85 222 Experimental Support for the RecBCD Pathway

wea25324_ch22_709-731.indd Page 712
712
12/20/10
4:42 PM user-f469
/Volume/204/MHDQ268/wea25324_disk1of1/0073525324/wea25324_pagefile
Chapter 22 / Homologous Recombination
3
1
(a) Cuts 1 and 2
2
4
(b) Cuts 3 and 4
+
+
Noncrossover (heteroduplex, or patch)
recombinants
Crossover (splice) recombinants
Figure 22.3 Resolution of a Holliday junction. The Holliday junction pictured at top can be resolved in two different ways, as indicated by the
numbered arrows. (a) Cuts 1 and 2 yield two duplex DNAs with patches of heteroduplex whose length corresponds to the distance covered by
branch migration before resolution. (b) Cuts 3 and 4 yield crossover recombinant molecules with the two parts joined by a staggered splice.
22.2 Experimental Support
for the RecBCD Pathway
Now that we have seen a brief overview of the RecBCD
pathway, let us look at the experimental evidence that supports this important bacterial recombination mechanism.
RecA
We have encountered RecA before, in our discussion of
induction of the l phage (Chapter 8). Indeed, this is a protein of many functions, but it was first discovered in the
context of recombination, and that is how it got its name.
In 1965, Alvin Clark and Ann Dee Margulies isolated two
E. coli mutants that could accept F plasmids, but could not
integrate their DNA permanently by recombination. These
mutants were also highly sensitive to ultraviolet light, presumably because they were defective in recombination repair of UV damage (Chapter 20). Characterizing these
mutants led ultimately to the discovery of two key proteins
in the RecBCD pathway: RecA and RecBCD.
The recA gene had been cloned and overexpressed, so
abundant RecA protein was available for study. It is a
38-kD protein that can promote a variety of strand
exchange reactions in vitro. Using such in vitro assays,
Charles Radding and colleagues discerned the three stages
of participation of RecA in strand exchange:
1. Presynapsis, in which RecA coats the singlestranded DNA.
2. Synapsis, or alignment of complementary sequences in
the single-stranded and double-stranded DNAs that
will participate in strand exchange.
3. Postsynapsis, or strand exchange, in which the singlestranded DNA replaces the (1) strand in the double-
stranded DNA to form a new double helix. An
intermediate in this process is a joint molecule in
which strand exchange has begun and the two DNAs
are intertwined with each other.
Presynapsis The best evidence for the association between
RecA and single-stranded DNA is visual. Radding and colleagues constructed a linear, double-stranded phage DNA
with single-stranded tails, incubated this DNA with RecA,
spread the complex on an electron microscope grid and
photographed it. Figure 22.4a shows that RecA bound
preferentially to the single-stranded ends, forming proteincoated DNA filaments, but leaving the double-stranded
DNA in the middle uncoated. These workers also incubated single-stranded circular M13 phage DNA with RecA
and subjected these complexes to the same procedure.
Figure 22.4b shows extended DNA circles coated uniformly with RecA. The magnification in panels (a) and (b)
is the same, so the thickness of the circular fiber in panel (b),
compared with the naked DNA and RecA–DNA complex
in panel (a), demonstrates clearly that these circular DNAs
are indeed coated with RecA.
Single-strand DNA-binding protein (SSB) also helps to
form the coated DNA fiber in the presynapsis process.
Radding and colleagues showed that the appearances of
DNA–protein complexes formed by mixing single-stranded
M13 phage DNA with SSB alone, or with SSB plus RecA
are clearly different, and the DNA–protein complexes with
both SSB and RecA strongly resemble those with RecA
alone in Figure 22.4. Furthermore, Radding and colleagues
showed that SSB accelerates the formation of the coated
DNA. In the presence of SSB plus RecA, formation of extended circular filaments was complete after only 10 min.
By contrast, the process had barely begun after 10 min
when SSB was absent.
wea25324_ch22_709-731.indd Page 713
12/20/10
4:42 PM user-f469
/Volume/204/MHDQ268/wea25324_disk1of1/0073525324/wea25324_pagefile
22.2 Experimental Support for the RecBCD Pathway
(a)
Figure 22.4 Binding of RecA to single-stranded DNA. Radding and
colleagues prepared (a) a linear double-stranded DNA with singlestranded ends and (b) a circular single-stranded phage DNA. Then
they added RecA, allowed time for a complex to form, spread the
complexes on coated electron microscope grids, and photographed
Because RecA can coat a single-stranded DNA by itself,
what is the role of SSB? It seems to be required to melt
secondary structure (hairpins) in the single-stranded DNA
that would otherwise impede the expansion of the RecAcoated fiber. Evidence for this notion comes from several
sources. Radding and colleagues assayed for strand exchange when RecA was incubated with single-stranded
DNA at low and high concentrations of MgCl2. Low
MgCl2 concentrations destabilize DNA secondary structure, but high MgCl2 concentrations stabilize it. In this experiment, SSB was required under high, but not low, MgCl2
concentration conditions, which is the result we expect if
SSB is needed to relax DNA secondary structure.
Later in this section, we will see that ATP hydrolysis is
normally required for strand exchange, but I.R. Lehman and
coworkers showed that ATPgS, an unhydrolyzable analog
of ATP, will support a limited amount of strand exchange if
SSB is present. Because ATPgS causes RecA to bind essentially irreversibly to both single- and double-stranded DNA,
Radding and colleagues posed the hypothesis that ATPgS
causes RecA to trap DNA in secondary structure that is
unfavorable for strand exchange. If this is true, then SSB
should be able to override this difficulty by removing secondary structure if it is added to DNA before RecA. As expected, SSB did indeed accelerate strand exchange if it was
added before RecA. This provided more evidence that the
role of SSB is to unwind secondary structure in a singlestranded DNA participating in recombination.
SUMMARY In the presynapsis step of recombina-
tion, RecA coats a single-stranded DNA that is
participating in recombination. SSB accelerates
713
(b)
them. The bar in panel (a) represents 500 nm for both panels. (Source:
Radding, C.M., J. Flory, A. Wu, R. Kahn, C. DasGupta, D. Gonda, M. Bianchi, and
S.S. Tsang, Three phases in homologous pairing: Polymerization of recA protein on
single-stranded DNA, synapsis, and polar strand exchange. Cold Spring Harbor
Symposia of Quantitative Biology 47 (1982) f. 3 f&j, p. 823.)
the recombination process, apparently by melting
secondary structure and preventing RecA from
trapping any secondary structure that would inhibit strand exchange later in the recombination
process.
Synapsis: Alignment of Complementary Sequences We
will see later in this section that RecA stimulates strand
exchange, which involves invasion of a duplex DNA by a
single strand from another DNA. In this process, the invading strand forms a new double helix with one of the strands
of the other duplex. But the step that precedes strand exchange, synapsis, entails a simple alignment of complementary sequences, without the formation of an intertwined
double helix. This process yields a less stable product and
is therefore more difficult to detect than strand exchange.
Nevertheless, Radding and colleagues presented good evidence for synapsis as early as 1980.
As in the presynapsis experiments, electron microscopy
was a key technique in the first demonstration of synapsis.
Radding and coworkers used a favorite pair of substrates
for their synapsis experiments: a single-stranded circular
phage DNA and a double-stranded linear DNA. However,
in this case, the single-stranded circular DNA was G4
phage DNA, and the double-stranded linear DNA was
M13 phage DNA with a 274-bp G4 phage DNA insert
near the middle. Because this target for the single-stranded
G4 phage DNA lay thousands of base pairs from either
end, and nicks were very rare in this DNA, it was unlikely
that true strand exchange could happen. Instead, simple
synapsis of complementary sequences could occur, as illustrated in Figure 22.5.
wea25324_ch22_709-731.indd Page 714
714
12/20/10
4:42 PM user-f469
Chapter 22 / Homologous Recombination
+
ssG4 DNA
Synapsis
Figure 22.5 Synapsis. Synapsis is shown between circular singlestranded G4 phage DNA (red) and linear double-stranded M13(G4)
DNA (M13 phage DNA [blue] with a 274-bp insert of G4 phage DNA
[red]). The synapsis does not involve any intertwining of the linear and
circular DNAs.
To measure synapsis between the two DNAs, Radding
and colleagues mixed the two DNAs in the presence and
absence of RecA and subjected the mixture to electron microscopy. In the presence of RecA, they found a significant
proportion of the DNA molecules undergoing synapsis, as
depicted in Figure 22.6. In most of the aligned molecules,
the length of the aligned regions was appropriate, and the
(a)
/Volume/204/MHDQ268/wea25324_disk1of1/0073525324/wea25324_pagefile
(b)
Figure 22.6 Demonstration of RecA-dependent synapsis in vitro.
Radding and colleagues mixed the circular single-stranded and
linear double-stranded DNAs described in Figure 22.5 with RecA and
then examined the products by electron microscopy. Panels (a–c)
show three different examples of aligned DNA molecules. Below
each electron micrograph is an interpretive diagram, showing the
position of the aligned region within the linear DNA was
correct. Furthermore, synapsis was reduced by 20–40-fold
with two DNAs that did not share a homologous region.
Synapsis also failed in the absence of RecA.
Although the percentage of nicked double-stranded
DNAs was very low in this experiment, it was conceivable
that nicks could create free ends in the linear DNA that
could allow formation of a true, intertwined plectonemic
double helix. If this had happened, the linkage between the
two DNAs would have been stable to temperatures approaching the melting point of DNA. However, the alignments shown in Figure 22.6 were destroyed by heating at
208C below the melting point for 5 min. Thus, the synapsis
observed here does not involve the formation of a basepaired Watson–Crick double helix. Instead, it probably involves a paranemic double helix in which the two aligned
DNA strands are side by side, but not intertwined. Further
support for the notion that nicks are not required for synapsis comes from the finding that supercoiled DNA (unnicked
by definition) works just as well as linear double-stranded
DNA in these experiments.
How much homology is necessary for synapsis to occur?
David Gonda and Radding provided an estimate by showing that a 151-bp homologous region gave just as efficient
synapsis as did 274 bp of homology. But DNAs with just
30 bp of homology gave only background levels of aligned
DNA molecules. Thus, the minimum degree of homology
for efficient synapsis is somewhere between 30 and 151 bp.
(c)
linear double-stranded DNA in blue and the circular single-stranded
DNA in red. Thick red lines denote the zones of synapsis between the
two DNAs in each case. (Source: DasGupta C., T. Shibata, R.P. Cunningham,
and C.M. Radding, The topology of homologous pairing promoted by recA
protein. Cell 22 (Nov 1980 Pt2) f. 9 d–f, p. 443. Reprinted by permission of
Elsevier Science.)
wea25324_ch22_709-731.indd Page 715
12/20/10
4:42 PM user-f469
/Volume/204/MHDQ268/wea25324_disk1of1/0073525324/wea25324_pagefile
22.2 Experimental Support for the RecBCD Pathway
SUMMARY Synapsis occurs when a single-stranded
Table 22.1
DNA finds a homologous region in a doublestranded DNA and aligns with it. No intertwining
of the two DNAs occurs at this point.
Duplex DNA
Postsynapsis: Strand Exchange We have learned that
RecA is required for the first two steps in strand exchange:
presynapsis and synapsis. Now we will see that it is also
required for the last: postsynapsis, or strand exchange itself. Lehman and colleagues measured strand exchange between double-stranded and single-stranded phage DNA by
a filter-binding assay for D-loop formation as follows:
They incubated 3H-labeled duplex P22 phage DNA with
unlabeled single-stranded P22 DNA in the presence or absence of RecA. They used high-salt and low-temperature
conditions that restricted branch migration, which would
have completely assimilated the single-stranded DNA and
eliminated the D-loops. Then they removed protein from
the DNA with a detergent (either Sarkosyl or sodium dodecyl sulfate). Finally, they filtered the mixture through a
nitrocellulose filter. If D-loops formed in the duplex DNA,
the single-stranded D-loop would cause the complex to
bind to the filter, so labeled DNA would be retained. If no
D-loop formed, the unlabeled single-stranded DNA would
stick to the filter, but the labeled duplex DNA would flow
through. The detergent prevented DNA from sticking to
the filter simply because of its association with RecA.
Lehman and colleagues also performed this assay with supercoiled duplex M13 phage DNA and linear M13 DNA.
In both cases, about 50% of the DNA duplexes formed
D-loops, but only in the presence of RecA. Without RecA,
retention of D-looped DNA was less than 1%. Also,
with a nonhomologous single-stranded DNA, retention of
D-looped DNA was only 2%.
To verify that D-loops had actually formed, these workers treated the complexes with S1 nuclease to remove singlestranded DNA, then filtered the product. This greatly
reduced the retention of the labeled DNA on the filter, suggesting that a D-loop was really involved. To be sure, they
directly visualized the D-loops by electron microscopy.
D-loops were clearly visible in both kinds of DNA—linear
and supercoiled. This experiment thus had the added benefit of demonstrating that supercoiled DNA is not required
for strand exchange.
Table 22.1 shows the effects of nucleotides on D-loop
formation, as measured by retention by nitrocellulose filtration. We see that ATP is required for D-loop formation,
and its role cannot be performed by GTP, UTP, or ATPgS.
The fact that ATPgS cannot substitute for ATP in D-loop
formation indicates that ATP hydrolysis is required. In fact,
it appears that ATP hydrolysis permits RecA to dissociate
from DNA, which permits the new base pairing that must
occur in strand exchange.
P22 phage
M13 phage
715
Requirements for D-loop Formation
Reaction
components
D-loops
formed (%)
Complete
2RecA
2ATP
2ATP 1 GTP
2ATP 1 UTP
2ATP 1 ATPgS
Complete
2RecA
2ATP
2ATP 1 GTP
100
,1
,1
,1
,1
,1
100
1
1
2
Thus, this experiment demonstrated that ATP hydrolysis is essential for D-loop formation. RecA, an incredibly
versatile protein, has ATPase activity, which cleaves ATP as
RecA falls off the DNAs to allow the D-loops to form.
SUMMARY RecA and ATP collaborate to promote
strand exchange between a single-stranded and
double-stranded DNA. ATP is necessary to clear
RecA off the synapsing DNAs to make way for formation of double-stranded DNA involving the single
strand and one of the strands of the DNA duplex.
RecBCD
In our discussion of RecA, we have been considering model
reactions involving a single-stranded and a duplex DNA.
The reason, of course, is that RecA requires a singlestranded DNA to initiate strand exchange. But naturally
recombining DNAs are usually both double-stranded, so
how does RecA get the single strand it needs? We have already learned that RecBCD provides it. Two elements intimately involved in this process are Chi sites on the DNA
and the DNA helicase activity of RecBCD. Let us consider
the evidence for these two things.
Chi sites were discovered in genetic experiments with
bacteriophage l. Lambda red gam phages lacked Chi sites,
but their efficient replication depended on recombination
by the RecBCD pathway. Because, as we will see, the
RecBCD pathway depends on Chi sites, these mutants
made small plaques. Franklin Stahl and colleagues showed
that certain l red gam mutants made large plaques, suggesting that these mutants had more active RecBCD recombination. Stahl and colleagues then discovered that
recombination was enhanced near the point of the mutation and named these mutated sites Chi, for crossover
hotspot instigator. The fact that the mutations promoted
wea25324_ch22_709-731.indd Page 716
716
12/20/10
6:15 PM user-f469
/Volume/204/MHDQ268/wea25324_disk1of1/0073525324/wea25324_pagefile
Chapter 22 / Homologous Recombination
recombination nearby suggested that these mutations did
not behave like ordinary gene mutations that occur in coding regions and change the structures of gene products. Instead, it appeared that these mutations created new Chi
sites that stimulated recombination nearby.
We know that Chi sites stimulate the RecBCD pathway,
but not the l Red (homologous recombination) pathway,
the l Int (site-specific recombination) pathway, or the E. coli
RecE and RecF pathways (both homologous). This points
strongly to the participation of the RecBCD protein at Chi
sites, because this is the only component of the RecBCD
pathway not found in any of the others. In fact, because
RecBCD has endonuclease activity, one attractive hypothesis
is that it nicks DNA near Chi sites to initiate recombination.
Gerald Smith and his colleagues found evidence for this
hypothesis. They generated a 39-end-labeled, doublestranded fragment of plasmid pBR322 with a Chi site near
its end. The labeled 39-end lay only about 80 bp from the
Chi site, as shown in Figure 22.7a. Then they added purified RecBCD protein. After heat-denaturing the DNAs in
some of the reactions, they electrophoresed the DNA products and looked for the 80-nt fragment that would be
Chi site
(a)
3′
5′
5′
3′
(b)
Chi site: –
RecBCD: –
Product boiled: +
–
+
+
+
–
+
+
+
+
+
–
–
+
+
–
+
+
–
generated by nicking at the Chi site. (They had to run the
reaction briefly to avoid general degradation of the DNA
by the nonspecific RecBCD nuclease activity.) Figure 22.7b
shows that the 80-nt product was indeed observed and that
its appearance depended on the presence of the RecBCD
protein. Heat-denaturation of the DNA was not necessary
to yield the 80-nt single-stranded DNA, which suggested
that the RecBCD protein was not only nicking, but also
unwinding the DNA beyond the nick. Smith and colleagues
also mapped the exact cleavage sites by running the labeled
¯ 80-nt fragment alongside sequencing lanes generated by
chemical cleavage of the same labeled substrate. They observed two bands, one nucleotide apart, that showed that
RecBCD cut this substrate in two places, as indicated by
the asterisks in this sequence:
59-GCTGGTGGGTT*G*CCT-39
Thus, RecBCD cut this substrate 3 and 4 nt to the 39side of the GCTGGTGG Chi site (underlined). Another
substrate could be cut in three places, 4, 5, and 6 nt to the
39-side of the Chi site. Thus, the exact cleavage sites depend
on the substrate.
These findings supported the idea that RecBCD nicks
DNA near a Chi site and also suggested that RecBCD can
unwind the DNA, starting at the nick. Further support for
the role of RecBCD in unwinding DNA came from work
by Stephen Kowalczykowski and colleagues. Figure 22.8
100
80
% joint molecules
Pre-existing
bands
60
40
20
1
2
3
4
5
6
7
Figure 22.7 Chi-specific nicking of DNA by RecBCD. (a) Substrate
for nicking assay. Smith and colleagues prepared a 1.58-kb EcoRIDdeI restriction fragment with a Chi site about 80 bp from the DdeI
end. They 39-end-labeled the DdeI end by end-filling with [32P]
nucleotide (red). (b) Nicking assay. Smith and colleagues incubated
the end-labeled DNA fragment in panel (a), designated “1” in the top
line, or a similar fragment lacking the Chi site, designated “2” in the
top line, with or without RecBCD (designated “1” or “2” in the
middle line) for 30 sec. Then they terminated the reactions and
electrophoresed the products. Some of the reaction products were
boiled for 3 min as indicated at top. The arrow at right denotes the
80-nt labeled fragment released by nicking at the Chi site. The
appearance of this product depended on RecBCD and a Chi site, but
not on boiling the product. (Source: (b) Ponticelli, A.S., D.W. Schultz,
A.F. Taylor, and G.R. Smith, Chi-dependent DNA strand cleavage by recBC enzyme.
Cell 41 (May 1985) f. 2, p. 146. Reprinted by permission of Elsevier Science.)
0
0
2
4
6
Time (min)
8
10
Figure 22.8 RecBCD-dependence of strand exchange between
two duplex DNAs. Kowalczykowski and coworkers incubated two
duplex DNAs with RecA, RecBCD, and SSB (red) and assayed for joint
molecules (strand exchange) by filter binding or by gel electrophoresis.
(The joint molecules have a lower electrophoretic mobility than the
nonrecombining DNAs.) They also assayed for joint molecules without
RecA or without RecBCD (orange and purple symbols at bottom). The
blue line shows the results when RecBCD was omitted, but one of the
DNAs was heat-denatured. The green line shows the results when all
components except RecA were preincubated together, then RecA
was added to start the reaction. RecBCD was added last in all other
reactions. (Source: Adapted from Roman, L.J., D.A. Dixon, and S.C.
Kowalczykowski, “RecBCD-dependent joint molecule formation promoted by the
Escherichia coli RecA and SSB proteins,” Proceedings of the National Academy of
Sciences USA 88:3367–71, April 1991.)
wea25324_ch22_709-731.indd Page 717
12/20/10
4:42 PM user-f469
/Volume/204/MHDQ268/wea25324_disk1of1/0073525324/wea25324_pagefile
717
22.2 Experimental Support for the RecBCD Pathway
presents the results of one of their experiments, which demonstrated that: (1) RecA alone, or even RecA plus SSB,
could not cause pairing between two homologous doublestranded DNAs. (2) However, with RecBCD in addition to
RecA and SSB, strand exchange, which depends on DNA
unwinding, occurred rapidly, as long as the two DNAs
were homologous. (3) RecBCD was dispensible if one of
the DNAs was heat-denatured. This last finding implied that
one function of RecBCD is to unwind one of the DNAs to
provide a free DNA end; RecA and SSB can coat and then
use this free DNA end to initiate strand invasion.
Stuart Linn and colleagues provided direct evidence for
the DNA helicase activity of RecBCD using electron microscopy to detect the unwound T7 phage DNA products.
When the experimenters added SSB and RecBCD together,
they observed forked DNAs with duplex DNA adjoining
two single strands. This implied that RecBCD began unwinding at the end of the duplex, and SSB trapped the two
single-stranded DNAs that were generated. As expected,
the forks grew longer with time.
SUMMARY RecBCD has a DNA endonuclease ac-
tivity that can nick double-stranded DNA, especially near Chi sites, and a DNA helicase activity
that can unwind double-stranded DNAs from their
ends. These activities help RecBCD to provide the
single-stranded DNA ends that RecA needs to initiate strand exchange.
1.
3′
2.
5′
3.
5′
4.
3′
5′
+
3′
+
3′
+
5′
Anneal
Holliday junction
Figure 22.9 Forming a synthetic Holliday junction. Oligonucleotides
1–4 are mixed under annealing conditions so the complementary parts
of each can base-pair. The 59-end of oligo 2 (red) is complementary to
the 39-end of oligo 1 (red), so those two half-molecules can base-pair,
but the 39-end of oligo 2 (blue) is complementary to the 59-end of oligo
4 (blue), so those two half-molecules form base pairs. Similarly, the two
ends of oligo 3 are complementary to the other ends of oligos 1 and 4,
so oligo 3 crosses over in its base pairing, in a manner complementary
to that of oligo 2. The result is a synthetic Holliday junction.
Polyacrylamide gel
Junction
RuvA (ng) – 25 100 – – 25 100 100
RuvB (ng) – – – 110 220 110 110 110
Duplex
– 100 – 100
–
– 220 110
RuvAB/DNA
complex
RuvA/DNA
complex
Junction
RuvA and RuvB
RuvA and RuvB form a DNA helicase that catalyzes the
branch migration of a Holliday junction. We have seen that
Holliday junctions can be created in vitro; in fact they are
a by-product of experiments that measure the effect of
RecA on strand exchange. Early work on RuvA and RuvB
used such RecA products as the Holliday junctions that
could interact with RuvA and RuvB. Later, Stephen West
and his colleagues devised a method for using four synthetic oligonucleotides whose sequences required that they
base-pair in such a way as to form a Holliday junction, as
illustrated in Figure 22.9.
Carol Parsons and West end-labeled such a synthetic
Holliday junction and used a gel mobility shift assay to
measure binding of RuvA and RuvB to the Holliday junction. Because branch migration was known to require ATP,
they used the unhydrolyzable ATP analog, ATPgS. In principle, this should allow assembly of RuvA and RuvB on the
DNA, but should prevent the branch migration that would
dissociate the Holliday junction. They were successful in
demonstrating a complex between RuvA and the Holliday
junction, but did not see a supershift with RuvB, which
would have indicated a RuvA–RuvB–Holliday junction
Duplex
a b c d e f g h i
j
k l
Figure 22.10 Detecting a RuvA–RuvB–Holliday junction complex.
Parsons and West constructed a labeled synthetic Holliday junction
and mixed it with varying amounts of RuvA and RuvB, as indicated at
top. All mixtures contained ATPgS except the one in lane h. Parsons
and West then treated the mixtures with glutaraldehyde to cross-link
proteins in the same complex and prevent their dissociation. Finally,
they subjected the complexes to polyacrylamide gel electrophoresis
and autoradiography to detect the labeled complexes. (Source: Parsons,
C.A. and S.C. West, Formation of a RuvAB–Holliday junction complex in vitro. Journal
of Molecular Biology 232 (1993) f. 2, p. 400, by permission of Elsevier.)
complex. This suggested that this ternary complex was too
unstable under these experimental conditions. To stabilize
the putative complex, they added glutaraldehyde, which
should cross-link the proteins in the complex and prevent
them from dissociating during gel electrophoresis.
Figure 22.10 demonstrates cooperative binding between
RuvA and RuvB. At low RuvA concentration (lane b), little,
if any, binding to the Holliday junction occurred. On the
wea25324_ch22_709-731.indd Page 718
718
12/20/10
4:42 PM user-f469
/Volume/204/MHDQ268/wea25324_disk1of1/0073525324/wea25324_pagefile
Chapter 22 / Homologous Recombination
other hand, at high concentration, abundant binding occurred. Furthermore, RuvB by itself, even at high concentration, could not bind to the Holliday junction (lane e), but
both proteins together could bind, even at a concentration
of RuvA that could not bind well by itself (lanes f and g).
Only RuvA could bind to the Holliday junction in the absence of ATPgS; the ternary RuvA–RuvB–Holliday junction
complex did not form (lane h). Finally, neither RuvA, nor
RuvB, nor both together could bind to an ordinary duplex
DNA of the same length as the Holliday junction (lanes j–l),
so these proteins bind specifically to Holliday junctions.
RuvB can drive branch migration by itself if it is present
in high enough concentration, so it has the DNA helicase
and attendant ATPase activity. What, then, is the role of
RuvA? It binds to the center of a Holliday junction and
facilitates binding of RuvB, so branch migration can take
place at a much lower RuvB concentration. Furthermore,
as we will soon see, it appears to hold the Holliday junction
in a square planar conformation that is favorable for rapid
branch migration.
What is the nature of the binding between RuvA and the
Holliday junction? David Rice and colleagues have bolstered
the hypothesis of a square planar conformation for the
RuvA–Holliday junction complex by performing x-ray crystallography on RuvA tetramers and showing that they have
a square planar shape. This is illustrated in Figure 22.11a,
in which we see that each monomer is roughly L-shaped,
containing a leg and a foot connected by a flexible loop of
indeterminate shape, represented by a dashed colored line.
The foot of the L of one monomer interacts with the leg of
the next to form a lobe; one of the four lobes is surrounded
by a white dashed line in the figure. These lobes are arranged in a four-fold symmetrical pattern, with a natural
groove between each pair of lobes; white dashed lines lie in
two of these grooves. Figure 22.11b shows a side view of
the tetramer, which reveals a concave surface on top and a
convex surface on the bottom.
Molecular modeling showed that this RuvA tetramer
could mate naturally with a Holliday junction in a corresponding square planar conformation, as shown in Figure
22.12. Note the neat fit between the DNA and the concave
surface of the protein. The four branches of the Holliday
junction could lie in the four grooves on the surface of the
protein. Four b turns, one on each monomer, form a hollowlooking pin that protrudes through the center of the
Holliday junction. This square planar shape would allow
rapid branch migration. Any deviation from this shape
would slow branch migration, which emphasizes the importance of the square planar shape of RuvA. What is the
relationship between the square planar Holliday junction
and the familiar, branched Holliday junction we have seen
so far? They are really just two representations of the same
structure, as we will soon see.
Stephen West and Edward Egelman, along with Xiong Yu,
performed electron microscopy of the RuvAB–Holliday
(a)
(b)
Figure 22.11 Structure of RuvA tetramer as revealed by x-ray
crystallography. (a) Top view. The four monomers are represented
by different-colored ribbons, and one of the four lobes in the square
planar structure is outlined with a dashed white line. The three domains
of the blue monomer are numbered, as is the third domain (the “foot”
of the L) of the green monomer. (b) Side view. The same structure,
represented by the same colored ribbons, is shown from the side. The
concave and convex surfaces at top and bottom, respectively, are
evident. (Source: Rafferty J.B., S.E. Sedelnikova, D. Hargreaves, P.J. Artymiuk,
P.J. Baker, G.J. Sharples, A.A. Mahdi, R.G. Lloyd, and D.W. Rice, Crystal structure
of DNA recombination protein RuvA and a model for its binding to the Holliday
junction. Science 274 (18 Oct 1996) f. 2 d–e, p. 417. Copyright © AAAS.)
junction complex. They made 100 micrographs of the
complex, scanned them, and combined them to create an
average image. Figure 22.13a presents a model based on
this image, with color-coded DNA strands added. As expected, the RuvA tetramer is in the center at the junction,
with two RuvB hexameric rings flanking it. Panel (b) shows
what happens on bending two of the arms of the complex
wea25324_ch22_709-731.indd Page 719
12/20/10
4:42 PM user-f469
/Volume/204/MHDQ268/wea25324_disk1of1/0073525324/wea25324_pagefile
22.2 Experimental Support for the RecBCD Pathway
719
(with the RuvA tetramer removed for clarity), and panel (c)
shows the result of rotating the bottom DNA duplex
through 180 degrees out of the plane of the paper (again
the RuvA tetramer is removed for clarity). The RuvB rings
on this familiar Holliday junction are poised to catalyze
branch migration by moving in the direction of the arrows.
Though it is harder to visualize, they can do the same with
the DNA in the shape shown in panel (a).
SUMMARY RuvA and RuvB form a DNA helicase
that can drive branch migration. A RuvA tetramer
with square planar symmetry recognizes the center
of a Holliday junction and binds to it. This presumably induces the Holliday junction itself to adopt a
square planar conformation, and promotes binding
of hexamer rings of RuvB to two diametrically opposed branches of the Holliday junction. Then RuvB
uses its ATPase to drive the DNA unwinding and
rewinding that is necessary for branch migration.
RuvC
Figure 22.12 Model for the interaction between RuvA and a
Holliday junction. The RuvA monomers are represented by green
tubes that trace the a-carbon backbones of the polypeptides. The
DNAs in the Holliday junction are represented by space-filling models
containing dark and light pink and blue backbones and silver base
pairs. The yellow balls denote the phosphate groups of one of the two
pairs of sites that can be cut by RuvC to resolve the Holliday
junction. (Source: Rafferty, J.B., S.E. Sedelnikova, D. Hargreaves, P.J. Artymiuk,
P.J. Baker, G.J. Sharples, A.A. Mahdi, R.G. Lloyd, and D.W. Rice, Crystal structure
of DNA recombination protein RuvA and a model for its binding to the Holliday
junction. Science 274 (18 Oct 1996) f. 3d, p. 418. Copyright © AAAS.)
What nuclease is responsible for making the cuts that resolve Holliday junctions? West and colleagues showed in
1991 that it is RuvC. They built the 32P-labeled synthetic
Holliday junction pictured in Figure 22.14a, with a short
(12 bp) homologous region (J) at the joint, but the rest of
the structure composed of nonhomologous regions. Next,
they used a gel mobility shift assay to test the ability of
RuvC to bind to the Holliday junction and to a linear duplex DNA. Figure 22.14b shows the results: As they added
more and more RuvC to the Holliday junction, West and
colleagues observed more and more DNA–protein complex, indicating RuvC–Holliday junction binding. But the
3′
5′ 3′
3′
3′
5′
5′
3′
5′
3′
5′
3′ 5′
(a)
5′
3′
5′
5′
3′
5′
3′
(b)
Figure 22.13 Model for RuvAB–Holliday junction complex based
on combining EM images of the complex. (a) Complex with DNA
branches perpendicular to each other. The DNA moves through the
complex in the directions indicated by the arrows. (b) The blue-yellow
and red-green branches from panel (a) have been rotated 90 degrees
in the plane of the paper, and the RuvA tetramer has been removed so
we can see the center of the junction. (c) The blue-green and blue-
5′
3′
3′
5′
(c)
yellow branches from panel (b) have been interchanged by rotating the
lower limb of the junction 180 degrees out of the plane of the paper.
This produces a familiar Holliday junction with RuvB hexamer rings in
position to catalyze branch migration by moving in the direction of the
arrows. (Source: Adapted from Yu, X., S.C. West, and E.H. Egelman, Structure
and subunit composition of the RuvAB–Holliday junction complex. Journal of
Molecular Biology 266:217–222, 1997.)
wea25324_ch22_709-731.indd Page 720
720
5′ 3′
12/20/10
4:42 PM user-f469
/Volume/204/MHDQ268/wea25324_disk1of1/0073525324/wea25324_pagefile
Chapter 22 / Homologous Recombination
Junction
3′ 5′
[RuvC]μM 0
A
Duplex
[RuvC]μM
Junction
0.065
0.26
0 0.032
0.13
Duplex
0 0.26
0.032 0.13
0.032 0.13
0.065 0.26 0
0.065 0.26
B
Junction
Complex
J
J
C
D
Junction
Duplex
3′ 5′
1 2
(a)
5′ 3′
3 4
a
b
c
d
e f
g
h
i
a
j
(b)
Figure 22.14 RuvC can resolve a synthetic Holliday junction.
(a) Structure of the synthetic Holliday junction. Only the 12-bp central
region (J, red) is formed from homologous DNA. The other parts of the
Holliday junction (A, B, C, and D) are nonhomologous, as indicated by
the different colors. (b) Binding of RuvC to the synthetic Holliday
junction. West and colleagues end-labeled the Holliday junction (and
a linear duplex DNA) and bound them to increasing amounts of pure
RuvC under noncleavage conditions (low temperature and absence of
MgCl2). Then they electrophoresed the products. RuvC binds to the
same assay showed no binding between RuvC and a linear
duplex DNA made from strand 1 and its complement.
Thus, RuvC binds specifically to a Holliday junction,
but can it resolve the junction? Figure 22.14c shows that it
can. West and coworkers added increasing amounts of
RuvC to the labeled Holliday junction, or to the duplex
DNA. They found that RuvC caused resolution of the Holliday junction to a labeled species with the same mobility
as the duplex DNA, which is what we expect for resolution.
More complex experiments that can distinguish between
patch or splice resolution showed that the splice products
predominate, at least in vitro.
Thanks to x-ray crystallography studies performed by
Kosuke Morikawa and colleagues, we now know the threedimensional structure of RuvC. It is a dimer, with its two
active sites 30 Å apart. That puts them right in position
to cleave the square planar Holliday junction at two sites,
as shown in Figure 22.15a. Figure 22.15b presents a
more detailed representation of a RuvC–Holliday junction complex.
Does RuvC act alone, as this model implies, or does it
act on a Holliday junction already bound to RuvB or RuvA
plus RuvB? The evidence strongly suggests the latter.
West and colleagues have reconstituted a system that
carries out the intermediate to late stages of recombination in vitro and have shown that monoclonal antibodies
against RuvA, RuvB, and RuvC each block resolution of
Holliday junctions.
Duplex
b
c
d
e
f
g
(c)
Holliday junction, but not to ordinary duplex DNA. (c) Resolution of
the Holliday junction by RuvC. West and coworkers mixed the labeled
Holliday junction (or linear duplex DNA) with increasing concentrations
of RuvC under cleavage conditions (378C and 5 mM MgCl2). Then they
electrophoresed the products. RuvC resolved some of the Holliday
junction to a linear duplex form. (Source: Dunderdale, H.J., F.E. Benson,
C A. Parsons, G.J. Sharples, R.G. Lloyd, and S.C. West, Formation resolution of
recombination intermediates by E. coli recA and RuvC proteins. Nature 354 (19–26
Dec 1991) f. 5b–c. p. 509. Copyright © Macmillan Magazines Ltd.)
(b)
(a)
Figure 22.15 Model for the interaction between RuvC and a
Holliday junction. (a) Schematic model showing the RuvC dimer
(gray) bound to the square planar Holliday junction. Scissors
symbols (green) denote the active sites on the two RuvC monomers.
Note how the location of these active sites fits with the positioning of
the DNA strands to be cleaved in resolving the complex. (b) Detailed
model. Gray tubes represent the carbon backbone of the RuvC
dimer. The Holliday junction is represented by the same blue and
pink backbone and silver base pairs as in Figure 22.12. (Source: From
Rafferty, J.B., S.E. Sedelnikova, D. Hargreaves, P.J. Artymiuk, P.J. Baker,
G.J. Sharples, A.A. Mahdi, R.G. Lloyd, and D.W. Rice, Crystal structure of DNA
recombination protein RuvA and a model for its binding to the Holliday junction.
Science 274 (18 Oct 1996) f. 3e, p. 418. Copyright © AAAS. Reprinted with
permission from AAAS.)