m⁶A mRNA methylation regulates testosterone synthesis through modulating autophagy in Leydig cells

ABSTRACT

Macroautophagy/autophagy is indispensable for testosterone synthesis in Leydig cells (LCs), and here we report a negative association between m⁶A modification and autophagy in LCs during testosterone synthesis. A gradual decrease of METTL14 (methyltransferase like 14) and an increase of ALKBH5 (alkB homolog 5, RNA demethylase) were observed in LCs during their differentiation from stem LCs to adult LCs. These events led to reduced mRNA methylation levels of N⁶-methyladenosine (m⁶A) and enhanced autophagy in LCs. Similar regulation of METTL14, ALKBH5, and m⁶A was also observed in LCs upon treatment with human chorionic gonadotropin (HsCG). Mechanistically, m⁶A modification promoted translation of PPM1A (protein phosphatase 1A, magnesium dependent, alpha isoform), a negative AMP-activated protein kinase (AMPK) regulator, but decreased expression of CAMKK2 (calcium/calmodulin-dependent protein kinase kinase 2, beta), a positive AMPK regulator, by reducing its RNA stability. Thus, m⁶A modification resulted in reduced AMPK activity and subsequent autophagy inhibition. We further demonstrated that ALKBH5 upregulation by HsCG was dependent on enhanced binding of the transcriptional factor CEBPB (CCAAT/enhancer binding protein [C/EBP], beta) and the TFEB (transcription factor EB) to its gene promoter. Moreover, HsCG treatment decreased METTL14 by reducing its stability. Collectively, this study highlights a vital role of m⁶A RNA methylation in the modulation of testosterone synthesis in LCs, providing insight into novel therapeutic strategies by exploiting m⁶A RNA methylation as targets for treating azoospermatism and oligospermatism patients with reduction in serum testosterone.

Abbreviations: 3-MA: 3-methyladenine; ACTB: Actin, beta; ALKBH5: alkB homolog 5, RNA demethylase; AMPK: AMP-activated protein kinase; BafA1: bafilomycin A_1; CAMKK2: calcium/calmodulin-dependent protein kinase kinase 2, beta; CEBPB: CCAAT/enhancer-binding protein (C/EBP), beta; ChIP: chromatin immunoprecipitation; FTO: fat mass and obesity associated; HsCG: human chorionic gonadotropin; HSD3B: 3β-hydroxysteroid dehydrogenase; LCs: Leydig cells; m⁶A: N⁶-methyladenosine; METTL14: methyltransferase like 14; METTL3: methyltransferase like 3; MTOR: mechanistic target of rapamycin kinase; PPM1A: protein phosphatase 1A, magnesium dependent, alpha isoform; PRKAA: 5ʹ-AMP-activated protein kinase catalytic subunit alpha; SQSTM1: sequestosome 1; STK11/LKB1: serine/threonine kinase 11; TFEB: transcription factor EB; ULK1: unc-51-like kinase 1; WTAP: Wilms tumor 1-associating protein; YTHDF: YTH N6-methyladenosine RNA binding protein

KEYWORDS:

• AMPK
• autophagy
• Leydig cells
• m⁶A modification
• testosterone synthesis

Introduction

Testosterone, an important male hormone, is required for male sexual development and characteristic maintenance [1,2], and its deficiency usually causes male sexual dysfunction and reduced reproductive capacity [3,4]. Moreover, some late-onset hypogonadism symptoms, including a decrease in lean body mass, diabetes, a decline in bone mineral density, cardiovascular disease, and sleep disturbances, are observed in males with a deficiency in serum testosterone [1,5,6]. Leydig cells (LCs), residing in the testis interstitium, are the primary source of testosterone in males [7]. During the development of mammals from infancy to adulthood, four distinct developmental stages of LCs, including stem LCs, progenitor LCs, immature LCs, and adult LCs, have been characterized [7]. Undifferentiated stem LCs are capable of doing indefinite self-renewal and differentiating to progenitor LCs that share the capacity to produce minimal levels of androgen. Moreover, progenitor LCs continue to differentiate into immature LCs with a higher level of biosynthetic activity of testosterone metabolites. Adult LCs are derived from immature LCs characterized by more efficient production of testosterone. With aging, the ability to synthesize testosterone in LCs progressively declines, resulting in a reduction of serum testosterone and concomitant alterations in physical function [5].

Up to date, molecular mechanisms associated with the regulation of testosterone in LCs remain elusive [3,8]. Growing evidence support the notion that autophagy plays an important role in regulating testosterone synthesis in LCs; this process is initiated in progenitor LCs, peaked in adult LCs, and attenuated in aged LCs [3,4,9]. As a normal cellular metabolic process, autophagy mediates the delivery of cellular components to lysosomes to contribute to cell survival under stress [10–13]. Previous studies have identified that AMP-activated protein kinase (AMPK), a major energy-sensing kinase, can promote autophagy to maintain energy homeostasis, whereas MTOR (mechanistic target of rapamycin kinase) can inhibit autophagy [14,15].

Cell growth and differentiation depend upon the fine-tuning regulation of gene expression at both transcriptional and translational levels [16]. It is noted that post-transcriptional modifications of eukaryotic mRNA play important roles in helping cells survive various stimuli [17]. N⁶-methyladenosine (m⁶A), the most prevalent internal modification in mRNA, has roles in various biological processes [18,19]. Usually, m⁶A modification is embedded within the consensus motif DRACH (D = A, G, U; R = A, G; H = A, C, U) and preferentially enriched around stop codons [20–22]. m⁶A modification is a dynamic and reversible process, which is mediated through “writer” and “eraser” proteins [21,23]. In mammalian cells, methyltransferase complex consisting of METTL3 (methyltransferase like 3), METTL14 (methyltransferase like 14), and a regulatory subunit WTAP (Wilms tumor 1-associating protein) catalyze m⁶A modification [24,25]. By contrast, ALKBH5 (alkB homolog 5, RNA demethylase) and FTO (fat mass and obesity associated) are capable of mediating the reversal of this methylation [26,27]. m⁶A exerts many of its functions through “reader” proteins that selectively recognize and bind to the m⁶A-modified transcripts [28–33]. One class of m⁶A readers is the YTHDF (YTH N6-methyladenosine RNA binding protein), which includes YTHDF1 and YTHDF2. YTHDF1 promotes translation of the target m⁶A-modified mRNA, while YTHDF2 decrease mRNA stability and mediate subcellular localization and alternative splicing of mRNA [23,34–36].

Recent works have uncovered that m⁶A modification regulates differentiation of embryonic stem cells [37–39], reprogramming of somatic cells [40], sex determination [37–39], spermatogenesis [42],41 tumorigenicity of cancer cells [23,43], and neuronal development [36]. Given key roles of m⁶A modification in regulating gene expression and its dynamic properties, it was reasonable to speculate that m⁶A modification regulates differentiation of LCs and testosterone synthesis. Here, we found a negative correlation of m⁶A modification with testosterone synthesis in LCs. Further studies revealed that m⁶A modification affected AMPK activity by promoting translation of the negative AMPK regulator PPM1A (protein phosphatase 1A, magnesium dependent, alpha isoform) while decreasing expression of the positive AMPK regulator CAMKK2 (calcium/calmodulin-dependent protein kinase kinase 2, beta) by reducing its mRNA stability, resulting in autophagy inhibition and subsequent suppressed testosterone synthesis in LCs.

Results

Autophagy is vital for testosterone synthesis in LCs

In this study, we measured the autophagic marker LC3B-II levels in stem LCs (post-natal days 10; D10), progenitor LCs (D20), immature LCs (D40), adult LCs (D90), and aged LCs (D540), and found increasing expression of LC3B-II in LCs during cellular differentiation, peaking in adult LCs (Figure 1A); however, cellular SQSTM1 (sequestosome 1) became markedly reduced during development from stem LCs to adult LCS (Figure 1A). Moreover, we observed an increasing number of LC3 puncta in LCs during cell differentiation from stem LCs to adult LCs (Figure 1B), accompanied by gradually increasing serum testosterone levels (Fig. S1). Furthermore, we also observed upregulated expression of LC3B-II, increased numbers of LC3 puncta, and a concurrent decrease of SQSTM1 in primary LCs following HsCG treatment (Figure 1C,D), accompanied by increased mRNA expression of steroidogenic enzymes, including Star (steroidogenic acute regulatory protein) and Hsd3b (3β-hydroxysteroid dehydrogenase) (Figure 1E). As predicted, HsCG treatment markedly increased testosterone production in LCs (Figure 1F). Similarly, HsCG also induced increased expression of LC3B-II but decreased expression of SQSTM1 in TM3 cells (Figure 1G). These cells are usually chosen as a surrogate for primary LCs based on their sharing many properties of primary LCs [44].

Figure 1. Autophagy is closely associated with testosterone synthesis in Leydig cells (LCs). Primary LCs from mouse testes at various developmental stages were isolated and cultured for 48 h. (A) Autophagy-related protein expression was analyzed by western blotting. The expression levels of the target proteins were determined by densitometry by normalizing to ACTB, and data are presented as the means ± SEM (n = 3). *P < 0.05 vs. post-natal days 10 (D10) group. (B) The expression of LC3B in LCs at different developmental stages was examined by immunofluorescence staining, and the average LC3B puncta number per cell was quantified. Data are presented as means ± SEM (n = 3). *P < 0.05 vs. D10 group. (C) Primary LCs were treated with HsCG for indicated times. The cell extracts were subjected to western blotting and quantitative analysis. Data are presented as means ± SEM (n = 3). *P < 0.05 vs. D10 group. (D) The expression of LC3B in LCs at 6 h was examined by immunofluorescence staining, and the average LC3B puncta number per cell was quantified. Data are presented as means ± SEM (n = 3). *P < 0.05 vs. the control cells. (E) mRNA levels of Hsd3b and Star were determined by qRT-PCR. Data are presented as means ± SEM (n = 3). *P < 0.05 vs. the control cells. (F) Testosterone levels were detected by radioimmunoassay. Data are presented as means ± SEM (n = 3). *P < 0.05 vs. the control cells. (G) TM3 cells were treated with HsCG for indicated times. The cell extracts were subjected to western blotting and quantitative analysis. Data are presented as means ± SEM (n = 3). *P < 0.05 vs. the control cells

Since the increase in LC3B-II could reflect an increase in autolysosome synthesis or reduction in autophagic flux in cells, we performed additional autophagy blockade experiments. We note that LC3B-II levels, LC3 puncta number per cell and testosterone production became substantially ameliorated after pre-treatment with the autophagy inhibitor 3-methyladenine (3-MA) in HsCG-treated primary LCs (Figure 2A-C). Bafilomycin A₁ (BafA1) pre-treatment resulted in the enrichment of LC3B-II and SQSTM1 but lowered the testosterone production (Figure 2D-F). Moreover, we also observed similar effects of 3-MA and BafA1 on HsCG-induced autophagic flux in primary LCs in TM3 cells (Figure 2G-L). To further investigate whether testosterone synthesis in LCs involved autophagy, we used small interfering RNA (siRNA) targeting Atg7 to knock down ATG7 in LCs. As predicted, knockdown of ATG7 alleviated HsCG-induced LC3B-II increase and testosterone production (Fig. S2), suggesting that autophagy mediates testosterone synthesis.

Figure 2. 3-methyladenine and bafilomycin A₁ regulate HsCG-induced autophagy in Leydig cells (LCs). Primary LCs were pre-treated with 3-methyladenine (3-MA, 5.0 mM) for 1 h followed by exposure to HsCG for 6 h. (A) The cell extracts were subjected to western blotting and quantitative analysis. (B) Immunofluorescence staining of LC3B and quantitative analyses of LC3 puncta per cell were performed. (C) Testosterone levels were measured by radioimmunoassay. Primary LCs were pre-treated with bafilomycin A₁ (BafA1, 10 nM) for 1 h followed by exposure to HsCG for 6 h. (D) The cell extracts were subjected to western blotting and quantitative analysis. (E) Immunofluorescence staining of LC3B and quantitative analyses of LC3 puncta per cell were performed. (F) Testosterone levels were measured by radioimmunoassay. TM3 cells were pre-treated with 3-MA for 1 h followed by exposure to HsCG for 6 h. (G) The cell extracts were subjected to western blotting and quantitative analysis. (H) Immunofluorescence staining of LC3B and quantitative analyses of LC3 puncta per cell were performed. (I) Testosterone levels were measured by radioimmunoassay. TM3 cells were pre-treated with BafA1 for 1 h followed by exposure to HsCG for 6 h. (J) The cell extracts were subjected to western blotting and quantitative analysis. (K) Immunofluorescence staining of LC3B and quantitative analyses of LC3 puncta per cell were performed. (L) Testosterone levels were measured by radioimmunoassay. Data are presented as means ± SEM (n = 3). *P < 0.05 vs. the control cells; #P < 0.05 vs. the HsCG-treated cells

Autophagy in LCs is dependent on the AMPK signaling pathway

In order to assess whether the MTOR-ULK1 (unc-51-like kinase 1) signaling pathway mediates autophagy and testosterone synthesis in LCs, we examined the expression of p-MTOR Ser2448 and its downstream target p-ULK1 Ser757. We observed no marked changes in Ser757 phosphorylation of ULK1 and Ser2448 phosphorylation of MTOR (Figure 1C,G). Interestingly, we observed enhanced phosphorylation of AMPK subunit PRKAA2 (protein kinase, AMP-activated, alpha 2 catalytic subunit) at Thr172 (p-PRKAA2 Thr172), a canonical initiator of the autophagic process, and increased expression of its downstream protein p-ULK1 Ser555 (Figure 1C,G). Furthermore, during the differentiation of LCs from stem LCs to adult LCs, cellular p-PRKAA2 Thr172 and p-ULK1 Ser555 levels were steadily increased (Figure 1A).

In this study, the treatment using MTOR inhibitor rapamycin failed to alter the levels of LC3B-II, SQSTM1, p-PRKAA2 Thr172, p-ULK1 Ser555, and p-ULK1 Ser757 in HsCG-treated TM3 cells and primary LCs (Figure 3A and S3A). However, rapamycin could effectively inhibit phosphorylation of RPS6KB1, another classical MTOR target not associated with autophagic flux [45], indicating that the autophagic process in LCs is MTOR-independent. The specific AMPK inhibitor compound C could effectively suppress a serious of events, including HsCG-induced increase in LC3B-II, Thr172 phosphorylation of PRKAA2, Ser555 phosphorylation of ULK1, and decrease of SQSTM1 (Figure 3B and S3B). Similarly, the ULK1 inhibitor SBI-0206965 significantly attenuated HsCG-induced increase of LC3B-II and Ser555 phosphorylation of ULK1, and decrease of SQSTM1 in both TM3 cells and primary LCs. However, we examined no inhibitory effects of SBI-0206965 on p-PRKAA2 T172 (Figure 3C and S3C). These results confirmed the AMPK-ULK1-dependency of autophagy induction in LCs during HsCG stimulus. Consistently, we confirmed the increased expression of p-PRKAA2 Thr172 in mouse LCs during development, peaking in adult LCs, by immunofluorescence assay (Figure 3D).

Figure 3. AMPK mediates autophagy in Leydig cells (LCs) upon the HsCG challenge. (A-C) TM3 cells were pre-treated with rapamycin (100 μM), compound C (1 μM) or SBI-0206965 (2 μM) for 1 h followed by exposure to HsCG for 6 h. The cell extracts were subjected to western blotting and quantitative analysis. Data are presented as means ± SEM (n = 3). *P < 0.05 vs. the control cells; #P < 0.05 vs. the HsCG-treated cells. (D) The expression of p-PRKAA2 in LCs at different developmental stages was examined by immunofluorescence staining. Fluorescence intensity was presented as means ± SEM (n = 3). *P < 0.05 vs. post-natal days 10 (D10) group. (E) TM3 cells were treated with HsCG for indicated times. The cell extracts were subjected to western blotting and quantitative analysis. Data are presented as means ± SEM (n = 3). *P < 0.05 vs. the control cells. (F) Protein levels of PPM1A, STK11/LKB1, and CAMKK2 in LCs at different developmental stages were analyzed by western blotting. *P < 0.05 vs. D10 group. (G and H) TM3 cells were transfected with siRNA targeting Camkk2 (si-Camkk2) or pcDNA3.1-Ppm1a vector (O/E Ppm1a; O/E, overexpression) for 36 h followed by exposure to HsCG for 6 h. The cell extracts were subjected to western blotting and quantitative analysis. Data are presented as means ± SEM (n = 3). *P < 0.05 vs. the control cells; #P < 0.05 vs. the HsCG-treated cells

CAMKK2/PPM1A is required for thr172 phosphorylation of PRKAA2

To determine the mechanisms underlying HsCG-induced enhanced expression of p-PRKAA2 Thr172 in LCs, we examined CAMKK2 and STK11/LKB1 (serine/threonine kinase 11), two classical kinases that phosphorylate PRKAA2 at Thr172, and PPM1A, a phosphatase that mediates Thr172 dephosphorylation of PRKAA2 [46]. Our data revealed that HsCG induced increased expression of CAMKK2, and reduced expression of PPM1A in a time-dependent manner in both TM3 cells and primary LCs using western blotting (Figure 3E and S4A). Furthermore, we also recorded upregulated expression of CAMKK2 and downregulated expression of PPM1A in adult LCs in comparison to stem LCs by western blotting and immunofluorescence assays (Figure 3F and S5). Also, TM3 cells and primary LCs following CAMKK2 knockdown or overexpression of PPM1A markedly suppressed HsCG-induced increased expression of LC3B-II, AMPK activation, and concurrent decrease of SQSTM1 (Figure 3G,H, and S4B, C).

Testosterone synthesis in LCs is regulated by m⁶A mRNA modification

With the development of LCs from stem LCs to adult LCs, we observed a gradual decrease of m⁶A levels (Figure 4A). Moreover, we recorded a gradual increase in Alkbh5 and Fto mRNA in LCs during their development, whereas Mettl3, Mettl14, and Wtap exhibited no remarkable differences (Fig. S6). The protein levels of FTO and ALKBH5 became upregulated, while METTL14 levels in LCs became downregulated during cellular differentiation from stem LCs to adult LCs confirmed by both western blotting and immunofluorescence analyses (Figure 4B, S7 and S8). Consistently, we revealed a gradual decrease of m⁶A levels in LCs using ELISA during the same cellular differentiation process (Figure 4C). Furthermore, we also observed a significantly increased ALKBH5 and FTO protein levels, and a reduction in METTL14 in both primary LCs and TM3 cells upon HsCG treatment accompanied with a decrease in m⁶A levels, in a time-dependent manner (Figure 4D, E, and S9).

Figure 4. The m⁶A mRNA methylation regulates AMPK activity in Leydig cells (LCs). (A) m⁶A levels in LCs were assessed by immunofluorescence staining. Fluorescence intensity was presented as means ± SEM (n = 3). *P < 0.05 vs. post-natal day 10 (D10) group. (B) Protein levels of m⁶A regulatory enzymes in LCs at various developmental stages were examined by western blotting and quantitatively analyzed. Data were presented as means ± SEM (n = 3). *P < 0.05 vs. D10 group. (C) Relative m⁶A levels were measured by ELISA-based m⁶A quantitative analyses. *P < 0.05 vs. D10 group. TM3 cells were treated with HsCG for indicated times. (D) The expression of m⁶A regulatory enzymes were examined by western blotting and quantitatively analyzed. *P < 0.05 vs. the control cells. (E) Relative m⁶A levels were measured by ELISA-based m⁶A quantitative analyses. *P < 0.05 vs. the control cells. (F) The expression of both HSD3B and m⁶A levels in LCs at 6 h was examined by immunofluorescence staining. Fluorescence intensity was presented as means ± SEM (n = 3). *P < 0.05 vs. the control cells. (G) TM3 cells were transfected with increasing doses of pcDNA3.1-Mettl14 vector (O/E Mettl14; O/E, overexpression) for 36 h followed by HsCG treatment for 6 h. The cell extracts were subjected to western blotting and quantitative analysis. Data are presented as the means ± SEM (n = 3). *P < 0.05 vs. the control cells; #P < 0.05 vs. the HsCG-treated cells. (H) TM3 cells were transfected with siRNA targeting Alkbh5 (si-Alkbh5) and cultured for 36 h followed by HsCG treatment for 6 h. The cell extracts were subjected to western blotting and quantitative analysis. Data are presented as means ± SEM (n = 3). *P < 0.05 vs. the control cells; #P < 0.05 vs. the HsCG-treated cells

We further demonstrated that HsCG decreased m⁶A levels and increased HSD3B levels in LCs using immunofluorescence analyses (Figure 4F). To understand how m⁶A modification modulates the synthesis of testosterone in LCs, we tried to assess the association between m⁶A methylation and autophagy. We observed that METTL14 overexpression could alleviate HsCG-induced m⁶A reduction, testosterone production, LC3B-II upregulation, and concurrent decrease of SQSTM1 (Figure 4G and S10). Additionally, knockdown of ALKBH5 but not FTO significantly inhibited HsCG-induced m⁶A decrease, testosterone production, and autophagy in LCs (Figure 4H, S10, and S11). Together, these data support the idea that by modulating autophagy in LCs, testosterone synthesis may correlate to reduced m⁶A mRNA modification.

m⁶A mRNA modification is reduced in LCs from patients with azoospermia or oligospermia

To further investigate how m⁶A modifications play important roles in regulating testosterone synthesis in LCs, we measured m⁶A methylation levels in LCs from patients with azoospermia or oligospermia. Our data demonstrated that expression of HSD3B, a measure of the capacity to synthesize testosterone [47,48], positively correlated with puncta numbers in LC3 cells, while negatively correlated with SQSTM1 in LCs (Figure 5A and S12). Furthermore, we observed a negative correlation between m⁶A and HSD3B expression in these patients, suggesting m⁶A modification might mediate testosterone synthesis in LCs (Figure 5B). As predicted, we observed increased expression of METTL14 and decreased expression of ALKBH5 in those LCs that had reduced levels of HSD3B (Figure 5C,D).

Figure 5. Upregulated m⁶A levels were observed in Leydig cells (LCs) from azoospermia or oligospermia patients with lowered levels of serum testosterone. (A) The expression of LC3B in LCs at 6 h was examined by immunofluorescence staining, and the average LC3B puncta number per cell was quantified. (B) m⁶A levels in LCs were assessed by immunofluorescence assay. (C and D) The protein levels of METTL14 and ALKBH5 in LCs were determined by immunofluorescence staining. Data are presented as means ± SEM. *P < 0.05

m⁶A methylation regulates the activation of AMPK

Given that we identified a negative correlation between m⁶A methylation and autophagy, we next evaluated the possible relationship between m⁶A methylation and activation of AMPK in LCs. Our data revealed that METTL14 knockdown enhanced phosphorylation of PRKAA2, upregulated expression of LC3B-II, and decreased expression of SQSTM1 (Figure 6A); however, compound C effectively attenuated these changes (Figure 6A). Moreover, we observed decreased p-PRKAA2 Thr172, CAMKK2, and LC3B-II levels, and increased PPM1A and SQSTM1 levels in LCs in response to METTL14 overexpression (Figure 6B). Additionally, knockdown of CAMKK2 effectively alleviated both METTL14 knockdown- and ALKBH5 overexpression-induced PRKAA2 activation, LC3B-II increase, and SQSTM1 reduction in both TM3 cells and primary LCs (Figure 6C, D, and S13). Similarly, ectopic expression of PPM1A also suppressed both METTL14 knockdown- and ALKBH5 overexpression-induced PRKAA activation, LC3B-II increase, and SQSTM1 reduction (Figure 6E, F, and S13). Together, we provided solid data which confirmed that CAMKK2 upregulation and PPM1A downregulation modulate autophagy in LCs owing to reduced m⁶A levels.

Figure 6. m⁶A mRNA methylation impacts the expression of regulators of the AMPK pathway. (A) TM3 cells were transfected with shRNA targeting Mettl14 (sh-Mettl14) followed by compound C treatment for 6 h. The cell extracts were subjected to western blotting and quantitative analysis. Data are presented as means ± SEM (n = 3). *P < 0.05 vs. the control cells; #P < 0.05 vs. the sh-Mettl14-treated cells. (B) TM3 cells were transfected with increasing doses of pcDNA3.1-Mettl14 vector (O/E Mettl14; O/E, overexpression). The cell extracts were subjected to western blotting and quantitative analysis. Data are presented as means ± SEM (n = 3). *P < 0.05 vs. the control cells. (C) TM3 cells were transfected with sh-Mettl14 before treatment with siRNA targeting Camkk2 (si-Camkk2), and cell lysate was subjected to western blotting and quantitative analysis. Data are presented as means ± SEM (n = 3). *P < 0.05 vs. the control cells; #P < 0.05 vs. the sh-Mettl14-treated cells. (D) TM3 cells were transfected with the pcDNA3.1-Alkbh5 (O/E Alkbh5) vector before treatment with si-Camkk2, and cell lysate was subjected to western blotting and quantitative analysis. Data are presented as means ± SEM (n = 3). *P < 0.05 vs. the control cells; #P < 0.05 vs. the O/E Alkbh5-treated cells. (E) TM3 cells were transfected with sh-Mettl14 before transfection with pcDNA3.1-Ppm1a (O/E Ppm1a), and cell lysate was subjected to western blotting and quantitative analysis. Data are presented as means ± SEM (n = 3). *P < 0.05 vs. the control cells; #P < 0.05 vs. the sh-Mettl14-treated cells. (F) TM3 cells were transfected with pcDNA3.1-Alkbh5 (O/E Alkbh5) before transfection with pcDNA3.1-Ppm1a (O/E Ppm1a), and cell lysate was subjected to western blotting and quantitative analysis. Data are presented as means ± SEM (n = 3). *P < 0.05 vs. the control cells; #P < 0.05 vs. the O/E Alkbh5-treated cells

m⁶A methylation affects the RNA stability of CAMKK2

We next investigated whether m⁶A modification could regulate the expression of CAMKK2 and PPM1A. Because m⁶A methylation appeared to reduce the expression of CAMKK2, we speculated that the Camkk2 transcript was a target of YTHDF2, a protein that is capable of reducing the stability of m⁶A methylated transcripts. Consistent with this hypothesis, knock down of YTHDF2 increased expression of Camkk2 at both mRNA and protein levels to a similar extent as knock down of METTL14 in both TM3 cells and primary LCs (Figure 7A–C and S14A-C). Conversely, m⁶A methylation appeared to upregulate PPM1A, suggesting that YTHDF1 could recognize the transcript of PPM1A, which promoted the translation of PPM1A. As expected, knock down of YTHDF1 resulted in lower PPM1A protein levels, while we observed no significant changes in mRNA (Figure 7A–C and S14A-C). We noted that knock down of YTHDF1 failed to change the expression of CAMKK2, while knock down of YTHDF2 was unable to affect the expression of PPM1A (Figure 7A–C and S14A-C). Given the specificity of YTHDF1 or YTHDF2 in regulating the expression of PPM1A or CAMKK2, respectively, we hypothesized that YTHDF1 and YTHDF2 functioned at the level of RNA recognition and interaction. Consistent with this hypothesis, YTHDF1 interacted stronger with Ppm1a than Camkk2, while YTHDF2 interacted stronger with Camkk2 than Ppm1a, as assessed by RNA-immunoprecipitation (RIP) analyses (Figure 7D, S15A, and B). Moreover, our data demonstrated that the YTHDF1 enrichment at Ppm1a transcripts and YTHDF2 enrichment at Camkk2 transcripts were significantly reduced in LCs upon METTL14 knockdown (Figure 7D). Furthermore, it was noted that both YTHDF1 and YTHDF2 immunoprecipitation efficiencies remained unchanged irrespective of the treatments (Fig. S15C and D).

Figure 7. m⁶A induces the decay of the Camkk2 transcript in Leydig cells (LCs). TM3 cells were treated with indicated siRNAs. (A and B) Cell lysates were subjected to western blotting and quantitative analysis. Data are presented as means ± SEM (n = 3). *P < 0.05 vs. the control cells. (C) mRNA levels of target genes were measured by qRT-PCR. *P < 0.05 vs. the control cells. (D) YTHDF1/YTHDF2 was immunoprecipitated, followed by qRT-PCR for assessing the association of the indicated transcripts with each protein. Data are presented as means ± SEM (n = 3). *P < 0.05 vs. the IgG group. (E) Schematic representation of the position of m⁶A motifs with Camkk2 transcript. (F) Abundance of Camkk2 transcript among mRNA immunoprecipitated with anti-m⁶A antibody was measured by qRT-PCR and normalized to input. Data are presented as means ± SEM (n = 3). *P < 0.05 vs. the IgG group. (G) Abundance of Camkk2 transcript among mRNA immunoprecipitated with anti-m⁶A antibody from cells with or without transfection of shRNA targeting Alkbh5 (sh-Alkbh5) was measured by qRT-PCR. Data are presented as means ± SEM (n = 3). *p < 0.05. (H) Camkk2-3ʹ-UTR of the wild-type or containing an m⁶A consensus sequence mutant (A to G) was fused with a luciferase reporter. (I) Luciferase activity of Camkk2-3ʹ-UTR was measured and normalized to Renilla luciferase activity. Data are presented as means ± SEM (n = 3). *P < 0.05 vs. the control cells. (J) TM3 cells pre-treated with indicated siRNAs were treated with actinomycin D (5 μg/mL) for 2 h, followed by measurement of Camkk2 and Ppm1a mRNA levels at indicated times. *P < 0.05 vs. the control cells

To further confirm whether the Camkk2 transcript could be m⁶A modified, we performed sequence analysis of the Camkk2 transcript and found four matches to the 5ʹ-RRACU-3ʹ m⁶A consensus sequence within 3ʹ-UTR of the Camkk2 transcript (Figure 7E). We next performed m⁶A RIP analyses and found that m⁶A was significantly enriched at site 1 and site 3 (Figure 7F). Moreover, m⁶A levels at both site 1 and site 3 of the Camkk2 transcript were significantly reduced in LCs upon treatment with HsCG; knock down of ALKBH5 effectively inhibited HsCG-induced m⁶A reduction (Figure 7G). Furthermore, we generated luciferase reporters containing a firefly luciferase placed before the wild-type or mutant Camkk2-3ʹ-UTR (Figure 7H). The luciferase reporter assay revealed that HsCG treatment could upregulate luciferase activity, while knock down of YTHDF2 effectively attenuated this increase (Figure 7I). Moreover, HsCG failed to increase the luciferase activity of the reporter construct bearing the mutant Camkk2-3ʹ-UTR (Figure 7I). To understand the molecular mechanism that accounted for the regulation of CAMKK2 by m⁶A methylation, we tried to assess whether m⁶A modification could affect Camkk2 mRNA decay. We observed decreases in the decay rates of Camkk2 mRNA upon knock down of YTHDF2 or METTL14 in both TM3 cells and primary LCs (Figure 7J and S14D, E). Conversely, knock down of YTHDF2 failed to attenuate the decay rates of Ppm1a mRNA (Figure 7J and S14D, E).

m⁶A methylation promotes translation of the PPM1A transcript

We observed decreased expression of PPM1A in LCs upon knock down of YTHDF1 or METTL14, while we observed no significant changes of Ppm1a mRNA levels (Figure 7A–C). Therefore, we speculated that the disruption of protein stability or reduction in translation efficiency might decrease PPM1A protein levels. To verify this hypothesis, we treated LCs with the protein translation inhibitor cycloheximide (CHX) and observed that the degradation rate of PPM1A in LCs following METTL14 knockdown was similar to those cells without METTL14 knock down (Figure 8A,B). Next, we pre-treated LCs with MG-132 to inhibit proteasome activity and then treated with HsCG. Our data demonstrated that HsCG reduced the expression of PPM1A in both primary LCs and TM3 cells (Figure 8C, D, S14F, and G). However, we did not observe the HsCG-induced decrease of PPM1A in LCs upon ALKBH5 knockdown (Figure 8C, D, S14F, and G). Together, these results suggested that m⁶A mediates PPM1A expression through regulating its translation without impacting protein stability. As predicted, HsCG treatment promoted binding between YTHDF1 and EIF3A (eukaryotic translation initiation factor 3) and EEF2 (eukaryotic elongation factor 2), as assessed by coimmunoprecipitation (Co-IP) analyses (Fig. S16).

Figure 8. m⁶A triggers the translation of the Ppm1a transcript. (A and B) TM3 cells with or without transfection of shRNA targeting Mettl14 (sh-Mettl14) were pre-treated with cycloheximide (CHX, 10 μg/mL) for 3 h followed by exposure to HsCG for indicated times. Expression levels of PPM1A were examined by western blotting and quantitatively analyzed. (C and D) TM3 cells with or without transfection of shRNA targeting Alkbh5 (sh-Alkbh5) were pre-treated with MG-132 (2 μM) for 1.5 h followed by exposure to HsCG for 6 h. Expression levels of PPM1A were examined by western blotting and quantitatively analyzed. Data are presented as means ± SEM (n = 3). *P < 0.05 vs. the control cells; #P < 0.05 vs. the MG-132-treated cells. (E) Schematic representation of the position of m⁶A motifs with Ppm1a transcript. (F) Abundance of the Ppm1a transcripts among mRNA immunoprecipitated with anti-m⁶A antibody was measured by qRT-PCR. Data are presented as means ± SEM (n = 3). *P < 0.05 vs. the IgG group. (G) Abundance of the Ppm1a transcript among mRNA immunoprecipitated with anti-m⁶A antibody from cells with or without transfection of sh-Alkbh5 was measured by qRT-PCR. Data are presented as means ± SEM (n = 3). *P < 0.05. (H) Ppm1a-3ʹ-UTR of the wild-type or containing an m⁶A consensus sequence mutant (A to G) were fused with a luciferase reporter. (I and J) Luciferase activity of Camkk2-3ʹ-UTR was measured and normalized to Renilla luciferase activity. Data are presented as means ± SEM (n = 3). *P < 0.05 vs. the control cells. (K) Schematic representation of mutation in pcDNA3.1-Ppm1a-CDS-3ʹUTR is shown. (L) TM3 cells were transfected with sh-Mettl14 followed by transfection with indicated vectors, and cell lysate was subjected to western blotting

Sequence analysis revealed three sites of m⁶A modification within the CDS region and four m⁶A sites in the 3ʹ-UTR (Figure 8E). m⁶A RIP analyses revealed that m⁶A significantly enriched at site 1, site 2, site 4, site 5, and site 6 (Figure 8F). Following HsCG treatment, m⁶A enrichment at both site 4 and site 5 became significantly reduced; however, ALKBH5 knockdown effectively suppressed these changes (Figure 8G). To study the potential roles of m⁶A modification in regulating the expression of PPM1A, we constructed a luciferase reporter containing Ppm1a-3ʹUTR (Figure 8H) and demonstrated that HsCG treatment significantly reduced luciferase activity of the reporter construct; however, we could not reproduce this phenomenon in those cells with YTHDF1 knockdown (Figure 8I). Moreover, we also mutated the potential m⁶A motif GGACU to GGGCU, and the luciferase assay revealed that HsCG treatment could still reduce the luciferase activity of the reporter construct bearing Ppm1a-3ʹ-UTR-mut1, Ppm1a-3ʹ-UTR-mut2, but not Ppm1a-3ʹ-UTR-mut3 (Figure 8 H,J). To further confirm the importance of the m⁶A motif in 3ʹ-UTR in regulating the expression of PPM1A, we generated an expression construct containing the coding sequence (CDS) and 3ʹUTR sequence. PPM1A expression in those LCs overexpressing pcDNA3.1-WT, pcDNA3.1-mut1, and pcDNA3.1-mut2 were markedly reduced upon knock down of METTL14; however, we observed no obvious changes in pcDNA3.1-mut3-transfected cells (Figure 8 K,L). Collectively, our results suggested that m⁶A modification at both site 4 and site 5 in 3ʹ-UTR could promote translation of the Ppm1a transcript.

HsCG decreases METT14 expression by affecting protein stability

In this study, we observed reduced METTL14 levels in LCs upon HsCG treatment (Figure 4D and S9), while we detected no significant changes in mRNA levels (Figure 9A). To investigate the molecular mechanism underlying HsCG-induced decreased expression of METTL14, we assessed the effects of HsCG on METTL14 protein stability. Our data demonstrated that HsCG treatment increased the degradation rate of METTL14 (Figure 9B,C). Moreover, METTL14 levels could be restored in those LCs with pre-treatment with MG-132 (Figure 9D), suggesting that HsCG reduced METTL14 protein levels by inducing its degradation.

Figure 9. HsCG induces transcription of Alkbh5 in Leydig cells (LCs). (A) TM3 cells were treated with HsCG for indicated times, and mRNA levels of Mettl14 and Alkbh5 were analyzed by qRT-PCR. *P < 0.05 vs. the control cells. (B and C) TM3 cells were pre-treated with cycloheximide (CHX, 10 μg/mL) for 3 h followed by exposure to HsCG for indicated times, and protein levels of METTL14 were assessed by western blotting and quantitatively analyzed. *P < 0.05 vs. the CHX-treatment cells. (D) TM3 cells were pre-treated with MG-132 (2 μM) for 1.5 h followed by exposure to HsCG for 6 h. Expression levels of METTL14 were examined by western blotting and quantitatively analyzed. Data are presented as means ± SEM (n = 3). *P < 0.05 vs. the control cells. (E) Potential transcription factor binding sites in the Alkbh5 promoter region were predicted. (F) TM3 cells were treated with HsCG for indicated times. Both cytoplasmic and nuclear proteins were isolated, followed by western blotting and quantitative analysis. ACTB and LMNB1 served as cytosolic and nuclear markers, respectively. Data are presented as means ± SEM (n = 3). *P < 0.05 vs. the control cells. (G and H) TM3 cells were transfected with siRNA targeting Cebpb (si-Cebpb) or si-Tfeb and cultured for 36 h followed by exposure to HsCG for 6 h. The expression of ALKBH5 was examined by western blotting and quantitatively analyzed. Data are presented as means ± SEM (n = 3). *P < 0.05 vs. the control cells; #P < 0.05 vs. the HsCG-treated cells. (I and J) TM3 cells were transfected with Alkbh5 promoter constructs harboring indicated mutations of TEFB binding sites (I) or CEBPB binding sites (J). Luciferase activity was measured, and data were presented as means ± SEM (n = 3). *P < 0.05. An “NS” denotes no statistically significant difference between the testing groups

HsCG induces ALKBH5 expression via enhancing its transcriptional levels

We noted that HsCG treatment upregulated the mRNA and protein levels of ALKBH5 (Figures 4D and Figures 9A). Therefore, we anticipated that HsCG treatment could upregulate ALKBH5 expression by increasing its transcriptional levels. As shown in Figure 9E, we observed multiple potential transcription factor binding sites in the Alkbh5 promoter, including SMAD3, SP1, FOXM1, CEBPB, TFEB, NR5A1/SF-1, RELA, MYC, and FOS. Next, we found that HsCG treatment markedly promoted nuclear translocation of NR5A1, CEBPB, and TFEB in both TM3 cells and primary LCs (Figure 9F and S17). Based on the fact that HsCG could promote expression of ALKBH5 as early as 1 h following treatment, we could thus rule out the possibility of NR5A1 in mediating HsCG-induced ALKBH5 expression. Additionally, LCs with CEBPB or TFEB knockdown effectively inhibited HsCG-induced increase in ALKBH5 (Figure 9G, H, and S17). To provide direct evidence confirming that TFEB could regulate Alkbh5 transcription, we performed site-directed mutagenesis to mutate these TFEB binding sites from CANNTG to CANNCC in the Alkbh5 luciferase reporter construct. The luciferase assay revealed that mutation of site 1 and site 3 significantly decreased luciferase activity, while mutation of site 2 had minor effects (Figure 9I). Moreover, we also mutated two potential CEBPB-binding sites from NTGCNN to NAAANN and observed that mutation of site 2 but not site 1 could significantly decrease the luciferase activity (Figure 9J). Consistently, HsCG treatment promoted binding of TFEB to site 1 and site 2 of the ALKBH5 promoter, and binding of CEBPB to site 2 of this promoter, as assessed by ChIP assay (Fig. S18). Together, our results strongly support that HsCG upregulates expression of ALKBH5 by promoting binding of TFEB and CEBPB to the Alkbh5 promoter and then enhancing its transcription levels. Moreover, we also observed reduced enrichment of both TFEB and CEBPB in the nuclei of testicular LCs in aged mice in comparison to adult mice, accompanied by a decline of HSD3B (Figure 10A,B).

Figure 10. Reduced nuclear translocation of TFEB and CEBPB is observed in aged Leydig cells (LCs). Enrichment of TFEB (A) and CEBPB (B) in the nucleus of LCs was determined by immunofluorescence analysis. The merging of the red and green fluorescence signals was quantified with Image J software. D10, post-natal days 10; D540, post-natal days 540. (C) A model for the critical link between METTL14/ALKBH5 and autophagy in HsCG-treated LCs is proposed. HsCG reduces METTL14 expression and increases ALKBH5 expression by suppressing the stability of METTL14 and enhancing Alkbh5 transcription, respectively. Both downregulation of METTL14 and upregulation of ALKBH5 lead to reduced m⁶A mRNA methylation levels. Reduction of m⁶A levels alleviates m⁶A-mediated Ppm1a translation and upregulates CAMKK2 expression by retarding m⁶A-mediated decay of the transcript, contributing to activation of PRKAA2 and subsequently autophagy initiation

Discussion

Cell growth and differentiation depend on the fine-tuning regulation of gene expression [49]. LCs in mammals progressively improve their capacity to synthesize testosterone during development and differentiation from stem LCs to the adult LCs, peaking in the adult LCs [3]. As mammals age, the ability to synthesize testosterone markedly decreases [1,6,7,9]. Based on these facts, we speculated that some genes involved in testosterone synthesis might be dynamically regulated during the differentiation of LCs. It has been previously revealed that m⁶A modification plays an important role in regulating gene expression by impacting RNA stability, translation efficiency, subcellular localization, and alternative splicing [23,35,36,43]. In this study, we observed a gradual decrease of m⁶A modification levels in LCs during their differentiation from stem LCs to adult LCs, suggesting a role for m⁶A in mediating differentiation of LCs.

Growing evidence has supported that autophagy is vital for testosterone synthesis in LCs [3,4]. In this study, we observed that autophagy initiated in progenitor LCs, progressively increased with differentiation, and peaked in adult LCs, suggesting that autophagy was dynamically modulated during differentiation of LCs. Further studies revealed that m⁶A negatively mediated autophagy in LCs. Consistently, it has recently been reported that m⁶A can decrease the expression of ULK1 and TFEB by affecting their mRNA stability, resulting in inhibition of autophagy [22,34].

It has been reported that binding of HsCG to the LHCGR (luteinizing hormone/choriogonadotropin receptor) is crucial for stimulating subsequent intracellular signaling pathways [50]. LHCGR is expressed primarily in the gonads, where it regulates steroidogenesis [50]. In the testis, LHCGR is detected in both Sertoli cells and LCs, with the latter being the source of testosterone [3,51]. Therefore, we focused on LCs in this study. Following HsCG treatment, we could confirm the dependence of AMPK-ULK1 for the autophagic process in LCs. Note that phosphorylation of AMPK can be enhanced by an upstream kinase, including STK11/LKB1 and CAMKK2, and antagonized by PPM1A phosphatase [46]. In this study, both upregulation of the CAMKK2 kinase and downregulation of the PPM1A phosphatase, which contributed to the activation of PRKAA2 and autophagy initiation, were observed in LCs upon HsCG treatment.

Further studies demonstrated that m⁶A could inhibit PRKAA2 activity by inducing m⁶A-dependent translation of Ppm1a and m⁶A-dependent degradation of the Camkk2 transcript. Following HsCG treatment, reduced m⁶A slowed down the regulation of m⁶A on these transcripts, leading to increased CAMKK2 levels but reduced PPM1A levels. Moreover, as both overexpression of PPM1A and knock down of CAMKK2 could attenuate HsCG-induced autophagy in LCs, we believe that both PPM1A and CAMKK2 are crucial for autophagy induction and this coordinated regulation may provide a fine-tuning system in modulating testosterone synthesis (Figure 10C). Moreover, m⁶A modification-dependent testosterone synthesis was also confirmed in those patients with oligospermia or azoospermia. Together, these results highlight a vital role of m⁶A mRNA methylation in regulating testosterone synthesis through modulating autophagy in LCs.

m⁶A modification is catalyzed and formed by a methyltransferase complex consisting of METTL3, METTL14, and WTAP; however, ALKBH5 and FTO can reverse m⁶A modification [20,21]. It is noted that METTL14 is not enzymatically active, and its main function is to facilitate RNA binding to METTL3 [20,21]. Accumulating evidence has confirmed that reduced expression of METTL3 or loss-of-function mutation in METTL14 may result in lowered m⁶A levels [23]. In this study, we observed decreased expression of METTL14 but not METTL3 in LCs upon HsCG treatment. Moreover, a progressive decrease of METTL14 was detected in LCs during their differentiation from stem LCs to adult LCs. Therefore, we speculated that the reduction of METTL14 might hinder the assembly of the METTL3-METTL14-WTAP complex, inhibiting the potential of this complex to catalyze the m⁶A methylation of mRNA during testosterone synthesis in LCs.

FTO has been identified to demethylate m¹A in specific tRNAs, N⁶ and 2ʹ-O-dimethyladenosine (m⁶A_m) in some snRNAs, and internal m⁶A and m⁶A_m at the second base adjacent to the 5ʹcap (cap-m⁶A_m) in mRNA [20,21]. Compared with FTO, ALKBH5 only possesses effective demethylation activity toward m⁶A [20,21]. In this study, we found that knock down of ALKBH5 but not FTO could effectively reverse HsCG-induced autophagy. Moreover, it has been reported that ALKBH5 is most highly expressed in the testes but has low expression in the heart and brain [27]. Together, both the decrease of METTL14 and the increase of ALKBH5 could synergistically reduce m⁶A modification in LCs during testosterone synthesis.

In this study, knock down of CEBPB or TFEB dramatically attenuated Alkbh5 mRNA induction. HsCG induced recruitment of CEBPB and TFEB to the Alkbh5 gene promoter to enhance its transcription was confirmed by ChIP assay. TFEB, as a master gene during the autophagic process [52], has been reported to regulate Alkbh5 promoter activity following hypoxia/reoxygenation treatment in cardiomyocytes [22]. It is reported that TFEB nuclear translocation is the first step for binding to the promoter of target genes and then promote gene transcription. TFEB can be phosphorylated by various kinases, including MTORC1, ERK2, or AKT, and sequestered in the cytoplasm, while the inactivation of these kinases can induce dephosphorylation of TFEB and subsequent translocation to the nucleus [53–55]. Calcium ion release from lysosomes also contributes to TFEB nuclear translocation through PPP3/calcineurin-mediated dephosphorylation of TFEB [55]. Moreover, numerous studies have confirmed reactive oxygen species-dependent nuclear translocation of TFEB and autophagy activation [56]. At the moment, we do not have any data to unravel the molecular mechanism underlying the nuclear translocation of TFEB in LCs during testosterone synthesis, which deserves a follow-up study.

YTHDF2 has been reported to reduce the stability of m⁶A-modified mRNA, and YTHDF2-binding sites are usually enriched around stop codons and in 3ʹUTRs [20,21]. In this study, we identified two m⁶A modification sites to exist in 3ʹUTR of the Camkk2 transcript. In contrast to YTHDF2, YTHDF1 was shown to promote translation of m⁶A-modified transcripts by interacting with the EIF3A, EEF1A1, and EEF2 [35,57]. Moreover, similar to YTHDF2, YTHDF1-binding sites are also concentrated around the stop codons and in 3ʹUTRs [21]. However, recent studies have found that m⁶A in the CDS sequence, but not 3ʹUTR, also triggers the translation of some m⁶A-modified mRNAs [35]. Our data revealed that HsCG treatment significantly reduced m⁶A levels at two sites within 3ʹUTR of the Ppm1a transcript, but displayed no impact on other m⁶A modification sites within CDS. Although HsCG does not change m⁶A levels of the two sites within CDS of the Ppm1a transcript, whether they are also involved in PPM1A gene regulation, such as acting as sensors, deserves a separate study.

In summary, we have provided compelling both in vitro and in vivo evidence to demonstrate that m⁶A modification reduced testosterone synthesis in LCs through the regulation of autophagy by affecting the stability of the Camkk2 transcript and promoting translational efficiency of Ppm1a (Figure 10C). Further studies revealed that METTL14 and ALKBH5 modulated m⁶A modification in LCs upon the HsCG challenge. The results presented here provide insight into novel therapeutic strategies by exploiting m⁶A RNA methylation as targets for treating azoospermatism and oligospermatism patients with lowered levels of serum testosterone.

Materials and methods

Main reagents and cell culture

DMEM/F12 (11,320,033), horse serum (16,050,122), penicillin-streptomycin (10,378,016), and fetal bovine serum (FBS; 10099133C) were obtained from Gibco. The autophagy inhibitor 3-MA (HY-19,312), BafA1 (HY-100,558), MTOR inhibitor rapamycin (HY-10,219), AMPK inhibitor compound C (HY-13418A), protein synthesis inhibitor CHX (HY-12,320), DNA transcription inhibitor actinomycin D (HY-17,559), and ULK1 inhibitor SBI-0206965 (HY-16,966) were purchased from Medchemexpress. The specific information about antibodies used in the present study was provided in Table 1. TM3 cells (ATCC, CRL-1714) were cultured in DMEM/F12 with 5% horse serum, 2.5% fetal bovine serum, and 1% penicillin-streptomycin. Primary LCs from 4-week-old mice were isolated based on previously established protocols [58,59].

Table 1. Primary antibodies used in this study

Source	Primary antibodies	Catalog no.	Working dilution
Abcam	Rabbit anti-LC3B	ab48394	WB, 1:2000 IF, 1: 1000
Abcam	Mouse anti-SQSTM1	ab56416	WB, 1:1000
Abcam	Rabbit anti-METTL3	ab195352	WB, 1:2000 IF, 1: 1000
Abcam	Mouse anti-FTO	ab92981	WB, 1:1000
Boster	Rabbit anti-ACTB	MO1263-2	WB,1:500
Cell Signaling Technology	Rabbit anti-ULK1	6439	WB,1:1000
Cell Signaling Technology	Rabbit anti-p-ULK1 S757	14,202	WB,1:1000
Cell Signaling Technology	Rabbit anti-p-ULK1 S555	5869	WB,1:1000
Cell Signaling Technology	Rabbit anti-PRKAA2	5831	WB,1:1000
Cell Signaling Technology	Rabbit anti-p-PRKAA2 T172	50,081	WB,1:1000 IF, 1:50
Cell Signaling Technology	Rabbit anti-MTOR	2983	WB,1:1000
Cell Signaling Technology	Rabbit anti-p-MTOR S2448	5536	WB,1:1000
Cell Signaling Technology	Rabbit anti-RPS6KB1	9202	WB,1:1000
Cell Signaling Technology	Rabbit anti-p-RPS6KB1	9205	WB,1:1000
Cell Signaling Technology	Rabbit anti-PPM1A	3549	WB,1:1000
Cell Signaling Technology	Rabbit anti-SMAD3	9523	WB,1:1000
Cell Signaling Technology	Rabbit anti-FOS	2250	WB,1:1000
Cell Signaling Technology	Rabbit anti-p-PIK3C3/VPS34	13,857	WB,1:1000
Santa Cruz Biotechnology	Mouse anti-HSD3B	sc-100,466	IF,1:50
Santa Cruz Biotechnology	Mouse anti-SP1	sc-420	WB,1:500
Santa Cruz Biotechnology	Mouse anti-FOXM1	sc-376,471	WB,1:500
Santa Cruz Biotechnology	Mouse anti-CEBPB	sc-7962	WB,1:500 ChIP,1:50
Santa Cruz Biotechnology	Mouse anti-TFEB	sc-166,736	WB,1:500 ChIP,1:50
Santa Cruz Biotechnology	Mouse anti-NR5A1	sc-393,592	WB,1:500
Santa Cruz Biotechnology	Mouse anti-MYC	sc-40	WB,1:500
Santa Cruz Biotechnology	Mouse anti-LMNB1	sc-374,015	WB,1:500
Santa Cruz Biotechnology	Mouse anti-EEF2	sc-166,415	WB,1:500
Santa Cruz Biotechnology	Mouse anti-EIF3A	sc-376,651	WB,1:500
Santa Cruz Biotechnology	Mouse anti-PIK3C3/VPS34	sc-365,404	WB,1:500
Proteintech	Rabbit anti-CAMKK2	11,549-1-AP	WB,1:1000 IF,1:50
Proteintech	Rabbit anti-STK11/LKB1	10,746-1-AP	WB,1:1000
Proteintech	Rabbit anti-ALKBH5	16,837-1-AP	WB,1:1000 IF,1:50
Proteintech	Rabbit anti-YTHDF1	17,479-1-AP	WB,1:1000 Co-IP, 1:50 RIP, 1:50
Proteintech	Rabbit anti-YTHDF2	24,744-1-AP	WB,1:1000 Co-IP, 1:50 RIP, 1:50
Proteintech	Rabbit anti-RELA	10,745-1-AP	WB,1:1000
Sigma	Rabbit anti-METTL14	HPA038002	WB,1:2000 IF,1:100
Synaptic Systems	Rabbit anti-m^6A	202,003	IF,1:200

Plasmid and siRNA transfection

Gene sequences coding for Mettl14 and Alkbh5 were cloned into the pcDNA3.1 vector (Invitrogen, V79020). The siRNA sequences used in this study were designed and chemically synthesized by RiboBio as: Mettl14, 5ʹ-GGAUGAGUUAA UAG CUAAA-3ʹ; Fto, 5ʹ-GAGCUGGCAUCACGAUGAA-3ʹ; Alkbh5, 5ʹ-CCTATG AGTCCTCGGAAGATT-3ʹ; Ythdf1, 5ʹ-GCACACAACCUCUAUCUUU-3ʹ; Ythdf2, 5ʹ-GACCAAGAAUGGCAUUGCA-3ʹ; Atg7, 5ʹ-GCCTGGCATTTGATAAATGT A-3ʹ; Camkk2, 5ʹ-GTATCCACTTGGGCATGGAAT-3ʹ; Cebpb, 5ʹ-CACCCTGCGGA ACTTGTTCAA-3ʹ; Tfeb, 5ʹ-GCAGGCTGTCATGCATTATAT-3ʹ. Transfection was performed using Lipofectamine 2000 (Invitrogen, 11,668,019) according to the manufacturer’s protocols.

Construction of the stable cell lines

Lentiviral vector expressing Mettl14-specific or Alkbh5-specific shRNA was designed and constructed by GENECHEM. The sequence for Mettl14-specific shRNA was 5ʹ-CCGGGGAUGAGUUAAUAGCUAAACTCGAGTTTAGCTATTAACTCATCCTTTTTG-3ʹ, and sequence for Alkbh5-specific shRNA was 5ʹ-CCGGCCTATGAGTC CTCGGAAGATTCTCGAGAATCTTCCGAGGACT CATAGGTTTTTG-3ʹ. TM3 cells were transfected with the overexpressing or knock down plasmids. After 36 h of transfection, cells were treated with 2 μg/mL puromycin (Medchemexpress, HY-B1743A) to obtain stable cell lines.

Quantitative real-time PCR (qRT-PCR) and RNA decay analyses

Total RNA from cells was extracted using Trizol reagent (Invitrogen, 15,596,018) following the manufacturer’s instructions. The cDNA was generated by HiScript Q RT SuperMix (Vazyme, R122-01). qRT-PCR was performed using ChamQ SYBR qPCR Master Mix (Vazyme, Q311-02). All primers used in this study are listed in Table S1, and Actb was chosen as internal control for normalization. Moreover, to assess the effects of m⁶A modification on RNA stability, cells were treated with actinomycin D at 5 μg/mL for indicated times. The stability of RNA was analyzed using the Click-iT nascent RNA capture kit according to the manufacturer’s instructions (Life Technologies, C10365).

Western blotting and coimmunoprecipitation (Co-IP) analyses

Western blotting analyses were performed as previously described [60]. Briefly, proteins from each sample were separated on SDS-PAGE gel (Bio-Rad, 3,540,124) and then electrophoretically transferred onto PVDF membranes (Millipore Sigma, IPVH00010). The membranes were incubated with primary antibodies overnight at 4°C and then incubated with HRP-conjugated secondary antibodies for 45 min at 37°C. Protein signals were detected using ECL solution (Millipore Sigma, 345,818), and band intensities were quantified using Image J software (National Institutes of Health). Additionally, ACTB and LMNB1 (lamin B1) were chosen as cytosolic and nuclear markers, respectively. To measure protein stability, TM3 cells were treated with CHX at 100 μg/mL for indicated times, after which western blotting was used to examine protein levels of METTL14 and PPM1A. Moreover, Co-IP was performed as previously established protocols to assess interaction of YTHDF1with EIF3A and EEF2 [60]. Briefly, anti-YTHDF1 antibody was used as the precipitating antibody to first isolate YTHDF1 from cell lysates by immunoprecipitation, followed by western blotting analyses using antibodies against EIF3A and EEF2.

RNA-immunoprecipitation (RIP)-qRT-PCR analyses

RIP analyses were conducted using EZ-Magna RIP^TM RNA-Binding Protein Immunoprecipitation Kit (Millipore Sigma, 17–700) according to the manufacturer’s instruction. Briefly, cells were lysed with RIP lysis buffer containing protease and RNase inhibitor, after which cell lysate supernatant was incubated with anti-YTHDF1/YTHDF2 antibody-conjugated beads overnight at 4°C. After being washed with IP buffer for three times, the beads were then treated with proteinase K (Millipore Sigma, 71,049) at 65°C for 0.5 h with occasional shaking. At last, the supernatant was mixed with TRIzol reagent and total RNA was then isolated. Camkk2 and Ppm1a mRNA levels were detected by qRT-PCR, and primer sequences were listed in Table S1.

m⁶A RIP-qRT-PCR analyses and measurement of cellular m⁶A levels

Magna MeRIP^TM m⁶A kit was chosen to assess m⁶A modification levels in target mRNA according to the manufacturer’s instructions (Millipore Sigma, 17–10,499). Briefly, the isolated RNA needs to be fragmented by RNA fragmentation buffer. After saving one tenth of the total RNA as input, the remaining RNAs were performed for immunoprecipitation with m⁶A antibody (Millipore Sigma, ABE572) coated on magnetic beads A/G (Millipore Sigma, 16–663). After being washed with immunoprecipitation buffer for three times, m⁶A-modified RNAs were eluted with elution buffer. Finally, eluted RNAs were recovered by ethanol precipitation and then subjected to qRT-PCR assay. Specific primer information about Camkk2 and Ppm1a was list in Table S2. Moreover, total cellular m⁶A modification levels were detected by EpiQuik m⁶A RNA Methylation Quantification Kit (Epigentek, P-9005-96), and measurement were carried out in triplicate according to manufacturer’s instructions.

Immunofluorescence analyses

Immunofluorescent staining of testicular tissues was performed as previously described [61]. Rabbit anti-m⁶A, mouse anti-HSD3B, rabbit-anti METTL14, rabbit-anti ALKBH5, rabbit-anti CAMKK2, and rabbit anti-PPM1A were added to the tissue sections and incubated at 4°C overnight, followed by incubation with secondary antibodies. The images were acquired using a FV3000 confocal fluorescence microscope (Olympus, Tokyo, Japan). For cellular immunofluorescence assay, the cells plated onto cover glasses were fixed with 4% paraformaldehyde, and incubated antibodies according to protocols above. In this study, a total of 3 independent samples were included for quantitative analysis, and 15 Leydig cells for each sample were collected.

Chromatin immunoprecipitation (ChIP) assay

Pierce^TM Agarose ChIP Kit was chosen to assess binding ability of CEBPB and TFEB to ALKBH5 promoter according to manufacturer’s instruction (Thermo Fisher Scientific, 26,156). DNA immunoprecipitated with anti-CEBPB or TFEB was subjected to qRT-PCR assay. Primer sets targeting those regions containing potential CEBPB or TFEB binding sites in ALKBH5 promoter were provided in Table S3.

Luciferase report assay

The indicated cells were seeded in 12-well plates and transfected with pmirGLO luciferase vector (Promega, E1330) fused with wild-type or mutated Ppm1a-3ʹUTR or Camkk2. The firefly luciferase and Renilla luciferase activity in each cell were detected by dual-luciferase reporter assay system (Promega, E1910). TM3 cells were transfected with pGL3-Basic vector (Promega, E1751) fused with Alkbh5 promoter sequence (−2000 ~ −1), after which promoter luciferase activity was measured. Moreover, luciferase activity needs to be normalized by co-transfection with pRL-TK-Renilla vector (Promega, E2241).

Animal experiments and patient sample collection

Male SPF BALB/c mice (qls02-0202) were purchased from Qinglongshan animal breeding farm. Mice were sacrificed by CO₂ asphyxiation and testes were obtained for following histopathological analyses. Moreover, serum testosterone levels were measured by radioimmunoassay. In this study, the 14 clinical testicular specimens from azoospermia and oligospermia patients were collected from Center of Reproductive Medicine at the First Affiliated Hospital of Nanjing Medical University (Nanjing, China) with appropriate informed consent from patients. Based on levels of serum testosterone, these patients were further classified into low testosterone group (testosterone < 10.4 nM, n = 9) and normal testosterone group (testosterone > 10.4 nM, n = 5).

Statistical analysis

The data were analyzed for statistical significance with SPSS 18.0 (SPSS, Chicago, IL, USA). One-way ANOVA with Duncan’s post hoc test was used to analyze the difference between groups; the Student’s t-test was used for paired comparisons. A p value of less than 0.05 was considered statistically significant.

Disclosure statement

No potential conflict of interest was reported by the authors.

Supplementary material

Supplemental data for this article can be accessed here.

Additional information

Funding

This work was supported by the National Natural Science Foundation of China (31901182, 31870492, 31971517 and 31670519) and Fundamental Research Funds for the Central Universities (0214-14380438).