.N

sen

 

’1 ”~70

 

 

 

 

 

 
 

 

2 - 4 LIBRARY
2 0'54 Michigan State
University

This is to certify that the
dissertation entitled

CHARACTERIZATION OF THE ARABIDOPSIS THALIANA
CBF1 TRANSCRIPTION FACTOR: FUNCTIONAL ROLE OF
TWO EVOLUTIONARILY CONSERVED SIGNATURE
SEQUENCES

presented by

Donatella Canella

has been accepted towards fulfillment
of the requirements for the

Doctoral degree in Cell and Molecular Biologx

 

Q:
“4/ 4/74 4A

/ 'Major P‘Fofessor’s Signature
///7- 6A ‘7"

ate

MSU is an affinnative-acfion, equal-opportunity employer

 

 

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

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6/07 p:/CIRCIDaIeDue.indd-p.1

CHARACTERIZATION OF THE ARABIDOPSIS THALIANA CBF1
TRANSCRIPTION FACTOR:
FUNCTIONAL ROLE OF TWO EVOLUTIONARILY CONSERVED SIGNATURE
SEQUENCES

By

Donatella Canella

A DISSERTATION

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

DOCTOR OF PHILOSOPHY
Program in Cell and Molecular Biology

2007

ABSTRACT
CHARACTERIZATION OF THE ARABIDOPSIS THALIANA CBF1
TRANSCRIPTION FACTOR:
FUNCTIONAL ROLE OF TWO EVOLUTIONARILY CONSERVED SIGNATURE
SEQUENCES
By

Donatella Canella

Over the course of evolution plants have adapted to the environment by
developing ways to cope with different biotic and abiotic stresses. Among the abiotic
stresses, low temperatures represent a major limiting factor for the growth and
development of plants. The CRT /DRE-Binding Factors (CBFs) are transcriptional
activators that are rapidly activated in response to low temperature and in turn induce the
expression of a battery of cold-regulated (COR) genes to increase plant freezing
tolerance. Arabidopsis plants overexpressing CBF1, CBFZ or CBF 3 are constitutively
freezing tolerant, indicating that these regulators are master regulators of cold adaptation.
CBF proteins belong to the APETALAZ/Ethylene-Response Binding Protein
(AP2/EREBP) family, which includes 145 members in Arabidopsis. Proteins in this
family share high similarity within their AP2/EREBP DNA-binding domain. A unique
feature of the CBF proteins is that they contain two conserved sequences flanking the
DNA-binding domain. These sequences, represented by the consensus motifs
PKK/RPAGRxKFxETRHP and DSAWR, are also found in CBF-like proteins fi'om
evolutionarily diverse plant species, suggesting that these “signature” sequences play a
role in CBF activity. Overexpression of wild type CBF1 in Arabidopsis results in

constitutive COR gene expression. Transgenic lines overexpressing CBF1 carrying

alanine substitutions in the signature sequences showed reduced or no COR gene
expression, indicating that these sequences play an important role in CBF1 activity.
Analysis of protein levels revealed that alanine mutations throughout the DSAWR motif
affect protein accumulation in planta, and could explain the lower COR gene
accumulation in those plants. On the contrary, mutations in the PKKPAGR motif did not
affect protein steady state levels; instead they impaired the ability of CBF1 to bind its
cognate CRT/DRE element from the COR gene promoters. The most pronounced effect
was observed when two conserved Arg and Phe residues were substituted with Lys and
Ala, respectively; these substitutions were sufficient to abrogate DNA binding, indicating
an essential role of those residues and potentially a base-specific recognition. Altogether
these results indicated that DNA binding activity in the AP2/EREBP family extends
beyond the canonical DNA-binding domain previously described to include the N-
flanking PKKPAGR region. Based on these observations and secondary structure
prediction studies, we developed a computational model describing CBF1 bound to the
DNA. According to this model, CBF1 binds the DNA major groove through a three-
stranded beta sheet, as described for other AP2/EREBP proteins. In addition, a helical
stretch within the PKKPAGR motif makes essential interactions with the DNA minor
groove in close proximity to a conserved thymine that is a specificity determinant in the
CRT/DRE element bound by the CBF proteins. Additional investigations will elucidate
whether residues within the PKKPAGR motif represent specificity switches for the
recognition of the CRT/DRE promoter element by CBF proteins and whether a similar

mechanism has been conserved in other protein of the APZ/EREBP family.

ACKNOWLEDGMENTS

I came to the United States to work in the Plant Research Laboratory as a visiting
scholar. The plan was to stay one year. Several years later, I prepare to leave with a
doctoral degree. Being here has been a great experience, at the professional and personal
level, and I am very gratefiil to the many people that have accompanied me in this

amazing journey.

First of all, I want to thank my mentor, Mike Thomashow, for his great guidance and
support. I am very grateful for the freedom, trust and encouragement that he has given me
all throughout my PhD. I feel privileged to have met him and very fortunate to have
worked in his lab. I have learned a lot from our meetings and discussions; these times
will always stay with me wherever I’ll go.

I would like to thank my committee members: Sheng Yang He, Lee Kroos, Ken
Keegstra, and Steve Triezenberg. Thank you for your time and your constructive
criticism. Your advice over the years has been critical in the completion of my thesis
work and has been an important contribution to my growth as a scientist.

I want to thank Leslie Kuhn, with whom I have collaborated for the last few years of my
PhD. It has been a great pleasure to work together; I have greatly benefited from your
scientific experience, and your encouraging attitude had made me a more confident
scientist.

Thank you to all the present and past members of the Thomashow lab: Sarah Gilmour,

Keenan Amundsen, Marcela Carvallo, Diane Constan, Daniel Cook, Colleen Doherty,

iv

Malia Dong, Carri Duncan, Sarah Fowler, Chin-Mei Lee, Michael Mikkelsen, Susan
Myers, Kanchan Pavangadkar, Ritu Shanna, Bonnie St.John, Heather Van Buskirk,
Jonathan Vogel, Vandana Yadav, Dan Zarka and Xin Zhang. I can’t believe I won’t be
coming to the lab next week. I will miss you.

A very big thank you to all my friends. It’s been wonderful to have you in my life.
Thanks to each and all of you for being with me during all this time and for making my
life so much richer and happier. It’s amazing to think of how many little everyday things
we have shared; all these times are such a big part of who I am today; there’s no doubt in
my heart I could not have come so far without you.

A special thank you to Alberto, who has been by my side with his love and support.
Thank you for never doubting that I could make it. Your strength, patience and constant
presence have been a wonderful gift that will stay with me forever.

A warm thank you to Kathryn and Edward Boucher, my American family. You have
been my home away from home. Thank you for the many suppers and for our chats after
supper. Your positive and strong spirit has touched me deeply; I will never forget it.

My family has always been there for me. Thank you mom, dad, Robi and Cira. You have
been my rock with your endless love, constant presence and encouragement. Thank you

for believing in me and being proud of me, no matter what. Thank you very much.

TABLE OF CONTENTS

LIST OF TABLES ........................................................................................................... viii

LIST OF FIGURES ........................................................................................................... ix

CHAPTER ONE Literature Review

Freeze-induced cell and role of cold acclimation .................................................. 1
Changes in gene expression underline the cold acclimation response ................... 3
The CBF cold-response pathway in Arabidopsis thaliana ..................................... 4
Upstream events in the regulation of the CBF pathway ........................................ 9
Conservation of the CBF pathway in other plant species .................................... 10
Mechanisms of transcriptional activation in the CBF family of proteins ............ 12
Conclusion .................................................................................. 18
Literature Cited ........................................................................... 19

CHAPTER TWO Functional roles of the PKKPAGR and DSAWR sequences in CBF1
transcriptional activity

Introduction .......................................................................................................... 27
Results .................................................................................................................. 32
FUNTIONAL ROLE OF THE PKKPAGR MOTIF IN CBF1 ACTIVITY ........ 37
Mutations in the PKKPAGR motif affect COR gene activation .......................... 37
Reduced COR gene induction in the pkkpagr lines is not caused by lower protein
levels ........................................................................................ 43
Role of the PKKPAGR motif in nuclear targeting of CBF1 ................................ 47
Identifying regions in CBF1 involved in nuclear transport ................................. 51
Role of the PKKPAGR motif in DNA binding .................................................... 57
Secondary structure predictions to investigate structural the preferences of the
PKKPAGR motif .................................................................................................. 61
Specific amino acids within the PKKPAGR motif have important roles in
binding to the CRT/DRE promoter element ........................................................ 67
Side chain conservation in AP2/EREBP-like proteins that use their flanking
sequences to stabilize DNA binding ..................................................................... 73
A working model for DNA binding by CBF1 .......................... . ................ 80
FUNTIONAL ROLE OF THE DSAWR MOTIF IN CBF1 ACTIVITY ............ 87
Mutations 1n the DSAWR sequence affect COR gene activation ........................ 86
CBF1 protein detection In dsawr lines .................................................. 89
Role of the DSAWR motif 1n DNA binding activity by CBF1 ...................... 91
Point mutations within the DSAWR region from CBF 2 reveal the importance

an Asp residue to CRT/DRE recognition by CBF proteins ................................. 93
DISCUSSION and FUTURE DIRECTIONS ...................................................... 98

vi

MATERIAL AND METHODS ......................................................................... 106

Mutagenesis of the PKKPAGR and DSAWR motifs ........................................ 106
Preparation of constructions for plant transformation .............................. 108
Growth conditions and Northern blot analysis of transgenic Arabidopsis

plants .................................................................................................................. 110
Analysis of variance (ANOVA) for COR/CBF1 transcript ratios ..................... lll
Subcloning of CBF1 mutant ORFs into bacterial expression vectors for
expression and purification of 6xHiszCBF1 and MBP:CBF1 proteins .............. 112
Protein isolation and immunoblot analyses ........................................................ 114
Electro-Mobility Shift Assays (EMSA) ............................................................. 115
Fluorescence imaging of Arabidopsis root tips overexpressing CBF1 :GFPzGUS
transgenes ................................................................................. 116
Literature Cited ........................................................................... l 18

vii

LIST OF TABLES

Table 2.1. Secondary structure predictions of the PKKPAGR motif ......................... 63
Table 2.2. Sequery analysis of the crystallographic structures observed for tetrapeptide
sequences in the PKKPAGR region of CBF1, and mutations to AAA in this region that
were tested experimentally ........................................................................ 65
Table 2.3. Quantification of CRT /DRE-containing probes bound by CBF1 wild type and
mutated proteins .................................................................................... 75

Images in this dissertation are presented in color.

viii

LIST OF FIGURES

Figure 2.1. PKKPAGR and DSAWR signature sequences in the CBF family of plant

transcription factors ................................................................................. 28
Figure 2.2. Site-directed mutagenesis was used to generate CBF1 ORFs containing
specific mutations or deletions of the signature sequences ................................... 33
Figure 2.3. Arabidopsis pqupagr and dsawr lines ............................................. 35
Figure 2.4 A. Analysis of COR gene induction in APKK, M1 and M2 pkkpagr lines by
Northern blot analysis ....................................................................................................... 38
Figure 2.4 B. Analysis of COR gene induction in M3, M4 and M5 pkkpagr lines by
Northern blot analysis ....................................................................................................... 39
Figure 2.5A. Analysis of variance of COR78/CBF1 transcript ratios in transgenic plants
overexpressing wild type or mutated CBF1 transgenes ....................................... 41
Figure 2.5B. Analysis of variance of COR6.6/CBF1 transcript ratios in transgenic plants
overexpressing wild type or mutated CBF1 transgenes ....................................... 42
Figure 2.6. Comparison of CBF1 transcript and protein levels in Arabidopsis plants
overexpressing 6xMyc:CBFl transgene ......................................................... 45
Figure 2.7. Northern blot analysis of Arabidopsis seedlings overexpressing different
CBF1 :GFPzGUS constructs ........................................................................ 48
Figure 2.8. Localization studies of CBF1 :GFPzGUS chimeras .............................. 50
Figure 2.9. Localization studies of 5’ deletions of CBF1 :GFPzGUS chimeras ............ 54

Figure 2.10. Western blot analysis of total protein extracts from Arabidopsis seedlings
overexpressing different CBF1 :GFPzGUS constructs or GFPzGUS alone .................. 56

Figure 2.11. Triple alanine mutations abrogate the binding activity of a full length
6xHis-CBF1 protein in vitro. ..................................................................... 58

Figure 2.12. Triple alanine mutations abrogate the binding activity of MBP-CBFZH 12
proteins in vitro ...................................................................................... 60

Figure 2.13 A. Effect of point mutations in the predicted RKKFRET helical region of
CBF1 on binding activity of the protein to the COR 78 gene promoter ..................... 69

ix

Figure 2.13 B. Effect of point mutations in the predicted RKKFRET helical region of

CBF1 on binding activity of the protein to the COR15a gene promoter ..................... 70
Figure 2.13 C. Effect of point mutations in the predicted RKKFRET helical region of
CBF1 on binding activity of the protein to the COR6. 6 gene promoter ..................... 71
Figure 2.14. Three-dimensional structures of proteins that use their three-stranded B-
sheets to recognize the major groove of their target DNAs .................................... 77
Figure 2.15. Comparison of B-sheet—flanking sequences experimentally shown to be
important in DNA binding .......................................................................... 78
Figure 2.16A. NMR structure of the AP2/EREBP DNA-binding domain of ERF] bound
to its cognate DNA .................................................................................. 81
Figure 2.16B. Working model describing DNA binding by CBF1 .......................... 82

Figure 2.17. Nucleotide sequences of the cis-acting elements in CBF and ERF
proteins. ............................................................................................................................ 85

Figure 2.18. Accumulation of CBF and COR transcripts in Arabidopsis plants
overexpressing wild type or dsawr CBF1 ........................................................ 88

Figure 2.19. Northern and western blot analysis of dsawr plants and plants
overexpressing wild type CBF1 ................................................................... 90

Figure 2.20. DNA binding activity of an MBP-CBFl protein carrying alanine
substitutions in the DSAWR region ............................................................... 92

Figure 2.21. Northern blot analysis of transgenic plants overexpressing wild type and
mutated CBF2 transgenes ........................................................................... 94

Figure 2.22. DNA binding activity of MBP:CBF2 proteins carrying point mutations in
the DSAWR motif ................................................................................... 96

CHAPTER ONE

LITERATURE REVIEW

Plant growth and development can be greatly affected by low temperatures. Plants
differ in their ability to cope with low temperatures. Plants native to tropical regions, such
as tomato and rice, are typically very sensitive to low temperatures and will suffer
chilling injuries when temperatures drop to the range of 0-12°C. In contrast, plants
originating from temperate regions, such as the model plant Arabidopsis thaliana, rye and
canola, are generally more resistant, and are not only chilling tolerant but also freezing
tolerant (Sakai and Larcher, 1987). The ability of plants to tolerate freezing can be
significantly enhanced by pre-exposure to low non-freezing temperatures, an adaptive
process called cold acclimation. For instance, upon cold acclimation wheat can increase
freezing tolerance from -5°C to -20°C (Thomashow, 1998).

The events occurring inside a plant upon cold acclimation reflect a complex
network of changes that the plant mounts against freeze-induced damage (Levitt, 1980;
Thomashow, 1999). These include changes at the physiological, biochemical and
transcriptional level. In this chapter, I will summarize the advances in the field of cold
acclimation, with specific focus on the CBF family of transcriptional regulators and their

role in the CBF cold response pathway in the model plant Arabidopsis.

Freeze-induced cell damage and role of cold acclimation.
The negative effects of freezing temperatures on cell survival have been long

known (Levitt, 1980). The plasma membrane is the primary target of freeze-induced cell

damage. This injury is mainly caused by freeze-induced dehydration, which occurs in the
cell upon exposure to sub-zero temperatures. The first event following freezing inside the
cell is ice formation in the extracellular space; this is due to the presence of ice nucleating
agents and lower solute concentration in the extracellular space relative to the
intracellular space. This process can cause membrane damage due to adhesion of
membranes. However, the most severe damage results from freeze-induced dehydration.
When ice crystals form at sub-zero temperatures, the chemical potential in the outer space
drops, and causes an outward movement of intracellular water to the outer space. The
water loss can be as severe as 90% at approximately -10°C (Thomashow, 1999). The type
of injury at the membrane site varies depending on the freezing temperatures: freeze-thaw
cycles between -2°C to -4°C can cause expansion-induced cell lysis; a temperature range
of —4°C to -10°C typically causes phase transition of bilayer lipids from larnellar to
hexagonal II; in the most severe cases (below -10°C), fracture jump lesions will arise
(Steponkus and Webb, 1992).

Membrane damage induced by freezing temperatures is greatly reduced upon cold
acclimation. The protective role of cold acclimation has been under scrutiny for many
years, and includes a plethora of changes at the biochemical, physiological and
transcriptional levels. Studies by several investigators have established that one of the
roles of cold acclimation is to prevent cellular damage by stabilizing membranes.
Changes in membrane lipid composition are among the first and best documented events
that underline cold acclimation (Thomashow, 1999). Additional metabolic changes also

accompany cold acclimation, including the accumulation of small cryoprotective

molecules such as soluble sugars and proline (Rudolph and Crowe, 1985; Strauss and

Hauser, 1986; Carpenter and Crowe, 1988).

Changes in gene expression underline the cold acclimation response.

In more recent years, changes in gene expression have been correlated to
adaptation to low temperatures. The first direct evidence that cold acclimation was
accompanied by changes in gene expression was presented by Guy and colleagues
(1985), who reported that differential pools of translatable mRNAs accumulate in non-
acclimated and acclimated spinach leaves (Guy et al. , 1985). This discovery was
followed by the identification of numerous cold-responsive genes in other plant species,
including Brassica napus (J ohnson-F lanagan and Singh, 1987), alfalfa (Mohapatra et al. ,
1987) and Arabidopsis (Gilmour et al., 1988; Kurkela et al. , 1988). Those genes were
given names such as low temperature-induced genes (LT 1), cold-regulated genes (COR),
KIN (cold-induced), or RD (responsive to dehydration) (reviewed in Thomashow, 1999).
Analysis of their expression patterns revealed a positive correlation between their
transcript accumulation in response to low temperatures and the level of freezing
tolerance achieved by plants (Guy and Haskell, 1987; Mohapatra et al. , 1987). Despite
their highly divergent sequences, the proteins encoded by these genes share a set of
properties that suggest that they might share a common mechanism of action in
protecting the cell against freezing. Indeed, some of those properties have been
previously described for cryoprotectant molecules, and include high hydrophilicity,
presence of amphipathic helices, and solubility in aqueous buffers at high temperatures

(V olger and Heber, 1975).

The observation that changes in gene expression accompany cold acclimation
suggested that such changes might represent the initial signal regulating the events
required to achieve freezing tolerance. However, overexpression of COR15a in
Arabidopsis had little effect on overall plant freezing tolerance, despite its significant role
in protecting the chloroplast envelope (Steponkus et al. , 1998). This observation
indicated that individual COR genes were not sufficient to trigger the whole set of
changes induced by cold acclimation. It was soon hypothesized that an upstream
regulator of gene expression was responsible for the activation of a whole set of COR
genes. Consistent with this idea, characterization of a battery of COR genes from
Arabidopsis suggested that they might be coordinatively regulated in response to low
temperatures. Consistent with this idea, COR6. 6, COR15a, COR4 7, and COR 78
expression followed similar kinetics; they were upregulated as early as four hours after a
treatment at 4°C, reached a peak at about 24 hours and remained upregulated as long as
the plants were kept under inducing conditions (Hajela et al. , 1990) (Horvath et al. ,
1993). Furthermore, analysis of COR gene promoters unveiled the presence of a common
cis-regulatory element, named C-repeat/Drought Responsive Element CRT/DRE (Baker
et al., 1994; Yamaguchi-Shinozaki and Shinozaki, 1994). Similar induction patterns and
the presence of a conserved element in their promoters suggested that the COR genes

might be regulated by a common factor.

The CBF cold-response pathway in Arabidopsis thaliana.

Discovery of the CBF pathway.

A major breakthrough in the field of cold acclimation was the discovery of the
transcriptional regulator CRT/DRE Binding F actorl (CBF1) (Stockinger et al., 1997) and
its key role in regulating cold acclimation in Arabidopsis. CBF1 was identified in a yeast-
one-hybrid screen in which the CR T /DRE promoter element was used as bait. CBF1
could specifically bind this cis-element in vitro and induce the expression of a
COR15a::lacZ reporter gene in yeast (Stockinger et al., 1997). Arabidopsis plants
overexpressing CBF1 constitutively expressed the whole battery of COR genes and
displayed constitutive fi'eezing tolerance similar to freezing tolerance of cold-acclimated
non-transgenic plants, indicating that this transcription factor is a master regulator of the
cold acclimation response in Arabidopsis (Jaglo-Ottosen et al. , 1998). The discovery of
CBF1 and its role in cold acclimation was significant because it provided the first
evidence that genetic manipulation was indeed possible to generate more freezing
tolerant plants, whereas years of breeding attempts had resulted in only modest
improvement of freezing tolerance. Furthermore, it revealed that the whole set of changes
necessary to induce freezing tolerance could be induced by overexpressing a single gene
(Thomashow, 1999).

Shortly after, it was discovered that Arabidopsis has three CBF genes, CBF1,
CBF2, and CBF3 — also known as DREBI B, DREBI C, and DREBIA, respectively -
arranged in tandem array on chromosome IV (Gilmour et al. , 1998; Liu et al. , 1998;
Gilmour et al. , 2000). They follow a similar induction pattern in response to low
temperatures, whereby they are induced within 10-15 minutes of cold treatment, peak at
2-3 hours, and return to a basal expression level by 24 hours of cold treatment.

Overexpression of CBF1, CBF2, or CBF 3 in Arabidopsis rendered plants constitutively

freezing tolerant (Jaglo-Ottosen et al., 1998; Liu et al. , 1998; Gilmour et al., 2000;
Gilmour et al. , 2004) and induced constitutive expression of the same pool of genes,
indicating that the three transcriptional activators have matching transcriptional activities.
Three additional CBF homologs were identified more recently. CBF 4/DREBI D (Haake
et al. , 2002; Sakuma etal., 2002), was described as a regulator of drought response in
Arabidopsis, in that its expression was up-regulated by drought stress, but not by low
temperature. Its overexpression in Arabidopsis resulted in constitutive accumulation of
CRT /DRE-containing genes, and conferred to plants both drought and freezing tolerance,
similarly to what observed for overexpression of the other CBFs (Haake et al. , 2002).
This result is not surprising, given the physiological overlap between dehydration and
freezing stress. DDF I/DREBI F and DDF 2/DREBI E (Sakuma et al. , 2002; Magome et
al. , 2004) also show high sequence homology to the CBF family, but little is known on
their role in response to abiotic stresses. DDF I/DREBI F expression is induced upon
salinity stress (Sakuma et al. , 2002; Magome et al. , 2004), similarly to what happens to
CBF1-3, suggesting that these genes may display overlapping activity. The role of

DDF I/DREBI F and DDF 2/DREBI E in low temperature response, however, has not been

elucidated.

Predominant role of CBF in configuring low temperature responses in Arabidopsis.
Technological advances and the availability of whole genome sequences have

made it possible to conduct high-throughput experiments to identify the molecular

changes at the transcriptional and post-transcriptional level occurring in response to low

temperature.

Microarray technology has proven a powerful tool for surveying global gene
expression in Arabidopsis in response to cold treatment. Transcriptome profiling studies
using cDNA microarrays and Affymetrix GeneChips have been very usefirl to identify
novel genes, determine their kinetics of induction, and help delineate the complex
network of cascades that are configured. These analyses have revealed that hundreds of
genes are regulated during cold acclimation in Arabidopsis (Seki et al. , 2001; Fowler and
Thomashow, 2002; Kreps et al., 2002; Seki et al., 2002; Gilmour et al., 2004; Maruyama
et al., 2004; Vogel et al. , 2005). Among the up-regulated genes are several transcription
factors, some of which are induced in parallel with CBF. An important conclusion from
these studies has been that additional cold response pathways exist besides the CBF
pathway (Seki et al., 2001; Fowler and Thomashow, 2002; Kreps et al. , 2002; Seki et al. ,
2002). This is consistent with previous reports of Arabidopsis mutants that are
constitutively more freezing tolerant than wild type plants despite the fact that COR
genes are not induced. One example of a CBF-independent response is represented by the
eskr'moI Arabidopsis mutant (Xin and Browse, 1998). eskimoI plants display constitutive
freezing tolerance and yet do not show constitutive expression of the CBF regulon (Xin
and Browse, 1998; Xin et al., 2007). Indeed, transcriptome profiling analysis showed an
overlap of about 12% in gene transcripts between the eskimoI mutants and CBF2-
overexpressing plants (Xin et al., 2007). Interestingly, eskimoI plants are not drought
tolerant, which makes them a valuable tool for understanding the players and
mechanisms that are specific for the cold response in Arabidopsis. Similarly to the
eskimoI mutation, ada2b mutant plants in Arabidopsis are constitutively more freezing

tolerant than wild type plants, and yet they do not show constitutive COR gene

expression (V lachonasios et al. , 2003); therefore ADA2b is implicated in a CBF-
independent pathway.

Despite the presence of additional cold response pathways, the CBF family plays
a predominant role in configuring the changes observed when plants respond to cold.
Comparison between genes that are cold- and CBF-responsive has helped define the CBF
regulon as the pool of approximately 100 genes that are regulated by both low
temperature and CBF (Fowler and Thomashow, 2002; Vogel et al. , 2005). Eighty five
genes in the CBF regulon are up-regulated and account for almost 30% of the transcripts
that accumulate in response to low temperatures. In addition, members of the CBF
regulon are among the most highly induced upon cold treatment (V ogel et al. , 2005). For
instance, 84% of the cold-responsive genes that are up-regulated more than 15 fold and
50% of the genes that are up-regulated between 5-10 fold belong to the CBF regulon.

Metabolite profiling in Arabidopsis has enabled the analysis of changes occurring
at low temperature to changes in small organic compounds associated with low
temperature response. The role of CBF in mediating metabolic changes that occur upon
cold exposure was monitored by metabolomic profiling of wild type Arabidopsis plants
and plants overexpressing CBF 3 . Major rearrangements were detected during cold
acclimation, with 75% of the metabolites analyzed showing changes. Of those changes,
the majority (79%) could be attributed to CBF 3 overexpression (Cook et al. , 2004).
Therefore, the CBF pathway plays a major role in configuring the changes that are
needed for plants to adapt to low temperatures, both at the transcriptional and post-

transcriptional level.

Upstream events in the regulation of the CBF pathway.

Much of the research effort in the field of cold acclimation has focused on
elucidating the upstream events regulating the induction of CBF genes. Reporter gene
analyses and mutant screens have proven very valuable to identify novel genes that play
important roles during cold acclimation.

Promoter analysis of CBF2 unveiled a minimal 125-bp promoter region that is
cold responsive (Zarka et al. , 2003). Within this promoter fragment, two regions named
ICErl and ICEr2 (induction of CBF expression region 1 and 2) cooperated in directing
cold regulation of CBF expression (Zarka er al., 2003). ICErl contains a cis-acting
element which is recognized by ICEl (Inducer of CBF Expression 1). ICE] was isolated
through a mutant screen of plants that were defective in CBF 3 cold-induction; however,
CBF1 or CBF 2 expression was not affected, suggesting that one of the other 139 bHLH
proteins present in Arabidopsis might be involved in the specific recognition of this
element at low temperatures (Chinnusamy et al., 2003). Interestingly, overexpression of
ICE] is not sufficient to confer constitutive freezing tolerance, but a cold stimulus is
required. Presumably, post-translational mechanisms or the presence of a co-activator
must occur to activate ICEl (Chinnusamy et al. , 2003). More recently, it was discovered
that a transcriptional regulator member of the MYB family, MYB15, can bind to the
promoter of CBF1, CBF 2, and CBF 3, and physically interacts with ICEl in vitro
(Agarwal et al., 2006). MYB15 overexpression negatively regulates CBF expression in
the cold and is most evident in the early stages of induction (approximately 3 hours).
However, MYB15 overexpression does not significantly influence COR gene expression

(Agarwal et al., 2006). The significance of these results remains unclear at the present.

Recently, Vogel et al. (2005) have characterized ZAT12, a zinc-fmger
transcription factor that appears to be coordinately regulated with CBF2 in response to
low temperature. Interestingly, ZAT12 overexpression in Arabidopsis has a negative
effect on cold-regulated induction of the CBF genes, whereas in the zat12 knock-out lines
CBF expression is induced. Presmnably this repressive activity is mediated by an EAR-
like motif present in ZAT12 (Ohta et al. , 2001). At the same time, ZAT12 induces the
expression of a Z4T12 regulon, which partly overlaps with the CBF regulon, suggesting
that CBF and ZAT12 might cooperate to induce expression of cold responsive genes
(V ogel et al., 2005).

It is evident from the recent advances that the regulation CBF expression during
cold stress is complex and involves many factors. Identification of additional components
and a better understanding of their mechanisms of action will help gain a more detailed

picture of how transcriptional networks are regulated in response to low temperatures.

Conservation of the CBF pathway in other plant species.

A major interest in the field of cold acclimation has been to determine whether the
CBF pathway has been conserved in other plant species. This is of particular interest for
agronorrrical application to crop plants. CBF-like genes have been identified in a wide
variety of plant species, including both chilling sensitive and freezing tolerant plants. A
central question is to determine whether these represent functional homologs of the
Arabidopsis CBFs, and whether any differences in the CBF pathway of these plants can

explain different tolerance to low temperatures.

10

Among freezing tolerant plants, CBF-like genes have been identified in Brassica
napus (Jaglo et al., 2001), poplar (Benedict et al. , 2006), barley (Skinner et al. , 2005),
maize (Qin et al., 2004) and others. Some of these CBFs have been shown to be
functional homologs of the Arabidopsis CBFs based on their sequence similarity,
transcriptional activity, and the ability to enhance freezing and drought tolerance when
overexpressed. Furthermore, a number of CR T/DRE-containing genes present in those
plants and their transcripts can be constitutively accumulated when CBF is overexpressed
(Choi et al., 1999; Dal Bosco et al., 2003).

Chilling sensitive plants also contain CBF-like genes. Some of these genes
represent functional homologs of the Arabidopsis CBFs, since they are cold-responsive
and can induce constitutive expression of a CBF regulon when overexpressed in
Arabidopsis, and results in increased tolerance to freezing and dorught. Some examples
include CBFs from tomato (Hsieh et al., 2002; Zhang et al. , 2004) and rice (Dubouzet et
al., 2003). The tomato CBF locus includes three CBF genes (Zhang et al., 2004).
LeCBFI is cold-responsive and can induce constitutive expression of a CBF regulon and
irrrpart freezing tolerance when overexpressed in Arabidopsis. Tomato plants
overexpressing LeCBFI or AtCBFI can also activate a CBF regulon. However,
microarray analysis of a quarter of the tomato genome indicates that, despite the fact that
similar classes of proteins are induced in the two plants, transgenic tomato plants can
only slightly increase their chilling tolerance (Zhang et al. , 2004). Current efforts are in
place to determine whether the inability of chilling sensitive plants to show a functional
CBF response can be attributed to smaller and consequently less diverse CBF regulons or

additional defects that limit CBF function.

11

Mechanisms of transcriptional activation in the CBF family of proteins.
The CBF family of proteins.

CBF proteins are members of the AP2/EREBP family of transcription factors
(Riechmann and Meyerowitz, 1998). Members of this family are widespread among
plants, but a few examples have been recently found in other organisms such as bacteria
and bacterial viruses (Magnarri et al., 2004; Wuitschick et al. , 2004; Balaji et al. , 2005).
AP2/EREBP proteins play a variety of roles in plant growth and development as well as
in biotic and abiotic stress responses (Riechmann and Meyerowitz, 1998). A common
feature of these proteins is the presence of a conserved AP2/EREBP DNA-binding
domain. The Arabidopsis AP2/EREBP family includes 145 predicted proteins sharing
high homology within this domain. Based on the sequence similarity within the DNA-
binding domain, Sakuma et al. (2002) have divided the Arabidopsis AP2/EREBP family
into 5 different groups: the DREBs (56 members); the ERFs (65 members); the AP2$ (14
members); the RAVs (6 members); and a fifth group (4 members). The DREB group
includes the CBF family. CBF proteins are characterized by an N-terminal domain
including 32 amino acids of unknown function, the AP2/EREBP DNA-binding domain,
and a C-terminal domain resembling acidic activation domains from other transcriptional
activators such as the herpesvirus VP16, the mammalian p53 protein and RelA
(Triezenberg, 1995). A distinctive feature of the CBF family in Arabidopsis is the
presence of two conserved signature sequences flanking the DNA-binding domain,
defined by the consensus PKKP/RAGRxKFxETRHP and DSAWR (J aglo et al., 2001).
Strikingly, these sequences have been highly conserved throughout evolution, and they

can be found in very diverse plant species (J aglo et al., 2001). This Observation suggests

12

that they might play an important role in mediating CBF transcriptional activity. The

functional characterization of these motifs will be presented in this dissertation.

DNA-binding properties of CBF proteins and other APZ/EREBP family members.

The DNA-binding activity of the AP2/EREBP proteins was frrst described by
Ohme-Takagi and Shinshi (1995). They reported that a 59-amino acid region that was
highly conserved among four ethylene-binding proteins was required for specific binding
to their target DNA, called the GCC box (Hart et al., 1993; Ohme-Takagi and Shinshi,
1995; Sato et al., 1996). Shortly after, additional transcription factors with the same
conserved domain were identified, and the conserved region named AP2/EREBP DNA-
binding domain (Riechmann and Meyerowitz, 1998). One of the most significant
contributions towards characterizing the binding activity of the AP2/EREBP proteins has
been the NMR analysis of this domain from AtERFl bound to its target cis-regulatory
element. This domain is defined by a three stranded B-sheet followed by a C-terrninal
helix; the B-sheet is packed against the major groove of the DNA and makes specific
contacts through seven amino acids, mainly arginines and tryptophans (Allen et al. ,
1998). The degree of similarity among the Arabidopsis AP2/EREBP proteins ranges from
approximately 60 to 90% and six of the seven amino acids involved in direct DNA
contact in ERF 1 are conserved in the AP2/EREBP proteins (Hao et al. , 2002). The high
degree of conservation implies that these domains fold similarly and might make similar
DNA contacts.

Very little information is available on the DNA binding preferences within the

AP2/EREBP family. One of the reasons is that the target cis-acting elements have been

13

identified only for a few AP2/EREBP proteins, including ABI4, ORCA, CBFs, DREBZS,
and ERFs (Stockinger et al. , 1997; Allen et al., 1998; Liu et a1. , 1998; van der Fits and
Memelink, 2001; Acevedo-Hernandez et al. , 2005). Most of the current knowledge on the
binding properties of the AP2/EREBP family has been derived from the analysis of the
ERF and CBF proteins. The cis-acting elements targeted by the CBF and ERF proteins
have long been known and are represented by the CR T/DRE and GCC box promoter
elements, respectively (Baker et al., 1994; Yamaguchi-Shinozaki and Shinozaki, 1994;
Ohme-Takagi and Shinshi, 1995). The specificity determinants in these two cis-acting
elements are defined by the two core sequences A/GCCGACNT and AGCCGCC for the
CBF and ERF family, respectively (Hao et al., 1998; Hao et al., 2002; Maruyama et al. ,
2004). Much remains to be understood in terms of specificity switches within proteins of
the two subfamilies. Most of the work has focused on elucidating the role of amino acids
within the AP2/EREBP domain that differ between ERFs and CBFs. One of the most
important findings is that an alanine residue that is highly conserved in the ERF
subfamily can impart specific sequence recognition to proteins in this subgroup. In fact,
when alanine is substituted with the corresponding valine residue from the CBF proteins,
the mutated ERF alters its binding specificity and can recognize the CR T/DRE promoter
element (Hao et al. , 2002). However, there is no evidence indicating the corresponding
mutation in the AP2/EREBP domain of CBF can modify its binding preference. Sakuma
et al. (2002) have proposed that two conserved amino acids within the AP2/EREBP
domain, which differ between ERFs and CBFs, function as specificity determinants. The
authors showed that swapping of the residues between the two groups of proteins could

abrogate binding to their respective cis-acting elements; however, there is no direct

l4

evidence that residue swapping can drive CBF proteins to the ERF promoter. Based on
the current studies, it is unclear whether additional residues are essential for DNA
binding within the AP2/EREBP family. Hao et al. (1998) observed that short stretches of
amino acids flanking the DNA-binding domain of several ERF proteins are important for
binding to the GCC box. Since the flanking regions of ERF proteins share little or no
similarity, it was suggested that the role of these residues is to stabilize the protein-DNA

complex (Hao et al., 1998).

Characterization of CBF1 trans-activating properties.

Despite the fact that CBF1 was identified a decade ago, not much is known on the
mechanism whereby CBF proteins activate COR gene expression. The C-terrninal
domain of CBF (CBF1116-213) functions as the activation domain in both yeast and
Arabidopsis (Wang et al., 2005). Within this domain are several hydrophobic clusters;
each of them provides some functional redundancy, presumably to maintain efficient
activation of the CBF regulon under cold stress conditions (Wang et al. , 2005).

More recently, investigators have explored the role of chromatin-modifying
factors in facilitating CBF trans-activating properties. The CBF1 activation domain is
acidic in nature, similar to that of other transcription factors such as VP16 and Gcn4
(Hope and Struhl, 1986; Seipel et al., 1994). These proteins stimulate transcription in part
by recruitment of chromatin-modifying complexes such as Ada or SAGA (Kuo et al. ,
1998; Ikeda et al. , 1999). The transcriptional adaptors ADA2 and ADA3 and the histone
acetyltransferase GCNS are part of these large multi-protein complexes (Grant et al. ,

1997). They were shown to be essential for CBF1 transcriptional activity in yeast

15

(Stockinger et al. , 2001). More recently, it was discovered that CBF1 can also interact in
vitro with AtGCN5 and AtADA2b, the Arabidopsis homologues of GCNS and ADA2
from yeast (Mao et al. , 2006). Contrary to what expected, these interaction occur through
the DNA-binding domain of CBF1, and not through the activation domain. Furthermore,
they are not specific to CBF1 but appear to be conserved within the AP2/EREBP family,
as a similar interaction occurs with another member of the family, TINY (Wilson et al. ,
1996; Mac et al., 2006). To better understand the role of AtGCN5 and AtADA2b in vivo,
T-DNA insertion mutations for the two genes were analyzed. Disruption of AtGCN5 and
AtADA2b affects cold-regulated induction of the CBF regulon in Arabidopsis, by
reducing the level of COR gene'expression and delaying the kinetics of induction

(V lachonasios et al. , 2003). Taken together, the emerging evidence is that these two
adaptor proteins might be recruited by CBF1 to mediate COR gene induction.

In the effort to identify components of the cold acclimation response in
Arabidopsis, a set of mutants sensitive to freezing was isolated (McKown et al. , 1996). In
one of those mutants, sfi6, cold-induced activation of CRT/DRE-containing genes was
affected in one of the steps that follow CBF1 induction (Knight et al., 1999). CBF1
transcript still accumulated in response to low temperatures, yet COR gene expression
was compromised (Knight et a1. , 1999; Boyce et al., 2003). The nature of the sfi'6
mutation is still unknown, and thus the contribution of this factor to the CBF pathway has
not been elucidated. Given that the CBF transcript is unaffected in sfi6 plants, the role of
the SFR6 protein is post-transcriptional. Among other explanations, it is reasonable that

SFR6 might act as a co-factor of CBF.

16

Additional post-transcriptional events regulating CBF activity.

One way to modulate the activity of transcription factors is by actively
controlling their cellular localization (Tran and Wente, 2006). Whether nuclear shuttling
of CBF1 is dependent on a nuclear localization signal (N LS), and whether this is a low
temperature-regulated process is not clear. It is expected that some CBF1 protein can
accumulate in the nucleus without low temperature treatment, since Arabidopsis lines
overexpressing CBF1, CBF 2 or CBF 3 can accumulate high levels of COR gene
transcripts compared to the non transgenic plants in the absence of a cold stimulus (J aglo-
Ottosen et al. , 1998; Gilmour et al. , 2000). On the other hand, when CBF-overexpressing
plants are cold treated, COR gene expression increases (Gilmour et al., 2000). Among
other reasons, this may be due to more efficient nuclear transport. Recent studies focused
on the subcellular localization of several members of the AP2/EREBP family have
revealed that nuclear targeting within this family can be either constitutive or stimulus-
dependent. For instance, localization studies on AINTEGUMENTA (Klucher et al. ,
1996) have indicated that nuclear targeting of this AP2/EREBP protein is constitutive and
depends on the presence of specific basic residues (Krizek and Sulli, 2006). On the
contrary, the cytokinin response factors CRFs accumulate in the nucleus in response to
cytokinin treatment (Rashotte et al. , 2006). Moreover, a high-throughput localization
study of Arabidopsis proteins has identified an AP2/EREBP protein (id. At2g22880) that
can be found both in the nucleus and in the cytoplasm and four that are constitutively
nuclear localized (id. At1g13260, At2g39250, At2g33710) under standard grth
conditions (Koroleva et al. , 2005). Altogether these studies indicate that nuclear

localization of proteins in the AP2/EREBP family is regulated differently, and a case by

17

case study is required to understand the mechanisms dictating nuclear import of each
protein.

Based on its similarity to known nuclear localization motifs found in other plant
and non-plant transcription factors, the PKKPAGR motif has been proposed to be the
nuclear targeting signal of CBF1 (Stockinger et al. , 1997). Whether this motif is required

and/or sufficient for nuclear import of CBF proteins remains to be determined.

Conclusion

Since the discovery of CBF 1 in Arabidopsis, several lines of evidence have
suggested that additional cold response pathways are present. While it is important to
define these alternative pathways, furthering our knowledge of the CBF action
mechanisms could better our understanding of how these novel pathways are activated
and promote freezing tolerance in plants. The relevance of the CBF pathway has been
emphasized over the recent years by large-scale analyses, such as transcriptome and
metabolome profiling, and by numerous reports showing that this pathway is conserved
across diverse plants species. Understanding how CBF proteins are activated and regulate
the changes that lead to freezing tolerance will help define some of the basic mechanisms
that have been conserved not only the CBF pathway but also in novel cold-responsive
pathways, and it will ultimately help establish how plants sense and respond to low
temperatures.

To contribute to this effort, in dissertation I will discuss the functional
characterization of the Arabidopsis CBF1 protein, by describing the functional properties

of two conserved domains in the CBF family of transcriptional activators.

18

LITERATURE CITED

Acevedo-Hernandez, G.J., Leon, P., and Herrera-Estrella, LR. (2005). Sugar and
ABA responsiveness of a minimal RBCS light-responsive unit is mediated by
direct binding of ABI4. The Plant Journal 43, 506-519.

Agarwal, M., Hao, Y., Kapoor, A., Dong, C.-H., Fujii, H., Zheng, X., and Zhu, J.-K.
(2006). A R2R3 type MYB transcription factor is involved in the cold regulation
of CBF genes and in acquired freezing tolerance. J. Biol. Chem. 281, 37636-
37645.

Allen, M., Yamasaki, K., Ohme-Takagi, M., Tateno, M., and Suzuki, M. (1998). A
novel mode of DNA recognition by a beta-sheet revealed by the solution structure
of the GCC-box binding domain in complex with DNA. EMBO J 17, 5484-5496.

Baker, 8., Wilhelm, K., and Thomashow, M. (1994). The 5'-region of Arabidopsis
thaliana corl 5a has cis-acting elements that confer cold-, drought- and ABA-
regulated gene expression. Plant Mol Biol 24, 701-713.

Balaji, S., Babu, M., Iyer, L., and Aravind, L. (2005). Discovery of the principal
specific transcription factors of Apicomplexa and their implication for the
evolution of the AP2-integrase DNA binding domains. Nucl. Acids Res. 33,
3994-4006.

Benedict, C., Skinner, J.S., Meng, R., Chang, Y., Bhalerao, R., Huner, N.P.A., Finn,
C.E., Chen, T.H.H., and Hurry, V. (2006). The CBF1-dependent low
temperature signalling pathway, regulon and increase in freeze tolerance are
conserved in Populus spp. Plant, Cell & Environment 29, 1259-1272.

Boyce, J., Knight, H., Deyholos, M., Openshaw, M., Galbraith, D., Warren, G., and
Knight, M. (2003). The sfr6 mutant ofArabidopsis is defective in transcriptional
activation via CBF/DREBI and DREB2 and shows sensitivity to osmotic stress.
The Plant Journal 34, 395-406.

Carpenter, J.F., and Crowe, J.H. (1988). The mechanism of cryoprotection of proteins
by solutes. Cryobiology 25, 244.

Chinnusamy, V., Ohta, M., Kanrar, S., Lee, B.-h., Hong, X., Agarwal, M., and Zhu,
J.-K. (2003). ICE]: a regulator of cold-induced transcriptome and freezing
tolerance in Arabidopsis. Genes Dev. 17, 1043-1054.

Choi, D., Zhu, B., and Close, T. (1999). The barley (Hordeum vulgare L.) dehydrin
multigene family: sequences, allele types, chromosome assignments, and
expression characteristics of 11 Dhn genes of cv Dicktoo. Theor. Appl. Genet. 98,
1234—1247.

19

Cook, D., Fowler, 8., Fiehn, 0., and Thomashow, M.F. (2004). A prominent role for
the CBF cold response pathway in configuring the low-temperature metabolome
of Arabidopsis. PNAS 101, 15243-15248.

Dal Bosco, C., Busconi, M., Govoni, C., Baldi, P., Stanca, A.M., Crosatti, C., Bassi,
R., and Cattivelli, L. (2003). car gene expression in barley mutants affected in

chloroplast development and photosynthetic electron transport. Plant Physiol 131,
793—802.

Dubouzet, J.G., Sakuma, Y., Ito, Y., Kasuga, M., Dubouzet, E.G., Miura, S., Seki,
M., Shinozaki, K., and Yamaguchi-Shinozaki, K. (2003). OsDREB genes in
rice, Oryza sativa L., encode transcription activators that function in drought-,
high-salt- and cold-responsive gene expression. The Plant Journal 33, 751-763.

Fowler, 8., and Thomashow, M.F. (2002). Arabidopsis transcriptome profiling indicates
that multiple regulatory pathways are activated during cold acclimation in
addition to the CBF cold response pathway. Plant Cell 14, 1675-1690.

Gilmour, S.J., Hajela, RK, and Thomashow, M.F. (1988). Cold Acclimation in
Arabidopsis thaliana. Plant Physiol. 87, 745-750.

Gilmour, S.J., Fowler, S.G., and Thomashow, M.F. (2004). Arabidopsis transcriptional
activators CBF1, CBF 2, and CBF3 have matching functional activities. Plant Mol
Biol 54, 768-781.

Gilmour, S.J., Sebolt, A.M., Salazar, M.P., Everard, J.D., and Thomashow, M.F.
(2000). Overexpression of the Arabidopsis CBF3 transcriptional activator mimics

multiple biochemical changes associated with cold acclimation. Plant Physiol.
124, 1854-1865.

Gilmour, S.J., Zarka, D.G., Stockinger, E.J., Salazar, M.P., Houghton, J.M., and
Thomashow, M.F. (1998). Low temperature regulation of the Arabidopsis CBF
family of AP2 transcriptional activators as an early step in cold-induced COR
gene expression. The Plant Journal 16, 433-442.

Grant, P.A., Duggan, L., Cote, J., Roberts, S.M., Brownell, J.E., Candau, R., Ohba,
R., Owen-Hughes, T., Allis, C.D., Winston, F., Berger, S.L., and Workman,
J .L. (1997). Yeast Gcn5 functions in two multisubunit complexes to acetylate

nucleosomal histones: characterization of an Ada complex and the SAGA
(Spt/Ada) complex. Genes Dev. 1 1, 1640-1650.

Guy, C.L., and Haskell, D. (1987). Induction of freezing tolerance in spinach is
associated with the synthesis of cold acclimation induced proteins. Plant Physiol.
84, 872-878.

Guy, C.L., Niemi, K.J., and Brambl, R. (1985). Altered gene expression during cold
acclimation of spinach. PNAS 82, 3673-3677.

20

Haake, V., Cook, D., Riechmann, J.L., Pineda, 0., Thomashow, M.F., and Zhang,
J.Z. (2002). Transcription factor CBF 4 is a regulator of drought adaptation in
Arabidopsis. Plant Physiol. 130, 639-648.

Hajela, R.K., Horvath, D.P., Gilmour, S.J., and Thomashow, M.F. (1990). Molecular
cloning and expression of cor (Cold-Regulated) genes in Arabidopsis thaliana.
Plant Physiol. 93, 1246-1252.

Hao, D., Ohme-Takagi, M., and Sarai, A. (1998). Unique mode of GCC box
recognition by the DNA-binding domain of ethylene-responsive element-binding
factor (ERF domain) in plant. J Biol Chem 273, 26857-26861.

Hao, D., Yamasaki, K., Sarai, A., and Ohme—Takagi, M. (2002). Determinants in the
sequence specific binding of two plant transcription factors, CBF1 and NtERF2,
to the DRE and GCC motifs. Biochemistry 41, 4202-4208.

Hart, C., Nagy, F., and Meins, F. (1993). A 61 bp enhancer element of the tobacco
beta-1,3-glucanase B gene interacts with one or more regulated nuclear proteins.
Plant Mol Biol 21, 121-131.

Hope, I.A., and Struhl, K. (1986). Functional dissection of a eukaryotic transcriptional
activator protein, GCN4 of Yeast. Cell 46, 885.

Horvath, D.P., McLarney, B.K., and Thomashow, M.F. (1993). Regulation of
Arabidopsis thaliana L. (Heyn) cor78 in response to low temperature. Plant
Physiol. 103, 1047-1053.

Hsieh, T.-H., Lee, J.-T., Yang, P.-T., Chiu, L.-H., Charng, Y.-y., Wang, Y.-C., and
Chan, M.-T. (2002). Heterology expression of the Arabidopsis C-
Repeat/Dehydration Response Element Binding Factor 1 gene confers elevated
tolerance to chilling and oxidative stresses in transgenic tomato. Plant Physiol.

129, 1086- 1 094.

Ikeda, K., Steger, D.J., Eberharter, A., and Workman, J.L. (1999). Activationd
domain-specific and general transcription stimulation by native histone
acetyltransferase complexes. Mol. Cell. Biol. 19, 855-863.

Jaglo-Ottosen, K.R., Gilmour, S.J., Zarka, D.G., Schabenberger, 0., and
Thomashow, M.F. (1998). Arabidopsis CBF1 overexpression induces COR
genes and enhances freezing tolerance. Science 280, 104-106.

Jaglo, K.R., Kleff, S., Amundsen, K.L., Zhang, X., Haake, V., Zhang, J.Z., Deits, T.,
and Thomashow, M.F. (2001). Components of the Arabidopsis C-
Repeat/Dehydration-Responsive Element Binding Factor cold-response pathway
are conserved in Brassica napus and other plant species. Plant Physiol. 127, 910-
91 7.

21

Johnson-Flanagan, A.M., and Singh, J. (1987). Alteration of gene expression during
the induction of freezing tolerance in Brassica napus suspension cultures. Plant
Physiol. 85, 699-705.

Klucher, K.M., Chow, H., Reiser, L., and Fischer, RL. (1996). The
AINTEGUMENTA gene of Arabidopsis required for ovule and female

gametophyte development is related to the floral homeotic gene APETALAZ.
Plant Cell 8, 137-153.

Knight, H., Veale, E.L., Warren, G.J., and Knight, MR. (1999). The sfi'6 mutation in
Arabidopsis suppresses low-temperature induction of genes dependent on the
CRT/DRE sequence motif. Plant Cell 11, 875-886.

Koroleva, 0., Tomlinson, M., Leader, D., Shaw, P., and Doonan, J. (2005). High-
throughput protein localization in Arabidopsis using Agrobacterium-mediated
transient expression of GFP-ORF fusions. Plant J 41, 162-174.

Kreps, J.A., Wu, Y., Chang, H.-S., Zhu, T., Wang, X., and Harper, J.F. (2002).
Transcriptome changes for Arabidopsis in response to salt, osmotic, and cold
stress. Plant Physiol. 130, 2129-2141.

Krizek, B., and Sulli, C. (2006). Mapping sequences required for nuclear localization
and the transcriptional activation function of the Arabidopsis protein
AINTEGUMENTA. Planta 224, 612-621.

Kuo, M.-H., Zhou, J., Jambeck, P., Churchill, M.E.A., and Allis, CD. (1998).
Histone acetyltransferase activity of yeast Gcn5p is required for the activation of
target genes in vivo. Genes Dev. 12, 627-639.

Kurkela, S., Franck, M., Heino, P., Lang, V., and Palva, E. (1988). Cold induced gene
expresssion in Arabidopsis thaliana L. Plant Cell Reports 7, 495-498.

Levitt, J. (1980). Responses of plants to environmental stresses. (New York: Academic
Press).

Liu, Q., Kasuga, M., Sakuma, Y., Abe, H., Miura, S., Yamaguchi-Shinozaki, K., and
Shinozaki, K. (1998). Two transcription factors, DREBI and DREB2, with an
EREBP/AP2 DNA binding domain separate two cellular signal transduction
pathways in drought- and low-temperature-responsive gene expression,
respectively, in Arabidopsis. Plant Cell 10, 1391-1406.

Magnani, E., Sjolander, K., and Hake, S. (2004). From endonucleases to transcription
factors: evolution of the AP2 DNA binding domain in plants. Plant Cell 16, 2265-
2277.

Magome, H., Yamaguchi, S., Hanada, A., Kamiya, Y., and Oda, K. (2004). dwarf and
delayed-flowering 1, a novel Arabidopsis mutant deficient in gibberellin

22

biosynthesis because of overexpression of a putative AP2 transcription factor. The
Plant Journal 37, 720-729.

Mao, Y., Pavangadkar, K., Thomashow, M., and Triezenberg, S. (2006). Physical and
functional interactions of Arabidopsis ADA2 transcriptional coactivator proteins
with the acetyltransferase GCNS and with the cold-induced transcription factor
CBF1. Biochirnica et Biophysica Acta 1759, 69-79.

Maruyama, K., Sakuma, Y., Kasuga, M., Ito, Y., Seki, M., Goda, H., Shimada, Y.,
Yoshida, S., Shinozaki, K., and Yamaguchi-Shinozaki, K. (2004).
Identification of cold-inducible downstream genes of the Arabidopsis

DREBlA/CBF3 transcriptional factor using two microarray systems. The Plant
Journal 38, 982-993.

McKown, R., Kuroki, G., and Warren, G. (1996). Cold response ofArabidopsis
mutants impaired in freezing tolerance. Journal of Experimental Botany 47, 1919-
1925.

Mohapatra, S.S., Poole, R.J., and Dhindsa, RS. (1987). Changes in protein patterns
and translatable messenger RNA populations during cold acclimation of alfalfa.
Plant Physiol. 84, 1 172-1 176.

Ohme-Takagi, M., and Shinshi, H. (1995). Ethylene-inducible DNA binding proteins
that interact with an ethylene-responsive element. Plant Cell 7, 173-182.

Ohta, M., Matsui, K., Hiratsu, K., Shinshi, H., and Ohme—Takagi, M. (2001).
Repression domains of Class II ERF transcriptional repressors share an essential
motif for active repression. Plant Cell 13, 1959-1968.

Qin, F., Sakuma, Y., Li, J., Liu, Q., Li, Y.-Q., Shinozaki, K., and Yamaguchi-
Shinozaki, K. (2004). Cloning and functional analysis of a novel DREBI/CBF
transcription factor involved in cold-responsive gene expression in Zea mays L.
Plant Cell Physiol. 45, 1042-1052.

Rashotte, A.M., Mason, M.G., Hutchison, C.E., Ferreira, F.J., Schaller, G.E., and
Kieber, J.J. (2006). A subset of Arabidopsis AP2 transcription factors mediates
cytokinin responses in concert with a two-component pathway. PNAS 103,
11081-11085.

Riechmann, J.L., and Meyerowitz, EM. (1998). The AP2/EREBP family of plant
transcription factors. Biol Chem 379, 633-646.

Rudolph, A.S., and Crowe, J.H. (1985). Membrane stabilization during freezing: The
role of two natural cryoprotectants, trehalose and proline. Cryobiology 22, 367.

Sakai, A., and Larcher, W. (1987). Frost survival of plants: responses and adaptation to
freezing stress. (Berlin: Springer-Verlag).

23

Sakuma, Y., Liu, Q., Dubouzet, J.G., Abe, H., Shinozaki, K., and Yamaguchi-
Shinozaki, K. (2002). DNA-binding specificity of the ERF/AP2 domain of
Arabidopsis DREBs, transcription factors involved in dehydration- and cold-
inducible gene expression. Biochem Biophys Res Commun 290, 998-1009.

Sato, F., Kitajima, S., and Koyama, T. (1996). Ethylene-induced gene expression of
osmotin-like protein, a neutral isoform of tobacco PR-5, is mediated by the
AGCCGCC cit-sequence. Plant Cell Physiol. 37, 249-255.

Seipel, K., Georgiev, 0., and Schaffner, W. (1994). A minimal transcription activation
domain consisting of a specific array of aspartic acid and leucine residues. Biol
Chem Hoppe Seyler 375, 463-470.

Seki, M., Narusaka, M., Abe, H., Kasuga, M., Yamaguchi-Shinozaki, K., Carninci,
P., Hayashizaki, Y., and Shinozaki, K. (2001). Monitoring the expression
pattern of 1300 Arabidopsis genes under drought and cold stresses by using a
Full-Length cDNA microarray. Plant Cell 13, 61-72.

Seki, M., Narusaka, M., Ishida, J., Nanjo, T., Fujita, M., Oono, Y., Kamiya, A.,
Nakajima, M., Enju, A., Sakurai, T., Satou, M., Akiyama, K., Taji, T.,
Yamaguchi-Shinozaki, K., Carninci, P., Kawai, J., Hayashizaki, Y., and
Shinozaki, K. (2002). Monitoring the expression profiles of 7000 Arabidopsis
genes under drought, cold and high-salinity stresses using a full-length cDNA
microarray. The Plant Journal 31, 279-292.

Skinner, J., van Zitzewitz, J., Szucs, P., Marquez-Cedillo, L., Filichkin, T.,
Amundsen, K., Stockinger, E., Thomashow, M., Chen, T., and PM., H.
(2005). Structural, functional, and phylogenetic characterization of a large CBF
gene family in barley. Plant Mol Biol 59, 533-551.

Steponkus, P., and Webb, M. (1992). Freeze-induced dehydration and membrane
destabilization in plants. (Berlin: Springer-Verlag).

Steponkus, P.L., Uemura, M., Joseph, R.A., Gilmour, S.J., and Thomashow, M.F.
(1998). Mode of action of the COR15a gene on the freezing tolerance of
Arabidopsis thaliana. PNAS 95, 14570-14575.

Stockinger, E.J., Gilmour, S.J., and Thomashow, M.F. (1997). Arabidopsis thaliana
CBF1 encodes an AP2 domain-containing transcriptional activator that binds to
the C-repeat/DRE, a cis-acting DNA regulatory element that stimulates
transcription in response to low temperature and water deficit. PNAS 94, 103 5-
1040.

Stockinger, E.J., Mao, Y., Regier, M.K., Triezenberg, S.J., and Thomashow, M.F.
(2001). Transcriptional adaptor and histone acetyltransferase proteins in
Arabidopsis and their interactions with CBF1 , a transcriptional activator involved
in cold-regulated gene expression. Nucl. Acids Res. 29, 1524-1533.

24

Strauss, G., and Hauser, H. (1986). Stabilization of lipid bilayer vesicles by sucrose
during freezing. PNAS 83, 2422-2426.

Thomashow, M.F. (1998). Role of cold-responsive genes in plant freezing tolerance.
Plant Physiol. 118, 1-8.

Thomashow, M.F. (1999). Plant cold acclimation: fi'eezing tolerance genes and
regulatory mechanisms. Annual Review of Plant Physiology and Plant Molecular
Biology 50, 571-599.

Tran, E.J., and Wente, S.R. (2006). Dynamic Nuclear Pore Complexes: Life on the
Edge. Cell 125, 1041.

Triezenberg, SJ. (1995). Structure and function of transcriptional activation domains.
Current Opinion in Genetics & Development 5, 190.

van der Fits, L., and Memelink, J. (2001). The jasmonate-inducible AP2/ERF-domain
transcription factor ORCA3 activates gene expression via interaction with a
jasmonate-responsive promoter element. The Plant Journal 25, 43-53.

Vlachonasios, K.E., Thomashow, M.F., and Triezenberg, SJ. (2003). Disruption
mutations of ADA2b and GCN5 transcriptional adaptor genes dramatically affect
Arabidopsis growth, development, and gene expression. Plant Cell 15, 626-638.

Vogel, J.T., Zarka, D.G., Van Buskirk, H.A., Fowler, S.G., and Thomashow, M.F.
(2005). Roles of the CBF2 and ZAT12 transcription factors in configuring the low
temperature transcriptome of Arabidopsis. The Plant Journal 41, 195-211.

Volger, H., and Heber, U. (1975). Cryoprotective leaf proteins. Biochim Biophys Acta
412, 335-349.

Wang, Z., Triezenberg, S., Thomashow, M., and EJ., S. (2005). Multiple hydrophobic
motifs in Arabidopsis CBF1 COOH-terminus provide functional redundancy in
trans-activation. Plant Mol Biol 58, 543-559.

Wilson, K., Long, D., Swinburne, J., and Coupland, G. (1996). A dissociation
insertion causes a semidominant mutation that increases expression of TINY, an
Arabidopsis gene related to APETALA2. Plant Cell 8, 659-671.

Wuitschick, J.D., Lindstrom, P.R., Meyer, A.E., and Karrer, KM. (2004). Homing
endonucleases encoded by germ line-limited genes in T etrahymena thermophila
have APETALA2 DNA binding domains. Eukaryotic Cell 3, 685-694.

Xin, Z., and Browse, J. (1998). eskimoI mutants ofArabidopsis are constitutively
freezing-tolerant. PNAS 95, 7799-7 804.

25

CHAPTER TWO
FUNCTIONAL ROLE OF THE CONSERVED PKKPAGR AND DSAWR MOTIFS IN
CBF1 ACTIVITY

INTRODUCTION

CBF proteins are members of the AP2/EREBP family of transcription factors in
Arabidopsis (Riechmann and Meyerowitz, 1998). This multi-gene family is highly
conserved among plants and is characterized by the presence of the AP2/EREBP DNA-
binding domain (Jofuku et al. , 1994; Ohme-Takagi and Shinshi, 1995). NMR studies on
AtERFl (Allen et al. , 1998) revealed that this DNA-binding domain is composed of a
three-strand B-sheet structure followed by an or-helix. The B-sheet, at the N-terrninus, is
the fold which mediates DNA recognition. High identity within the DNA-binding domain
and conservation of the critical amino acids required for protein-DNA interaction
suggests that proteins in the AP2/EREBP family likely adopt a similar 3D structure.

Within the AP2/EREBP family in Arabidopsis, CBF1/DREB1B, CBF2/DREB1C,
CBF3/DREB1A, CBF4/DREB1D, DREBlE/DDFI and DREBlF/DDF2 (Stockinger et
al., 1997; Gilmour et al. , 1998; Liu et al., 1998; Haake et al. , 2002; Sakuma et al. , 2002;
Magome et al. , 2004) define a small sub-family that is characterized by the presence of
two short polypeptide sequences that flank the AP2/EREBP domain:
PKKP/RAGRxKFxETRHP (abbreviated as PKKPAGR) and DSAWR, at the N-and C-
terrninus, respectively (Figure 2.1), designated the “signature” sequences (Jaglo et al.,

2001).

27

1 32 47 105110 213

 

Figure 2.1. PKKPAGR and DSAWR signature sequences in the CBF family of plant
transcription factors.

Top panel. Schematic diagram of CBF1 protein. Yellow boxes represent PKKPAGR and
DSAWR motifs. The grey box at the C-terminus represents the activation domain. The
AP2/EREBP DNA-binding domain is indicated

Bottom panel Sequence alignment of AtCBF1-3 and CBF-like proteins. The signature
sequences are included in the yellow boxes. The consensus sequence is indicated below.
At, Arabidopsis thaliana; B. napus, Brassica napus; N. tabacum, Nicotiana tabacum; S.
cereals, Secale cereale; T. aestivum, Tritium aestivran; G. max, Glycine max; H.
vulgare, Horderan vulgae.

28

Orthologues of the CBF genes are present in a wide variety of plant species. A
sequence alignment of CBF representatives from evolutionarily distant plant species
showed that most of the identity among the Arabidopsis CBF family and CBF-like
proteins from other species is due to a high degree of identity between their AP2/EREBP
domains and the signature sequences (J aglo et al., 2001). Conservation of these motifs
across evolutionarily diverse plants species suggests that they have an important
functional role in CBF activity.

When AtCBFl was first isolated, the PKKPAGR motif was proposed to
represent the signal responsible for nuclear localization of the protein (Stockinger et al. ,
1997). This observation was based on the similarity of this sequence to known nuclear
localization signals from other plant proteins, such as the early induced Aux/1AA genes
(Abel et al. , 1994), the photomorphogenic repressor protein COPl (von Arnim and Deng,
1994) and the heat shock factors, HSF (Lyck et al., 1997). In addition, localization
studies of the member of the AP2/EREBP family AINTEGUMENTA (Klucher et a1. ,
1996) have indicated that nuclear targeting can depend on the presence of specific basic
residues (Krizek and Sulli, 2006).

Whether nuclear shuttling of CBF1 is dependent on the PKKPAGR motif, and
whether this is a low temperature-regulated process is not clear. Recent studies focused
on the subcellular localization of several members of the AP2/EREBP family have
revealed that nuclear targeting within this family can be either constitutive or stirnulus-
dependent (Klucher et al. , 1996; Rashotte et al. , 2006); therefore it is not possible to draw
a general conclusion based on studies on other AP2/EREBP family members. It is

expected that some CBF protein can accumulate in the nucleus in the absence of a cold

29

stimulus, as Arabidopsis lines overexpressing CBF1, CBF 2 or CBF 3 display constitutive
accumulation of COR transcripts compared to the non-transgenic plants (J aglo-Ottosen et
al., 1998; Gilmour et al., 2000). On the other hand, when CBF-overexpressing plants are
cold-treated, COR gene expression increases (Gilmour et al. , 2000), suggesting that
nuclear transport may be more efficient at low temperatures.

For several families of DNA-binding proteins, regions flanking the DNA-binding
domain can be an essential extension important not only for binding affinity but for
binding specificity as Well (Crane-Robinson et al., 2006). Based on their proximity to the
AP2/EREBP domain, a reasonable hypothesis is that the PKKPAGR and DSAWR motifs
participate in binding to the CR T /DRE promoter element. In addition, the N-terrninal
PKKPAGR region of this motif is rich in basic residues, bearing positive charges that
may favor a DNA-protein interaction. Finally, deletion studies of ERFl have shown that
the presence of 10 amino acids immediately upstream and 8 amino acids downstream of
the AP2/EREBP domain can greatly stabilize DNA binding (Hao et al., 1998). Based on
these observations, we hypothesized that the two signature motifs might play a role in
CBF binding to its cognate CR T/DRE promoter element.

The main goal of the experiments described in this chapter was to investigate the
roles of the PKKPAGR and DSAWR motifs in CBF function. The question was
addressed by taking a mutational approach. Arabidopsis transgenic lines were generated
by overexpressing CBF1 transgenes harboring specific mutations in the PKKPAGR and
DSAWR motifs. As a hallmark of CBF1 activity, COR gene expression was tested. The
results demonstrate that both conserved motifs are required for CBF1 function, since

Arabidopsis plants overexpressing a CBF1 transgene carrying mutations in the signature

30

sequences show greatly diminished COR gene expression compared to plants
overexpressing the wild type CBF1 transgene. The effect of the mutations cannot be
explained by reduced protein levels, as indicated by western blot analysis of plants
overexpressing the 6xMyc: CBF1 transgenes. Contrary to what was originally
hypothesized, the PKKPAGR motif alone is not required for nuclear localization of a
CBF1 :GFPzGUS chimera. Electrophoretic-mobility gel shift experiments showed that the
PKKPAGR and DSAWR motifs play a role in mediating the recognition of the CRT/DRE
cis-acting element present in the COR gene promoters. Extensive mutational analysis
within the PKKPAGR motif combined with secondary structure prediction methods
suggested that certain residues in the predicted helical region represented by the
RKKFRET sequence are involved in direct protein-DNA interaction. A computational
model of CBF1 bound to its target CR T/DRE promoter element has led to the
identification of specific residues that are essential to DNA binding and might mediate

specific DNA recognition.

31

RESULTS

A MUTATIONAL APPROACH TO INVESTIGATE THE FUNCTIONAL

IMPORTANCE OF THE SIGNATURE SEQUENCES IN CBF1 FUNCTION.

To investigate the importance of the signature sequences in CBF function,
Arabidopsis transgenic lines were generated by overexpressing CBF1 transgenes
harboring specific mutations in the PKKPAGR and DSAWR motifs. AtCBFI was
selected as a representative model for the Arabidopsis CBF family, as it has been shown
that overexpression ofAtCBFI, AtCBF 2, or AtCBF 3 results in activation of the same
CBF regulon of genes, indicating that the three proteins have overlapping function
(Gilmour et al. , 2004). Alanine scanning was used to introduce stretches of three alanines
in place of the original amino acids within the PKKPAGR motif, which resulted in 5
mutant ORFs named M1-M5 (Figure 2.2). In addition, the entire PKKPAGR motif was
deleted, and the resulting ORF named APKK. The wild type and mutagenized versions of
CBF1 driven by the constitutive 35$ Cauliflower Mosaic Virus (3 SS CaMV) promoter
were transformed into Arabidopsis and are referred hereafter as pkkpagr (APKK, M1-M5)
and dsawr lines. For each construct, fifteen to twenty independent homozygous lines
were selected, and five to seven representative lines were chosen for further

characterization.

32

 

M51

142 M3 M4

105 no 213
II I

II I

WT DSAWR
(is art: AAAAA

Figure 2.2. Site-directed mutagenesis was used to generate CBF1 ORFs containing
specific mutations or deletions of the signature sequences.

Top. Schematic of full-length CBF1 and CBF1 lacking the PKKPAGR motif (APKK).
Bottom. Description of specific mutations within the signature sequences.

WT: wild type sequence for the two signature sequences.
M1-M5, dsawr: Arrrino acids in red indicate where the wild type CBF1 sequence has

been substituted with alanines.

33

Overexpression of AtCBF I , AtCBF 2, or AtCBF 3 leads to constitutive expression
of the CBF regulon, a group of about 85 genes containing the cis-acting CRT/DRE
promoter element (V ogel et al., 2005). As a result, CBF-overexpressing plants are
constitutively freezing-tolerant. However, overexpression of wild type CBF1 results in
transgenic plants that are smaller in stature and delayed in flowering compared to the
wild type plants or transgenic plants harboring the empty vector. In addition, higher
amounts of CBF transcripts correlate with smaller plants that are delayed in flowering
(Haake et al., 2002; Gilmour SJ et al., 2004; Gilmour et al., 2004).

Transgenic lines overexpressing CBF1-pkkpagr at levels similar to or higher than
plants overexpressing wild type CBF1 were selected by Northern blot analysis (Figure
2.3 C) and their growth phenotype was compared (representative examples are shown in

Figure 2.3 A).

34

\

 

Figure 2.3. Arabidopsis pkkpagr and dsawr lines.

Seedlings harboring the 35S::CBFI construct (026 and G39), CBF1 mutants (M1 —5,
APKK, and dsawr) or the empty vector (B6) were grown on plates for 14 days before
being transferred to soil. Pictures were taken 30 days after transplanting into soil

A pkkpagr lines.

B. Left, dsawr transgenic line.

C. Northern blot analysis of different 358::CBF1 Arabidopsis plants. Total RNA
extracted from two-week old seedlings was tested for the expression of CBF1. RNA
levels of the 188 ribosomal RNA were used as a normalization control.

35

Overall, overexpression of CBF1 carrying alanine mutations in the PKKPAGR
motif resulted in plants showing different degrees of growth retardation, as compared to
the control line B6, carrying the empty vector. However, the growth retardation observed
in the pklgoagr lines was always less severe than that observed in the WT CBF1 over-
expressing plants, when similar transcript levels were compared (compare plant growth
and CBF1 expression levels in Figures 2.3A and C). These observations suggested that
the PKKPAGR motif is required for CBF1 fimction.

As was observed in the pkkpagr lines, the negative effects of CBF1-dsawr
overexpression on plant growth were much milder than those in the wild type
overexpressor (Figure 2.3B). It must be noted that CBF1 transcript levels in the dsawr
plants were similar to or higher than in the plants overexpressing wild type CBF1
(representative example shown in Figure 2.3 C), suggesting that the DSAWR motif also
plays a role in CBF1 function.

In summary, overexpression of CBF1 transgenes mutated in the two signature
sequences altered of one of the hallmark phenotypes of CBF-overexpressing plants,
growth retardation, providing a first suggestion that these two conserved motifs play a

role in CBF1 activity.

36

FUNCTIONAL ROLE OF THE PKKPAGR MOTIF IN CBF1 ACTIVITY.

Mutations in the PKKPAGR motif affect COR gene activation.

To assess the effect of alanine mutations on CBF1 activity, wild type and pkkpagr
lines were analyzed for COR gene expression by northern blot analysis. Overexpression
of AtCBF 1 leads to constitutive expression of the CBF regulon, including COR6. 6,
COMM, COR47, and COR 78 (Jaglo-Ottosen et al., 1998). If the PKKPAGR motif is
required for CBF1 activity, we would expect that the ability of CBF1 to activate these
target genes would be impaired or absent in transgenic plants expressing CBF proteins
with mutations in this motif. This possibility was tested by determining expression levels
of target COR genes in transgenic plants overexpressing wild type and mutant versions of
the PKKPAGR sequence.

Total RNA was analyzed for the expression of CBF] and COR genes in
independent pkkpagr lines, as well as in transgenic lines overexpressing the wild type
CBF1 or harboring the empty vector. The experiment was carried out by growing
Arabidopsis seedlings at control temperature (22°C). At this temperature the endogenous
CBF and COR gene transcripts are barely detectable by northern blotting (Figure 2.4A
and B, see B6 line, which is transformed with the empty vector). In contrast, transgenic
lines overexpressing wild type CBF1 are able to activate constitutive expression of the

COR genes (Jaglo-Ottosen et al., 1998; Figure 2.4A and B, see all WT lines).

37

WT APKK M1 M2

 

v-I‘O
u-I
BEss-Nnean-s_s—nsza

COR6.6 -.. ."fltuo-"N-“fl

COR” 040-" """‘"‘“-' -oo.q~-.l

COR15a fl .-

rss ‘i’W-MQ

Figure 2.4 A. Analysis of COR gene induction in APKK, M1 and M2 pklrpagr lines by
Northern blot analysis. Total RNA was extracted from 2-week Old seedlings and tested
for the expression of CBF1, COR6. 6, COR 78, and COR15a. 18S ribosomal RNA levels
are shown as loading control.

B6, transgenic plants carrying an empty expression vector; WT, wild type CBF1
overexpressing plants; APKK, M1 and M2, transgenic lines overexpressing CBF1 lacking
the entire PKKPAGR region or carrying triple alanine mutations, respectively. Numbers
above lanes represent independent transgenic lines.

38

 

1 5

B6
G5
G6
G2
G26
G39

CBF1
COR6.6 - - n.-

188 uni...
n
COR78 nu...

COR15a ”I! Q.“

133 macaque.

M3 M4 M5

 

N M O—i
r-IMVIOb—r—iNINQ—I—IMBG—I—r

.. we”- a. -1" Man. U ""
.o .- "~i-i. .na---..§ "

7‘17
---....- u‘-..‘- -
H ’ I a it the .n

afiwfiidgflmmflhm

Figure 2.4 B. Analysis of COR gene induction in M3, M4 and M5 pkkpagr lines by
Northern blot analysis. Total RNA was extracted from 2-week old seedlings and tested
for the expression of CBF1, COR6. 6, COR 78, and COR15a. 18S ribosomal RNA levels

are shown as a loading control.

B6, transgenic plants canying an empty expression vector; WT, wild type CBF1
overexpressing plants; M3-M5, transgenic lines overexpressing CBF1 carrying triple
alanine mutations. Numbers above lanes represent independent transgenic lines.

39

The pkkpagr lines showed higher COR gene induction than the control line B6,
suggesting that the CBF1 mutated proteins had retained some ability to activate
transcription (Figures 2.4 A and B, compare pkkpagr lines Ml-MS, and APKK with B6).
However, this function was affected by alanine substitutions in the PKKPAGR motif. For
instance, C 0R1 5a induction was greatly reduced in all the pkkpagr lines as compared to
the transgenic plants overexpressing wild type CBF1 transgene (Figures 2.4 A and B,
Ml-M5 lines), and almost undetectable in the lines overexpressing CBF1 transgene
lacking the whole PKKPAGR motif (Figures 2.4 A, APKK lines). Induction of COR6.6
and COR78 appeared to be also affected, in that higher levels of mutated CBF1 transcript
were needed to reach comparable COR transcript levels (Figures 2.4 A and B, compare
COR6. 6 and COR 78 transcript levels and corresponding CBF1 transcript levels in WT
and pkkpagr lines M1-M5). To determine whether mutations in the PKKPAGR motif had
a significant effect on COR6.6 and COR 78 induction, we compared the COR/CBF1
transcript ratios in the transgenic lines overexpressing wild type and mutated CBF1 by

ANOVA (Figures 2.5A and B).

40

150-

100-

COR78/CBF1

50-

 

   

 

0.0 ‘ " ‘
WT APKK M1 M2
40004
3200—
24.00-

160)“

COR78/CBF1

800—

 

 

o'm __7 7,, . .- 7;..11 'i’i. . L"»' 1‘ ,w. 'z, ("MW "1‘51"".
WI" M3 M4 M5

 

 

Wl'versusAPl<K|V|1.MZ WTversusNBJWJVB
Genotype (knotype

oorrparison Estirrete tvalue PvaiuecorrpaismEstimatetvdue Pvdue
hWTvs.APKK 4.2876 -16.62_s.m011Wl'vs.NB -1.1174 -9.14 <.(D01_
_7 WTvs.Ml -1_.7986 -23.22 <.0001 Wl'vs.M_4 -1.0185 .798 <.ooo1q
. WTvs.MZ 4.2870 46.62 $0001 WTvs.N5 07671 -6.28_ <.0001,
APKKvs.M1 05109 ,‘6-59 0.9968 'NBvs.M4 00989 7-0.81 _O.4288_
APKKstz 43,-0006 o-oo .<-0001 revs-NB. 00660 -3-01 00076.
M1vs.MZ 05106 -6.59 <.CID1 M4vs.|VE —0.2514 -2CB 0.0054

 

 

 

 

 

 

 

 

 

 

 

 

Figure 2.5A. Analysis of variance of COR 78/CBF1 transcript ratios in transgenic plants
overexpressing wild type or mutated CBF1 transgenes.

Top. Bar graphs represent the average of COR 78/CBFI ratios for 5-6 independent
transforrnants for different transgenic lines. WT, Ml-M5, APKK, transgenic lines
overexpressing CBF1 wild type, its mutated versions or lacking the PKKPAGR motif.
Bottom. Differences of least square means of COR78/CBF1 transcript ratios for
transgenic lines overexpressing CBF1 wild type or its mutated versions.

41

31(1)

 

 

 

 

 

 

 

 

 

 

 

25.00 T
'5 20.00- i
9
co 15.00
E
O 10.00
0
5.0)
on) 4 _.—l'—"—f__“'_‘_l ,4
WT APKK M1 M2
30.00-
‘_ 25.(D~ I
u.
m 20.0)-
2 1
0. 15.CD- .
E
o 10.00-
0
5.11)—
0.00 7,4,7, ,, ' I ' I
WI” M3 M4 M5
Differences of Least Square Means - COR6.6/CBF1 ratios
WT versus APKK, M1. M2 WT versus M3, M4. M5
Genotype Genotype
comparison Estimate tvalue Pvalue comparison Estimate tvalue Pvalue
WTvs.APKK -1.0578 -14.37 <.0001 WTvs. M3 -1.2853 -7.04 <.0001
WTvs. M1 -1.2605 -17.12 <.0001 WTvs. M4 -1.1368 -5.96 <.0001
WTvs. M2 -1.1783 -16.01 <.0001 WTvs. M5 -1.1158 -6.11 <.0001
APKK vs. M1 -0.2026 -2.75 0.0142 M3 vs. M4 -0.1485 -0.81 0.4268
APKKvs. M2 -0.1204 -1.64 0.1213 M3 vs. M5 -0.1695 -0.97 0.3433
M1 vs. M2 -0£§21 -1.12 0.2807 M3 vs. M4 -0.0210 -0.11 0.9097

 

 

 

 

 

 

 

 

 

 

 

Figure 2.5B. Analysis of variance of C 0R6. 6/CBF I transcript ratios in transgenic plants
overexpressing wild type or mutated CBF1 transgenes. Loglo values were used for this
analysis. Top. Bar graphs represent the average of C 0R6. 6/CBF1 ratios for 5-6

independent transforrnants for different transgenic lines. WT, Ml-MS, APKK, transgenic
lines overexpressing CBF1 wild type, its mutated versions or lacking the PKKPAGR

motif.

Bottom. Differences of least square means of COR6. 6/CBF1 transcript ratios for
transgenic lines overexpressing CBF1 wild type or its mutated versions.

42

In all cases, the difference in COR 78/CBF1 and COR6. 6/CBF1 transcript ratios
between plants overexpressing the native CBF1 (control lines) and any of its mutated
version was highly significant (P<0.0001) according to ANOVA analysis. For instance,
COR 78/CBFI ratio values ranged between 20.5 and 37 in the control lines, whereas M5
mutation, which showed the smallest effect on COR 78 accumulation, (Figure 2.48,
compare WT and M5 lines) resulted in a reduction of COR 78/CBF1 ratio of 7.4 folds
(Figure 2.5A). Similarly, a COR6. 6/CBF1 ratio of approximately 23 folds in the control
lines ranged between 1-2 folds in the pklgoagr transgenic lines (Figure 2.5B).

In summary, all the mutations introduced in the PKKPAGR motif caused a
reduction in COR gene induction in the pkkpagr lines compared to the lines
overexpressing wild type CBF1. These results indicate that an intact PKKPAGR motif is

required for CBF1 transcriptional activity.

Reduced COR gene induction in the pkkpagr lines is not caused by lower protein levels.
The reduced COR gene expression observed by northern blot analysis could be due
to different reasons. One would be that the mutations generated within the PKKPAGR region
reduced CBF1 stability; i.e. for a given transcript level, less mutated protein accumulated
compared to the wild type protein. To test this possibility, we analyzed CBF1 protein levels
in Arabidopsis protein extracts of wild type and transgenic plants.

Antibodies raised against full-length CBF1 or specific peptides have failed to detect
CBF1 protein in either cold-acclimated wild type or 35S::CBF1 transgenic Arabidopsis lines

(J aglo-Ottosen PhD thesis; Canella, unpublished). Therefore, to conduct protein detection

43

experiments by western blot analysis, we generated Arabidopsis plants overexpressing the
6xMyc:CBF1 transgene under control of the 35S CaMV promoter (Figure 2.6A). This
particular tag was chosen because it has been successfully used to detect different proteins in

plant extracts by western blot analysis (DeWitt et al. , 1996; Rivas et al. , 2002).

44

6xMyc

 

 

 

WT M1 M2 M3 M4 M5

 

 

Col-O
G6
G 6

N 61849 31112 2 925181310313214232627

 

CBF1 7 "" w m- u . a __ fl..-e*~
CORISO

188 I'- --'w Deadbwg

51-. "

 

 

 

 

C, WTO Ml WT M2M3 M4 M5
228‘ng §es~a_2:a:a

 

rfl—v)! '- P
a-c-Myc fl. "' ” :_ "I " .... .. "' '
i .

. u“ . II. 8

Figure 2.6. Comparison of CBF 1 transcript and protein levels in Arabidopsis plants
overexpressing 6xMyc:CBF1 transgene.

A. Schematic of 6xMyczCBF1 construct. The N-terminal 6xMyc tag is indicated by the
dark grey boxes; yellow boxes represent the two signattu'e sequences flanking the DNA-
binding domain.

B. Northern blot analysis of 358::6xMyczCBF1 transgenic lines. Total RNA was
analyzed for CBF1 and CORISa expression. Transcript amounts were normalized using
the 18S ribosomal loading control.

C. Western blotting showing the presence of 6xMyc:CBFl protein in the transgenic
plants (detected with a monoclonal anti-c-Myc antibody).

Col-0, non-transgenic plants; G6 and G26, Arabidopsis plants overexpressing wild type
CBF1 without 6xMyc tag; WT and M1-M5, transgenic plants overexpressing a 6xMyc
tag fused to a wild type (WT) or mutated (M1-5)CBF1 transgene. Protein loading is
indicated by the Coomassie-stained membrane.

45

Several independent lines per construct were analyzed for the expression levels
of CBF] and COR genes. Northern blot analysis indicated that plants overexpressing a
wild type CBF1 transgene fused to a 6xMyc tag could accumulate COR15a levels
comparable to plants overexpressing the wild type CBF1 transgene (Figure 2.6B). In
contrast, overexpression of 6xMyc:CBFI transgenes carrying mutations in the
PKKPAGR motif resulted in a dramatic reduction of COR15a accumulation. These data
indicate that addition of a tag to the N-terrninus of CBF1 doesn’t significantly affect the
activity of CBF1 wild type or CBF1 mutated in the PKKPAGR motif.
To determine the levels of the CBF1 wild type of mutated protein in planta, total protein
extracts from two representative transgenic lines were chosen for western blot analysis.
Comparison of the northern and western blotting results indicated a good agreement
between CBF1 transcript and protein levels. For instance, the two independent lines for
the PKKPAGR mutation M2 (line #2 and line #25) show CBF1 transcript levels similar
to the control line WT18 (Figure 2.63). Consistently, the protein levels are very similar
(Figure 2.6C, right panel). Moreover, in transgenic lines that could accumulate high
levels of mutated CBF1 protein compared to the wild type protein, COR gene induction
was still much lower than in the wild type overexpressing lines (compare M3.1 and
M5.14 in Figure 2.6B and 265C, right panel). Therefore, we can conclude that the
reduced COR gene induction observed in the pkkpagr lines cannot be attributed to

reduced CBF1 protein accumulation.

46

Role of the PKKPAGR motif in nuclear targeting of CBF1.

As discussed above, one possible function of the PKKPAGR motif could be that
of mediating CBF targeting to the nucleus. To test this hypothesis, we generated
Arabidopsis plants overexpressing translational fusions of CBF1 to GFP:GUS. As a
control, Arabidopsis plants overexpressing GFP:GUS alone and Nuclear Inclusion
protein a (NIa):GFP:GUS (Carrington et al. , 1991) were generated. GFP:GUS and
NIa:GFP:GUS are known for being mainly localized to the cytoplasm and the nucleus,
respectively (Grebenok et al. , 1997). In addition to the full-length CBF1 ORF, a construct
lacking the PKKPAGR motif (CBFlAPKK) was designed to assess whether the nuclear
localization of CBF1 requires this region of the protein. To determine whether the
PKKPAGR motif is sufficient to drive nuclear entry of CBF1, the N-terminal fragment of
CBF1 containing the first 32 residues and an intact PKKPAGR motif (PKK:GFP:GUS)
was generated (Figure 2.7A). The localization of the proteins was monitored by laser
confocal microscopy.

Independent transgenic lines were selected by northern blot analysis and analyzed
by laser confocal microscopy. For the construct harboring the full-length CBF1 transgene
fused to GFP:GUS, it was not possible to isolate any transgenic plants, despite three
independent transformation events by the floral dip method (Clough and Bent, 1998).
One possible explanation for this is that those plants produced a highly stable CBF1
protein which is extremely toxic to plants. For the remaining constructs, northern blot
analysis of two representative lines confirmed the expression of the transgenes (Figure

2.7B, upper panel - see GFP probe).

47

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

1 32 47 105110 213
A. :5 l J-GFPGUS
a: PKK
fl l 1 APKK
[“1 u 1 802
L J 801
I: 800
:1 799
GFPGUS GFPGUS
B- Alone APKK Alone m
4 4 10 1 s (r) 1 3
CBF1 g .
_’_!
GFP I - -. .-
COR15a -
“S M .-
GFPGUS GFPGUS
Alone 799 800 Alone 80 I 802
68(‘)Nla237 22]2218(‘)345145
CBF1 ‘ .
GFP - I ¢ ‘ .

COR15a

183 m

 

Figure 2.7. Northern blot analysis of Arabidopsis seedlings overexpressing different
CBF1:GFP:GUS constructs.

A. Schematic of the different CBF1 :GFPzGUS constructs. CBF1(aal -213), full-length
protein, APKK, a construct in which the PKKPAGR region has been deleted; PKK, first
32 amino acids including the PKKPAGR motif. All the other constructs represent 5’
deletions of CBF1 ORF. Dark grey boxes represent the two signature sequences.

B. Northern blot analysis of transgenic lines overexpressing constructs indicated in the
top panel. Total RNA (10 ug) from 12-day-old seedlings was tested for CBF1, GFP, and
COR15a transcript levels. Numbers above the lanes indicate independent lines for each
given construct. The asterisk represent samples from cold-treated (4°C, 10 hours)
Arabidopsis seedlings overexpressing GFP:GUS alone.

48

To determine the cellular localization of different CBF1 :GFP:GUS chimeras, six-
to-seven day old Arabidopsis seedlings were grown under controlled (22°C) conditions
and root tips were analyzed by laser confocal microscopy (Figure 2.8). Analysis of the
control lines confirmed that GFP:GUS and NIa:GFP:GUS were mainly localized to the
cytoplasm and nucleus, respectively. The CBFIAPKKzGFPzGUS chimera could direct
the cytoplasmically localized GFP:GUS to the nucleus, indicating that nuclear
localization of the protein is not affected by the deletion of the PKKPAGR motif alone.
In addition, a fusion of the 47 N-terrninal residues of CBF1 to GFP:GUS could not direct
the chimera into the nucleus, indicating that the presence of the PKKPAGR motif in this
context is not sufficient to promote nuclear entry of CBF1:GFP:GUS.

That the PKKPAGR is neither required nor sufficient for nuclear localization of
CBF1 was a surprising finding, based on the resemblance to similar motifs that are
functional NLSs in other plant transcription factors, including the member of the
AP2/EREBP family AINTEGUMENTA (Abel et al. , 1994; Koroleva et al. , 2005; Krizek
and Sulli, 2006; Rashotte et al., 2006).

In smnmary, localization studies showed that, under the experimental conditions
presented, the lack of the PKKPAGR motif does not prevent CBF1 from entering the
nuclear compartment, and therefore decreased transcriptional activity of the protein

cannot be explained by failure to localize to the nucleus.

49

(iH'Ul S

\I’KK

 

Figure 2.8. Localization studies of CBF1 :GFP:GUS chimeras.

Laser confocal images of root tips of Arabidopsis plants overexpressing various
CBF1 :GFP:GUS chimeras.

Lefi, GFP fluorescence (green); center, Propidium Iodide fluorescence (red); right,
merged images (yellow color indicates overlap of red and green fluorescence). All
constructs described in Figure 2.6; Nla, Nuclear Inclusion protein A, a nuclear marker.

In addition, these results indicate that CBF1 can gain access to the nuclear
compartment through an active mechanism, since it can facilitate the nuclear import of
GFP:GUS (approximately 100KDa), whose size is well above the size cut-off for free
diffusion into the nucleus (Kaffrnan and O'Shea, 1999). For a facilitated transport across
the nuclear envelope, proteins need a recognition sequence for interaction with nuclear
receptors or directly with the nucleoporins that are an integral part of the Nuclear Pore

Complex (NPC).

Identifying regions in CBF1 involved in nuclear transport.

Several reports have indicated that it is not unusual to find multiple karyophilic
signals in nuclear proteins, leading to the suggestion that they could provide some
redundancy (Sudbeck and Scherer, 1997; Tang et al., 1997; Walther et al., 2005). For
instance, different NLS could provide independent mechanisms to target the protein to
the nucleus in reponse to different stimuli. If PKKPAGR represented one of multiple
NLSs in CBF1, then deletion of the PKKPAGR motif alone might not eliminate import
into the nucleus to a detectable level; however, simultaneous deletion of an additional
motif would affect this process, as it has been observed in the case of the thyroid
transcription factor-1 and HMG-containing proteins (Sudbeck and Scherer, 1997;
Christophe-Hobertus et al. , 1999). As a first approach to determine whether PKKPAGR
is one of multiple signals in CBF1 that are required for transport across the nuclear
membrane, 5’ deletions of CBF1 ORF fused to GFP:GUS were generated and

transformed into Arabidopsis plants for localization studies.

51

5’ deletions of CBF1 ORF were designed by analysis of the protein sequence. No
obvious nuclear localization motif could be identified in CBF1 based on the current
understanding of NLSs. In fact, the most commonly described nuclear localization signal
rich in basic residues, does not represent the consensus motif for nuclear localization, and
there have been numerous examples of nuclear import through novel targeting signals
(Poon and Jans, 2005). Consequently, the following 5’ deletions were chosen based on
motif conservation or on the presence of known domains in CBF proteins (Figure 2.7A):
deletion of residues 1-32 up to the PKKPAGR motif (construct 802), deletion up to the
C-terminal end of the PKKPAGR motif (construct 801 ), deletion of the DNA-binding
domain as a whole based on the presence of positive charges throughout the domain
(construct 800), and deletion of five additional residues that eliminated the DSAWR
motif (construct 799), based on its conservation in the CBF family of proteins.
Arabidopsis plants overexpressing each of these constructs were generated, and
overexpression of the transgene was confirmed by northern blot analysis for two to four
representative lines per construct (Figure 2.7B).

To determine the subcellular localization of the different CBF1 :GFP:GUS
chimeras, fluorescence images were captured by laser confocal microscopy by analysis of
the root tips of 6-7-day-old Arabidopsis seedlings grown at 22°C. Deletion of amino
acids 1-32 (construct 802), and of amino acids 1-47 (construct 801) did not appear to
significantly affect nuclear localization of CBF1 :GFP:GUS, confirming that the
PKKPAGR motif is not essential to this function (Figure 2.9). However, when the
AP2/EREBP DNA-binding domain (construct 800) or the DNA-binding domain and the

DSAWR motif (construct 799) were deleted, there was a detectable change in the cellular

52

distribution of the proteins, whereby GFP fluorescence could be detected not only in the
nucleus but throughout the cell. These observations indicated that the protein could no

longer be actively imported into, or retained in the nucleus.

53

 

Figure 2.9. Localization studies of 5’ deletions of CBF1 :GFP:GUS chimeras.

Laser confocal images of root tips of Arabidopsis plants overexpressing various
CBF1 :GFP:GUS chimeras.

Left, GFP fluorescence (green); center, Propidium Iodide fluorescence (red); right,
merged images (yellow color indicates overlap of red and green fluorescence). All
constructs described in Figure 2.6.

54

To verify that the fluorescence detected in planta indeed corresponded to the full-
length proteins expressed, and was not the result of unpredicted protein truncation or
degradation, total protein extracts were analyzed by western blot analysis. In all the
samples tested, it was possible to detect specific signals in which the observed molecular
weight corresponded approximately to the expected size for that construct, indicating that
the fluorescence detected by confocal imaging resulted from the localization of the
indicated proteins (Figure 2.10).

Taken together, these results indicate that efficient nuclear localization of
CBF1 :GFP:GUS can occur under warm conditions when the PKKPAGR region has been
deleted. The exact nature of the nuclear localization signal for CBF1 has yet to be
determined. The analysis presented here suggests that the AP2/EREBP DNA-binding
domain contains the signal required for efficient nuclear targeting and/or retention of
CBF1. Additional analysis will be required to identify the specific region(s) or residues

responsible for this process.

55

 

 

 

 

 

 

 

 

13 m hcarmrpzous
95 Ferrous
72‘ ..._. — ._...
55. - 72 -

Q

55
" * a f -,::CBF1:GFP:GUS
_ W *

 

Figure 2.10. Western blot analysis of total protein extracts from Arabidopsis seedlings
overexpressing different CBF1:GFP:GUS constructs or GFP:GUS alone.

Total protein extracts were prepared from 12-day-old seedlings grown at 22°C.
Monoclonal antibody raised against OF P was used for protein detection of 70 ug of total
protein extracts.

Top and bottom blots shown on the right represent shorter and longer exposures of the
same film, respectively.

ThetwobandsdetectedatSS and 72KDarepresentcmss-reactingbands,astheycanbe
detected in the protein sample from the non-transgenic plants represented by Arabidopsis
Ws-2.

56

Role of the PKKPAGR motif in DNA binding.

We hypothesized that the PKKPAGR motif might play a role in CBF] DNA
binding activity based on the proximity of this motif to the AP2/EREBP domain and the
presence of positively charged residues that might provide a favorable electrostatic
interaction with the negatively charged DNA. This hypothesis was tested in vitro using
electrophoretic-mobility shift assays to measure the binding activity of wild type and
mutated CBF1 proteins.

To conduct this analysis, recombinant full-length CBF1 proteins harboring the
same alanine mutations within the PKKPAGR region previously described, were fused to
a histidine tag (6xHis-CBF1 1:1,) for expression and purification from E. coli (Figure
2.11A). 6xHis-CBF1FL proteins were tested in a binding reaction in which 1 ng of 32P-
radiolabeled DNA probe containing the CRT/DRE binding site from the COR15a
promoter was mixed with 200 ng of each recombinant protein (Figure 2.11B). Equal
protein loading was tested by analyzing diluted protein aliquots directly taken from the
binding reaction by Western blotting, using a or-CBFI antibody (Figure 2.11C). As
previously reported (Stockinger et al., 1997), wild type CBF1 could cause a band shift of
a 20-bp fragment containing a monomer of the CR T/DRE element. The shift was
specifically abolished in the presence of unlabeled competitor DNA. The same unlabeled
DNA fragment carrying point mutations was unable to compete for binding, indicating
that CBF1 protein specifically binds the CRT/DRE cis-acting element. Unlike the wild
type protein, none of the mutant proteins were able to efficiently bind and cause a shift of

the 32P-labeled probe.

57

 

 

 

 

 

 

 

 

A. I] [j AP2 I I
be
$59— WT 5% SS) P .n
A—
_ _ W M —————— Competitor DNA
8' . ‘ 1. - ~ . ‘0. . .
COR15a
C.
. I,“ -~ or-CBF]
Empty
vector WT M3 M4 M5
' - ' " ot-CBFI

Figure 2.11. Triple alanine mutations abrogate the binding activity of a full-length
6xHis-CBF1 protein in vitro.

A. Schematic illustration of 6xHis-CBF1 constructs. Dark grey boxes, signature
sequences; light grey box, 6xHis tag; AP2, AP2/EREBP DNA-binding domain.

B. EMSA. 200 ng of each recombinant protein were used in a 12 ul binding reaction
containing 0.5 ng of radiolabeled CRT/DRE element from COR15a promoter. Where
indicated, 100 ng of unlabeled competitor DNA were added. W and M, wild type and
mutated competitor DNA.

C. Recombinant 6xHis-CBF1 proteins were analyzed by Western blot analysis using an
antibody raised against full-length CBF1. Three different protein amounts (50, 25 and
12.5 ng) were tested.

These data showed that the presence of an intact PKKPAGR sequence is crucial
to direct CBF1 to a CRT /DRE-containing element in vitro. The failure to bind could
result from improper folding, loss of side chain interaction, or both.

To test whether the PKKPAGR motif was sufficient to impart DNA binding
activity to the AP2/EREBP domain in the context of this domain alone, a new set of
constructs was designed, in which only the CBF 1 region including the DBD and the two
flanking signature sequences (CBFlzm 12) was used (Figure 2.12A). Since purification of
soluble full-length CBF1 firsed to a 6xHis tag had proven difficult, wild type and
PKKPAGR mutated versions of CBF127-112 were over-expressed as a translational fusion
to Maltose Binding Protein (MBP), which has been shown to increase protein solubility

when over-expressed in E. coli (Guana et al. , 1987).

59

27 112

 

 

 

 

 

 

 

A. l M Li AP2 I
my“ M1] miner“ M3 M4 M5
B.
UH
in!
C.

 

Figure 2.12. Triple alanine mutations abrogate the binding activity of MBP-CBFz-Hn
proteins in vitro.

A. Schematic illustration of MBP-CBF1 constructs. Dark grey boxes,signature
sequences; the MBP tag and the AP2/EREBP DNA-binding domain are indicated.

B. Electro mobility shift assays. Increasing amounts (0.2, 2.0 and 15 ug) of each
recombinantproteinwereusedina25ulbindingreactioninthepresenceof0.5 ngof
radiolabeled CRT/DRE element from COR15a promoter.

C. Western blot analysis of MBP alone and the different MBP-CBF1 proteins, reflecting
the increasing amormts of the recombinant proteins used in the electrophoretic-mobility
shift assay (anti-MBP antibody).

60

As previously observed in the context of the full-length wild type protein, MBP-
CBF127.1 12 was able to cause a specific shift of the labeled CRT/DRE probe (shown later,
Figure 2.24), indicating that wild type CBFlz-H 12 was sufficient for recognition of the
CR T/DRE promoter element. On the contrary, mutations in the PKKPAGR region caused
a dramatic reduction in binding affinity, as shown in the titration experiment of Figure
2.123.

In summary, a CBF1 fragment composed of the AP2/EREBP DNA-binding
domain and the two flanking signature sequences is sufficient to recognize a target
CRT /DRE promoter element. This activity is lost in CBF1 protein when the PKKPAGR

signature sequence is mutated.

Secondary structure predictions to investigate the structural preferences of the
PKKPAGR motif.

To gain a better understanding of how mutations within the PKKPAGR region
might affect DNA binding by CBF1, it is important to know the structural preferences
and spatial orientation of the native region relative to the target DNA. This information
will help to discriminate whether the contribution of PKKPAGR to binding is indirect, by
providing the proper folding required to fit on the COR gene promoters, direct, due to the
presence of essential side chain interactions that make DNA contacts, or both. To obtain
insights into the structural propensity of the PKKPAGR motif, we conducted secondary
structure prediction studies.

The search was carried out using a suite of web-based prediction servers that are

rated among the most accurate according to the independent evaluator EVA (Eyrich et

61

al., 2001). They are PredictProtein (Rost and Liu, 2003), Psipred (McGuffin et al., 2000),
SAM-T02 (Karplus et al., 2003), and SABLE (Adamczak et al., 2005). CBF1 residues 1-
60 were used as input, which includes the N-terminus of the protein, the PKKPAGR
motif and the first 12 residues of the B-sheet DNA-binding domain. This sequence length
represents the longest CBF1 fragment that restricts the automatic identification and
alignment of sequence homologs to close relatives of CBF1, but not other AP2/EREBP
proteins that do not contain the PKKPAGR motif. The secondary structure predictions
generated by the individual methods were compared to identify consistent predictions

across different methods (Table 2.1).

62

Table 2.1. Secondary structure predictions of the PKKPAGR motif

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

PROFsec, SAM-T02, SABLE, PSI PRED, Consensus,
P Cc +++d C ++ C +++ C +++ C
K C +++ C ++ C +++ C +++ C
K C +++ C +++ C +++ C +++ C
P C +++ C 'H' C +++ C ++ C
A C ++ d T, ++ C +++ C + C
G C ++ T ++ C +++ C + C
R C ++ Hc 0 C +++ C + C
K od H +d C ++ C + c/h
K 0 H + C + 0 c/ h
F 0 H ++ C ++ H + .2 c/h
R 0 H ++ C H H + c/ h
E o C ++ C ++ H + c/h
T 0 C “H" C +++ 0 C
R C ++ C ++ C +++ C 4+ C
H C +++ C ++ C +++ C +++ C
P C ++ C +++ C 4+ C ++ C ..

 

 

a PROF sec (Rost and Liu, 2003 ), SAM-T02 (Karplus et al., 2003), SABLE (Adamczak
et al., 2005), and PSIPRED (McGuffin et al. , 2000) are the prediction methods used.

b Agreement between different prediction methods. Capital letters listed in the consensus
column indicate agreement between the predictions, and lower case letters indicate

differences.

c C, random coil conformation; H, helicity, T, turn.

(1 The number of pluses refers to the strength of prediction; one, two and three pluses

indicate low, moderate and high confidence, respectively. Zero indicates that no
prediction could be made for those specific residues.
Individual residues of the PKKPAGR motif are indicated in the first column.

63

 

Overall, we observed a high confidence in predicting the secondary structure for
two clusters in the PKKPAGR motif, represented by residues PKKPAGR and RHP.
Based on the agreement between different methods, those two clusters are likely to be in
a coiled or irregular secondary structure (Table 2.1, very right column). On the contrary,
the confidence of prediction dropped for the central region represented by residues
KKFRET, with some helicity predicted by SAM-T02 and PsiPred, whereas SABLE and
JPred predicted coil, and PROFsec returned no prediction. Therefore, no secondary
structure could be assigned to the KKFRET residues within the PKKPAGR motif.

Since the above methods were principally developed to predict secondary
structure over full-length protein sequences, a local predictor - Sequery - was used for
further analysis of the KKFRET region (Collawn et al. , 1990; Craig et al. , 1998).
Sequery predicts the secondary structures of local motifs (typically 4-6 residues) by
identifying similar sequences in the Protein Data Bank, a database of experimentally
determined 3D structures. Sequery matches the query tetrapeptide to sequences in the
database and analyzes how often they occur in a particular secondary structure. To avoid
statistical bias due to the over-representation of structures from the same super-families, a
low-homology subset of protein chains with 525% sequence identity was generated.
Tetrapeptides within the N-flanking region of CBF1 were designed by progressively
sliding one residue at a time to cover the whole PKKPAGR region, resulting in 14 motif
searches, identified as exact matches (Table 2.2, left box). To ensure a significant number
of matches, we generated substitution patterns based on conservative substitutions and
the consensus motif described for the PKKPAGR motif (J aglo et al. , 2001). These

tetrapeptide substitutions were classified as close matches (Table 2.2, very lefi column).

64

Table 2.2. Sequery analysis of the 3D structures observed for tetrapeptide sequences in the

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

PKKPAGR region of CBF1.
Exact Matches. Close Matches.

Query, H (%). E (%). T (%). I (%), Hits; Queryc H (%) E (%) T (%) I (%) Hits
CPKK N/A N/A N/A N/A 0 (CS)(PS)(KR)(KR) 30 0 13 57 23
PKKP 0 0 100 2 (PS)(KR)(KR)(PS) 0 3 23 74 39
KKPA 0 100 4 (KRXKRXPSM 3 1 2 10 57 42
KPAG 14 O 86 7 (KR)(PS)AG 21 0 9 70 34
PAGR 0 22 78 9 (PS)AGR 14 0 19 67 21
AGRK 14 14 0 72 7 AGR(KR) 25 5 20 50 20
GRKK 38 0 0 62 8 GR(KR)(KR) 25 5 12 58 64
RKKF 56 0 22 22 9 R(KR)(KR)F 5 1 6 8 35 49
KKF R 75 0 25 4 (KR)(KR)F(KR) 56 12 6 26 34
KFRE 100 O O 0 6 (KR)F(KR)(ED) 56 2 7 35 73
FRET 100 0 0 3 F (KR)(ED)(TS) 47 3 21 29 38
RETR 33 17 17 33 6 (KRXEDXTSXKR) 30 7 12 51 67
ETRH 33 0 0 67 3 (EDXTS)(RK)H 36 0 36 28 14
TRHP 0 0 0 100 2 (TSXRKXHRK)? 0 0 32 68 38

Triple alanine substitutions.
Queryd H (%) E (%) T (%) I (%) Hits

RKKF —-» RAAA 83 0 0 14 22

KKFR —» AAAR 86 3 0 14 29

FRET —-» FAAA 64 5 5 26 22

TRHP —v TAAA 72 14 7 7 29

RHPI -+ AAAI 88 8 4 0 24

 

 

 

 

 

 

 

 

a,b,c,d Secondary structure predictions obtained by using tetrapeptide motifs covering the

whole PKKPAGR region. Those motifs represent either the exact sequence (exact

matches box, b), conservative substitutions based on the consensus motif described for

the PKKPAGR (Jaglo et al., 2001) (close matches box, c), or alanine substitutions
described in this study (triple alanine substitutions box, d).

e H, E, T, and 1 indicate helical, extended, turn and irregular structures, respectively.
f Number of matches retrieved by Sequery. Each tetrapeptide motif was used to search a
non-redundant database of low-homology (<25% pair-wise sequence identity) 3D
structures from the Protein Data Bank (Wang and Dunbrack, 2003).
The area of the table highlighted in grey indicates tetrapeptides showing strong helical
propensity according to Sequery results.

65

 

Sequery predictions for the PKKPAGR region were generally consistent with the
secondary structure prediction results obtained with the methods described above and
illustrated in Table 2.1. In fact, in both cases the PKKPAG and RHP clusters were
predicted to be coiled or irregular in structure (compare very right column of Table 2.1
with the two left columns of Table 2.2). In addition, Sequery results revealed a strong
tendency for a helical structure in the RKKFRET region: approximately 40-50% of the
close matches occurred in helices. The trends were even stronger when the exact matches
were used as input, with values ranging from 50 to 100%. To assess possible structural
perturbations by replacement with three consecutive alanines in the RKKFRET region,
Sequery analysis for these mutations was also conducted (lower section of Table 2.2, see
prediction for helical structure). Each of these substituted sequences was present in
helical conformation 64-88% of the time, which was expected based on the helical-
promoting effect of poly-alanines (Chou and Fasman, 1978; Zhang et al. , 1991;
Chakrabartty et al. , 1994). This result suggested that triple alanine substitutions within
the RKKFRET region caused little perturbation of the native structure.

In summary,secondary structure prediction studies suggested that two clusters in
the PKKPAGR motif, represented by PKKPAG and RHP, are likely to be irregular in
nature. Based on the helix-inducing propensity of poly-alanines (Chou and Fasman,
1978; O'Neil and DeGrado, 1990; Chakrabartty et a1. , 1994), it is likely that alanine
substitutions in these regions will affect the native structure. Therefore, it is not possible
to discriminate whether residues in this two regions are essential for protein folding or are

directly involved in interaction with the DNA, based on these substitutions. On the

66

contrary, the central RKKFRET region - rich in basic residues - showed strong helical
preference according to the predictor Sequery, suggesting that triple alanine mutations in
this region do not affect protein folding. Based on these observations and the notion that
patches of positively charged groups are energetically more stable at the protein surface
(Kim et al., 2005), we propose loss of binding due to mutations in the RKKFRET region
of CBF1 are likely due to a loss of side chain-DNA interactions required to bind the

CRT/DRE element.

Specific amino acids within the PKKPAGR motif have important roles in binding to
the CRT/DRE promoter element.

Electrophoretic-mobility shift assays and secondary structure predictions
suggested that residues within the RKKFRET motif might participate to DNA binding by
contributing important side chain interactions with the target CR T /DRE promoter
element. To test the contribution of individual residues in the predicted helical motif,
RKKFRET, we tested the role of specific amino acids by mutational analysis. The main
focus of our investigation was to assess the relevance of the positively charged side
chains in the DNA recognition by CBF1, since they are usually enriched in DNA-binding
domains (Nadassy et al. , 1999; Luscombe and Thornton, 2002). These residues could
favor the formation of a stable protein-DNA complex through electrostatic interactions
with the negatively charged phosphates of the DNA backbone, or participate in direct
recognition of DNA bases. In addition, we were interested in testing the role of the
conserved phenylalanine. This residue is not generally enriched at the protein-DNA

interface; however, when it is present, it can form hydrophobic contacts in complexes

67

with bent DNA, like in the case of the TATA-binding proteins [see (Kim et a1. , 1993) for
reference].

Mutations of the residues mentioned above were designed either to preserve the
side chain chemistry (conservative mutations) or to alter it (non-conservative mutations)
(Figure 2.13A, upper panel). In all cases but one, one additional constraint was applied:
the mutations should be compatible with the predicted helical structure of the RKKFRET
region, to exclude local protein unfolding in case reduced binding activity was observed
for those mutant proteins. The Phe4—>Pro mutation was designed to test the effect of a
helix-breaking residue on the stability of the protein-DNA complex. Based on these
criteria, a reciprocal ArgHLys substitution and Phe—*Tyr were chosen as conservative
mutations; Arg—>Ser, Lys—*Ala, and Phe—>Ala represented non-conservative
substitutions. Preserving the positive charge, in the case of Arg and Lys, and the aromatic
ring in the case of Phe—>Tyr, may result in a negligible effect on DNA binding; on the
contrary, the non-conservative mutations would directly test the importance of the
positive charge, in the case of Arg and Lys, or the contribution of the aromatic ring, in the
case of Phe. Recombinant proteins carrying point mutations were tested by gel mobility
shift assays in the presence of CRT/DRE promoter elements from COR15a, COR6. 6, and

COR78 (Figure 2.13A, B, C).

68

   

Conservative
Non-Conservative

A121 LysZ Lys3 Phe4 Arg5

 

 

WT Lys Ser Arg Ala Arg Ala Ala Pro Tyr Ser WT

_AAAAAAA —4l AAAé

 

 

 

 

 

 

Figure 2.13 A. Effect of point mutations in the predicted RKKFRET helical region of
CBF1 on binding activity of the protein to the COR 78 gene promoter.

Top panel. Substitution table indicating the point mutations designed within the
RKKFRET region of CBF1 . Conservative and non-conservative mutations are indicated.
H and C below the PKKPAGR motif indicate the predicted helical and coiled structures,
respectively.

Bottom panel. Gel mobility shift assays indicating the binding activity of different
MBP: CBF127-112 proteins. DNA binding reactions were carried out in a 15 ul reaction
containing increasing amounts (300 and 600 ng) of each recombinant protein in the
presence of 0. 5 ng of radiolabeled CRT/DRE-containing probe fi'om COR 78 promoter.

69

Arg] LysZ Lys3 Phe4 Arg5 Phe4

 

 

 

MBP WT Lys Ala Arg Ala Ala Pro Ser Tyr

AZEAAA _AAAA21

 

 

 

 

 

 

Figure 2.13 B. Effect of point mutations in the predicted RKKFRET helical region of
CBF1 on binding activity of the protein to the COR15a gene promoter.

Gel mobility shift assays indicating the binding activity of different MBP:CBF127-112
proteins. DNA binding reactions were carried out in a 15 ul reaction containing
increasing amounts (300 and 600 ng) of each recombinant protein in the presence of 0.5
ng of radiolabeled CRT/DRE—containing probe from COR15a promoter.

70

Argl Lysz Lys3 P_he4 A___rg_5

-2222222 _222

 

 

w

Argl Lys2 Lys3
WT Lys Ser Arg Ala Arg Ala

 

 

 

   

 

._u. ’1 .fi'. n--uw§<MBP:CBFl
Phe4 ArgS

Ala Pro Tyr Ser WT
MBP M/l

MBP > tw". ”:‘3. " H4 MBP:CBF1

 

Figure 2.13 C. Effect of point mutations in the predicted RKKFRET helical region of
CBF1 on binding activity of the protein to the COR6. 6 gene promoter.

Top panel. Gel mobility shift assays indicating the binding activity of different

MBP: CBF127- 112 proteins. DNA binding reactions were carried out in a 15 ul reaction
containing increasing amounts (300 and 600 ng) of each recombinant protein in the
presence of 0. 5 ng of radiolabeled CRT/DRE-containing probe from COR 78 promoter.
Bottom panel. Western blot analysis (or-MBP antibody) showing protein loading.
Twenty-five and 50 ng of each recombinant protein were analyzed.

71

The binding results observed at the three different promoters showed that most of
the residues tested play an important role in DNA binding by MBP:CBF127412. The most
striking observation was that the two mutations in Argl (Arg—*Lys and Arg—>Ser), and
two of the three mutations on Phe4 (Phe—>Ala and Phe—>Pro) caused almost complete
loss of DNA binding to the three promoters tested. The loss of binding by substitution of
Argl with Lys, of similar side-chain length and charge, was significant and suggested
that the interaction of this residue with DNA is possibly occurring through the DNA base.
Loss of binding for the Phe4—>Ala substitution suggested that the aromatic ring in this
residue is essential for the interaction with the DNA. Pro is a less conservative mutation
that is likely to interrupt helicity, and therefore loss of DNA binding was expected. The
conservative Phe—>Tyr mutation did not cause a significant change in binding activity; in
fact, this mutated protein displayed an increase in DNA binding affinity compared to the
wild type protein. This result suggests that the aromatic ring contributes favorable
interactions with DNA bases; in addition, the terminal hydroxyl group, with its H-
bonding donating accepting capacity, could add H-bonds between the side chain and
either the DNA bases of with part of the backbone. Overall, substitutions at LysZ and
LysB showed a moderate effect on DNA binding, and less effect when the mutant was
electrostatically conservative (Lys—>Arg) than when the side chain was truncated
(Lys—>A1a). These data suggested that the contribution of those residues to DNA binding
is less important, and they could likely be involved in salt bridging to the phosphate
backbone of the DNA. Finally, Arg5—>Ser substitution showed no significant effect on

DNA binding, indicating that this residue is not important for this function.

72

Interestingly, we could observe some differences in the binding affinity of the
mutant proteins to different promoters. For instance, Lys—>Arg mutation in LysZ affected
binding to the CRT/DRE from the COR 78 promoter more strongly than the COR15a
promoter (compare Figures 2.13A and B), and mutant CBF1 proteins carrying the
Lys—+Ala mutation displayed different binding affinities to the three promoters. It must
be noted, however, that binding to the COR 78 and the COR6. 6 promoters was tested only
one time. Additional technical replicates will elucidate whether these results are
reproducible.

Taken altogether, these data indicate that specific amino acids within the
PKKPAGR motif play a significant role in CBF1 binding to its target CR T/DRE promoter
element. Based on the moderate loss of DNA binding for substitutions at LysZ and Ly53,
it is likely that these residues are participating in electrostatic interactions at the protein-
DNA interface through their positively charged side chains. The more dramatic effects
observed for substitutions at Argl and Phe4 suggest that these amino acids might provide

side chain interactions that are base-specific.

Side chain conservation in AP2/EREBP-like proteins that use their flanking
sequences to stabilize DNA binding.

An important role of amino acid sequences flanking DNA-binding domains in
DNA recognition has been previously reported for a few families of proteins, such as the
homeodomain and high-mobility group families (Lnenicek-Allen et al., 1996; Wolberger,
1996; Dragan et al., 2004). The importance of specific amino acids in the PKKPAGR

motif in DNA binding, made us wonder whether this phenomenon is conserved in DNA-

73

binding proteins that, like the AP2/EREBP proteins, use their B-sheets to interact with
their target DNA. We addressed this question by searching the Protein Data Bank for
structures similar to the three [i-stranded domain described for AtERFl (Allen et al. ,
1998). Any consensus in this region would further support the idea that these residues sit
close to the DNA, and participate directly to the binding activity of the protein, possibly
by a common mechanism.

The atomic coordinates for the NMR structure of the AP2/EREBP domain of
AtERFl [PDB id. lgcc (Allen et al., 1998)] were used as a query to search the
comprehensive Protein Data Bank (http://rcsb.org), which includes 3D structures
obtained from crystallography and NMR studies. The method of choice was Dali (Holm
and Sander, 1993), a well-established website for protein structure comparison. The
server returned a list of 26 hits that shared significant structural similarity (Z-score >2.0)
with l gcc, including ERF] itself (Table 2.3). Six of the other 25 structures did not
contain any structural information on the sequences flanking the B-sheet at its N-
terminus, and were discarded (data not shown). Of the remaining hits, the analysis was

focused on nucleic acid-binding proteins, represented by ten structures.

74

Table 2.3. Summary of Dali analysis showing significant structural matches of nucleic
acid binding proteins with the AP2/EREBP DNA binding domain of AtERFl.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

PDB Z- % Nucleic acid-binding proteins Nucleic Acid Flanking
id. score, Idec containing a fl-sheet bound sequence,
Manure r

2gcc 11.8 100 AtERFI fragment 7 N/A
2bn8 4.2 16 Cell division activator (E. colt) - 9L
lkjk 3.9 10 Viral protein — intggrase \/ 7L
lzlb 3.8 9 DNA integrase mutant ‘J 2L
2bb8 3.4 10 Tn9l6 integrase 7 9L
2d35 2.8 13 Cell division activator (E. colz') \/ 7L
lrth 2.8 15 HIV—1 reverse transcriptase \/ 16L

lmkm 2.8 8 T. maritime 0065, IclR family - 2L12H
lbhi 2.5 8 DNA-bindinflotein cre-bpl - N/A
1di2 2.3 9 X laevis dsRNA binding protein A J 1L12H3L
lytb 2.2 6 S. cerevisiae TATA-box binding protein J H/L

 

2gcc (Allen et al., 1998); 2bn8 (Chen et al., 2005); 1kjk(Wojciak et al., 2002); lzlb
(Biswas et al., 2005); 2bb8 (Connolly et al., 1998); 2d35 (unpublished;
http://www.rcsb.org/pdb/explore/explore.do?structureId=2D35); 1rth(Ren et al. , 1995);
lmkm (Zhang et al., 2002); lbhi (Nagadoi et al., 1999); ldi2 (Ryter and Schultz, 1998);
lytb (Kim et al., 1993).

a Protein Data Bank identifier.

b Degree of significant structural similarity (Z-score >2.0), according to Dali Ca
matching.

c percentage of sequence identity over positions of structural homology.

d check marks under “Nucleic Acid bound in structure” indicate that the 3D structure of
the protein is in complex with either DNA or RNA.

e secondary structure of sequences N-terminal to their B-sheets, where available.
H=helix; L=loop; N/A=structure of this region of the protein is not present or not defined
in the crystal or NMR structure.

75

The similarity between three of them, represented by PDB identifiers lkjk and
lzlb (representing the same protein), and 2bb8 and the AP2/EREBP domain of ERF]
had previously been reported (Connolly et al., 1998; Wojciak et al. , 2002; Biswas et al.,
2005; Chen et al. , 2005). An extensive comparative analysis of the DNA-bound structure
of ERF 1 and Tn916 integrase (Wojciak et al., 1999) has previously indicated significant
similarities in the mode of DNA recognition by these B-sheet containing proteins
(Connolly et al. , 2000). This recognition involves a similar orientation of the B-sheet
relative to major groove of the target DNA, and conserved protein-DNA contacts within
the AP2/EREBP-like structure (Figure 2.14).

We wanted to extend our analysis to the N-flanking regions, in search of protein-
DNA complexes in which both the B-sheet and the N-flanking motif were involved in
DNA-binding. Among the AP2/EREBP-like structures described in Table 2.3, lambda
integrase and Tn916 integrase displayed this binding mode. NMR studies showed that the
secondary structure of the N-flanking regions of these proteins differs from the helical
structure predicted for the PKKPAGR region RKKFRET. This N-terminal region is
represented by a loop in Tn916 integrase and by an unstructured segment in lambda
integrase. However, despite the structural differences, some compelling similarities were

found when we analyzed the amino acid composition of these motifs (Figure 2.15A, B).

76

 

 

 

 

 

 

 

Connolly et al., 2000

Figure 2.14.Three-dimensional structures of proteins that use three-stranded B-sheets to
recognize the major groove of DNA.

Left. The three-stranded B-sheet complexes of the DNA binding domain fi'om Tn916
integrase (left) and AtERFl (right). The Figure emphasizes the conserved orientation of
the sheet structures relative to the DNA. The conserved B-sheets and the DNA backbone
are dark and light grey, respectively.

Right. DNA-binding domain of Tn916 in complex with DNA. labeled residues were
mutated and the DNA binding activity of the mutated protein was tested by
electrophoretic mobility shift assay. ArgS, in the N-flanking region of Tn916, contributes
to DNA binding aflinity. DNA and protein backbones are shown as tubes, and the three-
stranded B-sheet is represented by ribbons. N and C, N and C termini of the DNA-
binding domain including the N-flanking region

77

 

A
*
Tn916-DBD

B

+++ +- +
CBF1 (lgcc) PKKPAGRKKFRETRHPI - - - -YRGVRQR
K-Int (lkjk) MGR_RRSHERRDLPP---NYRNNGY
Tn916 (1tn9) MBKBR--DNRGRILKTGESQRKDG

' : : *

Figure 2.15. Comparison of B-sheet—flanking sequences experimentally shown to be
important in DNA binding.

A. Schematic of the NMR derived structure of the Integrases Tn916 and Lambda (k-DBD
and Tn9l6 DBD, respectively) and ERF] (GBD) as described by Wojciak et al. (1999).
The NMR structures of ERFl and Tn916 were solved in complex with the DNA. The
antiparallel B-sheet is blue, the C-terminal helix is red. Asterisks N-flanking motifs.

B. Alignment of N-flanking regions from Lambda and Tn916 integrases and ERF].
Alignment is based on optimizing charged residue correspondence. Underlined residues
are known to be important for binding. (*), exact conservation of residues; (z), strong
conservation; (.), some degree of conservation in the residue chemistry. (+) and (—),
positively and negatively charged residues, respectively. Area highlighted in grey,
residues at the 5’ end of the beta sheets. lgcc (Allen et al., 1998); lkjk (Wojciak et al.,
2002); 1b69 (Connolly et al., 2000).

78

Not only are the N-terminal residues proximal to or directly contacting the minor
groove of the DNA in the NMR structure (Figure 2.15A), mutational analysis of residues
in the N-flanking regions of these two proteins indicated that one or more positively
charged arginines play important roles in DNA binding. For instance, mutation of an
arginine residue in the N—flanking sequence of Tn9l6 integrase causes a significant loss
in DNA-binding affinity (Connolly et al. , 2000), whereas deletion of two arginines in the
[ES-sheet flanking region of lambda integrase could abrogate DNA-binding (Wojciak et al. ,
2002). When we analyzed the sequence composition of those motifs we observed that the
central core of those N-flanking regions includes a cluster of positively charged residues,
and one negatively charged residue.

In summary, we could identify a subgroup of B-sheet DNA-binding proteins in
which one or more residues present in the N-flanking region provide a significant
contribution to DNA-binding affinity. The presence of a positively charged cluster, their
proximity to the DNA and their contribution to DNA binding, is consistent with our
findings and supports the idea that the PKKPAGR motif provides a similar contribution

at the CR T /DRE promoter site.

79

A working model for DNA binding by CBF1.

Extensive mutagenesis and structural analysis of the PKKPAGR motif and
comparison to AP2/EREBP-like proteins that use their N-flanking regions to contact the
DNA, prompted us to describe CBF1 interaction with the DNA by computational
modeling. The experimental results obtained by point mutations within the RKKFRET
predicted helix drove the model design. The constraints used in modeling the orientation
of the helix relative to CBF and the DNA were the following: (1) The C-terminus of the
RKKFRET helix had to lie within two residues of the N-terminus of the B-stranded
region of CBF; (2) Argl and Phe4 residues, critical to DNA binding, were placed in close
proximity to the DNA, given that mutations to these positions have drastic effects on
DNA binding. (3) The helix should pack well against the DNA and CBF1. The other
features described below were derived from the placement of the RKKFRET helix to
meet these three constraints. The NMR structure of DNA bound to the CBF1 homolog
ERFl (PDB accession 1 gcc), was used as template to model the interaction between the
DNA minor groove and the RKKFRET N-terminal flanking helix (Figure 2.16A). Figure

2163 illustrates the working model of CBF1 bound to the DNA.

80

 

Figure 2.16A. NMR structure of the AP2/EREBP DNA-binding domain of ERF] bound
to its cognate DNA (Allen et al., 1998).

Lefi and right figures represent two different views of the DNA binding domain in
complex with its cognate promoter element. The three-stranded beta sheet is represented
by ribbons numbered 1-3 that contact the DNA by reaching into its major groove. N and
C termini of the AP2/EREBP domain are indicated by N and C, respectively. 5’ and 3’
end of the double-stranded DNA are indicated.

81

 

(Leslie Kuhn)

Figure 2.16B. Working model describing DNA-binding by CBF1. The NTVIR structure of
DNA bound to the CBF homolog ERF] (PDB id. lgcc), was used as the basis to model
the interaction between the minor groove of DNA and the RKKFRET N-terminal
flanking helix (yellow ribbon and side chains) relative to the main chain of the Iii-stranded
DNA-binding domain of CBF1 (orange ribbon). The DNA is shown in tubes, with
carbons in white, nitrogens in blue, oxygens in red, and phosphorus in magenta. In the
RKKFRET helix, Arg 1 and Phe 4 residues bind within the DNA minor groove, where
Argl forms hydrogen bonds and Phe4 in involved in stacking interactions. The main-
chain structure of the B-strand domain of CBF was modeled with SwissModel. Two
additional residues, Arg and His, bridge between the C-terminus of the KKFRET and the.
N-terminus of the beta-stranded DNA-binding domain, as indicated by the orange

line. This connection likely involves a slight reorientation of the first four residues of the
beta-stranded region (orange ribbon), as observed in the structures of two DNA
integrases (PDB id. lb69 and lzlb), which also use a B-stranded domain to bind DNA,
with an N-terminal flanking region that enters the minor groove.

82

The AP2/EREBP domain and the RKKFRET region (orange and yellow,
respectively) are placed against the DNA major and minor groove, respectively. The
AP2/EREBP domain (shown in orange vertically oriented), is placed into the major
groove, and its interaction with DNA forms a smooth pocket where the RKKFRET helix
is packed. The proximity of this region to the DNA is also derived by the physical
constraint of a short two-residue connection to the three-stranded B-sheet. Mutational
analysis indicated that Argl and Phe4 are critically important for DNA binding.
Argl——>Lys abrogated DNA binding, despite conserving side-chain length and charge.
For the Phe mutations, Ala is helix-propensive and hydrophobic but less bulky than Phe,
whereas Pro is a less conservative mutation and is likely to interrupt helicity. The loss of
DNA binding upon these single-site mutations suggests that these residues interact
intimately with the DNA, and both are positioned to bind within the minor groove.
Phe—*Tyr mutation didn’t affect DNA binding. This is consistent with the phenyl moiety
of Tyr maintaining favorable aromatic interactions with adjacent DNA bases, while its
terminal hydroxyl group can both donate and accept hydrogen bonds either with bases in
the groove of the DNA backbone. The phenyl moiety of Phe4 (or Tyr4) can make
aromatic ring stacking interactions with adjacent bases in the groove, which have been
reported when Phe interactions with DNA occur (Luscombe and Thornton, 2002;
Moravek et al. , 2002) and this is likely to distort the local DNA conformation (Moravek
et al. , 2002). Substitutions in the other residues of the RKKFRET motif resulted in a
more moderate effect on DNA binding. Accordingly, Lys2 is shown salt bridging to the
phosphate backbone and Lys3 is hydrogen bonding to a backbone ribose residue,

consistent with mutations to these residues having moderate (non-base-specific) effects

83

on DNA binding, and less effect when the mutant is electrostatically conservative
(Lys—>Arg) than when the side chain is truncated (Lys-+A1a). Arg5 and Glu6 are
positioned such that they can form a salt bridge that stabilizes the helix, on the opposite
side of the DNA. This is also consistent with Arg5—>Ser mutation having little effect on
DNA binding, and with the Ser mutation also being able to form a stabilizing intra-helix
hydrogen bond with Glu6.

Based on the experimental constraints imposed on the model, the helical
RKKFRET region can be placed close to the 3’end of the CRT/DRE motif -
(G/A)CCGACNT. According to the model, Argl can make hydrogen-bond and pi-cation
interactions with bases in the minor groove, corresponding to the region in which bases
B11-B13 (from chain B) pair with C14-C16 (from chain C) in the 1 gcc structure (Figure
2.17). This Bl 1-B13/Cl4-C16 sequence follows the conserved consensus GCCGCC,
representing the core sequence bound by ERF] via its B-sheet. Interestingly, the
corresponding region from the CRT/DRE promoter element contains a conserved thymine
that has been shown to be required for specific binding of CBFs to CR T/DRE-containing
promoter elements (Maruyama et al. , 2004; Sakuma et al. , 2006). The proximity of the
RKKFRET region to this DNA base suggests that either Argl or Phe4 could direct

specific recognition of the DNA by CBF proteins.

84

CBF cognate DNA (CORISA) 5'- T G G C C G A C C T G - 3'
Residue # in Chain B of lgcc 3 4 5 6 7 8 9 10 ll 12 13

ERF DNA in PDB lgcc 5'- T A G C C G C C A G C 3'

ERF DNA in PDB 1 gcc 3'- A T C G G C G G T C G 5'
Residue # in Chain C of lgcc 24 23 22 21 20 19 18 l7 16 15 I4

CBF cognate DNA (CORISA) 3'- A C C G G C T G G A C 5'

Figure 2.17. Nucleotide sequences of the cis-acting elements in CBF and ERF proteins.
The sequences shown for ERF DNA refer to the coding and complementary strand of
DNA used in the NMR studies by Allen et al. (1998), in which the structure of the DNA-
binding domain was solved in complex with the DNA. The conserved bases are shown in
bold.

85

In summary, by integrating secondary structure predictions and experimental
analysis, it was possible to describe CBF1 interaction with the DNA by computational
modeling. We propose that the interactions of N-terminal flanking residues with the
minor groove involve bases beyond those bound by the B-sheet, some of which might

confer specificity for COR gene promoters relative to ERE.

86

FUNCTIONAL ROLE OF THE DSAWR MOTIF IN CBF1 ACTIVITY.

Mutations in the DSAWR sequence affect COR gene activation.

To determine whether the DSAWR motif plays a role in CBF1 function,
Arabidopsis plants overexpressing dsawr mutant or wild type CBF1 transgenes were
generated. Morphological analysis of those plants provided a first indication that residues
within the conserved DSAWR motif might play a role in CBF1 function (Figure 2.3).

The role of the DSAWR motif in CBF1 activity was assessed by northern blot
analysis as described for the PKKPAGR motif. Representative results are shown in

Figure 2.18.

87

CBF1

COR78
COR1 5a

COR47

COR6.6
18$

3

dsawr

 

86
621
G26

 

 

 

 

 

 

 

Figure 2.18. Accumulation of CBF and COR transcripts in Arabidopsis plants
overexpressing wild type or dsawr CBF1.

Total RNA was isolated from warm—grown seedlings (22°C), and 5 ug were analyzed by
Northern blotting. Northern blot analyses were performed using probes for CBF1,

COR6. 6, CORISA, COR47, and COR78 transcripts.

B6, transgenic lines harboring the empty vector; WT, transgenic lines overexpressing
wild type CBF1; dsawr, transgenic lines overexpressing CBF1 transgene carrying alanine
substitutions in the DSAWR region. Numbers above lanes indicate independent

transgenic lines.

88

Similarly to what was observed for the pkkpagr lines (Figures 2.4 and 2.5),
Arabidopsis plants overexpressing CBF1-dsawr transgene still retained some ability to
activate transcription at COR gene promoters, compared to the control plants (B6)
harboring the empty vector only (Figure 2.18, compare B6 to independent dsawr lines).
However, COR transcript accumulation was reduced when dsawr lines were compared to
Arabidopsis lines overexpressing wild type CBF1 transgene Northern blot analysis
showed that higher transcript levels of the mutated CBF1 were needed to achieve COR
transcript accumulation similar to the wild type CBF1 overexpressing lines (Figure 2.18,
compare WT and dsawr lines).

Altogether these data indicated that alanine substitutions throughout the DSAWR

motif of CBF1 affected its transcriptional activity.

CBF1 protein detection in dsawr lines.

To determine whether the dsawr mutation affected CBF1 protein accumulation, it
was necessary to measure CBF1 protein levels in planta. The experiment was carried out
by generating transgenic plants overexpressing wild type and mutated CBF1 transgenes
fused to a Myc tag. Protein levels were measured by western blot analysis, as described

for the pkkpagr lines. CBF1 gene and protein expression levels are shown in Figure 2.19.

89

CBF1 WT CBF1-dsawr

 

 

o

l

O mooo—cmxo—rtx v-nrxooamxohoo

ONxor—v—NNNNmm WOv-tv-tv-‘v-‘F-‘NNNN
CBF1 'm‘ -'- “were . -4» ..
COR6.6 ~...u..r_..u g... .

188 0.000....” W.

CBF1 WT CBF1-dsawr

 

c?
'6‘ w v—i In l‘ V) \0 w
0 so .— N N .— N N N
a-Myc “h- .—
Ponceau an .: ”*Wfimtmmtzflwp~
Red ‘5"

Figure 2.19. Northern and western blot analysis of dsawr plants and plants
overexpressing wild type CBF1.

Top. Northern blot analysis of 353::6xMyczCBF1 transgenic lines. Total RNA was
analyzed for CBF1 and COR6. 6. Transcript amounts were normalized using the 183
loading control.

Bottom. Western blot analysis showing the presence of 6xMyczCBFl protein in the
transgenic plants (detected with a monoclonal anti-Myc antibody).

WT and dsawr, transgenic plants overexpressing a 6xMyc tag fused to a wild type (WT)
or mutated (dsawr) CBF1 transgene; Col-0, non-transgenic plants.

90

Several independent lines were analyzed for the expression of CBF1 and C 0R6. 6.
Representative lines overexpressing similar levels of My-CBFI were chosen for protein
detection from plant extracts. Western blot analysis clearly indicated that dsawr plants.
were affected in CBF1 accumulation when compared to Arabidopsis plants
overexpressing wild type CBF1 transgene at similar level (Figure 2.19, compare).

Taken together, these results indicate that the inability of the dsawr lines to induce

COR gene expression can be explained, at least in part, by reduced protein accumulation.

Role of DSAWR in DNA binding activity of CBF1.

Based on its proximity to the AP2/EREBP DNA-binding domain, and by analogy
to the PKKPAGR sequence, we postulated that the DSAWR sequence might play a role
in CR T/DRE binding by CBF1. Gel mobility shift assays were conducted to compare the
binding activity of the wild type and mutated protein.

Experiments were conducted as previously described for the PKKPAGR mutants.
Binding reactions were set up by incubating wild type of mutated CBF1 proteins with
Y32P-radiolabeled probe representing CRT/DRE elements from COR6. 6, COR15a, and
COR 78 promoters. In all cases, alanine substitutions greatly reduced the binding activity

of CBF1 , indicating that the DSAWR motif contributes to recognition (Figure 2.20).

91

WT dsawr WT dsawr WT dsawr

mp4] A AAMBP 414] ”0““

1
vol-.52..

 

I. “ ~u w

 

 

 

 

 

 

Free
probe

      

CORISA COR78 COR6.6

Figure 2.20. DNA binding activity of an MBP-CBFlzulz protein carrying alanine
substitutions in the DSAWR region.

Upper panel. Gel shift mobility assay in which binding ofMBP:CBF127-1 12 proteins to
COR6. 6, COR15a, and COR 78 is shown. 0.5 and 1.0 ug of protein were used in a 15.0 111
reaction in the presence of 0.5 ng of radiolabeled CRT/DRE-containing probes from
COR6. 6, COR15a, and COR78 promoters.

Lower panel. Western blot analysis showing protein loading. 25 and 50 ng of each
recombinant protein were analyzed.

Point mutations within the DSAWR region of CBF2 reveal the importance of an
Asp residue in CRT/DRE recognition.

To better understand the contribution of the DSAWR region to DNA binding by
the CBF proteins, analyzing the effect of point mutations can be informative. In a mutant
suppressor screen designed to identify regulators of the CBF pathway that act through the
CBF regulon, several mutations in the CBF 2 transgene were identified (Gilmour,
unpublished). Two mutations fell in the DSAWR region of CBF 2, Asp—>Asn and
Arg—>Trp. Overexpression of those mutated transgenes in Arabidopsis produced plants
that showed reduced accumulation of COR transcript compared to the plants

overexpressing the wild type gene (Figure 2.21).

93

WT D/N WT R/W

CBF1 MM '- --."" -rm..- 0!.

 

 

 

 

COR15a an-” ' * a... .. ..

18s err-u quid i.e.-ea; mags-99

“I"

 

 

 

 

 

 

 

 

 

 

Sarah Gilmour, unpublished

Figure 2.21. Northern blot analysis of transgenic plants overexpressing wild type and
mutated CBF2 transgenes.

Total RNA was isolated from two-week old warm-grown (22°C) seedlings. Northern blot
analysis was performed using probes for CBF1, COR15a, and 188 ribosomal RNA.

WT, RNA samples from independent lines overexpressing wild type CBF1; DIN and
WW, transgenic plants overexpressing a mutated CBF2 transgene harboring an Asp to
Asn or Arg to Trp mutation, respectively.

94

Transcriptome profiling by Gilmour and colleagues (2004) revealed that CBF1,
CBF 2, and CBF 3 display overlapping functions when overexpressed in Arabidopsis. In
addition, the DNA—binding domains of CBF1 and CBF2 only differ in two residues,
neither of which is involved in direct contact with DNA based on the AP2/EREBP
structure described for AtERFl, so presumably the two DNA-binding domains can be
considered equivalent. To determine the effect of Asp—>Asn and Arg—+Trp mutations on
CBF 2 binding affinity, MBP:CBF227-112 fusions were subcloned in E. coli and the
purified protein extracts tested in a gel mobility shift assay. As observed for CBF1, CBF2
could specifically bind a CRT /DRE-containing probe (Figure 2.22). The Asp—>Asn
mutation, conservative in structure but with no charge, resulted in a significant loss of
DNA binding, but retained its specific sequence recognition, as shown by the competition
assay. Surprisingly, the Arg—>Trp substitution, despite the significant change in side

chain structure, did not alter CBF2 binding affinity.

95

WT Asn/Arg Arg/Trp

 

 

 

Protein
—MBPA A A
————WTM——WTM——WTM C°"'P°“‘°"
DNA
-- In .- O

 

 

 

 

WT Asp/Asn Arg/Trp
MBP A /1 ._/_i

MBP» - Q Q ‘4 MBP:CBF1

Figure 2.22. DNA binding activity of MBP:CBF 227-1 12 proteins canying point mutations
in the DSAWR motif.

Upper panel. Gel shift assay showing different DNA binding activity. Increasing amounts
(150 and 450 ng) of wild type and mutated CBF2 proteins were tested in a 15 ul binding
reaction containing 0.5 ng of radiolabeled CRT/DRE-containing probe from COR1 5a
promoter. WT and M indicate a wild type and mutated unlabeled DNA (100 ng each),
respectively used for the competition experiments indicated.

Lower panel. Western blotting showing protein loading (15 and 45 ng of each protein
were analyzed). a-MBP was used for protein detection.

96

Taken together, these observations suggest that the reduced ability to induce COR
gene expression observed for the dsawr plants could be explained, in part, by inefficient
binding of CBF1 to CR T/DRE-containing promoters. Based on the effect on protein
accunulation observed when multiple alanines are introduced into the DSAWR motif,
reduced ability to induce COR gene expression also appears to be due to reduced

accumulation of the mutant protein.

97

DISCUSSION and FUTURE DIRECTIONS

The main goal of this study was to determine whether two conserved signature
sequences in the CBF family of proteins are required for CBF l transcriptional activity,
and, if so, to elucidate their role. A mutational approach was taken to generate transgenic
plants overexpressing wild type and mutated versions of the two motifs.

Induction of several members of the CBF regulon - COR6. 6, COR15A, and
COR 78 - was tested by northern blot analysis to evaluate the relevance of the signature
sequences on CBF1 activity. In all cases tested, reduced COR/CBF1 transcript ratios in
the pkkpagr lines compared to the transgenic plants overexpressing wild type CBF1
indicated that the PKKPAGR motif is required for CBF1 activity. Interestingly, we
noticed a difference on the effect of these mutations at different COR gene promoters.
For instance, COR15a induction was dramatically affected by all mutations tested
compared to COR6. 6 and COR 78 induction. One possible explanation is that PKKPAGR
confers higher binding affinity to different COR promoters, COR1 5a in this case. Once
this motif has been mutated, the DNA binding function is lost or reduced and the effect
will be more pronounced at the most highly induced genes. Alternatively, induction of
COR15a-like genes might represent a sub-group of the CBF regulon that requires a
CBF1-interacting factor that acts through the PKKPAGR motif. When the PKKPAGR
motif is mutated, the interaction is lost and CBF 1 ability to recruit the transcriptional
machinery will be greatly impaired. Yeast-two-hybrid screen by using the PKKPAGR

motif as bait could help identify such factor. Overexpression of CBF1 in T-DNA knock-

98

out lines for that gene might show that a subset of COR15a-like genes is affected,
similarly to what has been observed in the pkkpagr lines.

Several hypotheses were postulated to assign a functional role to the PKKPAGR
motif. Western blot analysis of transgenic plants overexpressing 6xMyc: CBF1 constructs
ruled out the possibility that the PKKPAGR motif lowers CBF1 protein steady-state
levels. On the contrary, substitution of the KKF residues with alanines resulted in higher
CBF1 protein accumulation (Figure 2.5, compare CBF1 transcript and protein levels in
WT and M3). Transcription factors tend to be very unstable proteins and it has been
extensively reported that their activity and their turn-over can be tightly correlated
(Tansey, 2001). Their degradation is usually mediated by the proteolytic activity of the
proteasome, which recognizes a poly-ubiquitin chain attached to a lysine residue of the
target protein. Increased CBF1 protein levels and reduced transcriptional activity when
KKF residues are converted to alanines are consistent with a role of the ubiquitin
pathway in modulating CBF1 activity. The obvious question is whether the difference in
protein levels in wild type and mutated CBF1 proteins can be attributed to this pathway.
To address this question it will be necessary to determine whether CBF1 is ubiquitylated
in vivo, and whether the protein can accumulate to higher levels in the presence of
specific proteasome inhibitors. If this is the case, the following step will be the
identification of the ubiquitin ligase that is responsible for the addition of the ubiquitin
chain that signals CBF1 degradation to the proteasome. Ultimately, the goal is to extend
this analysis to wild type Arabidopsis plants and study the effect of low temperature on

the regulation of CBF1 protein stability.

99

The PKKPAGR region, rich in basic residues, represents a potential NLS, and
resembles NLSs previously identified in other plant transcription factors, including the
member of the AP2/EREBP family AINTEGUMENTA (Krizek and Sulli, 2006). We
tested whether the PKKPAGR motif played a role in the transport of CBF1 across the
nuclear envelope by nuclear localization studies of Arabidopsis plants overexpressing
translational fusions of CBF1 to GFP:GUS. Laser confocal microscopy of Arabidopsis
plants overexpressing CBF1 APKK:GFPGUS revealed that this chimera is present in the
nucleus of root tip cells, indicating that the PKKPAGR motif is not required for nuclear
targeting of CBF1; on the contrary, localization studies of 5’ deletions of CBF1 fused to
GFP:GUS revealed that in the absence of its N-terminal domain (residues 1-104)

CBF1 :GFP:GUS loses its nuclear localization, thus mapping the nuclear targeting signal
to the N-terminal of the protein which includes the DNA-binding domain. These data
correlate with the fact that CBF1 1-1 15-VP1 6A1) can activate COR gene induction in stable
Arabidopsis lines, and therefore must contain a nuclear targeting motif (Wang et al. ,
2005). The involvement of the DNA-binding domain in nuclear targeting is not a new
finding; it has been calculated that for 79% of DNA-binding proteins in which both NLS
and DNA-binding domain have been characterized, the two functions map to the same
protein domain and sometime the very same residues (LaCasse and Lefebvre, 1995).
Some examples include the plant-specific SBP-domain proteins (Birkenbihl et al. , 2005),
the yeast transcription factor GAL4 (Silver et al. , 1984; Carey et al. , 1989; Birkenbihl et
al., 2005), and the transcription factors from the high-mobility group family (Sudbeck
and Scherer, 1997; Prieve et al., 1998). Due to lack of a consensus signal for nuclear

import, we could not identify any potential NLS in this region. However, the

100

AP2/EREBP domain contains several positively charged residues, mainly concentrated at
its N-terminus, next to the PKKPAGR motif. It is possible that both the PKKPAGR
region and this portion of the DNA-binding domain must be deleted to eliminate
transport across the nuclear membrane.

Based on its proximity to the DNA-binding domain, we hypothesized that the
PKKPAGR motif might be helping CBF1 to bind to the CRT /DRE element, and found
that triple alanine mutations within this region abrogated DNA-binding. Secondary
structure predictions suggested that several residues in the PKKPAGR motif are likely to
be coiled, while a seven-residue motif — RKKFRET - showed strong helical propensity.
Point mutations within the RKKFRET motif revealed that specific amino acids are
crucial for CBF1 DNA binding activity. Point mutations at either Argl or Phe4
(Arg—’Lys and Arg—>Ser and Phe—>Ala)’ caused the most dramatic loss in DNA-binding,
whereas point mutations at LysZ and Lys3 (Lys—rArg and Lys—>Ala) had a minor effect
and Arg5—*Ser substitution showed no significant effect. The CRT/DRE element is
specifically recognized both by CBF and DREBZA proteins (Liu et al., 1998). A
conserved thymine in the core sequence of this cis-acting element represents a specificity
determinant for CBF1-dependent DNA binding activity (Sakuma et al. , 2006). Based on
the NMR structure available for ERF 1 and the high homology among AP2/EREBP
proteins, it was proposed that residues within this domain are involved in the specific
binding to the CCGAC core of the CRT/DRE element (Allen et al., 1998 - verify);
however the conserved thymine is located downstream of these DNA bases and therefore
is unlikely to be contacted by residues within the AP2/EREBP domain. We integrated

these observations with experimental data, obtained from mutational analysis and

101

secondary structure predictions, in a computational model that illustrates how CBF1
interacts with the DNA major groove through its B-sheet and with the DNA minor groove
via the RKKFRET motif. In this model, the helical region of the PKKPAGR motif
reaches into the DNA minor groove with both Argl and Phe4 oriented in close proximity
to the conserved thymine. Arginines are frequently associated to base specific
recognition, when present at the protein-DNA interface; this interaction occurs mainly via
hydrogen bonding with a guanine, in fewer cases with thymine (Luscombe et al. , 2001;
Luscombe and Thornton, 2002). Phenylalanines are not frequently found; however, when
present they can contribute specific DNA recognition by providing stacking interactions
with adjacent bases in the minor groove, preferentially at a thymine-adenine pair
(Luscombe and Thornton, 2002; Moravek et al. , 2002). LysZ and Lys3, are placed in the
vicinity of the DNA phosphates to provide favorable electrostatic interactions, whereas
Arg5 does not participate to the protein-DNA complex. Overall, the experimental data
can be nicely fit into this working model and explain how the PKKPAGR motif provides
side chain interactions that are important or essential for CBF1 DNA binding activity.
One important question that remains to be answered is whether Argl and/or Phe4 indeed
represent specificity determinants and whether they are involved in direct recognition of
the conserved thymine found in the CRT/DRE promoter element. NMR studies of CBF1
in complex with its cognate DNA will help shed some light on the nature of those
interactions.

A role for N-flanking regions in DNA-binding affinity has been recognized in
homeodomain proteins (Wolberger, 1996). In this family, binding specificity is driven by

a-helices mainly via hydrophobic interactions in the major groove of the target DNA,

102

whereas the N-terrninal tail makes contact with the DNA minor groove (Wolberger,
1996). DNA-binding by B-sheets has been extensively characterized at the structural and
biochemical level in several DNA-binding proteins (Tateno et al. , 1998; Connolly et al. ,
1998). However, most of the research has focused on defining the conserved structural
features important for DNA binding within the B-sheet only. In contrast, very little is
known about the role of their N-flanking sequences. Analysis of N-flanking sequences
from other B-sheet-containing DNA-binding proteins revealed that these regions tend to
be enriched in basic residues (Figure 2.14). Moreover, one or more residues in these
regions are crucial for binding to the cognate DNA. These observations together with the
results obtained with our mutational analysis suggest that the formation of protein-DNA
complexes in the B-sheet family extends beyond the AP2/EREBP-like structure to
include residues from the N-flanking region that provide important side chain interactions
at the protein-DNA interface.

Along these lines, identification of the binding determinants in the PKKPAGR
motif could provide some insights into the mode of DNA binding in the AP2/EREBP
family. To date, most of the characterization has been limited to the binding activity of
the ERF and CBF proteins. It has been proposed that the specificity determinants in these
two subfamilies lie within the AP2/EREBP DNA-binding domain (Hao et al. , 1998;
Sakuma et al., 2002). However, high sequence similarity and the conservation of the
residues that make direct contact with the DNA in these proteins suggests that the
specificity switch may instead be found outside of this domain. Studies on several ERF
proteins indicated that 10 amino acid residues upstream of the AP2/EREBP domain were

essential for DNA binding (Hao et al. , 1998). Since proteins in the ERF group do not

103

share significant similarity in the region immediately upstream of the AP2/EREBP
domain, the authors concluded that these residues do not contribute to specific binding,
but instead their function is to stabilize the protein-DNA complex (Hao et al. , 1998).
Interestingly, ERF proteins contain a cluster of positively charged residues flanking the
AP2/EREBP domain at its N-terminus. It would be of interest to test the role of those
basic amino acid residues in DNA binding by the ERF proteins. A large-scale comparison
of the N-flanking sequences from members of this family combined with mutational
analysis might reveal the presence of similar amino acids in other proteins within the
AP2/EREBP family. In addition, sequence comparisons within different subgroups of the
AP2/EREBP family could indicate whether there are any conserved residues that are
specific for different subfamilies. If present, those residues would represent potential
specificity determinants.

Similarly to what had been observed for the pkkpagr lines, the growth retardation
in the dsawr plants was not as severe as in for wild type CBF1-overexpressing plants.
Consistently, COR gene induction was affected by this mutation. When overexpressed in
Arabidopsis, dsawr CBF1 proteins do not accumulate as well as wild type CBF1 proteins.
Inability to accumulate this protein in planta provides an explanation for reduced COR
gene expression in the dsawr lines compared to the wild type CBF1-overexpressing lines.
It is not clear at the moment whether this result is solely due to protein instability.
Alanine substitutions within the DSAWR motif impaired CBF1 DNA binding activity.
Analysis of two point mutation in the DSAWR motif - Asp—rAsn and Arg—>Trp -
suggest that at least one residue (Asp) might play an important role in DNA binding. In

fact the Asp—+Asn mutation, conservative in size though not in charge, may suggest that

104

no major change in protein folding occurred, and therefore a side chain interaction with
the partner DNA might have been lost. On the contrary, mutation of Arg—*Trp did not
show any effect on DNA-binding affinity. Interestingly, overexpression of a CBF 2
transgene carrying that mutation in Arabidopsis has a dampening effect on COR gene
expression as compared with overexpression of wild type CBF2 (Figure 2.23). Whether
protein stability is affected in these plants is currently unknown. An intriguing possibility
is that this residue is critical for protein-protein interaction that is required for

transcriptional activation at the COR gene promoters.

105

MATERIAL AND METHODS

Mutagenesis of the PKKPAGR and DSAWR motifs.
Alanine scanning mutagenesis.

Site-directed mutagenesis (Li and Wilkinson, 1997) was performed to convert the
original amino acids in the signature sequences of CBF1 to stretches of three alanines.
Primers were designed to introduce the desired mutations to the PKKPAGR motif of CBF1
protein using the QuikChange mutagenesis kit (Stratagene), and the protocol was carried
out according to the manufacturer instructions. A M restriction site was included in these
primers for the screening of plasmids containing the mutated ORFs. The template DNA
used was the full-length CBF1 ORF inserted into pBS/SK'.The primers used for the
PKKPAGR mutants are the following: Mutant] (PKK), fwd (MT640): 5’ gcc acg agt tgt
gcg gcc gca ccg gcg ggc cgt3’; rev. GVIT641) 5’acg gcc cgc cgg tgc ggc cgc aca act cgt
gch’; Mutant2 — (PAGR), fwd. (MT642): 5’cga gtt gtc cga aga aag egg ccg ccg cta aga agt
ttc gtg aga c3’; rev. (MT643): 5’ gtc tca cga aac ttc tta gcg gcg gcc gct ttc ttc gga caa ctt
g3’; Mutant3 -— (KKF), fwd. (MT644): 5’ccg gcg ggc cgt gcg gcc gct cgt gag act cgt3’; rev.
(MT645): 5’acg agt ctc acg agc ggc cgc acg gcc cgc cgg3’; Mutant4 (RET), fwd. (MT646):
S’gcg ggc cgt aag aag ttt gcg gcc gct cgt cac cca att tac ag3’; rev (MT647): 5’ctg taa att ggg
tga cga gcg gcc gca aac ttc tta cgg ccg cg3’; MutantS (RHP), fwd. (MT686): S’aag aag ttt
cgt gag act gcg gcc gca att tac aga gga gtt cgt3’; rev (MT687): 5’acg aac tcc tct gta aat tgc
ggc cgc agt ctc acg aaa ctt ctt3’. Primers used for DSAWR mutant: fwd. (MT817) 5’ gtc tca
act tcg ctg ccg cgg ccg cgg cgc tac gaa tcc cgg ag3’; rev. (MT818): 5’ctc cgg gat tcg tag

cgc cgc ggc cgc ggc agc gaa gtt gag ac3’.

106

The APKK mutant, that lacks the entire PKKPAGR region, was made using a modified
protocol based on the QuikChange method (Wang and Malcolm, 1999). To overcome the
tendency of the perfectly complementary mutagenic primers to anneal to each other rather
than to the target sequence, a two-stage PCR was performed, running two separate single-
primer reactions before the final PCR-amplification. Primers: fwd. (MT665) 5’taa atg tct
ggt caa gca gtt tct ttg agc cg3’; rev. (MT666): 5’tgt tga gca ccg gtt gca gcc tgt tat tag ag3 ’.
All constructs were verified by DNA sequencing.

Point mutations in the RKKFRE T region of CBF1 .

The point mutations in the RKKFRET motif were designed by using environment-
specific substitution tables (Overington et al., 1992), which allowed to choose substitutions
compatible with the helical structure. To introduce point mutations, the QuikChange
mutagenesis protocol was followed as described in the Strategene manual. The following
primers were used: Argl—>Lys, fwd: S’ccg aag aaa ccg gcg ggc aag aag aag ttt cgt gag act
cg3’; rev: 5’cga gtc tca cga aac ttc ttc ttg ccc gcc ggt tct tcg3’; Argl—>Ser, fwd: 5’cga aga
acc ggc ggg ctc gaa gaa gtt tcg tga gac tcg3’; rev: S’cga gtc tca cga aac ttc ttc gag ccc gcc
ggt ttc ttc g3’; Lys2—rArg, fwd: 5’ccg gcg ggc aga aga aag ttt cgt gag act cg3’; rev: 5’cga
gtc tca cga aac ttt ctt Actg ccc gcc gg3’; LysZ—>Ala, fwd: 5’ccg gcg ggc aga gcg aag ttt cgt
gag act cgt cac3’; rev: 5’gtg acg agt ctc acg aaa ctt cgc tct gcc cgc cgg3’; Lys3—Arg,
fwdz5’ccg gcg ggc aga aag aga ttt cgt gag act cgt cac c3’; revz5’ggt gac gag tct cac gaa atc
tct ttc tgc ccg cc gg3’; Lys3—>A1a, fwd: 5’ccg gcg ggc aga aag gcg ttt cgt gag act cgt cac
c3’; rev: 5’ ggt gac gag tct cac gaa acg cct ttc tgc ccg ccg g3’; Phe4—)Ala, fwd: S’ggg ccg
taa gaa ggc tcg aga gac tcg tca ccc3’; rev: 5’ ggg tga cga gtc tct cga gcc ttc tta cgg ccc3’;

Phe4—>Pro, fwd: 5’ gcg ggc cgt aag aag cct cga gag act cgt cac cc3’; rev: 5’ ggg tga cga gtc

107

tct cga ggc ttc tta egg ccc gc3’; Phe4—>Tyr, fwd: 5’ccg gcg ggc aga aag aag tac cgt gag act
cgt c3’; rev: 5’ gac gag tct cac ggt act tct ttc tgc ccg ccg g3’; Arg5—+8er, fwd: 5’ccg gcg ggc
aga aag aag ttt agt gag act cgt cac c3’; rev: 5’ ggt gac gag tct cac taa act tct ttc tgc ccg ccg
g3’. For the screening of the clones harboring the desired mutation, the primers were
designed to disrupt a pre-existing Sau96I restriction site in all mutant ORFs, except for

Phe4—+Ala and Phe4—rPro mutations, where a new XhoI restriction site was inserted.

Preparation of constructions for plant transformation.
Plant expression vector harboring wild type and mutated CBF1.

BglII restriction ends were added by PCR in order to subclone the ORF of wild
type or mutated CBF1 into pGA643 (An et al. , 1988) which harbors the constitutive 358
CaMV promoters. Primers BglII fwd: 5’gaa gat cta tga act cat ttt cag ctt ttt ctg3’; BglII
rev: 5’ gaa gat ctc tcg ttt cta caa caa taa aat aaa3’. A11 constructs were verified by DNA
sequencing.

Plant expression vector harboring 6xMyc: CBF1 ORFs.

F usional translation of wild type and mutated CBF1 to 6xMyc, were generated by
subcloning a SmaI/Sacl CBF1 ORF into the binary vector pKVB24 (V lachonasios et al. ,
2003) containing a 6xMyc tag, under the control of CaMV 35S promoter. The
translational fusion results in 6xMyc at the N-terminus of CBF1. Primers used to add
SmaI/SacI ends by PCR amplification were: MT708, fwd: 5’agc ccg ggg atg aac tca tt3’
MT709, rev.: 5’cag agc tct tac taa ctc ca3’.

Plant expression vector harboring CBF1:GFP:GUS.

108

Wild type CBF1 and CBFIAPKK ORFs were pEZT-CL(GUS). The vector was
engineered by subcloning a BamHI GUS insert into the pEZT-CL plant expression vector
(Schena et al. , 1991), resulting in an in frame fusion to GFP under the control of the CaMV
35 S promoter. The eGFP gene in pEZT is based on mGFP4 (Haseloff et al. , 1997) and
contains additional mutations (S65T, Y66H) to increase intrinsic GFP fluorescence
(Cormack et al. , 1996). BamHI restriction ends were added to GUS by PCR amplification
using the following primers: fwd. (MT794): 5’cag gat ccg cat cga taa gct tga att cac c3’;
rev. (MT795): 5’aga gga tcc cca att ccc gag gct gta3’. Full-length and 5’ deletions of CBF1
ORF were subcloned into the XhoI restriction site of pEZT-CL(GUS). In frame fusions of
CBF1 to GFP:GUS were generated by inserting an XhoI fragment into the binary vector. A
PCR approach was used to add the proper restriction ends, as follows: firll-length CBF 1
XhoI 5’ end (MT704): 5’ gcc tcg aga tga act cat ttt cag3’; XhoI 3’ end (MT825): 5’ccc tcg
agg cgt aac tcc aaa gcg3’For the 5’ ORF deletions the following forward primers were
used: construct 802 (MT802): S’acc tcg aga tgc cga aga aac cgg cgg gcc g3’; construct 801
(MT801): S’acc tcg aga tga ttt aca gag gag ttc g3’; construct 800 (MT800): 5’acc tcg aga
tgg act cgg ctt ggc ggc tac g3’; construct 799 (MT799): 5’acc tcg aga tgc tac gaa tcc cgg
agt caa c3’. The XhoI insert for NIa was PCR-amplified from the yeast plasmid pAVA367,
containing the in frame fusion Nia-GFP (Schena et al. , 1991), with the following primers:
fwd. (MT815): S’acc tcg aga tgg gga aga aga atc aga agc aca agc taa aga tga g3’; rev.
(MT826): 5’ccc tcg agt gac ctg tca atg gat cca 03’.

Plant transformation.
All the transgenic lines used in this study were generated by the floral dip method

described by Clough and Bent (1998). pEZT-CL(GUS) or pKVBZ4 constructs were

109

transformed into the A. tumefaciens strain LBA4404, whereas pGA643 constructs were
transformed into strain GV3101. Generation and selection of transgenic lines was
performed as follows. Arabidopsis plants (T0) were grown in soil to bolting and then
transformed by A. tumefaciens-mediated transformation using the floral dip method
(Clough and Bent, 1998). Seeds from T0 plants were screened for antibiotic or herbicide
resistance (600uM Basta; 50ug/ml Kanamycin). Resistant T1 seedlings were grown to
maturity to produce T2 seeds. T2 seeds were plated onto selective media containing
antibiotic or herbicide and screened for a 3:1 survival ratio, indicating that there was an
insertion in a single locus in the T1 genome. Homozygous lines were selected by growing

plants on selective media and screened for 100% survival.

Growth conditions and Northern blot analysis of transgenic Arabidopsis plants.
Seeds were sterilized and stratified for 3 days at 4°C, and then plated on
Gamborg’s plates supplemented with BS nutrients and 23.21 gr/liter of sucrose (Caisson
Laboratories, Inc., Rexburg, ID, USA) and 0.8% phytagar (Life Technologies Inc.,
Gaithersburg, MD, USA). Seedlings were grown under continuous cool-white fluorescent
light (100umol-m'2-s") at 22°C for two weeks. Tissue samples were harvested in liquid
Nitrogen and total RNA extracted using either the RN easy Plant Miniprep kit (QIAGEN)
or the Trizol reagent (Life Technologies, Gaithersburg, MD). Five to ten micrograms of
total RNA were fractionated in 1% formaldehyde gels and transferred onto nitrocellulose
membranes as described (Sambrook et al. , 1989). Membranes were hybridized in Church
buffer (1% BSA, 1 mM EDTA, 0.5 M NaPO4 pH 7.2, 7% SDS) (Church and Walter,

1984) at 65°C, overnight. Blots were probed with a 32P-labelled fragments prepared using

110

the Random Primers DNA Labeling System (Invitrogen), according to the manufacturer
instructions. Probes for all the transcripts were obtained by using the full-length ORF as a
probe. After hybridization, membranes were washed three times in .IXSSC, 0.1%SDS at
55°C, for 15 minutes. mRN A levels in different samples were normalized by comparing
them to the levels of 188 rRNA determined from the same blots. Following the washes,
the membranes were exposed to a phosphorimager screen; the screen was later scanned

and then quantified using a QuantityOne software from BioRad (Hercules, CA, USA).

Analysis of variance (ANOVA) of COR/CBF1 transcript ratios.

AN OVA tests were carried out using to determine the effect of mutations in the
PKKPAGR motif on COR78/CBF1 mRNA ratios. mRNA values for CBF1 and different
COR genes were obtained by exposing radioactive membranes to a phosphorimager
screen. The screen was later scanned and the photointensity of different bands was
measured using a QuantityOne software from BioRad (Hercules, CA, USA). Variables
were log-transformed to meet the normality assumptions of statistical analyses. Each
analysis was conducted by averaging ratios from two technical replicates. Each group
analyzed was represented by five or six independent transgenic lines. Statistical
difference in COR/CBF1 ratios of transgenic plants overexpressing a wild type CBF1
transgene or CBF1 transgenes mutated in the signature sequences was analyzed by
analysis of variance (AN OVA) using SAS Proc Mixed procedures (SAS Institute, Cary,
NC), version 9.1. To test for significant difference between COR/CBF1 ratios between

transgenic plants overexpressing wild type CBF1 or the mutated versions described

111

earlier we estimated least-square means for each genotype and compared them to the
least-square means of the control plants overexpressing wild type CBF1. These estimates
were used to calculate the t-values and the statistical significance at P<0.0001 for all the

transgenic lines analyzed.

Subcloning of CBF1 mutant ORFs into bacterial expression vectors for expression

_ and purification of 6xHiszCBFl and MBP:CBF1 proteins.
Expression and purification of 6xHis:CBF 1 proteins.

All CBF1 ORFs described earlier were introduced into the pET28a+ expression vector

(N ovagen/EMD, San Diego, CA) under the control of the bacteriophage IPTG-inducible T7
promoter. The constructs were generated by NcoI/XhoI cleavage of the pSJG6 construct
(ref), in which CBF1 is fused to 6His-T7 tag. The NcoI/XhoI fragments were ligated onto
pET28a+. Sequences were confirmed for all constructs and plasmids were transformed into
BL21-CodonPlus(DE3)-RIL (Stratagene, La Jolla, CA) E. coli strain for optimal expression
of the recombinant proteins (Kleber-Janke and Becker, 2000). Protein expression and
purification was started by inoculation of individual colonies containing plasmid DNA in
3mL of Luria-Bertani broth containing 50ug/ml Kanamycin and 40ug/ml Chloramphenicol
(Sigma-Aldrich, St.Louis, MO) and grown at 37°C, overnight (250 rpm). 50ul of that
culture were inoculated into 500ml of the same broth to an 0ng of approximately 0.9
before addition of lmM IPTG (Research Organics, Cleveland, OH) to induce protein
expression for 4 hours, at 37°C. Cells were harvested by centrifugation, and resuspended in
101111 of cell lysis buffer (SOmM NaHzPOr-HzO, 300mM NaCl, IOmM Imidazole, pH8.0).

Cells were lysed by sonication (BRANSON Sonifier150, Pittsburgh, PA) in the presence of

112

protease inhibitors (Complete EDTA-free protease inhibitor cocktail tablets, Roche,
Mannheim, Germany). The soluble protein fiaction was separated by centrifugation, at 4°C.
The supernatant fraction containing 6xHis-T7-CBF1 proteins was loaded over a nickel
column equilibrated With wash buffer (SOmM NaHzPO4'HzO, 300mM NaCl, 20mM
Imidazole, pH8.0). After three washes in wash buffer, 6xHis-T7-CBF1 proteins were eluted
by washing with elution buffer (SOmM NaHzPO4-HZO, 300mM NaCl, 250mM Imidazole,
pH8.0). Protein concentration was determined using the Bradford dye binding assay (Bio-
Rad, Hercules, CA), with BSA used as standard.
Expression and purification ofMBP:CBF127.1 12 proteins

A 258bp anI/Xbal CBF1 fragment (aa27-112) was cloned into pMAL-c2x
vector (New England BioLabs, Beverly, MA), downstream of the malE gene, which
encodes Maltose Binding Protein (MBP) under the control of the strong Pm promoter.
this resulted in the translational fusion of CBF1 to the C—terminus of MBP. The primers
used for wild type CBF1 and all the PKKPAGR mutants were: MT908(fwd): S’ATA
TTT TCT AGA TTC GTA GCC3’, and MT909(rev.): 5’CCA TGG AAG GAT TTC
GGC CAC GAG TTG T3’. For the dsawr CBF1 mutant, MT910 was used as the reverse
primer: 5’ATA TTT TCT AGA TTC GTA GCG C3’. Individual transformed cells were
grown at 37°C in 3ml Rich Medium (10g tryptone, 5 g yeast extract 5g NaCl, 2g glucose,
per liter) supplemented with lOOug/ml Ampicillin to an CD. of 0.5. Protein expression
was induced by addition of lmM IPTG to the bacterial suspension, and let grow for 3
additional hours. Cells were harvested by centrifugation (6,000 rpm, 10min., 4°C), and
lysed by sonication, as described above. The soluble protein fraction was separated by

centrifugation at 13,000rpm, 30 min., at 4°C. The supernatant containing the fusion

113

proteins was loaded over an amylase column for protein purification by affinity. The
amylase column was equilibrated in column buffer (20mM Tris-HCl, 200mM NaCl,
lmM EDTA). MBP-CBF1 proteins were eluted in column buffer containing lOmM

maltose. Protein concentration was determined using the Bradford dye binding assay

(Bio-Rad, Hercules, CA).

Protein isolation and immunoblot analyses.
Immunodetection of CBF1 protein fiom E. coli extracts.

Western Blotting and protein detection were carried out by first separating the
proteins on a 10% bis-acrylamide tricine gel. Proteins were transferred in Towbin buffer
(25mM2 Tris, 192mM Glycine, 15% MeOH, 0.02% SDS) to a PVDF membrane
(Irnmobilon-P, Millipore corporations, Bedford, MA) using by electrophoretic transfer
(Genie®, Idea Scientific, Minneapolis, MN), for 45min. at lAmp, 30 volts.

The membrane was blocked in PBS-Tween (per liter: 80g NaCl, 2gr KCl, 14,4gr
NazHPO4, 2.4gr KHzPO4, pH7.2, 0.1% (v/v) Tween 20) and 5% non-fat dried milk for 1h
at RT with gentle rocking. Incubation with the primary antibody was conducted for 1h at
room temperature in PBS-T and 5% non-fat dried milk for 1h at RT. Antibody dilutions
were: 1/7,500 for a-CBFI (F1, IgG purified CBF1 antiserum raised against the full-
length CBF1 protein fused to a 6xHis tag); 1/ 10,000 for a-MBP (New England Biolabs,
Beverly, MA). The presence of the protein on the membrane was assayed by
chemiluminescence detection (Amersham Biosciences, Piscataway, NJ).

Immunodetection of CBF1 protein fi'om plant extracts.

114

Total protein extracts were prepared from two-week old Arabidopsis seedlings by
grinding frozen tissue in protein extraction buffer (20 mM Tris-HCl, pH 8.0, SOmM
NaCl, 5 mM EDTA, 0.05 % SDS) supplemented with protease inhibitor cocktail from
Roche. Protein concentration was determined using the Bradford reagent (Bio—Rad,
Hercules, CA, USA) with BSA as the standard. Equal amounts of proteins were subjected
to SDS-polyacrylamide gel electrophoresis (PAGE) (4-20% gradient gel). After
electrophoresis, the proteins were transferred to polyvinylidene difluoride (PVDF)
membranes by electrotransfer. Membranes were blocked in 5% nonfat milk powder in
Tris-buffered saline (TBS) - 0.1% Tween-20 for l h at room temperature, and then
incubated overnight at 4°C with specific antibodies. Antibody dilutions were: 1/5,000 for
the a-Myc monoclonal antibody Roche-Mannheim, USA), and 1/10,000 for the a-GFP
monoclonal antibody (Pierce Biotechnology, Rockford, IL). The next day, membranes
were washed three times and incubated in horseradish peroxidase-coupled anti-mouse
IgG antibody [1/ 10,000 dilution of the rabbit anti-mouse IgG (Pierce Biotechnology, Inc.,
Rockford, IL)] for 1 h at room temperature. Immunoreactive bands were visualized by an

enhanced chemiluminesence assay following the manufacturer's inst2ructions.

Electrophoretic-Mobility Shift Assays (EMSA).

24-bp fragments from COR15a, COR 78 and COR6.6 promoters containing a
CR T/DRE element were prepared by synthesizing both strands. The oligomers were
suspended in 1X STE buffer (10mM Tris, pH8.0; 10mM NaCl; lmM EDTA), and the
complementary strands were annealed. The resulting double strand oligonucleotides were

labeled with [y-32P]ATP (NEN Life Science Products) by T4 polynucleotide kinase (New

115

England Biolabs, Beverly, MA) and purified through a Sephadex G-50 column. Mutant
and wild type sequences were prepared following the same procedure. The DNA
oligomers used for the binding assays are the following: COR15a: 5’ATT TCA TGG
w CTG CTT TTT3’; COR78: S’AAT ATA CTA w ATG AGT TCT 3’;

COR6. 6: S’AAA AAG CTA CCGAC ATA AGC CAA3’. A mutant version of COR15a

 

in which the entire C-repeat, CCGAC, had been altered was used for all the competition
assays: 5’ ATTTCATGGtatgtCTGCT'ITTT3’. The binding of 6xHis:CBF] or
MBP:CBF127-112 to the CRT/DRE-containing DNA probes was tested in a total volume of
12.0 111 as follows: 0.5 ng of a 32P-labeled DNA probe encoding the CRT/DRE binding
site was mixed with a gradient concentration of each recombinant protein (100 - 300 ng
of 6xHis:CBF 1 and 0.3 — 15 ug for MBP:CBF127-112) and incubated in binding buffer (20
mM Tris-HCl, pH 8.0; 100 ug/ml BSA; 30mM KCl; 5mM MgClz; 4% Glycerol; lmM
DTT) in the presence/absence of 100 ng unlabeled competitor DNA. Afier incubation at
room temperature for 20 min., samples were loaded into nondenaturing polyacrylamide
gels (4% wt/vol), and fractionated by electrophoresis at 150V, for 3 hours at 4°C. The
gels were dried at 80°C, 30 min., and exposed to a phosphorimager screen. The screen
was later scanned and the photointensity of complexed DNA was measured using a

QuantityOne software from BioRad (Hercules, CA, USA).

Fluorescence imaging of Arabidopsis root tips overexpressing CBF1 :GFP:GUS
transgenes.
For cellular localization of CBF1 :GFP:GUS proteins, Arabidopsis seeds were

sterilized, stratified for 3 days at 4°C, and plated on Gamborg agar plates supplemented

116

with sucrose as described above. The Petri dishes were oriented vertically and the
seedlings grown for 6-7 days before being imaged. Laser confocal images were collected
using an upright LSM Zeiss 510 META microscope equipped with a 40X oil immersion
objective. To visualize the nuclei, Arabidopsis seedlings were incubated with a solution
of 50 ug/ml Propidium Iodide (PI) and 5 rig/ml RNase, in the dark for 30 min — 1 hr.
Samples were rinsed with distilled water a couple of times and mounted in tap water.

GF P fluorescence images were obtained using Argon ion laser excitation of 488 nm with
a 505/530 nm bandpass filter. PI fluorescence images were collected using an excitation
line of 543 nm with a 560 nm longpass filter. Postacquisition image processing was done
with the LSM 5 Image Browser (Zeiss) and Adobe Photoshop 5.0 software (Mountain

View, CA).

117

LITERATURE CITED

Abel, S., Oeller, P., and Theologis, A. (1994). Early auxin-induced genes encode short-
lived nuclear proteins. PNAS 91, 326-330.

Adamczak, Porollo A, and J, M. (2005). Combining prediction of secondary structure

and solvent accessibility in proteins. Proteins: Structure, Function, and
Bioinforrnatics 59, 467-475.

Allen, M., Yamasaki, K., Ohme—Takagi, M., Tateno, M., and Suzuki, M. (1998). A
novel mode of DNA recognition by a beta-sheet revealed by the solution structure
of the GCC-box binding domain in complex with DNA. EMBO J 17, 5484-5496.

An, G., Ebert, P., Mitra, A., and Ha, S. (1988). Binary vectors. (Dordrecht: Kluwer
Academic Publishers).

Birkenbihl, R., Jach, G., Saedler, H., and Huijser, P. (2005). Functional dissection of
the plant-specific SBP-domain: overlap of the DNA binding and nuclear
localization domain. JMB 352, 585-596.

Biswas, T., Aihara, H., Radman-Livaja, M., Filman, D., Landy, A., and Ellenberger,
T. (2005). A structural basis for allosteric control of DNA recombination by
[lambda] integrase. Nature 435, 1059.

Carey, M., Kakidani, H., Leatherwood, J., Mostashari, F., and Ptashne, M. (1989).
An amino-terminal fragment of GAL4 binds DNA as a dimer. Journal of
Molecular Biology 209, 423.

Carrington, J.C., Freed, D.D., and Leinicke, A.J. (1991). Bipartite signal sequence

mediates nuclear translocation of the plant potyviral Nla protein. Plant Cell 3,
953-962.

Chakrabartty, A., Kortemme, T., and Baldwin, R.L. (1994). Helix propensities of the
amino acids measured in alanine-based peptides without helix-stabilizing side-
chain interactions. Protein Sci 3, 843-852.

Chen, H.A., Simpson, P., Huyton, T., Roper, D., and Matthews, S. (2005). Solution
structure and interactions of the Escherichia coli Cell Division Activator protein
CedA. Biochemistry 44, 673 8-6744.

Chou, P.Y., and Fasman, GD. (1978). Empirical predictions of protein conformation.
Annual Review of Biochemistry 47, 251-276.

Christophe-Hobertus, C., Duquesne, V., Pichon, B., Roger, RR, and Christophe, D.
(1999). Critical residues of the homeodomain involved in contacting DNA bases
also specify the nuclear accumulation of thyroid transcription factor-1. European

Journal of Biochemistry 265, 491-497.

118

Church, G., and Walter, G. (1984). Genomic sequencing. PNAS 81, 1991-1995.

Clough, S.J., and Bent, A.F. (1998). Floral dip: a simplified method for Agrobacterium-
mediated transformation of Arabidopsis thaliana. The Plant Journal 16, 735-743.

Collawn, J.F., Stangel, M., Kuhn, L.A., Esekogwu, V., Jing, S.Q., Trowbridge, 1.8.,
and Tainer, J.A. (1990). Transferrin receptor internalization sequence YXRF
implicates a tight turn as the structural recognition motif for endocytosis. Cell 63,
1061-1072. _

Connolly, K.M., Wojciak, J.M., and Clubb, RT. (1998). Site-specific DNA binding
using a variation of the double stranded RNA binding motif. Nat Struct Biol 5,
546-550.

Connolly, K.M., Ilangovan, U., Wojciak, J.M., Iwahara, M., and Clubb, RT. (2000).
Maj or groove recognition by three-stranded [beta]-sheets: afiinity determinants
and conserved structural features. Journal of Molecular Biology 300, 841-856.

Cormack, B.P., Valdivia, R.H., and Falkow, S. (1996). FACS-optimized mutants of the
green fluorescent protein (GFP). Gene Flourescent Proteins and Applications 173,
33-3 8.

Craig, L., Sanschagrin, P.C., Rozek, A., Lackie, S., Kuhn, L.A., and Scott, J.K.
(1998). The role of structure in antibody cross-reactivity between peptides and
folded proteins. Journal of Molecular Biology 281, 183-201.

Crane-RobinsOn, C., Dragan, A.I., and Privalov, P.L. (2006). The extended arms of
DNA-binding domains: a tale of tails. Trends in Biochemical Sciences 31, 547.

DeWitt, N.D., Hong, B., Sussman, M.R., and Harper, J.F. (1996). Targeting of two
Arabidopsis H+-ATPase isoforrns to the plasma membrane. Plant Physiol. 112,
833-844.

Dragan, A., Read, C., Makeyeva, E., Milgotina, E., Churchill, M., Crane-Robinson,
C., and Privalov, P. (2004). DNA binding and bending by HMG boxes: energetic
determinants of specificity. Journal of Molecular Biology 343, 371.

Eyrich, V.A., Marti-Renom, M.A., Przybylski, D., Madhusudhan, M.S., Fiser, A.,
Pazos, F., Valencia, A., Sali, A., and Rost, B. (2001). EVA: continuous
automatic evaluation of protein structure prediction servers. Bioinforrnatics 17,
1242-1243.

Gilmour, S.J., Fowler, S.G., and Thomashow, M.F. (2004). Arabidopsis transcriptional
activators CBF1, CBF2, and CBF3 have matching functional activities. Plant Mol
Biol 54, 768-781.

Gilmour, S.J., Sebolt, A.M., Salazar, M.P., Everard, J .D., and Thomashow, M.F.
(2000). Overexpression of the Arabidopsis CBF3 transcriptional activator mimics

119

multiple biochemical changes associated with cold acclimation. Plant Physiol.
124, 1854-1865.

Gilmour, S.J., Zarka, D.G., Stockinger, E.J., Salazar, M.P., Houghton, J.M., and
Thomashow, M.F. (1998). Low temperature regulation of the Arabidopsis CBF
family of AP2 transcriptional activators as an early step in cold-induced COR
gene expression. The Plant Journal 16, 433-442.

Gilmour SJ, Fowler SG, and MF, T. (2004). Arabidopsis transcriptional activators
CBF1, CBF2, and CBF3 have matching functional activities. Plant Mol Biol 54,
768-781.

Grebenok, R., Pierson, E., Lambert, G., Gong, F., Afonso, C., Haldeman-Cahill, R.,
Carrington, J., and Galbraith, D. (1997). Green-fluorescent protein fusions for
efficient characterization of nuclear targeting. Plant J 11, 573-5 86.

Guana, C., Lib, P., Riggsa, P., and Inouyeb, H. (1987). Vectors that facilitate the
expression and purification of foreign peptides in Escherichia coli by fusion to
maltose-binding protein. Gene 67, 21-30.

Haake, V., Cook, D., Riechmann, J.L., Pineda, O., Thomashow, M.F., and Zhang,
J.Z. (2002). Transcription factor CBF4 is a regulator of drought adaptation in
Arabidopsis. Plant Physiol. 130, 639-648.

Hao, D., Ohme-Takagi, M., and Sarai, A. (1998). Unique mode of GCC box
recognition by the DNA-binding domain of ethylene-responsive element-binding
factor (ERF domain) in plant. J Biol Chem 273, 26857-26861.

Haselofl', J., Siemering, K.R., Prasher, D.C., and Hodge, S. (1997). Removal of a
cryptic intron and subcellular localization of green fluorescent protein are
required to mark transgenic Arabidopsis plants brightly. PNAS 94, 2122-2127.

Holm, L., and Sander, C. (1993). Protein structure comparison by alignment of distance
matrices. J Mol. Biol. 233, 123-138.

Jaglo-Ottosen, K.R., Gilmour, S.J., Zarka, D.G., Schabenberger, 0., and
Thomashow, M.F. (1998). Arabidopsis CBF1 overexpression induces COR
genes and enhances freezing tolerance. Science 280, 104-106.

Jaglo, K.R., Kleff, S., Amundsen, K.L., Zhang, X., Haake, V., Zhang, J.Z., Deits, T.,
and Thomashow, M.F. (2001). Components of the Arabidopsis C-
Repeat/Dehydration-Responsive Element Binding Factor cold-response pathway
are conserved in Brassica napus and other plant species. Plant Physiol. 127, 910-
917.

Jofuku, K.D., Boer, B.G., Montagu, M.V., and Okamuro, J.K. (1994). Control of
Arabidopsis flower and seed development by the homeotic gene APETALA2.
PlantCell6, 1211-1225.

120

Kaffman, A., and O'Shea, E. (1999). Regulation of nuclear localization: a key to a door.
Annual Review of Cell and Developmental Biology 15, 291-339.

Karplus, K., Draper, J., Casper, J., Mandel-Gutfreund, Y., Diekhans, M., and
Hughey, R. (2003). Combining local-structure, fold-recognition, and new fold
methods for protein structure prediction. Proteins: Structure, Function, and
Genetics 53, 491-496.

Kim, J., Mao, J., and Gunner, M. (2005). Are acidic and basic groups in buried
proteins predicted to be ionized? Journal of Molecular Biology 348, 1283.

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

Kleber-Janke, T., and Becker, W.-M. (2000). Use of modified BL21(DE3) Escherichia
coli cells for high-level expression of recombinant peanut allergens affected by
poor codon usage. Protein Expression and Purification 19, 419.

Klucher, K.M., Chow, H., Reiser, L., and Fischer, R.L. (1996). The
AINTEGUMENTA gene of Arabidopsis required for ovule and female

gametophyte development is related to the floral homeotic gene APETALA2.
Plant Cell 8, 137-153.

Koroleva, 0., Tomlinson, M., Leader, D., Shaw, P., and Doonan, J. (2005). High-
throughput protein localization in Arabidopsis using Agrobacterium-mediated
transient expression of GFP-ORF fusions. Plant J 41, 162-174.

Krizek, B., and Sulli, C. (2006). Mapping sequences required for nuclear localization
and the transcriptional activation function of the Arabidopsis protein
AINTEGUMENTA. Planta 224, 612-621.

LaCasse, E., and Lefebvre, Y. (1995). Nuclear localization signals overlap DNA- or
RNA-binding domains in nucleic acid-binding proteins. Nucleic Acids Res. 23,
1647-1656.

Li, S., and Wilkinson, M. (1997). Site-directed mutagenesis: a two-step method using
PCR and DpnI. Biotechniques 23, 588-590.

Liu, Q., Kasuga, M., Sakuma, Y., Abe, H., Miura, S., Yamaguchi-Shinozaki, K., and
Shinozaki, K. (1998). Two transcription factors, DREBI and DREB2, with an
EREBP/AP2 DNA binding domain separate two cellular signal transduction
pathways in drought- and low-temperature-responsive gene expression,
respectively, in Arabidopsis. Plant Cell 10, 1391-1406.

Lnenicek-Allen, M., Read, C., and Crane-Robinson, C. (1996). The DNA bend angle
and binding affinity of an HMG box increased by the presence of short terminal
arms. Nucl. Acids Res. 24, 1047-1051.

121

Luscombe, N., and Thornton, J. (2002). Protein-DNA interactions: amino acid
conservation and the effects of mutations on binding specificity. Jom'nal of
Molecular Biology 320, 991 .

Luscombe, N., Laskowski, R., and Thornton, J. (2001). Amino acid-base interactions:
a three-dimensional analysis of protein-DNA interactions at an atomic level. Nucl.
Acids Res. 29, 2860-2874.

Lyck, R., Harmening, U., Hohfeld, 1., Trenter E, Scharf KD, and L, N. (1997).
Intracellular distribution and identification of the nuclear localization signals of
two plant heat-stress transcription factors. Planta 202, 117-125.

Magome, H., Yamaguchi, S., Hanada, A., Kamiya, Y., and Ode, K. (2004). dwarf and
delayed-flowering 1, a novel Arabidopsis mutant deficient in gibberellin
biosynthesis because of overexpression of a putative AP2 transcription factor. The
Plant Journal 37, 720—729.

Maruyama, K., Sakuma, Y., Kasuga, M., Ito, Y., Seki, M., Gods, H., Shimada, Y.,
Yoshida, S., Shinozaki, K., and Yamaguchi-Shinozaki, K. (2004).
Identification of cold-inducible downstream genes of the Arabidopsis
DREBlA/CBF3 transcriptional factor using two microarray systems. The Plant
Journal 38, 982-993.

McGuffin, L.J., Bryson, K., and Jones, D.T. (2000). The PSIPRED protein structure
prediction server. Bioinformatics 16, 404—405.

Molar/ck, 2., Neidle, 8., and Schneider, B. (2002). Protein and drug interactions in the
minor groove ofDNA. 30, 1182-1191.

Nadassy, K., Wodak, S., and Janin, J. (1999). Structural features of protein-nucleic
acid recognition sites. Biochemistry 38, 1999-2017.

Nagadoi, A., Nakazawa, K., Uda, H., Oknno, K., Maekawa, T., Ishii, S., and
Nishimura, Y. (1999). Solution structure of the transactivation domain of ATF-2
comprising a zinc finger-like subdomain and a flexible subdomain. J.Mol.Biol.
287, 593-607.

O'Neil, K., and DeGrado, W. (1990). A thermodynamic scale for the helix-forming
tendencies of the commonly occurring amino acids. 250, 646-651.

Ohme-Takagi, M., and Shinshi, H. (1995). Ethylene-inducible DNA binding proteins
that interact with an ethylene-responsive element. Plant Cell 7, 173-182.

Overington, J., Donnelly, D., Johnson, M.S., Sali, A., and Blundell, T.L. (1992).

Environment-specific amino acid substitution tables: Tertiary templates and
prediction of protein folds. Protein Sci 1, 216-226.

122

Poon, I.K.H., and Jans, D.A. (2005). Regulation of nuclear transport: central role in
development and transformation? Traffic 6, 173-186.

Prieve, M.G., Guttridge, K.L., Mnnguia, J., and Waterman, ML. (1998). Differential
importin-alpha recognition and nuclear transport by nuclear localization signals
within the High-Mobility-Group DNA binding domains of Lymphoid Enhancer
Factor 1 and T-Cell Factor 1. Mol. Cell. Biol. 18, 4819-4832.

Rashotte, A.M., Mason, M.G., Hutchison, C.E., Femira, F.J., Schaller, G.E., and
Kieber, J.J. (2006). A subset of Arabidopsis AP2 transcription factors mediates
cytokinin responses in concert with a two-component pathway. PNAS 103,
11081-11085.

Ren, J., Esnouf, R., German, E., Somers, D., Ross, C., Kirby, 1., Keeling, J., Darby,
G., Jones, Y., and Stuart, D. (1995). High resolution structures of HIV -1 RT
from four RT-inhibitor complexes. Nat Struct Biol. 2, 293-302.

Riechmann, J.L., and Meyerowitz, EM. (1998). The AP2/EREBP family of plant
transcription factors. Biol Chem 379, 633-646.

Rivas, S., Romeis, T., and Jones, J.D.G. (2002). The Cf-9 disease resistance protein is
present in an ~420—Kilodalton heteromultimeric membrane-associated complex at
one molecule per complex. Plant Cell 14, 689-702.

Rost, B., and Liu, J. (2003). The PredictProtein server. Nucl. Acids Res. 31, 3300-3304.

Ryter, J.M., and Schultz, S.C. (1998). Molecular basis of double-stranded RNA-protein
interactions: structure of a dsRNA-binding domain complexed with dsRNA.
Embo J 17, 7505-7513.

Sakuma, Y., Lin, 0., Dubouzet, J.G., Abe, H., Shinozaki, K., and Yamaguchi-
Shinozaki, K. (2002). DNA-binding specificity of the ERF/AP2 domain of
Arabidopsis DREBs, transcription factors involved in dehydration- and cold-
inducible gene expression. Biochem Biophys Res Commun 290, 998-1009.

Sakuma,Y., Maruyama,K.,Osahbe,Y.,Qin,F.,Seki,M.,Shinozaki,K.,and
Yamaguchi—Shinozaki, K. (2006). Functional analysis of an Arabidopsis
transcription factor, DREBZA, involved in drought-responsive gene expression.
Plant Cell 18, 1292-1309.

Sambrook, J., Fritsch, E., and Maniatis, T. (1989). (Cold Spring Harbor, NY: Cold
Spring Harbor Laboratory Press).

Schena, M., Pieard, D., and Yamamoto, K. (1991). Vectors for constitutive and
inducible gene expression in yeast. Methods Enzymology 194.

Silver, P.A., Keegan, L.P., and Ptashne, M. (1984). Amino terminus of the yeast GAL4
gene product is sufficient for nuclear localization. PNAS 81, 5951-5955.

123

Stockinger, E.J., Gilmour, S.J., and Thomashow, M.F. (1997). Arabidopsis thaliana
CBF1 encodes an AP2 domain-containing transcriptional activator that binds to
the C—repeat/DRE, a cis-acting DNA regulatory element that stimulates
transcription in response to low temperature and water deficit. PNAS 94, 1035-
1040.

Sudbeck, P., and Scherer, G. (1997). Two independent nuclear localization signals are
present in the DNA-binding High-mobility Group domains of SRY and SOX9. J.
Biol. Chem. 272, 27848-27852.

Tang, Y., Ramakrishnan, C., Thomas, J., and Deme DB. (1997). A role for
HDJ-2/HSDJ in correcting subnuclear trafficking, transactivation, and
transrepression defects of a glucocorticoid receptor zinc finger mutant. Mol Biol
Cell 8, 795-809.

Tansey, W.P. (2001). Transcriptional activation: risky business. Genes Dev. 15, 1045-
1050.

Vlaclronasios, K.E., Thomashow, M.F., and Triezenberg, SJ. (2003). Disruption
mutations of ADA2b and GCN5 transcriptional adaptor genes dramatically affect
Arabidopsis growth, development, and gene expression. Plant Cell 15, 626-638.

Vogel, J.T., Zarka, D.G., Van Buskirk, H.A., Fowler, S.G., and Thomashow, M.F.
(2005). Roles of the CBF2 and ZAT12 transcription factors in configuring the low
temperature transcriptome of Arabidopsis. The Plant Journal 41, 195-211.

von Arnim, A.G., and Deng, X.-W. (1994). Light inactivation of Arabidopsis
photomorphogenic repressor COPl involves a cell-specific regulation of its
nucleocytoplasmic partitioning. Cell 79, 1035-1045.

Walther, R.F., Atlas, E., Carrigan, A., Rouleau, Y., Edgeeombe, A., Visentin, L.,
Lamprecht, C., Addielrs, G.C., Hache, R.J.G., and Lefebvre, Y.A. (2005). A
Serine/I‘hreonine-rich motif is one of three nuclear localization signals that
determine unidirectional transport ofthe mineralocorticoid receptor to the
nucleus. J. Biol. Chem. 280, 17549-17561.

Wang, W., and Malcolm, B. (1999). Two-Stage PCR protocol allowing introduction of
multiple mutations, deletions and insertions using QuikChange site-directed
mutagenesis. BioTechniques 26, 680-682.

Wang, Z., Triezenberg, S., Thomashow, M., and EJ., S. (2005). Multiple hydrophobic
motifs in Arabidopsis CBF1 COOH-terminus provide frmctional redundancy in
trans-activation. Plant Mol Biol 58, 543-559.

Wojciak, J., Sarkar, D., Landy, A., and Clubb, R. (2002). Arm-site binding by lambda

-integrase: solution structure and functional characterization of rts amino-terminal
domain. PNAS 99, 3434-3439.

124

Wojciak, J.M., Connolly, K.M., and Clubb, R.T. (1999).‘NMR structm'e of the Tn916
integrase-DNA complex 6, 366-373.

Wolberger, C. (1996). Homeodomain interactions. Current Opinion in Structural
Biology. 6, 62-68.

Zhang, R.-g., Kim, Y., Skarina, T., Beasley, S., Laskowski, R., Arrowsmith, C.,
Edwards, A., Joachimiak, A., and Savchenko, A. (2002). Crystal structure of
Themotoga maritima 0065, a member of the IclR transcriptional factor family. J.
Biol. Chem. 277, 19183-19190.

Zhang, X., Basso, W., and Matthews, B. (1991). Toward a simplification of the protein

folding problem: a stabilizing polyalanine alpha-helix engineered in T4 lysozyme.
Biochemistry 30, 2012-2017.

125

lllllllllllllllllllllll LIBRAFilES

Willi/lllill/fl!!!W!!!1fill/Illll/lI/MW

31293 02956 4006