86 223 Meiotic Recombination

wea25324_ch22_709-731.indd Page 721
12/20/10
4:42 PM user-f469
/Volume/204/MHDQ268/wea25324_disk1of1/0073525324/wea25324_pagefile
22.3 Meiotic Recombination
One way to explain this result is that RuvA, RuvB, and
RuvC work together. If that is true, then the proteins probably naturally associate with one another, and one should
be able to cross-link them. So West and colleagues prepared
various mixtures of the two proteins, added glutaraldehyde
to cross-link them, and electrophoresed them to detect
cross-links. They found that RuvA and RuvB could be
cross-linked, as expected, and RuvB and RuvC could also
be cross-linked, but RuvA and RuvC could not. Thus,
RuvB can bind to both RuvA and RuvC, suggesting that all
three proteins can bind together to a Holliday junction.
This hypothesis of concerted action by RuvA, RuvB,
and RuvC is consistent with the notion that branch migration is necessary during resolution to help RuvC find its
preferred sites of cleavage. It is also consistent with x-ray
crystallography data showing that RuvA can associate
with a Holliday junction as a tetramer, as we have already
seen, or as an octamer, with tetramers on either side of the
DNA. West hypothesized that the complex involving the
RuvA octamer is specific for efficient branch migration
(Figure 22.16a). Later, RuvC could replace one of the RuvA
tetramers to form the putative RuvABC–junction complex,
or “resolvasome” (Figure 22.16b) that is specific for resolution of the Holliday junction.
Mutations in ruvA, ruvB, and ruvC all produce the
same phenotype: heightened sensitivity to UV light, ionizing radiation, and the antibiotic mitomycin C because of
(a) Alternate RuvAB–junction complex
2 RuvA
tetramers
721
defective recombination repair. But RuvA and B promote
branch migration, whereas RuvC catalyzes resolution of
Holliday junctions. Why should defects in all three of these
proteins have the same end result? One way to answer this
question would be to show that resolution depends on
branch migration. Then defective RuvA or RuvB would
block resolution indirectly by blocking branch migration.
West and colleagues did not exactly do that, but they
did show that hotspots for RuvC resolution occur, which
implies that branch migration is needed to reach those hotspots. To determine the sequences at the RuvC cutting sites,
West and coworkers performed primer extension analysis
on the RuvC products, using the same primers for DNA
sequencing of the same DNAs. In all, they identified 19 cutting sites, and they observed a clear consensus sequence:
59-(A/T)TT ↓ (G/C)-39. RuvA and B are presumably needed
to catalyze branch migration in vivo to reach such consensus sites. This hypothesis also implies that resolution to
patch or splice products depends on the frequencies of the
RuvC resolution sequence in the two DNA strands. Overall, this should be a 50/50 mix.
SUMMARY Resolution of Holliday junctions in
E. coli is catalyzed by the RuvC resolvase. This protein acts as a dimer to clip two DNA strands to yield
either patch or splice recombinant products. This
clipping occurs preferentially at the consensus
sequence 59-(A/T)TT ↓ (G/C)-39. Branch migration
is essential for efficient resolution of Holliday junctions, presumably because it is essential to reach the
preferred cutting sites. Accordingly, RuvA, B, and C
appear to work together in a complex to locate and
cut those sites.
RuvB
hexamer
22.3 Meiotic Recombination
(b) Putative RuvABC–junction complex
c
As mentioned early in this chapter, meiosis in most eukaryotes is accompanied by recombination. This process shares
many characteristics in common with homologous recombination in bacteria. In this section, we will examine the
mechanism of meiotic recombination in yeast.
c
Top view
Side view
Figure 22.16 Models of Ruv protein–junction complexes. (a) A
RuvAB–junction complex discovered by Pearl and colleagues. In
contrast to the complex with a RuvA tetramer described by other
workers, this one contains a RuvA octamer at the Holliday junction.
This could be the form the complex takes during active migration.
(b) West’s model of the RuvABC–junction complex, with RuvC in
purple. This could be the form the complex takes during resolution.
Top and side views are on the left and right, respectively, in both
panels. (Source: Adapted from West, S.C., RuvA gets x-rayed on Holliday.
Cell 94:700, 1998.)
The Mechanism of Meiotic Recombination:
Overview
Figure 22.17 presents a hypothesis for meiotic recombination
in budding yeast (Saccharomyces cerevisiae), where it has
been most thoroughly studied. The process starts with a chromosomal lesion: a double-stranded break. Next, an exonuclease recognizes the break and digests the 59-ends of two of the
strands, creating 39-single-stranded overhangs. One of these
single-stranded ends can then invade the other DNA duplex,
wea25324_ch22_709-731.indd Page 722
722
12/20/10
4:42 PM user-f469
/Volume/204/MHDQ268/wea25324_disk1of1/0073525324/wea25324_pagefile
Chapter 22 / Homologous Recombination
(a) Double-strand break
(b) Exonuclease (5′
5′
3′
5′
3′
3′
5′
3′)
3′
5′
(c) Strand invasion,with D-loop formation
(d) DNA repair synthesis
(e) Ligation and branch migration
(f) Resolution (noncrossover)
Noncrossover recombinants
(g) Resolution (crossover)
Crossover recombinants
Figure 22.17 Model for meiotic recombination in yeast. (a) A
double-stranded break occurs in one DNA duplex (blue), which is
paired with another DNA duplex (red). (b) An exonuclease digests the
DNA 59-ends at the newly created break. (c) A single-stranded 39-end
of the top duplex invades the bottom duplex, creating a D-loop.
(d) DNA repair synthesis extends the free 39-ends, with enlargement
of the D-loop. (e) Branch migration occurs both leftward and rightward
to yield two Holliday junctions. (f) The Holliday junctions are resolved
by cleaving the inside strands at both Holliday junctions, yielding
noncrossover recombinant DNAs with patches of heteroduplex, but
no exchange of DNA arms beyond the Holliday junctions. (g) The
Holliday junctions are resolved by cleaving the inside strands at the
left Holliday junction and the outside strands at the right Holliday
junction. This yields crossover recombinant DNAs with exchange of
DNA arms to the right of the right Holliday junction.
forming a D-loop as we observed in bacterial homologous
recombination. Next, DNA repair synthesis fills in the gaps in
the top duplex, expanding the D-loop in the process. Next,
branch migration can occur in both directions, leading to two
Holliday junctions. Finally, the Holliday junctions can be resolved to yield either a noncrossover recombinant with two
sections of heteroduplex, or a crossover recombinant that has
exchanged flanking DNA regions.
There is good evidence for most of these steps, but a
few features of the hypothesis are contradicted by experiment. In particular, the model predicts that hybrid DNA
will be produced on both sides of the double-stranded
break. However, when this prediction was tested genetically, hybrid DNA was usually found only on one side of
the break. In the few cases where hybrid DNA was found
on both sides of the break, it was in the same chromatid,
not in both chromatids as the model predicts. Thus, more
data are needed to resolve this discrepancy and perhaps
amend the hypothesis.
It is also worth noting that different organisms may do
things somewhat differently. The classical work on meiotic
recombination was performed in budding yeast, where the
double Holliday junction model seems to predominate.
However, Gerald Smith and colleagues reported in 2006
that the fission yeast Schizosaccharomyces pombe carries
out meiotic recombination through a single Holliday junction intermediate, such as that pictured in Figures 22.2 and
22.3. Furthermore, these authors suggested that this organism may initiate some meiotic recombination by singlestrand nicking, rather than double-strand breaks.
In another departure from the canonical model, Thorsten
Allers and Michael Licten reported in 2001 that noncrossover recombinants appeared at the same time as Holliday
junctions in budding yeast. Only later did crossover recombinants appear, through resolution of the Holliday junctions. This finding suggests that noncrossover recombinants
in this organism do not result primarily from resolution of
Holliday junctions, but from another mechanism that does
not involve Holliday junctions.
The Double-Stranded DNA Break
How do we know that recombination in yeast initiates with
a double-stranded DNA break (DSB)? Jack Szostak and
colleagues laid the groundwork for answering this question
in 1989 by mapping a recombination initiation site in the
ARG4 gene of the yeast Saccharomyces cerevisiae. They
did not look at recombination per se, but at meiotic gene
conversion, which depends on meiotic recombination in
yeast. Because both gene conversion and recombination
initiate at the same site, these workers could use gene conversion as a surrogate for recombination. We will examine
the mechanism of gene conversion later in this chapter.
Szostak and coworkers verified earlier work that had
shown that meiotic gene conversion in the ARG4 locus was
wea25324_ch22_709-731.indd Page 723
12/20/10
4:42 PM user-f469
/Volume/204/MHDQ268/wea25324_disk1of1/0073525324/wea25324_pagefile
723
22.3 Meiotic Recombination
polar: It was common (about 9% of total meioses) near the
59-end of the gene, and relatively rare (about 0.4% of total
meioses) at the 39-end. This behavior suggested that the initiation site of recombination lies near the 59-end of the gene.
So Szostak and colleagues made deletions in this region to
try to remove the initiation site and therefore to block gene
conversion. They found that deletions having their 39-ends
in the 2316 to 11 region all greatly decreased gene conversion rates, suggesting that the initiation site for recombination lies in the promoter region of the ARG4 gene.
This information allowed these workers to look for a
DNA break, either single- or double-stranded, in a very
restricted region of the yeast genome. Accordingly, they
cloned a 15-kb fragment of DNA, including the ARG4
gene, into a yeast plasmid, and introduced it into a strain of
yeast that carries out synchronous meiosis immediately on
transfer to sporulation medium. They extracted plasmid
DNA at various times after induction of sporulation and
subjected it to electrophoresis.
Figure 22.18 depicts the results of the electrophoresis.
At time zero, we can see mostly supercoiled monomers,
with some supercoiled dimers, relaxed circular monomers,
and a band of lower mobility that is probably some form of
dimer. These same bands appeared throughout the time
course after induction of sporulation. The one novelty during sporulation was a relatively faint band of linear monomer that first appeared at 3 h, peaked at 4 h, and decreased
after that time. This linear DNA must have been created by
a double-stranded break in the plasmid. The timing of this
DSB coincided with the timing of commitment to meiotic
recombination in these cells (2.5–5 h), and with the appearance of recombination products (4 h). These findings are all
consistent with the hypothesis that the first step in meiotic
recombination is the formation of a DSB.
Szostak and colleagues used restriction mapping to
demonstrate that DSBs occurred in three different locations
in the plasmid, indicated by arrows in Figure 22.18a. One
of these breaks (site 2) lies within a 216-bp restriction fragment in the control region just 59 of the ARG4 gene. Szostak
and colleagues’ previous work had shown that a 142-bp
deletion in this same region depressed the level of meiotic
gene conversion in the ARG4 gene, so they tested the same
deletion for effect on the DSB. They found that this deletion
did indeed eliminate the DSB at site 2, but had no effect on
the DSBs at sites 1 and 3. Thus, the ability to form the DSB
at site 2 is correlated with the efficiency of meiotic gene
conversion just downstream in the ARG4 gene.
If a DSB is the initiating event in meiotic recombination, then it should occur in a yeast chromosome, not just
in a plasmid. Therefore, Szostak and colleagues used restriction mapping in cells lacking the plasmid to search for
these same DSBs. They found that a DSB at site 2 (and at
site 1) also occurred in yeast chromosomal DNA, and that
the timing of appearance of these DSBs was the same as in
the plasmid.
(a)
L
1
2
R
3
Sphl
1 kb
(b)
Sphl
DED82 DED81
Time (h):
ARG4
0
CEN4
1
2
3
ARS1TRP1 URA3
4
5
6
7
Relaxed circular
monomers
Supercoiled dimers
Linear monomers
Supercoiled
monomers
Figure 22.18 Detecting a double-stranded DNA break in a plasmid
bearing a recombination initiation site. (a) Map of the plasmid used
to detect the DSB. The yellow bar represents the 15-kb insert of yeast
DNA containing the recombination initiation site. The other colored
bars represent loci within the vector, including the centromere (CEN4).
The locations of genes are indicated below the bars. L and R are the
locations of probes used to hybridize to blots. The arrows marked 1,
2, and 3, are the locations of DSBs mapped in this experiment.
(b) Electrophoresis results. Szostak and colleagues transformed yeast
cells with the plasmid depicted in panel (a), induced sporulation, then
electrophoresed samples of plasmid collected at the indicated times
after induction. Finally, they Southern blotted the DNAs and probed
them with a 32P-labeled probe that hybridized to the vector. The
identities of the bands are indicated at left. Note the appearance of
linear monomers around 4 h, which indicates a double-stranded
break. (Source: Sun, H., D. Treco, N.P. Schultes, and J.W. Szostak, Doublestrand breaks at an initiation site for meiotic gene conversion. Nature 338
(2 Mar 1989) f. 1, p. 88. Copyright © Macmillan Magazines Ltd.)
Nancy Kleckner and colleagues demonstrated similar
double-stranded breaks in a yeast chromosome into which
they had inserted a LEU2 gene next to a HIS4 gene to create a hotspot for meiotic recombination. Actually, two
DSBs occurred close together at this hotspot. They also
discovered a nonnull mutation (a mutation that does not
totally inactivate a gene) in the RAD50 gene (rad50S) that
caused a buildup of the fragments caused by the DSBs. This
mutation apparently blocked a step downstream of DSB
formation, so it allowed the DSBs to accumulate.
In 1995, Scott Keeney and Kleckner discovered that the
59-ends created by a DSB are covalently bound to a protein
in rad50S mutants. One attractive hypothesis to explain
wea25324_ch22_709-731.indd Page 724
4:42 PM user-f469
/Volume/204/MHDQ268/wea25324_disk1of1/0073525324/wea25324_pagefile
Chapter 22 / Homologous Recombination
O
Pstl
CENIII
(b)
DSB DSB
Site II Site I
1 kb
(a)
Pstl
HIS4
Probe
LEU2
(c)
α-HA:
SP
O
11
-H
A
SP
O
11
+
this behavior is that the catalytic protein that created the
DSB would normally dissociate from the DNA immediately, but in this mutant the protein remained bound to the
DNA ends it had created. If that is the case, then identifying the protein bound to the DSB ends would identify a
prime suspect for the endonuclease that created the DSB.
Accordingly, Kleckner and colleagues set out to identify
the protein or proteins covalently bound to the DSB. They
began by isolating nuclei from meiotic rad50S cells. Because of their accumulation of the protein–DSB complex,
these cells should provide a rich source of the DSB-bound
protein. To purify the bound protein, Kleckner and coworkers used a two-stage screening procedure: First, they
extracted the nuclei and denatured the protein with guanidine and detergent, then purified DNA and DNA–protein
complexes by CsCl2 gradient ultracentrifugation. Any proteins bound to DNA under these denaturing conditions
should be covalently attached. Then, they passed this mixture through a glass fiber filter that bound the DNA–protein
complexes but allowed pure DNA to flow through. The
material bound to the filter should be highly enriched in
covalent DNA–protein complexes, so Kleckner and colleagues digested the DNA in these complexes with a nuclease and subjected the liberated proteins to SDS-PAGE.
They observed several bands, two of which appeared in
rad50S cells, but not in a spo11D mutant that was blocked
in DSB formation. These same two bands appeared in
preparative-scale, as well as pilot-scale preparations, and
their appearance depended on nuclease treatment of DNA–
protein complexes.
Next, Kleckner and coworkers excised the two candidate bands (Mr 5 34 kD and 45 kD) from a preparative gel,
subjected them to trypic digestion, then sequenced some of
the trypic peptides. The short protein sequences obtained
from these peptides yielded the corresponding DNA sequences. Then, because the sequence of the entire yeast
genome was already known, these short DNA sequences
allowed for easy identification of the corresponding genes.
The 45-kD protein corresponded to (coincidentally) the
spo11 gene product, Spo11, and the 34-kD protein corresponded to a mixture of five different proteins, including
two ribosomal proteins.
Because Spo11 was already known to be required for
meiosis, it was an attractive candidate for the DSB-bound
protein. To reinforce this hypothesis, Kleckner and colleagues demonstrated that Spo11 is bound specifically to
DSBs, and not to bulk DNA. To do this, they used an epitope tagging approach. They created a Spo11 gene fused to
the coding region for an epitope of the protein hemagglutinin. The protein product of this gene (Spo11-HA), and any
DNA attached, could therefore be immunoprecipitated
with an antihemagglutinin antibody.
In a preliminary experiment, these workers isolated
DNA (with any covalently attached proteins) from meiotic
rad50S cells containing the same HIS4LEU2 recombination
SP
O
SP 11-H
O
11 A
+
724
12/20/10
–
+
–
+
Parent
Site II
Site I
Figure 22.19 Association of Spo11 with DSB fragments. (a) Map of
the hotspot region. A fragment (red and blue) with the LEU2 gene (red)
has been inserted adjacent to the HIS4 gene in yeast chromosome III.
The centromere (CENIII), the locations of the DSB sites I and II, and the
site to which the Southern blot probe hybridizes are shown, along with
the locations of two PstI sites flanking the DSBs. (b) Southern blot of
total DNA, cut with PstI, electrophoresed, blotted, and hybridized to
the probe. The parent fragment, as well as the subfragments generated
by DSBs, are present. (c) Southern blot of DNA, cut with PstI, then
immunoprecipitated with an anti-HA antibody, then blotted and probed
as in panel (b). The subfragments generated by DSBs are greatly
enriched relative to the parent fragment. (Source: Keeney, S., C. Giroux, and
N. Kleckner, Meiosis-specific DNA double-strand breaks are catalyzed by Spo11,
a member of a widely conserved protein family. Cell 88 (Feb 1997) f. 3, p. 378.
Reprinted by permission of Elsevier Science.)
hotspot as in their previous studies. They cut the DNA with
PstI, electrophoresed it, blotted the fragments, and probed
the blot with a DNA probe for the hotspot region. Figure
22.19a depicts a map of the hotspot region, showing the
two sites where DSBs occur during meiotic recombination,
two PstI sites flanking the DSB sites, and the location to
which the probe hybridizes downstream of HIS4LEU2.
Thus, if no DSBs occur, only the parent PstI fragment
should be observed. On the other hand, if the DSBs do occur, two additional smaller fragments, corresponding to
DSB sites I and II, should appear. Figure 22.19b demonstrates that both these smaller fragments do indeed appear,
both in wild-type cells (SPO111) and in SPO11-HA cells.
Next, Kleckner and colleagues checked to see whether
Spo11-HA was specifically bound to these fragments created by DSBs. They repeated the experiment we have just
discussed, but this time they immunoprecipitated Spo11HA–DNA complexes after cutting the DNA with PstI.
Figure 22.19c presents the results. It is clear that the two DNA
fragments created by DSBs (but very little of the parental
wea25324_ch22_709-731.indd Page 725
12/20/10
4:42 PM user-f469
/Volume/204/MHDQ268/wea25324_disk1of1/0073525324/wea25324_pagefile
22.3 Meiotic Recombination
fragment) were immunoprecipitated along with Spo11-HA.
However, they were not immunoprecipitated in the absence
of the anti-HA antibody, nor were they precipitated from a
yeast strain with the wild-type SPO11 gene with no HA tag
attached. Further analysis showed that these fragments
were not immunoprecipitated from a wild-type RAD50
strain that did not accumulate DSBs, nor from a mutant
strain that did not form DSBs at all.
If Spo11-HA merely bound nonspecifically to DNA, it
should have been attached to the parental DNA fragment
as well as the two subfragments created by DSBs, but the
subfragments were enriched over 600-fold relative to the
parental DNA in the immunoprecipitates. Thus, Spo11 appears to bind specifically to DSBs and is likely to be the
catalytic part of the enzyme that created the DSBs. Furthermore, Spo11 was known at that time to be homologous to
proteins in other organisms, including an archaeon, a fission yeast, and a roundworm. All four proteins have only
one conserved tyrosine, which is likely to be the catalytic
amino acid that becomes covalently attached to the DSB. In
this way, it would resemble the active site tyrosine of a
topoisomerase (Chapter 21). The conserved tyrosine in
Spo11 is Tyr-135 and, as expected, it is essential for activity. Accordingly, one can propose a model such as the one
in Figure 22.20, which calls for the participation of two
molecules of Spo11, one to attack each strand of the DNA
at slightly offset positions. This process creates the DSB,
and leaves a transient intermediate with a molecule of
SPO11 covalently attached through its active site tyrosine
to the newly created 59-phosphate on each strand. Thus,
the creation of DSBs appears not to occur by a simple hydrolysis, but by a transesterification, in which the attacking
group is a tyrosine residue of the enzyme, rather than a
water molecule.
The covalent association between Spo11 and the DSB
ends is only transient, so the two molecules of Spo11 must
be removed somehow. This process could occur by direct
hydrolysis of the protein–DNA bonds, or by the action of
an endonuclease that would remove the proteins along
with a short stretch of DNA from each end.
In 2005, Scott Keeney and colleagues showed that the
latter mechanism, illustrated in Figure 22.20, is correct.
Like Kleckner and colleagues, they engineered a yeast
strain to express Spo11 tagged with the hemagglutinin
(HA) epitope. Then they immunoprecipitated Spo11
from meiotic cells using an anti-HA antibody. To detect
oligonucleotides bound to Spo11, they treated the immunoprecipitates with terminal deoxynucleotidyl transferase (TdT) and 32P-labeled cordycepin triphosphate. TdT
adds nucleotides nonspecifically to the 39-ends of DNAs,
and cordycepin triphosphate (39-deoxyadenosine triphosphate) terminates the reaction because it provides
no 39-hydroxyl group to link to the next nucleotide.
Figure 22.21 shows the results. The asterisks mark
two bands that must not have anything to do with DSBs
HO
725
Spo11
Spo11
OH
(a) DNA cleavage
(Transient) 5′
3′
P
3′
OH
Spo11
5′
5′
Spo11
HO
P 3′
3′
5′
(b) Nicking DNA strands
5′
3′
P
3′
OH
Spo11
5′
5′
Spo11
HO
P 3′
3′
5′
(c) Release of Spo11-linked
oligonucleotides
P
5′
3′
5′
Spo11
3′
OH
HO
3′
Spo11
5′
3′
5′
P
Figure 22.20 Model for the participation of Spo11 in DSB
formation. (a) DNA cleavage. Two molecules of Spo11, with active
site tyrosines represented by their OH groups, attack the two DNA
strands at slightly offset positions. This transesterification reaction
breaks phosphodiester bonds within the DNA strands and creates
new phosphodiester bonds between the new DNA 59-ends and the
Spo11 tyrosines. (b) Nicking DNA strands. The nicking is asymmetric,
yielding two sizes of Spo11-linked oligonucleotides. (c) Release of
Spo11-linked oligonucleotides. The release could occur before DNA
end resection, as shown here, but there is evidence for a later release.
or Spo11, because they are seen in lane 1, in which no cell
extract or antibody was used. The arrows point to two
bands (in lane 3) that were Spo11-specific. This claim of
specificity comes from the fact that these bands were not
seen when mock-immunoprecipitation was performed
without the HA antibody (lane 2), or when Spo11 was
not tagged with the HA epitope (lane 4). We know that
the appearance of these bands also depended on the formation of DSBs, because they did not appear when the
catalytic tyrosine of Spo11 was changed to phenylalanine
(lane 5), or when DSBs were blocked by the mei4 mutation (lane 6).
The fact that the oligonucleotide-tagged Spo11-HA appeared in two bands suggested that Spo11 is associated
wea25324_ch22_709-731.indd Page 726
4:42 PM user-f469
/Volume/204/MHDQ268/wea25324_disk1of1/0073525324/wea25324_pagefile
+
+
100
Large oligos
% DSBs
2
50
kDa
Small oligos
250
150
0
*
100
*
75
50
37
Lane: 1
2
3
4
5
0
2
4
Time (h)
6
8
Labeled Spo11 (% of maximum)
+
4
DSBs (%)
+
SPO11-HA, mei4⌬
-
Y135F-HA
Anti-HA: -
SPO11
Extract:
SPO11-HA
Chapter 22 / Homologous Recombination
None
726
12/20/10
0
Figure 22.22 Time course of DSB and Spo11-oligonucleotide
formation and disappearance. DSBs as a percentage of total DNA
(green) were measured at the yeast HIS4LEU2 recombination hotspot.
The large (red) and small (blue) oligonucleotide-linked Spo11 species
are plotted as a percent of the maximum for the large species.
(Source: Adapted from Neale, M. J., et al., Endonucleolytic processing of covalent
protein-linked DNA double strand breaks. Nature 436: 1054, fig. 1f, 2005.)
6
Figure 22.21 Evidence for Spo11-linked oligonucleotides.
Keeney and colleagues immunoprecipitated yeast proteins from
cellular extracts with no antibody (lanes 1 and 2), or with anti-HA
antibody, as indicated at top. The genotypes of the cells from which
extracts were prepared, also indicated at top, were: lanes 2 and 3,
SPO11, fused to a coding region for an HA epitope; lane 4, wild-type
SPO11, including no HA epitope; lane 5, the SPO11 mutant Y135F, in
which the active site tyrosine (Y) is changed to phenylalanine (F), fused
to a coding region for an HA epitope; lane 6, the meiosis mutant
mei4D, in which DSBs are not formed, with SPO11 fused to a coding
region for an HA epitope. Keeney and colleagues terminally labeled
any oligonucleotides attached to the immunoprecipitates with TdT and
[a-32P]cordycepin triphosphate. Finally, they subjected the proteins to
SDS-PAGE and detected labeled proteins by autoradiography.
Asterisks indicate nonspecific labeled bands that appear even without
extracts or antibody. Arrows indicate Spo11-specific proteins that are
labeled only when DSBs form. (Source: Reprinted by permission from
Macmillan Publisher Ltd: Nature 436, 1053-1057, Thomas Schalch, Sylwia Duda,
David F. Sargent and Timothy J. Richmond, “Endonucleolytic processing of covalent
protein-linked DNA double-strand breaks,” fig. 16, p. 1054 copyright 2005.)
with oligonucleotides of two sizes. Accordingly, Keeney
and colleagues digested the protein from each band with
protease and electrophoresed the remaining oligonucleotides to determine their sizes. The upper band yielded a
smeared oligonucleotide band centered around 24–40 nt
long, and the lower band yielded a smeared oligonucleotide
band centered around 10–15 nt long. This result confirmed
that oligonucleotides of two sizes were bound to Spo11,
and the smearing suggested that the oligonucleotides were
of varying lengths, or that there was heterogeneity in the
cutting of Spo11 by protease. Allowing for an average of
three amino acids remaining attached to the oligonucleotides, and knowing that oligonucleotides less than 10 nt
long were not retained well by the gel, Keeney and
colleagues estimated that the two oligonucleotides were
about 21–37 and # 12 nt long. These workers also
obtained similar results in their studies on mouse DSB
processing, but the sizes of the two classes of oligonucleotides linked to the mouse Spo11 homolog were somewhat
different from those discovered in yeast.
Figure 22.22 shows that the timing of appearance of
the Spo11-oligonucleotides exactly corresponded to the
timing of appearance of Spo11-free resected DSBs, which is
what we expect if Spo11-oligonucleotides are natural products of DSB processing. Furthermore, the larger and smaller
bands were always in a strict 1:1 ratio, indicating that they
are produced simultaneously by the same process. These
findings lead to some intriguing tentative conclusions.
First, the accumulation and disappearance of Spo11oligonucleotides closely mirrors the accumulation and disappearance of Spo11-free resected DSBs. As already noted,
the correspondence of the accumulation of the species is
expected, but one would not have predicted that the disappearance of Spo11-oligonucleotides would coincide with
disappearance of Spo11-free DSBs. Instead, the simplest
model would be that Spo11 (with or without attached oligonucleotides) would be released from the DSB before resection (Figure 22.20c), which in turn would occur before
the DSBs disappeared due to formation of Holliday junctions. This model would therefore predict that destruction
of Spo11-oligonucleotides would begin before the loss of
Spo11-free DSBs. The fact that the two phenomena occur
simultaneously can be explained in two ways. First, it
could just be a coincidence that the destruction of Spo11oligonucleotides is slow enough that it occurs at the same
time that resected DSBs are forming Holliday junctions. But
the more interesting possibility is that Spo11-oligonucleotides
do not begin to be degraded until after resection because
they are not released until that time. This concept is illustrated in Figure 22.23.
wea25324_ch22_709-731.indd Page 727
12/20/10
4:42 PM user-f469
/Volume/204/MHDQ268/wea25324_disk1of1/0073525324/wea25324_pagefile
22.3 Meiotic Recombination
5′
3′
OH
Spo11
P
P
Spo11
HO
3′
5′
(a) Resection
5′
3′
OH
Spo11
P
P
Spo11
HO
3′
5′
(b) Recombinase
loading (asymmetric)
5′
3′
OH
Spo11
P
P
Spo11
HO
3′
5′
(c) Duplex invasion
5′
3′
3′
5′
OH
Spo11
P
P
Spo11
3′
5′
5′
3′
Figure 22.23 A model for DSB end resection prior to release of
Spo11-oligonucleotides. (a) Resection occurs on both strands,
using the nicks created at a previous step (Figure 22.20). (b) Both
recombinases (Rad51 and Dmc1) load asymmetrically onto the newly
created single-stranded regions, with one protein (blue) coating one
strand, and the other (orange) coating the other strand. At this point,
we do not know which protein promotes duplex invasion, so the
colors are arbitrary. (c) One of the proteins (blue) tags the coated free
39-end for invasion into a homologous duplex, initiating Holliday
complex formation. At this point, the Spo11-linked oligonucleotides
would dissociate and be degraded.
The second intriguing feature of Figure 22.22 is that the
two size classes of Spo11-linked oligonucleotides are produced in equal quantities. This suggests that the larger oligonucleotides come from one of the DSB ends, and the smaller
ones come from the other. This could mean that there is an
inherent asymmetry in the DSB that predetermines which
free 39-end will invade the other DNA duplex to initiate
Holliday junction formation. Keeney and colleagues envisioned a model similar to the one illustrated in Figure 22.23.
The asymmetry of cutting of the two strands leads to an
asymmetry in the lengths of the free 39-ends base-paired to
Spo11-linked oligonucleotides after resection. The one on
the right in this illustration is less tied up in base pairing, and
it could attract one of the two recombinases (Rad51 or
Dmc1), while the other would bind to the other recombinase.
727
This asymmetry could then dictate which free end invades
the homologous duplex to initiate Holliday junction formation. In this case, it is the one on the right.
We will see in the next section that a complex including
Rad50 and Mre11 is involved in resecting the DSB ends,
and the following evidence suggests that this complex also
contains the endonuclease that cuts the DNA near DSBs,
leading to the release of the Spo11-oligonucleotides: First,
we know that Mre11 has the required endonuclease activity; second, mutations in both the RAD50 and MRE11
genes block removal of Spo11 from DSB ends; and third,
the oligonucleotides attached to Spo11 have 39-hydroxyl
groups, which is consistent with the mechanism used by the
Mre11 endonuclease.
Now we know that the Spo11 gene is highly conserved
throughout the eukaryotic kingdom, including yeasts,
plants, and animals. Thus, it is very likely that the doublestranded break model for initiation of recombination is
also conserved. In one study that supports this conclusion,
Kim McKim and Aki Hayashi-Hagihara performed experiments similar to those of Kleckner and colleagues, looking
for mutations in Drosophila that blocked gene conversion.
In 1998, they reported that mutations in the mei-W68 gene
had this phenotype, and that mei-W68 is a Drosophila homolog of the yeast Spo11 gene. Interestingly, mei-W68 mutations affect gene conversion in somatic, as well as meiotic,
cells. Thus, in contrast to yeast, where Spo11 is required
only for meiotic recombination, mei-W68 is required for
both meiotic and somatic recombination. Thus, the doublestranded breaks that mei-W68 presumably induces appear
to be required for both meiotic and somatic recombination
in Drosophila.
SUMMARY The DSB model of meiotic recombina-
tion in yeast begins with a double-strand break.
DSBs can be directly observed in a plasmid DNA or
in chromosomal DNA. DSBs accumulate in rad50S
mutants, where Spo11 can be found covalently
bound to the DSB ends. Two molecules of Spo11
collaborate to create DSBs by cleaving both strands
at closely spaced sites. The cleavage operates
through transesterification reactions involving active site tyrosines on the two molecules of Spo11,
which leads to covalent bonds between the two
molecules of Spo11 and the newly formed DSBs.
Spo11 then could be released from the DSBs in a
complex with oligonucleotides ranging from about
12–37 nt long. The Spo11-oligonucleotides may
even be released after resection of the DSBs. The
cleavage by Spo11 appears to be asymmetric, yielding a longer free 39-end on one side of the DSB than
on the other. This may set up one free end for invasion of a homologous DNA duplex, which initiates
Holliday junction formation.