Voet D., Voet Ju.G. Biochemistry

Подождите немного. Документ загружается.

D. The SOS Response

Agents that damage DNA, such as UV radiation, alkylat-

ing agents, and cross-linking agents, induce a complex sys-

tem of cellular changes in E. coli known as the SOS re-

sponse. E. coli so treated cease dividing and increase their

capacity to repair damaged DNA.

a. LexA Protein Represses the SOS Response

Clues as to the nature of the SOS response were provided

by the observations that E. coli with mutant recA or lexA

genes have their SOS response permanently switched on.

RecA, a 353-residue protein that coats DNA as a multimeric

helical filament, plays a central role, as we shall see, in homol-

ogous recombination (Section 30-6Ab). When E. coli are ex-

posed to agents that damage DNA or inhibit DNA replica-

tion,their RecA specifically mediates the proteolytic cleavage

of LexA (202 residues) between its Asp 84 and Gly 85. RecA

is activated to do so on binding to ssDNA (it was initially as-

sumed that RecA catalyzes the proteolysis of LexA but sub-

sequent experiments by John Little indicate that activated

RecA stimulates LexA to cleave itself).Further investigations

indicated that LexA functions as a repressor of 43 genes that

participate in DNA repair and the control of cell division, in-

cluding recA, lexA, uvrA, and uvrB. DNA sequence analyses

of the LexA-repressible genes revealed that they are all pre-

ceded by a homologous 20-nt sequence, the so-called SOS

box, that has the palindromic symmetry characteristic of op-

erators (control sites to which repressors bind so as to inter-

fere with transcriptional initiation by RNA polymerase; Sec-

tion 5-4A). Indeed, LexA has been shown to specifically bind

the SOS boxes of recA and lexA.

The preceding observations suggest a model for the reg-

ulation of the SOS response (Fig. 30-62). During normal

Section 30-5. Repair of DNA 1221

Figure 30-62 Regulation of the SOS response in E. coli. In a

cell with undamaged DNA (above), LexA largely represses the

synthesis of LexA, RecA, UvrA, UvrB, and other proteins

involved in the SOS response. When there has been extensive

Induced state

Damaged DNA

Operator

mRNA

lexA recA uvrA

UvrA

RecA

Activated RecA

Cleaved LexA

UvrB

uvrB

LexA

Operator

To other genes

controlled by LexA

LexA binds to operators,

repressing the expression

of genes involved in the

SOS response

mRNA

ribosome

Uninduced state

Undamaged DNA

lexA recA uvrA uvrB

LexA

DNA damage (below), RecA is activated by binding to the

resulting single-stranded DNA to stimulate LexA self-cleavage.

The consequent synthesis of the SOS proteins results in the

repair of the DNA damage.

JWCL281_c30_1173-1259.qxd 8/10/10 9:12 PM Page 1221

growth, LexA largely represses the expression of the SOS

genes, including the lexA gene, by binding to their SOS

boxes so as to inhibit RNA polymerase from initiating the

transcription of these genes. When DNA damage has been

sufficient to produce postreplication gaps, however, this ss-

DNA binds to RecA so as to stimulate LexA cleavage.The

LexA-repressible genes are consequently released from re-

pression and direct the synthesis of SOS proteins including

that of LexA (although this repressor continues to be

cleaved through the influence of RecA). When the DNA

lesions have been eliminated, RecA ceases stimulating

LexA’s autoproteolysis. The newly synthesized LexA can

then function as a repressor, which permits the cell to re-

turn to normality.

b. SOS Repair Is Error Prone

The E. coli Pol III holoenzyme is unable to replicate

through a variety of lesions such as AP sites and thymine

dimers. On encountering such lesions, the replisome stalls

and disassembles by releasing its Pol III cores, a process

that is called replication fork “collapse.” Cells have two

general modes for restoring collapsed replication forks,

recombination repair and SOS repair. Recombination repair

circumvents the damaged template by using a homologous

chromosome as its template DNA in a process known as

homologous recombination, which also functions to gener-

ate genetic diversity. Hence we shall postpone our discussion

of recombination repair until after our consideration of

homologous recombination in Section 30-6A. In the follow-

ing paragraphs we discuss SOS repair.

In SOS repair, the Pol III core lost from the collapsed

replication fork is replaced by one of two so-called bypass

DNA polymerases, whose synthesis is induced by the SOS

response: DNA polymerase IV (Pol IV, the 336-residue

product of the dinB gene) or DNA polymerase V [Pol V;

the heterotrimeric product of the umuD and umuC genes,

UmuDⴕ

C (umu for UV mutagenesis), where UmuD¿ is

produced by the RecA-assisted self-cleavage of the 139-

residue UmuD to remove its N-terminal 24 residues, and

UmuC consists of 422 residues]. Both of these enzymes are

Y-family DNA polymerases, all of whose members lack

3¿S5¿ proofreading exonuclease activity and replicate un-

damaged DNA with poor fidelity and low processivity and

hence are also known as error-prone DNA polymerases.

Translesion synthesis (TLS) by Pol V, which was charac-

terized in large part by O’Donnell and Myron Goodman,

requires the simultaneous presence of the ␤

sliding clamp,

the ␥ complex (clamp loader), and SSB, together with a

RecA filament in complex with the ssDNA arising from

the action of helicase on the dsDNA ahead of the stalled

replication fork.This so-called Pol V mutasome tends to in-

corporate G about half as often as A opposite thymine

dimers and AP sites, with pyrimidines being installed infre-

quently. This process is, of course, highly mutagenic. But

even in replicating undamaged DNA, Pol V is at least 1000-

fold more error prone than is Pol I or Pol III holoenzyme.

However, after synthesizing ⬃7 nt, the Pol V mutasome is

replaced by Pol III holoenzyme, which commences normal

DNA replication after the now bypassed lesion. Pol II, a

TLS participant that accurately replicates DNA, is also in-

duced by the SOS response but it is synthesized well before

Pol V appears (see below). The role of Pol II appears to be

the mediation of error-free TLS, and only if this process

fails is it replaced by Pol V to carry out error-prone TLS.

There are numerous types of DNA lesions besides AP

sites and thymine dimers that interfere with normal DNA

replication. Depending on the type of lesion, Pol IV, which

is also error prone, may instead be recruited to carry out

TLS. With many lesions,TLS may skip over the altered nu-

cleotide, resulting in deletion of one or two bases in the

daughter strand opposite the lesion (yielding a frameshift

mutation, so called because it would change a structural

gene’s reading frame from that point onward; Section 5-

4Bd). Moreover, Pol IV is prone to generating frameshift

mutations even when replicating undamaged DNA.

The Y-family DNA polymerase Dpo4 from the archae-

bacterium Sulfolobus solfataricus P2, a homolog of E. coli

Pol IV and Pol V, misincorporates ⬃1 base per 500 repli-

cated nucleotides. The X-ray structure of a complex of

Dpo4 with a primer–template DNA that had been incu-

bated with ddATP (which is complementary to the tem-

plate base), determined by Wei Yang, reveals the struc-

tural basis for this low fidelity (Fig. 30-63).The 352-residue

protein contains the fingers, palm, and thumb domains

1222 Chapter 30. DNA Replication, Repair, and Recombination

Figure 30-63 X-ray structure of the bypass DNA polymerase

Dpo4 from Sulfolobus solfataricus P2 in complex with a

primer–template DNA and ddADP. The protein is drawn in

ribbon form with its fingers, palm, thumb, and little finger

domains blue, red, green, and purple, respectively. The DNA is

gold with its backbones drawn as ribbons and its bases

represented by rods.The ddADP, which is base-paired to a

template T in the enzyme’s active site, is shown in ball-and-stick

form colored according to atom type (C pink, N blue, O red, and

P magenta). [Courtesy of Wei Yang, NIH, Bethesda, Maryland.

PDBid 1JX4.]

JWCL281_c30_1173-1259.qxd 8/10/10 9:12 PM Page 1222

common to all known DNA polymerases (although their

orders differ in the sequences of the different families of

DNA polymerases) and, in addition, has a C-terminal do-

main unique to Y-family DNA polymerases that has been

dubbed the “little finger” domain. The enzyme, as ex-

pected, has incorporated a ddA residue at the 3¿ end of the

primer and, in addition, binds a ddADP in base-paired

complex to the new template T. The little finger domain

binds in the major groove of the DNA. However, the fin-

gers and thumb domains are small and stubby compared

to those of replicative DNA polymerases such as Klentaq1

(Fig. 30-9) and pol ␦ (Fig. 30-41), and the residues that con-

tact the base pair in the active site are all Gly and Ala

rather than the Phe, Tyr, and Arg that mainly do so in the

replicative DNA polymerases. Moreover, the bound DNA

is entirely in the B form rather than in the A form at the

active site as occurs in many replicative DNA poly-

merases. Since the minor groove is more accessible in A-

DNA than in B-DNA (Section 29-1B), this suggests that

error-prone DNA polymerases have relatively little facil-

ity to monitor the base-pairing fidelity of the incoming nu-

cleotide. This accounts for the ability of error-prone DNA

polymerases to accommodate distorted template DNA as

well as non-Watson–Crick base pairs at their active sites.

SOS repair is an error-prone and hence mutagenic

process. It is therefore a process of last resort that is only

initiated ⬃50 min after SOS induction if the DNA has not

already been repaired by other means. Yet, DNA damage

that normally activates the SOS response is nonmutagenic

in the recA

⫺

E. coli that survive.This is, as we saw, because

bypass DNA polymerases will replicate over a DNA lesion

even when there is no information as to which bases were

originally present. Indeed, most mutations in E. coli arise

from the actions of the SOS repair system, which is there-

fore a testimonial to the proposition that survival with a

chance of loss of function (and the possible gain of new

ones) is advantageous, in the Darwinian sense, over death,

although only a small fraction of cells actually survive this

process. It has therefore been suggested that, under condi-

tions of environmental stress, the SOS system functions to

increase the rate of mutation so as to increase the rate at

which the E. coli adapt to the new conditions. Finally, it

should be noted that the eukaryotic pols ␩, ␫, and ␬, all Y-

family members, and pol ␨, an X-family member, are impli-

cated in TLS and that pol ␩, the product of the XPV gene,

is defective in the XPV form of xeroderma pigmentosum

(Section 30-5Bb).

E. Double-Strand Break Repair

Double-strand breaks (DSBs) in DNA are produced when

a replication fork encounters a nick and by the reactive

oxygen species (ROS) by-products of oxidative metabo-

lism and ionizing radiation (which also produces ROS). In

fact, around 5 to 10% of dividing cells in culture exhibit at

least one chromosome break at any given time. Moreover,

DSBs are normal intermediates in certain specialized cel-

lular processes such as recombination during meiosis (Sec-

tion 1-4Ab) and V(D)J recombination in lymphoid cells,

which helps generate the vast diversity of antigen-binding

sites in antibodies and T-cell receptors (Section 35-2C).

Unrepaired or misrepaired DSBs can be lethal to cells or

cause chromosomal aberrations that may lead to cancer.

Hence, the efficient repair of DSBs is essential for cell via-

bility and genomic integrity.

Cells have two general modes to repair DSBs: recombi-

nation repair, which only occurs during the late S and G

phases of the cell cycle (when sister chromatids are present

to serve as templates), and nonhomologous end-joining

(NHEJ), which functions throughout the cell cycle. Here

we discuss NHEJ, a process which, as its name implies, di-

rectly rejoins DSBs. The recombination repair of DSBs is

discussed in Section 30-6Ag.

In NHEJ, the broken ends of the DSB must be aligned,

its frayed ends trimmed and/or filled in, and their strands

ligated.The core NHEJ machinery in eukaryotes includes

the DNA end-binding protein Ku (a heterodimer of ho-

mologous 70- and 83-kD subunits, Ku70 and Ku80), DNA

ligase IV, and the accessory protein Xrcc4. Ku, an abun-

dant nuclear protein, binds to a DSB, whether blunt or

with an overhang, and hence appears to be the cell’s pri-

mary DSB sensor. The X-ray structure of Ku in complex

with a 14-bp DNA, determined by Jonathan Goldberg, re-

veals that the protein cradles the dsDNA segment along

its entire length and encircles its central ⬃3 bp segment

(Fig. 30-64). The protein ring is also present in the closely

Section 30-5. Repair of DNA 1223

Figure 30-64 X-ray structure of human Ku protein in complex

with DNA containing 14 bp. The subunits of Ku70 (red helices

and yellow strands) and Ku80 (blue helices and green strands) are

viewed along the pseudo-2-fold axis relating them.The DNA,

viewed with its DSB pointing upward, is drawn in space-filling

form with its sugar–phosphate backbone dark gray and its base

pairs light gray. Note that the DNA is surrounded by a ring of

protein. [Courtesy of John Tainer,The Scripps Research Institute,

La Jolla, California. Based on an X-ray structure by Jonathan

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

New York. PDBid 1JEY.]

JWCL281_c30_1173-1259.qxd 8/10/10 9:12 PM Page 1223

similar X-ray structure of Ku alone, thereby explaining

why Ku that is bound to a dsDNA, which is then circular-

ized, becomes permanently associated with it. Ku makes

no specific contacts with the DNA’s bases and few with its

sugar–phosphate backbone, but instead fits snugly into

the DNA’s major and minor grooves so as to precisely ori-

ent it.

Ku–DNA complexes have been shown to dimerize so

as to align the members of a DSB, both blunt ended and

with short (1–4 bp) complementary single strands, for lig-

ation as is diagrammed in Fig. 30-65. The DNA ends are

exposed along one face of each Ku–DNA complex, pre-

sumably making them accessible to polymerases that fill

in gaps and to nucleases that trim excess and inappropri-

ate ends preparatory for ligation by DNA ligase IV in

complex with Xrcc4. Nucleotide trimming, which of

course generates mutations, appears to be carried out in

an ATP-dependent manner by the evolutionarily con-

served Mre11 complex, which consists of two Mre11 nu-

clease subunits and two Rad50 ATPase subunits. Ku is

eventually released from the rejoined DNA, perhaps by

proteolytic cleavage.

The reason that the mutations generated by NHEJ are

usually not unacceptably deleterious is that only a small

fraction of the mammalian genome is expressed (Section

34-2A). In fact, the genome in a somatic cell of a 70-year-

old human typically contains ⬃2000 “scars” caused by

NHEJ.

F. Identification of Carcinogens

Many forms of cancer are known to be caused by exposure

to certain chemical agents that are therefore known as car-

cinogens. It has been estimated that as much as 80% of hu-

man cancer arises in this fashion. There is considerable ev-

idence that the primary event in carcinogenesis is often

damage to DNA (carcinogenesis is discussed in Section 34-

4C). Carcinogens are consequently also likely to induce the

SOS response in bacteria and thus act as indirect muta-

genic agents. In fact, there is a high correlation between

carcinogenesis and mutagenesis (recall, e.g., the progress of

xeroderma pigmentosum; Section 30-5Bb).

There are presently over 80,000 man-made chemicals of

commercial importance and ⬃1000 new ones are intro-

duced each year.The standard animal tests for carcinogen-

esis, exposing rats or mice to high levels of the suspected

carcinogen and checking for cancer, are expensive and re-

quire ⬃3 years to complete.Thus, relatively few substances

have been tested in this manner.

a. The Ames Test Assays for Probable

Carcinogenicity

Bruce Ames devised a rapid and effective bacterial as-

say for carcinogenicity that is based on the high correla-

tion between carcinogenesis and mutagenesis. He con-

structed special tester strains of Salmonella typhimurium

that are his

⫺

(cannot synthesize histidine so that they are

unable to grow in its absence), have cell envelopes that

lack the lipopolysaccharide coating that renders normal

Salmonella impermeable to many substances (Section 11-

3Bc), and have inactivated excision repair systems. Muta-

genesis in these tester strains is indicated by their rever-

sion to the his

⫹

phenotype.

In the Ames test, ⬃10

tester strain bacteria are spread

on a culture plate that contains only a small amount of his-

tidine to permit the bacteria to initially grow and mutate.

Usually a mixture of several his

⫺

strains is used so that mu-

tations due to both base changes and nucleotide insertions

or deletions can be detected. A mutagen placed in the cul-

ture medium causes some of these his

⫺

bacteria to revert to

the his

⫹

phenotype, which is detected by their growth into

visible colonies after 2 days at 37°C (Fig. 30-66). The muta-

genicity of a substance is scored as the number of such

colonies less the few spontaneously revertant colonies that

occur in the absence of the mutagen.

1224 Chapter 30. DNA Replication, Repair, and Recombination

Processing enzymes

fill gap (not shown)

Xrcc4 – LigIV are recruited

Ku bridges ends

Xrcc4 – LigIV

DNA strands are repaired

Ku binds to ends

DNA with double-strand break

heterodimers

Figure 30-65 Schematic diagram of nonhomologous end-

joining (NHEJ). The left dsDNA fragment is missing a base and

the right fragment is blocked by a nonligatable group (filled

black circle).The two Ku heterodimers are drawn in two shades

of yellow and the Xrcc4–DNA ligase IV complexes are drawn in

two shades of blue. The newly repaired links in the DNA are

represented by pink circles. [After Jones, J.M., Gellert, M., and

Yang, W., Structure 9, 881 (2001).]

JWCL281_c30_1173-1259.qxd 8/10/10 9:12 PM Page 1224

Many noncarcinogens are converted to carcinogens in

the liver or in other tissues via a variety of detoxification

reactions (e.g., those catalyzed by the cytochromes P450;

Section 15-4Bc).A small amount of rat liver homogenate is

therefore included in the Ames test medium in an effort to

approximate the effects of mammalian metabolism.

b. Both Man-Made and Naturally Occurring

Substances Can Be Carcinogenic

There is an ⬃80% correspondence between the com-

pounds determined to be carcinogenic by animal tests

and those found to be mutagenic by the Ames test. Dose–

response curves, which are generated by testing a given

compound at a number of concentrations, are almost al-

ways linear and extrapolate back to zero, indicating that

there is no threshold concentration for mutagenesis. Sev-

eral compounds to which humans have been extensively

exposed that were found to be mutagenic by the Ames

test were later found to be carcinogenic in animal tests.

These include tris(2,3-dibromopropyl)phosphate, which

was used as a flame retardant on children’s sleepwear in

the mid-1970s and can be absorbed through the skin; and

furylfuramide, which was used in Japan in the 1960s and

1970s as an antibacterial additive in many prepared

foods (and had passed two animal tests before it was

found to be mutagenic). Carcinogens are not confined to

man-made compounds but also occur in nature. For ex-

ample, carcinogens are contained in many plants that are

common in the human diet, including alfalfa sprouts.

Aflatoxin B

one of the most potent carcinogens known, is produced

by fungi that grow on peanuts and corn. Charred or

browned food, such as occurs on broiled meats and

toasted bread, contains a variety of DNA-damaging

agents. Thus, with respect to carcinogenesis, as Ames has

written, “Nature is not benign.”

6 RECOMBINATION AND MOBILE

GENETIC ELEMENTS

The chromosome is not just a simple repository of genetic

information. If this were so, the unit of mutation would

have to be an entire chromosome rather than a gene be-

cause there would be no means of separating a mutated

gene from the other genes of the same chromosome. Chro-

mosomes would therefore accumulate deleterious muta-

tions until they became nonviable.

It has been known from some of the earliest genetic

studies that pairs of allelic genes may exchange chromoso-

mal locations by a process known as genetic recombination

(Section 1-4Cb). Mutated genes can thereby be individu-

ally tested, since their propagation is then not absolutely

dependent on the propagation of the genes with which they

had been previously associated. In this section, we consider

the mechanisms by which genetic elements can move, both

between chromosomes and within them.

A. Homologous Recombination

Homologous recombination (also called general recombi-

nation) is defined as the exchange of homologous segments

between two DNA molecules. Both genetic and cytological

studies have long indicated that such a crossing-over

process occurs in higher organisms during meiosis (Fig.1-27).

Bacteria, which are normally haploid, likewise have elabo-

rate mechanisms for the interchange of genetic informa-

tion. They can acquire foreign DNA through transforma-

tion (Section 5-2A), through a process called conjugation

(mating) in which DNA is directly transferred from one

cell to another via a cytoplasmic bridge (Section 31-1Ac),

and via transduction in which a defective bacteriophage

that has erroneously acquired a segment of bacterial DNA

rather than the viral chromosome transfers this DNA to

another bacterial cell. In all of these processes, the foreign

DNA is installed in the recipient’s chromosome or plasmid

OCH

Aflatoxin B

Section 30-6. Recombination and Mobile Genetic Elements 1225

Figure 30-66 The Ames test for mutagenesis. A filter paper

disk containing a mutagen, in this case the alkylating agent ethyl

methanesulfonate, is centered on a culture plate containing his

⫺

tester strains of Salmonella typhimurium in a medium that

initially contained only a small amount histidine. A dense halo of

revertant bacterial colonies appears around the disk from which

the mutagen diffused.The larger colonies distributed about the

culture plate are spontaneous revertants.The bacteria near the

disk have been killed by the toxic mutagen’s high concentration.

[Courtesy of Raymond Devoret, Institut Curie, Orsay, France.]

JWCL281_c30_1173-1259.qxd 8/10/10 9:12 PM Page 1225

A B

(l)

–

(a)

(

( )

(

(g)

through homologous recombination (to be propagated, a

DNA segment must be part of a replicon; that is, be associ-

ated with a replication origin such as occurs in a chromo-

some, a plasmid, or a virus).

a. Recombination Occurs via a Crossed-Over

Intermediate

The prototypical model for homologous recombination

(Fig. 30-67) was proposed by Robin Holliday in 1964 on the

basis of genetic studies on fungi.The corresponding strands

of two aligned homologous DNA duplexes are nicked, and

the nicked strands cross over to pair with the nearly com-

plementary strands of the homologous duplex after which

the nicks are sealed (Fig. 30-67a–e), thereby yielding a four-

way junction known as a Holliday junction (Fig. 30-67e). A

Holliday junction has, in fact, been observed in the X-ray

structure of d(CCGGTACCGG), determined Shing Ho

(Fig. 30-68), in which, perhaps unexpectedly, all the bases

form normal Watson–Crick base pairs without any appar-

ent strain.The crossover point can move in either direction,

often thousands of nucleotides, in a process known as

branch migration (Fig. 30-67e, f ) in which the four strands

exchange base-pairing partners.

A Holliday junction can be resolved into two duplex

DNAs in two equally probable ways (Fig. 30-67g–l ):

1. The cleavage of the strands that did not cross over

(right branch of Fig. 30-67j–l) exchanges the ends of the

original duplexes to form, after nick sealing, the traditional

recombinant DNA (Fig. 1-27b).

2. The cleavage of the strands that crossed over (left

branch of Fig. 30-67j–l) exchanges a pair of homologous

single-stranded segments.

The recombination of circular duplex DNAs results in

the types of structures diagrammed in Fig. 30-69. Electron

microscopic evidence for the existence of the postulated

1226 Chapter 30. DNA Replication, Repair, and Recombination

Figure 30-67 The Holliday model of homologous recombination between homologous DNA duplexes.

See the Animated Figures

JWCL281_c30_1173-1259.qxd 8/10/10 9:12 PM Page 1226

Section 30-6. Recombination and Mobile Genetic Elements 1227

Figure 30-68 X-ray structure of the self-complementary decameric DNA d(CCGGTACCGG).

(a) The secondary structure of the four-stranded Holliday junction formed by this sequence in which

the four strands, A, B, C, and D, are differently colored, their nucleotides are numbered 1 to 10 from

their 5¿ to 3¿ termini, and Watson–Crick base-pairing interactions are represented by black dashes.

The 2-fold axis relating the two helices of this so-called stacked-X conformation is represented by the

black lenticular symbol. (b) The observed three-dimensional structure of the Holliday junction, as

viewed along its 2-fold axis, in which the oligonucleotides are represented in stick form with their

backbones traced by ribbons, all colored as in Part a. With the exception of the backbones of strands

B and D at the crossovers, the two arms of this structure each form an undistorted B-DNA helix,

including the stacking of the base pairs flanking the crossovers. The two helices are inclined to each

other by 41°. Note that Fig. 30-67g is a schematic representation of the stacked-X conformation as

viewed perpendicular to both helices (from the side in this drawing and hence having the projected

appearance of the letter X).A Holliday junction can also assume a so-called open-X conformation,

which is represented by Fig. 30-67i. [Courtesy of Shing Ho, Oregon State University. PDBid 1DCW.]

Figure 30-69 Homologous recombination between two

circular DNA duplexes. This process can result either in two

circles of the original sizes or in a single composite circle.

(a)

(b)

(c)

(d)

(e)

(f)

(g)

(b)

(a)

JWCL281_c30_1173-1259.qxd 9/2/10 9:06 AM Page 1227

“figure-8” structures is shown in Fig. 30-70a. These figure-8

structures were shown not to be just twisted circles by cut-

ting them with a restriction endonuclease to yield chi struc-

tures (after their resemblance to the Greek letter ␹) such

as that pictured in Fig. 30-70b.

b. Homologous Recombination in E. coli Is

Catalyzed by RecA

The observation that recA

⫺

E. coli have a 10

-fold lower

recombination rate than the wild-type indicates that RecA

protein has an important function in recombination. In-

deed, RecA greatly increases the rate at which comple-

mentary strands renature in vitro. This versatile protein

(recall it also stimulates the autoproteolysis of LexA to

trigger the SOS response and is an essential participant in

the translesion synthesis of DNA; Section 30-5D) polymer-

izes cooperatively without regard to base sequence on

ssDNA or on dsDNA that has a single-stranded gap. The

resulting filaments, which may contain up to several thou-

sand RecA monomers, specifically bind the homologous

dsDNA, and, in an ATP-dependent reaction, catalyze

strand exchange.

EM studies by Edward Egelman revealed that RecA fil-

aments bound to ssDNA or dsDNA form a right-handed

helix with ⬃6.2 RecA monomers per turn and a pitch (rise

per turn) of 95 Å.The DNA in these filaments binds to the

protein with 3 nt (or bp) per RecA monomer and hence is

underwound with ⬃18.5 nt (or bp) per turn (vs 10 bp per

turn for cannonical B-DNA).

The formation of RecA–DNA filaments is highly coop-

erative; it requires five or six RecA protomers to form a

stable assembly. Consequently, attempts to crystallize

RecA–DNA filaments over many years were unsuccessful.

Nikola Pavletich ingeniously solved this conundrum by

linking five or six E. coli RecA genes (each corresponding

to residues 1–335 of this 353-residue protein) in tandem via

14-residue linkers and mutating the first and last RecA so

as to prevent them from forming longer filaments. These

fusion proteins, which had DNA-dependent ATPase and

strand-exchange activities comparable to that of

monomeric RecA, formed crystals containing both ssDNA

and dsDNA.

The X-ray structure of the RecA

–(ADP–AlF

⫺

)

–

(dT)

–(dA)

complex (Fig. 30-71;ADP–AlF

⫺

is a nonhy-

1228 Chapter 30. DNA Replication, Repair, and Recombination

Figure 30-70 Electron micrographs of intermediates in the

homologous recombination of two plasmids. (a) A figure-8

structure. This corresponds to Fig. 30-69d.(b) A chi structure that

results from the treatment of a figure-8 structure with a

restriction endonuclease. Note the thinner single-stranded

connections in the crossover region. [Courtesy of Huntington

Potter, University of South Florida, and David Dressler, Oxford

University, U.K.]

Figure 30-71 X-ray structure of the RecA

–(ADP–AlF

⫺

)

–

(dT)

–(dA)

complex viewed with its filament axis vertical. The

RecA units RecA

(the N-terminal unit) through RecA

are

colored green, cyan, magenta, and gray, respectively, with the

C-terminal unit, RecA

, colored in rainbow order from its

N-terminus (blue) to its C-terminus (red).The DNA and

ADP–AlF

⫺

are drawn in space-filling form with DNA C gray,

ADP C green, N blue, O red, P orange, F light blue, and Al

purple. [Based on an X-ray structure by Nikola Pavletich,

Memorial Sloan-Kettering Cancer Center, New York,

New York. PDBid 3CMX.]

(a)

(b)

JWCL281_c30_1173-1259.qxd 9/2/10 9:06 AM Page 1228

drolyzable ATP analog) exhibits a straight filament axis

with overall helical parameters that are closely similar to

those derived from EM studies. Each RecA unit consists of

a largely helical 30-residue N-terminal segment, a 240-

residue ␣/␤ ATPase core, and a 64-residue C-terminal glob-

ular domain. The linkers connecting adjacent RecA units

are disordered. Each RecA unit makes extensive contacts

with its nearest neighbors so as to form a filament with a

deep helical groove that exposes the DNA bound in its in-

terior (Fig. 30-71 is viewed looking into this groove).

The two DNA strands, which lie close to the filament

axis, form a complete set of Watson–Crick base pairs. How-

ever, rather than being smoothly stretched out, as had been

expected, the dsDNA assumes an irregular conformation in

which each 3-bp segment that is bound to a RecA unit

closely resembles B-DNA with the steps between succes-

sive base pairs in this triplet having an axial rise of ⬃3.4 Å

and a helical twist of ⬃30° (vs 3.4 Å and 36° for canonical B-

DNA; Table 29-1). In contrast, the step between successive

base pair triplets has an axial rise of 8.4 Å and a helical twist

of ⫺4°, thereby forming a 5-Å-high gap between successive

triplets that is partially filled by the side chain of the con-

served Ile 199.The sugar–phosphate backbone of the DNA

strand furthest from the viewer in Fig. 30-71 [the (dT)

]

makes extensive contacts with RecA. In contrast, the other

strand [the (dA)

] makes few contacts with the protein; it is

held in place almost entirely by base pairing with the first

strand. The ADP–AlF

⫺

is sandwiched between adjacent

␣/␤ ATPase cores, where it is completely buried.

The X-ray structure of the ssDNA-containing

RecA

–(ADP–AlF

⫺

)

–(dT)

complex closely resembles

that of the foregoing dsDNA-containing complex but with

the absence of the DNA strand closest to the viewer in Fig.

30-71. Thus, each RecA unit binds a (dT)

segment that is

held in a B-DNA-like conformation with successive (dT)

segments separated by a 7.8-Å axial rise.

How does RecA mediate DNA strand exchange between

single-stranded and duplex DNAs? On encountering a

dsDNA with a strand that is complementary to its bound

ssDNA, RecA partially unwinds the duplex and, in a reac-

tion driven by RecA-catalyzed ATP hydrolysis, exchanges

the ssDNA with the corresponding strand on the duplex.

This process tolerates only a limited degree of mispairing and

requires that one of the participating DNA strands have a free

end. The assimilation (exchange) of a single-stranded circle

with a strand on a linear duplex (Fig. 30-72) cannot proceed

past the 3¿ end of a highly mismatched segment in the com-

plementary strand. The invasion of the single strand must

Section 30-6. Recombination and Mobile Genetic Elements 1229

No assimilation of noncomplementary DNA

No assimilation

Assimilation stops at

heterologous

segment

Assimilation of 3' end of homologous DNA

Homologous

DNA

Heterologous

DNA

Figure 30-72 The RecA-catalyzed assimilation of a

single-stranded circle by a dsDNA can occur only if the dsDNA

has a 3ⴕ end that can base-pair with the circle (red strand).

Strand assimilation cannot proceed through a noncomplemen-

tary segment (purple and orange strands).

JWCL281_c30_1173-1259.qxd 8/10/10 9:12 PM Page 1229

therefore begin at its 5¿ end. A model for the consequent

branch migration process is diagrammed in Fig. 30-73. Of

course, two such strand exchange processes must occur si-

multaneously in a Holliday junction (Figs. 30-67 and 30-69).

The above structures suggest that the fidelity of homol-

ogous recombination arises from the B-DNA-like confor-

mation that RecA imposes on the otherwise flexible bound

ssDNA strand, which would exclude non-Watson–Crick

base pairs. Strand exchange, of course, requires the separa-

tion of the two strands of the incoming dsDNA to permit

one of its strands to sample base pairing with the ssDNA

substrate. The above structures suggest that this is facili-

tated by the disruption of base stacking between base pair

triplets in the RecA–DNA complex. However, the struc-

ture of the triple helical DNA intermediate in the strand

exchange reaction is, as yet, unknown.

c. Eukaryotes Have RecA-Like Proteins

Yeast RAD51 (339 residues) functions in the ATP-

dependent repair and recombination of DNA in much the

same way as does the 30% homologous E. coli RecA pro-

tein. The electron micrograph–based image reconstruction

of RAD51 in complex with double-stranded DNA is nearly

identical to that of RecA at low resolution: Both com-

plexes form helical filaments in which the DNA has an

⬃5.1-Å rise per base pair and 18.6 bp per turn. Since

RAD51 homologs occur in chickens, mice,and humans, it is

very likely that such filaments universally mediate DNA

repair and recombination.

d. RecBCD Initiates Recombination by Making

Single-Strand Nicks

The single-strand nicks to which RecA binds are made

by the RecBCD protein, the 330-kD heterotrimeric prod-

uct of the SOS genes recB, recC, and recD. RecB is both a

3¿S5¿ helicase and a nuclease, whereas RecD is a 5¿S3¿

helicase. The formation of a RecA binding site begins with

RecBCD binding to the end of a dsDNA and then unwind-

ing it via its two ATP-driven helicase functions (Fig. 30-74).

As it does so, RecB nucleolytically degrades the unwound

single strands behind it, with the 3¿-ending strand being

cleaved more often and hence broken down to smaller

fragments than the 5¿-ending strand. However, on RecC

encountering the sequence GCTGGTGG from its 3¿ end

(the so-called Chi sequence, which occurs about every

⬃5 kb in the E. coli genome), the enzyme pauses and

ceases its cleavage of the 3¿-ending strand but increases the

rate at which it cleaves the 5¿-ending strand. RecBCD then

helps load RecA onto the 3¿-ending strand before dissoci-

ating from the DNA.

1230 Chapter 30. DNA Replication, Repair, and Recombination

Figure 30-73 Hypothetical model for the RecA-mediated

strand exchange reaction. Homologous DNA molecules are

paired in advance of strand exchange in a three-stranded helix.

The ATP-driven rotation of the RecA filament about its helix

Figure 30-74 The generation of a 3ⴕ-ending single-strand

DNA segment by RecBCD to initiate recombination. (1)

RecBCD binds to a free end of a dsDNA and, in an ATP-driven

process, advances along the helix, unwinding the DNA and

degrading the resulting single strands behind it, with the

3¿-ending strand cleaved more often than the 5¿-ending strand.

(2) When RecBCD encounters a properly oriented Chi

sequence, it binds it and thus stops cleaving the 3¿-ending strand

but increases the frequency at which it cleaves the 5¿-ending

strand.This generates the potentially invasive 3¿-ending strand

segment to which RecA binds.

Spooling

out

Spooling

5′

3′

Three-stranded DNA

RecBCD

RecA

ATP

ADP + P

Chi

axis would cause duplex DNA to be “spooled in” to the filament,

right to left as drawn. [After West, S.C., Annu. Rev. Biochem. 61,

617 (1992).]

JWCL281_c30_1173-1259.qxd 8/10/10 9:12 PM Page 1230