gm:
.4: . .
‘su1. .. a
4.

V:

I... new. £2.33. . n.“
amwmflfimmmiy gum?
”(1‘4qiun‘1vqt—N ,

_ a ‘

H

.
r.
.2

WP; Mafia“; ..

1. r Q
a ”rd. nWwiwttfi
Epwf. sand.“ 1
q . 23W V
. El, 06 ‘
a t. a? dirt).

0%

.. 133. ii
. . .M. .3 ‘ z.
k a...¢§.w..2v§:
1.1. .5; 3..
o. ‘JI

i

nxhfiflma fifipf

i
1 33.714

(Iv... 531‘:
(2.3....

2!. :“v

1 )3.
fight}.

3131.1
1. .‘\»..a..

 

a3; a?

 

This is to certify that the
dissertation entitled

STRUCTURAL AND FUNCTIONAL STUDIES OF THE
ENZYMES INVOLVED IN A BACTERIAL GDP-D-RHAMNOSE
BIOSYNTHETIC PATHWAY

presented by

NICOLE A. WEBB

has been accepted towards fulfillment
of the requirements for the

Ph.D. degree in Biochemistry and Molecular
Biology

 

 

ZZZZJZ/Zv\

Major Professor’s Signature
/-:Z ' Z? —- (75"

Date

MSU is an Affirmative Action/Equal Opportunity Institution

LIBRARY
Michigan State
University

 

 

44-—.-o-n_‘-.-I--d-o-u-n-u-o-n—b—

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

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

2/05 p:/ClRC/DateDue.indd-p.1

STRUCTURAL AND FUNCTIONAL STUDIES OF THE ENZYMES INVOLVED
IN A BACTERIAL GDP-D-RHAMNOSE BIOSYNTHETIC PATHWAY

By

Nicole A. Webb

A DISSERTATION

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

DOCTOR OF PHILOSOPHY
Department of Biochemistry and Molecular Biology

2005

ABSTRACT

STRUCTURAL AND FUNCTIONAL STUDIES OF THE ENZYMES INVOLVED IN
A BACTERIAL GDP-D-RHAMNOSE BIOSYNTHETIC PATHWAY

By
Nicole A. Webb

D-rhamnose is a rare 6—deoxy sugar found primarily in the lipopolysaccharide
molecules of pathogenic gram-negative bacteria, where it is involved in host-bacterium
interactions and the establishment of infection. The biosynthetic pathway of D—rhamnose
may be a potential therapeutic target, as it is expected that inhibitors of the enzymes
would not affect human metabolic pathways. The biosynthesis of D-rhamnose proceeds
through the conversion of GDP-D-mannose by GDP-D-mannose 4,6-dehydratase (GMD)
to GDP-4—keto-6-de0xy-D-mannose, which is subsequently reduced to GDP-D-rhamnose
by the reductase (RMD). This study focuses on the structural and enzymatic
characterization of the two enzymes.

The X-ray crystal structure of GMD from Pseudomonas aeruginosa has been
determined in complex with NADPH and GDP. GMD belongs to the NDP-sugar
modifying subfamily of the short-chain dehydrogenase/reductase (SDR) enzymes, all of
which exhibit bidomain structures and a conserved catalytic triad (Ser/T hr, Tyr-XXX-
Lys). Although most members of this enzyme subfamily display homodimeric structures,
this bacterial GMD forms a tetramer in the same fashion as the plant Arabidopsis
thaliana GMD isoform, MURI. Based on the conservation of subunit interactions and
sequences in GMDS, evidence suggests that the tetrameric form of this enzyme is the
preferred, and perhaps functionally relevant. oligomeric state for most bacterial and

eukaryotic GMDs.

Initial X-ray analysis has been completed on RMD from P. aeruginosa as well as a
homolog from the bacterium Aneurinibacillus thermoaerophilus. While crystals of RMD
from P. aeruginosa diffracted to 3.7 A, crystals of RMD from A. thermoaerophilus
diffracted to 1.8 A. Complete data sets were collected on crystals from both species. The
P. aeruginosa RMD crystal belongs to the tetragonal space group P422 (or higher
symmetry) (a = b = 182, c = 250 A), with 12 molecules in the asymmetric unit. The A.
thermoaerophilus RMD crystal belongs to the triclinic space group P1 (a = 47, b = 56. c
= 79 A, or = 72, [3 = 83, y = 76°), with 2 molecules in the asymmetric unit. Amino acid
sequence homology details indicate that RMD is also a member of the SDR family,
whose structures are quite similar. Therefore, full structure determination by the
molecular replacement method using the coordinates from the most closely related SDR
protein, GMD from P. aeruginosa. is in progress. As no structure is currently available
for RMDs, crystallographic analysis of RMD should provide detailed information on the
active site of the enzyme and facilitate structure-based inhibitor design.

Finally, using a capillary-electrophoresis—based enzymatic assay, we have shown that
recombinant GMD and RMD from P. aeruginosa are able to convert GDP-D-mannose to
GDP-D-rhamnose, continuing the product structure by NMR analysis. Additionally, we
present a method for the enzymatic synthesis of GDP-D-rhamnose, an important
glycobiological building block not commercially available. Synthesis of GDP-D-
rhamnose is a crucial prerequisite for (l) the identification and biochemical studies of
corresponding glycosyltransferases, as well as (2) the determination of how this
deoxysugar, as a component of cell wall glycoconjugates, contributes to the virulence of

plant and human pathogens.

for my husband,
who has inspired me
to achieve more than I
ever dreamed possible

for my daughter
whose life
has provided new
meaning and purpose
to mine

iv

The proper rewards are not simply tacked on to the activity/or which they are given.
but are the activity itself in consummation.

-C.S. Lewis. The Weight of Glory

VI

ACKNOWLEDGMENTS

One of the pleasures of finally finishing my graduate program is the opportunity to
show my appreciation to those who were influential in my career as a graduate student. I
am grateful to many people in the academic community at Michigan State University. I
would first like to express my gratitude to my advisor and mentor, Dr. R. Michael
Garavito, whose guidance and encouragement provided me with an opportunity to grow
as a scientist. My past and current colleagues have imparted much needed support: Ms.
Amy Sharmen, Ms. Karen Poster-Verrill and Mr. Micheal Dumond shared with me their
extensive knowledge of lab techniques, Dr. Anne Mulichak helped walk me through
solving my first crystal structure, Dr. Ling Qin was always a great source of assistance in
collecting data, and Dr. Christine Harman and Dr. Rachel Powers were an invaluable
source of support both academically as well as personally. I also wish to thank my
committee members, Dr. Christoph Benning, Dr. James Geiger, Dr. Jack Priess and Dr.
John Wang, who provided a great deal of insight into my project. I could also not due
without the kindness and computer-savvy of Dr. Kaillathe “Pappan” Padmanabhan.

There are several members of the scientific community outside of Michigan State
University I would also like to acknowledge. My collaborator on the functional studies
was Dr. Karen Poon, a member of the lab of Dr. Joseph Lam, Department of Molecular
and Cellular Biology. University of Guelph, Guelph, Ontario, Canada. Furthermore, the
staff at the Advanced Photon Source, Argonne National Laboratory assisted in collecting

data and data processing. Specifically I would like to thank Dr. Stephan Ginell at the

vi

Structural Biology Center Collaborative Access Team and Dr. Zdzislaw Wawrzak at the
DuPont-Northwestern-Dow Collaborative Access Team.

I am blessed with a supportive family. I would like to thank those members who
guided my path to graduate school: my father— and mother-in-law, Dr. and Mrs. Charles
and Philippa Webb for planting the idea and my parents, Mr. and Mrs. Philip and
Beverley Cowan for their encouragement to attend. My parents have instilled in me the
importance of maintaining a high level of integrity and pressed me to use my God-given
talents to the best of my abilities. My parents-in—law have shown confidence in me even
when I have had none. My siblings and their families have been understanding of the
commitment level of the path I have chosen as well as the distance it has taken me, which
has often led to my absence in family activities.

Finally I am most grateful to my core family. When I first began at MSU, we were a
family of two. Now we are three. My daughter Elizabeth, or as we call her “Bizzy”, has
brought a new and unexpected level of joy to my life. Her sense of exploration and
experimentation warms a scientist’s heart. Ultimately, I would like to express my
appreciation for the friendship, encouragement and advice of my husband, David. I know
very few people as intelligent, logical, decisive. and devoted. His unique outlook on life
has not only influenced my graduate career but also my world-view, and I consider it a

privilege that he accepted me as his wife.

vii

TABLE OF CONTENTS

DEDICATION ........................................................................................................... iv
ACKNOWLEDGMENTS .......................................................................................... vi
TABLE OF CONTENTS ......................................................................................... viii
LIST OF TABLES ...................................................................................................... x
LIST OF FIGURES........ ............................................................................................ xi
LIST OF ABBREVIATIONS .................................................................................. xiii
INTRODUCTION
Deoxysugars and their biosynthesis ......................................................................... 1
Prevalence of deoxyhexoses ................................................................................. 1
GDP-deoxyhexoses biosynthetic pathways .......................................................... 5
dTDP-deoxyhexoses biosynthetic pathways ......................................................... 6
CDP-deoxyhexoses biosynthetic pathways ........................................................... 7
Structural biology of the deoxyhexose pathway enzymes ..................................... 8
GDP-D-rhamnose biosynthesis .............................................................................. 11
Rhamnose in P. aeruginosa LPS ......................................................................... 1 1
GDP-D—rhamnose biosynthetic pathway ............................................................. 13
CHAPTER 1: CRYSTAL STRUCTURE OF GMD
1.1. Introduction .................................................................................................... 20
1.2. Experimental Procedures ................................................................................. 21
1.3. Results and Discussion .................................................................................... 25

CHAPTER 2: CRYSTALLIZATION AND INITIAL X-RAY ANALYSIS OF RMD

2.1 . Introduction .................................................................................................... 47
2.2. Experimental Procedures ................................................................................. 48
2.3. Results and Discussion .................................................................................... 52

CHAPTER 3: FUNTIONAL CHARACTERIZATION OF GMD AND RMD

3.1. Introduction .................................................................................................... 61
3.2. Experimental Procedures ................................................................................. 62
3.3. Results ............................................................................................................ 64
3.4. Discussion ...................................................................................................... 73

CHAPTER 4: PROTEIN ENGINEERING WITH THE GOAL OF IMPROVING
PROTEIN EXPRESSION, PURIFICATION AND/OR CRYSTALLIZATION
4. 1 . Introduction .................................................................................................... 77
4.2. Experimental Procedures ................................................................................. 79

viii

4.3. Results and Discussion .................................................................................... 81

CHAPTER 5: FUTURE DIRECTIONS ..................................................................... 95
APPENDIX A: MBP-HuPLSCRl AND THE Abl-SH3 DOMAIN
A]. Introduction ................................................................................................. 100
A2. Experimental Procedures .............................................................................. 101
A3. Results ......................................................................................................... 102
LITERATURE CITED ............................................................................................ 107

ix

LIST OF TABLES

Table 1. X-ray crystal structures of NDP-sugar modifying enzymes involved in 6- and

3,6-di-deoxyhexose pathways .................................................................................. 9
Table 2. GMD X-ray diffraction data and refinement statistics ..................................... 24
Table 3. RMD X-ray diffraction data ............................................................................. 51
Table 4. Gel filtration results of GMD and RMD ........................................................... 66
Table 5. NMR analysis of GDP-D-rhamnose ................................................................. 73
Table 6. Amino acid sequence of the linker region in pMal-C2 and pMal-C 2L3 ............ 84

LIST OF FIGURES

Figure 1. Biosynthetic pathways for 6- and 3,6-di-deoxyhexoses ..................................... 3
Figure 2. Structures of 6- and 3,6-di-deoxyhexoses .......................................................... 4
Figure 3. SDR model of catalysis .................................................................................. 10
Figure 4. Lipopolysaccharide of P. aeruginosa .............................................................. 12
Figure 5. Biosynthetic pathway of GDP-D-rhamnose .................................................... 14
Figure 6. The alignment of GMD and RMD amino acid sequences from various species.
.............................................................................................................................. 17
Figure 7. SDS-PAGE of samples from a typical GMD purification ................................ 25
Figure 8. Crystals of GMD ............................................................................................ 26
Figure 9. Overall Structure of GMD ............................................................................. 27
Figure 10. Electron density at the GMD active site ........................................................ 30
Figure 11. GMD active site interactions ......................................................................... 31
Figure 12. The mechanism proposed for GDP-D-mannose 4,6 dehydratase ................... 36
Figure 13. The alignment of GMD amino acid sequences from various species ............. 37
Figure 14. RR loop of GMD and MUR] ........................................................................ 44
Figure 15. SDS-PAGE of samples from a typical PaRMD purification .......................... 52
Figure 16. SDS-PAGE of samples from a typical AtRMD purification .......................... 53
Figure 17. Crystals of PaRMD ....................................................................................... 54
Figure 18. Crystals of AtRMD ....................................................................................... 55
Figure 19. Gel filtration analysis of GMD and RMD ..................................................... 65
Figure 20. CE analysis of GMD dehydratase activity from pH 5 to 10 ........................... 67
Figure 21. CE analysis of GMD/RMD reactions ............................................................ 68

xi

Figure 22. CE analysis of the time—dependent GMD conversion of the keto-intermediate

in the presence of NADPH ..................................................................................... 69
Figure 23. HPLC purification of the GMD/RMD reaction product followed by CE
analysis .................................................................................................................. 71
Figure 24. NMR spectra for GDP-D-rhamnose. ............................................................. 72
Figure 25. SDS-PAGE of samples from a MBP-RMD purification and Factor Xa
cleavage ................................................................................................................. 83
Figure 26. Crystal structures of MBP and MBP-fusion proteins ..................................... 87
Figure 27. Model of MBP-8H3 binding domain sitezAbl-SH3 domain complex. ........... 88
Figure 28. SDS-PAGE of samples from a typical MBP-8H3 and MBP-SH3-RMD
purification ............................................................................................................ 89
Figure 29. SDS-PAGE of samples from a typical Abl-SH3 domain purification ............ 90
Figure 30. Gel filtration analysis of MBP-SH3-RMDzAbl—SH3 domain ......................... 91
Figure 31. Crystals of MBP-SH3-RMDzAbl-SH3 .......................................................... 92
Figure 32. SDS-PAGE of samples from a typical MBP-HuPLSCRl purification ......... 102
Figure 33. SDS-PAGE of MBP-HuPLSCRl/Abl-SH3 domain binding ....................... 104

xii

ADP
AGME

CDP

CF

Ered
FPLC

GalE
GDP
GFS
GMD
GMER
GTS

HEPES
HPLC

IPTG

LB
LPS

MES
MPD
MURI

NAD(H)
NADP(H)
NDP
Ni-NTA
NMR

LIST OF ABBREVIATIONS

adenine diphosphate
ADP-L-glycero-D-mannoheptose 6‘-epimerase

cytidine diphosphate
capillary electrophoresis
cystic fibrosis

deoxythymidine diphosphate

CDP-6-deoxy-L-threo-D-glycero-4-hexulose-3-dehydrase

E. reductase
CDP-3,6-dideoxy-D-glycero-D-glycero-4-hexulose-5-epimerase
CDP-D-glucose 4,6-dehydratase

glucose-1 -phosphate cytidylyltransferase
CDP-3,6-dide0xy-D-glycero-L-glycero-4-hexulose-4-reductase

fast protein liquid chromatography

UDP-galactose epimerase

guanosine diphosphate

GDP-fucose synthase

GDP-D-mannose 4,6-dehydratase
GDP-4-keto-6-deoxy-D-mannose 3,5-epimerase/reductase
GDP-6-deoxy-D-talose synthetase

4-(2-hydroxyethyl)-1-piperazine-ethanesulfonic acid
high-pressure/performance liquid chromatography

isopropyl-beta-D-thyogalactopyranoside

Luria broth
lipopolysaccharide

4-morpholineethanesulfonic acid
2-methyl-2,4-pentanediol
GMD isoform from Arabidopsis thaliana

nicotinamide-adenine dinucleotide
nicotinamide-adenine dinucleotide diphosphate
nucleoside disphosphate

nickel nitrilo-triacetic acid

nuclear magnetic resonance

xiii

OD

PCR
PDB
PEG
PEP

PL
PLSCRI

RMD

leA
leB/dTGDH
leC

leD

RMSD

SDR
SDS-PAGE
8H3

SQDI

T1 1
Tris

UDP
UV

optical density

polymerase chain reaction
protein data bank
polyethylene glycol
pentaerythritol propoxylate
phospholipid

phospholipid scramblase 1

GDP-4-keto-6-deoxy-D-mannose reductase
glucose- 1 -phosphate thymidylyltransferase
dTDP-D-glucose 4,6-dehydratase
dTDP—4-keto-6-deoxy-D-glucose 3,5-epimerase
dTDP-4-keto-6-deoxy-L-mannose reductase
root mean square deviation

short chain dehydrogenase/reductase

sodium dodecylsulfate polyacrylamide gel electrophoresis
Src homology 3

UDP-sulfoquinovose synthase

dTDP-4-keto—6-deoxy-L-mannose reductase
tris(hydroxymethyl)aminomethane

uridine diphosphate
ultraviolet

xiv

INTRODUCTION

Deoxysugars and their biosynthesis

Prevalence ofdeoxyhexoses

Lipopolysaccharide (LPS) molecules are major components of the outer membrane of
the cell walls of gram-negative bacteria and are intimately involved in host-bacterium
interactions and the viability of the pathogen. The permeability restrictions imposed by
the outer membrane contribute to the intrinsic resistance of gram-negative bacteria to
antibiotics. The LPS molecules, in particular, play a role in outer membrane stability and
impede the destruction of bacterial cells by serum components and phagocytic cells. In
addition, the LPS molecules may be involved in colonization or antigenic shifts that
determine the course and outcome of the infection.

LPS molecules share a common tripartite structure of a lipid A region, a core
oligosaccharide region and an O-antigen region. The O-antigen, which is the most
exposed region, consists of repeating units of varying monosaccharides. O-antigen can
be a linear or branched homOpolymer or heteropolymer, and can be strain-specific or vary
within a strain. A tremendous amount of effort has been exerted to determine the exact
chemical structures of O—antigen due to its contribution to bacterial virulence. In general,
the structures consist of neutral sugars, amino sugars, sugar acids and many different
deoxysugars. More recently. studies have expanded to include the enzymology involved
in the biosynthesis and assembly of O-antigen. Specifically, the biosynthesis of 6-
deoxyhexoses has received heightened interest after a few of them have shown to be

substantial LPS O-antigen components of several human pathogens. Humans lack

metabolism for many deoxyhexoses found in bacterial cell walls and would most likely
remain unaffected by inhibitors of bacterial cell wall metabolism. Since many of the
currently available antibiotics target the same cellular process or even the same target
enzyme, multiple-drug resistance of pathogens has become a serious problem. placing a
high demand on new therapeutic targets and/or strategies. The biosynthetic pathways of
6-deoxyhexoses may be targets for novel antibacterial therapeutics.

6-deoxyhexoses are formed from a common monosaccharide by replacing the 6—
hydroxyl group with a hydrogen atom, which can dramatically affect its biological
function. The committed step in the 6-deoxyhexose biosynthetic pathway is the
conversion of the precursor, the nucleotide diphosphate (NDP)-activated hexose, to an
NDP—activated 4-keto-6-deoxyhexose by an NDP-sugar 4,6-dehydratase. The resulting
NDP-4—-keto-6~deoxyhexose intermediate serves as a branching point for D- and L-
deoxysugar pathways. The intermediate may undergo subsequent steps such as
epimerization, reduction or methylation to produce mono-, di-, tri- and tetradeoxysugars
and branched-chain deoxysugars. More than one nucleotidyl sugar can be associated
with a given deoxyhexose. For instance, both dTDP—D-glucose and UDP-D-glucose serve
as precursors for L-rhamnose. However, it appears that only GDP-D-mannose serves as a
precursor for D-rhamnose. In addition to being present as constituents of bacterial LPS
molecules, deoxyhexoses can also be found in macrolide antibiotics. The general
pathways to several 6- and 3,6-di-deoxyhexoses are shown in Figure 1. The structures of

those deoxyhexoses are shown in Figure 2.

Glucose

GLK

Glucose-I-P "‘—— Glucose-S-P —’ Glucose-1-P

Ep 1 PGI 1 leA l
CDP-D-glucose Fructose-S-P dTDP-D-glucose

Eodl PMI 1 leBl
CDP-4-keto-6- Mannose-6-P dTDP4-keto-6—
deow-D-glucose deoxy-D-glucose
£1,531 PMM l leC 1 fed

CDP-3,6-di- Mannose-1 -P dTDP-4-keto-6-
deoxy-D-glucose deoxy- L-mannose

abe GDP leD
/ rms\Eep, Ered l dTDP-D-
GDP- D-mannose fucose
CDP-D- CDP-L-
abequose ascarylose GMD l dTDP L' dTDP-S-deoxy-
rhamnose L-talose
CDP-D-paratose GDP-4— keto-6-
deox -D-mannose
l M y
CDP-D- GMER/ RMD GTS \
tyvelose GDP-6-d
GDP-L-fucose eoxy-
D-talose

GDP-D-rhamnose
Figure l. Biosynthetic pathways for 6- and 3,6-di-deoxyhexoses

GLK, glucokinase, PGI, phosphoisomerase, PMI, phosphomannoisomerase, PMM,
phosphomannomutase, GDP, manno-l-phosphate guanylyltransferase, GMD, GDP-D-
mannose 4,6-dehydratase, GMER, GDP-4-keto-6-deoxy-D-mannose 3,5-
epimerase/reductase, RMD, GDP-4-keto-6-deoxy-D-mannose reductase, GTS, GDP-6-
deoxy-D-talose synthetase, PGM, phosphoglycomutase, leA, glucose-l-phosphate
thymidylyltransferase, leB, dTDP-D-glucose 4,6-dehydratase, leC, dTDP-4-keto-6-
deoxy-D-glucose 3,5-epimerase, leD, dTDP-4-keto-6-deoxy-L-mannose reductase,
T11, dTDP-4-keto-6-deoxy-L-mannose reductase, gene product of fed, dTDP-4-keto-6-
deoxy-D-glucose reductase, Ep, glucose-l-phosphate cytidylyltransferase, E01), CDP-D-
glucose 4,6-dehydratase, El, CDP-6-deoxy-L-threo-D-glycero-4-hexulose-3~dehydrase,
E3, E. reductase, Ecp, CDP-3,6-dideoxy-D-glycero-D-glycero-4-hexulose-S-epimerase,
Ema, CDP-3,6-dideoxy-D-glycero-L-glycero-4-hexulose-4-reductase, gene product of
r/bS, CDP-D-paratose synthase, GDP, guanosine diphosphate, dTDP, deoxythymidine
diphosphate, CDP, cytidine diphosphate.

OH
Me O
OH
OCDP

C BED-abequose

Me O
HO
OH
OCDP

C DP-D-paratose

e

M 0g
mm

OCDP

CDP-D-tylevose

Me 0 OCDP
HO
OH

C DP—l.-ascary0se

Me 0 OGDP

OH

HO
OH

G DP-l .-fucose

Me OH O

HO /
HO OGDP

GDP—D-rhamnose

OH
Me OH O

HO OGDP

G DP—b—(leoxy-D—talose

Figure 2. Structures of 6- and 3,6-di—deoxyhexoses

OH
Me O

HO OdTDP
OH

dTDP-DTucose

Me 0

OdTDP
HO
HO
OH
dTDP—l.—rhamnose
Me 0 OdTDP
OH HO OH

dTDP-6-de0xy-1.-talose

GDP-deox)thexoses biosynthetic pc‘ithways

The 6-deoxyhexoses originating from the GDP-D-mannose precursor whose pathways
have been studied include GDP-D-rhamnose, GDP-L-fucose and GDP-6-deoxy—D-talose
(Figure 1, middle). The enzyme GDP-D-mannose 4,6-dehydratase (GMD) is responsible
for catalyzing the conversion of GDP-D-mannose to GDP-4-keto-6-deoxy-D-mannose,
which serves as the common intermediate to these 6-deoxyhexoses. Enzymes possessing
GMD activity have been identified in bacteria [1-4|, plants [SI and animals [6, 7]. For
GDP—D-rhamnose synthesis, a 4-reductase (RMD) that targets the 4—keto group of the
intermediate has been identified in Pseudomonas aeruginosa [8] and Aneurinibacillus
thermoaerophilus [3, 8|. D-rhamnose is found mainly as a component of the LPS of
gram—negative bacteria, such as the plant pathogen Xanthomonas campestris [9| and the
human pathogens Helicobacter pylori [10] and P. aeruginosa [l l l. The 3,5-epimerase/4-
reductase (GMER) involved in the GDP-L-fucose pathway has been identified in bacteria
[12], plants [13] and animals [14]. L-Fucose is a substantial cell wall component of
bacteria, affects nodulation in rhizobial organisms [15. 16[ and is important in stem
development and strength in plants [5]. In humans, L-fucose is present in the human
ABO blood group antigens and Lewis (Le) glycans, playing a role in inflammation and
immune response [17]. The enzyme involved in the GDP-6-deoxy-D-talose pathway has
been identified in bacteria, where 6-deoxy-D-talose is a component of LPS [18, 19]. The
GDP—4-keto-6—deoxy-D-mannose reductase (GTS) involved in GDP-6-deoxy-D-talose .
biosynthesis has been characterized in Actinobacillus aetinomyeetemcomitans [18, 20].

In addition to 6—deoxyhexoses, GDP-4-keto-6-deoxy-D-mannose can also be converted to

the dideoxy amino sugar GDP-D-perosamine by GDP-D-perosamine synthetase (beE),

which has been identified in Vibrio clzolerae [ l9]

dTDP-deoxyhexoses biosynthetic pathways

The precursor dTDP-D-glucose leads to the following 6-deoxyhexoses: dTDP-L-
rhamnose, dTDP—6—deoxy-L-talose and dTDP-D-fucose (Figure 1, right). Similar to the
GDP-D-mannose pathways, these deoxyhexoses are synthesized from dTDP—D-glucose
via dTDP-4-keto—6—deoxy-D-glucose with the assistance of dTDP—D-glucose 4,6-
dehydratase (leB). The 4—keto-6—deoxyhexose again serves as a common intermediate
in the pathways to these deoxyhexoses. The enzymes involved in the biosynthesis of
dTDP-L—rhamnose have been studied extensively from Salmonella enterica [21—26] and
Escherichia coli [27-32]. The catalytic action of leB is related to GMD, dTDP-4—keto-
6-deoxy-D-glucose 3,5-epimerase (leC) performs a double epimerization at position C3
and C5 like GMER, and the direct analog to dTDP-4-keto-6-deoxy-L-mannose reductase
(leD) is RMD. L-rhamnose is a known component of cell walls of several pathogens
such as P. aeruginosa, Mvcobacterium tuberculosis, V. cholerae, Enterococcus feacalis
and Streptococcus mutans as reviewed in Maki et al. [33]. RmIB-C are also involved in
the pathway leading to 6-deoxy—I--talose. The remaining enzyme in the pathway is a
dTDP-4—keto-6-deoxy-L-mannose reductase like RmID; however, it provides the
stereoselectivity to reduce the intermediate to dTDP-6—deoxy-L—talose [34]. The gene for
dTDP—6-deoxy-I.-talose biosynthesis (product of the gene Ill) and its corresponding
protein has been identified in A. actinomyceremcomirons, one of the few species of
bacteria where 6-deoxy—L-talose is a component of the cell wall [34]. Finally, in dTDP-

D-fucose synthesis, the dTDP—4-keto-6—deoxy-D-glucose reductase (product of the gene

fcd) catalyzes the reduction of intermediate [35]. D-fucose is located in the cells and

capsule structures of a limited number of gram-negative and gram-positive bacteria.

CDP-deoxyhexoses biosynthetic pathways

The deoxyhexoses originating from the precursor CDP-D-glucose shown in Figure I
(left) are actually 3,6-dideoxyhexoses. Since 6-deoxyhexoses serve as the universal
precursors for higher reduced di— and tri-deoxyhexoses, the initial enzymatic steps are the
same. The enzymes involved in the pathway leading to L-ascarylose have been well
studied from the bacteria Yersinia pseudotuberculosis [36]. Again the committed step is
the conversion of CDP-D-glucose to the 4-keto-6-deoxy intermediate by the CDP-D-
glucose 4,6-dehydratase (E01,). The 4—keto-6—deoxyhexose is converted to the 3,6-
dideoxyhexose by sequential activities of CDP-6-deoxy-L-threo-D-glycero-4—hexulose-3-
dehydrase (El) and El reductase (E3). The 3,6-dideoxyhexose can undergo a C5
epimerization catalyzed by GDP-3,6-dideoxy-D-glycero-D-glycero-4—hexulose-5-
epimerase (Ecp) and a C4 reduction catalyzed by CDP-3,6—dideoxy-D-glycero—L-glycero-
4—hexulose-4-reductase (Emd) to yield CPD-I.—ascarylose. The 3,6-dideoxyhexose can
also undergo other subsequent epimerization and/or stereospecific reductions to yield
paratose, abequose and tyvelose. Two reductases, CDP-D-paratose synthase (encoded by
the rbe gene) from Salmonella typhi [37] and S. enterica [38] and CDP-D-abequose
synthase (encoded by the abe gene) from S. enterica [39] have been identified. Finally,
tyvelose epimerase, which is involved in the last step of the conversion of CDP-D-
paratose to CDRD-tyvelose, has been isolated from S. typhi and its structure has been
reported [40]. Each of these 3,6—dideoxyhexoses is known to be components of cell walls

and capsules of gram-negative and gram—positive bacteria.

Structural biology of'the deoxyhexose pathway enzymes

Of the enzymes involved in the deoxyhexose pathways mentioned above, several X-
ray crystal structures are available (Table 1). Interestingly, despite functional diversity
and low sequence identity, many of the enzymes exhibit strikingly similar three-
dimensional (3D) structures. The monomers fold into two domains: the N-terminal
NAD(H) or NADP(H) cofactor-binding domain and the C-terminal NDP-sugar substrate-
binding domain. The N-terminal domain consists of an alternating a/B motif known as a
Rossmann fold, which is commonly associated with dinucleotide binding. The cleft
between the two domains is the site of dinucleotide cofactor binding, sugar nucleotide
binding and where catalysis occurs. These structural characteristics place the enzymes in
the NDP-sugar modifying subfamily of the short chain dehydrogenase/reductase (SDR)
protein family. This is a large and diverse family of proteins. requiring NAD(H) or
NADP(H) as their cofactor to carry out a variety of reactions including reductions,
oxidations, dehydrations or epimerizations.

Although members of the NDP-sugar modifying subfamily of the SDR family share
amino acid sequence identities of only 15-30%. distinct sequence motifs allow for
tentative functional assignment of residues. For instance, a common N-terminal glycine-
rich region is a part of the nucleotide-binding region; this allows for the close packing of
the cofactor to the protein backbone. Also, the conserved catalytic triad of Thr/Ser and
Tyr-XXX—Lys allows one to build a model of catalysis for the initiating step (Figure 3).
Mutagenesis and kinetic studies on SDR proteins has revealed the importance of the
catalytic triad member Tyr, which acts as an active site base deprotonating the O4

hydroxyl of the substrate as the C4 position is oxidized by NAD/NADP. The catalytic

Lys lowers the pKa of Tyr by stabilizing it in its negatively charged state [32, 41]. The
role of the Ser/T hr catalytic triad member is less clearly defined. It has been proposed to
help orient the substrate in the active site and/or facilitate proton transfer [42-45]. The
oxidation at the C4 position prepares the sugar for further reactions by acidifying the
hydrogen atoms at positions 3 and 5. The 4-keto intermediate acts as a springboard for

other SDR reactions.

Table 1. X-ray crystal structures of NDP-sugar modifying enzymes involved in 6-
and 3,6—di-deoxyhexose pathways

 

 

Enzyme Substrate Product Protein Number
Family of
Species
GDP-D-mannose 4.6- GDP-4-keto-6-
dehydratase GDP—D-mannose deoxy-D- SDR 4
(GMD) mannose
GDP::;;::§°** GDP-4-keto—6-
. - deoxy-D- GDP—L—fucose SDR 1
epimerase/reductase mannose
(GMER)
dTDP-D-glucose 4,6

dTDP-4-keto-6—

dehydratase dTDP-D- glucose SDR 4
(leB) deoxy-D-glucose
dTDP-4—keto-6deoxy-
D-glucose 3,5- dTDP-4-keto-6— dTDP-4—keto-6— .new
. deoxy-L- epimerase l
epimerase deoxy-D- glucose .
(mm C) mannose family
dTDP-6-deoxy-L-lyxo- dTDP-4-keto-6— dTDP-L-
4-hexulose reductase deoxy-L- rhamnose SDR 1
(leD) mannose

CDP-D-glucose 4.6—
CDP-4— keto-6-

dehydratase CDP-D-glucose SDR l
deoxy-D-glucose
(BOD)

CDP—D—tyvelose 2-
epimerase CDP—D-paratose CDP-D-tyvelose SDR 1

(AW)

 

TYT Tyr

(0 HO

O-NDP O-NDP

 

NAD(P)+ NAD(P)H

Figure 3. SDR model of catalysis
The SDR catalytic triad member Tyr deprotonates the 04 position of the sugar as a

hydride is transferred from the C4 position of the sugar to NAD(P).

Nearly all the members of the SDR family exist as dimers where the monomers
interact, forming a four-helix bundle at the interface involving 2 helices of each
monomer. However, there are exceptions. leD exhibits a new Mgz’ldependent
dimerization mode and the GMDS seem to be following a new tetramerization pattern.
So although we can make assumptions based on the highly conserved sequence motifs of
some of these enzymes, we have already seen exceptions to the rule that makes the
structure/function relationship of these enzymes an important topic to be studied.

Furthermore, the fact that leC represents a new class of epimerases unrelated to the

10

SDR enzymes raises questions about other enzymes in the biosynthetic pathways of

deoxyhexoses.

GDP-D-rhamnose biosynthesis

Rhamnose in P. aeruginosa LPS

The focus of this dissertation is on structure/function studies of the enzymes involved
in the biosynthetic pathway of the NDP-activated 6-deoxyhexose, GDP-D-rhamnose,
from the bacterium P. aeruginosa. This opportunistic pathogen causes infection in
patients with impaired immune systems such as burn wound victims, cancer patients and
especially those with cystic fibrosis (CF). Due to its intrinsic resistance to many
antibiotics, infections by this bacterium are difficult to treat; therefore efforts have been
made to understand the factors involved in host-bacterium interactions and the
establishment of infection. The LPS, which plays a significant structural role in the outer
membrane, contributes to the pathogenesis of P. aeruginosa [46]. It is involved in
protecting the bacterium from phagocytosis [47] as well as serum-mediated killing [48].

As previously mentioned, the LPS of gram-negative bacteria is tripartite in nature: the
hydrophobic lipid A region, the core oligosaccharide region and the O-antigen or O
polysaccharide region (Figure 4). P. aeruginosa contains two variant forms of O-antigen
known as A-band and B-band. The serotype-specific B-band is a heteropolymer
composed of many different monosaccharides, while the common A-band LPS is a
homopolymer of D-rhamnose arranged as trisaccharide repeat units linked Oil—>2, Oil—>3,
Oil—>3 [8]. Interestingly, studies have shown that B-band O-antigen is absent or

expressed in smaller amounts in chronic P. aeruginosa isolates from CF patients, while

11

the level of A-band O-antigen is maintained [8]. Furthermore, there is also a correlation
between the presence of anti-A-band antibodies in CF patients with both lower
pulmonary function and increased duration of P. aeruginosa infections [8]. Other
organisms expressing the A-band O-antigen include two bacterial species known to be
pathogenic in CF patients, Burkholderia cepacia and Stenotrophomonas (Xanthomonas)
maltophilia, as well as Pseudomonas syringae pv. morsprunorum C28 and P. syringae

pv. cerasi 435 [8]. As a result, the biosynthesis of A-band LPS has gained interest as a

potential therapeutic target.
A
i B-band
l
l
O-antigenf A-band
Core i
region .
Lipid A l Outer

Leaflet

5E §§§§§§3§§§ Inner

>5 s‘gg s >§><2
Phosphollpidi §$<§Z§<>§§E $3 : ._ \
V I ‘2 2 ($223? I "‘5 Leaflet

Figure 4. Lipopolysaccharide of P. aeruginosa'

(A) Schematic of the outer membrane and (B) chemical structure of A-band O—antigen of
P. aeruginosa. Images in this dissertation are presented in color.

12

GDP-D-rhamnose biosynthetic pathway

The source of the D-rhamnose molecules found in P. aeruginosa A-band LPS is the
nucleotide-activated GDP-D—rhamnose. Markovitz proposed the biosynthetic pathway to
GDP-D-rhamnose in the 19605 [49]. The reaction proceeds through two steps: GDP-I)-
mannose 4,6-dehydratase (GMD) converts GDP-D-mannose to GDP-4-keto-6-deoxy—D-
mannose, and the reductase (RMD) subsequently reduces the intermediate to GDP-D-
rhamnose (Figure 5). The first step in the reaction is particularly important, as GMD is
also part of the GDP-L—fucose pathway. Defects in GDP-L-fucose synthesis, particularly
in GMD activity, have been linked to stem shoot development in plants [5]. In addition.
deficiencies in the biosynthesis of GDP-L-fucose in humans have resulted in the rare
immune disorder leukocytes adhesion deficiency type II (LADII) [50]. Characterization
of GMD has been reported from bacterial sources H. pylori [4], E. coli [2], Klebsiella
pneumoniae [l] and A. thermoaerophilus [3], plant source Arabidopsis thaliana (known
as MURI) [5], and mammalian sources porcine thyroid [6] and human [7, 51]. GMD
amino acid sequences are quite similar and the molecular mass of the monomers
generally lies between 40-55 kDa. Somoza et al. determined the 3D structure of GMD
from E. coli and definitively confirmed the structural relationship of GMD to the NDP-
sugar modifying subfamily of SDR enzymes [52]. However, GMDs from various species
tend to differ in their multimerization. Like SDR members, dimeric structures have been
reported for E. coli [52], K. pneumonie [1] and human GMDs [51]. Nevertheless, the
more recently published structure of GDP-D-mannose 4,6-dehydratase MURl from A.

thaliana [44] suggests it exists as a tetramer. Furthermore, there are reports that suggest

l3

that H. pylori GMD is a tetramer [4] and that porcine thyroid GMD may be a hexamer

l6I-

OH
OH
H
O O GDP-D-mannose
O-GDP OH
NA DP+
GMD
NADPH
OH
OH ,O
O GDP-4-keto-6-deoxy-
D-mannose
O-GDP CH3
N ADPH
RMD (
NADP+
OH
OH
OH
O GDP-D-rhamnose
O-GDP CH3

Figure 5. Biosynthetic pathway of GDP-D-rhamnose
GDP-D-mannose 4,6-dehydratase (GMD) catalyzes the conversion of GDP-D-mannose to

the intermediate GDP-4-keto-6-deoxy-D-mannose. The 4—reductase (RMD) catalyzes the
reduction of the intermediate to GDP-D-rhamnose.

A BLAST search with the amino acid sequence of P. aeruginosa GMD demonstrates
that it most closely matches dTDP-glucose 4,6-dehydratases (like leB), RMDs and

UDP-glucose 4-epimerases, which catalyze the interconversion of UDP-galactose and

UDP-glucose. These enzymes are also members of the NDP-sugar modifying subfamily

14

of the SDRs and contain a catalytic triad that includes Ser/Thr and Tyr—XXX~Lys as well
as a characteristic glycine-rich region at the N—terminus. Mutagenesis studies on E. coli
GMD have confirmed the role of Ser/Thr-Tyr-Lys in catalysis [52]. Like the initial
catalytic step of the SDR enzymes, Tyr deprotonates the O4 hydroxyl of the hexose ring
while NADP oxidizes the C4 position forming the 4—keto intermediate. The ensuing
dehydration step requires another active site base. Evidence suggests the second active
site base is a conserved Glu, which effectively implies that a catalytic quartet is required
for dehydratases [52]. To complete the action of oxidation and reduction, it has been
demonstrated in a bacterial GMD that the mannose C4 hydride is transferred from
NADPH back to the hexose C6 position to produce the 4—keto-6-deoxy intermediate
product [53].

Much fewer studies have focused on the second step of the GDP-D-rhamnose
biosynthetic pathway. In fact, RMD has only been characterized in the nonpathogenic,
gram-positive bacteria A. thermoaerophilus [3] and in P. aeruginosa in a coupled
reaction with H. pylori GMD [54]. The amino acid sequences of these RMDs indicate
the molecular masses of the monomers are ~35 kDa. Currently there are no 3D structures
available and the state of multimerization is unknown. A BLAST search with RMD
demonstrates its relatedness to SDR members: GMDs, dTDP-glucose 4,6-dehydratases
and UDP-glucose epimerases. The amino acid sequence of RMD most closely aligns
with that of its biosynthetic pathway mate GMD (Figure 6). The conservation of the
SDR residues is initially apparent: Gly-XX-Gly-XX-Gly and the catalytic triad of Ser and
Tyr-XXX-Lys (corresponding to Gly9—XX-Gly12-XX—Gly15 and Thr126 and Tyr150-

XXX-Ly3154 of P. aeruginosa GMD). If RMD fits into the “functional-mold” of the

15

SDR enzymes, the catalytic triad member Tyr would be expected to activate the substrate
by protonation facilitating the cofactor to reduce the 4—keto-6-deoxysugar to GDP-D-
rhamnose. This reaction mechanism has been suggested for the related enzyme dTDP—6-
deoxy-lyxo-4—hexulose reductase (leD), which is involved in dTDP-L-rhamnose
biosynthesis [26]. It is interesting to note that one of the catalytic residues of GMD,
Glu128 (GMD of P. aeruginosa), is not conserved across RMDs. In GMD catalysis,
Glu128 is proposed to abstract a hydrogen atom from the hydroxyl group at C5 of the
sugar. The C5 atom is not expected to be involved in the reduction as RMD functions by

reducing the keto group at the C4 position.

I Figure courtesy of Dr. Joseph Lam, Department of Molecular and Cellular Biology,
University of Guelph, Guelph, Ontario, Canada.

16

Figure 6. The alignment of GMD and RMD amino acid sequences from various
species

Conserved residues are highlighted in blue boxes; similar residues are highlighted in light
blue.

17

Oxm0.0mm>mzm¢¢m<¢aum4<
Hozwwfix>wom¢dzw>m¢flm>w
>mZUmHHOZOJUAmm<4JMm>¢
4mAO><mm>Z>B>3m<¢UKm4¢
<ma2>¢mmm2>>H3>4>JK¢¢U
<m42><mmwthH>m¢>Sfld>¢
dwzwwmmme>BszdwaMQ><
dmzommmm>Z>BH3IU>>Kfl>m
<IJOmmmm>Z>BH3$O>QKQ>O

and

Az>m
>IJH
JZQH
4&44
qmqh
HmAH
AKAH
HZB>
Jmh>

....mq¢mm¢0>qu>$IU¢Om

....qn43mema<4 ........
HmOHZEmemmqum>mwoxonqnmeqwmoem
Hmoezmmmmmm>Hom>m>oxmaanmaq>moaooeam
HamezHmmmmmxaom>m>642nAsmaqwmonooe>w.

Z>ZHOZ
m>ZHmm
HJZHOJ
H<>>Q<
>0Q>D<
HOQ<Z¢
Z<D>Q<
A49>OO
AOD>>O

mod

mma

A>mzzmmbm
wxeJZmQx
m<m0mm¢mm
womHmmm>4
>m<qumH¥
>mmhmmm>x
>mmmmmm><
quOm3m9¢
F>m023m<0

mm

 

:wZZfiXUdmamugflBUOXme >>mmmm>ma>m4mm¢mm....Oddfln
WWmEmmJOZmflm>mmmmdeZ$wmmmmOHBQHmUQQZWQ....m>4ud
finmemmHamNSHmJ<4m¢>OL>>DOmmH>thUmmOm¢....¥J¢UJ
LWWmQImmHPMmOQ.mm..EmubzmmmO<O>W¥>BmOmQHhmmm>¢md
flWmmmwmmBFN¥OmHmO..>x $Jmmhm¢0>mx>mZHAOUB..¥>43
wwwmmzmmH7NZOmHmJ..>m >4mmhm40>mmHMMmQOJH..m24wd
.mmemeHPM¥OmHmO..>JL%JmmBm¢O>mmfixme04m..demd
fwmmmwmmFPNXOmEm>..>x mzmmmmewmxdflmme....ma4mfi
mmmmmwmmFZMDOmméo..Hd m2mmbm40>m¢fimmmm0....mH4mA
and ova mna man man
IBH¢< I>>IHQ4¢¢UJ¢OWJOmdeHDUm>me¢Im ...........
mmxdd ImH>meHDWH>¥¥>mOmQZHQdeZW.>Zm ...........
>BOO¢ $H>¢Dm.meZJDOAmDmeJDWmIQ>...m ...........
ImOdd Z>>mflfixH>QH3mmqmm<QBJQ4>IquJ4¥Z>ZEQQH»HEZ
IWOdU Z>Hm9mx>mZHHK>AUHWDBJQU>$J¥EZUMHmdomzqumm
Im04< ZWHmPNXBhflHAmHJmeQHZ.medmmmmxmmmmdflqumfl
ImZ<O Z>>mQflO>MMAHmBAZMBQmJDOWIth.XQZUBmmflowafl
mmOdO Z>>mQ.OHXXHHd>dmmdDBJQOm>ABH¥QX ........ dmm
mmO<4 Z>>mOQO¢¥H>€m0>m04042900m>0HQOm ...... Hoamm
we on we an

A>¢OAIxO>mOm43

HZ>ZHaHmBmmHJU>>M>OxMJQded

WmOSOOb H

         

>mquBz oooooooooooooooooooooooooo
HJ¢memm>>BmQ<xm<HHmmmommmOzzmde
HA¢>ZmEmUOm>memOmm ..............
HQ4H¥WXZ .........................
HJ—dN/me ...........................
HJ—desz ...........................

mean

1...
u .

EWUQHOG>EGZ¢ ,..... HQHNHNmE ...........................
....dxdmzzmmhwm>m>ZOmHQ224J>xO>mO<>fl fiddmz ............................

a
>mme2mmm¢mmxme>m>wxmnaEm4q>manoa>m“

dmzmbflmmmm<>dom>m%0&MJJ4¥¢J%QUDOOEH
mv . .

on

ON

 

>1H4mmrfinz ..........................
ca w

DEM MdOZHmMm.m
92m 0mm402mmmfi.<
Q2“ dmOZHUDmm4.m

n20 mHmmOQHmdflfl
QED Zdlbfl

Q80 HMOQNQ.H

920 HAOU.M

DIG OMM£OSMMNH.¢
DID (mOZHODmM¢.m

92m M<OZHmNm.m
aim Camdozmmmfi.4
92m dwOZHUDmm<.m
DID mHmmOQHm¢H¢
GEO Zdlbm

DID Hmonflm.m

GEO HQOU.M

DIG ONM4OSKMNH.¢
DIG 4mOZHODNM¢.m

GEN M<UZHmMm.m
QZm OfimememB.<
92m 4mOZHODmm4.m

DIG mHmmOQHfl¢N¢
920 Z‘lbm

DID H“OQ¥A.N

GEO HA00."

DID OMN¢O£¢NMB.¢
n20 ‘mOZHUDNN‘.m

18

..................... dommeAHmmdmxmqu.mmM.meoxamxzmwHABm>mmmmeszq
................. mmm>mwm30ququm0xHBH.mmx.OBBQIdmdmmUm>mmOm4mmzmdmDO
mgmOO&¢DEWU¢D>J>¥mmxfiqmqoma>22x>dxmm0>.Ommioq>mx4xmdDUOQZO>mdQMm>mOQH
............ mwdzmZHmde>Q¢m>me>demd>.max.quonXBUDGOAmD>mBmmwwx40>
..m>mZKABNOUQOAWJZXOUquawmdzox>qm¥daw.mm>.Udmxmdwemawqqao>mbmmmwmmoH
.mquH¢>Q>Ummxaqmmxxddma024>2mm>zmeBH.max:qummdx8m00445m>m<mmmwmm0>
............... xaoxmHJDZOZQ>EZB>QOmmmemmZ:0A¥ZH¢Xm<QUAA>D>mdQMmmme>

................ >¢mH>mmmzwzqddeBQquB.xmm0940m24mzzwmdmmHamm—ZmDmDH
................. mmw>mmao4m>EEqumemH.mmKBOA>mO¢x¢mZOAQ>Q>m¢mmmmdeH

wan man man wow wow
rHs>mH> oooooooooooooooooooooooooooooooooooo mmmq<zqogngqgothmmomuquuwmwm
moH‘DHx 000000000000000000000000000000000000 >Z¢Z4AAJDQ>OOHmBOHomU>Zo o o o>>
>qum oooooooooooooooooooooooooooooooooooo >OdH04qquqmmeom00mU>Zoooow>
m>>ox3 .................................. zqwqwomm>oammm>emwmme<> ..... >w
B>m>xoew ................. xumo>mzmzxom3>HonHmammxm>mmm>mm>me¢H ..... >m
m>>HfiUBMJImJZJOA<Z¢m¥mQ>m¥HAUHWXHOBZOmMJQHUHmmkm2x>mDm>mEB¥DHdH ..... >w
€HH>DOQ¥ ....... >OQ¢DIOB>m>>HUfimm>0fi0mmmdxHOA04¢<Zm>mOm>m>O>UB¢H ..... >m
>>>omw .................................. zqw>mwm¢Hmummm>mmzmwe<H ..... >>
xqmom» oooooooooooooooooooooooooooooooooo ng>mmm¢HOUEDm>FHB>wB<> IIIII >>
men men omu

m0m0<qoqqd>>¢ou>flm>oqmmu440420>MQOQ40¥0<EHA<H0m4mm¢B<QOmOm<0H .
DOBxUNOm443><O>HQm>QBmmw>¢mQ70>xHHQMOmemHQ>HOM<mQOB>hUAmOUmOH1W
€0¢m0mmaama><md>:0>aqmfiww>DHQO>mAmZ<OQOOXZm<H0m<¢m<H>th¥OUmOHa.
DQmmmOOAZAZde>N.U¢mORfiwmdodzudmax.BOJU>¥Hqu<mHH¥mB>hszmmmmmq
QMQWQZOJZQZE¢M>MQx4mURAmX¢QAZUAmmU.meUA>H34>mmmmeh>hz<wmmmmmH.
QOmBmeq24$ZEx>>Qx¢mwzmwxdodzaqwqu.QBJZ><H¢W4<mHmeh>bhm0m>mmmWT
OmmOMOOJZZZOEx>y.x¢IUZ&nJmDZzUA>JU.mmaOO<HZ<H¢mEHmmB>thOmmmmmT

moaxmooqzqzzom>s.x<mozmwo<oa204mqm.genomeHm<>ooemee>mmHommmmmxx.nmeu
aHomm

      

  

u. .n

QQéXQOOJZJZZ<W>WQ04mOZFJ¥¢D>ZUAEAW.OOXOAXHm4><QH>MmH>hmH0mgammmE
ova wan on“ man mom mad and

t

 

92m mfiwszwm.m
92m OMMQOEMNEB.<
QSm 4mOZHODmm<.m
DIG mHmmOQHmddd
n80 debm

GIG Hmonhm.fl

020 HAOU.M

GEO ONӢOSMNNB.4
DID QmOZHODmM4.m

DEM m<OZHmNm.m
92m Ofimflozmmm8.4
92m dmOZHmem<.m

DID mHmQOQHmfldd
GEO 2¢Sbw

980 HmOAHm.N

020 HQOU."

QED OM”402¢NNB.4
DIG dmOZHGDMfl4.m

Q2“ W<OZHmNm.m
GEM 0mm¢02mmmfi.<
92m 4mOZHODmm4.m

980 mHmmOGHm4M¢
GIG ZKSDI

n20 HmOANA.fl

DEG HA00.”

QED ON“402¢MNB.¢
920 ‘mOZHODMM¢.m

6.2.88 .w 0.53..—

19

CHAPTER 1: CRYSTAL STRUCTURE OF GMD'

1.1. Introduction

GDP-D-mannose 4,6-dehydratase (GMD) is a member of the NDP-sugar modifying
subfamily of the SDR protein family. As a member of this subfamily, GMD binds its
cofactor NADP(H) in the N-terminal portion of the molecule where a common glycine-
rich region is present. Members share a handful of highly conserved residues that include
a catalytic triad of Tyr-XXX-Lys and Ser/Thr that are all important for catalysis.
Although the amino acid sequences of members of this subfamily are significantly
different, the three dimensional structures are quite homologous. Closely related
enzymes for which structures are known include UDP-glucose epimerase (GalE) from E.
coli and human [45, 55-62], GDP-fucose synthetase (GFS) [63] and dTDP-D-glucose 4.6-
dehydratase (dTGDH) [21, 42].

Somoza et al. [52] determined the three-dimensional structure of GMD from E. coli
and confirmed the structural relationship of GMD to the NDP-sugar modifying subfamily
of SDR enzymes. However their work suggested that active GMDs exist primarily as
dimers. In contrast, Mulichak et al. [44] showed that the structure of the MURl isoform
of GMD from A. thaliana was a tetramer where the NADP(H) binding site is intimately
involved in creating the subunit interface. The question thus arises as to whether
bacterial GMDs differ from eukaryotic GMDs in terms of their active oligomeric state
and, perhaps, the regulation of enzyme activity. Reported here is the crystal structure of a
GMD from the bacterium P. aeruginosa in the presence of the ligands NADPH and GDP.
Our analysis again confirms the structural homology between the GMDs from all three

species, but reveals that the GMD from the bacterium P. aeruginosa exists as a MURI-

20

like tetramer. Based on the conservation of subunit interactions and sequence in GMDs,
evidence suggests that the tetrameric form of this enzyme may be its active oligomeric

state in prokaryotes and eukaryotes.

1.2. Experimental Procedures

1.2.1. Cloning and Expression of GMD. The gmd gene, originally isolated by
Currie et al. [64], was amplified by PCR while incorporating a BamHI restriction site
over the natural start codon and a Hindlll restriction site over the natural stop codon. The
new PCR fragment was ligated into the pQE30 vector (Qiagen), placing a 6x-His tag at
the N—terminus. A glycerol stock of E. coli M15 cells harboring the recombinant plasmid
was used to inoculate a starter culture of 50 ml Luria Bertani (LB) broth. Following
cultivation at 37°C for 16 h, the starter culture was transferred to 1 L LB broth. Once an
optical density at 600 nm (ODOOO) of 0.5-0.6 was reached, the cells were induced with
0.25 mM IPTG and grown for an additional 16 h at room temperature before harvesting
by centrifugation. Cells were re-suspended in Lysis Buffer (50 mM HEPES, 300 mM
NaCl, 5 mM imidazole pH 8.0).

1.2.2. Fermentation of GMD. A low temperature glycerol stock of E. coli M15 cells
harboring the recombinant plasmid was used to streak M9 plates containing 2.5% (w/v)
glucose, 25 [lg/ml kanamycin and 100 ug/ml ampicillin. After a growth period of about
16 h at 37°C, an individual colony was chosen to inoculate 5 ml of starter medium (M9
salts plus 2.5% (w/v) glucose, 2 mM MgSO,, 25 ug/ml kanamycin, 100 ug/ml
ampicillin). The starter culture was placed in an incubator for 12 h at 37°C at a shaker
speed of 250 rpm. The culture was then transferred to a 250 ml Erlenmeyer flask

containing 100 ml of growth medium (22 mM KHzPO4, 42 mM NazHPO4, 8.3 mM NaCl,

21

18 mM NH4CI, 1 mM MgSO... 2.5% (w/v) glucose. 3 [1M thiamine-HCI, 25 ug/ml
kanamycin, 100 ug/ml ampicillin). The 100 ml culture was then placed back into the
incubator at 37°C at a shaker speed of 250 rpm until an OD,00 of between 2.0 and 3.0.
The 100 ml culture was used to inoculate l L of fermentation medium (33 mM KZHPO4,
10 mM citric acid monohydrate, 21 mM H2504, 0.3 g ammonium iron 11] citrate, 25
ug/ml kanamycin, 100 ug/ml ampicillin, pH 7.0) in the fermentation vessel hooked to the
BioFlow II Fermentor (New Brunswick Scientific). Using fed-batch fermentation, the
conditions that produced the best protein yield consisted of growth for 12 h at 37°C in the
fermentor followed by induction with 0.5 mM IPTG and a 12 h post induction growth
period at 25°C. The dissolved oxygen was kept constant at 10%, pH at 7.0 and the max
rpm at 750. Cells were harvested by centrifugation and re-suspended in Lysis Buffer.
1.2.3. Purification of GMD. Cells were lysed by sonication. The lysate was cleared
by centrifugation at 10,000 x g for 20 min and loaded onto a column of 4 m1 Ni-NTA
resin (Qiagen). The column was washed with Wash Buffer (50 mM HEPES, 300 mM
NaCl, 20 mM imidazole pH 7.5) and protein was eluted with Elution Buffer (50 mM
HEPES, 300 mM NaCl, 200 mM imidazole pH 7.5). Further purification was carried out
by FPLC using a HiTrap’“ Q anion exchange column (Amersham Pharmacia Biotech)
equilibrated with Buffer A (20 mM Tris pH 8.5). Protein was eluted over a 40 ml salt
gradient using Buffer B (20 mM Tris pH 8.5, l M NaCl). Fractions were analyzed by
SDS-PAGE. Unless otherwise noted, protein samples for SDS-PAGE analysis were
denatured using SDS and equal volumes were loaded onto a 4-12% acrylamide gradient

gel. Those shown to be homogenous were pooled and concentrated to 10 mg/ml.

22

1.2.4. Crystallization. Purified GMD was used to set up crystallization screens
(Crystal Screen 1 and 2 from Hampton Research) using the hanging drop vapor diffusion
method. Prior to data collection, crystals were looped directly from the drop and flash-
frozen in liquid propane.

1.2.5. Data Collection and Processing. X-ray diffraction data were collected to 2.2
A on a MAR CCD detector at the Advanced Photon Source beamline 5-1D (DND),
Argonne National Laboratory. During data collection the crystal was held at 100 K in a
cryostream and radiation was used at a 1.0 A wavelength. The data were processed using
HKL version 1.6 software [65]. Data collection statistics are presented in Table 2.

1.2.6. Structure Determination and Refinement. The structure of GMD was
determined by molecular replacement using the AMoRe programs [66] from the CCP4
suite [67], using a search model based on the known structure of MURl (PDB code
lN7H) [44]. Using a dimer as a search model, clear solutions for a tetramer in the
asymmetric unit were obtained. CNS v1.1 [68] was used to perform several rounds of
simulated annealing, followed by simple positional refinement and finally individual B-
factor refinement. Manual building was carried out between each round of refinement
using 2Fo-FC and FO-Fc maps in CHAIN [69]. Throughout refinement non-
crystallographic restraints were imposed on each of the four molecules in the asymmetric
unit. The final model (R,,,Cu,,=l7.3, R,,,_.c=l9.5) consists of a GMD tetramer, 693 water
molecules and four molecules each of NADPH and GDP, despite the fact that no ligand
or cofactor was added to the crystallization drop. There is clear electron density for

residues 3-323 of 323 possible residues in each monomer. The final model has good

23

stereochemistry as determined by PROCHECK [70]. The refinement statistics are

outlined in Table 2.

Table 2. GMD X-ray diffraction data and refinement statistics

 

Space group P3221
Resolution range (A) 30.0-2.2
Cell parameters (A, °) a=b=125.7, c=220.0
(1:13:90, y=120

Unique reflections

Working set 1 10,070 (92.7%)

Test set 5,398 (4.9%)
Rmerge 3.9 (7.6)
Completeness (%) 98.3 (92.8)

a
Rfactor 17.3
Rfreeb 19.5
Residues 1282/ 1 292
Water molecules 693
c

rmsd

Bonds (A) 0.005

Angles (deg) 1.3

Dihedrals (deg) 23

lmpropers (deg) 0.7
Ramachandran

Most favored (%) 90.2

Allowed (%) 9.8

 

aR-factor = 2|FO - Fcl/EFO, where F0 and PC are observed and calculated structure
. b . . .
factors, respectively. R-free IS the cross validation R-factor computed for the test set of

. c . . .
reflections. rmsd IS the root-mean-square dev1ation.

24

1.3. Results and Discussion

1.3.1. Expression, Purification and Crystallization. A typical protein yield from
the shake-flask method was 3-4 mg of purified protein per liter of cell culture.
Fermentation increased the protein yield over the shake-flask method by 10—fold.
producing about 30—40 mg of purified protein per liter of cell culture. The purification
protocol described above produced protein of a quality that could be crystallized. SDS-
PAGE gels of samples from a typical purification are shown in Figure 7. Protein after a
Ni-NTA column step did not produce crystals, so the next step of purification was added.
Bipyramidal crystals were grown in a drop containing equal parts protein and well
solution (50% 2-methyl—2,4-pentanediol, 100 mM Tris pH 8.5, 200 mM NH4H2PO4) and
equilibrated against a reservoir of well solution (Figure 8). The resulting crystals (0.4 x
0.4 x 0.3 mm) belong to the trigonal space group P3221 (a=b=125.7 A, c=220.0 A,

a=B=90°, y=120°) with four molecules in the asymmetric unit.

 

Figure 7. SDS-PAGE of samples from a typical GMD purification

(A) load (L), fractions (F) and pool (P) from the Ni-NTA column; (B) load (L), fractions
(F) and pool (P) from the anion exchange column. Molecular weight marker (M)
standards shown in kDa to the left of each gel.

25

 

Figure 8. Crystals of GMD
Crystals were grown in 50% 2-methyl-2,4-pentanediol, 100 mM Tris pH 8.5, 200 mM

NH4H2P04, 5 mM GDP-D-mannose, 2 mM NADP. The larger crystal is ~08 mm across
at the largest point and the smaller crystal is ~0.35 mm across at the largest point.

1.3.2. Overall Structure. As is characteristic of members of the NDP—sugar
modifying SDR subfamily, the GMD monomer folds into two domains: the N-terminal
cofactor-binding domain and the C—terminal substrate-binding domain (Figure 9). The
larger N-terminal domain consists of an alternating a/B motif of seven fi-strands in the
order 3-2-1-4—5-6-7 flanked by (it-helices, yielding a modified Rossmann fold, an element
generally associated with dinucleotide binding. Common in this subfamily is the
transition into the C—terminal domain following [36, providing an extension of an a-helix
and two B-strands. The chain then turns back to the N-terminal portion, adding another
a—helix and the seventh N-terminal B—strand before returning to and completing the C-
terminal domain. The smaller domain consists largely of three (it—helices with two sets of
mixed parallel and anti-parallel B—shects. The cleft that exists between the two domains
is the site of dinucleotide cofactor and nucleotide-sugar substrate binding and is where

catalysis occurs.

26

 

Figure 9. Overall Structure of GMD

Ribbon representation of the GMD monomer (A), the GMD dimer highlighting the
tetramer interface (B), and the GMD tetramer (C). For each drawing [Ii-strands are
colored in blue, and the areas colored in yellow represent the regions involved at the
tetramer interface. In the active site of the GMD monomer and dimer, the cofactor
NADPH and substrate GDP are shown in backbone representation. Only the cofactor
NADPH is included in the tetramer for clarity. Arrows in the GMD dimer indicate the
RR loops, and the asterisk highlights the close approach of the cofactors from each
monomer to one another. The asterisks in the tetramer highlight the four-helix bundle,
the dimerization mode common among SDR enzymes.

27

GMD crystallizes with four molecules in the asymmetric unit, forming a
homotetramer (Figure 9). Within the tetramer, two GMD monomers interact to form a
four-helix bundle at the interface involving (14 and (15 of each monomer. The contacts
between the long helices are mainly hydrophobic in nature, with a few hydrogen bonds
between Asn163 and the main chain carbonyl of Arg147*, as well as Glul66 to both
Tyr145* and Arg147* (*denotes residue from opposite monomer). Several closely
related enzymes, as well as nearly all the members of the SDR family, share the four-
helix bundle dimerization mode. GMD deviates from the typical homodimeric structure
as two of these “SDR dimers” sandwich together to form a tetramer. GMD is only the
third enzyme of the NDP—sugar modifying SDR subfamily to be observed as a tetramer
next to its plant homolog MURl [44] and the recently published structure of tyvelose
epimerase [40].

The tetramerization of GMD results in adjoining of the cofactor binding sites at the
interface such that the adenosyl phosphate moieties of bound NADPH molecules fall
within 7 A and 7.5 A respectively for GMD and MURl. Both GMD and MURl share a
flattened ellipsoidal shape (~ 95 x 75 x 60 A and ~lOO x 74 x 57 A respectively). In
contrast, tyvelose epimerase, which oligomerizes along the analogous surface, forms a
less compact tetramer (~100 x 110 x 60 A). Consequently, the tetramer interface of
tyvelose epimerase is less extensive and the adenine rings of bound NAD molecules are
separated by 11 A.

1.3.3. Cofactor Binding Site. There is well—ordered electron density in the active
site to unambiguously place an NADPH molecule in each monomer (Figure 10). The

nonplanar electron density corresponding to the nicotinamide ring suggests that the

reduced form of the cofactor is present. NADPH is found in the cofactor-binding sites of
the crystals when no cofactor was added to the crystallization conditions, suggesting tight
binding of the cofactor throughout purification. The tight binding of reduced form of the
cofactor is consistent with the characterization of both E. coli GMD and Y.
pseudotuberculosis CDP-D—glucose 4,6-dehydratase, which reveals that they bind their
reduced cofactor more tightly than they bind the oxidized form [52, 71]. This is clarified
with the catalytic mechanism (discussed in detail in section 1.3.5.), which displays an
intramolecular hydride transfer from the C4 carbon on the mannose ring to the C6
carbon, effectively regenerating NADP in the process and allowing it to remain bound
[53]. However, NADPH is seen bound in the cofactor-binding site even with an
abundance of NADP added to the crystallization conditions. One potential explanation
for the presence of the reduced form of the cofactor is the inadvertent reduction of the
dinucleotide by the buffer. Tris buffer has demonstrated reducing power in the case of
the UDP-glucose epimerase/NAD complex in the presence of UMP [72]. A schematic
diagram of the interactions of NADPH with the surrounding residues is shown in Figure
11.

NADPH binds in an extended conformation with the pyrophosphate positioned
adjacent to the positive dipole of (11 in a manner consistent with dinucleotide binding in a
Rossmann fold. The characteristic glycine-rich fingerprint sequence of Gly9-XX-Gly12-
XX-GlylS in this region allows for the close packing of the cofactor to the protein
backbone. This leads to the formation of hydrogen bonds of the pyrophosphate to the
main chain amide nitrogen atoms of Gln13 and Aspl4. Further hydrogen bonding occurs

from the pyrophosphate to Ser85.

29

 

«mus—0

maps»;

33:
one;
. , ham.» (will?
flficu
36.2

on Es...

353

.86 3:8 05 wfieaaobsm 8562 SE» 3953 ADO use. 93:93 Ema—<2
8: 952. 920 2: «a .335... 5.58:— .S 0.33%

30

mm 32

3 O ..
SN 6 +13: Zfl ~12

v\ + meow ......o
.. o C . x V
V 4 w 5.3 £2: / . . .oLf

WN—EO 0h_.r_m( mmamx‘

»
Emmi
025 3:3

.oEzw a ma AOHZC onEoo _MDE\omo:Em€-o-mQO\me<Z mom: 5 @2358 we? omega—z .020 5:3
omofiafiéiow 03:33 use ImQ<Z LowofloQ 2: 5233 meson sumac»: E828 of («o :5me ochfionom
28:228.: 8% 9:3“ GEO .: 25w:—

 

31

The nicotinamide moiety is bound in the syn conformation and may be stabilized in
this orientation by hydrogen bonding between the carboxyamide nitrogen and the
pyrophosphate. This conformation is consistent with other SDR enzymes, allowing a B-
side hydride transfer during catalysis. ln GMD the conformation may be further
stabilized by hydrogen bonds between the carboxyamide group of the nicotinamide
moiety to the main chain amide nitrogen of HislSO and the Arg185 side chain. The
nicotinamide ribose hydroxyls are within hydrogen bonding distance to the catalytic
residues Tyr150 and LyslS4, interactions that are highly conserved in SDR enzymes.

The adenine ring of the cofactor is largely coordinated by the negatively-charged side
chain Asp59, which makes a potential hydrogen bond to the adenosyl amino group, while
the main chain amide nitrogen of the subsequent residue, Met60, hydrogen bonds to the
N1 nitrogen of the adenine group. Both interactions are highly conserved among SDR
enzymes. Further coordination of the ring nitrogen atoms occurs through hydrogen
bonding with water molecules in the adenine ring pocket. The Arg36 side chain
hydrogen bonds to the 2’ phosphate group via both NE and NH1 nitrogen atoms. The
position of this arginine is conserved among GMDs and other NADP-binding SDRs such
as GFS, and is suggested to be responsible for discriminating between NADP and NAD
[63]. The hydroxyl of the adenosyl ribose is coordinated by the Thrll side chain and by
the main chain amide nitrogen of Gly12. The main chain amide nitrogen of Ala83 is
within hydrogen bonding distance of the ribose ring oxygen. Similar interactions with
the adenosyl ribose are seen in MURl and GFS [44, 63].

Most intriguing about the cofactor binding site is the involvement of the adenosyl end

in the tetramer interface, a feature also seen in the plant homolog MURl [44]. The RR

32

loop, a segment of nine residues (Arg35-Arg43), stretches into the neighboring monomer
making not only protein-protein interactions, but also contacts to the neighboring
cofactor. Protein-protein interactions include Arg35 hydrogen bonding to Ser85* and
Glu188*, Ser38 to Trp42*, and Arg43 to both Ser37* and Ser38*, as well as to the main
chain carbonyl of Ser38*. Protein-cofactor interactions include hydrogen bonding of
Ser37 to the neighboring adenosyl 2’-phosphate via the main chain nitrogen atom as well
as the side chain hydroxyl group. Arg35 also hydrogen bonds to the pyrophosphate of
the neighboring NADPH, further tethering the tetramer together. Furthermore, the Arg36
side chains from each monomer are involved in a parallel stacking arrangement between
the two adenine rings. They also help to coordinate the 2’-phosphate of their own
NADPH in a way similar to that observed in GFS and MURl.

1.3.4. Substrate Binding Site. The electron density corresponding to the substrate
reveals that GDP is present in the active site of each monomer (Figure 10). Attempts to
replace GDP with GDP-D-mannose by adding excessive amounts of the natural substrate
to the crystallization conditions have been unsuccessful. However, mannose could be
modeled in using the NADPH/GDP-D—rhamnose/MUR1 (PDB code 1N7G) complex as a
guide. Hydrogen bonding and van der Waals interactions were optimized while holding
the GDP moiety constant. Binding interactions of the GDP-D-mannose with GMD are
depicted in Figure 11. The orientation of the O6 hydroxyl of the GDP-D-mannose was
chosen based on its ability to make potential hydrogen bonds to Thr126, Ser127 and
Glu128. In the crystal structure, the presence of a water molecule close to the GDP-D-
rhamnose C5 atom that makes hydrogen bonds to Ser127 and Glu128 further supports

this proposed rotamer. Based on this model, both catalytic residues Thr126 and Tyr150

33

could hydrogen bond to the hexose O4 hydroxyl. Further coordination of the 02 and O3
hydroxyls and OS of the hexose ring occurs through the side chains of Arg185, Tyr150,
the main chain carbonyl oxygen of Ser85 and through water mediated hydrogen bonds.

The GDP moiety, which is completely buried in the small domain, is in the syn
conformation. This is an unusual conformation for this nucleotide and may be related to
substrate recognition. GDP is stabilized in this orientation by an intramolecular hydrogen
bond between the guanine amine nitrogen atom and the phosphate. The pyrophosphate
bridge of the GDP moiety abuts against the positive dipole of a6, in a manner similar to
that of dinucleotide binding in a Rossmann fold. This allows for hydrogen bonding to the
main chain amide nitrogen of Vall90. Further contacts to the GDP moiety include
Asn179, Lysl93, Arg218, Arg279 and Glu282, all highly conserved residues among
GMDs. The ability of Arg279 to hydrogen bond to both phosphates is a feature
conserved among several SDRs, although the exact position of the arginine is not
conserved. Furthermore, the hydrogen bonding of Glu282 to the ribose hydroxyls is also
observed in UDP—binding SDRs I43, 57].

1.3.5. Catalytic Mechanism. The proposed catalytic mechanism for GMD occurs
via a three-step process (Figure 12) [53]. The first step involves a hydride transfer from
the mannose C4 to the C4 of the cofactor. An active site base deprotonates the O4
hydroxyl to yield the GDP-4-ketomannose intermediate (b). The second step involves
the elimination of water from C6 to yield the GDP-4-keto-5.6—ene intermediate (c). In
the final step, the C5-C6 double bond is reduced as the cofactor returns the hydride to the
C6 position yielding GDP-4-keto-6—deoxy-D-mannose (d). The SDR enzymes share only

a few conserved residues that include the catalytic triad Tyr-XXX-Lys and Ser/Thr that

34

are important in catalysis (Figure 13) [73]. Additionally, a highly conserved Glu among
dehydratases has proven to be important in the dehydration step [42, 52]. Mutagenesis
and kinetics studies on E. coli GMD [52] as well as other SDR enzymes have supported
the roles of Tyr, Lys, Ser/Thr and Glu in the reaction mechanism. The GMD reaction
resembles that of other nucleotide hexose dehydratases, such as CDP-D—glucose 4,6-
dehydratase and dTDP-D-glucose 4,6—dehydratase that have been more extensively

studied [27, 29—32, 71 [.

35

2985.95

mcoéhéoxédow 8v Ino<z

8:93ch

2988.25
omoccmEmeédoO 3V

\JJMMIVMOOO

i

.J
Nam.

cocosuoc

noob mmoccmE-o->xooo
fo 0 espresso E
Ill
0 IO
IO
30-0
IO
All IO 0
mmoccwEé- m
:0 new:
L ..
A
+no<z
cozngo

ADV omoacmEéAxooeéouox

-viDO 22» 9 22.262 .bficc .98 AB 0862585 o:o-o.m-8§-v 2.: BE cozmavxcou Am: 0362535
987v 2: 62608 8 2v omoecmEédQO mo coca—cg ”38m 8:: BE BEZB on :8 @885 2:.
83.26.396.96 omeeuafiéiA—U .8.— Bmeaea Swizz—8:. 2:. .2 «But.—

36

Figure 13. The alignment of GMD amino acid sequences from various species

Conserved residues are highlighted in blue; similar residues are highlighted in light blue.
Secondary structural elements are marked according to P. aeruginosa GMD. Residues
involved in the dimer interface are boxed in copper, residues involved in the tetramer
interface are boxed in yellow. An asterisk marks catalytic residues Thr126, Tyr150, and
Ly3154.

37

mmBHMKOQHm0>¥0wqmmhm¢0>mx>mZHQOUH.
mfizmzoafimd>m0>dmmkmfiomeBKMmeAH.
QBFMXOQHmO>dO>QmmHm¢O>mmBMMmQOdh.
HQBFWMOQ. mmBmOmZNWWOdOM¥M>FMOmDHBIW
mu<nzomm meemx0a2m>>x0azmmmmdowmx<omxmz.
mmfimemm afizmaomwdoHQUhZMWFWflONhKmeme.
ova owa oma ova ona oua
I

a

           

Z><mAZ>
Z>¢mdz>€
Zodwzo>m
ZU¢LJU><

-

3|
'8

 

QXEZUWHE¢OmZ¥>Qmm
Jmmmxxmommqaquxo
JEh.¥&ZUHE&QO>HIQ
AXZQ¢M2>ZIQQHVHIZ

>JUBQQHJDCw
HJZWmQPZQDw
BAZMHQWJDDa

  

c.mH¥>mm0¢OAZ>HmHm¥>mZHH
. m>¥>$m0<<qszm9m¥BB<H
m><>xw2d042w>mam0>mma
m><>mm0<¢qz>>mom¥H>DH

 
  

      
    

mB¢>hmO¢OAZ>>QOOH¥MH 4>JmIJQHJDOm JHH¥D¥ ........ Jim
. m¢0>mm0¢¢dzw>mOQO¢¥H> >OHQUWHO ...... Qmm
ooa om on or om om
dAnquddnflfiqqdddnfiq .Allll deDDHQddNDfiq AIIIII dd
.8 a «a a

 

H0m>m>0¥mddmmdqmeQOOBHOBHA¢>zmmMUZNUDUmUM<...mmomémdmdz..
dom>m>042dqym<a>m00009>OBHA¢H¥mKZ .........................
H0m>m>0xmqqmm<qmeQOOB>OBHA4>wa .........................

J0m>m>0xmaqqxdd><OQOOBHUB>q¢mmBE ..........................

on ON OH H
Sad ill 3333 i
as a 8 a

 

 

nannouanau¢
afluosu.<
onoaamuoa.m

$38392
afluosu.¢
oaoaamsuou.m

adadm
auoflhm.m
“HOU.M
aanaouanaud
canonu.¢
uaouamfluoa.m

38

mnowuomaounH Hofimuuma

............. <ZQZBKEQm>Q¢$>Emm>JmQh<>.mmz324¥0M<¥FUDOOJmD>mB
.mwmzmaHflOUDOQquxoumXQoymJZQ¥>qdeQ>.mm>30am¥m<¥fimm©gqqa>mfi
mm4<H¢>Qwomquqmm¥x<¢MJDZ¢>me>EmKQBH.mmxBDAXmE¢¥BmOOJQBm>mQ
.mOOM¢OZ>O<Q>Q>Mmmx<dmgflma>zzx>4¥mm0>.Oaxzwd>m¥<¥m¢QOOQZQ>w<
.............. quxmHADEQZQ>ZEB>QOmmmH¥mmZSUQXZH<Mm<QOJQ>Q>m<
................ mMm>mmJQ<m>EZMHJmQQmB.MQXZGA>NO<K¢QZOAQ>Q>W¢

own oam oom omn
dddflqddUHQdANQdddflqugq ddfiqqddfiq .Llllll
c_5 as N_u

amwwxqa>e>m>xo ................. meomo>mzmzxomz>aeonmqamxm>amm>mm>mwed
amasmmoam>>Houqumquaxq<z<mxmo>mxquHmxaoezomquawammamzx>aem>mhexoe<
amawmmo>¢aa>o0m ....... x>omaomoe>m>>aoxmm>weommquHoqo¢¢<2m>aom>mwo>we¢
mMmeODHm>>QM3 .................................. ZJquOmm>QAmmm>FIOmmH¢
amaaxmo>>>>om> .................................. zqo>m0adamoamm>mmzmwe<
mMmmdeHMQmDMV .................................. QAO>Imm4HOUEfim>BBB>UB<

oow ohm own omm

‘III dddddddddqq All

__n we o_n

H>aoanmmzoqquzdm>wox¢mozomxaan209mmumqowq»me>mmmme9>mz<Oanmmm2quo
H>wouaHmaoqzamzzx>wog«mozomxamazoawnuoeqzwaaamaaaeaxme>aemom>mmmm2quw
H>aamEOmooqzaxozx>woa¢mozomqmnzzoqwnumanuo¢Hz<Hameaxme>memommmmm22mqaw
>>womaxmuonzn.a¢m>>awmaozommannzoaqueano>xHawaaxeme9>mzmommnmmm2aqHo
som>>om¢aoznmo<anzoqmaquamaxHm<>ooeaxme>mmHomxnmmmZEAHo

 

>>>on4xmooqzara m>>oo<aozamxdc>zoqmnmooxoaxHm4>aoe>xme>mmHomqnmmmzaqu
own omm own oam oou oma oma
.LIII ddflmqqddenfiq AIIIII .OII dddflfiqqdd&flnfiqq .All
on he an an as

mnofluumumunH Hufian

dufism
anoH>m.m
auoo.m
namoooanuua
oauonu.<
amonamsuou.m

ndadm
akOahm.=
aHou.m
mauaouanaua
oauonu.¢
ouonamauoa.m

nufium
aHoH%Q.m
aaou.u
nauoooanuua
ofiuonu.<
amonamfiuoo.m

6.283 .2 2:»:

39

In the oxidation step, the NADP cofactor abstracts a hydride from the C4 hydroxyl of
the mannose ring. The mannose modeled in the active site of GMD reveals a distance of
3.5 A between the C4 of the nicotinamide ring and C4 of the hexose. The angle formed
by the N1-C4 atoms of the nicotinamide ring and C4 of the hexose is 92°. Both values
are consistent with the positioning of the cofactor and substrate in other related enzymes
such as MURl [44], SQDI [43] and human GalE [45], where the distances range from
3.4 A to 3.7 A and the angles fall in the range of 81° to 96°.

To complete the oxidation step, an active site Tyr removes a proton from the O4
hydroxyl forming the 4-keto intermediate. The catalytic TyrlSO in GMD is in proper
position (distance of 2.7 A) relative to the mannose model for it to directly attack the O4
hydroxyl. Early studies of E. coli GalE [55] and dTGDH [21] showed that the distance of
Tyr to the O4 hydroxyl is too great for it to act directly as the base. The Ser/T hr catalytic
triad member had been proposed to act as a proton shuttle to complete the oxidation step.
However, more recent crystallographic studies of human GalE [45], dTGDH [42], SQDI
[43] and MURI [44], showed that Tyr is within proper hydrogen bonding distance to
directly attack the O4 hydroxyl. The Ser/T hr instead may orient the substrate in the
active site and/or facilitate proton transfer. The presence of hydrogen bonds between the
O4 and O6 hydroxyl and the Thr126 hydroxyl suggests that both these roles may be
accomplished. To further facilitate the oxidation step, the catalytic Lys may stabilize Tyr
in its negatively charged state. Studies suggest that Lys lowers the pKa of Tyr [32],
which is normally between 9-12. The measured pKa of Tyr in E. coli dTGDH [32] and
E. coli GalE [41] is 6.4 and 6.1 respectively. The distance between the phenolic oxygen

of Tyr150 and the amide nitrogen of LyslS4 is 4.4 A, a distance too far for hydrogen

40

bonding, but within the range of electrostatic interactions to effectively lower the pKa of
Tyr150.

The 4—keto intermediate acts as a springboard to other SDR reactions. In the case of
GMD, the ketone functionality serves to acidify the proton at C5 and permits the
dehydration from C6 to form the GDP-4—keto-5,6—ene intermediate. The presence of the
“cue” intermediate has recently been detected in the homologous dTGDH reaction [29].
For dehydration to occur, another active site base must be present to abstract a proton
from the C5 position. Studies of E. coli GMD [52] and dTGDH [30] showed that a
glutamic acid might fulfill this requirement. The corresponding Glu128 side chain in
GMD is within 3.6 A of the C5 carbon of the mannose model, a position that would
enable it to deprotonate C5. To complete the dehydration reaction, the C6 hydroxyl must
be protonated by an active site acid. An Asp residue has been proposed based on
structural analysis of dTGDH [42] and supported by mutagenesis experiments [27, 30].
The corresponding GMD residue, Ser127, is within 2.8 A of the modeled position of the
hexose O6 hydroxyl, and may assume a similar role. Alternatively, Glu128 is within 2.6
A, suggesting the possibility of this side chain playing a dual role in the dehydration step,
acting as both a general base and a general acid, as has been suggested in dTGDH and
MURI [30, 44]. Whereas the dehydration mechanism described here is the step-wise
water elimination mechanism as seen in D135N and D135A mutants of dTGDH,
dehydration may also occur through a concerted mechanism as seen in wild type dTGDH
[31 [. Further kinetic studies would need to be completed to determine which mechanism

of dehydration GMD actually utilizes.

41

The final step of the GMD reaction involves a hydride transfer from NADPH back to
the hexose C6 position. The distance between the nicotinamide C4 atom and the C4 and
C6 atoms of the hexose moiety (3.5A and 3.8A respectively) suggests that only modest
rotation of the hexose ring would be required to complete the hydride transfer.
Interestingly, because NADP is regenerated, the cofactor may remain bound through each
catalytic cycle. To finalize the reduction step, an active site acid is required for proton
addition to the C5 position of the hexose. Proposed residues to fulfill this role based on
structural analysis of dTGDH include the catalytic Tyr, Glu or Asp [42]. Of the
corresponding residues in GMD, Tyr150 and Ser127 (aligning with Asp) are >44 A to
the C5 position of the hexose model. Glu128 is 3.6 A away, but would move even
further with the rotation of the hexose ring towards the nicotinamide ring. However,
Thr126 of the catalytic triad and Asnl79 are positioned such that they may be able to
fulfill the role as the general acid to complete the reaction.

1.3.6. Structural Comparisons. The secondary structural elements between P.
aeruginosa GMD and MURI superimpose well with a root mean square deviation (rmsd)
of 1.2 A over Cu atoms. The secondary structural elements between P. aeruginosa GMD
and E. coli GMD do not superimpose as well (rmsd of 3 A) because E. coli GMD has no
substrate or cofactor bound in the active site. The main difference between the three is an
area of disorder present in MURI residues 76-81 and E. coli GMD residues 35-55 which
corresponds to a region that is highly variable in size and sequence among GMDs (Figure
13). This stretch in P. aeruginosa GMD is shorter and well ordered, forming a short
helix ((12) between [32 and [33 of the Rossmann fold. Immediately preceding this region

is a Gly-XX-ArgArg sequence that is conserved among all GMDs sequenced thus far

42

except for P. aeruginosa GMD. In contrast, P. aeruginosa GMD exhibits an arginine
shift resulting in the sequence Gly—XXX-ArgArg. The two positively charged arginine
residues mark the beginning of the RR loop that closes over the adenosyl phosphate end
of the cofactor, are important in cofactor binding and are involved in the tetramer
interface for P. aeruginosa GMD and MURI. The RR shift causes an interesting
rearrangement of interactions. In MURI the first arginine of the sequence, Arg60, adopts
a parallel stacking arrangement with Arg60* of the neighboring monomer,
simultaneously packing against the adenine ring and coordinating the 2’phosphate. In
GMD the second arginine of the sequence Arg36 also adopts a parallel stacking
arrangement with Arg36* of the neighboring monomer. However, due to the shift by two
residues, the side chain is oriented almost perpendicular to the adenine ring (Figure 14).
Despite the rearrangement in this region, Arg36 in P. aeruginosa GMD still maintains the
electrostatic interactions to coordinate the 2’phosphate. Based on the role that the
cofactor plays in the tetramer interface of P. aeruginosa GMD and MURl, cofactor
binding might also be expected to assist in ordering part of the RR loop in E. coli GMD
and may be essential for tetramer formation.

Other differences between P. aeruginosa GMD and E. coli GMD include the
positioning of the loop between the [34 strand and the (14 helix. This loop in the apo E.
coli GMD occupies a portion of the cofactor-binding site and would have to move as
much as 6 A to make room for the cofactor. Secondly, the smaller substrate—binding
domain in E. coli GMD adopts a more open conformation. The difference is apparent
when comparing the rmsd of 2.4 A with P. aeruginosa GMD for the C-terminal domain,

as opposed to just 1.8 A for the N-terminal domain. In addition to the more open

43

conformation of its C-terminal domain, E. coli GMD has an extended loop between the
(18 helix and the BI] strand. The loop, which inserts an additional 27 residues compared

to P. aeruginosa GMD, is variable in size among GMDs.

     
 

ArgSS' , i‘ ArgGO 5‘s"!
Ar961' .

 

Figure 14. RR loop of GMD and MURl
Stereoview of the superposition of the RR loops from GMD (highlighted in color) and

MURl (blue). GMD side chains are labeled in green, MURl side chains are labled in
blue, asterisks denote side chains from opposite monomer.

One of the intriguing features about P. aeruginosa GMD is its oligomeric state, as it
deviates from the canonical homodimeric structures seen in most other related enzymes
of the N DP—sugar modifying subfamily of the SDRs. The structures of all other members
of this subfamily, including E. coli GMD, have been observed as dimers, with the
exception of the previously mentioned tetrameric tyvelose epimerase [40] and ADP-L-
glycero—D L _‘ 5 r' , which is a pentamer [74]. P. aeruginosa GMD

and A. thaliana MURI can be seen as a dimer of canonical SDR dimers, which then

44

generates a new set of subunit interactions. An important subsequent question is whether
a significant number of tetramer interactions between GMD and MURl are conserved.
As previously mentioned, the RR loop of residues Arg35-Arg43 that is so intimately
involved in the tetramer interface and the cofactor binding sites, makes protein-protein
interactions as well as protein—cofactor interactions to the neighboring monomer. The
Arg35 to Ser85* and Ser37 to Arg43* hydrogen bonds are conserved in GMD and
MURl. Furthermore, these residues are highly conserved among the GMD sequences.
Also within the RR loop is hydrogen bonding between Ser38 and Arg43*. Although the
residue corresponding to Ser38 is an Asn in MURl, the interaction is conserved. The
sequences of several GMDs reveal that in most cases, a Ser is present in this position.
Away from the cofactor binding site overlap, Asp62 hydrogen bonds to the amide
nitrogen atoms of Val95* and Thr96*. Asp62 is highly conserved among GMDs. One of
the more interesting interactions involves residue Arg68, which is moderately conserved
across several GMD sequences. This residue is involved in hydrogen bonding to the
main chain carbonyl oxygen of Asn92*, an interaction also seen in MURI. The same
Arg in GMD and MURl extends towards the diagonally related monomer to hydrogen
bond to Glu] 10 and Argl 13, both highly conserved residues among GMD sequences.
1.3.7. Conclusions. In summary. the GDP-D-mannose 4,6-dehydratase MURI
isoform from Arabidopsis thaliana was shown to be a tetramer, while the first GMD
structure to be determined, E. coli apo-GMD, was observed as a dimer. This raised the
question as to whether or not prokaryotic and eukaryotic GMDs differed in oligomeric
state. We have determined the structure of P. aeruginosa GMD with NADPH and GDP

bound in the active site and found it to exist as a tetramer. The tetramer arises from the

45

dimerization of the canonical dimer seen for most members of the SDR superfamily, but
in a manner where the cofactor binding sites closely interact across the new interface.
The residues involved in the tetramer interactions are well conserved between the
prokaryotic GMD and the eukaryotic MURI. Moreover, a high degree of sequence
conservation among the residues within the tetramer interface is also observed across a
broad range of GMDs. These observations suggest that the tetramer may be a more

common oligomeric state for GMDs than previously thought.

1Portions of Chapter 1 were previously published, Webb et al.. (2004) Protein Science.

46

CHAPTER 2: CRYSTALLIZATION AND INITIAL X-RAY ANALYSIS OF RMD

2.1. Introduction

Genes coding for a particular bacterial polysaccharide are usually found in clusters.
In the case of both A. thermoaerophilus and P. aeruginosa, the rmd gene is located
adjacent to the gmd gene. The corresponding RMD enzymes have been characterized in
both the gram-negative bacterium P. aeruginosa and gram-positive bacterium A.
thermoaerophilus [3, 54]. A sequence alignment of the two RMDs shows 33% identity
and 54% similarity between pairs. A BLAST search with RMD shows significant
similarities to GDP-mannose dehydratases, dTDP-glucose dehydratases, UDP-galactose
epimerases and GDP-fucose synthetases, all members of the NDP-sugar modifying
subfamily of the SDRs. No RMD analogue has yet been found in the human genome
using P. aeruginosa RMD as a probe, which is in agreement with the fact that humans
lack rhamnosylation. Thus, the second step in GDP-D-rhamnose biosynthesis appears to
be a potential target for enzyme inhibition to prevent the synthesis of D-rhamnose-
containing LPS molecules in pathogenic bacteria. Structural information would aid in the
design of inhibitors. Since no structure is currently available for RMDs, efforts were
made to determine the X-ray structure of RMD from P. aeruginosa. As this proved to be
a difficult task, an ancillary effort was made to determine the structure of RMD from A.
thermoaerophilus, which has shown to be similar in function to the P. aeruginosa RMD.

Reported here are the cloning, expression, purification, crystallization and initial X-
ray analysis of the RMD crystals from both bacterial species. The amino acid sequence
of RMD most closely aligns with its biosynthetic pathway mate GMD. If the residues

align at the tertiary level, then we would expect that the core structure would mimic that

47

of GMD, or other SDR enzymes, since the three-dimensional structures tend to be highly
similar. Crystallographic analysis should allow us to see if RMD meets our expectations
or breaks from the mold of the typical SDR protein. Preliminary data suggests that P.
aeruginosa GMD may be an appropriate search model for the molecular placement
method of phase determination using data from A. thermoaerophilus RMD crystals.

indicating our hypothesis may be true.

2.2. Experimental Procedures

2.2.1. Cloning and Expression of RMD. The rmd gene from P. aeruginosa,
originally isolated by Lightfoot and Lam [75], was amplified by PCR while incorporating
a BamHI site over the natural start codon and a SalIII site over the natural stop codon.
The new PCR fragment was ligated into the pQE30 vector (Qiagen), placing a 6x-His tag
at the N-terminus. A glycerol stock of E. coli M15 cells harboring the recombinant
plasmid was used to inoculate a starter culture of 50 ml Luria Bertani (LB) broth.
Following cultivation at 37°C for 16 h, the starter culture was transferred to 1 L LB broth.
Once an optical density at 600 nm of 0.5—0.6 was reached, the cells were induced with
0.25 mM IPT G and grown for an additional 16 h at room temperature before harvesting
by centrifugation. Cells were re-suspended in Lysis Buffer (50 mM HEPES, 300 mM
NaCl, 5 mM imidazole pH 8.0).

The rmd gene from A. thermoaerophilus, originally isolated by Kneidinger et al. [3],
was also amplified by PCR while incorporating a BamHI site at the N-terminus and a
Kpnl site at the C-terminus. The PCR fragment was then ligated into the pQE80 vector

(Qiagen), which placed a 6x-His tag at the N-terminus. The new plasmid was

48

transfonned into E. coli BL21 DE3 cells. The same protocol as stated above was used to
express the protein in shake flasks.

2.2.2. Purification of RMD. The same protocol as in section 1.2.3 was used to
purify RMD resulting in the N-terminally His-tagged RMD from P. aeruginosa
(PaRMD) or the N -terminally His-tagged RMD from A. thermoaerophilus (AtRMD).

2.2.3. Crystallization of RMD. Purified PaRMD or AtRMD at 10 mg/ml were used
to set up crystallization screens using the microbatch method with the Impax I-S robot
(Douglas Instruments). A total of 198 conditions were screened by combining 1 ul of
purified protein with 1 ul precipitating solution. Crystallization conditions were refined
using the hanging drop vapor diffusion method. Prior to data collection, the PaRMD
crystals were treated with 0.1% glutaraldehyde, briefly transferred to a cryoprotectant
solution (10% PEG 10K, 100 mM Hepes pH 7.6, 7.5% glycerol, 2.75 M sodium formate.
5 mM GDP-D-mannose and 2 mM NADP) and flash-frozen in liquid nitrogen. The
AtRMD crystals were flash-frozen in liquid nitrogen directly from the crystallization
drop.

2.2.4. Cryoprotection Techniques. Through personal correspondence with Terese
Bergfors [76], a technique for streak-seeding for cryoconditions was created. Purified
PaRMD at 10 mg/ml was used to set up crystallization screens (Crystal Screen 1 and 2
from Hampton Research) using the hanging drop vapor diffusion method. A 1:1 mix of
protein and precipitating solution in a drop size of 4 ul was placed on the cover slip and
allowed to equilibrate over 500 pl of precipitating solution for one day. All drops were
then streak-seeded with their respective protein crystals in an effort to determine possible

cryoprotectants. This process was accomplished by touching a cat whisker to the side of

49

a previously obtained protein crystal in order to gather nuclei. Then the cat whisker was
streaked through the drops on the cover slips whether there was precipitation or not. The
drops were then monitored for their ability to grow crystals. Those conditions that
supported any crystal growth, whose precipitating solution contained a component used
for cryoprotecting, provided a starting point for optimizing the final cryosolution.

A technique was devised from an idea that stemmed from a method listed in the
CCP4 bulletin board (we-'w.ccp4.ztc.uk) titled “Summary of oil and cryo-protectant
combo”. After crystal(s) were obtained in a hanging drop vapor diffusion screen, an o-
ring greased with vacuum grease was placed on the cover slip around the drop containing
the crystal(s). The resulting reservoir was then filled with paraffin oil, covering the drop.
The crystal(s) were then directly looped out of the drop and flash-frozen, providing a
layer of oil as the cryoprotectant.

2.2.5. Data Collection and Processing. X—ray diffraction data from both the PaRMD
and AtRMD crystals were collected on a MAR CCD detector at the Advanced Photon
Source beamline 5—ID (DND), Argonne National Laboratory. During data collection the
crystals were held at 100 l( in a cryostream and radiation was used at a 1.0 A wavelength.
The data were processed using XDS software [77]. The AtRMD crystal diffraction data
were used to calculate a self-rotation function using the AMoRe programs [66] from the
CCP4 suite [67]. See Table 3 for statistics.

2.2.6. Preparation of Phasing Models. The secondary structure of P. aeruginosa
and A. thermoaerophilus RMD were predicted using the PSIPRED secondary structure
prediction method [78] using the PSIPRED server [79]. The fold recognition server 3D-

PSSM was used to predict tertiary structure by threading [80]. The tertiary structure was

50

also predicted using SWISS—MODEL [81], which uses comparative modeling. The
coordinates were checked visually and compared to other structures of SDR members.
Based on the secondary and tertiary structure prediction as well as sequence alignments,
molecular replacement models were constructed using P. aeruginosa GMD (PDB code
IRPN), E. coli GMD (PDB code 1DB3) and Arabia'opsis MURI (PDB code 1N7H) as
starting models. All differing residues were changed to alanines and portions were
removed in which 1) amino acid sequences differed significantly and/or 2) the predicted
structure differed. The main structural feature of the SDR enzymes, the Rossmann fold,

was maintained.

Table 3. RMD X-ray diffraction data

 

PaRMD AtRMD

 

Space group P422 (potentially P4212, 1"
P4122, P412|2, P4222,
P42212, P4322, P43212)

Unit cell parameters a=b=182, c=250 3:46.88, b: 55.74, c=79.24,
(A, °) a=B=y=90 =72.54, [3432.95, y=75.61
Resolution range (A) soc—3.7 30.0- 1 .8

No. observed reflections 209,657 215,748

No. unique reflections 44,486 64,129

Completeness (%) 93.9 (96.7)* 96.5 (95.2)*

Rmerge (%) 7.5 (21.8)* 10.6 (48.1)*

Average l/0(l) 5.8 (1.2)* 13.4 (3.7)*

VM (A Da 1) 2.46 2.66

Molecules per

asymmetric unit 12 2

Estimated solvent

content (%) 50.1 53.8

 

*Indicates statistic from highest resolution shell.

51

2.3. Results and Discussion

2.3.1. Expression and Purification. RMD from both P. aeruginosa and A.
thermoaerophilus were expressed in the pQE system as N—terminally His—tagged protein
(PaRMD and AtRMD, respectively). Both constructs were readily purified to 90—95% by
Ni—NTA column, and further purified to 95-98% by an anion exchange column. A
typical protein yield from the PaRMD preparations was 9-12 mg of purified protein per
liter of cell culture; the yield for AtRMD was less, only 2—3 mg of purified protein per
liter of cell culture. SDS—PAGE gels of samples from a typical purification of PaRMD
are shown in Figure 15 and AtRMD samples are shown in Figure 16. The gene products
appear to run on the SDS—PAGE gels as a 35 kDa protein, which closely matches the

molecular weight based on the amino acid sequence of 34.8 kDa for PaRMD and 35.9

kDa for AtRMD.

   

2 M L F1 F2 F3 F4 F5 F6HF-7 F8 F9

  

Figure 15. SDS-PAGE of samples from a typical PaRMD purification

(A) load (L), flow through (FT), wash (W) and fractions (F) from the Ni-NTA column;
(B) load (L) and fractions (F) from the anion exchange column. Molecular weight marker
(M) standards shown in kDa to the left of each gel. PaRMD runs as expected at ~35 kDa.

52

LFTWFI F2 F3 F4 F5 MLFI F2 F3F4F5
193 a. -‘
102
60

  

41

27

 

20
15

Figure 16. SDS-PAGE of samples from a typical AtRMD purification
The left-hand side of the gel shows the load (L), flow through (FT), wash (W) and

fractions (F) from the Ni-NTA column; the right-hand side of the gel shows the load (L)
and fractions (F) from the anion exchange column. Molecular weight marker (M)
standards shown in kDa to the left of the gel. AtRMD runs as expected at ~35 kDa.

2.3.2. Crystallization. The initial screening of crystallization conditions at room
temperature for PaRMD resulted in protein crystal clusters as shown in panels A, B and C
of Figure 17. The crystals in panels A and B were obtained in the following conditions:
50:50 mix of protein sample and well solution where the protein sample contained 10
mg/ml protein, 5 mM GDP and 2 mM NADP and the well solution contained 10% PEG
8K, 100 mM sodium/potassium phosphate pH 6.0, 200 mM NaCl. Although any sight of
protein crystals in a screen is encouraging, it is clear from the inspection of these crystals
under a low-powered light microscope that they are severely twinned. Crystallization
conditions such as growth temperature, the pH of the buffer and the molecular weight of
the PEGs were modified to find the optimal crystal growing condition. Single crystals
were obtained simply by replacing 5 mM GDP with the substrate analog 5 mM GDP-D-
mannose to the protein sample. Examples of these crystals can be seen in panels D and

E. The crystals in panel D are hexagonal in shape and about 0.2 x 0.2 x 0.1 mm in size.

53

Panel E shows thick rod-shaped crystals that were generally 0.15 mm in thickness and

0.4-0.5 mm in length.

 

         

Figure 17. Crystals of PaRMD
Conditions giving crystals in panels A and B were optimized to yield single crystals

shown in panels D and E by the addition of GDP-mannose instead of GDP. The
conditions to grow the crystals in panel C were optimized by changing the buffer and the
molecular weight of PEG to yield single crystals like that shown in panel F.

The PaRMD crystals in panel C of Figure 17 were obtained in a different screening
condition: 50:50 mix of protein sample and well solution where the protein sample
contained 10 mg/ml protein, 5 mM GDP-D-mannose and 2 mM NADP and the well
solution contained 12% PEG 20K and 100 mM MES pH 6.5. By optimizing the PEG

molecular weight and using a different buffer, single, non—twinned crystals were obtained

in the following conditions: 50:50 mix of protein sample and well solution where the

54

protein sample contained 10 mg/ml protein, 5 mM GDP-D-mannose, 2 mM NADP and
the well solution contained 19% PEG 10K and 100 mM HEPES pH 7.5. An example of
these bipyramidal-like crystals, which were generally 0.4 x 0.4 x 0.4 mm in size, is
shown in panel F.

A microbatch screen for crystallization conditions using AtRMD resulted in small,
plate-like crystals shown in Figure 18. The crystallization conditions were as follows:
50:50 mix of protein sample and well solution where the protein sample contained 10
mg/ml protein, 5 mM GDP-D-mannose and 2 mM NADPH and the well solution
contained 35% pentaerythritol propoxylate (PEP) (5/4 PO/OH), 100 mM MES pH 6.5
and 200 mM MgC12. Larger plate-like crystals were grown using the hanging drop vapor

method.

 

Figure 18. Crystals of AtRMD
Small, plate-like crystals of AtRMD obtained in a microbatch screen.

The PaRMD crystals tended to grow within a week whereas the AtRMD crystals
tended to take 3-4 weeks to grow. Reproducibility was a problem not only with protein
originating from different purifications, but also with protein from the same batch. In

addition, the crystals were prone to crumbling and dissolving, necessitating an arduous

55

search for a suitable cryoprotectant and/or cryoprotection technique that would not
damage the crystals.

2.3.3. Cryoprotection. Generally before X-ray data collection, crystals are briefly
transferred to cryoprotected mother liquor then flash-frozen to minimize radiation
damage during the experiment. The cryoprotectant ideally prevents the formation of ice
and maintains the crystallographic integrity of the crystal during the freezing process.
The selection of the cryoprotectant(s) as well as the percent composition involves some
trial and error. The PaRMD crystals grew in PEGs, which are known ice-preventing
agents. The X-ray data of the crystals that were frozen directly from the drop revealed
the formation of ice by the observed powder diffraction rings or “ice rings” on the X-ray
diffraction screen. Data from crystals that had been transferred to a drop with a higher
percentage of the respective PEG revealed poor-quality diffraction, potentially indicating
poor cryoconditions. Thus an assay of several cryoprotectants, concentrations and a test
of mixtures were required to determine the best cryocondition. To reduce the evaluation
time it took to find the appropriate cryoprotectant, a screen of crystallization conditions
was streak-seeded. First, a crystal screen is prepared and allowed to sit over night. Then
a cat whisker is stroked over the surface of a previously obtained crystal to dislodge and
collect the nuclei. Finally each drop of a screen is streaked with the cat whisker. The
crystal trays are then monitored for any crystal growth, whether large or small, twinned
or single, needles or chunks. If crystals appear, then it is assumed that the components of
the precipitating solution would be more helpful than harmful to the crystal lattice. With
any luck, one of the components is a cryoprotectant, giving a good starting point in the

search for the appropriate cryosolution. Nevertheless streak-seeding PaRMD for

56

cryoconditions yielded no positive results, so several of the more common
cryoprotectants were tested (i.e. sugars, MPD, ethylene glycol, glycerol and lower-
molecular-weight PEGs). However, in each case, the crystals crumbled before flash-
freezing.

Paraffin oil was also investigated as a suitable cryoprotectant, since it does not require
the transfer of the crystal from one drop to another, which could disturb the crystal
packing. A previous study has shown that crystals grown in PEG solutions are stripped
of the thin film of mother liquor when passed through oil [83]. So a low—tech technique
was created that would minimize manual manipulation of the PaRMD crystals. The
cover slip on which the crystals were grown was turned drop side up. A greased o-ring
was placed over the 4 ul drop and the reservoir was filled with paraffin oil. The crystal
was then directly loop out of the drop, swiped through the oil and flash—frozen in liquid
nitrogen. This method has the added advantage of providing extra handling-time as it
protects the drop from drying out. Even though it seemed to be a less invasive procedure,
the crystals were irreversibly damaged.

In the end, the best cryocondition for the PaRMD crystals was an initial treatment
with glutaraldehyde, then the use of a combination of the same molecular weight PEG in
which the crystals were grown, along with glycerol and sodium formate as well as GDP—
D-mannose and NADP. Glutaraldehyde is known to cross-link proteins through the
amino group of lysine, adding stability to fragile crystals for data collection. The final
conditions served to improve the quality of the diffraction by reducing spot twinning and

mosaicity.

57

The AtRMD crystals also proved to be very fragile. However, mesh litholoops
(Protein Wave Corp.) provided support for the thin plates during freezing. Since the
crystals grew in a high percentage of PEP, a known cryoprotectant, the crystals were
frozen directly from the crystallization drop. Fortunately, this proved to be a suitable
cryoprotectant and no further optimization was necessary.

2.3.4. Data Collection and Processing. A complete data set was collected from one
of the bipyramidal PaRMD crystals, like the one shown in panel F of Figure 17. These
crystals diffracted to 3.7 A, while all other PaRMD crystals diffracted to 8-10 A at best.
Although the signal-to-noise ratio is 1.2 in the final resolution shell (22 is desired), the
completeness (96.7%) and Rmerge (21.8%) are respectable (see Table 3). These crystals
belong to the P422 space group; it is unclear based on the systematic absences if the unit
cell possesses symmetry elements that would place it in a higher symmetry space group.
The number of molecules in the asymmetric unit was therefore estimated based on the
current working space group (P422), unit cell parameters (a=b=182 A, c=250 A,
a=B=y=90°) and molecular weight (34,825 Da) using the Matthews probability
calculation [84]. The median Matthews coefficient (VM) is 2.52 A3/Da with the vast
majority of proteins falling into the 2-3.5 A3/Da range; the median solvent content is 47%
with the vast majority of proteins falling in the 35-65% range [85]. For PaRMD a
Matthews coefficient of 2.46 A3/Da yields 12 molecules in the asymmetric unit with a
solvent content of 50.1%. However. certainly within reason is a VM of 2.96 or 2.11
A3/Da, which would allow for 10 or 14 molecules respectively in the asymmetric unit

with a solvent content of 58.4 or 41.8% respectively.

58

Though the resolution of the PaRMD data was high enough to determine phases, the
data would not provide much detail beyond the backbone trace. While attempts were
made to grow crystals in different conditions with the hope obtaining a different space
group, these hopes were not realized. Since it is difficult to give up on crystals that have
already been obtained, attempts were made to save the crystals and improve the
resolution by post-crystallization treatments. Two separate annealing procedures were
used: macromolecular crystal annealing (MCA) [86] and in situ annealing [86, 87].
MCA was able to improve the diffraction of the PaRMD crystals from 8-10 A to 6-7 A.
In situ or flash-annealing was able to help decrease the ice-rings, but had little affect on
the diffraction of the PaRMD. A “last resort” procedure, as described by the author of
the paper in which the procedure was presented, was fast desiccation of protein crystals
[88]. Although it was reported that poorly diffracting crystals showed “spectacular”
improvement of diffraction with this procedure, the same results did not occur with any
of the PaRMD crystals. This was the point at which efforts were directed at the RMD
homolog from A. thermoaerophilus.

Up until the point of X-ray data collection, the RMD protein from P. aeruginosa and
A. thermoaerophilus behaved in a similar fashion. Expression and purification were
comparable, even the crystals had fragility in common. However, once a single frame of
X-ray diffraction data was collected on the AtRMD crystals, the difference was readily
apparent, as they diffracted to 1.8 A. A complete data set was collected on the AtRMD
crystals (see Table 3 for statistics). The crystal falls into the P1 space group and have a
much smaller unit cell than that of the PaRMD crystals. As a result, based on a Matthews

coefficient of 2.66 A3/Da, the number of molecules per asymmetric unit is 2, with a

59

solvent content of 53.8%. A self-rotation function indicates that the molecules in the
asymmetric unit are related by a non-crystallographic twofold axis. suggesting AtRMD
exists as a dimer.

2.3.6. Initial Phasing Results. Structure determination by the molecular replacement
method using coordinates from various homologous structures was initially attempted
with data from the PaRMD crystals. With the use of the programs AMoRe [66] and
Phaser [89] from the CCP4 suite [67], no clear solution was evident in any of the possible
P422 space groups. On the other hand, using a dimer as a search model based on the
known structure of GMD from P. aeruginosa (PDB code lRPN), a solution for a dimer
in the asymmetric unit was obtained using the Phaser program from the CCP4 suite [67].
Currently efforts are underway to obtain experimental phases for comparison to those
determined by molecular replacement.

2.3.7. Conclusions. To summarize, the process of crystal structure determination of
RMD from P. aeruginosa and A. thermoaerophilus has been initiated. The cloning,
expression, and purification were successful, yielding plenty of protein pure enough for
crystallization trials. Single crystals were obtained of protein from both bacterial species
and X-ray diffraction data has been collected. While PaRMD crystals diffracted weakly
to only 3.7 A, AtRMD crystals diffracted to 1.8 A. Full structure determination by the
molecular replacement method using the coordinates from GMD from P. aeruginosa is in
progress. Our hope is that the AtRMD structure will serve as a successful search model

for the PaRMD data, allowing a comparison of the two structures.

60

CHAPTER 3: FUNTIONAL CHARACTERIZATION OF GMD AND RMDl

3.1. Introduction

Rhamnose is a deoxyhexose found in glycoconjugates of bacteria and plants but not
in humans. L-rhamnose is the more common isoform, found in cell walls and capsules of
a variety of bacteria. On the other hand, D-rhamnose has mainly been described as a
constituent of the LPS of gram-negative bacteria such as plant pathogens X. campestris
[9], and human pathogens, for example P. aeruginosa [11], H. pylori [10] and
Campylobacter fetus [90]. D-rhamnose molecules are arranged as repeating trisaccharide
units in the A-band LPS O-antigen of P. aeruginosa. The LPS molecules of this
opportunistic pathogen contribute to its virulence in immunocompromised patients
including those with cystic fibrosis, cancer and burn-wounds. Interestingly, the same D-
rhamnose polysaccharide structure seen in A-band LPS of P. aeruginosa has also been
detected in other opportunistic pathogens B. cepacia and S. maltophilia that are
problematic in cystic fibrosis patients [8].

The source of D-rhamnose comes from the nucleotide activated GDP-D-rhamnose,
which is made in two steps: GDP-D-mannose 4,6—dehydratase (GMD) first converts
GDP-D-mannose to GDP—4-keto—6—deoxy-D-mannose, which is subsequently reduced
GDP-D-rhamnose by the reductase (RMD). The genetics of GDP-D-rhamnose
biosynthesis in P. aeruginosa has been extensively studied. Lightfoot and Lam
determined that the P. aeruginosa gmd gene coded for a protein that functions as a GDP-
D-mannose dehydratase based on amino acid homology to other dehydratases. In
addition, paper chromatography experiments showed the conversion of ”C-GDP-D-

mannose to GDP-D-rhamnose from supernatants of an A” strain carrying the gmd gene on

61

a plasmid [91]. Rocchetta et al. confirmed the role of P. aeruginosa RMD in the
pathway from work that revealed that rmd knockout mutants lacked A-band O-antigen
[92]. Amino acid homology of RMD shows that it is similar to enzymes that modify
sugars at the C4 and C6 positions. Maki et al. verified the role of P. aeruginosa RMD in
the conversion of the 4-keto-6-deoxy intermediate to GDP-D-rhamnose by co-expressing
the Helicobacter pylori gmd and P. aeruginosa rmd genes in Saccharomyces cerevisiae
[54]. However. GMD from P. aeruginosa has not yet been functionally characterized.
Furthermore, the entire pathway has only been functionally characterized in A.
thermoaerophilus, where D-rhamnose is a constituent of the surface layer glycoprotein of
this bacterium [3].

In this study we developed a capiMary—electrophoresis (CE) -based enzymatic assay to
confirm the involvement of the enzymes GMD and RMD from P. aeruginosa in the
biosynthesis of GDP-D-rhamnose. To supplement the X-ray analysis experiments
(Chapter 2), assays were also completed on RMD from A. thermoaerophilus. We
demonstrate dehydratase activity in GMD and reductase activity in both GMD and RMD.
Furthermore, we present a method for the enzymatic synthesis of GDP-D-rhamnose, an

important glycobiological building block.

3.2. Experimental Procedures

3.2.1. Cloning, Expression and Purification of GMD and RMD. Procedures
outlined in Section 1.2.1. and 1.2.3. were used to produce purified GMD. Likewise.
procedures stated in Sections 2.2.1. and 2.2.2. were used to produce purified RMD from
both P. aeruginosa and A. thermoaerophilus. (Unless otherwise noted, discussion will

center on RMD from the bacterial source P. aeruginosa.)

62

3.2.2. Determination of the Oligomeric State by Gel Filtration. A HiPrep 16/60
Sephacryl S-200 High Resolution gel filtration column (Amersham Biosciences),
calibrated with high and low molecular weight calibration kits (Amersham Biosciences),
was used to determine the oligomerization of both GMD and RMD. A 1 ml sample was
prepared using protein in 20 mM Tris pH 8.5 and 300 mM NaCl with or without 10 mM
GDP-D-mannose or 5 mM NADP. Samples were applied to the column and protein
elution was detected at 280 nm. Molecular weights of the samples were estimated based
on the calibration curve of K8,. [(Ve-Vo)/(Vt-Vo)| vs. log molecular weight (where Ve is
the elution volume for the protein, V0 is the void volume and Vt is the total bed volume).

3.2.3. Assay of GMD and RMD activity. The standard reaction buffer was 40 mM
Tris-HCI (pH 7.5) and 10 mM MgCl2 with a total reaction volume of 35 to 60 uL.
Typically, the initial GMD reactions contained 1.0 uM GDP-D-mannose and 0.1 mM or
0.01 mM NADP. The reaction was initiated by the addition of l to 10 ug of purified
protein and incubated at 37°C. The completion of the GMD reaction was judged by CE
analysis (~1—2 h). For the coupled assay, 1 to 10 ug of RMD was added directly to the
reaction tube along with 0.1 mM or 0.01 mM NADPH. For the sequential assay, GMD
was removed by filtration using a microcon YM- 10 filter cartridge; the filtrate containing
the labile 4-keto intermediate was used as a substrate for the RMD/NADPH reaction. For
the determination of the pH optimum for GMD, the following reaction buffers were used:
MES (pH 5, 5.5, 6.5), Tris (pH 7.0, 7.5, 8.0) and Bis-Tris propane (pH 9.0, 10.0).

3.2.4. Analysis of Reaction Products by Capillary Electrophoresis. CE analysis
was performed with a P/ACE MDQ Glycoprotein System (Beckman Coulter) with UV

detection. A 75 um x 50 cm bare silica capillary was used with the UV detector mounted

63

at 50 cm of the capillary. The CE analysis was run in 25 mM sodium tetraborate (pH 9.5)
at 25°C, and sample was introduced by pressure injection for 8 sec. The separation was
performed at 22 kV, and the sugar-nucleotide substrate and product were detected at 254
nm. Peak integration was performed using the Beckman 32 Karat Software.

3.2.5. Purification of Sugar Nucleotide Product via HPLC. A preparative scale
enzymatic reaction containing 30 umol of GDP-D-mannose and 1.5 mg of purified GMD
was incubated for 2 h at 37°C in 40 mM Tris pH 7.5 and 10 mM Mngl. Protein was
subsequently removed from the completed reaction by unltrafiltration through a
Centriplus YM-3 cartridge (Millipore). The filtrate was subjected to an Econo-Pac High
Q anion exchange column over a 0-500 mM triethylammonium bicarbonate buffer (pH
8.0) gradient. Fractions that contained GDP-D-rhamnose were pooled, acidified to pH
4.8 with AG 50W-X4 resin (Bio-Rad) and lyophilized in preparation for NMR analysis.

3.2.6. NMR Analysis of Reactions Product. The lyophilized sample containing
GDP-D—rhamnose was resuspended in 160 pl of 99% D20 and analyzed by NMR
spectroscopy. All spectra were acquired using a Varian Inova 500 MHz
spectrophotometer equipped with a Z-gradient triple resonance (‘H, l3C, 3'P) probe. The
experiments were performed at 15°C with suppression of the water resonance. The
methyl resonance of acetone was used as an internal reference at 6,, 2.225 ppm and 6C
31.07 ppm. The COSY. TOCSY, HMQC and 3'P HSQC experiments were used to assign

the resonances [93].

3.3. Results

3.3.1 Protein Expression, Purification and Determination of Oligomeric Status.

The results for GMD and RMD cloning, expression and purification are summarized in

64

Section 1.3.1. and 2.3.1. respectively.

Gel filtration analysis of recombinant GMD

suggested that it exists as a tetramer in similar buffer conditions used for crystallization

(Chapter I) as well as functional studies (Figure 19a). In contrast, recombinant RMD

appears to exist as a dimer (Figure 1%). See Table 4 for elution volumes.

>
ii; 2A
‘1

I . ‘ A
r i i ' l,
i I
.‘ ,
2 f l
i l
. 2 ‘ l I ‘
r I l . I‘
~ I i '
i ‘ l
I I
8 . l V
l . . ‘.
. ./' ‘. l . .'
‘ . I ' L. . ‘ i

l i ‘ ~ 1 V

l '. . ' , ; j",

i ' . ‘ l l l ‘5

l ' . I l i
l ‘ I . f , .
. V . ‘ , l

v, ,- l i
l
. .
l “ u
l '2 3 h .
.7 "V A‘ f
2 l, \‘. \. 1'
r ‘ ‘ ~ ‘ \ '.' (,1
i ,4 I / \\.;\'r- . 1’ /I
. l ,- . I l l l l ‘7‘ ire-“M. ‘ (NW w l l l \
171"“ ""T . l l I '. , . l. T‘ t 4 ‘1‘ ‘r'h—e
. = l l 3 3 ' I ! ' l l

l t ,
l i I l .I
l
2 . : , .
, I
l l
.—. . , ,
‘ » I
lA ’I It I
l i .
l 5 , ‘ i
l l l '
l ‘ I . i
I i l x. I
. l ‘ I| '
I . i j l
l i i ll 1 I i
. ' l l , l ‘
t t l l I
« : I i .,
‘ l j l \
l t l | l
, _. .
J t ' y‘ l .
. ‘ I . ~ ‘
l l ; 3 l v .‘
1 n l ‘ ‘- u I
i , i ' - -. 19 ‘1 l‘ r, f '1 x
I l' \\ /’ \ i '. \ ‘ 1' ‘ \
J .I \ , ,‘ __ \\ N. l .‘ ‘ ‘l
I '/ \ ' 2“. \ ‘\ .‘ / ‘
'l 4- " \- 1/ 33“.. _ \ a i’ " \\
I l . ‘ ,1 ,‘i x; t '- 4 ~ JT/N' ~. —-L -I’ . ' 4 l "~-

I I l

l ' - ' ' ‘ .
l , .
t; *‘Jm-zzr—fm i‘_ ; L»: in- ii;- “L- . .i/ l lNT'h‘MW

' ' ' ' l ' l ' 5 l 1 l I I I l l .

l l l . , I

Figure 19. Gel filtration analysis of GMD and RMD

. I 1
63'4"... 1.1:: -. -.- flti- - F:t' . A.- .42 -:. 'i‘. " " lit—taltdfl"
I I

(A) Gel filtrations analysis of GMD and standards, and (B) RMD and standards. Peaks:
(1) ferritin, (2) catalase, (3) aldolase, (4) albumin, (5) ovalbumin, (6) chymotrypsinogen
A, (7) ribonuclease, (8) GMD, (9) RMD, (10) RMD/GDP-mannose/NADPH.

65

Table 4. Gel filtration results of GMD and RMD

 

 

Peak Sample Molecular Weight Elution Volume, Ve
(Da) (ml)
1 Ferritin 440,000 36.5
2 Catalase 232,000 42.7
3 Aldolase 158,000 45.5
4 Albumin 67,000 49.8
5 Ovalbumin 43,000 55.8
6 Chymotrypsinogen A 25,000 69.4
7 Ribonuclease 13,700 75.7
8 GMD 48.]
9 RMD 52.9
10 RMD/NADPH/GDP-man 56.5

 

3.3.2. GMD catalyzes the dehydration of GDP-D-mannose. CE analysis of the
GMD-catalyzed reaction containing GDP-D-mannose in the presence or absence of NAD
or NADP resulted in the appearance of a new peak (Figure 20). This new peak putatively
corresponds to the formation of the intermediate product of GMD, GDP-4-keto-6-deoxy-
D-mannose. No standard is available for this intermediate product, which is known to be
labile making it difficult to analyze. The apparent dehydratase reaction of GMD
proceeded quantitatively and irreversibly and required no exogenous cofactor for activity.
To test the pH optimum for GMD activity, the reaction was completed under pH range of

5 to 10; GMD showed optimum activity between pH 7 and 8 (Figure 20).

66

k)

>u—

1

[>5
0

>
>>
[2

F

>

1: 3:9
11
’l

j

a.u. A254

J 6.0

 

Time (min)

Figure 20. CE analysis of GMD dehydratase activity from pH 5 to 10
(1) GDP-D-mannose, (2) putative GDP-4-keto—6-deoxy-D-mannose.

3.3.3. RMD catalyzes the reduction of GDP-4-keto-6-deoxy-D-mannose. Due to
the instability of the intermediate product of GMD, the 4-keto-6-deoxy-D-mannose
intermediate was prepared in situ. First the GMD-catalyzed reaction was incubated for 2
h, and the enzyme was removed by filtration. The filtrate containing the 4-keto-
intermediate was followed by CE analysis after incubation with NADPH (Figure 21f),
NADP (Figure 21e), NADPH/RMD from P. aeruginosa (Figure 21c) and NADPH/RMD
from A. thermoaerophilus (Figure 21b). No new peak was observed in the control
reactions with just NADPH or NADP, ruling out the possibility that reduction of the
labile intermediate was either spontaneous or by any reducing power of the buffer.
However, a new peak, the putative GDP-D-rhamnose. was observed upon the addition of

NADPH/RMD from either bacterial species. The novel peak was verified by spiking the

67

final reaction with GDP-D-mannose (Figure 21a). Similar results were observed in the
coupled reactions (data not shown). The apparent reductase activity of RMD proceeded

quantitatively and required exogenous NADPH. No reverse reaction was observed.

 

A L ii A) + GDP-D-man

: B) NADPH/AtRMD

 

 
 

 

 

 

 

 

 

E 4‘4 . 1
<2 . C) NADPH/PaRMD
. ,l H.
.3 l________;(~jl j l D) NADPH/GMD
. A ILL , E) NADP
. L A F) NADPH
Time (min)

Figure 21. CE analysis of GMD/RMD reactions

The GDP-4-keto-6-deoxy intermediate was treated with the components noted; the final
reaction was spiked with GDP-D-mannose (A). Arrows: (1) GDP-D-mannose, (2) GDP-
D-rhamnose, (3) GDP-4-keto—6-deoxy-D-mannose, (4) NADP, (5) NADPH.

3.3.3. ls GMD bifunctional? Interestingly, CE analysis revealed that the same peak
corresponding to GDP-D-rhamnose was observed when GMD/NADPH was added to the
filtrate (containing just the 4-keto-intermediate) (Figure 21d) as when RMD/NADPH was
added, indicating GMD displays bifunctionality in vitro. To test this observation further,

the conversion of the 4~keto-intermediate by GMD was followed in a time-dependent

fashion at 0 h (Figure 22d), 1 h (Figure 22c) and 2.5 h (Figure 22b), revealing the

68

quantitative conversion of the 4-keto-intermediate to GDP-D-rhamnose. The novel peak
was again verified by spiking the reaction with GDP-D-mannose (Figure 22a). The
reductase activity of GMD was dependent upon the addition of exogenous NADPH to the

reaction and no reverse reaction was detected.

‘—
e
‘—
{—5
{—6.

 

M 1%.... ' ‘ A) +GDP-D-man
31 -

 

 

 

 

 

 

 

 

3. l
<N it B) 2.5 h
:3; i1 , A C) 1 h
. )1 n A D) 0 h
E) GDP-D-man

 

 

 

 

Time (min)

Figure 22. CE analysis of the time-dependent GMD conversion of the keto-
intermediate in the presence of NADPH

GMD reductase activity at (D) 0 h, (C) 1 h, (B) 2.5 and (A) final reaction spiked with
GDP-D-mannose; (E) GDP-D-man. Arrows: (1) GDP-D-mannose, (2) GDP-D-rhamnose,
(3) GDP-4-keto-6-deoxy-D-mannose, (4) NADP, (5) NADPH.

69

3.3.4. Isolation and identification of the reaction product. No commercial
standard for GDP-D-rhamnose was available, so the product had to be further
characterized. The nucleotide sugar was purified by HPLC using anion exchange
chromatography (Figure 23). Two major peaks were observed in the elution profile.
Peak 1 was subjected to CE analysis to reveal that a small amount of NADP co-eluted
with the product. However, the reaction product was unequivocally identified as GDP-D-
rhamnose by NMR spectroscopy. The N MR spectrum of the nucleotide sample are
presented in Figure 24. From a 3”P-HSQC experiment a correlation was observed to the
anomeric resonance of the sugar resonance at 5.43 ppm. Proton assignments for the
sugar were made using the COSY experiment. Assignments for 13C were made from an
HMQC experiment. Proton coupling constants were typical of a manno—pyranose
configuration. The 'H, ”C, and 3'P chemical shifts for the sugar nucleotide were similar
to the ones reported before for GDP-D-rhamnose [3]. The chemical shifts for the carbon,
proton and coupling constants for the rhamnose are shown in Table 5. The values are in

congruence with those previously reported for GDP-D-rhamnose by Maki et al. [18].

70

 

 

A254

 

 

 

    

ltlllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllll

Time (min)

 

 

 

 

 

 

B
l 2
v “f“;
'. 'l'Ll
v 1
V
In
N
< _ l
- V +_A__.a A
A 1

 

 

 

Ff“ : - .. , .____~
Time (min)
Figure 23. HPLC purification of the GMD/RMD reaction product followed by CE
analysis

(A) Anion exchange chromatograph of reaction product (inset is enlarged view of the
first peak), (B) CE analysis of fractions in the first peak. CE peaks: (1) GDP-D-
rhamnose, (2) NADP.

71

CH, acetone

0
H H
H o 0
on no 11 I1
H0 o—r—o~r~o~cm o a
l l "00
n n on on

H“ H“

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

00

d

 

Figure 24. NMR spectra for GDP-D-rhamnose

(A) Partial proton spectra, (B) one-dimensional-TOCSY for selective excitation of the H-
1 sugar resonance with a mixing time of 150 ms, (C) 3 IP HMQC spectrum, (D) '3 C
HSQC spectrum.

72

Table 5. NMR analysis of GDP-D-rhamnose

WrH and 13C chemical shifts, 6(ppm) and proton coupling constants, JH,H(Hz)
Compound H-1 H-2 H-3 H-4 H-5 H-6
C-1 C-2 C-3 C-4 C-5 C-6
J(l,2) Jl2.3) Jl3.4) Jl4.5) J(5.6)
GDP-D-rhamnose 5.. 5.43 4.03 3.87 3.42 3.90 1.26
by 97.3 71.2 70.4 72.9 60.6 17.6
Jim 1.2 3.5 9.7 9.8 6.1

 

 

 

3.4. Discussion

In this study, we developed a CE-based assay for the functional characterization of
the recombinant enzymes GMD and RMD from bacterial source P. aeruginosa.
Enzymes were expressed separately in E. coli as 6His-tagged protein and purified to
homogeneity. Using CE, HPLC and NMR analyses, we were able to demonstrate and
confirm their involvement in converting GDP—D-mannose to GDP-D-rhamnose.

Our results have shown that using recombinant GMD that was fresh or frozen for
over 1 year resulted in the conversion of GDP-D-mannose to the intermediate product
GDP-4~keto-6—deoxy-D-mannose. The long-term stability of GMD is in contrast to
previous studies, which show the inherent instability of GMDs, except in the case of the
plant GDP-D-mannose 4,6-dehydratase MURl [5]. The pH optimum for GMD activity is
between 7 and 8, which is comparable to the pH optimums reported for GMD from K.
pneumoniae (pH 7-7.5) [l], porcine thyroid (pH 6.5—8.0) [6] and recombinant GMD from
E. coli (pH 8.0) [2] and human (pH 7.5) [51]. In most cases, the GMDs use NADP(H),
which is in contrast to the dTDP— and CDP-glucose 4,6-dehydratases that show
preference for NAD(H). Some GMDS (i.e. K. pneumoniae, A. thermoaerophilus, and H.

pylori) are activated or require exogenous cofactor for activity, indicating that the

73

cofactor may be loosely bound [1, 3, 4|. However, no exogenous cofactor was required
for P. aeruginosa GMD dehydratase activity, which is consistent with the purification
data, the known mechanism and the reported structure. P. aeruginosa GMD crystallizes
with cofactor bound in the active when none was added to the crystallization conditions.
indicating that it was tightly bound throughout purification. This is consistent with
reports that the cofactor was difficult to remove from MURl, E. coli GMD and P.
aeruginosa GMD [44, 52, 94], and the cofactor was described as tightly bound in porcine
thyroid and human GMD as well. Considering the mechanism, the cofactor transfers a
hydride from the C4 to the C6 position of the sugar resulting in its regeneration,
suggesting that the cofactor may remain bound throughout the catalytic cycle. Many
enzymes of the NDP-sugar modifying subfamily of SDRs that internally recycle their
cofactor show that the cofactor is tightly bound in the active site [31, 60]. Furthermore,
structural evidence of P. aeruginosa GMD suggests that the cofactor is involved in
creating the tetramer interface. To support the crystallography data, gel filtration
experiments have shown that P. aeruginosa GMD is functional as a tetramer. The
cofactor seems to be important for tetramerization. and perhaps, the regulation of enzyme
activity.

Regarding the RMD-catalyzed reaction, the apparent instability of the substrate, the
4—keto—intermediate product of GMD, was addressed by preparing the intermediate in
situ. In both the coupled and sequential reactions, we demonstrated the reduction of the
intermediate to GDP-D-rhamnose upon the addition of recombinant RMD from either P.
aeruginosa or A. thermoaerophilus. Data suggest that it is an NADPH-dependent

reaction, which is consistent with previous reports on RMD from A. thermoaerophilus

74

[3|. The requirement of exogenous NADPH suggests that the cofactor must be readily
accessible to solvent to be released, allowing a new cofactor to bind for restoration of
enzymatic activity, as it is suspected that there is not a second substrate in the reaction
that regenerates the cofactor NADP. Finally, gel filtration experiments have revealed that
RMD from P. aeruginosa is functional as a dimer and that the dimer is its assumed
biological molecule. Interestingly, the data suggest that RMD adopts a more compact
structure upon binding cofactor and substrate, perhaps indicating domain movement.

P. aeruginosa GMD not only exhibited dehydratase activity, but also NADPH-
dependent reductase activity in vitro, creating the final product GDP-D-rhamnose.
GMDs possessing bifunctionality have also been demonstrated in A. thaliana and A.
thermoaerophilus in vitro [3, 5]. Whereas the reductase activity of A. thermoaerophilus
GMD is quite low compared to its RMD, the reductase activity of P. aeruginosa GMD
seems to be comparable to that of its RMD. We expect no GDP-D-rhamnose
contamination from the E. coli expression system by activity of an enzyme encoded by
the bacteria, as no RMD homolog exists in the E. coli genome. Based on the CE analysis
of the enzyme-substrate reactions, the pathway for GDP-D-rhamnose biosynthesis in P.
aeruginosa could be revised to show that GMD can be a bifunctional enzyme in vitro,
converting GDP-D-mannose to the 4-keto—intermediate, then to the product GDP—[)-
rhamnose.

To conclude, GDP-D-rhamnose is an important component of LPS molecules, whose
biosynthetic pathways may be targets for antimicrobial agents. The enzymatic synthesis
of GDP-D-rhamnose would allow further studies on the assemblies of LPS molecules.

We have shown that recombinant GMD and RMD are able to convert GDP-D-mannose to

75

GDP-D-rhamnose, confirming the product structure by NMR analysis. With the
enzymatic activities of GMD and RMD characterized, these proteins are ideal candidates
for scaling up the production of GDP-D-rhamnose for use in studying
rhamnosyltransferases, Wpr, WbpY and Wpr, involved in poly-D-rhamnan

biosynthesis.

1Portions of this Chapter are from a manuscript in preparation under the following title:
Poon, K.K.H.', Webb, N.A.2, McNally, D53, Brisson, .l-R.3, Garavito, RM.2 and Lam,
J.S.', “Functional characterization of GMD and RMD involved in the biosynthesis of
GDP-D—rhamnose from Pseudomonas aeruginosa”.

1Department of Molecular and Cellular Biology, University of Guelph, Guelph, Ontario
N1G 2W1, Canada.

2Department of Biochemistry and Molecular Biology, Michigan State University, East
Lansing, Michigan 48824-1319, USA

3Institute for Biological Sciences, National Research Council, Ottawa, Ontario K1A 0R6,

Canada

76

CHAPTER 4: PROTEIN ENGINEERING WITH THE GOAL OF IMPROVING

PROTEIN EXPRESSION, PURIFICATION AND/OR CRYSTALLIZATION

4.1. Introduction

The road to obtaining a protein crystal structure can have several obstacles along the
way. The protein must be expressed to several milligram quantities and it must be
protected from proteolysis and must fold properly, sometimes requiring chaperones. A
purification scheme must also be devised that leads to a sample suitable for
crystallization. Conditions must then be determined in which reproducible crystals can
be grown that lead to X-ray diffraction data of a quality acceptable for structure
determination. A protein whose characteristics present problems in any step of the way is
typically referred to in the structural genomics world as a “hi gh-hanging fruit”. This term
has been given primarily to membrane proteins; however, it can also refer to other
proteins with low solubility and/or stability.

The use of affinity tags for purification of recombinant proteins has become a
widespread tool since its inception in 1983 with Protein A. Some large affinity tags like
glutathione-S-transferase (GST) and maltose-binding protein (MBP) have the added
advantage as they may provide increased expression, greater solubility, protection from
cleavage by proteases, and assistance in proper folding of the protein to which it is fused.
For crystallization purposes, large affinity tags have generally been cleaved from the
protein of interest. However, a handful of crystal structures of chimeric proteins with
large affinity tags have been reported; human T cell leukemia virus type I ngI
ectodomain fragment (gp21) [95], Staphylococcus accessory regulator R (SarR) I96] and

the MATal protein [97'], which were resistant to crystallizing by themselves. It has been

77

suggested that this strategy may have a greater application in structural genomics on
challenging proteins [98].

We have created a strategy that makes use of a large affinity tag along with an
additional stabilizer protein for increasing the expression, enhancing the purification and
facilitating the successful crystallization of problematic proteins. The pMal protein
fusion and purification system (New England Biolabs), which makes use of the affinity
tag MBP, was chosen as a starting point. In addition to the aforementioned advantages,
MBP allows for affinity purification of the fusion protein over amylose resin. However,
low affinity of the fusion protein to the resin may be caused by interactions between
MBP and the protein of interest that block or distort the maltose-binding site. For this
reason, the length of the linker between MBP and the protein of interest was maintained
to reduce the likelihood of any inhibitory interactions. Yet, the presence of the linker
adds flexibility and may prevent a stable conformation of the partners that is needed to
facilitate crystallization. To address this issue, the linker in the pMal-C2 vector was
replaced with an Src homology 3 (SH3) domain-binding site. The stabilizer peptide, the
SH3 domain of Abelson {99] tyrosine kinase, which is known to tightly bind the SH3
domain-binding site, was added to the purified fusion protein before crystallization to
serve as a clamp, providing rigidity to the new complex. This approach for problematic
proteins was successful in the crystallization of P. aeruginosa RMD. Although
reproducibility was a problem, the results presented in this chapter lead to promising
suggestions for further optimization of this system, which may have a place in
expressing, purifying and crystallizing “high—hanging fruit” in the structural genomics

world.

78

4.2. Experimental Procedures

4.2.1. Construction of pMal—C2L3 Vector. The following oligos were used to
replace 50 nucleotides between the SacI and anI restriction sites in the linker region of
pMal-CZ: 5’-CGGCAGCACCGACTTACAGCCCACCACCGCCGCCGGGAAGG-3’
and 5’-CCI"TCCCGGCGGCGGTGGTGGGCTGTAAGTCGGTGCFGCCGAGCTC—3’.
The vector pMal-CZ was first digested with Sac] and anI, isolated by electrophoresis
on 0.8% agarose gel and purified using the QIAquick gel extraction kit (Qiagen). The
following reaction was allowed to incubate at room temperature overnight: 1 pg each of
the oligos, 100 ng of the cut vector, 1 ul of 10x ligation buffer (New England Biolabs), 1
ul T4 DNA ligase (New England Biolabs) and ddeO to 10 ul. The reaction was
transformed into DHSa E. coli cells and plated on an LB agar plate with 100 ug/ml
ampicillin. Overnight cultures were grown from the resulting colonies; DNA was
isolated and sequenced for confirmation of the insertion.

4.2.2. Cloning of RMD into pMal-C2 and pMal-C2L3. The RMD-pQE30 (section
2.2.1.) construct and was digested with BamHI and Sall, isolated by electrophoresis on
0.8% agarose gel and purified using the QIAquick gel extraction kit (Qiagen). The
fragment was cloned into either pMal-C2 or pMal-C2L3 digested with BamHI and SalI.
The new constructs were confirmed via sequencing.

4.2.3. Expression and Purification of RMD-pMalCZ and RMD-pMal-C2L3. A
glycerol stock of E. coli BL21 (DE3) cells transformed with RMD-pMalC2 or RMD-
pMal-C2L3 was used to inoculate 50 ml LB broth with 2 mg/ml glucose and 100 ug/ml
ampicillin. Following cultivation at 37°C for 16 h, the starter culture was transferred to l

L LB broth with 2 mg/ml glucose and 100 ug/ml ampicillin. Once an optical density at

79

600 nm of 0.5—0.6 was reached, the cells were induced with 0.3 mM IPT G and grown for
an additional 4 h at 37° before harvesting by centrifugation. Cells were re-suspended in
Column Buffer (20 mM Tris-HCI, 200 mM NaCl, 1 mM EDTA, l mM DTT pH 7.4).
After a freeze/thaw cycle and sonication, the lysate was cleared by centrifugation at
10,000 x g for 20 min and loaded onto a column of 10 ml amylose resin. The column
was washed with 20 ml Column Buffer; then the protein was eluted with Elution Buffer
(Column Buffer plus 10 mM maltose). Those fractions shown to be homogenous were
pooled and concentrated to 10 mg/ml. The protocol recommended in the pMal Protein
Fusion and Purification System manual for cleavage of the fusion protein with Factor Xa
was used to cleave MBP from MBP-RMD.

4.2.4. Expression and Purification of Abl-SH3. The SH3 domain of human Abl
tyrosine kinase had been cloned into pETle (Novagen) over the 6-His tag via the
NcoI/BamHI sites; the construct was a gift from Dr. Rhea Hudson, Department of
Structural Biology and Biochemistry, Hospital for Sick Children, Toronto, Ontario,
Canada. A glycerol stock of E. coli BL21 (DE3) cells harboring the Abl-SH3-pETle
construct was used to inoculate 50 ml LB broth with 100 ug/ml ampicillin. After growth
at 37° for 16 h, this starter culture was used to inoculate l L of LB broth with 100 ug/ml
ampicillin. Once an optical density at 600 nm of 0.5-0.6 was reached, the cells were
induced with 0.3 mM IPTG and grown for an additional 6 h at room temperature before
harvesting by centrifugation. The cells were re-suspended in Buffer A (20 mM Tris pH
8.5) and exposed to a freeze/thaw cycle. After sonication, the lysate was cleared by
centrifugation at 10,000 x g for 20 min. The supernatant was filtered over a 30 kDa

centricon; then the flow-through was concentrated over a 5 kDa centricon. The sample

80

was purified by FPLC using a Resource Q anion exchange column (Amersham
Pharmacia Biotech) equilibrated with Buffer A. Protein was eluted over a 60 ml salt
gradient using Buffer B (20 mM Tris pH 8.5, l M NaCl). Fractions were analyzed by
SDS-PAGE. Those shown to be homogenous were pooled and concentrated to 10
mg/ml.

4.2.5. In vitro binding of MBP-SH3—RMD and the Abl-SH3 domain. The Abl-
SH3 domain was added to MBP-SH3-RMD at an approximately 1:1 molar ratio and
allowed to sit on ice for l h. Unbound Abl—SH3 domain was separated from the MBP-
SH3—RMDzAbl-SH3 domain complex by FPLC on a Sephacryl 8200 gel filtration
column (Amersham Biosciences) equilibrated with S200 Column Buffer (20 mM Tris pH
8.5, 300 mM NaCl, 10 mM maltose, 0.3% sarcosyl). Fractions were analyzed by SDS-
PAGE.

4.2.6. Crystallization of MBP-RMD, MBP-SH3-RMD and MBP-SH3—RMDzAbI-
SH3 domain complex. Purified MBP-RMD, MBP-SH3-RMD or MBP-SH3-RMDzAbl-
SH3 domain complex at 10 mg/ml was used to set up crystallization screens using the
microbatch method with the Impax 1-5 robot (Douglas Instruments). A total of 198
commercially available conditions were screened by combining 1 ul of purified protein
with 1 ul precipitating solution. Crystallization conditions were refined using the
handing drop vapor diffusion method. To aid in crystallization, 0.5 mM maltose, 5 mM

GDP and/or 2 mM NADP(H) were added to the protein sample.

4.3. Results and Discussion

4.3.1. Crystallization trials on the MBP-RMD chimera. Previous attempts at

obtaining high quality crystals of P. aeruginosa RMD have been unsuccessful (see

81

Chapter 2). Therefore, RMD became the trial protein on which to experiment new
strategies for expression, purification and crystallization. Though solubility is not a
problem with RMD, the large affinity tag MBP was chosen to create a fusion protein for
added stability. MBP is known to increase expression, assist in folding, protect proteins
from proteolysis and facilitate affinity purification over amylose resin. The MBP tag
should also change the crystallization environment as a fusion with RMD, hopefully
facilitating the growth of crystals of higher quality. In addition, the three dimensional
structure of MBP may allow for phase determination of the fusion protein by molecular
replacement or provide known locations of heavy atom-binding sites for phase
determination via multiple isomorphous replacement. The structures of E.coli MBP in
apo- [100] and maltose-bound l 101] forms have been reported.

The gene encoding P. aeruginosa RMD was initially cloned into pMal-C2 to check
for expression of the fusion protein. Purification over amylose resin yielded 8 mg of
purified protein per liter of cell culture that was 90-95% pure. Fractions analyzed by
SDS-PAGE (Figure 25a) show the molecular weight of the fusion protein to be roughly
80 kDa, which closely matches 76 kDa, the molecular weight predicted from the amino
acid sequence. Attempts were made to cleave MBP from the fusion protein using Factor
Xa. The cleavage was carried out at a ratio of 1 mg Factor Xa per 100 mg protein at
room temperature and was followed by SDS-PAGE (Figure 25b). The results indicated
that partial cleavage began at 0.5 h and was nearly 75% complete at 18 h. However, the
cost of the protease makes this process somewhat prohibitive for the amount of protein
needed for crystallization purposes. The expense, along with the tedious task of

optimizing the cleavage conditions as well as precipitation of the target protein are

82

common problems encountered during the cleavage step. To avoid the cleavage process,
re-purification and the loss of protein, the tag may be left on for crystallization trials.
This approach was attempted for the full-length MBP-RMD fusion; however, no crystals
were observed in any of the conditions. This is consistent with the results from the
crystallization trials of the gp21 ectodomain fragment that had also been cloned into
pMal—C2 [102], and is most likely due to the flexibility caused by the linker region.

A B

C 0 0.5 1.5 18
........ 193

193 p

102 I

    
  

60 5

7 232:3?

Figure 25. SDS-PAGE of samples from a MBP-RMD purification and Factor Xa
cleavage

(A) MBP-RMD amylose column purification: load (L), flow through (FT), wash, (W)
and fractions (F); (B) Factor Xa cleavage of MBP-RMD: control (C) and samples at 0,
0.5, 1.5 and 18 h. Molecular weight marker (M) standards shown in kDa to the left of
each gel (MBP, 41 kDa, MBP-RMD, 76 kDa).

4.3.2. Reconstruction of the linker region. The amino acid sequence of the linker
region in the original pMal-C2 vector is shown in Table 6. This represents the region
between MBP and the protein of interest, which is 22 amino acids in length and contains
a Factor Xa cleavage site at the C—terminus. The length of the linker is important for

some fusion proteins to bind tightly to the amylose resin. Since the binding affinity is

already only in the micromolar range, the separation of MBP to the protein of interest is

83

crucial to avoid any interactions that might block MBP from binding to the amylose.
However, the length of the linker introduces conformational heterogeneity. To address
this issue, the concept of a clamp or stabilizer was intriguing when considering how to
restrain the flexibility of the MBP-fusion protein. Antibody fragments have been used in
co-crystallization experiments for this purpose, specifically with membrane proteins with
some success. While antibodies can cover a portion of the hydrophobic surface of the
membrane protein converting it to a region more likely to make crystal contacts, they
may also restrain flexible regions, essentially acting as a clamp, stabilizing the
conformation of the protein. This can facilitate crystal growth and/or improve the quality
of the crystals. However, generating antibodies can be laborious. Plus, a more universal

clamp is desirable.

Table 6. Amino acid sequence of the linker region in pMaI-C2 and pMal-C2L3

 

JJMaI-CZ

 

SacI
malE ACT AAT TCG AGC TCG AAC AAC AAC AAC AAT AAC
MBP Thr Asn Ser Ser Ser Asn Asn Asn Asn Asn Asn

anl
AAT AAC AAC AAC CTC GGG ATC GAG GGA AGG ATT TCA
Asn Asn Asn Asn Leu Gly Ile Glu Gly Arg Ile Ser
Factor Xa]

 

pMal-C2L3

 

Sacl

malE ACT AAT TCG AGC TCG GCA GCA CCG ACT TAC AGC
MBP Thr Asn Ser Ser Ser Ala Ala Pro Thr Tyr Ser

an1
CCA CCA CCG CCG CCG GGA AGG ATT TCA
Pro Pro Pro Pro Pro Gly Arg Ile Ser
Factor Xa]

 

84

While working on a side project, the expression, purification and crystallization of
human phospholipid scramblase l (HuPLSCRl) (see Appendix A), an idea presented
itself. HuPLSCRl contains multiple proline-rich motifs resembling SH3 domain-binding
sites. SH3 domains are non-catalytic protein modules present in a wide variety of
unrelated proteins that can be independently expressed, purified and even crystallized.
HuPLSCRI is known to have marked specificity for the Abl-SH3 domain; in fact, studies
suggest that HuPLSCRl is the normal substrate of Abl tyrosine kinase I103]. The Abl-
SH3 domain was available in milligram quantities for the studies with HuPLSCRl.
Work with HuPLSCRl and the Abl-SH3 domain was the inspiration for the following
idea: replace the linker in the pMal-C2 vector with an SH3 domain-binding site and the
Abl-SH3 domain could be used as the clamp or stabilizer for crystallization trials of the
fusion protein. An added advantage is that the crystal structure of the proline-rich
peptide from the SH3-binding protein 3BP—l bound to the Abl-SH3 domain is available
[104].

In general, SH3 domains bind proline-rich peptides with affinities of 0.2-50 uM and
are quite promiscuous. Pisabarro and Serrano [105] rationally designed a peptide ligand
for the Abl-SH3 domain with high-affinity and specificity by mutating the 3BP—l peptide
(APTMPPPLPP). Mutations were chosen based on the tendency of the peptide to adopt a
polyproline helix II (PPII) conformation and the ability of the residues to form favorable
interactions to the Abl-SH3 domain. In doing so they created a peptide APTYSPPPPP
that binds the Abl-SH3 domain with an affinity 100 times higher than the original

peptide. This sequence was chosen for the linker region.

85

The structures of the reported MBP fusion proteins along with the structure of MBP,
which are shown in Figure 26, were consulted before deciding on the final linker
sequence. In all three cases, the linker region had been shortened. The linker region was
substituted with AAA for gp21, AAAEF for SarR and AAAAA for MATal. The
rationale behind the short AAA linker in gp21 was to connect the C-terminal helix of
MBP to the predicted N-terminal helix of gp21. However, instead of forming a
continuous helix, the short linker provides a 90° turn between the two. None of the three
structures showed that a continuous helix was necessary for forming a rigid connection,
so we were confident with the sequence chosen for the linker, which was not necessarily
optimized for such a linkage. The Sad site at the N—terminus of the pMal-C2 linker
region along with the anI site at the C-terminus was used to incorporate the SH3
domain-binding site linker-coding region (shown in Table 6) creating the new vector
pMal-C2L3.

The expected tertiary structure of MBP plus the SH3 domain-binding site in complex
with the Abl-SH3 domain is shown in Figure 27. The structure of MBP is taken from the
coordinates of the Protein Data Bank entry lanf. The break in the final helix of MBP is
predicted based on the structures of MBP, SarR and Matal; gp21 broke one residue
proximal. The linker region is the ten-residue proline—rich peptide (APT MPPPLPP) from
the SH3-binding protein 3BP-1, whose structure is known in complex with the Abl-SH3
domain [104]. Although the sequence of the peptide encoded by pMal-C2L3 is different
by three residues (AP'TYSPPPPP), the Abl-SH3 domain should bind to the peptide in a
similar fashion based on modeling work of Pisabarro and Serrano [105]. The very C-

terminal portion of the linker marks the beginning of the target protein. One would

86

expect that MBP would assist in the formation of the tertiary structure of the target
protein. However, there is certainly a concern for the potential affect of MBP and/or the
Abl—SH3 domain on the quaternary structure of the target protein. The three MBP-fusion
protein structures that have been solved are all in their biologically relevant quaternary
state. Our hope is that our protein will assume its biologically relevant quaternary state
before adding the Abl-SH3 domain, and the addition of the peptide will stabilize the

overall structure to facilitate crystallization.

 

Figure 26. Crystal structures of MBP and MBP-fusion proteins

(A) MBP, (B) MBP-ngl, (C) MBP-Mata] and (D) MBP-SarR. MBP is shown in the
same orientation in each panel. The asterisks in panels B, C and D highlight the linker
region.

87

 

Figure 27. Model of MBP-SH3 binding domain site:AbI—SH3 domain complex
The MBP portion is rainbow-colored. An asterisk denotes the beginning of the linker

region, or the SH3-binding domain site, colored in red. The Abl-SH3 domain is colored
in blue.

4.3.3. Crystals of the suspected MBP-SH3-RMD:Abl-SH3 domain complex.
RMD was cloned into the newly constructed vector pMal-C2L3. Expression and
purification trials were performed on both RMD-pMal-C2L3 as well as the control pMal-
C2L3. SDS-PAGE analysis of samples from the amylose column are shown in Figure
28. MBP-SH3-RMD elution fractions run around 80 kDa, which matches closely to 78
kDa, the predicted molecular weight based on the amino acid sequence. MBP-SH3

elution fractions run around 45 kDa, which matches closely to its predicted molecular

88

weight of 43 kDa. While pMal-C2L3 yielded 20 mg of purified MBP—SH3 per liter of
cell culture, RMD-pMAL-C2L3 yielded only 8 mg of purified MBP—SH3—RMD per liter
of cell culture. This was most likely due in part to an unexpected result: the fusion
protein tended to aggregate. This is contrary to the results of other proteins expressed as
fusions with MBP, which tended to become more soluble upon the addition of MBP.
Nonetheless, it was obvious that the buffer would need some tweaking, i.e. addition of
detergent, to keep the complex soluble. Several detergents were tested; the best seemed
to be sarcosyl at a concentration of 0.3%. Also, it was noted that increasing the pH of the

Tris buffer from 7.4 to 8.5 helped keep the protein from aggregating.

      
 
    

L W F1. F2 F3, Fi.._..F,5 F6 F7
102

60

41

 

Figure 28. SDS-PAGE of samples from a typical MBP-SH3 and MBP-SH3—RMD
purification

(A) MBP-SH3 amylose column purification: load (L), wash (W) and fractions (F); (B)
MBP-SH3-RMD amylose column purification: load (L), wash (W) and fractions (F).
Molecular weight marker (M) standards shown in kDa to the left of each gel.

The gene encoding the Abl-SH3 domain had been cloned into the pE'I‘le vector over

the code for the 6-His tag, so it was purified over an anion exchange column rather than a

Ni-NTA column (Figure 29). Before loading the column, the sample was filtered over a

89

30 kDa centricon to decrease the amount of unwanted protein. Although this step
resulted in the loss of a significant amount of desired protein along with the unwanted
protein (see lane 3 of the SDS—PAGE gel in Figure 29), the sample in the end was more
pure. The purification process for the Abl-SH3 domain resulted in about 1.5 mg of

purified protein per liter of cell culture.

       

Figure 29. SDS-PAGE of samples from a typical Abl-SH3 domain purification
(1) Cell lysate, (2) 30 kDa centricon flow-through, (3) 30 kDa centricon retentate, (4)

column load (30 kDa centricon flow-through concentrated to 2 ml) and (5-9) fractions.
Molecular weight marker (M) standards shown in kDa to the right of the gel. The Abl-
SH3 domain is 6.8 kDa based on the amino acid sequence.

It is known that the SH3 binding—domain peptide encoded in the linker region of
pMal-C2L3 binds the Abl-SH3 domain with the affinity of Kd=0.4 uM [105]. The Abl-
SH3 domain was added to the pool of the MBP—SH3-RMD fractions from the amylose
resin at a 1:1 molar ratio and allowed to bind before running a gel filtration column. The
results from the 5200 gel filtration column are seen in Figure 30. A nice, single peak of

the complex is seen in the column trace, along with a smaller peak corresponding to

MBP. The Abl—SH3 domain does not show up on the SDS—PAGE gel shown most likely

90

because it is very low in concentration. Not having an antibody available made this

diffith to prove definitively.

ML12345678

lI 193 -
‘ 102 .-
l 60 - ' _--- '-
41 .. g ~—
8
N 27 on.
<
20 h.-
" t 15 an.
7 I

Figure 30. Gel filtration analysis of MBP-SH3-RMDzAhl-SH3 domain
SDS-PAGE samples: load (L), lanes 1-5 are fractions across the first peak and lanes 6-8

are fractions across the second peak. Molecular weight marker (M) standards shown in
kDa to the left of the gel.

Crystallization trials were set up with MBP-SH3-RMD, the Abl-SH3 domain and the
complex. While crystals were observed in a small number of conditions for the Abl-SH3
domain, none were observed for MBP-SH3-RMD alone. When crystallization trials were
set up with the complex, needles and small, plate-like crystals were observed (Figure 31).
The best results occurred upon the addition of 5 mM GDP and 2 mM NADP to the
protein sample before the trays were set up. Since MBP adopts a different conformation
upon binding maltose, 0.5 mM maltose was used in some of the trials as a variable in the
protein sample. However, no obvious differences were observed with or without it in the
sample. This is most likely because MBP had retained the maltose molecules during

purification. This was observed in the structures of MBP-gp21 and MBP-SarR where

MBP was seen in a closed conformation [95, 96]. In contrast, MBP was observed in an

91

open conformation in the MBP-Mata] structure, most likely because maltose was
stripped during purification over a strong cation exchange column [97]. The
crystallization condition of the best—formed crystals was 30% PEG 400, 100 mM sodium
acetate trihydrate pH 4.6 and 100 mM cadmium chloride dihydrate. This condition is
consistent with the conditions of the other MBP-fusion proteins: PEGs seem to be the
most successful precipitants, and the pH of the buffer in the solution is generally low.
Although several rounds of optimization trials were completed, reproducibility of the
crystals of the suspected complex was clearly a problem. In addition, the crystals

obtained were too small for x-ray analysis.

 

 

 

 

  

 

 

Figure 31. Crystals of MBP-SH3-RMD:AbI-SH3

Panel A: needle-like crystals in 28% PEG 400, 100 mM HEPES pH 7.5, 200 mM CaClz
dihydrate. Panel B: plate-like crystals in 18% PEG 8K, 100 mM sodium cacodylate pH
6.5. Panel C: rod-shaped crystals in 30% PEG 400, 100 mM sodium acetate trihydrate
pH 4.6, 100 mM cadmium chloride dihydrate.

92

4.3.4 Conclusions. To conclude, we have presented here a strategy of employing (l)
a large affinity tag, MBP, to assist in the expression and purification, along with (2) a
stabilizer protein, the Abl-SH3 domain, to assist in the crystallization of problematic
proteins. While many of the other strategies to improve expression, purification and/or
crystallization require the modification of the protein itself (i.e. point mutations,
truncations, etc.), this strategy was designed so that the protein of interest is left
unhindered in its secondary, tertiary and most likely quaternary structure formation.
Although reproducibility has been a problem, we were successful in obtaining crystals
upon the addition of the stabilizer protein, the Abl-SH3 domain, to the MBP—SH3-RMD
fusion protein. It appears the Abl-SH3 domain is providing rigidity to the MBP-fusion
protein, which did not crystallize on its own. This strategy may facilitate the expression,
purification and crystallization of other proteins that have proven to be difficult due to
solubility or stability problems.

Though MBP has certainly shown its advantages, it may have its limitations based on
the size of protein/peptide to which it is fused. The three fusion protein structures that
have been solved are all small in comparison to the size of MBP; gp21 has 88 residues,
SarR has 115 residues and MaTal has 50 residues compared to MBP, which has 368
residues. This is in contrast to our trial protein RMD, which has 313 residues. The
dominance of MBP over smaller proteins may direct crystal packing and more easily
facilitate structure determination by molecular replacement. Therefore, this method may
be more conducive to small proteins. Since it seems impractical to keep up a 3:1 tag to

protein size ratio (based on the largest fusion protein structure solved, SarR), perhaps a

93

different strategy would work for larger proteins. Work on new strategies in our lab is

ongoing.

94

CHAPTER 5: FUTURE DIRECTIONS

' Is the cofactor necessary for GMD to form a tetramer?

It was shown that unlike the typical homodimeric SDR structures, GMD crystallizes
as a tetramer where the cofactor is intimately involved in the tetramer interface. It would
be interesting to see if removal of the cofactor causes the tetramer to fall apart. To test
for this, one would need to remove the cofactor, which has proven to be difficult to do in
the case of P. aeruginosa GMD and A. thaliana MURl. A protocol has been reported for
stripping CDP-D-glucose 4,6-dehydratase of its cofactor by dialyzing it against phosphate
buffer containing potassium bromide for several days [71]. The same protocol coUId be
used on GMD to produce the apoenzyme and the process could by followed by gel
filtration experiments.

' Can the substrate specificity of GMD be determined and/or altered?

Efforts could be made to determine substrate specificity. It does not appear that
GMDs are particular about the sugar moiety they accept. They are able to bind their
natural substrate and intermediates, as well as GDP-D-rhamnose [5], GDP-D-fucose [2,
51, 52] and GDP-D-glucose [6]. However, GMDs are considerably more particular about
nucleotide specificity. This is supported by work completed on porcine thyroid GMD
where the following list of nucleotides were tested as potential inhibitors: GMP, GDP,
GTP, AMP, ADP, ATP, CMP, CDP, CT P, UMP, UDP UTP and ITP. Only GDP and
GTP showed any appreciable inhibitory action [6]. Initial look at the interactions show
that several residues in contact with GDP are conserved across GMDs: Asnl79, Lysl93,
Arg218, Arg279 and Glu282. Mutations on these residues could be generated to see

which are necessary for GDP binding. The activity of the resulting mutants could be

95

tested via CE (protocol presented in Chapter 3) and the structures could be studied via X-
ray crystallography. It would also be interesting to see whether or not the substrate-
binding site could be altered so that GMD could accept other nucleotide diphosphates or
NDP-sugars. If GMD could be modified to accept different substrates, it opens up the
potential to utilize GMD in making alternate deoxyhexoses for the synthesis of
glycoconjugates.
0 Do the structural characteristics of RMD place it into the SDR protein family?

Specific stretches of conserved amino acid sequences (i.e. G-XX-G-XX-G, S/T, Y-
XXX-L) suggest that RMD fits into the SDR protein family. It will be interesting to see
if the monomeric structure reveals that it folds into two domains, the N—terminal cofactor
binding domain and the C-terminal substrate binding domain, like the NDP-sugar
modifying subfamily of SDRs. X-ray structure determination using the data set from
AtRMD is currently in progress. A respectable molecular replacement solution has been
obtained using coordinates from structure of GMD from P. aeruginosa, suggesting that
the core structure of RMD will be similar to the SDR protein structures. Certainly the
first order of business is to perform model building/refinement on the molecular
replacement solution for AtRMD to check the solution. In addition, heavy-atom
derivatives will be prepared for use in experimental phase determination by isomorphous
replacement for comparison to those determined by the molecular replacement method.
0 What residues are involved in the active site of RMD?

AtRMD crystals were obtained in the presence of NADPH and product analog GDP—
D-mannose. Hopefully GDP-D-mannose or at least GDP (as seen in GMD) is bound in

the active site. The 4—keto-6-deoxy intermediate could also be modeled in using MURl

96

as a guide (as in P. aeruginosa GMD). Our hope is to get a picture of the GDP-sugar
bound in the active site to get an idea of what residues may be involved in substrate
binding and/or catalysis. Furthermore, we hope to see if the GDP moiety is bound in a
syn conformation, as seen in GMD and MURl, or the more commonly seen anti
conformation. The ability of the nucleotide to adopt this unusual, strained conformation
may be related to the substrate specificity for GMD or RMD enzymes.

0 Will the structures of AtRMD and PaRMD be comparable?

As previously mentioned, a data set was also collected on one of the PaRMD crystals.
The diffraction data is only 3.7 A and phase determination by the molecular replacement
method has been unsuccessful. However, AtRMD may serve as a better search model.
Although the resolution of the data will not be able to reveal much beyond the backbone
trace, perhaps we will be able to see evidence as to why the crystals diffract so poorly
compared to AtRMD (i.e. flexible loops, surface areas interfering with crystal packing,
etc.). This may lead to ideas for modifying the PaRMD in hopes to obtaining a version of
the protein that would provide better diffracting crystals.

0 What? GMD is bifunctional?

Recombinant GMD was able to convert GDP-D—mannose to GDP-D-rhamnose,
exhibiting both dehydratase and reductase activities. However, this raises several
questions. (1) Is there a minimal amount of reduced cofactor where GMD is still
bifunctional? Experiments were performed in excess NADPH conditions, far in excess
of natural conditions in vivo. The CE-based assays could be completed in various
NADPH concentrations to determine the minimal concentration necessary. (2) Can it be

assumed that the GMD tetramer falls apart to bind the reduced form of the cofactor? The

97

catalytic mechanism of GMD indicates that the NADP cofactor is regenerated with each
round of catalysis and remains tightly bound. It would be interesting to follow the
oligomeric state through these experiments via gel filtration, dynamic light scattering or
ultracentrifugation, and also determine if GMD possesses reductase activity as a dimer or
tetramer. (3) Last, but certainly not least, does GMD possess bifunctionality in vivo?
Our collaborators at the University of Guelph have a P. aeruginosa A-band LPS
knockout mutant available. The plasmid containing the gmd gene could be transformed
into the bacterium and the LPS analyzed for the presence of D-rhamnose to see if GMD

saves the less virulent phenotype.

98

APPENDIX

99

APPENDIX A: MBP-HuPLSCRl AND THE Abl-SH3 DOMAIN

A.l. Introduction

Human phospholipid (PL) scramblase I (HuPLSCRl) is a plasma membrane protein
that has been implicated in the transbilayer movement of plasma membrane PLs upon the
increase of intracellular Ca2+ due to cell activation, cell injury or apoptosis [106]. The
newly surface-exposed PLs play a central role in promoting blood coagulation and have
been implicated in the clearance of injured or apoptotic cells. HuPLSCRl is a 35 kDa
type 2 membrane protein with a single transmembrane segment near the C-terminus. An
EF-hand-like Cab-binding segment immediately precedes the transmembrane domain.
The cytoplasmic domain also includes multiple proline-rich motifs resembling Src
homology 3 (SH3) domain-binding sites. HuPLSCRl has marked specificity for binding
to the SH3 domain of Abelson [99] tyrosine kinase, which phosphorylates Tyr residues of
HuPLSCRl [103]. In fact, studies suggest that HuPLSCRl is the normal substrate of Abl
tyrosine kinase [103].

The DNA encoding HuPLSCRl and the Abl-SH3 domain were given to our lab for
the purpose of crystallizing the proteins that they encode. The original goal of this
project was to obtain the crystal structures of the MBP-HuPLSCR] fusion protein in
complex with the Abl-SH3 domain. This appendix contains a short summary of the work
that was completed to express, purify, crystallize and test the binding of MBP-
HuPLSCRl and the Abl-SH3 domain. This work was the inspiration that led to the
strategy for expression, purification and crystallization of problematic proteins presented

in Chapter 4.

100

A.2. Experimental Procedures

A.2.l. Expression and Purification of HuPLSCRl. The cDNA encoding
HuPLSCRl had been cloned into pMAL-C2 (New England Biolabs) as previously
reported [106] and provided to our lab from Dr. Peter Sims of Scripps Research Institute,
La Jolla, California. The expression protocol in section 4.2.3. was used to express MBP-
HuPLSCRl. The same protocol in section 4.2.3. for running the amylose column was
used for the first step in the purification of MBP-HuPLSCRl. Further purification was
carried out by FPLC using a HiTrap” Q anion exchange column (Amersham Pharmacia
Biotech) equilibrated with Buffer A (20 mM Tris pH 8.5). Protein was eluted over a 40
ml salt gradient using Buffer B (20 mM Tris pH 8.5, l M NaCl). Fractions were
analyzed by SDS-PAGE.

A.2.2. Expression and Purification of the Abl-SH3 domain. The Abl-SH3 domain
was expressed and purified using the protocol under section 4.2.4.

A.2.3. In vitro binding of MBP-HuPLSCRl to the Abl-SH3 domain. For binding,
100 pl of MBP-HuPLSCRl (10 mg/ml) was mixed with 10 pl of the Abl-SH3 domain
(10 mg/ml) for an approximately 1:] molar ratio. The mixture was allowed to sit on ice
for 2 h, and run over a Sephadex G-50 spin column to clear any excess Abl-SH3 domain;
then concentrated over a 30 kDa centricon. The samples were analyzed by an SDS-
PAGEgd.

A.2.4. Crystallization of MBP-HuPLSCRl, the Abl-SH3 domain and MBP-
HuPLSCRlebl-SH3 domain complex. Purified MBP-HuPLSCRl, the Abl-SH3
domain or MBP-HuPLSCRlebl-SH3 domain complex at 10 mg/ml was used to set up

crystallization screens using the microbatch method with the Impax I-5 robot (Douglas

101

Instruments). A total of I98 commercially available conditions were screened by
combining 1 pl of purified protein with 1 pl precipitating solution. Crystallization
conditions were refined using the handing drop vapor diffusion method. To aid in

crystallization, 0.5 mM maltose and/or 0.5 mM CaCl2 were added to the protein sample.

A.3. Results

Active HuPLSCRl had been previously cloned, expressed and purified as an MBP-
fusion protein [106]. A similar expression and amylose column purification scheme was
used. An additional step of purification over a HiTrap Q anion exchange column was
added in preparation for crystallization trials. The purification was followed by SDS-
PAGE (Figure 32). About 3-4 mg of purified MBP-HuPLSCRl (78 kDa) was obtained
from 1 L of cell culture. Results from the Abl—SH3 domain expression and purification

can be found in section 4.3.3 and Figure 29.

 

B
M LFT WFl F2 F3 F4 F5 F6“. M L FT F1 F2 F3 F4 F5 F6 F7
193 ~ 193 u
102 102 w
.;‘¢:-_: "n “ . . _“*
. “w veer ~
41 41 b ‘ '.
27 27 .b ,5
20 ' 20 in 1?

 

Figure 32. SDS-PAGE of samples from a typical MBP-HuPLSCRl purification

(A) Samples from the amylose column purification: load (L), flow through (FT), wash
(W) and elution fractions (F). (B) Samples from the anion exchange column purification:
load (L), flow through (FT), elution fractions (F). Molecular weight marker (M)
standards shown in kDa to the left of each gel.

102

Immunoprecipitation experiments have shown that HuPLSCRl interacts with Abl
tyrosine kinase, most likely through the interaction of the SH3 domain of Abl with one or
more of the proline-rich domains of HuPLSCRI [103]. Purified MBP-HuPLSCRl was
bound to the purified Abl—SH3 domain at a 1:1 molar ratio and run over a gel filtration
column to clear any excess Abl-SH3 domain. The complex was then concentrated using
a 30 kDa centricon. The process was followed by SDS-PAGE (Figure 33). MBP-
HuPLSCRl (78 kDa) and the Abl-SH3 domain (6 kDa) are seen in Lane 3 as the mixture
before Sephadex G-50 clarification and concentration over a 30 kDa centricon. Lane 4
shows the Sephadex G-50 flow through. After concentration of the flow through over a
30 kDa centricon, the Abl-SH3 domain can clearly be seen (Lane 5). As a control, Lane
7 shows the Abl-SH3 domain at the same concentration as that in Lane 3. Once certain
that MBP-HuPLSCRl was binding the Abl—SH3 domain, crystallization trials were set up
with the complex as well as MBP-HuPLSCRl and the Abl-SH3 domain as controls.
Small crystals were seen with all three samples; of note, thin rods were seen in differing
conditions for MBP-HuPLSCRl and the complex (data not shown). However,
reproducibility was a problem. This may be for one or both of the following reasons: (I)
lack of rigidity of the fusion protein due to the 22 amino acid linker between MBP and
HuPLSCRl and/or (2) heterogeneity of the complex due to the presence of two SH3
domain-binding sites at the N-terminus of HuPLSCRl, where the Abl-SH3 domain could

bind one or the other or both.

103

I93

102

60

41

27

 

Figure 33. SDS-PAGE of MBP-HuPLSCRl/Abl-SH3 domain binding

(1) 1:1 molar ratio of MBP-scarmblasezAbl-SH3; (2) the sample in (1) concentrated by
half; (3) the gel filtration flow through; (4) the 30 kDa retentate; (5) the 30 kDa flow
through; (6).the Abl-SH3 domain at a comparable concentration to that of (2). MBP-
HuPLSCRl is 78 kDa and the Abl-SH3 domain is 6 kDa. Molecular weight marker (M)
standards shown in kDa to the left of the gel.

To address the flexibility of the system, work was completed to cleave the linker and
the MBP tag from HuPLSCRl using Factor Xa. However, as mentioned in Chapter 4,
the protease cleavage step is expensive and optimization is tedious; also in this case, it
added heterogeneity to the system because cleavage was incomplete. Although we would
have liked to pursue the structure of the HuPLSCRI:Abl-SH3 domain complex with or
without MBP, it was apparent that it was problematic. Still interested in the structure of
HuPLSCRl alone, a different approach was considered. An alignment of HuPLSCRl to
HuPLSCR2 and two mouse homologs MuPLSCRl and MuPLSCR2 indicates that the

significant similarities between the amino acid sequences starts at residue number 85

104

(numbering based on HuPLSCRl). In fact, HuPLSCR2 lacks the entire N-terminal
portion containing the proline-rich SH3 domain-binding sites. Work has been started to
shorten HuPLSCRl by 85 residues and re-clone it into an expression vector in hopes of

obtaining a version of the protein that will crystallize.

105

 

LITERATURE CITED

106

LITERATURE CITED

1. Yamamoto, K., Katayama, I., Onoda, Y., Inarni, M., Kumagai, H. & Tochikura, T.
(1993) Evidence that the enzyme catalyzing the conversion of guanosine diphosphate D-
mannose to a 4-keto sugar nucleotide intermediate requires nicotinamide adenine
dinucleotide phosphate, Arch. Biochem. Biophys. 300, 694-698.

2. Sturla, L., Bisso, A., Zanardi, D., Benatti, U., De Flora, A. & Tonetti, M. (1997)
Expression, purification and characterization of GDP-D-mannose 4,6-dehydratase from
Escherichia coli, FEBS Lett. 412, 126-130.

3. Kneidinger, B., Graninger, M., Adams, G., Puchberger, M., Kosma, P., Zayni, S. &
Messner, P. (2001) Identification of two GDP-6-deoxy-D-lyxo-4-hexulose reductases
synthesizing GDP-D-rhamnose in Aneurinibacillus thermoaerophilus L420-91T, J. Biol.
Chem. 276, 5577-5583.

4. Wu, 3., Zhang, Y. & Wang, P. G. (2001) Identification and characterization of GDP-
D-mannose 4,6-dehydratase and GDP-L-fucose synthetase in a GDP-L-fucose
biosynthetic gene cluster from Helicobacter pylori, Biochem. Biophys. Res. Commun.

285, 364-371.

5. Bonin, C. P., Potter, 1., Vanzin, G. F. & Reiter, W.-D. (1997) The MURI gene of
Arabidopsis thaliana encodes an isoform of GDP-D-mannose-4,6-dehydratase,
catalyzing the first step in the de novo synthesis of GDP-L-fucose, Proc. Natl. Acad. Sci.
USA. 94, 2085-2090.

6. Broschat, K. 0., Chang, S. & Serif, G. (1985) Purification and characterization of
GDP-D-mannose 4,6-dehydratase from porcine thyroid, Eur. J Biochem. 153, 397-401.

7. Sullivan, F. X., Kumar, R., Kriz, R., Stahl, M. L., Xu, G.-Y., Rouse, 1., Chang, X.-j.,
Boodhoo, A., Potvin, B. & Cumming, D. A. (1998) Molecular cloning of human GDP-

mannose 4,6-dehydratase and reconstitution of GDP-fucose biosynthesis in vitro, J. Biol.
Chem. 273, 8193-8202.

8. Rocchetta, H. L., Burrows, L. L. & Lam, J. S. (1999) Genetics of O-antigen
biosynthesis in Pseudomonas aeruginosa, Microbiol. Mol. Biol. Rev. 63, 523-553.

9. Molicaro, A., Silip, A., Lanzetta, R., Newman, M. A., Dow, J. M. & Parrilli, M.
(2003) Structural elucidation of the O-chain of the lipopolysaccharide from Xanthomonas
campestris strain 8004, Carbohydr. Res. 338, 277-281.

10. Kocharova, N. A., Knirel, Y. A., Widmalm, G., Jansson, P. E. & Moran, A. P. (2000)

Structure of an atypical O-antigen polysaccharide of Helicobacter pylori containing a
novel monosaccharide 3-C-methyl-D-mannose, Biochemistry. 39, 4755-4760.

107

ll. Yokata, S., Kaya, S., Sawada, S., Kawamura, T., Araki, Y. & Ito, E. (1987)
Characterization of a polysaccharide component of lipopolysaccharide from
Pseudomonas aeruginosa IID 1008 (ATCC 27584) as D-rhamnan, Eur. J. Biochem. 167,
203-209.

12. Ginsburg, V. (1960) Formation of guanosine diphosphate L-fucose from guanosine
diphosphate D-mannose, J. Biol. Chem. 253, 2196-2201.

13. Liao, T.-H. & Barber, G. A. (1972) Purification of guanosine 5'-diphosphate D-
mannose oxidoreductase from Phaseolus vulgaris, Biochim. Biophys. Acta. 70, 794-798.

14. Overton, K. & Serif, G. (1981) Synthesis of L-fucose in thyroid tissue, Biochim.
Biophys. Acta. 675, 281-284.

15. Lopez-Lara, I. M., Blok-Tip, L., Quinto, C., Garcia, M. L., Bloemberg, G. V.,
Lamers, G. E., Lugtenberg, B. J., Thomas-Oates, J. & Spaink, H. P. (1996) NodZ of
Bradyrhizobium extends the nodulation host range of Rhizobium by adding a fucosyl
residue to nodulation signals, Mol. Microbiol. 21, 397-408.

16. Mergaert, P., Van Mantago, M. & Holsters, M. (1997) The modulation gene nolK of
Azorizobium caulinodans is involved in the formation of GDP-fucose from GDP-
mannose, FEBS Lett. 409, 312-316.

17. Varki, A. (1994) Selectin ligands, Proc. Natl. Acad. Sci. USA. 91, 7390-7497.

18. Maki, M., Jarvinen, N., Rabina, J., Maaheimo, H., Mattila, P. & Renkonen, R. (2003)
Cloning and functional expression of a novel GDP-6-deoxy-D-talose synthetase from
Actinobacillus actinomycetemcomitans, Glycobiology. 4, 295-303.

19. Alberrnann, C. & Piepersberg, W. (2001) Expression and identification of the beE
protein from V ibrio cholerae 01 and its use for the enzymes synthesis of GDP-D-
perosamine, Glycobiology. 11, 655-661.

20. Suzuki, N., Nakano, Y., Yoshido, Y., Nezu, T., Terada, Y., Yamashita, Y. & Koga,
T. (2002) Guanosine diphosphate-4-keto-6-deoxy-D-mannose reductase in the pathway
for the synthesis of GDP-6-deoxy-D-talose in Actinobacillus actinomycetemcomitans,
Eur. J. Biochem. 269, 5963-5971.

21. Allard, S. T. M., Giraud, M.-F., Whitfield, C., Graninger, M., Messner, P. &
Naismith, J. H. (2001) The crystal structure of dTDP-D-glucose 4,6-dehydratase (leB)

from Salmonell enterica serovar typhimurium, the second enzyme in the dTDP-L-
rhamnose pathway, J. Mol. Biol. 307, 283-295.

22. Allard, S. T. M., Giraud, M.-F., Whitfield, C., Messner, P. & Naismith, J. H. (2000)
The purification, crystallization and structural elucidation of dTDP-D-glucose 4,6-

108

dehydratase (leB), the second enzyme of the dTDP-L-rhamnose synthesis pathway
from Salmonella enterica serovar typhimurium, Acta Crystallogr. D. 56, 222-225.

23. Giraud, M.-F., Leonard, G. A., Field, R. A., Berlind, C. & Naismith, J. H. (2000)
leC, the third enzyme of dTDP-L-rhamnose pathway, is a new class of epimerase, Nat.
Struct. Biol. 7, 398-402.

24. Giraud, M.-F., McMiken, H. J., Leonard, G. A., Messner, P., Whitfield, C. &
Naismith, J. H. (1999) Overexpression, purification, crystallization and preliminary
structural study of dTDP-6-deoxy-L-lyxo-4-hexulose reductase (RmID), the fourth

enzyme of the dTDP-L-rhamnose synthesis pathway, from Salmonella enterica serovar
typhimurium, Acta Crystallogr. D. 55, 2043-2046.

25. Graninger, M., Nidetzky, B., Heinrichs, D. E., Whitfield, C. & Messner, P. (1999)
Characterization of dTDP-4-dehydrorhamnose 3,5-epimerase and dTDP-4-
dehydrorhamnose reductase, required for dTDP-L-rhamnose biosynthesis in Salmonella
enterica Serovar Typhimurium LT2, J. Biol. Chem. 274, 25069-25077.

26. Blankenfeldt, W., Kerr, 1. D., Giraud, M.-F., McMiken, H. J., Leonard, G. A.,
Whitfield, C., Messner, P., Graninger, M. & Naismith, J. H. (2002) Variation on a theme
of SDR: dTDP-6-deoxy-L-Iyxo-4-hexulose reductase (leD) shows a new Mg”-
dependent dimerization mode, Structure. 10, 773-786.

27. Gross, J. W., Hegeman, A. D., Gerratana, B. & Frey, P. A. (2001) Dehydration is
catalyzed by glutamate-136 and aspartic acid-135 active site residues in Escherichia coli
dTDP-glucose 4,6-dehydratase, Biochemistry. 40, 12497-12504.

28. Gross, J. W., Hegeman, A. D., Vestling, M. M. & Frey, P. A. (2000) Pre-steady-state
kinetics of deoxythymidine diphosphate glucose 4,6-dehydratase: detection and
quantification of reaction intermediates by mass spectroscopy, Biochemistry. 39, 1559.

29. Gross, J. W., Hegeman, A. D., Vestling, M. M. & Frey, P. A. (2000)
Characterization of enzymatic processes by rapid mix-quench mass spectrometry: the
case of dTDP-glucose 4,6-dehydratase, Biochemistry. 39, 13633-13640.

30. Hegeman, A. D., Gross, J. W. & Frey, P. A. (2001) Probing catalysis by Escherichia
coli dTDP-glucose-4,6-dehydratase: identification and preliminary characterization of
functional amino acid residues at the active site, Biochemistry. 40, 6598-6610.

31. Hegeman, A. D., Gross, J. W. & Frey, P. A. (2002) Concerted and stepwise
dehydration mechanisms observed in wild-type and mutated Escherichia coli dTDP-
glucose 4,6-dehydratase, Biochemistry. 41, 2797-2804.

32. Gerratana, B., Cleland, W. W. & Frey, P. A. (2001) Mechanistic roles of Thr134,

Tyr160, and Lysl64 in the reaction catalyzed by dTDP-glucose 4,6-dehydratase,
Biochemistry. 40, 9187-9195.

109

33. Maki, M. & Renkonen, R. (2004) Biosynthesis of 6-deoxyhexose glycans in bacteria,
Glycobiology. 14, lR-15R.

34. Nakano, Y., Suzuki, N., Yoshida, Y., Nezu, T., Yamashita, Y. & Koga, T. (2000)
Thymidine diphosphate-6-deoxy-L-lyxo-4-hexulose reductase synthesizing dTDP-6-

deoxy-L-talose from Actinobacillus actinomycetemcomitans, J. Biol. Chem. 275, 6806-
6812.

35. Yoshida, Y., Nakano, Y., Nezu, T., Yamashita, Y. & Koga, T. (1999) A novel NDP-
6-deoxyhexosyl-4-ulose reductase in the pathway for the synthesis of thymidine
diphosphate-D-fucose, J. Biol. Chem. 274, 16933-16939.

36. Liu, H.-w. & Thorson, J. S. (1994) Pathways and mechanisms in the biogenesis of
novel deoxysugars by bacteria, Annu. Rev. Microbiol. 48, 223-256.

37. Hallis, T. M., Lei, Y., Que, N. L. S. & Liu, H.-w. (1998) Mechanistic studies of the
biosynthesis of paratose: purification and characterization of CDP-paratose synthase,
Biochemistry. 3 7, 4935-4945.

38. Verma, N. & Reeves, P. R. (1989) Identification and sequence of rbe and rjbE,
which determine antigenic specificity of group A and group D salmonellae, J. Bacteriol.
171, 5694-701.

39. Wyk, P. & Reeves, P. R. (1989) Identification and sequence of the gene for abequose
synthase, which confers antigenic specificity on group B salmonellae: homology with
galactose epimerase, J. Bacteriol. 171, 5687-93.

40. Koropatkin, N. M., Liu, H.-w. & Holden, H. M. (2003) High resolution X-ray
structure of tyvelose epimerase from Salmonella typhi, J. Biol. Chem. 278, 20874-20881.

41. Liu, Y., Thoden, J. B., Kim, J., Berger, E., Gulick, A. M., Ruzicka, F. J., Holden, H.
M. & Frey, P. A. (1997) Mechanistic roles of tyrosine 149 and serine 124 in UDP-
galactose 4-epimerase from Escherichia coli, Biochemistry. 36, 10675-10684.

42. Allard, S. T. M., Beis, K., Giraud, M.-F., Hegeman, A. D., Gross, J. W., Wilmouth,
R. C., Whitfield, C., Graninger, M., Messner, P., Allen, A. G., Maskell, D. J. & Naismith,
J. H. (2002) Toward a structural understanding of the dehydratase mechanism, Structure.
10, 81-92.

43. Mulichak, A. M., Theisen, M. J ., Essigmann, B., Benning, C. & Garavito, R. M.
(1999) Crystal structure of SQDI , an enzyme involved in the biosynthesis of the plant
sulfolipid headgroup donor UDP-sulfoquinovose, Proc. Natl. Acad. Sci. USA. 96, 13097-
13102.

110

44. Mulichak, A. M., Bonin, C. P., Reiter, W.-D. & Garavito, R. M. (2002) Structure of
MURl GDP-mannose 4,6-dehydratase from Arabidopsis thaliana: implications for
ligand binding specificity, Biochemistry. 41, 15578-15589.

45. Thoden, J. B., Wohlers, T. M., Fridovich-Keil, J. L. & Holden, H. M. (2000)
Crystallographic evidence for Tyr 157 functioning as the active site base in human UDP-
galactose 4-epimerase, Biochemistry. 39, 5691-5 701.

46. Cryz, S. J., Pitt, T. L., Furer, E. & Germanier, R. (1984) Role of lipopolysaccharide
in virulence of Pseudomonas aeruginosa, Infect. Immun. 44, 508-513.

47. Engels, W., Endert, J., Kamps, M. A. & van Boven, C. P. (1985) Role of
lipopolysaccharide in opsonization and phagocytosis of Pseudomonas aeruginosa, Infect.
Immun. 49, 182-189.

48. Dasgupta, R., de Kievit, T. R., Masoud, H., Altman, E., Richards, J. C., Sadovskaya,
I., Speert, D. P. & Lam, J. S. (1994) Characterization of lipopolysaccharide-deficient
mutants of Pseudomonas aeruginosa derived from serotypes O3, O5, and O6, Infect.

Immun. 62, 809-817.

49. Markovitz, A. (1964) Biosynthesis of guanosine diphosphate D-rhamnose and
guanosine diphosphate D-talomethylose from guanosine diphosphate a-D-Mannose, J.
Biol. Chem. 239, 2091-2098.

50. Etzioni, A., F rydman, M., Pollack, S., Avidor, 1., Phillips, M. L., Paulson, J. C. &
Gershoni-Baruch, R. (1992) Brief report: recurrent severe infections caused by a novel
leukocyte adhesion deficiency, N Engl. J. Med 327, 1789-1792.

51. Bisso, A., Sturla, L., Zanardi, D., De Flora, A. & Tonetti, M. (1999) Structural and
enzymatic characterization of human recombinant GDP-D-mannose-4,6-dehydratase,
FEBS Lett. 456, 370-374.

52. Somoza, J. R., Menon, S., Schmidt, H., Joseph-McCarthy, D., Dessen, A., Stahl, M.
L., Somers, W. S. & Sullivan, F. X. (2000) Structural and kinetic analysis of Escherichia
coli GDP-mannose 4,6 dehydratase provides insights into the enzyme's catalytic
mechanism and regulation by GDP-fucose, Structure. 8, 123-135.

53. Oths, P. J., Mayer, R. M. & Floss, H. G. (1990) Stereochemistry and mechanism of
the GDP-mannose dehydratase reaction, Carbohydr. Res. 198, 91-100.

54. Maki, M., Jarvinen, N., Rabina, J., Roos, C., Maaheimo, H., Mattila, P. & Renkonen,

R. (2002) Functional expression of Pseudomonas aeruginosa GDP-4-keto-6-deoxy-D-
mannose reductase which synthesizes GDP-rhamnose, F EBS Lett. 269, 593-601.

55. Thoden, J. B., Frey, P. A. & Holden, H. M. (1996) High-resolution X-ray structure of
UDP-galactose 4-epimerase complexed with UDP-phenol, Protein Sci. 5, 2149-2161.

111

56. Thoden, J. B., Frey, P. A. & Holden, H. M. (1996) Molecular structure of the
NADH/UDP-glucose abortive complex of UDP-galactose 4-epimerase from Escherichia
coli: implications for the catalytic mechanism, Biochemistry. 35, 5137-5144.

57. Thoden, J. B., Frey, P. A. & Holden, H. M. (1996) Crystal structures of the oxidized
and reduced forms of UDP-galactose 4-epimerase isolated from Escherichia coli,
Biochemistry. 35, 2557-2566.

58. Thoden, J. B., Gulick, A. M. & Holden, H. M. (1997) Molecular structures of the
8124A, 8124T, and SlZ4V site-directed mutants of UDP-galactose 4-epimerase from
Escherichia coli, Biochemistry. 3 6, 10685-10695.

59. Thoden, J. B., Hegeman, A. D., Wesenberg, G., Chapeau, M. C., Frey, P. A. &
Holden, H. M. (1997) Structural analysis of UDP-sugar binding to UDP-galactose 4-
epimerase from Escherichia coli, Biochemistry. 36, 6294-6304.

60. Thoden, J. B. & Holden, H. M. (1998) Dramatic differences in the binding of UDP-
galactose and UDP-glucose to UDP-galactose 4-epimerase from Escherichia coli,
Biochemistry. 37, 11469-11477.

61. Thoden, J. B., Wohlers, T. M., Fridovich-Keil, J. L. & Holden, H. M. (2001)
Molecular basis for severe epimerase-deficiency galactosemia: X-ray structure of the
human V94M-substituted UDP-galactose 4-epimerase, J. Biol. Chem. 23, 20617-20623.

62. Thoden, J. B., Wohlers, T. M., Fridovich-Keil, J. L. & Holden, H. M. (2001) Human
UDP-galactose 4-epimerase. Accommodation of UDP-n-acetylglucosamine within the
active site, J. Biol. Chem. 276, 15131-15136.

63. Somers, W. S., Stahl, M. L. & Sullivan, F. X. (1998) GDP-fucose synthetase from
Escherichia coli: structure of a unique member of the short-chain

dehydrogenase/reductase family that catalyzes two distinct reactions at the same active
site, Structure. 6, 1601-1612.

64. Currie, H. L., Lightfoot, J. & Lam, J. S. (1995) Prevalence of gca, a gene involved in
synthesis of A-band common antigen polysaccharide in Pseudomonas aeruginosa, Clin.

Diagn. Lab. Immunol. 2, 554-562.

65. Otwinowski, Z. & Minor, W. (1997) Processing of X-ray diffraction data collected in
oscillation mode, Methods Enzymol. 276, 307-326.

66. Navaza, J. ( 1994) AMoRe: and automated package for molecular replacement, Acta
Crystallogr. A. 50, 157-163.

67. Collaborative Computational Project, N. (1994) The CCP4 Suite: Programs for
Protein Crystallography, A cta Crystallogr. D. 50, 760-763.

112

68. Brunger, A. T. (1998) Crystallography and NMR system (CNS): a new software
system for macromolecular structure determination, Acta Crystallogr. D. 54, 905-921.

69. CHAIN. (1995) CHAIN. Crystallographic Modeling Program Version 7. 0, Baylor
College of Medicine, Waco, TX.

70. Laskowski, R. A., MacArthur, M. W., Moss, D. S. & Thornton, J. M. (1993)
PROCHECK: a program to check the stereochemical quality of protein structures, J.
Appl. Crystallogr. 26, 283-291.

71. He, X., Thorson, J. S. & Liu, H.-w. (1996) Probing the coenzyme and substrate
binding events of CDP-D-glucose 4,6-dehydratase: mechanistic implications,
Biochemistry. 35, 4721-4731.

72. Konopka, J. M., Halkides, C. J ., Vanhooke, J. L., Gorenstein, D. G. & Frey, P. A.
(1989) UDP-galactose 4-epimerase. Phosphorus-31 nuclear magnetic resonance analysis
of NAD and NADH bound at the active site, Biochemistry. 28, 9071-9080.

73. Persson, B., Krook, M. & Jornvall, H. (1991) Characteristics of short-chain alcohol
dehyrdogenases and related enzymes, Eur. J. Biochem. 200, 537-543.

74. Deacon, A. M., Ni, Y. 8., Coleman Jr., W. G. & Ealick, S. E. (2000) The crystal
structure of ADP-L-glycero-D-mannoheptose 6-epimerase: catalysis with a twist,
Structure. 8, 453-462.

75. Lightfoot, J. & Lam, J. S. (1991) Molecular cloning of genes involved with
expression of A-band lipopolysaccharide, an antigenically conserved form in
Pseudomonas aeruginosa, J. Bacteriol. 173, 5624-5630.

76. Bergfors, T. (2003) Seeds to crystals, J. Struct. Biol. 142, 66-76.

77. Kabsch, W. J. (1993) Automatic processing of rotation diffraction data from crystals
of initially unknown symmetry and cell constants, .1. Appl. Cryst. 26, 795-800.

78. Jones, D. T. (1999) Protein secondary structure prediction based on position-specific
scoring matrices, J. Mol. Biol. 292, 195-202.

79. McGuffin, L. J., Bryson, K. & Jones, D. T. (2000) The PSIPRED protein structure
prediction server, Bioinformatics. 1 6, 404-405.

80. Kelley, L. A., MacCallum, R. M. & Stemberg, M. J. E. (2000) Enhanced genome
annotation using structural profiles in the program 3D-PSSM, J Mol. Biol. 299, 499-520.

81. Schwede, T., Kopp, J., Guex, ’N. & Peitsch, M. C. (2003) SWISS-MODEL: an
automated protein homology-modeling server, Nucleic Acids Res. 31, 3381-3385.

113

82. Luft, J. R. & DeTitta, G. T. (1995) Chaperone salts, polyethylene glycol and rates of
equilibration in vapor-diffusion crystallization, Acta Crystallogr. D. 51, 780-785.

83. Riboldi-Tunnicliffe, A. & Hidgenfeld, R. (1999) Crystallography with oil - an old
idea revived, J. Appl. Cryst. 32, 1003-1005.

84. Matthews, B. W. (1968) Solvent content of protein crystals, J. Mol. Biol. 33, 491-
197.

85. Kantardjieff, K. A. & Rupp, B. (2003) Matthews coefficient probabiliies: improved
estimates for unit cell contents of proteins, DNA and protein-nucleic acid complex
crystals, Protein Sci. 12, 1865-1871.

86. Harp, J. M., Tim, D. E. & Bunick, G. J. (1998) Macromolecular crystal annealing:

overcoming increased mosaicity associated with cryocrystallography, Acta Crystallogr.
D. 54, 622-628.

87. Yeh, J. L. & Hol, W. G. (1998) A flash-annealing technique to improve diffraction
limits and lower mosaicity in crystals of glycerol kinase, Acta Crystallogr. D. 54, 479-
480.

88. Abergel, C. (2004) Spectacular improvement of X-ray diffraction through fast
desiccation of protein crystals, Acta Crystallogr. D. 60, 1413-1416.

89. Storoni, L. C., McCoy, A. J. & Read, R. J. (2004) Likelihood-enhanced fast rotation
functions, Acta Crystallogr. D. 60, 432-438.

90. Senchenkova, S. N., Shashkov, A. S., Knirel, Y. A., McGovern, J. J. & Moran, A. P.
(1996) The O-specific polysaccharide chain of Campylobacter fetus serotype B

lipopolysaccharide is a D-rhamnan terminated with 3-O-methyl-D-rhamnose (D-
acofriose), Eur. J. Biochem. 239, 434-438.

91. Lightfoot, J. & Lam, J. S. (1993) Chromosomal mapping, expression and synthesis
of lipopolysaccaride in Pseudomonas aeruginosa: a role for guanosine diphosphate
(GDP)-D-mannose, Mol. Microbiol. 8, 771-782.

92. Rocchetta, H. L., Pacan, J. C. & Lam, J. S. (1998) Synthesis of the A-band
polysaccharide sugar D-rhamnose required RMD and Wpr: indentiflcation of multiple
AlgA homologues, Wpr and ORF488, in Pseudomonas aeruginosa, Mal. Microbial.
29, 1419-1434.

93. Brisson, J. R., Sue, S. C., Wu, W. G., McManus, G., Nghia, P. T. & Uhrin, D. (2002)

in NMR spectroscopy of glycoconjugates (Jimenez-Barbero, J. & Peters, T., eds) pp. 59-
93, Wiley-VCH, Weinhem.

114

94. Webb, N. A., Mulichak, A. M., Lam, J. S., Rocchetta, H. L. & Garavito, R. M.
(2004) Crystal structure of a tetrameric GDP-D-mannose 4,6-dehydratase from a
bacterial GDP-D-rhamnose biosynthetic pathway, Protein Sci. 13, 529-539.

95. Kobe, B., Center, R. J., Kemp, B. E. & Poumbourios, P. (1999) Crystal structure of
human T cell leukemia virus type 1 gp21 ectodomain crystallized as a maltose-binding

protein chimera reveals structural evolution of retroviral transmembrane proteins, Proc.
Natl. Acad. Sci. USA. 96, 4319-4324.

96. Liu, Y., Manna, A., Li, R., Martin, W. E., Murphy, R. C., Cheung, A. L. & Zhang, G.
(2001) Crystal structure of the SarR protein from Staphylococcus aureus, PNAS. 98,
6877-6882.

97. Ke, A. & Wolberger, C. (2003) Insights into binding cooperativity of
MATal/MATaZ from the crystal structure of a MATa] homeodomain-maltose binding
protein chimera, Protein Sci. 12, 306-312.

98. Smyth, D. R., Mrozkiewicz, M. K., McGrath, W. J., Listwan, P. & Kobe, B. (2003)
Crystal structures of fusion proteins with large-affinity tags, Protein Sci. 12, 1313-1322.

99. Zablackis, E., York, W. S., Pauly, M., Hantus, S., Reiter, W.-D., Chapple, C. C. S.,
Albersheim, P. & Darvill, A. G. (1996) Substitution of L-fucose by L-galactose in cell
walls of Arabidopsis murl, Science. 272, 1808-1810.

100. Sharff, A. J., Rodseth, L. E., Spurlino, J. C. & Quiocho, F. A. (1992)
Crystallographic evidence of a large ligand-induced hinge-twist motion between the two

domains of the maltodextrin binding protein involved in active transport and chemotaxis,
Biochemistry. 31, 10657-10663.

101. Spurlino, J. C., Lu, G.-Y. & Quiocho, F. A. (1991) The 2.3 A resolution structure of
the maltose- and maltodextrin-binding protein, a primary receptor of bacterial active
transport and chemotaxis, J. Biol. Chem. 266, 5202-5219.

102. Center, R. J., Kobe, B., Wilson, K. A., Teh, T., Howlett, G. J., Kemp, B. E. &
Poumbourios, P. (1998) Crystallization of a trimeric human T cell leukemia virus type I
gp21 ectodomain fragment as a chimera with maltose-binding protein, Protein Sci. 7,
1612-1619.

103. Sun, J., Zhao, J., Schwartz, M. A., Wang, J. Y. J., Wiedmer, T. & Sims, P. J. (2001)
c-Abl tyrosine kinase binds and phosphorylates phospholipid scramblase I, J. Biol. Chem.
276, 28984-28990.

104. Musacchio, A., Saraste, M. & Wihnanns, M. (1994) High-resolution crystal

structures of tyrosine kinase SH3 domains complexed with proline-rich peptides, Struct.
Biol. 1, 546-551.

115

105. Pisabarro, M. T. & Serrano, L. (1996) Rational design of specific high-affinity
peptide ligands for the Abl-SH3 domain, Biochemistry. 35, 10634-10640.

106. Zhou, Q., Zhao, J., Stout, J. G., Luhm, R. A., Wiedmer, T. & Sims, P. J. (1997)
Molecular cloning of human plasma membrane phospholipid scramblase, J. Biol. Chem.

272, 18240-18244.

116

MH'G

l1[ltllllttlllltfllillll11121111171