I

4
, p3".
.W'Iflf‘
a a n"
1‘ ‘ ‘2 ,‘3‘

.. {.hflcn
4511'. a

S
‘13!

r
3.
r
I

:9.

 

”flu. 4;

«
Vi“.‘¢ .

3’ '55}:
$3324
5 J u '

49E

«)3?

I“

«n.4,...

. m

‘

3:132
9

2
-.'i

-~ .7“
I. e

flu E

135:?

 

This is to certify that the

dissertation entitled

STUDIES ON THE CONFORMATION OF ESCHERICHIA COLI
CATABOLITE ACTIVATOR PROTEIN AND RNA POLYMERASE WHEN
THEY INTERACT WITH PROMOTER DNA

presented by

MARK HENRY SINTON

has been accepted towards fulfillment
of the requirements for

Ph .D. degree in Biochemistry

Mf W

Major professor

Date 9/;‘f/Q3
I

MS U is an Affirmative Action/Equal Opportunity Institution 0-1277 1

 

 

MICHIGAN STATE UNIVERSITY LIBRARIES

293 01022 1822

 

LIBRARY
Michigan State
Unlverslty

 

 

 

PLACE II RETUI-‘W DOXto mantras ctnckouflrun your meow.

TO AVOID FINES Mum on or before date duo.

DATE DUE DATE DUE DATE DUE
'l 1 9 lUUb' *

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

MSU I. An Affirmative Action/Equal Opportunity Inotttwon
mm:

 

 

 

 

 

STUDIES ON THE CONFORMATION OF ESQHERIQHIA QQLI
CATABOLITE ACTIVATOR PROTEIN AND RNA POLYMERASE WHEN
THEY INTERACT WITH PROMOTER DNA
By

Mark Henry Sinton

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

DOCTOR OF PHILOSOPHY

Department of Biochemistry

1993

ABSTRACT

STUDIES ON THE CONFORMATION OF ESCHERICHIA COLI
CATABOLITE ACTIVATOR PROTEIN AND RNA POLYMERASE WHEN
THEY INTERACT WITH PROMOTER DNA

By

Mark Henry Sinton

Although much research has been done on the interactions of
Escherichia 9911 catabolite activator protein (CAP) and RNA polymerase
when bound at promoter DNA, the study of the conformations of the proteins
in these complexes is a largely unexplored area. The precise conformation of
each protein is likely to play an important role in the regulation of
transcription initiation. In order to explore protein conformation, in
particular when the proteins are interacting with promoter DNA, both CAP
and RNA polymerase were labeled with fluorescent reporter groups, then
mixed with various promoter DNA fragments. Gel retardation assays and
runofi‘ transcription experiments indicate that fluorescent labeling of the
proteins does not grossly alter their function. In addition, the sites of probe
attachment were determined. Detailed kinetic analysis and runoff
transcription experiments show that, while the labels were attached at
regions that are not critical to protein function, they do influence protein
activity.

Fluorescence spectra from labeled protein-DN A solutions indicate a
heterogeneous mixture of specific and nonspecific complexes, as well as free
protein. The heterogeneous mixture makes interpretation of the fluorescent
spectra dificult, but some difi'erences in the “average” solution complex are

mggefled

Harem ;

 

mpiexes

‘average'

   

nudesti:

Distribute
immh;
3mm ;
the a Sr".

 

Wilmerasi
PIESL‘HCE 0'
WIFmeras
The Seque.
the Wiser.
Emma:
Mains u
beading S
Hansen-p.

suggested from the data. CAP in “average” CAP-DN A complexes may have a
difi‘erent conformation from CAP in “average” CAP-RN A polymerase-DNA
complexes at a particular promoter, for instance. The conformation of CAP in
“average” CAP-polymeraseDNA complexes may also depend on the precise
nucleotide sequence of the promoter.

CAP and RNA polymerase each induce DNA bending that may also
contribute to the control of transcription initiation. Using gel
electrophoresis, the protein-induced DNA bending at the lactose and
galactose promoter regions was determined. The data indicate differences in
the structure of the complexes. At the lag operon, the CAP- and RNA
polymerase-induced bends appear to add, yielding a larger bend in the
presence of both proteins. At the galactose operon, however, CAP and RNA
polymerase together do not produce a larger bend than either protein alone.
The sequence of the promoter may influence the total amount of bending in
the presence of both proteins. Lac-gal hybrid promoters (a 1353 CAP site
attached to a gal promoter sequence) seem to be bent in the presence of both
proteins to a similar extent as gay promoter DNA. The differences in
bending suggest that protein-induced DNA bending may not be critical to
transcription initiation and its regulation.

for Sharon

thank you

iv

ACKNOWLEDGMENTS

I want to thank Dr. Arnold Revzin for his support of me and this work,
and for giving me the freedom to explore on my own. I thank my committee
members, Drs. Lee Kroos, Zachary Burton, Tom Deits, and Loran Synder, for
their useful comments and help over the years. I also wish to thank Drs.
Jack Holland and Gale Strasburg for the generous use of their fluorometers,
as well as many helpful discussions on fluorescence. I owe a special note of
thanks to Joe Leykam and the folks of the Michigan State University
Macromolecular Structure, Sequence, and Synthesis Facility for their efforts
to help me with the amino acid sequencing, amino acid analysis, and HPLC
chromatography experiments. I thank Eugene Zaluzec of the Michigan State
University Mass Spectrometry Facility for performing the mass spectrum
experiments. I give thanks to my lab mates--Donald Lorimer, Hongyun Yu,
and Diane Cryderman--for the many lively discussions and support since I
joined the team. A special thank you goes to Richard Halberg for listening to
me, especially when I was frustrated by the difficulty of my research.

Finally, thank you to all of you within the department who I have had the
pleasure to learn with and from, both in and out of the laboratory. I have
enjoyed getting to know all of you.

 

h
ZGCLCI

Chaps:

TABLE OF CONTENTS

List of Figures ................................................................................................... viii
List of Tables ...................................................................................................... xi
List of Abbreviations ......................................................................................... xii
Chapter 1. Introduction and Literature Review ............................................. 1
Introduction ......................................................................... 1
Literature Review ............................................................... 3
Chapter 2. Fluorescent Labeling of CAP and RNA Polymerase .................. 28
Introduction ....................................................................... 28
Materials and Methods ..................................................... 30
Results ............................................................................... 44
Discussion .......................................................................... 65
Chapter 3. Characterization of Chemically Modified CAP and RNA

Chapter 4.

Chapter 5.

Polymerase ................................................................................... 69
Introduction ....................................................................... 69
Materials and Methods ..................................................... 71
Results ............................................................................... 81
Discussion ........................................................................ 105

Fluorescence of EITC- and LYIA-Labeled CAP and RNA

Polymerase ................................................................................. 113
Introduction ..................................................................... 113
Materials and Methods ................................................... 115
Results ............................................................................. 123
Discussion ........................................................................ 157

CAP and RNA Polymerase Induced Bending At Lactose and

Galactose Promoters .................................................................. 161
Introduction ..................................................................... 161
Materials and Methods ................................................... 163
Results ............................................................................. 172
Discussion ........................................................................ 193

vi

(“nap-1H

 

Append:

.L“ 3’ .
BIULAETG

 

Chapter 6. Summary .................................................................................... 200
Appendix A. ..................................................................................................... 205

Bibliography .................................................................................................... 208

vii

 

Figure 2.4

n
r.

are 2.,

I‘m

.nJ

Flé’l'e 3.2
Figure 3 9

 

Figure 2.1

Figure 2.2

Figure 2.3

Figure 2.4
Figure 2.5

Figure 2.6

Figure 2.7

Figure 3.1

Figure 3.2
Figure 3.3

Figure 3.4

Figure 3.5
Figure 3.6
Figure 3.7

Figure 4.1

LIST OF FIGURES

Structures of Eosin Isothiocyanate (EITC) and Lucifer Yellow
Iodoacetamide (LYIA) ....................................................... 34

Absorbance Spectra of Labeled CAP and RNA Polymerase ...... 45

Relative Fluorescence Efficiency Spectra of Labeled CAP and

RNA Polymerase ............................................................... 47
Mass Spectra of LYIA-Labeled and Unlabeled CAP .................. 50
Reverse Phase HPLC of Labeled and Unlabeled CAP ............... 57
EITC-Labeled CAP and RNA Polymerase Bind to Promoter

DNA ................................................................................... 6O
Transcription by Labeled CAP and RNA Polymerase ................ 62

Reverse Phase HPLC Separation of Digested EITC-Labeled
and LYIA labeled CAP ...................................................... 82

Mass Spectra of EITC-Labeled and Unlabeled N eurotensin ..... 85

Sequence of EITC-Labeled N eurotensin ..................................... 87
Representative Kinetic Experiments with EITC-Labeled CAP
and RNA Polymerase ........................................................ 94
Representative Graphs of Rate Experiment Data ...................... 96
EITC-Labeled Polymerase Transcription Time Course ........... 103

Location of EITC and LYIA Probes in the Sequence of CAP 110

Promoter Region Structure of Wild Type Lac, Wild Type Gal,
and Lac-Gal DNA Fragments .................................................... 1 16

viii

Figure 4.2
Figure 4.3

Figure 4.4

Figure 4.5

Figure 4.6
Figure 4.7

Figure 4.8
Figure 4.9

Figure 4.10

Figure 4.11

Figure 4.12

Figure 4.13

Figure 4.14

Figure 4.15

Figure 5.1

Figure 5.2

Figure 5.3

Titration of EITC-Labeled CAP with Lag Promoter DNA ........ 124

Titration of EITC-Labeled CAP with Various Promoter
DNA Fragments .............................................................. 126

Titration of EITC-Labeled CAP with Calf Thymus DNA ........ 128

Titration of EITC-Labeled RNA Polymerase with
Promoter DNA ................................................................. 130

Formation of LYIA-Labeled CAP Transcription Complexes... 132

Formation of Labeled CAP ”Transcription Complexes” with

Calf Thymus DNA ........................................................... 135
Heparin and RNA Polymerase Interact with Labeled CAP ..... 137
CAP Protection of L36 DNA From Q31 Digestion .................. 139
Transcription From Specific Complexes In “Fluorescence

Like” Samples .................................................................. 142
Sodium Iodide Quenching of LYIA-Labeled CAP-DNA
Complexes .................................................................................. 145
Sodium Iodide Quenching of LYIA-Labeled CAP-

Polymerase-DNA Complexes .......................................... 148
Sodium Iodide Quenching of EITC-Labeled CAP-DNA
Complexes .................................................................................. 150
Sodium Iodide Quenching of EITC-Labeled CAP-

Polymerase-DNA Complexes .......................................... 152
Acrylamide and Dithiothreitol Quenching of EITC-Labeled

CAP-DNA Complexes ...................................................... 154
Protein-Induced DNA Bending Affects DNA Gel Mobility ...... 164

Structure of the Various Promoter DNA Fragments
Used in This Work .......................................................... 167

CAP- Induced Bending of Circularly Permuted Wild Type
Galactose Promoter DNA ........................................................... 173

Rare '

Figure 5

ma

 

 

 

Figure 5.4

Figure 5.5

Figure 5.6

figure 5.7
Figure 5.8

Figure 5.9

Figure 5.10

RNA Polymerase-Induced Bending of Circularly Permuted
Galactose Promoter DNA in the Presence and

Absence of CAP ................................................................ 177
Protein-Induced Bending of IchVS-CAP+ AF and BF

Promoter Fragments ....................................................... 179
SDS Analysis of Circularly Permuted Galactose

Promoter DNA-Protein Complexes ................................. 182
SDS Analysis of ngL8-UV5-Protein Complexes ..................... 185

Protein-Induced Bending of 725bp Galactose Promoter
AF and BF Fragments .................................................... 187

Protein-Induced Bending of 725bp P2'CAP1' Lactose-
Galactose Hybrid Promoter AF and BF Fragments ...... 190

Possible CAP-RNA Polymerase-Gd Promoter DNA
Structures ........................................................................ 195

 

“at" “

l: C).

Table 2.1

Table 2.2

Table 2.3

Table 3.1

Table 3.2

Table 5.1

LIST OF TABLES

Spectral Properties of EITC and LYIA ....................................... 36
Mass Spectrum Analysis of Unlabeled and LYIA-

Labeled CAP ...................................................................... 52
Densitometry of Labeled and Unlabeled CAP Run into

SDS-PAGE Gels ................................................................ 55
Amino Acid Analysis of LYIA-Labeled CAP Peptides ................ 90
Kinetic Parameters of EITC-Labeled CAP and RNA

Polymerase Interactions with m Promoters .................. 99
Relative Mobility of Protein-DN A Complexes .......................... 175

LIST OF ABBREVIATIONS

SDS .................................................................................. Sodium Dodecylsulfate
PAGE .......................................................... Polyacrylamide Gel Electrophoresis
ddeO .............................................................................. Double Distilled Water

Chapter 1
Introduction and Literature Review

mm

The regulation of protein activity in a cell can occur at many levels.
The amount of protein can be controlled by the regulation of gene
transcription, for instance. If a protein function is not needed, there is little
reason for the cell to waste resources on its production through transcription
and translation. When a particular protein is needed, however, transcription
can be quickly initiated, and even stimulated, to yield a sufi'icient cellular
amount of the protein. Thus, transcription (and its regulation) has been
intensely studied over many years.

As a result, we now know that transcription is often controlled during
the initiation of mRN A synthesis. The initiation of transcription requires
several steps. RNA polymerase first locates and binds to a promoter DNA
sequence, usually located upstream of the gene to be transcribed. The
enzyme must then form a transcriptionally competent complex with the
promoter DNA in a process that may itself involve several chemical events.
Ultimately, nucleoside triphosphates bind to RNA polymerase and the
catalytic formation of mRN A ensues. Other proteins or small molecules may
influence transcription by interacting with the DNA and/or RNA polymerase
during formation of these complexes. Control of transcription initiation
could, in theory, occur at any of these points.

Much information is available concerning the efl'ect of DNA sequence
and conformation on the mntrol of transcription. Less is known about the
Precise conformation of proteins in transcriptionally competent complexes,

Which must certainly play a critical role. Many activator and repressor
1

2

proteins are thought to influence transcription by altering the conformation
of RNA polymerase, for instance. This thesis describes studies aimed at
learning more about the conformations of E. ggli RNA polymerase and the
catabolite activator protein (CAP), and how those conformations may change

when the proteins interact with different promoter regions.

3

I il l B .
The Lactose Operon. Elucidation of the regulation at the lactose
operon by Jacob and Monod was a large step toward understanding how cells

control their functions at the level of transcription (Jacob & Monod, 1961).
They coined the term “operon” to describe the coordinated expression of the
Escherichia 53211 enzymes needed for lactose sugar metabolism. Only in the
presence of lactose, and the absence of glucose, are three genes (LagZ, lacY,
and lacA) of the operon expressed at high levels. As the need to utilize
lactose as a carbon source arises, expression of the enzymes needed to
metabolize the sugar increases. Otherwise, the energy needed to produce
these proteins is conserved for other cellular functions. With the grouping of
related genes and a relatively simple control mechanism, whole sets of genes
can be easily expressed or not as the changing needs of the cell require.

Regulation of the lac operon involves both negative and positive
control at the level of transcription. Monod first described in 1947 the
observation that E. chi cells preferred glucose as a carbon source even if
other sugars were present (Monod, 1947). Only when the glucose had been
exhausted did the use of lactose begin. The mechanism of glucose “catabolite
repression” (Magasanik, 1961) was not obvious. While it was known that the
presence of lactose relieved the operon from negative control, something more
was needed to activate the lactose operon.

Adenosine 3’, 5'-cyclic monophosphate (cAMP) plays a key role in
stimulation of transcription of the lac Operon. The cellular concentration of
cAMP is inversely proportional to the glucose concentration (Makman &
Sutherland, 1965). IchMP is added to a growth medium containing both
glucose and lactose, good expression of the lactose enzymes is observed,
despite the presence of glucose (Ullman & Monod, 1968). How cAMP

4

moderates lactose operon expression was better understood after the isolation
of a protein that could bind cAMP and stimulate lactose operon expression
(Anderson et al., 1971; Beckwith et al., 1972). That protein was named the
catabolite activator protein (CAP) after its role in activating the lactose sugar
metabolism operon.

In terms of sites of action of these proteins on the chromosome, Jacob
proposed the idea of a “promoter” element necessary for proper initiation of
transcription by RNA polymerase (Jacob et al., 1964). It has since been
shown that RNA polymerase interacts with the promoter element to initiate
transcription at a specific site in the DNA sequence and that CAP, in complex
With cAMP, binds nearby to stimulate RNA polymerase transcription from
the operon.

That the proteins may bind to separate sites was suggested by the fact
that many of the known naturally occurring DNA mutations that affect
lactose operon transcription are clustered in two areas of the promoter region
(Arditti et al., 1968; Silverstone et al., 1970; Arditti et al., 1973). The IagUV5
mutation, for instance, exhibits high levels of transcription independent of
CAP or cAMP, and was placed in the downstream region of the presumed
Promoter element, near the start point of transcription. A cluster of
mutations (L1, L8, L37, and L29) that markedly reduced CAP-induced
transcription stimulation of the lactose operon was found to be in an
uDetzr'eam region of the promoter element.

With the advent of techniques to quickly mutate and sequence DNA,
the nucleotide structure of the promoter was determined. Comparison of
In«any promoters revealed a conserved structure (Hawley & McClure, 1983;
Harley & Reynolds, 1987 ). In each case, the promoter was located upstream
of the start site of transcription. Two regions of conserved sequence are

5

found around -10 and -35 (all numbering is relative to +1, the first nucleotide
incorporated into mRNA), which are separated by a 17 i 1bp spacer region of
nonconserved sequence. The - 10 consensus sequence is TATAAT, while the
-35 sequence is TTGACA. Other techniques for determining the binding sites
for proteins, such as DNaseI footprinting, confirm the notion that RNA
polymerase interaction with a promoter defines the start site of transcription.
Similar work with the CAP binding site produced a consensus sequence for
CAP binding (Ebright et al., 1984; Berg 8: von Hippel, 1988; Jayaraman et
al., 1989). The CAP consensus is aaaTGTGAtctagaTCACAttt, where the
capital letters indicate a strong consensus and the small letters less
conserved bases. At the lactose operon, the CAP binding site is indeed
located upstream of the promoter as was predicted by genetic analysis.

We now understand the nature of the mutations affecting lactose
operon transcription. For example, the L8 mutation is a single base change
within the CAP binding sequence that reduces the affinity of CAP-cAMP for
the site (Dickson et al., 1977). The change is from ...TGTGA...TCACA... to
...TGTAA...TCACA... (the bold type indicates the base change). The UV5
mutation is a double base change in the -10 region of the promoter, from
TATGTT to TATAAT (Reznikofi' & Abelson, 1980). This mutation creates a
Perfect -10 consensus match, and results in a strong promoter from which
transcription occurs independent of CAP and cAMP.

Further studies have shown that, in fact, there are two promoters at

the lac operon, and at other E. m operons as well (Musso et al., 1977;
Reznikofi' et al., 1982; Malan & McClure, 1984). m m, RNA polymerase
can bind to either the Lac P1 promoter, which directs initiation at +1, or the
P2 promoter that directs initiation from -22. Until recently, transcription
from P2 had not been observed in E119; however, Xiong et al. (1991) showed

6

the existence of P2 transcription in m from plasmids carrying portions of
the lactose promoter region. The P2 promoter is thus probably active within
the cell as well as in m. Other possible Lac promoter sites have been
observed. Xiong et al. also suggested the existence of a P3 promoter that
initiated transcription fi'om --15 in yin with the above described plasmids,
though there have been no reports of P3 transcription from -15 in m to
date. Lorimer and Revzin postulated a different P3 promoter downstream of
+1 (Lorimer & Revzin, 1986), and studied this system by exonuclease III
digestion experiments of RNA polymerase-lactose promoter region DNA
complexes. However, deletion of this potential downstream promoter did not
appear to alter P1 transcription (Lorimer et al., 1990). Perhaps RNA
polymerase can bind to many promoter “like” sequences, even if it does not
initiate transcription from such sites.

In the absence of CAP or cAMP, RNA polymerase binds to, and
initiates transcription fi‘om lag P2 (Malan & McClure, 1984). Some RNA
polymerase can apparently bind to P1, but does not form protein-DNA
complexes capable of transcription without the addition of CAP and cAMP
(Lorimer & Revzin, 1986). When CAP is bound to the promoter region, no P2
binding or transcription is observed-~only P1 polymerase binding and
transcription is seen (Reznikoff et al., 1982). Indeed, CAP stimulates P1
directed transcription by RNA polymerase about thirty fold m m
(Chambers & Zubay, 1969). The organization of the promoters and the CAP
site gives some insight to this phenomenon. The CAP binding site is located
around -60, and covers about 20bp (from -70 to -50) of the promoter region
DNA (Majors 1975; Spassky et al., 1984). RNA polymerase binding at P2
covers about 70bp, from -70 to +1, while at P1 it covers from about -50 to +20.

CAP binding, then, probably excludes RNA polymerase from the overlapping

7

P2 site (and from the P3 site described by Xiong et al.), thereby facilitating
RNA polymerase binding at the P1 promoter site.

We now know that repression of the operon operates on a similar
principle. The lactose repressor (product of the Incl gene) binds to sites in the
control region of the lactose operon called operator sites. Two such sites are
located about 93bp apart. When both are occupied with repressor molecules,
a structure is formed in which the DNA is looped (Kramer et al., 1987;
Brenowitz et al., 1991). While the role of the DNA loop is unclear, this
repressor complex does block transcription by RNA polymerase in the
absence of lactose. The binding of RNA polymerase does not seem to be
hindered by repressor, however (Lee 8r. Goldfarb, 1991). The lactose
repressor just physically blocks RNA polymerase from moving down the
DNA, although a small amount of transcription can occur in this state. The
operon is relieved of repression when any available lactose is converted into
allolactose (the first step of the metabolic pathway). Allolactose binds to the
repressor and reduces its affinity for operator DNA, thereby relieving
repression of the operon.

If P2 were a high afiinity site for RNA polymerase from which
transcription does not occur, then CAP could stimulate P1 transcription
simply by exclusion of polymerase from the non-productive P2 site. However,
this does not seem to be the case. Mutations of the P2 promoter that reduce
the affinity of RNA polymerase for the site do not alter P1 transcription in
the absence of CAP (Lorimer et al., 1990). A five base pair insertion at -50,
between the CAP site and the two promoter sites, eliminates the stimulatory
effect of CAP on P1 transcription (Mandecki & Caruthers, 1984; Straney et
al., 1989). Inserting ten base pairs of DNA between the sites, however,
restores the stimulatory effect of CAP. The ten base pair insertion may still

 

 

operon i

099m

8

allow CAP to exclude polymerase from P2, but the phasing of the insertion
indicates that CAP and RNA polymerase need to be on the same face of the
DNA for CAP to exert its efi‘ect on P1 transcription. Interestingly, a ten base
pair deletion disrupted transcription stimulation by CAP. Thus the specific
location of CAP relative to polymerase is important to the activation of
transcription by CAP, implying that CAP-RN A polymerase contacts may well
be critical in this process.

The Galactose Operon. CAP and cAMP together influence the level of
transcription fi'om many operons. After the lactose operon, the galactose
operon is perhaps the next best understood sugar metabolism operon. This
operon shares many features of the lactose operon, but also shows
considerable differences. There are two gal promoters (Musso et al., 1977).
RNA polymerase binds and initiates (at ~5) from P2 in the absence of
CAP/cAMP, but binds and initiates transcription (at +1) from P1 in its
presence. In this sense, positive regulation of the galactose aperon is similar
to that at the lactose operon. The structures of the promoter regions are
quite different, however. One major difference is that a CAP molecule binds
around ~40 (from about ~50 to ~30) at the gal operon (Taniguchi et al., 1979;
Busby et al., 1982). CAP binding does preclude P2 polymerase binding, as at
139 (and could potentially prevent P1 binding as well, since the CAP and
polymerase sites appear to overlap). The P1 promoter site does not have a
good ~35 consensus region, so that RNA polymerase-DNA contacts in this
region are probably weak. It appears that the CAP molecule bound at -40
does compete with RNA polymerase for DNA contacts in this region
(Shanblatt & Revzin, 1986). On the other hand, the presence of CAP does
not alter the RNA polymerase contacts in the ~10 region of the P1 promoter:
similar RNA polymerase-DNA contacts are observed for the lagUV5, lag", and

 

Wilma

9

gfi+ promoters. Promoters with a poor ~35 consensus sequence are often
positively regulated (Raibaud & Schwartz, 1984). Perhaps CAP makes
intimate contact with RNA polymerase in these situations, to somehow
compensate for contacts that RNA polymerase may not make with the DNA
in this region.

Another difference between control at the lag and gal promoters is that
at gal a second CAP molecule binds around ~60 (from ~50 to ~70), apparently
simultaneously with RNA polymerase binding at gal P1 (Shanblatt & Revzin,
1983). The second CAP binds only in the presence of RNA polymerase, and
does not require that cAMP be bound to it. Based on the observation that the
second gal CAP is located at precisely the same position (relative to +1) as
the one CAP at lag, it was proposed that these CAP molecules may interact
with RNA polymerase in a similar manner. All in all, it is likely that direct
contacts between RNA polymerase and one or two of the CAP molecules play
an important role in regulating gal transcription.

Comparison of the lactose and galactose operons shows other
difi‘erences as well. The first CAP at gal likely interacts with RNA
polymerase in a different fashion than does the lg CAP molecule, which
binds further upstream. Indeed, RNA polymerase must be rather flexible to
accommodate the propinquity of a CAP molecule bound at ~40, and still make
appropriate ~10 DNA contacts. The conformation of the first CAP and RNA
Polymerase are probably difi'erent when bound at the galactose operon then
When at the lactose operon. The second CAP at the galactose operon may
also be in an altered conformation due its interaction with the first CAP
molecule and with RNA polymerase. Thus, the end result, stimulation of P1
transcription mediated by CAP and cAMP, can probably be reached in
different ways.

10

RNA Polymerase. How RNA polymerase interacts with promoters has
been an area of intense research. RNA polymerase is a multisubunit enzyme
with the structure azflfl’o (Burgess, 1969; Burgess et al., 1969) (called
holoenzyme) with masses of 371038 (at), 156KDa (I3 and B’), and 70KDa (o)
(Burgess et al., 1986). Incorporation of ribonucleotides into the mRN A chain,
as well as chain initiation, is catalyzed by core polymerase (0935’). Promoter
recognition and specificity are the function of the a subunit. Without the a
subunit, core polymerase can only bind to DNA in a nonspecific manner.
Holoenzyme has a large promoter-specific binding constant, and a reduced

affinity for nonspecific DNA.
The formation of RNA polymerase-promoter DNA complexes capable of

transcription is a multistep process. RNA polymerase must first bind to the
DNA. This initial binding may not be at a promoter, as holoenzyme can bind
nonspecifically to random DNA sequences. How polymerase actually finds a
promoter in a sea of competing nonspecific sites is unclear, but probably

involves movement of the enzyme via a one-dimensional random walk along
the DNA until a promoter region is found. A simple model for describing
polymerase interaction with a promoter is shown below (Chamberlin, 1974):

kl k2
R+Pr = RPrc = RPI'O,
k_1 1a,
where R denotes RNA polymerase and Pr represents promoter DNA. The
formation of transcription complexes involves the binding of polymerase to
the promoter sequence to form a “closed” complex (RPrc), followed by
“melting” of the DNA (separation of part of the DNA into single strands) to

form an “open” complex (RPro). The melted region is from about ~10 to +3

Sléhih

intern

appeal
Ceglae

bend ar

 

 

 

 

DNA 001
scan/em
Ti Al p
DNA

New
RNA p(

with th
‘56 HR
“Isa:
Camila
IO [DOVE
Dom: of
the A55

11

(Siebenlist et al., 1980). Variations of this model involving additional
intermediates have been invoked to rationalize some data.

RNA polymerase-promoter binding and open complex formation
appears to induce a bend in the promoter DNA (Kuhnke et al., 1987;
Ceglarek & Revzin, 1989; Kuhnke et al., 1989 ). Details of formation of the
bend are yet to be elucidated, but it may allow the protein to make more
DNA contacts than would otherwise be possible. A neutron small angle
scattering study by Heumann et al. (1988) of RNA polymerase bound to the
T7 A1 promoter indicates that polymerase binds mostly on one side of the
DNA.

Transcription usually begins at an A or G in bacterial systems (Harley
81 Reynolds, 1987). After formation of the first few phosphodiester bonds,
RNA polymerase appears to proceed in two ways. The nascent mRN A
transcript, up to about the tenth incorporated nucleotide, may be aborted
with the enzyme not leaving the promoter (McClure et al., 1978). If aborted,
the mRNA chain is released, and polymerase may begin another round of
transcription from +1. Alternatively, 0' may dissociate from the initiated
complex at about the tenth nucleotide, after which core polymerase continues
to move down the DNA, catalyzing the elongation of the mRNA chain to the
point of termination. Termination involves release of the mRN A chain and
the dissociation of core polymerase from the DNA. The released a and core
are free to interact with other core or 0 molecules to begin another round of
transcription (Travers & Burgess, 1969; Wu et al., 1975). There may also be
other steps in this sequence of events. Straney and coworkers showed the
existence of several RNA polymerase-DN A complex intermediates between
Open complex and elongation complex formation ( Straney & Crothers, 1985;
Straney & Crothers, 1987; Straney & Crothers, 1987b). Initiated

12

intermediates, for instance, showed a loss of upstream DNA contacts as
compared to uninitiated open complexes.

Various techniques have been used to study the functions of the
individual subunits of RNA polymerase. Such dissection of the enzyme
indicates that each subunit contributes to the overall function of the enzyme.
Analysis of or mutants in which various sequences were deleted suggests that
subunit assembly is the function of the N ~terminal domain (Hayward et al.,
1991; Igarashi et al., 1991; Igarashi & Ishihama, 1991). Assembly of core
polymerase is believed to proceed as shown:

oc+ B -> aB—E—eczfl—Lazfib’.
Deletion of about one-third of a from the C-terminus does not alter subunit
assembly of core polymerase in yitm or it; yiya. The C-terminal portion of the
a subunit appears to interact with regulatory proteins such as CAP ( Igarashi
& Ishihama, 1991; Igarashi et al., 1991; Zou et al., 1992; Kolb et al., 1993).
Holoenzyme containing truncated 0: subunits is not stimulated by CAP m
m, suggesting an interaction between CAP and the or subunit. Further,
there appear to be two regions of or that interact with CAP, depending on the
position of CAP relative to the promoter site. As one might expect, CAP
located at ~60 interacts with a different area of on than when it is located at
~40. Some deletions of or eliminate CAP transcription stimulation when CAP
is at ~60, but do not do so when CAP is at ~40. 2 Such C-terminal deleted a
subunits do not appear to alter polymerase function at CAP-independent
promoters like lacUV5. A point mutation study of the C-terminal region of or
revealed that CAP positioned at ~40 contacts the or residues between 265 and
270 (Zou et al., 1992).

Rohrer et al., (1975) reported that a was ADP-ribosylated during T4
Phage infection of E. 99h. They suggested that or could be involved with the

13

promoter binding, since the ribosylated holoenzyme had a reduced affinity for
promoters. This could also be explained, however, by the subunit assembly
function of a. The on subunits probably make intimate contact with the other
subunits. The modified subunit could have an altered conformation that
allosterically induces structural changes in the rest of the holoenzyme that
are unfavorable for promoter binding.

The B and 8’ subunits are involved in many steps in the transcription
process. Affinity labeling shows that the 3 subunit binds ribonucleotides,
suggesting that the active site of mRN A synthesis resides within this subunit
(Grachev et al., 1989; Mustaev et al., 1991). A GMP afinity label was cross-
linked to lysines and a histidine in the 3 subunit. Localization of the cross-
linked residues indicates that lysines in the regions 1036 to 1066 (in
particular lysine 1065) and 1234 to 1242 are close to the active site. The
histidine residue was localized to residue 1237 . While lysine 1065 and
histidine 1237 are probably near the active site of the subunit, they are not
critical for initiation. Substitution of arginine for lysine 1065 and alanine for
histidine 1237 did not inhibit transcription initiation, but blocked the
transition from initiation to elongation. The nucleotide binding site may
undergo a conformation change during the initiation-elongation transition.

The 8 subunit may also bind the mRN A chain during polymerization
(Lee et al., 1991). Mutation of glycine 813 to lysine results in an increased
dependence on ribonucleotide concentration during elongation and an
increase in abortive initiation. This mutation also decreases the stability of
open complexes. Lee and coworkers suggest that glycine may contact the
mRN A chain as part of a bipartite nucleotide binding sequence (N-X-X-D-G-

X-X-X-X-G-K), which is found in many GTP and ATP binding proteins.

 

 

 

attach

14

The 8 subunit may contact promoter region DNA (Martin et al., 1992).
Deletion of ten residues from 435 to 444 results in a holoenzyme that
elongates properly and can recognize promoters, but has a reduced promoter
binding amnity. Once elongating a transcript, however, the core enzyme
appears to function correctly. This N ~terminal region appears to be a
conserved motif in 8 subunits of other organisms.

Less is known about the 8’ subunit, but it may bind the mRNA chain
during polymerization (Borukhov et al., 1991). Elongating core polymerase
can made to pause by the incorporation of a ribonucleotide analog into the
mRNA that does not have a free 3’ hydroxyl group. This stalls the core
polymerase from further elongation, but does not cause complex dissociation.
The ribonucleotide analog 8-azido-ATP is one such molecule suitable for this
purpose. The 8-azido-ATP can be photo-crosinnked to the paused core
polymerase (Borukhov et al., 1991). Such an experiment resulted in the
attachment of the 3’ end of the nascent mRNA chain to the 8’ subunit.
Specific chemical degradation of the cross-linked species revealed methionine

932 and tryptophane 1020 as the sites of attachment. In the initiation
complex, 8-azido-ATP photo-crosslinking of tri- and tetranucleotide mRN A
chains results in attachment to B, 8’, and c (Bowser & Hanna, 1991), but
attachment to only 8 and 3’ with pentanucleotide mRNA-initiated complexes.

Several compounds known to inhibit RNA polymerase function bind to
either the B or 3’ subunit. Heparin and rifampicin are two such compounds.
Heparin, a negatively charged polysugar, acts to block transcription by
inhibiting polymerase-promoter binding. Once polymerase is in an open
complex, however, heparin does not affect initiation, elongation, or
termination. Being negatively charged like DNA, heparin may compete with
DNA for RNA polymerase binding. Heparin apparently binds to the B’

subur

 

pm;

oftbe

Static
mutatr
the firs
whim“

 

limit-,5:
does 11c
be‘We-e
mm

mm:

15

subunit (Zillig et al., 1976), suggesting that this subunit can also bind
promoter DNA.

Rifampicin blocks transcription initiation by inhibiting the formation
of the phosphodiester bond between the first two ribonucleotides (Wehrli &
Staehlin, 1971). Rifampicin apparently binds to the 8 subunit, since
mutations of this subunit overcome rifampicin inhibition. This was one of
the first indications that the nucleotide incorporation site resided in the 8
subunit. The mechanism of rifampicin action is unclear, however. A
fluorescence energy transfer study indicates that the rifampicin site probably
does not overlap the active site of B (Kumar 8: Chatterji, 1990). A distance of
between 24 to 30A between the rifampicin site and the nucleotide site
(containing terbium (III)~GTP) was calculated. Rifampicin probably does not
inhibit bond formation by simple overlap of the 8 subunit active site.

The 8 and 8’ subunits each contain an octahedral coordinated zinc ion
(Giedroc & Coleman, 1986). The zinc ion within the 3 subunit may be
involved in binding the initiating ribonucleotide (Wu et al., 1992).
Rifampicin may disrupt the coordination of the 8 subunit zinc ion, reducing
the afiinity for the initiating nucleotide and thereby inhibiting bond
formation. The role of the 3’ subunit zinc ion is unclear, but it is probably
not directly involved with template recognition (Wu et al., 1992).

While the core polymerase catalyzes mRN A polymerization, the
primary function of the 0 subunit is promoter recognition and specificity.
The 0'70 subunit recognizes variants of the consensus promoter described
above (the superscript indicates the mass of the subunit). In this sense,
holoenzyme” containing 0'70 (E070) is the general purpose transcription

enzyme of E. can. However, other 6 subunits are known. Under conditions
of heat shock, E. ml; cells produce 032. E032 recognizes promoters with a

16

different ~10 consensus sequence than described above, but with a similar ~35
region (Landick et al., 1984). Another 6 subunit, 0'54, recognizes the
promoters for nitrogen regulated genes (Hirschman et al., 1985). The
separation of promoter recognition from the polymerization function allows
for transcriptional regulation simply by altering the relative abundance of
each a subunit within the cell as conditions warrant.

Certainly one way for 0' to confer promoter recognition to the core
enzyme is for a to actually contact the ~10 and ~35 sequences of a promoter.
Genetic evidence indicates that this is indeed the case (see below). It is likely
that a masking function keeps free a from binding to promoter DNA until it
is a part of the holoenzyme. Otherwise promiscuous binding to promoter
DNA could reduce the pool of free subunit available to form holoenzyme,
potentially blochng holoenzyme promoter binding and transcription. The
conformation of 0 changes upon binding to the core polymerase, suggesting
such a masking function (Wu et al., 1976), and would explain why extensive
efi‘orts have yet to reveal any convincing c—promoter DNA footprints. A
report of a complex between 0 and the A PR promoter (Ramesh & Meares,
1989) was later retracted (Wellman & Meares, 1991) due to the inability to
repeat the original experiment. The a subunit could also be involved in the
melting of DNA prior to transcription initiation. Simpson (1979) showed that
ultraviolet light-induced crosslinking of bromouracil lagUV5 DNA to bound
RNA polymerase holoenzyme involved the o and 8 subunits. The
nontemplate base ~3 was found to attach to the a subunit, a position that is
within the region of melted DNA of RNA polymerase-promoter DNA open
complexes.

Comparison of various 0 subunit sequences reveals four conserved

regions, numbered 1 through 4 (Helmann 8: Chamberlin, 1988). Through

 

genes.
unders
It? r'l'.
0 Der.
he er:J
These i

chi:

 

m
for pro
are en;

tinting
T
at“

“fliers:

17

genetic manipulations, the ftmction of each region is becoming better
understood. The N—terminal region 1 of 07° spans residues 1 to 139, and is
thought to be involved in masking the DNA promoter binding activity of free
a (Dombroski et al., 1992). C-terminal and internal deletions of 0 revealed
the existence of two DNA binding domains within the subunit sequence.
These domains could discriminate and specifically bind to promoter DNA, but
only if devoid of the N ~terminal conserved region. In the presence of the N -
terminal region, these domains could bind to DNA, but showed no preference
for promoter DNA over nonspecific DNA As the a subunit interacts with the
core enzyme, a conformation change in region 1 may allow the two DNA
binding domains to recognize promoter DNA.

The 070 subunit has a roughly 245 residue spacer between regions 1
and 2 (from residue 139 to 384). The function of the spacer is not well
understood, but may possibly be involved with several subunit functions. It
may be important for stability of the subunit (Hu & Gross, 1983), for
instance. A fourteen residue deletion within the spacer reduces the apparent
stability of the subunit, depending on the temperature of cell growth. At the
appropriate lower temperature, however, the subunit seems to function
normally. The region between 361 and 390 appears to be important to core
binding (Lesley & Burgess, 1989). Internal deletions and frame shift
mutations of this region resulted in a reduced 6 afinity for core polymerase.
Further, a synthetic peptide of this sequence could bind core polymerase.
However, the peptide could also bind intact 070 and holoenzyme, so the
intportance of this sequence in core binding is somewhat unclear.

The spacer region may also be involved in the reduction of the affinity

of holoenzyme for nonspecific DNA as compared to the core enzyme. The 0
subunit of B. aahtiha (043) does not have a large spacer region, and the

18

holoenzyme requires another protein (521) to reduce its nonspecific binding
('Iiiian et al., 1977). The E. call 0’70 spacer region may thus be homologous to
the 621 protein in function, though such homology has yet to be proven.
Region 2 can be divided into four subregions, and extends fiom about
residue 384 to about residue 464 in 670. While subregion 2.2 is the most
conserved span of the various region 2 sequences, its function is not clear.
The highly conserved nature of this subregion suggests that it is involved
with core polymerase binding, but no hard evidence exists to show this.
Subregion 2.4, on the other hand, is thought to make contacts with the ~10
region or promoter DNA (Siegele et al., 1989; Waldburger et al., 1990).
Mutant promoters were used in yiya to select for o subregion 2.4 mutants
capable of transcription from them. Three classes of o mutants were
observed. The first class of mutant subunits exhibited little change in
transcription from the mutant promoters as compared with wild type protein.
The second class showed some transcriptional increase independent of the
promoter mutation. The third class, however, was comprised of subunit
mutations that increased transcription depending on the position of the
promoter mutation. One region 2.4 mutation (threonine 440 to isoleucine)
enhanced transcription fium - 10 mutant promoters. Specifically, this residue
responded to the ~ 10 consensus sequence changes at - 12, CATAAT and
GATAAT. Another mutation in this region (glutamine 437 to histidine) also
responds to changes at position ~12. Since this region of c is thought to form
an «helix, glutamine 437 is one turn away from threonine 440, but on the
same helical face. These residues may contact the ~12 T in the consensus ~10
sequence TATAAT.
Subregions 2. 1 and 2.3 may be involved in the melting of promoter
region DNA, a conclusion based on the similarity of these regions with known

19

single-stranded DNA binding proteins (SSBs). Subregion 2.3 contains a
relatively high number of aromatic residues. Analysis of various known
prokaryotic SSBs suggests the importance of stacking between the bases and
aromatic protein residues (see Ollis & White, 1987, or Khamis et al., 1987).
Subregion 2.3 shares little sequence similarity with any of the prokaryotic
SSBs, but the abundance of aromatic residues in each suggests a similar
function for subregion 2.3. The sequence of subregion 2.3 is similar to the
aromatic consensus sequence found in many eukaryotic SSBs (Adam et al.,
1986; Swanson et al., 1987).

A similar line of reasoning suggests that subregion 2.1 may also bind
single-stranded DNA. This subregion, too, contains many aromatic residues.
Furthermore, the mutation of serine 389 to phenylalanine increases lactose
operon transcription in 3:in by about four fold (Siegele et al., 1988). Addition
of the aromatic residue may make subregion 2.1 more capable of interactions
with single-stranded region of melted DNA.

Region 3 is found in only a portion of the surveyed 6 subunits. In E.
call 670 this region extends from about residue 475 to 521. The function of
this region is not currently known. Region 4 spans from about residue 555 to
the subunit C-terminus, and can be divided into two subregions. The first
subregion, 4.1, is of unknown function. Subregion 4.2, however, is thought to
make contacts with the ~35 region of DNA promoters via a helix-turn-helix
DNA binding motif (Siegele et al., 1989; Gardella et al., 1989). In 2119
transcription from mutant promoters, as described above for subregion 2.4,
implies that arginine 584 and 588 contact bases in the ~35 consensus
sequence. Mutating arginine 584 to cysteine or histidine enhances
transcription fi'om promoters that are mutant within the ~35 region
(TTGACA). The arginine 588 to histidine mutation increases transcription

from-

 

tern;
motif

these:

 

20

from ~33 mutations ('I'I‘GACA) of the ~35 region. Arginine 584 is near the N-
terminus of the recognition helix of a proposed helix-turn-helix DNA binding
motif. Arginine 588 is near the middle of the recognition helix, roughly on
the same helical face as arginine 584.

Genetic analysis is a powerful tool for dissecting protein structure-
function relationships. It suffers the disadvantage that it is difficult to
determine precisely the efi‘ect of the amino acid sequence change. Is the
effect localized, or is there an allosteric conformation change involved? Given
the importance of the conformation of each of the subunits of RNA
polymerase in enzyme activity, physical approaches to studying its structure
are in order.

Catabolite Activator Protein. The search for the extra factor needed to
stimulate lactose operon transcription resulted in the discovery and
purification of the catabolite activator protein (CAP) (Anderson et al., 197 1;
Beckwith et al., 1972). lg 3119 transcription studies indicated that purified
CAP can significantly stimulate lactose operon transcription depending on
the concentration of cAMP. cAMP binding seems to unlock the specific DNA
binding activity of CAP (Krakow & Pastan, 1973; Takahashi et al., 1983).
Indeed, the conformation of CAP undergoes a change upon binding of cAMP
(Wu et al., 1974; Eilen et al., 197 8). CAP is relatively resistant to protease
attack in the absence of cAMP; in its presence, however, CAP is much more
readily digested. Fluorescence studies of labeled CAP also suggest a
conformational change upon binding cAMP. The structural change induced
by cAMP probably exposes the DNA binding region of the protein, as CAP
alone shows no specific DNA binding amnity. Thus CAP/cAMP binds to
promoter regions and stimulates transcription by RNA polymerase. The

21

molecular mechanism of CAP transcription activation is not fully understood,
and remains an intense area of study.

The molecular mass of CAP is 47.2KDa, composed of two identical
monomers. The solution of the crystal structure of the dimer plus cAMP (one
cAMP bound to each monomer) to 2.9A (McKay et al., 1982), and later to 2.5A
(Weber & Steitz, 1987), reveals several things about the dimer. Each
monomer contains two domains: a N ~terminal domain that binds cAMP and
within which the monomer-monomer contacts occur, and a C-terminal
domain with a helix-turn-helix DNA binding motif. The domains are
separated by a short hinge region. The dimer is asymmetric; one of the
monomers has a more compact structure than the other. Only one cAMP is
needed for CAP transcription stimulation (Garner & Revzin, 1982; Shanblatt
& Revzin, 1983), although the crystal structure implies that both cAMP
binding sites are available.

The crystal structure indicates that residues involved in cAMP binding
are located in a B—roll in the N-terminal region of the protein. The residues
that contribute to the binding of cAMP are glutamate 72, arginine 82, serine
83, arginine 123, threonine 127, and serine 128. Mutation of any of the first
four residues disrupts the ability of the protein to stimulate transcription
(Eschenlauer & Rezm'kofi', 1991; Moore et al., 1992), most likely by
disturbing cAMP binding.

Most of the intersubunit contacts of the dimer occur between two N ~
terminal helices (helix C). The two helices are slightly rotated (about 20°)
with respect to each other. Only three hydrogen bonds form between the two
monomers, and two of them are through the bound cAMP molecules. The
rest of the interactions are hydrophobic in nature. This suggests that the
conformation change induced by cAMP binding is brought about by the

 

CR“
\
, ~

E1"
9

its,
tL .

urc-

as»
VJ

22

formation of the two hydrogen bonds. The conformation change may be
transmitted to the C-terminal DNA binding domain by the hinge region.

Indeed, the hinge region (residues 134-139) appears critical to the
conformational change induced by cAMP. Mutation of asparagine 138 can
alter the cAMP-mediated conformation change m m and in. m (Kim et
al., 1992; Ryu et al., 1993). Three classes of asparagine 138 mutants were
observed: one class behaved like wild type CAP, another could bind cAMP
but not DNA, and the third could not bind cAMP or DNA. These classes
correlate with the polarity of the substitution. Polar substitutions resulted in
a wild type-like behavior, while nonpolar substitutions resulted in the
altered behavior described above.

The conformation change induced by cAMP probably alters the
orientation of the recognition helix (helix F) of the C-terminal helix-tum—
helix DNA binding motif (Kim et al., 1992). Substitution of alanine 144 with
a more bulky residue seems to mimic the activity of the CAP-cAMP complex
in yin, in that the mutant protein is independent of cAMP. The bulky
substitution may push the F helix away from another helix (the D helix) to
activate the DNA binding ability of CAP. A similar result was seen with the
substitution of glutamine for glycine 141 (Tan et al., 1991). cAMP binding
may do the same through the polar hinge region, moving the F helix away
from the D helix

How CAP-cAMP binds DNA is relatively well understood. From the
crystal structure of the protein, an electrostatic model of CAP-DNA
interaction was proposed (Weber & Steitz, 1984). The two F helices, as part
of the helix-turn-helix DNA binding region, fit nicely into the major groves of
the lactose operon CAP site. About 20bp of DNA was required to
accommodate both of the F helices, in agreement with the known length of

23

the consensus binding sequence. The addition of a bend in the DNA allowed
for more protein-DNA contacts, suggesting that CAP binding could induce a
DNA bend. It should be noted that while Weber and Steitz chose to interpret
their data via a bend in the DNA, a flex in the protein conformation may also
lead to additional protein-DN A contacts.

Many workers have since shown with a variety of techniques that CAP
does indeed bend DNA (Wu & Crothers, 1984; Porschke et al., 1984; Kotlarz
et al., 1986; Lin-Johnson et al., 1986). CAP binding to circularly permuted
DNA fi-agments resulted in large differences in mobility depending on the
podtion of the CAP site relative to the fragment ends. Fragments with a
CAP molecule bound to a site near one end ran further into the gel than
those with the CAP bound near the middle. In addition, the extent of
ligation of DNA minicircles containing a CAP site was markedly increased in
the presence of CAP and cAMP. Estimates of the degree of bending range
from 90° to 135°. The function of the bending is not clear, but it may allow
for extra protein-DNA interactions. In fact, about 30bp of DNA are required
to maximize the binding affinity of oligonucleotides for CAP, even though the
CAP consensus binding sequence is only about 20bp long (Lin-Johnson et al.,
1986). How CAP bends DNA was determined with the 3A solution of a CAP-
cAMP-DNA crystal (Schultz et al., 1991). The CAP-cAMP complex with 30bp
of CAP-site DNA revealed a bend of 90°, primarily final two 40° kinks. The
remaining 10° arises from a bend which allows an interaction between the
flanking DNA sequence and the protein. The DNA is bent toward the
protein, as suggested by the electrostatic model. Many of the residues of the
helix-turn-helix structure contact the DNA in agreement with the
electrostatic model. While the structure of the DNA is apparently highly
altered by CAP binding, the conformation of the protein is also likely

 

15 {alrj

5 less

proba

24

influenced. Extrinsic and intrinsic fluorescence data indicate that CAP does
indeed undergo a conformational change when binding to wild type lag DNA
(Wu et al., 1974; Takahashi et al., 1983).

While the influence of cAMP on CAP, as well as on CAP-DNA binding,
is fairly well understood, how CAP stimulates RNA polymerase transcription
is less clear. Exclusion of polymerase from a less active promoter site
probably plays a role, as described above. Moreover, DNA bending may play
a role, although the molecular details are obscure. DNA bending may
position CAP to contact RNA polymerase, thereby stimulating transcription
via protein-protein contacts. The CAP binding site at the lactose Operon,
however, can be replaced with naturally bent DNA sequences, which
stimulate RNA polymerase transcription from the lactose operon, although
not as well as does CAP (10 fold vs. 30 fold, respectively) (Gartenberg &
Crothers, 1991). The phasing of the bent DNA is critical, as is the position of
a bound CAP molecule (see above), indicating that the orientation of the bend
is important for stimulation to occur. The bend induced by CAP may allow
for RNA polymerase-upstream DNA contacts that are not otherwise
available. Zinkel and Crothers (1991) proposed that, in addition to the CAP-
induced DNA bend helping to form an open complex, the bend may store
energy to help break strong DNA-protein and/or protein-protein contacts
enabling RNA polymerase to more easily leave the promoter area during
transcription initiation.

Probably the main factors in transcription stimulation, however, are
CAP-RNA polymerase contacts. Several lines of evidence indicate that such
contacts are important for CAP-enhanced transcription. Pinkney and
Hoggett (1988) showed that CAP could interact with RNA polymerase in
solution. The anisotropy of fluorescently labeled CAP was altered in the

 

25

presence of RNA polymerase, suggesting the formation of a complex between
the two proteins. RNA polymerase binding is also known to stabilize CAP
binding to DNA (Mandecki & Caruthers, 1984). In the absence of RNA
polymerase, CAP is relatively easily dissociated from DNA. In the presence
of the protein, however, bound CAP is much less easily removed from the
DNA. A monoclonal antibody study provided additional evidence for CAP-
RNA polymerase interaction (Li & Krakow, 1987). The antibody bound to
CAP and CAP-cAMP quite well, but not at all to CAP in complex with DNA
and RNA polymerase, suggesting direct contacts between CAP and RNA
polymerase. Because of the bulk of the antibody, however, this result could
simply be due to polymerase sterically hindering antibody binding rather
than to tight protein-protein contacts. Ren et al. (1988) showed that a double
mutant CAP (arginine 142 to histidine and alanine 144 to threonine) could
stimulate transcription by RNA polymerase in vim or in vim in the absence
of cAMP, although the addition of cAMP enhanced the activity of the mutant.
DNase I protection experiments showed that the mutant CAP-cAMP complex
could bind wild type lag DNA in the absence of RNA polymerase. In the
absence of cAMP, however, mutant CAP binding could only be detected when
RNA polymerase was also present in the sample. These data, too, suggest an
interaction between the two proteins.

The portion of RNA polymerase with which CAP interacts appears to
be the on subunit (see above). Deletion analysis indicates that the C-terminal
region of the a subunit contacts CAP when both proteins are bound at a
promoter. Finally, mutations of CAP can reduce or eliminate in m and in
vim transcription stimulation while not interfering with the DNA binding of
the protein (Bell et al., 1990; Williams et al., 1991). Substitution of
glutamate 17 l with lysine reduces CAP transcription stimulation activity.

26

Changing histidine 159 to either leucine or isoleucine results in a complete
loss of transcription stimulation. Both of these residues may interact with
RNA polymerase, as the each mutant protein can still bind cAMP and
promoter region DNA.

The potential for different interactions between CAP and RNA
polymerase is quite possible, given the positions of CAP binding relative to
RNA polymerase at various promoters. The interactions at the lactose
operon may well be difi‘erent from those at the galactose operon where the
first CAP molecule binds at -40 (instead of at -60 as at the lag (see above)).
Deletion analysis of the or subunit of RNA polymerase indicates that CAP
does, in fact, interact with different areas of the C-terminal region of the
subunit, depending on the relative positions of CAP and RNA polymerase at
a promoter. The corresponding genetic analysis of CAP also reveals two
potential interaction surfaces. As mentioned above, mutation of histidine
159 eliminates the stimulatory activity of CAP without disrupting the DNA
binding activity of the protein (Bell et al., 1990; Williams et al., 1991). If
lysine 52 is mutated in addition to histidine 159, transcription stimulation
activity~ is restored, but only at the -40 CAP position. Transcription
stimulation from a -60 position is not reinstated, suggesting that different
areas of CAP interact with RNA polymerase depending on the relative
position of the two. In fact, the double mutant stimulates transcription
better from -40 than does the wild type protein. The conformation of CAP in
the complex may be different depending on which face of the protein
interacts with RNA polymerase.

Do CAP and RNA Polymerase Acquire Different Conformations and / or
Induce Different Bends At Various Promoters? Protein and DNA

conformations are surely critical elements in the transcription process. It

 

seer:
the:
p03:
the

—-——

ezxier.

Indus
of :12:
mix

prom

 

thus'u

tram

Chm
refiec
eXper
mm:
Como:
Probe
does -
56m
am:

We.

139105

mthe

27

seems likely that CAP and RNA polymerase have altered conformations
when bound at different promoter regions given the physical variation in the
positioning of the binding sites. Further, the DNA sequence may influence
the conformations of the proteins. After all, different DNA sequences are
essentially difl'erent ligands that could allosterically alter the proteins.
Induced DNA bending itself may be important in the molecular mechanism
of transcription, but there may also be an interplay between protein
conformation and DNA bending with each influencing the other. Studies of
protein conformation, as well as the DNA bending, at different promoters are
thus important for understanding of the molecular mechanism of
transcription and its regulation.

Fluorescence is a sensitive method to assess protein conformation.
Changes in protein conformation that alter a fluorophore’s surroundings are
reflected in altered fluorescence spectra. This thesis describes fluorescence
experiments to probe CAP and RNA polymerase when bound at various
promoter regions to determine whether the proteins do indeed have altered
conformations. The labeling of CAP and RNA polymerase with fluorescent
probes is detailed, as are experiments designed to show that such labeling
does not grossly alter either protein activity. The fluorescent moiety also
serves as an agent that can yield information about residues important to the
activity of the protein. Thus, considerable effort was expended to determine
precisely where the probes are covalently bound.

Finally, experiments designed to assess bending differences at the
lactose and galactose operons as detected by gel electrophoresis are
described. The data reveal interesting, if somewhat unexpected, differences
in the structure of transcription complexes at these promoters.

Chapter 2
Fluorescent Labeling of CAP and RNA Polymerase

Muslim;

The initiation of transcription is a multi-step process involving a
complicated series of interactions between RNA polymerase, DNA promoter
regions, and in many cases accessory proteins that can enhance or inhibit the
reaction. Elucidation of this process requires knowledge of the structures of
the complexes, and an understanding of the conformational changes that are
induced in the proteins and DNA as the complexes are formed. Studies of
CAP and RNA polymerase structure have, however, been rare due to the
often dificult and/or crude techniques available to probe protein
conformation. As a result, much of the research into DNA-protein
interactions has involved observing the effects of mutating the DNA (and in
some cases, the proteins).

Much of what has been done in this area has been with the use of
fluorescence as a monitor of protein conformation. Fluorescence is a sensitive
method to assess the conformation of proteins, as changes in protein
conformation can appear as altered fluorescent spectra. Takahashi et al.
(1983) studied the intrinsic fluorescence from CAP tryptophan residues
before and after the formation of CAP-lag promoter DNA complexes. An
increase in the intrinsic CAP fluorescence signal in the presence of Lag DNA
indicated the formation of complexes. Using intrinsic fluorescence has the
advantage of simplicity, but presents other problems. The wavelength region
for tryptophan absorbance overlaps that for the DNA. The DNA acts as an
inner filter, absorbing light that could instead excite the tryptophan residues.

Another difliculty for our system, is the size of RNA polymerase (about ten
28

 

 

 

DIODE
fluor
PFUH
alter
are I

DIES;

alien

29

fold greater mass than CAP). In the presence of both proteins, the intrinsic
signal from CAP would be masked by the polymerase fluorescence signal
(which is also ten fold larger). It would be diflicult to determine any
potential signal changes from CAP intrinsic fluorescence under such
conditions.

A way to avoid these limitations is to use extrinsic fluorescence
instead. With the appropriate choice of fluorescent probe, DNA absorbance
can be avoided. Indeed, Wu et al. (1974) labeled CAP with a fluorescent
reporter group (N-(iodoacetylaminoethyl)l-naphthylamine-S-sulfonate or
dansyl chloride), and observed a fluorescence spectral change when DNA
containing a wild type lac CAP site was added to the labeled protein. In
addition, conformational changes in one protein when in the presence of
other proteins can be monitored by labeling the protein of interest with a
fluorescent label. Wu et al. (197 5) labeled the 6 subunit of RNA polymerase
with dansyl chloride to follow 0 release from the holoenzyme during the
initiation of transcription.

This chapter describes the covalent modification of catabolite activator
protein (CAP) and RNA polymerase with fluorescent probes for use in
fluorescence studies of protein conformation at lactose and galactose
promoter regions. Because attachment of a fluorescent label to a protein can
alter the activity of the protein, studies are described that show that there
are not major changes in CAP or RNA polymerase activity due to the
presence of the probe. Finally, the potential sites of probe attachment are
discussed in light of the fact that the activity of each protein is not grossly
altered by the label.

30
MW

Proteins. CAP and RNA polymerase were isolated as previously
described (Garner & Revzin, 1981). Protein concentrations were determined
spectrophotometrically using extinction coefficients of 3.98 x 104 M'1 for CAP
and 3.0 x 105 M-1 for RNA polymerase at 280nm (Garner & Revzin, 1981).
Typical preparations were judged to be > 95% pure by SDS gel
electrophoresis. Purified protein was stored at -20°C in storage bufl'er
(10mM Tris, pH 8, 50% glycerol, 0.1mM EDTA, 0.1mM DTT, and either
100mM (RNA polymerase) or 200mM (CAP) N aCl). CAP preparations were
usually about 25% active in binding DNA and transcription stimulation,
while RNA polymerase stocks were about 10% active in DNA binding and
transcription (Garner 8: Revzin, 1981). The reason why the purified proteins
are less than 100% active is unclear.

DNA Fragments. Two related DNA promoter fragments were used in
characterizing the fluorescently labeled proteins. The first was a 203 base
pair (bp) lactose wild type promoter fragment isolated fiom a pMB9
recombinant plasmid kindly provided by Forrest Fuller (see the Appendix for
the sequence of this fragment and other DNA fragments used in this work).
This DNA fragment extends fiom -140 to +63, with a CAP binding site at
roughly -60 (all numbering is relative to the transcription start site at +1),
and two RNA polymerase sites (P1 and P2) which initiate transcription at +1
and -22, respectively. Transcription fiom P1 by RNA polymerase is
dependent on the CAP-cAMP complex binding, while P2 transcription is
repressed when the CAP-cAMP complex is bound. The other DNA promoter
fragment used is a 203bp lag double mutant called L8-UV5 (again, the
sequence can be found in Appendix). The UV5 “up” mutation in the - 10
region of P1 makes transcription by RNA polymerase fiom this site

31

independent of the presence of CAP-cAMP. The L8 mutation in the CAP
binding site reduces the affinity of CAP-cAMP for the site.

Purified DNA was stored at 4°C in TE (20:0.1) buffer (20mM Tris, pH
7.5, 0.1mM EDTA). Absorbance spectra were taken on the purified DNA
stock solutions to determine concentrations, using an extinction coefficient of
1.3 x 104M‘1bp'1at 260nm (Felsenfeld & Hirschman, 1965). Where needed,
DNA was end-labeled with 32p by calf intestinal phosphatase and T4 kinase
treatment using 7-32P-ATP (Maniatis et al., 1982). The activity of the end-
labeled DNA was determined by diluting lul of the DNA into 5ml of
scintillation fluid (9.5g 2,5-diphenyloxazole in 1425ml toluene and 473ml of
absolute ethanol) and measuring the radioactivity in a liquid scintillation
counter

. Protein-DNA Binding Reactions. Binding reactions involving CAP

and/or RNA polymerase with a DNA promoter fragment were performed as
follows. Proteins were added to binding buffer (40mM Tris, pH8, 10mM
MgClz, 0.1mM EDTA, 20uM cAMP (if CAP was present), and 100mM KCl)
containing the required amount of DNA for a total reaction volume of 25111.
The reactions were incubated at 37°C prior to loading onto a polyacrylamide
gel (see below).

When CAP was involved, a 5- to 20-fold excess concentration of protein
(over that of promoter DNA) was incubated for 15min with the DNA. RNA
polymerase, at a 10- to 20-fold excess of protein, was incubated for 30min
with promoter DNA. Where both CAP and RNA polymerase were present in
a sample, CAP was added and incubated prior to RNA polymerase addition.

Gel Retardation Assay. DNA-protein complexes were separated from
free DNA by gel electrophoresis as described by Garner and Revzin (1981).
After complexes were formed as described above, 1/10 volume of 10X DNA

 

 

DNA
king
‘1“ e.
“Gite.
‘0 im.
1‘1)!

3030

32

loading buffer (25% Ficoll (w/v), and 0.25% (w/v) each of bromophenol blue
and xylene cyanol) was added to the samples. The samples were then
carefully placed into a gel well, and the nondenaturing gel was subject to
electrophoresis at 10V/cm for 1.5h at room temperature with a TBE running
buffer (90mM Tris, pH 8.3, 90mM boric acid, 2.5mM EDTA). When samples
containing CAP were run, 2011M cAMP was added to the gel during casting
and to the running bufl‘er. Heparin was added to a final concentration of
100ng/ml to RNA polymerase samples prior to adding the loading buffer.
Heparin removes nonspecifically bound RNA polymerase and prevents free
polymerase from binding to DNA, but does not perturb preformed open
complexes.

DNA bands were visualized either by ethidium bromide staining or by
x-ray film exposure fi'om radioactive DNA. For ethidium bromide-stained
gels, the amount of DNA used was between 100ng and 500ng per reaction.
In experiments requiring radioactively labeled DNA, 2000 to 3000cpm of end-
labeled DNA (convenient for overnight film exposure at -70°C using an
intensifying screen) and 50ng of the unlabeled counterpart were used. This
amount of end-labeled DNA was negligible compared to the nonradioactive
DNA in the sample.

Runoff Transcription Assay. Polymerase-DNA or CAP-polymerase-
DNA complexes were formed at 37°C in binding buffer as described above
using lOng of promoter DNA in a 10111 reaction volume. After the
transcription complexes were formed, 2.5ul of a 5X ribonucleotide mix
containing heparin and the four ribonucleotides in binding buffer was added
to initiate transcription; concentrations in the 5X ribonucleotide mix were
1mM each ATP, GTP, and GTP, 125nM UTP, 5nCi a-32P-UTP
(3000Ci/mmole) per reaction, and 500ng/ml heparin. The transcription

33

reaction was allowed to proceed for 10min at 37 °C before being quenched by
an equal volume of 2X urea stop buffer (10M urea and 0.25% (w/v)
bromophenol blue and xylene cyanol). Prior to loading onto a denaturing gel
(Maniatis et al., 1982), all samples were heated at 90°C for 5min and quick
chilled on ice. The samples were then loaded onto the gels, and the gels run
at 55°C and constant power (50W) for 2-3h using TBE as the running buffer.
Either lac" or lagLS-UV5 promoter DNA was used, which results in a P1
transcript of 63nt and a P2 transcript of 85nt.

Fluorescent Labeling of CAP and RNA Polymerase. Two fluorescent
probes were used in this work. Eosin isotbiocyanate (EITC) and lucifer
yellow iodoacetamide (LYIA) are shown in Figure 2. 1, and important spectral
properties are given in Table 2.1. Each probe can react covalently with
amino acid residues according to the following reactions.

For EITC (Cherry et al., 1976)

s
R-N=C=S + NH,—P= R—NH—C-NH—Ror

S
R-N=C=S + HS-P = R—NH-é—s—P,
while the LYIA reaction is 9 :.-

Chemicals. Haugland. 1992)
R-CHz-I + HS-P = R-CHz-S-P + HI,

 

where R represents the fluorescent probe and P represents protein. EITC is
likely to attack primarily lysines, arginines, glutamines, asparagines, and
cysteines, while LYIA is cysteine-specific. The residues that react are
presumably on or near the surface of the protein. As each probe is light

 

34

Figure2 ..1 Structures of Eosin Isothiocyanate (EITC) and Lucifer Yellow
Iodoacetamide (LYIA). The structure of EITC rs fiom Cherry et al. (1976),
andLYIAfromtheHMuo . _ -_ . i; h'u ,:
(Haugland1992).

   

35

EITC - 704.9 g/mole

 

LYIA - 503.3 g/mole

iliHCHzcrizNHEcnzi

N
Q // \ ,0

//1 /I\

~/ \. / \
/ \\ V. \
i
. . /” \\ //

 

36

Table 2,1. Spectral Properties of EITC and LYIA. The absorbance and
fluorescence maxima wavelengths for fi'ee label, as well as the em” values,
are from the H k f Fl re n Pr Research h mi
(Haugland, 1992). Because both probes have a significant absorbance at
280nm, the concentration of protein in labeled protein stock solutions was
determined by subtracting out the contribution of the label from the overall
absorbance. The extinction coefficients at 280nm for each probe were
calculated using the formula 5280/8max = A280/Amax' 8 units are M'l.

EITC

LYIA

37

 

 

 

Absorbance Fluorescence em,“ 6280
Marn'mum (nm) Maximum (nm)
522 543 8.3 x 104 2.4 x 104
430 530 1.3 x 104 3.0 x 104

 

 

 

 

 

38

sensitive, such exposure was minimized when working with probe stock
solutions or labeled protein.

Eosin isothiocyanate was attached to CAP and RNA polymerase using
the method described by Austin et al. (1983). Briefly, 2mg to 4mg of CAP or
RNA polymerase was dialyzed against labeling bufi‘er (100mM N azCO3,
pH 9, 50% glycerol) to remove the Tris-based protein storage buffer. Freshly
prepared EITC (usually 10mM) in labeling buffer was then added to the
solution at 2.5 fold greater concentration than the protein. The resulting
solution was vortexed quickly and placed at -20°C for 3 days. The labeling
reaction was quenched by addition of Tris, pH 8, to a final concentration of
10mM.

Excess EITC was removed by passing the solution through spin
columns made with Bio-Gel P-6 swollen in TGED (10mM Tris, pH 7.5, 5%
glycerol (v/v), 0.1mM EDTA, 0.1mM DTT) + 100mM NaCl (for polymerase) or
200mM N aCl (for CAP). The columns were made by pouring enough Bio-Gel
P-6 into 5m] syringes, with glass wool plugs, for a 4ml column volume. The
columns were then washed with 5 column volumes of the appropriate TGED
buffer. The columns were placed into 15ml Corex round-bottomed centrifuge
tubes and spun once in a Sorvall SS34 fixed angle rotor at 3,500rpm at 4°C
for 4min in a Sorvall RC-5B Refrigerated Superspeed Centrifuge. This
procedure compacted the column bed and removed much of the TGED bufl‘er
(to avoid sample dilution). The labeling solution was layered onto column,
which was then centrifuged as above. Excess EITC entered the Bio-Gel P-6
and remained in the column, while the labeled protein passed through and
was collected. This was repeated at least once to ensure complete removal of
any excess probe. During this separation, the labeling buffer was exchanged
for TGED buffer.

39

A slightly modified procedure was used to label CAP with lucifer
yellow iodoacetamide. Again, 2mg to 4mg of CAP was dialyzed against
labeling bufl‘er to remove the DTT in the protein storage buffer which might
react with LYIA. Freshly prepared LYIA (usually 1mM) in labeling buffer
was added at 2.5 times the protein concentration. The labeling reaction was
then incubated at room temperature (instead of at -20°C) for 3 days. Excess
LYIA was removed as described above for EITC. Each labeled protein was
aliquoted and stored at -20°C after determination of labeling ratios by
absorbance (see below).

Attachment of the probes to the proteins was confirmed by running
aliquots onto SDS gels with unlabeled protein and free probe as controls.
Between 30ug and 50ng of labeled protein (enough to see the probe migrate
through the gel), 10ug of unlabeled protein, and 50uM free label were placed
in equal volumes of 2X SDS loading buffer (188mM Tris, pH 6.8, 30%
glycerol (v/v), 9% SDS (w/v), 15% 2-mercaptoethanol (v/v), and 0.1%

 

bromophenol blue (w/v)). The samples were then heated at 90°C for 15min
and loaded onto a 7.5% (polymerase) or a 15% (CAP) polyacrylamide gel.
SDS-PAGE was run at room temperature at 15V/cm with SDS running buffer
(50mM Tris, pH 8, 380mM glycine, and 0.1% SDS (w/v)). The gel was
carefully monitored since both EITC and LYIA bleach during electrophoresis.
After appropriate separation, and before probe bands faded to
background (usually 1.5-2.5h), electrophoresis was stopped, and the gel was
removed fi'om the electrophoresis apparatus. The wet gel was placed on a
light box to better observe the probe bands. Migration distances of the visible
probe bands were then measured from the top of the gel. The gel was then
stained as usual with Coomassie brilliant blue R. After suitable destaining,
the migration distance of the protein bands was also measured hour the top

 

 

 

mice
ni‘
135E114.
Sit
and i

Sm

40

of the gel. Since the gels expanded during the staining process, a correction
ratio was calculated from the total length of the gel measured before and
after staining. The migration distances were then compared to determine if
the probe ran with the protein bands.

Fluorometer and Sean Conditions. The fluorescence of the labeled
proteins was monitored using a unique fluorometer built by Dr. Jack Holland
(Holland et al., 1973). This instrument simultaneously measures absorbance
and fluorescence, to produce relative fluorescence emciency data, and
corrects fluorescence spectra as well so that dilution of the sample is not a
problem. Samples (in binding bufl'er) used with this instrument had
concentrations of fluorescent label of 1M. EITC-labeled protein emission
was monitored from 520nm to 580nm, with excitation at 500nm. LYIA-
labeled protein scan parameters were 500nm to 570nm for the emission, and
430nm for the excitation. The excitation and emission slits were set at 3nm
and 10m, respectively, for all scans. Each spectrum consisted of only one
scan pass, as this instrument cannot as yet produce spectra averaged from
many scans. However, each experiment was performed at least twice so that
spectra could be averaged by hand. All scans were performed at room
temperature.

HPLC Chromatography. Protein samples needing fractionation were
run over reverse phase HPLC columns. The Waters 600 HPLC instrument
was connected to a Waters 990 Photodiode Array detector to monitor column
eflluent anywhere fi'om 200nm to 600nm. Real time data collection, and off
line chromatograph analysis, were controlled by Waters 990 software and
data acquisition hardware.

Protein or peptide samples were loaded onto a C- 18 column for reverse
phase separation. Reverse phase gradients were from 100% 0.1%

 

41

trifluoroacetic acid (v/v) to 90% acetonitrile (v/v) over 90min with a 1ml/min
flow rate. Blanks were run before and after sample runs to ensure clean
columns.

Column eflluent was monitored at 490nm for EITC label or 430nm for
LYIA label, in addition to 2 14nm, 230nm, and/or 280nm for peptide content.

In the reverse phase buffer system, the absorbance maximum of EITC shifts

to 490nm. Fractions showing both protein and label absorbance were 7:
collected, reduced in volume using a Savant speed-vacuum if necessary, and
stored at -20°C.

Labeling Ratios. The number of probes attached per molecule of
protein was determined primarily by absorbance, but was checked by SDS a,
gel and mass analyses. The absorbance method assumes that the extinction
coeficient of each probe is not altered when attached to protein. SDS gel and
mass analysis checked the validity of this assumption.

To determine labeling ratios by absorbance, the absorbance spectrum
of each labeled protein was taken from 230nm to 700nm. Probe
concentrations were calculated from the absorbance maximum of 530nm for
EITC and 430nm for LYIA using the extinction coefficients fiom 522nm and
430nm, respectively (see Table 2. 1). Using these concentration values and
the calculated 280nm probe extinction coefficients, the absorbance
contributions at 280nm were determined. These absorbance values were
then subtracted from the overall 280nm absorbance values to obtain the
absorbance due to protein. The concentration of protein was then easily

 

determined fiom the known extinction coemcient at 280nm.

Since the absorbance method involved an assumption, labeling ratios
were check by two other methods. Molecular masses of unlabeled CAP and
LYIA-labeled CAP were determined from mass spectrometry and compared

42

with the known mass of LYIA and CAP. Spectra were obtained fiom the
Michigan State University Mass Spectrometry Facility using the laser
desorption mass spectrometry technique (Hillenkamp et al., 1991; Chait et
al., 1992). Briefly, unlabeled or LYIA-labeled CAP was mixed with sinapinic
acid and dried onto a probe tip. The tip was inserted into the mass
spectrometer and irradiated with short duration laser pulses. Energy from
the laser caused the matrix to vaporize, taking the protein with it into the
gas phase. During this process, the protein picked up one or more hydrogen
ions and became charged. The time-of-flight to reach a detector was recorded
and correlated to mass with internal protein standards. Prior to mass
determination, each protein was purified from bufl'er components that
interfere with this technique (primarily salt) by reverse phase HPLC. For
unknown reasons, determination of mass by the above approach did not
succeed for EITC-labeled CAP. Thus, SDS gel analysis was used to check the
labeling ratios calculated by absorbance.

If the assumption of unchanged probe extinction coefficients is wrong,
then label and protein concentration calculations based on that assumption
would be erroneous. A way around this problem is to use SDS-PAGE to
compare a sample with a known amount of unlabeled protein with a sample
containing presumably the same amount of labeled protein, based on
absorbance calculations. If similar, then the assumption of unaltered probe
extinction coefficients, and the concentration calculations based on the
assumption, is valid. EITC-labeled CAP and LYIA-labeled CAP were thus
run into a 15% polyacrylamide SDS gel, along with a known amount of
unlabeled CAP. In each case, samples containing 30ng and 60ng of protein
were run into the gel. After running the gel at 15V/cm for about 4h, the gel
was silver stained and dried between cellophane sheets. Densitometry scans

 

43

were then made on the dried gel with a Hoefer Scientific GSSOO
Transmitiance/Reflectance Scanning Densitometer, using GelScanner data
acquisition software. Each lane was scanned perpendicular to the band three
times: once at each end and once in the middle. These values were then
averaged to obtain an overall value for each protein band. This experiment
was performed twice for each labeled protein and the results averaged. The
average values were then compared to determine if similar amounts of

labeled and unlabeled CAP had been run into the gels.

 

44

Results

CAP Is Labeled with EITC or LYL4, as Is the aSubunit of RNA
Polymerase. Each labeled protein was analyzed by SDS polyacrylamide gel
electrophoresis to ensure probe attachment. The colored EITC and LYIA
bands ran with the 23.6KDa CAP monomer band (data not shown),
indicating probe attachment to the protein (free EITC and LYIA run at the
dye front). In the case of RNA polymerase, EITC was found to bind primarily
to the a subunit. This was quite surprising, considering that the other
polymerase subunits probably present large numbers of available surface
residues that could react with EITC.

Probe Attachment Changes the Spectral Properties of El TC and LYIA.
Figure 2.2 shows the absorbance of EITC and LYIA when free in solution and
when covalently linked to either protein. EITC attachment to either CAP or
RNA polymerase shifts its absorbance maximum from 522nm to 530nm.
LYIA attachment to CAP, on the other hand, does not shift the absorbance
maximum. The relative fluorescence efficiency of each labeled protein is
shown in Figure 2.3. EITC-labeled protein spectra show a slightly red-
shifted maximum, but otherwise does not differ significantly from the
spectrum of the corresponding free probe. The LYIA-labeled CAP spectrum
exhibits a small blue-shift, compared to the spectrum of free LYIA.

A shift to longer absorbance and fluorescence wavelengths is
consistent with the notion that the probes attach to the surface of the
proteins. Once attached to the protein surface, EITC is held exposed to the
solvent. In the case of the LYIA-labeled CAP, LYIA might be in a surface
recess that is somewhat less solvent exposed than where EITC attaches to
CAP, giving rise to the small blue-shift. Surface attachment is not

 

45

Figure 2,2. Absorbance Spectra of Labeled CAP and RNA Polymerase. The
absorbance spectra of fi'ee EITC and LYIA are also shown. Panel A shows
the EITC-labeled CAP spectrum. Panel B shows the LYIA-labeled CAP
spectrum. Panel C shows the EITC-labeled polymerase spectrum. Abs
denotes absorbance.

46

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

A.
1.6
— — — — Free EITC
1.2 , EITC-CAP
Abs 0.8
0.4 »
3‘ ___
0 —-———-—J
250 350 450 550 650 750
Wavelength (nm)
B.
1.2
————— Free LYIA
LYIA-CAP
0.8 »
a. / \
0.4 l /\ .
I ' \
i \ 3
\ \
‘\\ /"\\
x‘ . 34/ wk
0 ‘ ' ‘7 * ‘ — — ‘ ' ‘ — fl "
250 350 450 550 650 750
Wavelength (nm)
1
/ \
0.3 , / \ ————— Free EITC
/ \ EITC-RNA
0.6 f Polymerase
m
0.4 ‘ I], \\.
\ // \\
"‘\ / / \
0'2 > \x \‘ //.»‘ \i \\
0 ‘ ‘ ‘ - ~ — - - 4“»—
250 350 450 550 650 750
Wavelength (nm)

47

Figm 2.3. Relative Fluorescence Efficiency Spectra of Labeled CAP and
RNA Polymerase. The spectrum of the corresponding free probe is shown for
reference along with each labeled protein . Panel A shows EITC-labeled
CAP, Panel B shows EITC-labeled RNA polymerase, and Panel C shows the
LYIA-labeled CAP spectrum. RFE denotes relative fluorescence efficiency.

48

 

 

   
 

 

 

 

 

 

 

 

 

 

 

 

 

 

A.
40
30 r ,1/
’// {/,/
.’ /
RFE 20 - / EITC-Labeled
/ / CAP
, / ----- FreeEI'I‘C
10 l / /
V
0 4L _4 A
520 530 540 550 560 570 530
Wavelength (nm)
B.
30
/”"'”
20 » .\
// ,/ \‘\\\
RFE EITC-Labeled \\\
/ RNA Polymerase \-\
10 * ’ ----- Free EITC ‘\
1/
0 a L a
520 530 540 550 560 570 580
Wavelength (nm)
C.
160
120 i /’/:"’/ \\
///,/’/ \‘\\ “~~.\ ________
RFE 80 r/ /’ LYIA-Labeled \\
CAP
' ————— Free LYIA
40 »
0 A #1 ._r n n r
500 510 520 530 540 550 560 570

Wavelength (nm)

49

surprising considering that both probes are bulky and would thus be unlikely
to penetrate the interior of the protein.

CAP and RNA Polymerase Are Labeled with One Probe Per Molecule on
Average. The number of labels per protein molecule was determined in three
ways. The first method calculated probe and protein concentration from the
absorbance data of the labeled protein. The results suggest that on average
each protein molecule is labeled with one probe. This method assumes that
the probe’s extinction coemcient remains relatively unaffected after
attachment to a protein.

As a check on the possibility that the labeling ratios calculated from
absorbance data were in error, two other methods were used to determine the
number of labels per protein molecule. Mass analysis by laser desorption
allows for determination of the mass of whole proteins. Analysis of unlabeled
CAP showed several peaks (see Figure 2.4 and Table 2.2), each of which
corresponds to CAP monomer (the calculated monomer mass is 23,624Da).

N o dimer was observed at 47,248Da.

Figure 2.4 also shows the mass spectrum of LYIA-labeled CAP. Two
[M-i-2H'J"+ peaks, and two [M+3H]+++ peaks were observed, but no [M-i-H]+
peak was seen. The calculated mass of one molecule of lucifer yellow (ionic
form) is 457.4. One-half this value is 228.7, while one-third is 152.7. The
difference between the [M+2H]++ peaks and the [M+3H]+++ peaks
corresponds well with these values. This indicates that CAP was labeled
with one lucifer yellow molecule per CAP molecule. At least for LYIA-labeled
CAP, the assumption of a constant extinction coefficient appears to be valid.
However, for unknown reasons the mass of EITC-labeled CAP was not
observable with this technique.

 

50

Figure 2.4. Mass Spectra of LYIA-Labeled and Unlabeled CAP. Laser
desorption mass spectrometry of LYIA-labeled CAP and unlabeled CAP was
performed as described in Materials and Methods. The intensity scale is set
in arbitrary units. Panel A shows the unlabeled CAP spectrum. The peak at
11,476Da is a dimer of the internal insulin standard. Panel B shows the

LYIA-labeled CAP mass spectrum. The insulin dimer in this spectrum is at
11,465Da.

51

 

.TIS . tithes has.

t; :16

 

 

.2133 not. - use: -

than $0

938 - m

 

~ tel. 3. st: G‘u n."05l
0v 0....

 

-7500
-76004

33:85

 

-77OO ‘

 

fill-‘03;
one: . 55 . Eu

 

 

 

..c
titties m
fl‘s cl
1 “Ghana’subo‘u "OB—p In“

 

m p ltwv

~¢ F511}: .5135; “0'— p 0

11600

   

palate
+Ifl1n+ «H3 ..IU

”nah-
is. s ..

 

 

 

~7‘IOO‘
-7150‘

982:.

52

Table 2.2. Mass Spectrum Analysis of Unlabeled and LYIA-Labeled CAP.
The molecular weight of a CAP monomer is 23,624Da. The mass of LYIA is
457.4Da (ionic form).

 

53

[Mi-H]+ [M+2H]++ [M+3H]+++ Predicted Mass

 

CAP 23,538 1 1,763 7,845.7 23,624

 

 

 

 

 

LYIA-labeled CAP ND 11,987 7,992.8 24,127

 

 

ND = not detected

54

As a further verification on the number of probes per protein molecule
(especially for EITC-labeled CAP), the label and CAP concentrations
calculated fiom the absorbance data were checked by running a presumed
amount of labeled protein (determined from the absorbance-derived protein
concentration) into a SDS-PAGE gel along with a similar known amount of
unlabeled CAP. Table 2.3 shows the results of densitometry analysis of such
SDS gels. As can be seen from the data, the intensifies of both EITC-labeled
and LYIA-labeled CAP bands agree well with the unlabeled control.
Therefore, it is likely that the concentration values determined from the
absorbance data are accurate. The assumption of a constant extinction
coefficient for EITC, as well as for LYIA, appears valid, and the conclusion
that there is one label per protein molecule on average, as indicated by the
absorbance data, seems correct.

CAP Is Heterogeneously Labeled By EI TC. A surprising result fiom
the HPLC purification of the proteins prior to mass analysis was the
fi'actionation of EITC-labeled CAP. As can be seen in Figure 2.5, LYIA-
labeled CAP runs as a single peak indicating the presence of only one species
of labeled protein (Panel B). EITC-labeled CAP (Panel C), however, runs as
three peaks (a fourth peak is probably unlabeled CAP in this particular batch
of labeled protein, which happened to have a ratio of 0.5 EITC molecules per
protein). This suggests that three species of EITC-labeled CAP are present.
The major species (fraction 4) is likely labeled with one probe per protein
molecule, since the absorbance data indicate such a labeling ratio. The other
two minor species may be labeled with more than one probe molecule per
protein.

The Specific DNA Binding Activity of CAP and RNA Polymerase Is Not
Grossly Altered by the Fluorescent Labels. CAP and RNA polymerase

 

55

Table 2,3. Densitometry of Labeled and Unlabeled CAP Run into SDS-PAGE
Gels. The values presented (in arbitrary units) represent the average of two
separate experiments. The errors represent the average high and low values.

 

56

 

Expected Unlabeled CAP EITC-labeled LYIA-labeled
Amount Loaded CAP CAP
30 rig 2,4001800 1,900:i:200 2,3001800

 

 

 

 

 

60 3L 4,200: 1,000 4,1001600 4,000: 1,100

 

 

 

57

Figjge 2,5. Reverse Phase HPLC of Labeled and Unlabeled CAP. Reverse
phase HPLC of CAP was performed as detailed in the Materials and Methods
section (0.1% trifluoroacetic acid to 90% acetonitrile gradient over 90min).
Panel A shows the unlabeled CAP chromatograph at 214nm, 230nm, 254nm,
and 280nm. Panel B shows the LYIA-labeled CAP chromatograph at 214nm,

280nm, and 430nm. Panel C shows the EITC-labeled CAP chromatograph at
214nm, 280nm, and 490nm.

 

~ I- ! I: -

58

.~
A a.o.o .-

,. I‘ b '
g. 0.0.5.4-..A—asoomd.A.e-a-A-n-s.‘.o

. - on Q u .- .a .c R - n .

AAAAAA . . AAAA“-‘-‘. ‘.

 

AAAAAAAA

 

 

 

 

 

 

.‘l

 

I

'v‘vr0“--v-v'v-v'.v.-v~v-v<--'c~-o'r-'-r"t_'~~v-v-v31"" .-‘.. ..‘...-‘.v- .‘.vv.‘-.v.‘.-..év. . .‘vvv i""fl' -.-‘-. .‘.---'

-

m'nm'

 

 

 

 

 

 

“g

 

 

mp.

 

 

 

Mn 8‘ I
o-m S"

v

 

 

V'Vvvv vv v vvy vvvvlvvvvl‘v'v vvvvlv'vvlvvvv vvuvlvvvv vvvv 'vv‘ 'vv‘ vvvv

O. P F

no B

In?!“
900‘"

awn."

59

holoenzyme have several activities. Each binds specifically to DNA promoter
regions, and each participates in transcription initiation. Because the
covalent modification of CAP and RNA polymerase with EITC or LYIA could
interfere with protein function, the activity of each labeled protein was thus
determined in two assays. The first is a gel retardation assay to test the
specific DNA binding activity of each protein, and the second is a runoff
transcription assay to test labeled protein activity in transcription.

Gel retardation assays show that EITC-labeled and LYIA-labeled CAP
can bind Inf DNA to similar levels as the unlabeled protein (see Figure 2.6,
LYIA-labeled CAP data not shown). The cAMP dependence for specific DNA
binding of EITC- and LYIA-labeled CAP was also studied with gel binding
assays. Labeling of CAP with either probe does not appear to alter the
requirement for cAMP (data not shown). Moreover, EITC-labeled RNA
polymerase can bind to the lacL8-UV5 DNA fragment to the same level as
the unlabeled protein, but requires lower salt concentration (30mM vs.
100mM) to bind at similar levels to lac“ DNA (Panel B of Figure 2.6).
Attachment of either EITC or LYIA does not appear to alter markedly the
overall DNA binding activity of the proteins.

The Transcription Activity of Labeled Proteins Is Not Greatly Afi‘ected
by the Presence of the Label. Transcription by the labeled proteins was
studied to assess the effect of introducing a fluorescent moiety on their ability
to participate in the initiation of mRN A synthesis. Figure 2.7 shows runoff
transcription from Inf or lacLS-UV5 DNA with labeled CAP and RNA
polymerase. With these DNA fragments, transcripts initiated at P1 are 63nt
long, and those initiated fiom P2 are 85nt long.

60

Figge 2.6. EITC-Labeled CAP and RNA Polymerase Bind to Promoter DNA.
Complexes were formed at 37°C and separated on 4% nondenaturing
polyacrylamide gels as detailed in Materials and Methods. Panel A shows
various EITC-labeled RNA polymerase-promoter DNA and EITC-labeled
CAP-promoter DNA complexes, while Panel B shows a salt titration of EITC-
labeled RNA polymerase binding at the lac" P2 promoter.

61

A.
EITC-CAP - - — - - + + -
CAP - - - - + - - +
EITC-Polymerase - + - + - - - -
Polymerase + - + - + + - -
203bp Lac+ - - + + + + + +
203bp LacLB-UV5 + + - - - - - _

- - ”- .— .— 4— Polymerase

Complexes

Complexes

'-.~-_-__ -' {-Free DNA

B.

KC] (mM) 100 10 20 30 40 50 75 100 150 200
POIymerase + - - - - - - - - -

EITC- -+++++++++
Polymerase

- w - - - F -—- «Polymerase

Complexes

-------... «Free DNA

62

Figm 2,7. Transcription by Labeled CAP and RNA Polymerase. Complexes
were formed and transcription initiated at 37°C as detailed in Materials and
Methods. Transcription was performed with by“ or lagUV5 promoters at the
indicated salt concentration. Transcripts from P1 (63nt) and P2 (85nt) are
shown. Panel A shows transcription in the presence of labeled CAP (with
unlabeled polymerase), and Panel B shows transcription by labeled RNA
polymerase .

LYIA-CAP + - - -
Polymerase + + + +

KCl

CAP

EITC-CAP
Polymerase
EITC-Polymerase
203bp Lac+

203bp LacLS-UV5

I++I+

{—P2 (85nt)

'gg (—P1 (63nt)

 

30mM 100mM
+-+---
--++—+
++--+-
++++--
----++

c

Q . <—P2 (85nt)

..,.”<—P1 (63nt)

64

Labeled CAP is active in P1 transcription stimulation. Each labeled
CAP correctly represses P2 transcription while stimulating P1 transcription
with lag? DNA at 100mM salt and at 37°C. The level of P1 stimulation is
similar to that seen with unlabeled CAP. Thus EITC and LYIA are attached
to CAP in areas that are apparently not critical for interaction with RNA
polymerase, otherwise transcriptional activity would most likely drastically
decrease.

It can be seen that EITC-labeled RNA polymerase correctly transcribes
fi'om the strong lacLS-UVS promoter at 100mM salt and at 37°C (Figure 2.7,
Panel B). At the wild type lac promoters, the labeled polymerase correctly
transcribes fiom P1 or P2 depending on the presence of CAP at 30mM salt
and 37°C. In the absence of CAP, labeled RNA polymerase transcribes from
P2, as does the unlabeled enzyme. In the presence of CAP, EITC-labeled
RNA polymerase does not initiate from P2 as expected.

The maximum level of transcription from the lagUVS promoter is
similar for the labeled polymerase and for the unlabeled protein. However,
EITC-labeled RNA polymerase exhibits lower levels of transcription fi'om the
wild type lac P2 promoter than does the unlabeled polymerase. Further, the
presence of CAP at the wild type lac promoter seems to drastically reduce the
level of P1 transcription. The labeling of the a subunit may alter the ability
of RNA polymerase to interact with CAP at this promoter. Curiously, the use
of EITC-labeled CAP, instead of unlabeled CAP, seems to restore at least
some of the transcripts from P1. The modification of CAP appears to
compensate for whatever alteration that a labeling induces in the polymerase
holoenzyme.

65
D' .

Selection of Extrinsic Fluorescence. Many fluorescence studies of
protein conformation have relied on intrinsic fluorescence of tryptophan and
tyrosine residues. In the case of CAP and RNA polymerase, intrinsic
fluorescence presents two limitations. The first is that DNA absorbs in the
region for tryptophan and tyrosine, and thus DNA would act as an inner
filter. The second limitation is that where CAP and RNA polymerase are
both present, the fluorescence signal of CAP would be masked by the RNA
polymerase signal. Therefore, extrinsic fluorescence was chosen to study
CAP and RNA polymerase conformation when bound to DNA.

CAP and RNA polymerase were labeled with either eosin
isothiocyanate or lucifer yellow iodoacetamide. Covalent attachment of EITC
and LYIA to CAP and RNA polymerase was achieved with a relatively simple
procedure. For each protein, labeling stoichiometries were 1:1 on average.
The limited alteration of labeled protein spectral properties compared with
free probe suggests surface residue probe attachment.

RNA polymerase is apparently only labeled in the a subunit. This is
quite interesting, as it suggests that one a residue is more reactive with
EITC than the many other potential residues of the holoenzyme. One
residue of a is presumably quite solvent exposed, and thus easily reacts with
EITC. Once the probe is covalently attached, a slight conformation change in
a may make any other potential residues (in the entire protein) less reactive
toward free EITC.

Labeled Proteins Are Active in DNA Binding and Transcription
Initiation. The major problem with fluorescent modification of proteins is the
potential to significantly disrupt activity. A labeled, but inactive, protein is
not of much use. Two activities of CAP and RNA polymerase-DNA binding

66

and transcriptionuwhere thus checked to observed the effect of fluorescent
labeling. Each labeled protein binds promoter DNA to similar levels (using
the same amount of protein) as unlabeled protein. The need for similar
amounts of labeled and unlabeled protein indicates that labeled protein
molecules are binding to the promoter DNA If the labeled protein were
inactive and the DNA binding due to contamination by unlabeled molecules,
then a much higher concentration of labeled protein would be needed to
achieve the same amount of DNA binding. As this is not the case, we
conclude that the labeled protein does not have a substantially altered
overall DNA binding capacity.

The lower salt required to achieve strong binding of labeled RNA
polymerase to the wild type lac promoter points to the importance of DNA
sequence in polymerase-promoter interactions. EITC-labeled RNA
polymerase may have an altered conformation (see above) that does not
interact well with the wild type lactose P1 and P2 polymerase sites under
normal conditions. The mutant lacUVS promoter, however, allows labeled
polymerase interaction without any modification of buffer conditions.
Lowering the salt may change the conformation of labeled RNA polymerase
to a form better able to bind to the wild type sites. Alternatively, lower salt
may allow the labeled polymerase to melt wild type lac promoter DNA more
easily.

In addition to DNA binding, however, it is desirable that each protein
be active in transcription. Runoff transcription experiments show that each
labeled protein is indeed able to participate in transcription: labeled RNA
polymerase can transcribe promoter DNA, and labeled CAP can stimulate
RNA polymerase transcription from lag P1. Labeled CAP can stimulate
transcription to the same level as unlabeled CAP, indicating as much

67

transcriptional activity as the unlabeled protein possesses. That similar
levels of labeled and unlabeled CAP are needed to observe similar levels of
transcription stimulation again indicates that the labeled CAP molecules do
participate in transcription, and that the stimulation is not derived from
contamination by unlabeled molecules.

The labeled RNA polymerase, however, has somewhat altered activity.
While labeled enzyme can apparently transcribe well from the lacUV5
promoter, lower levels of P1 and P2 transcripts are seen with the wild type
lactose promoter, as compared with results using unlabeled RNA polymerase.
The lower salt concentration needed for wild type lactose promoter DNA
binding does not seem to overcome some block in the transcription process.
Moreover, allowing more time for transcription to initiate does not result in
an increase in the levels of P1 or P2 transcription from labeled polymerase-
lac promoter DNA complexes (see Chapter 3). This points to an altered
labeled polymerase conformation that slows some aspect of the transcription
initiation process at the wild type lac promoter. Thus the different DNA
sequences at lacUV5 vs. wild type lac can influence the transcriptional
activity of the labeled polymerase. Nevertheless, although the modified
polymerase has difficulty transmng from the wild type lac promoters,
EITC-labeled RNA polymerase can indeed participate in transcription.
Considering the level of transcription from the lacUV5 promoter,
transcription in labeled polymerase samples is not due to an unlabeled
contaminant. Overall then, the activities of CAP and RNA polymerase are
modified but not in a major way by labeling with EITC or LYIA.

Labeled Proteins Should be Useful in Fluorescence Studies. Each of
these proteins is apparently labeled at sites not critical for protein function.
This is both fortunate and unfortunate. It is presumably unfortunate that

68

labeling did not occur at sites of important interactions. It is fortunate
because the labeled proteins and their unlabeled counterparts behave in
similar ways, so that conclusions reached from fluorescence data should be
applicable to the unlabeled proteins. Each labeled protein should be useful
for fluorescence studies of protein conformation at promoter regions. The
average 1:1 stoichiometry of the labeled proteins should make for relatively
easy spectral determination.

The labeled a subunit of polymerase could be sensitive to DNA
promoter sequence variations, to the presence of bound CAP/cAMP, and to
intersubunit interactions within the holoenzyme. EITC-labeled and LYIA-
labeled CAP may respond to a variety of interactions: DNA-protein, protein-
protein, monomer-monomer, and cAMP-CAP contacts, for example.
Fluorescence studies with these proteins and promoter DNA are discussed in
Chapter 4.

Prior to the discussion of these experiments, however, the following
chapter describes work performed to determine precisely which CAP and 0'
amino acid residues are modified by EITC or LYIA. This information will aid
in the interpretation of fluorescence spectra, as well as providing knowledge
about sequences of CAP and RNA polymerase that are likely, or are not
likely, to be involved in contacts with DNA or with the other protein in
transcription complexes. Detailed kinetic analyses with the labeled proteins
were also performed, to characterize further the influence of the labels on the
DNA binding and transcriptional activity of the proteins.

Chapter 3
Characterization of Chemically Modified CAP and RNA Polymerase

Intmdnctian

Transcription by E. 99h RNA polymerase, and its regulation at certain
operons by the catabolite activator protein (CAP), are processes that are
critically affected by a variety of DNA-protein and protein-protein contacts.
Information about the areas of contact, and effects of conformational changes
that occur as the proteins interact at the promoter, is required for a better
understanding of the transcription process. A classical biochemical approach
to the problem involves chemically modifying a protein and observing the
effect on protein activity.

N arayanan and Krakow ( 1982) used trinitrobenzenesulfonic acid to
modify 0‘70 lysine residues. The modified subunit was then reconstituted
with core polymerase to determine if the modifications altered 0’ subunit
activity. Five lysines could be modified before reconstituted a lost the ability
to recognize promoter DNA. A similar study observed the effect a arginine,
cysteine, glutamic acid, or aspartic acid residue modification on 6 activity
(N arayana & Krakow, 1983 ). At least one of the arginines or cysteines is
vital for promoter recognition, since its reaction with N -ethylmaleimide
renders holoenzyme unable to bind to promoters. Alterations of glutamic or
aspartic acid residues also result in the inability to recognize promoter DNA.
The activity of CAP has not been probed, in general, in this manner.

As well as being useful for fluorescence studies, then, labeling of CAP
and RNA polymerase with fluorescent moieties also provides a chemical
probe of areas important for each protein’s activity. Knowledge of the

location of probe attachment, together with transcription and kinetic data,
69

70

provides insights into which amino acid sequences are important for CAP
and RNA polymerase function. This chapter summarizes transcription and
kinetic studies of the labeled proteins as well as the determination of the
location of probe attachment. The data provide a sense of the roles of the
labeled regions in transcription complex interactions.

7 1
MaterialsandMeMs
Proteins, DNA Fragments, Binding Reactions, and Retardation Gels.
CAP and RNA polymerase were prepared as described in Chapter 2, as were
the two DNA promoter fragments used (203bp lag DNA fragments, one
containing the wild type (lag?) promoter, the other the CAP-independent L8-
UV5 mutant promoter). Binding reactions for CAP and/or RNA polymerase
to these promoter fiagments were also described in Chapter 2. Gel
retardation assays, using a TBE running buffer, were performed as detailed
in the previous chapter. Where CAP samples were involved, 2011M cAMP
was added to the gel during casting and to the running bufl'er.

 

Kinetic Experiments. Association and dissociation rates of DNA-
protein complexes were measured using a gel retardation assay. Binding
reactions using radioactive end-labeled DNA were scaled up so that 25111
aliquots could be removed at various times and run into a gel. The gel allows
separation of complexes fiom unbound DNA, so electrophoresis of aliquots
taken at various times is a measure of the extent of reaction as a function of
time. Association and dissociation experiment procedures are similar,
though with some significant differences.

In association rate experiments, addition of the protein to end-labeled
(radioactive) DNA started the time course (time “zero”). After protein was
added, aliquots were removed at several time points, quenched, and then
loaded immediately onto a running gel. CAP reactions were quenched with
10 fold excess unlabeled promoter DNA, while RNA polymerase reactions
were quenched with heparin (final concentration lOOug/ml). Loading the
aliquots onto a running gel served to limit the amount of time the aliquots
remained in the gel well. These experiments were usually performed over

the course of 1h. Running the gel at lOV/cm allowed sumcient time to load

72

all the aliquots and achieve good separation of fiee DNA from DNA in
complexes.

To ascertain the dissociation rate, CAP or RNA polymerase was
allowed to bind the DNA prior to addition of an excess of a competitor (time
‘zero”). The competitor prevented any free protein from binding (or re-
binding) to the DNA. As in association rate experiments, heparin (final
concentration was 100ng/ml) was the competitor for RNA polymerase
samples, while unlabeled promoter DNA (10 fold excess) was used for CAP
samples. Once the samples were quenched, aliquots were again removed
over a period of time and loaded onto a running gel. For CAP dissociation
experiments, the gels were run at 10V/cm. RNA polymerase-promoter DNA
complexes are stable for hours, so the gels for RNA polymerase dissociation
experiments were run at 3V/cm to lengthen the amount of time aliquots could
be loaded onto the gel with good separation of fiee DNA fiom the complexed
DNA.

CAP rate experiments were done near 0°C (in an ice-water bath) and
the gels electrophoresed at 4°C (in a cold room). CAP interacts with DNA
quickly, so the low temperature conditions were used to slow the rates to
observable levels. Association rates were slowed even further by using 1/8
the normal amounts of DNA and protein. The kinetics of RNA polymerase-
DNA interactions were measured at 30°C, while the gels were run at room
temperature. For polymerase association experiments, 1/5 the usual
amounts of DNA and protein were used. Again, this served to slow the rates
to measurable levels.

Following electrophoresis, the gel was autoradiographed with Kodak
X-AR film (using an intensifying screen at -70°C), and DNA bands were
located by aligning the film and dried gel via radioactive ink dots placed onto

73

the dried gel prior to film exposure. Bands were excised and placed into 5ml
of scintillation fluid (one band per vial), and the radioactivity measured in a
liquid scintillation counter. Variations in the amount of radioactivity in the
aliquots loaded onto the gel were normalized as follows. The average total
counts per lane (free DNA counts plus complexed DNA counts (after
subtraction of the background, of course» was calculated. This value was
then used to determine a normalization factor for each aliquot: aliquot total
counts/average total counts. Using the normalized data, the percentage of
complex or fine promoter at each time point was then calculated. The
observed rate constant (kobs), corresponding to the particular reaction
mechanism indicated below, was then calculated.

Reaction Mechanisms. RNA polymerase interaction with a promoter
can be described by the following mechanism:

k1 l‘2
R+Pr .-. RPrc = RPr0

k_1 k_2
where R represents polymerase, Pr denotes promoter DNA, and RPrc and
RPr0 are closed and open complexes, respectively. The binding constant KB
(for closed complex formation) is defined as 1:1/k4, while k2 is the
isomerization constant for formation of the open complex from the closed
complex. In many cases, open complex formation is essentially irreversible,
so that k_2 is negligible. The “tau” analysis (McClure et al., 1978) allows
determination of several of these constants from association rate data.

As McClure et al. showed, the rate of appearance of RPro from the
above mechanism is

dRPro
dt

Under steady state conditions the rate of change of RPrc is taken to be zero:

 

= szPrc. (1)

74

‘11:? = k1R(Pr)- (k_1 + k2)RPrc = 0. (2)

The total amounts of promoter (Prt) and polymerase (Rt) (in the usual case

 

where polymerase is in excess) are
Pr, =Pr+RPrc+RPro, (3)
R, = R. (4)
Inserting equations (3 ) and (4) into equation (2) and solving for RPrc yields

dfifo = k0b8(Prt-Rpro)r (5)

where kobs equals

klszt
= - . 6
kObs klRt + k_1 + k2 ( )

Integrating equation (5) over time yields

PP
’kobst = ln[-P-ri-), (7)

where Pr0 is the amount of free promoter at time “zero” (which is equal to Prt)
and Pri is the amount of fine promoter at time i. The value of kobs is
obtained by plotting the natural log of (Pri/Pro) vs. time and determining the
slope of the resulting line (-kobs=slope).

In certain circumstances the kobs data can be further analyzed. Tau
(I) can be defined as

+————-1. ”‘2 (8)
t:=1}{‘obs k2 +k1 + kZRt
Assuming that k_1>>k2 (that is, closed complexes are not very stable

compared to open complexes)

1=i+ k‘l

k2 klszt (9’

75

Performing a linear regression on a tau plot (1 vs. 1/[polymerase]) yields a
slope equal to 1/KBk2 (since KB=k1/k_1) and a y-intercept of 1/k2. Association
rate experiments for labeled and unlabeled RNA polymerase were thus
performed over a range of polymerase concentrations to assess the effect of
EITC labeling on formation of open complexes.

The determination of the Open complex dissociation rate constant (k_2)
is less complicated. If re-forming of Open complexes is prevented by the
presence of a suitable competitor, then the rate of disappearance of open

complexes is

dR Pro
dt

Integrating equation (10) over time leads to

= k_2RPro. (10)

 

RPii

. 11
RPr0 ( )

—k_2t = ln

 

RPri is the amount of open complex at time i, and RPr0 is the amount at time
“zero”. Plots of the natural log of (RPri/RPro) vs. time yields a slope equal to
-k_2. The data fi'om chemically modified and unmodified polymerase
dissociation experiments were used to determine the effect of the chemical
probe on open complex stability.

The measurement of CAP-DN A kinetic parameters is done in a similar
manner. Here, the specific CAP interaction with DNA can be described by
the reaction mechanism:

C + Pr :1 CPr
k-l
where C represents CAP and Pr is promoter DNA (with a CAP site). The rate
over time of formation of CAP-DNA complexes is

76
dCPr

— = k C Pr. (12)
dt 1 ( )
The total amount of CAP (0,) is equal to
Ct =C, (13)

under conditions of excess protein. Plugging equation (13) into equation (12),
solving for Ct, and integrating over time yields

—kobsCtt= lug} (14)

where kobs=k1 and Pri and Pro are the concentration of fi'ee DNA binding
sites at times i and zero, respectively. A plot of the natural log of (Pri/Pro) vs.
the product of Ct and the time yields a slope equal to -k1. CAP association
rate experiments were carried out with EITC-labeled and unlabeled CAP to
observe the effect of the EITC label on formation of specific CAP-DNA
complexes. As kobs equals k1 in this case, tau plots for CAP complex
formation were unnecessary.

Evaluation of dissociation rates of specific CAP-DNA complexes follows
the same approach used for RNA polymerase. That is, disappearance of

CAP-DNA complexes is given by

JC—Pr- = k_ICPr, (15)

dt
again assuming that complexes are not allowed to re-form. Integration of

equation (15) over time yields the equation below

CPr
-k_ = l .
1t ln[CPro) (16)

CPri denotes the amount of CAP-DN A complex at time i, and CPr0 is the

 

amount of complex at time zero. Again, a plot of the natural log of
(CPri/CPro) vs. time results in a slope equal to -k_1. Dissociation experiments

77

using EITC-labeled and unlabeled CAP were performed to assess the effect of
the label on CAP-DNA complex stability.

Runoff Transcription Assays. Runoff transcription assays were
performed as described in Chapter 2. To follow the time course of
transcription, aliquots were removed from a transcribing sample and
quenched at various times after ribonucleotide mix addition. The samples
were scaled up to accommodate the number of 12.5ul aliquots needed for the
time course. In order to obtain the amount of transcription at each time
point, the radioactivity of each transcript band was determined as described
above, and correlated with the known number of uracils per runoff
transcript. These data were used to calculate the percentage of EITC-labeled
RNA polymerase transcription relative to unlabeled polymerase transcription
at each time point.

Chemical Modification of CAP and RNA Polymerase. CAP and RNA
polymerase were chemically modified with the fluorescent probes eosin
isothiocyanate (EITC) and lucifer yellow iodoacetamide (LYIA) as described
in the previous chapter. The chemical structure of the probes is also shown
in that chapter.

Isolation of sand Core Polymerase. EITC-labeled RNA polymerase
holoenzyme was separated into core polymerase and a subunit components
2 by the column chromatography method of Lowe et al. ( 1979). This method
shunts the effluent of a Bio-Rad BioRex 70 column to a Whatman DE-52
(DEAE-cellulose) column. Core polymerase remains bound to the BioRex 70
column, while 0 passes through and binds to the DE-52 column.

A 7cm BioRex 70 column was made by pouring the swollen material
into a 10cm Bio-Rad Econo column (0.5cm internal diameter). The lml

volume DE-52 column was made in a lml syringe with a glass wool plug.

78

Both columns were washed with five column volumes of TGED (10mM Tris,
pH 7.5, 5% glycerol (v/v), 0.1mM EDTA, 0.1mM DTT) «1» 100mM NaCl prior to
use. After connecting the columns and loading roughly 100ng of EITC-
labeled RNA polymerase onto the BioRex 70 column, the tandem columns
were washed with 40ml of TGED + 100mM N aCl. The columns were then
disconnected and each was washed with one column volume of TGED +
500mM NaCl to elute the proteins. Fractions were monitored by absorbance
to determine protein content. Protein-containing fractions were reduced in
volume in a Savant speed-vacuum and stored at -20°C until needed.
Proteolysis of Labeled Protein. Labeled CAP and a subunit were
double digested by proteases to obtain small labeled peptides suitable for
sequencing. CAP is relatively resistant to common proteases (yielding a
12.5KDa N-terminal core fragment), so a double digestion method was used.
In addition to the labeled proteins, unlabeled CAP was also digested as a
control. Trypsin digestions were performed at 37 °C for 8b in digestion buffer
(100mM NH4HCO3, pH 8.2, lmM CaClz, and 20uM cAMP) plus 0.1% SDS
with roughly 100ng of protein and 10ug of trypsin. Two additional 511g
aliquots of trypsin were added (each followed by 8h incubations). Reactions
were quenched by dilution of the samples to lml with double distilled ddeO.
Volumes of the samples were reduced in a Savant speed-vacuum lyophilizer.
Endoprotease glu-C (V8 protease) was then used to complete the
digestion procedure. Reactions were again performed at 37°C in digestion
buffer for 8b with 5ug of protease. Again, two additional 5ug aliquots of
endoprotease glu-C were added to the reactions as described above. These
digestions were also quenched by dilution of the samples to lml with ddeO
with subsequent volume reduction. Aliquots of digested protein run on 18%
SDS gels revealed digestion of each into smaller peptides (less than 8.5KDa).

79

Reverse Phase HPLC Chromatography. Reverse phase HPLC
chromatography was performed as described in Chapter 2 with the following
modification: the flow rate of the gradient runs was adjusted to 0.25ml/min
from lml/min.

Location of Probe Attachment by Amino Acid Sequencing. Peptides
from digested CAP and a subunit were first separated by HPLC reverse
phase chromatography. Labeled peptide aliquots were sequenced at the
Michigan State University Macromolecular Structure, Sequence, and
Synthesis Facility on either of two Applied Biosystems Model 477A protein
sequencers. These data were compared with the known sequences of each
protein to determine the site of probe attachment.

The LYIA-labeled aliquot, however, yielded no sequence. This fi'action
was then subjected to amino acid analysis performed by the MSU
Macromolecular Structure, Sequence and Synthesis Facility and to laser
desorption mass spectrometry (as described in Chapter 2) by the MSU Mass
Spectrometry Facility to determine its amino acid composition and mass.
Together, these data revealed the location of LYIA probe attachment within
the sequence of CAP.

As a control, additional experiments were performed to determine the
amount of protein eluted from the reverse phase column under these
separation conditions. A known amount of digested EITC-labeled CAP or
LYIA-labeled CAP was loaded onto the reverse phase column as usual.
Column flow-through and emuent fi'actions were collected over the gradient
time course. Each was then reduced in volume by freeze drying on a
Labconco Freeze Dryer 8. The absorbance spectrum of each was taken and
used to calculate the amount of peptide recovered.

80

Labeling and Sequence Analysis of Neurotensin. A peptide of known
sequence was labeled with EITC and sequenced as a control. The peptide
chosen was neurotensin, with the sequence p-E-L-Y-E-N-K-P-R—R-P-Y—I-L
(where p-E represents a pyroglutamyl residue), as it has several potential
labeling sites (N ,K, and R residues). The calculated mass of neurotensin is
1672.9Da.

Roughly 2mg of solid neurotensin was dissolved in labeling buffer and
incubated with 2.5 fold excess molar concentration of EITC for three days at
-20°C. The reaction was quenched with Tris as usual, but excess EITC was
then separated from the labeled peptide by reverse phase HPLC. Unlabeled
peptide and free EITC were run as HPLC controls. Aliquots of the labeled
and unlabeled peptide were submitted for mass determination as described
above.

The rest of the labeled peptide was digested with endoprotease glu-C
(see above) to yield a fiee N-terminus for sequencing. Again, the digested
peptide was fractionated by reverse phase H PLC along with digested
unlabeled peptide as a control. The labeled 9-mer was sequenced by the
MSU Macromolecular Structure, Sequence and Synthesis Facility.

81

Beams
Location of Probe Attachment. While the labeling ratio data described

in Chapter 2 suggest one probe attached to one protein molecule, they reveal
nothing about the location of probe attachment. Information on the exact site
of labeling was thus obtained by digesting labeled CAP and a and isolating
the peptide with label absorbance for sequence analysis. The sequence data
were then compared with the known sequences of CAP and a. Potential
residues for EITC attachment are lysines, arginines, glutamines,
asparagines, and cysteines. LYIA reacts only with cysteines. There are
numerous possible attachment sites in CAP and a for EITC, while there are
only three cysteines in a CAP monomer with which LYIA could react.

EI TC Appears to Modify Lysine 102 in the Sequence of CAP. Reverse
phase HPLC separation of digested EITC-labeled CAP resulted in several
fiactions having both protein and EITC absorbance (see Figure 3.1). This
was consistent with reverse phase HPLC results for whole EITC-labeled CAP
showing one major, and two minor labeled protein peaks (see Chapter 2).
The fiactions with the largest EITC absorbance were sequenced ( fractions 1,
13, 14, 15, 16, 25, 26, 27, and 31). One of the fractions (31) yielded no
sequence, possibly due to N-terminal blockage. This fraction was probably
undigested EITC-labeled CAP, as its peak was sharp, well separated from
the other fiactions, and eluted late in the gradient where whole CAP would
be expected to appear.

The other fiactions revealed several potential labeling sites (all of the
fi‘actions contained multiple peptides). The residues with free amines (or
cysteines) that could react with EITC are lysine 101, 102, and 189; arginine
116, 124, 143, 170, 181, and 186; asparagine 66 and 110; and glutamine 67,
105, 108, 126, and 175. While the number of residues is not surprising, it

82

Figure 3,1. Reverse Phase HPLC Separation of Digested EITC-Labeled and
LYIA-Labeled CAP. Labeled CAP digested with trypsin and end0protease
glu-C were run over a reverse phase HPLC column in order to fi'actionate the
digested peptides. The solvent gradient was fi'om 0.1% trifluoroacetic acid to
90% acetonitrile over 90min at a constant flow rate of 0.25ml/min (see
Materials and Methods). Panel A shows the digested EITC-labeled CAP
chromatograph at 230nm, 280nm, and 490nm. Fractions 1, 13, 14, 15, 25, 26,
27, and 31 were submitted for sequencing at the MSU Macromolecular
Structure, Sequence and Synthesis Facility. N 0 sequence information was
obtained from fraction 31, which may be undigested protein, presumably due
to N -terminal blockage. Panel B shows the digested LYIA-labeled CAP
chromatograph at 214nm, 230nm, 280nm, and 430nm. Since undigested
LYIA-labeled CAP ran as one reverse phase HPLC fraction (Chapter 2,
Figure 2.5), indicating homogeneous labeling of CAP by LYIA, only fraction 2
was subjected to amino acid analysis and mass spectrometry. Panel C shows
the chromatograph of digested EITC-labeled a at 214nm, 280nm and 490nm.
The only a fi'action with both EITC and protein absorbance was fraction 2.

83

5‘) 6“

 

 

 

 

 

 

 

 

 

 

 

 

 

cl.

 

 

 

 

 

 

 

 

I
-. ‘ in. w

”530‘!

 

 

 

 

 

 

C
r > rt » >
A O b
A > I
1
A v
9 .
A v
A v
A o .C
A v
u. .
A -
A v
A IL ‘
v
F v
A v
A v
i
v
A v
n A .
A v
. .-
v
$ v
v
A v
. ,d
A v
a. .
A v
A v
. .-
A v
a. .
v
v
. O
A v
”A v
A v
V
. 6
A v
(A v
A v
A v
I
A v
, v
A v
A Af ‘
. .0
A k v
I. v
A .v v
A w v
A A .I
A v v
.A . .
A v
# v
, .0
A v
FA .
A v
A v
. .-
v
o. v
A w
v
r .-
v
n. A v
A v
A v
A v k
A v
a. .
A v
A v
. 6
A v
QA v
A v
A v
A .0
A v
. v
.
1 “ H
t t t
o t
.2 .. w

 

 

 

 

84

seems unlikely that each one would be labeled to some extent by EITC. More
likely, one of the above residues is the major site of attachment. Which of the
above residues that could be is unclear, as an EITC-modified residue will
likely appear at an unexpected location during separation (by column
chromatography) of residues in the sequencing process. The presence of the
probe moiety probably alters residue properties that form the basis for the
column chromatography separation. Labeled arginines might appear in the
region expected for glutamine residues, for example.

As an aid in determining which CAP residues were labeled, a small
peptide was chosen for EITC labeling and sequencing. The peptide used was
neurotensin with the sequence p-E-L-Y-E-N-K-P-R—R—P-Y-I-L (where p-E
represents a pyroglutamyl residue). This peptide was labeled with EITC
under the same conditions as CAP and RNA polymerase. Mass spectral
analysis of labeled and unlabeled neurotensin (Figure 3.2) showed [Mi-HT“
peaks of 2376.5Da and 1674.8Da, respectively. Free EITC has a mass of
704.9Da, which agrees well with the difi'erence between 237 6.5Da and
1674.8Da. N eurotensin was labeled with only one EITC molecule per peptide
molecule.

After digestion to expose a fi'ee N-terminus, and HPLC purification,
the labeled peptide was sequenced. The resulting sequence was N-(V,W)-P-
R-R-P (see Figure 3.3). The second residue, which should have been a lysine
(K), appeared between valine and tryptophan instead. None of the other
residues showed this behavior. Apparently only the lysine was labeled by
EITC, and shifted to an earlier separation point (lysine is ordinarily the next
to last residue to elute from the separation column). Of the above potential
CAP residues, only lysine 102 appeared as a shifted residue (near valine).

85

Figjge 3.2. Mass Spectra of EITC-Labeled and Unlabeled N eurotensin.
Mass analysis by laser desorption mass spectrometry was performed as
detailed in the Materials and Methods section. The intensity scale has
arbitrary units. Panel A shows the mass spectrum of EITC-labeled
neurotensin, while Panel B shows the spectrum of unlabeled neurotensin.
The peak at 5734.8Da in the spectrum of EITC-labeled neurotensin is the
internal insulin standard ([M-i-H]+ form).

86

 

 

2200

(43:2 +AIAEV 1m
Denna-

 

 

 

m nadir»
..C. :6 gen

m 3:: -

 

 

 

0.050“ t

 

(o'JAC. 0 VM
++ .

«do: i

3.15
+H:.EU
tfiu e 0w.+kus+aobr 0.0:...“ i

_.J\_

 

 

coma t

A

rtssaiasfiji e . 2
on: pita «WEAK; «..PfifiE—H

 

 

 

 

 

 

 

1000«
o

-4000<

450004

3.2.3:. 5360:...

87

W. Sequence of EITC-Labeled Neurotensin. The sequence of labeled
neurotensin was determined as described in Materials and Methods. Panel A
shows the sequencer data for the first three residues of the labeled peptide.
Panel B shows similar data obtained by sequencing an aliquot of the same
peptide used in Panel A on a different sequencing instrument.

88

..u «in 0..“ n.n~ ..nu v.- 0.: v..—

..p- v..—

o.: v.n— 0.: ..o— 0.. a.» 0.0

 

 

 

 

 

 

 

 

1
lit 11 . I ..
A a a case > u A a u s a «in

co.oo~
.

a..oo.

c...o.

oo.eo.

. ucas -.-sua son-.-
..a. ...— o... ..A. a... ..o. ..o a.» 0..

 

 

 

 

bl

mu

 

. «It. . u m tidfinwuzA ..

.0...-

 

 

 

 

o 3.— guisg—n

 

 

 

 

 

 

..~ A... ...u a... o... ..uu o... ...A a... ...A o... ...a a... ..o. ..s .mwtho.¢
..u A a span a.. A t, it 4 u o a r...
. u..oo~
co.oo.
v...o.
to.ooo
«flange .3... . .m

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

t.-..Mttttodltttc1tittudtti 3
« .Qtt o o A e o u 4 o o utifi
d .
O
-twig-lt.d-»-t.~.rrtt.~.tt 3
a
1 <.. e «A, t..¢..“ 4, e .< .t A; + w s..« o “
o o
T: 1,
Av
0:32: .2: Ana—Ace‘s
rrtt.,m---..m-ttl.#-rtt.d -L .w
4+ o< e 4: 91 «9990.

A

 

 

 

89

None of the others were shifted. This suggests that EITC modified primarily
lysine 102 in the CAP sequence. Since the two arginines, the glutamine, and
the asparagine neurotensin residues were not labeled by EITC under the
same conditions used to label CAP, it is perhaps not surprising that little
EITC modification occurred at these CAP residues. It cannot be ruled out,
however, that other residues are also labeled. Indeed, as the HPLC
purification of undigested EITC-labeled CAP indicates the presence of three
EITC-labeled CAP species (one major species, and two minor species; see
Chapter 2), other CAP residues may be labeled to some extent.

CAP Residue Cysteine 93 Is Probably Modified by LYIA. No sequence
information was obtained from digested LYIA-peptide fi'action 2 (see Figure
3.1). Amino acid analysis, and mass spectrometry data, however, yielded
enough information to indicate the site of LYIA attachment to CAP. Laser
desorption mass spectrometry of the LYIA-labeled peptide fiaction yielded
three masses of 2337.3Da, 2972.8Da, and 3399.7Da. Apparently there was
more than one peptide in the fi'action after reverse phase HPLC
chromatography. The amino acid analysis of this fraction is thus complicated
due to the multiple peptides. However, comparison of the amino acid results
with the positions of the three cysteines and the likely protease digestion
sites ruled out two of them.

One of the three CAP cysteine residues (cysteine 17 9), for example, is
surrounded by the sequence Q-E-I-G-Q-I-V—GC-S—R—E-T. Digestion with
endoprotease glu-C and trypsin should yield the fragment I-G-Q-I-V-G-C-S-
R. However, the amino acid analysis did not indicate the presence of
glutamine residues (see Table 3.1). This suggests that cysteine 179 is not the
site of LYIA modification. In addition, cysteine (179) is in the turn region of
the C-terminal helix-turn-helix DNA binding domain. Labeling of this

90

Table 3.1. Amino Acid Analysis of LYIA-Labeled CAP Peptides. Amino acid
analysis of the digested LYIA-labeled CAP HPLC fraction 2 (see Figure 3.1,
Panel B) was performed by the Macromolecular Structure, Sequence and
Synthesis Facility as described in Materials and Methods. Since there were
several peptides in the sample, normalization to one of the amino acids to
determine both the mass of the peptide and the number of each residue
within the peptide sequence was not possible. Cysteine was not detected,
presumably due the presence of the LYIA moiety.

91

 

 

 

 

 

 

 

 

 

 

 

Residue Amount (pmole) Residue Amount (pmole)
Alanine 6,349 Leucine 7,475
Arginine 2,510 Lysine 5,937
Asparagine 0 Methionine 1,603
Aspartic Acid 10,957 Phenylalanine 3,262
Cysteine ND Proline 4,343
Glutamine 0 Serine 6,220
Glutamic Acid 1 1,237 Threonine 3,885

Glycine 6,905 Tryptophan 0

Histidine 1,841 Tyrosine 2,429
Isoleucine 3,637 Valine 4,608

 

 

 

 

ND = N at Detected

 

92

cysteine would probably inactivate specific binding. Indeed, CAP with more
than about one LYIA per CAP (made by incubating CAP with 10 fold excess
LYIA for three days at room temperature) inactivates specific DNA binding
(data not shown). Further, cysteine (19) is near a tryptophan residue. Since
the amino acid analysis reveals no tryptophan, cysteine 19 is also not likely
to be the site of LYIA labeling. This leaves cysteine 93 as the most likely
choice for the site of LYIA labeling. The CAP sequence A-K—T-A-C-E-V-A—E-
I-S-Y-K-K-F—R has a mass of 1826Da. Addition of the disodium form of LYIA
(503Da) brings this the mass of the peptide to 2329Da, which agrees well

with 2337.3Da bound by mass spectrum analysis.
The oResidue Lysine 251 Is Likely Modified by EITC. Separation of

digested EITC-labeled a by reverse phase HPLC chromatography resulted in
only one fiaction with EITC absorbance. This fi'action appeared to be a
dimer of sequence L-R. However, one of these residues, with an attached
EITC, probably is at an altered position in the sequencer’s column
chromatography separation. Another sequence run of EITC-labeled
neurotensin was performed as an aid to determine which residue of the above
dimer might be labeled with EITC. (The EITC-labeled CAP fractions and the
EITC-labeled a fraction happened to be run on two different sequencer
instruments.) The labeled neurotensin yielded the sequence N -(Y,P)-P-R-R-
P-Y-I. The modified lysine eluted from the sequencer column in a position
between tyrosine and proline. Based on the labeled neurotensin data, we
conclude that the “R” residue in the a dipeptide is probably the EITC-
modified lysine, so that the dipeptide sequence is L—K

The L-K tandem occurs twice within the a subunit. The first is near
the N-terminus of the protein (residues 9 and 10), and the second is at 250-
251. The first occurrence is an unlikely candidate as the nearest glutamate

93

residue is at position 2. Digestion with trypsin and endoprotease glu-C
should yield an eight residue peptide (fiom residue 3 to 10). Thus, a much
larger peptide would be expected if lysine 10 were EITC-labeled. The
sequence around position 250 is Q-E-E-I-L—K—L—S—E-V-F. Trypsin should cut
at the C-terminal side of lysine, while endoprotease glu-C should cut at the
C-terminal side of glutamate. This would yield a trimer of sequence I-L—K
However, endoprotease glu-C can sometimes cut at the C-terminal side of
other residues, one of which is isoleucine (Joe Leykam, personal
communication). Thus, EITC apparently attaches to lysine 251 within the a
sequence.

A potential pitfall here is that the fiactions collected from the reverse
phase separation prior to sequencing might represent only a minor labeling
site. Experiments were thus performed to determine how much digested
protein was recovered fiom typical HPLC runs. Such experiments suggest
that between 40% to 50% of the total protein loaded onto the column was
recovered with no protein being lost to the flow-through. The recovery of this
amount of peptide fiom reverse phase HPLC columns is typical (Joe Leykam,
personal communication). Thus the major sites of probe attachment are
probably represented in the collected labeled peptide fractions.

Effect of the Presence of the Probe on CAP and Polymerase Binding
Kinetics. A detailed kinetic analysis of the labeled protein binding activity
revealed some interesting results. Figure 3.4 shows representative gels of
such rate experiments with EITC-labeled CAP and EITC-labeled RNA
polymerase. After the amount of radioactivity in each free DNA and complex
DNA band was determined, these data were used to calculate the observed
rate constant for a particular reaction (see Materials and Methods under
Reaction Mechanisms). Typical plots of such data are shown in Figure 3.5,

94

Figgge 3.4. Representative Kinetic Experiments with EITC-Labeled CAP
and RNA Polymerase. Aliquots fiom a binding reaction were loaded at
various times onto running gels (5% polyacrylamide for CAP, 4% for
polymerase) as described in Materials and Methods. Panel A shows EITC-
labeled RNA polymerase association with the laeL8-UV5 at 30°C, while

Panel B shows EITC-labeled CAP dissociation from lag;+ DNA at 0°C (gel run
at 4°C).

95

Time(min) 0 0.5 1 2 3.5 5 10 15 30 60
-
-

EITC-Polymerase—y: ..A - - -

Complexes

Free DNA-> ------

Time (min) 0 0.5 1 2.5 5 10 15 22.5 30 60

EITC-CAP» —
Complexes -

FreeDNA—> ---—-’

96

Fiage 3.3. Representative Graphs of Rate Experiment Data. EITC-labeled,
as well as unlabeled, CAP and RNA polymerase association and dissociation
rate experiments were performed as detailed in Materials and Methods. The
percentage of bound DNA or free DNA was determined by measuring the
radioactivity of the corresponding DNA bands excised from a nondenaturing
polyacrylamide gel. These data were then plotted vs. time as described in the
Materials and Methods section. Panel A shows the dissociation of EITC-
labeled CAP fi'om wild type lag DNA. Panel B shows EITC-labeled RNA
polymerase association with preformed wild type lac promoter DNA-CAP
complexes. Panel C shows the EITC-labeled and unlabeled RNA polymerase
tau plot was generated from several association rate experiments (at
different concentrations of polymerase) such as one depicted in Panel B. Rt
represents the total polymerase concentration in any particular experiment.
The various constants describing CAP and polymerase interaction with
promoter DNA were determined from the slopes of the data.

97

 

 

 

 

 

 

A.
CPri/CPro
-25
0 5 10 15 20 25
Time (min)
B.
labeled.
)ciation 0 F
d5. The -0.2 »
the
’ . , . I
itunflg 0'4
edinthe Pri/Pro -0.6 .

{lg '03 F .\'\
1 .1 I

’ I
new6 - . . ,\

O 20 40 60 80 100

B. R1 Time (min)
intent

b

 

 

 

 

140
130 i /

120 ’ Polymerase //
110 '

Tau /
. 100 .
(mm) I A!
90 t
80 L EITC-Polymerase

0.004 0.003 0.012 0.016 0.02 0.024 0.028
1/[Rt] (1/nM)

 
    

 

 

 

98

along with a representative tau plot generated from labeled and unlabeled
RNA polymerase “observed” association rate constants obtained at several
polymerase concentrations.

Kinetic parameters for promoter binding and open complex formation
by labeled and unlabeled RNA polymerase are shown in Table 3.2. With
lacLS-UV5 promoter DNA, the presence of EITC affects both the
isomerization constant (k2) and the binding constant (KB) at 30°C, with a
larger effect on k2 (by about an order of magnitude) than on KB (roughly a
four fold difi‘erence compared to unmodified enzyme). It is noteworthy that
k2 is reduced, while KB is increased somewhat. The EITC labeling of 0
appears to confer a modest increase in the affinity of the enzyme for closed
promoter sites, compared to unlabeled protein. However, the labeled
polymerase does not form an Open complex as efficiently as the unlabeled
enzyme. The dissociation rate of RNA polymerase fi-om the lanV5 promoter
is apparently unaffected by covalent attachment of the EITC probe. Once
open complexes form with labeled RNA polymerase, they are just as stable as
unlabeled RNA polymerase open complexes.

At the wild type lag promoter region, association of polymerase at P2 is
not much altered by the attachment of EITC to a (in the absence of CAP
polymerase binds primarily to P2) at 30mM salt. The dissociation rate of
labeled polymerase from P2 complexes is also apparently not affected. In
general, open complex formation at P2 is slower than at laeL8-UV5.

In the presence of CAP, the EITC-labeling of a does not appear to
grossly alter the kinetic parameters of RNA polymerase holoenzyme. EITC-
labeled RNA polymerase has a slightly increased KB value (by about 4 fold as
compared to the unlabeled enzyme in the presence of CAP), while the

isomerization of closed to open complexes (k2) is relatively unchanged by the

99

Table 3.2. Kinetic Parameters for EITC-Labeled CAP and RNA Polymerase
Interactions with Q9 Promoters. The rate constants were calculated finm
data as described in Materials and Methods. The kinetics of CAP interaction
with wild type lae promoter DNA are based on binding reactions performed
at 0°C in binding buffer containing 100mM KCl. The constants describing
the interaction of RNA polymerase with the lagL8—UV5 promoter are drawn
from binding reactions incubated at 30°C in binding buffer plus 100mM KCl.
The kinetics of RNA polymerase interaction with wild type lac promoter DNA
are based on reactions performed in binding buffer plus 30mM KCl at 30°C.
EC represents EITC-labeled CAP; ER signifies EITC-labeled RNA
polymerase; CAP denotes the unlabeled protein; and R denotes unlabeled
RNA polymerase.

ierase
from
major
)rmed
.bing
lrawn
M KCl.
3r DNA
; 30°C.

eled

100

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Complexes kL(M‘ls'l) k_1 (8'1)
CAP + 12:: 4.6105 x 103 6.1113 x 104
EC Lig‘ 2.0104 x 103 1.1111 x 10'3
Complexes KR(M'1) k9 (8'1) k _9 (8'1)
R flagL8UV5 1.510.01 x 107 2.610.01 x 10'2 1.9101 x 10“
ER + lagL8~UV5 5.8113 x 107 3.8102 x 10 '3 2.3102 x 10“1
R + 13.2““ (P2) 6.7121 x 106 2.7108 x 10'4 2111.0 x 10*1
ER 4,139+ (P2) 3.6133 x 106 2.4101 x 10-4 3.2104 x 104
R + CAP + lsflPl) 1.6109 x 107 9.2126 x 104 1.0101 x 104
ER + CAP + Lag; (P1) 6011.1 x 107 9.8185 x 10‘1 1.9102 x 10‘1
R + EC + MP1) 7.5148 x 106 1611.0 x 10'3 1.2101 x 10“1
ER + EC + Lag: (P1) 2.5107 x 107 3310.5 x 10'4 1.6101 x 10‘1

 

 

 

 

 

101

presence of the EITC-label. CAP-labeled polymerase-lag promoter complexes
dissociate somewhat faster than the corresponding unlabeled polymerase-
CAP complexes. EITC-labeled CAP and RNA polymerase interaction at lag"
(P1) shows mixed effects. KB is somewhat larger for EITC-labeled RNA
polymerase in the presence of EITC-labeled CAP than for the unlabeled
holoenzyme. However, interaction with EITC-labeled CAP results in a
smaller EITC-labeled RNA polymerase k2 value (about 5 fold) than for the
corresponding interaction with unlabeled polymerase. The dissociation of
such double labeled complexes is about the same as EITC-labeled CAP-
unlabeled polymerase-lag DNA complexes.

EITC-labeling of CAP affects both association and dissociation rates
from its specific lab DNA site, but only by about two fold at 0°C. Data at
higher temperatures were difficult to obtain, since reaction rates become too
fast to measure by the gel retardation assay. EITC-labeled CAP binds about
half as fast as unlabeled CAP, while dissociating at about twice the rate.

EITC-labeled CAP also alters P1 open complex formation by RNA
polymerase at the lag promoter. The labeled CAP does not afl'ect KB, but
increases k2 slightly as compared with unlabeled CAP. EITC-labeled CAP
appears to be better at enhancing the rate of isomerization of closed to Open
complexes than unlabeled CAP. Once P1 complexes are formed with EITC-
labeled CAP, the complexes are as stable as those containing the unlabeled
protein.

What is interesting from the above discussion is that neither labeled
protein prohibits such interactions fi'om occurring, but only alters them to
some degree. That the kinetic parameters are altered somewhat indicates
that the labeled proteins are indeed participating in DNA binding. Binding

102

due to an unlabeled contaminant would result in the same parameters as
calculated for the unlabeled controls.

Transcription Using Labeled Proteins. The transcription data
presented in Chapter 2 indicate that labeled CAP and unlabeled CAP have
equivalent activities. Labeled RNA polymerase catalyzed transcription from
lan'VS as well as does the unlabeled enzyme. EITC-labeled polymerase
exhibits less transcriptional activity from wild type lactose promoter DNA
than the unlabeled protein, however. Time course transcription experiments
were performed to determine if labeled RNA polymerase-lag;+ DNA complexes
are delayed in transcription initiation, but could reach unlabeled polymerase
levels if given more time. Unlabeled RNA polymerase reaches its maximum
level of transcription between five and ten minutes after addition of the
ribonucleotide mix (see Figure 3.6). EITC-labeled RNA polymerase also
reaches a plateau between five and ten minutes, but does not reach the levels
of the unlabeled polymerase, even after forty minutes of transcription.
Careful analysis of transcription gels showed no apparent increase in
abortive products when comparing the labeled and unlabeled polymerases
(data not shown). So while EITC-labeled RNA polymerase can bind lag"
promoter DNA well and appears to clear promoters at a similar rate, its
transcriptional activity is apparently somehow reduced.

103

M33. EITC-Labeled Polymerase Transcription Time Course. Aliquots
from a transcribing sample containing lac“ promoter DNA were withdrawn
at the indicated times and run onto a gel as detailed in the Materials and
Methods section. Panel A shows the autoradiograph of the gel.

Transcription from P2 (85nt) and P1 (63nt) is indicated by the arrows. Panel
B shows the amount of transcription at each time point from P2. The ratio of
labeled polymerase transcription to unlabeled polymerase transcription at
each time point remains constant at about 20%, indicating similar rates of
transcription initiation by the labeled and unlabeled holoenzyme.

104

Time (min) 40 2010 5 1 40 2010
Polymerase - - - - - + + +
EITC-Polymerase + + + + + - - -

-.- ... --u—.:

51
++

 

’9’. <—P2 (85nt)
I q o
O C O O . <-P1 (63nt)

 

30

 

25

20
Amount of
Transcripts 15

(fmol)

 

 

0 10 20 30 40

Time (min)

105
Discretion

EITC-Labeling of Lysine 251 Influences C'Subunit Function. In
addition to being potentially valuable for fluorescence studies, as described in
the preceding chapter, EITC and LYIA are also useful chemical probes of
protein areas that may influence protein function. Kinetic and localization
studies with labeled CAP and RNA polymerase yield some interesting
information about the locations of the attachment sites for EITC and LYIA.

The sigma subunit of RNA polymerase is apparently labeled at lysine
251. While no crystal structure of o is available, much is known about the
various regions of the subunit. It is of value to consider that C is comprised
of four regions, based on sequence alignment with other bacterial 0’ factors
(Gitt et al., 1985; Helmann & Chamberlin, 1988). Lysine 251 is in the 245
residue spacer between the N-terminal region one (residues 1 to 139) and
region two (384 to 464). The role of the 245 spacer region is not known, but it
has been proposed to be possibly involved in core binding (Burgess et al.,

1987; Lesley & Burgess, 1989), 0' subunit stability (Hu & Gross, 1983;
Grossman et al., 1983), or reduction of holoenzyme nonspecific DNA affinity
('Ifiian et al., 197 7; Bernhard & Meares, 1986).

The region near lysine 25 l is probably quite solvent exposed, as this
residue is more reactive with EITC than the many other potential surface
residues available in the holoenzyme. This region is thus probably not
essential for core polymerase binding. If it were, probe attachment would
either perturb such an interaction and holoenzyme would not be formed, or
the site would not be labeled at all. The region near lysine 251 probably
does not appreciably afi‘ect the structural integrity of the subunit as the
labeled holoenzyme is active in specific DNA binding, and labeled
POIymerase-DNA complexes are just as stable as unlabeled protein

106

complexes. It is not apparent from the above data if EITC-labeling of lysine
251 alters the nonspecific affinity of polymerase. Because the labeled
holoenzyme is active in specific DNA binding, the region near lysine 251 is
most likely not involved with intimate contact with promoter DNA.

It is quite possible, however, that EITC-labeling of o alters its
conformation. Modification of C lysines with trinitrobenzenesulfonic acid
resulted in altered protease digestion patterns, suggesting changes in
conformation (Narayanan &1Krakow, 1982). Modification of o with EITC
does alter the binding of the labeled polymerase to promoter DNA. At a
strong promoter like lanV5, EITC-labeled RNA polymerase binds well, but
is slower than unlabeled protein in isomerization of closed complexes to open
complexes. Perhaps the modification of lysine 25 l transmits a conformation
change to the regions involved in DNA melting (2.1 and/or 2.3) (Simpson,
1979), causing open complex formation to be slowed. Labeling of lysine 251
results in a slight increase in the affinity of the enzyme for promoter DNA. It
may be that a conformation change transmitted to region 2.1 and/or 2.3 also
affects region 2.4, which is tliought to recognize the -10 region of promoters
(Siegele et al. 1989; Waldburger et al., 1990) (the UV5 mutations are in the
- 10 region of lag), or to region 4, which interacts with the -35 region of
promoters (Siegele et al., 1989; Gardella et al., 1989). The change that
negatively affects 2.1 and/or 2.3 seems to potentially affect region 2.4 or 4 in
a positive manner.

But at the wild type lag promoters (which differs from the lanV5
sequence), labeled polymerase does not form open complexes unless the salt
concentration is lowered from 100mM to 30mM. Perhaps the isomerization
from closed to open complexes is blocked at the higher salt. In the absence of

CAP, the presence of the label has little overall effect on promoter binding.

107

The reduced salt concentration may change the labeled C conformation in
such a way that isomerization can occur at the weaker promoters as fast as
for the unlabeled protein. Alternatively, the DNA may melt more easily at
lower salt, allowing the labeled enzyme to form open complexes. In the
presence of CAP, the labeled polymerase exhibits similar promoter binding
kinetics compared with CAP plus unlabeled enzyme. Certainly, the region
around lysine 25 1 is not likely involved in protein-protein interactions with
CAP important for polymerase promoter binding, otherwise we would expect
to observe little, if any, Open complexes formation in the presence of CAP.

Labeled polymerase appears to have a lowered overall transcription
activity at the wild type lag P1 and P2 promoters. At the lae“ P2 promoter
the labeled polymerase produces full length runoff transcripts, but does not
reach unlabeled polymerase levels, even after extended reaction time. Since
the unlabeled polymerase transcribes well at lower salt (30mM), this loss of
activity is not due to a salt effect on the conformation of the polymerase.
Only a fraction of the labeled polymerase-lag?“ complexes can evidently
overcome some barrier to transcription initiation. This suggests that there
are at least two subpopulations of the complexes that may differ in
conformation. The difference in DNA promoter sequence between the lag
wild type and lanV5 promoters may play an important role in polymerase
conformation and transcription initiation.

It is possible that EITC modification not only affects a subunit
conformation, but also the conformations of B and 3’ via intersubunit
contacts. These subunits are involved in many steps of transcription ([3:
Grachev et al., 1989; Rockwell & Gottesman, 1991; B’: Borukhov et al., 1991;
Bowser & Hanna, 1991; Petersen & Hansen, 1991) such as ribonucleotide
binding to B. Perhaps the [3 subunit is changed so that binding of the

108

initiating ribonucleotide is significantly altered, but once bound,
transcription can proceed normally. Similar results have been observed with
a ten residue deletion (residues 435-444) in a conserved N—terminal region of
B (Martin et al., 1992). This mutant polymerase binds to promoters with
reduced afinity, but once it has initiated transcription, the mutant
polymerase does elongate normally.

It is interesting that while CAP does not radically alter labeled
polymerase binding to lag:+ P1, these complexes do not transcribe well
(Chapter 2, Figure 2.7). CAP is thought to interact with the a subunit(s) of
polymerase (Igarashi & Ishihama, 1991; Igarashi et al., 1991; Zou, et al.,
1992; Kolb, et al., 1993), which may be important for transcription
stimulation by CAP. The modified 0’ subunit might change the conformation
of or so that critical CAP interactions are altered. Thus initiation would be
inhibited in CAP-labeled polymerase-DNA complexes. Changes in B and 3’
conformations as indicated above might also contribute to this result. Oddly,
EITC-labeled RNA polymerase transcribes somewhat better in combination
with EITC-labeled CAP than it does with the unlabeled CAP. Perhaps the
conformation of modified CAP is altered in a manner that is complementary
to any effects induced by the labeling of 0'. The modified CAP might be able
to interact somewhat better with the modified holoenzyme.

In summary, then, while it seems clear that the lysine 251 region is
not involved with direct DNA or CAP interactions important for polymerase
binding and transcription, some interactions important for activity are
altered by the presence of the label. This may be due to conformational
Changes in o and the other subunits brought about by the EITC modification
rather than through direct interactions of the labeled residue with DNA
and/or CAP.

109

Labeling of Lysine 102 or Cysteine 93 Does Not Significantly Influence
CAP Function. The major site of EITC attachment is likely lysine 102, but
other labeling sites may be present. Labeling of CAP with 2.5 to 3 EITC
probes (by reacting CAP with 10 fold excess EITC) does not appear to
significantly reduce specific DNA binding (data not shown). Lysine 26 is
probably not labeled under these conditions as this residue interacts with
DNA in the DNA-CAP crystal structure (Schultz et al., 1991). Reduced
specific DNA binding would be expected if lysine 26 were labeled. Lysine 102
is near the N—terminal and of the B helix (see Figure 3.7). This residue is
essentially on the Opposite side of the C-terminal DNA binding domain of the
protein.

CAP is apparently labeled at cysteine 93 by LYIA. Labeling of CAP
with more than one LYIA per molecule (again, by incubation with 10 fold
excess probe) inactivates specific DNA binding. This suggests that after
cysteine 93 labeling, cysteine 179 may be labeled. This residue is in the turn
of the helix-turn-helix DNA binding motif of the C-terminal domain of CAP.
Alternatively, cysteine 93 of the other monomer in a labeled dimer may be
labeled and cause a conformation change that inactivates CAP binding.
Attachment of a bulky probe molecule could alter the helix-tum-helix
conformation or simply sterically interfere with DNA binding. Cysteine 93 is
near the N -terminal end of strand 8 of the B-roll in the N-terminal domain of
the protein (see Figure 3.7).

Neither of these labeled CAP residues is probably near the site of
interaction with the a subunit of RNA polymerase. These CAP-or interactions
are thought to occur on either of two faces of the CAP dimer (see Figure 3.7):
near histidine 159 or glutamatic acid 171, which when deleted or changed
inhibit transcription stimulation when CAP binds at roughly ~60 or -40 (Bell

110

Figare 3.7. Location of EITC and LYIA Probes in the Sequence of CAP. A
CAP monomer is shown along with the likely locations of EITC or LYIA
attachment. The structure of the monomer is from Weber and Steitz (1987).
The two domain structure of the monomer is clearly seen: a C-terminal
domain that binds DNA via a helix-turn-helix structure (helices E and F),
and a N-terminal domain that binds cAMP. Both EITC and LYIA react with
residues in the N-terminal domain (lysine 102 and cysteine 93, respectively).
The locations of histidine 159 and glutamatic acid 171, part of a region that
is thought to interact with RNA polymerase when both are bound to
promoter DNA, are also shown. In addition, lysine 52 is shown. This residue
is part of another region that may interact with RNA polymerase. See text
for full details.

 

FL

 

 

 

\__

 

~ ‘.

'O

Glutamic acid 171

. “,1

/‘\

111

Histidine 159

.32;

t!—
43y

'\ ‘7

(l
‘1
\

/X.’
Cysteine 93 (LYIA) (

u!

C

 

 

F

 

8
V; _.__...Lysine 102 (EITC)

__)~

 

+ ‘ ~--
H,N \

112

et al., 1990; Williams et al., 1991); and near lysine 52, which in combination
with a histidine 159 mutation, restores transcription stimulation, but only at
CAP sites positioned around -40 (Bell et al., 1990; Williams et al., 1991).

The CAP and RNA polymerase residues modified by EITC or LYIA are
not critical for protein binding or transcription activity. Their modification
does, however, alter protein activity somewhat, and suggests that residues
far from active sites do influence CAP and polymerase activity. The
influence is probably transmitted to the active sites by protein
conformational changes induced by probe attachment. Protein conformation
is certainly not rigid, and so CAP and RNA polymerase conformation changes
induced by DNA binding and/or protein-protein interactions are quite
possible. The fluorescence studies described in the next chapter do, in fact,
suggest some DNA sequence-specific efi'ects on the structure of bound

proteins.

Chapter 4
Fluorescence of EITC- and LYIA-Labeled CAP and RNA Polymerase

131mm

Transcription and its regulation are likely to involve a number of
conformational changes in CAP, RNA polymerase, and DNA. In recent years
studies of DNA structures, and transformations between these structures,
have been popular. Less well studied are the effects on protein structure
caused by interactions with specific DNA sequences. Information about the
conformation changes of CAP and RNA polymerase when bound at promoter
regions will lead to a better understanding of the molecular mechanisms of
transcription, and of its control.

Fluorescence, either intrinsic or extrinsic, has often been used as a
means to monitor protein conformation. Wu et al. (1975, 1976) fluorescently
labeled the 0' subunit of RNA polymerase holoenzyme with dansyl chloride or
N -( 1-pyrene)maleimide, for instance. Using the dansyl chloride labeled
subunit, these workers followed the release of C during transcription
initiation, and its subsequent binding to free core polymerase. Using the N-
(l-pyrene )maleimide modified a, they observed a fluorescent spectral change
when the subunit interacted with core polymerase, indicating a potential 0
subunit conformational change during the formation of holoenzyme.
Further, Wu et al. (1974) used fluorescently labeled CAP to monitor its
binding to DNA containing a lag CAP site. The spectral data indicated that
CAP changed conformation upon binding to the DNA. Other researchers
have monitored the intrinsic fluorescence of CAP (Takahashi et al., 1983),

and observed fluorescence changes upon DNA binding.

113

114

The conformations of CAP and RNA polymerase no doubt are critical
factors in the molecular mechanism of transcription initiation. As discussed
earlier, CAP and RNA polymerase bind at different relative positions at
difi'erent promoters; in addition, difi'erent numbers of CAP may be bound.
There thus exists the potential for a variety of protein-protein and protein-
DNA contacts at catabolite sensitive operons. Does initiation of transcription
at these promoters involve different CAP-polymerase contacts, or are there
compensating structural changes, dependent on DNA sequence, that bring
the same regions of the proteins into proximity in each case? While it is
fairly clear that CAP, for instance, changes conformation upon DNA binding,
whether it has similar conformations at various promoters is unknown.
Likewise, whether RNA polymerase adopts similar conformations at each
promoter, or if its interaction with CAP at a promoter has any effect on its
conformation, is unclear. In order to study this aspect of CAP and RNA
polymerase protein structure, fluorescence experiments with CAP and RNA
polymerase labeled with fluorescent probes are described. The results are
complicated, but do hint at conformational differences depending on the
sequence of the DNA involved.

1 15
Mater-131mm

Proteins. CAP and RNA polymerase were purified and stored as
described in Chapter 2. On average, CAP and RNA polymerase were judged
to be 10 to 25% active in DNA binding and transcription assays (Garner &
Revzin, 1981).

DNA Fragments. In addition to the 203bp lag wild type and lag; -
UV5 promoter DNA fragments described in Chapter 2, three other DNA
fragments were used in this study. The first was a 230bp wild type gal
promoter fi‘agment, the second was a 2 10bp lag-gal hybrid promoter
fragment, and the third was calf thymus DNA (as a nonspecific control). The
Appendix shows the promoter region sequence of the gal and lag-gal
fragments. Figure 4.1 shows the structure of the four promoter fi‘agments.
All DNA stocks were stored in fluorescence binding buffer (see below) to
avoid buffer dilution of fluorescence samples.

The gal+ fi‘agment was isolated fiom the recombinant plasmid
pMHSZ30. pMHS230 was generated by ligating tandem head to tail dimers
of the 230bp gal sequence into the MI site of pUC 18. This fragment
contains the P1 and P2 RNA polymerase binding sites (transcription
initiation at the nucleotides designated +1 and -5, respectively) as well as
both gal promoter CAP sites (designated CAPl (at -40) and CAP2 (at -60)).

The lag-gal hybrid promoter fragment was constructed (performed by
Diane Cryderman) by attaching a wild type lae CAP site to a gal promoter
carrying P2' and CAPl‘ mutations. Lag DNA extends to about -70 to -49,
with gal DNA extending from -43 to +45. The region between -43 and -49 is
an E9931 linker (cut and end-filled). This hybrid promoter has the CAP site
positioned at -60, precisely as it is at the lag promoter. The gal P2‘ and
CAP 1' mutations allowed for the observation of the effect, if any, of the lag

116

Figjge 4,l. Promoter Region Structure of Wild Type Lag, Wild Type fig, and
Lag-Gal DNA Fragments. The 203bp wild type lac DNA fragment has a CAP
site located around -60, and two promoters, P1 and P2, that initiate from +1
-22, respectively. The position of RNA polymerase when bound at P2, in the
absence of CAP, is shown by the horizontally hatched box. In the presence of
CAP, RNA polymerase binding at P1 is shown by the stippled box. CAP
precludes RNA polymerase from the P2 site. The 230bp wild type gal DNA
fragment has two CAP sites. Polymerase binding at P1 (stippled box)
overlaps the first CAP site (CAPl). Again, CAP binding at CAPl precludes
RNA polymerase from the P2 promoter, but not from P1. The 210 bp lae-gfl
hybrid DNA fragment has a gfl P1 promoter region (P2 has been mutated so
that polymerase does not bind to the P2 promoter) attached to a lag CAP site.
The position of the CAP site is around -60, as it is at the wild type lag
promoter region. The CAPl site (at -40) of gfl portion of the hybrid has also
been mutated such that CAP has a reduced affinity for the -40 site.

117

has:+

RNA Polymerase

3mm

RNA Polymerase

.22P2

Gar

+134

 

RNA Polymerase

Lac-Gal

RNA Polymerase

 

pBR322 DNA 139931 Linker-J
DNA

118

CAP site on gal Pl transcription. In a sense, the hybrid construct can be
considered to be an attempt to turn a gfl promoter into a lag promoter.

Binding Reactions and Gel Retardation Assays. Binding of CAP
and/or RNA polymerase to promoter DNA was performed at the indicated
temperature, and the reactions analyzed on gels as appropriate, essentially
as described in Chapter 2. The only modifications were (a) the use of
fluorescence binding buffer (40mM Tris, pH 8, 10mM MgClz, 0.1mM EDTA,
2011M cAMP (if CAP was present)), plus either 30mM N aCl (polymerase) or
100mM NaCl (CAP), and (b) an increase in the cAMP concentration from
2011M to 2000M in samples, gels, running buffer, and fluorescence samples if
wild type gal promoter DNA was used.

Runoff Transcription. Runoff transcription assays were performed as
detailed in Chapter 2 with the following modifications. Samples similar to
those used in fluorescence experiments (see below) used a larger amount of
promoter DNA (111g). The ribonucleotide mix for these samples was
proportionally altered to 2mM ATP, GTP, and CTP, 250pM UTP, 10uCi a-
32P-UTP (3000Ci/mmole) per reaction, and 1mg/ml heparin.

Fluorescent Labeling of Proteins. Labeling of CAP and RNA
polymerase with eosin isothiocyanate (EITC) and lucifer yellow
iodoacetamide (LYIA), and characterization of the labeled proteins, has been
described in previous chapters.

AluI Protection as a Measure of CAP Binding. Within the CAP binding

Site at lag" is an Alal restriction site (see Appendix). CAP binding to the
203bp lag,” DNA fragment protects the DNA from Alal digestion at this site.
This is a convenient method for assaying CAP occupation of the lag;+ site.
Here, CAP (or a labeled counterpart) was allowed to bind to lag;+ as usual at
room temperature. At the end of the binding incubation, 10 units of A131

119

were added, and the sample incubated for another 60min at room
temperature. At the end of the second incubation, the digestion reaction was
quenched with 20mM EDTA. The DNA was then phenol and
phenol:chloroform (50:50) extracted and ethanol precipitated (Maniatis et al.,
1982). The DNA pellet was dissolved in 10|rl TE (20:0.1) (20mM Tris, pH 7.5,
0. 1mM EDTA) and run on a nondenaturing gel. The presence of a 160bp
DNA band indicated CAP binding. Control experiments with no CAP protein
present were performed to ensure complete Alal digestion of the lag;+ DNA
under these conditions.

Fluorometers, Scan Conditions, and Fluorescence Samples. Two
fluorometers were used in this study. The first fluorometer is described in
Chapter 2, and was used as described there. Labeled CAP fluorescence was
also monitored with a SLM Instruments 4800 fluorometer controlled, along
with data acquisition, by computer hardware and software from On-Line
Instrument Systems. This instrument produced uncorrected fluorescence
data.

Typical fluorescence samples used with the SLM instrument contained
300nM of EITC label for EITC-labeled CAP, and 400nM label for LYIA-
labeled CAP in fluorescence binding buffer. The instrument was set to 8nm
and 16nm excitation and emission slit widths, respectively. Emission scans
of EITC-labeled CAP samples were made fi'om 520nm to 570nm with
excitation at 500nm, while those for LYIA-labeled CAP were from 500nm to
550nm with excitation at 430nm. Each spectrum consisted of five scans
averaged together. All incubations and scans described below were at room
temperature.

Dilution of the samples fi'om added components caused fluorescence

signal loss. This signal loss was corrected by performing the following

 

120

experiments. For each labeled CAP, a sample was simply titrated with
fluorescence binding buffer, and the resulting spectra collected. Taking the
spectrum maximum at each titration point and calculating the percent signal
decrease allowed for plots of percent signal loss vs. percent sample dilution.
These data were then fit by linear least squares regression to yield an
equation to calculate the expected signal loss due to sample dilution. This
control was performed twice for each labeled CAP, and the resulting data
averaged to obtain the following equations:

EITC-labeled CAP-
%FC = 98 - 0.70(%VC), (1)

 

LYIA-labeled CAP-

%FC =97 - 0.72(%VC), (2)
where %FC and %VC represent the percent fluorescence change and percent
volume change for a particular spectrum. In experiments which utilized this
instrument, the percent fluorescence change fi‘om dilution was calculated
and used to determine the value to add back at each wavelength of a
Spectrum by the following equation:

lei = A, + (81, - (Sli(%FC / 100))), (3)
where Ad is the corrected fluorescence value at wavelength i, hi is the
uncorrected fluorescence value at wavelength i, and Sli is the fluorescence
value of the first (undiluted) spectrum at wavelength i.

DNA Promoter Effects on Labeled Protein Fluorescence. The effect of
DNA interaction on the conformation of labeled proteins was investigated by
titrating DNA into fluorescence samples. Typical experiments ran as follows:
after a “protein only” spectrum was obtained, promoter DNA was added and
the sample incubated for 15min (CAP samples) or 30min (RNA polymerase
samples) prior to taking another spectrum. This pattern was repeated

121

throughout the DNA titration range. The effect of different promoter
sequences was measured by using the five DNA fiagments described above.
For all DNA fragments the titration range was from 0.1 “promoter
equivalent” (that is, one promoter DNA fragment per ten protein molecules)
to 2 or 4 “promoter equivalents”.

The effect on labeled protein fluorescence of the presence of both
proteins at a promoter was monitored by first making a “protein only”
spectrum. For labeled CAP, the desired DNA promoter fiagment was added
(at 1 “promoter equivalent”) to the sample, and another spectrum taken after
a 15min incubation. Then unlabeled RNA polymerase was added (at 1:1:1
stoichiometry) and the sample incubated for another 30min prior to taking
another spectrum. Finally, heparin was added to 1mg/ml, and the last
spectrum taken after 5min incubation.

EITC-labeled RNA polymerase-unlabeled CAP-promoter DNA
complexes were formed in a slightly different order. After taking the “protein
only” spectrum, preformed CAP-promoter DNA complexes were added to the
EITC-labeled RNA polymerase sample. Again, the stoichiometry of the
unlabeled CAP, DNA fi‘agment, and EITC-labeled RNA polymerase was
1:1:1. The rest of the experiment was performed as described above. The
effect of different promoter sequences was also assessed by studies with each
of the five DNA fiagments. Control experiments were performed without
DNA to observe any potential interactions between CAP and RNA
polymerase that might interfere with interpretation.

Quenching of Labeled CAP Fluorescence. Quenching of labeled CAP
fluorescence when in DNA complexes was performed to gain more
information about CAP conformation. Labeled CAP-promoter complexes
were formed at 1:1 stoichiometry as detailed above prior to taking a base

 

122

spectrum. Then any of several quenchers was titrated into the sample. A
spectrum was taken after adding the quencher and incubating for 5min.
This was repeated at each quencher titration point. The fluorescence
maxima fiom the titration range were used to generate Stern-Volmer plots of
Fo/F vs. quencher concentration, where F0 is the base spectrum maximum,
and F is the spectrum maximum at individual titration points.

Various quenchers were used: N aI, dithiothreitol (DTT), and
acrylamide. NaI titrations ranged from 5mM to 100mM. The ionic strength
of the sample was not maintained throughout N aI titrations. Acrylamide
and D’I'I‘ titrations ranged fi'om 10mM to 500mM and 1M to 10mM,
respectively. The effect of these quenchers on complex stability was
determined by adding quencher to preformed complexes (in fluorescence
binding buffer at room temperature) and incubating for 5min. The samples

were then loaded and run on nondenaturing gels as previously described.

 

123

Results
Interactions Between Labeled Protein and Various DNA Fragments

Cause Similar Changes in the Fluorescence Spectra of the Labeled Proteins.
Labeled CAP and RNA polymerase were titrated with promoter DNA to
observe the effect of DNA binding as monitored by extrinsic fluorescence.
Figure 4.2 shows typical uncorrected fluorescence and relative fluorescence
efiiciency spectra for EITC-labeled CAP titrated with wild type lag promoter
DNA Increasing amounts of DNA causes a fluorescence signal loss, but no
major wavelength shifts are observed. Figure 4.3 shows the results of adding
other DNA fiagments to EITC-labeled CAP. As can be seen, the pattern of
signal loss with increasing amounts of DNA is repeated with each of the
promoter fragments (laeLS-UV5, lag-gal, and gal). Further, the same pattern
of signal loss is seen with calf thymus DNA, which probably does not contain
a significant number of CAP sites within its sequence (see Figure 4.4). The
results of titration experiments with LYIA-labeled CAP also show a decrease
in the fluorescence signal as DNA is added to the sample (data not shown).
Titration of EITC-labeled RNA polymerase with lag or lanV5 promoter DNA
results in little change fiom the “protein only” spectrum (Figure 4.5). The
sequence of the DNA does not seem to alter these patterns. Both
fluorometers yield similar spectra indicating that the results are not an
instrumental artifact.

Formation of labeled CAP-polymerase-DN A complexes also causes a
fluorescence signal loss with little wavelength change. Figure 4.6 shows
typical results with LYIA-labeled CAP, RNA polymerase, and promoter DNA
fiagments. As observed above, adding promoter DNA at 1: 1 stoichiometry
(one DNA promoter fragment per CAP molecule) to the labeled CAP results

 

124

Figlge 4.2. Titration of EITC-Labeled CAP with Lag Promoter DNA. EITC-
labeled CAP was titrated with lag+ DNA as described in Materials and
Methods. Here, FL denotes fluorescence (arbitrary units) and RFE denotes
relative fluorescence efficiency (arbitrary units). EC stands for EITC-labeled
CAP. Promoter indicates lae DNA. Panel A shows the fluorescence spectra
from the SLM 4800 fluorometer, and Panel B shows the relative fluorescence
efficiency spectra. The spectra of Panel B are as indicated in Panel A.

125

 

    

 

 

 

A.
2
1.5
FL 1
- - - 5 FIG/Promoter
----- 2 PIC/Promoter
0 5 - - — - l EC/Promoter
I — - - - 0.5 EC/Promoter
0 1 A r
525 530 535 540 545 650 555 560 565
Wavelength (nm)
B.

 

 

 

 

0 A r A n m J A

525 530 535 540 545 550 555 560 565
Wavelength (nm)

126

Figure 4,3. Titration of EITC-Labeled CAP with Various Promoter DNA
Fragments. EITC-labeled CAP was titrated with lanV5-L8, gal, or lag-gal
promoter DNA as described in Materials and Methods. Here, FL and EC are
as described in legend to Figure 4.2. Panel A shows the EITC-labeled CAP
fluorescence spectrum titrated with lanV5-L8 promoter DNA, Panel B
shows the titration of EITC-labeled CAP with gal DNA, and Panel C shows
the EITC-labeled CAP spectrum titrated with lag-gal promoter DNA. The
spectra of Panel B and C are as indicated in Panel A.

127

 

     
    
 

 

    
 

 

 

 

 

 

 

 

A.
2.5
2 L
1'5 l ECOnl
FL ’
1 . - - - 5 EC/Promoter
----- 2 EC/Promoter
0.5 - - - - l EC/Promoter
. — - - - 0.5 EC/Promoter
0 L . A - . - -
525 530 535 540 545 550 555 560 565
Wavelength (nm)
B.
2.5
O r I r A
525 530 535 540 545 550 555 560 565
Wavelength (nm)
C.
2.5

 

 

 

 

 

525 530 535 540 545 550 555 560 565
Wavelength (nm)

128

Figgle 4.4. Titration of EITC-Labeled CAP with Calf Thymus DNA. The
titration of the labeled CAP with calf thymus DNA was performed as
indicated in the Materials and Methods section. The abbreviations FL, RFE,
and EC are as in legend to Figure 4.2. Calf thymus DNA was added to the
samples to the appropriate protein to “promoter” ratio. Panel A shows the
uncorrected fluorescence, while Panel B shows the relative fluorescence
efficiency spectra.

129

 

  
 

 

  
  
 

  
 

   

 

 

565

 

 

    

 

 

A.
2.5
2 t
1.5 »
——EC Only
FL
— - - 5 EC/"Promoter"
1 ----- 2 EC/“Promoter”
— - - ' l EC/‘Promoter"
0.5 l- — - - - 0.5 EC/"Promoter"
0 1 r r r r
525 530 535 540 545 550 555 560
Wavelength (nm)
B.
30
25 t
20 t
RFE 15 »
10 - — - 10 EC/"Promoter"
----- l EC/"Promoter"
— - — '0.5 EC/"Promoter"
5
o r r
525 530 535 540 545 550 555 560

Wavelength (nm)

565

130

Figu_re 4.5. Titration of EITC-Labeled RNA Polymerase with Promoter DNA.
EITC-labeled RNA polymerase was titrated with either wild type lae or
lanV5 promoter DNA as detailed in the Materials and Methods section.
RFE denotes relative fluorescence efficiency, and ER EITC-labeled RNA
Polymerase. Panel A shows the relative fluorescence efficiency spectrum of
EITC-labeled RNA polymerase titrated with wild type lae promoter DNA,
and Panel B shows the labeled polymerase spectrum titrated with lacUV5
DNA. The spectra in Panel B are is indicated in Panel A.

131

 

25

20>

 

A

— ' - 'lER/Prornoter

— - - - IBR/Pmmotero
Heparin

A_ L

A

 

A

 

 

535

540 545 550
Wavelength (nm)

555

560

565

 

 

4A A

A

 

 

 

530

535

540 545 550
Wavelength (nm)

555

560

565

132

Fignle 4,3. Formation of LYIA-Labeled CAP Transcription Complexes.
Complexes of LYIA labeled CAP-polymerase-promoter DNA were formed and
monitored as described in Materials and Methods. P represents promoter
DNA , LC denotes LYIA-labeled CAP, and R indicates RNA polymerase. FL
denotes fluorescence. Panel A shows the LYIA-labeled CAP-RNA
polymerase-lag DNA transcription complexes; Panel B, the laeL8-UV5
complexes; gfl complexes appear in Panel C; and those with 1 -gal in Panel
D. The spectra for Panels B, C, and D are as indicated in Panel A.

133

A85 fines—983 A65 598—953

 

 

 

  

 

 

  

 

 

 

 

 

 

 

 

new ova ems one can one as as new man. can one one can can Be So mom
i , 7 c . . . . . . l o
a _
_
1 .
a,
.5... A Ah
1 'l'5e'l5050505050'050505-56505-505150‘0-‘IL om for”"lll"”l'llllli'lu"r'a'lllclll'cII.Ill- ''''' II-l hnehh'EJf
..fiufiuufi... .......... it... 1 a ...... .. s
5 m 5
.Q 0
A85 fines—263 A85 sumac—26>.»
new $3 ems can can can AS as new new ova one one has one as as mom.
i l . i l . o . . . i l q . o
stun—om
+§5§l.l
DAV ES‘Hfi ccccc r 0.0
am

am
e.
141-312991912th I 11.... Pruitt-nlfiufitiiichs

1/ l\ H

m4 m4

 

 

 

 

 

 

 

134

in a signal loss, regardless of the fragment used. Adding RNA polymerase (at
1:1:1 stoichiometry) followed by heparin to EITC-labeled CAP-promoter DNA
complexes only decreases the fluorescence signal further. The sequence of
the DNA fi‘agment again does not appear to influence the signal change. As
with the titration experiments, calf thymus DNA caused the same sorts of
EITC- or LYIA-labeled CAP signal changes in the presence of RNA
polymerase and heparin (see Figure 4.7). Further, in the absence of DNA,
heparin and RNA polymerase each cause a labeled CAP fluorescence loss,
complicating the interpretation of the above spectra (Figure 4.8). Our data,
then, presumably reflect interactions between heparin, polymerase, and
CAP, in addition to effects from DNA binding to the proteins.

Solutions of CAP, RNA Polymerase, and DNA Are Complicated
Mixtures. It seemed clear from the calf thymus titration data that
nonspecific-DN A protein complexes are present in fluorescence experiment
samples. The fact that heparin and RNA polymerase cause a fluorescence
signal change for labeled CAP in the absence of DNA similar to that seen in
the presence of DNA indicates significant interactions between these
molecules as well. It thus seemed in order to determine the degree to which
specific complexes are indeed present in solutions in which DNA and protein
are present at 1:1 stoichiometry. To this end, protection fiom A121 digestion
and runoff transcription experiments were performed.

At 1:1 stoichiometry, EITC-labeled CAP provides little protection of
wild type lag promoter DNA from A1111 attack (see Figure 4.9); hardly any of
the 160bp fiagment that is diagnostic for specific CAP-DNA binding is seen.
Only at a stoichiometry of 3:1 (EITC-labeled CAPzDNA) does the 160bp
fragment appear, indicating labeled CAP occupancy of the lag DNA binding

135

Figggg 4.7 . Formation of Labeled CAP “Transcription Complexes” with Calf
Thymus DNA. The “transcription complexes” were formed as described in
the Materials and Methods section. “P” represents calf thymus DNA , EC
denotes EITC-labeled CAP, and R indicates RNA polymerase. FL denotes
fluorescence. Panel A shows the EITC-labeled CAP spectra, while Panel B
shows the corresponding LYIA-labeled CAP spectra.

25

525

L2

08

FLO.6 »

0A

02 >

136

 

 

---lEC/"P"
-"'-- IIHJCPVR

— - — - l EC/"P'VR +
Heparin

A 4 A

    
   

 

 

530

535

540 545 550

Wavelength (nm)

555

560

565

 

 

 

L

 
   

o...-
...1

. “_____,,-_._._._.-.___,_,_ _____ ----.--_,J

 

 

505 510

515

520 525 530
Wavelength (nm)

535

540

545

137

m. Heparin and RNA Polymerase Interact with Labeled CAP.
Polymerase was added first to samples where both polymerase and heparin
were present. FL represents the uncorrected fluorescence, and EC or LC
stand for EITC-labeled CAP or LYIA—labeled CAP, respectively. Panels A
and B show heparin’s effect on EITC-labeled CAP and LYIA-labeled CAP,
respectively. Panels C and D show the effect of adding polymerase before
heparin to EITC-labeled CAP and LYIA-labeled CAP, respectively.

138

A85 guano—96>»

m¢m ova own one can can mfim on mom

 

{ 4 1 l a A.

5:93
+o-Eofih_cm+3 . n u u . Dd

conga—om + 3 l l l

 

 

 

 

 

 

7. 5153'
CoollallnilollI-IioI'lalilillo al.-IOl-OIIIOI-I‘fi
TI/ ...u l: u

9H
:5: 4.55333
mvm ova wmm one can can mfim ch mom
. 4 . . . . J o
cranoz+3l l l . m6
niouql
'5'55'5555' “‘lv
1/ Illllllll|\ .n
04

 

 

 

Ah

Arm

mmm own 0mm

:5: sauce—263
can Ea o3 mmm can. can

 

 

 

    
 

 

 

 

 

  

. 4 . l . o
5.395 1 ad
+ onuSEbcL+Um o . a n n _
sauna—om + 0w l l l a
Hm
mg
a
@N
:5: 5.9.2263
mwm own mom can mvm ch mmm one can
4 . w . . . o
. We
Sanger + 8 l l l
3:0 on H
Hm
@~
1 w
ed

 

 

 

139

W. CAP Protection of Laf DNA From $1 Digestion. Preformed
EITC-labeled and unlabeled CAP-lag;+ DNA complexes were digested with
A131, and the DNA separated on a 5% nondenaturing polyacrylamide gel.
The amounts of labeled and unlabeled CAP are indicated above the picture,
expressed as a multiple of the amount of promoter DNA in the samples. A
lane of undigested 203bp lag? DNA is shown for reference, as are finaII
digested pBR322 markers (denoted by M). The DNA bands due to A131
digestion of the lac DNA are indicated. The presence of a 160bp band
indicates protection from A131 attack by CAP binding.

140

 

 

Al <75 cosh

%Oéfim
- - - n20
m m H 508$ mo 9553

+
m

v+l
03+-
N+I
v-l-l-I

141

site. Samples with unlabeled CAP, however, show some protection at 1: 1
stoichiometry, and nearly complete Alul protection at five fold excess CAP.
Certainly some specific complexes are present at 1:1 stoichiometry with
unlabeled CAP. The lack of such complexes observed with EITC-labeled CAP
may be due to the fact that EITC-labeled CAP dissociates fiom wild type 1a;
promoter DNA about twice as fast as unlabeled CAP. The EITC-labeled CAP
complexes dissociate more rapidly, making more DNA available for attack by
$1, resulting in less of the 160bp fi‘agment. All in all, there may be a
limited number of specific complexes formed at 1: 1 stoichiometry with EITC-
labeled CAP.

Whether specific CAP complexes are formed when RNA polymerase is
present was unclear from the spectral data, so samples under conditions of
1: 1:1 stoichiometry were assayed by runoff transcription. Figure 4.10 shows
runofi' transcription from LYIA-labeled CAP and EITC-labeled CAP samples
under such conditions fiom the wild type lac promoter. Polymerase alone
initiates primarily at P2. Unlabeled CAP (at 1:1:1 stoichiometry) seems to
shift some, but not all, of the transcIiption to P1, as expected if specific CAP
binding occurred. LYIA-labeled and EITC-labeled CAP also shift some of the
transcription to P1. In the presence of labeled or unlabeled CAP, the amount
of P1 transcription (as detected by eye) relative to P2 transcription is about
10% to 30%. This suggests that some specific labeled CAP-polymerase-
promoter DNA complexes are indeed present under conditions of 1: 1: 1
stoichiometry.

Solutions of CAP, polymerase, and DNA are thus heterogeneous
mixtures of complexes. Fluorescence samples probably contain, in addition to
specific complexes, some nonspecific complexes, free protein, and perhaps
heparin-protein and protein-protein complexes as well. The spectrum from

 

142

W. Transcription From Specific Complexes In “Fluorescence Like”
Samples. CAP and RNA polymerase complexes were formed with lag
promoter DNA (at a stoichiometry of 1:1:1) and allowed to transcribe. The
proteins present in each sample lane are indicated above the picture. Control
samples under usual conditions of protein excess (see Chapter 2) show the P1
(63nt) and P2 (85nt) transcripts as indicated.

CAP
LYIA-CAP
EITC-CAP

Polymerase

 

143

 

144

such a mixture is an average over all the species present in the sample.
There does not appear to be an easy method to separate the specific
component of the fluorescence spectrum fi‘om those arising from free and
nonspecifically bound protein. Calculating difl‘erence spectra by subtracting
normalized calf thymus DNA-protein titration data from promoter DNA-
protein fluorescence data does not work for all of the promoter fragments
used, for example. While there is no simple way to separate the various
contributions to the overall signal, there are still some interesting things to
be learned from these “average” complexes.

Quenching of Labeled CAP Fluorescence Suggests Differences in
“Average” Solution Complexes. Since fluorescence samples appeared to be
mixtures of complexes, dynamic fluorescence quenching measurements of
labeled CAP samples were done to probe the possible differences in these
solutions. Figure 4.11 shows typical N aI quenching of LYIA-labeled CAP
complexes. The upward curvature exhibited by some of the data is probably
due to a static quenching component in addition to dynamic quenching.
Dynamic quenching occurs when a fluorophore absorbs light and then
collides with a quencher, in this case 1', before emitting fluorescent light; the
energy is transferred to the quencher instead. Static quenching occurs, on
the other hand, when the quencher forms a complex with the fluorophore
prior to excitation. The fluorophore then never has a chance to emit
fluorescent light, but instead directly transfers the energy from the absorbed
light to the quencher. As seen in Figure 4.11, preforming nonspecific
complexes with calf thymus DNA does not alter the quenching pattern of the
protein very much. More strikingly, “average” complexes with promoter DNA
show a more classic dynamic quenching profile (linear dependence on

quencher concentration).

as.
u—n
o

 

 

W.”

145

Figlge 4,11. Sodium Iodide Quenching of LIYA-Labeled CAP-DNA
Complexes. The Stern-Volmer plots were generated as described in
Materials and Methods. Best fits of the data were derived by linear or
nonlinear least squares regression as appropriate. Panel A shows the
quenching of labeled CAP-calf thymus DNA complexes compared with free
labeled CAP and free LYIA. Panel B shows the quenching of various LYIA-
labeled CAP complexes formed using each of the four promoter fiagments
described in Materials and Methods. LC denotes LYIA-labeled CAP.

146

 

1.8
1: FreeLYIA
1-7 ' 2: LC
6 3: LC+CalfThynnrs DNA
1. ~

1.5 ~
Fo/F 1.4 ~
1.3
1.2 ~

1.1

 

 

 

 

 

0 10 20 30 4o 50 60 70 80 90 100
Sodium Iodide (mM)
1 9
1.8 ‘ 1: LC+LactoseDNA ’4’]?
L 2: LC+LactoseL8-UV5 DNA 14""
1'7 3: LC+Galactose DNA /' '

1.6 . 4: LC + Lac-Gal Hybrid DNA
5: LC + Calf Thymus DNA 3 ,. 33d!” 1

 

 

 

 

o 10 20 30 4o 50 60 70 so 90 100
Sodium Iodide (mM)

147

These data certainly suggest that the “average” LYIA-labeled CAP complexes
with promoter and calf thymus DNA are difi‘erent.

Moreover, “average” LYIA-labeled CAP complexes formed with RNA
polymerase present exhibit slightly difi'erent quenching profiles as compared
with the corresponding CAP-only complexes. Figure 4. 12 shows these
results. “Average” calf thymus DNA-polymerase-labeled CAP complexes are
quenched to a somewhat greater extent than are “average” calf thymus DNA-
labeled CAP complexes, for instance. Here, too, sequence of the DNA
fiagment seems to have some influence on the quenching pattern. Formation
of complexes with each of the promoter DNA fragments results in difi‘ering
susceptibilities to NaI quenching.

Sodium iodide quenching of EITC-labeled CAP complexes is more
complicated. EITC-labeled CAP in “average” complexes exhibit quite
different quenching as compared with free EITC or labeled protein (see
Figure 4.13). The downward and upward curvature of some of the labeled
CAP data likely reflect a combination of several processes: conformation
changes due to protein-quencher interaction, multiple label sites with
difi‘erent quenching constants, and static quenching. Basically, EITC-labeled
CAP seems to possess an altered accessibility to the quencher when in DNA
“average” complexes. Figure 4.14 also shows N aI quenching of “average”
EITC-labeled CAP-polymerase-DNA complexes. As with LYIA-labeled CAP,
complexes with EITC-labeled CAP and RNA polymerase have different
quenching profiles compared with the labeled CAP-DN A complexes. The
sequence of the DNA appears to influence the quenching of these “average”
complexes.

Acrylamide and D'I'I‘ were also used to quench the fluorescence of
EITC-labeled CAP (see Figure 4.15). In fact, the downward curvature of

 

148

Figu_re 4.12. Sodium Iodide Quenching of LYIA-Labeled CAP-Polymerase-
DNA Complexes. The Stern-Volmer plots were derived from the fluorescence
data as detailed in Materials and Methods. Best fits were generated as
described in the legend to Figure 4.11. LC represent LYIA-labeled CAP.
Panel A shows calf thymus DNA “transcription complexes” compared with
labeled CAP and free LYIA. Panel B shows the transcription complexes with
the promoter DNA fiagments used in this work, as well as complexes with
calf thymus DNA.

149

   
 

 

1.8
1: Free LYIA

1.7 r 2: LC
3: LC + Calf Thymus DNA + Polymerase

 

1.6 -
1.5 »
Fo/F1.4 »
1.3 »
1.2

1.1

 

 

 

 

 

 

 

 

0 10 20 30 40 5O 60 70 80 90 100
Sodium Iodide (mM)
1.6
1: LC + Lactose DNA + Polymerase
2: LC + Lactose 1.8-U116 DNA + Polymerase
1.5 * 3: LC 4» Galactose DNA 4» Polymerase
4: LC + Lac-Gal Hybrid DNA + Polymerase
1.4 *
Fo/F 1.3 -
1.2 >
1.1 *
Maggi:
O
1 r r r
0 10 20 30 4O 50 60 7O 80 90 100

Sodium Iodide (mM)

150

Figure 4,13. Sodium Iodide Quenching of EITC-Labeled CAP-DNA
Complexes. The Stern-Volmer plots were derived from the fluorescence data
as described in Materials and Methods. Best fits were calculated as
described in the legend to Figure 4.11. Panel A shows the quenching EITC-
labeled CAP-calf thymus DNA complexes compared with labeled CAP and
free EITC. Panel B shows the sodium iodide quenching of complexes with
the promoter DNA fragments used in this work, as well as complexes with
calf thymus DNA. EC represent EITC-labeled CAP.

151

 

     

 

 

 

 

 

 

 

 

 

1.25
1.2 > 2 ID
/
I . I
C] I ,,,,,
1.15 » ’
I x .
FO/F I]. x"
1.1 » Cy
I
P;
. If 1: Free EITC
“’5 2: EC
3: EC + Calf Thymus DNA
1 1
o 10 20 30 40 5o 60 7o 80 90 100
Sodium Iodide (mM)
1'35 1: EC + Lactose DNA
2: EC + Lactose 1.8-UV5 DNA ‘_
1'3 ’ 3: EC+GalactoseDNA D__,..—--"
4:EC+Lac-GalHybridDNA 2 __..—-"'"p
1.25 ~
1.2 »
Fo/F
1.15
1.1
1.05 ~
' <>
1 J}: J 1 r 1 r r r L

30 40 50 60

Sodium Iodide (mM)

70 80 100

152

Figu_re 4,14. Sodium Iodide Quenching of EITC-Labeled CAP-Polymerase-
DNA Complexes. The Stern-Volmer plots were generated from the
fluorescence data as detailed in the Materials and Methods section, and best
fits were derived as described in the legend to Figure 4.11. Panel A shows
calf thymus DNA “transcription complexes” compared with labeled CAP and
free EITC, while Panel B shows the transcription complexes with each of the
promoter DNA fragments used in this work. EC represent EITC-labeled
CAP.

153

 

 
   

 

 

 

 

 

    
      
     

A.
1'25 1: Free EITC
2: EC
12 . 2 ,0 D‘\\3: EC+Calf1hymusDNA+Polymerase
/ D \
\
1.15
Fo/F

1.1

1.05
l
0 10 20 30 40 50 60 70 80 90 100
Sodium Iodide (mM)
B.
1.16
1: BC + Lactose DNA + Polymerase 3: EC + Galactose DNA + Polymerase

1 l4 _ 2: EC + Lactose 1.8-UV5 DNA + Polymc 4: BC + Lac-Gal Hybrid DNA + Polymerase

' 5: BC + Calf Thymus DNA + Polymerase
1.12

1.1

Fo/F 1.08 »
1.06 r ,. -
~-.-.-_- : :;':::.'L

1.04 O
1.02

 

 

0 10 20 30 40 50 60 70 so 90 100
Sodium Iodide (mM)

154

Figu_re 4.15. Acrylamide and Dithiothreitol Quenching of EITC-Labeled
CAP-DNA Complexes. The Stern-Volmer plots were generated from the
fluorescence data as described in the Materials and Methods section. Best
fits were calculated as described in the legend to Figure 4.11. EC denotes
EITC-labeled CAP. Panel A shows the acrylamide “quenching” of EITC-
labeled CAP-Lag promoter DNA and gal DNA complexes compared with free
probe and free labeled protein. Panel B shows the dithiothreitol “quenching”
of EITC-labeled CAP-lac promoter DNA and -calf thymus DNA complexes,
along with fi‘ee EITC and free labeled protein.

155

 

1.25 I

 

 

 

 

 

 

 

 

Fo/F
2: EC
3: EC-I-LactoseDNA
0'95 4: EC + Galatose DNA
0.9 1 1 4 1 1 L r r L
0 50 100 150 200 250 300 350 400 450 500
Acrylamide(mM)
1.4 [L
2
1.3 _______ ______________________ __
----- D 34>
1.2 4 ................................... 3
. ' O
4
Fo/Fm -------------------- -0- ..................... -4,
1 a
lereeEITC 1
09 2: EC
' 3: EC+CalfThymus DNA
4:EC+LactoseDNA
0.3 . J . . A , , J . a

Dithiothreitol (mM)

 

156

these data indicate that the main acrylamide and DTT effect on the labeled
protein alone is a conformational one through some interaction between the
quencher and the protein, rather than a dynamic quenching interaction. The
limited amount of data obtained reveal small difi'erences, apparently based
on the DNA sequence of the fi'agment used in forming the complexes.
“Average” complexes of EITC-labeled CAP and lag DNA are “quenched” by
acrylamide difl'erently than are labeled CAP-gal DNA complexes, for
instance.

While any of the quenchers used could cause a change in the extent of
labeled protein-DNA complex formation, this does not seem to be a problem.
It is difficult to assess the ability of the quenchers to disrupt or alter
nonspecific complexes, but gel retardation assays revealed that the
quenchers did not affect specific complexes. For example, gel shift
experiments showed that none of the quenchers caused specific CAP-lac+
DNA complexes to dissociate (data not shown).

 

157
D. .

The Heterogeneous Nature of CAP-Polymerase-DNA Solutions
Complicates Their Analysis by Fluorescence. The data described in the
previous section indicate that solutions of CAP, RNA polymerase, and DNA
are composed of a mixture of specific and nonspecific DNA-protein complexes,
fi'ee protein, and possibly protein-protein and protein-heparin complexes. In
fact, this rather bizarre behavior has been observed in the past, and has been
reported in the literature. As mentioned in the Materials and Methods
section, the CAP and RNA polymerase preparations used throughout this
work were on average 10 to 25% active in DNA binding and transcription.
Why the other 75 to 90% does not specifically bind to DNA or participate in
transcription is unclear. But, when monitored by fluorescence, DNA binding
activity seems to be quite high (see Figure 4.2, Panel B, for instance). Thus
the activity of the protein depends markedly on the technique used to define
and measure that activity. Various reports of 100% CAP activity in
fluorescence have been published (Wu et al., 1974; Takahashi et al., 1983).
Fried described a purification approach that purports to yield fully active
CAP in gel retardation assays, but we have not been able to reproduce this
result in our laboratory.

Even nonspecific CAP-DNA interactions are more complicated than
might be expected. For example, it was found that, in the absence of cAMP,
CAP forms very highly cooperative complexes with DNA of random sequence
(Saxe & Revzin, 1979). Further probing of these complexes, however,
indicates that more than one type of structure is formed. Hudson et al.
(1990) proposed yet another CAP-DN A binding mode. The complexes
described by these authors exhibited uniform binding of CAP to DNA (like
beads on a string) with little DNA bending. The cooperativity of the

 

158

complexes is low, suggesting that protein-protein interactions are not
important in their formation. These complexes are probably different than
those observed by Saxe and Revzin. Whether there are multiple forms of
CAP in these solutions remains to be determined.

The fact that various CAP and RNA polymerase complexes are formed
in solution means that our fluorescence data reflect an “average” complex.
The inability to separate one contribution fi'om another increases the
difficulty of extracting information from the overall signal. As a result, the
fluorescence experiments designed to probe CAP and RNA polymerase
conformational differences at various promoter DNA fragments are quite
difficult to interpret. And yet, keeping in mind the complicated nature of
these solutions, some useful information about protein conformations can
still be gleaned from the data.

The Conformation of CAP May Be Altered at Difi‘erent Promoters. The
quenching of labeled CAP fluorescence when in “average” complexes revealed
some unexpected results. CAP in complexes formed with calf thymus DNA
seem to have a difi'erent conformation than when bound to either Lag, gfl,
lagLS-UV5, or lag-m promoter DNA. Complexes formed with any of the
promoter fragments exhibited an altered fluorescence quenching pattern
compared with the calf thymus DNA complexes. A change in the ability of a
quencher to reduce the fluorescence of a fluorophore after a perturbation is
indicative of an environment change. In the case of proteins, environment
changes are likely due to changes in conformation. Our results do not reveal
whether the sequence of the promoter DNA influences the conformation of
CAP. “Average” complexes formed with the promoter DNA fragments all

showed a similar quenching profile, albeit a difl‘erent one than observed for

159

the nonspecific calf thymus DNA complexes; the resolution in these mixtures
was not adequate to distinguish variations among the promoters.

One explanation of these observations is that the specific CAP binding
site located within each of the promoter fragments causes an ordering of any
CAP molecules that are nonspecifically bound to the same fragment. The
specific binding of CAP to the wild type lac site could act as a nucleation site
for other CAP molecules, which might nonspecifically bind to the DNA
fi'agment in a manner that is difi'erent from the nonspecific complexes formed
on calf thymus DNA CAP is known to induce a bend in DNA when
specifically bound (Wu & Crothers, 1984). Perhaps the bend afi‘ects the
configuration of other CAP molecules bound to the fragment. Since calf
thymus DNA does not contain a specific CAP site, the nonspecific complexes
formed there could be difi‘erent from those formed on promoter fi'agments.

Moreover, CAP seems to be in a difl'erent conformation when RNA
polymerase is present compared with CAP-DNA complexes, as indicated from
the fluorescence quenching data. The fluorescence from LYIA-labeled CAP-
polymerase-lac” DNA complexes is quenched more readily than is the
fluorescence from LYIA-labeled CAP- + complexes, for example.
Interestingly, the conformation of CAP in CAP-polymerase-DN A complexes
seems to be altered by the DNA sequence of the fiagment used. Each of the
DNA fragments used resulted in somewhat different labeled CAP-
polymerase-DNA quenching profiles, in contrast with the CAP only
complexes (see above). Perhaps the precise promoter sequence is more
important to RNA polymerase conformation. RNA polymerase could have
altered conformations at the various promoters, and so might cause
alterations in the conformation of CAP through protein-protein interactions.
If this is true, then the ability to quench the fluorescence of EITC-labeled

 

160

CAP-RN A polymerase complexes should be dependent on the sequence of the
promoter DNA. RNA polymerase is also known to bend promoter DNA
(Kuhnke et al., 1987; Ceglarek & Revzin, 1989; Kuhnke et al., 1989). Ifthe
polymerase-induced DNA bending is DNA sequence-dependent, this could
also contribute to a complex with a difl'erent structure and protein
conformations.

These quenching data tantalizingly suggest that the conformation of
CAP may be altered by RNA polymerase promoter binding depending on the
sequence of the promoter DNA. If indeed true, then transcription complexes
may be more varied than previously believed. The next chapter describes
experiments designed to probe CAP- and RNA polymerase-induced bending

 

 

that indeed suggest difl'erent CAP and RNA polymerase conformations at the
lactose and galactose promoters.

Chapter 5
CAP- and RNA Polymerase-Induced Bending at the Lactose and
Galactose Promoters

1111mm

Early studies of DNA-protein interactions assumed a rigid-rod
structure for the DNA. Much work has since shown that DNA is quite
flexible, being able to assume a variety of interchangeable conformations. In
addition, DNA often contains sequence specific intrinsic bends (at A-T tracts,
for example). Furthermore, protein binding can induce DNA bending.
Catabolite activator protein (CAP) binding to lac DNA is the classic example
of such induced bending: CAP binding induces a 290° bend in the DNA (Wu
& Crothers, 1984; Porschke et al., 1984; Kotlarz et al., 1986; Liu-Johnson, et
al., 1986; Schultz et al., 1991). RNA polymerase can also cause bending
when it interacts at promoter region DNA (Kuhnke et al., 1987; Ceglarek &
Revzin, 1989; Kuhnke et al., 1989).

Many questions come to mind concerning protein-induced DNA
bending. Certainly two major questions are 1) is the bending by CAP and
RNA polymerase the same at different promoters, and 2) what role does
bending play in the mechanism of transcription. How DNA bending by CAP
affects transcription stimulation is not clearly understood, although a range
of hypotheses have been proposed (for a review see Kolb et al., 1993). A
sharp CAP-induced bend could facilitate DNA unwinding by RNA
polymerase. Another proposal suggests that the CAP-induced bend stores
energy that helps RNA polymerase escape from the promoter more efficiently
by breaking protein-protein and protein-DNA contacts. And while CAP-

161

 

162

induced bending is well documented for wild type lactose promoter DNA, less
is known about such bending at the galactose promoter.

This chapter describes experiments to probe CAP- and RNA
polymerase-induced bending at lag and gal promoters to more fully answer
both of the above questions. The data also provide indirect glimpses into the

conformation of the proteins at the two promoters.

 

163
WWW
Proteins. CAP and RNA polymerase were purified as described in

Chapter 2.

Binding Reactions and Gel Retardation Assays. Binding reactions
were performed as detailed in Chapter 2 at 37°C in binding buffer (40mM
Tris, pH 8, 10mM MgClz, 0.1mM EDTA, 2011M cAMP (if CAP was present),
and 100mM NaCl). Gels and retardation assays were also as described in
Chapter 2 with the following differences. In addition to TBE, indicated

 

nondenaturing gels were run with a TEAc running buffer (40mM Tris, pH
8.1, 20mM sodium acetate, and 2mM ETDA, plus 2011M cAMP if CAP was

 

present in the samples). The TEAc running buffer was recirculated ,
throughout the electrophoresis period by a peristaltic pump. One other
modification was the increase in cAMP concentration from 2011M to ZOOuM in
samples, gels, and running buffers when any of the gal DNA fragments were
used.

Circular Permutation Assay. The resolution of protein-induced DNA
bending by gel electrophoresis involves the use of a series of similar DNA
fragments. The fragments contain the same DNA sequences, but the protein
binding sites are located at different positions in the fragment (Wu &
Crothers, 1984): some fragments have a site near the fragment middle, while
others have a site near one fragment end (see Figure 5.1). The term “circular
permutation” was coined to described the cyclic nature of the fi‘agment series.
If the DNA is bent, then fragments with the bend in the middle will migrate
more slowly through a gel than will the corresponding fragment with the
bend near one end because the bend at the middle site causes a smaller mean
square fragment end-to-end distance than at the end site (McGhee &
Felsenfeld, 1980). Figure 5.1 illustrates the potential migration of a

164

Figure 5,1. Protein-Induced DNA Bending Affects DNA Gel Mobility. Panel
A shows the mobility of hypothetical DNA fragments with bound protein
(circles). Proteins that induce a DNA bend alter the relative mobility of the
protein-DNA complex depending on the position of the bend relative to the
fragment ends. An induced bend in the middle of a fragment reduces the
fragment mean square end-to-end distance, resulting in a slower migrating
protein-DNA complex than an induced bend near a fragment end, which has
a larger mean square end-tocend distance. In the absence of induced DNA
bending, the position (relative to the fragment ends) of the bound protein
does not influence the mobility of the protein-DNA complex. In this case,
protein binding does not change the mean square end-to—end distance of the
DNA fragment. Panel B shows one method to generate a series of circularly
permuted DNA fragments. A tandem head-to-tail dimer of a hypothetical
DNA fragment with a protein binding site (black box) is digested with several
restriction enzymes that out only once within the sequence of the fragment
(labeled 1 through 4). Digestion of the hypothetical dimer with the enzymes
1 through 4 results in the series of circularly permuted fragments shown
below the dimer.

165

 

 

 

 

 

 

 

 

 

FL ll ll H [—
Complexed DNA
Free DNA _, __
1 2 3 4 1 3 4 1
:— 1 J— 1 l I! 1 fl]
1 L
2 __-_
3 L
4 _-__

 

 

 

 

 

 

 

 

 

166

circularly permuted DNA fragment series when a protein induces a DNA
bend. Another way to think about this is that a fragment with a bend near
the end is more linear than one bent near the middle, hence behaves more
like free DNA when traveling through the gel matrix (Klevan & Wang, 1980).
That this assay represents actual DNA bending has been confirmed by
several independent approaches, including electrodichroic observation of
CAP-DNA complexes (Porschke et al., 1984), and measurements of rates of r.
ligation of DNA minicircles with an incorporated CAP binding site when in
the presence of CAP (Kotlarz et al., 1986). As can be seen in Figure 5.1, the
minimum number of DNA fragments in a circularly permuted series is two:

 

one with the site in the middle, and one with the site near a fragment end.
The relative mobility of a complex is defined as the migration distance of that
complex (measured from the bottom of the gel well) divided by the distance of
migration of the free DNA.

DNA Promoter Fragments. Several DNA promoter fragments were
used in this study. The 203bp fi‘agments containing the wild type lag and
lagL8-UV5 promoters, and the 230bp wild type gal promoter fragment were
described in Chapters 2 and 4. The 230bp gal promoter DNA was circularly
permuted by digesting pMHSZSO with the appropriate restriction enzymes
(see Figure 5.2).

In addition, several other fi'agments used were in this work. A 789bp
wild type lag fiagment, which includes within its sequence the 203bp wild
type lag DNA sequence, was used as a longer version of the lac promoter.
Permuted 725bp DNA containing the wild type gfl promoter region provided
longer DNA length variations of the circularly permuted 230bp gfl fragment
(see Figure 5.2). A permuted 7 68bp lacUV5-CAP" fi'agment was used to
observe the effect of a wild type CAP site coupled with the mutant lagUV5

167

Figlge 5,2. Structure of the Various Promoter DNA Fragments Used in This
Work. The RNA polymerase and CAP binding sites are shown for the 203bp
wild type lag fragment and the 230bp wild type gal fragment. The longer
fragments contain sequence variations of these two promoter regions. The
230bp gal promoter fragment was circularly permuted by digesting a tandem
head-to-tail dimer with M1, HM, M11, and HpaII. The two CAP sites
are shown (black and dark gray boxes), as is the P1 RNA polymerase site
(light gray box), for each of the permuted 230bp wild type gfl fragments. The
structure of the circular permuted (designated AF and BF) 725bp gfl
fragments shows the relative position of the wild type gal promoter region.
The permuted 768bp lag;UV5-CAP+ and 725bp lac-gal hybrid fragments have
similar structures as the 725bp gal fragments. The details of the 145bp gal,
lagUVS-CAP”, and lag-m hybrid promoter regions are given in the text.

168

 

 

 

 

 

 

 

 

 

 

 

 

203bp L25
RNA Polymerase (P1) ..
159-EV I» inn)
“ I. 42 W)
"RNA Polymerase (P2)
230bp 051*
RNA Polymerase (Pl)
1" +1 1)
_ r is”
'TmA Polymorue (P2)
mu m
m: ‘1 sun can :2 Hall m1
Lmll 1 mil 1
Wm
Hint! l
mu m
Hall m.
m m
725bp ®+
AF
R322 (287b ) (145 BR322 (293 )
BF
BR322 28 BR322 293b 145b

7 68bp LiUV5-CAP ”

AF
, 31312322 (287bp) ,flUVS-CAP”, pBR322 (293bp)

BF

“UV5-CA3: 2312322 5281922 £3322 993522

725bp I_»a_c-G_al

AF
Lac-G_al (145bp)
| 2311322 (287112) 1. : 11133322 (293bp!

BF
Lac-931. (145bp)

| gnaagg (271:2) ' 2333322 mam)

 

 

169

promoter site on DNA bending. Lastly, four permuted 725bp lag-gal hybrid
promoter DNA fragments were used to assess the affect of a lag CAP site
coupled to a gal promoter on the protein-induced DNA bending. These
fragments contained a wild type lag CAP site positioned at -60, as it is at the
wild type lac promoter region, relative to the gal P1 promoter. The sequences
of these fragments are listed in the Appendix.

The 725bp gal DNA fragments were constructed by ligating a 145bp
EQRI to WIN subfragment (~96 to + 49) of the gal“ 230bp sequence into E
pUC 18 along with two pieces of pBR322 DNA: the 287bp BamHI to HindIII F
and the 293bp BamHI to EQRI fragments. The “AF” version has the
structure BamHI-pBR322-EmRI-gall45-llingllII-pBR322-BamHI (see Figure
5.2). The “BF” version has the structure EmRI—gal145-HindIII-pBR322-
_Ba_mHI-pBR322-E_q;RI. These constructs were generously provided by Dr.
J ianli Cao.

Construction of the laggUV5-CAP+ 768 fiagments has been fully
detailed (Ceglarek, Master’s Thesis, Michigan State University, 1987). This
promoter region has the UV5 P1 “up” mutation described in Chapter 2

 

ligated to a lag?“ CAP site. The CAP site is positioned at -60, just as it is
located in the wild type lag; promoter. The AF and BF versions of this
promoter were cloned into pBR322, and generously provided by John
Ceglarek.

The four lag-gal hybrid fragments contained identical sequences,
except for single base pair mutations in the gal DNA. These mutations
eliminated CAP binding at the gal CAP1 site (positioned at -40) and/or RNA
polymerase binding at the gal P2 promoter (-5 transcription initiation). The
four variations were CAP1’P2' (similar to the lag-gal hybrid described in
Chapter 4), CAP1'P2+, CAP1+P2‘, and CAP1+P2+.

170

The 725 bp lag-gal hybrid promoter fragments (each in an “AF” and
“BF” version) were made by cloning in pUC 18 essentially as described above
for the 725bp gfl constructs. The lag-gal hybrid fragments (145bp) were
ligated into pUC 18 along with pBR322 fragments to generate the BF
versions of the 725bp fragments. The AF versions were generated by ligating
the hybrids into pDCMS (a pUC 18 derivative with EggRI' and HindIII'
polylinker sites and the pBR322 fragments attached to the plasmid at the _
BamHI polylinker site). The lac-gal hybrids and the EmRI'HdeII' mutant l—
pUC 18 plasmid were kindly provided by Diane Cryderman. Because of a
BamHI site located within the sequence of all of the lag-gal hybrids, isolation
of the 725bp lag-gal AF fragments fi'om their respective plasmids required
partial digestion with BamHI.

Gel Retardation-SDS Gel Assays. The protein content of complexes

 

separated by gel retardation assays was determined by running the
complexes on a subsequent SDS gel. Complexes were formed as described in
Chapter 2 with the scale-up of the reaction to lag of promoter DNA and
proportional amounts of protgin. These complexes were then separated on
nondenaturing polyacrylamide gels as previously described. Ethidium
bromide-stained gels were placed onto a UV transilluminator, and complex
bands were excised from the gels. Excised gel slices were placed into 10011.1 of
SDS loading buffer and incubated at room temperature for 2h. Prior to
loading the gel slices onto the SDS gel, they were heated at 90°C for 15min.
The gel slices were then loaded into the SDS gel wells along with the SDS
loading buffer and the SDS gel run at constant voltage (0.5V/cm) overnight.
The SDS gel system used a running buffer of 25mM Tris, pH 8, l90mM
glycine, and 0.1% SDS (w/v). "These SDS gels were silver stained by the
method of Guiliani et al. (1983) to visualize the protein bands. Comparison

 

171

of the protein bands in a lane arising from a gel slice with control lanes
indicated the presence, or absence, of CAP, and/or RNA polymerase, in the
gel slice.

 

 

172

Beams
CAP and RNA Polymerase Bending at Galactose Promoter Fragments.

Although CAP-induced DNA bending of lactose promoter DNA has been
known for some time (Wu & Crothers, 1984), whether CAP bending occurred
at the wild type galactose promoter was unknown. Using circularly
permuted 230bp gal DNA fragments to resolve potential CAP-induced
bending results in differently migrating CAP-DNA complex bands (see
Figure 5.3). These data indicate that CAP does indeed bend wild type
galactose promoter DNA, with the induced bend even larger than that
observed at lag (Stephanie Shanblatt, unpublished data). The results are

 

 

summarized in Table 5.1. ,

Suggestions of RNA polymerase-induced bending at wild type and
mutant galactose promoter DNA (Kuhnke et al., 1987) prompted the
investigation of polymerase bending with the circularly permuted 230bp gal
DNA fragments. RNA polymerase alone seems to induce little DNA bending
with the permuted fragments (see Figure 5.4). Further, the DNA bending in
CAP-RN A polymerase-permuted gal+ complexes is not apparently altered
from the polymerase only complexes. Indeed, similar results are seen with
203bp long fragments containing wild type lac or lacUV5 promoters (data not
shown; Lorimer & Revzin, 1986).

Figure 5.5 shows the bending by CAP and RNA polymerase at the
768bp lag:,UV5—CAP+ AF and BF fragments (see also Table 5.1). RNA
polymerase alone appears to cause a small DNA bend, while CAP induces a
typical large bend. When CAP and RNA polymerase are both present an
enhancement (more or less additive) of the total bending is seen. CAP and
RNA polymerase are functionally independent at this mutant lag promoter--
CAP binds well there but has little effect on transcription from the strong

173

Elgagafl. CAP-Induced Bending of Circularly Permuted Wild Type
Galactose Promoter DNA. CAP-DNA complexes were formed as described in
Materials and Methods with circularly permuted gfl DNA. Samples were
then run into a 5% polyacrylamide gel using a TBE running buffer.
Restriction enzymes used to yield particular permuted fragments are
indicated, as is the presence of CAP in the samples. The locations of free

' DNA and CAP complexes are also indicated.

174

MJQM—M
CAP-+-+-+-+

 

CAP-DNA
Complexes

{-Free DNA

175

Tabla 5.1. Relative Mobility of Protein-DNA Complexes. The relative
mobility of various CAP-, polymerase-, and CAP-polymerase-promoter DNA
complexes was measured as described in the Materials and Methods section.
All entries have about 10% error. ND stands for Not Determined.

176

 

DNA Fragment CAP-DNA Polymerase-DNA CAP-Polymerase-
Complexes Conylexes DNA Complexes
afl+ (230bp)
Hiafl 0.81 0.36 0.36
3:31.311 0.57 0.37 0.35
flpaII 0.52 0.36 0.38
MI 0.68 0.35 0.36
lao“ (203bp) 0.70 0.32 0.31
lacL8-UV5 (203bp) ND 0.33 0.33
lagzUV5-CAP+ ( 768bp)
AF 0.64 0.31 0.13
BF 0.80 0.38 0.32
gal+ (725bp)
AF ND 0.45 0.44
BF ND 0.49 0.32
lag-gal (725bp)
P2'CAP1'
AF ND , 0.29 0.30
BF ND 0.38 0.39
P2+CAP1’
AF ND 0.30 0.30
BF ND 0.38 0.39
P2‘CAP1+
AF ND 0.28 0.30
BF ND 0.36 0.34
P2+CAP1+
AF ND 0.30 0.31
BF ND 0.36 0.36

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

177

W. RNA Polymerase-Induced Bending of Circularly Permuted
Galactose Promoter DNA in the Presence and Absence of CAP. Protein-DNA
complexes were formed as previously described and run into a 4%
polyacrylamide gel using a TEAc running buffer. Restriction enzymes used
to produce the permuted 230bp gal fragments are indicated. Free DNA and
RNA polymerase- and CAP-polymerase—DNA complexes are also indicated.

178

 

I

<75 meal I. .1.. ..ll.]].

2:83:80

<zo5§oas£d<olv Is) )1 .1 Is 1.
d <2Dém§ofihom

4‘s“

++-++-++-++
+--+--+--+

ass 33m :85 mam

 

 

 

$30839”
n20

179

Figara 5,5. Protein-Induced Bending of LagUV5-CAP+ AF and BF Promoter
Fragments. CAP and polymerase complexes were formed as described above
and run into 4% polyacrylamide gels with a TBE running buffer. The AF
version has the protein binding sites near the middle of the fragment, while
the BF version has the sites near one end. Free DNA and protein-DNA
complexes are indicated, as are the presence of CAP and RNA polymerase.
Panel A shows the lagUV5-CAP” AF samples, and Panel B shows the BF
samples.

Polymerase + +

180

CAP+-+-

(— CAP-Polymerase-DNA
Complexes

«Polymerase-DNA
Complexes

(— CAP- DNA
Complexes

Free DNA

 

    
  
 
  

CAP-Polymerase-DNA
(— Complexes

(— Polymerase-DN A
Complexes

(— CAP-DNA
Complexes

<- Free DNA

 

181

lagUVS promoter. Similar experiments with the 789bp lag; wild type
promoter fragment or with 783bp lag," AF and BF type fragments show
results consistent with those from the lagUV5-CAP" fragments (laf 789 data
not shown; lag‘” 783 AF and BF results are unpublished data of Dr. Jianli
Cao). CAP and RNA polymerase show an additive bending of the DNA with
these longer length lac promoter fragments.

Data on bending induced by CAP and RNA polymerase binding at the
lag promoter appear to be a function of fragment length. Two explanations
for these results come to mind. Since this assay depends on the mean square
end-to-end distance of the DNA, the length of the DNA may be critical for
assay sensitivity. If an induced bend is small, short length permuted
fragments will all appear to be roughly linear and migrate similar distances
through a gel. Similarly, if the bend leads to DNA wrapping around the
protein, creating a “nucleosome like” DNA-protein loop, then all of the short
fragments may again migrate the same distance. Longer length permuted
fragments, on the other hand, would reveal the bending. An alternative
explanation is that CAP might unexpectedly be absent from presumed CAP-
polymerase-short length DNA complexes, which would preclude cooperative
complex DNA bending. Thus, each of these possibilities was tested, as
described below.

SDS Analysis of Protein-DNA Complexes. Protein-DN A complexes
from nondenaturing gels were run into SDS gels to observe which proteins
were present in the complexes. Such an analysis indicates that CAP and
RNA polymerase are present where expected. Figure 5.6 shows the results
for complexes formed with gal DNA. CAP is clearly present in the presumed
CAP-RN A polymerase complexes. Analysis of the corresponding 203bp lag;+

 

182

M55. SDS Analysis of Circularly Permuted Galactose Promoter DNA-
Protein Complexes. CAP-RNA polymerase-gal DNA complexes were formed
as described above and run into a 4% polyacrylamide gel. Complex bands

were excised and then run on a 12.5% SDS gel and silver stained as detailed

in the Materials and Methods section. Circularly permuted gal fragments
are indicated, as are the proteins.

183

 

%Olv
«4an?

51v

51v

.nsnlv

 

 

 

 

 

- - + + + + + Hlsm enema
+ u + + + + .. onshoEbom
- + + + + + - $6

nail: main nan 58m

184

complexes also indicates the presence of each protein (data not shown).

Thus, lack of CAP does not explain the bending results. Rather, the length of
the DNA does seem to be the limiting factor with these particular complexes
at the lag promoter.

An interesting side point is the presence of CAP in complex with RNA
polymerase and lagLS-UV5 DNA (Figure 5.7). Little CAP binding is seen at
the mutated L8 CAP site in the absence of RNA polymerase (data not shown).
However, polymerase can apparently stabilize CAP binding in such a
complex, suggesting an interaction between the two proteins. This agrees
with labeled CAP fluorescence data and with other results in the literature
(Pinkney & Hoggett, 1988; see Chapter 1).

Bending Induced by CAP and RNA Polymerase at Longer Length Gal
DNA Fragments. Since the SDS analyses indicated that complexes contained
the appropriate proteins, the bending experiments were repeated using the
7 25bp AF and BF gal promoter fragments. Results similar to those obtained
with the 768bp laggUV5-CAP+ were expected; that is, polymerase and CAP
would cooperate to yield a larger bend than either induces in the absence of
the other. As seen in Figure 5.8, however, the 725bp gal+ RNA polymerase
and CAP-polymerase complexes exhibit almost the same pattern seen with
the shorter gal+ fragments (see also Table 5.1). Polymerase does not seem to
induce a large bend, and CAP-polymerase-DNA complexes migrate the same
distance as the polymerase only complexes. The length of the 725bp gal
fragments should be quite adequate to distinguish any induced bending.
Thus the bending induced by polymerase and CAP at the galactose promoter
is different than that at the lactose promoter.

 

185

Figare 5.2. SDS Analysis of L_a_cL8-UV5-Protein Complexes. Protein-DNA
complexes were formed and separated from unbound DNA on a 4%
polyacrylamide gel. Complex bands were excised and um into a 12.5% SDS

gel and silver stained as described in Materials and Methods. The positions
of the protein bands are indicated.

 

186

CAP -
RNA Polymerase -
LacLB-UV5 +

+++
++
I
I

<—s'& B’
{—0

{-a

  
  
 

<- CAP
<— Free DNA

 

 

187

Figaga 5,8. Protein-Induced Bending of 725bp Galactose Promoter AF and
BF Fragments. RNA polymerase- and CAP-RNA polymerase-DNA complexes
were formed and run into a 4% polyacrylamide gel using a TEAc running
buffer as detailed in Materials and Methods. AF and BF versions of the DNA
fragment are indicated, as are the proteins present in each sample. In
addition, the positions of the free DNA and complex bands are shown.

 

188

CAP
Polymerase

Polymerase-DNA &
CAP-Polymerase—DNA
Complexes

Free DNA

 

 

 

189

Bending at Lag-Gal Hybrid Fragments. Because of the apparent
difference in protein-induced bending at lag and m, experiments with the
725bp lag-gal hybrid AF and BF fragments were performed to observe the
degree of DNA bending when a wild type lactose CAP site is at -60 relative to
the gfl P1 promoter site. A P2'CAP1’ hybrid was used to eliminate potential
interference due to bending from polymerase binding at P2, and especially
from CAP binding at CAP1.

CAP binds to its lac site at the lag-gal hybrid and induces a sizable
bend, although not quite as large as with wild type lactose promoter DNA
(Diane Cryderman, unpublished results). This could be due to the limited
amount of flanldng lactose sequence around the CAP site, as such sequence
is known to contribute to the extent of bending (Liu-Johnson et al., 1986).
The smaller bend could also be due to some influence of the gal DNA portion
of the hybrid sequence on CAP-induced bending.

Extending these experiments to include RNA polymerase yields
interesting results. Polymerase- and CAP-polymerase-induced bending are
shown in Figure 5.9 for the 725bp lag-gal P2'CAPl' version of the hybrid. It
can be immediately seen that polymerase- and CAP-polymeraseinduced
DNA bending at this fragment resembles that observed with the 725bp wild
type gal promoter fragments. Polymerase induces a small bend, but no
enhanced bending is seen when both proteins are bound. Indeed, all of the
hybrid lag-gal combinations yielded similar results: none of the CAP-
polymerase-hybrid complexes show the enhanced bending that would be
predicted by the presence of CAP (see Table 5.1). Whether the gal P2 and
CAP1 sites are wild type or mutated does not appear to make a difference as
far as the degree of bending in these complexes. Numerous experiments

indicate that CAP and RNA polymerase are indeed bound as expected to each

 

190

W. Protein-Induced Bending of 725bp P2'CAP1' Lactose-Galactose
Hybrid Promoter AF and BF Fragments. RNA polymerase- and CAP-RNA
polymerase-DNA complexes were formed with the P2'CAP1' version of the
hybrid and run into a 4% polyacrylamide gel using a TBE running buffer as
detailed above. Free and complexed DNA are indicated, in addition to the
proteins present in each sample.

191

<29 omen:

moonEoO

<zmswemaboa¢<o
a <ZQ5mSoE~om

 

 

onseofifiom
A20

192

of these hybrid fiagments (Diane Cryderman, unpublished data). Thus the
lack of bending is not due to lack of one of the proteins in the complexes.

193
D . .

CAP Bends Wild Type Galactose Promoter DNA. CAP appears to bend
wild type galactose promoter DNA as it does wild type lactose promoter DNA.
While the amount of bending was not precisely quantified for the gal
promoter, the extent of CAP bending is even larger than that induced at the

lactose CAP site. Data from x-ray crystallography of CAP-DNA complexes
shows that CAP bends DNA toward the protein molecule by 900 (Schultz et

 

al., 1991). The bend arises mainly from two 40° kinks. Perhaps CAP bends r
galactose promoter DNA in a similar manner
RNA Polymerase Bends Promoter DNA RNA polymerase appears to
induce a bend in each of the promoter DNA fi'agments studied. The extent of
i-

the bend is small since the mobility differences between circularly permuted
promoter-containing fragments is small. The additive bending observed in
the presence of both CAP and RNA polymerase at lag promoter DNA suggests
that a highly bent, “nucleosome like” looped complex is unlikely to be induced
by polymerase alone, since CAP-induced bending probably would not
increase the already large amount of bending in such a looped complex. In
addition, numerous independent studies of polymerase-DN A complexes
provide no indication that the protein and DNA interact along the complete
length of the DNA fiagment. For example, footprints of RNA polymerase-
promoter complexes are about 50bp long (Straney & Crothers, 1985;
Shanblatt & Revzin, 1983). Moreover, Buckle et al. (1991) did not observe
Photo-induced DNA cross-linking to CAP and RNA polymerase along the
entire length of their DNA fragment, but only near the expected binding
Sites. The orientation of the polymerase-induced bend is likely similar to
that induced by CAP: the DNA bends towards the protein molecule, thereby
maximizing the opportunity for DNA-protein contacts. This is consistent

194

with neutron scattering data showing RNA polymerase mainly on one side of
the T7 A1 promoter (Heumann et al., 1988). It is interesting to note that the
polymerase-induced bend in the 725bp gal DNA fiagment is apparently less
than that reported by Kuhnke et al. (1989). This group observed a
polymerase-induced bending using circularly permuted wild type and mutant
galactose promoter fi-agments of about 400bp. The reason for this
discrepancy is unclear.

CAP Plus RNA Polymerase Bending Is Different at the Galactose and
Lactose Promoter Regions. CAP and RNA polymerase appear to additively
bend lactose promoter DNA. At the galactose promoter, however, CAP and
RNA polymerase behave differently; the mobility of the CAP-polymerasegal
DNA complex varies little with the position of the promoter in the fragment.
At least two possibilities can be envisioned to explain the results at the wild
type galactose promoter (see Figure 5.10). First, polymerase binding could
reduce the size of the CAP-induced bend, lowering the total degree of bending
at the galactose promoter region (Figure 5. 10, Panel B1). Second,
interactions between the molecules may cause a DNA bend away from RNA
polymerase, thus effectively reducing the overall complex bend (Figure 5.10,
Panel B2).

Both of these possibilities have implications as to the role of DNA
bending in the control of transcription. In the first case, RNA polymerase
reduction of the CAP bend would require CAP to give up at least some of the
energetically favorable interactions that form the bend. CAP interaction
with polymerase could replace some or all of this loss, however, resulting in a
smaller overall complex bend. The second case requires a bend away from
RNA polymerase to increase the DNA mean square end-to-end distance
yielding a “linear-like” complex. How this might happen is not clear,

 

 

195

W. Possible CAP-RNA Polymerase-Gal Promoter DNA Structures.
CAP is represented by the small dark dimers, RNA polymerase by the large
light oval, and promoter DNA by the dark line. Panel A shows a hypothetical
structure of a bent CAP-polymerase-lactose promoter DNA complex. Panel
B1 shows a potential galactose promoter DNA complex with only a small
degree of DNA bending. Panel B2 shows a highly bent galactose promoter
DNA structure. In this structure one of the CAP molecules (perhaps the CAP
at -60) is rotated relative to the other two proteins, giving rise to the S “like”
path of the DNA. The rotation of one of the proteins could be such that the
DNA moves out of the plane of the page; that is, the rotation of the protein
need not be a full 180° as shown in the panel. The distance between the
fragment ends would thus be increased from that shown in Panel A,
resulting in a more “linear-like” structure.

196

 

197

however. The potential for CAP to bend the DNA away from itself seems
unlikely given the x-ray crystallographic evidence showing otherwise
(Schultz et al., 1991). Perhaps protein-protein interaction between the two
CAP molecules and RNA polymerase rotates at least one of the proteins away
fiom the axes of the others. The energy required could be ofl‘set by new,
favorable interactions between the proteins. Locally, each protein would
induce its usual bend, but DNA ends are rotated away from one another,

-. . l ‘1'"

resulting in an overall increase in mean square end-to-end distance. Which

of these possibilities represents the wild type galactose promoter complex ,

cannot be distinguished with this data.
CAP and RNA Polymerase Bend Lactose-Galactose Hybrid Promoter l-

DNA Like Wild Type Galactose DNA. Surprisingly, CAP and RNA

polymerase do not appear to additively bend lag-w hybrid DNA. The

presence of the wild type galactose CAP1 site does not alter the extent of

hybrid bending, nor does the presence of a wild type P2 polymerase site. It

therefore appears that the degree of complex bending is determined by the

 

galactose portion of the hybrid promoters, causing complexes with all of the
hybrids to show the same protein-induced DNA bending as does the wild type
gal promoter region. Indeed, the hybrid data suggest that the gal P1
polymerase site is the determining sequence. That the DNA sequence
influences protein-induced DNA bending is consistent with the findings for
CAP (Liu-Johnson et al., 1986).

On closer inspection, however, these data suggest that protein-induced
bending may not play as active a role in transcription initiation as has been
proposed. Zinkel and Crothers (1991) suggested that CAP-induced DNA
bending may store energy to facilitate the disruption of protein-protein and
protein-DNA contacts that leads to escape of RNA polymerase from a

 

198

promoter. Not all of the hybrid promoters exhibit P1 transcription
stimulation. In general, with the lag;+ CAP site located at —60, only the
hybrids with a CAP l+ site (that are likely to bind two CAP molecules)
stimulate P1 transcription (Diane Cryderman, unpublished data). The
hybrids that resemble a wild type lactose promoter, with only one CAP site,
do not show CAP-enhanced stimulation of P1 transcription. Since the
presence of the wild type gal CAP 1 site (or P2 promoter) within a hybrid
promoter sequence does not appear to influence the total degree of bending,
the hybrid DNA is presumably bent in a like manner at all of the hybrid
promoters. And yet, only in the presence of the CAP1 site is P1 transcription
stimulation observed: the presence of a CAP molecule at CAP1 seems to be
more important than protein-induced DNA bending. This leads to the
conclusion that protein-protein contacts are vital to CAP-induced
transcription stimulation, and that DNA bending is of lesser importance.
Perhaps protein-induced DNA bending influences transcription by stabilizing
protein-DNA complexes (by increasing the length of DNA that can interact
with a protein), but has no direct role in the regulation of transcription
initiation.

Bending Data Suggest Differences in CAP-RNA Polymerase-Promoter
DNA Complexes. The data described above strongly suggests that CAP and
RNA polymerase together bend DNA differently at the galactose and lactose
promoter regions. At the lactose promoter, the overall bend is additive, while
at the galactose promoter, CAP binding apparently does not alter the bend
induced by polymerase alone. This is quite significant as each protein alone
seems to bend DNA in a relatively consistent manner when comparing the
lag and gal promoter regions. The lack of additive bends at the gal promoter
suggests altered DNA-protein and protein-protein interactions as compared

 

199

with the lactose promoter. It is quite possible that the conformations of CAP
and RNA polymerase are different at the two promoters. This agrees with
the results discussed in Chapter 4 that suggest “average” solution CAP-DNA,
RNA polymerase-DN A, and CAP-polymerase-DNA complexes may have
different protein conformations influenced by the DNA sequence. Further
studies of CAP and RNA polymerase are clearly needed to understand what

role protein conformation plays in the control of transcription initiation.

 

Chapter 6

Summary

In order to study the conformation of catabolite activator protein (CAP)
and RNA polymerase when bound at promoter region DNA, each protein was
fluorescently labeled for use in fluorescence experiments. Chapter 2
described the labeling of CAP and RNA polymerase with eosin isothiocyanate
or lucifer yellow iodoacetamide. Extrinsic fluorescence overcomes some
limitations inherent in intrinsic fluorescence studies of protein-DNA systems.

Fluorescent labeling results were as follows:

 

1) each protein is labeled with one probe per protein molecule on 9*
average;

2) only the a subunit of RNA polymerase is labeled, suggesting the
presence of a highly reactive (or exposed) residue within the
sequence of o;

3) absorbance and fluorescence spectra from the labeled proteins
indicate that each probe is probably attached to a residue on the
surface of the protein; and

4) gel retardation and runoff transcription assays indicate that the
labeled proteins retain full activity, and so would allow
interpretation of fluorescence experiments in terms of properties
of the unlabeled proteins.

As the binding and transcription activities of each protein are not

markedly changed by the labeling, the areas of probe attachment are not
critical for protein function, but may nonetheless influence it. Chapter 3

described experiments aimed at determining the precise amino acid residues

200

201

to which the labels are covalently bound. Kinetic and transcription

experiments were performed to determine how labeling these apparently

noncritical areas influenced protein activity. The experiments in Chapter 3

revealed several interesting observations:

1)

2)

3)

4)

5)

6)

eosin isothiocyanate attaches to CAP residue lysine 102 (on the
opposite side of the protein from the DNA binding domain), while
lucifer yellow iodoacetamide modifies cysteine 93 of CAP (at one
end of one strand that contributes to the 5 roll that binds cAMP);
eosin isothiocyanate modifies lysine 251 within the 0' subunit
sequence (in the spacer region between conserved regions 1 and 2;
modification of lysine 102 of CAP causes a slight alteration in the
kinetics of DNA binding, perhaps brought about by a
conformational change induced by the attachment of the label;
neither lysine 102 nor cysteine 93 in CAP are probably involved
with RNA polymerase contacts as each labeled protein retains full
transcriptional activity;

alteration of 0' lysine 251 does not affect RNA polymerase
interaction with the lagUV5 promoter, but does appear to change
the ability of RNA polymerase to bind wild type lag DNA, which
can be overcome by lowering the salt concentration; and
modification of this a residue reduces the overall transcription

level by polymerase from wild type lactose promoter DNA.

While the CAP modifications appear to influence DNA binding

somewhat, the influence of lysine 251 on 0' function is much more

pronounced. Modification of this residue may alter promoter binding by

inducing a different subunit conformation, indicating the importance of this

region in subunit function. Further, the integration of a into the holoenzyme

 

202

must be such that lysine 251 is quite solvent exposed, so that its reactivity

with eosin isothiocyanate is higher than that of other lysine residues.

These data point to the importance of protein conformation in the
molecular mechanism of transcriptional control. More information about the

conformation of the proteins interacting with promoter DNA could be useful

in understanding the transcription initiation process. The fluorescence of
labeled CAP and RNA polymerase were thus studied to determine if the

proteins have altered conformations at different promoter regions. These

experiments are presented in Chapter 4. The data indicated the following:

1)

2)

3)

4)

5)

solutions of CAP, or RNA polymerase, and promoter DNA are
heterogeneous mixtures of specific and nonspecific complexes, as
well as free protein, and hence provide information about the
“average” solution complex;

solutions of CAP plus RNA polymerase along with promoter DNA
are also heterogeneous mixtures with specific complexes, protein-
protein complexes, complexes containing heparin, and free
protein, again resulting in an average signal;

the average CAP-RN A polymerase-promoter DNA complex may
be different depending on the promoter DNA used;

based on fluorescence quenching of CAP-DNA complexes it
appears that average specific complexes are different from
average nonspecific complexes, but that with different promoter
fragments the average specific complexes are similar; and

based on fluorescence quenching of CAP-RN A polymerase-
promoter DNA complexes it appears that differences exist
between these average complexes and average CAP-DNA

complexes.

. 8.3.: _-. "H'-

 

III:

203

These data suggest that CAP may have a single conformation when
bound at any promoter (irrespective of the DNA sequence) in the absence of
RNA polymerase. The binding of RNA polymerase may lead to changes in
the structure of both proteins, depending on the sequence of the promoter
region DNA. These potential conformation differences may be crucial to
proper regulation of transcription initiation at difi‘erent promoters.

Indeed, the amount of DNA bending induced by CAP and RNA
polymerase binding may be important not only to the potential configuration
of the complexes formed at various promoters, but also in the control of
transcription initiation. Chapter 5 described experiments to determine the
amount of bending at lactose and galactose promoter region DNA.
Experiments with a lactose-galactose hybrid promoter DNA construct were
also performed to observe the effect of this mixed sequence on DNA bending.
The data indicated several enticing results:

1) CAP and RNA polymerase bending at lactose promoter DNA is
greater in the presence of both proteins than with either protein
alone;

2) CAP and RNA polymerase do bend galactose promoter region
DNA, but together generate a bend that is much smaller than the
sum of the individual bends; and

3) CAP and RNA polymerase-induced bending at the hybrid
promoter DNA resembles that at the wild type gal DNA promoter,
implying that the galactose promoter sequence determines the
total amount of DNA bending in complexes with both proteins.

CAP and RNA polymerase bend DNA differently at the lactose and
galactose promoter regions, suggesting differences in the structure of the

complexes. The sequence of the RNA polymerase binding site seems to be

 

204

important in determining the amount of induced bending. These data, along
with the fluorescence, indicate that bending and protein conformation may
well be different at different promoter sequences. Bending, however, may
play only a small direct role in the molecular mechanism of transcription
initiation and its regulation. Further study of both should yield information
vital to the full comprehension of transcription initiation.

m1

 

Appendix

The upper strand sequences of the promoter DNA fragments used in

this work are shown below.
The sequence of the 203bp wild type lag DNA fragment is:

 

-140 -130 £121 -120 ~110
l I I I
CCGA’l'I‘CATT AATGCAGC'I‘G GCACGACAGG 'I'I'I'CCCGAC'I‘
-100 -90 -8|0 310
I
GGAAAGCGGG CAGTGAGCAC AACGCAATI‘A ATGTGAGTTA
$11
-60 -50 40 -30
I s/ | I I r’PZ
GCTCAC'I‘CAT TAGGCACCCC AGGCT'l'l‘ACA C'l'l‘l‘ATGC'l'I‘
-2lo 4'0 +1 ’1’] +1.0
CCGGCTCGTA TG'I'I‘GTGTGG AATI‘GTGACC CGGATAACAA
Aid
+20 +30 +40 +50 +60

l I l l I
'I'I'I‘CACACAG GAAACAGCTA TGACCATGAT TACGGA’I'I‘CA CTGG

The 203bp lagL8-UV5 sequence is the same as the 203bp lag’r sequence
except for mutations at -66 (G to A, L8) and -9 and -8 (G to A and T to A,
UV5). The promoter region of the 789bp wild type lag; DNA fragment has the
same sequence as shown above for the 203bp laf. In addition the lagUV5-
CAP“ promoter region sequence is the same as for laf except for the two

mutations at -9 and -8 (UV5, see above).
The promoter region for wild type gal is:

205

206
-90 so -70 so

I I I I
CATAAAAAAC GGCTAAA’I'I‘C ’l'l‘GTGTAAAC GA'I'I‘CCACTA

-50 40 30 -20
I I I
A'I'I'I‘A‘I'IICCA 'I‘G'I‘CACACTT 'I'l‘CGCATC'I'l‘ TG'I'I‘ATGCTA
4'0 If; +fl_’ P1 +10 +20
TGG’I'I‘A'I'I‘TC ATACCATAAG CCTAATGGAG CGAA‘I'I‘ATGA
+30
I
GAG'I'I‘CTGGT

For the lag-gal hybrid promoter region DNA sequences, greyed bases
represent lag DNA, and black bases denote gal DNA. The CAPl’ mutation is
C to T at -37, and the P2' mutation is A to C at -16 (indicated by the
underlined bases).

CAP 1'P2' lag-w:

-7Io -60 -5'0 -40
| I
A’i‘G’i‘(-}.:Xij}’i"i‘A {3f{'1"3‘C.r‘«..€j°“i"{3?{§"’i‘AATTCCCCATG'I'l‘ACAC'I'l‘

-30 -20 -10 +[1_* P1
'l‘TCGCATCTT TGTTCTGCTA 'l‘GG'I'I‘A’I'I'I‘C ATACCATAAG

CAP1'P2+ lag-gal:

-70 -60 -50 -4IO

.«ti'i‘ifiTGr-XG'E .T. ,‘s.’ art. ”..i‘ij': ‘ttfiT {3 .‘xT 'iz‘aA'I'I‘CCCCA'l‘G'l'I‘ACAC'I'I‘

-3|0 -20 -10 Pl_2__> +r1_, P1
'l‘TCGCA’I‘C’I'I‘ TGTTATGCTA TGG'I'l‘A’l'l‘TC ATACCATAAG

CAP1+P2‘ lag-gal:

 

A

 

207

-7|0 -60 -5|0 -4|:0
|
$4.2. 5.5.242": 1“,:th25. 4,212.7 4:”: 7EIAA'I'I‘CCCCATGTCACAC’I'I‘

-3|0 -2|0 ~1|0+r1_.P1
’l'I‘CGCATCTT ’l‘GTTCTGCTA TGG'I'I‘ATTTC ATACCATAAG

CAP1+P2+ lag-gal:

-70 -60 -5|0 410
| |
A'E‘G’i‘il.<~‘*~..t.'£’.2."i‘:’>. {32:73.1 TA( 5732 1.2“} l4ATTCCCCATGTCACAC'IT

-30 -2|0 -1|0 P2 +1

'I'I‘CGCATCTT TGTTATGCTA 'l‘GG'l'I‘A'l'I'I‘C ATACCATAAG

BIBLIOGRAPHY

 

 

10.

11.

12.

 

10.

11.

12.

BIBLIOGRAPHY

Adam, S.A., Nakagawa, T., Swanson, M.S., Woodmff, T.K., and
Dreyfuss, G. (1986)M9_1.&e11_3i91. 5, 2932-2943.

Anderson, W.A., Schneider, A.B., Emmer, M., Perlman, KL, and
Pastan, I. (1971) mm 255, 5929-5937.

Arditti, R.R., Scaife, J .G., and Beckwith, J .R. (1968) M 35,
421-426.

Arditti, R., Grodzicker, T., and Beckwith, J. (1973) m 1J1, 652-
655.

Austin, R.H., Karohl, J ., and Jovin, TM. (1983) Bimbgm, 22, 3082-
3090.

Beckwith, J ., Grodzicker, T., and Arditfi, R. (1972) 51. M91, Biol. £9,
155- 160.

Bell, A., Gaston, K., Williams, R., Chapman, K., Kolb, A., Buc, H.,
Minchin, S., Williams, J ., and Busby, S. (1990) W L8,
7243-7250.

Berg, 0., and von Hippel, P. (1988) M 2.0.0, 709-723.
Bernhard, S.L., and Meares, CF. (1986) Biochem 25, 5914-5919.

Borukhov, S., Lee, J ., and Goldfarb, A. (1991)J,_Bi91,_Qhem, 206,
23932-23935.

Bowser, C.A., and Hanna, M.M. (1991) J. Mol, Biol, 220, 227-239.

Brenowitz, M., Pickar, A., and Jamison, E. (1991) Men}, 30, 5986-
5998.

208

 

13.

14.

15.

16.

17.

18.

19.
20.
21.

22.

23.

24.

26.

27.

28.

13.

14.

15.

16.

17.

18.

19.
20.
21.

22.

23.

24.

25.

26.

27.

28.

209

Buckle, M., Geiselmann, J ., Kolb, A., and Buc, H. (1991) M515
figs, 19, 833-840.

Burgess, RR. (1969) M 24.-4., 6168-6176.

Burgess, R. R., Erickson, B, Gentry, D, Gribskov, M., Hager, D.,
Lesley, S, Strickland, M, and Thompson, N. (1986)1n RNA

2: a =. :«un Remikofi'wWS”
Burgess, R.R., Dahlberg, J. E, Gross, C. A., Record, M. T., Jr, and
Wickens, M. P. eds., pp 3-15, Elsevier, New York.

 

Burgess, R.R., Travers, A.A., Dunn, J .J ., and Bautz, EKE. (1969)
mm 221. 43-46-

Busby, S., Aiba, H., and deCrombrugghe, B. (1982) M M,
211-227.

Ceglarek, J. (1987) Master of Science Thesis, Michigan State
University.

Ceglarek, J .A., and Revzin, A. (1989) W515 1Q, 360-365.
Chait, B.T., and Kent, SB. (1992) Sam 251, 1885-1894.
Chamberlin, M.J. (1974) mm 43, 721-775.

Chambers, D.A., and Zubay, G. (1969) WM 5.3,
118-122.

Cherry, R.J., Cogoli, A., Oppliger, M., Schneider, G., and Semenza, G.
(1976) 51mm 1.5, 3653-3656.

Dickson, R.C., Abelson, J ., and Johnson, P. (1977) M 111, 65-
75.

Dombroski, A.J., Walter, W.A., Record, T.M., Jr., Siegele, D.A., and
Gross, CA. (1992) (2:11 IQ, 501-512.

Ebright, R.H., Cossart, P., Gicquel-Sanzey, B., and Beckwith, J. (1984)
New 3.1;, 232-235.

Eilen, E., Pampeno, C., and Krakow, J .S. (1978) Mm, 11, 2469-
2473.

Eschenlauer, A.C., and Reznikofl‘, W.S. (1991) m 113, 5024-5029.

29.

30.

31.
32.

33.

34.

35.

36.

37.

38.

39.

40.

41.

42.

43.

29.

30.

31.
32.

33.

34.

35.

36.

37.

38.

39.

40.

41.

42.

43.

210

Felsenfeld, G., and Hirschman, S.Z. (1965) M 13, 407-427.

Gardella, T., Moyle, H., and Susskind, M.M. (1989)J,_1\Lol_._B_i_QL 295,
579-590.

Garner, M.M., and Revzin, A. (1981) W 9, 3047-3060.
Garner, M.M., and Revzin, A. (1982) Wm. 21, 6032-6036.

Gartenberg, M.R., and Crothers, D.M. (1991)5LM91,_B';QL 219, 217-
230.

Giedroc, DR, and Colemann, J .E. (1986) m 2.5, 4969-4978.

Gitt, M.A., Wang, L-F., and Doi, RH. (1985) 1W 2% 7178-
7185.

Grachev, M.A., Lukhtanov, E.A., Mustaev, A.A., Zaychikov, E.F.,
Abdukayumov, M.N., Robinov, I.V., Richter, V.I., Skoblov, Y.S., and
Chistyakov, PG. (1989) W 180. 577-585.

Grossman, A.D., Burgess, R.R., Walter, W., and Gross, GA. (1983) Cell
32, 151-159.

Guilian, G.G., Moss, R.L., and Greaser, M. (1983) mm
129, 277-287.

Harley, CB, and Reynolds, RP. (1987) W 15, 2343-
2361.

Haugland R P in WW5
Chemicals, Larison, KD. ed. ,pp 9-13, Molecular Probes, Eugene,

Oregon.

Hawley, D.K., and McClure, W.S. (1983) W5. 11, 2237-
2255.

Hayward, R.S., Igarashi, K., and Ishihama, A. (1991) J, M91, Biol, 221,
23-29.

Helmann, J .D., and Chamberlin, M.J. (19881W 51,
839-872.

 

 

45.

46.

47.

48.

49.

50.

51.

52.
53.

54.

56.

57.

58.

59.

60.

45.

46.

47.

48.

49.

50.

51.

52.

53.

54.

55.

56.

57.

58.

59.

60.

211

Heumann, H., Lederer, H., Baer, G., May, R.P., Kjems, J .K., and
Crespi, H. (1988) sLMQLflQL 2.0.1, 115-125.

Hillenkamp, F., Karas, M., Beavis, R.C., and Chait, RT. (1991) m
Chgm, 63, 1193- 1202.

Hirschman, J ., Wang, P.-K., Sei, K., Keener, J ., and Kustu, S. (1985)
W 82, 7527-7529.

Holland, J .F., Teets, RE, and Timnick, A. (1973) mm 45, 145-
153.

Hu, J .C., and Gross, CA. (1983) W 1.9.1, 492-498.

Hudson, J .M., Crowe, LG, and Fried, MG. (1990) M 265,
3219-3225.

Igarashi, K., Fujita, N., and Ishihama, A. (1991) M 2L8, 1-6.

Igarashi, K., Hanamura, A., Makino, K., Aiba, H., Mizuno, T., Nakata,
A., and Ishihama, A. (1991b) WM 88, 8958-8962.

Igarashi, K, and Ishihama, A. (1991) £19.11 Q5, 1015- 1022.
Jacob, F., and Monod, J. (1961) M 3, 318-356.

Jacob, F., Ullman, A., and Monod, J. (1964) W 2%, 3125-
3128.

Jayaraman, P.-S., Cole, J ., and Busby, S. (1989) MM _l_7,
135-146.

Khamis, M.I., Casas-Finet, J .R., Maki, A.H., Murphy, J .B., and Chase,
J .W. (1987) mm 252, 10938- 10945.

Kim, J ., Adhya, 3., and Garges, 5- (1992) W
89, 9700-9704.

Klevan, L., and Wang, J .C. ( 1980) Mm. 1.9, 5229-5234.

Kolb, A., Busby, S., Buc, H., Garges, S., and Adhya, S. (1993) Ann.
mm in-preSS-

Kolb, A., Igarashi, K., Ishihama, A., Lavigne, M., Buckle, M., and Buc,
H. (1993) We. 2.1, 319-326-

 

 

 

61.

62.

63.

65.

66.

67.

68.

69.
70.

71.

72.

73.

74.

.75.

76.

77.

212

Kotlarz, D., Fritsch, A., and Buc, H. (1986) M 8, 799-803.

Krakow, J -S-, and Pastan, I. (1973) Wm 10,
2529-2533.

Kuhnke, G., Fritz, H-J., and Ehring, R. (1987) EMECLL 5., 507-513.

Kuhnke, G., Theres, C., Fritz, H-J., and Ehring, R. (1989) EMBLLL 8,
1247-1255.

Kumar, KB, and Chatterji, D. (1990) Bigchem. 2.9, 317-322.

 

Landick, R., Vaughn, V., Lau, E.T., van Bogelen, R., Erickson, J .W.,
and Heirnhardt, F.C. (1984)Qe11 3.8., 175-182.

 

Lee, J., and Goldfarb, A. (1991) £2ng 52. 793-798. E

Lee, J ., Kashlev, M., Borukhov, S., and Goldfarb, A. (1991) M11.
M E, 6018-6022.

Lesley, SA, and Burgess, RR. (1989) 31m 28, 7728-7734.
Li, X-M., and Krakow, J .S. (1987) W 252, 8383-8389.

Lin-Johnson, H-N., Gartenberg, M.R., and Crothers, D.M. (1986) Cell
47, 995- 1005.

Lorimer, DD, and Revzin, A. (1986) Ma. 14, 2921-2938.

Lorimer, D.D., Cao, J ., and Revzin, A. (1990) sLMQLBiQL 2L8, 275-
287.

Lowe, P.A., Hager, D.A., and Burgess, RR. (1979) 8329,8943, L8, 1344-
1352.

Kramer, H ., Niemiiller, M., Amouyal, M., Revet, B., van Wilcken-
Bergmann, B., and Muller-Hill, B. (1987) EMBQJL 8, 1481-1491.

Magasanik. B. (1961) MW 25, 248-
256.

Majors, J. (1975) flange 288, 672-674.

78.

79.

80.

81.

82.

83.

85.
86.

87.

88.

89.

90.

91.

92.
93.

94.

213

Makman, RS, and Sutherland, E.Q. (1965) mm 2249, 1309-
1314.

Malan, TR, and McClure, W.R. (1984)Q§_11 89, 173- 180.

Mandecki, W., and Caruthers, M.H. (1984) £938 8;, 263-267.
Maniatis, T, Fritsch, E. F., and Sambrook, J. (1982) mm
WM Cold Spring Habor Laboratory, Cold
Spring Habor, New York.

Martin, E., Sagitov, V., Burova, E., Nikiforov, V., and Goldfarb, A. 2::
(1992) W 281, 20175-20180. .’

McClure, W.R., Cech, C.L., and Johnston, DE. (1978) mm
288, 8941-8948.

McGhee, J .D., and Felsenfeld, G. (1980) mm 49, 1115- g
1156.

 

McKay, D.B., and Steitz, T.A. (1981) m 2.9.0., 744-749.
Monod, J. (1947) Gm 11, 223-289.

Moore, J ., Kantorow, M., Vanderzwaag, D., and McKenney, K. (1992)
Lfiagt, 172, 8030-8035.

Musso, R.E., DiLauro, R., Adhya, S., and deCrombrugghe, B. (197 7)
Qfill 12, 847 -854.

Mustaev, A., Kashlev, M., Lee, J ., Polyakov, A., Lebedev, A.,
Zalenskaya, K, Grachev, M., Goldfarb, A., and Nikiforov, V. (1991) 9L
81m 288, 23927-23931.

Narayanan, CS, and Krakow, J .S. (1982) 819911113, 21, 6103-6111.

Narayanan, CS, and Krakow, J .S. (1983) W 1.1, 2701-
2716.

Ollis, D.L., and White, SW. (1987) M 81, 981-995.
Petersen, SK, and Hansen, PG. (1991) m 118, 5200-5206.

Pinkney, M., and Hoggett, J .G. (1988) m1. 25.0, 897-902.

 

en

en

PA

PA

9"

95.

96.

97.

98.

99.

100.

101.

102.

103.

104.

105.

106.

107.

108.
109.

110.

214

Porschke, D., Hillen, W., and Takahashi, M. (1984) EMBQ J, 8, 2873-
2878.

Raibaud, 0., and Schwartz, M. (1984) Wm}. 18, 178-206.

Ramesh, U., and Meares, CF. (1989) mm
W lsz, 121-125.

Ren, Y.L., Garges, S., Adhya, S., and Krakow, J .S. (1988) Mall.
M 85, 4128-4142.

Reznikofl', W.S., and Abelson, J .N. (1980) in W Miller, J .A,
and Reznikofi‘, W.S. eds., pp 221-243, Cold Spring Habor Laboratory,
Cold Spring Habor, New York.

Reznikofl‘, W. S, Maquat, L. E., Munson, L. M., Johnson, R. C, and
Mandecki, W. (1982) m W,
Rodriguez, R. L, and Chamberlin, M ..J eds., pp 80- 95, Praeger
Publishers, New York.

Rockwell, P., and Gottesman, ME. (1991) ,LMQLBJQL 222, 189-196.

Rohrer, H., Zillig, W., and Mailhammer, R. (1975) W 88,
227-238.

Ryu, S., Kim, J ., Adhya, S., and Garges, S. (1993) MW
LISA m, 75-79.

Saxe, S.A., and Revzin, A. (1979833112111. 18, 255-263.

Schultz, S.C., Shields, G.C., and Steitz, T.A. (1990) M 218,
159-166.

Shanblatt, S.H., and Revzin, A. (1983) W 88,
1594-1598.

Shanblatt, S.H., and Revzin, A. (1986) mm 281, 10885-
10890.

Siebenlist, U., Simpson, RB, and Gilbert, W. (1980) £211 2.0, 269-281.
Siegele, D.A., Hu, J .C., and Gross, C.A. (1988) M 293, 29-37.

Siegele, D.A., Hu, J .C., Walter, W.A., and Gross, CA. (1989) sLMQL
Big], 296, 591-603.

111.

112.

113.

114.

115.

116.

117.

118.

1 19.

120.

121.

122.

123.

124.

125.

126.

127.

128.

215

Silverstone, A.E., Arditti, RR, and Magasanik, B. (1970) MIL
M 88, 773-779-

Simpson, R.B. (1979) Q82 18, 277-285.

Spassky, A., Busby, S., and Buc, H. (1984) EMBQJ. 8, 43-50.
Straney, D.C., and Crothers, D.M. (1985) QQLI 13, 449-459.

Straney, DC, and Crothers, D.M. (1987) M 1.98, 267-278.
Straney, DC, and Crothers, D.M. (198%)1M 1.93., 279-292.

Straney, D.C., Straney, SB, and Crothers, D.M. (1989) J.MQLBiQL
M, 41-57.

Swanson, M.S., Nakagawa, T.Y., LeVan, K., and Dreyfuss, G. (1987)
M81521. 1, 1731-1739.

Takahashi, M., Blazy, B., Baudras, A., and Hillen, W. ( 1983) J, M81,
L01, 187, 895-899.

Tan, G., Kelly, R, Kim, J ., and Wartell, RM. (1991) Bjmhgm, 39,
5076-5080.

Taniguchi, T, O’Neill, M., and deCrombrugghe, B. (1979) Mall.
M 16, 5090-5094-

Tfiiian, R., Losick, R., Pero, J ., and Hinnebusch, A. (1977 ) 812.41.
Bimbem. 7A., 149-154—

Travers, A.A., and Burgess, RR. (1969) Range 222, 537-540.
Ullman, A., and Monod, J. (1968) W 2, 57-60.

Waldburger, C., Gardella, T., Wong, R., and Susskind, M.M. (1990) l,
MgLfliQL 215, 267-276.

Weber, I.T., and Steitz, T.A. ( 1984) W88. 81,
3973-3977.

Weber, I.T., and Steitz, T.A. (1987) LMQLBiQL 128, 311-326.

Wehrli, D., and Staehlin, M. (1971) 8391.391. 88, 290-309.

129.

130.

131.

132.

133.

134.

135.

136.

137.

138.

139.

216

Wellman, A., and Meares, C.F. (1991)Bu2chem1_cal_and_3mhm
W 111, 140-144.

Williams, R., Bell, A., Sims, G., and Busby, S. (1991) W
29, 6705-6712.

Wu, C-W., Yarbrough, L.R., Hillel, Z., and Wu, F.Y-H. (1975) hog.
W 12, 3019-3023-

Wu, F.Y-H., Huang, W—J., Sinclair, RB, and Powers, L. (1992) J, Bigl,
Chgm, 281, 25560-25567.

Wu, F.Y—H., Nath, K, and Wu, C-W. (1974) Mm. 13., 2567-2572.

Wu, F.Y-H., Yarbrough, LR, and Wu, C-W. (1976) W _18, 3254-
3258.

Wu, H-M., and Crothers, D.M. (1984) Nam 398, 509-513.

Xiong, X, delaCruz, N ., and Reznikofl‘, W.S. (1991) M 118, 4570-
4577.

Zillig, W., Palm, P., and Heil, A. (1976) in W, Losick, R.,
and Chamberlin, M. eds., pp 101-125, Cold Spring Harbor Laboratory,
Cold Spring Harbor, New York.

Zinkel, SS, and Crothers, D.M. (1991) M £9, 201-215.

Zou, C., Fujita, N ., Igarashi, K., and Ishihama, A. (1992) Mglgmflm:
58512121521. 8, 2599-2605-

"Im:212122121122:ES