Mammalian SWI/SNF chromatin remodeler is essential for reductional meiosis in males

BRG1, a catalytic subunit of the mammalian SWI/SNF nucleosome remodeler is essential for male meiosis1. In addition to BRG1, multiple subunits (~10-14) some of which are mutually exclusive, constitute biochemically distinct SWI/SNF subcomplexes, whose functions in gametogenesis remain unknown. Here, we identify a role for the PBAF (Polybromo - Brg1 Associated Factor) complex in the regulation of meiotic cell division. The germ cell-specific depletion of PBAF specific subunit, ARID2 resulted in a metaphase-I arrest. Arid2cKO metaphase-I spermatocytes displayed defects in chromosome organization and spindle assembly. Additionally, mutant centromeres were devoid of Polo-like kinase1 (PLK1), a known regulator of the spindle assembly checkpoint (SAC)2. The loss of PLK1 coincided with an abnormal chromosome-wide expansion of centromeric chromatin modifications such as Histone H3 threonine3 phosphorylation (H3T3P) and Histone H2A threonine120 phosphorylation (H2AT120P) that are critical for chromosome segregation3,4. Consistent with the known role of these histone modifications in chromosome passenger complex (CPC) recruitment, Arid2cKO metaphase-I chromosomes display defects in CPC association. We propose that ARID2 facilitates metaphase-I exit by regulating spindle assembly and centromeric chromatin.


57
Consistent with a putative role in cell division, we were able to detect abundant expression of both 58 ARID2 and BRG1 in metaphase-I spermatocytes by immunofluorescence (IF). They appeared 59 localized peripherally to the metaphase plate displaying limited chromosomal association 60 (Extended Data Fig .1d, top row). While transitioning to anaphase-I, both ARID2 and BRG1 were 61 detected at the spindle mid-zone while maintaining their peripheral association with segregating 62 chromosomes (Extended Data Fig .1d, bottom row). The primarily cytosolic distribution of 63 SWI/SNF combined with its narrow localization to chromosomes might suggest a preferential 64 association with the spindle and or kinetochores. Based on the mutant phenotype and spatial 65 distribution of PBAF subunits in late meiosis-I spermatocytes, we hypothesized that ARID2 66 regulates processes essential for chromosome segregation. These include chromosome 67 condensation, alignment and microtubule-kinetochore attachment.

68
We began by monitoring phosphorylated histone H3 on serine 10 (H3S10P), a known marker of

76
Metaphase-I spermatocytes that lacked spindle were demonstrably deficient in ARID2 relative to 77 those that displayed spindle in Arid2 cKO tubules (Extended Fig. 2b). The latter therefore represent 78 cells that might have undergone inefficient knockout and were considered internal controls.

82
Normally, a failure to establish microtubule -kinetochore attachments would activate the spindle 83 assembly checkpoint (SAC) which delays metaphase exit 14,15 . Given that Arid2 cKO spermatocytes 84 are deficient in microtubules we were curious to determine the status of SAC proteins many of 85 which associate with kinetochores. For this purpose, we chose Polo-like kinase1 (PLK1), a well-86 known SAC regulator 2,16 that also influences spindle assembly 17-22 . In control (Arid2 Het ) 87 metaphase-I spermatocytes, PLK1 foci were seen overlapping with Centromere protein A 88 (CENPA), a known histone H3 variant and component of functional centromeres 23 (Fig. 2b). In 89 contrast, 60 % of the scored metaphase-I spermatocytes in Arid2 cKO tubules lacked centromeric 90 PLK1 (PLK1 -, Fig. 2b). Those that still displayed centromeric PLK1 (PLK1 + ) appeared to express 91 ARID2 at almost normal levels, indicating that they were internal controls (Extended Fig. 2c).

92
Since PLK1 is known to be expressed prior to metaphase-I 24 , we monitored PLK1 in diplotene 93 spermatocytes where desynapsing axial elements are marked by HORMAD1 25 . Here, PLK1 foci 94 PLK1 was specific to Arid2 cKO metaphase-I spermatocytes. Interestingly, PLK1 was detected in 96 ARID2 and BRG1 co-immunoprecipitations (Co-IP) with nuclear lysates from spermatogenic cells 97 harvested at 3-weeks post-partum (Fig .2c). This suggests that PBAF might facilitate the 98 recruitment of PLK1 to centromeres.

100
Given that PLK1 is known to influence centromeric chromatin during mitosis 26 , we were curious 101 to determine the impact of ARID2 on histone modifications known to regulate cell division. These 102 include Histone H3 threonine3 phosphorylation (H3T3P) and Histone H2A threonine120 103 phosphorylation (H2AT120P), that are concomitantly deposited by HASPIN kinase 3,27 and spindle 104 checkpoint kinase BUB1 4 respectively. Interestingly, perturbation of H3T3P levels have been 105 shown to be detrimental to oocyte maturation in mouse 27 . Therefore, we first monitored H3T3P 106 in Arid2 WT (normal) metaphase-I spermatocyte squashes. Similar to metaphase-I oocytes, H3T3P 107 overlapped centromeres and was distributed along the inter-chromatid axes (ICA) (Fig. 3a, 1st 108 row). In Arid2 cKO preparations, the absence of centromeric PLK1 was used to differentiate the 109 mutant metaphase-I spermatocytes from internal controls (Fig .2b, Extended Fig. 2c). H3T3P 110 distribution was limited in internal controls (10%) resembling normal metaphase-I spermatocytes 111 (Fig. 3a,2nd row). In contrast to normal spermatocytes and internal controls, H3T3P was spread 112 chromosome-wide in majority of mutant metaphase-I spermatocytes (60%), with fewer mutants 113 (30%) displaying low levels of chromatid bound H3T3P (Fig. 3a, row 3,4). We also monitored 114 H3T3P abundance in spermatogenic histone extracts obtained from control and Arid2 cKO testes 115 at time points spanning the appearance of metaphase spermatocytes. By western blot, we failed 116 to notice enhanced H3T3P levels upon the loss of ARID2 (Extended Fig. 3a, top panel). Therefore, 117 the chromosome-wide expansion of H3T3P in mutants is unlikely to occur from increased HASPIN 118 activity.

124
Therefore, ARID2 is required to ensure the centromeric enrichment of both H3T3P and 125 H2AT120P.

135
To understand how ARID2 might restrict centromeric chromatin, we chose to investigate its role 136 in H3T3P regulation. During mitosis, H3T3P is initially detected chromosome-wide, only to be 137 later targeted to centromeres by the phosphatase activity of the CDCA2-PP1 complex, where 138 CDCA2 (also known as Repo-Man) is essential for targeting of the PP1 phosphatase 30,31 . While 139 the meiotic role of CDCA2 is unknown, it appeared to be highly transcribed at the meiosis-I to

145
Interestingly, these defects in CDCA2 targeting appeared consistent with the changes in H3T3P 146 distribution observed in ARID2 deficient metaphase-I spermatocytes (Fig. 3a, c). Therefore, 147 ARID2 might constrain H3T3P by regulating the chromosomal targeting of CDCA2.

149
To further investigate this mechanism, we examined BRG1 interacting proteins that were 150 previously identified by immunoprecipitation-mass spectrometry (IP-MS) 5 . While CDCA2 was 151 absent from this list of candidates, we noticed an association between BRG1 and Protein 152 phosphatase 2 regulatory subunit A alpha (PPP2R1A), a scaffolding subunit of PP2A, whose 153 phosphatase activity is essential for the chromosomal targeting of CDCA2 32 (Extended Fig. 4a;

168
In fact, the majority of Arid2 cKO metaphase-I spermatocytes with reduced PPP2CA levels (83%) 169 displayed H3T3P spreading (Fig. 3f). We conclude that ARID2 impedes H3T3P spreading during and internal controls from Arid2 cKO testis. In contrast, pINCENP was displaced from ICA and 181 centromeres, adopting a predominantly cytosolic localization pattern in mutant metaphase-I 182 spermatocytes (87%) (Fig. 4a). Similar to pINCENP, its meiotic kinase, AURKC was also depleted 183 from chromatin and enriched cytosolically in mutants (70%) compared to internal controls and 184 normal metaphase-I spermatocytes (Fig. 4b). Since AURKC is known to be partially compensated 185 by AURKB during female meiosis 36 , we were curious to examine AURKB localization in the 186 absence of ARID2. In normal metaphase-I spermatocytes, AURKB was restricted to 187 chromosomes, displaying prominent centromeric enrichment along with diffuse chromatid 188 localization. In contrast, we observed a striking loss of chromosomal AURKB in ARID2 deficient 189 metaphase-I spermatocytes (82%) (Fig. 4c). Therefore, CPC localization is severely impacted in 190 the absence of ARID2. We hypothesize that the mis-regulation of centromeric chromatin 191 contributes to the defect in CPC recruitment.

193
Our studies highlight a crucial requirement for ARID2 and therefore the PBAF complex in 194 reductional male meiosis. This adds to previously identified roles of SWI/SNF in spermatogonial 195 stem cell maintenance and meiotic prophase-I progression 5 . We show that ARID2 is essential for 196 (i) normal spindle assembly, (ii) PLK1 association with centromeres, (iii) maintaining centromere 197 identity and (iv) the chromosomal targeting of the CPC during metaphase-I, activities that normally 198 promote metaphase exit (Fig. 5). Therefore, it is reasonable to assume that PBAF might function 199 to alleviate the SAC in normal metaphase-I spermatocytes. Interestingly, the loss of ARID2 200 mimics spindle poisons, that inhibit microtubule synthesis and activate a SAC induced metaphase 201 arrest. In contrast to its meiotic role, the PBAF complex prevents aneuploidy during mitosis 41 , 202 implying a role in SAC activation. These distinct roles might be attributed to differences in mitotic 203 and meiotic PBAF associations. In spermatocytes we demonstrate that ARID2 interacts with 204 PLK1, a kinetochore associated protein, known to regulate anaphase onset during mitosis 42 and 205 oocyte meiosis 17 . Therefore, ARID2 might mitigate SAC by facilitating PLK1 recruitment to 206 centromeres. Additionally, the resolution of unattached kinetochores is necessary to overcome 207 SAC. Here, the role of ARID2 in spindle assembly is instructive. Interestingly, ARID2 is abundantly 208 detected at spindle poles in metaphase-I spermatocytes (Extended Fig. 2b). These sites usually 209 represent centrosomes, a microtubule organizing center, known to be present in spermatocytes 43 .

210
It is intriguing to speculate whether ARID2 regulates spindle assembly by promoting centrosome 211 activity. Alternatively, known mechanisms of spindle formation involving PLK1 and the CPC 18-212 20,44 , whose localization are dependent on ARID2 in metaphase-I spermatocytes cannot be ruled 213 out .

215
In addition to spindle assembly, PBAF also limits H3T3P and H2AT120P in spermatocytes. In the 216 case of H3T3P, ARID2 appears to employ a conserved mitotic mechanism involving the PP2A 217 regulated chromosomal targeting of CDCA2-PP1, a known H3T3P phosphatase 31,32 . The fact that 218 this pathway is governed by an association between PBAF and PP2A scaffolding subunit, 219 PPP2R1A is novel. Distinct mechanisms might govern the regulation of H2A120P by ARID2, 220 given that H2AT120P is thought to precede H3T3P. In this regard, the mitotic role of Remodeling

297
Preparation of nuclear lysates for immunoprecipitation:

298
Nuclear lysates were obtained from spermatocyte enriched populations isolated from 3-week-old 299 males by methods described previously 55,56 . Protein extracts for immunoprecipitation were 300 isolated from nuclei using a high salt extraction method 5 .

Co-Immunoprecipitation (Co-IP):
303 Co-IP were performed exactly as described previously 5 with minor modifications. These include 304 (i) Incubation of antibodies with nuclear lysates (500 -600 μg/IP) prior to capture of antibody-305 antigen complexes with magnetic protein A beads for 3 hours over a rotator at 4 C, (ii) Milder IP 306 wash conditions involving two IP buffer washes, followed by sequential washes once in high salt 307 wash buffer, low salt wash buffer and final wash buffer.