1
Homologous recombination induced by a replication fork barrier requires
1
cooperation between strand invasion and strand annealing activities
2
3
Léa Marie
1
and Lorraine S. Symington
1,2
*
4
1
Department of Microbiology & Immunology, Columbia University Irving Medical Center, New
5
York, NY10032
6
2
Department of Genetics & Development, Columbia University Irving Medical Center, New York,
7
NY10032
8
9
*Corresponding author: Lorraine S. Symington, e-mail: lss5@cumc.columbia.edu
10
11
12
13
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprintthis version posted August 20, 2021. ; https://doi.org/10.1101/2021.08.20.456128doi: bioRxiv preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprintthis version posted August 20, 2021. ; https://doi.org/10.1101/2021.08.20.456128doi: bioRxiv preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprintthis version posted August 20, 2021. ; https://doi.org/10.1101/2021.08.20.456128doi: bioRxiv preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprintthis version posted August 20, 2021. ; https://doi.org/10.1101/2021.08.20.456128doi: bioRxiv preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprintthis version posted August 20, 2021. ; https://doi.org/10.1101/2021.08.20.456128doi: bioRxiv preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprintthis version posted August 20, 2021. ; https://doi.org/10.1101/2021.08.20.456128doi: bioRxiv preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprintthis version posted August 20, 2021. ; https://doi.org/10.1101/2021.08.20.456128doi: bioRxiv preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprintthis version posted August 20, 2021. ; https://doi.org/10.1101/2021.08.20.456128doi: bioRxiv preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprintthis version posted August 20, 2021. ; https://doi.org/10.1101/2021.08.20.456128doi: bioRxiv preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprintthis version posted August 20, 2021. ; https://doi.org/10.1101/2021.08.20.456128doi: bioRxiv preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprintthis version posted August 20, 2021. ; https://doi.org/10.1101/2021.08.20.456128doi: bioRxiv preprint
2
ABSTRACT
14
15
Replication stress and abundant repetitive sequences have emerged as primary conditions underlying
16
genomic instability in eukaryotes. Elucidating the mechanism of recombination between repeated
17
sequences in the context of replication stress is essential to understanding how genome
18
rearrangements occur. To gain insight into this process, we used a prokaryotic Tus/Ter barrier
19
designed to induce transient replication fork stalling near inverted repeats in the budding yeast
20
genome. Remarkably, we show that the replication fork block stimulates a unique recombination
21
pathway dependent on Rad51 strand invasion and Rad52-Rad59 strand annealing activities, as well
22
as Mph1/Rad5 fork remodelers, Mre11/Exo1 short and long-range resection machineries, Rad1-
23
Rad10 nuclease and DNA polymerase δ. Furthermore, we show recombination at stalled replication
24
forks is limited by the Srs2 helicase and Mus81-Mms4/Yen1 structure-selective nucleases. Physical
25
analysis of replication-associated recombinants revealed that half are associated with an inversion of
26
sequence between the repeats. Based on our extensive genetic characterization, we propose a model
27
for recombination of closely linked repeats at stalled replication forks that can actively contribute to
28
genomic rearrangements.
29
30
31
INTRODUCTION
32
33
Maintaining genome integrity is essential for accurate transmission of genetic information and cell
34
survival. Replication stress has emerged as a major driver of genomic instability in normal and cancer
35
cells. Replication forks become stressed as a result of DNA lesions, spontaneous formation of
36
secondary structures, RNA-DNA hybrids, protein-DNA complexes, activation of oncogenes, or
37
depletion of nucleotides
1-3
. These obstacles to the progression of replication can cause forks to slow
38
down, stall and collapse. Consequently, multiple mechanisms have evolved to handle perturbed
39
replication forks to ensure genomic stability
4
.
40
In eukaryotes, the presence of multiple replication origins, including dormant origins that are
41
fired in response to replication stress, is one way to ensure complete genome duplication
5,6
.
42
Alternatively, the obstacle can be bypassed by translesion polymerases or by legitimate template
43
switching. The latter is a strand exchange reaction mediated by homologous recombination (HR)
44
proteins, consisting of annealing a nascent strand to its undamaged sister chromatid to template new
45
DNA synthesis
7
. In recent years, replication fork reversal has also emerged as a central remodeling
46
process in the recovery of replication in both eukaryotes and bacteria
8-12
. This process allows stalled
47
replication forks to reverse their progression through the unwinding and annealing of the two nascent
48
strands concomitant with reannealing of the parental duplex DNA, resulting in the formation of a four-
49
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprintthis version posted August 20, 2021. ; https://doi.org/10.1101/2021.08.20.456128doi: bioRxiv preprint
3
way-junction, sometimes called a chicken-foot structure. Consequently, the lesion can be bypassed
50
by extension of the leading strand using the lagging strand as a template followed by branch migration
51
of the reversed structure. Alternatively, the extruded nascent strands can undergo HR-dependent
52
invasion of the homologous sequence in the reformed parental dsDNA, resulting in the formation of a
53
D-loop to restart replication. In bacteria, the replisome is reassembled on the D-loop structure
13
,
54
whereas in eukaryotes DNA synthesis within the D-loop can extend to the telomere or be terminated
55
by a converging replication fork
5
. In addition, relocation of a lesion back into the parental duplex could
56
facilitate repair by the excision repair pathways
14
.
57
Thus, along with its critical role in DNA repair and segregation of chromosome homologs
58
during meiosis, HR is involved in multiple replication restart mechanisms, which contribute to the
59
preservation of genome integrity. However, HR can also be a source of instability as it occasionally
60
occurs between chromosome homologs in diploid mitotic cells, resulting in loss of heterozygosity.
61
Moreover, non-allelic HR (NAHR) between dispersed repeats can cause genome rearrangements
15-
62
18
. A significant factor underlying chromosome rearrangements is the abundance of repeated
63
sequences in eukaryotic genomes. Approximately 45% of the human genome is composed of
64
repetitive sequences including transposon-derived repeats, processed pseudogenes, simple
65
sequence repeats, tandemly repeated sequences and low-copy repeats (LCRs) distributed across all
66
chromosomes
19,20
. NAHR between repeated sequences can lead to deletions, duplications,
67
inversions or translocations
21-27
. Consequently, NAHR has been associated with many genomic
68
disorders
28,29
and is a major contributor to copy-number variation (CNV) in humans.
69
It is well established that rearrangements due to NAHR can result from the repair of double
70
strand breaks (DSBs)
30-34
. However, studies in yeast, human and bacteria have shown that such
71
genomic alterations can also arise during replication
18,24,35,36
. Notably, studies in
72
Schizosaccharomyces pombe have shown that a protein-induced, site-specific replication fork barrier
73
can cause a high frequency of genomic rearrangements in the absence of a long-lived DSB
74
intermediate
24,37
, consistent with the idea that replication stress contributes to NAHR. Elucidating the
75
molecular mechanisms of NAHR occurring during the processing and restart of stressed replication
76
forks remains crucial to understanding how genome rearrangements occur.
77
In Saccharomyces cerevisiae, spontaneous HR between repeated sequences shows different
78
genetic requirements depending on the genomic location of the repeats. Inter-chromosomal
79
recombination is generally Rad51 dependent, whereas recombination between tandem direct repeats
80
can occur by Rad51-independent single-strand annealing (SSA)
38
. It has been shown that repeats in
81
inverted orientation can spontaneously recombine by Rad51-dependent and Rad51-independent
82
mechanisms
39
, and these two pathways generate different recombination products. Rad51-mediated
83
recombination results in gene conversion, which maintains the intervening sequence in the original
84
configuration, whereas Rad51-independent recombination leads to inversion of the intervening DNA.
85
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprintthis version posted August 20, 2021. ; https://doi.org/10.1101/2021.08.20.456128doi: bioRxiv preprint
4
The inversion events require Rad52 and Rad59
40
, which are known to catalyze annealing of RPA-
86
coated single-stranded DNA (ssDNA) in vitro, and are required for SSA in vivo. Because DSB-induced
87
recombination between inverted repeats is dependent on Rad51
41
, it was proposed that the
88
spontaneous Rad51-independent inversions could be the result of annealing between exposed ssDNA
89
at stressed replication forks
42
.
90
To elucidate the mechanism of NAHR between inverted repeats in the context of replication
91
stress, we investigated the role of a protein-induced replication fork barrier in promoting inverted
92
repeats recombination. Previous studies have shown that the Escherichia coli Tus/Ter complex can
93
function as a DNA replication fork barrier when engineered into the genome of yeast or mouse cells
94
43-45
. Here, we demonstrate that a polar replication fork barrier engineered to induce fork stalling
95
downstream of inverted repeats is sufficient to trigger NAHR. Physical analysis of the recombinants
96
showed that gene conversion and inversion events were stimulated to the same extent. Unlike
97
spontaneous events, we found that replication-associated NAHR unexpectedly relies on a unique
98
pathway dependent on Rad51 strand invasion and Rad52-Rad59 strand annealing activities. We
99
discuss a model to account for dependence on both Rad51 and Rad52-Rad59 and formation of gene
100
conversion or inversion outcomes.
101
102
RESULTS
103
104
A polar replication fork barrier stimulates NAHR
105
To assess NAHR, we used a recombination reporter composed of two ade2 heteroalleles oriented as
106
inverted repeats
39
. The inverted repeat cassette was inserted at the HIS2 locus, 4 kb centromere
107
distal to the efficient ARS607 replication origin, on chromosome 6. The origin-proximal ade2-n allele
108
contains a +2 frameshift located 370 bp away from the stop codon and is transcribed by the native
109
ADE2 promoter. The origin-distal allele, ade2Δ5’, has a deletion of the first 176 nucleotides along with
110
the promoter. The two repeats share 1.8 kb of homology and are separated by 1.4 kb containing a
111
TRP1 gene transcribed by its native promoter (Fig 1A).
112
To analyze recombination in the context of a unique stressed replication fork, in the absence
113
of any genome-wide stress or global checkpoint activation, we took advantage of the galactose-
114
inducible Tus/Ter replication fork barrier
43,46
. We inserted 14 TerB repeats (hereafter referred to as
115
14 Ter) in the permissive or blocking orientation relative to ARS607, 120 bp or 170 bp distal to the
116
ade2Δ5’ repeat, respectively (Fig 1A). The location was selected based on a previous study showing
117
that Tus/Ter induces mutagenesis of the newly replicated region behind the stalled fork
47
. The P
GAL1
-
118
Tus cassette was integrated at the LEU2 locus.
119
In cells containing 14 Ter repeats in the blocking orientation, an elevated proportion of colonies
120
developing white sectors, indicative of an Ade
+
phenotype, was noticeable on plates containing
121
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprintthis version posted August 20, 2021. ; https://doi.org/10.1101/2021.08.20.456128doi: bioRxiv preprint
5
galactose (Fig 1B). Consistently, quantification of Ade
+
recombinants arising in this strain showed that
122
expression of the Tus protein stimulated recombination frequency from 0.62% to 8.08% (Fig 1C; Table
123
S1). We confirmed that the induction of the Tus protein expression had no effect on recombination
124
frequency in cells containing no Ter repeats or 14 Ter repeats in the permissive orientation (Fig 1C;
125
Table S1). By two-dimensional (2D) gel analysis of a 5 kb fragment encompassing part of the ade2
126
reporter and the Ter repeats, we confirmed that induction of the Tus protein expression generates a
127
significant replication fork arrest in the strain containing 14 Ter repeats in the blocking orientation (Fig
128
1D; Fig S1). Thus, replication fork stalling at a polar Tus/Ter barrier stimulates recombination between
129
inverted repeats, more than 10-fold. We investigated the nature of the Tus/Ter-induced events by a
130
PCR-based method (Fig 1E). Gene conversions and inversions were equivalently induced upon
131
expression of the Tus protein, representing 47.5% and 52.5% of the Ade
+
recombinants, respectively
132
(Fig 1F).
133
The role of genome-wide replication stress in stimulation of NAHR was assessed by growing
134
cells with the ade2 reporter on media containing DNA damaging agents known to induce replication
135
stress, namely, methyl methanesulfanate (MMS), camptothecin (CPT) or hydroxyurea (HU). Within
136
three days, an increased proportion of colonies containing white sectors, indicative of an Ade
+
137
phenotype, was clearly visible in the presence of MMS and CPT (Fig S2A). Consistently, quantification
138
of Ade
+
recombination frequencies under normal conditions (0.62% spontaneous recombination) and
139
genotoxic conditions (16.15% with MMS, 9.79% with CPT, 1.5% with HU) revealed a strong stimulation
140
of recombination between the inverted ade2 repeats in presence of MMS and CPT (Fig S2B). The
141
types of recombination events induced by MMS or CPT were determined by PCR analysis of
142
independent recombinants. In the presence of MMS, the frequency of gene conversions was 30-fold
143
higher (10.8%), whereas the frequency of inversions was increased by a factor 18 (5.4%). In presence
144
of CPT, the frequency of gene conversions was 9 times higher (3.2%), whereas inversions were
145
induced 22-fold (6.36%) (Fig S2C). We note that in the presence of CPT, the nature of the
146
recombination event of a small proportion of recombinants could not be easily determined by the PCR
147
method employed and these were not analyzed further. We detected a moderate induction of
148
recombination frequency by HU and the distribution of gene conversions and inversions appeared
149
similar to normal conditions (Fig S2B and C). Since replication fork arrest by a protein block is effective
150
in stimulating NAHR, we suggest that the attenuated induction of Ade
+
recombinants in response to
151
HU is due to the dNTP requirement for DNA synthesis associated with recombination-dependent fork
152
restart.
153
Together, these results indicate that NAHR between long inverted repeats, leading to gene
154
conversion or inversion of the intervening sequence, can be generated by genome-wide replication
155
stress or by a localized replication fork barrier, consistent with prior studies in S. pombe and mouse
156
cells
24,37,44
.
157
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprintthis version posted August 20, 2021. ; https://doi.org/10.1101/2021.08.20.456128doi: bioRxiv preprint