SEN1 promotes meiotic progression, sporulation and spore viability in yeast

Because SEN1 is essential, a meiotic depletion allele (sen1-md) was created by placing SEN1 under transcriptional control of the CLB2 promoter, which is not active during meiosis [55,56]. Sen1 protein levels were monitored using a newly developed α-Sen1 antibody. Protein extracts were generated from SEN1 cells at different times after transfer to sporulation (Spo) medium and probed with α-Sen1 antibodies. A single band corresponding to Sen1’s predicted molecular weight of 253 kD was observed and Sen1 protein levels were constant throughout meiosis (Figs 1A and S1A). In contrast, the amount of Sen1 protein in vegetative cells at 0 hour in the sen1-md diploid was reduced relative to SEN1 and decreased below the limit of detection by 3 hours in Spo medium (Fig 1A).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 1. Meiotic depletion of Sen1 results in a prophase I delay and reduces sporulation and spore viability.

(A) Protein extracts from the indicated meiotic timepoints were generated from SEN1 (NH716), sen1-md (NH2667) and P_REC8-SEN1 (NH2667::pNH410) diploids and probed on immunoblots with either α-Sen1 or α-Arp7 antibodies (Arp7 was used as a loading control). The white line indicates two different blots containing the same samples probed with the indicated antibodies. (B) Meiotic progression was followed by staining fixed cells at the indicated timepoints with DAPI and determining the frequency of MI (binucleate) and MII (tetranucleate) cells by fluorescence microscopy. 200 cells were counted at each timepoint for each replicate. SEN1 (n = 7); sen1-md (n = 8); P_REC8-SEN1 (n = 14). The averages of the biological replicates for each strain were plotted. (C) Whole cell Red1 immunofluorescence. To determine the frequency of cells in prophase I, whole cell immunofluorescence using α-Red1 antibodies was performed at the indicated timepoints in SEN1, sen1-md and P_REC8-SEN1 diploids. The average frequencies of Red1⁺ cells from six replicates of each strain were plotted. (D) T₅₀ analysis of Red1⁺ cells. The T₅₀ value indicates the time it took after transfer to Spo medium to reach the half-maximal % Red1⁺ value. In some replicates, this calculation could not be done due to the absence of data from the 1 hour time point. Data were analyzed for statistical significance using an unpaired, two-tailed Student’s t-test. * = p = 0.0291. (E) Sporulation. Cultures derived from different single colonies of the indicated strains were sporulated in liquid Spo medium and the number of asci out of 200 cells was determined by light microscopy. Each dot represents a biological replicate. SEN1 (n = 25); sen1-md (n = 29); P_REC8-SEN1 (n = 17) (F) Spore viability. Tetrads from the cultures sporulated in Panel E were dissected to determine the frequency of viable spores. Each dot represents the spore viability of one biological replicate containing 23-28 dissected tetrads. The statistical significance of differences between the strains in Panels E and F was determined using the Mann-Whitney test (**** = p < 0.0001). ns = not significant. (G) Distribution of viable spores in tetrads. The total number of tetrads for each strain was SEN1 (623), sen1-md (728) and P_REC8-SEN1 (395).

https://doi.org/10.1371/journal.pgen.1011684.g001

To validate the antibody, SEN1 and sen1-∆N (an allele that encodes an N-terminal truncation that lacks the first 1003 amino acids out of 2231) were placed under the control of the meiosis-specific REC8 promoter and transformed into a sen1-md diploid [57]. (Throughout the text, sen1-md::P_REC8-SEN1 strains are referred to simply as P_REC8-SEN1). The peptide used to generate the Sen1 antibody contains amino acids 2002–2019 and therefore the antibody should recognize both the full length and truncated Sen1 proteins that exhibit different mobilities on an immunoblot. In the P_REC8-SEN1 diploid, Sen1 was induced by 4 hours in Spo medium compared to vegetative cells (0 hours) (Figs 1A and S1A). In the extracts from the P_REC8-sen1-∆N diploid, the ~ 250 kD band was replaced by a less abundant protein of ~150 kD, which is close to the expected molecular weight predicted for Sen1^1004-2231 (135 kD), confirming the antibody’s specificity for Sen1 (S1A Fig).

Phenotypic characterization of the sen1-md diploid revealed several meiotic defects. Meiotic progression was examined by staining the DNA with 4’,6-diamidino-2-phenylindole (DAPI) and counting the number of nuclei per cell. Binucleate cells indicate completion of Meiosis I (MI), while tetranucleate cells indicate completion of Meiosis II (MII) (S2 Fig). On average, < 30% of sen1-md cells progressed past MI, and those that did were delayed relative to SEN1 (Fig 1B). In addition, both sporulation and spore viability were significantly reduced in the sen1-md diploid (Fig 1E and 1F).

Mutants that decrease interhomolog crossovers exhibit higher levels of inviable spores due to MI non-disjunction. As a result, the number of tetrads in which all four spores are viable goes down, with a coordinated increase in the number of tetrads with two and zero viable spores [58]. This pattern was not observed for sen1-md (Fig 1G). While the number of tetrads with 4 viable spores (4:0) was greatly reduced, all other tetrad classes (3:1, 2:2, 1:3, 0:4) were increased, indicating that sen1-md spore lethality is unlikely due to MI chromosome mis-segregation (Fig 1G).

While the reduced amount of Sen1 protein present in vegetative sen1-md cells is sufficient for viability, it may not be sufficient for other SEN1 functions (Fig 1A). This fact raises the possibility that the meiotic progression delay, reduced sporulation and decreased spore viability phenotypes result from DNA damage incurred during vegetative growth, rather than defects in meiotic processes. To test this idea, a P_REC8-SEN1 strain was examined to see if expressing SEN1 after the initiation of meiosis can complement sen1-md meiotic phenotypes. In fact, P_REC8-SEN1 partially rescued the meiotic progression delay and sporulation defect (Fig 1B and 1E) and restored spore viability to SEN1 levels (Fig 1F and 1G). Thus, meiosis-specific expression of SEN1 is sufficient to reverse the meiotic defects of the sen1-md mutant, suggesting a requirement for Sen1 activity during normal meiosis.

A delay in meiotic progression was previously observed in the absence of RNase H activity (rnh1∆ rnh201∆) where it is known that DNA:RNA hybrids form on the single- stranded tails of DSBs, thereby making them difficult to repair [17,31]. The MRC delays exit from prophase I when a threshold number of DSBs is present. If the meiotic progression delay in sen1-md is also due to impaired meiotic DSB repair, then cells should take longer to exit prophase I. Prophase I cells can be distinguished using whole cell immunofluorescence with antibodies that detect Red1, a meiosis-specific axial element protein that is degraded upon exit from pachynema [54,59,60]. This analysis found that both entry into and exit from prophase I were affected in the sen1-md diploid, indicating there are two separate components to the meiotic progression delay. First, the appearance of Red1⁺ cells in sen1-md was delayed by ~2 hours compared to SEN1, indicating that SEN1 is necessary for efficient entry into meiosis (Fig 1C). Second, the sen1-md downward slope due to exit from prophase I occurred later in sen1-md relative to SEN1 (Fig 1C). A diploid ectopically expressing SEN1 using P_REC8-SEN1 exhibited similar prophase I kinetics as SEN1, indicating that the MRC was no longer being triggered (Fig 1C and 1D). These results indicate that SEN1 is also required for timely progression out of prophase I.

SEN1 promotes expression of IME1 regulated genes

REC8 is an early meiotic gene whose transcription is activated by Ime1 [38,39,42,61]. One explanation for why P_REC8-SEN1 only partially rescues the sen1-md meiotic progression delay is that SEN1 promotes the timely transcription of genes regulated by Ime1. As a first test of this hypothesis, the timing of expression for several early meiosis-specific proteins was examined.

HOP1, RED1 and REC8 encode meiosis-specific components of the axial elements formed along sister chromatids [59,61–63]. MEK1/MRE4 encodes a meiosis-specific kinase that regulates many different steps of meiotic recombination, as well as the MRC [52,64–70]. Hed1 is a meiosis-specific protein that binds to Rad51 to inhibit its strand exchange activity during prophase I [71,72]. Mek1 phosphorylation of Hed1 threonine 40 stabilizes the protein and can be used as a marker for Mek1 kinase activity [53,73].

In the SEN1 diploid, Rec8 protein was first observed after 2 hours in Spo medium, its level peaked at 4 hours and was nearly undetectable by 8 hours (S1B Fig). By 4 hours, most of the cells were in prophase I, as indicated by peak levels of Red1 and meiotic DSBs [indirectly indicated by Hop1 phosphorylation (pHop1)] (Figs 1C and S1B) [74]. Mek1 activity, indicated by phosphorylated Hed1 (pHed1), also peaked at 4 hours, in keeping with the larger amount of phosphorylated Hop1 (S1B Fig) [73]. Red1 was nearly gone by 6 hours consistent with cells progressing into MI (Figs 1B, 1C and S1B). In contrast, all 5 proteins in the sen1-md diploid peaked later than 4 hours and were still present at 10 hours, consistent with a portion of the cells being delayed in prophase I. Notably, the Rec8 protein from the P_REC8-SEN1 diploid exhibited similar expression kinetics as in sen1-md (S1B Fig).

To more directly test the hypothesis that SEN1 promotes early gene transcription, RNA sequencing analysis was performed using cells taken at different times after transfer to Spo medium. As was previously observed, meiotic progression in the sen1-md diploid was delayed in this time course relative to SEN1 (Fig 2A). Genes were clustered by k-means based on RNA expression patterns in the SEN1 diploid. The heat map for the SEN1 strain exhibited gene clusters that agreed well with the literature (Fig 2B) [38,42]. When a subset of early genes from the SEN1 diploid known to function in prophase I was examined, their expression nearly uniformly peaked at 3–4 hours after transfer to Spo medium (Fig 2C). In contrast, expression of the same set of genes in the sen1-md mutant was delayed, with broader peaks occurring between 4 and 6 hours. The delay in early gene transcription correlated with the later appearance of IME1 transcripts, the master regulator for early gene transcription, in the sen1-md strain compared to SEN1 (Fig 2C). As expected, the delay in early gene transcription coordinately delayed the expression of NDT80-regulated middle genes (Fig 2B and 2C). We propose that meiotic progression proceeds more slowly in sen1-md for two reasons: (1) a delay in the onset of early gene transcription, and (2) activation of the MRC due to the presence of unrepaired DSBs. This hypothesis would explain the partial rescue of the meiotic progression defect in P_REC8-SEN1. Ectopic expression of SEN1 during meiosis using P_REC8-SEN1 would allow DSB repair, thereby deactivating the MRC and allowing cells to progress faster than sen1-md. However, because SEN1 transcripts produced using the Ime1-regulated REC8 promoter appeared too late to fix the early gene transcription defect, the cells still exhibited a delay in progression through the early stages of meiosis.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 2. SEN1 specifically regulates IME1 expression during meiosis.

SEN1 (NH2473) and sen1-md (NH2520) were transferred to Spo medium and cells were taken at various timepoints. (A) Meiotic progression. Fixed cells were stained with DAPI and examined by fluorescence microscopy for the presence of MI and MII cells. (B) RNA-seq analysis. mRNA was isolated at the indicated timepoints from the timecourse shown in Panel A. Numbers above each column indicate hours after transfer to Spo medium. The z value for each timepoint (n) was calculated as [(FPKM_n-FPKM_ave)/SD], where FPKM = fragments per kilobase per million, SD = standard deviation and ave = average. Genes were sorted based on the RNA expression patterns in the SEN1 diploid by clustering analysis (k-means) using Cluster 3.0. The heat map was generated using Java TreeView. Green indicates z values < 0, while red indicates z values > 0. (C) A subset of previously identified IME1-dependent early genes and NDT80-dependent middle genes were selected and clustered as in Panel B. Gene names in orange encode proteins that were examined in S1B Fig, while IME1 is written in blue. (D) Meiotic progression in P_CUP1-IME1 diploids. Diploids containing P_CUP1-IME1 (39807; n = 4), P_CUP1-IME1 sen1-md (NH2842; n = 5) or P_CUP1-IME1 sen1-md::P_REC8-SEN1 (NH2842::pBG45; n = 5) were incubated in Spo medium for 2 hours at which time CuSO₄ was added to a final concentration of 50 μM to induce IME1 expression. Cells at the indicated timepoints were stained with DAPI and meiotic progression analyzed as in Fig 1B. The 0 hour timepoint is the time of copper addition. The average values for each timepoint were graphed. (E) T₅₀ analysis of the timecourses from Panel D. The T₅₀ value indicates the time it took after the addition of copper at which the half-maximal %MI + MII value was reached. Data were analyzed for statistical significance using an unpaired, two-tailed Student’s t-test. **** = p < 0.0001.

https://doi.org/10.1371/journal.pgen.1011684.g002

To test whether SEN1 is involved specifically in the regulation of IME1 expression, the sen1-md mutant was introduced into a strain in which IME1 was under the control of the copper-inducible CUP1 promoter [75]. If SEN1 regulation of the IME1 promoter is responsible for the early gene transcriptional delay, then P_CUP1-IME1 should eliminate this component of the delay. The P_CUP1-IME1 sen1-md diploid exhibited delayed progression compared to P_CUP1-IME1 after the addition of copper, as expected if the MRC was being triggered by unrepaired prophase I DSBs (Fig 2D). However, when P_REC8-SEN1 was present in the P_CUP1-IME1 sen1-md background, the kinetics of meiotic progression was the same as P_CUP1-IME1 (Fig 2D and 2E). These results confirm the hypothesis that SEN1 specifically promotes IME1 expression to allow entry into meiosis.

SEN1 suppresses aberrant transcription of regulatory sequences upstream of IME1

The regulation of IME1 expression is complex, as it requires integration of signals relating to both mating type and nutrients (reviewed in [76,77]. The IME1 promoter is approximately 2 kb in length [78,79]. Within the first 1.1 kb are multiple regulatory sites, including ones that repress IME1 transcription when glucose is present or activate transcription when cells are exposed to acetate [79]. In haploid cells, transcription of a long noncoding RNA called IRT1 beginning 1.4 kb upstream of the IME1 gene creates a repressive chromatin state that inhibits IME1 transcription (Fig 3A) [80]. IRT1 transcription is activated by a repressor of meiosis, RME1, a transcription factor that binds to two sites upstream of IRT1 [80,81]. Transcription of RME1 itself is repressed in diploids by the a1/α2 repressor encoded by MATa and MATα, thereby preventing transcription of IRT1 and allowing expression of IME1 when diploid cells are in Spo medium (Fig 3A) [81]. In case the repression of RME1 expression is leaky, transcription of a second long non-coding RNA called IRT2, located immediately upstream of IRT1, prevents Rme1 from binding to its sites in the IME1 promoter, thereby ensuring that IRT1 transcription is low (Fig 3A) [82]. IRT2 transcription is activated by a complex formed between Ume6 and Ime1 [82].

Download:

• PNG
  larger image
• TIFF
  original image

Fig 3. Direct RNA sequencing in SEN1 and sen1-md diploids after 2 hours in Spo medium.

(A) Schematic model of the IME1 region. In haploid cells, Rme1 activates transcription of the noncoding IRT1 RNA that inhibits transcription of IME1 by creating a “repressive chromatin state” (black bar). In diploid cells in Spo medium, RME1 expression is repressed and glucose repression is removed, lowering the amount of IRT1, thereby allowing IME1 transcription. Ime1 then binds to Ume6 upstream of IRT2 and activates transcription of this non-coding RNA. (B) Total RNA extracted from SEN1 diploid cells after 2 hours in Spo medium was fractionated on a 1% agarose gel in 1xTAE containing 1% bleach and stained with SYBR Gold. (C) Total RNA was incubated with the indicated biotinylated antisense RNAs targeting specific rRNAs, followed by precipitation using streptavidin-coated magnetic beads. (D) Individual RNA molecules detected by dRNA-seq from SEN1 and sen1-md were sorted by their 3’-end positions and subsequently by their 5’-end positions before plotting along the indicated chromosomal region. The step size of the read counts in sen1-md was set to 1.2 times that of SEN1 to normalize for the number of uniquely mapped reads for the combined replicates (12,637,870 and 15,252,836, respectively). Red lines: Watson strand. Blue lines: Crick strand. (E) Heatmaps of read coverage (normalized to count per million) for SEN1 and sen1-md were plotted across 918 annotated CUT regions. Each region was scaled to equal length and aligned at its annotated transcription start (TSS) and termination sites (TTS), with 500 bp of flanking sequence. Data were sorted according to transcript abundance within the CUT regions. (F) Same as E, except that 833 SUT regions were plotted. (G) Same as D, except that the region encompassing IME1 was plotted. Grey lines: intronic regions. The tests for statistical significance for conclusions based on Panels D and G are described in the Methods.

https://doi.org/10.1371/journal.pgen.1011684.g003

To investigate how SEN1 contributes to the regulation of IME1, nanopore direct RNA sequencing (dRNA-seq) was performed on coding and non-coding transcripts isolated from SEN1 and sen1-md cells in Spo medium [83]. dRNA-seq requires that each transcript has a 3’ poly(A) tail to which a sequencing adapter and motor protein that threads the RNA through a pore can be ligated. Non-coding RNAs, particularly those that fail to terminate properly in the absence of the Nrd1-Nab3-Sen1 (NNS) complex, lack polyA tails and therefore would be underrepresented [84]. To capture these transcripts, poly(A) tails need to be added in vitro. However the high abundance of the four ribosomal RNAs (rRNAs) (25S, 18S, 5.8S and 5S) in total RNA preparations would overwhelm this reaction and so must be removed prior to polyadenylation (Fig 3B). The four rRNAs were depleted from total RNA isolated from SEN1 and sen1-md cells by hybridization to the combined biotinylated antisense transcripts for each rRNA, followed by streptavidin-based removal (Fig 3C). The resulting rRNA-depleted RNA was then polyadenylated in vitro using E. coli poly(A) polymerase prior to nanopore library preparation.

Two biological replicates were performed for SEN1 and sen1-md strains after two hours in Spo medium, yielding ~6–7 million uniquely aligned reads per replicate to the S. cerevisiae SK1 genome. To assess data quality, sequencing reads mapped within 5,496 annotated transcript regions were quantified using featureCounts, followed by normalization and analysis with DESeq2 [85–87]. Strong correlations (r > 0.95) were observed between biological replicates of the same genotype, whereas lower correlations between SEN1 and sen1-md samples highlight condition-specific transcriptional changes (S3A Fig). In dRNA-seq, reads initiate at the poly(A)-anchored 3’ end and proceed in a 3’-to-5’ direction; therefore RNA fragmentation or degradation can yield 5’-truncated reads and under-representation of 5’ ends. As a result, transcripts sharing the same 3’ end frequently have heterogeneous 5’ ends. This is a known limitation of dRNA-seq on the Oxford Nanopore platform [83].

In wild-type cells, SEN1 negatively regulates the activity of the NNS complex by promoting premature termination of NRD1 transcription [88]. Consistent with this role, NRD1 transcript levels in the sen1-md mutant were 4.6-fold higher than in SEN1 (Figs 3D and S3B). No significant difference in transcript levels was observed for the adjacent RAD50 gene, although the transcription initiation site was altered to give a longer 5’ untranslated region in sen1-md (Figs 3D and S3B). The NNS complex is also required to suppress the expression of non-coding RNAs by coupling SEN1-dependent termination to TRAMP-mediated exosome degradation [22]. Consistent with this fact, sen1-md exhibited an accumulation of cryptic unstable transcripts (CUTs) and, to a lesser extent, stable unannotated transcriptions (SUTs) (Fig 3E and 3F) [89–91]. In addition, while the amount of sense transcripts for protein coding genes was similar between SEN1 and sen1-md, antisense transcripts were modestly elevated in sen1-md (S3D Fig). These results indicate that there were widespread termination defects caused by impaired Sen1 function.

The SEN1 diploid exhibited the expected transcript pattern in the IME1 region, i.e., a high level of IME1 and IRT2 RNAs and a lower level of IRT1 RNA (Fig 3G) [82]. In addition, the RPL43B gene immediately upstream of IRT2 was transcribed in the opposite orientation as expected. This pattern was significantly altered in sen1-md. First, IME1 transcripts were significantly reduced in sen1-md relative to SEN1 which explains why it takes longer for sen1-md cells to activate transcription of the early genes and enter meiosis (Figs 3G and S3B). Second, novel transcripts originating within IRT1 with opposite polarity spanned the IRT2 sequence (Fig 3G). The genomic location of these antisense transcripts was annotated as IRT2-R. In addition, a high level of antisense transcripts spanning RPL43B was also observed (annotated RPL43B-R) (Fig 3G). These results demonstrate that SEN1 suppresses the accumulation of antisense transcripts upstream of IME1 that may interfere with IME1 transcription.

The sen1-md delay in exiting prophase I is due to the meiotic recombination checkpoint

The MRC prevents cells from exiting prophase I when DSB levels are high because activated Mek1 inhibits Ndt80-dependent transcription [52,60,92,93]. Mek1 interacts with Ndt80 through a five-amino acid sequence in the transcription factor and deletion of this sequence abrogates the MRC [52]. The NDT80 allele containing this five-codon deletion is referred to as NDT80-mid (Mek1 Interaction Defective). NDT80-mid specifically disrupts the MRC without affecting Mek1 kinase activity and cells progress through meiosis with the same kinetics as NDT80 strains [60]. Therefore NDT80-mid is a useful tool for determining whether the MRC is triggered by a particular mutant.

In the SEN1 strain, MI cells peaked at ~13% after only 5 hours in Spo medium (Fig 4A). In contrast only ~7% of the sen1-md cells had completed MI after 12 hours (Fig 4A). The onset of MI occurred later than SEN1 in both NDT80-mid sen1-md and sen1-md due to the delay in early gene transcription. However, NDT80-mid sen1-md cells completed MI faster than sen1-md, peaking with ~18% cells at 8 hours (Fig 4A). The fact that the MRC is responsible for the prophase I arrest/delay observed when Sen1 is meiotically depleted indicates that SEN1 is required for timely DSB repair.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 4. Meiotic progression and sporulation of sen1-md diploids in the absence of the meiotic recombination checkpoint.

SEN1 (n = 13), sen1-md (n = 20), NDT80-mid sen1-md (NH2667::pNH317²) (n = 4), mek1∆ (NH729) (n = 6), and mek1∆ sen1-md (NH2669) (n = 10) diploids were induced to undergo meiosis and the average frequency of the indicated cell types determined at various time points as described in Fig 1B. (A, E) binucleate cells that have completed MI; (B, F) Post-MI includes both MII cells (tetranucleate) and cells with fragmented nuclei (>4 DAPI foci). (C, G) Cells with fragmented nuclei. (D, H) Different colonies from the strains used in Panels A and E, as well as NDT80-mid P_REC8-SEN1 (NH2667::pNH410::pNH317²), mek1∆ P_REC8-SEN1 (NH2669::pNH410), and mek1∆ P_SEN1-SEN1 (NH2669::pBG28), were sporulated at 30°C on solid medium. Each dot represents a biological replicate for which the percentage of sporulated cells was determined for 200 cells. The SEN1 and sen1-md data are repeated in A-H. Error bars indicate the means and standard deviations. Statistical significance of differences between strains in D and H was determined using the Mann-Whitney test (* = p < 0.02; ** = p < 0.002; *** = p < 0.001; **** = p < 0.0001).

https://doi.org/10.1371/journal.pgen.1011684.g004

Although NDT80-mid allowed some sen1-md cells to exit prophase I, it did not rescue the sporulation defect (Fig 4D). This may be because allowing cells to proceed through the meiotic divisions with unrepaired DSBs resulted in fragmented nuclei, as indicated by the presence of cells with more than 4 DAPI staining foci, similar to what is observed when the MRC is inactivated in dmc1∆ or sae2∆ diploids (S2 Fig) [92,94]. Cells containing either four nuclei (i.e., completed MII) or fragmented nuclei were classified as “post-MI”. Greater than 80% of SEN1 cells were post-MI by 12 hours with <5% showing fragmented nuclei (Fig 4B and 4C). In contrast, only ~25% of sen1-md and NDT80-mid sen1-md cells proceeded past MI (Fig 4B). A major difference between these two strains, however, was that almost all the NDT80-mid sen1-md post-MI cells contained fragmented nuclei, compared to <5% of sen1-md cells (Fig 4C). Therefore, the prophase I delay created by the MRC in the sen1-md diploid provided time for some DSBs to be repaired.

MEK1 controls not only the MRC, but also interhomolog bias and inhibition of Rad51 strand exchange activity [67,69,73]. For example, when MEK1 is inactive in a dmc1∆ mutant, cells progress through meiosis with intact chromosomes because Rad51 repairs DSBs using sister chromatids, thereby removing the signal that activates the MRC [49,74]. In contrast, if DSBs are unrepairable using either homologs or sister chromatids as templates (e.g., sae2∆ or rad50S), cells progress through meiosis with fragmented nuclei in the absence of the MRC [49,94]. Similar to NDT80-mid, combining mek1∆ with sen1-md resulted in faster entry into MI compared to sen1-md with a reduced number of post-MI cells (Fig 4E and 4F). The increased frequency of fragmented nuclei in the mek1∆ sen1-md diploid suggests that DSBs persisted because they were unrepairable (Fig 4G). The sporulation defects of NDT80-mid sen1-md and mek1∆ sen1-md were rescued by P_REC8-SEN1 to the same extent as the constitutively expressed P_SEN1-SEN1, consistent with a need for SEN1 during prophase I to allow DSB repair (Fig 4D and 4H).

SEN1 plays a minor role in crossover formation at the HIS4LEU2 DSB hotspot

To determine directly whether SEN1 promotes meiotic recombination, physical analyses using the HIS4LEU2 hotspot were performed. This well characterized hotspot has asymmetric XhoI sites that give rise to distinct bands corresponding to DSBs and crossovers (COs) that can be detected using Southern blots (Fig 5A) [95]. In addition, there is an NgoMIV site on one homolog adjacent to the region where Spo11 makes DSBs. Digestion with both XhoI and NgoMIV therefore also allows detection of noncrossovers (NCOs) resulting from gene conversion events involving the NgoMIV site [96].

Download:

• PNG
  larger image
• TIFF
  original image

Fig 5. Physical analysis of DSBs, crossover and noncrossover formation during meiosis in sen1-md and P_REC8-SEN1 diploids.

(A) Schematic of the HIS4LEU2 hotspot (taken from [64]). The homologs, P1 and P2, have XhoI sites located at asymmetric positions. The dotted vertical line indicates where Spo11 makes DSBs. The NgoMIV site is adjacent to the DSB site on P1. The black box indicates the sequence used as a probe in Southern blot analysis. (B) DSBs in sae2∆ diploids. DNA was isolated from sae2∆ (NH1054), sae2∆ sen1-md (NH2846), and sae2∆ sen1-md::P_REC8-SEN1 (NH2846::pNH410)(n = 5 for all) at the indicated timepoints and digested with XhoI to detect DSBs. The XhoI sites on the two homologs are slightly offset, resulting in two DSB fragments. One replicate is shown. (C) Quantification of the total DSBs from the experiment in B. The amount of DNA in the two DSB fragments was combined and divided by the total amount of DNA for the five replicates in each strain at each timepoint. None of the pairwise combinations at either 6 or 8 hours were significantly different based on an unpaired, two-tailed Student’s T test. (D) Timecourses of SEN1 (n = 2), sen1-md (n = 3) and P_REC8-SEN1 (n = 3) diploids were performed and genomic DNA isolated from cells at the indicated timepoints. Genomic DNA was digested with XhoI to detect Spo11-generated breaks. The entire gel for the representative replicate shown here is presented in S4 Fig. (E) Average DSB values for the different replicates normalized to the total amount of DNA. (F) CO and NCO: Genomic DNA was digested with XhoI and NgoMIV. CO1 and CO2 indicate crossover bands, while NCO1 and NCO2 represent noncrossover bands. (G) average %CO2 values. (H) average %NCO1 values. The asterisk indicates a band containing a flanking XhoI site gene conversion.

https://doi.org/10.1371/journal.pgen.1011684.g005

A reduction in recombination products will be observed if there is a defect in DSB formation. Whether SEN1 promotes DSB formation was tested using an sae2∆/com1∆ mutant, in which DSBs are generated, but Spo11 remains covalently attached to the 5’ ends, thereby preventing resection and repair [97,98]. No significant differences were observed for the levels of DSBs in sae2∆, sae2∆ sen1-md or sae2∆ P_REC8-SEN1 strains after both 6 and 8 hours in Spo medium. SEN1 therefore does not promote Spo11-mediated DSB formation at the HIS4LEU2 hotspot (Fig 5B and 5C).

In the SEN1 diploid, DSBs peaked at 3 hours after transfer to Spo medium and then rapidly disappeared (Fig 5D and 5E) (S4 Fig). In contrast, DSB formation was delayed in both sen1-md and P_REC8-SEN1. This was expected as transcription of most of the genes required for making DSBs (e.g., SPO11, REC114, REC102 and REC104) is dependent on IME1. DSBs also persisted longer in sen1-md than in either SEN1 or P_REC8-SEN1 and were still detectable after 12 hours (Fig 5E). CO2 and NCO1 fragments that can be unambiguously identified on the XhoI/NgoMIV gel were quantified (Fig 5F). In the SEN1 strain, CO2 formation plateaued by 6 hours. In contrast, CO formation in sen1-md was delayed and reduced by 25% at 12 hours but still had not reached a plateau (Fig 5G). The kinetics of CO formation in the P_REC8-SEN1 strain was intermediate between SEN1 and sen1-md, as expected if the sen1-md early gene transcription delay was followed by more efficient DSB repair due to the presence of SEN1. However, for reasons that are unclear, the level of COs in the P_REC8-SEN1 strain resembled that of sen1-md and did not reach SEN1 levels.

A different phenotype was observed for NCOs in the sen1-md diploid. Instead of being decreased, there were more NCOs at 12 hours compared to SEN1 (Fig 5F and 5H). This result could be because prophase I length is longer in many of the sen1-md cells, which delays the expression of the polo-like kinase, CDC5, required for resolution of the double Holliday junctions into COs [54]. Furthermore, cells arrested in pachynema continue to make DSBs and produce NCOs by synthesis-dependent strand annealing [99–101]. The idea that the increase in NCO formation in sen1-md is due to a delay in prophase I exit is supported by the observation that in the P_REC8-SEN1 diploid that was able to progress, NCOs plateaued by 8 hours (Fig 5F and 5H).

The effects of sen1-md on interhomolog recombination are small and therefore their significance is difficult to determine with a limited number of replicates. We therefore repeated the CO and NCO analyses at the 8 and 12 hour timepoints using 5 replicates from the SEN1, sen1-md and P_REC8-SEN1 diploids. Sporulation and spore viability in the replicates from the three strains were consistent with previous results (S5A and S5B Fig). A slight reduction in the CO2 band was observed between the SEN1 and sen1-md diploids (~20%) at both the 8 and 12 hour timepoints, although only the 8 hour values were statistically significant (S5C and S5D Fig). In contrast, the NCO1 values were similar for SEN1 and sen1-md at both timepoints (Figs 5E and S5C). For reasons that are unclear, the P_REC8-SEN1 COs and NCOs were reduced compared to sen1-md in contrast to the previous timecourses. SEN1 may have a differential effect on DSBs that form crossovers, but this effect is minor.

SEN1 promotes chromosome synapsis during meiosis

In budding yeast, meiotic DSB repair by a crossover specific pathway containing a functionally diverse set of “ZMM” proteins is required for SC formation [102–107]. The SC is a tripartite supramolecular chromosomal complex that initiates assembly from discrete chromosome sites associated with recombination (as well as from centromeres) and extends along the full lengths of aligned homologous chromosome axes [108,109]. SC formation in yeast is completely dependent on DSB formation and early processing. Diploids containing spo11∆, sae2∆ or rad50S mutants exhibit only aggregates of SC structural proteins (polycomplexes) instead of linear SC structures, while the DSB repair defects of dmc1∆ or msh4∆ are associated with an overall decrease and delay in SC formation, rather than a complete abrogation of SC [110–116].

Diploids homozygous for ndt80∆ arrest in pachynema, the stage of prophase I when chromosomes are fully synapsed [93]. The ability of ndt80∆ sen1-md diploids to assemble SCs was assessed by examining the distribution of SC structural components Gmc2 (part of the SC central element) and Zip1 (an SC transverse filament protein) on surface-spread meiotic chromosomes that displayed maximal Red1 accumulation after 5 hours in Spo medium [117,118]. At this timepoint, 80% of the ndt80∆ control nuclei exhibited full-length or nearly full-length linear structures of Gmc2 throughout the entirety of the mid-prophase nucleus, while the remaining spread chromosomes showed shorter linear assemblies of Gmc2 (Fig 6A and 6C). By contrast, only 6% of the surface-spread nuclei from the ndt80∆ sen1-md strain exhibited full-length linear Gmc2 structures, even though most spreads displayed Red1 accumulation on chromosome axes to the same extent as the ndt80∆ control. The remaining nuclei were nearly evenly divided between those that had only Gmc2 foci (48%) and those with shorter linear Gmc2 assemblies (46%) (Fig 6A and 6C). In ndt80∆ sen1-md cells, the Gmc2 linear assemblies detected on chromatin co-localized with Zip1, indicating that these structures correspond to mature (albeit partial) SCs (Fig 6B). Expression of SEN1 from the REC8 promoter partially rescued the sen1-md synapsis defect, increasing the frequency of Red1⁺ nuclei with complete SCs from 6% to 64% (Fig 6A and 6C). The partial complementation may be due to variability in the timing of expression by the REC8 promoter, with SEN1 being expressed too late in some cells. SEN1 therefore indirectly plays an important role in promoting chromosome synapsis during yeast meiosis.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 6. SC assembly in ndt80∆, ndt80∆ sen1-md and ndt80∆ P_REC8-SEN1 diploids.

The ndt80∆ (NH2188, 156 nuclei), ndt80∆ sen1-md (AM6697, 169 nuclei) and ndt80∆ P_REC8-SEN1 (AM6697::pNH410, 75 nuclei) diploids were transferred to Spo medium for 5 hours at which time cells were processed to make chromosome spreads. (A) Immunofluorescence on surface-spread meiotic chromosomes to detect chromosome axes (Red1) and the central element (Gmc2). Each row corresponds to a different surface-spread nucleus, with the genotypes indicated on the left. DNA is labeled with DAPI (white, first column), α-Red1 (magenta, second column) or α-Gmc2 (green). The last column shows the merged image. Bar, 1 μm. (B) Same as A except that α-Zip1 antibodies were used to detect the central region of the SC. (C) Histogram of the fraction of total nuclei at 5 hours with completely synapsed chromosomes (Comp. linear = full linear structures of Gmc2 that extend throughout the full spread nucleus), incomplete SC (incomp. linear = shorter linear structures of Gmc2 that only partially extend throughout the surface-spread nucleus), or Gmc2 foci. Only those surface-spread nuclei with maximal Red1 accumulation on chromosome axes were analyzed.

https://doi.org/10.1371/journal.pgen.1011684.g006

SEN1 prevents formation of SPO11-independent DSBs during meiosis

If Spo11-generated meiotic DSBs are responsible for the MRC-mediated prophase I delay in the sen1-md diploid, then deleting SPO11 should rescue this delay. In fact, the sen1-md spo11∆ diploid progressed faster through MI than sen1-md, consistent with the MRC no longer being triggered (Fig 7A), However, as no Spo11 DSBs have been made, the chromosomes should be intact and therefore enter MII as efficiently as spo11∆, but this was not the case (Fig 7B). Instead, the sen1-md spo11∆ diploid exhibited a similar number of post-MI cells as sen1-md, but the number of fragmented nuclei was much higher and sporulation was significantly reduced (Fig 7B, 7C and 7G). The fragmented nuclei show that DSBs were present in the sen1-md spo11∆ cells that were not directly created by Spo11.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 7. Meiotic progression and sporulation of sen1-md diploids containing spo11∆.

SEN1, sen1-md, spo11∆ (NH1055, n = 10) and spo11∆ sen1-md (NH2689, n = 15) diploids were induced to undergo meiosis and the average frequency of the indicated cell types was determined at various time points as described in Fig 3. The SEN1 and sen1-md data for panels A-C are the same as Fig 3. (A) Cells that have completed MI; (B) Post-MI includes both MII cells and cells with fragmented nuclei (>4 DAPI foci). (C) Cells with fragmented nuclei. (D-F) Comparison of %MI, % post-MI and % cells with fragmented nuclei from sen1-md rad53∆ sml1∆ (NH2843) and sen1-md spo11∆ rad53∆ sml1∆ (NH2844) diploids with the timecourses shown in A-C. Dotted lines are the same as in A-C. (G) Different colonies from the diploids used in Panels A-C, spo11∆ P_REC8-SEN1 (NH2689::pNH410) and spo11∆ P_SEN1-SEN1 (NH2689::pBG28) were sporulated at 30°C on solid medium. Each dot represents a biological replicate for which the frequency of sporulated cells was determined for 200 cells. The data for SEN1 and sen1-md from Fig 4 are repeated in all of the graphs. Error bars indicate the means and standard deviations. Statistical significance of differences between strains was determined using the Mann-Whitney test (* = p < 0.02; ** = p < 0.002; *** = p < 0.001; **** = p < 0.0001).

https://doi.org/10.1371/journal.pgen.1011684.g007

Meiotic DSBs are a critical part of prophase I and so do not trigger the Rad9-Rad53-mediated DNA damage checkpoint, and their repair is instead regulated by Mek1 [66,92,119]. However, DSBs that persist past MI trigger the DNA damage checkpoint and delay entry into MII [120]. Therefore, one explanation for why sen1-md spo11∆ cells did not efficiently proceed through MII would be if DSBs were present post-MI that triggered the RAD53-dependent DNA damage checkpoint. This idea was tested by looking at meiotic progression in sen1-md and sen1-md spo11∆ diploids containing deletions of both RAD53 and SML1, since rad53∆ cells are inviable when SML1 is present [121]. (For simplicity, the rad53∆ sml1∆ genotype is referred as rad53∆).

This analysis revealed that sen1-md not only triggers the MRC during prophase I, it also triggers the DNA damage checkpoint post-MI. While the slow progression through MI was similar in the sen1-md and rad53∆ sen1-md diploids due to the MRC, the rad53∆ sen1-md cells entered MII more quickly than sen1-md (Fig 7D and 7E). Furthermore, there was an increase in the frequency of cells with fragmented nuclei in rad53∆ sen1-md, indicating that the absence of the post-MI DNA damage checkpoint allowed some DSBs to escape repair (Fig 7F).

An unexpected result was observed in the rad53∆ sen1-md spo11∆ diploid. Since the DNA damage checkpoint does not function during prophase I, deletion of rad53∆ should have no effect on the kinetics of progression through MI (Fig 7D) [120]. However, entry into MI was significantly faster and more efficient in rad53∆ sen1-md spo11∆ compared to sen1-md spo11∆ (Fig 7D). Significantly more rad53∆ sen1-md spo11∆ cells progressed through MII than the sen1-md spo11∆ strain, perhaps because more cells completed MI (Fig 7E). The frequency of cells with fragmented nuclei increased as well (Fig 7F). These observations suggest that in sen1-md spo11∆ diploids, RAD53 functions prior to prophase I in response to SPO11-independent DSBs to delay progression into prophase I and the first meiotic division.

In contrast to NDT80-mid sen1-md and mek1∆ sen1-md, P_REC8-SEN1 only weakly complemented the sporulation defect of sen1-md spo11∆ (Fig 7G). However, the sen1-md spo11∆ sporulation defect was completely rescued when SEN1 was constitutively expressed using P_SEN1-SEN1. This difference indicates that the timing of SEN1 transcription matters in the absence of SPO11 and argues that DSBs present in sen1-md spo11∆ are temporally separable and occur earlier than the Spo11-generated DSBs in sen1-md.

SEN1 prevents accumulation of DNA:RNA hybrids during premeiotic S phase and prophase I

Previous work demonstrated that in the absence of RNaseH activity, DNA:RNA hybrids form at the ends of Spo11-generated DSBs in prophase I, resulting in a meiotic progression delay and reduced spore viability [17,31]. To test whether SEN1 similarly prevents DNA:RNA hybrid accumulation during meiosis, the frequency of DNA:RNA hybrids (some of which may be R-loops) was compared between SEN1 and sen1-md at different times after transfer to Spo medium. DNA:RNA hybrids were detected as foci on chromosome spreads using the S9.6 antibody (Fig 8A) [32]. The rnh1∆ rnh201∆ hpr1∆ diploid was used as a positive control as it has previously been shown that vegetative cells with this genotype exhibit numerous DNA:RNA hybrid foci using this antibody (Fig 8A) [17]. To further validate the assay, chromosome spreads from rnh1∆ rnh201∆ hpr1∆ cells collected 9 hours after transfer to Spo medium were treated with RNase H prior to antibody staining. This treatment led to a significant reduction of S9.6 foci, confirming that the foci are DNA:RNA hybrids (Fig 8B).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 8. SEN1 decreases DNA:RNA hybrids during meiosis.

(A) Chromosome spreads were prepared from vegetatively growing SEN1, sen1-md and rnh1∆ rnh201∆ hpr1∆ (LZY2919) diploids. The DNA was stained with DAPI (blue) and DNA:RNA hybrids were detected by immunofluorescence (red foci) using the S9.6 antibody. (B) Chromosome spreads from rnh1∆ rnh201∆ hpr1∆ cells at the 9 hr meiotic timepoint were treated with either bovine serum albumin (-RNase H) or RNase H (+RNase H) and then stained with the S9.6 antibody (n = 2). (C) DNA:RNA hybrid foci were detected on chromosome spreads from SEN1 (n = 4 or 5), sen1-md (n = 4, 5 or 6), spo11∆ (n = 3) and sen1-md spo11∆ (n = 3) at the indicated timepoints using the S9.6 antibody. n varies for some strains depending upon the timepoint. Two hundred spreads were examined for each biological replicate. Green bars indicate the mean values. Statistical significance was determined using the Mann-Whitney test (* = p < 0.02; ** = p < 0.002; *** = p < 0.001; **** = p < 0.0001).

https://doi.org/10.1371/journal.pgen.1011684.g008

The sen1-md diploid exhibited a statistically significant increase in S9.6 foci at the 0 hour timepoint compared to SEN1 (Fig 8C). This observation suggests that the sen1-md mutant is hypomorphic in vegetative cells; that is, while viable, the reduced amount of Sen1 protein has an effect on R-loop levels. An even higher frequency of DNA:RNA hybrids was observed in sen1-md starting at 2 hours and remained high even after 9 hours in Spo medium (Fig 8C). At 5 and 7 hours, the SEN1 diploid exhibited a significantly higher number of S9.6 foci compared to spo11∆, which is unable to make prophase I DSBs (Fig 8C). However, this difference was not present after 9 hours, as expected if the DSBs in the SEN1 strain were repaired by then (Figs 5D and 8C). These results support the idea that the MRC is triggered in the sen1-md mutant by DNA:RNA hybrids present at the ends of SPO11-dependent DSBs.

A previous study used the S9.6 antibody to do ssDRIP-seq (single strand DNA:RNA hybrid immunoprecipitation) to map the specific locations of DNA:RNA hybrids in meiotic yeast cells in an rnh1∆ rnh201∆ diploid [31]. At later timepoints (e.g., 5 hour), DNA:RNA hybrids were observed that corresponded to the positions of Spo11 oligos indicating meiotic DSBs. In addition, at an early timepoint (2 hour), DNA:RNA hybrids (proposed to be R-loops) were found at places where the transcription of highly transcribed genes collides head on with replication (TRCs) [31]. The frequency of these hybrids was increased in the spo11∆ diploid.

To see if SEN1 regulates the presence of R-loops during premeiotic S phase, S9.6 foci were examined after 2 hours in Spo medium. The timing of DNA replication in the SEN1, sen1-md, spo11∆ and sen1-md spo11∆ diploids was determined using flow cytometry. Replicating cells were present at the 2 hour timepoint in SEN1, but replication was delayed in the sen1-md mutant by an hour due to the delay in IME1 transcription (S6 Fig). Consistent with the literature, the spo11∆ diploid proceeded faster through premeiotic S phase, and this partially counteracted the sen1-md delay such that there were replicating cells after 2 hours in the sen1-md spo11∆ strain (S6 Fig) [122]. At the 2 hour timepoint, the frequency of S9.6 foci was significantly increased in the sen1-md spo11∆ strain compared to either the sen1-md or spo11∆ mutant alone (Fig 8C). SEN1 therefore functions in conjunction with SPO11 to suppress the presence of DNA:RNA hybrids during premeiotic S phase.

SEN1 is a regulator of meiotic early gene expression

The sen1-md delay in meiotic progression was only partially rescued by P_REC8-SEN1, revealing the unexpected result that SEN1 regulates entry into meiosis. RNA-seq analysis showed that transcription of the IME1 meiotic early gene master regulator was delayed in the sen1-md diploid, resulting in a coordinate delay in early and middle gene expression. Furthermore, SEN1 directly affects transcription of IME1 through regulation of the IME1 promoter, because substituting the IME1 promoter with the CUP1 promoter allowed sen1-md::P_REC8-SEN1 to progress through meiosis with the same kinetics as SEN1. IME1 transcription is negatively regulated by the presence of the non-coding RNA IRT1, which is downregulated in meiosis by the absence of Rme1 and the expression of the upstream non-coding RNA IRT2 [80,82]. Given Sen1’s role in terminating non-coding RNAs, one hypothesis to explain these results would be if there was readthrough of the IRT2 transcript into the IRT1 region when SEN1 function was impaired. To test this idea, individual transcripts starting in IRT2 and continuing into IRT1 would need to be detected. Towards this end, a protocol was developed for looking at non-coding RNAs present after 2 hours in Spo medium by dRNA sequencing.

The dRNA sequencing analysis confirmed that IME1 transcripts were reduced in sen1-md, explaining why it takes longer for sufficient Ime1 to accumulate for cells to enter meiosis. While IRT2 readthrough was not observed, ruling out our initial hypothesis, dRNA-seq revealed instead a dramatic change in the expression pattern of the IME1 region. Novel antisense transcripts starting in IRT1 and proceeding through IRT2 were detected, as well as antisense transcripts that spanned a ribosomal protein gene upstream of IRT2. These results suggest that SEN1 suppresses antisense transcription in this region, perhaps by premature transcriptional termination of the antisense RNAs. Further study is necessary to determine the mechanism by which SEN1 suppresses antisense transcription upstream of IME1, as well as whether this aberrant transcription negatively affects IME1 expression.

SEN1 functions prior to prophase I to prevent SPO11-independent DSBs

During S phase in vegetatively growing cells, R-loops accumulate at places where replication collides head-on with transcription (TRCs) [6,9,123–127]. When a TRC occurs, positive supercoils accumulate between the replication fork and transcription machinery, stalling both processes. The transcribed RNAs may then base pair with the template strand of the DNA to make R-loops [126]. Sen1 associates with the replisome and helps prevent R-loops from forming at TRCs both by displacing RNA Pol II as well as coordinating the activities of topoisomerases I and II to relieve topological stress [123,124,126,127].

R-loops also accumulate at TRCs during premeiotic S phase [31]. When the precise genomic locations of R-loops were mapped in an rnh1∆ rnh201∆ SK1 diploid after transfer to Spo medium, an increase in R-loops at TRCs was observed at the 2 hour timepoint when premeiotic S was occurring [31,122]. No increase in R-loop signal was detected when the orientation of a specific TRC was changed such that replication and transcription were co-directional. The R-loops at premeiotic TRCs are distinct from the DNA:RNA hybrids formed at Spo11-generated DSBs in that they do not co-localize with Spo11 hotspots and their levels are reduced at later timepoints when chromosomes have fully synapsed [31]. Interestingly deletion of SPO11 increased the frequency of R-loops at the premeiotic TRCs and these R-loops exhibited a similar distribution as the R-loops observed in the rnh1∆ rnh201∆ diploid [31].

Several observations support the idea that SEN1 also plays a role in preventing/removing R-loops at TRCs during premeiotic S phase. (1) sen1-md diploids exhibited a severe sporulation defect, in part because some cells arrest in prophase I due to unrepaired DSBs triggering the MRC. If the sen1-md sporulation defect was due solely to unrepaired Spo11 DSBs, then spo11∆ should have rescued sporulation in the sen1-md diploid, but this was not the case. (2) In the sen1-md spo11∆ strain, the MRC was abrogated as expected, with an increase in MI cells compared to sen1-md. Surprisingly, most of the cells that progressed past MII exhibited fragmented nuclei, demonstrating the presence of DSBs. These DSBs were not generated directly by Spo11 since the protein is not there. (3) Progression through MI was faster and more efficient when the DNA damage checkpoint kinase RAD53 was deleted from sen1-md spo11∆, suggesting that DSBs formed during premeiotic S phase are triggering this checkpoint prior to prophase I. (4) P_REC8-SEN1 only weakly rescued the sporulation defect of sen1-md spo11∆, while P_SEN1-SEN1 rescued completely, arguing that the timing of P_REC8-SEN1 expression was too late in the sen1-md spo11∆ diploid to effectively complement sen1-md. In contrast, P_REC8-SEN1 expression was not too late to allow repair of Spo11-generated DSBs in NDT80-mid sen1-md or mek1∆ sen1-md strains. These results suggest that SPO11-independent DSBs arise at an earlier time than SPO11-dependent prophase I DSBs. (5) After 2 hours in Spo medium, DNA:RNA hybrid foci were increased in sen1-md spo11∆ compared to either sen1-md or spo11∆. The latter finding is informative because spo11∆ diploids do not make the prophase I DSBs that are used for recombination and therefore the S9.6 foci in the sen1-md spo11∆ cannot be attributed to DNA:RNA hybrids at resected Spo11-dependent DSB ends.

We propose that the SPO11-independent DSBs in the sen1-md spo11∆ diploid arise because (1) R-loops formed at TRCs during premeiotic S phase get converted to DSBs that trigger the DNA damage checkpoint and (2) SEN1 is not present to prevent/eliminate R-loops at premeiotic TRCs. Sen1 could help prevent R-loop formation by removing RNA Pol II from TRCs or by coordinating topoisomerase I and II activity as it does during vegetative S phase. However, SEN1 is not sufficient to completely prevent R-loop formation at TRCs, otherwise the R-loop frequency at TRCs would not increase in rnh1∆ rnh201∆ diploids where SEN1 is functional [31]. Therefore, it is also possible that Sen1 may disassemble R-loops at premeiotic TRCs using its helicase activity. The multi-functional roles that SEN1 may play at premeiotic TRCs could explain why sen1-md is more defective in sporulation and spore viability than the rnh1∆ rnh201∆ mutant, since RNase H1/2 can only remove R-loops but not prevent their formation [17,31].

An interesting question is what SPO11 is doing at TRCs in premeiotic S phase. During wild-type meiosis, Spo11 is recruited to chromosome axes after DNA replication where it generates prophase I DSBs [128–130]. During premeiotic S phase, Spo11 may generate DSBs to relieve the topological constraints of the positive supercoils on unreplicated DNA, which would be a novel function for this highly conserved endonuclease.

The fact that progression through MI in NDT80-mid sen1-md, mek1∆ sen1-md and sen1-md spo11∆ diploids was similar indicates that SPO11-independent DSBs do not trigger the MRC. Spo11-generated DSBs occur on the chromosome axes, after which the Mec1/Tel1 checkpoint kinases phosphorylate Hop1 [48]. Mek1 binds to phosphorylated Hop1 via its FHA domain and then activates itself by phosphorylation of the Mek1 activation loop in trans [50,51,131]. We propose that the SPO11-independent DSBs arise from processing of persistent R-loops formed at TRCs during premeiotic S phase and thus are likely not forming on chromosome axes, in which case Mek1 (and therefore the MRC) would not be activated [50,51,131]. A similar phenomenon was observed in C. elegans mutants lacking RNase H activity where R-loops accumulated in the germ line [26]. An increased frequency of DSBs was observed in rnh-1.0 rnh-2; spo11 worms compared to spo11 alone and those DSBs failed to trigger the DNA damage checkpoint [26]. SPO11-independent DSBs have also been observed during premeiotic S phase in mouse spermatocytes [132]. Therefore, it is possible that a non-canonical function for Spo11 in regulating R-loops during meiosis is conserved.

SEN1 promotes the repair of SPO11 DSBs during prophase I

Several studies have shown that DNA:RNA hybrids can form on the resected ends of DSBs in a variety of species both in vegetative and meiotic prophase I cells [13,14,16,17,26,31,33,34]. In budding yeast meiosis, DNA:RNA hybrids that accumulated at the resected ends of DSBs in an rnh1∆ rnh201∆ hpr1∆ diploid impaired recombinase binding and interfered with DNA repair [17].

Our work indicates that SEN1 also functions to promote repair of SPO11 DSBs. Meiotic progression in sen1-md was delayed due to the MRC that is triggered by Spo11-generated DSBs [50–53]. Furthermore, when sen1-md diploids were allowed to progress through meiosis without the MRC-mediated delay, nuclei were fragmented, indicating the presence of broken chromosomes. DNA:RNA hybrid foci were elevated in sen1-md strains throughout prophase I. Chromosome synapsis was defective in the sen1-md mutant but this defect was significantly improved when SEN1 was expressed using the REC8 promoter. We propose that SEN1 removes DNA:RNA hybrids from the ends of Spo11 prophase I DSBs and that this removal is necessary for recombinases to bind, similar to what has been observed in rnh1∆ rnh201∆ hpr1∆ diploids [17], In this case, the synapsis defect is an indirect effect of a problem with DSB repair.

Interestingly, despite the strong synapsis defect, COs were reduced only by ~25% in the sen1-md diploid. This value is not very different from the 40% reduction in COs in the rnh1∆ rnh201∆ hpr1∆ diploid (Fig 5) [17]. One explanation for the relatively minor CO defects is that RNase H1/2 and Sen1 function redundantly in removing the DNA:RNA hybrids at the ends of DSBs, although the rnh1∆ rnh201∆ sen1-md diploid was too sick to directly test this idea. One difference between the RNaseH mutant is that it decreased both COs and NCOs, whereas only COs were decreased by sen1-md (Fig 5) [17]. The ZMM crossover pathway is required for chromosome synapsis [64,102,106]. An intriguing idea is that Sen1 specifically removes DNA:RNA hybrids from DSB ends that are destined to be repaired by the ZMM pathway. In this case, most of the COs observed in the sen1-md mutant would be generated by the action of structure-selective nucleases such as Mus81-Mms4 that do not contribute to synapsis. RNase H may instead remove DNA:RNA hybrids from DSBs that are processed by other pathways to form either COs or NCOs. Further work is necessary to test this idea.

Strains and media

All the strains used in this study were derived from the SK1 background. Genotypes of each strain are listed in S1 Table. Growth media are described in [133]. Sporulation (Spo) medium consisted of 2% potassium acetate. All experiments were carried out at 30°C.

Gene deletions were made using polymerase chain reaction (PCR) generated fragments to replace open reading frames (ORFs) with either kanMX6 (confers G418 resistance), natMX4 [confers nourseothricin (NAT) resistance], or hphMX4 (confers Hygromycin B resistance) using the plasmids pFA6a-kanMX6, p4339, or pAG32, respectively. Deletions were verified by PCR using a forward primer upstream of the deleted ORF and a reverse primer within the drug marker. In addition, the absence of the wild-type gene was confirmed by failure to detect a PCR product using the same forward primer and a reverse primer in the ORF. For the ndt80∆::hphMX4 haploid parents of AM6697, flanking primers generated different sized PCR products for the NDT80 and ndt80∆::hphMX4 alleles.

To make sen1-md strains, a 2.5 kb fragment was amplified from pMJ787 containing kanMX6 133 bp upstream of P_CLB2-3xHA. This fragment was designed with homologies such that it replaces the 50 bp immediately upstream of the SEN1 ATG with the CLB2 promoter. G418^R haploid transformants of both mating types were screened by colony PCR to confirm the insertion of the fragment directly upstream of SEN1 using a forward primer located 814 bp upstream of SEN1 and a reverse primer 740 bp downstream of the SEN1 start codon.

The P_REC8-SEN1 and P_REC8-sen1-∆N URA3 integrating plasmids, pNH410 and pBG27, respectively, were digested with PshAI to target integration to codon 1811 of SEN1 and transformed into the sen1-md diploid, NH2667. Integration of the plasmids was confirmed by complementation of the sen1-md sporulation defect and/or immunoblots probed with α-Sen1 antibodies.

The sen1-md NDT80-mid diploid, NH2667::pNH317² was created by transforming the haploid parents of NH2667 with the URA3 NDT80-mid integrating plasmid, pNH317, digested with SnaBI to target integration upstream of the NDT80 ORF (note that NDT80-mid is dominant to NDT80) [52]. Proper plasmid integration was confirmed by PCR with a forward M13 primer present in the vector sequence with a reverse primer 1.9 kb upstream of the NDT80-mid ORF.

Diploids containing rad53∆::kanMX6 were constructed in several steps. For rad53∆ strains to be viable, the SML1 gene must be deleted [121]. The SML1 gene was replaced with 3HA epitopes using pMPY-3HA as a template to amplify a 3HA-URA3-3HA cassette flanked by sequences directly upstream and downstream of the SML1 ORF. This cassette was transformed into S2683 selecting for Ura⁺ colonies. The cassette was popped out using 5-fluororotic acid, which selects against cells containing the URA3 gene, replacing the SML1 ORF with the 3HA sequence [134,135]. RAD53 was then deleted using kanMX6. This haploid was crossed to NHY1210 sen1-md spo11∆::natMX4 (a haploid parent of NH2689). Spore colonies from tetrad dissection of this diploid were screened for G418 resistant non-parental ditypes that should contain both rad53∆::kanMX6 and kanMX6::P_CLB2-SEN1. The presence of both genes was confirmed by PCR. For sml1∆::3HA, the PCRs used primers that flank the SML1 ORF. The NH2840-23-3 and NH2840-71-1 segregants were NAT sensitive and therefore contained SPO11. The NH2840-23-4 and NH2840-33-4 segregants were resistant to NAT and therefore contained spo11∆::natMX4. The appropriate haploid pairs were mated to make NH2843 and NH2844, respectively.

To make a P_CUP1-IME1 sen1-md diploid, a PCR fragment containing kanMX:P_CLB2-SEN1⁷⁴¹ was amplified using genomic DNA from NH2667 as the template and primers located 571 bp upstream of kanMX and 741 bp downstream of the SEN1 ATG. This fragment was transformed into the P_CUP1-IME1 haploids, 39806 and 39823, and the presence of the sen1-md allele confirmed by PCR. The haploids were then mated to generate the NH2842 diploid. The P_REC8-SEN1 allele was integrated into NH2842 using PshAI digested pBG45.

The sae2∆ sen1-md diploid, NH2846, was constructed by first replacing SAE2 with natMX4 in the haploid parents of NH716. The resulting two sae2∆::natMX4 haploids were transformed with the kanMX::P_CLB2-SEN1⁷⁴¹ fragment described above to introduce the sen1-md mutation. The two haploids were then mated to make NH2846.

Plasmids

The genotypes of plasmids are listed in S2 Table. Any sequences amplified by PCR were sequenced in the final plasmids by the Stony Brook University DNA Sequencing Facility or by Plasmidsaurus (https://www.plasmidsaurus.com/).

The SEN1 gene was fused to the REC8 promoter in pNH257 to make pNH410 using the NEBuilder HiFi DNA Assembly Master Mix (hereafter referred to as Gibson assembly or GA) (New England Biolabs E2621). A 2μ SEN1 plasmid containing a genomic fragment from chromosome XII with the coordinates 989631-1000863 was used as the template for PCR [136]. The 6696 bp SEN1 ORF was amplified as three separate fragments of approximately 2 kb. The SEN1a fragment contains 25 bp of homology to the BamHI cut site of pNH257 (immediately downstream of P_REC8) fused to SEN1 codons 1–790. SEN1b contains codons 783–1555 and SEN1c contains codons 1547–2231 along with 210 bp of 3’ untranslated sequence and 25 bp of homology to the EcoRI cut site of pNH257. The three PCR fragments were incubated with BamHI/EcoRI-digested pNH257 in a GA reaction following the manufacturer’s instructions. The P_REC8-SEN1 allele was moved into a TRP1 integrating vector by subcloning a 7.2 kb XhoI/NotI fragment from pNH410 into XhoI/NotI digested pRS304 to make pBG45. The P_REC8-sen1-∆N allele in pBG27 lacks the first 1003 codons of SEN1. pNH410 was used as the template to generate a 1691 bp PCR fragment containing homology to the EcoRI site on pNH257 along with an ATG and SEN1 codons 1004–1555. This fragment was then joined to the SEN1c fragment and ligated into pNH257 using GA as described for pNH410. The P_SEN1-SEN1 plasmid, pBG28, was constructed using a SEN1a fragment that had 25 bp of homology to the NotI site in pRS306 fused to a 2.8 kb sequence containing 0.8 kb located immediately upstream of the SEN1 ORF, as well as the first 791 codons of the gene. The SEN1b fragment was the same as for pNH410, while the SEN1c fragment contained the same fragment of SEN1 with 192 bp 3’ untranslated sequence and 3’ homology to the SalI site in pRS306.

Timecourses

Liquid sporulation was performed as described in [133] with the following changes. Diploids were streaked out from the -80°C freezer onto YPD medium supplemented with complete powder [133] and incubated at 30^oC for 2–3 days. Single colonies were inoculated into 8 mL YPD and incubated at 30^oC for 24 h. 1.0 mL and 1.7 mL of inoculum were then diluted into 200 mL YPA in two 2 L flasks and incubated at 240 revolutions per minute (RPM) for 18 hours at 30^oC. The optical density at 660 nm (OD₆₆₀) of a 2-fold dilution in YPA of the culture was read using a SPECTRONIC 200 spectrophotometer. Diluted cultures with OD₆₆₀ readings between 1.6 and 2.3 were pelleted, washed with water and resuspended in the volume of Spo medium required for a cell density of 3 x 10⁷ cells/mL in 2 L flasks. The cells were incubated in a 30°C shaker rotating at 250 RPM. At each timepoint, cells were fixed with a 1/10 volume of 37% formaldehyde for DAPI staining (Vector Laboratories H-1200-10). Formaldehyde-fixed cells were processed for whole cell immunofluorescence as described in (Ziesel et al. 2022). For protein and RNA sequencing samples, 5 ml sporulating culture were pelleted in a 15 ml conical tube, the supernatant was discarded, and the cell pellets stored at -80°C. For DNA physical analysis, 10 mL culture was added to 1 mL 0.5 M EDTA and 10 mL 100% ethanol and stored at -20°C. For nuclear spreads, 6.7 mL of sporulating culture were immediately processed as described in [137]. Cells were prepared for flow cytometry analysis as described in [138]. The cells were processed by the Stony Brook University Flow Cytometry Facility using a Becton Dickenson LSRFortessa flow cytometer. The data were analyzed using Beckman Coulter Kaluza software.

To induce IME1 transcription in the P_CUP1-IME1 diploids, cells were transferred to Spo medium for 2 hours after which 100 mM CuSO₄ (Sigma #209198) was added to a final concentration of 50 μM.

Immunoblots and antibodies

Total protein was isolated from frozen cell pellets and utilized for immunoblot analyses as described in (Weng, Wan et al. 2024). The sources and conditions used for various antibodies are listed in S3 Table. The Sen1 antibody was generated by Covance Research Products (now Labcorp Drug Development) in a guinea pig using the peptide Ac-CFSDDVSFIPRNDEPEIK-amide (amino acids 2002–2019).

Nuclear spreads

For localization of meiotic axis and central region proteins of the SC (Fig 6): Strains homozygous for ndt80::∆hphMX4 were cultured overnight in BYTA (1% yeast extract, 2% bactopeptone, 1% potassium acetate, 50 mM potassium phthalate) then sporulated in 1% potassium acetate for 5 hours. Nuclei from these cells were surface-spread on glass slides and imaged as described in [139]. Affinity purified rabbit anti-Zip1 (1:150), mouse anti-Gmc2 (1:800), and rabbit anti-Red1 (1:200) were used [59]. Secondary antibodies conjugated with Alexa Fluor dyes (Jackson ImmunoResearch) were used at a 1:200 dilution. Microscopy and image processing were performed using a Deltavision RT imaging system (General Electric) adapted to an Olympus (IX71) microscope.

For localization of DNA:RNA hybrids (Fig 8): Meiotic cultures were spheroplasted and their nuclei spread and immunostained as described in [137]. Slides treated with 5 units of RNase H (New England Bioscience M0297L) were incubated at 37^oC for 1 h in a moist slide box after blocking and then washed with 1xTBS prior to incubation with antibodies. Spreads were incubated with antibodies at 4^oC in a moist slide box. Slides were first stained for DNA:RNA hybrids with the mouse DNA:RNA hybrid antibody, S9.6, (Kerafast ENH001) diluted 1:1000 in 1% BSA and incubated overnight. After washing slides as described, goat α-mouse IgG-Alexa 594 (Fisher A32742) secondary antibody diluted 1:1000 in 1% BSA was applied to slides and incubated for two hours. Micrographs were taken with a ZEISS Axio Imager.Z2 with a Zeiss Plan-Apochromat 100X objective and analyzed using ZEN 3.5 software.

Physical analysis of DSBs and repair products

Genomic DNA was isolated from ethanol-fixed sporulation samples using the MasterPure Yeast DNA Purifiation Kit (Biosearch Technologies MPY80200) and quantified with the Qubit 1x dsDNA HS Assay Kit (Invitrogen Q33231). Digestion of samples, gel electrophoresis, Southern blotting, and hybridization with radiolabeled probe for analysis of the HIS4LEU2 hotspot were performed as described in [140] with the following changes: the blotted membrane was incubated in prehybridization solution [0.5% SDS, 0.5 mg/mL sheared salmon sperm DNA, 6xSSC (29.22 mg/mL sodium chloride, 14.7 mg/mL sodium citrate, HCl to pH 7.0)] at 68^oC for 2 hours. ³²P-labeled probe with at least 1x10⁶ counts per minute was added to the hybridization solution (0.5% SDS, 0.5 mg/mL sheared salmon sperm DNA, 6xSSC, 0.1 mg/mL dextran sulfate) and incubated with the membrane for 16 h at 68^oC. The membrane was washed at 68^oC three times for 30 minutes with ~40 mL of Wash Buffer 1 (2x SSC, 0.1% SDS) and then twice with Wash Buffer 2 (0.2x SSC, 0.1% SDS). It was then washed at room temperature for 5 minutes each with Wash Buffer 2 and Wash Buffer 3 (0.1x SSC). The blots were imaged using a FujiFilm FLA-7000 phosphoimager and Multi-Gage software.

Illumina RNA sequencing analysis

After growth in YPA at 30°C, the SEN1 and sen1-md diploids were transferred to Spo medium at a density of 3 X 10⁷ cells/ml and 5 ml aliquots were taken at various timepoints, pelleted and frozen at -80°C. The cell pellets were sent to GENEWIZ (now Azenta) for RNA extraction and polyA RNA enrichment. The kit used to prepare the libraries was the NEBNext Ultra RNA Library Prep, PolyA (New England Biolabs). The adapter sequences used to generate the RNA sequencing data were:

1. Read1: AGATCGGAAGAGCACACGTCTGAACTCCAGTCAC
2. Read2: AGATCGGAAGAGCGTCGTGTAGGGAAAGAGTGTA

The RNA sequencing data were mapped to the yeast genome using Bowtie 1.1.2 and custom Python scripts with Yeast PacBio 2016 as the reference genome (https://yjx1217.github.io/Yeast_PacBio_2016/data/) [141, 142]. Fragments/kb transcript/million (fpkm) values were determined using TopHat2 [143] which were then used to calculate the standard scores (Z-scores) for each gene using the formula (fpkm_n - fpkm_ave)/SD, where n = timepoint, fpkm_ave= average of the fpkm values for each timepoint and SD is the standard deviation. Genes were clustered based on their pattern of Z values using Clustal 3.0 and then visualized as a heat map plotted by JavaTree. The sen1-md strain was not weighted and these genes were “passengers” whose location was set by the SEN1 cluster. The Illumina RNA sequencing data are available in the National Institutes of Health Gene Expression Omnibus (GEO) with the accession number GSE292703. (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE292703).

Nanopore dRNA sequencing analysis

Data processing

Base-calling was conducted on a custom Linux workstation equipped with an AMD Threadripper PRO 5955WS CPU, dual Nvidia RTX 4090 GPUs (MSI V510-007R), and 256 GB of RAM. The resulting FASTQ files were analyzed using the Epi2Me Labs wf-transcriptomes Nextflow pipeline (version 25.04.06) (https://nanoporetech.com/products/analyse/epi2me) with the -direct_rna option enabled. Reads were aligned to the SK1 reference genome using minimap2 with a preset override of -x splice -uf -k 14 -G1k, optimized for direct RNA reads with short introns. The reference genome was indexed in minimap2 with the -k 15 setting. The workflow produced aligned BAM files for downstream analysis.

The genomic coordinates of CUTs, SUTs, and ORF-T were originally annotated based on the S288C genome (version V56, 2007-04-06) [87]. These coordinates were first converted to genome version V64 (2011-02-3) using liftOver and subsequently remapped to the SK1 genome. This was achieved by retrieving the DNA sequences corresponding to the annotated regions with bedtools getfasta and aligned to the SK1 genome using minimap2. The resulting SAM alignment files were converted to sorted and indexed BAM files with samtools, which were then transformed to BED format using bedtools bamtobed for downstream analysis.

Differential gene expression was performed as follows. Read count quantification was determined using featureCounts (v2.1.1; Subread package). Reads were assigned to annotated ORF-T features defined in the SK1 genome using the following options: -F SAF-format annotation) and -O (minimum MAPQ > 20). The resulting count matrix was analyzed in R Studio (version 2025.09.1) with the DESeq2 package. Raw counts from two SEN1 and two sen1-md dRNA-seq replicates were normalized for library size, and differential expression analysis was performed using DESeq2’s negative binomial model with Wald test statistics. Genes with a false discovery rate (FDR) < 0.01 and an absolute log₂ fold change > 1 were considered signficantly differentially expressed.

To generate the normalized read count plots shown in Fig 3D and 3G, BAM alignments from replicates samples of SEN1 and sen1-md were merged using samtools merge. Primary mapped reads were then extracted from the merged BAM files using samtools view with option -F 0x904 for regions chrXIV:173000-184300 and chrX:603000-609000, respectively. The filtered BAM files were converted to BED format and sorted first by the 3’-end position and then by 5’-end position of each read. Because the total number of aligned reads in the wild-type sample (15,252,836) were larger than that in sen1-md (12,637,870), the step size for individual reads was set at 1 for SEN1 and 1.2 for sen1-md to normalize read density across samples. Datagraph was used to plot these figures.

To generate the heatmaps shown in Figs 3E, 3F, S3C, and S3D, deeptools bamCoverage was used to calculate normalized read coverage with the parameters --normalizeUsing CPM, --samFlagExclude 16 (for Watson strand) or –samFlagInclude 16 (for Crick strand), -bs 1, --effectiveGenomeSize 12053285. The resulting bigwig files were then used to generate the matrix files with the deeptools computeMatrix function, followed by heatmap visualization using deeptools plotHeatmap. The Nanopore dRNA sequencing data have been deposited in GEO with the accession number GSE310121. https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE310121.

Statistical analysis of the direct RNA-seq data was performed using DESeq2’s negative binomial model and Wald test statistics [DOI 10.1186/s13059-014-0550-8]. Genes with an adjusted p-value (i.e., false discovery rate) < 0.01 and an absolute log2 fold change > 1 were considered significantly different in our analysis. Under these criteria, RAD50 is not significantly changed, as it shows a log2 fold change of 0.61 in sen1-md compared with SEN1 and an adjusted p-value of 0.0191, which does not meet our predefined significance thresholds (Fig 3D). IME1 transcript levels were determined to be significantly reduced in sen1-md relative to SEN1 based on DESeq2’s negative binomial model and Wald test statistics. IME1 shows a log2 fold change of –1.25 and an adjusted p-value of 5.56 × 10 ⁻ ⁸. Therefore, the decrease in IME1 expression is statistically significant under our defined criteria (Fig 3G).