Date

0-7639

   
  

   
 
  

L 13 RA R Y
M 1' ' ‘- mn State

Li. i.;~u’TCISIty

This is to certify that the

thesis entitled
AN X-RAY CRYSTALLOGRAPHIC STUDY OF THE CHEMICAL
DENATURATI ON 0F a-CHYMOTRYPSI N

presented by

LYNDON STANLEY HIBBARD

has been accepted towards fulfillment
of the requirements for

 

PH.D. degree in CHEMISTRY

a .133.“ L7

I
Major professor

 

aufiu“? ‘7‘ qu7-
U

AN X-RAY CRYSTALLOGRAPHIC STUDY OF THE CHEMICAL
DENATURATION OF a—CHYMOTRYPSIN

By

Lyndon Stanley Hibbard

A DISSERTATION

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

DOCTOR OF PHILOSOPHY

g .

Department of Chemistry

1977

U?

3f

ABSTRACT
AN X-RAY CRYSTALLOGRAPHIC STUDY OF THE CHEMICAL
DENATURATION 0F a-CHYMOTRYPSIN
. By
Lyndon Stanley Hibbard

Protein denaturation is a process in which the three-
dimensional structure of a protein molecule is changed
from that of the native conformation to a more disordered
conformation. It is an aspect of the phenomenon of protein
folding and has been studied for a number of proteins in
solution to determine the mechanism by which the polypeptide
chains of a protein are organized to form a particular
tertiary structure. Urea and guanidine HCl (GdnHCl) are
particularly powerful protein denaturants and it is the
effects of these compounds on the three-dimensional structure
of the enzyme a-chymotrypsin (CHT) that form the principal
results reported in this dissertation.

Crystals of native CHT in solutions 75% saturated in
ammonium sulfate at pH 3.6 were eXposed to gradually-
increasing concentrations of the denaturants and 29-scans
of the axial reflections were recorded at various
concentration increments and compared with those of the
native protein crystals. When significant changes were
observed, three-dimensional X-ray intensity data were

measured to 2.8 A. resolution. and ”best” difference

Lyndon Stanley Hibbard

electron density maps1 were calculated using the coefficients

where AF is the difference between the derivative and the
native structure amplitudes, dN is the native phase angle.
and m is a function of error in aN (figure of merit) which
is included to weight the terms of the difference Fourier.
The difference Fourier maps for two derivatives,
2.0 M, GdnHCl and 3.0 M, urea, were very complicated
revealing large numbers of changes over many parts of the
CHT molecule. A graphical representation of the difference
peaks, the difference diagonal plot (DDP).2 was used to
characterize the locations of, and the extent of the most
significant changes in the derivative maps with respect to
the CHT amino acid sequence. Several GdnHCl difference
map peaks near the aromatic substrate binding site created
a distinctive array of DDP features similar to those
observed in the DDP's of inhibitor-CRT complexes.3 An
examination of the urea DDP revealed several groups of
changes which occurred in the interior of the CRT molecule.
The peaks which appeared in the GdnHCl difference
map were located exclusively on the surface of the protein.
A number of difference peaks appeared in the dimer interface
region near the substrate binding site and represent
changes in the local solvent structure with the possible

binding of a guanidinium ion (Gdn+) near sulfate ions

Lyndon Stanley Hibbard
hydrogen bonded to the Ser 195 hydroxyl groups of both CHT
molecules. A Gdn+ was also observed in the uranyl binding
site regiong’s forming an intermolecular salt bridge
between two CHT molecules.

The urea difference displayed a complex picture of
changes occurring at sites in the nonpolar interior of the
CHT molecule as well as on the surface. The nature and
disposition of many of the peaks strongly suggests that
urea molecules penetrated the interior of the protein to
interact with a number of aliphatic side chains. Disruptions
of the native structure were observed in the A-chain/B-chain
contact region, in a segment of B-domain B-sheet near the
tryptOphan cluster Trp 27, Trp 29, and Trp 237, and in a
segment of A-domain B-sheet near the aromatic cluster

Trp 51, Phe 89, and Trp 237°

1. D. M. Blow and F. H. c. Crick, Acta Cryst” 12;. 79L» (1959).
2. M. N. Liebman, Ph.D. Dissertation, Michigan State
University, 1977.
3. A. Tulinsky, I. Mavridis, and R. Mann. J. Biol. Chem.,
in the press.
h. A. Tulinsky, N. V. Mani, C. N. Morimoto, and
R. L. Vandlen, Acta Cryst., egg. 1309 (1973).
5. R. L. Vandlen and A. Tulinsky, Biochemistgy, lg, 4193
(1973).

To my wife,

Susan

Ho
Ho

Acknowledgements

With deep appreciation and gratitude the author wishes
to thank Professor Alexander Tulinsky for his guidance,
support, and encouragement.

The author is grateful to Dr. N. V. Raghavan and to
Dr. Stephen R. Ernst for many stimulating discussions and
helpful contributions.

To Dr. M. N. Liebman for his contribution of the
difference diagonal plot and for many useful discussions,
the author is grateful.

To Mr. Lawrence Weber for his contributions to the
difference diagonal plot calculations, the author extends
his sincere thanks.

For their financial support, the author thanks the
Molecular Biology Section of the National Science Foundation
and the Department of Chemistry, Michigan State University.

The author wishes to thank his wife, Susan, for her
constant support and encouragement through four difficult

but rewarding years.

iii

II.

III.

IV.

TABLE OF CONTENTS

Denaturation and the Mechanism of Protein Folding:
An Introduction

The Crystal and Molecular Structure of
Chymotrypsin

1. The Activation and Primary Structure

of Chymotrypsin
2. The Crystal Structure of Chymotrypsin
3. The Molecular Structure of Chymotrypsin

Secondary Structure
Folding Domains and the
Distance Diagonal Plot

(
(
(
(

O‘ m
VV vv

c Nonpolar Interdomain Region
d The Dimer Interface Region
Experimental

The Crystal Soaking Experiments

The X-Ray Intensity Measurements

The Reduction of Measured Intensities
to Structure Amplitudes

. Scaling the Reduced Structure
Amplitudes

4? WNH

The Difference Fourier Synthesis

1. The Difference Fourier Synthesis

2. Errors and Peak Heights in the
Difference Fourier Synthesis

The 0.01 M, Trichlbroacetate Derivative

1. The 0.01 M, Trichloroacetate Difference
Electron Density

2. The 0.01 M, Trichloroacetate Difference
Diagonal Plot

The 2.0 M, Guanidine HCl Derivative

l. The Difference Fourier Map
2. The Difference Diagonal Plot

iv

11
13
l3
lh
19
21
24

2h
30

3Q
38
no
40
43
48

48
51
56

56
57

Table of Contents (Continued)

3. Changes in the Dimer Interface Region 62
4. Changes on the Dimer Surface 70
VII. The 3.0 M, Urea Derivative 75
1. The Difference Electron Density Map 75
2. The Difference Diagonal Plot 76
3. Changes on the Dimer Surface 8“
4. Changes in the Dimer Interface Region 89
5. Changes in the A—Chain/B—Chain Contact
Region 95
6. Changes in the Nonpolar Interdomain
Region 99
VIII. Summary of Results and Discussion 105
l. The 2.0 M, Guanidine HCl Derivative 105
2. The 3.0 M, Urea Derivative lOJ

LIST OF TABLES

Table Page
1 Summary of X-ray Intensity Data Collection,

Reduction, and Scaling Parameters 36
2 The Most Significant Features in the 0.01 M,

TCA Difference Map 49
3 The 2.0 M. GdnHCl Derivative - the Most

Significant Difference Peaks in the Dimer

Interface Region 64
4 The 2.0 M, GdnHCl Derivative - the Most

Significant Difference Peaks on the Surface

of the CHT Dimer 7l
5 The 3.0 M, Urea Derivative - the Most

Significant Difference Peaks on the Surface

of the CHT Dimer 86
6 The 3.0 M, Urea Derivative — the Most

Significant Difference Peaks in the Dimer

Interface Region 90
7 The 3.0 M, Urea Derivative - the Most-

Significant Difference Peaks in the

A-Chain/B-Chain Contact Region 96
8 The 3.0 M Urea Derivative - the Most Significant

Difference Peaks in the Nonpolar Interdomain
Region 101

vi

LIST OF FIGURES

Figure

1

The optical rotation of four proteins at
365 nm. is shown as a function of Urea ( C)
and Guanidine HCl ( O ) concentration.

(a) Ribonuclease at pH 6.6. (b) Lysozyme
at pH 2.9. (c) a-Chymotrypsin at pH 4.3.
(d) s-Lactoglobulin at pH 3.2. (From
Reference 3)

The amino acid sequence of CHT.

The packing of CHT molecules in the crystal.
(a) Projection on the yz-plane. The
asterisks denote the active site regions.
(b) Projection on the xz-plane.

Schematic illustration of the folding domains
in the CHT molecule. Cylinders 1 and 2.
Denoted, reSpectively. .

The distance diagonal plot of CHT. Plot
contours are 5.0, 10.0, and 15.0 A.
Interatomic distances, ri', were calculated
and plotted using the CEMCOMGRAF facility,
De artment of Chemistry, Michigan State

University. Plot courtesy of Dr. S. R. Ernst

The nonpolar interdomain region of CHT shown
is projection along the local two-fold
rotation axis. The residues shown occur

in the x-interval 45/76 to 65/76.

Page

10

12

15

. 17

20

Schematic representation of the dimer interface
interactions projected on a plane perpendicular

to the yz-plane and containing the local two-
fold rotation axis (horizontal line through
center of figure). (From Reference 52).

The soaking cell used for the preparation of
the denatured derivative CHT crystals.

vii

22

26

List of Figures (Continued)

Figure

9

10

11

12

13

14

15

l6

17

18

19

The axial 26- -scans for the 0. 01 M. TCA,
2. 0 M. GdnHCl, and 3. 0 M. urea derivatives,
and native CHT, respectively.

The distributions of average reduced
derivative and native intensities in
intervals of 29.

Argand diagram of the native and derivative
structure factor vectors. (a) The centro-
symmetric case. (b) The noncentrosymmetric
case.

The difference diagonal plot of the 0. 01 M.
TCA derivative.

The difference diagonal plot of the 2. 0 M.
GdnHCl derivative.

The difference diagonal multiplicity plot
of the 2. 0 M. GdnHCl derivative.

A feature of the 2.0 M, GdnHCl difference
map located in the dimer interface region.

The difference diagonal plot of the 2. 0 M.
GdnHCl derivative, calculated only for the
difference peaks in the dimer interface
region, IDistances ri calculated for each
map point > 0.2 aA.. -5 within each peak.

The difference diagonal plot of the 3.0 M,
Urea derivative.

The difference diagonal multiplicity plot
of the 3.0 M, Urea derivative.

Difference peaks observed in the A- Chain/
B-Chain contact regiOn. The difference map
contours are the thick gray lines. (a)
Sections of electron density in the

x- interval 66/76 to 75/76. (b) Sections
of electron density in the x— interval

55/76 to 65/76

viii

Page

31

39

42

53

58

61

66

69

77

78

98

Chapter I

Denaturation and the Mechanism of Protein Folding: An
Introduction

Protein denaturation is a process in which the three-
dimensional structure of a protein molecule is changed from
that of the native conformation to a more disordered structure.1
This unfolding is accompanied by loss of the native
function and by dramatic changes in many of the physical and
chemical prOperties of the protein. Denaturation can be
induced by eXposing the protein to heat, extremes of pH,
solvent—air boundaries, and a wide variety of chemicals which
includes some inorganic salts (e.g., LiBr and KSCN), deter-
gents, organic solvents (especially halogenated acetic acids
and ethanols), and a class of organonitrogen compounds (urea,
guanidine HCl (hereafter referred to as GdnHCl), and related
compounds). Susceptibility to denaturation by each of these
varies from protein to protein. The regeneration of the
native conformation from the fully-unfolded (random coil)
form also varies from one protein to another, and with the
denaturant for a given protein. Acids, heat, GdnHCl, and,
urea are the most frequently used protein denaturants with
GdnHCl and urea being more.effective than the former. A
greater degree of unfolding is generally observed with GdnHCl
than with urea solutions of the same concentration.2 Two of
these observations are illustrated by Figure 1 in which the
reduced Specific rotation at 365 nm. is plotted against

denaturant concentration for the proteins lysozyme, ribonuclease,

l

 

 

 

240-
230-
220'
210-
rQ200-
L4190.
. 1 '180-
230.11 .-.11 128'
0123 78910 150

 

 

 

 

 

 

4 6
MOLARWY
(a) Ribonuclease at pH 6.6.

 

(b) Lysozyme at pH 2.9.

 

 

 

 

 

 

 

 

 

29C> . reg . 11
28C) 320. 1
270- 3KK)* 4
26C) 28C)‘ «
250 - 260 ~ 1
:nZSC)' gag. .
5'3 230 » S; 220 r «
220 » fi' 200 - ‘
210 , 180 ' 'l
30c). 16C)P ‘
190 ~ 138’
‘ l’
180 1 . A . .111 1 100 1 i . . . 11111
0’1 2 £5 4 55 6»'7 8 C) I 2 £544 5 E5 7 8 53
MOLARUY MOLARUY' _
(c) cx-Chymotrypsin at pH 4.3. (d) £3~Lactoglobu1in at pH 3.2.

Figure 1. The Optical rotation of four proteins at 365 .
is shown as a function of Urea (. ) and Guanidine HCl (non )
concentration. (From Reference 3)

a-chymotrypsin, and B-lactoglobulin.3 All the proteins
undergo significant structural changes at lower concentrations
of GdnHCl than urea, and the concentrations at which the
transitions occur vary somewhat among the four proteins.

The literature in this field is extensive and a number of
very helpful review articles have been published including

those of Edelhoch and Osborne,” Franks and Eaglund,5 Pace,2

6.7

Tanford, and Kauzmann.l
The mechanism of protein denaturation can be considered

half of one of the fundamental problems of biochemistry--the

mechanism of protein folding. Because many proteins can

undergo denaturation reversibly Mg £1222 to regain the fully

functional native structure from the denatured conformation,

it is presumed that the amino acid sequence contains all the

information necessary to determine the three-dimensional

structure. The manner in which this is accomplished from

the randomly-coiled polypeptide chain is not well—understood;

a random search by the molecule of all the possible conformations

is not the mechanism as such a search would take prohibitively

long.8 The most p0pu1ar theory currently is that of a

folding mechanism in whichthe polypeptide chain is guided

to the final conformation through the formation of local

structures in different parts of the molecule (nucleation).

It is generally assumed that the folding nuclei are elements

of secondary structure (a-helix, B—sheet, B-bends) which are

'formed earliest and which then direct the rest of the protein

9

molecule into place.

L,
Thermodynamic studies have generally supported a two-

state folding/unfolding mechanism in which the native con-
formation (N) and the denatured conformation (D) are directly

interconvertible as indicated by the equilibrium expression7’lo’ll

It;3IL

12 studied the thermal

For example, Privalov and Khechinashvili
denaturation of five proteins (metmyoglobin, ribonuclease,
cytochrome c, lysozyme, and a-chymotrypsin) by scanning
microcalorimetry techniques. They found that to a good
approximation the thermal denaturation of these proteins could
be described by the two-state mechanism, and that the small
deviations from the model which were observed must be due

to folding intermediates present in small concentrations.
Further, they produced the interesting result that the
differences in free energy between the native and the denatured
states for the five proteins are all in the range 7-13 kcal./mole
of protein at 37°C.

Kinetic studies of denaturation processes present a more
complicated picture. The presumption that a protein molecule
must proceed through partly-folded intermediates in
transit between the native and denatured states is supported
by the results of many kinetics eXperiments.2 Commonly
observed in many therma1-, urea-, and GdnHCl-induced denatur-
ation processes are simultaneous fast and slow unfolding and
refolding reactions whose rates may be different by two

orders of magnitude.13 As one example, Baldwin and coworkersln

5
observed two reactions with rates differing by a factor of
450 inthe refolding of GdnHCl-denatured ribonuclease at
pH 5.8 and at 25°C.

Brandts and coworkers13 have suggested the possibility
that the slow folding (or unfolding) reaction may be due to
cis-trans isomerism of the amino acid residue proline. They
found from a review of the known protein structures deter-
mined by X-ray crystallography that the trans-conformation
of proline is that most often observed (90-95% of all prolines
observed) in native proteins. Therefore, the normally-trans
proline residues in the native protein may be in equilibrium
with the cis-conformation in the random-coil protein, and
the cis-prolines may impede the normal refolding rate of
the protein molecule. They suggest further that the slow
unfolding reaction is really a transition among two or more
partly unfolded forms. The partially unfolded intermediates
formed as a result of a cis-trans equilibrium should be
detectable in kinetics experiments but not in calorimetric
eXperiments, as the enthalpies of the cis—trans isomers are
about the same.

The enzyme<x-chymotrypsin (hereafter referred to as CHT
except when distinguished from other chymotrypsins) and its
enzymatically-inactive precursor chymotrypsinogen (hereafter
referred to as CHTgen) have been the subjects of many denatura-
tion studies. The native-denatured protein equilibrium of

15

CHTgen was studied by Brandts over the pH range 0.5—3.7

and over the temperature range 0-65°C. He found that the

6

heat capacities of the native and denatured forms differ by
several thousand ca1./deg. mole and that a salt present in
the CHTgen solutions could act either to stabilize or
denature the native protein depending on the pH of the solution.
Biltonen and Lumry16 studied the denaturation of CHT in the
acid pH range and found two distinct native-denatured equilibria.
Porter and Prestonl7 found that CHT retains its catalytic
activity in dilute solutions of the detergent sodium dodecyl
sulfate (SDS). Finally, the change in free energy for
the acid-thermal unfolding of CHT at pH 3 and at 25°C. has
been determined to be 7.4 kcal./mole CHT.15

The study of denatured proteins by X-ray methods was
first undertaken by Astbury and coworkers in 1935 who obtained
and analyzed X-ray photographs of fibers of denatured egg
albumin and seed proteins.l9 They observed that the conformation
state of the denatured proteins was that of the extended
polypeptide chain as Opposed to the globular conformation
exhibited by native proteins in solution. The first account
of protein denaturation in a crystal is that of Snape, et.
a]...20 who observed changes in the X-ray diffraction pattern
of lysozyme at several concentrations of urea. They were
able to produce crystals‘which diffracted well even when
soaked in urea solutions as concentrated as 9M, However,
they concluded that the changes in structure, even at lower
urea concentrations such as 3M, were so great as to preclude
a detailed analysis based on conventional difference electron

density maps. A low resolution (3.0 A.) difference map

7
revealed widespread changes throughout the lysozyme molecule
More recently, the results of some crystallographic denaturation

21’22 have been published.

studies by Yonath and coworkers
They have examined the effects of several denaturants (urea,
KCNS, bromoethanol, and SDS) on triclinic crystals of lysozyme
cross-linked by glutaraldehyde. They find, for example,

that SDS penetrates the hydrophobic interior of the lysozyme
molecule causing the two "wings" to spread apart.

At the outset of the work to be reported here, the
effects of several chemically-distinct denaturants on the
three-dimensional structure of CHT were undertaken for
investigation. Urea and GdnHCl were obvious choices.
Trichloroacetic acid (TCA), which has been used in the
separation of protein inhibitors from the other proteins
present in many varieties of beans, was another choice

presumably because it has special denaturing abilities not

possessed by the more powerful mineral acids.

Chapter II

The Crystal and Molecular Structure of Chymotrypsin.
1. The Activation and Primary Structure of Chymotrypsin.

The function of the enzyme CHT is the degradation of
other proteins during digestion. Specifically, CHT catalyzes
the hydrolytic cleavage of peptide bonds whose carbonyl
groups belong to amino acid residues with aromatic side
chains (Trp, Tyr, Phe). It is a serine protease as its
catalytic prOperties are due to an eSpecially reactive
serine residue. It is functionally and structurally similar
to a number of other serine protease enzymes including the
digestive enzymes trypsin and elastase, the blood clotting
enzymes factor X and thrombin, and the bacterial proteases
alpha-1ytic protease and subtilisin.

a-CHT A is produced in the mammalian pancreas as the
catalytically-inactive precursor (zymogen) CHTgen A. Bovine
CHTgen A is a single-stranded polypeptide chain containing
245 amino acid residues and cross-linked by 5 disulfide
bonds. The amino acid sequence, or primary structure, of

23’24’25 and confirmed

CHTgen A was determined by Hartley
Meloun, 23. al.26 There exists a second bovine CHTgen,
CHTgen B, also found in pancreatic extracts, whose amino
acid sequence contains 245 residues and is very similar to
the sequence of CHTgen A,27 and can be activated by trypsin
to form the active enzyme CHT B.

Bovine CHTgen A can be activated to yield several CHT

A's by the enzymatic cleavage of certain peptide bonds by

8

9
trypsin and CHT. n-CHT results from the cleavage by

trypsin of the CHTgen peptide bond Arg 15-I1e 16. Subsequent
cleavage by CHT of the peptide bond Leu 13-Ser 14 and the
removal of the dipeptide Ser 14-Arg 15 gives 5-CHT. Further
cleavage at Tyr l46-Thr 147 and Asn 148-Ala 149 by CHT
results in the removal of the dipeptide Thr 147-Asn 148
and yields a-CHT.28'29 The key cleavage step is the first
one as both'N- and d-CHT are fully active enzymes having,
in fact, significantly higher activities thana-CHT.30 A
fourth CHT, y-CHT, is known and has been shown to be a pH
conformer ofa-CHT.31

X-ray crystallographic methods have shown that the n-,
5-, and Y-forms of CHT have isomorphous crystal structures
indicating that these forms of the enzyme have very similar

32

conformations. In fact, the overall folding of the poly-
peptide chains in CHTgen and all the forms of CHT is simi-
lar.33’Bu As a result of comparisons of these protein
structures, the changes which occur upon activation are
relatively small in number and result mainly in an improved
substrate binding site which is only partially formed in

34

CHTgen and small changes in the orientation of the active

Ser 195 side chain.35
The a-CHT molecule is composed of 241 amino acid residues

in three polypeptide chains—-the A-chain (residues 1-13),

the B-chain (residues 16-146), and the C-chain (residues

l49-245)—-with a molecular weight of approximately

25,300. The amion acid sequence is shown in Figure 2, with

Figure 2.

The

Pro
Gln

11¢

Val-Leu-Scr-Cly-Leu
10

13

Glu'Ann-Ila-Leu'Ser'Gly°Cly'CyI'Ph¢'H1l'

Asn 50

£5

1.()

11¢
16

Trp’Val'Val'Thr'Ala'Aia

55

Ile'Lys'Leu'Lys'Cln'Ile

All

80

Lys-Val-Phe-Lys-Asn-Ser

90

Cys'Val°Ala-Ser°Val-Thr°61n'5er

Leu
Pro

Ser 125

Asp

120

Gly-

Trp
61y
s.
Th:

Val

115

Leu

-Val°Asn.Gly-Clu-

20 25
Phe-Cly°Tht'Lye-Alp'cln'Leu'Ser°
i 60 35
S

'Hll'Cyl'Cly'Val'Thr'Thr'Ser'Anp'Val'an'an
60 65

All

Ser‘Cly‘Gln'Aep‘Phe'Glu'Gly
75 70

‘Lys'Clu‘Ser’Ser'

-Lys-Tyr'Asn-Ser-Len-Thr-Ile-Asn-Asn-Asp

95 100 11¢
Thr
-Ph¢°Ser-A1a-Ala-Thr'Set-Leu'LyS'Leu'Leu
110 105
°Thr'Arg-Tyt

165 150 155

Clu-AII°VII'PIO°Cly'SCt'TIP'PrO‘Trp

Gin 30

Val

Ala'Asn'Thr-Pro'Asp'Arg~Leu'CIn°Cln°Ala

Set

Thr-Gly-Irp-Tyt-Lye-Lys-Cys‘AIn-Thr-Asn-Set-Leu-Leu-Pro-Leu

. 170 S 165
Lye 175 I
- S
Ile-Lys'Asp-Ala-Het~Ile-Cye-Ale-Gly-Ala-Set-61y
18 180

Phe-Ala-Ala-Gly-Thr-Thr-Cys-S-S-Cye-Val'LeurPro.cly-Gly-Set-Aspocly-Het-Cys-Ser

130

135

265

i

. 200 195 S 190
Lye

O s
Lye
Asn Thr
ciy 205 set

Ale-Trp-Thr-Leu-Vsl'cly-Ile'Vel-Ser'TIp-Gly-Ser
210 215

240

amino aciEFSequence of CHT.

V11
Si:

221

230

160

CyI-Ser-Thr-Ser

Thr
Pro
61y

Val

Asn-Ala-Aln-Leu-Thr-Gln-Gln-Val°Trp°Asn-Val'Lou°Ala-Thr°Val-Ar3°A1l°Tyr

225

11
the numbering of the residues being the same as that of the
amino acid sequence of CHTgen. Five disulfide bonds connect
residues 1 and 122, 42 and 58,136 and 201, 168 and 182,

and 191 and 220 forming both inter- and intra-chain linkages.

2. The Crystal Structure of Chymotrypsin.

CHT crystallizes as monoclinic plates from solutions
50% saturated in ammonium sulfate at pH 3.6. The crystal
structure has the symmetry of Space group P2l with four
molecules per unit cell or two molecules per asymmetric
unit. This arrangement is illustrated in Figure 3. The
molecules form "chains" parallel to the c-axis such that
the molecules of one chain are related to the molecules
of neighboring chains by two-fold screw axes coincident with
and parallel to the b-axis. Thus, Molecules I and I' are
related.tx> molecules II and II' respectively by the two-fold
screw axes (Figure 3b). Molecules in the same chain are
related by two different non-crystallographic or local
two-fold rotation axes36 labelled A and B (Figure 3a). The
local two-fold axis A relates molecules I and II to I' and
II' reSpectively. The local dyad B also relates I to I'
but in a different way.

Molecules of CHT form dimers in solution at lower pH
and from comparisons of the results of solution dimerization
equilibrium eXperiments with the X-ray crystallographic
structure, it has been concluded that the dimer in

solution is the one formed by dyad A in the crysta1.37'38

12

 

 

 

 

 

 

 

 

Figure 3. The packing of CHT molecules in the crystal.
(a) Projection on the yz-plane. The asterisks denote the
active site regions. (b) Projection on the xz-plane.

13
Therefore, all further references to the CHT dimer refer
to this pair of molecules.
Another important aSpect of the crystal structure is

the solvent channels parallel to the c-axis which result
from the manner in which the CHT molecules pack in the
crystal. The channels are large enough to permit the diffusion
of small molecules, i.e., heavy-atom compounds, enzyme
inhibitors, denaturant molecules, fluorescent probes,

etc., into the crystal lattice to interact with the

protein.

3. The Molecular Structure of Chymotrypsin.

The structure of the CHT molecule has been determined
independently using X-ray crystallographic methods by Blow
and coworkers at the Medical Research Council Laboratories,
Cambridge, England39’l+0 and by Tulinsky and coworkers at

“1’42 The details of the structure

Michigan State University.
are recorded in the papers cited. The intention here is to
describe briefly those aSpects of the CHT structure most

pertinent to the results to be presented later.

(a) Secondary Structure.

The CHT molecule consists largely of extended poly-
peptide chains, often running parallel or antiparallel to
one another, forming several sections of ill-formed B-sheet.
The extended chains are often terminated by B-bends--sharp
turns in the polypeptide backbone in which a hydrogen bond

may occur involving a peptide carbonyl oxygen and an<x—amino

14
hydrogen three residues away. There are two short helices
in the molecule: three turns of a-helix made up of the
polypeptide segment Leu 234-Asn 245 occurs at the carboxy- '
terminal end of the C-chain on the exterior of the molecule
on the side Opposite the active site: the other helix is
one-and-a-half turns of distorted a-helix between Ser 164

and Gly 173, also occurring on the surface of the molecule.

(b) Folding Domains and the Distance Diagonal Plot.

The tertiary structure of a protein arises from the
organization by the protein Of its elements of secondary
structure to form a globular molecule. From an examination
Of its B-sheet structure, it is evident that the CHT molecule
is organized into two folding domains. These have been

40 and

described in the literature by Birktoft and Blow
called Cylinders 1 and 2. Cylinder l is made up Of B-sheet
from polypeptide segments of the B-chain. The axis of
Cylinder 1 lies nearly in the plane defined by the local
two-fold axes at an angle Of about 400 with respect to the
two-fold axes (Figure 4). The axis of Cylinder 2 makes an
angle of about 450 with respect to the ac—plane of the

crystal at right angles to the axis of the Cylinder l. The
two cylinders, or domains, thus formed divide the CHT molecule
roughly into halves such that the molecule may be bisected

by a plane parallel to the bc-plane Of the crystal close

to the active site. The arrangement of the cylinders

with reSpect to theoverall shape of the CHT molecule is

illustrated schematically in Figure 4.

tN

 

 

 

 

     
    

ACTHEZ
EHTE
REGKMV

 

 

J

 

Figure 4. Schema 1c illustrallon of the folding domains
in the CHT molecule. Cylinders 1 and 2. Denoted,
respectively. ‘

16

The interiors Of both cylinders consist almost entirely
Of aliphatic side chains (Val, Leu, Ile) except for a small
number of polar side chains occurring in both cylinders
near the center of the molecule. Two serines--Ser 54 and
Ser 104--occur in the innermost end of Cylinder 1. Three
polar side chains--Thr 138, Ser 190, and Tyr 228--are found
in the innermost end of Cylinder 2.

A concise and dramatic representation of protein
tertiary structure is the distance diagonal plot.43’uu The

distance diagonal plot is an NxN matrix, the elements of

which are r.

1j' the distances between the i-th and the j-th

a-carbon atoms, where N is the number of amino acid residues
in the protein molecule. The array thus calculated may
be contoured at convenient distance levels. The distance
diagonal plot for CHT is shown in Figure 5. Since rij = rji’
the matrix is symmetrical and only one-half is unique.
Features of the distance diagonal plot correspond to the
close approach(below an arbitrary limit) of onecx-carbon
atom to another and are a unique two-dimensional representation
of the three-dimensional folding.

Elements of secondary structure are immediately evident
from the distance diagonal plot. An a-helix appears as a
thick feature on the diagonal and the C-terminal O-helix
(residues 234 to 245) of CHT is distinctly visible. Segments
Of parallel B-sheet are fatures parallel to the diagonal
at some distance from it. Anti-parallel B-sheet is

represented by lengths Of density perpendicular to the

 

 

V
16?

“£3735 ”we r'

.......................................................

 

 

 

Figure 5. The distance diagonal plot.of CHT. Plot contours
are 5.0, 10.0, and 15.0 A. Interatom1c distances, rij, were
calculated and plotted using the CEMCOMGRAF facility,

Department Of Chemistry,*Michigan State University. Plot
courtesy of Dr. S. R. Ernst.

18
diagonal and connected to the diagonal. Other features
occurring further from the diagonal represent the close
approach of polypeptide strands and lOOps which are not
hydrogen-bonded secondary structural features. Detailed
interpretations of these have been given e1sewhere.u'5'46

The CHT folding domains formed from segments of parallel
and antiparallel B-sheet appear in the distance diagonal
plot as two groups of blocks stacked together on the
diagonal - the two largest features in the plot. A further
interesting feature of the CHT distance diagonal plot is
the approximate mirror of symmetry perpendicular to the
diagonal at about residue 123. The two domain features are
clearly mirror images suggesting that the amino acid sequence
1-122 might be homologous to the reversed sequence of 123-245
(palindromic homology). Conventional measures Of sequence
homology,47 however, have not borne out this relationshipfia’49
That folding domains are a widespread feature among
proteins, and that the distance diagonal plot may be useful
for studying them has been pointed out by Rossman50 and by
50

Kuntz.”5 Rossman and Liljas showed that particular
structural domains may be associated with a particular
function, and that the corresponding parts of different
enzymes performing the same function may have the same
three-dimensionalfolding (as indicated by the distance
diagonal plot) while exhibiting little amino acid sequence

homology. Further, Liebman”6 has discovered an example

Of similar folding occurring in two proteins having

19
Opposing functions, i.e., CHT (enzyme) and pancreatic
trypsin inhibitor (substrate) where the overall folding
of the latter approximates that Of the first half of
Cylinder 1 Of CHT.

(c) Nonpolar Interdomain Region.

The two Eysheet cylinders with interior nonpolar amino
acid side chains constitute two large hydrOphObic volumes
in the CHT molecule. There exists also a third hydrOphObic
region--a semicircular band of nonpolar side chains wrapping
around the inner ends Of both cylinders. In contrast to
the cylinders which are filled with aliphatic side chains,
this region has in addition to aliphatic residues a large
number Of aromatic groups, most of which are arranged in
clusters. This region also contains a nonpolar cavity,
about 8 A. in diameter, just inside the molecule near the
C-terminal a-helix. The nonpolar amino acid residues which
form this unusual feature are highly conserved among the
other serine proteases whose tertiary structures are similar

to that of CHT.”2

The arrangement Of the nonpolar groups
and the nonpolar cavity are illustrated in Figure 6.

The CHT molecule contains four clusters Of aromatic
side chains, three of which occur in the interdomain region.
The tryptOphan cluster--Trp 27, Trp 29, and Trp 207--along
with the three prolines--Pro 4, Pro 8, and Pro 28--con-
stitutes a binding site for some substrate-like molecules51

as well as capping one end of the hydrophobic cavity.

The residues Trp 51, Phe 89, and Trp 237 cap the other end

20

 

 

 

 

     

Pro 8

4
20? p.12:
Trp 29 Prol98Trp14'

Prm His 40
Phe 41

KM

Gay
Trp 51

Tr‘p 237

Phe 89

   
    

 

 

Figure 6. The nonpolar interdomain region Of CHT shown
is projection along the local two-fold rotation axis. The
residues shown occur in the x-interval 45/76 to 65/76.

21
of the hydrophobic cavity. The group His 40, Phe 41, and
Trp 141 extend toward the active site region completing
the semicircular interdomain arrangement of aromatic
residues. A fourth aromatic cluster-—Tyr 171, Trp 172,
and Trp 215—-occurs just inside the surface of the molecule
near the 1.5 turns of distorted a-helix formed by residues

Ser 164-Gly 173.

(d) The Dimer Interface Region.

The dimer interface region, or the region in which the
two CHT monomers make contact about dyad A as shown in
Figure 3a, is formed as a result of the complementary fit of
the monomers. It is characterized by a number of mutual
close contacts between the amino acid side chains Of both
molecules involving van der Waals interactions, hydrogen
bonding, and ion-pair bond formation. These interactions
are summarized schematically in Figure 7. The polar inter-
actions involve pairs of residues in a mutually reciprocal
manner while the nonpolar interactions involve residues
of one molecule with their exact counterparts in the other
molecule.52

At pH 3.6, the e-amino group of lysine is protonated
making possible the two ion-pairs Lys 36'-ASp 153 and Lys
36-Asp 153', where the primes indicate that a particular
amino acid residue belongs to CHT molecule I'. The terminal
amino group at Ala 149 is also protonated at this pH
resulting in the ion-pairs ASp 64'—Ala 149 and Asp 64-Ala

149'. Extensive hydrogen bonding also occurs. The B-chain

22

.Amn monoummmm Eoumv .Aousmflm mo Hopsoo gwsonnw mafia Hapsouanozv

mflxm COHPMPoH cachiosp Hmooa one msazflmpcoo can osmamuuh may OP pmHsOHonmhom Osman

a so umpomwOMQ msoapomumpsw oommnmpsfi nosflc one mo COHvMpcmmmunmh oavmsmgom .m shaman
ogvdm< :m3;.<J<

e: .5: .. an 2:

 

ea cum- :~ .55 o: .23.: 3. .55
. «a. .5: . _ _ an .2...
Eu cum L.“ gun «a. 5: “cm nee—m on E...
.8. cum 2.. .22:
1.... .9: -2; E: T... a? -on .95

a: «.2 .. em a:

i3
carboxy-terminal residue Tyr 146 is located such that it
may form hydrogen bonds with the carbonyl of His 57 of the
other CHT molecule and with the hydroxyl-group of Ser 195
Of the other CHT molecule through a sulfate ion bridge.
Finally, van der Waals forces determine the equilibrium
positions of the phenyl groups of Phe 39 and Phe' 39 so
as to minimize close contacts between them.

The two residues Phe 39 and Phe' 39 display a further
general property of the dimer interactions. The local two-
fold symmetry in this region is far from exact and differences
in conformation occur between the two molecules. When CHT
molecules dimerize, the phenyl rings of Phe 39 and Phe' 39
must rotate about their reSpective Qu'CB bonds because of
close contacts and do not display local two—fold symmetry.
This lack of two-fold symmetry has been termed variability in

41,42 The

structure and has been discussed in some detail.
variability in structure is Often reflected in chemical
behavior as well, both in the asymmetric binding of small
molecules to the protein and in changes in the protein

molecules accompanying derivative formation. This phenomenon

is displayed by the TCA, the GdnHCl, and the urea derivatives.

Chapter III.
Experimental.
1. The Crystal Soaking EXperiments.

The denaturation experiments were performed by exposing
crystals Of CHT to gradually-increasing concentrations of
several chemical denaturants and by Observing the diffraction
patterns Of individual crystals at periodic intervals of
time. If substantial changes were evident with reSpect
to the diffraction pattern Of crystals of native CHT, the
three-dimensional X-ray intensities were measured at 2.8 A.
resolution, reduced to structure amplitudes, and scaled to
the native intensities, and difference electron density
maps were calculated and examined.

The CHT used in all the denaturation eXperiments was
bovine a-CHT, three times crystallized, from Worthington
Biochemical Corporation. All the denaturants and other
chemicals were commercially-available, reagent grade quality,
and used without further purification. Crystals Of CHT
were grown from aqueous solutions approximately 50% saturated
in ammonium sulfate at pH 3.6 with a protein concentration
Of about 5 mg./ml. of solution. The tubes containing the
mother liquor were stored at room temperature and the
crystals--diamondeshaped plates--usually appeared within
four to seven days after the solutions were prepared.

After two weeks time, crystals Of suitable size (1.0-1.5 mm.
along the longer face diagonal) and apparent quality (clear

and without cracks) were harvested and placed in tubes

24

25
containing aqueous solutions 75% saturated in ammonium
sulfate with the pH adjusted to 3.6. The crystals were then
stored in a water bath at l4-16°C. Crystals Of CHT may be
stored in this manner indefinitely. Those having a regular
shape and as thick as possible (0.2—0.3 mm.) were chosen for
further soaking experiments from this stock of stored crystals.
The soaking eXperiments were carried out in two-
chambered soaking cells (Figure 8). The cells consisted
Of a lower chamber (a 5 ml. snap-cap vial) containing the
crystals bathed in the storage solution, covered by a
cellophane dialysis membrane, and an upper chamber fashioned
from a short length of glass tubing having the same outside
diameter as the vial. To assemble the cell, the crystals
selected for the experiments were placed in the vial and
covered by the 75% saturated ammonium sulfate solution.
The vial was filled completely and covered by the membrane
so that no air bubbles were trapped under the membrane.
The membrane was then secured in place by a sleeve made
from Tygon tubing. The upper chamber was then filled with
solution and capped with a rubber stOpper or parafilm.
The total volume of the cells were about 10 ml. with the
volumes Of the upper and lower chambers being about equal.
The soaking eXperiment is begun with the upper chamber
filled with the 75% saturated ammonium sulfate solution.
A 1-2 ml. aliquot of denaturant dissolved in 75% ammonium

sulfate at pH 3.6 is exchanged for an equal volume of

26

 

 

 

 

 

 

GLASS TUBING
/14mm.00. X 50mm.
. l ,
Salty‘é‘iSANEl 4.4 ------ l (TYGON TUBING 32— LD.
MEMBRANE
l DRAM SNAP-CAP
‘ VIAL (VOLUME2 5ml.)
annual

 

 

K CRYSTALS

Figure 8. The soaking call used for the preparation Of the
denatured derivative CHT crystals.

27
the solution already in the upper chamber. After this has
been done daily for several days and the denaturant con-
centration in the cell approaches that Of the stock denaturant
solution, the solutions may be exchanged by simply pouring
out the upper chamber and refilling it with the stock at
the end of 1 1/2-2 weeks, the concentration of the denaturant
in the cell is essentially that of the stock denaturant
solution. At this time, a crystal may be withdrawn for an
examination of its diffraction pattern. If no changes are
evident, either with reSpect to the native pattern or that
of some previous concentration derivative, the soaking cell
is reassembled and the aliquot-exchanging process is resumed
with the denaturant at some higher concentration. This pro-
cedure is continued until substantial changes have appeared
in the diffraction pattern of the denatured protein crystals.
The pH of the solution in the upper chamber was measured
before each aliquot was exchanged to provide a check on the
course of the experiment. A variation in pH measurements
Of 0.1-0.2 pH units from one day to the next was not unusual,
but the pH was never allowed to drift from 3.6 by more
than 0.3 pH units.

There are several advantages to the use Of this soaking
cell over the use Of a test tube. The dialysis membrane
impedes the physical mixing of solutions reducing the
possibility Of damage to the crystals by a sudden change in
the solvent environment. The crystals are physically isolated

from the pH meter electrode and transfer pipettes reducing

28
the danger of their being accidentally jarred or crushed.
Finally, exchanging aliquots is convenient when one can
simply pour out the upper chamber without jostling the
crystals about.

The choice of starting concentration for each denaturant
was determined from literature accounts of protein denaturation
experiments in solution along with the expectation that
changes in the structure of CHT may occur at lower denaturant
concentrations than those necessary to obtain a fully-
unfolded, random-coil conformation. The urea and GdnHCl
experiments began at 1.0 M. concentrations and proceeded
to higher concentrations, going in 1.0 M. increments for
the urea and 0.5 M. increments for the GdnHCl. Both these
compounds are very soluble in 75% saturated ammonium sulfate
solutions such that 3.0 M. GdnHCl and 5.0 M. urea solutions
are easily produced. TCA also dissolved readily in the
ammonium sulfate solutions, but because less was known
about its denaturing abilities, a starting concentration
Of 0.5 M. was chosen. The pH Of the TCA solutions was 3.6
like all the other derivative soaking solutions.

The urea eXperiment proceeded in 1.0 M. steps beginning
at 1.0 M, and terminating at 5.0 M. Substantial intensity
changes first appear in the axial reflections Of the 2.0 M.
derivative and reappear in a generally more pronounced way
in the 3.0 M, derivative. Further changes which appear in

the 4.0 M. diffraction pattern are generally continued in

29
the 5.0 M, derivative. Since only small changes are
evident between the 2.0 and the 3.0 M, Ze-scans, the three-
dimensional intensity data for the 3.0 M. derivative were
measured at 2.8 A. resolution.

Scans were recorded and examined for GdnHCl derivatives
at the following concentrations: 0.5 M,, 1.0 M,, 1.5 M,,
2.0 M,, and 2.5 M, Small changes first appear in the axial
reflections of the 1.0 M, crystals so that three-dimensional
intensity data were measured for this derivative at 2.8 A.
resolution. At 2.0 M, GdnHCl, further changes with respect
to the native diffraction pattern were Observed so three-
dimensional intensity data were also measured for this
derivative. At 2.5 M,, the crystals had become Opaque
with cracks and the diffraction pattern had sharply dete-
riorated making data collection impractical.

Crystals of CHT soaked in 0.5 M, TCA solutions 73%
saturated in ammonium sulfate at pH 3.6 became badly cracked
and disintegrated within 24 hours. Crystals soaked in 0.1
M. TCA solutions became thoroughly cracked but withstood
mounting in a capillary and the Ze-scans were recorded.
Dramatic differences with respect to the native diffraction
pattern, even at low 29-angles, were evident on all three
axes. The rapid decline in intensity with increasing
29-angle as well as the high background, which itself had
a peak at about 220 in 29, indicated the relatively poor
quality of this crystal. Finally, crystals soaked in 0.01
M, TCA solutions produced good quality diffraction patterns

3o

exhibiting significant changes with respect to the native
diffraction pattern, and three-dimensional X-ray intensities
were measured for this derivative at 2.8 A. resolution.

The axial Ze-scans for the 2.0 M, GdnHCl, the 3.0
M, urea, and the 0.01 M, TCA derivatives along with those

Of the native CHT crystals are shown in Figure 9.

2. X-ray Intensity Measurements.

Accurate measurement of the X-ray intensities which are
subsequently converted to structure factor amplitudes re-
quires the careful consideration Of.a number of experimental
problems, many of which arise because of the delicate
nature of a protein crystal. Protein crystals must be
enclosed in a sealed glass capillary in contact with the
mother liquor because if allowed to dry, the crystal lattice
breaks down and the diffraction pattern is lost. Protein

53

crystals mounted in this manner are held against the

inside wall of the capillary by surface tension Of the mother
liquor and occasionally move becoming misaligned during the
course Of data collection. If not corrected, this misalign-
ment will cause the intensity measurement to be in error,
reduced from the true value. Protein crystals are sensi-
tive to radiation damage so that in developing an optimal
data collection procedure, the reduction in intensity due

to radiation damage must be balanced against an adequate
amount Of eXposure for reliable precision. These and other

such problems have been overcome to the extent that it is

now routine practice in our laboratory to measure about 6400

31

0.01:4 TCA

.. llml lull . . l:

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

L—
O

 

Figure 9. The axial 29-scans for the 0.01 M. TCA, 2.0 M.
GdnHCl, and 3.0 M. urea derivatives, and native CHT,
respectively.

32

X—ray intensities from a single crystal corresponding to
about 75 hours of X-ray eXposure and typically eXperiencing
10 to 20% decay in intensity due to X-ray damage. Many of
the details of the experimental methods used in our laboratory
have already been described elsewhere.ul’Su

The X-ray intensity measurements to be described here
were made using a Picker FACS—l 4-Circle Automatic Diffracto-
meter controlled by a Digital Equipment Corporation (DEC)
PDP-8 computer interfaced also with a DEC 32K DF32 Disc
File and an Ampex TMZ 7-track tape transport. The soft-
ware system used in our laboratory is based on the program
package originally supplied with the FACS—l system by Picker
but adapted to perform protein data collections more suit-

54

ably. The data collected by this system is written on
magnetic tape and transferred to the Michigan State University
Computer Center for calculations on the CDC 6500 computer.

Due to the fact that protein crystals frequently move
during data collection, an automatic crystal aligning sub-
routine is an essential part Of the data collection soft-

54

ware. Motion Of the crystal is monitored by periodically
measuring the intensities of several reflections during the
course Of data collection. The monitored reflections are
chosen such that their intensities are very sensitive to
small changes in the angular coordinates, and in such a

way that they define the orientation Of the crystal. The

(0.18.0) reflection at the two ¢-angles correSponding to

+* +* . o _ .
the a and the c axes (wh1ch are 78 apart 1n ¢) 18 used

33

to define the coincidence of the crystal 3* axis and the
¢-axis Of the diffractometer. The intensities of this
reflection at these two O—angles are extremely sensitive to
small changes in m (0.050 change in w leads to about 10%
loss in intensity). Another reflection in general reciprocal
Space (non-Special angles) at high 26 is chosen to monitor
the ¢-angles to protect against a twisting motion about the
g* axis. These three standard reflections are measured
about once every hour during data collection, and if the
intensity of any one of them drOpS below a preset limit
(90% Of their initial intensity) the data collection program
transfer control to the crystal aligning subroutine.

The data collecting program is based on an m step-
scan procedure using balanced Ni/CO filters to Obtain a
correction for background radiation. The step-scan is
performed with respect to the most sensitive angle, w.
The diffractometer is driven to the calculated 29, X: and
¢ coordinates of a reflection and scans through a 0.150
range in beginning at -0.070 and going to +0.08O in.» in
0.030 steps. The intensity is measured for a preset time
at each step and the four largest intensities are summed
to give the intensity of the reflection (count-six—drOp-
two step scan).55 The crystal is then returned to the w-
angle correSponding to the highest intensity and the back-
ground is measured with a balanced CO filter. This back-
ground measurement is multiplied by 4 to Obtain the total

background intensity.

34

Occasionally, the crystal will undergo a slight mis-
alignment such that the reflection peak is not at w = 0.000.
TO compensate for this, the step-scan program checks the w—
angle for the step of highest intensity. If the maximum is
at, or next to, either end of the w-step range, the inten-
sities of one or two additional 0.03o steps are measured
beyond the end Of the step range. If the maximum is one
step from either end of the range, only one additional step
is measured. If the maximum is at the end of the range, two
additional steps are taken. The final reflection intensity

is still taken as the sum of the four highest intensities.

3. The Reduction of Measured Intensities to Structure
Amplitudes.
In principle, the relative structure amplitude |F(hkl)|
for a single crystal having a completely isotrOpic absorption
behavior may be expressed as a function of the X—ray intensity

by the relation
1
a
F(hkl) = [12:1 ]

where I(hkl) is the measured X-ray intensity and L and p are
the Lorentz and polarization corrections, respectively.

The polarization correction is necessary because the efficiency
of reflection of a non-polarized incident X-ray beam varies

with the diffraction angle 29 since it becomes plane-polarized

35

at 900 and has the functional form

_ 1 + cOSZ(29)
P ‘“ 2 '

which is independent of the method of intensity measurement.
The functional form of the Lorentz factor depends upon the
geometry of the diffraction eXperiment and for diffractometer

measurements is given by

_ 1
L _ Sin(29) '

The need for the Lorentz correction arises because the
incident beam is divergent.

Several other corrections are necessary for the proper
reduction Of CHT data. The most important source of system-
atic error in the X-ray intensities is the absorption of
X-rayS by the mother liquor, glass capillary, and by the
crystal itself. The absorption correction used in our work
is semi-empirical,56 based on the variation of the relative
transmission Of X-rays with respect to the azimuthal angle
(¢). The intensities of four (0,k,0) reflections (k = 2,

4, 6, 18) are measured versus ¢ and are used to calculate

an absorption correction for all general (hkl) reflections
depending upon d, 29, and k-index (reciprocal lattice level).
Since crystals of CHT are twinned along the 2* axis, the

(Okl) intensities are the sum of the intensities of the

larger crystal and its twin. This systematic effect is

36

w

(\l

m.~+
0.0
0.0
0.0:
.< m

mm.s
Hm.a
ma.m
ms.H

admom

.mzwesop mopmacm gamma on» an ompmflsofimo whouum pudendum on» was momeL»CdLap CM whenrzzo

 

 

- - flavss.HoH amass.me AoHVo~.ao Asvs~.oe are dense:
muwveec
c.n me.~ AHVS.HOH Awsvma.mo Rosana.ao Amvma.o: -oaoaroaee .m Ho.o
o.oH me.H lmve.aoH Aaavom.ee lavas.so Asses.se «on: .w o.n
o.~ oo.H Aaaa.aoa Amvno.eo Loves.so Anvs~.os Horace .m o.~
o.o as.“ olavm.aoa .< Amvma.oo .< loves.ao .< Anvsn.as Horses .m 0."
llqu Ava m m m w w>Hum>Mueo
hmomo prmm
cure oncoumeoEHQ Hfioo pus:

.muoudrwsmd wzwflmom ucm .20Hpozvmm .coflpOdHHoo memo szmCOHCH hmhnx mo zsmsczm .H dfinmk

37

corrected in data reduction by use of the relationship

kt
I(0kl) -- ITk—t' ‘I(0kl)'.

where kt is the ratio of the crystal intensity to the twin
intensity, I(0kl)' is the raw intensity, and I(0kl) is the
corrected intensity. Background radiation is measured so
that a correction to the observed intensity can be made.
In our eXperiments, background radiation is the result of
coherent and incoherent scattering and is measured through
the use of balanced Ni/CO filters. In actual practice,
however, the filters are not quite balanced, and a lack-
of—balance correction is applied to the Observed background.
A summary of pertinent data collection and reduction

parameters is given in Table l.

4. Scaling the Reduced Structure Amplitudes.

The data reduction calculations just described place
the individual intensities in a data set on a common basis.
The entire three-dimensional data set, however, must be
placed on a scale comparable to that of the native structure
amplitudes. There is also the further problem that the
X-ray scattering power of a crystal decreases due to
disorder with increasing 29-angle and that the rate of
decrease varies from native tO derivative sets of data.
Scaling is accomplished by fitting the distribution of

+

average reduced derivative intensities <|FD|2> to the
+

distribution of average native reduced intensities <|FN|2>

which have already been placed on an absolute scale

38
(with a B = 27 A.2, see reference 41). This is accomplished
by fitting the distribution of relative average derivative
intensities <|FD|2>, in intervals Of 29, to the native

intensity distribution varying the parameters k and B in

the equation

2

F 2 ( ) - F 2> (rel) 'k (2B ‘ 2 2
<| DI > abs — <| DI eXp Sin 9/1 )

to Obtain the best fit (Figure 10)- Table 1 contains the
.+
k and B values for the denatured derivatives where <|FDl2>(abs)
+
and <|FD|2>(rel) are the scaled and relative average

derivative intensities, respectively.

39

 

 

 

 

.. ' NATIVE
2 - 0 TCA
(IFI >X '0 2 a GdnHCI
O UREA
. 200!
9
I50 a \
\3
I00 '4 '
50 I I I I I I
O 5 IO 20 25 30

<2 9)

Figure 10. The distributions Of average reduced derivative
and native intensities in intervals Of 29.

Chapter IV

The Difference Fourier Synthesis.

1. The Difference Fourier Synthesis.

The large volumes of solvent which exist in the
protein crystal lattice enable small molecules or ions to
diffuse throughout the lattice and interact with the
protein molecules. The chemist may exploit this property
to obtain information about the protein beyond that immediately
deriveable from the structure Of the native protein alone.
The term native structure denotes the parent protein
structure determined by the method of multiple isomorphous
replacement at a particular set of solvent conditions. The
term derivative structure will denote structures differing
from the native due to changes in solvent composition or the
binding of small molecules or ions regardless of whether or
not a covalent bond is formed between the ligand and the
protein. The changes in structure accompanying the
formation of the derivative may be examined by the difference
Fourier synthesis.

The electron density at a point (x,y,z), o(x,y,z),
in a unit cell Of volume V in a monoclinic crystal is given

by the Fourier summation

F(hkl) exp [121: i(hx+ky+lz )], (1)

WMB

I-‘MB

0(X,y,Z) 1");

827948
II

0

O

40

41
where each term contains a complex structure factor EIhkl)
and the limits of the summation are such that the sum is
confined to the unique reflections. A difference Fourier

synthesis is used to compare a derivative electron density

map tO a native map,
L} [+ ..
Ap(x,y,z) = V E g f FD(hkl) — FN(hkl) '

eXp[-2Ni(hx+ky+1z)], (2)

where the coefficients are the differences between the
->

derivative structure factor, FD(hkl), and the native

+ +

struiture factor, FN(hkl). Figure 11 shows the vectors FD

and FN for a given (hkl) in the complex plane with the

phase angles 9D and “N' and moduli [Eb] and [EN], respectively.
Equation (2) may be rewritten with the structure factors

expressed as the products of structure amplitudes and phase

factors
_ 4 ' + ° 0
Ap(x,y,z) — V 2 E f [IFDlexp(1aD) - IFNIeXp(1aN)]
h

eXp[42ni(hx+ky+lZ)]. (3)

'5'.

where the (hkl) notation has been dropped forkconvenience.
The Fourier synthesis Of equation (3) is the exact difference
Fourier because both the derivative and native phases are
included. In practice, however, GD is not known and the
assumption is made that since the differences between the

derivative and the native structure amplitudes are small,

42

 

 

 

IMAGINARYAXIS
'F' I? I? r?’
I NI N AI A REAL
-o -r AXIS
IFOI FO

 

(a) The centrosymmetric case.

 

 

 

(b) The noncentrosymmetric case.

Figure 11. Argand diagram of the native and derivative
structure factor vectors.

43
the corresponding differences between 9D and “N will

also be small. Therefore, ON is used for “D in equation
(3). Further, each phase has associated with it a figure

Of merit, m, which is a measure of the error in “N’ and is
used to weight the terms of the Fourier series to give the
"best" electron density map. The "best" electron density
has the lowest r.m.s. error with respect to a map calculated
with exact phases.57 The difference Fourier synthesis

used in the derivative studies reported here correspond to

the "best" difference electron density and are

4 I .
Ao(X.y.Z) = - z z 2: m[IFD I - IF I] exP(1a )°
V h k 1 N N

SXp[-2fli(hx+ky+lz)]. (4)

2. Errors and Peak Heights in the Difference Fourier
Synthesis.

There are three main sources of error in the difference
Fourier-—the error in the native phases, the errors in the
intensity measurements, and the errors due to the use of
the coefficients AFeXp(iaN ) (where AF - —IF D|--|FNI) instead
Of the exact coefficients [IFD lexP(iaD)— IFNIeXp(iaN)].

(a) Errors in the native phases. Henderson and Moffat58
have shown that the average contribution of the difference
coefficient mAFeXp(iaN) to the height of features in difference
electron density maps is mzAF. The figure of merit is a
function of 29-angle and varies among reflections according

to the heavy-atom phasing SO that only an approximate

44

average estimate Of the error is possible. The average
figure of merit for 2.8 A. resolution CHT phases with
m _> 0.7 is 0.89. I

(b) Errors in the intensity measurements. The mean

square error, 62, in an electron density map resulting from

errors in the amplitudes is estimated by the formula59'6o
2 ‘* .
52 = 53-2 2 z o (IFI), (5)
V h k l

+ ->
where o(|F|) is the standard error in IFI and the summation
is over the independent reflections for a noncentric

monoclinic space group. For a difference Fourier with

amplitudes (IFDI - IFNI), the mean square error is
4 2 I 2 I
62 = __ z z z o (IFDII + o (IFNII . (6)
A 2
V h k 1

For difference coefficients to be significant,they must

at least satisfy
(IF - F ) 2( F ) + 2( F ) 9 (7)
pl I NI > 0 I DI 0 I NI °

(0) Errors due to the use of inexact coefficients.

Figure 11a illustrates the relationships among FN, FD, and
F; for the centrosymmetric case where the phase angles are
restricted to the values 0 and n. Changes of phase upon
derivative formation are rare for centric reflections SO
that dB and “N are probably the same. The coefficient,

AFexp(LuN) is then the exact difference structure factor.

45

In addition, the figures of merit for most centric reflections
are close to 1.0 so the centric terms in equation (4)
contribute at full weight and exactly to the difference
density.

For noncentrosymmetric reflections, the situation is
more complicated. Referring to Figure 11b, the structure
amplitudes IEDI' IFNI, and IFAI are related by

+

+ -> +
IFDI2 = IF'NI2 + IF‘IA2 " 2IFNIIFAI COSE" - (GA - O'N)]O (8)

Rearranging (8) and using the relation
COS(O:A - OLN) = -$- exp{l(ozA - ON)} + eXp{-i(<1A - aN)}] ,

an expression for the difference Fourier coefficient is

 

 

 

Obtained61 from which several Observations can be made.
" 2
. IFAI . .
AFeXp(1a ) = 1 1 eXp(1o ) (1)
N IF + IF I N
DI N
+ —>
IF IIF I
1 N 3 eXp(iaA) (ii) (9)
IFDI + IFNI
-> —>
IF IIF I
4 N &=— eXp[i(2a - a )1 (iii)
IF I + IF I N A
D N
+ -> + -> +
0n the average, IFNI and IFDI >> IFAI and since IFDI = IFNI,

term (i) will not contribute Significantly to the electron

density. Term (iii) is approximately

46

1
ngl exp[i(2aN — aA)]

 

and should simply contribute to the background Since “N

and “A are uncorrelated and their difference is, in general,

a random number. Term (ii) is approximately

+

IFI
A C
2 exP(1dA)

 

and makes the greatest contribution to the difference electron
density. The form Of term (ii) indicates that the peak
heights in the difference Fourier using the coefficients
AFexp(laN) will be about one-half the heights Of peaks in

an exact difference Fourier. This is consistent with the

62 that the peak heights Of atoms

Observation by Luzzati
in electron density maps not included in the phasing will
appear with peak heights approaching about one-half their
true height.

Blundell and Johnson conclude that the error due to
the use of approximate coefficients is prOportional to the

61 Henderson and Moffat58 considered the problem

r.m.s. AF.
in detail, but the formula they derived to estimate the
error in difference maps has since been shown to calculate
values that are low by a factor of 2. Ford, et.al.,63
report a corrected formula for the mean square error in
difference maps calculaed with the coefficients mAFexp(idN)

compared to maps calculated with the true coefficients

-+ -+
g;[|FDIexp(iaD) - IFNlexP(idN)] which is

2 2
2 4 [(AF) 5A
<Ao > = - z z z + . (10)
v2 h k 1 2 E

where 6A2 is the mean-square error in AF due to errors in
intensity measurements given by equation (6), and where the
summation is over the unique monoclinic reflections.

An upper limit to the error in the difference map is

the observed r.m.s. difference density

2. l.
<Ap2>0bs2 = l [E Z Z Ap(X,y,z)2]2 (11)
x y z

<:

calculated over the unit cell.

Finally, a discussion Of the errors in the difference
Fourier should include the Observation that the difference
Fourier is a sensitive function whose error level is nor-
mally a small fraction Of that of the parent protein
electron density. This is partly due to the cancellation of
series termination errors. Secondly, the error in a Fourier
series is in general prOportional to the magnitudes Of the
coefficients and because AF is on the average much smaller
than either IFNI or IFDI, the error in the difference
Fourier is correspondingly less than in eitherthe-FN or
the FD electron density maps. Difference Fouriers commonly
contain significant features with peak heights that are
less than one—tenth the features in the parent electron

density.58

Chapter V.

The 0.01 M, Trichloroacetate Derivative.
1. The 0.01 M, Trichloroacetate Difference Electron Density.
The difference electron density map of the 0.01 M,
TCA derivative revealed several unexpected features. First,
there is no evidence in this map that the structures Of the
CHT molecules are significantly affected by the presence Of
the TCA anion even though the b-axis shrinks by 0.45 A.
There are only three Significant difference peaks in the
asymmetric unit, all positive, and all suggesting binding
Of TCA to the protein. There are no significant negative
peaks to indicate movement by any part of either CHT
molecule in response to the TCA. Secondly, two Of the
binding sites are in the active site regions Of Molecule I
and I' near the substrate specificity sites in which the
aromatic Side chains of a substrate protein are located.
These two difference peaks are approximately related by
the local two-fold rotation axis. The other binding Site
is on the surface of Molecule 1. This site is intermolecular
and is formed by amino acid side chains of surface residues
of both Molecule I and a neighboring CHT molecule. The
unit cell coordinates and heights of these peaks are given
in Table 2. The Observed r.m.s.zn> calculated over a
featureless region of the asymmetric unit is 0.03 eA.-3.
Peak 1, the TCA binding site on the surface of Molecule
1, is 3.0—4.0 A. from the carbonyl group Of Thr 134 and
4.0—5.0 A. from the pyrrolidine ring of Pro 161. The

48

Table 2.

49
The Most Significant Features in the 0.01 M. TCA

Difference Map

Height
1 M Ie.A.‘3) Comments

0.057 0.192 0.30 Surface of Molecule I
0.283 0.385 0.27 Active Site Of Molecule I
0.330 0.947 0.29 Active Site of Molecule 1'

50
shape Of the peak is roughly Spherical and does not suggest
an orientation for the TCA anion. This peak does not have
a local two-fold equivalent in.MOlecule I', as the inter-
molecular binding site itself does not have a local two-
fold related counterpart. This behavior is similar to
that of the uranyl heavy atom derivative used in the deter—
mination of the 2.8 A. phases (binding Site U, Table 3,
Reference 41).

The two active Site substitution peaks are also approxi-
mately Spherical, but show a Slight tapering along the
positive direction of the a* axis (local two-fold direction).
Kendrew models of the TCA anion were constructed and placed

64 with the

into the difference peaks using a Richards' box
carboxyl groups in the tapered side of the peaks. The TCA
anions are approximately related by the local dyad both
with respect to location and with reSpect to orientation

of the anions. The TCA corresponding to Peak 2 in Molecule

I is oriented such that the CC1 group is about 3.0 A. from

3
the peptide bond joining Trp 215 and Gly 216 and the
carboxyl group is 2.0-3.0 A. from the main chain atoms of
21195

group. In Molecule 1', the TCA Of Peak 3 is oriented such

Cys 191 and Met 192 and 2.0-3.0 A. from the 804

that the CCl3 group is about 5.0 A. from the o-carbon atom
of Gly' 216 and the TCA carboxyl group is 2.0-3.0 A. from
both the sou2‘—195' and the y-OH of Ser' 195. The carboxyl
group is also about 3.0 A. from the main chain atoms at

Ser' 195. In both binding Sites, the TCA carboxyl groups

51
closely approach carbonyl groups of the protein. However,
both CHT molecules accomodate the TCA anions without any
apparent difficulty as the difference map shows no evidence
of movement by any part Of either CHT molecule.

The pKa Of TCA is about l (250 C.) SO that the carboxyl
group is charged at pH 3.6. The forces reSponsible for the
binding may involve either hydrogen bonding or ion-dipole
interactions between the TCA carboxyl group and the neighboring
protein carbonyl groups, and hydrogen bonding interactions
with the sulfate groups. Van der Waals interactions are
probably responsible for orienting the 0013 group near the

main chain atoms at Trp 215 and Ser 216.

2. The 0.01 M, Trichloroacetate Difference Diagonal Plot.
The representation of the difference density on a
diagonal plot, or the difference diagonal plot (hereafter
abbreviated as DDP), is a two-dimensional representation of
the changes occurring in the three—dimensional difference
map and is an extension of the concept of the distance
diagonal plot described in Chapter II. The DDP wasconceived
and originally used by Dr. M. N. Liebman“6 of our laboratory.
The DDP is an.NxN matrix. whose elements r.. are the moduli

+ 13
of the vectors rij which are

-> + + ->

I'ij : (p " C(11) + (p - Coj)

-r ->

1
where °°i’ Cdj, and p are the coordinate vectors Of the
i—th and the j-th alpha-carbon atoms and a difference map

peak, respectively, in an orthogonal coordinate system.

52
The number N designates the amino acid residue. For a
difference map containing m peaks, there will be m vectors

+ +

p and m vectors rij for each ij-pair Of a-carbon.atoms.

In our DDP program only the smallest Of the moduli Of the

+
m vectors r.. is recorded. The plot thus calculated may

1

then be contiured at convenient distance intervals. The
features which appear on the DDP represent the close approach
(below an arbitrary limit) of a difference peak to a given
pair of a-carbon atoms, and are a unique two-dimensional
representation of the three-dimensional difference density

.)

associated with the molecule. Because the vectors r.lj and
+

rji are Of equal length, only one-half Of the NxN matrix is
unique. Therefore, the full matrix may be used to represent
changes occurring in both CHT molecules Of the dimer. In
all the DDP's calculated in our work, the area below the
diagonal is devoted to Molecule I and the area above the
diagonal to Molecule I'. The o-carbon coordinates Of
Molecule 1' were obtained by performing a two—fold rotation
about the local dyad A Of Figure 3a on the a-carbon atom
coordinates of Molecule I.

Since the TCA difference map was relatively simple,
the DDP of TCA is correspondingly simple and is shown in

Figure 12. The TCA DDP is normally contoured at 4.5, 6.0,
and 9.0 A. levels, but only the 9.0 A. features are shown
in Figure 12. The vertical and horizontal lines dividing
the plot into quarters are drawn in at residue 123 and

denote the folding domains A (residues 1—122) and B

53

 

 

 

 

 

0 ‘90 1 .
O l l
I
100 -
4' \\\
l \
I
- I I ..
\\ I, \\
\\~"/ I, \
I
I I
. o I l
0 I I I
I I
. I'I
\\ I]
2CK¥
Figure 12. The difference diagonal plot of the 0.01 M,

TCA derivative.

 

54

(residues 123-245) Of CHT. Features in the DDP which are
reflected across the diagonal represent changes occurring
in both CHT molecules at the same amino acid residues. DDP
features not having a diagonal-mirror plane equivalent
represent changes occurring in one CHT molecule but not the
other, and are usually related to the variability in the
native structure. The TCA DDP contains examples of both
kinds of features.

The two difference peaks in the active site regions of
Molecules I and I' are the dominant features in the DDP
and appear in the lower-right quarters of the plot. The
two features on both sides of the diagonal at residues 191
to 196 represent the close approach of both TCA difference
peaks to the d—carbon atoms of the polypeptide segment
Cys 191-Met 192-Gly 193-Asp l94-Ser 195-Gly 196. The two
features are not exact mirror images since neither the CHT
molecules nor the two difference peaks are exactly alike.
However, the similar binding behavior of the two TCA
molecules gives rise to two very similar features in the DDP.
The two Off-diagonal features nearby also Show an approxi-
mate symmetry with respect to the diagonal. Thus, similar
structural changes are occurring in the two CHT molecules.

The TCA substitution on the surface of Molecule I,
Peak 1, creates the group of features in the diagonal plot
enclosed by broken lines. These features are contoured
at 12.0 A. level. Peak 1 is located in the secondary

binding site Observed for the phenylalanine-CHT derivative,

55

and the pattern of peaks observed for this TCA substitution
is very Similar to those created by the Phe secondary
substitution in the Phe-CHT DDP. Clearly, the DDP provides
a convenient method for comparing or contrasting the
binding behavior of small molecule substituents.

The DDP provides a quick and concise means for
diSplaying changes observed in a difference Fourier. As
will be seen with CHT, the DDP dramatically demonstrates
structural variability in CHT derivatives. In addition,
it may be used to estimate the location and extent of
changes occurring in the derivatives. One difference peak
near several a-carbon atoms, or a group of peaks near several
d-carbon atoms, will result in more extended DDP features
than those arising from a single peak near a small number Of
a-carbon atoms. Examples of both extended and localized

interactions are found in the DDP of TCA.

Chapter VI.

The 2.0 M, Guanidine Derivative.
1. The Difference Fourier Map.

Approximately 60 peaks with heights 3 0.20 e.A.-3 (40)
were observed in the 2.0 M, GdnHCl difference map calcu-
lated over the CHT dimer and the surrounding solvent regions.
Of these, about 30 peaks (both positive and negative) were
located in or near the two CHT molecules 4.5 A. or less
from an d-carbon atom as determined by a DDP calculation.
The remaining peaks were located entirely in the inter-
molecular contact regions shared with neighboring CHT mole-
cules: no significant density was Observed in the solvent
regions. The latter further enhanced the credibility Of
the map. The Observed changes occurred in two regionS--the
dimer interface region and on the surface Of the dimer.

Most Of the peaks were small in volume and occurred singly
at scattered locations. The interiors Of the molecules
were essentially unchanged from the native structure.

Due to the nature Of the changes displayed in the
difference map, a detailed interpretation Of many Of the
peaks was impossible. There were, however, several examples
Of extended density with relatively large peak heights
(0.2—0.3 e.A.'3) in both the dimer interface region and on
the surface Of the dimer. These were examined in some detail.

The observed r.m.s. of the difference density of the

2.0 M, GdnHCl map calculated over the entire asymmetric unit

56

57
is 0.05 e.A.’3. Only those peaks with heights 3 0.20 e.A.’3

(40) were examined.

2. The Difference Diagonal Plot.

The DDP calculated for all the difference peaks (63)
in the asymmetric unit with heights 1 0.20 e.A."3 is shown
in Figure 13. The DDP has been contoured only at the 9.0
A. level as before. Since the GdnHCl derivative is more
complex than the TCA derivative, its DDP is correspondingly
more complicated than that of TCA. However, several quali-
tative Observations may be made immediately. The Observed
changes are wide-spread with reSpect to the amino acid
sequence over both molecules in the dimer. Further, the
changes occur almost entirely within one domain or the other
since there are very few features in the lower-left and the
upper-right quarters of the plot which would indicate changes
involving amino acid residues of both domains. There are
also several DDP features without a diagonal-mirror counter-
part, reflecting asymmetric response and the structural
variability between the independent CHT molecules.

The DDP can give an indication of whether key amino
acid residues were affected by derivative formation. Some
key residues include those involved in catalytic action,
dimer formation, and other residues which proved to be
prominent in earlier derivative studies. The 2.0 M, GdnHCl
DDP contains numerous examples of structural changes near
key residues. Difference peaks near His 40 and His' 40

gave rise to a set of DDP features in Figure 13 labelled

\‘n

200

 

 

 

 

200-

 

 

I"'""-“
l'

L-----....._.J

 

Figure 13. The difference diagonal plot Of the 2.0 M,

GdnHCl derivative.

 

59
”1". The aromatic substrate binding site is partly formed
by two polypeptide segments located near one another-—Ser
190-Cys 191-Met 192—Gly l93—Asp 194 and Trp 215-Gly 216-
Ser 217. A characteristic set of features appear in the
DDP for difference peaks occurring either in the substrate
binding site (as in the TCA derivative) or in the dimer
interface region near the substrate binding Site (as in the
GdnHCl derivative). These are the features in the rectangular
array in Figure 13 labelled "2". There are other character—
istic arrays which occur for difference peaks located in
other parts of the CHT molecule.

The qualitative results of the DDP are readily-obtained
and may be useful in characterizing, for instance, the
members of a set of related CHT derivatives.°°’°5 However,
upon closer examination Of the DDP and comparison with the
difference Fourier, it is apparent that extensive DDP
features do not always represent the most significant
features in the difference map, and conversely, the most
significant features in the difference map dO not always
create Spectacular and extended features in the DDP. One
of the most extensive DDP features in GdnHCl involves the
C-terminal a-helix of Molecule 1, residues 234 to 245 (see
Figure 13). The features result from two small difference
peakS—-approximate1y 1.0 A. in diameter at the 0.2 e.A.-3
contour--situated inside the o-helix. These two peaks_do
not comprise either significant or extended difference

density features. Also, several very significant difference

6O

peaks Occurring on the surface Of Molecule I in the uranyl
binding Site region“1 give rise to only a single point in
the DDP. Thus, the DDP does not always represent the
Significance Of difference peaks accurately.

This problem is partly alleviated by the use Of the
difference diagonal multiplicity plot (DDMP). This is an
NxN matrix whose elements mij are the number Of difference

peaks for whom the inequality, r..

13 i k, where k 1s an

arbitrary limit, is satisfied for each ij-pair Of a-carbon

atoms. The quantities r. are calculated for the DDP as

13'
described in Chapter V. For each element rij which appears

in the DDP, there is a corresponding element mij in the

DDMP. For the 2.0 M, GdnHCl DDP, the limit of rij is 9.0
A. and each correSponding mij of the DDMP is the number of
difference peaks whose distances to the i-th and the j-th
d-carbon atoms add up to 9.0 A. or less. The number of
difference peaks close to the ij-pair of d-carbons (below
an arbitrary limit)--a multiplicity--gives an additional
measure of the extent Of change. The DDMP for the 2.0 M,
GdnHCl derivative is Shown in Figure 14, where only those
elements mij 3 2 have been plotted.

Comparing the GdnHCl DDMP to the DDP reveals two sets
of DDP features containing points with peak multiplicities
Of 2 or more. Parts of four of the features in group "2"
in the DDP (Figure 13) are due to groups Of two or three

peaks. The identity and map coordinates of each peak

involved is listed along with the DDMP so that features in

 

 

 

20

 

 

61

 

 

O 100 200
O 1 l 1 l l l 1 1
100 -
-I
200 -
-I

 

 

 

 

Figure 14. The difference diagonal multiplicity plot Of the
2.0 M, GdnHCl derivative.

62
the DDP may be cross-referenced with peaks in the difference
map. The peaks giving rise to multiplicities Of 2 or
more do in fact constitute the most significant set Of
changes in the dimer interface region. The only other
feature in the DDMP occurs on the diagonal at residues
Ser' 115 to Thr' 117 indicating that significant changes
have Occurred around this polypeptide segment which is
located on the surface of Moleule I'.

The DDMP does not solve the problem Of significant
difference density occurring on the surface of the protein
molecule which fails to contribute significantly to the DDP
features. By raising the limit Of rij' these surface peaks
would contribute more heavily, but the DDP features already
present would become much more enhanced with the concomi-
tant loss of some Of the discriminating ability of the DDP.
The 9.0 A. limit is an Optimal value detennined by calculating
DDP'S at various limits. Thus, significant density

occurring on the surface Of the CHT molecule may be

unavoidably underweighted by the DDP.

3. Changes in the Dimer Interface Region.

The dimer interface region contains a number Of
strained close contacts and is very sensitive to changes
in the solvent environment.3°'67 This region is also
notable for numerous features in the native electron density
which do not correspond to protein groups and have been
attributed to ordered solvent molecules and ions. The

locations of four sulfate ions in the dimer interface region

63
have been unambiguously determined by a SOuZ":SeOu2' exchange
eXperiment.66 The unit cell coordinates and peak heights
of the most significant difference peaks in the dimer
interface region are listed in Table 3.

Peak 1 is about 3.5 A. in length (at the 0.20 e.A.‘3
contour) extending from a point 1.0 A. from the sulfur atom
Of'Met 192, through the local two-fold axis, to a point 1.0
A. from the sulfur atom of Met' 192. Peak 2 is 3.0 A. from
Peak 1 and 1.0 A. from SOu2--l95'. There is no negative
density occurring on the native density at the methionine
side chains, so there is no direct evidence that the side
chains of either Met 192 or Met' 192 have moved. The two
methionine sulfurs are 4.0 A. apart in the native dimer and
are connected in the native electron density at the 0.5
e.A."3 contour, and it is in this density that Peak 1 is
located. There is a cresent-shaped positive difference
peak 2.0 A. from Peak 1 in the negative x-direction ranging
in height from 0.16 to 0.20 e.A.-3. This peak extends in
an are 6.0 A. in length (at the 0.16 e.A.-3 contour) from
a point 2.0 A. from the sulfur Of Met' 192, passing within
1.0 to 2.0 A. of the sulfur of Met 192, and ending at a
point 2.0 A. from SOu2"-195'. The total picture presented
by Peaks 1 and 2 and the cresent-shaped positive peak is
that of a reorganization of the solvent structure around the
methionine Side chains and SOu2'-l95'. The native density
joining the methionine side chains is probably due to close

contact of the sulfur atoms, and the location Of Peak 1

64

 

 

 

oaoaaoasu o: .nam sous .s o.H mm.o- mmm.o omm.o Hmo.o a

oaoaaoaea as was sous .< o.H nm.o- mms.o scm.o mos.o o

oooesoesa sm.o+ mos.o mom.o mms.o m

Hosea ca aea>ao eso>aon ca sacs cannon mm.o+ mam.o mm~.o oss.o a

nae .soq so cameo oosn sous .< o.m sm.o- Nos.o mam.o ans.o m

.moa--msom sous .< o.H om.o- mmm.o mom.o mao.o N.

unease ooan mos .eos one was so: goes .< o.a sm.o- acs.o mom.o mmo.o H
nesoesoo an-.<.o0esmaum m M m sacs

.coawom oommhoch Hmswn one
ca memos oososoeeum esaoaeasmem enos as» I o>aeo>anom Humseo «a o.m are .m ounce

65

suggests that these side chains may have been slightly
disturbed.

Peak 3 is an oval peak 3.0 A. in length partly occupying
a segment of native density which has been tentatively
assigned to a counterion for SOuz'-l95'. It is about 3.0 A.
from the side chain Of Leu' 143, 2.0 A. from the side chain
of Met 192, and 2.0 A. from both Peaks 4 and 5. Peaks 4
and 5 are positive peaks whose centers are 2.0 A. apart but

3

are joined at the 0.20 e.A.' contour to form an extended
segment of positive density whose projection in the yz-plane
has the shape of an inverted "V", spanning a distance of
about 5.0 A. at its widest part, and forming one of the most
interesting features in the 2.0 M, GdnHCl difference map.
Peaks 4 and 5 are shown in Figure 15. Peaks 4 and 5 are
located in a solvent cavity in the dimer interface which is
bounded by the side chains Of Phe' 39, Thr 151, and Leu 143,
and by the main chain segments His 40-Phe 41, Met 192-Gly 193,
and His' 40-Phe' 4l-Cys' 42. Peak 4 is an oval-shaped peak
3.0 A. in length, 3.0 A. from the peptide group joining

Met 192 and Gly 193, and 3.0 A. from the main chain at

His 40. Peak 5 is also 3.0 A. in length and is located
21195'

counterion and about 3.0 A. from the main chain atoms at

2.0 A. from the native peak attributed to the SOn

Cys' 42. A detailed interpretation of the changes represented
by Peaks 3, 4, and 5 is difficult. Since none of the
surrounding protein groups appear to have been perturbed,

these peaks must represent changes in the solvent. Either

66

GILJI42 C 5’42-C 5’58
Trp I41 Leu143 Ploe’cIIL:J Oisul%de

       

 

Cgs42-Cgs 58 504-495 Peaks 4 6. 5
O'Isulf‘ide Phe 41 Counterion

Figure 15. A feature of the 2.0 M, GdnHCl difference
map located in the dimer interface region.

-67
Peak 4 or 5 could represent a guanidinium ion (hereafter
abbreviated Gdn+) as there is adequate room for the ion
and numerous hydrogen bonding possibilities are available
near either peak: however, little information is available
from the peak's shapes to suggest an orientation for a Gdn+
ion.

Peaks 6 and 7 represent changes in the imidazole side
chains of His 40 and His' 40, respectively. Peak 6 is
partly superimposed in the His 40 imidazole density, 1.0 A.
from the N61 atom. Peak 7 is 1.0 A. from the B-carbon atom
of His' 40. There is no significant accompanying positive
difference density for either negative peak. Histidine
imidazolium ions have a pKa of about 6.5 so that at pH
3.6, the imidazole groups both bear a positive charge. The
difference peaks may represent some disruption of the solvent
structure around the charged side chains. Alternatively,
and more probably, the difference peaks represent small
movements by the imidazole side chains away from the inter-
face region and toward the interiors of their respective
CHT molecules. If either of Peaks 4 or 5 are bound Gdn+
ions the movement of the histidine Side chains would be in
response to the close approach of a like charge--Peak 4 is
5.0 to 6.0 A. from the His 40 side chain and Peak 5 is
about 5.0 A. from the imidazole of His' 40.

Another group of changes were observed in the dimer
interface region. There are several small difference peaks

(<2.0 A.in diameter at the 0.20 e.A.'3 contour) occurring

68
in and near the native electron density of the polypeptide
segment Gly 216-Ser 217-Ser 218 in.Molecule I and the
corresponding segment of Molecule I'. These polypeptide
segments display marked deviations from local two-fold

42

symmetry in the native structure. This region also
contains 6 hydroxyl groups from various side chains in a
volume about 6.0 A. in diameter. It may be Significant that
a region of the dimer which is crowded and stained, and which
may have a complicated solvent structure because of the
hydroxyl groups, should be affected by a chemical denaturant.
A DDP calculated for the difference peaks in the dimer
interface region only is Shown in Figure 16, contoured at
the 9.0 A. level. The features created by the sane difference
peak, or by the same group of peaks, have been grouped
together by broken lines. Groups 1 and 2 represent the
changes occurring near His 40 (Peak 6) and His' 40 (Peak 7),
respectively. Peaks 1, 2, 3, 4, and 5 cannot be separated
because they all occur close together and near a number of
the same a-carbon atoms. The features labelled "3" represent
the close approach of Peaks 1 and 3 to the d-carbon atoms
of'Met 192, Gly 193, Ser 217, and Ser 218. The group
labelled ”4" arises from the close approach of Peaks 1,
2, 3, 4, and 5 to the polypeptide segments Cys 191-Met 192-
Gly 193 and Trp 215-Gly 216—Ser 217-Ser 218. From this DDP,
it is possible to isolate the contribution of difference
peaks in the dimer interface region. The characteristic

arrays of DDP features described in Section 2 are also

69

 

 

 

(3 K1) 2CK)
O 1 1 1 1 j l 1 1
I- (V) I'I
I I 2 II
-I I I II
II' 'I
I ' II
I II
I I I I
'I 5J1 II
II
'I
100« II
II
I
I
'I
I
,I
_ I
II
I:
I\
-I I \\
I \\
\
4 l.‘
200‘ I I
' I
I3 I
. \:____

 

 

 

 

Figure 16. The difference diagonal plot of the 2.0 M,
GdnHCl derivative, calculated only for the difference
peaks in the dimer interface region. Distances ri'
calculated for each map point 3 0.2 e.A.'3 within each

peak.

7O
evident here. Finally, the changes Show an approximate
local two-fold symmetry, but it is clear that the difference
peaks in the dimer interface region are closer to the main

chain atoms of Molecule IL The full implication of this

result is not clear at this time.

4. Changes on the Dimer Surface.

Most of the difference peaks observed on the surface
of the CHT dimer with peak heights of 0.20 e.A.'3 or more
are small in volume and have no significant accompanying
difference density of the opposite sign. These probably
represent small reorientations of side chains, changes in
the solvent structure surrounding a group of atoms, or
small movements by main chain atoms, and are not inter-
pretable in more detail. There are, however, several
examples of more extended difference density peaks. The
most significant difference features appearing on the dimer
surface are listed in Table 4.

Peaks 4, 5, and 6 are two-fold screw equivalent peaks
of l, 2, and 3, respectively, occurring in two intermolecular
contact regions which are two-fold screw related. Peak 1,
about 4.0 A. in length, is located 2.0 A. from the carboxyl
group of Glu 21 and about 3.5 A. from the Gdn+ group of
Arg 154. Peak 4, identical to 1 in Size and shape, occurs
2.0 A. from the carboxyl group of Asp' 153 and 5.0 A.
from the carboxyamide group between Gly' 74 and Ser' 75.
Peak 2 is less than 2.0 A. wide at the 0.20 e.A.-3 contour,
but is a ”footprint"-shaped peak 5.0 A. from heel to toe

1
7

AAHIImsom ea Seansoo o>aeas one 2H

sma .mus mo cacao mean +soo sous .< o.H
mzothINoo and .Qm< Scam .< 0.: .pCm>Hom CH
ssoamIImoo mud .sns sous .s o.m

mSwH o:woaaonhmn :N ohm

Scum .< 0.: ”mango moan :ma wp< Scum .< o.m
o.H

.¢

cwmno ovwm +cou :ma wh< sohm
saoam- moo Hm ado scam .< o.~

BN.OI
0N.0I
3N.0I
mm.o+

0N.0I
:N. 0'
nm.o+

 

mesossoo

one so named oososoeean esooaeaswam sec: on» I o>aea>eson Houses

AMI.<.mVP£mem

Hmm.o
mow.o
mum.o
moo.o

mmm.o
mms.o
mam.o

N

Hm:.o
oom.o
mmm.o
mam.o

ooo.o

mmo.o

aao.o
m

.2 o.N one

mo:.o
amm.o
mno.o
nno.o

ooo.o

sem.o

ssm.o
_m

.so2an emu one so codessm

:rmxorx

 

good

.s canoe

72
at the 0.16 e.A."3 contour. It is located on the edge of,
and along the entire length of, the native electron density
of the Arg 154 side chain. Peak 5, identical to Peak 2,
is located such that its peak is about 4.0 A. from the
Asp' 153 carboxyl group with the rest of the peak extending
away from the surface of Molecule I' toward a neighboring
CHT molecule. Peak 3, which is approximately 2.0 A. in
length, is located in the electron density of a neighboring
CHT molecule, 9.0 A. from the Arg 154 Gdn+ group and 7.0 A.
from the pyrrolidine range of Pro 24. Peak 6, identical to
Peak 3, is situated in the native electron density of the
Arg' 154 side chain.

This intermolecular binding site has been prominent in
previous work. The point of highest electron density in
Peak 1 is 1.0 A. from the location of the dominant inter-
molecular binding site of the uranyl heavy atom derivative
(site ”U" in Table 3, Reference 41) which was notable for
its high occupancy and its lack of a local two-fold equivalent.
The structure of this binding site has been described in

some detail previously.38 In the uranyl derivative, the

2+
2

Molecule I and Asp 153 of the neighboring CHT molecule

U0 ion complexes with the carboxylate groups of Glu 21 Of

(itself the two-fold screw equivalent of Molecule I').

Only one such binding Site exists per dimer because the

local two-fold related region is incapable of binding an

ion because the CHT molecules are farther apart. Interestingly,

Peaks 4, 5, and 6 are located only about 5.0 A. from the

73
positions they would have occupied had they been related to
Peaks 1, 2, and 3 by the local two—fold rotation.

The uranyl binding Site is also noted as one of the
major areas in which changes in structure have been observed
with change in pH.3°’52 Apparently, as the pH is raised
from 3.6 to 5.4, the ion—pair consisting of the Arg 154
Gdn+ ion and a sulfate ion is disrupted and a new ion-pair
is formed involving the Arg 154 Gdn+ ion and the carboxyl
group of Glu 21. Additional difference map features for the
pH 5.4 derivative create a complicated picture of changes
in both the protein and solvent structures in this region.

The locations of Peaks 1 through 6 afford the convenient
identification of segments of the native density belonging
to the nearest neighbor CHT molecule which contribute to
the formation of this intermolecular binding site. Because
Peak 6 is near the Arg' 154 side chain, Peak 3 must also
be near an Arg 154 side chain on the surface of the CHT
molecule which is the nearest neighbor to Molecule I in
the negative y-direction. A Kendrew model of the polypeptide
segment Asp 153-Arg 154-Len l55-Gln 156 was constructed and
fitted into the native density using a Richards' box.°°
The result gives the structure Of this binding site -
a cavity 6.0 to 8.0 A. in diameter defined by the side
chains of two Arg 154 residues and by the carboxyl side
chains Of Glu 21 of Molecule I and Asp 153 of flueneighboring
CHT molecule. The Arg 154 Gdn+ groups are about 6.0 A.
apart and the Glu 21 carboxyl group is 6.0 A. from the
carboxyl group of ASp 153. The cavity thus formed is Open

to the solvent on two sides.

74

The picture which emerges from the disposition of the
Gdn.HCl difference peaks, the native structure of the binding
site, its past history as an effective ligand of large
positive ions, and its sensitivity to solvent changes, is
that a Gdn+ ion is binding in the uranyl binding site.

Peaks 1 and 4 probably represent the Gdn+ ions: a model of
the Gdn+ ion is readily accommodated by Peak 1 and is located
at reasonable distances from the nearby ASp' 153, Arg' 154,
Glu 21, and Arg 154 side chains. Peaks 2 and 5 represent

the movement of the positively-charged Arg 154 Gdn+ group

of Molecule I away from the bound Gdn+ ion in the solvent,
and toward the sulfate ion with which it is presumed to

form an ion-pair in the native structure.38 Peaks 3 and 6
represent small movements of the Arg' 154 side chain on
Molecule I'. The Gdn+ ion so situated is between the carboxyl
groups of Glu 21 (Molecule I) and Asp' 153 (the neighboring
CHT molecule which is the two-fold screw equivalent of
Molecule I'), and approximately half-way between them

forming a kind of intermolecular salt bridge. Peak 7 is
located squarely on the native density which has been shown
to be a sulfate ion located 2.5 A. from the terminal amino
group of the Lys 177 side chain (SOu2'-177, Reference 66).
This peak shows simply that the sulfate ion has been removed
from its position in the native structure. Since there is

no positive density nearby, a shift to a new position is

not indicated.

Chapter VII.

The 3.0 M, Urea Derivative.
l. The Difference Electron Density Map.

The difference Fourier map calculated over the CHT
‘dimer and the surrounding solvent regions contained a
larger number of peaks than was observed in the GdnHCl
difference map. Approximately 160 peaks with heights
1 0.2 e.A.-3 (3.00) were Observed in the map, of which
approximately 120 peaks occurred at a distance of 4.5 A. or
less from an o-carbon atom. Numerous changes occurred on
the dimer surface, in the dimer interface region, but unlike
the GdnHCl derivative, changes also occurred in the hydro-
phobic interior of both CHT molecules. NO significant
difference density was observed in the solvent, but numerous
peaks were observed in the intermolecular contact regions
shared by Molecules I and I' with the neighboring CHT
molecules. Though widely-distributed, the difference
density was not randomly-distributed, occurring in specific
regions of the dimer. The urea difference map contains
fewer examples of extended, individual peaks like those
Observed in the GdnHCl map. The urea derivative difference
peaks tended to be smaller in volume and clustered together
at various locations in the molecule. Like the GdnHCl
difference map, a detailed interpretation of many of the
peaks is not possible, but the location and distribution of

the peaks is significant and they will be discussed by
groups.

75

76

Groups of peaks were Observed in the following regions:
the dimer surface, the dimer interface region, the A-Chain/
B-Chain contact region, and in two areas of the nonpolar
interdomain region. The changes occurring in the interiors
of the CHT molecules were studied in detail for Molecule I
only.

The observed r.m.s. do for the difference map is
0.07 e.A.‘3 and only peaks with heights 3 0.20 e.A."3

were examined and input to the DDP calculation.

2. The Difference Diagonal Plot.

The DDP forthe 3.0 M, urea derivative is shown in
Figure 17: the correSponding DDMP is shown in Figure 18.
The features appearing in the DDP correspond to difference

rij : 9.0 A. The features appearing in the

peaks for which
DDMP correspond only to those elements rij of the DDP for
which 3 or more difference peaks satisfy the 9.0 A. limit.
The 3.0 M, urea DDP is considerably more complicated
than the GdnHCl DDP. Changes have occurred throughout both
folding domains of both CHT molecules. Also, there are
considerably more changes occurring in the interdomain
regions of both molecules (lower-left and upper-right
quarters of the DDP) for the urea derivative. Generally,
most plot feature which occur in one CHT molecule are
reflected across the diagonal, indicating that the deviations

from local two-fold symmetry in the difference map are more

a matter of degree than kind. The difference peaks are

77

 

 

 

 

r_________.___.___.

 

3‘1”“

' ——_-—

 

 

 

 

 

     

  

 

.//
/
/;' 4

C

 

 

 

  

200‘

II I
' 0-

The difference diagonal plot

 

Figure 17.

.o m.

the 3

of

Urea derivative.

'78

 

.,
I
'I
I
I
I
I
I
I
I
I
I
I
2I
I
I
I
I
I
I
I

F”1

I o
u I

IIII IIIIII.I_ 3

/
/
4

l__u__".__

 

 

fir-———---
c—’
I
I
I
I
I
I
I
I
I
I
I
I
i
I
I
I
l

T- -— ---A>- -

 

IIIIIIIIIL

2

l______._".____"_

I

 

agonal multiplici

 

 

200-

y plot of the

J.
U

0
J.

The difference d
Urea derivative.

Figure 18.
.0 M,

3

79
closer to more pairs of 0-carbon.atoms in Molecule I' than in
iMOlecule I: the number of rij §_9.0 A. is 538 forIMolecule
1' versus 369 for Molecule I. There is also a larger
number of points above the diagonal (29) than below the
diagonal (17) in the DDMP. These two results taken together
indicate that the structure Of Molecule 1' is substantially
more perturbed by the urea than.Molecule I.

The DDP contains some indications of disruption of
secondary structure in both molecules. There are several
plot features, perpendicular to, and extending away from
the diagonal, on both sides of the diagonal Which represent
difference density occurring near strands of anti—parallel
B-sheet. There is some difference density near the C-
terminal o-helix of Molecule I', but it may not represent
significant changes Since there is no feature in this region
of the DDMP.

A most useful result obtainable from the DDP is an
indication of those regions of the CHT molecule where
difference peaks are concentrated. From the discussionsof
the TCA and GdnHCl derivative DDP's, difference peaks
occurring in particular locations of the molecule may create
distinct patterns of features. Examples include the
rectangular array of features Observed in both the TCA and
the GdnHCl DDP'S created by difference peaks in or near the
substrate binding site, and other combinations Of vertical
and horizontal bands of DDP features occurring for difference

peaks located in the interiors of the CHT molecules. However,

80
large features (or groups of features) may be created by
small difference peaks so that the multiplicities of DDP
features should also be considered. Therefore, to represent
significant and extended difference density peaks, DDP
features should satisfy two criteria: (1) the DDP feature
should have a correSponding feature in the DDMP (subject to
arbitrary limits appropriate for the derivative) and, (2) the
I DDP feature Should extend over a number of amino acid
residues, or even better, be one Of several DDP features on
approximately the same residue number ordinate or abscissa.
In this way, one might "survey" a complex difference map
to determine those regions of the CHT molecule in which the
most significant changes might be expected to occur.

The 3.0 M, urea DDP and DDMP contain indications that
there are several areas of extended difference density in
both CHT molecules. In Molecule I of the DDMP, two features
at (103, 55) and at (103-104, 102-103), where the coordinates
are eXpressed as (row, column), correSpond to two features
in the DDP at the same positions. The first is an off-diagonal
feature and the second is an extended feature on the diagonal.
These two features, with two others nearby, join the L-
shaped group labelled "15 in Figure 17. This group of DDP
features was created by a group of peaks concentrated in
and around a segment of anti-parallel B-sheet composed of
four polypeptide segments - Gly 44-Ser 45-Leu 46, Trp 51-Va1
52-Va1 53-Thr 54, Thr 104-Leu 105-Leu lO6-Lys 107-Leu 108-Ser
109 and Ile 85-Ala 86-Lys 87-Val 88-Phe 89-Lys 90. These

81

difference peaks make up one of the two principal changes
observed in the hydrOphObic core of the CHT molecule.

Four DDMP features at (118, 25), (118-119, 28),
(118, 69-70), and (118-120, 115-119), all clustered in the
abscissa interval 118-120 in Figure 18, lead to another
L-shaped group of features in the DDP labelled "2" in Figure
17. This group of features indicates that changes have
occurred involving the amino acid residue segments
Cys l-Pro 8, Val 23-G1n 30, and Gln 116-Cys 122. Examination
of the difference map revealed that these DDP features are
due to a group of difference peaks (including an example of
one of the most extended difference peaks in the map)
occurring in the A-Chain/B-Chain contact region.

Two smaller clusters of DDP features labelled "3" and
"4" in Figure 17 have been suggested by DDMP features
occurring at (162, 131) and (215, 214) in Figure 18,
respectively. Cluster 3 arises from a number of difference
peaks occurring in and around a segment of anti-parallel
B-sheet partly made up of the polypeptide sgements Leu 155-
Leu 160 and Val l37-Gly 140. These difference peaks
constitute the other major group of changes observed in the
nonpolar interdomain region. Cluster 4 arises from a small
number of difference peaks in the neighborhood of the
polypeptide segments Try 215—Ser 217 and Gly 226-Tyr 228,
and also near the aromatic cluster Tyr 171, Trp 172, and
Trp 215. The group of difference peaks involved in Cluster
4 are less extensive than those grbups of peaks reSponsible

for Clusters l, 2, and 3.

82

The features in the Molecule I' portion of the DDMP
have been grouped together in three clusters, 1', 2', and
3'. Cluster 1' consists of four closely-spaced DDMP features
at (30, 31). (31. 33). (31-34, 65-66), and (31,68). The
corresponding DDP features indicate changes have taken place
in a part of the A folding domain involving the polypeptide
segments Glu' 30-Asp' 35. Gly' 38-Ser' 45, and Val' 65-Glu' 70.
This portion of the A-domain borders the nonpolar inter-
domain region and is near (or includes) a number of nonpolar
and/or aromatic residues including Phe' 39, His' 40, Phe' 41,
and Trp' 141. Cluster 2' on the DDMP consists Of two features
at (20-22, 156-157) and (155-156, 156-157). The correSponding
Cluster on the DDP indicates that there are changes involving
the polypeptide segments Gly' l9-Val' 23, Cys' l36-Trp' 141,
and Leu' l55-Ala' 158. These polypeptide segments are
nearest neighbors and together form a small segment Of
twisted anti-parallel B-sheet bordering on the nonpolar
interdomain region.

The final group of Molecule I' DDMP features correSponding
to Cluster 3' in the DDP consists of seven separate features
including the largest single feature in the 3.0 M, urea
DDMP - a feature located at (231-235, 234-236). The
coordinates of the other six features are: (126, 233),

(182, 183). (180, 228), (182, 227), (228-229, 229-230),
and (231, 232). The correSponding DDP features in Figure 17
are quite extensive and indicate that extensive changes in

structure have probably occurred over a large volume of the

83
B-domain. The polypeptide segments especially implicated as
the sites of these changes are Leu' 123-Ala' 126, ASp' 178-
val' 188, Ile' 212-Ser' 218, and Cys' 220-Trp' 237. The
segment Leu' 123-Ala' 126 occurs on the surface of Molecule
I' adjacent to a portion of the Cys' 220-Trp' 237 segment.
The three segments ASp' 178-Val' 188, Ile' 212-Ser' 218,
and Cys'220-Trp' 237 are adjacent to one another and together
form a large segment of anti-parallel B-sheet, which is in
turn a substantial portion of the B-domain.

With a detailed examination of the Molecule 1' portion
of the DDP, another result emerges — a greater degree of
variability in local two-fold symmetry is now evident. The
polypeptide segments implicated by each cluster are listed

below and correspondences are indicated.

Molecule I Molecule 1'
Cluster Polypeptide Cluster Polypeptide
Segment Segment
1 Gly 44-Leu 46 lu' 30-ASp' 35

  
  
  

Trp 5l-Thr 54 Gly' 38-Ser' 45

Thr 104-Ser 109 Val' 65-Glu' 70

Ile 85-Lys 90 Gly' l9—Val'23

Cys' l36-Trp' 141
jéfiereu' 155-A1a' 158
Glu 116-G1u 30 3' Leu'-123-Ala' 126

3 Leu 155-Leu 160;;;7 Asp' l78-Val' 188

Val 137-G1y 140

Val 23—Glu 3O

””””’,.Ile' 212-Ser' 218
4 Trp 215—Ser 217"”””’,,vas' 220-Trp' 237

Gly 226-Tyr 228

84
Clearly, substantial changes have taken place in both
molecules which are not reflected in the other.

In conclusion, the DDP used in conjunction with the DDMP
gives a reliable estimate of the location and extent of
changes in structure represented by the peaks in a complicated
electron density map. The 3.0 M, urea DDP indicated correctly
the polypeptide segments where the difference peaks were
most numerous - 3 of the 4 clusters of DDP features correSpond
to extensive sets of difference peaks discussed in separate
subsequent sections of this chapter. The many small
features of the DDP which have not been discussed can be
categorized into two groups - features on or close to the
ordinate or abscissa of a feature in a cluster already
discussed, and features created by peaks isolated from other
peaks or occurring near only a small number of a-carbon
atoms. Peaks creating features in the first category may
be members of significant and extensive groups of peaks.
Significant difference peaks occurring in the intermolecular
contact regions may create DDP features in the second
category. In any case, the DDP does not contain all the
information of the difference map, but may serve as an aid
to the organization of the information contained in the

difference map.

3. Changes on the Dimer Surface.
There are a substantial number of difference peaks

occurring on the dimer surface and they appear to represent

85

mostly changes in the solvent structure, and small, localized
changes in the positions of main chain and side chain atoms.
Most of the peaks are small in extent (3 2.0 A. in diameter
at the 0.20 e.A.'3 contour) and were observed at scattered
locations over the surface.: Peaks which appeared to belong
to a group of peaks extending into the interior of the
molecule are grouped with those peaks and are not discussed
here. The unit cell coordinates and heights of the most
significant surface peaks are listed in Table 5.

Peak 1 indicates that urea, like GdnHCl, displaces the

2-
804

ion near Lys 177 from its position in the native
structure. There is no significant positive density nearby
to indicate that the sulfate ion has undergone a small
change in position - it has simply been removed. Peaks 2
and 3 represent a small shift in the main chain atoms at

the peptide link between Thr 174 and Lys 175 - no significant
difference density was observed in the native density of the
side chains of either residue. Peak 4, located just outside
the native density at the Tyr 171 phenol may represent a
change in the solvent structure rather than a motion of

the side chain.

Peaks 5 through 8 all Odour in a contact interface
region shared by Molecule I with a neighboring CHT molecule
(nearest neighbor in the negative y-direction). These
four peaks, and an eQual number of smaller peaks (<2.0 A.

across), were observed in a small solvent cavity formed

by the polypeptide segments Thr 135-Cys 136—Val 137 and

86

sous ao NHH was p<

mzohm HhCODMmO NHH ma< BOHH .< o.N

30H pom cam oma onm

£909 pm Swmno Cams SoHH .< o.m .Pso>aom :H
Sumac ocwm mma how on» 809m .d o.m

cacao moan mma use one .s o.m

“msohw moapmom ma zoAINH mac 809% .< o.m
cwmno moan and how

sons .< o.m .saoso moan nma use sous < o.n
namgo moan 00H pom scum

.< o.m .nEopa cacao sacs mma sac ache .< o.m
macaw Hononm dud Mme thm .< o.m

ssoam ooaesos nae nsSISsH has gone .« o.m
macaw mz ouaEdhxonumO mud man P¢
msHIINSom ep<

 

mesossoo

one no agnom mononommwn szowmwswwm Pmos one I o>wpm>whon mos: aw o.m one

mm.0I
0N.0

mm.o
0N.0I

0N.0

0N.0

um.o
mm.0:
0N.0
0N.0I
3N.0I

NNN.0
NHH.0

mHH.o
mam.o

msa.o

mam.o

mom.o
son.o
Hmm.o
na~.o
mam.o

 

~m-.<.ovemmHom N
.sosan emu one so comessm

omn.o
onm.o

NHN.0
mmo.o

000.0
500.0

mam.o
osa.o
onm.o
mun.o
Hms.o

w

oso.o
moo.o

003.0
Hum.o

o:n.o

m¢:.o

smm.o
omm.o
umm.o
smm.o
emm.o
_m

HH
0H

NOW-3"“

 

secs

.0 wands

87

:.....pm zPHmCoc Coppooao o>Hpmc one CH: msmos Cowpmooa hMHSOfiehmm m :....P<:*

mzomw Hhxonnmo mma .Qm< Scam .< o.n

msouw +ccu :ma mh<
809% .< o.N .mzohw Hhxonhmo mm Qm< sane .¢ o.m

 

upcoesoo

cm.o mmm.o

om.o was.o

 

AMI.<.oqemm«om N

 

aam.o mma.o ma
sHo.o mom.o NH
M w. room

oossaesoo .m ounce

88

Ala 158-Ser 159-Lou l60-Pro 161, and residues of the
neighboring CHT molecule. Peaks 5, 6, and 7 are all about
the same size, shape, and height, and may represent urea
molecules engaged in hydrogen bonding interactions with
groups on both proteins. Peak 8 is 2.0 A. from the hydroxyl
side chain of Ser 159 and 2.0 A. from the peptide group
joining Leu 10 and Ser 11 and may represent some reorganization
of solvent structure: there was no Significant difference
density observed in the nearby native electron density.
Peak 8 is also only several Angstroms from some of the
peaks occurring in the A-Chain/B-Chain contact region and
may be connected with the changes occurring in that region.

Peak 9 is shaped like a footprint, is about 3.0 A. in
length, and is one of a number of Similarly-shaped peaks
observed throughout the map. It is located in a crevice on
the surface defined by the polypeptide segments Ser 127-Asp
128-Asp 129-Phe l30-Ala 131 and Leu l63-Ser 164-Asn 165
and is about 2.0 A. from the carbonyl of Phe 130. No
Significant difference density was Observed in any of the
surrounding native density. Thus, peak 9 may represent a
urea molecule hydrogen bonded to the Phe 130 carbonyl.

Peaks 10 and 11 form a gradient at the main chain
atoms of Ala 112, clearly indicating that the polypeptide
backbone at this residue has moved in the direction of the
solvent.

Finally, one peak was observed in each of the uranyl

binding Sites discussed-in Chapter VI. They are significant

89
features and are listed as Peaks 12 and 13 in Table 5.
Peak 13 is the two-fold screw equivalent of Peak 12, and
both peaks are about 3.0 A. from the GdnHCl Peaks 1 and 4
(Chapter VI, Section h), respectively. There is no
accompanying negative difference density at either site so
that Peaks 12 and 13 may represent urea molecules hydrogen

bonded to the Glu 21 and/or the Arg 154 side chains.

4. Changes in the Dimer Interface Region.

This region of the difference map contains a substantial
number of difference peaks («30) with absolute heights 1
0.2 e.A.'3 There are no examples of single, extended
difference features in this region, but rather groups of
small peaks (~2.0 A. across at the 0.2 e.A."3 contour)
were observed at several locations in the interface.

Several of these groups of peaks appeared to represent
significant changes in structure in both Molecules I and I'.
These peaks are listed in Table 6.

One group of peaks occurred in and around the native
density of the polypeptide segment Gly' 216-Ser' 217-Ser' 218-
Thr' 219-Cys' 220-Ser' 221-Thr' 222-Ser' 223-Thr' 224.

This segment contains seven hydroxyl groups within a
relatively small volume and portions of it display large
deviations from local two-fold symmetry.“2 The unit cell
coordinates and heights of the most significant peaks in
this region are Peaks 1-11 in Table 6. Interestingly, none
of these peaks have a local dyad equivalent in Molecule I -

there are no significant features in the difference map at

9O

 

 

 

33mm spfimcmc m>apac cofiompczoo .mmfl--m3om p3 mm.o- mam.o mmm.o_ Ham.o 3H
cause mafia mma pm: sash .3 o.H o~.o “33.0 3om.o Hmo.o ma
.mma--m3om

Scum .3 o.m «asoum Hammad 63H use gang .3 o.m mm.o- Nam.o mom.o mHo.o NH
macaw asconnmo cam .sao scum .3 o.m m~.o mom.o mom.o mmm.o HH
omm .mao mo agopm gamma came 93 m~.o- «03.0 m~3.o oom.o 0H

macaw Haconnmo mam .pmm p3 3m.o- Hm3.o mnm.o mam.o a

cam .aao mo macaw cause came gone .3 o.H w~.o mmm.o o3m.o oom.o m
:Hmno mafia mmm .nna scum .3 o.H mm.o- o3m.o 3Hn.o mm3.o a

Cause muwm mma .hmm P¢ mm.o mmm.o .m::.o 353.0 0

cause moflm mmm .uae gone .3 o.H 3N.o 063.0 oom.o H03.o m

muonw Hacopnmo Hmm .pmm p3 5N.o- nm3.o mn3.o 533.0 3

aopm 50 3mm .naa scam .< o.H nm.o Hmm.o mom.o 3m3.c m

sop“ do 3mm .pa» 93 -.o- mon.o 3n3.o Hm3.o N

macaw moapamd mam .oum-3~m .gaa song .3 o.m mm.o mmn.o m33.o mo3.o H

mesmesoo Am-.<.mgpgmflmm _m .M- .m xmom

.Cowwmm mommuoan umsan

map Cw memm monohommflm panowmflcmHm pmoz one u m>wvm>wpoa won: am o.m 629 .w wanna

91

cwmzo ocwm mm use on» Scam .3 o.mno.a

macaw Hzconnmo mm haw Souk .< o.m

o3 mamnmn mam pm sauna Came Scum .< o.m

wean mcflowaouahn NnH .oum Scum .< o.m
mzownopssoo .mmauum3om new

mmHnnmaom apon Souk .< o.m hpw>mo pam>aom CH
c3330 mufim «ma pm:

gone .3 o.~ a33.33550 moa--~3om gone .3 o.m

 

mHCoEEoo

mm.o
mm.o
mm.o-
mm.o

uN.o

mm.o-

3mm.o
mom.o
wm3.o
mmm.o

553.0

Nc:.o

 

an-.3.mapawfimm N

umm.o ooo.o
. mam.o moa.o
mmm.o mfim.o
mm:.o mmm.o
NmN.o mmm.o
~m~.o 3Nu.o
M M
cm52wvcoo

om
ma
ma
ma

0H

ma
33mm

 

.0 wands

92
residues Gly 216 to Thr 224. Due to the complicated solvent
structure which exists around these residues, and because
this part of the CHT molecule is strained by the close
approach of its local dyad equivalent during dimer formation,
this segment appears to be very sensitive to changes in
the solvent environment, e.g., changes in pH (38, 67).
Significant changes in this region were also observed in
the GdnHCl derivative (Chapter VI).

Though the entire segment Gly' 216—Thr' 224 is perturbed
by the urea, most of the difference peaks were clustered
near residues Ser' 221-Thr' 222-Ser' 223-Thr' 224. They
represent complicated changes involving both main chain
and side chain atoms in a number of small adjustments in
position. Changes in the solvent structure are evident as
well and most of the peaks were small as well (m2.0 A. at
the 0.2 e.A.'3 contour). Residues Ser' 221-Thr' 222-Ser'
223-Thr' 224 form a B—bend with a hydrogen bond between the
Ser' 221 carbonyl group and the Thr' 224 peptide amino group.
Peaks 2 and 4 represent movements by the main chain atoms
at Ser' 221 and Thr' 224 away from each other (they cannot
move closer) probably breaking the 3-bend hydrogen bond.
Similarly, the negative peak at the Ser' 218 carbonyl group
may represent a movement by the main chain at Ser'218-Thr' 219
resulting in the breaking of the hydrogen bond between the
peptide amino group of Thr' 219 and souz’-217'. The

2-

so” -217' itself has net been affected.

93

The three largest difference map features in this group
of peaks are peaks 8, 10, and 11. Peak 10 is 3.5 A. in
length at the 0.2 e.A."3 contour and is in the native density
at the main chain at Cys' 220, indicating some disordering
of the main chain atoms as there is no accompanying positive
density. There is no indication that the disulfide Cys' 191-
Cys' 220 has been affected. Peaks 8 and 11 are oval-shaped
peaks, each about 2.5 A. in length, and neither have any
significant accompanying negative density. They were
observed on Opposide sides of the main chain at Gly' 216
and may represent a pair of urea molecules binding to the
CHT molecule since there are many polar groups on the protein
here with which urea molecules could form hydrogen bonds.

Another set of difference peaks were observed near the
active sites of both molecules in a pattern similar to that
of some of the 2.0 fl. GdnHCl difference peaks. Two of the
peaks may represent bound urea molecules and the rest of
the peaks represent changes in the solvent structure as
there was no significant difference density observed in the
native density of the protein in this region. The changes
are those represented by Peaks 12—16 in Table 6.

.Peaks 13 and 16 may represent urea molecules binding in
the dimer interface. Peak 13 is an oval-shaped peak 2.0 A.
in length; peak 16 is an oval about 2.5 A. in length.

Peak 13 is 1.0 A. from the native density at the Met 192
side chain and may represent a urea molecule forming a

hydrogen bond between an amino group and a non-bonding

94
pair of electrons on the Met 192 sulfur. Peak 16 may have
come about because of a urea molecule hydrogen bonded to
one of the counterions of either SOuZ'-l95 or SOnZ'-l95'.
Peaks 12, 14, and 15 probably represent changes in the
local solvent structure around the two sulfate ions. The
locations of these peaks are somewhat similar to
those of the GdnHCl peaks observed in this region. For
example, peak 16 is located very near the position occupied
by one lobe of the positive double peak observed in the
GdnHCl difference map (peak 5, Chapter VI, Section 3).
There correspondences observed between peaks of the two
derivatives are listed below, where the GdnHCl peak numbers

are those of the discussion in Chapter VII, Section 3.

Urea Peak GdnHCl Peak
12 2
l4, l5 3
16 5

Further, a small negative urea peak is located very close
(<l.O A.) to the negative GdnHCl peak 1, and the positive
urea peak 13 is coincident with a portion of the positive,
cresent-shaped (0.16 e.A.-3Lcontour) GdnHCl peak described

in Chapter VI near the Met 192 and Met' 192 side chains.

These similarities between the urea and the GdnHCl derivatives
are unusual - there are very few examples of peaks in one

difference map coinciding with those of the other.

95

Finally, there are a number of peaks located in the
dimer interface solvent cavity near residues Gly 38-Phe 39-His
40. These are Peaks 17-20 in Table 6. Peaks 19 and 20 are
in the solvent and may represent urea molecules hydrogen
bonded with the carbonyl oxygen of Gly 38 and the side
chain hydroxyl group of Thr 37, respectively. The rest of
the peaks observed here represent small changes in the

protein and the solvent and are difficult to interpret.

5. Changes in the A-Chain/B-Chain Contact Region.

 

The A-chain (residues 1—13) is attached to the B—chain

covalently at one point - the disulfide bond Cys l-Cys 122-
and by non-covalent interactions at many other points. The
A-chain is composed chiefly of amino acid residues with
nonpolar side chains and folds over the surface of the CHT
molecule such that it forms a number of van der Waals
contacts with nonpolar side chains of B-chain residues.
The pyrrolidine side chains of Pro 4 and Pro 8 are directed
toward the interior of the CHT molecule, and specifically
toward a cluster of tryptOphan side chains (Trp 27, Trp 29,
Trp 237) which belong to the hydrOphObic interdomain region
described in Chapter II. This region was very much affected
by the urea; the difference map contained a number of
significant features in a relatively small portion of the
CHT molecule. These features' unit cell coordinates and
peak heights are listed in Table 7.

Peaks 1 and 2 join at the 0.2 e.A.-3 contour to form a

continuous feature about 5.0 A. in length, roughly perpendicular

 

96

macaw thonpmo OH 5mg Scam .< 0.3 mm.ou

macaw ocwpmmm m Hm>um cum p< mm.ou

seem 50 m Hs> p3 Hm.o-

macaw Hhsonhmo muapmom 5 saw 93 mm.ou

cause seam mm 33> p3 mm.o-

cacao scam w oHH map 8099 .< o.N mm.o

msHsso seam mm 33> sad m Hs> seem .3 o.m mm.o

 

 

mpCmEEOU

Anu.<.mvvgwwmm

n3m.o
mmH.o
nma.o

mma.o

NON.O
MNN.0
mmm.o

N

mmo.o
wmo.o
ooo.o
mmo.o
mmo.o
mmo.o
moo.o

N

350.0
oss.o
mom.o
mom.o
mam.o
n3m.o
mom.o
.m

.Cowwmm PomPcOo Cwmnolm\cwmnol<

esp s3 eased eeseseeman esseamasmam ewes esp - e>aes>asea ads: as o.m e39

HNMSWON

 

33mm

.5 mdnme

97

to the yz-plane. This feature is located in a crevice on
the surface of the CHT molecule formed by the two polypeptide
segments Ile 6-Glu 7—Pro 8-Val 9 and Ala 22-Va1 23-Pro 24-
Gly 25-Ser 26, and is about 2.0 A. from the side chains of
Ile 6, Val 9, and Val 23. Peaks 3-7 represent changes in
the orientation of the Val 23 side chain and a general
disruption of the A-chain residues Glu 7-Pro 8—Val 9-Leu 10.
The terminal methyl group of the Ile 6 side chain can be
placed into the positive density of peak 2, but none of the
neighboring side chains can be repositioned to fit peak 1.
It is unlikely that peaks 1 and 2 were the result of the
A-chain moving closer to the B-chain - the negative density
around the A-chain indicates small, localized movements by
individual groups. Peaks 1 and 2 were not formed by a move-
ment of the B-chain toward the A-chain as there is no
significant difference density in this region of the B-chain,
except at the Val 23 side chain. The most satisfying
explanation for these changes is that peaks 1 and 2 represent
a pair of urea molecules - possibly hydrogen bonded to one
another since their peak centers are bout 2.0 A. apart -
binding by van der Waals forces in the aliphatic "pocket"
formed by the side chains of Ile 6, Val 9, and Val 23.
Stable urea-hydrocarbon complexes are known and the
structures of some have been determined by X-ray
crystallography.68 These features appear in Figure 19(a).

I Peak 3 arises because of a motion by the Val 23 side

chain about the Ca-CB bond to move the is0pr0pyl group

98

 
  

~‘\\

‘1 .
’II‘I-jl .,‘.

:4

g g
r. -;.»‘ J's"
f
IE , 47'»)...

[a “ “\\"\\
(l \ NM“:
‘ ‘31.}.
‘\ of ”'52 '7 ‘7 . ,
V ' I '5

 
 
 
 
 
  

.

     
 

  

“x\.-

I/1\

C

  

5..) )
1 /‘

 

Ala5 Ile6 Gln? Urea Val23 Pro 24

(a) Sections of electron density in the x—interval 66/76
to 75/76.

Gin? Pr‘08 Va|9 Gln157

    
  
 
  
  

  
 
  

    
 

'1'
,1 '
N

_ , ‘\\".4.

’ I
‘
‘\ . L \-
/ .
. 4 -. ‘
- " I I \ ;_‘ 9:. '

(b) Sections of electron density in the x—interval 55/76
to 65/76.

Figure '19. Difference peaks observed in the A-chain/B—chain
contact region. The difference map contours are the thick
gray lines.

\
.-

 

fl"

    

99
further from the bound urea. Its position in the derivative
structure is unclear because there is no significant positive
density accompanying negative peak 3; it may be disordered
in the derivative structure. Peaks 4 through 7 represent
movements by groups in the polypeptide segment Gln 7-Leu 10
in reSponse to the close approach of the urea molecules.
Like peak 3, however, peaks 4-7 have no accompanying
significant positive density indicating that these residues
have become disordered in the derivative structure.
Interestingly, residues Ser ll-Gly lZ-Leu 13 are also
disordered but in the native structure as no definitive

52

electron density exists for them.

6. Changes in the Nonpolar Interdomain Region.

The striking structural features of this region of the
CHT molecule have been described in Chapter II. It contains
three clusters of aromatic residues, several proline side
chains, and a number of aliphatic side chains arranged
roughly in the shape of a "horseshoe" surrounding the inner
ends of the two folding domains (cylinders). A concentration
of difference peaks was observed in this region and the
peaks occurred in two groups. One group of small peaks
represents scattered changes occurring in a segment of
ELsheet belonging to the B-domain near the tryptOphen
cluster Trp 27, Trp 29, and Trp 207. The second group of
peaks were observed in and around a segment of B—sheet
belonging to the A-domain and near the atomatic cluster

Trp 51, Phe 89, and Trp 237.

100

The members of the first group of peaks were observed
at scattered locations in and around a small segment of
anti-parallel B-sheet composed of the polypeptide segments
Val 137-Thr 138-Thr l39-Gly 140, Gln 156-Gln 157-Ala 158-
Ser l59-Leu 160, and Gly l97—Pro l98-Leu l99-Val 200. The
peaks of this group are peaks 1-5 in Table 8. Though the
exact nature of the changes here are unclear, most of the
difference peaks listed and an equal number of smaller peaks
represent small local shifts in position of main chain and
side chain atoms. Peaks 1 and 5 are similar in appearance -
both are cylindrical and 2.0-3.0 A. in length and both are
1.0-2.0 A. from the native density of an aliphatic and an
aromatic side chain, respectively. There are several other
peaks observed in the difference map having sizes, shapes,
and heights very similar to peaks 1 and 5, and are located
close to, but not in, the native density at various residues.
Because of the similarities of appearance, peaks 1 and 5
may represent urea molecules. Urea is capable of interacting
with a variety of protein groups, including nonpolar groups
so that the penetration of the hydrOphObic core of the CHT
molecule by urea is not at all unreasonable. If peak 1 is
a urea molecule participating in van der Waals interactions
with the Leu 155 side chain, peaks 2, 3, and a small negative
peak at the Leu 155 a-carbon atom (all within about 5.0 A.
of peak 1) represent small movements by the nearest protein
groups to accomodate the urea molecule. Likewise peak 5

may represent a urea molecule sandwiched between the native

101

03000 0033000 003 003-303 :03 e3

Cacao 003m m03 033 Sony .3 o.3

msohm 30209300 0m 03¢ Scum .3 0.3

£3020 0030 cm 003 scum .< o.m

0:000 0033000 00 s33r30 use 33

scum do :03 has p¢

seem so 003 303 seem .3 0.3

macaw 0:380 ocwpmmm 003 :03 P<

cacao 0030 m33 pom

seen .3 0.0 .msosw 030033 00 see 3000 .3 0.0
MC3H 0:3033ouuhm 003 ohm

603% .< o.m .mzouw o3oozw mm Age 263% .< o.~
seed .0 003 use 33

seem s0 00 use 33

swmno 0030 mmH son scum .3 o.m

09208800

03» :3 03000 0030003330 03003033030 3002 033 - e>3pm>3sea sea: «a 0.0 030

00.0:
mm.o
3N.o
nm.o
HN.0I
3N.OI
nm.o
m~.ou

NN.o

00.0
00.0-
00.0-
00.0

 

30-.3.00330300 0

030.0
003.0
000.0
000.0
030.0
000.0
003.0
030.0

033.0

003.0
00~.o
000.0
000.0

033.0
033.0
003.0
000.0
000.0
033.0
033.0
033.0

003.0

000.0
003.0
033.0
033.0
cm

000.0
000.0
330.0
300.0
300.0
000.0
000.0
030.0

N3m.o

000.0
330.0
030.0
030.0
_m

.Cowwom SamsovhmPCH uwaomcoz

m3
NH
Ha

\OBQQ

HNnd’

 

3000

.m manna

102

00000 0033000 00 030-00 033 s033 .3 0.0
s030 30 00 003 33
03000 0030 003 033 s003 .3 0.3

macaw 0038003onpmo 230:0 003m 303 £03 »<

 

mpCmESoo

00.0 000.0
00.0- 000.0
00.0 003.0
30.0- 000.0

 

30-.3.03330300 0

003.0 000.0 03
000.0 000.0 03
000.0 000.0 03

Nu:.o M3m.o :3

%

005:3PCoo

N

 

3000
.0 0300a

103
density of the Trp 29 indole side chain and that of the
A-chain backbone at Gly 2. No significant negative difference
density was observed in the native density of any of the
nearby amino acid residues - apparently peak 5 represents
a substitution requiring little accomodation by the protein.

The second group of peaks occurred in and around a
segment of anti-parallel B-sheet in the A-domain made up
of the polypeptide segments Trp 5l-Val 52-Val 53-Thr 54,
Thr 104-Leu 105-Leu 106-Lys 107-Leu 108-Ser 109, and
Ile 85-Ala 86-Lys 87-Va1 88-Phe 89-Lys 90. There are
larger peaks in this group than in the previous group, and
they combine to form more extended changes. They are listed
in Table 8 as Peaks 6-17.

This group of peaks displays a feature not observed
anywhere else in the urea difference map - an extended
density feature at the 0.16 e.A.'3 level under several
small peaks at the 0.2 e.A.-3 level. Three negative peaks,
including peaks 8 and 9, and a small peak at the Phe 89
carbonyl group combine to form a negative feature
approximately 11.0 A. in length involving peptide groups in
three chains — peptide Thr 54-Ala 55, Thr 104-Leu 105, and
Phe 89-Lys 90. There is one'significant positive peak -
peak 7 - approximately 2.0 A. from peak 8, which together
with peak 8 may represent a movement by the main chain
atoms at Thr 104 into the hydrophobic cavity described in
Chapter II. The motions of the other two polypeptide

segments are uncertain, but it is clear that this segment

104

of B-sheet has been disrupted. Three other nearby negative
peaks indicate local shifts by main chain and side chain
atoms at Asn 101 (peaks l3 and 14) and at Asn 95 (peak 16).
These residues are on the surface of the CHT molecule and
may be the result of local changes in solvent structure and
not directly related to the extended negative feature just
described.

This region also contains several positive, oval-shaped
peaks having no significant accompanying negative density.
The peaks are in the solvent near the Lys 90 side chain
(peak 10), between the side chains of Ile 103 and Asn 91
(peak 12), and in the solvent 2.0 A. from the peptide group
joining Ala 56 and His 57 (peak 17). These peaks may
represent bound urea molecules (especially peak 12) or
changes in the solvent structure at points close to the

surface of the protein molecule.

Chapter VIII.

Summary of Results and Discussion
1. The 2.0 M, Guanidine HCl Derivative.

The changes observed in the 2.0 M, GdnHCl derivative
were confined entirely to the surface of the molecule
(including the dimer interface region) and no significant
difference density was observed in the hydrOphObic inter—
domain region or in either of the folding domains. Most
of the difference peaks were small in extent and indicative
of localized movements of side chain and main chain atoms
and changes in the local solvent structure. There is also
evidence that the Gdn+ ion binds to the proteins in the
dimer interface near the sulfates SOuZ'-l95 and 8042'-l95',
and in the uranyl binding site. However, the latter binding
site is an artifact of the crystal packing and may not be
significant to the chemistry of GdnHCl denaturationo
The sulfate ion near Lys 177 is also removed from the
surface of the molecule to the bulk solvent in the GdnHCl
derivative.

The chief effect of the GdnHCl at the 2.0 molar con-
centration on the structurejof the CHT molecule, seems to
be to cause changes in the solvent shell around the molecule.
The protein reSponded to the presence of the GdnHCl by
undergoing structural adjustments in those parts of the
molecule already known to be solvent-sensitive from the pH
derivative studies. Concentrations of GdnHCl typically

required to effect the complete unfolding of a protein

105

106

in solution are about 6.0 M,69’70’7l Clearly, changes in
structure occur at lower concentrations, and while the
changes observed in the 2.0 M, GdnHCl derivative are
significant and may accompany the onset of denaturation,

it is not possible to extend these results to a model for
GdnHCl-induced unfolding. Soaking CHT crystals in solutions
with higher concentrations of GdnHCl will not provide
answers either as an abrupt deterioration of the crystals
occurs between 2.0 and 2.3 M, GdnHCl concentrations. With
the extension of the CHT native phases from 2.8 A. to 1.8 A.
72

resolution it may be possible to learn more about the
solvent-protein interactions in the native structure. In
a 1.8 A. resolution 2.0 M, GdnHCl derivative, some of the
details of the solvent-protein interactions in the dimer

interface region may be elucidated.

2. The 3.0 M, Urea Derivative.

_The 3.0 M. urea derivative is more interesting than
the GdnHCl. Significant changes were observed not only on
the surface of the CHT molecule, but in hydrOphObic interior
regions as well. The changes observed on the surface were
local in nature affecting both polar and nonpolar groups,
but no trends were evident in the behavior of the surface
residues. A similar observation can be made with regard to
the difference Fourier features in the dimer interface. The
latter difference peaks represented local changes involving
the amino acid residues in the polypeptide segment Trp' 215-
Thr'224, the side chains of Met 192 and Met' 192, the

107
sulfate ions 8042'-l95 and SOnZ'-l95' and the surrounding
solvent.

A special subset of the set of surface changes are
those observed in the A-chain/B-chain contact region. The
A-chain and the B-chain interact via van der Waals forces
between aliphatic side chains, and these interactions
are disturbed by what may be the intrusion of urea molecules
into a space between the A-chain and segments of the B-chaino
The A-chain folds over a portion of the hydrOphObic inter-
domain region at the tryptophan cluster Trp 27, Trp 29, and
Trp 207, and the changes observed in the A-chain/B-chain
region are in close proximity to some of the difference
peaks Observed in the B-domain g-sheet. Here again, the
difference map indicates small changes involving individual
groups on the protein. There is also some evidence that
urea molecules are binding near nonpolar side chains.

The final group of difference peaks observed in the
interior of the CHT molecule were those near a segment of
B-sheet in the A—domain near the aromatic cluster
Trp 5l-Phe 89-Trp 237. Several negative peaks, taken as
a group, comprise the most extended difference feature in
the entire map and indicate a general disruption of this
segment of B-sheet. Here as previously, the appearance of
substantial positive peaks without significant accompanying
negative density suggests that urea molecules may be
binding noncovalently to protein groups. Even if some of

these peaks (especially the large peak near the Ile 103 side

108

chain) do represent urea molecules, it is not always clear
how specific urea substitutions affect the protein groups
nearest to them. Once again, this kind of detail may be
obtainable from a 1.8 A. resolution 3.0 M, urea difference
Fourier. Also, since 5.0 M, urea-CHT crystals diffract
well, the changes observable in a 5.0 M, derivative may

elaborate upon those observed in the 3.0 M, derivative.

LIST OF REFERENCES

1.
2.

3-

11.

12.

130

14.

15.

LIST OF REEEBENCES

 

W. Kauzmann, Adv. Prot. Chem., M3, i (1959).
C. N. Pace, CRC Crit. Rev. Biochem., 3, 1 (1975).

R. F. Greene, Jr., and C. N. Pace, J. Biol. Chem.,
232. 5388 (1974).

H. Edelhoch and J. C. Osborne, Jr., Adv. Prot. Chem.,
22. 183 (1976)-

F. Franks and D. Eaglund, CRC Crit. Rev. Biochem.,
l. 165 (1975).

C. Tanford,Adv. Prot. Chem., g3, 121 (1968).
C. Tanford, Adv. Prot. Chem., 24, 1 (1970).
D. B. Wetlaufer, Proc. Nat. Acad. Sci. USA, 29, 697 (1970).

A. V. Finkelstein and 0. B. Ptitsyn, J. Mol. Biol., 103I
15 (1976)-

R. Lumry and R. Biltonen, Structure and Stability of
Biological Macromolecules, vol. 2, S.N. Timasheff
and G. D. Fasman, eds., Marcel Dekker Inc., New York,

1969, p. 65.

J. F. Brandts, Structure and Stability of Biological
Macromolecules, vol. 2, S. N. Timasheff and G. D. Fasman,
eds., Marcel Dekker Inc., New York, 1969, p. 213.

 

P. L. Privalov and N. N. Khechinashvili, J. Mol.
Biol., 86, 665 (1974).

J. F. Brandts, H. R. Halvorson, and M. Brennan,
Biochemistry, M3, 4953 (1975).

J. R. Garel, B. T. Nall, and R. L. Baldwin, Proc. Nat.
Acad. Sci. USA, 23, 1853 (1976).

J. F. Brandts, J. Am. Chem. Soc.,‘§§, 4291 (1964).

109

16.

17.

18.

19.

20.

21.

22.

23.
24.

25.

26.

27.

28.

29.

30.

31.

32.

110

R. Biltonen and R. Lumry, J. Am. Chem. Soc., 21,
4251 (1969).

W. H. Porter and J. L. Preston, Anal. Biochem., 66,
69 (1975).

R. Biltonen and R. Lumry, J. Am. Chem. Soc., 2;,
4256 (1969)-

 

W. T. Astbury, S. Dickinson, and K. Bailey, Biochem.
6., 22. 2351 (1935).

K. W. Snape, R. Tjian, C. C. F. Blake, and D. E. Koshland,
Jr., Nature (London), 250, 295 (1974).

A. Yonath, A. Sielecki, J. Moult, A. Podjarny, and
W. Traub, Biochemistry, M6, 1413 (1977).

 

A. Yonath, A. Podjarny, B. Houig, A. Sielecki, and
W. Traub, Biochemistry, M6, 1418 (1977).

 

B. S. Hartley, Nature (London), 201, 1284 (1964).

 

B. S. Hartley and D. L. Kaufmann, Biochem. J., 101,
229 (1966).

D. M. Blow, J. J. Birktoft, and B. S. Hartley, Nature
(London), 221, 337 (1969).

 

B. Meloun, I. Kluh, V. Kostka, L. Moravek, Z. Prusik,
J. Vanacek, B. Keil, and F. Sorm, Biochim. Biophys.
Acta, 130, 543 (1966).

 

L. B. Smilie, A. Furka, N. Nagabhushan, K. J. Stevenson,
and C. 0. Parkes, Nature (London), 218, 343 (1968).

W. J. Dreyer and H. Neurath, J. Biol. Chem., 17,
527 (1955 .

M. Rovery, A. Curnier, and P. Desnuelle, Biochim.
Bigphys. Acta. l1. 565 (1955)-

P. Ibsnuelle, The Enzymes, vol. IV, Second Ed.,
P. D. Boyer, H. Lardy, and K. Myrback, eds., Academic
Press, New York, 1960, p.93.

 

R. B. Corey, 0. Battgay, D. A. Breuckner, and
F. G. Mark, Biochim. Bi0phy_. Acta, 24, 535 (1965).

J. Kraut, H. T. Wright, M. Kellerman, and S. T. Freer,
Proc. Nat. Acad. Sci. USA, 56, 304 (1967).

33-

34.

36.

37.

38.

39.

40.

41.

42.

43.

[+140

45.
46.

“7o

48.
49.

50.

111

H. T. Wright, J. Kraut, and P. E. Wilcox, J. Mol.
B1010. 22' 363 (1968).

S. T. Freer, J.Kraut, J. D. Robertus, H. T. Wright,
and Ng. H. Xuong, Biochemistry, 2, 1997 (1970).

J. J. Birktoft, J. Kraut, and S. T. Freer, Biochemistry.
;§, 4481 (1976).

D. M. Blow, M. G. Rossman, and B. A. Jeffrey, J. Mol.
Biol., 6, 65 (1964).

K. C. Aune and S. N. Timasheff, Biochemistry, Mg,
1609 (1972).

 

R. L. Vandlen and A. Tulinsky, Biochemistry, M2,
4193 (1973)-

B. W. Matthews, P. B. Sigler, R. Henderson, and
D. M. Blow, Nature (Londor), 214, 652 (1967).

 

 

J. J. Birktoft and D. M. Blow, J. Mol. Biol., 66,
187 (1972).

A. Tulinsky, N. V. Mani, C. N. Morimoto, and R. L. Vandlen
Acta Cryst., B22, 1309 (1973).

A. Tulinsky, R. L. Vandlen, C. N. Morimoto, N. V. Mani,
and L. H. Wright, Biochemistry, M2, 4185 (1973).

 

D. C. Phillips, British Biochemistry Past and Present,
T. W. Goodwin, ed., Academic Press, London, 1970, p. 11.

 

K. Nishikawa, T. Ooi, Y. Isogai, and N. Saito, 6.
Phys. Soc. Japan, 22, 1331 (1972).

I. D. Kuntz, J. Am. Chem. Soc., 22, 4362 (1975).

 

M. N. Liebman, Ph.D. Dissertation, Michigan State
University, 1977-

M. 0. Daynoff, Atlas of Protein Squence and Structure,
vol. 5, National Biomedical Research Foundation,
Silver Spring, Maryland, 1972.

K. Nishikawa and T. Ooi, J. Theor. Biol., 42, 351 (1974).

N. V. Raghavan and A. Tulinsky, unpublished results
of this laboratory.

M. G. Rossman and A. Liljas, J. Mol. Biol., 65,
177 (1974).

510

52.

53-
54.

55-

56.

57.

58.
59.
60.

61.

62.
63.

64.
65.

66.

67.

68.
69.

112

A. Tulinsky, P. W. Codding, and R. L. Vandlen, Abstracts,
American Crystallographic Association, Columbia, S.C.,
Feb. 1972, No. K3.

 

R. L. Vandlen, Ph.D. Dissertation, Michigan State
University, 1972.

M. V. King, Acta Cryst.,_z, 601 (1954).

R. L. Vandlen and A. Tulinsky, Acta Cryst., B27,-
437 (1971)-

H. W. Wyckoff, M. Doscher, D. Tsernoglou, T. Inagami,
L. N. Johnson, K. D. Hardman, N. M. Allewell, D. M. Kelly,
and F. M. Richards, J. Mol. Biol., 22, 563 (1967).

 

 

A. C. T. North, D. C. Phillips, and F. S. Mathews,
Acta Cryst., A24, 351 (1968).

 

D. M. Blow and F. H. C. Crick, Acta Cryst., 22, 794
(1959)-

R. Henderson and K. Moffat, Acta Cryst., B22, 1414 (1971).
D. W. J. Cruickshank, Acta Cryst., 2, 65 (1949).
D. W. J. Cruickshank, Acta Cryst., 2, 72 (1950).

 

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

V. Luzzati, Acta Cryst., 6, 142 (1953).

 

L. 0. Ford, L. N. Johnson, P. A. Machin, D. C. Phillips,
and R. Tjian, J. M01. Biol., 66, 349 (1974).

F. M. Richards, J. Mol. Biol., 22, 225 (1968).

 

A. Tulinsky, I. Mavridis, and RJ'Mann, J. Biol. Chem.,
in the press.

 

A. Tulinsky and L. H. Wright, J. Mol. Biol., 62,
47 (1973).

A. Mavridis, A. Tulinsky, and M. N. Liebman, Biochemistry,
$2. 3661 (1974)-

A. E. Smith, Acta Cryst., 2, 224 (1952).

 

 

C. Tanford, K. Kawahara, and S. Lapanje, J. Am. Chem.
Soc., 62, 729 (1967).

 

113

70. Y. Nozaki and C. Tanford, J. Am. Chem. Soc., 62,
742 (1967).

71. C. Tanford, K. Kawahara, S. Lapanje, T. M. Hooker, Jr.,
M. H. Zarlengo, A. Salahuddin, K. C. Aune, and
T. Takagi, J. Am. Chem. Soc., 62, 5023 (1967).

72. N. V. Raghavan and A. Tulinsky, Abstracts, American
Crystallographic Association, East Lansing, Michigan,
August, 1977, No. PB22.