.ﬁ‘ 61.2.11, . . . $332.»,
:53... “it? . . . . .1.
it... .3. .r . . . . . . 3:55.51?“
5 x: s.» 27...; .
.5 .. . 1.3mm} . , 1.2:... i
.1 {£311} {a . . .
. . . .5 :33. ER» . .
Airfﬂﬂomx. .
«use £32.: . . . . .w... ”nun“... a. 2. x: .
‘33.:1.‘ . . $5.... . S .3: .9.
. . i a. .531...
. . .3. . 55....5.
2.}. . .. _ . ‘ .Elxﬁwi:
2.: , :1... 3...: 5..
M 5 v in. I ~ i
.213... . . (iv:
givﬁv 5.1.. : .5 x .L. .5 . {$52.
. s E... c . .
. .LE‘iuiukrﬂiitsc. \L
{.3 gnu»... s...
. . 15.5.:
a . {$535.5
.313511 . .i. .1. . a}: . . .
35.51.... 324'. 35.9 .. 5:55:12}: . w
§v|pllllohsplai \~kIJ‘~ULSF§- 4 : . 1411-332: .gnip-lnﬁk . Eff-35:5,
rlou';§-;\.: - [‘530: . ﬂ .i-La . nu {.(Jir?
.1 . 1.... .1Es..:..i.i 5.15. 35.. 33:5... . .
“a! 5x1 :§c?x.€i1‘»i~lx§.}!.ts n
5: .3...&. 5:55:55: 5: .nnﬁqrxn..usz
1. . . . .3 i 5 , .. .5.
whirl: 1‘ c is x s . PF? 1.3%.}:
SE 3. 3...;ti
. Si... .:
3:15...

3

x r:

. EZSG:

4.5"“ 3, $55.5: 5 : _
15...... . 1:133:21xii4. @4453“.

n-1§~1ri~l v‘thl .ril. ti %2\ 5 g on

5:}..de ,1.._E..“..i{krijitrngsﬁ .drv

. 33.}. 6.1311. 111 1:2. .
, {munclgtélﬂwﬁnh : . .1S5’....«vluvuq§» w
ai}..i¢§ Ext! 3.....{155275

.E.s..«:.:.$i.za§.. .5.

. E». t\-l!\og. its<
Spicaiﬁr Y: _..1.\;$:2Lcnu.:3.
2.53.!)

z

:33.) 133’.

{5:5 Fifi”;
[itivuul‘ t5.
.3} .<¥(k§.: .

'3‘ I»)!
5.5. i. .
3.3.5:, 2.:
55:23... 122...:
54.55. 4......
:5....i 255%; .
a“: Kan». ﬁnes». . 5..
(Lt . .. . . . il.....vdlr.v..
, . .3252...
. L s

p 3!...‘0‘11‘
5.215»).

Kira/{(7. ‘

. . c.5ié..un.

. ’23)! . . ,4.

5.3.21 5

.5 555.1}. p24,: 1
t. 2

.(
3:. .12.)...
..)_§~:.)Z. .

v J. n
.555. 55.5?
Ctnxéxﬁ‘. ..

5.3.1.1131!) 22‘}
.{rfgti’xvlrlﬁlifz .
f7. . 3:115:73

71...... E i; I.

 

 

 

 

 

 

 

 

 

 

lllllllllllllllllll

This is to certify that the
dissertation entitled

Structural and Functional Comparison of the
Transcriptional Activating Domains of VP16
and RELA

presented by

Peter J. Horn

has been accepted towards fulﬁllment
of the requirements for

Ph.D. degree in Biochemistry

Maj/or {)rofeéxir

Or

Date 4 7 y

MSUI'J un Aﬂirmuliw Anion/Equal Opportunity lnrmmmn 042771

 

 

 

LIBRARY __

Miehigan State
University

 

 

STRUCTURAL AND FUNCTIONAL COMPARISON OF THE TRANSCRIPTIONAL
ACTIVATING DOMAINS OF VP16 AND RELA

By

Peter J. Horn

A DISSERTATION

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

DOCTOR OF PHILOSOPHY
Department of Biochemistry

1999

ABSTRACT
STRUCTURAL AND FUNCTIONAL COMPARISON OF THE TRANSCRIPTIONAL
ACTIVATING DOMAINS OF VP16 AND RELA
By

Peter J. Horn

The activation domains of many eukaryotic transcriptional activator proteins are
modular in nature. These domains are constructed of multimerized motifs, each
independently capable of directing activation. The activation domain of the herpes
Simplex virus activator, VP16, consists of two such motifs or "subdomains". Each is
sufficient to independently enhance transcription in vivo and in vitro. Similarly, the
activation domain‘of the mammalian activator RelA is separable into three independent
subdomains. For both activators, the transcriptional activity Of each subdomain is
critically dependent upon a Short cluster of hydrophobic residues. These Observations
raise at least two pertinent questions. First, what primary structural characteristics
contribute to subdomain ﬁmction? Second, does each subdomain function identically, or
do different subdomains contribute to activation through distinct mechanisms?

To evaluate these questions, both subdomains of VP16 (VP16N and VP16C) and
two Of the three subdomains OfRelA (RelAII and RelAIH) were examined by a
combination of site-directed and/or random mutational strategies. While each subdomain
was found to be dependent upon short hydrophobic segments for function, subtle
structural distinctions were observed. The VP16 subdomains each consist of a central,
critical F(D/E) dipeptide ﬂanked by hydrophobic residues of variable importance. The

precise spacing of the hydrophobic residues within the primary structure is distinct for

each subdomain. In contrast to either VP16 subdomain, the RelA subdomains are
dependent only upon hydrophobic residues for function. Both RelAII and RelAIII
exhibited a pattern of critical hydrophobic residues similar to that present in VP16C.

Despite some common structural properties, several results argue that the VP16
subdomains are not functionally identical. VP16N, and not VP16C, stimulated an
integrated reporter gene. VP16C efﬁciently activated a promoter containing an Initiator
(Inr) element, but not a promoter containing a TATA element, whereas VP16N directed
equal activation of both promoters. Finally, the VP16C subdomain, but not the VP16N
subdomain, stimulated formation of a TFIID-TFIIA-DNA (DA) complex. Similar
experiments showed that the RelAII and RelAIII subdomains, while effectively activating
only an Inr-containing promoter, did not enhance DA complex formation, indicating that
these two activities are unlikely to be functionally related.

These results demonstrate that the VP16 subdomains ﬁinction through distinct
mechanisms. However, both subdomains have similar structural properties. The VP16N
and VP16C subdomains differ only in the spacing of hydrophobic residues within their
core sequences. To test whether this Spacing is critical for subdomain-speciﬁc function,
the spacing of the hydrophobic residues within VP16N, was altered to resemble the
spacing observed in VP16C. This VP16N mutant was incapable of stimulating DA
complex formation, suggesting that the spacing Of the core hydrophobic residues within
the VP16 subdomains is insufﬁcient to confer subdomain-speciﬁc function. Residues

external to these core motifs must provide functional speciﬁcity.

 

to Ellen

 

ACKNOWLEDGMENTS

I would like to thank my committee members for their support and assistance.
Their time and effort is greatly appreciated. I would also like to thank Steve for his
encouragement and patience throughout my time in his lab. His efforts to allow me to

develop independently have been appreciated.

I would also like tO acknowledge members of the Triezenberg lab past and
present, for their companionship, friendship, and scientiﬁc help and advice. Iwould also
like to thank Naoko Kobayashi and Arnie Berk for their stimulating collaborations and

kind hospitality during my brief time at UCLA.

Most importantly, I would like to thank my wife, family and friends. They have
provided a constant source of support throughout the "ups-and downs" Of scientiﬁc

research. Without their help, none of this would have been achievable.

ACKNOWLEDGMENTS

I would like to thank my committee members for their support and assistance.
Their time and effort is greatly appreciated. I would also like to thank Steve for his
encouragement and patience throughout my time in his lab. His efforts to allow me to

develop independently have been appreciated.

I would also like to acknowledge members of the Triezenberg lab past and
present, for their companionship, friendship, and scientiﬁc help and advice. I would also
like to thank Naoko Kobayashi and Arnie Berk for their stimulating collaborations and

kind hospitality during my brief time at UCLA.

Most importantly, I would like to thank my wife, family and friends. They have
provided a constant source Of support throughout the "ups-and downs" of scientiﬁc

research. Without their help, none of this would have been achievable.

 

TABLE OF CONTENTS

 

 

 

List of Figures .............................................................................................................. viii
List of Symbols and Abbreviations ................................................................................. ix
Chapter One
Introduction and Literature Review ............................................................................ 1-50
Stepwise Model for Pre-initiation Complex Formation ......................................... 7
Holoenzyme Model of Transcription Initiation ..................................................... 9
Assembly of Pre-initiation Complexes on Inr Promoters .................................... 11
TBP binding constitutes the ﬁrst regulated step in PIC formation ....................... 12
TBP-associated factors may govern activator response ....................................... 21
TFIIB recruitment may stimulate PIC formation ................................................ 31
The basal transcription factor TFHA is required for activator response ............... 36
Activation may also involve non-basal auxiliary factors ..................................... 41
Other complexes are also potential coactivator candidates .................................. 46
Potential models for the function of activators ................................................... 48
Chapter Two
Activation Domins Contain Similar Yet Distinct Hydrophobic Motifs .................... 51-87
Introduction ....................................................................................................... 51
Experimental Results ......................................................................................... 55
The activation domains of VP16 are dependent upon similar
hydrophobic motifs ..................................................................... 55
A related “acidic” activator, RelA, also contains similar motifs .............. 64
Discussion and Conclusions ............................................................................... 75
Experimental Methods ....................................................................................... 8O
Recombinant DNA manipulations .......................................................... 80
PCR mutagenesis and screening of VP16C mutants ................................ 84
Mouse L cell transfection ....................................................................... 85
Chloramphenicol Acetyltransferase (CAT) Assays ................................. 86
Chapter Three
Distinct Activation Domains Function through Distinct Mechanisms ..................... 88-140
Introduction ....................................................................................................... 88

Experimental Results ......................................................................................... 90

VP16N and not VP16C activates transcription from an integrated

reporter ....................................................................................... 90
The VP16C activation domain is sensitive to basal promoter structure 93
The VP16C subdomain recruits TFIID and TFHA to a promoter

in vitro ........................................................................................ 99
Expression Of VP16N or VP16C activation domains inhibits
transcription in vivo .................................................................. 104
Are the RelA subdomains mechanistically similar to VP16C? .............. 111
The RelA activation domain displays Inr promoter preference .............. 111
RelA does not enhance DA recruitment to a promoter in vitro .............. 112
Can alteration of the hydrophobic core of VP16N reconstitute
VP16C activity? ........................................................................ 1 13
Discussion and Conclusions ............................................................................. 118
Experimental Methods ..................................................................................... 124
Recombinant DNA Manipulations ........................................................ 124
Tissue Culture Cell Manipulation ......................................................... 126
RNA Isolation ...................................................................................... 127
Primer Extension Analysis ................................................................... 128
Preparation of Gal4-Fusion Proteins ..................................................... 130
Preparation of eTFIID .......................................................................... 132
Preparation of His—tagged TFIIA .......................................................... 136
Preparation Of EMSA Probes ................................................................ 138
Gal4 EMSA Assay ............................................................................... 138
TFHA-TBP EMSA Assay ..................................................................... 139
DA Complex EMSA Assay .................................................................. 140
Chapter Four
Activation domains are composed of multiple, small modular elements ........... 141
Activator-target protein-protein complexes appear predominantly
hydrophobic ......................................................................................... 142
Modularity of activation domain structure ........................................................ 148
Bibliography ......................................................................................................... 159-184

vii

LIST OF FIGURES

Figure 1. VP16C mutations identiﬁed by genetic selection ............................................ 58

Figure 2. Relative activity of VP16N alanine substitutions in the activation domain

core region ......................................................................................................... 62
Figure 3. RelA Activation Domain Alignment ............................................................... 67
Figure 4. Comparison of VP16 and RelA sequences ...................................................... 68

Figure 5. Relative activity OfRelAII substitutions in the activation domain core
region ................................................................................................................ 70

Figure 6. Relative activity of RelAIII substitutions in the activation domain core

region ................................................................................................................ 72
Figure 7. Activity Of VP16 activation domains on an integrated reporter ....................... 92
Figure 8. Promoter Speciﬁcity Of individual VP16 and RelA subdomains ...................... 96
Figure 9. TFIIA stimulates TBP recognition of the AdML promoter TATA box ............ 98
Figure 10. Saturation Of the EMSA probe with Gal-AD fusion proteins ....................... 102

Figure 11. Comparison of the capacity of VP16N, VP16C and a mutant VP16N
subdomain to enhance DA complex formation ................................................. 103

Figure 12. Gal4-Vpl6C stimulates formation Of a TFIID-promoter complex

through recruitment of TFHA to DNA ............................................................. 105
Figure 13. In vivo competition experiments with NLS-VP16 and Gal4-VP16

expression vectors ............................................................................................ 108
Figure 14. The RelA activation domain does not stimulate DA complex formation ...... 115

Figure 15. Deletion of VP16N amino acids D441 and L444 produces a VP16C-like
core motif ........................................................................................................ 116

Figure 16. A VP16N mutant, VP16NAD/L, does not stimulate DA complex
formation ......................................................................................................... 117

viii

AdML
bp
CAT
EMSA
HCA
HEPES
HSV
IPTG
kDa
OD
PAGE
PBS
RNAP
SDS-PAGE
TAF
TFII

tk

VP16

LIST OF ABBREVIATIONS

activation domain

adenovirus major late

base pair

chloramphenicol acetyltransferase

electrophoretic mobility shift assay

hydrophobic cluster analysis
N-2-hydroxyethylpiperiazine-N'-2-ethanesulfonic acid
herpes simplex virus
isopropyl-B-D-thiolgalactopyranoside

kilODaltons

optical density

polyacrylamide gel electrophoresis

phosphate buffered saline

RNA polymerase

sodium-dodecyl-sulfate polyacrylamide gel electrophoresis
TBP associated factor

transcription factor, RNA polymerase H

thymidine kinase

virion protein 16

ix

Chapter One

Introduction and Literature Review

Accurate control of gene expression is essential for a wide variety Of biological
processes including cell growth and development, tissue specialization, metabolic
adaptation, immune response, angiogenesis and oncogenesis. As such, a thorough
understanding of the ﬁindarnental mechanisms governing this process is a central issue of
current biological research. Regulation of gene expression occurs through a variety of
transcriptional and post-transcriptional mechanisms. Transcriptional initiation and
elongation, differential mRNA splicing, mRNA stability and transport, translational
eﬁiciency, and poSt-translational protein modiﬁcations all modulate gene expression.
Transcriptional regulation of gene expression has been one Of the most extensively studied
aspects of gene regulation.

Unlike prokaryotic systems, the task of RNA production in eukaryotic nuclei is
divided between three polymerases. Ribosomal RNAS are transcribed by RNA polymerase
I. Transcription of protein—coding genes is performed by RNA polymerase H (RNAPII),
while RNA polymerase IH transcribes most small RNAS (some snRNAs are actually
RNAPH transcribed). While prokaryotic RNA polymerases are capable of promoter
recognition in the absence of additional factors, (the prokaryotic promoter recognition
factor, the polymerase 0' subunit, is a component of the polymerase itself), the eukaryotic
RNA polymerases require auxiliary factors for promoter recognition. These general
transcription factors (GTFs, also called basal transcription factors) are required to

coordinate recruitment of RNAPH to the promoter. The RNAPH GTF proteins are

thought to assemble in a sequential fashion into a promoter associated complex, referred
to as the pre-initiation complex (PIC). This PIC comprises the GTFs TBP (the TATA
binding protein of TFHD), TFHA, TFIIB, TFIIE, TFIIF, and TFIIH (Zawel, L. and
Reinberg, D., 1993; Orphanides, G. etal., 1996; Conaway, RC. and Conaway, J .W.,
1993). The assembly of the PIC is described in more detail in following sections.

Regulation of transcription from RNAPII promoters requires an additional set of
proteins, often referred to as speciﬁc transcription factors. These factors comprise two
general groups: transcriptional activators and transcriptional repressors (Hahn, S., 1993;
Triezenberg, S.J., 1995; Hanna-Rose, W. and Hansen, U., 1996). Eukaryotic activators,
the major topic of this text, are generally bipartite in structure, composed of two
independent functiOnal domains (Ptashne, M., 1988; Mitchell, R]. and Tjian, R, 1989).
One domain is essential for promoter recognition (either through direct protein-DNA or
protein-protein interaction with additional DNA-binding factors), while the second domain
ﬁmctions to activate transcription. Artiﬁcial chimeric activators composed of ﬁrsions of a
DNA-binding domain and an activation domain are functional in vitro (Carey, M. et al.,
1990; Lin, Y.S. et al., 1988) and in vivo (Sadowski, I. et al., 1988; Walker, S. et al., 1993;
Seipel, K. et al., 1992). For example, a fusion of the DNA-binding domain of the yeast
activator Gal4 with the herpes simplex virus (HSV) transactivator VP16 produces an
unusually potent transcriptional activator (Sadowski, I. et al., 1988). This synthetic
activator has been utilized as a prototype in a variety of studies, many of which are
reviewed in this text.

While the structural features essential for activation domain function remain

somewhat poorly understood, their function is clearly dependent upon the hydrophobic

arnino acids Of these domains (Triezenberg, S.J., 1995; Hahn, S., 1993). Early
experiments with the yeast activator GAL4 suggested that acidic amino acids in one of
two GAL4 activation domains were of principal importance to transcriptional activity
(Ma, J. and Ptashne, M., 1987; Gill, G. and Ptashne, M., 1987). Deletion analysis of
another yeast activator GCN4 also localized the activation domain to a large region of
extensive acidic character (Hope, LA. and Struhl, K., 1986). These early observations led
to the supposition that the sole structural determinant for activation domain function was a
primary structure rich in acidic residues (Hope, IA. and K. Struhl, 1986).
Characterization of other transcriptional activators has also resulted in the identiﬁcation of
activation domains rich in glutamine, proline, or isoleucine residues. This led to a
proposal for classiﬁcation of activator classes deﬁned by the predominance of these
residues in activation domain sequences (Mitchell, RI. and R. Tjian, 1989).

However, studies using the acidic HSV activator VP16 identiﬁed hydrophobic
residues, and not acidic residues, as critical for transactivation (Cress, W.D. and
Triezenberg, S.J., 1991). In addition, more reﬁned analyses of the activation domain of
the yeast activator GCN4 by site-directed mutagenesis identiﬁed hydrophobic amino acids
within this region as the critical residues for transactivation (Drysdale, C.M. et al., 1995;
Jackson, B.M. et al., 1996). Other prototypical "acidic" activators also display similar
characteristics. The mammalian tumor suppressor p53 requires hydrophobic residues
within the N-terminal transactivation domain for activation (Lin, J. et al., 1994), and the
NF-KB family member RelA requires multiple hydrophobic residues for function (Schmitz,

ML. et al., 1994; Schmitz, M. et al., 1995; Blair, W.S. et al., 1994b). Unrelated (“non-

acidic”) activators such as the glutamine-rich activators Spl (Gill, G. et al., 1994) and
Oct-1 (Tanaka, M. and Herr, W., 1994) also require hydrophobic residues for function.

Classiﬁcation of activator subtypes merely by amino acid content may not be
particularly meaningful. Instead, classiﬁcation based upon activator function might be a
more useful method. Several studies suggest that activators may function through distinct
mechanisms. The Schaffner laboratory compared the ability of various Gal4-activation
domain ﬁrsions to ﬁrnction from cis elements distal (enhancer) and proximal (immediately
5’ of TATA sequences) to an eukaryotic promoter. While some activators (e. g. VP16)
were insensitive to promoter positioning, others (6. g. Spl) required a proximal location
for efficient function. All activation domains examined functioned from promoter
proximal positions (Seipel, K. et al., 1992). Similar results were also observed in yeast
(Kunzler, M. et al., 1994).

Other Observations also suggest that different activators function through distinct
mechanisms. Studies Of several activators by the Bentley (Blau, J. et al., 1996; Yankulov,
K. et al., 1994) and Groudine (Krumm, A. et al., 1995) laboratories established that
polymerase elongation constitutes a rate—limiting step that some, but not all activators
stimulate. Blau et al. proposed classiﬁcation of activators into three groups based upon
their ability to stimulate initiation (e. g. Spl), elongation (e. g. Tat) or both (e. g. VP16,
HSF) events. In vitro and in vivo studies by the Smale laboratory identiﬁed differential
sensitivity of activators to basal promoter architecture, again suggesting that activators
may enhance transcription through distinct mechanisms. The VP16N activation domain

(see below) demonstrated no dependence upon basal promoter structure. In contrast, the

Spl activation domain was highly dependent upon initiator sequences for efﬁcient function
(Emami, K.H. et al., 1995).

Diﬂ‘erent activation domains also display distinct functions in vitro. TAFs are
required for activated transcription in vitro, and when assembled with TBP, comprise the
transcription factor TFIID (reviewed in Burley, SK. and Roeder, KG, 1996). TBP is
essential for transcription by all three RNA polymerases (Hernandez, N., 1993). Speciﬁc
TAFs are likely to provide speciﬁcity to TBP complexes in vivo, as distinct TBP-TAF
complexes have been isolated (Lee, TI. and Young, RA, 1998). TFIID subcomplexes
containing partial TAF complexes are also differentially responsive to diﬁerent activators,
consistent with the idea that different activators target distinct TAFs for recruitment of
TFIID. Distinct partial T AF complexes mediate for NTF-l and Spl function in vitro
(Chen, J. et al., 1994). The VP16C subdomain also interacts with the human TAFS
hTAFn32 and hTAFu7O (Klemrn, R. et al., 1995), as well as the Drosophila homologue
dTAFn40 (Goodrich, J.A. et al., 1993). These interactions are speciﬁc for VP16C. The
VP16N subdomain does not bind either Of these TAFS. The Epstein-Barr virus activator
Zta and HSV VP16 also stimulate formation of a TFIID-TFIIA-promoter (DA) complex
in vitro, as assessed by an agarose EMSA technique (Lieberman, RM. and Berk, A.J.,
1994; Kobayashi, N. et al., 1995). The C-terminal half of the VP16 activation domain
(VP16C, amino acids 456-490) stimulates formation of this complex, while the N-terminal
half (VP16N, amino acids 410-456) does not stimulate formation of this complex
(Kobayashi, N. et al., 1995, see below). This activity has not been observed in any other
activators. This result suggests that even separate domains within a single activator may

function through distinct mechanisms. Finally, a variety of studies have observed

differential eﬂ‘ects of reporter chromosomal integration on activator ﬁrnction in vivo,
suggesting a differential capacity of activators to antagonize the repressive effects of
chromatin organization (reviewed in Smith, CL. and Hager, G.L., 1997).

The precise mechanism of function of any speciﬁc activator has remained elusive
despite a great deal of experimental effort by a large number of investigators.
Conceptually, activators may function at any step in transcription. Chromatin structure is
commonly regarded to be inhibitory to transcription. Acetylation of nucleOsomal histones
correlates with enhancement of gene expression. Derepression of chromatin packaged
genes may be facilitated by recruitment of histone acetyltransferases such as the
GCNS/ SAGA or other acetyltransferase complexes (reviewed in Struhl, K., 1998).
Activators may also recruit chromatin remodeling complexes such as SWI/SNF or RSC to
genes (reviewed in Kingston, RB. et al., 1996). These complexes facilitate access Of
nucleosomal DNA to DNA binding proteins and may be required prior to activator and
PIC recognition of regulatory or promoter DNA sequences.

Original models for activator function largely involved stimulation Of either the
rate of formation or stability of the PIC. Consistent with these proposals, protein-protein
interactions between activation domains and GTFs have been observed (reviewed in
Roeder, KG, 1996). Activators may also induce isomerization of the PIC or a partial
PIC to an activated state, essentially acting as allosteric effectors of PIC activity (Chi, T.
and Carey, M., 1996). Conceptually, activators could also enhance promoter escape
(Kumar, K.P. et al., 1998; Goodrich, IA. and Tjian, R., 1994), facilitating the conversion
of the initiation complex to an elongation complex and ﬁnally, activators may function

through stimulation of elongation (Bentley, BL, 1995). Any or all of these mechanisms

may be Operative. The following sections will focus on a brief description Of the initiation-
elongation cycle followed by a more detailed survey Of the potential cross talk between

activators and speciﬁc GTFS or regulatory complexes.

Stepwise Model for Pro-initiation Complex Formation

In vitro studies utilizing puriﬁed and recombinant basal transcription factors have
led to the proposal of a stepwise model (Buratowski, S. et al., 1989; reviewed in Roeder,
R. G., 1996; Conaway, RC. and J.W. Conaway, 1993) for formation of an RNAPH pre-
initiation complex (PIC). These studies have largely been conﬁned to a subset of
promoters, primarily strong viral promoters containing either consensus or near consensus
TATA elements (i.e. the adenovirus Elb, major late, and E4 promoters). Recent studies
have led to an appreciation of the complexity of eukaryotic promoters (reviewed in Smale,
ST, 1997). Eukaryotic promoters appear to be composite in nature. In addition to well
recognizable TATA sequences, sequences within the initiation site (initiator, or Inr
element) (Oshea-Greenﬁeld, A. and Smale, ST, 1992) as well as downstream of the
initiation site (DPE; downstream promoter element) contribute to promoter ﬁrnction and
strength (Burke, T.W. and Kadonaga, J .T., 1996). Thus, current models, which have been
elucidated through study of strong viral promoters may be representative Of events at only
a subset of eukaryotic promoters. A limited number of studies have been performed with
TATA-less (Inr- containing) promoters, and some data suggest that PIC formation
pathways may be distinct for TATA and Inr promoters (for a review of Inr-mediated PIC

assembly, see Smale, ST, 1997).

Pre-initiation complex formation on the adenovirus major late promoter (AdML
promoter) initiates with recognition of the TATA box by TFIID. The TFIID complex is
composed Of the TATA binding protein (TBP) and a collection Of 8-12 additional
polypeptides (TAFs, TBP associated factors). Recognition of the promoter by TFIID
(Lieberman, PM. and A]. Berk, 1994; Ozer, J. et al., 1998) or TBP (Cortes, P. et al.,
1992; Imbalzano, A.N. et al., 1994b) is facilitated in vitro under certain experimental
conditions by the auxiliary transcription factor TFHA. This TFIID-TFIIA-TATA ("DA"
complex) can recruit TFIIB to form a "DAB" complex, which in turn recruits RNAPII.
Recruitment of RNAPH is thought to occur through the association of TFIIF to the DAB
complex. Interestingly, TFHB can be isolated in a RNAPH complex and TFHF is tightly
associated with this RNAPII complex. These observations suggest a model whereby
RNAPII recruitment is facilitated by TFIIB and TFIIF subunits of a larger RNAPII
complex (see Conaway, RC. and J .W. Conaway, 1993; an early proposal of the
competing holoenzyrne model for PIC formation). Association Of TFHB and TFIIH with
the promoter bound RNAPII complex completes PIC formation, facilitating the release of
the promoter bound complex.

Formation of the PIC is insufﬁcient for transcription initiation. Efﬁcient
transcription of RN APII is unique among eukaryotic RNA polymerases in its requirement
for hydrolysis of the [3-7 bond of ATP (Bunick, D. et al., 1982). Transition to an initiation
complex may require ATP hydrolysis to facilitate opening of the DNA helix at the start
site of transcription, as evidenced by the increased reactivity of promoter sequences with
the single-stranded DNA speciﬁc reagent KMnO4 (Wang, W.D. et al., 1992). Formation

of this "open complex" is followed by a period of abortive initiation, characterized by the

 

production and release of short RNA molecules (Bunick, D. et al., 1982). Tjian and
coworkers argue that release from this state, in a TFIIB and TFIIH dependent fashiOn,
completes the initiation cycle (Goodrich, J .A. and R. Tjian, 1994). TFIIF may also be
involved in this process. The RAP74 subunit is essential for assembly of the PIC but also
plays an essential role in early elongation (Chang, C. et al., 1993). Release ofRNAPII
correlates with the hyperphosphorylation Of a C-terminal domain (CTD) of the largest
subunit ofRNAPII (Payne, J .M. et al., 1989; Layboum, PI. and Dahmus, ME, 1990).
The activity Of this elongation complex can be inﬂuenced by a number Of additional
transcriptional elongation factors including TFIIS, elongin and TFIIF (for review, see Aso,
T. et al., 1995). Some activators are also capable of stimulating elongation (Yankulov, K.

et al., 1994).

Holoengme Model of Transcription Initiation

A competing model for initiation complex formation has gained considerable
attention in the last three to ﬁve years (Koleske, A]. and Young, RA, 1995; Orphanides,
G. et al., 1996). AS mentioned above, early Observations suggested the possibility that the
RNAPH machinery may exist in a larger complex or "holoenzyme". Genetic and
biochemical studies Of the large subunit Of RNAPII by Young and coworkers provided
convincing evidence for the existence of a holoenzyme in yeast. Young and colleagues
were studying the role of the CTD of RNAPH. Deletion of the CTD results in cell death
and an allele with a synthetic truncation of half of the CTD produces a conditional lethal
phenotype. Young and coworkers isolated suppressor mutations of this phenotype in

what they have called SRB genes (Suppressor ofRNA polymerase B). The proteins

 

encoded by these genes are associated with RN APH in cell-free extracts (Thompson, C.M.
et al., 1993), although only a subset of RNAPH appears to be present in these holoenzyme
complexes. Young and coworkers have estimated that 2-4,000 SRB molecules exist per
yeast cell, while RNAPII numbers 10-20,000 molecules (see Lee, TI. and RA. Young,
1998). Kornberg and coworkers simultaneously identiﬁed a high molecular weight
complex that associates with the CTD and is sufﬁcient to confer transcriptional response
to activators in vitro (Kelleher III, R]. et al., 1990). This "mediator" complex contains
the SRB proteins (Kim, Y.-J. et al., 1994). Characterization of the Young holoenzyme
has veriﬁed that initiation factors TFIIB, TFIIF, TFIIH, SRBS and the previously
identiﬁed SUGI and GAL]! gene products are present in this complex (see Koleske, A].
and RA. Young, 1995). Since quantitation of cellular levels of proteins by comparative
Western blotting of RNAPH and SRB proteins has determined that only a portion of
RNAPH (20-40%) is present in holoenzyme complexes, it is possible that a subset of
RNAPII (perhaps a specialized species), is present in a pre-assembled initiation complex
lacking only TFIID and TFIIA. These results also suggest that RNAPII recognition of a
promoter can occur in a two-step fashion: TFIID recognition of the promoter followed by
RNAPII holoenzyme recognition of TFIID.

Interestingly, the holoenzyme supports activated transcription in vitro (Koleske,
Al and Young, RA, 1994). This observation has been conﬁrmed using analogous
metazoan preparations (Maldonado, E. et al., 1996). In fact, the mediator activity isolated
by Kornberg and colleagues was originally identiﬁed as a column fraction capable of
restoring activator response to core RNAPH in the presence of TBP (Kelleher III, R]. et

al., 1990). These results seem to suggest that TAF presence is unnecessary for activator

10

response in vitro. This observation is contrary to several other in vitro analyses (see

below; TAF section).

Assembly of Pre-initiation complexes on Inr promoters

In contrast to the better understood events occurring on TATA containing
promoters, the process of PIC formation on TATA-less promoters is less well understood.
Comparatively little work has been performed on Inr containing promoters, and most
studies of Inr function have been performed using TATA-containing promoters that also
contain Inr elements (for instance the viral AdML promoter).

Proposals for PIC assembly on Inr containing promoters have suggested an
analogous but distinct step-wise pathway to that elucidated for TATA containing
promoters (reviewed in Smale, ST, 1997). Work by Smale and coworkers suggests that
TATA-less promoter ﬁrnction requires the same ensemble of GTFS, with one notable
exception. TATA-less basal transcription requires TAFS, whereas basal TATA ﬁinction
occurs with TBP (Kauﬁnann, J. and Smale, ST, 1994; Verrijzer, C.P. et al., 1995;
Martinez, E. et al., 1994). Since Inr promoter recognition cannot occur through TATA
sequences, this result suggests that the TAFs themselves may function in promoter
recognition. Roeder and colleagues also Obtained evidence that Inr-function may require
factors in addition to TAFs for efﬁcient TAT A-less transcription (Martinez, E. et al.,
1998a). The identity of these “TIC” factors remains unknown. Other models for
promoter recognition include recognition of the Inr element by Inr-binding proteins, a
number of which have been proposed. TFII-I (Roy, AL. et al., 1993), YYl (Usheva, A.

and Shenk, T., 1994), and even RNAPII itself (Carcamo, J. et al., 1991) have all been

11

demonstrated to recognize Inr-like elements. In the near future, a clearer picture of PIC

formation on TATA-less promoters is likely to emerge.

TBP binding constitutes the first regulated step in PIC formation

The ﬁrst step in PIC formation on the majority Of promoters is recognition of the
TATA box by the TATA- Binding Protein (TBP). The TBP protein is composed of two
domains. The N-terminal domain is less well conserved among eukaryotic species. The
C-terminal domain (Often referred to as the "core" domain) is highly conserved in all
eukaryotic species analyzed (Hernandez, N., 1993). The DNA sequence of this domain is
composed of two direct repeats, suggesting that TBP may have evolved by gene fusion of
two monomeric ancestors (Yamamoto, T. et al., 1992). Crystal structures of the core
domain of the yeast (Kim, Y.C. et al., 1993), Arabidopsis (Nikolov, D.B. et al., 1992), and
later human (Nikolov, D.B. et al., 1996) TBP molecules reinforced this conjecture. The
structure of the molecule resembles a saddle, with a convex upper surface and a lower
concave surface Of sufﬁcient size to accommodate the DNA helix. The molecule displays
two-fold pseudo-symmetry consistent with the sequence similarity of the two repeats.
Each repeat folds into a twisted beta sheet, topped by a long C-terrninal alpha helix. The
ﬁrst and second beta strands are interrupted by a short alpha helix, producing a loop
between the adjoining strands forming a "stirrup" on either side of the molecule. These
stirrups are involved in protein-protein interactions of TBP with TFHA and TFIIB.

Although a consensus TATA sequence has been identiﬁed, DNA sequences
recognized by TBP are relatively degenerate. Even non-TATA sequences at TATA

positions in apparently TATA-less promoters contact TBP in some cases (see Martinez, E.

12

 

et al., 1994 for discussion). The crystal structure Of TBP with DNA suggests a possible
molecular basis for this promiscuity (Kim, Y.C. et al., 1993). The concave underside of
the TBP molecule, composed of the twisted beta sheets Of the two repeats, binds the
TATA motif, producing an extensive distortion of the DNA helix evidenced as a broad
widening of the minor groove and introduction Of a severe kink in the path of the DNA
helix. Since few sequence Speciﬁc contacts were Observed, Sigler proposed that the
capacity of a DNA sequence to accommodate this conformation determines the quality of
a TATA motif (Kim, Y.C. et al., 1993). The general transcription factors TFHA and
TFIIB associate with TBP and DNA, contributing additional protein-DNA contacts likely
to contribute signiﬁcantly to PIC stability. This is apparent ﬁom the results Of structural
studies from the Sigler and Burley laboratories that provide a glimpse of the early stages
of PIC assembly. The structures of both a TBP-TFHB-DNA (Nikolov, D.B. et al., 1995)
and TBP-TFIIA—DNA (Geiger, J .H. et al., 1996) complex have been solved.
Superposition of these complexes has provided a high-resolution model for the TBP-
TFIIA-TFHB-DNA complex (Nikolov, DB. and Burley, SK, 1997; Geiger, J .H. et al.,
1996). The two GTFS bind Opposite faces of the TBP-DNA complex, making direct
protein-protein contacts with TBP. TFIIA binds to TBP near the N-terminal stirrup,
associating with DNA upstream of the TATA box, while TFHB clamps the C-terminal
stirrup. The TFIIB N-terminus is oriented toward the start site consistent with genetic
studies in yeast that have implicated TFIIB in start site selection (Pinto, I. et al., 1992).
The mammalian transcription factor TFIID was originally identiﬁed as a
component Of a chromatographic fraction isolated ﬁom HeLa nuclear extracts (Reinberg,

D. et al., 1987). This activity was demonstrated to protect the promoter region of the

13

AdML promoter, including the TATA box and downstream sites in the vicinity of the
initiation site (Horikoshi, M. et al., 1988b). Shortly thereafter, a TATA-binding protein
was identiﬁed in Saccharomyces cerevisiae, puriﬁed to homogeneity and cloned
(Cavallini, B. et al., 1989; Hahn, S. et al., 1989; Horikoshi, M. et al., 1989; Schmidt, MC.
et al., 1989). Identiﬁcation of this gene rapidly facilitated the identiﬁcation ofDrosophila
(Hoey, T. et al., 1990; Muhich, ML. et al., 1990) and mammalian TFIID (Homnann, A.
et al., 1990; Kao, C.C. et al., 1990; Peterson, MC. et al., 1990). Metazoan TFIID was
subsequently identiﬁed as a high molecular weight complex of TBP and TAFs (some 8-12
polypeptides dependent upon isolation method and species; reviewed in Tansey, WP. and
Herr, W., 1997b; Lee, TI. and RA. Young, 1998). The role of TAFs in PIC formation
and transcriptional regulation will be reviewed separately below. Two TBP genes have
been identiﬁed in Arabidopsis (Gasch, A. et al., 1990). Gene duplication in plants is a
relatively common phenomenon, suggesting that the existence of multiple TBP proteins
may be unique to this kingdom. However, a second TBP-like gene has also been
identiﬁed in Drosophila (Crowley, T.E. et al., 1993). This related gene encodes a TBP
variant with extensive C-terrninal sequence identity and nonconserved N-terminal domain.
The second TBP gene is expressed in neuronal tissues and appears to be involved in
tissue-speciﬁc transcription. Early efforts to substitute this TBP variant for the
ubiquitously expressed basal TBP suggested that the variant was not capable Of
substituting as a GTF. However, Tjian and colleagues recently reported that this variant
TBP is capable of ﬁrnctioning as a GTF. The TBP-like protein appears to be associated
with a distinct set of TAFs, suggesting that tissue-speciﬁc transcription in Drosophila

neuronal tissue is mediated by a tissue-speciﬁc TFIID complex (Hansen, S.K. et al., 1997).

14

TBP has been implicated as a target of transcriptional activators (and repressors)
by both genetic and biochemical methodologies. Consistent with a possible role for TBP
in activator response, TBP has been demonstrated to directly associate with a number of
transcriptional activators in vitro. This interaction was ﬁrst demonstrated using VP16
(Stringer, K.F. et al., 1990). Mutations in VP16 that reduce transactivation in vivo also
abolish TBP binding in vitro, consistent with a role for the TBP-VP16 interaction in vivo
(Ingles, O]. et al., 1991). Physical interactions between TBP and activators have also
been documented with Spl (Emili, A. et al., 1994), the Oct-1 and Oct-2 POU speciﬁcity
domains (Zwilling, S. et al., 1994), c-Rel (Xu, X. et al., 1993), p53 (Chang, J. et al.,
1995), the Epstein-Barr Virus transactivator Zta (Lieberman, PM. and Berk, A.J., 1991),
and E2F (Emili, A; and Ingles, C.J., 1995), among many others. Most of these
experiments were performed using recombinant TBP without TAF subunits. Since the
TFIID complex comprises TBP and TAFs, the relevance of these interactions has been
challenged. Given the large number of TAPS, it is reasonable to presume that TAFs might
occlude any activator binding site. However, the interaction Of Zta and TBP has been
demonstrated to occur in the presence of TAFS (Lieberman, PM. and AI. Berk, 1991).
TBP-repressor interactions have also been described. Notably, the p53-TBP interaction
observed in vitro has been implicated in p5 3-mediated repression of transcription (Seto, E.
et al., 1992). The Observation that repressors are also capable of binding TBP suggests
that simple recruitment to a promoter per se, is not sufﬁcient to activate transcription.
Interestingly, TBP residues implicated in VP16 binding map to the DNA-binding surface
on the underside of the TBP “saddle” (Kim, T.K. et al., 1994). This suggests that VP16-

binding and DNA binding may be competitive, an intuitively contradictory result.

15

However, TBP (Coleman, RA. et al., 1995) and TFIID (Taggart, AKP. and Pugh, BE,
1996) can dimerize through the DNA-binding surface. Thus, VP16 may enhance PIC
formation by facilitating TBP dimer dissociation. Consistent with this possibility,
dissociation Of dirneric TBP has been implicated as a slow step in TATA—recognition in
vitro (Coleman, RA. and Pugh, BE, 1997).

A number of genetic experiments in yeast have also implicated TBP as a potential
target for transcriptional activators. Struhl and co-workers have identiﬁed TBP-TATA
recognition as a limiting step in transcription in vivo that can be enhanced by activators
(Klein, C., Struhl, K., 1994). This result is consistent with mammalian studies suggesting
that TFHD recognition of a promoter may be limiting for transcription in vivo. Colgan
and Manley foundthat overexpression of TBP in vivo stimulated transcription ﬁ'om TATA
containing reporter genes (Colgan, J. and Manley, IL, 1992). Interestingly, artiﬁcial
tethering of TBP to a composite promoter containing LexA binding sites (by expression of
a LexA-TBP fusion protein) activated transcription, suggesting that enhancement of TBP
binding to a promoter is sufficient to activate a gene (Chatterjee, Struhl, K., 1995). It is
unknown whether the LexA-TBP ﬁrnctions as native TBP (to nucleate formation of a
PIC), or whether the fusion fortuitously produced an artifactual activation domain.
Veriﬁcation that gene fusions to DNA-binding domains can serendipitously produce
artiﬁcial activators comes from early studies by the Ptashne laboratory. Synthetic
activators were identiﬁed by screening a library of Gal4 ﬁisions produced from random E.
coli DNA (Ruden, D.M. et al., 1991). It would be useful to know if TBP mutants that
diminish DNA binding also diminish the activity of this LexA-TBP ﬁrsion protein. Klages

and Strubin produced an artiﬁcial recruitment of TBP by introduction Of a leucine zipper

l6

into TBP. Binding of a leucine zipper protein upstream Of a promoter activated
transcription by the altered TBP, suggesting that TBP recruitment is sufﬁcient to activate

transcription in vivo (Klages, Strubin, M., 1995).

Classical genetic experiments have also yielded a great deal Of information
regarding TBP residues involved in essential protein-protein and protein-DNA
interactions. Identiﬁed TBP mutant alleles can be generally grouped into four classes
(reviewed in Hampsey, M., 1998): (1) DNA-binding speciﬁcity mutants, (2) TFHA
interaction mutants, (3) TFIIB interaction mutants, and (4) activation-defective mutants

with normal basal function.

The DNA-binding speciﬁcity mutants of TBP were the ﬁrst identiﬁed. SPT15 was
originally identiﬁed as a gene essential for accurate promoter usage by the transposable Ty
element of Saccharomyces cerevisiae. The 579115-122 allele contains a L205F mutation
that alters TBP afﬁnity for non-consensus TATA boxes, with transcription switching to a
downstream HIS3 non-consensus CATAAA box (Amdt, K.M. et al., 1992). Introduction
of an analogous mutation in the N-terminal TBP repeat results in preference for a
TATAAG sequence, an indication that the polarity of TATA recognition was inherent to
TBP and not conferred in combination with other components of the basal machinery

(Amdt, KM. et al., 1994).

Interestingly, a genetic selection constructed to identify activation-defective TBP
alleles that did not interfere with basal function isolated mutations within the DNA-binding
interface (Lee, M. and Struhl, K., 1995). All four mutants displayed alterations in DNA

binding and normal TFHB and Gal4-VP16 interactions. Another screen designed to

17

identify Ind and Gal' TBP alleles also identiﬁed residues involved in DNA binding (Reddy,
P. and Hahn, S., 1991). These phenotypes are commonly associated with defects in
transcriptional activation (Amdt, K.M. et al., 1995). Collectively, these results argue that
TBP-TATA recognition itself may be an active contributor to activator response in viva.
Perhaps certain activators prefer certain core promoter elements, and alteration of TBP-

TATA interactions inﬂuences activator-core promoter cooperation.

The second class of TBP mutants involves disruption Of the TBP-TFHA interface.
Perhaps one of the most intriguing TBP mutants is a member of this class. A genetic
screen devised to identify temperature-sensitive (ts) TBP alleles with normal RNAPIII
transcription identiﬁed a K138T/Y 139A double mutation. This double mutant does not
respond to several activators, and fails to interact with TFHA, suggesting that TBP-TFHA
association is critical for activated transcription in vivo (Stargell, LA. and Struhl, K.,
1995). This result correlates with the requirement for TFHA for in vitro transactivation in
mammalian systems (see TFHA section below). Interestingly, these residues are located
on the H2 helix on the convex "top" of the protein, nowhere near residues involVed in
TFIIA contact as indicated in the TBP-TFHA crystal structures (Geiger, J.H. et al., 1996;
Tan, S. et al., 1996). However, the TFIIA derivatives used for crystallization lack a
central protein segment likely to be disordered in the native protein. This region was
excluded to facilitate crystallization. Perhaps a segment of this disordered domain
contacts the H2 helix directly. Consistent with this possibility, ﬂuorimetric and NMR
Studies using synthetic peptides corresponding to this region of TBP and a TFHA segment

from this central lOOp appear to associate in vitro (Ramboarina, S. et al., 1998). These

18

data agree quite well with the phenotypes of similar site-directed substitutions. Two
double mutants of TBP (K133/K138 and K133/K145 mutants) are defective for TFIIA
binding as evidenced by disruption of TBP-TFHA—DNA complexes in gel shift
experiments (Buratowski, S. and Zhou, H., 1992). These studies are also consistent with
in vitro Observations in mammalian transcription systems that implicate TFHA as a critical

component for activator response (see TFIIA section below).

Another interesting TBP allele appears to affect the TBP-TFHA interaction. Blair
and Cullen identiﬁed a TBP mutant with enhanced RNAPII activity (Blair, W.S. and
Cullen, BR, 1997). Gain-of-function mutants of the transcriptional apparatus are
extremely rare. The phenotype arises from a single substitution, N69S. This mutation
enhanced transcription from weak promoters without altering DNA binding affinity.
Secondary mutations that diminished TBP-TFHA interaction blocked the N698

phenotype, consistent with the possibility that TFHA is involved in this process.

The third class of TBP mutants involves disruption of the TBP-TFIIB interface.
Certain yeast TBP mutations (L1 14K, L189K, or K271L) were found to diminish Gal4-
VP16 activator response in vivo (Kim, T.K. et al., 1994). These mutants also displayed in
vitro TBP-TFIIB interaction defects. Coupled with previous results suggesting that VP16
may directly bind TBP (Stringer, K.F. et al., 1990), the authors suggest that Gal4-VP16
may stimulate both TBP and TFIIB recruitment (Kim, T.K. et al., 1994). This proposal is
consistent with the Observation by Green and coworkers that TFIIB recruitment to a
promoter can be enhanced by activators (Lin, Y.S. and Green, M.R., 1991). A similar

conclusion was reached by Tansey and Herr. They designed an E284R mutation to impair

l9

 

the TBP-TFIIB interaction. This mutation blocked TBP-TFIIB interaction in vitro and
severely impacted activator response in viva. An additional R169E mutation in TFIIB
produced an allele-speciﬁc suppressor of the TBP mutant. The authors interpret this as
evidence that activators function through recruitment of TFIIB (Tansey, WP. and Herr,

W., 1997a).

In contrast, Lee and Struhl identiﬁed several TBP mutants (E186A, E188A, and
L189A) that are defective in TBP-TFIIB interaction in vitro, yet are competent to support
activated transcription by several activators in viva (Lee, M. and Struhl, K., 1997). The
reason for this disparity is unclear, although the nature of the substitutions may play a role.
Perhaps the charge reversal introduced by Tansey and Herr produces a more severe
phenotype which activators can overcome, whereas the alanine substitutions introduced by
Lee and Struhl are less disruptive and therefore do not require activator for compensation.
Consistent with this idea, Berk and colleagues have observed similar activator response
defects in human TBP mutants with alterations of these corresponding human TBP
residues (E284R, E286K or L287E) when charge reversal mutations of the type

employed by Tansey and Herr were introduced.

This particular study by Berk and colleagues (Bryant, G0. at al., 1996) is the
most comprehensive study of TBP residues involved in transcriptional response that has
been performed in viva. They produced 89 separate substitutions (charge reversals or
introduction of large polar residues for hydrophobic residues) Of residues predicted by
crystal structure to be surface exposed in human TBP. With the Objective of identifying

TBP residues involved in activator response in viva, they discarded those that abolished

20

 

DNA-binding in vitro, as well as those that radically altered global protein structure.
Activation by Gal4-VP16 and Gal4-E1A(CR3) was assessed with two different basal
promoters in viva. The authors identiﬁed four clusters of TBP residues essential for
activator function. Two patches, one on the upstream side of TBP (relative to the start
site) and one on the stirrup of the second repeat of TBP, correspond to identiﬁed TFIIA
and TFIIB binding sites, respectively. The other two sites, representative of the fourth
class of TBP mutations, are less easily explained. One set contains residues in the H2
helix analogous to those described by Stargell and Struhl (Stargell, LA. and K. Struhl,
1995). The second cluster occurs on the downstream side of TBP. The authors postulate
that these sites are likely to be involved in interactions with TAFs. However, in light of
the observations of Stargell and Struhl, the H2 region could also be involved in secondary

contacts with TFHA.

TBP-associated factors may govern activator response

The observation that activators directly interact with TBP suggests that activators
may mediate TBP recruitment to promoters. However, TBP is actually associated with
multiple accessory proteins (TAFs) in viva. Two lines of investigation suggest that these
TAFS are involved in transcriptional regulation. The ﬁrst and perhaps most direct
evidence evolved from the original analysis and identiﬁcation of TBP. Yeast TBP can be
isolated from cell—free extracts as a single polypeptide. This allowed its puriﬁcation and
identiﬁcation as the SPT15 gene product. However, metazoan TBP was more difﬁcult to

purify to homogeneity from cell-free extracts or nuclear extracts.

21

Additionally, recombinant human TBP or yeast TBP did not support activated
transcription in vitro (Pugh, BF. and Tjian, R., 1990), while the impure TFIID ﬁaction
(the 0.85M KCl eluate from a P-ll fractionated nuclear extract) robustly supported
activated transcription (Horikoshi, M. et al., 1988a; Horikoshi, M. et al., 1988b;
Sawadogo, M. and Roeder, R.G., 1985). This inconsistency was resolved by the
identiﬁcation of a collection of TBP-associated factors (TAFS) that restore activator
response in vitro in conjunction with puriﬁed or recombinant TBP in Drosophila
(Dynlacht, B.D. et al., 1991). These polypeptides were tightly associated with TBP and
could only be dissociated from TBP by mildly denaturing conditions (urea plus salt
washes).

The identiﬁcation of this high-molecular weight TFIID complex precipitated a
series Of attempts to reconstitute activation in vitro using more deﬁned systems composed
of recombinant and puriﬁed, native TFIID. The Berk and Roeder laboratories have
constructed stable HeLa cell lines that express TBP with an N-terminal epitope tag (using
an inﬂuenza virus hemagluttinin (HA) epitope). This epitope is recognized by a
monoclonal antibody (12CA5) that can be used to irnmunopurify the TFIID complex from
nuclear extracts (Zhou, P. et al., 1992). The Tjian and Roeder laboratories have also used
anti-TBP monoclonals to purify TFHD (Dynlacht, B.D. et al., 1991). Isolation of these
complexes has allowed detailed analysis of the role(s) of TAF S in transcriptional activation
as well as analysis of the role of TAFS in earlier steps of PIC formation. TAFS have also
been demonstrated to be essential for activator stimulated formation of a TFIID-TFHA-
DNA complex in vitro (Lieberman, PM. and A]. Berk, 1994). Multiple mechanisms

through which TAFs might effect activation have been suggested. TAFs may function

22

through direct protein-protein interaction with activators, acting as adapters for activation
domains. They may stabilize the PIC through additional supportive protein-DNA
contacts, or may ﬁrnction as sensors of activator presence allowing isomerization Of the
PIC to an activated state (reviewed in Roeder, R.G., 1996). Evidence consistent with all
these possibilities exists.

The earliest model for TAF ﬁrnction came from studies performed primarily in the
Tjian laboratory. The metazoan activator Spl is active only in metazoan organisms.
Presumably, yeast lacks a factor(s) essential for Spl function. The initial lack Of
evidence for TAF polypeptides in yeast (yeast TAFS were eventually identiﬁed) suggested
a potential role for TAFS in this species-speciﬁc coactivation of Spl. Consistent with this
hypothesis, Hoey et a1. identiﬁed a 110 kDa Drosophila TAF that speciﬁcally binds Spl
(Hoey, T. et al., 1993). Since that initial description, several TAF-activator interactions
have been described. VP16 (as well as p53 and RelA) binds dTAFn4O (Goodrich, J .A. et
al., 1993) and the human homologue, hTAFn32 (Klemrn, RB. et al., 1995). These TAFS
dimerize in vitro with dTAFn6O and hTAFn80 (Weinzierl, R.O.J. et al., 1993). This
dirnerization occurs through histone H3 and H4 related domains. The acidic activators
p53 and RelA also bind both TAF S in vitro (Thut, C]. et al., 1995; Lu, H. and Levine,
A.J., 1995; Burley, SK. and KO. Roeder, 1996). The Ile-rich activator NTF-l binds
dTAFn150 and dTAFn150 (Chen, J. et al., 1994). Human TAFn3O associates with the
estrogen receptor (J acq, X. et al., 1994), and the CREB activator has been found to bind
dTAFn110 in vitro (Ferreri, K. et al., 1994).

Cloning of the Drosophila TAF S has allowed the reconstitution of completely

recombinant partial TFHD complexes. Importantly, reconstitution of these partial TFIID

23

complexes with minimal sets of TAFS has allowed identiﬁcation Of TAFS necessary and
suﬁicient for function Of individual activators (Chen, J. et al., 1994). The dTAFu230 and
dTAFn150 polypeptides appear to form the core of the complex, as both are required to
recruit additional TAFS into these recombinant complexes. The addition of TAFnl 10 to
this subcomplex allows activation by Spl, suggesting that Spl and TAFnl 10 interaction is
suﬂicient for activator function in vitro. Subcomplexes assembled with dTAFn230 and
dTAFn150, or with dTAFn230 and dTAFu60, support activation by Gal4-NTF-1 fusion
proteins, but not by Spl, verifying that the dTAFnl 10-Sp1 interaction is also required for
Spl activation in vitro. This result is also consistent with the observation of protein-
protein interactions between the NTF-l activation domain and dTAFn150 and dTAFn60.
Activation by Gal4-VP16C, Gal4-Jun, and Gal4-CTF fusions required the complete
assembly Of the recombinant TFIID complex. These experiments with recombinant TAFS
verify that activated transcription requires the presence of speciﬁc TAFS in vitro.
Furthermore, these required TAFS correspond to those observed to participate in speciﬁc
activator-TAF interactions.

A second set of data suggests that TAFS may be directly involved in formation of
the TFIID-promoter nucleoprotein complex, augmenting TBP-TATA DNA contacts.
Original studies Of TFIID-promoter binding identiﬁed a DNAse protection proﬁle
considerably more extensive than that observed with recombinant TBP (Sawadogo, M.
and R. G. Roeder, 1985; Horikoshi, M. et al., 1988b). TBP efﬁciently protected the
TATA box, but the TFIID footprint extends downstream beyond the initiation site, while
also inducing a number of hypersensitive sites throughout this region (Horikoshi, M. et al.,

1988b). Interestingly, this extensive pattern of protection and hypersensitivity could be

24

stimulated by the activator ATF. This suggests that TFIID TAFS may directly contact
DNA. Consistent with this hypothesis, dTAFn150 has been Observed to produce a similar
pattern of DNAse protection over the initiation site (Verrijzer, C.P. et al., 1994).

A second Observation also suggests that TAFS may be involved in control of the
DNA binding activity of TFIID. Early Observations by Roeder suggested that hTAFn250
inhibited TBP recognition of the TATA box (Kokubo, T. et al., 1993 a). This inhibition is
mediated by a short N-terminal domain of the TAF (Kokubo, T. et al., 1993b). A recent
NMR study identiﬁes how the N-temrinal domain mediates this inhibition. Interestingly,
the domain appears to mimic TATA DNA, producing a protein surface complementary to
the concave DNA-binding surface of TBP (Liu, DJ. et al., 1998). This domain could
effectively act as a competitive inhibitor to DNA binding. Consistent with observed
interactions between VP16 and the DNA-binding concave surface of TBP, VP16 and
hTAFn250 competitively bind the DNA-binding site of TBP (Nishikawa, J. et al., 1997).
This produces a simple model whereby activators might stimulate transcription by
alleviating the effects of a TBP associated repressor that is actually a component Of TFIID
itself.

The third possible role of TAFS in transcriptional regulation is perhaps the most
intriguing. The hypothesis that TAFS may be involved in isomerizations to and from
active complexes was initially suggested by Roeder and colleagues. Since that time,
similar ideas have been advanced by other investigators (Chi, T. and M. Carey, 1996).
Carey and coworkers have described a DNAse protection pattern similar to that observed
by Horikoshi, et al. (Horikoshi, M. et al., 1988a; Horikoshi, M. et al., 1988b). They have

proposed that this DNAse sensitivity pattern reﬂects isomerization of a template

25

associated TFIID complex to a species with enhanced downstream interactions.
Hypothetically, this could reﬂect initiator binding by the human homologue of dTAFnl 50.
Formation of this DNAse protection pattern is enhanced by the Epstein- Barr virus
activator Zta. This isomerization activity requires the presence of TFIIA, and formation
Of the isomerized complex correlates with activation of transcription.

Downstream promoter recognition by TAFS is consistent with recognition Of start
site sequences by dTAFn150, although several other TAFS may have DNA binding
activity. Cloning of the Drosophila TAFS dTAFn40 and dTAFn60, and their human
homologues hTAFn32 and hTAFn70, revealed sequence similarity to histones H3 and H4
respectively (Kokubo, T. et al., 1994; Hisatake, K. et al., 1995; Hoffmann, A. et al.,
1996). FurthermOre, cloning of hTAFn22 revealed sequence similarity to histone HZB
(Choi, B. et al., 1996). Interestingly, crystal structures of these H3 and H4-related TAF
proteins have been Obtained as a complex (Xie, X.L. et al., 1996). The crystal structure
reveals that the two TAFS contain a classical histone-fold core domain, as well as
conserved histone-like TAF-TAF interaction domains. This suggests that these two TAFS
in conjunction with the H2B-related TAF may form a histone octamer-like structure.
Consistent with this possibility, the TAF:TBP stoichiometry of these TAFS within TFIED
has been Observed to be 4: 1, and a histone-like pattern of protein-protein interactions has
been Observed (Hoﬁ‘mann, A. et al., 1996). While a histone-like octameric TAF complex
may form, the DNA-binding capacity of the complex remains an unsolved issue. Studies
by Roeder and colleagues suggest that TFIID recognition of the AdML promoter results
in wrapping of DNA (Oelgeschlager, T. et al., 1996). Conceivably, this wrapping could

be accounted for through association of DNA into a TAF "nucleosome". However,

26

critical arginine residues in both histones H3 and H4 are not conserved in the TAF
polypeptides. These arginine residues (H3 residue R83 and H4 residue R45) are both
involved in critical histone-DNA contacts. If the TAF octamer binds DNA, it must do so
in a fashion distinct from that employed by the histone octamer (see Burley, SK. and R. G.
Roeder, 1996; Luger, K. et al., 1997).

Another line of investigation suggests that TAFS could play a signal transduction
role in transactivation. Work on Drosophila dTAFn230 by Tjian and colleagues has
identiﬁed enzymatic activity. TFIIF is phosphorylated by dTAFn230 (Dikstein, R. et al.,
1996). This activity requires two non-contiguous domains Of the protein, and the protein
exhibits no sequence Similarity to any characterized kinase. Drosophila TAFu23O has also
been demonstrated to acetylate histones (Mizzen, C.A. et al., 1996). Notably, the kinase
activity requires the C-terminal bromodomain, which is not conserved in the homologous
yeast TAF (Dikstein, R. et al., 1996). What role(s) these activities may play in activator
function remains to be elucidated.

Work on yeast TAFS has also provided some Of the most controversial and
interesting results with regard to TAF ﬁrnction in viva. Despite early results suggesting
that yeast lack a TAF-containing TFIID complex, yeast TAFS were eventually identiﬁed
(Poon, D. et al., 1995; Reese, J.C., Apone, L., Walker, S. S., Grifﬁn, L. A., Green, M. R.,
1994). Deletion of several of the yeast TAF genes is lethal (Poon, D. et al., 1995). The
yeast TAFS display a reduced afﬁnity for TBP and purify as a separate complex.
Reconstitution of this yeast complex produces a yeast TFIID complex similar to its
metazoan counterpart. Identiﬁcation and cloning of the yeast TAFS allowed analysis of

TAF function in viva. Walker, et al. isolated a number oftemperature—sensitive alleles of

27

yTAFn145. After a shift to non-permissive temperature, cell growth ceased, but
transcription of a large number of analyzed genes was unchanged (Walker, S.S. et al.,
1996). Similar results were Obtained using galactose-regulated expression of various
TAFS. TAFS were expressed from the GAL] promoter in the presence of galactose. A
shift to glucose-containing media results in repression of this promoter and depletion of
the expressed TAFS. Western blots conﬁrmed reductions in TAF levels, although
extremely low levels of TAF expression below Westem detection levels are difficult to
rule out for in such experiments.

Moqtaderi, et al. describe a similar set of in viva experiments using a copper-
regulated system which selectively degrades speciﬁcally tagged TAF proteins (Moqtaderi,
Z. et al., 1996). Strains containing ubiquitin fusion proteins of various TAFS were
constructed. A copper-inducible promoter was used to induce overexpression of UBRl.
UBRl encodes the N-end recognition protein of the ubiquitin degradation pathway. Thus, .
TAF expression could be down regulated by copper induction through subsequent
degradation of the targeted TAF. Several yeast TAF s were examined, with similar results.
Depletion of TAFS did not signiﬁcantly reduce transcription of several genes examined.
Furthermore, TAFS appear non-essential for transcriptional activation in response tO
several yeast activators. The authors suggest that TAFS may represent only one potential
pathway of activation. While TAF-function might be redundant with multiple pathways in
viva, these supporting pathways are non-functional in vitro. This would result in an
apparent absolute requirement for TAF s for activation of transcription in vitro, with no
apparent requirement in vivo. In contrast, Buratowski and coworkers performed similar

experiments with isolated ts alleles Of the yeast histone—like TAFS (Michel, B. et al., 1998).

28

These TAFS are essential for transcription in viva. Shifting to non-permissive conditions
disrupted production of a wide variety Of transcripts. The reason for this contradiction is
unknown. Perhaps, only the histone-like TAFS are essential. The histone-like TAFS are
also components of the SAGA histone acetyltransferase complex (Grant, RA. et al.,
1998). They may be essential in this role.

Results contradictory to the yeast studies have also been Obtained through genetic
studies in Drosophila (Sauer, F. et al., 1996). Sauer et a1. conducted a genetic screen
designed to identify EMS-induced dominant mutations that alter the expression of a Rasl
reporter construct in the eye. Mutations that reduce or enhance this expression were
mapped to the dTAFn110 and dTAFn60 genes. These mutations also reduced expression
of several other eye-speciﬁc genes, but did not eﬂ‘ect expression of all genes in the eye,
suggesting that TAFS may be involved in recognition or regulation of a subset of genes in
Drosophila. Furthermore, heterozygous ﬂies expressing either a TAFnl 10 partially
deleted allele, or one transdorninant TAFn60 allele displayed reduced expression of two
bicoid responsive gene products. These results are consistent with previous Observations
that these TAFS are required for synergistic function of the two bicoid activation domains
(Sauer, F. et al., 1995).

If TAFS are non-essential for activated transcription in yeast in viva, why is the
phenotype Of TAF null alleles so dramatic? Recall that deletion of TAF genes is lethal. A
potential cause for this phenotype was identiﬁed by Walker, et a1. and Apone et a1.
(Apone, L.M. et al., 1996), although a hint was actually provided by data Obtained almost
a decade earlier (Sekiguchi, T. et al., 1988; Sekiguchi, T. et al., 1991). Apone et a1.

identiﬁed a temperature-sensitive allele of yTAFn90. Temperature-shift results in ngM

29

cell cycle arrest. Walker et a1. Observed a similar arrest with ts yTAFn145 alleles, although
arrest occurred in G]. A ts allele of T SM] (the yeast TAFn150 homologue) also produced
cells arrested in 62 (Walker, 8.8. et al., 1996). These phenotypes are similar to those
described for the hamster cell-line tsl3. This cell-line was isolated as a cell-cycle ts
mutant, incapable of cell cycle progression. The defect was complemented by a human
clone designated CCGI (Sekiguchi, T. et al., 1988). Cloning Of dTAF1123O revealed that a
close human homologue had previously been cloned as CCGI (Ruppert, S. et al., 1993).
Cloning of the human TAF homologue (hTAFn250) veriﬁed its identity to CCGI
(Hisatake, K. et al., 1993).

Overall, these results suggest that TAFS are involved in regulation of only a subset
Of genes, probably involved in cell-cycle regulation in yeast and metazoans, and perhaps in
development and tissue-pattemed expression in metazoans. Although one cannot rule out
the possibility that the cell cycle phenotypes of TAFS are not related to their
transcriptional role in vitro, it is difficult to rationalize a mechanism whereby cell-cycle
regulatory proteins could allow transactivation in vitro. The ability of partially
reconstituted recombinant TFIID complexes to facilitate activator function in vitro is
particularly difﬁcult to rationalize, without proposing a true transcriptional ﬁrnction for
these proteins (Chen, J. et al., 1994).

One potentially interesting explanation for TAF activity involves the enzymatic
acetyltransferase activity associated with the TFIID complex (dTAFn23 0, hTAFn250, and
yTAFn145 are histone acetyltransferases). This is interesting in light of the converse
observation that certain TAFS are also present in other histone acetyltransferase (HAT)

complexes. TAFS have been identiﬁed in both the yeast (Grant, RA. et al., 1998) and

30

human SAGA HAT complexes (Martinez, E. et al., 1998b). The reason for their presence
in these complexes is currently unknown. Perhaps the TAFS serve only a structural role.
Alternatively, they may be tightly associated substrates. Notably, the current deﬁnition of
a HAT is somewhat broad, and the speciﬁcity of these enzymes for histones should be
regarded with caution. Other proteins may also be acetylation targets for these enzymes.
The p53 tumor suppressor is speciﬁcally acetylated by the P/CAF HAT complex in vitro
(Gu, W. and Roeder, R.G., 1997). These same p53 residues are also acetylated in viva
(Sakaguchi, K. et al., 1998).

Combined, these results suggest that TFIID plays a critical role in modulation of
activated transcription. While experiments suggest that TBP recruitment may be limiting
for transcription in viva, TBP mutants that are presumed to affect subsequent steps in PIC
formation also display activation-defective phenotypes. These Observations reinforce that

steps later in PIC formation may also be enhanced by activators.

TFIIB recruitment may stimulate PIC formation
The transcription factor TFIIB enters the PIC following TFIID recognition of the

core promoter. The structure of the protein has been determined both in isolation (Bagby,
S. et al., 1995) and in a TBP-TFIIB-DNA complex (Nikolov, D.B. et al., 1995). The
factor binds directly to TBP and contacts DNA in the promoter complex immediately
upstream and downstream of the TATA box. Consistent with this observation, Reinberg
and coworkers have identiﬁed promoter sequences that contribute to AdML basal
promoter strength, apparently through enhancement of TFIIB-DNA interactions

(Lagrange, T. et al., 1998). TFIIB has also been implicated in start site selection in yeast

31

(Pinto, I. et al., 1992), and can be isolated from crude extracts in complex with RNAPII.
These results have been interpreted as evidence that TFIIB may play a critical role in
recruitment Of RNAPH to promoters (for review see Conaway, RC. and J .W. Conaway,
1993).

A study by Lin and Green provided the ﬁrst suggestion that TFIIB recruitment
may be regulated by transcriptional activators (Lin, Y.S. and MR. Green, 1991).
Formation of a PIC in vitro is a slow process, requiring 15-30 minutes for completion.
The authors attempted to determine the "rate-limiting step" in PIC formation. The
reasoning was that activators were likely to activate transcription by enhancing the rate of
formation of the PIC, by analogy to enzymatic catalysis. The experiments were performed
in a HeLa nuclear extract by ﬁrst preincubating a promoter fragment in the extract in the
presence or absence of activators. Following re-isolation of the promoter fragment, the
collection of basal factors required for continued PIC formation was assessed by
continuation of PIC assembly with deﬁned combinations of partially puriﬁed basal factors.
Slow recruitment of any essential factor (a "rate-limiting" step) would presumably be
reﬂected in the lack of that factor in the partially assembled PIC, and a requirement for
supplementation of that factor in the second incubation. If activators kinetically enhance
assembly of the PIC, then the presence of an activator should alleviate this requirement.

Pre-incubation of a Gal4-responsive Adenovirus E4T promoter template with
nuclear extract resulted in formation of a partial complex that was still sensitive to
activator presence in the second reaction. The second incubation did not require TFIID,
but did require TFIIB, TFIIF/RNAPH, and TFIIE for activation. Omission of the

activator from the ﬁrst incubation did not change this requirement. This suggested that

32

 

TFIID recognition was not rate-limiting (TFIID was not required in the second
incubation), and that activator was not required to recruit TFHD (even PICS assembled
without activator did not require TFIID in the second reaction). The authors also
concluded that PIC formation stalls at the point of incorporation of TFIIB into the
complex. More importantly, the assembly of a partial TFIID-TFIIB complex required
activator. Based upon these results, Lin and Green concluded that TFIIB incorporation
into the PIC was a slow step that required activator participation in vitro, and that
activators enhance transcription through recruitment of TFHB to an assembling PIC. This
idea was strengthened by the Observation that Gal4-VP16 was capable of depleting an
essential component(s) for transcription from nuclear extract. Addition of TFIIB to this
depleted extract restored activity, consistent with their proposal. A similar study
performed later by Choy and Green, employed Western blots to analyze partially
assembled PIC complexes in the presence or absence of activator to conﬁrm these earlier
results (Choy, B. and Green, M.R., 1994). The authors veriﬁed recruitment of TFIIB by
activator (Gal4-AH) to a promoter in vitro. In addition, the authors identiﬁed a second
slow step after TFIIB association that could be enhanced in vitro by Gal4-AH, consistent
with several proposals that activators facilitate multiple steps in PIC formation (Ptashne,
M., 1988; Ptashne, M. and Gann, A., 1997; He, S. and Weintraub, S.J., 1998; Blair, W.S.
et al., 1996; Brown, S.A. et al., 17; Chi, T.H. et al., 1995).

Direct TFIIB-activator interactions have also been demonstrated using several
transcriptional activators. As mentioned above, a VP16 protein afﬁnity column depletes a
HeLa nuclear extract of an essential factor(s). Western blotting of material eluted from

this aﬁinity column conﬁrms the presence of TFIIB in the bound ﬁaction. This interaction

33

appears to be direct, as recombinant TFIIB also binds VP16. Furthermore, VP16 mutants
which reduce transcriptional activity in vitro and in viva (FP442 and truncation of the C
subdomain) abolish this interaction (Lin, Y.S. et al., 1991). Experiments performed by
O'Hare and coworkers conﬂict with these conclusions. Mutations within VP16C, which
signiﬁcantly impair VP16 activity, do not appear to affect VP16 binding to TFIIB in
Similar experiments with HeLa nuclear extract (Walker, S. et al., 1993). However,
additional factors present in the extract may stabilize TFIIB-VP16 contacts. Experiments
assessing VP16 mutants performed by Lin et al. utilized recombinant or puriﬁed TFIIB.
Signiﬁcantly, TFIIB mutants defective for the observed TFIIB-VP16 interaction are also
defective for activated transcription in vitro (Roberts, S.G.E. et al., 1993). However,
Goodrich et al. reported that Drosophila TFIIB failed to interact speciﬁcally with VP16 in
vitro. Mutations which reduce binding of mammalian TFIIB, do not effect the binding of
Drosophila TFIIB (Goodrich, J .A. et al., 1993). This Observation is inconsistent with the
robust activity of VP16 in the Drosophila system, suggesting that TFIIB interaction is not
essential for VP16 ﬁrnction. Consistent with the observations by Lin et al., a number of
other transcriptional activators also bind TFIIB in vitro. These include the Drosophila
fushi-tarazu activator (Colgan, J. et al., 1993), the yeast activator Gal4 (Wu, Y. et al.,
1996), the NF-xB family member RelA (Paal, K. et al., 1997), and several members of the
steroid receptor superfamily (COUP-TF, Progesterone (PR) and Estrogen (ER)
Receptors) (Ing, N.H. et al., 1992). In this last study, TFHB binding was found to be
ligand independent for the PR and ER, suggesting that TFIIB interaction is not sufﬁcient

to explain transactivation by the liganded receptor.

34

 

Interestingly, a very similar experiment to that performed by Lin and Green was
performed by John White in Pierre Chambon's laboratory, with a noticeably diﬁ‘erent
conclusion (White, J. et al., 1992). Utilizing a HeLa nuclear extract and Gal4-VP16
activator, the authors identiﬁed formation Of a partial PIC that was activation domain
dependent. They employed an antibody that recognizes the VP16 activation domain.
They reasoned that if activators accelerate formation of a critical intermediate in PIC
formation, then following the formation of this critical intermediate, the activation domain
should be dispensable for further PIC formation. Thus, masking the activation domain by
an antibody at this point should have no effect on further PIC formation. Such an
intermediate was identiﬁed. The kinetics of formation of this intermediate are consistent
with those previorisly observed for formation of a template-committed complex, an event
attributed to TFIID. Using partially puriﬁed basal transcription factor fractions, the
authors were able to reconstitute a seemingly identical complex (as judged by antibody
resistance) using TFIIA and TFIID. TFIIB had no effect on the formation Of this
complex. Furthermore, TFIID and TFIIB could not form this antibody resistant complex.
The authors conclude that VP16 acts through a TFIID-TFHA—DNA complex, independent
of TFIIB.

Another Observation is inconsistent with the conclusions of Lin and Green. Struhl
and colleagues have analyzed the effect of a number of TBP mutants on transcription in
yeast (Lee, M. and K. Struhl, 1997). These mutants were originally described by Ebright
and Reinberg (Tang, H. et al., 1996). Lee and Struhl employed corresponding mutants of
yeast TBP to study the importance of these in vitro interactions to transcriptional

activation in yeast. Several yeast TBP mutants defective for TFIIB complex formation in

35

vitro, supported activated transcription by several activators in viva. The authors argue
that this result makes it unlikely that TFIIB recruitment to a TBP-DNA complex is

limiting for transcription in viva.

The Basal Transcription Factor TFIIA is reguired for activator response

The transcription factor TFHA has had a controversial and enigmatic history.
Shortly after TFIIA was identiﬁed as a factor essential for speciﬁc initiation in vitro,
experiments using partially puriﬁed systems suggested that the factor was not required for
basal transcription and played only a minor stimulatory role in transcription. This
suggestion seemed at odds with in viva studies in Saccharomyces cerevisiae. Yeast
TFIIA comprises two polypeptides encoded by two genes, T 0A1 and T 0A2. Deletion of
either T 0.41 or T 0A2 is lethal (Ranish, J.A. et al., 1992). This severe phenotype suggests
that TFIIA plays a criticalrole in transcription in viva.

Simultaneous eﬂ‘orts by three laboratories shed light on the molecular basis for
these inconsistencies. The initial work utilized either partially puriﬁed or recombinant
basal transcription factors. Curiously, TFIIA stimulated transcription when partially
puriﬁed TFIID was used, but not when TBP replaced these TFIID fractions (Ma, D. et al.,
1993; Yokomori, K. et al., 1994; Sun, X. et al., 1994). The same result was also Observed
with epitope-puriﬁed TFIID. These Observations imply that the stimulatory effect of
TFIIA requires TAFS (Zhou, Q. et al., 1993).

An additional complexity is suggested by recent work from the Smale and Roeder
laboratories. Transcription of a core promoter lacking a TATA box, with a ﬁrnctional

terminal deoxynucleotidyltransferase (TdT) Inr cannot be reconstituted in vitro with the

36

current ensemble of puriﬁed basal factors (TFHA, TFIIB, TFIID, TFIIE, TFIIF, and
TFIIH). Smale and coworkers identiﬁed a column fraction activity which when
supplemented with these factors reconstituted Inr promoter activity. They attributed this
activity to the human homologue of dTAFn150, a TAF implicated by the Roeder and Tjian
laboratories as being involved in downstream promoter recognition by TFIID (Kaufrnann,
J. et al., 1996). Signiﬁcantly, this TAF appears to be missing from hTFIID. Martinez, et
al. have also succeeded in reconstituting a cell-free system capable of transcription from a
TdT Inr template in vitro (Martinez, E. et al., 1998a).

In contrast to the results of Smale and coworkers, Martinez, et al. found that
hTAFn150 stabily associates with TFIID. The previous failure to identify the TAF in
previous preparatiOn resulted from the fact that the TAF migrates aberrantly as a 135 kDa
polypeptide in SDS PAGE. This results in comigration with hTAFn135, a previously
identiﬁed homologue of dTAFnl 10. This hTFIID preparation, despite containing
hTAF315O, did not reconstitute Inr basal activity. Further fractionation of the TFIID-
containing crude fraction identiﬁed three non-TAF components (which they term TIC-
1/2/3) necessary to ﬁrlly reconstitute Inr activity. Thus, Inr function requires accessory
factors, in addition to the set of basal factors sufficient for TATA containing promoters,
although the precise identity of these factors is unknown.

This observation suggests a potential explanation for the variable requirement for
TFIIA function in previous studies. The AdML promoter is a composite TATA + Inr
promoter. When the function of the AdML promoter is analyzed, only partial activity is
observed with puriﬁed hTFIID, since the AdML promoter functions only through the

TATA box in the absence of TIC and TFHA. Addition of these factors (by use Of cruder

37

fractions) allows the Inr element to function synergistically with the TATA element. Thus,
studies employing TFHA-depleted nuclear extracts or crudely puriﬁed TFIID would
display high levels of TFHA dependent activity due to TATA and Inr synergy. Further
puriﬁcation of TFIID eliminates TIC and results in increasingly reduced TFHA
dependence. Yokomori et a1. provide an alternative explanation Of the variable
requirement for TFHA (Yokomori, K. et al., 1993). The authors isolated TFIIA as a
component Of Drosophila TFIID. Apparently, TFHA associates with TBP with relatively
high afﬁnity even in the absence of DNA. The TFIID-associated TFHA could be stripped
from TBP by a moderate salt wash (3 00 mM KCl). Thus, variability in the ionic
conditions used for TFIID preparation could signiﬁcantly alter TFHA requirement.
Regardless of the potential role of TFIIA in basal transcription, TFIIA is clearly
required for activator response in vitro (Ma, D. et al., 1993; Yokomori, K. et al., 1994;
Sun, X. et al., 1994; Zhou, O. et al., 1993; Ozer, J. et al., 1994). Reconstitution of
activated transcription can be Observed with model activators (either Gal4-AH or Gal4-
VP16) and puriﬁed basal factors only in the presence TAFS and TFHA. In reconstituted
TFIID systems, TAF-dependent activation was determined to be completely dependent
upon the addition of TFIIA to transcription reactions. Chi et al. argue that this eﬁ‘ect of
TFHA is the result of prevention Of the inhibitory effect of TAFS on TFIID DNA
recognition (Chi, T.H. et al., 1995). Consistent with this possibility, Ozer et a1. identiﬁed
TFIIA mutants incapable of activating transcription. These mutants were still capable of
formation of a TBP-TFllA-DNA complex, but incapable of associating with TFHD and
DNA (Ozer, J. et al., 1996). This effect could be reproduced with hTAFu250 alone. The

authors argue that TFHA, in an activator dependent fashion, competitively displaces

38

hTAFn250, stimulating TFIID recognition of the TATA box on targeted promoters (Ozer,
J. et al., 1998)

Chiang and coworkers have described a TFHA-dependent, TAF-independent
activator response to Gal4-VP16 in vitro (Wu, S. et al., 1998). Interestingly, this
activator response does not require holoenzyme. The Kornberg and Young laboratories
describe TAF-independent activation in vitro using preparations of holoenzyme (Koleske,
A.J. and RA. Young, 1994). However, Chiang and coworkers (Wu, S. et al., 1998)
succeeded in generating an activator-responsive in vitro system using extensively puriﬁed
RNAPII (RNAP II was epitope puriﬁed and washed with 0.85M KCl and 1M urea),
TFIIA and PC4 (a ssDNA binding protein implicated as a coactivator; see below). This
suggests that TFHA may ﬁrnction in a TAF-independent capacity as well.

A role for TFHA in response to activators has also been indicated by other results
utilizing in vitro transcription systems. TFIIA has been demonstrated to enhance the
affinity of TBP for DNA in certain EMSA systems (Maldonado, E. et al., 1990). Berk
and coworkers (Lieberman, PM. and A.J. Berk, 1994) developed an agarose EMSA
system for analysis of complex formation using epitope puriﬁed TFIID. In this system,
TFIID-DNA binding is strongly dependent upon TFHA, and can be stimulated by
activators such as the Epstein-Barr virus activator Zta or VP16 (Kobayashi, N. et al.,
1995). The hypothesis is that TFIIA enhances the stability of this TFIID-TFIIA—DNA
complex (DA complex) to electrophoretic conditions. The enhancement of this DA
complex by Zta is activation domain dependent, and appears to depend upon the ability of
Zta to recruit TFIIA. Zta mutants that do not bind TFIIA, do not enhance formation of

the DA complex (Ozer, J. et al., 1996). The HSV activator VP16 can also enhance DA

39

complex formation. This activity has been localized to the VP16C subdomain, and VP16C
has been shown to bind TFIIA (Kobayashi, N. et al., 1995; Kobayashi, N. et al., 1998).
An interesting hypothesis related to this Observation has been proposed by Carey
and coworkers (Chi, T. and M. Carey, 1996). They have proposed that TFIIA aﬂ‘ects
activator response by inducing an isomerization of the DA complex. This isomerized
complex was suggested by the appearance of DNAse protection as well as hypersensitivity
downstream of the TATA box extending through the initiation Site (Sawadogo, M. and
RC. Roeder, 1985). This pattern was originally observed by Roeder and coworkers when
footprinting with crude preparations of TFIID (see above). The appearance of these sites
was attributed to DNA binding by TFIID and could be enhanced by the transactivator
ATF (Horikoshi, M. et al., 1988b). Similarly, TFHA and Zta enhance the appearance of
the downstream pattern of DNAse sensitivity seen by Chi et a1. It has been proposed, by
Roeder originally and later by Carey, that this pattern of hypersensitivity reﬂects an
isomerization of TFHD to a more extensive TFHD-DNA nucleoprotein complex.
Accumulation of this isomerized complex correlated with transactivation, although an
inability to directly trap this intermediate complex precludes any direct observation of its
activity relative to unisomerized PIC. While only Zta has been reported to initiate this
isomerization, the authors note that VP16 also induces DA isomerization, consistent with
the observed activity of VP16 in DA complex formation assays (Kobayashi, N. et al.,
1998). While the exact relationship between this isomerized DA complex and the
electrophoretically stable DA complex originally identiﬁed by Lieberman and Berk is
unknown, these parallels are interesting. Perhaps the stable DA complex is in fact an

isomerized DA complex.

40

Activation may also involve non-basal auxiliaﬂ factors

While the majority Of the early investigations regarding activator function involved
studies of activator interactions with basal transcription factors, a variety of genetic and
biochemical studies have focussed recent attention on an expanding group of auxiliary
cofactors. These coactivators have been identiﬁed in both yeast and mammalian systems.
One of the earliest observations of coactivator complexes came from genetic studies in
yeast. Berger et al. utilized a genetic selection based upon reversal Of the slow growth
phenotype of yeast overexpressing a chimeric Gal4-VP16 activator (Berger, S.L. et al.,
1992). Mutations {within the activation domain alleviated the phenotype, suggesting that it
was transcriptionally linked. The selection has resulted in the identiﬁcation of several
yeast genes proposed to serve as transcriptional adapters: ADA 1, ADA2, ADA3, ADA4
(GCN5), and ADA5 (SPT 20) (Pina, B. et al., 1993; Marcus, G.A. et al., 1994; Marcus,
G.A. et al., 1996). Mutations of any of the ADA genes alleviate the slow growth
phenotype. The gene proteins appear to form a multi-protein complex, as protein-protein
interactions between Ada2p and Ada3p have been reported (Candau, R. and Berger, S.L.,
1996; Horiuchi, J. et al., 1995). They are required for efﬁcient function of several
activators (i.e. VP16, Gal4, Gcn4) and their deletion has no effect upon basal general
transcription. This suggests that the proteins ﬁrnction as bridging factors, or adapters,
between activators and the basal transcription machinery. Complexes of the Ada proteins
and Spt proteins have also been identiﬁed in vitro (Grant, RA. et al., 1997).

An additional hint as to the ﬁrnction of these proteins arose from a line of

biochemical investigations in T etrahymena. This multinucleate organism has served as a

41

model for the regulation of gene expression through nuclear structure. The macronucleus
is transcriptionally active, while the micronucleus is maintained in a transcriptionally silent
state. Biochemical studies aimed at comparison of the structural packaging of
micronuclear and macronuclear DNA implicated post-translational modiﬁcation of
histones in this suppression of gene expression. Macronuclear histones are extensively
acetylated. Correlations between histone acetylation and transcriptionally active loci have
been Observed in other species as well (Csordas, A., 1990; Wolffe, A., 1994; Turner,
B.M., 1991; Loidl, P., 1994).

This link between acetylation and the transcriptional competence of chromatin
precipitated a search for the cellular machinery responsible for this post-translational
modiﬁcation. Biochemical studies of histone acetyltransferases (HATS) in T etrahymena
identiﬁed both cytoplasmic (type B) and nuclear (type A) enzymes. Allis and coworkers
puriﬁed the HAT A enzyme and cloned the HAT A gene. Comparison of the predicted
gene product with the GenBank database identiﬁed similarity to the yeast GCN5 gene
product (Brownell, J.E. et al., 1996). This observation provided the ﬁrst concrete
evidence that histone acetyltransferases function as transcriptional regulatory components.

Genetic studies of Ty transcription in yeast have also produced additional evidence
linking histone acetylation to gene regulation. The SPT genes were identiﬁed as
suppressors of a TyzHIS3 integrant of Saccharamyces cerevisiae. The Spt mutants favor
usage of the natural HIS3 promoter, producing stable HIS3 mRNA and protein. These
SPT genes have been divided into two major subgroups: an TBP-associated group and a
histone associated group. The ADA and SPT studies converged with the identiﬁcation of

ADA5 as SPT 20 (Marcus, G.A. et al., 1996; Roberts, SM. and Winston, F., 1996).

42

Biochemical analysis identiﬁed a high molecular weight complex of ADA and SPT gene
products (Grant, RA. et al., 1997). This acetyltransferase complex (the SAGA complex)
ﬁrnctions as a histone acetyltransferase in vitro, acetylating a mononucleosome containing
a Gal4 binding site in a Gal4-VP16N dependent fashion (Utley, R.T. et al., 1998).
Mammalian studies have also provided another link between histone acetylation
and the ﬁrnction of an expanding number of transcriptional activators. The CBP/p3 00
family of coactivators was initially identiﬁed as a collection of related proteins capable of
binding and enhancing activation by the mammalian CREB (Kwolg R.P.S. et al., 1994)
and Adenovirus ElA activators (Arany, Z. et al., 1994). The family has since been
implicated in the function of an increasing number Of activators (reviewed in Ogryzko,
V.V. et al., 1996) including c-Jun (Arias, J. et al., 1994), c-Myb (Dai, P. et al., 1996), c-
Fos (Bannister, A.J. et al., 1995), MyoD (Yuan, W. et al., 1996), and the nuclear hormone
receptors (Kamei, Y. et al., 1996). In vitro studies identiﬁed CBP as a histone
acetyltransferase (Ogryzko, V.V. et al., 1996). Roeder and colleagues also reported that
CBP acetylates p53 (611, W. and KO. Roeder, 1997). This acetylation has also been
observed in viva (Sakaguchi, K. et al., 1998). Studies with the CBP/p300 proteins
identiﬁed an additional protein (P/CAF; p3 OO/QBP Associated F actor) which is thought to
mediate p300 function. P300/CBP proteins can be found as part of a stable complex with
P/CAF. Cloning and sequencing of the P/CAF gene identiﬁed extensive sequence
homology to yeast GCN5 (Yang, X.-J. et al., 1996). A human GCN5 homologue has also
been identiﬁed,-suggesting that metazoans may have evolved enhanced regulatory capacity
through duplication and specialization of these enzymes (Inoue, M. et al., 1996). The

proteins are homologous throughout their C-terrninal HAT domains, but diverge in their

43

N—terminal sequences. While these studies provide evidence for multiple HAT enzymes in
higher metazoans, it should be emphasized that demonstration of HAT activity does not
necessarily imply in viva relevance for the activity. Additional acetylation targets are
likely to exist (for instance, the p53 target of CBP). Acetylation of TFIIB and TFIIF has
also been reported (Imhof, A. et al., 1997). In the end, acetylation may prove to be a
general nuclear signal transduction pathway.

The family of coactivators and adapters is not limited to acetyltransferase
components. Tjian and coworkers obtained evidence of a requirement for additional
factors for activated transcription in vitro (Dynlacht, B.D. et al., 1991). This observation
led to the identiﬁcation of TAFS (see above) as non-basal factors required for activated
transcription in vitro. Recent genetic experiments in yeast have conﬁrmed that at least a
subset of these TAFS is required for activator function in viva (Michel, B. et al., 1998).
Biochemical studies in yeast have also identiﬁed a set of proteins believed to facilitate
transcriptional activation. A complex of these proteins was isolated biochemically from
Saccharamyces cerevisiae. This mediator complex is necessary and sufﬁcient to allow
activator response in vitro (Kelleher III, R]. et al., 1990). Cloning of the mediator genes
identiﬁed overlap between these genes and components of the yeast holoenzyme.

Another class of coactivators has been identiﬁed through biochemical studies in
mammalian systems. Roeder and colleagues have dissected the transcriptional
components biochemically through fractionation and reconstitution of transcription
reactions in vitro. This process led to the identiﬁcation of an Upstream Stimulatory
Activity (USA) that enhances activator function. This USA, which is a relatively complex

column fraction, contains several factors that differentially modulate transcription (in some

44

cases basal transcription, in other cases only activated transcription). Two positive control
(PC) factors have been characterized, PC] and PC4. The best characterized is PC4. Ge
and Roeder identiﬁed PC4 as a single-stranded DNA binding protein, which enhances
transcription in cooperation with activators (Ge, H. and Roeder, R. G., 1994). The DNA
binding activity is essential for PC4 function (Kretzschmar, M. et al., 1994).

The USA fraction characterization has also led to identiﬁcation of two negative
regulatory factors that may contribute to activator function in vitro. The Dr2 protein was
identiﬁed as Topoisomerase I (Merino, A. et al., 1993). The better-characterized
repressor, Drl, functions as a heterodimer with DRAP]. The protein complex binds TBP
preventing TATA binding and represses basal transcription (Inostroza, J .A. et al., 1992;
Yeung, K.C., et a1, 1994). Both of these repressors probably augment the function of
activators by repressing basal transcription in vitro. The basal factor TFIIA appears to be
able to overcome this repression in vitro (Roeder, R.G., 1991).

A third TBP-directed repressor, MOT 1 was identiﬁed through genetic studies in
yeast. The MOT 1 gene product removes TBP from DNA in vitro, in an ATP-dependent
reaction. Overexpression of TFHA suppresses a mat] mutation, suggesting that MOT]
may function in opposition to TFHA in viva (Auble, D.T., et al., 1994). These repressors
may cooperate to diminish transcription in the absence of a stimulatory signal. The ability
of activators to derepress this inhibited basal state may be an essential component of

activation in viva.

Other complexes are also potential coactivator candidates

45

Transcription is also repressed in vitro and presumably in viva by chromatin
structure (reviewed in Workman, IL. and Kingston, RE, 1998). DNA is packaged in
viva into nucleosomes, each consisting of 146 bp of DNA associated with an octameric
complex of four histones. The structure of this core histone particle associated with a 5S
rDNA sequence has been solved (Luger, K., Maeder, A.W., Richmond, R.K., Sargent,
DE, and Richmond, T.J., 1997). The crystal structure reveals extensive, periodic minor
groove contacts between DNA and histones. This protein-DNA interaction obscures one
face of the DNA helix as it is wrapped about the histone core. This nucleosomal
packaging inhibits DNA binding proteins, signiﬁcantly reducing afﬁnity for their DNA
binding sites. This inhibition of DNA binding likely extends to the PIC and its
components. TBP binding is inhibited by nucleosome formation (Imbalzano, A.N. et al.,
1994a), and pre-binding of TFIID partially alleviates the inhibitory eﬂ‘ects of nucleosomes
in vitro (Felsenfeld, G., 1992).

Counteracting these inhibitory eﬁ‘ects of nucleosome structure must be accounted
for in any viable model for transcriptional regulation. Histone tail acetylation may provide
one means to modify chromatin structure. The histone tails directly bind DNA and are
likely to stabilize histone-DNA contacts signiﬁcantly (Luger, K., et al., 1997), as well as
inﬂuence higher order chromatin ﬁber structure. In vitro, under proper ionic strength
conditions, arrays spontaneously fold into compact ﬁbers (Fletcher, TM. and Hansen,

J .C., 1995). This process is impaired in trypsinized nucleosomal arrays lacking histone
tails (Logic, C.L. et al., 1998). Similar, but less dramatic effects are also seen when arrays
are constructed with hyperacetylated histones. However, accessibility of DNA binding

proteins to mononucleosomes can also be enhanced by acetylation, suggesting that higher

46

 

order structure alone is insufﬁcient to ﬁrlly explain the repressive effects of histones (Lee,
D.Y. et al., 1993).

While the direct effects of acetylation may be sufficient to allow PIC formation on
nucleosomal DNA, a class of nucleosomal remodeling factors may also enhance the
accessibility of nucleosomal DNA sequences. These factors were initially identiﬁed
through genetic studies in yeast. The Herskowitz laboratory described a set of S W] genes
essential for mating type switching in yeast (Peterson, CL. and Herskowitz, I., 1992;
Stem, M., et al., 1984). Simultaneously, the Winston laboratory characterized a set of
yeast genes (SNF genes) essential for growth on sucrose (Winston, F. and Carlson, M.,
1992). The products of these genes are found in a single high molecular weight complex
in cell-free lysates (Peterson, C.L. et al., 1994; Cairns, B.R. et al., 1994). The SWI/SNF
complex alters the DNAse footprint of mononucleosomes, enhancing DNAse cleavage of
the histone bound face of the DNA helix in an ATP-dependent fashion. This activity is
also manifest by increased accessibility Of nucleosomal DNA sequences to DNA-binding
proteins (Cote, J. et al., 1994) and restriction enzymes (Logic, C. and Peterson, CL,
1997). An expanding number of SWI/SNF related ATP-dependent remodeling complexes
have also been identiﬁed (Kingston, RB. et al., 1996). How these complexes are targeted
to DNA loci remains unknown, however activators could conceivably play a direct or

indirect role in this recruitment.

Potential models for the function of activators

How do activators function? The previous sections were intended to provide a

survey of the components of the basal transcriptional machinery and regulatory

47

 

components, as well as evidence for their participation in transcriptional activation. Most
models that have been advanced to explain activator function involve activator directed
protein-protein interactions that stabilize or modify the basal apparatus. Activators have
been proposed to stimulate the kinetics of PIC formation by recruitment of TBP, TFHA,
TFIIB, or the holoenzyme itself Alternatively, activators may stabilize the PIC or induce
its isomerization to an activated state once it has formed. This isomerization could be
induced by conformational changes induced by complex formation between activators and
their targets. The composite nature of eukaryotic promoters provides an additional
complexity. Individual promoters may require stimulation of different steps of PIC
formation. TATA-less promoters may be more dependent upOn activators that stimulate
TFIID recruitment and evidence suggests that these promoters are TAF dependent.

Activators are also likely to function prior to PIC assembly by facilitating access to
DNA binding sites (either promoter elements or activator sites) obscured by the packaging
Of DNA into nucleosomes. Recruitment Of histone acetyltransferase (HAT) complexes or
chromatin remodeling complexes such as the yeast SWI/SNF complex may be involved in
this process. Other derepression models have also been proposed. Activators may
antagonize the repressive effects of repressors such as Dr] or Dr2.

The discussion above focuses primarily upon activation through stimulation of PIC
formation. However, activators are also likely to effect other steps downstream of PIC
assembly. Activators such as Tat, SRF, and VP16 can function through stimulation of
elongation (Yankulov, K. et al., 1994; Blau, J. et al., 1996; Spencer, CA. and Groudine,
M., 1990). SRF has been proposed to recruit TFIIF to promoters (Zhu, H. et al., 1994).

How Tat or VP16 stimulate elongation remains unknown. Other factors (TFHS, elongin,

48

etc.) also effect either the rate of RNAPII catalysis or the capacity of the enzyme to
progress through pause sites, and loading of these factors could theoretically enhance
transcription through RNAPII pause or arrest sites. These activators could also stimulate
elongation through stimulation of CTD phosphorylation. Elongation may also be
enhanced by recruitment Of TFIIH. The VP16 activation domain binds TFIIH in vitro
(Xiao, H. et al., 1994). The TFIIH activity comprises a complex that contains both a
helicase and CTD kinase activity (Drapkin, R. and Reinberg, D., 1994). Phosphorylation
of the CTD has been proposed to facilitate elongation. This capacity of activators to
stimulate elongation could introduce an additional level of promoter-speciﬁcity. Genes
with promoter proximal pause sites such as the c-myc (Krumm, A. et al., 1992) or hsp70
(O'Brien, T. and Lis, J .T., 1991) promoters are likely more dependent upon activators that
stimulate elongation.

The mechanism of activator function has been an extensively investigated topic.
The majority of the collected evidence suggests that activators work through multiple
mechanisms, stimulating multiple steps of transcription upstream and downstream of PIC
complex formation. Furthermore, different activators are distinct in their capacity to
stimulate these different Steps. Not all activators are capable of stimulating integrated
templates, suggesting that some may be incapable of antagonizing the repressive effects of
chromatin (reviewed in Smith, CL. and G.L. Hager, 1997). Activators also differ in their
capacity to interact with certain coactivator components. Finally, activators are
diﬂ‘erentially capable of stimulating elongation of the RNAPH complex (Yankulov, K. et

al., 1994; Blau, J. et al., 1996).

49

 

 

We are interested in the primary structural elements that deﬁne activator function.
We have utilized two model activators, VP16 and the mammalian activator RelA to study
the structural requirements for activator function. Each activation domain is composed of
multiple, independent subdomains, composed of short segments of hydrophobic amino
acids. Each subdomain displays relatively similar structural requirements, with only subtle
differences observed. Despite this similarity, evidence suggests that these subdomains do

function through distinct mechanisms in vitro and in viva.

50

Chapter Two

Activation Domains Contain Similar Yet Distinct Hydrophobic Motifs

Introduction

Herpes simplex virus (HSV) serves as a useful model for the study of the
regulation of gene expression. Lytic infection involves coordinated expression of over 70
viral gene products, requiring intricate temporal control of three sets of viral genes. These
genes have been classiﬁed according to their time of induction into three major classes:
immediate-early, delayed-early, and late genes (Hayward, GS, 1993). Immediate-early
genes, activated shortly after infection, encode viral regulatory proteins that control
subsequent expression of delayed-early and late genes. Induction of the immediate-early
genes is triggered in part by VP16. The protein is biﬁmctional, serving both this
regulatory role, as well as a structural role as a tegument component of the virion. The
gene is essential (Weinheimer, S.P. et al., 1992), although an HSV strain bearing a VP16
mutant lacking the activation domain (an amino acid 410-490 deletion) is infective in
tissue culture systems (Rath Pichyangkura, unpublished results).

The VP16 protein contains 490 amino acids comprising two domains. Deletion
analysis of the C-terminus of VP16 has identiﬁed an activation domain within the C-
terminal 90 amino acids (amino acids 413-490). This C-terminal domain is both necessary
for VP16 activation (Triezenberg, S.J. et al., 1988) as well as sufficient to confer activator
function to a heterologous DNA binding domain. Fusion of the C-terminal activation
domain (amino acids 410-490) to the DNA binding domain of the yeast activator GAL4

(Gal4-VP16) produces a potent activator of transcription both in viva (Sadowski, I. et al.,

51

1988) and in vitro in mammals (Carey, M. et al., 1990) and yeast (Chasman, D.I. et al.,
1989). The N-terminal 410 amino acid domain is suﬂicient to recognize VP16 responsive
cis-elements in the viral irnmediate-early enhancers. These elements are hybrid Oct-1
binding sites recognized by a trimeric complex consisting of Oct-1, a host cell factor
(HCF), and VP16 (Preston, C.M. et al., 1988; Gerster, T. and Roeder, R.G., 1988). The
N-terminal domain is also sufﬁcient for tegument assembly (Rath Pichyangkura,
unpublished results).

The activation domain of VP16 contains an unusually high content of acidic amino
acids. This observation has led to its classiﬁcation as an "acidic" activator, a class for
which it has served as prototype. Deletion of the C-terminal 40 amino acids results in
approximately 50 per cent residual activity, while further deletion of an additional 40
residues results in full loss of activity in viva (Triezenberg, 5.]. et al., 1988), suggesting
the existence of two functional motifs within the intact activation domain. These two
"subdomains" will be referred to as VP16N (amino acids 410-456) and VP16C (amino
acids 456-490). Both of these domains are capable of activating transcription
independently in viva (Emami, K.H. and Carey, M., 1992; Seipel, K. et al., 1992).

Initial mutational studies on the VP16 activation domain were performed on the
VP16N subdomain. Cress and Triezenberg identiﬁed a critical hydrophobic residue, F442
that was essential for activation by VP16 (Cress, W.D. and SJ. Triezenberg, 1991). A
rough correlation between acidity and the transcriptional activity of the domain was also
suggested by the deleterious eﬂ‘ect of asparagine substitutions for aspartic acid residues.
However, this effect required simultaneous alteration of multiple residues; no single acidic

residue was required for function. Apparently, only overall acidity of the domain is

52

 

important for function. Regier et al. demonstrated the importance of two additional
hydrophobic residues within the VP16N subdomain (Regier, J .L. et al., 1993). These two
residues, L439 and L444, ﬂank the critical F442 residue and partially contribute to VP16N
function. Neither appears as critical for ﬁrnction as F 442. Attempts to identify critical
residues within the VP16C subdomain demonstrated weak eﬁ‘ects of alteration of two
phenylalanine residues (F473 and F475) within the subdomain, but interpretation of these
results was complicated by the presence of an impaired VP16N subdomain in these
constructs.

We are interested in precisely deﬁning the essential primary structural requirements
for activation domain function. This topic has been extensively studied. Studies of VP16
(Cress, W.D. and SJ. Triezenberg, 1991; Regier, J .L. et al., 1993; Walker, S. et al.,
1993), RelA (Blair, W.S. et al., 1994b; Schmitz, ML. et al., 1994), Spl (Gill, G. et al.,
1994), p53 (Lin, J. et al., 1994), as well as the yeast activators Gcn4 (Drysdale, C.M. et
al., 1995; Jackson, B.M. et al., 1996) and Gal4 (Leuther, K.K. et al., 1993), have each
identiﬁed common structural features. These activators appear to require multiple
hydrophobic residues for activation domain function. Furthermore, in many of these
cases, activators appear to contain multiple activation motifs (subdomains) capable Of
independent function (e. g. VP16, RelA, p53, GCN4). These studies imply that activation
domains are constructed ﬁom repeated motifs, each functionally dependent upon their
hydrophobic character.

Work by Schaﬂher and colleagues on VP16N illustrates that these functional
motifs may actually be surprisingly small. Seipel et al. found that a synthetic Gal4-fusion

composed of a direct repeat of an 11 amino acid motif based upon the VP16N subdomain

53

 

sequence (surrounding the critical F442 residue) is suﬁicient to direct activated
transcription (Seipel, K. et al., 1994). This result implies that this 11 amino acid motif
contains all sequence information necessary to confer function. Baeuerle and coworkers
have also reconstituted activator function using a Gal4-fusion of a similarly compact
segment of RelA (Schmitz, ML. et al., 1994). Collectively, these results suggest that
activation domains are composite in nature, containing multiple compact hydrophobic
motifs.

Despite these similarities, examination of activation domains has failed to identify
any consensus motif characteristic of these domains. The only apparent similarity revealed
is a common requirement for the overall amino acid composition of these motifs. This has
led to the proposal that the critical structural features for activation domain function are
somewhat ﬂexible. - Perhaps any DNA-binding domain associated, surface-exposed
sequence rich in hydrophobic residues is sufﬁcient for activator function (Ruden, D.M. et
al., 1991; Sigler, P., 1988). This is poignantly demonstrated by the observation that
insertion of an alanine residue into the center of the C—terminal activation domain of RelA
does not signiﬁcantly disrupt the strength of this activation domain (Schmitz, ML. et al.,
1994). Alternatively, the apparent lack of conservation between different activation
domains may be the result of comparison of unrelated and ﬁrnctionally distinct activation
domains. A lack of structural conservation between functionally unrelated activation
domains would not be surprising.

To examine these structural questions, we utilized a strategy of mutational analysis
to investigate the primary structural requirements for activation domain ﬁrnction, using

two well—characterized activators, VP16 and RelA. Both are members of the acidic class

54

 

of transcriptional activators. Our results indicate that VP16 and RelA share a similar
requirement for hydrophobic and aromatic amino acids for function. Furthermore, both
activation domains are composite in nature. Despite these similarities, mutational analysis
suggests subtle differences in the structural requirements of the VP16 and RelA
subdomains. Together, these results suggest that activation domains are reliant upon

similar, yet individually unique, core elements for function.

Experimental Results

The activation domains of VP16 are dependent upon similar hydrophobic motifs

T o characterize the essential primary structural features of the VP16C subdomain,
we employed a genetic strategy based on reversal of a slow-grth phenotype in yeast.
Overexpression of a Gal4-VP16 fusion is relatively toxic to yeast (Berger, S.L. et al.,
1992). This results in a slow growth phenotype, exhibited by reduced colony size of yeast
expressing a Gal4-VP16C fusion from a high-copy yeast vector. The vector used in the
experiments reported here (pPJHl3) is derived from a yeast 2p circle. This DNA is a
recombinant derivative of an endogenous yeast plasmid that maintains a copy number of
25-50 DNA molecules per cell. The plasmid expresses a Gal4-VP16 ﬁrsion from the
strong yeast ADH] promoter. A Similar selection strategy has previously been used to
identify yeast mutants deﬁcient for activator ﬁrnction in viva (Berger, S.L. et al., 1992).
Berger et al. demonstrated that mutations within the VP16 activation domain that reduce
transcriptional activity also reduce the severity of the toxicity phenotype. This linkage
suggested that reversal of toxicity could be used to isolate potential VP16 transcriptional

mutants.

55

 

Our isolation strategy is brieﬂy outlined as follows: (1) introduction of random
mutations within the VP16C subdomain, (2) subcloning of these mutants into pPJH13 to
produce a library of Gal4-VP16C fusions, (3) transformation of this library into yeast, (4)
selection of yeast bearing a transformed plasmid and displaying robust growth, (5)
isolation of the plasmid from identiﬁed colonies and sequencing of the activation domain,
(6) assay of the transcriptional activity of the VP16C mutant in yeast.

To introduce mutations within the VP16C subdomain, Taq DNA polymerase was
used to amplify VP16C under conditions that reduce polymerase ﬁdelity. The VP16C
PCR products were subcloned into pPJH13, producing a library of Gal4-PCR product
fusions. This pool of pPJH13 variants was transformed into yeast strain BP] (It/L4 Ta
ade1-100 ura3-52 leu2-3,112 his4-519) and cells were plated onto Leu' media for
selection of pPJH13 transforrnants. High level expression of this wild-type VP16C ﬁrsion
results in reduced colony size relative to empty vector controls ("wild-type VP16C" will
be used to refer to Gal4-fusions containing VP16 sequences which correspond to the
native VP16 gene sequence).

Colonies were selected for secondary screening based upon visual inspection of
colony size. Large colonies were selected and patched onto selective media to verify this
phenotype. DNA was isolated from candidate colonies and transformed into E. coli for
ampliﬁcation. The plasmids were then isolated from E. coli and sequenced to identify any
activation domain mutations. To assess the transcriptional phenotype of each mutant
Obtained, candidate VP16C activation domain mutants were subcloned into low copy
yeast expression vectors (ARS/CEN vectors). These vectors also express Gal4-VP16C

ﬁrsions under the control of the ADH] promoter, but are maintained at copy numbers of

56'

 

1-2 per cell. This allows lower levels of expression and reduces the toxicity of the fusion
proteins, allowing more accurate quantitation of their transcriptional activity. Activity was
quantitated by cotransformation of these Gal4-VP16C fusion plasmids with the plasmid
pLGSDS (Guarente, L. et al., 1982). This reporter plasmid contains a bacterial lacZ gene
under the control of the GAL] -] 0 UAS and a CYC] TATA box. The GAL] -] 0 Upstream
Activating Sequence (U AS) corresponds to the natural yeast sequences regulating the
GAL] and GALIO genes. The UAS contains four Gal4-binding sites. Three individual
transformants of each mutant were assayed by the method of Miller et a1. (Miller, J H
1972)

The collection of mutations identiﬁed as well as their relative transcriptional
activities appears in Figure 1. Of 14 mutants isolated, 11 have mutations occurring
between amino acids 470 and 480, highlighting the probable importance of this region for
activation domain ﬁrnction. Nine Of 13 missense mutations have lesions at one or more
phenylalanines within the activation domain. Eight of these phenylalanine mutations
identiﬁed occur within the core 470-480 region (F473, F475 or F479). Three of four
mutants classiﬁed as exhibiting a severe phenotype (less than 25% wild-type activity)
contained substitutions of one or more hydrophobic residues within this core region by
polar amino acids. Signiﬁcantly, conservative mutations of these hydrophobic residues
(e. g. F479L, F457L/F479L, F47 5L) exhibited little deleterious effect, suggesting that
retention of hydrophobic character at these positions is sufficient to confer activation
domain function. The remaining mutant exhibiting a severe phenotype consisted of a
prematurely terminated open reading frame due to mutation of the glutamine 477 codon to

a stop codon, reinforcing that the entire core region is essential for function.

57

 

 

$an 5 woueomvﬁ 8e 89596:; 3238.595 9.8 58m 3 3322: 8a Soho 3:23:85: 0: no 23: £3, 3:832

en: E 938me 03 53:3 Raozmtomgb E 82838 838 warms—“Ev 2384 Locum 28: Emu 05 E 33% 3—38 2E.
25502 E concave me 5338 3:23:85: Com cabana 803 3883 couezoe 38:8 Baa—onsm .260 Box 05:8 Bro—-m—wﬁm
E 8:865 225533 Bow 058“ 2: £3» 65883 begun £5 323 328me 8a 203838 eoaucoa .38 Bow 05:3 cote—
-2wEm 5 a9 2.: $82.... Deanna EeEoBa 03% 05 .«o 8325a SEE 2F 05% oczwm 05 E can: 08 @3352 30:8an
05. 303838 Enact 858,38 33:28 @962 9 30538 203 mEeEoe c2338 2:. 30508 E wantomow me @8023.
803 moabocona 9.22033 2: 8m oratoomoc $5338 USE/-225 £2338 ocosw .3 35:52 28:85: US?» ._ ouawwm

AXVN LOPWN‘FVO DDDDDDDDDDDDDNDDDDDDDDDDDDDDDDDDDDDDDDDDD

D
D
D
D
D
D
D
D
D
D
D
U)
D
D
D
D
D
D
D

:02 85> -.»».>»»...»».»p.»»»».».»O»»».»»»....»».»
gm? wcvmd DDDDDDDDDDDDDDDDDDDDDDQDDDDDDDDDDDDDDDDDD
:2 P38 trrrtpptprpppmppp.-p»..->.»>>»p»»».»»rrpt
:2 $422.8va »»»»»>»»»».»>»»»p..pprmprpp».».»».»»»»».r
as: mus: --prprppprrprrtqtppprp»»»»>>»»»»»»»»».-»»
$8 ﬁEaéSOm »»»»»».».»».»»»AO»»..rrpppprrprrprrtrtptp
:8 Eng ..ppprrrtppppppptmtrr.>ttpttprtrtrr».i-»
$2 $48.?va ..p»...».r.»»p>»Om..r»..-»>»»>»..p»».p»r»
§v asvmmﬁﬁzﬁvma »»».»».ptprmhptmrpr...t.»»p»»»».».p»»p>»»
$2 £48.8va tiprpppttopmrtppptrtrr.-,.--»ptprpptrptpp
32: 84.5.5315 pppprpp..trq».»»....»».»»..»>.r»»a»»»>»»»
$8 stud .»»»»»».»»»4»»»».r.»».»»>>»»»>>>»»»»pr»pp
E353 mzoﬂﬁbs oceans“Oa<nam20mmmmn<znq<o>m<mnmmsm0mommno

 

58

Mutants of intermediate phenotype include a F 457S substitution, a L468P mutation, a
F473S mutation, a L468P/M47 8V mutation, and a double E474G/F475L mutation.
Signiﬁcantly, non-conservative substitutions of aromatic residues outside the core region,
Y46SC and F457S had relatively mild eﬁ‘ects on activation. In addition, introduction of a
basic amino acid within the core region (Q477R) had little effect on activation. The fact
that introduction of a basic residue in the center of this highly acidic region has no impact
on transcriptional activation implies that the acidic character of the motif is not of critical
importance.

The frequency of observed mutations under these PCR conditions is worth noting.
The occasional isolation of wild-type sequences as well as the isolation of activation
domain mutants with wild-type phenotypes (for instance F479L) probably reﬂects the
subjective nature of the growth phenotype selection. A high percentage of A:T to G:C
transition mutations was Observed. There was a low incidence of G:C to A:T transitions,
as well as transversions in general. Only one frameshift was observed. The cumulative
error rate was determined to be 10'2 errorpr sequenced. It is important to remember
. however, that these data are representative of the screened population, not the entire
population of ampliﬁed DNA.

A high frequency of transition mutations has been reported for Taq DNA
polymerase. This has been attributed to a greater efﬁciency of extension of transition
mispairings compared to transversion mispairings (Huang, M. et al., 1992). Additionally,
A:T to G:C transition mutations are naturally favored due to the unusual stability of the
non— Watson-Crick G:T base pair. Also, in this mutagenic PCR reaction nucleotide

concentrations were not equal. The dATP concentrations were lowered to enhance

59

mutation frequency. This low concentration of dATP could contribute to the Observed
bias.

Overall, these results suggest that the function Of the VP16C activation domain is
dependent upon a critical collection of aromatic residues within a region spanning amino
acids 470 to 480. Substitution of polar residues for any of the hydrophobic residues
within this "core motif" signiﬁcantly reduces transcriptional function. These mutations do
not however allow us to discern the relative contribution of each hydrophobic residue, due
to the high percentage of mutations with multiple amino acid changes. Hydrophobic
character, and not aromatic character per se, appears to be important at residues 473, 475
and 479. Furthermore, the mild effects of non-conservative mutations of hydrophobic
residues outside of the 470-480 core suggests that sequences external to this core are not
critical for function.

These results are consistent with the results of subsequent experiments performed
in our laboratory by Susan Sullivan in collaboration with Richard Ebright (Rutgers
University). A collection of alanine mutations (an "alanine-scan") was produced at each
position within the VP16C activation domain. The activity of each mutant was
determined by cotransformation assays in yeast (using pLGSDS as described above) and
by cotransfection assays in mammalian cells (CAT assays as described in following
sections). The results in yeast and mammalian cells were essentially concordant.
Mutations within the 470-480 core region displayed the greatest reductions in activity.
Speciﬁcally, mutants in which an alanine replaced phenylalanines at positions 473, 47 5 or
479 were signiﬁcantly reduced in activity. In addition, an E476A mutation was also found

to be deleterious to activity. These results highlight the importance of multiple

60

 

phenylalanines within the core amino acid 470-480 region, as well as a single acidic
residue (E476).

These combined mutational results suggest a similar amino acid requirement to
that previously observed for VP16N (Cress, W.D. and SJ. Triezenberg, 1991; Regier, J.L.
et al., 1993). Activation by the VP16N subdomain is critically dependent upon F442. An
F442A mutation essentially abolishes activity. In addition, two ﬂanking hydrophobic
residues (L43 9, L444) appear to contribute to the strength of the domain. While no single
amino acid substitution in VP16C has an effect of the magnitude of a F 442A substitution
on VP16N, the overall composition of the two subdomains is relatively similar; both
require multiple aromatic or hydrophobic residues for function. Thus, these two
subdomains appear to differ in structure only by relatively subtle criteria. In contrast to
VP16N, the VP16C subdomain appears to rely upon multiple phenylalanine residues for
ﬁrnction. In addition, the exact spacing or phasing Of the hydrophobic residues within the .
two subdomains differs. These results seem to imply that the subdomains may be
structurally redundant.

To more directly compare these two subdomains, we performed an alanine-scan of
VP16N analogous to that performed on VP16C. Alanine mutations were introduced by
site-directed mutagenesis within the VP16N region ﬂanking the critical F442 residue
(amino acids 43 9-447). Transcriptional activity was assessed by cotransfection assays
with a bacterial chloramphenicol acetyltransferase (CAT) reporter construct. This
construct (pG5T+I/CAT) contains ﬁve Gal4 binding sites and a composite promoter
composed of the adenovirus major late TATA box and the terminal

deoxynucleotidyltransferase Inr

61

Figure 2. Relative activity of VP16N alanine substitutions in the activation domain
core region. Panel A. Alanine substitutions were introduced at each position throughout
the core of the VP16N activation domain. The activity of each was assessed as a Gal4-
VP16N fusion protein by cotransfection of a mammalian Gal4-VP16N expression vector
and a reporter construct which expresses chloramphenicol acetyltransferase (CAT) from a
composite TATA + Inr promoter downstream of ﬁve Gal4 binding sites. CAT activity
was determined by a ﬂuor diffusion method, using 3H—acetyl-COA and unlabelled
chloramphenicol. Activity (y-axis) is expressed as nmol acetyl-COA acetylated per hour,
per mg cell extract. The activity of the wild-type VP16N activation domain appears on the
left. The position of each alanine substitution mutant is indicated along the x-axis. Panel
B. The activity of various substitutions of amino acid 442. Activities were assessed as
described for Panel A. For each mutant, the amino acid substitution examined appears
along the x-axis.

62

 

rm

 

_ <25
«6sz
mmeva
{£va 1
<33 mass
«5va l
<3: - dad
I <35
. ﬁe;
€6va l
. «$va Dab 6:3

 

a

    

 

 

160 ~-

093 ES? 4 l L l + l _
_ _ k _ 0 o 0 0 0 0 0
. _ l _ _ _ _ _ 2 0 8 6 4 2
0 0 0 0 0 0 0 0 1 1
441 2 0 8 6 4 2
1 1 .
B 55.8 2:. 23 .x.

as?“ 2:. 2;» .x.

63

element upstream of the CAT gene. Enzymatic CAT activity in cell-free lysates was used
as a measure of transcriptional activity.

The results are relatively similar to those obtained with VP16C (see Figure 2A).
Although F442A is by far the most deleterious mutation, additional hydrophobic residues
(L43 9, L444, L447, and to a lesser effect M446) also contribute signiﬁcantly to
transcriptional activity. Interestingly, the acidic amino acid immediately following F442 is
also important for activity. This parallels results Obtained with VP16C. Analysis of

F442L, F442A, F442S, and F442P mutations were similar, albeit somewhat more severe

 

than previously Obtained results using VP16(1-456) truncations. A F442L mutant
exhibited some residual activity while F442A, F4428 and F442P substitutions completely
abolished activity (see Figure 2B). Overall, these results argue that the VP16 activation
subdomains are each composed of a pair of critical residues, a phenylalanine followed by
an acidic amino acid, which are augmented by a collection of ﬂanking hydrophobic

residues.

A related "acidic" activator RelA also contains similar motifs

 

RelA, the p65 subunit of the NF-KB family of speciﬁc transcription factors, is
involved in a variety of immune events including B cell development, T-cell activation and
inﬂammatory response (reviewed in Siebenlist, U. et al., 1994). Others have previously
suggested that the RelA and VP16 activation domains may be structurally and/or
functionally similar based upon several observations. Both activators are active in yeast as
well as mammalian cells. Both activators are dependent upon the yeast ADA genes for

ﬁrnction in Saccharamyces cerevisiae. Both activators have highly acidic activation

64

 

domains, and are unusually potent. Collectively, these results imply that VP16 and RelA
may contain structurally related activation domains (Blair, W.S. et al., 1994b; Blair, W.S.
et al., 1994a).

The RelA activation domain is composed of three independently ﬁrnctional
segments as indicated by deletion analysis (Moore, RA. et al., 1993). These subdomains
reside between residues 417-459 (referred to as RelAI in this text), residues 459-521
(RelAII), and residues 508-551 (RelAIH). Studies by Cullen and coworkers have
established the importance of multiple phenylalanine residues throughout the activation
domain (Blair, W.S. et al., 1994b). Blair et al. proposed that VP16 and RelA are
ﬁrnctionally similar and are likely to contain homologous activation domains. Based upon
an alignment of VP16N and RelA sequences, the authors suggested that amino acid
segment 43 6—453 (LQFDDEDLGALL) and 535-545 (SIADMDFSALL) within the RelA
activation domain are structurally related to VP16. The 43 6-453 and 535-545 sequences
are present in the RelAII and RelAIII subdomains, respectively. Direct ﬁrsion of either of
these segments to Gal4(1-147) produced synthetic activators capable Of function in yeast.
Mutational analyses performed on the Gal4-RelA(535-545) constnrct veriﬁed that multiple
hydrophobic and acidic amino acids within this RelA segment were essential for activity
(Blair, W.S. et al., 1994b). The central F542 residue within RelAIII required aromatic
character for function, and alteration of the preceding aspartic acid to either glutamine or
lysine signiﬁcantly impaired activity. A L545A mutation also diminished the activity of the
subdomain. Interestingly, introduction of an additional phenylalanine or aspartic acid
residue dramatically increased activity. While these mutational studies were

comprehensive, no analyses of these mutants were performed in mammalian cells.

65

 

 

Baeuerle and coworkers have also analyzed the structural requirements for RelAIII
transactivation using a Gal4-RelAIII chimeric activator (Schmitz, ML. et al., 1994). Site-
directed mutagenesis largely conﬁrmed the Observations of Blair et al. The studies were
performed in mammalian cells, however the Baeuerle data set is less thorough than that
examined by Blair et al. The authors verify that F542A or F 542P mutations negate
activity in viva. Furthermore, a D541M/DS43M double mutant displayed little activity,
and a pair of alanine substitutions at D53 land D533 diminished activity suggesting that
the acidity of the domain contributes to activity. However, substitutions of these acidic
residues were not analyzed individually. Based on their results, the authors propose that
RelAIII is composed of a pair of direct repeats, with a critical DF core element. The same
authors describe a similar motif in RelAII based on structural comparison, although this
hypothesis was nOt tested by mutagenesis (Schmitz, M. et al., 1995).

While these results indicate that RelA activation domain function is dependent
upon F542 and a number of ﬂanking hydrophobic and acidic residues, the studies do not
present a systematic analysis that can be directly compared to our VP16 results.
Therefore, we desired to analyze RelA in a similar fashion by construction of an alanine
scan Of the activation domain. Based on the supposition that the VP16 and RelA
activation domains are ﬁrnctionally analogous and therefore potentially homologous in
activation domain structure, we attempted to identify segments of the RelA activation
domain that resembled VP16N or VP16C in primary structure.

RelA homologues have been cloned from four species. The protein is composed

of a very well conserved rel homology domain Of approximately 300 amino acids and a

66

 

 

 

 

 

 

 

 

 

PQAVAPPAPKPTQAGEG T—qﬁf
PQSLSAPVPKSTQAGEG, s H
P SQSDHD SMLN P P
P ASSPTPEP .1 VGGSGPPPSTTTTT
E L! r-NST
E LLGNST
EML
TTTQCALGGGIPDPGGSP aye

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

EB I PVAPHT
SMS HS
S-Ss SAALQI‘
EPPG IGAGGGFGELLSLPTNFGDP PSS TFG'
r- v AQ PDPAPAP P
‘ I v SQ PDPAPTP TS
GTsoch ‘ Sr NEDE
P Ls- y vac GSGGGGP P PPQDL
LP 3 D
LP 5 D
ERIDS I FDIS
g P PPQPT

 

 

 

 

Figure 3. RelA Activation Domain Alignment. The GCG program pileup was used to align
RelA activation sequences. Alignments are arranged relative to the human sequence. Four
sequences of the human, mouse, Xenopus, and chicken cDNAs are arranged in order top to
bottom. Only activation domain sequences beginning with human position 417 (or the
corresponding amino acid ﬁ'om the other species) are depicted. Boxes indicate regions of
similarity in all four homologues. Bold characters denote amino acids which are identical in at
least three of four homologues. Spaces correspond to gaps in the alignment.

67

 

  
 

550
]| 551

VP16C F EaFlE Q
RelAII S E Fi
RelAIII M DIES A I. L S

N

 

Figure 4. Comparison of VP16 and RelA sequences. Panel A. HCA program output
of the VP16 activation domain. Hydrophobic activation domain amino acids are
highlighted in yellow. Amino acids are indicated by Single-letter amino acid code.
Proline residues are represented by a star, and glycines are indicated by a diamond.
Threonine residues are indicated by a box, and serine residues by a box with an internal
dot. Panel B. HCA program output of the RelA activation domain. Hydrophobic
residues within the two RelA segments proposed to be related to VP16C are shaded
yellow. Symbols are as indicated in Panel A. Panel C. Linear alignment of VP16C,
RelAII, and RelAIII amino acid sequences. Amino acids 473—480 of VP16C, 472-479 of
RelAII, and 540-547 of RelAIII are shown. Hydrophobic residues conserved within the
proposed alignment are indicated by yellow shading.

C-terrninal extension that contains the activation domain. While the activation domain
sequences are not as well conserved, alignment of the activation domains by the GCG
algorithm GAP does identify four regions of conservation within the activation domain
(see Figure 3). The ﬁrst segment is interrupted in two Of four species. The remaining
three correspond to sequences within the RelAII and RelAIII subdomains.

To facilitate the search for RelA sequences that resembled VP16 sequences in
primary structure, we utilized a method of schematic comparison called Hydrophobic
Cluster Analysis (HCA). This computer-assisted method displays the primary structure of
a protein as a continuous helical projection. The projection is displayed in two dimensions
as if the helix was opened along its length and laid ﬂat. A duplicate helix is reproduced
below the original in order to preserve the continuity of all residues (Lemesle-Varloot, L.
et al., 1990). By visual inspection, comparison of patterns of hydrophobic residues can be
readily performed. A sample of the HCA patterns Of VP16N and VP16C sequences
appears in Figure 4A. Note that the relatively subtle differences in primary structure
between VP16N (LDDFDLDML) and VP16C (FEFEQMF) are easily discerned as a
distinct shape or "cluster" of hydrophobic residues. Using this method to assist sequence
analysis, we identiﬁed two segments, one within RelAII and one within RelAIII, that
resemble VP16C in primary structure (see Figure 4B). The ﬁrst segment occurs in
sequences between residues 474 and 482. The second, less Obvious similarity occurs
between residues 540 and 55 1. Both of these regions correspond to highly conserved
regions of the RelA activation domain. A linear alignment of these sequences with VP16C

appears in Figure 4C.

69

200-;
180 ‘

g, 160}

E 140-,

33

E 120-

E 100 1;

3

e\°

L478A F

N471A
S472A If
E473A
F474A |
Q475A .
Q476A
L477A

Wild type

Figure 5. Relative activity of RelAII alanine substitutions in the activation
domain core region. Site-directed mutagenesis was employed to introduce
single-amino acid alanine substitutions at each position throughout the core of the
RelAII activation domain. The activity of each was assessed as a Gal4-RelAII
fusion protein by cotransfection of a mammalian Gal4-RelAII expression vector
and a reporter construct which expresses chloramphenicol acetyltransferase from
a composite TATA + Inr promoter downstream of ﬁve Gal4 binding sites.
Activity (y-axis) is expressed as nmol acetyl-coA converted per hour, per mg cell
extract. Note the log axis to allow comparison of the F473A, L475A and L476A
mutants. Activity was determined by a ﬂuor diffusion method, using 3H-acetyl-
coA and unlabelled chloramphenicol. The activity of the wild-type RelAII
activation domain appears on the extreme leﬁ. The position of each alanine
substitution is indicated along the x-axis.

70

TO analyze these domains in a fashion comparable to our VP16 results, we
constructed an alanine-scan Of both of these regions Of RelA. Transcriptional activities
were determined by cotransfection assays in mammalian tissue culture using mouse L
ﬁbroblasts. The reporter used was the previously described pG5T+I/CAT. Amino acids
469 to 478 (469-VDNSEFQQLL-478) were analyzed in the context Of a Gal4-RelAII
fusion protein (amino acids 1-147 Of Gal4 ﬁrsed to amino acids 459-521 OfRelA). Based
upon the sequence comparison to VP16C, alanine substitutions of F474, L477, L478, and
perhaps V469 would be expected to diminish activity Of the domain (see alignment, Figure
4C). The results Of the cotransfection assays are depicted in Figure 5. Consistent with the
predictions, a V469A mutation displayed a moderate reduction in activity, while F474A,
L47 7A, or L478A mutations nearly abolished activity. These results suggest that the
RelAII activation domain is similar in its structural requirements to the VP16C activation
domain. The subdomain is dependent upon multiple hydrophobic residues for ﬁrnction,
with a consensus Fxx‘l"? (where the symbol ‘1’ represents a hydrophobic amino acid and x
denotes a position insensitive to alanine substitution) motif that resembles that Observed in
VP16C.

The RelAII] activation domain also contains a sequence segment that somewhat
resembles VP16C in primary structure. While this similarity is less obvious, the RelAII]
domain is the best characterized by previous studies and thus of particular interest. Two
studies have suggested similarity between this subdomain and VP16N. Blair et al.
concluded that RelAIII contains a VP16N related domain based on the activity of the
domain in yeast as well as mutagenesis results in yeast (Blair, W.S. et al., 1994b). Schmitz

et al.

71

 

 

120 1‘
,5; 100 i
.2 1
i l
a so
0 l
O- l
b 1
2 so ‘
3
.\°

N 43

O C O

% J_L l.
4

M539A —
.y

<
ax
v
In
I—l

o A

stir-$.53.“ Sﬁﬁsﬁéﬁé
Z‘nnnm VVVmVVVV
Inlnlnln mmm<mmmm

Figure 6. Relative activity of RelAIII alanine substitutions in the activation
domain core region. Site-directed mutagenesis was employed to introduce
single-amino acid alanine substitutions at each position throughout the core of the
RelAIII activation domain. The activity of each was assessed as a Gal4—RelAIH
fusion protein by cotransfection of a mammalian Gal4-RelAIII expression vector
and a reporter construct which expresses chloramphenicol acetyltransferase from
a composite TATA + Inr promoter downstream of ﬁve Gal4 binding sites.
Activity (y-axis) is expressed as nmol acetyl-coA converted per hour, per mg cell
extract. Activity was determined by a ﬂuor diffusion method, using H-acetyl-
coA and unlabelled chloramphenicol. The activity of the wild-type RelAIII
activation domain appears on the extreme left. The position of each alanine
substitution is indicated along the x-axis.

72

also noted that RelAIII is composed Of a pair of direct repeats Of critical DF residues, a
motif that appears in many activation domains including VP16 (Schmitz, ML. et al.,
1994)

We constructed an alanine-scan of the RelAIII domain from residues 534 to 549.
The DF repeats highlighted by Schmitz, et al. correspond to the F534 (an aspartic acid
precedes this residue) and D541, F 542 sequences. Mutants were assayed in mammalian
cells as Gal4-RelAIII fusions (amino acids 1-147 of Gal4 ﬁrsed to amino acids 508-551 of
RelA). The activities of these alanine substitution mutants are shown in Figure 6.
Although multiple hydrophobic residues throughout the subdomain contribute to activity,
residues within the second repeat proposed by Schmitz, et al. appear more critical for
ﬁrnction (compare the phenotypes of F534A and F542A mutations). In conﬂict with the
proposal, the two repeats also differ with regard tO the degree of contribution Of
additional ﬂanking hydrophobic residues. While F 534 appears sufficient for relatively
robust activity (the 1537A and M539A mutants retain 30% or greater activity), the second
repeat contains two additional critical hydrophobic residues. Both L545A and L546A
substitutions resulted in defects nearly as severe as that exhibited by the F 542A mutation.
Blair et al. also noted the greater contribution of the C-terminal repeat in yeast, as
evidenced by the greater effect Of a F 542A mutation relative to a F 534A mutation (Blair,
W.S. et al., 1994b). These authors also reported that both 1537F/D539N/D541N and
D539K/D541K RelAIH mutants displayed severely impaired activity in yeast. In contrast,
our D541A mutation has limited effect on activity suggesting that this acidic residue
contributes only partially to function. This discrepancy may reﬂect a diﬁ‘erence in

activation domain ﬁrnction in yeast and metazoan systems or more likely may simply be a

73

 

 

consequence Of the substitutions themselves. The more severe effects associated with the
double mutants are not surprising, given the fact that a D539A mutant displays slightly
reduced transcriptional activity. Alternatively, the phenotype Observed by Blair et al. may
result from the substitution Of a larger amino acid for aspartate. Perhaps the larger
residues are sterically detrimental while substitution of the smaller alanine side-chain is
more permissive.

Our mutational analysis is also consistent with our proposed alignment of VP16C,
RelAII and RelAII]. Both the Cullen and Baeuerle laboratories noted sequence Similarity
between VP16 and RelAIII. Schmitz et al. also noted a similar motif within RelAII,
although no mutational analysis Of this domain was conducted (Schmitz, M. et al., 1995).
Both groups suggested that a critical DF sequence motif, and the observed contribution of
neighboring hydrOphObic residues, resembles the structural requirements of the VP16N
subdomain. Based on our own sequence analysis, we predicted that both RelA
subdomains bear a greater resemblance to VP16C. This is suggested by two Observations.
First, the spacing of the aromatic and hydrophobic residues within the RelA domains more
directly resembles VP16C. Comparison suggests that a common Fxx‘IJ‘I’ core motif
(where ‘1’ denotes a hydrophobic amino acid) is essential for VP16C (F475, M478, F479),
RelAII (F474, L477, L478) and RelAIII (F 542, L545, L546) ﬁrnction. Second, the
aspartic acid residue preceding F 542 in RelAIII (suggested by Blair et al. and Schmitz et
al. to be critical for RelAIII function) is relatively insensitive to alanine substitution. This
property resembles the sensitivity Of E474 in VP16C and E473 in RelAII. While each of
these subdomains contains a similar hydrophobic core motif, they also contain critical

differences. In contrast to the RelA subdomains, both the VP16N and VP16C subdomains

74

 

contain a Single critical acidic residue immediately C-terminal to an essential phenylalanine
residue (F442, D443 in VP16C and F475, E476 in VP16C). Neither RelA subdomain
contains a corresponding residue within the core motif. RelAII activity is independent of
any Of the hydrophilic residues within the sequences examined, and mutational analysis Of
RelAIII revealed only one acidic residue with any Signiﬁcant role. This residue, D541,
resides immediate N-terminal to the critical F542 residue. A serine residue resides in the
C-terrninal ﬂanking position and is insensitive to substitution by alanine. These results
illustrate that the VP16C and RelA subdomains contain similar sequence elements
essential for their function. While these motifs are similar both in their requirement for
multiple hydrophobic residues for function as well as in the precise phasing Of these
essential residues within the primary structure, the results also reveal a number of

differences between these motifs.

MSSIOH and Conclusions

Our laboratory has been interested in the structure/ﬁrmction requirements of
transcriptional activators. To investigate these requirements we have used three principal
models, two of which are the topic Of this dissertation. Our studies have focussed on the
HSV lytic-cycle activator VP16, and the mammalian activator RelA, a member of the
metazoan NF-KB family Of transcription factors. In addition, the laboratory is involved in
yeast studies employing Hap4, a transcriptional activator involved in metabolic adaptation
to non-fermentative environments.

The primary structural requirement for activation domain function has been an

intensively studied topic. Previous work on the VP16N subdomain identiﬁed a critical

75

 

phenylalanine (Cress, W.D. and SJ. Triezenberg, 199]) and ﬂanking hydrophobic residues
that are important for VP16N function (Regier, J.L. et al., 1993). Site-directed
mutagenesis of a small region Of the VP16C subdomain identiﬁed a Similar qualitative
requirement for hydrophobic residues, however the studies failed to identify a mutation
with similar severity to the F442A mutation previously identiﬁed in VP16N (Regier, J .L.
et al., 1993). While these results suggested a qualitative similarity between the two
subdomains, they seemed to indicate that other features within VP16C might also
contribute to function.

To ascertain the structural elements critical for VP16C function we employed a
genetic strategy designed to select potential VP16C activation domain mutants. The
strategy exploited the fact that overexpression of a Gal4-VP16 fusion is partially toxic to
yeast cell growth. This phenotype appears to require activation domain ﬁrnction, as an
F442A mutation within VP16N diminishes this phenotype (Berger, S.L. et al., 1992). We
produced a library of VP16C mutants by PCR under low ﬁdelity conditions and
introduced this library as a collection of high-copy Gal4-VP16C expression vectors into
yeast. We reasoned that by selection of yeast that express the VP16C ﬁrsion but still
display robust growth, we could Obtain VP16C mutants with reduced transcriptional
activity. DNA was recovered from candidate yeast colonies and sequenced to identify
mutations within the VP16C activation domain. The transcriptional activity of the
encoded activation domain was then assessed using a cotransformation assay in yeast.
Using this strategy, we identiﬁed several VP16C mutants.

Of 14 mutants identiﬁed, 11 contained mutations within a Single short segment of

the VP16C subdomain. Three of four mutants with particularly severe phenotypes had

76

 

non-conservative substitutions Of one or more phenylalanine residues within this 470-480
region. The fourth involved truncation of the subdomain at residue 477 due to pre-mature
termination. Three mutants with phenylalanine to leucine substitutions exhibited robust
activity, suggesting that hydrophobic character is sufficient for subdomain function.
Furthermore, even non-conservative mutations of aromatic residues outside of the core
470-480 region exhibited little loss of activity. Collectively these results argue that
VP16C requires multiple hydrophobic residues within a core region (amino acids 470-480)
for activator function. These results are largely consistent with the results of an alanine-
scan of VP16C conducted by Susan Sullivan (Sullivan, SM. et al., 1998). Alanine
substitutions of F473, F475, E476, or F479 signiﬁcantly reduced the activity of VP16C.
Furthermore, the phenotypes of these mutants were generally similar in yeast and
mammalian cells, suggesting that comparison Of previous VP16N mutagenesis results in
mammalian cells to these VP16C results in yeast is justiﬁable.

These results, combined with those of Sullivan, et al., suggest both an element of
similarity (requirement for hydrophobic residues for function) and distinction between
VP16N and VP16C (e. g. the lack of any Single alanine substitution within VP16C, that
displayed a severity comparable to the VP16N F442A substitution). To allow a more
direct comparison of the VP16N and VP16C subdomain, we produced an alanine-scan of
VP16N analogous to that constructed in VP16C. We concentrated on a region from
amino acids 439 to 447 that contained multiple hydrophobic residues surrounding the
critical F422 residue. Two leucine residues C-terrninal to F442 were also found to be of
signiﬁcant importance for ﬁmction. In addition, a single acidic residue appears signiﬁcant

as evidenced by the severely reduced activity of a D443A mutant activation domain.

77

 

These results mirror those Obtained with VP16C. The two subdomains require a central
dipeptide (a phenylalanine residue followed by an acidic residue) and ﬂanking hydrophobic
residues for function. The only signiﬁcant difference between the two domains appears to
be the relative spacing Of these hydrophobic residues within the domain. These results
extend the Observations of Regier et al, identifying additional residues Of signiﬁcance for
VP16N function.

We have also investigated the structural requirements of the NF-KB family member
RelA. This activator has been suggested to resemble VP16 both structurally and
ﬁmctionally. The RelA activation domain has been subdivided into three independently
functional subdomains by deletion analysis (Moore, RA. et al., 1993). Schmitz et al.
proposed that the most C-terminal Of these activation domains (RelAIII) was constructed
of a direct repeat, with a critical DF motif central to each repeat (Schmitz, ML. et al.,
1994). The authors also Observe that this motif can be found in several transactivation
domains, including both subdomains of VP16 and a sequence motif within the RelAII
subdomain (Schmitz, M. et al., 1995). Their alignment predicts that VP16N residue D441
and VP16C residue D472 would be critical for function. Our observations indicate that
neither of these residues is sensitive to alanine substitution (see Figure 2A and Sullivan,
SM. et al., 1998). Based on the Observed ﬁmctional similarities (function in yeast, ADA
dependence in yeast, extremely robust activity) between VP16 and RelA, we were
interested in identifying potential VP16-like motifs within the RelA activation domain.

To search for similarity we employed a computer algorithm, HCA, which employs
a visual pattern recognition technique (Lemesle-Varloot, L. et al., 1990). Using this

algorithm we identiﬁed two activation domain segments, one within RelAII and one within

78

 

RelAIII, that contained a patch of hydrophobic residues similar to that observed in VP16C
(see Figure 4). Comparison of these sequences predicts that a critical Fxx‘IJ‘P motif
(where ‘I’ denotes a hydrophobic residue) should be essential for activity. TO test this
hypothesis, we constructed alanine scans of both the RelAII and RelAHI subdomains.
Assay of these mutants conﬁrmed these predictions. In RelAII, F474A, L477A and
L478A mutations severally reduced transactivation. Similarly, in the RelAIII subdomain,
F542A, L545A, and L546A substitutions severely compromised activity. Notably, both
RelA motifs lack an acidic residue immediately following the critical phenylalanine residue.
We have identiﬁed common structural requirements for activation domain function
in two potentially related activators. These motifs require a critical phenylalanine residue
as well as C-terminal hydrophobic residues. However, the motifs also have individual
characteristics that suggest subtle differences in structure. The VP16N and VP16C
subdomains differ in the precise Spacing of the critical hydrophobic residues within the
motif. Furthermore, the VP16C subdomain contains multiple phenylalanine residues that
appear to be equally critical to function as evidenced by alanine-scanning results (Sullivan,
SM. et al., 1998). In contrast, VP16N contains a single, dominant phenylalanine residue.
The RelAII subdomain closely resembles VP16C in primary structure, however the residue
immediately C-tenninal to the critical phenylalanine residue in RelAII is a non-acidic
glutamine residue that is insensitive to alteration. This contrasts with the motifs Observed
in both VP16N and VP16C. In contrast to all three of these subdomains, RelAIII also
contains an N-terrninal extension with an additional phenylalanine residue. However, a
F534A substitution has a relatively small effect on RelAIII ﬁrnction, when compared to

the dramatic effects of F542A, L545A, or L546A mutations. The residue immediately C-

79

 

terminal to the F 542 residue in RelAIII is also not acidic. An SS43A substitution has little
effect on transcriptional activity.

Overall, these results suggest that activation domains contain similar, yet
individually distinct hydrophobic elements critical to their function. The motifs are
composed of short segments of non-polar residues that appear to consist of approximately
10 residues. Surrounding residues are variably important for activity. This Observation is
interesting in light of studies performed by Schaffner's laboratory (Seipel, K. et al., 1994).
Seipel et al. found that an 11 amino acid segment Of the VP16N subdomain was sufﬁcient
to activate transcription when multimerized as a Gal4 fusion protein. This result implies
that all essential structural information is contained in this short peptide sequence. This
idea is consistent with the model that activation domains are composed of critical short
peptide segments, or core elements, that when combined can ﬁmction to efficiently

activate transcription.

Experimental Methods

Recombinant DNA manipulations

The VP16C coding sequences were subcloned into the phage mpJR92 (Regier,
J .L. et al., 1993), an M13 derivative. The mpJR92 phage contains an insertion of a BglII
linker into the SmaI site of the M13 polylinker, downstream of the coding sequence Of the
full activation domain of VP16 (413-490). The full activation domain sequences Of VP16
(SalI-BamHI fragment) were replaced with a C subdomain-encoding Sall-BamHI

fragment from pLA101-VP16ASma to generate the phage mpPJH18.

80

The recipient vector for library construction, pPJH13, was derived from pr92
(J. Regier, S. Triezenberg, unpublished), to provide unique SalI and BglII sites for
subcloning of mutagenized PCR products. Parental pr92 was cut with PstI and blunt-
ended using T4 DNA polymerase. This plasmid was ligated with a T4 DNA polymerase
blunted XbaI-XbaI fragment from the same plasmid. This manipulation results in the
deletion Of a small segment of pEMBL polylinker present in pr92 that contains two SalI
sites. The plasmid expresses Gal4-VP16C sequences from the strong ADH] promoter and
contains a LE U2 marker gene for selection of the plasmid in Leu' yeast strains.

The Gal4-VP16N expression vector (pSGVPN) was constructed by insertion of
Gal4-VP16N coding sequences into the mammalian expression vector pSGp65(307-551)
(Moore, RA. et al., 1993) replacing the entire RelA activation domain sequence. The
plasmid pACB41 (A. Cress, S. Triezenberg, unpublished) was linearized with BamHI and
blunted with Klenow enzyme. The Gal4-VP16N sequences were then excised from
pACB41 by digestion with XhaI. The plasmid pSGp65(307-551) was digested with XbaI,
blunted using Klenow enzyme, and then digested with Xhal. Ligation Of the pSGp65(307-
551) vector with pACB41 Gal4-VP16 sequences generated pSGVPN. The ligation of the
blunt BamI-II and XbaI sequences regenerates the BamHI site in VP16, producing a fusion
of VP16N sequences to the poly stop linker of pSGp65(307-551). A variant of pSGVPN
designated pSGVPN(Sac’) was constructed by digestion of pSGVPN at a unique site with
Sacl, T4 DNA polymerase polishing to remove the Sacl 3'-overhang, and religation. This
process deleted the unique SacI site in VP16N and facilitated subcloning of the VP16N
mutants. The VP16N mutations were introduced by conventional Oligonucleotide site-

directed mutagenesis (Kunkel, TA, 1985) using the M13 phage pDCl. Following

81

identiﬁcation of mutations by sequencing, the activation domain sequences were
subcloned ﬁom M13. Replicating form Of M13 was prepared, and VP16N sequences
were excised by SalI and BamHI digestion. These fragments were subcloned into Sall-
BamHI digested pSGVPN(Sac'). Recombinants were identiﬁed by SacI screening and
veriﬁed by sequencing.

The Gal4-RelAII mammalian expression vector was constructed in two steps. A
BamHI-Smal RelA activation domain fragment was ﬁrst subcloned into pB S/KS+. The
pBS/KS+ polylinker was digested with Spel followed by blunting of the fragment with
Klenow enzyme and restriction with BamHI. Insertion of the BamHI- Smal sequences
into the pB S/KS+ vector to produce the vector pBSp65(459-521) produces a stop codon
in frame with RelA sequences upstream of pB S/KS+ polylinker sequences. A BamHI-
XbaI fragment was then excised from pBSp65(459-521) and subcloned into BamHI-Xbal
digested pSGp65(307-551). This vector, pPJH65(459-521), was utilized for all
subsequent RelAII manipulations. The Gal4-RelAIII was a gift from Paul Moore (Moore,
RA. et al., 1993).

Mutagenesis Of RelA sequences was performed by PCR. For production Of RelAII
mutations, a two step PCR protocol was adopted. This procedure requires two templates
with diﬁerent nucleotide sequences 5' and 3' of RelA coding sequences. A PCR reaction
was performed using pPJH65(459-521) as template. The RelA primers used were
constructed with alanine codon substitutions at each desired position within RelAII. The
position of the missense codon is at the center of the Oligonucleotide. The primers were
selected to minimize hairpin formation as well as optimize PCR performance in both PCR

reactions. This requires that the PCR primers be relatively large (3 0-35 nucleotides) and

82

hybridize strongly at both the 5' end and 3' end. First PCR reactions were performed for
20 cycles (95°C melt 15 seconds, 45°C anneal 15 seconds, 72°C elongate 30 seconds)
using the mismatched RelAII primers and ST-97. ST-97 hybridizes to Gal4 coding
sequences 5' Of the Gal4-RelAII cloning boundary.

The product of the ﬁrst PCR reaction was gel puriﬁed on agarose and isolated
using Qiaex (Qiagen) per the manufacturer's instructions. One half of the isolated material
was used in a second PCR reaction. The second PCR reaction was performed using this
"maxi-primer", Oligonucleotides ST-97 and ST-S, and pBS/p65(459-521) as template.
The ST-5 primer hybridizes to polylinker sequences in pB S/KS+ (extension from ST-5
synthesizes template strand RelA sequences). The PCR reaction was performed as a two
step PCR. First, a. two cycle, two temperature initial PCR reaction (95°C melt 15
seconds, 72°C anneal/elongate) was performed to produce a hybrid maxi-primer
pBS/p65(459-521) template for the second PCR step. This step was immediately
followed by a 20 cycle three temperature PCR (95°C melt 15 seconds, 45°C anneal 15
seconds, 72°C elongate 1 minute) to amplify the hybrid product. PCR products were
phenol chloroform extracted, digested with BamHI and XbaI, gel puriﬁed and isolated.
Processed fragments were introduced into BamHI and XbaI cut pSGp65(307-551).
Mutants were identiﬁed by restriction digestion and DNA sequencing.

Mutations within RelAIII were introduced by a similar PCR strategy. The ﬁrst
PCR was performed using pBS/p65(417-551) as template. The mutagenic primers contain
RelAIII coding sequences with appropriate alanine codon substitutions. The 3' primer
used was ST-212, a primer complimentary to pBS polylinker sequences that extends

during PCR from polylinker sequences into the 3' end of RelA sequences. This PCR

83

 

product was agarose gel puriﬁed and used as a maxi-primer in the second PCR reaction as
described above. The second template used was pSGp65(508-551) which encodes a
Gal4-RelAIII ﬁrsion. The reaction was performed using primers ST-281 (a Gal4 coding
strand primer) and ST-212. The PCR product was subcloned by BamHI-XbaI digestion

and ligation into BamHI-XbaI digested pPJH(417-551).

PCR mutagenesis and screening of VP16C mutants

 

To introduce mutations, Taq DNA polymerase was used to amplify VP16C
activation domain sequences under low ﬁdelity conditions. The ﬁdelity of the enzyme was
reduced by incorporation Oan2+ into the PCR reaction (Leung, D.W. et al., 1989), as
well as by use of an unequal concentration of nucleotides. The PCR was performed using
the double-stranded replicating form of a VP16C containing M13 phage (mpPJI-I18) and
primers ST-S (M 13 polylinker forward primer) and ST-39 (polylinker reverse primer), in a
PCR buffer containing 10 mM Tris-HCl, pH 8.8, 50 mM KCl, 100 pig/ml. gelatin, 0.5 mM
MgC12, 0.5 mM MnClz, 0.4 mM dATP, 3.2 mM dCTP, 3.2 mM dGTP, and 3.2 mM dTTP
(all ﬁnal concentrations).

The VP16C PCR products were subcloned into SalI and Bng digested pPJH13 tO
produce a library Of Gal4-PCR product fusions. This pool of pPJH13 variants was grown
in E. coli and DNA was isolated by plasmid preparation using CsClz gradients. Plasmid
DNA was transformed into yeast strain BF] and cells were plated onto Leu' media for
selection of pPJH13 transformants. Candidate yeast colonies were chosen based on visual
inspection of colony size. Large colonies were patched onto Leu' media to verify a normal

growth phenotype relative to yeast expressing wild-type VP16C fusions. Colonies that

84

 

passed the second screen were lysed mechanically by vortexing colonies suspended in 10
uL of phenol-chloroform containing approximately a 10 uL volume of glass beads. Yeast
DNA (1-2 uL) isolated in this manner was then transformed into E. coli. DNA isolated
from these E. coli clones was sequenced to identify activation domain mutations. TO
assess the transcriptional phenotype of each mutant, candidate VP16C activation domains
were subcloned into low copy yeast expression vectors (ARS/CEN vectors). Activity was
quantitated by cotransfection of these Gal4-VP16C ﬁrsions with the plasmid pLGSDS
(Guarente, L. et al., 1982). Transformed yeast were grown on solid media selective for
both plasmids. Three individual transformants were selected and grown in liquid selective
medium. Following growth to mid-log (OD 0.6 to 0.8), yeast were harvested, cell-free
extracts were prepared, and B-galactosidase assays were performed colorimetrically as

described by Miller (Miller, J .H., 1972).

Mopse L cell transfection

All mouse L cell transfections were performed using a standard DEAE-dextran
technique. Cells (106 cells) were seeded and incubated overnight in DMEM media
(GibcoBRL) plus 10% Fetal Bovine Serum (FBS; Atlanta Biologicals) at 37°C in a 10%
CO; environment on P60 tissue culture plates to allow attachment. The resulting cultures
displayed ~60- 75% conﬂuence. All DNA used was puriﬁed by ultracentrifugation over
CSC12 gradients twice, and quantitated by ﬂuorescence using a Hoeffer miniﬂuorimeter
and calf thymus DNA as a standard. This procedure eliminates any uncertainty due to
residual contaminating RNA. DNA was aliquoted into 3 mL polypropylene tubes and

resuspended in 1.5 mL ofDMEM/I-IEPES (DMEM containing 20 mM HEPES-NaOH pH

85

7.2), plus 200 ug/mL DEAE-dextran. The DMEM was aspirated from each P60 plate,
and the cell monolayers were rinsed with 2 mL of DMEM/HEPES. The DNA suspensions
were mixed by inversion several times and overlaid onto the cells. Following a 5 hour
incubation at 37°C, 10% C02, plates were aspirated and overlaid with 1 mL of 1x HBS
containing 10% DMSO. Following a 3 minute DMSO shock, cells were rinsed with 2 mL
DMEM/HEPES. After rinsing, cells were grown 2 days in DMEM/HEPES containing

10% PB S, before harvesting and analysis.

Chloramphenicol Acetyltransferase (CAT) Assays

Cell monolayers were washed with 5 mL of PBS and overlaid with 0.5 mL of PBS.
Cells were detached by scraping with a rubber policeman. The cell suspension was
removed to a microﬁrge tube, and the P60 was rinsed a second time with another 0.5 mL
of PBS to collect residual cells. Cells were spun at 4K for 1 minute in an Eppendorf
microfuge. The pellet was resuspended in 150 uL of 0.25M Tris-HCl, pH 7.8. The cell
suspension was lysed by repeated freeze thawing. Cells were frozen in a dry ice/ethanol
bath for 5 minutes. The frozen cell suspension was then thawed at 37°C for 5 minutes.
This freeze-thaw cycle was repeated three times. Cell debris was removed by
centriﬁrgation at 14K in an Eppendorf microfuge at 4°C for 10 minutes. Ten microliters
of this clariﬁed extract was used for protein estimation by a modiﬁed Bradford method
(Stoscheck, C.M., 1990).

Extracts were treated at 65°C for 10 minutes to remove endogenous acetyl-COA
metabolizing enzymes, followed by centriﬁrgation in a microﬁrge at 14K for 10 minutes at

4°C to remove debris. Fifteen microliters Of lmg/mL BSA in water was added to each

86

 

supernatant. Activity was assayed in 10 mL scintillation vials by a continuous organic
extraction as described (Nielsen, D.A. et al., 1989). Lysates were pre-incubated with
chloramphenicol, followed by initiation Of the reaction with 3H-acetyl-COA. Typically, a
60 uL volume Of lysate was incubated with 20 uL of 5 mM chloramphenicol (1 mM ﬁnal
concentration) at room temperature. Reactions were initiated at 30 second intervals with
20 uL osz-acetyl-COA (20 uCi/mL; 150 uM) diluted in 75 M HCl. Vials were counted
1 9 seconds at 30 minute and hour intervals. T 0 ensure that CAT enzymatic initial

velocities were obtained, only data displaying approximately twice the cpm at the longer

time point were retained for analysis.

87

 

Chapter Three

Distinct Activation Domains Function through Distinct Mechanisms

Introduction

The mutational studies outlined in the previous chapter established several
common characteristics critical for the ﬁmction of the two subdomains of VP16 and the
potentially related subdomains of RelA, highlighting the importance Of numerous
hydrophobic residues throughout these activation domains. However, the VP16
subdomain and RelA subdomains display Slightly different structural requirements; one
acidic residue immediately following the most critical phenylalanine residue was Observed
to be important for VP16N and VP16C ﬁmction. This feature contrasts with the RelA
subdomains. In RelA activation domain sequences, the critical F474 residue of RelAII is
followed by a glutamine residue which is insensitive to alteration, and the critical F542
residue of RelAIH is followed by a serine residue.

These results imply that activation domains are relatively similar in their structural
requirements, which may result from functional similarity. However, an early Observation
suggested that VP16N and VP16C might be functionally distinct. A Gal4-VP16C chimera
is a robust activator in yeast cells, but displays only weak activity in mammalian cells. In
contrast, a Gal4-VP16N chimera is a very potent activator in mammalian cells and
relatively weak in yeast cells. This inverse relationship suggests that VP16N and VP16C
may be functionally distinct. Perhaps different subdomains facilitate different steps in PIC
assembly and differentially impact transcription in these species. This possibility seemed

even more attractive given the Observed speciﬁcity of TAF interactions with the two

88

subdomains of VP16. The TAF S are essential and sufficient for transcriptional activation
in vitro (Dynlacht, B.D. et al., 1991). Furthermore, the histone-like TAFS are essential for
transcriptional activation in viva in yeast (Michel, B. et al., 1998). The VP16C
subdomain, and not the VP16N subdomain, speciﬁcally interacts with the histone-related
TAF S, human TAFn32 (Klemm, R. et al., 1995) and Drosophila TAFn40 (Goodrich, J .A.
et al., 1993). This result suggests that VP16C and not VP16N may participate in TAF-
mediated activation, and emphasizes the potential for distinct mechanistic ﬁmctions of the
individual subdomains. Given these Observations, it would have been satisfying to have '
observed distinct structural features for VP16N and VP16C. Perhaps these domains are in
fact structurally and functionally redundant. Alternatively, the subdomains are ﬁmctionally
distinct and the minor structural variations observed are sufficient to confer this
subdomain speciﬁcity.

To examine these questions, we sought to more closely examine the possibility that
VP16N and VP16C might be mechanistically distinct. We investigated the activities of
these two activation domains using three methods. The results suggest that VP16N and
VP16C function by distinct mechanisms. First, VP16N and not VP16C was capable of
activating transcription from an integrated template in mammalian cells. Second, VP16C
requires a speciﬁc promoter architecture for function. In contrast, the VP16N subdomain
is insensitive to core promoter structure. Third, VP16C and not VP16N facilitates
recruitment Of TFIID to a promoter in a TFIIA dependent fashion in vitro.

We have also examined the activation domain of the NF-xB family member RelA.
These structural investigations verify that two subdomains of RelA are critically dependent

upon a cluster of hydrophobic amino acids for function. These critical sequences resemble

89

 

VP16C with regard to the pattern of hydrophobic residues within the motif (Fxx‘I’T,
where ‘1’ denotes a hydrophobic amino acid). To determine if this shared motif is
indicative of ﬁrnctional similarity, we assessed RelA function by the promoter speciﬁcity
and DA shift assays. The RelAII and RelAIII subdomains displayed identical core
promoter speciﬁcity to that Observed with VP16C. However, in contrast to expectations,
the RelA activation domain was incapable of facilitating TFIID recruitment in vitro. To
further investigate the hypothesis that the Fxx‘IJ‘P motif confers speciﬁc function, we
produced a VP16N mutant that contained the VP16C-like Fxx‘P‘P motif by deletion of
two VP16N residues. This Gal4-VP16N mutant was incapable Of facilitating DA
recruitment to a promoter in vitro. These results suggest that the core motif itself is

insufﬁcient to confer speciﬁc VP16C-like function to an activation domain.

Experimental Results

VP16N and not VPl6gctivaftes transcription from an integrated reporter

Genetic analyses identiﬁed a collection of yeast mutant genes that suppress the
toxicity of overexpression of Gal4-VP16 chimeras in yeast (Berger, S.L. et al., 1992).
These genes (which include ADA2, ADA3, GCN5 and ADA5) are believed to encode
adapter molecules for AD function. The identiﬁcation of the GCN5 gene product as a
histone acetyltransferase (HAT) complex suggested that the ADA genes may be involved
in the modulation of gene expression via chromatin modiﬁcation (Brownell, J .E. et al.,
1996). Indeed, these gene products associate in viva with other proteins to form a high

molecular weight complex, which displays activator-sensitive HAT activity in vitro (Grant,

90

RA. et al., 1997). This observation suggests that one mechanism by which VP16 might
function is through acetylation of nucleosomes by recruitment of this complex.

While transiently transfected DNA is chromatin packaged, it displays aberrant
micrococcal nuclease digestion patterns, suggesting a structure distinct from native
chromatin (Jeong, S. and Stein, A., 1994). The transcriptional activity of this abberant
chromatin may be considerably different than that observed for native chromatin. In fact,
clear differences between the activity of integrated and transiently transfected promoters
have been Observed (reviewed in Smith, CL. and GL. Hager, 1997). Conceivably,
differences Observed in the capacity of activators to function on integrated and non-
integrated reporters might reﬂect mechanistic differences between these activators. We
compared the ability of Gal4-VP16 subdomain fusions to activate both transfected and
integrated CAT reporters. The cell line NIHZO8, derived from NIH 3T3 cells, contains a
stable integrant of the reporter plasmid pGSBCAT. The pGSBCAT reporter contains ﬁve
Gal4 binding sites and the adenovirus Elb T ATA sequences upstream of the bacterial
CAT gene. A Gal4-VP16 ﬁrsion expression vector (N, C, or FL) was cotransfected into
NIH3T3 or NIH208 cells with 1 ug of pGSBCAT reporter or carrier DNA, respectively.
The results appear in Figure 7. While Gal4-VP16FL, Gal4-VP16N and Gal4-VP16C
displayed similar activity in the parental N1H3 T3 cell line, only Gal4-VP16FL and Gal4-
VP16N displayed signiﬁcant activity in NIHZO8 cells (VP16C displayed only 6% of the
activity of VP16N). We conclude that VP16N and not VP16C is capable of activating the
integrated reporter. This suggests that VP16N and not VP16C employs a mechanism(s)

capable of efﬁciently antagonizing the repressive effects of chromatin.

91

nmol Acetyl-coA conv./h/mg

 

 

 

NIH208 NIH3 T3

Figure 7. Activity of VP16 activation domains on an integrated reporter. Gal4-VP16
expression vectors (100 ng) encoding fusions of VP16N, VP16C or VP16FL activation
domain sequences were transfected with 1 ug of carrier DNA (pBR322) into NIH208
cells (which contain an integrated pGSBCAT reporter gene) or with 1 ug of pGSBCAT
reporter plasmid into the parental NH-I 3T3 cell line. The reporter construct contains ﬁve
Gal4 sites upstream of an Elb TATA box and the bacterial chloramphenicol
acetyltransferase (CAT) gene. CAT activity was determined by a ﬂuor diffusion method
using 3H-acetyl-COA and unlabelled chloramphenicol, and is expressed as nmol acetyl-
CoA converted per hour, per mg cell extract. Bars (open bars, VP16N; solid bars,
VP16C; grey bars, VP16FL) represent means plus or minus standard deviations of three
transfections. The left set represent results using the NIH 208 cell line. NIH 3T3 results
are represented by the right set of bars.

92

The VP16C activation domain is sensitive to bas_al pronloter structure

 

Another potential distinction between the two subdomains of VP16 was suggested
by results from the Smale and Gralla laboratories. Smale and coworkers (Emami, K.H. et
al., 1995) assessed the ability of Gal4-VP16N and Gal4-Spl fusions to activate
transcription from reporter constructs that varied only in their basal promoter elements.
Gal4-VP16N displayed 12-fold higher activation from a TATA+Inr containing construct
relative to an Inr-only construct. In contrast, a Gal4-Spl ﬁrsion displayed Similar activity
using either reporter construct. The authors suggest that this behavior reﬂects diﬁ‘erences
in the mechanisms of the two activation domains. In contrast, using a Gal4-VP16FL
chimeric activator, Chang and Gralla (Chang, C. and Gralla, J .D., 1993) reported only 2-3
fold greater transactivation of a TATA+Inr promoter relative to an Inr-only reporter.
These relative activities are more reminiscent ofGal4-Sp1 behavior than the Gal4-VP16N
behavior Observed by Smale and coworkers. This difference in basal promoter preference
exhibited by Gal4-VP16FL could be attributable to an additional Inr-speciﬁc activity
contributed by the VP16C subdomain. These published observations suggest that VP16N
and VP16C may display different basal promoter speciﬁcity.

We tested the promoter speciﬁcity of the VP16N and VP16C subdomains using
the plasmids constructed by the Smale laboratory (Emami, K.H. et al., 1995). The
reporter plasmids contain ﬁve Gal4 sites immediately upstream of two different basal
promoters. The TATA-containing promoter was originally constructed by insertion of a
synthetic Oligonucleotide containing the adenovirus major late promoter TATA sequences.
The Inr-containing plasmid was constructed using a synthetic Oligonucleotide bearing the

Inr sequence of the terminal deoxynucleotidyltransferase gene. The promoter sequences

93

 

are fused to an HSV thymidine kinase (tk)- glyCOproteinH gene ﬁ'agment. Activity was
assessed by primer extension using an HSV tk speciﬁc primer. The Inr-reporter produces
a product of 84 nucleotides, while the TATA-reporter produces a 75 nucleotide product.
The vectors also contain SV40 origins of replication as well as the polyoma early
transcript unit, allowing replication in a wide range of cell lines. Our analyses were
performed in a mouse L tk' cell line.

To examine the promoter speciﬁcity of the VP16N, VP16C, and VP16FL
activation domains, expression vectors encoding Gal4-activation domain ﬁrsions were
cotransfected with either the TATA or Inr reporter plasmid. The results are depicted in
Figure 8A. Transfection with the Inr reporter resulted in two extra primer extension
products longer than those predicted to arise from promoter-speciﬁc initiation. It should
be noted that these reporter constructs are derived from an autonomously replicating
polyoma derivative and contain polyoma enhancer sequences. We believe this set of larger
extension products originates from partial extension of an upstream transcript, likely to
originate in polyoma sequences. Primer extension through this region appears to be
difﬁcult resulting in premature termination by AMV reverse transcriptase. This problem
occasionally partially obscures the Inr signal. Perhaps RNase or 81 protection might
produce better results.

Despite this difficulty, an activator-speciﬁc primer extension product can be
resolved (compare the no activator control, lane 6, with the activator containing
cotransfections, lanes 7-10). The Gal4-VP16C transfection produced detectable products
only when cotransfected with the Inr reporter plasmid (compare lanes 3 and 5, to lanes 8

and 10). In contrast, the Gal4-VP16N and Gal4-VP16FL transfections produced robust

94

 

 

activity with either the TATA (lanes 2 and 4) or Inr reporter plasmid (lanes 7 and 9).
Thus, the VP16C subdomain demonstrates Inr-promoter speciﬁc function. In contrast,
VP16N and the full-length VP16 activation domain are insensitive to basal promoter
composition (compare lanes 2 and 4, to lanes 7 and 9). These differences might suggest
differences in ﬁrnction between the activation domains. However, diﬁ‘erences in
transfection efficiency could also provide a simple explanation for the results observed.

To eliminate any complications due to inconsistencies between transfections, we
examined the promoter preference of each activation domain when simultaneously
cotransfected with both reporter plasmids. Gal4-fusions of VP16N, VP16C, and VP16FL
were cotransfected with both the TATA and Int-reporters. Simultaneous cotransfection
with both reporters allows direct comparison of the relative strengths of the two reporters,
since their extension products differ in length. The results are shown in Figure 8B. The
lower arrow marks the TATA-only extension product. The upper arrow marks the Inr-
only product. The results demonstrate clearly that the VP16C fusion protein activates
only the Inr-reporter (see lanes 8-11). In contrast, both VP16N and VP16FL activate
either promoter construct efficiently (see lanes 3-6 and lanes 13-16).

This result could be due to a differential requirement for accumulation of a
sufﬁcient amount of mRNA product to Obtain primer extension signal, with a threshold
requirement that was higher for the TATA reporter. If this were the case, then a weak
activator would display apparent promoter speciﬁcity due to its inability to sufﬁciently
stimulate transcription past the threshold level for the TATA reporter. To test for such
potential threshold phenomena, we varied the quantity of activator expression vector in

the transfections. In each case, expression vector quantities were titrated from conditions

95

 

8:8:0: 4&0:

-8: .8 w: : :5 <>_0:-<H<H 05 .8 w: _ A: 054 5:0 :0t0:0: 2T<H<H .8 w: 2 ”m 05.: 5:0 :0t0:0: 5-8: 05 .8 m1 ~ ”v 054
.2885: 8-8: .8 m1 : :5 <>_0:-<H<H 05 .8 w: _ “m 054 .2885: 8:098: .8 w: : :5 xv<H<H 2: .8 ma: Mm 054 .8283:
03:5 :0 m: g :5 292.5; 2: .8 mi: H: 05.: .2858 85088 Angst/-350 2:. .8 w: _ 5:580 80:00:88: =< 80:00:08
:0>t0: 25> 2:030: =5 25— 885.: <30: 50m 4.2098: :5 <>_0:-<H<H £02538: :0:0_0:-9:0b0: J5:E:800: :05

:5 m0:0w 808:0: 23:: :5 2T<H<H 2: :23 game 86:00.00 88.0: .8 838800 .0 :05: .28 8:8:0: 2T<H<H .8 ma:
:5 882, 85088 Hagar/-350 .8 w: 3 85:00 38-8: .980 :0t0:0: xt<H<H .8 m: g :5 880> 85088 ESE/450 .8
010 :88 $5.55 0228 oz .30: 2 02 so: 01 m e5 .N ._ .30 ._ .8 0202430 5 .20.: e :2 so: .0: 8m :5
.02 .2: .8 .o: 52.5-28 .20: 9 02 so: 0: om Pa .2 .m .N .0 02.3-38 .20: 8 :2 so: .01 m Pa .N ._ .8: .2 .8
2292-350 .8535 0 Z ”Emu 0:. 0:0— 80: 85.: 508555: E 552:5 2 8:00.55: :50 :_ 285—: 8:300:20 5:00.55:

.8 >850: 05. 228:0: 03-8: :5 2T<H<H 2: .8 £00 .8 w: _ 5:5 2800:8356 880.5558 53 Am05_ .8 8m :50 085 0:0:
2: 3 8865 a :3228 8 52.3400 02.5-38 £2,548 2:2 8:52: 8:885 555.: :22: Peace
853005-306 .8 08: 86:08:: :< 808590 080:8: 089 .m _0::.— .2 05: 285—: 850820 E0M-Ee.0 0 .8 w: m 8

.3: 05: 285—: 85088 Auger/-300 0 .8 w: m AU 05: 285—: 8:88:00 0:75-350 0 .8 m1 m AZ 05: 285—: 850820
GEN/-300 5 .8 m: m .8 05c 8528: 0: 5:5 880.5558 53 Am05_ .8 “0.0. Ewt ”m1 3 :0t0:0: 8H 8 @05— .8 :00 0:0— mm:

C :0t0:0: <H<H 2: 855 .< .055 0.00:8: 86:38:28.: :0880::.<H<:. 0:609. 05 88:0: _05: :50 :_ 30:5 030— 05.
80:8: 86:38 08:: 88:88:: 0:605 05 88:0: _05: :50 :_ B05 :0::: 05. .305 5 .3 :08:0: 2 3:80:55 08005
.8 828: 05 05: :50 E 058885: 0208: >32 :5 08:: 0:609: 0: 5 m5? 8:80:00 5:8: .3 :0880: 0:03 32:85.:
0m_00:m .:0:00:m5b-:8: 200: m: 2.00 :080mm5b 80:: :0505: 53 <72 2:5 .320: 352:5 5v 885—: 008:0: A05 05::—
05253: 08:88-8: 8:5 <H<H 5:,» 2.00 A 082: 8:: 380.550 003 20:03 85088 82:48 555:: £58033.
:50 .8 00:80:08 88:88 05 0:580: 0 H .mﬁaEeBa 50m :5 of?» 3:23.05 :0 56500:: 20850.5 .a 0.5m;

96

nix—0m

 

omvmmﬁ

Hm: : m>

.U

 

 

 

 

88> 88> .
m
:duzo :dozo
8 <8:
.4

5288

97

17 ng yTBP 33 ng yTBP

TFHA TFHA

 

free probe

on »
mun ,

_ A '3‘ ,
k, 3 s. 2‘
‘1-

 

Figure 9. TFIIA stimulates TBP recognition of the AdML promoter TATA
box. To test the activity of recombinant TFIIA, TFIIA was incubated with two
quantities (17 ng, lanes 2-7; 33 ng, lanes 8-13) of yeast TBP (yTBP). Under these
electrophoretic conditions (see Methods) yTBP does not associate with the AdML
TATA box in the absence of TFIIA (lanes 2 and 8). Addition of TFIIA results in
stable binding to the TATA box in a TFIIA dose dependent fashion (see lanes 3-7
and 9-13). TFIIA alone does not bind DNA (lane 14).

98

 

The polyoma enhancer produces not only the technical problem of upstream start sites,
but also a possible scientiﬁc complexity. The two VP16 subdomains may differ not in
promoter recognition, but rather in their ability to synergize with the polyoma enhancer.
To test the contribution of polyoma sequences, we removed all of the polyoma sequences
from the reporter constructs. Reporters bearing or lacking polyoma sequences were
cotransfected into mouse L cells. The polyoma bearing reporters were included as internal
controls for transfection and primer extension. In each transfection, a Gal4-VP16FL
expression vector served as activator (VP16FL does not display basal promoter
sensitivity; see above). The results appear in Figure 8C. In all cases, primer extension
signals were evident only from the polyoma bearing reporter plasmids. No primer
extension products were detected from cotransfected plasmids lacking the polyoma
enhancer. These results conﬁrm that polyoma sequences are essential to observe
detectable signal from these synthetic reporters. Either these results reﬂect a direct effect
of basal promoter structure on activator function, or they reﬂect a basal promoter effect
on activator-enhancer synergy. Regardless, the primer extension results suggest

differences in the mechanism of function of the two subdomains

The VP16C subdomain recruits TFIID and TFHA to a promoter in vitro

A number of observations suggest a role for TFHA in response to activators
(reviewed in introduction). Berk and coworkers have developed an in vitro TFIID
electrophoretic mobility shift assay (EMSA) system. The size of the TFHD complex
precludes the use of traditional polyacrylamide EMSA systems, since TFIID-DNA

complexes do not migrate from the loading wells in these systems. To circumvent this

99

 

difﬁculty, Lieberman et al. performed EMSA in a 1.4% agarose gel in the presence of
Mg2+ to stabilize TFIID-DNA complexes (Lieberman, RM. and A.J. Berk, 1994). In this
EMSA system, TFHA dramatically enhances the apparent afﬁnity of TFIID for DNA. The
TFHA polypeptides appear to enhance the stability of TFIID on DNA by forming a
TFIID-TFIlA-DNA complex (DA complex). Results from the Berk laboratory suggest
that certain activators can also stimulate formation of this DA complex. The Epstein-Barr
Virus activator Zta was the ﬁrst activator demonstrated to enhance DA complex
formation (Lieberman, P., 1994). Gal4-VP16 fusions can also stimulate DA complex
formation (Kobayashi, N. et al., 1995; Kobayashi, N. et al., 1998). This activity appears
to localize to the VP16C subdomain, since the VP16C subdomain strongly stimulates DA
complex formation while the VP16N subdomain has no effect (see below and Kobayashi,
N. et al., 1998).

We wished to examine the capacity of the VP16 and RelA subdomains to stimulate
DA complex formation. We ﬁrst needed to establish appropriate conditions for the
EMSA experiments. Human TFHA, human epitope-tagged TFIID (eTFIID), and Gal4-
activation domain ﬁJsions were prepared as outlined in Methods. The TFHA was assayed
for the ability to enhance the affinity of recombinant yeast TBP for the AdML promoter in
a polyacrylamide EMSA system (Figure 9). Radiolabeled AdML promoter DNA was
incubated with yeast TBP (yTBP) in the absence (lane 2, 17 ng yTBP and lane 8, 33 ng of
yTBP) or presence of an increasing concentration of recombinant human TFIIA (lanes 3-7
and lanes 9-13). Under these electrophoretic conditions, DNA-binding by yTBP is
completely dependent upon TFHA (Maldonado, E. et al., 1990). The TFHA preparation

efﬁciently stimulated yTBP recognition of the promoter.

100

Gal4-VP16N and Gal4-VP16C ﬁrsions were prepared as described in Methods,
and tested for DNA-binding activity. The EMSA probe used for this and all subsequent
EMSA experiments is a 225 bp restriction fragment of pG5E4TCAT containing ﬁve Gal4
sites and the adenovirus E4T promoter upstream of a segment of the bacterial CAT gene.
Performance of the agarose EMSA experiments requires prior determination of the
quantity of each Gal4-fusion protein necessary to saturate all ﬁve Gal4 binding sites. This
was determined using a conventional polyacrylamide EMSA system. An example of such
an experiment appears in Figure 10. Conditions saturating for all ﬁve sites in this
experiment were used to determine appropriate Gal4-fusion quantities for the agarose

EMSA experiments.

To test the capacity of each VP16 subdomain to stimulate DA complex formation,
we performed an agarose EMSA (Figure 11). Radiolabeled DNA was incubated with
immunopuriﬁed TFIID and recombinant TFIIA in the absence or presence of Gal4-VP16N
or Gal4-VP16C (one, two, or four times the amount of protein necessary to saturate the
probe). DA complexes were resolved by electrophoresis in a Mg” containing
electrophoresis buffer on a 1.4% (w/v) agarose gel followed by autoradiography. A small
degree of DA complex formation was apparent even in the absence of activator (lane 3).
In the presence of Gal4-VP16C, the extent of this DA complex formation was signiﬁcantly
enhanced (lanes 5-7). However in the presence of Gal4-VP16N, DA complex formation
was actually inhibited (lanes 11-13). In each case, formation of these complexes was
completely TFIIA dependent (see lanes 4, 9, and 15). These results agree with those

previously observed by Kobayashi, et al. (Kobayashi, N. et al., 1995).

101

Gal4-VP16N Gal4-VP16C

free probe

  

Figure 10. Saturation of the EMSA probe with Gal4-AD fusion proteins. Gal4-
VP16N or Gal4-VP16C fusion proteins were incubated with radiolabeled adenovirus
E4T DNA under conditions identical to those employed in the agarose EMSA
experiments, followed by electrophoresis on 4% PAGE in leBE. Lane 1: free probe.
Lanes 2-9; increasing concentration of the Gal4-VP16N fusion protein. Lanes 10-18;
increasing concentration of the Gal4-VP16C fusion protein. The probe contains ﬁve
Gal4 binding sites. Species with 1,2,3,4 or 5 occupied sites are each resolved.

VP16C VP16N VP16NAD/L

A AAAA AAAA AAAA
D DDD D DDD D DDD D

 

Figure 11. Comparison of the capacity of VP16N, VP16C and a mutant VP16N
subdomain to enhance DA complex formation. Recombinant Gal4-VP16C (lanes 4-9),
Gal4-VP16N (lanes 10-15), or Gal4-VP16NAD/L (lanes 11-16) were incubated with a
radiolabeled adenovirus E4T promoter probe, in the presence or absence of TFIIA (A; 20
ng) and/or TFIID (D; 1 uL). Following incubation, complexes were separated on a 1.4%
agarose gel in a Mg—containing TBE buffer. The Gal4-fusions were present at a
concentration as indicated by the ﬁlled triangles above each set of lanes (0.5 times, 1
times, or 2 times the concentration required to saturate the Gal4 sites under identical
incubation conditions). TFIIA and TFIID without activator (lane 3) weakly associated
with the probe. Addition of VP16C (lanes 5-7), but not VP16N (1 1-13) or VP16NAD/L
fusions (17-19), enhanced DA complex formation. In all cases, eﬁicient DA complex
formation required both TFIIA (lanes 2, 9, 15 and 21 lack TFIIA) and TFIID (lanes 8, 14
and 20 lack TFIID).

l03

In collaboration with Arnold Berk and Naoko Kobayashi, we assessed a series of VP16C
mutants for the ability to stimulate DA complex formation. Using a collection of our
VP16C mutants, Naoko Kobayashi performed DA complex agarose EMSA assays. The
VP16C mutants were chosen to display a range of transcriptional strength from nearly
wild type to completely defective as assayed in yeast and mammalian cells in vivo. The
results appear in Figure 12. The VP16C mutants are arranged in order left to right, from
strongest to weakest (as determined by the in vivo transcription assays). The in vivo
transcriptional activities of the mutants correspond well with their ability to stimulate DA
complex formation in vitro.

We further tested the ability of wild type and mutant VP16C to bind directly to
TFIIA. Gal4-VP16C proteins were in vitro translated, incubated with myc-tagged TFIIA
and coimmunoprecipitated using an anti-myc epitope MAb. The results demonstrated a
direct correlation between the ability of the VP16C mutants to bind TFHA, and their
transcriptional activity (see Figure 12). In contrast, coimmunoprecipitation experiments
performed with TAFn3 2, TBP, TFIIB and the p65 subunit of TFIIH did not display this
correlation. These results suggest that VP16C stimulates DA complex formation through
recruitment of TFHA to an assembling PIC. Furthermore, the VP16N activation domain
does not stimulate DA complex formation, demonstrating that VP16N and VP16C can

ﬁinction through distinct mechanisms in vitro.

Expression of VP16N or VP16C activation domains inhibits transcription in vivo

To determine whether or not the VP16 subdomains function through.

similar mechanisms in vivo, we also employed a strategy originally used by Chambon

104

 

 

Figure 12. Gal4-VP16C stimulates formation of a TFIID-promoter complex through
recruitment of TFHA to DNA. Panel A. An agarose EMSA experiment was performed
using Gal4-fusions of wild type VP16C and several VP16C mutants. Gal4-VP16C fusions
were incubated for 20 minutes at 30°C with radiolabeled adenovirus E4T promoter probe
in the presence of TFIIA and/or TFIID (as indicated above each lane). Following the
incubation, complexes were separated on a 1.4% agarose gel using a Mg-containing TBE
buffer. The mutants are arranged in descending order of in vivo transcriptional potency,
ﬁ'om left to right. The positions of free probe and various complexes are indicated at the
right of the ﬁgure. Panel-B. Phosphor Imager quantitation of the image of panel A.
Mutants: C1, D472A; C2, Q477A; C3, F479A; C5, F473 S/E474G; C6,

F47 5S/M478T/F479S; C7, F473A/F475A; C8, F479L; C9, F479S/D486G. Panel C.
Interaction of TFIIA with Gal4-VP16C mutants. The mutants are arranged in descending
order of in viva transcriptional potency, from left to right. Gal4-VP16C (or the indicated
mutant) was incubated in the presence (+, lanes 3, 6, 9, 12, 15, 18, 21, 24, and 27) or
absence (-, lanes 2, 5, 8, 11, 14, 17, 20, 23, and 26) of Myc-epitope tagged TFIIA.
Following incubation, monoclonal 9E10 (a Myc-epitope monoclonal) was used to
immunoprecipitate TFIIA. Co-irnmunoprecipitated material was subjected to SDS-PAGE.
Bound 3SS-Met labeled Gal4-VP16C was quantiﬁed by Phosphorimager, using 5% of

input Gal4-VP16C (lanes 1) as a reference. Panel D. Numerical quantitation of the

autoradiogram in Panel C. Means and standard deviations of three experiments are
shown.

(Figures obtained from Kobayashi, et al., 1998)

105

 

 

eTFIlD_+++++++++++++—+—
A. rTFllA--+++++++++++-—+—
G4VP160 — — —wrcs C2C1CSC3 C9c7csw-wrwrwr-

 
 
 

Q (—|ID+IIA+G4VP16C
. (lIDHIA
‘ * <—G4VP160

”LufFreeprobe
12 3 4 5 6 7 8 91011121314151617

B 100-

80‘
60‘

 

%Wild type
N A
o 9

 

 

WT CB C2 C1 CS C3 CS C7 CS

 

C 5‘” “”50 WT CB C2 C1 C5 C3 C9 C7 CG
. I — +Il — +I“-\._. +ll| >— +lll — +ll| — +Ill — +lll — +lll — +1
“‘ ~ and... “Masai “Can... a . ..,... w p.

12 3 4 5 6 7 8 9101112131415161718192021222324252627

D. 120

1 00
80

60

96 Wild type

40

20

 

F‘ 12 °
lgure ' WT C8 C2 C1 C5 C3 C9 C7 C6

106

and coworkers (Tasset, D. et al., 1990) to analyze the glucocorticoid receptor. An in vivo
competition strategy was employed to assess the mechanistic similarity of activation
domains from the glucocorticoid receptor (GR) and estrogen receptor (ER). Activation
domains lacking DNA binding domain sequences were overexpressed as chimeric proteins
containing N-terminal fusions of the SV40 T antigen nuclear localization sequence (NLS)
to activation domain sequences. They reasoned that if these constructs are still capable of
recognition of their nuclear target yet lack DNA-binding domains, then their expression
should result in the progressive titration of the target from nuclear pools. This titration
should prevent activation by any activator that utilizes the same target protein. Using this
strategy, Tasset, et al. argued that the N-terminal activation domains of the GR and ER
function through the same mechanism. The strategy has been employed in several other
cases (Schmitz, M. et al., 1995; Carruth, L.M., et al., 1994; De Felice, M. et al., 1995;
Folkers, G.E. et al., 1995). Using plasmids obtained from Chambon, we produced
mammalian expression vectors for NLS-VP16N (pNLS-VP16N) and NLS-VP16C
(pNLS-VP16C) chimeric proteins (“competitors”) to test their ability to interfere with the
activity of either Gal4-VP16N or Gal4-VP16C (“eﬁ‘ectors”). An example of the results of
an experiment using Gal4-VP16C is shown in Figure 13A. Competitor and effector DNAs
were cotransfected into mouse L cells by the DEAE-dextran method. Cotransfections
were performed with a Gal4-VP16C effector (100 ng), varied competitor DNA quantities
(as indicated in the ﬁgure legend) and 1 ug of the pGST+I/CAT reporter construct
(described previously; see Chapter 2). In each transfection, DNA levels were maintained

constant by the use of pSGS as carrier DNA. The pSGS plasmid is the parental DNA

107

 

 

Figure 13. In viva competition experiments with NLS-VP16 and Gal4-VP16
expression vectors. Panel A. Mouse L cells were cotransfected with 100 ng of a Gal4-
VP16C expression vector, 1 ug of a reporter construct which expresses chloramphenicol
acetyltransferase (CAT) from a composite TATA + Inr promoter downstream of ﬁve Gal4
binding sites, and an increasing dose (0 to 10 ug) of one of two constructs expressing
either the VP16N (set of bars on the left) or VP16C activation domains (bars on right)
fused to a nuclear localization sequence. The total DNA quantity transfected was held
constant by inclusion of the parental NLS-fusion expression vector pSGS as a carrier
DNA. Bars represent the average of three separate transfections plus or minus the
standard deviation. CAT activity was determined by a ﬂoor diffusion method, using 3H-
acetyl-CoA and unlabelled chloramphenicol. The activity is expressed as arbitrary units
normalized to the control transfection (0 ug of competitor and 10 ug of pSGS). Black
bars: 0 ug of competitor and 10 ug of pSGS. White bars: 1 ug of competitor and 9 pg of
pSGS. Grey bars: 5 ug of competitor and 5 ug of pSGS. Right cross-hatched bars: 10 ug
of competitor and 0 ug of pSGS. Panel B. same as Panel A, except that the experiment
was performed with 100 ng of a Gal4-VP16N expression vector.

108

 

 
   

NLS-VP16N NLS-VP16C
NLS-Competitor

nmol acetylcoA conv./h/mg

 

NLS-VP16N NLS-VP16C
NLS—Competitor

Figure 13.

109

 

of the competitor plasmids, and contains an SV40 enhancer upstream of an empty multiple
cloning site. This polylinker is the site of insertion of activation domain sequences in the
competitor constructs. While 1 pg of pNLS-VP16C was sufficient to inhibit Gal4-VP16C
activity 65%, the same quantity of pNLS-VP16N had little observable effect (<6%
inhibition). Thus speciﬁcity, as evidenced by the greater effect of the pNLS-VP16C
competitor on Gal4-VP16C, was observed at low levels of competitor DNA. However, at
higher quantities of DNA, both competitors inhibited Gal4-VP16C activity (compare 10
ug bars, Figure 13A). Results from a similar experiment performed with a Gal4-VP16N
effector DNA displayed different results. Inhibition was not observed at competitor levels
below 5 ug, and at 5 ug both competitors inhibited VP16N activity slightly (see Figure
13B). If competitOr levels were increased to 10 ug, pNLS-VP16N displayed greater
inhibition than pNLS-VP16C.

These results are consistent with partially independent mechanisms of activation by
VP16N and VP16C. The speciﬁc effect of the NLS-VP16C competitor on Gal4-VP16C
activity at low competitor dose (Figure 13A) suggests that the VP16C competitor titrates
a limiting factor essential for VP16C ﬁrnction. Furthermore, the absence of an effect of
the VP16N competitor does not appear to result from a lack of expression of the NLS-
VP16N ﬁision, since the VP16N competitor is capable of inhibition of Gal4-VP16C at
higher concentrations (Figure 13B). The inhibition of Gal4-VP16C observed by both
competitors at high competitor DNA levels, could be explained by either of two models.
The activation domains might contact multiple targets and share at least one target. This
hypothetical common target would have low afﬁnity for VP16N and high affinity for

VP16C. Alternatively, the inhibition observed at higher levels of competitor could be non-

110

 

 

speciﬁc due to the presence of the strong SV40 enhancer/promoter on the competitor
plasmids resulting in lowered expression of the chimeric activator and reduced
transactivation of the reporter construct. This second model points out a potential
limitation in the experimental strategy. A primary assumption is that cotransfection of
each of these DNA constructs results in coexpression of these constructs in each
transfected cell. Another critical assumption is that alteration of the amount of competitor
DNA transfected results in a corresponding alteration of competitor protein synthesized in
each cell. Both assumptions are difﬁcult to verify. Thus, while these results suggest that
VP16N and VP16C utilize distinct cellular targets, clear interpretation of these results

remains complex.

We RelA spbdomains mechanistically similar to VP16C?

The mutagenesis studies described in the previous chapter suggest that VP16C and
RelA subdomains H and III share similar structural requirements for function.
Speciﬁcally, both contain a similar sequence motif of critical residues: Fxx‘IAI’ (where F
denotes phenylalanine and ‘1’ denotes a hydrophobic amino acid). This structural similarity
raises the possibility that VP16C and RelA may be functionally similar. To examine this
possibility we examined some of the ﬁinctional characteristics of the RelA activation
domain. We chose to examine RelA using two assays previously used to examine VP16N

and VP16C.

The RelA activation domain displays Inr promoter preference

111

 

 

 

The VP16C subdomain displays core promoter speciﬁc function. Gal4-VP16N
efficiently activates promoters containing either TATA or Inr elements in viva. In
contrast, Gal4-VP16C activates only Inr containing promoters. We tested the sensitivity
of the RelAII and RelAIII subdomains to promoter structure, as described for previous
experiments with Gal4-VP16 fusions. A Gal4-RelAII expression vector was cotransfected
with either the TATA or Inr-containing reporter plasmid. The Gal4-RelAII results are
shown in Figure 8A. The RelAII fusion displayed transcriptional activity only when
cotransfected with the Inr reporter plasmid (lane 10). No activity was detected when the
TATA reporter plasmid was used (lane 5). We also tested the ability of a Gal4-RelAHI
ﬁision protein to activate the reporter constructs. A Gal4-RelAIII expression vector was
cotransfected with both the TATA and Inr reporter plasmids simultaneously. The result of
the cotransfection experiments appears in Figure 8B. The Gal4-RelAIII ﬁision protein
directs Inr transcription preferentially, as was observed with VP16C fusions. These results
argue that the RelAII, RelAIII and VP16C subdomains share similar mechanism(s) of

activation.

RelA does not enhance DA recruitment to a promoter in vitro

The previous results suggest that the RelA and VP16C activation domains may
function at least in part through similar mechanisms. In addition to the promoter
speciﬁcity observed in primer extension experiments, the VP16C subdomain also enhances
DA complex formation. To ﬁrrther test for functional similarities between VP16C and the
RelA activation domain, we evaluated RelA for this DA stabilization activity. A plasmid

was constructed that encodes a fusion protein containing a full-length RelA activation

112

 

 

domain (amino acids 417-551) fused to a minimal Gal4 DNA binding domain (amino acids
1-92). The full length activation domain was chosen for two reasons. First, use of this
fusion protein allows evaluation of both VP16C-like domains without the preparation of
two ﬁision proteins. Second, the strength of the intact domain is likely to be greater than
the individual subdomains, improving the potential signal strength. This eﬁ'ect was
previously observed with VP16. Despite the fact that VP16N does not stimulate DA
complex formation, the full length VP16 activation domain enhances DA complex
formation more than VP16C alone (N aoko Kobayashi, personal communication).

Epitope puriﬁed TFIID (0.5 to 2 uL) was incubated with recombinant TFHA
either in the presence or absence of a recombinant Gal4-RelA fusion protein, and
separated on an agarose gel. The results are shown in Figure 14. As seen previously,
some DA complex was observed in the absence of activator (lane 4). The presence of
Gal4-RelA reduced the extent of DA complex formation under all TFIID conditions
examined (see lanes 7-9). This phenotype is common with activators that do not stimulate
DA complex formation, as noted previously for Gal4-VP16N (Figure 11). In contrast, a
Gal4-VP16C fusion protein included as a positive control signiﬁcantly enhanced the extent
of DA complex formation (lane 12). As observed previously, formation of the DA
complex is TFHA dependent (see lanes 5, 10 and 11). These results strongly argue that

RelA does not stimulate DA complex formation.

Can alteration of the hydrophobic core of VP16N reconstitute VP16C activity?

Although the RelA activation domain contains two subdomains, RelAII and

113

 

RelAIII, that share a critical sequence motif (Fxxww) with VP16C, the RelA activation
domain does not stabilize DA complex formation. This result implies that the Fxx‘P‘I’
motif is insufﬁcient to confer this in vitro activity to an activation domain. To directly test
this hypothesis, we altered VP16N sequences such that the subdomain contained a
sequence with this motif. Codons specifying Asp 441 and Leu 444 of VP16N were
deleted, producing a primary structure that resembles VP16C with respect to the
positioning of the acidic and hydrophobic residues within the sequence. For reference, the
sequences of the core regions of VP16N, VP16C, and the altered subdomain
(VP16NAD/L) appear in Figure 15. We assayed a Gal4-VP16NAD/L ﬁrsion protein for
the ability to stabilize DA complex formation. As outlined above for the RelA agarose
EMSA assay, epitope puriﬁed TFIID (0.5 to 2 1.1L) was incubated with recombinant
TFHA either in the presence or absence of a recombinant Gal4-VP16NAD/L ﬁrsion
protein, followed by separation on an agarose gel. The results are shown in Figure 16. As
observed previously, high TFIID levels allowed some DA complex formation even
without activator (see lanes 2-4). The presence of Gal4-VP16NAD/L reduced the extent
of DA complex formation (see lanes 7-9). Under identical conditions, a Gal4-VP16C
fusion protein enhanced DA complex formation (lane 12). Formation of the DA complex
was in all cases, TFIIA dependent (see lanes 5 and 10). These results conﬁrm the
conclusions suggested by the RelA DA complex experiments. The presence of the

F xx‘P‘I’ primary structural element does not confer the capacity to stabilize formation of a

ternary DA promoter complex in vitro.

114

 

RelA VP16C
a, D D D
g A A.-
0.

AAA AAA A

 

 

Figure 14. The RelA activation domain does not stimulate DA complex formation.
Gal4-RelA (lanes 6-10) or Gal4-VP16C (lanes 11-13) was incubated in the presence of
TFIID and/or TFIIA (as indicated by the text D or A; 20 ng) with a radiolabeled
adenovirus E4T promoter fragment. Samples were separated on 1.4% agarose gel in a
Mg—containing buffer. TFIID quantities were varied (0.5 11L, 1 uL, and 2 uL) as
indicated by the ﬁlled triangles. Two microliters of TFIID and 20 ng of TFIIA readily
formed DA complex in the absence of activator (lane 4). The presence of Gal4-VP16C
enhanced the formation of this complex (lane 12). In contrast, Gal4-RelA reduced the
extent of DA complex formation (lanes 7-9). In all cases, TFIIA was required to observe
signiﬁcant complex formation (see lanes 5, 10 and 11).

 

VP16N:

(Amino acids 439 to 447)

LDDFDLDML

VP16C:

(Amino acids 470 to 479)

MADFEFEQMF

VPl 6N AD/ L:
(Amino acids 439 to 447)

LDFDDML

Figure 15. Deletion of VP16N amino acids D441 and L444 produces a VP16C-like
core motif. The amino acid sequences of VP16N and VP16C are represented by single-
letter amino acid code. Codons 441 and 444 were deleted from the VP16N core sequence
to produce the VP16N mutant VP16NAD/L (bottom line). This core sequence resembles
the VP16C core sequence, but contains VP16N sequences ﬂanking the altered core
sequence.

116

 

VP16AD/L
D D

 

C

D
(D
'8“-
5 A

 

Figure 16. A VP16N mutant, VP16NAD/L,does not stimulate DA complex formation.
Gal4-VP16NAD/L (lanes 6-11) or Gal4-VP16C (lanes 12) was incubated in the presence
of TFIID and/or TFIIA (as indicated by the text D or A; 20 ng) with a radiolabeled
adenovirus E4T promoter fragment. Samples were separated on 1.4% agarose gel in a
Mg-containing TBE buffer. TFIID quantities were varied (0.5 uL, l uL, and 2 uL) as
indicated by the ﬁlled triangles. Two microliters of TFIID and 20 ng of TFIIA readily
formed DA complex in the absence of activator (lane 4). The presence of Gal4-VP16C
enhanced the formation of this complex (lane 12). In contrast, Gal4-VP16NAD/L
reduced the extent of DA complex formation (lanes 7-9). In all cases, TFIIA was
required to observe signiﬁcant complex formation (see lanes 5 and 10). TFIIA alone did
not produce an observable shift (lane 11).

ll7

Discussion and Conclusions

The results presented in the previous chapter highlight the critical features of the
primary structures of the VP16N and VP16C subdomains, identifying similar features
essential for ﬁinction. However, minor differences in activation domain structure were
also observed. Speciﬁcally, the spacing of hydrophobic residues about the central critical
phenylalanine residue in the two subdomains differ. Sequence and mutational analysis of
RelA identiﬁed a similar motif in two segments of the activation domain. These motifs
within RelA more closely resemble VP16C than VP16N suggesting a possible ﬁinctional
parallel between RelA and VP16C. Sequence analysis and mutational results identiﬁed a
motif (Fxx‘P‘I’) within VP16C, RelAII and RelAIII that is essential for function. We were
interested in determining whether or not the subtle structural differences in the VP16N
and VP16C subdomains were paralleled by differences in ﬁmction. Does this difference,
however subtle, confer distinguishable ﬁmction to each subdomain? Alternatively, the
VP16N and VP16C subdomains may be functionally redundant. The structural differences
observed between the two subdomains may simply reﬂect an inherent ﬂexibility in the
structural requirements for activation domain ﬁinction.

As an initial means to assess the possibility that VP16N and VP16C activate
through distinct mechanisms, we tested the relative activity of VP16N and VP16C in the
cell line NII-1208, which contains an integrated pGSBCAT reporter. The VP16N but not
the VP16C fusion proteins displayed robust activity in NIH208 cells. In contrast, both
activators strongly stimulated transcription of a cotransfected pGSBCAT reporter in the
parental NIH3 T3 cell line. This result suggests a difference in the capacity of VP16N and

VP16C to eﬁiciently antagonize the repressive effects of chromatin in viva.

118

 

We also examined the capability of VP16N and VP16C to stimulate transcription
from two reporters with different basal promoter architecture. Results from the Smale
(Emami, K.H. et al., 1995) and Gralla (Chang, C. and JD. Gralla, 1993) laboratories
implied that the VP16N and VP16C subdomains differ in their capacity to mediate
transcriptional activation through different promoter elements. Analysis of VP16N and
VP16C subdomains by cotransfection of two reporters containing either a TATA element
or Inr element conﬁrmed this inference. While either activator efficiently transactivated
the Inr-containing reporter, only VP16N activated the TATA reporter. These results are
also consistent with the idea that VP16N and VP16C mediate transcriptional activation by
distinct mechanisms.

Another result reinforces this conclusion. While neither previous observation
suggests a speciﬁc mechanism for either subdomain, experiments performed in
collaboration with Berk and coworkers suggest a mechanism for VP16C in vitro
(Kobayashi, N. et al., 1998). Lieberman et a1. developed an in vitro assay for TFIID
recruitment (Lieberman, PM. and A.J. Berk, 1994). This assay employs recombinant
TFIIA and immunoafﬁnity puriﬁed native TFIID complex. Formation of the DA complex
in this assay system is highly TFIIA dependent and can be stimulated by the Epstein-Barr
virus activator Zta (Lieberman, P., 1994; Kobayashi, N. et al., 1995) and HSV VP16. The
VP16 activity has been localized to the VP16C subdomain (Kobayashi, N. et al., 1995).
Using a panel of our VP16C mutants, Kobayashi et al. (Kobayashi, N. et al., 1998)
established a correlation between activation potential in viva and DA complex formation
in vitro. A correlation was also observed between the capacity to stimulate DA complex

formation and TFHA binding by these mutants. These experiments suggest that VP16C

119

 

activates transcription through TFIID recruitment to promoters via a TFIIA dependent
mechanism. Furthermore, the correlation between the in viva and in vitro phenotypes of
the VP16C mutants implies that this mechanism may be signiﬁcant in viva.

This result is also interesting with regard to the observation that VP16C displays
Inr-containing promoter speciﬁcity. The in vitro EMSA results establish that VP16C is
capable of recruiting TFIID to a promoter in a TFIIA dependent fashion. Since the
primary sequence speciﬁc DNA-binding subunit of TFIID is TBP, removal of TATA
sequences would be expected to markedly reduce TFIID affinity for TATA-less
promoters. An activator that recruits TFIIA could stabilize TBP interaction with non-
TATA sequences by facilitating a TATA-like interaction of TBP with non—TATA
sequences. If the VP16C subdomain ﬁmctions by recruitment of TFHA, this could explain
the preference for Inr-containing promoters. In the presence of a TATA box, TFIIA
activity may be of lesser importance due to the presence of already strong TBP-DNA
interactions. This would result in activators that recruit TFHA having a reduced effect on
PIC stability. This model would predict that TATA-containing promoters do not require
TFHA and are not activated by TFHA-recruiting activators. In contrast, Inr-containing
factors, limited by their aﬁinity for TFIID, would be highly dependent upon TFHA and
TFHA-recruiting activators.

The DA complex assay employs an agarose gel EMSA separation of nucleoprotein
complexes in the presence of Mg”. This assay system appears to be particularly sensitive
to the stability of the DA complex, as prolonged electrophoresis appears to dissociate the
complex signiﬁcantly. This suggests that some activators may function by stabilizing a

TFIID-TFIIA-DNA complex, rather than increasing the rate of TFIID recruitment. While

120

 

the precise nature of this stabilized complex is unknown, it is interesting to note that
Chambon and coworkers (White, J. et al., 1992) implicated an early step in PIC formation
as a step stimulated by VP16. This "template-committed" complex was sufﬁcient to
enhance transcription even after blockage of the VP16 activation domain by a monoclonal
antibody. Carey and colleagues (Chi, T. and M. Carey, 1996) have also suggested that
formation of an altered TFIID-DNA complex, as evidenced by an extended DNAseI
footprint, correlates with the enhancement of activation provided by Zta. This event also
requires TFIIA. It is inviting to speculate that all three of these observations may be
related. Regardless of the identity of the EMSA stabilized DA complex, the assay
provides a USBﬁJl means to determine the ﬁmctional similarity of activation domains.

We utilized the DA assay to assess the ﬁrnction of the subdomains of VP16 as well
as the RelA activation domain. The VP16C and not the VP16N subdomain efﬁciently
stimulates DA complex formation. The RelA activation domain contains two sequence
segments that contain a VP16C-like Fxx‘P‘P motif. When a Gal4-RelA full length
activation domain was tested for DA complex stabilization, the activator actually reduced
DA complex formation. This phenotype is common for activators that do not stimulate
DA complex formation, and is also exhibited by a Gal4-VP16N fusion protein. This result
demonstrates that the Fxx‘P‘I’ motifs present in the RelA activation domain are not
sufficient to confer DA complex stabilization activity to the activation domain.

Another observation supports the conclusion that the Fxx‘P‘P motif is not a
suﬁicient determinant for DA complex stabilization. A VP16N mutant was designed to
simulate the primary structure of the VP16C subdomain. Deletion of D441 and L444 of

VP16N produces a VP16N mutant (VP16AD/L) with a primary structure that resembles

121

 

 

 

the Fxx‘P‘I’ motif of VP16C (see Figure 15 for sequences). The Gal4-VP16NAD/L fusion
protein did not enhance DA complex stability. This result conﬁrms that the Fxx‘I’T motif
is not suﬁicient to confer DA complex stabilization activity. Collectively, these results
suggest that while the Fxx‘P‘P motif may be necessary for transcriptional activity, it is not
sufﬁcient for specifying the DA complex stimulatory activity.

Although the RelA activation domain did not stimulate DA complex formation, the
RelAIII subdomain did resemble VP16C with regard to the promoter speciﬁcity test.

Both Gal4-VP16C and Gal4-RelAIII ﬁisions directed transcription from an Inr-containing
promoter, and not from a TATA-containing promoter in viva. In contrast, Gal4-VP16N
stimulated transcription from either an Inr-containing or TATA-containing promoter. The
lack of a correlation between the promoter speciﬁcity results and DA complex results is
surprising. The capability of the VP16C subdomain to stabilize TFIID recognition of a
promoter via TFHA recruitment seems to present a reasonable explanation for the
observed Inr speciﬁcity (see above). The observed Inr-speciﬁcity and lack of DA complex
stabilization activity observed with RelA suggests that these two activities may not be
related.

This conclusion is reinforced by inspection of the Adenovirus E4 promoter (the
promoter used for the DA complex EMSA studies). The AdE4 promoter contains a
thymidine-rich tract of nucleotides at the initiation site. This sequence does not resemble
the TdT Inr sequence (Smale, ST, 1997). Since the DA complex EMSA assay is
performed using an Int-lacking promoter, it is unlikely that this activity can be described
as Inr-speciﬁc. Consistent with this objection, Smale and colleagues failed to demonstrate

TFIID recognition of a TdT Inr promoter using this DA complex EMSA system (Emami,

122

 

 

K.H. et al., 1997). However, these studies were performed in the absence of activators.
It remains possible that an activator will facilitate TFIID recognition of an Inr containing
promoter in vitro.

Another line of investigation suggests a link between Inr function, TAFS, and
TFIIA. The Tjian laboratory obtained evidence that TFHA may facilitate promoter
speciﬁc function in a TAF dependent fashion (Hansen, SK. and Tjian, R., 1995). The
Drosophila Adh promoter is transcribed from two tandem promoters. These promoters
are differentially utilized temporally throughout development. This promoter speciﬁc
function can be reconstituted in vitro from appropriate stage embryos. The stage speciﬁc
function of the distal promoter is Inr element dependent. In vitro this Inr-mediated
speciﬁcity is dependent upon TAFs and TFIIA (Hansen, SK. and R. Tjian, 1995). These
results suggest that Inr function is mediated by TFHA through TAFs

In summary, our experiments lead to the following conclusions. The VP16N and
VP16C subdomains display functional properties both in viva and in vitro that suggest
distinct mechanisms of ﬁinction. These include distinct basal promoter architecture
preferences, differential performance on an integrated reporter, and a differential capacity
to stabilize TFIID promoter recognition through TFIIA recruitment. The RelA activation
domain, which contains two Fxx‘I’ﬂ’ motifs similar to VP16C (one in RelAH and one in
RelAIII), displays the same promoter speciﬁcity as VP16C. Despite this common
promoter-speciﬁcity, the RelA activation domain lacks the ability to stimulate DA complex
formation, an observation inconsistent with the supposed similarity of function between

VP16C and RelA. Furthermore, a VP16N mutant (VP16NAD/L) which contains a

FxxW motif similar to VP16C does not direct DA complex stabilization. These results

123

 

suggest that while the identiﬁed F xx‘IJ‘I’ motif may be necessary for activity, it is

insufﬁcient to confer speciﬁc ﬁmction to an activation domain.

Experimental Methods

Recombinant DNA Manipulations

The pNLS-VP16N fusion vector was produced by modiﬁcation of the plasmid
VP16(N) previously constructed by Tasset et al. (Tasset, D. et al., 1990). This vector
contains VP16 410-490 sequences. To delete VP16C sequences, the plasmid was cut
with Smal and KpnI. The digested DNA was blunted with T4 DNA polymerase, followed
by agarose gel electrophoresis and ligation of the plasmid DNA to recircularize with T4
DNA ligase. pNLS-VP16C was prepared from VP16(N) by deletion of VP16 410-452
sequences. The VP16(N) plasmid was digested with AvaI. AvaI cleaves both at the SV40
NLS junction as well as at the VP16N-VP16C boundary (VP16 codon 453). The 5'
overhangs were ﬁlled with Klenow enzyme. Plasmid DNA was puriﬁed by agarose gel
electrophoresis and recircularized with T4 DNA ligase. The resulting constructs express
either an NLS-VP16N or NLS-VP16C fusion from the SV40 early enhancer.

The alternate promoter reporter constructs were obtained from Steven Smale
(UCLA). They consist of synthetic Oligonucleotides with appropriate TATA or Inr
sequences inserted upstream of HSV thymidine kinase (tk) sequences. The plasmids also
contain an SV40 minimal origin of replication and a polyoma fragment which encompasses
the entire early transcript unit. To remove these sequences, we excised the entire polyoma

coding region by BamI-II digestion and religated the linearized plasmid.

124

 

The VP16NAD/L E. coli expression plasmid, pG4(1-92)VP16NAD/L, was
constructed by PCR using a modiﬁcation of the three primer method used to construct
RelA point mutants (see Methods, Chapter 2). An activation domain PCR product
lacking L444 was ﬁrst constructed using primer ST-27 7 that contains activation domain
coding sequences lacking a L444 codon. This DNA was ampliﬁed from pSGVPN using
ST-277 and a second primer ST-246, which hybridizes within SV40 polyadenylation
sequences 3' of the VP16 coding sequences in pSGVPN. The product was gel puriﬁed
and reampliﬁed using ST-246 and ST-278, an Oligonucleotide containing VP16 coding
sequences, but lacking both codon D441 and codon L444. This PCR product was
agarose gel puriﬁed and used as a maxiprimer for ampliﬁcation from the E. coli expression
vector pJGRl 10-1 using PCR primers ST-246 (SV40 primer) and ST-281, a primer which
contains Gal4 sequences upstream of the pJGRl 10-1 cloning junction with VP16. This
PCR converted the PCR product to a Gal4(1-92)-VP16 chimeric sequence. The PCR
product was digested with XhaI in the Gal4 coding sequences, and Bell within polylinker
sequences at the 3' end of VP16 sequences in pSGVPN. The fragment was introduced
into XhaI-BamHI cut pVP16CX, an E. coli Gal4-VP16C expression vector, and
sequenced to verify the codon 441 and 444 deletions.

The Gal4(1-92)-RelA(417-551) expression plasmid (pG4(1-92)-RelAFL) was
constructed in several steps. The activation domain (amino acids 417-551) was ampliﬁed
from pSGp65(307-551) (a gift of Paul Moore and Brian Cullen; Moore, RA. et al., 1993)
by PCR using Oligonucleotides ST-205 and ST-206. The ST-205 Oligonucleotide contains
a BamHI site placed in frame with pSG424 polylinker and RelA sequences. The ST-206

Oligonucleotide ampliﬁes from outside of the XbaI cloning site in pSGp65(307-551). This

125

 

PCR product was ligated into the EcaRV polylinker site of pB S/KS+ to generate the
plasmid pBSp65(417-551). The pBSp65(417-551) BamI-II-XbaI excised activation
domain sequences were then subcloned into BamI-II and XbaI cut pSGp65(307-551) to
generate the plasmid pPJH65(417-551), a mammalian expression vector. The E. coli
expression vector was produced from pPJH65(417-551) and pJL2. pPJH65(417-551)
was linearized with BamHI and the 5'-overhang was blunted by treatment with Klenow
enzyme. The activation domain sequences were then excised with XbaI and gel puriﬁed
by agarose electrophoresis. The E. coli expression vector pJL2 was cut with HpaI and
XbaI, excising VP16 and HSV tk sequences, and the RelA activation domain sequences

were introduced by ligation to generate pG4(1-92)-RelAFL.

Tissue Culture Cell Manipulations

All mouse L cell transfections were performed by a standard DEAE-dextran
technique as described in Chapter Two. All CAT assays were also performed as described
in Chapter Two. NIH 208 and NIH 3T3 cell transfections were performed by a
Lipofectarrrine (GibcoBRL) technique as advised by the manufacturer. In a P60 tissue
culture plate, cells were seeded at 2x105 cells per plate. Cells were grown in DMEM
media plus 10% PBS at 37°C in a 10% C02 environment to approximately 60%-80-%
conﬂuence. DNA (see Figure legends) was suspended in 100 uL of DMEM in a 1.5 mL
microfuge tube. Lipofectamine (4 uL per transfection) was suspended in DMEM (100 uL
per transfection). A volume of 100 uL of the Lipofectamine/DMEM suspension was
transferred to each DNA suspension, followed by mixing of the contents by ﬂicking of the

microfuge tube several times with a ﬁnger. These suspensions were incubated at room

126

 

 

temperature approximately 30 minutes, followed by addition of 0.8 mL of serum-free
DMEM. Cell cultures were rinsed with 2 mL of serum-ﬂee DMEM, and overlaid with the
diluted DNA-lipid complexes. Cell cultures were incubated at 37 °C/ 10% C02 for ﬁve
hours followed by addition of 1 mL of DMEM containing 20% FBS. Lipofectamine
containing media was removed following grth overnight and replaced with fresh
DMEM containing 10% FBS. Cultures were harvested and analyzed after an additional 2

days of growth.

RNA Isolation

RNA from transfected monolayers was isolated using TRI Reagent (Molecular
Research Center, Inc.) following the manufacturer’s instructions with the exception that
volumes throughout were reduced two-fold for use in P60 tissue culture plates.
Monolayers were aspirated to remove media. A 500 uL aliquot of TRI reagent was
pipetted into each plate. The TRI Reagent was pipetted repeatedly onto the monolayer,
until cell disruption was observed. Treated monolayers ﬁrst displayed a glassy
appearance, and were subsequently observed to detach. Detached cells were transferred
to a microfuge tube. Tubes were either stored at -80°C for subsequent use or immediately
lysed by incubated at room temperature for 5 minutes. Frozen cells were thawed at room
temperature and incubated an additional 15 minutes to be certain cellular dissociation had
occurred. A 100 uL volume of chloroform was added to each lysate. Tubes were
vortexed 15 seconds and centrifuged at 14K in an Eppendorf ﬁxed angle microﬁlge for 15
minutes at 4°C. The upper, aqueous phase of ~280 uL was collected from each sample,

and 500 uL of isopropanol was added to each tube. Tubes were incubated at room

127

 

 

temperature ﬁve to ten minutes and RNA was precipitated by centriﬁigation in a ﬁxed
angle microfuge at 14K for 8 minutes at 4°C. Isolated RNA was washed with 80%

ethanol and dissolved in 20 uL of RNAse free TE buffer.

Primer Extension Analysis

Preparation of the radiolabeled probe was performed using T4 DNA kinase. A 10
pmol aliquot of Oligonucleotide ST-136 was dried in a Speed-vac apparatus to remove
residual acetonitrile. The Oligonucleotide was dissolved in 7 uL of water and 30 uCi 32P-
y-adenosine-triphosphate (8000 Ci/mmol; New England Nuclear; Amersham), 10 units of
T4 kinase (GibcoBRL), and 3 uL of 5x Forward Reaction Buffer (GibcoBRL) were added
(15 [IL total volume). The solution was incubated at 37°C for 1 hour. After incubation,
the reaction was stopped by addition of 2 DL of 0.5 M EDTA and 50 uL of TE buffer. A
DE-52 (Whatman) column was prepared in a 1 mL pipette tip. A 20 uL volume of AG
50W-X8 (BioRad) cation exchange (100-200 mesh) resin was poured onto a glass wool
plug as a support matrix for the DE-52. Approximately 50 uL (bed volume) of DE-52
(Whatman) was overlaid and the column was equilibrated with 10 mL of TEN100 (10 mM

Tris-HCl, pH 8, 1 mM EDTA, 100 mM NaCl). The labeled DNA was applied to the

 

column and ﬂowthrough was reloaded onto the column a second time. The column was
then washed with 1 mL of TEN100 twice, followed by 0.5 mL of TEN300 (TENlOO
buffer with 300 mM NaCl). Oligonucleotide was eluted with 400 uL of TEN600

(TEN100 with 600 mM NaCl).

128

 

RNA was analyzed by primer extension analysis. Each RNA sample (10 uL) was
hybridized with radiolabeled Oligonucleotide ST-136 (3.5 uL; prepared as above) in a 15
uL ﬁnal volume of 10 mM Tris-HCl, pH 8.3, containing 150 mM KCl, and 1 mM EDTA.
These samples were immersed in a 65°C water bath. After 1 hour, samples were removed
and allowed to cool to room temperature (30 minutes-1 hour). Samples were extended
with AMV Reverse Transcriptase (Life Sciences, Inc., St. Petersburg, FL) in a 45 uL
reaction containing (ﬁnal concentrations) 20 mM Tris-HCl, pH 8.3, 10 mM MgC12, 6 mM
DTT, 150 mg/mL actinomycin D (Boehringer—Mannheim), 0.3 mM each of dATP, dCTP,
dTTP and dGTP, and 5 units of AMV reverse transcriptase. Extensions were performed
for 1 hour at 42°C. Following extension, 100 uL of an RNAse solution (20 ug/mL
RNAse A, 0.1 mg/mL herring sperm DNA) was added to each extension. RNAse
incubations were continued at 37°C for 15 minutes. A 15 uL volume of 3M
NaCH3COOH, pH 5.2 was added to each sample, followed by phenol/chloroform
extraction and ethanol precipitation of the ﬁnal products. Precipitated DNA was washed
with 70% ethanol, air dried ten minutes and dissolved in 10 uL of loading buﬂ‘er (0.05%
(w/v) bromophenol blue, 0.05% (w/v) xylene cyanol, 20 mM EDTA, pH 8 dissolved in
formamide). Samples were separated on a 9% PAGE gel containing 7M Urea for 1 hour
and 50 minutes at a constant power of 60 Watts. Gels were dried and visualized by

autoradiography.

Preparation of Gal4-Fusion Proteins
Gal4-VP16 ﬁlsion proteins (VP16N, VP16C, and VP16NAD/L ﬁisions) were

prepared by selective precipitation and ion exchange chromatographies. Cultures (4 x IL

129

 

 

cultures) of E. coli transformants of pJGRl 10-1 (Gal4-VP16N), pJGR81-ASma (Gal4-
VP16C), or pVP16NXAD/L (Gal4-VP16NAD/L) expression vectors were grown on LB
medium containing 80 ug/mL ampicillin at 37°C to an OD500 of 0.7. Expression was
induced by addition of 240 mg of IPTG per liter of culture. One mL of 0.1 M
Zn(CH3COO)2 was also added to supply sufficient Zn” for Gal4 DNA binding domain
coordination. After an additional 3 hours of grth at 37°C, cells were harvested by
centrifugation at 5,000 rpm in a Sorvall GSA rotor for 10 minutes. Supernatant was
decanted and cells were either immediately lysed or stored at -80°C for subsequent use.
Lysis was performed by sonication. Cells were resuspended in 160 mL of Buffer
A (20 mM HEPES, pH 7.5, 10 DM Zn(CH3COO)2, 20 mM B-mercaptoethanol)
containing 200 mM NaCl, and protease inhibitors (2 ug/mL Pepstatin A, 2 ug/mL
Leupeptin, 1 mM benzamidine and 1 mM phenylmethylsulfonylﬂuoride). Cells were
sonicated at 4°C at ﬁill power with a ﬂat tip probe in 30 second bursts with 1 minute rest
between each cycle. Sonication was repeated until greater than 90% of all cells were lysed
as determined by comparison of OD600 before and after sonication. The extract was
clariﬁed by centriﬁigation at 10,000 rpm in a Sorvall GSA rotor for 30 minutes (all
subsequent centrifugation steps were performed at 4°C unless noted). The supernatant
was retained and clariﬁcation was repeated if necessary. The volume was measured and
polyethyleneimine (PEI) was added to a ﬁnal concentration of 0.3% (v/v) while stirring at
4°C. The extract was stirred 15 minutes at 4°C. The PEI and associated material was
precipitated by centrifugation at 10,000 rpm in a GSA rotor for 15 minutes. The

supernatant was discarded and the PEI pellet was resuspended in 100 mL of Buffer A

130

 

 

containing 200 mM NaCl and protease inhibitors using a rubber policeman. The washed
PEI pellet was reprecipitated from this suspension by centrifugation at 10,000 rpm in a
GSA rotor for 15 minutes. This pellet was then resuspended in 100 mL of Buﬁ‘er A
containing 750 mM NaCl and protease inhibitors to elute the fusion protein. The PEI was
pelleted at 10,000 rpm in a GSA rotor, and the supernatant containing the Gal4-VP16
fusions was retained. Ammonium sulfate was added to 35% saturation while stirring at
4°C. This solution was stirred one hour to overnight. Protein was precipitated in a
Sorvall SS-34 rotor at 10,000 rpm for 30 minutes. Precipitated protein was dissolved in
approximately 30 mL of Buffer A containing 200 mM NaCl and protease inhibitors.

The dissolved protein fraction was dialyzed against SCB (20 mM HEPES, pH 7.5,
10 M Zn(CH3COO)2, 1 mM DTT) containing 0.1 M NaCl. The protein ﬁaction was
loaded at a ﬂow rate of 10 mL/hr onto a Whatman P-ll phosphocellulose column (bed
volume approximately 10 mL) equilibrated in SCB containing 0.1 M NaCl. Following
loading, the column was washed overnight at 20 mL/hr with SCB + 0.1 M NaCl. The
protein was eluted at a ﬂow rate of 30 mL/hr with a linear gradient (4x column volume)
from 0.1 to 1.0 M NaCl in SCB. Fractions containing Gal4-fusions were identiﬁed by
SDS-PAGE and Western blot using a Gal4-VP16 reactive rabbit antiserum, LA2-3.

Gal4-RelA fUSiOIIS were further puriﬁed by Heparin-aﬁnity and DEAE-ion
exchange chromatography. Protein ﬁ'actions eluted from the phosphocellulose column
were dialyzed against Buffer SCB containing 0.1 M NaCl. Protein was loaded at a ﬂow
rate of 15 mL/hr onto a 4 mL Heparin-agarose column equilibrated with SCB containing
0.1 M NaCl, and washed with the same buffer. Protein was eluted with a linear NaCl

gradient (4x column volume) from 0.1 M to 1.0 M NaCl in Buffer SCB. Samples were

131

 

 

 

analyzed by SDS-PAGE. Samples that contained Gal4-RelA were pooled for subsequent
chromatography. The pool was dialyzed against Buffer SzCB (20 mM HEPES, pH 7.9,
10 M Zn(CH3COO)2, 1 mM DTT) containing 0.1 M NaCl. The protein was loaded at a
ﬂow rate of 10 mL/hr onto a 3 mL Whatman DE-52 column equilibrated with SzCB
containing 0.1 M NaCl. The column was washed with the same buffer and eluted using a
linear gradient (4x column volume) of 0.1 to 1.0M NaCl in Buﬁ‘er $2CB. Samples were

analyzed by SDS-PAGE. Aliquots were frozen in liquid nitrogen and stored at -80°C.

Preparation of eTFIID

The TFIID was epitope puriﬁed by irnmunopuriﬁcation from a stable cell line
prepared for this purpose. The cell line (LTRoc3) expresses a TBP gene with an N-
terminal ﬁision of the inﬂuenza virus Hemagluttinin (HA) epitope. This small synthetic tag
allows recognition of the TBP fusion by the monoclonal antibody 12CA5. The
immunopuriﬁed TFIID contains an apparently full compliment of TAFs (Zhou, Q. et al.,
1993)

The LTRor3 cell line was grown in SMEM (Gibco/BRL) + 1% glucose,
supplemented with 1x Pen/ Strep (Gibco/BRL) in spinner ﬂasks. Cells were counted using
a hemacytometer and daily diluted to 4 x 107 cells/ mL of media. LTRa3 cells were
harvested at a volume of 15-30 liters by centrifugation at 2,000 rpm in a swinging bucket
rotor for 10 minutes. Following harvest, cell pellets were pooled and pelleted in a conical
bottom ﬂask to measure cell pellet volume. The LTRor3 pellet was suspended in ﬁve
times this pellet volume of phosphate buffered saline (PBS). PBS washed cells were

centrifuged at 2,000 rpm and the PBS was discarded. The cell pellet was resuspended in

132

 

 

 

ﬁve volumes of Buffer A (10 mM HEPES pH 7.9, 1.5 mM MgC12, 10 mM KCl, 10 mM
B-mercaptoethanol). The cell suspension was incubated at 4°C for 10 minutes, followed
by centrifugation at 2,000 rpm for 10 minutes. At this stage the cell pellet swelled
approximately 2-fold in volume. The Buffer A was aspirated and LTRor3 cells were
suspended in 2 pellet volumes (Qﬂginal pellet volume) of fresh Buffer A. The cell pellet
was lysed by Dounce homogenization using a Kontes all glass homogenizer with a B type
pestle. Lysis was conﬁrmed by microscopy. Crude nuclei were harvested by
centriﬁigation at 25,000 x g in an SS-34 rotor for 20 min. at 4°C. A 5/3 volume (original
cell pellet volume) of Buffer C (20 mM HEPES pH 7.9, 25% glycerol, 0.42M NaCl, 1.5
mM MgC12, 0.2 mM EDTA, 0.5 mM PMSF, 10 mM B-mercaptoethanol) was added to
the crude nuclei pellet. The pellet (at this stage gelatinous) was resuspended by Dounce
homogenization 10 times. Following resuspension, the sample was nutated 30 min. at
4°C, followed by clariﬁcation at 25,000 x g in an SS-34 rotor for 30 min. at 4°C. This
extract was loaded onto a phosphocellulose P-ll column (~1 mL bed volume/ 20 mg of
total protein) equilibrated with buffer D100 (20 mM HEPES pH 7.9, 0.2 mM EDTA, 100
mM KCl, 20% glycerol, 10 mM B-mercaptoethanol). The P-ll column was washed with
2 column bed volumes of buffer D300 (D100 with 300 mM KCl), followed by 5 column
volumes of D500 (D100 with 500 mM KCl). The phosphocellulose “D fraction” was
eluted with ~1.5 column volumes of D1000 (D100 with 1 M KCl). Two milliliter
fractions of eluate were collected. The eTFIID containing fractions were identiﬁed by
Western blot using the Hemaglutinin (HA)-epitope monoclonal 12CA5.

Monoclonal 12CA5 coupled Protein A-Sepharose beads for immunoprecipitation

were prepared as follows. Dry Protein A- Sepharose CL—4B (Pharrnacia; 1.5 g) was

133

 

swollen by addition of 15 mL of water. The CL—4B media was washed four times by
exchange of ~50 mL of fresh water. Swollen Protein A- Sepharose CL-4B was stored as
a 50% slurry in PBS plus 20% (v/v) ethanol until subsequent use. For coupling, two
milliliters of the 50% slurry was removed and decanted into a 15 mL conical plastic tube.
The Sepharose was pelleted at 4000 rpm in a table top centriﬁige for 5 minutes followed
by removal of supernatant. The Protein A-Sepharose then was washed with 14 mL of
fresh PBS, followed by centriﬁigation at 4,000 rpm for 5 minutes. The PBS was
aspirated, and the wash procedure was repeated four times. A 3 .5 mL volume of
monoclonal antibody 12CA5 (Hemagluttinin antibody; 2 mg/mL in PBS) was added to the
Protein A-Sepharose and the slurry was incubated at room temperature for two hours with
gentle rocking. The Sepharose was pelleted at 4,000 rpm for 5 minutes and the
supernatant was removed. Beads were washed with 20 mL of 0.2 M sodium borate, pH
9.0, followed by centrifugation at 4,000 rpm for 10 minutes. The supernatant was
decanted and the beads were resuspended in 20 mL of 0.2 M sodium borate, pH 9.0
containing 0.104 g of dimethylpimelimidate (20 mM ﬁnal concentration). This slurry was
rocked at room temperature for 30 minutes to allow crosslinking of the 12CA5 antibody
to Protein A. Beads were pelleted at 4,000 rpm for 10 minutes and washed with 20 mL of
0.2 M ethanolamine, pH 8.0 followed by a repeat of the centrifugation. The supernatant
was discarded and the 12CA5-Protein A-Sepharose beads were resuspended in 20 mL of
0.2 M ethanolamine and gently rocked at room temperature for 2 hours to stop the cross-
linking reaction. Crosslinked beads were pelleted at 4,000 rpm and washed with 20 mL of
0.1 M glycine-HCI pH 3.0 followed by exchange with fresh PBS. Coupled, washed beads

were suspended in 2 mL of PBS containing 0.01% (w/v) merthiolate for storage until use.

134

 

Coupled beads were pelleted at 4,000 rpm and washed twice with 15 mL of buffer
D500 (D100 with 500 mM KCl) containing only 5 mM B-mercaptoethanol to prevent light
chain dissociation. The phosphocellulose D fraction was split into two, six milliliter pools.
A seventy microliter bed volume of 12CA5 beads was aliquoted to each of 12 tubes and
one milliliter aliquots of the P-1 1 TFIID fraction were added to each tube. These samples
were nutated overnight at 4°C. Beads were precipitated by centrifugation for 5 minutes at
14,000 rpm in an Eppendorf ﬁxed-angle microfuge, and consolidated into six tubes.
Pellets were washed three times with buffer D500 containing 0.05% NP-40 and 5 mM [3-
mercaptoethanol to dissociate endogenous TFHA. The supernatant was discarded. Beads
were washed twice with 1 mL of D100 containing 5 mM B-mercaptoethanol, and eTFIID
was eluted with buffer D100 containing 1 mg/mL HA peptide and 5 mM [3-
mercaptoethanol. After a one hour incubation with gentle rocking at room temperature,
holes were pierced in the bottom of each tube with a 20 gauge needle followed by
centrifugation of all liquid from the tube at 14,000 rpm until the 12CA5 Sepharose beads

were completely dried. Eluted eTFIID was frozen in liquid nitrogen and stored at -80°C.

Preparation of His-tagged TFIIA

E. coli strain M15 transformants of thTFHAor, thTFIIAB, or thTFIIAy,
were grown at 37°C in LB media (1 liter) containing 100 ug/mL ampicillin and 25 ug/mL
kanamycin to an OD600 of 0.7 (E. coli strain and plasmids a kind gift of Arnold Berk,
UCLA). Cultures were induced with 0.5 mM IPTG and grown an additional 1.5 hours at

30°C. Cells were harvested at 5,000 rpm in a Sorvall GS-3 rotor for 5 minutes. The cell

135

 

pellets were used immediately or stored at -80°C. Pellets from 1 liter cultures were
resuspended by manual swirling in 12 mL of Qiabuffer A (10 mM Tris base, 0.1 M
NaH2P04, 6 M guanidine HCl, pH 8.0) containing 7 mM B-mercaptoethanol and 1 mM
phenylmethylsulfonylﬂuoride (PMSF). Suspended cells were rocked at 4°C for 1 hour in
Teﬂon Oakridge tubes.

Ni-NTA resin (Qiagen) was prepared as follows. Two milliliters of a 50% slurry
of resin was precipitated at 1,500 rpm for 5 minutes in a 15 mL Corex tube in a Sorvall
table top centrifuge. This resin (1 mL resin bed volume) was resuspended in 12 mL of
Qiabutfer A. The resin was precipitated with a second 5 minute spin. The supernatant
was discarded and the resin was suspended in 12 mL of Qiabuffer F (6 M guanidine HCl,
0.2 M acetic acid). Following this wash, the resin was pelleted by a 5 minute spin. The
acid washed resin was resuspended in 12 mL of Qiabuﬂ‘er A to wash out residual acetic
acid and the resin was precipitated by centrifugation for an additional 5 minutes. This
wash step was performed an additional two times.

The E. coli lysate was transferred to a 15 mL Corex tube containing the
equilibrated Ni-NTA resin. The lysate was rocked at 4°C for 1 hour to allow Ni-His6 tag
binding. Following this incubation, the resin was precipitated by centriﬁigation at 1,5 00

rpm in a tabletop centriﬁige. The supernatant was discarded. The resin was washed by

 

rocking 15 minutes at 4°C, with 10 mL of Qiabuffer A containing 7 mM B-

mercaptoethanol and 1 mM PMSF. The resin was then precipitated by centriﬁrgation and
the process repeated, followed by washing of the resin by rocking in 10 mL of Qiabuffer B
(10 mM Tris base, 0.1 M NaHzPO4, 8 M urea, pH 8.0) containing B-mercaptoethanol and

PMSF. The resin was precipitated by centrifugation and the wash step was repeated. The

136

 

resin was then suspended in 5 mL of Qiabuffer B containing B-mercaptoethanol and
PMSF and transferred to a disposable mini-column. Following transfer, the column was
washed with an additional 10 mL of Qiabuffer B. All subsequently used chromatography
buﬁ‘ers contained 7 mM B-mercaptoethanol and 1 mM PMSF. The column was washed
with 30 mL of Qiabuffer B, adjusted to pH 6.3, followed by 12 mL of Qiabuffer B
containing 10 mM imidazole. Following these wash steps, the proteins were eluted with
Qiabuffer B containing 200 mM imidazole. One milliliter fractions were collected and
analyzed by SDS-PAGE.

TFIIA was renatured by step dialysis to remove urea. Protein was estimated by

 

Bradford assay and the subunits were mixed in approximately equimolar amounts (1 mg of
each of the large subunits and 0.5 mg of the small subunit, in an undiluted volume of
approximately 5 mL). This mixture was dialyzed overnight against one liter of Buﬂ‘er
D100 (20 mM HEPES-KOH, pH 7.9, 20% glycerol, 0.2 mM EDTA, 100 mM KCl, 7 mM

B-mercaptoethanol, 1 mM PMSF) containing 2M urea. The mixture was then dialyzed

 

against one liter of D100 containing 1M urea for 4 hours, followed by a change of
dialysate to D100 plus 0.5M urea. Following an additional 4 hour dialysis against the
0.5M urea buffer, the sample was dialyzed against 3 x 1 liter of Buffer D100. Insoluble
material was removed by centriﬁrgation at 14,000 rpm in an Eppendorf microﬁrge.
Soluble TFIIA protein concentration was determined by a Bradford protein assay.

Aliquots were frozen in liquid nitrogen and stored at -80°C.

Preparation of EMSA Probes

137

 

The TBP-TFHA EMSA probe used was a 452 bp XhaI-HinDIII AdML promoter
fragment, excised from the vector pML (obtained from Dr. Zachary Burton, Michigan
State University). The fragment was gel puriﬁed and end-labeled with 32P-y-ATP by the
exchange reaction using T4 DNA kinase. The agarose EMSA probe used was a 225 bp
EcaRI-HinDIII fragment excised from the plasmid pG5E4TCAT (obtained from Arnold
Berk, UCLA). This fragment was incubated with calf alkaline phosphatase for 30
minutes, agarose gel puriﬁed, and end-labeled with 32P-y-ATP by the forward reaction of
T4 DNA kinase to Optimize speciﬁc activity of the probe. Probes were extracted with

phenol/chloroform, ethanol precipitated and puriﬁed by gel ﬁltration on a ~6cm Bio-Gel

 

P6-DG column poured in a Pasteur pipette. One microliter samples of the 100 DL
fractions were quantitated by Cerenkov counting in a scintillation counter. Speciﬁc

activity was determined assuming 100% recovery of the DNA.

Gal4 EMSA Assay
Binding reactions were assembled in 10 mM HEPES-KOH, pH 7.8, 60 mM KCl,

12.5% glycerol, 5 mM MgC12, 10 mM B-mercaptoethanol, 1 mg/mL bovine serum

 

albumin, 40 ug/mL polde:dC, and 500 cpm of probe (ﬁnal concentrations). The probe
used is a 240 bp fragment of pG5E4TCAT (see probe preparation section above)
encompassing 5 upstream Gal4 sites, the adenovirus E4 TATA box and downstream
promoter sequences, as well as a segment of the bacterial chloramphenicol
acetyltransferase gene. Gal4-fusion proteins were diluted in buffer D400 (20mM HEPES-
KOI-I, pH 7.9, 20% (v/v) glycerol, 0.2 mM EDTA, and 400 mM KCl) as indicated in text

and/or legends. One microliter of each dilution was used per reaction. Incubations were

138

 

performed at 30°C for 30 minutes, followed by electrophoresis on a 4% PAGE gel in 1x
TBE (90 mM Tris base, 90 mM boric acid, 2 mM EDTA) at 150 V for 1 hour and 45
minutes. Gels were dried under vacuum at 80°C onto Whatman 3MM paper. Results

were visualized by autoradiography.

TFHA-TBP EMSA Assay
TFHA-TBP gel shifts were performed as described by Maldonado and Reinberg

(Maldonado, E. et al., 1990). Yeast TBP (0.6 or 1.1 pmol) was incubated in binding
buffer with hTFIIA (0.07-1.49 pmol) at 30°C for 30 minutes. The binding buffer
contained 10 mM HEPES, pH 7.9, 4 mM MgC12, 5 mM (NH4)2SO4, 8% (v/v) glycerol,
2% (w/v) polyethyleneglycol (average molecular weight = 8000 Daltons), 50 mM KCl, 5
mM B-mercaptoethanol, 0.2 mM EDTA, 25 ug/mL poldede and l frnol of the 3’2P
labeled AdML promoter DNA probe. Following the incubations, samples were loaded
onto a 0.5x TBE gel and electrophoresed at 150 Volts for 1 hour and 45 minutes. Gels

were dried onto Whatman 3MM paper and visualized by autoradiography.

DA Complex EMSA Assay
Binding reactions were assembled in 10 mM HEPES-KOH, pH 7.8, 60 mM KCl,

12.5% glycerol, 5 mM MgC12, 10 mM B-mercaptoethanol, 1 mg/mL bovine serum
albumin, 40 ug/mL polde1dC, and 500 cpm of probe (ﬁnal concentrations). The probe is
identical to that described above (Gal4 EMSA methods section). As indicated in text and
legends, reactions contained 1-2 uL of TFHD, 20 ng of TFIIA and/or Gal4-ﬁ1sion protein

(0.5x, 1.0x, 1.5x protein required to saturate all ﬁve Gal4 binding sites as determined

139

 

empirically by titration of Gal4-ﬁisions in a 4% PAGE EMSA. The Gal4-fusion proteins
were dialyzed against buffer D400 (20mM HEPES-KOH, pH 7 .9, 20% (v/v) glycerol, 0.2
mM EDTA, and 400 mM KCl) prior to use. Reactions were assembled on ice, and
incubated at 30°C for 30 minutes prior to electrophoresis. Samples were loaded directly
without added loading dye to avoid shadowing by xylene cyanol. Electrophoresis was
performed in 1x Buffer G (45 mM Tris base, 45 mM Boric acid, 0.5 mM EDTA, 5mM
MgAcetate) on a 1.4% agarose (low EEO, Boehringer-Mannheim Biochemical) gel at
70V for 3 to 3.5 hours. After electrophoresis, gels were dried by vacuum at 80°C onto
DEAE-cellulose paper reinforced with Whatman 3MM paper. Results were visualized by

autoradiography.

140

 

Chapter Four

Activation domains are composed of multiple, small modular elements

The results described in the previous chapters provide evidence that activation
domains are composite in nature, composed of repeated, perhaps related functional motifs.
These essential elements are relatively small (approximately 10 residues) hydrophobic
sequences. Elements ﬁ'om different activators (e. g. VP16 and RelA) are relatively similar
in terms of their common requirement for hydrophobic amino acids for function.

However, such elements also display some structural distinctions; both VP16C

 

subdomains contain a critical dipeptide (a phenylalanine followed by an acidic residue),
whereas neither RelA subdomain examined contains an identical motif.

Despite similar structural requirements, not all activation domains are functionally
equal. This is apparent by comparison of a number of functional characteristics of VP16N
and VP16C. Their variable potency in different species, their distinct capacity to
derepress chromatin templates and their differential basal promoter speciﬁcities all suggest
that VP16N and VP16C ﬁrnction through distinct mechanisms in viva. Perhaps more

directly, in vitro studies establish that VP16C, but not VP16N, is capable of stimulating

 

the formation of a TFIID-TFIIA-DNA (DA) complex.

These results present a paradox. How do two VP16 motifs of such similar
structure impart distinct functional attributes to each subdomain? One possibility is that
the limited structural differences within these core motifs are responsible. To test this
possibility, we constructed a VP16N mutant (VP16NAD/L) which contained a core

primary structure similar to VP16C. However, this VP16NAD/L mutant did not stimulate

141

 

DA complex formation, arguing that the spacing of hydrophobic residues within an
activation motif is not the determinant of this speciﬁc function. The RelA results also
support this conclusion. We identiﬁed two motifs within the RelA activation domain that
appear similar to the VP16C core motif. An alanine scan of each RelA subdomain veriﬁed
these predictions in part. The ﬁinction of each RelA subdomain was critically dependent
upon a similar cluster of hydrophobic residues. Interestingly, VP16C and RelAIII display
identical promoter speciﬁcity in viva. However, the RelA activation domain is incapable
of stimulating DA complex formation. These results argue that while some common
mechanistic features of the RelA and VP16C activation domains may exist, DA
recruitment is not responsible for the observed promoter speciﬁcity. These results
illustrate that prediction of function through such structural analyses is unlikely to be
ﬁ'uitful.

Clearly, the pattern of hydrophobic residues within the subdomain is not solely
responsible for ﬁmctional speciﬁcity. Yet, the alanine scan of VP16C demonstrates that
the hydrophobic residues within the core FxexMF motif are primarily responsible for in
viva activity. What is the role(s) of these hydrophobic residues? At least two formal
possibilities exist. Either the hydrophobic residues are essential for nucleation of a critical
tertiary structure within the activation domain, or the residues are involved directly in
formation of a protein-protein interface between the domain and its target protein(s). The
latter hypothesis seems likely given the results of stmctural studies of a number of

complexes of activation domains and their proposed cellular targets.

Activator-target protein-protein complexes appear predominantly hydrophobic

142

The crystal structure of a peptide from the p53 activation domain bound to MDM2
illustrates how an activation domain might complex with its cellular target (Kussie, P.H. et
al., 1996). The MDM2 protein is the product of a p53 inducible gene. MDM2 binds the
activation domain of p53 and masks its ﬁmction. This interaction appears to involve the
same residues of p53 essential for transactivation (Lin, J. et al., 1994), suggesting that the
mode of binding of p53 to its transcriptional target may be similar to that observed in the
MDM2 co-crystal. In the crystal structure, the activation domain peptide forms a short
two-tum helix. Three hydrophobic residues (F19, W23 and L26) insert into a deep cleft in
the MDM2 interaction domain. Each of these residues lies on the same face of the helix.
Only two intermolecular hydrogen bonds are observed in the entire complex. One
involves MDM2 hydrogen bonding to a p53 main-chain amide at the entrance to the cleft,
and a second MDM2 hydrogen bond forms with the tryptophan indole group at the
bottom of the cleft. This paucity of hydrogen bonding suggests that speciﬁc recognition
must rely principally upon recognition of the hydrophobic residues. Consistent with this
observation, the surface complementarity observed between p53 and MDM2 is very
precise. The three critical hydrophobic residues in p53 pack against each other, producing
a surface exactly complementary to the MDM2 binding pocket (Kussie, P.H. et al., 1996).

The structure of the CREB-CBP complex has recently been solved by NMR
(Radhakrishnan, I. et al., 1997). The CREB activator requires the coactivator CBP for
full in viva function (Kwok, R.P.S. et al., 1994). Consistent with this observation, a
domain of CBP (the KIX domain) speciﬁcally interacts with a segment of the CREB
activation domain (pKID). The pKID-KIX interaction resembles the p53-MDM2

interaction in that the activation domain-target interface is predominantly composed of the

143

 

hydrophobic face of an amphipathic alpha helix, inserted into a hydrophobic channel on
the surface of the KIX target protein. However, this structure also differs in some
signiﬁcant aspects. The pKID domain consists of two helices connected by a short ﬂexible
loop. This 100p contains a Protein Kinase A (PKA) phosphorylation site.
Phosphorylation of this site is necessary for the pKID-KD( interaction (Zhong, H. et al.,
1998). The N-terminal helix makes only minor contacts with the KIX domain. The C-
terrninal helix is predominantly responsible for formation of the interface.

This helix of pKID lies in a relatively shallow hydrophobic groove across the

surface of the KDC domain packing at an angle across the surface of two KIX helices.

 

This contrasts with the p53-MDM2 interaction, in which the p53 helix inserts into a very
deep cleft between two helices. In addition to this distinction, the pKID-KD( interaction
is signiﬁcantly augmented by electrostatic interactions. Three ionic interactions are
observed in the NMR structure. Two involve aspartic acid residues in the second helix of
pKID and lysine residues in KIX, while a third occurs between an arginine residue of the
ﬁrst helix and a glutamate residue of the KIX domain. The PKA-phosphorylated serine
residue of the central loop of pKID also hydrogen bonds to a tyrosine residue of the KIX

domain. Thus, while the principal mode of interaction remains similar to the p53-MDM2

 

interaction, the CREB-CBP interface also contains multiple ionic interactions that
contribute signiﬁcantly to the speciﬁcity of the CREB-CBP interface.

A third relevant structure has also been solved. The steroid receptors activate
transcription from their target genes in a ligand dependent fashion. This activation is
thought to occur via a ligand-induced conformational change that results in the

recruitment of a family of related coactivator molecules (the p160 family of coactivators,

144

 

 

including SRC-l, GRIP-1, and ACTR). This interaction is thought to be dependent upon
a common motif in the p160 proteins. The motif consists of a short LxxLL sequence.

The crystal structure of the PPAR-y receptor (a steroid receptor superfamily member)
complexed with a segment of SRC-l which contains two of these LxxLL motifs was
recently reported (N olte, R.T. et al., 1998). The two SRC-l motifs contact each of the
ligand binding domains of a PPAR-y homodimer. Interestingly, in many ways the PPAR-
SRC-l interaction resembles the reverse of the p53-MDM2 and CREB-CBP interactions.
In this case, the ligand-binding domain of the activator (PPAR-y) forms a globular domain
that recognizes a short two-tum amphipathic alpha helix in the target (SRC-l). The
remainder of SRC-l is disordered and appears to form a ﬂexible tether between the
LxxLL motifs. The SRC-l helices pack into a hydrophobic interface at the junction of
four helices in the PPAR receptor ligand-binding domain. The interaction is
predominantly hydrophobic in nature and is supplemented by the hydrogen bonding of two
invariant receptor charged residues with the main-chain of the LxxLL motif of SRC-l at
the beginning and end of the helical motif.

This structure resembles a similar interaction between the thyroid hormone
receptor (TRB) and the LxxLL motif of GRIP-l (another p160 family member)
(Darimont, B.D. et al., 1998). Four residues of the LiorLL helix insert into a hydrophobic
patch constructed of the interface of four helices of the TRB receptor. Again, the
interface of the two proteins consists of a globular TRB domain that recognizes the
hydrophobic face of an amphipathic helix in GRIP-l. In contrast to the CREB-CBP and

p53-MDM2 complexes, the role of activator and target is reversed. The target, rather

I45

 

 

than the activation domain consists of a small helical motif which is speciﬁcally recognized
by a more complex globular domain (the receptor ligand-binding domain).

All four of these complexes are relatively simple molecular interactions, with small
interfacial areas, dominated by one of two partners. One partner (more commonly the
activation domain) folds upon target recognition to form a short (2-3 turn) amphipathic
helix, the hydrophobic face of which inserts into a complimentary pocket of the relatively
more complex target protein. The target protein is principally responsible for speciﬁcity of
recognition, providing a signiﬁcantly more complex interface that is dependent upon the
tertiary structure of the target protein. Induced folding of the activation domain is
observed in both the p53-MDM2 (Kussie, P.H. et al., 1996) and CREB-CBP complexes
(Radhakrishnan, I. et al., 1997). Speciﬁcity is achieved principally through
complementary of the interacting surfaces, although additional electrostatic interactions
also provide speciﬁcity to the CREB-CBP and receptor-p160 coactivator complexes.

The predominance of one partner in these activator-target interactions is more
reminiscent of a protein-peptide interaction than a typical protein-protein interaction (such
as those observed in multisubunit enzymes and oligomeric proteins). Like the structures
discussed above, protein-protein interfaces of oligomeric proteins are almost exclusively
hydrophobic and contain large interfacial areas ranging in size from 1,400 to greater than
6,000 A2 in size (Janin, J. et al., 1988). In contrast, protein-peptide interactions are
characteristically less hydrophobic, augmented by some additional ionic interactions (such
as is observed in the KIX-pKID interaction), and contain signiﬁcantly smaller interaction
surfaces (Janin, J. and Chothia, C., 1990). These characteristics are necessary for

transient protein-protein interactions such as antibody-antigen and protease-substrate

146

 

interactions, as these proteins must be stable in the absence of their partners. This requires
a smaller, relatively less hydrophobic surface area that is not energetically unfavorable
when exposed to solvent (Jones, S. and Thornton, J.M., 1996). These same properties
would be expected of activator and target proteins. Consistent with these requirements,
the interfaces observed in all three of the activator-target complexes is relatively small,
ranging from 1,200 A2 for the CREB-CBP interaction to 1,500 A2 for the p53-MDM2
interface. These ﬁgures are actually slightly smaller than the mean ﬁgure reported for
heterocomplexes (~1,900 A2) in a 1996 Brookhaven PDB survey performed by Jones and
Thornton (Jones, S. and J .M. Thornton, 1996). However, even smaller interfaces have
been observed. The SH3 domains of signal transduction proteins, which recognize a short
polyproline helix, typically bury ~800 A2 of surface area upon complex formation (see
review, Kuriyan, J. and Cowbum, D., 1997).

The number of transient protein-protein complexes with known structure has been
increasing steadily. While protease-inhibitor complexes and antibody Fab ﬁagments
provided initial examples of such structures, advances in signal transduction research have
provided a recent wealth of structural information. The structures of SH2, SH3, and PTB
domains complexed with their targets have been solved (for review see Kuriyan, J. and D.
Cowbum, 1997). No common theme emerges with respect to the structural requirements
for these proteins. While SH3 domains recognize the hydrophobic face of a short
polyproline helix (similar to the activation domain-target interfaces), the SH2 domains
recognize an extended polypeptide chain. This extended chain is bound across an SH2-
domain surface groove. Phosphotyrosine binds in a deep pocket at one end of the groove,

and speciﬁcity is determined by the geometry and chemical composition of the groove.

147

 

,Jv

 

 

These properties variably favor the presence of hydrophobic or polar residues at positions
C-terminal to the phosphotyrosine. This results in a binding pocket that is in essence
capable of "reading" the primary structure of a short peptide motif. The PTB domains
provide yet one more distinct method of protein-peptide interaction. The peptide ligand
forms a [3 sheet that is incorporated into an antiparallel sheet of the PTB domain. The
interaction is extensively stabilized by intermolecular hydrogen bonding between these two
antiparallel B sheets. Nature has clearly found many approaches suitable for the formation

of such transient protein-protein complexes.

Modularity of activation domain structure

In both VP16 and RelA, the activation domain appears to be modular in structure,
composed of repeated motifs with similar structural requirements. Yet, these subdomains
do appear to be functionally discemable. How is this "speciﬁcity" achieved? Two
possibilities exist. The speciﬁcity may result from speciﬁc recognition or "readout" of the
primary structure, as is observed with SH2 domains. Alternatively, the core motifs may
provide the driving force of the molecular interaction, without directly contributing to
speciﬁcity. The determinants of speciﬁcity may reside within surrounding sequences. Our
studies seem to favor the second model. The RelA activation domain contains two amino
acid segments that contain a F xx‘I’T motif that is also found in VP16C. Despite this
structural similarity, the RelA activation domain does not share the capacity to stimulate
formation of a promoter DA complex. Furthermore, a VP16N mutant engineered to

contain a hydrophobic core with a core motif similar to that observed in VP16C does not

148

 

display this property either. These results argue that the precise primary structure of the
Fxx‘I’T core motif is not a suﬁicient determinant for speciﬁc function.

Given these results it seems likely that the primary determinants of speciﬁcity
reside outside of the core motif. Studies performed on the steroid receptor coactivators
SRC-l and GRIPl provide a precedent that could be applicable to other activation
domains. Darimont, et al. describe a set of biochemical and crystallographic studies of the
GRIPl-TRB interaction (Darimont, B.D. et al., 1998). Sequences both N-terminal and C-
terminal to the LxxLL core are required for high affinity binding of GRIPl to TRB.
Despite this contribution, crystallographic studies reveal no direct interaction between
these regions and the ligand-binding domain. Rather, the authors attribute the effects of
ﬂanking sequences to favorable electrostatics between GRIP] residues and the ligand-
binding domain. 1

A role for core-ﬂanking sequences in speciﬁc steroid receptor recognition has also
been observed with SRC-l. McInerney, et al. tested the capacity of rnicroinjected SRC-l
mutants to mediate receptor ﬁmction in viva (McInemey, E.M. et al., 1998). Both the
LxxLL core and ﬂanking sequences were found critical for function in viva. Interestingly,
mutations affecting speciﬁc residues C-terminal of one core motif of SRC-l differentially
affected coactivation of the estrogen and retinoic acid receptors. Apparently, distinct
residues within these C-terminal sequences differentially contribute to coactivation of
speciﬁc receptors. Similar to the situation observed with the TRB receptor, the PPAR}!-
SRC-l crystal structure failed to discern any direct contact between these ﬂanking

sequences and PPARy.

149

 

 

 

 

Signal transduction proteins provide another example of a strategy employed by
nature to provide speciﬁcity to these protein-protein interactions. Like the activation
domain-target complexes, SH2 and SH3 domains recognize small peptide motifs. Because
of the small size of these interfaces, the strength of these complexes is limited, with typical
complexes displaying high nanomolar to micromolar dissociation constants. These
properties result in weak, transient interactions, which these proteins appear to have
adapted to their advantage (reviewed in Kuriyan, J. and D. Cowbum, 1997). Analysis of
the kinetics of SH2-phosphopeptide interactions reveals rapid on and off rates, allowing
the rapid and repetitive sampling of alternate sites to facilitate the process of selective
pairing of SHZ domains and phosphorylated targets.

These signaling proteins share another property with the transcriptional activators;
they are modular in structure. This modularity is thought to enhance both afﬁnity and
speciﬁcity, through cooperative function of individual domains. One example of this
cooperativity is seen in T cell receptor signaling. The T-cell receptor is composed of
several polypeptide chains, which contain speciﬁc phosphotyrosine motifs (ITAM motifs).
The ITAM motifs are recognized by ZAP-70, a tyrosine kinase that contains two SHZ
domains that are connected by a rigid linker. The juxtapositioning of the two domains
provides a combinatorial ITAM/ZAP-70 interface that is speciﬁc for a very precise head to
tail repeat of two phosphotyrosine motifs. This leads to cooperativity between the two
protein-protein contacts (Hatada, M.H. et al., 1995). A similar example is provided by the
SH-PTP2 tyrosine phosphatase. Binding of peptides to the SH2 domains of this protein
leads to activation of the phosphatase. ,While single phosphotyrosine peptides can activate

the phosphatase, activation is dramatically enhanced by peptides that contain tandem

150

 

phosphotyrosine motifs. The structure of SH2-PTP2 reveals that the two SH2 domains
are rigidly constrained, once more selecting for the precise positioning of two
phosphotyrosine motifs for optimal effect (Eck, M.J. et al., 1996). The GRB2 adapter
provides yet one more example of the repetitive, modular structure of these proteins.
GRB2 contains one SH2 and two SH3 motifs. In contrast to the previous two examples,
the domains are ﬂexibly linked, providing a molecule suited to more promiscuous
interactions (Maignan, S. et al., 1995). Thus, these signaling molecules have evolved to
combine multiple peptide recognition domains in speciﬁc structural combinations to

facilitate either the ﬂexibility or speciﬁcity of these protein-protein interactions.

 

The structure of activation domains suggests that activation domain targets may
exploit a similar strategy. RelA provides a good example of the modularity of activation
domains. The domain is separable into three, independently functional subdomains. Each
subdomain is similar in terms of required structural properties (all are highly dependent
upon critical hydrophobic residues) and each is well conserved evolutionarily. Perhaps the
activation domain has evolved by duplication of a common ancestral motif followed by

specialization of these duplicated sequences into individual, distinct subdomains. These

 

subdomains may cooperate to recognize a single target with enhanced affinity, or may
have evolved to recognize independent targets.

The VP16 activation domain is also composed of repeated, ﬁinctionally distinct
subdomains. Multimerization of these motifs may enhance activation domain potency by
simultaneously allowing the activator to recruit multiple targets to regulated promoters.
This model could explain the greater than additive eﬁ‘ect, or synergy, of multiple activators

to transcription (Chi, T.H. et al., 1995). Consistent with this model, He and Weintraub

151

 

report that multirnerization of two activation domains of distinct ﬁmction (VP16N and an
Spl activation domain fragment) results in synergistic increases in activator potency (He,
S. and SJ. Weintraub, 1998). The ﬁmctional differences between the VP16N and VP16C
subdomains seem to argue that the VP16N and VP16C subdomains may ﬁinction in a
similar, complementary fashion. However, Emami et al. demonstrated that a similar eﬁ‘ect
is observed upon multirnerization of two or more VP16N subdomains on a Gal4 chimeric
protein (Emami, K.H. and M. Carey, 1992). Thus, strictly speaking, synergy does not
require two distinct subdomains. Perhaps, the VP16N (and other subdomains) may have
evolved to ﬁllﬁll several mechanistic roles.

The cooperative nature of activator-target interactions is underscored by studies
performed with the CREB activator. CREB provides an excellent model for activator
ﬁmctional studies. ' A functionally relevant target (CBP) has been identiﬁed and the
mechanism of CREB-CBP interaction is understood in molecular detail (see above,
Radhakrishnan, I. et al., 1997). Despite the well-characterized nature of the CREB-CBP
interaction, it is clear that CREB-CBP interaction alone is insufﬁcient for CREB
activation. Phosphorylation dependent, CBP-dependent CREB activation can be
reconstituted in vitro in HeLa extracts. The CREB activation domain is also bipartite.
The pKID domain (discussed above) is responsible for CBP interaction. A second
glutamine-rich domain, Q2, was found to bind dTAFn110 (F erreri, K. et al., 1994). Both
interactions appear to required for activation in vitro, as antibodies against hTAFn130
blocked CREB activation (Nakajima, T. et al., 1997).

The coactivator CBP provides another relevant observation illustrating that

cooperativity of distinct activation domains results in transcriptional synergy. The Spl

152

 

and SREBP-la (Sterol-Response-Element binding protein) activators synergize to activate
the LDLR (low density lipapratein receptor) promoter in vitro. Synergistic activation is
observed in a crude nuclear extract, but is not observed in a transcription reaction
reconstituted with puriﬁed basal factors. Robust synergy can be reproduced using a
chromatin template in the presence of a partially puriﬁed CBP complex. This synergistic
activity is dependent upon CBP, as CBP peptides can block this activation. Furthermore,
TAFS are required for this synergy as well, since TBP cannot replace TFIID in the
transcription reactions (N aar, AM. et al., 1998). These results imply that efﬁcient
synergy of these two activators requires multiple coactivator components (CBP and
TAFS).

The VP16C results suggest a potential mechanism for VP16N and VP16C
synergy. The VP16N subdomain eﬁiciently derepresses an integrated reporter. In
contrast, VP16C is inemcient when assayed from an integrated, chromosomal reporter. In
contrast, VP16C stimulates TFIID recruitment via TFHA in vitro. This activity is not
present in VP16N. VP16C also displays promoter speciﬁc function, requiring an Inr
element for function in viva. Conceptually, these two observations could be related. The
primary defect expected for a TATA-less promoter would be a reduced ability to recruit
TFHD to the promoter. Thus, an activator that stimulates TFIID recruitment would be
expected to enhance TATA-less promoter function. If this activity was unnecessary due
to the presence of a consensus TATA sequences, a TFHA-dependent activator (VP16C)
would be expected to provide little activation. Given this model, the VP16C Inr-

preference can be rationalized by its TFHA-dependent TFIID (DA) recruitment activity.

153

 

The two subdomains of VP16 may synergize to produce a highly potent activation domain
due to these two complementary activities.

This model suffers from a signiﬁcant deﬁciency. The Adenovirus E4 promoter
utilized for the DA recruitment assays does not contain an Inr element. Furthermore,
Emami, et al. failed to detect stable TFIID binding to an Inr promoter in vitro, even in the
presence of TFHA (Emami, K.H. et al., 1997). This is perhaps not surprising. While
TFIIA does bind DNA in the TBP-TFIIA-DNA complex, these DNA-interactions are
relatively limited (Geiger, J .H. et al., 1996). Perhaps TFIIA, being unable to speciﬁcally
recognize promoter sequences, requires recruitment via a sequence-speciﬁc activator. It
may prove interesting to test the ability of Gal4-VP16C to enhance TFIID recognition of
an Inr promoter in the DA agarose EMSA. Results from experiments performed by
Hansen and Tjian also bolster this hypothesis (Hansen, SK. and R. Tjian, 1995). Hansen
found that TFHA, in conjunction with TAFS, mediate core-promoter speciﬁcity of Adh
promoter usage in Drosophila. The TFIIA preferred promoter contains an Inr element
critical for the observed speciﬁcity. These results imply that TFIIA recruitment could
conceivably mediate Inr-speciﬁcity.

Recent developments with RelA are also of potential interest. RelA is a NF-KB
family member. These transcription factors are regulated by multiple mechanisms.
Primary control is maintained through a mechanism of cellular localization. A family of
IKB proteins masks the nuclear localization signal and tethers the NF-KB proteins in the
cytoplasm. In response to a wide variety of signal transduction pathways, the IKB

proteins are degraded by proteolysis allowing nuclear translocation of the NF-KB

transcription factors. In addition to this mechanism, NF-KB is also regulated by

154

 

phosphorylation. A subset of cellular protein kinase A (PKA) is found in a stable complex
with NF-KB and IKB (Zhong, H. et al., 1997). Activation of this PKA species occurs by a
mechanism independent of CAMP levels.

The cellular coactivator CBP also potentiates RelA transcriptional activity. The
CBP protein was originally described as a necessary adapter for CREB transactivation.
The structure of the CREB-CBP interface has been described above. Binding of pKID to
KIX is dependent upon phosphorylation of a critical serine residue. Similarly,
transactivation by RelA is stimulated by phosphorylation of Ser 276 within the rel
homology domain (Zhong, H. et al., 1998). The crystal structure of the rel homology
domains of a p50/RelA heterodimer has been solved (Chen, F.E. et al., 1998). The
structure of this region of the rel homology domain displays no structural homology to the
pKID structure observed in the NMR studies of CBP and CREB. The phosphorylation
site resides in a loop between two antiparallel beta strands at the N-terminus of the last
beta sheet in the dimerization domain. However, S276A mutations seriously debilitate
transcription in cooperation with CBP in viva (Zhong, H. et al., 1998). Interestingly,
Zhong et al. identiﬁed via sequence alignment a region of similarity between RelA and
CREB immediately C-terminal to this phosphorylation site. This sequence includes a
second, potential phosphorylation site (Thr 308). Perhaps phosphorylation of Ser 276
induces a conformational change exposing these C-terminal sequences to PKA, another
kinase, or CBP itself. Phosphorylation of this Thr 308 residue might then allow
interaction of this pKID-like sequence with the KIX domain of CBP. For this reason,
mutagenesis of these C-terminal sequences, particularly Thr 308 would have been of

interest.

155

 

These studies also produced evidence of the potential for cooperative interactions
between multiple RelA domains and separate CBP domains. Zhong et al. assayed CBP
coactivator activity using a cotransfection assay in viva. Cotransfection of a CBP
expression vector enhances NF-KB transactivation. This enhancement correlates with the
ability of CBP to bind RelA in vitro. The authors found that two RelA activation domain
segments bind adjacent domains of the CBP adapter. A segment of RelAH binds the N-
terminus of CBP, while the Ser 276 dependent segment of the rel homology domain binds
the KD( domain of CBP immediately adjacent to this N-terminal segment. This
modularity of both signaling partner (the activator) and signal transduction target (the
CBP adapter) parallels the strategies employed by SH2 and SH3 domain containing signal
transduction components. Whether these properties also result in cooperativity of
activator-target protein-protein interactions remains to be addressed. These observations
also suggest that the investigation of the effects of our RelA mutants in the natural context
of the RelA activator may be worthwhile. Perhaps the RelAH region we have targeted is
involved in CBP binding.

The least understood aspects of these protein-protein interactions are perhaps the
most fundamental. Traditionally, studies of protein-protein interactions in transcriptional
regulation have been primarily phenomenological. Studies have been biased toward
identiﬁcation of potential targets. Although target recruitment through protein-protein
interactions has long been suggested as the mode of action of transcriptional activators,
the lack of identiﬁed targets has severely hampered our understanding of the nature of
these binding events. The recent identiﬁcation of the CBP/p300 family as coactivators of

a large number of activators has allowed a more detailed study of these protein complexes.

156

 

More quantitative studies may allow a more detailed understanding of the molecular
details of these events.

More detailed quantitative studies of VP16-target interactions may also provide
insight into VP16 structure and function. The VP16 activation domain binds multiple
proteins of potential transcriptional relevance. Interactions with TBP (Stringer, K.F. et
al., 1990), TFIIB (Roberts, S.G.E. et al., 1993), hTAFn32 (Klemrn, R.D. et al., 1995),
TFHA (Kobayashi, N. et al., 1998), TFIIH (Xiao, H. et al., 1994) have all been reported.
The activation domain may have evolved to produce a ﬂexible interface capable of
interaction with a variety of transcriptional targets. This ﬂexibility would be advantageous
to a viral protein, allowing strong activation within a genetically compact molecule.
Alternatively, the complications of in vitro biochemistry using a protein that is both highly
charged and highly hydrophobic may have contributed to the generation of artifacts.
Typically, studies have utilized VP16 mutants to correlate in vitro and in viva phenotypes
to minimize these concerns. Rarely have the energetics of these protein-protein
interactions been investigated. Quantitation could provide a means to compare and
contrast these interactions.

The mechanism of DA complex formation may be another aspect that could beneﬁt
from more quantitative studies. The inability of VP16NAD/L or RelA to stabilize the DA
complex suggests that the VP16C Fxx‘P‘I’ motif is not sufﬁcient for this ﬁrnction.
Additional residues within the subdomain must be necessary for this activity. A set of
alanine substitutions throughout the entire VP16C subdomain has been constructed
(Sullivan, SM. et al., 1998). Comparison of the effects of these mutations on DA

complex formation may be informative. In addition, the component within the DA

157

 

 

complex that is recognized by VP16C remains unclear. Kobayashi, et al. established a
correlation between VP16C mutant in viva potency and their capability to bind TFHA and
stimulate DA complex formation. However, other components may be involved in this
process, most notably TAFn32, which has been shown to bind VP16C (Klemm, RD. et
al., 1995). Comparison of TAF binding, TFHA binding and DA complex formation with
the alanine scan set of mutations could provide a detailed data set capable of resolving
these issues. More quantitative studies using the alanine scan mutants may help identify
the structural elements that specify DA complex stabilization. Perhaps sequences external
to the FxxW motif of VP16C participate in TFIIA binding in a way similar to that

observed in the SRC-l and GRIP] LxxLL motifs.

158

 

 

 

BIBLIOGRAPHY

159

 

BIBLIOGRAPHY

Apone, L.M., C.-M.A. Virbasius, J .C. Reese and MR Green (1996). “Yeast TAFII90 is
required for cell-cycle progression through G2/M but not for general transcription
activation.” Genes and Dev. 10: 2368-23 80.

Arany, Z., W.R. Sellers, D.M. Livingston and R. Eckner (1994). “ElA-associated p300
and CREB—associate CBP belong to a conserved family of coactivators.” Cgﬂ 77: 799-
800.

Arias, J., A.S. Alberts, P. Brindle, F.X. Claret, T. Smeal, M. Karin, J. Feramisco and M.
Montminy (1994). “Activation of CAMP and mitogen responsive genes relies on a
common nuclear factor.” Nature 370: 226-229.

Amdt, K.M., S.L. Ricupero, D.M. Eisenmann and F. Winston (1992). “Biochemical and
genetic characterization of a yeast TFIID mutant that alters transcription in vivo and
DNA binding in vitro.” Mol. Cell. Biol. 12: 2372-2382.

Amdt, K.M., S. Ricupero-Hovasse and F. Winston (1995). “TBP mutants defective in
activated transcription in viva.” EMBO J. 14(7): 1490-1497.

Amdt, K.M., C.R. Wobbe, S. Ricupero-Hovasse, K. Struhl and F. Winston (1994).
“Equivalent mutations in the two repeats of yeast TATA-binding protein confer distinct
TATA recognition speciﬁcities.” Mol. Cell. Biol. 14(6): 3719-3728.

Aso, T., J .W. Conaway and RC. Conaway (1995). “The RNA polymerase elongation
complex.” FASEB J. 9: 1419-1428.

Auble, D.T., Hansen, K. E., Mueller, C. G. F., Lane, W. S., Thomer, J., Hahn, S. (1994).
“Motl, a global repressor of RNA polymerase II transcription, inhibits TBP binding to
DNA by an ATP-dependent mechanism.” Genes and Development 8(16): 1920-1934.

Bagby, 8., S]. Kim, E. Maldonado, K.I. Tong, D. Reinberg and M. Ikura (1995).
“Solution structure of the C-temrinal core domain of human TFIIB: Similarity to cyclin A
and interaction with TATA binding protein.” C_el_l 82: 857-867.

Bannister, A.J., T. Oehler, D. Wilhelm, P. Angel and T. Kouzarides (1995). “Stimulation
of c-Jun activity by CBP: c-Jun residues Ser63/73 are required for CBP induced
stimulation of in viva and CBP binding in vitro.” Oncogene 11: 2509-2514.

Bentley, D.L. (1995). “Regulation of transcriptional elongation by RNA polymerase II.”
Curr. Opin. Genetics Dev. 5(2): 210-216.

160

 

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.” Cel_l 70:
25 1-265.

Blair, W.S., H. Bogerd and RR. Cullen (1994a). “Genetic analysis indicates that the
human foamy virus bel-l protein contains a transcriptional activation domain of the acidic
class.” J Virol. 68: 3803-3808.

 

Blair, W.S., H.P. Bogerd, S.J. Madore and BK Cullen (1994b). “Mutational analysis of
the transcription activation domain of RelA: Identiﬁcation of a highly synergistic minimal
acidic activation module.” Mol. Cell. Biol. 14(11): 7226-7234.

Blair, W.S. and ER. Cullen (1997). “A yeast TATA-binding protein mutant that
selectively enhances gene expression from weak RNA polymerase II promoters.” Mol.
Cell. Biol. 17: 2888-2896.

Blair, W.S., RA. Fridell and ER. Cullen (1996). “Synergistic enhancement of both
initiation and elongation by acidic transcription activation domains.” EMBO J. 15: 1658-
1665.

Blau, J., H. Xiao, S. McCracken, P. O'Hare, J. Greenblatt and D. Bentley (1996). “Three
functional classes of transcriptional activation domains.” Mol. Cell. Biol. 16: 2044-2055.

Brown, S.A., C.S. Weirich, EM. Newton and RE. Kingston (1997). “Transcriptional
activation domains stimulate initiation and elongation at different times and via different
residues.” EMBO J 17: 3146-3154.

Brownell, J .E., J. Zhou, T. Ranalli, R. Kobayashi, D.G. Edmondson, S.Y. Roth and CD.
Allis (1996). “Tetrahymena histone acetyltransferase A: A homolog to yeast Gcn5p
linking histone acetylation to gene activation.” C_eﬂ 84: 843-851.

Bryant, G.O., L.S. Martel, S.K. Burley and A.J. Berk (1996). “Radical mutations reveal
TATA-box binding protein surfaces required for activated transcription in viva.” Genes
Dev. 10: 2491-2504.

Bunick, D., R. Zandomeni, S. Ackerman and R. Weinmann (1982). “Mechanism of RNA
polymerase II-speciﬁc initiation of transcription in vitro: ATP requirement and uncapped
run-off transcripts.” Cell 29: 877-886.

Buratowski, S., S. Hahn, L. Guarente and PA. Sharp (1989). “Five intermediate
complexes in transcription initiation by RNA polymerase H.” Cell 56: 549-561.

Buratowski, S. and H. Zhou (1992). “Transcription factor IID mutants defective for
interaction with transcription factor IIA.” Science 255: 1130-1132.

 

161

 

Burke, T.W. and J .T. Kadonaga (1996). “Drosophila TFIID binds to a conserved
downstream basal promoter element that is present in many TATA-box—deﬁcient
promoters.” Genes Dev. 10: 711-724.

Burley, SK. and RC. Roeder (1996). “Biochemistry and structural biology of
transcription factor 11]) (TFIID).” Annu. Rev. Biochem. 65: 769-799.

Cairns, B.R., Y]. Kim, M.H. Sayre, B.C. Laurent and RD. Kornberg (1994). “A
multisubunit complex containing the SWIl/ADR6, SWIZ/SNFZ, SW13, SNFS, and SNF6
gene products isolated from yeast.” Proc. Natl. Acad. Sci. USA 91: 1950-1954.

Candau, R. and S.L. Berger (1996). “Structural and functional analysis of yeast putative
adaptors - evidence for an adaptor complex in viva.” J. Biol. Chem. 271: 5237-5245.

Carcamo, J ., L. Buckbinder and D. Reinberg (1991). “The initiator directs the assembly of
a transcription factor-IID-dependent transcription complex.” Proc. Natl. Acad. Sci. USA
88: 8052—8056.

Carey, M., J. Leathervvood and M. Ptashne (1990). “A potent GAL4 derivative activates
transcription at a distance in vitro.” Science 247 : 710-711.

 

Carruth, L.M., Hardwick, J. M., Morse, B. A., Clements, J. E. (1994). “Visna virus tat
protein: a potent transcription factor with both activator and suppressor domains.” Journal -
of Virology 68(10): 6137-6146.

Cavallini, B., I. Faus, H. Matthes, J .M. Chipoulet, B. Winsor, J .M. Egly and P. Chambon
(1989). “Cloning of the gene encoding the yeast protein BTFlY, which can substitute for
the human TATA box-binding factor.” Proc. Natl. Acad. Sci. USA 86: 9803—9807.

Chang, C. and J .D. Gralla (1993). “Properties of initiator -associated transcription
mediated by GAL4-VP16.” Mol. Cell. Biol. 13(12): 7469-7475.

Chang, C., C.F. Kostrub and Z.F. Burton (1993). “RAP30/74 (transcription factor IIF) is
required for promoter escape by RNA polymerase II.” J. Biol. Chem. 268: 20482-20489.

Chang, J ., D.H. Kim, S.W. Lee, K.Y. Choi and Y.C. Sung (1995). “Transactivation ability
of p53 transcriptional activation domain is directly related to the binding afﬁnity to TATA-
binding protein.” J. Biol. Chem. 270: 25014-25019.

Chasman, D.I., J. Leatherwood, M. Carey, M. Ptashne and RD. Kornberg (1989).

“Activation of yeast polymerase H transcription by herpesvirus VP16 and GAL4
derivatives in vitro.” Mol. Cell. Biol. 9: 4746-4749.

162

 

 

Chatterjee, and K. Struhl. (1995). “Connecting a promoter-bound protein to TBP
bypasses the need for a transcriptional activation domain.” Nature 347: 820-822.

Chen, F .E., D.-B. Huang, Y.-Q. Chen and G. Ghosh (1998). “Crystal structure of p50/p65
heterodimer of transcription factor NF-kappaB bound to DNA.” Nature 391: 410-413.

Chen, J ., L.D. Attardi, C.P. Verrijzer, K. Yokomori and R. Tjian (1994). “Assembly of
recombinant TFIID reveals differential coactivator requirements for distinct transcriptional
activators.” QB 79(1): 93-105.

Chi, T. and M. Carey (1996). “Assembly of the isomerized TFHA-TFIID-TATA ternary
complex is necessary and sufﬁcient fo gene activation.” Genes Dev. 10: 2540-2550.

Chi, T.H., P. Lieberman, K. Ellwood and M. Carey (1995). “A general mechanism for
transcriptional synergy by eukaryotic activators.” Nature 377: 254-257.

Choi, B., M. Bando, S. Hasegawa and M. Horikoshi (1996). “Isolation and
characterization of a cDNA encoding a novel human TFHD subunit containing similarities
with histones H2B and H3.” Gene 169: 263-267.

Choy, B. and MR. Green (1994). “Eukaryotic activators ﬁmction during multiple steps of
preinitiation complex assembly.” Nature 366: 531-536.

Coleman, RA. and BF. Pugh (1997). “Slow dimer dissociation of the TATA binding
protein dictates the kinetics of DNA binding.” Proc. Natl. Acad. Sci. USA 94: 7221-7226.

Coleman, R.A., A.K.P. Taggart, L.R. Benjamin and BF. Pugh (1995). “Dimerization of
the TATA binding protein.” J. Biol. Chem. 270: 13842-13849.

Colgan, J. and J .L. Manley (1992). “TFHD can be rate limiting in viva for TATA-
containing, but not TATA-lacking, RNA polymerase II promoters.” Genes Devel. 6: 304-
315.

Colgan, J ., S. Wampler and J .L. Manley (1993). “Interaction between a transcriptional
activator and transcription factor-IIB in viva.” Nature 362(6420): 549-553.

 

Conaway, RC. and J .W. Conaway (1993). “General initiation factors for RNA
polymerase H.” Ann. Rev. Biochem. 62: 161-190.

Cortes, P., O. Flores and D. Reinberg (1992). “Factors involved in speciﬁc transcription

by mammalian RNA polymerase H: puriﬁcation and analysis of transcription factor HA
and identiﬁcation of transcription factor HJ.” Mol. Cell. Biol. 12: 413-421.

163

Cote, J ., J. Quinn, J .L. Workman and CL. Peterson (1994). “Stimulation of GAL4
derivative binding to nucleosomal DNA by the yeast SWI/SNF complex.” Science 265:
53-60.

 

Cress, W.D. and SJ. Triezenberg (1991). “Critical structural elements of the VP16
transcriptional activation domain.” Science 251: 87-90.

Crowley, T.E., T. Hoey, J .K. Liu, Y.N. Jan, L.Y. Jan and R. Tjian (1993). “A new factor
related to TATA-binding protein has highly restricted expression patterns in Drosophila.”
Nature 361: 557-561.

Csordas, A. (1990). “On the biological role of histone acetylation.” Biochem J. 265: 23-
28.

Dai, P., H. Akimaru, Y. Tanaka, D.X. Hou, T. Yasukawa, C. Kanei-Ishii, T. Takahashi
and S. Ishii (1996). “CBP as a transcriptional coactivator of c-Myb.” Genes and Dev. 10:
528-540.

Darimont, B.D., R.L. Wagner, J .W. Apriletti, M.R. Stallcup, P.J. Kushner, J .D. Baxter,
R.J. Fletterick and KR. Yamamoto (1998). “Structure and speciﬁcity of nuclear receptor-
coactivator interactions.” Genes Dev. 12: 3343-3356.

De Felice, M., G. Damante, M. Zannini, H. Francis-Lang and R. Di Lauro (1995).
“Redundant domains contribute to the transcriptional activity of the thyroid transcription
factor 1.” J. Biol. Chem. 270: 26649-26656.

Dikstein, R., S. Ruppert and R. Tjian (1996). “TAFIIZSO is a bipartite protein kinase that
phosphorylates the basal transcription factor RAP74.” Cgﬂ 84: 7 8 1-790.

Drapkin, R. and D. Reinberg (1994). “The multifunctional TFHH complex and
transcriptional control.” Trends Biochem. Sci. 19: 504-508.

Drysdale, C.M., E. Duenas, B.M. Jackson, U. Reusser, G.H. Braus and AG. Hinnebusch
(1995). “The transcriptional activator GCN4 contains multiple activation domains that are
dependent on hydrophobic amino acids.” MCB 15(3): 1220-1233.

Dynlacht, B.D., T. Hoey and R. Tjian (1991). “Isolation of coactivators associated with
the TATA-binding protein that mediate transcriptional activation.” Q11 66: 563-5 76.

Eck, M.J., S. Pluskey, T. Trueb, S.C. Harrison and SE. Shoelson (1996). “Spatial
constraints on the recognition of phosphoproteins by the tandem SH2 domains of the
phosphatase SH-PTP2.” Nature 379: 277-280.

 

Emami, K.H. and M. Carey (1992). “A synergistic increase in potency of a multimerized
VP16 transcriptional activation domain.” EMBO J. 11(13): 5005-5012.

164

Emami, K.H., A. Jain and ST Smale (1997). “Mechanism of synergy between TATA and
initiator: synergistic binding of TFHD following a putative TFHA-induced isomerization.”
Genes and Dev. 11: 3007-3019.

Emami, K.H., W.W. Navarre and ST. Smale (1995). “Core promoter speciﬁcities of the
Spl and VP16 transcriptional activation domains.” Mol. Cell. Biol. 15: 5906-5916.

Emili, A., J. Greenblatt and J .C. Ingles (1994). “Species-speciﬁc interaction of the
glutamine-rich activation domains of Spl with the TATA box-binding protein.” Mol. Cell.
Biol. 14(3): 1582-1593.

Emili, A. and OJ. Ingles (1995). “Promoter-dependent photocross-linking of the acidic
transcriptional activator E2F-1 to the TATA-binding protein.” J. Biol. Chem. 270(23):
13674-13680.

F elsenfeld, G. (1992). “Chromatin as an essential part of the transcriptional mechanism.”
Nature 355: 219-224.

Ferreri, K., G. Gill and M. Montminy (1994). “The CAMP-regulated transcription factor
CREB interacts with a component of the TFIID complex.” Proc. Natl. Acad. Sci. USA
91: 1210-1213.

Fletcher, TM. and J .C. Hansen (1995). “Core histone tail domains mediate
oligonucleosome folding and nucleosomal DNA organization through distinct
mechanisms.” J. Biol. Chem. 43: 25359-25362.

Folkers, G.E., E.C. Vanheerde and PT. Vandersaag (1995). “Activation function 1 of
retinoic acid receptor beta 2 is an acidic activator resembling VP16.” J. Biol. Chem. 270:
23552-23559.

Gasch, A., A. Hofﬁnann, M. Horikoshi, R.G. Roeder and N.-H. Chua (1990).
“Arabidopsis thaliana contains two genes for TFIID.” Nature 346: 390-3 94.

Ge, H. and RC. Roeder (1994). “Puriﬁcation, cloning, and characterization of a human
coactivator, PC4, that mediates transcriptional activation of class H genes.” C§l_l 78(3):
5 13-523 .

Geiger, J .H., S. Hahn, S. Lee and PB. Sigler (1996). “Crystal structure of the yeast
TFHA/TBP/DNA complex.” Science 272: 830-836.

 

Gerster, T. and R. G. Roeder (198 8). “A herpesvirus trans-activating protein interacts with
transcription factor OTF-l and other cellular proteins.” Proc. Natl. Acad. Sci. USA 85:
63 47 -63 5 1 . .

165

 

Gill, G., E. Pascal, Z.H. Tseng and R. Tjian (1994). “A glutamine-rich hydrophobic patch
in transcription factor Spl contacts the dTAF H1 10 component of the Drosophila TFHD
complex and mediates transcriptional activation.” Proc. Natl. Acad. Sci. USA 91: 192-
196.

Gill, G. and M. Ptashne (1987). “Mutants of GAL4 protein altered in an activation
ﬁinction.” Call 51: 121-126.

Goodrich, J .A., T. Hoey, C.J. Thut, A. Admon and R. Tjian (1993). “Drosophila
TAF(H)40 interacts with both a VP16 activation domain and the basal transcription factor
TFIIB.” Lell 75: 519-530.

Goodrich, J.A. and R. Tjian (1994). “Transcription factors IIE and IHI and ATP
hydrolysis direct promoter clearance by RNA polymerase H.” C_ell 7 7: 145-156.

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 (1997). “Yeast
Gcn5 ﬁrnctions in two multisubunit complexes to acetylate nucleosomal histones:
Characterization of an Ada complex and the SAGA (Spt/Ada) complex.” Genes Dev. 11:
1640-1650.

Grant, P.A., D. Schieltz, M.G. Pray-Grant, D.J. Steger, J .C. Reese, J .R. Yates ID and J.L.
Workman (1998). “A subset of TAFHs are integral components of the SAGA complex
rquired for nucleosome acetylation and transcriptional stimulation.” Ce_ll 94: 45-53.

Gu, W. and R. G. Roeder (1997). “Activation of p53 sequence-speciﬁc DNA binding by
acetylation of the p53 C-terrninal domain.” Ccﬂ 90: 595-606.

Guarente, L., RR. Yocum and P. Gifford (1982). “A GALIO-CYCl hybrid yeast
promoter identiﬁes the GAL4 regulatory region as an upstream site.” Proc. Natl. Acad
Sci. USA 79: 7410-7414.

Hahn, S. (1993). “Structure(?) and function of acidic transcription activators.” Cgﬂ 72(4):
481-483.

Hahn, S., S. Buratowski, P.A. Sharp and L. Guarente (1989). “Isolation of the gene
encoding the yeast TATA binding protein TFIID: A gene identical to the SPT15
suppressor of Ty element insertions.” Ce_ll 58: 1173-1181.

Hampsey, M. (1998). “Molecular genetics of the RNA polymerase H general transcription
machinery.” Microbiology and Molecuﬁrr Biolagy Reviews 62: 465-503.

Hanna-Rose, W. and U. Hansen (1996). “Active Repression mechanisms of eukaryotic
transcription repressors.” Trends Genet. 12: 229-234.

166

Hansen, S.K., S. Takada, RH. Jacobson, J .T. Lis and R. Tjian (1997). “Transcription
properties of a cell type-speciﬁc TATA-binding protein.” Ce_ll 91: 71-83.

Hansen, SK. and R. Tjian (1995). “TAFS and TFHA mediate differential utilization of the
tandem Adh promoters.” Len 82: 565-575.

Hatada, M.H., X. Lu, E.R. Laird, J. Green and J .P. Morgenstem (1995). “Molecular basis
for the interaction of the protein tyrosine kinase ZAP-70 with the T-cell receptor.” Nature
377 : 32-38.

Hayward, GS. (1993). “Immediate-early gene regulation in herpes simplex virus.” Sem.
Virol. 4(1): 239-251.

He, S. and SJ. Weintraub (1998). “Stepwise recruitment of components of the
preinitiation complex by upstream activators in viva.” Mol. Cell. Biol. 18: 2876-2883.

Hernandez, N. (1993). “TBP, a universal eukaryotic transcription factor?” Genes and
Development 7(7b): 1291-1308.

Hisatake, K., S. Hasegawa, R. Takada, Y. Nakatani, M. Horikoshi and RC. Roeder
(1993). “The p250 subunit of native TATA box-binding factor-TFHD is the cell-cycle
regulatory protein-CCGI .” Nature 362(6416): 179-181.

 

Hisatake, K., T. Ohta, R. Takada, M. Guerrnah, M. Horikoshi, Y. Nakatani and RC.
Roeder (1995). “Evolutionary conservation of human TAT A-binding-polypeptide-
associated factors TAF(H)31 and TAF(H)80 and interactions of TAF(H)80 with other

T AF 3 and with general transcription factors.” Proc. Natl. Acad. Sci. USA 92: 8195-8199.

Hoey, T., B.D. Dynlacht, MD. Peterson, B.F. Pugh and R. Tjian (1990). “Isolation and
characterization of the drosophila gene encoding the TATA box binding protein, TFIID.’
Ce_ll 61: 1179-1186.

‘)

Hoey, T., R.O.J. Weinzierl, G. Gill, J .L. Chen, B.D. Dynlacht and R. Tjian (1993).
“Molecular cloning and functional analysis of drosophila TAF l 10 reveal properties
expected of coactivators.” Call 72(2): 247-260.

Hoﬁinann, A., C.M. Chiang, T. Oelgeschlager, X.L. Xie, S.K. Burley, Y. Nakatani and
R. G. Roeder (1996). “A histone octamer-like structure within TFIID.” Nature 380: 356-
359.

Hoffmann, A., E. Sinn, T. Yamamoto, J. Wang, A. Roy, M. Horikoshi and RC. Roeder

(1990). “Highly conserved core domain and unique N terminus with presumptive
regulatory motifs in a human TATA factor (TFIID).” Nature 346: 387-3 90.

167

 

Hope, IA. and K. Struhl (1986). “Functional dissection of a eukaryotic transcriptional
activator protein, GCN4 of yeast.” Lell 46: 885-894.

Horikoshi, M., M.F. Carey, H. Kakidani and RC. Roeder (1988a). “Mechanism of action
of a yeast activator: direct effect of GAL4 derivatives on mammalian TFIID-promoter
interactions.” Ce_ll 54: 665-669.

Horikoshi, M., T. Hai, Y.S. Lin, M.R. Green and RC. Roeder (1988b). “Transcription
factor ATF interacts with the TATA factor to facilitate establishment of a preinitiation
complex.” C_ell 54: 1033-1042.

Horikoshi, M., C.K. Wang, H. Fujii, J .A. Cromlish, P.A. Weil and R.G. Roeder (1989).
“Cloning and structure of a yeast gene encoding a general transcription initiation factor
TFIID that binds to the TATA box.” Nature 341: 299-303.

Horiuchi, J ., N. Silverman, G.A. Marcus and L. Guarente (1995). “ADA3, a putative
transcriptional adaptor, consists of two separable domains and interacts with ADA2 and
GCN5 in a trimeric complex.” Mol. Cell. Biol. 15: 1203-1209.

Huang, M., N. Amheim and M.F. Goodman (1992). “Extension of base mispairs by Taq
DNA polymerase: implications for single nucleotide discrimination in PCR.” Nucleic Acids
M 20: 4567-4573.

Imbalzano, A.N., H. Kwon, M.R. Green and RE. Kingston (1994a). “Facilitated binding
of TATA-binding protein to nucleosomal DNA.” Nature 370: 481-485.

Imbalzano, A.N., K.S. Zaret and RE. Kingston (1994b). “Transcription factor TFIIB and
TFIIA can independently increase the afﬁnity of the TATA-binding protein for DNA.” L
Biol. Chem. 269(11): 8280-8286.

Imhof, A., X. Yang, V.V. Ogryzko, Y. Nakatani, A.P. Wolffe and H. Ge (1997).
“Acetylation of general transcription factors by histone acetyltransferases.” Current Biol.
7: 689-692.

Ing, N.H., J .M. Beekman, S.Y. Tsai, M.J. Tsai and B.W. O'Malley (1992). “Members of
the steroid hormone receptor superfamily interact with TFIIB (S300-II).” J. Biol. Chem.
267(25): 17617-17623.

Ingles, C.J., M. Shales, W.D. Cress, S.J. Triezenberg and J. Greenblatt (1991). “Reduced
binding of TFIID to transcriptionally compromised mutants of VP16.” Nature 351: 588-
590.

Inostroza, J .A., F.H. Merrnelstein, 1. Ha, W.S. Lane and D. Reinberg (1992). “Drl, a

TATA-binding protein-associated phosphoprotein and inhibitor of class H gene
transcription.” C_ell 70: 477-489.

168

 

Inoue, M., M. Isomura, S. Ikegawa, T. Fujiwara, S. Shin, H. Moriya and Y. Nakamura
(1996). “Isolation and characterization of a human cDNA clone (GCN5L1) homologous
to GCN5, a yeast transcription activator.” Cytogenetics Cell Genetics 73: 134-136.

Jackson, B.M., C.M. Drysdale, K. Natarajan and A. G. Hinnebusch (1996). “Identiﬁcation
of seven hydrophobic clusters in GCN4 making redundant contributions to transcriptional
activation.” Mol. Cell. Biol 16: 5557-5571.

Jacq, X., C. Brou, Y. Lutz, I. Davidson, P. Chambon and L. Tora (1994). “Human
TAFII3O is present in a distinct TFIID complex and is required for transcriptional

activation by the estrogen receptor.” C_ell 7 9(1): 107-117.

Janin, J. and C. Chothia (1990). “The structure of protein-protein recognition sites.” L
Biol. Chem. 265: 16027-16030.

 

J anin, J., S. Miller and C. Chothia (1988). “Surface, subunit interfaces and interior of
oligomeric proteins.” J. Mol. Biol. 204: 155-164.

J eong, S. and A. Stein (1994). “Micrococcal nuclease digestion of nuclei reveals extended
nucleosome ladders having anomalous DNA lengths for chromatin assembled on non-
replicating plasmids in transfected cells.” Nucleic Acids Res. 22: 370-3 75.

Jones, S. and J .M. Thornton (1996). “Principles of protein-protein interactions.” Proc.
lﬁrﬂ. Acad. Sci. USA 93: 13-20.

Kamei, Y., L. Xu, T. Heinzel, J. Torchia, R. Kurokawa, B. Gloss, S.C. Lin, RA. Heyman,
D.W. Rose, C.K. Glass and MG. Rosenfeld (1996). “A CBP integrator complex mediates
transcriptional activation and AP-l inhibition by nuclear receptors.” Call 85: 403-414.

Kao, C.C., P.M. Lieberman, M.C. Schmidt, Q. Zhou, R. Pei and A.J. Berk (1990).
“Cloning of a transcriptionally active human TATA binding factor.” Science 248: 1646-
1650.

 

Kaufrnann, J. and ST. Smale (1994). “Direct recognition of initiator elements by a
component of the transcription factor HD complex.” Genes Dev. 8(7): 821-829.

Kauﬁnann, J ., C.P. Verrijzer, J. Shao and ST Smale (1996). “CIF, an essential cofactor
for TFIID-dependent initiator function.” Genes Dev. 10: 873-886.

Kelleher HI, R.J., P.M. Flanagan and RD. Kornberg (1990). “A novel mediator between

activator proteins and the RNA polymerase H transcription apparatus.” Call 61: 1209-
121 5.

169

Kim, T.K., S. Hashimoto, RJ. Kelleher, HI, P.M. Flanagan, RD. Kornberg, M. Horikoshi
and R. G. Roeder (1994). “Effects of activation-defective TBP mutations on transcription
initiation in yeast.” Nature 369: 252-255.

 

Kim, Y.C., J.H. Geiger, S. Hahn and PB. Sigler (1993). “Crystal structure of a yeast
TBP/TATA-Box complex.” Nature 365: 512-520.

Kim, Y.-J., S. Bjorklund, Y. Li, M.H. Sayre and RD. Kornberg (1994). “A multiprotein
mediator of transcriptional activation and its interaction with the c-terrninal repeat domain
of RNA polymerase H.” C_ell 77(4): 599-608.

Kingston, R.E., C.A. Bunker and AN. Irnbalzano (1996). “Repression and activation by
multiprotein complexes that alter chromatin structure.” Genes Dev. 10: 905-920.

Klages, N.a.S., M. (1995). “Stimulation of RNA polymerase H transcription initiation by
recruitment of TBP in viva.” Nature 374: 822-823.

 

Klein, C., Struhl, K. (1994). “Increased recruitment of TATA-binding protein to the
promoter by transcriptional activation domains in viva.” Science 266: 280-282.

 

Klemm, R, J. Goodrich, S. Zhou and R. Tjian (1995). “Molecular cloning and expression
of the 32-kDa subunit of human TFIID reveals interactions with VP16 and TFIIB that
mediate transcriptional activation.” Proc. Natl. Acad. Sci. USA. 92(13): 5788-5792.

Klemm, R.D., J .A. Goodrich, S.L. Zhou and R. Tjian (1995). “Molecular cloning and
expression of the 32-kDa subunit of human TFHD reveals interactions with VP16 and
TFHB that mediate transcriptional activation.” Proc. Natl. Acafd. Sci. USA. 92(13):
5788-5792.

Kobayashi, N., T.G. Boyer and A.J. Berk (1995). “A class of activation domains interacts
directly with TFHA and stimulates TFHA-TFIID-promoter complex assembly.” Mol. Cell.
Biol. 15: 6465-6473.

Kobayashi, N., P]. Horn, S.M. Sullivan, S.J. Triezenberg, T.G. Boyer and A.J. Berk
(1998). “DA-complex assembly activity required for VP16C transcriptional activation.”
Mol. Cell. Biol. 18: 4023-4031.

Kokubo, T., D.W. Gong, S. Yamashita, M. Horikoshi, RG. Roeder and Y. Nakatani
(1993a). “Drosophila 230-kD TFIID subunit, a functional homolog of the human cell
cycle gene product, negatively regulates DNA binding of the TATA box-binding subunit
ofTFIID.” _Ge_nesm 7(6): 1033-1046.

Kokubo, T., D.-W. Gong, J .C. Wooton, M. Horikoshi, RG. Roeder and Y. Nakatani
(1994). “Molecular cloning of the Drosophila TFHD subunits.” Nature 367(3): 484-487.

170

 

 

Kokubo, T., S. Yamashita, M. Horikoshi, RG. Roeder and Y. Nakatani (1993b).
“Interaction between the N-terminal domain of the 23 O-kDa subunit and the TATA-box
binding subunit of TFIID negatively regulates TATA-box binding.” Proc. Natl. Acad. Sci.
LISA 91: 3520-3524.

Koleske, A.J. and RA. Young (1994). “An RNA polymerase H holoenzyme responsive to
activators.” Nature 368: 466-469.

Koleske, A.J. and RA. Young (1995). “The RNA polymerase H holoenzyme and its
implications for gene regulation.” Trends Biochem. Sci. 20: 113-116.

Kretzschmar, M., K. Kaiser, F. Lottspeich and M. Meisteremst (1994). “A novel mediator
of class H gene transcription with homology to viral immediate-early transcriptional
regulators.” Ce_ll 78: 525-534.

Krumm, A, L.B. Hickey and M. Groudine (1995). “Protomer—proximal pausing of RNA
polymerase H deﬁnes a general rate-limiting step after transcriptional initiation.” Genes
and Dev. 9: 559-572.

Krumm, A., T. Meulia, M. Brunvand and M. Groudine (1992). “The block to
transcriptional elongation within the human c-myc gene is determined in the promoter -
proximal region.” Genes and Development 6(11): 2201-2213.

Kumar, K.P., S. Akouliychev and D. Reinberg (1998). “Promoter-proximal stalling results
from the inability to recruit transcription factor III-I to the transcription complex and is a
regulated event.” Proc. Natl. Acad. Sci. USA 95: 9767-9772.

Kunkel, TA. (1985). “Rapid and eﬁicient site-speciﬁc mutagenesis without phenotypic
selection.” Proc. Natl. Acad. Sci. USA 82: 488-492.

Kunzler, M., G.H. Braus, O. Georgiev, K. Seipel and W. Schaffirer (1994). “Functional
differences beteen mammalian transcription activation domains at the yeast Gall
promoter.” EMBO J. 13(3): 641-645.

Kuriyan, J. and D. Cowbum (1997). “Modular peptide recognition domains in eukaryotic
signaling.” Annu. Rev. Biophys. Biomol. Struct. 26: 259-288.

Kussie, P.H., S. Gorina, V. Marechal, B. Elenbaas, J. Moreau, A.J. Levine and NP.
Pavletich (1996). “Structure of the MDM2 oncoprotein bound to the p53 tumor
suppressor transactivation domain.” Science 274: 948-953.

 

Kwok, RP.S., J .R. Lundblad, J.C. Chrivia, J .P. Richards, H.P. Bachinger, RG. Brennan,
S.G.E. Roberts, M.R Green and RH. Goodman (1994). “Nuclear protein CBP is a
coactivator for the transcription factor CREB.” Nature 370: 223 -226.

171

 

Lagrange, T., A.N. Kapanidis, H. Tang, D. Reinberg and RH. Ebright (1998). “New core
promoter element in RNA polymerase H-dependent transcription: sequence-speciﬁc DNA
binding by transcription factor HB.” Genes Dev. 12: 34-44.

Layboum, RJ. and ME. Dahmus (1990). “Phosphorylation of RNA polymerase IIA
occurs subsequent to interaction with the promoter and before the initiation of
transcription.” J. Biol. Chem. 265: 13165-13173.

Lee, D.Y., J .J . Hayes, D. Pruss and AP. Wolffe (1993). “A positive role for histone
acetylation in transcription factor access to nucleosomal DNA.” Call 72(1): 73-84.

Lee, M. and K. Struhl (1995). “Mutations on the DNA-binding surface of TATA-binding
protein can speciﬁcally impair the response to acidic activators in viva.” Mol. Cell. Biol.
15: 5461-5469.

Lee, M. and K. Struhl (1997). “A severely defective TATA-binding protein-TFHB
interaction does not preclude transcriptional activation in viva.” Mol. Cell. Biol. 17: 1336-
1345.

Lee, TI. and RA. Young (1998). “Regulation of gene expression by TBP-associated
proteins.” Genes Dev. 12: 1398-1408.

Lemesle-Varloot, L., B. Henrissat, C. Gaboriaud, V. Bissery, A. Morgat and J .-P. Mornon
(1990). “Hydrophobic cluster analysis: procedures to derive structural and functional
information ﬁ'om 2-D-representation of protein sequences.” Biochimie 72: 555-574.

Leung, D.W., B. Chen and D.V. Goeddel (1989). “A method for random mutagenesis of a
deﬁned DNA segment using a modiﬁed polymerase chain reaction.” Technique 1: 11-15.

Leuther, K.K., J.M. Salmeron and SA. Johnston (1993). “Genetic evidence that an
activation domain of GAL4 does not require acidity and may form a beta sheet.” Cell
72(4): 575-585.

Lieberman, P. (1994). “Identiﬁcation of functional targets of the Zta transcriptional
activator by formation of stable preinition complex intermediates.” Mol. Cell. Biol.
14(12): 8365-8375.

Lieberman, RM. and A.J. Berk (1991). “The Zta trans-activator protein stabilizes TFIID
association with promoter DNA by direct protein-protein interaction.” Genes Dev. 5:
2441-2454.

Lieberman, PM. and A.J. Berk (1994). “A mechanism for TAFS in transcriptional
activation: activation domain enhancement of TFIID-TFHA-promoter DNA complex
formation.” Genes Dev. 8: 995-1006.

172

Lin, J ., J. Chen, B. Elenbaas and A.J. Levine (1994). “Several hydrophobic amino acids in
the p53 amino-terminal domain are required for transcriptional activation, binding to
mdm-2 and the adenovirus 5 ElB 55-kD protein.” Genes Dev. 8(10): 1235-1246.

Lin, Y.S., M.F. Carey, M. Ptashne and MR. Green (1988). “GAL4 derivatives ﬁmction
alone and synergistically with mammalian activators in vitro.” C_ell 54: 659-664.

Lin, Y.S. and MR. Green (1991). “Mechanism of action of an acidic transcriptional
activator in vitro.” C_ell 64: 971-981.

Lin, Y.S., 1. Ha, E. Maldonado, D. Reinberg and MR. Green (1991). “Binding of general
transcription factor TFIIB to an acidic activating region.” Nature 353: 569-5 71.

Liu, D.J., 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(H)230 complex:
Protein mimicry of the minor groove surface of the TATA box unwound by TBP.” Call
94: 573-583.

Logie, C. and CL. Peterson (1997). “Catalytic activity of the yeast SWI/SNF complex on
reconstituted nucleosome arrays.” EMBO J 16(22): 6772-6782.

Logie, CL, C. Tse, J .C. Hansen and CL. Peterson (1998). “The core histone N-terrninal
domains are required for catalytic chromatin remodeling by the SWI/SNF and RSC
complexes.” : submitted.

Loidl, P. (1994). “Histone acetylation. Facts and questions.” Chromosoma 103: 441-449.

Lu, H. and A.J. Levine (1995). “Human TAF(II) 31 protein is a transcriptional coactivator
of the p53 protein.” Proc. Neil. Acad. Sci. USA 92(11): 5154-5158.

Luger, K., Maeder, A.W., Richmond, R.K., Sargent, DE, and Richmond, TJ. (1997).
“Crystal structure of the nucleosome core particle at 2.8 angstrom resolution.” Nature
389: 251-260.

Ma, D., H. Watanbe, F . Merrnelstein, A. Admon, K. Oguri, X. Sun, T. Wada, T. Imai, S.
Toshiﬁimi, D. Reinbeg and H. Handa (1993). “Isolation of a cDNA encoding the largest
subunit of TFHA reveals functions important for activated transcription.” Genes Devel 7:
2246-2257.

Ma, J. and M. Ptashne (1987). “Deletion analysis of GAL4 deﬁnes two transcriptional
activating segments.” C_ell 48: 847-853.

Maignan, S., J .-P. Guilloteau, N. Fromage, B. Amoux, J. Becquart and A. Decruix
(1995). “Crystal structure of the mammalian GRB2 adaptor.” Science 268: 291-293.

 

173

Maldonado, E., 1. Ha, P. Cortes, L. Weis and D. Reinberg (1990). “Factors involved in
speciﬁc transcription by mammalian RNA polymerase H: role of transcription factors HA,
HD, and IIB during formation of a transcription-competent complex.” Mol. Cell. Biol. 10:
633 5-6347.

Maldonado, E., R. Shiekhattar, M. Sheldon, H. Cho, R. Drapkin, P. Rickert, E. Lees,
A.C. W., S. Linn and D. Reinberg (1996). “A human RNA polymerase H complex
associated with SRB and DNA-repair proteins.” Nature 381: 86-89.

Marcus, G.A., J. Horiuchi, N. Silverman and L. Guarente (1996). “ADA5/SPT20 links the
ADA and SPT genes, which are involved in yeast transcription.” Mol. Cell. Biol. 16:
3 197-3 205.

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: 4807-4815.

Martinez, E., C. Chiang, H. Ge and RG. Roeder (1994). “TATA-binding protein-
associated factor(s) in TFIID function through the initiator to direct basal transcription
from a TATA-less class H promoter.” EMBO J 13: 3115-3126.

Martinez, E., H. Ge, Y. Tao, C. Yuan, V. Palhan and R. g. Roeder (1998a). “Novel
cofactors and TFHA mediate functional core promoter selectivity by the human
TAFH150-containing TFIID complex.” Mol. Cell. Biol. 18: 6571-6583.

Martinez, E., T.K. Kundu, J. Fu and RG. Roeder (1998b). “A human SPT3-TAF(H)31-
GCNS-L acetylase complex distinct from transcription factor IID.” J. Biol. Chem. 273:
23781-23785.

McInemey, B.M., D.W. Rose, S.E. Flynn, S. Westin, T. Mullen, A. Krones, J. Inostroza,
J. Torchia, R. Nolte, N. Assa-Munt, M.V. Milbum, C.K. Glass and MG. Rosenfeld
(1998). “Determinants of coactivator LXXLL motif speciﬁcity in nuclear receptor
transcriptional activation.” Genes and Dev. 12: 3357-3368.

Merino, A., K.R Madden, W.S. Lane, J.J. Champoux and D. Reinberg (1993). “DNA
topoisomerase I is involved in both repression and activation of transcription.” Nature
365: 227-234.

Michel, B., P. Komarnitsky and S. Buratowski (1998). “Histone-like TAFS are essential
for transcription in viva.” Mol. Cell 2: 663-673.

Miller, J .H. (1972). Experiments in Molecular Genetic; Cold Spring Harbor, New York.

Mitchell, RJ. and R. Tjian (1989). “Transcriptional regulation in mammalian cells by
sequence-speciﬁc DNA binding proteins.” Science 245: 371-378.

174

 

Mizzen, C.A., X. Yang, T. Kokubo, J .E. Brownell, A.J. Bannister, T. Owen-Hughes, J.
Workman, L. Wang, S.L. Berger, T. Kouzarides, Y. Nakatani and CD. Allis (1996). “The
TAFHZSO subunit of TFIID has histone acetyltransferase activity.” Cell 87: 1261-1270.

Moore, P.A., S.M. Ruben and CA. Rosen (1993). “Conservation of transcriptional
activation functions of the NF-kB p50 and p65 subunits in mammalian cells and
Saccharomyces cerevisiae.” Mol. Cell. Biol. 13: 1666-1674.

Moqtaderi, Z., Y. Bai, D. Poon, P.A. Weil and K. Struhl (1996). “TBP-associated factors
are not generally required for transcriptional activation in yeast.” Nature 383: 188-191.

Muhich, M.L., C.T. Iida, M. Horikoshi, RG. Roeder and CS. Parker (1990). “cDNA
clone encoding drosophila transcription factor-TFIID.” Proc. hilt]. Acad. Sci. USA 87:
9148-9152.

Naar, A.M., P.A. Beaurang, K.M. Robinson, J .D. Oliner, D. Avizonis, S. Scheek, J.
Zwicker, J .T. Kadonaga and R Tjian (1998). “Chromatin, TAFS, and a novel multiprotein
coactivator are required for synergistic activation by Spl and SREBP-la in vitro.” Genes
and Dev. 12: 3020-3031.

Nakajima, T., C. Uchida, S.F. Anderson, J .D. Parvin and M. Montminy (1997). “Analysis
of a CAMP-responsive activator reveals a two-component mechanism for transcriptional
induction via signal-dependent factors.” Genes and Dev. 11: 73 8-747.

Nielsen, D.A., T. Chang and DJ. Shapiro (1989). “A highly sensitive mixed phase assay
for chloramphenicol acetyltransferase activity in transfected cells.” Anal. Biochem. 179:
12-23.

Nikolov, DB. and SK. Burley (1997). “RNA polymerase H transcription initiation: a
structural view.” Proc. Natl. Acad. Sci. 94: 15-22.

 

Nikolov, D.B., H. Chen, E.D. Halay, A. Hoffrnann, RG. Roeder and SK. Burley (1996).
“Crystal structure of a human TATA box-binding protein/TATA element complex.” Proc.
Nat. Acad. Sci. USA 93: 4862-4867.

Nikolov, D.B., H. Chen, E.D. Halay, A.A. Usheva, K. Hisatake, D.K. Lee, RG. Roeder
and SK. Burley (1995). “Crystal structure of a TFIIB-TBP-TATA-element ternary
complex.” Nature 377 : 119-128.

 

Nikolov, D.B., S. Hu, J. Lin, A. Gasch, A. Hoffrnann, M. Horikoshi, N. Chua, RG.
Roeder and SK. Burley (1992). “Crystal structure of TFHD TATA-box binding protein.”
Nature 360: 40-46.

175

 

Nishikawa, J ., T. Kokubo, M. Horikoshi, RG. Roeder and Y. Nakatani (1997).
“Drosophila TAF(H)230 and the transcriptional activator VP16 bind competitively to the
TATA box-binding domain of the TATA box-binding protien.” Proc. Natl. Acad. Sci. 94:
85-90.

Nolte, R.T., G.B. Wisely, S. Westin, J .E. Cobbs, M.H. Lambert, R. Kurokawa, M.G.

Rosenfeld, T.M. Willson, C.K. Glass and M.V. Milburn (1998). “Ligand binding and co-
activator assembly of the peroxisome proliferator-activated receptor-gamma.” Nature 395:
137-143.

O'Brien, T. and J .T. Lis (1991). “RNA polymerase H pauses at the 5'end of the
transcriptionally induced Drosophila hsp70 gene.” Mol. Cell. Biol. 11: 5285-5290.

Oelgeschlager, T., C.M. Chiang and RG. Roeder (1996). “Topology and reorganization
of a human TFIID-promoter complex.” Nature 382: 73 5-738.

 

Ogryzko, V.V., R.L. Schiltz, V. Russanova, B.H. Howard and Y. Nakatani (1996). “The
transcriptional coactivators p300 and CBP are histone acetyltransferases.” Ce_ll 87: 953-
959.

Orphanides, G., T. Lagrange and D. Reinberg (1996). “The general transcription factors
of RNA polymerase H.” Genes Dev. 10: 2657-2683.

Oshea-Greenﬁeld, A. and ST. Smale (1992). “Roles of TATA and initiator elements in
determining the start site location and direction of RNA polymerase H transcription.” J_.
Biol. Chem. 267: 1391-1402.

Ozer, J ., A.H. Bolden and RM. Lieberman (1996). “Transcription factor HA mutations
show activator-speciﬁc defects and reveal a HA function distinct ﬁ'om stimulation of
TBA-DNA binding.” J. Biol. Chem. 271: 11182-1 1 190.

Ozer, J ., K. Mitsouras, D. Zerby, M. Carey and RM. Lieberman (1998). “Transcription
factor HA derepresses TATA-binding protein (TBP)-associated factor inhibition of TBP-
DNA binding.” J. Biol. Chem. 273: 14293-14300.

Ozer, J ., P.A. Moore, A.H. Bolden, A. Lee, C.A. Rosen and PM. Lieberman (1994).
“Molecular cloning of the small (gamma) subunit of human TFHA reveals ﬁmctions
critical for activated transcription.” Genes Dev. 8(19): 2324-2335.

Paal, K., P.A. Baeuerle and ML. Schmitz (1997). “Basal transcription factors TBP and

TFIIB and the viral coactivator ElA 13S bind with distinct afﬁnities and kinetics to the
transactivation domain of NF-kappa B p65.” Nucleic Acids Res. 25: 1050-1055.

176

 

Payne, J .M., R]. Laybourn and ME. Dahmus (1989). “The transition of RNA polymerase
H from initiation to elongation is associated with phosphorylation of the carboxyl-terminal
domain of subunit Ha.” J. Biol. Chem. 264: 19621-19629.

Peterson, CL, A. Dingwall and MP. Scott (1994). “Five SWI/SNF gene products are
components of a large multisubunit complex required for transcriptional enhancement.”
Proc. Neal. Acad. Sci. USA 91: 2905-2908.

Peterson, CL. and I. Herskowitz (1992). “Characterization of the yeast SW11, SW12 and
SW13 genes, which encode a global activator of transcription.” C_ell 68: 573-583.

Peterson, M.G., N. Tanese, B.F. Pugh and R. Tjian (1990). “Functional domains and
upstream activation properties of cloned human TATA binding protein.” Science 248:
1625-1630.

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 similar to and
different from those of ADA2.” Mol. Cell. Biol. 13(10): 5981-5989.

Pinto, 1., DE. Ware and M. Hampsey (1992). “The yeast SUA7 gene encodes a homolog
of human transcription factor TFIIB and is required for normal start site selection in
vivo.” C_eﬂ 68: 977-988.

Poon, D., Y. Bai, A.M. Campbell, S. Bjorklund, Y]. Kim, S. Zhou, RD. Kornberg and
RA. Weil (1995). “Identiﬁcation and characterization of a TFHD-like multiprotein
complex from Saccharomyces cerevisiae.” Proc. Natl. Acad. Sci. USA 92: 8224-8228.

Preston, C.M., M.C. Frame and M.E.M. Campbell (1988). “A complex formed between
cell components and an HSV structural polypeptide binds to a viral immediate early gene
regulatory DNA sequence.” C_ell 52: 425-434.

Ptashne, M. (1988). “How eukaryotic transcriptional activators work.” Nature 335: 683-
689.

Ptashne, M. and A. Gann (1997). “Transcriptional activation by recruitment.” Nature 386:
569-577.

Pugh, BF. and R. Tjian (1990). “Mechanism of transcriptional activation by Spl: evidence
for coactivators.” Q11 61: 1187-1197.

Radhakrishnan, I., GO Perez-Alvarado, D. Parker, H.J. Dyson, M.R. Montminy and PE.
Wright (1997). “Solution structure of the KIX domain of CBP bound to the
transactivation domain of CREB: A model for activatorzcoactivator interactions.” C_eﬂ 91:
741-752.

177

 

 

Ramboarina, S., N. Morellet, P. Petitjean, B.P. Roques and MC. Fournie-Zaluski (1998).

“Model of the complex formed between the H2 domain of the TATA box binidng protein

and the L(281-301) fragment of the human transcription factor TFIIA.” Protein Engng 11:
729-738.

Ranish, J .A., W.S. Lane and S. Hahn (1992). “Isolation of two genes that encode subunits
of the yeast transcription factor IIA.” Science 255: 1127-1129.

Reddy, P. and S. Hahn (1991). “Dominant negative mutations in yeast TFIID deﬁne a
bipartite DNA binding region.” Q 65: 349-357.

Reese, J .C., Apone, L., Walker, S. S., Grifﬁn, L. A., Green, M. R. (1994). “Yeast
TAF(II)s in a multisubunit complex required for activated transcription .” Nature
371(6497): 523-527.

Regier, J .L., F. Shen and SJ. Triezenberg (1993). “Pattern of aromatic and hydrophobic
amino acids critical for one of two subdomains of the VP16 transcriptional activator.”
Proc. Natl. Acad. Sci. USA 90: 883-887.

Reinberg, D., M. Horikoshi and RG. Roeder (1987). “Factors involved in speciﬁc
transcription by mammalian RNA polymerase 11. Functional analysis of initiation factors
HA and HD and identiﬁcation of a new factor operating at sequences downstream of the
initiation site.” J. Biol. Chem. 262: 3322-3330.

Roberts, S.G.E., 1. Ha, E. Maldonado, D. Reinberg and MR. Green (1993). “Interaction
between an acidic activator and transcription factor IIB is required for transcriptional
activation.” Nature 363(6431): 741-744.

 

Roberts, SM. and F. Winston (1996). “SPT20/ADA5 encodes a novel protein ﬁrnctionally
related to the TATA-binding protein and important for the transcription in Saccharomyces
cerevisiae.” Mol. Cell. Biol. 16: 3206-3213.

Roeder, RG. ( 1991). “The complexities of eukaryotic transcription initiation: regulation
of preinitiation complex assembly.” TIBS 16: 402-407.

Roeder, RG. (1996). “The role of general initiation factors in transcription by RNA
polymerase 11.” Trends Biochem. Sci. 21: 327-335.

Roy, A.L., S. Malik, M. Meisteremst and RG. Roeder (1993). “An alternative pathway
for transcription initiation involving TFII-I.” Nature 365(6444): 355-3 59.

 

Ruden, D.M., J. Ma, Y. Li, K. Wood and M. Ptashne (1991). “Generating yeast
transcriptional activators containing no yeast protein sequences.” Nature 21: 250-252.

 

178

 

 

 

Ruppert, S., B.H. Wang and R Tjian (1993). “Cloning and expression of human
TAF(H)250 - a TBP-associated factor implicated in cell-cycle regulation.” Nature
362(6416): 175-179.

Sadowski, I., J. Ma, S. Triezenberg and M. Ptashne (1988). “GAL4-VP16 is an unusually
potent transcriptional activator.” Nature 335: 563-564.

 

Sakaguchi, K., J.E. Herrera, S. Saito, T. Miki, M. Bustin, A. Vassilev, C.W. Anderson
and E. Appella (1998). “DNA damage activates p53 through a phosphorylation-
acetylation cascade.” Genes and Dev. 12: 2831-2841.

Sauer, F., S.K. Hansen and R. Tjian (1995). “Multiple TAFIIs direction synergistic
activation of transcription.” Science 270: 1783-1788.

 

Sauer, F., D.A. Wassarrnan, G.M. Rubin and R. Tjian (1996). “TAFIIs mediate activation
of transcription in the Drosophila embryo.” Cell 87 : 1271-1284.

Sawadogo, M. and R. G. Roeder (1985). “Interaction of a gene-speciﬁc transcription
factor with the adenovirus major late promoter upstream of the TATA box region.” Ceﬂ
43: 165-175.

Schmidt, M.C., C.,C. Kao, R. Pei and A.J. Berk (1989). “Yeast TATA-box transcription
factor gene.” Proc. Md. Acad. Sci. USA 86: 7785-7789.

Schmitz, M., M. Silva and P. Baeuerle (1995). “Transactivation domain 2 (TA(2)) of p65
NF-kappa B-Sirnilarity to TA(1) and phorbol ester-stimulated activity and phosphorylation
in intact cells.” J. of Bio. Chemistg 270(26): 15576-15584.

Schmitz, M.L., M.A. dos Santos Silva, H. Altmann, M. Czisch, T.A. Holak and RA.
Baeuerle (1994). “Structural and ﬁrnctional analysis of the NF-kappaB p65 C terminus: an
acidic and modular transactivation domain with the potential to adopt an alpha-helical
conformation.” J. Biol. Chem. 269: 25613-25620.

Seipel, K., O. Geogriev and W. Schaffner (1994). “A minimal transcription activation
domain consisting of a speciﬁc array of aspartic acid and leucine residues.” Biol. Chem.
Hoppe—Seyler 375(7): 463-470.

Seipel, K., O. Georgiev and W. Schaffner (1992). “Different activation domains stimulate
transcription from remote (enhancer) and proximal (promoter) positions.” EMBO J.
11(13): 4961-4968.

Sekiguchi, T., T. Miyata and T. Nishimoto (1988). “Molecular cloning of the cDNA of a

human X chromosomal gene (CCGI) which complements the temperature-sensitive G1
mutants, tsBN462 and ts13, of the BHK cell line.” EMBO J 7: 1683-1687.

179

Sekiguchi, T., Y. Nohiro, Y. Nalamura, N. Hisamoto and T. Nishimoto (1991). “The
human CCGI gene, essential for progression of the G1 phase, encodes a 210 kilodalton
nuclear DNA-binding protein.” Mol. Cell. Biol. 11: 3317-3325.

Seto, E., A. Usheva, G.P. Zambetti, J. Momand, N. Horikoshi, R. Weinmann, A.J. Levine
and T. Shenk (1992). “Wild-type p53 binds to the TATA-binding protein and represses
transcription.” Procﬂatl. Acad. Sci. USA 89(24): 12028-12032.

Siebenlist, U., G. Franzoso and K. Brown (1994). “Structure, regulation and ﬁinction of
NF-kB.” Annp. Rev. Cell Biol. 10: 405-455.

Sigler, P. (1988). “Acid blobs and negative noodles.” Nature 333: 210-212.

 

Smale, ST (1997). “Transcription initiation from TATA-less promoters within eukaryotic
protein-coding genes.” Biochim. et Biophys. Acta 1351: 73-88.

Smith, CL. and G.L. Hager (1997). “Transcriptional regulation of mammalian genes in
viva.” J. Biol. Chem. 272: 27493-27496.

Spencer, CA. and M. Groudine (1990). “Transcription elongation and eukaryotic gene
regulation.” Oncogene 5: 777-786.

Stargell, LA. and K. Struhl (1995). “The TBP-TFHA interaction in the response to acidic
activators in viva.” Science 269: 75-78.

Stern, M., Jensen, R, Herskowitz, I. (1984). “Five SWI genes are required for expression
of the H0 gene in yeast.” J Mol. Biol. 178(4): 853-868.

Stoscheck, CM. (1990). “Quantitation of protein.” Methods Enzymol 182: 50-68.

Stringer, K.F., C.J. Ingles and J. Greenblatt (1990). “Direct and selective binding of an
acidic transcriptional activation domain to the TATA-box factor TFHD.” Nature 345:
783-786.

Struhl, K. (1998). “Histone acetylation and transcriptional regulatory mechanisms.” Genes
and Dev. 12: 599-606.

Sullivan, S.M., P.J. Horn, V.A. Olson, A.H. Koop, W. Nu, RH. Ebright and SJ.
Triezenberg (1998). “Mutational analysis of a transcriptional activation region of the
VP16 protein of herpes simplex virus.” Nucleic Acids Res.: submitted.

Sun, X., D. Ma, M. Sheldon, K. Yeung and D. Reinberg (1994). “Reconstitution of
human TFHA activity ﬁ'om recombinant polypeptides: a role in TFHA-mediated
transcription.” Genes Dev 8(19): 2336-2348.

180

 

Taggart, A.K.P. and BF. Pugh (1996). “Dimerization of TFHD when not bound to
DNA.” Science 272: 1331-1333.

 

Tan, S., Y. Hunziker, D.F. Sargent and TI. Richmond (1996). “Crystal structure of a
yeast TFHA/TBP/DNA complex.” Nature 381: 127-134.

 

Tanaka, M. and W. Herr (1994). “Reconstitution of transcriptional activation domains by
reiteration of short peptide segments reveals the modular organization of a glutamine-rich
activation domain.” Mol. Cell. Biol. 14(9): 6056-6067.

Tang, H., X. Sun, D. Reinberg and RB. Ebright (1996). “Protein-protein interactions in
eukaryotic transcription initiation: structure of the preinitiation complex.” Proc. Natl.
Acad. Sci. USA 93: 1119-1124.

Tansey, WP. and W. Herr (1997a). “Selective use of TBP and TFIIB revealed by a
TATA-TBP-TFIIB array with altered speciﬁcity.” Science 7: 829-831.

Tansey, WP. and W. Herr (1997b). “TAFS; Guilt by association.” C_ell 88: 729-732.

Tasset, D., L. Tora, C. Fromental, E. Scheer and P. Chambon (1990). “Distinct classes of
transcriptional activating domains function by different mechanisms.” Cgﬂ 62: 1177-1187.

Thompson, C.M., A.J. Koleske, D.M. Chao and RA. Young (1993). “A multisubunit
complex associated with the RNA polymerase H CTD and TATA-Binding protein in
yeast.” Call 73: 1361-1375.

Thut, C.J., J. Chen, R. Klemm and R. Tjian (1995). “p53 transcriptional activation
mediated by coactivators TAF1140 and TAFII60.” Science 267: 100-104.

 

Triezenberg, SJ. (1995). “Structure and ﬁinction of transcriptional activation domains.”
Curr. Opin. Genetics Dev. 5: 190-196.

Triezenberg, S.J., RC. Kingsbury and S.L. McKnight (1988). “Functional dissection of
VP16, the trans-activator of herpes simplex virus immediate early gene expression.”
Genes Dev. 2: 718-729.

Turner, BM. (1991). “Histone acetylation and and control of gene expression.” J. Cell
_S_cl 99: 13-20.

Usheva, A. and T. Shenk (1994). “TATA-binding protein-independent initiation: YYI,

TFHB, and RNA polymerase H direct basal transcription on supercoiled template DNA.”
C_ell 76: 1115-1121.

181

 

Utley, RT., K. Ikeda, P.A. Grant, J. Cote, D]. Steger, A. Eberharter, S. John and J .L.
Workman (1998). “Transcriptional activators direct histone acetyltransferase complexes to
nucleosomes.” Nature 394: 498-502.

Verrijzer, C.P., J .-L. Chen, K. Yokomori and R. Tjian (1995). “Binding of TAFS to core
elements directs promoter selectivity by RNA polymerase H.” Ce_ll 81(7): 1115-1125.

Verrijzer, C.P., K. Yokomori, J .L. Chen and R. Tjian (1994). “Drosophila TAFH 150:
similarity to yeast gene TSM-l and speciﬁc binding to core promoter DNA.” Science 264:
933-941.

 

Walker, 8., R. Greaves and P. O'Hare (1993). “Transcriptional activation by the acidic
domain of Vmw65 requires the integrity of the domain and involves additional
determinants distinct from those necessary for TFIIB binding.” Mol. Cell. Biol. 13(9):
5233-5244.

Walker, 8.8., J .C. Reese, L.M. Apone and MR. Green (1996). “Transcriptional activation
in cells lacking TAF(H)s.” Nature 383: 185-188.

 

Wang, W.D., M. Carey and J .D. Gralla (1992). “Polymerase-H promoter activation -
closed complex formation and ATP-driven start site opening.” Science 255: 450-453.

Weinheirner, S.P., B.A. Boyd, S.K. Durham, J .L. Resnick and DR. O'Boyle (1992).
“Deletion of the VP16 open reading frame of herpes simplex virus type 1.” J. Virol. 66:
25 8-269.

Weinzierl, R.O.J., S. Ruppert, B.D. Dynlacht, N. Tanese and R. Tjian (1993). “Cloning
and expression of Drosophila TAF 1170 reveal conserved interactions with other subunits

ofTFIID.” EMBO J 12: 5303-5309.

White, J ., C. Brou, J. Wu, Y. Lutz, V. Moncollin and P. Chambon (1992). “The acidic
transcriptional activator GAL-VP16 acts on preformed template-committed complexes.”
EMBO J. 11: 2229-2240.

Winston, F. and M. Carlson (1992). “Yeast SNF/SWI transcriptional activtors and the
SPT/SIN chromatin connection.” Trends Genet. 8(11): 387-391.

Wolffe, A. (1994). “Nucleosome positioning and modiﬁcation: chromatin structures that
potentiate transcription.” Trends Biol. Sci. 19: 240-244.

Workman, J.L. and RE. Kingston (1998). “Alteration of nucleosome structure as a
mechanism of transcriptional regulation.” Annp. Rev. Biochem. 67: 545-579.

Wu, 8., E. Kershnar and C. Chiang (1998). “TAFH-independent activation mediated by
human TBP in the presence of the positive cofactor PC4.” EMBO J 17: 4478-4490.

182

 

 

 

 

 

Wu, Y., RJ. Reece and M. Ptashne (1996). “Quantitation of putative activator-target
afﬁnities predicts transcriptional activating potentials.” EMBO J. 15: 3951-3963.

Xiao, H., A. Pearson, B. Coulombe, R. Truant, S. Zhang, J .L. Regier, S.J. Triezenberg, D.
Reinberg, O. Flores, C.J. Ingles and J. Greenblatt (1994). “Binding of basal transcription
factor TFIIH to the acidic activation domains of VP16 and p53.” Mol. Cell. Biol. 14(10):
7013-7024.

Xie, X.L., T. Kokubo, S.L. Cohen, U.A. Mirza, A. Hoffrnann, B.T. Chait, RG. Roeder,
Y. Nakatani and SK. Burley (1996). “Structural similarity between TAFS and the
heterotetrameric core of the histone octamer.” Nature 380: 316-322.

Xu, X., C. Prorock, H. Ishikawa, E. Maldonado, Y. Ito and C. Gelinas (1993).
“Functional interaction of the v-Rel and c-Rel oncoproteins with the TATA-binding
protein and association with transcription factor-HB.” Mol. Cell. Biol. 13(11): 6733-6741.

Yamamoto, T., M. Horikoshi, J. Wang, S. Hasegawa, P.A. Weil and RG. Roeder (1992).
“A bipartite DNA binding domain composed of direct repeats in the TATA box binding
factor TFHD.” Proc. Natl. Acad. Sci. USA 89: 2844-2848.

Yang, X.-J., V.V. Ogryzko, J.-I. Nishikawa, B.H. Howard and Y. Nakatani (1996). “A
p3 00/CBP-associated factor that competes with the adenoviral oncoprotein ElA.” Nature
382: 319-324. -

Yankulov, K., J. Blau, T. Purton, S. Roberts and D.L. Bentley (1994). “Transcriptional
elongation by RNA polymerase II is stimulated by transactivators.” Ce_ll 77(5): 749-759.

Yeung, K.C., Inostroza, J. A., Merrnelstein, F. H., Kannabiran, C., Reinberg, D. (1994).
“Structure-ﬁrnction analysis of the TBP-binding protein Drl reveals a mechanism for
repression of class H gene transcription.” Genes and Development 8(17): 2097-2109.

Yokomori, K., A. Admon, J .A. Goodrich, J. Chen and R. Tjian (1993). “Drosophila
TFIIA-L is processed into two subunits that are associated with the TBP/TAF complex.”
Genes Dev. 7: 2235-2245.

Yokomori, K., M.P. Zeidler, J. Chen, C.P. Verrijzer, M. Mlodzik and R. Tjian (1994).
“Drosophila TFHA directs cooperative DNA binding with TBP and mediates
transcriptional activation.” M 8(19): 2313-2323.

Yuan, W., G. Condorelli, M. Caruso, A. Felsani and A. Giordano (1996). “Human p300

protein is a coactivator for the transcription factor MyoD.” J. Biol. Chem. 271: 9009-
901 3 .

183

 

Zawel, L. and D. Reinberg (1993). “Initiation of transcription by RNA polymerase II: a
multi-step process.” Prog. Nucleic Acids Res. Mol. Biol. 44: 67-108.

Zhong, H., H. SuYang, H. Erdjument-Bromage, P. Tempst and S. Ghosh (1997). “The
transcriptional activity of NF-kappaB is regulated by the IkappaB-associated PKAc
subunit through a cyclic AMP-independent mechanism.” Ccﬂ 89: 413-424.

Zhong, H., RE. V011 and S. Ghosh (1998). “Phosphorylation of NF-kB p65 by PKA
stimulates transcriptional activity by promoting a novel bivalent interaction with with the
coactivator CBP/p300.” Mol. Cell 1: 661-671.

Zhou, R, P.M. Lieberman, T.G. Boyer and A.J. Berk (1992). “Holo-TFIID supports
transcriptional stimulation by diverse activators and ﬁom a TATA-less promoter.” Genes
Dev. 6: 1964-1974.

Zhou, Q., T.G. Boyer and A.J. Berk (1993). “Factors (TAFS) required for activated
transcription interact with TATA box-binding protein conserved core domain.” Genes
Dev. 7(2): 180-187.

Zhu, H., V. Joliet and R. Prywes (1994). “Role of transcription factor TFHF in serum
response factor-activated transcription.” J. Biol. Chem. 269: 3489-3497.

Zwilling, S., A. Annweiler and T. Wirth (1994). “The POU domains of the Octl and 0th
transcription factors mediate speciﬁc interaction with TBP.” Nucleic Acids Res. 22: 1655-
1662.

184

 

 

UNIV
1'11'11111'.
3020

1 {H

11
1'1
3129

1' 1

NICHIGQN STQTE