Voet D., Voet Ju.G. Biochemistry

Подождите немного. Документ загружается.

Phillips first determined the X-ray structure of met re-

pressor in the absence of DNA. Model building studies

aimed at elucidating how met repressor binds to its palin-

dromic target DNA assumed that the 2-fold rotation axes

of both molecules would be coincident, as they are in all

prokaryotic protein–DNA complexes of known structure.

There were, consequently, two reasonable choices: (1) The

protein could dock to the DNA with the above pairs of ␤

ribbons entering successive major grooves; or (2) a symme-

try-related pair of protruding ␣ helices on the opposite face

of the protein could do so in a manner resembling the way

in which the recognition helices of HTH motifs interact

with DNA. A variety of structural criteria suggested that

the ␣ helices make significantly better contacts with the

DNA than do the ␤ ribbons.Thus, the observation that it is,

in fact, the ␤ ribbons that bind to the DNA provides an im-

portant lesson: The results of model building studies must

be treated with utmost caution. This is because our impre-

cise understanding of the energetics of intermolecular in-

teractions (Sections 8-4 and 29-2) prevents us from reliably

predicting how associating macromolecules conform to

one another. In the case of the met repressor, unpredicted

mutual structural accommodations of the protein and

DNA yielded a significantly more extensive interface than

had been predicted by simply docking the uncomplexed

Met repressor to canonical B-DNA.

The numerous prokaryotic transcriptional regulators

of known structure either contain an HTH motif or pairs

of ␤ ribbons like the met repressor (although numerous

prokaryotic DNA-binding proteins, including CAP, con-

tain an elaboration of the HTH motif known as the

winged helix motif in which two protein loops, one of

which contacts the DNA’s minor groove, flank the HTH

recognition helix like the wings of a butterfly). Moreover,

most of these proteins are homodimers that bind to palin-

dromic or pseudopalindromic DNA target sequences.

However, eukaryotic transcription factors, as we shall see

in Section 34-3B, employ a much wider variety of struc-

tural motifs to bind their target DNAs, many of which

lack symmetry.

E. araBAD Operon: Positive and Negative Control

by the Same Protein

Humans neither metabolize nor intestinally absorb the

plant sugar

L-arabinose. Hence, the E. coli that normally in-

habit the human gut are periodically presented with a ban-

quet of this pentose. Three of the five E. coli enzymes that

Section 31-3. Control of Transcription in Prokaryotes 1291

Figure 31-35 X-ray structure of the E. coli met

repressor–SAM–operator complex. (a) The overall structure of

the complex as viewed along its 2-fold axis of symmetry. The

self-complementary 18-bp DNA is drawn in stick form, and

SAM, which must be bound to the repressor for it to also bind

DNA, is shown in space-filling form with the DNA C white,

SAM C green, N blue, O red, P orange, and S yellow. The DNA

binds four identical 104-residue repressor subunits. Pairs of

subunits (light cyan and lavender) form symmetric dimers in

which each subunit donates one strand of the 2-stranded

antiparallel ␤ ribbon that is inserted in the DNA’s major groove

(upper left and lower right).Two such dimers pair across the

complex’s 2-fold axis via their antiparallel N-terminal helices,

which contact one another over the DNA’s minor groove. (b)

Detailed view of the lower half of Part a showing the 2-stranded

antiparallel ␤ ribbon (residues 21–29) inserted into the DNA’s

major groove, as viewed along its local 2-fold axis (rotated

relative to Part a by 50° about the vertical axis).The DNA is

shown in space-filling form and the polypeptide chains are

drawn in stick form with C green. [Based on an X-ray structure

by Simon Phillips, University of Leeds, U.K. PDBid 1CMA.]

See Interactive Exercise 41

(b)

(a)

JWCL281_c31_1260-1337.qxd 10/19/10 11:26 AM Page 1291

metabolize arabinose are products of the catabolite re-

pressible araBAD operon (Fig. 31-36).

The araBAD operon, as Robert Schleif has shown, con-

tains, moving upstream from its transcriptional start site,

the araI, araO

, and araO

control sites (Fig. 31-37a). The

araI site (I for inducer) consists of two closely similar 17-bp

half-sites, araI

and araI

, that are direct repeats separated

by 4 bp and are oriented such that araI

, which overlaps the

⫺35 region of the araBAD promoter, is downstream of

araI

. Likewise, araO

consists of two directly repeating

half-sites, O

and O

. Intriguingly, however, araO

con-

sists of a single half-site that is located in a noncoding up-

stream region of the araC gene (see below), at position

⫺270 relative to the araBAD start site.

The transcription of the araBAD operon is regulated by

both CAP–cAMP and the arabinose-binding protein

AraC. Each 292-subunit of the homodimeric AraC con-

sists of an N-terminal, arabinose-binding, dimerization do-

main (residues 1–170) connected via a flexible linker to a

C-terminal DNA-binding domain (residues 178–292).

Regulation of the araBAD operon occurs as follows

(Fig. 31-37):

1. In the absence of AraC, RNA polymerase initiates

transcription of the araC gene in the direction away from

its upstream neighbor, araBAD. The araBAD operon is ex-

pressed at a low basal level.

2. When AraC is present, but neither arabinose nor

CAP–cAMP (high glucose),AraC binds to araO

and araI

via two HTH motifs in each of its subunits. The binding of

AraC to araI

prevents RNAP from initiating transcription

of the araBAD operon (negative control).A series of dele-

tion mutations indicate that the presence of araO

is also

required for the repression of araBAD. The remarkably

large 210-bp separation between araO

and araI

therefore

strongly suggests that the DNA between them is looped

such that a dimeric molecule of AraC protein simultane-

ously binds to both araO

and araI

.This is corroborated by

the observation that the level of repression is greatly

diminished by the insertion of 5 bp (half a turn) of DNA

between these two sites, thereby transferring araO

to the

opposite face of the DNA helix relative to araI

in the

putative loop.Yet, the insertion of 11 bp (one turn) of DNA

has no such effect. Moreover, looping does not readily oc-

cur unless the DNA is supercoiled, which presumably

drives the looping process. The AraC dimer also binds to

araO

, the operator of the araC gene, so as to block the

transcription of araC but only at high concentrations.Thus,

it is likely that DNA looping itself represses the transcrip-

tion of araC. In either case, the expression of araC is

autoregulatory.

3. When arabinose is present, it allosterically induces

the AraC subunit bound to araO

to instead bind to araI

This activates RNAP to transcribe the araBAD genes (pos-

itive control). When the cAMP level is high (low glucose),

CAP–cAMP, whose presence is required to achieve the

maximum level of transcriptional activation, binds to a site

between araO

and araI

, where it functions to help break

the loop between araO

and araI

and hence to increase the

affinity of AraC for araI

.The orientation of araO

with re-

spect to araC is opposite to that of araI with respect to

araBAD, and hence the binding of AraC–arabinose at

araO

blocks RNAP binding at the araC promoter, that is,

it represses the expression of AraC.

If the araI

subsite is mutated so as to increase AraC’s

affinity for it, arabinose is no longer required for transcrip-

tional activation.This suggests that arabinose does not con-

formationally transform AraC to an activator but, rather,

weakens its binding affinity for araO

. If the araI site is

1292 Chapter 31. Transcription

Activator

Repressor

-Arabinose

araC mRNA

araC araO

araO

araI araB araA araD

Structural genes

Control sites

Regulatory gene

araBAD mRNA

L-ribulokinase

L-ribulose-5-P

epimerase

L-arabinose

isomerase

permease

system

L-Arabinose

(external)

L-Arabinose

(internal)

L-Ribulose L-Ribulose-5-P D -Xylulose-5-P

ADPATP

CAP

Figure 31-36 A genetic map of the E. coli araC and araBAD operons. The map indicates the

proteins these operons encode and the reactions in which these proteins participate. The permease

system, which transports arabinose into the cell, is the product of the araE and araF genes,

which occur in two independent operons.The pathway product, xylulose-5-phosphate, is

converted, via the pentose phosphate pathway, to the glycolytic intermediates fructose-6-

phosphate and glyceraldehyde-3-phosphate (Section 23-4). [After Lee, N., in Miller, J.H. and

Rezinkoff, W.S. (Eds.), The Operon, p. 390, Cold Spring Harbor Laboratory Press (1979).]

JWCL281_c31_1260-1337.qxd 8/11/10 9:48 PM Page 1292

turned around or if it is moved upstream so that araI

does

not overlap the araBAD promoter, AraC cannot stimulate

transcription. Evidently, AraC activates RNAP through spe-

cific and relatively inflexible protein–protein interactions.

The X-ray structures of the N-terminal domain of AraC

(residues 2–178), in both the presence and the absence of

arabinose, were determined by Schleif and Cynthia Wol-

berger. In the presence of arabinose, this domain consists

of an 8-stranded ␤ barrel followed by two antiparallel ␣

helices (Fig. 31-38). Two such domains associate via an an-

tiparallel coiled coil between each of their C-terminal

helices to form the protein’s dimerization interface. An

arabinose molecule binds in a pocket of each ␤ barrel via a

network of direct and water-mediated hydrogen bonds

with side chains that line the pocket. Residues 7 to 18 of the

N-terminal arm lie across the mouth of the sugar-binding

pocket (residues 2–6 are disordered), thereby fully enclos-

ing the arabinose. The structure of the N-terminal domain

in the absence of arabinose is largely superimposable on

that in the complex with arabinose, with the exception that

Section 31-3. Control of Transcription in Prokaryotes 1293

araO

(a) When AraC is absent,

araC is transcribed and

araBAD is transcribed

at a basal level

araO

araI

araBAD

araC

mRNA

araO

AraC

C-terminal

DNA-binding domain

N-terminal arm

Linker

Arabinose-binding

pocket

N-terminal

dimerization domain

araI

araBAD

(b)

When cAMP and L-arabinose

are low, AraC represses

araBAD transcription

are

abundant, araBAD

transcription is activated

araO

araC

araI

araBAD

RNA polymerase

araBAD mRNA

araBAD mRNA (basal level)

RNA polymerase

CAP

araC

AraC–arabinose

CAP

CAP–cAMP

araC

Figure 31-37 The mechanism of araBAD regulation. (a) In

the absence of AraC, RNAP initiates the transcription of araC.

araBAD is also expressed but at a low basal level. (b) When

AraC is present, but not

L-arabinose or cAMP, AraC links

together araO

and araI

to form a DNA loop, thereby repressing

both araC and araBAD.(c) When AraC and

L-arabinose are

Figure 31-38 X-ray structure of E. coli AraC in complex with

L-arabinose. The homodimeric protein is viewed along its 2-fold

axis with each of its subunits colored in rainbow order from

N-terminus (blue) to C-terminus (red).The arabinose is drawn in

space-filling form with C green and O red. [Based on an X-ray

structure by Robert Schleif and Cynthia Wolberger, Johns

Hopkins University. PDBid 2ARC.]

both present and cAMP is abundant, the resulting AraC–

arabinose complex releases araO

and instead binds araI

thereby activating araBAD transcription. This process is

facilitated by the binding of CAP–cAMP. araC is repressed by

the binding of AraC–arabinose to araO

JWCL281_c31_1260-1337.qxd 8/11/10 9:48 PM Page 1293

Hinge helix

N-subdomain

C-subdomain

Tetramerization helix

Headpiece

Inducer-binding

pocket

the N-terminal arm is disordered, a not unexpected obser-

vation considering that it interacts with bound arabinose

via a series of hydrogen bonds.

How does arabinose binding induce the AraC subunit

bound at araO

to instead bind to araI

? Several lines of

evidence indicate that, in the absence of arabinose, AraC’s

N-terminal arm binds to its DNA-binding domain in a

way that favors loop formation: (1) the deletion of the

N-terminal arm beyond its sixth residue makes AraC act as

if arabinose is present; (2) mutations to surface residues on

the DNA-binding domain that presumably eliminate its

binding of the N-terminal arm also constitutively activate

AraC; and (3) mutations in the DNA-binding domain that

weaken the binding of arabinose to the protein, presum-

ably by strengthening the binding of the N-terminal arm,

can be suppressed by a second mutation in the N-terminal

arm or by the deletion of its five N-terminal residues. Evi-

dently, the binding of the N-terminal arms to the DNA-binding

domains in the absence of arabinose rigidifies the AraC

dimer such that it cannot simultaneously bind to the directly

repeated araI

and araI

and hence induce the transcription

of araBAD. This is corroborated by the observations that

(1) joining two AraC DNA-binding domains by flexible

polypeptide linkers yields proteins that behave like AraC

in the presence of arabinose, and (2) a construct consisting

of two double-stranded araI

half-sites flexibly connected

by a 24-nt segment of ssDNA binds wild-type AraC with an

affinity that is unaffected by arabinose.

F. lac Repressor II: Structure

Here we continue our discussions of the lac repressor, but

now in terms of the concepts learned in Sections 31-3C–E.

a. Loop Formation Is Important in the Expression

of the lac Operon

DNA loop formation, which is now known to occur in

numerous bacterial and eukaryotic systems, apparently

permits several regulatory proteins and/or regulatory sites

on one protein to simultaneously influence transcription

initiation by RNAP. In fact, the lac repressor has three bind-

ing sites on the lac operon: the primary operator (Fig. 31-

28), now known as O

, and two so-called pseudo-operators

(previously thought to be nonfunctional evolutionary fos-

sils), O

and O

, which are located 401 bp downstream and

92 bp upstream of O

(within the lacZ gene and overlap-

ping the CAP binding site, respectively). Müller-Hill deter-

mined the relative contributions of these various operators

to the repression of the lac operon through the construc-

tion of a set of eight plasmids: Each contained the lacZ

gene under the control of the natural lac promoter as well

as the three lac operators (O

, O

, and O

), which were ei-

ther active or mutagenically inactive in all possible combi-

nations. When all three operators are active, lacZ expres-

sion is repressed 1300-fold relative to when all three

operators are inactive.The inactivation of only O

results in

almost complete loss of repression whereas the inactiva-

tion of only O

or O

causes only a ⬃2-fold loss in repres-

sion. However, when O

and O

are both inactive, repres-

sion is decreased ⬃70-fold. These results suggest that effi-

cient repression requires the formation of a DNA loop be-

tween O

and either O

or O

. Indeed, such loop formation,

and/or the cooperativity of repressor binding arising from

it, appears to be a greater contributor to repression than

repressor binding to O

alone, which provides only 19-fold

repression.

b. The lac Repressor Is a Dimer of Dimers

Ponzy Lu and Mitchell Lewis determined the X-ray

structures of the lac repressor alone, in its complex with

IPTG, and in its complex with a 21-bp duplex DNA seg-

ment whose sequence is a palindrome of the left half of O

(Fig. 31-28). Each repressor subunit consists of five func-

tional units (Fig. 31-39): (1) an N-terminal DNA-binding

domain (residues 1–49) which is known as the “headpiece”

because it is readily proteolytically cleaved away from the

remaining still tetrameric “core” protein; (2) a hinge helix

(residues 50–58) that also binds to the DNA; (3 and 4) a

sugar-binding domain (residues 62–333) that is divided into

an N-subdomain and a C-subdomain; and (5) a C-terminal

tetramerization helix (residues 340–360).

1294 Chapter 31. Transcription

Figure 31-39 X-ray structure of the lac repressor subunit. The

DNA-binding domain (the headpiece), which contains an HTH

motif, is red, the DNA-binding hinge helix is yellow, the

N-subdomain of the sugar-binding domain is light blue, its

C-subdomain is dark blue, and the tetramerization helix is

purple. [Courtesy of Ponzy Lu and Mitchell Lewis, University

of Pennsylvania. PDBid 1LBI.]

JWCL281_c31_1260-1337.qxd 8/11/10 9:48 PM Page 1294

The lac repressor has an unusual quaternary structure

(Fig. 31-40a).Whereas nearly all homotetrameric nonmem-

brane proteins of known structure have D

symmetry

(three mutually perpendicular 2-fold axes; Fig. 8-65b), lac

repressor is a V-shaped protein that has only 2-fold symme-

try. Each leg of the V consists of a locally symmetric dimer

of closely associated repressor subunits. Two such dimers

associate rather tenuously, but with 2-fold symmetry, at the

base (point) of the V to form a dimer of dimers.

In the structures of lac repressor alone and that of its

IPTG complex, the DNA-binding domain is not visible,

apparently because the hinge region that loosely tethers it

to the rest of the protein is disordered. However, in the

DNA complex, in which one DNA duplex binds to each of

the two dimers forming the repressor tetramer, the DNA

domain forms a compact globule containing three helices,

the first two of which form a helix–turn–helix (HTH) mo-

tif. The two DNA-binding domains extending from each

repressor dimer (at the top of each leg of the V) bind in

successive major grooves of a DNA molecule via their

HTH motifs, much as is seen, for example, in the com-

plexes of 434 phage repressor and trp repressor with their

target DNAs (Figs. 31-32 and 31-34). The binding of the

lac repressor distorts the operator DNA such that it

bends away from the DNA-binding domain with an ⬃60

Å radius of curvature due to an ⬃45° kink at the center of

the operator that widens the DNA’s minor groove to over

11 Å and reduces its depth to less than 1 Å. These distor-

tions permit the now ordered hinge helix to bind in the

minor groove so as to contact the identically bound hinge

helix from the other subunit of the same dimer. NMR

structures by Robert Kaptein and Rolf Boelens reveal

that the DNA-binding domain, when cleaved from the re-

pressor, binds to the lac operator without distorting the

DNA, but that the DNA-binding domain together with

the hinge helix forms a complex with the lac operator in

which the hinge helix binds in the DNA’s distorted minor

groove (Fig. 31-40b) as in the X-ray structure. Thus, the

binding of the two hinge helices to the lac operator ap-

pears necessary for DNA distortion. The two DNA du-

plexes that are bound to each repressor tetramer are ⬃25

Å apart and do not interact.

The sugar-binding domain consists of two topologi-

cally similar subdomains that are bridged by three

polypeptide segments (Fig. 31-39).The two sugar-binding

domains of a dimer make extensive contacts (Fig. 31-

40a). IPTG binds to each sugar-binding domain between

its subdomains. This does not significantly change the

conformations of these subdomains, but it changes the

angle between them.Although the hinge helix is not visi-

ble in the IPTG complex, model building indicates that,

since the dimer’s two hinge helices extend from its sugar-

binding domains, this conformation change levers apart

these hinge helices by 3.5 Å such that they and their at-

tached HTH motifs can no longer simultaneously bind to

their operator half-sites. Thus, inducer binding, which is

allosteric within the dimer (has a positive homotropic ef-

fect; Section 10-4), greatly loosens the repressor’s grip on

the operator.

The C-terminal helices from each subunit, which are lo-

cated on the opposite end of each subunit from the DNA-

binding portion (at the point of the V), associate to form a

bundle of four parallel helices that holds together the two

repressor dimers, thereby forming the tetramer (Fig. 31-

40a). The allosteric effects of inducer binding within each

dimer are apparently not transmitted between dimers.

Section 31-3. Control of Transcription in Prokaryotes 1295

Figure 31-40 The structure of the lac repressor in complex

with DNA. (a) The X-ray structure of the lac repressor tetramer

bound to two 21-bp segments of symmetric lac operator DNA.

The protein subunits are shown in ribbon form in yellow, cyan,

green, and orange and the dsDNA segments are drawn in space-

filling form with C white, N blue, O red, and P orange. [Courtesy

of Ponzy Lu and Mitchell Lewis with coordinates generated by

Benjamin Weider, University of Pennsylvania. PDBid 1LBG.]

(b) The NMR structure of the 23-bp O

lac operator DNA in

complex with two identical segments of the lac repressor

consisting of its DNA-binding domain and its hinge helix. Each

of the protein subunits is drawn in ribbon form colored in

rainbow order from its N-terminus (blue) to its C-terminus (red).

The DNA is represented 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.The complex is viewed with its 2-fold

axis vertical. Note that the protein dimer’s two HTH motifs are

inserted in successive major grooves at the periphery of the

complex and that the insertion of the two centrally located hinge

helices into the DNA’s minor groove greatly widens and flattens

the minor groove at this point and kinks the DNA in an upward

bend. [Based on an NMR structure by Robert Kaptein and Rolf

Boelens, Utrecht University, The Netherlands. PDBid 2KEI.]

(b)

(a)

JWCL281_c31_1260-1337.qxd 8/11/10 9:48 PM Page 1295

Moreover, the E. coli purine repressor (PurR), which is

homologous to the lac repressor but lacks its C-terminal

helix, crystallizes as a dimer whose X-ray structure closely

resembles that of the lac repressor dimer.What then is the

function of lac repressor tetramerization?

Model building suggests that when the lac repressor

tetramer simultaneously binds to both the O

and O

oper-

ators, the 93-bp DNA segment containing them forms a

loop ⬃80 Å in diameter (Fig. 31-41). Furthermore, the

CAP–cAMP binding site is exposed on the inner surface

of the loop. Adding the CAP–cAMP at its proper position

to this model reveals that the ⬃90° curvature which

CAP–cAMP binding imposes on DNA (Fig. 31-31) has the

correct direction and magnitude to stabilize the DNA

loop, thereby stabilizing this putative CAP–cAMP–lac

repressor–DNA complex.It may seem paradoxical that the

binding of CAP–cAMP, a transcriptional activator, stabi-

lizes the repressor–DNA complex. However, when both

glucose and lactose are in short supply, it is important that

the bacterium lower its basal rate of lac operon expression

in order to conserve energy. The binding site (promoter)

for RNAP is also located on the inner surface of the loop.

Thus, the large size of the RNAP molecule would prevent

it from fully engaging the promoter in this looped complex,

thereby maximizing repression.

c. Combining Genetic and Structural Studies of

the lac Repressor Reveals Its Allosterically

Important Residues

The phenotypes of 4042 point mutations of the lac re-

pressor, which encompass nearly all of its 360 residues

(making the lac repressor the most exhaustively mutation-

ally characterized protein known) have been mapped onto

its X-ray structure. Mutations with an I

⫺

phenotype (lac re-

pressors that fail to bind to the lac operator, so that ␤-

galactosidase is constitutively synthesized) are located at

the lac repressor’s DNA-binding interface, at its dimer in-

terface, or at internal residues of its inducer-binding core

domain. Residues whose mutations result in the I

pheno-

type (S for super-repressed; lac repressors that, in the pres-

ence of inducer, continue to repress the synthesis of ␤-

galactosidase) appear to be of two types: (1) residues that

are in direct contact with the inducer, whose alteration

therefore interferes with inducer binding; and (2) residues

at the dimer interface that are ⬎8 Å from (not in direct

contact with) the inducer-binding site. These latter muta-

tions reveal which residues mediate the lac repressor’s

allosteric mechanism rather than directly binding the inducer

or the DNA. Most of the allosterically important residues

are located at the dimer interface and are members of the

N-subdomain of the core domain, which links the inducer-

binding sites to the operator DNA-binding sites.This is con-

sistent with the observation that inducer binding causes a

relative twist and translation of the N-subdomain, a move-

ment which is propagated to the hinge helix and DNA-

binding domain. This study demonstrates the power of

combining genetic analysis with structural studies to eluci-

date structure–function relationships.

G. trp Operon: Attenuation

We now discuss a sophisticated transcriptional control

mechanism named attenuation through which bacteria reg-

ulate the expression of certain operons involved in amino

acid biosynthesis. This mechanism was discovered through

the study of the E. coli trp operon (Fig. 31-42), which en-

codes five polypeptides comprising three enzymes that me-

diate the synthesis of tryptophan from chorismate (Section

26-5Bc). Charles Yanofsky established that the trp operon

genes are coordinately expressed under the control of the

trp repressor, a dimeric protein of identical 107-residue sub-

units that is the product of the trpR gene (which forms an

independent operon). The trp repressor binds

L-tryptophan,

the pathway’s end product, to form a complex that specifi-

cally binds to trp operator (trpO, Fig. 31-43) so as to reduce

the rate of trp operon transcription 70-fold. The X-ray struc-

ture of the trp repressor–operator complex (Section 31-3Da)

indicates that tryptophan binding allosterically orients trp

repressor’s two symmetry related helix–turn– helix “DNA

reading heads” so that they can simultaneously bind to

trpO (Fig. 31-34). Moreover, the bound tryptophan forms a

hydrogen bond to a DNA phosphate group, thereby

1296 Chapter 31. Transcription

Figure 31-41 Model of the 93-bp DNA loop formed when lac

repressor binds to O

and O

. The proteins are represented by

their C

␣

backbones and the DNA is drawn in stick form with its

sugar–phosphate backbones traced by helical ribbons.The model

was constructed from the X-ray structure of the lac repressor

(magenta) in complex with two 21-bp operator DNA segments

(red) and the X-ray structure of CAP–cAMP (blue) in complex

with its 30-bp target DNA (cyan; Fig. 31-28).The remainder of

the DNA loop was generated by applying a smooth curvature to

canonical B-DNA (white) with the ⫺10 and ⫺35 regions of the

lac promoter highlighted in green. [Courtesy of Ponzy Lu and

Mitchell Lewis, University of Pennsylvania.]

JWCL281_c31_1260-1337.qxd 8/11/10 9:48 PM Page 1296

strengthening the repressor–operator association. Trypto-

phan therefore acts as a corepressor; its presence prevents

what is then superfluous tryptophan biosynthesis (SAM

similarly functions as a corepressor with the met repressor;

Fig. 31-35a). The trp repressor also controls the synthesis

of at least two other operons: the trpR operon and the

aroH operon (which encodes one of three isozymes that

catalyze the initial reaction of chorismate biosynthesis;

Section 26-5Bc).

a. Tryptophan Biosynthesis Is Also Regulated

by Attenuation

The trp repressor–operator system was at first thought

to fully account for the regulation of tryptophan biosynthe-

sis in E. coli. However, the discovery of trp deletion mu-

tants located downstream from trpO that increase trp

operon expression 6-fold indicated the existence of an ad-

ditional transcriptional control element. Sequence analysis

established that trpE, the trp operon’s leading structural

gene, is preceded by a 162-nucleotide leader sequence

(trpL). Genetic analysis indicated that the new control ele-

ment is located in trpL, ⬃30 to 60 nucleotides upstream of

trpE (Fig. 31-42).

When tryptophan is scarce, the entire 6720-nucleotide

polycistronic trp mRNA, including the trpL sequence, is

synthesized. As the tryptophan concentration increases,

the rate of trp transcription decreases as a result of the

trp repressor–corepressor complex’s consequent greater

abundance. Of the trp mRNA that is transcribed, how-

ever, an increasing proportion consists of only a 140-nu-

cleotide segment corresponding to the 5¿ end of trpL. The

availability of tryptophan therefore results in the prema-

ture termination of trp operon transcription. The control

element responsible for this effect is consequently termed

an attenuator.

b. The trp Attenuator’s Transcription Terminator Is

Masked when Tryptophan Is Scarce

What is the mechanism of attenuation? The attenuator

transcript contains four complementary segments that can

form one of two sets of mutually exclusive base-paired

Section 31-3. Control of Transcription in Prokaryotes 1297

anthranilate

synthase

component I

anthranilate

synthase

component II

tryptophan

synthase

β subunit

tryptophan

synthase

α subunit

anthranilate

synthase

N-(5′-phosphoribosyl)-

anthranilate isomerase,

Indole-3-glycerol

phosphate synthase

tryptophan synthase

(α

)

(CoI

CoII

)

Chorismate Anthranilate N

-(5′-Phosphoribosyl)-

anthranilate

Enol-1-o-carboxy-

phenylamino-

1-deoxyribulose

phosphate

Indole-3-glycerol-P

Glutamine Glutamate

Pyruvate

PRPP PP

L-Tryptophan

L-Serine Glyceraldehyde-3-P

Control

sites

trpP,O trpL trpE trpD trpC trpB trpA

Attenuator

Structural genes

mRNA

Leader

mRNA

CG A T GT A GAATGACTCAGAAC T CA

GC T A CA T CTTACTGACTTGTG A GT

⫺20 ⫺10 ⫹1

Figure 31-42 A genetic map of the E. coli trp operon

indicating the enzymes it specifies and the reactions they

catalyze. The gene product of trpC catalyzes two sequential

Figure 31-43 The base sequence of the trp operator. The

nearly palindromic sequence is boxed and its ⫺10 region is

overscored.

reactions in the synthesis of tryptophan. [After Yanofsky, C., J.

Am. Med. Assoc. 218, 1027 (1971).]

JWCL281_c31_1260-1337.qxd 8/11/10 9:48 PM Page 1297

hairpins (Fig. 31-44). Segments 3 and 4 together with the suc-

ceeding residues comprise a normal intrinsic transcription

terminator (Section 31-2Da): a G ⫹ C–rich sequence that

can form a self-complementary hairpin structure followed

by several sequential U’s (compare with Fig. 31-18). Tran-

scription rarely proceeds beyond this termination site unless

tryptophan is in short supply.

A section of the leader sequence, which includes seg-

ment 1 of the attenuator, is translated to form a 14-residue

polypeptide that contains two consecutive Trp residues

(Fig. 31-44, left). The position of this particularly rare

dipeptide segment (1.1% of the residues in E. coli proteins

are Trp;Table 4-1) provided an important clue to the mech-

anism of attenuation.An additional essential aspect of this

mechanism is that ribosomes commence the translation of

a prokaryotic mRNA shortly after its 5¿ end has been syn-

thesized.

The above considerations led Yanofsky to propose the

following model of attenuation (Fig. 31-45).An RNA poly-

merase that has escaped repression initiates trp operon

transcription. Soon after the ribosomal initiation site of the

trpL gene has been transcribed, a ribosome attaches to it

and begins translation of the leader peptide. When trypto-

phan is abundant, so that there is a plentiful supply of tryp-

tophanyl–tRNA

Tr p

(the transfer RNA specific for Trp with

an attached Trp residue; Section 32-2C), the ribosome fol-

lows closely behind the transcribing RNA polymerase so as

to sterically block the formation of the 2 ⴢ 3 hairpin. In-

deed, RNA polymerase pauses past position 92 of the tran-

script and only continues transcription on the approach of

a ribosome, thereby ensuring the proximity of these two

entities at this critical position.The prevention of 2 ⴢ 3 hair-

pin formation permits the formation of the 3 ⴢ 4 hairpin,

the transcription terminator pause site, which results in the

termination of transcription (Fig. 31-45a). When trypto-

phan is scarce, however, the ribosome stalls at the tandem

UGG codons (which specify Trp; Table 5-3) because of the

lack of tryptophanyl–tRNA

Tr p

. As transcription continues,

the newly synthesized segments 2 and 3 form a hairpin be-

cause the stalled ribosome prevents the otherwise compet-

itive formation of the 1 ⴢ 2 hairpin (Fig. 31-45b).The forma-

tion of the transcriptional terminator’s 3 ⴢ 4 hairpin is

thereby pre-empted for sufficient time for RNA poly-

merase to transcribe through it and consequently through

the remainder of the trp operon. The cell is thus provided

with a regulatory mechanism that is responsive to the tryp-

tophanyl–tRNA

Tr p

level, which, in turn, depends on the

protein synthesis rate as well as on the tryptophan supply.

There is considerable evidence supporting this model of

attenuation. The trpL transcript is resistant to limited

1298 Chapter 31. Transcription

Figure 31-44 The alternative secondary structures of trpL

mRNA. The formation of the base paired 2 ⴢ 3 (antiterminator)

hairpin (right) precludes the formation of the 1 ⴢ 2 and 3 ⴢ 4

(terminator) hairpins (left) and vice versa.Attenuation results in

the premature termination of transcription immediately after

Trp

UUUUUU

U U U U U U

130

110

“terminator”

“antiterminator”

nucleotide 140 when the 3 ⴢ 4 hairpin is present.The arrow

indicates the mRNA site past which RNA polymerase pauses

until approached by an active ribosome. [After Fisher, R.F. and

Yanofsky, C., J. Biol. Chem. 258, 8147 (1983).]

JWCL281_c31_1260-1337.qxd 8/11/10 9:48 PM Page 1298

RNase T1 digestion, indicating that it has extensive second-

ary structure. The significance of the tandem Trp codons in

the trpL transcript is corroborated by their presence in trp

leader regions of several other bacterial species. Moreover,

the leader peptides of the five other amino acid–biosynthe-

sizing operons known to be regulated by attenuation (most

exclusively so) are all rich in their corresponding amino

acid residues (Table 31-3). For example, the E. coli his

operon, which specifies enzymes synthesizing histidine

(Fig. 26-65), has seven tandem His residues in its leader

peptide whereas the ilv operon, which specifies enzymes

participating in isoleucine, leucine, and valine biosynthesis

(Fig. 26-61), has five Ile’s, three Leu’s, and six Val’s in its

leader peptide. Finally, the leader transcripts of these oper-

ons resemble that of the trp operon in their capacity to

form two alternative secondary structures, one of which

contains a trailing termination structure.

H. Riboswitches Are Metabolite-Sensing RNAs

We have just seen how the formation of secondary struc-

ture in a growing RNA transcript can regulate gene expres-

sion though attenuation. The conformational flexibility of

mRNA also allows it to regulate genes by directly interacting

Section 31-3. Control of Transcription in Prokaryotes 1299

Figure 31-45 Attenuation in the trp operon. (a) When

tryptophanyl–tRNA

Tr p

is abundant, the ribosome translates trpL

mRNA. The presence of the ribosome on segment 2 prevents the

formation of the base-paired 2 ⴢ 3 hairpin.The 3 ⴢ 4 hairpin, an

essential component of the transcriptional terminator, can

thereby form, thus aborting transcription. (b) When tryptophanyl–

High tryptophan(a)

(b)

Leader

peptide

Ribosome transcribing

the leader peptide mRNA

Transcription

terminator

“Terminated”

RNA polymerase

trpL mRNA

Low tryptophan

Antiterminator

Ribosome stalled

at tandem

Trp codons

trp operon mRNA

Transcribing

RNA

polymerase

DNA encoding

operontrp

Table 31-3 Amino Acid Sequences of Some Leader Peptides in Operons

Subject to Attenuation

Operon Amino Acid Sequence

trp Met-Lys-Ala-Ile-Phe-Val-Leu-Lys-Gly-TRP-TRP-Arg-Thr-Ser

pheA Met-Lys-His-Ile-Pro-PHE-PHE-PHE-Ala-PHE-PHE-PHE-Thr-PHE-Pro

his Met-Thr-Arg-Val-Gln-Phe-Lys-HIS-HIS-HIS-HIS-HIS-HIS-HIS-Pro-Asp

leu Met-Ser-His-Ile-Val-Arg-Phe-Thr-Gly-LEU-LEU-LEU-LEU-Asn-Ala-

Phe-Ile-Val-Arg-Gly-Arg-Pro-Val-Gly-Gly-Ile-Gln-His

thr Met-Lys-Arg-ILE-Ser-THR-THR-ILE-THR-THR-THR-ILE-THR-ILE-

THR-THR-Gln-Asn-Gly-Ala-Gly

ilv Met-Thr-Ala-LEU-LEU-Arg-VAL-ILE-Ser-LEU-VAL-VAL-ILE-Ser-

VAL-VAL-VAL-ILE-ILE-ILE-Pro-Pro-Cys-Gly-Ala-Ala-Leu-Gly-Arg-

Gly-Lys-Ala

Residues in uppercase are synthesized in the pathway catalyzed by the operon’s gene products.

Source: Yanofsky, C., Nature 289, 753 (1981).

tRNA

Tr p

is scarce, the ribosome stalls on the tandem Trp codons

of segment 1.This situation permits the formation of the 2 ⴢ 3

hairpin which, in turn, precludes the formation of the 3 ⴢ 4

hairpin. RNA polymerase therefore transcribes through this

unformed terminator and continues trp operon transcription.

JWCL281_c31_1260-1337.qxd 8/11/10 9:48 PM Page 1299

with certain cellular metabolites, thereby eliminating the

need for sensor proteins such as the lac repressor,CAP, and

the trp repressor.

In E. coli, the biosynthesis of thiamine pyrophosphate

(TPP; Section 17-3Ba) requires the action of several pro-

teins whose levels vary according to the cell’s need for TPP.

In at least two of the relevant genes the untranslated re-

gions at the 5¿ end of the mRNA include a highly conserved

sequence called the thi box. The susceptibility of the thi

box to chemical or enzymatic cleavage, as Ronald Breaker

showed, differs in the presence and absence of TPP, sug-

gesting that the RNA changes its secondary structure when

TPP binds to it (the binding of a metabolite by RNA is not

unprecedented; synthetic oligonucleotides known as ap-

tamers bind specific molecules with high specificity and

affinity; Section 7-6C). The TPP-sensing mRNA element

has been dubbed a riboswitch.

The predicted secondary structure of the TPP-sensing

riboswitch and its proposed mechanism are shown in Fig.

31-46a. In the absence of TPP, the mRNA assumes a con-

formation that allows a ribosome to begin translation. In

the presence of TPP, an alternative secondary structure

masks the sequence that identifies its translational initia-

tion site to the ribosome (its so-called Shine–Dalgarno

sequence; Section 32-3Cb) so that the ribosome cannot ini-

tiate the mRNA’s translation. Thus, the concentration of a

metabolite can regulate the expression of genes required for

its synthesis. The X-ray structure of the 80-nt TPP-binding

domain from the E. coli TPP-sensing riboswitch, deter-

mined by Breaker and Dinshaw Patel, reveals an intri-

cately folded RNA that binds TPP in an extended confor-

mation (Fig. 31-46b).

Over 20 classes of riboswitches have as yet been identi-

fied, including those that regulate the expression of en-

zymes involved in the metabolism of coenzyme B

(Fig.

25-21), riboflavin (Fig. 16-8), S-adenosylmethionine (SAM;

Fig. 26-18), lysine, and adenine. In general, they consist of

two components, an aptamer that binds an effector and a

so-called expression platform that transduces effector

binding to a change in gene expression. In some cases, the

1300 Chapter 31. Transcription

Figure 31-46 Structure of the TPP-sensing riboswitch from

E. coli. (a) The predicted secondary structure of a 165-residue

segment at the 5¿ end of the thiM gene is shown in the absence

(left) and presence (right) of TPP. The TPP-binding conformation

masks the Shine–Dalgarno sequence (orange) required by the

ribosome to initiate translation at the AUG start codon (red) just

downstream. [After Winkler,W., Nahvi,A., and Breaker, R.R.,

Nature 419, 952 (2002).] (b) The X-ray structure of the

riboswitch’s 90-nt TPP-sensing domain. The RNA is drawn in

CUGA

GACU

CCUUUUCC

AAG

GACU

TPP

Ribosome

125

3´

5´

TPP

(a)

(b)

cartoon form with its sugar–phosphate backbone represented

by an orange rod and its bases represented by paddles with

C green, N blue, and O red. The TPP is drawn in space-filling

form with C cyan, N blue, O red, and S yellow. Mg

2⫹

ions are

represented by lavender spheres. [Based on an X-ray structure

by Ronald Breaker,Yale University, and Dinshaw Patel,

Memorial Sloan-Kettering Cancer Center, New York, New York.

PDBid 2GDI.]

JWCL281_c31_1260-1337.qxd 8/11/10 9:48 PM Page 1300