Loss of tyrosine kinase receptor Ephb2 impairs proliferation and stem cell activity of spermatogonia in culture

Abstract

Germline stem and progenitor cells can be extracted from the adult mouse testis and maintained long-term in vitro. Yet, the optimal culture conditions for preserving stem cell activity are unknown. Recently, multiple members of the Eph receptor family were detected in murine spermatogonia, but their roles remain obscure. One such gene, Ephb2, is crucial for maintenance of somatic stem cells and was previously found enriched at the level of mRNA in murine spermatogonia. We detected Ephb2 mRNA and protein in primary adult spermatogonial cultures and hypothesized that Ephb2 plays a role in maintenance of stem cells in vitro. We employed CRISPR-Cas9 targeting and generated stable mutant SSC lines with complete loss of Ephb2. The characteristics of Ephb2-KO cells were interrogated using phenotypic and functional assays. Ephb2-KO SSCs exhibited reduced proliferation compared to wild-type cells, while apoptosis was unaffected. Therefore, we examined whether Ephb2 loss correlates with activity of canonical pathways involved in stem cell self-renewal and proliferation. Ephb2-KO cells had reduced ERK MAPK signaling. Using a lentiviral transgene, Ephb2 expression was rescued in Ephb2-KO cells, which partially restored signaling and proliferation. Transplantation analysis revealed that Ephb2-KO SSCs cultures formed significantly fewer colonies than WT, indicating a role for Ephb2 in preserving stem cell activity of cultured cells. Transcriptome analysis of wild-type and Ephb2-KO SSCs identified Dppa4 and Bnc1 as differentially expressed, Ephb2-dependent genes that are potentially involved in stem cell function. These data uncover for the first time a crucial role for Ephb2 signaling in cultured SSCs.

The tyrosine kinase receptor Ephb2 is required for normal proliferation and stem cell activity of adult mouse spermatogonia in long-term culture.

Introduction

Near life-long production of fertile gametes in the mammalian testis is supported by a complex hierarchy of spermatogenic cells that reside in the seminiferous tubules, where undifferentiated spermatogonia (Spg) are situated along the basement membrane [1, 2]. Within the pool of undifferentiated mouse Spg, classification of cells is based on state of differentiation. Accordingly, A single (A_s) Spg are considered less differentiated than A paired (A_pr) and A aligned (A_al) Spg [3]. Therefore, the pool of spermatogonial stem cells (SSC) is thought to be largely contained within the A_s Spg. However, only a small fraction of A_s Spg are SSCs, and substantial functional plasticity has been observed within the undifferentiated Spg pool in vivo [4, 5]. Prospective identification of adult mouse SSCs has been challenging historically due to a dearth of highly specific cell surface markers, such that either genetic reporters or multiple surface markers are commonly required [6–10].

The maintenance of SSCs, both in vivo and in culture systems, relies on their self-renewal capacity. The promise of limitless cultures of pure SSCs from different species, including humans, would enable a large variety of yet unrealized experimental and clinical opportunities. Two important components for SSC culture are glial cell line-derived neurotrophic factor (GDNF) and fibroblast growth factor (FGF), which are produced in vivo by testicular somatic cells, including Sertoli cells [11] and peritubular myoid cells [12]. SSCs express the GDNF receptor complex RET/GFRα1 and FGF receptors [11, 13]. Ligand-receptor binding induces activation of the MAPK/ERK and AKT/PI3K pathways [14, 15], leading to the expression of downstream targets, including transcription factors Etv5, Lhx1, Fos, and Bcl6b that orchestrate self-renewal [16–18]. A variety of feeder cell types and media formulations have been employed to derive and maintain long-term primary cell lines, typically from early post-natal mice. Such lines preserve SSC activity upon transplantation analysis [19, 20]. Yet, only a minority of Spg in culture possesses SSC activity, for reasons that remain unclear [21]. However, recent work has revealed further unanticipated complexity of both SSC culture systems and the transplantation assay by which SSC cultures are evaluated. First, SSCs transition in vitro between distinct molecular phenotypes [4]. Second, the challenged environment of the regenerating recipient testis enables differentiation-primed cells to exhibit stem cell activity [2, 22]. Thus, additional study of culture systems is urgently required to understand these dynamic changes and improve current approaches for optimal pre-clinical modeling.

We previously exploited a CD34⁺ somatic feeder cell line to efficiently derive long-term, transplantable mouse SSC lines, specifically from adults [23, 24]. A critical goal is to enhance adult SSC culture protocols by increasing the proportion of functional stem cells. Toward that end, we searched for novel genes that could co-regulate known self-renewal and/or proliferation signals.

The Eph receptor gene family is the largest known family of tyrosine kinase receptors and includes several members expressed in the male germline [25, 26]. Eph receptors control cell growth, survival, migration, or differentiation in a variety of physiological and pathological contexts [27, 28]. Activation of Eph receptors relies on binding with plasma membrane-tethered ephrin ligands. Thus, Eph receptors regulate heterotypic or homotypic cell–cell communication between neighbors, although the Eph receptor/ligand interaction can also occur in a cell-autonomous manner (i.e., within the same membrane) [29]. Eph receptor genes are classified into two groups: nine Eph receptor A (EphA) genes (Epha1, Epha2, Epha3, Epha4, Epha5, Epha6, Epha7, Epha8, and Epha10) and five Eph receptor B (EphB) genes (Ephb1, Ephb2, Ephb3, Ephb4, and Ephb6) [30, 31]. In general, the Eph receptors bind to ligands within the EphrinA and EphrinB subfamilies, respectively, but some exceptions exist. All Eph receptors have a common structure composed of extracellular domains (ligand binding; Sushi; EGF-like; Fibronectin type), a transmembrane domain, and intracellular domains [juxtamembrane; kinase; sterile alpha motif (SAM); PDZ]. A unique feature of the Eph receptor/Ephrin interaction is that signaling can occur downstream of the receptor (forward) and/or downstream of the ligand (reverse) [32–34]. A plethora of context-dependent downstream targets have been described in the literature, such as PI3K/AKT, MAPK/ERK, Rho GTPases, STAT2, and STAT3 [32, 35, 36].

Ephrin signaling is important for stem cell maintenance in several organs. In the brain, Ephrin signaling maintains neuronal stem cells and supports their interaction with the niche [37, 38]. In the intestine, Ephb2 is necessary for self-renewal of colonic stem cells [39]. In the testis, expression of Eph receptors has been noted in a handful of studies. At the transcriptional level, Mutoji et al. [26] found greater expression of Epha2 in an SSC-enriched germ cell population. Furthermore, EPHA2 selection was found to enrich for SSCs [40]. Guo et al. [25] demonstrated elevated Ephb2 expression in Spg by RT-PCR. More recently, the presence of EPHB2 and EPHB4 in germ cells was shown by immunostaining of mouse testicular tissue [41]. The prior association between Ephb2 and stem cells in the colon and the expression of Ephb2 in Spg together raised the possibility that Ephb2 could contribute to SSC self-renewal.

Here, we sought to investigate the functional significance of Ephb2 in SSCs. First, we detected EPHB2 in humans and mouse spermatogonia by immunostaining and flow cytometry, respectively. We found that Ephb2 loss in cultured SSCs significantly abrogated cell proliferation. In parallel, transplantation experiments showed that SSCs lacking Ephb2 had significantly reduced colonization capacity. Using transcriptomic analyses and rescue experiments, we identified Dppa4 and Bnc1 as potential downstream targets of the EPHB2 signaling in cultured SSCs. Overall, these results suggest that EPHB2 signaling regulates spermatogonial maintenance in vitro.

Materials and methods

Ethics

All mouse experiments were performed in accordance with national guidelines as per the Weill Cornell Medical College Institutional Animal Care and Use Committee and Society for the Study of Reproduction. Normal human cadaver testis tissue was obtained from the New York Organ Donor Network with consent of next of kin, as described previously [42].

SSC lines and cell culture

Primary SSC lines were derived from adult mice on a mixed C57Bl6/129 background and maintained on mitotically inactivated CD34⁺ somatic feeder cells [23]. SSC cultures were established from freshly dissected testes by two-step enzymatic digestion. Briefly, detunicated testes collected in a sterile fashion were minced and digested in two subsequent steps with collagenase I (Worthington Biochemical) followed by trypsin. The dissociated testis suspensions were plated in gelatin-coated 6-well plates at a density of 1/8 testis per well in 2 ml final volume of cell culture media. Half of the media was changed after 72 h and every 48 h afterward preserving at least 1 ml of conditioned media in each passage. One to four weeks after plating, SSC colonies were manually picked and plated into mitotically inactivated CD34⁺ feeder cells for long-term culture and expansion using appropriately weekly passaging onto fresh feeder cells [23]. Cell culture media for SSC maintenance consisted of StemPro-34 (Thermo Fisher Scientific, No 10640) with nutrient supplement (Thermo Fisher Scientific, No 10641) plus additional supplements and growth factors at the following concentrations: 20 ng/ml GDNF (Novus Biologicals), 10 ng/ml FGF2 (Thermo Fisher Scientific and Novus Biologicals), 25 μg/ml human insulin (Sigma-Aldrich), and 20 ng/ml EGF (Thermo Fisher Scientific) as described previously [43].

Lentiviral transduction

Labeled SSC lines were generated by transduction with a lentivirus carrying mCherry driven by the human PGK promoter (pCCL4-PGK-mCherry). Rescue SSC lines were generated with lentiviruses carrying either an Ephb2 transgene driven by the human PGK promoter (pCCL4-PGK-Ephb2) or a Dppa4 transgene driven by the murine EF1A promoter (pLV-EF1A-Dppa4). Transduction was carried out in feeder-free SSC cultures overnight in cell culture media containing polybrene (5 μg/ml). In respective experiments, WT, KO, and Rescue SSC lines were transduced simultaneously, at the same passage number. A lentivirus driving mCherry at an equivalent dose served as a negative control for re-expression of Ephb2 or Dppa4. Transduction efficiency was monitored by fluorescence microscopy for the mCherry reporter and by immunoblot for Ephb2 and Dppa4 transgenes.

CRISPR-Cas9-mediated knockout induction in SSCs

To disrupt gene expression, 10⁵ SSCs were transiently transfected by electroporation with 0.5 μg of plasmid pSp-Cas9(BB)-2A-EGFP backbone, containing the Cas9 enzyme, sgRNA targeting Ephb2 (or no sgRNA for wild-type control), and the fluorescent marker GFP (Neon Transfection System, Thermo Fisher). On the day after transfection, GFP-positive cells were sorted by FACS with a BD FACS Aria II instrument (Beckman Coulter) and cultured for several days for expansion. Cellular clones were then isolated, by serial dilution of a cell suspension followed by single-cell plating, so as each clone could then be expanded and analyzed for the presence of a genetic mutation at the targeted locus and for EPHB2 level. Potential off-target genes corresponding to out gRNA sequence are listed in Supplementary Table S1.

Fluorescence-activated cell sorting

For EPHB2 FACS, cultured cells were collected by trypsinization and then incubated with anti-EPHB2 antibody (AF467, R&D Systems) at 1 μg/ml for 1 h at 4 °C. For p-ERK, p-SRC, and Cleaved Caspase-3, an intracellular flow cytometry protocol was used (Supplementary Table S2). Cultured cells were trypsinized, then fixed in PFA (4%) and permeabilized with 90% methanol. Cells were then incubated with p-ERK antibody (#9101, Cell Signaling, 1:200), p-SRC antibody (ab185617, Abcam, 1:50), or cleaved caspase-3 antibody (#9661, Cell signaling, 1:800) for 1 h at 4 °C.

Cell suspensions were then subject to FACS analysis on Attune NxT Flow cytometer (Thermo Fisher Scientific).

For MCAM FACS, testes from Gfra1-CreERT2; Rosa26-LSL-tdTomato mice were dissociated <3 months after tamoxifen administration (100 mg/kg for 4 days) [43]. Single-cell suspensions were generated from testes dissected from adult mice by two-step enzymatic digestion as described above, replacing collagenase I by collagenase IV. The single-cell suspensions were incubated with Alexa Fluor 647 anti-MCAM antibody (ME-9F1, BioLegend) at a concentration of 6 mg/ml for 1 h at 4 °C. Cell suspensions with immunocomplexes were subject to FACS with a BD FACS Aria II instrument (Beckman Coulter). DAPI was used for live/dead cell discrimination.

SSC transplantation

SSC transplantation assays were performed as described previously using busulfan-conditioned mice [43, 44]. Briefly, C57BL/6J male mice at 6–8 weeks of age were treated with 40 mg/kg of busulfan and used 4–8 weeks later as recipients. Cultured SSCs labeled by transduction with an mCherry fluorescent reporter (as described above) were dissociated with 0.05% trypsin/EDTA, treated with 1 mg/ml DNase I, resuspended in standard culture medium containing DNase I, and sterile-filtered trypan blue. We used 7000 SSCs in 10 μl total volume for transplantation into each testis by microinjection into the efferent ducts injecting the same number between experimental and control lines for each experiment. Red, mCherry-positive colonies were counted 6–8 weeks later under a fluorescent stereoscope (Nikon) in detunicated testes.

Immunofluorescent staining of testicular tissue

For the human staining, cryopreserved testicular tissue was fixed with 4% PFA, then blocked with 10% normal donkey serum before incubation with antibodies targeting EPHB2 (ab150652, Abcam) and DDX4 (AF2030, R&D Systems) overnight at 4 °C in a humidified chamber. For the mouse whole mount staining, testes were fixed with 4% PFA overnight at 4 °C, then incubated in blocking buffer (PBS 0.1% Tween, 3% BSA) before incubation with antibodies for EPHB2 (AF467, R&D Systems) and DDX4 (ab13840, Abcam). Detection of primary antibodies was performed using Biotin-SP Donkey Anti-Goat IgG (705-065-147, Jackson ImmunoResearch) and Biotin-SP Donkey Anti-Rabbit IgG (711-065-152, Jackson ImmunoResearch) at 1 μg/ml for 90 min at room temperature (RT) followed by 1 μg/ml of Streptavidin Alexa Fluor 555 (S-21381, Thermo Fisher Scientific) or Alexa Fluor 488 (016-540-084, Jackson ImmunoResearch) for 30 min at RT. Slides were counterstained and mounted with VECTASHIELD HardSet Antifade Mounting Medium with DAPI (H-1500, Vector Laboratories) and analyzed using a Zeiss LSM 800 confocal microscope (Carl Zeiss Inc.).

Immunoblot

Feeder-free SSCs were suspended in lysis buffer containing PMSF (Sigma-Aldrich) and protease and phosphatase inhibitor cocktails (Sigma-Aldrich). Total cell lysates were separated by SDS-PAGE and transferred to PVDF membrane (BioRad Laboratories). Membranes were blocked in Tris-buffered saline containing 0.1% Tween-20 (TBST) with 3% BSA and incubated at 4 °C overnight with the following primary antibodies: EPHB2 (AF467, R&D Systems) or DPPA4 (AF3730, R&D systems). ACTB (A1978, Sigma-Aldrich) and PPIB (PA1-027A, Thermo Fisher) were used as loading controls. Blots were washed with TBST and incubated at room temperature with the appropriate secondary antibody conjugated to horseradish peroxidase. Chemiluminescent immunodetection was performed with ECL system (GE Healthcare Bio-Sciences), and membranes were exposed to X-ray film, and digitized scans of the blots were analyzed using ImageJ (National Institutes of Health, USA for densitometry).

EdU assay

SSCs were plated in gelatin-coated wells at 2–3 × 10³ cells/mm² and maintained in complete culture media in the presence of 5uM EdU (Click-iT Plus Edu 488 Imaging Kit) for 16 h. Cells were collected and subject to the fixation/permeabilization protocol (see FACS section). EdU detection was then performed according to the Click-iT protocol, and cells were analyzed by FACS on the FACS Attune NxT instrument.

Reverse transcription quantitative reverse transcription polymerase chain reaction

Total RNA was extracted from feeder-free cultured SSCs or MCAM+ populations isolated by FACS using the RNeasy Plus micro extraction kit (Qiagen) or Arcturus Pico Pure with on-column DNA digestion (Qiagen), respectively. RNA was reverse transcribed using qScript cDNA SuperMix (Quanta Biosciences). QPCR was performed in triplicates in a LightCycler 480II real-time System (Roche Molecular Systems) using Sybr Select Master Mix (Applied Biosystems). For quantification, each technical triplicate was normalized to endogenous Actb and Gapdh, and relative transcript expression was calculated using the comparative CT method (2^−ΔΔCt method) using one of the technical triplicates of the control condition as the reference value. Presented data were normalized to Actb. The following genes were detected using indicated forward (F) and reverse (R) primer sequences:

Actb : F 5′-GAGAAGATCTGGCACCACACC-3′, R 5′-GGTCTCAAACATGATCTGGGTC-3′; Bcl6b: F 5′-CATCCAAGCCAGCTATGAACC-3′, R 5′-AGTAGGTGGGTCTGGGTGTCC-3′; Bnc1: F 5′-GTGAAGACGAGCCAGAGGAC-3′, R 5′-TGTTCAGCCTCCCTCTCAGT-3′; Ephb2: F 5′-ACTATGGCGGCTGTATGTCC-3′, R 5′-GCACATCCACTTCTTCAGCA-3′. Etv5: F 5′-TGCCTAACTGCCAGTCATCC-3′, R 5′-GATACATGGTGGGCTCCTGC-3′; Dppa4: F 5′-ATGTGGTGTGTGGTTCATGG-3′, R 5′-CATGTTGTCCTCCAGGTGTG-3′; Gfra1: F 5′-AGCAACAGTGGCAATGACCTG-3′, R 5′-AGTGGTAGTCGTGGCAGTGG-3′. Id4: F 5′-ACTACATCCTGGACCTGCAG-3′, R 5′-TGCTGTCACCCTGCTTGTTC-3′. Zbtb16: F 5′-TTTGCGACTGAGAATGCATTTAC-3′, R 5′-ACCGCATTGATCACACACAAAG-3′; Lhx1: F 5′-AGAGCGCGAAGCAAAGTGTT-3′, R 5′-CTCTTTGGCGACACTGCTGTT-3′; Efnb1: F 5′-GGCAAGCATGAGACTGTGAA-3′, R 5′-GCTTGCGGAGCTTGAGTAGT-3′; Efnb2: F 5′-GTGGGAGGTGACTGACTGGT-3′, R 5′-GCTGCGCTTTTTATTTCCAG-3′; Efnb3: F 5′CAGGCCCAATTTGTCTTGTT-3′, R 5′-CCATGGACCACCTCTTCACT-3′.

RNA sequencing

RNA extraction was performed on SSCs from three WT cellular clones and three Ephb2-KO cellular clones (n = 3). Mean RNA Integrity Number (RIN) was 9.7 ± 0.33 (mean ± SD). Sequencing was performed on a NextSeq500 with 81 bp single-end protocol. Statistical analysis was performed by comparing the FKPM (normalized gene expression) for each condition using DESeq2. Differentially expressed genes were included based on statistical significance (P < 0.05).

Data availability

RNA-seq files are available from the GEO database (series accession number GSE141401).

Statistical analyses

Results are presented as mean ± SD. At least three biological replicates and three technical replicates were performed for each experiment unless otherwise indicated in the text. The statistical significance was determined with GraphPad Prism (GraphPad Software), version 7.0 using the Student’s t-test or as otherwise indicated in each figure. Results were considered significant at P value < 0.05 (P < 0.05 is indicated as ^*; P < 0.01 is indicated as ^**; P < 0.001 is indicated as ^***).

Results

Ephb2 is expressed in undifferentiated Spg in vivo and in long-term cultured SSCs

Members of the Eph receptor family were previously found in adult humans and mouse germ cells [25, 26, 41]. By immunostaining with an EPHB2-specific antibody, we visualized EPHB2 immunoreactivity in normal adult human testis in a proportion of DDX4⁺ cells localized to the basement membrane of the seminiferous tubules (Figure 1A, left panel; white arrowheads). Similarly, whole mount staining of adult mouse seminiferous tubules revealed EPHB2 expression in DDX4⁺ spermatogonia-like cells (Figure 1A, right panel; white arrowheads). Consistent with these results, scrutiny of a recently published single-cell RNA-sequencing datasets revealed that Ephb2 mRNA expression was enriched in SSC-like populations of humans and mouse spermatogonia (Supplementary Figures S1 and S2) [45, 46].

Figure 1.

Ephb2 expression in Spg and induction of knockout by CRISPR-Cas9. (A) Anti-EPHB2 (red) and anti-DDX4 (green) as a marker for germ cells. Corresponding goat (Gt) and rabbit (Rb) IgG-negative controls are shown in the insets. Left: adult human testicular tissue section. Right: whole mount adult mouse testicular tissue. White arrowheads indicate selected EPHB2+ germ cells. Bar = 50 μm. (B) Cell suspensions were labeled with anti-MCAM and FACS sorted into fractions with varying MCAM intensity (P1–P4). (C) RT-qPCR showing Ephb2 expression in MCAM^Bright vs. MCAM^Medium or MCAM^Dim germ cells normalized to MCAM^Neg cells, using testicular cells from a wild-type (WT) adult mouse (left panel) or using germ cells from an adult Gfra1-CreER^T2; Rosa26-lox-stop-lox-tdTomato mouse (GCRT) 3 months after tamoxifen-induced tdTomato reporter recombination (right panel). (D) Schematic of SSC derivation. After a two-step enzymatic digestion, testicular cells from wild-type animals were plated. After 10 days, SSC colonies were manually picked, re-plated on a layer of feeder cells, and expanded with serial passaging to produce SSC lines. (E) EPHB2 expression in SSCs was measured by FACS using a goat polyclonal EPHB2 antibody and a normal goat IgG-negative control. (F) To disrupt the Ephb2 gene, SSCs were transiently transfected with the pSpCas9-2A-GFP plasmid containing sgRNA targeting Ephb2 exon 3 (or without sgRNA, to generate wild-type controls). GFP-positive SSCs were sorted, single-cell cloned, and expanded in vitro. After expansion, EPHB2 protein expression was assessed. Tables: list of expanded WT and Ephb2-KO SSC clonal lines and their EPHB2 protein status by FACS, with “++” denoting ~ 1 log increased fluorescence intensity vs. “null” or IgG control. (G) EPHB2 protein expression in representative WT and Ephb2-KO SSC clonal lines by FACS and immunoblot. (H) Ephb2 gene targeting; gDNA sequencing from KO cell clones, showing insertions and deletions at the expected locus on both alleles.

To characterize Ephb2 gene expression in adult undifferentiated mouse Spg in vivo, testicular cells were sorted by FACS based on MCAM expression level, which negatively correlates with germ cell differentiation [43, 47]. We then measured Ephb2 mRNA in the different MCAM-positive populations by RT-qPCR (Figure 1B). Ephb2 exhibited a ~ 50-fold, a ~ 30-fold, and a ~ 6-fold higher expression in the MCAM^Bright, MCAM^Med, and MCAM^Dim populations, respectively, compared to the MCAM^Neg population (Figure 1C, left panel). However, MCAM is also present in a fraction of testicular somatic cells [43], and a small number of MCAM-bright somatic cells with strong Ephb2 expression could potentially confound these results. Therefore, we utilized a transgenic mouse model with a Gfra1 promoter-driven, tamoxifen-dependent Cre plus an inducible tdTomato reporter (referred to as GCRT, described by Yamada et al. [43]), to obtain highly enriched undifferentiated Spg, free of somatic cells [43]. Enrichment for Ephb2 expression was found in the MCAM^Bright and MCAM^Med germ cell population, compared to the MCAM^Neg and MCAM^Dim germ populations (Figure 1C, right panel). Taken together, these data indicate that less differentiated Spg populations express higher levels of Ephb2 than more differentiated germ cells in humans and mouse.

We then investigated Ephb2 expression in cultured Spg in vitro. Our CD34⁺ feeder system enables maintenance of long-term adult SSC lines, preserving stem cell activity in a subpopulation of cells while enabling genetic manipulation prior to transplantation [23, 24, 43, 44, 48]. Briefly, testicular cells derived from adult mouse dissociated testes were plated in SSC growth medium without feeders for the first 1–2 weeks. Emerging germ cell colonies were then manually picked, expanded, and cryopreserved or serially maintained on CD34⁺ feeder-cells, as performed previously [43, 44] (Figure 1D). Colonization activity of expanded SSC lines was assessed in transplant experiments, confirming the presence of functional stem cells in our SSC cultures (see Figures 2, 3, and 5; [44]). We then analyzed EPHB2 in vitro by FACS analysis and observed expression in a large majority of cultured adult SSCs (Figure 1E).

Figure 2.

Ephb2 loss alters SSC proliferation rate and colonization capacity. (A) Cell recovery assay of WT and Ephb2-KO SSCs cultured on feeder cells over 7 weeks. (Mean ± SD of three experiments. ^*P < 0.05.) (B) Cell recovery assay of WT and Ephb2-KO SSCs on extracellular matrix (Matrigel) over 3 weeks. (Mean ± SD of three experiments. ^*P < 0.05.) (C) Representative phase contrast image of WT and Ephb2-KO SSCs in culture on feeder cells. Bar = 50 μm. (D) Quantification of apoptotic cells by staining for cleaved caspase-3 and FACS analysis in wild-type (WT) and Ephb2-KO SSCs (KO). (Bars = mean ± SD of three experiments.) Colony formation assay by transplant of mCherry-labeled SSCs into busulfan-treated mice. (E) Left: representative image of EdU staining of SSCs (cytospin) after 16 h of EdU incubation in wild-type (WT) and Ephb2-KO SSCs (KO). Right: representative FACS analysis of EdU+ SSCs after 16 h of EdU incubation. Quantification of EdU-positive SSCs by FACS analysis in wild-type SCCs (WT) and Ephb2-KO SSCs (KO). (Bars represent mean ± SD of three experiments. ^*P < 0.05, paired t-test.) (F) Representative image of whole-mount testes colonized with WT or Ephb2-KO SSCs (white arrows: colonies formed by transplanted SSCs). (G) Transplantation: representative experiment showing number of colonies formed by WT (n = 13) vs. Ephb2-KO SSCs (n = 9; unpaired t-test: ^***P < 0.001). (H) Transplantation: bars represent mean ± SD of three experiments (paired t-test: ^**P < 0.01). (I) RT-qPCR showing Bcl6b, Lhx1, Etv5, Gfra1, Zbtb16/Plzf, and Id4 in Ephb2-KO SSCs, normalized to WT samples. Bars represent mean ± SD of three experiments.

Figure 3.

Ephb2 re-expression restores proliferation but not stem cell activity. (A) Immunoblot showing EPHB2 expression in WT and Ephb2-KO SSCs after transduction with lentiviral mCherry (control) or hEphb2 (Ephb2 Rescue). PPIB = loading control. (B) Cell recovery assay of WT, Ephb2-KO, and Ephb2-Rescue SSCs cultured on feeder cells over 3 weeks. (Mean ± SD of three experiments; paired t-test: WT vs. Ephb2-KO: ^*P < 0.05; Ephb2-KO vs. Ephb2-Rescue: ^*P < 0.05.) (C) Quantification of EdU-positive SSCs by FACS analysis in wild-type (WT), Ephb2-KO (KO), and Ephb2-Rescue SSCs. (Bars represent mean ± SD of three experiments, t-test: ^*P < 0.05.) (D) Scatter plot of one representative experiment out of two showing the number of colonies formed by WT (n = 6), Ephb2-KO (n = 8), and Ephb2-Rescue (n = 8) SSCs. (t-test: WT vs. Ephb2-KO: ^***P < 0.001, WT vs. Ephb2-Rescue: ^***P < 0.001).

Figure 5.

Dppa4 re-expression restores proliferation but not colonization. (A) Left: representative immunoblot showing DPPA4 expression in wild-type and Ephb2-KO SSCs. PPIB = loading control. Right: quantification of DPPA4 expression in Ephb2-KO SSCs, normalized to WT SSCs. Bars represent mean ± SD of three experiments (paired t-test: ^**P < 0.01). (B) Immunoblot showing DPPA4 expression in WT and Ephb2-KO SSCs after transduction with a pCCL-PGK-mCherry plasmid (Ctrl) or a pCCL-PGK-Dppa4 plasmid (Dppa4-Rescue). PPIB = loading control. (C) Cell recovery assay of WT, Ephb2-KO, and Dppa4-Rescue SSCs cultured on feeder cells over 3 weeks. (Mean ± SD of three experiments; paired t-test: WT vs. Ephb2-KO: ^*P < 0.05; Ephb2-KO vs. Dppa4-Rescue: ^*P < 0.05.) (D) Quantification of EdU-positive SSCs by FACS analysis in WT, Ephb2-KO (KO), and Ephb2-Rescue SSCs. (Bars represent mean ± SD of three experiments; paired t-test: ^*P < 0.05 is indicated.) (E) Representative whole-mount image of testes after transplant of WT, Ephb2-KO (KO), or Dppa4-Rescue SSCs (white arrows: colonies formed by transplanted SSCs). (F) Scatter plot of one representative experiment of two showing the number of colonies formed by WT (n = 5), Ephb2-KO (n = 5), and Dppa4-Rescue (n = 5) SSCs. (unpaired t-test WT vs. Ephb2-KO: ^***P < 0.001, WT vs. Dppa4-Rescue: ^***P < 0.001).

Ephb2 knockout by CRISPR-Cas9 targeting in cultured SSCs

To investigate the role of Ephb2 in SSC self-renewal, we disrupted Ephb2 using CRISPR-Cas9. A plasmid co-encoding the Cas9 enzyme, a guide RNA targeting the exon 3 of Ephb2 and a GFP reporter, was transfected into SSCs in culture, just prior to FACS sorting for GFP-positive cells. Single-cell clones were then expanded for several passages prior to screening (Figure 1F). Eleven cellular clones electroporated with the plasmid containing the gRNA were expanded. Ten of those 11 clones exhibited loss of EPHB2 protein expression (Ephb2-KO SSCs) by FACS analysis. In parallel, five cellular clones electroporated with the same plasmid backbone lacking gRNA were expanded and screened as wild type (WT SSCs; Figure 1F). These results were confirmed by FACS and immunoblot (Figure 1G). Efficient targeting was then confirmed by sequencing the region of interest in Ephb2 exon 3. Bi-allelic mutations were observed in Ephb2-KO clones, whereas WT showed no mutation at the Ephb2 locus (Figure 1H).

Ephb2 loss-of-function affects SSC proliferation without modifying the apoptosis rate

To analyze growth of Ephb2-KO SSCs, mutant and WT cell lines were passaged, counted, and re-plated weekly at the same density. Ephb2-KO SSCs proliferated 35 ± 6% (mean ± SD) slower than WT after 1 week. After 6 weeks, significantly reduced cell mass was found for Ephb2-KO SSCs compared to WT (Figure 2A). To address a potential non-cell-autonomous role for feeder cells in the Ephb2 KO phenotype, we also compared the growth of WT and Ephb2-KO SSCs cultured on a protein extracellular matrix (Matrigel) without feeders for 3 weeks. In these challenged growth conditions, in which cells grew slower than in standard conditions on feeders, we again observed a significant difference with Ephb2-KO SSCs growing ~ 55 ± 8% slower than WT SSCs, suggesting that the Ephb2-KO phenotype does not depend on interactions with somatic feeder cells (Figure 2B). Colonies from Ephb2-KO SSCs on CD34⁺ feeders appeared morphologically similar to WT SSCs, with no increase in floating or fragmented cells (Figure 2C).

To confirm that the cell mass difference observed was not due to an increase in cell death, we assessed apoptosis by measuring the proportion of cleaved caspase-3-positive cells. No significant difference was observed between WT and Ephb2-KO SSCs (1.19 ± 0.7% [mean ± SD] and 1.2 ± 0.6%, respectively; Figure 2D). Hence, we measured proliferation rates in Ephb2-KO and WT SSCs by EdU incorporation assays, revealing that the Ephb2-KO cell population had 11 ± 3% fewer proliferating cells than the corresponding WT (Figure 2E). Analysis of cell cycle by propidium iodide incorporation also revealed a decrease in the proportion of S-phase cells in the Ephb2-KO compared to WT (Supplementary Figure S3). Taken together, these data suggest that ablation of Ephb2 in SSC cultures impairs cell expansion, not by altering survival, but rather by decreasing proliferation.

Stem cell activity of SSCs is reduced by Ephb2 loss

Only a small proportion of spermatogonia in long-term culture is capable of colonization in vivo, and cell mass in vitro is poorly reflective of changes in the functional stem cell fraction [21]. Stem cell activity measurement can be extrapolated from the quantification of colonies formed after transplantation of cells into the seminiferous tubules of a recipient mouse [49]. We investigated the effect of Ephb2 loss on stem cell activity by transplanting WT and Ephb2-KO SSCs into busulfan-treated recipients. Prior to transplantation, both genotypes were simultaneously transduced with a plasmid expressing mCherry fluorescent protein, driven by a human PGK promoter fragment (hPGK). Testes were recovered after ~ 2 months, and newly formed mCherry-positive donor-derived colonies were counted (Figure 2F). Although both WT and KO lines had stem cell activity, Ephb2-KO cells formed ~ 5.4-fold fewer colonies than WT SSCs (46.4 ± 20.5 [mean ± SD] vs. 252 ± 46.3 per 100 000 cells injected, respectively; Figure 2G and H). These results indicate that Ephb2-KO SSC cultures have greatly reduced stem cell activity. To further characterize the population of cells in culture, we analyzed the gene expression of selected canonical stem cell or undifferentiated Spg markers to assess whether loss of those markers could explain the reduced stem cell activity. However, no substantial differences were found in the expression of Bcl6b, Lhx1, Etv5, Id4, or Zbtb16/Plzf between WT and Ephb2-KO SSCs (Figure 2I).

Ephb2 re-expression restores proliferation but not colonization capacity

In order to confirm the specificity of the Ephb2 loss phenotype, we performed a rescue experiment. Ephb2 was re-expressed in KO SSCs by lentiviral transduction of a human EPHB2 cDNA (or an mCherry control cDNA) driven by hPGK. Of note, the human EPHB2 gene shares more than 99% homology with the mouse gene. Successful EPHB2 protein expression was confirmed by immunoblot (Figure 3A). While the proliferation of KO SSCs transduced with the mCherry control [5 ± 0.46-fold (mean ± SD) per week] remained significantly slower than WT mCherry control SSCs (6 ± 0.22-fold per week), the proliferation of KO SSCs transduced with hEPHB2 (EPHB2-Rescue) was significantly increased [6.5 ± 0.38-fold (mean ± SD) per week; Figure 3B]. The proliferation rate was also measured by EdU incorporation, demonstrating a significant increase in EPHB2-Rescue SSCs (60.4 ± 4.5%) compared to Ephb2-KO mCherry controls (43.9 ± 5%; Figure 3C). Next, we performed transplantation experiments comparing the colonizing ability of mCherry-labeled EPHB2-Rescue SSCs to WT and Ephb2-KO SSCs (Figure 3D). The number of colonies formed by rescued cells remained unchanged relative to the Ephb2-KO cells, suggesting that the mechanism underlying the reduction of stem cell activity when Ephb2 is ablated is different from that affecting in vitro proliferation.

Germ cell-associated transcription factors are regulated by Ephb2

Given the strong phenotype observed after Ephb2-KO SSCs transplantation, we hypothesized that substantial transcriptional modifications occur in SSCs lacking Ephb2, compared to WT. RNA sequencing on matched pairs (n = 3) of WT and Ephb2-KO cellular clones showed that Ephb2 loss induced unexpectedly subtle transcriptional changes in SSCs (Supplementary Figure S4). Of 27 426 analyzed transcripts, 68 were significantly downregulated in Ephb2-KO SSCs and 25 were upregulated (Supplementary Table S3). Given the dearth of significantly regulated genes, no single pathway was clearly identified. Among the 10 most differentially expressed transcripts, we focused our attention on two genes potentially involved in germ cell maintenance (Figure 4A). Dppa4 is known as a pluripotency marker in embryonic stem cells [50, 51]. However, at the adult stage, Dppa4 expression is restricted to germ cells [51, 52]. Bnc1 is a transcription factor expressed mainly in germ cells and has a role in the maintenance of germ cell proliferation rate [53]. We then confirmed that both Dppa4 and Bnc1 were downregulated in the absence of Ephb2, at the mRNA level by RT-qPCR. Re-expression of Ephb2 in KO SSCs restored the expression of both genes, confirming a regulatory link between Ephb2 and those two potential target genes (Figure 4B).

Figure 4.

Transcriptional targets of the Ephb2 pathway in SSCs. (A) Heatmap representing the 10 genes for which expression was most significantly altered as revealed by RNA sequencing (P < 0.05 by DESeq2) of three WT cellular clones and three Ephb2-KO cellular clones. The color key scale bar shows Z-score values for the heatmap (significance: P < 0.05). (B) RT-qPCR showing Dppa4 and Bnc1 in wild-type, Ephb2-KO, and Ephb2-Rescue SSCs. Bars represent mean ± SD of four experiments (paired t-test: ^*P < 0.05; ^**P < 0.01).

Dppa4 expression rescues proliferation of Ephb2-KO SSCs but not stem cell activity

Dppa4 is a marker of undifferentiated Spg in the marmoset monkey [54], but a specific functional role for Dppa4 has not been described in the adult rodent or primate testis. Consistent with the mRNA results, the level of DPPA4 protein was also significantly reduced in Ephb2-KO SSCs [57 ± 13% (mean ± SD); Figure 5A]. To determine whether Dppa4 is responsible for Ephb2-mediated proliferation in vitro and regulation of stem cell activity, we overexpressed Dppa4 in Ephb2-KO SSCs (Figure 5B). Ephb2-KO SSCs were transduced with lentivirus driving murine Dppa4 expression and evaluated by immunoblot (Figure 5B). We then measured cell recovery and EdU incorporation, revealing that Dppa4 over-expression significantly increased the proliferation rate of Ephb2-KO SSCs (Figure 5C and D).

To determine whether Dppa4 rescues the defect in colonization of Ephb2-KO SSCs, we transplanted mCherry-labeled Ephb2-KO SSCs overexpressing Dppa4 into recipient animals. However, the stem cell activity remained significantly lower than WT SSCs, with no difference compared to Ephb2-KO SSCs (Figure 5E and F). These results confirm the discrepancy between the in vitro proliferation and colonization phenotypes. Thus, Ephb2 seems to play a role in stem cell function that cannot be rescued by the re-expression of transgenes.

Ephb2 loss-of-function reduces the activity of the ERK MAPK and SRC signaling pathways

The stem cell activity of SSCs relies on concerted activation of intracellular signaling pathways by specific growth factors (GFs), including GDNF or FGF. Through their cognate tyrosine kinase receptors, GFs induce the activation of ERK MAPK and PI3K/AKT pathways and induce the expression of downstream mediators, including transcription factors related to self-renewal and proliferation [17, 18, 55, 56]. In the context of cancer, EPHB2 downstream signaling was linked to the RAS/RAF/MAPK pathway [57, 58]. Therefore, we interrogated the expression of critical phospho-proteins in those pathways in Ephb2-KO SSCs. We focused on the phosphorylation of SRC kinase, which functionally binds to EPHB2 in different models [59, 60] and on the phosphorylation of ERK1/2, a downstream effector of the MAPK pathway. In each condition, we measured the percentage of cells with p-ERK1/2 (Thr202/Tyr204) or p-SRC (Tyr419) at a higher level than the median in WT SSCs (p-ERK^High and p-SRC^High, respectively; Figure 6). WT SSCs were 60 ± 11% (mean ± SD) p-ERK^High, while Ephb2-KO SSCs were 30 ± 16% p-ERK^High. When Ephb2 was re-expressed (Ephb2-Rescue SSCs), the p-ERK^High population increased to 49 ± 14%. Similarly, WT SSCs were 54 ± 1% p-SRC^High and Ephb2-KO SSCs were 20 ± 16% p-SRC^High, while Ephb2-Rescue SSCs were 32 ± 26% p-SRC^High. These data suggest that Ephb2 is necessary to maintain optimal activity of canonical pathways that regulate SSC self-renewal.

Figure 6.

Effect of Ephb2 ablation on ERK and SRC signaling in SSCs. (A) Measurement of signaling pathway activity by FACS. WT, Ephb2-KO, and Ephb2-Rescue SSCs were stained with rabbit anti-p-ERK (top row) or anti p-SRC antibodies (bottom row) and conjugated secondary anti-rabbit antibody. Left: representative FACS histograms of p-ERK and p-SRC intensity WT (black), Ephb2-KO SSCs (red), and Ephb2-Rescue (dash). Right: percentage of cells with p-ERK or p-SRC expression higher than WT median expression. [Mean ± SD of six experiments (WT and Ephb2-KO) and three experiments (Ephb2-Rescue); paired t-test: ^*P < 0.05, ^**P < 0.01.]

Discussion

Whereas a handful of cell surface receptors have been studied in depth in germline stem cells, many others remain unexplored. The tyrosine kinase receptor Ephb2 was detected in testicular germ cells, but its function was not examined [25, 41]. Here, we showed that Ephb2 is expressed in undifferentiated spermatogonia in vivo and in cultured SSCs. Introduction of loss-of-function mutations in the Ephb2 gene in cultured SSCs disrupted cell proliferation and stem cell activity. The absence of Ephb2 led to a decrease in p-ERK and p-SRC, thought to be critical for SSC self-renewal and colonization. Ephb2 loss also induced a decrease in the expression of germ cell-associated transcription factors potentially involved in SSC maintenance.

SSC proliferation and stem cell activity rely on the concerted effects of MAPK and PI3K/AKT signaling, among others [15, 61]. Cancer studies support paradoxical roles for Ephb2 as both an activator and a repressor of these pathways [62–64]. Our results show that the absence of Ephb2 significantly reduces the phosphorylation of ERK1/2, a critical downstream effector of the MAPK pathway, known to induce expression of numerous targets. These findings stand in contrast to those in neuronal cells, in which EPHB2 negatively regulates p-ERK1/2 [65]. Further experiments will be needed to identify the unique intermediate partners that connect EPHB2 to phosphorylation of ERK1/2 in SSCs.

Although Ephb2 loss did not affect expression of canonical factors associated with SSC self-renewal, such as Blc6b, Lhx1, and Etv5, our transcriptomic data revealed potential novel downstream targets that could affect adult SSC maintenance. Dppa4 is a transcription factor that plays a key role in the development of numerous organs [51, 66]. However, its expression in adults is restricted to germ cells [50, 52]. To date, there are no reports of a specific role for Dppa4 in function of adult germ cells, although it is considered as a marker of undifferentiated Spg in some species, such as the marmoset monkey [54]. Given its essential role in development, few data are available from Dppa4 knockout models, and the consequences of Dppa4 loss on germline stem cell maintenance are unknown. Yet, similar to our findings in cultured Spg, Dppa4 is required for embryonic stem cell proliferation, suggesting a conserved function in these two cell types [50]. The transcription factor Bnc1 was also significantly repressed in EPHB2-KO SSCs, according to our transcriptomic data. Primarily expressed in germ cells, Bnc1 was previously described as essential to proper germ cell maintenance and spermatogenesis [53, 67]. Bnc1 KO mice, although fertile, show significantly altered germ cell proliferation and reduced testis size [53]. Interestingly, a microarray expression study on ovarian germ cells lacking Bnc1 demonstrated decreased Dppa4 expression [68], raising the possibility that these two transcription factors share a common genetic pathway downstream of Ephb2. In conclusion, the reduction in mRNA for Dppa4 and Bnc1 in Ephb2-KO SSCs followed by the restoration of expression of both genes when Ephb2 was re-expressed supports the concept that Dppa4 and Bnc1 are candidate Ephb2 targets.

Previous studies showed that transgenic mice with a developmental knockout of Ephb2, generated by homologous recombination of a protein null allele in embryonic stem cells, are fertile and lack gross reproductive defects [69–71]. We performed preliminary morphological analysis of testicular tissue from Ephb2 knockout vs. wild-type control littermates, to detect abnormalities in the seminiferous tubules, revealing no gross phenotype or difference in number of ZBTB16-expressing undifferentiated spermatogonia (data not shown). In one sense, these findings may not be surprising, since mouse genetic studies of Eph receptors using developmental knockout alleles have revealed substantial redundancy among Eph family members in other organ systems [72–74]. Furthermore, compensation at the level of mRNA is increasingly recognized as a global confounding factor in developmental knockout animals [75]. Nonetheless, the microenvironment in vitro is artificial and substantially different from the stem cell niche in vivo, and Ephb2 could be particularly relevant for the former compared to the latter. Thus, genetic compensation could differentially alter the in vivo phenotype of SSCs compared to that observed in culture. These issues could be resolved in the future using conditional ablation of Ephb2 in adult SSCs in vivo.

Ephb2 is essential to the maintenance of stem cells in the nervous system, the intestine, and bone [37, 39, 69, 76]. However, the detailed mechanisms regarding its role in those contexts are only beginning to emerge. In the colon, the absence of Ephb2 led to decreased cyclin D2 expression, with a concomitant decrease in stem cell proliferation [39]. Our analysis of cyclin expression in SSCs showed no differential expression of cyclins D1, D2, or D3 in Ephb2-KO cells (not shown). Thus, the explanation for the difference in cell cycle between Ephb2-KO cells and WT remains cryptic.

A complete picture of the function of Eph receptors would require elucidation of the biology underlying their cognate Ephrin ligands. Experimentally, this is a challenging problem due to extensive ligand-receptor promiscuity. The three EphrinB ligands are expressed in vivo in the testis and in SSCs in vitro (Supplementary Figure S5). The presence of membrane bound ligands implies that most Eph/Ephrin signaling is due to interactions between the adjacent cells. Among the best described roles for Ephb2 is its mediation of axon guidance during development, which depends on cell–cell interactions [69, 77]. Our results on SSC proliferation in the presence or absence of feeder cells support the possibility of homotypic Spg–Spg interactions, but Spg–somatic cell interactions could also be important in vivo. In a study on osteogenesis, Efnb1 is described as a necessary ligand for EPHB2 to maintain hematopoietic stem cells and proper bone formation [76]. While preliminary studies using non-pre-clustered, recombinant EFNB1 in our SSC system showed no significant effects on signaling (data not shown), all EPHB2 ligands should be considered in the future.

Re-expression of Ephb2 in Ephb2-KO cells was sufficient to rescue the SSC proliferation defect but not the loss of stem cell activity. Transplantation experiments rely on the ability of SSCs to colonize seminiferous tubules previously depleted of endogenous stem cells. For successful engraftment to occur, injected SSCs must migrate from the tubular lumen to the basement membrane, through a process of homing that is apparently unique to the transplantation assay [78, 79]. It is possible that the loss of stem cell activity in Ephb2-KO cells could be related to homing. As a matter of speculation, the challenging conditions of the transplantation assay could be too stringent for the engineered EPHB2-rescued cells to exhibit a normal phenotype. In addition, the activity of the Ephb2 transgene might be sufficient for in vitro signaling but not for yet undefined in vivo cell–cell interactions necessary for recolonization of the seminiferous tubules. Whereas certain genes (e.g., Zbtb16/Plzf) are required for SSCs in vivo but not in vitro [80], our data suggest that the opposite is true for Ephb2, emphasizing the limitations of the in vitro culture model. A final issue is that the rescue experiments herein employed constitutive promoters, bypassing the normal gene regulatory apparatus Ephb2, which could be critical for dynamic expression and stem cell activity but not simple cell proliferation. Moreover, the use of a constitutive promoter in the rescue experiment might have led to overexpression of Ephb2, which could be detrimental to SSC differentiation and colony expansion. However, we were not able to verify this hypothesis histologically after transplantation.

Our data indicate that the Ephb2 receptor is a key kinase receptor for the maintenance of SSC proliferation. In Ephb2 KO SSCs, we identified defects in MAPK and SRC signaling and in the expression of transcription factors such as Dppa4 and Bnc1. Ephb2 also plays a crucial role in colonization of recipient testes in a manner that maybe distinct from its regulation of cell proliferation in vitro. Additional studies will be required to understand these different functions. We propose that experimental manipulation of Ephb2, Dppa4, and possibly Bnc1 could provide key insights into in vitro expansion of Spg from other species, such as humans in which long-term propagation of SSCs has been particularly challenging.

Conflict of interest

The authors have declared that no conflict of interest exists.