ARF6, a component of intercellular bridges, is essential for spermatogenesis in mice

Abstract

Male germ cells are connected by intercellular bridges (ICBs) in a syncytium due to incomplete cytokinesis. Syncytium is thought to be important for synchronized germ cell development by interchange of cytoplasmic factors via ICBs. Mammalian ADP-ribosylation factor 6 (ARF6) is a small GTPase that is involved in many cellular mechanisms including but not limited to regulating cellular structure, motility, vesicle trafficking and cytokinesis. ARF6 localizes to ICBs in spermatogonia and spermatocytes in mice. Here we report that mice with global depletion of ARF6 in adulthood using Ubc-CreERT2 display no observable phenotypes but are male sterile. ARF6-deficient males display a progressive loss of germ cells, including LIN28A-expressing spermatogonia, and ultimately develop Sertoli-cell-only syndrome. Specifically, intercellular bridges are lost in ARF6-deficient testis. Furthermore, germ cell-specific inactivation using the Ddx4-CreERT2 results in the same testicular morphological phenotype, showing the germ cell-intrinsic requirement of ARF6. Therefore, ARF6 is essential for spermatogenesis in mice and this function is conserved from Drosophila to mammals.

Graphical Abstract

Introduction

Spermatogenesis is a dynamic and complex process that produces sperm in males through adulthood. The continuity of this process depends on the balance among spermatogonial stem cells (SSCs) renewal, proliferation, and differentiation. Within seminiferous tubules, SSCs residing at the basement membrane divide mitotically to generate more stem cells to maintain stem cell populations and spermatogonia that differentiate. The expansion of diploid spermatogonia via a series of mitotic divisions increases the number of spermatogonia entering meiosis. Each primary spermatocyte undergoes two rounds of meiotic divisions to produce four haploid spermatids. Finally, spermatids undergo spermiogenesis to develop into spermatozoa that are released into the lumen of seminiferous tubules (Fayomi & Orwig, 2018; Jan et al., 2012; Oatley & Brinster, 2012).

Cytokinesis is the final stage of the cell division process by which two daughter cells result from a single eukaryotic cell (Gerhold et al., 2022; Gould, 2016; Prekeris & Gould, 2008). It involves mitotic spindle reorganization, formation of the central spindle, cleavage furrow formation, actin-myosin ring assembly, intercellular bridge formation, contraction and abscission. This process is highly coordinated and the molecular mechanisms determining this dynamic process have been investigated in detail (Fededa & Gerlich, 2012; Mierzwa & Gerlich, 2014).

In contrast to somatic cell cytokinesis, abscission is incomplete and intercellular bridge remains in germ cells, resulting in the formation of germ cell syncytium (Greenbaum et al., 2007; Weber & Russell, 1987). The syncytium resulting from intercellular bridges (ICBs) is a unique feature of mammalian spermatogenesis. Stable ICBs are thought to be necessary for normal spermatogenesis, providing structural channels that connect germ cells for the exchange of cytoplasmic materials and signaling molecules to ensure synchronization of spermatogenesis (Greenbaum et al., 2011; Mäkelä & Toppari, 2018; Rezende-Melo et al., 2020; Weber & Russell, 1987). The testis-expressed gene 14 (TEX14) was identified as a germ cell-specific gene (Wang et al., 2001). The TEX14 protein localizes to ICBs and is necessary for maintaining ICB stability (Greenbaum et al., 2006). TEX14 prevents abscission by competitively binding to CEP55 to prevent the recruitment of ALIX and TSC101, which stabilizes ICBs (Iwamori et al., 2010; Kim et al., 2015). Tex14-null mice lack ICBs and exhibit spermatogenic arrest before the first meiotic division, resulting in male infertility (Greenbaum et al., 2006). The absence of TEX14 was also reported in infertile patients diagnosed with Sertoli cell-only syndrome (Borjian Boroujeni et al., 2018). In addition, RACGAP1 (also known as MgcRacGAP) localizes to ICBs in germ cells. The Racgap1 conditional deletion mouse mutants lacked ICBs and showed germ cell proliferation arrest and male infertility (Lorès et al., 2014). These studies show that formation of intercellular bridges is essential for male germ cell development.

The small ADP-ribosylation factor (ARF) GTPase family members participate in cytokinesis (Frémont & Echard, 2018; Jackson & Bouvet, 2014; Militello & Colombo, 2013). ARF1 plays an important role in Golgi disassembly, which is a prerequisite step prior to cytokinesis (Jackson, 2018). ARF6 localizes to the plasma membrane at the site of cleavage furrow ingression and is essential for cytokinesis in somatic cells (Schweitzer & D’Souza-Schorey, 2002, 2005). ARF6 promotes cleavage furrow formation by membrane addition via vesicular transport, regulates the localization of septin ring at the site of cytokinesis (Frémont & Echard, 2018; Jackson, 2018; Sztul et al., 2019) and protects the post-mitotic midbody from 14–3–3 mediated disintegration (Joseph et al., 2012). Activated ARF6 interacts with mitosis kinesin-like protein (MKLP1 or KIF23) (Dyer et al., 2007) and is present at the Flemming body at the late phase of cytokinesis to link the plasma membrane to the microtubules at the cleavage plane for cell separation (Makyio et al., 2012). A failure in cytokinesis results in increased number of binucleated and multinucleated cells in cultured Arf6 knockout mouse embryonic fibroblasts (Makyio et al., 2012). In Drosophila, Arf6 is necessary for proper cytokinesis in fly spermatocytes; its deletion resulted in failure of cleavage furrow formation and ring canal or ICB establishment, resulting in a four-wheel-drive or multinucleated phenotype and male fly infertility (Dyer et al., 2007). Interestingly, the mouse ARF6 protein localizes to intercellular bridges in spermatogonia and spermatocyte (Katsumata et al., 2017), suggesting that ARF6 may also play a role in spermatogenesis in mouse.

We previously generated a floxed Arf6 or Arf6^tm1.1Lbh mouse model to study its role in kidney glomerulus biology (Lin et al., 2017). To circumvent its embryonic lethality (Suzuki et al., 2006), we inactivated Arf6 using an inducible Cre (Ubc-CreERT2). We find that ARF6 is dispensable for viability in mice when it is deleted in adulthood but that conditional deletion of Arf6 in adult mice results in male infertility.

Results

Conditional deletion of mouse Arf6 results in male infertility

Arf6 floxed (Arf6^f) mice were crossed with transgenic mice expressing tamoxifen regulated recombinase under the control of Ubiquitin C promoter (Ubc-CreERT2^Tg/+) to obtain mice in which the first two exons could be deleted in all tissues by recombination following treatment with tamoxifen. Three-month-old male mice homo- or heterozygous for Arf6 recombined alleles were identified one month after tamoxifen treatment using the PCR-based screening strategy outlined in Figure 1a and 1b. We initially observed that a male tamoxifen-treated Arf6^f/f, Ubc-CreERT2^Tg/+ (referred to as Arf6 uKO) mouse did not sire offspring during 12 months of breeding despite numerous changes of female breeders, while a tamoxifen-treated Arf6^f/+, Ubc-CreERT2^Tg/+(Arf6^+/−) littermate sired multiple litters during the same period. In a formal experiment to confirm this observation, tamoxifen-treated male Arf6 uKO mice (n=5 independent Arf6 uKO and n=4 control male mice) were mated with two young virgin females during a three-month period (Table 1, First Mating). Mating partners were switched to allow experienced females to mate with Arf6 uKO male mice for another two months (Table 1, Second Mating). Male Arf6 uKO mice did not sire any offspring with either virgin or experienced females. Infertility in Arf6 uKO mice was specific to males since Arf6 uKO females had normal fecundity during a 3-month breeding period (Table 2). At six months post tamoxifen, Arf6 uKO male mice displayed smaller testes while other reproductive structures were comparable to littermate controls (Fig. 1c). We performed histological analysis of testes. While tamoxifen-treated control adult testis contained a full spectrum of germ cells (i.e., spermatogonia, spermatocytes, and post-meiotic spermatids), Arf6 uKO testis lacked germ cells. (Fig. 1d). The observation that Arf6 deleted adult mice were viable was surprising given the broad tissue expression of ARF6 protein (Akiyama et al., 2010; Katsumata et al., 2017; Tsuchiyas et al., 1991). In conclusion, ARF6 deleted in adult mice is dispensable for viability yet is essential for male fertility.

Fig. 1. Conditional Arf6 null male mice display germ cell aplasia.

(a) ARF6 mouse line generation. Exon 1 and exon 2 are floxed by two loxP sequences (open triangles) and a FRT sequence (filled triangle). Coding sequence of Arf6 is indicated by a solid bar within exon 2. Primer locations and orientation used for genotyping are indicated with arrows. Floxed Arf6 (f-Arf6) mouse was crossed with transgenic mice carrying an Ubc-CreERT2 allele and their progenies were intercrossed to generate various Arf6 genotyped mice. The CRE recombinase expression was induced by tamoxifen to generate whole body ARF6 knockout or uKO mouse.

(b) Genotyping of various Arf6 mice by PCR strategy using primer set consisting of either F1, R1 and R2 primers (top panel) or F2, R2, β-globin Fwd and β-globin Rev primers (middle panel) on tail DNA. Myh9 primers were included as internal control in Ubc-CreERT2 allele (Tg) PCR reaction. The relative mobility and the resulting amplicons are indicated.

(c) Gross morphology of Arf6 null (uKO) and littermate control (Ctrl) paraformaldehyde fixed male gonads at 6 months post tamoxifen treatment. SV, seminal vesicle; B, bladder; testis and epididymis are indicated.

(d) Low magnification histology of paraffin-embedded sections of littermate control (Ctrl) and Arf6 uKO testes after hematoxylin and eosin staining. BOX, enlargement of region marked by black dashed box. Sg, spermatogonia; SpC, spermatocytes; SpT, spermatids and Ser, Sertoli cells are indicated. Embedded scale bar:100 μm.

Table 1.

Fertility of Arf6 male mice evaluated during a five-month period. Two to three-month old male mice of Arf6^f/f, Ubc-CreERT2^Tg/+ (uKO, n=5) and littermate control (n=4) were given three doses of tamoxifen over a five-day period. One month post tamoxifen, each male mouse was housed with two virgin females during a three-month period (First Mating). After a week of separation, breeding partners were switched to mate for another two months (Second Mating).

		First Mating		Second Mating
Male genotype	Fertile/total (n)	Litters (n)	Total pups (n)	Litters (n)	Total pups (n)
Arf6 uKO	0/5	0	0	0	0
Arf6 control	3/4	10	51	6	14

Table 2:

Fertility of Arf6 female mice. Three-month-old female mice of Arf6^f/f, Ubc-CreERT2^Tg/+ (uKO, N=4) and littermates (Control, n=4) were given three doses of tamoxifen over a five-day period. One month post tamoxifen, a stud male mouse was housed with one Arf6 KO and one control females over a three-month period.

Female genotype	Fertile/total (n)	Litters (n)	Total pups (n)
Arf6 uKO	4/4	6	28
Arf6 control	3/4	3	14

ARF6 is essential for spermatogenesis

To investigate the loss of germ cells in Arf6 uKO males, we performed a time-course analysis of the early effect of Arf6 deletion induced by tamoxifen in adult mice between two to three months of age (Fig. 2a) (Shi et al., 2019). Small testes were readily observed by 20 days post tamoxifen (dpt) treatment (Fig. 2b), while there was no significant effect on body weight (data not shown). The reduction of testis weight was statistically significant as early as 12 dpt and mean testes weight decreased progressively by day 20 (Fig. 2c). At 8 dpt, Western blotting did not detect a reduction of ARF6 protein abundance in Arf6 uKO testis compared with control. By 12 dpt, ARF6 protein was reduced and was further reduced at day 16 and day 20 (Fig. 2d). Reduced abundance of spermatocyte-specific proteins SCP3 was detected at 12 dpt and thereafter in Arf6 uKO testicular lysate. The relative abundance of germ cell-specific proteins DDX4, and TEX14, showed a similar expression pattern as ARF6 protein (Fig. 2d), suggesting that germ cells might be progressively depleted with time after tamoxifen treatment.

Fig. 2. Conditional deletion of Arf6 in adult male mice resulted in testicular atrophy.

(a) Experimental design. Homozygous floxed Arf6^f/f, Ubc-CreERT2^Tg/+ or Arf6^f/−, Ubc-CreERT2^Tg/+ (uKO) male mice between two and three months of age and their littermate controls (Ctrl) were given 3 doses of tamoxifen (TAM) over a 5-day period. Mice were sacrificed at indicated day post tamoxifen.

(b) Representative gross testicular morphology of Arf6 uKO mice and littermate control (Ctrl) 20 days after final tamoxifen treatment.

(c) Box and whisker plot of testis weight at day 2 (control testes, n=12, uKO testes, n=12), day 4 (n= 8 and n=10, respectively), day 6 (n= 10 and n=8, respectively), day 8 (n= 6 and n=12, respectively), day 12 (n= 10 and n=12, respectively), day 16 (n= 14 and n=15, respectively), and day 20 (n= 11 and n=19, respectively) post tamoxifen treatment. *, P<0.001; **, P<0.0001; and ***, P<0.00001 (t-test).

(d) Tamoxifen treated-testicular protein lysate were immunoblotted with indicated antisera. Each lane represents individual mouse testis lysate collected at indicated times post tamoxifen treatment. Indicated protein abundance relative to control at each time point is indicated below blots. Protein mobility standards are indicated.

We examined the morphological effect of ubiquitous Arf6 deletion on spermatogenesis. A binary decision key roadmap was used to assess spermatogenic stages of hematoxylin-eosin stained seminiferous tubules (Meistrich & Hess, 2013). For simplicity, we grouped our histological assessments within spermatogenic stages I-VI, VII-VIII, IX-X, XI and XII (Fig. 3) (Russell et al., 1990). Figure 4c summarizes the time-dependent progressive morphological changes observed by cell type following tamoxifen exposure in Arf6 uKO male mice. By 2 dpt, we observed a few intense eosin cytoplasmic stained degrading germ cells following tamoxifen treatment (Fig 3, d2 uKO). Binucleated cells and a reduction of pachytene spermatocytes (in stages I-VI, VII-VIII) and zygotene spermatocytes (in stages XI and XII) were observed in 4 dpt Arf6 uKO seminiferous tubules (Fig. 3, d4 uKO). By 6 dpt, we observed a complete absence of zygotene spermatocytes in stage XI-XII, secondary spermatocytes in stage XII and binucleated cells in stage I-VI, IX-X, and XII in Arf6 uKO testis (Fig. 3, black arrows). Consistent with the changes observed by 6 dpt, at 8 dpt, there was additional loss of leptotene spermatocytes (stages IX-X), absence of zygotene spermatocytes (stage XI-XII), and appearance of vacuoles; Type B/Intermediate spermatogonia were rarely observed. By 12 dpt, the loss of preleptotene through diplotene spermatocytes and the presence of densely stained degenerating germ cells and vacuoles were apparent. Multinucleated giant cells in stages I-VI, IX-X and XI were readily observed in Arf6 uKO mouse testes (Fig. 3, d12 uKO black arrows). Degeneration of cells was evident in stage XII, marked by cytoplasm that stained intensely with eosin (Fig. 3, d12 uKO red arrows). By day 16, vacuoles, round spermatids, elongating spermatids and Sertoli cells were prominent among seminiferous tubules. Only stage VII-VIII and IX-X tubules were identifiable due to their characteristic luminal lining with elongated spermatids and the presence of oval spermatids, respectively (Fig. 3). By day 20, only stage VII-VIII tubules were identifiable among other tubules consisting mainly of elongating spermatids and Sertoli cells (Fig. 3).

Fig. 3. Arf6 is essential for spermatogenesis in male mice.

(a-b) Morphological analysis of testes of tamoxifen treated adult male mice. (a) Hematoxylin-eosin-stained cross sections of Stage I-VI, VII-VIII, IX-X, XI and XII seminiferous tubules from Ctrl testis at 12 dpt and Arf6 uKO testes from 2 to 12 dpt are shown. Red arrows indicate apoptotic cells of intense cytoplasmic eosin staining. Multinucleated giant cells are marked by black arrows. Sg, spermatogonia; In, intermediate spermatogonia; B, type B spermatogonia; Pl, preleptotene; L, leptotene; Z, zygotene; P, pachytene; D, diplotene spermatocytes; Mi, metaphase spermatocyte; S2, secondary spermatocyte; RS, round spermatids; ES, elongated spermatids; and Ser, Sertoli cell are indicated. (b) Representative images of Arf6 uKO testes at 16 and 20 dpt and Ctrl testis at day 20 post tamoxifen are shown. Recognizable stages are indicated, and vacuoles are marked by asterisks.

Fig. 4. Progressive loss of GCNA1 positive spermatogonia after Arf6 deletion.

(a) Paraffin-embedded sections (7 μm) of Ctrl and Arf6 uKO mouse testis collected from 4 to 20 dpt were examined by indirect immunofluorescence microscopy using antisera against ARF6 (green) and GCNA1 (magenta) and by staining with peanut agglutinin (PNA, white). Enlargement of regions is marked by white dashed boxes. White arrows indicate ARF6 signals localized between prominent GCNA1-positive cells in the basal compartment (marked by two dashed lines); yellow arrows indicate ARF6 signals localized between spermatocytes of GCNA1-lo signals in the adluminal compartment; arrowheads indicate ARF6-immunoreactive ring-like structures. Tubules without PNA signals are marked by asterisks. Scale bar: 25 μm.

(b) Density of GCNA1-hi cells was assessed in seminiferous tubules, categorized into three categories (>50%, <50%, or none). Between 67 to 417 tubules were assessed at each time point. Solid-filled bars represent observations from tamoxifen-treated control testes, and dotted pattern bars are observations taken from Arf6 uKO testes. *, P<0.001 (Fisher’s exact test).

(c) Summary of time-dependent progressive cell loss by type following tamoxifen exposure in Arf6 uKO male mice based on morphological analysis and GCNA1 immunofluorescence analysis. Solid lines indicate the presence of cells and cell attenuation is marked by dotted lines. The appearance of either apoptotic cells, vacuole, or multinucleated giant cells is noted.

We examined control and Arf6 uKO testes by immunofluorescence microscopy to validate the observation above that ARF6 deletion results in loss of spermatogonia and failure of spermatogenesis. GCNA1, is a mouse germ cell nuclear antigen that is expressed in all germ cells except spermatids beyond step 10 (Enders & May, 1994). Consistent with published observations, GCNA1 was abundantly expressed in wild type spermatogonia, leptotene and zygotene spermatocytes within the basal compartment (referred to as GCNA1-hi here) as compared to later stages of developing germ cells in the adluminal compartment (referred to as GCNA1-lo here). Staining with peanut agglutinin (PNA) was included to orient our observations by marking acrosomes of spermatids (Fig. 4a) (Nakata et al., 2017). In control mouse testes, most cells were marked with strong immunoreactivity to GCNA1 (or GCNA1-hi) within the basal compartment after tamoxifen treatment. ARF6 signals were prominent between GCNA1-positive cells in ring-like structures (Fig. 4a, arrows and arrowheads), consistent with localization to germ cell intercellular bridges (Greenbaum et al., 2006). In 6 dpt Arf6 uKO tubules, we started to observe a relative decrease of GCNA1-hi cells within the basal compartment although ARF6 signal was still prominent. By 8 dpt, we mainly observed ARF6 signals localized between the GCNA1-lo cells. By 12 dpt, ARF6 immunoreactivity was no longer detected within the seminiferous tubules and GCNA1 signals were mostly co-localized with PNA signals that mark spermatids. This observation correlates to the histological analysis by hematoxylin-eosin staining where only round and elongated spermatids persisted after tamoxifen induced ARF6 deletion (Fig 3). By 20 dpt, tubules without any signals to GCNA1, PNA or ARF6 were observed among tubules that displayed PNA only signals in Arf6 uKO tubules. We conducted a semi-quantitative analysis of GCNA1-hi cell occupancy within the basal compartment (Fig. 4b). We observed a rapid disappearance of GCNA1-hi cells or spermatogonia and early spermatocytes with ARF6 deletion. We also examined LIN28A expression by immunofluorescence microscopy in this context. LIN28A is a pluripotency factor whose expression is specific to undifferentiated progenitor spermatogonia and early differentiating spermatogonia in mice (Chakraborty et al., 2014; Zheng et al., 2009). In control mice, we observed ARF6 localization between LIN28A-positive spermatogonia, again confirming its localization at ICB of spermatogonia (Fig. 5). Consistent with our observations of GCNA1 expression, deletion of Arf6 resulted in depletion of LIN28A-expressing spermatogonia as early as 4 dpt, while ARF6 protein expression persisted in spermatocytes to 8 dpt (Fig. 5, arrowheads). By 12 dpt, both ARF6 and LIN28A expression were absent within the Arf6 uKO tubules. Together, our morphological analysis and immunofluorescence study demonstrates that deletion of Arf6 in adult mice resulted in a failure of spermatogenesis.

Fig. 5. Arf6 inactivation leads to a rapid loss of LIN28A expression.

Paraffin-embedded sections (7 μm) of Ctrl and Arf6 uKO mouse testis collected at 4, 6, 8, 12, 16 and 20 dpt were examined by indirect immunofluorescence microscopy using antisera against ARF6 (green) and LIN28A (magenta). BOX, enlargement of region marked by white dashed box. White arrows indicate the detection of ARF6 signals localized between LIN28A positive cells; arrowheads indicate ARF6-immunoreactive ring-like structures; and white asterisks mark non-specific staining at the periphery of seminiferous tubules and interstitial space. Scale bar: 50 μm.

Deletion of Arf6 results increased programmed cell death

We examined whether Arf6 deletion resulted in programmed cell death of spermatogonia by TUNEL assay. Relative to tamoxifen-treated control tubules, spermatogonia in Arf6 knockout tissue stained positive by TUNEL assay as early as 2 dpt and 4 dpt and continued through 12 dpt in Arf6 uKO tubules compared to tamoxifen-treated control tubules (Fig. 6). The prevalence of identifiable TUNEL stained cells declined at later time points to 20 dpt, at which point spermatogonia were largely absent.

Fig. 6. Progressive apoptosis determined by TUNEL Assay.

Paraffin-embedded sections (7 μm) of control and Arf6 uKO mouse testis collected at 2, 4, 6, 8, and 12 dpt were analyzed by TUNEL assay. Sections were processed according to the manufacturer’s directions for TUNEL analysis and counterstained with PNA (green channel). Representative images are shown. Magenta asterisks indicate tubules containing TUNEL-positive cells. BOX, enlargement of regions marked by white dashed box. Scale bar: 100 μm.

Progressive loss of ICB in Arf6 uKO tubules

ARF6 was previously shown to localize to TEX14-positive ICBs of PLZF-positive spermatogonia primarily occupying the basal compartment of seminiferous tubule, and to ICBs in SYCP3-positive spermatocytes (Katsumata et al., 2017). We confirmed these observations (Fig. 7, Ctrl). Tamoxifen had no acute effect on this co-localization in control testes (Fig. 7). Prominent TEX14 signals, often appearing as ring-like structures, were also readily observed in germ cell layers (Fig. 7, magenta arrowheads). At 8 dpt in Arf6 uKO tubules, co-localization of ARF6 and TEX14 was observed (Fig. 7, d8 uKO). By 12 dpt, ARF6 was no longer detected within the seminiferous tubule, while TEX14-only signals were predominantly observed, mostly in ring-like structures, showing that TEX14 localization is independent of ARF6. By 16 dpt, distinct TEX14 signals in ring-like structures were rarely detected, consistent with immunoblotting data (Fig. 2d) at these time points.

Fig. 7. Immunofluorescent analysis of TEX14 in Arf6 uKO mouse testis.

Paraffin-embedded sections (7 μm) of Ctrl and Arf6 uKO mouse testis collected at 8, 12, and 16 dpt were examined by indirect immunofluorescence microscopy using antisera against ARF6 (green) and TEX14 (magenta). Enlargement of regions is marked by a white dashed box. White arrows indicate the detection of both ARF6 and TEX14 signals; several TEX14-only immunoreactive structures are indicated by magenta arrowheads. Scale bar: 50 μm.

Given the known role of TEX14 in maintaining intercellular bridge structure in the syncytia of germ cells and the observation that TEX14 expression were rarely detected in 16 dpt Arf6 uKO tubules (Fig. 7), we used transmission electron microscopy to examine ICB structures. ICBs were readily identifiable among type B spermatogonia, primary spermatocytes and round spermatids in tamoxifen-treated control testis (Fig. 8b−d). In Arf6 uKO testes at 16 dpt, we observed prominent round and elongated spermatids among Sertoli cells along the basal lamina (Fig. 8e and 8f). At the same time point, degenerating spermatocytes were identified (Fig. 8g), some round spermatids displayed plasma membrane blebbing (Fig. 8h−i), and ICBs were not found.

Fig. 8. Transmission electron microscopy of Arf6 uKO and control mouse testes.

Testes were collected from 16 days post tamoxifen-treated male control mice (a to d) and Arf6 uKO mice (e to i) and examined by transmission electron microscopy. (a) Type A spermatogonia (A) and Sertoli cell (Ser) resting on basal lamina (BL), and intercellular bridge (ICB) connected Type B spermatogonia (b), primary spermatocytes (c) and round spermatids (d) from control testes are shown. ICB are labeled and indicated with an arrow. Enlargement of dashed areas is shown. Panels e to g show the basal region of seminiferous tubules from 16 dpt Arf6 uKO mouse testes consisting of Sertoli cells (Ser), round spermatids (RS), and elongated spermatids (ES). Two degrading cells in panel g are indicated by arrowheads. Panels h and I show round spermatids with blebbed plasma membrane indicated with double arrowheads. m, mitochondria; a, acrosome; and cb, chromatoid bodies. Magnifications for panels e to g were taken at 5000 X and panels a to d and h to i at 10,000 X. Scale bars: 1 μm for panels a to d, h and i and 2 μm for panels e to g.

Germ cell-specific inactivation of Arf6 causes loss of spermatogenic cells

To determine whether the Sertoli cell-only phenotype observed in the Arf6 uKO mice resulted from loss of ARF6 specifically in germ cells or testicular somatic cells, we crossed Arf6 floxed mice with inducible Ddx4-CreERT2^Tg/+ mice to generate germ cell-specific Arf6 knockout (referred to as Arf6 gKO) mice (John et al., 2008) (Fig. 9a). Tamoxifen was administered by oral gavage as detailed in Figure 2a and testes were harvested at days 2, 4, 6, 8, 12, 16, and 20 after treatment. Like Arf6 uKO mice, Arf6 gKO mice were grossly indistinguishable from their littermate controls. Within testes, the Arf6 gKO morphological phenotype was like that found in ubiquitous Arf6 uKO testes. The weight decrease found in Arf6 uKO testes was more pronounced than that in Arf6 gKO testes (e.g., compare day 20 in Fig. 2c to Fig. 9b, 40 mg ± 3 mg vs 65 mg ± 2.6 mg). Reduced relative abundance of DDX4 and TEX14 in Arf6 gKO testes was first observed at 16 dpt, whereas it was first observed in Arf6 uKO testes at 12 dpt (compare Fig. 9c to Fig. 2d). Immunoblotting revealed a small decrease in ARF6 protein abundance in Arf6 gKO testes at 20 dpt. The persistence of ARF6 in Arf6 gKO testis was likely due to expression in Sertoli and Leydig cells after germ cell-specific deletion.

Fig. 9. Germ cell-specific inactivation of Arf6 leads to depletion of spermatogenic cells.

(a) ARF6 mouse line generation. Exon 1 and exon 2 were floxed by two loxP sequences (open triangles) and a FRT sequence (filled triangle). Coding sequence of Arf6 is indicated by a solid bar within exon 2. Floxed Arf6 (f-Arf6) mouse was crossed with transgenic mice carrying an Ddx4-CreERT2 allele and their progenies were intercrossed to generate various germ cell-specific Arf6 genotypes. CRE recombinase expression was induced by tamoxifen to generate gKO mouse.

(b) Box and whisker plot of testes weight at day 2 (control testes, n=4, gKO testes, n=4), day 4 (n= 6 and n=4, respectively), day 6 (n=4 and n=12, respectively), day 8 (n=6 and n=8, respectively), day 12 (n=10 and n=20, respectively), day 16 (n= 10 and n=15, respectively), and day 20 (n= 6 and n=14, respectively) post tamoxifen treatment. *, P<0.001, and #, P<0.0002 (t-test).

(c) Tamoxifen treated-testicular protein lysate were immunoblotted with indicated antisera. Each lane represents individual mouse testicular lysate collected at indicated times post tamoxifen treatment. Indicated protein abundance relative to control at each time point is indicated below blots. Protein mobility standards are indicated.

(d) Hematoxylin-eosin-stained cross sections of Stage I-VI, VII-VIII, IX-X, XI and XII seminiferous tubules from Ctrl testis at 16 dpt and Arf6 gKO testes from 8, 12 and 16 dpt are shown. Red arrows indicate apoptotic cells of intense cytoplasmic eosin staining. Multinucleated giant cells are marked by black arrows. Sg, spermatogonia; In, intermediate spermatogonia; B, spermatogonia B; Pl, preleptotene; L, leptotene; Z, zygotene; P, pachytene; D, diplotene spermatocytes; Mi, metaphase spermatocyte; S2, secondary spermatocyte; RS, round spermatids; ES, elongated spermatids; and Ser, Sertoli cell are indicated.

(e) Representative image of hematoxylin-eosin stained Arf6 gKO testes at day 20 post tamoxifen is shown. Recognizable stages are indicated, and vacuoles are marked by asterisks.

(f) Summary of time-dependent progressive cell loss by type following tamoxifen exposure in Arf6 gKO male mice based on morphological analysis and GCNA1 immunofluorescence microscopy (Fig. 10). Solid lines indicate the presence of cells and cell attenuation are marked by dotted lines. The appearance of either apoptotic cells, multinucleated giant cells or vacuole are noted.

We examined time-dependent progressive germ cell loss in Arf6 gKO testes by morphological analysis (Fig. 9d−f). We observed a reduction of preleptotene, leptotene, zygotene and pachytene spermatocytes in Arf6 gKO testes at 8 dpt (Fig. 9d). In addition, at 8 dpt, we also observed binucleated cells and multinucleated giant cells in stage I-VI tubules, binucleated round spermatid cells in Stage VII-VIII, and apoptotic cells with intense eosin-staining in Stage XII tubules. By 12 dpt, preleptotene, leptotene, and zygotene spermatocytes were no longer observed, and pachytene, diplotene and metaphase spermatocyte abundance was clearly reduced. Like Arf6 uKO testes, we observed an increased number of multinucleated giant cells with intense eosin cytoplasmic staining and secondary spermatocytes in stage XII with vacuole formation. At 16 dpt Arf6 gKO testes, pachytene spermatocytes were not observed and there was progressive reduction of diplotene and metaphase spermatocytes. Although seminiferous tubules at 16 dpt could still be readily staged, multinucleated giant cells were prominent in stages VII to XII, and tubular cellularity was greatly decreased due to loss of cell populations as compared to control (Fig. 9d). This loss of cellularity persisted to 20 dpt (Fig. 9e).

Figure 9f summarizes the time-dependent progressive morphological changes observed by cell type following tamoxifen exposure in Arf6 gKO male mice. We examined the spermatogonia population in the Arf6 gKO mouse testis by performing immunofluorescence microscopy using antisera against ARF6 and GCNA1, and PNA lectin staining. In tamoxifen-treated control mouse testes, most tubules were marked with cells of strong immunoreactivity to GCNA1 (or GCNA1-hi) within the basal compartment. Prominent ARF6 signals were observed localized between GCNA1-positive cells and in ring-like structures (Fig. 10, arrows and arrowheads). Both 6 dpt and 8 dpt Arf6 gKO tubules displayed ARF6 and GCNA1 immunoreactivity like control tubules, although deletion of GCNA1-hi spermatogonia was suggested by the observation that this staining was less densely spaced. By 12 dpt, heterogeneity of GCNA1 and ARF6 immunoreactivity was more apparent; tubules devoid of GCNA1-hi positive spermatogonia were among normal tubules. Deletion of ARF6 protein appeared to occur in some seminiferous tubules and not others and GCNA1 expression showed a similar trend. This observation became more apparent in 20 dpt gKO tubules (Fig. 10). Taken together, the Arf6 gKO phenotype was like that obtained in Arf6 uKO testes, although gKO testes appeared heterogeneously effected.

Fig. 10. Progressive loss of GCNA1 positive spermatogonia induced by Arf6 deletion.

(a) Paraffin-embedded sections (7 μm) of Ctrl and Arf6 gKO mouse testis collected from 6 to 20 dpt were examined by immunofluorescence microscopy using antisera against ARF6 (green) and GCNA1 (magenta) and by staining with peanut agglutinin (PNA, white). Enlargement of regions is marked by white dashed boxes. White arrows indicate ARF6 signals localized between prominent GCNA1-positive cells in the basal compartment (marked by two dashed lines), yellow arrows indicate ARF6 signals localized between spermatocytes of GCNA1-lo signals in the adluminal compartment; arrowheads indicate ARF6-immunoreactive ring-like structures. Scale bar: 25 μm.

Like the Arf6 uKO, loss of tubular cellularity corresponded with a clear increase in programmed cell death by TUNEL assay (Fig. 11). By histology, Arf6 gKO mouse testis appeared to phenocopy Arf6 uKO mouse testis, albeit with delayed progression of germ cell loss and more heterogeneity. These results demonstrate that ARF6 plays an essential germ cell-intrinsic role in spermatogenesis.

Fig. 11. Progressive apoptosis determined by TUNEL Assay.

Paraffin-embedded sections (7 μm) of Ctrl and Arf6 gKO mouse testis collected at 2, 4, 6, 8, and 12 dpt were examined by TUNEL assay analysis. They were processed according to the manufacturer’s directions for TUNEL analysis and counterstained with PNA. Representative images are shown. Magenta asterisks indicate tubules containing TUNEL-positive cells. BOX, enlargement of regions marked by white dashed box. Scale bar: 100 μm.

Discussion

Here we show that ADP-ribosylation factor GTPase ARF6 is essential for mammalian spermatogenesis. Arf6 deletion from adult germ cells causes a rapid loss of spermatogonia and spermatocytes, resulting in a Sertoli cell-only syndrome. The mechanism by which Arf6 deletion results in this phenotype is uncertain. It is possible that loss of ARF6 might impair ICB formation or function.

In adult mice, we and others showed that ARF6 was detected in the intercellular bridges of spermatogonia and spermatocytes but not in mature spermatozoa or spermatids (Katsumata et al., 2017). Stable ICBs are required for synchronization of spermatogenesis by maintaining these cells in syncytia and are important for the expansion of the undifferentiated spermatogonia and subsequent differentiation (Rezende-Melo et al., 2020). ICBs are thought to provide structural channels ranging from 0.5- to 3-μm in diameter connecting germ cells for the exchange of cytoplasmic materials and signaling molecules. Normal ICB dynamics are regarded as a prerequisite for commitment to differentiation in spermatogonial stem cells (Greenbaum et al., 2007, 2011). Dyer et al. demonstrated Drosophila ARF6 was necessary for cleavage furrow formation and ring canal establishment during cytokinesis in spermatocytes (Dyer et al., 2007). Katsumata et al. extended this observation to mammals, reporting that mouse ARF6 localizes to ICBs of spermatogonia and spermatocytes (Katsumata et al., 2017). Our study demonstrates the essential function of ARF6 in mouse spermatogenesis and ICB formation. Therefore, the function of ARF6 in spermatogenesis is evolutionarily conserved in fly and mouse and possibly in other species.

An association between ICBs and male infertility was demonstrated in two previously published mouse knockout models: Tex14 knockout and conditional male germ cell-specific RacGap1 knockout (Greenbaum et al., 2006; Lorès et al., 2014). In Tex14-deficient mice, spermatogenesis was arrested at meiosis and ICBs failed to form (Greenbaum et al., 2006) and a reduction of undifferentiated spermatogonia was observed (Rezende-Melo et al., 2020). TEX14 is thought to stabilize ICBs by competitively interacting with CEP55 to prevent the recruitment of ALIX and TSG101 to the midbody, which in turn, is necessary for abscission and completion of cell division (Iwamori et al., 2010; Kim et al., 2015). The RacGap1 mutant mice lacked TEX14 within ICBs, displayed a Sertoli cell-only syndrome phenotype, and exhibited multinucleated cells, suggesting a failure in cytokinesis (Lorès et al., 2014). The multinucleated cell phenotype was observed in cultured Arf6-deficient mouse embryonic fibroblasts and Arf6-deficient Drosophila spermatocytes (Dyer et al., 2007; Makyio et al., 2012). Consistent with these observations, we observed a similar phenotype that included loss of ICBs, loss of TEX14 protein, and multinucleated giant cells in both Arf6 uKO or gKO adult mice.

In conclusion, ARF6-dependent signaling is essential for spermatogenesis in mice. ARF6 is a ubiquitously expressed protein and yet its deletion in adult mice reveals a profound tissue-specific testicular phenotype. In humans, the molecular basis of Sertoli cell-only syndrome is variable and remains undefined in most cases. The observation that ARF6 is essential for spermatogenesis provides a foundation for exploring associated cellular mechanisms that might provide insights into the causes of infertility.

Materials and methods

Mice

Mice were maintained and used for experiments according to the protocol approved by the Institutional Care and Use Committee of the University of Pennsylvania.

Floxed Arf6 mice were generated as described (Lin et al., 2017). Both floxed Arf6 and Ubc-creERT2 alleles (JAX stock number: 007001) were backcrossed at least 10 generations onto C57Bl/6J background. Homozygous floxed Arf6 (Arf6^f/f) mice were first bred to Ubc-CreERT2^Tg/+ mice to generate heterozygous floxed Arf6 (Arf6^f/+, Ubc-CreERT2^Tg/+), and these mice were treated with tamoxifen to generate heterozygous null Arf6 (Arf6^+/−, Ubc-CreERT2^Tg/+) breeders. Breeding of Homozygous floxed Arf6 (Arf6^f/f) mice and heterozygous null Arf6 (Arf6^+/−, Ubc-CreERT2^Tg/+) generated experimental Arf6^f/f, Ubc-CreERT2^Tg/+ and Arf6^f/−, Ubc-CreERT2^Tg/+ and littermate control Arf6^f/f, Ubc-CreERT2^+/+ and Arf6^f/−, Ubc-CreERT2^+/+ mice. To generate whole body Arf6 knockout (uKO) mice, three doses of tamoxifen at 3 mg/10 g body weight over a five-day period were given orally via a 21G feeding tube connected to a 1 ml syringe to induce CRE-mediated deletion of the floxed exons to generate Arf6 null allele. . Littermate controls were similarly treated with tamoxifen.

Homozygous Arf6^f/f mice were bred to Ddx4-CreERT2^Tg/+ mice (JAX, stock number: 006954) and their progenies were interbred to generate tamoxifen inducible germ-cell specific Arf6^f/f, Ddx4-CreERT2^Tg/+ knockout (gKO) and littermate control Arf6^f/f, Ddx4-CreERT2^+/+ mice. These mice were backcrossed 2 generations onto C57Bl/6J background.

Mice were genotyped by two PCR strategies of either tail or testicular DNA (Lin et al., 2017). Strategy 1: Wild-type Arf6 allele (285 bp amplicon) and floxed Arf6 allele (295 bp amplicon) were assayed using primers F1 and R1, 5’-CCAAGCTTGACTTGATACCTGG-3’ and 5’- GACGAGATGACTGTGGGTAAAG-3’ respectively. Null Arf6 allele (456 bp amplicon) was assayed by PCR strategy with primers F1 and R2 ,5’- GCCCTTATTGTCAAACAAGC-3’.

Strategy 2: Wild-type Arf6 allele (406 bp amplicon) and floxed Arf6 allele (527 bp amplicon) were assayed using primers R2 and F2 5’- GTTTCACCGGGCTTTTGACAACTG-3’. β-globin amplicon (494 bp) served as internal control using primers 5’- ATCTGCTCACACAGGATAGAGAGG-3’ and 5’- CTTGAGGCTGTCCAAGTGATTCAG-3’

CreERT2 allele (300 bp amplicon) was assayed by PCR strategy with primers 5’- GCATAACCAGTGAAACAGCATTGCTG-3’ and 5’-GGACATGTTCAGGGATCGCCAGGCG-3’ and a 386 bp Myh9 amplicon using primers 5’- GCATAACCAGTGAAACAGCATTGCTG-3’ and 5’-GGACATGTTCAGGGATCGCCAGGCG-3’ serves as internal control.

Reagents

Anti-TEX14 (goat, used at 1:2000) (Greenbaum et al., 2006), anti-DDX4 (rabbit, Proteintech 51042–1–AP, used at 1:1000), anti-ARF6 (rabbit, Proteintech 20225–1–AP, used at 1:1000), anti-calnexin (rabbit, StressMarq SPC-108, used at 1:5000) antibodies, anti-SCP3 (mouse, Santa Cruz sc-74569 used at 1:1000) and horseradish peroxidase -conjugated secondary antibodies (Fisher Scientific PI31460 and Santa Cruz sc-2020 used at 1:5000) were used for immunoblotting analysis. The GCNA1 (Germ Cell nuclear antigen 1,10D9G11, used at 1:250) monoclonal antibody developed by George Enders at the University of Kansas Medical School was obtained from the Developmental Studies Hybridoma Bank, created by the NICHD of the NIH and maintained at the University of Iowa, Department of Biology, Iowa Cit, IA 52242. Anti-TEX14 (goat, used at 1:1000) (Greenbaum et al., 2006), anti-ARF6 (rabbit, Abcam ab131261, used at 1:250) and anti-LIN28 (goat, R&D Systems AF3757, used at 1:250) antisera, Alexa Fluor −488, −594 and −647 conjugated secondary antibodies (Invitrogen A11034, A11007 and A-21447, used at 1:2000) and Prolong Gold antifade mounting media with DAPI (Molecular Probes P36935) were used for immunofluorescence analysis. Fluorescein Peanut Agglutinin (PNA, FL-1071) purchased from Vector Laboratories was used at 1:5000. Tamoxifen solution was prepared by dissolving one gram of tamoxifen (Sigma, T5648) in 3 ml of absolute ethanol at 60°C and diluted with 30 ml of pre-warmed corn oil (Sigma, C8267) to a final concentration of approximately 30 mg/ml. Tamoxifen solution was aliquoted and stored at −20°C. TUNEL assay was performed on paraffin-embedded sections using In Situ Cell Death Detection kit (Millipore Sigma 12156792910) according to the manufacturer’s protocol.

Assessment of Fertility

Arf6^f/f, Ubc-CreERT2^Tg/+ males and littermates between two to three months of age were treated with tamoxifen. One month post tamoxifen, each tamoxifen-treated male mouse was initially housed with two young virgin females for three months. At the end of this first mating, mice were separated for a week and mating partners were switched for another two months. The results were tabulated in Table 1. Three-months old Arf6^f/f, Ubc-CreERT2^Tg/+ female mice and littermate controls were treated with tamoxifen. One month post tamoxifen, one stud male was mated with one Arf6 knockout female and one littermate control female. Data for female breeding was tabulated in Table 2.

Immunoblotting analysis

Immunoblotting experiments were performed using the procedures described previously (Mata et al., 1996). Briefly, testes were minced with a pair of surgical scissors in lysis buffer (50 mM HEPES pH7.5, 150 mM NaCl, 1.5 mL MgCl2, 1 mM sodium vanadate, 50 mM sodium fluoride, 20 mM b-glycerophosphate, 10% glycerol, 1% NP40, 0.5% sodium deoxycholate, 0.1% SDS and supplement with EDTA-free protease inhibitor cocktail per manufacturer’s recommendation (Millipore Sigma, 11836170001). The lysate was sheared through a 21-gauge needle ten times followed by centrifugation for 10 min at 15,000 rpm in a microcentrifuge at 4°C. Samples containing 30 μg of protein lysate were electrophoresed on an SDS-polyacrylamide gel and transferred onto a nitrocellulose membrane. The membranes were blocked, probed with indicated antibodies and then horseradish peroxidase-conjugated secondary antibodies. Membranes were incubated with the Clarity ECL chemiluminescent reagents (BioRad 1705061) and visualized by autoradiography as per manufacturer’s instruction.

Histological and Immunofluorescence analyses, image acquisition and processing

Freshly isolated whole testes were rinsed briefly in PBS buffer, fixed in Bouin’s solution (Ricca Chemical, 112016) and embedded in paraffin as described (Shi et al., 2019). Paraffin embedded testes were sectioned at 7 μm onto glass slides and stained with either hematoxylin and eosin or proceeded to indirect immunofluorescence microscopy with indicated antisera as described with modification (George et al., 2014). Antigen retrieval was done in Tris-EDTA buffer (pH 9.0) by heating sections to 95°C for 40 min.

Histology images were acquired using Leica DM5500 microscope with a DFC450 digital color camera (Leica Microsystems) and EVOS FL Auto Imaging System at using its CMOS color digital camera (Life Technologies). Scale bars were embedded onto images during acquisition. Confocal immunofluorescence images were acquired at 20X magnification using Leica TCS SP8 multiphoton confocal microscope. Histograms stretching was used equally for each channel to adjust image’s brightness and contrast using Fiji ImageJ software version 1.54f.

Transmission electron microscopy, image acquisition and processing

Freshly dissected testes were immediately immersed in a solution of 100mM Sodium Cacodylate buffer pH7.4 with 2% PFA, 2.5% glutaradehyde and 0.02% picric acid incubated overnight at 4oC. After subsequent buffer washes, the samples were post-fixed in 2.0% osmium tetroxide with 1.5% potassium ferricyanide for 1 hour at room temperature and rinsed in distilled water. After dehydration through a graded ethanol series, the tissues were infiltrated and embedded in EMbed-812 (Electron Microscopy Sciences, Fort Washington, PA). Thin sections (70 nm) were stained with uranyl acetate and SATO’s lead citrate and mounted onto copper grids. Transmission electron images were acquired using JEOL JEM-1010 electron microscope fitted with a Hamamatsu digital camera and AMT Advantage NanoSprint500 software at 80 kV at both 5000x and 10000x magnifications. TEM images were processed using CLAHE (Enhance Local contrast) function in Fiji ImageJ software version 1.54f.

Statistical analysis and atlas browsers

Data were presented as Mean ± S.E.M. box and whisker plot and analyzed using 2 tailed unpaired student t tests. Human Testis Atlas Browser by Cairns lab at Utah (https://humantestisatlas.shinyapps.io/humantestisatlas1/) was used to browse for ARF6 expression in human germ cells (Guo et al., 2018). Basal layer of strongly GCNA1 immunofluorescence immunoreactivity of seminiferous tubules were categorized as >50%, <50% or none from images acquired by Leica DM5500 microscope. Fisher’s exact test was used to calculate the p-values at each time point to compare the distribution of counts in each category between KO vs Control groups.

Highlights.

• Arf6 null adult male mice were infertile, exhibiting a Sertoli-cell-only phenotype.

• Germline-specific Arf6 deletion results in similar progressive loss of germ cells.

• ARF6 is a component of intercellular bridges in male germ cells.

• Arf6 deletion resulted in disruption of germ cell intercellular bridges.

• ARF6 plays an essential germ-cell-autonomous role in spermatogenesis.