5,5

b
‘2

t1
Hwy

.Au‘fi: 4‘
‘guk H".
”,thtn' 4'

 

THE”

 

 

Scout ‘_ ‘-—- T‘—
5" ’ W LIBRARY ‘
Michigan State '
University
L. L I .__.

 

This is to certify that the
dissertation entitled

Characterization of the transcriptional coactivator proteins
ADA2 and GCN5 in Arabidopsis

presented by
Yaopan Mac

has been accepted towards fulfillment
of the requirements for the

Doctoral degree in Genetics
' Cellular and Molecular BiologL

Majdr ngnature
1} / 5‘03

 

 

 

 

Date

MSU is an Afiirmative Action/Equal Opportunity Institution

PLACE IN RETURN Box to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE

DATE DUE

DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6/01 c-JCIRC/DateOuepSS-pJS

 

333W

CHARACTERIZATION OF THE TRANSCRIPTIONAL COACTIVATOR PROTEINS
ADA2 AND GCN5 IN ARABIDOPSIS

By
Yaopan Mac

A DISSERTATION

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY
Genetics Graduate Program
and

Graduate program in Cell and molecular biology

2003

Tll" ‘ “1"
...1t‘\\ ‘.

~
‘ Flim-
Lc‘u'yltn

9‘
w
u

:450
we

kit”: '5

~-
‘ "\ -

-~ ‘ x 1
g... J: a:
'.

'-~I§3tc

 

ABSTRACT

CHARACTERIZATION OF THE TRANSCRIPTIONAL COACTIVATOR PROTEINS
ADA2 AND GCN5 IN ARABIDOPSIS

By
Yaopan Mao

Chromatin modifications play important roles in regulating transcriptional activation
in eukaryotes. The GCN5-ADA2 protein complexes are core histone acetylation
modifiers that have been described in yeast and metazoans, where the GCN5 protein is a
histone acetyltransferase (HAT). The GCN5-ADA2 complexes are involved in the
activation of many genes in these organisms.

This study has been focused on the GCN 5 and ADA2 homologous proteins in the
plant Arabidopsis thaliana. Arabidopsis has two ADA2 homologs termed ADA2a and
ADA2b and one GCN5 homolog.

The ArabidopsisGCNS is a HAT that acetylates histone H3 primarily. GCN5
interacts with the ADA2 proteins. The regions of each protein involved in GCN5-ADA2
interaction have been defined. The ADA2 proteins enhance GCN5 HAT activity on both
free core histones and nucleosomal histones.

The ADA2 proteins interact with the protein CBF 1 , a cold stress response activator
in Arabidopsis. In addition, the interaction domain of CBF] with the ADA2 proteins has
been defined to the DNA binding domain of CBF]. These findings suggest that GCN5-
ADAZ complexes may play roles in plant cold acclimation.

The ArabidopsisADAZ proteins can be acetylated by GCN5 in vitro. The acetylation

site has been identified as lysine 215 in the ADA2b protein. The sequence around lysine

i

l mm
7 ‘TEECILA
-hmt

‘ a v."
"'K m 35‘s;
in - 5

215 in ADA2b is conserved in plant ADA2 proteins and is similar to that of the N-
terminal tail of histone H3. Mutant ADA2b that can no longer be acetylated by GCN5
can complement a T-DNA disruption allele of ADA2b for most growth phenotypes. This

leaves the significance of ADA2 acetylation an open question.

to Lin Mei

iv

.
mp“:

‘slthU

\
t.
s. ”it

ACKNOWLEDGEMENTS

I would like to thank my advisor Dr. Steve Triezenberg for his excellent and diligent
guidance and encouragement as well as his patience through my time in his laboratory.
His support and help are appreciated with my whole heart forever. I would like to thank
my committee members for their support and assistance. Their time and effort are greatly
appreciated. I would also like to thank Dr. Michael Thomashow for his advice on my
research.

I would like to acknowledge the members of the Triezenberg lab for their
companionship, friendship and scientific help and advice. Their participation in creating a
fiiendly and active atmosphere is appreciated very much. I would also like to thank Dr.
Kostas Vlachonasios for his help in establishment of transgenic plants.

I would like to thank my wife, family and friends. They have provided a constant
source of support through my life. Their support makes me go through all kinds

difficulties.

(ham
11: 61

TABLE OF CONTENTS

List of Figures ......................................................................................... x
List of tables ......................................................................................... xii
List of Symbols and Abbreviations .............................................................. xiii
Chapter One
Introduction and Literature Review ............................................................ 1-30
Pol II and the general transcriptional factors ............................................. 1
Activators ..................................................................................... 2
Coactivators .................................................................................. 3
TFIID .......................................................................................... 3
Mediators ..................................................................................... 6
The mechanisms of transcription activation by activators ............................. 7
Chromatin and chromatin modifiers ..................................................... 10
The function of chromatin acetylation .................................................. 12
The GCN5-containing histone modifiers ................................................ l4
Histone methylation ........................................................................ l8
Histone phosphorylation .................................................................. 23
Histone code and the cross talk of histone tail modification ......................... 23
The plant world ............................................................................ 28
Chapter Two
The GCN5 Gene and protein of Arabidopsis ................................................ 31-45
Introduction ................................................................................. 31
Experimental Methods ..................................................................... 32
Determining the genomic sequence of GCN5 gene ........................... 32
Arabidopsis suspension cell lysate .............................................. 33
Western blotting ................................................................... 33
Purification of recombinant GCN5 ............................................. 34
HAT asSay ........................................................... ’ ............... 35
Results ....................................................................................... 36
The ArabidopsisGCNS protein is a HAT ...................................... 36
The genomic sequence of GCN5 gene .......................................... 37
The endogenous GCN5 protein ................................................. 37
Discussion .................................................................................. 43
Chapter Three
The Interaction between GCN5 and ADA2 ................................................. 46-65
Introduction .................................................................................. 46
Experimental Methods ..................................................................... 47

vi

Re

Chap

an

GST pull-down assay ............................................................. 47

Yeast two-hybrid assay ........................................................... 48
Results ....................................................................................... 49
ADA2 interaction with GCN5 ................................................... 49
Defining the interaction domains/regions
between GCN5 and ADA2 ....................................................... 53
Discussion ................................................................................... 58
Chapter Four
Stimulation of GCN5 Enzyme Activity by ADA2b ........................................ 66-96
Part I Purification of recombinant protein of GCN5, ADA2a and ADA2b. . . . . . .66-74
Part II Both ADA2a and ADA2b enhance the HAT activity of GCN5 ................. 75-96
Introduction ................................................................................. 66
Part 1. Purification of recombinant protein of GCN5, ADA2a and ADA2b ....... 66
Experimental Methods ..................................................................... 67
Expressing plasmid construction ................................................ 67
Recombinant protein purification ............................................... 67
Western blotting ................................................................... 69
Experimental results ......................................................................... 69
Discussion for Part I ........................................................................ 73
Part II. ADA2 enhances the HAT activity of GCN5 .................................. 74
Experimental Methods ..................................................................... 74
Free core histones ................................................................. 74
Free histone acetylation assay ................................................... 75
Nucleosome purification ......................................................... 75
Nucleosomal histone acetylation assay ......................................... 77
Results ........................................................................................ 79
ADA2b enhanced GCN5 HAT activity ........................................ 79
The N-terminus and the middle part of
ADA2a/b enhance GCN5 HAT activity ........................................ 81
ADA2b enhances GCN 5 HAT activity on nucleosomal substrate ......... 83
The mechanisms of ADA2b to enhance GCN5 HAT activity ............... 89
Discussion for Part II ...................................................................... 89
Chapter Five
The Function of the Unique N-terminus of the ArabidopsisGCNS Protein ........... 97-133
Introduction ................................................................................. 97
Experimental methods ..................................................................... 98
The yeast two-hybrid system .................................................... 98
Construction of plasmids ......................................................... 98
Western blotting of bait protein GCN5 (1-210) expressed in yeast. . . . . ....99
cDNA libraries .................................................................... 100
Yeast cDNA library transformation ........................................... 101
Western blotting to check if MSI4
has the 50 amino acids extension .............................................. 102
Prey plasmid (library cDNA plasmid) DNA preparation .................. 103

vii

llipter

.JC lfllf

Confirmation of positive candidates .......................................... 104

Yeast cell plasmid transformation (mini-scale) .............................. 104
GST pull-down assay ............................................................ 105
Results ...................................................................................... 105
Yeast two-hybrid screening for proteins
interacting with GCN5 (1-210) ................................................. 105
Yeast two-hybrid screening for proteins
interacting with full-length GCN5 ........................................................... 1 11
Candidate #133 ................................................................... 1.17
Candidate #81 .................................................................... 121
Candidate #77 .................................................................... 123
Screening library CD4-10 using GCN5 (full length) as the bait .......... 125
Testing interaction with PCATs ............................................... 127
Discussion ................................................................................. 128
Chapter Six
The Interaction between CBF] and ADA2 ................................................ 134-157
Introduction ................................................................................ 134
Experimental Methods ................................................................... 134
Plasmid construction ............................................................ 134
Recombinant protein expression and purification ........................... 135
GST pull-down assay ............................................................ 136
Yeast two-hybrid assay ......................................................... 136
Gel mobility shift assay ......................................................... 137
Results ........ , .............................................................................. 138
Test the physical interaction between
CBF 1 and ADA2a/b or GCN5 ................................................. 138
Test the interacting region of CBF 1
with ADA2a/b and GCN5 ...................................................... 140
Determining the interacting motif ......
in the N-terminus of CBF] with ADA2 ...................................... 143
The AP2 domain ofTINY interacts with ADA2a and ADA2b............148
Detect ADA2b-CBF 1 -DNA tertiary complex .............................. 151
Discussion ................................................................................. 153
Chapter Seven
Acetylation of ADA2a and ADA2b by GCN 5 ........................................... 158-207
Part I Acetylation of ADA2a/b by GCN5 in vitro ....................................... 158-182
Part H Test the significance of ADA2b acetyaltion in vivo ............................ 182-207
Introduction ................................................................................ 158
Experimental Methods (Part I and Par 11) .............................................. 160
Purification of recombinant GCN5, ADA2b and CBF] proteins ......... 160
Purification of recombinant yADA2N ....................................... 160
Purification of ADA2b (1-1 79, 28-225, 28-179, 1-208) .................... 160
Acetylation assay for ADA2b .................................................. 161
Acetylation assay for ADA2bN,

viii

ADA2bM, ADA2bC and ADA2bN deletions ............................... 161
Mutations of putative acetylation site in ADA2a and ADA2b ............ 162
Acetylation assays for free and

nucleosomal histones with mutant ADA2b163

ADA2bN acetylation for mass spectrometry ................................. 164
Construction of plasmid pGEX6P-2bNR
which expresses ADA2bNR .................................................... 164
ADA2bNR acetylation for mass spectrometry ............................... 164
Constructing transformation plasmid for plant .............................. 165
Arabidopsis plant transformation .............................................. 165
Growth of transgenic plants .................................................... 166
Genotyping the offspring of the transformed plants ........................ 166
RNA analysis ..................................................................... 167
Results ...................................................................................... 167
Part I Acetylation of ADA2 by GCN 5 in vitro ....................................... 167
ADA2b rs acetylated by GCN5 in vitro.. ...167
The N-terminus of ADA2b rs acetylated by GCN5 in vitro ............... 170
K215 of ADA2b 1s necessary for acetylation ................................ 172
Determination of the acetylation
site of ADA2bN by mass spectrometry ....................................... 177
Part 11 Test the significance of ADA2b acetyaltion in viva ........................ 182
Mutant ADA2b is able to enhance GNCS HAT activity ................... 182
Dwarf and sterile ADA2b-/- Arabidopsis plants ............................. 182
Transgenic strategy .............................................................. 185
ADA2b (K215R) can complement ADA2b-/- plant phenotype ............ 187
ADA2b (K215A) can complement ADA2b-/- plant phenotype ............ 196
Discussion ................................................................................. 203
Chapter Eight
Perspective and Hypothesis .................................................................. 208-214
The GCN5-ADA2 complexes in Arabidopsis ........................................ 208
The targets of GCN5-ADA2 complexes .............................................. 210
The potential function of the unique
N-terminus of the GCN5 protein .................................................... ...212
The acetylation of ADA2 proteins by GCN5 ......................................... 213
Bibliography .................................................................................. 215-230

ix

F...”

. .iudb l-
h

" ‘N

»"N

“Av-b an
U

q
influx
. .‘&b-..
.

line I
”@363
i’ lire .'

LIST OF FIGURES

Images in this dissertation are presented in color.

Chapter Two
Figure l. Arabidopsis GCN5 protein is a histone acetyltransferase. ........................ 37
Figure 2. Diagram shows GCN5 gene and its encoded protein GCN5. ..................... 39
F igure3. Detection of endogenous GCN5 protein ............................................... 42
Chapter Three
Figure 1. GCN5 interacts with ADA2a and ADA2b. ......................................... 51
Figure 2. GCN5 interacts with ADA2a and ADA2b. ......................................... 52
Figure 3. The GCN5 HAT domain interacts with ADAZa and ADA2b. ................... 54
Figure 4. The middle part of ADA2a and ADA2b suffices to

interact with the HAT domain of GCN5. ........................................... 56
Figure 5. The N-terminus of ADA2 interacts with GCN5 and the middle

part of ADA2 interacts with the HAT domain of GCN5. ......................... 57
Figure 6. Differential interaction of the middle part of ADA2a

and ADA2b with the HAT domain of GCN 5 ........................................ 59
Figure 7. The N-terminus of ADA2b but not ADAZa interacts with GCN 5 (3 69-

568). . . . .60

Figure 8. The extended N-terminus of GCN 5 interacts with ADAZa but not ADA2b. ..61

Chapter Four
Figure l. The purified recombinant proteins GCN5, ADA2a and ADA2b .................. 71
Figure 2. Confirmation of the purification of recombinant protein ADA2b. . . . . . .72
Figure 3. ADA2b enhances the HAT activity of GCN5 on fi'ee core histones. ............... 80
Figure 4. The N-terminus and the middle part of both ADA2a and ADA2b enhance
the HAT activity of GCN5 on free core histones. ................................. 82
Figure 5. The combinatorial effect of the N-terminus and the middle part of ADA2b in
enhancing the HAT enzyme activity of GCN5 on free core histones ................ 84
Figure 6. Checking the purified nucleosomes from turkey erythrocytes .................... 85
Figure 7. Comparing the HAT activity on free core histones of Arabidopsis
GCN5-ADA2 with that of yeast HAT complexes SAGA and NuA4 .............. 87
Figure 8. Arabidopsis GCN5 does not acetylate nucleosomal
core histone while GCN5 plus ADA2b does weakly. ............................ 88
Figure 9. Differential HAT activities of GCN5-ADA2b to
varying amount of Ac-CoA and core histotons ..................................... 90
Chapter Five
Figure 1. Bait protein GDBD-GCNS (1-210) was expressed in yeast host
strain AH109. .......................................................................... 108
Figure 2. The protein GCN5 (1-210) was not toxic, and did not activate
reporter genes in yeast host strain AH109. ....................................... 109

Figure 3. The protein GCN5 (full length) was not toxic, and did not activate

reporter genes in yeast host strain AH109. ........................................ 1 13
Figure 4. Confirmation of the candidates from the two-hybrid screening

using the full-length GCN5 bait ...................................................... 115
Figure 5. The isolated ADA2a and ADA2b candidates by GCN5 (fl) bait
in yeast two hybrid screening ........................................................ 1 16
Figure 6. Native protein of ArabidopsisMSI4 contains the 50 amino
acids in its N-terminus. .............................................................. 120
Figure 7. Candidate #133 lost interaction with GCN5 when the extra leading
20 amino acids were deleted. ........................................................ 122
Figure 8. The full length of candidate #77 protein interacted with the
HAT domain and the H-B region of GCN5. ....................................... 124
Figure 9. Candidate #77 was a potential activator and interacted with
the HAT domain GCN5. It is a potential activator. ............................... 126
Figure 10. Testing the interaction of GCN5 with PCAT] and PCAT2,
the Arabidopsis orthologs of CBP/p300 ............................................ 129
Chapter Six
Figure 1. The transcriptional activation of protein CBF] is dependent
on yeast protein ADA2, ADA3, and GCN5. ..................................... 139
Figure 2. Interaction of CBF] with ADA2a, ADA2b and GCN5 ........................... 141
Figure 3. The C-terminus of CBF 1 is a transcriptional activator. ......................... 142
Figure 4. Structure of the AP2 DNA-binding domain ........ 144
Figure 5. The N-terminus of CBF] interacts with ADA2b and GCN5. .................. 145
Figure 6. Defining the minimal interaction domain of CBF 1 with ADA2a and
ADA2b ................................................................................... 147
Figure 7. Differential interaction of AP2 domain with ADA2 proteins .................... 149
Figure 8. Testing the CBF 1 protein binding to DNA probe .................................. 152
Figure 9. Detect ADA2b-CBFl-DNA tertiary complex ...................................... 154
Chapter Seven
Figure 1. ADA2b is acetylated by GCN5. .................................................... 169
Figure 2. The N-terminus of ADA2b is acetylated by GCN5 ............................... 171
Figure 3. The N-terminus of Arabidopsis ADA2a is acetylated by GCN 5. .............. 173
Figure 4. Defining the acetylation site of ADA2b N-terminus ............................... 174
Figure 5. Yeast Gcn5 acetylates ArabidopsisADAZb. ....................................... 176

Figure 6. Lysine residues in ADA2a (1-274) and ADA2b necessary for acetylation. ..178
Figure 7. The ADA2b mutants enhance the HAT activity of GCN5 on

free core histones ...................................................................... 183
Figure 8. Mutant ADA2b proteins stimulate GCN5 HAT activity on
nucleosomal histone .................................................................... 184
Figure 9. The dwarf phenotype of Arabidopsis with homozygous T-DNA
insertion at the ADA2b locus ......................................................... 186
Figure 10. Genotyping the offspring of transformed plants .................................. 188

Figure l 1. Comparison of the silique length of R12 plant to that of wild type plant. . ...190
Figure 12. Over-expression of transgenic cDNA of ADA2b (KZISR)

xi

did not rescue the ADA2b-/- phenotype. ......................................... 191
Figure 13. Over expression of transgene is required to rescue the ADA2b-/-

phenotype .............................................................................. 192
Figure 14. The phenotype rescue of plant ADA2b-/- by the expression of
transgenic cDNA of ADA2b (K215R) in line R16 ............................... 194
Figure 15. The acetylation of ADA2b at lysine 215 in vivo is not required
for plant development ............................................................... 195
Figure 16. Over expression of transgenic cDNA is required to rescue
ADA2b-l- phenotype ................................................................. 197
Figure 17. Expression of transgenic cDNA (K215A) rescues plant
ADA2b-/- phenotype. .............................................................. 199
Figure 18. Partial rescue of phenotype of plant ADA2b-l- by over-expression
of transgenic cDNA of ADA2b (K215A). ........................................ 200
Figure 19. Partial rescue of the phenotype of plant ADA2b-/- by over-expression of
transgenic cDNA of ADA2b (K215A). ........................................... 201
Figure 20. Complementation variation with over-expression of transgenic cDNA
to rescue ADA2b-/- phenotype. .................................................... 202
CONTENTS OF TABLES
Table 5-1 Summary of the candidates .......................................................... 1 18
Table 7-1 Q-TOF mass data for ADA2bNR ................................................... 18]

xii

i

ii

PD“
3.0
2280
ill
’31]
0.-
1E?
{31)
WAGE
1.5;? T8?

T]

”L.

AD

bp

CPM
GAD
GDBD
HAT
IPTG
kDa
MET

OD
SDS-PAGE
TAF TBP
TFII

LIST OF SYMBOLS AND ABBREVIATIONS

activation domain

base pair

counting per minute

Gal4 activation domain

Gal4 DNA binding domain

histone acetyltransferase
isopropyl-2-D-thiolgalactopyranoside
kiloDalton

methyltransferase

optical density

sodium-dedecyl-sulfate polyacrylamide gel dlectropheresis
associated factor

transcriptional factor, RNA polymerase II

xiii

itizil’

TI” 1

..L-.., _

:“gga
‘nl‘l\

nyr

Chapter One

Introduction and Literature Review

Pol II and the general transcriptional factors

The transcription of protein-coding genes in eukaryotic nuclei is carried out by RNA
polymerase II (pol 11). Pol II consists of 10-12 subunits, which are well-conserved across
species (Hampsey, 1998, Woychik and Hampsey, 2002). One eminent feature of po] 11 is
the carboxyl-terminal domain (CTD) of the largest subunit, a heptapeptide repeat ranging
fi'om 26 repeats in yeast to 52 repeats in human. The CTD is under extensive and
differential phosphorylation regulation during transcription and thus provides a switch to
regulate pol II activity during transcription (Carlson, 1997, Hampsey, 1998‘, Majello and
Napolitano, 200]). For example, hyper-phosphorylation of the CTD accompanies the
transition of pol II from transcriptional initiation to elongation, and hypo-phosphorylation
of the CTD is associated with the termination of elongation.

The general transcriptional factors (GTFs) of RNA pol I] include TFIIA, TFIIB,
TFIID, TFIIB, TFIIF and TFIIH‘(Hampsey, 1998, Orphanides, et al., 1996) TFIID can
also be categorized as a co-activator (see below). TFIID contains the TATA binding
factor TBP, and is the first GTF to bind to promoter to initiate the assembly of
transcriptional machinery (Burley and Roeder, 1996). These GTFs are identified through
in vitro studies of transcription constitution, where the adding of these GTFs enabled pol
II to initiate transcription precisely from the promoter in a DNA template (Roeder, 1996).

The GTFs recruit pol II to the core promoter and participate in promoter melting,

DNA template unwinding and pol II dispatch from initiation to elongation (Orphanides,

one fan“
mm C
Ill and II
3.1:: and l.
mind 10 3"
14mm 1
his is sum“
raripiion

raisin Srbl.

m

et al., 1996).

One form of pol II is the so-called holoenzyme. The biochemically purified yeast
holoenzyme consists of some GTFs (including TFIIB, TFIIB, TFIIH, TFIIF, but not
TFIIA and TFIID) and the Srb/Mediator complex which associates with the CTD of pol
II (Lee and Young, 2000). The holoenzyme, unlike the simpler form of pol II, is able to
respond to activators to initiate transcription in vitro (Koleske and Young, 1994). The
holoenzyme is globally required for transcription in viva (Thompson and Young, 1995).
This is supported by the microarray data (Holstege, et al., 1998), in which the
transcription of more than 93% of the 5361 tested yeast genes were dependent on the Srb
protein Srb4. Srb4 is known to be part of pol 11 holoenzyme exclusively (Myers, et al.,
1998).

Similar pol II holoenzymes have been purified from human and murine cells.
Interestingly, a broader range of proteins are found in mammalian pol 11 holoenzyme,
including the chromatin modifiers CBP (Neish, et al., 1998), the SWI/SNF complex
(Wilson, et al., 1996) and DNA repair proteins (Maldonado, et al., 1996). Like the
situation in yeast, human cells have several forms of Srb/Mediator complexes, including
SMCC/TRAP (Ito, et al., 1999), ARC (Naar, et al., 1999), DRIP (Rachel, et al., 1999),
and CRSP (Ryu, et al., 1999). These complexes are the interacting targets of numerous

transcriptional activators and nuclear hormone receptors (Rachez and Freedman, 2001).

Activators
The transcriptional activation of protein-coding genes usually starts with transcriptional
activators binding to promoters or/and enhancer elements (Bryant and Ptashne, 2003).

Transcriptional activators are typically modular having both a DNA-binding domain,

_,.-u ,'
night

fia‘tfil
i-‘-\i 5|
I

ECU):

IL,

m
'T“

- «as:
l .J ‘1 a]

.
~11

Etna

which lead activators to specific genes, and an activation domain (AD), which activates
transcription (Ptashne and Gann, 1997, Triezenberg, 1995). Multiple activators generally
coordinate the transcriptional activation of a specific gene (Carey, 1998), and one specific

activator can participate in activation of multiple genes.

Coactivators

Co-activators are broadly defined as transcriptional factors that facilitate
transcriptional activation but lack cognate DNA binding activities. The studies of
eukaryotic protein-coding gene transcription could be roughly divided into two ages: the
DNA age, and the chromatin age which began nearly a decade ago. Both ages have
uncovered many co-activators. Co-activators uncovered in the DNA age include TFIID
and Srb/Med complexes, which interact with the pol II transcriptional machinery. Co-
activators found in the chromatin age are chromatin modification/remodeling complexes,
such as SWI/SNF and SAGA complexes.

Their functions can be broadly divided into two categories: directing activator
recruitment of the transcriptional machinery and chromatin remodeling or modifying
(Naar, et al., 2001). This might reflect that chromatin remodeling and general
transcription machinery assembly is an integrated business during transcription initiation.
Co-activator TFIID and Mediator are discussed in more detail below, and chromatin

modifiers are discussed later.

TFIID

TFIID is composed of TBP and about 14 TAFs (for TBP associated factors) (Burley

and Roeder, 1996). These components are well conserved across species. The

multifunctional activities of TFIID include core promoter binding, activator interaction,
histone acetyltransferase (HAT) activity, kinase activity and ubiquitin ligase activity
(Burley and Roeder, 1996).

Although TBP, a component of TFIID, is considered a GTF, the TAF components of
TFIID have the characteristic of co-activators (Albright and Tjian, 2000). TFIID binds
promoter DNA. TBP binds the TATA element, and some TAFs bind the promoter
element of Initiator (INR) and down stream element (DPE). Because of its promoter
binding capacity, TFIID is generally believed the first recruited GTF to nucleate the
assembly of transcriptional initiation machinery in target genes (Naar, et al., 2001). It has
been proposed that TFIID is globally required for gene transcription (Pugh and Tjian,
1992). Recently, Mencia et al proposed that TFIID is required for transcription from
promoters with weak or nonexistent TATA elements including most ribosomal protein
genes (Mencia, et al., 2002).

The largest TAF in humans (TAFn250 or TAF 1) has two tandem bromodomains that
can bind acetylated histone H4 tail at lysine 5 and 12 in vitra (Jacobson, et al., 2000).
This suggests that TFIID can bind to, besides DNA, the histone tail at promoters in viva.
Remarkably, the acetylation state of histone tails at promoters marks the transcription
competence of genes. How TFIID incorporates the acetylation status of promoters
through bromodomain binding into transcription regulation is not known.

Not only does TFIID bind to chromatin DNA and histone tails, one component of
TFIID has been proposed to mimic DNA structure. The largest TAF inhibits TBP binding
to the TATA box in vitro (Kokubo, et al., 1998). The N-terminus of the Drasaphila

largest TAF (TAF11230) has been found to resemble the minor groove surface of the

{Ali but in h

:i‘rizon' et‘t‘cc

Willi 10 W
l

. m“

fillies in VI!
mmmmi
tilii. hiti R‘dl
Tit-[150. sug
fighting l

The Dras-
fi‘lii. H1 is ll
£130th the
iiquilinylalil
iii: and Sa
mtipliona

In lIlTO c'
timing dir
will. Direct
“13in pm.

All 10 gr!
Emillfrg. w
me bindi

““ 31in M

TATA box in binding to the concave side of TBP protein (Liu, et al., 1998). The auto-
inhibitory effect in TFIID is proposed to provide a “hand-ofi” switch for activators such
as VP16 to recruit TFIID in transcriptional activation of genes (N ishikawa, et al., 1997).

The largest TAF in TFIID also has HAT (histone acetyltransferase) and kinase
activities in vitra (Wassarman and Sauer, 2001). Whether TFIID acetylates promoter
histones in viva is unknown, but clearly this HAT activity is important. Mutant TAF11250
which has reduced HAT activity can not complement a temperature sensitive mutation of
TAF11250, suggesting that acetyltransferase activity is required for cell cycle progression
by regulating the expression of essential proliferative control genes (Dunphy, et al., 2000).

The Drasaphila TAF11230 mono-ubiquitinylates histone H1 in viva (Pham and Sauer,
2000). H] is the linker histone which is important for higher order structure of chromatin.
Although the function of H1 ubiqintinylation is not known, TAF 11250 defective in mono-
ubiquitinylation reduces expression of genes targeted by the maternal activator Dorsal
(Pharn and Sauer, 2000). This suggests the importance of H] ubiquitinylation in
transcriptional activation.

In vitra data have found many transcriptional activators interact with various TAFs,
indicating directed recruitment of TFIID in transcriptional activation in viva (Naar, et al.,
2001). Direct evidence in viva is the recruitment of TFIID by activator Rap] to ribosomal
protein promoters recently reported by the Struhl group (Mencia, et al., 2002).

All together, a scenario is emerging for TFIID. Activators target TFIID to core
promoters, where TFIID binds precisely to specific promoters through its DNA and
histone binding capacity. The histone acetylation and ubiquitinylation activities of TFIID

may then help to decompress chromatin structure. TFIID then initiates the assembly of

liedizlnr
’

lit 3
13315?-
E A LVC

lit: amt

V

.9." 4'
'll ‘
&.§|~‘\Ll

transcriptional machinery.

Mediators

The yeast Mediator was originally isolated biochemically for its ability to stimulate
activator-dependent transcription activity in reconstituted transcriptional reactions (Malik
and Roeder, 2000). There are three protein groups in the yeast Mediator: Srb proteins,
Med proteins, and a diverse group of proteins (Myers and Kornberg, 2000). The Srb
proteins are genetically identified as the suppressors of pol II CTD truncation mutants
(Myers, et al., 1998). The yeast Mediator exists either as s fi'ee complex or associated
with po] 11 through physical interactions with the pol II CTD. Yeast Mediator is generally
required for transcriptional regulation in viva (Lee and Young, 2000).

Mediator-like complexes have also been isolated from mammalian cells.
Interestingly, mammalian Mediators are more diverse (Myers and Kornberg, 2000, Naar,
et al., 2001), including complexes such as SMCC/TRAP, NAT, CRSP, ARC, and DRIP.
All these mammalian Mediators share a subset of common components (Rachez and
Freedman, 2001). Therefore, they might be the isoforms of a core Mediator dependent on
the status of cells or purification methods (Naar, et al., 200]).

Exactly how Mediators work in transcriptional activation is not well known. Yeast
Mediator stimulates TFIIH kinase activity up to 30 fold (Kim, et al., 1994). This suggests
that Mediator stimulates pol H to escape from promoters and proceed to elongation.
Because of their interaction with both activators and pol 1], Mediator proteins might act
as an interface between activators and the general transcriptional machinery (Myers and
Kornberg, 2000, Naar, et al., 2001).

Interestingly, the Mediator protein Nut] from yeast is found to be able to acetylate

.
Tz"‘vlnr
I-‘ At.

3.73431;

1’).

‘.

1.. "'
Ethyld‘»
w

n

histones, and it belongs to the GCN5-related GNAT HAT family (Larch, et al., 2000).
This might indicate that Mediator might also modify chromatin during the regulation of

transcription.

The mechanisms of transcription activation by activators

Activators activate transcription by exploiting the facts that (i) pol II can not initiate
transcription automatically from DNA template, (ii) GTFs can guide pol II to genes’
promoters to initiate transcription (Woychik and Hampsey, 2002), and (iii) the chromatin
generally possesses inhibitory effect on pol II transcription (Struhl, 1999). A few
mechanisms are currently understood for activators to work.

First, activators activate transcription by stimulating assembly of transcriptional
initiation machinery at core promoters (Ptashne and Gann, 1997). This could be achieved
by the stepwise and concerted assembly of GTFs and po] 11 at core promoters. The
stimulated assembly can also be accomplished by one-step loading of pol H holoenzyme
at core promoters. It is possible that in viva the activator-stimulated assembly of
transcriptional initiation machinery at core promoters is a multiple-step procedure, which
is in between the two extremes described above (Lee and Young, 2000).

This mechanism of activation is suggested by the in vitra observations that activators
interact physically with proteins from GTFs, po] 11 and the Srb/Mediator (Barberis and
Gaudreau, 1998, Burley and Roeder, 1996, Lee and Young, 2000, Ptashne and Gann,
1997). In support of this mechanism, when the components of GTFs, pol II and the
Srb/Mediator are artificially brought to the core promoter by firsed DNA binding domains,

the requirement of activators is no longer necessary (Keaveney and Struhl, 1998). The

ugh
9‘39 H'

a: 1 A' ’
" I) .
1A.. Jpn“!

tested factors for artificial recruitment have included TBP (Chatterjee and Struhl, 1995,
Klages and Strubin, 1995), Gall] (Farrell, et al., 1996, Gaudreau, et al., 1997), TBP
associated factor (TAFIIs) (Gonzalez-Como, et al., 1997, Keaveney and Struhl, 1998),
and TFIIB (Gonzalez-Como, et al., 1997).

Secondly, activators recruit chromatin-remodeling complexes to facilitate
transcription. Various chromatin remodeling complexes, including ATP-dependent
nucleosomal modifiers (such as SWI/SNF complex) and histone covalent modifiers (such
as SAGA complex) have been reported to be recruited by numerous activators. For
example, the HAT complex of SAGA can be recruited by Gal4 (Larschan and Winston,
2001), VP16 (Utley, et al., 1998), Gcn4 (Swanson, et al., 2003), and Pho4 (Barbaric, et
al., 2003); the SWI/SNF complex is recruited by SwiS (Cosma, et al., 1999), EKLF
(Kadam, et al., 2000), VP16, Gcn4 and Hap4 (Neely, et al., 1999); and co-activator CBP
by nuclear hormone receptors, CREB, EZF, and p53 (Chan and La Thangue, 200]).

The importance of co-activator chromatin modifiers in transcription is strongly
supported by two facts: (i) the existing physical interaction between activators and the
components of chromatin modifiers, (ii) dysfunctional chromatin modifiers leading to
defective transcriptional activation of target genes. For instance, reporter gene
transcription from in virra assembled chromatin is dependent on the SAGA complex and
acetyl-CoA (Utley, et al., 1998). In viva, the GCN4 dependent activation of HIS3 in yeast
is dependent on SAGA HAT activity (Kuo, et al., 1998, Wang, et al., 1998).

Activators can also use other mechanisms to activate transcription. For instance, the
HIV Tat activates transcription by stimulating the processivity of pol II (Garber and

Jones, 1999, Jeang, et al., 1999, Karn, 1999). Activators also stimulate promoter escape

it“ 5102;:
£33:th

m 31: ‘re

attain;
II]: 1. {ml
.in docs
3CD 3103
Still. 2::
Wm

Kim

I“ .

mater
PM a

’c‘ihlIEd
wired

and elongation rate (Blau, et al., 1996, Brown, et al., 1998, Krumm, et al., 1995).
Activators might also induce changes in structure of co-activators to enable them to
activate transcription. For instance, the co-activator ARC becomes a smaller co-activator
CRSP when induced by activator VP] 6 or SREBP-la, and the induced CRSP is
contrastingly different from ARC in three dirnensional conformations (Taatjes, et al.,
2002). Interestingly, only the CRSP is able to embark on activated transcription while
ARC does not (Taatjes, et al., 2002).

These mechanisms described above are not mutually exclusive for a specific
activator. For instance, the typical acidic activator VP16 is able to recruit TBP (Hall and
Struhl, 2002), histone acetyltransferase complex SAGA (Utley, et al., 1998) and
SWI/SNF complex (Neely, et al., 1999).

Many activators can recruit multiplicity of co-activators for transcriptional activation.
For instance, the yeast GCN 4 recruits as many as seven co-activators to activate
transcription of various genes, including co-activator SAGA, SWI/SNF and Mediator
(Swanson, et al., 2003). Gcn4 seems able to recruit all these co-activators directly,
because direct physical interaction with these co-activators exists (Natarajan, et al., 1999,
Swanson, et al., 2003). Interesting, Swanson et al propose that not all recruited co-
activators are necessary for a specific promoter, only a subset of co-activators are
required. This multitude of recruitment might be explained by the fact that Gcn4 is
required for hundreds of gene promoters and for each promoter only some co-activators
are needed.

In summary, the action of activators is to influence transcription either

thermodynamically by recruitment of transcriptional factors or kinetically by alteration of

 

1 - ‘
35:733.

first it

.I: ‘3’ T '
.4. ii“

the activity of transcriptional factors.

On the other hand, the activation of a specific gene usually makes use of a
combination of the mechanisms, where activators bring in chromatin modifiers and the
transcriptional initiation machinery in an orchestra] manner (Orphanides and Reinberg,
2002). This is well illustrated in the activation of yeast H0 gene (Cosma, et al., 2001,
Cosma, et al., 1999). In the HO promoter, the activator Swi5 recruits SWI/SNF complex
first, which then facilitates the recruitment of SAGA histone acetylase complex in the
absence of Swi5. The remodeled chromatin is then able to bind a second activator SBF
which orchestrates assembly of the transcriptional initiation machinery.

One theme of the concerted recruitment is the requirement of multitude of functions.
In fact, many co-activators are multifunctional. For instance, the yeast SAGA complex
has at least two activities: chromatin remodeling and TBP binding (Sterner, et al., 1999).
Some co-activators such as CBP/p300 possess these functions in one molecule (Goodman
and Smolik, 2000). This design, presumably through the fusion of multiple proteins

during evolution, simplifies the multiple steps of recruitments.

Chromatin and chromatin modifiers

In the eukaryotic nucleus, the genomic DNA is packaged into a well-organized
nucleoprotein structure known as chromatin. The basic building unit of chromatin is the
nucleosome, which consists of a core histone octamer (two copies of each of H2A, H2B,
H3 and H4) wrapped by 147 base pairs of DNA (Luger, et al., 1997). Although the
nucleosome structure has been resolved (Luger, et al., 1997), the higher ordered structure
of chromatin is still poorly understood. There are linker histone H1 and many non-histone

proteins incorporated into chromatin.

10

l ‘— V
si‘CL-‘I
' '1
,rfIELAC
*i M; .
yuan.
D
iii-1‘09

: it

“0.".

er- ‘
:AT

'J‘V'Fr -

an.“ .I

It is now clear that the chromatin is not just a way of storaging genomic DNA, it is
subject to dynamic re-organization and plays crucial role in regulating transcription
(Jenuwein and Allis, 200]). The chromatin maintains the transcriptional restrictive
ground state by preventing some transcriptional factors from binding (Struhl, 1999). The
nucleosome has a general inhibitory effect on transcriptional initiation in vitra [(Lorch, et
al., 1987, Workman and Roeder, 1987), and loss of nucleosomes in viva increases
transcriptional initiation (Han and Grunstein, 1988).

Each histone tail (including the N-termini of all four core histones and the C-
terminus of H2A) is strikingly conserved, and very basic in amino acid composition.
These tails comprise about 25% of the core histone mass. They are not required for the
nucleosome assembly, but they are, together with linker histone H1, critical for higher
order structure of chromatin (Wolffe and Hayes, 1999). Indeed, various tail-DNA
interactions have been observed in vitra for proper chromatin assembly and transcription
regulation (Lee and Hayes, 1997, Loyola, et al., 2001, Tse and Hansen, 1997).
Genetically, all the four tails are required for basal transcriptional repression (Lenfant, et
al., 1996).

The sophisticated organization of chromatin provides a platform for modification
and remodeling in regulating transcription. There are two major enzymatic modifications
of the highly compact chromatin, (i) ATP-dependent remodeling of the nucleosomes
which is carried out by the SWI/SNF like remodelers (Peterson and Workman, 2000), (ii)
covalent modification of the histone proteins, including acetylation, methylation,
phosphorylation, ADP- and ubiquitinylation (Allen, et al., 1998, Spencer and Davie,

1999).

11

Hison

~~ ' I
4.0- «.nh ‘

:L Mtlzk

T ‘9"!91‘,
33.33. mm;

III: I

Histone acetylation has been the most extensively investigated among all the histone
modifications. Histone acetylation happens in the N-tails of all four core histones. The
consequence of histone acetylation is the acetyl moiety from acetyl-CoA is covalently
attached to the amide group of lysine residue. This attachment provides the platform for
subsequent signaling which eventually leads to the remodeling of targets (Fischle, et al.,

2003).

The function of chromatin acetylation

It has been known for decades that histone acetylation has intimate connection with
transcriptional regulation (Grunstein, 1997, Struhl, 1998), and acetylated histones are
usually a hallmark of transcriptionally active or competent chromatin region (Hebbes, et
al., 1988).

Numerous studies have revealed that chromatin acetylation is related to active
transcription. Kuo et al illustrated that H3 and H4 are hyper-acetylated at
transcriptionally active promoters and enhancers (Kuo, et al., 1998). At the B-globin locus
of chicken and mammalian erythroid cells, there is general background low level of
acetylation throughout the active locus, hyper-acetylation of H3 and H4 occurs only at
the regulatory site and the actively transcribed genes (Bulger, et al., 2002, Litt, et al.,
2001). Histone acetylation also maintains chromatin competent for transcriptional
activation. For instance, the chicken B-globin HS4 insulator keeps insulated chromatin
region at high acetylation state (Mutskov, et al., 2002).

At promoters, HAT recruitment and hyper-acetylation with transcriptional activation

initiation is well documented (I-Iassan, et al., 200]). For instance, the transcription from a

12

flux for!
LII—oi Lani:
WW" t
3.x.'lt‘.il'

11-33::

maze

“>le

“$110:

chromatin template in vitro is stimulated by recruited local acetylation (Utley, et al.,
1998). In viva, hyperacetylation precedes transcriptional activation at yeast PH08
promoter (Reinke, et al., 200]), and HIS3 promoter (Kuo, et al., 2000).

Acetylation of histones at promoters is generally believed to facilitate the binding of
other transcriptional factors, such as GTFs and other chromatin modifiers. In vitra, local
acetylation by SAGA helps to recruit SWI/SNF complex (Hassan, et al., 2001). In viva,
hyperacetylation at promoter improves TBP binding in human estrogen-responsive p82
promoter (Sewack, et al., 2001), improves the recruitment of transcriptional activator
Adr] in yeast ADH2 promoter (V erdone, et al., 2002), improves the recruitment of CBP-
Pol II holoenzyme complex (Agalioti, et al., 2000), and improves the recruitment of
SWI/SNF or TFIID (Agalioti, et al., 2002). Interestingly, differential acetylation of
histones seems to have different meanings. Agalioti et a] have shown that, in the
promoter of IFN-beta, acetylation of histone H4 K8 mediates recruitment of the
SWI/SNF complex whereas acetylation of K9 and K14 in histone H3 is critical for the
recruitment of TFIID (Agalioti, et al., 2002).

Recently, the Horz group found that, during the activation of the PH05 gene in yeast,
the histones altogether dissociate from the promoter region of PH05 gene following
hyperacetylation by the recruited SAGA (Reinke, et al., 200]). This finding suggests
additional functions of promoter hyperacetylation.

Compared to the hyperacetylation at regulatory elements, the importance of histone
hyperacetylation at genes’ coding regions is relatively less clear. There is high level
genomic wide chromatin acetylation in yeast, and this acetylation is severely affected by

mutations in HAT enzymes GCN5 and Esal (Kristjuhan, et al., 2002). In vitra,

l3

Kl) ill}:

'l‘Mn-a

L. .1115;

acetylation of H3 and H4 tails relieves chromatin repressive effect on T7 transcription
elongation on nucleosomal template (Protacio, et al., 2000). This suggests that the
progressive transcription along the coding regions require acetyaltion. A HAT complex
known as elongator has been found to associated with elongating pol II in yeast

(W ittschieben, et al., 1999), and similar human elongator has been shown to facilitate po]
11 transcription through chromatin in vitra (Kim, et al., 2002). Severe histone H3
hypoacetylation in the coding region can be achieved by double mutation of Gcn5 and
Elp3. This hypoacetylation coincides with transcription inhibition, suggesting
hyperacetylation in coding regions is important for sufficient elongation (Kristjuhan, et

al., 2002).

The GCN5-containing histone modifiers

Numerous histone acetylation modifiers have been identified from budding yeast to
human to plants (Roth, et al., 2001, Stockinger, et al., 200], Tanner, et al., 2000). These
histone acetylation proteins and complexes cover a wide range of activity in transcription,
such as TFIID which is a basic component of transcriptional machinery, Mediators which
are either an independent complex or associated with po] 11 holoenzyme, and nuclear
receptor co-factors such as CBP/p300 which are involved in many aspects of
transcriptional activation.

Most of the histone modifiers are highly conserved among species. For instance,
CBP/p300 and their HAT activity are conserved from C. elegans to human and to plants
(Bordoli, et al., 2001, Calvo, et al., 2001). Another feature of these histone modifiers is

that they are mostly multi-peptide complexes. The most complicated one might be yeast

14

rm 0

.1 a.-
. .

I
36""
I ‘5. a C
5

Lair?

" I
lr‘f“, ' ‘
VLI- 4L

‘m,~.. .

‘- - .4 Ofi

(r ’1.“~ ~h|
A

Mediator which contains more than 20 subunits (Woychik, 1998).

Based on their acetyltransferase domain sequence similarity and function, five
groups of histone acetyltransferases have been described (Roth, et al., 200] , Sterner and
Berger, 2000), including CBP/p300, TAF l/T AF [1250, ACTR/SRC, the GNAT family,
and the MY ST family that contains enzymes such as Esal, 8352 and Sas3 in yeast.

GCN 5 and its homologues such as PCAF form the HAT family of GNAT which is
found in all eukaryotes (Roth, et al., 200] , Sterner and Berger, 2000). Recombinant yeast
Gcn5 acetylates preferentially histone H3 lysine l4 and H4 lysines 8 and 16, but it has no
apparent activity on nucleosomal histones (Kuo, et al., 1996).

There are at least three distinct GCN5-containing complexes in yeast, namely ADA
SALSA, and SAGA (Eberharter, et al., 1999, Grant, et al., 1997, Kuo, et al., 1996,
Sterner, et al., 2002). The ADA complex (0.8 MDa) contains proteins Ada2, Ada3, Ahcl
and Gcn5. The SAGA complex (for Spt-Ada-GcnS-acetyltransferase) (1.8 MDa) is
composed of four classes of proteins, transcriptional adapter proteins (Adal , Ada2, Ada3,
Ada4/Gcn5, and Ada5/Spt20), Spt proteins (Spt3, Spt7, Spt8), TBP associated factors
(T afn90, —68/61, -60, -25/23, and -20/17), and Tra]. The SALSA complex is very similar
to SAGA except it lacks Spt8 and contains a truncated version of Spt7. There is also a
forth Gcn5-containing complex reported in yeast called A2 that consist of proteins Ada2,
Ada3 and Gcn5 (Sendra, et al., 2000).

Recombinant Gcn5 alone can not acetylate nucleosomal histones in virra. But ADA
and SAGA complexes are able to acetylate nucleosomal histones. This indicates the
functional participation of the additional components in both communication with

nucleosome and regulating Gcn5 activity (Balasubramanian, et al., 2002). In these

15

D.
AR'NL

HIO'
tel“
11 Cl
.‘C

C...
4111a;

complexes, Gcn5 acetylates more lysine residues on histone H3, and also acetylates H2B
and H4 (Eberharter, et al., 1999, Grant, et al., 1997, Kusch, et al., 2003).

The Ada proteins were originally identified genetically as suppressors of toxicity
caused by transcriptional activator VP16 in yeast (Berger, et al., 1992, Pina, et al., 1993).
The AdaZ protein fi'om yeast and metazoans has two conserved domains at its N-ter'minus,
the ZZ finger domain and the SANT/Myb domain. The SANT domain is also found in
proteins Swi3, TFIIIB and nuclear co-repressor NCo-R (Aasland, et al., 1996). The
SANT domain is required for AdaZ to support transcriptional activation (Candau, et al.,
1997) and recent findings have indicated that SANT domain is essential for Gcn5
acetylation activity on nucleosomal histones (Boyer, et al., 2002, Sterner, et al., 2002). In
yeast, the ZZ finger is essential for interaction with Ada3 and Gcn5 in vitra (Candau, et
al., 1 997).

The Spt proteins were originally isolated in genetic screens for mutants that
compensate for the defects caused by transposon element insertion at yeast promoters.
They are firnctionally connected with TBP, and Spt3 interacts with TBP (F assler and
Winston, 1988, Winston and Sudarsanam, 1998).

The largest protein in SAGA is the Tral protein which has more than 2000 amino
acids. Protein Tral belongs to ATM (ataxia telangiectasia mutated)/DNA-
PK/phosphatidylinositol 3-kinase family including human TRRAP protein, implicated in
cell cycle checkpoint signaling and cellular response to DNA damage (Abraham, 2001,
Grant and Berger, 1999, Khanna, et al., 2001, Vassilev, et al., 1998).

There are at least two fimctions associated with SAGA: the HAT activity to acetylate

histones and the interaction with TBP to regulate the assembly of transcriptional

16

ifif. 10 I?

1'?" \H
3.51.. t. g
l

machinery. Genetic and biochemical analyses in yeast also reveal distinct roles of SAGA
proteins (Berger, et al., 1992, Grant, etal., 1997, Sterner, et al., 1999). Null mutation in
ADA 1, SPT7 or ADA5/SPT 20 result in severe growth and transcription phenotype with
loss of SAGA complex, while null mutations in GCN5, ADA2, ADA3, SPT3 and SPT8
lead to moderate phenotype and the SAGA complex still remain intact. Loss of both
Gcn5/Spt3 or Gcn5/Spt8 produces similar phenotype as that of Ada] , Spt7 or Spt20.
These results demonstrate that there are two distinct protein groups in SAGA complex.
Similar SAGA complexes have been purified from human cells and from Drasaphila
(Kusch, et al., 2003). The human complexes include PCAF, STAGA and TFTC (Brand,
et al., 2001, Martinez, et al., 2001, Ogryzko, et al., 1998). The situation of these
complexes in higher eukaryotes is more complicated, indicating more cellular functions
carried out by the complexes. First, human has two GCN5 homologues, GCN5 and PCAF.
Human GCN5 and PCAF also have intimate association with CBP/p300 independently.
Secondly, two ADA2 proteins are found in human, mouse, Drasaphila, as well as plant
Arabidopsis (Kusch, et al., 2003, Muratoglu, et al., 2003, Stockinger, et al., 2001). In
Drasaphila, only one ADA2 is a component of SAGA complex. The other ADA2 variant,
ADA2a, can associate with GCN5 and ADA3 0(usch, et al., 2003). More intriguing, the
ADA2a protein in Drasaphila is part of GCN5-independent complexes, which are
concentrated at transcriptionally active regions in polytene chromosomes (Kusch, et al.,
2003). This suggests that the two ADA2s have distinct function in higher eukaryotes.
This is also revealed by the distinct effect of mutations in the two Arabidopsis ADA2
genes. ADA2b-null plants have severe phenotype whereas ADA2a-null plants seem not

to have any defects (V lachonasios, et al., 2003).

17

-. uh
.- “1:"
Qflfl§|k i.

“Blunt

2113:0511'

m
‘l “:3“
«11km.

The functional divergence among SAGA-like complexes is also reflected by the
observations that PCAF is dispensable for mouse development, while GCN5 null

mutation leads to death early in embryogenesis (Xu, et al., 2000, Yarnauchi, et al., 2000).

Histone methylation

The role of histone methylation in transcription regulation has emerged as an
important issue in the last few years. Like the situation of histone acetylation, this
explosion was triggered by the identification of histone methylation enzymes CARMI
(Chen, et al., 1999) and human Suv39h1 (Rea, et al., 2000). The availability of antibodies
that specifically recognize methylated lysine or arginine residues also has facilitated the
research.

Histone methylation plays important role in both transcriptional activation and
repression. Histone methylation is implicated in many biological processes such as
transcription, cell cycle regulation, development and even tumor genesis (Kouzarides,
2002, Lachner and Jenuwein, 2002, Zhang and Reinberg, 200]).

In contrast to other histone modifications, there are a few unique features to histone
methylation (Zhang and Reinberg, 200]). First, this covalent modification is known to
occur on lysine and arginine residues of histone H3 and H4 N-termini, including K4, K9,
K27, K36, K79, R2, R17 and R26 in H3, and K20 and R3 in H4. Secondly, the
modification can be mono-, di- or tri-methylation on lysine, and mono- or di-methylation
on arginine either symmetrically or asymmetrically. How these variant forms of
methylation differ in regulating chromatin function is not well known.

Thirdly, the modification is currently believed to be permanent. That is, no enzymes

18

iii-.1716 II]

‘
T‘J‘h

cam.”

25'} a
b L?

CAR

have been identified that can remove the methyl group from histones. This modification
is removed only after DNA replication. This permanency property might explain that
histone methylation is intimately associated with long-term epigenetic marks (J enuwein,
2001). However, recent clues indicate that transient methylation might be possible. Some
genes that are rapid turned on and off are controlled partially by promoter methylation,
such as the cyclin E gene and nuclear hormone receptor regulated gene p82 (Bannister, et
al., 2002). In addition, The N-terminal domain of yeast HAT enzyme Elp3 is suggested to
be a demethylase (Chinenov, 2002), implicating the existence of possible histone de-
methylation enzyme.

Fourthly, methylation of histones has dual effects on chromatin. It is connected with
both gene activation and repression (see more description below). This is in contrast with
histone acetylation that is generally associated with gene activation (F ischle, et al., 2003).

Finally, most histone methyltransferases (HMTs) function as single polypeptide to
methylate histones. The only known exception is the Enhancer of Zeste, a Polycomb-
group transcriptional repressor, which requires three other proteins to be enzymatically
active (Kuzrnichev, et al., 2002).

Two groups of HMT enzymes can methylate H3 and H4 tails (Zhang and Reinberg,
2001). Group one methylates arginine residues, including enzymes PRMT l, PRMT2, etc.
Group two methylates lysine residues.

In group one, the best-characterized are PRMT] and PRMT4 (also known as
CARMl) (Schurter, et al., 2001 ). PRMT] is a nuclear hormone receptor co-activator. It
specifically methylates histone H4 R3 both in vitra and in viva (Strahl, et al., 1999, Wang,

et al., 200]). PRMT] stimulates transcription from chromatin in Xenapus oocyte system

19

:aHill dept
moriplion ll
m ill 101
CAR)“ i
raptor 5i gna'
£1336 lSthur
med rec
l3 rebirth
more: “a
Scaring ll
n16 860
iii? a com
i'itulai hi
SET domai:
153:3 diSIii
SET domai
“3 Drosq
EaiSET

T .

”31%le

in a HMT dependent manner, directly illustrating the involvement of R3 methylation in
transcription (Wang, et al., 2001). PRMT] can be recruited directly by transcriptional
factor YY] to facilitate transcriptional activation (Rezai-Zadeh, et al., 2003).

CARM] is associated with GRIP, a p160 family co-activator in nuclear hormone
receptor signaling (Chevillard-Briet, et al., 2002). In vitra, CARMl methylates H3 R2, 17
and 26 (Schurter, et al., 2001). Direct in viva methylation of R17 by CARM] was
illustrated recently (Daujat, et al., 2002). Interestingly, this methylation depends on prior
H3 acetylation by CBP. In the activation of p82 gene, histone H3 R17 methylation at the
promoter was observed to coincide with the association of CARMl (Bauer, et al., 2002),
suggesting that CARM] methylation of H3 R17 plays roles in transcriptional activation.

The second group of HMTs methylates specifically lysine residues. These enzymes
share a common motif termed the SET domain. The specificity of these enzymes for
particular histone substrates is suggested to be dependent on the pre-SET or/and post-
SET domain (Rea, et al., 2000, Zhang and Reinberg, 2001). Kouzarides (Kouzarides,
2002) distinguished four families in this group based on the detailed sequences of the
SET domains. (i) The Suv39 family, including human Suv39hl and fission yeast C1r4
and Drasaphila Su(var)3-9, is able to methylate H3 K9. (ii) The SET] family, including
yeast SET], Polycomb proteins EZHl and EZH2 (containing SAN T domain), and
Trithorax protein MLL (containing bromodomain), is able to methylate H3 K4. (iii) The
SETZ family, including Drasaphila ASH] , is able to methylate H3. (iv)The RIZ family,
including RIZ, BLIMP and PFM] , does not contain pre-SET and post-SET domains.

The best characterized methylated lysine residues are histone H3 K4 and K9. K9

methylation is generally associated with transcriptional repression while K4 methylation

20

is genial} a

. lama
.ii the 1
iii. a prole
tun. reciii
HP: rise bit
imhmr
fIt homolo
3312111an
more Supt
icmhmr
P'Tiitess for

H3 R9
:5 Wm H
rmlli‘d 1C
85?ch ii]:
3m le It
EutlmSOm
tmchron

Sm are ol

is generally associated with transcriptional activation.

Mammalian HMT Suv39h1 specifically methylates H3 K9. The HP] protein may
mediate the repression of K9 methylation in heterochromatin (Jenuwein and Allis, 2001).
HP], a protein universally associated with heterochromatin (Eissenberg and Elgin, 2000),
binds specifically to H3 methylated at K9 (Bannister, et al., 2001, Lachner, et al., 200]).
HP] also binds with Suv39h], and null Suv39h1 leads to the disassociation of HP] from
heterochromatin (Lachner, et al., 2001). In fission yeast, crippled HMT activity of CLR4
(the homologue of mammalian Suv3 9) results in decreased HP] in heterochromatin. On
model (Jenuwein and Allis, 2001) proposes that initial K9 methylation brings HP] , and
more Svu39hl and HP] are recruited until this propagation creates locally repressive
heterochromatin. The chromodomain of HP] is probably play important role in this
process for it binds specifically to H3 tail with methylated K9 (Lachner, et al., 2001).

H3 K9 methylation is also evident in local repression of euchromatic genes. Suv3 9h]
is known to cooperate with Rb protein to repress E2F activity and Suv39h1 could be
recruited to E2F ] through its interaction with Rb (V andel, et al., 200]). Neilsen el al
showed that Rb, which is recruited by E2F] at the cyclin E promoter, recruits Suv39h]
and HP] to repress transcription of cyclin E, and that the local K9 methylation (only one
nucleosome) is Rb dependent (Nielsen, et al., 200] ). The involvement of HP] repression
in euchromatic genes probably is not restricted to Rb regulated promoters, for many HP]
sites are observed in Drasaphila euchromatin (Kouzarides, 2002). This suggests a wider
range of usage of HP] in gene repression.

In contrast to H3 K9 methylation, H3 K4 methylation is associated with

transcriptionally active chromatin (Lachner and Jenuwein, 2002) (Jenuwein and Allis,

21

LES E

m.-

“‘93 1
8‘51.er

(tr-s i
m. ,1.

2001). H3 K4 methylation is highly conserved among Tetrahymena, yeast and human,
and H3 K4 methylation in Tetrahymena is associated with active transcription (Strahl, et
al., 1999). At the mating-type locus (47 kb) in fission yeast, H3 K4 methylation is strictly
found in active euchromatin, while H3 K9 methylation is highly enriched in flanking
silent heterochromatin (Noma and Grewal, 2002). In another case of the chicken B-globin
locus and its flanking loci, H3 K9 methylation is associated with constitutive condensed
chromatin and inactive globin genes during erythropoiesis, while H4 K4 methylation
correlates with active chromatin which is hyper-acetylated (Litt, et al., 2001).

But this rule is not universal. H3 K4 methylation by Set] in budding yeast is required
for rDNA silencing (Briggs, et al., 2002, Bryk, et al., 2002). On the other hand, K9 and
K4 methylation are not exclusive from each other in chromatin. For instance, Ash] , the
epigenetic activator of the Trithorax group, methylates H3 K4 and K9, and H4 K20
simultaneously. Transcriptional activation by Ash] coincides with the methylation of
these three lysine residues at the promoter of Ash] target genes (Beisel, et al., 2002). This
reveals that combinatorial K methylations might have different meanings.

Recently, H3 K79 was also identified to be methylated by the enzyme Dot] (N g, et
al., 2002). Both the enzyme and K79 methylation are found in yeast and human cells.
Curiously, Dot] does not have a SET domain, and thus represents a distinct class of
lysine methylation enzymes. K79 methylation is important for telomere and mating-type
silencing in yeast, but K79 hyper-methylation is also a mark of euchromatin in both yeast
and mammalian cells (N g, et al., 2003), suggesting complicated roles played by K79
methylation. Interestingly, efficient K79 methylation is dependent on ubiquitinylation of

H2B K123 (Briggs, et al., 2002, Ng, et al., 2002).

22

ln summa
met; at 1’6

mated to l

HlSl'OllC ]
:imclcrlzcd
awesome
mall as for
ii. 1999 1.
Elm ionizin
Piifithoryla'
'Chtung. e1

Proteins
minalian
31.2001). [1
Willi HAT (
31'1ij Snfl

hZ'P‘Mlna

Hi

Al lea
2b

‘IQUllln\r]E

In summary, distinct patterns of methylation have different meaning. But how these
markers are read differently is still not clear. In addition, histone methylation is closely

connected to histone acetylation (see below).

Histone phosphoglation

Histone phosphorylation, compared to histone acetylation and methylation, is less
characterized (Cheung, et al., 2000). The H3 S10 phosphorylation is required for
chromosome condensation and segregation during mitosis and meiosis (Wei, et al., 1998),
as well as for transcriptional activation for certain genes (Lo, et al., 2000, Sassone-Corsi,
et al., 1999). The phosphorylation of H2A 8139 is connected with DNA break repair
upon ionizing radiation and apoptosis (Thomson, et al., 2001). The HZB 814 is also
phosphorylated during apoptosis and the kinase identified (Mstl) is caspase activated
(Cheung, et al., 2003).

Proteins Rsk and Msk have been identified as potential H3 810 kinases in
mammalian cells in response to mitogen stimuli (Sassone-Corsi, et al., 1999, Thomson, et
al., 2001). In yeast, Snfl was identified as an H3 810 kinase (Lo, et al., 2001). Together
with HAT GCN5, this kinase regulates gene 1N0] expression. The phosphorylation of
$10 by Snfl and acetylation of K14 by GCN5 on histone H3 at the [NO] promoter

happen in a sequential manner (Lo, et al., 2001).

Histone code and the cross talk of histone tail modification

At least five histone tail modifications are known, including acetylation, methylation,

ubiquitinylation, phosphorylation and ADP-ribosylation. There can be many covalent

23

a

attachment patterns on single histone tail, on single nucleosome, and on nucleosomal
clusters in a sequential and combinatorial manner. The histone code hypothesis (Strahl
and Allis, 2000, Turner, 2002) proposes that various modifications exert different
downstream functions by either recruiting various proteins or affecting chromatin
structure, thereby to execute various cellular fimctions such as transcriptional regulation
(Turner, 2002).

Based on the histone code concept, one specific modification may have different
functions, dependent on the modification context. One example is H3 S 1 0
phosphorylation (Strahl and Allis, 2000, Turner, 2002). This S10 phosphorylation plays
dual roles: it is required for chromatin condensation during mitosis as well as for certain
immediate early genes activation in response to mitogen stimulus. The coupled
acetylation of H3 K14 might make S ] 0 phosphorylation a gene activation marker
(Cheung, et al., 2000).

Another prediction is that various proteins or complexes bind to specific patterns of
modifications and carry out the specific downstream functions. There are a few known
proteins or domains that bind to particular covalent tags on histone tails. The
chromodomain of HP] binds to K9 methylated H3 tail as described above (Jacobs and
Khorasanizadeh, 2002). Bromodomains bind acetylated H3 and H4 tails (Dhalluin, et al.,
1999). Bromodomains are found in many transcriptional factors and complexes, such as
SAGA, TAF250, CBP/p300, SWI/SNF, and RSC (Jeanmougin, et al., 1997), as well as
some histone methylases (Kouzarides, 2002). One function of bromodomains in binding
to acetylated H3 and H4 tails might be to keep HAT complexes at local promoters in

order to maintain local hyperacetylation status. Hassen et al showed that in vitra retention

24

of SAGA and SWI/SNF on acetylated nucleosome template is dependent on their
bromodomain, and this is also true in viva at promoters targeted by SAGA and SWI/SNF
(Hassan, et al., 200]).

The double bromodomain in human TAFnZSO binds preferentially to H4 with a
specific acetylation pattern that is most frequently found in actively transcribed genes
(Jacobson, et al., 2000). Crystallographic information shows that the two binding pockets
of the tandem bromodomains prefer acetylated lysine residues that are seven amino acids
apart. In fact, these two bromodomains in TAF1|250 bind best to H4 tails with double
methylations of either K5-K12 or K8-K16. This case well exemplifies the histone code
idea that specific patterns of modification are recognized by specific proteins.

The histone code also predicts that one modification can inhibit or enhance the other
modification. For example, NuRD is a histone deacetylase complex associated with
transcription repression. Its preferential binding to the H3 tail is inhibited by K4
methylation but not by K9 methylation (N ishioka and Reinberg, 2001, Zegerman, et al.,
2002). HZB K123 ubiquitinylation in budding yeast is a prerequisite for H3 K4
methylation. H3 K4 methylation is abolished in the H2B K123R mutant (Dover, et al.,
2002, Nishioka and Reinberg, 200], Sun and Allis, 2002, Zegerman, et al., 2002). The
H2B K123R mutation perturbs silencing at the telomere, illustrating functional links
between Rad6-mediated H2B (K123) ubiquitinylation, SETl-mediated H3 (Lys 4)
methylation, and transcriptional silencing. HZB K123 ubiquitinylation is also required for
efficient methylation of H3 K79 (Briggs, et al., 2002, Ng, et al., 2002). Mutation of H2B
K123R also results in severe loss of H3 K79 methylation. Interestingly, these two sites

(H3 K79 and H2B K123) are proximate to each other in the nucleosome. Both K4 and

25

K79 methylation are involved in gene silencing. H2B ubiquitination might act as a switch
that controls the site-selective histone methylation patterns responsible for this silencing
(Briggs, et al., 2002).

Besides cross talk in trans as described above, there are also intra-tail
communications in histone modifications. H3 K14 acetylation by GCN5, PCAF and p300
is facilitated by prior S]0 phosphorylation (Cheung, et al., 2000, Lo, et al., 2000). The
crystal structure reveals close contact between the arginine residue (R164) near the yeast
Gcn5 reaction center and H3 S10. Indeed, the mutation of Gcn5 (R164A) resulted in
acetylation of H3 K14 independent of SIG-phosphorylation, and this mutation led to
decreased Gcn5 acetylation activity in target promoters (Lo, et al., 2000).

This meaningful cross talk between histone tail modifications in cis has also been
described in a few more cases. Upon epidermal growth factor (EGF) treatment in
mammalian cells, H3 810 phosphorylation precedes K14 acetylation and this di-
modification is tightly associated with target promoters after EGF treatment (Cheung, et
al., 2000). In yeast 1N0] promoter, as mentioned above Snf] phosphorylation at H3 810
and Gcn5 acetylation at H3 K14 occur in a strict sequential manner to enhance
transcription of [NO] gene (Lo, et al., 2001). Coupled H3 810 phosphorylation and H3
K14 acetylation is also seen in thyroid hormone receptor (TR) targets where reduction in
both 810 phosphorylation and K14 acetylation coincides with ligand-unbound TR status
whereas increase of both modifications coincides with ligand-bound state, indicating
these coupled-modifications are a mark of active gene (Li, et al., 2002).

Not only does H3 810 phosphorylation cross talk with K14 acetylation, but also with

lysine methylation. H3 810 phosphorylation inhibits H3 K9 methylation by Suv39h1 in

26

I'M/.3110 l'll

1513110:
med 1n
bra hem
.. 2002)
tabled pi
.9le11 im
athiiatior

H3 K4
ling. cl:
rile Hi i
fiscusscd i
‘s’fies whii

M} 15 We

vitra, and vice versa (Rea, et al., 2000).

Instances of cross-talk between histone acetylation and methylation have also been
reported. In vitra, H3 K9 and K14 acetylation improves H3 K4 methylation by MLL, the
human homolog of Drasaphila trithorax which is a SET-domain-containing HMT (Milne,
et al., 2002). This linked modification might also occur in viva, since histones at MLL
regulated promoters have these modifications. In another case, H4 R3 methylation by
PRMT] improves H4 K8 and K12 acetylation by p300, but acetylated H4 inhibits R3
methylation by PRMT] (Wang, et al., 200] ).

H3 K4 methylation and K9 methylation were also reported to be mutually inhibitory
(Wang, et al., 2001). Methylated H3 at K4 improves subsequent p300 acetylation on H3,
while H3 K9 methylation inhibits H3 K14, K18 and K23 acetylation by p300. As
discussed in histone methylation section, H3 K4 methylation is associated with active
genes while H3 K9 methylation with repressive chromatin, and histone acetylation by
p300 is well documented with gene activation (Sterner and Berger, 2000), indicating that
H3 K4 and K9 methylation have very different functions.

Moreover, the cross-talk in cis between histone acetylation and arginine methylation
was also reported recently (Daujat, et al., 2002). After estrogen treatment, p300 was
observed at the pS2 promoter and H3 K18 was acetylated, followed by H3 K23
acetylation. CARM] was then recruited, leading to the methylation of R17. This
sequential acetylation and methylation was abolished if CBP HAT activity was crippled,
clearly demonstrating that H3 K18 and K23 acetylation is prerequisite for R17
methylation.

Histone modifications can also crosstalk to pol 11. Two groups report that H3 K36

27

methylation was dependent on pa] 1] CTD phosphorylation in yeast (Li, et al., 2002, Xiao,
et al., 2003). Partial deletion or CTD kinase knock-out resulted in total abolition of K36
methylation, whereas H3 K4 and K79 methylation is not affected.

In summary, the covalent tags on chromatin histones establish various patterns of
chromatin modification that may serve as marks for specific downstream functions in
regulating genes’ expression. These patterns can be dynamic depends on multiple cellular
stimuli and the cross talks between these modifications. How these specific modifications

are interpreted in detail is the subject of intense study currently.

The plant world

Most knowledge about the regulation of transcription activation has been learned
from the studies of yeast and metazoans. Relatively less has been known in the plant
kingdom.

The genome sequence of Arabidopsis thaliana [the Arabidopsis initiative, Nature
408, 796 - 815 (2000)] reveals that the genome encodes nuclear DNA-dependent RNA
polymerase I, II and III that are typical of eukaryotes. This suggests that plants use the
common mechanisms of gene transcription.

Plants, like yeast and metazoans, also have various chromatin remodeling
modification factors (Li, et al., 2002) and some of these factors have been found to play
crucial roles in plant gene expression (V erbsky and Richards, 2001), such as Sufi-type
chromatin remodeling ATPases, histone acetyltransferase that include the GNAT family,
the MYST family, and TAF 250 or CBP-like proteins, and histone deacetylases that

include de3 or Sir2-like proteins. Arabidopsis CBP-like protein PCAT2 has been shown

28

. . ..
m. "Ti
. 9...}
flag!"
“I 1.. L
"A‘POJ
«Lille
”193*

.s>A~L‘

1, Wg-r
53““;

to be able to acetylate histones in vitra although it lacks the bromodomain (Bordoli, et al.,
2001). Arabidopsis Snf5-like protein BSH can complement partially the snf5 mutant in
budding yeast (Brzeski, et al., 1999), and the Swi2-like protein MOM has been found
important in gene silencing (Amedeo, et al., 2000). Meanwhile, limited studies available
todate indicate that chromatin modification plays important roles in regulating gene
expression in plants. Hyperacetylation of promoter histones of the pea plastocyanin gene
correlated directly with gene activation (Chua, et al., 200] ). All these findings indicate
that plants make use of similar chromatin remodeling mechanisms in regulating gene
expression.

But plants, due to their vital difference in life cycle fiom animals, might have
evolved plant-specific mechanisms to regulate gene transcription (Lusser, 2002). For
instance, Arabidopsis lacks most transcriptional factors for polymerase I, and it has a
fourth class of the largest subunit and the second-largest subunit for nuclear RNA
polymerase which is not found in other eukaryotes [the Arabidopsis initiative]. Moreover,
the Arabidopsis genome encodes about 1500 transcriptional factors whereas Drasaphila
melanagaster and C. elegans have about 650 (Riechmann, et al., 2000). Among these
transcriptional factors, at least five families are not found in yeast and metazoans
(Riechmann and Ratcliffe, 2000), such as the AP2/EREBP family (~l40 members)
implicated in flower development, cell proliferation, stress response, ABA and ethylene
response, the NAC (~100 members) family implicated in pattern formation and organ
development, and the WRKY family (~70 members) implicated in defense response.

In addition, owing to its sessile nature, plants are affected greatly by both biotic and

abiotic environmental signals, such as drought, high salt, cold temperature and pathogen

29

attacks. These aboitic and biotic stresses are the major losses in crop productivity
worldwide (Singh, et al., 2002). Transcriptional regulations in plants in response to these
stressful conditions take part in vital roles in plants’ life (Singlr, et al., 2002). This makes
the studies of transcription regulation of plants in response to stressful condition very
significant.

One example is the transcriptional regulation in plants in response to cold
temperature. In Arabidopsis, many cold regulated (COR) genes are transcriptionally
activated that enable the plant to accommodate cold temperature (Thomashow, 1998).
One key transcriptional activator of COR genes CBF] , belongs to the AP2/EREBP
family of to plant transcriptional factors (Stockinger, et al., 1997). Constitutive
expression of CBF 1 induced COR gene expression and increased the freezing tolerance
of non-acclimated Arabidopsis plants (Jaglo-Ottosen, et al., 1998). Very intriguingly, the
transcriptional activity of CBF l was found to depend on the Ada2, Ada3 and Gcn5
proteins in yeast (Stockinger, et al., 2001).

This initial finding of CBF ] dependence on yeast co-activator proteins led to the
hypothesis that Arabidopsis has similar Ada2 and Gcn5 proteins and that these proteins
might modify chromatin in facilitating plant COR genes’ activation. In addition, the
limited knowledge about plant histone acetyltransferases, the indicated importance of
transcriptional regulation in plant life, as well as the indications of the divergence of
transcriptional activation in plants altogether led to this thesis study ofArabidapsis

ADA2 and GCN5 homologous proteins.

30

W

The GCN5 Gene and protein of Arabidopsis

Introduction

The Gcn5 protein was first identified genetically from budding yeast. Under the
growth condition of amino acid starvation, yeast genes related to amino acid synthesis are
turned on by activator Gcn4. The efficient activation activity of Gcn4 required protein
Gcn5 (Georgakopoulos, et al., 1995). Gcn5 was also required by chimerical activator
Gal4-VP16 for efficient activity in yeast (Marcus, et al., 1994). Mutation in GCN 5
reduced the transcriptional activity of Gcn4 and Gal4-VP16, suggesting that Gcn5 might
function in facilitating transcription ad co-activator protein. Meanwhile, Gcn5 was found
to function together with protein Ada2 and Ada3 as a unit. Budding yeast with mutation
in GCN5 showed similar phenotypes as the mutation in ADA2 or ADA3 as well as
double mutations in ADA2 and GCN5, and each mutation could relieve the toxicity
caused by the over-expression of Gal4-VP16 (Berger, et al., 1992, Georgakopoulos, et al.,
1995)

The mechanism of yeast Gcn5 in facilitating transcription remained elusive until the
milestone discovery in 1996 that the yeast Gcn5 was found to be a histone
acetyltransferase (HAT) that acetylated histone H3 (Brownell, etal., 1996). Chromatin
acetylation had been observed to connect with gene activation (Turner, 1991). This
finding shed light on the mechanism of Gcn5 in facilitating transcription by modifying
chromatin. The GCN5 protein is conserved in yeast and metazoan with two conserved

domains, the catalytic HAT domain and the bromodomain (Xu, et al., 1998).

31

But very little was known about GCN5 proteins in plant world. A cDNA from plant
A rabidapsis thaliana was previously cloned at our laboratory that encodes a protein
homologous to GCN5 with a conserved HAT domain and bromodomain, as well as a
unique N-terminus (200 amino acids). Recombinant GCN5 protein from Tetrahymena,
yeast and human can acetylate free histone H3 in vitra (Brownell, et al., 1996, Xu, et al.,
1998, Yang, et al., 1996). It was fundamental for us to determine whether A rabidapsis
GCN5 possessed histone acetyltransferase activity in vitra.

Research work was also performed to determine the genomic sequence of
Arabidopsis GCN5 gene. It was not clear whether the known N-terminal extension of
Arabidopsis GCN 5 was complete for mammalian GCN5 homologues have longer N-
terminus extension (about 400 amino acids), although the predicted N-terminal extension
of Arabidopsis GCN5 shared no obvious sequence similarity to that of mammalian GCN5
homologues. The GCN 5 genomic sequence could also reveal the boundaries of exons and
introns of GCN5 gene that would help in the firture plan in our lab of screening

Arabidopsis lines with T-DNA insertion in the GCN5 gene.

Exmrimental Methods

Determining the genomic sguence of GCN5 gene
A genomic DNA library of Arabidopsis thaliana was provided kindly by Dr.

Christopher Benning of our department, in which the genomic DNA was digested
partially with Hind III and subcloned into cosmic vector pBIC20. The library had an
average insert size of 20 kB. The cosmids had been already introduced into E. cali strain

NM554.

32

The library was plated on LB medium with 10 ug/ml tetracycline at a density of
about 100 colonies per plate (100 x 15 mm). The colonies were transferred to
nitrocellulose. Standard colony hybridization procedures were followed using the GCN5
cDNA (full length) as probe. Cosmid DNA prepared from the positive colonies was
screened again after digested with Hind III by standard Southern blotting procedure using
GCN5 (full length) cDNA probe. Two positive DNA fragments were identified and sub-
cloned to vector pBS (KS+). The two fragments were DNA sequenced by the sequencing
center at Michigan State University using various primers that ensured that every DNA

region was read in both 5’ and 3’ directions.

Arabidopsis suspension cell usate
Arabidopsis leaf suspension cell culture was originally provided by Dr. Natasha

Raikhel’s laboratory from Plant Research Laboratory at Michigan State University. The
cell culture was maintained in Garnborg’s B-5 (GibcoBRL) liquid medium with 1.8
rig/ml 2, 4-dichloro-phenoxyacetic acid. The suspension cells were passaged every 10
days in 50 ml medium at 25°C and 24-hour light cycle.

To make cell lysates, the cells were resuspended in Tris buffer (50 mM Tris pH8.0, 1
mM EDTA, 100 mM KC], 10% glycerol, 2 mM MgC12, 5 mM DTT) with protease
inhibitor cocktail (Roche) and lmM PMSF. The cells were treated with French Press
once. The broken cells were either used directly as “crude lysate”, or centrifuged at 7,600

x g for 15 min, and the supernatant was collected as “supernatant cell lysate”.

Western blotting

33

Two rabbit antisera were used. Antiserum #84-6 was raised with GCN5 peptide
antigen (amino acids 383-400) and the antiserum was collected after boosting for the
sixth time. Antiserum #91-3 was raised with recombinant GCN5 antigen (amino acids
155-568) and the antiserum was collected after boosting for the third time.

In] cell lysate or supernatant cell lysate was resolved in 7.5% SDS-PAGE gel (20
mA current for 3 hours), and then transferred to nitrocellulose membrane. The membrane
was incubated with antiserum #84-6 (1 :10000 dilutions) or #91-3 (1 :10000 dilutions).
Then the membrane was incubated with goat against rabbit second antibody (1:4000
dilutions, BioRad company, affinity purified and HRP conjugated). The signal was

detected using Lumi-light Western Blotting Substrate kit (Roche).

Purification of recombinant GCN5

For purifying His6-GCN5 (155-568), cDNA fragments encoding GCN5 (155—568)
was subcloned into His6-tagged vector pET-28c (N ovagen), and the constructed plasmid
was transformed into E. cali strain BL2](DE3). 10 ml overnight culture of transformed E.
cali was transferred to 500 m] LB medium with 50 ug/ml kanamycin, and the culture was
induced with 90 uM IPTG for 16 hours at room temperature. The cells were collected by
centrifugation and washed once with Na-PO4 buffer (50 mM Sodium phosphate, pH7.8,
500 mM NaCl, 10% glycerol and 5 mM 2-mercaptoethanol).

The cell pellet was resuspended in 15 ml Na-PO4 buffer with protease inhibitor
cocktail and 12 mM irnidazole, and treated with French Press once to make cell lysate.
The supernatant was collected from the lysate after centrifugation at 20,000 x g for 20

minutes. Appropriate amount of lysate supernatant was mixed with Ni-NTA beads

34

(Qiagen) at cold room for 1 hour, and then the beads was washed with Na-PO4 buffer
with 0.1% NP-40 for several times. The beads were washed again with Na-PO4 buffer
with 40 mM imidazole once. The His6-GCN5 (155-568) was eluted with Na-PO4 buffer
with 250 mM imidazole.

For purification of recombinant GCN5 (1-568, 205-568, or 205-384), DNA
fragments encoding the respective GCN5 proteins were subcloned into GST-tagged
vector pGEX-6p (Pharmacia) and transformed into E. cali strain BL2](DE3) Codon Plus.
10 ml overnight culture of transformed E. coli was transferred to 1000 ml LB culture, and
the culture was induced with 70 uM IPTG for about 14 hours at room temperature. The
cells were collected by centrifugation and washed once with PBS buffer (140 mM NaCl,
2.7 mM KCl, pH7.3, 10 mM Na2P04, 1.8 mM KH2P04, 10% glycerol, and 7 mM 2-
mercaptoethanol).

The cell pellet was resuspended in about 20 ml PBS buffer with protease inhibitor
cocktail and 1 mM EDTA, and treated with French Press once to make lysate. The
supernatant was collected fiom the lysate afier centrifugation at 20,000 x g for 20
minutes. Appropriate amount of lysate supernatant was mixed with GST Glutathione
Sepharose beads at cold room for 1 hour, and then the beads were washed with PBS
buffer for several times. To recover GCN5 proteins from the beads, the beads were cut
with PreScision (Pharmacia) protease at cold room for 4 hours and the supernatant was

collected.

I-IAT assay

The HAT assays were carried out in 30 u] reactions containing 50 mM Tris-HCI pH

35

8.0, 10% glycerol, 1 mM DTT, 1 mM PMSF, 0.] mM EDTA, 50 mM NaCl, 10 mM
butyric acid, 20 pg calf thymus histones (Sigma), 0.15 or 0.2 uCi [3H]acetyl-C0A (NEN
Life Science Products, 0.] uCi/ul) and about 2.9 pg GCN5(155-568), or 250 ng GCN5(1-
568), or 70 ng GCN5(205-568), or 70 ng GCN5(205-384) or 2.8 ug purified recombinant
human TAF31 (as a negative control). The reaction mixture was incubated at 27 or 30°C
for 30 min or 40 min, after which the reactions were terminated by adding 10 or 20 u] of
SDS—PAGE loading buffer. 20 u] from each terminated reaction mixture was
electrophoresed on 15% or 18% SDS—polyacrylamide gels. Gels were fixed (40%
methanol, 10% acetic acid solution), soaked in autoradiographic enhancer (NEN Life

Science Products), vacuum dried and exposed to X-ray film.

Results

The Arabidopsis GCN5 protein is a HAT

The predicted amino acid sequence of Arabidopsis GCN5 protein based on the
cDNA sequence (Stockinger, et al., 2001) suggests that it is a HAT enzyme. An in vitra
HAT assay was carried out to test directly whether the Arabidopsis CGN5 protein is a
HAT.

Recombinant GCN5 proteins as either full length or truncated versions were purified
fiom expressed E. cali cells. As shown in Figure 1B, relatively pure recombinant proteins
of GCN5 were obtained, including GCN5 (1-568, full length), GCN5 (155-568) that still
contained part of the unique N-terminus, GCN5 (203-568) that contained both the HAT
domain and the bromodomain, and GCN5 (203-3 84) that contained the HAT domain

only.

36

 

 

 

 

  

1 2 3 4
1 203 384 458 568 kDa
HAT Br as _ -‘
1-568
L 44 -
155-568 “ ‘
-
203-568
28 -
203-384 '
my -<
19 -
C GCN5
(155663) TAF31 1-568 205-568 203-384
m -- Q ~-
H213 '
HZA m ma 43%-
H4

Figure l. Arabidopsis GCN5 protein is a histone acetyltransferase.

(A) Diagram of the GCN5 protein. The predicted HAT domain and bromodomain (Br)
are indicated. The bars show the GCN5 proteins in various lengths that were purified as
recombinant for HAT assays. (B) The purified recombinant proteins of GCN5 in various
lengths. The recombinant proteins were electrophoresed in 10% SDS-PAGE gel and were
either Coomassie Blue stained (lane 2, 3, and 4) or silver stained (lane 1). The respective
GCN5 proteins were indicated by the arrows. l: His6-GCN5 (155-568), 2: GCN5 (1-
568), 3: GCN5 (203-568), 4: GCN5 (203-384). Note that the protein mark applies only
to lane 1 and that the unindicated bands in lane 1 were contaminated proteins. (C) The
results of the HAT assays with recombinant GCN5 proteins in various lengths. The HAT
reactions were electrophoresed in SDS-PAGE gel. The histones in gel were first
visualized by Coommasie Blue staining as represented on the left panel, then the
acetylation signals were detected by enhanced fluorography as shown on the right two
panels. Note that the histones between H4 and H2A on the left panel were also histone
H4.

37

The HAT assays were then performed using the four recombinant GCN5 proteins,
free core histones and [3 H] acetyl CoA. The HAT assay results (Figure 1C) showed that
purified recombinant GCN5 proteins were able to acetylate free core histons, and core
histone H3 was the major acetylated substrate. Moreover, the HAT domain alone was

sufficient to acetylate core histone H3.

The genomic spguence of GCN5 gene

An Arabidopsis genomic library (provided by Dr. Christopher Benning) was
screened with GCN5 cDNA probe to identify the GCN5 gene. Seven positive library
colonies were identified. To identify fragments specific to the GCN5 gene, these cosmid
DNAs from these colonies were digested with Hind III and subject to Southern blotting
with the GCN5 full-length cDNA probe. Two common fragments were identified as
GCN5 gene-specific. These two fragments were subcloned into vector pBS (KS+). The
two fragments were then DNA sequenced using serial specific primers. The GCN5 gene
sequence was then assembled using the computer program Sequencher. The genomic
sequence of GCN5 gene was deposit in the GenBank (accession number gi: 13591701).

By matching the genomic sequence to that of the GCN5 cDNA, the boundaries of
exons and introns were determined. Detailed map information and the location of specific
primers used for sequencing are shown in Figure 2. The result revealed no apparent open
reading We at the 5’ end besides the previous determined cDNA sequence of GCN5.
But interestingly, the N-terminus of GCN5 (amino acids 1-211) protein is encoded by a

single exon.

The endogenous GCN5 protein

38

Figure 2. Diagram shows GCN5 gene and its encoded protein GCN5.

(A) Diagram of GCN5 gene. The lightly shaded areas in upper panel represent exons
and the open areas represent introns, with boundaries indicated by the sequence
numbers. Various primers and the sequences read by them are shown at the top where
the arrows indicate the relative sequence length and direction obtained using each
primer. Three Hind III restriction enzyme sites defined the two genomic DNA
fragments identified by Southern blotting of cosmid DNA. The lower panel indicates
the encoded GCN5 protein corresponding to the exons in the upper panel.

(B) Schematic alignment of Arabidopsis GCN5 against its homologues fi'om other
species. The HAT domain and the bromodomain (Br) are indicated. The unique N-
terminus of Arabidopsis GCN5 is indicated by blue box, while the N-termini of
mammalian and Drasaphila homologues are indicated by the unfilled box.

39

mzuc siesta
mzoc 33:
mZUmV ESE—u:
36m 8.52
:0h 52:5:
mzee 5.5.:
mzuc 38>
mzoe 33:35.4

 

 

 

awn wmv 5n 3N

35a:— eeEEm _ 52.5: .55 l _
...—na-;..-_#:-.m-+:mluia..-
w .l _

33. n3... 2:. 8% 33.. an». 3.2 35 E: 22 32 EE 32 NS _

40

 

 

le N: 35:
-~ as: $2.3 Ni 35%
AIEEWII . <

Both the genomic sequence and the cDNA sequence revealed that Arabidopsis
GCN5 protein comprises 568 amino acids in full length. The calculated molecular weight
of GCN5 protein is 63 kDa, and the apparent molecular weight of recombinant GCN5
protein is about 75 kDa in SDS-PAGE gel (Figure 3A).

To check the expression of endogenous GCN5 protein, cell lysates from Arabidopsis
cell suspension culture were analyzed by Western blotting. The antibodies used were
raised against either purified recombinant GCN5 antigen (amino acids 155-568) or
peptide antigen (amino acids 383-400) which was right after the HAT domain. The
peptide antigen sequence is not inside the conserved HAT domain and is specific to
Arabidopsis GCN 5. Therefore the antiserum against this peptide antigen was more
specific than the one raised against recombinant GCN5 antigen which contained both the
conserved HAT domain and bromodomain.

The Western blotting detected a predominant protein by both antibodies with
approximate molecular weight of 75 kDa (Figure 3B). This size is consistent with the
migration of recombinant full-length GCN5 protein expressed from the cloned cDNA.

Curiously, the antiserum against recombinant GCN5 antigen also detected a larger
protein that is unlikely a longer version of GCN5 for it was not detected by the more
specific peptide antiserum. It might be a protein that also contains the conserved HAT
domain or bromodomain.

Interestingly, a protein, with a little bit smaller size, was weakly detected by peptide
antiserum but not by antiserum against recombinant GCN5 antigen ( aa 155-568) (Figure

3B).

41

83
' e. O
62 g g a W
' I
47.5
r i
B Antiserum against Antiserum against

GCN5 (—a__a 155-568) GCN5 (aa 383-400)

 

Figure 3. Detection of endogenous GCN5 protein.

(A) Recombinant GCN5 protein in 10% SDS-PAGE gel. All the proteins were
Coomassie Blue stained. 1: 0.5 pg BSA, 2: lpg BSA, 3: recombinant GCN5
protein. (B) The endogenous GCN5 protein was detected by Western blotting.
Crude lysate of Arabidopsis leaf suspension cells (lane 4 and 6) and the
supernatant of the crude lysate (lane 5 and 7) were resolved in 7.5% SDS-PAGE
gel. The proteins in gel were then transferred to nitrocellulose membrane and
detected by antiserum against either recombinant GCN5 antigen (aa 155-568) or
GCN 5 peptide antigen (aa 383-400). The detected GCN5 protein is indicated by
the arrows. There are two close bands in lane 6 and 7 as indicated by the black
and gray arrows. An unknown protein cross reacted with antiserum against
recombinant GCN5 antigen, as indicated by the asterisk marker in lane 4 and 5.

42

Discussion

The results described in this chapter indicate that the predicted GCN5 homologue of
Arabidopsis cDNA has histone acetyltransferase activity. This GCN 5 protein
preferentially acetylates histone H3. The predicated HAT domain of Arabidopsis GCN 5
is sufficient for this activity. In addition, the genomic locus of Arabidopsis GCN5 was
cloned and sequenced. The GCN5 gene contains thirteen exons and 12 exons.

Recombinant Arabidopsis GCN5 acetylates mainly core histone H3 and weakly HZB
and H4 (Figure 1), a shared enzymatic activity characteristic of recombinant protein of
GCN 5 and PCAF fiom other species when core histone mixture is provided as the
substrate. Which lysine residue on histone H3 is acetylated by Arabidopsis GCN 5 was
not determined. Recombinant yeast Gcn5 and human PCAF acetylate mainly lysine
residue 14 (K14) on histone H3 (Kuo, et al., 1996, Schiltz, et al., 1999). It is reasonable
to infer that recombinant Arabidapsis GCN 5 protein also acetylates mainly H3 K14 in
vitra, based on the facts that (i) the highly sequence conservation in the Arabidopsis
GCN5 HAT domain (60% identity and 72% similarity to the yeast Gcn5 HAT domain),
(ii) the conservation of the A, B and D motifs but not the C motif in the Arabidopsis HAT
domain, like that of other species’ GCN5 and PCAF protein (Neuwald and Landsman,
1997), (iii) the sufficiency of the Arabidopsis GCN5 HAT domain for enzymatic activity
(Figure 3B).

The sufficiency of the HAT domain of Arabidopsis GCN 5 to acetylate histones
suggests that the unique N-terminus, the conserved bromodomain and the region between
the bromodomain and the HAT domain (a 384-458, designated as H-B) are not

important for GCN5 HAT activity in virra. The H-B regions in GCN5 and PCAF from

43

different species also show sequence conservation. The Arabidopsis H-B has 50%
similarity to that of yeast Gcn5. However, the function of H-B is not known.

The bromodomain was also not necessary for yeast Gcn5 HAT activity in vitra
(Candau, et al., 1997). However, efficient acetylation of nucleosomal substrate by the
yeast SAGA in vitra requires the Gcn5 bromodomain (Sterner, et al., 2002). Similar
situation was also observed for HAT protein CBP. CBP does not belong to the GNAT
HAT family where GCN5 proteins belong (Deng, et al., 2003). In viva, the bromodomain
was required for the transcriptional activation of some genes in yeast (Georgakopoulos, et
al., 1995, Sterner, et al., 2002). These findings reveal that, although there is a general
dispensability of this bromodomain in vitra for GCN5 to acetylate free core histones, it is
important for GCN5 function in viva.

Evidence suggests that the bromodomain is more involved in communicating with
acetylated histone H3 and H4 tails. Yeast Gcn5 bromodomain bound to acetylated H3
and H4 tail (Omaghi, et al., 1999), and NMR structural research revealed that the human
PCAF bromodomain formed a hydrophobic pocket to which acetylated lysine residue
from H3 or H4 tail bound (Dhalluin, et al., 1999). More recently, bromodomains within
the catalytic subunits of SAGA and SWI/SNF has been shown to anchor these complexes
to acetylated promoter nucleosomes (Hassan, et al., 2001). The bromodomain of human
PCAF is also bind to acetylated lysine residue of the HIV Tat protein, and this interaction
synergizes Tat and PCAF action on transcriptional activation of HIV promoter (Dorr, et
al., 2002). This binding to non-histone protein of bromodomain proposes that the general
role of bromodomain is to recognize acetylated lysine residues, in analogue to the SH2

domain in signal transduction pathways (Horn and Peterson, 200]).

44

Our genomic sequence of GCN5 gene is identical to that released more than one year
later by the “the Arabidopsis initiative” [the Arabidopsis initiative, Nature 408, 796-815
2000] which detemrined the whole genomic DNA sequence of Arabidopsis. There are
twelve introns in the GCN5 gene. Interestingly, there is no intron in the first exon which
encodes the first 21 1 amino acids that encompass the whole unique N-terminus.

The peptide antiserum detected two closely positioned proteins in the Western
blotting of endogenous GCN5 proteins (Figure 33). It is not known whether the smaller-
sized one is a shorter version of GCN5, or the bigger-sized one is a covalently modified

version of GCN5, or one of detected proteins is not GCN5 protein.

45

Chapter Three

The Interaction between GCN5 and ADA2,

Introduction

In budding yeast, the GCN5 protein is an integral component of at least two distinct
complexes, termed ADA and SAGA, that acetylate nucleosomal histone H3 and H2B
(Eberharter, et al., 1999, Grant, et al., 1997). The 1.8 MDa SAGA contains Ada adapter
(or co-activator) proteins (Adal , Ada2, Ada3, and Gcn5), Spt proteins (Spt3, Spt7, Spt8,
and Spt20/AdaS), TBP associated proteins (TAF 11) pr0teins (TAFS/yTAFn90,
TAF6/yTAFn60, TAF9/T AFn17/20, TAF 10/yTAF1125, and TAF 12/yTAF 1161/68), and
the Tral protein that is related to the members of the ATM/DNA-
PK/phosphatidylinosito] 3-kinase family (Abraham, 2001, Khanna, et al., 200], Vassilev,
et al., 1998) . The smaller ADA complex (approximately 0.8 MDa) contains the Ada2,
Ada3, Gcn5 and Abel proteins. SAGA-like complexes are conserved in yeast and
metazoan organisms. At least three distinct SAGA like complexes have been identified
from human cells, including PCAF (Ogryzko, et al., 1998), TFTC (Brand, et al., 200]),
and STAGA (Martinez, et al., 200]).

The yeast Ada2 and Gcn5 proteins can interact with each other (Candau and Berger,
1996, Candau, et al., 1996, Candau, et al., 1997). Biochemically purified ADA and
SAGA complexes contain both ADA2 and GCN 5. ADA2 is lost from these two
complexes in yeast lacking Gcn5 protein (Grant, et al., 1997).

At present, no direct biochemical evidence indicates whether Arabidopsis has similar
GCN5-containing HAT complex (es). Chapter II shows that Arabidopsis GCN5 is a

histone acetyltransferase (HAT). Previously, our lab cloned cDNAs from Arabidopsis

46

encoding proteins homologous to ADA2 protein. Arabidopsis has two ADA2
homologues, namely ADA2a and ADA2b. We hypothesized these two ADA2 proteins
interact with GCN 5 based on the fact that yeast Ada2 interacts with Gcn5. The
experiments described in this chapter test the physical interaction of the A rabidapsis

GCN5 protein with the ADA2a and ADA2b proteins.

Exmrimenta] Methods

GST pull-down assay

For constructing plasmids expressing GCN5 GST-firsion proteins, the cDNAs
encoding various fragments of GCN5 protein (amino acids 1—568, 1—150 or 155—568)
were suncloned into GST vector pGEX (Pharmacia). GST-fusion proteins were expressed
in E. cali strains XA90 or BL2](DE3) Codon Plus following induction with 40—60 pg/ml
IPTG at 20°C for 20 h. The E. coli cells were then lysed by French press in Tris buffer
(20 mM Tris pH 7.9, 150 mM NaCl, ] mM EDTA and 10% glycerol) with protease
inhibitor cocktail (Roche) and 1 pg/ml pepstatin. Triton X-100 and DTT were added to
1% (v/v) and 10 mM final concentration, respectively, and the lysates were rocked at 4°C
for 40 min. The lysates were centrifuged at 20,000 x g for 15 nrin and the supematants
were collected. To check the expression of GST-fusion protein, 1 ml of the supernatant
was shaken together with 60 pl glutathione—Sepharose beads (Pharmacia) for 90 min at
4°C. The beads were then washed eight times with 1 ml Tris buffer containing 10 mM 13-
mercaptoethanol, 0.2% Triton X-100, ] pg/ml pepstatin, 50 pg/ml PMSF, 1 pg/ml
leupeptin and 2 pg/ml aprotinin. The bead-bound protein was eluted with 2% SDS and

visualized by Coomassie Brilliant Blue staining after fractionation by SDS—PAGE.

47

To express the Arabidopsis ADAZa/ADAZb proteins in vitra, the corresponding
cDNAs were cloned into His6-tagged vector pET-28 (Novagen). The ADA23/ADA2b
proteins were then expressed using CsCl gradient purified plasmids templates and the
TNT T7 transcription/translation system (Promega). [3 5 S] methionine was included in the

system to radiolabel the expressed proteins.

Depending on the specific experiment, GST or GST fusion protein (about 1-2 pg)
bound to glutathione—Sepharose beads was incubated with 4 pl of radiolabeled ADA2a or
ADA2b translation mixture for 1 h at 25°C in 200 pl of binding buffer (40 mM HEPES
pH 7.6, 100 mM NaAc, 5 mM MgC12, 1 mM EDTA, 10% glycerol, 0.2% Triton X-100, 1
pg/ml pepstatin, 50 pg/ml PMSF, 1 pg/ml leupeptin and 2 pg/ml aprotinin) plus 8%
bovine serum. After washing six times with 800 pl of binding buffer (with 300 or 400
mM NaCl) at 4°C or room temperature, the bound proteins were eluted with 2% SDS and

detected by autoradiography after SDS-PAGE.

Yeast two-hybrid assay

GAL4 System 3 (Clontech) was used for the yeast two-hybrid assays, including prey
vector pGADT7 that expresses the Gal4 activation domain (GAD) and has Leu2 and
Ampr markers, and bait vector pGBKT7 that expresses Gal4 DNA binding domain
(GDBD) and has Trp] and Kanr markers. cDNAs encoding the firll length or different
parts of GCN5 were inserted into bait vector pGBKT7 in frame with that of GDBD.
cDNAs encoding the full length or different parts of ADA2a or ADA2b were inserted

into prey vector pGADT7 in fiame with that GAD. All the cDNA encoding different

48

parts of ADA2a, ADA2b and GCN5 were made by PCR from full length cDNA
templates, and the DNA sequences were checked to make sure that there were no

mutations introduced during PCR reaction.

The constructed bait plasmids were prefixed by “pGDBD-”. For instance, pGDBD-
GCN5(fl) meant bait plasmid with the cDNA encoding GCN 5 in full length. The
constructed prey plasmids were “pGAD-”. For instance, pGAD-ADA2b meant prey
plasmid with cDNA encoding ADA2b protein.

To test protein-protein interaction, the transformed yeast cells (strain AH109) with
respective bait and prey plasmids were first grown on solid synthetic dropout (SD)
medium lacking leucine and tryptophan (Leu- Trp-). Colonies from SD (Leu- Trp-)
medium were then streaked on SD medium (Ade- His- Leu- Trp-) containing X-or-gal for
testing protein-protein interaction, as well as SD medium (Leu- Trp- ) to make sure the
introduced ADA2 and GCN 5 proteins were not toxic to yeast cells. The yeast colonies
were allowed to grow for three to four days at 30° C before the pictures were taken.
Three selective markers were used simultaneously as criteria to judge the positive
interaction between prey and bait proteins, the expression of ADE2 and HIS3 genes, as
well as the expression of MEL] gene whose protein product was secreted to medium and

catalyzed substrate x-a— gal into blue color product.

Results
ADA2 interaction with GCN5
The initial approach to investigate possible physical interactions between GCN5 and

ADAZa or ADA2b was to use a GST pull-down assay. In this assay, GCN5 proteins

49

(amino acids 1-568, ]-1 50, and 155-568) were expressed in E. cali as GST-fusions.
Approximately the same amount of each fusion protein was purified from E. cali and
immobilized on glutathione-Sepharose beads. The ADA2a and ADA2b proteins were
expressed and [353] methionine labeled in vitra using the TNT T7 transcription/translation
system (Promega). The GCN5 fusion proteins bound to beads were incubated with the
ADA2a and ADA2b proteins. After washing to remove non-specific proteins, the ADA2a
and ADA2b proteins bound to the beads were eluted with SDS-PAGE loading buffer and
detected by autoradiography.

As shown in Figure 1, both the full length GCN5 (aa 1-568) and the N-terrninally
truncated version of GCN5 (aa 155-568) pulled down both ADAZa and ADA2b.
Comparable amounts of the N-terminal protein of GCN5 (aa 1-150) and the control GST
protein did not pull down the ADA2 protein. We concluded that Arabidopsis GCN5 can
directly interact with both ADA2a and ADA2b, and that the N-terminal region of GCN 5
is neither necessary nor sufficient for these interactions.

To corroborate and extend this result, a yeast two-hybrid assay was employed. The
cDNAs encoding full length or truncated GCN5 proteins as bait were inserted into vector
pGBKT7 in frame with that of Gal4 DNA binding domain (DBD). cDNAs encoding
ADA2 proteins as prey were subcloned into vector pGADT7 in frame with that of Gal4
activation domain (AD). Yeast strain AH109 cells transformed with respective bait and
prey plasmids were grown on selective SD medium for three to four days at 30° C to test
protein-protein interaction.

As shown in Figure 2, both ADA2a and ADA2b interacted with the full length of

GCN5, but not the unique N-terminus of GCN5 (amino acids 1-210). This data confirmed

50

 

«in». §

Figure l. GCN5 interacts with ADA2a and ADA2b.

GST pull-down assay was performed to test GCN5 interaction with ADA2a
and ADA2b. ADA2a and ADA2b proteins, expressed and [35S]-methionine
radiolabeled in vitra, were pulled down by equivalent amount of GST-GCN5
and GST proteins, that were bound to glutathione-Sepharose beads. The pull-
down ADA2a and ADA2b proteins were visualized by autoradiography after
being analyzed in SDS-PAGE gel. 1: 50% Input, 2: GST-—GCN5(l-568),

3: GST--GCN5(155-568), 4: GST--GCN5(l-150), 5: GST.

51

 

Figure 2. GCN5 interacts with ADA2a and ADA2b.

Yeast two-hybrid assays were performed to test the interaction. The interaction of
GCN5 with ADA2a is shown in panel A and B, and ADA2b in panel C and D.
The pictures were taken after the yeast cells had been grown on selective solid
medium (SD Trp- Lcu- His- Ade- /X-a-gal) (A and C) or on plasmid maintaining
solid medium (SD Trp- Leu-) (B and D) for 4 days at 30°C. 1, 2, 3, 4: yeast co-
transforrned with prey plasmid pGAD-ADA2a and bait plasmid pGDBD-GCNS
(1-210), pGDBD-Lamin, pGBKT7, and pGDBD-GCN5(fl) respectively. 6, 7, 8,
9: yeast co-transformed with prey plasmid pGAD-ADA2b and bait plasmid
pGDBD-GCN5 (1-210), pGDBD-Lamin, pGBKT7, and pGDBD-GCN5(fl)
respectively. 5: yeast co-transformed with plasmid pGDBD-p53 and pGAD-LT
as positive control, where LT stands for SV40 large T antigen. Plasmid pGBKT7
was the empty bait vector. In the plasmid nomenclature, GDBD stands for Gal4
DNA binding domain and GAD for Gal4 activation domain.

52

the result of GST pull-down experiment of GCN5 interaction with ADA2a and ADA2b,
and further found that the N-terminus of GCN5 (amino acids 1-210) was not sufficient
for interaction with ADA2a and ADA2b. Based on yeast colony growth, the ADA2b
protein seemed to have stronger interaction with the GCN5 protein than the ADA2a

protein.

Defining the interaction domains/rpgjons between GCN5 and ADA2

Yeast two-hybrid assays were performed to further define the interacting regions or
domains between GCN5 and ADA2a or ADA2b. GCN5 has distinguishable domains
including the bromodomain (aa 451-568) and HAT domain (a 203 -3 84), and its unique
N-terminus (aa 1-200). For the yeast two hybrid assay, the GCN5 protein was split into
the N-terminus (aa 1-210), the HAT domain (a 203-384), the bromodomain (aa 451-568)
and the region between the HAT and the bromodomain (H-B, aa 361-468). The cDNAs
of these region of GCN 5 (except the N-terminus) were PCR amplified and subcloned into
prey plasmid pGADT7 in fiame with that of Gal4 AD.

These plasmids were transformed into yeast strain AH109 together with ADA2a or
ADA2b plasmids, and assayed in selective plates as described above.

The results in Figure 3 showed that the HAT domain of GCN5 (amino acids 203-384)
was sufficient to interact with ADA2a or ADA2b. Neither the bromodomain nor the H-B
region showed interaction. GCN5 (aa 230-468) failed to show interaction with ADA2a/2b
(Figure 3), suggesting that amino acids 203 -230 in the HAT domain is necessary for

interaction with ADA2a or ADA2b.

Similar experiments were set up to test which region in ADAZa and ADA2b

53

 

C
E
380 458
II \I - lirnmn
HAT Bromo 1450-568)
Harms)

GCN5 1230-568)

Figure 3. The GCN5 HAT domain interacts with ADA2a and ADA2b.

Yeast two-hybrid assays were performed to test the interaction. The interaction of
GCN5 with ADA2a is shown in panel A and B, and ADA2b in panel C and D. The
pictures were taken after the yeast cells had been grown on selective solid medium

(SD Trp- Leu- His- Ade- IX-a-gal) (A and C) or on plasmid maintaining solid medium
(SD Trp- Leu-) (B and D) for 4 days at 30°C. 1, 2, 3, 4: yeast co-transformed with prey
plasmid pGAD-ADAZa and bait plasmid pGDBD-GCNS (HAT), pGDBD-GCN5(H-B),
pGDBD-GCNS (230-468), and pGDBD-GCNS (Bromo) respectively. 7, 8, 9, 10: yeast
co-transformed with prey plasmid pGAD-ADAZb and bait plasmid pGDBD-GCNS
(HAT), pGDBD-GCNS (H-B), pGDBD-GCNS (230—468), and pGDBD-GCNS (Bromo)
respectively. 5: yeast co-transformed with plasmid pGDBD-GCNS (HAT) and plasmid
pGADT7. 6: yeast co-transformed with plasmid pGDBD-p53 and pGAD-LT as positive
control, where LT stands for SV40 large T antigen. Plasmid pGADT7 was the empty
prey vector. In the plasmid nomenclature, GDBD stands for Gal4 DNA binding domain
and GAD for Gal4 activation domain. (E) Diagram shows the respective GCN5 regions
tested in the yeast two-hybrid assays.

54

interacted with GCN5 and its respective domains. ADA2 proteins, including Arabidopsis
ADAZa and ADA2b, have three recognizable conserved regions, respectively named as
the N-terminus, the middle part and the C-terminus. In order to test their detailed
interaction with GCN5, ADA2a and ADA2b were divided accordingly into three
conserved parts as illustrated in Figure 4A, the N-terminus (ADA2a aa 2-267, ADA2b aa
2-229), the middle part (ADA2a aa 267-418, ADA2b aa 226-377) and the C-terminus
(ADA2a aa 415-548 ADA2b aa 354-486). The cDNAs of these regions were PCR
amplified and subcloned into prey plasmid pGADT7 in frame with that of Gal4 AD.

The results of yeast two-hybrid assays, as shown at Figure 4B, indicate that the
middle part of ADA2a or ADA2b was found sufficient to interact with the HAT domain
of GCN5. The C-terminal of ADA2a and ADA2b did not interact with GCN 5. Curiously,
the N-terminus of both ADA2a (ADA2aN) and ADA2b (ADA2bN) showed interaction
only to the full length of GCN5. Neither the bromodomain nor the HAT domain of GCN5
was sufficient for interaction with the N-terminus of ADA2a and ADA2b. The interaction
between ADA2aN and GCN5 seemed weak because the corresponding yeast colonies
were pretty small in growth compared to the ADA2bN-GCN 5 yeast colonies (Figure 4B).

The regions of ADA2a and ADA2b that interact with GCN5 were more narrowly
defined by testing additional deletion constructs. These interactions were confirmed again
as shown in Figure 5, where the N-terminus of ADA2 (ADA2a: aa 2-250, 2-267, and 2-
272; ADA2b: aa 2-208, 2-225, 2-229) showed interaction with the full length of GCN5,
and the middle part of ADA2 (ADA2a: aa 267-418, 273-418; ADA2b: aa 226-377, 230-

377) showed interaction with the HAT domain of GCN5.

The results shown thus far indicate that the interaction pattern of ADA2a with GCN5

55

A 6 268 418

 

 

      

WU

Br G5 r GPSI
Figure 4. The middle part of ADA2a and ADA2b is sufficient to interact
with the HAT domain of GCN5.

Yeast two-hybrid assays were performed to test the interactions. The different
parts of GCN5 as bait protein and the different parts of ADA2a/b as prey protein
are shown schematically in panel A. The GCN5 protein was tested for the full
length (G5), the HAT domain and the bromodomain (Br). The ADA2a/b proteins
were tested for the three conserved parts, the arnino-terminus (aN or bN), the
middle part (aM or bM) and the carboxyl-terminus (aC or bC). The yeast two-
hybrid assays results are shown at panel B, where the yeast cells have been
grown on selective solid medium (SD Trp- Leu- His- Ade- /X-or-gal) for 4 days
at 30°C. Note: there is tiny colony growth in the group “aN G5” which is not
visible in the picture

56

 

Figure 5. The N-terminus of ADA2 interacts with GCN5 and the middle part of
ADA2 interacts with the HAT domain of GCN5.

Yeast two-hybrid assays were performed to test the interactions. Shown are the interaction
of GCN5 (full length, fl) with various regions of ADA2a (A) or ADA2b (C), and the HAT
domain of GCN5 with various regions of ADA2a (B) or ADA2b (D). The pictures were
taken afier the yeast cells had been grown on selective solid medium (SD Trp- Leu- His-
Ade- lX-a-gal) (the left picture for each panel) or on plasmid maintaining solid medium
(SD Trp- Leu-) (the right picture for each pane) for 4 days at 30°C. 1, 2, 3, 4, 5: yeast co-
transfonned with prey plasmid pGAD-ADAZa (273-418, 261-418, 1-272, 1-267 or 1-250
respectively) and bait plasmid pGDBD-GCNS (fl) (A) or pGDBD-GCNS (HAT) (B). 6,

7, 8, 9, 10: yeast co-transformed with prey plasmid pGAD-ADAZb (230-377, 225-377,
1-229, 1-225 or 1-208 respectively) and bait plasmid pGDBD-GCN5 (fl) (C) or pGDBD-
GCN5 (HAT) (D). In the plasmid nomenclature, GDBD stands for Gal4 DNA binding
domain and GAD for Gal4 activation domain.

57

was the same as that of ADA2b with GCN5, except that the N-terminus of ADA2a
showed weaker interaction with the full length GCN5. The following experiments
identified three exceptions to this pattern, indicating subtle distinctions in the interactions
of GCN5 with ADA2a and ADA2b.

As shown in Figure 6B and C, the GCN5 HAT domain (aa 203-3 84) was repeatedly
observed to interact with the middle part of both ADA23 and ADA2b. However, a short
version of the GCN5 HAT domain (a 203-368), interacted only with the middle part of
ADA2a (Figure 6B), but not the middle part of ADA2b (Figure 6C).

As shown in Figure 7, the N-terminus of ADA2b (1-229), but not ADA2a (1-267),
retained the ability to interact with the truncated GCN5 (aa 369-568) that contained the
H-B region and the bromodomain. But neither the bromodomain nor the H-B region
alone was suffrcient for interaction with the N-terminus of ADA2b.

Moreover, the extended N-terminus of GCN5 (amino acids 1-250, compared to
previous 1-210 in Figure 2) showed interaction with the full length of ADA2a but not the
full length of ADA2b (Figure 8). Amino acids 210-250 of GCN 5 are inside the HAT
domain which had previously found to interact with the middle part of ADA2 (Figure 5).
But the new yeast two-hybrid result show that the middle part of ADA2a was not

sufficient for interaction with the extended N-terminus of GCN5 (Figure 8).

Discussion
The experiments described in this chapter indicate that the Arabidopsis GCN 5
protein can interact with both ADA2a and ADA2b both in vitra and in yeast interaction

assays. The HAT domain of GCN5 interacts with the middle part of ADA2a and ADA2b,

58

A 68 418 8
aM (267 l )

 

GCN5

1 AT 203-384

360-468)
4 (ppm, 203-368)
5 (369-568]
6 r 451-568)

B
C

 

4

Figure 6. Differential interaction of the middle part of ADA2a and ADA2b with
the HAT domain of GCN5.

Yeast two-hybrid assays were performed to test the interactions. The different parts of
GCN5 (l, 2, 3, 4, and 5) as bait protein and the middle conserved part of ADAZa (aM)
and ADA2b (bM) as prey protein are shown schematically in panel A. The yeast two-
hybrid assays results are shown at panels B (for aM) and C (for bM), where the yeast
cells have been grown on selective solid medium (SD Trp- Leu- His- Ade- /X-or-gal)
(the left picture for each panel) or on plasmid maintaining solid medium (SD Trp- Leu-)
(the right picture for each panel) for 4 days at 30°C.

59

 

 

2 l369—568)
answer-468)
4 (ET, 203-384)
B
C

 

Figure 7. The N-terminus of ADA2b but not ADA2a interacts with GCN5 ($69-$68).
Yeast two-hybrid assays were performed to test the interactions. The different parts of
GCN5 (l, 2, 3, and 4) as bait protein and the middle conserved part of ADA2a (aN)

and ADA2b (bN) as prey protein are shown schematically in panel A. The yeast two-
hybrid assays results are shown at panels B (for 3N) and C (for bN), where the yeast
colonies have been grown on selective solid medium (SD Trp- Leu- His- Ade- /X-a-gal)
(the left picture for each panel) or on plasmid maintaining solid medium (SD Trp- Leu-)
(the right picture for each panel) for 4 days at 30°C.

60

 

Figure 8. The extended N-terminus of GCN5 interacts with ADA2a,

but not ADA2b.

Yeast two-hybrid assays were performed to test the interactions. The growth

of the yeast colonies with respective bait and prey plasmids on selective solid
medium (SD Trp- Leu- His— Ade- /X-a-gal) is shown at the lefi, and on plasmid
maintaining solid medium (SD Trp- Leu-) at the right. The pictures were taken
after the colonies had been growing at 30°C for 4 days. 1, 2, 3, 4: yeast co-
transformed with bait plasmid pGDBD-GCNS (1-250) and prey plasmid pGAD-
ADA2a (full length), pGAD-ADAZa (267-418), pGAD-ADAZb (full length), or
pGAD—ADAZb (226-377) respectively.

61

and the N-termini of ADA2a and ADA2b interact with the full length GCN5 protein.

These interactions can be taken as preliminary evidence for the presence of ADA2-
GCNS complex(es) in Arabidopsis in viva. To what extent the in viva complex(es) is
similar to the yeast SAGA and human PCAF complexes is unknown. The genomic
sequence of Arabidopsis reveals homologous proteins of yeast Tral protein and the TAFs
(T AF 5, 6, 9, 10 and 12) which are components of yeast SAGA. Spts. But curiously, the
Arabidopsis genome encodes no apparent homologues components of yeast SAGA
complex including Ada], AdaB, AdaS, Spt3, Spt7 and Spt8, suggesting the GCN5-ADA2
complex in Arabidopsis might be different in composition.

It is unlikely that both ADA2a and ADA2b are present in the same complex in viva
based on the observation that both ADA2a and ADA2b bind to a common part of GCN5.
Rather, Arabidopsis likely contains two GCN5-ADA2 complexes: one containing
ADAZa and GCN5 and the other containing ADA2b and GCN5. Supporting evidence
comes from the characterization of ADA2 proteins in Drasaphila. Drasaphila, like
Arabidopsis, has two ADAZS designated as ADA2a and ADA2b. Biochemical
purification of Drasaphila ADA2 associated proteins illustrates that ADA2b is in a
SAGA-like complex but ADA2a is not (Kusch, et al., 2003, Muratoglu, et al., 2003),
although ADA2a also shows interaction with Drasaphila GCN5. The enrooted
phylogenetic tree shows that the plant ADAZS have diverged into a separate branch
different from that of the metazoan ADAZS (Kusch, et al., 2003). This suggests that the
Arabidopsis ADA2-GCN5 complex(es) might also have its own uniqueness in
compositions.

In addition, genetic data from our lab also shows functional distinctions between

62

ADAZa and ADA2b in Arabidopsis. Arabidopsis mutants with homozygous disruption in
the GCN5 gene has very similar defective phenotype to that of ADA2b disruption
mutants (V lachonasios, et al., 2003), while no obvious defective phenotype has been
traced to plant with homozygous disruption of ADA2a gene.

Although the Arabidopsis ADA2a and ADA2b proteins interact with GCN5, the
details of those interaction differ in three ways, (i) ADA2a, but not ADA2b binds to
GCN5 (aa 1-250), (ii) the N-terminus of ADA2b, but not the N-terminus of ADA2a binds
to truncated GCN5 (aa 369-568), (iii) the middle part of ADA2a, but not ADA2b binds to
GCN5 (aa 203 -368). Whether these differences have biological relevance is not known.

In budding yeast, the conserved N-terminus of Ada2 protein interacts with Gcn5
protein and the HAT domain of Gcn5 interacts with the ADA2 protein (Candau and
Berger, 1996). These interactions are similar to what have been found in the interactions
of Arabidopsis GCN5 with ADA2 and ADA2b. But there are some specific interactions
of Arabidopsis GCN5 with ADA2 and ADA2b in detail as discussed below.

The GCN5 C-terminal (aa 369—568) is sufficient to interact with the N-terminus of
ADA2b (Figure 7). This interaction has not reported from other system. This C-terminal
of GCN5 includes the conserved H-B region and the bromodomain. However, neither H-
B nor bromodomain is sufficient to interact with the N-terminus of ADA2b (Figure 7 and
Figure 4). One explanation of these observations is that the H-B region can not fold
correctly unless bromodomain is connected or that the N-terminus of ADA2b needs to
contact both H-B and bromodomain for sufficient interaction.

The physical interaction of the middle part of ADA2a and ADA2b with GCN5 HAT

has also not been reported from other model organisms. This specific interaction is also

63

confirmed in separate yeast two-hybrid screening experiment. When GCN5 in full length
was used as bait to screening for interacting proteins expressed from Arabidopsis
seedling cDNA library, the major prey proteins identified were truncated ADA2a and
ADA2b protein that converge on the conserved middle part (see Chapter V for more
information). A very recent report (Sterner, et al., 2002) showed that in yeast partial
deletion of the conserved middle part of the Ada2 protein results in disintegration of
Gcn5 and Ada3 in purified SAGA complex. This observation might suggest indirectly
that the middle part of Ada2 is involved in the interaction with Gcn5 in budding yeast.
Curiously, the middle part of ADA2b loses interaction with GCN5 HAT domain
when 16 amino acids (369-364) are deleted from the HAT domain. This is reminiscent of
what has been observed in budding yeast (Candau, et al., 1997). In budding yeast, the
HAT activity of Gcn5 is defined to amino acids 1-261. The region right after the HAT
domain (aa 262-280) in yeast Gcn5 is necessary for interaction with Ada2 in an in vitra
co-expression and co-immunoprecipitation assay. This interacting region in Gcn5 with
Ada2 is critical for the in viva fimction of GCN5. In gcn5 null mutant, GCN5 (1-280) is
able to complement for growth in minimum medium, able to potentiate the activator
function of Gal4-VP] 6 in low copy plasmid, able to restore the toxicity of Gal4-VP16 in
high copy plasmid to yeast. All these fimctions are lost when GCN5 (1-280) is truncated
to GCN5 (1-261). In protein sequence alignment, the region of amino acids 262-280 in
yeast Gcn5 overlaps with the region of amino acids 369-3 84 of Arabidopsis GCN5, and
both regions are located right at the end of HAT domain. This overlapping suggests that
there might be common motif important in the GCN5 protein for interaction with ADA2.

That the N-terrninus interacts with GCN5 and the middle part of ADA2 interacts

64

with the HAT domain of GCN5 suggests that the ADA2 and GCN5 interaction involves
multi-site contacts. One of the significances of these physical interactions is illustrated by
the fact that both the N-terminus and the middle part of ADA2 enhance GCN5 HAT

activity (see Chapter IV for more detail).

65

Chapter Four

Stimulation of GCN5 Epzyme Activity by ADA2b

Part 1. Purification of recombinant protein of GCN5. ADA2a and ADA2b
Part 11. Both ADA2a and ADA2b enhance the HAT activig of GCN5.

Introduction

The role of ADA2 in GCN5-ADA2 complex is not yet well understood. In yeast,
ADA2 is not an essential gene. The SAGA complex can be purified intact in yeast strains
with null mutations in ADA2 (Grant, et al., 1997). Although early models proposed that
ADA2 serves as a handle for transcriptional activators to recruit SAGA complex to target
promoters (Barlev, et al., 1995), more recent work indicates that found that
transcriptional activators contact SAGA complex by its component, instead of Ada2,
ADA], TAF6, TAF 12 and Tral (Brown, et al., 2001, Hall and Struhl, 2002). On the other
hand, the GCN5 transcriptional activation activity has been shown to depend on Ada2
interaction in yeast (Candau, et al., 1997).

Recombinant yeast Gcn5 protein is active on free histone but not nucleosomal
histones (Kuo, et al., 1996), while SAGA is able to acetylate nucleosomal histones (Grant,
et al., 1997) in vitra. Therefore, other components in SAGA must enable Gcn5 to
acetylate nucleosomal substrates. All these observations led us to test the hypothesis that

Arabidopsis ADA2 might regulate GCN5 enzyme activity.

Part 1. Purification of recombinant protein of GCN5, ADA2a and ADA2b

66

Experimental Methods

Expressing plasmid construction

GST-tagged expression vector pGEX-6P-l (Amersham Biosciences) was used to
construct all GST-fusion plasmids. The GST tag in the expressed GST fiision proteins
can be cut off by PreScission protease which cuts the amino acid sequence Leu-Glu-Val-
Leu—Phe-Gln /Gly-Pro.

cDNA fragments of the Open reading frame of GCN5, ADA2a and ADA2b were
inserted in frame into vector pGEX-6P-l . The cDNA fragments included those encoding
GCN5 (aa 1-568, full length), ADA2b (the full length of wild type, alanine or arginine
point mutation at K215), the three individual parts of ADA2b (the N-terminus, aa 1-225;
the middle part, a 226-3 77; and the C-terminus, aa 354-486), and the three individual
parts of ADA2a (the N-terminus, aa 1-267; the middle part, a 267-418; and the C-
terminus, aa 415-548). For ADA2b, the cDNA encoding the firll-length protein was also

inserted to His6-tagged vector pET-28 (N ovagen) to express the His6-tagged ADA2b

protein.

Recombinant protein purification

All recombinant proteins were expressed in E. cali strain BL21(DE3) Codon Plus
(Stratagene). The purification procedure was carried out at 4°C unless indicated.

Transformed E. cali cells with respective plasmids were first grown in 10 ml liquid
LB with ampicillin (100 rig/m1) and chloramphenicol (40 ug/ml) at room temperature
overnight. The 10 ml culture was transferred to 1000 ml liquid LB with the same

concentration of antibiotics. The cells were induced at room temperature (around 20°C)

67

for 18 hours with 60-90 uM IPTG. The cells were then collected by centrifugation, and
washed once with cold HEPES buffer (40 mM HEPES, 50 mM KCl, 10% glycerol, 7
mM 2-mercaptoethanol). The washed cells were resuspended in about 20 ml of HEPES
buffer plus 150 mM KCl with protease inhibitor cocktail tablets (Roche). Cell lysates
were made by French press treatment. In order to improve the solubility of the expressed
fusion protein, Triton X-100 was usually added to the lysates to final concentration of
0.2% and the lysate was rocked at 4°C for 30 minutes. Cleared supernatant was recovered
after centrifugation at 20,000 x g for 20 min at 4°C and stored at -80°C.

To purify GST-fusion proteins, the supernatant was incubated with appropriate
volume of glutathione-Sepharose beads (Amersham Biosciences) in cold room for about
one hour. The beads were then washed with an excess volume of HEPES buffer plus 150
mM KCl and 0.001% Triton X-l 00, and then washed once with HEPES buffer.

GST fusion proteins were eluted from the Sepharose beads with HEPES buffer
containing 20 mM glutathione. The directly eluted GST fusions included GST-ADA2b
(the full length of wild type, the K215A and K215R mutants), GST-ADA2bN (the N-
terminus), GST-ADA2bM (the middle part), GST-ADA2bC (the C-terminus), GST-
ADA2aN, GST-ADA2aM and GST-ADAZaC. GST protein alone as control was purified
in the same way.

For purifying recombinant GCN5, the GCN5 protein was cut off from GST fusion
protein directly while it was still bound to the Sepharose beads by PreScission protease
(Amersham Biosciences) in HEPES buffer for four hours at cold room.

For the purification of His6-ADA2b, the procedure of the expression and purification

of GST fusions described above was followed with the following modifications: (i) Ni-

68

NTA beads (Qiagen) were used instead of Sepharose beads, (ii) buffer A (20 mM HEPES,
25 mM KCl, 5% glycerol, 7 mM 2-mercatpoethanol) was used, (iii) 20 mM imidazole
was included in the cell lysates, (iv) 200 mM KAc included in the washing solutions and
40 mM imidazole was also included in the final washing step, (v) the His6-ADA2b
protein was eluted from the Ni-NTA beads with 225 mM imidazole in buffer A. The
eluted His6-ADA2b was either used directly or dialyzed in HEPES buffer at cold room.
Because of the considerable contamination of E. cali proteins in purified His6-
ADA2b protein, the same purification procedure was followed to purify the contaminated

proteins from E. cali cells transformed with the vector pET-28.

Western blotting
Protein samples of His6-ADA2b and GST-ADA2b were electrophoresed in 7.5%

SDS-PAGE gel. The proteins were detected by standard Western blotting. The first
antibody was either ADA2b specific IgG (1:4000 dilutions) purified from rabbit
antiserum (rabbit #108) against ADA2b peptide antigen (aa 392-405), GST antibody
(BioRad, 122000 dilutions) and His6-tag monoclonal antibody (Sigma, 1:3000 dilutions).

The signal was detected using Lumi-light Western Blotting Substrate kit (Roche).

Exmrimental results

In order to test the possible regulation role of ADA2 on GCN5 enzyme activity, the
Arabidopsis GCN 5, ADA2a, and ADA2b proteins were first expressed and purified fi'om
E. cali either as His6-tagged or as GST-tagged fusion protein.

The GCN5 protein was purified as a GST fusion protein and then the GST tag was

69

subsequently cut off with protease. As shown in Figure 1A, relatively pure GCN5 protein
was obtained. Acquire

The ADA2 protein has been very hard to express and purify. After many trials,
reasonable amounts of recombinant ADA2b protein were obtained using low
concentrations of IPTG (60-90uM) and prolonged induction time (about 18 hours) at
room temperature. The expressed recombinant ADA2b, as GST-tagged or His6-tagged
fusion protein, was then purified as shown in Figure 1B and Figure 1C. Recombinant
ADA2a protein in full length could not be successfully purified regardless of the
modification of expression and purification methods.

There was considerable contamination in the isolated ADA2b protein either in GST
tagged or His6-tgged form. The presence of GST-ADA2b or His6-ADA2b protein was
confirmed by Western blotting assay using either ADA2b peptide antibody, His6
monoclonal antibody or GST antibody (Figure 2). The Western blotting result also
detected some truncated ADA2b proteins in the purified preparation.

Recombinant ADA2a and ADA2b protein were also purified in their truncated forms.
Based on the amino acid sequence similarity as discussed in chapter 111, ADA2b protein
was divided into three regions, the N-terminus (2bN, amino acids 1-225), the middle part
(2bM, amino acids 226-3 77) and the C-terrninus (2bC, amino acids 353-486). Relatively
pure GST-tagged proteins of all these ADA2b regions were isolated (Figure 1D).

Similarly, the three regions of ADA2a (2aN: 1-267, 2aM: 267-418, and 23C: 418-
548) were also obtained as GST-tagged fusions in relative purity (Figure 1E).

For the three regions of both ADA2a and ADA2b, in addition to the relatively high

purity, the yield was also relatively high: GST-ADA2bN (up to7 ug/ul), GST-ADAZbM

70

kDa
83 a
62 ,, .. .- “ ""
47.5
iii
D E
B Sul C 2.53
kDannrs 12 NMCkDaNMC
175»- kDa 83- 83—
83~--=< 83' ..
” :: '9‘ 62—-< 62-
62 "“' 62-
as .4 , _ >-
475 -- 47.5- - 47-5‘ 475 >-

Figure l. The purified recombinant proteins of GCN5, ADA2a and ADA2b.
The recombinant proteins were analyzed in SDS-PAGE gel and Coomassie Blue
stained. Shown are purified recombinant proteins GCN5 (A) where BSA protein
was included to indicate the relative quantity of the GCN5 protein, GST-ADA2b
(B), His6-ADA2b (C, lane 2) where proteins purified from expressed E. cali
transformed with His6 vector only is also shown (lane 1), the three parts as GST
fusions of ADA2a (D) and ADA2b (E). The positions of the respective proteins
in the gels are indicated by the arrows. (B) f1, f2, f3: elution fraction 1, 2, and 3
respectively from glutathione Sepharose beads. (D, E) N: the N-terminus, M: the
middle part, C: the C-temrinus.

71

 

A t'd B
”“3 ' ° anti-GST anti-His6
antibody , .
kDa 1:4000 antimy antibody
1:2 :
175- "km kDa "'72....1 _r!~39100. _
175- '
83 - -
g P ..— - - a
62 - ’ - ‘p— ." -
62" " '
47.5. a .— _ _
47.5- ... °
4 ‘ "
32.5 - f; «o
I t 32 5. u — . a ..a
q .
25 .. __
25- f” . ‘ --
l 2 3 4 1 2 3 4

Figure 2. Confirmation of the purification of recombinant protein of
Arabidopsis ADA2b.

Western blotting assays were performed to detect the recombinant ADA2b
protein purified either as GST-tagged firsion protein (GST-ADA2b) or His6-
tagged fusion protein (His6-ADA2b). The recombinant ADA2b proteins
were electrophoresed in 7.5% SDS-PAGE gel, and detected by antibodies
against either ADA2b peptide antigen (A) or GST tag and His6 tag antigens
(B). The detected His6-ADA2b protein is indicated by the black arrow while
the GST-ADA2b protein by the blue arrow. For the GST-ADA2b protein
preparation, either 0.75 pl (lane 1) or 0.38 ul (lane 2) was loaded in the
SDS-PAGE gel. For His6-ADA2b protein preparation, either 1.5 ul (lane 3)
or 0.75 ul (lane 4) was loaded.

72

(about 0.5 ug/ul), GST-ADA2bC (up to 7 ug/ul), GST-ADA2aN (up to 2 ug/ul), GST-

ADA2aM (267-418) (about 0.5 ug/ul), and GST-ADA2aC (up to 20 ug/ul).

Discussion for Part I

The purification ADA2 proteins of other species has been hard. For instance,
reasonable amount of yeast ADA2 was purified only when yeast GCN5 is co-expressed
in E. coli (Balasubramanian, et al., 2002).

Reasonable amount of recombinant fusion proteins GST-ADA2b and His6-ADA2b
were obtained from E. cali cells under the mild expression conditions of low temperature
(around 20 °C) and prolonged induction (around 18 hours) with low amount of IPTG
(60-90 uM), although the purity is low. However, expression and purification of full-
length ADA2a protein has not been successful.

The preparations of GST-ADA2b or His-ADA2b had much contamination (Figure 1).
No further purifications were applied to remove the contamination. But the existence of
purified His6-ADA2b or GST-ADA2b in the contaminated mixture was confirmed by
Western blotting, using antibodies against the tag of His6 or GST or ADA2b specific
antibodies. The Western blotting showed that the purified ADA2b protein preparation
contained the full-length ADA2b protein and some truncated ADA2b proteins which
were probably the products of incomplete translation or proteolysis.

For unknown reasons, early experiments fount that phosphate buffer (140 mM NaCl,
2.7mM KCl, 10 mM NazHPO4, 1.8 mM KH2P04, pH 7.3, for GST-ADA2b; and 50 mM
NaHzPO4, 300 mM NaCl, for His6-ADA2b) was found not very compatible with ADA2b

protein, which sometimes could lead to either poor elution of ADA2b protein from

73

glutathione Sepharose and Ni-NTA beads, or precipitation of the fusion proteins after
dialysis. The chelating reagent EDTA also caused similar problems. This incompatibility
might be due to the putative zinc finger at the ADA2b N-terminus, for EDTA sequesters
zinc ions and phosphate precipitates zinc ions. Therefore, HEPES buffer (40 mM HEPES,
pH 7.3, 50 mM KCl, 10% glycerol) was used subsequently in purification of all ADA2
and GCN5 recombinant proteins.

Using HEPES buffer, however, only a very small fraction of GST-ADA2b or His6-
ADA2b could be eluted from glutathione Sepharose or Ni-NTA beads. But relatively
concentrated amounts of the three parts of protein ADA2b could be eluted fi'om
glutathione Sepharose beads. The GST tag could also be readily removed by protease
digestion from both GST-ADA2bN and GST-ADA2bC proteins.

For ADA2a, it was unable to purify more than trace amounts of either His6-tagged or
GST-tagged protein regardless of using phosphate or HEPES buffer. However, like the
three parts of ADA2b, very clean and concentrated purification of the three regions of

ADA2a were achieved using HEPES buffer (Figure 1).

Part II. ADA2 enhances the HAT activity of GCN;
Exmrimental Methods
Free core histonee

Core histone mixture (H3, H23, H2A, and H4) was purchased fiom Sigma (Cat. No.
H-9250, purified from calf thymus). The mixture was dissolved in HEPES buffer (40 mM
HEPES, 50 mM KCl, 10% glycerol, 7 mM 2-mercaptoethanol) to final concentration of

10 mg/ml.

74

Free histone acetylation assay

For testing the enhancing activity of GST-ADA2b, the reaction was carried out in 51
pl HEPES buffer, containing about 4 pM (0.25 pg) GCN5, 15 pg core histone mixture,
and 0.1 pCi [3 H] Ac-CoA. Varying amounts of GST-ADA2b ranging from about 1.2 to 6
pg were added. The GST protein (ranging from about 2 to 10 pg) was added in the
controls. The reaction proceeded for 5 or 10 min and was quenched with 12 ul of 2N HCl.
15.5p1 from each reaction was spotted onto P81 Whatrnan filter paper (half of filter paper
disc with the diameter of 2.5 cm). The paper discs were washed in 500 ml of 50 mM
NaHCO3 (pH adjusted to 9.1) for three times. Each disc was then soaked in 5 ml of
Ecolume liquid scintillation cocktail (ICN company). Scintillation activity was read for
one minute three days later by scintillation counter (LKB, 1208 rackbeta). For each 63 pl
reaction (51 pl reaction plus 12 pl quenching HCl), three aliquots of 15.5 pl were
counted.

The remaining 16.5 pl of each reaction was analyzed in 15% SDS-PAGE gel. The
gel was first stained with Coomassie Blue to visualize the proteins, and then the gel was
soaked in autoradiographic enhancer (NEN Life Science Products), vacuum dried and
exposed to X-ray film for three days for visualization of acetylated histones and ADA2
proteins.

For testing the stimulatory activity of the three parts of ADA2b or ADA2a, the same

procedure was followed varying amounts of the three parts.

N ucleosome purification

Mono and oligonucleosomes were purified from turkey erythrocytes, following the

75

protocol (Cote, et a1, Methods de Molecular Genetics 1995, 108-127) with some
modification. All procedure was done at 4°C or on ice. About 60 ml blood was collected
fi'om turkey, and the erythrocytes were collected after centrifugation at 3,000 x g. The
cells were suspended in cell lysis buffer [20 mM HEPES, pH 7.5, 0.25 M sucrose, 3 mM
Mng, 0.2% (v/v) Nonidet P-40, 3 mM 2-mercaptoethanol, 0.4 mM
phenylmethylsulfonyl fluoride (PMSF), 1 pM pepstatin A, 1 pM leupeptin]. The nuclei
were pelleted at 3,000 x g and washed with the lysis buffer and then with buffer B (20
mM HEPES, pH 7.5, 3 mM MgC12, 0.2 mM EGTA, 3 mM 2-mercaptoethanol, 0.4 mM
PMSF, 1 pM pepstatin A, 1 pM leupeptin). The nuclei were then resuspended in 5 ml
buffer B, to which an additional 3 ml buffer B containing 0.6 M KCl and 10% glycerol
was added dropwise while stirring. The nuclei were pelleted again at 17,500 x g for 30
min, and washed with buffer MSB (20 mM HEPES, pH7.5, 0.4 M NaCl, 1 mM EDTA, 1
mM 2-mercaptoethanol, 0.5 mM PMSF, 5% glycerol).

The nuclei were then resuspended in buffer H83 (20 mM HEPES, pH7.5, 0.65 M
NaCl, 1 mM EDTA, 1 mM 2-mercaptoethanol, 0.5 mM PMSF, 0.34 M sucrose). The
nuclei were broken by gentle Dounce homogenization (B pestle, stroked about 100 times)
to release high molecular weight chromatin. The homogenate was centrifiiged at 10,000 x
g for 20 min. The supernatant was dialyzed overnight in buffer LSB (20m M HEPES,
pH7.5, 0.1 M NaCl, 1 mM EDTA, 1 mM 2-mercaptoethanol, 0.5 mM PMSF). 0.03 x
volumes of 100 mM CaClz were added to the dialyzed chromatin (final volume was 8.0
ml). 4.0 ml chromatin was digested with 105 units of micrococcal nuclease (Amersham
Bioosciences) at 37°C for 6 min, and the digestion was quenched with 440 pl 0.5 M

EGTA. 4 M NaCl was added to final concentration of 0.6 M. The supernatant was

76

collected afier centrifugation at 150,000 x g for 30 min.

3.5 ml of the supernatant was applied to Sepharose CL-6B (Amersham Biosciences)
column (65 x 2.5 cm) which was equilibrated in HSB buffer without sucrose. The column
was run at 12 ml/hr and 1.1 ml fractions were collected. Fractions were checked for
nucleosome integrity by monitoring DNA length and core histone composition. For
checking the DNA length, the fraction samples were first extracted with
phenol/chloroform and then 40 pl from each extraction was loaded in 1% agarose gel.
For checking the histone composition, 5 pl of each fraction was mixed with 4 x SDS-
PAGE loading buffer and boiled for 10 minutes before loading to 15% SDS-PAGE gel.
The proteins in gel were visualized with Coomassie Blue staining.

Fractions 51-54 were pooled as nucleosome B and fi'actions 58-61 as nucleosome C.
Nucleosomes B and C were dialyzed against buffer C (40 mM HEPES, pH 7.4, 20m M
KCl, 2% glycerol, 1 mM EDTA, 0.7 pg/ml pepstatin, 1 pg/ml leupeptin, 1 pg/ml
aproptinin, 0.5 mM PMSF, 7 mM 2-mercaptoethanol) overnight. The dialyzed
nucleosome solutions, while still inside dialysis bags, was concentrated by soaking in dry
polyethylene glycol (Sigma, molecular weight 15,000-20,000 Da) for four hours. The

concentrated nucleosome solutions were kept at -80°C until used.

Nucleosomal histone acetylation assay

The yeast HAT complexes SAGA and NuA4 were included as positive controls,
which were kindly provided by Dr. Shelley Berger at Wistar Institute in Philadelphia.
For comparing the HAT activity of Arabidopsis GCN5 with that of SAGA and NuA4,

similar free histone acetylation assays as that above were performed. The assays were

77

carried out in 50 pl HEPES buffer with or without 6 mM sodium butyrate, 15 pg free
core histone mixture, 0.] pCi [3 H] Ac-CoA. 4 pl SAGA or NuA4, or 4 picomole GCN5,
or 4 picomole GCN5 plus about 10 picomole GST-ADA2b was included. The reaction
was incubated at 25°C for 15 minutes before quenched with 12 pl 2N HCl. Three aliquots
of 15 pl from the reaction were read for acetylation activity by the method of scintillation
counting on filter paper as described above. The remaining 17 pl reaction was
electrophoresed in 15% SDS-PAGE gel for fluorographic picture.

For nucleosomal histone acetylation assay, the reaction was performed in 24 p1
HEPES buffer containing 6 mM sodium butyrate, 8 pl nucleosome preparation, 0.05 pCi
[3 H] Ac-CoA, about 10 pM GST-ADA2b or His-ADA2b, and 96 pM, 24 pM or 6 pM
GCN5. The mixture reacted for 30 min at 25°C and was stopped by adding 16 p1 of 6 x
SDS-PAGE loading buffer and boiled for 5 min. As control, 2 pl of SAGA or NuA4 was
added to the reaction mixture instead of GCN5 and GST-ADA2 or His6-ADA2b. The
reactions were boiled and analyzed in 15% SDS-PAGE gel. Then the gel was soaked in
autoradiographic enhancer (NEN Life Science Products), vacuum dried and exposed to
X-ray film for visualization of acetylated histones and ADA2 proteins.

For testing mutant ADA2b effect of GCN5 activity on nucleosomal substrate, the
reaction was carried out in 27 pl HEPES buffer containing 6mM sodium butyrate, 8p]
nucleosome preparation, 0.2 pC tritiated Ac-CoA, about 8 pM GCN5 and about 4 pM or
8pM GST-ADA2b (wild type, K215R and K215A mutant). The reaction proceeded for
30 min at 25°C and was stopped by adding 16 pl of 6x SDS-PAGE loading bufl‘er and
boiled for 5 min. equivalent amount of GST was included instead of GST-ADA2b as

negative control. The reactions were resolved in 15% SDS-PAGE gel, and the acetylated

78

histones were detected by fluorography for 4 months at -80°C.

Results

ADA2b enhanced GCN5 HAT activig.

To test whether ADA2b could enhance the enzymatic activity of GCN5, HAT assays
were performed using the purified recombinant proteins as described in Part I.

For a typical HAT assay, the reaction mixture included about 4 pM (0.25 pg) GCN 5,
15 pg core histone mixture, [3 H] Ac-CoA, and varying amounts of GST-ADA2b protein
or His6-ADA2b protein which had a different repertoire of contaminated proteins (Figure
1C) from that GST-ADA2b proteins. GST or protein preparation fi'om E. cali cells
transformed with His6-tag vector only that contained identical contamination proteins
found in His6-ADA2b preparation (Figure 1C) was used as a negative control in parallel
reactions. The mixture was incubated for 10 minutes at 25°C. The acetylation signals of
core histones were measured by liquid scintillation counting on filter paper. As shown in
Figure 3, using free core histone mixture (H2A, H2B, H3 and H4) as the substrate, the
HAT activity of GCN5 was enhanced up to 5 fold by the presence of GST-ADA2b or
His6-ADA2b.

The filter paper scintillation counting method counted all acetylated proteins, and
did not indicate whether the acetylation was enhanced on one core histone or an
expanded substrate spectrum. To address this uncertainty, aliquots of the HAT assay
reactions were electrophoresed in a 15% SDS-PAGE gel. The proteins were first stained
with Coomassie blue to show their position, and the acetylation signals were visualized

by enhanced fluorography. As shown in Figure 3C, GCN5 alone (with GST protein

79

At rzrna
7000
0000
5000
4000
3000
2000
1000

 

 

 

 

GST— ’ ‘ was.»
A A AD A2b P

GST GST-ADA2b

 

7000
6000
5000
4000
3000
2000
1000

 

 

 

   

 

His6-tag His6-ADA2b

Figure 3. ADA2b enhances the HAT enzyme activity of GCN5

on free core histories.

HAT assays were carried out with recombinant GCN5, [3H] Ac-CoA, free core histones
(H2A, HZB, H3 and H4) and the recombinant ADA2b protein or controls. The HAT
activity was either quantitatively measured by liquid scintillation counting as shown by
CPM readings (A and B) or visualized by fluorography (C). (A) The results of HAT
assays using recombinant GST-ADA2b (ranging from about 0.6 to l. 5 pg) or equivalent
amounts of control GST. (B) The results of HAT assays using recombinant Hisé-ADAZb
(about 0.4 pg) or equivalent amounts of control His6-tag which was purified from
expressed E. cali cells that contained His6-tag vector only (see figure 1C lane 1). (C)
Fluorogram of the HAT assays. Parellel assays of panel A were analyzed in 15% SDS-
PAGE gel and the proteins were first visualized by Coomassie Blue staining as shown at
the left, and then the acetylation signals were detected by enhanced fluorography as
shown at a the right. 1: GST, 2: GST-ADA2b.

80

present as control) acetylated histone H3, HZB and H4 weakly. This activity was
enhanced by the addition of GST-ADA2b protein. The pattern of histone acetylation did
not change. Therefore, the enhancement activity mainly came from stronger acetylation

of current substrate histones. ADA2b did not change GCN5 substrate specificity.

The N-terminus and the middle part of ADA2a/b enhance GCN5 HAT activity.
As alluded in Part 1 above, both the ADA2a and ADA2b protein could be divided

into the three conserved parts, the N-terminus, the middle part and the C-terminus (C). As
discovered at Chapter III, both the N-terminus and the middle part of ADA2a and
ADA2b interact with GCN5. A following question was which part(s) of ADA2a and
ADA2b are necessary and sufficient to enhance GCN5 HAT activity. To answer this
question, similar HAT assays were performed using purified GST-fusion proteins of the
three parts. As shown on Figure 4A, both the N-terminus and the middle part of ADA2a
and ADA2b enhanced GCN5 HAT activity, whereas the C-terminus of both ADA2a and
ADA2b did not (the result of the C-terminus of ADA2b is not shown). The N-terminus
and the middle part of ADA2b had slightly better enhancement activity than their
counterparts of ADA2a.

To test whether the pattern of histone acetylation was altered by these two parts of
ADA2a and ADA2b, core histones were checked for their acetylation by fluorography.
As shown in Figure 4B, both the N-terminus and the middle part enhanced GCN5
acetylation activity on core histones H3, H2B and H4, but did not change GCN5 substrate
specificity. The N-terminus of ADA2b is acetylated by GCN5 (see Chapter VII), but like

the situation in ADA2b (wild type, full length), the enhanced acetylation signal by

81

 

 

 

 

1 2 3 4 5 6

123456 123456
- - 4
S -- 4
.. -2“:

$3260.03! ‘9‘-"

114—“ ‘nfi

Figure 4. The N-terminus and the middle part of both ADA2a and ADA2b
enhance the HAT enzyme activity of GCN5 on free core histories.

HAT assays were carried out with recombinant GCN5, [3H] Ac-CoA, free core histones
(H2A, H28, H3 and H4) and the three regions of protein ADA2a or ADA2b. Control
GST was also included. The HAT activity for each assay was either quantitatively
measured by liquid scintillation counting as shown by CPM readings (A) or visualized
by fluorography (B). 1: GST, 2: GST-ADA2bN, 3: GST-ADA2bM, 3: GST-ADA2aN,
4: GST-ADA2aM,5: GST-ADA2aM, 6: GST-ADA2aC: ADA2a, N: the N terminus, M:
the middle part, C: the C terminus. (A) The results of HAT assays using recombinant
GST (about 20-160 pM), GST-ADA2bN (about 8 to 60 pM), GST-ADAZbM (about 12
-90 pM), GST-ADA2aN (about 9-70 pM), GST-ADA2aM (about 13-100 pM), and
GST-ADA2bC (about 13-100 pM). (B) F luorogram of the HAT assays. Parellel assays
of panel A were analyzed in 15% SDS-PAGE gel and the proteins were first visualized
by Coomassie Blue staining as shown at the left, and then the acetylation signals were
detected by enhanced fluorography as shown at a the right.The acetylated
GST-ADA2bN and ADA2aC are indicated by the arrows.

82

ADA2bN mainly came from the acetylation of core histones. Interestingly, the C-
terrninus of ADA2a but not ADA2b was acetylated by GCN5 (Figure 4B, and data not
shown).

Having found that both the N-terminus and the middle-part of ADA2b enhanced
GCN 5 HAT activity, it was not known whether both parts together enhanced GCN5 HAT
activity better than alone. Similar HAT assays were performed with the two parts
included in the assays simultaneously. Figure 5 showed an additive effect on GCN5 HAT
activity when both the N-terminus and the middle-part of ADA2b were added
simultaneously. This primary result shows that the N-terminus and the middle part might

use different mechanisms to enhance GCN5 HAT activity.

ADA2b enhances GCN5 HAT activity on nucleosomal substrate
The preceding experiments indicate that Arabidopsis ADA2 proteins can enhance the

GCN5 HAT activity using free core histones as substrate. However, the native substrate
for GCN5 is nucleosomal histones. Therefore, studies were followed to investigate
whether ADA2 proteins can influence GCN5 HAT activity on nucleosomal histones.

The nucleosomes were purified from turkey erythrocytes. The purified
chromatographic nucleosomes fiactions contained equal amounts of the four core
histones (Figure 6A) and contained DNA fragments with length ranging from 200 base
pairs to a few kilobase pairs (Figure 6B). These nucleosomes were devoid of linker
histone H1. Fractions of mononucleosome or dinucleosomes (based on the DNA length)
from the gel-filtration were pooled and used in all subsequent acetylation assays.

As a positive control for the nucleosome acetylation assay, yeast HAT complexes

83

 

 

 

 

N M
C C

N
M
GST

N
M

032

Figure 5. The combinatorial effect of the N-terminus and the middle
part of ADA2b in enhancing the HAT enzyme activity of GCN5 on
free core histories.

HAT assays were carried out with recombinant GCN5, [3H] Ac-CoA, free
core histones (H2A, HZB, H3 and H4) and the various combinations of the
three regions of protein ADA2b as well as the control GST protein.
Equivalent amount of each protein was included in each assay. The HAT
activity for each assay was quantitatively measured by liquid scintillation
counting as shown by the CPM readings. N: GST-ADA2bN, M: GST-
ADA2bM, C: GST-ADA2bC,

84

47.5 ,3

32.5

    

25
16.5 /H3
. . Q ! 3:3
h ‘— anne- .n -—H4
6.5 \
\

43 51 58 61 71

31 33 35 37 39 41 43 45 47 49 51 53 55 57 59 6163 65 67 69 717375

 

Figure 6. Checking the purified nucleosomes from turkey blood erythrocytes.
Different chromatographic fractions of nucleosome preparations as indicted by the
numbers were either analyzed for the composition of histone proteins (A) or for the
length of genomic DNA fragments (B). (A) The fractions were analyzed in 15%
SDS-PAGE gel and stained with Coomassie Blue in gel. (B) The fractions were
electrophoresed in 1% agarose gel after phenol/chloroform extraction. The DNA
fragments were stained by ethidium bromide.

85

SAGA and NuA4 were included, which are known to acetylate nucleosomal histone H3,
H2B and H4. The SAGA and NuA4 complexes were provided kindly by Dr. Shelley
Berger from the Wistar Institute.

In order to use appropriate amounts of SAGA and NuA4 in the nucleosomal HAT
assay, the activity of SAGA and NuA4 on fi'ee core histone substrate was compared to
that of Arabidopsis GCN5 protein plus ADA2b protein. When free core histone mixture
was used as the substrate, about 2 pM GCN5 protein plus about 5 pM GST-ADA2b
protein had similar HAT activity to that of 2 pl of yeast SAGA or NuA4 (Figure 7).

Nucleosomal HAT assays were then carried out using the purified turkey
nucleosomes and the recombinant GCN5 protein and ADA2b protein as well as the
positive controls. As shown in Figure 8, the prepared turkey nucleosomes were readily
acetylated by 2 pl of yeast HAT complex SAGA or NuA4. Core histones H3, H28, H4
were acetylated. GCN5 alone did not acetylate the core histones in nucleosomes, even
using as much as about 96 picomole of GCN5 enzyme. However, in the simultaneous
presence of recombinant GST-ADA2b protein in the reaction, core histone H3 from the
nucleosomes was acetylated, even at the lowest amount of GCN5 enzyme tested (about 6
picomoles). Because of the concentration limitation of recombinant GST-ADA2b protein,
it was impractical to add more ADA2b to the HAT assay reaction. Since the purified
GST-ADA2b protein preparation had numerous contaminating proteins, a parallel assay
was performed using recombinant His6-ADA2b instead of GST-ADA2b. The purified
His6-ADA2b preparation had a very different contamination protein profile fi'om that of
GST-ADA2b (Figure 1). As shown in Figure 8, GCN5 was also stimulated to acetylate

nucleosomal histone H3 when recombinant His6-ADA2b protein was added,

86

annn
JUUU

4000"
3000"
2000"
1000

 

 

GCN5 NuA4 SAGA
ADA2b

GCN5 h I»
ADA2b 9“» 6’0

H3

- 1' “ <HZB

—H4

Figure 7. Comparing the HAT activity on free core histories ofArabidopsis
GCN5-ADA2 with that of yeast HAT complexes SAGA and NuA4.

HAT assays were carried out with [311] Ac-CoA, free core histones

(H2A, HZB, H3 and H4) and Arabidopsis recombinant proteins GCN5 (~ 2 pM)
and GST-ADA2b (~5pM), or purified yeast HAT complex SAGA (2 pl) or NuA4
(~2 pl). The HAT activity for each assay was either quantitatively measured by
liquid scintillation counting as shown by the CPM readings (A ) or visualized

87

 

 

 

 

H3
. HZB
' \HZA
\H4
B
acetyalted
ADA2b
by GCN5
M —-J
m «e
GCN5 ‘__‘
His6-ADA2b + + +
GST-ADA2b + + +
NuA4 +
SAGA +

Figure 8. Arabidopsis GCN5 does not acetylate nucleosomal core histone while
GCN5 plus ADA2b does weakly.
Nucleosomes obtained from turkey erythrocytes were incubated, in the presence of

[3H]-Ac-CoA,with recombinant Arabidopsis GCN5 protein (lane 1-9, ~6, 24 and 96
pM) supplemented with GST-ADA2b (lane 4-6, ~10 pM) or His6-ADA2b (lane 7-9),
or yeast HAT complex NuA4 (lane 10, 2 pl) or SAGA (lane 11, 2 l). Reacted mixtures
were analyzed by SDS-PAGE, stained by Coomassie Blue (panel A) and then subjected
to enhanced fluorography for 30 days (panel B).

88

demonstrating the effect was ADA2b protein specific.
Nevertheless, compared to positive control SAGA which strongly acetylated
nucleosomal histone H3, 2B and H4, Arabidopsis GCN5 together with ADA2b acetylated

nucleosomal histone weakly and only acetylated H3.

The mechanisms of ADA2b to enhance GCN5 HAT activity
The mechanisms by which ADA2 enhances GCN5 HAT activity on free core

histones are unknown. One possibility is that ADA2b improves GCN5 affinity for its
substrate histones or acetyl CoA. To test this hypothesis, similar HAT assays were
performed with constant amount of GCN 5 enzyme, GST-ADA2b protein or control GST
protein, and varying amount of core histone or acetyl CoA.

As shown in Figure 9A, the GCN 5 HAT activity increased in the presence of control
GST or GST-ADA2b with increasing amount of acetyl CoA and constant amount of core
histones, and the fold of enhancement of GST-ADA2b over control GST remained stable.
In contrast, as shown in Figure 93, with increasing amount of core histones and constant
amount of acetyl CoA, the GCN5 HAT activity decreased slightly in the presence of
GST-ADA2b, but the HAT activity increased in the presence of control GST. These
preliminary results indicate that ADA2b protein might enhance GCN5 HAT activity by

reducing the GCN5 Km for core histones.

Discussion for Part II
The experiments described in this chapter indicate that both Arabidopsis proteins

ADA2a and ADA2b can enhance the histone acetyltransferase activity of GCN5 on free

89

 

3000‘

2000‘

1000‘

 

 

 

e GCN5
e GCN5 + GST-ADA2

 

 

0.01

CPM

0.1 1010, Ac-CoA)

 

30001

2000‘

10004

 

 

 

e GCN5
O GCN5 + GST-ADA26I

 

 

10 100 (pg, histories)

Figure 9. Differential HAT activities of GCN5-ADA2b to varying

amount of Ac-CoA and core histotons.

HAT assays were carried out with recombinant GCN5 protein in the presence or absence
of recombinant GST-ADA2b protein, and either constant amount of free core histones

(H2A, HZB, H3 and H4) and varying amount of [3H] Ac-CoA (A) , or constant [3H] Ac-
CoA with varying amount of core histones (B). The HAT activities were quantitatively

measured by liquid scintillation counting as shown by the CPM readings.[3H] Ac-CoA
ranged from 0.0125 pCi to 0.2 pCi, and core histones fiom 1.25 pg to 33.75 pg.

90

core histones in vitra, and ADA2b stimulates GCN5 acetylation on nucleosomal histone
H3. The enhancing activity on free histones is attributed to the conserved N-terminus and
the middle part of ADA2a and ADA2b proteins. This functionality helps interpret the
earlier finding that both ADAZa and ADA2b interact with GCN5 in vitra and that the N-
terminus and the middle part of ADA2a and ADA2b interact with GCN5 (chapter 111).
We propose that in viva ADA2a and ADA2b regulate GCN5 enzyme activity in
Arabidopsis.

Compared to yeast SAGA complex that strongly acetylates core histone H3, HZB
and H4 from the purified turkey nucleosomes, the Arabidopsis protein GCN5 together
with ADA2b acetylates nucleosomal histones weakly and is limited to histone H3 (Figure
8). This suggests that in “viva the Arabidopsis GCN5-ADA2 complex, like the complex of
yeast SAGA or human PCAF, has more components that make it capable of efficient
acetylation of nucleosomal substrate.

In support of this suggestion, Song Tan group (Balasubramanian, et al., 2002)
reported last year that in yeast, recombinant protein complex containing Ada2 and Gcn5,
or AdaZ and Gcn5 and Ada3 had about 5-fold HAT activity over Gcn5 alone on free core
histone substrate, and Ada2 enhanced GCN5 HAT activity on nucleosomal histones
about 15 fold. Moreover, the recombinant complex Ada2-Ada3-Gcn5, compared to
SAGA complex, had similar HAT activity on nucleosomal substrate as well as the same
trend of acetylation spectrum of lysine residues on H3 tail (Lys-18 > Lys-14 > Lys-

9 > Lys-23). This similar observation in yeast Ada2 and Gcn5 suggests the general
mechanistic conservation among species. The weak acetylation activity of Arabidopsis

GCN5-ADA2 on nucleosomes suggests that more components may affect this GCN5-

91

ADA2 complex in viva, although Arabidopsis genomic sequence reveals there is no
apparent homolog of yeast Ada3 protein.

Arabidopsis GCN5 protein alone acetylated mainly core histone H3, and acetylated
HZB as well as H4 weakly when free core histones were assayed. ADA2b as well as the
N-terminus and middle part of ADA2a and ADA2b enhanced GCN5 HAT activity on all
these three core histones (Figure 3 and 4). There was no striking change of substrate
repertoire with the addition of these ADA2 proteins. The apparent identical enhancement
effect on GCN5 of the N-terminus and the middle part of ADA2a or ADA2b was
surprising in considering that the N-terrninus had different interaction nature with GCN5
from that of the middle part (Chapter III). For instance, the N-terminus of ADA2b
interacts the c-terminal of GCN5 (amino acids 369-568, including the H-B and the
bromodomain), while the middle part of ADA2b interacts the HAT domain of GCN5
(amino acids 203-3 84).

Although there was no substrate change in enhanced HAT activity of GCN 5, it was
not known for a specific substrate, such as H3, whether the improved signal of
acetylation was due to more acetylation of one specific residue or due to expansion of
acetylated residues on the specific tail. This question might be answered by Western
blotting using specific antibodies against specific acetylated residues of histone tails or by
Edman degradation sequencing.

Another general trend was that the N-terminus and middle part of ADA2b had better
enhancement activity than their counterparts of ADA2a, as revealed in Figure 5C. The
significance of this difference is not known

In the related MYST family HATS, most members such as Esal and Sas3 have a

92

cysteine rich, zinc-binding module at the N-terminus of their HAT domain. The module
is essential for the HAT activity (Akhtar and Becker, 2001, Marrnorstein, 2001). The
HAT activity of Esal is reduced more than 90% if the cysteine module is mutated
(Takechi and Nakayama, 1999). The HAT domain of CBP/p300 also has a cysteine rich,
zinc-binding module. The module is not present in the HAT domain of GNAT family
members including GCN5. Curiously, the conserved N-terminus of ADA2 has a similar
cysteine rich, zinc-binding module. The fact that the N-terminus of Arabidopsis ADA2
enhances GCN5 HAT activity suggests that the module might be important for GCN5
proper activity.

Meanwhile, the enhancement of GCN5 activity by the N-terminus of ADA2a and
ADA2b may also involve the SANT domain. The SANT domain is also conserved in the
N-terminus of ADA2. It locates right after the cysteine rich, zinc-binding module. In
yeast, purified SAGA with SAN T domain deletion in Ada2 is inactive in nucleosomal
HAT assays (Boyer, et al., 2002). In vitra, the SANT domain in yeast ADA2 has also

been demonstrated to be required for acetylation of histone H3 (Grant, et al., 1997).

Very interestingly, our preliminary data (Figure 9) suggests that ADA2b enhances
GCN5 HAT activity by lowering GCN5 Km for histone substrate but not acetyl CoA.
This prediction is in agreement with observations that yeast ADA2 improves GCN5 HAT
activity on fi'ee histone H3 tail by decreasing Km for H3 tail (3.8 fold) and increasing
Vmax (2.6 fold), but yeast ADA2 does not change GCN5 affinity for acetyl-CoA (Boyer,

et al., 2002).

The following discussion describes possible ways that the GCN5 enzyme activity

could be altered.

93

Crystal structures of the HAT domains of GNAT family members (yeast HATl,
yeast Gcn5, Tetrahymena GCN5, human PCAF) and the distantly related MYST family
member Esal reveal a striking conserved core domain, consisting of 4 B-sheets, a 100p
and a a-helix (Marrnorstein, 2001, Roth, et al., 2001). This core domain is responsible for
Ac-CoA binding and catalysis. Variable N- and C-termini, juxtaposed at the periphery of
the core domain, form two hydrophobic clefis together with the core domain. The tail of
H3 and Ac-CoA bind to the clefts. The variable N- and C- termini mediate histone
substrate binding (Marrnorstein, 2001 , Roth, et al., 2001). The constancy of the core
domain and the variability of the N- and C-termini suggest that proteins, by influencing
the structure of the N- and C-termini, may change GCN5 enzyme catalytic activity and/or
substrate specificity. Indeed, subtle conformational changes in the HAT domain are
observed upon substrate Ac-CoA binding (Marrnorstein, 2001).

Biochemical studies show an ordered substrate binding and product release in yeast
Gcn5 and human PCAF (Tanner, et al., 2000, Tanner, et al., 2000), in which Ac-CoA
binds first and then H3 tail. The acetylation reaction is carried out by an ordered Bi-Bi
mechanism, where the acetyl moiety is transferred directly to the s-amino group of lysine
14 residue on histone H3 (H3 K14). After the reaction, acetylated H3 tail is released first
followed by CoA. This strictly ordered mechanism suggests that any facilitation by other
protein in substrate binding or product release may enhances GCN5 HAT activity,
although which step is limiting is not known.

In the GCN5 histone acetylation reaction, the acetyl moiety is transferred directly to
the s-amino group of H3 K14 via nucleophilic attach. Under physiological pH (7-8), the

e-amide group of H3 K14 (pKa~10) is protonated, which hinders the nucleophilic attach

94

by acetyl-CoA. Therefore, any influential action on the protonation of the c-amide group
will improve the HAT enzyme activity by changing Kcat. A critical glutamic acid residue
is conserved in HAT domain of the GCN5 and PCAF proteins. In yeast Gcn5 protein,
mutation of the critical glutamic acid (residue 173) to glutamine almost abolishes the
enzyme activity (Kuo, et al., 1998, Wang, et al., 1998). This mutation does not change
the Km for either histone tail or acetyl CoA substrate, but Kcat drops nearly 360 fold
(Tanner, et al., 1999). The crystal structure reveals the critical Glu residue in human
PCAF to be closely positioned to the e-amino group of H3 k14, suggesting its action of
abstracting a proton from the e-amino group (Clements, et al., 1999). Therefore, any
proteins (such as ADA2) which improve the deprontonation process will enhance GCN5
HAT activity.

On the other hand, the histone tails are notoriously non-structural random coils
(Luger, et al., 1997), whereas the crystal structure of PCAF HAT domain with bound
histone H3 tail illustrates that the histone H3 tail adapts a distinct path (Clements, et al., ).
This suggests that proteins such as ADA2 might help to load the histone tail to the
catalytic site of HAT domain to facilitate HAT enzyme activity. In support of this
suggestion, ADA2 is indeed observed to lower GCN5 Km for histone substrate (Boyer, et
al., 2002) (and my preliminary data in Figure 9).

In summary, substrate binding, direct attack at the s-amino group by acetyl moiety,
and products release could all be the possible mechanisms for regulating Arabidopsis
GCN5 activity on free histone substrate. These possible ways might be carried out by
ADA2 protein through direct or indirect modulating of the variable N- and C-termini of

GCN5 HAT domain.

95

 

As to the histone tail substrate from nucleosomes, another layer of possible
regulation may be carried out by Arabidopsis ADA2b. The histone tails are notoriously
resistant to being acetylated by GCN5 alone. One possible hindrance might be the
inaccessibility of histone tails that might caused by the interaction between the positive
tail and the negative DNA. Therefore, ADA2 protein might, as suggested by John Denu
and Craig Petersom (Boyer, et al., 2002), helps present histone tail to GCN5 for .

acetylation.

 

96

Chapter Five
The Function of the Unigue N-terminus of the Arabidopsis GCN5 Proteip

 

Introduction

The transcriptional coactivator protein GCN5, originally identified in yeast, has well
conserved homologs across broad phylogenetic distances, including Tetrahymena,
Drasaphila, mouse and human. Mammals have two closely related proteins GCN5 and
PCAF. They all contain a HAT (histone acetyltransferase) domain and a bromodomain.
An N-terminal region (about 400 amino acids) is highly conserved among mouse and
human GCN5/PCAF as well as Drasaphila GCN5, but the N-terminus is not present in
yeast GCN5. In contrast, Arabidopsis GCN5 has a unique N-terrninus (amino acids 1-
203), which shares no sequence similarity to that of human GCN5 (as illustrated in
Figure 1 in Chapter II).

GCN5 homologues from rice, maize and alfalfa also have N-terminal regions that are
similar in sequence to the N-terminus of Arabidopsis GCN5, suggesting some specific
function in plant world. One notable feature of the N-terminus of Arabidopsis GCN5 is
the repetitive stretches of serine residues (residue frequency 22 %).

The unique N-terminus of Arabidopsis GCN5 is not required for its HAT activity on
fi'ee core histones (Chapter II), and it does not interact with ADA2a or ADA2b (Chapter
III). Virtually no fimction of the N-terminus of Arabidopsis GCN5 has been known.
Intriguingly, the human and mouse GCN5 N-terminus interacts with CBP/p300 in vitra
and the Arabidopsis genome also encodes CBP/p300 homologues (Bordoli, et al., 2001,

Xu, et al., 1998, Yang, et al., 1996). The experiments described in this chapter were

97

 

undertaken to explore the potential function of the unique N-terminus ofArabidapsis

GCN5.

Exgrimental methods

The yeast two-hybrid gystem
The MATCHMAKER Two-Hybrid System 3 (Clontech) was used for yeast two-

hybrid screening. This system is Gal4-based, in which the bait gene is expressed as a
fusion to Gal4 DNA-binding domain. The system features yeast strain AH109, which
reduces the fi'equency of false positives by using three reporters—ADEZ, HIS3, and
MEL] . MEL] encodes a-galactosidase which is secreted into medium and turns X-a-gal
into blue color. The system also includes vector pGBKT7 for construction of bait plasmid
and vector pGADT7 for construction of prey plasmid.

Vector plasmid pGADT7 is derived from vector pACT, which is the vector used to
construct the two cDNA libraries used in the screening. pGADT7 and pACT both have
the 2 micron replication origin, leucine auxotrophic and ampicillin resistant markers, and
express the Gal4 activation domain driven by ADH] promoter. The major difference is
that pGADT7 had more restriction enzyme sites in the multiple cloning sites, integration
of HA epitope tag and incorporation of in vitra translation start sequence right after the
Gal4 activation domain. Vector pGBKT7 has tryptophan auxotrophic and kanamycin

resistant markers and expresses Gal4 DNA binding domain.

Construction of plasmids
To construct bait plasmid pGDBD-GCNS (1-568, the full length; or 1—210, the N-

98

terminus), cDNA fragments were inserted into plasmid pGBKT7 in frame with that of
Gal4 DNA binding domain. The cDNA fragment of the N-terminus of GCN5 was PCR
amplified first from relevant plasmid. The two bait plasmids were named as pGDBD-
GCN5 and pGDBD-GCNS (1-210).

To construct plasmid pGAD-MSI4 (full length, 1-507), pGAD-AMSI4 (missing the
50 amino acids in the beginning, 51-507), PCR fragment (Nca I - Sca I fragment) of
cDNA encoding MSI4 (aa l-70, with the first 50 amino acids encoded) or MSI4 (aa 51-
70, without the first 50 amino acids encoded) were sub-cloned into vector pGADT7,
together with the rest cDNA of MSI4 which was cut from candidate #133 plasmid (cut by
Sca I and Xha I). Both the PCR fragments in the constructed plasmids were DNA
sequenced to make sure no mutation was introduced. The encoding cDNAs in the
constructed plasmids were then cut out with Hg] 11 and Xha I and subcloned into vector
pGADT7 again which was cut with BamH I and Xha I. The reason for the second

subcloning was to introduce longer spacer between Gal4 activation domain and MSI4

proteins.

Western blotting of bait protein GCN5 11-210) expressed in yeast

Yeast AH] 09 transformed with plasmid pGBKT7 expressing Gal4 DNA binding
domain GDBD or pGDBD-GCN5(1-210) were grown in 5 ml liquid synthetic dropout
medium (SD Trp-) overnight, and then the 5 ml culture was transferred to fresh 40 ml SD
Trp— medium for another four hours. The yeast cells were collected by centrifugation, and
washed with chilled water. The cells were then mixed with 4-fold volume of glass beads

(Sigma,#8772-6, 425-600 pm, acid treated) and 3-fold cracking buffer (8M urea, 5%

99

SDS, 50mM Tris, pH6.8, lmM EDTA, 10% glycerol), and vortexed vigorously in the
cold room for about 8 min. The supematants from the vortexed cells were collected as
cell lysates. 5 pl of each sample was resolved in 10% SDS-PAGE gel, and the proteins in
gel were transferred to nitrocellulose membrane. The Gal4 and Gal4-GCN5 proteins were
detected by standard Western blotting using first antibody LA2-3(1 : 100000) (rabbit
antiserum). Antibody LA2-3 had been raised against antigen Gal4 DNA binding domain
in fusion with VP16 activation domain, therefore the antibody could detect Gal4 DNA-

binding domain.

mm

The first cDNA library was made using polyA RNA from 3-day-old etiolated
Arabidopsis (Kim, et al., 1997). Samples of this library as plasmids were provided by Dr.
Peter Quail.

The CD4-10 phage library (LACT) was provided by Arabidopsis Biological
Research Center at Ohio State University (ABRC order # 8485). This library was
constructed using random primed RNA from mature Arabidopsis leaves and roots. The
cDNA fragments bigger than 300 bp were inserted into the T-filled Xha I site of phage
vector LACT.

To convert the phage library to plasmid library, 0.5 ml E. cali strain BNN132 from
50 m1 LB/0.2% maltose overnight culture was transferred to 50m] LB/0.2% maltose and
cultured for three hours until OD600 reached 0.620. For BNN132, one OD600 contained
about 8.6 x 108 cell/ml. 30 ml culture of the above culture was spun down and

resuspended in 10 ml 10 mM MgSO4. About 3.2 x 109 E. cali cells were mixed with

100

2.4x108 PFU of CD4-10 library phage, and the mixture was incubated for 30 min at 30°C.
Liquid LB medium was then added to the infected cells to final volume of 6 ml. A 20 pl
sample was taken to estimate how many E. cali cells were infected, and the result was 2.6
x 107 cells were infected by the library phage (the original transformants were 9 x106
when the library was made). Then the phage infected cells were incubated at 30°C for 70
min with slow shaking. The cell culture was then spread on 24 plates (150 x 15mm) with
LB/0.2% glucose/Amp (50 pg/ml) and the plates were incubated for 22 hours at 37°C.

E. cali lawn was scraped from all the plates and the plates were washed twice with
LB/Amp (50pg/ml). All the E. cali cells were added into 6 liters of LB/Amp (50pg/ml)
and cultured for 5 hours. Library plasmid was prepared from the 6-liter culture by
plasmid DNA mega-preparation method. The library plasmid DNA was purified twice
with CsCl/ ethidium bromide gradient centrifugation. The library DNA was finally
resuspended in 3 ml TE (pH 7.5) after ethanol precipitation. The final DNA concentration

was lpg/pl. The library cDNA fragments sized from 300bp to 2kb.

Yeast cDNA library transformation
The transformation methods followed the protocols recommended by the

MATCHMAKER Two-Hybrid System 3. Yeast AH109 (Clontech) carrying bait plasmid
pGDBD-GCNS (1-210) was grown in 280 ml liquid SD Trp- medium for 23 hours
(OD600=1.18). The culture was transferred to 1000 ml YPDA medium (YPD medium
plus 0.002% adenine) and cultured for 3 hours and 10 min at 30°C (OD600=0.666). The
cells were collected by centrifugation and washed with 700 ml water. The cells were

suspended in 7 ml LiAc/TE buffer (100 mM LiAc, l x TE, pH 7.5). The cells were then

101

mixed with 100-200pg library DNA, 20 mg carrier DNA and 30 ml PEG/LiAc/TE buffer
(PEG, final concentration 40%, MW3350). The mixture was incubated for 30 min at
30°C with gentle rocking. Then 7 ml DMSO was added and mixed gently. The mixture
was incubated at 42°C for 20 min with gentle rocking occasionally. The mixture was then
shocked in ice for 10 min. The cells were spun down and resuspended in 5 ml YPDA
with 6 ml TE buffer. 25 pl was taken from the resuspension to plate on SD (Trp- Leu-)
solid medium plates to determine the transformation efficiency. The rest was spread on
solid medium of SD (Trp- Leu- His- Ade-). Totally about one million transformants were
obtained. The plates were incubated at 30°C for two weeks, and grown positive colonies
were streaked on SD (Trp- Leu- His- Ade-) plates. Then these positive colonies were
picked up and re-streaked on SD Trp- Leu- His- Ade-lX-a-gal plates to isolate single
colony.

The subsequent library transformation for low stringency screening, for bait GCN5

in full length or for CD4-10 library followed the same protocol.

Western blotting to check if M814 has the 50 amino acids extension

To construct in vitra transcription/translation plasmids, PCR fragment (Nca I-Sca I
fragment) of cDNA encoding MSI4 (aa 1-70, with the first 50 amino acids encoded) or
MSI4 (a 51-70, without the first 50 amino acids encoded) was sub-cloned into vector
pET-28(b) (Novagen), together with the rest cDNA of MSI4 which was cut from
candidate #133 plasmid by Sca I and Xha I. The PCR fragments were inserted directly
into Nca I site in vector pET-28(b) which is right after the ribosomal binding sequence,

so that His and T7 tags were eliminated to make sure that the translated products only

102

contained MSI4 protein. Both the PCR fragments in the constructed plasmids were DNA
sequenced later to make sure no mutation was introduced.

The in vitra translated proteins were made by using TNT T7 Quick Coupled
Transcription/Translation system (Promega) and the constructed plasmids. The cell lysate
of Arabidopsis leaf suspension cell culture was prepared as Chapter II.

For performing Western blotting, 1 pl or 4 p1 cell lysate 1.25 pl from translated TNT
system were resolved in 10% SDS-PAGE gel and transferred to nitrocellulose membrane.
The M814 proteins were detected by standard Western blotting method. The first
antibody was provided by Dr. William Folk at the Department of Biochemistry in
University of Missouri-Columbia The antiserum was raised in rabbit against antigen

MSIA4 (with the first 50 amino acids missing) and affinity purified.

Pr_ey plasmid (p’bragy cDNA plasmid) DNA preparation

Positive colony candidates were picked up from the initial screening plates. They
were all patch-streaked on solid medium SD (Trp-Leu-His-Ade-) plates. Then each
grown candidate was streaked again on SD (Trp-Leu-His-Ade-) plates to isolate single
colony. Single colony from each streaked candidate was grown in liquid medium SD
(Trp- Leu- His- Ade-), and the plasmid was extracted from the liquid culture. For plasmid
extraction, the yeast cells was spun down from 1.5 ml overnight culture, and then the
cells were mixed with 200 pl QAD buffer (2% Triton X-100, 1%SDS, 100 mM NaCl, 10
mM Tris-HCl, pH 8.0, 1 mM EDTA), about 100 pl glass beads and 100 pl
phenol/chloroform. The cell mixture was vortexed for 2 min. The supernatant was

collected after 5 min centrifugation at highest speed on bench centrifuge. Plasmid DNA

103

 

was precipitated by 2 x volume of ethanol, and dissolved in 70 pl 1x TE.

To recover the prey plasmid, E. cali (DHSa strain) was transformed by the plasmid
extraction obtained above and grown on solid LB/Ampicillin plate. Single colony from
solid LB/Ampicillin plate was grown in 5 ml liquid LB/Ampicillin medium. The prey

plasmid was then nriniprepared from the E. cali culture.

Confirmation of positive candidates
To confirm the candidates, AH109 yeast cells were transformed again with the prey

plasmid and empty plasmid pGBKT7 (only the Gal4 DNA-binding domain) or bait
plasmid. The transformed yeast cells were grown on SD Trp- Leu- solid medium first to
select positive transformants and then the grown colonies were streaked onto SD Trp-

Leu- His- Ade- IX-a-gal solid medium and allowed to grow for 4 days at 30°C.

Yeast cell plasmid transformation (mini-scale)

All the procedure was carried out at room temperature unless indicated. AH109 cells
were grown in 10 ml liquid YPDA medium overnight at 30°C. The YPDA medium was
YPD medium (20 g/l Difco peptone and 10 g/l yeast extract) supplemented with 0.003%
adenine hemisulfate (Sigma #A-9126). The overnight culture was then transferred to
fresh 60 ml YPDA for about 3 hours until OD600 reached about 0.6. The cells were then
spun down at 1000 x g and washed with 20 ml H20. The cell then was resuspended in 1x
TE buffer (10 mM Tris, 1 mM EDTA, pH 7.5, 100 mM LiAc). About 0.5 pg DNA each
of prey or/and bait plasmid was mixed gently with 100 pg Herring testes carrier DNA,

and then was vortexed together with the 100 pl prepared yeast suspension. 600 pl

104

PEG/LiAc/TE buffer (TE buffer plus 40% PEG MW4000) was then added to the cell
mixture by vortexing. The mixture was incubated at 30°C for 30 min at gentle shaking.

70 pl DMSO was added to the cell mixture and gentle inversion was performed to mix
DMSO evenly. The cell mixture was then heat shocked at 42°C water bath for 15 min
and then cold shocked on ice for 5 min. The cells were the recovered by brief
centrifugation and resuspended in 200-300 pl 1x TE buffer. About 150 pl suspended cells

were spread on appropriate selective solid medium.

GST pull-down assay.
The detailed procedure was described in chapter I].

Results

Yeast two-hybrid screening for proteins interacting with GCN5 (1-210)

To understand the function of the N-terminus of Arabidopsis GCN5, one convenient
way is to identify proteins interacting with the N-terminus. The yeast two-hybrid
screening strategy was employed in search for interacting proteins with the N-terminus.

The MATCHMAKER Two-Hybrid System 3 (Clontech) was used for yeast two-
hybrid screening. This system is based on the Gal4 activation domain and DNA-binding
domain. The system includes bait plasmid pGBKT7 (which encodes the Gal4 DNA
binding domain GDBD, and carries the Trpl selective marker) and prey plasmid
pGADT7 (which encodes the Gal4 activation domain GAD, and carries the Leu2 marker).
The yeast host strain AH]09 has four auxotrophic mutations (tryptophan, leucine,

histidine and adenine), and carries three Gal4 responsive integrated reporter genes to

105

detect bait-prey protein interaction: ADEZ, HIS3, and MEL] . MEL] encodes or-
galactosidase which is secreted into the medium and turned X-a-gal into blue color.
These three selection criteria have an advantage of eliminating most false positive
interactions between bait and prey proteins.

The cDNA encoding the amino acids 1-210 of the N-terminus of GCN5 was inserted
into the plasmid pGBKT7 in frame with that of GDBD, creating the bait plasmid
pGDBD-GCNS (1-210). The cDNA library was provided by Dr. Peter Quail at the
Department of Plant and Microbial Biology, University of California, Berkeley. The
cDNA library was derived from 3-day-old etiolated Arabidopsis seedlings, in which the
first strand cDNA was synthesized using oligo (dT) primer and afterwards the library
cDNA was inserted into plasmid pACT. Plasmid pACT was an early version of plasmid
pGADT7. The major difference is that GADT7 has more restriction enzyme choices at
the multiple cloning sites and that there is a HA epitope tag at the end of Gal4 activation
domain.

Before starting the two-hybrid screening, the bait protein GCN5(1-210) was first
checked in host AHA109 for its authentic expression, non-toxicity to host strain, and
incapability to activate reporter genes. Strain AH109 was transformed with plasmid
pGDBD-GCNS (1-210) to check the expression of bait protein. The transformed cells
were grown in liquid synthetic dropout medium lacking tryptophan minus (SD Trp-)
overnight, and then the cells were lysed using glass beads. Standard Western blotting was
performed to detect bait protein using the Gal4 DNA binding domain specific antibody,
LA2-3. Control cell lysate from AH109 cells transformed with the empty plasmid

pGBKT7 was also included in the Western blotting experiment to differentiate bait

106

protein size. As shown in Figure l, the Western blotting result showed that bait protein
GCN5 (1-210) in fusion with Gal4 DNA-binding protein was expressed in host strain of
AH109.

To test whether the bait protein GDBD-GCN5(1-210) could not activate reporter
genes, yeast cells transformed with bait plasmid pGDBD-GCN5(1-210) were grown on
solid medium SD (Trp-, SD Trp- His-) and SD (Trp- Ade-). The transformed yeast cells
grew only on SD (Trp-), but not SD (Trp- His-) and SD (Trp- Ade-) solid media after 4
days incubation at 30°C (Figure 2). This result confirmed that GCN5 (1-210) could not
activate reporter genes. On the other hand, yeast cells with plasmid pGDBD-GCNS (1-
210) on SD (Trp-) solid medium appeared similar in size to that of yeast colonies with
plasmid pGBKT7 on SD (Trp-) solid medium, confirming that the bait protein was not
toxic to the yeast cells (Figure 2).

After it was clear that bait protein GCN5 (1-210) was not toxic, nor did it activate
reporter genes in yeast host strain AH109, yeast two-hybrid screening was performed.
The yeast AH109 cells, with the bait plasmid pGDBD-GCN 5(1 -210), was transformed
with the cDNA library following the protocol of library transformation recommended for
the MATCHMAKER Two-Hybrid System 3. The transformants were spread on solid
medium of SD (Trp- Leu- His- Ade-), in which markers Trp- Leu- were to keep the prey
and bait plasmids and markers His- Ade- were to select positive candidates. A small
portion of transformed cells was spread on non-selective solid medium SD (Trp- Leu-) to
estimate the total transformants. About one million transformed yeast cells were screened
on selective solid medium. The cells on medium SD (Trp- Leu- His- Ade-) were allowed

to grow for two weeks at 30°C.

107

kDa

I ., 68

r g, .3

GDBD-GCN5 (1.210) ‘ -' .
' w 43

r1 .
' 29

GDBD >4.-

Figure 1. Bait protein GDBD-GCN5 (1-210) was expressed in yeast host
strain AH109.

Cell lysates fiom yeast cells transformed with plasmid pGBKT7 or plasmid
pGDBD—GCN5( 1-2 I 0) were resolved in 10% SDS-PAGE gel, and the proteins
in gel were transferred to nitrocellulose membrane. The protein GDBD and
fusion protein GDBD-GCN5 (1-210) were detected by Western blotting using
rabbit antiserum LA2-3 which had been raised against antigen GDBD in fusion
with VP16 activation domain. The positions of both GDBD and pGDBD-GCN 5
(1-210) are indicated by the filled triangles. GDBD: Gal4 DNA binding domain.

108

 
  
   

SD (Trp- Ade-) /

l ' .
pGDBD- pGBKT7-53
GCN5(1-210) pGADT7-T
SD (Trp- His-)

H pGDBD- pGBKT7-53
GCN5(1'210) pGADT7-T

 

pGDBD-
GCN5(1-210)

Figure 2. The protein GCN5(1-210) was not toxic, and did not activate

reporter genes in yeast host strain AH109.

The transformed yeast cells with respective plasmids were grown on solid

SD medium of Trp-, or Trp- Ade-, or Trp- His- for 4 days at 30°C. Plasmid
pGDBD-GCN5(l-210) expressed bait protein GCN5 (amino acids 1-210) in
fusion with Gal4 DNA binding domain (GDBD), plasmid pGBKT7 expressed
GDBD only, and plasmid pair pGBKT7-53 and pGADT7-LT, as positive
control, expressed p53 protein and SV40 large T antigen. Trp- medium was to

maintain plasmids pGBKT7 and pGDBD-GCNS which had a Trpl+ marker.

109

Since both selective markers His- and Ade- were included, the screening was under
high stringent selection. Totally 102 positive candidate colonies were picked up from the
selective medium during the 2-week incubation period.

The cDNA prey plasmids were recovered from the positive yeast colonies. First, the
plasmids (including bait and cDNA prey plasmids) were prepared from the 102 candidate
yeast cells by standard plasmid extraction method. Then, E. cali cells were transformed
with the prepared plasmids and were grown to single colony on solid LB medium with
ampicillin to select cDNA prey plasmid. The cDNA prey plasmids were prepared from
the transformed E. cali by minipreparation.

To ensure that the cDNA encoded prey protein interacted only with bait protein
GCN5 (1-210) but not the Gal4 DNA binding domain, the prey plasmids were
transformed again to yeast AH109 cells which carried either plasmid pGBKT7 or bait
plasmid pGDBD-GCNS (1-210). The transformed yeast cells were grown on SD (T rp-
Leu-) solid medium first to keep the bait and prey plasmids and subsequently were
streaked on solid medium SD (Trp- Leu- His- Ade- IX-a-gal ) to test bait-prey interaction.
However, the test result showed that proteins from all 102 candidate prey plasmids
showed interaction with both the Gal4 DNA binding domain and GDBD-GCN5 bait.
This result indicated that either these 102 candidates interacted with GDBD only or they
alone could activate reporter genes. None of them had specific interaction with bait
GCN5 (1-210).

The screening markers above used two reporters (Ade- His-) in the initial screening.

This high screening stringency might have missed proteins which interacting weakly with

110

bait GCN5 (1-210). Therefore, a second round of yeast two-hybrid screening was
repeated with lower selective stringency using only the His- selective marker.

About 2 million yeast transformants were screened on SD (Trp- Leu- His-) solid
medium. Transformed yeast cells were allowed to grow for 10 days at 30°C. About 1500
positive colonies were picked up from the SD (Trp- Leu- His-) solid medium and patch-
streaked on SD (Trp- Leu- His- Ade- /X-a-Gal) solid medium. Apparently, lower
selective stringency retrieved more initial positive candidates. But only about 80
candidates grew well and turn blue color on high stringent selective medium SD (Trp-
Leu- His- Ade- /X-a-Gal). For these 80 candidates, prey plasmids were prepared from
the yeast cells, and they were checked, as done before, for the interaction with GAL4
DNA binding domain only. The result showed that all of the 80 candidates interacted
with both the Gal4 DNA binding domain and the bait protein GCN5 (1-210). This result
again indicated that no candidates interacting specifically with GCN5 (1-210) were

identified when the initial screening stringency was lowered.

Yeast two-hybrid screening for proteins interacting with full-length GCN5

The two genetic screens described above uncovered no potential positive candidate
when the N-terminus of GCN5 was used as the bait. One possible defect using the N-
terminus of GCN5 as the bait was that the N-terminus might not be in its native
conformation when the rest of GCN5 was absent. Therefore, the full length of GCN 5 was
employed as bait to repeat the yeast two-hybrid screening. There was another advantage
of using the full length GCN5 as bait. GCN5 in full length was known to interact with

proteins ADA2a and ADA2b as shown in chapter 111. ADA2a and ADA2b candidates

1]]

should be screened out from the library when GCN5 full length was used as the bait.
Therefore, successful recovery of ADA2a and ADA2b candidates from the cDNA library
served as positive control, in making sure that the two-hybrid screening had no systemic
problem.

The cDNA of full length GCN5 was inserted into plasmid pGBKT7 in frame with
that of Gal4 DNA binding domain to construct bait plasmid pGDBD-GCNS. As done
with the N-terminus above, the full length of GCN5 protein bait was checked first for
non-toxicity in the host strain as well as its incapability to activate reporter genes. Yeast
AH109 cells were transformed with bait plasmid pGDBD-GCNS and prey plasmid
pACTlor pACT2. These two prey plasmids were purified randomly from the cDNA
library with unknown library cDNA insertions. The transformed yeast cells, compared
with that of positive controls, grew well on SD Trp- Leu- solid medium after 4 days
culture at 30°C, confirming the bait GCN5 protein was not toxic to host yeast cells.
Meanwhile, the transformants failed to grow on selective medium of SD (Trp- Leu- His),
SD (Trp- Leu- Ade-), and SD (Trp- Leu- His -Ade-) solid medium while the positive
control did, confirming that GCN5 protein did not activate reporter genes (Figure 3).

After clarifying that bait GCN5 protein was neither toxic nor capable of activating
reporter genes, 3 third round of library screening was carried out. Yeast AH109 cells with
bait plasmid pGDB-GCNS were transformed with 200 pg library DNA. The transformed
yeast cells were spread on SD (Trp- Leu- His—) solid medium (low stringency) and
incubated for 13 days at 30°C. Totally about 1.8 million transformants were screened,
and about 550 initial positive candidates were picked up and patch-streaked on SD (Trp-

Leu- His- Ade- fX-a-Gal) solid medium to confirm their candidacy. Based on the criteria

112

 

      

SD (Ade- His— Leu- Trp-) SD (Leu- Trp-)

Figure 3. The protein GCN5 (full length) was not toxic, and did not activate
reporter genes in yeast host strain AH109.

Yeast AH109 cells co-transformed with respective plasmids were grown on on the
medium as indicated at 30 °C for 4 days before the pictures were taken. 1: pGDBD-
GCN5+pACT1, 2: pGDBD-GCN5+pACT‘2, 3: pGBKT7-p53+pGADT7-LT, 4:
pGBKT7+pACT1. Bait plasmid pGDBD-GCNS expressed bait protein GCN5 in
fusion with 0314 DNA binding domain (GDBD). Prey plasmid pACTland pACT2
were purified randomly fiom the cDNA library, which were empty pACT vectors
with possible unknown library cDNA insertions, therefore expressed Gal4 activation
domain GAD in fusion with possible unknown protein. The plasmids pGBKT7-p53
(expressing p53 protein) and pGADT7-LT (expressing SV40 large T antigen)

were positive control, and plasmids pGBKT7 (expressing GDBD) and pACTl were
negative control.

113

of good colony growth and capability of turning blue color on SD (Trp- Leu- His- Ade-
/X- or -Gal) medium, totally 156 potential candidates were recovered.

These156 potential candidates were colony-streaked again on SD Trp- Leu- Ade-
His- fX-or-gal solid medium to isolate single colonies for the preparation of prey plasmids
as done for that of bait GCN5 (1-210).

To confirm specific interaction with GCN5, the prey plasmids were prepared as
described above and co-transformed into yeast AH109 cells again with empty plasmid
pGBKT7 (expressing Gal4 DNA-binding domain only) or bait plasmid pGDBD-GCN5
(expressing GDBD-GCN5). The transformed yeast cells were streaked on SD (Trp- Leu-
Ade- His- /X- a-gal) solid medium to confirm real candidacy by the criterion of colony
growth with bait plasmid and no colony growth with empty plasmid after 4-day
incubation at 30°C. Out of these 156 candidates, 28 candidates showed specific
interaction with GCN 5 bait only, the rest showed either interaction with both GCN5 bait
and GDBD or no interaction with either GCN5 bait or GDBD.

Among these 28 candidates, candidates #2, 3, 4, 5, 7, 8, 9, 10, 11, 13, 16, 20, 26, 30,
73, 77, 104 and 133 showed big-size-colony growth and bright blue colonies with bait
GCN5 (some of them were shown on Figure 4). Candidates #8], 145, 137, 141, 155, and
161 showed big-size-colony growth and moderate blue colonies with bait GCN5.
Candidates 82 showed moderate colony gowth and moderate blue colonies with bait.
Candidate #38, 46, 106 showed small colony growth and little blue colonies with bait.

DNA sequencing indicated that candidates #2, 3, 4, 5, 7, 9, 10, 11, 13, 16, 20, 26, 30,
104, were cDNAs of ADA2b, and # 8 was that of ADA2a. Details of these cDNAs

related to the full length of Arabidopsis ADA2a or ADA2b are shown in Figure 5. This

114

 

10

Figure 4. Confirmation of the candidates from the two-hybrid screening
using the full-length GCN5 bait.

These candidate plasmids (as indicated by the numbers in the pictures) were
co-transformed to yeast strain AH109 cells with plasmid pGDBD-GCN5
(expressing GCN5 full length in fusion with Gal4 DNA binding domain),
pGDBD-GCNS (1 -210) (expressing the N-terrninus of GCN5 in fusion with
Gal4 DNA binding domain) or pGBKT7 (expressing Gal4 DNA binding
domain), and the transformed cells were grown on selective solid medium

SD Ade- His -Leu- Trp- /x-a-Gal for 4 days at 30°C. The candidates numbers
are colored differently to indicated the co-transformed plasmids where the red
color is for pGDBD-GCNS (1-210), blue color for pGDBD-GCNS, and black
color for pGBKT7. Candidate #74 did not interacted with GCN5 and Gal4
DNA binding domain, and #104 interacted only with the Gal4 DNA binding
domain. 53-T was a positive control of p53 protein and SV40 large T antigen.

115

1 160 225 _ 408486

 

 

 

 

 

 

 

 

 

Figure 5. The isolated ADA2a and ADA2b candidates by GCN5 (fl) bait
in yeast two hybrid screening.

Each line represents the ADA2a or ADA2b isolate from yeast two-hybrid
screening with the GCN5 firll length as the bait. The length of each line
indicates the relative region encoded by the candidate cDNA in respective

of the fill] length of protein ADA2a or ADA2b. The numbers are the amino
acid positions for respective proteins. The numbers in the brackets indicate
the number of isolates of each prey plasmid. The conserved N-terminus, the
middle part and the C-terminus of ADAZa and ADA2b protein are represented
by the shaded areas inside the two bars.

116

result illustrates that the performed yeast two-hybrid screening was technically working.

Candidates #38, 46, 73, 77, 81, 106, 133, 141, 145, 161 were checked again for their
specific interaction with GCN5. The candidate plasmids were co-transformed into yeast
AH109 cells with plasmid pGDBD-GCNS, pGDBD-GCNS (1-210) and pGBKT7. The
transformants were grown on SD (Trp- Leu- Ade- His- /X- a -gal) solid medium. They
showed interaction only with GCN5 in full length. None of them interacted with the N-
terminus of GCN5 (Figure 4).

These remaining candidates were DNA sequenced. A summary of these candidates is
listed in Table 5-1, including the putative proteins and their functions flom Blast search
of GenBank. Among all the candidates, candidates #133 and #81 were most interesting in

considering the apparent frmction of Arabidopsis GCN 5 as histone tail modifier.

Candidate #133

The #133 cDNA encoded a 507 amino acid protein. The cDNA contained the whole
open reading flame as well as the polyA tail and upstream untranslated 5’ sequences. The
protein encoded by candidate #133 cDNA was identical to the Arabidopsis protein named
MSI4 (Kenzior and Folk, 1998) except that #133 had additional 50 amino acids at its N-
terrninus. Arabidopsis MSI4 is homologous to human protein RbAp46/48 (for Rb
associated protein), which belongs to the Msil-like family with a featured domain called
WD-40.

Many Msil-like proteins are components of complexes involved in the assembly and
modification of chromatin (V igrrali, et al., 2000). Msil, RbAp48 and related the related

protein p55 are components of chromatin assembly complexes of yeast, humans and

117

U 505:8:
8:50:50 0:: :0 :8m 8: 0:: 05:88

.nZUO :55 5580885 8:
8088: 805 “20 :e 80 «.230 .m 5: 8:88 cm; 8 ”a .mZUO :55 558885 .8: .0888: 805 “1:0:0 80 «.290

.m 5: 8:88 on: 8 ”0:. 8580.88 0:0m 8:089: :8 .800: <2: 88858.: 80:88.: 058888500508 58885:
0808:: 550:8 <58 8: 8588088.: 850:: w5::_05 85—3808 <2: :0 88:8 88.85 5 8395 08 880:0:
:8: 9.55 8E. .858 550:: w5:55:: 5 8235 80 880:0: HD 80:0 80: 8:88: 8:88-: a m558:8 :8: 80:
8:80 8 5 88.085 .8888: 05.8 53.9 05 5 :::o: 85 0:: 6.5.: 5588 53:888. :0 :05850 58080.88 :8
5::0::0:0:5 050: 8 80::0 505. .8888: 058m :0 .588: 0 80 88:85 5:88: 08:85 ”m :80: :00 :08888:
w58: 88:5 0: 8 :::o: :00: 8: 55: 88:: :0 08208:: 8088 < 8.88:8 80:0::::m 05:80: 5:55 8:888:00
58:58:88: :8 08888—8: 05 5:55 85:00:: 5:058:08 .588: 850:: a 0: 8 850:: 5 5.5: .80: :8 :05: 0:: 5
.5888 :88 888:8 5 5 :m:o::0 5:0: 88:: 0:: 5 8:8: :8 5 888:8 0: 8 8505: 5 588: 5:: .8: m5:00 0::
8:: E5058: 5 55: :o 8:08: 0::. 88858: :8 88880 :8: 88: 588: 55: :0 85800 5:88: 5: :. .55: 2..

 

a: 8888 8:8: .5 8:8 as :o a an .888... 28888-: a 238 N38 8..
83562 :88 68658... .88.: 8-8% 288.088-838-988?m 8 RE: 888 8;. .8:
m 6888 .8 632:8 3888 8 82 see: .8858»: a 8:18» 8285 2;. .2: a
.8888 .8: 5.8 :8
Dam—O 0:: 0:305 5:8: 55 :0 80:80:: 8 Sm .0588: 680:0:
888:8 .8: 5.8.98: .98: .85 «.2822: 2888.828... 8 535 $85.: 3.5
5:558: 0808888 :8 :555:
:68 88.5. 888:8 656888: 8 8m .3828 a 2818 8.8.8“: 8;.
882.: a: $888 8&2 3.95 .8 Sn .8: a .8 88 5889.. mm;
3 8 8:8 :3 :82: .888: a 8.8.8 8:82 8;.
:OEU 588: 850:0 88:: .580
68858 888 88: 028 888: a. E. a. a 8m); 8888 :8
5888 8:83.88 88.82.82 8 2m .5825 :8: a m _ E: 3:85 E
van: 8855 a 88mm 8882 E
<8: 3 an .522: 88-8: a 538 on: 82 as.
on: : 8:4

508:: .8 588: :85?—

05n .588: «.5580

8:880 0 55:58 a 5:85:09 082::8

 

8:82:80 0:. :e 8:88:m Tm 038,—.

118

Drosophila respectively. The Drosophila p55 protein also is a subunit of NURF, an ATP-
dependent nucleosome remodeling complex. In humans, Rbap48 is a retinoblastoma
interacting protein that acts as an apparent histone H4 chaperone (Roth and Allis, 1996)
and a subunit of human chromatin assembly factor CAF -1(V erreault, et al., 1996).

As mentioned, candidate #133 had additional 50 more amino acids at its N-terminus
compared to the reported Arabidopsis MSI4 (Kenzior and Folk, 1998). The first 60 amino
acids of candidate #133 were MESDEAAAVS PQA'ITPSGGT GASGPKKRGR
KPKTKEDSQT PSSQQQSDVK MKESGKKTQQ.

To test whether the endogenous MSI4 had the 50 amino acids extension, MSI4
protein from Arabidopsis suspension cell lysate was compared to that of in vitro
translated MSI4 with or without this 50 amino acids extension by Western blotting. To
construct in vitro expression plasmids, PCR products from candidate #133 encoding
protein MSI4 with or without the 50 amino acids were subcloned into His-tag expressing
vector pET28 where the His tag was removed so that the expressed protein contained no
extra amino acids. The two proteins were expressed in vitro using the constructed
plasmids and the TNT T7 transcription/translation system (Promega). The Arabidopsis
suspension cell lysate was prepared previously (Chapter II). The Western blotting result,
using MSI4 specific antibody, showed that endogenous MSI4 did have this 50 amino
acids extension (Figure 6A), where protein size of endogenous MSI4 was the same as
that of in vitro translated MSI4 with the 50 amino acids. The protein size of in vitro
translated MSI4 without the 50 amino acids was smaller than that of endogenous MSI4.

Candidate #133 cDNA included 60 nucleotides 5’ to the start ATG codon of MSI4.

These 60 nucleotides encoded 20 amino acids in frame following the cDNA of Gal4

119

200 -
98 -
6... my."
43 -
28 .. "“ a

l 51 507

Gal4
activation

domain

 

 

Figure 6. Native protein ofArabidopsis MSI4 contains the 50 amino
acids in its N-terminus.

(A) Western blotting was carried out to confirm that the native MSI4 protein
had the 50 amino acids extension. The native endogenous MSI4 protein was
from the Arabidopsis leaf suspension cell lysates. The in vitro MSI4 proteins
(amino acids 1-507 or 51-507) were made in TNT T7 Quick Coupled Trans-
cription/Translation system (Promega) using their respective expressing
plasmids. The protein samples were resolved in 7.5% SDS-PAGE gel and the
proteins were detected by Western blotting using specific antibody against
MSI4 (51-107). 1: In] of Arabidopsis leaf suspensin cell lysate, 2: 4n] of
Arabidopsis leaf suspensin cell lysate, 3: in vitro translated MSI4 protein (a
51-507), 4: in vitro translated MSI4 protein (1-507). The blue arrow points
to the MSI4 proteins. (B) The diagram of candidate #133 prey plasmid.
Various encoding regions are indicated in the plasmid, including the regions
encoding the first 50 amino acids of native MSI4 protein (gray area) and the
extra leading 20 amino acids (blue area). The numbers refer to the amino acids
positions in the MSI4 protein.

120

activation domain (Figure 6B). Therefore, it was legitimate to determine if these 20
amino acids were necessary for the interaction with GCN5 bait protein.

To test this necessity, a yeast two-hybrid assay was performed. The cDNA of MSI4
(full length or missing the first 50 amino acids) were subcloned into yeast prey vector
pGADT7 in frame with that of the Gal4 activation domain. Yeast AH109 cells were
transformed by bait plasmid pGDBD-GCNS and respective constructed MSI4 plasmids,
and grown on selective solid medium SD (Trp- Leu- His- Ade- /X-a-gal) for 4 day at
30°C. Surprisingly, as shown in Figure 7, candidate #133 lost interaction with GCN5
when these 20 amino acids were not present. The control #133 with these 20 amino acids
still showed interaction with GCN5. Therefore the 20 amino acids were necessary for the
interaction with protein GCN5, and very likely that protein MSI4 does not interact with
GCN5, although the MSI4 protein expression in yeast cells was not checked. The 20
amino acids sequence were RYRDTESKDQKRKIKKRERK. (Note: The 5 amino acids
encoded by the adaptor sequence for Xho I, which was used to construct the cDNA

library, were not included in these 20 amino acids.)

Candidate #8]

Based on the experience with candidate #133, other were checked for the existence
of extra 5’ sequence. Only candidate #8] was found to have S’untranslated sequence,
which encoded 41 amino acids in frame with Gal4 activation domain and prior to the
legitimate start codon.

As for candidate 133, a yeast two-hybrid assay was performed to check the

involvement of these 41 amino acids interaction with GCN5. Indeed these 40 amino acids

121

' "fit?“ a
ll
ya

      

Leu-Trp-Ade-His- IX-a-gal Leu-Trp-

Figure 7. Candidate #133 lost interaction with GCN5 when the extra leading

20 amino acids were deleted.

Yeast two-hybrid assay was performed to check the interaction of GCN5 with the
different lengths of candidate #133. Yeast AH l 09 cells were co-transformed with
bait plasmid pGDBD-GCNS and the respective constructed #133 candidate plasmids.
The transformed yeast cells were grown on selective solid medium SD (Trp- Leu-
His- Ade- /X-a-gal) for 4 day at 30°C.The transformed cells were also grown on non-
selective solid medium SD (Trp- Leu-)to confirm proteins were not toxic to the host
cells.

1: pGAD-ADA2b+pGDBD-GCN5 (positive control),

2: pGAD-MSl4 (l-507)+pGDBD-GCN5,

3: pGAD-MSI4 (SO-507)+pGDBD-GCN5.

4: #133+pGDBD-GCN5,

#133: the original prey plasmid fi'om the library screening.

l22

were necessary for interaction with GCN5 bait protein (data not shown), and candidate
#81 also did not interact with GCN5 legitimately. The 41 amino acids were
PRDLPPNPTTTI‘CIKKKIYSLSFFDQYIRQGSRQSPQHLHH.

In conclusion, candidate #133 and #81 were false positives. Curiously, there is a
common sequence in the cDNA encoded extra peptide of candidate #133 and #81: IKKK
in #81 and IKKR in #133. It is not known if this common sequence is responsible for the

interaction with GCN5 bait protein. However, none of the other candidates have this

common sequence. But curiously, Arabidopsis ADA2a does have IKKK at its C-terminus.

Candidate #77

Candidate #77 encodes a putative bHLH transcriptional activator. The full length
cDNA in GenBank encodes a protein of 315 amino acids (termed as 77fl). The candidate
isolated by the two-hybrid screen encoded residues 29-315.

To test which part of GCN5, except the N-terminus, interacted with protein 77fl, an
in vitro GST pull-down assay was performed. Recombinant GST fusion proteins of
various GCN5 parts bound to Sepharose beads were tested to pull down protein 771]
which was in vitro translated and radiolabeled by [8”] methionine. As shown in Figure 8,
protein 77f] showed interaction with GCN5, the HAT domain and the H-B region but not
the bromodomain. A yeast two-hybrid assay also was performed to test various GCN5
parts’ interaction with 77fl, and two overlapping regions 3 of GCN5 (1-250, and the HAT
domain) were found to interact with 77fl (Figure 10). These results suggested that the
HAT domain of GCN5 interacted with 7711.

Since 77fl had a putative DND binding domain, it was also tested for being a

123

61 5 23 4 12345
'0
.9
’9. “.I. M 3"
V b
' a“
M In

Figure 8. The full length of candidate #77 protein interacted with the
HAT domain and the the H-B region of GCN5.

An in vitro GST pull-down assays were performed. Various recombinant
GST-GCN5 proteins bound to Sepharose beads were tested to pull down
candidate #77 in full length which was translated in vitro and radiolabeled

by [S35] methionine using the TNT T7 Translation/Transcription Couple
system (Promega). The pulled-down proteins were eluted fi'om the beads
by boiling in SDS-PAGE loading buffer. Then the eluates were electropho-
resed in 10% SDS-PAGE. The gel was then dried to expose to X-ray film
for autoradiogram, as shown in (A). The amount of GST-GCN5 proteins
as well as control GST protein used in each pull-down assays were shown
in (B) that were stained with Coomassie Blue. The arrow points to the #77
protein in full length. 1: GST-GCN5, 2: GST-GCN5 (HAT, 203-384), 3:
GST-GCN5 (H-B, 360-486),4: GST-GCN5 (Br, 451-568), 5: GST,

6: 50% input.

124

transcriptional activator. Yeast one-hybrid assay was performed where truncated versions
of 77fl were fused to 6314 DNA binding domain to test their potentiality activating
reporter genes. As shown in Figure 9, protein 77 could activate reporter genes in yeast
one-hybrid assays. The N-terminus (1-82) was sufficient to activate reporter genes. This
indicated that the N-terminus contained an activation domain. Based on yeast colony
growth, the firll length and the part including the helix-loop-helix DNA-binding domain
(1-173) did not activate as well as that of N-terminus (1—82). This might be attributed to
the variation of protein expression level, the presence of repression domain(s), or that the
DNA binding domain sequestered the expressed protein elsewhere in the yeast genome.

Further research of candidate 77 was not pursued.

Screening library CD4-10 using GCN5 (full length) as the bait

After these three rounds of two-hybrid screening, no potential proteins interacting
with the N-terminus of GCN5 were identified. The mRNAs to make the cDNA library
used above were made from 3-day-old etiolated Arabidopsis seedlings, and the first
strand cDNA was synthesized using oligo-T primer. This library might not include
cDNAs for some genes in other growth conditions, such as light which is very important
for plant growth. Therefore, a second cDNA library called CD4-10 was screened. This
library was derived from adult Arabidopsis plant roots and leaves grown in light, and the
first strand cDNA was synthesized using random primers. The phage library was
provided by Arabidopsis Biological Research Center at Ohio State University. The
plasmid cDNA library was prepared. The plasmid vector was pACT, the same as that of

the cDNA library used above.

125

A Leu-Trp-Ade-His-lX—a-gal

 

Figure 9. Candidate #77 was a potential activator and cinteracted with

the HAT domain GCN5. It is a potential activator.

The candidate #77 cDNA encoded amino acids 29-315 of the full length protein
of amino acids 1-315. The full length cDNA was cloned from the screened
library by PCR, and various cDNA lengths as indicated were subcloned into
vector pGBKT7 or vector pGADT7. Yeast AH 109 cells co-transformed with
respective plasmids were grown on selective solid medium SD (Trp- Leu- His-
Ade- IX-a-gal) for 4 day at 30°C. The transformed cells were also grown on
non-selective solid medium SD (Trp- Leu-) to check the expressed proteins
were not toxic to the host cells. 1 , 2, 3, 4, 5, and 6 tested the interaction of GCN5
with candidate #77, and 7, 8, 9 and 10 tested candidate #77 as a potential trans-
criptional activator where the co-transformation of plasmid pGADT7 (carried
the marker of Leu2+) was to make the transformed yeast cells grow on Leu-
solid medium. 1: pGAD-77(FL)+pGDBD-GCN5(bromo), 2: pGAD-77(FL)+
pGDBD-GCN5(230-468), 3: pGAD-77(FL)+pGDBD-GCN5(HAT, 203-384),
4: pGAD-77(FL)+pGDBD-GCN5(1-250), 5: pGAD-77(FL)+pGDBD-GCN5
(l-210), 6: pGAD-77(FL)+pGDBD-GCN5(H-B, 360-486), 7: pGADT7+
pGDBD-77(F L), 8: pGADT7+pGDBD-77(l-l73), 9: pGADT7+pGDBD-77
(1-82), l0: pGADT7+pGBKT7

126

Similar yeast two-hybrid screening was carried out. Yeast AH109 cells with bait
plasmid pGDB-GCNS were transformed with 600 ug CD4-lO library DNA. The
transformed yeast cells were spread on SD (Trp- Leu- His-) solid medium and incubated
for 11 days at 30°C to identify candidates. Totally about 2 million transformants were
screened, and about 260 initial positive candidates were picked up and patch-streaked on
SD (Trp- Leu- His- Ade- /X-a-Gal) solid medium. Totally 58 potential candidates were
recovered from the second screening of patch-streaking based on their good growth and
turning to blue color.

These 58 potential candidates were colony-streaked again on SD (Trp- Leu- Ade-
His— IX-a-Gal) solid medium to isolate single colony, and prey plasmids were then
prepared as described before. To confirm these candidates’ specific interaction with
GCN5 bait, the prepared prey plasmids were co-transformed into yeast AH109 cells
again with empty plasmid pGBKT (expressing Gal4 DNA-binding domain, GDBD) or
bait plasmid pGDBD-GCNS (expressing GDBD-GCN5). The transformed yeast cells
were streaked on SD (Trp. Leu— Ade- His- /X-a-Gal) solid medium. Out of these 5 8
candidates, all showed interaction with both the GCN5 bait and GDBD. This result

illustrated none of the 58 candidates was real.

Testing interaction with PCAT3
Arabidopsis has four CBP/p300 orthologs namely PCATl, PCAT2, PCAT3 and

PCAT4 (Bordoli, et al., 2001). These PCATs have about 600 amino acids at their C-
terminus highly homologous to the C-terminus of CBP/300 (amino acids 1201-1807),

including the C/H2 and C/H3 domains. The corresponding C/HZ and C/H3 regions are

127

amino acids 1078-1350 and 1596-1743 in PCATl, and 986-1256 and 1507-1654 in
PACT2. The homologous 600 amino acids region in PCAT2, like that of CBP/p300, was
shown to have HAT activity and ElA binding prOperty (Bordoli, et al., 2001).

As mentioned in the introduction, the N-terminus of mammalian GCN5 interacted
with CBP/p300 in vitro. This led to the hypothesis that the N-terminus of Arabidopsis
GCN5, although very different from its animal counterpart in sequence, interacts with
PCATs. Yeast two-hybrid assay was performed to test the possible interaction, in which
the full length of GCN5 was used as the bait.

The cDNA clones encompassing the full length of PCAT2 and C/HB domain of
PCATl were provided kindly by Dr. Richard Eckner from the Institute for Molecular
Biology, University of Zurich, Zurich, Switzerland. Various cDNA fi'agments were
amplified by PCR and inserted into vector pGADT7 in frame with the Gal4 activation
domain. These cDNA fragments encoded PCAT2 regions of 1-704, 611-994, 977-1362,
1323-1654, and the C/H3 domain of PCATl (amino acids 1588-1745).

Yeast AH109 cells were co-transformed with GCN5 bait plasmid pGDBD-GCNS
together with the plasmids containing various parts of PCATl or PCAT2. The
transformed cells were grown on SD (Trp- Leu- Ade- His— /X-a-Gal) solid selective
medium for 4 days at 30°C as well as non-selective medium SD (Trp- Leu-). As shown in

Figure 10, no interaction was found between GCN5 and any parts of PCATI or PCAT2.

Discussion

In summary, no proteins were identified that interact with the N-terminus of GCN5

protein through the strategy of yeast two-hybrid screening, leaving the function of GCN5

128

 

Figure 10. Testing the interaction of GCN5 with PCATl and PCAT2,
the Arabidopsis orthologs of CBP/p300.

Yeast AH109 cells were co-transformed with GCN5 bait plasmid pGDBD-
GCNS (fl) together with the plasmids expressing various parts of PCAT]
or PCAT2. The transformed cells were grown on SD (Trp- Leu- Ade- His-
/X-a-gal) solid selective medium for 4 days at 30°C as shown at the left,
as well as on solid medium SD (Trp- Leu—) as shown at the right.

: pGAD-PCAT1(C/H3) + pGDBD-GCNS,

' pGAD-PCAT2(1-704) + pGDBD-GCNS,

pGAD-PCAT2(61 1-994) + pGDBD-GCNS,

pGAD-PCAT2(977-l362) + pGDBD-GCNS,
pGAD-PCAT2(1323-1654) + pGDBD-GCNS,

pGAD-ADA2b(fl) + pGDBD-GCNS (positive control),

' pGADT7 + pGDBD-GCNS (negative control).

SQMAHP"

129

N-terminus an open question. The N-terminus of GCN5 also does not interact with the
CBP/P300-like proteins from Arabidopsis.

For the first cDNA library, the screening was saturated. First, the same ADA2b
candidates were isolated several times. Secondly, about 2 million yeast colonies were
screened, whereas Arabidopsis has about 25,000 genes. For the CD4-10 library, the
screening was very likely saturated but the library might not be comprehensive because
no ADA2 candidates have been fished out. Attempt to amplify ADA2 cDNA by PCR
directly from the library also failed (data not shown).

All the yeast two-hybrid screenings were carried out at temperature of 30°C. This
could be one reason for failing to detect interacting proteins with GCN5, because some
Arabidopsis proteins might not like this temperature in regard that the favorite ambient
temperature for Arabidopsis plants to grow is below this temperature. Meanwhile, if the
N-terminus is supposedly involved in plant response to abiotic and biotic stresses, the
targets’ cDNAs may not be present in the two cDNA library which were not made under
stressful growth conditions. This might also lead to the failure to identify interacting
protein with the N-terminus of GCN5.

Some function learned from the N-terminus of mammalian GCN5 might be
interesting to mention here. The mouse GCN5 alone is able to acetylate nucleosomal
substrates, suggesting that the unique N-terminal domains of mammalian P/CAF and
GCN5 may provide additional fimctions important to recognition of chromatin substrates
and the regulation of gene expression (Xu, et al., 1998). The same group has found that
TRF2 (telomere related factor 2) can interact with the N-terminus of mouse GCN5

through yeast two-hybrid screening (personal communication).

130

Some potential candidates were identified that interact with the full length of GCN5,
as listed in Table 6. Since the most two interesting candidates #133 and #81 turned out to
be false positive, further investigation was not pursued for the rest of the candidates.
However, discussion of some of the candidates is revealing.

Candidate #77 encodes a transcriptional activator with putative a helix-loop-helix
DNA binding domain. The future identification of the cognate DNA sequence to which
the native protein binds may reveal some roles played by the GCN5-ADA2 complex.

Candidate #141 encodes a putative ethylene-responsive DNA/RNA helicase protein.
The whole protein has 501 amino acids, where the candidate #141 recovered from yeast
two-hybrid encodes about 300 amino acids, beginning at amino acid 182. The protein
contains a domain called DEAD/DEAH located at amino acids 101-305, and another
overlapping domain called SrmB (from Superfamily 11 DNA and RNA helicases) that is
located at amino acids 72-430. The protein is implicated as an ATP-driven RNA or DNA
helicase participating in a variety of cellular processes such as DNA replication and
recombination and repair, transcription, translation, ribosomal structure and biogenesis,
and pre-RNA splicing.

At first, it is hard to understand why GCN5 might interact with this functionally
remote protein. However, Roeder and colleagues have recently reported that the human
SAGA-like complex STAGA has a loose association with protein SAP130 (Martinez, et
al., 2001). Protein SAP130 is a component of the splicing factor SF3b, a U2 snRNP-
associated protein complex that is essential for spliceosome assembly (Martinez, et al.,
2001). They also find that STAGA associates with components of UV-damaged DNA

binding complex (UV -DDB). UV-DDB recognizes UV-damaged DNA and links to DNA

131

damage repair. In addition, two transcriptional HAT complexes, the human T1P60 and
yeast IN080, both contain proteins related to DNA helicase and both complexes have
DNA helicase activity (Ikura, et al., 2000, Legube, et al., 2002, Shen, et al., 2000). Cells
with impaired HAT activity in TIP6O have defective double-strand break repair (Ikura, et
al., 2000), and yeast with null INO80 shows hypersensitivity to DNA damage (Shen, et
al., 2000). Quite curiously, loss of GCN5 protein in mouse embryo results in increased
apoptosis that is intimately connected to DNA damage (Xu, et al., 2000). Actually, the
huge protein of Tral in SAGA belongs to the protein family of ATM (ataxia
telangiectasia mutated)/DNA-PK/phosphatidylinositol 3-kinase implicated in cell cycle
checkpoint signaling and cellular response to DNA damage (Abraham, 2001, Khanna, et
al., 2001, Vassilev, et al., 1998). Taken together, these results implicate that GCN5-
containg complex might also participate in DNA damage repair. All these fimctional
connections might suggest that candidate #141 is a legitimate interacting protein with
GCN5. The interaction of GCN5 with the candidate #141 is weak based on yeast colony
growth in yeast two-hybrid assay. This might reflect that the interaction is transient.

Candidate #73 is a chloroplast pyruvate kinase isoenzyme. It seems unlikely that
Arabidopsis GCN5, a histone acetyltransferase, also works in the chloroplast.
Interestingly, an analysis of subcellular localization of the yeast proteins (Kumar, et al.,
2002) shows that the majority of yeast GCN5 (70%) is in nucleus, while a small fraction
(12%) is associated in mitochondria.

Candidate #137 encodes a protein related to chromosome segregation ATPase and
membrane-bound metallopeptidase that are involved in cell division and chromosome

partitioning. Candidate #145 and #155 also encodes protein that is similar to tripeptidyl

132

peptidase. It is curious why peptidases were isolated three times.

Candidate #46 encodes a Pirin-like protein. Pirin is a nuclear protein and the N-
terminal half of Pirin is significantly conserved between mammals, fimgi and plants
(Wendler, et al., 1997). Human Pirin stabilizes the formation of complexes containing
Bel-3, the anti-apoptotic transcription factor NF -kappaB and its DNA target sequences in
vitro. Moreover, and Bcl-3 mediated transcription is stimulated by HAT protein Tip60
(Dechend, et al., 1999), suggesting a role for HATS in Pirin-associated biological

activities.

133

Chapter Six
The interaction between CBF] and ADA2

Introduction

The expression of many genes is induced in Arabidopsis plants under cold stress.
COR genes (for c_o_ld response) are among these genes which are actively transcribed
when plant Arabidopsis is exposed to cold temperature, and are quickly turned off when
the plant is shifted back to warm temperature. One of the transcriptional activators of
COR genes is CBF 1 (Stockinger, et al., 1997), (for Q—repeat Binding Factor).

When protein CBF 1 was assayed for transcriptional activity in budding yeast, its
activity was dependent on yeast proteins Ada2, Ada3 and Gcn5 (Stockinger, et al., 2001)
(Figure 1). These yeast proteins are components of HAT complexes ADA and SAGA
(Grant, et al., 1997). Since CBFl is a transcriptional activator of Arabidopsis and
evidence suggests that Arabidopsis has similar ADA2 and GCN 5 containing complex (es)
in vivo (as discussed in Chapter III and IV), the experiments performed in this chapter
were conducted to test whether there is functional interaction between Arabidopsis

protein CBF] and ADA2 or/and GCN5.

Experimental Methods

Plasmid construction
To construct GST-CBF] expressing plasmids, cDNAs encoding CBFl (amino acids
1-213, fl; 1-115, N; 115-213, C) were PCR amplified and inserted into GST vector

pGEX-6P in fiame with that of GST. The constructed plasmids were correspondingly

134

named pGEX-6P-CBF1(fl, 1-115 or N, 115-213 or C).
To constructed His6-tagged expressing plasmids, cDNAs flagrnents of respective
proteins were inserted into vector pET-28 in flame.

To construct yeast two-hybrid plasmids of CBF 1 , cDNA fragments (encoding amino
acids 1-78, 24-115, 48-115, 97-213, 78-213, and 48-78) were amplified by PCR and
inserted into prey plasmid pGADT7 in flame to that of Gal4 activation domain. All the
cDNAs flagrnents in constructed plasmids were checked by DNA sequencing to make
sure no mutation was introduced. Other yeast two-hybrid plasmids were described in
Chapter III.

For constructing prey plasmids expressing TINY (aal-95), ERF4 (1-85), ERFS (1-
219), or DRAB2a (1-139), the corresponding DNA flagrnents were amplified directly
flom genomic DNA by PCR using specific primer pairs and subcloned into yeast two-
hybrid prey vector pGADT7 in flame with Gal4 activation domain. All the cloned DNA
flagments encodes AP2 domain. The sequences of the DNA fragments were continued to

ensure the accuracy of PCR products.

Recombinant protein expression and purification
The expression and purification of GST-tag and His6-tag proteins of ADA2b, GCN5

and CBFl were essentially as described in chapter IV. The GST tag was subsequently cut
off with protease thrombin (for CBF l) or PreScission protease (for GCN5 and ADA2b).
PBS buffer with 10% glycerol and 1 mM DTT or HEPES buffer (40 mM HEPES, 50 mM
NaCl, pH 7.3, 10 % glycerol and 7 mM 2-mercaptoethanol) was used dependent on

individual purification procedure.

135

Recombinant His6-CBF l and His6-ADA2b were purified using Ni-NTA beads
(Qiagen). The eluted protein solution was dialyzed against HEPES buffer or otherwise

indicated specifically.

GST pull-down assay

To express in vitro radiolabeled proteins, about 0.5-1 ug of the corresponding pET-
28 vector-derived plasmids was mixed with Transcription/Translation coupled TNT T7
system (Promega) in a final volume of 50 u] that included [3 5 S] methionine. The mixture
was allowed to react for two hours at 30°C.

Depending on the specific experiment, about 1-2 ug GST or GST fusion protein was
bound to glutathione Sepharose beads by incubating the beads with appropriate amount
expressed E. coli cell lysate for about one hour and subsequent washing with HEPES
buffer (40 mM HEPES pH 7.4, 100 mM NaAc, 5 mM Mng, 1 mM EDTA, 10%
glycerol, 0.02% Triton X-100). Then the beads were incubated with 4 ul of radiolabeled
translation mixture of target protein for l h at 25°C in 200 pl of binding buffer (HEPES
buffer with 1 ug/ml pepstatin, 50 ug/ml PMSF, 1 ug/ml leupeptin and 2 ug/ml aprotinin)
plus 8% bovine serum. Afier washing six times with 800 pl of binding buffer (with salt
raised to 300 or 450 mM NaCl) at room temperature, the bound proteins were eluted by
boiling in 50 pl 4x SDS-PAGE loading buffer. The eluates were electrophoresed in 10 or
12% SDS-PAGE gel. Pulled down proteins were detected by autoradiography after

exposure to X-ray film for 1-2 days.

Yeast two-hybrid assay

136

The yeast two-hybrid system MACTCHMAKER Gal4 system 3 (Clontech) was used
for two hybrid assays. Yeast AH109 cells were transformed with respective bait and prey
plasmids. The transformed yeast cell were grown on selective solid medium SD (Ade-
His- Leu- Trp-) to find protein-protein interaction, as well as on non-selective solid
medium SD (Leu- Trp-) to check the expressed foreign proteins were not toxic to yeast

cells.

Gel mobility shift assay

The DNA probe was a double 17-mer oligonucleotide derived from the promoter
sequence of gene COR15a. The wild type sequence was ATTTCATGG
CCGACCTGCTTTTT. To make wild type probe, synthetic oligonucleotides (ST567 and
ST568) were annealed, and were end labeled with y-[32P]-ATP and T4 nucleotide kinase
(Gibico 18004-010). The mutant sequence was ATTTCATGGtatgtCTGCTI‘I‘TT. The
corresponding synthetic oligonucleotide primers ST567 and ST568 were also annealed
for competition assay.

The DNA-CBF] complex formation was carried out in 20 u] buffer (20 mM HEPES,
50 mM KCI, 2.5 mM MgC12, 1 mM EDTA, 10% glycerol, lmM DTT) containing 4 ug
BSA, 1 ug poly (dI-dC), 1 u] probe and about 2 pM CBF] . The reactions were
electrophoresed in native polyacrylamide gel (40:1, in 0.5 x TBE buffer). The gel was
pre-run at 150V for 30 minutes at cold room before adding the samples. The samples
were run at 150V for 4 hours at cold temperature. The TBE buffer was changed to new
buffer afier sample running for 2 hours. Afier gel running, the signal was visualized by

auto-radiography.

137

 

 

IEL'

ad

3C

l——l

For ADA2b and GCN5 interacting with CBFl-DNA complex, 1, 2, 4 p1 of
recombinant ADA2b (5 pl contained about 0.5 pg) or 0.5 pl GCN5 (about 0.5 pg) was
added to the 20 pl reaction at the beginning of the assembly. 0.15 pCi [H3] labeled
acetyl-CoA was also included in a duplicated assay with both GCN5 and ADA2b in the
beginning of reaction, and this assembly was allowed to react for 40 min at 27°C before

adding probe. BSA 2pg was also included in one control assay.

Results

Test the physical interaction between CBF] and ADA2a/b or GCN5
In order to test the physical interaction between CBF] and ADA2 and/or GCN5,

GST pull—down assay was performed. Purified recombinant fusion protein GST-CBF]
was attached to glutathione Sepharose beads and was used to pull down Arabidopsis
ADA2a, ADA2b and GCN5 proteins. ADA2a, ADA2b and GCN5 proteins were
expressed and radiolabeled by [3 5 S] methionine using in vitro transcription/translation
coupled TNT T7 system (Promega). The bound proteins were detected by
autoradiography after electrophoresis in SDS—PAGE gel.

As shown in Figure l, recombinant GST-CBF I pulled proteins ADA2a, ADA2b and
GCN 5, but not control protein luciferase. The same amount of protein of control GST
pulled down much less of the target proteins. A later experiment was repeated where
more stringent washing condition was applied: the salt NaCl was raised to 450 mM flom
original 150 mM, 0.2% Triton X-100 was replaced with 1% NP-40 and the result was the

same. The results suggest that CBF 1 interacts with ADA2a, ADA2b and GCN5.

Test the transcriptional activation domain of CBF].

138

nmol min '1 mg proteins

B-galactosidase

 

WT Aada2 Aada3 Agcns

Figure l. The transcriptional activation of protein CBFl is dependent

on yeast protein ADA2, ADA3, and GCN5.

B-galactosidase assays of yeast cell lysates were performed to detect CBFl
transcriptional activity. Protein CBF l was expressed in yeast strains (wild
type WT, Aada2, Aada3, and Agcn5) under the control of ADC] promoter.
The yeast strains contained an integrated reporter gene LacZ which had two
direct repeat copies of the CRT/DRE sequence at its promoter to which CBF]
bound. This experiment was performed by Dr. Eric Stockinger at Dr. Michael
Thomashow laboratory of Plant Research Laboratory [Stockinger et al, Trans-
criptional adaptor and histone acetyltransferase proteins in Arabidopsis and
their interactions with CBF] , a transcriptional activator involved in cold-
regulated gene expression. Nucleic Acids Res. 2001 29:1524-33.]

139

 

CE

we

of

After finding that CBF 1 interacting with ADA2a, ADA2b and GCN5, investigation
was carried out further to find which part(s) of CBF] was responsible for interaction with
ADA2 and GCN5 protein. Protein CBF 1 (full length of 213 amino acids), according its
sequence homology, is composed of two domains: the activation domain (amino acids
115-213, the C-terminus) and the DNA binding domain (amino acids 1-115, the N-
terrninus). The N-terminus contains a well conserved DNA domain called AP2 (AP for
apetala) (Allen, et al., 1998).

To confirm this domain boundary prediction, the N-terminus and the C-terminus of
CBF] were fused separately to the Gal4 DNA binding domain and the firsion proteins
were tested for their ability to activate the reporter genes in yeast. The results indicate
that the C-terminus activated the transcription of reporter genes, whereas the N-terminus
ofCBFl (aa 1-115) did not (Figure 3). This confirmed that the C-terminus (115-213) is a
transcriptional activation domain. Gel mobility shifi assays (see below) also confirm that

the N-terminus has the capacity of binding to specific DNA motif.

Test the interacting rgfl'on of CBF] with ADA2a/b and GCN5
After roughly defining the boundaries of the N- and C- terminus of CBF 1 protein,

study was carried out to determine which parts of CBF] (N or/and C) could interact with
GCN5 and ADA23/b using the method of GST pull-down assays. Recombinant protein
GST-ADA2b and GST-GCN 5 were used to pull down the full length, the N-terminus and
the C-terminus of CBF] which were translated and radiolabeled in vitro using the TNT
T7 system.

As shown in Figure 2B, GST-ADA2b and GST-GCN5 pulled down the full length of

140

 

 

 

Fl ( G C t t‘ I l .

50% input GST-CBFl _GL_
+ +

ADA2a + + +
ADA2b + + + + +
GCN5 + + + + + + +
Luciferase + + +
'5'” - CID—-.Z'
’l -
‘9’ .- .
‘-
r~l
B CBFl-FL CBFl—N CBFl-C

123412341234

F .
mi

1: 20% input, 2: GST-GCN5 3: GST-ADA2b 4: GST

 

Figure 2. Interaction of CBFl with ADA2a, ADA2b and GCN5.

GST pull-down assays were performed to detect the interactions.(A) The Arabi-
dopsis ADA2a, ADA2b and GCN5 proteins as well as luciferase (as a negative
control) were expressed and [358] methionine radiolabeled using the in vitro
transcription/translation coupled TNT T7 system (Promega). Equivalent amounts
of GST or the GST—CBF 1 fusion protein bound to glutathione Sepharose beads
were incubated with ADA2a, ADA2b and GCN5 solutions. Afier washing, the
bound proteins were eluted with SDS-PAGE loading buffer by boiling. The elua-
tes were electrophoresed in SDS-PAGE gel, and the radiolabeled proteins were
detected by autoradiography. (B) Similar GST pull-down assays to find out the N
-terminus of CBF] interact with ADA2b and GCN5. The major difference fiom
(A) was that ADA2b and GCN5 were in the form of GST fusion protein while
CBF 1 proteins were expressed and radiolabeled in the TNT T7 system. CBF l-N:
amino acids 1-115, CBFl-FL: amino acids 1-213, CBFl-C: amino acids 115-213.

141

 

 

 

 

 

 

 

N
Trp-Leu-Ade-His-lX-a-gal Trp-Leu-
B FL
N C
1 24 48 78 96 115 -
activation domain / E
A
B
F D

 

 

E

Figure 3. The C-terminus of CBF] is a transcriptional activator.

(A) Diagram shows the CBF 1 protein and its various regions. (B) Determine the trans-
criptional activation domain of CBF l in yeast cells. cDNAs encoding different regions
of CBF] (as shown in panel A) were subcloned into vector pGBKT7 in flame with that
of Gal4 DNA binding domain. CBF l plasmids were co-transformed into yeast AH109
cells with plasmid pGADT7, and the transformed cells were grown on selective solid
medium SD (Ade— His- Leu- Trp- IX-a-gal) to test transcriptional activation activity
for various regions of CBF], and on non-selective solid medium SD Leu- Trp- to
check the expressed CBF 1 proteins were not toxic to yeast cells. The co-transformation
of plasmid of pGADT7 (with Leu+ marker) enable the yeast cells to grow on Leu- solid
medium.

142

CBF] , confirming the earlier result that the full length CBF 1 interacted with ADA2b and
GCN5. Surprisingly, in contrary to the general model that co-activators usually contact

the activation domain, the N-terminus of CBF 1 (1-115) but not the C-terminus interacted

with ADA2b and GCN5 (Figure 2B).

Determining the interacting motif in the N-terminus of CBF] with ADA2

CBF] is a member of the AP2 domain-containing transcriptional factors in
Arabidopsis, such as ERF 1, ERF 4, ERFS, DREB2a, and TINY. The AP2 (apetala)
domain is a DNA binding domain found in many transcriptional factors and is unique to
plants (Riechmann, et al., 2000). The NMR structure of ERF] AP2 domain reveals it is
composed of an a-helix and a three B-sheets, where the three B-sheet bundle contacts the
DNA GCC-box motif (Allen, et al., 1998) (and see Figure 4A). Based on CBF] ’8
sequence homology to the DNA binding domain of ERF 1 and CBF] ’3 binding to similar
GCC-box DNA motif (Stockinger, et al., 1997), the CBF 1 N-terminus was divided into
distinct regions as shown in Figure 4B.

GST pull-down assays were performed to test which part of the N-terminus of CBF 1
was necessary for interaction with GCN5 and ADA2b. As showed in Figure 5,
recombinant fusion proteins GST-GCN5 and GST-ADA2b pulled down the three
overlapping regions of CBF 1 N-terminus, amino acids 24-115, 48-115 and ]-78. Control
GST alone did not pull down any of these three regions. This result suggested that the
minimal interaction region of CBF] was in the region of amino acids 48-78, the 3-[3

sheets.

The above date indicted that the putative three B-sheet region of CBF] was sufficient

143

 

B
: : 96 115 213

mn- l / a

Figure 4. Structure of the AP2 DNA-binding domain

(A) Shown is the NMR structure of GCC-box binding domain of Arabidopsis protein
ERF] (for ethylene response factor). The domain has about 60 amino acids, its three-
B-sheet bundle contacts the DNA motif (AGCCGCC) underneath. The picture is
modified from M. D. Allen et al EMBO J. 17, 5484-9 1998.

(B) The defined regions of protein CBF]. The region (amino acids 48-96) in CBF] is
highly homologous to the GCC-box binding domain of Arabidopsis protein ERF], and
CBF 1 protein has been shown to bind to similar GCC-box (GGCCGACC).

CBFl 24-115 CBFl 48-115 CBFl 1-78
123412341234

43- -- e
T Ma ’Q’“

~

‘ #

GST-GCN5 \. p a...
GST-ADA2b - :g-L' . 2.7-e .. w“!
GST — g 1 “g d

 

Figure 5. The N-terminus of CBF] interacts with ADA2b and GCN5.
GST pull-down assays were performed to detect the interactions. Recombinant
GST—ADA2 and GST-GCN5 fusion proteins and GST alone were bound to
glutathione Sepharose beads. The three N-terrninal regions of CBF] a 24-115,

48-115 and 1-78 were expressed and [S35] methionine radiolabeled using the
in vitro transcription/translation coupled TNT T7 system (Promega). GST,
GST-ADA2b or GST-GCN5 fusion protein bound to beads was incubated with
the solutions of three CBF] proteins. After washing, the bound proteins were
eluted with SDS-PAGE loading buffer. The eluates were electrophoresed in
SDS-PAGE gel, and the pulled down proteins were detected by autoradiogra-
phy, as shown in the upper panel. The protein amounts of GST-ADA2b, GST-
GCNS and GST bound to beads tested were shown in the lower panel that were
stained with Coomassie Blue. 1: 25% Input, 2: GST-ADA2b, 3: GST-GCN5,

4: GST.

145

 

hrhr
hnhé
imop

\anm

conm
inmn
dour

testa

HCCC:
Sufii

inter.

“as1

QC)

13316:

and;

for interacting with either GCN5 or ADA2b. A yeast two-hybrid assay was carried out to
further confirm the result. cDNA fragments encoding various CBF] regions were inserted
into plasmid pGADT7 (encoding Gal4 activation domain) to make prey plasmids. These
various domains and regions included CBF] amino acids 1-78, 24-115, 48-115, 96-213
and 78-213. Yeast strain AH109 cells were co-transformed with respective CBF] prey
plasmids together bait plasmid encoding the full-length GCN5, ADA2a or ADA2b. The
transformed yeast cells were grown on selective solid medium SD (Trp- Leu- His— Ade-
/X-a-gal) for 4 days at 30°C.

As shown in Figure 6B, ADA2b showed interaction with the CBF] regions that
contained the putative three B-sheet region. The C-terminus of activation domain did not
interact with ADA2b. This result correlated with earlier result obtained flom GST pull-
down assays. The yeast two-hybrid assay result also reveals that ADA2a, which was not
tested in the previous GST pull-down assays, had similar interaction properties as that of
ADA2b.

The results in Figure 6B showed that three-B-sheet bundle of CBF ] (aa 48-78) was
necessary for interaction with ADA2a and ADA2b. To find whether this bundle was
sufficient for interaction with ADA2a and ADA2b, the bundle alone was tested for its
interaction in yeast two-hybrid assay. As shown in Figure 6C, the three B-sheet region
was sufficient for interaction with ADA2a and ADA2b.

Surprisingly, the yeast two-hybrid assays did not indicate any interaction between
GCN5 and CBF l, regardless which part of CBF] (N-terminus and C-terminus) was
tested (Figure 6D). This failure of interaction was not due to the lack of expression of bait

and prey proteins in transformed yeast cells, for GCN5 bait protein is known to be

146

 

 
 

4 7 96 115 . 213
w.- m 1 activation domain [L]
a I

 

 

 

 

 

 

pGDBD-ADAZa pGDBD-ADAZb pGDBD-GCNS

Figure 6. Defining the minimal interaction domain of CBF] with ADA2a and
ADA2b.

Yeast two-hybrid assays were performed to detect the interactions. cDNA encoding the
various CBF] regions (as indicated in panel A) were subcloned into vector pGADT7 to
create prey plasmids that expressed the respective CBF 1 protein regions in fusion with
Gal4 activation domain.Yeast strain AH109 cells were co-transformed with CBF] prey
plasmids together with bait plasmids of GCN5, ADA2a or ADA2b. The transformed
cells were grown on selective solid medium (SD Trp- Leu- His— Ade-lX-or-gal) for 4
days at 30°C. (A) Diagmr shows the CBF] protein in full length and the various regions
(a, b, c, d, e) tested for interaction with ADA2 and GCN5 proteins. c: amino acid 1-78,
a: 24-115, b: 48-115, d: 96-213, e: 78-213, and f: 48-78.

(B, C, D) Shown is the growth of transformed yeast colonies on the selective medium.

147

expressed and flee to interact with other proteins (see chapter 111) and all the CBF 1

regions and domains were known to express in the yeast cells (Figure 6B and Figure 3B).

The AP2 domain of TINY interacts with ADA2a and ADA2b.

There are about 140 AP2 domain-containing proteins in Arabidopsis that are
implicated in variety of biological activities (Riechmann, et al., 2000). The interaction of
tlle AP2 domain of BCF 1 with ADAZa and AD2b raises the question whether the
interaction is specific to CBF] or is common to all AP2 domain-containing proteins.

To address whether the interaction of CBF] domain with ADA2 is specific, four
transcription factor proteins flom Arabidopsis, ERF4, ERFS, TINY and DREB2a were

tested for their interaction with ADA2, ADA2b and GCN 5 by yeast two-hybrid assay.
The ERF4/5 proteins are ethylene response factors, DREAb2a is implicated in
Arabidopsis response to drought (Liu, et al., 1998, Sakuma, et al., 2002), and over
CXpression of TINY leads to tiny plants (Wilson, et al., 1996). These four proteins all
contained very similar AP2 domains as that of CBF 1, as shown in Figure 7A. DNAs
encoding AP2 domains of TINY (amino acids 1-95), ERF 4 (1-85), ERFS (1-219) and
DREBZa (1-139) were amplified flom Arabidopsis genomic DNA. Since these genomic
DNA flagrnents did not have any introns, they were directly subcloned into vector
PGADT? to construct prey plasmids. Yeast two-hybrid assays were performed as before.
Interestingly, only TINY (1-95) showed interaction with ADA2a and ADA2b (Figure 7B),
demonstrating that ADA2 proteins interact with only specific AP2 domain. As observed

for CBF ] , none of them including TINY showed any interaction with GCN5 (Figure 7B).

148

 

[.i «LII Fl IrraL .5 "V1; 0:11.; ..1.1ur.\.rhu

Figu re 7. Differential interaction of AP2 domain with ADA2 proteins.

(A) Amino acids sequence alignment shows the high homology among the AP2 domains
of Arabidopsis transcriptional factors TINY, ERF4, ERF 5, DREB2a and CBF l. The dash
lines indicate the identical residues and the dots indicate where residues are absent for the
respective AP2 domain. (B) Yeast two hybrid assays were performed to detect protein-
protein interactions. DNA flagrnents encoding respective AP2-domain containing regions
of TINY (amino acids 1-95), ERF4 (1-85), ERFS (1-219) and DREBZa (1-139) were in-
serted in to prey vector pGADT7 in flame with that Gal4 activation domain. Yeast cells
were co-transformed with the corresponding prey plasmids and the bait plasmids of ADA

2a (pGDBD-ADA2a), ADA2b (pGDBD-ADAZb), or GCN5 (pGDBD-GCNS), and were
grown on selective medium for 4 days at 30°C.

1 : pGAD-BRF4 (1-85),

2: pGAD-ERFS (1-219),

3: pGAD-DRBBZa(1-l39),

4: pGAD-T1NY(1-95),

5: pGAD-GCNS (HAT, 203-384),
6: pGADT7, 7: pGAD-ADAZb.

149

286.5% 534.52% ¢~<n<dmnea

 

 

Emqurmmlzfilmrnrqlo---u----->-52-4-a-7¢§l=m rrrrrrr we 38-4333
Hurlerm-{ETrural--quarlr->rwox-Iornrafilfi rrrrrrr an 33-3de
mam}-mrgmruzrulurarmluolm-11.7mm .m----H-¢lle ...... hmo 2.3-2.. «amaze
manuraxruarream--arwrrrrnrurorrruu>rr.--oraruradumuzmru ..... mH Ammrmuvvmmn
nomrrruuraran nun u. u H a x ..... >m Acm-vmvazua

mamzrog§>§aouaogufi . ggmgmzmoggum 83..va Emu

<

150

Detect ADA2b-CBF1-DNA tertiagy complex
The observation of physical interaction between ADA2a/b and CBfl suggests a

model that when CBF] activates the transcription of COR genes, it recruits the ADA2-
GCNS chromatin modifier to facilitate transcription.

The data described above show that the putative three B-sheet bundle of CBF 1
interacts with proteins ADA2a and ADA2b. On other hand, this three B-sheet bundle,
based on the NMR structure of the homologous AP2 domain of EFRl with bound DNA
GCC-box, likely binds directly to DNA motifs of CRE element flom CBF] responsive
genes such as COR genes. It is not known whether CBF] could interact with its cognate
DNA motif and ADA2 proteins simultaneously, or if there exists competitive binding to
CBF] between ADA2 proteins and DNA motif.

To address this question, gel mobility shift assays were carried out to test whether
adding of ADA2b protein results in either supershifi or disappearing of CBF l-DNA
complex. In the gel mobility shift assays, the DNA probe was 17-mer oligonucleotides
which contained the CBF 1 -binding motif flom the COR15a gene. A mutant probe was
also introduced as the control. The probes were radiolabeled by [32F]. Recombinant His6-
ADA2b and His6-CBF](aa 1-213) proteins were purified as described in Chapter IV,
using HEPES buffer (40 mM HEPES, 50 mM KC], 10% glycerol, 7mM 2-
mercaptoethanol). The proteins of His6-CBF] and His6-ADA2b were eluted flom Ni-
NTA beads by 220 mM imidazole in HEPES buffer. The proteins solution was then
dialyzed in HEPES buffer to remove the imidazole. The purified His6-ADA2b and His6-
CBF] proteins are shown in Figure 8A. There was considerable contamination in the

purified His6-ADA2b. The presence of His-ADA2b in the purified preparation was

151

A kDa 12

83 — 3 w
62 I » ‘
47.5 -
32 5 _ -< 1: His6-ADA2b
. .. 2: His6-CBFl(l-213)
25 _
B Wild WPC probe mutant robe

 

CBFl(ul) 01/321/161/8 1/4 1/2 1 0 1/32 1/16 1/8 1/4 1/2 1_

  

Figure 8. Testing the CBF] protein binding to DNA probe.

(A) The purified recombinant proteins CBF] and ADA2b as His6-tagged fusions.
The protein preparations were electrophoresed in 10% SDS-PAGE gel and visua-
lized by Coomassie Blue staining. The protein position of CBF] and ADA2b is
indicated by the filled triangle. These two recombinant proteins were used in gel
shifi mobility assays in panel B and Figure 9. (B) Gel mobility shift assay to de-
tect the binding of CBF ] with DNA probe. The [32F] radiolabeled DNA probe
(wild type or mutant) was incubate with recombinant His-CBF] protein, and
then electrophoresed in native polyacrylamide gel. The positions of the DNA
probe in gel were detected by autoradiography. The flee probe and CBF l-probe
are indicated by the arrows.

152

 

confrr.
Chap:

T
probe:
forme
bindir
mobil

ADA

protei
recon
W, w
Show.
detec‘
acety

Sallie

confirmed by Western blotting and by its ability to enhance GCN5 HAT activity (sees
Chapter IV).

The gel mobility shifi assays were carried out in HEPES buffer with the labeled
probes and recombinant His6-CBF 1 protein. As shown in Figure 8B, the CBF] proteins
formed complexes with wild type probe but not the mutant one, illustrating the specific
binding of CBF] with the wild type probe. Adding of the ADA2b protein alters the
mobility of the radiolabeled DNA probe, as shown in Figure 9. However, the adding of
ADA2 protein also did not disrupt the CBF 1 -DNA complex (Figure 9).

Since the GCN5 protein interacts with the ADA2b protein, the presence of the GCN5
protein might change ADA2b properties in binding to other partners. Therefore, the
recombinant GCN5 protein, which was purified in HEPES bufl‘er as described in Chapter
IV, was also added together with ADA2b in the gel mobility shifi assays. The result
showed that the CBF] -DNA complex remained stable and no supershifi complexes were
detected (Figure 9). Considering that ADA2b is acetylated by GCN5 (see Chapter VII),
acetyl-CoA was also included in another two parallel reactions. The result was again the

same as above (Figure 9).

Discussion
The experiments described in this chapter show that Arabidopsis protein CBF 1 can
interact with co-activator proteins ADA2a and ADA2b using the yeast two-hybrid assay
or GST pull-down assay. The putative three B-sheet bundle in the AP2 domain of CBF]
i S necessary and sufficient for interacting with ADA2a and ADA2b. This interaction

Suggests a model. When plant Arabidopsis is under cold stress, activator CBF] activates

153

 

Ac-

 

Tertiary
complex?

CBFl-DNA>

free probe b

 

1234567891011121314

Figure 9. Detect ADA2b-CBF1-DNA tertiary complex.

Gel mobility shift assays were performed to detect the potential tertiary complex. [32P]
radiolabeled DNA probe (wild type) was incubate with the CBF] protein in combination
with the ADA2b protein (lane 3-8) or the ADA2b protein plus the GCN5 protein (lane 9-
12). The last two lanes also contained Acetyl-CoA in the reaction in order to make the
ADA2b protein acetylated. The incubated mixtures were then electrophoresed in native
polyacrylamide gel. Two different purified batches of the ADA2b protein (batch l and
batch 2) were tested. Both the recombinant proteins CBF] (1-213) and ADA2bwere puri-
fied as His-tgged fusions, while recombinant GCN5 was purified as GST fusion and the
GST tag was removed subsequently. The free probe in the last two lanes seemed to diff-
use for unknown reason.

154

the transcription of cold response genes in recruiting GCN-ADA2 complex that
acetylates histones at the local promoter for facilitating transcriptional activation. In
support of this model, Arabidopsis plants with crippled ADA2b or GCN5 protein show
delayed transcription of COR genes in response to cold temperature treatment
(V lachonasios, et al., 2003).

GST pull-down assays showed that GCN5 can also contact the CBF] N-ternrinus.
But yeast two hybrid does not show this interaction. This lack of interaction of GCN5
with any region and domains of CBF 1 in yeast two-hybrid assay was not due to the failed
expression of bait or prey proteins in yeast cells. Bait protein GCN5 in fusion of Gal4
DNA binding domain is expressed in yeast (see Figure 3 at chapter 111). Prey protein
CBF 1 (amino acids 24-115, 48-115 and 1-78) are expressed because they show
interaction with ADA2a and ADA2b in the parallel assays. We conclude that the
interaction between GCN5 and CBF] as showed by GST pull-down assay is artificial.

The interaction of CBF l to ADA2 proteins is specific. Among very similar AP2
domains flom Arabidopsis transcriptional factors TINY, ERF 4, ERFS and DREB2a, only
TINY shows interaction with ADA2a and ADA2b. This might be explained by the fact
that putative three B-sheets and a a-helix domain of TINY have the highest similarity in
amino acids sequence to that of CBF 1 (see Figure 7). A less likely explanation is that the
yeast cells failed to express these AP2 domain-containing factors.

Regardless, the interaction of ADA2 adaptor protein with AP2 DNA binding domain
suggests that ADA2-GCN5 complex might play multiple roles in plants. The AP2
domain is unique to plant world, and is found in a wide range of proteins in Arabidopsis,

including important regulators of several developmental processes involved in floral

155

organ identity determination or control of leaf epidermal cell identity, and many key
responding factors to various types of biotic and environmental stress such as cold stress,
dehydration and pathogen attack (Chang and Shockey, 1999, Memelink, et al., 2001,
Riechmann, et al., 2000, Shinozaki and Yarnaguchi-Shinozaki, 2000). This possible broad
range of action is also supported by the finding that disruption of the ADA2b or GCN5
gene Arabidopsis resulted in severe phenotypes such as dwarf and sterility (V lachonasios,
et al., 2003). Interestingly, a recent report shows that the ADA2b protein in Arabidopsis
is involved in cell differentiation in response to auxin and cytokinin concentration
variation (Sieberer, et al., 2003), and a AP2 domain containing factor is involved in shoot
differentiation in combination with cytokinin action (Banno, et al., 2001).

The failure of detecting an ADA2b-CBF1-DNA complex by gel mobility shifi assays
could be a technical problem such as the buffer conditions are not appropriate, and does
nor necessary mean that in viva CBF] can not recruit the ADA2-GCN5 complex.

The AP2 domain of CBF 1, which binds DNA motif of COR genes’ promoter
directly, contacts the co-activator ADA2 protein. This observation is contrary to the
standard model in which the activation domains of transcriptional activators contact and
recruit transcriptional co-activators. But increasing evidence supports that the DNA
binding domain recruitment of co-activator is also widely employed. One example is
EKLF, a transcriptional activator in erythroid cell differentiation, in which the DNA
binding domain of EKLF contacts SWI/SNF complex directly and the recruited Swi/Snf
complex sufficiently remodels local chromatin (Kadam, et al., 2000). Interestingly, for
transcription to start, the activation domain of EKLF is additionally required. Another

case is the DNA binding domain of nuclear receptors RXR/RAR heterodirner which

156

contacts histone modifier PCAF directly (Blanco, et al., 1998).

On the other hand, the contact of the ADA2 proteins by transcriptional activator in
Arabidopsis also suggests that similar complex in other species such as yeast SAGA
might also be recruited by activators through interaction of the ADA2 protein in addition
to the known handles of GCN5 and Tral proteins (Brown, et al., 2001, Kulesza, et al.,

2002)

157

Chapter Seven
Aceglation of ADA2a and ADA2b by GCN5
Part I Acetylation of ADA2a/b by GCN5 in vitro
Part 11 Test the sigificance of ADA2b acetyaltion in viva

Introduction

Although discovered initially as histone acetyltransferses, the mammalian CBP/p300
and PCAF proteins can also acetylate other proteins including general transcriptional
factors and regulatory factors such as TFIIB, TFIIF, p53, EKLF, GATA-l, HMG I (Y),
HMG17, ACTR, Tat, MyoD and E2F1(Spilianakis, et al., 2000). The biological
significance of acetylation for some non-histone proteins has been reported, including the
regulation of transcriptional factor activity and the regulation of protein cellular location
(F reiman and Tjian, 2003, Kouzarides, 2000, Sterner and Berger, 2000).

For instance, acetylation affects protein cellular localization. Protein POP in
Caenarhabdr’tis elegans is a homologue of LEF/T CF , a nuclear effector of the Wnt
signaling pathway that controls cell fate decisions and embryonic patterning in animal
development. Recently, the Shi group found that POP flom Caenarhabditis elegans
embryos is acetylated by CBP/P300. Acetylated POP enhances its nuclear retention
through blocking nuclear export pathway (Gay, et al., 2003). CIITA is a tansactivator of
MHC class 11 genes. CIITA associates with CBP/PCAF and is acetylated by CBP/PCAF.
Interestingly, the acetylated lysines are within a nuclear localization signal (NLS). The
acetylated NLS enhances CHTA nuclear import (Spilianakis, et al., 2000).

Another case of protein acetylation affecting cellular localization is adenovirus 12 S

158

ElA (Madison, et al., 2002). EIA is acetylated at lysine 239 by CBP in transfected cells.
E] A with acetylation at lysine 239 or substitution mutation at lysine 239 resides
predominantly in cytoplasm. Unacetylated E] A has much higher affinity for importin-a3
than acetylated ElA. This might be the first case indicating that acetylation of protein
results in disruption of protein-protein interaction.

Acetylation of tumor suppressor protein p53 is another case in which acetylation
regulates protein function. Human p53 is acetylated at lysines 373 and 382 within its C-
terrninal domain known to regulate p53 DNA binding activity. Acetylated p53 enhances
its DNA binding activity in vitro (Gu and Roeder, 1997). In viva, acetylation of p53
promotes recruitment of co-activators such as CBP and TRRAP and therefore potentiates
p53 activity as a transcriptional activator (Barlev, et al., 2001).

A similar case is HIV Tat acetylation. Acetylation of Tat at lysine 50 by p300/PCAF
results in its dissociation from TAR RNA and promotes fomration of a multiprotein
complex comprised of Tat, p300/CBP, and PCAF (Deng, et al., 2000). This event enables
the stalled transcriptional machinery complex to elongate efficiently on the HIV DNA
template (Garber and Jones, 1999).

Given that many transcriptional activators are acetylated, we hypothesized that the
Arabidopsis activator CBF 1 may be acetylated by GCN5. In vitro acetylation assays
found no support for this hypothesis. Serendipitously, ADA2b, which was included
originally to facilitate acetylation, was itself acetylated by GCN5. This unexpected
finding resulted in detailed study of the Arabidopsis ADA2 acetylation in vitro and its

significance in viva in this chapter.

159

Experimental Methods

Purification of recombinant GCN5, ADA2b and CBFl proteins
Recombinant full length an truncated GCN5, ADA2b and ADA2a were prepared as

described in Chapter IV. CBF 1 was also expressed as a GST fusion protein flom a
plasmid named pGEX-CBF ] (fl). The GST-CBF] firsion protein was affinity purified by
glutathione Sepharose beads and then the fusion protein was digested with protease
thrombin for 5 hours in the cold room in PBS buffer, and the thrombin was subsequently
absorbed away by incubating the digestion supernatant with Streptavidin agarose

(N ovagen) at room temperature for 30 min.

Purification of recombinant yADA2N

The cDNA encoding the yeast ADA2 N-terminus (amino acids 1-176) was
subcloned into pGEX vector to express GST fusion protein GST-yADA2(1 -1 76). As
control, GST tagged ADA2b N-terminus GST-ADA2bN(1-225) as well as GST proteins
were purified. The expression and purification procedure for GST, GST-ADA2b (1-225)
and GST-yADA2 (1-1760 was the same as that of purification of ADA2bN (see chapter
IV) except that the GST fusion proteins bound to Sepharose beads were eluted with GSH

Tris buffer (20 mM glutathione, 50 mM Tris, pH 8.0, 50 mM NaCl, 1 mM DTT).

Purification of ADA2b (1-179, 28-225, 28-179, 1-208)
cDNA encoding the deleted versions of ADA2b N-terminus (amino acids 1-179, 28-

225, 28-179, 1-208) were amplified by PCR and subcloned into pGEX-6P vector in

flame with that of GST. The expression condition and purification procedure was the

160

same as described in Chapter IV except that induction condition was 80 pM IPTG for 17

hours. The GST tag was removed by protease PreScission (digestion for 3 and half hours).

Acetylation assay for ADA2b
The acetylation was carried out in Tris buffer (50 mM Tris, pH 7.8 or 7.4 or 7.0, 40

mM NaCl 10 % glycerol, 0.1 mM EDTA, and 1 mM DTT) or HEPES buffer (40 mM
HEPES, 50 mM KC1, pH 7.3, 10% glycerol and 1 mM DTT). About 0.3-0.5 pg GCN5
was used in the assay together with appropriate amount of ADA2b proteins as needed.
The mixture was allowed to react for 30 min at 27°C or otherwise specified. After the
reaction was finished, the mixture was analyzed in SDS-PAGE gel. The proteins in gel
were first stained with Coomassie Blue and the acetylation signals were then detected by
enhanced fluorography. For lower pH assay, the same assay was repeated except that the

Tris pH was adjusted to 7.0, or 7.4.

Acetylation assay for ADA2bN, ADA2bM, ADA2bC and ADA2bN deletions
About 0.5 pg GCN5, 0.3 pg ADA2b, 2 pg ADA2bN, 0.5 pg ADA2bM, 2 pg

ADA2bC, and 0.15 pCi H3-Ac-CoA were added as needed in 30 pl Tris buffer (50 mM
Tris, pH7.4, 40 mM NaCl 10 % glycerol, 0.1 mM EDTA, and 1 mM DTT). The
acetylation reaction proceeded for 40 min at 27°C. In some experiments, about 0.5 or lpg
ADA2bM (225-377) was also included improve acetylation efficiency. The reaction
products were resolved in a 12% SDS-PAGE gel and the proteins were Coomassie
stained. The protein gel was then soaked in autoradiographic enhancer (N EN Life

Science Products), vacuum dried and exposed to X-ray film at -70°C for fluorographic

16]

 

pictur

min
pro:
pun'

DH

or)

€112

Pia

3px
D.\

picture.

Mutations of putative acetylation site in ADA2a and ADA2b

Site-directed mutagenesis was used to change lysine codons to arginine or alanine
codons in ADA2a and ADA2b expressing plasmids. Plasmid pGEX-ADA2a encoding
GST fused to ADA2a (full length) was mixed with primer pairs ST660-ST666 (K/A at
K257), ST661-ST667 (K/R at K257), ST662-ST668 (K/A at K265), or ST663-ST669
(K/R at K265). Plasmid pGEX-ADA2b encoding GST-ADA2b (1486, full length) was
mixed with primer pairs ST664-ST670 (K/A at K215) or ST665-ST671 (K/R at K215).
The PCR reaction was cycled for 16 times at settings of 95°C 1 min, 55°C 1 min, 72°C 6
min and 5 unit PFU (Turbo) polymerase enzyme in 50 pl reaction volume. Then the PCR
product was cut with 20 units of Dpn l for 80 nrin to digest the templates. DNA was
purified flom the digestion by ethanol precipitation and used to transform E. cali strain
DHSa.

Positive clones were identified by digestion with restriction enzyme Hae III (GG/CC)
or Him I (GC/GC), for the mutated plasmids contained new restriction sites for these
enzymes.

To make mutant GST fusion expressing plasmid for ADA2b, the Nca I-EcaR I
flagrnent of ADA2b (aa172-288, mutated site K215) was cut flom the positive mutated
plasmid and used to replace the same region in wild type plasmid pGEX-ADA2b. This
approach avoided introducing unnecessary other mutated sites and reduced the work of
DNA sequencing---only the Nca I-EcaR I region was sequence checked.

To make mutant GST firsion expressing plasmid for ADA2a, the Hind III-Psi I

162

 

mu

wh

W3

sec.

C0(

W3

the

“'8

ICE

nu

TEE
Pr

He

fragment of ADA2a (aa210-313, mutated site K265 or K257) was cut flom the positive
mutated plasmid and used to replace the same region in wild type plasmid pET-ADA23
which expressed His6-tagged ADA2a. Then cDNA encoding the mutated ADA2a (1-273)
was subcloned into GST vector to make GST fusion expressing plasmid. The DNA
sequences were checked.

In summary, for ADA2aN(1-273), wild type codon AAG at K257 was changed to
codon GCC for alanine or CGC for arginine, codon AAA at K265 was changed to codon
GCC for alanine or CGC for arginine. For ADA2b (fl), wild type codon AAG at K215

was changed to codon GCC for alanine or CGC for arginine.

Ace_tylation assays for free and nucleosomal histones with mutant ADA2b
Recombinant GST-ADA2b (K215R or K215A) was purified essentially the same as

that of GST-ADA2b (wild type) (see chapter IV). GST-ADA2b (wild type) was prepared
in chapter IV. The flee core histones’ acetylation assays were also performed the same
way as described in chapter IV.

For testing mutant ADA2b effect of GCN5 activity on nucleosomal substrate, the
reactions were carried out in 27 pl HEPES buffer containing 6mM sodium butyrate, 8p]
nucleosome preparation (prepare in chapter IV), 0.2 pC [3 H] Ac-CoA, about 8 pM GCN5
and about 4 pM or 8pM GST-ADA2b (wild type, K215R and K215A mutant). The
reaction proceeded for 30 min at 25°C and was stopped by adding 16 pl of 6 x SDS-
PAGE loading buffer and boiled for 5 min. Equivalent amount of GST was included as
negative control. The reaction products were resolved in a 15% SDS-PAGE gel, and the

acetylated histones were detected by fluorography for 4 months at -80°C.

163

 

 

I‘L.

ADA2bN acetylation for mass spectrometry
The acetylation reaction was carried out in 16 p] buffer (40 mM Tris, 50 mM KC]

10% glycerol, 7 mM 2-mercaptoethanol, pH 7.4) containing ~2pg ADA2bN (1-225),
~l.5 pg ADA2bM (225-377), ~0.5 pg GCN5, and 60 nM Ac-CoA. The reaction was
allowed to proceed for 30 nrin at 25°C. The acetylated protein was resolved in 12% SDS-
PAGE gel. The protein band was cut out from the gel after being stained with Coomassie
Blue and sent to the mass spectrometry facility at our department. There the acetylated
protein was digested in gel by trypsin. The recovered trypsin-digested faction was

subjected to mass spectrometry (LC/MS/MS). This was done by Dr. Brett Phinney.

Construction of plasmid pGEX6P-2bNR which expresses ADA2bNR

A cDNA flagrnent encoding ADA2b(1-172) cut flom plasmid pGAD-ADAZb (1-
208) by EcaR I and Nca I, and cDNA encoding ADA2b 173-221 (cut with EcaR I and
Xho I flom PCR product amplified flom ADA2b full length cDNA by primer pairs
ST486-ST707) were subcloned into GST vector pGEX-6P-l by EcoR I and Xha]. The
encoded GST fusion protein sequence is read as GST-ADA2b(1-221)-SGRIV TD,
including the 7 amino acids encoded by the plasmid vector. The acetylation site of K2] 5
reads as mRSFVDRSFGGKZ”KPVSTSSGRIVTD. The new ADA2b (1-221) plus tailed

vector sequence was named ADA2bNR.

ADA2bNR acetylation for mass smtrometly

The protocol was basically the same as that of ADA2bN except the reactions were

performed in HEPES buffer (40 mM HEPES, pH 7.4, 50 mM NaCl, 10 % glycerol and 7

164

 

IA

mM 2-mercaptoethanol).

Constructing transformation plasmid for plant

Binary plasmid pKVA31was used to construct transformation plasmids with mutant
cDNAs of ADA2b. Plasmid pKVA31, constructed by Dr. Kostas Vlachonasios, is a
derivative of plasmid pCAMBIA 3301 (CAMBIA, Canberra, Australia) and contains the
cDNA of wild type ADA2b under the control of the strong promoter 358, and the bar]
(herbicide basta) selection marker. The cDNAs of ADA2b (K215R) and ADA2b (K215A)

were subcloned into pKVA31 to the wild type ADA2b cDNA.

Arabidopsis plant transformation

The constructed plasmids with mutated ADA2b cDNAs were introduced into
bacterium Agrabacterium tumefaciens strain GV3101. To make transgenic plants, the
transformed A grobacteria were used to transform Arabidopsis plants which were in the
early floral stage by the method floral dip (Clough SJ, Bent AF). The genotype of the
transformed plants was heterozygous for T-DNA insertion at the ADA2b gene locus.

Four plants were transformed by each constructed plasmid containing mutated
cDNA of ADA2b (K215R), ADA2b (K215A) or wild type ADA2b.

T1 seeds were collected flom the transformed plants. The seeds collected flom plants
transformed by wild type ADA2b plasmid were named as A] , A2, A3, and A4, and by
that ofADA2b (K215A) as B], B2, B3 and B4, and by that ofADA2b (K215R) as C1,

C2, C3 and C4.

165

 

6°C
med
“hi
can.

“all

pro.
thal
PC}
AD

the

Pia
3]]

1.

U.)

Growth of transgenic plants

The collected seeds were first sterilized in 30% bleach (Clorox) with 0.5 % Tween 20 for
20-30 rrrin, and washed with water for 4 times. The seeds were then left in cold room (4-
6°C) for a week to synchronize germination time. The seeds were then put on solid
medium of Garnborg’s B-5 with Basta (4 mg/L) to germinate. 7 to 10 days old seedlings
which were Basta resistant were planted in soil for future screening under the growth
condition of 24-hour light cycle. The plant in soil pots were watered every 8 days with

water, and with nutrient-solution in the middle of every 8 days.

Genotyping the offspring of the transformed plants

DNA samples were prepared flom the growing plant rosette leaves in soil using
Wizard Genomic DNA Purification Kit (Promega). To identify the plants’ genotype, the
genomic DNA was PCR amplified with specific primers and afterwards the PCR
products were analyzed by agarose gel electrophoresis. The primers were designed such
that the genomic DNA flagrnent of the ADA 2b gene across the T-DNA insertion site was
PCR amplified. The primers were SJT780 and STQ29. The PCR product flom the
ADA2b gene with T-DNA insertion was about 8.2 kb long, whereas PCR product flom
the wild type ADA2b gene was about 2.7 kb long.

To check the transgenic integration of the ADA2b cDNA by transformation, the
plant genomic DNA was amplified by PCR with another pair of primers, SJT780 and
STQ29, where SJ T780 was a 358 promoter specific primer. The PCR product was about

1.3 kb.

166

RNA analysis
Total RNA was prepared flom the plant rosette leaves using the RNeasy Plant Mini

kit (Qiagen). About 5 pg of total RNA was electrophoresed in 1% agarose gel and
transferred to Hybond N+ membrane (Amersham Biosciences) by the standard method of
capillary transfer (Frederick M Ausubel et a], Current Protocols in Molecular Biology,
4.95-4.96). The RNA on the membrane was cross-linked by UV light. The membrane
was hybridized with [32P] labeled cDNA probe of ADA2b (cDNA nucleotides 1434).
Hybridization was performed in PerfectHbeM (SIGMA) at 68°C overnight with 100
pg/ml sheared herring testis single strand DNA as non-specific blocker. The hybridized
membrane was washed subsequently in 2 x SSC, 0.1% SDS and 0.01% sodium
pyrophosphate briefly at room temperature, and washed in 0.5 x SSC, 0.1% SDS and
0.01% sodium pyrophosphate at 65°C for 30 min. The signal of hybridization was

visualized with phosphor imager.

Results

Part I Acetylation of ADA2 by GCN5 in vitro
ADA2b is aceglated by GCN5 in vitro

Triggered by the observation that many transcriptional activators can be acetylated
by PCAF , a sibling of GCN5, we tested if the Arabidopsis transcriptional activator CBF 1
was acetylated by GCN5. In vitro acetylation assays were performed to test this
hypothesis.

Recombinant proteins CBF 1 , ADA2b and GCN5 of Arabidopsis were expressed as

167

 

GST fusion
GST tag w
and GST-A
purified pr
The quality
of this C ha]
The ac
proteins. A
that AD A2
bridges CE
assay Was .
acCtylated
Surprj
C0ntrol b(
acetylatior
This 1
“Md be 4
al., 1996)
that the p
by GCN5
ace'l‘ilatir

the reacri

As d

GST fusions in E. cali. After affinity purification on glutathione Sepharose beads, the
GST tag was subsequently removed by proteases, where GST-CBF] was cut by thrombin
and GST-ADA2b and GST-GCN5 were cut by the Prescission protease. The three
purified proteins were stored in PBS buffer (pH 7.4, with 10% glycerol and 1 mM DTT).
The quality of the purified proteins is documented in Figure 9 of Chapter VI and Figure 1
of this Chapter.

The acetylation assays were performed in Tris buffer (pH 7.8) with the recombinant

proteins. As shown in Figure 1A, CBF 1 alone was not acetylated by GCN5. Considering

 

that ADA2b enhances GCN5 HAT on histone substrate activity and that ADA2b protein L
bridges CBF] and GCN 5 interaction by its virtue of contacting both proteins, the HAT
assay was also performed with the addition of ADA2b protein. But again CBF 1 was not
acetylated by GCN5 (Figure 1A).

Surprisingly, the added ADA2b was found serendipitously to be acetylated by GCN5.
Control bovine serum albumin was not acetylated, indicating the specificity of GCN 5
acetylation on the ADA2b protein.

This preliminary acetylating assay was carried out at pH 7.8 in Tris buffer. There
could be GCN5-independent chemical acetylation activity at this high pH value (Kuo, et
al., 1996). Therefore, the assay was repeated using the same reaction conditions except
that the pH was lowered to7 .4 or 7.0. The results showed that ADA2b was still acetylated
by GCN5 at the lower pH (Figure 1C). In addition, GCN5 was found necessary for the
acetylation. No acetylation occurred to ADA2b alone without the presence of GCN5 in
the reaction.

As discussed in Chapter III, Arabidopsis GCN5 protein is composed of the unique

168

1 2 3 4 F_' __P'_
, 2 5 5* 2 5 5*
b ->. 97_ I
“'1 contaminating
62" u. - - - - "‘ protein

44- - - “GCN5(203-568)
> - ‘
28‘- u -< GCN5(I-IAT,
19_ 203-384)
- -
C
ADA2b BSA ADA2b

GCN5 HAT 203-568 fl fl
pH 7.0 7.4 7.0 7.4 7.07.4 7.8 7.0 7.4 7.8

Figure 1. ADA2b is acetylated by GCN5.

(A) ADA2b but not CBF] was acetylated by GCN5. The acetylation assays were perfor-
med with GCNS, [3H] Ac-CoA and respective proteins CBF], ADA2b and BSA at pH
7.8. The reaction products were analyzed in a 10% SDS-PAGE gel. The proteins in gel
were stained by Coomassie Blue as shown at the upper where the ADA2b and CBF]
proteins are indicated by the red and green arrows respectively, and the acetylation
signals were detected by enhanced fluorography as shown at the bottom. 1: CBF], 2:
CBF1+ADA2b, 3: BSA, 4: BSA+ADA2b. (B) Shown are purified recombinant GCN5
(203-384, HAT domain), or GCN5(203-568) which contains the HAT domain and the
bromodomain. The proteins were stained with Coomassie Blue after SDS-PAGE. These
GCN5 proteins were used as acetylation enzyme in panel C. The GCN5 proteins are
indicated by the blue arrows *: the sample was incubated at room temperature for one
hour to check the protein stability. (C) Acetylation assays of ADA2b at lower pH and
with truncated GCN5 versions. Similar acetylation assays as in panel A were performed
at pH 7.0 and 7.4 with full-length (fl) and truncated versions of GCN5. The black arrow
points to the acetylated ADA2b.

169

N-terminus, HAT domain and the bromodomain. To test if the acetylation of ADA2b
required the intact GCN5 protein or if the HAT domain was sufficient, acetylation assays
were repeated using purified recombinant GCN5 HAT domain or GCN5 (203-568) which
lacks the N-terminus. As shown in Figure 1C, the HAT domain sufficed to acetylate
ADA2b.

Other experiments validated the finding of the acetylation of ADA2b by GCN5 in
vitro. For example, in the HAT assays described in Chapter IV, where ADA2b was
shown to enhance GCN5 acetylation activity on core histones, ADA2b was repeatedly

found acetylated simultaneously (see Figure 3 of Chapter IV).

The N-terminus of ADA2b is acetylated by GCN5 in vitro.
Having found that ADA2b is acetylated by GCN5 in vitro, experiments were

performed to determine which region of ADA2b was acetylated. The ADA2b protein was
split into three regions as described in Chapter III, ADA2bN (amino acids 1-225),
ADA2bM (aa 225-377), and ADA2bC (aa 353-486). Recombinant proteins of these three
different regions were purified flom E. cali as GST fusion proteins, and the GST tag was
subsequently removed by protease PreScission. Acetyaltion assays were performed as
before in which the three regions were assayed either alone or in combination. The result
showed that ADA2bN, but not ADA2bM and ADA2bC, was acetylated by GCN5 (Figure
2). Interestingly, the HAT assays also found that ADA2bM could improve the acetylation
efficiency of ADA2bN by GCN5 (Figure 2B, compare lane N to lane N plus M).

No ADA2 flom other organisms has been reported to be acetylated by its partner

GCN5 and or other HAT enzymes. Since the N-terminus is highly conserved among

170

 

 

A B
N N
N M N M N M N M
N M C M C C 2b C N M C M C C 2b C
_ __ _ -
V.‘ - Or .I F
Coomassie Blue staining Fluorogram

Figure 2. The N-terminus of ADA2b is acetylated by GCN5.

In the acetylation assays, about 2 pg ADA2bN, 0.5 pg ADA2bM, or 2 pg
ADA2bC alone or in combination were acetylated by about 0.3 pg GCN5 with
0.15 pCi [3H] Ac-CoA. The acetylation reactions were resolved in a 12% SDS-
PAGE gel and the proteins in gel were Coomassie blue stained, as shown at panel
A. The proteins ADA2bN, ADA2bM and ADA2bC are indicated by the blue
arrows. The acetylated proteins in gel were then visualized by enhanced fluoro-
graphy as shown at panel B. ADA2b in full length (2b) was also included as a
positive control. N: ADA2b (aa 1-225), M: ADA2b (aa 225-377), C: ADA2b

(aa 353-486).

171

ADA2 flom different species, we tested whether the acetylation of the N-terminus of
ADA2b was specific. The corresponding N-terminus of Arabidopsis ADA2a and budding
yeast ADA2 was tested for acetylation by Arabidopsis GCN5. Recombinant proteins of
ADA2a (amino acids 1-267) and GST-yAda2 (yeast Ada2, amino acids 1-176, as GST
flrsion ) were purified. Similar acetylation assays found that the N-terminus of ADA2a,
but not that of the budding yeast ADA2, was acetylated by Arabidopsis GCN5 (Figure 3).
Again, the acetylation of ADA2aN was improved with the presence of ADA2bM (see
Figure3). Based on the facts that the yeast GCN5 was capable to acetylate ADA2b (see
below) and that yeast GCN5 HAT domain is very similar to that of Arabidopsis GCN5, it
is very likely that yeast GCN5 can not acetylate the N-terminus of yeast Ada2. Therefore,

the acetylation is limited to the Arabidopsis ADA2 proteins.

K215 of ADA2b is necessagy for acetylation
The previous results show that the acetylation site of ADA2a or ADA2b was located

at the N-terminus. The results also suggested that the acetylated residue(s) might not
reside in the conserved sequences of ADA2 N-terminus. Indeed, there are extensions at
the amino and carboxyl ends of Arabidopsis ADA2 N-terminus that are not found in
yeast ADA2. Therefore, serial deletions at the both ends of ADA2bN were assayed to test
the necessity of these extensions in acetylation. GST fusion plasmids expressing serial
deletions of the N-terminus of ADA2b (amino acids 1-179, 28-225, 28-179, 1-208) were
constructed, and the recombinant proteins were expressed and purified flom E. coli
(Figure 4).

Acetylation assays were performed as before with similar amounts of GCN5 and

172

 

«\ ,\ ’ \ x I
"Tee“,bq}x $90 39 Q, ’60
”k ‘r v N 4“ 9
Q '3 y? 9 ék D Q?
£99185 VYK bygflvafa
r9r90°0°0° $090909
. .
.- ’- O .
-c- -0
-: p O ' ~
I. -
. "‘ T <

12345 6789

12345 6789

Figure 3. The N-terminus of Arabidopsis ADA2a is acetylated by GCN5.
Acetylation assays were carried out using about 5 pg of each protein substrate and about
0.3 pg GCN5 plus 0.15 pCi [3H] Ac-CoA. The acetylation reactions were analyzed in a
12% SDS-PAGE gel and the proteins in gel were Coomassie Blue stained first, as shown
in the left panel. The acetylated proteins in gel were then visualized by enhanced fluoro-
graphy as shown in the right panel. The protein positions of yeast N-terminus as GST
filsion [GST-yADA2a (1-1 76)] is indicated by blue arrows. Lane 6, 7, 8, and 9 were
actually the repeat of lane 2, 3, 4 and 5, except that lane 6, 7, 8, and 9 had the middle
part of Ada (ADA2bM, amino acids 225-377) included in the reaction to improve the
acetylation by GCN5. The ADA2bM is indicated by the red arrow in the protein gel in
the right panel. Positive control of lane 1 also contained ADA2bM.

173

B

ADA2b rv DRerGK215Kvarsvm225SvaLsm
Histone 33 Ta Rxs'rGGK“ apart

ADA2a gs DRSVGEK257KLRLPGEK265VPLVTELYGYN
Histone H3 Ta Rxs'rGGK“ APRK

Figure 4. Defining the acetylation site of ADA2b N-terminus.

(A) Acetylation assay of various deletions of the N-terminus of ADA2b. About
3 pg ADA2b (1-225, 28-225, 28-179, 1-208) and 2pg ADA2b(l-l79) were
acetylated by about 0.3 pg GCN5 with 0.15 pCi [3H] Ac-CoA. About lpg
ADA2bM (amino acids 225-377) was also included in the reactions to enhance
the acetylation. The acetylation reactions were resolved in a 12% SDS-PAGE
gel and the proteins in gel were Coomassie Blue stained first, as shown in the
lower panel. The included ADA2bM is indicated by the filled triangle. The
truncated proteins of ADA2bN are indicated by the arrows. The acetylated
proteins in gel were then visualized by enhanced fluorography as shown at the
upper panel. (B) Sequence alignment of the partial ADA2a and ADA2b N-
termini against the N-tail of histone H3. The partial sequence of ADA2b N-
terrninus includes the region of amino acids 208-225 required for acetylation.
The serine and lysine residues are colored.

174

 

acetyl CoA. The results showed that only two deleted version of ADA2b (amino acids 1-
225, and 28-225) could be acetylated (Figure 4). This suggested that the peptide region at
arrrino acids 208-225 was necessary for acetylation by GCN5. Interestingly, this region is
not present in ADA2 of yeast or metazoans, suggesting the uniqueness of ADA2
acetylation in Arabidopsis. Moreover, a stretch of nine amino acids within this region
was found very similar to that of histone H3 N-tail (Figure 4B). The histone H3 N-tail
contains two noteworthy residues, lysine 14 and serine 10. The lysinel4 is the major site
acetylated by yeast GCN 5 in vitro. The serine 10 is phophorylated and its

phosphorylation enhances lysine 14 acetylation by Gcn5 in yeast (Lo, et al., 2000).

 

Corresponding residues also exist in the nine amino acids region of the N-terminus of
ADA2b. A similar sequence is also present in the corresponding region of the N-terminus
of ADA2a, but the ADA2aN seemed to have two similar stretches (Figure 4B).

The GCN5/PCAF protein does not recognize a precise motif to acetylate, but a
general consensus sequence of GKXP has been suggested (Marrnorstein, 2001), where
the lysine is the acetylated residue. Sequence checking of the histone H3 tail-like region
of ADA2a and ADA2b suggested that K215 might be the potential acetylation sites in
ADA2b, and K257 and K265 in ADA2a (Figure 4B).

An alternative approach was also carried out to support this prediction. A parallel
HAT assay showed that Gcn5 of budding yeast was able to acetylate Arabidopsis ADA2b
(Figure 5), although the yeast Gcn5 was less efficient in acetylating Arabidopsis ADA2b
than Arabidopsis GCN5. This revealed a common acetylation theme in protein ADA2b
by GCN5 family protein.

In order to test the prediction, mutations were introduced into full-length ADA2b

175

yGCNS AtGCNS yGCNS AtGCNS

 

-. 1 2 3 1 2 3 1 2 3 1 2 3
In " v - . _ .
in U.” haw-:- f‘: i
1 a l u . 5
. H
--. r“ '

4'" ' ‘ .
-v-III

Figure 5. Yeast Gcn5 acetylates Arabidopsis ADA2b.

The HAT assay was performed in 26 p1 HEPES buffer with about
0.5 pg GST-ADA2b (wt,and mutant K215A) or 20 pg core histories,
0.15 pCi [3H] Ac-CoA, 0.5 pg Arabidopsis GCN5 (AtGCNS) or
about 1 pg purified recombinant yeast Gcn5 (yGCNS). The reaction
proceeded for 30 min at 26°C. Then the reactions were analyzed by
10% SDS-PAGE gel. The proteins in gel were first stained with
Coomassie Blue as shown in the lefi. Then the acetylation signals of
GST-ADA2b and histones were detected by enhanced fluorography,
as shown in the right. 1: core histones, 2: GST-ADA2b (K215A),

3: GST-ADA2b (wt).

176

 

such that K215 was changed to arginine or alanine. The mutant proteins were assayed for
acetylation by GCN5. The results showed that this point mutation of K215 either to
arginine or alanine abolished acetylation of ADA2b by GCN5, while control wild type of
ADA2b remained acetylated (Figure 6). Since arginine is very similar to lysine, there
should not be big conformational change of ADA2b protein when K215 was changed to
arginine.

A similar strategy was also applied to ADA2a. Since its fill] length had been hard to

 

purify, the assay was still focused on the N-terminus of ADA23 (amino acids 1-272).

Residues K257 or K265 in the N-terminus of ADA2a were changed to arginine or alanine.

In. -

The mutant recombinant proteins were purified flom E. cali as GST fusions. Acetylation
assay showed that the N-terminus was no longer acetylated by GCN 5 when K257 was
changed to arginine or alanine, whereas the acetylation capacity still persisted when
K265 was changed to arginine or alanine (Figure 6).

This mutational analysis suggested that K215 in ADA2b and K257 in ADA2a are
the sites of acetylation by GCN5 in vitro. Curiously, about 20 amino acids around the
lysine residue are also found to exist in ADA2 homologous flom plant rice and maize,

but not in ADA2$ flom budding yeast or any know metazoan organisms (Figure 6B).

Determination of the aceglation site of ADA2bN by mass spectrometly
The HAT assay result of mutant ADA2b demonstrated only the necessity of K215

for acetylation. It did not prove the sufficiency of K215 for acetylation. To directly
determine which lysine residue(s) in ADA2b was acetylated by GCN5, the method of

mass spectrometry was approached. To simplify the analysis, only the N-terminus of

177

A ADA2b(f'I) ADA2a(1-274) ADA2b(fI) ADA2a(l-274)
6? «3' «Vi-3'63? a? 41' «V 5’6“?"
x x e b e b
«it a a «M sum 4“ era‘sw’ewe:
' ,1 'm -.. - —- -

‘3 ifs-h m

J.

V.“
in... ..-

u‘, .--~.

B athADAZa QSD G 57RLRLPGEK265VPLVTELYGYN
athADAZb FVD 15xpvsrsvnn SLV ELSNYN
zmaADAZ nvn 53xparsanze P SL TELSGYN
OsaADA2 nvo xparsannc P SL TELSGYN
Histone 33 T 4 323x

Figure 6. Lysine residues in ADA2a (1-274) and ADA2b necessary for acetylation.
(A) Testing the necessary lysine residues in ADA2a and ADA2b. The recombinant GST
fusion proteins GST-ADA2a (1-274) and GST-ADA2b were either wild type (wt) or
mutant where lysine residue at position 215 (for ADA2b), or 257 and 265 (for the N-
temlinus of ADA2a) was changed to arginine or alanine. These wild type and mutant
recombinants bound to Sepharose beads were incubated with about 0.5 pg GCN5, 0.2
pCi [3H] Ac-CoA for acetylation. After the acetylation was finished, the bound GST
fusion proteins were eluted by boiling in SDS-PAGE loading buffer and electrophoresed
in a 10% SDS-PAGE gel. The resolved protein in gel were stained with Coomassie Blue,
as shown in the right. Then the acetylated signals were visualized by enhanced fluoro-
graphy, as shown in the left. WT: wild type, K215R: ADA2b with lysine residue at
position 215 mutated to arginine, and so on. (B) Sequence alignment shows that a stretch
of amino acids around the lysine residues required for acetylation are found in ADA2
homologs flom plants. athADA2: Arabidopsis ADA2, zmaADA2: corn ADA2,
osaADA2: rice ADA2.

178

 

ADA2b (1-225, ADA2bN) was subjected to mass spectrometry.

In the mass spectrometry assay, ADA2bN was acetylated by GCN5 using cold
acetyl CoA. Then the acetylated ADA2bN was SDS-PAGE purified and sent to the mass
spectrometry facility at our department for further analysis, where the ADA2bN was first
digested in gel by trypsin and analyzed by LC/MS/MS mass spectrometry.

The mass spectra showed that two peptide flagments were singly acetylated. One
was the flagrnent of SKmEQCLEH, where the K137 was acetylated. The second flagment
was SFGGKZ’SKPVSTSVNNSZ” GSIELER, in which the GSIELER was the expressed
peptide flom the GST vector sequence. For the second peptide flagment, it was
ambiguous from the data as to whether K215 or K216 was acetylated because of the
flagment was too big.

Since the uncertainty of K215 or K216 acetylation was due to the large-sized peptide,
smaller recombinant ADA2bN was prepared for mass spectrometry again. To do so, the
original ADA2bN (1 -225) expressing plasmid was re-subcloned so that 4 amino acids of
VNNSZ” at the C-terminal end were deleted. The shorted ADA2bN was named
ADA2bNR and the amino acids at the C-terrninal end flom expressed plasmid
were ...RSFVDRSFGGK2'5KPVSTSSGRIVTD where peptide of GRIVTD was flom
expressed vector sequence. Now the trypsin digestion peptide flom ADA2bNR flanking
K215 was shorter: SFGGKZ'SKPVSTSSGR. The recombinant protein of ADA2bNR was
expressed and purified as for ADA2bN. For mass spectrometry assay, the recombinant
ADA2bNR was acetylated by GCN5 using tritium-labeled acetyl CoA. The reason for
using tritium-labeled Ac-CoA was to label the acetylated flagrnent that could be purified

after HPLC treatment. As done for ADA2bN, the acetylated ADA2bNR was SDS-PAGE

179

 

gel purified and sent to the mass spectrometry facility of our department, where the
sample was digested with trypsin in gel and flactionated in two separate HPLC runs.

The HPLC flactions with peak tritium activity were subjected to MALDI mass
spectrometry. MALDI results of the sample identified three peptides with possible
acetylation. M/Z 1335.17 Da corresponded to Arabidopsis ADA2b a 45-52 with one
acetylated residue; M/Z 1350.08 Da corresponded to aa 17 8-1 87 with two acetylated
residues; M/Z 1436.69 Da corresponded to aa 211-224 with one acetylated residue. aa
211-224 is the sequence of SFGGKZISKPVSTSSGR.

Since the MALDI data did not tell which K was acetylated in peptide aa 211-224,
one saved HPLC flaction with peak tritium activity was further subjected to Q-TOF
analysis. From Mascot Search, ADA2b was identified as the top hit with a score of over
75 by the Q-TOF mass data. M/Z 1350 Da was found to contain no acetylation sites, (for
the real mass plus two acetylations should be 1333.4. It is not known how it was picked
up as the peptide aa 178-187 with two acetylations in previous MALDI assay). MS/MS
of 1437 Da identified the predicted peptide SF GGK215 KPVSTSSGR with one acetylation
site. The detailed calculated masses for this peptide are listed in Table 7-1. The
highlighted data in the table were the identified masses flom Q-TOF. The Q—TOF mass
data only matched the prediction that the K5 (i.e. K215 of ADA2b) was acetylated. The
data did not match acetylation at K6, or null acetylation (see the y and b ions data in
Table 7-1 below). Therefore it was certain that K215 of ADA2b was the acetylation site.
The key mass data to determine that K5 was acetylated rather than K6 were the mass data
underlined in the table.

In surrunary, the mass spectrometry data identified that K215 was acetylated in

180

w-

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

2 2w 2: E m _ 2: 2:: E m _ No: we: .1 m
N 3: N.NNe 2 e N _.NNN Nos 2 o N NNNN «$2 2 o
N N52 38 Na m N Na: eéNz Na m N no: N82 N_ m
4 SN Sew : m 4 Nos. 3:: : m e 4.84 NS: : m

a SN 33 S .N m Son 22: 2 e m 3% Na: 2 e
e NNoN 33. o m e «room 38 o m e 3% 38 o m
N 46% nNNo N > N 3% 2% w > N 3% 3% N >
N 3% N: N a m 4.85 3.: N a N 23 3.: N a
a 3.3. oeNN e v. o we: 2% c on a ea; 2% e a
S “.me 28 m a 2 3&2 Nam m z 2 News 3: m v.
: 5% No: a. o : one: Noam o o : Nee: was a o
N_ N.NS 3.1 N o N_ 982 eNoN N 0 Na N.8N_ NNaN N o
2 $3 2: N a 2 59.: EN N a 2 new: nmNN N a
3 no: as. a m 3 SN: 9% a m 3 EN: 5w _ m
a a e a com a a c a as a a e a com

+N was. .1 one: i we...

 

 

“ZaNaNA—d. ..8 3.3 9.5: «HQ—NO min 03.3.

181

ADA2b. Some evidence suggests acetylation on lysine residues such as K137 (in peptide
SKmEQCLEH ). But it seems unlikely that K137 is acetylated for the deletion and

mutant acetylation assays described earlier does not support this finding.

Part 11 Test the sigpificance of ADA2b acetyaltion in viva
Although the previous section clearly shows that ADA2a and ADA2b proteins can

be acetylated by GCN5 in vitro, it is not known whether ADA2 is acetylated in viva, nor
is the biological significance of acetylation known. Experiments were carried out to test
whether the mutant ADA2b is able to enhance GCN5 HAT activity and to be functional

in viva to understand some of the biological function of ADA2 acetylation.

Mutant ADA2b is able to enhance GNCS HAT activity.
Recombinant GST-ADA2b (K/A or K/R) proteins were purified the same way as

that of wild type GST-ADA2b (wild type, see chapter IV). HAT assays were carried out
using the same conditions as for wild type GST-ADA2b (see chapter IV). As shown in
Figure 7A, these two ADA2b mutants enhanced GCN5 HAT activity on flee core
histones as good as wild type ADA2b did, and the mutants did not change GCN5
substrate specificity (Figure 78). Meanwhile, nucleosomal HAT assays also showed that
mutant ADA2b proteins behaved like wild type ADA2b in stimulating GCN5 HAT
(Figure 8). Therefore, lysine 215 in ADA2b is not important for the enhancement of

GCN5 HAT activity.

Dwarf and sterile ADAZbJ- Arabidopsis plants

182

 

-.
..
!|—-—

 

A CPM

 

 

 

 

7000
6000‘
5000‘
4000‘
3000‘
2000i
1000‘
‘ A A A 4‘
GST WT K/R K/A
B .-» .-_.___ .._.___.,
W" " "' “'ga
“in

a“ .v.;c

      

12341234

Figure 7. The ADA2b mutants enhance the HAT activity of GCN5 on
free core histories.

HAT assays were carried out with purified recombinant GCN5, [H3] labeled
Ac-CoA, flee core histones and the purified recombinant ADA2b proteins as
GST fusion in wild type (WT), or alanine (K/A) or arginine (K/R) substitution
at lysine 215. The the HAT activities wer quantitated by liquid scintillation
counting on filter paper as shown by CPM readings in chart A. The reactions
were also analyzed in a 15% SDS-PAGE gel. The proteins in gel were stained
with Coomassie Blue, as shown in the right of panel B and then the acetylation
signals were detected by enhanced fluorography. The acetylated wild type
ADA2b is indicated by the arrow in panel B.

183

 

 

8M 4M —8"M —4PM
_.P__P_ WAKGWAKG
WAKGWAKG .- .

acetyalted
"- 3| I m
‘2 by GCN5

d

    
 

 

Figure 8. Mutant ADA2b proteins stimulate GCN5 HAT activity on

nucleosomal histone.

The HAT assays were performed with purified turkey erythrocyte nucleosomes, [3 H]
labeled Ac-CoA, GCN5 and GST fusion protein of wild type ADA2b (W), or alanine
(A) or arginine (R) substitution at lysine K215 of ADA2b, or control GST (G). About

8 pm and 4 pm ADA2b as GST fusions were used in the assays]. The reaction products
were analyzed by SDS-PAGE, and stained by Coomassie Blue (as shown in the lefi
panel) and then subjected to enhanced fluorography for 4 months (as shown in the right
panel).

184

 

Our lab has an Arabidopsis line which has an insertion of T-DNA at the ADA 2b gene.
The insertion site is at the fifth intron, and the ADA2b gene can only express the first
four exons that encode approximately 160 amino acids of the N-terminal of ADA2b
protein (V lachonasios, et al., 2003). Phenotypically, plants with heterozygous T-DNA
insertion at ADA2b loci (designated as ADA2b +/-) are wild type, while plants with
homozygous T-DNA insertion at ADA2b loci (designated as ADA2b-/-, or adaZb-I in the
literature) show pleiotropic phenotypes such as dwarf and sterility, as shown in Figure 9

(V lachonasios, et al., 2003). Under normal laboratory growth conditions, the two severe

 

phenotypes of ADA 2b-/- plants are rescued by transgenic overexpression of ADA2b

ll."—

cDNA encoding the flrll- length ADA2b protein.

Transgenic stratgn

Therefore, the ADA 2b-/- plants provide a good system to test the function of mutant
ADA2b which can no longer be acetylated by GCN5 in vitro. The approach was to
express mutant ADA2b (K215R and K215A) in ADA2b-/- plants to investigate the degree
of phenotype rescue.

To do so, ADA2b +/- plants (ADA2b-/- plants are sterile) were transformed with a
specific binary plasrrrid that contained the mutated cDNA encoding ADA2b (K215R) or
ADA2b (K215A). The transformation resulted in random insertion of the mutated cDNAs
into ADA 2b +/- plant genome together with an herbicide resistant marker bar] . The
expression of both the mutated cDNA and the bar] gene were under the control of the
strong 358 promoter derived flom cauliflower mosaic virus. Therefore, positive

transgenic plant offspring could be screened by the feature of being resistant to herbicide

185

 

Figure 9. The dwarf phenotype of Arabidopsis with homozygous T-DNA
insertion at the ADA2b locus.

The big plant is wild type with heterozygous T-DNA insertion at the ADA2b locus
(genotype: ADA2b+/-). The three dwarf plants have homozygous T-DNA insertion
at the ADA2b locus (genotype: ADA2b-/-).

Picture is provided by Dr. Kostas Vlachonasios.

186

Basta. A positive control encoding wild type ADA2b was also included in parallel
transformants.

Totally 4 ADA2b +/- plants were transformed each by the plasmids containing cDNA
of protein ADA2b (K215R), ADA2b (K215A), and ADA2b (wt, wild type). Seeds were
collected flom these plants as stage Tl . These T] seeds were allowed to germinate on
solid Gamborg’s B-5 medium with Basta selection. Resistant seedlings (about 7-10 days)
were transferred to soil to grow for both genotype identification and grth phenotype
examination.

Plant offspring with the genotype of ADA2b-/- were identified by the method of PCR
screening. Single plant genomic DNA was extracted flom one rosette leaf, and the
ADA2b gene flagment spanning the T-DNA insertion site was PCR amplified using
primers flanking the site of T-DNA insertion. The ADA 2b gene with T-DNA insertion
gave a PCR product of about 8.2 kb in size and the ADA2b gene without T-DNA
insertion gave a PCR product of about 2.7 kb. An example of determining plant
genotypes is shown in Figure 10A. Afier determining plant genotype, the integration of
transgenic cDNA of ADA2b was also checked by PCR method using an ADA2b cDNA
and 35S promoter specific primers which gave a PCR product of about 1.3 kb (Figure

103).

ADA2b 15R can com lement ADA2b-/- Iant heno
To check the function of protein ADA2b (K215R) in viva, plants with genotype of
ADA2b-/- and integration of ADA2b (K215) cDNA [designated as ADA2b-/-:ADA2b

(K215R)] were first identified and then checked for their growth phenotype.

187

 

 

 

sr7so T-DNA sroz9
i ADA2b gene
ADA2b
gene promoter
B
ST596 STQ29
ADA2b cDNA

 

prorsnoter

 

Figure 10. Genotyping the offspring of transformed plants.

The plants’ genomic DNA was amplified by PCR and the products were analyzed in 1%
agarose gel and the PCR products were visualized by ethidium bromide staining.

(A) PCR analysis to determine the plants’ genotypes. The used PCR primers’ positions
(ST780 and STQ29) in ADA2b gene in respective of the T—DNA insertion site are
indicated in the right diagram. The 8.2 kb and 2.7 kb PCR products are indicated by the
arrows. The genotypes of 1, 2, 3, and 4 plants are listed as the followings, l, 2:
ADA2b+/—, 3: ADA2b-/-, 4: ADA2b+/+. (B) Determining the integration of ADA2b
cDNA into the genome in the transformed plants. The PCR primers (ST596 and STQ29)
are 358 promoter and ADA2b cDNA specific as indicated in the right diagram. The 1.3
kb PCR products are indicated by the arrow.

188

Three such plants were identified, designated as R3, R5 and R12. Line R3 and R5
showed the mutant phenotypes, whereas R12 displayed a near-normal growth phenotype
with reduced seed yields (Figure 11). This result indicated that the K215R mutant protein
can suffice for most of the discemable biological activity of ADA2b protein, and thus the
acetylation of this K215 is not essential for those biological activities.

One explanation for the lack of complement in line R3 and R5 might be that the
transgene was integrated but not expressed. To test this hypothesis, the steady-state levels

of ADA2b mRNA were assayed by Northern blotting. As shown in Figure 12, the mRNA

 

levels of ADA2b mutant in line R5 are comparable to that in line K61 (genotype:
ADA2b-/-:ADA2b (wt)), which carried the wild type transgene and showed normal
growth phenotype. Expression of the K215R transgene in line R3 is much lower, but still
comparable to the levels of that of endogenous ADA2b. These results suggest that levels
of expression of the transgene might important in performing complementation.

To further characterize the ability of the K215R transgene to rescue the ADA 2b-/-
phenotype, nine T1 lines were selected with genotype of ADA2b +/-:ADA2b (K215R). T2
progeny of these lines were screened by PCR to identify plants with genotype of ADA2b-
/-:ADA2b (K215R). For two T] lines (designated as R45 and R48), 8 out of 8 dwarf T2 1
plants had the genotype of ADA2b-/-:ADA2b (K215R), whereas none of the 28 normal
T2 plants examined had the genotype of ADA2b-/-:ADA2b (K215R). Expression of the
K215R transgene was low or undetectable in these dwarf T2 plants as shown in Figure 13,
reinforcing the conclusion that the expression level of the transgene might be critical for
complementation.

Another T1 line R44 yielded T2 progeny with the genotype of ADA2b-/-:ADA2b

189

 

 

Figure 11. Comparison of the silique length of R12 plant to that of wild type plant.
Plant R12 seeds were in T2 stage flom parents that had the genotype of ADA2b-/- with
the integration of the cDNA of mutant ADA2b (K215R) [designated as ADA2b-/-:
ADA2b (K215R)]. The wild type and R12 plants seeds were geminated on Garnborg B-5
solid medium and planted in soil. The pictures were taken when the plants had been
growing in the soil for 23 days. The genotype of plant R12 was ADA2b-/-:ADA2b
(K215R). The short siliques in plant R12 are indicated by the arrows. Both figures are
shown at the same scale.

190

wt R1 R3 R5 58 K6] mut

 
 
  

wild type mRNA
flafiéeflfc cDNA
partial mRNA
of wild type

..ggue-ueeOU

 

Figure 12. Over-expression of transgenic cDNA of ADA2b(K215R)

did not rescue the ADA2b-l- phenotype.

Nothem blotting was performed to detect the transcripts of ADA2b cDNA that was
in RNA samples were prepared from the plants on 2] day old (a), 23 day old (b)
and 25 day (c). After electrophoresis in a 1% agarose gel , the rRNAs in gel were
stained with ethidium bromide as shown in the lower picture. The Norther results
are shown in the upper picture. mRNA of The genotype of these plants were listed
above. R1: ADA2b+/-:ADA2b (K215R), R3: ADA2b-/-:ADA2b (K215R),

R5: ADA2b-/-:ADA2b (K215R), mut: ADA2b-/-, 58: ADA2b+/-,

K61: ADA2b-/-:ADA2b (wild type).

191

 

>th e9 e

:4.

R44 R44 R44 R44 K30 K15 R48 R48 R44 R48 R48 R48 R48 R45 R45

.- «and..-~ - .. -..

 

Figure 13. Over expression of transgene is required to rescue the ADA2b-l-
phenotype.

Northern blotting was performed to detect the expression level of transgenic cDNA.
About 5 pg total RNA sample, that was extracted flom rosette leaf at day 20 afier
the seedling was transfened to soil, was loaded for each plant as indicted by the
rRNAs in the lower panel. The detected mRNA levels of transgenic cDNA either as
wild type (plant K15 and K30) or mutant (K215R, the rest plants) are shown in the
upper panel as indicated by the arrow. All plants were dwarf except plant K15 and
K30 which were normal in size, plant R44-13 which was small but not dwarf, and
plant R44-5 which was dwarf and died prematurely. The genotype of the plants are
listed as the followings, K15, K30: ADA2b-/-:ADA2b(wil type), the rest: ADA2b-/-
:ADA2b(K215R).

192

(K215R). Some of these, displayed a near normal phenotype although the siliques were
often reduced in size and number (such as line R44-21), while five others showed the
mutant phenotypes (designated as line R44-5, -6, -7, -13 and -14). The mRNA levels of
the transgene in line R44-6, 77 and -14 were low, while the mRNA level in line R44-13
was comparable to that of line K15 and K30 that were genotype ADA2b-l- and rescued by
transgenic overexpression of wild type ADA2b (Figure 13). Line R44-5, which had a
dwarf phenotype but high levels of transgene expression (Figure 13), died before

maturity for unknown reason. But lines R44-5, -13 and R5 as described above suggest
that there is exception to the general observation that expression of K215R transgene is
able to complement.

In another T1 line (designated as R16), three T2 plants flom all examined 12 plants
were identified as ADA2b-/-:ADA2b (K215R) (the rest had wild type ADA2b gene).
These three plants (designated as R16-32, R16-44, and R16-48) showed normal growth
and seed production, although one plant had somewhat smaller roseate leaves, and all the
three plants had fewer siliques than the control with the genotype ADA2b +/-. This results
showed that expression of K215R in T2 can complement in ADA2b-l- plants. The growth
state of the three plants after 22 days in soil was shown in Figure 14.

For lines R12, R44-21 and R16-48 (as described above), general grth and
transgene expression were examined again in their T3 plants in comparison to that of
plants phenotypically complemented by wild type transgene (line K51 and K6], genotype
of ADA2b-/-:ADA2b(wt)). As shown in Figure 15, these plants grew to normal size, and
produced good amount of seeds (comparable to that of wild type). The steady state

mRNA levels of the transgenes in these plants were also similar and they all were

193

 

Figure 14. The phenotype rescue of plant ADA2b-l- by the expression of
transgenic cDNA of ADA2b (K215R) in line R16.

These plants were at T2 stage flom T1 line R16. The growth status of the rescued plant
R16-48, R16-45 and R16-32 is compared to that of the wild type plants which are the
rest plants. The picture was taken when the plants had been growing in the pot for 22
days. The genotype of R16-48, R16-45 and R16-32 plants are ADA2b-/-:ADA2b
(K215R).

194

 

 

ST486 T6 K51 T3
wild type ADA2b-/-:ADA2b (wild type)

    

v

 

R16-48 T3 ADA2b-F
ADA2b-/-: ADA2b (K215R)

Figure 15. The acetylation of ADA2b at lysine 215 in vivo is not required

for plant development.

These plant seeds germinated on solid Gamborg’s B-5 medium without Basta
selection. The geminated seedlings were transferred to soil pots. Pictures were
taken after the plant had been growing in the pot for 10 days. The plants genotype
as well as T stage are indicated. The last figure shows the severe dwarf phenotype
of plant with homozygous T-DNA insertion at ADA2b locus.

195

overexpressed (Figure 16). The most obvious distinction flom wild type plants was that
these plants (including K51 and K61) had some shorter siliques along the shoots (see
Figure 1] for illustration) and these plants were delayed in flowering by 2-3 days. Since
this defect also happened in plants K51 and K6], it was not unique to plant R12, R44-21,
and R16-48. The expression of transgene in R12, R44-21 and R16-48 was as the same
level as that of line K51 and K6], but they all had much higher expression than

endogenous ADA2b gene.

ADA2b 15A can com lement ADA2b-A Iant heno e

The described data above suggest that acetylation of ADA2b at K215 in viva is not
required for normal function. Similar complementation experiments were performed to
test whether the K215 is important for ADA2b by changing the residue to alanine.

As performed for K215R screening, T1 plants flom transformed parents with K215A
cDNA were screened by PCR as described above to identify genotype of ADA2b-l-
:ADA2b (K215A). In one screening, 34 T1 plants (17 normal size and 17 dwarves) were
screened. Of these, 5 dwarf plants were genotyped as ADA2b-/-:ADA2b (K215A), none
of the normal size plant had this genotype.

In another screening, 24 normal-sized Tl plants were screened (dwarf plants were
not screened). One plant, designated as A14, had the genotype of ADA2b-/-:ADA2b
(K215A). Plant A14 had normal size and produced similar seeds as that of plant R16
identified above with genotype of ADA2b-/-:ADA2b(K/R).

To assess the grth of A14, T2 stage plants of A14 and T4 plants of R16-48 and

K5] as well as wild type were grown for comparison. These plant seeds were germinated

196

 

 

 

wt K51 Rl6R44 R12 K61 wt wt K51 Rl6R44 R12 K61 wt
-48 -21 -48 -21

   

Figure 16. Over expression of transgenic cDNA is required to rescue

ADA2b-l- phenotype.

Northern blotting was performed to show the mRNA level of either wild type transgenic
cDNA in plant K61 and K5], or mutated transgenic cDNA (K215R) in plants R12, R16-
48 and R44-21. All the plants were nonna] in size. Two wild type plants were also
included. About 5 pg total RNA, that was extract from the plant rosette leaf at day 20
afier the seedling was transferred to soil, was loaded in the agarose gel for each plant.
The right panel shows the rRNAs for each sample after electrophoresis in agarose gel,
indicating the equal loading for each sample. The result of Northern blotting by auto-
radiography is shown in the left panel. The black arrow indicates the probed mRNA of
ADA2b transgenic cDNA. The probed mRNA of wild type endogenous ADA2b gene is
indicated by blue arrow. The genotype for plants K61 and K5] was ADA2b-/-:ADA2b
(wild type), and for plants R12, R16-48, R44-21 is ADA2b-/-:ADA2b(K215R).

wt: wild type.

197

 

in Gamborg’s medium without Basta selection. For A14, 132 T2 seeds were germinated,
39 yielded dwarf plants indicating heterozygous status of the transgene. The T2 plants of
A14 growmg in soil were very similar to that of R16-48 and K5] as shown in Figure 17.
All plants of R16-48, K51 and K14, compared to wild type control, had some shorter
siliques, and were delayed 2-3 days in flowering. The results illustrate that ADA2b
(K215A) can rescue the phenotype of ADA2b-/— plant, suggesting the lysine residue at
215 is not important for plant normal growth under laboratory conditions.

In another screening of T2 plants flom T1 parent plant (designated as A4) that had
the genotype ADA2b +/-:ADA2b (K215A), 25 T2 plants (including 4 dwarves, 2 small
plants and 3 short plants) were genotyped. Only the three short plants and the two small
plants were found to be the genotype of ADA2b-/-:ADA2b (K/A). These short plants had
normal-sized and dark leaves, the leaves were curved, and these three plants were a few
days delayed in inflorescent shooting. These three plants never produced seeds. The two
small plants were about the double size of dwarf plant of genotype of ADA2b-/-. These
small and short plants were all sterile. The growth of these small and short plants was
shown in Figure 18 and Figure 19 in comparison with wild type plants.

Similarly, in another screening of T2 plants flom T1 parent plant (designated as A21)
that had genotype of ADA2b +/-:ADA2b (K215A), 24 T2 plants (including 2 dwarf and 2
small and 4 short plants) were screened. All the dwarf, small and short plants were
identified to be the genotype of ADA 2b-/-:ADA2b (K215A). These 4 short plants had the
same phenotype as that of the short plants described above for A4.

To examine the expression of the transgene in these dwarf, short and small plants,

Northern blotting was performed. As shown in Figure 20, they all had overexpression of

198

      

plant K51 T4
ADA2b-l-: ADA2b (wild type) ADA2b-l-= ADA2b (K215R)

   

plant A14 T2
ADA2b-l—: ADA2b (K215A)

wild type and dwarf

Figure 17. Expression of transgenic cDNA (K215A) rescues plant

ADA2b-l- phenotype.

These plant seeds germinated on solid Gamborg’s B-S medium without Basta
selection. The germinated seedlings were transferred to soil pots. Pictures were
taken afier the plant had been growing in the pot for 10 days. The plants genotype
as well as the T stage are indicated. The right bottom figure also shows a dwarf
plant with genotype of ADA2b-/-.

199

 

 

 

Figure 18. Partial rescue of phenotype of plant ADA2b-l— by over-expression
of transgenic cDNA of ADA2b (K215A).

The plants were all at T2 stage flom A4 transformed parent. The growth status of
three short plants (A, B and C) and two small plants (D and B) were compared to
that of the three wild type plants (X,Y and Z). The short plants had normal leaf
size but short inflorescent shoots. The small plants were about the double size of
ADA2b-/- dwarf plants. The picture was taken when the plants had been growing
in the pot for 18 days. The genotype of the small and short plants was
ADA2b-/-:ADA2b (K215A).

200

 

Figure 19. Partial rescue of the phenotype of plant ADA2b-l- by overexpression of
transgenic cDNA of ADA2b (K215A).

The growth status of the short plant C was compared to that of the wild type plant Y in
later growth stage. Their growth in earlier stage was shown in Figure 18. The insert
shows the whole pot with all the plants shown in Figure 18. The short plant had shorter
and bushy shoots, and was sterile although it did bear some short siliques. The wild type
plant bore mature siliques. The picture was taken when the plants had been growing in
the pot for 30 days. The genotype of the small C plant was ADA2b-/-:ADA2b (K215A).

201

 

 

Q .. ”mm-om-

K15 1 2 5 38 8 4s 4b 21s 21b 21b2 K30K61 r7 K51

_..- “---~----e~ m aw.-

 

Figure 20. Complementation variation with over-expression of transgenic cDNA

to rescue ADA2b-l- phenotype.

Northern blotting was performed to detect the mRNA levels of the transgenic cDNA in
the ADA2b-/- plants. The RNA samples were prepared when the plants were 27 day old
in soil. The RAN samples were loaded in equal total amount for each plant, as indicated
by the rRNAs in lower panel which were stained by ethidium bromide. The results of

the Northern blotting is shown in the upper panel. All plants were in T2 stage except

K15 and K30 which were in T1 stage. Plant 4s and 4b were the “small” and “short” plant
respectively referred in the results part for line A4, and the same case for plant 21s and
21b. Plant 21b2 was a second “short” plant of A21 line. The rest plants were identified in
parallel screenings unless specified. The genotypes of all plants were ADA2b-l- except
plant 38 (ADA2b+/-), and the transgenic cDNA for each plant and its growth phenotypes
is listed in the following. 1: :ADA2b(K215R), dwarf and sterile, 2: :ADA2b(K215A),
dwarf and sterile, 5: :ADA2b(K215R), dwarf and sterile, 8: :ADA2b(K215R), small size,
sterile, 45, 21s: :ADA2b(K215A), about the double size of dwarf plant and sterile, 4b,
21b, 21b2: :ADA2b(K215A), wild type leaf, short and delayed fluorescent shoots, K51,
K15, K30: :ADA2b(wt), wild type, K61: :ADA2b(wt), wild type size, some shorter
siliques, K61: :ADA2b(wt), wild type size, some shorter siliques, f7: :ADA2b(wt), wild
type size leaf, delayed shooting and half normal length of shoot, and sterile.

202

 

 

the transgene, and the levels were even to that of plant K6] , which, as mentioned
above, had genotype of ADA2b-/-:ADA2b (wt) and the phenotype was complemented.

All the results suggest that both K21R and K215A mutants can complement wild
type ADA2b in viva when they are over-expression. It is not known whether similar
endogenous level expression of mutant proteins can also complement. However, the
overexpression of K215R or K215A mutant can not always complement as described
above. The reason for this variation is not known. Curiously, the phenotype
complementation variation is not unique to the transgene of K2] SR and K215A, but also
happened during the complementation by the wild type transgene. For example, one
screening was performed to identify ADA 2b-/- plants complemented by the wild type
transgene. Totally 117 normal plants and three short T2 plants were screened by PCR
from 11 lines that were genotyped as ADA 2b +/-:ADA2b (wt). Only these three short
plants (designated as a3, e9 and f7) were genotyped as ADA2b-/-:ADA2b (wt), the
remaining 117 plants were genotyped as either ADA2b+/- or ADA2b+/+. Plant a3 had
small leaf (but not dwarf) and dwarf inflorescent shoots. Plants e9 and f7 had normal size
leaf but darker in color, dwarf shoots in early stage, and their shoots grew to wild type in
later stage (42 days afier the seedlings transferred to soil), but had shorter and sterile
siliques. The mRNA level of the wild type transgene in plant f7 was examined. As shown

in Figure 20, plant f7 had similar level of transgene expression as that of plant K61 .

Discussion

The major conclusions of this work are that (i) Arabidopsis protein ADA2, including

ADA2a and ADA2b, can be acetylated by GCN5 in vitro, and this acetylation does not

203

happen to yeast ADA2 protein, (ii) the acetylation site was identified to be residue of
lysine 215 in ADA2b, (iii) the peptide sequence surrounding the acetylation site is similar
to the N-tail of histone H3 and is unique to plant ADA2, (iv) the acetylation of lysine 215
is not essential for ADA2b protein flmction in viva for normal plant growth under
laboratory conditions.

That the mutant ADA2b protein (K215R and K215A) behaves like wild type in viva
raises a few questions. First, the acetylation of Arabidopsis ADA2 protein may be just an
in vitro artificial effect. Secondly, acetylation of ADA2 may occur in viva but its
importance may not be demonstrated in the assays we have used. For instance, ADA2
acetylation might happen only in specific biological responses in plants such as pathogen
attach response, UV damaged DNA repair, and so on. Since only the normal laboratory
growth condition has been tested to check the function of ADA2b mutants, this can not
exclude the possibility that in other stressful growth conditions the acetylation of ADA2
is required. Interestingly, very preliminary data indicate that plant seedlings of ADA2b-/-
:ADA2b (K215A) is weaker in resisting to rrricro-organisms attach than seedlings of
ADA2b-/-:ADA2b (wt). More careful testing is needed to support this finding. Curiously,
many stress response transcriptional factors in Arabidopsis implicated in pathogen attack,
dehydration and high salt treatment contain AP2 domain (Park, et al., 2001, Riechmann
and Meyerowitz, 1998, Stockinger, et al., 1997), and as shown in Chapter VI, ADA2
proteins can interact with some AP2 domain- containing transcriptional activators.

Another possible explanation for the complementation by K215 mutants in viva is
that the over-expression of transgene of ADA2b mutant might conceal the acetylation

importance, although it is not known whether the protein level of mutant protein is also

204

over-expressed in transgenic plants or not. Although ADA2a can not complement
ADA2b-/- plants, but the ADA2b acetylation per se might be compensated for by ADA23
acetylation. A recent research in budding yeast (Gu, et al., 2003) indicates that, for some
duplicated genes, the genetic defect can only been seen when both the duplicated genes
are not functional.

Regardless, the observation that K215 mutants can function in viva raises the key of
question whether ADA2 proteins in vivo are really acetylated. Theoretically, this question
can be answered in the following way. ADA2 protein could be first isolated flom plant
cell lysates by immuno-precipitation, perhaps in combination with other biochemical
method such as chromatography. Then the purified protein could be subjected to mass
spectrometry to determine if ADA2 is acetylated. Alternatively, the purified ADA2b
protein acetylation status could be determined by specific antibodies. The antibodies
could be raised against synthetic ADA2b peptide along the acetylation site in which the
lysine residue is either acetylated or unacetylated (as a negative control). Technically,
there are a few good sides of this approach. First, ADA2a or ADA2b antigen to make the
antibody is not limiting for, as shown at Chapter IV, the recombinant N-terminus and the
C-terminus of ADA2a and ADA2b could be purified in very good purity and quantity.
Secondly, the current mass spectrometry techniques are able to detect picogram scale
protein, therefore no huge amount of Arabidopsis plant material required. Finally,
specific antibodies against acetylated and unacetylated peptides have been successfully
used in many other similar occasions. The major concern of the above approach is that
the acetylation of ADA2 might occur only in unknown special biological responses.

The ADA2b-/- phenotype is generally rescued by mutant ADA2b protein, but some

205

phenotypic distinctions remained. Some siliques are shorter (as shown in Figure 11), and
flowering is delayed by about 2-3 days. These defects are also observed in ADA2b-/-
plants rescued by wild type ADA2 (such as in K61 and K5] plants). These defects are not
likely to arise due to simply the overexpression of ADA2b protein, since overexpression
of wild type or mutant ADA2b (K215R) does not produce these problems in plants with
wild type genotype.

That mutant ADA2b (K215R) behaves like ADA2b (K215A) is somewhat surprising.
Arginine is very similar to lysine, so substitution of K2] 5 by arginine should not change
ADA2b protein conformation very much, while alanine substitution might have dramatic
effect on the overall structure of ADA2b protein. However, two arguments support the
notion that the region flanking K215 in ADA2b may not take a defined structure. First,
the histone H3 N-tail, the native substrate of GCN5 protein, is a random coil (Luger, et al.,
1997). Secondly, the region flanking K215 in ADA2b is only found in plant ADA2, and
is not inside the conserved parts of ADA2 protein. These two facts might argue that the
region in ADA2b dose not take a specific conformation, therefore alanine substitution to

K215 will not affect the overall conformation of ADA2b protein.

As mentioned in the introduction, HIV Tat protein is acetylated by PCAF. The
acetylated sites in Tat are at K28 and K50 (Deng, et al., 2000). The sequence surrounding
K50 is very similar to that of ADA2b. In Tat the sequence is SYGRKK, and in ADA2b
the sequence is SFGGKK. Recently, acetylated Tat at K50 has been shown to bind
specifically to the bromodomain of PCAF and that this interaction competes effectively
against HIV -1 TAR RNA binding to the acetylated Tat (Mujtaba, et al., 2002).

Bromodomains are known to bind acetylated lysine residue (Zeng and Zhou, 2002) and

206

this domain is present in nearly all nuclear histone acetyltransferases including

Arabidopsis GCN5.

207

Chapter Eight

Perspective and Hypothesis

The GCN5-ADA2 complexes in Arabidopsis

This study show that the Arabidopsis GCN5 protein is a histone acetyltransferase
and associates with Arabidopsis ADA2a or ADA2b proteins in vitro. Based on these two
facts, it is hypothesized that there are GCN5-ADA2 complexes in Arabidopsis, and one
complex contains GCN5 and ADA2a proteins and the other contains GCN5 and ADA2b
proteins.

Recombinant GCN5 and ADA2b proteins together acetylate nucleosomal histone H3
weakly in comparison to the budding yeast HAT complex SAGA that has more than ten
subunits. The Arabidopsis GCN5-ADA2 complexes are hypothesized to contain
unidentified protein components in addition to the proteins GCN 5 and ADA2. One
function of the additional proteins is hypothesized to enable GCN5 to acetylate
nucleosomal histones effectively. The Arabidopsis genome sequence
[http://www.tigr.org/tdb/e2k1lathl/athl .shtrnl] encodes putative proteins homologous to
the proteins TAFS, TAF 6, TAF 9, TAF10, TAF12, and Tral found in budding yeast
SAGA complex, but the genome does not encode apparent homologues of Ada], Ada3,
AdaS/Spt20, Spt3, Spt7 and Spt8 also found in budding yeast SAGA complex (Grant, et
al., 1997) and Ahc] that is the component of budding yeast ADA complex (Eberharter, et
al., 1999). Therefore, it is likely that the Arabidopsis GCN5-ADA2 complexes have
distinct components. It is eminent in the next step to investigate what the other protein

Components are in the Arabidopsis GCN5-ADA2 complexes.

208

 

 

The elucidation of the components of Arabidopsis GCN5-ADA2 complexes will be
beneficial in many aspects. One key goal in the future research is to find out what the
biological roles the complexes are exactly play in viva. The identification of the
components might shed light on some biological roles the complexes play in viva. In
addition, the SAGA complex is well conserved in metazoans (Kusch, et al., 2003,
Ogryzko, et al., 1998). It is not known whether that Arabidopsis GCN5-ADA2 like
complexes are conserved in plants. GCN5 and ADA2 homologs have been found in
maize (Bhat, et al., 2003), implicating those other plants might have similar GCN5-
AAD2 complexes. The identification of components ofArabidapsis GCN5-ADA2
complexes might also pave the way for characterization of related GCN5-ADA2
complexes in other plants.

On the other hand, it will be very interesting to know how similar the GCN5-ADA2a
complex is to the GCN5-ADA2b complex in terms of the protein components. The thesis
study finds differences in the detailed interactions between GCN5-ADA2a and GCN5-
ADA2b. Genetics studies discovered that Arabidopsis plants with homozygous disruption
of the ADA2b gene have pleiotropic phenotypes while plants with homozygous
disruption of the ADA2a gene dud not have any visible defects (Vlachonasios, et al.,
2003). Moreover, Arabidopsis plants with homozygous disruption of GCN5 gene show
pleiotropic phenotypes that are similar but do not overlap completely with the phenotypes
in plants with homozygous disruption of the ADA2b gene (V lachonasios, et al., 2003).
All these facts lead to the hypothesis that the GCN5-ADA2a complex is very likely
different flom the GCN5-ADA2b complex. This hypothesis is also supported by the

recent description that the GCN5-ADA2a complex is different flom the GCN5-ADA2b

209

 

 

in fly (Kusch, et al., 2003). Interestingly, Arabidopsis has two homologues for each of
budding yeast Tral , TAF 6, and TAF12. If these homologous proteins are components of
GCN5-ADA2 complexes, it is not known whether one group of these homologues is
exclusively in one GCN5-ADA2 complex. Therefore, during the endeavor of identifying
the components of Arabidopsis GCN5-ADA2 complexes, it is equally important to
distinguish the components of GCN5-ADA2a complex flom that of the GCN5-ADA2b

complex.

The tapgets of GCN5-ADA2 complexes
Although the Arabidopsis GCN5-ADA2 complexes are featured as chromatin

covalent modifiers, the specific targets for the modification by the complexes are largely
unknown. Arabidopsis plants with homozygous disruption of ADA2b or GCN5 show
multiple grth defects. Microarray data indicated that the transcription of many genes
involved in a variety of cellular processes is up- or down-regulated in these disrupted
mutants (V lachonasios, et al., 2003). This suggests that the GCN5-ADA2 complexes are
targeted to many biological activities, but it is unlikely that all these effects found in
microarray data are primary.

The thesis study shows that two Arabidopsis proteins TINY and CBF 1 physically
interact with ADA2 proteins in vitro. Both TINY and CBF] proteins are AP2 domain
containing transcriptional factors Jaglo-Ottosen, 1998 #96] (Wilson, et al., 1996), The
AP2 domain is a DNA binding domain only found in plants. There are about 140 AP2
domain-containing proteins in Arabidopsis (Riechmann, et al., 2000). Only five AP2

domain-containing proteins (ERF4, ERFS, DREb2a, TINY and CBFI) were tested for

210

their interaction with ADA2 proteins, and two show interaction. It is hypothesized that
more AP2 proteins may interact with the GCN5-ADA2 complexes. The AP2
transcriptional factors are involved in a variety of biological functions in plants such as
response to abiotic and biotic stresses and hormonal regulation (Riechmann and
Meyerowitz, 1998). The proteins CBF 1 a transcriptional activator involved in
Arabidopsis response to cold stress (Stockinger, et al., 1997), and over expression of
protein TINY results in tiny plants (Wilson, et al., 1996). Therefore it hypothesized that
the GCN5-ADA2 complexes participate in many cellular activities associated with AP2
proteins, and the study of searching for more interacting AP2 proteins with GCN5-ADA2
complex promise to be fluitful.

In addition, the two candidates isolated by yeast two-hybrid screening with GCN5 as
bait are interesting for further investigation (see chapter V). One is candidate #77, a basic
helix-loop-helix transcriptional activator. It is not known what genes this activator
activates. It is also curious whether other basic helix-loop-helix transcriptional activators
interact with the GCN5-ADA2 complexes. Another candidate #141 implicated as ATP-
driven RNA or DNA helicase participating in variety of cellular processes such as DNA
recombination and repair. Increasing evidences show that nuclear HAT complexes might
participate in DNA-damage repair. The human SAGA-like complex STAGA associates
with components of UV-damaged DNA binding complex (UV -DDB) (Martinez, et al.,
2001). And two transcriptional HAT complexes, the human TIP60 and yeast INOSO, both
contain proteins related to DNA helicase and both complexes have DNA helicase activity
(Ikura, et al., 2000, Legube, et al., 2002, Martinez, et al., 2001, Shen, et al., 2000).

Human cells with impaired HAT activity in TIP60 have defective double-strand break

211

repair (Ikura, et al., 2000), and yeast with null INO80 shows hypersensitivity to DNA
damage (Shen, et al., 2000). Moreover, loss of GCN5 protein in mouse embryos results in
increased apoptosis that is intimately reconnected to DNA damage (Xu, et al., 2000). The
huge Tral protein in yeast SAGA belongs to the ATM family implicated in cell cycle
checkpoint signaling and cellular response to DNA damage (Abraham, 2001, Khanna, et
al., 2001, Vassilev, et al., 1998). Therefore, the Arabidopsis GCN5-ADA2 complexes are
very likely involved in DNA damage repair, and this might be a potential area for future
study as addition to the biological functions of the complexes. Approachable experiments
might be to investigate the capacity of DNA-damage repair in the mutant ADA2b or

GCN5 plants as described above.

The potential function of the unigue N-terminus of the GCN5 protein

The N-terminus of Arabidopsis GCN5 (aa 1-200) is completely different to the N-
terminus of other animal GCN5/PCAF proteins. The fimction of this N-terminus
remained unknown. It is not required for GCN5 HAT activity in vitro and no interacting
proteins have been identified by the strategy of yeast two-hybrid screening. Similar N-
terminal regions also exists in GCN5 proteins of rice and maize, suggesting the
importance of this N-terminus to plant GCN5 proteins. And curiously, the N-terminus in
Arabidopsis is encoded on exon (chapter 1), suggesting the N-terminus might function as
unit.

The identification of other components of GCN5-ADA2 complexes might provide
clues as to the possible role of the N-terminus of GCN5, and perhaps it has to wait to

until the identification of the components of GCN5-ADA2 complexes before there is a

212

glimpse of the filnction of the N-terminus. However, since plants with disrupted GCN5
gene show pleiotropic phenotypes that can be rescued by transgenic expression of GCN5
protein (V lachonasios, et al., 2003), it is practical to investigate whether GCN5 protein

with the N-terminus deleted will behave like wild type GCN5 in the mutant GCN5 plants.

The ace latiou of ADA2 roteins b GCN5 F
The serendipitous finding that both ADA2a and ADA2b can be acetylated by GCN5

 

in vitro is novel, and no similar cases have been reported in yeast or metazoan ADA2
proteins. Since the sequence along the acetylation site is only conserved in plant ADA2 E
proteins, it is hypothesized that the acetylation event is unique to plant ADA2 proteins.
As discussed in chapter VII, one major question is whether ADA2 proteins are acetylated
in viva, and this has to be answered first to investigate the significance of ADA2
acetylation. Nevertheless, the conservation of the sequence flanking the acetylation site in
plant ADA2 proteins suggests that there be functions associated with the region. One
approach to investigate the significance of this region might be to test the filnction of the
ADA2b protein that has multiple mutations at the conserved region, such as simultaneous
mutations at serine 210 and lysine 215 for both residues are conserved in similar histone
H3 N-tail.
Although mutant ADA2b that is not acetylated by GCN5 behaves like wild type
ADA2b in enhancing GCN5 HAT activity on nucleosomal histones, it is still not known
whether this residue is important in native GCN5-ADA2 complexes that are

hypothetically have more components. On the other hand, the significance of ADA2

 

proteins acetylation in vivo is still an open question. First, only overexpression (in the

213

level of mRN A) of ADA2b mutants was tested, and overexpression might conceal the

defects associated with mutant ADA2b if the protein level is also overexprssed. It is

obligate to test whether similar level expression of mutant to that of endogenous wild

type gene can perform complementation in ADA2b-mutated plants. But, as the positive

control, it is fundamental to find out first whether similar level of transgenic expression

of wild type ADA2b cDNA is able to complement. Secondly, only the normal laboratory

growth conditions were tested and only gross phenotypes were examined for the

complementary function of the mutant ADA2b. It is not known whether the mutant
ADA2b protein can complement in other grth conditions and whether there are subtle
differences in complemented plants in microscopic detail. Therefore, it is meaningful to
investigate the complement ability of mutant ADA2b under other growth conditions such

UV-light exposure and pathogen attack.

 

214

Bibliography

1. Aasland, R., A. F. Stewart, and T. Gibson 1996. The SANT domain: a putative
DNA-binding domain in the SWI-SNF and ADA complexes, the transcriptional co-
repressor N-CoR and TFIIIB Trends Biochem Sci. 21:87-8.

2. Abraham, R. T. 2001. Cell cycle checkpoint signaling through the ATM and
ATR kinases Genes Dev. 15:2177-96.

3. Agalioti, T., G. Chen, and D. Thanos 2002. Deciphering the transcriptional
histone acetylation code for a human gene Cell. 111:381-92.

4. Agalioti, T., S. Lomvardas, B. Parekh, J. Yie, T. Maniatis, and D. Thanos
2000. Ordered recruitment of chromatin modifying and general transcription factors to
the IFN-beta promoter Cell. 103:667-78.

5. Akhtar, A., and P. B. Becker 2001. The histone H4 acetyltransferase MOF uses
a C2HC zinc finger for substrate recognition EMBO Rep. 2:113-8.

6. Albright, S. R., and R. Tjian 2000. TAFs revisited: more data reveal new twists
and confirm old ideas Gene. 242:1-13.

7. Allen, M. D., K. Yamasaki, M. Ohme-Takagi, M. Tateno, and M. Suzuki
1998. A novel mode of DNA recognition by a beta-sheet revealed by the solution
structure of the GCC-box binding domain in complex with DNA Embo J. 17:5484-96.

8. Amedeo, P., Y. Habu, K. Afsar, O. M. Scheid, and J. Paszkowski 2000.
Disruption of the plant gene MOM releases transcriptional silencing of methylated genes
Nature. 405:203-6.

9. Balasubramanian, R., M. G. Pray-Grant, W. Selleck, P. A. Grant, and S. Tan
2002. Role of the Ada2 and Ada3 transcriptional coactivators in histone acetylation J Biol
Chem. 277:7989—95.

10. Bannister, A. J., R. Schneider, and T. Kouzarides 2002. Histone methylation:
dynamic or static? Cell. 109:801-6.

1]. Bannister, A. J., P. Zegerman, J. F. Partridge, E. A. Miska, J. 0. Thomas, R.
C. Allshire, and T. Kouzarides 2001. Selective recognition of methylated lysine 9 on
histone H3 by the HP] chromo domain Nature. 410: 120-4.

12. Banno, H., Y. Ikeda, Q. W. Niu, and N. H. Chua 2001. Overexpression of
Arabidopsis ESR] induces initiation of shoot regeneration Plant Cell. 13:2609-18.

13. Barbaric, S., H. Reinke, and W. Horz 2003. Multiple mechanistically distinct
functions of SAGA at the PHOS promoter Mol Cell Biol. 23:3468-76.

215

14. Barberis, A., and L. Gaudreau 1998. Recruitment of the RNA polymerase 11
holoenzyme and its implications in gene regulation Biol Chem. 379: 1397-405.

15. Barlev, N. A., R. Candau, L. Wang, P. Darpino, N. Silverman, and S. L.
Berger 1995. Characterization of physical interactions of the putative transcriptional
adaptor, ADA2, with acidic activation domains and TATA-binding protein J Biol Chem.
270:19337-44.

l6. Barlev, N. A., L. Liu, N. H. Chehab, K. Mansfield, K. G. Harris, T. D.
Halazonetis, and S. L. Berger 2001. Acetylation of p53 activates transcription through
recruitment of coactivators/histone acetyltransferases Mol Cell. 8:1243-54.

l7. Bauer, U. M., S. Daujat, S. J. Nielsen, K. Nightingale, and T. Kouzarides
2002. Methylation at arginine 17 of histone H3 is linked to gene activation EMBO Rep.
3:39-44.

18. Beisel, C., A. Imhof, J. Greene, E. Kremmer, and F. Sauer 2002. Histone
methylation by the Drosophila epigenetic transcriptional regulator Ash] Nature. 419:857-
62.

19. Berger, S. L., B. Pina, N. Silverman, G. A. Marcus, J. Agapite, J. L. Regier, S.
J. Triezenberg, and L. Guarente 1992. Genetic isolation of ADA2: a potential
transcriptional adaptor required for function of certain acidic activation domains Cell.
70:25 1-65.

20. Bhat, R. A., M. Riehl, G. Santandrea, R. Velasco, S. Slocombe, G. Donn, H.
H. Steinbiss, R. D. Thompson, and H. A. Becker 2003. Alteration of GCN5 levels in
maize reveals dynamic responses to manipulating histone acetylation Plant J. 33:455-69.

21. Blanco, J. C., S. Minucci, J. Lu, X. J. Yang, K. K. Walker, H. Chen, R. M.
Evans, Y. Nakatani, and K. Ozato 1998. The histone acetylase PCAF is a nuclear
receptor coactivator Genes Dev. 12: 1638-51.

22. Blau, J., H. Xiao, S. McCracken, P. O'Hare, J. Greenblatt, and D. Bentley
1996. Three functional classes of transcriptional activation domain Mol Cell Biol.
16:2044-55.

23. Bordoli, L., M. Netsch, U. Luthi, W. Lutz, and R. Eckner 2001. Plant
orthologs of p300/CBP: conservation of a core domain in metazoan p300/CBP
acetyltransferase-related proteins Nucleic Acids Res. 29:589-97.

24. Boyer, L. A., M. R. Langer, K. A. Crowley, S. Tan, J. M. Denu, and C. L.

Peterson 2002. Essential role for the SAN T domain in the functioning of multiple
chromatin remodeling enzymes Mol Cell. 10:935-42.

216

25. Brand, M., J. G. Moggs, M. Oulad-Abdelghani, F. Lejeune, F. J. Dilworth, J.
Stevenin, G. Almouzni, and L. Tora 2001. UV-damaged DNA-binding protein in the

TFT C complex links DNA damage recognition to nucleosome acetylation Embo J.
20:3187-96.

26. Briggs, S. D., T. Xiao, Z. W. Sun, J. A. Caldwell, J. Shabanowitz, D. F. Hunt,
C. D. Allis, and B. D. Strah12002. Gene silencing: trans-histone regulatory pathway in
chromatin Nature. 418:498.

27. Brown, C. E., L. Howe, K. Sousa, S. C. Alley, M. J. Carrozza, S. Tan, and J.
L. Workman 2001. Recruitment of HAT complexes by direct activator interactions with
the ATM-related Tral subunit Science. 292:2333-7.

28. Brown, S. A., C. S. Weirich, E. M. Newton, and R. E. Kingston 1998.
Transcriptional activation domains stimulate initiation and elongation at different times
and via different residues Embo J. 17:3146-54.

29. Brownell, J. E., J. Zhou, T. Ranalli, R. Kobayashi, D. G. Edmondson, S. Y.
Roth, and C. D. Allis 1996. Tetrahymena histone acetyltransferase A: a homolog to
yeast Gcn5p linking histone acetylation to gene activation Cell. 84:843-51.

30. Bryant, G. O., and M. Ptashne 2003. Independent recruitment in vivo by Gal4
of two complexes required for transcription Mol Cell. 11:1301-9.

31. Bryk, M., S. D. Briggs, B. D. Strahl, M. J. Curcio, C. D. Allis, and F. Winston
2002. Evidence that Set] , a factor required for methylation of histone H3, regulates
rDNA silencing in S. cerevisiae by a Sir2-independent mechanism Curr Biol. 12:165-70.

32. Brzeski, J., W. Podstolski, K. Olezak, and A. Jerzmanowski 1999.
Identification and analysis of the Arabidopsis thaliana BSH gene, a member of the SNFS
gene family Nucleic Acids Res. 27:2393-9.

33. Bulger, M., T. Sawado, D. Schubeler, and M. Groudine 2002. ChIPs of the
beta-globin locus: unraveling gene regulation within an active domain Curr Opin Genet
Dev. 12:170-7.

34. Burley, S. K., and R. G. Roeder 1996. Biochemistry and structural biology of
transcription factor IID (TFIID) Annu Rev Biochem. 65:769-99.

35. Calvo, D., M. Victor, F. Gay, G. Sui, M. P. Luke, P. Dufourcq, G. Wen, M.
Maduro, J. Rothman, and Y. Shi 2001. A POP-l repressor complex restricts
inappropriate cell type-specific gene transcription during Caenorhabditis elegans
embryogenesis Embo J. 20:7197-208.

36. Candau, R., and S. L. Berger 1996. Structural and functional analysis of yeast
putative adaptors. Evidence for an adaptor complex in vivo J Biol Chem. 271:5237-45.

217

37. Candau, R., P. A. Moore, L. Wang, N. Barlev, C. Y. Ying, C. A. Rosen, and S.
L. Berger 1996. Identification of human proteins functionally conserved with the yeast
putative adaptors ADA2 and GCN5 Mol Cell Biol. 16:593-602.

38. Candau, R., J. X. Zhou, C. D. Allis, and S. L. Berger 1997. Histone
acetyltransferase activity and interaction with ADA2 are critical for GCN5 function in
vivo Embo J. 16:555-65.

39. Carey, M. 1998. The enhanceosome and transcriptional synergy Cell. 92:5-8.

40. Carlson, M. 1997. Genetics of transcriptional regulation in yeast: connections to
the RNA polymerase II CTD Annu Rev Cell Dev Biol. 13:1-23.

4]. Chan, H. M., and N. B. La Thangue 2001. p300/CBP proteins: HATs for
transcriptional bridges and scaffolds J Cell Sci. 114:2363-73.

42. Chang, C., and J. A. Shockey 1999. The ethylene-response pathway: signal
perception to gene regulation Curr Opin Plant Biol. 2:352-8.

43. Chatterjee, S., and K. Struhl 1995. Connecting a promoter-bound protein to
TBP bypasses the need for a transcriptional activation domain Nature. 374:820-2.

44. Chen, D., H. Ma, H. Hong, S. S. Koh, S. M. Huang, B. T. Schurter, D. W.
Aswad, and M. R. Stallcup 1999. Regulation of transcription by a protein
methyltransferase Science. 284:2174-7.

45. Cheung, P., K. G. Tanner, W. L. Cheung, P. Sassone-Corsi, J. M. Denu, and
C. D. Allis 2000. Synergistic coupling of histone H3 phosphorylation and acetylation in
response to epidermal growth factor stimulation Mol Cell. 5:905-15.

46. Cheung, W. L., K. Ajiro, K. Samejima, M. Kloc, P. Cheung, C. A. Mizzen, A.
Beeser, L. D. Etkin, J. Chernoff, W. C. Earnshaw, and C. D. Allis 2003. Apoptotic
phosphorylation of histone H2B is mediated by mammalian sterile twenty kinase Cell.

1 13:507-1 7.

47. Chevillard-Briet, M., D. Trouche, and L. Vandel 2002. Control of CBP co-
activating activity by arginine methylation Embo J. 21:5457-66.

48. Chinenov, Y. 2002. A second catalytic domain in the Elp3 histone
acetyltransferases: a candidate for histone demethylase activity? Trends Biochem Sci.
27 :1 15-7.

49. Chua, Y. L., A. P. Brown, and J. C. Gray 2001. Targeted histone acetylation
and altered nuclease accessibility over short regions of the pea plastocyanin gene Plant
Cell. 13:599-612.

218

50. Clements, A., J. R. Rojas, R. C. Trievel, L. Wang, S. L. Berger, and R.
Marmorstein 1999. Crystal structure of the histone acetyltransferase domain of the
human PCAF transcriptional regulator bound to coenzyme A Embo J. 18:3521-32.

51. Cosma, M. P., S. Panizza, and K. Nasmyth 2001. Cdkl triggers association of

RNA polymerase to cell cycle promoters only after recruitment of the mediator by SBF
Mol Cell. 7:1213-20.

52. Cosma, M. P., T. Tanaka, and K. Nasmyth 1999. Ordered recruitment of
transcription and chromatin remodeling factors to a cell cycle- and developmentally
regulated promoter Cell. 97:299-311.

53. Daujat, 8., U. M. Bauer, V. Shah, B. Turner, S. Berger, and T. Kouzarides
2002. Crosstalk between CARM] Methylation and CBP Acetylation on Histone H3 Curr
Biol. 12:2090-7.

54. Dechend, R., F. Hirano, K. Lehmann, V. Heissmeyer, S. Ansieau, F. G.
Wulczyn, C. Seheidereit, and A. Leutz 1999. The Bcl-3 oncoprotein acts as a bridging
factor between NF-kappaB/Rel and nuclear co-regulators Oncogene. 18:3316-23.

55. Deng, L., C. de la Fuente, P. Fu, L. Wang, R. Donnelly, J. D. Wade, P.
Lambert, H. Li, C. G. Lee, and F. Kashanchi 2000. Acetylation of HIV-1 Tat by
CBP/P300 increases transcription of integrated HIV-l genome and enhances binding to
core histones Virology. 277 :278-95.

56. Deng, 2., C. J. Chen, M. Chamberlin, F. Lu, G. A. Blobel, D. Speicher, L. A.
Cirillo, K. S. Zaret, and P. M. Lieberman 2003. The CBP bromodomain and
nucleosome targeting are required for Zta-directed nucleosome acetylation and
transcription activation Mol Cell Biol. 23:2633-44.

57. Dhalluin, C., J. E. Carlson, L. Zeng, C. He, A. K. Aggarwal, and M. M. Zhou
1999. Structure and ligand of a histone acetyltransferase bromodomain Nature. 399:491-6.

58. Dorr, A., V. Kiermer, A. Pedal, H. R. Rackwitz, P. Henklein, U. Schubert, M.
M. Zhou, E. Verdin, and M. Ott 2002. Transcriptional synergy between Tat and PCAF
is dependent on the binding of acetylated Tat to the PCAF bromodomain Embo J.
21:2715-23.

59. Dover, J., J. Schneider, M. A. Tawiah-Boateng, A. Wood, K. Dean, M.
Johnston, and A. Shilatifard 2002. Methylation of histone H3 by COMPASS requires
ubiquitination of histone H28 by Rad6 J Biol Chem. 277 :28368-71.

60. Dunphy, E. L., T. Johnson, S. S. Auerbach, and E. H. Wang 2000.

Requirement for TAF(II)250 acetyltransferase activity in cell cycle progression Mol Cell
Biol. 20:1134-9.

219

61. Eberharter, A., D. E. Sterner, D. Schieltz, A. Hassan, J. R. Yates, 3rd, S. L.
Berger, and J. L. Workman 1999. The ADA complex is a distinct histone
acetyltransferase complex in Saccharomyces cerevisiae Mol Cell Biol. 19:6621-31.

62. Eissenberg, J. C., and S. C. Elgin 2000. The HP] protein family: getting a grip
on chromatin Curr Opin Genet Dev. 10:204-10.

63. Farrell, S., N. Simkovich, Y. Wu, A. Barberis, and M. Ptashne 1996. Gene
activation by recruitment of the RNA polymerase 11 holoenzyme Genes Dev. 10:2359-67.

64. Fassler, J. S., and F. Winston 1988. Isolation and analysis of a novel class of
suppressor of Ty insertion mutations in Saccharomyces cerevisiae Genetics. 118:203-12.

65. Fischle, W., Y. Wang, and C. D. Allis 2003. Histone and chromatin cross-talk
Curr Opin Cell Biol. 15:172-83.

66. Freiman, R. N., and R. Tjian 2003. Regulating the regulators: lysine
modifications make their mark Cell. 112:11-7.

67. Garber, M. E., and K. A. Jones 1999. HIV -1 Tat: coping with negative
elongation factors Curr Opin Irnmunol. 11:460-5.

68. Gaudreau, L., A. Schmid, D. Blaschke, M. Ptashne, and W. Horz 1997. RNA
polymerase 11 holoenzyme recruitment is sufficient to remodel chromatin at the yeast
PHOS promoter Cell. 89:55-62.

69. Gay, F., D. Calvo, M. C. Lo, J. Ceron, M. Maduro, R. Lin, and Y. Shi 2003.
Acetylation regulates subcellular localization of the Wnt signaling nuclear effector POP-1
Genes Dev. 17:717-22.

70. Georgakopoulos, T., N. Gounalaki, and G. Thireos 1995. Genetic evidence for

the interaction of the yeast transcriptional co-activator proteins GCN5 and ADA2 Mol
Gen Genet. 246:723-8.

71. Gonzalez-Canto, E., N. Klages, and M. Strubin 1997. Synergistic and
promoter-selective activation of transcription by recruitment of transcription factors

TFIID and TFIIB Proc Natl Acad Sci U S A. 94:8036—41.

72. Goodman, R. H., and S. Smolik 2000. CBP/p300 in cell growth, transformation,
and development Genes Dev. 14: 1553-77.

73. Grant, P. A., and S. L. Berger 1999. Histone acetyltransferase complexes Semin
Cell Dev Biol. 10:169-77.

74. Grant, P. A., L. Duggan, J. Cote, S. M. Roberts, J. E. Brownell, R. Candau, R.
Ohba, T. Owen-Hughes, C. D. Allis, F. Winston, S. L. Berger, and J. L. Workman

220

1997. Yeast Gcn5 functions in two multisubunit complexes to acetylate nucleosomal
histones: characterization of an Ada complex and the SAGA (Spt/Ada) complex Genes
Dev. 11:1640-50.

75. Grunstein, M. 1997. Histone acetylation in chromatin structure and transcription
Nature. 389:349-52.

76. Cu, W., and R. G. Roeder 1997. Activation of p53 sequence-specific DNA
binding by acetylation of the p53 C-terminal domain Cell. 90:595-606.

77. Gu, Z., L. M. Steinmetz, X. Gu, C. Scharfe, R. W. Davis, and W. H. Li 2003.
Role of duplicate genes in genetic robustness against null mutations Nature. 421:63 -6.

78. Hall, D. B., and K. Struhl 2002. The VP16 activation domain interacts with
multiple transcriptional components as determined by protein-protein cross-linking in
vivo J Biol Chem. 277:46043-50.

79. Hampsey, M. 1998. Molecular genetics of the RNA polymerase 11 general
transcriptional machinery Microbiol Mol Biol Rev. 62:465-503.

80. Han, M., and M. Grunstein 1988. Nucleosome loss activates yeast downstream
promoters in vivo Cell. 55:1137-45.

81. Hassan, A. H., K. E. Neely, M. Vignali, J. C. Reese, and J. L. Workman 2001.
Promoter targeting of chromatin-modifying complexes Front Biosci. 6:D1054-64.

82. Hassan, A. H., K. E. Neely, and J. L. Workman 2001. Histone acetyltransferase
complexes stabilize swi/snf binding to promoter nucleosomes Cell. 104:8] 7-27.

83. Hebbes, T. R., A. W. Thorne, and C. Crane-Robinson 1988. A direct link
between core histone acetylation and transcriptionally active chromatin Embo J. 7: 1395-
402.

84. Holstege, F. C., E. G. Jennings, J. J. Wyrick, T. 1. Lee, C. J. Hengartner, M.
R. Green, T. R. Golub, E. S. Lander, and R. A. Young 1998. Dissecting the regulatory
circuitry of a eukaryotic genome Cell. 95:717-28.

85. Horn, P. J., and C. L. Peterson 2001. The bromodomain: a regulator of ATP-
dependent chromatin remodeling? Front Biosci. 6:D1019-23.

86. Ikura, T., V. V. Ogryzko, M. Grigoriev, R. Groisman, J. Wang, M. Horikoshi,
R. Scully, J. Qin, and Y. N akatani 2000. Involvement of the TIP60 histone acetylase
complex in DNA repair and apoptosis Cell. 102:463-73.

87. Ito, M., C. X. Yuan, S. Malik, W. Gu, J. D. Fondell, S. Yamamura, Z. Y. F n,
X. Zhang, J. Qin, and R. G. Roeder 1999. Identity between TRAP and SMCC

221

complexes indicates novel pathways for the function of nuclear receptors and diverse
mammalian activators Mol Cell. 3:361-70.

88. Jacobs, S. A., and S. Khorasanizadeh 2002. Structure of HP] chromodomain
bound to a lysine 9-methylated histone H3 tail Science. 295:2080-3.

89. Jacobson, R. H., A. G. Ladurner, D. S. King, and R. Tjian 2000. Structure and
function of a human TAF 11250 double bromodomain module Science. 288: 1422-5.

90. Jaglo-Ottosen, K. R., S. J. Gilmour, D. G. Zarka, O. Schabenberger, and M.
F. Thomashow 1998. Arabidopsis CBF] overexpression induces COR genes and
enhances fleezing tolerance Science. 280:104-6.

91. Jeang, K. T., H. Xiao, and E. A. Rich 1999. Multifaceted activities of the HIV-1
transactivator of transcription, Tat J Biol Chem. 274:28837-40.

92. Jeanmougin, F., J. M. Wurtz, B. Le Douarin, P. Chambon, and R. Losson
1997. The bromodomain revisited Trends Biochem Sci. 22: 1 51-3.

93. Jenuwein, T. 2001. Re-SET-ting heterochromatin by histone methyltransferases
Trends Cell Biol. 11:266-73.

94. Jenuwein, T., and C. D. Allis 2001. Translating the histone code Science.
293: 1074-80.

95. Kadam, S., G. S. McAlpine, M. L. Phelan, R. E. Kingston, K. A. Jones, and B.
M. Emerson 2000. Functional selectivity of recombinant mammalian SWT/SNF subunits
Genes Dev. 14:2441-51.

96. Karn, J. 1999. Tackling Tat J Mol Biol. 293:235-54.

97. Keaveney, M., and K. Struhl 1998. Activator-mediated recruitment of the RNA
polymerase II machinery is the predominant mechanism for transcriptional activation in
yeast Mo] Cell. 1:917-24.

98. Kenzior, A. L., and W. R. Folk 1998. AtMSI4 and RbAp48 WD-40 repeat
proteins bind metal ions F EBS Lett. 440:425-9.

99. Khanna, K. K., M. F. Lavin, S. P. Jackson, and T. D. Mulhern 2001. ATM, a
central controller of cellular responses to DNA damage Cell Death Differ. 8: 1052-65.

100. Kim, J., K. Harter, and A. Theologis 1997. Protein-protein interactions among
the Aux/1AA proteins Proc Natl Acad Sci U S A. 94:11786-91.

222

10]. Kim, J. H., W. S. Lane, and D. Reinberg 2002. Human Elongator facilitates
RNA polymerase II transcription through chromatin Proc Natl Acad Sci U S A. 99:1241-
6.

102. Kim, Y. J., S. Bjorklund, Y. Li, M. H. Sayre, and R. D. Kornberg 1994. A
multiprotein mediator of transcriptional activation and its interaction with the C-terminal
repeat domain of RNA polymerase 11 Cell. 77:599-608.

103. Klages, N., and M. Strubin 1995. Stimulation of RNA polymerase II
transcription initiation by recruitment of TBP in vivo Nature. 374:822-3.

104. Kokubo, T., M. J. Swanson, J. I. Nishikawa, A. G. Hinnebusch, and Y.
Nakatani 1998. The yeast TAF145 inhibitory domain and TFIIA competitively bind to
TATA-binding protein Mol Cell Biol. 18:1003-12.

105. Koleske, A. J., and R. A. Young 1994. An RNA polymerase 11 holoenzyme
responsive to activators Nature. 368:466-9.

106. Kouzarides, T. 2000. Acetylation: a regulatory modification to rival
phosphorylation? Embo J. 19:] 176-9.

107. Kouzarides, T. 2002. Histone methylation in transcriptional control Curr Opin
Genet Dev. 12:198-209.

108. Kristjuhan, A., J. Walker, N. Suka, M. Grunstein, D. Roberts, B. R. Cairns,
and J. Q. Svejstrup 2002. Transcriptional inhibition of genes with severe histone h3
hypoacetylation in the coding region Mol Cell. 10:925-33.

109. Krumm, A., L. B. Hickey, and M. Groudine 1995. Promoter-proximal pausing
of RNA polymerase 11 defines a general rate-limiting step afier transcription initiation
Genes Dev. 9:559-72.

110. Kulesza, C. A., H. A. Van Buskirk, M. D. Cole, J. C. Reese, M. M. Smith, and
D. A. Euge12002. Adenovirus ElA requires the yeast SAGA histone acetyltransferase
complex and associates with SAGA components Gcn5 and Tral Oncogene. 21:14] 1-22.

111. Kumar, A., S. Agarwal, J. A. Heyman, S. Matson, M. Heidtman, S. Piccirillo,
L. Umansky, A. Drawid, R. Jansen, Y. Liu, K. H. Cheung, P. Miller, M. Gerstein, G.

S. Roeder, and M. Snyder 2002. Subcellular localization of the yeast proteome Genes
Dev. 16:707-19.

112. Kuo, M. H., J. E. Brownell, R. E. Sobel, T. A. Ranalli, R. G. Cook, D. G.
Edmondson, S. Y. Roth, and C. D. Allis 1996. Transcription-linked acetylation by
Gcn5p of histones H3 and H4 at specific lysines Nature. 383:269-72.

113. Kuo, M. H., E. vom Baur, K. Struhl, and C. D. Allis 2000. Gcn4 activator

223

targets Gcn5 histone acetyltransferase to specific promoters independently of
transcription Mol Cell. 6: 1309-20.

114. Kuo, M. H., J. Zhou, P. Jambeck, M. E. Churchill, and C. D. Allis 1998.
Histone acetyltransferase activity of yeast Gcn5p is required for the activation of target
genes in vivo Genes Dev. 12:627-39.

115. Kusch, T., S. Guelman, S. M. Abmayr, and J. L. Workman 2003. Two
Drosophila Ada2 homologues function in different multiprotein complexes Mol Cell Biol.
23:3305-19.

116. Kuzmichev, A., K. Nishioka, H. Erdjument-Bromage, P. Tempst, and D.
Reinberg 2002. Histone methyltransferase activity associated with a human multiprotein
complex containing the Enhancer of Zeste protein Genes Dev. 16:2893-905.

117. Lachner, M., and T. Jenuwein 2002. The many faces of histone lysine
methylation Curr Opin Cell Biol. 14:286-98.

118. Lachner, M., D. O'Carroll, S. Rea, K. Mechtler, and T. Jenuwein 2001.
Methylation of histone H3 lysine 9 creates a binding site for HP] proteins Nature.
410:116-20.

119. Larschan, E., and F. Winston 2001. The S. cerevisiae SAGA complex functions
in vivo as a coactivator for transcriptional activation by Gal4 Genes Dev. 15: 1946-56.

120. Lee, K. M., and J. J. Hayes 1997. The N-terrninal tail of histone H2A binds to
two distinct sites within the nucleosome core Proc Natl Acad Sci U S A. 94:8959-64.

121. Lee, T. 1., and R. A. Young 2000. Transcription of eukaryotic protein-coding
genes Annu Rev Genet. 34:77-137.

122. Legube, G., L. K. Linares, C. Lemercier, M. Schefl'ner, S. Khochbin, and D.
Trouche 2002. Tip60 is targeted to proteasome-mediated degradation by Mdm2 and
accumulates after UV irradiation Embo J. 21: 1704-12.

123. Lenfant, F., R. K. Mann, B. Thomsen, X. Ling, and M. Grunstein 1996. All
four core histone N-termini contain sequences required for the repression of basal
transcription in yeast Embo J. 15:3974-85.

124. Li, G., T. C. Hall, and R. Holmes-Davis 2002. Plant chromatin: development
and gene control Bioessays. 24:234-43.

125. Li, J., Q. Lin, H. G. Yoon, Z. Q. Huang, B. D. Strahl, C. D. Allis, and J.

Wong 2002. Involvement of histone methylation and phosphorylation in regulation of
transcription by thyroid hormone receptor Mol Cell Biol. 22:5688-97.

224

 

126.
Correla
chicker

127.
2001. ".
neighh

128.

Kay, 1
comp]
TBP C

129.

and ls’
EREE
droug'
Plant

130.
Shielr
the hi

13].

Man
funct
Cell.

132.
Med?

133.
initii
Cell.

134.
Ram

Chro

135,
199:
339;

136.

126. Litt, M. D., M. Simpson, M. Gaszner, C. D. Allis, and G. Felsenfeld 2001.
Correlation between histone lysine methylation and developmental changes at the
chicken beta-globin locus Science. 293:2453-5.

127. Litt, M. D., M. Simpson, F. Recillas-Targa, M. N. Prioleau, and G. Felsenfeld
2001. Transitions in histone acetylation reveal boundaries of three separately regulated
neighboring loci Embo J. 20:2224-35.

128. Lin, D., R. Ishima, K. I. Tong, S. Bagby, T. Kokubo, D. R. Muhandiram, L. E.
Kay, Y. Nakatani, and M. Ikura 1998. Solution structure of a TBP-TAF(II)230
complex: protein mimicry of the minor groove surface of the TATA box unwound by
TBP Cell. 94:573-83.

129. Liu, 0., M. Kasuga, Y. Sakuma, H. Abe, S. Miura, K. Yamaguchi-Shinozaki,
and K. Shinozaki 1998. Two transcription factors, DREB] and DREB2, with an
EREBP/AP2 DNA binding domain separate two cellular signal transduction pathways in
drought- and low-temperature-responsive gene expression, respectively, in Arabidopsis
Plant Cell. 10:1391-406.

130. La, W. S., L. Duggan, N. C. Tolga, Emre, R. Belotserkovskya, W. S. Lane, R.
Shiekhattar, and S. L. Berger 2001. Snfl -a histone kinase that works in concert with
the histone acetyltransferase Gcn5 to regulate transcription Science. 293:1142-6.

13]. Lo, W. S., R. C. Trievel, J. R. Rojas, L. Duggan, J. Y. Hsu, C. D. Allis, R.
Marmorstein, and S. L. Berger 2000. Phosphorylation of serine 10 in histone H3 is
functionally linked in vitro and in vivo to Gcn5-mediated acetylation at lysine 14 Mol
Cell. 5:917-26.

132. Lorch, Y., J. Beve, C. M. Gustafsson, L. C. Myers, and R. D. Kornberg 2000.
Mediator-nucleosome interaction Mol Cell. 6:197-201.

133. Lorch, Y., J. W. LaPointe, and R. D. Kornberg 1987. Nucleosomes inhibit the
initiation of transcription but allow chain elongation with the displacement of histones
Cell. 49:203-10.

134. Loyola, A., G. LeRoy, Y. H. Wang, and D. Reinberg 2001. Reconstitution of
recombinant chromatin establishes a requirement for histone-tail modifications during

chromatin assembly and transcription Genes Dev. 15:2837-51.

135. Luger, K., A. W. Mader, R. K. Richmond, D. F. Sargent, and T. J. Richmond
1997. Crystal structure of the nucleosome core particle at 2.8 A resolution Nature.
389:25 1 -60.

136. Lusser, A. 2002. Acetylated, methylated, remodeled: chromatin states for gene
regulation Curr Opin Plant Biol. 5:437-43.

225

 

r555

 

 

137. Madison, D. L., P. Yaciuk, R. P. Kwok, and J. R. Lundblad 2002. Acetylation
of the adenovirus-transforming protein E] A determines nuclear localization by disrupting
association with irnportin-alpha J Biol Chem. 277 :38755-63.

138. Majello, B., and G. Napolitano 2001. Control of RNA polymerase II activity by
dedicated CTD kinases and phosphatases Front Biosci. 6:D1358-68.

139. Maldonado, E., R. Shiekhattar, M. Sheldon, H. Cho, R. Drapkin, P. Rickert,
E. Lees, C. W. Anderson, S. Linn, and D. Reinberg 1996. A human RNA polymerase
II complex associated with SRB and DNA-repair proteins Nature. 381:86-9.

140. Malik, S., and R. G. Roeder 2000. Transcriptional regulation through Mediator-
like coactivators in yeast and metazoan cells Trends Biochem Sci. 25:277-83.

14]. Marcus, G. A., N. Silverman, S. L. Berger, J. Horiuchi, and L. Guarente 1994.
Functional similarity and physical association between GCN5 and ADA2: putative
transcriptional adaptors Embo J. 13:4807-15.

142. Marmorstein, R. 2001. Structure of histone acetyltransferases J Mol Biol.
311:433-44.

143. Martinez, E., V. B. Palhan, A. Tjernberg, E. S. Lymar, A. M. Gamper, T. K.
Kundu, B. T. Chait, and R. G. Roeder 2001. Human STAGA complex is a chromatin-
acetylating transcription coactivator that interacts with pre-mRN A splicing and DNA
damage-binding factors in vivo Mol Cell Biol. 21:6782-95.

144. Memelink, J., R. Verpoorte, and J. W. Kijne 2001. ORCAnization of
jasmonate-responsive gene expression in alkaloid metabolism Trends Plant Sci. 6:212-9.

145. Mencia, M., Z. Moqtaderi, J. V. Geisberg, L. Kuras, and K. Struh12002.
Activator-specific recruitment of TFIID and regulation of ribosomal protein genes in
yeast Mol Cell. 9:823-33.

146. Milne, T. A., S. D. Briggs, H. W. Brock, M. E. Martin, D. Gibbs, C. D. Allis,
and J. L. Hess 2002. MLL targets SET domain methyltransferase activity to Hox gene
promoters Mol Cell. 10:1107-17.

147. Mujtaba, S., Y. He, L. Zeng, A. Farooq, J. E. Carlson, M. Ott, E. Verdin, and
M. M. Zhou 2002. Structural basis of lysine-acetylated HIV-1 Tat recognition by PCAF
bromodomain Mol Cell. 9:575-86.

148. Muratoglu, S., S. Georgieva, G. Papai, E. Scheer, I. Enunlu, O. Komonyi, I.
Cserpan, L. Lebedeva, E. Nabirochkina, A. Udvardy, L. Tara, and I. Boros 2003.
Two different Drosophila ADA2 homologues are present in distinct GCN5 histone
acetyltransferase-containing complexes Mol Cell Biol. 23:306-21.

226

 

149. Mutskov, V. J., C. M. Farrell, P. A. Wade, A. P. Wolffe, and G. Felsenfeld
2002. The barrier fimction of an insulator couples high histone acetylation levels with
specific protection of promoter DNA flom methylation Genes Dev. 16:1540-54.

150. Myers, L. C., C. M. Gustafsson, D. A. Bushnell, M. Lui, H. Erdjument-
Bromage, P. Tempst, and R. D. Kornberg 1998. The Med proteins of yeast and their
function through the RNA polymerase II carboxy-terminal domain Genes Dev. 12:45-54.

15]. Myers, L. C., and R. D. Kornberg 2000. Mediator of transcriptional regulation
Annu Rev Biochem. 69:729-49.

152. Naar, A. M., P. A. Beaurang, S. Zhou, S. Abraham, W. Solomon, and R.
Tjian 1999. Composite co-activator ARC mediates chromatin-directed transcriptional
activation Nature. 398:828-32.

153. Naar, A. M., B. D. Lemon, and R. Tjian 2001. Transcriptional coactivator
complexes Annu Rev Biochem. 70:475-501.

154. Natarajan, K., B. M. Jackson, H. Zhou, F. Winston, and A. G. Hinnebusch
1999. Transcriptional activation by Gcn4p involves independent interactions with the
SWI/SNF complex and the SRB/mediator Mol Cell. 4:657-64.

155. Neely, K. E., A. H. Hassan, A. E. Wallberg, D. J. Steger, B. R. Cairns, A. P.
Wright, and J. L. Workman 1999. Activation domain-mediated targeting of the
SWI/SNF complex to promoters stimulates transcription flom nucleosome arrays Mol
Cell. 4:649-55.

156. Neish, A. S., S. F. Anderson, B. P. Schlegel, W. Wei, and J. D. Parvin 1998.
Factors associated with the mammalian RNA polymerase 11 holoenzyme Nucleic Acids
Res. 26:847-53.

157. Neuwald, A. F., and D. Landsman 1997. GCN5-related histone N-
acetyltransferases belong to a diverse superfarnily that includes the yeast SPT10 protein
Trends Biochem Sci. 22: 154-5.

158. Ng, H. H., D. N. Ciccone, K. B. Morshead, M. A. Oettinger, and K. Struhl
2003. Lysine-79 of histone H3 is hypomethylated at silenced loci in yeast and

mammalian cells: a potential mechanism for position-effect variegation Proc Natl Acad
Sci U S A. 100:1820-5.

159. Ng, H. H., Q. Feng, H. Wang, H. Erdjument-Bromage, P. Tempst, Y. Zhang,
and K. Struhl 2002. Lysine methylation within the globular domain of histone H3 by

Dot] is important for telomeric silencing and Sir protein association Genes Dev.
16:1518-27.

227

 

 

 

 

160.
1128 by y
J Biol C

161. l
O'Carr
2001. R

162. l‘
Drosopl
TATA l
94:85-9

163.

regulatr

164.
by Setl
Acad S

165.
H. “or
aCEtyla

 

l 66 .
bromc
histon

] 67 .
facto‘

168.
Cell

169

160. N g, H. H., R. M. Xu, Y. Zhang, and K. Struhl 2002. Ubiquitination of histone
H23 by Rad6 is required for efficient Dotl-mediated methylation of histone H3 lysine 79
J Biol Chem. 277:34655-7.

161. Nielsen, S. J., R. Schneider, U. M. Bauer, A. J. Bannister, A. Morrison, D.
O'Carroll, R. Firestein, M. Cleary, T. Jenuwein, R. E. Herrera, and T. Kouzarides
2001. Rb targets histone H3 methylation and HP] to promoters Nature. 412:561-5.

162. Nishikawa, J., T. Kokubo, M. Horikoshi, R. G. Roeder, and Y. Nakatani 1997.
Drosophila TAF(II)230 and the transcriptional activator VP16 bind competitively to the
TATA box-binding domain of the TATA box-binding protein Proc Natl Acad Sci U S A.
94:85-90.

163. Nishioka, K., and D. Reinberg 2001. Transcription. Switching partners in a
regulatory tango Science. 294:2497-8.

164. N oma, K., and S. I. Grewal 2002. Histone H3 lysine 4 methylation is mediated
by Set] and promotes maintenance of active chromatin states in fission yeast Proc Natl
Acad Sci U S A. 99:16438-45.

165. Ogryzko, V. V., T. Kotani, X. Zhang, R. L. Schiltz, T. Howard, X. J. Yang, B.
H. Howard, J. Qin, and Y. Nakatani 1998. Histone-like TAFs within the PCAF histone
acetylase complex Cell. 94:35-44.

166. Omaghi, P., P. Ballario, A. M. Lena, A. Gonzalez, and P. Filetici 1999. The
bromodomain of Gcn5p interacts in vitro with specific residues in the N terminus of
histone H4 J Mol Biol. 287:1-7.

167. Orphanides, G., T. Lagrange, and D. Reinberg 1996. The general transcription
factors of RNA polymerase 1] Genes Dev. 10:2657-83.

168. Orphanides, G., and D. Reinberg 2002. A unified theory of gene expression
Cell. 108:439-51.

169. Park, J. M., C. J. Park, S. B. Lee, B. K. Ham, R. Shin, and K. H. Paek 2001.
Overexpression of the tobacco Tsi] gene encoding an EREBP/AP2-type transcription

factor enhances resistance against pathogen attack and osmotic stress in tobacco Plant
Cell. 13:1035-46.

170. Peterson, C. L., and J. L. Workman 2000. Promoter targeting and chromatin
remodeling by the SWI/SNF complex Cru'r Opin Genet Dev. 10: 1 87-92.

171. Pham, A. D., and F. Sauer 2000. Ubiquitin-activating/conjugating activity of

TAF 11250, a mediator of activation of gene expression in Drosophila Science. 289:2357-
60.

228

 

172. Pina, B., S. Berger, G. A. Marcus, N. Silverman, J. Agapite, and L. Guarente
1993. ADA3: a gene, identified by resistance to GAL4-VP16, with properties sirrrilar to
and different flom those of ADA2 Mol Cell Biol. 13:5981-9.

173. Protacio, R. U., G. Li, P. T. Lowary, and J. Widom 2000. Effects of histone tail
domains on the rate of transcriptional elongation through a nucleosome Mol Cell Biol.
20:8866-78.

174. Ptashne, M., and A. Gann 1997. Transcriptional activation by recruitment
Nature. 386:569-77.

175. Pugh, B. F., and R. Tjiau 1992. Diverse transcriptional functions of the
multisubunit eukaryotic TFIID complex J Biol Chem. 267:679-82.

176. Rachez, C., and L. P. Freedman 2001. Mediator complexes and transcription
Curr Opin Cell Biol. 13:274-80.

177. Rachez, C., B. D. Lemon, Z. Suldan, V. Bromleigh, M. Gamble, A. M. Naar,
H. Erdjument-Bromage, P. Tempst, and L. P. Freedman 1999. Ligand-dependent
transcription activation by nuclear receptors requires the DRIP complex Nature. 398:824-
8.

178. Rea, S., F. Eisenhaber, D. O'Carroll, B. D. Strahl, Z. W. Sun, M. Schmid, S.
Opravil, K. Mechtler, C. P. Ponting, C. D. Allis, and T. Jenuwein 2000. Regulation of
chromatin structure by site-specific histone H3 methyltransferases Nature. 406:593-9.

179. Reinke, H., P. D. Gregory, and W. Horz 2001. A transient histone
hyperacetylation signal marks nucleosomes for remodeling at the PHO8 promoter in vivo
Mol Cell. 7:529—38.

180. Rezai-Zadeh, N., X. Zhang, F. Namour, G. Fejer, Y. D. Wen, Y. L. Yao, I.
Gyory, K. Wright, and E. Seto 2003. Targeted recruitment of a histone H4-specific
methylflansferase by the transcription factor YYl Genes Dev. 17: 1019-29.

181. Riechmann, J. L., J. Heard, G. Martin, L. Reuber, C. Jiang, J. Keddie, L.
Adam, 0. Pineda, O. J. Ratcliffe, R. R. Samaha, R. Creelmau, M. Pilgrim, P. Broun,
J. Z. Zhang, D. Ghandehari, B. K. Sherman, and G. Yu 2000. Arabidopsis

transcription factors: genome-wide comparative analysis among eukaryotes Science.
290:2] 05 -1 0.

182. Riechmann, J. L., and E. M. Meyerowitz 1998. The AP2/EREBP family of
plant transcription factors Biol Chem. 379:633-46.

183. Riechmann, J. L., and O. J. Ratcliffe 2000. A genomic perspective on plant
transcription factors Curr Opin Plant Biol. 3:423-34.

229

 

 

 

184.
Rihhr

185.
single

186.
Rev B

187.

cofac.
Natur

188.

Yam
gene
189.

acti»

190.
Nak
COM

1 9]
A. I
Met
Bio.

192
275

193
Pro

B10

1 94
Tel]-
195
deh
Pat]

184. Roeder, R. G. 1996. The role of general initiation factors in transcription by
RNA polymerase 11 Trends Biochem Sci. 21:327-35.

185. Roth, S. Y., and C. D. Allis 1996. Histone acetylation and chromatin assembly: a
single escort, multiple dances? Cell. 87:5-8.

186. Roth, S. Y., J. M. Denu, and C. D. Allis 2001. Histone acetyltransferases Annu
Rev Biochem. 70:81-120.

187. Ryu, S., S. Zhou, A. G. Ladurner, and R. Tjian 1999. The transcriptional
cofactor complex CRSP is required for activity of the enhancer-binding protein Spl
Nature. 397 :446-50.

188. Sakuma, Y., Q. Liu, J. G. Dubouzet, H. Abe, K. Shinozaki, and K.
Yamaguchi-Shinozaki 2002. DNA-binding specificity of the ERF/AP2 domain of
Arabidopsis DREBs, transcription factors involved in dehydration- and cold-inducible
gene expression Biochem Biophys Res Commun. 290:998-1009.

189. Sassone-Corsi, P., C. A. Mizzen, P. Cheung, C. Crosio, L. Monaco, S. Jacquot,
A. Hanauer, and C. D. Allis 1999. Requirement of Rsk-2 for epidermal growth factor-
activated phosphorylation of histone H3 Science. 285:886-91.

190. Schiltz, R. L., C. A. Mizzen, A. Vassilev, R. G. Cook, C. D. Allis, and Y.
Nakatani 1999. Overlapping but distinct patterns of histone acetylation by the human
coactivators p300 and PCAF within nucleosomal substrates J Biol Chem. 274:1189-92.

191. Schurter, B. T., S. S. Koh, D. Chen, G. J. Bunick, J. M. Harp, B. L. Hanson,
A. Henschen-Edman, D. R. Mackay, M. R. Stallcup, and D. W. Aswad 2001.

Methylation of histone H3 by coactivator-associated arginine methyltransferase 1
Biochemistry. 40:5747-56.

192. Sendra, R., C. Tse, and J. C. Hansen 2000. The yeast histone acetyltransferase

A2 complex, but not flee Gcn5p, binds stably to nucleosomal arrays J Biol Chem.
275:24928-34.

193. Sewack, G. F., T. W. Ellis, and U. Hansen 2001. Binding of TATA binding
protein to a naturally positioned nucleosome is facilitated by histone acetylation Mol Cell
Biol. 21:1404-15.

194. Shen, X., G. Mizuguchi, A. Hamiche, and C. Wu 2000. A chromatin
remodelling complex involved in transcription and DNA processing Nature. 406:541-4.

195. Shinozaki, K., and K. Yamaguchi-Shinozaki 2000. Molecular responses to

dehydration and low temperature: differences and cross-talk between two stress signaling
pathways Curr Opin Plant Biol. 3:217-23.

230

196. Sieberer, T., M. T. Hauser, G. J. Seifert, and C. Luschnig 2003. PROPORZ] ,
a Putative Arabidopsis Transcriptional Adaptor Protein, Mediates Auxin and Cytokinin
Signals in the Control of Cell Proliferation Curr Biol. 13:837-42.

197. Singh, K., R. C. Foley, and L. Onate-Sanchez 2002. Transcription factors in
plant defense and stress responses Curr Opin Plant Biol. 5:430-6.

198. Spencer, V. A., and J. R. Davie 1999. Role of covalent modifications of histones
in regulating gene expression Gene. 240: 1-12.

199. Spilianakis, C., J. Papamatheakis, and A. Kretsovali 2000. Acetylation by
PCAF enhances CIITA nuclear accumulation and transactivation of major
histocompatibility complex class 11 genes Mol Cell Biol. 20:8489-98.

200. Sterner, D. E., R. Belotserkovskaya, and S. L. Berger 2002. SALSA, a variant
of yeast SAGA, contains truncated Spt7, which correlates with activated transcription
Proc Natl Acad Sci U S A. 99:11622-7.

201. Sterner, D. E., and S. L. Berger 2000. Acetylation of histones and transcription-
related factors Microbiol Mol Biol Rev. 64:435-59.

202. Sterner, D. E., P. A. Grant, S. M. Roberts, L. J. Duggan, R. Belotserkovskaya,
L. A. Pacella, F. Winston, J. L. Workman, and S. L. Berger 1999. Functional
organization of the yeast SAGA complex: distinct components involved in structural

integrity, nucleosome acetylation, and TATA-binding protein interaction Mol Cell Biol.
19:86-98.

203. Sterner, D. E., X. Wang, M. H. Bloom, G. M. Simon, and S. L. Berger 2002.
The SAN T domain of Ada2 is required for normal acetylation of histones by the yeast
SAGA complex J Biol Chem. 277:8178-86.

204. Stockinger, E. J., S. J. Gilmour, and M. F. Thomashow 1997. Arabidopsis
thaliana CBF] encodes an AP2 domain-containing transcriptional activator that binds to
the C-repeat/DRE, a cis-acting DNA regulatory element that stimulates transcription in
response to low temperature and water deficit Proc Natl Acad Sci U S A. 94: 1035-40.

205. Stockinger, E. J., Y. Mao, M. K. Regier, S. J. Triezenberg, and M. F.
Thomashow 2001. Transcriptional adaptor and histone acetyltransferase proteins in
Arabidopsis and their interactions with CBF] , a transcriptional activator involved in cold-
regulated gene expression Nucleic Acids Res. 29:]524-33.

206. Strahl, B. D., and C. D. Allis 2000. The language of covalent histone
modifications Nature. 403:4] -5.

231

207. Strahl, B. D., R. Ohba, R. G. Cook, and C. D. Allis 1999. Methylation of
histone H3 at lysine 4 is highly conserved and correlates with transcriptionally active
nuclei in Tetrahymena Proc Natl Acad Sci U S A. 96: 14967-72.

208. Struhl, K. 1999. Fundamentally different logic of gene regulation in eukaryotes
and prokaryotes Cell. 98: 1-4.

209. Struhl, K. 1998. Histone acetylation and transcriptional regulatory mechanisms
Genes Dev. 12:599-606.

210. Sun, Z. W., and C. D. Allis 2002. Ubiquitination of histone H28 regulates H3
methylation and gene silencing in yeast Nature. 418:104-8.

211. Swanson, M. J., H. Qiu, L. Sumibcay, A. Krueger, S. J. Kim, K. Natarajan, S.
Yoon, and A. G. Hinnebusch 2003. A multiplicity of coactivators is required by Gcn4p
at individual promoters in vivo Mol Cell Biol. 23:2800-20.

212. Taatjes, D. J., A. M. Naar, F. Andel, 3rd, E. Nogales, and R. Tjian 2002.
Structure, function, and activator-induced conformations of the CRSP coactivator
Science. 295: 1058-62.

213. Takechi, S., and T. Nakayama 1999. Sas3 is a histone acetyltransferase and
requires a zinc finger motif Biochem Biophys Res Commun. 266:405-10.

214. Tanner, K. G., M. R. Langer, and J. M. Denu 2000. Kinetic mechanism of
human histone acetyltransferase P/CAF Biochemistry. 39:11961-9.

215. Tanner, K. G., M. R. Langer, Y. Kim, and J. M. Denu 2000. Kinetic
mechanism of the histone acetyltransferase GCN5 flom yeast J Biol Chem. 275:22048-55.

216. Tanner, K. G., R. C. Trievel, M. H. Kuo, R. M. Howard, S. L. Berger, C. D.
Allis, R. Marmorstein, and J. M. Denu 1999. Catalytic mechanism and function of
invariant glutamic acid 173 flom the histone acetyltransferase GCN5 transcriptional
coactivator J Biol Chem. 274:18157-60.

217. Thomashow, M. F. 1998. Role of cold-responsive genes in plant freezing
tolerance Plant Physiol. 118:1-8.

218. Thompson, C. M., and R. A. Young 1995. General requirement for RNA
polymerase II holoenzymes in vivo Proc Natl Acad Sci U S A. 92:4587-90.

219. Thomson, S., A. L. Clayton, and L. C. Mahadevan 2001. Independent dynamic

regulation of histone phosphorylation and acetylation during irnmediate-early gene
induction Mol Cell. 8:123 1-41 .

232

 

220. Triezenberg, S. J. 1995. Structure and function of transcriptional activation
domains Curr Opin Genet Dev. 5: 190-6.

221. Tse, C., and J. C. Hansen 1997. Hybrid trypsinized nucleosomal arrays:
identification of multiple ftmctional roles of the H2A/H2B and H3/I-I4 N-termini in
chromatin fiber compaction Biochemistry. 36:11381-8.

222. Turner, B. M. 2002. Cellular memory and the histone code Cell. 111:285-91.

223. Turner, B. M. 199] . Histone acetylation and control of gene expression J Cell
Sci. 99:13-20.

224. Utley, R. T., K. Ikeda, P. A. Grant, J. Cote, D. J. Steger, A. Eberharter, S.
John, and J. L. Workman 1998. Transcriptional activators direct histone
acetyltransferase complexes to nucleosomes Nature. 394:498-502.

225. Vandel, L., E. Nicolas, O. Vaute, R. Ferreira, S. Ait-Si-Ali, and D. Trouche
2001. Transcriptional repression by the retinoblastoma protein through the recruitment of
a histone methyltransferase Mol Cell Biol. 21:6484-94.

226. Vassilev, A., J. Yamauchi, T. Kotani, C. Prives, M. L. Avantaggati, J. Qin,
and Y. Nakatani 1998. The 400 kDa subunit of the PCAF histone acetylase complex
belongs to the ATM superfarnily Mol Cell. 2:869-75.

227. Verbsky, M. L., and E. J. Richards 2001. Chromatin remodeling in plants Curr
Opin Plant Biol. 4:494-500.

228. Verdone, L., J. Wu, K. van Riper, N. Kacherovsky, M. Vogelauer, E. T.
Young, M. Grunstein, E. Di Mauro, and M. Caserta 2002. Hyperacetylation of

chromatin at the ADH2 promoter allows Adrl to bind in repressed conditions Embo J.
21 :1 101-1 1.

229. Verreault, A., P. D. Kaufman, R. Kobayashi, and B. Stillman 1996.
Nucleosome assembly by a complex of CAF-l and acetylated histones H3/H4 Cell.
87:95-104.

230. Vignali, M., A. H. Hassan, K. E. Neely, and J. L. Workman 2000. ATP-
dependent chromatin-remodeling complexes Mol Cell Biol. 20: 1899-910.

231. Vlachonasios, K. E., M. F. Thomashow, and S. J. Triezenberg 2003.
Disruption mutations of ADA2b and GCN5 transcriptional adaptor genes dramatically
affect Arabidopsis growth, deve10pment, and gene expression Plant Cell. 15:626-38.

232. Wang, H., R. Cao, L. Xia, H. Erdjument-Bromage, C. Borchers, P. Tempst,

and Y. Zhang 2001. Purification and filnctional characterization of a histone H3-lysine
4-specific methyltransferase Mol Cell. 8:1207-17.

233

233. Wang, H., Z. Q. Huang, L. Xia, Q. Feng, H. Erdjument-Bromage, B. D.
Strahl, S. D. Briggs, C. D. Allis, J. Wong, P. Tempst, and Y. Zhang 2001.
Methylation of histone H4 at arginine 3 facilitating transcriptional activation by nuclear
hormone receptor Science. 293:853-7.

234. Wang, L., L. Liu, and S. L. Berger 1998. Critical residues for histone
acetylation by Gcn5, functioning in Ada and SAGA complexes, are also required for

transcriptional function in vivo Genes Dev. 12:640-53.

235. Wassarman, D. A., and F. Sauer 2001. TAF(II)250: a transcription toolbox J
Cell Sci. 114:2895-902.

236. Wei, Y., C. A. Mizzen, R. G. Cook, M. A. Gorovsky, and C. D. Allis 1998.
Phosphorylation of histone H3 at serine 10 is correlated with chromosome condensation
during mitosis and meiosis in Tetrahymena Proc Natl Acad Sci U S A. 95:7480-4.

237. Wendler, W. M., E. Kremmer, R. Forster, and E. L. Winnacker 1997.
Identification of pirin, a novel highly conserved nuclear protein J Biol Chem. 272:8482-9.

238. Wilson, C. J., D. M. Chao, A. N. Imbalzano, G. R. Schnitzler, R. E. Kingston,
and R. A. Young 1996. RNA polymerase 11 holoenzyme contains SWI/SNF regulators
involved in chromatin remodeling Cell. 84:235-44.

239. Wilson, K., D. Long, J. Swinburne, and G. Coupland 1996. A Dissociation
insertion causes a semidominant mutation that increases expression of TINY, an
Arabidopsis gene related to APETALA2 Plant Cell. 8:659-71.

240. Winston, F., and P. Sudarsanam 1998. The SAGA of Spt proteins and
transcriptional analysis in yeast: past, present, and fllture Cold Spring Harb Symp Quant
Biol. 63:553-61.

24]. Wittsehieben, B. 0., G. Otero, T. de Bizemont, J. Fellows, H. Erdjument-
Bromage, R. Ohba, Y. Li, C. D. Allis, P. Tempst, and J. Q. Svejstrup 1999. A novel

histone acetyltransferase is an integral subunit of elongating RNA polymerase 11
holoenzyme Mol Cell. 4: 123-8.

242. Wolfl'e, A. P., and J. J. Hayes 1999. Chromatin disruption and modification
Nucleic Acids Res. 27:711-20.

243. Workman, J. L., and R. G. Roeder 1987. Binding of transcription factor TFIID
to the major late promoter during in vitro nucleosome assembly potentiates subsequent

initiation by RNA polymerase 11 Cell. 51:613-22.

244. Woychik, N. A. 1998. Fractions to functions: RNA polymerase II thirty years
later Cold Spring Harb Symp Quant Biol. 63:311-7.

234

245. Woychik, N. A., and M. Hampsey 2002. The RNA polymerase II machinery:
structure illuminates function Cell. 108:453-63.

246. Xiao, T., H. Hall, K. O. Kizer, Y. Shibata, M. C. Hall, C. H. Borchers, and B.
D. Strahl 2003. Phosphorylation of RNA polymerase II CTD regulates H3 methylation
in yeast Genes Dev. 17:654-63.

247. Xu, W., D. G. Edmondson, Y. A. Evrard, M. Wakamiya, R. R. Behringer,
and S. Y. Roth 2000. Loss of Gcn512 leads to increased apoptosis and mesoderrnal
defects during mouse development Nat Genet. 26:229-32.

248. Xu, W., D. G. Edmondson, and S. Y. Roth 1998. Mammalian GCN5 and
P/CAF acetyltransferases have homologous amino-terminal domains important for

recognition of nucleosomal substrates Mol Cell Biol. 18:5659-69.

249. Yamauchi, T., J. Yamauchi, T. Kuwata, T. Tamura, T. Yamashita, N. Bae, H.
Westphal, K. Ozato, and Y. Nakatani 2000. Distinct but overlapping roles of histone
acetylase PCAF and of the closely related PCAF-B/GCNS in mouse embryogenesis Proc
Nat] Acad Sci U S A. 97:11303-6.

250. Yang, X. J., V. V. Ogryzko, J. Nishikawa, B. H. Howard, and Y. Nakatani
1996. A p300/CBP-associated factor that competes with the adenoviral oncoprotein ElA
Nature. 382:3 19-24.

251. Zegerman, P., B. Canes, D. Pappin, and T. Kouzarides 2002. Histone H3
lysine 4 methylation disrupts binding of nucleosome remodeling and deacetylase (NuRD)
repressor complex J Biol Chem. 277:11621-4.

252. Zeng, L., and M. M. Zhou 2002. Bromodomain: an acetyl-lysine binding
domain FEBS Lett. 513:124-8.

253. Zhang, Y., and D. Reinberg 2001. Transcription regulation by histone
methylation: interplay between different covalent modifications of the core histone tails
Genes Dev. 15:2343-60.

235