Molecular mechanisms for environmentally induced plasticity in the positioning of meiotic recombination at hotspots

In meiosis, Spo11/Rec12-initiated homologous recombination is clustered at hotspots that regulate its frequency and distribution across the genome. Intriguingly, the intensities and positions of recombination hotspots can change dramatically in response to intracellular and extracellular conditions, and can display epigenetic memory. Here, using the fission yeast Schizosaccharomyces pombe, we reveal mechanisms for hotspot plasticity. We show that each of six hotspot-activating proteins (transcription factors Atf1, Pcr1, Php2, Php3, Php5, Rst2) is rate-limiting for promoting recombination at its own DNA binding site, allowing each class of hotspot to be regulated independently by agonistic and antagonistic signals. We also discovered that the regulatory protein-DNA complexes can establish a recombinationally poised epigenetic state before meiosis. Notably, Atf1 and Pcr1 controlled the activation of DNA sequence-dependent hotspots to which they do not bind; and they do so by regulating the expression of other hotspot-activating proteins. Thus, while each transcription factor activates its own class of DNA sequence-dependent hotspots directly in cis, cross-talk between regulatory networks modulates in trans the frequency and positioning of recombination at other classes of DNA sequence-dependent hotspots. We posit that such mechanisms allow cells to alter the frequency distribution of meiotic recombination in response to metabolic states and environmental cues.


48
Introduction 5 the PCR primers listed in Table 3; reactions were carried out using a CFX96 Real Time System

206
When measuring the impact of protein deletions on hotspot activation, we included all possible 207 pair-wise combinations of hotspot-activating proteins and DNA sequence motifs. The removal 208 of the Php2, Php3 or Php5 proteins (which bind to and promote recombination at the CCAAT 209 motif) had no significant impact on rates of recombination at the M26 hotspot ( Figure 1C).

210
Similarly, the removal of Php2 or Php3 had no significant impact on rates of recombination at 211 the Oligo-C hotspot ( Figure 1E). Likewise, the removal of the Rst2 protein (which binds to and 212 promotes recombination at the Oligo-C motif) had no significant impact on the rates of 213 recombination at the M26 hotspot ( Figure 1C) or at the CCAAT hotspot ( Figure 1D). These 214 findings suggest that these two DNA sequence-specific, protein-DNA complexes functions with 215 high specificity to promote recombination.

217
The binding of Atf1-Pcr1 heterodimer to M26 DNA sites directly activates this class of hotspots 218 (5,34,35). Each subunit of this basic, leucine-zipper (bZIP) heterodimer binds one half-site of 219 the M26 DNA site (50) and each subunit is essential for hotspot activity (e.g., Figure 1C).

220
Strikingly, the removal of Pcr1 significantly reduced (by 74% and 70%, respectively) rates of 221 recombination at the CCAAT and Oligo-C hotspots (Figure 1D-1E). Removing Atf1 also 222 reduced recombination (by 46% and 30%, respectively) for CCAAT and Oligo-C ( Figure 1D-223 1E). The differential effects of Atf1 and Pcr1 on activation of the CCAAT and Oligo-C hotspots 224 are notable, given that the Atf1-Pcr1 heterodimer regulates M26-class hotspots, but these 225 differential effects are readily explained by discovery of underlying molecular mechanisms 226 (described subsequently). [These mechanisms involve the fact that Atf1 and Pcr1 are each 227 multifunctional (e.g., they regulate both recombination and transcription), and they form both 228 homodimers and heterodimers (34,47,50,51)]. Similarly, the removal of Php5 led to a 229 significant, albeit modest reduction in recombination at the Oligo-C hotspot ( Figure 1E). We 230 conclude that hotspot-activating transcription factors can control the activation of heterologous 231 DNA sequence-dependent hotspots to which they do not bind. The most parsimonious model is 232 that transcription factor "A" might regulate the expression level of transcription factor "B", 233 thereby affecting recombination at "B" DNA sites. This model predicts that DNA sequence 234 motif-dependent hotspots should be sensitive to the abundance of their binding/activating 235 proteins, which we tested as follows.

237
Hotspot-activating proteins are rate-limiting for promoting recombination at their own 238 DNA binding sites

240
To test for dose-dependent responses, we compared rates of recombination for each hotspot 241 DNA sequence motif (M26, CCAAT, Oligo-C) in meioses that were homozygous wild-type, 242 heterozygous wild-type/null mutant, and homozygous null mutant for their respective binding 243 proteins. Essentially identical results were observed for each of the six different hotspot-244 activating proteins (Atf1, Pcr1, Php2, Php3, Php5, Rst2). In every case, the rate of 245 recombination in the heterozygotes was intermediate between that of homozygous wild-type regression analysis of the entire data set revealed a robust, positive correlation between dose 7 using a second, independent experimental approach that attenuates the expression of genes 250 without changing their copy numbers (described in a subsequent section). The dose-dependent 251 responses, observed using two different approaches, support a very important conclusion: each of a particular, DNA sequence-specific, hotspot-activating protein will therefore affect rates of 255 recombination at the corresponding class of sequence-dependent hotspots in the genome.

257
Parental imprinting affects hotspot activation

259
Analyses of DNA binding in vitro and in vivo (47,51-55), as well as chromatin mapping 260 experiments (29,46,47,51), revealed that the hotspot binding/activating proteins occupy their 261 own DNA binding sites before meiosis. Moreover, the cis-acting regulatory modules (sequence-262 specific protein-DNA complexes) each promote recombination via the modification of chromatin 263 structure before and during meiosis (30). For these reasons, we conducted our analyses of 264 heterozygous wild-type/null mutant meiosis in two reciprocal configurations.

271
Following mating and during meiosis, each of the two configurations would be heterozygous for 272 the protein (e.g., atf1 + /atf1D) and for the alleles used to measure recombination (e.g., ade6-273 M26/ade6-M210). In other words, in meiosis the genotypes were identical for each paired,

279
For each of the six different hotspot binding/activating proteins (Atf1, Pcr1, Php2, Php3, Php5, 280 Rst2), the rate of recombination was higher when the protein was expressed in the haploid 281 parent with the hotspot DNA sequence motif (e.g., M26) than when the protein was expressed  Php3-Php5 complex) and the Oligo-C hotspot (Rst2). In fission yeast the depletion of nitrogen 296 source triggers sexual differentiation and the expression of key meiotic drivers (e.g., ste11) (51), 297 so we analyzed transcript levels in cells that had been cultured in nitrogen-free media for 60 8 expected, no atf1 transcript was detected in atf1D mutants and no pcr1 transcript was detected 301 in pcr1D mutants.

303
It was reported previously, based on quantitative Northern blot analyses of biological samples like those used in this study, that Atf1 promotes the expression of pcr1 and that Pcr1 represses 305 the expression of atf1 (47). We obtained the same result using qRT-PCR: the removal of Atf1 306 led to a reduction in the abundance of the pcr1 transcript ( Figure 3A) and the removal of Pcr1 expression levels that we observed in the mutants versus wild-type (0.6 and 2.9, respectively) 309 were similar to those observed previously (0.5 and 3.4, respectively), which provides reciprocal 310 confirmation of findings in the two different studies and further validates the qRT-PCR approach 311 used in this study.

313
Overall, considering all six genes analyzed, the removal of Atf1 reduced significantly the 314 expression of pcr1, php2 and php3 ( Figure 3A). The removal of Pcr1 significantly increased the 315 expression of atf1 and reduced that of rst2 ( Figure 3B). We conclude that some hotspot-316 activating proteins can regulate the expression of other hotspot-activating proteins. We also 317 found that Atf1 had no significant impact on the expression of php5 and rst2 ( Figure 3A) and 318 that Pcr1 did not affect significantly the expression of php2, php3 and php5 ( Figure 3B). Our

326
The finding that Atf1 controls the expression of proteins that bind to and activate the CCAAT

340
Interestingly, our qRT-PCR data also indicated that this cannot be the sole mechanism for 341 regulating hotspots in trans. Pcr1 was strongly required for the activation of the CCAAT box The modulation of expression levels by mutating other factors was also informative for a third 352 reason. In each case where expression of the hotspot-binding/activating protein was reduced (Figure 3), there was a corresponding reduction in rates of recombination at its DNA binding 354 site (Figure 1). This provides independent support for the conclusion, reported above using a 355 different experimental approach (Figure 2), that the hotspot-binding/activating proteins are rate-356 limiting for promoting recombination at their own DNA binding sites.

360
In diverse taxa, environmental cues and metabolic states can reshape the frequency distribution 361 of meiotic recombination across the genome (see Introduction). We reasoned that previously 362 unknown mechanisms for this plasticity must be related to the way that cells position 363 recombination at hotspots. Therefore, this study analyzed all six of the fission yeast proteins

382
We found that all six of the proteins known to directly regulate meiotic recombination hotspots in 383 fission yeast are rate-limiting for promoting recombination at their own DNA binding sites 384 (Figure 2). This property has also been reported for DNA sequence-specific, hotspot activating 385 proteins Rap1 of budding yeast (59) and Prdm9 of mammals (at least in some genetic contexts) 386 (9). Consequently, any signals that affect the functions of a given hotspot binding/activating 387 protein will affect the rate at which it promotes recombination at its own DNA binding sites

438
Multiple mechanisms for modulating recombination rates via hotspot-activating proteins

440
One prediction of our model is that hotspot-activating proteins (which are key 441 responders/effectors of environmental and metabolic cues) can regulate the activity of 442 heterologous, DNA sequence-dependent hotspots to which they do not bind. This prediction 443 was confirmed in our study. Atf1 and Pcr1 each controlled rates of recombination at the CCAAT 444 and Oligo-C hotspots (Figure 1D-1E), which are, respectively, bound and activated by the

459
The first type of mechanism for controlling hotspot activity is exerted through controlling the 460 abundance of the activator proteins. The exemplar here is that transcription factor "A" can 461 regulate the expression of transcription factor "B" (Figure 3), which is rate-limiting for hotspot 462 activity at "B" DNA sites (Figure 2). We documented the use of this same type of fundamental, 463 sequential, transcription-dependent pathway mechanism for two different classes of hotspots 464 (atf1 à php2, php3 à recombination at CCAAT hotspot; pcr1 à rst2 à recombination at Oligo-465 C hotspot; see pathway diagram in Figure 3C). We note that any other factors (cellular or 466 environmental) whose signals modulate the expression of the rate-limiting, hotspot-467 binding/activating proteins would also modulate rates of recombination at their DNA sites of 468 action. We do not mean to imply that the regulation of hotspots is exclusively mediated by

543
Third, the poised epigenetic states provide a potential mechanism for the observation that prior The number after each primer name indicates position of the 5' end of the primer relative to the first nucleotide in the start codon of the gene. F, forward primer; R, reverse primer.
binding sites and at DNA sequence-dependent hotspots to which they do not bind. (A)

844
Diagram of genetic crosses showing relative positions of ade6 alleles (see Table 1 for

900
local recombination rate ≥ 10), as is often done in the literature, the number of annotated