The RNA-binding protein LIN28A is required for maintaining tissue homeostasis, including in the reproductive system, but the underlying mechanisms on how LIN28A regulates germline progenitors remain unclear. Here, we dissected LIN28A-binding targets using high-throughput sequencing of RNAs isolated by crosslinking immunoprecipitation (HITS-CLIP) in the mouse testes. LIN28A preferentially binds to mRNA coding sequence (CDS) or 3′UTR regions at sites enriched with GGAG(A) sequences. Further investigation of Lin28a-null mouse testes indicated that meiosis-associated mRNAs bound by LIN28A were differentially expressed. Next, ribosome profiling revealed that the mRNA levels of these targets were significantly reduced in the polysome fractions, and their protein expression levels decreased, in Lin28a-null mouse testes, even when meiotic arrest in the null mouse testes was not apparent. Collectively, these findings provide a set of LIN28A-regulated target mRNAs, and show that LIN28A binding might be a mechanism through which LIN28A acts to regulate undifferentiated spermatogonia fates and male fertility in mammals.
Post-transcriptional regulation plays an important role in organizing cellular function, which usually requires RNA-binding proteins (RBPs) to regulate cellular processes. The RNA-binding protein LIN28A was originally identified as a key regulator of developmental timing in Caenorhabditis elegans (Moss et al., 1997), and its expression is tightly regulated during morphogenetic processes (Moss and Tang, 2003). In early male germ cells of mice and humans, LIN28A is required for diverse biological processes, including embryogenesis, growth and tumorigenesis (Peng et al., 2011; Polesskaya et al., 2007; Viswanathan et al., 2009). LIN28A is also highly expressed in human embryonic stem cells (hESCs) and facilitates the reprogramming of fibroblasts to induced pluripotent stem cells (iPSCs) (Yu et al., 2007). Moreover, a study demonstrated that transgenic mice that overexpress LIN28A exhibit overgrowth and delayed onset of puberty because of impaired glucose metabolism and insulin sensitivity (Zhu et al., 2010), confirming the role of LIN28A in cell growth and survival. In the germline, LIN28A is a critical determinant of cell fate and proliferation, and has been shown to be specifically expressed in undifferentiated spermatogonia in both mice and humans. Its expression has also been associated with pluripotency (Aeckerle et al., 2012; Zheng et al., 2009). A conditional knockout (cKO) of Lin28a in germ cells results in a significant decrease in the number of primordial germ cells (PGCs), resulting in a marked reduction in testis mass and sperm number, which in turn leads to impaired fertility in adult Lin28a mutant males (Chakraborty et al., 2014; Shinoda et al., 2013a,b), thereby revealing that LIN28A acts as a stem cell factor in the reproductive system.
At the molecular level, LIN28A has been identified as a key regulator of developmental timing, controlling ESC self-renewal not only by regulating miRNA let-7 precursors but also by binding to mRNA targets (Cho et al., 2012; Newman and Hammond, 2010). In mouse testis, the LIN28A–let-7 pathway is critical for proper germ cell pool development at the PGC stage in vivo, whereas LIN28A is dispensable to postnatal germ cell expansion (Shinoda et al., 2013a,b). Independently of the let-7 pathway, LIN28A regulates the translation of Oct4 (also known as Pou5f1) and Dhx9 (also known as RHA) by directly binding to mRNA of ESCs at high-affinity sites within their coding region (Qiu et al., 2010). In addition, LIN28 positively influences proliferation, in part by binding to and enhancing the translation of mRNAs related to the growth and survival of ESCs (Peng et al., 2011). Thus far, ESCs have been utilized in research studies to reveal LIN28A-targeted RNAs and binding site sequences through high-throughput sequencing of RNA isolated by crosslinking immunoprecipitation (HITS-CLIP), which is the most effective method for studying the target RNAs of RBPs (Cho et al., 2012; Mayr and Heinemann, 2013; Wilbert et al., 2012). This technique introduces covalent bonds between bases and amino acids that are in close proximity, capturing the physiological state of RNA–protein interactions (Kishore et al., 2011). It has been reported that LIN28A binds to GGAGA sequences enriched within the loop structures at the 3′UTR of target mRNAs, such as TDP-43 (also known as Tardbp) or Fus (also known as TLS), and controls their protein abundance (Wilbert et al., 2012). Although LIN28A deletion in the mouse testes results in subfertility, the underlying mechanisms by which LIN28A regulates its targets in the germline remain unclear.
To address this question, we first generated LIN28A protein–RNA interaction maps to analyze LIN28A-binding RNAs through HITS-CLIP for the 10-day post-partum (dpp) mouse testes. Of note, we discovered over 12,000 mRNA targets of LIN28A in undifferentiated mouse spermatogonia. Our analysis of binding sites revealed the LIN28A-bound mRNA sequences were enriched in GGAG or GGAGA motifs. Combined with RNA-seq in Lin28a cKO mouse testes, we found that LIN28A binds to and regulates certain target mRNAs that are preferentially involved in meiosis. In this study, we confirmed that LIN28A binds to GGAG(A) motifs within the meiotic genes Hormad1, Terb1 and Prdm9 at their 3′UTR, and LIN28A acts as an enhancer of target mRNA translation in meiosis.
Lin28a knockdown does not affect the repopulation capability of germ cells
LIN28A is one of the critical regulators of reproductive function and acts by regulating germ cell development and playing a direct role in the growth and survival of PGCs (Shinoda et al., 2013a,b). To investigate the effect of LIN28A on male mouse germ cells, we established three lines of cultured germ cells that formed clusters with typical grape-shaped morphology from transgenic mice with a β-galactosidase reporter gene (Fig. 1A). Then, we knocked down Lin28a in these established cultured germ cells using two different lentivirus-mediated small hairpin RNAs (Lin28a shRNA575 and shRNA746). Western blot analysis confirmed that LIN28A protein levels were significantly decreased in cultured cells transfected with both Lin28a shRNA575 and shRNA746 (Fig. 1B). To assess whether knocking down Lin28a affects the capacity of germ cells to undertake spermatogenesis, we transplanted an established germ cell culture from Rosa transgenic mice that had Lin28a knocked down (>75%) into recipient testes. Donor-derived regeneration colonies were visualized by X-gal staining (Fig. 1C). However, we did not find any difference in colony numbers between the transplanted Lin28a shRNA and the control germ cells (vector plasmid transfected, Fig. 1D), indicating that partial loss of LIN28A in vitro did not alter the functional properties of germ cells upon transplantation. Similar observations were made in a previous study that showed that Lin28a is not required for the self-renewal potential of undifferentiated spermatogonia in the mouse testes (Chakraborty et al., 2014). These findings further support our previous results that LIN28A does not affect the self-renewal of germ cells (Zheng et al., 2009).
Capturing LIN28A RNA targets in the mouse testes using HITS-CLIP
To systematically explore the global mRNA binding of LIN28A in vivo, we first performed three biologically independent HITS-CLIP experiments (RL5i1, RL5i3 and RL5i7) in 10 dpp mouse testes, in which LIN28A is abundant in the spermatogonia (Zheng et al., 2009). Using a reliable antibody (Cho et al., 2012; Wilbert et al., 2012), this method takes advantage of 254 nm UV-mediated crosslinking between interacting RNA and proteins in live cells, immunoprecipitation (IP) of protein–RNA complexes and deep sequencing of the co-precipitated RNAs (Fig. 2A), thereby providing a global view of the interacting transcriptome and allowed the precise mapping of binding sites in the target RNAs. Our IP data showed that there was a substantial amount of LIN28A in both the no-UV and UV crosslinked groups (Fig. 2B). Meanwhile, autoradiography of 32P-labeled RNAs (as shown in red boxes) was performed on RNPs separated on a gel, which were then purified. These RNAs then had connectors added in order to establish a cDNA library through high-throughput sequencing (Fig. 2C). The correlation among the three CLIP-seq libraries found was evaluated to be >60%, indicating that the experiments were highly reproducible (Fig. S1), and most of the LIN28A CLIP targets originated from LIN28A-interacting RNAs. By analyzing the types of RNAs that interact with LIN28A in three CLIP-seq libraries (Table S1), we determined that, as expected, mRNA was the predominant RNA form, and comprised ∼90% of the RNAs, and the second most abundant class consisted of miscRNA (Fig. 2D), confirming that importance of LIN28A regulation by directly binding to its mRNA targets, in addition to let-7 precursors (Wilbert et al., 2012). By assessing the annotation of the nucleotides found, we found that there was a significant enrichment of LIN28A binding within the CDS and 3′UTR of target mRNAs (Fig. 2E). These findings were identical to the proportion of the different types of mRNA and the proportion of the regions within those mRNA that were discovered in ESCs and HEK293 cells (Cho et al., 2012; Wilbert et al., 2012). We next selected replicates (RL5i1 and RL5i3) that had a high number of overlapping genes with previously published mRNAs from mESCs CLIP-seq experiments (Cho et al., 2012). A total of 12,574 genes were shared among all three data sets, and an average of 75% of the previously identified targets was detected in the mouse testes CLIP-seq data set (Fig. 2F), revealing that the different mRNAs bound by LIN28A and the mechanism of LIN28A binding may be highly similar between the mouse testes and mESCs. Thus, these findings indicate that mRNAs are the major interactors of LIN28A protein that bind to the CDS or 3′UTR of the target mRNAs.
LIN28A mediates mRNA expression of meiotic genes in the mouse testes
A previous study has shown that Lin28a deficiency compromises the germ cell pool in both males and females by affecting PGC proliferation during embryogenesis (Shinoda et al., 2013a,b). To explore LIN28A-mediated mRNA expression of LIN28A in the mouse testes, we reproduced a germline-specific Lin28a knockout mouse by crossing mice that carried a floxed allele of Lin28a (purchased from Jackson Laboratory, Bar Harbor, ME, USA; catalog J023913) with mice that expressed Vasa-Cre as previously described (Shinoda et al., 2013a,b). Similar to what was found in that report, the testes of adult Lin28a cKO mice were smaller than those from the wild-type (WT) mice (Fig. 3A). To better understand the functional significance of LIN28A target mRNAs, high-throughput deep sequencing was applied to analyze 10 dpp WT and Lin28a cKO mouse testes with three samples in each group. We then used the UCSC genome browser to visualize RNA sequences from the deep sequencing, revealing that the second exon of the Lin28a gene in the cKO testes was specifically knocked out (Fig. 3B), confirming that the strategy of cKO was successful. Further analysis of RNA-seq data indicated that the mRNA species that were significantly differentially expressed (fold change>1.5 and P<0.05) in response to LIN28A depletion were expressed in both WT and cKO testes (Fig. 3C). Furthermore, we compared the two replicates of CLIP (RL5i1 and RL5i3) with the significant differentially expressed genes found from the 10 dpp mouse testes RNA-seq, and a total of 132 candidate genes were identified in the above three data sets (Fig. 3D), suggesting that these mRNAs interact and are directly regulated by LIN28A. Next, we performed gene ontology (GO) analysis of these candidate mRNA targets, to assess the GO process with which they were associated and so gain insight into the functional roles of LIN28A. The GO terms enriched with candidate genes included meiotic cell cycle, synapsis, male meiosis, and spermatogenesis. Of note, the GO term meiotic cell cycle was strongly over-represented in our data (Fig. 3E), which comprised nine genes (Dmc1, Hormad1, Prdm9, Spo11, Terb1, Ccnb3, Dpep3, Sycp1 and Sycp2). Next, we investigated the effect of lack of LIN28A expression on the expression of these targets, and we found that the transcript levels of some of these meiotic target genes were significantly changed in 10 dpp Lin28a deficiency mouse testes (Fig. 3F). Overall, these results suggest that candidate genes selected by our analysis involved in meiotic events are most likely to be targets of LIN28A.
LIN28A interacts with GGAG(A) motifs at the 3′UTR within mRNA sequences
Previous studies have shown that LIN28A binds to GGAG(A) sequences within mRNAs and provided evidence of LIN28A autoregulation by directly binding to its own mRNA in human ESCs (Wilbert et al., 2012). To better understand LIN28A–RNA interactions in mouse testes, we focused our analysis on mRNA targets with LIN28A-binding sites at their 3′UTR. Resolution of binding sites afforded by HITS-CLIP identified potential motifs involved in the interaction of LIN28A with mRNA sequences. We further used a differential motif discovery algorithm (HOMER; http://homer.ucsd.edu/homer/ngs/index.html; Heinz et al., 2010) to identify the consensus sequence of LIN28A-binding sites and discovered that GGAG or GGAGA motifs were enriched in LIN28A-binding sites within mRNA sequences. Of the most enriched 4-mer or 5-mer motifs, the GGAG and GGAGA ranked top with a much higher statistical significance than other motifs (Fig. 4A). This is consistent with results from a previous study showing that the specific binding sites of LIN28A are GGAG or GGAGA motifs (Wilbert et al., 2012). To precisely map LIN28A-binding sites, we used a dual-luciferase assay to evaluate meiotic target mRNAs such as Dmc1, Hormad1 and Terb1, which contain GGAGA motifs, and Sycp2 and Prdm9, which contain GGAG motifs at their 3′UTR. We then mutated the motifs GGAG(A) to CCTC(T) in the 3′UTR of target mRNAs, and the cloned regions of the 3′UTRs were inserted downstream of the luciferase reporters (Fig. 4B, left panel). In contrast to WT reporters, co-transfection of mutated (Mut) reporters for Hormad1, Terb1 and Prdm9, but not Dmc1 or Sycp2, with a plasmid expressing the LIN28A CDS led to an enhancement in luciferase activity (Fig. 4B, right panel). Transfection of WT and Mut reporters with a control plasmid lacking the LIN28A CDS (Empty) caused no effect on luciferase activity. These findings indicate that LIN28A binds to 3′UTR around a G-rich GGAG(A) motif in Hormad1, Terb1 and Prdm9. Thus, our results indicate that LIN28A as a mediator that preferentially interacts with consensus GGAG(A) motifs within target mRNA transcripts at the 3′UTR.
LIN28A facilitates propagation of cultured germ cells in vitro
To investigate the physiological function of LIN28A in cultured mouse germ cells, we generated Lin28a cKO cultured cells in vitro using validated protocols from Lin28a germline knockout mice (Kubota and Brinster, 2008). Western blot analysis of cultured cKO germ cells showed a significant reduction in LIN28A protein expression levels, whereas there was no significant change in the protein expression levels of LIN28B, a homolog of LIN28A (Fig. 5A). In addition, the cultured cells were positive for LIN28A and ZBTB16; however, cKO cells were positive for ZBTB16 but negative for LIN28A (Fig. 5B). To validate the effect of Lin28a deficiency on germ cell propagation, we counted the number of cultured cells for more than 1 month. The propagation rate of germ cells in the cKO group was significantly slower than that of WT after 6 days of culture in vitro (P<0.05), and the number of cells in the cKO group was half that of the WT on day 36 (Fig. 5C), confirming that loss of LIN28A negatively affects germ cell propagation. Because the transcript levels of some meiotic target genes had significantly changed in Lin28a deficiency mouse testes (Fig. 3F), we hypothesize that increased mRNA expression of meiotic genes may inhibit Lin28a cKO germ cell propagation. We found, that the transcript levels of multiple meiotic target genes, including Hormad1, Prdm9 and Terb1, were increased in cultured Lin28a cKO germ cells as compared to WT cells (Fig. 5D). In summary, our in vitro study demonstrates that LIN28A plays a pivotal role in germ cell maintenance, at least in part by recognizing specific sequence elements within multiple meiotic mRNAs.
LIN28A modulates translation of its target mRNAs in vivo
At the molecular level, several previous reports have suggested that LIN28A is a positive regulator of mRNA translation (Balzer and Moss, 2007; Jin et al., 2011; Mendez and Richter, 2001; Polesskaya et al., 2007). It is, however, unclear how LIN28A modulates mRNA translation and to what extent this function affects physiological events in the mouse testes. To further explore LIN28A-binding target mRNAs and monitor translational efficiency at the genomic level, we performed cell polysome separation with 10 dpp WT and Lin28a cKO mouse testes. RNA was isolated from the fractions of free ribonucleoprotein (RNP) and polysomes by gradient centrifugation and fractionation (Fig. 6A, top panel). Western blot analysis confirmed the absence of ribosomal protein L6 (RPL6) and tubulin from RNP and polysome fractions, respectively (Fig. 6A, bottom). The level of several of LIN28A putative target mRNAs in the polysome fractions, including those for Dmc1, Sycp2, Terb1, Hormad1 and Prdm9 in the cKO group markedly decreased (Fig. 6B). If LIN28A binding to target mRNAs reflects a mode of gene regulation, then we hypothesize that deleting LIN28A would alter the protein expression of these targets and promote the initiation of meiosis during spermatogenesis. As expected, the expression levels of LIN28A meiotic target proteins, DMC1 and SYCP2, was significantly reduced in Lin28a cKO mouse testes compared to in WT (Fig. 6C and data not shown), whereas that of germ cell-associated protein ZBTB16 did not significantly change, which coincides with LIN28A being a translational regulator of its target mRNAs. These findings support a role for LIN28A in the translational regulation of certain mRNAs. Because meiotic gene expression was decreased in Lin28a deficiency cells, we determined whether deleting LIN28A induces meiotic arrest. A previous study reported that Dmc1 mutants exhibit high levels of chromosome asynapsis, and synaptic defects often lead to meiotic arrest at the pachytene stage in males (Mahadevaiah et al., 2008). We analyzed this particular meiotic assembly by immunostaining spermatocytes using anti-SYCP1 and SYCP3 antibodies (Fig. 6D), among which meiotic stages of spermatocytes were determined on the basis of SYCP3 immunolabeling. Contrary to our expectation, the meiotic chromosome spreads did not reveal detectable differences in the proportion of meiotic-stage cells in the Lin28a cKO mice compared to the WT (Fig. 6E), suggesting that it is likely that Lin28a is dispensable in meiotic recombination. Nonetheless, our data indicate that LIN28A modulates the translation of meiotic target mRNAs during spermatogenesis in the mouse testes, potentially elucidating the mechanism of LIN28A in regulating mammalian undifferentiated spermatogonia fates and male fertility.
RNA-binding proteins directly interact with RNA transcripts in cells to exert various forms of regulation such as alternative splicing, turnover, localization and translation. LIN28A is a highly conserved RNA-binding protein that plays a crucial role in regulating developmental timing in C. elegans and mammals (Moss et al., 1997; Shinoda et al., 2013a,b; Viswanathan et al., 2009). Our previous studies have shown that LIN28A is specifically expressed in undifferentiated spermatogonia of the mouse testes (Li et al., 2016; Zheng et al., 2009). However, the molecular mechanisms by which LIN28A binds to its direct targets and regulates stem cell fate in the germline remain elusive. This study has shown that the propagation of cultured germ cells from Lin28a-null mouse testes is significantly reduced (Fig. 5C), whereas the characteristics of stem cells are not significantly affected by transplantation (Fig. 1D), confirming that LIN28A promotes male germ cell propagation in vitro.
Several studies have shown that LIN28A exerts its biological effects mostly by directly inhibiting let-7-dependent pathways in stem cells (Moss and Tang, 2003; Rybak et al., 2008). However, independent of prerequisite alterations in let-7 levels, several studies have also shown that LIN28A has important regulatory functions that are mediated by direct interactions with target mRNAs. For example, LIN28A directly binds to Oct4 mRNA through high-affinity sites to facilitate its expression at the post-transcriptional level (Qiu et al., 2010). Conversely, a previous study has reported that LIN28A, as a suppressor, contributes to the global translation of mRNAs involved in the secretory pathway of human ESCs (Cho et al., 2012). In addition, the MAPK/ERK signaling pathway phosphorylates Ser200 of LIN28A to increase protein stability and regulate pluripotency at the post-transcriptional level (Tsanov et al., 2017). Moreover, LIN28A preferentially binds to GGAGA motif-enriched sites within mRNA sequences that encode RNA processing and splicing factors (Wilbert et al., 2012). Thus, we wanted to determine how LIN28A regulates mRNA processing and affects mouse germline cells by identifying its specific target RNAs and binding sites. The HITS-CLIP approach allowed us to systematically identify and map the binding sites of a RNA-binding protein at a genomic scale by taking advantage of single-nucleotide resolution in vivo (Chi et al., 2009; Licatalosi et al., 2008). When comparing data on annotation of nucleotides with our genome-wide HITS-CLIP screen data, we found that the binding regions of LIN28A in the mouse testes within target mRNAs were significantly enriched in CDS and 3′UTR regions (Fig. 2E). Systematic identification of LIN28A-binding sites revealed that more than 12,000 genes are candidate targets of LIN28A in the mouse testes, and may provide a more accurate list of LIN28A targets. Furthermore, our analyses indicate that GGAG or GGAGA motifs are present within LIN28A-binding mRNA sites, which resemble the sequence identified in ESCs (Wilbert et al., 2012). In addition, the three LIN28A target mRNAs of Hormad1, Terb1 and Prdm9, were found to contain GGAG(A) motifs at their 3′UTR. The meiotic gene Hormad1 is a critical component of the synaptonemal complex and is essential for mammalian gametogenesis, as knockout male mice are infertile, with deficient testes undergoing meiotic arrest at the early pachytene stage, and synaptonemal complexes cannot be visualized (Shin et al., 2010). In addition, Terb1 is a meiosis-specific telomere-associated protein, and disruption of Terb1 in mice abolishes meiotic chromosomal movement and impairs homolog pairing, synapsis and recombination, thereby causing infertility in male mice due to lack of postmeiotic cells in the testes (Shibuya et al., 2014). Furthermore, Prdm9 is the only known vertebrate hybrid-sterility gene that determines where DNA breaks should occur during meiotic prophase, thereby causing failure of meiotic chromosome synapsis and infertility in male hybrids (Flachs et al., 2012). These studies suggested that LIN28A might regulate such meiotic mRNA targets to regulate the meiosis fate decision. How does Lin28a influence meiotic genes? One possibility was that Lin28a regulates and/or binds to meiotic genes. To validate the mechanism though which LIN28A acts, we tested whether mutations of the GGAG(A) motif in the 3′UTR affected the LIN28A affinity for these sites, and found that such mutations strongly reduced binding, confirming that the identified motifs provide genuine direct binding sites for LIN28A in the 3′UTRs of the meiotic mRNAs Hormad1, Terb1 and Prdm9 (Fig. 4B). This suggests that the GGAG(A) motifs identified in our analyses may indeed be sufficient for LIN28A binding. Our results provide evidence that LIN28A regulates gene networks through direct binding to GGAG(A) motifs at the 3′UTR of its meiotic mRNA targets.
Because mRNAs that are actively being translated are associated with polysomes, and LIN28A positively influences (i.e. enhances) the translation of its target mRNAs into proteins in ESCs (Jin et al., 2011), we hypothesized that LIN28A in the mouse testes would be associated with multiple meiotic mRNAs in polysomes. Consistent with this prediction, ribosome profiling provided evidence that mRNA and protein levels of candidate targets in polysomes, such as Sycp2 and Dmc1, decrease when Lin28a is knocked out (Fig. 6B,C), indicating that certain meiotic genes are the major functional targets of LIN28A during translational control. Of note, SYCP2 is required for normal meiotic chromosome synapsis during oocyte and spermatocyte development, and plays a role in the assembly of synaptonemal complexes (Yang et al., 2006). Sycp2-deficient male mice are sterile because of the absence of axial element formation and the subsequent disruption of chromosome synapsis in prophase I spermatocytes. Nevertheless, Sycp2 heterozygous males are fertile (Yang et al., 2006). Likewise, DMC1 is required for the resolution of meiotic double-strand breaks and may participate in meiotic recombination, specifically in homologous strand assimilation (Pezza et al., 2007). Dmc1-deficient male mice have no germ cells that develop further than the zygotene stage, leading to infertility; however, in Dmc1 heterozygous testes, normal proliferation and spermatogenesis occurs, and mature spermatozoa are observed (Yoshida et al., 1998). Owing to their clear roles in meiosis, we then hypothesized that LIN28A may influence meiotic progress. Surprisingly, contrary to our expectation, meiotic chromosome spreads showed that Lin28a deficiency has no effect on meiotic events during spermatogenesis (Fig. 6D,E). Although Lin28a deficiency resulted in a decrease in SYCP2 and DMC1 protein expression levels, meiosis in Lin28a deficiency male mice was not abnormal, possibly because unusual synapsis did not occur between nonhomologs in both Sycp2 and Dmc1 heterozygous meiotic nuclei.
GGAG sequences in the terminal loop of let-7 precursors serve as the binding sites of CCHC zinc finger domains of LIN28A, and are critical for LIN28A repressing maturation of let-7 microRNAs (Heo et al., 2009). Thus, we hypothesize that LIN28A acts as a suppressor of the biogenesis of target mRNAs. In support of the notion, transcript levels of meiotic target genes (Dmc1, Hormad1, Prdm9, Terb1 and Sycp2) significantly increased in Lin28a-null mouse testes (Fig. 5D), suggesting that LIN28A recognizes GGAG(A) sequence motifs in the 3′UTR to promote the degradation of its target mRNAs. However, LIN28A is also a positive regulator of translation of mRNAs such as Igf2 and Oct4 (Polesskaya et al., 2007; Qiu et al., 2010). Consistent with our previous observations, the expression of meiotic target genes decreased in polysome fractions of Lin28a-null testes (Fig. 6B), indicating that LIN28A can drive mRNA into polysomes and increase the efficiency of translation of target mRNAs. In addition, SYCP2 and DMC1 protein level changes were larger compared to their respective mRNA level changes (Fig. 6B,C), suggesting that the differences between the observed mRNA and protein level changes are most likely the result of impaired translation due to the deletion of LIN28A, which is consistent with the role of LIN28A as a translational enhancer of mRNA (Polesskaya et al., 2007). Furthermore, LIN28A is one of the critical regulators of reproductive function, regulating germ cell development, timing of sexual maturation and reproductive behavior throughout adult life (Aeckerle et al., 2012; Chakraborty et al., 2014; Shinoda et al., 2013a,b). In this study, based on the conditional reproductive system Lin28a-knockout mice, we further demonstrate that Lin28a cKO germ cell propagation is significantly inhibited (Fig. 5C). Our present study suggests that LIN28A performs the role of ‘translational regulator’ of meiotic genes during the maintenance of mouse germ cells (Fig. 6F).
Collectively, our results shed new light on the molecular mechanisms of LIN28A action, showing that it binds GGAG(A) motifs of target mRNAs at their 3′UTR and enhances the translation of these target mRNAs into proteins involved in meiosis. These findings may help us to better understand the critical reproductive function of LIN28A during male fertility.
MATERIALS AND METHODS
Germ cell isolation and culture
Germ cell cultures were established as previously described from 6 to 8 dpp mouse testes of C57BL/6 and C57BL/6×129S-Gt (Rosa) 26Sor/J mice (Jackson Laboratory) (Wang et al., 2017; Wu et al., 2009). Briefly, CD90.2-positive cells were enriched by magnetic cell separation with antibody-conjugated microbeads (Miltenyi Biotec, Bergisch Gladbach, Germany). Cultures were maintained on mitotically inactivated STO feeder layers in a serum-free medium (Kubota and Brinster, 2008) and at 37°C in an incubator with a humidified 5% CO2 and 95% air atmosphere. The medium was supplemented with 20 ng/ml recombinant GDNF (R&D Systems, Minneapolis, MN), 150 ng/ml GFRA1 (R&D Systems), and 1 ng/ml FGF2 (BD Biosciences, San Jose, CA). Germ cells were gently dislodged and collected using a pipetting method that yielded germ cell preparations of high (95%) purity for all experiments (Oatley et al., 2006).
All the procedures involving animals were approved by the Institutional Animal Care and Use Committee (IACUC) of Nanjing Medical University (ID: 2011082112). Lin28aFlox/Flox mice and Vasa-Cre mice were purchased from The Jackson Laboratory. Lin28aFlox/Flox mice were crossed with transgenic mice that expressed Cre recombinase to generate Lin28aFlox/−-Cre offspring. Genotyping for Lin28a floxed and Cre alleles were performed separately on genomic DNA that were isolated from tails using the following primers: Lin28aFlox/Flox or Lin28aFlox/−, 5′-TGGTGTTCCAGTGGCTTG-3′ (forward) and Lin28aFlox/Flox, 5′-GCAGCTGGTAAGAAGAAACCTG-3′ (reverse), Lin-28aFlox/−, 5′-TCAGCTCACACGTTGATCC-3′ (reverse); Vasa-Cre, 5′-CACGTGCAGCCGTTTAAGCCGCGT-3′ (forward) and 5′-TTCCCATTCTAAACAACACCCTGAA-3′ (reverse). PCR conditions were as follows: Lin28aFlox/Flox and Lin28aFlox/−: 98°C, 3 min, 98°C, 20 s, 60°C, 20 s, 68°C, 30 s, 35 cycles, 68°C, 5 min, 10°C, and then pause; Vasa-Cre: 95°C, 5 min, 95°C, 30 s, 58°C, 45 s, 72°C, 45 s, 35 cycles, 72°C, 5 min, 10°C, and then pause. All of the experiments were performed in accordance with the institutional guidelines of Nanjing Medical University.
RNA isolation, RT-PCR and RNA sequencing
All of the RNA extractions were performed with TRIzol reagent (Thermo Fisher Scientific, Shanghai, China), using cells or decapsulated testes, according to the manufacturer's instructions, unless otherwise specified. The appropriate RNA was reverse transcribed using a cDNA reverse transcription kit (Takara Bio, Otsu, Japan). Quantitative RT-PCR was performed using SYBR green master mix (Vazyme Biotech, Nanjing, China) and StepOne Plus (Applied Biosystems, Shanghai, China). RT-PCR conditions were as follows: 95°C, 5 min, 95°C, 10 s, 60°C, 30 s, 40 cycles, 95°C, 15 s, 60°C, 60 s, 95°C, 15 s, and then pause. Relative gene expression was analyzed based on the 2−ΔΔCt method with actin (Actb) as internal control. Primer sequences were as follows: Lin28a, 5′-CAACGTGCGCATGGGGTTCG-3′ (forward) and 5′-TTTTGGCCGCCGCTCACTCC-3′ (reverse); Dmc1, 5′-AGCAACTATGACCTTTCAGGC-3′ (forward) and 5′-TCATTTTCAGGCATCTCGGG-3′ (reverse); Ccnb3, 5′-ACCACCACTACTACCAAAAG-3′ (forward) and 5′-GAGTTGAAGATGAAGATGAAGATGC-3′ (reverse); Sycp1, 5′-AGTCGGGAAAACATTGATAAAGATC-3′ (forward) and 5′-ATACAGTCTGCTCATGGCTC-3′ (reverse); Sycp2, 5′-TTAATCTTGGGAGCCACACTC-3′ (forward) and 5′-TTCAAAGTTAGAGTCAGTAGCTTCT-3′ (reverse); Hormad1, 5′-AGCTTCCCTGAGTGCATTGG-3′ (forward) and 5′-TTGTCCCATAAGCACGTTCTG-3′ (reverse); Prdm9, 5′-CTGAATACAAGTGGCTCAGAACA-3′ (forward) and 5′-CCTCATAGGCAAGGCCCTTTC-3′ (reverse); Spo11, 5′-ATTCTGTCGGCCTTCGGATG-3′ (forward) and 5′-TTGCATAAGTGTCGCTCTGTATT-3′ (reverse); Terb1, 5′-ACTTCCAGGTTTGGAGGCAC-3′ (forward) and 5′-GTCTACAGCCCTTCGTCCTTT-3′ (reverse); Dpep3, 5′-CTGCTGGGGGTGTTACTTCTA-3′ (forward) and 5′-CGTGGCCGTGTCATTAGGAA-3′ (reverse). For RNA sequencing, the resulting cDNA libraries were sequenced using an Illumina HiSeq 2500 machine (San Diego, CA).
LIN28A HITS-CLIP was performed essentially as described previously (Vourekas et al., 2015). For each CLIP replicate, 15 pairs of testes from 10 dpp mice were detunicated, immediately UV crosslinked twice (400 mJ/cm2), after which the cell pellet was flash-frozen in liquid nitrogen and stored at −80°C. The UV crosslinked cells were lysed with 300 μl 1× PMPG buffer [1× PBS (no Mg2+ and no Ca2+) and 2% Empigen] with protease inhibitors (Roche, Mannheim, Germany) and 7 μl RNasin inhibitor (Promega, Madison, WI), treated with 10 μl RQ1 DNase (Promega) for 5 min at 37°C, and then centrifuged at 90,000 g for 30 min at 4°C. For each immunoprecipitation, the supernatant from the treated lysates was precleared by using rabbit IgG and 10 μg of anti-LIN28A rabbit monoclonal IgG using protein A Dynabeads (Thermo Fisher Scientific) for 3 h at 4°C. Antibody-bound beads were washed three times with 1× PMPG buffer and then incubated with lysates for 3 h at 4°C. RNA linkers (RL3 and RL5) were labeled with 32P and ligated to calf intestinal phosphatase (CIP)-treated RNA on beads as previously described (Vourekas et al., 2015). Immunoprecipitation beads were eluted using Novex reducing loading buffer, and then samples were separated by electrophoresis in NuPAGE gels (4%–12% gradient with MOPS buffer). Crosslinked RNA–protein complexes were transferred onto nitrocellulose membranes (Thermo Fisher Scientific), which were then exposed to film. Membrane fragments containing the main radioactive signal were excised. RNA extraction, 5′ linker ligation, RT-PCR, a second PCR, and extraction were performed as described previously (Vourekas et al., 2015). For deep sequencing, cDNA libraries were sequenced on an Illumina HiSeq 2500 machine.
Cultured germ cells were lysed with RIPA buffer (Beyotime, Shanghai, China) containing a protease inhibitor cocktail (Roche). Cell lysates were rotated at 4°C for 10 min and centrifuged at 21,130 g for 30 min at 4°C. Lysates were boiled at 95°C for 5 min in loading buffer (Beyotime) for degeneration. Primary antibodies were: anti-ZBTB16 (1:1000, R&D Systems), anti-LIN28A (1:2000, Abcam, Shanghai, China), anti-LIN28B (1:500, Cell Signaling Technology, Danvers, MA, USA), anti-β-tubulin (1:1000, Cell Signaling Technology), anti-SYCP2 (1:1000, ABclone, Wuhan, China), anti-DMC1 (1:1000, ABclone), anti-RPL6 (1:1000, ABclone). Secondary antibodies were purchased from Zhongshan Biotechnology Co, Beijing, China.
Germ cell transplantation
Transplantation was performed as previously described (Wu et al., 2012). Briefly, germ cells (GT-Rosa 26Sor/J) were transplanted into the testis of 129/SvCP×C57BL/6 male mice, in which endogenous spermatogenesis had been depleted by treatment with 55 mg/kg busulfan (Sigma-Aldrich) 8 weeks before transplant to abolish endogenous spermatogenesis. At 2 months after transplantation, the testes were harvested, and donor cell-derived colonies were identified using X-gal (Sigma-Aldrich) staining.
Lentivirus transfection knockdown
The pLKO.1 shRNA lentivirus vector and lentivirus packaging plasmids were provided by Dr Chen Dahua (State Key Laboratory of Reproductive Biology, Beijing, China). Lentivirus particles were generated by co-transfection of shRNA plasmids and packaging plasmids into HEK293T cells using CaCl2 transfection as described previously (Li et al., 2016). We designed two different shRNAs for Lin28a knockdown. The primer sequences were as follows: Lin28a-shRNA575, 5′-CCGGCCCAGTAAGAATGCAACTTAACTCGAGTTAAGTTGCATTCTTACTGGGTTTTTG-3′ (forward) and 5′-AATTCAAAAACCCAGTAAGAATGCAACTTAACTCGAGTTAAGTTGCATTCTTACTGGG-3′ (reverse); Lin28a-shRNA746, 5′-CCGGGAAGCGAAACAAGTGTCAAACCTCGAGGTTTGACACTTGTTTCGCTTCTTTTTG-3′ (forward) and 5′-AATTCAAAAAGAAGCGAAACAAGTGTCAAACCTCGAGGTTTGACACTTGTTTCGCTTC-3′ (reverse). A pLKO.1 empty vector was used as a negative control. Lentivirus-containing supernatant was collected 48 h after transfection, centrifuged at 1500 g for 10 min, filtered by a 0.45-μm filter membrane, and then stored at −80°C until use. For viral transduction, 300,000 germ cells were plated onto 12-well plates pre-coated with 0.1% gelatin (Sigma-Aldrich) and incubated with a 1:1 mixture of culture medium and viral supernatant, supplemented with 5 μg/ml polybrene. After 12 h transduction, cells were re-plated onto STO feeder layers and cultured in germ cell medium (see germ cell culture section). RNA was isolated 72 h after lentiviral transduction.
Dual-luciferase reporter assay
Various 3′UTR sequences from LIN28A targets and the Lin28a CDS were amplified by PCR. The 3′UTR of target mRNAs were inserted into psiCHECK-2 destination plasmid (Promega), and the Lin28a CDS were inserted into the pcDNA3.1 (+) destination plasmid (Thermo Fisher Scientific) using the ClonExpress MultiS One Step Cloning Kit (Vazyme Biotech). Mutant sites (Mut) of 3′UTR were generated by the Tsingke Biological Technology (Nanjing, China). Primer sequences for the Lin28a CDS and WT 3′UTRs were as follows: Lin28a CDS, 5′-TTGGTACCGAGCTCGGATCCATGGGCTCGGTGTCCAACC-3′ (forward) and 5′-CACACTGGACTAGTGGATCCTCAATTCTGGGCTTCTGGGAGC-3′ (reverse); Sycp2-3′UTR, 5′-ATTCTAGGCGATCGCTCGAGAATCTGGCTGCTGTCAATGAATTTTAT-3′ (forward) and 5′-AACGAATTCCCGGGCTCGAGCATTTGAAGGACATTTGTTTGGGCC-3′ (reverse); Hormad1-3′UTR, 5′-ATTCTAGGCGATCGCTCGAGCAAACAAGATGCCCTGGACTTGG-3′ (forward) and 5′-AACGAATTCCCGGGCTCGAGATGAGCAGAATCAATGACATTTGTATTTTT-3′ (reverse); Dmc1-3′UTR, 5′-ATTCTAGGCGATCGCTCGAGGTGGGGAATTGGTACAGACTGC-3′ (forward) and 5′-AACGAATTCCCGGGCTCGAGTTTAGTAGTAAATACTTTACTTTTTGGATTTTCATTT-3′ (reverse); Terb1-3′UTR, 5′-ATTCTAGGCGATCGCTCGAGCTGGACCTCTTTCGGTCCTTTG-3′ (forward) and 5′-AACGAATTCCCGGGCTCGAGCAATTAACAAAATGCCATCTGTTTATTTGTTGG-3′ (reverse); Prdm9-3′UTR, 5′-ATTCTAGGCGATCGCTCGAGTATATTTTCGAAAAGAATGAGAAAGCCA-3′ (forward) and 5′-AACGAATTCCCGGGCTCGAGTCAAGCTGTCACATTTTTATTATTCCCAGG-3′ (reverse); PCR conditions were as follows: 98°C, 2 min, 98°C, 30 s, 60°C, 30 s, 72°C, 90 s, 35 cycles, 72°C, 10 min, 10°C, and then pause. Wild-type and mutant plasmids were transfected, together with the Lin28a CDS into HEK293T cells simultaneously using CaCl2 transfection. After transfection for 36 h, the cells were collected and luciferase activities were assayed according to the manufacturer's instructions of the Dual-Luciferase Reporter Assay System (Promega).
For immunofluorescence, cultured germ cells were fixed with 4% paraformaldehyde (PFA) for 15 min and blocked with 1% bovine serum albumin (BSA, Sigma-Aldrich) containing 0.1% Triton X-100 in Dulbecco's PBS for 1 h at room temperature. The primary antibodies anti-ZBTB16 (1:100, AF2944, R&D Systems) and anti-LIN28A (1:100, ab46020, Abcam) were incubated for 2–3 h at room temperature, and cells were washed twice with Dulbecco's PBS. Secondary antibodies (bovine anti-goat-IgG conjugated to TRITC 1:1000 and donkey anti-rabbit-IgG conjugated to FITC 1:1000) were incubated for 1–2 h at room temperature. Samples were analyzed using confocal microscopy (LSM 700, Zeiss, Pleasanton, USA).
Chromosome spreads of spermatocytes were made as previously described (Kolas et al., 2005; Peters et al., 1997). Briefly, seminiferous tubules of the mouse testes were gently minced with tweezers in DMEM, and cells were mechanically separated. The cellular suspension was then spun to pellet the cellular debris, and the nuclear suspension was pipetted onto slides. The slides were then fixed for 3 min each in freshly prepared 2% PFA in PBS containing 0.03% SDS, and in 2% PFA alone. The slides were rinsed twice for 1 min each in 0.4% PHOTO-FLO 200 solution (Eastman Kodak Company, Rochester, NY), dried, and then blocked in TBST containing 10% goat serum. The slides were then incubated with primary antibodies for 1 h at 37°C or overnight at 4°C. Primary antibodies used for immunofluorescence were as follows: rabbit anti-SYCP1 antibody (1:100, ab15090, Abcam), mouse anti-SYCP3 antibody (1:100, ab97672, Abcam).
Sucrose density gradient fractionation
Testes lysates were prepared essentially as described previously (Li et al., 2016). For each replicate, five pairs of decapsulated testes from 10 dpp mice were homogenized in MBC buffer. Lysate was kept on ice for 15 min and centrifuged at 10,000 g for 10 min. The supernatant was loaded on 20%–50% (w/v) linear density sucrose gradient (Gradient Master, Biocomp, Fredericton, NB, Canada) and centrifuged for 3 h at 180,000 g in a SW41 rotor (Beckman Coulter Optima L-100 XP Ultracentrifuge, Brea, CA, USA), followed by collection of RNP, 40S to 80S ribosome and polysome fractions (0.5 ml) using a piston gradient fractionator (Biocomp). For quantitative RT-PCR analysis of mRNA species, total RNAs were isolated from 200 μl of each fraction. The efficiency of polysome separation was verified by western blot analysis of individual fractions using antibodies directed against RPL6 (ABclone, Wuhan, China) and tubulin (Cell Signaling Technology).
Data processing and statistical analysis
Image data were assembled using Adobe Photoshop 7.0 software (Adobe, San Jose, CA). All of the data were reported as means±s.d. unless otherwise noted in the figure legends. Sequencing, gene expression, transfection, and transplantation experiments were performed in duplicate or triplicate using independently established germ cell cultures. Significance was tested by using unpaired Student's t-test using Excel (Microsoft, Redmond, WA) or Prism software (GraphPad Software, La Jolla, CA). A difference was considered significant when the P-value was <0.05.
Conceptualization: X.W.; Methodology: Mei Wang, F.Y., Min Wang, L.L., X.W.; Validation: X.W.; Formal analysis: X.W.; Investigation: Mei Wang, L.Y., F.Y., Min Wang, L.L., X.W.; Data curation: S.W., Min Wang; Writing - original draft: Mei Wang, X.W.; Writing - review & editing: X.W.; Visualization: Mei Wang, L.Y., S.W., L.L., X.W.; Supervision: X.W.; Project administration: X.W.; Funding acquisition: Mei Wang, L.L., X.W.
This work was supported by the National Key R&D Program of China (2018YFC1003302), the National Natural Science Foundation of China (grant 31872844 to X.W., grant 31601195 to L.L.), Natural Science Foundation of Jiangsu Province for Youth (BK20150991 to L.L.), and the Innovative and Entrepreneurial Program of Jiangsu Province to Mei Wang.
All of the RNA sequencing and high-throughput sequencing data reported in this paper are accessible through the NCBI Gene Expression Omnibus (GEO) accession number GSE134033.
Peer review history
The peer review history is available online at https://jcs.biologists.org/lookup/doi/10.1242/jcs.242701.reviewer-comments.pdf
The authors declare no competing or financial interests.