Voet D., Voet Ju.G. Biochemistry

Подождите немного. Документ загружается.

to stall. How, then, do RNAPs avoid accumulating at dam-

aged or mispaired sites, which, if it occurred on an essential

gene, would be lethal?

RNAPs do not monotonically move forward along the

template DNA. Instead, they frequently backtrack such

that the RNA’s penultimate nucleotide, which was in the

i ⫺ 1 position, has re-entered the i ⫹ 1 position and the 3¿-

nucleotide, now in the i ⫹ 2 position, enters the secondary

channel where it binds in the so-called P (for proofreading)

site. If the forward movement of the RNA is impeded by

damage to the template or by mispairing, further back-

tracking becomes favored so that several more ribonu-

cleotides enter the secondary channel.The backtracking of

only one or a few nucleotides is reversible. Otherwise,tran-

scription is arrested until the RNA is hydrolytically cleaved

at the active site. In E. coli, this requires the assistance of

the homologous proteins GreA and/or GreB, whereas with

RNAP II, this function is carried out by the unrelated

protein TFIIS. These proteins induce the RNAP active site

to hydrolyze the phosphodiester bond between the ribonu-

cleotides in the i ⫹ 1 and i ⫺ 1 positions (a reaction that is

not the reverse of the polymerase reaction since this would

be pyrophosphorolysis). In this way, RNAP can correct its

mistakes and resume RNA synthesis. RNAP I and RNAP

III also efficiently correct their mistakes.

Kornberg determined the X-ray structure of RNAP II

in complex with TFIIS, a 28-nt template DNA, a 14-nt non-

template DNA, and a 13-nt RNA that is complementary to

5¿ end of the template DNA except for the last two residues

at the RNA’s 3¿ end, which are mismatched (Fig. 31-25).

The RNA–DNA hybrid has backtracked such that these

latter residues occupy the i ⫹ 1 and i ⫹ 2 positions. The C-

terminal domain of TFIIS is bound in the RNAP’s funnel

with one of its loops insinuated through the pore to inter-

act with the RNAP’s active site residues. There it presum-

Section 31-2. RNA Polymerase 1281

Figure 31-24 The A and E sites and the trigger loop in RNA

polymerase II. A cutaway view of the transcribing complex

viewed as in Fig. 31-22b. Its bound nucleic acids and nucleotides

are colored differently with template DNA cyan, nontemplate

DNA green, newly transcribed RNA red, GTP in the A site

Figure 31-25 X-ray structure of backtracked RNA polymerase

II in complex with DNA, RNA, and TFIIS. The RNAP, which is

represented by its semitransparent surface diagram, is viewed as

in Figs. 31-22b and 31-24 with its subunits colored as in Fig. 31-

21a.The DNA and RNA are shown in ladder form with the RNA

(13 nt) red, the template DNA (28 nt) green, and the nontem-

plate DNA (14 nt) blue.The base pairs at the i ⫹ 1 and

i ⫹ 2 positions of the RNA ⴢ DNA hybrid helix are mismatched.

An Mg

2⫹

ion, which is represented by a red sphere near the i ⫹ 1

position, marks the RNAP’s active site.TFIIS is drawn in cartoon

form in magenta. Its C-terminal domain is inserted into the

RNAP’s funnel (Fig. 31-22c) with a loop occupying the pore,

where it is in proximity to the RNAP’s active site. [Based on an

X-ray structure by Roger Kornberg, Stanford University. PDBid

3GTM.]

orange, and ATP in the E site blue.The trigger loop is magenta,

the bridge helix is green, and the two Mg

2⫹

ions at the active site

are represented by magenta spheres.The RNAP II surface is

gray. [Courtesy of Dong Wang and Roger Kornberg, Stanford

University. PDBid 2E2H.]

JWCL281_c31_1260-1337.qxd 8/11/10 9:48 PM Page 1281

ably facilitates the hydrolytic reaction, perhaps by liganding

the active site Mg

2⫹

ion that normally accompanies the in-

coming NTP (Section 31-2Ea) and/or positioning a hy-

drolytic water molecule.The trigger loop in this structure is

in the “open” conformation. Interestingly, the cryo-

EM–based structure of E. coli RNAP in complex with

GreB indicates that GreB likewise inserts an extended

protein finger in the RNAPs active site via its secondary

channel, even though the structures of GreB and TFIIS are

unrelated.

Despite the foregoing,transcription is less accurate than

DNA replication: RNAPs incorporate one incorrect base

for every ⬃10

transcribed, whereas, for example, E. coli

Pol I incorporates one incorrect base in ⬃10

(Section 30-

2Ab). Cells can tolerate the former rate because most

genes are repeatedly transcribed. In contrast, errors in

DNA synthesis alter all the affected gene’s transcripts in

the cell in which the error occurred and all of its progeny.

d. Mammalian RNA Polymerase I Has

a Bipartite Promoter

Since, as we shall see in Section 31-4B, the numerous

rRNA genes in a given eukaryotic cell have essentially

identical sequences, its RNAP I only recognizes one pro-

moter. Yet, in contrast to the case for RNAPs II and III,

RNAP I promoters are species specific, that is, an RNAP I

only recognizes its own promoter and those of closely re-

lated species. This is because only closely related species

exhibit recognizable sequence identities near the transcrip-

tional start sites of their rRNA genes. RNAP I promoters

were therefore identified by determining how the tran-

scription rate of an rRNA gene is affected by a series of in-

creasingly longer deletions approaching its start site from

either its upstream or its downstream sides. Such studies

have indicated, for example, that mammalian RNAPs I re-

quire the presence of a so-called core promoter element,

which spans positions ⫺31 to ⫹6 and hence overlaps the

transcribed region. However, efficient transcription addi-

tionally requires an upstream promoter element, which is

located between residues ⫺187 and ⫺107. These elements,

which are G ⫹ C-rich and ⬃85% identical, are bound by

specific transcription factors which then recruit RNAP I to

the transcription start site.

e. RNA Polymerase II Promoters Are

Complex and Diverse

The promoters recognized by RNAP II are considerably

longer and more diverse than those of prokaryotic genes

but have not yet been fully described. The structural genes

expressed in all tissues, the so-called housekeeping genes,

which are thought to be constitutively transcribed, have

one or more copies of the sequence GGGCGG or its com-

plement (the GC box) located upstream from their tran-

scription start sites. The analysis of deletion and point mu-

tations in eukaryotic viruses such as SV40 indicates that

GC boxes function analogously to prokaryotic promoters.

On the other hand, structural genes that are selectively ex-

pressed in one or a few types of cells often lack these GC-

rich sequences. Rather, many contain a conserved AT-rich

sequence located 25 to 30 bp upstream from their transcrip-

tion start sites (Fig. 31-26). Note that this so-called TATA

box resembles the ⫺10 region of prokaryotic promoters

(TATAAT), although they differ in their locations relative

to the transcription start site (⫺27 vs ⫺10). The functions

of these two promoter elements are not strictly analogous,

however, since the deletion of the TATA box does not nec-

essarily eliminate transcription. Rather, TATA box dele-

tion or mutation generates heterogeneities in the transcrip-

tional start site, thereby indicating that the TATA box

participates in selecting this site.

The gene region extending between about ⫺50 and ⫺110

also contains promoter elements. For instance,many eukary-

otic structural genes, including those encoding the various

globins, have a conserved sequence of consensus CCAAT

(the CCAAT box) located between about ⫺70 and ⫺90

whose alteration greatly reduces the gene’s transcription

rate. Globin genes have, in addition, a conserved CACCC

box upstream from the CCAAT box that has also been im-

plicated in transcriptional initiation. Evidently, the pro-

moter sequences upstream of the TATA box form the initial

DNA-binding sites for RNA polymerase II and the other

proteins involved in transcriptional initiation (see below).

f. Enhancers Are Transcriptional Activators That

Can Have Variable Positions and Orientations

Perhaps the most surprising aspect of eukaryotic tran-

scriptional control elements is that some of them need not

1282 Chapter 31. Transcription

Figure 31-26 The promoter sequences of

selected eukaryotic structural genes. The

homologous segment, the TATA box, is

shaded in red with the base at position ⫺27

underlined and the initial nucleotide to be

transcribed (⫹1) shaded in green.The

bottom row indicates the consensus

sequence of several such promoters with

the subscripts indicating the percent

occurrence of the corresponding base.

[After Gannon, F., et al., Nature 278, 433

(1978).]

Chicken

ovalbumin

Adenovirus

late

Rabbit

β globin

Mouse β

globin major

T AT

JWCL281_c31_1260-1337.qxd 8/11/10 9:48 PM Page 1282

Inactive

chromosome

segment

Active

chromosome

segment

RNA polymerase

Ribosome

mRNA

Direction of

protein

synthesis

(translation)

direction of

RNA synthesis

(transcription)

have fixed positions and orientations relative to their corre-

sponding transcribed sequences. For example, the SV40

genome, in which such elements were first discovered, con-

tains two repeated sequences of 72 bp each that are located

upstream from the promoter for early gene expression.

Transcription is unaffected if one of these repeats is

deleted but is nearly eliminated when both are absent.The

analysis of a series of SV40 mutants containing only one of

these repeats demonstrated that its ability to stimulate

transcription from its corresponding promoter is all but in-

dependent of its position and orientation. Indeed, transcrip-

tion is unimpaired when this segment is several thousand

base pairs upstream or downstream from the transcription

start site. Gene segments with such properties are named

enhancers to indicate that they differ from promoters, with

which they must be associated in order to trigger site-

specific and strand-specific transcription initiation (al-

though the characterization of numerous promoters and

enhancers indicates that their functional properties are

similar). Enhancers occur in both eukaryotic viruses and

cellular genes.

Enhancers are required for the full activities of their cog-

nate promoters. It was originally thought that enhancers

somehow acted as entry points on DNA for RNAP II (per-

haps by altering DNA’s local conformation or through a

lack of binding affinity for the histones that normally coat

eukaryotic DNA; Section 34-1B). However, it is now clear

that enhancers are recognized by specific transcription fac-

tors that stimulate RNA polymerase II to bind to the corre-

sponding but distant promoter. This requires that the DNA

between the enhancer and promoter loop around so that

the transcription factor can simultaneously contact the en-

hancer and the RNAP II and/or its associated proteins at

the promoter. Most cellular enhancers are associated with

genes that are selectively expressed in specific tissues. It

therefore seems, as we discuss in Section 34-3B, that

enhancers mediate much of the selective gene expression in

eukaryotes.

g. RNA Polymerase III Promoters Can Be Located

Downstream from Their Transcription Start Sites

The promoters of genes transcribed by RNAP III can be

located entirely within the genes’ transcribed regions. Don-

ald Brown established this through the construction of a

series of deletion mutants of a Xenopus borealis 5S RNA

gene. Deletions of base sequences that start from outside

one or the other end of the transcribed portion of the 5S

gene only prevent transcription if they extend into the seg-

ment between nucleotides ⫹40 and ⫹80. Indeed, a frag-

ment of the 5S RNA gene consisting of only nucleotides 41

to 87, when cloned in a bacterial plasmid, is sufficient to di-

rect specific initiation by RNAP III at an upstream site.

This is because, as was subsequently demonstrated, the se-

quence contains the binding site for transcription factors

that stimulate the upstream binding of RNAP III. Further

studies have shown, however, that the promoters of other

RNAP III–transcribed genes lie entirely upstream of their

start sites. These upstream sites also bind transcription fac-

tors that recruit RNAP III.

3 CONTROL OF TRANSCRIPTION

IN PROKARYOTES

Prokaryotes respond to sudden environmental changes,

such as the influx of nutrients, by inducing the synthesis of

the appropriate proteins. This process takes only minutes

because transcription and translation in prokaryotes are

closely coupled: Ribosomes commence translation near the

5¿ end of a nascent mRNA soon after it is extruded from

RNA polymerase (Fig. 31-27). Moreover, most prokaryotic

mRNAs are enzymatically degraded within 1 to 3 min of

their synthesis, thereby eliminating the wasteful synthesis

of unneeded proteins after a change in conditions (protein

degradation is discussed in Section 32-6). In fact, the 5¿

ends of some mRNAs are degraded before their 3¿ ends

have been synthesized.

In contrast, the induction of new proteins in eukaryotic

cells frequently takes hours or days, in part because tran-

scription takes place in the nucleus and the resulting

mRNAs must be transported to the cytoplasm, where

translation occurs. However, eukaryotic cells, particularly

those of multicellular organisms, have relatively stable en-

vironments; major changes in their transcription patterns

usually occur only during cell differentiation.

Section 31-3. Control of Transcription in Prokaryotes 1283

Figure 31-27 An electron micrograph and its interpretive

drawing showing the simultaneous transcription and translation

of an E. coli gene. RNA polymerase molecules are transcribing

the DNA from right to left while ribosomes are translating the

nascent RNAs (mostly from bottom to top). [Courtesy of Oscar

L. Miller, Jr. and Barbara Hamkalo, University of Virginia.]

JWCL281_c31_1260-1337.qxd 8/11/10 9:48 PM Page 1283

In this section we examine some of the ways in which

prokaryotic gene expression is regulated through tran-

scriptional control. Eukaryotes, being vastly more complex

creatures than are prokaryotes, have a correspondingly

more complicated transcriptional control system whose

general outlines are beginning to come into focus. We

therefore defer discussion of eukaryotic transcriptional

control until Section 34-3B, where it can be considered in

light of what we know about the structure and organization

of the eukaryotic chromosome.

A. Promoters

In the presence of high concentrations of inducer, the lac

operon (Section 31-1Ab) is rapidly transcribed. In contrast,

the lacI gene is transcribed at such a low rate that a typical

E. coli cell contains ⬍10 molecules of the lac repressor.Yet,

the I gene has no repressor. Rather, it has such an ineffi-

cient promoter (Fig. 31-10) that it is transcribed an average

of about once per bacterial generation. Genes that are tran-

scribed at high rates have efficient promoters. In general, the

more efficient a promoter, the more closely its sequence

resembles that of the corresponding consensus sequence.

a. Gene Expression Can Be Controlled by

a Succession of ␴ Factors

The processes of development and differentiation involve

the temporally ordered expression of sets of genes according

to genetically specified programs. Phage infections are

among the simplest examples of developmental processes.

Typically, only a subset of the phage genome, often referred

to as early genes, are expressed in the host immediately af-

ter phage infection.As time passes, middle genes start to be

expressed, and the early genes as well as the bacterial genes

are turned off. In the final stages of phage infection, the

middle genes give way to the late genes. Of course some

phage types express more than three sets of genes and

some genes may be expressed in more than one stage of an

infection.

One way in which families of genes are sequentially ex-

pressed is through “cascades” of ␴ factors. In the infection

of Bacillus subtilis by bacteriophage SP01, for example, the

early gene promoters are recognized by the bacterial

RNAP holoenzyme. Among these early genes is gene 28,

whose gene product is a new ␴ subunit, designated ␴

gp28

that displaces the bacterial ␴ subunit from the core en-

zyme. The reconstituted holoenzyme recognizes only the

phage middle gene promoters, which all have similar ⫺35

and ⫺10 regions but bear little resemblance to the corre-

sponding regions of bacterial and phage early genes. The

early genes therefore become inactive once their corre-

sponding mRNAs have been degraded. The phage middle

genes include genes 33 and 34, which together specify yet

another ␴ factor, ␴

gp33/34

, which, in turn, permits the tran-

scription of only late phage genes.

Most bacteria, including E. coli and B. subtilis, likewise

have several different ␴ factors (E. coli has seven). These

are not necessarily utilized in a sequential manner. Rather,

those that differ from the predominant or primary ␴ factor

(␴

in E. coli) control the transcription of coordinately ex-

pressed groups of special purpose genes, whose promoters

are quite different from those recognized by the primary ␴

factor. For example, in E. coli, the alternative ␴ factor ␴

the master regulator of the heat shock response (Section

9–2C), whereas ␴

directs the expression of proteins in-

volved in nitrogen assimilation. Likewise, sporulation in B.

subtilis, a process in which the bacterial cell is asymmetri-

cally partitioned into two compartments, the forespore

(which becomes the spore, a germline cell from which sub-

sequent progeny arise) and the mother cell (which synthe-

sizes the spore’s protective cell wall and is eventually dis-

carded), is governed by five ␴ factors in addition to that of

the vegetative (nonsporulating) cell: one that is active be-

fore cell partition occurs, two that are sequentially active in

the forespore, and two that are sequentially active in the

mother cell. Cross-regulation of the compartmentalized ␴

factors permits the forespore and mother cell to tightly co-

ordinate this differentiation process.

B. lac Repressor I: Binding

In 1966, Benno Müller-Hill and Walter Gilbert isolated lac

repressor on the basis of its ability to bind

C-labeled

IPTG (Section 31-1Aa) and demonstrated that it is a pro-

tein. This was an exceedingly difficult task because lac

repressor comprises only ⬃0.002% of the protein in wild-

type E. coli. Now, however, lac repressor is available in

quantity via molecular cloning techniques (Section 5-5G).

a. lac Repressor Finds Its Operator by

Sliding Along DNA

The lac repressor is a tetramer of identical 360-residue

subunits, each of which binds one IPTG molecule with a

dissociation constant of K ⫽ 10

⫺6

M. In the absence of in-

ducer, the repressor tetramer nonspecifically binds duplex

DNA with a dissociation constant of K ⬇ 10

⫺4

M. How-

ever, it specifically binds to the lac operator with far

greater affinity: K

⬇ 10

⫺13

M. Limited proteolysis of lac

repressor with trypsin reveals that each subunit consists of

two functional domains: Its 58-residue N-terminal peptide

binds DNA but not IPTG, whereas the remaining “core

tetramer” binds only IPTG.

The observed rate constant for the binding of lac repres-

sor to lac operator is k

⬇ 10

⫺1

.This “on” rate is much

greater than that calculated for the diffusion-controlled process

in solution: k

⬇ 10

⫺1

for molecules the size of lac re-

pressor. Since it is impossible for a reaction to proceed faster

than its diffusion-controlled rate, the lac repressor must not

encounter operator from solution in a random three-dimen-

sional search. Rather, it appears that lac repressor finds oper-

ator by nonspecifically binding to DNA and diffusing along it

in a far more efficient one-dimensional search.

b. lac Operator Has a Nearly Palindromic Sequence

The availability of large quantities of lac repressor made

it possible to characterize the lac operator. E. coli DNA

1284 Chapter 31. Transcription

JWCL281_c31_1260-1337.qxd 8/11/10 9:48 PM Page 1284

DNA

sequence

CAP-cAMP

binding site

Operator

Z gene

+30+20+10+1–10–20–30–40–50–60–70–80

Promoter

mRNA

fMet

Thr

Met

GAAAGCGGGCAGTGACCGCAACGCAATTAATGTGAGTTAGCTCACTCATTAGGCACCCCAGGCTTTACACTTTATGCTTCCGGCTCGTATGTTGTGTGGAATTGTGAGCGGATAACAATTTCACACAGGAAACAGCTATGACCATG

CTTTCGCCCGTCACTGGCGTTGCGTTAATTACACTCAATCGAGTGAGTAATCCGTGGGGTCCGAAATGTGAAATACGAAGGCCGAGCATACAACACACCTTAACACTCGCCTATTGTTAAAGTGTGTCCT TTGTCGATACTGGTAC

I gene

Glu

Gln

Stop

Ser

Gly

that had been sonicated to small fragments was mixed with

lac repressor and passed through a nitrocellulose filter.

Protein, with or without bound DNA, sticks to nitrocellu-

lose, whereas duplex DNA, by itself, does not. The DNA

was released from the filter-bound protein by washing it

with IPTG solution, recombined with lac repressor, and the

resulting complex treated with DNase I. The DNA frag-

ment that lac repressor protects from nuclease degradation

consists of a run of 26 bp that is embedded in a nearly 2-

fold symmetric sequence of 35 bp (Fig. 31-28, top). Such

palindromic symmetry is a common feature of DNA seg-

ments that are specifically bound by proteins (recall, for

example, that restriction endonuclease recognition sites

are also palindromic; Section 5-5Aa).

Palindromic DNA sequences, as we have seen, bind to

proteins that have matching 2-fold symmetry. However,

methylation protection experiments on the lac repressor–

operator system do not fully support this model: There is

an asymmetric pattern of differences between free and

repressor-bound operator in the susceptibility of its bases

to reaction with DMS (Fig. 31-28). Furthermore, point mu-

tations that render it operator constitutive (O

), and that

invariably weaken the binding of repressor to operator,

may increase as well as decrease the operator’s 2-fold

symmetry (Fig. 31-28).

c. lac Repressor Prevents RNA Polymerase from

Forming a Productive Initiation Complex

Operator occupies positions ⫺7 through ⫹28 of the lac

operon relative to the transcription start site (Fig. 31-29).

Nuclease protection studies, it will be recalled, indicate

that, in the initiation complex, RNA polymerase tightly

binds to the DNA between positions ⫺20 and ⫹20 (Sec-

tion 31-2Aa).Thus, the lac operator and promoter sites over-

lap. It was therefore widely assumed for many years that

lac repressor simply physically obstructs the binding of

RNA polymerase to the lac promoter. However, the obser-

vation that lac repressor and RNA polymerase can simul-

taneously bind to the lac operon indicates that lac repres-

sor must act by somehow interfering with the initiation

process. Closer investigation of this phenomenon revealed

that, in the presence of bound lac repressor, RNA poly-

merase holoenzyme still abortively synthesizes oligonu-

cleotides, although they tend to be shorter than those made

in the absence of repressor. Evidently, lac repressor acts by

somehow increasing the already high kinetic barrier for

RNA polymerase to generate the open complex and com-

mence processive elongation.

We discuss the lac repressor structure and further as-

pects of lac operator organization in Section 31-3F.

C. Catabolite Repression: An Example

of Gene Activation

Glucose is E. coli’s metabolite of choice; the availability of

adequate amounts of glucose prevents the full expression of

⬎100 genes that encode proteins involved in the fermentation

Section 31-3. Control of Transcription in Prokaryotes 1285

Figure 31-28 The base sequence of the lac operator. The

symmetry related regions (red) comprise 28 of its 35 bp. A “⫹”

denotes positions at which repressor binding enhances

methylation by dimethyl sulfate (which methylates G at N7 and

A at N3) and a “⫺” indicates where this footprinting reaction is

inhibited.The bottom row indicates the positions and identities

of different point mutations that prevent lac repressor binding

mutants).Those in red increase the operator’s symmetry.

[After Sobell, H.M., in Goldberger, R.F. (Ed.), Biological

Regulation and Development, Vol. 1, p. 193, Plenum Press

(1979).]

Figure 31-29 The nucleotide sequence of the E. coli lac

promoter–operator region. The region extends from the

C-terminal portion of lacI (left) to the N-terminal portion of

lacZ (right).The palindromic sequences of the operator and

Protected by lac repressor

mutations

AATTTTGC

TTAAAACG

5' TG G G AA A AAAAA A A3'AG G GGGGCCCCCTT T T T TTTT

3' AC C C TT T TTTTT T T5'TC C CCCCGGGGGAA A A A AAAA

+–– –

++ ––––

the CAP-binding site (Section 31-3C) are overscored or

underscored. [After Dickson, R.C.,Abelson, J., Barnes, W.M.,

and Reznikoff, W.A., Science 187, 32 (1975).]

JWCL281_c31_1260-1337.qxd 8/11/10 9:48 PM Page 1285

of numerous other catabolites, including lactose (Fig. 31-30),

arabinose, and galactose, even when these metabolites are

present in high concentrations. This phenomenon, which is

known as catabolite repression, prevents the wasteful dupli-

cation of energy-producing enzyme systems.

a. cAMP Signals the Lack of Glucose

The first indication of the mechanism of catabolite re-

pression was the observation that, in E. coli, the level of

cAMP, which was known to be a second messenger in ani-

mal cells (Section 18-3Cb), is greatly diminished in the

presence of glucose.This observation led to the finding that

the addition of cAMP to E. coli cultures overcomes catabo-

lite repression by glucose. Recall that, in E. coli, adenylate

cyclase is activated by the phosphorylated enzyme EIIA

glc

(or possibly inactivated by dephospho-EIIA

glc

), which is

dephosphorylated on the transport of glucose across the

cell membrane (Section 20-3D). The presence of glucose,

therefore, normally lowers the cAMP level in E. coli.

b. CAP–cAMP Complex Stimulates the Transcription

of Catabolite Repressed Operons

Certain E. coli mutants, in which the absence of glucose

does not relieve catabolite repression, are missing a cAMP-

binding protein that is synonymously named catabolite

gene activator protein (CAP) and cAMP receptor protein

(CRP). CAP is a homodimer of 209-residue subunits that

undergoes a large conformational change on binding

cAMP. Its function was elucidated by Ira Pastan, who

showed that CAP–cAMP complex, but not CAP itself,

binds to the lac operon (among others) and stimulates tran-

scription from its otherwise low-efficiency promoter in the

absence of lac repressor. CAP is therefore a positive regula-

tor (turns on transcription), in contrast to lac repressor,

which is a negative regulator (turns off transcription).

The X-ray structure, by Thomas Steitz, of CAP–cAMP

in complex with a palindromic 30-bp segment of duplex

DNA whose sequence resembles that of the CAP binding

sequence (Fig. 31-29) reveals that the DNA is bent by ⬃90°

around the protein (Fig. 31-31a). The bend arises from two

⬃45° kinks in the DNA between the fifth and sixth bases

out from the complex’s 2-fold axis in both directions. This

distortion results in the closing of the major groove and an

enormous widening of the minor groove at each kink.

Why is the CAP–cAMP complex necessary to stimulate

the transcription of its target operons? And how does it do

so? The lac operon has a weak (low efficiency) promoter;

its ⫺10 and ⫺35 sequences (TATGTT and TTTACA; Fig.

31-10) differ significantly from the corresponding consen-

sus sequences of strong (high-efficiency) promoters

(TATAAT and TTGACA; Fig. 31-10). Such weak promot-

ers evidently require some sort of help for efficient tran-

scriptional initiation.

Richard Ebright has shown that CAP interacts directly

with RNAP via the C-terminal domain of its 85-residue ␣

subunit (␣CTD) in a way that stimulates RNAP to initiate

transcription from a nearby promoter. The ␣CTD also

binds dsDNA nonspecifically but does so with higher affin-

ity at A ⫹ T–rich sites such as those of UP elements (Sec-

tion 31-2Aa). It is flexibly linked to the rest of the ␣ subunit

and hence is not seen in the X-ray structure of Tth RNAP

(Fig. 31-11) due to disorder.

Three classes of the over one hundred CAP-dependent

promoters have been characterized:

1. Class I promoters, such as that of the lac operon, re-

quire only CAP–cAMP for transcriptional activation. The

CAP binding site on the DNA can be located at various

distances from the promoter provided that CAP and

RNAP bind to the same face of the DNA helix. Thus,

CAP–cAMP activates the transcription of the lac operon if

its DNA binding site is centered near positions ⫺62 (its

wild-type position; Fig. 31-29), ⫺72, ⫺83, ⫺93, or ⫺103, all

of which are one helical turn apart. For the latter sites, this

requires that the DNA loop around to permit CAP–cAMP

to contact the ␣CTD. Such looping is likely to be facilitated

by the bending of the DNA around CAP–cAMP.

2. Class II promoters also require only CAP–cAMP for

transcriptional activation. However, in class II promoters,

the CAP binding site only occupies a fixed position that

overlaps the RNAP binding site, apparently by replacing

the promoter’s ⫺35 promoter region. CAP then interacts

1286 Chapter 31. Transcription

Minutes after IPTG addition

lac mRNA

0246810

Glucose

Figure 31-30 The kinetics of lac operon mRNA synthesis

following its induction with IPTG, and of its degradation after

glucose addition. E. coli were grown on a medium containing

glycerol as their only carbon-energy source and

H-labeled

uridine. IPTG was added to the medium at the beginning of the

experiment to induce the synthesis of the lac enzymes.After

3 min, glucose was added to stop the synthesis. The amount of

H-labeled lac RNA was determined by hybridization with DNA

containing the lacZ and lacY genes. [After Adesnik, M. and

Levinthal, C., Cold Spring Harbor Symp. Quant. Biol. 35, 457

(1970).]

JWCL281_c31_1260-1337.qxd 8/11/10 9:48 PM Page 1286

with RNAP via interactions with both the ␣CTD and the ␣

subunit’s N-terminal domain.

3. Class III promoters require multiple activators to

maximally stimulate transcription. These may be two or

more CAP–cAMP complexes or a CAP–cAMP complex

acting in concert with promoter-specific activators as oc-

curs in the araBAD operon (Section 31-3E).

The X-ray structure of CAP–cAMP in complex with the

E. coli ␣CTD and a 44-bp palindromic DNA containing the

22-bp CAP–cAMP binding site and 5¿-AAAAAA-3¿ at each

end, determined by Helen Berman and Ebright, reveals how

these components interact (Fig.31-31b).The 2-fold symmet-

ric CAP–cAMP–␣CTD complex contains two differently lo-

cated pairs of ␣CTDs. Each member of the pair designated

␣CTD

CAP,DNA

binds to both CAP and to the DNA. CAP and

␣CTD

CAP,DNA

interact over a surprisingly small surface area

involving only six residues on each protein that mutagenesis

experiments had previously implicated. ␣CTD

CAP,DNA

also

interacts with the minor groove of a 6-bp segment of the

DNA (5¿-AAAAAG-3¿) centered 19 bp from the center of

the DNA. Each member of the other pair of ␣CTDs, desig-

nated ␣CTD

DNA

, interacts with the minor groove of an UP

element-like sequence (5¿-GAAAAA-3¿) that is fortu-

itously present in the DNA but it makes no contacts with

other protein molecules. The common portions of the two

CAP complexes pictured in Fig. 31-31a,b are closely super-

imposable, thereby indicating that the conformation of CAP

and its interaction with DNA are not significantly altered by

its association with the ␣CTD. Evidently, CAP–cAMP tran-

scriptionally activates RNAP via a simple “adhesive” mech-

anism that facilitates and/or stabilizes its interaction with the

Section 31-3. Control of Transcription in Prokaryotes 1287

Figure 31-31 X-ray structures of CAP–cAMP–dsDNA

complexes. The dsDNA and cAMP in these 2-fold symmetric

complexes are colored according to atom type with DNA C

white, cAMP green, N blue, O red, and P orange. (a) CAP–cAMP

in complex with a palindromic 30-bp self-complementary DNA

viewed with its 2-fold axis horizontal.The protein is drawn in

ribbon form with its identical subunits pink and blue and with

their C-terminal domains in darker shades.The DNA is shown in

stick form with successive P atoms in the same strand connected

by orange rods and with the cAMP drawn in space-filling form.

(b) CAP–cAMP in complex with a 44-bp palindromic DNA and

four ␣CTD subunits.The DNA, CAP, and cAMP are viewed as in

Part a with the DNA drawn in space-filling form. The ␣CTD

subunits are drawn in ribbon form with the ␣CTD

CAP,DNA

green

and the ␣CTD

DNA

dark green. (c) The same structure as in Part a

showing the binding of the CAP dimer’s two helix–turn–helix

(HTH) motifs in successive major grooves of the DNA. The view

is rotated 45° about the vertical axis relative to Part a. Note how

CAP’s F (recognition) helix is inserted into the DNA’s major

groove, as can also be seen in Parts a and b. [Parts a and c based

on an X-ray structure by Thomas Steitz, Yale University. PDBid

1CGP. Part b based on an X-ray structure by Helen Berman and

Richard Ebright, Rutgers University. PDBid 1LB2.]

See Interactive Exercise 38

(a)

(b)

(c)

JWCL281_c31_1260-1337.qxd 10/19/10 10:37 AM Page 1287

promoter DNA.The structures of ␣CTD

CAP,DNA

and ␣CTD-

CAP

and their interactions with DNA are nearly identical,

thereby suggesting that they are representative of the inter-

action of an ␣CTD with an UP element.

D. Sequence-Specific Protein–DNA Interactions

Since genetic expression is controlled by proteins such as

CAP and lac repressor, an important issue in the study of

gene regulation is how these proteins recognize their target

base sequences on DNA. Sequence-specific DNA-binding

proteins generally do not disrupt the base pairs of the

duplex DNA to which they bind. Consequently, these

proteins can only discriminate among the four base pairs

(A ⴢ T, T ⴢ A, G ⴢ C, and C ⴢ G) according to the functional

groups of these base pairs that project into DNA’s major

and minor grooves. An inspection of Fig. 5-12 reveals that

the groups exposed in the major groove have a greater

variation in their types and arrangements than do those

that are exposed in the minor groove. Indeed, the positions

of the hydrogen bonding acceptors in the major groove

vary with both the identity and orientation of the base pair,

whereas in the minor groove they are largely sequence

independent. Moreover, the ⬃5-Å-wide and ⬃8-Å-deep

minor groove of canonical (ideal) B-DNA is too narrow to

admit protein structural elements such as an ␣ helix,

whereas its ⬃12-Å-wide and ⬃8-Å-deep major groove can

do so. Thus, in the absence of major conformational

changes to B-DNA, it would be expected that proteins

could more readily differentiate base sequences from its

major groove than from its minor groove. We shall see

below that this is, in fact, the case.

a. The Helix–Turn–Helix Motif Is a Common DNA

Recognition Element in Prokaryotes

See Guided Exploration 30: Transcription factor–DNA interactions

The CAP dimer’s two symmetrically disposed F helices

protrude from the protein surface in such a way that they

fit into successive major grooves of B-DNA (Fig. 31-31).

CAP’s E and F helices form a helix–turn–helix (HTH)

motif (supersecondary structure) that conformationally re-

sembles analogous HTH motifs in numerous other prokary-

otic repressors of known X-ray and NMR structure, includ-

ing the lac repressor, the E. coli trp repressor (Section

31-3G), and the cI repressors and Cro proteins from bac-

teriophages ␭ and 434 (Section 33-3D). HTH motifs are

⬃20-residue polypeptide segments that form two ␣ helices

which cross at ⬃120° (Fig. 31-31c). They occur as compo-

nents of domains that otherwise have widely varying struc-

tures, although all of them bind DNA. Note that HTH mo-

tifs are structurally stable only when they are components

of larger proteins.

The X-ray and NMR structures of a number of

protein–DNA complexes (see below) indicate that DNA-

binding proteins containing an HTH motif associate with

their target base pairs mainly via the side chains extending

from the second helix of the HTH motif, the so-called recog-

nition helix (helix F in CAP, E in trp repressor, and ␣3 in

the phage proteins). Indeed, replacing the outward-facing

residues of the 434 repressor’s recognition helix with the

corresponding residues of the related bacteriophage P22

yields a hybrid repressor that binds to P22 operators but

not to those of 434. Moreover, the HTH motifs in all these

proteins have amino acid sequences that are similar to each

other and to polypeptide segments in numerous other

prokaryotic DNA-binding proteins, including lac repres-

sor. Evidently, these proteins are evolutionarily related and

bind their target DNAs in a similar manner.

How does the recognition helix recognize its target se-

quence? Since each base pair presents a different and pre-

sumably readily differentiated constellation of hydrogen

bonding groups in DNA’s major groove, it seemed likely

that there would be a simple correspondence, analogous to

Watson–Crick base pairing, between the amino acid

residues of the recognition helix and the bases they contact

in forming sequence-specific associations.The above X-ray

structures, however, indicate this idea to be incorrect.

Rather, base sequence recognition arises from complex

structural interactions. For instance:

1. The X-ray structures of the 48% identical N-termi-

nal domain of 434 repressor (residues 1–69) and the en-

tire 71-residue 434 Cro protein in their complexes with

the identical 20-bp target DNA (the expression of phage

434 is regulated through the differential binding of these

proteins to the same DNA segments; Section 33-3Db)

were both determined by Stephen Harrison. Both ho-

modimeric proteins, as seen for CAP (Fig. 31-31), associ-

ate with the DNA in a 2-fold symmetric manner with their

recognition helices bound in successive turns of the

DNA’s major groove (Figs. 31-32 and 31-33). In both com-

plexes, the protein closely conforms to the DNA surface

and interacts with its paired bases and sugar–phosphate

chains through elaborate networks of hydrogen bonds, salt

bridges, and van der Waals contacts. Nevertheless, the de-

tailed geometries of these associations are significantly

different. In the repressor–DNA complex (Fig. 31-32), the

DNA bends around the protein in an arc of radius ⬃65 Å

which compresses the minor groove by ⬃2.5 Å near its

center (between the two protein monomers) and widens

it by ⬃2.5 Å toward its ends. In contrast, the DNA in com-

plex with Cro (Fig. 31-33), although also bent, is nearly

straight at its center and has a less compressed minor

groove (compare Figs. 31-32a and 31-33a). This explains

why the simultaneous replacement of three residues in

the repressor’s recognition helix with those occurring in

Cro does not cause the resulting hybrid protein to bind

DNA with Cro-like affinity: The different conformations

of the DNA in the repressor and Cro complexes prevents

any particular side chain from interacting identically with

the DNA in the two complexes.

2. Paul Sigler determined the X-ray structure of E. coli

trp repressor in complex with a DNA containing an 18-bp

palindrome (TGTA

CTAGTTAACTAGTAC, where the

trp repressor’s target sequence is underlined) that closely

resembles the trp operator (Section 31-3G). The dimeric

protein’s recognition helices bind, as expected, in succes-

sive major grooves of the DNA, each in contact with an

1288 Chapter 31. Transcription

JWCL281_c31_1260-1337.qxd 8/11/10 9:48 PM Page 1288

Section 31-3. Control of Transcription in Prokaryotes 1289

Figure 31-32 X-ray structure of the 69-residue N-terminal

domain of 434 phage repressor in complex with a 20-bp dsDNA

containing its target sequence. One strand of the DNA has the

sequence d(TATACAAGAAAGTTTGTACT). (a) The complex

viewed with the homodimeric protein’s 2-fold axis horizontal.

The protein is drawn in ribbon form with one of its two identical

subunits blue and the other red and with their helix–turn–helix

(HTH) motifs in darker shades.The DNA is drawn in stick form

with C white, N blue, O red, and P orange, and with successive P

atoms in the same chain connected by orange rods. (b) A

(a)

(b)

Figure 31-33 X-ray structure of the 71-residue 434 Cro

protein in complex with the same 20-bp DNA shown in

Fig. 31-32. Parts a and b correspond to those in Fig. 31-32. Note

the close but not identical correspondence between the two

structures and, in particular, the difference in the widths of the

schematic drawing indicating how the HTH motif, which

encompasses helices ␣2 and ␣3, interacts with its target DNA.

Short bars emanating from the polypeptide chain represent

peptide NH groups, hydrogen bonds are represented by dashed

lines, and DNA phosphates are represented by numbered circles.

The small circle is a water molecule. [Part a based on an X-ray

structure by and Part b courtesy of Aneel Aggarwal, John

Anderson, and Stephen Harrison, Harvard University. PDBid

2OR1.]

See Interactive Exercise 39 and Kinemage

Exercise 18-1

minor groove between the two subunits in each structure. [Part a

based on an X-ray structure by and Part b courtesy of Alfonso

Mondragón, Cynthia Wolberger, and Stephen Harrison, Harvard

University. PDBid 3CRO.]

(b)

(a)

JWCL281_c31_1260-1337.qxd 10/27/10 1:34 PM Page 1289

operator half-site (A

CTAGT; Fig. 31-34). There are nu-

merous hydrogen bonding contacts between the trp re-

pressor and its bound DNA’s nonesterified phosphate

oxygens. Astoundingly, however, there are no direct hydro-

gen bonds or nonpolar contacts that can explain the repres-

sor’s specificity for its operator. Rather, all but one of the

side chain–base hydrogen bonding interactions are medi-

ated by bridging water molecules (the one direct interac-

tion involves a base that can be mutated without greatly

affecting repressor binding affinity). Such buried water

molecules have therefore been described as “honorary”

protein side chains. In addition, the operator contains sev-

eral base pairs that are not in contact with the repressor

but whose mutation nevertheless greatly decreases repres-

sor binding affinity. This suggests that the operator as-

sumes a sequence-specific conformation that makes favor-

able contacts with the repressor. Indeed, comparison of

the X-ray structure of an uncomplexed 10-bp self-comple-

mentary DNA containing the trp operator’s half-site

(CCA

CTAGTGG) with that of the DNA in the trp repres-

sor–operator complex reveals that the ACTAGT half-site

assumes nearly identical idiosyncratic conformations and

patterns of hydration in both structures. However, the B-

DNA helix, which is straight in the DNA 10-mer, is bent by

15° toward the major groove in each operator half-site of

the repressor–operator complex. Other DNA sequences

could conceivably assume the repressor-bound operator’s

conformation but at too high an energy cost to form a sta-

ble complex with repressor (trp repressor’s measured 10

fold preference for its operator over other DNAs implies

an ⬃23 kJ ⴢ mol

⫺1

difference in their binding free ener-

gies).This phenomenon,in which a protein senses the base

sequence of DNA through the DNA’s backbone confor-

mation and/or flexibility, is referred to as indirect readout.

434 repressor apparently also employs indirect readout:

Replacing the central A ⴢ T base pair of the operator

shown in Fig. 31-32 with G ⴢ C reduces repressor binding

affinity by 50-fold even though 434 repressor does not

contact this region of the DNA.

It therefore appears that there are no simple rules govern-

ing how particular amino acid residues interact with bases.

Rather, sequence specificity results from an ensemble of mu-

tually favorable interactions between a protein and its target

DNA.

b. The met Repressor Contains a Two-Stranded

Antiparallel ␤ Sheet That Binds in Its Target

DNA’s Major Groove

The E. coli met repressor (MetJ), when complexed

with S-adenosylmethionine (SAM; Fig. 26-18), represses

the transcription of its own gene and those encoding en-

zymes involved in the synthesis of methionine (Fig. 26-

60) and SAM. The X-ray structure of the met repres-

sor–SAM–operator complex (Fig. 31-35), determined by

Simon Phillips, reveals a symmetric dimer of intertwined

homodimers that lacks an HTH motif. Rather, met re-

pressor binds to its palindromic target DNA sequence

through two symmetry-related pairs of symmetrical two-

stranded antiparallel ␤ sheets (called ␤ ribbons) that are

inserted in successive major grooves of the DNA. Each

␤ ribbon makes sequence-specific contacts with its tar-

get DNA sequence via hydrogen bonding and, probably,

indirect readout.

1290 Chapter 31. Transcription

Figure 31-34 X-ray structure of an E. coli trp repressor–

operator–tryptophan complex. The complex is viewed with its

molecular 2-fold axis horizontal.The protein’s two identical

subunits are drawn in ribbon form colored pink and blue with

their HTH motifs (helices D and E) more deeply colored.The

18-bp-containing self-complementary dsDNA is shown in stick

form with C white, N blue, O red, P orange, and with successive

P atoms in the same chain connected by orange rods.The trp

repressor binds its operator only when

L-tryptophan, drawn in

space-filling form with C green, is simultaneously bound. Note

that the protein’s recognition helices (E) bind, as expected, in

successive major grooves of the DNA but extend approximately

perpendicular to the DNA duplex axis. In contrast, the

recognition helices of 434 repressor and Cro proteins are nearly

parallel to the major grooves of their bound DNAs (Figs. 31-32

and 31-33), whereas those of CAP assume an intermediate

orientation (Fig. 31-31). [Based on an X-ray structure by Paul

Sigler,Yale University. PDBid 1TRO.]

See Interactive Exercise 40

JWCL281_c31_1260-1337.qxd 10/19/10 10:54 AM Page 1290