w
an?» .5...
fining?

I , \
...m.1.% ...
. V . ““9432
V ..
rump? . . . . .
VA .ninԤAvI~t ,
i .i. a. V

..

. . 0mm”.
. Q13.
... . , . A
’7 u n
. 931:: w .,
n.5,
‘ . ’

.. ‘
fiat:

‘ I. A
53 V A?»

u. .
“H?
WNW. V .155. in...
{a . V. V 54.2%.
? .v...t..nm~rn...i: Aim“... ..
fiwflumflfim . 2}.
avg; ..LV?s V

Iv~
yo

, 5...er
x (3
fig”; 13.6.3.
:s .
.‘o...:.v.«nniv.
"WI... . Ana... _
‘LI‘ 1‘ [“‘l{

b.
.11.:\ .7 V

ht. .3. Ju?“ V... t

n 14 um, ‘.I 4

. xfifiwmmmhxmmymm»
.w W34 ‘V

1.
”Vi 4.

It;

£6
1» ‘ .flflhh; . )l)
affihxmrayuauu.

.‘l
x 51 X
2.... V :a 1.

firmku} H

.V

V k?

 

 

 

1P .

. .. Havana?

, if. , at}.
. . . . . . Vv Inhklt
. i»:
u .1 51“.. M33“
. A; . £4.95. , , , R... h . 1, s...
. . V uh Ll. V "5.33:? .331} : 1.3.1 0? who .ul.‘

......r.:.:_5krb p ....PP1 L spy... ..!~$.:1L .‘ V s 3 A, s 'L. . >5. . , , 4 a. v.

 

THESE! 5-”.-

"/
2 (XJI

This is to certify that the

dissertation entitled
STRUCTURAL ANALYSIS OF BRANCHING ENZYME AND

ADP-GLUCOSE PYROPHOSPHORYLASE

presented by

K IM BINDERUP

has been accepted towards fulfillment
of the requirements for

 

 

Ph.D. . Biochemistry
degree 1n
.../ €
\3 deli AW

 

Major professor

Date / 511/” //C9O

MS U is an Affirmative Action/Equal Opportunity Institution 0-12771

 

LIBRARY i

Michigan State
University

 

 

 

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

 

AUG 613 $ 7m?

 

PM f t 8092

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6/01 c-JCIRC/DateDue.p65«p.15

STRUCTURAL ANALYSIS OF BRANCHING ENZYME AND ADP-GLUCOSE
PYROPHOSPHORYLASE

By

Kim Binderup

A DISSERTATION

Submitted to
Michigan State University
In partial fulfillment of the requirements
For the degree of

DOCTOR OF PHILOSOPHY

Department of Biochemistry

2000

ABSTRACT

STRUCTURAL ANALYSIS OF BRANCHIN G ENZYME AND ADP-GLUCOSE
PYROPHOSPHORYLASE

By
Kim Binderup

The biosynthesis of starch and bacterial glycogen involves the enzymes ADP-
glucose pyrophosphorylase, starch or glycogen synthase, and branching enzyme.

Branching enzyme catalyzes the cleavage and subsequent transfer of a-l,4-glucan
to a-l,6 branch points. Hence, the enzyme plays an important role in the biosynthesis of
glycogen and the amylopectin of starch. There is limited structure-function information
available for branching enzyme, but its homology with other enzymes that cleave and/or
transfer chains of a-glucan has provided some insight. The enzyme is believed to contain
a central (on/Bk barrel with extensions at the N- and C-termini. We have investigated the
role of three residues at or near the pr0posed active site and characterized the purified
mutant enzymes. We find the activity of all replacements affected the enzyme activity
and in some cases, the substrate specificity.

We describe the characterization of the first inhibitor of branching enzyme. The
compound BAY e4609 was found to be a slow-binding inhibitor of branching enzyme,
similar to its action in a-amylase and related enzymes. We discuss how this may indicate
a close functional relationship between a-amylase and branching enzyme, and how we

may benefit from BAY e4609 in crystallization experiments.

To investigate the role of the N- and C-terminal of branching enzyme, we
subjected the native 84 kDa enzyme from E. coli to limited proteolysis. We found a
relatively stable 70 kDa fragment, which had an intact C-terminal, but a 111- or 113
residue truncation at the amino terminal. Characterization of this enzyme revealed it had
40-60% residual branching activity. The chain transfer patterns of the native and
truncated enzymes were compared by several independent techniques and were found to
differ significantly. Furthermore, the truncated branching enzyme was crystallized and
diffracted X-rays to about 2.0 A. As a consequence, this work is leading us to the first
three-dimensional structure of a branching enzyme.

The rate of biosynthesis is controlled at the level of ADP-glucose pyrophosphorylase, as
the enzyme is modulated by allosteric effectors. The enzyme from plants and
cyanobacteria is commonly activated by 3-phosphoglycerate and inhibited by
orthophosphate, whereas the enzyme from most bacteria is activated by fructose 1,6-
bisphosphate and inhibited by adenosine-monophosphate. Furthermore, although
tetrameric in structure, the enzyme’s subunit composition differs between sources. ADP-
glucose pyrophosphorylase from bacteria and cyanobacteria is a homotetramer, whereas
the enzyme from plant is a heterotetrameric structure ((1202). However, there is no
structure of ADP-glucose pyrophosphorylase to explain its complex allosteric behavior or
subunit interactions. Here, we describe the crystallization of ADP-glucose
pyrophosphorylase from E. coli and potato tuber, and discuss how this data is leading us

to the first structure of this important enzyme.

Copyfightby
Kim Binderup

2000

ACKNOWLEDGMENTS

This work has been carried out in the laboratory of Dr. Jack Preiss, director of the
Starch Bioengineering Group, Department of Biochemistry, Michigan State University.

I would like to thank Jack Preiss for his valuable support. Jack has given me space
to develop and pursue my own research ideas, yet provided guidance wherever needed.
As a consequence, I have learned more than research techniques- it has enabled me to
become an independent researcher.

Several people have been directly involved in the research described in this
dissertation. I would like to thank René Mikkelsen, who has been a great colleague and
friend. René Mikkelsen performed several of the experiments described in Chapter 2 and
4-6, and assisted in many others. Nathalie Libessart helped me characterize the inhibitor
described in Chapter 3. Furthermore, I would like to thank her for correcting my
dissertation. The crystallization of the truncated E. coli branching enzyme (Chapter 4)
was done in collaboration with Dr. James Geiger and his graduate student Marta Abad.
Marta Abad grew the crystals of branching enzyme and performed the X-ray data
collection and processing. Dr. James Geiger, Dr. Raghuvir Ami, Marta Abad, and
Leandra Watanabe all contributed in the project of crystallizing ADP-glucose
pyrophosphorylase (Chapter 7).

I would also like to thank my committee members Drs. Raghuvir Arni, James
Geiger, Leslie Kuhn, Shelagh Ferguson-Miller, and Honggao Yan for their guidance and

valuable cements.

I wish to thank my parents and brother for their support, and for never doubting
my abilities. Last but not least, I wish to thank my friend, fiancée, and life-companion

Nathalie for her love and unconditional support.

vi

TABLE OF CONTENTS

LIST OF TABLES ................................................................................... xi

LIST OF FIGURES ............................................................................... xiii

LIST OF ABBREVIATIONS .................................................................. xvii

INTRODUCTION ................................................................................. 1

References ............................................................................... 4
CHAPTER 1

LITERATURE REVIEW ......................................................................... 5

Starch and Bacterial Glycogen ...................................................... 6

The metabolic flow of Carbon into Starch ................................. 6

Structure, Organization, and Biosynthesis of Storage Starch ........... 11

Other components of starch .................................................. 19

Bacterial Glycogen ............................................................ 20

Biosynthesis of Starch and Bacterial Glycogen .................................... 22

Starch/ Glycogen Synthase .................................................. 24

Branching Enzyme ............................................................ 26

Assay of Branching Enzyme Activity ............................. 26

Branching Enzyme Isoforms ....................................... 31

Structure-Function Relationships .................................. 32

ADP-Glucose Pyrophosphorylase .......................................... 37

Allosteric Regulation ................................................ 37

ADP-Glucose Pyrophosphorylase is a Tetramer ................. 38

Structure of the Pyrophosphorylase Domain ..................... 39

Active and Regulatory Sites ......................................... 49

References ............................................................................... 56

CHAPTER 2.

SITE-DIRECTED MUTAGENESIS STUDIES OF TYR-300, GLU-459,

AND GLU-527 IN ESCHERICIA COL] BRANCHING ENZYME ....................... 66
Abstract .................................................................................. 67
Introduction .............................................................................. 68
Materials and methods ................................................................. 70

E. coli BE expression vector ................................................... 7O
Constructions of mutants ....................................................... 70
Expression of wild-type and mutant E. coli BE ............................ 70
Purification of E. coli BE wild-type and mutants ........................... 71
Assay of BB activity ............................................................ 71
Analysis of the chain length distribution by high performance
anion exchange chromatography ............................................. 71
Results .................................................................................... 73
Identification of targets for mutagenesis .................................... 73

Vii

Expression and purification of mutant enzymes ........................... 73

Replacement of Glu-459 ...................................................... 75
Replacement of Glu-527 ...................................................... 80
Replacement of Tyr-3 00 ...................................................... 83
Discussion ........................................................................... 88
References ........................................................................... 90

CHAPTER 3.

SLOW-BINDING INHIBITION OF BRANCHING ENZYME BY THE

PSEUDOOLIGOSACCHARIDE BAY-E4609 ............................................... 93

Abstract ................................................................................... 94
Introduction .............................................................................. 95
Materials and methods ................................................................. 97
Expression and purification of E. coli BE ................................. 97
Assay of BE activity ......................................................... 98
Gel filtration .................................................................. 98
Results .................................................................................. 100
Fractionation of BAY-e4609 ............................................... 107
Discussion . ............................................................................ l 10
References .............................................................................. 1 12

CHAPTER 4.

LIMITED PROTEOLYSIS OF ESCHERICIA C 0L1 BRANCHING ENZYME 115
Abstract . ................................................................................. 116
Introduction . ............................................................................ 1 17
Materials and methods . ............................................................... 119

E. coli BE expression vector ................................................. 119
Limited proteolysis of branching enzyme .................................. 119
Purification of proteinase K treated branching enzyme ................... 119
Construction of Nd“ 12 truncated BE ........................................ l 19
Expression of WT and truncated E. coli BE ............................... 120
Expression of selcno-methionine substituted Nd” 12 ..................... 120
Purification of BE WT and Nd-1 12 ........................................... 121
SDS-PAGE and western blotting ............................................ 121
Protein sequencing ............................................................ 122
Assay of BE activity ........................................................... 122
Protein assay ................................................................... 123
Mass spectrometry ............................................................ 123
Protein crystallization ......................................................... 123
X-ray diffraction analysis .................................................... 123
Results ................................................................................... 126
Purification of E. coli BE WT ................................................ 126
Proteolysis of E. coli branching enzyme .................................... 126
Analysis of the 70 kDa fragment ............................................ 129
Activity 0f PK-BE and Nd1.112 ............................................... 132
Crystallization of E. coli branching enzyme ............................... 132

viii

Production of selenium-methionine Nd“ 12 ................................ 135

Discussion ............................................................................... 138
References ............................................................................... 142
CHAPTER 5.
TRUNCATION OF THE AMINO TERMINAL DOMAIN RESULTS IN
ALTERED CHAIN TRANSFER PATTERN OF BRANCHING ENZYME ........... 144
Abstract ................................................................................. 145
Introduction ............................................................................. 146
Materials and methods ................................................................ 148
Plasmids ......................................................................... 148
Expression of wild-type and truncated BE in E. coli
strain AC71 (glgB') ............................................................ 148
Purification of branching enzyme ........................................... 148
Branching enzyme activity ................................................... 148
Glycogen synthase assay . .................................................... 149
Protein assay ................................................................... 149
Analysis of branched chain length .......................................... 150
HPAEC analysis ............................................................... 150
Analysis of chain length by gel filtration ................................... 150
Glucan-iodine complex ........................................................ 150
Average a—glucan chain length determination ............................. 151
Results .................................................................................. 152
Chain transfer by PK-BE, Nd“ 12 and BE-WT ............................. 152
Discussion ............................................................................... 167
References .............................................................................. 1 70
CHAPTER 6.
AN IN VITRO RECONSTITUTION SYSTEM FOR GLYCOGEN
BIOSYNTHESIS ................................................................................ 172
Abstract .................................................................................. 173
Introduction ............................................................................. 1 74
Materials and methods ................................................................ 177
Purification of enzyme material ............................................. 177
Branching enzyme activity ................................................... 177
Glycogen synthase activity ................................................... 177
Phosphorylase activity ........................................................ 177
Protein assay ................................................................... 178
SDS-PAGE analysis .......................................................... 178
Synthesis of a-D-glucan by branching enzyme
and glycogen synthase ........................................................ 178
Synthesis of a-D-glucan by branching enzyme
and phosphorylase a ............................................................ 178
Analysis of branched chain length ........................................... 179
HPAEC analysis ............................................................... 179
Results ................................................................................. 180

ix

Enzymatic activities . ......................................................... 180

Synthesis of a-D-glucan . .................................................... 182
Structures of a-D-glucans .................................................... 184
Discussion .............................................................................. 196
References .............................................................................. l 98

CHAPTER 7.

CRYSTALLIZATION OF ADP-GLUCOSE PYROPHOSPHORYLASE ............. 200
Abstract .................................................................................. 201
Introduction ............................................................................. 202
Materials and methods . ............................................................... 204

Reagents ........................................................................ 204
Plasmids and bacteriophages ................................................ 204
Expression of E. coli ADP-Glc PPase ...................................... 204
Purification of E. coli ADP-Glc PPases .................................... 205
Expression and purification of potato tuber a4 and 0:213; ................ 206
ADP—Glc PPase assay ........................................................ 207

Protein assay . .................................................................. 208
Protein crystallization . ....................................................... 208

X-ray diffraction analysis ................................................... 208
Results ................................................................................... 210
Expression and purification of ADP-Glc PPases .......................... 210
Crystallization of E. coli ADP-Glc PPase WT ............................ 212
Crystallization of potato tuber ADP-Glc PPase ........................... 219
Crystallization of E. coli ADP-Glc PPase mutant G336D ............... 222
Discussion .............................................................................. 224
References ............................................................................... 226

CHAPTER 8.

CONCLUDING REMARKS AND RESEARCH PERSPECTIVES ..................... 228

References ............................................................................... 234

LIST OF TABLES

CHAPTER 1.
1. Properties of potato tuber small subunit and the native enzyme .............. 38

2. Matrix of amino acid sequence identity/ similarity of ADP-glucose

pyrophosphorylases ................................................................... 40
CHAPTER 2.
1. Specific activities of E459 mutants measured using assay A, B and C ....... 76
2. Kinetic parameters of WT E. coli and mutant E459D BEs ...................... 78
3. Specific activities of E527 mutans measured using assay A, B and C ....... 82
4. Specific activities of Y300 mutants measured using assay A, B and C 84
CHAPTER 4.
1. Purification table of BE-WT ....................................................... 125
2. Purification table of Nd“ .2 ........................................................ 131
3. Specific activities of E. coli BE WT, PK-BE and Nd 1-1 I 2 ..................... 133
4. Kinetic parameters of E. coli BE WT, PK-BE and NdH 12 ................... 133
5. Statistics for the branching enzyme X-ray diffraction data collection 134
6. Branching enzyme crystal properties ............................................ 135
7. Mass spectrometry of truncated branching enzyme ........................... 137
CHAPTER 5.
1. Expression of BE WT and truncated branching enzyme in E. coli AC71
(gIgB') and properties of the synthesized a-glucans ............................ 162
CHAPTER 7.
1. Pyrophosphorolysis assay of ADP-Glc PPase .................................. 208

xi

. Summary of single enzyme purifications ........................................ 212
. Mass spectrometry analysis of E. coli ADP-Glc PPase ....................... 212

Data collection and processing statistics for potato tuber small subunit

crystal forms a4#l and a4#2 ....................................................... 221

xii

LIST OF FIGURES
CHAPTER 1.

1. Simplified model of plant starch synthesis in photosynthetic cells
and storage cells ........................................................................ 7

2. The flow of carbon into starch by the hypothetical SS-ADP-Glc

pathway, as proposed by Pozueta-Romoeo et a1 .................................. 10
3. Structure of amylose and amylopectin ............................................ 12
4. Overview of the starch granule organization, adapted from Bali et al. ...... 15

5. Discontinuous synthesis model for amylopectin biosynthesis,

adapted from Ball et a1 .............................................................. 17
6. Model of the organization of amylose and amylopectin in the starch
granule as proposed by Lineback ................................................. 18
7. Biosynthetic pathway of starch and bacterial glycogen ......................... 23
8. The iodine stain assay ................................................................ 27
9. The phosphorylase A stimulation assay .......................................... 28
10. The branching linkage assay ........................................................ 29
11. Schematic representation of some branching enzymes from
plants and bacteria . ................................................................. 34
12. Amino acid alignment of several ADP-glucose pyrophosphorylases ........ 41

13. Amino acid alignment of ADP-glucose pyrophosphorylases and
N-acetylglucoseamine l-phosphate uridylyltransferase ........................ 43

14. Structure of the pyrophosphorylase domain of N-acetylglucoseamine

1-phosphate uridylyltransferase ................................................... 46
15. G-l-P binding site of ADP-Glc PPase ............................................. 50
16. Proposed ATP binding site of ADP-Glc PPase .................................. 51
17. Activator sites of cyanobacterial and plant ADP-Glc PPases .................. 52
18. E. coli ADP-Glc PPase NdHl ...................................................... 53

xiii

19. Proposed Anabaena phosphate binding site ...................................... 54

20. Position of G336 of E. coli ADP-Glc PPase, indicated in bold ................ 55
CHAPTER 2.
1. Partial sequence alignment of several branching enzymes from
various organisms ................................................................... 74
2. HPAEC analysis of the debranched a-glucan formed by the action of
E. coli BE WT on reduced amylose and of E459D ............................. 79
3. pH profile of E. coli BE WT and E459A mutant ................................ 81
4. Heat stability curve of BB WT and Y300F ....................................... 86
5. Relative activity of BE WT and Y300F at elevated temperatures ............. 87
CHAPTER 3.
1. Structures of the pseudo-oligosaccharides acarbose and BAY e4609 ........ 96
2. Inhibition of E. coli BE by BAY e4609 ......................................... 101
3. Lineweaver-Burk plot of E. coli BE in presence of BAY e4609 ............. 103
4. Rate of loss of branching enzyme activity ...................................... 105
5. Preincubation of E. coli BE with BAY e4609 .................................. 106
6. Progress curve of E. coli BE with BAY e4609 ................................. 108
7. Gel filtration of BAY e4609 ...................................................... 109
CHAPTER 4.
1. SDS-PAGE and branching activity of E. coli BE treated with
1:1000 molar ratio of proteinase K to BE at various incubation times . .. 127
2. SDS-PAGE of BE-WT treated with various proteases as described

in materials and methods .......................................................... 128

xiv

3. SDS-PAGE and western blot analysis of wild-type and
truncated branching enzymes ..................................................... 130
4. Crystal of the truncated E. coli branching enzyme ............................. 136
5. Schematic diagram of several branching enzymes ............................. 140
CHAPTER 5.
1. Analysis of transferred chain length by gel filtration after 2 hrs and 4 hrs
of branching using reduced amylose AS-320 as substrate .................... 153
2. Relative distribution of transferred oligosaccharide chain sizes after
2 hrs and 4 hrs branching based on the data from gel filtration .............. 155
3. High Performance Anion Exchange Chromatography of debranched
a-glucans formed by the action of BE-WT and Nd“ 12 on reduced
amylose AS-320 after 2 hrs and 16 hrs of branching .......................... 157
4. Difference in chain distribution between Nd“ 12 and BE-WT resulting
from branching amylose AS-320 for 16 hrs .................................... 159
5. Staining of cell colonies with iodine vapor ..................................... 160
6. High Performance Anion Exchange Chromatography of debranched
a-glucans purified from E. coli B+pET23d, AC71+pGBE,
AC71+pTRC, and AC71+pET23d ............................................... 163
7. Difference in chain distribution between a-glucans purified from
E. coli B, AC71+pGBE and AC71+pTRC ...................................... 166
CHAPTER 6.
1. Reaction scheme of de novo glucan synthesis .................................. 176
2. SDS-PAGE of purified enzymes for glycogen reconstitution system 181
3. Effects of different ratios of enzymatic activities on a-glucan yields 183
4. Effect of reaction time on glucan yields ......................................... 185
5. Analysis of transferred chain lengths after 1 hr, 3 hrs and 16 hrs of

branching, using 10 units of BE activity and 1 unit of GS/PA activity ...... 186

XV

6. Difference plot showing the effect of BE-WT vs. NdH 12
on the glucan structure ............................................................. 189

7. Difference plot showing the effect of reaction time on the
synthesized glucan structures ..................................................... 190

8. Difference plot showing the effect of phosphorylase A

vs. glycogen synthase ............................................................... 191
9. Analysis of transferred chains using BE:GS and BEzPA
ratios of3:1,6:1,and 20:1 ........................................................ 193
10. Difference plot showing the effect of enzyme ratio on
the synthesized glucan structures ................................................. 195
CHAPTER 7.
1. SDS-PAGE of purified enzymes for protein crystallization ................. 21 1
2. Crystallization conditions for E. coli ADP-Glc PPase WT .................. 214
3. X-ray data set of E. coli ADP-Glc PPase crystal form ECWT#l ........... 215
4. X-ray data set of E. coli ADP-Glc PPase crystal form ECWT#2 ........... 217
5. X-ray data set of E. coli ADP-Glc PPase crystal form ECWT#3 ........... 218
6. Crystallization conditions for potato tuber a4 ADP-Glc PPase .............. 220
7. Crystallization conditions for E. coli ADP-Glc PPase G336D ............... 223
CHAPTER 8.
1. Proposed catalytic mechanism for branching enzyme ......................... 230

xvi

LIST OF ABBREVIATIONS

3-PGA: 3-phosphoglycerate

ADP-Glc PPase: ADP-Glucose pyrophosphorylase.

Am: Amylose

AMP: Adenosine monophosphate.

Ap: Amylopectin.

BE: Branching enzyme.

BE-WT: E. coli wild-type branching enzyme.

c.l.: Chain length.

CGTase: Cyclodextrin glucanotransferase.

d.p.: Degree of polymerization.

DBE: Debranching enzyme

DTNB: 5,5’ dithiobis-2-nitrobenoiz acid.

DTT: Dithiothreitol.

EDTA: Ethylenediaminetetracetic acid.

FBP: Fructose 1,6-bisphosphate.

G-l-P: Glucose l-phosphate.

G-6-P: Glucose 6-phosphate.

GBSS: Granule bound starch synthase.

GlmU: N-acetylglucoseamine l-phosphate uridylyltransferase
GS: Glycogen synthase.

HPAEC: High performance anion exchange chromatography.
IPTG: Isopropyl—D-thiogalactoside.

MAD: Multiple anomalous dispersion.

mBE: Maize endosperm branching enzyme.

mBEI: Maize branching enzyme isoform 1.

mBEII: Maize branching enzyme isoform 11.

MW: Molecular weight.

NdHn: E. coli branching enzyme with 112 amino acids deleted at N-terminal.
PA: Phosphorylase A.

PAD: Pulsed amperometric detector.

Pi: Orthophosphate.

PK-BE: Proteinase K cleaved E. coli branching enzyme.
PLP: Pyridoxal phosphate.

PPA: Porcine pancreatic a-amylase.

Se-met: Selenium-methionine.

SS: Starch synthase.

SSS: Soluble starch synthase.

TAA: Taka a-amylase.

TP: Triosephosphates.

WT: Wild-type.

xvii

INTRODUCTION.

Starch is not only the primary intake of calories in a normal human diet but also
the raw material for conversion to many useful commercial products by chemical and
biochemical reactions (1). In the food-industry for example, the starch is converted into
maltodextrins, cyclodextrins and glucose-syrup, and in the nonfood-industry it has found
use in biodegradable packaging material, paper-manufacturing, cosmetics, pharmaceutics,
and much more. Although it is evident that starch is important, its metabolism in plants is
not fully understood as the exact roles of the biosynthetic enzymes remain unknown (2,
3).

The biosynthesis of plant starch and bacterial glycogen involves at least three
enzymes: ADP-glucose pyrophosphorylase (ADP-Glc PPase), starch synthase (SS) or
glycogen synthase (GS), and branching enzyme (BE). ADP-Glc PPase synthesizes ADP-
glucose from glucose-l-phosphate (G-l -P) and ATP, and is believed to be the major
regulatory enzyme in the biosynthetic pathway (2, 4, 5). SS or GS form the or-l,4-linkages
of glucan by transfer of the glycosyl unit of ADP-glucose to the non-reducing terminal of
an a-1,4-glucan primer. Finally, BE forms the a-1,6-linkages of starch and glycogen by
cleavage and subsequent transfer of a-l ,4 glucan to a-1,6 branch points. Starch synthase
and branching enzymes are believed to be major determinants of the starch structure in
plants, although other enzymes are almost certainly involved (6-8).

In chapter 1, a brief summary of our current understanding of the structure and
biosynthesis of plant starch and bacterial glycogen is provided. A detailed description of

protein structure-fimction relationships of branching enzyme and ADP-glucose

pyrophosphorylase is presented. For completion, glycogen synthase and starch synthase
are briefly described, but for detailed information of these enzymes please refer to a
number of recent reviews (4, 5, 7-9).

Our laboratory focuses on elucidating the structure-function relationship of the
starch and glycogen biosynthetic enzymes, using conventional biochemical and molecular
biological approaches. Much of the work described here has aimed at increasing the
understanding of branching enzyme structure-function relationships. Not only do we lack
a structure of branching enzyme, but also the regions responsible for the properties of the
enzyme, including transferred chain length and substrate specificity, are largely unknown.
Chapter 2 describes site-directed mutagenesis studies of the E. coli branching enzyme,
which includes substitution of three residues near or at the proposed active site. In chapter
3, the identification and characterization of the first inhibitor of branching enzyme is
described. This work indicates a close relationship of branching enzyme to a-amylase, as
the compound investigated has the same mechanism of inhibition in both enzymes. The
limited proteolysis of E. coli branching enzyme is presented in chapter 4. We find that
cleavage of 112 amino acids at the amino terminal produces a branching enzyme with a
40-60% reduction in activity. In contrast to the full-length E. coli BE, crystallization
experiments of the truncated enzyme was successful in producing single crystals suitable
for X-ray diffraction analysis. We present a native data set complete to 2.3 A and the
production of selenium-methionine protein necessary for solving the first structure of
branching enzyme. In chapter 5, the truncated branching enzyme is analyzed in detail
with respect to its chain transfer pattern. We find the truncated enzyme transfers longer a-

glucan chains than the native E. coli enzyme, and conclude the amino terminal domain is

involved in determining the chain transfer pattern of branching enzymes. A glycogen
biosynthetic reconstitution system is described in chapter 6. Here, we use purified
branching enzymes, glycogen synthase, and phosphorylase to synthesize a-glucan
structures in vitro. The a-glucans are then analyzed by high performance anion exchange
chromatography (HPAEC) and the chromatograms compared, allowing us to draw
conclusions about multiple parameters affecting on the final a-glucan product.

Despite being intensely studied for decades, there is not currently any three-
dimensional structure of ADP-Glc PPase available. Crystals of the enzyme from E. coli
has been reported (10), but the crystals were subjected to rapid decay in the X-ray beam
and furthermore diffracted poorly (> 5A). Chapter 7 concentrates on the challenge of
obtaining a three-dimensional structure of ADP-Glc PPase. We have expressed
recombinant ADP-Glc PPases from E. coli and potato tuber, and purified the proteins to
near-homogeneity. Crystallization trials have produced single protein crystals of the
enzyme from E. coli and the potato ADP-Glc PPase small unit. We discuss the various
crystal forms obtained and their potential in solving a structure of this enzyme.
Furthermore, selenium-methionine E. coli ADP-Glc PPase has been produced and
multiple anomalous dispertion (MAD) data sets collected of two separate crystal forms.

Finally in Chapter 8, I discuss the perspectives for future studies of Branching
Enzyme and ADP-glucose pyrophosphorylase with emphasis of obtaining the three-
dimensional structures of these enzymes. In addition, I propose a catalytic mechanism for
branching enzyme.

Kim Binderup - East Lansing, December 2000.

10.

References

. Hizukuri S. (1995) in Carbohydrates in Food (Elliasson, A.C, Ed.). Acad. Press, NY.

pp. 347-428.

Preiss, J. and Sivak, MN. (1998) in Comprehensive Natural Products Chemistry
(Pinto, B.M, Ed.) Pergamon Press, Oxford, Vol. 3, pp. 441-495.

Sivak, MN. and Preiss, J. (1998) Advances in Food and Nutrition Research,
Academic Press, San Diego, CA., Vol 41.

Preiss, J. (1996). Regulation of Glycogen Synthesis. in Escherichia coli and
Salmonella typhimurium: Cellular and Molecular Biology, second edition, Amer. Soc.
Microbiol., Washington, DC. Vol. 1, pp. 1015-1024.

Preiss, J. (1984). Annu Rev. Microbiol. 38, pp. 419-458.

Ball, 8., Guan, H.P, James, M., Myers, A., Keeling, P., Mouille, G., Buleon, A.,
Colonna, P., and Preiss, J. (1996). Cell. 86, 349-352.

Smith, A.M., Denyer, K., and Martin, C. (1997). Annu. Rev. Plant Physiol. Plant.
Mol. Biol. 48, 67-87.

Myers, A.M., Morel], M.K., James, M.G., and Ball, 8.6. (2000). Plant Physiol. 122,
989-997.

Preiss, J. (1991) Oxf Surv. Plant. Mol. Cell Biol. 7, 59-114.

Mulichak, A.M., Skizypzak-Jankun, E., Rydel, TJ., Tulinsky, T., Preiss, J. (1988). J.

Biol. Chem. 263, 17237-1723 8.

CHAPTER 1

LITERATURE REVIEW

1. Starch and Bacterial Glycogen.
1.1 The Metabolic Flow of Carbon into Starch.

Starch is a major reserve polysaccharide in green plants and probably the second
most abundant carbohydrate in nature after cellulose (1 ).

Starch is an end-product of carbon fixation in photosynthesis and is found in
almost all green plants (1-4). It is present in virtually every type of plant tissue and organ
including leaves, roots, fruits, pollen, tubers, grains, and stems. A distinction is made
between transient starch, mainly synthesized in leaves, and storage starch synthesized in
the storage organs of the plant (5). Transient starch is synthesized in the chloroplast of
photosynthetically active tissue, where triose-phosphate (TP) from the Calvin cycle via
gluconeogenesis is converted into hexose monophosphate, which then enters the starch
biosynthetic pathway (reviewed in ref. 6; Figure 1). Transient starch is deposited in
granules in the chloroplasts during active photosynthesis throughout the day and degraded
by respiration during the night, thus ensuring a constant availability of photosynthates to
the whole plant (2, 7). In the plant, fixed carbon destined for export is transported from
the chloroplast to the cytosol as triose-phosphate via the TP/Pi translocator (8, 9; Figure
1). The triose-phophate is then converted to sucrose, which is transported from the
photosynthetic tissue and delivered through the phloem to the rest of the plant. The rapid
turnover of transient starch in a day-night cycle makes it an important energy-buffer in
growing plants. In Arabidopsis thaliana, for example, mutants unable to synthesize starch
can grow at continuous light, but show a dramatic decrease in growth-rate when grown in

a day-night cycle (7).

Figure 1. Simplified model of plant starch synthesis in photosynthetic cells
(transient starch, top half of figure) and storage cells (storage starch, lower half of
figure). In photosynthetic tissue, the fixed C02 can be stored as starch in the
chloroplast through the use of pathway 1. For export of fixed carbon from
photosynthetic cells, triosephosphates (TP) are transported from the chloroplast to
the cytosol via the TP/Pi translocator, and converted to sucrose by pathway 11.
The sucrose is then transported from the photosynthetic tissue, often over long
distances in the phloem, where it can be unloaded into storage cells.
Subsequently, sucrose is converted to hexosephosphate (pathway III) and
transported into the amyloplasts by a hexosephosphate translocator. The
amyloplast is able to store large amounts of carbon by conversion of the

hexosephosphate into starch by pathway IV.

Some proteins involved in pathways I-IV are indicated as numbered:
1) Phosphoglucoisomerase 6) UDP-glucose pyrophosphorylase
2) Phosphoglucomutase 7) Sucrose phosphate synthase

3) ADP-glucose pyrophosphorylase 8) Sucrose phosphatase

4) Starch synthases and 9) Sucrose synthase
branching enzymes 10) F ructokinase
5) Triosephosphate/Pi translocator 11) Hexosephosphate translocator

 

 

 

Chloro last
CO2 p Photosynthetic cell

Ru1,5P2
3-PGA

Calvin

RuSP cyde T}: W5 TP - - -;
L F-6-P «J Pi Pi F”
11 II

 

 

 

 

 

 

 

G-6-P 0‘” II
21 12
G-l-P G-l-P UTP
3 ATP I g
r 7 UDP-Glc
ADP-G16 suc-6-P
4: 8i
V
Starch Sucrose
V
Sucrose
Amyloplast V
Sucrose
IV 9
11 UDi—glc fructose
G-1-P/ G-6-P 0-1-1: 111 110
V2 12
G-l-P
G-6-P <———t F—6-P
3rATP I
ADP-Glc
4 . ATP
Sta'rch Storage cell

 

 

 

 

 

 

The major site of starch accumulation is in the storage organs (as storage starch)
including seeds, tubers and storage roots. In contrast to transient starch, the metabolic
turnover of storage starch is very slow and accumulates over long periods, sometimes
several months of the growing season. Once synthesized, it normally remains unchanged
until the next growing season, when it is rapidly mobilized to support new growth.
Storage starch is synthesized in specialized plastids of non-photosynthetic tissue named
amyloplasts, which are committed primarily to starch production. These develop directly
from proplastids and have little internal lamella structure (10). The flow of carbon into
the amyloplast is in the form of the hexose monophosphates glucose-l-phosphate (G-l-P)
or glucose-6-phosphate (G-6-P; 11-13; Figure 1). In the cytosol of the non-photosynthetic
tissue, sucrose is first converted into Glc-l-P or Glc-6-P before transport into the
amyloplast by specific translocators (14, 15). Some studies have suggested that in
addition to hexose phosphates, C3 intermediates is also transported into amyloplasts for
starch synthesis (16, 17), perhaps via a modified TP/Pi transporter. Detailed carbon
labeling studies have since concluded the direct route of hexose phosphate, rather than C3
intermediates, is the far-most likely candidate for entry into the amyloplasts (13, 18).

A proposed alternative pathway of starch synthesis deserves mention here.
Akazawa and coworkers have shown the presence of an ADP/ATP translocater in
chloroplasts and amyloplasts of spinach that also appeared capable of transporting ADP-
glucose (1 9). A mechanism was subsequently proposed where ADP-glucose is
synthesized in the cytosol directly from sucrose by sucrose synthase and then transported
in the amyloplasts where it is utilized directly by starch synthase (Figure 2). This

alternative pathway, called the SS-ADP-Glc pathway (Sucrose Synthase-ADP-glucose

Amyloplast Cytosol

ADP-Glc <———-> ADP-Glc Fructose
Starch
sucrose ADP

 

 

Figure 2. The flow of carbon into starch by the hypothetical
SS-ADP-Glc pathway, as proposed by Pozueta-Romeo et al (19).
The pathway uses a cytosolic sucrose synthase for formation of
ADP-glucose, which is translocated to the amyloplast. The
ADP-glucose is used for glucan synthesis by starch synthase,
and therefore the pathway predicts ADP-glucose
pyrophosphorylase to have a minor role in starch synthesis.

10

pathway), depends on sucrose synthase for formation of ADP-glucose and hence predicts
ADP-Glc PPase to have a minor role in starch synthesis. These predictions do not satisfy
results obtained from mutants deficient in ADP-Glc PPase (7, 20) or antisense studies
targeting the ADP-Glc PPase genes (21). In Arabidopsis, for example, there were
undetectable levels of starch in a mutant plant lacking one of the two subunits of ADP-
Glc PPase (7). Also, the expression of a bacterial ADP-Glc PPase, desensitized to
allosteric regulation, in tobacco, tomato, and potato dramatically increased the starch
content in these plants (22). In light of these results we can conclude that the ADP-
glucose used in starch synthesis is formed via. the ADP-Glc PPase, and that the SS-ADP-

Glc pathway plays little, if any, role in this process.

1.2 Structure, Organization, and Biosynthesis of Storage Starch.

Starch accumulates as a complex granular structure made of a-glucans (or-1,4-
linked and a-1,6-branched) both in leaf cell chloroplasts (transient starch) and in the
amyloplasts of the plant storage tissue cell (storage starch; 5). Starch at first appears to be
a simple compound because it is comprised only of glucose, however, its structure is far
from being fully understood. The storage polysaccharide is usually defined as a mixture
of two distinct fractions: amylose and amylopectin (2, 23, 24). Amylose is mainly found
as linear chains of 840 to 22,000 a-1,4 linked D-glucopyranosyl units with a molecular
weight of 105-106 (Figure 3). Although first defined as a linear polymer, some molecules
found in the amylose fraction are branched (a-l,6-D-glucopyranose; one per 170 to 500
units). Amylopectin is the major polymer fraction of starch, usually making up ~70 % of

the starch, although this number varies highly depending on the plant

11

Figure 3. Structure of amylose and amylopectin. (A) Simple representation of the
mostly linear amylose and the highly branched amylopectin. Amylose is
composed of 0t-1 ,4-linked glucose, whereas amylopectin is a-l ,4-linked and a-
l,6-branched. (B) Detail of the Ot-l,4-gIUCOSlC11C linkage present in both amylose
and amylopectin, and the a-1,6-g1ucosidic linkage of amylopectin. (C)
Diagrammatic representation of an amylopectin molecule, adapted from ref. 73.
a-l ,4 lined glucans are attached by (rt-1,6 linkages to from a highly branched
structure. The outer short glucan chains (A chains) are unbranched and linked to
the multiple branched B chains. Only the single C-chain has a reducing terminal

in each amylopectin molecule.

12

amylose

amylopectin

H O a-1,6 li nlrege

_0» OH H H/“WM‘”

 

a- l ,4 linkege
between two
glucose units

C

__._:1 .___————L1 ______Jl

*1

 

 

 

I

i"

T

 

 

 

 

 

i i»

C-chain

A-chain

B-chains

Edge of granule

species, variety, and storage organ (25). It is a highly branched polymer with 4 to 5 % on-
1,6 glucosidic linkages and a molecular weight of 107 to 109. The amylopectin forms a
highly regular and organized structure with three types of a-glucan chains (Figure 3). The
A chains are the “outer” chains of the amylopectin molecule that are not substituted on
the 6-OH position. The A chains are a-1,6-linked to the inner branches termed B chains,
which may be branched at one or several points. Only a single chain (the C chain) has a
free reducing end per amylopectin molecule. The a-glucan chains of amy10pectin are not
randomly arranged, but form a repeating motif of crystalline and amorph lamella of
approximately 9 nm (Figure 4; 26). Interestingly, this motif is seen throughout the plant
kingdom, and suggests the existence of a highly ordered and well-conserved biosynthetic
pathway (27). Many different proposed structures have attempted to explain the semi-
crystalline nature of amylopectin (1, 28-31), but only the most satisfactory in fitting
experimental data, and widely accepted model, is discussed here. The crystalline lamella
consists of linear and parallel a-glucans that intertwine to form double helicies, as
determined by X-ray diffraction analysis (32, 33). These parallel a-glucans are the A- and
B chains of amylopectin. Starch from various plant sources is known to have ranging
degrees of crystallinity. This is probably a function of their different average branch
lengths (i.e. size of B and A chains), and indeed the length of the crystalline lamella has
been reported to range 4-6 nm, which is equivalent of 12-18 glucose residues arranged in
a helix (34, 35).

Most of the a-1-6-branch points are believed as clustered in the amorphous
lamella approximately 3 nm in length. How such an asymmetric distribution of the a-1,6-

linkages is achieved in the plant cell has been object of much speculation. Only

14

 

 

 

 

 

 

 

 

 

 

 

 

 

Figure 4. Overview of the starch granule organization, adapted
from Ball at at. (36).

(A) The starch granule (1.5 pm thick) with the amorphous (dark
regions) and semi-crystalline growth rings (clear regions). Each
semi-crystalline growth ring consist of a repeating motif of amor-
phous and crystalline lamella.

(B) An enlargement of a segment of the semi-crystalline growth
rings, where each amorphous and crystalline lamella is
represented as a shaded or clear region, respectively.

(C) The relation between the lamella and the amylopectin
structure. The dotted lines define the limits between the
crystalline (1) and amorphous (2) lamella.

(D) The panel shows the helical organization of the glucan
chains in the crystalline lamella (1) and the clustering of the

a-1,6-branch points in the amorphous lamella.

15

Jnm

6am

recently have researchers gained insight into this mechanism. Work by Ball and
coworkers has demonstrated a discontinuous synthesis model for amylopectin
biosynthesis (3 6). The model is based on the sugary (suI) mutants of maize, long known
to be defective in some aspect of starch biosynthesis (37). These mutants do not produce
starch, but rather a glucan structure reminiscent of glycogen and was hence named
phytoglycogen (38, 39). Later, the enzyme responsible for the sugary phenotype was
demonstrated to be a defect in debranching enzyme (40-43). The model by Ball et al
involves the action of debranching enzymes in amylopectin biosynthesis (in addition to
ADP-Glc PPase, SS and BE) to perform glucan trimming of the growing carbohydrate
polymer. The glucan trimming is required to generate order in the amorphous lamella for
subsequent synthesis of the crystalline lattice (Figure 5). This action at the surface of the
growth rings presumably prevents the formation of phytoglycogen, and allows for a
continuous elongation of the highly ordered amylopectin molecule.

Although there has been much progress in understanding the structures of
amylopectin and amylose, their assembly into the final starch structure is still largely
unknown. A model by Lineback (31) proposes that several amylopectin molecules, each
consisting of 20-40 repeating motifs of crystalline and amorph lamella, align much like
the spokes of a wheel (Figure 6). The reducing ends of the C-chains are believed to face
the hilum (center) of the starch granule, whereas the branches extend in a radial
orientation toward the edge of the granule. In Lineback’s model, the amylose is packed
alongside the amylopectin, but this fails to explain how amylose is protected from
branching enzyme and is not consistent with current views of its synthesis (44, 45). The

amylose is now believed to pack in the amorphous regions, where it is protected from the

16

to

Crystallization

 

Figure 5. Discontinuous synthesis model for amylopectin
biosynthesis, adapted from Ball at al. (36). (A) The chains of
amylopectin in the water-soluble, amorphous lamella are elongated
by starch synthases (SS). (B) When the chains are sufficiently long
to serve as substrate, branching enzymes (BE) perform extensive
random branching of the polymer. (C) Debranching enzymes (DBE)
work simultaneously to trim the loosely branched glucans. (D) The
trimming by DBE will enable crystallization of elongated chains and
the remaining branches will form the next amorphous lamella.

 

Figure 6. Model of the organization of arnylose (Am) and amylopectin (Ap)
in the starch granule as proposed by Lineback (31). The reducing termini
of the C-chains face the hilum (center) of the starch granule, and the
amylopectin polymers align radially towards the edge of the starch

granule. In this particular model, the arnylose is proposed to pack parallel

to and between the amylopectin molecules.

starch branching enzymes, that have been found located in the stroma near the edge of the
starch granule by electron microscopy (46). Furthermore, the packing of amylose in the
amorphous lamella is consistent with its proposed biosynthesis. Amylose is synthesized
from amylopectin acting as a primer by the granule bound starch synthase I (GBSSI), as
has been demonstrated in numerous waxy (wx) mutant plants and the sta2 mutant of
Chlamydomonas, all deficient in normal amylose synthesis (47-50). The waxy (GBSSI’)
mutants all accumulate normal amounts of starch granules containing amylopectin with
wild-type crystalline organization (34). These results imply that amylose is not required
for the biogenesis of normal granules. Furthermore, GBSSI is located throughout the
starch granule where it is believed to be immobilized on the carbohydrate matrix, thus
being able to deposit the synthesized amylose directly into the amorphous lamella (see
Figure 9 in ref. 44 for detailed model). The above model for the biogenesis of storage
starch (here greatly simplified) is currently the most favored by researchers, but is by no
means a complete description, as many questions remain unanswered. The exact roles of
the many different enzyme isofonns in starch biosynthesis remains to be determined, and
it is very likely that additional enzymes and isoforms, yet undiscovered, are involved in

the making of starch.

1.3. Other components of starch.

In addition to carbohydrate, the starch granule contains small amounts (0.2-0.5 %)
of protein (including the enzymes of starch biosynthesis) and 0.5-1.0 % of lipids. Also,
some amylopectins, notably amylopectin from potato tubers, are known to be

phosphorylated, with up to 1 phosphate per 300 glucose residues (51).

19

1.4 Bacterial Glycogen.

In this report, we will only consider glycogen from bacteria, as its metabolism in
mammals differs completely. However, the biosynthetic pathway of bacterial glycogen
shares many features of plant starch biosynthesis, as described below. Many bacteria will
store glycogen as a carbohydrate reserve when cell growth is slow and carbon is in
abundance. The rate of glycogen accumulation is often highest in the stationary phase,
when growth ceases due to depletion of a nutrient such as nitrogen, sulfur, or phosphate
(see review in ref. 52). E. coli B in particular, has been shown to accumulate large
amounts of glycogen in nitrogen-limited media with glucose as the carbon source (53).
However, the inability of bacteria to synthesize glycogen is not lethal, and bacteria
deficient in the glycogen biosynthetic enzymes grow as well as their normal parent
strains. The glycogen is generally considered an energy-storage polymer, which becomes
rapidly degraded upon carbon starvation. One example is glycogen-containing E. coli
cells that have prolonged viability in media with no exogenous carbon source compared
to their glycogen-less counterparts (54). Also, for some spore-forming bacteria, the
glycogen may have a role as endogenous source of carbon and energy for spore formation
as seen in Bacillus cereus (52, 55).

Glycogen is similar to amylopectin by being (rt-1,4 linked and or-l,6 branched.
Glycogen is more branched than amylopectin with approximately 10 % a-1,6 glucosidic
linkages and a degree of polymerization (d.p.) of about 104 units. Perhaps the most
significant difference between the structure of glycogen and amylopectin is that
amylopectin is a highly organized semi-crystalline molecule. In contrast, glycogen has a

random arrangement of regular branching lengths, making the smaller glycogen granule

20

highly water-soluble. It is possible to calculate the maximum size attainable by the
glycogen granule before steric hindrance impairs further growth. Those calculations are in
good agreement with the size (~25 nm in diameter) measured for these particles (56).
Thus, the highly water-soluble glycogen and the semi-crystalline starch differ mainly by
the decreased branching observed in the plant polysaccharide and its asymmetric
distribution of the ct-l,6-linkages.

It is widely accepted that the glycosyl donor in biosynthesis of bacterial glycogen
is ADP-glucose and that the pathway consists of three enzymes: ADP-Glucose
pyrophosphorylase, glycogen synthase and branching enzyme (52, 57, 58). These
enzymes are believed to be necessary and sufficient for glycogen biosynthesis, although
the presence of a bacterial glycogenin-like protein was hypothesized in the 19708 (59),
but later proofed incorrect (60). However, some bacteria can form a-glucans from sucrose
or maltose by amylosucrase and amylomaltase, respectively (52, 61, 62). These enzymes
are not believed to be significant in a-glucan formation under normal conditions, as they
are scarce in bacteria and their enzymatic activities repressed in presence of glucose (52).
The ADP-Glc PPase, glycogen synthase and branching enzyme is each encoded by a
single gene in bacteria (63, 64). This contrasts the equivalent enzymes in plants, where
multiple isoforms are always found (2). Probably this difference can be attributed to the
higher degree of control and regulation needed in plant starch biosynthesis and the
increased complexity of the storage polymer. Furthermore, the allosteric regulation of the
key enzyme ADP-Glucose PPase differs between bacteria and plants, as will be discussed

in detail below.

21

Several plant enzymes involved in starch biosynthesis have been expressed in E.
coli, including ADP-Glc PPase (65, 66), Starch Synthase (67), and Branching enzyme
(68, 69). Of particular interest was the complementation by recombinant maize starch BE
isoforms of the E. coli mutant AC71 (glgb’) lacking the endogenous branching enzyme
(70). These experiments failed to reproduce the typical asymmetric pattern of
amylopectin in E. coli, but resulted instead in glycogen-like products. These experiments
clearly demonstrate the biosynthesis of starch is more complex and probably involves

more enzymes, than the glycogen biosynthesis of bacteria.

2. Biosynthesis of Starch and Bacterial Glycogen

Glycogen synthesis in bacteria and starch synthesis in plants follow similar
pathways and have at least three enzymes in common: ADP-Glc pyrophosphorylase,
glycogen synthase or starch synthase, and branching enzyme (Figure 7). First, ADP-
glucose pyrophosphorylase activates a glucose molecule by forming the sugar-nucleotide
ADP-glucose. Glycogen/starch synthase elongates the growing (rt-glucan chain by forming
an a-(1,4) glucosidic linkage between the activated sugar and a primer. Finally, branching
enzyme catalyses the formation of branch points by cleavage and transfer of the a-1,4
glucan chain to a-1,6 branch points.

As will be discussed in detail, ADP-glucose pyrophosphorylase is an allosteric
regulated enzyme and is considered a major regulatory step in the rate of glycogen and
starch synthesis in bacteria and plants, respectively (2). It is thought that the specificities

of glycogen/ starch synthases and branching enzymes are mainly responsible for the

22

A: ADP-glucose pyrophosphorylase

CHZOH

o CHZOH
o \ 8 ATP PPi O
H \
OH - E Z “0 ii 1?
H0 012:0 HO OH . PO' Adenosine
A 0‘ 0'
Glucose- 1 -phosphate ADP-glucose
+
CHZOH CHZOH
O O
HO OH
HO 0“ 0 HO 0“
B: Starch/glycogen synthase B ADP
CHZOH CHZOH CHZOH
O O O
C' Branchin enzyme i C
' g CHZOH

HO
HO OH
CHZOH o

Y'l;oé\ $.-
HO OH
HO OH O HO 0“

Figure 7. Biosynthetic pathway of starch and bacerial glycogen. ADP-glucose
pyrophophorylase catalyzes the formation of ADP-glucose (A), which is
added on to a primer by starch/ glycogen synthase (B). Branching enzyme

then forms the branch points of amylopectin and glycogen.

23

different structures of glycogen and starch. A brief summary of our current level of

understanding of these enzymes is provided next.

3. Starch! Glycogen Synthase.

In bacterial systems (e.g. E. coli) only one glycogen synthase isoform has been
found (71), whereas in plants there are at least four different starch synthase genes and
therefore four different enzymes (67, 72). Some starch synthases are bound to the starch
granule (granule bound starch synthases, GBSS), while others are found in the soluble
portion of the cell extracts (soluble starch synthases, SSS). It is not clear whether some of
the soluble and granule bound starch synthases are indeed the same proteins. The roles of
each of these starch synthases in the synthesis of the starch granule components, amylose
and amylopectin, is not well understood. However, plants defective in one or more starch
synthase genes suggest the various soluble and granule bound starch synthases have
different roles in amylose and amylopectin biosynthesis. Probably the most studied, and
best understood starch synthase is GBSSI which now is well established as being
involved in amylose biosynthesis (see section 1.2). The role of the various soluble starch
synthases remains confusing, especially as the different SSS isoforms appear to dominate
in various plant systems (73). For example, in vitro measurements of SSS activity from
maize indicate $881 as the dominant activity, whereas similar experiments in pea and
potato tuber found 88811 and SSSIII, respectively, as having the greatest enzyme activity
(74). It is clear that more work is needed, but as the native (rt-glucan primers are yet
uncharacterized and the specificities of SSS isoforms partly overlap, this remains a

challenging research subject.

24

Until recently, the only 1,4 a-D-glucan synthase reported to be cloned and over-
expressed was the E. coli and S. typhimurium enzymes (71, 75). The E. coli enzyme can
be purified to homogeneity with good yield, which has facilitated biochemical analysis
including studies using chemical modification, kinetic measurements, and site-directed
mutagenesis (76-78). Crystallization of the recombinant E. coli glycogen synthase (GS)
has also been attempted, but with limited success (Binderup and Preiss, unpublished
results). Glycogen synthase is associated with the glycogen in E. coli and released by O.-
amylase treatment during its purification (76). The resulting maltooligosaccharides in the
purified enzyme preparation are speculated to inhibit formation of protein crystals due to
microheterogeneity. The expression of GS in the glycogen-less E. coli mutant strain
G6MD3 (deficient in ADP-Glc PPase) should therefore be performed.

The bacterial enzyme shows several regions of similarity with plant starch
synthases. Chemical modification studies and site-directed mutagenesis of the E. coli
enzyme has demonstrated the ADP-Glc binding site to be located at the amino terminal of
the protein, specifically at a conserved Lys-X-Gly-Gly motif (76-78). F urthermore,
Holmes and Preiss (79) have described chemical modification of GS with the sulflrydryl
specific reagents iodoacetate and 5,5’ dithiobis-Z-nitrobenoiz acid (DTNB). ADP or
ADP-glucose protected the enzyme against iodoacetate, whereas glycogen protected the
enzyme against DTNB. The results indicated two distinct Cys residues may be located at
or near the binding site for ADP-glucose and glycogen. Besides these observations

structure-function information of this enzyme remains very limited.

25

4. Branching Enzyme.
4.1 Assay of Branching Enzyme Activity.

There are several assays available for measuring branching activity. The iodine
stain assay (Figure 8) is based on the decrease in absorbance of a glucan-iodine complex
with increased branching (80). Although the assay is quite insensitive it allows for testing
of different substrate (i.e. maltodextrins, amyloses, and amylopectins). The phosphorylase
stimulation assay (Figure 9) is based on incorporation of MC labeled glucose into a-
glucan as branching enzyme increases the number of non-reducing ends (81). Although
an indirect measure of branching activity, the phosphorylase stimulation assay is very
sensitive and the assay that best imitates the function of branching enzyme in vivo. The
branching linkage assay is the most quantitative assay (82), as its determines the number
of branching events on an arnylose substrate, making the assay suitable for kinetic studies
(Figure 10). After branching of the substrate, the chains transferred by BB are cleaved at
the a-l ,6-branch points by isoamylase and the reducing ends detected by a modified Park-
Johnson method (83). Alternatively, the cleaved chains can be analyzed by gel-filtration
or high performance anion exchange chromatography (HPAEC) to determined by
transferred chain pattern by branching enzyme (Figure 10). For a complete
characterization of branching enzyme it is best to employ all three assays. However, it is
important to work with enzyme preparations free from a-glucan degradative enzymes, as

these interfere with all three branching enzyme assays.

26

\_/ _____________________ c /
Helical Iodine / a-D-glucan complex
. 5 <00660> for
BranCh'ng amylose/ iodine complex
Amyiose / enzyme iodine
Amylopectin
5 <OD530> for

amylopectin/ Iodine complex

Fig. 8. The iodine stain assay (assay C). The substrate, amylase or
amylopectin, is branched by branching enzyme. This gives rise to a
lower absorbance of the helical iodine/or-D-glucan complex. The
wavelengths used are 530 nm and 660 nm for amylopectin and amylose,

respectively.

27

Phophorylase A EG-G@

Branching Enzyme a_ (1’4) Glucan

*
G-1-P / 6443 > intermediate formed
by phosphorylase A.

as 4,

. 0' 9
0’ . 0’
——> [EGG-666666439
\o ‘0

% ©

74C incorporated Glycogen

 

Fig. 9. The phosphorylase A stimulation assay (assay A). Phosphorylase
A elongates a primer from the non-reducing end . Branching enzyme
produces for each branch point an additional non-reducing terminal,

which then phosphorylase A elongates further. The synthesized polymer
has incorporated l4C-labeled glucose (*G). The reducing end is

indicated©

28

Figure 10. The branching linkage assay (assay B) and high performance anion
exchange chromatography (HPAEC). The (rt-(1,6) linkages are introduced into the
amylose substrate by branching enzyme. After heat inactivation of the reaction
mixture, the polymer is debranched by isoamylase, producing one reducing
terminal, illustrated as an encircled G, per chain. In assay B, the reducing termini
are quantified by a modified Park-Johnson method. In HPAEC and gel filtration

analysis, the distribution of the transferred chains is monitored.

29

@G-[G]n-G-G© Branching @c 53

Enzyme 6‘ 4?
Reduced amylose . 09: 9
-G-G-G-G-G-G-G
e: . a
“g <9
a-( 1,6) branched or-(1,4) glucan
lsoam lase
_ @G 4? — Y
c 9
e e
0’ e
e \(o'
@G-G-G-G-G-G-G-G©

16‘“? fqo
“ ©

 

 

L— or-( 1,6) cleavage by isoamylase —-4-

HPAEC G
@c@ @6—
a-(1,4) glucans
Analysis of transferred
chain length distribution fiffflnfm
method
Number of reducing

ends ( ) determined
spectrophotometrically

30

4.2 Branching Enzyme Isoforms
Branching enzymes fi‘om a variety of sources have been identified and described,
including E. coli (84, 85), maize (86, 87), and rice endosperm (88, 89). In contrast to
bacteria, multiple isoforms of the branching enzyme, encoded by distinct genes, have
been reported for many plants, including maize, rice, and Arabidopsis (81, 85, 89, 90).
These isoforms appear to fall into two classes of branching enzymes that can
distinguished by their biochemical properties (91 ). Class I branching enzymes transfer
long a-glucan chains and have a relative high activity with amylase as substrate.
Members of class I branching enzymes include maize isoform 1, pea isoform B, rice
isoform 1, and potato isoform H. Class H branching enzymes transfer short a-glucan
chains and preferentially use amylopectin as substrate. Members of class II branching
enzymes include maize isoform II, pea isoform A, rice isoform HI, and potato isoform 1.
The division of branching enzymes into classes according to their biochemical
properties, and their proposed distinct roles in starch synthesis, evolved from studies of
the maize branching enzyme isoforms. The investigation of the mode of chain transfer
and substrate specificity in the maize endosperm branching enzyme I (mBEI) and
branching enzyme 11 (mBEII: 91) led to the hypothesis that these enzymes have different
roles in the synthesis of amylopectin in viva. The mBEI can transfer long branches (d.p.
40 to 100) whereas mBEII transfers shorter chains (d.p. 6 to 14). Furthermore, the
preferred substrate for mBEI is amylose, whereas mBEII prefers amylopectin. The
differences between mBEI and mBEII isoforms suggest that mBEI first branches the
growing a-glucan polymer transferring long chains and mBEII, in turn, uses this product

for further branching to amylopectin. There are also mutant plants deficient in class II

31

branching enzymes that support this theory. Both the rugosus (r) mutant of pea, originally
used in genetic crossing experiments by Mendel (92-94), and the amylose extender (ae)
mutants of maize and rice (95, 96) are deficient in class H BE activity. Total starch
content in these plants are reduced by 20 % in maize and 50 % in pea, and the
amylopectin content is lowered from ~70 % to ~30 %. The amylopectin in these mutants
has longer branches than the wild-type plants, and the degree of polymerization is
reduced. The effect of the ae and r mutations on starch synthesis furthermore suggest that
class I branching enzymes cannot fully compensate for loss of class II branching enzyme
activity.

In addition to the distinct biochemical properties of class I and class H branching
enzymes there are differences in the primary structures of the proteins. Enzymes from the
two families are structurally similar and related to glycogen branching enzymes of
bacteria. However, class II branching enzymes contain an extension of ~50 residues at the
amino terminal compared to class I enzymes. Very recent work has demonstrated that
deletion of 39 residues of this extension in mBEH lowers the specific activity to ~70 % of
wild-type levels, whereas the substrate specificity and chain transfer pattern is unaltered
(97). These results may indicate this extension is not responsible for the different

properties of the two branching enzyme families.

4.3 Structure-Function Relationships.
There is no three-dimensional structure of branching enzyme available, although
the crystallization of a truncated E. coli branching enzyme has been reported (98). This

report was published simultaneously with our work on the limited proteolysis of E. coli

32

branching enzyme (99; Chapter 4). However the study by Hilden et al (98) was limited in
its biochemical analysis of the truncated branching enzyme, and the poor diffraction
quality of the crystal form (4.1 A at the EPS synchrotron source) makes it an unsuitable
candidate for a three-dimensional structure.

Although the structure of a branching enzyme has not been determined, its
homology to amylolytic enzymes, defined as enzymes involved in cleavage and/ or
transfer of a-glucans, has provided important information about its structure- function
relationships. Several structures of amylolytic enzymes including a-amylase, cyclodextrin
glucanotransferase, isoamylase, and pullulanase have been solved, and all consist of a
central a/B-barrel structure (100-103). Alignments and secondary structure prediction of
branching enzyme show this protein also conforms to this structure. This allows for
prediction of which residues are involved in catalysis, as the catalytic residues of a/B-
barrels are typically found in the loops interconnecting each central B-strand with the
following a-helix (104).

Homology between branching enzymes and other amylolytic enzymes was first
described by Romeo et al. (64). Later it was established that branching enzymes belonged
to the family of amylolytic enzymes as they contain four highly conserved regions that are
also found in a-amylases, pullulanase, isoamylase and cyclodextrin glucanotransferases
(CGTase; 105, 106). Figure 11 shows a schematic representation of some branching
enzymes and an alignment of the four conserved regions found in all amylolytic enzymes.
The four conserved regions contain seven amino acids invariantly found in all amylolytic

enzymes, and these residues have been targeted for site-directed mutagenesis studies.

33

N-ternrinal (ix/13): barrel C-terminal

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

E. coli BE 243 83.
l I
I l
mBE I
188 aa " 191 a.a.
mBEII 200 a.a. 127 aa.
Bacillus Sterathermaphilus 149 a.a. 7 A if." .‘ 137 a.a.
(18
W '\
B3 —)H3100p Bit—+114 loop 65—)H5 loop B7 —)H‘7 loop
Region 1 Region 2 Region 3 Region 4
TAA 120TTKDWANH -'77 aa. 200DGLRID'I'V -20 aa. ZZBIGKVLD -59 aa. 292mm)
PPA 99TYVDAVINH -94 aa. 192AGFRLDAS -31 an. 231FQSVID -58 aa. 295mm
CGTase 167VIIDFAPNH -83 an. 253DGIRVDAV -23 aa. 284mm -63 aa. 253mm
Maize BEI 280VLMDVVHSH -63 aa. 346DGFRFDGV -50 aa. 404VAIDVS -60 an. 4'70YAESED
Maize BEll 318VLMDVVHSH -61 aa. 382DGFRFDGV -50 aa. 4BSIGRDVS -59 aa. 504YAESHD
Ecolt' BE 331VILDWVPGH -59 aa. 399DALRVDAV -49 an. 456mg? -58 aa. 520LPLSHD
Scoe/icolor BE" 354VIMDWVPAH -60 aa. 4IGDGLRVDAV -49 aa. 4721MEST -60 aa. 537LPISHD

Figure 11. Shematic representation of some branching enzymes fi'om plants and
bacteria. The central (G/B)3 barrel is shaded gray. The N- and C-terminal
extensions relative to the barrel are indicated for each protein. The lower part of
the figure shows the four highly conserved regions shared between enzymes of
the amylolytic family, including four branching enzymes. The location of each
region in the B—mt loops of the (or/Bk barrel is indicated. The 7 residues of region
1-4 that are completely conserved in the a—amylase family, are indicated in bold
letters. Key: TAA, Taka or-amylase; PPA, porcine pancreatic (rt-amylase; CGTase,

cyclodextrin glucano-transferase.

34

Site-directed mutagenesis of the two conserved Asp residues in region 2 and 4 and
the Glu residue of region 3 indicate these amino acids are important for branching activity
(107, 108). These three residues are proposed to be catalytic residues in the amylolytic
family, based on the crystal structures and biochemical evidence from (rt-amylase,
CGTase, and pullulanase (100-102, 109). Although the exact role of these residues in
branching enzyme is unknown, the site-directed mutagenesis studies indicate that part of
the mechanism in all these enzymes is shared.

Chemical modification of mBEII with phenylglyoxal, a reagent specific for arginine,
caused inactivation of the enzyme (110). The phenylglyoxal inactivation could be
prevented by the presence of amylase during the experiment, suggesting a potential role
in substrate binding. Subsequently, the Arg residue of region 2 (R384 in mBEII) was
targeted for site-directed mutagenesis studies (111). All substitutions produced inactive
enzymes, except the conservative substitution R384K, which retained 5 % of wild-type
activity. However, kinetics of the purified R384K mutant indicated only a minor increase
in Km of the amylose substrate (111). Thus, whereas the Arg of region 2 appears to play
an important role in branching enzyme catalysis, it may not be directly involved in
substrate binding.

Chemical modification of mBEII with the histidine specific reagent
diethylpyrocarbonate caused rapid inactivation of mBEII, but the inactivation could be
prevented by the presence of amylase (112). Treatment of the diethylpyrocarbonate-
inactivated enzyme with hydroxylamine rapidly restored the enzyme activity. The
conserved His residues of regions 1 and 4 (H320 and H508 in mBEH, respectively) were

individually substituted by Ala via site-directed mutagenesis (112). Kinetic data of the

35

purified mutants H320A and H508A showed little effect on Vmax, but a 10-fold increase
in Km. Thus, the results suggested an important role of the conserved histidine residues in
substrate binding.

The proposed functions for the conserved residues described above correlate well
with site-directed mutagenesis of (rt-amylase and cyclodextrin glucanotransferase as well
as their three-dimensional structures (100, 101, 109). Apart from these regions there is
limited sequence homology between various branching enzymes. Typically the similarity
is 25-40 % identity across species and 40-70 % between isozymes from the same source.

All branching enzymes contain extension as the N- and C-terrnini relative to the
central a/B-barrel domain (113, 114). In contrast to the central a/B-barrel, these extension
are not well conserved between different branching enzymes. For example, between the
maize isoforms mBEI and mBEII, the sequence identity is 58 % overall but 67 % in the
central a/B-barrel doimain. Kuriki et al investigated the roles of these extensions by
construction of chimeric enzymes out of mBEI and mBEII (115). A chimeric enzyme
with the N-terminal and central or/B-barrel of mBE H and the C-terrninal domain of mBEI
produced an active enzyme with the chain transfer pattern of mBEII and the substrate
specificity of mBEI (115). These results suggest that the C-terminal domain is involved in
substrate specificity and the N-terrninal and/ or central CUB-barrel region is involved in
chain transfer action of branching enzymes. Although the study was limited by the large
number of inactive chimeric constructs, these results has provided much insight into

which domains are responsible for the properties of branching enzymes.

36

5. ADP-Glucose Pyrophosphorylase.
5.1 Allosteric Regulation.

The reaction of ADP-Glc PPase was first described in 1962 by Espada in soya
bean (116) and was subsequently found in many plant tissues and bacterial extracts
(reviewed in ref. 2, 6, and 52). The enzyme is highly regulated but its specificity toward
the effectors depends upon the source of the enzyme. ADP-Glc PPase from most bacteria
requires the glycolytic intermediate fi'uctose 1,6-bisphosphate (F BP) for optimal activity
and is inhibited by adenosine monophosphate (AMP; 58). The presence of F BP may be
considered as an indicator of carbon in excess of energy need, acting as a signal that
under these circumstances, carbon may be used for an energy storage compound.
Conversely, AMP can be considered to indicate a low energy-charge of the cell, where
synthesis of energy storage compounds should be repressed. The enzyme from
cyanobacteria and almost all higher plant sources is activated by 3-phosphoglyceric acid
(3-PGA) and is inhibited by orthophosphate (Pi; 117, 118). A high ratio of 3-PGA/ Pi is
observed during photosynthetic carbon fixation, which produces 3-PGA as the first stable
product in many plants, meanwhile the Pi concentration is low due to
photophosphorylation. Thus, it appears that in many cases there is a correlation between
the nature of the activator seen for the organism’s ADP-Glc PPase and its major
assimilation pathway.

The activator can have multi-fold effect on enzyme activity, either increasing Vmax
and/ or lowering the Klm of the enzyme for the substrates. In addition, the activator can

reverse the effect of the allosteric inhibitor. The concentration for half-maximal activation

(A05) and inhibition (105) for ADP-Glc PPase is generally of the order of 11M.

37

5.2 ADP-Glucose Pyrophosphorylase is a Tetramer.

Studies on the subunit structure of ADP-Glc PPases have been carried out to some
extent (58, 66). For all bacterial and cyanobacterial enzymes studied so far, the native
enzyme is a homotetrarner (0a) with a subunit mass of approximately 50 kDa. In contrast,
the plant enzyme consists of two related but distinct subunits having masses in the 50-60
kDa range forming a (1202 structure (119). The smaller subunit (50-55 kDa) is highly
conserved with an amino acid sequence identity of 85-95% across plant species. In
contrast, different larger subunits (50-60 kDa) are less conserved (SO-60%; 120). When
the small and large subunit from the same plant are compared, the identity is typically in
the range of 50-60%.

Recent experiments indicate that the plant small subunit is the catalytic subunit, as
it has enzymatic activity when expressed alone. The large subunit functions by
modulating the regulatory properties of the small subunit (Table 1; 66, 121). The large
subunit does not have any detectable catalytic activity when expressed in absence of the

small subunit.

Table 1: Properties of potato tuber small subunit and the native enzyme.

 

 

Potato small (cu) Potato small + large (azfiz)
Specific Activity 50 U/ mg 100 U/ mg
A05 for 3-PGA 2.4 mM 160 uM
[0,5 for Pi (in presence of 3 mM 3-PGA) 80 “M 640 ”M

 

From Ballicora et al. (66). Enzymes were purified to homogeneity. Assay was in ADP-

Glc synthesis direction.

38

The potato tuber small subunit expressed in absence of the large subunit forms a stable or.
structure with high enzymatic activity. The ac, enzyme is less sensitive to the activator
3-PGA and more sensitive to the inhibitor, Pi, compared to the native enzyme. Thus, a
possible role for the plant large subunit in potato tuber is to increase the sensitivity for the
activator and decrease the sensitivity for the inhibitor. Table 2 presents a matrix of amino
acid sequence identity/ similarity of various ADP-Glc PPases based on pair-wise
alignments. The enzymes from heterotrophic bacteria are highly conserved (80-98 %) but
only 30-35 % identical to the enzyme from cyanobacteria and plants. However, the
enzymes of bacterial and plants do exhibit many conserved domains (Figure 12). There
are regions where the relationship between the bacterial and plant enzymes is minimal or
nonexistent, and both bacterial and plant enzymes exhibit significant identity among
themselves. Examples of this include the extreme N-tenninal and C-terminal regions of
the enzyme. One might speculate that the universally conserved regions are catalytic
while the regions conserved only in plants or bacteria are regulatory elements. As

discussed below, a large number of biochemical data supports this argument.

5.3 Structure of the Pyrophosphorylase Domain.

There is no reported structure of ADP-glucose pyrophosphorylase. However,
recent work has produced the three-dimensional structure of the E. coli N-
acetylglucosamine l-phosphate uridylyltransferase (GlmU) involved in the biosynthesis
of peptidoglycan in bacteria (122). The crystal structure of a truncated form of E. coli

GlmU was solved to 2.25 A and 2.3 A resolution for the apo-GlmU and

39

£553 :85
.828 880A nowa— Jom 3533 ema— afioflg nowa— g £553 :95. angina "—35 g

.8038 8:268 ogéfim no women Be macaw

 

 

30:8 - - - - - =8... .5
i830 8.23.0 - - - - awe. é<
8.23.0 3.05.0 3.88.0 - - - :8: .8...
R285 $23.0 @235 3.28.0 - - sees a
$22.0 3.2.2.0 $05.0 $82.0 $82.0 - use
as 8m as. .5 a3 .2... =8; .3. see... a use

 

.mcéosmmosmea omens—wda *0 banana. \bucoE oucoscom Que 0% mo x132 "N 033.

40

.2553 owns .693 2.88 ”omega

32253 :25 .893 888 #3838 £553 own: .32 38.235 ”NEQEQEV Howagfla 325:8 :25 33.52: «NQQEQEV
Haemfifla ”afieneé Hosanna 58 .m ”Sacco ”moms—boimonmoinomooflwia .3 80.58 8 >3— .»on E 33%

0.5 20223525 gas/.2828 fine #85 5 33% mm 0:282 3538 < 63555 0.838 05 mo 8: .3 BEE—2&2 Pa
@223: 8 53202 .x. o3 $8583 338200 .momflbofimofiea emcee—win? 3.55» .«o 258:3? Eon 92:2 .N— onE

 

 

 

 

 

emfieauoo emueuuon

eMueanueue emueanueue

Hal-ouch . H, .. Sea-ac.—

uueEenuE—e ,... KG. . Sausage

Pinon: ,: , .egé-m -: 22:12.... 0931.3

£00. _ . _ I. _. . a ‘53:, :‘ . £908

emueauon . . m3 enuenuoa

emudanueun . .. ....c<h3.+...; h: 081136.:

315303 I. “as... .2 33.0 «I .. a: .3330.-

.ndelenueuu .. . . . :..VE..._..;..;I..._5.2z . .. 0: 316534.:

0.1923 ,..,_....£....qc4.5.i£2.20. .. on 0.5334

oak—cue . . . won 83000

emueauon ... - . , . .V.. . «2:2,. fa. . . 09:39.—

emuegueuc fin»: imam-o H E: _ 09330.3

32—30“ $222.31»;— .5222 t 2.32: ”.2832. 38.305

.242 r a: _ 3.3333: 222.2 93. .32: .32 315939:

...:...z...5 82934 3:32.22 5...... chance

, 8.3000 , . our—000

,, . .. .., omueauou . omueauoa

H . 09359.3 592.532.. emueEueue
,. mgmgi ”3 "132224” 33302 H2....a.u..2§22.2.a.x_>iee<oooa: .2332—
...e; 2......1122 3302.234 :23 25:. 2.5.5232. M2305: sausage
_ , , N . 0.1345. 5 525a: Z. L .: 11,: 2,... gauge
. .. Pincus 2., i. al.... 4...... 829000

 

 

4r

GlmU/UDP-GlcNAc complex, respectively. GlmU is a bifunctional enzyme that

catalyzes two reactions in presence of MgClzz

(1) GlcN-l-P + acetyl-CoA —) GlcNAc-l-P + CoA

(2) GlcNAc + UTP —> UDP-GlcNAc + PPi

The first reaction involves an acetyltransferase activity, whereas the second reaction
involves a pyrophosphorylase activity to synthesize the nucleotide-activated form of N-
acetylglucosamine. The truncated GlmU, for which the crystal structure was solved, is
defective in acetyltransferase activity but has normal uridylyltransferase activity (122).
Biochemical experiments have shown GlmU is organized into two domains (123): The
pyrophosphorylase domain is encoded by the N-terminal (Asn3-Asn227) of GlmU and
shares homology with other pyrophosphorylases, whereas the C-terminal encodes the
acetyltransferase domain with 23 times the hexapeptide repeat (LIV)-(GAED)-X2-
(STAV)-X typical of bacterial acetyltransferases (124). The alignment of E. coli GlmU
with ADP-Glc PPases from diverse sources shows the conservation of several residues,
although the overall homology is weak (Figure 13). However, secondary structure
predictions of ADP-Glc PPases closely match the actual structure of the GlmU
pyrophosphorylase domain (125), which further validates GlmU as a prototypic template
for the pyrophosphorylase superfamily. The alignment only spans the N-terminal part of
ADP-Glc PPase, which encompasses the proposed active site (125). The remaining ADP-
Glc PPase sequence is probably involved in its allosteric regulation based on biochemical

evidence, as discussed in a later section.

42

Figure 13. Amino acid sequence alignment of ADP-glucose pyrophosphorylases and N-
acetylglucoseamine l-phosphate uridylyltransferase. Only the amino terminal regions
that share homology are shown. Several residues of E. coli N-acetylglucoseamine 1-
phosphate uridylyltransferase with proposed involvement in substrate binding or catalysis
are indicated in bold.

GlmU: E. coli N-acetylglucoseamine l-phosphate uridylyltransferase. Sources of ADP-
glucose pyrophosphorylases: anabe: Anabaena; stuss: Potato tuber small subunit; agrob:
Agrobacterium; r_sph: Rhodospirilum; ecoli: E. coli; bst_c: Bacillus stearothermophilus.

Consen: consensus sequence of ADP-glucose pyrophosphorylases.

43

 

an:
zz>om>mmqm0mm>¢>Hmmoilmow<quHiloewamooaz

-uueououz» uuuuuuuuuuuuuuuu a uuuuuuuuuu mHuunomo .........

mm>eo>oxz»om...........mm»¢mmmxxmm442mH>oxomommm..

€MABw>QmSMmm<QmOmO>Umqmmmm4>¢Awdm.BHXmHAQXGQOmm..
doHBw>om3>mm9m4082>0w0mhmm¢2¢xwz¥.>H<mHAQmwwommm..
¢DHBO>Qm3>mmmmhomm>UmQ¢hmm¢>¢M03m.>HMmHHQ&Umommm..
deBUHQm3M00...........%AM¢O>mZOAmB¢0mH>mmwma.....
<mHBwHszwoa...........mAM¢O>ZEQ.M¢¢DmHmewmQ.....

va— :Qflw

.mm
.2m
.2m
.9m
.z¢
.Bm

doxmfl>HUB>¥IwZ|mmBHmQNOBmQQAIIX>BAAGHUOOm¥¢0m4m049m>
ImXINIh IIIIIIIIII lewmlld IIIIIIIIII Bllflllllm IIIIII >0
<mx.&¢mm>Hm2wm2922H.Ummmflmm3m>mH>mHH>O¢O¥¢HmNQAEmO>D
fimx.n>hmHHMQZMQ><Z>.UhflmflmmHm>m204>80mdoxm>moHAZMmMD
{mm.NAmQBHmAZ¢D>mZ>.Umdh¢mdmm>BAUUHB>D<Oemum0mAZHme
dmx.MHmQHHmoxmz>mE>.Umwfidmzmm>mqowHB>Q<Umo>m0042wmwo
Om¥.fl<mmHHmwmmoH&EA.Umdhdmxmozm444>eHQ<QBmmm¢Omemwo
xmx.flmmDH>mOmZQszd.whamdmmooHmH>mABHD¢ZBmmmmOHhAmMQ

<oxqqowomo>>qm>m¢¢oqllmz¢<|loH>zo>z¢x

IIIIIIIIIIIIIIIIIII AIIIIIOBIIIIIIHImZIIZWIIIoHImlllll
02meAQS¢..meUHrmmAAAmOWOBA>O>BQHQmZEUZmABmQHHm»...M
mzmezmmm..30mOHm0>48mmOVOBH>02mmH0mZHUZmAdmoHHmh...¥
mzmmmmhmz..SOmOHmmHAmmdxmoedHflzmeUmZEéZmAdeHHmd...&
mzmmmmmmo..SOmOAmmHAmmdxwoed>DHmmmeZA<ZmA<moHHad...x
szx>0022mdwdmmqmmzqmflmZhOBA>>HXWHZWZQUZm>mHQHAmw..Zfl
www.mwmmz..medemzqmfimszBA>wH¥mHmmZH02m>mHQHAmM...M

s—Z

 

ZZB>Z¢AS¢XEQ<UZ€HAHOBZHMOHomO mOB
IIIIIIIIIIIIIIIII NHQZW¢IIIIIIIIIIIIIIIIIIIIIIIIIII
Z¢ZQHoqwoqumSZhHwszm<AZm.....................&mm
ammomququwofloh>wHmedqummoszm..............mmz
omdomHAAQmAh¢SDh>>H02m<4<z¢mowmH0..............mmo
Qfldomm>¢mzqufimh>wHOZmdqdbomzumH0..............mmo

wmmxomAAZAZ>QXWH>MHmedemZMX¢mXQDAOAHBBD>MZ¢OAOMG
.madeAAXHA>QXMh>wH02m¢HMmOmmdommBAGA>BBD>¢2¥BA¢E0

8:. :6 e5
mqu>aoquqHomaommmm¢¢ot012¢mweo qom<0q>zzq
nuwumootq uuuuuuuuuuuuuuu ot><u<eouuz uuuuuuuuuu quit-
2x»Hmoomq>q>womzwomHwszow><z<eomwsx>omm>mwmmq>e>o
owamao<gH>>wmmxwmmHHoqzoe>¢o¢eomxszmoxzmoo<mqqo>m
zxw>mno44HH>waowmo>HoHzoe>¢o¢eoqwzmzmo>mom¢mqHog»
zmmeno¢4H>zwmm<wmmHHoHzox>¢o¢eomwzoemm>mom¢mqHamm
zqumooaqHqwmq>emmmquqwom>¢o<900m30mzmmmoo¢¢4>m>m

mequOmAHAhmo>03m04234>0m>40¢900m3zmzmm900<4q>m>h

awn a. vzu .1—

o<qemq>zmlqomwilzmeoxo¢¢qH>>mz¢zzqzaiaeaeezezezee:
uuuu><mxauuuequiqmuouoouq uuuuuuuu zeieeeeeeeeeeeeee
oomm>¢mu¢H2994mmqmmoooo¢qqz¢Hoxmxz:zeezezzezzezzee
oohm>¢mx¢mxzeqoxqmeomoo<qHq<>qumqom¢42gmozxoqm>z¢
momw><mm<mmoaquqmmomuo¢4>m¢z<o¢aqmqmmomx2zztiara:
momw><mm<mmoaququmomoo¢q>wmz¢om¢q...mo>mxmmzeeeez
o.qm>¢mx¢mxxeqqumeomooquHoq>mmm¢omo4090m20mom><z
o¢>m>¢mx<queqmwqmeomowquH<q>xx2ziaeeaeezzzzeeeze

oeHo
mGmeOU
Autumnv
Aflaooov
Anamiuv
.nouamv
*mmsumv
Amnmcmv

asao
mamCOU
Autumne
*fiaooov
Anamiuv
Anoummv
*mmsumw
Amnmcmv

seam
mamCOU
Autumnw
Awaoomv
Azania.
Anoummv
Ammsumv
Amnmcmv

44

The structure of the GlmU pyrophosphorylase domain consists of a central seven-
stranded mixed B-sheet surrounded by 6 (it-helices, 3 on each side of the B-sheet (Figure
14A). The uridylyltranserase active site, identified by the presence of soaked UDP~
GlcNAc, is in a large open pocket. This 21 A long and 13 A deep pocket is formed by
two lobes: The first lobe consists of residue Asn3-Vall 11 and the surface loop Hi5216-
Asn227, whereas the second lobe consists of the remaining residues of the
pyrophosphorylase domain (Figure 14A,B). The interaction between GlmU and the
substrate UDP-GlcNAc is shown in detail in Figure 14C,D. The first lobe exposes Leul l,
Gly14, Gly16, Arg18, and LysZS towards the pocket along with Gln76-Thr82. The second
lobe exposes the region Gly138-Asp157 to the pocket, whereas the floor of the active site
is made from the regions TyrlO3-Asp105 and Val223-Asn227. The sugar and nucleotide
moieties of UDP-GlcNAc contacts the protein, whereas the phosphate groups are totally
solvent accessible. Based on the structure and biochemical evidence, the authors propose
Arg18 and possibly LysZS as catalytic residues, which are conserved in all ADP-Glc
PPases (Figure 13). However, the article does not propose a catalytic mechanism for the
enzyme. The assigned amino acids for nucleotide binding are Leul 1, Ala13, Gln76, and
AsplOS, where the leucine and aspartate residues are also found in ADP-Glc PPases. The
sugar-binding site of GlmU is composed of Gly81, AsplOS, Tyr139, Glu154, Asn169,
Tyr197, and Thrl99. In ADP-Glc PPase we only find Gly81, AsplOS and Glu154 as
being uniquely conserved. It is possible the sequence variations between ADP-Glc PPases
and GlmU of the above residues are largely due to their different substrate specificity,
whereas catalytic residues and amino acids important for substrate interaction but not

sugar- or nucleotide recognition are strictly conserved in the enzyme family. Certainly,

45

Figure 14. Structure of the pyrophosphorylase domain of N-acetylglucosamine 1-
phosphate uridylyltransferase. Adapted from, Brown et al (122). This figure is
presented in color.

(A) Ribbon diagram of the pyrophosphorylase domain showing the first lobe
(Asn3-Vall l 1) in blue and the second lobe (Glu112-Gln213) in magenta, the
bound UDP-GlcNAc in orange and the consensus motif including Arg18 and
LysZS in red.

(B) Electrostatic potential mapped on the molecular surface from —7kT (red) to
+7kT (blue) with the bound UDP-GlcNAc within the pocket.

(C) A close-up view of the pyrophosphorylase domain (orientation as in Figure
14A) with UDPGlcNAc (orange), and the side chain residues Arg18, LysZS
(catalytic residues; red), Leul 1, Ala13, Gln76, and Asp105 (nucleotide
binding site; pink), Gly81, AsplOS, Tyr139, Glu154, Asn169, Tyr197, and
Thr199 (sugar binding site, green) and Gln193, Glu195, Asn227, and Gln231
(yellow).

(D) Superposition of the apo-GlmU (yellow) and the GlmU/UDP-GlcNAc
complex (green). Backbone regions that deviate significantly are highlighted
with their side chains. A rigid-body motion of regions Va113 1-Gln166 and
Va1187-Tyr197 is observed, along with significant movements of Tyr103,
Tyr139 and Gln193 side chains within the active site.

46

 

47

 

48

the conservation of multiple residues proposed in substrate binding and/or catalysis
suggests there is a shared mechanism between XDP pyrophosphorylases and that useful
information can be extracted from studying GlmU. This includes design of site-directed
mutagenesis experiments of the ADP-Glc PPase enzymes to identify residues important
for catalysis and substrate binding. In fact, site-directed mutagenesis studies of E. coli
ADP-Glc PPase residue D142 that aligns with D105 of GlmU (Figure 13) is currently
being performed in our laboratory. Preliminary results strongly suggest that D142 of E.
coli ADP-Glc PPase has a direct role in catalysis rather than substrate binding as
proposed by the authors of the GlmU structure.

The structure of GlmU may even serve as a model in solving the crystal structure
of ADP-Glc PPase, although the authors have not released its atomic coordinates. Still,
the known topology of GlmU will facilitate the tracing of an ADP-Glc PPase electron

density map.

5.4 Active and Regulatory Sites.

A large amount of effort has focused on answering which domains are responsible
for allosteric regulation, substrate binding, catalysis, and oligomerization in the ADP-Glc
PPases. Site-directed mutagenesis, chemical modification, limited proteolysis, and
chimeric enzyme construction are some of the techniques that have been used to address
these questions.

As discussed above, several regions in all ADP-Glc PPases are conserved. When
the invariant residue K195 (Figure 15) in the E. coli enzyme was subjected to site-

directed mutagenesis, the apparent affinity (S05) for Glc-l -P decreased dramatically,

49

suggesting involvement in Glc-l-P binding (126). When the equivalent position in the
potato tuber small subunit (K198) was changed to Arg, Ala or Glu, the S05 for G-l-P
decreased 135-550 fold, while the apparent affinities for allosteric effectors and other
substrates were unaffected (127). Of importance is the conservation of this Lys in GlmU
(as LyslS6; Figure 13) but it has not been assigned a ftmction in this enzyme, nor has the
authors described its conservation. However, it is in close proximity to GlmU residue

G1u154 involved in binding of the sugar moiety.

ecopyro EFVEKPANPP
anapyro DFSEKPKGEA
arathsmall EFSEKPKGEH
potsmall EFAEKPQGEQ
arathlarge SFSEKPKGDD
potlarge QFAEKPKGFD

Figure 15. G-l-P binding site of ADP-Glc PPase. The E. coli ADP-Glc PPase residue
K195 and equivalent positions are indicated in bold letters. Refer to legend of Figure 12

for key to sources of ADP-Glc PPases.

Interestingly, mutagenesis of the equivalent lysine, K213, in the large subunit of potato
tuber had no effect on the 80.5 for G-l-P, and thus it seems unlikely that the large subunit
participates in Glc-l-P binding (127). These results are consistent with the small subunit
being involved in catalysis and the large subunit being a modulator of the regulatory

properties of the small subunit of plant ADP-Glc PPases.

50

The site for ATP binding has been investigated by chemical modification using
the substrate analog 8-azido-AMP with the E. coli enzyme (128). The major binding site
of 8-azido-AMP was found in the region indicated in Figure 16, specifically at the Y114
residue. When the residue Y114 was changed by site-directed mutagenesis to
phenylalanine, a 10-20 fold effect on 805 for ATP was seen, but the apparent affinities for
other substrates and allosteric effectors was changed as well (129). Furthermore, this
Y114 is not conserved, as the plant ADP-Glc PPases have a Phe residue at this position
and GlmU has a deletion in this particular region. These results indicate that Y114

probably is not the ATP binding site, although it may be close to it.

ecopyro Q..RMKGENW YRGTADAVTQ
anapyro QTP..ENPNW FQGTADAVRQ
arathsmall QSP. .ENPNW FQGTADAVTD
potsmall QSP..ENPDW FQGTADAVRQ
arathlarge QTPGESGKRW FQGTADAVRQ
potlarg QTPGEAGKKW FQGTADAVRK

Figure 16. Proposed ATP binding site of ADP-Glc PPase. Residue Y114 of E. coli ADP-
Glc PPase and the equivalent positions are indicated in bold letters. Refer to legend of

Figure 12 for key to sources of ADP-Glc PPases.

Pyridoxal phosphate (PLP) has been found to be an activator of many ADP-Glc
PPases (52, 130, 131). PLP probably mimics the 2 phosphate groups of FBP and the
phosphate and carboxylate group of 3-PGA, respectively (132). Using PLP in chemical
modification experiments it was found that residue K39 in the E. coli enzyme, K440 in

the small subunit of spinach leaf enzyme and K419 of the Anabaena enzyme was

51

involved in the binding of their respective activators. Further studies by PLP modification
showed that also K3 82 of the Anabaena enzyme (131) and the equivalent lysine in the
spinach leaf large subunit (130) was involved in activation. In conclusion, there appears
to be two activator sites for 3-PGA in the C-terminal (Figure 17). Site-directed
mutagenesis studies on the potato tuber enzyme have showed that the activator sites in the

small subunit are more important than the equivalent sites in the large subunit (121).

Activator site 1 Activator site 2

 

ecopyro NAEEDARRFY RSEEGIVLVT ~~~~~~~~~~~~~~~~
anapyro IRRAIIDKNA RIGHDVKIIN WVLKNAVIT DGTII~
arathsmall IKRAIIDKNS RIGDNVKIIN VTVIKDALIP TGTVI"
potsmall IKRAIIDKNA RIGDNVKIIN VTVIKDALIP SGIII”
arathlarge IQECIIDKNA RVGKNVIIAN TVILKNSVIK DGWI*
potlarge IRKCIIDKNA KIGKNVSIIN IIILEKATIR DGTVI"

Figure 17: Activator sites of cyanobacterial and plant ADP-Glc PPases. Refer to legend of

Figure 12 for key to sources of ADP-Glc PPases.

Limited proteolysis by proteinase K of E. coli ADP-Glc PPase in presence of
Mg”, ADP-Glc, and FBP, generated an enzyme with 10 to 13 amino acids removed at the
N-terminal and 2 amino acids removed at the C-terminal (133). This truncated enzyme
was almost independent of FBP for maximal activity and insensitive to AMP inhibition,
whereas the apparent affinities for substrates were unaltered. Using genetic engineering,
an enzyme (NM 1, Figure 18) lacking the first 11 amino acids was constructed and found

to have properties identical to the proteolyzed enzyme (134).

52

l 10
E. coli native MVSLEKNDH LMLARQLP
E. COli Nd1-11 LARQLP

Figure 18: E. coli ADP-Glc PPase NM“.

The 11 amino acids that appear to be important for allosteric regulation contain a lysine
residue (K6 of E. coli enzyme) found in several bacterial ADP-Glc PPases which are
activated by FBP. It remains to be determined by site-directed mutagenesis if this lysine
residue can be changed with resulting loss of allosteric regulation. However, the above
results suggest that elements at the N-terminal are involved in allosteric regulation of
bacterial enzymes.

In a study by Sheng and Preiss (135), the Anabaena enzyme was treated with the
arginine specific reagent phenylglyoxal, which resulted in a time- and concentration
dependent loss of activity. The inhibitor phosphate most effectively protected the enzyme
from phenylglyoxal inactivation. Five arginine residues conserved in higher plant- and
cyanobacterial ADP-Glc PPases (but not in bacterial enzymes) were individually changed
to alanine. This approached produced a mutant R294A with IOO-fold lower affinity for
the inhibitor in the presence of activator, 3-PGA, and 40-fold in absence of activator.
Because the R294A substitution had little impact on the kinetic constants for substrates
and activator, these results suggest that R294 is specifically involved in inhibitor binding
(Figure 19). However, when the potato tuber small subunit was subjected to site-directed
mutagenesis at the equivalent position (Figure 19), the resulting mutant showed no effect

on phosphate sensitivity but unexpectedly lowered 5-10 fold the apparent affinity for

53

3-PGA (136). Clearly, more work is needed in order to locate exactly what the phosphate
binding site is, and why the effect of changing this residue apparently differs between

cyanobacteria and plants.

ecopyro YNESLPPAKF
ana ro RARYLPPTKL
arathsmall QPRYLPPSKM
pot smal l QPRYLPPSKM
arathlarge SRRNLPPSKI
potlarge SPRFLPPTKI

Figure 19: Proposed Anabaena Phosphate binding site. Residue R294 and conserved
equivalent positions are indicated in bold letters. Refer to legend of Figure 12 for key to

sources of ADP-Glc PPases.

The ADP-Glc PPases of a number of allosteric mutants have been cloned (137-
139). In almost all cases these mutations occur at sites different from the allosteric and
substrate binding sites. Thus the regions where the mutations occur may be involved in
maintaining the conformation of the inactive WT enzyme (T state in Monod’s allosteric
model; 140). Here, I will only describe the ADP-Glc PPase mutant deduced from E. coli
strain 618, which accumulates 300 % more glycogen than the WT (E. coli K-12) strain
(138). The nucleotide sequence of the ADP-Glc PPase gene showed a mutation, resulting
in a substitution of G336 to aspartate (Figure 20). The single mutation G336D was solely
responsible for the altered allosteric properties, as determined by site-directed

mutagenesis of the WT enzyme to G336D. G336D had identical kinetics to the enzyme of

54

strain 618 (139). G336D has a 5-fold lower A05 for FBP and a 12-fold higher 105 for AMP
compared to the WT enzyme. In addition, only a 2-fold activation by FBP was observed
compared to a 35-fold activation as seen by the WT enzyme (141). Thus, the substitution

G336D appears to result in “conformational steering” towards the active form (R-state)

compared to the WT enzyme.

E.Coli G336D TLNSLVSDGC
E.Coli WT TLNSLVSGGC
anapyro VTESIIGEGC
arathsmall VTDSVIGEGC
arathlarge LIDSIISHGS

Figure 20: Position of G336 of E. coli ADP-Glc PPase, indicated in bold. Refer to legend

of Figure 12 for key to sources of ADP-Glc PPases.

55

10.

11.

12.

I3.

14.

15.

References

. Hizukuri S. (1995) in Carbohydrates in Food (Elliasson, A.C, Ed.) Acad. Press, NY.

pp. 347-428.

Preiss, J. and Sivak, MN. (1998) in Comprehensive Natural Products Chemistry
(Pinto, B.M, Ed.) Pergamon Press, Oxford, Vol. 3, pp. 441-495.

Sivak, MN. and Preiss, J. (1998) Advances in Food and Nutrition Research,
Academic Press, San Diego, CA., V0141.

Smith, A.M., Denyer, K., and Martin, C. (1997). Annu. Rev. Plant Phisiol. Plant.

Mol. Biol. 48, 67-87.

. Nelson, OE. and Pan, D. (1995) Plant Mol. Biol. 46, 475-496.

Preiss, J. (1991) Oxf Surv. Plant. Mol. Cell Biol. 7, 59-114.
Lin, T.P., Casper, T., Sommerville, C. Preiss, J. (1988). Plant Physiol. 86, 1131-1135.
Heldt, H.W., Flugge, UL and Borchert, S. (1991) Plant Physiol. 95, 341-343.

Heldt, H.W. and F lugge, UL (1987) in Biochemistry of Plants (Stumpf, PK. and
Conn, E.E., Eds.) Academic Press, NY, Vol. 12, pp. 49-85.

Badenhuizen, NP. (1965) in The Biogenesis of Starch Granules in Higher Plants.
Appleton-Century Crofts, NY.

Tyson, RH. and ap Rees, T. (1988) Planta 175, 33-38.

Hill, L.M. and Smith, A.M. (1991) Planta 185, 91-96.

Viola, K, Davies, H.V. and Chudeck, AR. (1991) Planta 183, 202-208.

Preiss, J. (1988) in The Biochemistry of Plants. (Preiss, J ., Ed.) Academic Press, NY.
Vol. 14, pp. 181-254.

Okita, T.W. (1992) Plant Physiol. 100, 560-564.

56

16.

17.

18.

19.

20.

21.

22.

23.

24.

25.

26.

27.

28.

29.

30.

31.

32.

33.

Liu, T.T. and Shannon, J .C. (1981) Plant Physiol. 67, 525-529.

MacDonald, RD. and ap Rees, T. (1983) Phytochemistry 22, 1141-1143.

Keeling, P.L., Wood, J .R., Tyson, RH, and Bridges, LG. (1988) Plant Physiol. 87,
311-319.

Pozueta-Romeo, J ., Ardila, F. and Akazawa, T. (1991) Plant Physiol. 97, 1565-1572.
Ball, 8., Marianne, T., Dirick, L., Fresnoy, M., Delrue, B., and Decq, A. (1991)
Planta 185, 17-26.

Muller-Rober, B., Sonnewald, U. and Willmitzer, L. (1992) EMBOJ. 11, 1229-1238.
Stark, D.M., Timmerman, K.P., Barry, G.F., Preiss, J ., and Kishore, GM. (1992)
Science 258, 287-292.

Preiss, J. Sivak, M. (1995) in Seed Development and Germination 1St ed., Marcel
Dekker, NY. pp. 139-168.

Manners, DJ. (1989) Carbohydr. Polym. 11, 87-112.

Detherage, W.L., MacMasters, M.M., and Rist, CE. (1955) Trans. Am. Assoc.
Cereal. Chem. 13, 31-42.

Gallant, D.J., Bouchet, B., and Baldwin, PM. (1997) Carbohydr. Polym. 32, 177-191.
Jenkins, P.J., Cameron, RE, and Donald, A.M. (1993) Starke 45, 417-420.

Nikuni, Z., Chorikagaku 2, 6-23.

French, D. (1972) Denpun Kagaku 19, 8-33.

Hizukuri, S. (1986) Carbohydr. Res. 147, 342-347.

Lineback, D. R. (1984) Baker ’s Dig. 58, 16-21.

Katz, J .R. and Van Italie, TE. (1930) Z. Phys. Chem. Abt. A150, 90-100.

Hizukuri, S. (1985) Carbohydr. Res. 141, 295-306.

57

34.

35.

36.

37.

38.

39.

40.

41.

42.

43.

44.

45.

46.

Buleon, A., Gallant, D.J., Bouchet, B., Mouille, G., D’Hulst, C., Kossman, J ., and
Ball, S.G. (1998) Plant Physiol. 115, 949-957.

Smith, A.M. and Martin, C. (1993) in Biosynthesis and Manipulation of Plant
Products (Grierson, D., Ed.) Blackie Academic and Prof. Pub. Glascow. Vol. 3, pp. 1-
54.

Ball, S., Guan, H.P., James, M., Myers, A., Keeling, P., Mouille, G., Buleon, A.,
Colonna, P., and Preiss, J. (1996) Cell 86, 349-352.

Correns, C. (1901) Bibl. Bot. 53, 1-161.

Sumner, J .B. and Somer, GP, (1944) Arch Biochem. 4, 4-7.

Erlander, S. (1958) Enzymologia 19, 273-283.

Nakamura, Y., Umemoto, Y., Takahata, Y., Komae, K., Amano, E., and Satoh, H.
(1996) Plant Physiol. 97, 491-498.

James, M.G., Robertson, D.S., and Myers, A.M. (1995) Plant Cell 7, 417-429.
Mouille, G., Maddelein, M.L., Libessart, N., Talaga, P., Decq, A., Delrue, B., and
Ball, S. (1996) Plant Cell 8, 1353-1366.

Nakamura, Y. (1996) Plant Sci. 121, 1-18.

Van de Wal, M. D’Hulst, C., Vincken, J .P., Buleon, A., Visser, R., and Ball, S.

( 1998) J. Biol. Chem. 273, 22232-22240.

Maddelein, M.L., Libessart, N., Bellanger, F ., Delrue, B., D’Hulst, C., Koomhuyse,
N., Bossu, J .P., Fontaine, T., Wieruszeski, J .M., Decq, A., and Ball, S. (1994) J. Biol.
Chem. 269, 25150-25157.

Kram, A.M., Ooertergetel, GT, and Van Bruggen, E.F.J. (1993) Plant Physiol. 101,

23 7-243.

58

47. Nakamura, T., Yamamori, M., Hirano, H., Hirada, S., and Nagamine, T. (1995) Mol.
Gen. Genet. 248, 253.

48. Delrue, B., Fontaine, T., Routier, F., Decq, A., Wieruszeski, J .M., Koomhuyse, N.,
Maddelein, M.L., Foumet, B., and Ball, S. (1992) J. Bacteriol. 174, 3612-3620.

49. Weatherwax, P. (1922) Genetics 7, 568-572.

50. Murata, T., Sugiyarna, T., and Akazawa, T. (1965) Biochem. Biophys. Res. Comm. 18,
371-3 76.

51. Takeda, Y. and Hizukuri, S. (1982) Carbohydr. Res. 102, 312-327.

52. Preiss, J. (1984) Ann. Rev. Microbiol. 38, 419-458.

53. Holme, T. (1957) Acta Chem. Scand. 11, 763-775.

54. Strange, RE. (1958) Nature 220, 606-607.

55. Slock, J .A. and Stahly, DD. (1974) J. Bacteriol. 120, 399-406.

56. Geddes, R. (1985) in The Polysaccharides (Aspinall, G.O., Ed) Academic Press,
Sandiego, CA. Vol. 3, pp. 283-336.

57. Recondo, E. and Leloir, LP. (1961) Biochem Biophys Res. Comm. 6, 85-88.

58. Preiss, J ., Shen, L., Greenberg, E. and Gentner, N. (1966). Biochemistry 5, 1833-
1845.

59. Barengo, R., F lawia, M., and Krisman, CR. (1975) FEBS Lett. 53, 274-278.

60. Kawaguchi, K., Fox, J ., Holmes, E., Boyer, C., and Preiss, J. (1978) Arch. Biochem
Biophys. 190, 385-397.

61. Hestrin, S. (1960) in The Bacteria (Gunsalus, I.C., Stanier, R.Y, Eds.) Academic

Press, NY. Vol. 3, pp. 373-388.

59

62.

63.

64.

65.

66.

67.

68.

69.

70.

71.

72.

73.

74.

75.

Preiss, J. and Walsh, DA. (1981) in Biology of Carbohydrates (Ginsburg, V.,
Robbins, P., Eds) Wiley, NY, pp. 199-314.

Okita, T.W., Rodriguez, R.L. and Preiss, J, (1981) J. Biol. Chem. 256, 6944-6952.
Romeo, T., Kumar, A. and Preiss, J. (1988) Gene 70, 363-376.

Iglesias, A.A., Barry, G.F., Meyer, C., Bloksberg, L., Nakata, P.A., Greene, T.,
Laughlin, M.J., Okita, T.W., Kishore, GM. and Preiss, J. (1993). J. Biol. Chem. 268,
1081-1086.

Ballicora, M.A., Laughlin, M.J., Fu, Y., Okita, T.W., Barry, GR and Preiss, J.
(1995). Plant Physiol. 109, 245-251.

Irnparl-Radosevich, J .M., Nichols, D.J., Li, P., McKean, A.L., Keeling, P. and Guan,
HP. (1999) Arch. Biochem. Biophys. 362, 131-138.

Guan, H.P., Baba, T. and Preiss, J. (1994) Plant Physiol. 104, 1449-1453.

Guan, H.P., Baba, T. and Preiss, J. (1994) Cell. Mol. Biol. 40, 981-988.

Guan, H.P., Kuriki, T., Sivak, M., Preiss, J. (1995) Proc. Natl. Acad. Sci. 92, 964-
967.

Kumar, A., Larsen, C. E., Preiss, J. (1986). J. Biol. Chem. 261, 16256-16259.

Cao, H., Imparl-Radosevich, J .M., Guan, H.P., Keeling, P., James, M., and Myers,
A.M. (1999) Plant Physiol. 120, 205-216.

Martin, C. and Smith, A.M. (1995) Plant Cell 7, 971-985.

Myers, A.M., Morell, M.K., James, M.G., and Ball, S.G. (2000) Plant Physiol. 122,
989-997.

Okita, T.W., Rodriquez, R.L. and Preiss, J. (1981) J. Biol. Chem. 256, 6944.

60

76.

77.

78.

79.

80.

81.

82.

83.

84.

85.

86.

87.

88.

F urukawa, K., Tagaya, M., Inouye, M., Preiss, J. and Fukui, T. (1990) J. Biol. Chem.
265, 2086-2090.

Furukawa, K., Tagaya, M., Tanaziwa, K., and Fukui, T. (1993) J. Biol. Chem. 268,
23837-23842.

Furukawa, K., Tagaya, M., Tanaziwa, K., and Fukui, T. (1994) J. Biol. Chem. 269,
868-871.

Holmes, E. and Preiss, J (1982). Arch. Biochem. Biophys. 216, 736-740.

Krisman, CR. (1962) Anal. Biochem. 4, 17-23.

Hawker, J. S, Ozbun, J. L., Ozaki, H., Greenberg, E., Preiss, J. (1974). Arch.
Biochem. Biophys. 160, 530-551.

Takeda, Y., Guan, H. P. and Preiss, J. (1993). Carbohydr. Res. 240, 253-263.
Hizukuri, S, Takeda, Y., Yasuda, M., and Suzuki, A. (1981) Carbohydr. Res. 205-
213.

Boyer, C. and Preiss, J. (1977). Biochem. 16, 3693-3699.

Baecker, P.A., Greenberg, E. and Preiss, J. (1986). J. Biol. Chem. 261, 8738-8743.
Boyer, C. and Preiss, J. (1978). Carbohydr. Res. 61, 321-334.

Singh, B. K. and Preiss, J. (1985). Plant Physiol. 79, 34—40.

Mizuno, K., Kawasaki, M., Shimada, H., Satoh, H., Kobayashi, E., Okumura, S.,

Arai, Y. and Baba, T. (1993). J. Biol. Chem. 268, 19084-19091.

89. Nakamura, A., Haga, K., Ogawa, S., Kuwano, K., Kimura, K., and Yamane, K.

90.

91.

(1992). FEBS Lett. 296, 37-40.
Dang, P. L., and Boyer, c. D. (1988) Phytochem. 27, 1255-1259.

Guan, H. P. and Preiss, J. (1993). Plant Physiol. 102, 1269-1273.

61

92. Mendel, G. (1865) Vehr. Nat. Forsh. Ver. Brunn. 4, 3-47.

93. Kooistra, E. (1962) Euphytica 11, 357-373.

94. Bhattacharyya, M.K., Martin, C. and Smith, A.M. (1990) Plant Mol. Biol. 22, 525-
531.

95. Shannon, J .C. and Garwood, D.L. (1984) in Starch: Chemistry and Technology
(Whistler, R.L., BeMiller, J .N., Pascal], E.F. Eds) Academic Press, Orlando, FL. pp.
25-86.

96. Mizuno, K., Kawasaki, T., Shimada, H., Satoh, H., Kobayashi, E., Okumura, S., Arai,
Y. and Baba, T. (1993).]. Biol. Chem. 268, 19084-1919091.

97. Hong, S., Mikkelsen, R. and Preiss, J. (2000) Arch. Biochem Biophys. In Press.

98. Hilden, I. Leggio, L., Larsen, S. and Poulsen, P. (2000) Eur J Biochem. 267, 2150-
2155.

99. Binderup, K., Mikkelsen, R. and Preiss, J. (2000) Arch. Biochem. Biophys. 377, 366-
371.

100. Matsuura, Y., Kusunoki, M., Harada, W. and Kakudo, M. (1984). J. Biochem. 95,
697-702.

101. Buisson, G., Duee, E., Haser, R. and Payan, F. (1987). EMBOJ. 6, 3909-3916.

102. Klein, C. and Schulz, G. E. (1991). J. Mol. Biol. 217, 737-750.

103. Katsuya, Y., Mezaki, Y., Kubota, M. and Matsuura, Y. (1998).]. Mol. Biol. 281,
885-897.

104. Farber, G. and Petsko, GA. (1990) Trends. Biochem. Sci. 15, 228-234.

105. Baba, T., Kimura, K., Mizuno, K., Etoh, H., Ishida, Y., Shida, O. and Arai, Y.,

(1991). Biochem. Biophys. Res. Commun. 181, 87-94.

62

106. Svensson, B. (1994). Plant. Mol. Bio. 25, 141-157.

107. Kuriki, T., Takata, H., Okada, S. and Imanaka, T. (1991). J. Bacteriol. 173, 6147-
6152.

108. Takata, H., Kuriki, T., Okada, S.,Takesada, Y., Iizuka, M., Minamiura, N. and
Imanaka, T. (1992). J. Biol. Chem. 267, 18447-18452.

109. Uitdehaag, J .C.M., Mosi, R., Kalk, K.H., Van der Veen, B.A., Dijkhuizen, L.,
Withers, S.G. and Dijkstra, B.W. (1999) Nat. Struc. Biol. 6, 432-436.

110. Cao, H. and Preiss, J. (1996) J. Protein Chem. 15, 291-304.

111. Libessart, N. and Preiss, J. (1998) Arch. Biochem. Bi0phys. 360, 135-141.

112. Funane, K., Libessart, N., Stewart, D., Michishita, T. and Preiss, J. (1998).].
Protein Chem. 17, 579-590.

113. Jespersen, H.M., MacGregor, B.A., Sierks, MR. and Svennson, B. (1991)
Biochem. J. 280, 51-55.

114. Janecek, S., Svennson, B. and Henrissart, B. (1997) J. Mol. Evol. 45, 322-331.

115. Kuriki, T., Stewart, DC. and Preiss, J. (1997).]. Biol. Chem. 272, 28999-29004.

116. Espada, J. (1962). J. Biol. Chem. 237, 3577-3581.

117. Ghosh, HP. and Preiss, J. (1966). J. Biol. Chem. 241, 4491-4504.

118. Plaxton, W. C. and Preiss, J. (1987). Plant Physiol. 83, 105-112.

1 19. Preiss, J. (1997). In Engineering Improved Carbon and Nitrogen Resources.
Taylor and Francis, London, pp 81-104.

120. Smith-White, B. S. and Preiss, J. (1992). J. Mol. Evol. 34, 449-464.

121. Ballicora, M.A., Fu, Y., Nesbitt, N. M. and Preiss, J. (1998). Plant Physiol. 118,

265-274.

63

122. Brown, K., Pompeo, F ., Dixon, S., Mengin-Lecreulx, D., Cambillau, C. and
Boume, Y. (1999) EMBO J. 18, 4096-4107.

123. Gehring, A.M., Lees, W.J., Miniola, D.J., Malsh, QT. and Brown, ED. (1996)
Biochemistry 35, 579-585.

124. Vaara, M. (1992) FEMS Microbial Lett. 76, 249-254.

125. Ballicora, M.A., Fu, Y., Wu, M., Sheng, J ., Nesbitt, NM. and Preiss, J. (1996) in
Regulation and Manipulation of Starch and Sucrose Metabolism in Plants.
(Nakamura, Y., Ed.) Tsukuba, Japan. pp 5-11.

126. Hill, M. A., Kaufmann, K., Otero, J ., Preiss, J. (1991). J. Biol. Chem. 266, 12455-
12460.

127. Pu, Y., Ballicora, MA. and Preiss, J. (1998). Plant Physiol. 117, 989-996.

128. Lee, Y.M. and Preiss, J. (1986). J. Biol. Chem. 261, 1058-1064.

129. Kumar A., Tanaka, T., Lee Y.M. and Preiss, J. (1988). J. Biol. Chem. 263, 14634-
14639.

130. Morell, M., Bloom, M, Preiss, J. (1988). J. Biol. Chem. 263, 633-637.

131. Chamg, Y., Iglesias, AA. and Preiss, J. (1994). J. Biol. Chem. 261, 1058-1064.

132. Preiss, J. (1996). Regulation of Glycogen Synthesis. In Escherichia coli and
Salmonella typhimurium: Cellular and Molecular Biology, second edition, Amer. Soc.
Microbiol., Washington, DC. Vol. 1, pp. 1015-1024.

133. Wu, M. and Preiss, J. (1998). Arch. Biochem. Biophys. 358, 182-188.

134. Wu, M. and Preiss, J. (2000) Arch. Biochem. Biophys. In Press.

135. Sheng, J ., Preiss and J. (1997). Biochemistry, 36, 13077-13084.

136. Ballicora, M. and Preiss, J. Unpublished results

64

137. Preiss, J. Romeo, T. (1989). In Advances in Microbial Physiology, vol. 30, pp
183-238. Academic Press, New York/ London.

138. Leung, P., Lee, Y.M., Greenberg, E., Esch, K., Boylan, S., Preiss, J. (1986). J.
Bacteriol. 167, 82-88.

139. Meyer, C.R., Ghosh, P., Nadler, S., Preiss, J. (1993). Arch. Biochem. Biophys.
302, 64-71.

140. Monod, J ., Wyman, J ., Changeux, J. P. (1965). J. Mol. Biol. 12, 88-118

141. Meyer, C.R., Bork, J .A., Nadler, S., Yirsa, J ., Preiss, J. (1998). Arch. Biochem.

Biophys. 353, 152-159.

65

CHAPTER 2
SITE-DIRECTED MUTAGENESIS STUDIES OF TYR-300,

GLU-459, and GLU-527 in ESCHERICIA COLI BRANCHING ENZYME.

This data has been published in two separate articles:

Binderup, K. and Preiss, J. (1998).“ Glu-459 is Important for E. coli Branching Enzyme

Function”, Biochemistry 37, 9033-9037.

Mikkelsen, R., Binderup, K. and Preiss, J. (2001) “ Tyrosine Residue 300 is Important for

Activity and Stability of Branching Enzyme from Escherichia coli”. Arch. Biochem.

Biophys. Accepted for publication- in press.

66

Abstract.

The branching enzyme belongs to the amylolytic family, a group of enzymes that
cleave and/ or transfer chains of glucan. The amylolytic enzymes are homologous and all
contain 4 conserved regions, proposed to contain the active site.

By primary structure analysis the 3 conserved residues Tyr-300, Glu-459, and
Glu-527 in E. coli branching enzymes were identified. These residues are near the
conserved regions 1, 2 and 4, respectively, and hence map to the active site. We used site-
directed mutagenesis of the Glu-459, Tyr-300 and Glu-527 residues in the E. coli
branching enzyme in order to determine the significance of these conserved positions. A
substitution of Glu-459 to Asp resulted in increased specific activity compared to wild
type, suggesting that the mutation had created a more efficient enzyme. Changing Glu-
459 to Ala, Lys or Gln lowered the specific activities and altered the preferred substrate
from amylose to amylopectin. Mutagenesis of Glu-527 and in particular Tyr-300 affected
the enzymatic activity, indicating these residues are important for obtaining maximal
branching activity. In addition, substitution of Tyr-300 to Phe was found to affect the

thermal stability of branching enzyme.

67

Introduction

Branching enzymes catalyze the formation of 01-1 ,6-glucosidic linkages of
glycogen in bacteria (1) and of starch in plants (2). The enzyme hence plays an important
role in the biosynthesis of starch and glycogen (2). Branching enzymes from a variety of
sources have been identified and described including E. coli (3-5), maize (6-8), and rice
endosperm (9, 10).

Homology between the branching enzymes and other amylolytic enzymes was
first described by Romeo et al. (11). Subsequently, Baba et al. (12) established that
branching enzymes contain four highly conserved regions in the central portion of the
enzyme, which are thought to constitute the active site. These regions now define the
amylolytic enzyme family, which includes a-amylase, pullulanase, isoamylase,
cyclodextrin glucanotransferase and branching enzyme. Jespersen et al. (13) predicted all

of the above enzymes to fold into a (OI/13): barrel similar to that observed in the crystal

structures of a-amylases (14-16) and cyclodextrin glucanotransferases (17, 18). Site-
directed mutagenesis studies of the active center of neopullulanase (19) suggested that the
catalytic mechanism of the amylolytic enzymes was similar (20). In branching enzyme,
most of the seven amino acids invariantly found in the four conserved regions have been
subjected to site-directed mutagenesis and found crucial for catalysis or substrate binding
(21 -23).

Even though the members of the amylolytic family may share structural features
and regions of conserved amino acids, they catalyze different reactions and this poses the
question of which residues are responsible for the distinct catalytic properties. This work

describes studies on three conserved amino acids in branching enzymes. We investigated

68

the two residues Glu-459 and Glu-527, located immediately after the catalytic residues
Glu-458 (region 3) and Asp-526 (region 4), respectively, hereby forming two acid pairs
near the proposed catalytic site. These amino acids are not conserved in the other
members of the amylolytic family. For Glu-459, we investigate the properties of the
constructed E. coli branching enzyme mutants E459D, E459A, E459K, and E459Q and
conclude that the efficiency of catalysis as well as the preferred substrate for branching
activity is affected. Substitution of Glu-527 resulted in enzymes with 10-25 % of vvild-
type branching enzyme activity. Tyr-300 is found in all amylolytic enzymes prior to
region 1, and has been proposed to play a role in substrate binding based on the crystal
structures of a-amylase, isoamylase, and cyclodextrin glucanotransferase (24). We have
investigated the function of this Tyr in E. coli branching enzyme by its replacement to
Ala, Asp, Phe, Ser, and Trp. The major finding here is the dramatic loss of activity and

the decreased thermal stability upon substitution of Tyr-300 with Phe.

69

Materials and methods

E. coli BE expression vector. H.P. Guan kindly provided the plasmid pEXSB,
the construction of which has been described previously (5). The pEXSB plasmid is a
pET-23d derived plasmid containing the origin pl 5A, T7 promotor, T7 terminator,
kanamycin resistance, and the E. coli glgB gene encoding the E. coli BE.

Construction of mutants. The pEXSB plasmid purified from Epicurian Coli
XL2-Blue cells (Stratagene Corp., La Jolla, CA) was used as dsDNA template in site-
directed mutagenesis (Quickchange site-directed mutagenesis kit, Stratagene Corp).
Oligonucleotides were synthesized using an Applied Biosystems model 380A DNA
synthesizer (Macro-molecular Facility, Department of Biochemistry, Michigan State
University). The mutations were confirmed by DNA sequencing and secondary mutations
were ruled out by sequencing the entire gene for all constructions. DNA sequencing was
done by the dideoxy chain-terminating method (25) using the dsDNA as template. The
primers used were the T7 universal primers or synthetic 21-mer Oligonucleotides.

Expression of wild type (WT) and mutant E. coli BE. An overnight culture of
the transformed cells (BL21[DE3]) carrying the recombinant plasmid encoding the gene
for WT BE or the constructed mutants was diluted 1:20 (v/v) in fresh Luria Broth
medium containing 50 ug'ml‘I kanamycin. The cells were grown at 37°C to A600 = 0.5
and allowed to cool down to 25°C. Flasks were transferred to a shaker at room
temperature and T7 RNA polymerase expression induced by adding isopropyl-D-
thiogalactoside to 0.5 mM. After 2 hrs incubation endogenous expression was turned off
by adding rifampicin to 150 ug-ml". Incubation was continued for 16 hrs at room

temperature and cells were then harvested in a refrigerated centrifuge.

70

Purification of E. coli BE wild-type and mutants. The cell pastes obtained were
about 1.5 g per 1 l of culture. The purification of the E. coli BE WT and the mutants was
performed as described by Guan et al. (5). Protein concentration was measured with the
BCA protein assay reagent (26), using BSA as the standard.

Assay of BE activity. BE activity was measured by three different assays as
described by Guan and Preiss (8).

Assay A. The phosphorylase A stimulation assay is based on the stimulation by
BE of the synthesis of a-D—glucan from a-D-Glc-l-P catalyzed by rabbit phosphorylase a
(27). Reaction mixtures contained, in a final volume of 100 pl, 100 mM citrate (pH 7.0),
10 mM AMP, 0.2 mg phosphorylase a (Sigma Chemical Co.) and 50 mM D-[”C]-Glc-1-
P (50 DPM-nmol") and were initiated by adding an appropiate amount of BB. One unit is
defined as 1 umol of Glc incorporated into a-D-glucan per min at 30°C.

Assay B. The branching linkage assay determines the number of branching points
introduced by BB into the substrate, reduced amylose (28). Substrates were prepared by
the reduction of enzymatically synthesized amylose (AS-320, degree of polymerization
(d.p.) 1815) as described by Takeda et al. (28). Reaction volumes contained 25 mM
MOPS (pH 7.5) and an appropriate amount of substrate in a final volume of 100 pl. The
reaction was initiated by adding BE. One unit is defined as 1 umol of branching linkages
formed per min at 30°C.

Assay C. The iodine stain assay is based on measuring the decrease in absorbance
of the glucan-iodine complex resulting from the branching of the substrate, amylose
(potato type III [d.p. 1176, chain length (c.1.) 884]) or amylopectin (corn, Sigma

Chemical Co.) (6, 8). Reaction mixtures contained 50 mM citrate (pH 7.0) and 0.1 mg

71

substrate in a final volume of 500 pl. The reaction was initiated by the addition of an
appropriate amount of enzyme. One unit is defined as the decrease in absorbance of 1.0
per min at 30°C.

Analysis of the chain length distribution by high performance anion
exchange chromatography. Reduced amylose (5 mg, AS-320, d.p. 1815) was incubated
in 25 mM MOPS, pH 7.5 (500 1.11) at 30°C with BB (5 mU of E. coli BE WT and mutants
by assay b). After 1 and 4 hrs, the reaction in a 200 111 aliquot was terminated by boiling
for 2 min. The synthesized a-glucan was then debranched by adding 1 M acetate buffer
(pH 3.5, 20 ul) and isoamylase (10 ul, 1180 U/ul). After 90 min at 45°C the solution was
boiled for 5 min. The solution was filtered through a 0.22 pm membrane and 25 ul
samples injected in the HPAEC (BioLC, Dionex, Sunnyvale, CA), using a CarboPac PA-
1 column (250 - 4 mm). Eluent A was 150 mM soditun hydroxide; eluent B was 150 mM
sodium hydroxide containing 500 mM sodium acetate. The gradient program was: 25%
eluent B (75 % eluent A) at Time 0, 45% at 15 min, 60% at 45 min, 70% at 80 min, and
80% at 100 min, with a flow rate of 0.3 til-min". The standard was 25 pl of 5 mg-ul'l

maltodextrin (d.p. 1-20, Aldrich, WI).

72

Results
Identification of targets for mutagenesis. Figure 1 shows a partial sequence
alignment of several branching enzymes from a variety of bacteria and plants. Only the
sequences in the vicinity of the regions 1, 3 and 4, conserved in the amylolytic family as
defined by Baba et al. (12), are shown. The residue immediately following E458 (E. coli
BE numbering) in region 3 is a conserved acid in all branching enzymes, a fact first
pointed out by Jespersen et al. (13). For prokaryotes this acid is most frequently Glu,

while higher plants have Asp in this position. This residue, Glu-459, is located in the [35-

HS loop of the (ct/[3) barrel based on secondary structure prediction using the software
PHD (29). The mutants E459A, E459D, E459K, and E459Q were constructed as
described in materials and methods.
We also identified a conserved residue located immediately after the active site in region
4 (Figure 1). Position 527 in E. coli BB is conserved as a Glu in all bacterial branching
enzymes, whereas the equivalent position in plant and other eucaryotic enzymes is
conserved as Gln. Predictions show E527 maps to the center of the 137-H7 loop. The
mutants E527A, E527D, E527N, and E527Q were constructed in order to probe the role
of this amino acid. Finally, The Y300 residue of E. coli branching enzyme was targeted
due to its conservation in all amylolytic enzymes, where is has been proposed to play a
role in substrate binding (24). Y3 00 is located in the BZ-HZ loop of the central (oz/B)
barrel domain. The mutants constructed at this site were Y300A, Y300D, Y300F, Y300L,
Y3OOS, and Y300W.

Expression and purification of mutant enzymes. The BL21(DE3) strain used

for expression of the E. coli BE WT and mutants is glgB+. This poses the problem that

73

.202
E @88me Pa .mosEmE “533:3 :05 use Ahmmm can mmvm .896 ago 330589 :8 .m 05 E ammocowfisfi c8026
.8? 8m @8038 “2an 05:3 05. 682.895 o8 AND :3 B Bum .3 won—woe mm v :2on mo mm<-m_m Batman 05 can m .8me

@8528 :5 wotomcoo 23. .mEmENwS 32.5, Bob 3:358 wfinofifi 332% .«o «58:93 3:038 35m ._ "5&5

 

2mm :8 .m sea :8 .m 89 :8 .m
a a a
5 mooIMm mg. m N m zom>nw¢>H a o m 4 2:53 me o m N 89 H E82 82
Bacolmmmfi Sm zom>amo>3 mmm ZEN/much ma cum :5 a Eamon $385.94
>38? mm» m m e 2853 H e> n m e ze>mwwm mm m e N :9 = .882 Be:
Emommmmfi o S zom>nm§>e m3 ZEN/meow ma mmN AN: _§eoa 8a:
m>>mmmmmmq m H e zoemnmfizz w v N 309632 0 o N 89 3:838 358%

m>>mMmmmmq m N 4 so a mawfizn n v m oawouozmm m N N as 3385688; 323%
m>>mmMm H mg m m m smegma e> m w 4 659638 N m m 2.9 Eauaxssa assassin.
m>>mmmmqmq H Nm moemmmng v m 4 029833 e m N 3 :8 careeam

 

mm m
w GONmom m GOHmom

74

the recorded activity from the constructed mutants potentially could arise from the
endogenous E. coli BE. Therefore, we used rifampicin during the induction to inhibit
endogenous transcription. Rifampicin is a potent inhibitor of bacterial RNA polymerases
(3 0) but does not affect the T7 RNA polymerase that transcribes the recombinant
plasmid. An immunoblot with 0.2 ug crude extract of E . coli BE WT and the mutants
showed for each, one distinct band, while the negative control (BL21[DE3] carrying
plasmid pET23d) even when loaded with excess crude extract , i.e. 5 1.1g or even 15 ug,
showed no bands (data not shown). Furthermore, the negative control lacked any
detectable BE activity in the crude extract. These observations indicate rifampicin
blocked the expression of endogenous branching enzyme and the detected BE activity
arose solely from the expressed mutant plasmids. When compared to the original method
published by Guan et al. (5), this method increased the specific activity of the crude
extract three-fold due to less endogenous expression (data not illustrated).

The WT E. coli BE and the constructed mutants were purified to near
homogeneity to compare their properties. All the purified proteins showed one band of
about 85 kDa on 3 SDS PAGE gel when 1 pg was loaded on the gel. The specific activity
of the WT enzyme in the phosphorylase A assay was similar compared to the 1111 U-mg'

' obtained by Guan et al. (5) (Tables 1 and 3).

Replacement of Glu-459. Table 1 compares the specific activities of the E. coli
BE WT and the E459D, E459D, E459K, and E459Q mutants. Interestingly, the E459D
showed a higher specific activity when compared to the WT in all assays (185%, 160%,

145%, and 140% relative to wild-type activity in the phosphorylase A assay, branching

75

5330252 adage—>85 93 82.38 55 >83 :83 2562 06 88:3 NO 98 .0 a
.21 S we? omm-m< 0:533 mo 3388088 05 >38 owe—d: mafia—«.5 05 5 .

 

 

2 N _.N who ”no 65
3 EN 2; EN 3: 62
92 an «N 4.2 E: 6?.
08 88A NS $2 eNa m2
burrow mo 23%
a: 2 NN R 3 No 268223 5S,
2 3 : a: a 6 823a 5;?
0 beam 2:3
Ed adv _N.o M: 2 m humane: wezocem
02 8 N _ m Ne: 82 < s anofimoi
93m vane 4&me 99$ .5 Nae...

 

.U 98 m .< Emma mafia 35.808 3982: mmvm mo 82>an 25025 A 033.

76

linkage assay, iodine assay with amylose, and iodine assay with amylopectin,
respectively).

Table 1 includes the ratios of activity for the three different assays. Guan and
Preiss (8) showed that these ratios are useful for differentiation of the properties of
branching enzymes. The WT and E459D enzymes have essentially similar ratios of
activity, indicating that the properties of the E459D mutant are similar to the WT. The
other mutants show lower specific activities as compared to WT, the E459K being the
less active especially in the branching linkage assay. The C2/C1 ratio increases from a
0.58 in the WT to 2.1, 2.9, and 1.5 in E459A, E459K, and E459Q, respectively, meaning
these mutations altered the substrate preference from amylose to amylopectin.

The kinetic properties of E. coli BE had not been studied previously, and a full
characterization was needed for this study. Using reduced amylose AS-320 as substrate,
we investigated the kinetic parameters of E. coli BE WT and the E459D mutant. The
computed Km values displayed in Table 2 show a lower K.,,(11.7 11M) for the WT as
compared to that of the E459D mutant (23.5 11M). The specific activities (V max per mg of
protein) were 3.7 U/mg for the wild-type and 5.9 U/mg for the E459D. The estimated
specific activities as well as the high Km of E459D are approximate due to the low
solubility of amylose that does not allow the use of high concentrations of substrate.
Based on this fact, the indicated specific activity of the E459D mutant is a low estimate.

The Km and specific activities for the maize branching enzymes have been
published previously with assays using AS-320 as substrate (31). The reported Km values
were 59 uM for both mBEI and mBEII but due to a computational error in amount of

substrate, these numbers are off by a factor of 7. The corrected values for Km are 8.4 uM

77

for both mBEI and mBEII. The specific activity was 3.3 U/mg for mBEI and 0.62 U/mg
for mBEII. Therefore, the kinetic parameters of E. coli BE WT closely resembles the
mBEI. Fersht (32) defined the substrate specificity as Kw/ Km. This ratio is similar for

WT and E459D (4.5-105 s'1-M" and 3.5105 s"-M" respectively, Table 2).

Table 2. Kinetic parameters of WT E. coli and mutant E459D BEs.

 

 

WT E459D
Specific activity3
(umol x min'l x mg'l) 3.7d:0.2 5.9i0.8
Km (11M) 11.7i0.7 23.5:t5.6
L.,/K.,, (3'1 x M") 4.5 x 105 3.5 x 105

 

8 Activity was measured by the branching linkage assay (assay b) using reduced amylose
AS-320 as substrate.

Modification of the structure of reduced amylose AS-320 by BB was analyzed for
the WT and all mutants except E459K, due to its low activity. Figure 2 shows the
HPAEC pulsed amperiometric detector (PAD).response chromatograms of WT and
E459D after 4 hrs of branching as described in Materials and Methods. The mutant
enzymes E459A and E459Q yielded similar results to the WT (data not illustrated). The
only significant difference in the chromatograms in Figure 2 is presence of a d.p. 4 peak
in the chain length distribution of E459D which is not observed for WT enzyme.
Therefore the HPAEC chromatogram was rerun with 1.5 ug of maltotetraose included in
the sample to test if the retention time of the maltotetrose corresponded to that of the

observed peak. This increased the proposed d.p. 4 peak two-fold in size and we

78

A 7 P400

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

* l-
o. 3
4
o <‘r.’
< E
n. :1
'6
O
(D
l 1 J 4 l L_ 100
0 20 4O 60 80
7
B + {-400
........ .350 A
a) """"""""""" 2
co ........... E
C ..... -- V
2% -3°° a
s 1 ~250 ’3'
D _ , <
(
0- -200 g
- "'3
U U (j UU -150 w
l L l 1 1 .1m
0 20 40 60 ' 80

Retention time (min)

Figure 2. HPAEC analysis of the debranched a-glucan formed by action of E. coli

BE WT on reduced amylose AS-320 (A) and of E459D (B). Number on peak
indicates the corresponding c.l. The sodium acetate gradient is superposed on

chromato gram.

79

concluded the E459D does differ from WT in transferring this particular chain length. A
similar study of the E. coli BE WT was done recently by Guan et al. (5) were they
registered d.p. 4 and even d.p. 3, however without presenting the corresponding PAD
chromatograms. The relative peak area for all E459 mutants based on the PAD response
chromatograms were calculated and showed that there is little difference in the branching
products of the enzymes tested, the most commonly transferred chain length being 11
(data not shown).

As the focus of our study is a conserved acid in the branching enzymes, we
studied the activity over a pH range for the E. coli BE WT and E459A (Figure 3). The
effect of pH on E. coli BE activity had not been reported previously. Figure 3 indicates
that the activity profile of the WT and E459A are similar over the pH range tested with
an optimum near pH 7.7. The reported pH optimum for the maize branching enzyme
isoforms is 7.5 (28). If the Glu-459 was involved in acid-base catalysis, a substitution
would create a shift in the relative activity at a particular pH, but this is not the case.
Thus, a role of Glu459 as destabilizing the de-protonated form of the neighboring Glu-

45 8, proposed to be a catalytic residue, can be ruled out.

Replacement of Glu-527. Glu-527 was substituted with Ala, Asp, Asn, and Gln
as described above and the corresponding mutant enzymes purified to homogeneity. All
the E527 mutants showed lower activity than the wild-type enzyme varying from 10 %
(E527N) to 25 % (E527D) (Table 3). As for E459 we find that the Cz/Cl ratios change
significantly for E527N and E527Q, indicating a shift in the preference of substrate in

assay B from amylose to amylopectin. However, this did not affect the chain transfer

80

1.2- w

 

 

 

- 0.12
1.0-
E’ 0 8 0 ~ 0.10 8’
5 \- 5
_ ~ - 0.08 _
.3 0.6- 4 .3;
2 ._>.
g < a 0.06 "g
.2 0.4— o 4 .2
8 ‘ O a 0.04 g
a 0.2- /L « 3
‘0 O\O - 0.02
0.0— -—-—---- 4
1 ' l V l 1 1 1 1 ' t ' 1 ' 0-00
5.5 6.0 6.5 7.0 7.5 8.0 8.5 9.0

Figure 3. pH profile of E. coli BE WT (I) and E459A (0) mutant. Left Y-axis is
specific activity for WT and right Y-axis for E459A as defined by the branching

linkage assay (assay B).

81

3:330:32 £38338“ Ea owe—>85 53> >83 cam 25:: ofi mBE—ow NO was .0 e

.21 2 we? omm-m< 89:33 .«o 5383088 65 >83 ammo—d: wanSwfi 05 5 e
EN EN end SS 36 .96

 

 

Na 2: .1. 3: EN No).
a: wNN 3 we ed .02
see man :3 NS 82 m2
5338 .8 23M
92 E: 5.2 ”.2 N: No seaeaea
NN 3. Se N.mN m. a G 8255
0 5.58m 2563
owe a; £8 Ne 24 m news; waneeam
as N: SN ”2 8: < .e eaEeeemei
dNNmm ENE ENE fiNmm e3 5&4

 

.U 28 m .< Emma mam: 8582: 953:8 nmmm CO 83338 23025 .m 033.

82

pattern, as all the E527 mutants had chain transfer patterns identical to BE WT as
determined by HPAEC. Using the branching linkage assay the Km for E527A was
measured to 14.2 uM 21:1.7 compared to 11.7 uM i0.7 for the wild-type enzyme, and

therefore E527 is not likely to be involved in substrate binding.

Replacement of Tyr-300. Several mutants of Y3 00 were constructed and the
corresponding enzymes purified and analyzed with respect to the three assays available
for detecting branching activity (Table 4). The majority of the mutants had levels of
branching activities below the detection level of assays B and C. In particular, for mutant
Y3 00W the activity was at background levels even in the more sensitive assay A. Only
Y3OOF with 25 % of BB WT activity could be analyzed in further detail. The possible
involvement of Y300 in substrate binding was investigated through the kinetic
parameters of this mutant. The specific activities of BB WT and Y3OOF were 3.2 U/mg
and 0.8 U/mg, respectively. The computed Km value of 17.5 uM for Y3OOF was very
similar to the WT BE Km of 14.2 uM, indicating the Y300 is not directly involved in the
binding of substrate in branching enzyme.

From the crystal structure of isoamylase, is has been suggested the conserved Tyr may
form a hydrogen bond in the active site to the catalytic Asp in region 1 (24). The inability
to form such a hydrogen bond could result in lower heat stability and enzymatic activity
at elevated temperatures compared to the wild-type enzyme. Therefore, we measured the
heat stability of E. coli BE, which has not been studied previously. We measured the heat
stability of WT BE and Y3OOF by heating samples to temperatures between 30 and 65°C

for 10 min. Residual enzymatic activity was assayed by the phosphorylase a assay

83

038023 8: 6.:

.>_o>u882 avenue—>3 93 owe—>83 5?: >83 588 253m 2: 385% NO 23 .U a

.21 E 33 omm-m< 2338 we 59352.8 05 >83 owe—a: 32035 05 5 .

 

 

86 who .96
3: 2m 62
m2 ofi .02
- 8? 8% cm: m? a? 82 m2
338.6 88
.3 .3 .3 n: .3 3 in No 388:3
3 .3 .3 m2 3 .3 8.3 6 083a
0 93333
adv 8.9 8.? 3o 86v adv :._ m ..ow§5_w=3§m
:v t: S E 3 h: 8: < .a 330385
389 moo; 48? ”so? 959 <89 .3 33

 

.0 33 m .< >83 mama: 38308 $388 03$ «0 333:3 250on .v 2an

84

(Figure 4). These results show that the heat stability of Y300F is lowered by
approximately 5 degrees compared to the WT enzyme, indicating that the hydroxyl group
of the Tyr is important for stability of the enzyme. To analyze this further we measured
the relative activity of WT BE and Y3OOF at temperatures between 30 and 50°C using the
branching linkage assay, assay B (Figure 5). Y3OOF showed a significantly lower relative
activity at higher temperatures compared to WT. Both WT and Y3OOF were stable at
50°C for at least 30 min in the presence of substrate (data not shown). Combined, these
measurements demonstrate the importance of the hydroxyl group on the Tyr residue for

stability of the active site.

85

 

 

O\
o\° 80‘ I
\
2:
IE 60-
*5 I
(U
:5 40~
:5
20-
O
O‘F r I I I I I i I fi fi ' I\rT1

30 35 4o 45 50 55 60 65
0
Temperature/ C

Figure 4. Heat stability curve of BB WT (I) and Y3OOF (0). Samples were
heated to the indicated temperature for 10 min followed by cooling on ice for 5
min. The remaining activity was determined by the phosphorylase A stimulation

assay (assay A). The activity obtained at 30°C was set to 100 %.

86

 

 

100~o
XI
0
o\° 804
3?
5 so- -
(U
“>J
2.: 401
L“ o
63 .
20—
O\
O
O l ' l I r v T 1 I
30 35 4O 45 50

Temperature, °C

Figure 5. Relative activity of BE WT (I) and Y3OOF (O) at elevated
temperatures. Specific activities of BE WT and Y3OOF were determined at the

indicated temperature using the branching linkage assay (assay B). The activity

obtained at 30°C was set to 100%.

87

Discussion.

Based on the results presented here, we conclude that Glu-459, located in the
conserved region 3, is important for the function of the E. coli BE. When Glu-459 is
changed to other amino acids, a significant change in the properties of the enzyme is seen
with respect to specific activities, kinetic parameters and substrate specificity. The
conservative change from Glu to Asp even led to higher specific activities compared to
the WT in all three assays available. It is worth noting that all higher plant branching
enzymes (as well as a few from prokaryotes) have an Asp at this position. It would be
interesting to study the consequences of the reciprocal conservative substitution at this
position in a plant branching enzyme.

Glu-459 is not likely to be involved in chain transfer, as the HPAEC-PAD
chromatograms were virtually identical for all constructed mutants. It is also evident that
Glu-459 is not an essential catalytic residue, as some activity remained even when it was
substituted with an amino acid of opposite charge (E459K). However, Glu-459 could be
involved in catalysis of a step which is not rate limiting, based on our observations of
altered specific activity and substrate preference. The pH profiles of the WT and the
E459A mutant rules out the possibility that Glu-459 is involved in acid-base catalysis as
the pH profiles as the profiles can be superimposed.

The residues Y300 and E527 are both required for maximum branching activity,
although they are evidently not catalytic residues. Previously it has been suggested that
Y300 plays a role in substrate binding (24), but kinetic analysis of BB WT and Y3OOF
revealed similar Km values. These findings rule out any direct involvement of the

hydroxyl group of Y300 in binding of substrate. However, the Tyr could participate in

88

binding and stabilization of the substrate via hydrophobic stacking interactions with the
aromatic ring. Such stacking interactions between the substrate and the corresponding
Tyr residue have been reported in cyclodextrin glucanotransferase (33). We did see an
effect on heat stability of approximately 5°C and branching activity at elevated
temperatures when analyzing the Y3OOF mutant. These findings indicate the formation of
a hydrogen bond between Y300 and another residue in or near the active site as being
important for the enzyme stability.

A three-dimensional structure is required before we can assign specific roles for
the residues investigated here by site-directed mutagenesis, and in turn, the reason for

these positions being conserved in all known branching enzymes.

89

10.

11.

12.

13.

14.

15.

References.

. Preiss, J. (1984) Annu. Rev. Microbio. 38, 419-458.

Preiss, J. (1991) in Plant Molecular and Cell Biology (Mifflin, 3., Ed), vol 7, pp 59-

114, Oxford University Press, Oxford.

. Boyer, C., and Preiss, J. (1977) Biochem. 16, 3693-3699.

Baecker, P.A., Greenberg, E., Preiss, J. (1986) J. Biol. Chem. 261, 8738-8743.
Guan, H.P, Li, P., Imparl-Radosevich, J ., Preiss, J ., Keeling, P. (1997) Arch.
Biochem. Biophys. 342, 92-98.

Boyer, C., and Preiss, J. (1978) Carbohydr. Res. 61, 321-334.

Singh, B.K. and Preiss, J. (1985) Plant Physiol. 79, 34-40.

Guan, HP, and Preiss, J. (1993) Plant Physiol. 102, 1269-1273.

Nakamura, A., Haga, K., Ogawa, S., Kuwano, K., Kimura, K., Yamane, K. (1992)
FEBS Lett. 296, 37-40.

Mizuno, K., Kawasaki, M., Shimada, H., Satoh, H., Kobayashi, E., Okumura, S.,
Arai, Y., Baba, T. (1993) J. Biol. Chem. 268, 19084-19091.

Romeo, T., Kumar, A., Preiss, J. (1988) Gene 70, 363-376.

Baba, T., Kimura, K., Mizuno, K., Etoh, H., Ishida, Y., Shida, 0., Arai, Y., (1991)
Biochem. Biophys. Res. Commun. 181, 87-94.

Jespersen, H.M, Ann MacGregor, E., Henrissat, B., Sierks, M.R., Svensson, B. (1993)
J. Protein Chem. 12, 781-805.

Matsuura, Y., Kusunoki, M., Harada, W., Kakudo, M. (1984) J. Biochem. 95, 697-
702.

Buisson, G., Duee, E., Haser, R., Payan, F. (1987) EMBO J. 6, 3909-3916.

90

16.

17.

18.

19.

20.

21.

22.

23.

24.

25.

26.

27.

28.

29.

30.

Boel, E., Brady, L., Brozozowski, A.M., Derewenda, Z., Dodson, G.G., Jansen, V.J.,
Petersen, S.B., Swifi, H., Thim, L., Woldike, H.F. ( 1990) Biochem. 29, 6244-6249.
Klein, C., and Schulz, GE. (1991) J. Mol. Biol. 217, 737-750.

Kubota, M., Matsuura, Y., Sakai, S., Katsube, Y. (1991) Denpun Kagaku 38, 141-
146.

Kuriki, T., Takata, H., Okada, S., Imanaka, T. (1991) J. Bacteriol. 173, 6147-6152.
Takata, H., Kuriki, T., Okada, S.,Takesada, Y., Iizuka, M., Minamiura, N., Imanaka,
T. (1992) J. Biol. Chem. 267, 18447-18452.

Kuriki, T., Guan, H.P., Sivak, M., Preiss, J. (1996) J. Protein Chem. 15, 305-313.
Funane, K., Libessart, N., Stewart, D., Michishita, T., Preiss, J. (1998). J. Prot. Chem.
17, 579-590.

Libessart, N., Preiss, J. (1998). Arch. Biochem. Biophys. 360, 135-141.

Katsuya, Y., Mezaki, Y., Kubota, M., Matsuura, Y. (1998). J. Mol. Biol. 281, 885-
897.

Sanger, F ., Nicklen, S., Coulson. (1977) Proc. Natl. Acad. Sci. USA 74, 5463-5467.
Smith, P.K., Krohn, R.I., Hermanson, G.T., Mallia, A.K., Gartner, F .H., Provenzano,
M.D., Fujimoto E.K., Goeke, N.M., Olson, B.J., Klenk, DC. (1985) Anal. Biochem.
150, 76-85.

Hawker, J .S, Ozbun, J .L., Ozaki, H., Greenberg, E., Preiss, J. (1974) Arch. Biochem.
Biophys. 160, 530-551.

Takeda, Y., Guan, H.P., Preiss, J. (1993) Carbohydr. Res. 240, 253-263.

Burkhard, R., Sander, C. (1993). J. Mol. Biol. 232, 584-599.

Fromageot, HP, and Zinder, ND. (1968) Prod. Natl. Acad. Sci. 61, 184-191.

91

31. Kuriki, T., Stewart, D.C., Preiss, J. (1997) J. Biol. Chem. 272, 28999-29004.

32. F ersht, AR. (1985) in Enzyme structure and Mechanism, Freeman, New York.

33. Uitdehaag, J. C. M., Mosi, R., Kalk, K. H., Van der Veen, B. A., Dijkhuizen, L.,
Withers, S. G., Dijkstra, B. W. (1999) Nature Struc. Biol. 5, 432-436.

34. Homerova, D., and Kormanec, J. (1994) Biochem. Biophys. Acta 1200, 334-336.

35. Takata, H., Takaha, T., Kuriki, T., Okada, S., Takagi, M., Imanaka, T. (1994) Appl.

Envir. Microbio. 60, 3096-3104.

36. Kiel, J. A., Boels, J. M., Beldman, G. and Venema, G. (1992) DNA seq. 3, 221-232.

37. Fisher, D.K., Gao, M., Kim, K.N., Boyer, C.D., Guiltinan, M.J. (1996) Plant Mol.
Bio. 30, 97-108.

38. Fisher, D.K., Boyer, C.D., Hannah, LC. (1993) Plant Physiol. 102, 1045-1046.

39. Mizuno, K., Kimura, K., Arai, Y., Kawasaki, M., Shimada, H., Baba, T. (1992) J.

Biochem. 112, 643-651.

92

CHAPTER 3
SLOW-BINDING INHIBITION OF BRANCHING ENZYME

BY THE PSEUDOOLIGOSACCHARIDE BAY-e4609.

This data has been published.
Binderup, K., Libessart N., and Preiss, J. (2000) “BAYe4609, a Slow-Binding Inhibitor

of Branching Enzyme. Functional Evidence for Relationship with Amylolytic Enzymes.”

Arch. Biochem Biophys. 374, 73-78.

93

Abstract.

Branching enzyme from E. coli is shown to be inhibited by the pseudo-
oligosaccharide BAY 64609. The mechanism of binding is studied in detail by kinetics
using reduced amylose as substrate. Lineweaver-Burk plots suggest the mechanism of a
non-competitive or slow-binding inhibitor. Further studies by progress curves and rate of
loss of branching activity allows us to conclude BAY e4609 as being a slow-binding
inhibitor of branching enzyme. We discuss how these results parallel the inhibition of a-
amylase by acarbose and the significance for branching enzyme as belonging to the

amylolytic family.

94

Introduction.

Branching enzyme catalyzes the formation of a-l ,6-glucosidic linkages of
glycogen in bacteria (1) and starch in plants (2-4). Hence, the enzyme plays an important
role in the biosynthesis of starch and glycogen (2-4). Many starch/ glycogen hydrolases
and related enzymes have been assigned to the a-amylase family as they contain a
characteristic (on/Bk barrel domain (5-7). It has been suggested that branching enzymes
belong to the family of amylolytic enzymes (5-7) by virtue of containing in the central
portion of the enzyme, four conserved regions also present in other amylolytic enzymes,
amylase, pullulanase, isoamylase, and cyclodextrin transferase. By site-directed
mutagenesis of branching enzymes it was suggested that in addition to structural
similarities of the amylolytic enzymes, the catalytic mechanisms among them are shared
(8). Moreover, the variations in substrate specificities and products can be ascribed to the
relationship between their catalytic centers and different subsite structures (9, 10).

Several inhibitors of a-amylases have been reported including tendamistate/ Hoe
467 (11), trestatins (12), and acarbose (13, 14). Investigation of carbohydrate-splitting
enzyme inhibitors have largely been driven through medical interest in the treatment of
metabolic diseases such as diabetes (For a review of inhibitors of a-glucosidases see
Truscheit et al. (15)). In particular, the pseudo-tetrasaccharide acarbose has been found to
inhibit CGTases as well as a-amylases (15), and recently X-ray structures of porcine
pancreatic a-amylase and Bacillus circulans CGTase both with bound acarbose have
been solved (16, 17). In both cases acarbose was found at the catalytic site of the enzyme.
Whereas Qian and co-workers proposed acarbose to be a transition state analog (16),

Strokopytov found it more likely to be a substrate analog (17) even though the mode of

95

binding of the inhibitor in the two structures was very similar. Despite these differences,

the interpretation of the roles of amino acids surrounding the catalytic cleft including

identifying residues involved in substrate binding and direct catalysis was very similar.
To date no published study has reported any inhibitors of BB. The present study

reports the identification of an acarbose-like inhibitor of BB, BAY e4609 (Figure 1).

+1on CH20H CH3 cuzort
o H
H
OH m OH on OH n

 

 

acarbose: m=0, n=2 - BAY e4609: m+n= [7-30] ,

 

Figure 1: Structures of the pseudo-oligosaccharides acarbose and BAY e4609. m
and n designates the number of glucose units at the non-reducing and reducing end,
respectively. Acarbose is a pseudo-maltotetrose, whereas BAY e4609 is a pseudo-

oligosaccharide of chain length ranging from 9 to 32.

BAY e4609 is a pseudo-oligosaccharide resembling amylose with a
hydroxymethylconduritol unit and a 4-amino-4-deoxy-D-chinovose residue linked to a
varying number of a-D-glucose units from 7 to 30. It is a secondary metabolite purified
from Actinoplanes strain 50/13 (1 3) and has been shown to be an even more potent
inhibitor of pig pancreatic oc-amylase than acarbose (15). We investigate the mode of
binding of the inhibitor to BE and discuss how our results strengthen the hypothesis of

BE belonging to the amylolytic family beyond sequence conservation.

96

Materials and Methods.

Expression and purification of E. coli BE. An overnight culture of the
transformed cells (BL21[DE3]) carrying the recombinant plasmid encoding the gene for
BB was diluted 1:20 (v/v) in fresh Luria broth medium containing 50 pg ml'l kanamycin.
The cells were grown at 37°C to A600 = 0.5 and allowed to cool down to 25°C. Flasks
were transferred to shaker at room temperature and T7 RNA polymerase expression
induced by adding isopropyl-D-thiogalactoside to 0.5 mM. Afier 2 hrs incubation
endogenous expression was turned off by adding rifampicin to 150 ug ml'l as described
by Binderup and Preiss (18). Incubation was continued for 16 hrs at room temperature
and cells were then harvested in a refrigerated centrifuge. The purification of the E. coli
BE was performed as described by Guan et al. (19) with some modifications: The cell
paste (about 1.5 g per 1 l of culture) was resuspended in 10 ml buffer A (50 mM Tris-
acetate pH 7.5; 10 mM ethylenediaminetetracetic acid [EDTA]; 2.5 mM dithiothreitol
[DTT] ) and lysed by sonication. The homogenate was centrifuged at 10000 g for 15 min
at 4° followed by precipitation with 1.6 M ammonium sulfate by addition of a 3.8 M
stock solution. After centrifugation the pellet was resuspended in 10 ml buffer A and
dialyzed overnight in buffer A. The sample was then loaded on MonoQ HR 10/10
column and eluted with 0-0.3 M KCl in buffer A. Fractions containing branching enzyme
activity were pooled and BE precipitated with 1.6 M ammonium sulfate. The precipitate
was resuspended in 10 ml buffer B (10 mM triethanolamine pH 8.0; 1 mM EDTA; 2.5
mM DTT) and dialyzed overnight in buffer B. The sample was loaded on a MonoQ HR

5/5 column and eluted with O-0.4 M KCl in buffer B. Fractions were pooled and

97

concentrated to 2 mg ml'l. Protein concentration was measured with the bicinchoninic
acid protein assay reagent (20), using bovine serum albumin as the standard.

Assay of BE activity. BE activity was measured using the phosphorylase A
stimulation assay (assay A) and the branching linkage assay (assay B) as described
previously (21).

Assay A. The phosphorylase A stimulation assay is based on the stimulation by
BE of the rate of synthesis of a-D-glucan from a-D-Glucose-l-P catalyzed by rabbit
phosphorylase a (22). Reaction mixtures contained in a final volume of 100 pl, 100 mM
citrate (pH 7.0), 10 mM adenosine monophosphate, 0.2 mg phosphorylase a (Sigma
Chemical Co.) 50 mM D-[14C]-Glucose-l-P (so DPM nmol"), and various amounts of
inhibitor. The reaction was initiated by adding an appropiate amount of BB. One unit is
defined as 1 umol of glucose incorporated into a-D-glucan per min. at 30°C.

Assay B. The branching linkage assay determines the number of branching points
introduced by BB into the substrate, reduced amylose (23). Substrates were prepared by
the reduction of enzymatically synthesized amylose (AS-320, degree of polymerization
1815) as described (23). Reaction mixtures contained 25 mM MOPS (pH 7.5) and an
appropriate amount of substrate (35 uM unless stated otherwise) and inhibitor in a final
volume of 100 pl. The reaction was initiated by adding an appropiate amount of BB.
Debranching was performed by using isoamylase from Pseudomonas amyloderamosa
(Hayashibara Biochemical Laboratories, Japan). One unit is defined as 1 mo] of
branching linkages formed per min. at 30°C.

Gel filtration. A column (1.6 x 55 cm) was packed with Fractogel TSK HW-SOF

resin (EM Science) and equilibrated with H20. BAY e4609 (15 mg) was applied on the

98

column and 2-ml fractions were collected at a flow rate of 20 ml/hr. The average degree
of polymerization (d.p.) was determined by the ratio of total carbohydrate to reducing
sugar, as measured by the phenol-H2804 assay (24) and Park-Johnson method (25),

respectively.

99

Results.

Activity of E. coli BE in presence of inhibitor was initially performed by the
phosphorylase a stimulation assay (Figure 2). The phosphorylase a incorporates glucose
units into the nonreducing ends of a glucose polymer. In contrast to acarbose, BAY
e4609 has a nonreducing end (Figure l) which will serve as substrate for phosphorylase
a, so all experiments were performed with controls lacking BE. Figure 2 shows how
residual activity was reduced to less than 10% after 30 min incubation when using a BAY
e4609 concentration of 250 ug ml"l (BAY e4609 covers a range of sizes corresponding to
a molecular weight (MW) of 1500 to 5000, for which reason the concentration of BAY
e4609 is designated in pg ml").

The high sensitivity and ease of the phosphorylase a assay makes it ideal for a
qualitative estimate of branching activity, however it is not suitable for kinetic studies as
substrate concentration cannot be varied. In contrast, the branching linkage assay allows
for detailed kinetic studies but is complicated by the fact that it utilizes the enzyme
isoamylase [E.C. 3.2.1.68] (a a-[1,6] glucosidic-linkage hydrolyzing enzyme) which is
homologous to BE (6, 7). The lower branching activity when using inhibitor could be due
in part to BAY e4609 interaction with isoamylase. Consequently, the activity of
isoamylase in presence of BAY e4609 was measured using glycogen as substrate and
found to be unaffected (data not shown).

A Lineweaver-Burk plot was used to identify the type of inhibition. Figure 3
shows the E. coli BE in presence of various concentrations of BAY e4609. When fitted to
linear equations, the lines appear to intersect on the X-axis, indicative of a non-

competitive type of inhibitor, for which the inhibitor does not affect the binding of

100

 

 

 

 

 

 

 

 

100 7' 1 I .—

80- A

o
g. 60‘ O O

3
a o J
20- U C]
01 X X
0 20 40 60 80
'I'Ime, min.

Figure 2: Inhibition of E. coli BE by BAY e4609. (I) no inhibitor; (0) 2.5
ug/ml; (O) 10 ug/ml (O) 25 pig/ml; (El) 100 ug/ml; (x) 250 ug/ml. Activity was
measured by phosphorylase a stimulation assay. The activity of BB without

inhibitor was set as 100 % activity for each time point. 1 unit (U) by assay A of

BE was used. T = 30 °C, pH = 7.0.

101

substrate to enzyme, but reduces the Vmax. The reaction pathway for such a system can be

written as in equation 1.

k +13 k
E ="= ES —-—> E + products

*4
k +3illk -3 k -51' k +5i (1)
+4s

EI ~—— EIS
k -4

La.-

Two distinct types of non-competitive inhibitors exit depending on the fate of the
complex EIS. For a fully non-competitive inhibitor, EIS does not break down and the
effect is equivalent to a reduction in the amount of active enzyme. In case of a partially
non-competitive inhibitor EIS breaks down at a velocity different of BS and the rate of
reaction is the sum of the two.

However, an irreversible inhibitor may appear as a noncompetitive inhibitor in a
Lineweaver-Burk plot. E1 is a complex which is fully active and rapidly reversible during
the reaction, whereas EI" is the inactive BE-inhibitor complex which is very slowly
reversible (k.2<< k2). A true irreversible inhibitor has a k.2 value of zero (for example,
when a covalent E-I bond is formed) whereas a slow-binding inhibitor is defined as
having a slow release rate as seen by a low k.2 value (26). Thus, the differentiation of an

inhibitor as being irreversible or slow-binding is merely a matter of degree.

 

*1 k2
E+I *— EI _“--—— EI“ (2)
k_1 k-)

102

3.5-

3.01

1
2.51

l
2.04

1N 1.5) I
1.0-

0.54 g
0.0 . 1+1 a r . r

' I T r 1
-0.05 0.00 0.05 0.10 0.15 0.20 0.25 0.30
1/[81

 

 

 

Figure 3: Lineweaver—Burk plot of E. coli BE in presence of BAY e4609. (I) no
inhibitor; (0) 25 ug/ml; (C) 50 ug/ml. Activity was measured by branching
linkage assay (assay B) using reduced amylose AS-320 as substrate. Reactions
were started with 1 mU of BB (assay B) diluted into substrate containing

inhibitor. Product was measured after 30 min incubation. T = 30 °C, pH = 7.5.

103

As analysis of primary plots does not allow differentiation between these models,
additional information was sought by monitoring the rate of loss of activity.
The rate of BAY e4609 inactivation of E. coli BE was investigated by incubating BE
with excess inhibitor at 30 °C for up to 5 min. At various times, aliquots were removed
into tubes containing reduced amylose AS-320 without added inhibitor and branching
activity measured by assay B for 30 min as described in materials and methods. Figure 4
shows the rate of loss of branching activity being of first order. This suggests dissociation
of the inactive BE-inhibitor complex is slow, fitting reaction pathway 2 (27, 28).
Continuous kinetic assays using BAY e4609 were done by pre-incubating BE
with inhibitor on ice and starting the reaction with substrate. A classical reversible
competitive inhibitor in this type of experiment will show linear product formation as the
initial enzyme-inhibitor complex readily will establish a dynamic equilibrium. In
contrast, Figure 5 shows an initial velocity less than the steady state velocity with a
significant lag period characteristic of a slow-binding inhibitor (26). As shown in Figure
5, the progress curve for BB in absence of inhibitor is linear as expected, but clearly the
initial inhibitor-enzyme complex dissociates slowly as the steady state velocity is not
reached until about 30 min of incubation with 35 uM substrate. This also points towards
reaction pathway 2 being the most plausible mechanism of BAY e4609 and furthermore
indicates that it is a slow-binding inhibitor rather than irreversible (k.2>0). This was
confirmed by the observation that E. coli BE does not bind BAY e4609 covalently as
determined by the lack of a gel-shift when enzyme pre-incubated with inhibitor is run on

a SDS-PAGE gel under denaturing conditions (data not shown).

104

log loss of activity

 

 

 

I ‘ I ' I

2 3 4 5
time, min.

0
4'1

Figure 4: Rate of loss of branching enzyme activity. 0.5 mU of E. coli BE was
pre-incubated with a final concentration of 50 ug/ml of BAY e4609 at 30 °C. At
various times, aliquots were taken and diluted into 35 uM substrate lacking
inhibitor. The activity of BB without inhibitor is defined as 100 %. Activity was
measured by the branching linkage assay after 30 min incubation. T = 30 °C, pH
= 7.5. There was no loss of activity during incubation of the BE in the absence of

inhibitor.

105

60-
50-
40-
l

30- I

20-

M

Product, nmol reducing sugars

107 /=
1 , EQH/s
o 1": " , . , . , . , . .
o 10 20 30 4o 50 60
time, min.

 

 

Figure 5: Preincubation of E. coli BE with BAY e4609. (I) no inhibitor; (0) 25

ug/ml; (D) 37.5 ug/ml; (C) 50 ug/ml. 0.5 mU (assay B) of E.coli BE was

incubated for 90 min on ice with inhibitor and diluted into the substrate amylose
AS-320 lacking additional inhibitor. Branching activity was measured by assay B

after 0 to 60 min incubation at 30 °C. pH = 7.5, [AS-320] = 35 uM.

106

Progress curves for the inhibition of BB by BAY e4609 was performed by starting
the reaction with enzyme diluted into the substrate containing inhibitor (Figure 6). In the
presence of a classical rapid reversible inhibitor the reaction rates will remain linear
during the period where there is no substrate depletion. For a slow-binding inhibitor there
will be an initial burst of reaction followed by a lower steady-state rate, reflecting the
slow establishment of equilibrium between enzyme and inhibitor. The curves presented in
figure 6 show in presence of inhibitor an initial rapid formation of product, which afler a
few minutes drops to a much reduced rate. The rate of reaction in absence of inhibitor is
linear throughout the time of incubation. In conclusion, the data presented in this report is

consistent with BAY e4609 being a slow binding inhibitor of E. coli BE.

Fractionation of BAY e4609. In order to determine if a subset of the BAY e4609
chains enriched in inhibitory activity could be isolated, we performed fractionation of the
oligosaccharide mixture by gel filtration (Figure 7). Fractions 29-35 (pool 1), 36-39 (pool
2) and 40-45 (pool 3) were combined and concentrated as described under materials and
methods. The average oligosaccharide size was determined to be d.p. 33 for pool 1, d.p.
17 for pool 2, and d.p. 10 for pool 3, indicating fractionation according to chain size of
BAY e4609. Inhibition experiments using assay B with a constant amount of each of the
three pooled inhibitor fractions showed that all chain sizes of BAY e4609 inhibited
branching enzyme. The population of average d.p. 33 inhibited branching enzyme 1.4 and
3.4 times more efficiently than the average d.p. 17 and d.p. 10 populations, respectively
(data not shown). This suggests that whereas all chain sizes of BAY e4609 inhibit

branching enzyme, the longer chains constitute a more active portion of BAY e4609.

107

35-
30-
25-

20- I

0

15-

fi
“

\\

nmol reducing sugar

10-

 

 

d
d
_
d
d
d

.1
-

time, min

Figure 6: Progress curve of E. coli BE with BAY e4609. (I) no inhibitor; (0)
0.5 ug/ml, (A) 1.5 mg/ml. Concentration of substrate reduced amylose AS-320
was 10 uM. Reaction was started with 2.5 mU of BE and branching activity

measured by the branching linkage assay (assay B) at 30 °C. pH = 7.5.

108

 

3858
r
r
r

”9"”

total carbohydrate,
' o
'H.\_
o/O/O
/./.
a a;

 

 

 

Figure 7. Gel filtration of BAY e4609. The X-axis shows the fraction number.

Total carbohydrate (I) was measured by the phenol-sulfuric acid method. The
total carbohydrate scale (pg/ml) is shown on the left Y-axis. hm“ values of the

iodine-oligosaccharide complex (0) is shown on the right Y-axis. The degree of

polymerization (d.p., 9K) was determined as described under materials and

methods.

109

Discussion.

We have investigated the mechanism of inhibition of E. coli BE by the pseudo-
oligosaccharide BAY e4609 and found it to belong to the class of slow-binding
inhibitors. The nitrogenous sugars acarbose and deoxynojirimycin had no inhibitory
effect on branching enzyme, although both are potent inhibitors of other enzymes of the
a-amylase family. This can be thought to be due to their smaller size relative to BAY
64609, as the minimum chain length required for E. coli BE branching activity has been
found to be of 12 glucose residues (19). BAY e4609 is a mixture of oligosaccharides of
ranging size, and the efficiency of inhibition was found to vary according to the number
of glucose units present in the inhibitor, the longer chains being more effective. In
addition, the relative position of the hydroxymethylconduritol and 4-amino-4-deoxy-D-
chinovose residues in a BAY e4609 molecule of a given chain length might be essential
for effective inhibition. We have also seen the inhibition of maize branching enzyme 11
by BAY e4609, although it is inhibited to a lesser extend than BE from E. coli (data not
shown). The analysis was limited to construction of a Lineweaver-Burk plot, which
appeared similar to that of E. coli BE with BAY e4609. However, it leaves the exact
mode of binding undetermined. It would be interesting to characterize various BE’s with
respect to BAY e4609 and acarbose to see if relationship with inhibition and enzymatic
properties (e. g. substrate preference, chain length transfer) would emerge.

The BB belongs to the family of amylolytic enzymes, which all cleave and/ or
transfer chains of a-(1,4)-glucan. This relationship has been based on sequence
conservation in the central part of the BE, thought to fold as an (oz/Bk barrel as well as

site-directed mutagenesis studies of the proposed active site residues. Interestingly,

110

Wilcox and Whitaker (29) investigated the mechanism of inhibition of bovine pancreatic
a-amylase by acarbose and concluded it to be slow, tight-binding. Our observation that
the mechanism of inhibition of BB by BAY e4609 is similar to that of a-amylase by
acarbose thus provides direct functional evidence for BE being an amylolytic enzyme.
Furthermore, it suggests that one or more steps in the reaction path could be shared
between a-amylase and BE, as the crystal structure of porcine pancreatic a-amylase
complexed with acarbose concluded acarbose to be a transition state analog (16).
Interestingly, we found no inhibitory effect of Pseudomonas amyloderamosa isoamylase
by BAY e4609. This is significant, as isoamylase has been proposed to belong to the
family of amylolytic enzymes based on sequence comparison (6, 7). In effect, acarbose/
BAY e4609 can be speculated to be useful in further differentiating the groups of

enzymes considered to be in the a-amylase family.

111

References.
l. Preiss, J. (1984). Annu. Rev. Microbio. 38, 419-458.
2. Preiss, J. (1991). in Plant Molecular and Cell Biology (Mifflin, 3., Ed), vol 7, pp 59-
114, Oxford University Press, Oxford.
3. Preiss, J. and Sivak, M. N. (1986). in Genetic engineering (Stelow, J. K., Ed.), Vol. 20,
pp 177-223, Plenum, New York.
4. Preiss, J. and Sivak, M. N. (1996). in Photoassimilate Distribution in Plants and Crops:
Source-Sink Relationships (Zamski, E. and Shaffer, A. A., Eds), pp 63-96, Dekker, New
York.
5. Baba, T., Kimura, K., Mizuno, K., Etoh, H., Ishida, Y., Shida, O. and Arai, Y., (1991).
Biochem. Biophys. Res. Commun. 181, 87-94.
6. Jespersen, H. M., MacGregor, A. E., Svennson, B. ( 1991). Biochemistry 280, 51-55
7. Jespersen, H. M, MacGregor, A. E., Henrissat, B., Sierks, M. R., Svensson, B. (1993).
J. Protein Chem. 12, 781-805.
8. Kuriki, T., Takata, H., Okada, S. and Imanaka, T. (1991). J. Bacteriol. 173, 6147-6152.
9. Nakamura, Y., Takeichi, T., Kawaguchi, K., and Yamanouchi, H. (1992). Plant
Physiol. 84, 329-335.
10. Holm, L., Koivula, A. K., Lehtovaara, P. M., Hemminki, A., and Knowles, J. K. C.
(1990). Protein Eng. 3, 181-191.
11. Meyer, B. H., Muller, F. O. and Clur, B. K. (1983) J. Clin. Pharmac. 16, 145-148.
12. Mayo, J. W. and Carlson, D. M. (1974). Arch. Biochem. Biophys. 163, 498.
13. F rommer, W., Puls, W., Schafer, D. and Schmidt, D. D. (1970). German priority,

December 28, BAYER AG.

112

14. Puls, W., Keup, U, Krause, H. P., Muller, L., Schmidt, D. D., Thomas, G. and
Truscheit, E. (1980). Front. Horm. Res. 7, 235.

15. Truscheit, E., F rommer, W., Junge, B., Muller, L, Scmidt, D. D. and Wingender, W.
(1981). Angew. Chem. Int. Ed. Eng]. 20, 744-761.

16. Qian, M., Haser, R., Buisson, G., Duee, E. and Payan, F. (1994). Biochemistry 33,
6284-6294.

17. Strok0pytov, B., Penninga, D., Rozeboom, H. J., Kalk, K. H., Dijkhuizen, L. and
Dijkstra, B. W. (1995). Biochemistry 34, 2234-2240.

18. Binderup, K. and Preiss, J. (1998). Biochemistry 37, 9033-9037.

19. Guan, H. P., Li, P., Imparl-Radosevich, J ., Preiss, J. and Keeling, P. (1997). Arch.
Biochem. Biophys. 342. 92-98.

20. Smith, P.K., Krohn, R.I., Hermanson, G.T., Mallia, A.K., Gartner, F.H., Provenzano,
M.D., Fujimoto E.K., Goeke, N.M., Olson, B.J. and Klenk, DC. (1985). Anal. Biochem.
150, 76-85.

21. Guan, HP. and Preiss, J. (1993). Plant Physiol. 102, 1269-1273.

22. Hawker, J .S, Ozbun, J .L., Ozaki, H., Greenberg, E. and Preiss, J. (1974). Arch.
Biochem. Biophys. 160, 530-551.

23. Takeda, Y., Guan, HP. and Preiss, J. (1993). Carbohydr. Res. 240, 253-263.

24. Dubois, M., Gilles, K. A., Hamilton, J. K., Rebers, P. A., and Smith, F. (1956). Anal.
Chem. 28, 350.

25. Park, J. T., and Johnson, M. J. (1949). J. Biol. Chem. 181, 149.

26. Morrison, J. F. and Walsh C. T. (1988). Adv. Enzymol. 62, 201-301.

113

27. Uehara, Y., Tonomure, B. and Hiromi, K. (1980). Arch. Biochem. Biophys. 202, 250-
258.

28. Gutfreund, H. (1972). in Enzymes: Physical Principles pp. 20. Wiley-Interscience,
New York.

29. Wilcox, E. R. and Whitaker, J. R. (1994). Biochemistry 23, 1783-1791.

114

CHAPTER 4
LIMITED PROTEOLYSIS OF ESCHERICIA COLI BRANCHING ENZYME.

Part of this data has been published:
Binderup, K., Mikkelsen, R., and Preiss, J. (2000) “ Limited Proteolysis of Branching

Enzyme from Escherichia coli”. Arch. Biochem. Biophys. 377, 366-371.

115

Abstract

Branching enzyme is involved in determining the structure of starch and glycogen. It
catalyzes the formation of branch points by cleavage and transfer of a-(l ,4) glucan chains
to a-(1,6) branch points. Limited proteolysis of the branching enzyme from E. coli (84
kDa) by proteinase K produced a truncated protein of 70 kDa, which still retained 40-60
% of branching activity, depending on the type of assay used. Amino acid sequencing
showed that the 70 kDa protein lacked 111 or 113 residues at the amino terminal, whereas
the carboxy terminal was still intact. We purified this truncated enzyme to homogeneity
and analyzed its properties. The enzyme had a 3- to 4-fold lower catalytic efficiency
compared to the native enzyme, whereas the substrate specificity was unaltered.
Furthermore, a branching enzyme with 112 residues deleted at the amino terminal was
constructed by recombinant technology and found to have properties identical to the
proteolyzed enzyme. We have obtained high-resolution diffiacting crystals of this

truncated branching enzyme form and have collected a complete native data set to 2.3 A.

116

Introduction

Branching enzyme (l,4-a-D-glucan:1,4-oc-D-glucan 6-a-D-(l ,4-a-D-glucano)-
transferase; EC 2.4.1.18) catalyzes the cleavage of an a-1,4 glucosidic linkage and the
subsequent transfer of ot-l,4-glucan to a-1,6 branch points. The enzyme hence plays an
important role in determining the structure of starch and glycogen (1). Several bacterial
branching enzymes have been identified and characterized (2). Also, multiple isoforms of
the branching enzyme encoded by distinct genes have been reported for many plants,
including maize, rice and spinach (3-5). The investigation of the mode of chain transfer
and substrate specificity in the maize endosperm isoforms BE I (mBEI) and BE 11
(mBEII) led to the hypothesis that these enzymes have different roles in the synthesis of
amylopectin in vivo (6,7). The mBEI can transfer long branches (d.p. 40 to 100) whereas
mBEII transfers shorter chains (d.p. 6 to 14). Furthermore, the preferred substrate for
mBEI is amylose, whereas mBEH prefers amylopectin. The differences between mBEI
and mBEH isoforms suggest that mBEI first branches the amylose transferring long
chains and mBEII, in turn, uses this product for further branching to amylopectin.

BE has been shown to belong to the amylolytic family of enzymes by amino acid
sequence alignments and site-directed mutagenesis experiments. The amylolytic family,
consisting of enzymes that cleave and/or transfer chains of glucan, contains a central
(a/B)3-barrel as observed in on-amylase and cyclodextrin glucano transferase (8).
Branching enzymes have variable extensions at the amino- and carboxy termini relative to
the conserved (Mme-barrel region with little sequence conservation across sources. Our
knowledge of the function and significance of these domains is limited. A study using

hybrid enzyme construction of two maize branching enzyme isoforms produced a

117

chimeric enzyme with the N-terminal and central (010)8-barrel of mBEH and the C-
terminal domain of mBEI. This approach produced an active enzyme with the chain
transfer pattern of mBEII and the substrate specificity of mBEI (9), suggesting that the C-
terrninal domain is involved in substrate specificity. Another chimeric enzyme
construction with the mBEI sequence at the N-terminal and the central (or/[3)3-barrel plus
the C-terminal domain of mBEH gave an active enzyme having an alteration in size of
oligosaccharide chain transfer (9). Thus it was suggested that the N-terminal and/ or
central a/B-barrel region is involved in determining the size of chain transferred.

Here, we use the technique of limited proteolysis in order to study the role of the
amino- and carboxy-terminal of E. coli branching enzyme. The major finding of this
investigation was a branching enzyme truncated at the amino terminal with altered
activity. Furthermore, crystallization experiments of the truncated E. coli branching
enzyme formed single well-diffracting crystals which is allowing us to pursue the first

three-dimensional structure of a branching enzyme.

118

Materials and methods

E. coli BE expression vector. H. P. Guan kindly provided the plasmid pEXSB.
The pEXSB plasmid is a pET-23d (Novagen) derived expression vector containing the
origin plSA, T7 promotor, T7 terminator, kanamycin resistance, and the E. coli glgB
gene encoding the E. coli BE (10).

Limited proteolysis of branching enzyme. All proteases were obtained from
Boehinger Mannheim except carboxypeptidase Y from Pierce. Purified branching enzyme
(0.5 mg/ml) was digested with the various proteases under the following conditions: 25
°C, 1 mM CaClz and 50 mM Tris-Cl pH 7.5 for proteinase K, chymotiypsin, trypsin, and
subtilisin. Papain: 25 °C, 1 mM CaClz and 50 mM Na citrate pH 6.5. Protease V8 : 25 °C,
1 mM CaClz and K-PO4 or NH4HC03 (50 mM, pH 7.8). Elastase: 25 °C, 1 mM CaClz
and 50 mM Tris-Cl pH 8.0. Carboxypeptidase Y: 37 °C, 1 mM CaClz and 50 mM Na
citrate pH 6.5. Time of incubation and concentration of the proteases were varied as
indicated in each experiment. Proteolysis was stopped by adding
phenylrnethylsulfonylfluoride to 0.5 mM and aliquots assayed for activity.

Purification of proteinase K treated branching enzyme Purified E. coli BE was
digested by proteinase K as described above for 90 min at 1:1000 molar ratio. The
proteolyzed enzyme was loaded on a Mono Q HR 5/5 column equilibrated in 50 mM
Tris-Acetate pH 8.0, 5 mM EDTA and 2.5 mM DTT (buffer A). The enzyme eluted at 0.1
M KCl by a gradient of 0-50 % buffer B (buffer A + 1 M KCl).

Construction of NdHn truncated BE. The pEXSB plasmid purified from
Epicurian Coli XL2-Blue cells (Stratagene) was used as dsDNA template for PCR using

the following primers: 5’- GAT GCC TGG CCC ATG GCT GAA GGT ACT CAC CTG

119

CG -3’ and 5’- GCT AGT TAT TGC TCA GCG G -3’. The resulting DNA fragment and
pET23d vector were both digested with NcoI and SalI restriction enzymes (New England
Biolabs). The PCR fragment and the purified major DNA band from the pET23d
digestion were subsequently ligated using the T4 ligase (Boehinger Mannheim). This
produced plasmid pTRCl, encoding a 112 amino acid deleted branching enzyme (N d 1.1 12)
with a serine to alanine mutation at position 2 due to construction of the NcoI restriction
site. The construct was confirmed by restriction enzyme analysis and DNA sequencing of
the entire coding region.

Expression of WT and truncated E. coli BE. A 200 m1 overnight culture of the
transformed cells (BL21[DE3]) carrying pEXSB or pTRCl was added to a Biostat E10
ferrnentor containing 12 1 fresh LB with 25 ug - ml'l kanamycin or 100 ug - rnl'l
ampicillin, respectively. The cells were grown at maximum aeration with 200 rpm stirring
at 37°C to A600 = 0.4. The temperature was adjusted to 25°C and expression of the T7
RNA polymerase induced by adding isopropyl-D-thiogalactoside to 0.5 mM. After 5 hrs
induction, the cells were harvested by centrifugation at 100,000 g - min.

Expression of seleno-methionine substituted Nd1.m. The gene encoding the
truncated branching enzyme was excised from pTRCl and ligated into a pET28a vector
using the NcoI/SalI restriction sites. The plasmid was transformed into E. coli strain CT-
13, kindly provided by Dr. Honggao Yan, Michigan State University. E. coli strain CT-13
is a methionine-auxotroph BL21(DE3) derivative and tetracyclin resistant. The culture
medium (minimal medium) consisted of IX Eagle Basal medium (GibcoBRL), 0.2 %
glucose, 0.5 mM MgSO4, 100 uM CaClz, 10 M ZnClz, 10 uM F6Cl3, 0.4 mg/l thiamine,

4 mg/l d-biotin, 50 ug/ml kanamycin, and 10 ug/ml tetracyclin. Cells from 500 ml

120

minimal medium overnight culture containing 100 ug/ml L-methionine were washed and
added to a Biostat E10 ferrnentor containing 12 1 fresh minimal medium with 50 ug/ml
selenium-methionine (Sigma). The cells were grown at maximum aeration with 200 rpm
stirring at 37°C to A600 = 0.5 before addition of 1 mM isopropyl-D-thiogalactoside and
selenium-methionine to 100 ug/ml. After 15 hrs induction at 25°C, cells were harvested
and stored at -80°C.

Purification of BE WT and Ndmu. The purification procedure was modified
from Guan et al. (10). After sonication the BE WT was precipitated with 40 %
(NI-I4)2SO4, whereas the Nd” 12 enzyme was precipitated with 30 % (NH4)2804. The
MonoQ HR 16/10 pH 7.5 column was replaced with a DEAE fractogel column (EM
science) equilibrated in 50 mM Tris-Acetate pH 8.0, 10 mM EDTA and 2.5 mM DTT
(buffer C). The enzyme was eluted with a gradient of 0-50 % buffer D (buffer C + 1 M
KCl). The remaining part of the purification was identical to that reported previously
(10). BE-WT and Ndmlz eluted at 0.25 M KCl and 0.1 M KCl, respectively, for both the
DEAE fractogel and Mono Q column. For truncated branching enzyme substituted with
selenium-methionine the sample was concentrated after the MonoQ column to 1 ml. Then
200 pl batches were injected onto a Superose 12 size exclusion column (Pharrnacia,
Sweden) pre-equilibrated with buffer B (25 mM HEPES pH 7.5 and 5 mM DTT), and
eluted with 1.5 bed volume buffer B. The fractions with highest BE purities were pooled
and concentrated to 5 mg/ml.

SDS-PAGE and Western Blotting. SDS-PAGE was performed by 4-15 %
polyacrylamide gels according to the method of Laemmli (l 1). Western blotting was

carried out according to the method of Bumette (12). The primary rabbit antibody anti-BE

121

was diluted 1:1000 in 25 mM K-PO4 pH 7.2, 150 mM NaCl and 3 % gelatin. The
antigen- antibody complex was detected using anti-rabbit IgG conjugated with alkaline
phosphatase (US Biochemicals, diluted 1:10000) with a chromogenic substrate
(Boehinger Mannheim). Molecular weights were determined using Perfect Protein
markers (Mr 15, 25, 35, 50, 75, 100, and 150 kDa) for both SDS-PAGE and western

blots.
Protein sequencing. N-terrninal and C-terrninal protein sequencings were

performed by Joe Leykarn at the Macro-molecular Facility, Department of Biochemistry,
Michigan State University.

Assay of BE activity. BE activity was measured by three different assays as
described by Guan and Preiss (7).

Assay A. The phosphorylase A stimulation assay is based on the stimulation by BB
of the synthesis of a-D-glucan from a-D-Glc—l-P catalyzed by rabbit phosphorylase a (5).
Reaction mixtures contained in a final volume of 100 pl, 100 mM citrate (pH 7.0), 10
mM adenosine monophosphate, 0.2 mg phosphorylase a (Sigma) and 50 mM D-[MC]-
Glc-l-P (50 DPM -nmol'1) and the assay was initiated by an appropriate amount of BB.
One unit is defined as 1 pmol of Glc incorporated into or-D-glucan per min at 30°C.

Assay B. The branching linkage assay determines the number of branching points
introduced by BE into the substrate, reduced amylose (6). Substrates were prepared by the
reduction of enzymatically synthesized amylose (AS-320, d.p. 1815) as described by
Takeda er al. (6). Reaction volumes contained in a final volume of 100 pl, 25 mM MOPS

(pH 7.5) and an appropriate amount of substrate. The reaction was initiated by adding an

122

appropriate amount of BE. One unit is defined as 1 pmol of branching linkages formed
per min at 30°C.

Assay C. The iodine stain assay is based on the monitoring the decrease in
absorbance of the glucan-iodine complex resulting from the branching of the substrate,
amylose or amylopectin (potato type III [d.p. 1176, chain length 884] and corn
respectively, Sigma) (3,7). Reaction mixtures contained in a final volume of 500 pl, 50
mM citrate (pH 7.0) and 0.1 mg substrate. The reaction was initiated by the addition of an
appropriate amount of enzyme. One unit is defined as the decrease in absorbance of 1.0
per min at 30°C.

Protein assay. Protein concentration was measured with the BCA protein assay
reagent (13), using BSA as the standard.

Mass spectrometry. Protein samples were dialyzed against H20 and analyzed by
mass spectrometry by Stacy Trzos at the Macro-molecular Facility, Department of
Biochemistry, Michigan State University.

Protein crystallization. Initial crystallization experiments were carried out at 4°C
and 25°C by the sparse-matrix approach using the hanging drop vapor-diffusion method
described by J ancarik and Kim (14). Drops contained 2 pl precipitant solution (100 mM
HEPES pH 7.2) and 2 pl protein solution at 5 mg/ml.

X-ray diffraction analysis. Crystals were transferred to a cryo-protectant solution
containing 25 % (v/v) 2-methyl-2,4-pentanediol, 2 % (w/v) polyethylene glycol 4000 and
100 mM HEPES pH 7.5. Crystals were mounted in nylon cryo-loops (Hampton, CA) and
stored in liquid N2. High resolution data were collected at the Advanced Photon Source at
Argonne on the Structural Biology Center ID-19 beamline. Intensity data was collected in

123

a 3x3 (3072x3072 pixel) CCD area detector to a resolution of 2.3 A. The crystal to
detector distance was set to 220 mm and 160° of data was collected with an oscillation
range of 05°. Diffraction data was indexed and integrated using DENZO and scaled by

the software SCALEPACK (15).

124

A< Emacs Emma 833385 a owe—bonmmofi 05 .3 3:5on mm? .5338 08.03””.—

 

 

: 82 888 o8 we 0 252
am one 88% 2: 8m aweafinfia
a 3m 85? comm e: .Omafizv as 9.5
2: mi 82% 82. § saga 880
a. 5223.? . a.a.. as: me E
22> 3:52». 2&0on 55:95. £895 0820> 98m Ecumenism

 

HBflm mo 038 cowwoctsm A 033.

125

Results

Purification of E. coli BE-WT. The expression and purification procedure
detailed in materials and methods greatly improved the yield of E. coli branching enzyme
(Table 1). The BE-WT enzyme constituted 20 % of total protein in crude extract and
appeared as the major band on a SDS-PAGE gel (data not shown). A total yield of 250
mg BE-WT was obtained from a single 12 l batch representing a 20-fold improvement of
that reported previously (10).

Proteolysis of E. coli branching enzyme. The action of proteinase K on BE was
monitored by SDS-PAGE and branching activity. Initial proteolysis experiments were
performed using 1:50 molar ratio of proteinase K to branching enzyme (data not shown).
These experiments showed a rapid digestion of the native enzyme (84 kDa) to yield a
truncated enzyme (70 kDa) subsequently degraded to several smaller fragments.
Branching activity under these conditions was rapidly lost. Use of a 1:1000 molar ratio of
proteinase K to BE produced the 70 kDa fragment as the major product in 90 min (Figure
1A), with small amounts of fragments of 42, 35, 32, and 13 kDa. Although branching
enzyme activity decreased during the proteolysis experiment, 30 % activity remained
(Assay A) when no native enzyme was apparent on a SDS-PAGE gel after 90 min (Figure
1B). To test if branching enzyme could be protected from proteinase K cleavage by
substrate, we performed digestion experiments in presence of glycogen, amylopectin or
amylose, but neither changed the proteolytic rate or the cleavage pattern (data not shown).
In an effort to produce alternative digestion patterns, we subjected branching enzyme to
several proteases with a range of specificities (Figure 2). Whereas branching enzyme was

found sensitive to proteinase K and subtilisin, the other proteases utilized had low rates of

126

 

 

a;
.3 100 - O
'5 so -
E 60 -
i? .
g 40 -
a .
a 20 l
o

 

 

éo ' Jo ' éo ' éo ' 1bo

time/ min

on

Figure 1. SDS-PAGE (A) and branching activity (B) of E. coli BE treated with
1:1000 molar ratio of proteinase K to BE at various incubation times. A: 3 pg of total
protein was loaded per lane. The gel has been stained with coomasie brilliant blue. B:
Branching activity of BE-WT control (0) and proteinase K treated BE (I) by

phoshorylase a stimulation assay (Assay A).

127

 

150
100
75
50

35
25 ‘

 

15

 

 

 

Figure 2. SDS-PAGE of BE-WT treated with various proteases as described in
materials and methods. 3 pg of total protein was loaded per lane and the gel
stained with coomasie brilliant blue. The ratio of protease to BE-WT and
incubation time is indicated in parentheses. Control BE-WT (lane 1).
Proteinase K (lane 2; 1:1000; 90 min). Protease V8 w. P04 buffer

(lane 3; 1:30; 1 hr). Protease V8 w. NH4HC03 buffer (lane 4; 1:20; 2 hrs).
Elastase (lane 5; 1:2; 3 hrs). Chymotrypsin (lane 6; 1:30; 2 hrs). Subtilisin
(lane 7; 1:500; 90 min). Trypsin (lane 8; 1:10; 3 hrs). Carboxypeptidase Y
(lane 9; 1:3; 2 hrs).

128

digestion. Even 3 hours of treatment using a 1:2 molar ratio of elastase to BE failed to
cleave the native enzyme completely. Carboxypeptidase Y also had a low proteolytic
effect on branching enzyme and high concentrations (1:3 molar ratio) produced a
digestion pattern of an endopeptidase, probably due to small contaminations of other
proteases (Figure 2, lane 9). All proteases tested displayed similar digestion patterns,
producing a 70 kDa fragment with the notable exception of trypsin (Figure 2, lane 8). For
the latter, assays showed that neither fragment (45, 40, and 30 kDa) exhibited branching
enzyme activity. Residual branching activity for BB treated with the remaining proteases
indicated the major fragment (approx. 70 kDa) had a comparable level of activity in all
cases.

Analysis of the 70 kDa fragment. In order to analyze the 70 kDa fragment in
greater detail, we cleaved 4 mg of purified BE with proteinase K and subsequently
purified 1.6 mg of the truncated enzyme, PK-BE, as described in materials and methods
(Figure 3). N-terrninal microsequencing of purified PK—BE yielded a 1:1 molar ratio of
Leu-Leu—Ser-Glu-Gly-Thr-His-Leu- and Ser-Glu-Gly-Thr-His-Leu-Arg-Pro-, indicating
proteinase K performed a staggered cut cleaving either at position Trpm-l-Leu112 or
Leul l3-l-Serl 14. C-tenninal sequencing of both wild-type BE and PK-BE yielded Glu-
Ala-Glu-Arg-, indicating the C-terrninal of PK-BE was still intact. This gives PK-BE a
calculated MW of 71.6 kDa. In order to confirm the results from limited proteolysis, we
subsequently generated a truncated BE by inserting a start codon at position 113 via
recombinant DNA techniques as described in materials and methods. This truncated

enzyme, Nd1-n2, was expressed and purified to near-homogeneity (Figure 3, Table 2).

129

>
w

 

 

Std. 1 2 3 4 Std, 1 2 3 4
150 *- 15o ..
10° *- 1oo «-
75 ~ 5-..... 75 r-‘-.-.‘..—--——
50 I.“ M“ 50 -
35 ... : N 35 4'05
25 fi' 25 .- 1' " '
15 “m m 15

 

 

 

 

 

 

Figure 3. SDS-PAGE (A) and western blot analysis (B) of wild-type and
truncated branching enzymes. Control BE-WT (lane 1). Proteinase K
treated BE-WT (lane 2). Purified PK—BE (lane 3). Purified Ndi—nz

(lane 4). Loading per lane was 3 pg total protein in A and 100 ng total

protein in B.

130

A< .9803 Emma 5:22:53 m omflfionamona 05 .3 @9532: mm? brine ogm

 

z. 5 2.8” we 8 o as:
E 2% 2.3.: 03 can 3353-38
2: OR comma as m: 63.27: 2. 23
2: New 258: 22a cam 52a... 2580
a sack ems . 2.23 a.a.: we as
22> 2&8 528% 2:8... 222a a§_a> 2% aeaameaa

 

.2 :62 mo 033 seasoning .N 058.

131

Both SDS-PAGE and western blot analysis indicate PK-BE and Nd1.112 have similar

migration patterns and are free of any native BE.

Activity of PK-BE and Nd1-u2. Activities of pure PK-BE, Nd1-112 and BE-WT
were characterized with respect to three assays available for branching enzyme (Table 3).
Specific activities of PK-BE and Nd1-112 were very similar and consistently lower than
BE-WT (6O % of wild-type activity for assay A and 40 % for assay B and assay C).
Calculating the ratio of activity has been useful in comparing the properties of various
branching enzymes (7,16). In particular the iodine assay has been employed to establish a
measure of preference between the two substrates amylose and amylopectin. Here, the
enzymes have a similar C2/C1 ratio (iodine assay with amylopectin and amylose,
respectively), suggesting the substrate preference is unaltered for both PK-BE and Nd1-n2
relative to BE-WT. As for BE-WT, both truncated branching enzymes showed negligible
levels of amylolytic activity.
Kinetic parameters of the enzymes were determined using the branching linkage assay
(Assay B) with reduced amylose AS-320 as substrate (Table 4). Whereas the Km values
were essentially unchanged, both PK-BE and Nd1-n2 have a 3-4 fold lower specific

activity and catalytic efficiency than BE-WT.

Crystallization of E. coli branching enzyme. Previous attempts to crystallize E.
coli BE WT were unsuccessful in providing single protein crystals suitable for X-ray
diffraction analysis. When subjecting NdH 12 to crystallization trials by the sparse matrix

screening, we were able to produce single protein crystals of maximum size 0.3x0.1x0.1

132

Table 3. Specific activities of E. coli BE WT, PK-BE and Nd1-n2.

 

Specific Activities, (U/mg)

 

 

Assay WT PK-BE Nd1-n2

Phosphorylase A, A 1100 650 684

Branching linkage, B 1.10 0.39 0.47
Iodine Stain“, C

Amylose, C1 91.5 34.9 35.1

Amylopectin, C2 5 1 .2 22.1 21 .6
Ratio of activity

AIB 1000 1667 1455

NC; 12.0 18.6 19.5

A/C2 21.5 29.4 31.7

C2/C1 0.56 0.63 0.62

 

1:!The C1 and C2 denotes the iodine staining assay with amylose and amylopectin as

substrate, respectively.

Table 4. Kinetic parameters of E. coli BE WT, PK-BE and Nd1-n2.

 

WT PK-BE Nd1-112

 

Specific activity, (pmolxmin'lxmg’l) 3.7V. 0.2 0.98 ‘7- 0.05 1.25 ’7- 0.04

K.,, (pM) 11.7 ‘7. 0.7 12.7 ‘7. 1.6 12.6 +/. 1.2

k,,,/ K... (s"><M") 4.5x105 1.1x105 1.4><105

 

Activity was measured by the branching linkage assay (Assay B) using reduced amylose

133

Table 5. Statistics for the branching enzyme X-Ray diffraction data collectiona

 

Wavelength (A) 0.97794

Resolution range (A) 35.0 — 2.3 (2.38 - 2.30)
Number of measured reflections 1,671,259

Number of independent reflections 152,002

Completeness (%) 99.6 (98.6)

Rmerge (I)b (%) 8.6 (30.3)

<I>/<oI> (%) 10.43 (2.57)

 

8‘ Values in parenthesis refer to the lowest resolution shell
b Rmerge = 21 III - (I)| / 2(1), where 11 is an individual intensity measurement and (I) is

the average intensity for this reflection, with summation over all data

134

mm (Figure 4). A high-resolution native data set was collected at the Argonne Advanced
Photon Source complete to 2.3 A (Table 5). The branching enzyme crystals belong to the
P21 space group with unit cell parameters a = 91.44, b = 102.58, c = 185,41 A, and B =
91 .38°. Based on these results and the size of the enzyme (71.6 kDa monomer in
solution), we assume to have 4 molecules in the assymmetric unit. The corresponding
crystal volume per protein mass (Vm) and solvent content are within the range of values
observed for protein crystals (1 7). The crystal properties are summarized in Table 6. The
data was 99.6% complete in a 35.0 to 2.3 A resolution range with a Rmerge of 0.086 for

152002 unique reflections derived from a total of 1617259 individual reflections.

Table 6. Branching enzyme crystal properties.

 

Unit cell (A, °) A = 91.44, b = 102.58, c = 185.41
B = 91.38

Space group P2‘

Z 4

vm (A3/Da) 3.1

Solvent content (%) 56.5

 

Production of selenium-methionine Ndmlz. In order to solve the structure we
decided to obtain phasing information via multiple anomalous dispertion data derived
from selenium-methionine substituted branching enzyme crystals. Using the pTRCl
plasmid we measured that 10 % of the final branching enzyme yield was obtained prior to

induction, which would hinder full incorporation of selenium-methionine into Nd” 12

135

 

4. Crystal of the truncated E. coli branching enzyme.

Figure

136

(data not shown). Furthermore, methionine auxotroph CT-13 cells did not grow when
transformed with pTRC 1. The construct based on the double-repressed pET28 vector as
described in materials and methods effectively solved these problems. The yield from 12 1
minimal media was 5 mg of highly pure branching enzyme which was fully substituted as

determined by mass spectrometry (Table 7).

Table 7. Mass spectrometry of truncated branching enzyme.

 

Experimental MW“ Theoretical MW Error

 

Native Nd1.u2 71681.2 Da 71588.2 Da 0.130 %

Se-Met Nd1-n2 72406.5 Da 72338.6 Da 0.094 %

 

aAverage of three individual experiments.

Crystals of selenium-methionine substituted truncated branching enzyme have now been

produced and will be analyzed at the Argonne Advanced Photon Source.

137

Discussion.

Branching enzyme is important for determining the structure of glycogen and
starch. There is considerable variation with respect to specific activities, chain transfer
patterns and substrate preferences of branching enzymes from diverse sources, but little is
known of what structural features are responsible for these observations. By deleting 111
to 113 amino acids at the N-terrninal of E. coli branching enzyme by either proteolysis or
recombinant techniques, we have produced an active enzyme with altered specific
activity. We found that most proteases produced the same proteolytic pattern with a
relatively stable fragment of 70 kDa. This suggests part of the amino terminal region is
exposed to the surface of the protein as proteolysis occur at flexible loci along the peptide
chain (18). Secondary structure prediction by the program PHD (19) indicates proteinase
K cleaves the E. coli branching enzyme at the carboxy end of an or helix covering residues
105-113, but with a low reliability index. The region of residues Pheloo to Pro243 of E.
coli BE has been shown to share homology with the amino terminal (domain N) of
isoamylase from Pseudomonas amyloderamosa and these regions are proposed to share a
common structural domain (8, 20). The recent three-dimensional structure of isoamylase
shows domain N contains 160 residues, where residues 1-83 form six [3 strands and one or
helix, whereas residue 83-160 has no secondary structure (21).

It was concluded by Kuriki et al. (9) that the substrate specificity of the maize
branching enzyme isoforms is determined by the C-terrninal region. Our analysis of the
N-tenninal truncated E. coli BE indicates substrate binding is unaffected by the deletion
as the Km was essentially constant between PK—BE, Nd1.n2 and BE-WT. Also, glycogen,

amylopectin or amylose could not protect the N-tenninal region from digestion by

138

proteinase K. The data from iodine assay furthermore indicate the substrate preference of
PK-BE and Nd1-112 is unaltered from the native enzyme. These observations support the
argument that the N-tenrrinal 111-113 amino acids have limited involvement in substrate
binding for E. coli branching enzyme. The N-terminal has been proposed to be involved
in the chain transfer pattern of branching enzyme based on construction of a hybrid
enzyme from the maize isoforms (9). We are currently investigating if the N-terrninal 112
amino acids are involved in chain transfer by analyzing the branching pattern of the
truncated branching enzyme relative to the native enzyme.

A schematic representation of several branching enzymes constructed by amino
acid sequence alignments and secondary structure predictions, illustrates the diversity of
the amino and carboxy termini of branching enzymes (Figure 5). E. coli BB is
homologous to Bacillus Stearothermophillus BE (45 % identity/ 57 % similarity) and the
two enzymes differ at least with respect to therrnostability and temperature optima at
30°C and 50°C, respectively (7,22). As the B. Stearothermophillus BE has a shorter N-
terrninal by 94 residues but otherwise is very similar to E. coli BE, we measured if the
truncated E. coli branching enzymes had altered therrnostability compared to BE-WT.
However, we found the enzymes were identical in this respect. There appears to be
considerable variation of the lengths of both the N- and C-terrninal regions between
various branching enzymes (Figure 5). The enzyme from Bacillus stearothermophilus has
the shortest amino terminal region (149 amino acids) of any known branching enzyme.
Interestingly, PK-BE has only 130 amino acids at the N-terrninal, but still has branching
activity. It would be of interest to construct a series of deletion mutants of branching

enzyme at the N-terrninal toward the conserved (01/ IDs-barrel to investigate if branching

139

N-terminal R1 32 RB R4 C-terrninal

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

E. coli BE-WT L 243

mBEI I 188 191 l
mBEII [ 200

B. Stearotherm. [ 1‘9 l
E. coli PK-BE 130

 

 

Figure 5. Schematic diagram of several branching enzymes. Location of the
conserved (01/ IDs-barrel (grey) containing the four conserved regions (RI-R4) was
determined by alignment to amylolytic enzymes of known structure and secondary
structure prediction by PHD (19). The number of arrrino acids at the amino- and carboxy
terminals are indicated. mBEI: maize BE isoform 1. mBEII: maize BE isoform II. B.

Stearotherm.: Bacillus Stearothermophilus BB.

140

enzyme activity will be further altered or lost, and these experiments are currently in
progress in our laboratory. Several enzymes involving cleavage or formation of a-l ,6
glucan linkages including branching enzyme, isoamylase and pullulanase, have an amino
terminal extension relative to the conserved (oz/[3)3-barrel region. Other amylolytic
enzymes, including several a-amylases, do not possess any N-terminal region, leading to
speculation if a branching enzyme could be altered in its function by truncation
experiments.

There is currently no structure available of branching enzyme. Considerable efforts have
concentrated on crystallizing the native enzyme from E. coli with limited success. We
were hopeful that the truncated enzyme’s resistance to further proteolysis indicated a
more compact and stable enzyme hereby enabling us to crystallize it. The results from
crystallizing the NdH 12 truncated enzyme are very promising and although much future
work is needed we believe this work ultimately will lead us to the first three-dimensional

structure of branching enzyme.

141

References
l. Guan, H. P., Kuriki, T., Sivak, M. N. and Preiss, J. (1995) Proc. Natl. Acad. Sci. 92,
964-967.
2. Preiss, J. (1984) Annu. Rev. Microbiol. 38, 419-458.
3. Boyer, C. D. and Preiss, J. (1978) Carbohydr. Res. 61, 321-334.
4. Nakamura, Y., Takeichi, T., Kawaguchi, K and Yamanouchi, H (1992) Plant Physiol.
84, 329-335.
5. Hawker, J. S., Ozbun, J. L., Ozaki, H., Greenberg, E. and Preiss, J (1974) Arch.
Biochem. Biophys. 160, 530-551.
6. Takeda, Y., Guan, H. P. and Preiss, J. (1993) Carbohyd. Res. 240, 253-263.
7. Guan, H. P. and Preiss, J. (1993) Plant Physiol. 102, 1269-1273.
8. Jespersen, H. M., MacGregor, A. E., Henrissat, B., Sierks, M. R. and Svensson, B.
(1993).]. Prat. Chem. 12, 791-805.
9. Kuriki, T. Stewart, D. C. and Preiss, J. (1997) J. Biol. Chem. 272, 28999-29004.
10. Guan, H. P., Li, P., Imparl-Radosevich, J ., Preiss, J. and Keeling, P. (1997) Arch.
Biochem. Biophys. 342, 92-98.
1 1. Laemmli, U. K. (1970) Nature 227, 680-685.
12. Bumette, W. W. (1981)Anal. Biochem. 112, 195-203.
13. Smith, P. K., Krohn, R. 1., Hermanson, G. T., Mallia, A. K., Gartner, F. H.,
Provenzano, M. D., Fujimoto, E. K., Goeke, N. M., Olson, B. J. and Klenk, D. C. ( 1985)
Anal. Biochem. 150, 76-85.
14. Jancarik, J, Kim, S-H 1991. J. Appl. Cryst. 24; 409—41 1.

15. Otwinowski, Z, Minor, W (1997). Methods Enzymol. 276; 307-326.

142

16. Binderup, K. and Preiss, J. (1998) Biochemistry 37, 9033-9037.

17. Matthews, B. (1968). J. Mol. Biol. 33, 491-497.

18. Fontana, A., Fassina, G., Vita, C., Dalzoppo, D., Zamai, M. and Zambonin, M. (1986)
Biochemistry 25, 1847-1851.

19. Burkhard, R. and Sander, C. (1993).]. Mol. Biol. 232, 584-599.

20. Svenson, B. (1994) Plant Mol. Biol. 25, 141-157.

21. Katsuya, Y., Mezaki, Y., Kubota, M. and Matsuura, Y. (1998) J. Mol. Biol. 281, 885-
897.

22. Takata, H., Takaha, T, Kuriki, T., Okada, S., Takagi, M. and Imanaka, T. (1994)

Appl. Environ. Microbiol. 60, 3096-3104.

143

CHAPTER 5
TRUNCATION OF THE AMINO TERMINAL DOMAIN RESULTS IN

ALTERED CHAIN TRANSFER PATTERN OF BRANCHIN G ENZYME.

Patent pending: Binderup, K. and Preiss, J. “Truncation of the amino terminal domain

results in the altered chain transfer pattern of branching enzyme.”

144

Abstract

Previous work has reported the production of an E. coli branching enzyme with a
l 12 residue deletion at the amino terminal by limited proteolysis. Here, we investigate
the chain transfer pattern by in vitro branching of amylose using purified truncated or
native branching enzyme. Gel permeation chromatography of the synthesized a-glucan
afier 2 and 4 hours of branching shows the truncated branching enzyme transfers fewer
short chains (degree of polymerization [d.p.] < 20) and a greater proportion of
intermediate size chains (d.p. 30-90) than the native enzyme. High performance anion
exchange chromatography (HPAEC) of the branching limited product indicates the
truncated branching enzyme transfers a smaller proportion of chains d.p. 5-11 and more
chains longer than d.p. 12. Also, the genes encoding native or truncated branching
enzyme were individually expressed in a branching enzyme deficient mutant, AC71
(glgB'). By HPAEC analysis of the purified a-glucan we find that truncated branching
transfers relatively fewer chains of d.p. 4-11 and more chains d.p. longer than d.p. 12.
These observations allow us to conclude that truncation of the amino terminal has
produced an enzyme with altered branching pattern. Our results are consistent with the

construction of hybrid branching enzymes from the maize isoforms.

14S

Introduction

The biosynthesis of plant starch and bacterial glycogen requires the involvement
of at least three enzymes: ADPglucose pyrophosphorylase, starch synthase or glycogen
synthase (GS) and branching enzyme (BE; 1, 2). The rate of glycogen and starch
synthesis is allosteric regulated at the step of ADPglucose pyrophosphorylase. However,
the structures of glycogen and starch are mainly determined by the properties of glycogen
synthase or starch synthase and branching enzyme (3), although other enzymes are
involved (4, 5).

BE catalyzes the cleavage of a a-(1,4) glucosidic linkage and subsequent transfer
to a a-(l,6) branch point (6). The enzyme has been identified and characterized from a
variety of sources including E. coli, maize and rice endosperm (7-9). In many plants
multiple isoforms of the enzyme encoded by distinct genes have been reported including
maize, rice and pea seed (8-10). The characterization of maize endosperm BE isoforms 1
(mBEI) and II (mBEII) with respect to chain transfer pattern and substrate specificity, led
to the hypothesis that these enzymes have different roles in amylopectin synthesis in vivo
(8, 11). mBEI transfers long chains (d.p. 40 to 100), whereas mBEII transfers short
chains (d.p. 6 to 14). Also, the preferred substrate for mBEI was amylose whereas mBEII
showed higher activity with amylopectin. These observations led researchers to suggest
that in viva, mBEI transfers long chains which are then further branched by mBEII.
Homology in the primary structure between BE and other enzymes cleaving and/or
transferring chains of glucan, suggests a central catalytic (Ct/Bk barrel domain (12, 13).
This domain contains seven residues located in four regions that are invariably found in

branching enzymes, a-amylases, pullulanases, isoamylases, and cyclodextrin

146

glucanotransferases (13). In contrast, the size of extensions at the amino- and carboxy
termini vary greatly between the amylolytic enzymes (14). By constructing a hybrid
branching enzyme out of mBEI and mBEII, an enzyme containing the amino terminal
region and central (or/(3)3 barrel of mBEII and carboxy terminal from mBEI was obtained
(15). This resulted in an active branching enzyme with the chain transfer pattern of
mBEII and a substrate specificity of mBEI, suggesting the amino terminal being involved
in the branching pattern.

Recently, we have reported the construction of a truncated E. coli branching
enzyme by limited proteolysis using proteinase K (16). This produced an enzyme (PK-
BE) with 111 or 113 residues at the N-terminal, whereas the C-terminal was intact. The
truncated enzyme had 40-60 % activity of the native E. coli BE depending on the type of
assay and unaltered substrate specificity. Furthermore, a N-terminal 112 residue deletion
of E. coli BE constructed by recombinant DNA technology (N dH 1 2) was found to have
properties identical to PK-BE. In this study we investigate the chain transfer pattern of
the truncated E. coli branching enzyme using either in vitro branching of amylose or

expression in AC71 (glgB) followed by analysis of the synthesized a-glucans.

147

Materials and methods
Plasmids. glgB was excised from plasmid pEXSB (17) with restriction enzymes
Neal and Sal] and ligated into the NcoI and Sal] sites of vector pET23d (N ovagen),
forming plasmid pGBE, encoding the native E. coli branching enzyme. The construction
of plasmid pTRCl encoding an enzyme with a 112 residue deletion at the amino terminal
of E. coli BE was described earlier (16).

Expression of wild type and truncated BE in E. coli strain AC71 (glgB').
Overnight cultures of E. coli B containing vector pET23d or strain AC71 (glgB', 3)
transformed with plasmids pET23d, pGBE and pTRCl were diluted 1:20 into fresh
Kornberg Broth pH 7.4 containing 100 ug/ml arnpicillin and 2 % glucose. The cells were
grown at 37°C to OD6OO = 0.5 before induction with iSOpropyl B-D-thiogalactoside to 0.5
mM. Incubation was continued at 25°C for 12 hrs before harvest. The a-glucan was
isolated from about 25 g of cell paste according to the method of Preiss et al. (18). The
yield (crude extract and purified product) was determined by the complete hydrolysis of
a-glucan by glucoamylase and (it-amylase (19) and subsequent quantification by NADPH
formation through action of hexokinase and glucose-6-phosphate dehydrogenase (20).

Purification of branching enzymes: The native E. coli branching (BE-WT), E.
coli branching enzyme treated with proteinase K (PK-BE) and the E. coli branching
enzyme with 112 residues deleted at the N-terminal by recombinant DNA techniques
(N dH 1 2) were purified to homogeneity as described earlier (16).

Branching enzyme activity. The branching enzyme activity was measured by
either the phosphorylase stimulation assay (assay A) or the branching linkage assay

(assay B).

148

Assay A. The phosphorylase A stimulation assay is based on the stimulation by
BE of the synthesis of a-D-glucan from a-D-Glc-l-P catalyzed by rabbit phosphorylase a
(21). Reaction mixtures contained in a final volume of 100 pl, 100 mM citrate (pH 7.0),
10 mM adenosine monophosphate, 0.2 mg phosphorylase a (Sigma) and 50 mM D-[‘4C]-
Glc-l-P (50 DPM - nmol") and the assay was initiated by an appropriate amount of BB.
One unit is defined as 1 nmol of Glc incorporated into a-D-glucan per min at 30°C.

Assay B. The branching linkage assay determines the number of branching points
introduced by BB into the substrate, reduced amylose (11). Substrates were prepared by
the reduction of enzymatically synthesized amylose (AS-320, d.p. 1815) as described by
Takeda et al. (11). Reaction volumes contained in a final volume of 100 pl, 25 mM
MOPS (pH 7.5) and an appropriate amount of substrate. The reaction was initiated by
adding an appropriate amount of BB. One unit is defined as 1 nmol of branching linkages
formed per min at 30°C.

Glycogen synthase assay. The glycogen synthase activity was measured by the
primed glycogen synthesis reaction (22). The reaction mixture (200 pl) contained 300
nmol ADP-[14C]glucose (350 DPM - nmol'l), 0.5 mg rabbit liver glycogen, 2 nmol GSH,
100 ug BSA, 0.1 nmol magnesium acetate and 5 nmol potassium acetate in 0.1 M
biscine pH 8.0. The reaction was initiated by adding an appropriate amount of glycogen
synthase. One unit is defined as the incorporation of 1 nmol glucose into glycogen per
min at 37°C.

Protein assay. Protein concentration was measured with the BCA protein assay

reagent (23), using BSA as the standard.

149

Analysis of branched chain length Reduced amylose (10 mg, AS-320, d.p.
1815) was incubated in 25 mM MOPS, pH 7.5 (1 ml) at 30°C with branching enzyme (10
mU of E. coli BE WT, PK-BE and NdH 12 by assay B). After 2, 4 and 16 hrs, aliquots of
the reaction was heated in boiling water bath for 2 min. The synthesized a-glucan was
then debranched by adding isoamylase (100 U/ml) and Na-acetate pH 3.5 to a final
concentration of 100 mM. After 90 min at 45°C, the solution was heated in a boiling
water bath for 5 min and immediately analyzed by HPAEC and gel filtration.

HPAEC analysis. The debranched a-glucan solution was filtered through 0.22
pm membrane and 25 pl samples injected in the BioLC HPAEC (Dionex), using a
CarboPac PA-l column (250 x 4 mm). Eluent A was 150 mM sodium hydroxide; eluent
B was 150 mM sodium hydroxide containing 500 mM sodium acetate. The gradient
program was: 25 % eluent B at time 0, 45 % at 15 min, 60 % at 45 min, 70 % at 80 min,
and 80 % at 100 min with a flow rate of 0.3 ml - min". The standard was 25 pl of 5
mg - ml'l maltodextrin (d.p. 1-20, Aldrich).

Analysis of chain length by gel filtration. The debranched (it-glucan solution
was briefly centrifuged before loading 900 pl (containing 8 mg debranched sample or 3
mg substrate AS-320) on a TSK HW 50 gel filtration column (dextn'n separation range
MW 500-20000, O 1.6 cm X 55 cm) equilibrated in H20. Eluent was H20 at flow rate 20
ml - hr". 2 ml fractions were analyzed for kmax and maximal absorption of the iodine-
polysaccharide complex and total carbohydrate content by the phenol-H2804 method
using glucose as standard (24).

Glucan-iodine complex kmax of the glucan-iodine complex was performed as
described by Krisman (25) in presence or absence of CaClz.

150

Average a-glucan chain length determination. First, the amount of a-(l ,6)
linkages were determined by isoamylase debranching followed by a Park-Johnson
analysis as described above. Then, a-glucan samples (200 pl) were hydrolyzed to glucose
by the concerted action of glucoamylase and a-amylase as described by Huijing (19) and
total glucose measured by the Park-Johnson method. Average chain length was calculated

as the ratio of total glucose to a-(1,6) linkages.

151

Results

Chain transfer pattern by PK—BE, Nd1-m, and BE-WT. Modification of the
structure of reduced amylose AS-320 by the branching enzymes was monitored by gel
filtration and HPAEC. We analyzed the transferred chains by gel filtration after 2 and 4
hours branching for BE-WT, PK-BE, and Nd“ 12 (Figure l). The control, substrate
amylose AS-320, eluted as a sharp peak with calculated average chain length of d.p.
1500-2000. After 2 hours branching, PK-BE and NdH 12 showed a much broader range of
transferred chains relative to BE-WT (Figure lB-D). As the same number of units of
activity (by assay B) was added for all three enzymes, this observation can not be
attributed to differences in activity. Furthermore, the deduced average chain size for the
collected fractions indicate the substrate has been modified by branching. In case of PK-
BE and NdH 12, the maximal optical density of the iodine-polysaccharide complex
(Figure lC-D, solid line) shows a bimodal peak after 2 hours of branching, probably due
to the strong absorption of long chains. After 4 hours branching, BE-WT (Figure 1F)
shows a narrow peak compared to PK—BE and NdH 12, which furthermore is shifted to the
right, indicating transfer of smaller oligosaccharide chains. The maximal optical density
of BE-WT is much reduced compared to the truncated branching enzymes, further
suggesting absence of long chains.

The data from gel filtration is summarized in Figure 2. Both truncated branching
enzymes have very similar chain transfer patterns and are markedly different than BE-
WT. For both 2 and 4 hours of branching the BE-WT has a higher population of small
chains (d.p. <20) and lower amounts of intermediate chains (d.p. 20-30 and d.p. 30-90)

compared to the truncated branching enzymes. All enzymes show a similar population of

152

Figure 1. Analysis of transferred chain length by gel filtration after 2 hrs (A-D)
and 4 hrs (E-H) of branching using reduced amylose AS-320 as substrate. The
conditions for sample preparation, column- and debranching conditions is detailed
in materials and methods. The x-axis shows the elution volume scale (ml). Total
carbohydrate (O) of each 2 ml fractions was measured by the phenol sulfuric acid
method. The total carbohydrate scale (pg/m1) is shown at left on the Y-axis. lmax
values are shown as the broken line. The Y-axis on the right represents the kmax
wavelength (nanometers). The optical density (O.D.) of the iodine-polysaccharide
complex (solid line) was measured at Xmax for each fraction. The optical density
scale is shown at far right on the Y-axis. The degree of polymerization (d.p., I)
was determined by using kmax values of the debranched glucans according to
Banks et al. (26). Note the logarithmic scale of the deduced d.p.

A, E: Substrate reduced amylose AS-320.

B, F: BE-WT after 2 and 4 hrs of branching, respectively.

C, G: PK-BE after 2 and 4 hrs branching, respectively.

D, H: Nd1-112 after 2 and 4 hrs branching, respectively.

153

 

 

 

 

 

i

 

-15

 

d4m250

"n. [Edrt70
““5 ..

 

 

 

700 l7

 

Tao

4700

 

.
son 3

 

 

 

-700

 

 

800

500 t3

400 '1

 

elution volume, ml

160

 

 

#400 ”2

 

 

 

 

-‘700

-500 '3

+400 -1

 

 

 

 

 

- 700

-800

b
-5oo 3

 

 

 

 

 

-700

-600

-5oo ’3

Moo -1

 

 

40 so ' ab 166

elution volume, ml

154

 

130

population, %
N w A
‘P . ‘?

 

  

 

 

 

t‘fi

l1: ‘1

 

‘ ,

< 20 20-30 30-90 > 90

population,%
to to A 01
9.9? ‘5’.

AL

  

 

‘Vrfl

< 20 203 30-90 > 90
d.p.

 

 

.4
.09

 

Figure 2. Relative distribution of transferred oligosaccharide chain sizes after
2 hrs (A) and4 hrs (B) branching based on data from gel filtration. BE-WT
(black). PK-BE (gray). Ndl-112 (white). For explanation of deduced d.p.,
refer to legend of Fig. l. The relative population was determined by
measuring total carbohydrate of fractions of a given chain size. Sum of

carbohydrate in all fiactions was set to 100 %.

155

long chains (d.p. >90), most likely reflecting equal amounts of partly reacted substrate
AS-320.

Further investigation of transferred short chains was performed by HPAEC.
Figure 3 shows the HPAEC pulsed amperiometric detector (PAD) response
chromatogram of BE-WT and NdH .2 after 2 and 16 hrs of branching. There is a clear
difference in chain transfer pattern between BE-WT and Nd” 12 after 2 hrs of branching
(Figure 3A,B). The chromatogram of BE-WT (Figure 3A) shows a glucan distribution
with d.p. 11 being most transferred, and few chains are detected above d.p. 25. In
contrast, NdH 12 shows a bimodal distribution centered at d.p. 11 and d.p. 16 with
transferred chains up to d.p. 40 easily detected. These chromatograms are in agreement
with the results obtained from gel filtration after 2 hrs of branching (Figure 2A). The
glucan product after 16 hrs of branching is shown in Figure 3C,D. These chromatograms
illustrate the branching limited products of amylose AS-320 by BE-WT and NdH 12 as
branching enzyme activity was still present after 16 hrs incubation and addition of 10-
fold the branching activity yielded identical glucan structures (data not shown). kmax of
the synthesized glycogen was 460 nm and 474 nm for BE-WT and NdH 12, respectively.
The peak areas from the HPAEC chromatograms after 16 hrs were calculated and the
relative peak area difference between NdH 12 and BE-WT is shown in Figure 4 for easy
comparison. NdH 12 transfers a relatively greater proportion of chains longer than d.p. 12
and fewer chains d.p. 4 to d.p. 11 compared to the native E. coli branching enzyme.

In order to investigate an in vivo comparison of BE-WT and truncated BE, the
genes coding these were expressed individually in the glgB deficient E. coli strain AC71.

Figure 5 shows iodine staining of E. coli B containing pE'123d, AC71 transformed with

156

Figure 3. High Performance Anion Exchange Chromatography of debranched or-
glucan formed by the action of BE-WT and Ndl-112 on reduced amylose AS-320
after 2 hrs (AB) and 16 hrs (CD) of branching. Numbers above peaks indicate
particular chain sizes.

A, C: BE-WT after 2 and 16 hrs of branching, respectively.

B, D: Ndl-112 after 2 and 16 hrs branching, respectively.

157

11

 

Retention time

 

omcoamm: Q<n_

158

 

relative peak area (%)
“9
/
I
I
I\.

n l

 

 

. . 1 . I v r . r ' I ' I
0 5 1O 15 20 25 30 35 40
Chain length, d.p.

Figure 4. Difference in chain distribution between Nd1_112 and BE-WT

resulting from branching amylose AS-320 for 16 hrs. The difference

was the relative peak areas from Nd1_112 minus the relative peak area

from BE-WT for a given chain length. The sum of chains with d.p. 1-40

was set as 100 %.

159

 

Figure 5. Staining of cell colonies with iodine vapor. (A) E. coli B transformed
with pET23d. E. coli strain AC71 (glgB-) was transformed individually with
control pET23d (B), pGBE encoding BE-WT (C) and pTRCl encoding
truncated BE (D). Overnight cultures were dispensed on Kornberg plates

(pH 7.4) containing 100 mg/ml ampicillin, 1 mM isopropyl b-D-

thiogalactoside and 2 % glucose. After 15 hrs at 37°C the plates were

inverted over a lawn of iodine crystals for cell staining.

control pET23d, pGBE encoding BE WT and pTRCl encoding truncated BE.
AC71+pET23d results in a green-blue staining of the iodine-glucan complex
characteristic of amylose (25). The order of brown iodine-glycogen color development
was AC71+pGBE >> AC71+pTRC] > E. coli B. These staining patterns could be due to
varying amounts or different structures of the or-glucans. Subsequently, we isolated a-
glucan from E. coli B and AC71-504 with pETQBd, pGBE or pTRC] as described in
materials and methods (Table 1). The amount of produced a-glucan was consistently
different between the strains E. coli B and AC71 with pET23d, pGBE, or pTRC]. The
measured glucan amount rules out the weak iodine staining pattern of E. coli B and
AC71+pTRC] as being due to low glucan amounts, as both have higher glucan content
than the dark-staining AC71+pGBE. Mm values of the purified or-glucan in absence of
CaClz were consistent with the iodine staining patterns observed (Table 1). In presence of
CaClz a large shift in km“ was seen for AC71+pGBE. Average chain length of
AC71+pET23d was 38, indicating a largely linear a-glucan polymer. In contrast, average
chain length of AC71+pTRC] was slightly higher than E. coli B and AC71+pGBE. To
further analyze the synthesized a-glucan, we performed HPAEC of the debranched
polysaccharides as described in materials and methods (Figure 6). This analysis
demonstrates effective complementation of AC71 with pGBE and pTRCl and allows for
comparison with the control strain E. coli B. As in E. coli B, the most transferred chain
size of AC71 complemented with pGBE and pTRC] was d.p. 8. However, AC71+pTRC]
shows a higher proportion of longer chains. Detection of short chains in the control strain
AC71 was profoundly difficult, even at 50-100 times the usual HPAEC sensitivity

(Figure 6D). The relative peak areas between AC71+pTRC] and AC71+pGBE shows

161

.8866 £55 5 @8388 33 E088 502MB 98 53:8 owafimNcm .835 382w: 203 ficogoaxm

 

 

m3 m: _.m _ Nwod owhd vm fiuga + KU<
at. m _ m Nd ES 486 e mmoa + :3.
5mm own mm God Noodv «A 3.ng + KU<
wmv owe w.: mwod and 2: m :8 .m
E: E: .53 ESE ESE ME

293 + 30.6 I fimc2 :85 .33qu m0 @363 mm anon—wé :Ebm

 

demos—mac mammogram 2.: mo

momtoaoa can hmwNMv _ >U< :8 .m E 08.35 @20ch @8855 98 H? mm mo commmoaxm A 2an

162

Figure 6. High Performance Anion Exchange Chromatography of debranched 0L-
glucan purified from E. coli B+pET23d (A), AC71+pGBE (B), AC71+pTRC (C)
and AC71+pET23d (D). Isolation of the (it-glucan was performed as described in
materials and methods. Numbers above peaks indicate particular chain sizes.
Sensitivity of detector in (D) was 50-100 times that of (A-C) in order to detect

product and large peak on left (D) is due to buffer.

163

12

8

12
5
“All“

8

 

mmcoame 9E

 

LNRM/wwwmuu

Retention time

164

that the truncated branching enzyme in host strain AC71 transfers a smaller proportion of
d.p. 5 to d.p. 11 and a greater proportion of chains longer than d.p. 12 compared to the
native branching enzyme (Figure 7). We found the glycogen structure of AC71
complemented with pGBE was overall very similar to that isolated from a wild type E.
coli B strain, although glycogen from E. coli strain B was slightly more branched than the
product from AC71+pGBE (Figure 7).

Branching enzyme activity in crude extracts varied considerably as determined by
the phosphorylase a stimulation assay but was consistent between isolations (Table 1).
AC71+pTRC] had higher branching activity than either E. coli B or AC71+pGBE but
still transferred longer chains than either, ruling out attributing the different glycogen
structure to incomplete branching. Interestingly, the 20-fold difference in branching
activity between E. coli B and AC71+pGBE resulted in only a minor difference in
glycogen structure. Glycogen synthase activity, unaffected in strain AC71, was measured

as a control and found to be at similar levels in all strains.

165

 

 

 

 

1- I
a .3 ”s...
g 0/ O I -I
‘5 0‘ / 000 o
x 4
g 'i.
a. 4, /
9 ,I
': 4 '-
(D
s. _2q

I
‘3 ' I ' I ' C j T
0 10 20 30 40

Chain length, d.p.

Figure 7. Difference in chain distribution between or-glucan purified from
E. coli B, AC71+pGBE and AC71+pTRC. (I) Difference plotted as the
relative peak areas from Ndl-HZ minus the relative peak area from
BE-WT for a given chain length. (0) Difference plotted as the relative
peak areas from E. coli B minus the relative peak area from BE-WT for
a given chain length. The graphs are average of two separate isolations

of a-glucan. The sum of chains with d.p. 1-40 was set as 100 %.

166

Discussion.

By deleting 111 to 113 amino acids at the N-terminal of E. coli branching enzyme
by either proteolysis or recombinant techniques, we have previously shown that the
activity of the enzyme is reduced. In the present study, HPAEC and gel filtration
experiments indicate that the transfer pattern of the truncated branching enzyme differ
significantly from that of native E. coli branching enzyme in transfer of short and
intermediate chain sizes. We base these conclusions on the glycogen structures resulting
from branching amylose in vitro with purified branching enzyme in addition to isolation
of or-glucan from glgB deficient strain AC71 complemented with either native or
truncated BE. The in vitro branched amylose and the glycogen isolated from BE
complemented AC71 have similar, although not identical structures, possibly due to the
interplay of additional enzymes in vivo. However, in both cases we find that truncated
branching enzyme transfers fewer chains shorter than d.p. 11 and more chains of d.p. 12
and longer relative to the native BE (compare Figure 4 and Figure 7).

For both enzymes we observe some relative long chains (d.p.>15), even after 16
hrs of branching of the amylose substrate. Also, there are long chains in the glucan
product purified from the AC71 mutant complemented with WT or truncated BE. Rather
than indicating incomplete branching, this probably reflects steric hindrance within the
dense glucan polymer for further BE action. We believe the structures of these branching
limited products indicate that the observed differences in chain transfer patterns between
BE-WT and the N-terminally truncated branching enzyme are due to intrinsic properties
of the enzymes.

We can not fully explain the iodine staining patterns (Figure 5) and the Mm

167

values of the purified glucan polymers (Table 1). Based on average chain length and
HPAEC, E. coli B and AC71+pGBE should have similar staining with iodine and kmax
values, both slightly below km, of AC71+pTRC]. In fact, when branching amylose with
BE-WT or Nd” 12, we find 7km“ values consistent with the glycogen structures of 460 nm
and 474 nm, respectively. Our observations could be explained by the presence of small
amounts of linear chains or longer branches in AC71+pGBE, as the specific absorbance
of such chains is high compared to that of glycogen (27). The effect of CaClz in the
absorbance of the glucan-iodine complex is a 10-fold increase for glycogen, whereas the
absorbance of amylose is diminished (25). When measuring km“ of AC71+pGBE in
presence of CaClz we observe a much shorter wavelength. This shift indicates the
presence of both glycogen and linear chains in case of AC71+pGBE. The ratio of BE
activity to GS activity in crude extract is 6:1 and 9:1 in E. coli B and AC71+pTRC1,
respectively, whereas the ratio in AC71+pGBE is 1:3 (Table 1). This allows for
speculation if the relatively large amount of glycogen synthase activity could have
produced a small fraction of linear chains in the latter strain, explaining the staining
patterns with iodine.

Our findings are consistent with chimeric enzyme constructions between the
maize isoforms, where the amino terminal of branching enzyme was concluded to play a
role in the size of oligosaccharide chains transferred (15). The low sequence conservation
between branching enzymes complicates the construction of hybrid branching enzymes,
especially in absence of a three-dimension structure. The study presented here establishes
an alternative method for deducing valuable structure-function relationships of branching

enzymes. Giving the large size variability of the amino and carboxy terminal regions of

168

branching enzymes (14, 16), it would be of interest to construct additional deletions by
recombinant DNA technology. We believe this would allow for further investigation of
the determinants of substrate specificity and transfer patterns, and these experiments are
currently in progress in our laboratory. This approach may enable rational design of chain
transfer patterns by branching enzymes, leading to desirable polysaccharides for specific

industrial uses.

169

References
1. Preiss J. (1991) Oxford Surv. Plant Mol. Cell Biol. 7, 59-114.
2. Preiss, J. (1984) Annu. Rev. Microbiol. 38, 419-458.
3. Guan, H. P., Kuriki, T., Sivak, M. N. and Preiss, J. (1995) Proc. Natl. Acad. Sci. 92,
964-967.
4. Ball, 8., Guan, H. P., James, M., Myers, A., Keeling, P., Mouille, G., Buleon, A.,
Colonna, P. and Preiss, J. (1996) Cell 86, 349-352.
5. Colleoni, C., Dauvillee, D., Mouille, G., Buleon, A., Gallant, D., Bouchet, B., Morell,
M., Samuel, M., Delrue, B., d’Huelst, C., Bliard, C., Nuzillard, J-M. and Ball, S, (1999)
Plant Physiol. 120, 993-1003.
6. Borovsky, D., Smith, E. C. and Whelan, W. J. (1976) Eur. J. Biochem. 62, 307-312.
7. Boyer, C. and Preiss, J. (1977) Biochemistry 16, 3693-3699.
8. Guan, H. P. and Preiss, J. (1993) Plant Physiol. 102, 1269-1273.
9. Nakamura, Y., Takeichi, T., Kawaguchi, K. and Yamanouchi, H. (1992) Plant Physiol.
84, 329-335.
10. Smith, A. M. (1988) Planta 175, 270-279.
11. Takeda, Y., Guan, H. P. and Preiss, J. (1993) Carbohyd. Res. 240, 253-263.
12. Baba, T., Kimura, K., Etoh, H., Ishida, Y., Shida, O. and Arau, Y. (1991) Biochem.
Biophys. Res. Commun. 181, 87-94.
13. Jespersen, H. M., MacGregor, A. E., Henrissat, B., Sierks, M. R. and Svensson, B.
(1993).]. Prat. Chem. 12, 791-805.
14. Jespersen, H. M., MacGregor, A. E., Henrissat, B., Sierks, M. R. and Svensson, B.

(1991) Biochem. J. 280, 51-55.

170

15. Kuriki, T. Stewart, D. C. and Preiss, J. (1997).]. Biol. Chem. 272, 28999-29004.

16. Binderup, K., Mikkelsen, R. and Preiss, J. (2000) Arch. Biochem. Biophys. In Press.
17. Guan, H. P., Li, P., Imparl-Radosevich, J ., Preiss, J. and Keeling, P. (1997) Arch.
Biochem. Biophys. 342, 92-98.

18. Preiss, J., Greenberg, E. and Sabraw, A. (1975) J. Biol. Chem. 250, 7631-7638.

19. Huijing, F. (1970) Clin. Chim. Acta 30, 567-572.

20. DiPietro D. L. and Weinhouse S. (1960).]. Biol. Chem. 235, 2542-2445.

21. Hawker, J. S., Ozbun, J. L., Ozaki, H., Greenberg, E. and Preiss, J (1974) Arch.
Biochem. Biophys. 160, 530-551.

22. Holmes, E. and Preiss, J. (1979) Arch. Biochem. Biophys. 196, 436-448.

23. Smith, P. K., Krohn, R. I., Hermanson, G. T., Mallia, A. K., Gartner, F. H.,
Provenzano, M. D., Fujimoto, E. K., Goeke, N. M., Olson, B. J. and Klenk, D. C. (1985)
Anal. Biochem. 150, 76-85.

24. Dubois, M., Gilles, K. A., Hamilton, J. K., Rebers, P. A. and Smith, F. (1956) Anal.
Chem. 28, 350.

25. Krisman, C. R. (1962) Anal. Biochem. 4, 17-23.

26. Banks, W., Greenwood, C. T. and Khan, K. M. (1971) Carbohyd. Res. 17, 25-33.

27. Murdoch, K. A. (1992) Carbohydr. Res. 233, 161-174.

171

CHAPTER 6

AN IN VIT R0 RECONSTITUTION SYSTEM FOR GLYCOGEN

BIOSYNTHESIS

172

Abstract.

We have constructed a reconstitution system for biosynthesis of bacterial
glycogen in vitro. The E. coli glycogen synthase and two different branching enzymes
were purified to homogeneity and used in glucan synthesis from ADP-glucose. We also
used branching enzymes and phosphorylase in synthesis of glucan with glucose-1-
phosphate as substrate for comparison. The in vitro synthesized polysaccharides were
purified before debranching and subsequent analysis of chain lengths by high
performance anion exchange chromatography. From the resulting elution profiles we are

able to evaluate several parameters suspected to affect the polyglucan yield and structure.

173

Introduction.

Glycogen is a polymer of glucose with 90 % a-(l ,4) glucosidic linkages and 10 %
a-(1,6) glucosidic linkages at branch points. Many bacteria are able to accumulate
glycogen as an energy storage polymer (1 ), and biosynthesis of bacterial glycogen
involves at least three enzymes. ADP-glucose pyrophosphorylase catalyzes the formation
of ADP-glucose from glucose-l -phosphate and ATP, releasing PPi as a by-product (2).
The rate of glycogen biosynthesis is allosterically regulated at the step of ADP-glucose
pyrophosphorylase, and in most bacteria the enzyme is activated by fructose-1,6-
bisphosphate and inhibited by adenosine monophosphate (3). Glycogen synthase (GS)
forms the or-(l ,4) glucosidic linkages of glycogen using ADP-glucose as a substrate (4).
Finally branching enzyme (BE) catalyzes the cleavage and subsequent transfer of a-(l ,4)
glucan chains to a-(1,6) branch points (5). It is believed the properties of glycogen
synthase and branching enzyme mainly define the or-glucan structure in vivo. Previously,
it has been suggested that a glycoprotein might be involved in the priming of the or-(l ,4)
glucan synthesis in E. coli (6), as in the case with glycogenin in yeast (7,8). However,
evidence has since accumulated that the three enzymes described above are sufficient for
E. coli glycogen biosynthesis, and that an additional primer is not required, apart from the
endogenous or-(l ,4)-glucan primer associated with glycogen synthase (9).

Carl and Gerti Cori first described in vitro synthesis of a polyglucan by use of
phosphorylase and glucose-l-phosphate as substrate (10), which led researchers to
believe phosphorylase was a biosynthetic enzyme. Phosphorylase has now long been
known as involved in glycogen degradation, and the three enzymes described above as

constituting the biosynthetic pathway. However, no studies have reported in vitro

174

glycogen synthesis based on the biosynthetic enzymes. This report describes a glycogen
biosynthetic reconstitution system based on purified preparations of glycogen synthase
and branching enzyme with ADP-glucose-as the substrate (Figure 1A). This allows us to
study several parameters suspected to influence the final polyglucan structure. First, we
use the native E. coli branching enzyme (BE WT) as well as purified truncated E. coli
branching enzyme (N d“ 12; Chapter 4), which has been shown to have an altered chain
transfer pattern (Chapter 5). This enables us to measure if branching enzymes with
different properties result in distinct polysaccharide structures from de novo synthesis.
Furthermore, we use branching enzyme and phosphorylase a (PA) with glucose-1-
phosphate as substrate, in order to compare or-(1,4)-glucan synthesis by GS vs. PA and
the corresponding effects on the polysaccharide structures (Figure 1B). We also vary the
incubation time to measure its impact on or-glucan yields and structures. Finally, we
analyze how using different enzymatic ratios of branching enzyme and GS or PA may
affect the structure and yields of the polysaccharides. Our results indicate several
parameters influence the polysaccharide product and that we may be able to design

specific a-glucan structures by the de novo synthesis method described here.

175

25508 0:0 £2,588 5 0035000 00 392338930 028388

comma 005080802“ :wE .3 03.405 003 638m 502m 33235-6. 3-5 60031va-6 35300." 05 CO 88250 2:. AN:

-62V 083:0 3328.5 020055 :00 .m 05 .8 2.3 mm: 08.3% mama—0:05 093-23, :8 N 3 0083800 003 35a 30:03

0. To 05 mo accustom .Ae 03—bonamonq cowoobw .3 223839- 708033 80¢ .8 A3 0005:? cowoobw .3 00002wifl<

20533 05 80¢ 00200583 0003 832:: 0m20002w Aw. 3-6 22. 200550 502m 98: 0.0 MO 08038 530.00% ._ 0.5me

5320 8550-658
.Bxééde
52:0 Bfifiesdé
.Bxééda

58% 86.59658
.Buazéde
seem Bfi:§-€.c-a
6835.38

N _ T _ U2
03605085 .. ..
.00 5020-205 T Banana _ 382w 2 0322.8:
N — T _ CZ
2062525
. . All 080: w- 000 8?. Ewe?»
Imm go—JwIAV —VI.U — mg + a TU

<

176

Materials and methods
Purification of enzyme material E. coli BE-WT and Nd|-112 were purified as
described by Binderup et al (11). The purification of E. coli glycogen synthase was
performed as described by Furukawa et al (12).
Branching enzyme activity. The branching enzyme activity was measured by the
phosphorylase stimulation assay (assay A).

Assay A. The phosphorylase A stimulation assay is based on the stimulation by
BE of the synthesis of or-D-glucan from or-D-Glc-l-P catalyzed by rabbit phosphorylase a
(13). Reaction mixtures contained in a final volume of 100 pl, 100 mM citrate (pH 7.0),
10 mM adenosine monophosphate, 0.2 mg phosphorylase a (Sigma) and 50 mM D-[14C]-
Glc-1 -P (50 DPM - nmol'l) and the assay was initiated by an appropriate amount of BB.
One unit is defined as 1 pmol of Glc incorporated into a-D-glucan per min at 30°C.

Glycogen synthase assay. The glycogen synthase activity was measured by the
primed glycogen synthesis reaction (14). The reaction mixture (200 pl) contained 300
nmol ADP-[”C]glucose (350 DPM - nmol'l), 0.5 mg rabbit liver glycogen, 2 pmol GSH,
100 pg BSA, in 0.1 M citrate pH 7.0. The reaction was initiated by adding an appropriate
amount of glycogen synthase. One unit is defined as the incorporation of 1 pmol glucose
into glycogen per min at 37°C.

Phosphorylase activity assay. The phosphorylase activity was measured by
synthesis of or-D-glucan from a-D-glucose-l-phosphate into a glycogen primer. The
reaction mixture (200 pl) contained 100 mM D-[14C]-Glc-1-P (50 DPM - nmol'l), 10 mM

adenosine monophosphate and 200 pg glycogen in 0.1 M citrate pH 7.0. The assay was

177

initiated by adding phosphorylase a (Sigma). One unit is defined as 1 pmol of Glc

incorporated into a-D—glucan per min at 30°C.

Protein assay. Protein concentration was measured with the BCA protein assay
reagent (15), using BSA as the standard.

SDS-PAGE analysis. SDS-PAGE was performed by 4-15 % polyacrylamide gels
according to the method of Laemmli (16).

Synthesis of a-D-glucan by branching enzyme and glycogen synthase.
Synthesis of a-D-glucan through the concerted action of glycogen synthase and
branching enzyme was performed using ADP-glucose as substrate. The reaction mixture
(200 pl) contained 67 mM ADP-['4C]-glucose (3.1 DPM - nmol"), 2 pmol GSH, and 100
pg BSA, in 0.1 M citrate pH 7.0. The incubation time at 30°C and amount of branching
enzyme and glycogen synthase was varied as indicated in each experiment. At each time
point, the reaction was terminated by boiling and the a-D-glucan product washed with
methanol and ethanol, before drying overnight in a vacuum desiccator.

Synthesis of or-D-glucan by branching enzyme and phosphorylase a.
Synthesis of a-D-glucan through the concerted action of phosphorylase a and branching
enzyme was performed using glucose-l-phosphate as substrate. The reaction mixture
(200 pl) contained 125 mM D-[14C]-glucose-1-phosphate (2.0 DPM - nmol'l) and 10 mM
adenosine monophosphate in 0.1 M citrate pH 7.0. The incubation time at 30°C and
amount of branching enzyme and phosphorylase a was varied as indicated in each
experiment. At each time point, the reaction was terminated by boiling and the a-D-
glucan product washed with methanol and ethanol, before drying overnight in a vacuum

desiccator.

178

Analysis of branched chain length. The synthesized a-glucan was solubilized in
H20 and debranched by adding isoamylase (500 U/ml) and Na-acetate pH 3.5 to a final
concentration of 100 mM. After 4 hrs at 45°C, the solution was heated in a boiling water
bath for 5 min and immediately analyzed by HPAEC.

HPAEC Analysis. The debranched or-glucan solution was filtered through 0.22
pm membrane and 25 pl samples injected in the BioLC HPAEC (Dionex), using a
CarboPac PA-l column (250 x 4 mm). Eluent A was 150 mM sodium hydroxide; eluent
B was 150 mM sodium hydroxide containing 500 mM sodium acetate. The gradient
program was: 25 % eluent B at time 0, 45 % at 15 min, 60 % at 45 min, 70 % at 80 min,
and 80 % at 100 min with a flow rate of 0.3 ml - min". The standard was 25 pl of 5

mg - m1" maltodextrin (d.p. 1-20, Aldrich).

179

Results.

Enzymatic activities. All enzymes used in this study were purified to near-
homogeneity to prevent any interfering reactions from enzymes involved in carbohydrate
modification or catabolism (Figure 2). The specific activities of the enzyme preparations
were determined in the conditions used during the oc-glucan synthesis in presence of a
glycogen primer, in order to perform direct comparisons of enzymatic activities. The
specific activities of E. coli BE WT and Nd1-112 branching enzymes were 1100 U/mg and
650 U/mg, respectively. The activity of E. coli glycogen synthase was measured under
standard conditions (3 7°C, 0.1 M biscine pH 8.0) and the conditions used for a-glucan
synthesis (30°C, 0.1 M citrate pH 7.0), and corresponding specific activities were 540
U/mg and 450 U/mg, respectively. Phosphorylase a was measured in the enzyme’s non-
physiological synthesis direction and the specific activity determined as 12.5 mU/mg.
Although the substrates of glycocen synthase (ADP-glucose) and phosphorylase (Glc-1-
P) are different, their specific activities under substrate-saturated conditions may be
directly compared, as both assays measure the incorporation of glucose into an existing
glycogen primer. With decreasing substrate concentrations the comparison of glycogen
synthase and phosphorylase a enzymatic activities becomes meaningless due to the
reversibility of the phosphorylase a reaction (equilibrium at 70 % towards synthesis at pH
7.0 (17)), whereas the reaction of glycogen synthase at pH 7.0 is essentially irreversible
(18). Furthermore, the Km value of glycogen synthase for ADP-glucose (18) is much

lower than the Km value of phosphorylase a for Glc-l-P (17).

180

150

100
75

35
25

15

 

Figure 2. SDS-PAGE of purified enzymes for glycogen
reconstitution system. (A), E. coli wild-type branching enzyme;
(B), E. coli truncated branching enzyme; (C), E. coli glycogen
synthase; (D), rabbbit liver glycogen phosphorylase.

Loading was 2-4 pg total protein per lane.

181

Synthesis of or-D-glucan. Previous results indicated the in vivo ratio of activity
between E. coli branching enzyme and E. coli glycogen synthase to range from 4:1 to
10:1. We measured a specific activity of 0.51 U/mg and 0.083 U/mg in vivo for branching
enzyme and glycogen synthase, respectively, corresponding to an enzymatic activity ratio
of 6: 1. In order to measure the effects of different enzymatic ratios of the amount of in
vitro synthesized a-glucan, we monitored the incorporation of l4C-glucose into the
glycogen product (Figure 3). The rate of glucan synthesis for BB WT when in
combination with either glycogen synthase (Figure 3A) or phosphorylase a (Figure BB)
were very similar. In either case, at 10-fold the activity of branching enzyme compared to
glycogen synthase or phosphorylase the rate of glucan synthesis becomes independent of
branching enzyme in these conditions. At this point, approximately 45 % (1 unit GS/Pa)
or 30 % (0.5 unit GS/Pa) of the substrate has become incorporated into the growing
polysaccharide. The result obtained for Nd“ 12 shows a very different rate of or-glucan
synthesis when in combination with glycogen synthase (Figure 3C) than phosphorylase a
(Figure 3D). The concerted action of glycogen synthase and Nd1-112 yields much less
polysaccharide synthesis for a specific amount of branching enzyme used, than in the
case of BB WT. It is evident from the “lag” phase that there is no significant or-glucan
synthesis with low branching enzyme amounts, which can be partly compensated by
addition of more glycogen synthase. The kinetics from Figure 3C suggests that Nd 1. 1 12
may require a longer primer before transfer of chains can be initiated. After glycogen
synthase synthesizes a primer of sufficient length for Nd” 12, the simultaneous action of

branching enzyme and glycogen synthase will then lead to rapid glucan accumulation.

182

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

DPM °/o DPM B 04150
24m A 7 2411*
J .
. ‘ ‘__________A l
ZIID- // . 40 20010- 0 /0 -—-o \o - 4o
4 ‘ i 4 /O / ¢
. 1811)-
1am / .. a J 0 - m
1311). A 0/0 Ufa . tzID- / I”. I“. R. l
/
i 1 /0 / . 20 10; ‘ 20
am‘ m1
Cl
0 3 fl W r r T 0 O 1 fl 1 I t f r O
O 5 10 15 Z) O 5 1O 15 Z)
m’tsE uitsE
DPM % DPM %
240m. 1 240001 .
. Vvflv/V‘V 4 50 . 50
2(IID- I 2m. 0 O O ‘
1 / 40 l O /
V 1 . 40
18113- ‘ 4 16000. /
. /A/ \A O I ____...—--l \ ‘
12CID~ ‘/ 120m- /
I I
< V / O ‘
spool ‘ 2° amt ‘ 2°
. D /U _U #0 l
. v 0 ‘ ” 10 -
401) / ‘/ '1 W1 10
‘ A/U ‘ i? l
O 'Tfl_+ I v I zl/L'fl—v—rm—r—Fr'r—r—r-V' o O r - I r r w r Y I 0
0 5 10 20 I) 40 50 60 7O 0 5 10 15 20

Figure 3. Effect of different ratios of enzymatic activity on a-glucan yields.
Various amounts of branching enzyme were incubated with glycogen

synthase (GS) or phosphorylase A (PA) and the apprOpriate substrate for 30
min. 0.5 units GS (D); 1.0 unit OS (A); 2.0 units GS (V); 0.5 units PA (I);
1.0 unit PA (0). The incorporation of 14C labeled substrate is indicated
in DPM (left Y-axis) and as percentage of total substrate (right Y-axis).
A: BE WT and GS using ADP-glucose as substrate.

B: BE WT and PA using glucose-l-phosphate as substrate.

C: NdH .2 and GS using ADP-glucose as substrate.

D: Nd“ ,2 and PA using glucose-l-phosphate as substrate.

183

This theory may be verified by adding small amounts of malto-oligosaccharides as a
primer for the reaction. Figure 3D shows that in combination with phosphorylase, NdH .2
leads to rapid glucan synthesis. The absence of a “lag” phase here due to the requirement
of Nd” 12 for a long primer, is most likely due to the presence of glucan primers which is
invariantly found in the commercial preparations of purified phosphorylase a.

The synthesis of or-glucan at various incubation times using 10 units of branching
enzyme activity and 1 unit of GS or Pa activity is shown in Figure 4. Using BE WT or
NdH 12 in concert with glycogen synthase led to 80 % conversion of the substrate ADP-
glucose into the polysaccharide in 3 hrs (Figure 4A). There was a late on-set of a-glucan
synthesis for Nd“ 12 compared to BE WT, consistent with the requirement of a longer
oligosaccharide primer for the truncated branching enzyme. a-glucan synthesis from BE
WT or Nd“ 12 in combination with phosphorylase a reached the maximum conversion of
Glc-l-P to the polysaccharide product in 2 hrs under these conditions (Figure 4B). The
observed equilibrium point of 60 % towards synthesis direction of phophorylase a is

close to the theoretical value (70%) at this particular pH (17, 19).

Structures of synthesized or-glucans. We investigated the effects of incubation
time on the structure of the synthesized a-glucan structures, while keeping the amounts
of branching enzyme (10 units) and GS/PA (1 unit) constant (Figure 5). Under these
conditions, the synthesis of a-glucan proceeded for l, 3, or 16 hrs and the resulting

polysaccharide was debranched and analyzed by HPAEC as described in materials and

methods. The PAD chromatogram for each polysaccharide structure was integrated and

184

 

 

 

 

 

 

40000- j90
35000: A /° ‘80
' I/Iéa/ ‘
soooo- / /0 :70
25000- 0 160
20000- / 450
- C)
E . / J40
0 15000- 0 +30
I / / :
10CKX31 C) _ 2f)
. I /
5000‘! o 410
Glycvr/I ' I ' j *fifi T ' n ' I ' I ' I ' 0
o 20 4o 60 so 100 120 140 160 180 200
min
35000- ~70
' 1
ENJOCX)- IEEE; ‘2:::::::£]::==-—{)-==:::ii _ E“)
4 :::==== .
250004 79 450
20000- /O -40
2 - O/ .
Esl '15CKX3n ‘ 3Cl
. .
10000- ~20
4 J
500‘” +10
otffi I j I ' I ' I r I ' I ' I ' I ' I Y o
0 20 4o 60 so 100 120 140 160 180 200
min

Figure 4. Effect of reaction time on glucan yields. 1.0 unit of glycogen
synthase (A)or phosphorylase A (B) was incubated with 10 units

BE-WT (I) or Nd (O) and the appropriated substrate. The

1-112
incorporation of 14C labeled substrate is indicated in DPM (left Y-axis)

or as percentage of total substrate (right Y-axis).

185

Figure 5. Analysis of transferred chains after 1 hr (Black), 3 hrs (White) and 16
hrs (Grey) of branching, using 10 units of BB activity and 1 unit of GS/PA
activity. The synthesized glucans were debranched and analyzed by HPAEC. The
relative peak areas were calculated using the sum of d.p. 1-40 as 100 %.

A - C: BE-WT and GS using ADP-glucose as substrate

D — F: NdH 12 and GS using ADP-glucose as substrate.

G - I: BE-WT and PA glucose-l-phosphate as substrate.

J — L: Nd“ 12 and PA glucose-l-phosphate as substrate.

186

 

2,

Relative peak area (%

D‘“ E

    
  
  

 

“ ‘ ‘ 0
5 10 15 2025 30 3540 5 1015 20 25 30 3540

H

   
  

510152025303540 510152025303540

J 8 K

510152025303540 510152025303540

0
510152025 30 35 40

 

 

 

51015 20 25 30 35 40

L

 

51015 20 25 30 35 40

Tra nsferred ch ai n length (d . p.)

187

the relative peak area of a particular chain length calculated, to allow for direct
comparison of the chain transfer patterns between various conditions. Figure 5 allows us
to draw several conclusions of variables affecting the or-glucan structures by construction
of peak area difference plots. Figure 6 illustrates the difference in or-glucan structures
based on the use of either wild-type or truncated E. coli branching enzyme, derived from
Figure 5C and Figure 5F.Nd1-m transfers fewer short chains (d.p. 4 - 16) and a higher
proportion of long chains (d.p. > 16) relative to the native E. coli branching enzyme. This
conclusion of Nd“ 12 transferring longer chains than BE WT is also reached when
branching enzyme is in combination with phosphorylase a (compare Figure 51 and Figure
5L), although the observed effect is less dramatic.

The effect of reaction time on the structure of the synthesized a-glucan is
illustrated in Figure 7. This difference plot, based on Figure 5B and Figure 5C, shows
that a longer time of incubation will increase the number of short chains of the
polysaccharide. It suggests the or-glucan structure, while being synthesized, is
continuously processed by branching enzyme, leading to a more compact and dense
molecule. The most dramatic effect on the polysaccharide structure is in the use of
glycogen synthase or phosphorylase a in glucan synthesis (compare Figure 5A-C with
Figure SG-I (BE WT), or Figure 5D-F with Figure SJ-L (Nd1-m)). A difference plot
between the polysaccharide structures of BB WT in combination with either glycogen
synthase or phosphorylase a clearly shows how using phosphorylase a causes a high
degree of branching (Figure 8). In fact, more than 85 % of all the chains are of d.p. S 15

after 16 hrs of glucan synthesis with BE WT and phosphorylase a (Figure 51).

188

 

 

2i
o\° .
0” ll II'll.l
a 1‘ I I...
Q) l
.E . [I I...I
“r I
:6 0"...IT I 17' I I . I I . I
g l) 10 1.9 25 30 35 4o
3; -1- /
<0 I
8. ‘ I/
.“.>: '2‘ ' /
*5 3 \ I.
E I
a; -3~ [I

l 'I /

.41 I

Chain length, d.p.

Figure 6. Difference plot showing the effect of BE-WT vs. NdH 12 on the glucan
structure. Difference in chain distribution between BE-WT and NdH 12 after 16
hrs in combination with glycogen synthase (GS) using a BE:GS ratio of 10:1. The
difference was the relative peak area from NdH 12 minus the relative peak area
from BE-WT for a given chain length. The sum of chains with d.p. 1-40 was set
as 100 %.

189

 

 

2.0-
. /"

c\°n 15‘ I' \
é .

1.0- .
§ I'-
a: ‘ \-
'6‘ 0.5- \
3 OO-lll', I I\F' ' ' . I T 1
g l) 5 10 15 \I 20 25 30 flu-I10
E 05- \I """II
a ‘ l I
2 4.0- ‘IIIII

-15- Chain length, d.p-

Figure 7. Difference plot showing the effect of reaction time on the synthesized

glucan structures. The difference was the relative peak area using BE-WT after 16
hrs minus the relative peak area using BE-WT after 3 hrs for a given chain length.
The ratio of BE:GS was 10:1. The sum of chains with d.p. 1-40 was set as 100 %.

190

 

 

0“..- I I 1* I . I . I ' “II-I...
20 2.5-“313' 35 4o

\ l.-
l I.
III.

I
N
l
I

 

Relative peak area difference, °/o

Chain length , d.p.

Figure 8. Difference plot showing the effect of phosphorylase A (PA) vs.
glycogen synthase (GS). The difference was the relative peak area using BE-WT
in combination with PA minus the relative peak area from BE-WT in combination
with GS for a given chain length. The ratio of BEzPA or BE:GS was 10:1 and the

incubation time was 16 hrs. The sum of chains with d.p. 1—40 was set as 100 %.

191

We also investigated the effects of different enzymatic ratios of activity on the
structure of the synthesized or-glucan structures (Figure 9). This was performed by use of
3, 6, or 20 units of branching enzyme (BE WT or Nd“ 12) with a constant amount of
GS/Pa (1 unit). Different ratios of branching enzyme and GS/Pa activity resulted in a
large effect of the polysaccharide structures. The difference plot of transferred chains
when using 3-fold or 20—fold excess of branching enzyme is shown in Figure 10. A
greater proportion of short chains (d.p. 4 —15) is observed with 20 units of branching
enzyme, meaning higher branching activity leads to a more branched structure. From
Figure 9 we can verify that Nd“ 12 transfers longer chains than BE WT (compare Figure
9A-C with Figure 9D-F). Also, we again observe that phosphorylase a causes a more
branched glucan structure compared to when glycogen synthase is utilized (compare

Figure 9A-C with Figure 9A-I), consistent with our observations earlier.

192

Figure 9. Analysis of transferred chains using BE:GS and BEzPA enzyme activity
ratios of 3:1 (Black), 6:1 (White) and 20:1 (Grey). Branching enzyme was
incubated with glycogen synthase or phosphorylase A and the appropriate
substrate for 16 hrs. at 30°C prior to debranching and analysis by HPAEC. The
relative peak areas were calculated using the sum of d.p. 1-40 as 100 %.

A - C: BE-WT and GS using ADP-glucose as substrate

D — F: NdH 12 and GS using ADP-glucose as substrate.

G - I: BE-WT and PA using glucose-l-phosphate as substrate.

J - L: NdH 12 and PA using glucose-l-phosphate as substrate.

193

Relative peak area (%)

 

    

w
0

 

   

510153253335400510152351154) 5101523533354)

E F

  
  
  

 

51015215303510 5101521533354)

 

510153533354)

J

51015205303540

K L

    

  

5101521533354) 5101521530354) 510153511354)

Transferred chain length (d.p.)

194

 

 

 

\° '\
o. 3. /.
Q)
a - \
93. 2 '1
$3 '1.
”5 ‘ \
3 . I
:a 1 \
"If: I l\
a) /
Q. 0 my 1 r T ! ' I ' I f l r I flli
g in 10 1 20 25 30 'm' 40
*5 \I I"
2:3 -1. ‘5 II.-
1 III...
-2.
Chain length , d.p.

Figure 10. Difference plot showing the effect of enzyme ratio on the synthesized
glucan structures. The difference was the relative peak area using a BE-WTzGS
ratio of 20:1 minus the relative peak area using BE-WTzGS ratio of 3:1 for a
given chain length. The incubation time was 16 hrs. The sum of chains with d.p.

1-40 was set as 100 %.

195

Discussion.

The work presented here is the first attempt to create a reconstitution system for in
vitro bacterial glycogen biosynthesis. In order to limit the number of variables and
simplify the data analysis we have not included the E. coli ADP-glucose
perphosphorylase in this study. Still, we are able to make several interesting conclusions
related to the parameters that affect the structure of the synthesized a-glucans. All our
experiments indicate that the truncated branching enzyme transfers longer chains than BE
WT, consistent with our conclusions from in vivo purified a-glucan structures and in
vitro branching of amylose (Chapter 5). The fact that the de novo oc-glucan synthesis
described in this study leads to distinct structures depending on BE WT or Nd“ 12 was
used, is additional proof of a difference in chain transfer pattern between these enzymes.
An unexpected observation was the late onset of glycogen synthesis when the truncated
branching enzyme was combined with glycogen synthase. This brings up new questions
about the substrate required for NdH 12 branching. It would be of interest to investigate
further if the minimum chain length for NdH 12 branching differs from BE WT, and if this
is the origin of the distinct polysaccharide structures produced by the two enzymes.

Increasing incubation time not only affected the yield of the oc-glucan, but also
resulted in a more branched structure, as branching enzyme continuously processed the
polysaccharide. However, the effect of incubation time on the a-glucan structure did not
appear as large as the other parameters investigated (e. g. source of branching enzyme, GS
vs. PA and enzymatic activity ratios). The use of phosphorylase a instead of glycogen
synthase caused a much higher degree of branching. This is possibly due to the

reversibility of the phosphorylase a reaction, which may trim the overall a-glucan

196

structure. This argument is also supported by the fact that the difference between the BE
WT or NdH 12 polysaccharide structures was not as great when using phosphorylase a
compared to E. coli glycogen synthase.

It has been proposed that the chain transfer and substrate specificities of
branching enzyme are the primary determinants of the glycogen structure (20). However,
we observed that changing the ratio of enzymatic activity between branching enzyme and
GS/PA highly affected the ot-glucan product. This suggests the interplay of branching
enzyme and glycogen synthase is an important factor affecting the final polysaccharide
structure. We suggest the branching enzymes and glycogen synthase to be co-expressed
in E. coli and their expression levels regulated through using different promotor systems.
The activity levels could then be measured and the resulting a-glucan analyzed to see if
novel polysaccharide structures could be produced in viva.

Recent advancements in the study of enzymes modifying a-glucan have spiked a
renewed interest in the starch industry for biochemical processing of the raw materials.
We believe our result demonstrate that polysaccharides with quite distinct physical

properties can be manufactured by controlling a small set of parameters.

197

10.

11.

12.

13.

14.

15.

References.

. Preiss, J. (1984). Annu Rev. Microbiol. 38, pp. 419-458.

Espada, J. (1962). J. Biol. Chem. 237, 3577-3581.

. Preiss, J. (1996). Regulation of Glycogen Synthesis. In Escherichia coli and

Salmonella typhimurium: Cellular and Molecular Biology, second edition, Amer. Soc.
Microbiol., Washington, D.C. Vol. 1, pp. 1015-1024.

Greenberg, E. and Preiss, J. (1964). J. Biol. Chem. 239, 4314.

Boyer, C., Preiss, J. (1977). Biochem. 16, 3693-3699.

Barengo, R., Flawia, M., and Krisman, C. R. (1975). FEBS left. 53, 274-278.
Chrisman, C. R. and Barengo, R. (1975). Eur. J. Biochem. 52, 117.

Lomako, J ., Lomako, W. M., and Whelan, W. J. (1988). FASEB J. 2, 3097.
Kawaguchi, K., Fox, J., Holmes, E., Boyer, C., and Preiss, J. (1978). Arch. Biochem.
Biophys. 190, 385-397.

Cori, J. F. and Cori, G. R. (1939). J. Biol. Chem. 131, 397.

Binderup, K, Mikkelsen, R., and Preiss, J. (2000). Arch Biochem. Biophys. 377, 366-
371.

Furukawa, K., Tagaya, M., Inouye, M., Preiss, J ., and F ukui, T. (1990). J. Biol. Chem.
265, 2086-2090.

Takeda, Y., Guan, H.P., Preiss, J. (1993) Carbohydr. Res. 240, 253-263.

Holmes, E. and Preiss, J. (1979). Arch. Biochem. Biophys. 196, 436-448.

Smith, P.K., Krohn, R.I., Hermanson, G.T., Mallia, A.K., Gartner, F.H., Provenzano,
M.D., Fujimoto E.K., Goeke, N.M., Olson, B.J., Klenk, D.C. (1985) Anal. Biochem.

150, 76-85.

198

16. Laemmli, U. K. (1970) Nature 227, 680-685.

17. Madsen, N. B. (1991). in A study of Enzymes, (Kuby, S. A., Ed.), CRC Press, New
York, Vol II, pp. 139-158.

18. Fox, J ., Kawaguchi, K., Greenberg, E., and Preiss, J. (1976). Biochemistry 15, 849.

19. Cori, C.F., Cori, G.T., and Green, AA (1943). J. Biol. Chem. 151, 39.

20. Guan, H. P., Li, P., Imparl-Radosevich, J ., Preiss, J. and Keeling, P. (1997). Arch.

Biochem. Biophys. 342, 92-98.

199

CHAPTER 7

CRYSTALLIZATION OF ADP-GLUCOSE PYROPHOSPHORYLASE

Part of this data is published:
Binderup, K., Watanabe, L., Polikarpov, I., Preiss, J. and Ami, R. K. (2000)
“Crystallization and Preliminary X-ray Analysis of the Catalytic Subunit of Potato Tuber

ADP-Glucose Pyrophosphorylase” Acta Cryst. D 56, 192-194.

200

Abstract
ADP-glucose pyrophosphorylase is the key regulatory enzyme in the biosynthesis of
starch in plants and glycogen in bacteria. There is no three-dimensional structure of this
important enzyme to help explain its catalysis or complex allosteric behavior. Here we
describe the expression, purification, and crystallization of the enzyme from Escherichia
coli and the 50 kDa catalytic subunit from potato tuber. Two multiple anomalous
dispersion data sets were collected to 2.9 A and 3.5 A of E. coli ADP-glucose
pyrophosphorylase from crystal forms belonging to the space group P222 and P622,
respectively. Furthermore, native data sets from two different potato small subunit crystal

forms have been obtained.

201

Introduction

ADP-glucose pyrophosphorylase (ADPGlc PPase, EC 2.7.7.27) catalyzes the
synthesis of ADP-glucose from ATP and glucose-l-phosphate, releasing pyrophosphate
as a product (1). This enzyme plays the key regulatory role in the biosynthesis of starch
in plants (2) and of glycogen in bacteria (3). ADPGlc PPase from all sources is found to
be allosterically controlled, but the specificity toward the effector depends upon the
source of enzyme. Enzyme from enteric bacteria is activated by fructose 1,6-bisphosphate
(FBP) and inhibited by adenosine monophosphate (4), whereas the enzyme from almost
all higher plant sources is activated by 3-phosphoglyceric acid and inhibited by
orthophosphate (5).

ADPGlc PPase from all sources is found to exist as a tetramer in solution. The
enzyme from enteric bacteria is homotetrameric (6), whereas the enzyme from higher
plants is comprised of two subunits, forming a 0962 structure (7). The enzyme from
Escherichia coli has been cloned and expressed using the M13mp18RF bacteriophage
(8,9). A combination of chemical modification and site-directed mutagenesis studies has
identified several amino acids important for catalysis or allosteric regulation of the E. coli
ADP-Glc PPase (10). The cDNAs of two potato (Solanum tuberosum L.) tuber ADPGlc
PPase subunits have been cloned from separate genes (11). The subunits have dissimilar
sizes of 50 kDa and 51 kDa and display an overall protein sequence identity of 59 %. The
amino acid sequence identity of ADPGlc PPases from different plant sources ranges
between 85-90 % for the small subunit and 50-60 % for the large subunit. The enzyme
from potato tuber exists physiologically as a heterotetramer (azfiz ). However, the

recombinant smaller subunit (50 kDa) forms an enzymatically active homotetramer when

202

expressed in the absence of the large subunit (12). The small subunit expressed alone has
higher enzymic activity with lower affinity for the activator 3-phosphoglyceric acid and
higher sensitivity toward orthophosphate inhibition. In contrast, the large subunit is not
by itself active. Furthermore the small subunit has proved more stable upon storage and
easier to purify in large quantities. The above observations have led to the suggestion that
the plant ADPGlc PPase small subunit functions as the catalytic subunit whereas the
large subunit is involved in allosteric regulation.

Despite being extensively studied for decades, there is not a three-dimensional
structure available for ADP-Glc PPase from any organism. Earlier attempts to determine
the structure of ADPGlc PPase from E. coli were unsuccessful due to the extremely rapid
decay and the fact that crystals diffracted to low resolution (13). Here we present initial
crystallization and X-ray diffraction analysis of the E. coli and potato tuber small subunit

ADPGlc PPases.

203

Materials and methods.

Reagents. 32PPi was obtained from DuPont-New England Nuclear. l4C-Glc-l-P
was from ICN. Chemicals for protein crystallization were obtained from Hampton
Research including Crystal Screen I and II. All other reagents were of the highest quality
available.

Plamids and bacteriophages.

pMLIO. pMLlO is an expression vector containing the mature potato tuber small
subunit ADP-Glc PPase gene under control of the lac promotor (12). It has kanamycin
resistance and ori-pACYC from pACYC177 (Novagen).

pMONI 7336. pMON17336 is an expression vector derived from pBR327 that
contains an insert encoding the large subunit of potato tuber ADP-Glc PPase (14). It has
the spectinomycin/ streptomycin (Spe/Str) resistance gene from Tn7, ori-327 and the
PrecA and GlOL expression cassette (15).

pGLGC. The WT E. coli ADP-Glc PPase gene (glgc) was subcloned from the
plasmid pOP12 (l6) and inserted in the NcoI/SacI sites of pET28. This was achieved by
PCR using the primers 5’-GGAGTTAGCCATGGTTAGTTTAGAGAAGAACG-3’

(J P403) and 5’-GAACATACATGGTCGACCTGCATTATCGCTCC-3’ (JP404) and
pOP12 as DNA template.

M13mp18RF. Previous work provided the mutant E. coli ADP-Glc G336D gene
inserted in M13mp18RF bacteriophage (17). The bacteriophage contains the coding
region of the respective genes as well as the promoter regions necessary for expression.

Expression of E. coli ADP-Glc PPase. E. coli strain BL21(DE3) was

transformed with pGLGC for expression of WT E. coli ADP Glc PPase. 300 ml of

204

overnight cultures was added to a Biostat E10 ferrnentor containing 12 1 fresh LB with 50
pg - ml'I kanamycin. Cells were grown at 37°C at 200 rpm to A600 = 0.5 before induction

with 0.5 mM isopropyl-D-thiogalactoside (IPTG). The incubation was continued at 25°C

for 16 hrs before harvesting by centrifugation.

For production of seleno-methionine substituted E. coli WT ADP-Glc PPase, The
pGLGC plasmid was transformed into E. coli strain CT-13, kindly provided by Dr.
Honggao Yan, Michigan State University. E. coli strain CT-13 is a methionine-auxotroph
BL21(DE3) derivative and tetracyclin resistant. The culture medium (minimal medium)
consisted of IX Eagle Basal medium (GibcoBRL), 0.2 % glucose, 0.5 mM MgSO4, 100
uM CaClz, 10 uM ZnClz, 10 uM FeCl3, 0.4 mg/l thiamine, 4 mg/l d-biotin, 50 pg/ml
kanamycin, and 10 pg/ml tetracyclin. Cells from 500 ml minimal medium overnight
culture containing 100 pig/ml L-methionine were washed and added to a Biostat E10
fermentor containing 12 1 fresh minimal medium with 50 ug/ml selenium-methionine
(Se-met, Sigma). The cells were grown at maximum aeration with 200 rpm stirring at
37°C to A600 = 0.5 before addition of 1 mM IPTG and selenium-methionine to 100
ug/ml. After 15 hrs induction at 25°C, cells were harvested and stored at -80°C.

The expression of the ADP-Glc PPase mutant G336D in the E. coli strain G6MD3
was achieved as described earlier by Meyer et al (17).

Purification of E. coli ADP-Glc PPases. Purification of the WT and mutant
G336D enzyme was performed as described earlier (17). Cells were resuspended in
buffer (50 mM glycylglycine pH 7.5, 1 mM EDTA and 5 mM DTT) before disruption by
sonication. Afier centrifugation, the supernatant was subjected to a heat treatment at 60°C

(WT) or 55°C (G336D) for 5 min and re-centrifuged. The resulting supernatant was taken

205

to 60 % saturation of ammonium sulfate and centrifuged. The pellets were dissolved in
dialysis buffer (50 mM glycylglycine pH 7.5, 1 mM EDTA, 2.5 mM DTT, 5% glycerol)
and dialyzed against 2 x 2 l of the same buffer. The protein was applied on a DEAE
fractogel colurrm (EM Science) and washed with buffer (30 mM K-phosphate pH 7.5, 5
% glycerol, 1 mM DTT) followed by a wash with 60 mM K-phosphate pH 7.3, 100 mM
KCl, 5 % glycerol and 1 mM DTT (Buffer A). The enzyme was eluted with a 1 1 linear
gradient of buffer A to 200 mM K-phosphate pH 7.0, 300 mM KCl, 5 % glycerol and 1
mM DTT. Fractions containing pyrophosphorolysis activity were pooled and taken to 60
% ammonium sulfate. Afier centrifugation, the pellet was resuspended in buffer B (50
mM Tris-Cl pH 7 .5, 1 mM EDTA and 1 mM DTT) and applied on a MonoQ 16/ 10
column. The protein was eluted with a linear gradient (400 ml) of buffer C to 40 % buffer

C (as buffer B + 1 M KCl). The purified protein was concentrated by ultrafiltration and
stored at —80°C.

Expression and purification of potato tuber a4 and azflz. E. coli strain
AC70R1-504 deficient in the endogenous E. coli ADP-Glc PPase (gIgC‘) was used for
expression of both the small subunit and heterotetramer of the potato tuber enzyme as
described earlier (14). For expression of the small subtmit, kanamycin (25 ug/ml final)
and 0.5 mM final concentration IPTG was used for selection and induction of the pMLlO
plasmid, respectively. In case of expression of the heterotetramer, both kanamycin (25
1.1ng final) and spectinomycin (70 pg/ml) was used for selection of pMLlO and
pMONl7336, respectively. The two plasmids were coordinately induced by addition of
IPTG (0.5 mM final) and nalidixic acid (5 ug/ml final), but apart from these differences,

the expression and purification methods were identical (12).

206

Overnight cultures of 5 ml selective LB media was diluted 1:1000 v/v into flasks of 1 1
fresh LB and grown at 37°C until OD600 reached 1.0-1.2. At this time expression of the
ADP-Glc PPases was induced as described above, and the cultures allowed continuos
shaking for an additional 40 hrs at 25°C. The cells were harvested by centrifugation,
resuspended in buffer A (50 mM HEPES pH 7.5, 5 mM MgC12, 1 mM EDTA, 20 %
sucrose) and disrupted by sonication. The crude extract was subjected to heat treatment at
65°C for 5 min and centrifuged. The supernatant was applied on a DEAE fractogel
column and eluted with a linear gradient from buffer A to 40 % buffer B (as buffer A + 1
M KCl). Active fractions were pooled and taken to 60 % saturation of ammonium sulfate.
The resulting pellet was dissolved in buffer C (20 mM Bis-Tris-propane pH 7.0, 5 mM
K-phosphate, 5 mM MgC12, 1 mM EDTA, 10 % sucrose) and dialyzed against buffer C.
The sample was applied on a MonoQ 10/10 column and eluted with a linear gradient of
O-O.5 M KCl in buffer C. Active fractions were concentrated by ultrafiltration, diluted 1:2
with 2 M K~phosphate pH 7.0 and applied on an aminopropyl-agarose column (C-3
column, 1x12 cm). The enzyme was eluted with a gradient of 1—0.05 M K-phosphate
buffer (pH 7.0). The purified enzyme was concentrated by ultrafiltration and stored at -
80°C.

ADP-Glc PPase assay. Pyrophosphorolysis of ADP-Glc was determined by the
formation of [32P]ATP from 32PPi. The reaction was done in presence of 3-PGA or FBP

according to the source of enzyme (Table 1).

207

Table l: Pyrophosphorolysis assay of ADP-Glc PPase.

 

 

Potato tuber (x4, 0:232 E. coli WT, G336D
Buffer 80 mM glycylglycine pH 8.0 40 mM HEPES pH 7.0
Bovine serum albumin 0.2 mg/ml 0.4 mg/ml
NaF 10 mM 5 mM
MgC12 8 mM 8 mM
ADP-Glc 2 mM 0.8 mM
32PPi (1—2 x103 cpm/nmol) 1.5 mM 2 mM
Dithiothreitol (DTT) 3 mM - not added -
3-PGA 1 mM - not added -
FBP - not added - 1.2 mM

 

All concentrations are in 250 pl final volume.

The reactions were initiated by addition of an appropriate amount of enzyme and after 10
min incubation at 37°C, the reaction was terminated by addition of 3 ml cold 5%
trichloroacetic acid. The [32P]ATP formed was measured as described by Morell and
coworkers (7). One unit of ADP-Glc pyrophosphorolysis activity is defined as the
amount of enzyme forming 1 pmol of ATP/ min under the conditions described.

Protein assay. Protein concentration was determined using bicinchoninic acid
reagent (18, Pierce) with bovine serum albumin as the standard.

Protein crystallization. Initial crystallization experiments were set up at 25°C
and 4°C by the sparse-matrix approach using the hanging drop vapor-diffusion method
described by J anick and Kim (19).

X-ray diffraction analysis.

Rotating anode X-ray source: Crystals were harvested using the reservoir solution

and mounted in thin-walled glass capillary tubes or mounted in nylon cryo-Ioops. X-rays

208

were generated by a Rigaku RU-200 rotating anode generator operating at 100 mA and
50 kV and monochromatic CuKa radiation was obtained using the MSC/ Yale mirrors.
Crystals were cooled to 4°C during data collection. Diffraction intensities were measured
by utilizing an R-AXIS II imaging plate detector. Data was indexed with DENZO scaled
and reduced using SCALEPACK (20).

Syncrotron X-ray source. Crystals were mounted in nylon cryo-loops (Hampton,
CA) and stored in liquid N2. High resolution data were collected at the Advanced Photon
Source at Argonne on the Structural Biology Center ID-l9 beamline. Intensity data was
collected in a 3x3 (3072x3072 pixel) CCD area detector. The crystal to detector distance
was set to 220 mm and 160° of data was collected with an oscillation range of 05°.
Diffraction data was indexed and integrated using DENZO and scaled by the sofiware

SCALEPACK (20).

209

Results.

Expression and purification of ADP-Glc PPases. A subcloning of E. coli ADP-
Glc PPase into the pET28 expression vector system was performed in order to improve
the yield and purity of the protein, as well as to facilitate the production of selenium-
substituted protein. Initially, the E. coli ADP-Glc PPase gene was subcloned into pET23,
but the BL21(DE3) cells hosting this construct were unable to grow, due to expression of
the plasmid even in absence of IPTG (data not shown). Expression of ADP-Glc PPase is
known to greatly affect the growth rates of E. coli, as is seen by the lack of colonies when
E. coli cells hosting an ADP-Glc PPase gene are plated on solid medium containing
inducer, e. g. IPTG (M. Ballicora, personal communication). The selection of the double-
repressed pET28 expression vector system circumvented the problems of culture growth,
and the optimized expression conditions in LB media produced 44 mg/l of purified E.
coli ADP-Glc PPase, representing an 8-fold improvement compared to earlier expression
techniques based on M13mpl 8RF (17).

In order to crystallize ADP-Glc PPase, enzyme from both E. coli (wild-type and
mutant G336D) and potato tuber (a4 and oczfiz) were purified as described in materials
and methods. Table 2 summarizes the yields of single purification batches, many of
which were conducted several times in order to provide sufficient material of the desired
purity. The purity of the enzymes was typically 80-90 % as judged by SDS-PAGE
protein gels stained by Coomassie blue (Figure 1). Table 2 reflects the level of difficulty
associated with the particular protein purification, as the level of expression and required
effort varies significantly. The expression of potato tuber ADP-Glc PPase is relatively

low with 0.4 % and 4 % of the total protein for the on and a282, respectively.

210

150

100
75

50

35
25

15

A B CD E F

 

 

Lat-ix'm'vsi

 

 

Figure 1. SDS-PAGE of purified enzymes for protein crystallization.

Loading was 13 pg protein per lane. A: E. coli BE WT; B: E. coli
truncated BE; C: E. coli ADP-Glc PPase; D: E. coli ADP-Glc PPase
G336D mutant. E: potato a4 ADP-Glc PPase; F: potato OLZBZ ADP-

Glc PPase.

211

Table 2. Summary of single protein purifications.

 

 

Protein Volume Purification folda Recoveryb Yieldc
E. coli ADP-Glc PPase, WT 4 liters 3 6O % 175 mg
E. coli ADP-Glc PPase, G336D 12 liters 38 39 % 40 mg
Potato tuber a4 30 liters 250 20 % 8 mg

Potato tuber (12132 30 liters 25 10 % 10 mg

 

aPurification fold indicates the ratio of specific activity of purified protein to that of the
crude extract. bRecovery is number of units of purified protein relative to that of crude
extract. cYield is the final mass of purified protein.

Selenium-substituted E. coli ADP-Glc PPase was produced by the methionine
auxotroph E. coli strain CT-13 hosting the pGLGC plamid. Afier purification, a total of
51 mg of Se-Met substituted protein was obtained from 4 liters. Mass spectrometry of
native and Se-met protein suggested the protein was fully substituted (Table 3).

Furthermore, HPLC analysis indicated the complete absence of methionine in the

selenium-substituted enzyme (data not shown).

Table 3. Mass spectrometry analysis of E. coli ADP-Glc PPase.

 

Theoretical MW2| Measured MWa Deviation, %

 

Native enzyme 48740.76 Da 48779.16 Da 0.08
Se-Met enzyme 49397.22 Da 49434.76 Da 0.08

 

8 MW per monomer.

Crystallization of E. coli ADP-Glc PPase WT. Initial screening of the protein

either uncomplexed or in presence of 1.5 mM FBP was performed and several conditions

212

produced small crystals directly from Hampton Crystal Screen I and II. Optimization of
initial crystallization conditions including temperature, protein concentration and
crystallization reagents, led to the crystals depicted in Figure 2.

a. Crystal form ECWT#I was derived from the conditions reported previously
(13) and was found to difi‘ract to a maximal resolution of 3.0 A. The cell parameters and
space group (P212121) were identical to those reported by Mulichak and coworkers
(Figure 3). The crystals appeared after 2-4 weeks and were very sensitive to small
changes in conditions of the experimental set-up. Furthermore, the crystals frequently
appeared cracked or hollow and were very easy to disrupt during mounting. The crystals
decayed rapidly in the X-ray beam (1-2 frames), and cryo-cooling conditions were sought
in order to stabilize the crystal. By including 30 % v/v glycerol in the well solution, it
was possible to collect a native data set by cryo-cooling (-150°C) from a crystal of
dimensions 0.2x0.2x0.7 mm (Figure 3). The quality of the processed data set is very poor
with low redundancy and high R-factors even in the lower resolution shells. Given the
problems of obtaining good crystals and the quality of the data set, this crystal form was
not analyzed further.

b. Crystal form ECWT#2 grew very large (up to l.0x0.5x0.5 mm) by
macroseeding of smaller crystals. The crystal belongs to the hexagonal system (P622),
and diffracted to a maximum resolution of 3.0 A. However, this particular crystal form
diffracts weakly and requires a powerful X-ray source for measuring data. Initially,
crystals decayed rapidly in the X-ray beam, but using 30 % glycerol as cryo-protectant
stabilized the crystals sufficiently during data collection. Crystals grown from selenium-

substituted protein were analyzed at the Argonne synchrotron source. The X-ray

213

 

Crystal form ECWT#I

Crystallization conditions:
5 mg/ml protein, 278 K
15% PEG 8000

0.1 M HEPES pH 7.1

0.2 M ammonium sulfate
size (mm): 0.2XO.2X0.7

    

’41: >' r

Crystal form ECWT#3

Crystallization conditions:
9 mg/ml protein, 298 K

20 % PEG 4000

5 % iso-propanol

0.1 M sodium citrate pH 6.2
size (mm) O.3X0.3XO.3

  
    

Crystal form ECWT#2

Crystallization conditions:
5 mg/ml protein, 278 K
1.6 M ammonium sulfate
0.1 M HEPES pH 7.5

0.2 M sodium acetate

size (mm) 0.5X0.5Xl .0

Crystal form ECWT#4

Crystallization conditions:
9 mg/ml protein, 278 K,
1.5 mM FBP

30 % PEG 1500

size (mm) 0.1X0.0SX0.05

Figure 2. Crystallization conditions of E. coli ADP-Glc PPase WT.

214

Space group: P212121
Unit Cell, A: a = 149.05, b = 149.33, c = 172.69

Summary of reflection intensities and R-factors by resolution shells.

 

 

 

Lower shell Upper shell Average Error Chi2 Linear Square
A A I R-factor R-factor
15.0 7.27 2079.9 134.2 1.042 0.044 0.044
7.27 5.89 630.7 84.8 1.052 0.124 0.104
5.89 5.18 522.2 99.4 0.973 0.124 0.123
5.18 4.72 584.5 121.0 1.036 0.153 0.128
4.72 4.39 550.5 139.6 1.155 0.197 0.161
4.39 4.14 429.7 157.4 1.085 0.267 0.208
4.14 3.93 278.3 184.0 1.160 0.469 0.358
3.93 3.77 245.3 225.0 1.114 0.642 0.502
3.77 3.62 205.1 265.2 0.978 0.798 0.622
3.62 3.50 192.2 287.7 1.042 0.000 0.812
All reflections 629.9 161.5 1.053 0.126 0.097

 

R linear = £(ABS(I-<I>))/2(I)
R square = 2((l-<l>)’)/ :02)
Chi2 = £((I-<I>)2)/ (error2*N/(N-1)))

Redundancy of measured data

 

 

 

Shell % of reflections with given number of observations
lower upper 0 1 2 3 4 5-6 7-8 9- 12 >12 Total
15.0 7.27 13.5 15.2 32.3 15.0 23.7 0.0 0.0 0.0 0.0 86.3
7.27 5.89 15.3 13.0 34.8 19.2 17.7 0.0 0.0 0.0 0.0 84.7
5.89 5.18 19.3 18.2 34.6 17.7 10.3 0.0 0.0 0.0 0.0 80.7
5.18 4.72 21.4 27.3 33.5 13.3 4.8 0.0 0.0 0.0 0.0 78.6
4.72 4.39 25.5 33.4 30.1 8.3 2.8 0.0 0.0 0.0 0.0 74.5
4.39 4.14 30.0 35.1 25.7 7.4 1.8 0.0 0.0 0.0 0.0 70.0
4.14 3.93 33.9 35.7 22.5 6.8 1.1 0.0 0.0 0.0 0.0 66.1
3.93 3.77 36.5 36.8 20.1 5.9 0.7 0.0 0.0 0.0 0.0 63.5
3.77 3.62 39.1 37.5 17.8 5.2 0.4 0.0 0.0 0.0 0.0 60.9
3.62 3.50 40.5 36.8 17.4 4.9 0.4 0.0 0.0 0.0 0.0 59.5

All hkl 27.4 28.8 26.9 10.4 6.5 0.0 0.0 0.0 0.0 72.6

 

Figure 3. X-ray data set of E. coli ADP-Glc PPase crystal form ECWT#l.

215

absorption spectrum confirmed the enzyme contained selenium and had a sharp
absorption edge. The cell parameters and processing statistics of the multiple anomalous
dispersion (MAD) data set is shown in Figure 4. The diffraction does not extend beyond
3.5 A, as seen by the sharp drop in percentage of measured reflections in the 3.57-3.41 A
resolution shell. The poor quality of this particular data has hindered finding the expected
64 selenium sites in the asymmetric unit, although this is sufficiently low for modern
software. However, as we observe a large variation in crystal diffraction limits and data
quality, more crystals should be analyzed at a synchrotron radiation source.

c. With crystal form ECWT#3, initial experiments using small crystals
(0.1x0.1x0.2 mm) revealed X-ray diffraction to about 3.3 A and a cryo-cooling condition
was found by including 10 % glycerol in the solution. Selenium-substituted protein
crystals (0.3x0.3x0.3 mm) were brought to the Argonne Advanced photon source and two
MAD data sets were collected to 3.0 and 3.3 A completion. Some decay occurred during
data collection, as the crystals initially showed a maximal diffraction of 2.6 A. The two
data sets were merged to enhance the completion and redundancy of the measured data
(Figure 5). This data set is very complete to 3.0 A with reasonable R-factors, but diffracts
poorly beyond the 3.00-2.90 A resolution shell. The crystal form has very large unit cell
dimensions and at least 250 selenium sites are expected in the asymmetric unit, posing
problems for calculating an electron density map.

(1. ECWT#4 was too small to test for X-ray diffraction. Here, the ADP-Glc PPase
is bound to the activator FBP included in the protein sample (1.5 mM final

concentration). Optimizing this condition as well as crystal seeding techniques in order to

216

Space group: P622

Unit Cell, A: a =b = 180.9, c = 225.4

# expected selenium sites = 64

Summary of reflection intensities and R-factors by resolution shells.

 

 

 

Lower shell Upper shell Average Error Chi2 Linear Square
A A I R-factor R-factor
99.0 6.82 917.5 38.1 3.658 0.093 0.135
6.82 5.41 233.8 16.4 1.699 0.140 0.146
5.41 4.73 249.0 19.6 1.656 0.159 0.336
4.73 4.30 259.0 22.7 1.634 0.175 0.189
4.30 3.99 151.5 23.9 1.334 0.284 0.702
3.99 3.75 109.4 27.5 1.294 0.437 0.449
3.75 3.57 100.4 30.7 1.254 0.538 0.000
3.57 3.41 55.3 32.4 1.088 0.882 0.777
All reflections 285.4 25.9 1.758 0.180 0.189

 

R linear = 2(ABS(I-<I>))/2(I)

R square = 2((I-<I>)2)/ 20’)
Chi2 = 2((I-<I>)2)/ (errorz‘N/(N-l»)

Redundancy of measured data

 

 

 

Shell % of reflections with given number of observations
lower upper 0 1 2 3 4 5-6 7-8 9-12 >12 Total
99.0 6.82 2.7 1.9 l 1.8 12.0 22.4 44.8 4.3 0.0 0.0 97.3
6.82 5.41 0.9 1.2 7.0 9.5 21.3 55.4 4.7 0.0 0.0 99.1
5.41 4.73 0.7 1.6 6.2 8.1 20.1 60.2 3.2 0.0 0.0 99.3
4.73 4.30 0.6 1.3 6.4 6.4 19.9 63.1 2.2 0.0 0.0 99.4
4.30 3.99 0.4 1.3 5.9 6.5 21.2 62.9 1.7 0.1 0.0 99.6
3.99 3.75 0.5 1.1 6.1 7.6 21.5 61.2 1.8 0.2 0.0 99.5
3.75 3.57 0.4 1.5 5.4 7.8 22.1 60.3 2.2 0.2 0.0 99.6
3.57 3.41 70.7 1.2 2.2 2.9 6.4 15.9 0.6 0.0 0.0 29.3
3.41 3.28 100 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
3.28 3.17 100 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0

All hkl 27.2 1.1 5.2 6.2 15.6 46.2 2.1 0.0 0.0 72.8

 

Figure 4. X-ray data set of E. coli ADP-Glc PPase crystal form ECWT#2.

217

Space group: P21212, Z = 16-20
Unit Cell, A: a = 154.4, b = 273.5, c = 209.1
Mosaicity = 0.339

# selenium sites = 250+

Summary of reflection intensities and R-factors by resolution shells.

 

Lower shell, A Upper shell, A Average, I a 1/6 Norm. R-factor

 

 

30.00 6.23 3033.1 104.2 29.0 0.031
6.23 4.95 1640.1 65.9 25.0 0.052
4.95 4.33 2614.4 95.1 27.0 0.049
4.33 3.93 1838.8 86.9 21.0 0.066
3.93 3.65 1341.7 86.7 15.0 0.095
3.65 3.44 920.1 86.7 10.0 0.128
3.44 3.27 580.9 86.2 6.7 0.192
3.27 3.12 394.2 86.1 4.5 0.279
3.12 3.00 292.8 86.2 3.3 0.354
3.00 2.90 220.5 87.1 2.5 0.461

All reflections 1266.5 87.0 14.5 0.085

 

R linear = £(ABS(I-<I>))/2(I)
R square = 2((I-<l>)2)/ :02)
Chi2 = £((I-<I>)2)/ (errorZ’N/(N-l )))

Redundancy of measured data

 

Shell % of reflections with given number of observations
lower upper 0 1 2 3 4 5-6 7-8 9-12 13-19 >19 Total

 

40.0 6.24 2.8 5.4 23.5 3.5 7.6 6.0 7.6 10.5 33.1 0.0 97.2
6.24 4.96 0.4 3.7 24.3 3.8 8.0 6.4 7.1 12.8 33.5 0.0 99.6
4.96 4.33 0.2 3.1 23.4 3.9 8.4 6.7 7.2 13.7 33.4 0.0 99.8
4.33 3.94 0.1 3.0 22.4 4.3 8.5 6.8 7.3 14.2 33.5 0.0 99.9
3.94 3.65 0.1 2.8 21.7 4.3 8.6 6.9 7.8 15.1 32.8 0.0 99.9
3.65 3.44 0.1 2.6 21.1 3.9 9.2 6.9 8.0 15.5 32.6 0.0 99.9
3.44 3.27 0.1 2.9 20.0 3.9 9.5 7.2 8.3 17.3 30.8 0.0 99.9
3.27 3.12 0.4 3.0 19.1 4.0 9.8 7.2 8.4 21.5 26.6 0.0 99.6
3.12 3.00 1.4 6.3 14.6 4.6 9.2 7.9 10.1 26.5 19.6 0.0 98.6
3.00 2.90 18.3 4.9 8.8 3.4 5.4 14.4 12.8 24.6 7.5 0.0 81.7

 

Allhkl 2.4 3.8 19.9 4.0 8.4 7.6 8.4 17.1 28.4 0.0 97.6

 

Figure 5. X-ray data set of E. coli ADP-Glc PPase crystal form ECWT#3.

218

produce larger crystals were unsuccesful, and thus the crystal form was not pursued

further.

Crystallization of potato tuber ADP-Glc PPase. The out form was screened for
crystallization conditions in the uncomplexed form or with Cry-ATP (0.3 mM final) or
ADP-Glc (3 mM final) included in the protein sample. Figure 6 depicts crystals from four
separate conditions.

a. Crystal form a4#1 was obtained after 1-2 months and grew as very large rods
(>1 mm in length) or as plates. This crystal form diffracted X-rays well (2.2 A maximum)
and found to be reasonably stable during data collection. A native data set was collected
to 3.4 A without cryo-cooling and the data is summarized in Table 4. Attempts to refine
conditions so crystals could be reproduced in a shorter time have been unsuccessful,
complicating further analysis.

b. The 014#2 crystal form is a reasonable candidate for solving the structure of a
potato ADP-Glc PPase small subunit. a4#2 diffracted X-rays well (2.8 A), grew large
(0.5x0.2x0.2 mm) in 1-2 weeks and was found to be stable during data collection.
Furthermore, a cryo-cooling condition has been obtained by including 20 % glycerol in
the well solution, which may prevent crystal decay altogether. A native data set has been
collected to 3.0 A (94 % complete) and is summarized in Table 4.

c. Crystal form a4#3 were obtained with ADP-Glc included in the protein sample.
These were obtained with relative ease in 1-2 days, but unfortunately diffracted X-rays

poorly (4.5 A) and were concluded to be unsuitable for X-ray diffraction analysis.

219

    

Crystal form a4#l

crystallization conditions:

5 mg/ml protein, 278 K,
0.3 mM Cir-ATP

30 % PEG 4000

0.2 Mammonium acetate
0.1 Msodium citrate pH 5.6
size (mm) 0.1XD.lXD.l

 

Crystal form a4# 3

Crystallization conditions:

5 mg/ml protein, 298 K

3 mM ADP-Glucose

ll % PEG 8000

0.1 Mmagnesium chloride

0.1 Msodium cacodylate pH 6.5
size (mm) 0.1XD.2XD.4

 

Crystal form a4# 2

Crystallization conditions:

5 mg/ml protein, 278 K

20 % PEG MME 2000

0.15 Mammonium sulfate
0.1 Msodium acetate pH 4.6
size (mm) 0.2XO.2XD.5

    

Crystal form (141% 4

Crystallization conditions:
5 mg/ml protein, 298 K
18 % PEG 8000

0.1 Mmagnesium chloride
0.1 Msodium cacodylate
pH 6.5

size (mm) 0.1XD.1XD.2

Figure 6. Crystallization conditions of potato tuber a4 ADP-Glc PPase.

220

Table 4. Data collection and processing statistics for potato tuber small subunit

crystal forms a4#1 and 014#2

 

 

Crystal form a4#1 014#2
Space group P41 P2
Cell constants (A) a=b=110.57 c=190. 14 a=80.06
b=138.84
c=92.20
B=112.40°
Maximum resolution (A) 2.2 2.8
Resolution of dataset (A) 3.4 3.0
Number of unique reflections 24429 30342
‘ Rmerge (%) 12.1 9.5
Completeness (%) 79 82
VM 2.9 2.4
Number of molecules per 4 4

asymmetric unit

 

a Rmcrgc=22|1(h)i - <I(h)>|/2<I(h)>; I(h) is the observed intensity of the ith
measurement of reflection h, and <I(h)> the mean intensity of reflection h

calculated afier scale.

221

(1. Crystal form a4#4 diffracted X-rays reasonably well (3.5 A), but we could not
produce crystals of a sufficient size by refining conditions or even seeding techniques.
Also, these crystals were very sensitive to handling and decayed rapidly during data
collection.

In addition, the potato tuber 01202 protein has been screened for crystallization
conditions. Although small crystals have been produced, these are yet too small for X-ray

diffraction analysis.

Crystallization of E. coli ADP-Glc PPase mutant G336D. G336D was screened
uncomplexed or with 0.5 mM FBP included in the sample for crystallization conditions.
Figure 7 shows an example of a crystal produced in absence of FBP. Despite a decent
appearance and reasonable size (0.2x0.2x0.2 mm) these crystal diffracted X-rays very
poorly (>7A) and hence were concluded to be unsuitable for diffraction analysis. In other
conditions, small crystals of G336D have been produced and these are currently being

optimized.

222

 

E. coli ADP-Glc PPase G336D

Crystallization conditions:
4 mg/ml protein, 298 K.
20 % PEG MME 550

0.1 M MES pH 6.5

10 mM zinc sulfate

size (mm): 0.2x 0.15x 0.15

Figure 7. Crystallization conditions of E. coli ADP-Glc PPase G336D.

223

Discussion.

It is evident from the results described above that obtaining a three-dimensional
structure of ADP-Glc PPase poses a challenging task. Despite numerous crystal forms of
both ADP-Glc PPase from E. coli and the potato small subunit, and a large effort in data
collection and processing, we have not yet been able to produce an electron density map
of the enzyme. However, we have obtained several results that could make a solution of
the structure eventually possible. This includes increasing the protein yield and purity of
the E. coli ADP-Glc PPase by subcloning the corresponding gene into a pET28
expression vector. As protein crystallization demands large amounts of purified material,
this is an important prerequisite for solving the structure. Furthermore, the pET28
expression system has enabled us to produce protein substituted with selenium
methionine with good yields, necessary for our strategy in obtaining phasing information.

The low level of potato on expression complicates its crystallization but
experiments to produce larger amounts of soluble protein have been unsuccessful
(Ballicora and Preiss, unpublished results). As there are no atomic coordinates available
for an ADP-Glc PPase, we are seeking phasing information by measurements of MAD
data obtained from selenium-methionine substituted crystals. However, frequently a 10-
fold reduction in protein expression is observed as a result of using the minimal media
required in producing selenium substituted proteins. This would render the production of
selenium-substituted potato on; enzyme exceedingly difficult. In fact, the limitation on
potato on protein yields leaves the E. coli ADP-Glc PPase currently the only candidate

for producing the first three-dimensional structure of the enzyme.

224

The two crystal forms P622 (ECWT#2) and P21212 (ECWT#3) of E. coli ADP-
Glc PPase have successfully been produced with selenium-substituted protein. For crystal
form P21212, this has resulted in a very complete 3.0 A MAD data set with a sharp
absorption edge for selenium. In most circumstances this information would be adequate
for constructing an electron density map. However, the high number of expected
selenium sites in the assymmetric unit (250+) has hindered the calculation of a map, due
to limitations in software capability. By obtaining an additional source of phasing
information (i.e. isomorphous replacement) it is plausible the selenium sites in crystal
form P21212 can be found, enabling the calculation of an electron density map.

Alternatively, the structure could be solved based on the hexagonal crystal form
of E. coli ADP-Glc PPase, as the lower number of expected selenium sites (64 sites)
makes locating their positions in the assymmetric unit possible. The weak diffraction of
this particular crystal form requires the most powerfill bearnline (SBC) at the Argonne
Advanced Photon Source in order to collect X-ray data of sufficient quality, which in this
case has hindered rapid progress. The ADP-Glc PPase is a crystallographic challenge but
the results presented in this report encourage us to continue working on solving its three-

dimensional structure.

225

10.

References.

. Espada, J. (1962). J. Biol. Chem. 237, 3577-3581.

Preiss, J. (1991). Oxford Surveys of Plant Molecular and Cell Biology, Vol. 7 edited

by B. Mifflin, pp. 59-114. Oxford.

. Preiss, J. (1984). Annu. Rev. Microbiol. 38, 419-458.

Preiss, J ., Shen, L., Greenberg, E. and Gentner, N. (1966). Biochemistry 5, 1833-

1845.

. Ghosh H. P. and Preiss, J. (1966). J. Biol. Chem. 241, 4491-4504.

Haugen, T. H., Ishaque, A. and Preiss, J. (1976). J. Biol. Chem. 251, 7880-7885.
Morell M. K., Bloom, M., Knowles, V. and Preiss, J. (1987). Plant Physiol. 85, 182-
187.

Baeker, P.A., Furlong, GE, and Preiss, J. (1983).]. Biol. Chem. 258, 5084-5088.
Hill, M.A., Kaufrnann, K., Otero, J ., and Preiss, J. (1991) J. Biol. Chem. 266, 12455-
12460.

Preiss, J. and Sivak, M. (1998) in Comprehensive Natural Products Chemistry (B. M.

Pinto, Ed.) Vol. 3, pp. 441-495, Pergamon Press, Oxford.

11. Nakata P. A., Greene, T.W., Anderson, J. M., Smith-White, B. J ., Okita, T. W. and

12.

13.

Preiss, J. (1991). Plant Mol. Biol. 17, 1089-1093.

Ballicora, M. A., Laughlin, M. J ., Fu, Y., Okita, T. W., Barry, G. F. and Preiss, J.
(1995). Plant Physiol. 109, 245-251.

Mulichak, A.M., Skrzypzak-Jankun, E., Rydel, T.J., Tulinsky, T., and Preiss, J.

(1988). J. Biol. Chem. 263,17237-17238.

226

14. Iglesias, A.A, Barry, G.F., Meyer, C., Bloksberg, L., Nakata, P.A., Greene, T.,
Laughlin, M.J., Okita, T.W., Kishore, G.M., and Preiss, J. (1993) J. Biol. Chem. 268,
1081-1086.

15. Olins, PO. and Rangwala, SH. (1989) J. Biol. Chem. 264, 16973-16976.

16. Okita, T.W., Rodriguez, R.L., and Preiss, J. (1981).]. Biol. Chem. 256, 6944-6952.

17. Meyer, C.R., Bork, J .A., Nadler, S., Yirsa, J. and Preiss, J. (1998) Arch. Biochem.
Biophys. 353, 152-159.

18. Smith, P.K., Krohn, R.I., Hermanson, G.T., Mallia, A.K., Gartner, F.H., Provenzano,
M.D., Fujimoto E.K., Goeke, N.M., Olson, B.J., and Klenk, D.C. (1985) Anal.
Biochem. 150, 76-85.

19. Jancarik, J. and Kim S-H (1991). Sparce matrix sampling; a screening method for
crystallization of proteins. .1. Appl. Cryst. 24, 409-411.

20. Otwinowski, Z. andMinor, W. (1997) Methods Enzymol. 276, 307-326.

227

CHAPTER 8

CONCLUDING REMARKS AND RESEARCH PERSPECTIVES

228

The work described above has aimed at increasing our understanding of the
structure-function relationships of branching enzyme and ADP-glucose
pyrophosphorylase. This objective has been pursued via X-ray crystallography to solve
the three-dimensional structures of both enzymes, as well as conventional biochemical
approaches for branching enzyme. Solving the crystal structures of branching enzyme
and ADP-glucose pyrophosphorylase currently represents the most important research
challenges in our laboratory. For branching enzyme, we believe the crystal form of the
truncated enzyme will enable us to achieve this goal. The quality of the native data set,
the availability of selenium-methionine substituted protein, and the sufficiently low
number of selenium sites in the asymmetric unit, all indicate we can solve the structure,
although much work clearly remains.

Based on sequence homology with the a-amylase family I would like to conclude
my thesis by discussion of a possible catalytic mechanism for branching enzyme which is
a variation of the proposed catalytic mechanism for CGTase (1; Figure 1). This
hypothetical branching enzyme mechanism utilizes Asp335 of region 2 (B4-H4 loop) and
Glu458 of region 3 (BS-H5 loop) conserved in the (a/B)s-barrel. The mechanism
proceeds via two half-chair oxo-carbenium transition states (T81 and TS2) and a covalent
intermediate with Asp335. A covalent intermediate involving this Asp has been observed
in the crystal structure of CGTase by soaking crystals in 4-deoxy-maltotriosyl-a-flouride
(1). The reaction of all enzymes in the or-amylase family may have this first part of the
reaction in common. However, the last part of the reaction differs between the amylolytic

enzymes. In branching enzyme, for example, the nucleophile attacking the T82 half-chair

229

Figure 1. Proposed catalytic mechanism for branching enzyme. The two
conserved amino acids Asp335 and Glu458 of regions 2 and 3, respectively, act as
catalytic residues. Initially, the a-1,4 glucosidic linkage is cleaved by transfer of a
proton from Glu458. The transition state TSl involves an oxo-carbenium in a
half-chair conformation. Asp335 forms a covalent linkage with the C-1 as an
intermediate. The next transition state (TS2) involves a second half-chair oxo-
carbenium. Glu458 acts here as a base and the new (1-1,6 linkage is formed by

nucleophilic attack of the 6-OH oxygen.

230

 

 

 

T'—
C/ Asp335
0‘ \ _ / Asp335
OH 8) 0 -> 0.40\
OH 0'
HO OH O
0"? 0m \(‘9
3” HO OH HO OH OH
/o—H O
...—-c\ V 0' HO
\ /
Glu458 0 ”0* HO 0”
Glu458 O
Substrate / / TSl
/ Asp335 / Asp335
0‘C O""C\
OH 81 \o > OH 0'
O
Hm ‘~.@
OH HO OH
OH H) 0
0' HO O F
c/ o o
_....- /
\\ HO CH ._—-—C HO
O \\0 HO OH
Glu458 Glu458
Intermediate '1‘S2 _
0’ / Asp335
OH \0'
O
HO OH
O
O
Glu458 \\o ”0 0“

 

 

231

 

 

 

is oxygen of the 6-OH group, whereas it is the oxygen of 4-OH in CGTase and H20 in or-
amylase. The challenge is to explain how such a specificity is maintained in these related
enzymes.

For branching enzyme in particular, dual-labeling studies have shown that transfer
between two glucan molecules is possible (2). In other words, the donor and acceptor on-
glucans are not necessarily from the same molecule. It is therefore possible branching
enzyme has an additional or-glucan binding site. Obviously, it would be desirable to
obtain a crystal structure of branching enzyme with bound substrate, or substrate
analogue to answer some of these questions. The BAYe4609 inhibitor described in this
thesis may be utilized in such crystallization trials. As BAYe4609 contains a mixture of
varying size glucan-like chains, purification or even chemical synthesis of BAYe4609
may be required.

Another approach involves creating the double mutant D335N-E458Q of the E.
coli branching enzyme and soak crystals with commercially available arnyloses of high
purity. The approach involving a double Asp-Glu mutant has previously been successful
in obtaining an enzyme-substrate complex of CGTase, although the enzyme had a very

low turn-over (3).

A considerable work effort has concentrated on crystallization of ADP-glucose
pyrophosphorylase. Despite that many different crystal forms have been obtained of both
the E. coli enzyme and the potato tuber small subunit, the majority of these crystals have

proved unsuitable for X-ray diffraction analysis. Low protein yields have also posed

problems, in particular for the potato tuber on; enzyme. The expression of the E. coli

232

ADP-glucose pyrophosphorylase in a pET expression vector system has not only
improved the yield, but also facilitated the production of selenium-methionine protein.
This has led to MAD data sets of two E. coli ADP-glucose pyrophosphorylase crystal
fbnns.

Although diffracting X-rays reasonably well, the many selenium sites of E. coli
ADP-glucose pyrophosphorylase crystal form P21212 hinders the construction of an
electron density map. There are at least two different strategies for obtaining an E. coli
ADP-glucose pyrophosphorylase structure: A) Isomorphous replacement by heavy atom
soaks of P21212 crystals can provide additional information, thereby assisting in finding
the selenium sites. B) The structure can be solved initially with crystal form P622, before
working with the higher-resolution data of the P21212 crystal form.

The three-dimensional structure of E. coli N-acetylglucosamine l-phosphate
uridylyltransferase (GlmU) does provide us with some structural information about the
pyrophosphorylase domain. Although the atomic coordinates are not available, many
experiments based on the GlmU structure combined with sequence alignments, can be
designed. One example is site-directed mutagenesis of Asp142 in E. coli ADP-Glc PPase,
already described in Chapter 1. However, for understanding the complex allosteric
regulation and subunit interactions of ADP-glucose pyrophosphorylase, obtaining a
structural model remains an absolute requirement, and therefore an important research

challenge.

233

References.
1. Uitdehaag, J .C.M., Mosi, R., Kalk, K.H., Van der Veen, B.A., Dijkhuizen, L.,
Withers, S.G. and Dijkstra, B.W. (1999) Nat. Struc. Biol. 6, 432-436.
2. Knegtel, R.M.A, Strokopytov, B., Penninga, D., Farber, O.G., Rozeboom, H.J.,
Kalk, K.H., Djikhuizen, L., and Dijkstra, B.W. (1995). J. Biol. Chem. 270, 29256-29264.
3. Nielsen, A.N., Blennow, A., Nielsen, TH, and Moeller, B.L. ( 1998).

117, 869-875.

234