Fibroblast growth factor homologous factor 1 stimulates Leydig cell regeneration from stem cells in male rats

Abstract Fibroblast growth factor homologous factor 1 (FHF1) is an intracellular protein that does not bind to cell surface fibroblast growth factor receptor. Here, we report that FHF1 is abundantly present in Leydig cells with up‐regulation during its development. Adult male Sprague Dawley rats were intraperitoneally injected with 75 mg/kg ethane dimethane sulphonate (EDS) to ablate Leydig cells to initiate their regeneration. Then, rats daily received intratesticular injection of FHF1 (0, 10 and 100 ng/testis) from post‐EDS day 14 for 14 days. FHF1 increased serum testosterone levels without affecting the levels of luteinizing hormone and follicle‐stimulating hormone. FHF1 increased the cell number staining with HSD11B1, a biomarker for Leydig cells at the advanced stage, without affecting the cell number staining with CYP11A1, a biomarker for all Leydig cells. FHF1 did not affect PCNA‐labelling index in Leydig cells. FHF1 increased Leydig cell mRNA (Lhcgr, Scarb1, Star, Cyp11a1, Hsd3b1, Cyp17a1, Hsd17b3, Insl3, Nr5a1 and Hsd11b1) and their protein levels in vivo. FHF1 increased preadipocyte biomarker Dlk1 mRNA level and decreased fully differentiated adipocyte biomarker (Fabp4 and Lpl) mRNA and their protein levels. In conclusion, FHF1 promotes Leydig cell regeneration from stem cells while inhibiting the differentiation of preadipocyte/stem cells into adipocytes in EDS‐treated testis.

(LH) receptor (LHCGR) for receiving LH trophic stimulation and express cholesterol transport proteins, high-density lipoprotein receptor (SCARB1) and steroidogenic acute regulatory protein (STAR), as well as steroidogenic enzymes, including cytochrome P450 cholesterol side chain cleavage enzyme (CYP11A1), 3β-hydroxysteroid dehydrogenase isoform 1 (HSD3B1), cytochrome P450 17α-hydroxylase/17,20-lyase (CYP17A1) and 17β-hydroxysteroid dehydrogenase isoform 3 (HSD17B3). 1 ALCs also secrete insulin-like 3 to regulate spermatogenesis. 6 The LC regeneration is very similar to the LC developmental process during puberty. 7 SLCs begin to commit into progenitor Leydig cells (PLCs) on post-EDS day 14 and then they further differentiate into immature Leydig cells (ILCs) on post-EDS day 28, when the biomarker 11β-hydroxysteroid dehydrogenase isoform 1 (HSD11B1) begins to be expressed in these advanced cells. 7 The proliferation and differentiation of SLCs in the LC lineage is controlled by a set of growth factors and hormones. 1 Although several critical growth factors such as platelet-derived growth factor AA, 8 dessert hedgehog, 9 and kit ligand, 10 have been reported, the regulatory growth factors are largely unknown.
By re-analyzing the transcriptome of LC lineage cells, 11 we identify fibroblast growth factor homologous factor 1 (FHF1), which is significantly up-regulated from SLCs/PLCs into ILCs and further into ALCs, indicating that this peptide might exert autocrine effects on LC development. FHFs form a subfamily of proteins, which have sequences and structures similar to fibroblast growth factors (FGFs). 12 FHFs are sometimes named according to FGF nomenclature (FHF1 = FGF12, FHF2 = FGF13, FHF3 = FGF11, FHF4 = FGF14).
However, FHFs and FGFs have unrelated functions. FGFs bind to the cell surface receptor with tyrosine kinase activity. 13 However, FHFs are expressed as intracellular peptides that bind to some intracellular proteins 14 or cytoplasmic tails of some ion channels. 15,16 FHF members play a critical role in developmental cell processes and neuron excitability. 15,17 FHF1 is one of the FHF members that are widely expressed in many tissues such as cartilaginous skeleton, neuron, heart and testis, suggesting a role in the development of these tissues. 14,18 Here, we report that FHF1 stimulates LC development in rats.

| FHF1 increases serum T levels in vivo
After re-analyzing the transcriptome of SLCs, PLCs, ILCs and ALCs, 11 we found that the main expression levels (arbitrary unit) of Fhf1  20,21 We intratesticularly injected FHF1 (0, 10 or 100 ng/testis/ day) starting on post-EDS day 14 up to day 28 ( Figure 1A). After the treatment, FHF1 did not affect bodyweights and testis weights when compared to the control (Table S1). FHF1 dose-dependently increased serum T levels with significance recorded at 100 ng/testis on post-EDS day 28 ( Figure 1B). However, it did not alter serum LH ( Figure 1C) and FSH ( Figure 1D) levels. These data suggest that FHF1 promotes LC regeneration primarily via direct action within the testis.

| FHF1 increases HSD11B1-positive LC number in vivo
Elevation of serum T levels could be contributed by the increase of LC number at the advanced stage. We used two biomarkers to label F I G U R E 1 FHF1 experimental protocol and serum testosterone (T), LH and FSH levels after in vivo FHF1 treatment. A, Experimental protocol; B-D, Serum T, LH, and FSH levels. Mean ± SEM, n = 8. Asterisks (*) designate significant difference from the control (FHF1, 0 ng/ testis) at P < 0.05 respectively LCs: CYP11A1 (representing all LCs) and HSD11B1 (representing LCs at the advanced stage). 7,20 As shown in Figure 2, FHF1 did not change the number of CYP11A1-positive LCs. However, FHF1 significantly increased the number of HSD11B1-positive LCs at 100 ng/ testis. This indicates that more LCs reach the advanced stage in the LC lineage after FHF1 treatment. We also measured SOX9-positive cells and we found that FHF1 did not alter SOX9-positive Sertoli cell number ( Figure 2L).

| FHF1 does not increase the PCNA labelling index in LCs in vivo
We used PCNA to label the proliferating cells and CYP11A1 to stain all LCs. As shown in Figure S1, we did not find that FHF1 increased the ratio of PCNA labelling in CYP11A1-positive LCs.
These results indicate that the increased number of HSD11B1positive cells is not contributed by LC proliferation but possibly by its differentiation.
We asked whether the increased number of HSD11B1-positive LCs came from the proliferation of PLCs. PLCs have a higher capacity of cell division. 22 We isolated PLCs and performed cell cycle analysis after in vitro FHF1 treatment. FHF1 did not affect the percentage of cells entering the S and G2 phases ( Figure S2).
The result indicates that FHF1 does not alter PLC mitosis.

| FHF1 promotes LC differentiation in vivo
We performed RNA sequencing analysis to explore the effects of FHF1 (100 ng/testis) on LC gene expression. We sequenced 14,028 transcripts in the testis of two groups (0 and 100 ng/testis FHF1). Among these transcripts, 197 transcripts were significantly up-regulated (P < 0.05) and 99 transcripts were significantly down-regulated (P < 0.05) in the FHF1 group when compared to the control ( Figure 3A,B). GO analysis showed that most up-regulated genes are related to the categories including response to gonadotropin, dioxin metabolic process and steroid metabolic F I G U R E 2 Leydig cell (LC) and Sertoli cell (SC) numbers in the testes after in vivo FHF1 treatment Immunohistochemical staining of CYP11A1 (Panels A-C), HSD11B1 (Panels E-G) and SOX9 (Panels I-K) of the testes from the rats treated with 0, 10 and 100 ng/testis FHF1 on post-EDS day 28. Panels A, E and I: the control; Panels B, F and J: 10 ng/testis FHF1; Panels C, G and K: 100 ng/testis FHF1; Panel D, H and L: quantitative data. Black arrow indicates CYP11A1-and HSD11B1-positive LCs and black arrowhead indicates SOX9-positive SCs. Bar = 50 mm. Mean ± SEM, n = 8, * P < 0.05 when compared to the control Figure 3C) and most down-regulated genes are related to the categories including positive regulation of T cell differentiation, negative regulation of protein processing and regulation of cell development ( Figure 3D).
In the LC steroidogenic pathway, gene expression of LC genes (Star, Cyp11a1, Hsd3b1, Cyp17a1 and Insl3) was up-regulated by ≥2fold ( Figure 4). We further verified the LC gene expression and compared it to the Sertoli cell genes using qPCR. As shown in Figure 5, all these LC genes were up-regulated whereas Sertoli cell genes (Sox9, Fshr, Amh and Dhh) were not altered.
We measured the levels of these LC gene products using Western blot. We found that these proteins had similar changes in their respective mRNA levels ( Figure 6). In addition, we also used the semi-quantitative measurement of CYP11A1, HSD11B1, and SOX9 densities in the testis and we found that their densities were similar to the Western blotting data (Figure 7).
When SLCs and PLCs differentiate into ILCs, they increase cell size and cytoplasmic size. 23 We used CYP11A1 to stain LCs to measure the LC metrics. As shown in Figure 7E,J,O, FHF1 significantly increased LC size and cytoplasmic size without affecting the nuclear size in the 100 ng/kg group, indicating that FHF1 stimulates the growth of LCs. These data suggest that FHF1 promotes LC regeneration.

| FHF1 prevents the transition of stem cells into adipocyte in vivo
Our previous study demonstrated that SLCs are multipotent stem cells and can differentiate into adipocytes. 24 We performed a

| FHF1 regulates the signalling pathways
We measured the levels of total proteins (SIRT1, PGC-1α and AKT1) and the phosphorylated protein (pAKT1) in the testis after FHF1 treatment. AKT1 level was not changed, whereas phosphorylated AKT1 (pAKT1) levels were significantly increased at 100 ng/testis FHF1 group ( Figure 9). FHF1 significantly increased SIRT1 and PGC-1α levels at 100 ng/testis ( Figure 9). We further treated PLCs with 10 and 100 ng/mL FHF1 for 24 hours and found that FHF1

| D ISCUSS I ON
Although classic FGFs belong to a family of signalling proteins that bind to the cell surface FGF receptors and play diverse roles in cell growth, differentiation, morphogenesis and developmental processes, [25][26][27][28] FHFs are not exact FGF members and are intracellular non-secretory proteins. 29 FHFs lack a signal sequence and cannot be released from cells 30 to activate FGF receptors. 31 FHF1 binds to islet brain-2 and voltage-gated sodium channels and plays a critical role in the membrane targeting and ion channel function. 14   The previous study demonstrated that SLCs are multipotent stem cells and are able to differentiate into either the LC lineage or the adipocyte lineage. 24 Using RNA sequencing and qPCR as well as Western blotting, we found that biomarkers of mature adipocytes  The SIRT1/PGC-1α signalling may also be involved in FHF1mediated action. SIRT1 is an NAD-dependent class III histone deacetylase and plays significant roles in many biological activities, including development, gene modification and metabolism via deactivation. 48    This time-course of administration regimen was adopted because

| Serum T measurement
Immulite2000 Total T kit was employed to measure the serum T concentrations. The lower detection limit of serum T concentrations was 0.2 ng/mL.

| ELISA for serum LH and FSH levels
According to the manufacturer's instructions (Chemicon CA), each ELISA kit was used to detect the serum levels of LH and FSH. Briefly, sample and assay diluent was mixed, washed and incubated with peroxidase-conjugated IgG anti-LH or anti-FSH. Then, the conjugated complex was washed and substrate was added for the reaction. The levels of LH and FSH were measured at 550 nm using a microplate reader with a correction wavelength at 450 nm.

| RNA sequencing
Sequencing analysis and base calling were conducted using Solexa pipeline v1.8 (Off-Line Base Caller software, v1.8, Illumina, Foster City, CA). Sequence quality was examined using the FastQC software. 57 The trimmed reads (trimmed 5′, 3′-adaptor bases) were aligned to a reference genome using Hisat2 software. 58 The transcript abundance for each sample was estimated with StringTie 59 and the FPKM 60 value for gene and transcript levels were calculated with R package Ballgown. 61 The differentially expressed genes and transcripts were filtered using R package Ballgown. 61 The novel genes and transcripts were predicted from assembled results by comparing to the reference annotation using StringTie and Ballgown and the coding potential of those sequences was then assessed using CPAT. 60 Principle Component Analysis and correlation analysis were performed according to gene expression level. Hierarchical Clustering, Gene Ontology, Pathway analysis, scatter plots and volcano plots were performed with the differentially expressed genes in R, Python or shell environment for statistical computing and graphics.

| Biological pathway analysis
Biological pathway analysis was performed as previously described. 62 GenMAPP2.1 (San Francisco, CA) was used to map the signal pathways of potential pathways. We illustrated the biological pathways containing differentially expressed genes by importing our statistical results into the program. The results of differential gene expression profiles were confirmed using qPCR.

| qPCR
Total RNAs were isolated from testes using Trizol. The concentrations of total RNAs were measured using NanoDrop 2000 (Thermo Scientific, Shanghai, China). The first-strand cDNA was synthesized and used as the template for qPCR as previously described. 63  and RNase-free water. The procedure for qPCR was set as following: 95°C for 5 minutes, followed by 40 cycles of 95°C for 10 seconds and 60°C for 30 seconds. Ribosomal protein S16 (Rps16) was used as the internal control, which was the house-keeping gene. The mRNA level of each gene was read as the Ct value and calculated using a standard curve method and was normalized to Rps16 as previously described. 64 The primers and gene names are listed in Table S2.

| Western blot
Western blot was performed as previously described. 65   . The house-keeping protein, ACTB, serves as a control. The density of target protein calculated by using J-Software was normalized to ACTB. All the antibodies used were listed in Table S3. In order to enumerate CYP11A1-or HSD11B1-positive LC numbers or SOX9-positive Sertoli cell number, sampling of the testis was performed according to a fractionator technique as previously described. 66 Briefly, each testis was cut in eight discs and two discs were randomly selected. Then, discs were cut into four pieces and one piece was randomly selected from a total of eight pieces. These pieces of testis were embedded in paraffin in a tissue array as above.

| Immunohistochemical staining of the testis and enumeration of LCs
Paraffin blocks were sectioned into 6-μm-thick sections. Ten sections were randomly sampled from each testis per rat. Sections were used for immunohistochemical staining. Images were taken using a digital camera, under a 10 × objective and total microscopic fields per section were counted. The histochemical staining was performed as above. The total number of LCs or Sertoli cells was calculated by multiplying the number of LCs or Sertoli cells counted in a known fraction of the testis by the inverse of the sampling probability.

| Immunofluorescent staining of the testis
Immunofluorescent staining was performed to investigate the effects of FHF1 on the proliferation of LCs. Sections were incubated with the primary antibody of PCNA for 60 minutes and then washed and incubated with the CYP11A1 antibody for double staining. Fluorescent secondary antibody (Alexa-conjugated anti-rabbit or anti-mouse IgG, 1:500) were used after the primary antibody. Sections were counterstained with mounting medium containing DAPI. Sections were visualized under a fluorescent microscope (Olympus, Tokyo, Japan). The CYP11A1 (green colour) was used to label LCs and PCNA (red colour) was used to label proliferating cell nucleus.

| PLC isolation
PLCs were isolated as described previously. 67 Eighteen 21-dayold Sprague Dawley rats were killed by asphyxiation with CO 2 . In brief, the removed testis was digested with collagenase and DNase substrate. 68 More than 95% of PLCs were stained, indicating that the purity of PLCs was high.

| PLC culture and cell cycle assay
PLCs were seeded into the 6-well culture plates after isolation with a cell density of 10 6 cells/well. PLCs were treated with 0, 10 and

| Statistical analysis
Data are expressed as mean ± SEM. P < 0.05 was considered statistically significant. The differences among groups were evaluated by unpaired student t test when two groups were compared or by oneway ANOVA followed by ad hoc Dunnett's multiple comparisons to compare with the control when three or more groups were compared. GraphPad version 6 software was used for statistical analysis (GraphPad Inc, CA). Science and Technology Bureau (ZS2017009).

CO N FLI C T O F I NTE R E S T
The authors declare that they have no conflict of interest to declare.