. _ .2 .3...
.533. . . «3m... £3“... an.
..

u

$1 5‘12. .maflafi‘

.. g. .m? E... “ma...
. 3....

v.
“'1‘“

w , a.
#2.; way:
. ‘2. .P , u...

«2‘!

. it .
.7I\..
1?...) v.1

ytivxia:
3 5”."

5.0.1. .. 1......13111...
wn-I 1952111§pfl
. .33!!! I
y . ‘ hf
r gym...) L.

. .
«trurvdw 3. n5
’211h1».vv2..h.v. :nn
. 395.11”? .u

0 {~33

 

..z...m...u!...n£un
1....” vflnfiumu ha
, $.31. 3‘ inn
.0 O: F . .
.. 2..
vvpar...¢....rii...h flan
‘ ‘2’!" 3.". t.

It.“ V‘...
a . ‘33:}... 0.:
2 i!h.l.£l\n¥1~ '3.
*‘x‘t‘lxzz ‘1
. A.) In 9.1:?!
1 3a a $3..
3 .

 

 

 

§.. .127“... 1...;
a. 25.1.39. i.
z .13.}:
ivlvti
3:1‘ xvi:
lizi‘lsxaafv

 

 

.s : : :::

 

mi. :59 {:45 ELLE... 2 ._ 41...... ‘ 3. Li»:

T1515 15
Z
20€>7’

This is to certify that the

dissertation entitled

X-RAY CRYSTALLOGRAPHIC STUDIES OF MIP SYNTHASE

presented by
Adam J. Stein

has been accepted towards fulfillment
of the requirements for

PhD degree in Chemi S try

 

xx

Major professof #

 

Date ”[3,0]...

 

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

 

 

LIBRARY

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

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6/01 cJCIRC/DateDuepss-p. 1 5

X-RAY CRYSTALLOGRAPHIC STUDIES OF MIP SYNTHASE
BY

Adam Joshua Stein

A DISSERTATION
Submitted to
Michigan State University
In partial fulfillment of the requirements
for the degree of
DOCTOR OF PHILOSOPHY
Department of Chemistry

2002

ABSTRACT
X-RAY CRYSTALLOGRAPHIC STUDIES OF MIP SYNTHASE
BY

Adam Joshua Stein

1L-myo-inositol 1-phosphate synthase catalyzes the conversion of D-
glucose 6-phosphate to 1L-myo-inositol 1-phosphate. This represents the first
step in the biosynthesis of all inositol-containing compounds. The reaction
involves a complex series of transformations including oxidation, intramolecular
aldol condensation, and reduction, all of which occur in the same active site of
the enzyme. Two crucial crystal structures of MIP synthase have been solved: (1)
the holo form and (2) an enzyme/inhibitor complex with the inhibitor, 2-deoxy-
glucitol 6-phosphate. While 58 amino acids are disordered in the holo form of the
enzyme near the active site, the inhibitor nucleates the folding of this domain
serving to completely encapsulate it within the enzyme. Three helices and a long
beta strand are formed in this process. A mechanism for this binding is proposed
that first involves the binding of the inhibitor to the ordered part of the enzyme
followed by the nucleation and folding of the disordered region. We further
postulate a mechanism for the conversion based on the structure of the inhibitor-

bound complex.

To my wife, for all of your love and support in finishing this part of my life.

ACKNOWLEDGEMENTS

I would like to acknowledge all the people that have helped me get to this
point. First and foremost, I would like to thank my advisor, Dr. James H. Geiger.
Without your intellect and guidance, this thesis would not be possible. Thanks for
all the times spent on the graphics trying to figure out the final resting place of the
inhibitor. In addition, thanks for helping me become a better scientist and more
importantly, a good crystallographer.

I would also like to acknowledge the support of my coworkers, past and
present members of the Geiger group, for their friendship, helpful discussions,
and encouragement. More specifically, I would like to sincerely thank Marta C.
Abad and Stacy Hovde for all the lunches, advice, and discussions involving our
structures. To both of you, good luck with all your endeavors in whatever road
you choose to travel.

I am deeply indebted to Dr. Alexander Tulinsky and Dr. Jorge Rios for
steering me in the right direction. Doc, your knowledge and insight have inspired
me. Jorge, you showed me the essentials and for that, I will always remember
you.

I would also like to say thanks to Dr. John W. Frost and Chad Hansen for
supplying the inhibitors for this study and for the verification of MIP synthase
sequence. Without you guys, we would probably still be wondering where the

remaining fifty-residues were.

Financially, I would like to thank both the NSF and the MSU IRGP for
funding this project. Without their support, this project would not have been
possible.

Finally, I would like to thank my family and most importantly, my wife.
Without you Shannon, life would be empty. Your help and support over the years
have enabled me to make this possible and I want you to know that. Whenever I
had an issue or a problem, you were there for me to lean on. Just remember that

I will love you “always and forever.”

TABLE OF CONTENTS

LIST OF TABLES ............................................................................................... ix
LIST OF FIGURES ............................................................................................. x
LIST OF ABBREVIATIONS .............................................................................. xiii
CHAPTER 1
INTRODUCTION ................................................................................................ 1
BACKGROUND ............................................................................................ 1
General Overview .................................................................................... 1
Inositols and Cellular Signaling ................................................................ 2
The Biological Role of MIP Synthase Activity ........................................... 5
MIP Synthase Homology, Classification, and Characteristics .................. 8
Mechanistic Aspects of MIP Synthase ................................................... 17
Remaining Mechanism Issues ............................................................... 22

AN IN DEPTH LOOK INTO PREVIOUS INHIBITOR STUDIES AND

THEIR POTENTIAL ROLES IN MIP SYNTHASE CATALYSIS ................... 28
An Introduction to Previous Inhibitor Studies .......................................... 28
Does the Enzyme Bind the Cyclic or Acyclic Form of Glucose 6-
phosphate? What is the Anomeric Configuration of the C1 carbon

if the Enzyme Does Bind the Cyclic Form? ............................................ 28
What is the Nature of the Various Proton Donors and Acceptors
in the Reaction Pathway? ....................................................................... 31
What is the Nature of the Enzyme-catalyzed Aldol Cyclization? ............ 33
How Does MIP Synthase Bind NAD? ..................................................... 34
GOAL OF THIS THESIS .............................................................................. 36
CHAPTER 2
THE NATIVE MIP SYNTHASE STRUCTURE .................................................. 37
EXPERIMENTAL ........................................................................................ 37
Crystallization and Data Collection Analysis of the Native
MIP Synthase ......................................................................................... 37
Crystallization and Data Collection Analysis of the Se-Met
MIP synthase ......................................................................................... 41
A Primer in MAD Phasing ....................................................................... 43
MAD Phasing Experimental Results ...................................................... 47
Refinement Analysis for the Native Structure ......................................... 50

vi

STRUCTURAL RESULTS AND DISCUSSION ........................................... 52

The Structure of the M IP Synthase Tetramer ......................................... 52
The Structure of the MIP Synthase Dimer .............................................. 52
The Structure of the MIP Synthase Monomer ........................................ 56
NAD Binding ........................................................................................... 6O
MIP Synthase and its Unique Rossmann Fold ....................................... 65
Conclusions ............................................................................................ 65
MATERIALS AND METHODS .................................................................... 69
Expression, Purification, and Characterization for the Native
MIP Synthase ......................................................................................... 69
Overexpression and Purification of Se-Met MIP Synthase ..................... 73
Modifications to the Purification Protocol as a Result of Se-Met
Overexpression ...................................................................................... 76
Multiple Isomorphous Replacement (MIR) Experimental Results .......... 78
An Attempt to Solve the P21 Crystal Form ............................................. 83
CHAPTER 3
THE INHIBITOR/MIP SYNTHASE STRUCTURE ............................................. 85
EXPERIMENTAL ........................................................................................ 85
Crystallization and Data Collection Analysis for the Inhibitor-bound
MIP Synthase ......................................................................................... 85
Solving the Inhibitor Complex and Refinement Analysis ........................ 87
STRUCTURAL RESULTS AND DISCUSSION ........................................... 90
The Structural Changes in MIP Synthase that Occur Upon Inhibitor
Binding ................................................................................................... 90
The Interactions Between MIP Synthase and DgtoIP ............................. 97
Modeling of the Aldol Condensation and a Proposal for the
Mechanism of MIP Synthase ................................................................ 102
Is the New Domain Due to NAD or Substrate Binding? ....................... 103
MIP Synthase, a New Mechanism for Induced-fit? ............................... 108
MATERIALS AND METHODS .................................................................. 109
An Attempt in Solving Another Inhibitor Structure, the Z-
vinylhomophosphonate/MIP synthase Complex ................................... 109
CHAPTER 4
VALPROATE, A POTENTIAL THERAPEUTIC DRUG TARGET IN THE
TREATMENT OF MANIC DEPRESSION ....................................................... 110
GENERAL OVERVIEW ............................................................................ 110
An Introduction to Bipolar Disease ....................................................... 110
The lnositol Depletion Hypothesis ........................................................ 11 1
Recent Date Published on Valproate Inhibition of MIP Synthase ......... 111

vii

ASSAY RESULTS AND DISCUSSION ..................................................... 115

Potential Valproate Binding Site in MIP Synthase ................................ 115
Assay Results ...................................................................................... 1 16
MATERIALS AND METHODS .................................................................. 121
Activity Assay Experimental to Determine the K. for Valproate
Inhibition ............................................................................................. 121
REFERENCES ............................................................................................... 123

viii

LIST OF TABLES

Table 1. Data collection statistics for the two crystal forms of the native

MIP synthase .................................................................................. 39
Table 2. Data analysis for the Se-Met MIP synthase MAD phasing
experiment 42
Table 3. Statistics for the Se-Met sites found by SOLVE .............................. 48
Table 4. Se-Met atomic coordinates and occupancies for the sites

found by SOLVE ............................................................................. 49
Table 5. Refinement statistics for the native MIP synthase structure ............ 51
Table 6. Native MIP synthase tetramer interactions ...................................... 53
Table 7. Native MIP synthase dimer interactions .......................................... 55
Table 8. Residues that bind NAD in both yeast and archae form of MIP

synthase as seen in the holo structure ............................................ 62
Table 9. A list of the insertion residues that completely encapsulate

the NAD in both the yeast and archae forms as seen in the

holo structure .................................................................................. 63
Table 10. The make-up of the 2X M9 media for Se-Met overexpression ........ 74
Table 11. Heavy atom screening results for the most promising

denvafives ....................................................................................... 81
Table 12. Positions and occupacies for the HgCIz derivative .......................... 82
Table 13. Crystal statistics for the dgtoIP/NADH/MIP complex ....................... 86
Table 14. Refinement statistics for the dgtoIP/MIP synthase structure ........... 88
Table 15. New tetramer interactions formed upon inhibitor binding ................ 92
Table 16. New dimer interactions upon inhibitor binding ................................. 93
Table 17. Additional NAD interactions in both yeast and archae MIP ............. 93
Table 18. Components used in the valproic acid K. assay ............................ 122

Figure 1.
Figure 2.
Figure 3.

Figure 4.

Figure 5.
Figure 6.

Figure 7.

Figure 8.

Figure 9.

Figure 10.

Figure 11.

Figure 12.

LIST OF FIGURES

The biosynthesis of inositols and inositol phosphates ...................... 4
Sequence homology between all known MIP species .................... 11
The proposed mechanism of MIP synthase .................................... 19
Two possible modes of rearrangement by the cyclic

substrate ......................................................................................... 23
Schematic for a type I aldolase reaction ......................................... 26
Schematic for a type II aldolase reaction ........................................ 26

Acyclic substrate analogs with inhibition constants (K.) shown
below each compound .................................................................... 29

A series of cyclic compounds, all inactive towards MIP
synthase .......................................................................................... 30

E and Z series of vinyl homophosphonate inhibitors. A
and B are the corresponding intermediates in Figure 3 ................... 32

Mechanism and inhibitors of MIP synthase. Arrows
point from proposed intermediate to active enzyme
inhibitors .......................................................................................... 35

Crystal images. (a) MIP synthase crystal form I. Note:

crystal nucleated around a fiber. (b) MIP synthase crystal of

form II. (c) MIP synthase crystals produced via microseeding

at a dilution of 106-5 ......................................................................... 40

The structure factor diagram for the reflection of a protein crystal
that contains one kind of anomalously scattering atoms. F3 is the
contribution to the structure factor by the nonanomalously

scattering atoms, FA is the nonanomalous contribution of the
anomalously scattering atoms, and the complete nonanomalous
part is seen in equation 1. Equation 2 is the anomalous

contribution of the anomalously scattering atoms. The

anomalous component is exaggerated in this figure.

The dotted lines are for the mirror image of the Friedel

mate ................................................................................................ 45

Figure 13. Ribbons model of the MIP synthase tetramer. The top two
monomers are colored red and blue, make up one
asymmetric unit, and are related by a non-crystallographic
two-fold axis that is roughly in the plane of the page.
The bottom two monomers are green and are related
to the top two monomers by a crystallographic two-fold
axis running roughly perpendicular to the page .............................. 57

Figure 14. A front view of the native MIP synthase dimer ................................ 58

Figure 15. Ribbons model of the native MIP synthase monomer.
The various regions are colored: red for the N-terminal
region from residues 10-65; purple for the NAD-binding
domain encompassing residues 66-326; green for the
catalytic or tetramerization region containing residues
327-441 (with residues 352-409 disordered); blue, the C-
tenninal region containing residues 442-533. Note: the
N-terminal, C-terminal, and an insertion from the NAD-
binding regions, are intimately associated. Together,
they make up the central domain of the protein .............................. 59

Figure 16. The DAPDH monomer (PDB accession code 1DAP) (top) in
comparison to the holo MIP synthase monomer (bottom). .............. 64

Figure 17. The Rossmann fold search results. These structures
represent the types of Rossmann folds as seen in the
PDB Data Bank. Each protein has its pdb accession
code listed below its name .............................................................. 67

Figure 18. A space-filling model of MIP synthase in its substrate-
unbound form. The protein atoms are red while the NAD
atoms are green .............................................................................. 68

Figure 19. Purification of MIP synthase. (a) SDS-PAGE gel of MIP
synthase after each chromatographic step. (b) Purity of
native MIP synthase using the previous method ............................. 72

Figure 20. A comparison between the minimal and 2X M9 media
growths in the overexpression of Se-Met MIP synthase ................. 77

Figure 21. Final 2Fo - Fc electron density map contoured at 1.20
showing the dgtolP location within the active site. The
inhibitor is in red, active site residues in green, and
NADH/NH4 are in blue ..................................................................... 89

xi

Figure 22.

Figure 23.

Figure 24.

Figure 25.

Figure 26.

Figure 27.

Figure 28.

Figure 29.

Figure 30.

Ribbons depiction of inhibitor-bound MIP synthase.
The amino acids that were ordered in the holo form of

the enzyme are green while the newly ordered residues
(from 352-409) are red .................................................................... 94

Space-filling model of the inhibitor-bound MIP synthase.

The inhibitor dgtolP is colored yellow, NADH is green,

and the protein atoms are red. Note that both dgtolP and

NADH are completely obscured by the enzyme .............................. 95

A view of the loop encompassing residues 191-199 that

flips out upon inhibitor binding. The structures of MIP

synthase bound (red ribbons) and unbound (green ribbons)

by dgtolP are overlayed. Interactions between this loop

and the newly formed domain are shown. Dotted lines

denote hydrogen bonds. Residues are colored by atom

type: green, carbon; red, oxygen; blue, nitrogen ............................ 96

The interactions of dgtolP with the enzyme as seen in the

structure. Dotted lines denote hydrogen bonds. Atoms

are colored by atom type as in Figure 22 with the addition: magenta,
phosphorous. Bonds of the inhibitor are aqua.

Only residues that make direct hydrogen bonds to

dgtolP are shown for clarity blue ..................................................... 98

Overlay of the nicotinamide rings of alcohol

dehydrogenase (orange), DHQ synthase (cyano),

and MIP synthase (blue). The spheres represent

the relative positions of the divalent metal

(ADH and DHQ) and the monovalent ammonium

ion of MIP synthase ....................................................................... 101

Modeling of dgtolP in the active site of MIP synthase
to a conformation consistent with aldol cyclization and
subsequent reduction at CS. Atoms are colored by
atom type as in Figure 22. DgtoIP bonds are aqua

and NADH bonds are gold ............................................................ 105
Proposed mechanism for the transformation catalyzed

by MIP synthase ........................................................................... 106
An alternative mechanism for the catalysis of MIP synthase ........ 107
Chemical structure of valproic acid ............................................... 115

xii

Figure 31. Potential valproate binding site. The map is contoured
at 1.20 ........................................................................................... 118

Figure 32. A double reciprocal plot to determine K. and type of
inhibition ........................................................................................ 119

Figure 33. A plot of SN versus S to determine the KM/Vmax value
for the competitive inhibition equation used in calculating a K. ....... 120

** Some of the images in this dissertation are presented in color.

xiii

MIP

NAD

NADH
MI/Ins
CDP-DAG
Ptdlns
Ptdlns(4,5)P2
Ins(1,4,5,)P3
GDP

GTP

PLC-B

ER

GPI

DAPDH
G6P

DHQ
NADPH

SQD1
NaAc
PEG
Se

LIST OF ABBREVIATIONS

myo-inositol 1-phosphate
nicotinamide adenine dinucleotide
nicotinamide adenine dinucleotide, reduced form
myo-inositol

cytodine diphosphate-diacylglycerol
phosphatidyl inositol
phosphatidylinositoI-4,5-bisphosphate
myo-inositoI-1 ,4,5-triphosphate
guanosine diphosphate

guanosine triphosphate
phospholipase C-B

endoplasmic reticulum

glycosyl phosphatidyl inositol
diaminopimilate dehydrogenase
D-glucose 6-phosphate
dehydroquinate synthase

nicotinamide adenine dinucleotide phosphate,
reduced form

sulfolipid biosynthesis protein
sodium acetate
polyethylene glycol

selenium

xiv

SeMm
MAD
LB
AMP
IPTG
BME
TCA
NaIO4
Na2803
DTT
ddH20
dgtolP
MIR
p-CMB
FOM
ALDH
OXRED
UPS
ADH
PKC

seleno—methionine

multi-wavelength anomalous dispersion
Lauria-Bertani medium

ampicillin
isopropy-B-D-thiogalactopyranoside, dioxane free
B-mercaptoethanol

trichloroacetic acid

sodium periodate

sodium sulfite

dithiothreitol

double distilled water

2-deoxy-glucitol 6-phosphate

multiple isomorphous replacement
para-chloromercuriobenzoate

figure of merit

aldehyde dehydrogenase
oxidoreductase

undecaprenyl pyrophosphate synthase
alcohol dehydrogenase

protein kinase C

XV

CHAPTER 1
INTRODUCTION

BACKGROUND

General Overview

myo—lnositol is the starting material for the biosynthesis of a host of
critically important signaling molecules including the poly-phosphorylated myo-
inositols and the precursor phospholnositides‘. myo-Inositol 1-phosphate
synthase (MIP synthase, EC 5.5.1.4) converts D-glucose 6—phosphate to myo-
inositol 1-phosphate. This is the first committed and rate limiting step in the de
novo biosynthesis of all inositols in eukaryotesz. This complex transformation
occurs via a multi-step reaction mechanism involving an oxidation, reduction, and
intramolecular aldol condensation. This mechanism occurs stereospecifically
within a single active site with no dissociation of intermediates. The aldol
condensation is novel in that it seems to involve neither lysine Schiff-base
formation nor metal activation, rather a monovalent cation for catalysis". A more
detailed mechanism analysis will be presented in a later section.

MIP synthase is found in many eukaryotes including protozoa, fungi,
algae, plants, mammals, and humans. The gene encoding MIP synthase has
been cloned from several organisms including yeast (INO1 gene), amoebas, and
several plants4'37. In all organisms, MIP synthase codes for an approximate 60
kD monomer polypeptide. Very homologous complete sequence data has been

obtained, via a BLAST search, from mice and humans as wells.

MIP synthase is a member of a unique class of enzymes that use NAD not
as a cosubstrate, but as a co-catalyste. NAD is reduced to NADH and then
reoxidized back to NAD in the same catalytic cycle. Though a mechanistic
scheme has been proposed, there was no structural data illuminating the details
of this enzyme’s precise role in this important conversion until the inhibitor/MIP
structure was solved. Even though the three-dimensional fold of the protein is
now known, the enzyme’s amino acid sequence has no homology to proteins
which utilize a NAD-binding domain.

Strikingly, MIP synthase may be a target of therapeutic importance.
Currently, lithium is used in the treatment of manic depression’. Although the
precise mechanism by which this occurs is unknown, one theory is that lithium
acts by inhibiting MIP phosphatase and reducing inositol levels in the brain.
Other inhibitors of the phosphatase have already been tested in animal models
and are found to have similar effects to those of lithium, adding credence to this
mode of actions. Inhibiting MIP synthase may also lower inositol levels in the
brain and have a similar therapeutic effect. Recent data seems to suggest that

valproate may inhibit the MIP synthase pathway“.

Inositols and Cellular Signaling
myo-Inositol (MI or Ins.) is a critical component of all eukaryotic
membranes. The de novo biosynthesis of myo-inositol begins with the conversion
of glucose 6-phosphate to myo-inositol 1-phosphate (MIP) by MIP synthase,

followed by dephosphorylation of this product by MIP phosphatase to give myo-

inositol (Figure 1). MI is then combined with cytodine diphosphate-diacylglycerol
(CDP-DAG) in a reaction catalyzed by phosphatidyl inositol synthase to form
phosphatidyl inositol (Ptdlns)9. Ptdlns is the most abundant inositol lipid in
nature, making up 5% of the total membrane phospholipid content in
eukaryotes‘. Subsequent phosphorylations on the inositol moiety of Ptdlns
produce a variety of polyphosphorylated inositides including phosphatidylinositol-
4,5-bisphosphate (Ptdlns(4,5)P2), which is the precursor of a central second
messenger, myo-inositol 1,4,5-triphosphate (lns(1,4,5)P3) (Figure 1).

This second messenger is released when extracellular agonists bind to a
heterotrimeric G-protein-coupled receptor, causing the exchange of GDP for GTP
and subsequent activation of phospholipase C-B (PLC-B). Hydrolysis of
Ptdlns(4,5)P2 by activated PLC-B liberates lns(1,4,5)P3, which binds to its
receptor buried in the membrane of the endoplasmic reticulum (ER) and
stimulates the release of Ca”. Ca“2 release in turn causes a diverse set of
cellular responses including secretion, excitation, contraction, growth, and
proliferation”. MIP synthase’s role in this process is to maintain the
concentration of myo-inositol necessary for these critical signal transduction

processes to occur. The pathway described above is depicted in Figure 1.

 

 

 

 

 

 

 

 

HO OH HO OH HO O oEDP-DAG
HO O MlPsyntIuse HOflOH MIP phosphatase HO U
HO ——> H0 0903' +’ H0 0H] m
053032" Li inhibition 22 mijssé'iasaéf
Glucose 6—phosphatc MIP Inositol (Ins)
«AA/w HO O OPO32. am:
am H0 =WMAM§
:W 0903 2' o 2- 3 ~
vtvvva HO 0903 3W2
3W ———‘> HQ wwvw
WM VVWVV“
3m _ 2 op032- WVVVVVVQ ER
3% Ptdlns(4, SW: 03 PO "\ W2
3% W2
V’VWVV‘ \ W ++
W Ins(l,4,5)P3 ‘—_— C3
vwvw| [IR—
vW/W' VVVVVV‘
VWO WM
gWO W

 

 
 

 

R: G protein coupled receptor

 

 

 

IR: Ins(1, 4 ,5)P3 receptor (inserted m the ER membrane)

 

 

‘?
2:
W,_®
mw D Extracellular agonist

(((((((

 

ER: Endoplasmic reticulum
GTP GDP Gaby: Heterotrimeric G protein

Figure 1. The biosynthesis of inositols and inositol phosphates.

The Biological Role of MIP Synthase Activity

Recent studies have focused on the regulation of the enzyme in vivo in
both plants and fungi. In the latter case, the transcriptional regulation of the gene
has proven to be both complex and responsive. Transcription of the yeast INO1
gene is repressed in the presence of an inositol source and responds in an
intricate manner to the levels of the various inositol-phospholipids and their
precursors, inositol and choline""5. Several transcription factors have been
implicated in the process, including both those that appear to be specific for
phosphoinositide biosynthesis, as well as those that are involved in regulation of
other genes as well‘6‘22. These studies show that the activity of MIP synthase is
carefully regulated in response to cellular environment and stress. These studies
also show this activity to be an important determinant in cellular regulation in
yeast

Recent studies in plants have demonstrated that MIP synthase plays
several critical roles in these organisms as well. MIP synthase activity has been
shown to be rapidly up regulated by the plant growth hormone abscisic acid
during hormone—induced morphogenic response“. MIP synthase also plays an
essential role in seed development in rice by directing the synthesis of phytin
both chronologically and temporally”. MIP synthase is also differentially
regulated during embryonic and post-embryonic development of P. vulgan‘sz‘.

lnositol metabolism plays an important role in plant salinity tolerance. The
inositol metabolites ononitol and pinitol, as well as inositol itself, lower the

cytoplasmic osmotic potential and balance sodium accumulation in the vacuole.

MIP synthase’s role in this response is not merely in synthesizing the precursor
MIP, but also in regulating sodium uptake in the roots with photosynthetic activity
in the leaves. The enzyme is expressed in high levels in the leaves, maintains a
gradient in the phloem, and is down regulated in the roots. This causes root
growth and concomitant salt uptake to be dependent on the plant’s capacity to
produce glucose by photosynthesis. The glucose is then converted to myo-
inositol by MIP synthase and freely transported to the roots via the phloem. In
short, MIP synthase activity is thought to be the linchpin of an important
regulatory system in plants, intimately involved in communication between
photosynthetic activity and root metabolism” 2526.

All of this data indicates that MIP synthase activity plays important roles in
plant development, response to environmental stress, long-range
communication, and intracellular signaling. Interest in the role of these enzymes
in plant biology is intensifying as most of these results have come within the last
decade”.

More recently, MIP synthase has been cloned from a parasitic protozoan
Entamoeba histolytica. The enzyme’s action is critical to the parasitic function of
these organisms because they use glycosyl phosphatidyl inositol (GPI) to anchor
macromolecules to their exterior membrane. Also, by changing the E. histolytica
MIP synthase gene, the evolution of MIP synthase along the phylogenetic line
may be traced. Many of these macromolecules are necessary for interaction with

the target host”.

Even though the gene encoding for MIP synthase in mice and humans
has been cloned, little is known regarding the regulation of this enzyme in these
higher eukaryotes. Indeed, since most of the inositols in mammals and higher
eukaryotes come from dietary sources, it is not expected that the enzyme will be
generally expressed. However, MIP synthase activity exists and can be purified
from rat and bovine testis, rat and human brain, rat liver and kidneys, and rat
mammary glandzw‘. MIP synthase activity in the liver is reduced by thyroid
removal, but this reduction can be reversed by hormone treatment”. This
indicates that MIP synthase may be part of an important regulatory pathway in
some organs. Since dietary inositol does not traverse the blood/brain barrier, MIP
synthase activity is probably important in maintaining inositol levels there,
especially since brain and nerve tissue actively use lns(1,4,5)P2 as a second
messenger in neuronal stimulation. Several reports have shown that changes in
brain inositol levels can have profound effects on various aspects of behavior,
indicating that inositol synthesis is an important determinant of neurochemical
brain function“ 33'”.

In short, all of the available data indicates that MIP synthase behaves not
as a housekeeping gene, whose expression and activity remain level through the

life of the cell, but as a gene that has important regulatory functions in response

to stress, cellular signaling, and development.

MIP Synthase Homology, Classification, and Characteristics

The amino acid sequence of MIP synthase has been remarkably well
conserved throughout evolution, from archae to human, over virtually its entire
sequencez. For example, the S. cerevisae (yeast) and D. melanogaster (fly)
amino acid sequences are 45% identical in overall sequence. Similarly, with the
archae species, there is also a high sequence homology. Interestingly, in the
active site, the yeast and archae species are 75% homologous even though the
overall sequences differ by 150 amino acids. In contrast to this high homology,
BLAST searches of all available gene databases using the Stanford yeast MIP
synthase sequence reveal convincing homology to all other MIP synthase genes
whose sequences are known, but show no convincing homology to any other
proteins. In addition, the amino acid sequence of MIP synthase was subjected to
a battery of threading and fold-recognition programs available from the UCLA-
DOE structure prediction server”, and again, no convincing matches were found.
There was one interesting result from the folding recognition searches. The
enzyme diaminopimilate dehydrogenase (DAPDH), an enzyme involved in L-
arginine synthesis, has no sequence homology but does have similar folds. This
enzyme consists of a Rossmann-fold, dimerization interface, and an insertion
domain, each characteristic of the MIP synthase structure. These results are in
agreement with those of the literature, suggesting that MIP synthase is
structurally uniquez. Figure 2 depicts the sequence alignment between all known

species of MIP synthase.

MIP synthase is a member of a small class of enzymes that utilize NAD
catalytically. This class of enzymes includes the epimerases, nucleoside
diphosphohexose oxidoreductase, S-adenosylehomocysteine hydrolase,
dehydroquinate synthase (DHQ synthase), 8001, and omithine cyclase39. The
strategy of all these enzymes is to transiently oxidize an alcohol or amine to a
carbonyl or imine. Carbon-carbon bonds are either formed or broken at positions
adjoining the oxidized group. NADH or NADPH then reduces the oxidized group
back to its original form to complete the catalytic cycle. Of all of these enzymes,
the X-ray structures of DHQ synthase, UDP-galactose-4—epimerase, $001, and
S-adenosylehomocysteine hydrolase are known‘o'“. While the epimerase and
SQD1 have similar overall folds, both DHQ synthase and S —
adenosylehomocysteine hydrolase are structurally unique. The one structural
characteristic that all of these enzymes share is a Rossmann fold-like NAD
binding domain. Even though these enzymes share an NAD-binding domain in
their structures, no convincing sequence homology can be detected between it
and any of these enzymes in pairwise alignments using the DNASTAR program
package.

In most cases, the MIP synthase gene encodes for a protein of around 530
amino acids (533 in yeast) corresponding to a subunit weight of around 60 kD.
Purification and characterization of both native and recombinant proteins from
several species have demonstrated several salient features of the enzyme:

(a) It is always oligomeric in solution. While the chloroplast and

fungal enzymes are trimers, most other forms behave as
tetramers.

(b) It is activated by ammonium ions (five fold) and inhibited by
glucitol 6-phosphate, 2-deoxy-glucose 6-phosphate, and E-
vinylhomophosphonate.
(c) It operates in a pH range of 7.0-8.4.
(d) Requires NAD for activity, but uses it catalytically. not as a
co-substrate.
As the crystal structures of MIP synthase will show, features (a) and (b) are
essential in understanding the chemistry behind MIP synthase catalysis. These
features demonstrate how similarty the enzyme behaves from species to species.
The combination of the high sequence conservation shown above, and the fact
that the enzymes behave so similarly, imply that structural insights will be
immediately transferable to other MIP synthase enzymes from other organisms

based on the structure of the yeast MIP synthase.

10

A...........¢...._._._..p_....a_.

 

Figure 2. Sequence homology between several MIP synthase species.

11

39

“888§:“88882382

     

"K-KVPKLG ----- VMLVGWGGNNGSTLTA ---------------- GVLANKHGLSWATKT Mauw
............... Yeast
. human
................ C paradisi
ida
e.hist.pm
ammehm
............... M crystalinum
............ VA. v o mycobact?.pro
OTNEKDEYEEKDEVY 'R ornithinedecarbp
RNYAHWV 0 pvulgarispro
_ _ . . . . LL ................ S pida paslof.
R- VPV ----V --------------- L trypanosome
------- YG---- VT - - GL------ Arches
-GVOQANYFGSLTQASTIRLGSDAE-GEEVYAPF ------------------------- K Mfiww

 

1n 1o 10 I

______ . v.‘K

 

S NMNL ADAMRRAQVL D

    

S- - LL PMVNPDDI VVGG ------------------------ VIDI
' 1| 1.:

  

'1

' COCO-o..-
'—

Figure 2 (cont’d)

12

  
 

YY P """""""""""" DFI 'AA’NQG
'I

......... I DLQKOLYPYMELLVPLPSI - -

   
    
   
 
 
 

    

  

TV -
D G . Lamamnensls
- - K - EQVE K . ‘
PE-— - - -M VS- GHV VAP"KD F -- methanobact’lpr
PP ------- T Vt} RGPT (113- - I mycobaet?.pro
NLSRDRI I FAN KSI NSLI Y RK NI N omrltrirledecarbp
- - - - - - ------- .pro
. . pastor
-------- Arches

VLWTANT- ERYSEVI PGVNDTAE-- NLLASI KAGEEEI-
so 2&0

- V

L

C
FC
V

Figure 2 (cont’d)

13

   

LEGCPFI NGSP omrvpcu EL AEKKGVFI GGDDFKS --------- GQTKI KSVLADFL Majority
2§o 360

Yeast

E .

  

 
    

...... .- VV V-' ..........._
------ -P PY human
------ -P v PY‘
...... -N PR‘ PY‘ Operadisi
----- PIYK CahrgIda
------ - P 9. .pro
------ -P g PY Enlamoebahist
------ '3 PY Lamazomnsls
------ P V PY Maydallnum
------ NI 6' P ---------- urethanobem."
------ II GPS GVl ---------- bact?.pro
o TKIN DN ornlthinedocerbp
----------- ' .pro
---------------- pompastor.
------------ Archao

Figure 2 (cont’d)

14

378

 

-------------- AMDEYTSEI FLGGHNTI VL HNVTC- EDSLLAAPLI | DLVVLTELA

     
     
  

TRVOYKA- DE-K ------ FESFHPVLSILSYL--LKAPLVP-PGTP ------- VVNALNK
so .,. so

svvaOD annihkiéi'vv'sb'si'vb
~ INK ......

--------------- SFt'SFSFSAY-
RLoV

.- LL ECR ----------------- mEHv -----
.................. HEQFVVLKEw----------------------------~YS

Figure 2 (cont’d)

15

Yeast

human

we ..

C. paradrsi

C nda

o.hist.pro

Entamoeba hist.
. crysta inum
methanobad”

omithinedecarbp
p.vulgaris.pro

‘ ‘ pastor.

.............................................. ‘u
- - - I LGQPCSVESVTNGKKLHANGH ------- SNGSAKL AT N- GNGH
- - - ML NKKGPVPAAT NGCT GDANGH ------- L QEEPPMPTT
......................................... T-EHFF
......................................... T-EHFF
............................................ NGK
-- - - - - - - - -AG-ERER
---------------------------------------- lD-TK

Figure 2 (cont’d)

16

Majority

Yeast

fly

human

plant

C paradisi
Candida
ehlslpro
Entamoeba hist
L. amazonensis
M. crystallinum
methanobact?.pr
mycobact?.pro
omithinedecarbp
p.vulgarrs.pro
plCIa pastor.
trypanosome
Archae

Mechanistic Aspects of MIP Synthase

The elucidation of the intricate set of reactions catalyzed in the active site
of MIP synthase has been the subject of intense investigation for the past thirty
years. These studies have involved the use of radiolabeled substrate and NAD in
feeding and biochemical studies, as well as mechanism based inhibitorsz' 39' “"5.
Mechanistic studies indicate that MIP synthase from various sources share the
common mechanism originally predicted by Loewus and Kelly6b. These
experiments have resulted in the proposed mechanism shown in Figure 3.

The first step of this process is either binding of the acyclic form of glucose
6-phosphate or binding of the more prevalent cyclic form followed by acid or base
catalyzed ring opening in the active site. NAD then oxidizes this acyclic
compound in the C-4 position to yield the keto intermediate B. Enolization of this
ketone to form intermediate C sets up the cyclization reaction by aldol
condensation to yield D. NADH then reduces the C-5 keto group back to an
alcohol and enzyme release yields the product 1L-myo-inositol 1-phosphate.

The mechanistic studies were carried out with partially purified enzymes
obtained from different sources. Most of the findings were similar with regard to
the mechanism, suggesting a similar mechanism for all species. Initial
mechanistic studies concluded that myo-inositol 1-phosphate synthase requires
NAD for activity. Treatment with charcoal removes NAD from the rat testis
enzyme, leaving a totally inactive apoenzyme7"1°"‘°2. Reconstitution of the
apoenzyme with added NAD restores 80% of the original activity. NADH

generated during the reaction is tightly bound to the enzyme and therefore does

17

not seem to be exchanged between the enzyme-bound NADH and free NADH in
the medium. The hydrogen on the C-5 of D-glucose 6-phosphate (G6P) is
transferred to the C-4 position of the dihydropyridine ring of NADH during the
oxidation of G6P, and the same hydrogen is delivered back to the product at the
same position during the reduction of 2-inosose 1-phosphate. Adding NADH(3H)
to an incubation medium of MIP synthase and G6P failed to introduce tritium into
the reaction substrate and product‘o" 103. None of the intermediates have been
isolated, suggesting all intermediates are tightly bound and not released until the
final reduction to myo-inositol 1-phosphate. When myo-inositol 1-phosphate
synthase from bovine testis was incubated simultaneously with a mixture of
deuterated D-glucose 6-phosphate-dy and nondeuterated D-glucose 6-
phosphate-do, there was no crossover of the label from deuterated to
nondeuterated product”.

Incubation of partially purified apoenzyme from rat testis with 5-keto-G6P
and [RS-4-3H]-NADH led to the formation of D-glucose 5-3H 6-phosphate and
trace levels of tritiated myo-inositol 1-phosphate‘°2. This apoenzyme also
catalyzed the reduction of 5-keto-D-glucitol 6-phosphate by [RS-4-3H]-NADH to
form D-glucitol 5-3H 6-phosphate. Additional evidence for the involvement of 5-
keto-G6P was found in another experiment in which base treatment of 5-keto-
G6P yielded two cyclose phosphates, which after reduction with NaBH., gave a
mixture of two cyclitol monophosphates, L-myo-inositol 1-phosphate and epi-

inositol 3-phosphate‘°5.

18

OH 1
HO 3 OH
O

5
HO 6

opoan2

/Glucose 6-phosphate
HO O OH
HO_%0 HO
0 0P0st
01303142
HO H
H A myo-inositoI1-phosphate(MlP)
(Oxidation) NAD (Reduction)
phi/tad

OH
0%,, HO 0 0H
”CPOMZ HO 0P03H2

HO
HO
B H

\ myo-2-inosose 1-phosphate
Enolization D
“0%0 /
I

HO \ OP 03H dol cyclization

C

H
HO

HO

Figure 3. The proposed mechanism of MIP synthase.

19

By addition of NaB3H4 to a solution containing homogenous rat testis MIP
synthase, GSP, and NAD, 2-inosose 1-phosphate was trapped as a mixture of
tritiated myo-inositol and scyIIo-inositol‘os. Iditol and glucitol, epimeric alditols
representing 5-keto-D-glucose 6-phosphonate, were not significantly labeled. Its
cyclization to 2-inosose 1-phosphate occurs as soon as its formation. Its
reduction to putative NADH in the enzymatic reaction must be the rate-limiting
step. In the presence of apo-MIP synthase reconstituted with NADH, myo-2-
inosose 1-phosphate is converted into myo-inositol 1-phosphate along with
oxidation of the enzyme-bound NADH.

In terms of kinetic isotope effects, with [5-3H] GBP as a substrate, an
isotope effect of 0.2 to 0.48 was observed‘°3. An isotope effect in the reaction of
D-[6-3H] GGP was also reported, suggesting that removal of the hydrogen at C-6
to be partially rate determining. No isotope effects at C-1, C-2, C-3, and C-4 were
detected101, 103-104.

When reactions were undertaken with D-[5-‘BO] G6P in H20 or with
unlabeled substrate in H2180 using enzyme from various sources, mass spectral
analysis of the products showed no exchange of the C-5 oxygen with H20 during
the reactionab'ad'31'50. This result is inconsistent with the involvement of a Schiff
base between an active site residue and the C-5 keto group of intermediate B
(Figure 3). If a Schiff base reaction were to occur, one would expect to see an
exchange between the C-5 oxygen and H20.

When the reaction was carried out with both D-[(6S)-6-3H] G6P and o-

[(6R)—6-3H] G6P as substrate, Floss reported that the tritium was preferentially

20

removed from the pro-R position by all three enzymes purified from bovine testis,
L. Iongiflorum, and S. flavopen'cus‘s. Furthermore, Floss concludes that the ring
closure occurs in a retention mode at C-6 of the substrate, a finding that
contradicts eanier reports but agrees with the stereochemistry determined in the
S. flavopericus reaction and with general aldolase reactions. This idea of
retention of the C-6 position on the substrate turns out to be highly controversial.
Other experiments contradict the Floss results. Working with the charcoal treated
partially purified rat testis and mammalian enzymes, Byun and Jenness observed
an uptake of tritium in both G6P and myo-inositol 1-phosphate from [(4S)-4-3H]-
NADH‘°7. No appearance of tritium in either G6P or myo-inositol 1-phosphate
was observed when [(4R)-4-3H]-NADH was used. This result led to the
conclusion that hydrogen transfer involves the pro-S hydrogen of NADH.
However, this experimental result is not consistent with the idea that NADH
remains tightly bound to the enzyme throughout the reaction. Similarly, whether
the enzyme could process an enzyme-NADH-G6P complex is still questionable.
While many experiments have been done to try and elucidate the complex
mechanism employed by MIP synthase, there is no discrete chemical
explanation for this transformation. Although a minimal basis is available, many
questions about the enzyme's mechanism remain unchallenged. Solving the
structure of the inhibitor/MIP synthase complex has aided in eliminating some of
the confusion. The inhibitor/MIP structure will show that the acyclic form of the

substrate binds suggesting that the repositioning of the phosphate does occur

21

before binding. The remaining mechanistic issues are discussed in detail in the

proceeding section.

Remaining Mechanism Issues

Several issues regarding this mechanism remain controversial or
speculative:

(a) Does the enzyme bind the cyclic or acyclic form of glucose 6-
phosphate? What is the anomeric configuration of the C-1
carbon of glucose 6-phosphate if the enzyme does bind the
cyclic form?

It has long been assumed that MIP synthase first binds the cyclic form of
glucose 6-phosphate and subsequently catalyzes the conversion to the acyclic
form within the active site because the solution equilibrium of glucose 6-
phosphate favors the cyclic form with a Keq > 1 x 103. This hypothesis presents a
problem however, because it is expected that phosphate interaction with the
enzyme is one of the most important anchoring points not only of the substrate,
but of all subsequent intermediates as well. As the inhibitor structure
demonstrates, the phosphate is the most important anchor in the substrate
binding to the enzyme. The question arises as to how the substrate rearranges
itself after ring opening to obtain a conformation consistent with the cyclization
step. While it is still unclear, Figure 4a shows the simplest answer, which is
rotation of the phosphate and hydroxyl about the 04-05 bond resulting in
phosphate repositioning. The other alternative is the conformational

rearrangement proposed by Floss, which is depicted in Figure 4b. Floss believes

that if the substrate is bound as the 8 anomer in the conformation shown, rotation

22

about the C4-05 bond can occur without repositioning of the phosphate moiety“.
With this conformation, he further postulates that a single residue base in the
enzyme could deprotonate the O1 and O5 oxygens in the ring opening step;
deprotonate C6 during the enolization step; and finally, protonate 01 during the
aldol cyclization step, because all of these proton transfers occur on the same
edge of the substrate. On the other hand, inhibition studies (Frost lab) indicate
that, save the inosose intermediate D, only acyclic compounds effectively inhibit
the enzyme, leading to the supposition that it is actually the acyclic form of the
substrate that binds to the enzyme“. This would be surprising, however,

because in solution, no more than about 1% of the substrate is in the acyclic

 

 

 

form.
a.
O Hoy§o
HO
0 Ring Opening: mm“ pVOPOSHa
—>
HO HO HO
A
OP03H2 OPO3H2
Glucose 6-phosphate
b.
O OH O \o HO}’~\~0
HO HO HO
0 _> OH :_
HO 0 HO 0 HO 0
\ \ \ _
P=O _ P—0
d \ /\ O A d \
HO HO HO OH

Figure 4. Two possible modes of rearrangement by the cyclic substrate. (a) The
simplest rearrangement where the phosphate repositions upon ring
opening. (b) An alternative rearrangement as proposed by Loewus
and Floss.

23

(b) What is the nature of the various proton donors and acceptors in
the reaction pathway?

While there has been much speculation regarding the role of the enzyme
in various proton donation and abstraction events in the pathway, there is
insufficient data to elaborate the details of this process. For example, two
possibilities have been proposed for the identity of the base that abstracts the
proton from intermediate B to form enolate C (Figure 3). According to Floss, a
single base on the enzyme could both abstract the proton during the enol
formation and then subsequently transfer the same proton to the ketone
intermediate C during the aldol condensation. Similarly, this same base could
also catalyze the ring opening of the glucose 6-phosphate by transferring the
proton on the C1 alcohol to ether oxygen in the hemiacetal ring. Conversely,
Frost and co-workers have performed inhibition studies that seem to imply that
MIP synthase binds only the acyclic form of the substrate, negating the necessity
for catalyzing ring opening. In addition, their studies have indicated that the
phosphate group of the substrate may be the base that abstracts the proton
during enol formation. The inhibitor/MIP structure complies with the Frost notion
that the phosphate is the active site base that pulls off the proton during
enolization and not an active site residue base.

(c) What is the nature of the enzyme-catalyzed aldol condensation?

Almost all enzymatically catalyzed aldol condensations utilize one of two
methods for activating the aldehyde carbonyl prior to attack by the enol.

Aldolases are categorized as type I or type II enzymes based on this division“.

24

While type I aldolases are primarily found in higher plants and animals, type II
aldolases are predominantly found in bacteria and fungi. In type I aldolases, the
ketone is activated by a Schiff base formation with a lysine residue (or histidine)
from the enzyme (Figure 5)“ ‘8. No metal cofactor is necessary for this
activation. In type II aldolases, a metal cofactor, usually zinc, acts as a Lewis
acid to stabilize the developing negative charge on the keto oxygen and can
chelate the enolate to stabilize this negative charge (Figure 6)“ ‘9. Neither of
these mechanisms appears to explain the intramolecular aldol condensation
catalyzed by MIP synthase (i.e. the transformation from intermediate B to C in
Figure 3). Though some preliminary studies seemed to indicate a Schiff base
intermediate, indicative of a type I aldol condensation, NaBH4 reduction failed to
trap the expected intermediates, nor was there any exchange of radioactivity to
the product when the reaction was run in H2180 as would be expected with the
formation of a Schiff base intermediate as discussed earlier“. Since yeast MIP
synthase requires no divalent metal, nor is it inhibited by high concentrations of
EDTA, a type II aldol condensation pathway is also unlikely, leaving the
mechanism of this important step in the reaction to be essentially not
understood” 3" 5°. Indeed, it is intriguing to note that MIP synthase is able to
efficiently catalyze this cyclization without resorting to these forms of substrate
activation. Although neither mechanism is likely, the inhibitor/MIP structure
indicates that a third aldolase mechanism may be the driving force for the aldol
condensation. The third mechanism, a type III aldolase, utilizes a monovalent

cation as a Lewis acid to stabilize the negative charge on the keto oxygen and

25

can chelate the enolate to stabilize the negative charge. In the inhibitor/MIP
structure, the monovalent cation ion is an ammonium ion. This makes sense

since yeast M lP synthase is activated five fold by ammonium ions.

 

OH O 'N' -

Figure 6. Schematic for a type II aldolase reaction.
(d) How does MIP synthase bind NAD?

Given the prevalence of the Rossmann-fold in most other NAD-binding
proteins, it is no surprise to find that MIP synthase does contain a Rossmann
fold-like domain. Though sections of the amino acid sequence of MIP synthase
contain conserved glycine-rich regions, characteristic of this domain, no
sequence homology can be detected between MIP synthase and any protein
whose structures are knownz. In addition, none of these glycine-rich regions
matches well with the characteristic pattern of small hydrophobic and glycine
residues found in these domains“ 52. Of additional interest, the structure of DHQ
synthase has revealed an unusual interaction between the nucleotide and the
Rossmann fold. This results in the active site of the enzyme being located on the
opposite side of the beta sheet relative to all other known structures of this
type“. Of all the enzymes of known structure, the reaction catalyzed by DHQ
synthase is most similar to that of MIP synthase in that an oxidation, reduction,
and aldol condensation all occurs in the enzyme’s active site. These similarities
beg the question of whether there are structural similarities in these two
enzymes, particularly in the way they bind the nucleotide. Sequence alignments
of the two proteins, however, reveal no homology to one another. The
inhibitor/MIP structure has revealed that MIP synthase binds NAD in a fashion
different from that of DHQ synthase: (1) an insertion loop in the MIP structure

(residues 149-215) completely encompasses the adenine part of NAD molecule

27

and (2) the orientation of the nicotinamide ring is flipped 120° to that of the
nicotinamide ring of DHQ synthase (see Figure 26).

AN IN DEPTH LOOK INTO PREVIOUS INHIBITOR STUDIES AND
THEIR POTENTIAL RO|_.ES IN MIP SYNTHASE CATALYSIS

An Introduction to Previous Inhibitor Studies

As stated earlier, there are several questions concerning the complex
mechanism employed by MIP synthase that remain unanswered. In order to shed
some light on those questions, crystallographic inhibitor studies were
implemented with the hopes of developing a structural basis for the MIP synthase
mechanism. The Frost group at Michigan State University were kind enough to
aid in these studies with the design and synthesis of potent inhibitors that mimic
most of the proposed intermediates in the catalytic cycle. Additionally, inhibitors
were designed to answer specific questions regarding the enzyme’s role in
catalysis. The major outstanding questions regarding the mechanism of MIP
synthase, as mentioned earlier in the introduction, will be further investigated

here by describing the role(s) of the proposed inhibitor/enzyme complexes.

Does the Enzyme Bind the Cyclic of Acyclic Form of Glucose 6-phosphate?
What is the Anomeric Configuration of the C1 carbon if the Enzyme Does
Bind the Cyclic Form?

If MIP synthase first binds the cyclic form of glucose 6-phosphate and
subsequently catalyzes ring opening to the acyclic form to produce intermediate

A in Figure 3, the most likely mechanism for this conformational rearrangement is

that depicted in Figure 4b involving rotation about the C4-C5 bond without

28

significant repositioning of the phosphate group. If this is the case, inhibitors that
mimic intermediate A should be bound in an unique conformation with the
phosphate held anti to the C5-C6 bond as shown in Figure 4b. Such acyclic
inhibitors were available, either commercially or via the collaboration with the
Frost group. Figure 7 shows this series of inhibitors and their K.s56' 72'". Note that
all of these molecules are inhibitors because they lack an aldehyde at the
terminal carbon, and are therefore incapable of undergoing the aldol cyclization.
Additionally, it is interesting to note that removal of the hydroxyl group at the CZ
position improves inhibition significantly. Since the substrate obviously has a
hydroxyl group at this position, the reason for this improvement is completely

unclear.

     

on
.\\OH OH
= on OH NW
H203PO 6H n OH
D-glucitol 6-phosphate 5
1
K. = 150x106 Ki :1 ioxiotS Ki =5.8xio*5 Ki =17x10"5 Ki =47x10'6

(R3CH2PO3H2)

Figure 7. Acyclic substrate analogs with inhibition constants (K.) shown below
each compound.

To test whether MIP synthase binds the cyclic form of the substrate, a

series of molecules designed to mimic the cyclic form of the substrate, but unable

to undergo ring-opening, were synthesized (Figure 8) and all were found to be

29

ineffective inhibitors towards MIP synthase. This led to the supposition that in
fact, MIP synthase binds only the acyclic form of the substrate, negating the

requirement for the unique conformation described earlier.

 
       

 
     

H203PO 5H H203PO 5H H203PO C=)H H203PO 6H

Figure 8. A series of cyclic compounds, all inactive towards MIP synthase.

To further test this hypothesis, a series of molecules were designed and
synthesized, that restrict the position of the phosphate group relative to the rest
of the molecule72' 73. This was done by replacing the phosphate moiety with a
phosphonate group and introducing either a cis- or trans- double bound between
the phosphonate and the rest of the molecule (Figure 9). For reference, a
phosphonate is identical to a phosphate save for the replacement of the ether
linkage with a methylene. According to the cyclic substrate-binding model, the Z-
series inhibitors should more accurately mimic the structure of the acyclic
intermediate in the active site. However, it is only the E-series of compounds that
inhibit MIP synthase. As proof that these inhibitors accurately mimic true
enzymatic intermediates, it was shown that these inhibitors are oxidized to the
ketone by the enzymes depicted in Figure 9. Structures of one or more of these
inhibitor/enzyme complexes and comparison with the above proposed structures

would further verify that they undergo substrate-like binding to the enzyme. Since

30

they are oxidized in the active site, these structures should mimic the binding of

intermediate B in the active site.

What is the Nature of the Various Proton Donors and Acceptors in the
Reaction Pathway?

The combination of the above-proposed structures would be extremely
informative in delineating the side chains of the enzyme that are involved in
various protonations and deprotonations that occur during the catalytic cycle. The
combination of a wide variety of enzyme/inhibitor structures, all of which are
based on mechanistic assumptions regarding the enzyme, should add
considerable weight to these assignments. The E-phosphonate series should be
especially informative in determining whether it is the substrate phosphate or the
enzyme that provides the catalyst for removal of the proton necessary for
formation of the enol intermediate C (Figure 3). At present, the phosphate base
hypothesis is based solely on the fact that the E-phosphonate series of inhibitors
may indicate that the phosphate is positioned correctly to be the agent
responsible for this abstraction (Figure 9). This, as described in the dgtolP/MIP
synthase complex structure section, turns out to be not a hypothesis, but a

structural fact.

31

Z series inhibitors HO OH

(inactive) NAD NADH HO
6 R = CHO n / o
7 R = CH20H

HO H
NAD NADH H0
o— H
_}O

D-Glucose 6-phosphate B
(keto form, A)

E series inhibitors

(active) OH NAD NADH
8 R1=CHO, R2=OH

K. =1.1x10'3
9 R.=CH20H, 122:0}:®
K. =30x1045

Hgl OH
10 R.=CHO,R2=H n NAD NADH
K.=100x10'6 2 R / $4
(.3) I OH

1 l R1=CH20H, R2=H
K. =0.67x10'6

 

Figure 9. E and Z series of vinyl homophosphonate inhibitors. A and B are the
corresponding intermediates in Figure 3.

32

What is the Nature of the Enzyme-catalyzed Aldol Cyclization?

As explained earlier, the nature of the aldol cyclization catalyzed by this
enzyme remains mysterious as it appears to employ neither of the two well-
known aldol mechanisms found in enzymes but rather a type III aldolase
mechanism. The issues in this step include the following:

1. How does the enzyme orient the two ends of the molecule to optimize

cyclization?

2. How does the enzyme activate the aldehyde sufficiently for aldol

condensation without the use of metal cofactors?

3. How does the enzyme’s active site optimally shift the keto-enol

equilibrium to favor the enol form necessary for the condensation?
The mechanism of this condensation should be effectively attacked because
there is significant access to inhibitors that mimic intermediates that are on both
sides of the aldol condensation. The inhibitors described above (compounds 1, 2,
3, and 4 in Figure 7; compounds 9, 10, and 11 in Figure 9) together effectively
mimic the intermediates (A, B, and C in Figure 3) that occur before the aldol
condensation. From these stmctures, a general understanding of the structural
details of how the acyclic intermediates are bound in the active site should be
evident. Compound 10 could be especially interesting because it is the only
inhibitor of MIP synthase that contains an aldehyde at C1, representing the
electrophile in the aldol reaction. Additionally, intermediate D, myo-2-inosose 1-
phosphate, has also been synthesized and found to be an effective inhibitor (K. =
3.6x10'5 M)“. This molecule is an inhibitor because it binds the NAD form of the
enzyme and therefore cannot be reduced to the final product in the active site. In

summary, Figure 10 juxtaposes tight-binding inhibitors of MIP synthase with most

of the intermediates in the catalytic cycle, displaying a structural map in

33

developing answers to many of the outstanding questions concerning the

mechanism of MIP synthase.

How Does MIP Synthase Bind NAD?

As previously described, treatment of MIP synthase with activated
charcoal followed by introduction of NADH is known to give the NADH form of the
enzyme. This form of the enzyme should also be crystallized to elucidate the
structural transformations incurred upon reduction of the cofactor. The inability of
the NAD form of MIP synthase to release bound myo-inosose 1-phosphate
(intermediate D) from the active site has led to the hypothesis that a significant
conformational change may indeed occur upon reduction of the cofactor NAD to
NADH. This conformational change would then result in the release of the final
product MIP. These structures would give clear answers to the questions
regarding the nature of the enzyme’s association with its cofactor and elucidate

the details of the proposed conformational change.

 

 

OH Ho/‘Qi/FOH
DVCHzPoaH OPOgH
2
HO H dgtolP

Glucose 6-phosphate
\ HO O OH

(5H0
—%O/ HO 0P03H2

OPO3H2

myo-l-Inositol 1-phosphate (MIP)

:(:xidat:n)k ”:5: +13% (Reduction)
OH
HO HO

'. B H myo-2-lnosose 1-phosphate
" Enolization D i.
’ Hoff/W) I

 

 

 

OH \ OPOgH dol cyclization .‘
HO O_[/\OH HO
C O 0 OH
HO POW “0 090 H
a 2
E-vinylhg'mophosphonate HO
myo-2-inosose 1-phosphate

   

 

 

Figure 10. Mechanism and inhibitors of MIP synthase. Arrows point from
proposed intermediate to active enzyme inhibitors.

35

GOAL OF THIS THESIS

Due to the lack of structural data, the structure of yeast MIP synthase was
solved via x-ray crystallography. Similarly, to shed light on some of the
mechanism inconsistencies, an inhibitor/MIP synthase structure was also solved.
The purification, crystallization, structure determination, and stmctural results of

both of these structures will be discussed in this thesis.

36

CHAPTER 2
THE NATIVE MIP SYNTHASE STRUCTURE

EXPERIMENTAL

Crystallization and Data Collection Analysis for the Native MIP Synthase

Two crystal forms were grown from hanging drops, via the vapor-diffusion
technique at room temperature”. The drops were prepared by mixing 2 pL of
protein solution (10 mg mL‘1 in buffer A) with 2 pL of reservoir solution. The drops
were equilibrated against 1 mL of reservoir solution. For crystallization of crystal
form I, MIP synthase was equilibrated against a reservoir solution containing 2-
5% (v/v) PEG 8000 and 100 mM sodium acetate (NaAc) pH 4.5. For crystal form
ll MIP synthase was again equilibrated against a reservoir solution containing 5-
8% (v/v) PEG 8000 and 100 mM NaAc pH 5.0-6.0. Crystals from both forms were
observed in 1-2 days.

For data collection, both forms can be transferred to cryoprotectant mother
liquors (5% PEG 8000, 100mM NaAc at pH 4.5 or pH 5.0 depending on crystal
form and 30% glycerol). Complete data sets were collected at —150°C (123K)
from a single, flash frozen crystal of each form.

Native data were collected at home on an R-AXIS lIc imaging-plate
system with Cu Ka x-rays generated with a Rigaku RU200 rotating-anode
generator operated at 50 kV and 100 mA. Data reduction and scaling were

performed using DENZO and SCALEPACK, respectively”.

37

X-ray diffraction quality crystals of MIP synthase were generated using the
outlined conditions. Form I crystals grew to dimensions of 0.2 x 0.4 x 0.7 mm
(Figure 11a). Form ll crystals have typical dimensions of 0.1 x 0.3 x 0.6 mm
(Figure 11b). Since spontaneous nucleation of both crystal forms was difficult, a
microseeding protocol was developed for the form I crystals that consistently
yields large, single, and well-diffracting crystals. In microseeding, 2-3 crystals
from a previous crystallization were crushed in a 10 pL drop of a mother liquor
(5% PEG 8000, 100 mM NaAc pH 4.5). These crushed crystals were then diluted
in the mother liquor in a series ranging from 10'3 to 10". The ideal dilution range
for yielding the aforementioned crystals was 105'5 to 10“. Consequently, the
microseeded crystals were better defined morphologically with typical dimensions
of 0.2 x 0.3 x 0.4 mm (Figure 11c). Crystal form I diffracted to 2.5 A with a data
set completeness of 96.7% with a Rmerge of 6.6%. Form II crystals diffracted to
2.9 A with a data set completeness of 99.9%. The Rmerge of form II was 10.7%.
Both crystal forms are monoclinic with form I belonging to space group C2 and
form ll belonging to P21. Unit-cell parameters and data statistics are given in
Table 1 for the data sets collected on each form. The volume of the unit cell for
the C2 form is consistent with a 120 kD dimer in the asymmetric unit and a
solvent content of 51 %59. Similarly, the volume of the unit cell for the P2. form is
consistent with a 240 kD tetramer in the asymmetric unit and a solvent content of

50%”.

38

Table 1. Data-collection statistics for the two crystal forms of the native MIP

synthase.

Form It

Form IIt

 

X-ray Source

Space group
Wavelength (A)
Unit-cell parameters
a(A)
b(A)
C(A)
l3(°)
Mosaiclty (°)
Resolution (A)
Last resolution shell
Completeness (%)
Last resolution shell
Rama 1
Last resolution shell
l/c(l)
Last resolution shell

Rigaku RU200,
50 W, 100 mA

02
1.54

153.0
96.6
122.6
126.4
0.4
40—2.5
2.6-2.5
96.7
98.9
0.066
0.28
15.1
3.6

Rigaku RU200,
50 W. 100 mA

P2.
1.54

94.5
186.2
86.5
110.5
0.5
40-2.9
2.9-2.8
99.9
100.0
0.107
0.52
13.0
2.7

 

I This data was collected at home.
* Rmerge is defined as 2 lr..-<r.,>l/z I...

39

 

Figure 11. Crystal images. (a) MIP synthase crystal form I. Note: crystal
nucleated around a fiber. (b) MIP synthase crystal of form II.
(c) MIP synthase crystals produced via microseeding at a
dilution of 105-5.

40

Crystallization and Data Collection Analysis of the Se-Met MIP Synthase

In crystallizing the Se-Met MIP synthase, the crystallization conditions
were the same as the native protein form I. Due to the number of molecules in
the asymmetric unit and the difference in diffraction resolution, the C2 space
group (form I) was the sole crystal form pursued in all Se-Met structural
experiments. These crystals were taken to the synchrotron to utilize the SBC-ID-
19 Beamline at Argonne National Laboratory in Chicago, Illinois. The crystals
described above are also depicted in Figure 11c.

At the APS, with the help of R. Zhang, a three-wavelength MAD data set
was collected. In order to collect a more complete data set, a reverse beam
experiment was also performed at all three wavelengths. In all, six data sets were
collected (at 123K) using one crystal. As with the case of the native protein, the
crystal dimensions, space group, and cryoprotectant conditions were similar. The
Se—Met crystals diffracted toa little higher nominal resolution of 2.4 A. After
processing the data using DENZO, the two data sets for each wavelength were
merged together and scaled using the “SCALE ANOM” command in
SCALEPACK“. Both of the programs used were part of the HKL2000 suite

package. The results of the Se-Met data collection are described in Table 2.

41

Table 2. Data analysis for the Se-Met MIP synthase MAD phasing experiment.

it. 2.2 2.3
MA) 0.9794 0.9796 0.9464
No. of total reflections 90,842 93,132 91,546
No. of unique reflections 49,240 49,965 49,388
Percent complete 94.1 (92.4)‘3 95.0 (93.6) 94.5 (92.6)
R... (%) b 4.3 (26.7) 4.7 (32.1) 4.8 (33.8)
I/c(l) 12.7 (2.3) 14.2 (2.4) 12.6 (2.0)
Space group C2
Cell dimensions (A) a=153.00, b=96.43, c=122.66, y=126.10°

’ The parentheses denote those values for the last resolution shell.
b R3,... = XI I0 - <l> |/lo, where lo is the observed intensity and <l> is the

average intensity obtained from multiple observations of symmetry
related reflections.

42

A Primer in MAD Phasing

In solving a macromolecular structure, one needs to be able to determine
the phase angle. If a protein has anomalous scatterers in its molecule, the
difference in intensity between the Bijvoet pairs, I Fh (+) I 2 and I F..(-) I 2, can be
exploited to determine the phase angle. In the multiple wavelength method, the
wavelength dependence of the anomalous scattering is used. Similarly, a protein
should contain an element that gives a sufficiently strong anomalous signal.
Therefore, the elements in the upper rows of the periodic table are not suitable.
Hendrickson showed that the presence of one Se (selenium) atom in a protein of
not more than approximately 150 amino acid residues is sufficient for successful
applications of MAD”. One way to introduce Se into a protein is by growing a
microorganism on a seleno-methionine (Se-Met) substrate instead of a
methionine-containing substrate. Condition for application of this method is that
the wavelengths are carefully chosen to optimize the difference in intensity
between Bijvoet pairs and between the diffraction at the selected wavelengths“.

Mathematically, one can describe the most frequently occurring situation
where there is one type of anomalously scattering atom present. This was
described by Karle in 1980‘°°. In his theory, the nonanomalous scattering of all
atoms in the structure is separated from the wavelength-dependent part. Each
anomalously scattering atom has an atomic scattering factor off = fo + Af + if. In
Figure 12, F3 is the contribution to the structure factor by the nonanomalously

scattering atoms, F... is the nonanomalous contribution of the anomalously

43

scattering atoms, and the complete nonanomalous part is F3). = F3 + F... The

anomalous scattering contribution is:
Af/foxFA+if’/foxFA=a (1)
$3,. is the phase angle of F3... (I). of vector FA, and ta of vector a. A4) = (t... + (1800 -
4)..) = (180° + (it... - ¢a)). Through mathematical manipulations, one can derive the
following expression:
IFI2= IFBAI2 +PIFAI° + IFBAI + IFAI X [0008(033- ¢A) 1' TSIUIIIIBA' ¢A)I (2)
with
p = ((2.02 + (()2)/f,-°-, q = 2 AW... and r = 2 (If,
p, q, and r are functions of the wavelength and can be derived form the atomic
absorption coefficient. The IF I 2 values are different for Friedel mates but they
can be determined experimentally. The unknown quantities are I F3A| , I FAI , and
(t1).3A - In). all three independent of wavelength and equal for Friedel mates except
for the sign of (4).... - it...) Therefore, a data set at one wavelength gives two sets of
equations for these three unknowns and in principle, measurements at two
different wavelengths are sufficient to find I F3A| , I FAI , and (cl)... - 4)..) for each
reflection. To calculate the electron density map of the protein, it... is needed.
This is obtained by solving the A-structure, that is by locating the anomalously
scattering atoms from a Patterson map with coefficients I F... I 2 or by direct
methods. From the A-structure, (I... can be calculated and ¢3A from the known

value of (it... - ¢A).

 

At = [o.. + (180° - ¢,)] if lfo x FA

   
 

 

 

1. FBA = Fe + FA
2. a = Am. x i=A + i(f"/fo) x r=A J

 

Fe

 

 

 

Figure 12. The structure factor diagram for the reflection of a protein crystal that
contains one kind of anomalously scattering atoms. F3 is the
contribution to the structure factor by the nonanomalously scattering
atoms, F... is the nonanomalous contribution of the anomalously
scattering atoms, and the complete nonanomalous part is seen in
equation 1. Equation 2 is the anomalous contribution of the
anomalously scattering atoms. The anomalous component is
exaggerated in this figure. The dotted lines are for the mirror image
of the Friedel mate.

45

Because no anomalous scattering is taken into account for the calculation
of the A-stmcture, the real structure or its enantiomorph is obtained. The solution
of this problem is to calculate 4).. angles for both structures. This gives two sets of
()3... angles and two protein electron density maps from which the best one must
be selected. An advantage of this method is that nonisomorphism does not play
a role here. All data can be collected on a single crystal if the lifetime of the
crystal allows this.

Practically speaking, the MAD method requires that great care be
exercised in the collection and processing of the x-ray diffraction intensities,
because the intensity differences are rather small. The choice of wavelengths
should be such that the differences between I F(h k l) I and I F(-h -k -l) I as well
as the dispersive differences AF = < I F(/I,-) I > - < I HA.) I > should be optimized.
< I F()..) I > is the average of I F(h k I) I and I F(-h -k -I) I at wavelength ,1,- .

Since MIP synthase can be overexpressed in E. coli, growth of cultures in
minimal media containing Se-Met should produce protein fully substituted with
Se-Met. MIP synthase’s 11 methionines should give ample phasing power in a
MAD experiment when converted to Se-Met. However, since both crystal forms
contain more than one molecule in the asymmetric unit, the difficulty will be
locating these sites in Patterson maps. In the case of the more tractable C2
crystal form, 22 sites are expected from the two molecules in the asymmetric

unit.

46

MAD Phasing Experimental Results

After collecting and processing the three-wavelength MAD data sets, the
data was then brought back to MSU and the Se-Met sites were found using the
SOLVE package”. Through the automated search, SOLVE found 18 of the 22
sites expected from the amino acid sequence (11 per monomer and 22 per
asymmetric unit). As expected, the N-terrninal methionines were disordered,
leaving two Se-Met sites unaccounted for. The missing sites would be defined
later (non N-terrninal) in the crystal structure of the inhibitor-bound complex, as
these sites were located in the disordered substrate-binding domain of the
structure. The final statistics for the sites found by SOLVE are tabulated in Table
3. In addition, in Table 4, the actual atomic positions are listed along with their
relevant occupancies.

After finding the sites, the sites and the generated mtz file were then
employed in the calculation of a solvent flattened map using the CCP4
package“. Subsequent maps and symmetry equivalent sites were then

calculated and viewed using the TURBO-FRODO and 0 graphics programs“.

47

Table 3. Statistics for the Se-Met sites found by SOLVE.

M=FOM= |F(hkl)bestI/IF(hk|) I65

Figure of Merit with Resolution (FOM)
DMIN: TOTAL 8.52 5.54 4.38 3.74 3.31 3.00 2.77 2.58

N: 43525 2192 3746 4810 5662 6335 6758 7039 6983
MEAN FOM: 0.60 0.84 0.86 0.80 0.77 0.69 0.55 0.41 0.29

48

Table 4. Se-met atomic coordinates and occupancies for the sites found by

Site

10
11
12
13
14
15
16
17

18

SOLVE.

X
0.391
0.305
0.897
0.745
0.217
0.984
0.259
0.277
0.251
0.006
0.235
0.190
0.014
0.816
0.818
0.542
0.994
0.008

Y
0.999
0.034
0.315
0.400
0.223
0.299
0.223
0.163
0.182
0.334
0.192
0.227
0.297
0.366
0.384
0.119
0.200
0.312

Z
0.210
0.289
0.116
0.380
0.184
0.289
0.349
0.171
0.133
0.288
0.335
0.286
0.249
0.041
0.077
0.352
0.010
0.026

49

Occupancy

0.877
0.725
0.765
1.002
0.802
0.837
0.705
0.775
0.502
0.890
0.896
0.684
0.693
0.609
0.538
0.918
0.678
0.299

Height/o
20.0
19.8
16.4
15.6
17.5
19.3
17.6
16.6
15.0
15.2
15.3
11.4
14.8
15.1
13.8
12.7
10.3
8.00

Refinement Analysis for the Native Structure

Once the 18 Se-Met sites were found, heavy atom refinement and
calculation of an interpretable electron density map was performed using
SOLVE. After solvent flattening, the map was then traced using 0 and TURBO-
FRODO. Subsequently, using CNS, refinement began after tracing was
completed63. Early rounds of refinement consisted of annealing, rigid-body
calculations, simulated annealing omit maps, minimizations, and group b-factor
refinements. After multiple rounds of calculations, the later rounds mainly
consisted of minimizations, individual b-factor refinements, and bulk solvent
corrections. Some 2-fold averaging was performed to see if any of the disordered
regions were better defined. Unfortunately, 2-fold averaging did not help the
quality of the maps. Similarly, 3Fo-2Fc maps were also calculated to improve the
phasing of the disordered regions. Contrary to the averaged maps, these maps
helped resolve 2 disordered loops in the determination of the final structure.
Once the R-factor was reasonable (R/Rm 25% and 28%, respectively), waters
were then added using both the CNS and TURBO packages. The final model at
2.4 A has an R and Rm of 20.5% and 24.3% respectively. The final model
contains residues 10-351 and 410-533 for both molecules in the asymmetric unit
and 543 water molecules. All but 6 (1.1%) residues lie within the most favored or
allowed regions of the Ramachandran plot. The parameters evaluated by
PROCHECK were well within the bounds established from well-refined structures
at the equivalent resolution“. The final refinement statistics are shown in Table

5.

50

Table 5. Refinement statistics for the native MIP synthase structure.

Refinementa
Resolution range (A) 1002.4
R/R... (%)” 205/243
Number of waters 543

R.M.S. Deviation
Bond angles (°) 1.9270
Bond lengths (A) 0.0095
Average B-factor (protein) (A2) 24.3

a Data collected at the SBC lD-19 beamline at Argonne National
Laboratory.

bR={£I IFobsl - IFcach I/ElFobsl}and Rm={2I IFobsI ' IFcach I/XIFobsl}r

where all reflections belong to a test set of 10% randomly selected data.
Also, all refinement statistics were calculated with a 20 cutoff.

51

STRUCTURAL RESULTS AND DISCUSSION

The Structure of the MIP Synthase Tetramer

MIP synthase is a homotetramer both in solution and in the solid state
(Figure 13)2' 56. As shown in Figure 13, the tetramer has 222 symmetry with a
non-crystallographic two-fold axis relating the two molecules in the asymmetric
unit and a crystallographic two fold axis relating the two molecules at the top of
Figure 13 (the red and blue monomers) with those at the bottom (the two green
monomers). The interface between the non-crystallographic dimer (the
dimerization interface) is both large and intricate, burying 11,700 A2 of surface
area between the two. Interactions are made throughout the molecule, from the
top of the structure to the bottom. The surface is composed mostly of
hydrophobic residues making it unlikely that there is any dissociation of the

tetramer. The interactions along the tetramer interface are listed in Table 6.

The Structure of the MIP Synthase Dimer
Though not as involved as the dimerization interface, the interface
between the dimers is also significant, burying around 6,000 A2 of surface area.
A front view of the dimer is shown in Figure 14. All the interactions in this
interface emanate from the juxtaposition of two large beta sheets, one from each
dimer. Most of the interactions among the dimer interface are hydrophobic in
nature with some tight salt bridges. All of the dimer interactions are listed in Table

7.

52

Table 6. Native MIP synthase tetramer interactions.

095A—-----Y166B
K97A-------L164B
E98A<------>Q126B
Q162A------K97B
E165A<---->K98B
Y166A-------QQ5A
Y349A<- ----- >Y349A
M415A ----- Y349A
E41 7A<------>Y349A

H433A<---->E417A

QQ5B-------Y1 66A

K978 L164A

 

E98B<------>Q1 62A

 

Q1 62 B------K97A
E1 6SB< >K98A
Y1 66B------095B

Y3493<.....-..>Y349B
M41 58—-----Y3498
E417B<- ----- >Y3493

H433B<------>E41 7B

<-----> Indicates a hydrogen bond interaction

------ Indicates a hydrophobic interaction

53

Table 7. Native MIP synthase dimer interactions.

 

 

 

 

S11A< >R44B K101A< ------ >E421 B I341A---------G107B
V12A ----- F458 F106A ----- L422B L422A-------L440B
K1 3A< >D46B G1 07A ----- A339B L422A-------F106B
V14A ---- V47B S113A<---->D338B M423A<---->T443B
V15A ---- V47B L117A ------ F45B M423A<----->SB48
Y32A<---->E529B L117A - ------ V47B L424A-------F94B
E33A<---->K18B G118A ----- A35B L424A-------L1643
N34A<---->E529B G1 1 8A ««««« V37B L424A-- ----- P1 03B
A35A ------ l1 19B I119A ---------V36B G425A-------F94B
V36A ------- l1 19B I119A -- ----- V37B G425A------L4403
V37A ----- G118B N124A< ------- >H498B H427A< >E98B
V37A - ------ V126B V126A ------- V37B N428A<---->N4348
R44A<----->S1 1B A128A --------- F45B R429A<----->H433B
F45A ------ V12B L164A -------- L424B R429A<---->N434B
D46A<---->K13B L328A ————— F3353 R429A <---->V435B
V47A ----- V14B V331A -------- F 335B S431A <----->H433B
V47A -------- L117B L332A ------ L328B I431A I4323
V47A ------ V15B F335A ------ L328B H433A<------>S431 B
P49A -------- V15B D338A<------>R507B N434A<------>R42QB
<---------> Indicates a hydrogen bond interaction

------------- Indicates a hydrophobic interaction

 

F56A-------F528B
F94A-------L424B
C436A< >N428B
L440A------L422B
L440A------G426B

T443A<---->M423B
F477A-------L532B
Y478A<----->R531 B
T482A<----->E5303
T482A<---->R531 B
R494A<------>E53OB
H498A<------>N1 248

Table 7 (cont’d)

A339A—----G1 07B
G340A-----F1 06B
L503A------F33SB

N504A< >N5043

 

K505A<----->E525B
R507A<----->D3383
T508A-------E5253
N51 2A<---->N524B

F513A----------L526B
F513A----------L526B
L520A---------V47B

8522A<------->8522B

 

 

V435A< >R428B
C436A< >H427B
N524A< >Y3OB

 

E525A<---->K5053
L526A-------F 51 3B
K527A<----->Y3ZB
F 528A--—---F 56B
E529A<----->Y32B
E530A<----->R4943

R531 A<----->T482B

<-------> Indicates a hydrogen bond interaction

----------- Indicates a hydrophobic interaction

55

The Structure of the MIP Synthase Monomer

The MIP synthase monomer can be divided into three major domains,
each with its own distinct function (Figure 15). A central domain consisting of the
N- and C-terrnini make the majority of contacts between monomers across the
non-crystallographic two-fold and stabilize the relative orientation of the other two
domains. A domain reminiscent of an NAD-binding or Rossmann fold domain
(encompassing residues 66 - 326) contains a parallel five stranded beta sheet,
four surrounding helices, and 2 additional extensions, one that completely
surrounds the adenine of NAD preventing ready dissociation of the nucleotide
(residues 149-215) and one insertion between the first and second strands of the
Rossmann fold (residues 93-140) that folds into the central domain of the
enzyme. The third domain (encompassing residues 327 - 441) contains the

beta sheet involved in the tetramerization interface and the catalytic domain.

56

 

Figure 13. Ribbons model of the MIP synthase tetramer. The top two monomers
are colored red and blue, make up one asymmetric unit, and are
related by a non-crystallographic two-fold axis that is roughly in the
plane of the page. The bottom two monomers are green and are
related to the top two monomers by a crystallographic two-fold axis
running roughly perpendicular to the page.

57

 

Figure 14. A front view of the native MIP synthase dimer.

58

 

Figure 15. Ribbons model of the MIP synthase monomer. The various regions
are colored: red for the N-terminal region from residue 10-65; purple
for the NAD-binding domain encompassing residues 66-326; green
for the catalytic or tetramerization region containing residues 327-
441 (with residues 352-409 disordered); blue, the C-terminal region
containing residues 442-533. Note: the N-terminal, C-terminal, and
an insertion from the NAD-binding regions are intimately associated.
Together, they make up the central domain of the protein.

59

Numerous interactions between these three domains create a rigid overall
structure with little possibility for relative motion between these domains.
Inspection of Figure 2 indicates that though all eukaryotic MIP syntheses are
highly conserved throughout their length and will almost certainly have very
similar structure throughout, a significant portion of the N-terrninus is missing in
the archae enzyme. This will lead to differences in the structure of the central
domain of this enzyme relative to that seen here. As mentioned earlier, the fold of
MIP synthase bears some resemblance to that of diaminopimilate
dehydrogenase (DAPDH), especially the structure and relative position of the
catalytic and Rossmann fold domainssg' 7° (Figure 16). DAPDH is a dimer in
solution and shares a similar dimerization interface with MIP synthase as well.
Interestingly, the insertion domain (Figure 16) in DAPDH corresponds to the
disordered part of the native MIP synthase structure where some of the active

site residues are located.

NAD Binding
Interaction of MIP synthase with NAD is similar to that seen in other
Rossmann fold NAD-binding enzymes. NAD runs across the bottom of the
parallel beta sheet of the nucleotide-binding domain. Instead of the more
common GXGXXG motif, MIP synthase has a GXGGXXG motif (GLGGNNG)
(encompassing residues 72-78) in the loop connecting the B3 strand and a1
helix. This loop makes interactions with the phosphodiester backbone as seen in

other structures. The amide of the nicotinamide is rotated to the phosphodiester

60

side and makes a tight hydrogen bond with the phosphodiester backbone serving
to stabilize this orientation. Numerous interactions between NAD and the enzyme
are observed throughout the nucleotide-binding and catalytic domains of MIP
synthase. In turn, most of these residues are highly conserved amongst the
eukaryotic enzymes. The nucleotide binding domains of the archaebacterial
enzymes are less highly conserved and many of the interacting residues are
different in these enzymes. The significance of these differences must await the
structure determination of these enzymes. A complete listing of NAD-binding
residues is shown in Table 8. As shown in Table 8, the differences in sequence
between the archaebacterial and yeast forms of the enzyme are striking. Of the
28 interacting residues, 13 are different in the archae form (46% of the interacting
residues). Interestingly, as stated in the previous section, residues 149-215
completely encapsulate the NAD and prevent it from dissociation. The residues
involved in NAD-binding and encapsulation are listed in Table 9. Noticeably,
there are several differences in the binding from yeast to archae. In fact, of the 9

residues, 6 are different in the archae form (67% of the interacting residues).

61

Table 8. Residues that bind NAD in both the yeast and archae form of MIP
synthase as seen in the holo structure.

 

 

 

 

 

 

 

 

 

 

 

 

62

 

Invariant least ELIE?
I71 G72 > S13

G75 G74 > T15
W147 N76 > M17
D148 N77 > V18
I149 S184 > G98
R160 N194 > S105
T244 0195 > G106
A245 D196 > I107
T247 E197 > K108
S296 W243 > N146
P297 N246 > S149
D320 L321 > G224
K322 $323 > T226
D438
A442

Table 9. A list of the insertion residues that completely encapsulate the NAD in
both the yeast and archae forms as seen in the holo structure.

 

 

 

 

 

Invariant Legit _A_r_qh_a_g
I149 S184 > G98
N150 I185 > T99
R160 N194 > S105

0195 > G106
D196 > I107
E197 > K108

 

63

DAPDH Monomer

   

Catalytic Floor Domain

   
   

MIP Synthase Monomer

Missing Residues and

Catalytic Floor Domain DAPDH'hke Insertion

Domain

Figure 16. The DAPDH monomer (PDB accession code 1DAP) (top) in
comparison to the holo MIP synthase monomer (bottom).

64

MIP Synthase and its Unique Rossmann Fold

Through an exhaustive search of the PDB Data Bank, a list of the various
Rossmann-fold motifs was compiled. After inspection of the list, MIP synthase
seems to have a unique Rossmann-fold. No other Rossmann-fold was found to
have a 5,4-motif with two insertions (5 B-sheet — 4 a-helix, or more commonly, B-
a-B-a-B-a-B-a-B motif). The search revealed that Rossmann-folds can be
classified into three categories: 5,4-motif with one (or none) insertion, the typical
6,5-motif, and a 7,8-motif. There is no other known structure with a 5,4-motif and
two insertions. The closest ancestor is the 5,4-motif with one insertion. This type
of motif is seen in the enzyme aldehyde dehydrogenase (ALDH) (pdb accession
code 1AD3). ALDH is interesting in that it seems to have two Rossmann-folds
per monomer, even though there is only one NAD molecule per monomer. Figure

17 summarizes the results of the Rossmann-fold search.

Conclusions

The 2.4 A structure of MIP synthase as purified from E. coli71 revealed
partial NAD occupancy in the NAD-binding domain and allowed us to define the
catalytic region of the enzyme based on the position of the nicotinamide.
Surprisingly, the nicotinamide was completely exposed on the surface of the
structure with no discernable active site cleft or cavity surrounding it (Figure 18).
Additionally, we were unable to locate electron density for 58 residues, from 352
- 409. These residues represent the most conserved region of the enzyme,

sharing 73% identity from archae to humans. The beginning and end of this

65

region are located near the putative catalytic region defined by the position of the
nicotinamide. In an effort to shed light on this unusual combination of
observations, we grew crystals of MIP synthase in the presence of 10 mM dgtolP

(Figure 3).

66

 

Aldehyde Dehydrogenase (ALDH) DHQ Synthase (DHQ)
1AD3

  
 
  

5,4 motif
3 Roam arm
0 insertions

 

Oxldoreductase (OXRED) Sultollpld Blosynthesls Protein ($001)
1 D74 ORR

   

6,5 motif
1 Rossmann
0 insertions

Undecaprenyl Pyrophosphate Synthase (UPS)
1JP3

Figure 17. A thorough Rossmann-fold search results. These structures represent
the types of Rossmann-folds as seen in the PDB Data Bank.

67

 

Figure 18. A space-filling model of MIP synthase in its substrate-unbound form.
The protein atoms are red while the NAD atoms are green.

68

MATERIALS AND METHODS

Expression, Purification, and Characterization for Native MIP Synthase

After analysis of an earlier protocol for the overexpression and purification
of MIP synthase, modifications to the sequence of events leading up to the
formation of pure native protein were made. Included in the earlier procedure
were the following: an (NH4)2804 precipitation step, dialysis, a long DEAE
column (10 x 30 cm, 2 L gradient), BioGeI column (10 x 150 cm, 6-8 hours), and
a Blue A Affinity column“: 5‘. Use of this protocol yielded 3 mg/L protein, poor
protein purity, as well as time consuming. This prep was clearly not enough for a
crystallographic study on the enzyme. In turn, modifications were made to both
improve the purity and to increase the yield. The following protocol for
overexpression and purification of native yeast MIP synthase was implemented.

Previous studies yielded the cloned and purified plasmid (pT-7-7/MIPSYN)
and upon initiation of this project, the plasmid was given to us via the Frost lab at
Michigan State University. Yeast MIP synthase was overexpressed in E. coli in
the efficient BL21 (DE3) overproducing strain.

The transformation protocol was as follows: After thawing out an aliquot of
BL21 (DE3) competent cells, 10 ng (1 ttL) of the pT-7-7/MIPSYN plasmid was
added to an eppendorf tube. After tapping this tube to mix it, the tube was then
cooled on ice for 30 minutes. The cells were then heat shocked at 42°C for 90

seconds and then cooled for 2 minutes on ice. After cooling on ice, 80 pL of LB

media was added and the tube was incubated at 37°C for 40 minutes. After the

69

incubation period, the cells were then plated onto LB/AMP plates in volumes of
20 uL and 80 uL and then incubated at 37°C overnight. Glycerol stocks of viable
colonies were used initially, but stopped after further tests revealed that fresh
transformations yielded more proficient cells.

MIP synthase was then expressed in LB media following a 3-hour
induction with 60 mg/L of IPTG (Isopropy-B-D-thiogalactopyranoside, dioxane
free). Cell pellets were re-suspended in buffer A (20 mM NH4CI, 10 mM Tris, 10
mM B-mercaptoethanol (BME)) in a volume proportional to 2 mL buffer A per
gram of dry cells and stored at —80°C.

After thawing on ice for 30 minutes, the cells were sonicated in (3) one-
minute intervals and centrifuged at 4,000 rpm for 10 minutes. The resulting
supernatant was purified to homogeneity using 4 chromatography steps. (i)
Phast-Q chromatography (Pharrnacia): protein was eluted using a linear salt
gradient (high: 300 mM NH4CI, 10 mM Tris, 10 mM BME, low: 20 mM NH4CI, 10
mM Tris, 10 mM BME). The fractions containing MIP synthase were pooled and
diluted with Buffer A to lower the salt concentration. (ii) Anion-exchange
chromatography: diluted fractions from (i) were applied to a SOURCE-Q column
(8 mL total volume, Pharrnacia) in Buffer A (20 mM NH4CI, 10 mM Tris, 10 mM
BME, 10% glycerol). Elution with a linear gradient from Buffer A to Buffer B (1 M
NH4CI, 10 mM Tris, 10 mM BME, 10% glycerol) gave peak fractions, which were
pooled and concentrated to 2 mL for gel filtration. (iii) Gel filtration
chromatography: after concentration, the protein was applied to a gel filtration

column (Superdex 200 16I75, Pharrnacia) and eluted in the original Buffer A

70

(without glycerol). MIP synthase eluted from this column as a 240 kD tetramer,
consistent with previous reports“: 5‘. (iv) Blue A affinity chromatography: pooled
fractions from (iii) flowed through the affinity resin to remove the residual
indogenous glucose 6-phosphate dehydrogenase that co-purifies with MIP
synthase. The enzyme was then concentrated to 10 mglmL for crystallization
experiments.

In concentrating MIP synthase, there were a few complications. The
original prep used high-pressure Amicon filters for concentration, which yielded a
loss of 50-70% of the protein. When these filters were used with the new preps,
the protein precipitated out of solution. As a result, to minimize loss and to
prevent precipitation, the Amicon Centriprep (Centriprep-30) and Centricon
(Centricon-30) concentrators were implemented. These filters increased the final
yield of protein and were less expensive. In getting the protein to 10 mglmL, one
has to be careful. The protein will fall out of solution at 20 mg/mL. This is the limit
of concentration. Concentration of all steps in the purification, considered this to
help reduce the loss of protein. Similarly, after each concentration, a loss of
about 30-50% per run was noted.

MIP synthase was further characterized by an enzyme activity assay“: 56.
This assay involved incubation of MIP synthase with 5 mM glucose 6-phosphate
and 1 mM NAD. Aliquots were removed every two minutes, added to TCA,
incubated with aqueous NalO. for 1 hour at 37°C, and quenched with Nazsos.
Released inorganic phosphate was determined colorimetricallyss. Defining one

unit of activity as 1 pM of MIP per minute at 37°C per milligram of enzyme, our

71

number of 0.26 uM min'Img'1 compares well with typical literature values of 0.22
uM min'Img‘1 56. The present purification gives as much as a three-fold
improvement (from 3 mg/L to 9 mg/L) in purification over previous methods and

produces enzyme of superior purity (Figure 19)“ 56.

(a)

64 kD
MIP Synthase

 

Phast O Source 0 Superdex Original
200 purification

Figure 19. Purification of MIP Synthase. (a) SDS-PAGE gel of MIP synthase
after each chromatographic step. (b) Purity of native MIP synthase
using the previous method.

72

Overexpression and Purification of Se-Met MIP Synthase

In order to produce Se-Met protein, initial experiments centered on the use
of minimal media (including individual amino acid supplements) as the source for
culture growths“ 6‘. Unfortunately, these experiments rarely yielded functional
protein (1 out of every 20 L of cell culture). All growths resulted in cell lysis
presumeably attributed to the significant change in pH upon amino acid
supplementation. The active range of MIP synthase in solution ranges from 7-
8.4. While in minimal media, the pH was 8.4, on the high end of the range thus
conceivably causing the lysis. Although some overexpression of MIP synthase
was obtained using the amino acid supplement method, it was not nearly as
much as one would have hoped for in a crystallographic experiment. As a result,
a different media (or an auxotrophic strain) was needed to help reduce the high
toxicity levels caused by the minimal media.

The new media, called 2X M9 media, was an offshoot of the often-
successful minimal media. In this media, all amino acid supplements were
withdrawn. As seen in Table 10, the only constituents of the media were a
mixture of salts, trace vitamins, and glucose. Other additives include 100 mglmL
carbenicilin and most importantly, the Se-Met dissolved in BM KOH. Carbenicilin
was chosen over ampicilin to increase the potency of the antibiotic.

To make one liter of media, the solids were dissolved in 800 mL of ddH20,
20 mL of the 100X Vitamin solution (Gibco) were added, and finally, the proper
amounts of the salt solutions. To bring the final volume of the media to one liter

double distilled water (ddHZO) was added. One important thing to remember is

73

that the vitamin solution is light sensitive so it must be protected by wrapping it in

aluminum foil and storing it at 4°C.

Table 10. The make-up of the M9 media for Se-Met overexpression.

Solids
(NH4)2S04
NazHPO4
KH2P04
NaCl
Solutions
10 mM FeCI3
10 mM ZnCl2
100 mM CaCI2

Se-Met in 8 M KOH
100 mglmL Carbenicilin

so mglmL IPTG

100x Basal Eagle Medium Vitamin
20 % Glucose

Amount in 1L
49
13.49
69
49

200 pL (autoclaved)
200 uL (autoclaved)
2 mL (autoclaved)

1 mL (autoclaved)

100 mg dissolved in 3-4 drops of KOH
1 mL (sterilize filtered)

1 mL (sterilize filtered)

20 mL
20 mL (autoclaved)

The overexpression of Se-Met MIP synthase is a two-day experiment and

the following procedure should be implemented:

(a) Day 1 — Make all the desired media. Once this is done,
autoclave the media in the following manner: Place 5 mL of
culture into 10 mL test tubes, 100 mL into 250 mL flasks, and 1
L into Fembachs. For each liter of growth, the must be one 5 mL
overnight and one 100 mL secondary culture.

74

(b) Once all of the solutions are autoclaved, set-up the overnight
cultures by taking the autoclaved 5 mL of media and add the
following to it: 200 uL of 20% glucose, 200 uL of the vitamin
solution, 5 LL of carbenicilin, and 2-3 colonies of MIP synthase
form a fresh transformation. Grow overnight at 37°C and for at
least 18 hours.

(0) Day 2 — When the overnight is a “cloudy-white” color, prepare
the media for the secondary growth: To the 100 mL pre-
autoclaved media, add 2 mL of 20% glucose, 2 mL of vitamin
solution, and 100 LtL of carbenicilin. Once these are added,
take the 5 mL overnight and add the entire culture directly to the
100 mL secondary growth and grow at 37°C until an 0055.. of
0.5-0.6. This usually takes 5-6 hours.

(d) During the intermittent period, prepare the 1 L growth: To the
pre-autoclaved Fernbach, add 20 mL of 20% glucose, 20 mL of
the vitamin solution, and 1 mL of the carbenicilin.

(e) When the right cell density is reached, add 50 mL culture to the
1 L of media and allow the new culture to grow until an 00553 of
0.5-0.6 at 37°C. This will take anywhere from 3-4 hours.

(f) At the right 00553,, you are ready to induce. This is the most
critical step. Add to the culture 1 mL of IPTG and 100 mg of Se-
Met dissolved in a few drops of 8 M KOH. Once the Se—Met is
added, grow for 3 hours at 37°C. After three hours, resuspend in
Buffer A and follow the same protocol for purification, as
described earlier. For the Se-Met preps, DTT was substituted
for BME in all buffers used in the purification. This was done to
help maintain the pH of the environment. DTT is a better
reducing agent.

Typical yields for the Se-Met MIP synthase prep range from 20 to 25

mglL, a 2-fold increase from the native prep. As shown in Figure 20, a

comparison between the minimal and 2X M9 growths is presented in terms of a

gel. As seen in the gel, overexpression in the 2X M9 media is higher than that in

the minimal media (Lane 2 as compared to Lane 3 in Figure 20). To show

incorporation of the Se-Met, a mass spectrum of the substituted protein was

taken by the Macromolecular Structure Facility in the Biochemistry Department at

Michigan State University. From the resulting mass spectrum, the difference in

weight between the Se-Met and the native MIP synthase correlates to the

75

amount of Se-Met incorporated. In this case, the percent incorporation was about

96%.

Modifications to the Purification Protocol as a Result of Se-Met
Overexpression

In lieu of such a high yield of Se—Met protein, subsequent native preps
involved the incorporation of the 2X M9 media. Instead of the typical LB media,
native MIP synthase can now be overexpressed in 2X M9 media with a higher
yield. Within this new expression system, a few modifications were made to the
purification procedure.

The first modification involved substituting DTT for BME for the same
reasons as stated in the Se-Met overexpression protocol. Similariy, the
SOURCE-Q step was eliminated from the purification. Since the protein was pure
enough to run over a gel filtration column after the Phast-Q column, the
SOURCE-Q step was unnecessary. Finally, in the Phast—Q step, a higher
gradient was used (high: 750 mM NH4CI, low: 20 mM NH4CI). The native protein
in 2X M9 media seemed to bind more tightly to the Phast-Q resin, thus
necessitating the increase in concentration of the high-end salt. Other than these

changes, the purification protocol remained the same.

76

65 RD
60 kD

 

Lane 1: Kaleidoscope molecular weight standard
Lane 2: 2X M9 overexpression

Lane 3: Minimal medla overexpreeelon

Lane 4: Source 0 for mlnlmal medla

Lane 5: Source 0 for 2X M9 medla

Figure 20. A comparison between the minimal and 2X M9 media growths
in the overexpression of Se-Met MIP synthase.

77

Multiple Isomorphous Replacement (MIR) Experimental Results

Though obtaining reproducible and well diffracting crystals of a protein or
macromolecular complex is critical to solving the x-ray structure, reasonably
accurate stmcture factor phases must also be determined to calculate an
interpretable electron density map. In solving a macromolecular structure, the
first method one might try is a heavy atom MIR experiment. Since reproducible
and well diffracting crystals were easily obtainable, an attempt to get an
isomorphous heavy atom derivative by screening a heavy atom library was
made. This was pursued before any attempt in using MAD phasing to solve the
native structure.

For this technique to work, most, if not all, of the protein molecules within
the crystal must be identically bound by a heavy-atom-containing compound (i.e.,
Hg, Sm, U, etc.), with essentially no change in the structure of the crystalline
lattice”: 98. This is usually accomplished by soaking pre-grown crystals in a wide
variety of heavy atom reagents at various times and concentrations, while
occasionally it is necessary to co-crystallize the protein with the heavy atom
reagent.

Quite a large number of reagents have previously been used to produce
satisfactory heavy atom derivatives: these include the lanthanides, actinides and
uranyls, as well as mercury, gold, and platinum containing compoundsez' 98. While
uranyl acetate, the lanthanides, and actinides are considered “hard” ions that
form primarily ionic interactions with proteins, the mercurials, platinates, and

aurates are “soft” ions that primarily react with the sulfhydryls of cysteine

78

residues, the deprotonated nitrogens on histidine residues, or occasionally with
the sulfur atom of a methionine residue. Since the interaction is covalent, the
latter compounds tend to bind a little more specifically and are often better
ordered than derivatives produced with “hard” ions.

Through an exhaustive screen, it was determined that the MIP synthase
crystals were highly reactive toward all mercurial compounds tested.
Concentrations of reagents as low as 1 uM crack the crystals irreversibly and
severely compromise the diffraction pattern. With six cysteine residues per
monomer and twelve per asymmetric unit, this is not surprising given the well-
known reactivity of mercurials with sulfhydryls”. The most likely explanations for
this behavior are: (1) one or more of the substitutions causes debilitating
changes in intermolecular crystal packing interactions or (2) the rate of the
substitution is too fast for othenrvise acceptable and subtle lattice changes to
keep up, leading to lattice destabilization and discontinuity. Table 11 summarizes
the results of the most promising heavy atom derivatives obtained via soaking of
the C2 crystals.

Using very low concentrations of HgClz in ten-hour soaks significantly
alters the intensity differences between native and derivatized crystals. Complete
data sets on these crystals have resulted in overall intensity differences of about
20% on I when compared to the native data. Though the resulting difference
Patterson maps were noisy and difficult to interpret, as many as nine unique
heavy atom positions were located using the SOLVE automated heavy atom

search program“.

79

Unfortunately, all of these sites were of relatively low occupancy resulting
in a derivative of relatively low phasing power. Though this derivative could be
useful in locating heavy atoms in more complicated derivatives using the
difference Fourier synthesis, it did not produce a map of sufficient quality to use
in determining envelopes for noncrystallographic symmetry averaging, even after
solvent flattening. These sites were found using the MLPHARE program“. The
atomic positions and occupancies for the mercury derivative sites are described

in Table 12.

80

Table 11. Heavy atom screening results for the most promising derivatives.

Derivative
p-CMB

Mercury (II) acetate

Concentration

1mM

1mM

Mercury (ll) chloride 0.2 mM

Mercury (II) chloride 2.0 pM

Samarium (ll) acetate 1.0 mM

K2PtCI4

KPtCI3

1.0 mM

1.0 mM

Uranium (II) acetate 1.0 mM

KAUCI4
K2Pt(CN)4
KAu(CN)2

1.0 mM
1.0 mM

1.0 mM

Lg
ON
ON
ON
ON
ON
ON
ON
ON
ON
ON

ON

Resolution (A)

7.0

7.0

3.8 wk.

3.5 str.

No diffraction
No diffraction
7.0

No diffraction
No diffraction
3.0

No diffraction

ON = overnight, wk. = weak, str. = strong.

81

Deforrnities

No cracks
Degraded
Slight crack
No cracks
No cracks
Yellow color
Yellow color
Green color
Yellow color
Degraded

No cracks

Table 12. Positions and occupancies for the HgCIz derivative.

Site

$0.3M

X
0.3371

0.9488
0.9264
0.7327
0.2874
0.1043
0.2304
0.7405

Position

Y
0.0017

0.1328
0.1204
0.3117
0.1507
0.0843
0.1862
0.4786

2
0.0412
0.5712
0.1816
0.1141
0.4836
0.0939
0.2528
0.3737

82

Occupancy
0.32
0.22
0.18
0.22
0.34
0.17
0.21
0.31

B-factor
60.0
15.0
49.0
26.0
60.0
57.0
60.0
60.0

In order to help stabilize the lattice and try to get a more isomorphous
derivative, two experiments involving other heavy atoms were performed. The
first involved the use of bulkier mercurials. Sometimes, more bulky heavy atom
reagents will bind only to a subset of the sites that the smaller reagents bind. The
types of compounds that represent the bulkier mercurials that were tested were
mersalyl and phenyl mercurate derivatives. The use of the bulkier mercurials did
not help in finding an isomorphous derivative. The second experiment involved
controlling the rate of substitution. By gradually increasing the concentration of
the heavy atom soak over a longer period of time, the rate of substitution may be
more manageable. This procedure had been implemented both by gradual
addition over several days and by dialysis against higher concentrations of the
heavy atom derivative. Changing the rate of substitution did not improve the
isomorphism necessary for a good phasing derivative. Due to poor MIR results,

attempts in synthesizing a MAD phasing derivative were made.

An Attempt to Solve the P2. Crystal Form
Even though the C2 structure was solved, an attempt to solve the P2.
crystal form was also made. Once it was discovered that the active site was
disordered in the holo structure, interest in the enzyme turned toward solving the
other form. Using AMORE in the CCP4 package, a molecular replacement
solution was found“. Using the C2 monomer, the resulting solution contained the
four anticipated molecules that make up the expected tetramer in the asymmetric

unit. The correlation coefficient and the R-factor were 63.5 and 41.6%

83

respectively. Once the solutions were put into the right orientation using the
lsqkab script, maps using the native P2. data set and the AMORE tetramer
solution were calculated to 2.9 A. Unfortunately, there were two major collisions
among symmetry equivalent tetramers. Coincidentally, it was noted that there did
seem to be some potential structural changes both where the collisions occurred
and at the tetramer interface. In addition, partial NAD and a disordered active site
also existed as previously seen in the C2 structure. Conclusively, it was decided
that this crystal structure yielded no other important structural information
pertaining to the existence of the active site. As a result, experiments shifted the
focus on obtaining inhibitor/MIP synthase structures and no other work was done

with the P2. structure.

CHAPTER 3
THE INHIBITOR/MIP SYNTHASE STRUCTURE

EXPERIMENTAL

Crystallization and Data Collection Analysis for the Inhibitor-bound MIP
Synthase

In order to get a better understanding of how the complex transformations
used by MIP synthase work, an inhibitor-bound protein was also crystallized. The
first inhibitor used was 2—deoxy-glucitoI-6-phosphate (dgtolP) due to its tight
binding capability (K. = 6x1045 M) and its structural similarity to that of the acyclic
form of the substrate.

The crystals of the inhibitor-bound complex were produced via co-
crystallization with 20 mM NAD and 20 mM dgtolP. The crystallization conditions
were similar to that of both the native and Se-Met protein. The crystal used for
data collection at the SBC ID-19 Beamline (again with the help of R. Zhang) was
crystallized in 3% PEG 8000 and 100mM NaAc at a pH of 4.5. Interestingly, the
resolution of the inhibitor-bound complex was 0.2 A higher than that of the native
crystal. Diffraction was improved from 2.4 A to 2.2 A allowing for a more
complete atomic understanding of the structure. This data was also processed
using DENZO and scaled using SCALEPACK in the HKL2000 suite package.

The crystal statistics for the dgtolP complex are described in Table 13.

85

Table 13. Crystal statistics for the dgtolP/NADHIMIP complex.

Resolution (A) 2.2

No. of total reflections 76,517

Percent complete (%) 94.9 (96.3) a

Rsym (%) " 8.2 (30.4)

l/o(l) 13.9 (4.6)

Space group C2

Cell dimensions (A) a=152.73, b=98.31, c=121.86,
y=126.18°

a The parentheses denote those values for the last resolution shell.
" R5,... = XI lo - <l>I/lo, where I0 is the observed intensity and <l> is the

average intensity obtained from multiple observations of symmetry
related reflections.

86

Solving the Inhibitor Complex and Refinement Analysis

To solve the structure of the dgtolP inhibitor complex, a difference map (Fo
- F0) was calculated using the native pdb file and the reflection file from the
inhibitor complex crystal. After an initial round of refinement and calculation of a
difference Fourier map using CNS, clear electron density was evident for the
inhibitor, NADH, and all of the previously disordered region (residues 352-409) in
a 1.20 contoured map calculated at 2.5 A resolution. Subsequent rounds of
rebuilding, refinement, and resolution extension resulted in the final structure.
The refinement rounds consisted mainly of minimizations, individual b-factor
refinements, and bulk solvent corrections. The reduction of NAD to NADH and
oxidation of C5 to the ketone was discernable in the electron density map. The
final model contains residues 9-533, 618 water molecules, NADH, dgtolP, and
one ammonium ion (per monomer). It has an R and R-free of 20.8% and 27.6%
respectively. 97.8% of the entire structure lies within the most favored or allowed
regions of the Ramachandran plot. The parameters evaluated by PROCHECK
are well within the bounds established from well-refined structures at the
equivalent resolution. Table 14 summarizes the final refinement statistics for the
inhibitor-bound MIP synthase complex. Similariy, Figure 21 shows the final 2Fo -

Fc electron density map contoured at 1.20 near the active site.

87

Table 14. Refinement statistics for the dgtolP/MIP synthase structure.

Refinementa
Resolution range 10.0-2.2
R/R... (%)” 20.8/27.6
Number of waters 618

R.M.S. Deviation
Bond angles (°) 1.7580
Bond lengths (A) 0.0099

Average B-factor (protein) (A2) 32.2

a Data collected at the SBC lD-19 beamline at Argonne National
Laboratory.

bR={£I IFobsI ' IFcach I/ZIFobsI}and Rm={2I IFobsI ' IlecI I/XIFobsI}.

where all reflections belong to a test set of 10% randomly selected data.
Also, all refinement statistics were calculated with a 20 cutoff.

88

 

Figure 21. Final 2Fo — Fc electron density map contoured at 1.20 showing the
dgtolP location within the active site. The inhibitor is in red, active
site residues in green, and NADH/NH. are in blue.

89

STRUCTURAL RESULTS AND DISCUSSION

The Structural Changes in MIP Synthase that Occur Upon Inhibitor Binding
The inhibitor/MIP structure yields a lot of structural information. Most
importantly, the active site becomes ordered upon inhibitor binding. All of the
residues in the disordered region (352-409) are cleariy evident in the electron
density map. These residues create an entire subdomain that folds around the
inhibitor, completely encapsulating it (Figure 22). Once this subdomain is
ordered, there is no access to the interior of the active site, even for a solvent
molecule (Figure 23). Thus a rather large cavity is created within the structure of
the enzyme. One could conclude that the formation of the MIP active site
represents an extreme example of induced fit where the substrate nucleates the
folding of its own active site and its complete encapsulation within the enzyme.
Though there is little change in the structure of the rest of MIP synthase upon
inhibitor binding, the loop that includes residues 191-198 is an exception, as the
folding of the new catalytic domain forces this loop to flip out and create space
for the newly ordered region (Figure 24). Several new contacts are made
between this loop and the newly ordered subdomain as shown in Figure 24.
Similarly, several new contacts are introduced between the
crystallographic dimers that make up the tetramer and the monomers that make
up the dimer. These new interactions are also concentrated along the
tetramerization interface as previously reported in the native structure discussion.

The new tetramer and dimer interactions are listed in Table 15 and 16. Also,

90

listed in Table 17 are the new interactions between the resulting NADH
nicotinamide ring and the new substrate-binding domain. Interestingly, 60% of

the new interacting residues are different in the archae form of the enzyme.

91

 

 

 

N350A< >Y349A
H351A< >Y349A
D397A------P407A
H398A< >M405A

C399A<---->M405A
V401 A--------V401 A
K403A<---->Y349A
M405A<---->C399A
P407A------V344A

V408A------—--Y41 9A

A

 

Table 15. New tetramer interactions formed upon inhibitor binding.

 

N35OB< >Y3493
H351 B<---->Y3498

D397B—-----P407B

 

H398B< >M4053
C3993<---->M405B
V401 B-------V401 B
K403B<----->Y3498
M405B<----->C399B
P407B--------V3448

V408B----------Y41 QB

> Indicates a hydrogen bond interaction

Indicates a hydrophobic interaction

92

Table 16. New dimer interactions upon inhibitor binding.

S383A<----->Y127B
N384A<---->T1 14B

I386A-------P1 298
L387A—------F 1 06B
L392A------F1 068

A

> Indicates a hydrogen bond interaction

 

Indicates a hydrophobic interaction

Table 17. Additional NADH interactions in both the yeast and archae forms of
‘ MIP synthase upon inhibitor binding.

 

Invariant ms; m
D356 R198 ------>E109
K369 N354 >D259

N355 ------>Y260

93

 

Figure 22. Ribbons depiction of inhibitor-bound MIP synthase. The amino acids
that were ordered in the holo form of the enzyme are green while the
newly ordered residues (from 352-409) are red.

94

 

Figure 23. Space-filling model of the inhibitor-bound MIP synthase. The inhibitor
dgtolP is colored yellow, NADH is green, and the protein atoms are

red. Note that both dgtolP and NADH are completely obscured by the
enzyme.

95

A
Q
4

 

Figure 24. A view of the loop encompassing residues 191-199 that flips out upon
inhibitor binding. The structures of MIP synthase bound (red ribbons)
and unbound (green ribbons) by dgtolP are overlayed. Interactions
between this loop and the newly formed domain are shown. Dotted
lines denote hydrogen bonds. Residues are colored by atom type:
green, carbon; red, oxygen, blue, nitrogen.

96

The Interactions Between MIP Synthase and DgtoIP

The inhibitor dgtolP is bound to the enzyme in an extended conformation,
with the phosphate group in a transoid conformation relative to the inhibitor
carbon backbone (Figures 25 and 21). This conformation is consistent with
several of the inhibitor studies conducted. For one, though there are many
examples of exclusively acyclic substrate mimics that make good inhibitors of the
enzyme, no exclusively cyclic mimics of the substrate bind the enzyme with any
affinity”. As discussed earlier, the inhibitor studies suggest that MIP synthase
binds exclusively the acyclic form of the substrate, even though the equilibrium
favors the cyclic tautomer. Secondly, while 2-deoxy-D-glucitol 6-(E)-
vinylhomophosphonate, an E-vinyl-phosphonate substrate mimic, strongly
inhibits yeast MIP synthase, the Z enantiomer has no affinity for the enzyme”.
Given these data, it is not surprising to find dgtolP bound in an acyclic, extended
conformation with the phosphate in a transoid conformation. Consistent with the
biochemical data, dgtolP is found to be oxidized to the C5 ketone derivative and
the NAD to be reduced to NADH”. Evidently, we have a mimic of the first
intermediate in the reaction pathway, after the first oxidation, but before

enolization and intramolecular aldol cyclization.

97

 
      

' DgtolP

, 03

 

A

Figure 25. The interactions of dgtolP with the enzyme as seen in the structure.
Dotted lines denote hydrogen bonds. Atoms are colored by atom
type as in Figure 24 with the addition: magenta, phosphorous. Bonds
of the inhibitor are aqua. Only residues that make direct hydrogen
bonds to dgtolP are shown for clarity.

98

The dgtolP is well nestled in the cavity, with each of the hydroxyl groups
hydrogen bonded to a highly conserved residue of MIP synthase. In fact, all but
one of these residues is completely conserved in all MIP synthases. The
exception is 0325, which is conserved in all eukaryotic MIP synthases, but is
changed to E230 in the A. fulgidus enzyme. This residue interacts with 01 of the
inhibitor, which would be an aldehyde oxygen in the substrate. It is possible that
the substrate is bound in a less extended conformation, at least in the
archaebacterial enzyme, as the aldehyde would not hydrogen bond to the
glutamate.

A relatively large 40' peak of electron density was located proximal to the
phosphate group of the inhibitor and the residues D438 and N354 (Figures 25
and 21). This peak is not seen in the unbound structure as N354 is disordered.
Based on its proximity to both D438 and the phosphate, one could postulate that
this density is due to a monovalent cation, probably an ammonium ion. It is
believed that this density is an ammonium because ammonium ions increase the
rate of the enzyme five-fold. Similarly, ammonium ions are present in the
crystallization solution”. This putative ammonium ion is roughly tetrahedrally
coordinated to D438, N354, a phosphate oxygen, and 06 of the inhibitor. It is
interesting to note that while D438 is absolutely conserved in all MIP synthases
from archae to human, N354, while absolutely conserved in eukaryotic MIP
synthases, is occupied by D259 in A. fulgidus. One could postulate that the

addition of a second charged ligand changes the cation specificity from

99

monovalent to divalent, and may explain why the A. fulgidus MIP synthase
requires a divalent cation such as Zn” or an" for activity.

There are two other enzymes that have both NAD and a metal ion in their
active site: dehydroquinate synthase (DHQ synthase) and alcohol
dehydrogenase (ADH). When the nicotinamide rings of these two enzymes are
overlayed, the metal ion positions are almost identical. Further, when the
nicotinamide rings of DHQ synthase and MIP synthase are oveliayed, the
position of the ammonium ion is within 1.0 A of both the DHQ synthase and
alcohol dehydrogenase zinc atoms (0.89 A and 0.90 A respectively) (Figure 26).
This is in spite of the fact that none of these enzymes have any structural or
sequence similarity throughout their length. Given this positional similarity, it is
expected that the ammonium ion seen in MIP synthase has a role that is
analogous to that seen in the other two enzymes, which is to act as a Lewis acid
stabilizing negative charge on a keto oxygen atom.

DgtolP binding by MIP synthase involves both residues that were ordered
in both structures and additional residues that become ordered upon inhibitor
binding (Figures 25 and 2). While Q325, K412, D438, and K489 were ordered in
the unbound enzyme, K369 and N354 lie within the refolded subdomain (Figure
25). Five hydrogen bonds are made between the inhibitor and the enzyme: 0325
to 01 of dgtolP, K489 to O3, K412 to O4 and O5, and K369 to a phosphate
oxygen of dgtolP. The rest of the interactions are hydrophobic in nature and
define a hydrophobic core of the subdomain consisting of both enzyme and

inhibitor.

100

 

 

Figure 26. Overlay of the nicotinamide rings of alcohol dehydrogenase (orange),
DHQ synthase (cyan), and MIP synthase (blue). The spheres

represent the relative positions of the divalent metal (ADH and DHQ)
and the monovalent ammonium ion of MIP synthase.

101

Modeling of the Aldol Cyclization and a Proposal for the Mechanism of MIP
Synthase

The conformation of dgtolP in the dgtolP/MIP synthase structure is cleariy
not representative of the conformation of the substrate during the aldol
cyclization. A simple rearrangement of dgtolP was performed in an attempt to
mimic the conformation necessary for cyclization. To do this modeling, it was
assumed that the phosphate and C6 remain fixed while the inhibitor was
repositioned in a conformation consistent with the stereochemistry of the
reaction. Additionally, the 5-keto group was oriented with the nicotinamide of
NAD in a position consistent with the final reduction step. Similarly, it was
assumed that the structure of the enzyme is unchanged. The result of this
modeling is striking (Figure 27). Absolutely conserved residues now surround the
modeled inhibitor and make hydrogen bonds with all of the OH groups. In
addition, the 5-keto group is well positioned to interact with the ammonium ion
and K369 is in close contact with 01 of the inhibitor. Based on this modeling, two
mechanisms for the transformation catalyzed by MIP synthase can be proposed

In the first mechanism (Figure 28), the first step involves oxidation at C5
with the substrate molecule in an extended conformation similar to that seen in
the inhibitor complex structure. Subsequently, the substrate is reoriented to the
conformation shown in Figure 28 where the phosphate-mediated enolization
occurs. The developing negative charge on 05 is stabilized by the ammonium
cation to catalyze this process. Nucleophilic attack by C6 on C1 is promoted by
K369 stabilization of the developing negative charge on 01. Subsequent

protonation at 01, possibly mediated by K369, and reduction of C5 by NAD,

102

 

yields the 1L-myo-inositol 1-phosphate product. MIP synthase is therefore an
example of a type III aldolase, where a monovalent cation acts as the Lewis acid
in the aldol condensation. In the case of the archae enzyme, a similar
mechanism can be envisioned, except that the ammonium ion is replaced with a
zinc or manganese divalent cation, making this more characteristic of a type II
aldolase. All of the other residues interacting with the substrate and proposed to
be involved in the mechanism, are absolutely conserved in archae as well as
virtually all other MIP synthases whose sequences are known. Given the
presence of the divalent cation at this position, one would expect the residues
corresponding to 350-354 (MIP synthase) in archae (255 - 259) to be ordered
even in the absence of inhibitor or substrate. This would however, still leave
significant access to the active site when the remainder is disordered. Verification
of this proposal must await both mutational enzymology and further
enzyme/inhibitor structural results.

The other possible mechanism is the scenario in which the substrate
enters the active site in the alcohol form at C-5. When this occurs, the
protonation is through the mediation of D438. Although the phosphate still pulls
off the proton during enolization and the ammonium acts as a Lewis base
stabilizing the developing negative charge on C-5, the aspartate allows the
proton transfer to occur during aldol condensation. This mechanism is depicted in

Figure 29.

103

 

Is the New Domain Due to NAD or Substrate Binding?

Solving the dgtolP/MIP synthase structure inevitably invokes the question
of whether the presence of the new 58 residues is a result of the NAD or
substrate binding. Clearly, in the presence of partial NAD occupancy, none of the
substrate-binding domain residues is seen. Similarly, the loop encompassing
residues 191-198 is hovering over the NAD and not flipped out making contacts
with the new domain. However, what happens if there is complete NAD
occupancy without the presence of substrate? To answer this question, a
MIP/NAD crystal was made by soaking in 20 mM NAD into the MIP synthase
crystal lattice. A data set was collected and processed to 2.7 A. Upon calculation
of a F0 — Fc map, none of the substrate-binding domain residues was seen. Also,
the loop that flipped out upon substrate binding, was in its native position as seen
in the holo structure. Though diffusion of NAD into the lattice caused a
compromise in diffraction, a well-diffracting crystal was still attainable.
Conversely, co-crystallization of MIP with NAD was difficult. No crystals were
obtained via co-crystallization. A more complete crystallization experiment on this

complex would be beneficial to completely refute the NAD-binding argument.

104

 

Figure 27. Modeling of dgtolP in the active site of MIP synthase to a conformation
consistent with aldol cyclization and subsequent reduction at C5.
Atoms are colored by atom type as in Fig. 23. DgtolP bonds are aqua

and NADH bonds are gold.

105

D438 N354

_ ‘ ’,o/ NH2
K489 ‘~ '
V\+ .NI‘I4° K369
~63 ° °_\ + N
‘~ O'fl‘o'fi'wa
3323 -H0 \ x s -
H:° O 3 0
,0 .\ mo
H II
0
D438 N354
0A0. '0/ NH2
K489 ‘~ °
V‘nfi Nil-4° K369
\3 _ 1" “ + N
31 °HO 0%(3. ° - "N313
Sg/OH' °° \ O °°\~‘\O
‘. \ .n‘ ‘-
HO fi‘OH
O
:43: N354
0A0. ,0 NH2
K489 '1
+ .NH-l
W53 6 NH A99
‘\ 0' - :':- \ 3
31° ..-HO ‘~ -
°° H: Pit0\ .1650
0
NAD
NADH
HO HO OH OH
O\ .s\\\\\0
p.

HO "“0

Figure 28. Proposed mechanism for the transformation catalyzed by
MIP synthase.

106

D438 N354
W
NH2 Ho gm
., O

. Nil.’ . .
\ myo-inosrtol 1-phosphate (MIP)
HO" °

(9 K369 +

D438 N354
Lo OLNW
OH NH: ,.-'

N354
Litre /

 

Figure 29. An alternative mechanism for the catalysis of MIP synthase.

107

 

MIP Synthase, a New Mechanism for Induced-fit?

The concept of induced-fit where "the substrate causes an appreciable
change in the three dimensional relationship of the amino acids in the active site"
was proposed more than 40 years ago by Koshland75' 76. An exhaustive search of
domain movements in proteins categorizes all of these movements as either
hinge movements or shear movements". The structures of the NAD-bound and
inhibitor/NAD-bound MIP synthase, have revealed an extraordinary example of
induced-fit that complies with neither of these mechanisms. Instead, a folding
event occurs where almost fifty-eight residues become ordered upon binding to
the inhibitor. Once folded, the enzyme fully encapsulates its substrate, leaving no
access to the active site without unfolding some or all of this subdomain. Though
there are examples of ligand-induced structural organization, such as the
formation of two helices upon DNA binding in the leucine zipper proteins, it is
highly unusual to see a refolding event of the magnitude seen here in an enzyme
with a small molecule substrate” 79. This binding mechanism seems
counterintuitive, but allows the enzyme to completely encapsulate its substrate in
three dimensions. Encapsulation appears to be necessary for catalysis of this
complex transformation. It also invites the possibility that small molecules can be
used as protein folding scaffolds, both naturally and in protein engineering

applications.

108

MATERIALS AND METHODS

An Attempt in Solving Another Inhibitor Structure, the E-
vinylhomophosphonatelMlP synthase Complex

Once the dgtolP/MIP synthase structure was solved, an attempt to solve
the vinyl/MIP complex was made. The inhibitor, E-vinylhomophosphonate,
mimics the intermediate during the enolization step of MIP synthase catalysis
(Figure 10, mimic of intermediate B). A co-crystallization experiment was used
(same conditions as the dgtolP/MIP complex) to generate the crystal used for
data collection. After collecting data to 2.2 A of a vinyl/MIP complex at the SBC
Beamline, a difference map using the dgoIP/MIP synthase pdb and the vinyl
reflection file was calculated using CNS. Upon viewing the map, there was some
broken density for the vinyl inhibitor. Unfortunately, something happened to the
crystal during data collection, rendering a data set with non-discrete spots. As a
result, the data for this crystal had to be truncated to 2.9 A to get any value from
it. This truncation severely distorted any attempt to solve this structure. Another
data set of this complex was not collected again. A crystal structure of this
complex would definitely absolve any question on the conformation of the

substrate during the enolization and aldol condensation steps of catalysis.

109

CHAPTER 4
VALPROATE, A POTENTIAL THERAPEUTIC DRUG TARGET IN THE
TREATMENT OF MANIC DEPRESSION
GENERAL OVERVIEW

An Introduction to Bipolar Disease

Bipolar disorder, or manic-depressive illness, is a common condition with
a lifetime prevalence of 1-2%8°. It is characterized by recurring bouts of mania
and depression, which have deleterious effects on career and interpersonal
relationships. Approximately 15% of those afflicted commit suicide. Others suffer
with bouts of schizophrenia. Similariy, mortality rates are also increased due to
physical disorders“ 82. For decades, lithium has been the most effective agent
for the treatment of bipolar illness”. Despite the marked benefit that many
patients obtain from lithium therapy, 20-40% of patients fail to show a satisfactory
antimaniac response to lithium, and many patients suffer significant morbidity“.
More recently, the branched fatty acid valproate has been used for treatment of
bipolar disorder”. Like lithium, it is not completely effective, and the molecular
dynamics underlying its therapeutic effects have not been elucidated. Lithium
and valproate exert a variety of biochemical effects, only some of which are likely
to be related to their therapeutic mechanisms of action. Identifying common
targets of lithium and valproate is an approach that may more directly address

the therapeutic mechanisms underlying their efficacywm.

110

 

The lnositol Depletion Hypothesis

The inositol depletion hypothesis proposes that lithium acts by depletion of
inositol from the brain. This is based on the observed uncompetitive inhibition of
inositol monophosphatases by lithium, resulting in decreased inositol levels, an
increase in inositol phosphates, and subsequent down-regulation of the
phosphoinositide cycle". Because the brain obtains inositol primarily from
phosphoinositide turnover and de novo synthesis, it is highly sensitive to
fluctuations of the phosphoinositide cycle. Although there is considerable
evidence that lithium affects the phosphoinositide second messenger system, a
connection between this effect and the therapeutic mechanism of lithium has not
been established91'93. If inositol depletion formed the basis for the therapeutic
effect, then valproate might be expected to deplete inositol, as well. Previous
studies have shown that valproate does not inhibit bovine brain or yeast inositol
monophosphatase activity, nor does it have an effect on receptor-mediated
phosphoinositide tumoverm’e. Additionally, valproate does not lead to large

accumulations of inositol mono- or bisphosphates, as seen with lithium”.

Recent Data Published on Valproate Inhibition of MIP Synthase

It has been shown by Vaden et.al., that both lithium and valproate have a
profound effect on inositol metabolism in the eukaryote S. cerevisae53. Both
drugs, in therapeutically relevant concentrations, cause a decrease in
intracellular inositol mass and an increase in expression for a structural (INO1)

and a regulatory (INOZ) gene required for inositol synthesis. The mechanism of

111

 

 

inositol depletion by lithium, is most likely due to the inhibition of inositol
monophosphatase, as previously reported”. On the other hand, the mechanism
of inositol depletion by valproate is caused by the inhibition of myo-inositol 1-
phosphate synthase”.

Furthermore, a 40 pM inositol concentration, completely reverses the 35-
fold valproate-mediated increase in INO1 expression. Conversely, this
concentration only partly reverses a smaller lithium-induced increase in IN0153.
Interestingly, both lithium and valproate down-regulate the level of sodium-
dependent high affinity myo-inositol transporter in astrocyte-Iike cells. The effects
of these drugs on inositol uptake and transport in yeast have not been
determined, although preliminary data suggests that they do not cause
decreased inositol uptake”.

In terms of a mechanism, Vaden proposes that valproate leads to
decreased inositol levels by inhibition of the myo-inositol 1-phosphate synthase
pathway”. Several experiments demonstrated that inositol monophosphate
levels are reduced by valproate but not by lithium. Inhibition by valproate via this
pathway is further supported by the observation that the opi1 mutant does not
exhibit increased resistance to valproate, despite constitutive expression of INOt
and increased levels of MIP synthase. Previous findings that valproate does not
inhibit inositol monophosphatase or cause an accumulation of inositol
phosphates has been cited as evidence against the inositol-depletion

hypothesis“ 97. Vaden’s experiments conclusively suggest the mechanism of

112

inositol depletion by valproate is by inhibition of the rate-limiting step in the de
novo synthesis of inositol, via the catalysis of MIP synthase.

Other plausible mechanisms for inositol-depletion include the down-
regulating expression or activity of protein kinase C (PKC)53. PKC is highly
enriched in the brain and plays a major role in regulating, pre- and post-synaptic
aspects of neurotransmissionm. It is now know to exist as a family of closely
related sub-species, has a heterogeneous distribution in the brain, and plays a
major part in he regulation of neuronal excitability, neurotransmitter release, and
long-term alterations in gene expression and plasticity”. To date, only a few
studies have directly examined PKC in bipolar disorder. Although there is no
evidence for PKC expression in yeast, valproate does decrease mammalian
protein kinase C expression”. Protein kinase C appears to be required for
inositol synthesis, as yeast protein kinase C mutants are inositol auxotrophs.
Therefore, a valproate-mediated decrease in protein kinase C would result in
decreased intracellular inositol and thus a further reduction in synthesis of
inositol.

Vaden’s results present one with an interesting dilemma. If valproate does
inhibit the MIP synthase pathway at therapeutic levels, a potential therapeutic
drug design target may have been discovered. Unfortunately, Vaden did not
present any data as to the K. value for valproate inhibition. If the K. shows tight
binding of valproate, crystallographic experiments may be warranted. Since no
inhibition data was available, inhibition studies were initiated to find out what type

of inhibition was present and the K. value for valproate inhibition of MIP synthase.

113

 

Before assays were done to determine the inhibition constant, a quick glance
was taken into the active site of the dgtolP/MIP synthase complex to try to
determine if there was a potential valproate-binding site. Interestingly enough,
there was an identifiable hydrophobic pocket, consistent with the chemical

properties of valproate.

 

114

ASSAY RESULTS AND DISCUSSION

Potential Valproate Binding Site in MIP Synthase
Valproate (2-propyI-pentanoic acid), at least chemically, seems to mimic
the structure of the glucose 6-phosphate. As seen in Figure 30, valproic acid has
a negatively charged carboxcylic acid end, which could conceivably mimic the
phosphate group of the glucose 6-phosphate. The other end of the molecule is
highly hydrophobic consisting of a branched carbon backbone. As a result, one
can envision that the presence of a highly hydrophobic pocket in the active site

would serve as a viable binding site for the valproate molecule.

 

Valproic Acid

0 O

 

 

 

Figure 30. Chemical structure of valproic acid.

Coincidentally, a large hydrophobic surface consisting of residues L352,
L360, I400, and I402 aligns one side of the binding cavity. These hydrophobic
residues are completely conserved throughout evolution. Structurally, this

hydrophobic pocket is depicted in Figure 31. In order to substantiate this claim

115

and to see if the relevant crystallization experiment was worthy, activity assays to
determine the K. for valproate inhibition were performed using the Ames

colorimetric method as described earlier”.

Assay Results
After determining the activity constants for each trial of inhibitor
concentration, plots were made to determine the K. for valproate inhibition. From
these plots (Figures 32 and 33), a K. of 0.7 mM was calculated. To determine
what type of inhibition was present, a Lineweaver-Burke plot was made

(Equation 3).

1/v = [K31[1 + [l]lK.]Nmax x 1/[S]] + 1Nmax (3)

When 1/v was plotted against 1/S, all lines intersected at a single point
with each concentration of inhibitor (Figure 32), consistent with competitive
inhibition. To further verify a competitive inhibition mode, a Hanes-Woolf plot was
made (equation 4). In this plot, for competitive inhibition, lines at various
concentrations of inhibitor are parallel with each other. As shown in Figure 33, all

lines were parallel to each other, signifying competitive inhibition.

[SI/V = Km“ T [II/KIINmax + 1[Vmax X [S] (4)

116

This technique was also tested against previous inhibition assays to
determine its validity. Those experiments yielded no significant deviation from
previous assays on those inhibitors with published K.’s. For example, when an
inhibitor assay was done with dgtolP on MIP synthase with the technique used in

the valproate inhibition assay, a K. of 5.5 pM was calculated. This value is in

agreement with the reported value of 5.8 p.M by the Frost group.

117

 

Figure 31. Potential valproate binding site. The map is contoured at 1.26.

118

1lv versus 1IS

Competitive Inhibition

1lv (min mg pM")

No Inhibitor

 

 

o I T T I I
0 0.0002 0.0004 0.0006 0.0008 0.001

11s (pM'1)

K.- — [l]/[1- ((Vmax/Kmxslopelll

Figure 32. A double reciprocal plot to determine K. and type of inhibition.

119

Slv (min mg)

Slv versus 8

25000 -

20000 4

15000 I

10000 a

   
 

5000 l o Inhibitor

 

 

0 1000 2000 3000 4000 5000

8 (0M)

Intercept = Km(app)/Vmax

 

Figure 33. A plot of SN versus S to determine the KM/Vmax value for the
competitive inhibition equation used in calculating a K..

120

MATERIALS AND METHODS

Activity Assay Experimental to Determine the K. for Valproate Inhibition
In order to determine the K. for valproate inhibition, several plots are
necessary. In this case, a simple 00595 versus time plot was used to determine
the activity constant, an S/v versus 8 plot to determine the KMNmax, and a double
reciprocal plot of 1/v versus 1/8 to determine the type of inhibition and the final
K.. Similarly, as listed in Table 18, are all the components necessary for the
assay experiment. The protocol used for the experiment went as follows:

(a) Incubate the enzyme assay solution, specific G6P
concentration, and specific NAD+ concentration at 37°C for ten
minutes.

(b) Take out 100 uL of the assay solution and add 50 uL of 20%
TCA to a new eppendorf tube. This is the time point at time 0.

(c) Add enzyme and inhibitor and reincubate for three minutes at
37°C. The typical amount of enzyme used was about 0.4-0.5
mg. The inhibitor concentrations ranged from 0 to 250 pM.

(d) After 3 minutes, take out another 100 uL sample and again, add
to another new tube containing 50 pL of 20% TCA.

(e) Do this for 30 minutes taking 100 pL samples out every 3
minutes.

(f) After the 30-minute period, add 100 pL of 0.2 M Nal04 to each
of the tubes containing a time point.

(9) Vortex the tubes and incubate them at 37°C for 30 minutes.

(h) After 30 minutes. add 150 pL of 1.5 M Na2803 and 600 pL of
the P. assay solution.

(i) Vortex the tubes again and incubate them at 42°C for 20
minutes.

(j) After 20 minutes, take the CD of each tube at 595 nm.

(k) Note: The tubes should gradually get a blue tint, which indicates
the release of the phosphate from the product MIP. This is what
is measured.

121

Table 18. Colorimetric assay components used in the determination of the K. for
valproic acid inhibition.

Enzyme Assay Solution 20% TCA
0.2 mM DTT 0.2 M Nal04
2 mM NH4C| 1.5 M Nast3

50 mM Tris HCI, pH=7.7

GGP
Various concentrations, ideal value, 5 mM

NAD+
Various concentrations, ideal value, 1 mM

P. assay solution

6:1 ratio of ammonium molybdate-Malachite green reagent to
ascorbic acid

122

REFERENCES

. Arganoff, B. W. and Fisher, S. K. (1991) Inositol Phosphates and Derivatives,
Edited by A. B. Reitz, pp.20-32. Washington D.C.: American Chemical
Society.

. Majumder, A. L., Johnson, M. D., and Henry, S. A. (1997) Biochim. Biophys.
Acta. 1348(1-2), 245-256.

. (a) Loewus, M. W., and Loewus, F. A. (1973) Plant Sciences Letters. 1, 368-
371. (b) Mauck, L. A., Wong, Y. H., and Sherman, W. R. (1980) Biochemistry.
19(15), 3623-3629. (c) Pittner, F. and Hoffmann, O. O. (1978) Hoppe Seyers
Z. Physiol. Chem. 359(10), 1395-1400. (d) Sherman, W. R., Loewus, M. W.,
Pina, M. 2., and Wong, Y. H. (1981) Biochim. Biophys. Acta. 660(2), 299-305.

. (a) Abu-Abied, M. and Holland, D. (1994) Plant Physiol. 106(4), 1689. (b)
Dean-Johnson, A. and Henry, S. A. (1989) J. Biol. Chem. 264(2), 1274-1283.
(c) lshitani, M., et al. (1996) Plant J. 9, 279-293. (d) Johnson, M. D. (1994)
Plant Physiol. 105(3), 1203-1204. (e) Johnson, M. D. and Burk, D. H. (1995)
Plant Physiol. 109, 721. (f) Smart, C. C. and Fleming, A. J. (1996) Plant J. 4,
279-293. (9) Wang, X. and Johnson, M. D. (1995) Plant Physiol. 110, 336.

. Altschul, S. F., et al. (1997) Nucleic Acids Res. 25, 3389-3402.

. (a) Loewus, M. W., Bedgar, D. L., and Loewus, F. A. (1984) J. Biol. Chem.
259(12), 7644-7647. (b) Loewus, F. A., and Kelly, S. (1962) Biochem.
Biophys. Res. Commun. 7, 204-208.

. (a) Belmaker, R. H., Berdsudsky, Y., Agam, G., Levine, J., and Koffman, O.
(1996) Annu. Rev. Med. 47, 47-56. (b) Berridge, M. J., Downes, C. P., and
Hanley, M. R. (1982) Biochem. J. 206, 587-595. (c) Berridge, M. J., and
Irvine, R. F. (1989) Nature. 341, 197-205.

. (a) Koffman, 0., et al. (1991) Psychophannacol. Bull. 27(3), 185-190.
(D) Koffman, 0., and Belmaker, R. H. (1990) Brain Res. 534(1-2), 345-347.

. Parthasarathy, R., and Eisenberg, G. J. (1991) Inositol Phosphates and
Derivatives, Vol. 463, 1tt 50., Edited by A. B. Reitz, pp. 1-19. Washington
DC, American Chemical Society.

10. Exton, J. H. (1986) Adv. Cyclic Nucleotide Protein Phosphorylation Res. 20,

211, 262.

11. Klig, L. S., and Henry, S. A. (1984) Proc. Natl. Acad. Sci. U. S. A. 81(12),

381 6-3820.

123

12.

13.

14.

15.
16.

17.

18.

19.

20.

21.

22.

23.

24.

25.

26.

27.
28.

Griac, P., Swede, M. J., and Henry, S. A. (1996) J. Biol. Chem. 271(41),
25692-25698.

Hirsch, J. P., and Henry, S. A. (1986) Mol. Cell Biol. 6(10), 3320-3328.

Klig, L. S., Homann, M. J., Carman, G. M., and Henry, S. A. (1985) J.
Bacteriol. 162(3), 1135-11341.

Greeenberg, M. L., and Lopes, J. M. (1996) Microbiol. Rev. 60, 1-20.
Nikoloff, D. M., and Henry, S. A. (1994) J. Biol. Chem. 269(10), 7402-741 1.

Hoshizaki, D. K., Hill, J. E., and Henry, S. A. (1990) J. Biol. Chem. 265(8),
4736-4745.

White, M. J., Hirsch, J. P., and Henry, S. A. (1991) J. Biol. Chem. 266(2),
863-872.

Klig, L. S., Hoshizaki, D. K., and Henry, S. A. (1988) Curr. Genet. 13(1), 7-
14.

Slekar, K. H., and Henry, S. A. (1995), Nucleic Acids Res. 23(1 1), 1964-
1969.

Lamping, E., Luckl, J., Paltauf, F., Henry, S. A., and Kohlwein, SD. (1994)
Genetics. 137(1), 55-65.

Hudak, K. A., Lopes, J. M., and Henry, S. A. (1994) Genetics. 136(2), 475-
483.

Yoshida, K. T., Wada, T., Koyama, H., Mizobuchi-Fukuoka, R., and Naito, S.
(1999) Plant Physiology. 119, 65-72.

Johnson, M. D., and Wang, X. (1996) J. Biol. Chem. 271 (29), 17215-17218.

Nelson, D. E., Koukoumanos, M., and Bohnert, H.J. (1999) Plant Physiology.
119, 165-172.

Nelson, D. E., Rammesmayer, G., and Bohnert, H. J. (1998) Plant Cell. 10,
753-764.

Drobak, B. K., (1992) Biochem. J. 288(Pt. 3), 697-712.

Lohia, A., Hait, N. C., and Majumder, A. L. (1999) Mol. Biochem. Parasit. 98,
67-79.

124

29.

30.
31.

32.

33.

35.

36.

37.

38.

39.

40.

41.

42.

43.

44.

Pittner, F ., and Hoffmann-Ostenhof, O. (1979) Mol. Cell Biochem. 255(18),
23-26.

Maeda, T., and Eisenberg, F. Jr. (1980) J. Biol. Chem. 255(18), 8458-8464.

Mauck, L. A., Wong, Y. H., and Sherman, W. R. (1980) Biochemistry. 19,
3623-3629.

Hasegawa, R., and Eisenberg, F., Jr. (1981) Proc. Natl. Acad. Sci. U. S. A.
78, 4863-4866.

Bersudsky, Y., Kaplan, Z., Shapiro, Y., Agam, G., Kofman, 0., and
Belmaker, R. H. (1994) Eur. Neuropsychopharmacol. 4(4), 463467.

. Benjamin, J., Agam, G, Levine, J., Bersudsky, Y., Kofman, 0., and

Belmaker, R. H. (1995) Psychophannacol. Bull. 31(1), 167-175.

Belmaker, R. H., Bersudsky, Y., Benjamin, J., Agam, G, Levine, J., and
Kofman, O. (1995) Adv. Biochem. Psychophamracol. 49, 67-84.

Fux., M., Levine, J., Aviv, A., and Belmaker, RH. (1996) Am. J. Psychiatry.
153(9), 1219-1221.

Hussain, A., Bourgeois, J., Polvi, S., Tsang, E., Keller, W. A., and Georges,
F. (1997) Submitted.

Fischer, 0., and Eisenberg, D. (1996) Protein Science. 5, 947-955.

Frey, PA. (1987) in Pyridine Nucleoside Coenzymes Part 8., Vol. 2, Edited
by D. Dolphin, R. Poulson, and O. Avramovic, pp. 461-511. New York, John
Wiley & Sons.

Carpenter, E. P., Hawkins, A. R., Frost, J. W., and Brown, K. A. (1998)
Nature. 394, 299-302.

Thoden, J. B., Frey, P. A., and Holden, H. M. (1996) Biochemistry. 35, 5137-
5144.

Bauer, A. J., Rayment, |., Frey, P. A., and Holden, H. M. (1992) Proteins:
Struct. Func. Genet. 12, 372-381.

Thoden, J. B., Frey, P. A., and Holden, H. M. (1996) Biochemistry. 35,
2557.

Migaud, M. E., and Frost, J. W. (1995) J. Am. Chem. Soc. 117, 5154-5155.

125

45.

46.

47.

48.

49.
50.

51.

52.

53.

55.
56.

57.

58.
59.
60.

61.

Loewus, M. W., Loewus, F. A., Brillinger, G. U., Otsuka, H., and Floss, H. G.
(1980) J Biol. Chem. 255, 11710-11712.

Horecker, B. L., Tsolas, 0., and Lai, C. Y. (1972) The Enzymes, Vol. 7,
Edited by PD. Boyer, pp. 213-258. New York, Academic Press.

Faber, K. (1997) Biotransfonnations in Organic Chemistry. Springer-Veriag,
Berlin.

Littlechild, J. A., and Watson, J. C. (1993) Trends in Biochemical Science.
18, 36-39.

Dreyer, M. K., and Schultz, G. E. (1996) J. MOI. Biol. 259, 458-466.

Shaman, W. R., Loewus, M. W., Pina, M. Z., and Wong, Y. H. (1981)
Biochim. Biophys. Acta. 660(2), 299-305.

Creighton, T. E. (1993) Proteins, Structure and Molecular Properties. New
York, W.H. Freeman and Co.

Wierenga, R. K., Terpstra, P., and Hol, W. G. (1986) J. Mol. Biol.
187(1),101-107.

Vaden, D. L., Ding, 0., Peterson, 8., and Greenberg, M. L. (2001) J. Biol.
Chem. 276, 15466-15471.

Donahue, T. F., and Henry, S. A. (1981) J. Biol. Chem. 256(13), 7077-7085.
Ames, B. N. (1966) Methods Enzymol. 8, 115.
Migaud, M. E., and Frost, J. W. (1996) J. Am. Chem. Soc. 118(3), 495-501.

Wlodawer, A., and Hodgson, K. O. (1975) Proc. Natl. Acad. Sci. U. S. A., 72,
398-399.

Otwinowski, Z., and Minor, W. (1 997) Methods Enzymol. 276, 307-326.
Matthews, B. W. (1968) J. Mol. Biol. 33, 491 -497.

Yang, W., Hendrickson, W. A., Kalman, E. T., and Crouch, R. J. (1990) J.
Biol. Chem. 265, 13553-1 3559.

Leahy, D. J., Erickson, H. P., Aukhil, l., Joshi, P., and Hendrickson, W. A.
(1994) PROTEINS, Stmcture, Function, and Genetics. 19,48-54.

126

62.

63.

65.
66.

67.

68.

69.

70.

71.
72.

73.

74.

75.
76.

77.

78.

Blundell, T. L. and Johnson, L. N. (1976) Protein Crystallography, Academic
Press, New York.

Terwilliger, T. C., and Berendzen, J. (1999) Acta Crystallogr. 055, 849-861.

Collaborative Computational Project, No.4. (1994) Acta Crystallogr. 050,
760-763.

Drenth, J. (1994) Principles of Protein X-ray Crystallography. Springer, 292.

Jones, T. A., Zou, J. Y., and Kjelgard, M. (1991)Acta Crystallogr. A47, 110-
119.

Brunger, A. T., Adams, P. D., Clore, G. M. DeLano, W. L., Gros, P., Grosse-
Kunstleve, R. W., Jiang, J. S., Kuszewski, J., Nilges, M., Pannu, N. 8.,
Read, R. J., Rice, L. M., Simonson, T., and Warren, G. L. (1998) Acta
Crystallogr. 054, 905-921.

Laskowski, R. A., Macarthur, M. W., Moss, D. S., and Thornton, J. M. (1993)
J. Appl. Crystallogr. 26, 283-291.

Scapin, G., Cirilli, M., Reddy, S.G., Gao, Y., Vederas, J. C., and Blanchard,
J. S. (1998) Biochemistry. 37, 3278-3285.

Scapin, G., Reddy, S. G., and Blanchard, J. S. (1996) Biochemistry. 35,
13540-13551 .

Stein, A. J., and Geiger, J. H. (2000) Acta Crystallogr. D56, 348-350.
Tian, F., Migaud, M. E., and Frost, J. W. (1999) J. Am. Chem. Soc.

Tian, F., Arthur, J. C., Peng, J., Migaud, M. E., and Frost, J. W. (1999)
Abstracts of Papers of the American Chemical Society. 217, U202.

Barnett, J. E. G., Rasheed, A., and Corina, D. L. (1973) Biochem. Soc.
Trans. 1, 1267.

Koshland, D. E., Jr. (1958) Proc. Natl. Acad. Sci. U. S. A. 44, 98-105.
Koshland, D. E. (1995) Angew. Chem. Int. Ed. Eng. 33, 2375-2378.

Gerstein, M., Lesk, A. M., and Chothia, C. (1994) Biochemistry 33,6739
6749.

Benevides, J. M., Li, T. 8., Lu, X. J., Srinivasan, A. R., Olson, W. K., Weiss,
M. A., and Thomas, G. J. (2000) Biochemistry 39, 548-556.

127

79

80.

81.

82.

83.

85.

86.

87.

88.

89.

90.

91.

92.

93.

94.

Pabo, C. 0., and Sauer, R. T. (1992) Annu. Rev. Biochem. 61, 1053-1095.

Weissman, M. M., Leaf, P. J., Tischler, G. L., Kamo, M., Bruce, M. L., and
Florio, L. P. (1988) Psychol. Med. 18, 141-153.

Angst, J. (1986) Psychopathology 19, Suppl. 2, 47-52.

American Psychiatry Association (1987) in DSM-lll-R: Diagnostic and
Statistical Manual of Mental Disorders (Williams, J. B. W., ed.) 3rd Ed., pp.
213-233, American Psychiatry Association, Washington, D. C.

Goodwin, F. K. and Jamison, KR. (1990) Manic-depressive Illness, pp.
665-674, Oxford University Press, New York.

Chou, J. C. (1991) J. Clin. Psychopharmacol. 11, 3—21.

Bowden, C. L., Brugger, A. M., Swann, A. A., Calabrese, J. R., Janicak, P.
G., Petty, F., Dilsaver, S. C., Davis, J. J., Rush, A. J., and Small, J. G.
(1994) JAMA 271, 918-924.

Chen, G., Huang, L.-D., Jiang, Y.-M., and Manji, H. K. (1994) J. Neurochem.
63, 2361 -2364.

Manji, H. K., Bersudsky, Y., Chen, G., Belmaker, R. H., and Potter, W. Z.
(1 996) Neuropsychopharmacology 15, 370-381 .

Manji, H. K., Chen, G., Hsiae, J. K., Risby, E. D., Masana, M. l., and Potter,
W. Z. (1996) J. Clin. Psych. 57, 34-46.

Chen, 6., Huang, L.-D., Jiang, Y.-M., and Manji, H. K. (1999) J. Neurochem.
72, 1327-1330.

Chen, G. Yuan, P.-X., Jiang, Y.-M, Huang, L.-D., and Manji, H. K. (1999)
Mol. Brain Res. 64, 52-58.

Kendall, E. D., and Nahorski, S. R. (1997) J. Phannacol. Exp. Ther. 241,
1023-1027.

Casebolt, T. L., and Jope, R. S. (1989) Bio. Psychiatry 25, 329-340.

Godfrey, P. P., McClue, S. J., White, A. M., Wood, A. J., and Grahame, S. D.
(1989) J. Neurochem. 52, 498-506.

Vadnal, R., and Parthasarathy, R. (1995) Neuropsychopharmacology12,
277-285.

128

95.
96.

97.

98.

99.

100

101.

102.

103.
104.

105.
106.
107.

108.

Murray, M., and Greenberg, M. L. (2000) Mol. Microbial. 36, 651-661.

Li. R., Wing, L. L., Wyatt, R. J., and Kirsch, D. G. (1993) Phannacol.
Biochem. Behav. 46, 323-329.

Dixon, J. F., and Hokin, L. E. (1997) Proc. Natl. Acad. Sci. U. S. A. 94,
4757-4760.

Petsko, G. A. (1988) in Diffraction Methods for Biological Macromolecules
Part A, Vol. 114, Edited by H. W. Wychoff, C. H. W. Hirs, and S. N.
Tinnshef, pp. 147-156. Orlando, Academic Press.

Hendrickson, W. A., Horton, J. R., and LeMAster, D. M. (1990) EMBO J. 5,
1665-1672.

. Kane, J. (1980) Int. J. Quant. Chem. Symp. 7, 357-367.
Barnett, J. E. G. and Corina, D. L. (1968) Biochem. J. 108, 125.

Barnett, J. E. G., Rasheed, A., and Corina, D. L. (1973) Biochem. J. 131,
21.

Loewus, M. W. (1977) J. Biol. Chem. 252, 7221.

Sherman, W. R., Stewart, M. A., and Zinbo, M. (1969) J. Biol. Chem. 244,
5703.

Kiely, D. E. and Sherman, W. R. (1975) J. Am. Chem. Soc. 97, 6810.
Chen, C, H. —J. and Eisenberg, F. J. (1975) J. Biol. Chem. 250, 2963.
Byun, S. N. and Jenness, R. (1981) Biochemistry 20, 5174.

Manji, H. K., Moore, G. J., and Chen, G. (2001) Brit. J. Psych. 178 (suppl.
41), $107-$119.

129