MORC3, a novel MIWI2 association partner, is an epigenetic regulator of 1 piRNA-dependent transposon silencing. 2

2 The PIWI (P-element-induced wimpy testis)-interacting-RNA (piRNA) pathway 3 plays a crucial role in the repression of TE (transposable element) expression via de 4 novo DNA methylation in mouse embryonic male germ cells. Various proteins, including 5 MIWI2 are involved in the process. TE silencing is ensured by piRNA-guided MIWI2 6 that recruits some effector proteins of the DNA methylation machinery to TE regions. 7 However, the molecular mechanism underlying the methylation is complex and has not 8 been fully elucidated. Here, we identified MORC3 as a novel associating partner of 9 MIWI2 and also a nuclear effector of retrotransposon silencing via piRNA-dependent de 10 novo DNA methylation in embryonic testis. Moreover, we show that MORC3 is 11 important for transcription of piRNA precursors and subsequently affects piRNA 12 production. Thus, we provide the first mechanistic insights into the role of this effector 13 protein in the first stage of piRNA biogenesis in embryonic TE silencing mechanism.


Introduction 16
genomic stability and induce damage to germlines, resulting in infertility 2 . Among the 23 TEs, LINE1 (long interspersed nuclear element-1) retrotransposons are the most 24 3-week-old MORC3 mutant mouse. 1 Next, DNA methylation levels of the LINE1 and IAP retrotransposons in the 2 spermatocytes, before the pachytene stage, of the 12-day-old MORC3 mutant mice were 3 examined by bisulfite sequencing (Fig. 3B and C). A significant reduction in the CpG 4 methylation levels of type A and TF LINE1 genes was observed in the MORC3 mutant 5 male germ cells, compared to the high DNA methylation of the control germ cells. 6 Meanwhile, DNA methylation of the IAP gene was not impaired by MORC3 mutation 7 and, correspondingly, no significant difference in IAP transcript levels was observed 8 (Fig. 3A). The similarity in the phenotypes, i.e., low DNA methylation at LINE1 regions 9 found in the MORC3 mutant, was observed in not only the postnatal testicular germ 10 cells but also in the sperm ( Fig. 3D and E). However, the degree of DNA methylation at 11 LINE1 regions in the MORC3 mutant sperm was slightly higher than that in the male 12 germ cells. These data suggested that some germ cells that have quite low methylation 13 levels at LINE1 regions would have been eliminated during the maturation process of 14 spermatogenesis in the MORC3 mutant mice by mechanisms yet to be elucidated. 15 16 Role of MORC3 in piRNA biogenesis 17 DNA methylation marks on LINE1 repeat elements are linked to the nuclear 18 events of the piRNA-dependent DNA methylation pathway, in which MIWI2 is involved. 19 To assess whether MORC3 plays any role in piRNA biogenesis, we purified the total 20 small RNAs from wild type as well as the MORC3 mutant embryonic testes (E16.5) and 21 subjected them to deep-sequencing. After trimming of miRNAs, rRNAs, and tRNAs 22 from both total small RNA libraries using the CLC Genomics Workbench, the reads 23 were mapped to the mouse genome (mm10, UCSC) and normalized to the read-depth of each library. Comparison of small RNA size distribution profiles among their libraries 1 revealed that the number of 25-31 nt small RNAs (corresponding to the length of 2 piRNAs) was significantly reduced in the MORC3 mutant testes ( Fig.4A and B). The 3 expression of miRNAs, used as an internal control, was almost equal in them 4 (supplementary Fig.2A) and the genomic annotations for 25-31 nt small RNAs using 5 the control and MORC3 mutant libraries indicated that much of reads were originated 6 from retrotransposon-related sequences. 7 Analysis of piRNAs of the MORC3 mutant testes revealed that a significant 8 decrease was largely contributed to the reduction of small RNAs derived from LINE 9 elements (LINE), LTR elements except for IAP (LTR (non-IAP)), and SINE elements 10 (SINE) (Fig.4C). To evaluate the expression of piRNAs corresponding to LINE1 in detail, 11 we mapped 25-31 nt small RNAs to type A and TF LINE1. While both sense and 12 antisense piRNAs levels were decreased to about half, the proportion of 1 st U sense and 13 10 th A antisense piRNAs was not altered in the MORC3 mutant mice compared to the 14 control mice ( Fig. 4D and E). 15 16 Reduction in MIWI2 and MILI loaded piRNAs in the MORC3 mutants 17 To analyze the function of MORC3 in more detail, we examined the piRNAs bound 18 to MIWI2 or MILI in embryonic testes. We performed immunoprecipitation of the germ 19 cells of wild type and the MORC3 mutant embryonic testes (E16.5) using the 20 anti-MIWI2 or -MILI antibodies and subjected them to deep-sequencing. Western blot 21 analysis revealed that the amount of MIWI2 and MILI proteins immunoprecipitated 22 with the corresponding antibodies were almost the same in the control and the mutant 23 testes (supplementary Fig.2B). However, the levels of MIWI2-and MILI-bound piRNAs 4, 5, and 6 primer sets were located around the retrotransposon sequence whereas Chr 1 10-1 and 2 primers were not. 2 The transcript levels of the regions close to LINE, LTR, and SINE sequences were 3 significantly decreased in the MORC3 mutant testes. However, in regions close to IAP 4 the transcript levels were not different between the control and the MORC3 mutant 5 testes. Essentially, similar patterns were also found in Chr 7 (2) and 8 clusters 6 (supplementary Fig.3). These data show that MORC3 has some retrotransposon 7 class-specific effect to regulate the transcription of piRNA precursor from embryonic 8 piRNA clusters. is the epigenetic regulator of piRNA-independent transposon gene silencing 34,35,39,40 . 16 Results of the current study identified MORC3 as a component of the MIWI2 protein 17 complex and a regulator of transposon repression in male germ cells. MORC3 is 18 essential for repression of retrotransposon via de novo DNA methylation involved in the 19 piRNA pathway. Herein, we show that MORC3 is a new player in piRNA biogenesis and 20 acts as an epigenetic regulator of transposon gene silencing via the piRNA pathway. 21 Immunoprecipitation of the sample from the ZF-MIWI2 Tg testis with anti-Flag 22 antibody and subsequent mass spectrometry analysis revealed a physical association 23 between ZF-MIWI2 and MORC3 (Fig.1A). This association likely involves direct binding, 24 as the experiments using 293T cells expressing MORC3 and MIWI2 confirmed the binding phenomena (Fig.1C). Basically, the other MIWI2 binding components expressed 1 in male embryonic germ cells are scarcely expressed under this condition. We found that 2 MORC3 expression in the fetal testis was restricted to germ cells and the protein was 3 predominantly localized to the nucleus and piP-body. This localization pattern is quite 4 similar to that of MIWI2 (Fig.1D). 5 The data of physical binding of MORC3 with MIWI2 prompted us to determine the 6 physiological role of MORC3. We noticed that the DNA methylation level of some classes 7 of retrotransposons was significantly lower and both MIWI2-and MILI-associating 8 piRNAs were reduced by about half in the MORC3 mutant embryonic testes . 9 Thus, MORC3 is at least partially required for piRNA-directed de novo DNA 10 methylation. It is conceivable that the loss of MIWI2 loaded piRNAs would be the 11 reason for the incomplete de novo DNA methylation in the MORC3 mutants (Fig.6C).

12
MORC3 mutants lead to de-silencing of LINE1 but there was no-effect on IAP 13 expression. The other proteins involved in piRNA production, such as TDRD1, TDRD9, 14 TDRKH, and EXD1, also regulate LINE1 expression rather than IAP [16][17][18]44,45 . In 15 contrast, in MILI and MVH mutants, de-suppression of IAP was observed to be as 16 robust as that of LINE1 11,14,19 . The loss of piRNA-dependent DNA methylation, which 17 was detected in MILI-and MVH-null testes, leads to meiotic arrest at the pachytene 18 stage in spermatogenesis 19,46 . In contrast, the vast majority of the MORC3 mutant 19 male germ cells escape such apoptosis and survive through the meiotic prophase ( Fig.2A  20 and B). However, the pregnancy rate of the normal female mice crossed with the 21 MORC3 mutant male mice was significantly lower and the fertility of the MORC3 22 mutant male mice was slightly impaired ( Fig.2C and supplementary Fig.1D). 23 Comparing to the phenotype of the MORC3 mutant mice, impaired fertility was not observed in EXD1 mutant mice, despite the reduction of MIWI2-bound piRNAs, 1 especially against LINE1. The reason for the normal fertility in EXD1 mutant mice 2 could be the milder effect of de-repression of retrotransposons 45 . Taken together, the 3 high survival rate of the MORC3 mutant germ cells in spermatogenesis is presumably 4 due to the milder degree of the deregulation of LINE1 and reduced levels of piRNAs, 5 compared to those in MILI-and MVH-null mutants. Although it is difficult to conclude 6 that the fertility defect of the MORC3 mutant male mice is dependent on the impaired 7 retrotransposon expression, some association may be anticipated. 8 Lastly, we wanted to understand the mechanism by which the impaired 9 retrotransposon expression was effected in the MORC3 mutant testis. Previous studies 10 revealed that piRNA precursors are conventional RNA pol II transcripts bearing 5′ caps 11 and 3′ poly (A) tails. The lengths of these nascent RNAs, which are transcribed from 12 transcriptional start sites in the piRNA clusters, are varied 26 . The coordinated increase 13 of pachytene piRNA precursor transcripts is controlled by common transcriptional 14 factors, such as the A-MYB protein, during spermatogenesis, especially around the 15 pachytene stage. It has been suggested that this regulation would be caused by the 16 compartmentalization and reorganization of TAD (topologically associating domain) 26,47 . 17 However, the mechanism of transcriptional regulation for embryonic piRNA precursors 18 remains unclear. 19 Our RT-qPCR data indicated that transcription of piRNA precursors originated 20 from retrotransposon regions in piRNA clusters and that the transcripts except those 21 of the IAP regions were significantly reduced in the MORC3 mutant embryonic testis 22 ( Fig.6A and B, and supplementary Fig.3). These data support the notion that the 23 regulatory mechanism of transcription including IAP regions on the piRNA clusters is different from that of transcription from the other retrotransposon regions in 1 gonocytes. 2 The H3K4me3 (H3 tri-methylated lysine 4) signals, that are histone modification 3 associated with RNA pol II transcription start sites, are highly observed in promoters of 4 retrotransposons including piRNA clusters in E16 testes 48 . In addition, the promoter 5 regions of piRNA clusters show much stronger H3K4me3 signals than coding gene 6 promoters in embryonic testis 49 . MORC3 has the PHD X/ZF CW domain, which 7 recognizes and binds to H3K4me3 marks, and the GHKL ATPase domain, which is 8 involved in ATP binding and hydrolysis 50,51 . Enrichment of MORC3 is found in 9 H3K4me3 sites of active gene promoters in the genome-wide ES cell analysis 51 . Taken 10 these observations together, our findings suggested that MORC3 might recognize and 11 bind preferentially to H3K4me3 marks at the promoter region of not only 12 retrotransposon genes but also piRNA clusters and regulate the transcription of piRNA 13 precursors via chromatin remodeling by hydrolyzing ATP in embryonic testis (Fig.6C). 14 Considering that MIWI2 is localized at the promoter regions of retrotransposons, it 15 is speculated that MORC3 proteins would be localized to the similar regions through a 16 protein complex formation and would play a role in the embryonic transposon silencing 17 via piRNA pathway 31 . Therefore, notably, MORC3 would have a direct contribution to 18 the initial stages of piRNA biogenesis rather than subsequent secondary piRNA 19 pathway, in which MIWI2 is involved. Thus, MORC3 affects not only secondary but also 20 primary piRNA biogenesis and regulates the transcription of piRNA precursors. 21 MIWI2 affects the chromatin state by targeting nascent RNAs transcribed from 22 piRNA-dependent retrotransposon regions using piRNAs as guides. The piRNAs may be 23 generated in part from the nascent RNAs transcribed by the regulation of MORC3.
Thus, MIWI2 plays the role as a recruiter of proteins related to de novo DNA 1 methylation on retrotransposons, while MORC3 controls the transcription of piRNA 2 precursors from retrotransposon related piRNA cluster regions of the entire genome. 3 MORC3 and MIWI2 might play an individual role on piRNA-dependent retrotransposon 4 regions, at the first step of primary piRNA biogenesis and after secondary piRNA 5 biogenesis, respectively. Thus, apart from their direct association as the protein complex, 6 the functional relationship between MORC3 and MIWI2 presumably is indirectly in this 7 transposon silencing system. 8 In this study, we have identified MORC3 as a nuclear effector of retrotransposon 9 silencing via piRNA-dependent de novo DNA methylation. Our data suggest that 10 MORC3 acts as an active regulator of piRNA precursors in the piRNA pathway and 11 subsequently affects piRNA production. Thus, we provide the first mechanistic insights 12 into effector protein at the first step of piRNA biogenesis in embryonic transposon 13 silencing mechanism. We generated conditional knockout mouse lines in which we inserted the targeting 5 vector (pDTMorc3neo) from the bacterial artificial chromosome clone (RP23-119J20) 6 containing mouse genomic DNA using the double Red recombination method as 7 described previously 52 . This targeting vector was constructed to produce an appropriate 8 conditional allele for Morc3. Parallel LoxP sites were inserted such that they flanked a 9 critical portion of the gene: the LoxP sites flanked exon 2 (containing a part of Histidine 10 kinase/HSP90-like ATPase superfamily domain) of Morc3 (supplementary Fig.4). The 11 DNA sequence of the targeting vector (pDTMorc3neo) is shown in supplementary Table  12 1. The linearized targeting vector was introduced into mouse ES cells (lab-made ES cells 13 originated from hybrid embryos derived from C57BL/6J and C57BL/6N mice) by 14 electroporation (GenePulser; Bio-Rad). The ES cells were grown on feeder cells in an 15 appropriate medium. Each colony was picked up and expanded, and the genomic DNA 16 of each clone was purified. To identify true homologous recombinant colonies, we 17 performed direct sequencing through PCR amplification using the outward primers for 18 the homologous region and the inward primers for the neomycin resistance gene. 19 Targeted ES cells were used for generating chimeric mice. The resulting chimeric mice 20 were backcrossed to C57BL/6J mice for over 5 generations. 21

22
Isolation of testicular germ cells 23 Whole testes (D12 after birth) were removed and treated with 1 mg/mL collagenase type 1 II and DNase1 for 15 min at 37°C. These samples were suspended with 0.25% trypsin 2 for 10 min at 37°C and washed with Hank's Stock Solutions (HBSS) (Nacalai Tesque 3 Inc., Kyoto, Japan). Anti-EpCAM (PE) antibody (CD326) (G8.8, BioLegend) was added 4 to the cell suspension with 5% BSA/PBS. After stirring for 90 min at 4°C, the immune 5 complex was washed with HBSS. The germ cells were sorted using a BD FACSAria 6 system (BD Biosciences, Franklin Lakes, NJ, USA) after immunostaining with the 7 anti-EpCAM antibody. The germ cell purity was verified by rerunning the sample after 8 sorting and was determined to be > 90%. 9 Isolation of sperm 11 Mouse sperm were isolated from the cauda epididymis of more than 10-week-old wild 12 type (C57Bl/6) and the MORC3 mutant mice by dissecting tissue in PBS. Sperm were 13 allowed to swim up for 30 min in 37°C with 5% CO2. To avoid the contamination of 14 somatic cells, the upper fraction was collected for bisulfite sequence analysis. Rockford, IL, USA) were used as the secondary antibody, and the signal was detected 5 using ECL Western Blotting detection reagents (GE Healthcare, Chalfont, Bucks, UK). 6 7 Bisulfite sequencing 8 Sorted germ cells (D12 after birth) were treated with bisulfite using the EpiTect Fast 9 DNA Bisulfite Kit (Qiagen, Valencia, CA, USA). The primers that we used for the 10 bisulfite sequencing of type A LINE-1 and type TF LINE-1 loci in this experiment 11 recognize more than several thousand loci in the mouse genome. Meanwhile, among 12 IAP retrotransposon species, only 1Δ1-type IAP, which is only 5-6 % of total IAP 13 elements, shows piRNA-dependent DNA methylation in 5′-LTR 53 . Unfortunately, 14 5′-LTR sequences of 1Δ1-type IAP could not be distinguished from those of the other 15 types of IAP. Therefore, we have used the primers specifically recognizing one locus of 16 1Δ1-type IAP (in this experiment, a locus in Chromosome 3). The PCR primer sequences 17 are described in the Supplementary information. The first and second rounds of PCR 18 amplification of the IAP and the H19 (GenBank accession no. U19619) were carried out 19 using Ex Taq (Takara, Japan). The PCR conditions were as follows: an initial round of 2 20 min at 94°C followed by 30 cycles of 30 s at 94°C, 30 s at 50°C (for H19) or 1 min at 55°C 21 (for IAP), and 1 min at 68°C, and the second round of 2 min at 95°C followed by 15 22 cycles of 30 s at 95°C, 30 s at 50°C (for H19) or 1 min at 56°C (for IAP), and 1 min at 23 72°C. Nested PCR was performed to amplify the H19 differentially methylated regions (DMRs). The PCR for the type A LINE-1 (GenBank accession. no. M13002) and type TF 1 LINE-1 (GenBank acc. no. D84391) was carried out using EpiTaq HS (for 2 bisulfite-treated DNA) (Takara, Japan) under the following conditions: 2 min at 94°C 3 followed by 30 cycles of 30 s at 94°C, 30 s at 55°C (for type A LINE-1) or 30 s at 50°C (for 4 type TF LINE-1), and 1 min at 68°C. The PCR products were purified using the 5 Eugene, OR, USA) were used as the secondary antibody for 1 h at room temperature. 10 Nuclei were counterstained with 1 μg/mL DAPI. Immunostained cryosections were 11 examined under a confocal microscope (LSM5Pascal, Carl Zeiss Co. Ltd). 12

HE staining 14
Eight-week-old testes and caudal epididymis of the Morc3 homozygous mutant and wild 15 type male mice were dissected and fixed in 4% paraformaldehyde for 2 h at 4°C. After 16 washing in PBS containing 10% and 20% sucrose, they were embedded in OCT 17 compound. The sections were stained with hematoxylin (Wako, Japan) and eosin. After 18 staining, the sections were treated with ethanol and xylene. The image was obtained 19 using the BZ-X710 microscope (Keyence, UK) 20 21 Immunoprecipitation, SDS-PAGE, and silver staining 22 The E16.5 or 3-week-old testes of wild type and the mutant mice carrying the 23 buffer (20 mM Tris [pH 7.5], 150 mM KCl, 0.5% TritonX, 2.5 mM MgCl2, 1 mM DTT, 1 RNasin (Promega), and a protease inhibitor tablet (Roche)). The lysates were 2 freeze-thawed and treated with benzonase. The homogenates were centrifuged at 15000 3 × rpm for 10 min at 4°C and the supernatant was subjected to immunoprecipitation 4 using anti-FLAG (FLA1; MBL) antibody overnight at 4°C and protein G Dynabeads 5 (Invitrogen, US) for 1.5 h at 4°C. The immune complex was washed with the lysis buffer 6 three times. Immunoprecipitates were subjected to SDS-PAGE on a 7.5% gel. The gels 7 were subjected to silver staining using the Silver Stain MS Kit (Wako, Japan) or 8 western blotting. 9 The transfected 293T cells were treated with lysis buffer (20 mM HEPES [pH 7.5], 0.1% 10 NP-40, 150 mM NaCl, 2.5 mM MgCl2, 1 mM DTT, and protease inhibitor tablet (Roche)). 11 The lysates were precleared with Protein G Dynabeads up to 2 mismatches after removing adaptor sequence, low-quality reads, and reads below length (15 nt) and above length (45nt). The trimmed 25-31 nt length small RNA 1 reads (corresponding to the length of piRNAs) were mapped to repetitive DNA 2 consensus sequence using Dfam database allowing up to 3 mismatches after mapping 3 the mouse genome allowing 0 mismatches (UCSC / mm10) and mapped to L1A sequence 4 (M13002), L1TF sequence (D84391), and IAP sequence (M17551) allowing up to 2 5 mismatches. The number of mapped read counts was normalized by the read-depth of 6 each library.     Quantitative RT-PCR for the expression analysis of piRNA precursors transcribed from 10 embryonic piRNA clusters using E16.5 wild type and MORC3-cKO embryonic testes 11 (bottom). Data is normalized by -actin and is shown as means and SD (Error bar) from 12 more than triplicate PCR reactions. Significant differences ( * p < 0.05 by the t test) 13 between wild type and MORC3-cKO data using primer sets for the Chr 7 (1)