in“...

:.

a6.
.1

. on . . .
Ed .
3.5”...) at... .
.33....» .9. .5: . ﬂank“...
. . . .unuunﬁuﬁu.
mam?

g
.
4%“...
1 ..}-€

aw

. ..
.11"; t L.
inukg. ..
3. .1. l..’-b..l:.u.:..\;s|8£!l -
Ila: 5.2.5.3.. ‘0.is
. 3 I1 .3:
-2 -11... f 12.5. .
. .. .. . t . ......i....
- . A I 23A! .
.t... . ......n . 33 .2. .
‘I .
...... t...§.....!..u. x “a". 9...... .m:..:§tl.ﬁih
.
it... 2 52m: 2 1.x...aesﬁi. {gang n.4, i...
{via-2.4. .3..." D. a- .II. {A .1
520.... I. c. t. M“
1.. It}... . .2.
3....»

a... \\.3I£\z.l
$1,082....Itan9vﬁl

(3...!
{In .1}!

t: ,
.zﬁa...:huuu.ﬁ...¥1h§v n9...
2. . .: .. .
ti-..

4 -. 13.013!
duct-\‘Eiéi
till}... 3...
I {.5vﬁlhnquﬂuvthna1l
.35.: 2:... a
. h .V..... .

($9... is
1A.. 90;.) I
NHL-til}:

It....\.§'.\
Saliimwxbaxs.
£3. 5...
. 1.. .
$333.:

 

 

 

 

 

 

 

 

 

 

.3... .31....5... . .r: L... . .

 

 

This is to certify that the

dissertation entitled

The X-ray crystallographic structures of

branching enzyme and angiostatin
presented by

Marta Cristina Abad Rivera

has been accepted towards fulﬁllment
of the requirements for

Ph . D. degree in Chemistry

//» MW

 

Major professor II/

1

“/4 or

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 c:/C|FiC/DateDue.p65-p.15

THE X-RAY CRYSTALLOGRAPI—IIC STRUCTURES OF BRANCHING ENZYME
AND ANGIOSTATIN
By

Marta Cristina Abad Rivera

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
THE X-RAY CRYSTALLOGRAPHIC STRUCTURES OF BRANCHING ENZYME
AND ANGIOSTATIN
By

Marta Cristina Abad Rivera

X-ray crystallographic studies have been performed for the structure
determination of two proteins, branching enzyme and angiostatin. In these studies
isomorphous replacement, anomalous dispersion and molecular replacement methods
were used to calculate the electron density maps of the aforementioned proteins.

Branching enzyme is one of three enzymes involved in the biosynthesis of starch
in plants and glycogen in animals and bacteria. It has an important role in the
determination of the ﬁnal structures of starch and glycogen. This enzyme catalyzes the
cleavage of or-l ,4 glucosidic bonds and subsequently transfers this chain into the (Jr-1,6
position. The conversion of this linear polysaccharide into a branched network not only
makes starch and glycogen more reactive to both synthesis and digestion, it also assures
its solubility in the cell. Insoluble glycogen caused by mutations in the branching
enzyme gene (Glycogen Storage Disease type IV) is a lethal genetic disease for which no
clinical treatment is known.

Escherichia coli branching enzyme was crystallized and high-resolution data to
2.3 A resolution was collected. Phasing information was obtained using isomorphous
replacement and anomalous dispersion methods. This, in addition to four fold averaging,

led to the calculation of an electron density map. The structure shows that branching

enzyme presents the central (or/B) barrel catalytic domain that is conserved among
members of the a-amylase family of enzymes, to which branching enzyme belongs. In
addition. a mechanism for branching enzyme has been proposed based on sugar substrate
modeling and comparison of the branching enzyme structure with other members of the

(Jr-amylase family of enzymes.

Angiostatin is a protein that inhibits angiogenesis, a process in which new blood
vessels form from pre-existing ones. Three decades ago Dr. Judah F olkman proposed
that tumor growth and metastasis dissemination are angiogenesis-dependent processes,
and the idea of angiogenesis inhibitors for cancer treatment was born. Angiostatin was
catapulted to the forefront of anticancer treatment drugs when it was ﬁrst discovered in
the mid 19903. The structure of human angiostatin was determined by molecular
replacement at 1.75 A resolution. The structure revealed that all three kringle lysine-
binding sites contain a bound bicine molecule, while those of kringle 2 and kringle 3 are
cofacial. Moreover, the separation of the kringle 2 and kringle 3 lysine binding sites is
sufﬁcient to accommodate the or-helix of the 30 residue peptide VEK-3O found in the
kringle 2/VEK-30 complex. Together the three kringles produce a central cavity

suggestive of a unique domain where they may function in concert.

T0 my-Mother and Father

AKNOWLEDGMENTS

I thank my advisor Jim Geiger for his guidance, help, and support. I couldn’t
have asked for a better advisor. I was also honored to have the opportunity to work with
Dr. Tulinsky, a renowned scientist, and pioneer in the ﬁeld of protein crystallography.
Dr. Tulinsky thank you for letting me become part of your team and enlightening me with
your interesting tales about crystallography in the old days. I will like to Show my
appreciation to Dr. Raghuvir Ami for his help in the reﬁnement of the angiostatin
structure. I was also very fortunate to work with Dr. Preiss, an authority in the ﬁeld of
starch and glycogen biosynthesis, and who introduced me to the marvelous ﬁeld of the
glycogen biosynthetic enzymes. I am in gratitude to the work of Dr. Kim Binderup in
providing the puriﬁed branching enzyme protein used for crystallization. I will also like
to offer my appreciation to my good friend and collaborator Dr. Jorge Rios. Dr. Rios
collected the selenium methionine data set that was crucial for the determination of the
structure of branching enzyme.

We all know that graduate school is not easy but having a lab full of nice people
makes a huge difference. Thanks to a group of phenomenal lab mates Xiangshu, Stacy,
Erika, Sara, Aimee, Adam, Mike, Tyra, Michelle and Elena thank you to all of you who
always kept me laughing even while writing this dissertation. I will like to give special
thanks to Stacy Hovde who took the time to revise every single page of this thesis and
pretty much everything I wrote. Thank you Stacy for all your help, for being my spell

checker and over all for a wonderful friendship.

I also thank my husband for his support and encouragement. Jose, we started this
together and we are both ﬁnishing in triumph. I will also like to thank my parents and
brothers: Francisco, Juan and Emilio for their love, encouragement and unconditional
support.

I am in gratitude for the ﬁnancial support provided by the Hispanic Scholarship
Fund, MSU’s Equal Opportunity Fellowship and especially to the Bill Gates Foundation

who supported the last two years of my graduate school.

vi

TABLE OF CONTENTS

LIST OF TABLES
LIST OF FIGURES
ABBREVIATIONS

Chapter I: INTRODUCTION

1.1 Branching enzyme
1.1.1 Glycogen
1.1.2 Biosynthesis of bacterial glycogen
1.1.3 Branching enzyme
1.1.4 Truncated branching enzyme

1.2 Angiostatin
1.2.1 Angiogenesis
1.2.2 Angiostatin

1.3 Literature Cited

Chapter II: X-RAY STRUCTURE DETERMINATION
2.1 Branching Enzyme
2.1.1 Crystallization
2.1.2 Structure determination
2.1.3 Structure reﬁnement -
2.1.4 Materials and methods
2.2 Angiostatin
2.2.1 Crystallization and data collection
2.2.2 Molecular replacement and structure reﬁnement
2.2.3 Materials and methods
2.3 Literature cited

Chapter III: THE THREE DIMENSIONAL STRUCTURE OF BRANCHING
ENZYME
3.1 Overall structure
3.2 Structural differences among members of the a-amylase family
3.3 Residues associated with the GSDIV
3.4 Proposed mechanism
3.5 Electrostatic potential surface
3.6 Conclusions
3.7 Literature cited

Chapter IV: THE THREE DIMENSIONAL STRUCTURE OF ANGIOSTATIN
4.1 Overall structure of angiostatin
4.2 The electrostatic surface of angiostatin

vii

ix

xi

xvi

Wr—tu—t

18
18
21
3O

36
38
4O
49
53
56
56
56
63
65

67
73
85
88
92
96
97

98
101

4.3 Ligand speciﬁcity of the kringle LBS
4.4 Angiostatin binding to protein domains
4.5 The inter-Kringle disulﬁde bond

4.6 Conclusions '

4.7 Literature cited

APPENDIX
Appendix 4.1 Kg/Kg and Kg/inter Kg peptide interactions of Angiostatin.

viii

105
111
117
117
119

122

LIST OF TABLES

CHAPTER 1: INTRODUCTION
Table 1.1 Mutational studies performed in branching enzyme from maize

endosperm isoform 1] and E. coli.

Table 1.2 Various molecules involved in angiogenesis activation or inhibition

CHAPTER II: X-RAY STRUCTURE DETERMINATION
Table 2.1 Crystal parameters for the branching enzyme crystal.

Table 2.2 Statistics for the branching enzyme X-ray diffraction data
collection

Table 2.3 List of all the heavy atom compounds tried

Table 2.4 Phasing power of the mercury and selenium methionine derivatives

Table 2.5 Reﬁnement statistics of BE

Table 2.6 Crystal parameters for the angiostatin crystal

Table 2.7 Statistics for the Angiostatin X-ray diffraction data collection

Table 2.8 Reﬁnement Statistics of angiostatin

CHAPTER III: THE THREE DIMENSIONAL STRUCTURE OF BRANCHING
ENZYME

Table 3.1 Residues responsible for causing the GSDIV their location and

effect.

Table 3.2 Protein interactions with modeled substrate

15

20

38

39

44

47

50

58

58

62

87

91

CHAPTER IV: THE THREE DIMENSIONAL STRUCTURE OF ANGIOSTATIN

Table 4.1 Rmsd values of the supperposition of the Cor positions of individual 103
Kgs and the Kgs in angiostatin

Table 4.2 Summary of the Kg-Kg interactions and inter-Kg peptide Kg 1 18
interaction of angiostatin. The interactions are determined with a
cutoff distance of <40 A.

LIST OF FIGURES

Images in this dissertation are presented in color.
CHAPTER I: INTRODUCTION

Figure 1.1

Figure 1.2

Figure 1.3

Figure 1.4

Figure 1.5

Figure 1.6

Figure 1.7

Starch and glycogen are formed by D-glucose units linked by a-1,4
and a-l ,6 glucosidic bonds. Amylose and amylopectin are the two
molecules that form starch and glycogen.

Biosynthetic pathway of starch and glycogen synthesis

Chimeric enzymes constructed from mBEI and mBEII. The “X”
represents the inactive mutants. N-mBEI, C-mBEII and 01/131
represent the amino terminal, carboxylate terminal and all} barrel
of mBEI, respectively. The same nomenclature applies for mBEII.

Conserved catalytic residues in the a-amylase family of enzymes.
The residues enclosed by the boxes are involved in substrate
binding and the shaded residues are involved in catalysis. The
labels in the sequence alignment stand for E. coli for E. coli BE;
human for Homo Sapiens BE; mBEI and mBEII maize endosperm
BE 1 and II, respectively; isoa is isoamylase from Pseudomonas
amyloderamosa, a-Asp and a-Por are a-amylase from Aspergillus
Oryzae and Porcine Pancreatic and CGT is cyclodextrin
glucanotransferase from Bascillus C irculans.

a) Carboxylate lysine residue. b) The carboxylate lysine analog, 8-
aminocaproic acid (EACA).

X-ray structure of Kl-EACA. K1 is colored green and the residues
encompassing the LBS are shown in atom color (nitrogen, blue;
oxygen, red and carbon, green) and EACA is colored lavender.
Side chains are labeled using plasminogen numbering.

X-ray crystallographic structure of K2-VEK30. K2 is colored
green and the VEK3O peptide is colored cyan. Side chains are
labeled using plasminogen numbering.

CHAPTER II: X-RAY STRUCTURE DETERMINATION

Figure 2.1

Figure 2.2

Crystallization using the hanging drop vapor diffusion method

A monoclinic crystal of glycogen branching enzyme. The crystals

xi

11

22

24

25

37

37

Figure 2.3

Figure 2.4

Figure 2.5

Figure 2.6

Figure 2.7

Figure 2.8

have dimensions of 0.3 x 0.1 x 0.1 mm3.

A section of the electron density map. a) Initial experimental
electron density map b) Electron density map after four fold
averaging c) Final electron density map after reﬁnement.

An example of the ﬁnal 2F0 - Fc electron density map.

Ramachandran plot of BE. Showing one of the four molecules for
clarity.

A tetragonal crystal of human angiostatin (Kg1-3). The crystals
have dimensions of 0.7 x 0.7 x 0.4 mm3.

Ramachandran plot of angiostatin
An example of the ﬁnal 2Fo - F c electron density map of

angiostatin. The map is centered at residue W315 and also shows
residues H317, W325 and Y327.

CHAPTER III: THE THREE DIMENSIONAL STRUCTURE OF BRANCHING

Figure 3.1

Figure 3.2

Figure 3.3

Figure 3.4

Figure 3.5

Figure 3.6

Figure 3.7

ENZYME

Three dimensional structure of e. coli BE truncated at the amino
terminus at amino acid 113.

The elements of secondary structure in the three domains of BE.
The [3 sheets from the N and C terminals are identiﬁed with an N
and a C, respectively.

The amino acid sequence of the truncated BE with their respective
element of secondary structure.

There are four molecules in the BE asymmetric unit.

The reactions catalyzed by the members of the a-amylase family
of enzymes. a) a-amylase hydrolyses a-l ,4 bonds. b) isoamylase
cleaves a-1,6 bonds. c) CGT catalyzes the formation of
cyclodextrins and (I) BE catalyzes the formation of OH ,6 bonds.

X-ray structures of members of the a-amylase family of enzymes.

Comparison between domains of isoamylase and BE. The

domains in isoamylase have been rotated to match the orientation
ofBE.

xii

48

51

52

57

61

62

68

70

71

72

74

75

76

Figure 3.8

Figure 3.9

Figure 3.10

Figure 3.11

Figure 3.12

Figure 3.13

Figure 3.14
Figure 3.15
Figure 3.16

Figure 3.17

Figure 3.18

a) Superposition of the structure of isoamylase in blue onto BE
shown in gold. b) Superposition of the structure of a-amylase
depicted in lavender onto 2. coli BE.

The structures of a) (Jr-amylase, b) CGT and c) isoamylase are
overlaid onto BE also showing are the a-1,4 cleaved sugar and the
incoming sugar oriented to form the branch point. BE is shown in
red and (at-amylase, CGT and isoamylase in gray. This mimic was
based on the substrate and intermediate bound structures of other
members of the a-amylase family taking into account the unique
loop structure of BE.

Comparison of the loops that surround the (or/B) barrel cavity. BB
is shown in red, isoamylase in lavender, CGT in green and or-
amylase in blue.

The B domain lies between [33 and 0L3. a) The loop between [33
and a3 in BE is shown in red b) A comparison between the B

domain of (it-amylase in blue, isoamylase in lavender, CGT in
green and BE in red.

a) Residues involved in BE catalysis. b) Position of these residues
in the barrel

a) Superposition of the conserved residues from BE, isoamylase,
or-amylase and CGT not bound to substrate. b) Comparison
between the residues from BE and the ones from a-amylase apo
and substrate bound.

Residues responsible for causing the GSDIV.
Proposed mechanism for BE catalysis

a) Orientation of catalytic residues before substrate binding. b)
Proposed substrate and c) intermediate interactions by modeling of
the substrate and intermediate from CGT.

Proposed mode of action for BE catalysis. a) substrate binding b)
intermediate formation and 0) model of the position that the
incoming sugar must have to form the a-1,6 branch

Electrostatic potential surface picture of BE a) Looking down the
barrel and b) rotated 180°. The EPS calculation corresponds to
10kT/e for the blue color, -10kT/e for red and an EPS ~ 0 is white,
where 10kT ~ 6 kcal/mol.

xiii

78

79

80

82

83

84

86

89

90

93

94

Figure 3.19 EPS of members of the (it-amylase family of enzymes. The

CHAPTER IV: THE THREE DIMENSIONAL STRUCTURE OF ANGIOSTATIN

Figure 4.1

Figure 4.2

Figure 4.3

Figure 4.4

Figure 4.5

Figure 4.6

Figure 4.7

structures are oriented looking straight into the central barrel
domain.

Three different representations of the overall structure of
angiostatin. (a) Ribbon picture showing Kgl , orange; Kg2,
magenta; Kg3, cyan; inter-kg peptide between Kgl and Kg2, blue;
inter-Kg peptide between Kg2 and Kg3, green; bicines, green with
atoms in atom colors (nitrogen, blue and oxygen, red); intKg
disulﬁde, yellow. LBS side groups also in atom colors. (b) Space
ﬁlling view of angiostatin. The LBS in each of the three kringles is
colored gold. All other atoms are red. (c) Stereo view of the Ca
trace.

The disulﬁde links in angiostatin. Angiostatin shown in red with
all disulﬁde bonds in yellow

Superposition of various Kgs from plasminogen. This ﬁgure
includes Kgl, Kg2 and Kg3 from angiostatin and the individual
Kgl and Kg2.

Electrostatic potential surface (EPS) of angiostatin with the bicines
omitted. a) Same orientation as in Figure 4.1. b) Rotated 180° to
show Kgl ’3 LBS

a) Carboxylate lysine residue. b) The carboxylate lysine analog,
e-aminocaproic acid. c) Bicine

Interaction of the three angiostatin LBS's with bicine. All three
depictions are in the same orientation. Hydrogen bonds and salt
bridge contacts are shown by dotted lines. Residues from
angiostatin are shown in green with atom colors, bicines are
shown in yellow. (a) Comparison of the binding of Kgl to bicine
and EACA. EACA is shown in lavender. (b) Angiostatin Kg2 and
Kg2 from the Kg2/VEK-3O structure are overlayed. Residues from
the VEK-3O peptide are not shown for clarity. (c) The angiostatin
Kg3 LBS with bound bicine.

a) EACA molecule was modeled onto the Kg3 LBS by overlaying

the structures of Kgl onto Kg3. b) Different conformation of the
bicine in Kgl depicted in red and the bicine in Kg3 shown in blue.

xiv

95

99

100

102

104

106

107

110

Figure 4.8

Figure 4.9

Figure 4.10

A Ribbons depiction of the modeled angiostatin/VEK30 complex. 112
The Kg2 of the Kg2/VEK30 complex was overlayed on

angiostatin Kg2. Angiostatin is colored green while VEK30 is

colored lavender. Side groups are labeled appropriately.

Endostatin modeled onto the angiostatin. This was done by 113
overlaying the helices of endostatin and VEK30. Endostatin is

colored purple and angiostatin green. b) Close view of the section

of angiostatin that harbors the residues that may be involved in

endostatin binding.

a) Structure of (1433 integrin. The av subunit is shown in red and 116
the [33 subunit in green. Also showing the residues involved in

angiostatin binding. b) Close view of the section of [33 that harbors

the residues involved in angiostatin binding.

XV

LIST OF ABBREVIATIONS

A - alanine

ADP - Adenosine diphosphate

ADPGlc Ppase - ADP glucose pyrophosphorylase
aFGF - acidic ﬁbroblast growth factor

APS - Advanced Photon Source

ATP - Adenosine triphosphate

BE - branching enzyme
bFGF - basic ﬁbroblast growth factor
bicine - N,N Bis(2hydroxyethyl) glycine

C - cysteine
CGT - cyclodextrin glucanotransferase
C terminal - carboxy terminal

D - aspartic acid
DEPC - diethylpyrocarbonate

E - glutamic acid

EACA - s-aminocaproic acid

E. coli - Escherichia coli

ED50 - half life maximum concentration to achieve 50% inhibition of bFGF-stimulated
bovine capillary endothelial cell growth.

EPS - electrostatic potential surface

F - phenylalanine

Fcalc - calculated structure factors

. F GF - ﬁbroblast growth factor

FH - structure factor contribution of the derivative
Fhk. - structure factor for a reﬂection labeled hkl
Fobs - observed structure factors

F pH - structure factor of the protein plus derivative

G - glycine

G(S)S - glycogen (starch) synthase
GSDIV - glycogen storage disease type IV

H - histidine
hepes - N-[2-hydroxyethyl] piperazine-N’-[ethane sulfonic acid]

xvi

I - isoleucine
I - intensity
interkringle - inter-Kg

K - lysine
Kg - kringle

L - leucine
LBS - lysine binding site

M - methionine

m - mass in mg

mBEI - maize endosperm BE isoform I
mBEII - maize endosperm BE isoform II
MPD - 2-Methyl-2,4-pentanediol

N - asparagine
N113BE - BE from E. coli lacking the ﬁrst 112 residues at the amino terminal
N terminal - amino terminal

P — proline

PAM - group A pathological Streptococcal surface protein
PEG - polyethylene glycol

PP - phasing power

Q - glutamine
R - arginine
rmsd - root mean square deviation

S - serine

SAD - single wavelength anomalous dispersion
SBC - Structural Biology Center

SeMet - selenium methionine

SIR - single isomorphous replacement

T - threonine
TSPl - thrombospodin 1
TSP2 - thrombospodin 2

UDP - Uridine diphosphate
UDPGlc Ppase - UDP glucose pyrophosphorylase

V - valine

v - volume in m1
VEGF - vascular endothelial growth factor

xvii

W — tryptophan
WT - wild type

Y - tyrosine

xviii

CHAPTER I: INTRODUCTION

1.1 Branching Enzyme
1.1.1 Glycogen

Glycogen, the energy storage polysaccharide in animals and bacteria cells.
accumulates under environmental conditions that limit growth and offer excess in carbon
supply (1-4). The major accumulation of glycogen occurs at the stationary phase of the
growth cycle under nitrogen reduction conditions. Bacterial mutants with defective
glycogen biosynthetic enzymes are viable under a glucose rich medium, showing that
glycogen is not required for growth (4).

Starch is the form in which plants store the energy accumulated by carbon
ﬁxation via photosynthesis. Starch and glycogen are composed of polyglucose chains
linked by a-l.4 and a-1-6 bonds (Figure 1.1). Amylose and amylopectin are the two
molecules that form starch and glycogen (Figure 1.1). Amylose is the mainly linear
polysaccharide formed by a-1,4 glucosidic bonds, although it contains some (it-1,6
branches (less than 0.6% of all glucosidic linkages). Amylose chains vary in length from
840 to 22,000 glucose units (5). Amylopectin is a highly branched polymer with a
molecular weight of up to one million. The a-l ,6 branches occur every 8 to 21 residues
in glycogen and every 24 to 30 residues in starch. Although there are many similarities
between the metabolism of starch and bacterial glycogen there is a difference in its ﬁnal
structure. Glycogen is more branched with twice (10% of all glucosidic bonds) the

amount of a-l .6 links compared to starch.

 

 

 

 

 

 

 

 

 

 

\
CH20H CHZOH rCHZQH CHZOH CHZOH
O 0 o o 0
OH OH OH OH
/ OH
O O O O‘ O
OH
OH \ OH L OH Jn OH OH
reducin
(X-l,4 Amylase end g
f \
012:“ CHZOH CHon CHZOH Amylopectin
O
\ \
OH \ NOH .x. OH \ OH
a \o’ H \o” \O :(wl
OH K OH Jn OH
CH20HH CH20H rCHZOH \ CHZOH
O
H

 

 

 

Figure 1.1 Starch and glycogen are formed by

D-glucose units linked by or-l ,4 and

(it-1.6 glucosidic bonds. Amylose and amylopectin are the two molecules that

form starch and glycogen.

IQ

 

Energy storage in the form of a glucose polymer offers a compact and efﬁcient
way of storing the easily mobilized glucose molecule. Moreover. the formation of the
(it-1,6 branches increases the energetic capacity of glycogen. Glycogen and starch are
synthesized and hydrolyzed by removing or adding the non-reducing glucose units (sugar
without a free anomeric carbon) (Figure 1.1). With more branches, there are more non-
reducing ends, this means that (it-amylase and starch synthase will have more substrates
to perform digestion or synthesis, increasing the molecule reactivity towards both

synthesis and digestion.

1.1.2 Biosynthesis ofBacterial Glycogen

Glycogen biosynthesis occurs in bacteria during non-growing periods when
carbon is in excess. During the non-growing phase energy is required for various
essential processes like chemotactic response, maintenance of motility and intracellular
pH, osmotic regulation, and translation among others processes. Glycogen plays an
important role in preserving cell integrity in bacteria under harsh conditions (4). Studies
performed on Escherichia coli (E. coli) and Enterobacter aerogenes show that mutants
unable to synthesize glycogen degrade their RNA and protein in media that is devoid of a
carbon source. When wild type bacteria are grown under the same conditions, they
preserve their cellular constituents (4).

The biosynthesis of bacterial glycogen follows a pathway similar to starch
synthesis in plants, although there is a higher similarity between the ﬁnal structures of
bacterial and mammalian glycogen. The starch and glycogen synthetic pathways consist

of three steps, starting with the formation of the sugar nucleotide glucosyl donor,

 
     
 

ADP-glucose
pyrophosphorylase

 

 

    

Glucose-1 -phosphate CHon
0\
ll OH \>_ ﬁ ﬁ
\\ ,/
OHi—j : T O T O Adenosine
OH 0 O
ADP-glucose
CH20H CHZOH
O O
\
OH >. OH
OH OH
b) Starch/Glycogen
synthase ADP
CH2OH
0 CHon CHZOH CHZOH
OH \ ’i—ox o o
\l / Branching «'/ OH >\ OH OH
OH [\\ / \
\ / O O
of) OH I H
OH OH OH
CH20H CHZOH

@5440

Figure 1.2 Biosynthetic pathway of starch and glycogen synthesis

followed by the elongation of the 0t-1,4 polyglucose chain, and ﬁnally the rearrangement
of the polysaccharide (Figure 1.2).

The activation of a glucose l-phosphate molecule into adenosine diphosphate
(ADP)-glucose is catalyzed by the enzyme ADP glucose pyrophosphorylase (ADPGlc
Ppase) (Figure 1a). In plants and bacteria cells, an ADP-glucose molecule is the glucose
donor in the next step of the reaction. In mammalian glycogen synthesis, the glucose
donor is UDP-glucose and its formation is catalyzed by UDPGlc Ppase. ADPGlc Ppase
from plants consists of two isoenzymes forming a heterotetramer. For example, ADPGlc
Ppase from potato tuber is formed by a small subunit involved in catalysis and a large
subunit responsible for allosteric activation and inhibition (6). Bacterial and mammalian
phosphorylases consist of a single unit forming a homotetramer and an homooctamer,
respectively.

The elongation of the glucan chain is catalyzed by glycogen or starch synthase
(G(S)S) as shown in Figure 1.2b. This enzyme forms an (it-1,4 glucosidic bond between
the anomeric carbon of the sugar nucleotide and a primer glucose molecule. In plant and
animal cells the number of isoenzymes may vary depending on the species. In bacteria a
single enzyme produces the elongation of the chain.

Tandecarz and Cardini were the ﬁrst to propose that glycogen synthesis must be
initiated by a self-glucosylated protein (7). This protein was later isolated and named
glycogenin. Glycogenin ﬁrst performs a self glucosylation that is followed by the
elongation of the glucose chain to up to 8 glucose units (8). Glycogenin has been isolated

from liver and muscle cells (8-10). There is no evidence of the existence of a self

glucosylated protein or if its even needed for the initiation of starch or bacterial glycogen
synthesis.

In the rearrangement step, branching enzyme (BE) is responsible for the
formation of the a-1.6 branch points. This is achieved by the cleavage of the a-1,4 bond
and the subsequent transfer of a glucose chain to the (1-1,6 position. This enzyme will be
discussed in detail in the next section.

ADPGlc Ppase is the only allosterically regulated enzyme in both the bacterial
glycogen and starch biosynthetic pathways. In e. coli it is activated by fructose 1,6
biphosphate, an indicator of carbon excess during glucogenesis, and inhibited by AMP,
ADP, or inorganic phosphate, all indicators of low energy in the cell. The regulation of
mammalian glycogen synthesis occurs at the GS step because GS catalyzes the ﬁrst
unique reaction in the pathway. UDP-glucose, the glucosyl donor on mammalian GS, is
used as a precursor for synthesis of cellular constituents. Mammalian GS differs from the
bacterial and plant enzymes not only in that it uses UDP-glucose as a substrate, but it also
exhibits regulatory activity. Mammalian GS exists in a phosphorylated or
dephosphorylated form that is either active or inactive, respectively. Because ADP-
glucose formation is used solely for starch and bacterial glycogen synthesis, the

regulation in the ﬁrst step is energetically more efﬁcient by conserving ATP.

1.1.3 Branching enzyme
Branching enzyme (1,4-0t-glucan : 1,4-0t-glucan 6-glucosyltransferase; EC
2.4.1.18) has an important role in the determination of the ﬁnal structure of starch and

glycogen. This enzyme catalyzes the formation of the 0t-1,6 branch points, transforming

a linear polysaccharide into a branched network. This is achieved by cleavage of the oc-
1.4-glucosidic linkage. yielding a non-reducing end polysaccharide chain, and subsequent
attachment to the or-l ,6 position.

The gene of the E. coli branching enzyme has been cloned and the nucleotide
sequence determined (11). The gene consisted of 2,1 84 base pairs, coding 728 amino
acids with a molecular weight of 84,231 Da. In bacteria there is only one BE while in
plants there can be multiple isoforms. Several of these plant isoenzymes have been
identiﬁed; four forms in rice seed. three in wheat endosperm and three in maize
endosperm seeds (12-1 5).

The unique feature of branching enzyme’s action lies in its speciﬁcity for the
length of the glucan chain transferred. Glycogen branching enzyme from E. coli has a
preference for transferring shorter chains between 5 and 16 glucose units from a non
reducing end of at least 11 units. On the other hand, starch branching enzyme from
maize transfers a wider range of chains from 6 to 30 glucose units (16). This speciﬁcity
is consistent with the denser structure of glycogen due to double the number of a-l ,6
links compared with starch.

The two isoenzymes from maize endosperm, mBEl and mBEII have been
extensively studied (17,18). Although mBEI and mBEII are 58% identical, they differ in
substrate preference, branching pattern and catalytic activity. The mBEI isoform has a
preference for transferring longer chains of 11 glucose units or longer with substrate
preference for amylose over amylopectin . On the other hand, mBEII transfers chains 6

glucose units or longer in length. Even though amylose is a favorable substrate, mBEII

has a higher afﬁnity for amylopectin. Also, mBEI has a ﬁve to six fold higher Vmax when
compared to mBEII.

Branching enzyme is divided into three domains based on secondary structure
prediction and its connection with the amylotic family of enzymes. These domains are;
an amino (N) terminal domain, a carboxyl (C) terminal domain and a central (or/B) barrel
catalytic domain. In an attempt to understand the role the three domains have in
branching enzyme activity, several chimeric mutants were made (Figure 1.3) (18). From
all eight mutants constructed, only three, F, G and H, showed some activity. Mutant F
exhibited a higher activity than the wild type mBEI and mBEII. This mutant also has a
higher speciﬁcity for amylose over amylopectin and a catalytic activity similar to mBEI.
In contrast, its branch transfer pattern was similar to mBEII. Analysis showed that
mutant G transferred more of the longer chains (11 glucose units or longer) similar to
mBEl. From these studies it was clear that the C terminal is involved in substrate
speciﬁcity and catalytic capacity, while the N terminal is involved in determining the size
of the chain transferred. This study also established the importance of the amino and
carboxyl termini, challenging the belief that the catalytic center of BB is limited to the
(Ct/B) barrel domain.

As mentioned before, BE belongs to the a-amylase family of enzymes (19,20).
Members of this group include a-amylases, pullulanase/isoamylase, cyclodextrin
glucanotransferase ( C GT) and branching enzymes. They have the common function of
cleaving and/or transferring glucose chains (21). The or-amylases and
pullulanase/isoamylase catalyze the hydrolysis of a-l ,4 and (X-I,6 glucosidic bonds,

respectively. Branching enzyme and C GT are the two members that catalyze

 

Optimum length

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

N-terminal (or/[3) Barrel C -terminal 0f glucose chain
transfer
aa # l 240 372 728
E. coli L l rtrtwirtrfririrtetvirtd “’6 glucose units
1 239 474 758 .
mBEI r N-mBEI 1 01/131 1 C-mBEI J ~12 glucose units
1 276 519 747 .
mBEII1 N-mBEII 1 1 01/1311 1 C-mBEII 1 76 glucose units
X 1 NmBEI 1 a/BI 1 C-mBEII 1
X 1 N-mBEI 1 a/BII 1 a/BI 1 C-mBEII j
X I N-mBEII LOL/BI 1 01/1311 1 C-mBEI 1
X l N-mBEI 1 OL/BII - 1 C-mBEI j
X L N-mBEII 1 a/BI 1 C-mBEII ]
F 1 N-mBEII 1 ' ~ awn-g1 C-mBEI ] ~6 glucose units
G 1 N-mBEI 1 a/B 11 , 1 C—mBEII 1 ~12 glucose units
Unable to
H [ N-mBEII j a/BI l C-mBEI ] determine

 

Figure 1.3 Chimeric enzymes constructed from mBEI and mBEII (18). The “X”

represents the inactive mutants and aa# stands for amino acid number. N-mBEI, C-mBEI
and (1/[31 represent the amino terminal, carboxylate terminal and 0t/13 barrel of mBEI,

respectively. The same nomenclature applies for mBEII.

transglycosylation reactions, with BE being the only one with speciﬁcity for two different
glucosidic bonds. CGT catalyzes the formation of cyclodextrins by cleavage and
subsequent transglycosylation of ct-l .4 links. X-ray structures of a-amylases, isoamylase
and C GTs show that these enzymes have a common (or/B) barrel domain that contain the
enzymes catalytic center (22-26). Based on biochemical data and X-ray structures of
apo and substrate bound (it-amylase and CGT, the catalytic center has been deﬁned to be
composed of seven residues; D335, H340, R403, D405, E458, H525 and D526 (E. coli
branching enzyme numbering) (Figure 1.4)(27,28). These residues are conserved among
members of this family and BE from various organisms (21,29). In an attempt to
understand the catalytic relevance of these conserved residues, studies have been
performed on maize endosperm BE by using site directed mutagenesis. When D405,
E458 and D526 (386, 441 and 509 mBEII numbering) were mutated to their respective
amide or alternate acid form BE activity disappeared (30).

Chemical modiﬁcation experiments using the arginine speciﬁc reagent,
phenylglyoxal. were performed to study the possibility of essential arginine residues
present (31). Phenylglyoxal inactivated both mBEI and mBEII. In both cases the
presence of amylose and amylopectin protected the enzyme from inactivation, although
amylopectin was a better substrate. The sequence alignment of several branching
enzymes were analyzed and two arginine residues, R312 (291 mBEII numbering) and
R403 (384), conserved among BEs and located in the (or/13) barrel catalytic domain, were
proposed to be the ones involved in catalysis. Site directed mutagenesis experiments
performed on mBEII demonstrated that R312 is not necessary for BE activity (32).

However. R403 mutations to either alanine, serine, glutamine or glutamic acid produced

10

 

 

 

 

 

 

 

 

 

E.coli 297 SWGTQPT 330LNVILDWVPGH
human 248 SFGYQIT 281IIVLLDVVHSH
mBEl BOBSFGYHVT 336LRVLMDVVHSH
mBE11277SFGYHVT 31OLLVLMDVVHSH
isoa 253 YWGYMTE 287|KVYMDVVYNH
a-Asp 79 YHGYWND112MYLMVDVVANH
a-Por 59 WERYQPV 91VR|YVDAV|NH
cor 97 YHGKWAR 130|KV11£FAPN_l-l
E.coli 398 GIDALWVDAVA 454 VTMAEEST
human 350 RFDGFRFDGVT 408|TIAEDVS
mBEI 408 MFDGFRFDGVT 466 TVVAEDVS
mBEII 379KFDGFRFDGVT 437 VTIGEDVS
isoa 368 GVDGFRFDLAS 431DLFAEPWA
a-Asp 199 SIDGLRIDTVK 226 YCIGEVLD
a-Por 190 GVAGFRLDASK 229 FIFQEVID
cor 222 GIDGI&MDAVK 253 FTFGEWFL
E.co|1520VLPLSHDEV
human 475 AYAESHDQA
mBEI 475 AYAESHDQA
mBEll 503 TYAESHDQA
isoa 504 NFIDVHDGM
a-Asp 291TFVENHDNP
a-Por 294 VFVDNHDNQ
CGT 322 TFIDNHDME

 

 

 

Figure 1.4 Conserved catalytic residues in the a-amylase family of enzymes. The
residues enclosed by the boxes are involved in substrate binding and the shaded residues
are involved in catalysis. The labels in the sequence alignment are E. coli for E. coli BE;
human for Homo Sapiens BE; mBEI and mBEII maize endosperm for BE I and II,
respectively; isoa is isoamylase from Pseudomonas amyloderamosa, a-Asp and a-Por
are a-amylase from Aspergillus Oryzae and Porcine Pancreatic and CGT is cyclodextrin

glucanotransferase from Bascillus Circulans.

ll

an inactive enzyme. This demonstrates that R403 plays a direct role in BE catalysis (32).
However, when R403 was replaced with lysine it retained approximately 5% of the wild
type (WT) enzyme activity. The two enzymes’ kinetic parameters, substrate speciﬁcity
and size of chain transferred were compared. The R403K mutant and WT enzyme have
similar substrate speciﬁcity preferring amylopectin as the substrate. The distribution of
size of the chains transferred and the Km value was similar to that of WT.

Analysis of the two histidine residues, H340 (320) and H525 (508), conserved in
the amylase family, was performed on mBE by site directed mutagenesis and chemical
modiﬁcation using diethylpyrocarbonate (DEPC) (33). DEPC reacts with histidine
residues as well as with other amino acids like tyrosine, lysine and cysteine (34,35).
Hydroxylamine removes the DEPC from modiﬁed hystidyl and tyrosyl residues, but not
from lysyl or cysteine residues. In these studies BE was inactivated after adding DEPC
and reactivated when hydroxylamine was added to the reaction mixture, ruling out
lysines or cysteines as the residues responsible for the loss of activity. The presence of
DEPC modiﬁed histidine increases the UV absorbance at 240 nm compared to the
umodiﬁed enzyme, while the absorbance at 270 nm is decreased by the presence of
DEPC modiﬁed tyrosine. These experiments showed an increase in the UV difference
spectrum at 240 nm with no change observed at 270 nm. This result showed that the
inactivation of mBEI and mBEII was caused by chemical modiﬁcation of the histidine
residues. Amylose and amylopectin protected mBEI and mBEII against DEPC
inactivation, with amylopectin being the better substrate. Subsequently, mutational
experiments were performed on H340 and H525. Speciﬁc activity of the H340A and

H525A mutants was reduced to 0.45% and 0.15% of that of wild type activity,

12

respectively. These mutants have higher Km values compare to the WT enzyme
indicating that these two residues are involved in BE substrate binding.

Mutational studies performed on E. coli BE included Y300 which is conserved in
the a-amylase family (36.37). Replacement of Y300 with A, D, L, S or W resulted in
mutant enzymes with less than 1% of wild type activity (36). The Y300F mutant retained
25% of wild type activity with comparable Km values and substrate preference. Based on
the crystal structures of the members of the (it-amylase family it has been proposed that
the conserved tyrosine forms a hydrogen bond with the conserved E458 orienting this
residue in the optimal position for catalysis (26). The inability of the Y300F mutant to
form the important hydrogen bond reﬂects differences in heat stability and enzymatic
activities at elevated temperatures. Indeed the heat stability of Y300F was found to be
lower by approximately 5°C. than that of WT, indicating the importance of the hydroxyl
group for protein stability (36). In summary, the conserved catalytic residues Y300,
H340, R403. D405. E458. H525 and D526 are indeed necessary for branching enzyme’s
activity.

Although members of the a-amylase family of enzymes share structural features
and conserved amino acids involved in catalysis, the fact that they catalyze different
reactions generates the question of which residues are responsible for the distinct
catalytic properties. Analysis of the amino acid sequence alignment reveals a conserved
position unique to branching enzymes. The residue E459 is either an aspartate (mostly in
eukaryotes) or a glutamic acid depending on the organism. This residue is located after
the E45 8 conserved in the (it-amylase family and necessary for BE activity. Mutation of

E459 to A. N or K dramatically lowered the speciﬁc activity to 30%, 12% and 6%, of

13

wild type respectively (37). These mutants altered the preference of the substrate from
amylose to amylopectin as well as its kinetic parameters. The E459D conservative
mutation increases the speciﬁc activity of BE, with kinetic properties similar to those of
the WT enzyme, although this mutation has an effect in the glucose chain transfer pattern.
Comparison of the pH proﬁles for the WT and E459D mutants rules out the possibility of
E459 being involved in acid-base catalysis. All the mutational data previously presented
is summarized in Table 1.1.

Glycogen branching not only increases the number of non-reducing ends, thus
making glycogen more reactive to synthesis and digestion, but it is also essential for
assuring glycogen solubility in the cell. Glycogen in its linear form precipitates in the
cell. Accumulation of insoluble glycogen in the cell is known as glycogen storage
disease type IV (GSD IV) and is caused by mutations in the gene of the ubiquitously
expressed glycogen branching enzyme (38,39). These mutations result in an impaired
glycogen metabolism that forbids the formation of the branch points in glycogen,
producing an insoluble polymer. This genetic disease occurs in different allelic variants
with various clinical presentations. GSD IV in its different forms affects the liver,
muscular tissue and the central and peripheral nervous system (40). The classical form of
the disease presents progressive liver cirrhosis leading to death in children at the age of
ﬁve years old. It is caused by either one of the two mutations R515C, F257L (refers to
546 and 306 in E. coli numbering) or the C terminal truncation at residue 524 (R524Ter,
555 in E. coli numbering). The ﬁrst two mutations leaves the enzyme with activity

between 20% to 27%, while the R524ter truncation produces an inactive enzyme. There

14

Table 1.1 Mutational studies performed in branching enzyme from maize endosperm
isoform II and E. coli (30-33,37,4l).

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Mutation Approximate Speciﬁc organism E. coli numbering
Activity (%)
D386E 2 maize (mBEII) 405
D386N 2 maize (mBEII) 405
E441 D 2 maize (mBEII) 458
E44] Q 2 maize (mBEII) 458
D509E 2 maize (mBEII) 526
D509N 2 maize (mBEII) 526
R384A 2 maize (mBEII) 403
R3 848 1 maize (mBEII) 403
R3 84Q 3 maize (mBEII) 403
R3 84E 1 maize (mBEII) 403
R3 84K 12 maize (mBEII) 403
H320A 4 maize (mBEII) 340
H508A 0 maize (mBEII) 525
Y300A 0 E. coli
Y300D 0 E. coli
Y300L 0 E. coli
Y3OOS 0 E. coli
Y300W 0 E. coli
Y300F 25 E. coli
E459A 30 E. coli
E459K 6 E. coli
E459N 12 E. coli
E459D 185 E. coli
N113GBE 60 E. coli

 

 

 

 

15

 

is a less common slow progressive form that presents liver dysfunction but no liver
failure and is associated with two mutations, L224P and Y329S, leaving a totally inactive
or a 50% active enzyme, respectively. The neuromuscular form is caused by a 70 residue
deletion between amino acids 262 to 331 (311 to 379). It is expressed at birth, with death
in the neonatal period. The Y329S mutation is also responsible for a late onset slowly
progressive form of the disease that affects the central and peripheral nervous system.
There is also a combined hepatic and muscular disorder caused by a single mutation of
R524N (555). Some of these mutations present a mild progressive form while others are
lethal in early childhood. With few exceptions, GSDIV is a progressive and lethal disease
(39,42). Although the human BE gene localized in chromosome 3 has been identiﬁed,
these mutations have not been tested in bacteria or plant BE (43). As it will be shown
later, these mutations are likely to cause the unfolding of the protein, inactivating the

enzyme.

1.1.4 Truncated branching enzyme

The determination of the three dimensional structure of branching enzyme has
been a goal in Dr. Preiss’s lab for many years. Previous attempts to crystallize the WT
enzyme were performed by members of his group and were not successful. The
technique of limited proteolysis was used to identify a more compact and stable form of
the enzyme that would be successful in crystallization (44). Branching enzyme from E.
coli was subjected to several proteases with a range of speciﬁcities. These proteases
were proteinase K, protease V8, elastase, chymotrypsin, subtilisin, trypsin and

carboxypeptidase Y. All proteases displayed similar digestion patterns producing a 71.6

16

kDa fragment, with the exception of trypsin. BE lacking the ﬁrst 112 residues at the
amino terminus (N113BE) was analyzed and its properties compared to the WT enzyme.
Despite the fact that N1 13BE was only 60% active, its substrate preference and Km value
were similar to the WT. N1 13BE presents an altered branching pattern with a higher
afﬁnity for longer chains of 12 glucose units or more.

Although glycogen synthesis has been a ﬁeld of study since the 1940’s and
progress has been achieved in the determination of its mechanism, its chemistry is not
fully understood. This is mainly due to the lack of structural models. There are no
structures of any of the enzymes involved either in glycogen or starch biosynthesis. The
three dimensional structure of branching enzyme is a de novo structure that will reveal

valuable information that will aid in the understanding of this biosynthetic pathway.

17

1.2 Angiostatin
1.2.1 Angiogenesis

In mammals, blood vessels are formed by two processes, vasculogenesis and
angiogenesis. Vasculogenesis is the new formation of blood vessels through
differentiation of precursor cells into endothelial cells, forming a vascular network (45).
This process occurs during embryonic development. Angiogenesis is the sprouting of
blood vessels from existing capillary beds (46). In adults, new blood vessels are formed
exclusively via angiogenesis, with the exception of the transient process of
neovascularization that occurs in the female reproductive system. Angiogenesis is
essential for development and wound healing. This process is triggered by a growth
stimulus that enters the endothelial dormant cells into the cell cycle. The process begins
with the degradation of the basement membrane in endothelial cells and lumen formation.
Simultaneously, the endothelial cells change their morphology, proliferate, migrate, form
microtubes and sprout new capillaries. This is a highly regulated process involving
multiple controls that can be turned on or off throughout the process.

Pathological angiogenesis occurs in rheumatoid arthritis, diabetic retinopathy,
tumor development, and in metastasis dissemination (47). In rheumatoid arthritis,
capillaries invade the joints and destroy the cartilage. In diabetic retinopathy, new
capillaries invade the retina causing blindness. Ocular vascularization is the most
common cause of blindness, responsible for approximately 20 eye diseases. Tumor
growth and metastasis dissemination are angiogenesis dependent processes as well.
Avascular tumors can not grow beyond 2 to 3 mm3, but once the tumor is vascularized,

rapid growth is observed (46,47). This network of blood vessels embedding the tumor

18

not only promotes tumor growth, it also serves as a portal for tumor cells to enter the
blood stream and to metastasize to other organs.

Angiogenesis is controlled by speciﬁc molecules; angiogenesis activators trigger
the process. while angiogenesis inhibitors stop it. These molecules act in concert to
maintain a balance in the vasculature with a turnover that can last for years, although
during wound healing the turnover lasts days. Once enough tumor cells have switched to
the angiogenic phenotype, the tumor itself stimulates the up regulation of angiogenesis
activators and down regulates angiogenesis inhibitors (48-51). Some of the currently
known angiogenesis activators and inhibitors and their properties are listed in Table 1.2.
Among the most common angiogenesis activators found in tumor cells are the vascular
endothelial growth factor (VEGF) and the family of ﬁbroblast growth factors (F GF),
which includes the basic and acid ﬁbroblast growth factors (bFGF and aFGF). VEGF is a
speciﬁc activator of endothelial cells, while F GF can act on a variety of cell types. High
levels of VEGF have been detected in the majority of human tumors including bladder,
breast, lung, gastrointestinal, ovary, prostate, glioblastoma, hemangioma and
retinoblastoma (48). High concentrations of bFGF have also been found in cancer
patients (49). Moreover, the angiogenesis inhibitor thrombospodin 1 (TSPI) has been
found to be down regulated in several tumors (50,51).

Once it was determined that tumor growth and metastasis dissemination were
angiogenesis dependent processes, it was proposed that blocking angiogenesis was the
best strategy for cancer treatment. This presented a major breakthrough in cancer
therapy, which for decades was targeted to destroy tumor cells using cytotoxic agents.

The high mutational rate of tumor cells, in addition to the high toxicity that these agents

19

Table 1.2 Various molecules involved in angiogenesis activation or inhibition

 

 

 

(46,52-54)

Activators Function

VEGF stimulates permeability and adhesion
bFGF and aFGF stimulates angiogenesis

 

angiopoietin l + Tie2 endothelium receptor

stabilize vessels and inhibits permeability

 

angiopoietin 2 +VEGF

stimulates angiogenesis

 

tumor necrosis factor

induces production of bFGF

 

OLV135, 0th3 integrins

mediate cell migration and lumen
formation

 

platelet derived GF

recruits smooth cells

 

VEGF receptors

integrate angiogenic and survival signals

 

transforming growth factor

stimulates extracellular matrix production

 

 

platelet endothelial cell adhesion protein

 

endothelial junctional protein

 

 

Inhibitors

Function

 

antithrombin III (fragment), angiostatin,
endostatin and interferon-13

inhibits proliferation and/or migration of
endothelial cells

 

 

 

TSPl and 2 endothelial cell proliferation and migration
prolactin inhibits VEGF and bFGF
VEG inhibitor modulates cell growth

 

granulocyte macrophage colony
stimulating factor

mobilization of endothelial precursor

 

angiopoietin 2

cause apoptosis of the vessel

 

 

P53

 

induces transcription of TSPl

 

20

 

 

present against normal cells, makes chemotherapy less effective and risky.
Antiangiogenic therapy targets the proliferation of normal endothelial cells, which
represent a uniform target, unlike the fast mutating cancerous cells. In a normal adult
only 0.01% of all endothelial cells undergo division at a speciﬁc time. Antiangiogenic
therapy will inhibit tumor vascularization selectively without affecting normal
vasculature. Also, endothelial cells can be easily reached through the blood stream. All
these reasons have directed attention to angiogenesis inhibitors as potential anti-cancer
agents. Among the angiogenesis inhibitors, angiostatin has been shown to inhibit tumor
growth and metastasis dissemination in animal models (55,56). Angiostatin causes no
side effects, toxicity or weight loss and is currently in phase I clinical trials at the Thomas

Jefferson University Hospital (Philadelphia, PA).

1.2.2 Angiostatin

The angiogenesis inhibitor angiostatin is an N-terminal fragment of plasminogen
with a molecular weight of 38 kDa (57). Although angiostatin is a proteolytic fragment
of plasminogen, plasminogen is inactive in the inhibition of endothelial cell growth,
neovascularization and metastatic tumor growth (55,58). Plasminogen is a zymogen of
plasmin activated after a single peptide bond between residues R561 and V562 (human
plasminogen numbering) is cleaved. Plasmin is a 92 kDa serine protease that catalyzes
the dissolution of blood clots. Plasminogen consists of a catalytic domain and ﬁve highly
homologous triple disulﬁde binding domains, known as kringles (Kg). These kringle

domains speciﬁcally bind the

21

C-terminus

 

 

a) T T O

—-—N——CH—C—-N—CH—C——O'

|

H o 2

amino acid chain CH2

CH2

CH2

NH,+

 

 

 

carboxylate lysine residue

b) ll

 

Figure 1.5 a) Carboxylate lysine residue. b) The carboxylate lysine analog,

s-aminocaproic acid (EACA).

22

carboxylate lysyl residues in ﬁbrin, the skeleton of blood clots (Figure 1.5a). Upon ﬁbrin
binding. plasmin exerts its proteolytic activity, solubilizing ﬁbrin and consequently

dissolving the blood clot.

X-ray crystal structures of four of the ﬁve individual plasminogen kringle
domains, Kgl , Kg2, Kg4 and Kg5, have been previously determined (59-62). Their
binding modes for lysine-like ligands have also been studied structurally and by site-
directed mutagenesis (61-64). The binding center of kringles is known as the lysine
binding site (LBS). It consists of a cationic and an anionic center that stabilizes the
carboxyl and amino group of a carboxylate lysyl residue (Figure 1.5 and 1.6). The LBS
is deﬁned by residues R115, R153, D137, and D139 in Kgl; R234, D219, and E221 in
Kg2 and R290, R324, D309 and H317 in Kg3. Binding experiments using the C-terminal
lysine analog e-aminocaproic acid (EACA) showed that Kgl has high afﬁnity for EACA
with a KB of 15.5 11M (Figure 1.5b and 1.6) (62,65,66). Kg2 has a lower afﬁnity for
EACA (401 11M) and Kg3 shows no afﬁnity at all. Kg2 also binds the VEK30 peptide, a
30 residue peptide encompassing residues 85 to 114 within the sequence of the group A
pathological Streptococcal surface protein (PAM) (67). Kg2 binds VEK30 strongly (KD
= 0.46 11M), presenting a model for protein binding at the surface of bacteria. This
peptide consists of a 5 turn a-helix that runs between the anionic and cationic centers of
Kg2’s LBS (Figure 1.7) (59). VEK30 forms a pseudo lysyl residue with amino acids

R101 and E104 that ﬁts in the Kg2’s LBS.

Effects of individual or combined kringles on endothelial cell proliferation in

vitro have been studied as well (57). This study revealed that Kgl -3 has a two fold

23

 

Figure 1.6 Structure of Kl-EACA (61). K1 is colored green and the residues
encompassing the LBS are shown in atom color (nitrogen, blue; oxygen, red and carbon,
green) and EACA is colored lavender. Side chains are labeled using plasminogen

numbering.

24

 

Figure 1.7 X-ray crystallographic structure of K2-VEK30 (59). K2 is colored green and
the VEK30 peptide is colored cyan. Side chains are labeled using plasminogen

numbering.

25

inhibitory activity with a half life maximum concentration (ED50) of 70 nM versus 135
nM for Kgl -4. The half life maximum concentration refers to the concentration to
achieve 50% inhibition of bFGF-stimulated bovine capillary endothelial cell growth.
Kgl has the highest inhibitory activity, with an ED50 of 320 nM. Recombinant Kg2-3
exerts inhibitory activity similar to Kg2 alone, although enhancement of inhibition is
observed when individual Kg2 and Kg3 are added together. Other studies performed on
animal models showed that agents containing Kgl -3, Kgl-4, Kgl -5, and Kgl-4 plus a
fragment of Kg5 have potent antiangiogenic and/or anti-tumor growth activity (57,68-
70). These fragments, as well as individual kringle domains, are also inhibitory toward
endothelial cell migration and/or proliferation in vitro. Studies with recombinant
angiostatin show that the maximum tumor inhibitory activity resides in a fragment of
angiostatin containing Kgl-3 (29.77 kDa; residues 79 to 338) (71). This smaller

fragment is the one currently used in clinical trials.

It has been observed that once a primary tumor is removed, dormant distant
metastasis may grow (72-76). A number of animal experiments showed that primary
tumors sometimes inhibits the growth of their metastasis (77-79). Although some
hypotheses have been proposed to explain this phenomenon, none of them has provided a
molecular explanation for this mechanism. The best explanation for this phenomenon
occurred after angiostatin was isolated from serum and urine of mice bearing Lewis lung
carcinoma (58). These experiments demonstrated that in fact, tumors promote the release
of angiostatin into the bloodstream to inhibit the development of metastasis and therefore

prevent competition.

26

Three molecules that mediate cleavage of plasminogen into angiostatin in vivo
have been identiﬁed. In Lewis lung carcinoma, the proteolytic activity of elastase
catalyses this transformation (57,58). It was also found that tumor cells up regulate the
production of elastase. In human prostate carcinoma, plasmin works as an enzyme and
substrate to generate angiostatin in the presence of free sulfhydral donors (70). The
antiangiogenic properties of pharmacological sulﬂtydril donors such as D-penicillamine
and captopril have been previously reported (80-83). Another proposed mechanism for
the generation of angiostatin involves the enzyme phosphoglycerate kinase (84). This
enzyme reduces the disulﬁde links in Kg5, a process that triggers the cleavage of
plasminogen into angiostatin. High levels of phosphoglycerate kinase have been
observed in plasma of mice bearing ﬁbrosarcoma tumors. Moreover, administration of
recombinant phosphoglycerate kinase increases the plasma levels of angiostatin by 86%
and inhibits tumor growth between 50% to 70%, depending on the type of tumor. Also,

reduced ﬁbrinolysis and hypercoagulation is often observed in cancer patients (85).

Human angiostatin potently inhibits the growth of transplanted tumors in mice
(55.56). The growth of Lewis lung carcinoma, T241 ﬁbrosarcoma, and reticulum cell
sarcoma tumors were inhibited by an average of 84% at doses of 100 mg/kg/day. These
tumors had poor response to other therapies. Angiostatin also inhibits the growth of
human breast carcinoma by 95%, colon carcinoma by 97% and prostate carcinoma by
almost 100% (55). Angiostatin also reduced the size of brain tumors in mice by more
than 71% (69). Brain tumors are known to be problematic to treat due to the difﬁculty of
crossing the brain-blood barrier, which hinders drug delivery. Angiostatin treatment does

not result in weight loss or other toxicity in mice.

27

The mechanisms by which angiostatin inhibits angiogenesis are still unclear. It
has been observed that in angiostatin treated tumors, the rate of apoptosis (process by
which a cell actively commits suicide) in tumor cells increases to ﬁve times that of the
control (58.86) (55). Although the mechanism by which angiostatin therapy leads to an
increase in death rate of tumor cells is not known, it is believed that angiostatin may
bring the tumor to a dormant stage between cell proliferation and apoptosis. It has also
been observed that angiostatin interacts with several proteins involved in endothelial cell
lysis, migration or proliferation. Angiostatin binds the a/ [3 subunit of ATP synthase,
angiomotin and the (1,133 integrin receptor (87,88). The a/B subunit of ATP synthase is
found on the surface of several tumor cell lines. ATP synthase catalyzes the transport of
H” across the membrane, resulting in tumor cell lysis. Furthermore, previous studies
demonstrated that the addition of ATP synthase to cultures of tumor cell lines induces
lysis of the cell (89-92). In this case angiostatin binding may mediate the mechanism for
inhibition of endothelial cell growth and tumor apoptosis. It has been reported recently
that angiostatin also binds angiomotin, a protein involved in endothelial cell migration
(88). Moreover, angiostatin inhibited migration in angiomotin expressing cells but not in
the absence of angiomotin. This indicates that angiostatin inhibits cell migration by
interfering with angiomotin activity in endothelial cells. An important speciﬁc binding is
its interaction with the (M33 integrin receptor (93). (1,133 is an endothelial cell surface
receptor implicated in the activation of angiogenesis. The angiostatin interaction can be
blocked by EACA at a concentration high enough to occupy the LBS in Kg2 (much
higher than that needed for Kgl). This indicates that Kg2 is more important than Kgl for

integrin binding. Although the mechanism of action for angiostatin is still unknown. It is

28

known that angiostatin is one of the most potent inhibitors for angiogenesis and it has
proved to be successful in decreasing tumor size and metastasis in animal models. To
better understand the structure and function of angiostatin, we have determined its three
dimensional structure to a resolution of 1.75 A. We hope that this structure will facilitate

the development of more effective anti-cancer therapies.

29

1 .3 Literature Cited

10.

ll.

l2.

13.

14.

15.

16.

Preiss, J., and Romeo, T. (1989) Adv Microb Physiol 30, 183-238.
Dawes, E. A., and Senior, P. J. (1973) Adv Microb Physiol 10, 135-266.
Preiss, J. (1984) Annu Rev Microbiol 38, 419-458.

Preiss, J. (1989) in Bacteria in Nature (Poindexter, J. S., and Leadbetter, E., eds)
Vol. 3, pp. 189-258, Plenum Publishing Corp., New York

Preiss, J. (1996) in Cellular and Molecular Biology Vol. 1, pp. 1015-1024, Amer.
Soc. Microbiol., Washington, DC

Ballicora, M. A., Laughlin, M. J., Fu, Y., Okita, T. W., Barry, G. F., and Preiss, J.
(1995) Plant Physiol 109(1), 245-251.

Tandecarz, J. S., and Cardini, C. E. (1978) Biochim Biophys Acta 543(4), 423-
429.

Krisman, C. R., and Barengo, R. (1975) Eur J Biochem 52(1), 117-123.

Lomako, J., Lomako, W. M., and Whelan, W. J. (1988) Faseb J2(15), 3097-
3103.

Pitcher, J., Smythe, C., Campbell, D. G., and Cohen, P. (1987) Eur JBiochem
169(3), 497-502.

Baecker, P. A., Greenberg, E., and Preiss, J. (1986) JBiol Chem 261(19), 8738-
8743.

Mizuno, K., Kimura, K., Arai, Y., Kawasaki, T., Shimada, H., and Baba, T.
(1992) J Biochem (Tokyo) 112(5), 643-651.

Morell, M. K., Blennow, A., Kosar-Hashemi, B., and Samuel, M. S. (1997) Plant
Physiol 113(1), 201-208.

Fisher, D. K., Gao, M., Kim, K. N., Boyer, C. D., and Guiltinan, M. J. (1996)
Plant Mol Biol 30(1), 97-108.

Gao, M., Fisher, D. K., Kim, K. N., Shannon, J. C., and Guiltinan, M. J. (1997)
Plant Physiol 114(1), 69-78.

Guan, H., Li. P., Imparl-Radosevich. J., Preiss, J., and Keeling, P. (1997) Arch.
Biochem. Biophys. 342, 92-100.

30

17.

18.

19.

20.

21.

22.

23.

24.

25.

26.

27.

28.

29.

30.

31.

32.

33.

34.

Hong, S., and Preiss, J. (2000) Arch Biochem Biophys 378(2), 349-355.

Kuriki, T., Stewart, D. C., and Preiss, J. (1997) J Biol Chem 272(46), 28999-
29004.

Baba, T., Kimura, K., Mizuno, K., Etoh, H., Ishida, Y., Shida, O., and Arai, Y.
(1991) Biochem Biophys Res C ommun 181(1), 87-94.

Romeo, T.. Kumar, A., and Preiss, J. (1988) Gene 70(2), 363-376.
Svensson, B. (1994) Plant Mol Biol 25(2), 141-157.

Matsuura, Y., Kusunoki, M., Harada, W., and Kakudo, M. (1984) J Biochem
(Tokyo) 95(3), 697-702.

Buisson, G., Duee, E., Haser, R., and Payan, F. (1987) Embo J 6(13), 3909-3916.
Boel, E., Brady, L., Brzozowski, A. M., Derewenda, Z., Dodson, G. G., Jensen,

V. J., Petersen, S. B., Swift, H., Thim, L., and Woldike, H. F. (1990) Biochemistry
29(26), 6244-6249.

Klein, C., and Schulz, G. E. (1991) JMol Biol 217(4), 737-750.

Katsuya, Y., Mezaki, Y., Kubota, M., and Matsuura, Y. (1998) J Mol Biol 281(5),
885-897.

Uitdehaag, J. C., van Alebeek, G. J., van Der Veen, B. A., Dijkhuizen, L., and
Dijkstra, B. W. (2000) Biochemistry 39(26), 7772-7780.

Brzozowski, A. M., and Davies, G. J. (1997) Biochemistry 36(36), 10837-10845.

Jespersen, H. M., MacGregor, E. A., Sierks, M. R., and Svensson, B. (1991)
Biochem J 280, 51-55.

Kuriki, T., Guan, H., Sivak, M., and Preiss, J. (1996) J Protein Chem 15(3), 305-
313.

Cao, H., and Preiss, J. (1996) J Protein Chem 15(3), 291-304.

Libessart, N., and Preiss, J. (1998) Arch Biochem Biophys 360(1), 135-141.

Funane, K., Libessart, N., Stewart, D., Michishita, T., and Preiss, J. (1998) J
Protein Chem 17(7), 579-590.

Miles, E. W. (1977) Meth. Enzymol. 47, 431-442

31

35.
36.

37.

38.

40.

41.

42.

43.

44.

45.
46.
47.
48.

49.

50.

51.

Lundblad, R. L. (1991) Chemical Reagentsfor Protein Modification, 2nd Ed.
Mikkelsen, R., Binderup, K., and Preiss, J. (2001) Arch Biochem Biophys 385(2),
372-3 77.

Binderup, K., and Preiss, J. (1998) Biochemistry 37(25), 9033-9037.

DiMauro, S., and Tsujino, S. (1994) in Myology (A.G., E., and C., F.-A., eds) Vol.
2, pp. 1554-1576.

Chen, Y. T., and Burchell, A. (1995) in The metabolic and molecular basis of
inherited diseases (QR, S., A.L., B., W.S., S., and D., V., eds) Vol. 1, pp. 935-
965.

Bao, Y., Kishnani, P., Wu, J. Y., and Chen, Y. T. (1996) J Clin Invest 97(4), 941-
948.

Binderup. K., Mikkelsen, R., and Preiss, J. (2001) Arch. Biochem. Biophys. , in
press

Schroder, J. M., May, R., Shin, Y. S., Sigmund. M., and Nase-Huppmeier, S.
(1993) Acta Neuropathol 85(4), 419-430.

Thon, V. J., Khalil, M., and Cannon, J. F. (1993) JBiol Chem 268(10), 7509-
7513.

Binderup, K., Mikkelsen, R., and Preiss, J. (2000) Arch Biochem Biophys 377(2),
366-371.

Risau, W. (1995) Faseb J 9(10), 926-933.

Folkman, J ., and Shing, Y. (1992) J Biol Chem 267(16), 10931-10934.
Folkman, J. (1995) Nat Med 1(1), 27-31.

F errara, N., and Davis-Smyth, T. (1997) Endocr Rev 18(1), 4-25.

Nguyen, M., Watanabe. H., Budson, A. E., Richie, J. P., Hayes, D. F., and
Folkman, J. (1994) J Natl Cancer Inst 86(5), 356-361.

Grossfeld, G. D., Ginsberg, D. A., Stein, J. P., Bochner, B. H., Esrig, D., Groshen,
S., Dunn, M., Nichols, P. W., Taylor, C. R., Skinner, D. G., and Cote, R. J. (1997)
J Natl Cancer Inst 89(3), 219-227.

Dameron, K. M., Volpert, O. V., Tainsky, M. A., and Bouck, N. (1994) Science
265(5178), 1582-1584.

32

52.

53.

54.

55.

56.

57.

58.

59.

60.

61.

62.

63.

64.

65.

66.

67.

Yancopoulos, G. D., Davis, 8., Gale, N. W., Rudge, J. S., Wiegand, S. J., and
Holash, J. (2000) Nature 407(6801), 242-248.

Carmeliet, P., and Jain, R. K. (2000) Nature 407(6801), 249-257.
Zhang, L., Chen, Q. R., and Mixson, A. J. (2000) Curr. Genomics 1(2), 117-133.

O'Reilly, M. S., Holmgren, L., Chen, C., and Folkman, J. (1996) Nat Med 2(6),
689-692.

Sim, B. K., O'Reilly, M. S., Liang, H., Fortier, A. H., He, W., Madsen, J. W.,
Lapcevich, R., and Nacy, C. A. (1997) Cancer Res 57(7), 1329-1334.

Cao, Y., Ji, R. W., Davidson, D., Schaller, J., Marti, D., Sohndel, S., McCance, S.
G., O'Reilly, M. S., Llinas, M., and Folkman, J. (1996) J Biol Chem 271(46),
29461-29467.

O'Reilly, M. S., Holmgren, L., Shing, Y., Chen, C., Rosenthal, R. A., Moses, M.,
Lane, W. S., Cao, Y., Sage, E. H., and Folkman, J. (1994) Cell 79(2), 315-328.

Rios-Steiner, J. L., Schenone, M., Mochalkin, I., Tulinsky, A., and Castellino, F.
J. (2001) J Mol Biol 308(4), 705-719.

Mulichak, A. M., Tulinsky. A., and Ravichandran, K. G. (1991) Biochemistry
30(43), 10576-10588.

Mathews, I., Vanderhoff-Hanaver, P., Castellino, F. J., and Tulinsky, A. (1996)
Biochemistry 35(8), 2567-2576.

Chang, Y., Mochalkin, I., McCance, S. G., Cheng, B., Tulinsky, A., and
Castellino, F. J. (1998) Biochemistry 37(10), 3258-3271.

Wu, T. P., Padmanabhan, K., Tulinsky, A., and Mulichak, A. M. (1991)
Biochemistry 30(43), 10589-10594.

McCance, S. G., Menhart, N., and Castellino, F. J. (1994) J Biol Chem 269(51),
32405-32410.

Marti, D., Schaller, J ., Ochensberger, B., and Rickli, E. E. (1994) Eur J Biochem
219(1-2), 455-462.

Burgin, J., and Schaller, J. (1999) Cell Mol Life Sci 55(1), 135-141.

Nilsen, S. L., Prorok, M., and Castellino, F. J. (1999) J Biol Chem 274(32),
22380-22386.

33

68.

69.

70.

71.

72.

73.

74.

75.

76.

77.

78.

79.

80.

81.

82.

83.

Cao, R. H., Wu, H. L., Veitonmaki, N., Linden, P., Famebo, J., Shi, G. Y., and
Cao, Y. H. (1999) Proc Nat Acad Sci USA 96(10), 5728-5733.

Joe, Y. A., Hong, Y. K., Chung, D. S., Yang, Y. J., Kang, J. K., Lee, Y. S.,
Chang, S. I., You, W. K., Lee, H., and Chung, S. I. (1999) IntJCancer 82(5),
694-699.

Gately, S., Twardowski, P., Stack, M. S., Cundiff, D. L., Grella, D., Castellino, F.
J ., Enghild, J ., Kwaan, H. C., Lee, F., Kramer, R. A., Volpert, O., Bouck, N., and
Soff, G. A. (1997) Proc Natl Acad Sci U S A 94(20), 10868-10872.

MacDonald, N. J., Murad, A. C., Fogler, W. E., Lu, Y. Y., and Sim, B. K. L.
(1999) Biochem Biophys Res Comm 264(2), 469-477.

Southam, C., and Brunschwig, A. (1961) Cancer 14, 971-978.
Warren, B. A., Khan, 8., and Chauvin, W. J. (1977) Bibl Anat (16(Pt 2)), 396-402.

Woodruff, M. (1980) The interactions of cancer and host, Grune and Stratton,
New York

Fidler, I. J., and Balch, C. M. (1987) Curr Probl Surg 24(3), 129-209.

Clark, W. H., Jr., Elder, D. E., Guerry, D. t., Braitman, L. E., Trock, B. J.,
Schultz, D., Synnestvedt, M., and Halpem, A. C. (1989) J Natl Cancer Inst
81(24), 1893-1904.

Gorelik, E., Segal, S., and Feldman, M. (1978) Int JCancer 21(5), 617-625.
Gorelik, E., Segal, S., and Fredman, M. (1980) J. Natl. Cancer Inst. 89, 219-227.

Bonﬁl, R. D., Ruggiero, R. A., Bustuoabad, O. D., Meiss. R. P., and Pasqualini,
C. D. (1988) Int JCancer 41(3), 415-422.

Volpert, O. V., Ward, W. F ., Lingen, M. W., Chesler, L., Solt, D. B., Johnson, M.
D., Molteni, A., Polverini, P. J., and Bouck, N. P. (1996) J Clin Invest 98(3), 671 -
679.

Brem, S. S., Zagzag, D., Tsanaclis, A. M., Gately, S., Elkouby, M. P., and Brien,
S. E. (1990) Am JPathol 137(5), 1121-1142.

Matsubara, T., and Ziff, M. (1987) J C [in Invest 79(5), 1440-1446.

Matsubara, T., Saura, R.. Hirohata, K., and Ziff, M. (1989) J C [in Invest 83(1),
158-167.

34

84.

85.

86.

87.

88.

89.

90.

91.

92.

93.

Lay, A. J., Jiang, X. M., Kisker, O., Flynn, E., Underwood, A., Condron, R., and
HOgg, P. J. (2000) Nature 408(6814), 869-873.

Green, K. B.. and Silverstein, R. L. (1996) Hematol ()ncol Clin North Am 10(2),
499-530.

Holmgren, L., O'Reilly, M. S., and Folkman, J. (1995) Nat Med 1(2), 149-153.
Moser, T. L., Stack, M. S., Asplin, 1., Enghild, J. J., Hojrup, P., Everitt, L.,
Hubchak, S., Schnaper, H. W., and Pizzo, S. V. (1999) Proc Nat Acad Sci USA
96(6), 2811-2816.

Troyanovsky, B., Levchenko, T., Mansson, G., Matvijenko, O., and Holmgren, L.
(2001) JCell Biol 152(6), 1247-1254.

Rozengurt, E.. Heppel, L. A., and Friedberg, I. (1977) J Biol Chem 252(13),
4584-4590.

Rozengurt. E., and Heppel, L. A. (1979) J Biol Chem 254(3), 708-714.
Chahwala, S. B., and Cantley, L. C. (1984) J Biol Chem 259(22), 13717-13722.

Saribas, A. S., Lustig, K. D., Zhang, X., and Weisman, G. A. (1993) Anal
Biochem 209(1), 45-52.

Tarui, T., Miles, L. a., and Takada, Y. (2001) J. Mol. Biol. 276(43), 39562-39568.

35

CHAPTER II: X-RAY CRYSTAL STRUCTURE DETERMINATION

2.1 Branching enzyme

The ﬁrst step in every structure determination is the production of single and well
diffracting crystals. For this purpose it is essential to have access to large quantities of
highly pure material. The protein is set up for crystallization using sparse matrices of
precipitating solutions that have been proven successful in crystallizing other proteins.
The most common method for the crystallization of macromolecules is the hanging drop
vapor diffusion method (Figure 2.1). In this method, a drop containing a mixture of
protein and precipitating solution is equilibrated against a reservoir containing the
precipitant. The protein is slowly precipitated and the molecules adopt identical
orientations forming an orderly three dimensional array of molecules held together by
non-covalent interactions. The crystallization process not only involves setting up
thousands of drops, but it also involves the constant monitoring of these drops. It is
important to look for precipitation behavior, relative solubility and the appearance of
crystals. Based on the observations from the initial sparse screens, new optimized
screens can be made. This turns into an iterative process that can produce crystals
suitable for X-ray diffraction data collection. Such a strategy was followed in the

crystallization of BE and angiostatin.

36

 
   
   

vacuum grease

precipitating

\__.____,./ _
‘\ / solution

Figure 2.1 Crystallization using the hanging drop vapor diffusion method

 

Figure 2.2 A monoclinic crystal of glycogen branching enzyme. The crystal has

dimensions of 0.3 x 0.1 x 0.1 m3.

37

2.1.1 Crystallization

The recombinant native and selenium methionine (SeMet) substituted BE lacking
the ﬁrst 1 12 residues were overexpressed and puriﬁed by Dr. Binderup, a member of Dr.
Preiss’s laboratory (1,2).

The native (Figure 2.2) and SeMet BEs were crystallized and X-ray data was collected at
cryogenic temperatures (123K), to prevent crystal damage.

The branching enzyme crystals belong to the P21 space group with unit cell
parameters a = 91.44 A, b = 102.58 A, c = 185.41 A, and B = 91.380. Assuming four
molecules of branching enzyme (71.6 kDa) per asymmetric unit, the crystal volume per
protein mass is 3.1 A3 Da'l corresponding to approximately 56.5% solvent in the crystal.
This value is within the range observed for protein crystals (3). The crystal parameters
for the branching enzyme crystal are listed in Table 2.1. Data was 99.6% complete for
152,002 unique reﬂections derived from a total of 499,161 reﬂections. Detailed data

collection statistics are found in Table 2.2.

Table 2.1 Crystal parameters for the branching enzyme crystal.

 

 

Crystal form Monoclinic
Space group P2.
Unit cell a=91.44b= 102.58c= 185.41 A

a=y=90° and B =91.38°
Solvent content 56%

Molecules per asymmetric unit 4

 

 

 

38

Table 2.2 Statistics for the branching enzyme X-ray diffraction data collection

‘1

 

 

Cell parameters (A, deg)

(a. b, c and [3)

Completeness (%)
8...... (I) ¥ (%)

 

 

91.48, 102.56,
185.10 and 91.45
99.6 (98.6)
8.6 (30.3)

10.4 (2.6)

 

91.65, 102.48,
195.92 and 91.68
94.2 (77.5)
10.1 (29.2)

9.3 (1.5)

 

Native 1 Se-Met 2 Hg soak 3
Wavelength (A)0.97794 0.97938 1.54180
Resolution range (A) 35.0 — 2.3 20 — 2.5 40 — 3.5
(last resolution shell) (2.38 — 2.30) (2.59 — 2.50) (3.63 — 3.50)

91.57, 102.79,
185.58 and 91.68
91.5 (86.3)
22.8 (50.0)

5.1 (2.3)

 

1. Data collected at the Advanced Photon Source, Structural Biology Center ID19

beamline

2. Data collected at the Advanced Photon Source, IMCA beamline ID17

3. Data collected at Michigan State University Macromolecular X-ray Facility home

8011106

1 Values in parentheses refer to the last resolution shell

¥ Rmerge = Z 11 - (1)1 / Z I, where [is an individual intensity measurement and (I) is the

average intensity for this reﬂection, with summation over all data

39

 

2.1.2 Structure determination

Once high resolution data have been obtained, an electron density map is
calculated. It is from this electron density map that the three dimensional structure of the
protein is deduced. This map not only has the positions of the main and side chain atoms
but also includes solvent molecules, ions, substrates and any other molecules that form

the lattice. The electron density is represented by the following equation,

1 —7r + +2
p(xayaZ)=‘—I;Z;;Fhkte 201x ky I)
h

The structure factor (F hkl) for each reﬂection labeled hkl, is a complete description of all
the atoms that contribute to that reﬂection. Fhk. is a wave function with frequency,
amplitude and phase. The frequency is the same as the X-ray source and the amplitude is
proportional to the square root of the measured intensity of the reﬂection. The only
information that is unknown is the phase; this is known in crystallography as the phase
problem.

The phase problem can be solved by any of the following methods; molecular
replacement, isomorphous replacement or multiple wavelength anomalous dispersion.
The molecular replacement method is used when the protein is homologous to another
protein for which the structure has been solved.

1n the case of a de novo structure determination, anomalous dispersion,
isomorphous replacement, or a combination of both methods can be used. BE was a de
novo structure determination and its phases were a hard and challenging problem to
solve. This was because BE is not only a large enzyme (616 residues for the truncated

form), but there are four molecules of BE in the asymmetric unit. The asymmetric unit is

40

the simplest volume of the unit cell that can generate the complete lattice by symmetry
operations. The solution of an X-ray crystal structure involves the determination of the
contents of the asymmetric unit. The asymmetric unit can consist of one, or more than
one, molecule and each of the molecules within the asymmetric unit will be related to
each other by non-crystallographic symmetry operators. The structure of BE was solved

by a combination of the isomorphous replacement and anomalous dispersion methods.

2.1.2.1 Heavy atom isomorphous replacement

In this technique, differences in the diffraction pattern are measured after the
introduction of a heavy atom. In order to be able to record these differences reliably, the
atom or group of atoms must have many electrons. If hard ions (Ianthanides and
actinides) are use as derivatives, the interaction between the protein and the heavy atom
is usually ionic. On the other hand, soft ions (mercury, gold, platinum, etc.) tend to react
with sulfurs on cysteines, deprotonated nitrogen on histidines or even with sulfur from
methionine (4). These interactions are covalent, so they tend to bind more speciﬁcally.
For this technique to work, it is important that the protein molecules within the crystal be
identically bound by the heavy atom compound, with essentially no change in the
structure of the native crystal lattice.

The coordinates and the diffraction pattern of the heavy atom alone can be
determined by calculating the difference between the diffraction patterns of the native
crystal and the heavy atom replaced crystal. This is because the contributions from every
atom to a reﬂection are combined in an additive way. With the diffraction pattern of just

the heavy atom, we have a small number of atoms and an easier structure to be

41

determined. For example, BE has four Hg atoms in its asymmetric unit, a rather simpler
pattern than that of the more than 20,000 non-hydrogen atoms in the protein structure.

A systematic and exhaustive search for isomorphous heavy atom derivatives was
performed; all 37 heavy atoms tried are listed in Table 2.3. Two different methods were
used to obtain the heavy atom derivative. In the ﬁrst method, pre-grown crystals were
soaked with the heavy atom reagents at various concentrations through different time
spans. Another method involved the co-crystallization of the crystal in the presence of
the heavy atom at various concentrations. However, crystals could not be grown in the
presence of any of the heavy atoms, not even at heavy atom concentrations as low as 1
nM.

The search for heavy atoms was done by soaking the crystals in a lmM heavy
atom solution overnight. Visible signs of reaction or crystal degradation such as cracking
or color change were looked for. Based on these results, repeated soaks were done at
lower concentrations. Compounds that show no evidence of degradation were tried at a
higher concentration. The soaking time was varied in both cases as well. Data sets were
compared with the native for intensity differences and isomorphism. From all the heavy
atoms tried, only mercuric chloride, mercuric acetate, mercurochrome, uranyl nitrate, p-
chloro mercuri benzoic acid and p-chloro mercuri benzene sulphonate were isomorphous
and presented signiﬁcant intensity differences.

The automated heavy atom Patterson search routines from the program SOLVE
were used to ﬁnd the heavy atom positions (5). The Patterson (Pm...) search is performed
by the calculation of a vector map known as the Patterson function. This Patterson map

is calculated based on the product of the electron density at two points separated by a

42

distance U,

P... = lp(x, y,2)p(x + u, y + W + W)d V
V

PWW = __1_ 22 21172th le—2m'(hu+kv+lw)
V h k 1

The Patterson map is calculated using only the contribution from the heavy atom.

Puvw = iZZZIFZh/d' e-Zm'(hu+kv+lw)
V h k l

H

where |F 1,111” is the structure factor contribution of the derivative. A peak will only be
observed when there is an electron density contribution from the heavy atom at both
positions (xyz and x + u, y + v, x + w). The result obtained from the Patterson function,
using only the contribution of the heavy atom, are peaks located at positions deﬁned by
interatomic vectors between heavy atoms. Symmetry within the unit cell is used to
transform these interatomic vectors into crystallographic coordinates.
This process, although simpliﬁed from the original protein structure, can be cumbersome
and time consuming. Today we have the advantage of fast and automated programs that
will perform these calculations.

Although the mercury positions were determined, this experiment didn’t provide
enough phasing information to solve the structure. The poor diffraction of the derivatives

(3.5 to 5.0 A) resulted in phases that were useful only to low resolution.

43

Table 2.3 List of all the heavy atom compounds tried

ﬂ

OO\10\tlt-I>- b)

\O

11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37

Hng
H8(C2H302)2
C13H17H8N06

CngHgOZS
CongBrzHg06N32
C6H4HgClSO3
CH3Cng

H gC l
H8(SCN)2
HgO
C7H5C1HgOz
K2PICI4
K2PIC16
Kth(CN)4
KAUCI4
KAu(CN)2
(CH3)3Pb(C2H302)
C6H9L306
LaCl3
La(NO3)3
LazO3
U02(CH3C02)2
U02(NO3)2
Sm (CH3COz)3
8111203
Tl(C2H3Oz)
PT(C2H3O2)3
C€(C2H302)3
N32WO4
YbCl3

NbCl3
K20$CI6

EI'CI3

KI

EUCI3

Hglz

LUCI3

Mercuric chloride

Mercuric acetate
2-(N-(3-hydroxymercuri-2-
methoxypropyl)carbamoyl) phenoxyacetic acid
Thimerosal (EMTS)
Mercurochrome
p-Chloromercuri benzene sulphonate
Methyl mercury chloride
Mercurous chloride

Mercuric thiocyanate

Mercuric oxide
p-Chloromercuri benzoic acid
Platinum (II) potassium chloride
Platinum (IV) potassium chloride
Potassium tetracyano platinate
Potassium tetracyano aureate III
Potassium dicyano aureate
Trimethyl lead acetate
Lanthanum acetate

Lanthanum chloride

Lanthanum nitrate

Lanthanum oxide

Uranyl acetate

Uranyl nitrate

Samarium acetate

Samarium oxide

Thallium acetate

Praseodymium acetate

Cerium acetate

Sodium tungstate

Ytterbium chloride

Niobium chloride

Potassium hexachloroosmiate
Erbium chloride

Potassium iodine

Europium chloride

Mercury iodine

Lutetium chloride

44

2.1.2.2 Anomalous dispersion

Like the isomorphous replacement method, the strategy in anomalous dispersion
is to solve a smaller and simpler structure. In this case, anomalous scatterers are
introduced in the protein molecule. In BE, we substituted the sulfurs in the methionines
with selenium by expressing the protein in the presence of SeMet. Advantage is taken of
the anomalous differences by irradiating the crystal with X-ray radiation at the absorption
edge of the scatterer (0.98 A for Se).

A single wavelength anomalous dispersion (SAD) experiment was performed at
the selenium absorption edge to a resolution of 2.5 A. BE has a total of 64 Se (16
Se/monomer), so this substructure was somewhat complex to solve. In this case, the 64
Se will provide enough phasing information, but the difﬁculty resides in ﬁnding the
positions of all 64 seleniums. To simplify the problem, we combined the SAD data with
the single isomorphous replacement (SIR) experimental data, previously collected.
Search routines were used with the SAD data and different combinations of the six SIR
data sets (5). Of all the SIR data sets, the p-chloro mercuri benzoic acid soak was the
most successful in locating Se sites. Four mercury sites and 27 of the total 64 selenium
sites were eventually identiﬁed (5). These sites were reﬁned and an initial set of phases
was calculated using the program SHARP (6). This information was used to identify a
total of 61 selenium sites (6).

Among all the programs available, SHARP presents the most powerful software
by using the maximum likelihood method and taking into account variables previously
ignored (6). The maximum likelihood method follows the principle of least squares with

the difference that the calculated and observed values are actually distributions and not

45

set values. The parameters are then varied in such a way that the calculated values
approach the observed ones. The ﬁrst programs that were use to determine the phases
employed a straight implementation of the least square method. Those programs took the
phase information generated by the derivative in the reﬁnement and in the recalculation
of the structure factors of the native data, introducing bias into the observable. Also,
previously developed programs assumed that the native amplitude measurement is error
free causing the measurement errors to be combined with the error caused by the lack of
isomorphism of the derivative, underestimating the phasing power of the derivative.
SHARP assigns a probability distribution not only to the phase but also to the structure
factor. This new generation of programs made the determination of the structure of BE
possible, something that under the same circunstances would not have been possible ten
years ago.

The contribution of the derivative to the phase determination is known as the
phasing power (PP). The PP is deﬁned by the following equation,

ZIFWIHZ

hkl

Z thkl,obs

hkl

)2
PH

where thkllll is the structure factor contribution of the derivative, |F 1,1,71,01,51,)” is the

 

 

PH — thkl,calc

observed structure factor of the protein plus derivative and IFhkLca1clpH is the structure
factor of the protein plus derivative calculated based on the phases determined by the
derivative. A table of the phasing power versus resolution for the mercury derivative,

and the selenium SAD data set is presented in Table 2.4.

46

With this information, an electron density map was calculated; and although the

quality of this map was poor, it was good enough to determine the boundaries of each

molecule in the asymmetric unit (Figure 2.3a). The non-crystallographic symmetry

elements were determined and the quality of the initial electron density map was

improved by applying non-crystallographic four fold symmetry averaging (Figure 2.3b).

Using this map we were able to build all four molecules in the asymmetric unit of BE

(Figure 2.3b). Figure 2.3 shows the electron density map calculated before averaging

(Figure 2.3a), after averaging (Figure 2.3b) and the ﬁnal map after reﬁnement (Figure

2.3c).

Table 2.4 Phasing power of the mercury and selenium methionine derivatives

 

 

 

Resolution, A Hg Phasing power Se Phasing power Se Phasing power
isomorphous anomalous
10.76 1.97 2.08 2.24
6.85 1.71 2.09 1.92
5.34 1.26 1.45 1.47
4.52 0.965 1.09 1.23
4.00 0.800 0.948 1.00
3.61 0.705 0.815 0.834
3.33 0.812 0.769
2.63 0.788 0.712

 

 

 

 

47

 

  
 

  

 
   
 

A
~12in
his

5?? &
{EEG-g»
i

 
   

l

 

c)

Figure 2.3 A section of the electron density map. a) Initial experimental electron density
map b) Electron density map after four fold averaging c) Final electron density map aﬁer

reﬁnement.

48

2.1 .3 Structure refinement

Once the BE structure was traced, it was subjected to multiple rounds of structure
reﬁnement. Structure reﬁnement involves the adjustment of the model to ﬁnd a closer
agreement between the calculated and the observed structure factors. The agreement
index between the calculated and observed structure factors is represented by the Ram),

described below:

 

F obs I —

 

F calc

 

 

El
factor = 2 F

obs l

 

R x100

 

The BE structure was reﬁned using the simulated annealing method (7,8). This
method uses molecular dynamics to simulate the various parameters in the
conformational space of the molecule. In the annealing process the molecule is heated
until all particles arrange themselves in the liquid phase. This is followed by slowly
cooling, so that all particles will arrange in the lowest energy state. The target function
consists of an empirical potential energy which is described by the stereochemistry and

nonbonding interactions in the macromolecule.

Reﬁnement was followed by the addition of water molecules and resolution
extension to 2.3A. The ﬁnal reﬁnement parameters are listed in Table 2.5. The ﬁnal
model consists of 19,323 non hydrogen atoms with two disordered regions between
residues 361 to 373 and 414 to 429 in all four molecules. The structure also includes
1,142 water molecules that represent ordered water molecules that form part of the

lattice. An example of the ﬁnal 2FO - FC electron density map is shown in Figure 2.4.

49

Figure 2.5 presents the Ramachandran plot of the BE structure. A Ramachandran
diagram is a plot of (1) (angle between N and C01) versus (p (angle between C and Ca).
According to geometry and steric restrictions the 41 and (p angles must be within certain
values, which are marked by the different shadows of gray in Figure 2.5. All residues,
with the exception of glycines and prolines, that don’t have a side chain, must lie within
these allowed regions. In the BE structure only 5 residues out of the 505 non glycine
residues lie in disallowed regions. This corresponds to only 1% of the structure a very
small fraction for a structure of this magnitude. Among these residues M472 and L475

lie in a sequence rich in glycines which could explain their troubled geometry.

Table 2.5 Reﬁnement statistics of BE

 

 

Rl‘actor 20.00%
Rfree 26.53%
Resolution 2.3 A
rmsd bonds 0.0075 A
rmsd angles 1.450

 

 

 

50

 

Figure 2.4 Stereoview example of the ﬁnal 2F0 - Fe electron density map.

51

 

Psi (degrees)
0

 

 

0
Phi (deems)

 

Figure 2.5 Ramachandran plot of BE. Showing one of the four molecules for clarity.

52

2.1.4 Materials and methods
2.1.4.1 Crystallization

The BE enzyme was puriﬁed in Dr. Preiss’s laboratory according to protocols
previously published. The enzyme activity was determined by using three different
assays (9,10). The puriﬁed protein was buffer exchanged in 25mM
N-(2-hydroxyethyl)piperazine-N-ethane sulfonic acid (hepes), pH 7.5, and concentrated
to approximately 5 mg/ml.

A homogeneous and active protein was screened for crystallization by using the
hanging drop vapor diffusion method (Figure 2.1). The reservoir contained 650 1.1L of the
precipitation solution and the 4 11L hanging drop consisted of a 1:1 protein to
precipitation solution ratio. The search for initial crystallization conditions was
performed through sparse matrix sampling by using different crystallization screens at
298 K and 277 K (11,12). Crystals were formed at 277 K from a solution containing 100
mM Hepes at a pH of 7.2. The ﬁrst crystals appeared after two weeks and reached a

maximum size of 0.3 x 0.1 x 0.1 mm3 in four weeks (Figure 2.2).

2.1.4.2 Native data collection

The crystals were transferred to a cryoprotectant solution containing 25% (v/v) 2-Methyl-
2,4-pentanediol (MPD), 2% (m/v) polyethylene glycol (PEG) 4,000 and 100 mM Hepes,
pH 7.5. The crystals were then mounted in nylon cryo-loops and ﬂash frozen by
immersion in liquid nitrogen. A high-resolution native data set was collected at the
Advanced Photon Source (APS) at The Argonne National Laboratories (Argonne, IL) on

the Structural Biology Center ID-l9 (SBC) beamline (Tables 2.1 and 2.2). Intensity data

53

were collected by using a 3 x 3 array (3072 x 3072 pixels) custom made CCD area
detector to a resolution of 2.3 A. The crystal to detector distance was set to 220 mm, and
160° of data was collected with an oscillation angle of 05°. Diffraction data was

processed by using the HKL2000 program package (13.14).

2.1.4.3 SIR and SAD data collections

An SIR experiment was also carried out, in which a native crystal was soaked for
18 hours in a solution containing 10% MPD, 0.1M Hepes pH 7.20 with 10 uM of p-
chloro mercuri benzoic acid. The crystals were cryoprotected and frozen, as previously
described. Data were collected over 160° with oscillations of 1°. A total of 118,955
reﬂections were measured at our home source by using a Rigaku R-AXIS IV” image
plate detector (Table 2.2). Cu Koc radiation was generated by a Rigaku RU-200 rotating
anode source operating at 50 kV and 90 mA.

In addition to the SIR data, a SAD experiment was also performed. The Se-Met
substituted protein was crystallized and cryoprotected with the same conditions as the
native crystals and a SAD experiment was performed at the selenium absorption edge.
Anomalous data, to a resolution of 2.5 A, were collected in a single element 165 mm
MAR CC D detector from beamline 17-ID in the facilities of the Industrial
Macromolecular Crystallography Association Collaborative Access Team (IMCA-CAT)
at the APS. The crystal to detector distance was set to 190 mm, and 180° of data was
collected with a 05° oscillation (Table 2.2).

The automated heavy atom search routines from the program SOLVE were used

to ﬁnd the four mercury and 27 initial Se positions (5). The program SHARP was used to

54

calculate an initial set of phases and ﬁnd a total of 61 Se sites (Table 2.4) (6). An
improved electron density map was calculated with the aid of four fold non-
crystallographic symmetry averaging using the Dork program package graciously
provided by the author, Dr. Greg Van Dyne, University of Pennsylvania. All model
building was done using TURBO FRODO (15) (16) and reﬁnement and map calculations

were carried out using CNS (Table 2.5) (17).

55

2.2 Angiostatin
2.2.1 Crystallization and data collection

EntreMed Inc., (Rockville, MD) provided the human angiostatin protein
containing Kgl-3. EntreMed Inc., has the patent on angiostatin and is running its clinical
trials. This protein was crystallized (Figure 2.6) and a complete data set to a resolution of
1.75 A was collected. The crystals are tetragonal and belong to the P422 space group
with unit cell parameters a = b = 56.94 A and c = 192.97 A. Assuming one molecule of
angiostatin (29.77 kDa) per asymmetric unit, the crystal volume per protein mass is 2.65,
which corresponds to approximately 51% solvent content in the crystal. This value is
within the range observed for protein crystals (3). The crystal parameters of the
angiostatin crystal are listed in Table 2.6. A synchrotron X-ray diffraction data set to a
resolution of 1.75 A, with an overall 1/0 of 19.5, was obtained. The X-ray diffraction
data was 92.8% complete with an Rmerge of 7% for 30,370 unique reﬂections from a

total of 217,983 measured reﬂections. Detailed data collection statistics are found in

Table 2.7.

2.2.2 Molecular replacement and structure reﬁnement

The structure was solved by molecular replacement using the structures of Kgl
and Kg2 as models (PDB id ICEA and 115K) (18,19). The molecular replacement
method uses phases from a known protein structure as an initial estimate of the phases of
the new protein. The challenge of this method lies in ﬁnding the correct orientation and

position of the model in the unit cell of the new protein. Although unit cell dimensions

56

 

Figure 2.6 Tetragonal crystals of human angiostatin(Kg1-3). The square plate crystals

have dimensions of 0.7 x 0.7 x 0.4 m3.

57

Table 2.6 Crystal parameters for the angiostatin crystal

 

 

Crystal form Tetragonal
Space group P41212
Unit cell a = b = 56.94, c = 192.97 A

anda=0=y=90°
Solvent content 51%

Molecules per asymmetric unit 1

 

 

 

Table 2.7 Statistics for the angiostatin X-ray diffraction data collection

 

 

Wavelength, A 1.0332
Resolution range, A 50 — 1.75
(last resolution shell) (1.81 — 1.75 )
Completeness, % 92.8

(last resolution shell) (83.0)
1/0' 19.5

(last resolution shell) (5-1)
Rmerged , % 7.0

(last resolution shell) (34.1)
Unique reﬂections 30,370
Measured reﬂections 217,983

 

 

 

58

and symmetry place some constraints, the process is still complicated. For this reason it
is divided in two steps; calculation of a rotation function that is followed by a translation
function. The rotation function uses Patterson maps to determine the correct orientation
of the model. As was previously discussed, a Patterson map is independent of the
position of the structure in the unit cell as long as the orientation does not change. The
solution process involves the calculation and evaluation of Patterson maps. This is
performed by the automated Patterson search routine of the program AMoRe (20). Once
the correct orientation of the model has been found the correct position can be determined
using the translation function. The translation search is done systematically over the unit
cell and evaluated based on the agreement of the calculated structure factors with

observed ones.

The structure factors of a properly positioned model are then calculated and a
correlation coefﬁcient (C) and an R value will determine if a correct solution have been
found. These two values are deﬁned below, where Fobs and Fcalc are the observed and

calculated structure factors.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

2 F obs _ Fla/c
R = h“ x 100
Z bes
hkl
Z( F‘obsz— obs _—2-1-72)( calc calc —2)
C : hkl 2 2 “2
2(170092— obs 22) 2(F2 calc calc )
hkl hkl

59

The human plasminogen Kgl and Kg2 structures were used as the protein models
(18,19). Both models gave rotation solutions that were later used in the translation
function. A translation search with Kg2 gave two solutions; a search with Kgl also gave
two solutions, one of which was unique, relative to the initial Kg2 search. Examination
of the packing of the Kg2 solutions showed them to be disulﬁde linked Kg2 and Kg3.
These solutions had a correlation factor of 32.6 and 30.8 and an R of 49. 1% and 49.3%,
respectively. Fixing the positions of Kg2 and Kg3 and calculating an electron density
map revealed density corresponding to the unique Kgl solution, indicating it to be Kgl.
This was corroborated with a new translation search that also produced the unique

solution that corresponded to Kgl.

This multi kringle angiostatin structure contains the ﬁrst structure of Kg3 to be
determined. The high homology among Kgs (50% average identity) allowed use of Kgl
to approximate Kg3 in the electron density. Residues different from Kgl were mutated
to alanine and an electron density map of Kgl, Kg2 and Kg3 was calculated by using the
program CNS (17). Analysis of the map allowed replacement of the previously mutated
residues to the corresponding amino acids in Kg3. The interkringle peptides connecting

Kgl to Kg2 and Kg2 to Kg3 were also built.

Multiple rounds of structure reﬁnement using the simulated annealing method
followed by the addition of water molecules and resolution extension resulted in the ﬁnal
reﬁnement parameters listed in Table 2.8. The ﬁnal model includes 253 residues (from
amino acid 81-333) and 398 water molecules. In addition to the 398 water molecules,

electron density for a bicine molecule, the buffer used in crystallization, was located in

60

each of the three Kg LBS’s. The Ramachandran plot of the angiostatin structure shows
no residues in disallowed regions (Figure 2.7). An example of the ﬁnal 2F0 - Fc electron

density map is shown in Figure 2.8.

135

Psi (degrees)

              

—135 0 45 90 :0

Phi (degrees)

-180 -90 -45

Figure 2.7 Ramachandran plot of angiostatin

61

Table 2.8 Reﬁnement statistics of angiostatin

 

 

 

 

R-factor 1 9.58%
Rﬁu 26.25%
Resolution 8.0-1.75 A
rmsd Bonds .011 A
rmsd Angles l.6°

 

 

Figure 2.8 An example of the ﬁnal 2Fo - Fc electron density map of angiostatin. The map

is centered at residue W315 and also shows residues H317 and W325.

62

2.2.3 Materials and Methods

The human angiostatin mutant N289E (this mutant lacks N-linked glycosylation)
containing Kgl-3 was expressed in Pichia pastoris and puriﬁed as previously described
(21 ). The puriﬁed protein, provided by the company EntreMed Inc. Rockville, MD, was
buffer exchanged into saline buffer (0.15 M NaCl) and concentrated to 15 mg/ml.

The protein was extensively screened for crystallization by using the hanging
drop vapor diffusion method (Figure 2.1). The search for well diffracting crystals was
performed using several sparse crystallization screens at two different temperatures, 298
K and 277 K (11,12). A 4 11L hanging drop in a 1:3 protein to precipitant solution ratio
was equilibrated against 650 11L of precipitant solution. The best crystals were obtained
at 277 K, with a precipitant solution containing 10% PEG 20,000, 2% (v/v) dioxane and
100 mM N,N Bis(2hydroxyethyl) glycine (bicine) buffer at a pH of 9.0. The crystals ﬁrst
appear overnight and grow to a maximum size of 0.7 x 0.7 x 0.4 mm3 in three days
(Figure 2.6).

The crystals were transferred to a cryoprotectant solution containing 35% (v/v)
glycerol, 10% PEG 20,000, 2% dioxane and 100 mM bicine pH of 9.0 and ﬂash frozen
by immersion in liquid nitrogen. X-ray diffraction data to a resolution of 1.75 A was
collected at the SBC beamline at the APS. Data were collected by using a custom built 3
x 3 array (3072 x 3072 pixels) CCD area detector. The crystal to detector distance was
150 mm and 120° of data was collected with an oscillation of 03°. Diffraction data was

processed using HKL2000 (14).

63

The structure was solved by molecular replacement using the program AMoRe
(20). All model building was done by using TURBO FRODO (15,16) and reﬁnement

and map calculations were carried out by using CNS (Table 2.8) ( 1 7).

64

2.3 Literature cited

DJ

10.

11.

12.

13.

14.

15.

16.

17.

Guan, H., Li, P., Imparl-Radosevich, J., Preiss, J ., and Keeling, P. (1997) Arch.
Biochem. Biophys. 342, 92-100.

Binderup, K., Mikkelsen, R., and Preiss, J. (2000) Arch Biochem Biophys 377(2),
366-371.

Mathews, B. W. (1968).]. Mol. Biol 33, 491-497.

Blundell, T. L., and Johnson, L. N. (1976) in Mol. Biol, Int. series of books and
monographs, pp. 183-239, Academic Press, New York

Terwilliger, T. C ., and Berendzen, J. (1999) Acta Cryst. D55, 849-861.

La F ortelle, E., and Bricogne, G. (1997) Methods Enzymol 276, 472-494.
Kirkpatrick, S., Gelatt, C ., and Vecchi, M. (1983) Science 220, 671-680.
Brunger, A. T., Kuriyan, J., and Karplus, M. (1987) Science 235, 458-460.
Binderup, K., and Preiss, I. (1998) Biochemistry 37(25), 9033-9037.

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

Cudney, R., Patel, 8., Wesisgraber, K., Newhouse, Y., and McPherson, A. (1994)
Acta Cryst. D50, 414-423.

Jancarik, J., and Kim, S. H. (1991).]. Appl. Cryst. 24, 409-411.

Otwinowski, Z. (1993) in Data Collection and Processing (Sawyer, L., Issacs, N.,
and Bailey, 8., eds), pp. 56-62, SERC Daresbury Laboratory, Daresbury, U.K.

Otwinowski, Z. M., W. (1997) Methods Enzymol 276, 307-326.
Jones, T. A. (1985) Methods Enzymol. 115, 157-171.

Jones, T. A., Zou, J. Y., Cowan, S. W., and Kjeldgaard, M. (1991) Acta Cryst.
A47, 110-119.

Brunger, A. T. (1992) X-PLOR, version 3.1, a Systemfor X -ray Crystallography
and NMR., Yale University Press, New Haven, CT

65

l8.

19.

20.

21.

Mathews, I., Vanderhoff-Hanaver, P., Castellino, F. J ., and Tulinsky, A. (1996)
Biochemistry 35(8), 2567-2576.

Rios-Steiner, J. L., Schenone, M., Mochalkin, I., Tulinsky, A., and Castellino, F.
J. (2001) J Mol Biol 308(4), 705-719.

Navaza, J. (1994) Acta C ryst A50, 157-163.
Shepard, S. R., Boucher, R., Johnston, J ., Boemer, R., Koch, G., Madsen, J. W.,

Grella, D., Sim, B. K. L., and Schrimsher, J. L. (2000) Prot. Exp. Purif. 20(2),
216-227.

66

CHAPTER III: THE THREE DIMENSIONAL STRUCTURE OF

BRANCHING ENZYME

This chapter presents the three dimensional structure of BE from e. coli. This is
the ﬁrst structure of any enzyme involved in glycogen biosynthesis and the only member
of the amylotic family of enzymes of which the structure was still unknown. Analysis of
the structure provided valuable information that allowed us to propose a mechanism for
BE catalytic action. This structure and the information obtained from it represent one of

the most important pieces of information obtained in the ﬁeld of glycogen biosynthesis.

3. 1 Overall structure

The enzyme used in the structure presents a truncation at the amino terminal
lacking the ﬁrst 112 residues. It portrays an altered branching pattern when compared to
WT BE. This truncated enzyme has a higher propensity for transferring glucose chains
of 12 units while the WT BE has a higher propensity for transferring branches of 6
glucose units (1).

The four molecules in BE asymmetric unit consists of 19,323 residues and 1,142
water molecules. Each molecule extends from residue 117 to residue 728, with the ﬁrst
four residues disordered. There are two disordered regions between residues 361 to 373
and 414 to 429 in all four molecules. The overall elliptical structure of BB has
dimensions of 87.7 A by 42.6 A and 42.0 A in depth.

The structure of BB is organized into three domains: the C-terrninal, N-terrninal

and (or/B) barrel (Figure 3.1). The C-terminal domain consists of 116 residues organized

67

N-terrninal

 

 

(Cl/B) barrel

 

Resolution = 2.3 A
Rfactor = 20.00%

Rfree = 26.53%

 

 

 

Figure 3.1 Three dimensional structure of e. coli BE truncated at the amino

terminus at amino acid 113.

68

in seven [3 strands. The N-terminal domain contains a 13 sandwich fold, which was
previously predicted by sequence analysis (2). This N-terminal domain is composed of
128 residues arranged in seven 13 strands. The central (or/B) barrel domain common in
members of the or-amylase family of enzymes extends from residue 241 to residue 612
comprising a total of 372 residues. This domain contains the residues involved in
catalysis and presents a substrate binding cavity with dimensions of 30.5 A x 17.7 A x
17.7 A big enough to accommodate branched glucose chains.

A complete (01/13) barrel should contain eight or-helices and eight B-strands. The
(or/13) barrel domain in BE is missing a5, the a-helix between [3 sheet number ﬁve (05)
and six ([36). This barrel also has three extra helices inserted; 011 a, 016a and a7a. The
011a helix located between 131 and a1 and the a7a located between 137 and a7 are both
one turn helices. The a6a helix positioned between a6 and [37 is a three turn helix. This
variation of the (01/ 13) barrel domain is also observed in isoamylase but not in other
members of the family. There are two loops connecting the domains in BE. The loop
that connects the N-terrninal domain to the (Ct/B) barrel is eighteen residues long (223-
240), and the loop joining the end of the (ct/13) barrel to the C-terminal domain is thirteen
residues long (613-625). The organization of the elements of secondary structure is
depicted in Figure 3.2 and summarized in the amino acid sequence diagram of Figure 3.3.

There are four molecules in the asymmetric unit of BE, which are oriented as
shown in Figure 3.4. There is a two fold rotation that relates molecule A to C, and B to
D, but all four molecules are not related by a perfect four fold. Instead the two folds are

coupled with a translation that relates both two folds.

69

 

Figure 3.2 The elements of secondary structure in the three domains of BE. The 0 sheets

from the N and C terminals are identiﬁed with an N and a C, respectively.

70

1 l3
LSEGTI ILR

121 l’Yl-Z‘l‘l-(i/\llz\l) 'I'MlXiV’l'Gl‘Rl“ SVWAI’NARRV SVVGQI’NYWI) (iRRlll’lVlRlRK ESGIWEIJ‘II’

181 (1A1 INGQLYKY EMIDANGNLR LKSI)PYAFEA QMRPE'I‘ASLI CGI.PEKVVQT EERKKANQFD

 

 

F Bl J( (118 0 L oTT Q
241 APISIYEVHL GSWRRHTDNN FWLSYRELAD QLVPYAKWMG rrnreurm EHPFDGSWGY

 

C <12 O l 133 l
301 QP'I'GI.YAPTR RFG'I‘RDDFRY FIDAAHAAGL anowvmn FPTDDFALAE FDGTNLYEHS

 

 

1 break 1 C 013 T) I break

r
361 DPRl—ZGYHQDW N'l‘l.lYNYGRR EVSNI’LVGNA LYWIERI’GID ALRVDAVASM IYRDYSRKEG

 

 

——1 a4 0 l 16
421 l-IWII’NI€I“GGR ENIEAIEFLR NTNRIIAIEQV SGAVTMAEliS 'I‘DFPGVSRPQ DMGGLGFWYK

 

 

<16 O -1
481 WNLGWMHDTL DYMKLDPVYR QYHHDKLTFG ILYNYTENFV LPLSHDEVVH GKKSILDRMP

 

 

C 01.7 O ( (18 -
541 GDAWQKFANL RAYYGWMWAF PGKKLLFMGN EFAQGREWNH DASLDWHLLE GGDNWHHGVQ

 

o

601 RI.VRDI.N1.’I‘Y RIIIIKAMI IEID li‘DI’YGl’I-IWLV VDDKI‘IRSVLI FVRRI)K1€(]NE IIVASNFTPV

 

[C‘9411C951

661 PRIIDYRFGIN QI’GKWREIIN TDSMHYHGSN AGNGG'I‘Vl-ISD EIASHGRQHS LSLTLPPLAT

721 [\VINRIZAE

Figure 3.3 The amino acid sequence of the truncated BE with elements of secondary

structure.

71

Molecule C Molecule D

 

Molecule A

Figure 3.4 There are four molecules in the BE asymmetric unit.

3.2 Structural differences among members of the a-amylase family

BE belongs to the a-amylase family of enzymes, which includes a-amylases,
CGTs, isoamylases and BEs. Even though members of this family share some structural
features, like the (01/13) barrel domain and the conserved catalytic residues, each enzyme
performs different reactions (Figure 3.5). (Jr-Amylase and isoamylase hydrolyze 01-1,4
and CX-I,6 glucosidic bonds, respectively. Branching enzyme and CGT catalyze
transglycosylation reactions with BB being the only one with speciﬁcity for two different
glucosidic bonds, 01-1.4 and (it-1,6. CGT catalyzes the formation of cyclodextrins by
cleavage and subsequent transglycosylation of a-l ,4 links. The similarity among the
primary structure of members of the a-amylase family is very low, preserved only among
the four regions that contain the conserved residues.

X-ray structures of members of the family showed the existence of a conserved
barrel domain (Figure 3.6) (3) (4,5). BE was the only member of the a-amylase family
for which the three dimensional structure was still unknown. The structure of BE
revealed that indeed BE has a central (01/13) barrel that contains the catalytic residues
(Figure 3.1). This domain is similar to the (01/13) barrels of other members of the family
(Figure 3.6). Also, the C-terminal domain of BE is structurally similar to the C-terminals
of isoamylase and a-amylase while the N-terrninal [3 sandwich domain of BE is
analogous to the N-tenninal domain in isoamylase (4,5). BE and isoamylase are
therefore the more structurally similar members of the (it-amylase family. A comparison

between domains is presented in Figure 3.7 where each domain in isoamylase has been

rotated and positioned in the same orientation as is BE. Isoamylase and BE are the only

73

0 OX Ot— amylase
10“ 0H 0H0
0 \H X0 OOH

 

OH OH
H OH
b) C 20 CHZOH 0“on CH20H CHZOH
\OH \ OH
clD‘LHI/ o
H O
OH H isoamylase
CHZOH CH2 CHZOH
o 01‘1on CHZOH
OH OH OH
0 0 O OH
OH OH CH
C) CHon CH20H CHZOH
O O Q CGT p Cyclic sugar chain
’OH \\ OH OH
0 0
OH
OH OH OH OH
linear sugar chain
CH20H CHZOH
d) CHZOH CHZOH CHZOH BBQOOHOO\
O o O
OH OH OH
0 0
OH 0H_—__——> CH20HH CH OH
OH OH OH 2
OH
O
OH

Figure 3.5 The reactions catalyzed by the members of the a-amylase family of enzymes.

a) a-amylase hydrolyses (1-1,4 bonds. b) isoamylase cleaves 01-1 ,6 bonds. c) CGT

74

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Figure 3.6 X-ray structures of members of the tit-amylase family of enzymes (3) (4,5).

75

Branching Enzyme Isoamylase

  

(a/B) - barrel

 

C-terminal

Figure 3.7 Comparison between domains of isoamylase and BE (5). The domains in

isoamylase have been rotated to match the orientation of BE.

76

members in the family catalyzes the formation of cyclodextrins and (1) BE catalyzes the
formation of a-1,6 branches.that bind sugars in the a-1,6 position. In addition, Figure
3.8 shows a superposition of isoamylase (Figure 3.8a) and a-amylase (Figure 3.8b) onto
BE (rmsd ~1A). It can be observed in the ﬁgure that although the positions of the
elements of secondary structure lie in proximity to one another, their relative orientation
is quite different. Also, the loops connecting those secondary structure elements are
different among the structures.

There are marked differences between the loops connecting elements of
secondary structure among members of the a-amylase family of enzymes. These loops
might be responsible for the distinct catalytic properties between the enzymes.
Comparison of the loop structures between members of the family revealed that BE has
shorter loops, presenting a more open cavity for the binding of a bulkier sugar as is the
case of a branched sugar. Overall BE has a more accessible cavity for sugar binding
compared to CGT, isoamylase and a-amylase (Figure 3.9). A model of the OH ,4 cleaved
sugar and the incoming sugar oriented to form the branch point was modeled, as shown in
Figure 3.9a. When the structures of a-amylase, CGT and isoamylase are overlaid onto
BE the modeled sugar oriented to form the branch point run into the extended loops. The
incoming a-1,6 sugar collides with the [37/(17 loop of isoamylase and the 65/66 loops of
CGT and a-amylase. It is important to remember that BE not only binds already

branched sugars but it also requires access to sugars properly oriented for forming the oz-

1.6 links.

77

 

Figure 3.8 a) Superposition of the structure of isoamylase in blue onto BE shown in gold

(5). b) Superposition of the structure of (it-amylase depicted in lavender onto e. coli BE.

78

 

Figure 3.9 The structures of a) a—amylase, b) CGT and c) isoamylase are overlaid onto

BE also showing are the (X-l,4 cleaved sugar and the incoming sugar oriented to form the
branch point. BB is shown in red and a-amylase, CGT and isoamylase in gray. This
mimic was based on the substrate and intermediate bound structures of other members of

the or-amylase family taking into account the unique loop structure of BE.

79

 

Figure 3.10 Comparison of the loops that surround the (oz/B) barrel cavity. BB is shown

in red, isoamylase in lavender, CGT in green and a—amylase in blue.

80

There are six loop structures observed in a-amylase, CGT and isoamylase not
present in BE that block the access of incoming sugars. These loops lie between [38/0t8,
[37/(17, B7/a7a, B2/or2. B3/a3, BS/B6 and a6a/B7. The loop between [35 and B6 in
isoamylase is moved away from the sugar binding channel. A comparison of all these
loops can be seen in Figure 3.10. There is also an extra domain of approximately sixty
residues, named domain B, inserted in the barrel between B3 and a3 (Figure 3.1 1). This
domain is present in a-amylase and CGT but not in BE. Isoamylase depicts a loop
extension consisting of as many residues as a-amylase and CGT, but topologically it can
not be considered a domain because it lacks elements of secondary structure. Although
in the BE structure there are 13 disordered residues in the loop between [33 and 0L3, this
loop is only 40 residues long, not long enough to account for the whole B domain.

Based on the structures of a-amylases and primary structure analysis, seven
conserved residues among the amylotic enzymes were deﬁned. These residues are D335,
H340, R403, D405, E458, H525 and D526 (Figure 3.12). It was established early on that
these conserved residues are involved in catalysis and substrate binding. Upon analysis
of the BE active site and after comparison with other members of the family we noticed
that although D335 is conserved it does not make any contacts with the substrates in any
of the bound structures. We also noticed that Y300 is not only conserved but it also
interacts with the modeled substrate and, as will be discussed later, has a crucial role in
the mechanism of BE. Figure 3.12 shows the conserved residues and their orientation in
the barrel.

The structures of BE, isoamylase, a-amylase and CGT were overlaid and the

conserved residues were compared (Figure 3.13) (3) (4,5). A comparison between

81

 

Figure 3.1l The B domain lies between B3 and a3. a) The loop between B3 and G3 in BB
is shown in red b) A comparison between the B domain of a-amylase in blue, isoamylase

in lavender, CGT in green and BE in red.

82

 

Figure 3.12 a) Residues involved in BE catalysis. b) Position of these residues in the

barrel

83

a)

as W
%l ”>—

H526
H340 b)

D525

E458
2r...» Y’>—
isoamylase g

D405

I (it-amylase

IBE

substrate bound a-am lase
E y H340.
ﬂ CGT

Figure 3.13 a) Superposition of the conserved residues from BE, isoamylase, a—amylase
and CGT not bound to substrate. b) Comparison between the residues from BE and the

ones from (it-amylase apo and substrate bound (4-6).

84

structures not bound to substrate is shown in Figure 3.13a. The position and orientation
of residues R403, H525 and D526 are conserved in all structures. Inspection of residues
E458, D405 and H340 show a drastic motion of their side chain in BE. We also
compared those residues in BE with apo and substrate bound a-amylase. Interestingly,
the position of those three residues does not change in both a-amylase structures after the
substrate binds (Figure 3.13b). The change in position of these residues is only observed
in BE. This may be important for the type of reaction that BE catalyzes. We must await

further studies of substrate bound BE to understand this difference in orientation.

3.3 Residues associated with the GSDIV

Mutations in BE are responsible for the GSDIV, a genetic disease that produces
an inactive BE causing the glycogen to precipitate in the cell (7,8). With few exceptions,
GSDIV is a progressive and lethal disease (9,10). The residues involved in these
mutations are V273, Y306, Y377, 0555 and K546 (e. coli numbering) depicted in Figure
3.14. Table 3.1 presents a summary of the residues in human and e. coli numbering, the
respective mutation and their location in the structure. These residues are conserved in
e. coli with the exception of R524 in humans, which is a glycine in e. coli. None of these
residues are located within the catalytic cavity, but rather are spread throughout the
structure. K546 is located at the beginning of helix (1.7 and an although exposed it makes
a salt bridge with the carbonyl of a proline in the turn of the helix holding the loop in
position. The two tyrosines, Y306 and Y377, hold the loops connecting B2 to a2 and B3
to a3, respectively. The valine is located in or], interacting with residues in a2 and

holding both helices next to each other. In conclusion, these mutations, in addition to the

85

deletions and truncations, are likely to cause the unfolding of the protein, inactivating the

enzyme.

 

Figure 3.14 Residues responsible for causing the GSDIV (7,8) (9,10).

86

Table 3.1 Residues responsible for causing the GSDIV their location and effect (7-10).

 

 

 

 

 

 

 

 

homo sapiens e. coli mutation Location Effect
L224 V273 P or] slow progressive
F257 Y306 L B2/or2 lethal
Y329 Y3 77 S B3/a3 slow progressive
R515 K546 C (17 lethal
R524 G555 N or a6a/B7 lethal
truncation a7a to C-terminal
residues residues deletion ala to (12 lethal
262-331 311-379

 

 

 

 

 

87

 

3.4 Proposed mechanism

Based on the proposed mechanism for CGT, the only member of the a-amylase
family besides BE that catalyzes a transglycosylation reaction, we have been able to
propose a mechanism for BE. The CGT mechanism was deduced from the analysis of
the substrate and intermediate bound X-ray structures (6). The BE structure was
superimposed on both CGT structures by overlaying the conserved catalytic residues
(rmsd ~1.1 A). The position and interactions observed when modeling the substrate and
intermediate presented a possible model for BE catalytic action. The proposed
mechanism for the reaction catalyzed by BE is shown in Figure 3.15. As previously
described, BE performs a transglycosylation reaction in which an a-l ,4 bond is cleaved
and an 0L-1,6 bond is subsequently formed (Figure 3.5d).

Before substrate binding, E458 and D526 are held in position by a hydrogen bond
network between the conserved waters 1017 and 809 and residue Y300 (Figure 3.16a).
Once the substrate binds in the cavity it pushes the water molecules away. The carbonyl
side chain of E458 rotates and is now oriented properly for interaction with the glycosidic
oxygen. The side chain of D526 also rotates to be able to interact with the oxygens in the
sugar. The sugar protein interactions are shown in Figure 3.16b and listed in Table 3.2.
The hydrolysis of the (I-I,4 bond is then initiated by the protonation of the glycosidic
oxygen by E458 acting as the proton donor to form an oxocarbenium ion.

This is followed by the nucleophilic attack of D405 to the C1 of the sugar, forming a
covalent bond in the intermediate formation (Figures 3.15 and 3.16c). This covalent
intermediate has been observed by X-ray crystallography in CGT and by NMR in (1.-

amylase (6) (11). Once the sugar is cleaved it diffuses away and a new polysaccharide

88

 

OH

   

og/

C substrate C First transition state
E458 /C{‘2 ———JQ /c(+2 /

 

 

HO

C
0 o/ \o o
O OH
9 °~
HO HO OH Hof
OH HO
O

0%.! intermediate

CH2

/

 

 

0 OH
OH
HO
HO
OH
Second
CH2 transition state
C
OH O/ \O
0 OH
OH
O
HO
HO 0
OH
H
/ OHHO
O
O§c/
C42 product
Figure 3.15 Proposed mechanism for BE catalysis /

89

Y3 00

H340 O

 

  
    
 
 

Figure 3.16 a) Orientation of catalytic residues before b)
substrate binding. b) Proposed substrate and

c) intermediate interactions by modeling 0f the

substrate and intermediate from CGT. w

 

9O

Table 3.2 Protein interactions with modeled substrate

 

 

 

 

 

 

 

 

 

 

Residue Substrate atom
H340 06
R403 OZ
D405 C l
D405 OS
E458 Glycosidic-O
H525 02
D526 02
D526 03

 

 

 

comes in properly oriented to form the a-l ,6 bond. At this point E45 8, which acted as a
proton donor in the ﬁrst step of the reaction, will now act at as proton acceptor,
deprotonating the hydroxyl group in the C6 of the new incoming sugar. Once this occurs,
the a-l,6 glycosidic bond is formed and the glutamic acid is regenerated.

In the case of isoamylase and or-amylase, where there is not a bond formation, a
water molecule acts as the proton donor hydroxylating the C 1. This brings up the
question of what makes CGT and BE perform a transglycosylation reaction and not just a
hydrolysis with a water molecule acting as the proton donor. A possible explanation for
this could be that in isoamylase and or-amylase, a water molecule is sequestered and
clamped in position. It has been mentioned that residue D526 could be the one to
activate the water molecule that hydroxylates the C1 of the sugar (12). However, a

comparison of the water molecules in the vicinity of D526 and the C l of the sugar,

91

among members of the family, does not provide any conclusive information, as a water
molecule in this region that would only be particular to the hydrolases was not observed.
Another explanation would be the simultaneous binding of the substrate and the sugar
that will act as the proton donor.

In Figure 3.17 the different sugars involved in the reaction have been modeled
onto the BE structure. We base this mimic on the substrate and intermediate bound
structures of other members of the or-amylase family, taking into account the unique loop
structure of BE. We also modeled the position and orientation that the incoming sugar

must be in order to be able to form the a-1,6 branch.

3.5 Electrostatic potential surface

An analysis of the surface charge distribution of BE was performed by the
calculation of the electrostatic potential surface (EPS) (Figure 3.18). The overall surface
of BB is quite electronegative and the cavity formed in the (a/ B) barrel domain is the
most electronegative feature of the surface. This cavity contains four electronegative
residues; D335, D405, E458 and D526 known to be involved in substrate binding and
catalysis. The EPS of a-amylase, CGT and isoamylase was also calculated (Figure 3.19).
It can be observed in Figure 3.19 that all members of the a-amylase family present an
overall electronegatively charged surface. The highly electronegative character of the
(a/ B) barrel domain in all four members of the family indicates that this negatively

charged catalytic cavity is important for sugar-protein interactions.

92

 

Figure 3.17 Proposed mode of action for BE catalysis. a) substrate binding b)

intermediate formation and c) model of the position that the incoming sugar must have to

form the (X-l,6 branch.

93

 

 

Figure 3.18 Electrostatic potential surface picture of BE a) Looking down the barrel and
b) rotated 180°. The EPS calculation corresponds to lOkT/e for the blue color, -10kT/e

for red and an EPS ~ 0 is white, where lOkT ~ 6 kcal/mol.

94

 

 

       

(It-amylase

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Figure 3.19 EPS of members of the a—amylase family of enzymes. The structures are

oriented looking straight into the central barrel domain.

95

3.6 Conclusions

Branching enzyme has a central catalytic (a/ B) domain like all the other members
ofthe or—amylase family. Comparison of the conserved catalytic residues in BE with other

members of the a-amylase family show a different orientation for residues H340, D405
and E458. We believe that this might be important for the type of reaction that BE

catalyzes.

Analysis of the residues responsible for the development of GSDIV revealed that
these mutations, in addition to the deletions and truncations, are likely to cause the

unfolding of the protein, inactivating the enzyme.

When a polysaccharide molecule is modeled in branching enzyme, we observe
that the catalytic residues are oriented properly for the reaction to proceed. Based on
these results we have been able to propose a mechanism for BE. The different sugars
involved in the reaction were modeled in BE’s active site. The determination of the
structure of BE with a substrate bound in the active site will provide the real picture of

the protein sugar interaction.

96

3.7 Literature cited

[\J

10.

ll.

12.

Binderup. K., Mikkelsen, R., and Preiss, J. (2001) Arch. Biochem. Biophys. , in
press

Jespersen, H. M., MacGregor, E. A., Sierks, M. R., and Svensson, B. (1991)
Biochem J 280, 51-55.

Uitdehaag, J. C., van Alebeek, G. J ., van Der Veen, B. A., Dijkhuizen, L., and
Dijkstra, B. W. (2000) Biochemistry 39(26), 7772-7780.

Brzozowski, A. M., and Davies, G. J. (1997) Biochemistry 36(36), 10837-10845.

Katsuya, Y., Mezaki. Y., Kubota, M., and Matsuura, Y. (1998) J Mol Biol 281(5),
885-897.

Uitdehaag, J. C., Mosi, R., Kalk, K. H., van der Veen, B. A., Dijkhuizen, L.,
Withers, S. G., and Dijkstra, B. W. (1999) Nat Struct Biol 6(5), 432-436.

DiMauro, S., and Tsujino, S. (1994) in Myology (A.G., E., and C., F.-A., eds) Vol.
2, pp. 1554-1576.

Bao, Y., Kishnani, P., Wu, J. Y., and Chen, Y. T. (1996) JClin Invest 97(4), 941-
948.

Chen, Y. T., and Burchell, A. (1995) in The metabolic and molecular basis of
inherited diseases (QR, S., A.L., B., W.S., S., and D., V., eds) Vol. 1, pp. 935-
965.

Schroder, J. M., May, R., Shin, Y. S., Sigmund, M., and Nase-Huppmeier, S.
(1993) Acta Neuropathol 85(4), 419-430.

Tao, B. Y., Reilly. P. J., and Robyt, J. F. (1989) Biochim Biophys Acta 995(3),
214-220.

Svensson, B. (1994) Plant M01 Biol 25(2), 141-157.

97

CHAPTER IV: THE THREE DIMENSIONAL STRUCTURE OF ANGIOSTATIN

4.1 Overall structure of angiostatin.

The three dimensional structure of angiostatin represents an important source of
information for the understanding of the recently-born ﬁeld of angiogenesis. Also, this is
the ﬁrst multikringle structure to be solved and the ﬁrst structure of Kg3. The complete
structure of human angiostatin consists of 253 residues that extend from amino acids 81
to 333. In addition, it contains 398 waters and three bicine molecules that form part of

the crystal lattice.

The overall structure of angiostatin can be described as a triangle with dimensions
of 60 x 57 x 45 A and 32 A in depth (Figures 4.1 and 4.2). Each Kg domain contains
three B strands connected by a series of 100ps. The kringle is held together by disulﬁde
links; there are three disulﬁdes per Kg, as shown in Figure 4.2. These disulﬁdes are
formed between C84-C162, C105-C145 and C133-C157 in Kgl; C166-C243, C187-
C226 and C215-C238 in Kg2; and C256-C333, C277-C316 and C305-C328 in Kg3.
There is also an inter-kringle (inter—Kg) disulﬁde between C169 in Kg2 and C297 in Kg3.
These Kg domains are connected to each other by inter-Kg peptides (Figures 413 and
4.2). The short inter-Kg peptide connecting Kgl to Kg2 consists of three glutamates and

the longer inter-Kg peptide between Kg2 and Kg3 contains 12 residues.

98

 

Figure 4.1 Three different representations of the overall structure of angiostatin. (a)

Ribbon picture showing Kgl, orange; Kg2, magenta; Kg3, cyan; inter-kg peptide
between Kgl and Kg2, blue; inter-Kg peptide between Kg2 and Kg3, green; bicines,
green with atoms in atom colors (nitrogen, blue and oxygen, red); intKg disulﬁde,
yellow. LBS side groups also in atom colors. (b) Space ﬁlling view of angiostatin. The
LBS in each of the three kringles is colored gold. All other atoms are red. (c) Stereo

view of the Ca trace.

99

 

Figure 4.2 The disulﬁde links in angiostatin. Angiostatin shown in red with all

disulﬁde bonds in yellow

100

The Kg domains are homologous, not only throughout their amino acid sequence,
but also structurally (Figure 4.3). A superposition of the Kgs gives rmsd values between
0.40 to 0.46 A, and the same is observed between the individual Kgs and the Kgs in
angiostatin (Figure 4.3). The results of the Ca superposition are listed in Table 4.1.
Beside the structural homology between Kgs, this also shows that the structures of
individual Kgs are good representations of the Kgs found in multi-Kg structures like

angiostatin.

In angiostatin, we observe that Kg2 and Kg3 are oriented with their LBSs facing
each other, forming a 20 A cavity. We believe that this cavity may be an important
binding site in angiostatin, which will be elaborated on later. Their LBS’s are related by
a 112° rotation about an axis between them, coupled with a 1.6 A translation (Figure 4.1).
The anionic centers between Kg2 and Kg3 are about 13.5 A apart, while the cationic ones
are separated further at 25 A. The corresponding transformations between the other
kringles of angiostatin are: Kgl to Kg2: 136°, 0.9 A; Kgl to Kg3: 163°, 1.2 A. The LBS

of Kgl is facing towards the back of the molecule as shown in Figure 4.1.

4.2 The electrostatic surface of angiostatin

An analysis of the surface charge distribution of angiostatin was performed by the

calculation of the electrostatic potential surface (EPS) (Figure 4.4). The EPS calculation

101

   

I Angiostatin Kgl
D Angiostatin Kg2
- Angiostatin Kg3
individual Kgl

individual Kg2

Figure 4.3 Superposition of various Kgs from plasminogen. This ﬁgure includes Kgl ,

Kg2 and Kg3 from angiostatin and the individual Kgl and Kg2 (1,2).

102

Table 4.1 Rmsd values of the superposition of the Ca positions of individual Kgs and the

Kgs in angiostatin

 

 

 

 

 

 

 

Kgl- Kg2- Kg3- Kgl- Kg2-
angiostatin angiostatin angiostatin individual individual
Kgl-angiostatin - 0.45 0.40 0.41 0.37
Kg2-angiostatin 0.45 - 0.37 0.37 0.46
Kg3-angiostatin 0.40 0.37 - 0.40 0.41
Kgl -individual 0.41 0.37 0.40 - 0.52
Kg2-individual 0.37 0.46 0.41 0.52 -

 

 

 

 

 

 

corresponds to 10kT/e for the blue color, -10kT/e for red and an EPS ~ 0 for white, where
10kT ~ 6 kcal/mol. The EPS shows a neutral overall structure with prominent bipolar
character in the LBSs of Kgl and Kg2. The most outstanding electronic feature is the
highly electropositive LBS of Kg3, compared to the other Kgs. In Figure 4.4b, the EPS
was rotated 1800 to display the back of the structure showing the LBS of Kgl. Inspection
of this side of the EPS reveals the dipolar character of the Kgl LBS, an electronegative
charge cluster corresponding to the Kgl-Kg2 linker (with three consecutive glutamates)
and the non-polar faces of the Kg2 and Kg3. In addition, there is an electropositive
crescent created by R223 and R242 of Kg2 (Figure 4.4b). The central cavity formed by
all three Kgs seems to form a non-polar binding surface that might be important for

interacting with other ligands.

103

 

a)

4}"

2 LBS Kg3 LBS

 

Figure 4.4 Electrostatic potential surface (EPS) of angiostatin with the bicines omitted.

a) Same orientation as in Figure 4.1. b) Rotated 180° to show Kgl ’8 LBS

4.3 Ligand speciﬁcity of the kringle LBS

The LBS is deﬁned by residues 115-119, 137-139, 144-146, and 153-155 in Kgl;
residues 201-207, 219-221, 224-228, and 234-237 in Kg2, and residues 287-291, 308-
312, 314-318, and 324-327 in Kg3. In Figure 4.1b, the residues in the LBS of each Kg
are depicted in yellow, while all other residues are shown in red and in Figure 4.2 the
LBS of each Kg is marked by magenta bars in the sequence alignment. Each LBS in
angiostatin has a bound bicine molecule; where bicine was the buffer used for
crystallization. The three bicines were modeled in the electron density and in the reﬁned

structure.

Although bicine is not a carboxylate lysyl analog, it does have a carboxyl end
(Figure 4.5). The carboxyl group of the bicine interacts with the cationic center in the
LBS in the same way EACA does, as shown in the Kgl/EACA structure (Figure 4.6a)
(1). The bound bicine molecules provided important information that helped in
understanding the various binding afﬁnities of each LBS for EACA, a mimic for the
amino terminal lysine in ﬁbrin. Among the three Kgs, Kgl has the highest binding
afﬁnity for EACA with a KD of 15.5 uM, while Kg2 has a KD of only 401 uM. Kg3
shows no afﬁnity for EACA (3-5). The binding afﬁnity of Kgl for bicine was measured

to be in the low millimolar range (6).

In the Kgl/EACA structure, R1 17 and R153 stabilize the carboxylate and D137
and D139 stabilize the ammonium end of EACA, while the hydrophobic residues F118,
W144, Y154, and Y156 stabilize the ﬁve carbon methene chain by hydrophobic

interactions (Figure 4.63). Analysis of the interactions between angiostatin’s Kgl and

105

C-terminus

 

 

 

 

 

 

 

 

a) R T i
- T—CH—C—LN (IzH—c—O'
.. l i”
amino ac1d chain CH2
CH2
CH2
NH;
carboxylate lysine residue
O
b) H
c) H
CH2 THz—c—o-
CH2 CH2
CH2 I I
CH2 H20
CH2
”0 OH
NH3+

Figure 4.6 a) Carboxylate lysine residue. b) The carboxylate lysine analog,

e-aminocaproic acid. c) Bicine

106

 

b)
Y156 W235
Y200
ﬁWZZS <
E221 D219
R220

I Kg2 from angiostatin
I Kg2 from KgZNEKBO
U Bicine

 

Figure 4.6 Interaction of the three angiostatin LBS's with bicine. All three depictions are
in the same orientation. Hydrogen bonds and salt bridge contacts are shown by dotted
lines. Residues from angiostatin are shown in green with atom colors, bicines are shown
in yellow. (a) Comparison of the binding of Kgl to bicine and EACA. EACA is shown in
lavender (l). (b) Angiostatin Kg2 and Kg2 from the Kg2NEK-30 structure are overlaid.
Residues from the VEK—30 peptide are not shown for clarity. (2) (c) The angiostatin Kg3

LBS with bound bicine.

107

bicine shows that the catalytic center formed by R] 15 and R153 stabilizes the
carboxylate head of the bicine by ionic interactions (Figure 4.6a). This carboxylate head
of the bicine is oriented like EACA in the Kgl/EACA, as shown in Figure 4.6a (l). The

hydroxyl tails of bicine are stabilized by interactions with W144, Y154, F118, and D137.

The bicine molecule in Kg2 is similarly oriented to the bicine in Kgl with R234
forming a salt bridge with the carboxylate end of bicine (Figure 4.6b). Nonetheless there
is a difference in conformation between the conserved aspartates of Kgl (D137) and Kg2
(D219). In Kgl , D137 is oriented towards the LBS, ready to make an ionic interaction
with the ammonium group of EACA. However, the D219 in Kg2 interacts with the non-
conserved R220 (an aspargine in Kgl and a glycine in Kg3) rotating its side chain out of
the LBS. The position of the D219 side chain is too far away to interact with the
ammonium group of EACA. This leaves an incomplete LBS with only one functional
residue, E221, in the anionic center and a cationic center with only one residue (R234) as
well. We believe this to be the reason why Kg2 has a lower EACA afﬁnity compared to
Kgl. The residues W225, W235 and E221 interact with the bis ethyl hydroxy end of the

bicine molecule (Figure 4.6b).

One of the most interesting results obtained from the structure of angiostatin
comes from the analysis of the LBS in Kg3. The limitation of not having a structure of
Kg3 made it impossible to explain why Kg3 is the only Kg of plasminogen with no
afﬁnity for EACA (5). Even though Kg3 has no afﬁnity for EACA, its LBS is occupied
by a bicine molecule. The carboxylate part of the bicine interacts ionically with the

cationic center formed by R324 and R290. The LBS of Kg3 is highly electropositive

108

with a lysine substituting for one of the aspartates in the anionic center. The positively
charged amino terminal tail of EACA will cause electrostatic repulsion between EACA
and the highly electropositive LBS of Kg3. Also, inspection of the LBS shows that this
lysine (K311) ﬁlls half of the. LBS, preventing the binding of long molecules like EACA.
This can be observed in Figure 4.7a where an EACA molecule was modeled onto the
Kg3 LBS by overlaying the structure of Kgl onto Kg3. There is a salt bridge between
K31 l and D309 that holds K311 in position. The bicine binds in a totally different
conformation compared to the other bicines of angiostatin to avoid steric clashes (Figure
4.7). The bicine in Kg3 is rotated 90° around its CB atom as shown in Figure 4.7b. The
LBS in Kg3 is reduced in size, quite positively charged and without bipolar character.

This explains the non-afﬁnity of Kg3 for EACA.

Studies performed on the K311D mutant of Kg3 shows some afﬁnity for EACA
and other small molecule C-terminal lysine analogs, indicating that K311 inhibits binding
of the molecules (4,5). These observations suggest that the Kg3 LBS is ideally suited to
bind only carboxylate-containing ligands such as aspartate or glutamate and not extended
bipolar ligands such as EACA or C-terminal lysine residues. This represents a new
binding mode, speciﬁc to Kg3 like kringles, that results from the highly electropositive

nature of this LBS.

It is not clear what role the LBS plays in the angiogenesis inhibitory activity of
angiostatin. Previous studies show no correlation between the antiangiogenic activity of

individual Kgs and their EACA binding afﬁnity. For example, Kg2 has a higher afﬁnity

109

R324

W325

   

b)

 

Figure 4.7 a) EACA molecule was modeled onto the Kg3 LBS by overlaying the
structures of Kgl onto Kg3. b) Different conformation of the bicine in Kgl of angiostatin

depicted in red and the bicine in Kg3 shown in blue.

110

for EACA than Kg3, but its inhibitory activity is lower than Kg3 (7). Other studies
performed in the angiostatin C169S/C297S double mutant resulted in the loss of EACA
binding by Kg2 without altering its antiangiogenic activity (8). It is important to mention
that the lysine afﬁnity monitored is that of EACA, which is a good model for carboxy
terminal lysines. It may be that angiostatin binds internal lysine residues, or it could even

have some other binding speciﬁcity thus far unknown.

4.4 Angiostatin binding to protein domains

It has been determined that Kgs not only bind six carbon zwitterions such as
lysine and EACA, but also protein domains (2,9,10). The recently determined structure
of Kg2, bound to a peptide sequence (VEK30) of the Streptococcal surface protein PAM,
presented a model for protein binding at the surface of bacteria (2). Since angiostatin
offers a more realistic model of the physiological target of the Streptococcal surface
protein, we modeled the VEK30 peptide onto angiostatin. In order to do this, we overlaid
the Kg2 from the Kg2/VEK30 structure onto angiostatin’s Kg2 (rmsd = 0.41) (Figure
4.8). The a-helix of the VEK30 peptide is accommodated without collisions in the 20 A
cavity between Kgs with enough space left to ﬁt another VEK30 helix. Kg2 from the
Kg2/VEK30 structure was also superimposed onto angiostatin’s Kg3 (rmsd 0.40),
showing that both helices ﬁt well, with Kg3 interacting with VEK30 similarly to Kg2.
The VEK30 peptide has an arginine and a glutamate separated by one turn of a helix.
This resembles a carboxylate lysine with a positive (R101) and a negative (E104) end that

interacts with Kg2’s E221 and R234, respectively. Interestingly, upon VEK30

lll

 

Figure 4.8 A Ribbons depiction of the modeled angiostatinNEK30 complex. The Kg2 of
the Kg2/VEK30 complex was overlaid on angiostatin Kg2 (2). Angiostatin is colored

green while VEK30 is colored lavender. Side groups are labeled appropriately.

112

 

Figure 4.9 Endostatin modeled onto the angiostatin (11). This was done by overlaying
the helices of endostatin and VEK30. Endostatin is colored purple and angiostatin green.
b) Close view of the section of angiostatin that harbors the residues that may be involved

in endostatin binding.

113

binding, the salt bridge between R220 and D219 in Kg2 of angiostatin would be
disrupted by the helix. This would force D219 to move into the LBS where it could
interact with VEK30 K98. The R220 could also move to make a hydrogen bond with

VEK30 Q95.

The ability of angiostatin to bind a protein domain led us to explore other possible
molecules relevant to angiostatin’s mode of action. The X-ray structure of the
angiogenesis inhibitor endostatin reveals that it contains an a-helix with the RGAD
sequence (11). R158 and D161 form a pseudo lysyl site, similar to the one in the VEK30
peptide. Endostatin was modeled in the cavity between Kg2 and Kg3 (Figure 4.9).
Endostatin ﬁlls the cavity with very few collisions. Moreover, inspection of Kg3’s LBS
shows that endostatin’s E272 interacts with Kg3’s R290 and R324. Validating the
previously suggested observation that Kg3 is suited to bind short carboxylate ligands.
Endostatin was modeled in the cavity between Kg2 and Kg3 (Figure 4.9). Endostatin
ﬁlls the cavity with very few collisions. Although there is no biochemical data indicating
that these two molecules bind, there have been observations of an increase in tumor

reduction when both inhibitors are given in combination to cancer patients (12).

Marneros and Olsen proposed a mechanism to explain the role of endostatin as an
angiogenesis inhibitor (1 3). This mechanism is based on the binding afﬁnity of the
endostatin domain of collagen XVIII. Collagen XVIII is involved in the activation of cell
migration by interactions with extracellular components through its endostatin domain. It
is proposed in this model that endostatin binding competes for the binding of matrix

components, inhibiting endothelial cell migration and consequently angiogenesis. We

114

propose that in the same way, angiostatin can interact with the endostatin domain of
collagen XVIII by the interactions previously proposed, and this could possibly be the

inhibitory mechanism of angiostatin.

Recent studies indicate that angiostatin binds the B 3 subunit of the endothelial cell
surface receptor, OLVB3 (14). The interaction is inhibited by EACA, but only at
concentrations high enough to occupy Kg2. This reinforces the importance of the cavity
between Kg2 and Kg3 for interaction with protein domains. We have identiﬁed a
sequence KKVEE in an exposed or-helix of the B3 subunit (Figure 4.10). We believe that
this sequence can represent a pseudo-lysyl site similar to the one in the VEK30 peptide.
However, the Kg2-Kg3 cavity must open further for the 0th3 integrin to ﬁt in the cavity.
It is believed that the binding of angiostatin to OM33 might perturb a critical signaling
pathway for endothelial cells, therefore inhibiting angiogenesis. Some of these results
reinforce the importance of the Kg2-Kg3 cavity in the protein domain binding. This

could mean that the angiostatin inhibitory activity resides in this cavity and not in the

LBS.

115

 

 

Figure 4.10 a) Structure of 0!.ng integrin (15). The at, subunit is shown in red and the B3

subunit in green; also shown are the residues that could be involved in angiostatin
binding. b) Close view of the section of B3 that harbors the residues involved in possible

angiostatin binding.

116

4.5 The inter-kringle disulﬁde bond

Studies performed on angiostatin as an inhibitor of endothelial cell proliferation
revealed that the Kg2-Kg3 fragment exhibits inhibitory activity, similar to Kg2 alone (7).
Also, an enhancement in inhibitory activity is observed with individual Kg2 and Kg3
versus the Kg2-Kg3 fragment (7). It was initially suggested that it is necessary to open
the Kg2-Kg3 disulﬁde bond in order to obtain maximum antiangiogenic activity. With
this in mind, the angiostatin double mutant C169S/C297S, which eliminates the inter-Kg
disulﬁde bond, was constructed (8). This mutant had little effect in antiangiogenesis
activity. Analysis of the interactions between Kg2 and Kg3 in angiostatin revealed that
there are numerous contacts between Kg2 and Kg3 and that disruption of the disulﬁde
bond should not alter the structure of angiostatin. Appendix 4.1 shows a complete list of
interactions between Kgs and their inter-Kg peptides. Some of the Kg2/Kg3 interactions
include a salt bridge between E163 and H168, four hydrogen bonds between the inter-Kg
peptide between Kg2 and Kg3 and Kg2 and Kg3. These are only some of the 102
contacts between all three Kgs and inter-Kg peptides (Table 4.2). Various interactions
within this structure indicate that angiostatin forms a molecular entity much like a single

domain protein that might function cooperatively.

4.6 Conclusions

The numerous interactions between the Kgs in angiostatin produce a unique

domain that harbors a recognition site important for angiogenesis inhibition. This

117

recognition site could be the cofacial orientation of the LBSs in Kg2 and Kg3, which
forms a cavity 20 A wide. The models of angiostatin with VEK30 and endostatin
illustrate how the interaction may occur. This could mean that the inhibitory activity of
angiostatin does not reside in the individual LBSs, but in this cavity, with the Kg domains
working in concert. Also, the orientations of all three LBSs in angiostatin demonstrate

that the LBSs in this multi-Kg structure remain functionally viable.

The structure of angiostatin has also provided an explanation for the inability of
Kg3 to bind EACA. The structure showed that the LBS of Kg3 is reduced in size, quite
positively charged and without bipolar character. Additionally, the low afﬁnity of Kg2

for EACA is explained by the rotation of the D219 side chain outside the LBS.

Table 4.2 Summary of the Kg-Kg interactions and inter-Kg peptide Kg interaction of

angiostatin. The interactions are determined with a cutoff distance of <40 A.

 

 

 

Interactions between Number of interactions
Kg2/Kg3 28
Kg2/inter-KgCKg1-Kg2) peptide 19
Kg2/inter-Kg(Kg2-Kg3) peptide 12
Kg3/inter-Kg(Kg2-Kg3) peptide 43

 

 

 

118

4.7 Literature cited

Ix)

10.

ll.

l2.

l3.

l4.

Mathews, I., Vanderhoff-Hanaver, P., Castellino, F. J ., and Tulinsky, A. (1996)
Biochemistry 35(8), 2567-2576.

Rios-Steiner, J. L., Schenone, M., Mochalkin, I., Tulinsky, A., and Castellino, F.
J. (2001) J Mol Biol 308(4), 705-719.

Chang, Y., Mochalkin, I., McCance, S. G., Cheng, B., Tulinsky, A., and
Castellino, F. J. (1998) Biochemistry 37(10), 3258-3271.

Burgin, J., and Schaller, J. (1999) Cell Mol Life Sci 55(1), 135-141.

Marti, D., Schaller, J ., Ochensberger, B., and Rickli, E. E. (1994) Eur J Biochem
219(1-2), 455-462.

C astellino, F. J. (2001), University of Notre Dame, personal communication

Cao, Y., Ji, R. W., Davidson, D., Schaller, J., Marti, D., Sohndel, S., McCance, S.
G., O'Reilly, M. S., Llinas, M., and Folkman, J. (1996) JBiol Chem 271(46),
29461 -29467.

Lee, H., Kim, H. K., Lee, J. H., You, W. K., Chung, S. I., Chang, S. 1., Park, M.
H., Hong, Y. K., and Joe, Y. A. (2000) Arch Biochem Biophys 375(2), 359-363.

Moser, T. L., Stack, M. S., Asplin, 1., Enghild, J. J., Hojrup, P., Everitt, L.,
Hubchak, S., Schnaper, H. W., and Pizzo, S. V. (1999) Proc Nat Acad Sci USA
96(6), 2811-2816.

Troyanovsky, B., Levchenko, T., Mansson, G., Matvijenko, O., and Holmgren, L.
(2001) J Cell Biol 152(6), 1247-1254.

Hohenester, E., Sasaki, T., Olsen, B. R., and Timpl, R. (1998) Embo J 17(6),
1656-1664.

Yokoyama, Y., Dhanabal, M., Grifﬁoen, A. W., Sukhatrne, V. P., and
Ramakrishnan, S. (2000) Cancer Research 60(8), 2190-2196.

Marneros, A. G., and Olsen, B. R. (2001) Matrix Biol 20(5-6), 337-345.

Tarui, T., Miles, L. a., and Takada, Y. (2001) J. Mol. Biol. 276(43), 39562-39568.

119

15. Xiong, J. P., Stehle, T., Diefenbach, B., Zhang, R., Dunker, R., Scott, D. L.,
Joachimiak, A., Goodman, S. L., and Amaout, M. A. (2001) Science 294(5541),
339-345.

120

APPENDIX

121

Appendix 4.1 Kg/Kg and Kg/intKg peptide interactions of angiostatin. The interactions

are determined with a cutoff distance of 4.0 A.

kgl/intKg(Kgl—K92)

# name atom # name atom distance (A)
161 GLU O 163 GLU N 3.60
162 CYS N 163 GLU N 3.61
162 CYS CA 163 GLU N 2.43
162 CYS CA 163 GLU CA 3.84
162 CYS C 163 GLU N 1.34
162 CYS C 163 GLU CA 2.47
162 CYS C 163 GLU CB 3.58
162 CYS C 163 GLU CG 3.82
162 CYS C 163 GLU C 3.44
162 CYS C 164 GLU N 3.82
162 CYS O 163 GLU N 2.27
162 CYS O 163 GLU CA 2.84
162 CYS O 163 GLU C 3.80
162 CYS O 164 GLU N 3.85
162 CYS CB 163 GLU N 3.24
162 CYS CB 164 GLU OE2 3.83

ng/intKg(Kgl-K92) .
# name atom # name atom distance (A)

166 CYS N 164 GLU C 3.41
166 CYS N 164 GLU O 3.61
166 CYS N 165 GLU N 2.84
166 CYS N 165 GLU CA 2.40
166 CYS N 165 GLU CB 3.12
166 CYS N 165 GLU C 1.32
166 CYS N 165 GLU O 2.24
166 CYS CA 165 GLU CA 3.78
166 CYS CA 165 GLU C 2.43
166 CYS CA 165 GLU O 2.78
166 CYS C 165 GLU C 3.59
166 CYS O 163 GLU CA 3.86
166 CYS O 163 GLU CB 3.47
166 CYS O 163 GLU C 3.71
166 CYS O 163 GLU O 3.95
166 CYS O 165 GLU C 3.80
166 CYS CB 165 GLU C 3.28
166 CYS CB 165 GLU O 3.52
167 MET CA 163 GLU OEl 3.53
167 MET CB 163 GLU OE1 3.91
167 MET C 163 GLU OEl 3.80
168 HIS N 163 GLU CD 3.75
168 HIS N 163 GLU OEl 3.10
168 HIS CB 163 GLU OE2 3.73
168 HIS ND1 163 GLU CD 3.96
168 HIS NDl 163 GLU OE2 3.77

122

173
173
173
173
173
174
174
174
174
174
175
175
175
175
175
175
176
176
176

ASN
ASN
ASN
ASN
ASN
TYR
TYR
TYR
TYR
TYR
ASP
ASP
ASP
ASP
ASP
ASP
GLY
GLY
GLY

Z<DC)O(3()

CD1
CEl
C
N
CA
CB
C

O

O
N
CA
CA

ng/intKg(KgZ-K93)
name atom

#

168
168
168
168
168
168
168
168
168
168
242
242
243
243
243
243
243
243
243
243
243
243
243
243
243
243

ng/kgB
#
169

169
169

HIS
HIS
HIS
HIS
HIS
HIS
HIS
HIS
HIS
HIS
ARG
ARG
CYS
CYS
CYS
CYS
CYS
CYS
CYS
CYS
CYS
CYS
CYS
CYS
CYS
CYS

CB
CG
CD2
NDl
CEl
NE2
NE2
NE2
NE2
NE2
C
O
N
CA

0
P

F)OCDC>O(D()0(3()O

name atom

CYS
CYS
CYS

N
CA
CA

163
163
163
163
163
163
163
163
164
163
163
163
163
163
164
164
163
164
164

248
248
247
248
246
245
245
246
246
247
244
244
244
244
244
244
244
244
244
245
244
244
244
245
245
244

297
296
297

GLU
GLU
GLU
GLU
GLU
GLU
GLU
GLU
GLU
GLU
GLU
GLU
GLU
GLU
GLU
GLU
GLU
GLU
GLU

name
PRO
PRO
PRO
PRO
PRO
THR
THR
PRO
PRO
PRO
THR
THR
THR
THR
THR
THR
THR
THR
THR
THR
THR
THR
THR
THR
THR
THR

name
CYS
PRO
CYS

123

OE2
OE1

OE1

atom

C)O(3()Q(3
B Cjﬁ'Utj

P

y

UJW

2505363953250(16)Z(725253250(D()O

atom

CG
SG

.71
.67
.76
.75
.64
.77
.89
.69
.78
.91
.85
.49
.81
.44
.89
.17
.32
.95
.61

wwwwwwwwmwwwwwwwwww

distance (A)
.93
.63
.56
.74
.82
.79
.70
.83
.24
.73
.91
.51
.54
.42
.80
.34
.44
.69
.28
.49
.27
.78
.42
.86
.95
.29

wwwwmmwwwMI—Iwwwwwwwwwwwwwww

distance (A)
3.51
3.91
3.29

 

169
169
169
169
169
169
170
222
222
222
222
222
222
242
242
242

CYS
CYS
CYS
CYS
CYS
CYS
SER
LEU
LEU
LEU
LEU
LEU
LEU
ARG
ARG
ARG

O
O
CB
CB
SG
SG
CB
CD1
CD2
CD2
CD2
CD2
CD2
CZ
NHl
NH2

Kg3/intKg(Kg2-Kg3)
name atom

#

256
256
256
256
256
256
256
256
256
256
256
256
256
256
256
256
256
256
256
256
258
258
258
258
258
258
263
263
263
263
263
263
263
263
263

CYS
CYS
CYS
CYS
CYS
CYS
CYS
CYS
CYS
CYS
CYS
CYS
CYS
CYS
CYS
CYS
CYS
CYS
CYS
CYS
LYS
LYS
LYS
LYS
LYS
LYS
ASN
ASN
ASN
ASN
ASN
ASN
ASN
ASN
ASN

O
CB
CB
SG
CG
CD
CD
CD
CE
CE
CB
CG
ODl
ND2
ND2
ND2
C
O
O

296
296
297
297
297
297
294
289
289
289
289
294
295
293
293
293

254
254
254
255
255
255
255
255
255
255
255
255
255
254
254
254
255
255
255
254
254
254
254
254
254
254
252
254
254
252
254
254
254
253
253

PRO
PRO
CYS
CYS
CYS
CYS
ASN
GLU
GLU
GLU
GLU
ASN
PHE
GLU
GLU
GLU

name
TYR
TYR
TYR
GLN
GLN
GLN
GLN
GLN
GLN
GLN
GLN
GLN
GLN
TYR
TYR
TYR
GLN
GLN
GLN
TYR
TYR
TYR
TYR
TYR
TYR
TYR
PRO
TYR
TYR
PRO
TYR
TYR
TYR
THR
THR

124

CD
CG
CB
SG
CB
SG

OE1
CB
OE2

ND2
CZ

atom

002000
mu, m

>

0000090000200

000
Dow
k)N

CE2
CZ
CE2
CZ

CD2
CD2

CD2
CE2

CA
CB

wwwwwwwwwwmwwwww

distance
.80
.10
.49
.55
.43
.56
.34
.26
.99
.83
.47
.84
.58
.29
.90
.99
.83
.41
.72
.92
.88
.38
.18
.78
.77
.83
.45
.75
.92
.97
.60
.55
.96
.10
.15

towwwwwwwwwwwwwwwwwwwwwwwmwwwHwwwwww

.91
.72
.84
.10
.01
.05
.93
.96
.69
.71
.97
.84
.94
.66
.71
.94

(A)

263
263
263
263
264
264
264
264
264
264
264
264
265
265
265
265
265
265
265
265
265
265
265
266
266
266
297
333

ASN
ASN
ASN
ASN
TYR
TYR
TYR
TYR
TYR
TYR
TYR
TYR
ARG
ARG
ARG
ARG
ARG
ARG
ARG
ARG
ARG
ARG
ARG
GLY
GLY
GLY
CYS
CYS

0000

253
253
254
254
254
254
255
254
255

255

255
254
254
254
254
253
253
253
254
255
255
255
255
254
255
255
247
254

THR
THR
TYR
TYR
TYR
TYR
GLN
TYR
GLN
GLN
GLN
TYR
TYR
TYR
TYR
THR
THR
THR
TYR
GLN
GLN
GLN
GLN
TYR
GLN
GLN
PRO
TYR

125

D
[O

000000900000200

0000
now
NH

NE2
CD

OE1
NE2

NE2
NE2
CB
CB

wwwwwwwwwwwwwwmawwwwwwwwwwww

.49
.46
.88
.89
.44
.15
.93
.78
.89
.52
.52
.60
.00
.83
.81
.38
.77
.47
.93
.96
.97
.82
.52
.66
.88
.27
.85
.98

IIIIIIIIIIIIIIIIIIIIIIIIIIIIII

111111111lllllllll111111111111111111111111

3 1293 02334 23