Control of meiotic pairing and recombination by chromosomally tethered 26S proteasome

See allHide authors and affiliations

Science  27 Jan 2017:
Vol. 355, Issue 6323, pp. 408-411
DOI: 10.1126/science.aaf4778

Proteasomes and SUMO wrestle chromosomes

Meiosis is the double cell division that generates haploid gametes from diploid parental cells. Pairing of homologous chromosomes during the first meiotic division ensures that each gamete receives a complete set of chromosomes. The proteasome, on the other hand, is a molecular machine that degrades proteins tagged for destruction within the cell (see the Perspective by Zetka). Ahuja et al. show that the proteasome is also involved in ensuring that homologous chromosomes pair with each other during meiosis. Rao et al. show that the SUMO (small ubiquitin-like modifier) protein, ubiquitin, and the proteasome localize to the axes between homologous chromosomes. In this location, they help mediate chromosome pairing and recombination between homologs.

Science, this issue p. 349, p. 408; see also p. 403


During meiosis, paired homologous chromosomes (homologs) become linked via the synaptonemal complex (SC) and crossovers. Crossovers mediate homolog segregation and arise from self-inflicted double-strand breaks (DSBs). Here, we identified a role for the proteasome, the multisubunit protease that degrades proteins in the nucleus and cytoplasm, in homolog juxtaposition and crossing over. Without proteasome function, homologs failed to pair and instead remained associated with nonhomologous chromosomes. Although dispensable for noncrossover formation, a functional proteasome was required for a coordinated transition that entails SC assembly between longitudinally organized chromosome axes and stable strand exchange of crossover-designated DSBs. Notably, proteolytic core and regulatory proteasome particles were recruited to chromosomes by Zip3, the ortholog of mammalian E3 ligase RNF212, and SC protein Zip1 . We conclude that proteasome functions along meiotic chromosomes are evolutionarily conserved.

Global homolog juxtaposition and local recombination during meiosis are temporally and spatially coordinated via multiple interdependencies. Pairing between homolog arms occurs at substantial levels before double-strand break (DSB) formation and is further stabilized by assembly of the synaptonemal complex (SC) between proteinaceous chromosome axes (1). Centromeres, by contrast, are initially coupled with one or several nonhomologous partners, as best documented in yeast and plants (2, 3). As SC assembly reaches completion, centromere coupling is replaced by homologous centromere pairing (2, 4). Recombination involves processing of Spo11-induced DSBs into crossovers (COs) and noncrossovers (NCOs), i.e., interhomolog recombination products with or without exchange of flanking chromosome arms (1). In several organisms, recombination mediates SC assembly (1) and release of nonhomologous centromere coupling (24). The ZMM group of proteins, including yeast SC protein Zip1 and E3 ligase Zip3, coordinately control synapsis and CO-specific strand exchange (5, 6). Before their role in synapsis, SC proteins, including Zip1, provide coupling between nonhomologous centromeres (2, 3).

In budding yeast, meiotic progression in zmm deletion mutants is temperature sensitive (ts) (5). A screen for ts mutants (7) identified a transposon insertion in PRE9, which encodes the universally conserved α3 subunit of the proteasome’s core particle (CP) (8). Although constitutively expressed like other CP components, α3Pre9 is nonessential (9, 10). The pre9Δ-proteasome is modestly impaired for assembly and proteolytic capacity (9, 11). pre9Δ diploids grow normally at higher temperatures, yet undergo prophase I arrest when shifted to ≥30°C following completion of premeiotic replication. Prophase I arrest also occurred in wild type (WT) exposed to the proteasome inhibitor MG132 [Fig. 1A, (i) to (iii), and fig. S1B] (12). Abrogation of DSBs restored meiotic progression in pre9Δ [Fig. 1A, (iv)]. Thus, at restrictive temperature, the pre9Δ-proteasome is functional for promoting meiotic divisions but is defective for resolving Spo11-induced events.

To delineate proteasome functions during meiosis, we first examined pairing between homologous chromosome loci tagged with green fluorescent protein (GFP). At times when WT exhibits maximum pairing, as indicated by predominance of fused GFP signals, pairing in pre9Δ along chromosome arms and at centromeres was markedly reduced (Fig. 1, B and C, and fig. S1D). Nonetheless, in both pre9Δ and WT, we detected ~16 signals of kinetochore protein Mtw1, which indicates pairwise centromere association (Fig. 1, B and D) (2, 4). Thus, in pre9Δ, unlike WT, centromeres are frequently coupled with nonhomologous partners.

Fig. 1 The proteasome controls meiotic progression and homolog pairing.

(A) Meiotic progression in WT, pre9Δ, or zip1Δ at (i) 23°C and (ii) 33°C, in the presence of MG132 (iii) (pink arrow), and (iv) in pre9Δ spo11Δ versus pre9Δ. N = 2; error bars indicate data range. (B) Localization of GFP-tagged centromere V (cenV, arrows) and total kinetochore signals (Mtw1-13xMyc) in nuclei with homologous centromere pairing (WT, left) or homology-independent centromere coupling (pre9Δ, right). Blue lines (DNA) trace edges of spread nuclei stained with 4′,6-diamidino-2-phenylindole (DAPI). Bar, 1 μm. (C) Fraction of prophase I–arrested nuclei with one, two, or 0 or >2 (“other”) GFP signals in strains homozygous for tetO arrays (i) near cenV, on the long arms of (ii) chromosome V, or (iii) on chromosome II; or (iv) heterozygous for tetO at nonallelic arm loci on chromosomes II and V (26). N = 2; error bars indicate data range (see also fig. S1D). (D) Number of kinetochore signals (mean ± SD) in WT (14.1 ± 2.5) and in pre9Δ (12.9 ± 3.8). For nucleus numbers, see table S1.

Pairing defects, including persistence of nonhomologous centromere coupling, are frequently associated with compromised DSB formation (1, 2). Physical analysis showed that indeed, both pre9Δ and MG132 exposure decreased DSB levels by ~35% (Fig. 2A and fig. S2). DSB reductions in a spo11 hypomorph (13) by >50%, however, were not accompanied by comparable defects in homologous centromere pairing, indicating distinct proteasome roles in DSB formation and homolog pairing [Fig. 1C, (i) and Fig. 2A].

Fig. 2 The proteasome controls DSB formation and the crossover-specific DSB-to-SEI transition (33°C).

(A) Cumulative DSB signals at five recombination hotspots at t = 7.5 hours in a sae2Δ background. N = 3, error bars represent SD (see also fig. S2). (B) One-dimensional gel Southern analyses of DSBs and COs [(i) and (ii)] and NCOs (iii) in pre9Δ, zip1Δ, and WT. Time points are 0, 2.5, 4, 5, 6, 7, 8.5, 10, 11, and 24 hours [asterisk in (iii): inverted loading order of zip1Δ samples at 2.5 and 4 hours]. (C) Quantitation of DSBs, COs, and NCOs in pre9Δ, zip1Δ, and WT. Dotted vertical line, t = 5 hours (see also fig. S4). (D) Two-dimensional gel Southern analyses of JM intermediates in WT and pre9Δ. (E) Quantitative analysis of JMs in WT, pre9Δ, and zip1Δ (see also fig. S5, A and B).

Processing of DSBs was also similarly affected by pre9Δ and MG132 exposure, as indicated by analysis at the HIS4::LEU2 recombination hotspot (fig. S3). Increased DSB steady-state levels observed under both conditions (Fig. 2, B and C, and fig. S4, A and B) were likely due to delayed DSB processing because proteasome defects did not increase DSB formation (above). Furthermore, proteasome defects reduced COs to <50% of WT levels, whereas NCOs formed normally (Fig. 2, B and C, and fig. S4). Thus, a structurally and functionally intact proteasome is required for DSB processing into COs, but not into NCOs. Efficient NCO formation in pre9Δ further intimates that homolog arms can undergo allelic recombination despite compromised pairing.

Crossover formation involves two successive strand invasions of 5′ resected DSBs, generating first interhomolog single end invasions (IH-SEIs) and then double Holliday junctions (IH-dHJs) (1). Three findings in pre9Δ suggested that a structurally intact proteasome mediates the CO-specific transition from DSBs to IH-SEIs: (i) DSBs persisted at late time points (Fig. 2C); (ii) both IH-SEIs and IH-dHJs were reduced and delayed; whereas (iii) intersister joint molecules (IS-JMs) appeared with normal timing (Fig. 2, D and E, and fig. S5, A and B). By inference, in pre9Δ as in zip1Δ (5), fewer IH-JMs form, but they accumulate to substantial levels at late times because of their increased life span. Recombination defects in MG132-treated cultures were qualitatively similar, but less severe (supplementary text and fig. S5, A and B). Thus, during meiosis, the proteasome mediates DSB processing specifically along the CO pathway.

At 23°C, COs were only modestly reduced in pre9Δ [fig. S6 and supplementary text]. By implication, high levels of chromosome missegregation observed in pre9Δ under permissive conditions are likely due to recombination-independent proteasome effects.

A role of proteasome function in synapsis was examined by monitoring chromosomal localization of Zip1, Zip3, and axis protein Red1. During WT leptonema, Zip1 colocalized with 10 to 20 Zip3 foci at centromeres (class I, Fig. 3, A and B; fig. S7, A to C) (14), followed by appearance of zygotene nuclei exhibiting Zip1 foci and lines along with ~40 Zip3 foci (class II). Pachytene entry was indicated by (semi)continuous Zip1 lines that contain >35 Zip3 foci (class III). Red1 also localized in multiple foci with little linear organization at earlier stages, but became organized as semicontinuous lines during pachynema (Fig. 3C) (15).

Fig. 3 The proteasome controls axis morphogenesis and synapsis (33°C).

(A) Zip1 and Zip3-GFP localization in WT class I, II, or III nuclear spreads; in pre9Δ; and in the presence of MG132 (bottom). Blue dotted lines as described in Fig. 1B. Bar, 1 μm. (B) Zip1 localization in WT and pre9Δ cultures. For Zip1 classes, see (A). (C) Red1-staining classes. Bar, 1 μm. (D) Quantitation of Red1 classes at t = 6 hours in PRE9 and pre9Δ in ndt80Δ background. “Other” indicates poorly spread nuclei. n ≥ 3; error bars represent SD; *P < 0.03, two-tailed t test. (E) Zip3-GFP focus numbers in a pdr5Δ culture treated with dimethyl sulfoxide (mock; 37.4 ± 16.1) or with MG132 (34.8 ± 15.0) in pre9Δ (38.3 ± 14.3) or in an untagged WT strain (5.3 ± 3.1) at the time of WT pachytene (t = 6 hours). Error bars represent SD (see also fig. S7C). (F) Zip1 localization in PRE9 spo11Δ (n = 3) and pre9Δ spo11Δ (n = 3) nuclei. Arrows indicate polycomplexes. Bar, 1 μm. (G) Quantitation of Zip1 localization in a spo11Δ background. #Polycomplexes (PC) were scored independently of patterns of chromatin-associated Zip1. Fractions are averages from two cultures at t = 3.5 and 5 hours or t = 4 and 6 hours, respectively; error bars represent SD; *P < 0.03, two-tailed t test. For nucleus numbers, see table S1.

Proteasome defects resulted in a marked reduction and/or delay in synapsis. Pachytene/class III nuclei reached only ~one-fifth (pre9Δ) or half (MG132) of WT peak levels, whereas class I and/or class II nuclei remained abundant (Fig. 3, A and B, and fig. S7D). Moreover, in pre9Δ, (i) Zip1 foci in class I nuclei remained associated with centromeres even at late times, a configuration normally limited to pre-DSB nuclei (fig. S7, A and B); (ii) short stretches of Zip1 polymerization occurred despite lack of homologous pairing (fig. S7, E and F); and (iii) Red1 failed to assume its longitudinally organized pachytene morphology, despite chromatin association at normal levels indicating defective axis morphogenesis (Fig. 3, C and D, and fig. S7G). Zip3 foci along chromosome arms, by contrast, were detected at normal numbers in pre9Δ or MG132 (fig. S7C) (t = 6 hours; Fig. 3, A and E), although we did not determine the extent of Zip3 association with its known targets (16). Thus, the proteasome mediates a coordinated transition involving homolog pairing, synapsis of longitudinally organized chromosome axes, and CO-specific strand exchange. Efficient Zip3 recruitment and NCO formation in pre9Δ and MG132 further argue for specific proteasome requirements in meiotic processes rather than generic cytotoxicity of proteasome inhibition.

Notably, pre9Δ affected Zip1 localization even in the absence of DSBs. In spo11Δ, Zip1 association with chromatin was limited to centromeres, whereas bulk Zip1 aggregated into polycomplexes (PCs) not associated with chromatin (Fig. 3F) (2). A less specific Zip1 association with chromatin in pre9Δ spo11Δ was indicated by the prevalence of nuclei with a multitude (>20) of Zip1 foci (“speckled”), whereas PCs were reduced, despite normal Zip1 abundance (Fig. 3, F and G, and fig. S7H). Thus, the proteasome controls association of Zip1 with chromatin independently of its effects on recombination.

How does the proteasome control chromosomal events during meiosis? In the 26S proteasome, the barrel-shaped 20S CP is flanked by one or two 19S regulatory particles (RPs) (8). The RP recognizes polyubiquitinated substrates and feeds them into the CP’s proteolytic tunnel. Meiosis-specific recruitment of fully assembled 26S proteasome to chromatin was indicated by localization analysis of fully functional, constitutive CP20S component α5 (α5Pup2-GFP; Fig. 4, A and D, and fig. S8) (17) and RP component Rpn12 [GFP-Rpn12 or hemagglutinin (HA)–Rpn12; Fig. 4, B and E, and fig. S9] (18). Low numbers of CP20S and RP19S foci in premeiotic cells (t = 0 hours) was consistent with proteasome occupancy of many genome regions in vegetative cells (Fig. 4, A and B) (19). Proteasome foci increased during premeiotic S phase and/or leptonema (t = 2 hours), with an intermittent dip or steadily, reaching peak levels of 29.2 (± 14.4) CP20S and 32.6 (± 16.6) RP19S foci, respectively. Proteasome foci peaked 1 hour after pachynema (Fig. 4, D and E) and were highest in nuclei exhibiting fragmented Zip1 staining as typically observed at pachytene exit. Thus, RP19S and/or CP20S recruitment to chromatin occurs in two waves, the first during leptonema and the second during pachytene exit. Both CP20S and RP19S tended to localize to chromatin regions devoid of Zip1 (Fig. 4, A and B), consistent with a proteasome role in displacing Zip1 from the corresponding chromosome regions.

Fig. 4 Meiosis-specific recruitment of the 26S proteasome to chromosomes is evolutionarily conserved.

(A) Localization of the proteasome CP (α5Pup2) and Zip1 on WT G0/1-arrested (t = 0 hours), leptotene (t = 2 hours), early pachytene (t = 5 hours), and late pachytene/diplotene (t = 6 hours) nuclear spreads. Bar 1, μm. (B) Same as (A), but for proteasome RP component Rpn12-GFP without the sample at 2 hours. (C) CP (α5Pup2-GFP) and Zip1 localization in spo11-yf (t = 5 hours), zip3Δ (t = 6 hours), and zip1Δ (t = 6 hours). Arrows, CP foci associated with polycomplexes. Bar, 1 μm. (D) CP focus counts in WT, spo11-yf, zip3Δ, and zip1Δ. Pachytene (Zip1 class III; red), meiotic divisions (blue). §For focus scoring see (12) and fig. S11. Asterisks indicate significant differences versus corresponding WT sample (two-tailed Wilcoxon rank sum test): **P < 0.01; *P < 0.05; n.s., P > 0.05. (E) RP focus counts in WT (see also Fig. 4D). (F) Requirement for C. elegans α3PAS-3 CP for SC axis and central element assembly. Region 1 (transition zone) excerpts from (i) WT and (ii) pas-3(RNAi) knockdown animals stained with HTP-3 and SYP-1. Arrows indicate HTP-3/SYP-1 coaggregates (see also fig. S12). Bar, 4 μm. (G) Squashed pachytene nucleus stained with antibodies against CP20S and SYP-1. Overlap is similar in ≥99% of nuclei (n > 100 nuclei) from 20 germ lines (see also fig. S13B). Bar, 2 μm. (H) Proteasome and SYCP3 localization along surface-spread mouse spermatocytes at the leptotene (n = 5), zygotene (n = 34), pachytene (n = 115), or diplotene stages (n = 47). Bars, 5 μm. Insets show magnified views of boxed areas (see also fig. S13, C and D). (I) Model for proteasome functions in homolog pairing, synapsis (SC), and crossover (CO) formation (see text for details).

CP20S recruitment to chromatin is controlled by a subset of meiosis-specific factors. Efficient recruitment of CP20S in a strain expressing catalytically inactive spo11-yf (13) suggest that meiotic DSBs are dispensable, at least for early CP20S recruitment, whereas α5Pup2 foci were substantially reduced at all times in both zip1Δ and zip3Δ (Fig. 4, C and D, and fig. S10). Polycomplexes in zip3Δ and spo11-yf also frequently contained at least one CP20S focus (>90%; Fig. 4C), indicating a potential Zip1 role in CP20S recruitment. However, a role of synapsis itself in CP20S recruitment is unlikely because spo11-yf and zip3Δ show similar defects in SC assembly (2, 5). More likely, Zip1 and Zip3 control CP20S recruitment via their spatial and/or functional association with coupled centromeres (2, 14, 16).

A conserved meiotic proteasome function was garnered from examination of the Caenorhabditis elegans germ line. Depletion of the α3Pre9 ortholog PAS-3 disrupted both early and late SC morphogenesis, as suggested by increased abundance in zygotene nuclei of PCs containing both axis protein HTP-3 and central element protein SYP-1 (Fig. 4F and fig. S12) (20, 21). Moreover, during pachynema, SYP-1 underwent premature and excessive polarization [fig. S12A, (iii)], a distribution normally limited to diplonema (21). Consistent with evolutionarily conserved functions along meiotic chromosomes, the CP20S extensively associated with SYP-1–decorated, pachytene chromosome squashes (Fig. 4G) [see also accompanying paper (22)].

Meiosis-specific proteasome association with chromosome axes was also detected in mouse spermatocytes, in which CP20S signal overlaps with SYCP3-stained chromosome axes (3, 22) at all meiotic stages except leptonema (Fig. 4H), whereas staining was absent in nonmeiotic testicular cells (fig. S13C).

Our findings support the following model: Before the DSB-induced homology search, nonhomologous interactions can become stabilized by promiscuous association with chromosomes of SC proteins, including yeast Zip1 [Fig. 4I, (i)]. Following its recruitment, the proteasome displaces SC proteins, restricting the latter to centromeres [Fig. 4I, (ii)]. At this stage, the proteasome may also ensure high DSB levels via its role in axis morphogenesis (1) and/or via removal of proteins that normally render DSB formation by Spo11 reversible (1, 23). Nonhomologous centromere coupling could provide a structural barrier against precocious nonallelic pairing (3). This pairing block is then destabilized by proteasome-mediated removal of SC proteins [Fig. 4I, (iii)]. Once integrated into the meiotic program, chromosomally tethered proteasome may have acquired lineage-specific functions and localization patterns. Our model postulates independent proteasome functions in homolog pairing, axis morphogenesis, and DSB formation that in turn control SC assembly and CO formation. The proteasome may also have distinct roles in the two latter processes. Notably, proteasome effects cannot simply be attributed to DSB reduction or failed Zip1 recruitment (supplementary text).

Interactions between the proteasome and polyubiquitinated substrates are assumed to be stochastic (8). Our findings suggest that the proteasome is targeted to chromosomal sites by appropriately modified substrates, analogous to its association, e.g., with the endoplasmic reticulum or with sites of DNA damage (8, 17, 19, 24). Notably, meiotic recombination sites in many organisms are occupied by RING finger E3 ubiquitin and/or SUMO ligases (6). Moreover, the ubiquitin-SUMO-proteasome system controls CO formation and positioning (7, 22, 25). Our work identifies chromosomal tethering of 26S proteasome during meiosis as an evolutionarily conserved mechanism for controlling protein dynamics at distinct chromosomal sites, thus ensuring homolog juxtaposition and CO exchange.

Supplementary Materials

Materials and Methods

Supplementary Text

Figs. S1 to S13

Tables S1 and S2

References (2741)

References and Notes

  1. Materials and methods are available as supplementary materials.
  2. Acknowledgments: We thank V. Matthews for assistance; A. Severson, A. Tartakoff, A. Almasan, B. Li, M. Lichten, R. Pezza, and R. Kondratov for advice; N. Hunter for communicating unpublished results; and A. Amon, S. Ben-Aroya, D. Dawson, S. Keeney, A. Peyroche, A. Villeneuve, and M. Zetka for strains and antibodies. Research was supported by NIH grants R15GM099056 (G.V.B.), R01GM104007 (J.L.Y.) and R01HD083177 (P.A.H.). Additional funding was provided by CSU’s Faculty Research Development Program, John Vitullo’s Bridge Funding Program, and the Center for Gene Regulation in Health and Disease (to G.V.B.).
View Abstract

Stay Connected to Science

Navigate This Article