Dual regulatory actions of LncBMP4 on BMP4 promote chicken primordial germ cell formation

Abstract

The unique characteristics of chicken primordial germ cells (PGCs) provide potential strategies for transgenic animal generation; however, insufficient PGC availability has limited their application. Regulation of bone morphogenic protein 4 (BMP4), a crucial factor for PGCs formation, may provide new strategies for PGC generation. We here identify a long noncoding RNA (lncRNA) that targets BMP4 (LncBMP4). LncBMP4 has similar functions as BMP4, in that it facilitates the formation and migration of PGCs. LncBMP4 promotes BMP4 expression by adsorbing the miRNA gga‐mir‐12211, thus reducing its inhibitory effect on BMP4 expression. In addition, the small peptide EPC5 encoded by LncBMP4 promotes the transcription of BMP4. The competing endogenous RNA (ceRNA) effect of LncBMP4 requires N6‐methyladenosine (m6A) modification, in a dose‐dependent manner, and high levels of m6A modification hinder EPC5 translation. Understanding the molecular mechanisms through which LncBMP4 promotes BMP4 expression during PGC formation may provide new avenues for efficient PGC generation.

The lncRNA lncBMP4 binds gga‐miR‐12211, which increases the expression of BMP4 and promotes PGC formation. LncBMP4 also encodes a small peptide, EPC5, that promotes PGC formation, and both activities are modified by lncBMP4 m6A levels.

Introduction

Primordial germ cells (PGCs) are a type of reproductive stem cell with both pluripotent and reproductive characteristics and are the precursor cells for gametes, able to transmit genetic information within the same generation (Saitou & Yamaji, 2012; Tagami et al, 2017). The unique migration characteristics of chicken PGCs through blood circulation provides a promising strategy for the production of transgenic chicken models and the conservation of poultry species resources; however, this approach has been limited by an insufficient number of PGC sources and uncertainty regarding the molecular mechanisms that regulate PGC formation (De Melo et al, 2012). Studies have shown that BMP4, which encodes bone morphogenic protein 4 (BMP4), is an important factor during PGC origination (Neave et al, 1997; Lawson et al, 1999), and altered BMP4 expression levels, both in vivo and in vitro, can significantly affect the formation efficiency of PGCs (Winnier et al, 1995; Shi et al, 2014). Therefore, the exploration of factors that affect BMP4 expression is crucial to addressing the shortage of PGC sources. Currently, two basic theories exist regarding the formation of PGCs: preformation and epigenesis (Extavour & Akam, 2003). Some cells in the mammalian mesoderm can be induced to form PGCs through specific signals, such as BMP4, which support the epigenetics theory (Extavour & Akam, 2003), and many studies have suggested that the formation mechanism of chicken PGCs occurs through similar mechanisms as mammalian PGCs, which are also induced by BMP4 (Winnier et al, 1995; Shi et al, 2014). However, the factors that regulate BMP4 have not been well‐studied. Michael et al found that the transcription factor fibroblast growth factor receptor 3 (FGR3) could inhibit BMP4 expression during osteoblast development, but this regulatory pathway did not exist during the formation of chicken PGCs (Naski et al, 1998). Therefore, we must further explore the factors that regulate BMP4 expression from other perspectives, to solve the problem of insufficient PGC sources.

Long, non‐coding RNAs (LncRNAs) are epigenetic factors that regulate gene expression (Djebali et al, 2012). Studies have shown that LncRNAs can change the binding characteristics of target gene promoters and RNA polymerase and can bind proteins, changing the conformation or subcellular localization of modifying enzymes to regulate gene expression (Willingham et al, 2005; Hirota et al, 2008). Studies have also shown that LncRNAs can regulate mRNA expression through the adsorption of micro‐RNAs (miRNAs), acting in a competing endogenous RNA (ceRNA) system (Fan et al, 2013). Moreover, LncRNAs can also regulate gene expression by encoding micropeptides (Huang et al, 2017; Matsumoto et al, 2017; Jiang et al, 2018). Recently, numerous studies have demonstrated that LncRNAs play indispensable roles during germ cell differentiation. Magnusdottir found that 313 of the 5,046 Blimp1/Prdm1 binding sites in mouse PGCs were associated with LncRNAs (Erna & M Azim, 2014). Li et al used high‐throughput sequencing to examine the expression profiles of mRNAs, LncRNAs, and circular RNAs (circRNAs) in mouse germ stem cells and found that the LncRNA MEG3 and the circRNA Igf1r competitively bound the miRNA miRNA‐15a‐5p, which increased the expression of target genes (Inha, Acsl3, Kif21b, and Igfbp2), promoting the differentiation of germ stem cells (Li et al, 2017). In a previous study, we used high‐throughput sequencing to analyze the expression of LncRNAs in chicken germ cells, identifying 7 LncRNAs that target BMP4. However, the functions of these LncRNAs and the molecular mechanisms through which they regulate BMP4 expression must still be systematically studied.

N6‐methyladenosine (m6A) represents one of the most common epigenetic modifications made to mRNA and non‐coding RNA (ncRNA) through the specific recognition motif RRACH, affecting RNA translation and stability (Dominissini et al, 2012; Chen et al, 2015). Studies have shown that m6A is involved in a variety of cell differentiation and developmental processes. Many master genes required for embryonic stem cell (ESC) maintenance and differentiation are modified by m6A, such as Nanog, Sox2, Nr5a2, Eomes, and FoxA2 (Young, 2011; Dunn et al, 2014). Batista showed that Mettl3 knockout could reduce m6A levels, which upregulated the expression of pluripotent genes, promoted the self‐renewal of ESCs, and maintained their pluripotent state (Batista et al, 2014). Aguilo also showed that ZFP217 expression could promote the reprogramming of somatic cells by preventing m6A methylation. To date, no studies have examined the effects of m6A during the development of chicken PGCs (Aguilo et al, 2015). However, our analysis found a large number of RRACH sites on the LncRNA TCONS_00874170, which targets BMP4, suggesting that m6A may be involved in the LncRNA‐mediated regulation of BMP4 expression, which remains to be systematically explored.

Based on these findings, LncBMP4, a LncRNA that was identified as a regulator of BMP4 expression, was explored using a multi‐group analysis in the present study. LncBMP4 was found to be involved in the origin and migration of PGCs, both in vivo and in vitro, demonstrating a similar function as that for BMP4. Further experiments found that m6A regulated the binding of LncBMP4 to the miRNA gga‐mir‐12211, in a dose‐dependent manner, to upregulate the expression of BMP4. LncBMP4 also upregulated the transcriptional activity of BMP4 through the LncBMP4‐encoded peptide EPC5, which could be inhibited by m6A modification. In conclusion, our study systematically explored the factors affecting the BMP4 expression, clarifying the regulatory network of epigenetic modifications that act on the BMP4 regulation of PGCs and providing a new strategy for approaching the issue of PGCs insufficiency.

Results

LncBMP4 regulates BMP4 during the formation of chicken PGCs

BMP4 is a crucial factor involved in PGCs origination; however, the factors that regulate BMP4 expression have not been explored in detail. In this study, to dissect the key regulatory factors for BMP4 during PGC formation, chromatin immunoprecipitation (ChIP) data regarding DNA methylation (He et al, 2018a; Data ref: He et al, 2018b), the dimethylation of lysine 4 on histone 3 (H3K4me2) (Zhang et al, 2021a; Data ref: Zhang et al, 2021b) and the LncRNA sequencing results (Gao et al, 2020a; Data ref: Gao et al, 2020b) during germ cell formation were collected for the chicken model [in ESCs, PGCs, and spermatogonial stem cells (SSCs)]. The results showed that the enrichment of DNA methylation and H3K4me2 in the promoter region of BMP4 had no effects, which means DNA methylation and H3K4Me2 may not be the main regulatory factors for BMP4 expression (Appendix Fig S1A). Interestingly, 7 LncRNAs located within 100 kb upstream and downstream of the BMP4 gene locus were identified based on cis prediction models, the expression patterns of which were consistent with that of BMP4, with significant expression levels in PGCs (Fig 1A and B; Table EV1). These results preliminarily indicated that these LncRNAs may act as the primary regulatory factors for BMP4.

Figure 1. LncBMP4 acts as a regulatory factor for BMP4 during the formation of chicken PGCs.

1. Seven LncRNAs targeted for BMP4 were identified by RNA‐seq, and the relative positions of these 7 LncRNAs and BMP4 are shown.
2. FPKM of the 7 LncRNAs identified in the RNA‐seq with high expression in PGCs. The solid lines at each end of the box plot represent the maximum and minimum values, respectively. The three dotted lines in the middle of the violin diagram represent the 75% percentile, the mean, and the 25% percentile in turn.
3. LncRNA classification. According to the positions of the LncRNA and the targeted mRNA, LncRNA‐mRNA pairs were divided into 12 biotypes (see methods and Appendix Fig S1B for details), and the 7 LncRNAs were distributed in the XT, SU, and SD biotypes.
4. GO analysis showed that the LncRNA‐mRNA in the XT biotype was primarily involved in cell development and differentiation, the LncRNA‐mRNA in the SU biotype was primarily involved in cell migration and cell development, and the LncRNA‐mRNA in the SD biotype was primarily involved in gonadal development and cell migration.
5. Correlation analysis between BMP4 and the SD biotype LncRNA showed a significant correlation between TCONS_00874170 (LncBMP4) expression and BMP4 expression (see Appendix Fig S1C for details).
6. The existence of LncBMP4 was confirmed by northern blot, the results showed that LncBMP4 was significantly expressed in PGCs, but weakly expressed in CEF and ESCs. H₂O was used as a negative control (n = 3 independent experiments).

Source data are available online for this figure.

Because the function of a LncRNA is associated with the relative position of the targeted mRNA (Sigova et al, 2013), we divided the LncRNA/mRNA combinations into 12 locus biotypes according to the relative positioning of the LncRNA and its target mRNA (Sigova et al, 2013; Luo et al, 2016) (Fig 1C; Appendix Fig S1B), including LncRNAs transcribed in an antisense direction (designated “X”) and LncRNAs transcribed in the same direction as the nearest gene (designated “S”). The 7 LncRNAs for BMP4 were distributed in the XT, SU, and SD biotypes (Fig 1C; Table EV1; Datasets [Link], [Link], [Link]). Gene ontology (GO) analysis of the target genes showed that the SD biotype was significantly involved in the regulation of cell development (especially cell migration, Fig 1D). From this biotype, TCONS_00874170 was chosen as the primary regulatory factor for BMP4, based on correlation analysis, and named LncBMP4 (Fig 1E; Appendix Fig S1C).

Then, we confirmed the existence and expression trends for LncBMP4 in ESCs and PGCs, using quantitative reverse transcription polymerase chain reaction (qRT–PCR) and northern blot analyses (Fig 1F; Appendix Fig S1D), the results of which were consistent with the results of high‐throughput sequencing. Further expression profiling performed in different tissues indicated that LncBMP4 was highly expressed in the gonads (Appendix Fig S1E). The subcellular localization results showed that LncBMP4 was expressed in both the nucleus and the cytoplasm of PGCs, but the nuclear expression level was significantly higher than that in the cytoplasm (Appendix Fig S1F). Together, these results suggested that LncBMP4 may play an important role in germ cell formation and act as an important factor for the regulation of BMP4 expression.

LncBMP4 promotes the origination of PGCs by targeting BMP4 in vitro

To study the function of LncBMP4 during the formation of PGCs, we constructed active overexpression (oe) and interference [short hairpin (sh)] vectors for LncBMP4, named oe‐LncBMP4 and sh‐LncBMP4, respectively (Appendix Fig S2A–C). Oe‐LncBMP4 and sh‐LncBMP4 were transfected into ESCs in vitro, and the effects of LncBMP4 on PGCs formation were explored using the BMP4 induction system (Shi et al, 2014; Fig 2A–C). First, the expression levels of LncBMP4 and BMP4 were determined, throughout the entire induction process, which showed a consistent expression pattern between LncBMP4 and BMP4 during the induction process. The expression of BMP4 was downregulated in the presence of sh‐LncBMP4, whereas BMP4 expression was upregulated in the presence of oe‐LncBMP4 (Fig 2B).

Figure 2. LncBMP4 promotes the origination of PGCs by targeting BMP4, in vitro .

1. Up: The schematic diagram of PGCs induced from chicken ESCs in vitro. Down: Morphological observation of the number of embryoid bodies in the in vitro PGCs induction model after LncBMP4 overexpression and interference; oeLncBMP4 indicates the overexpression of LncBMP4; shLncBMP4 indicates the interference of LncBMP4; BMP4 induction was regarded as the control. Scale bar: 60 µm (n = 3 independent experiments).
2. The expression of reproductive marker genes (Cvh and C‐kit), LncBMP4, and BMP4 were detected by qRT–PCR after LncBMP4 overexpression and interference, in vitro (data are shown as mean ± SEM, n = 3 independent experiments, *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001, one‐way ANOVA).
3. The efficiency of PGCs differentiation was detected by IF 4 days after induction, using C‐KIT and CVH antibodies. Scale bar: 60 µm (n = 3 independent experiments).
4. Detection of PGCs formation efficiency after LncBMP4 overexpression or interference, in vitro, by flow cytometry using the CVH antibody (data are shown as mean ± SEM, n = 3 independent experiments, ****P < 0.0001, one‐way ANOVA. Gating strategies and Blank were supplied in Appendix Fig S14A–D).

Source data are available online for this figure.

During the entire induction process, the results of cell morphological observations showed that embryoid bodies (EBs) appeared on day 4 following BMP4 induction (Fig 2A), and the numbers and sizes of EBs increased by day 6. After transfection with sh‐LncBMP4, no EBs appeared, throughout the entire process. However, the number of EBs that appeared 4 and 6 days after transfection with oe‐LncBMP4 was significantly higher than that in control cells following the BMP4 induction process (Fig 2A; Appendix Fig S2D). The expression levels of Cvh and C‐kit during the induction process were detected by qRT–PCR, which showed that Cvh and C‐kit expression significantly increased (P < 0.01) following LncBMP4 overexpression in the BMP4 induction system, whereas LncBMP4 interference resulted in the opposite effect. These results preliminarily indicated that LncBMP4 could promote the origination of PGCs, in vitro (Fig 2B). To clarify the effects of LncBMP4 on the origination of PGCs, immunofluorescence (IF) and flow cytometry (FC) analyses were performed on the cells collected 4 days after induction (Fig 2C and D; Gating strategies is shown in Appendix Fig S14). The results showed that, compared with normal BMP4 induction, the numbers of CVH⁺C‐KIT⁺ cells significantly decreased after LncBMP4 interference. However, after LncBMP4 overexpression, the numbers of CVH⁺C‐KIT⁺ cells significantly increased (Fig 2C and D). Similar results were observed on day 6 of induction (Appendix Fig S3A). We also collected the cells from different induction groups for PAS staining on 4 days, which could dye PGCs red, while ESCs were not. The results showed that the positive cells in BMP4 induction group and BMP4+oeLncBMP4 group were significantly higher than those in BMP4+shLncBMP4 group, which was consistent with the results of IF and FC (Appendix Fig S3B). These results indicated that LncBMP4 is an important factor during PGC formation.

LncBMP4 regulates the migration of PGCs in vivo

To further clarify the function of LncBMP4 during the formation of PGCs, oe‐LncBMP4 and sh‐LncBMP4 were injected into the blood vessels of chicken embryos (2.5 days), which were then incubated to 4.5 days. The qRT–PCR results showed that the expression levels of Cvh, C‐kit, and BMP4 significantly increased after LncBMP4 overexpression (P < 0.01), whereas these expression levels were significantly downregulated after LncBMP4 interference (P < 0.01, Appendix Fig S4A). Chicken embryos that were subjected to different treatments were collected at 4.5 days, and paraffin sections were prepared. Immunohistochemistry (IHC) assays using a DEAD‐box helicase 4 (DDX4) antibody showed that the number of PGCs in the genital ridge was significantly reduced after LncBMP4 interference compared with the normal incubation process, whereas the number of PGCs in the genital ridge significantly increased after LncBMP4 overexpression (Fig 3A). The PGCs in the collected genital ridge were counted by FC, based on DDX4 antibody staining, which showed that the number of PGCs was significantly reduced (P < 0.01) after LncBMP4 interference and significantly increased after LncBMP4 overexpression (P < 0.01, Fig 3B). PGCs have begun to originate in blastocysts by the 2.5‐days time point used for embryo injection; therefore, changes in the numbers of PGCs in the genital ridge are likely the result of changes in the migration and proliferation of PGCs.

Figure 3. LncBMP4 regulates the migration of PGCs in vivo .

1. The results of IHC examining using anti‐DDX4 antibody shows that the overexpression of LncBMP4 increases the number of PGCs in the genital ridge, and interference with LncBMP4 reduces the number of PGCs in the genital ridge. The number of PGCs was counted through the Segmentation module of Image J software. Scale bars: 200 μm (top row), 40 μm (bottom row) (data are shown as mean ± SEM, n = 3 independent experiments, *P < 0.05, **P < 0.01, ****P < 0.0001, one‐way ANOVA).
2. The number of PGCs in the genital ridge, following the overexpression or interference of LncBMP4, as determined by flow cytometry (data are shown as mean ± SEM, n = 3 independent experiments, ****P < 0.0001, one‐way ANOVA).
3. The number of PGCs labeled with PkH26 (red) that migrated to the genital ridge in a frozen section, the number of PGCs labeled with PkH26 (red) was counted using ImageJ software. Scale bar: 70 µm (data are shown as mean ± SEM, n = 6 independent experiments, **P < 0.01, ****P < 0.0001, one‐way ANOVA.).
4. The effect of overexpression and interference of LncBMP4 on the proliferation of PGCs was detected by EDU proliferation assay. Scale bar: 100 µm (n = 3 independent experiments).

Source data are available online for this figure.

We detected the proliferation capacity of PGCs transfected with oe‐LncBMP4 and sh‐LncBMP4 using a cell counting kit 8 (CCK8). The results showed that neither the loss nor the overexpression of LncBMP4 affected PGCs proliferation in vitro (Appendix Fig S4B). This result was also confirmed by EDU test (Fig 3D). PGCs transfected with oe‐LncBMP4 and sh‐LncBMP4 were also labeled with the red fluorescent dye PKH26 and injected into chicken embryonic blood vessels. The genital ridge was collected at 4.5 days, and frozen sections were examined to observe the migration of red‐labeled PGCs in the genital ridge under different LncBMP4 expression conditions. The results showed that LncBMP4 overexpression promoted the migration of PGCs, whereas LncBMP4 interference inhibited the migration of PGCs (Fig 3C; Appendix Fig S4C). Combined with the in vitro results, we concluded that LncBMP4 is not only involved in the origination of early PGCs but also regulates PGCs migration.

gga‐mir‐12211 regulates the formation of PGCs by targeting LncBMP4 and BMP4

The interactions that occur between LncRNAs and miRNAs represent a classical model for the regulation of target gene expression. Therefore, the study used DIANA TOOLS to identify two miRNAs, gga‐mir‐12211 and gga‐mir‐12258, which were both predicted to target LncBMP4 and BMP4 (Table EV2; Fig 4A and B). To study the functions of gga‐mir‐12211 and gga‐mir‐12258 during the formation of PGCs, active overexpression and interference expression vectors were constructed for gga‐mir‐12211 and gga‐mir‐12258 (Appendix Fig S5A and B). ESCs transfected with the overexpression and interference expression vectors for gga‐mir‐12211 and gga‐mir‐12258 were induced by BMP4. Cell morphological observations showed no EBs on day 2 or day 4 after gga‐mir‐12211 overexpression, and only a small amount of EBs was observed on day 6. After gga‐mir‐12211 interference, however, a large number of EBs appeared on day 4, which increased by day 6. In contrast, the overexpression and interference of gga‐mir‐12258 had no significant effects on the number of EBs produced during the BMP4 induction process (Fig 4C; Appendix Fig S5C). These results preliminarily indicated that gga‐mir‐12211 inhibited the formation of PGCs in vitro, whereas gga‐mir‐12258 was not involved in the formation of PGCs. Gene expression analysis showed similar results. Compared with the normal BMP4 induction process, the expression levels of LncBMP4 and BMP4 significantly increased over time after gga‐mir‐12211 interference (P < 0.01), and the expression levels of the PGCs marker genes Cvh and C‐kit were significantly upregulated. Following gga‐mir‐12211 overexpression, the complete opposite experimental results were observed (Fig 4D). However, neither gga‐mir‐12258 interference nor overexpression significantly affected the expression of any of the examined genes (Appendix Fig S5D). Further IF and FC analyses also indicated that gga‐mir‐12211 could inhibit the formation of PGCs, whereas gga‐mir‐12258 was not involved in the formation of PGCs in vitro (Fig 4E and F; Appendix Fig S5D and E).

Figure 4. gga‐mir‐12211 regulates the formation of PGCs by targeting LncBMP4 and BMP4 .

• A, B
  Screening of miRNAs that target both BMP4‐3'UTR and LncBMP4. (A) Two miRNAs (gga‐mir‐12211 and gga‐mir‐12258) simultaneously target BMP4‐3'UTR and LncBMP4. (B) Schematic diagram of base sequence and structure of gga‐mir‐12211 and gga‐mir‐12258.
• C
  Morphological observation of the number of embryoid bodies in the PGC in vitro induction model after the overexpression and interference of gga‐mir‐12211. BMP4 induction was regarded as control. Scale bar: 60 µm (n = 3 independent experiments).
• D
  The expression of reproductive marker genes (Cvh and C‐kit), LncBMP4, and BMP4 were detected by qRT–PCR in the genital ridge after gga‐mir‐12211 overexpression and interference in vitro (data are shown as mean ± SEM, n = 3 independent experiments, *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001, one‐way ANOVA).
• E
  The efficiency of PGC differentiation was detected by IF, 4 days after induction, using C‐KIT and CVH antibodies. Scale bar: 60 µm (n = 3 independent experiments).
• F
  Detection of PGC formation efficiency after the overexpression or interference of gga‐mir‐12211, in vitro, as assessed by flow cytometry using the CVH antibody (data are shown as mean ± SEM, n = 3 independent experiments, **P < 0.01, ***P < 0.001, ****P < 0.0001, one‐way ANOVA.).

Source data are available online for this figure.

m6A modification promotes the ceRNA function of LncBMP4 to regulate BMP4 expression

Next, we systematically explored the regulatory relationships among LncBMP4, BMP4, and gga‐mir‐12211. Functional study results showed that the function of gga‐mir‐12211 was opposite those of LncBMP4 and BMP4 (Figs 2, 3, 4) and that gga‐mir‐12211 overexpression could inhibit the expression of LncBMP4 and BMP4 (Fig 4E). Combined with the bioinformatics prediction results (Fig 3A), gga‐mir‐12211 was preliminarily confirmed to simultaneously target LncBMP4 and BMP4 expression. The results of a dual‐luciferase reporting assay showed that overexpression of gga‐mir‐12211 could significantly reduce the activity of the BMP4 3’ untranslated region (BMP4‐3’UTR) (Fig 5A), and gga‐mir‐12211 interference was able to significantly rescue BMP4‐3'UTR activity, which indicated that gga‐mir‐12211 could bind to the BMP4‐3'UTR and inhibit the expression of BMP4. The qRT–PCR results showed that gga‐mir‐12211 overexpression could significantly reduce LncBMP4 expression (P < 0.01), indicating that gga‐mir‐12211 may bind to LncBMP4 and inhibit LncBMP4 expression (Fig 5B). To verify the existence of a ceRNA system among LncBMP4, BMP4, and gga‐mir‐12211, DF‐1 cells were co‐transfected with BMP4‐3’UTR, oe‐gga‐mir‐12211, and oe‐LncBMP4. The results showed that gga‐mir‐12211 overexpression significantly reduced BMP4‐3’UTR activity (P < 0.01), but LncBMP4 overexpression was able to significantly rescue BMP4‐3’UTR activity (P < 0.01, Fig 5C). In order to further study the binding relationship between gga‐mir‐12211 to LncBMP4 and BMP4‐3'UTR, we fused the mir‐12211 sequence and the AGO coding sequence into pcDNA3.0 to construct the pcDNA3.0‐mir‐12211‐AGO vector. After the vector was transfected into PGCs, RIP experiment was carried out with AGO antibody. The results showed that RNA enriched in AGO contained LncBMP4 and BMP4‐3'UTR (Appendix Fig S6). Together, these results indicated that LncBMP4 could competitively bind to gga‐mir‐12211 to rescue the expression of BMP4.

Figure 5. m6A modification promotes LncBMP4 as a ceRNA, to regulate the expression of BMP4 .

• A
  A luciferase reporting system was used to detect the regulatory effects of gga‐mir‐12211 on the activity of BMP4‐3'UTR (data are shown as mean ± SEM, n = 3 independent experiments, *P < 0.05, ****P < 0.0001, one‐way ANOVA).
• B
  qRT–PCR was used to detect the regulatory effects of gga‐mir‐12211 on LncBMP4 expression (data are shown as mean ± SEM, n = 3 independent experiments, ***P < 0.001, ****P < 0.0001, one‐way ANOVA).
• C
  The results of the luciferase reporting system showed that LncBMP4 could rescue the gga‐mir‐12211 inhibitory effects against BMP4‐3'UTR (data are shown as mean ± SEM, n = 3 independent experiments, *P < 0.05, ****P < 0.0001, one‐way ANOVA).
• D
  Distribution diagram of RRACH sites on LncBMP4, BMP4‐3'UTR, and gga‐mir‐12211.
• E, F
  Enrichment levels of m6A in LncBMP4 (E) and BMP4‐3'UTR (F) were detected by RIP‐qPCR (data are shown as mean ± SEM, n = 3 independent experiments, ****P < 0.0001, one‐way ANOVA).
• G
  The effects of MA and NPC on m6A enrichment levels in the P2 region of LncBMP4 were detected by RIP‐qPCR (data are shown as mean ± SEM, n = 3 independent experiments, ***P < 0.001, ****P < 0.0001, one‐way ANOVA.).
• H, I
  The effects of m6A modifications of LncBMP4 on the ceRNA system were detected by the luciferase reporting system (data are shown as mean ± SEM, n = 3 independent experiments, *P < 0.05, ****P < 0.0001, one‐way ANOVA.).

Source data are available online for this figure.

Studies have shown that m6A is a key factor that affects the ceRNA system (Yang et al, 2018). Recently, a study used MeRIP‐Seq to map the whole transcriptome landscape of m6A in chickens (GEO Accession: GSE166240), which analyzed the dynamics of m6A before and after CEF infection with Marek's disease virus (MDV; Sun et al, 2021a; Data ref: Sun et al, 2021b). The genome matching rate of the sequencing results was above 80% (Table EV3), which proved that the sequencing results were authentic and reliable. From the results, we found that the transcript of LncBMP4 in this study also had dynamic changes of m6A modification (Appendix Fig S7A), which preliminarily proved that LncBMP4 could be modified by m6A. To explore the molecular mechanisms through which m6A regulates the ceRNA system, we analyzed the distribution of RRACH modification sites on gga‐mir‐12211, LncBMP4, and BMP4‐3’UTR. The results identified 16 m6A modification sites on the LncBMP4 sequence, 3 sites on the BMP4‐3'UTR sequence, and no sites on gga‐mir‐12211, which indicated that m6A may regulate the ceRNA system by modifying BMP4‐3’UTR and LncBMP4 (Fig 5D). RNA immunoprecipitation chip (RIP) results showed that m6A enrichment on LncBMP4 at sites 6–8 was upregulated during the formation of PGCs, but no significant differences in m6A enrichment were observed for BMP4‐3’UTR (Fig 5E and F), which indicated that m6A could only modify LncBMP4 to regulate the ceRNA system. Therefore, the effects of the m6A methylation activator (meclofenamic acid, MA; Huang et al, 2019) and the m6A methylation inhibitor (Neplanocin A, NPC; Uddin et al, 2019) on the ceRNA system were explored using the luciferase reporting system and RIP–qPCR (Fig 5G and H). The addition of NPC significantly decreased BMP4‐3’UTR activity (P < 0.01), whereas the addition of MA significantly increased BMP4‐3’UTR activity (P < 0.01). However, m6A methylation cannot affect BMP4‐3’UTR independently (Fig 5I). Importantly, After PGCs were treated with STM2457, a specific inhibitor of METTL3 at 10 μM concentration (Yankova et al, 2021), it was found that the enrichment level of m6A in the fragment corresponding to P2 primer (m6A sites 6–8 of LncBMP4) in PGCs was significantly downregulated (Appendix Fig S7B), but there was no significant change in other fragments (P1, P3, and P4). All these results indicated that m6A can participate in the ceRNA system.

m6A regulates ceRNA in a dose‐dependent manner at the RRACH site of LncBMP4

To explore the molecular mechanisms associated with the m6A modification of LncBMP4 and its effects on the ceRNA system, we proposed two hypotheses: first, that m6A modification affects LncBMP4 stability; and second, that m6A modification affects the binding between LncBMP4 and gga‐mir‐12211. To test these hypotheses, we added MA and NPC to DF‐1 cells overexpressing LncBMP4. The qRT–PCR results showed that neither NPC nor MA affected LncBMP4 transcription levels (Fig 6A). Similar results were obtained by interfering with the expression of the m6A modifying enzymes Mettl3 and Mettl14, interference with which could reduce the modification level of m6A at sites 6–8 on LncBMP4 (Appendix Fig S8A–D), which indicated that m6A modification did not affect the stability of LncBMP4. To explore the effect of m6A modifications on the ability of LncBMP4 to bind with gga‐mir‐12211 based on the prediction of biological information, we found that the binding site of gga‐mir‐12211 was located between 6–8 m6A sites on LncBMP4, which did not overlap with m6A modification sites (Appendix Fig S9). Therefore, we speculated that m6A modification on LncBMP4 could change the binding ability of LncBMP4 and gga‐mir‐12211 by changing the spatial structure of LncBMP4. To confirm this speculation, a luciferase reporter vector was constructed for LncBMP4 (Luc‐LncBMP4), which was co‐transfected into DF‐1 with oe‐gga‐mir‐12211. Luciferase activity was detected after the addition of both NPC and MA, and the results showed that the fluorescence activity associated with Luc‐LncBMP4 decreased significantly after the addition of MA, whereas the luciferase activity of Luc‐LncBMP4 increased significantly after the addition of NPC (Fig 6B and C), which is consistent with the results following Mettl3 and Mettl14 interference (Appendix Fig S8D–F). These results indicated that high levels of m6A modification can promote the binding between LncBMP4 and gga‐mir‐12211.

Figure 6. m6A regulates ceRNA, in a dose‐dependent manner, at the RRACH sites of LncBMP4 .

1. The influence of m6A on the stability of LncBMP4 was detected by qRT–PCR (data are shown as mean ± SEM, n = 3 independent experiments, ****P < 0.0001, one‐way ANOVA).
2. The effects of m6A modification on the binding ability of LncBMP4 with gga‐mir‐12211 were detected using a luciferase reporting system (data are shown as mean ± SEM, n = 3 independent experiments, *P < 0.05, ****P < 0.0001, one‐way ANOVA).
3. Effect of NPC and MA treatment on the activity of Luc‐LncBMP4 was detected using a luciferase reporting system (data are shown as mean ± SEM, n = 3 independent experiments, ****P < 0.0001, one‐way ANOVA).
4. The enrichment level of m6A, for LncBMP4 variants with the progressive accumulation of RRACH site mutations, was detected by RIP‐qPCR (data are shown as mean ± SEM, n = 3 independent experiments, *P < 0.05, **P < 0.01, one‐way ANOVA).
5. The ability of LncBMP4, modified with different levels of m6A, to bind gga‐mir‐12211 was detected by the luciferase reporting system (data are shown as mean ± SEM, n = 3 independent experiments, **P < 0.01, ****P < 0.0001, one‐way ANOVA.).

Source data are available online for this figure.

During previous analyses, we identified 16 m6A modification sites on LncBMP4 (Fig 5D). To investigate the specific effects of these loci on the binding between LncBMP4 and gga‐mir‐12211, we performed the progressive mutation of RRACH regions in the Luc‐LncBMP4 vector and transfected these mutated vectors into DF‐1 cells. The RIP results showed no significant changes in the m6A enrichment levels with the progressive accumulation of mutations to m6A modification sites 1–5; however, significant decreases in LncBMP4 m6A enrichment levels were observed when modification sites 6, 7, and 8 were mutated (P < 0.01). However, no significant changes were observed when mutations accumulated at sites 9–16, which indicated that the RRACH sites at positions 6, 7, and 8 were the key sites for m6A enrichment on LncBMP4 (Fig 6D). When the progressive accumulation mutant Luc‐LncBMP4 vector was co‐transfected with the gga‐mir‐12211 overexpression vector in DF‐1 cells, mutations to sites 1–6 had no significant effect on the Luc‐LncBMP4 luciferase activity. However, Luc‐LncBMP4 luciferase activity significantly increased with the progressive accumulation of mutations to sites 6–8 (P < 0.01), whereas Luc‐LncBMP4 luciferase activity was not significantly affected by the progressive accumulation of mutations to sites 9–16 (Fig 6E). These results suggested that the ability of m6A to promote the adsorption of LncBMP4 to gga‐mir‐12211 was dose‐dependent, based on the availability of m6A modification sites.

EPC5, which is encoded by LncBMP4, exhibits similar BMP4 regulatory functions as LncBMP4

Most LncRNAs cannot encode proteins (Derrien et al, 2012); however, we predicted an open reading frame (ORF) from 117 to 347 bp in LncBMP4 (Appendix Fig S10A). To determine whether the ORF encodes any proteins, we inserted the ORF into a polyclonal restriction site between two His tags in the pet‐28a (+) vector (Appendix Fig S10B). The ORF was found to encode a 12.4‐kD protein, which was named expression in PGCs, Chr5 (EPC5), based on the genetic location of LncBMP4 (Appendix Fig S10C).

To determine whether LncBMP4 or EPC5 plays a role in the formation of PGCs, we constructed an EPC5 overexpression vector (oe‐EPC5) and a LncBMP4 expression vector containing a mutation in the transcriptional start site locus of the EPC ORF (oe‐LncBMP4‐mut, Appendix Fig S10D). Oe‐LncBMP4, oe‐EPC5, and oe‐LncBMP4‐mut were transfected into ESCs and induced by BMP4. Morphological observations showed a large number of EBs were produced during the induction process when LncBMP4, EPC5, and LncBMP4‐mut were overexpressed, with LncBMP4 overexpression having the largest effect (Fig 7A; Appendix Fig S10E). These results preliminarily indicated that both the LncBMP4 skeleton chain itself and the EPC5 micropeptide can promote the formation of PGCs. The gene expression analysis showed similar results, with significantly increased expression levels observed for Cvh, c‐kit, and BMP4 with the overexpression of LncBMP4, EPC5, and LncBMP4‐mut (Fig 7B). However, the expression levels of Cvh, c‐kit, and BMP4 were lower with the overexpression of EPC5 and LncBMP4‐mut than those with LncBMP4 overexpression. IF and FC analyses showed that the overexpression of LncBMP4, EPC5, and LncBMP4‐mut all promoted the formation of PGCs in the BMP4‐induced model (Fig 7C and D). It was expected that the simultaneous overexpression of LncBMP4‐mut and EPC5 had a stronger effect on the induction of PGCs than the overexpression of LncBMP4 alone in the PGC induction model in vitro (Appendix Fig S11). Therefore, we confirmed that LncBMP4 and its encoded peptides can perform similar functions during the formation of PGCs.

Figure 7. EPC5, encoded by LncBMP4, exhibits similar functions as LncBMP4 .

1. Morphological observation of small peptides and LncBMP4 backbone chains, promoting PGC formation in vitro, scale bar: 60 µm (n = 3 independent experiments).
2. qRT–PCR was used to detect the expression of related genes in each group, during induction (data are shown as mean ± SEM, n = 3 independent experiments, *P < 0.05, **P < 0.01, ***P < 0.001, one‐way ANOVA).
3. The efficiency of PGC differentiation was detected by IF after 4 days of induction using C‐KIT and CVH antibodies. Scale bar: 60 µm (n = 3 independent experiments).
4. The number of CVH‐positive cells was determined by flow cytometry (data are shown as mean ± SEM, n = 3 independent experiments, **P < 0.01, **P < 0.01, ***P < 0.001, ****P < 0.0001, one‐way ANOVA.).

Source data are available online for this figure.

m6A inhibits the translation of LncBMP4 into EPC5, to regulate BMP4 transcription

Given the regulation of BMP4 expression by LncBMP4, we examined whether EPC5, the small peptide encoded by LncBMP4, can also regulate the expression of BMP4. We transfected oe‐EPC5 (with His tag) into ESCs and PGCs. We designed 15 pairs of ChIP‐qPCR primers, based on the BMP4 promoter sequence, and the 14^th pair was able to detect binding between EPC5 and the BMP4 promoter. EPC5 was enriched on the BMP4 promoter, and the enrichment level observed in PGCs was significantly higher than that in ESCs (Fig 8A). To further examine the regulatory effects of EPC5 on BMP4 promoter activity, a dual‐luciferase vector including the BMP4 promoter (PGL‐BMP4) and PGL‐BMP4 without the EPC5 binding site (PGL‐BMP4‐mut) were constructed and co‐transfected with oe‐EPC5 into DF‐1 (Appendix Fig S12A–D). The double‐luciferase assay results showed that the overexpression of EPC5 could significantly increase the activity of the BMP4 promoter (P < 0.01). The EPC5 binding site mutation significantly reduced the activity of the BMP4 promoter (P < 0.01), which could not be rescued by EPC5 overexpression. Together, these results suggested that LncBMP4 can regulate the transcription of BMP4 through the expression of EPC5 (Fig 8B).

Figure 8. m6A inhibits LncBMP4 translation into EPC5, to regulate BMP4 transcription.

• A
  The binding of the small peptide EPC5 with the BMP4 promoter region was detected by ChIP‐qPCR in ESCs and PGCs (data are shown as mean ± SEM, n = 3 independent experiments, ****P < 0.0001, one‐way ANOVA).
• B
  Double‐luciferase detection of small peptides that affect BMP4 transcription (data are shown as mean ± SEM, n = 3 independent experiments, **P < 0.01, ****P < 0.0001, one‐way ANOVA).
• C
  Effects of m6A on small peptide expression, assessed by Western blot (up), ImageJ software was used for gray‐level analysis (down) (data are shown as mean ± SEM, n = 3 independent experiments, ****P < 0.0001, one‐way ANOVA).
• D, E
  Western blot (up) was used to detect gga‐mir‐12211 and m6A affect the ability of LncBMP4 to translate EPC5. ImageJ software was used for gray‐level analysis (down) (data are shown as mean ± SEM, n = 3 independent experiments, ***P < 0.001, ****P < 0.0001, one‐way ANOVA.).

Source data are available online for this figure.

Previously, we found that m6A modifications on LncBMP4 can affect its ability to bind to gga‐mir‐12211. Here, we investigated whether m6A affects the ability of LncBMP4 to produce EPC5. We inserted His tags at both ends of the ORF in the oe‐LncBMP4 vector (oe‐LncORF‐His) (Appendix Fig S12E and F) and transfected this vector into PGCs treated with NPC and MA. Western blot analysis showed that the expression of EPC5 increased with the addition of NPC but decreased with the addition of MA. These results indicated that high levels of m6A modification can inhibit LncBMP4 from expressing EPC5. Based on previous findings, we speculated that the m6A modification‐mediated inhibition of EPC5 expression may be due to the ability of m6A‐modified LncBMP4 to adsorb gga‐mir‐12211, which may affect the translation of EPC5 (Fig 8C).

To confirm this speculation, we co‐transfected oe‐LncORF‐His with the overexpression and interfering gga‐mir‐12211 vectors in DF‐1 cells, and the results showed that compared with the DF‐1 cells transfected with oe‐LncORF‐His alone, EPC5 expression decreased with gga‐mir‐12211 overexpression and increased after gga‐mir‐12211 interference (Fig 8D). Interestingly, the addition of NPC to DF‐1 cells co‐transfected with oe‐LncORF‐His and gga‐mir‐12211 overexpression vectors was able to rescue the expression of EPC5, which is consistent with our conjecture (Fig 8E). LncBMP4 can regulate the transcription of BMP4 through the expression of the small peptide EPC5, which can be inhibited by m6A modification.

Discussion

In this study, we identified a novel LncRNA (LncBMP4), which is specifically expressed during the formation of PGCs. LncBMP4 is involved in the formation of chicken PGCs through the regulation of BMP4 expression, and m6A modification regulates this process through two pathways. First, LncBMP4 that was modified with high levels of m6A demonstrated enhanced binding to gga‐mir‐12211, which upregulated the expression of BMP4. Second, LncBMP4 can regulate the transcriptional activity of BMP4 through the encoded small peptide EPC5, and high m6A levels on LncBMP4 inhibited the translation of EPC5.

The formation of PGCs is a complex process that is regulated by many factors. Increasing research has identified and explored an increasing number of key genes and signaling pathways that regulate PGCs formation (Irie & Surani, 2017; Tan & Tee, 2019). Recently, researchers have turned their attention to epigenetic modifications, including DNA methylation (Hill et al, 2018), histone methylation (Ohta et al, 2017), and LncRNAs (Magnúsdóttir & Surani, 2014), which may be involved in PGCs formation. Studies have demonstrated that LncRNAs can regulate gene expression through a variety of methods, participating in a wide range of developmental processes, including germ cell development. In 2013, Bao et al studied changes in the expression of LncRNA and mRNA at 6 points during the development of male lines, using the Arraystar Mouse LncRNA microarray. They found that LncRNA expression was dynamically regulated during the development of male germ stem cells (Bao et al, 2013). In 2018, He et al found that lincRNA5 can activate the expression of MAPKAPK5 by binding to its promoter during the formation of chicken PGCs, based on transcriptome data from chicken ESCs, PGCs, and SSCs (He et al, 2018a). In our study, we examined a LncRNA, identified by high‐throughput sequencing to target BMP4 (LncBMP4), a factor involved in the origination of PGCs. Functional verification of LncBMP4 showed similar functions as those observed for BMP4. Although the origin of chicken PGCs remains unclear, studies have shown that the formation of chicken PGCs is also induced by BMP4, similar to the process observed in mammals. Therefore, studying factors that regulate BMP4 expression can provide a new understanding of the origins of chicken PGCs.

The ceRNA system of LncRNAs is a conventional method for the regulation of gene expression, which also determines the tissue specificity of LncRNAs (Wang et al, 2010; Guttman & Rinn, 2012). The LncRNA HULC is considered one of the most important upregulated transcripts in hepatocellular carcinoma (HCC). The HULC transcript contains a binding site for the miRNA mir‐372, which can reduce mir‐372 activity in the HCC cell line Hep3B following the overexpression of HULC (Wang et al, 2010). The adsorption of mir‐372 by HULC reduces the inhibitory effects of mir‐372 against the translation of its target gene, Prkacb (Fan et al, 2013). In our study, a similar phenomenon occurred. LncBMP4 was specifically expressed in PGCs, where it regulated BMP4 expression by binding to gga‐mir‐12211. Interestingly, the ceRNA system of LncBMP4 is regulated by the m6A modification of LncBMP4. The m6A modification is the most important RNA modification, able to affect RNA stability and regulate RNA translation. Studies have shown that RNA m6A modification levels are proportional to the number of RRACH motifs (Chen et al, 2015). Yang et al found that the m6A modification levels of linc1281 increased during mESC differentiation and were primarily located on the three RRACH motifs found in the last exon of linc1281. The study also found that mutations in the RRACH motif did not affect the transcription level of linc1281 but, instead, affected its binding to let‐7 (Yang et al, 2018). A similar biological phenomenon was observed in our study. When the RRACH motifs at sites 6, 7, and 8 on LncBMP4 were mutated, the expression of LncBMP4 was not affected; however, its adsorption of gga‐mir‐12211 was weakened, which inhibited the expression of BMP4. In the process of analyzing the mechanism by which m6a regulates the ceRNA system of lncRNA, it was found that LncRNA was a single‐stranded linear RNA with a certain spatial structure. Nian Liu et al have shown that m6A can affect the spatial structure of single‐stranded RNA (Liu et al, 2015), while there are multiple stem loops in the spatial structure of LncBMP4, and the key region is the m6A modification site. The stem‐loop structure could affect the binding of gga‐mir‐12211. After m6A modification, the spatial structure of the binding site of gga‐mir‐12211 on LncBMP4 was opened, so it was helpful for gga‐mir‐12211 to bind with LncBMP4 and play corresponding functions. Therefore, our study showed that in addition to LncBMP4 and gga‐mir‐12211, m6A also played an important role in determining the expression of BMP4 during the formation of PGCs. This finding laid a foundation for analyzing the specific mechanisms involved in the formation of PGCs and also provided new ideas for improving the formation efficiency of PGCs, both in vivo and in vitro.

Studies have confirmed that LncRNAs can encode short peptides that have key biological functions when a small ORF exists; therefore, LncRNAs that encode functional small peptides often play a dual regulatory role during the developmental process (Ulveling et al, 2011; Nam et al, 2016). Matsumoto et al found that although the expression of SPAR LncRNA did not change, the small peptide encoded by SPAR played a key role in the downregulation of the mTORC1 signal, which promoted the muscle regeneration and repair of acute skeletal muscle injuries (Matsumoto et al, 2017). Huang et al found that the small peptide encoded by the LncRNA hoxb‐as3 was downregulated in cancer, and the hoxb‐as3 peptide, rather than the LncRNA, inhibited the growth, colony formation, migration, and invasion of cancer cells by inhibiting the differential splicing of hnRNPA1 into PKM (Huang et al, 2017). Although an increasing number of LncRNAs have been found to play biological functions by encoding small peptides, no relevant studies have been reported for PGCs. In our study, in addition to regulating the expression of BMP4 through the ceRNA system, LncBMP4 promoted the expression of BMP4 through the binding of EPC5, which is encoded by LncBMP4, to the promoter region of BMP4. However, high levels of m6A modification on LncBMP4 inhibited the translation of EPC5. The analysis showed that the binding site between gga‐mir‐12211 and LncBMP4 was located in the ORF region and that m6A‐modified LncBMP4 absorbed gga‐mir‐12211, which blocked the translation of LncBMP4. However, it is significant to determine which way is the main factor regulating the expression of BMP4. The detection of endogenous EPC5 expression showed that the expression of EPC5 is upregulated during the formation of PGCs, which is consistent with the expression of LncBMP4 (Appendix Fig S13). This is because mir‐12211 in a state of low expression during the formation of PGCs (in this case, EPC5, rather than the ceRNA system of LncBMP4, is the factor that activates BMP4 expression). The significance of the existence of the ceRNA of LncBMP4 is that when mir‐12211 is expressed, BMP4 can be still activated after EPC5 is inhibited, which ensure the normal occurrence of PGCs. Therefore, we believe that the existence of these two mechanisms of LncBMP4 regulating the expression of BMP4 is the guarantee of precise regulation of the formation of PGCs.

Materials and Methods

Ethics approval

All procedures involving the care and use of animals conformed to U.S. National Institute of Health guidelines (NIH Pub. No. 85‐23, revised 1996) and were approved by the Laboratory Animal Management and Experimental Animal Ethics Committee of Yangzhou University. The experiments were conducted under the approval of the Ethics Committee of Yangzhou University for Laboratory and Experimental Animals (SYXK(Su)2016‐0020). The chickens used in this study were obtained from the Institute of Poultry Science, Chinese Academy of Agriculture Sciences.

Reagents and Tools table

Reagent/Resource	Reference or Source	Identifier or Catalog number
LncBMP4 F:CCCAAGCTTAACATTCTGGAAGTGTAATCAC	This study	Table EV4
LncBMP4 R:CCGCTCGAGCATTAGTTCCCCTTGTATAAGAAGT	This study	Table EV4
EPC5 F:CCCGAATTCATGAACAGTCCCTCGTCT	This study	Table EV4
EPC5 R:CCCCTCGAGTCACAACAGGCCAGGGAA	This study	Table EV4
shLncBMP4‐1 F:gatccGCAATGCAATCTGTCTTTAACttcaagagaGTTAAAGACAGATTGCATTGCTTTTTTg	This study	Table EV4
shLncBMP4‐1 R:aattcAAAAAAGCAATGCAATCTGTCTTTAACtctcttgaaGTTAAAGACAGATTGCATTGCg	This study	Table EV4
shLncBMP4‐2 F:gatccGCCAAGGGTTATCATGGATGTttcaagagaACATCCATGATAACCCTTGGCTTTTTTg	This study	Table EV4
shLncBMP4‐2 R:aattcAAAAAAGCCAAGGGTTATCATGGATGTtctcttgaaACATCCATGATAACCCTTGGCg	This study	Table EV4
shLncBMP4‐3 F:gatccGGGACTGTTCATCTACCAACAttcaagagaTGTTGGTAGATGAACAGTCCCTTTTTTg	This study	Table EV4
shLncBMP4‐3 R:aattcAAAAAAGGGACTGTTCATCTACCAACAtctcttgaaTGTTGGTAGATGAACAGTCCCg	This study	Table EV4
shNC: GGGTGAACTCACGTCAGAAC	This study	Table EV4
miR12211 F:ACCGGTGCGGCCGCGAATTCTGAGTGCGTTTGGAATGCTCG	This study	Table EV4
miR12211 R:GATCGCAGATCCTTGGATCCTGGAAGTTCTGATGACACAGCCTG	This study	Table EV4
miR12258 F:ACCGGTGCGGCCGCGAATTCTCTGCCCACCTGCACAACTAAG	This study	Table EV4
miR12258 R:GATCGCAGATCCTTGGATCCCCATTTCCAAATCACGCTCCG	This study	Table EV4
shmiR12211 F:GACGGCGCTAGGATCATCAACAGTATTGGCCATATCTTGTCCTCTAACAAGTATTCTGGTCACAGAATACAACAGTATTGGCCATATCTTGTCCTCTAACAAGATGATCCTAGCGCCGTCTTTTTT	This study	Table EV4
shmiR12211 R:AAAAAAGACGGCGCTAGGATCATCTTGTTAGAGGACAAGATATGGCCAATACTGTTGTATTCTGTGACCAGAATACTTGTTAGAGGACAAGATATGGCCAATACTGTTGATGATCCTAGCGCCGTC	This study	Table EV4
shmiR12258 F:GACGGCGCTAGGATCATCAACGCACATGGCTGCATCTGTCCTCCTCCCAAGTATTCTGGTCACAGAATACAACGCACATGGCTGCATCTGTCCTCCTCCCAAGATGATCCTAGCGCCGTCTTTTTT	This study	Table EV4
shmiR12258 R:AAAAAAGACGGCGCTAGGATCATCTTGGGAGGAGGACAGATGCAGCCATGTGCGTTGTATTCTGTGACCAGAATACTTGGGAGGAGGACAGATGCAGCCATGTGCGTTGATGATCCTAGCGCCGTC	This study	Table EV4
shMettl3‐1 F:gatccGCAATGCAATCTGTCTTTAACttcaagagaGTTAAAGACAGATTGCATTGCTTTTTTg	This study	Table EV4
shMettl3‐1 R:aattcAAAAAAGCGATTGCTCCTTCCTCAACAtctcttgaaTGTTGAGGAAGGAGCAATCGCg	This study	Table EV4
shMettl3‐2 F:gatccGGATGGGATCATCTCCAAACCttcaagagaGGTTTGGAGATGATCCCATCCTTTTTTg	This study	Table EV4
shMettl3‐2 R:aattcAAAAAAGGATGGGATCATCTCCAAACCtctcttgaaGGTTTGGAGATGATCCCATCCg	This study	Table EV4
shMettl3‐3 F:gatccGATCTACGGGATGATCGAACGttcaagagaCGTTCGATCATCCCGTAGATCTTTTTTg	This study	Table EV4
shMettl3‐3 R:aattcAAAAAAGATCTACGGGATGATCGAACGtctcttgaaCGTTCGATCATCCCGTAGATCg	This study	Table EV4
shMettl14‐1 F:gatccAAGAGAAACCTGTAGGGCTTCTTATttcaagagaATAAGAAGCCCTACAGGTTTCTCTTTTTTTTg	This study	Table EV4
shMettl14‐1 R:aattcAAAAAAAAGAGAAACCTGTAGGGCTTCTTATtctcttgaaATAAGAAGCCCTACAGGTTTCTCTTg	This study	Table EV4
shMettl14‐2 F:gatccGAGCCACCACTGGAAGAATACTATAttcaagagaTATAGTATTCTTCCAGTGGTGGCTCTTTTTTg	This study	Table EV4
shMettl14‐2 R:aattcAAAAAAGAGCCACCACTGGAAGAATACTATAtctcttgaaTATAGTATTCTTCCAGTGGTGGCTCg	This study	Table EV4
shMettl14‐3 F:gatccCCCACCCTCACAAACAGTAATTTCAttcaagagaTGAAATTACTGTTTGTGAGGGTGGGTTTTTTg	This study	Table EV4
shMettl14‐3 R:aattcAAAAAACCCACCCTCACAAACAGTAATTTCAtctcttgaaTGAAATTACTGTTTGTGAGGGTGGGg	This study	Table EV4
LncBMP4 F:CCAAGGGGATCATTAGAT	This study	Table EV5
LncBMP4 R:AGTGTTACCCGTTCATTT	This study	Table EV5
BMP4 F:AGTGATGAAGCCGCTGTCG	This study	Table EV5
BMP4 R:GCCCTGATGAGTCTGTGCC	This study	Table EV5
Cvh F:TTCTTGTGGCAACTTCGG	This study	Table EV5
Cvh R:CAACGACCAGTTCGTCCA	This study	Table EV5
C‐kit F:GCGAACTTCACCTTACCCGATTA	This study	Table EV5
C‐kit R:TGTCATTGCCGAGCATATCCA	This study	Table EV5
U1 F:ACATGGTGTACAACAAGCGC	This study	Table EV5
U1 R:CTCACCGCTCATCGTATCGG	This study	Table EV5
GAPDH F:GCCCAGAACATCATCCCA	This study	Table EV5
GAPDH R:CGGCAGGTCAGGTCAACA	This study	Table EV5
Mettl3 F:CAACCCGCAAGGCTTCAA	This study	Table EV5
Mettl3 R:GCGTTCGATCATCCCGTAG	This study	Table EV5
Mettl14 F:GCGTTCGATCATCCCGTAG	This study	Table EV5
Mettl14 R:AAGTATCATAAGAAGCCCTA	This study	Table EV5
β‐actin F:CAGCCATCTTTCTTGGGTAT	This study	Table EV5
β‐actin R:CTGTGATCTCCTTCTGCATCC	This study	Table EV5
LncBMP4‐P1 F:GTGTAATCACAGCCTTTG	This study	Table EV5
LncBMP4‐P1 R:CACGATATGTACTTACCCT	This study	Table EV5
LncBMP4‐P2 F:ATCGTGCCTCCTCTACCA	This study	Table EV6
LncBMP4‐P2 R:TCCCCGAGCAGTATTGAA	This study	Table EV6
LncBMP4‐P3 F:TTGTGAGGGCTGATAAGA	This study	Table EV6
LncBMP4‐P3 R:ATTCATTATCCTTCCTTG	This study	Table EV6
LncBMP4‐P4 F:AGGAAGGATAATGAATCT	This study	Table EV6
LncBMP4‐P4 R:CATTAGTTCCCCTTGTATA	This study	Table EV6
BMP4‐3’UTR F:TCCCTCCCGTCCCACC	This study	Table EV6
BMP4‐3’UTR R:CATTTGCACATAAAGTCATAAA	This study	Table EV6
BMP4 Promoter‐1 F:CTTTATTTACTTATTTACGGCT	This study	Table EV7
BMP4 Promoter‐1 R:CAGTTCTTGTATCCTGTGTTAT	This study	Table EV7
BMP4 Promoter‐2 F:ACAAGAACTGCTTTGGGGAGGT	This study	Table EV7
BMP4 Promoter‐2 R:GTGCCACATCTCCACGGCTCTG	This study	Table EV7
BMP4 Promoter‐3 F:GGCTGGGGTTGGGCTTGGGGAT	This study	Table EV7
BMP4 Promoter‐3 R:TTGTGCATGTACAGACCTCGAT	This study	Table EV7
BMP4 Promoter‐4 F:ACCTTTTTCCTCTGGCTGTTCT	This study	Table EV7
BMP4 Promoter‐4 R:ACATTTCACGTTGGGTCTTAGG	This study	Table EV7
BMP4 Promoter‐5 F:CAGCTCGGAGATGAGCAGTCCT	This study	Table EV7
BMP4 Promoter‐5 R:TTCAGTGGTAGATACAACATTG	This study	Table EV7
BMP4 Promoter‐6 F:TGCTTTTCAAACCTTCCACATT	This study	Table EV7
BMP4 Promoter‐6 R:GAAATAGTGTAGCATTTACTGA	This study	Table EV7
BMP4 Promoter‐7 F:ATGCTACACTATTTCCATCCGT	This study	Table EV7
BMP4 Promoter‐7 R:GCTGGTTGGCGACCCTGCACAT	This study	Table EV7
BMP4 Promoter‐8 F:AACCCCCTGCTATGTGCAGGGT	This study	Table EV7
BMP4 Promoter‐8 R:TCTGTGCAAGGGCCTGTATAGC	This study	Table EV7
BMP4 Promoter‐9 F:CAGACTTTGTTTGTAGGTGGGG	This study	Table EV7
BMP4 Promoter‐9 R:GCTGACACTGTAATATCTCTGA	This study	Table EV7
BMP4 Promoter‐10 F:TTTTATCTGCGTGCGCCCTTTC	This study	Table EV7
BMP4 Promoter‐10 R:TCAGTTTGTAGATGAACCGAGA	This study	Table EV7
BMP4 Promoter‐11 F:CAGGTTGGATATTAGGAAACAT	This study	Table EV7
BMP4 Promoter‐11 R:ATTGGTGGGATTACAGGACCAA	This study	Table EV7
BMP4 Promoter‐12 F:TCTATGGGTGGTGGTTTGGCCA	This study	Table EV7
BMP4 Promoter‐12 R:ATTTGGGGATGAGTTGAGAGCC	This study	Table EV7
BMP4 Promoter‐13 F:GTCATTGCAACCTTTGGCCACG	This study	Table EV7
BMP4 Promoter‐13 R:CTCGGCTGGGCGGCACCGTGGT	This study	Table EV7
BMP4 Promoter‐14 F:GGTTGAGCGGCTCCACGTG	This study	Table EV7
BMP4 Promoter‐14 R:GCCAAGGCGAGGAAGAAG	This study	Table EV7
BMP4 Promoter‐15 F:TTTTTATGCCGATCCTTTATTA	This study	Table EV7
BMP4 Promoter‐15 R:AATACATGGCAAAGCGCGATGA	This study	Table EV7

Methods and Protocols

Cell culture

High‐glycemic Dulbecco’s modified Eagle medium (DMEM, Gibco, 21013024), containing 10% fetal bovine serum (FBS, Gibco, 16140071), 100 U/ml penicillin, and 100 U/ml streptomycin (Gibco, 15140148), was used to culture DF‐1 cells. Knockout (KO)‐DMEM (Gibco, A3181501), containing 10% FBS, 2.5% chicken serum (Sigma, C5405), 100 U/ml penicillin, 100 U/ml streptolysin, 0.4 µmol/l non‐essential amino acids (Sigma, M7145), 2 mmol/l glutamine (Sigma, G7513), 0.1 mmol/l β‐mercaptoethanol (Sigma, M6250), 10 ng/µl mouse leukemia inhibitory factor (mLIF, Sigma, ESG1106), 10 ng/µl basic fibroblast growth factor (bFGF, Sigma, GF446), 5 ng/ml human stem cell factor (hSCF, Sigma, GF021), was used to culture ESCs and PGCs. The separation and culture methods used to obtain ESCs and PGCs are described in a previous study (Zhang et al, 2016). High‐glucose DMEM, containing 10% FBS, 2.5% chicken serum, 100 U/ml penicillin, 100 U/ml streptolysin, 0.4 µmol/l non‐essential amino acid, 2 mmol/l glutamine, 0.1 mmol/l β‐mercaptoethanol, and 40 ng/ml BMP4 protein, was used to induce ESC differentiation into PGCs, in vitro (Shi et al, 2014).

Vector construction

RNA from chicken PGCs was extracted using the TRIzol method (Invitrogen, 15596018). RNA was reverse transcribed into cDNA according to the first strand synthesis premix kit instructions (Biolab, JN0004). The full length of LncBMP4 was amplified with specific primers, including the restriction sites HindIII and XhoI. The PcDNA3.1 vector and the LncBMP4 fragment were digested with HindIII and XhoI. OeLncBMP4 was constructed by linking the purified PcDNA3.1 vector with the LncBMP4 fragment. Oe‐LncBMP4 was transfected into DF‐1 using Fugene HD transfection reagent (Roche, 4709705001). Transfection conditions were Fugene_(V):oe‐LncBMP4 _(M) = 3:1. The oe‐LncBMP4 activity was detected by qRT–PCR. Three interference targets were screened based on the full‐length sequence of LncBMP4, and upper and lower primers were designed, including BamHI and EcoRI restriction sites at the 5' ends, respectively. Primers were annealed to form double‐stranded oligomers and attached to the pGMLV‐SC5 skeleton to construct shLncBMP41, shLncBMP42, and shLncBMP43. shLncBMP41, shLncBMP42, and shLncBMP43 were transfected into DF‐1 cells. After 48 h, cell samples were collected and RNA was extracted and reverse transcribed into cDNA. The interference effects of each interference vector on LncBMP4 expression was quantitatively detected by qRT–PCR. The vector with the highest interference effect was named shLncBMP4 and used for all subsequent experiments.

The construction process used to generate the BMP4‐3'UTR luciferase reporter vector (Luc‐LncBMP4), the RRACH site mutant vector (Luc‐LncBMP4mut), the EPC5 overexpression vector (oe‐EPC5, fusion with His tag), LncBMP4 overexpression vector with a mutation at the ORF site (oe‐LncBMP4‐mut), and the overexpression and interference expression vectors for gga‐mir‐12211 and gga‐mir‐12258 were the same as described above, All primer sequences can be found in Table EV4.

Bioinformatics analysis

Burrows Wheeler Aligner was used to compare clean reads to the chicken genome (galgal6) and set the input control (CHIP‐seq of H3K4me2). MACS2 software were used to complete peak calling analysis (P ≤ 0.005), counted the number, width, and distribution of the peaks, then selected the peaks in BMP4 promoter (DATASET: GSE147255; Data ref: Zhang et al, 2021b).

The MBD‐seq data were evaluated and trimmed off for high‐quality assurance and then aligned to the galGal6 reference genome by bowtie v.1.1.1. A combination of procedures available in SAMtools and BEDtools was applied for data manipulation, filtration, and format conversion. Peaks of DNA methylation were called using MACS1.4.2, then selected the peaks in BMP4 promoter. (DATASET: PRJNA449166; Data ref: He et al, 2018b).

Based on the prediction of the function of lncRNAs, the 100 kb upstream and downstream genes with protein‐coding ability and expression pattern consistent with lncRNAs were defined as potential cis‐regulated target genes. (DATASET: PRJNA644100; Data ref: Gao et al, 2020b).

Northern blot

Primers F: 5' CAAGGCAACATTTCAGGC 3' and R: 5' CCTCCAGTG TTACCCGTTC 3' (704 bp) were designed based on the LncBMP4 sequence. CDNA from PGCs was used as a template, and the PCR DIG Probe Synthesis Kit (Roche, 11636090910) was used to amplify fragments, coupled with dUTP labeling. RNA from chicken ESCs, PGCs, SSCs, and chicken embryo fibroblasts were extracted using the TRIzol method (Invitrogen, 15596018) and 1% formaldehyde‐denatured gel electrophoresis was performed, at a constant voltage of 25 V at a low temperature, overnight. RNA was transferred to a membrane using the capillary method, and the membrane was baked at 80°C, for 2 h, to fix RNA molecules. A fresh, denatured probe was added to the 10.0 ml DIG Easy Hyb for hybridization. After hybridization, the membrane was washed and photographed in an imager (BIO‐RAD, 733BR2897).

PGCs formation‐related genes were detected by qRT–PCR

At 0, 2, 4, and 6 days after BMP4 model induction, cells under different treatment conditions were collected. Genital ridge samples were collected on day 4.5 of the in vivo incubation. RNA was extracted using the TRIzol method and reverse transcribed to obtain cDNA for qRT–PCR. The expression of related genes (Cvh, c‐kit, BMP4, and LncBMP4) during the formation of PGCs was detected by qRT–PCR, according to the instructions of the SuperReal color fluorescent quantitative premix reagent (SYBR Green, Tiangen, fp215‐02). mRNA levels were normalized against the level of β‐actin. Data were analyzed using the 2^{−ΔΔ C t} relative quantitative method, using Microsoft Excel. Primers can be found in Table EV5.

Immunofluorescence (IF)

On day 4 of the BMP4 induction model, cells subjected to different treatments were washed 2 times with phosphate‐buffered saline (PBS). The washed cells were fixed with 200 µl 4% paraformaldehyde for 15 min and then washed with PBS 2 times. Then, cells were treated with 200 µl 0.5% Triton X‐100 (Solarbio, China, vT8200) for 20 min at room temperature, followed by the addition of 10% FBS‐PBS (Gibco, USA, 10099141), to block the cells, at room temperature for 30 min. Diluted anti‐MVH (1:100; Abcam, UK, ab13840), and anti‐c‐kit (1:100; Invitrogen, USA, 14‐1172‐81) primary antibodies were used to cover the cell surface, at 4°C, overnight. PBST was used to wash off the primary antibody, followed by the addition of goat anti‐rat IgG (1:1,000; Proteintech, USA, SA00003‐11; fluorescein isothiocyanate [FITC] labeled) and goat anti‐mouse IgG (1:1,000; Proteintech, USA, SA00003‐12; tetramethylrhodamine [TRITC] labeled), incubation in a cassette for 2 h, washing with PBST, and the addition of 200 l DAPI (Beyotime, China, C1002), which was incubated for 15 min and then washed. Fluorescence was observed and photographed under a fluorescence inverted microscope (Olympus, Japan, FV1200).

Flow cytometry (FC) to detect labeled proteins in PGCs

The BMP4‐induced cells, after 4 days of induction, and cell suspensions from the genital ridge (E4.5), following the application of various treatment conditions, were collected and subjected to centrifugation at 300 g for 6 min. The cell pellet was resuspended in anti‐CVH (1:100; Abcam, UK, ab13840) antibody dilution, incubated at 37°C for 2 h. Then, the cells were centrifuged at 300 g rpm for 6 min. The cell pellet was washed, resuspended in PBST three times, and incubated with goat anti‐mouse IgG (1:1,000; Proteintech, USA, SA00003‐12; TRITC‐labeled), at 37°C for 2 h in the dark, to detect the proportion of MVH‐positive cells by FC.

PAS staining

Induced PGCs were fixed in a 24‐well plate with PAS fixator for 10–15 min. Wash twice with PBS. After drying, the cells were stained with a PAS staining kit (Solarbio, Beijing, China, G1280). Then, the samples were treated with oxidizer for 10 min and washed twice with PBS. When the sample dried, which was added with the Schiff staining solution for 15 min. Then, the samples were washed twice with sulfite with PBS, stained with hematoxylin for 2 min.

EDU proliferation detection

Primordial germ cells with good growth status were inoculated with 1 × 10⁵ cells per well in 24‐well plates and then transfected with oeLncBMP4 and shLncBMP4, respectively, and then, the PGCs were cultured to normal growth stage, and the cells were labeled with 50 μm EDU medium. After labeling, the DNA was fixed with 4% paraformaldehyde and stained with Apollo (Ribbio, Guangzhou, China). Finally, the DNA was stained with Hoechst33342 (Ribbio, Guangzhou, China) and observed with inverted fluorescence microscope.

Vascular injection of chicken embryos and immunohistochemistry (IHC)

Embryos were collected from hatched eggs at 2.5 days (13–17 HH), by making a round hole in the eggshell, with a diameter not exceeding 0.5 cm, using the blunt end of tweezers. The chicken embryo was exposed and a micro‐pipettor was used to inject the processed PGCs or the encased transfection vector into the embryonic blood vessel of the chicken embryo. The eggshell was cross‐sealed with medical tape, to continue incubation. The embryos were collected at 4.5 days, and paraffin sections were prepared according to previously described procedures. The paraffin sections were placed in 0.01 M sodium citrate buffer solution, pH 6.0, boiled in a microwave oven, allowed to cool to room temperature. This procedure was repeated four times. Sections were incubated in CVH primary antibody, diluted in 5% bovine serum, and incubated overnight at 4°C. Sections were washed with PBS three times and incubated with secondary antibody, for 1 h at room temperature. Sections were washed 3 times with PBS, followed by incubation with diaminobenzidine (DAB)‐H₂O₂ for 10 min. Mayer staining was performed for 30 s, differentiation with hydrochloric acid and alcohol was performed for 3 s, sections were immersed in running water for 15 min and treated with acetic acid for 2 min, ethanol for 2 min, and xylene for 5 min. Sections were covered with neutral gum and sections were observed and photographed using a microscope (Nikon).

miRNA screening, targeting LncBMP4 and BMP4‐3'UTR

miRNA screening, targeting both LncBMP4 and BMP4‐3'UTR, was predicted using DIANA TOOLS. Overexpression and interference expression vectors were constructed for the identified miRNAs, and their functions during the formation of PGCs were verified by in vivo and in vitro experiments, to identify key miRNAs.

Luciferase reporter assay

DF‐1 cells were transfected with BMP4‐3'UTR and the following: gga‐mir‐12211 overexpression or interference vectors; LncBMP4 overexpression or interference vectors; gga‐mir‐12211 and LncBMP4 overexpression vectors; or NPC (inhibitor of m6A methylation) (Uddin et al, 2019) and MA (activator of m6A methylation) (Huang et al, 2019) together. Cells for each group were collected, centrifuged for 6 min at 300 g, and the supernatant was discarded. According to the instructions provided with the Luciferase Reporter Assay Kit (E1910, Promega), 70 µl diluted lysate was added to each tube, the cells were resuspended, incubated at room temperature for 10 min, and then transferred to a 96‐well plate. After 70 µl firefly fluorescence solution was added to each well respectively, and luciferase activity was determined using a microplate reader. The regulation modes utilized against BMP4‐3'UTR, by LncBMP4 and gga‐mir‐12211, were analyzed according to the results. The effects of m6A on LncBMP4 binding with gga‐mir‐12211 were detected using the same method. The effect of EPC5 on the activity of the BMP4 promoter was assessed using the same experimental procedures.

RIP

ESCs, PGCs, and DF‐1 cells, transfected with Luc‐LncBMP4 containing mutated RRACH sites, and PGCs treated with STM2457 (GLPBIO, GC19771) were collected. m6A enrichment on LncBMP4 was detected using the RIP assay, with an anti‐m6A antibody. According to the RIP kit instructions (BML, RN1001), 500 µl lysis buffer (+) was added to the collected cells, which were then mixed and incubated at 4°C for 10 min. A/G magnetic beads, mixed with IgG antibody (CWBIO, CW0107S), were added to prepare the antibody‐protein‐A/G bead complex for enriching RNA. Then, 400 µl Master Mix Solution was added to the bead complex, mixed by vortex, and 250 µl Solution III was added, to isolate RNA. The LncBMP4 expression level was detected by qRT–PCR. Primers are listed in Table EV6. The sequence of miR‐12211 fused with AGO sequence was inserted into pcDNA3.0 vector and the (pcDNA3.0‐miR‐12211‐ago). PGCs in good growth state were transfected with pcDNA3.0‐miR‐12211‐AGO2 vector and then collected cells were used for RIP test with AGO antibody (Abcam, Cambridge, UK, ab186733), and qRTPCR was performed with LncBMP4 primer (P2, Table EV6) and BMP4‐3’UTR primer (Table EV6).

LncBMP4 coding capability detection

The online software ORF Finder (http://www.bioinformatics.org/sms2/orf_find.html) was used to predict the ORF in LncBMP4. The amplification primers were designed according to the ORF position. EcoR I and Xho I restriction sites were used to clone the ORF fragments, using the cDNA from PGCs as a template, and the ORF fragments were ligated into the pet‐28a (+) vector. After the constructed vector was transformed into BL strain, protein expression was induced, and protein was collected and the concentration was determined by the bicinchoninic acid method. The 35‐kD His fusion protein was inserted into the same vector, in the same position, as a positive control. Western blot analysis was performed. The protein loading volume was set at 20 µg, which was added to 6 µl sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS–PAGE) protein loading buffer. Denaturation was performed at 100°C for 30 min. A 12% Sure PAGE™ Prefabricated plastic E (M01210C, GenScript, Nanjing, China) gel was used to perform electrophoresis. Transblotting (Trans‐Blot SD Cell, Bio‐Rad, California, USA) was used to transfer the protein to a PVDF membrane (YA1701, Solarbio, Shanghai, China). Membranes were incubated with 5% skim milk (1706404, Bio‐Rad, California, USA) overnight, rinsed with TBST (Cat# T1082, Solarbio, Shanghai, China) 3 times, and incubated overnight with anti‐His antibody (Cell Signaling, 12698S). Membranes were incubated with goat anti‐rabbit IgG (CWBIO, CW0107) secondary for 2 h. Enhanced chemiluminescence (ECL) color rendering (CW0049, CWBIO, Beijing, China) was used to image membranes, using a chemiluminescent imager. Primers are indicated in Table EV4. The effects of m6A on LncBMP4 protein expression were studied, in vitro, using the same experimental methods, anti‐β‐Catenin (CWBIO, Beijing, China, CW0096 M).

ChIP‐qPCR

EPC5 was transfected in ESCs and PGCs (His tag was fused in the vector), and the binding of EPC5 to the promoter region of BMP4 was detected by ChIP‐qPCR. Cells were collected according to the ChIP kit instructions (Millipore, 17‐371rf) and incubated at room temperature, for 10 min, with 500 µl 16% formaldehyde. After the addition of 800 µl 10× glycine, samples were incubated on ice for 5 min, centrifuged at 300 g for 5 min, and the supernatant was discarded. The cells were washed 2 times with PBS containing protease inhibitors and then collected by centrifugation. A volume of 1 ml SDS cell lysis solution was added to resuspend the cells. Cell DNA was fractured by ultrasound, and 900 µl dilution buffer was added (containing protease inhibitors). Then, the samples were divided into an IgG group, a positive control group, and a His antibody group. In each group, 60 µl G agarose glycoprotein was added and incubated at 4°C for 1 h, after which the agarose was precipitated by centrifugation at 1,200 g. For each sample, 10 µl of the supernatant was used as input. The remaining supernatant was collected, and 1 µg mouse IgG antibody was added to the IgG group, 1 µg His antibody was added to the His antibody group. Overnight rotating incubation was performed at 4°C. On the second day, 60 µl agarose glycoprotein was added to each tube, incubated at 4°C for 1 h, and agarose was precipitated at 1,200 g. Then, the DNA was eluted after the supernatant was discarded. The enrichment of EPC5 bound to the promoter region of BMP4 was detected by qPCR. Primers can be found in Table EV7.

Data analysis

qRT–PCR, luciferase reporter assays, FC, ChIP‐qPCR, and RIP were each repeated 3 times. The results were analyzed using the Student’s t‐test, using the SPSS19.0 software package. Diagrams were generated in GraphPad Prism 6. During the experiment, the principle of random grouping was followed. In all of the experiments, the measurement of the sample was performed blindly by other colleagues. The data collection was blinded to group assignments. Data analysis was done by a person who was blinded to the treatments of samples.

Author contributions

BL conceived and designed the experiments. JZ and JJ performed the experiments. QZ analyzed the data. QZ wrote the manuscript. YZ, WW, and GC edited the manuscript.

Conflict of interest

The authors declare that they have no conflict of interest.

Supporting information

Appendix

Table EV1

Table EV2

Table EV3

Table EV4

Table EV5

Table EV6

Table EV7

Dataset EV1

Dataset EV2

Dataset EV3

Source Data for Appendix

Source Data for Figure 1

Source Data for Figure 2

Source Data for Figure 3

Source Data for Figure 4

Source Data for Figure 5

Source Data for Figure 6

Source Data for Figure 7

Source Data for Figure 8

EMBO reports (2022) 23: e52491.

Data availability

Data and code related to this paper may be requested from the authors or supplied in the manuscript. The accession numbers for the datasets reported in NCBI BioProject and GEO database: PRJNA449166 (https://www.ncbi.nlm.nih.gov/bioproject/PRJNA449166) for the DNA methylation sequencing data of three types of stem cells; the GSE147255 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE147255) for H3K4me2 ChIP‐seq for three types of stem cells; the PRJNA644100 (https://www.ncbi.nlm.nih.gov/bioproject/PRJNA644100) for RNA‐seq for three types of stem cells.