Graphical abstract

   graphic file with name fx1.jpg
   [49]Open in a new tab

Highlights

     * •
       NANOS2 is a 3′UTR mRNA binding protein and recognizes the AUKAAWU
       consensus
     * •
       CCR4-NOT deadenylase complex is the primary interactor of NANOS2 in
       spermatogonia
     * •
       NANOS2 binding results in mRNA destabilization
     * •
       NANOS2 regulates key spermatogonial signaling and metabolic
       pathways
     __________________________________________________________________

   Biological sciences; Molecular biology; Developmental biology; Omics;
   Transcriptomics

Introduction

   The maintenance of spermatogonial stem cells (SSCs) is essential to
   sustain life-long spermatogenesis and adult male fertility. In mouse,
   this population of stem cells resides within undifferentiated type A
   spermatogonia ([50]Huckins, 1971; [51]Oakberg, 1971). Undifferentiated
   spermatogonia comprise A[single] (A[s] or isolated cells) and A[paired]
   (A[pr], a pair of connected cells) or A[aligned] (A[al], chains of 4,
   8, or 16 connected cells) cells that arise due to incomplete
   cytokinesis and remain attached by intercellular bridges. The
   glial-cell-line-derived neurotrophic factor (GDNF) receptor Gfrα1 marks
   a subset of A[s], A[pr], and A[al4] that contains SSCs ([52]Hara
   et al., 2014; [53]Nakagawa et al., 2010). Gfrα1-positive A[s] are
   heterogeneous and include a subset of cells expressing ID4 and PAX7,
   which have SSC activity ([54]Aloisio et al., 2014; [55]Chan et al.,
   2014; [56]Helsel et al., 2017; [57]Sun et al., 2015). Upon Gfrα1
   downregulation, SSCs give rise to an intermediate population of
   spermatogenic precursors, marked by Ngn3, which are responsive to
   retinoic acid (RA) signaling and thus can differentiate into
   c-Kit-positive cells ([58]Ikami et al., 2015; [59]Nakagawa et al.,
   2010). Mechanistically, RA instructs spermatogonia differentiation
   through the activation of the PI3K/AKT/mTORC1 signaling pathway, which
   in turn enhances the translation of differentiation-related genes such
   as c-Kit ([60]Busada et al., 2014, [61]2015a, [62]2015a; [63]2015a).
   SSC self-renewal is extrinsically dependent upon growth factors, such
   as GDNF and fibroblast growth factor (FGF) ([64]Kitadate et al., 2019;
   [65]Meng et al., 2000). Furthermore, GDNF, FGF2, epidermal growth
   factor (EGF), and leukemia inhibitory factor (LIF) are key to promote
   SSC maintenance in long-term cultures ([66]Kanatsu-Shinohara et al.,
   2003; [67]Kubota et al., 2004; [68]Takashima and Shinohara, 2018).
   Downstream signaling cascades activated by these growth factors, such
   as phosphatidylinositol 3-kinase (PI3K)-Akt and ERK/MAPK pathways are
   also essential for SSC self-renewal ([69]Hasegawa et al., 2013;
   [70]Ishii et al., 2012; [71]Lee et al., 2007). The signaling cascades
   activated by the aforementioned growth factors input into the mTOR
   pathway ([72]Laplante and Sabatini, 2012; [73]Liu and Sabatini, 2020;
   [74]Yu and Cui, 2016). Tight control of mTOR complex 1 (mTORC1)
   activity is critical in regulating the balance between self-renewal and
   differentiation in SSCs ([75]Busada et al., 2015b; [76]Hobbs et al.,
   2010; [77]Serra et al., 2017), as shown by the progressive
   spermatogenic failure associated with the alteration of the mTORC1
   component RAPTOR or the regulatory TSC1/2 complex ([78]Hobbs et al.,
   2015; [79]Serra et al., 2019; [80]Wang et al., 2016).

   The RNA-binding protein (RBP) NANOS2 is mostly expressed in A[s] and
   A[pr] spermatogonia ([81]Suzuki et al., 2009) and is required for the
   self-renewal of SSCs ([82]Sada et al., 2009). Conditional ablation of
   Nanos2 in adult testis results in the rapid depletion of
   undifferentiated type A spermatogonia and the progressive loss of
   spermatogenesis, whereas the overexpression of Nanos2 results in the
   accumulation of undifferentiated type A spermatogonia ([83]Sada et al.,
   2009). Nanos2 is also required for the survival of male mouse gonocytes
   during embryonic development ([84]Tsuda et al., 2003), where one
   function of NANOS2 is to supress aberrant entry into meiosis
   ([85]Suzuki and Saga, 2008). In summary, NANOS2 is a key intrinsic
   regulator of the male germline.

   Nanos2 encodes a protein of 136 amino acids that contains two CCHC type
   zinc fingers and belongs to the family of NANOS RBPs with evolutionary
   conserved functions in the germline ([86]De Keuckelaere et al., 2018;
   [87]Tsuda et al., 2003). In mouse male embryonic gonocytes, NANOS2
   interacts with the CCR4-NOT deadenylation complex and may promote
   transcript degradation in P-bodies ([88]Suzuki et al., 2010). Moreover,
   CNOT1-NANOS2 interaction is essential for NANOS2 function during
   embryonic development ([89]Suzuki et al., 2012). In postnatal SSCs,
   NANOS2 has been linked to the translational repression of several
   transcripts encoding proteins associated with SSC differentiation
   ([90]Zhou et al., 2015b). NANOS2 via protein-protein interaction also
   sequesters mTOR in cytoplasmic messenger ribonucleoproteins to limit
   its activity ([91]Zhou et al., 2015b). However, stringent biochemical
   approaches have not been employed yet to identify the full complement
   of bona fide NANOS2 RNA targets in SSCs, raising the possibility that
   additional and overlooked molecular mechanisms can underlie NANOS2
   function in mouse SSCs.

Results

Generation and validation of the epitope-tagged Nanos2^TAG allele

   To explore the function of NANOS2 in SSCs, we generated an endogenously
   tagged allele of Nanos2 (Nanos2^TAG) in the mouse. We inserted a
   complex tag that consisted of the V5 tag, the Myc tag, the precision
   protease site, the His tag, and the enhanced green fluorescent protein
   (EGFP) that generates an N-terminal NANOS2 fusion protein ([92]Figures
   1A and [93]S1). This versatile tag enables many applications including
   immunofluorescence and immunoprecipitation approaches. Homozygous
   Nanos2^TAG mice were fertile and presented normal testicular weight as
   well as seminiferous tubule histology ([94]Figures 1B–1D). Despite the
   size of the tag, the lack of phenotype in Nanos2^TAG mice demonstrates
   the functionality of the epitope-tagged allele. This genetic
   complementation is in stark contrast to the impact of Nanos2
   deficiency. Nanos2 is essential for both the development of the male
   gonocytes and self-renewal of SSCs ([95]Sada et al., 2009; [96]Tsuda
   et al., 2003). In the adult testis Nanos2 expression is restricted to
   undifferentiated spermatogonia, many of which are GFRα1-positive
   ([97]Sada et al., 2009; [98]Suzuki et al., 2009). We used anti-GFP
   antibodies to examine the expression of TAG-NANOS2 in whole-mount
   preparations of seminiferous tubules, which confirmed the restricted
   cellular expression profile. TAG-NANOS2 expression was restricted to
   undifferentiated spermatogonia, most of which were Gfrα1 positive
   ([99]Figures 1E and 1F). Indeed, all Gfrα1-positive spermatogonia
   expressed TAG-NANOS2 ([100]Figure 1F), as was previously reported for
   endogenous NANOS2 ([101]Suzuki et al., 2009). PLZF staining that marks
   a wide population of progenitor spermatogonia from A[s] to very long
   chains revealed that TAG-NANOS2 expression is restricted predominantly
   to A[s] and A[pr] rather than longer spermatogonial chains
   ([102]Figure S2). Importantly, TAG-NANOS2 was not identified in
   c-Kit-positive differentiating spermatogonia ([103]Figure 1G). Similar
   to what has been reported for endogenous NANOS2 ([104]Sada et al.,
   2009; [105]Suzuki et al., 2009; [106]Zhou et al., 2015b), TAG-NANOS2 is
   localized to the cytoplasm and also found in P-bodies ([107]Figure 1H).
   In summary, we conclude that the Nanos2^TAG allele is functional and
   faithfully recapitulates the reported expression of Nanos2.

Figure 1.

   [108]Figure 1
   [109]Open in a new tab

   The Nanos2^TAG mouse allele is functional

   (A) Schematic representation of the NANOS2 protein and the TAG-NANOS2
   fusion protein, with corresponding lengths of amino acid sequences
   (aa).

   (B) Number of litters per animal from Nanos2^CTL(Nanos2^+/+) and
   Nanos2^TAG (Nanos2^TAG/TAG) mice. Data are mean and s.d., n = 6 for
   both genotypes. n.s. indicates not significant, p value > 0.05 using
   two-tailed Student's t-test.

   (C) Testicular weight of Nanos2^CTL and Nanos2^TAG from six-months-old
   mice. Data are mean and s.d., n = 5 for both genotypes. Statistical
   evaluation as in panel b.

   (D) Representative testis cross-sections stained with hematoxylin and
   eosin (H&E) from six-months-old Nanos2^CTL and Nanos2^TAG mice. Scale
   bar, 100 μm.

   (E) Representative immunofluorescent images of Nanos2^TAG seminiferous
   tubules stained with anti-GFP (green) and anti-GFRα1 (red) antibodies.
   Representative examples of GFRα1^Pos, NANOS2^Pos cells (A[s] and A[pr])
   are highlighted (white boxes). Scale bar, 50 μm.

   (F) Enumeration of GFRα1^Pos NANOS2^Pos and GFRα1^Neg NANOS2^Pos
   testicular populations. The number of cells (N) present in each cluster
   is normalized to 1,000 Sertoli cells (y axis). Error bars represent SEM
   (standard error of mean).

   (G) Representative immunofluorescent images of Nanos2^TAG seminiferous
   tubules stained with anti-GFP (green) and anti-c-KIT (red) antibodies.
   Representative example of NANOS2^Pos, c-KIT^Neg cell (A[s]) is
   highlighted (white box). Scale bar, 50 μm.

   (H) Representative immunofluorescent image of an A[s] cell from
   Nanos2^TAG seminiferous tubules stained with with DAPI (blue), anti-GFP
   antibody (green) and anti-DCP1a (red). Scale bar, 5 μm.

   (I) Representative bright field images of Nanos2^CTL and Nanos2^TAG SSC
   lines cultured on mouse embryonic fibroblasts (MEF) feeders.

   (J) Western blot using anti-GFP and anti-Actin antibodies on Nanos2^CTL
   and Nanos2^TAG SSC lines lysates.

   (K) Flow cytometry analysis of Nanos2^CTL and Nanos2^TAG SSC lines. The
   GFP intensity of the Nanos2^TAG population is represented in a
   single-parameter histogram, with the overlay of the Nanos2^CTL control.

   The transcriptome occupancy or the complement of transcripts bound by
   NANOS2 as well as a possible consensus binding site for NANOS2 remains
   uncharacterized by stringent cross-linking immunoprecipitation
   techniques coupled to high-throughput sequencing (CLIP-seq) approaches.
   RNA immunoprecipitation (RIP) of NANOS2 coupled with microarray
   hybridization from P7 testis ([110]Zhou et al., 2015b) or from E14.5
   fetal testis ([111]Saba et al., 2014) had previously been used to
   define NANOS2 target transcripts. However, RIP is the least stringent
   method for target identification because the RNA is not cross-linked to
   the RBP and low-stringency washes can only be applied. In the case of
   NANOS2, this method is more problematic given that a portion of NANOS2
   resides in P-bodies ([112]Suzuki et al., 2010; [113]Zhou et al.,
   2015b), and as such, NANOS2 RIP may readout many RNAs found therein.
   Furthermore, because cross-linking is not applied prior to the lysis of
   cells, RIP-based approaches are prone to artifacts due to the mixing of
   cellular compartments and the reassembly of new complexes with
   nonphysiological targets ([114]Lee and Ule, 2018; [115]Mili and Steitz,
   2004). One limitation of CLIP-seq is the requirement of large amounts
   of input material due to the low efficient UV cross-linking of RNA to
   RBPs ([116]Ramanathan et al., 2019). This constraint precludes the
   application of CLIP-seq from ex vivo isolated NANOS2-expressing
   spermatogonia. Fortunately, SSC lines derived from neonatal testis
   ([117]Kanatsu-Shinohara et al., 2003) can be expanded in culture for
   prolonged periods of time, retaining their spermatogonial identity
   ([118]Kanatsu-Shinohara et al., 2003). These cultures display
   functional heterogeneity as only a small fraction of these cells have
   the ability to produce colonies in recipient mouse testis in
   transplantation experiments ([119]Kanatsu-Shinohara and Shinohara,
   2013). They can also display heterogeneity in terms of different
   spermatogonia marker expression ([120]Kanatsu-Shinohara et al., 2003).
   Nonetheless SSC cultures represent a valuable surrogate system for the
   study of SSCs and spermatogonia in vitro ([121]Kanatsu-Shinohara and
   Shinohara, 2013). Thus, we derived control Nanos2 wild type
   (Nanos2^CTL) and experimental Nanos2^TAG/TAG or Nanos2^TAG/+
   (Nanos2^TAG) SSC lines from postnatal testis ([122]Figure 1I). All cell
   lines, independent of genotype, displayed the expected morphology
   ([123]Figure 1I). TAG-NANOS2 could be detected by Western blotting from
   the SSC lines ([124]Figure 1J), and more importantly, the Nanos2^TAG
   cell lines uniformly expressed TAG-NANOS2 as determined by FACS
   analysis using the fused EGFP ([125]Figure 1K).

CRAC reveals NANOS2 transcriptome occupancy in SSC lines

   To gain insight into the mechanism by which NANOS2 maintains SSC
   self-renewal, we aimed to define NANOS2 transcriptome occupancy in SSCs
   by using cross-linking and analysis of cDNAs (CRAC) ([126]Granneman
   et al., 2009). CRAC is one of the most stringent methods to identify
   high confident RNA-protein interactions due to the fact that only RBPs
   and RNAs at zero distance are cross-linked and that the
   cross-linked-RBP-RNA complexes are purified in multiple steps, with the
   last two being under denaturing conditions ([127]Granneman et al.,
   2009; [128]Lee and Ule, 2018; [129]Ramanathan et al., 2019)
   ([130]Figure S3). The Nanos2^TAG SSC lines permit the use of CRAC
   thanks to the presence of the V5 epitope tag, the PreScission protease
   recognition site and the His6 tag within TAG-NANOS2 ([131]Granneman
   et al., 2009). After TAG-NANOS2 was UV-C cross-linked to RNAs in SSCs,
   TAG-NANOS2-RNA complexes were sequentially purified and analyzed by
   autoradiography. Coprecipitated RNA was recovered from Nanos2^TAG but
   not Nanos2^WT SSC lines ([132]Figure 2A). RNA was isolated and
   converted into cDNA to generate libraries for high-throughput
   sequencing. After mapping sequencing reads to the mouse transcriptome,
   we analyzed the similarities among replicates and samples by
   calculating the Spearman's correlation coefficient. Nanos2^TAG
   replicates highly correlated with each other, while no correlation was
   observed between Nanos2^TAG and Nanos2^CTL ([133]Figure S4A),
   indicating the specificity of the CRAC experiments. CRAC analysis
   revealed that NANOS2 predominantly bound (>88%) to mRNA
   ([134]Figure 2B; [135]Table S1), generally with one binding site per
   transcript ([136]Figure 2C). Importantly, the vast majority of
   NANOS2-biding sites (81%) were located in the 3′ UTRs of transcripts
   ([137]Figures 2D and 2E). For the transcripts with multiple NANOS2
   binding sites, the length of the 3′UTR was not a determinant of the
   number of binding sites ([138]Figure S4B).

Figure 2.

   [139]Figure 2
   [140]Open in a new tab

   CRAC reveals NANOS2 occupancy, consensus binding sequence and
   preference for mRNA 3′ UTR residency in SSC lines

   (A) Autoradiograph showing the TAG-NANOS2 protein purified from
   homozygous Nanos2^TAG SSC lines by SDS-PAGE and cross-linked to
   radioactively labeled RNA. Dashed boxes indicate the area excised for
   the elution of RNA-protein complexes.

   (B) Transcriptomic distribution of mapped CRAC reads among different
   classes of RNAs as indicated is shown.

   (C) The number of NANOS2 peaks in individual mRNAs identified within
   sequencing reads is shown.

   (D)Metagene profile of NANOS2 CRAC peaks along the 5′ UTR, CDS, and 3′
   UTR of its targets. The average length of these regions is shown.

   (E) Percentages of NANOS2 peaks within 5′ UTR, CDS, and 3′ UTR are
   shown.

   (F) Logo representation of the NANOS2-binding site present in the top
   500 NANOS2 targets using reads with deletions. The logo is adapted from
   MEME-ChIP.

   (G) Position of the AUKAAWU motif within sequencing reads (adapted from
   Centrimo). Reads were centered on deletions.

   (H) Sequence alignment of a subset of reads with deletions, from the
   top 500 NANOS2 targets. The AUKAAWU motif is highlighted in red.
   Sequences are centered on deletions, which correspond to the
   cross-linking sites, and the most probable NANOS2 direct binding site
   (central ‘U’, in bold red).

   The sequence information from CRAC experiments can be used to define
   the binding site of RBPs ([141]Granneman et al., 2009). UV
   cross-linking generally results in covalent binding of a nucleotide
   (most commonly a pyrimidine) to one amino acid residue of the RBP
   ([142]Sugimoto et al., 2012). The sites of amino acid-nucleotide
   cross-linking are problematic for reverse transcriptases. The enzyme
   can terminate at those positions or readthrough by overcoming the
   cross-linked amino acid, potentially incorporating an error or
   microdeletion into the nascent cDNA ([143]Granneman et al., 2009;
   [144]Ule et al., 2003, [145]2005). Among mutations contained in read
   through cDNA, single-nucleotide deletions reflect most accurately the
   cross-linking site, since these are very rarely generated through PCR
   or sequencing errors ([146]Granneman et al., 2009; [147]Zhang and
   Darnell, 2011). Thus, we analyzed CRAC reads with deletions to
   precisely identify NANOS2-binding sites. We first normalized
   CRAC-binding site intensities for transcript expression levels, as
   determined by RNA-seq. This approach is commonly used as CRAC read
   counts are partially dependent on the expression level of the
   respective transcripts present in cells ([148]König et al., 2012).
   Comparing the top 500 normalized CRAC TAG-NANOS2-bound transcripts in
   SSC lines ([149]Table S2), we found a significant overlap with the
   RIP-identified NANOS2 target transcripts from P7 testis and E14.5 fetal
   testis ([150]Zhou et al., 2015b; [151]Saba et al., 2014) ([152]Figures
   S4C and S4D, [153]Tables S3 and [154]S4). MEME-ChIP analysis
   ([155]Machanick and Bailey, 2011) was performed on the top 500 most
   highly enriched transcripts ([156]Table S2) that contained deletions.
   This identified a putative consensus, seven-nucleotide-binding site for
   NANOS2 (AUKAAWU; with K = G or U, W = A or U) ([157]Figure 2F), that
   was significantly enriched in the proximity of deletions (p = 4.4 x
   10^−58) ([158]Figures 2G, 2H, and [159]S4E). In summary, in SSCs,
   NANOS2 predominantly binds mRNAs, generally at a single site in the
   3′UTR, with a preference for the consensus sequence AUKAAWU.

NANOS2 regulates many transcripts important for SSC metabolism and
self-renewal

   Having defined transcripts bound by NANOS2, we next sought to
   understand how this contributes to the maintenance of SSCs. Gene
   ontology analysis of the top 500 NANOS2-bound transcripts revealed very
   significant enrichment for metabolic and biosynthetic processes
   ([160]Figure 3A). We next applied an Ingenuity Pathway Analysis
   ([161]Krämer et al., 2014), which revealed enrichment for seven
   signaling pathways: PI3K/AKT, RAC, Integrin, ERK/MAPK, p70S6K,
   regulation of eiF4 and p70S6K as well as mTOR ([162]Figure 3B). These
   pathway classifications were consistent with the identification of many
   specific NANOS2 target mRNAs implicated in the respective pathways
   ([163]Figure 3C). Importantly, this analysis included pathways whose
   activity and regulation are known to be essential for the maintenance
   of SSCs ([164]Hasegawa et al., 2013; [165]Hobbs et al., 2010;
   [166]Ishii et al., 2012; [167]Lee et al., 2007; [168]Oatley et al.,
   2007) or for spermatogonia proliferation ([169]Feng et al., 2000):
   PI3K/AKT, ERK/MAPK, p70S6K, and mTOR ([170]Figure 3D). Besides its role
   in SSC maintenance ([171]Goertz et al., 2011; [172]Oatley et al.,
   2007), PI3K/AKT is also involved in mediating spermatogonia
   differentiation in cells responsive to RA ([173]Busada et al., 2015b).
   Many of these signaling pathways are directly stimulated by growth
   factors which are also essential for SSCs. SSCs are absolutely
   dependent on GDNF – Gfrα1 signaling ([174]Kanatsu-Shinohara et al.,
   2003; [175]Oatley et al., 2007). Indeed heterozygosity of GDNF results
   in the progressive loss of SSCs whereas its overexpression results in
   expansion of SSCs ([176]Meng et al., 2000). FGF signaling is also
   important for the self-renewal of SSCs as is EGF ([177]Ishii et al.,
   2012; [178]Kanatsu-Shinohara et al., 2003; [179]Kitadate et al., 2019;
   [180]Takashima et al., 2015). Among NANOS2 targets, were mRNAs encoding
   key signaling transduction molecules, as exemplified by the RAS-related
   proteins RAP1A/B, RALB, and RRAS2 and the PIK3CB and PTEN proteins.
   These factors act downstream of the GDNF, FGF, EGF receptor tyrosine
   kinases ([181]Figures 3C–3E) and, importantly, RAS and PTEN are
   essential for SSC self-renewal ([182]Lee et al., 2009; [183]Zhou
   et al., 2015a). These growth factors signaling pathways input into the
   mTOR pathway ([184]Meng et al., 2018), which regulates cell growth and
   metabolism ([185]Laplante and Sabatini, 2012). Moreover, mTOR activity
   supports SSC differentiation at the expenses of self-renewal
   ([186]Busada et al., 2015b; [187]Hobbs et al., 2010; [188]Wang et al.,
   2016). Notably, the mTORC2 component RICTOR is essential for
   spermatogonia differentiation ([189]Bai et al., 2018), whereas the
   mTORC1 component RAPTOR and the mTORC1 negative regulator TSC1/2 are
   essential for the maintenance of the SSC pool ([190]Hobbs et al., 2015;
   [191]Serra et al., 2019; [192]Wang et al., 2016). Interestingly, we
   identified the transcripts for the mTORC2 component RICTOR and mTORC1
   regulators RHEB and LAMTOR3 as NANOS2 targets ([193]Figures 3C–3E).
   Furthermore, many transcripts that encode proteins downstream of mTOR
   in the regulation of protein synthesis are targets of NANOS2, as
   exemplified by PPP2CA, PPP2R2A, RPS6KA5, EIF3E/F, RPS23, and EIF2S1
   ([194]Figures 3D and 3E). In summary, a subset of NANOS2 target
   transcripts encompasses key components of signaling and metabolic
   pathways whose regulated activity is known to be essential for SSC
   self-renewal ([195]Hobbs et al., 2015; [196]Lee et al., 2009;
   [197]Serra et al., 2019; [198]Wang et al., 2016; [199]Zhou et al.,
   2015a).

Figure 3.

   [200]Figure 3
   [201]Open in a new tab

   NANOS2 binds to mRNAs encoding proteins involved in metabolic and
   signaling pathways critical for SSC maintenance

   (A) Gene ontology (GO) analysis performed on the top 500 NANOS2 CRAC
   targets. The significance of enrichment of Biological Processes was
   calculated by Fisher's exact test and is shown as -log[10] p value. The
   dashed line shows the threshold for the significance (p value < 0.05).

   (B) Ingenuity pathways enrichment analysis performed on the top 500
   NANOS2 CRAC targets. The enrichment of signaling pathways is shown in
   the second column (p value < 0.05). The number and percentage of NANOS2
   targets in each pathway are also shown (third column).

   (C) Overlapping NANOS2 targets within the seven signaling pathways are
   shown as indicated.

   (D) The scheme shows some of the molecular relationships within these
   pathways with NANOS2 targets in blue, and factors critical for SSC
   maintenance in green.

   (E) Representative examples of NANOS2 peaks in four mRNA targets
   (Eif2s1, Pik3cb, Lamtor3 and Rheb) encoding proteins involved in the
   signaling pathways described in (B and C).

NANOS2 binds the CCR4-NOT complex and reduces mRNA half-lives in SSCs

   We next sought to understand the mechanism by which NANOS2 regulates
   its target transcripts in SSCs. CRAC revealed that the majority of
   NANOS2 binding occurs within the 3′ UTR of mRNAs. 3′ UTRs are
   frequently bound by protein complexes that can modulate mRNA
   translation, deadenylation, and decay ([202]Mayya and Duchaine, 2019).
   To understand which of these processes were involved in the regulation
   of NANOS2 targets, we aimed to identify NANOS2 interacting proteins in
   SSCs. To this end, we performed immunoprecipitation coupled to mass
   spectrometry (IP-MS) from SSC lines ([203]Figure 4A). Using stringent
   criteria of greater than four times enrichment and significance of p
   value < 0.05, we found 8 factors ([204]Figure 4A, [205]Table S5). Among
   these were 6 subunits of the CCR4-NOT deadenylase complex, including
   the catalytic subunits CNOT7 and CNOT8. NANOS2 interaction with the
   CCR4-NOT complex was independent of RNA, as treatment of the IP with
   RNase A and T1 still resulted in the enrichment of many CCR4-NOT
   subunits ([206]Figure 4B, [207]Table S5). The CCR4-NOT complex is the
   major cytoplasmic deadenylase and mRNA deadenylation constitutes the
   first, rate-limiting step of RNA degradation ([208]Bresson and
   Tollervey, 2018). These results suggest that NANOS2 recruits the
   CCR4-NOT complex to its mRNA targets and stimulate their degradation.
   To test this, we performed SLAM-seq ([209]Herzog et al., 2017) in SSC
   lines, which enables the measurement of mRNA half-life
   transcriptome-wide. SLAM-seq analysis revealed that the average
   transcript half-life is 224–225 min ([210]Figure 4C). Strikingly,
   NANOS2-bound transcripts had a much shorter half-life (180 min)
   relative to transcripts that were not bound by NANOS2 ([211]Figure 4D).
   NANOS2 is essential for SSC maintenance, whereas its overexpression is
   compatible with spermatogonial survival albeit with the loss of their
   ability to differentiate ([212]Sada et al., 2009). Thus, unlike
   Nanos2-deficiency, its overexpression is not required for SSC line
   survival ([213]Zhou et al., 2015b). NANOS2 overexpression in SSC lines
   ([214]Zhou et al., 2015b) resulted in an overall reduction in the
   abundance of NANOS2 target transcripts relative to other mRNAs
   ([215]Figures 4E and 4F). We conclude that in SSCs, NANOS2 recruits the
   CCR4-NOT deadenylase complex to mRNAs, which in turn promotes a
   reduction in transcript half-life.

Figure 4.

   [216]Figure 4
   [217]Open in a new tab

   NANOS2 interacts with CNOT proteins and its binding to mRNA is
   associated with transcript destabilization in SSC lines

   (A and B) Volcano plots showing results from the immunoprecipitation
   (using anti-GFP beads) followed by mass spectrometry analysis from
   Nanos2^CTL and Nanos2^TAG/TAG SSC lines. IPs were treated without (−)
   (A) or with (+) (B) RNase as indicated. Each TAG-NANOS2 interactor is
   shown as a dot and significantly enriched proteins are labeled (except
   for NANOS2 in B). The x axis shows the enrichment of the interactors in
   the IPs (log2(LFQ(Nanos2^TAG)/LFQ(Nanos2^WT); the y axis shows the
   confidence (-log[10](p value of two-sided Student's t-test)). CNOT
   proteins are highlighted in red. Each plot shows results from three
   replicates (N = 3).

   (C) Median decay curves showing the global mRNA stability in SSC lines
   for two replicates. mRNA half-life (t[1/2]) was determined for 5,806
   transcripts by counting the T to C conversion rate in sequencing reads,
   over time (minutes). The shaded areas indicate the first and third
   quantile decay curves range for each replicate. Transcript half-life
   median for each replicate are indicated with horizontal dotted lines
   and are also shown at the panel top.

   (D) Cumulative distribution of mRNA half-life in replicate 1 (rep1)
   (solid lines) and 2 (rep2) (dashed lines) for NANOS2 targets. ‘Bound’
   targets, in red, are top 500 NANOS2 CRAC targets; ‘Not bound’, in
   black, are mRNAs not bound by NANOS2. The indicated significance of
   difference between the NANOS2-bound and unbound transcripts was
   computed by Mann-Whitney U test.

   (E) Violin plots showing expression log[2] fold change in Nanos2
   overexpressing compared to wild-type SSCs. Putative top-500
   NANOS2-bound transcripts (annotated “Bound”) are compared to
   transcripts lacking CRAC peaks (annotated “Not bound”). The upper and
   lower quartiles and the median are indicated for each group. p value
   for group difference computed by Mann-Whitney U test.

   (F) Cumulative distribution plots showing log[2] fold expression change
   in Nanos2 overexpressing cells. Transcript groups are as in (E) above:
   top-500 NANOS2-bound (“Bound”) transcripts compared to unbound
   transcripts (“Not bound,” lacking NANOS2-CRAC peaks). p value for group
   difference computed by Mann-Whitney U test.

Discussion

   We have generated a versatile epitope-tagged Nanos2 allele that has
   enabled the molecular exploration of its function in SSCs. Notably, the
   use of transcriptomics and proteomics methods allowed us to explore
   NANOS2 mechanism of function in SSCs. The application of the stringent
   CRAC defined the NANOS2-bound target mRNAs transcriptome-wide. It also
   revealed that mouse NANOS2 directly binds RNA and has a well-defined
   consensus sequence, AUKAAWU. The length of the consensus is consistent
   with RNA binding by both zinc fingers of NANOS2, since each is
   predicted to recognize three nucleotides ([218]Choo and Klug, 1994).
   Drosophila NANOS does not bind RNA on its own, but only in cooperation
   with PUMILIO ([219]Sonoda and Wharton, 1999; [220]Weidmann et al.,
   2016). First, PUMILIO binds mRNAs through an eight-nucleotide consensus
   motif, and NANOS subsequently joins the complex ([221]Murata and
   Wharton, 1995; [222]Sonoda and Wharton, 1999; [223]Wharton and Struhl,
   1991). After that, PUMILIO consensus changes by enriching for A/U
   elements upstream of its binding site ([224]Weidmann et al., 2016). In
   contrast, our data reveal that mouse NANOS2 can directly bind RNA and
   also with sequence-specificity. Thus, despite NANOS proteins being
   highly conserved across evolution, the mode of NANOS2 binding has
   diverged from invertebrates to mammals. This conclusion is supported by
   the fact that the mammalian PUM2 is not required for mouse fertility
   ([225]Xu et al., 2007) and that PUM1, although it is expressed
   throughout spermatogenesis, does not have a defined spermatogonial
   function ([226]Chen et al., 2012). Furthermore, our protein-interaction
   studies in SSCs did not identify PUM1/2 or any other RBP as NANOS2
   partners and the NANOS2-PUM1/2 interaction was not observed from E15.5
   ([227]Suzuki et al., 2016). This supports the model that NANOS2 is
   sufficient for the binding and selection of its RNAs targets in mouse
   SSCs.

   Our data showed that NANOS2 mainly binds mRNAs within their 3′ UTRs,
   suggesting that NANOS2 could regulate the fate of its mRNA targets
   through post-transcriptional mechanisms ([228]Mayya and Duchaine,
   2019). Indeed, we found that many components of CCR4-NOT deadenylase
   complex copurified with NANOS2 in extracts from SSC lines. Association
   of NANOS2 with the CCR4-NOT complex was previously reported in
   embryonic gonads ([229]Suzuki et al., 2010), and NANOS2 was shown to
   interact with CNOT9 by co-IP in SSCs ([230]Zhou et al., 2015b).
   Moreover, the interaction between NANOS2 and CNOT1 is essential for the
   development of male germ cells ([231]Suzuki et al., 2012). In vitro
   deadenylase assays showed that immunoprecipitated NANOS2 from testis
   can deadenylate RNA substrates ([232]Suzuki et al., 2010), suggesting
   that NANOS2 may promote transcript degradation. Importantly, by using a
   transcriptome-wide approach to measure mRNA half-life in SSC lines, we
   have now demonstrated that NANOS2 binding is associated with an average
   20% reduction of transcript half-life, which across its many targets is
   likely to significantly impact the proteome. We therefore propose that
   in addition to the translational and sequestering mechanisms previously
   presented ([233]Zhou et al., 2015b), an important mechanism of NANOS2
   function in SSCs is to promote transcript turnover.

   By employing CRAC, we have now defined the full complement of
   NANOS2-bound transcripts in SSC lines. The SSC cultures contain a
   mixture of SSCs and progenitor cells, and as such, the NANOS2 target
   transcripts identified will be from both spermatogonial stages. We did
   observe a significant overlap between our study and the RIP-based
   studies form E14.5 and P7 testis ([234]Saba et al., 2014; [235]Zhou
   et al., 2015b) ([236]Figures S4C and S4D, [237]Tables S2 and [238]S3),
   which provides an independent in vivo validation of our approach. The
   differences in NANOS2 targets between the various datasets likely arise
   from the methodology used as well as the cell type examined, which
   differed between all the respective studies. The NANOS2-binding sites
   defined herein also provide a more thorough understanding of the genome
   through the possible interpretation of single-nucleotide polymorphisms
   or mutations associated with male infertility. It was previously shown
   by RIP that NANOS2 binds specific mRNAs encoding proteins important for
   SSC differentiation, including Sohlh2, Dazl and Taf7l ([239]Zhou
   et al., 2015b). CRAC also revealed many other NANOS2 targets, which are
   strongly over-represented for mRNAs encoding proteins involved in
   cellular metabolism and biosynthetic processes. The precise regulation
   of cellular metabolism is key for spermatogenesis ([240]Rato et al.,
   2012) and also for SSC self-renewal ([241]Kanatsu-Shinohara et al.,
   2016; [242]Morimoto et al., 2015). Many adult stem cells, including
   SSCs, must tightly control their proliferation rate in order to prevent
   premature exhaustion of the stem cell pool over time. By sensing and
   integrating multiple growth signals, mTOR balances many biosynthetic
   processes and thus sustains the anabolic growth and proliferation
   ([243]Ben-Sahra et al., 2013; [244]Kim and Guan, 2019; [245]Laplante
   and Sabatini, 2012). Indeed, mTOR signaling stimulates spermatogonia
   proliferation and differentiation ([246]Busada et al., 2015b; [247]Feng
   et al., 2000; [248]Hobbs et al., 2015), and consistently, long-term
   maintenance of SSCs requires that mTOR activity be retained at minimal
   levels ([249]Hobbs et al., 2010; [250]Wang et al., 2016). Importantly,
   we found that NANOS2 targets are enriched for mRNAs encoding proteins
   involved in mTOR signaling and other pathways which are essential for
   the regulation of SSC self-renewal. We conclude that selective binding
   by NANOS2 regulates the half-life of key mRNA targets, which are
   directly and indirectly involved in the control of the metabolic status
   and growth of the cell. In summary, we propose that NANOS2 represses
   many targets in order to regulate SSC quiescence and protect
   self-renewal potential.

Limitations of the study

   We have identified NANOS2-bound transcripts by using CRAC, a stringent
   biochemical method, in mouse SSC lines. However, we could not directly
   validate these NANOS2-RNA interactions in vivo by the same methodology
   due to the limiting amount of material that can be obtained from
   ex vivo-isolated spermatogonial populations. However, many of these
   CRAC-identified transcripts were also found as NANOS2 targets in
   NANOS2-RIP experiments performed in mouse embryonic and postnatal
   testes ([251]Saba et al., 2014) ([252]Zhou et al., 2015b), suggesting
   that SSC lines represent a good surrogate for their in vivo
   counterparts. Among NANOS2-bound transcripts identified by CRAC, we
   found many mRNAs involved in metabolic pathways which were previously
   shown to be important for SSC self-renewal. However, in this study, we
   have not investigated the contribution of NANOS2-mediated regulation of
   individual target transcripts to SSC maintenance. In the future, it
   will be interesting to see if mutations within NANOS2 consensus sites
   in target genes are associated with male infertility in mouse models or
   in humans.

STAR★Methods

Key resources table

   REAGENT or RESOURCE SOURCE IDENTIFIER
   Antibodies
     __________________________________________________________________

   Mouse monoclonal anti GFP (Clones 7.1 and 13.1) Roche Cat#11 814 460
   001; RRID: AB_390913
   Mouse monoclonal anti V5 Invitrogen Cat#R960-CUS; RRID: [253]AB_2556564
   Chicken Polyclonal anti GFP Aves Cat#GFP-1010; RRID: [254]AB_2307313
   Goat anti GFRα1 Neuromics Cat#GT15004; RRID: [255]AB_2307379
   Goat Polyclonal α-GFP ThermoFisher Cat#A-11122; RRID: [256]AB_221569
   Goat polyclonal anti C-kit R&D Systems Cat#AF1356-SP
   Rabbit polyclonal anti PLZF Santa Cruz Biotechnology Cat#sc-22839;
   RRID: AB_2304760
   Rabbit polyclonal anti GFP ThermoFisher Cat#A11122; RRID:
   [257]AB_221569
   Mouse monoclonal anti Dcp1a Sigma-Aldrich Cat#WH0055802M6; RRID:
   AB_1843673
     __________________________________________________________________

   Chemicals, peptides, and recombinant proteins
     __________________________________________________________________

   TAG-NANOS2 (N-terminal tagged NANOS2 protein) This paper N/A
     __________________________________________________________________

   Critical commercial assays
     __________________________________________________________________

   SLAMseq Explorer Kit – Cell viability Titration Module Lexogen
   Cat#SKU: 059.24
   SLAMseq Kinetics Kit – Catabolic Kinetics Module Lexogen Cat#SKU:
   062.24
   QuantSeq 3′ mRNA-Seq Library Prep Kit FWD Lexogen Cat#SKU: 015.24
   SENSE Total RNA-Seq Library Prep Kit Lexogen Cat#SKU: 042.24
   Ni-NTA Superflow QIAGEN Cat#30410
   RNace-It Ribonuclease cocktail Agilent Cat#400720
   RNasin Ribonuclease Inhibitor Promega Cat#N2115
   Pierce spin columns snap cap Thermo Scientific Cat#69725
   MetaPhor agarose Lonza Cat#50180
   LA Taq Takara Cat#RR002M
   MinElute Gel Extraction kit QIAGEN Cat#28604
     __________________________________________________________________

   Deposited data
     __________________________________________________________________

   CRAC-seq This paper GEO accession: [258]GSE149835
   RNA-seq This paper GEO accession: [259]GSE149835
   SLAM-seq This paper GEO accession: [260]GSE149835
   NANOS2 protein interactome in SSCs (IP-Mass spectrometry) This paper
   [261]Table S5
     __________________________________________________________________

   Experimental models: Cell lines
     __________________________________________________________________

   Mouse spermatogonial stem cells: Nanos2^CTL This paper N/A
   Mouse spermatogonial stem cells: Nanos2^TAG/+ This paper N/A
   Mouse spermatogonial stem cells: Nanos2^TAG/"tnTAG This paper N/A
   Mouse embryonic fibroblasts (feeders for SSCs) This paper N/A
     __________________________________________________________________

   Experimental models: Organisms/strains
     __________________________________________________________________

   Mouse: Nanos2^TAG DBA/2J x C57Bl/6 hybrid This paper (DBA/2J strain
   from Charles River Laboratories) DBA/2J strain code:
   625
   Mouse: Nanos2^TAG C57Bl/6 This paper N/A
     __________________________________________________________________

   Recombinant DNA
     __________________________________________________________________

   V5-c-Myc-PS-His6X-EGFP-FRT-neo-FRT (PS is PreScission) This paper N/A
     __________________________________________________________________

   Software and algorithms
     __________________________________________________________________

   FastQC software
   [262]http://www.bioinformatics.babraham.ac.uk/projects/fastqc/ N/A
   FASTX-collapser v0.0.14) [263]http://hannonlab.cshl.edu/fastx_toolkit/
   N/A
   pyCRAC ([264]Webb et al., 2014)
   [265]https://bitbucket.org/sgrann/pycrac
   flexbar v3.4.0 ([266]Dodt et al., 2012) N/A
   STAR v. 2.5.3a aligner ([267]Dobin et al., 2013) N/A
   Salmon v. 0.13.1 quasi-quantification ([268]Patro et al., 2017) N/A
   HTSeq v.0.11.2 ([269]Anders et al., 2015) N/A
   bamCoverage v3.1.3 (deepTools package) ([270]Ramírez et al., 2016) N/A
   SAMtools v1.9 ([271]Li et al., 2009) [272]http://www.htslib.org/
   Novoalign v2.07.00 Novocraft [273]http://www.novocraft.com
   Integrative Genomics Viewer Broad Institute
   [274]http://software.broadinstitute.org/software/igv/;
   RRID:[275]SCR_011793
   python 2.7 Jupiter notebooks, python libraries (pandas v0.19.2, NumPy
   v1.16.0, scipy v1.2.0) ([276]Turowski et al., 2016). update of gwide
   toolkit v0.5.27 [277]https://github.com/tturowski/gwide
   Bioconductor Limma Package ([278]Ritchie et al., 2015)
   [279]https://bioconductor.org/packages/release/bioc/html/limma.html
   Bioconductor topGO package Bioconductor
   [280]https://bioconductor.org/packages/release/bioc/html/topGO.html
   stats R package R project [281]https://www.r-project.org/
   MEME-ChIP [282]Machanick and Bailey, 2011
   [283]https://meme-suite.org/meme/tools/meme-chip
   SlamDunk pipeline ([284]Neumann et al., 2019)
   [285]https://github.com/t-neumann/slamdunk
   MaxQuant LFQ algorithm ([286]Cox et al., 2014) N/A
   Perseus version 1.6.0.2 ([287]Tyanova et al., 2016) N/A
   ingenuity-pathway-analysis IPA (QIAGEN Inc.) ([288]Krämer et al., 2014)
   [289]https://www.qiagenbio-informatics.com/products/ingenuity-pathway-a
   nalysis
   Affinity designer [290]https://affinity.serif.com/en-gb/ N/A
   FlowJo software [291]https://www.flowjo.com/solutions/flowjo/downloads
   N/A
   Fiji ImageJ. [292]https://imagej.net/Downloads N/A
   [293]Open in a new tab

Resource availability

Lead contact

   Further information and requests for resources and reagents should be
   directed to and will be fulfilled by the lead contact, Dónal O’Carroll
   ([294]donal.ocarroll@ed.ac.uk).

Materials availability

   Reagents generated in this study are available upon request from the
   Lead Contact.

Data and code availability

   The CRAC, RNA-seq, and SLAM-seq data sets generated in this study are
   available at GEO, accession [295]GSE149835. The data supporting this
   study are available in the [296]supplementary information.

Experimental models and subject details

Mouse generation and maintenance

   For the generation of the Nanos2^TAG allele, the sequence encoding
   V5-c-myc-PreScission-6xHis-EGFP (named ‘TAG’) was inserted after the
   endogenous ATG start codon of Nanos2. The targeting construct including
   the TAG sequence was genetically modified to include homology arms and
   an FRT-flanked neomycin cassette 3' of the 3' UTR. Individual A9 ESC
   clones were screened for homologous recombinants by Southern blotting
   of NheI-digested DNA with a 3' probe. Southern blotting was performed
   as described ([297]Morgan et al., 2017). The Nanos2^WT locus generated
   a DNA fragment of 6.6 kb, whereas the targeted Nanos2^TAG-neo locus
   produced a 3.8 kb fragment due to the presence of an additional NheI
   site, introduced after the integration of the 3′ FRT-flanked neomycin
   cassette in the Nanos2^TAG-neo locus. To generate mice A9-targeted ESCs
   were injected into C57BL/6 eight-cell-stage embryos, as described
   ([298]De Fazio et al., 2011). Nanos2^TAG-neo mice were crossed to
   FLP-expressing transgenic mice (FLPeR ([299]Farley et al., 2000)) to
   induce FLP-mediated excision of the FRT-flanked neomycin cassette, thus
   generating Nanos2^TAG mice. Female and male Nanos2^TAG mice were
   maintained on the C57BL/6 genetic background. For SSCs derivation,
   female Nanos2^TAG mice were crossed with DBA/2J male mice (purchased
   from the Charles River Laboratories) to generate mice in a mixed
   background. Female and male Nanos2^WT and Nanos2^TAG alleles were
   genotyped by PCR after extracting genomic DNA from ear biopsies using
   the following primers: “FW1_N2”: 5'-AACCTGGGGAATAACCTGCT-3', “FW2_N2”:
   5'-TGCTGCTGAATAAAGCGTTG-3', “RV_N2”: 5'-TCCCAGTCAGACGACTTGTG-3'.
   Fertility of male mice was assessed by setting up matings with 2 months
   old C57Bl/6 females and litter size was analyzed for the following
   3-4 months, at pups’ weaning age.

   Mice were generated at the EMBL Mouse Biology Unit, Monterotondo, and
   later bred and maintained at the Scottish Centre for Regenerative
   Medicine, University of Edinburgh. All procedures were done in
   accordance with the Italian legislation (Art. 9, 27. Jan 1992, nu116)
   or under UK Home Office authorization.

Derivation and maintenance of SSC lines

   SSCs were derived from testis of postnatal day 7 mice (males) and
   maintained as described in ([300]Kanatsu-Shinohara et al., 2003), with
   some modifications to the media. Cells were cultured in the following
   cell media (SSC media): Stem Pro-34 SFM medium with Supplements (2,6
   %), Invitrogen (10639-011); 6 mg/ml D-(+)-Glucose, Sigma (G7021);
   25 μg/ml Insulin, Sigma (I5500); 5 ml/ml BSA, MP-Biomedicals (810661);
   100 μg/ml apo-Transferrin human (diluted in Stem Pro-34 SFM medium),
   Sigma (T1147); 30 nM Sodium Selenite (diluted in Stem Pro-34 SFM
   medium), Sigma (S5261); 1X MEM Vitamin solution, Invitrogen (11120037);
   10 μg/ml D-Biotin (diluted in Stem Pro-34 SFM medium), Sigma (B4501);
   60 ng/ml Progesterone (diluted in 100% Ethanol), Sigma (P8783);
   30 ng/ml β-estradiol (diluted in 100% Ethanol), Sigma (E2758); 1X
   Sodium Pyruvate, Sigma (P2256-5G); 60 μM Putrescine (diluted in Stem
   Pro-34 SFM medium), Sigma (P7505); 2 mM L-Glutamine, Invitrogen
   (25030-024); 5.7 10^-7 M β-Mercaptoethanol, Bio-Rad (1610710); 1 μl/ml
   Lactate, Sigma (L4263); 1% FBS (HyClone Fetal Bovine Serum
   Characterized, 12379802, Fisher Scientific); 20 ng/ml epidermal growth
   factor (EGF), (EMBL); 10 ng/ml, fibroblast growth factor 2 (FGF2),
   (EMBL); 10 ng/ml glial cell line-derived neurotrophic factor (GDNF),
   (EMBL); 20 ng/ml Leukemia Inhibitory Factor (LIF), (EMBL); 100 U/ml
   Penicillin-Streptomycin, Invitrogen (15140-122). Cytokines were
   produced at the ‘Protein Expression & Purification’ Core facility, EMBL
   Heidelberg, Germany. SSCs were frozen in 50% FBS (12379802), 40% SSC
   media, 10% DMSO (D2650, Sigma). Both MEFs and SSCs were cultured at
   37°C and 7.5% CO2 and were tested for mycoplasma contamination
   routinely.

Method details

Histology

   Testes were fixed in Bouin’s solution (16045-1, Polysciences) and
   embedded in paraffin. Testes were sectioned with a microtome to obtain
   4-μm-thick slices, every ∼300-400 μm, to have ∼15 sections for testis.
   Testes slices were obtained from the beginning, a quarter and half of
   the testis longitudinal length. Slices were placed on a glass slide and
   these were stained with hematoxylin and eosin (H&E), according to
   standard protocols. Images of tubules’ sections were acquired with
   Zeiss Axio Scan Slide Scanner and the presence of spermatogenic cells
   was evaluated by analyzing images with the Zen Airyscan and Fiji ImageJ
   softwares.

Whole-mount immunofluorescence on seminiferous tubules

   Whole-mount immunofluorescence of seminiferous tubules was performed as
   described ([301]Carrieri et al., 2017). For the expression analysis of
   TAG-NANOS2 and GFRα1, co-staining of tubules was done with two primary
   antibodies: α-GFP, 1:500 (GFP-1010, Aves), α-GFRα1, 1:50 (GT15004,
   Neuromics). For the expression analysis of TAG-NANOS2 and c-KIT,
   co-staining was done with: α-GFP, 1:200 (A-11122, ThermoFisher) and
   α-C-kit, 1:250 (AF1356-SP, R&D Systems). For the expression analysis of
   TAG-NANOS2 and PLZF: α-GFP, 1:500 (GFP-1010, Aves), α-PLZF, 1:100
   (sc22839, Santa Cruz Biotechnology). For the expression analysis of
   TAG-NANOS2 and Dcp1a: α-GFP, 1:500 (GFP-1010, Aves), α-Dcp1a 1:500
   (WH0055802M6, Sigma-Aldrich). Secondary antibodies used were all
   Invitrogen: goat anti-Chicken, Alexa Fluor 488 (A-11039); donkey
   anti-Rabbit, Alexa Fluor 488 (A-21206); donkey anti-goat, Alexa Fluor
   568 (A-11057). Images were acquired by using a Confocal microscope, by
   taking Z-stacks, and with Z-stepsize set at 0.34 μm. Images were
   analyzed with Fiji ImageJ. A[s] and A[pr] spermatogonia were
   distinguished by using the 25 μm topographical criteria ([302]Huckins,
   1971). If the internuclear distance between two spermatogonia was over
   25 μm, cells were assigned to the A[s] category. On the contrary,
   spermatogonia whose internuclear distance was smaller than 25 μm, were
   considered to belong to the same chain.

FACS analysis and sorting of SSC lines

   For the analysis of GFP expression in SSC lines, MEFs were depleted by
   a brief sedimentation of MEF clumps and collection of supernatants,
   which were enriched for SSCs. SSCs were gated from residual MEFs and
   cell debris by selecting a cell population with low side and forward
   scatter, by excluding cell doublets, and by selecting only cells
   negative for the live cell dye DAPI. The analysis was performed with a
   BD Accuri C6 flow cytometer (BD Biosciences). SSC sorting was performed
   on FACSAria II Cell Sorter (BD Biosciences) using the same gating
   strategy detailed above. Data were analyzed using the FlowJo software.

CRAC and analysis

   The CRAC protocol was performed as described ([303]Granneman et al.,
   2009; [304]Turowski et al., 2016), with the following modifications.
   SSCs were grown on feeders in 150 cm^2 culture dishes for ∼7 days and
   irradiated with UV-C light (254 nm), with energy set at 0.4 J/cm^2,
   using a Stratalinker. TAG-NANOS2 was immunopurified from cell lysate by
   using an anti-V5 antibody (R960-CUS, Invitrogen) coupled with Dynabeads
   (11206D, Life technologies). TAG-NANOS2 protein was released from the
   antibody by proteolytic cleavage, using 40 μg of 3C-GST Protease, at
   4°C. Purified libraries were sequenced on an Illumina NextSeq500
   instrument, high output 75 bp, single-end run.

   Illumina sequencing data were demultiplexed using in-line barcodes and
   in this form were submitted to GEO. The first quality control step was
   performed using FastQC software
   ([305]http://www.bioinformatics.babraham.ac.uk/projects/fastqc/)
   considering specificity of CRAC data. Raw reads were collapsed to
   remove PCR duplicates using FASTX-collapser v0.0.14
   ([306]http://hannonlab.cshl.edu/fastx_toolkit/) then in-line barcodes
   were removed using pyBarcodeFilter.py script from pyCRAC package v3.0
   ([307]Webb et al., 2014). The 3’ adapters were removed using flexbar
   v3.4.0 ([308]Dodt et al., 2012) including flexbar quality filter.
   Pre-processed reads were mapped to the mouse genome (mm10) by using
   STAR v. 2.5.3a aligner ([309]Dobin et al., 2013) and Salmon v. 0.13.1
   quasi-quantification ([310]Patro et al., 2017). Mapped reads were used
   to determine their distribution among different classes of RNAs using
   pyReadCounter (pyCRAC package) and HTSeq v.0.11.2 ([311]Anders et al.,
   2015). BigWig files were generated using bamCoverage v3.1.3 script from
   deepTools package ([312]Ramírez et al., 2016) and visually inspected.
   Sam files operations were performed using SAMtools v1.9 ([313]Li
   et al., 2009).

   For the analysis of binding motifs preprocessed Nanos2^TAG/TAG reads
   were mapped to the mouse transcriptome database generated using Biomart
   ([314]Smedley et al., 2015). Reads were aligned to the mouse
   transcriptome using Novoalign v2.07.00 ([315]http://www.novocraft.com)
   with –r random and saved as sam files. Deletions, reflecting sites of
   direct RNA-protein contact, were extracted from aligned reads using
   custom made python script. Further steps were performed for both, full
   length mapped reads and deletions only. Data were converted to BigWig
   files using bamCoverage programme. Downstream analysis was performed
   using python 2.7 Jupiter notebooks, python libraries (pandas v0.19.2,
   NumPy v1.16.0, scipy v1.2.0) and in-house functions and scripts
   submitted as an update of gwide toolkit v0.5.27
   ([316]https://github.com/tturowski/gwide) ([317]Turowski et al., 2016).
   Peaks were called for the Nanos2^TAG/TAG profiles (FDR < 0.001) using
   argrelextrema function from signal processing library scipy.signal
   (v1.3.0) using order value 20. The total peak score for each gene was
   calculated by summing the intensities of one or more unique peaks
   present in each transcript and was expressed as reads per million (rpm)
   and averaging across two replicate CRAC experiments. These averaged
   peak scores were then normalized by transcript FPKM (fragments per
   kilobase million)averaged from two biological replicate mRNA-seq
   datasets (from Nanos2^TAG/+ and Nanos2^TAG/TAG SSC lines).
   Non-expressed genes, FPKM = 0, were eliminated from the analysis. The
   resulting ratio values were used to rank the final NANOS2 CRAC list
   (1428 genes). To define NANOS2 binding site we filtered sequencing
   reads associated with the top 500 most highly enriched targets, for the
   ones containing deletions. 3844 reads, corresponding to 498 out of 500
   targets, were identified as having deletions. This subset of reads was
   centered on deletions and only the region spanning 20 or 100 bases
   around deletions was considered. Motif analysis was performed by using
   MEME-ChIP ([318]Machanick and Bailey, 2011), and DREME ([319]Bailey,
   2011). Centrimo ([320]Bailey and Machanick, 2012) was also used to
   evaluate the central enrichment of the motifs found. As a background,
   we used the same sequences shuffled by MEME-ChIP. We looked for motif
   with E-value < 0.05, width 6-10 bp, with ‘0-1’ or ‘1’ occurrence per
   sequence, and we scanned only the strand provided.

Comparison between CRAC and published RIP datasets

   To identify NANOS2 targets from NANOS2 RIP-array experiments in E14.5
   male gonads ([321]Saba et al., 2014) and P7 testis ([322]Zhou et al.,
   2015b), the original array data sets (NCBI GEO accessions [323]GSE37718
   and [324]GSE61807) were analyzed. Probe data were background corrected
   and quantile normalized, followed by fitting of a differential
   intensity model, using the limma package in R ([325]Ritchie et al.,
   2015). NANOS2-bound targets were identified as transcripts with a
   greater than 2-fold excess in signal intensity in the NANOS2 IP versus
   control IP and by false discovery rate (FDR) of less than 0.05. These
   target lists were overlapped with our NANOS2 CRAC list. Violin and
   cumulative distribution plots were generated comparing top-500
   NANOS2-bound transcripts (ascertained by normalized CRAC scores) to
   transcripts not showing NANOS2 binding. Significant differences were
   detected using Mann-Whitney U tests.

RNA isolation and RNA-seq

   RNA was isolated from sorted SSCs by using QIAzol Lysis Reagent (79306,
   Qiagen) and following the manufacturer’s instructions. To generate
   RNA-seq libraries total RNA was treated with DNAse I, in 10X Buffer
   (AMPD1, Sigma). RNA was purified using Rneasy MinElute columns (74204,
   Qiagen) and ribosomal RNA depletion was performed with the RiboCop kit
   (Lexogen). Ribo-depleted RNA was used to generate libraries with the
   SENSE Total RNA-Seq Library Prep Kit (Lexogen) following the
   manufacturer’s instructions. Libraries were sequenced with an Illumina
   HiSeq2500 on a 50 bp, single-end run.

Gene ontology (GO) analysis

   The gene ontology (GO) enrichment of the top 500 NANOS2 CRAC targets
   was determined by using the topGO R package. The enrichment for the
   biological process ontology was assessed by using Fisher’s exact test
   (P-value < 0.05).

Ingenuity pathways analysis

   The pathways enrichment analysis of the top 500 NANOS2 CRAC targets was
   generated through the use of IPA (QIAGEN Inc.,
   [326]https://www.qiagenbio-informatics.com/products/ingenuity-pathway-a
   nalysis) ([327]Krämer et al., 2014). The significance of the enrichment
   for the signaling pathways was defined by setting -log P-value > 1.3,
   which corresponds to a 0.05 significance threshold.

Immunoprecipitation followed by mass spectrometry (IP-MS)

   IP-MS was performed from three replicates of Nanos2^CTL and
   Nanos2^TAG/TAG SSC lines. TAG-NANOS2 was immunoprecipitated with
   Anti-GFP antibody (11 814 460 001, Roche) cross-linked to Protein G
   magnetic beads (88848, Thermo) and processed as described ([328]Much
   et al., 2016). Peptides were separated on an ultra-high resolution
   nano-flow liquid chromatography nanoLC Ultimate 3000 unit fitted with
   an Easyspray (50 cm, 2 μm particles) column coupled to the high
   resolution/accurate-mass mass-spectrometer Orbitrap Fusion Lumos
   operated in DDA(data-dependent-acquisition)-mode (Thermo Fisher
   Scientific). Raw data were processed using MaxQuant version 1.6.1.0.
   Label-free quantitation (LFQ) was performed using the MaxQuant LFQ
   algorithm ([329]Cox et al., 2014). Peptides were searched against the
   mouse UniProt database (date 21.07.2017) with commonly observed
   contaminants (e.g. trypsin, keratins, etc.) removed during Perseus
   analysis ([330]Cox et al., 2014; [331]Hubner et al., 2010; [332]Tyanova
   et al., 2016; [333]The UniProt Consortium, 2017). For visualization,
   LFQ intensities were imported into Perseus version 1.6.0.2
   ([334]Tyanova et al., 2016) and processed as described ([335]Hubner
   et al., 2010).

SLAM-seq

   The optimal concentration of 4SU (4-Thiouridine) for SSC lines (100 μM)
   was determined by using the cell viability titration assay in the
   Lexogen SLAMseq Explorer Kit. SSCs labeling with 4SU was performed as
   followed: media with 4SU was supplied twice to SSCs, 24 and 3 hours
   before time 0 (t[0]). At t[0], 4SU-media was replaced with
   4SU-free-media and SSCs were isolated 0, 30 minutes, 1, 2, 9 and 24
   hours later.

   1.5 x 10^6 SSCs were isolated by FACS for each sample and timepoint.
   SLAM-seq libraries were prepared by using the Lexogen SLAMseq Kinetics
   Kit – Catabolic Kinetics Module and the Lexogen QuantSeq 3′ mRNA-Seq
   Library Prep Kit FWD for Illumina, following the manufacturer’s’
   instructions. Libraries were sequenced using an Illumina HiSeq2500
   platform on a 50 bp, single-end run. Two biological replicates were
   used to generate libraries with each biological replicate was composed
   of two technical replicates.

   Analysis of SLAM-seq libraries was performed by using the SlamDunk
   pipeline ([336]Neumann et al., 2019). T to C conversion rates obtained
   from different time points were normalized to t[0] for each gene and
   were used to fit a first-order decay reaction, with the R stats package
   nls function. The two technical replicates present in each biological
   replicate were collapsed before calculation of half-life.

Quantification and statistical analysis

   All details of statistical analyses, including replicates, statistical
   tests and outcomes, are described in the Method details, main results
   section, and Figure legends.

Acknowledgments