A nowhere-to-hide mechanism ensures complete piRNA-directed DNA methylation

Abstract

The mouse PIWI-interacting RNA (piRNA) pathway provides sustained anti-transposon immunity to the developing male germline by directing transposon DNA methylation^1-3. The first step in this process is the recruitment of SPOCD1 to young LINE1 loci⁴. Thereafter, piRNA-mediated tethering of the PIWI protein MIWI2 (also known as PIWIL4) to the nascent transposon transcript recruits the DNA methylation machinery^5,6. The piRNA pathway needs to methylate all active transposon copies but how this is achieved remains unknown. Here we show that nuclear piRNA and de novo methylation factors are all euchromatic, exposing constitutive heterochromatin as a genomic blind spot for the piRNA pathway. We discover a 'nowhere-to-hide' mechanism that enables piRNA pathway-mediated LINE1 surveillance of the entire genome. We find that SPOCD1 directly interacts with the nuclear pore component TPR, which forms heterochromatin exclusion zones adjacent to nuclear pores⁷. In fetal gonocytes undergoing piRNA-directed DNA methylation, TPR is found both at the nuclear periphery and throughout the nucleoplasm. We find that the SPOCD1-TPR interaction is required for complete non-stochastic piRNA-directed LINE1 methylation. The loss of the SPOCD1-TPR interaction results in a fraction of SPOCD1 and other chromatin-bound piRNA factors relocalizing to constitutive heterochromatin where they are no longer accessible to MIWI2 and the de novo methylation machinery. In summary, the piRNA pathway has co-opted TPR to guarantee that LINE1s are accessible to the piRNA and de novo methylation machineries.

Extended Data Figure 1. Nuclear piRNA factors and the de novo methylation machinery are euchromatic

a, Representative E16.5 gonocyte, somatic cell and corresponding seminiferous cord stained for DAPI, HP1B (red) and H3K9me3 (green, top row) or HP1A (green, bottom row) in foetal testis sections from wild-type mice. Scale bars, 2 μm (top row), 10 μm (bottom row). White rectangles on images of seminiferous cords highlight cells shown zoomed-in. b-e, Representative E16.5 seminiferous cord stained for HA (green), HP1B (red, top row) or H3K4me3 (red, bottom row) and DAPI (blue) in foetal testis sections from Spocd1^HA/+ (b), Miwi2^HA/+ (c), Dnmt3l^HA/HA (d) and Dnmt3c^HA/HA (e) mice. f-g, Representative E16.5 seminiferous cord stained for C19ORF84 (f) or SPIN1 (g) (both green), HP1B (red, top row) or H3K4me3 (red, bottom row) and DAPI (blue) in foetal testis sections from wild-type mice. Also shown in b-g are the channel intensity profiles (RFU: Relative Fluorescence Units) along the indicated line passing through major heterochromatin domains (for HP1B) or avoiding nucleoli (for H3K4me3) for the representative gonocytes presented in Fig. 1. Images shown in both rows of f are from the same quadruple stain (DAPI, C19ORF84, H3K4me3 and HP1B) of a seminiferous cord. Images shown in both rows of g are from another quadruple stain (DAPI, SPIN1, H3K4me3 and HP1B) of a seminiferous cord. Images in a through g are representative of n = 4 (b, c, e) or n = 3 (a, d, f, g) biological replicates; scale bars, 10 μm for seminiferous cords and 2 μm for single gonocytes. White rectangles highlight cells shown in Fig. 1 and in the profile plots.

Extended Data Figure 2. piRNA pathway factors are euchromatic from initiation of expression.

a-c, Representative seminiferous cords stained for HA (SPOCD1-HA) (a), MIWI2 (b) and DNMT3L (c), H3K4me3 and DAPI from Spocd1^HA/+ (a) or wild-type mice (b-c) at the indicated developmental stages. Shown to the right are boxplots of Pearson’s correlation coefficients of SPOCD1 (a), MIWI2 (b) and DNMT3L (c) with H3K4me3 in gonocyte nuclei at the indicated developmental stages. MIWI2 and DNMT3L are not expressed at E14.5, so co-localisation was not calculated for this developmental stage. Scale bars, 10 μm. Images in a through c are representative of n = 3 biological replicates. Box indicates interquartile range, central line represents the mean, and whiskers extend to median ± 1.5× the interquartile range or data limits, whichever is smaller.

Extended Data Figure 3. TPR is highly expressed in male foetal germ cells undergoing de novo genome methylation

a, Volcano plot showing enrichment (log₂(mean LFQ ratio of anti-HA immunoprecipitates from Miwi2^HA/HA/wild-type E16.5 foetal testis lysates) and statistical confidence of proteins co-purifying with HA-MIWI2 (n = 4 with 50 testes per replicate, from previously published data ). b, Volcano plot showing enrichment (log₂(mean LFQ ratio of anti-C19ORF84 immunoprecipitated/anti-rabbit serum immunoprecipitated from wild-type E16.5 foetal testis lysates) and statistical confidence of proteins co-purifying with C19ORF84 (n = 3 with 25 testes per replicate, from previously published data ). c, Single representative E16.5 germ and somatic cells stained for TPR (red) and DAPI (blue) in Spocd1^HA/+ foetal testis sections. Scale bars, 2 μm. d, Representative E16.5 seminiferous cord stained for HA (green), TPR (red) and DAPI in foetal testis sections from Spocd1^HA/+ mice. White rectangles highlight cells shown in Fig. 2f and Extended Data Fig. 3c. Scale bars, 10 μm. e, Representative gonocytes stained for TPR (red) and DAPI (blue) from wild-type mice at the indicated developmental stages. Scale bars, 2 μm. e, Representative seminiferous cords stained for TPR (red) and DAPI (blue) from wild-type mice at the indicated developmental stages. White rectangles highlight cells shown in Extended Data Fig. 3e. Scale bars, 10 μm. Images in c through f are representative of n = 3 biological replicates.

Extended Data Figure 4. SPOCD1-K464A does not interact with TPR

a, Multiple sequence alignment of the SPOCD1 TFIIS-M domains from the indicated species. Red rectangles highlight lysine residues making direct crosslinks with TPR-M in the CL-MS data. Sequence identity conservation is shown by depth of colour. b, Representative Coomassie-stained gel image of n = 3 MBP pull-down experiments with the indicated recombinant mouse SPOCD1 and human TPR fragments. For uncropped source gel image, see Supplementary Fig. 1d.

Extended Data Figure 5. The Spocd1^K464A mouse allele

a, Schematic representations of the mouse Spocd1 locus and encoded 1015 amino acid protein are shown, along with a schematic of the CRISPR targeting strategy showing the location of single-stranded oligo DNA donor (ssODN) and homology arms (HA) used. The sgRNA used for generation of the Spocd1^K464A allele (red), adjacent PAM sites (red) and the primers used for genotyping (blue) are indicated, along with a representative sequencing trace of Spocd1^K464A exon 7 harbouring the 3bp mutation encoding the K464A mutation (highlighted in red). Sequencing was performed on n = 3 F₁ animals. b, Representative image of PCR genotyping result for Spocd1^+/+, Spocd1^+/K464A and Spocd1^K464A mice. PCR genotyping was performed for over 600 mice with similar results. c, Volcano plot showing enrichment (LFQ ratio of anti-SPOCD1 immunoprecipitates from wild-type/Spocd1^K464A E16.5 foetal testis lysates) and statistical confidence (−log₁₀(P-value of two-sided Student’s t-test)) of proteins co-purifying with wild-type SPOCD1 (right quadrant) or SPOCD1-K464A (left quadrant) (n = 3 with 24 foetal testes per replicate per genotype). All identified proteins meeting the enrichment cut-off are listed in Supplementary Table 2. d, Representative adult testis sections of n = 3 wild type, Spocd1^K464A and Spocd1^-/- stained in blue for DAPI and red for the DNA damage marker γH2AX (top) or apoptotic cells by terminal deoxynucleotidyl transferase dUTP nick end labelling (TUNEL) assay (bottom). Scale bars, 50 μm.

Extended Data Figure 6. The SPOCD1-TPR interaction ensures that SPOCD1-SPIN1-C19ORF84 avoids heterochromatin

a-c, e-g, Representative E16.5 seminiferous cord stained for DAPI, HP1B (red) and SPOCD1 (a), SPIN1 (b), C19ORF84 (c), MIWI2 (e), DNMT3L (f), or HA (HA-DNMT3C) (g) (all green) stains in wild-type (top row) or Spocd1^K464A (bottom row) foetal testis sections from littermates. Images are representative of n = 8 (a), n = 4 (c) or n = 3 (b, d-f) biological replicates of each genotype. White rectangles highlight cells shown zoomed-in in Fig. 4a-f. d, Representative E16.5 gonocyte (top two rows) and seminiferous cords (bottom two rows) stained for DAPI (blue), H3K4me3 (red or green), SPOCD1 (green) and HP1B (red) stains in wild-type (first and third row) or Spocd1^K464A (second and last row) foetal testis sections from littermates. Images are representative of n = 4 biological replicates of each genotype. White rectangles in the last two rows highlight cells shown in the first two rows. Also shown below are channel intensity profiles (RFU: Relative Fluorescence Units) along the indicated line passing through major heterochromatic domains in the representative gonocyte. Shown to the right are boxplots of Pearson’s correlation coefficients of SPOCD1 with H3K4me3 in gonocyte nuclei of the indicated genotypes. Box indicates interquartile range, central line represents the mean, and whiskers extend to median ± 1.5× the interquartile range or data limits, whichever is smaller. P-values are from unpaired, two-tailed Student’s T-tests. h, Representative E16.5 seminiferous cord stained for DAPI, SPOCD1 and LINE1 DNA in foetal testis sections of the indicated genotypes. Images are representative of n = 6 biological replicates of each genotype. White rectangles highlight cells shown in Fig. 4g. a-h, Scale bars, 10 μm for seminiferous cords and 2 μm for single gonocytes.

Extended Data Figure 7. The SPOCD1-TPR interaction is necessary for Rasgrf1 methylation

a, Heatmap presentation of mean CpG methylation of imprinted loci in P14 undifferentiated spermatogonia of the indicated genotypes. The imprint control region (ICR) of Rasgrf1 is shown in detail on the bottom. b, Cartoon representation of piRNA-directed epigenetic silencing of LINE1 elements in mice. TPR ensures SPOCD1-targeted LINE1 loci are present in euchromatin and accessible to the piRNA-directed DNA methylation machinery.

Figure 1. Nuclear piRNA factors and the de novo methylation machinery are euchromatic

a-d, Representative E16.5 gonocyte stained for Haemagglutinin epitope tag (HA, green), HP1B (red, top row) or H3K4me3 (red, bottom row) and DAPI in foetal testis sections from Spocd1^HA/+ (a), Miwi2^HA/+ (b), Dnmt3l^HA/HA (c) and Dnmt3c^HA/HA (d) mice are shown. Shown to the right are boxplots of Pearson’s correlation coefficients of SPOCD1 (a), MIWI2 (b), DNMT3L (c) and DNMT3C (d) with HP1B and H3K4me3 in gonocyte nuclei. e-f, Representative E16.5 gonocyte stained for C19ORF84 (e) or SPIN1 (f) (both green), HP1B (red, top row) or H3K4me3 (red, bottom row) and DAPI in foetal testis sections from wild-type mice. Shown to the right are boxplots of Pearson’s correlation coefficients of C19ORF84 (e) and SPIN1 (f) with HP1B and H3K4me3 in gonocyte nuclei. Images in a through f are representative of n = 4 (a, b, d) or n = 3 (c, e, f) biological replicates; scale bars, 2 μm. Box indicates interquartile range, central line represents the mean, and whiskers extend to median ± 1.5× the interquartile range or data limits, whichever is smaller.

Figure 2. SPOCD1 directly interacts with TPR

a, Volcano plot showing enrichment (log₂(mean label-free quantification (LFQ) ratio of anti-HA immunoprecipitates from Spocd1^HA/HA/wild-type E16.5 foetal testis lysates) and statistical confidence (−log₁₀(P-value of two-sided Student’s t-test)) of proteins co-purifying with SPOCD1-HA (n = 4 with 50 testes per replicate, previously published data from ). b, Volcano plot showing enrichment (log₂(mean LFQ ratio of anti-TPR immunoprecipitated/anti-rabbit serum immunoprecipitated from wild-type E16.5 foetal testis lysates) and statistical confidence of proteins co-purifying with TPR (n = 3 with 25 foetal testes per replicate). All identified proteins meeting the enrichment cut-off are listed in Supplementary Table 1. Grey dots indicate IgG fragments. c, Volcano plots showing enrichment (log₂(mean LFQ ratio of anti-HA immunoprecipitates from lysates of mouse embryonic stem cells (mESCs) ectopically expressing SPOCD1-cHA or the indicated domain deletion mutants/wild-type mESCs) and statistical confidence of proteins co-purifying with SPOCD1-HA (left), SPOCD1-ΔTFIISM-HA (centre) and SPOCD1-ΔSPOC-HA (right) (n = 3 independently derived SPOCD1-HA-expressing mESCs). d, Representative western blot image for n = 3 pull-down experiments with indicated mouse SPOCD1-TFIISM and lysates of HEK cells expressing the indicated human TPR fragments. For uncropped source gel image, see Supplementary Fig. 1a. e, Crosslinking mass spectrometry (CL-MS) of SPOCD1-TFIISM and recombinant human TPR-M. Inter-protein crosslinks and cross-linked lysine residues are shown in green. Grey regions indicate helical secondary structure, while blue regions indicate β-strands. More details about identified crosslinks are included in Supplementary Table 3. f, Representative E16.5 gonocyte stained for HA (SPOCD1-HA) (green), TPR (red) and DAPI in foetal testis sections from Spocd1^HA/+ mice. Images are representative of n = 3 biological replicates. Scale bars, 2 μm.

Figure 3. Defective de novo LINE1 methylation in Spocd1^K464A mice

a, ConSurf conservation scores (left) and PyMOL predicted surface charge (right) of the mouse SPOCD1 TFIISM domain (amino acids 407-568, UniProt ID B1ASB6, AlphaFold 2 predicted structure). b, Representative Coomassie Blue-stained gel of n = 3 MBP pull-down experiments with the indicated recombinant mouse SPOCD1 and human TPR fragments. For uncropped source gel image, see Supplementary Fig. 1c. c, Representative western blot of n = 3 E16.5 foetal testis samples of the indicated genotypes indicating SPOCD1, MIWI2 and TPR protein levels in a single foetal testis. Alpha-tubulin served as loading control. For uncropped whole blot source images, see Supplementary Fig. 1b. d, Number of pups born per plug (mean ± SEM) fathered by n = 3 wild-type (6 plugs total) and n = 4 Spocd1^K464A studs (7 plugs total) mated to wild-type CD-1 females. e, Testis weight (mean ± SEM) of n = 8 wild-type and n = 16 Spocd1^K464A 12 week-old mice. Insert shows a representative image of testes from wild-type (left) and Spocd1^K464A (right) mice. Also shown are representative haematoxylin and periodic acid-Schiff (H&PAS)-stained epididymal tubules of n = 3 wild-type (top) and n = 4 Spocd1^K464A (bottom) mice. Scale bars, 20 μm. P-values in d and e are from unpaired two-tailed Student’s T-tests. f, Representative images of H&PAS-stained testes sections of wild-type and n = 3 Spocd1^K464A adult mice between 9 and 12 weeks of age, with different types of spermatogenic arrest observed in tubules of Spocd1^K464A testes. Scale bars, 20 μm. g, Representative adult testis sections of n = 3 wild type and Spocd1^K464A mice stained for DAPI (blue) and LINE1 ORF1p (red, left) or IAP GAG protein (red, right). Scale bars, 50 μm. h-j, Genomic CpG methylation analysis of P14 undifferentiated spermatogonia from wild-type (n = 4) and Spocd1^K464A (n = 3) mice. h, Percentages of CpG methylation levels of the indicated genomic features (with genic, promoter and CpG island (CGI) regions defined as those not overlapping transposable elements, and intergenic regions as those not overlapping transposable elements or genes) or transposable elements (not overlapping genes) are shown as box plots. Boxes represent interquartile range from 25^th to 75^th percentile, the horizontal line the median, whiskers the data range of the median ± 2× interquartile range. Significant differences (P-values < 0.05 of Benjamini-Hochberg-corrected unpaired two-tailed Student’s T-tests) to wild-type are indicated. i, Metaplots of mean CpG methylation over the consensus sequence of the indicated transposable elements and the adjacent 2 kb, comparing wild-type (n = 4), Spocd1^K464A, Spocd1^-/-, C19orf84^-/- and Miwi2^-/- P14 spermatogonia (all n = 3). P-values of Benjamini-Hochberg-corrected unpaired two-tailed Student’s T-tests comparing average CpG methylation of the promoter region (first 25 bins) to wild-type for Spocd1^K464A. j, Mean CpG methylation loss relative to wild-type for individual transposons of the indicated families correlated with divergence from their consensus sequence in Spocd1^K464A spermatogonia.

Figure 4. The SPOCD1-TPR interaction ensures LINE1 and Rasgrf1 accessibility to MIWI2 and the de novo methylation machinery.

a-f, Representative E16.5 gonocyte stained for DAPI, HP1B (red) and SPOCD1 (a), SPIN1 (b), C19ORF84 (c), MIWI2 (d), DNMT3L (e), or HA (HA-DNMT3C) (f) (all green) in wild-type (top row) or Spocd1^K464A (bottom row) foetal testis sections from littermates (n = 8 (a), n = 4 (c) or n = 3 (b, d-f) for each genotype). Shown to the right in a-c is the fraction of gonocytes with foci (mean ± SEM) for the stained proteins in sections from the indicated genotypes. Shown to the right in a-f are boxplots of Pearson’s correlation coefficients of the indicated proteins with HP1B in all wild-type and foci-containing Spocd1^K464A gonocyte nuclei (a-c), or all wild-type and Spocd1^K464A gonocyte nuclei (d-f). Scale bars, 2 μm. g, Representative E16.5 gonocyte (1^st and 3^rd rows) and zoomed-in view of region of interest (white rectangle on gonocyte nucleus) (2^nd and 4^th rows) stained for DAPI, SPOCD1 and LINE1 DNA in foetal testis sections of the indicated genotypes (n = 6 for each genotype). Shown to the right is the fraction (mean ± SD) of SPOCD1-K464A foci overlapping LINE1 DNA, as well as boxplots of Pearson’s correlation coefficients of SPOCD1 with LINE1 DNA in all wild-type and foci-containing Spocd1^K464A gonocyte nuclei. a-g, Box extends from 25^th to 75^th percentile, central line indicates the mean, whiskers extend between data limits or 1.5× the interquartile range, whichever is smaller. Scale bars, 2 μm. h-i, Representative E16.5 gonocyte with DAPI (greyscale) and DNA FISH for the Rasgrf1 locus on Chr. 9 (h), or a control locus on Chr. 9 (i) (both red) in wild-type and Spocd1^K464A E16.5 foetal testis sections. An example of a heterochromatin- and an euchromatin- overlapping FISH signal in gonocytes of each genotype are shown as indicated (n = 3 for each genotype). Shown to the right is the fraction (mean ± SEM) of heterochromatin-overlapping signals in gonocytes in sections from the indicated genotypes. Scale bars, 2 μm. j, Median CpG methylation levels (%) (mean ± SEM) of the indicated LINE1 families in n = 3 biological replicates from the indicated genotypes. P-values are from unpaired two-tailed Student’s T-tests. k, Variance in CpG methylation levels (%²) of LINE1 loci across all replicates of the indicated genotypes (n = 3 for each genotype), summed over all loci within the indicated LINE1 families. P-values in a-g and h-j are from unpaired, two-tailed Student’s T-tests.