Schizosaccharomyces pombe Assays to Study Mitotic Recombination Outcomes

The fission yeast—Schizosaccharomyces pombe—has emerged as a powerful tractable system for studying DNA damage repair. Over the last few decades, several powerful in vivo genetic assays have been developed to study outcomes of mitotic recombination, the major repair mechanism of DNA double strand breaks and stalled or collapsed DNA replication forks. These assays have significantly increased our understanding of the molecular mechanisms underlying the DNA damage response pathways. Here, we review the assays that have been developed in fission yeast to study mitotic recombination.


Introduction
An accurate DNA damage response (DDR) is fundamentally important for cellular homeostasis. The DDR involves first detection of the damage and activation of the DNA damage checkpoints, then recruitment of the repair machinery and repair of the damage. In humans, defects in DDR can lead to a variety of syndromes and diseases [1].
The DNA double strand break (DSB) constitutes severing of the chromosome into two parts. Threatening the viability of a cell, DSBs can lead to loss of essential chromosomal regions. In multicellular organisms, unrepaired breaks may cause cells to undergo apoptosis and new cells can be regenerated [2,3]. However, inappropriate repair of breaks may give rise to translocations, deletions, duplications and inversions [4,5] which have been identified in cancer cells [6].
DSBs may be produced by endogenous or exogeneous sources but are generally repaired by the same mechanisms [7]. Most endogenous breaks occur as a consequence of replication stress. DNA replication forks can stall or collapse as they pass through heterochromatin regions [8,9], collide with RNA polymerases [10] or pass through other fragile sites characterized by repetitive elements or euchromatin to chromatin boundaries [11,12]. The recombination machinery evolved to rescue stalled or collapsed replication forks [13].
The multitude of repair mechanisms that eukaryotic cells evolved to deal with these breaks have been previously extensively reviewed [7,10,[14][15][16][17][18][19] and can be generally broken into two major pathways, non-homologous end joining (NHEJ) and homologous recombination (HR). NHEJ involves localized repair of breaks with no major sequence rearrangements. Repair can result in small deletions or alterations of the sequences neighboring the break so NHEJ is considered an error-prone form of repair [20]. HR can be subdivided into several mechanisms that are both genetically and biochemically related such as single strand annealing (SSA), break induced replication (BIR) and the two closely  (1). The DSB may be repaired by non-homologous end joining (NHEJ) (2). When repair occurs by homologous recombination (HR), the DSB is first resected (3) to expose areas of single stranded DNA. If direct repeats (green areas) exist on the same chromosome, the break may be repaired by single strand annealing (SSA) (4). If homology is not found, the chromosome may be entirely lost (5). When homology is found elsewhere, the broken ends may invade this region (6). In synthesis dependent strand annealing (SDSA) (7) the invading strand may copy a small region then release and re-anneal. In break induced replication (BIR) (8) the invading strand may copy to the end of the red chromosome. In this case the right part of the broken black chromosome is lost. Occasionally, a more complex double Holiday junction (dHJ) may be established (9), the resolution of which can result in crossovers (CO) or non-crossovers (NCO).
Schizosaccharomyces pombe (fission yeast) diverged from Saccharomyces cerevisiae (budding yeast) approximately a billion years ago [22,23]. Many genes have been identified in S. pombe that show similarity to genes involved in human disease [24]. This is comparable to the level of conservation between S. cerevisiae and humans [25]. However, S. pombe appears to show higher conservations in chromosome structure and function genes [26] making S. pombe a great model system for studying chromosomal dynamics.
Specific to the study of chromosomal double strand break repair, fission yeast is great for several reasons. First, DDR genes are highly conserved from yeast to humans [27,28]. Second, in fission yeast, repetitive DNA elements, which are often a reason for endogenous breaks in human cells [29], are found only at centromeres and telomeres [30]. The overall structure of centromeres is conserved between yeast and humans. We and others have previously shown that replication trough the centromere does lead to chromosome breakage [31][32][33][34][35]. The presence of these repeats in only a few well characterized regions allows the better monitoring of the events that govern DSB repair. Third, haploid fission yeast has only three chromosomes [36] making it easier to track translocations and other aberrations resulting from inappropriate break repair. Fourth, fission yeast is as well suited for genetic screens as baker's yeast [37]. Finally, working with fission yeast is inexpensive and easy to learn.
A plethora of assays have been developed to study DNA damage in various systems and some are described in several recent elegant reviews [38][39][40][41]. Here, we focus on S. pombe and review primarily the in vivo assays that have been developed in this system to study mitotic recombination. We briefly summarize these assays and direct the reader to the original publications for more details.  (1). The DSB may be repaired by non-homologous end joining (NHEJ) (2). When repair occurs by homologous recombination (HR), the DSB is first resected (3) to expose areas of single stranded DNA. If direct repeats (green areas) exist on the same chromosome, the break may be repaired by single strand annealing (SSA) (4). If homology is not found, the chromosome may be entirely lost (5). When homology is found elsewhere, the broken ends may invade this region (6). In synthesis dependent strand annealing (SDSA) (7) the invading strand may copy a small region then release and re-anneal. In break induced replication (BIR) (8) the invading strand may copy to the end of the red chromosome. In this case the right part of the broken black chromosome is lost. Occasionally, a more complex double Holiday junction (dHJ) may be established (9), the resolution of which can result in crossovers (CO) or non-crossovers (NCO).

Mini-Chromosome Assays
Schizosaccharomyces pombe (fission yeast) diverged from Saccharomyces cerevisiae (budding yeast) approximately a billion years ago [22,23]. Many genes have been identified in S. pombe that show similarity to genes involved in human disease [24]. This is comparable to the level of conservation between S. cerevisiae and humans [25]. However, S. pombe appears to show higher conservations in chromosome structure and function genes [26] making S. pombe a great model system for studying chromosomal dynamics.
Specific to the study of chromosomal double strand break repair, fission yeast is great for several reasons. First, DDR genes are highly conserved from yeast to humans [27,28]. Second, in fission yeast, repetitive DNA elements, which are often a reason for endogenous breaks in human cells [29], are found only at centromeres and telomeres [30]. The overall structure of centromeres is conserved between yeast and humans. We and others have previously shown that replication trough the centromere does lead to chromosome breakage [31][32][33][34][35]. The presence of these repeats in only a few well characterized regions allows the better monitoring of the events that govern DSB repair. Third, haploid fission yeast has only three chromosomes [36] making it easier to track translocations and other aberrations resulting from inappropriate break repair. Fourth, fission yeast is as well suited for genetic screens as baker's yeast [37]. Finally, working with fission yeast is inexpensive and easy to learn.
A plethora of assays have been developed to study DNA damage in various systems and some are described in several recent elegant reviews [38][39][40][41]. Here, we focus on S. pombe and review primarily the in vivo assays that have been developed in this system to study mitotic recombination. We briefly summarize these assays and direct the reader to the original publications for more details.

Mini-Chromosome Assays
Mini-chromosome assays have been instrumental in elucidating many recombination pathways. Chromosome III (Ch.III), the smallest S. pombe chromosome, houses rDNA repeats at both ends next to the telomeres and has the longest centromere [36] (Figure 2A). Using γ irradiation, Niwa et al. [42,43] isolated a truncated Ch.III (termed Ch 16 ) in haploid yeast ( Figure 2B). Unlike S. cerevisiae which has point centromeres [44], S. pombe centromeres are characterized by repetitive elements resembling higher metazoans [30]. Since both the centromere and the rDNA repeats fluctuate in number between strains, the size of Ch.III cannot be determined precisely making it approximately 3.5 Mb. Initial genetic maps estimated the centromere to be 100 Kb [45][46][47] but latter structural analysis identified a 110 Kb centromere [48]. The fission yeast genome sequence [24] set the centromere size at 68 Kb with the caveat that seven 6760 bp repeats were missing. The addition of these repeats to the 68 Kb region brings the size of CEN III to 110 Kb which is what PomBase (the scientific database for fission yeast) reports [49,50]. The centromere spans between the meu27 + and ppc1 + loci (Figure 2A).
The truncated mini-chromosome Ch 16 should retain an intact Ch.III centromere but not have rDNA repeats. Pulse field gel electrophoresis shows that this mini-chromosome is about 530 Kb [42,43,51]. The chromosome can be maintained in haploid yeast by intra-allelic complementation between ade6-M210 (Ch.III) and ade6-M216 (Ch 16 ) (Figure 2A,B). Genetic mapping by Niwa et al. [42] showed that the left arm does not include the ade10 + locus but the yps1 + locus is present [51]. There are approximately 30 Kb between yps1 + and ade10 + and 84 Kb between yps1 + and meu27 + , so it appears that the left arm is about 110 Kb. On the right arm, ade6 + is 172 Kb from ppc1 + . Genetic mapping shows the presence of the ags1 + locus (formerly tps16 + [52]) 58 Kb telomere proximal from ade6 + but not tps14 + which is approximately 200 Kb from ags1 + [42]. cid2 + is within 5 Kb of ags1 + . Southern blotting has shown the presence of the SPCC61.05 locus 75 Kb from ags1 + [53]. Thus, the right arm is about 305 Kb.
Niwa et al. determined that Ch 16 is stable in mitotically dividing haploid cells at one copy per cell but unstable at two copies [42]. Remarkably, cells were able to simultaneously stably propagate both Ch 16 and a 100 Kb shorter derivative (Ch 16D1 ) suggesting that yeast cells may be able to determine ploidy by the size of the chromosome. Ch 16 loss monitored by appearance of adenine auxotrophs occurred at a frequency of 1 in 10 4 cells.
The Humphrey lab has produced several derivatives of Ch 16 to study recombination outcomes in vivo. They initially placed the S. cerevisiae homothallic endonuclease (HO) restriction site (MATa) marked with the kanamycin antibiotic resistance gene (KAN) at the rad21 + locus 27 Kb telomere distal of ade6-M216 producing Ch 16 -MG ( Figure 2C) [54]. KAN confers resistance to G418 (Geneticin). Upon induction of a single DSB, gene conversion of the KAN locus can be monitored by screening for adenine prototrophic and G418 sensitive (ade+G418 S ) colonies while chromosome loss will lose both markers. Long track gene conversion produces the same phenotype as chromosome loss, so it was necessary to distinguish between the two by pulse field gel electrophoresis (PFGE). Similarly, because an ade + G418 R phenotype could result from repair by NHEJ or failure of HO cutting, it was necessary to sequence across the DSB site.
Ch 16 -MG was improved by adding the his3 + marker at the cid2 + locus approximately 25 Kb from the MATa site creating Ch 16 -MGH ( Figure 2D) [53]. The advantage of this is that BIR can be investigated by monitoring the loss of both KAN and his + markers but retention of the adenine prototrophy (ade+G418 S his-). In BIR, the break is repaired by copying the missing information from Ch.III which is his3 − . PFGE followed by southern blotting showed both a larger mini-chromosome termed Ch x and the presence of the ade5 + marker, which could have only been transferred onto the Ch 16 from Ch.III by BIR. A mini-chromosome with MATa-HPH marker (Ch 16 -MHH) was also constructed which behaved identically to Ch 16 -MGH. HPH confers resistance to hygromycin abbreviated as HYG in this paper. Bioneer has generated a deletion mutant library marked with KanMX [55] and this Ch 16 -MHH chromosome is a powerful tool in screening genes that affect repair. loss of all markers on the right arm but retention of the arg3 + marker. Thus, the arg3 + marker distinguishes.  A third marker (arg3 + ) was introduced on the left arm of Ch 16 -MGH at the yps1 + locus 84 Kb from the meu27 + locus. An additional modification was moving the MATa site 33 Kb left of the ade6-M216 marker. This created Ch 16 -RMGAH ( Figure 2E) [51] which can monitor extensive BIR that results in loss of all markers on the right arm but retention of the arg3 + marker. Thus, the arg3 + marker distinguishes.
Extensive BIR from chromosome loss. To ensure that the new placement of the MATa and arg3 + marker on the left arm did not cause a locus-specific form of repair, they generated chromosome Ch 16 -YAMGH where the MATa site was left at the rad21 + locus and an HPH marker was put 4 Kb from the centromere ( Figure 2F). Remarkably, PFGE analysis of the arg+G418 S ade-his-or HYG R G418 S ade-hiscolonies showed that the two outcomes were identical suggesting that the mechanism of repair was independent of the break position.
Further analysis showed that following break induction, telomere distal sequences of the break are lost while extensive resection of the centromere proximal sequences occurs. Inverted centromeric repeats facilitate formation of an isochromosome which results from invasion and duplication of both the centromere and the left arm of the chromosome onto the right [51]. Nakamura et al. [33] constructed the ChL mini-chromosome ( Figure 2G) to study these isochromosomes and they found that they can occur in the absence of a break. Both groups showed that Rad51 suppresses isochromosome formation [35,51]. Using the ChL chromosome, we have also shown that perturbation of the replication machinery or the centromere heterochromatin increases these isochromosome events. This suggests that isochromosomes originate from improperly repaired random breaks that occur during replication [31]. Remarkably, we also identified some more drastic copy number variations in double chromatin and replication mutants which entirely lost Ch.III and acquired other smaller chromosome fragments. The structure and mechanism of these smaller chromosome fragments remains to be elucidated.
Finally, Ch 16 -LMYAU was generated to screen Bioneer mutants for chromosome loss phenotypes ( Figure 2H) [56]. Previously, the HO endonuclease was propagated on a plasmid but here the investigators introduced it at the yps1 + locus marked with the S. cerevisiae LEU2. The ura4 + marker was placed at the cid2 + locus. Chromosome loss was assayed by red/white sectoring in response to break induction. Detailed experimental procedures for using these assays were published [57]. Although we do not discuss it here, a chromosome loss assay in diploid cells has also been designed [58]. In addition to chromosome loss, this assay also monitors intra-homologue recombination.

Recombination at Repetitive Elements
Another series of assays have been devised in S. pombe to study chromosomal recombination at non-tandem repeats ( Figure 3). Repetitive elements had been previously shown to cause deletions, inversions or duplications [59,60]. It was known from studies in several species including S. cerevisiae that certain genetic mutations act as hotspots for meiotic recombination [61,62]. The ade6-M26 (G135T) allele in S. pombe creates such a hotspot [63]. Remarkably, the ade6-M375 (G132T) and the ade6-L469 (C1467T) as well as several other alleles did not serve as recombination hotspots. The G135T, G132T and C1467T mutations inactivate the gene because they introduce stop codons.
Schuchert and Kohli designed a clever non-tandem repeat assay to study crossover frequency at the ade6-M26 locus ( Figure 3A) [64]. They positioned the ade6-L469 3 end mutation on the left side and the ade6-M26 or ade6-M375 5 end mutation on the right side while placing the functional ura4 + gene in between. Then, the investigators assayed for deletion or conversion. Deletion (ade+ura-) can be predicted to arise by several mechanisms. Unequal sister chromatid exchange should generate both ade-ura+ and ade+ura-phenotypes ( Figure 3A(1)). A break between the two ade6 alleles, followed by resection in both directions past the mutations and annealing can also generate a deletion through SSA ( Figure 3A(2)). Since both ade6 mutations are proximal to the ura4 + gene, it is possible to reconstitute a wildtype ade6 + allele. Deletion may also occur through an intra-chromosomal crossover and loopout ( Figure 3A(3)). An ade+ura+ phenotype arises by replacing the mutation in one allele with the wildtype region of the other allele. The mechanism by which this happened remained elusive until later when the Whitby lab proposed a fork regression and mismatch repair model (see below).
To study mitotic recombination induced by a double strand break the Subramani group introduced the MATa site either in the unique sequence between ura4 + and ade6-M26/M375 or within the ade6-L469 allele ( Figure 3B) [65,66]. Upon induction of the DSB, two outcomes predominated (ade+ura-and ade-ura-) which the authors explained to arise by SSA. Subsequent genotyping by restriction digestion or backcrossing identified the exact allele in the ade-ura-phenotypes. ade+ura+ and ade-ura+ recombinants (not shown) also arose at a much lower rate and only when the HO restriction site was placed within the ade6-L469 allele suggesting that the position of the break determines the repair mechanism. These assays were used to analyze the role of several recombination genes in DSB repair [65].
The Whitby lab has subsequently been influential in designing several variations of the tandem repeat assay and have been able to explain some of the more complex outcomes. These assays have been used to identify and characterize genes involved in DNA damage repair [67][68][69][70][71][72][73][74][75][76][77][78]. The investigators initially designed a system similar to the Schuchert and Kholi assay except that instead of ura4 + they placed a his3 + marker between the ade6 alleles ( Figure 3C) [79]. This assay was used to show that the ura+his+ recombination outcomes may arise due to recombination dependent restart of stalled replication forks usually through BIR. Inactivation of various cellular processes such as Holliday junction resolution or the checkpoint changes the recombination outcomes [70,[79][80][81][82]. In a more recent report [83], the Whitby lab introduced a nicking site for the gpII M13 phage enzyme. This enzyme makes a nick that is converted to a DSB by the replication fork which approaches from the right (see Figure 3D,E) [84,85]. This assay was used to study restart of stalled replication forks due to the DSB produced by this nick.
To further investigate the effect of replication fork stalling on recombination outcomes, the Whitby lab modified the assay by placing RTS1, the naturally occurring replication fork termination site at the mating type loci [86], either between the his3 + and the ade6-M375 alleles or within the ade6-L469 allele ( Figure 3D) [87,88]. RTS1 is polar meaning that it can only stall forks in one direction, so strains were constructed with different RTS1 orientations, RTS1-AO (active orientation) and RTS1-IO (inactive orientation). A cluster of origins of replication are found to the right of the ade6-M375 allele and the replication fork is predicted to approach from the right ( Figure 3E). Two-dimensional gel electrophoresis showed that the RTS1-AO efficiently stalls replication forks [89]. The investigators used this assay to study the function of various recombination genes in fork restart and to show that rescue of collapsed replication forks can cause BIR dependent template switching that can generate chromosomal rearrangements. [76,87,88,90,91]. A protocol was published with extensive details on the use of these elegant assays [92].
By inserting the ade-his-ade cassette at different distances from the RTS1 pause site ( Figure 3E) the investigators showed that template switching can occur up to 75 Kb from collapsed forks [93]. Collisions between Pol III which transcribes tRNA and replication can cause chromosomal rearrangements [94]. A tRNA GLU gene was inserted between the his3 + and the ade6-M375 allele to show that collisions between Pol III and replication machinery increases the frequency of recombination ( Figure 3F) [91,93]. An increase in template switching was observed when both RTS1 and tRNA GLU were introduced in the same construct and the orientation of transcription of tRNA GLU faced RTS1 head on [93]. To monitor interaction of fluorescently tagged recombination genes (rad52 + , rad51 + , etc.), the LacO array was also introduced at different positions within the ade-his-ade repeat ( Figure 3G) [88,89,95,96].
induction of gene expression in S. pombe that we do not discuss here have also been engineered more recently [102,103].
We also designed an assay to study intrachromosomal deletions at direct repeats ( Figure 3I) [31,104]. A functional his3 + gene was placed between two truncated ura4 alleles with 200 bp of overlapping sequence. We showed that this assay can only detect deletions and not conversion. An assay to study recombination at non-tandem direct repeats. The ade6-L469 and ade6-M26 or ade6-M375 inactivating alleles were placed on either side of a functional ura4 + gene. White dots represent the positions of the mutations in the ade6 alleles. The strain is adenine auxotrophic and uracil prototrophic (ade-ura+). Different repair pathways lead to various phenotypic outcomes. Unequal sister chromatid exchange (1) produces both ade+ura-and ade-ura+ phenotypes while single strand annealing (2) or loop out (3) produces only ade+ura-. ade+ura+ phenotypes also arose for which the Whitby lab proposed a fork regression and mismatch repair model. (B) An assay similar to (A) but with the S. cerevisiae HO endonuclease restriction site (MATa) either between the ura4 + and the ade6-M26/375 alleles or in the ade6-L469 allele. Shown are phenotypes that may arise by single strand annealing. (C) An assay similar to (A) except that ura4 + was replaced with his3 + . The assay was further modified by Osman et al., 2016 to introduce a nick site between the his3 + and ade6-M375 repeat. The nick is created by the M13 bacteriophage gpII enzyme. A replication fork which approaches from the right converts this nick into a DSB. (D) The assay in (C) was modified by introducing the RTS1 pause site (black arrows) in between the his3 + and ade6-M375 or within the ade6-L469 allele. Because RTS1 is polar, two different strains were made for RTS1(A) and RTS1(B) each with a different orientation AO = active orientation, IO = inactive orientation) of the pause site. (E) The RTS1 was cloned near ori1323 while the ade-his-ade cassette was placed at different distances (indicated in kilobases) from the RTS1. The relative positions of endogenous replication origins are also shown. (F) An assay to test collisions of replication machinery with RNA Pol III. The tRNA GLU gene was placed in either of the two orientations between the his3 + and the ade6-M375 allele while the RTS1 pause site was placed on the right side of ade6-M375. (G) Three LacI arrays using the ade-his-ade cassette to allow microscopic visualization of repair dynamics. (H) An assay to study single strand annealing. Two LEU2 fragments with overlapping regions (black arrows) were placed on either side of a functional his3 + gene. The MATa site was placed on the 5 end of the his3 + gene. (I) An assay similar to (H) except that his3 + is flanked by ura4 with overlapping regions (arrows) instead of LEU2. The MATa site was placed on the 3 end of his3 + .
An assay to study SSA in S. pombe was designed by Watson et al. (Figure 3H) [97]. Two S. cerevisiae LEU2 fragments with overlapping regions were placed on either side of a functional his3 + gene. The MATa sequence was cloned at the 5 end of his3 + right before the start codon. Using this assay, the investigators showed that SSA is rad52 + dependent, confirming previous findings. In their report Watson and colleagues also designed an elegant system that allows fast transcriptional induction in S. pombe. Historically, transcriptional induction in S. pombe relied on the nmt1 promoter which is repressed by thiamine. Removal of thiamine de-represses the promoter but it takes anywhere between 14-20 h for full induction [98,99]. The new system which is based on upregulation of the urg1 promoter allows induction within 30 min mirroring the S. cerevisiae GAL induction system [100]. The urg1 system was optimized in a subsequent publication [101]. Other systems for faster induction of gene expression in S. pombe that we do not discuss here have also been engineered more recently [102,103].
We also designed an assay to study intrachromosomal deletions at direct repeats ( Figure 3I) [31,104]. A functional his3 + gene was placed between two truncated ura4 alleles with 200 bp of overlapping sequence. We showed that this assay can only detect deletions and not conversion.

Chromosomal Rearrangements Caused by Stalled or Collapsed Replication Forks at Inverted Repeats
A series of other assays have been designed to study chromosomal rearrangements resulting from stalled replication forks by the Carr, Murray and Lambert labs. Carr and colleagues placed the RTS1 on either side of ura4 + gene on Ch.III ( Figure 4A) [105]. Using 2-D gel electrophoresis they showed that these constructs can efficiently stall forks in the vicinity of the ura4 + gene. Deletion of several recombination genes including rad51 + decreases cell viability suggesting that homologous recombination is required for rescue of stalled forks. PFGE and PCR showed that some of the outcomes resulted in ura4 + loss through gene conversion without crossover, while others through a crossover between Ch.III and Ch.II produce a reciprocal translocation ( Figure 4B). In all cases, information was exchanged between Ch.III RTS1 and the endogenous Ch.II RTS1. To monitor anaphase bridges by microscopy, the Lambert lab also placed a LacO array next to the RTS1 pause site ( Figure 4C) [106].
constructs, it was shown that fork slippage, but not translesion synthesis or mismatch repair, is responsible for the restoration of the functional ura4 + cassette. The intra-S phase DNA damage checkpoint (Rad3) represses fork slippage and microhomology mediated repair at stalled forks [77]. Further, the NHEJ factor Ku appears to regulate recombination at arrested forks by controlling end resection [112].  1) The RTS1 pause site was placed on the right of a functional ura4+ gene (uraR), (2) left (Rura) or (3) both sides (RuraR). Fork stalling was assayed using pulse field gel electrophoresis (PFGE). (B) Chromosomal rearrangements of constructs in (A) can occur by two mechanisms: gene conversion without crossover and gene conversion with crossover. In both cases there is a recombination event between the RTS1 site next to the ura4 + cassette and the endogenous RTS1 at the MAT locus on Ch.II. Both types of events result in loss of ura4 + . (C) Modification of construct in (A) to monitor chromosomal dynamics by fluorescence microscopy. A LacO array was placed on the telomere side of the RTS1/ura4 + cassettes. (D) Repetitive elements constructs to study chromosomal rearrangements. (1) RTS1 cassettes were placed on both sides of two directional ura4 + repeats (Ru(dir)uR). (2) RTS1 was placed on both sides of inverted ura4 + repeats spaced by 14 bp of unique sequence (RuiuR). A construct without the 14bp spacer was also made (RuuR) (3) In oRuiuR the RTS1 is in opposite direction from the RuiuR construct. (4) In the RuhR construct, one of the ura4 + repeats was replaced with his3 + . (E) Several constructs to test the effect of repeat length and the position of the replication pause sites in producing rearrangements. (1) Two ura4 + inverted repeats spaced by 14 bp of unique sequence were engineered with 0.2 Kb (Rpal1R) or 1.8 Kb (Rpal2R) his3 + sequence between the right ura4 + repeat and the RTS1 pause site. An additional short ura4 + was placed between the left RTS1 and the ura4 + repeat (yellow arrow). (2) Tpal1R and Tpal2R constructs are similar to constructs in (1) except that the left RTS1 was replaced with three TER2/3 sequences. (3) In this construct the right ura4 + repeat was replaced by a unique sequence. (4) Each of the ura4 + repeat sizes were varied between 30-1200 base pairs. (5) The spacer between the inverted repeats was varied between 7-250 base pairs. (6) ura4 + cassette flanked by the TER2/3 termination sites. (7). Direct repeats but one of the ura4 + repeats was replaced with a ura5 + repeat. (F) The ura4 + and RTS1 pause sites were placed next to known origins of replication (3004, 3005, 3006/7). (1) Co-directional ura4 + and RTS1 with RTS1 placed either after (1) or in front (2) Figure 4D) [107]. Using these constructs, the investigators showed that recombination dependent rescue of stalled replication forks at inverted repeats can produce dicentric and acentric isochromosomes. In subsequent and even more sophisticated studies, the investigators generated a series of more complex constructs to precisely analyze the mechanisms of chromosomal rearrangements at inverted repeats ( Figure 4E) [108]. The length of the repeats as well as the gap between the repeats was varied, unique sequences were introduced at either side of the repeats, RTS1 sites were placed at different distances from the repeats and the TER2/3 ribosomal fork barrier was also tested. Their findings showed that inverted repeats cause forks to turn around or "execute a U-turn" and generate gross chromosomal rearrangements. Remarkably, the DNA damage checkpoint does not appear to detect the recombination intermediates that cause these rearrangements, at least in the cell cycle in which they occur [109].
The position of the RTS1 was varied relative to several Ch.III replication origions ( Figure 4F) [110,111] to show that when a replication fork approaching from one of these origins collides with the RTS1 site, it causes deletions in addition to gross chromosomal translocations. To test for replication fork slippage, the investigators designed a construct where the ura4 + gene was interrupted by 20 bp repeats flanked by 5 bp of microhomology sequences ( Figure 4G) [77,110]. A slightly modified construct has 22 bp repeats flanked by 4 bp of microhomology ( Figure 4G). Both constructs inactivate the ura4 + gene. Microhomology mediated repair results in a functional ura4 + . Using these constructs, it was shown that fork slippage, but not translesion synthesis or mismatch repair, is responsible for the restoration of the functional ura4 + cassette. The intra-S phase DNA damage checkpoint (Rad3) represses fork slippage and microhomology mediated repair at stalled forks [77]. Further, the NHEJ factor Ku appears to regulate recombination at arrested forks by controlling end resection [112].

Mating Type Loci Serve as a Natural Site for Studying Collapsed Replication Forks
The mating type of a S. pombe cell is determined by the allele present at the mat1 cassette ( Figure 5A) [113]. This allele can be either mat1M (M cell, M stands for minus) or mat1P (P cell, P stands for plus) [114,115]. Switching between mat1M and mat1P is accomplished by copying information from the silent mat2-P and mat3-M cassettes [116][117][118][119]. In addition to the allele present at the mat1 locus, the mating type of a population of cells is also determined by the ability to switch and the alleles present at the silent mat2-P and mat3-M regions. Wild type h 90 cells can switch information at the mat1 locus and are mixture of M and P cells [120][121][122][123]. A population that has the M allele at the mat1 locus (mat1M) and lost the ability to switch is h -. A population may also be hif it lost the mat2-P locus and repairs the mat1M with the same information (e.g., from mat3-M). Alternatively, a population of cells are h + if the P allele is expressed from the mat1 locus (mat1P) and the strain is unable to switch (e.g., mat1P∆17). The mating type region has three cassettes mat1, mat2-P(plus) and mat3-M(minus). Only mat1 is expressed with either P or M while mat2-P and mat3-M are silenced. The shading of mat2-P and mat3-M diagrams represents this silencing. Information switches from mat2-P and mat3-M onto mat1 so that mat1 expresses either P or M. All three cassettes are flanked by homologous regions (H1, H2) that facilitate recombination. In this diagram, we begin with a hypothetical example where mat1 expresses the M information. After one round of replication, an imprint consisting of insertion of one or two ribonucleotides or a single strand nick is made next to the mat1-M cassette on the lagging strand (blue star). For simplicity, the top diagram does not have the imprint. In the second round of replication, the leading strand collides with the imprint and produces a blunt one-ended DSB. To ensure that only the leading strand passes through the imprint, the replication fork from the other direction is paused and terminated by the RTS1 site. The one-ended break is repaired by copying information from the mat2-P or mat3-M regions. Since in our example the M information was expressed at mat1, the break may be repaired using mat2-P and the information at mat1 switches. Note that the imprint is always placed on the lagging strand with each DNA replication. (B) Exploiting the mating type loci to study The mating type region has three cassettes mat1, mat2-P(plus) and mat3-M(minus). Only mat1 is expressed with either P or M while mat2-P and mat3-M are silenced. The shading of mat2-P and mat3-M diagrams represents this silencing. Information switches from mat2-P and mat3-M onto mat1 so that mat1 expresses either P or M. All three cassettes are flanked by homologous regions (H1, H2) that facilitate recombination. In this diagram, we begin with a hypothetical example where mat1 expresses the M information. After one round of replication, an imprint consisting of insertion of one or two ribonucleotides or a single strand nick is made next to the mat1-M cassette on the lagging strand (blue star). For simplicity, the top diagram does not have the imprint. In the second round of replication, the leading strand collides with the imprint and produces a blunt one-ended DSB. To ensure that only the leading strand passes through the imprint, the replication fork from the other direction is paused and terminated by the RTS1 site. The one-ended break is repaired by copying information from the mat2-P or mat3-M regions. Since in our example the M information was expressed at mat1, the break may be repaired using mat2-P and the information at mat1 switches. Note that the imprint is always placed on the lagging strand with each DNA replication. (B) Exploiting the mating type loci to study rescue of collapsed forks. Collision of the leading strand with the imprint produces a blunt one-ended break that can be studied in WT and other control strains. h 90 is a wild type strain capable of switching that represents a mix of M and P cells. The smt-0 has a deletion that eliminates the imprint while M2,3∆ have deletions in the mat2-M and mat3-P regions. smt-0 M2,3∆ is a double mutant. The P∆17 mutant is similar to smt-0 except that in this strain the mat1 locus expresses the P information. (C) An assay to study silencing at the mating type loci. The functional ura4 + gene was places 150 base pairs from mat3-M. If repressed, the strain can grow on 5-fluoroorotic acid (5-FOA) which negatively selects against uracil prototrophs and die on minimal media lacking uracil. If de-repressed, the strain will die on 5-FOA and grow on minimal media lacking uracil.
In wild type populations, switching occurs in only 25% of the cells [124]. This is because unlike in S. cerevisiae where switching is initiated by a double strand break generated by the HO endonuclease [125], in S. pombe switching begins with an imprint in the lagging strand during an initial round of DNA replication. Two concurrent models exist to explain the nature of the imprint. The imprint may be caused by the introduction of one or two ribonuclotides in the DNA sequence [126][127][128] or the generation of a single strand break [129] or possibly both at the same time. Regardless, the imprint is converted to a one-ended double strand break in the next round of replication [129][130][131][132][133]. However, only one of the replication forks may convert the imprint into a DSB. To ensure that the other replication fork does not pass through the imprint, replication is terminated by the above mentioned RTS1 site [86,134]. This process allows unidirectional transfer of information from the two mat2-P and mat3-M cassettes onto mat1 in only one of the four cells.
Investigation of the mating type loci in S. pombe, which increased our understanding of replication, recombination and gene silencing has been reviewed previously [120,135]. This natural system is a remarkable assay for the study of the role of DNA recombination in restart of collapsed replication forks. Perhaps even more important is the fact that the system produces one-ended DSBs which more accurately resemble the type of substrates resulting from collapsed replication forks [136]. Using wild type cells or several mating type defective mutants ( Figure 5B) [130,137,138] the role of various DNA damage response genes have been investigated [136,139,140]. Work has also led to the identification of several recombination and replication genes, particularly the swi (switching) genes [117,118,141,142]. Additionally, various mechanisms of gene silencing were identified (reviewed in reference [143]). For example, unlike the centromeres, the mating type loci are silenced through an RNAi independent mechanism [144][145][146]. Work on silencing led to the identification of the clr (cryptic loci regulator) genes [147][148][149]. To investigate silencing, Thon and Klar placed the ura4 + gene 150 bp from the mat3 cassette [147] (Figure 5C). Finally, as Klar et al. point out in their review [135], the asymmetric mating type switching in S. pombe may also explain how genes are differentially regulated during development of higher eukaryotic organisms.

Other Fluorescence and Biochemical Assays
To investigate centromere dynamics, Nabeshima et al. placed a LacO array at the lys1 + locus, 30 Kb from the centromere ( Figure 6A) [150]. The Russell lab modified this assay by placing the HO endonuclease restriction site 1.5 Mb away at the arg3 + locus within 2.8 Kb of the LacO array ( Figure 6B) [151]. Using these assays, they showed that Crb2-YFP (yellow fluorescent protein) co-localizes with the LacI-GFP (green fluorescent protein) bound to the LacO array suggesting that Crb2 interacts with the chromosome at the site of the DSB. When the break was at the arg3 + locus, no co-localization was observed because the Crb2-YFP interacts with a DNA sequence too far from the LacO array. Several other constructs were made without the LacO array to investigate foci of various fluorescent-tagged proteins in response to a DSB ( Figure 6B). For finer analysis of repair proteins interacting with the break, the investigators next turned to Chromatin IP. Primers were designed to amplify sequences by PCR at 0.2, 2.0, 9.0 and 16.0 Kb and the interaction of a variety of proteins with the break were tested biochemically ( Figure 6C) [152,153]. Using these assays, the investigators have unraveled the role of various DNA damage checkpoint and repair proteins in processing of DSBs [136,152,[154][155][156][157][158][159][160]. Not discussed here are some other biochemical assays that the Russell lab engineered, including some assays to study resection [161] with some adaptations from S. cerevisiae [162].  the HO restriction site anywhere in the yeast genome. Please see text for details. (E) An assay to study YFP tagged proteins interacting with a DSB. A lacO array that interacts with LacI-mCherry was placed 30 Kb from the HO endonuclease restriction site. The strain also contains Rad52 fused with cyan fluorescent protein (Rad52-CFP). A library of YFP tagged proteins was transformed in the cells and colocalization of all three foci (red, blue, yellow) was monitored.
Another ingenious assay was designed by Leland and King to introduce a LacO array and the HO restriction site anywhere in the genome ( Figure 6D) [163]. This technique requires several steps. First, the HO restriction site marked with the HPH marker and flanked by two homology regions is placed at a desired location in the genome ( Figure 6D(1)). Then, a sequence harboring the LacO array marked with ura4 + and flanked by the same homologous sequences as the HO restriction site is integrated at the HO locus ( Figure 6D(2)). The HO restriction site is re-integrated at a location neighboring the LacO array ( Figure 6D(3)). This produces a construct with an HO restriction site next to the LacO array ( Figure 6C(4)) similar to the Du et al. assay ( Figure 6B). The advantage of this technique is that the LacO and the HO restriction site can be introduced anywhere in the genome.
Yu et al. has designed an assay to screen for proteins interacting with a DSB ( Figure 6E) [164]. They engineered the HO endonuclease restriction site 20 Kb from a LacO array that binds LacI-Cherry. The strain also expresses Rad52-CFP (cyan fluorescent protein) which interacts with DSBs. Upon DSB induction, co-localization of red and blue foci indicates that Rad52-CFP interacted with the DSB. When they transformed a library of S. pombe YFP tagged proteins [165], they were able to screen for other proteins that interact with the DSB by monitoring co-localization of all three foci (red, blue, yellow).

Non-Homologous Repair
Non-homologous repair can occur in the absence of homologous regions. Several assays have been designed to test non-homologous repair in S. pombe. In an NHEJ assay designed by Goedecke and colleagues, a plasmid is linearized with various restriction sites to produce non-homologous ends ( Figure 7A) [166]. The restriction sites used to generate one of the ends are different from those used to generate the other end, thus ensuring that the ends cannot be re-ligated. The linear fragment is transformed into living cells and allowed to re-circularize. The junction is amplified by PCR and sequenced. Using this assay, the investigators showed that the ends are resected up to 14 bps prior to rejoining. A similar assay by Manolis et al. relies on transforming linearized sequences with just one restriction enzyme (PvuII) ( Figure 7B) [167]. Subsequent sequence analysis of rejoined products identified various sequence alterations at the junction. The investigators also showed that in S. pombe rad50 + , mre11 + and the DNA damage checkpoint is not required for NHEJ. These plasmid type NHEJ assays were instrumental in demonstrating that key components of NHEJ are conserved form yeast to humans [168,169].
In a variation of these assays, Anabelle Decottignies used PCR to amplify a linear fragment ( Figure 7C) [170]. The fragment was transformed and allowed to re-circularize in living cells followed by junction analysis. Remarkably, she identified mitochondrial DNA sequences at the junctions. As she points out, the advantage of this assay is that primers with different microhomology overhangs can be engineered to test recircularization of different ends. This assay was successfully used by the Du lab to identify the XRCC4 NHEJ protein in a genome wide screen [171]. The Du lab also designed an assay that relies on the HO restriction enzyme ( Figure 7D) [172] similarly to their previous assays ( Figure 6B). In this assay, the 24 base pair HO restriction site marked with the natMX cassette (nat confers resistance to nourseothricin) was cloned at the arg3 + locus on Ch.I. The nmt1-HO endonuclease was integrated at ars1 + . Because the cells were grown under continuous expression of the HO endonuclease (no thiamine in the media), only those cells with imprecise repair of the junction (either deletions or insertions) could survive. These cells had destroyed the HO endonuclease restriction site, in effect inactivating the function of the enzyme. The junction of surviving cells was amplified by PCR and sequenced either by the Sanger methods or by Illumina sequencing. This assay is a useful tool for studying break repair that does not rely on homologous recombination.  The HO restriction site marked with natMX was cloned at the arg1 + locus on Ch.I. Continuous expression of HO endonuclease will kill all cells that do not repair the break or repair it correctly because it reconstitutes the restriction site and are vulnerable to re-cutting. Incorrect repair destroys the restriction site and allows cells to live. The authors used both Sanger sequencing and Illumina next generation sequencing to analyze the junction. (E) A transposon-based assay to study genome integration and excision. A plasmid encoding the transposase is co-transformed with another plasmid with transposon integration sequences (Hermes) flanking a KanMX cassette. De-repression of the transposase allows integration of the KanMX cassette randomly in the genome. Following integration, re-expression of the transposase will cause excision of the transposon. Analysis of the junction sequences can determine the form of repair. (F) An assay to test the role of telomeres in preventing chromosome end fusions. Two head to head telomeric repeats were cloned in the second intron of the his3 + gene. Each repeat also contains 80 bp of subtelomeric sequences. This telomere repeat interruption does not disrupt transcription of the his3 + gene. Consequently, the plasmid is his+leu+. The plasmid is linearized (destroying transcription of his3 + ) between the telomeric repeats and introduced into cells. Because each end is capped by telomeres, the fragment can be maintained as a mini-chromosome, which is leu+his-. Deletion of telomerase causes telomere attrition and fusion of the ends which reconstitutes the his3 + gene. Such fusions can be selected for on media lacking both histidine and leucine. Analysis of the fusion junction can determine the precision of repair.
A different NHEJ assay was designed by Li et al. that relies on analyzing the genome insertion and excision of a transposon ( Figure 7E) [173]. Two plasmids, one encoding the transposase and the other the transposon are co-transformed in the cells. De-repression of the transposase causes random insertion of the transposon. The role of NHEJ factors can be investigated by monitoring the efficiency of insertion. Alternatively, re-expression of the transposase will cause excision of the transposon from the genome. The transposon is characterized by 8 bp of homology on either end. Analysis of the repaired junction can determine the nature of the microhomology mediated repair.
Finally, we want to mention an assay to study chromosome end fusions due to telomere erosion ( Figure 7F) [174]. The his3 + is one of the genes in S. pombe that has introns. The investigators introduced head to head telomeric repeats in the second intron of his3 + . This insertion does not affect the function of his3 + . The plasmid is linearized between the telomeric repeats and transformed into living cells. The construct may be propagated as a linear fragment because it has telomeres and can be selected for on media lacking leucine. However, the linear fragment interrupts transcription of the his3 + gene. Deletion of telomerase causes telomere attrition and re-circularization of the fragment which can be selected for on media lacking histidine because it reconstitutes the function of the his3 + gene. The plasmid can be recovered and the junction analyzed by sequencing.

Concluding Remarks
Although none of the assays mentioned here can independently unravel the function of every repair pathway, when the data are put together, a clearer picture emerges. This is particularly true if combined with results from other systems such as S. cerevisiae. Nevertheless, the work done in S. pombe has not just been complementary to the work done in other model systems but instrumental in discovering and defining new mechanisms of DSB repair.
Nevertheless, there are many unanswered questions that remain. For example, the role of chromatin remodeling in modulating DSB repair requires further investigation. The fact that histone modifications play a role in biasing repair towards different pathways is clear. However, how these modifications choreograph repair is not well understood [175]. This is because histone modifications are transient and hard to capture. Experiments using some of the assays described here that either knock out histone modifying enzymes or substitute modifiable histone residues for unmodifiable ones have provided some insight but these are likely to cause genome wide changes which complicates interpretation. Some recent experiments have also shown that RNA plays a major role in DSB repair [176] but the exact dynamics are poorly understood. With advances in technology, such as CRISPR, it is almost certain that in the next few years we will see much more complex assays that are likely to elucidate some of these questions.