Autologous transplantation of spermatogonial stem cells restores fertility in congenitally infertile mice

Significance

Stem cell transplantation is widely used to rescue defects in stem cell-derived cells. However, it is generally impossible to rescue tissue dysfunction caused by defective microenvironment. In this study, we report that autologous spermatogonial stem cell (SSC) transplantation rescues congenital male infertility caused by Cldn11 deficiency. Cldn11-deficient mice lack spermatogenesis due to defects in the blood–testis barrier. However, WT or Cldn11-deficient SSC transplantation allowed development of fertile sperm from the donor cells in chemically castrated Cldn11-deficient mice. Because in vivo depletion of Cldn3 or Cldn5 restored endogenous spermatogenesis, complete spermatogenesis may be inhibited by the imbalance of claudin expression caused by Cldn11 deficiency. Our result suggests that some forms of male infertility can be rescued by autologous SSC transplantation.

Abstract

The blood–testis barrier (BTB) is thought to be indispensable for spermatogenesis because it creates a special environment for meiosis and protects haploid cells from the immune system. The BTB divides the seminiferous tubules into the adluminal and basal compartments. Spermatogonial stem cells (SSCs) have a unique ability to transmigrate from the adluminal compartment to the basal compartment through the BTB upon transplantation into the seminiferous tubule. Here, we analyzed the role of Cldn11, a major component of the BTB, in spermatogenesis using spermatogonial transplantation. Cldn11-deficient mice are infertile due to the cessation of spermatogenesis at the spermatocyte stage. Cldn11-deficient SSCs failed to colonize wild-type testes efficiently, and Cldn11-deficient SSCs that underwent double depletion of Cldn3 and Cldn5 showed minimal colonization, suggesting that claudins on SSCs are necessary for transmigration. However, Cldn11-deficient Sertoli cells increased SSC homing efficiency by >3-fold, suggesting that CLDN11 in Sertoli cells inhibits transmigration of SSCs through the BTB. In contrast to endogenous SSCs in intact Cldn11-deficient testes, those from WT or Cldn11-deficient testes regenerated sperm in Cldn11-deficient testes. The success of this autologous transplantation appears to depend on removal of endogenous germ cells for recipient preparation, which reprogrammed claudin expression patterns in Sertoli cells. Consistent with this idea, in vivo depletion of Cldn3/5 regenerated endogenous spermatogenesis in Cldn11-deficient mice. Thus, coordinated claudin expression in both SSCs and Sertoli cells expression is necessary for SSC homing and regeneration of spermatogenesis, and autologous stem cell transplantation can rescue congenital defects of a self-renewing tissue.

Spermatogenesis is maintained by close interactions between germ cells and somatic cells. Any defects in this interaction result in defective spermatogenesis, leading to infertility. Spermatogonial stem cells (SSCs) undergo self-renewal divisions (1, 2) and can recolonize empty seminiferous tubules and regenerate spermatogenesis (3). SSCs from a congenitally defective microenvironment can rescue SSC defects caused by Kit mutations such that normal offspring are born (4). However, spermatogenic defects can occur due to a number of mutations, most of which cannot be explained at the molecular level. This is particularly true in clinical cases, in which many responsible genes have not yet been identified.

The blood–testis barrier (BTB) forms between the Sertoli cells at ∼12–14 d postpartum (dpp) in mice, when they stop mitotic proliferation (5, 6). The BTB divides the seminiferous tubules into adluminal and basal compartments (7). After mitotic division of SSCs, a clone of an interconnected spermatocyte transmigrates through the BTB by continuous dynamic restructuring, and haploid cells eventually develop in the adluminal compartment (8). These spermatocytes are temporarily enclosed in an intermediate compartment and transported into the adluminal side. It is considered that the integrity of the BTB is essential for normal spermatogenesis because it creates a special environment for meiosis and also protects haploid germ cells from the immune system (5). Thus, the BTB is unique among blood–tissue barriers in the body in terms of its cell biology and immunological aspects, and understanding the molecular mechanism underlying the germ cell–Sertoli cell interaction has important implications for our understanding of infertility.

Research over the last decade has revealed the molecular structure of the BTB. Although the tight junctions of the BTB are formed between Sertoli cells, the functional BTB is composed of the Sertoli cell tight junctions, a physiological permeability barrier, and an immunological barrier (5, 9). Several tight junction proteins (TJPs) are identified, and the phenotypes of knockout (KO) mice for these components vary from normal, as seen in F11r KO mice, to slowly degenerative, as seen in Ocln KO mice (10), to sterile in Cldn11 KO mice (11). All animals with BTB defects are infertile because these defects likely cause immunological or other types of damages to the meiotic and postmeiotic cells (5, 9). Although the BTB is formed between Sertoli cells, spermatogonia and spermatocytes also express several TJPs (12). However, the roles of these TJPs are unknown because germ cells do not form tight junction by themselves. Because germ cells also express TJPs, defective spermatogenesis in TJP KO mice may be a result of defects in both the germ cells and Sertoli cells.

Here, we used spermatogonial transplantation to analyze the role of TJPs in Cldn11 KO mice, which completely lack the BTB. SSCs have the unique ability to transmigrate through the BTB, and SSCs regenerated from the transplanted SSCs can complete normal spermatogenesis (3). Therefore, this technique has been used to analyze the germ cell–Sertoli cell interaction. Of the several TJP-related KO mice, Cldn11 KO mice show the most prominent effects, because spermatogenesis in Cldn11 KO mice does not proceed beyond the spermatocyte stage (11, 13). In our attempt to analyze germ cell–Sertoli cell interaction using this model, we found that autologous SSC transplantation restores fertility.

Results

Immunohistochemical Analysis of Cldn11 KO Mice.

Cldn11 KO testes were significantly smaller than wild-type (WT) mouse testes when the testes were collected from 11-wk-old mice (Fig. 1 A and B). Although germ cells were found in the seminiferous tubules (Fig. 1C, SI Appendix, Fig. S1A), no spermatozoa were found in the epididymis (SI Appendix, Fig. S1B). Close examination by lectin immunostaining showed a lack of peanut agglutinin (PNA)-expressing haploid cells in the mutant testes (Fig. 1D). Interestingly, we occasionally found clusters of Sertoli cells in the tubule lumen (Fig. 1C, Inset) (14). Terminal deoxynucleotidyl transferase-mediated dUTP nick end-labeling (TUNEL) staining showed no evidence of increased apoptosis in these Sertoli cells (SI Appendix, Fig. S1C).

Fig. 1.

Characterization of Cldn11 KO mouse testes. (A and B) Appearance (A) and testis weight (B; n = 4–8) of Cldn11 KO mouse testis. (C and D) Histological appearance of testis (C). Sertoli cell clusters (arrow) are shown in Inset. (D) Lectin (PNA) immunostaining. (E and F) Immunostaining of CLDN3 (E) and CLDN5 (F) in busulfan-treated Cldn11 KO mouse testes. (G) Immunohistochemical analysis of KIT⁺ spermatogonia. (H) Quantification of KIT⁺ cells. At least 11 tubules were counted. (I and J) Immunohistochemical analysis of apoptotic CLGN⁺ (I) and SYCP⁺ (J) cells by TUNEL staining. Arrowheads indicate apoptotic cells. (Scale bars: A, 1 mm; C, 200 μm; D, G, I, and J, 50 μm; E and F, 20 μm.) Stain: hematoxylin & eosin (C), Hoechst 33342 (D–G, I, and J).

We also examined the impact of Cldn11 deficiency on the distribution of other TJPs after busulfan treatment, which destroys germ cells. Busulfan treatment did not influence the functional BTB because biotin microinjected into the interstitial tissue did not penetrate into the adluminal compartment (SI Appendix, Fig. S2A). Immunohistochemical analysis not only confirmed the lack of CLDN11 (SI Appendix, Fig. S1D) but also showed reduced expression of CLDN3 and OCLN after busulfan treatment (Fig. 1E and SI Appendix, Fig. S1E). In contrast, CLDN5 is more widely expressed in Sertoli cells on the basement membrane (Fig. 1F). These results suggested that TJP expression patterns in the BTB were significantly influenced by germ cells and CLDN11 (SI Appendix, Table S1).

To examine the impact of Cldn11 deficiency on the spermatogonial population, immunohistochemistry was carried out using antibodies against several spermatogonia markers. Although Cldn11 KO testes contained a reduced number of CDH1⁺ undifferentiated spermatogonia, no statistically significant difference was found (SI Appendix, Fig. S3A). However, the number of KIT⁺ differentiating spermatogonia was significantly decreased (Fig. 1 G and H). Nevertheless, the proportion of cells expressing MKI67 (proliferation marker) was comparable between the Cldn11 KO and the control testes (SI Appendix, Fig. S3A). These results show that the loss of CLDN11 influences premeiotic germ cells that are outside of the BTB.

Because a significant number of germ cells undergo apoptosis during meiosis, Cldn11 KO testes lack haploid cells. TUNEL staining was carried out and an analysis was performed to determine the number of apoptotic cells in WT mice. Quantification of TUNEL⁺ cells revealed that Cldn11 KO testes contained a large number of apoptotic cells, of which 20.5 ± 10.5% (n = 5) and 58.0 ± 16.2% (n = 3) were CLGN⁺ (spermatocytes) and SYCP3⁺ (spermatocytes to elongating spermatids) cells, respectively (Fig. 1 I and J). However, ZBTB16⁺ (undifferentiated spermatogonia) and KIT⁺ (differentiating spermatogonia) cells in both Cldn11 KO and control testes did not show increased apoptosis (SI Appendix, Fig. S3B). These results suggested that spermatocytes are the major cell type that undergoes apoptosis due to Cldn11 deficiency.

SSC Activity of Cldn11 KO Mice.

In the first set of transplantation experiments, Cldn11 KO mice were used as donors to examine whether Cldn11 deficiency influences SSC activity. To introduce a donor cell marker, Cldn11 KO mice were crossed with the transgenic mouse line C57BL6/Tg14 (act-EGFP-OsbY01) (green mouse). The testis cells were collected from both KO and littermate control WT mice. Total cell recovery from Cldn11 KO testis cells was significantly decreased (Fig. 2A). The cells were then transplanted into the seminiferous tubules of busulfan-treated mouse testes after dissociation into single cells.

Fig. 2.

Functional analysis of SSCs in Cldn11 KO mice. (A) Cell recovery (n = 4). (B) Appearance of recipient testis transplanted with Cldn11 KO mouse testis cells. (C) Colony counts (n = 22 for Cldn11 KO; n = 20 for WT). (D) Total SSC number in Cldn11 mouse testis (n = 4). (E) Lectin (PNA) and SYCP3 immunostaining of recipient testis. (Scale bars: B, 1 mm; E, 20 μm.) Stain: Hoechst 33342 (E).

Two months after transplantation, the recipient mice were killed and their testes were examined under ultraviolet (UV) light to visualize donor cell colonization (Fig. 2B). The numbers of colonies generated by the Cldn11 KO and control testis cells were 5.5 and 1.8 per 10⁵ transplanted cells, respectively (n = 22 for Cldn11 KO; n = 20 for WT) (Fig. 2C). The difference was statistically significant. When the total number of SSCs per testis was calculated (cell recovery × colony counts), it was significantly smaller in Cldn11 KO mouse testes than that in WT testes (831.2 vs. 2160.0 per testis) (Fig. 2D). Immunohistochemical staining confirmed normal spermatogenesis from the transplanted SSCs (Fig. 2E). These results suggested that Cldn11 KO testes contain a smaller number of SSCs, which can differentiate normally into sperm once they are provided with a normal environment.

Impact of Cldn11 Deficiency in SSC Homing.

The results in the preceding section showed the enrichment of SSCs in Cldn11 KO mice. However, it was possible that the SSCs were enriched due to the lack of haploid cells. Alternatively, a lack of Cldn11 might have influenced SSC homing. To directly test the impact of Cldn11 dosage on SSC homing, we used germ-line stem (GS) cells, cultured spermatogonia with enriched SSC activity (15). We first examined the impact of Cldn11 overexpression (OE). In these experiments, GS cells that expressed enhanced green fluorescent protein (EGFP) were transfected with Cldn11 cDNA by lentivirus, and GS cells that stably expressed Cldn11 (Cldn11 OE GS cells) were genetically selected by drug selection (Fig. 3A). Real-time PCR analysis showed that the transfected cells expressed Cldn11 at ∼160-fold (Fig. 3B). The infected cells were subsequently transplanted into WBB6F1-W/W^v (W) mice. W mice have mutations in the Kit tyrosine kinase and contain only a small number of undifferentiated spermatogonia (4). However, they can support spermatogenesis after transplantation of WT SSCs. Quantification of colony numbers revealed a comparable number of germ cell colonies generated by Cldn11 OE and control cells, i.e., 120.0 ± 34.0 and 120.0 ± 35.0 (n = 10) per 10⁵ transplanted cells (Fig. 3 C and D), respectively, suggesting that Cldn11 OE did not influence colony formation.

Fig. 3.

Impact of Cldn11 dosage in GS cells on SSC colonization. (A) Appearance of Cldn11 OE GS cells. (B) Real-time PCR analysis of Cldn11 expression in GS cells transfected with Cldn11 (n = 4). (C) Appearance of recipient testes transplanted with Cldn11 OE GS cells. (D) Colony count (n = 10). (E) Appearance or recipient testes transplanted with Cldn11 KO GS cells. (F) Colony count (n = 17). (G) Appearance of recipient testis transplanted with Cldn3/5 KD-Cldn11 KO GS cells. (H) Colony count (n = 18–19). (Scale bars: A, 50 μm; C, E, and G, 1 mm.)

Next, we examined whether decreased Cldn11 expression influenced the SSC activity of GS cells. We derived GS cells from Cldn11 KO mice (Cldn11 KO GS cells). The cells were labeled with a Venus-expressing lentivirus, and transplantation experiments were conducted. The recipient testes revealed reduced colonization of Cldn11 KO GS cells (Fig. 3 E and F). The numbers of colonies generated by the Cldn11 KO and control GS cells were 38.8 and 158.8 per 10⁵ transplanted cells, respectively (n = 17). The difference between the experimental and control samples was significant. We also depleted CLDN3 and CLDN5 in Cldn11 KO GS cells, which further decreased the colonization levels (Fig. 3 G and H), suggesting that CLDN3 and CLDN5 were necessary for colonization of Cldn11 KO SSCs. These results indicated that a synergy between the three claudins in SSCs was necessary for successful colonization, and that increased colonization of Cldn11 KO testis cells was likely caused by a lack of differentiating germ cells, considering the lower seeding frequency of Cldn11 KO SSCs.

Enhanced Colonization of SSCs in Cldn11 KO Mouse Testes.

Next, we used Cldn11 KO mice as recipients to examine the impact of Cldn11 on SSC colonization. We also investigated the effect of a GnRH analog (leuprolide acetate) treatment, which increases donor cell colony numbers and length by suppressing the hypothalamus-pituitary axis (16). Because CLDN11 is reportedly influenced by testosterone and testosterone regulates the permeability of the BTB (17, 18), we anticipated that leuprolide might interfere with the integrity of the BTB and enhance donor cell colonization.

Cldn11 KO mice were treated with busulfan when the animals were 7 wk old. Some of the KO and control mice were also treated with leuprolide at 4 and 9 wk after busulfan treatment for maximum effect (19). Donor cells were prepared from green mouse testes, and the cells were transplanted into Cldn11 KO mice at least 4 wk after the busulfan treatment. In the leuprolide experiments, the donor cells from green mice were transplanted the week after the last leuprolide treatment. Leuprolide treatment down-regulated testosterone levels, but did not disrupt the functional BTB, which was confirmed by biotin tracer experiments (SI Appendix, Fig. S2 A and B).

The number of colonies generated in leuprolide-treated Cldn11 KO, untreated Cldn11 KO, leuprolide-treated WT, and untreated WT mice were 9.6, 5.2, 2.6, and 1.6 per 10⁵ transplanted testis cells, respectively (n = 14 for Cldn11 KO; n = 16 for WT) (Fig. 4 A and B). Because the same donor cells generated significantly more colonies in Cldn11 KO mice than in the control mice, this confirmed that CLDN11 expression in Sertoli cells inhibits SSC colonization. Although the number of colonies was increased in leuprolide-treated WT mice, the difference between the leuprolide-treated and untreated mice was not statistically significant. However, the difference between leuprolide-treated and untreated Cldn11 KO mice was statistically significant. These results indicated that leuprolide treatment increased donor cell colonization in Cldn11 KO mice, suggesting that leuprolide-mediated colonization enhancement does not depend on Cldn11.

Fig. 4.

Enhanced colonization of SSCs in Cldn11 KO mouse testes. (A) Appearance of Cldn11 KO recipient testes. (B) Colony count (n = 14–16). L, leuprolide. (C) Immunostaining of CLDN11 in a W mouse testis that received shRNA against Cldn11 5 d after microinjection. (D) Quantification of fluorescence intensity after Cldn11 KD. At least 17 cells were counted. (E) Appearance of busulfan-treated WT recipient testes that were depleted of Cldn11 expression. (F) Colony count (n = 17–18). (G) Real-time PCR analysis of chemokine expression in Cldn11 KO mouse testis (n = 4). Both untreated and busulfan-treated mice were used. (Scale bars: A and E, 1 mm; C, 50 μm.) Stain: Hoechst 33342 (C).

Enhanced Colonization of Donor SSCs by Cldn11 Depletion in Recipient Testes.

In the next set of experiments, we examined the feasibility of manipulating Cldn11 in WT seminiferous tubules, which already have an established BTB. Here, we used W mice. Although conflicting observations were reported for the BTB in W mice (20–22), complete spermatogenesis from transplanted SSCs suggested that spermatogenesis can occur even with defective BTB (4). To test whether Cldn11 knockdown (KD) could enhance SSC colonization in W mice, lentivirus particles expressing a short hairpin RNA (shRNA) against Cldn11 were microinjected into the seminiferous tubules. Immunohistochemical analysis showed that the expression of Cldn11 decreased to 61.1% in the recipient testes 5 d after microinjection (Fig. 4 C and D). GS cells were then transplanted into the seminiferous tubules 7 d after lentivirus injection.

We found significantly increased colony numbers when we injected shRNA against Cldn11 prior to transplantation (Fig. 4 E and F). The numbers of colonies generated in Cldn11 KD and control recipient testes were 273.5 and 166.7 per 10⁵ transplanted cells, respectively (n = 17 for Cldn11 KD; n = 18 for control). These results suggested that Cldn11 down-regulation can increase colonization efficiency even after BTB formation.

Deregulated Expression of Chemokines in Cldn11 KO Mouse Testes.

Although these experiments suggested that the lack of Cldn11 enhances SSC colonization, it was still possible that Cldn11 KO mouse testes were secreting large volumes of chemokines involved in SSC homing. Increased chemokine expression might have attracted more SSCs than in the WT environment. To understand the molecular mechanism underlying the increased SSC colonization in Cldn11 KO mice, we analyzed the expression of several chemokines, including Gdnf, Cxcl12, Ccl5, Ccl7, Ccl9, and Ccl12, in both untreated and busulfan-treated mice. These genes are reportedly implicated in SSC migration into niches (23–26).

Real-time PCR analyses showed that busulfan treatment significantly altered the expression patterns of these chemokines (Fig. 4G). In untreated testes, only Ccl9 levels were comparable between the Cldn11 KO and WT testes. Although Cxcl12, Ccl7, and Ccl12 levels were significantly increased in the Cldn11 KO mouse testes, Gdnf and Cxcl5 levels were significantly decreased. However, analysis after busulfan treatment showed a significant decrease in the levels of Gdnf, Cxcl12, and Cxcl5 in Cldn11 KO testes. Gdnf expression in busulfan-treated testes was decreased to 0.4-fold relative to that in untreated mice. Cxcl12, which is involved in primordial germ cell or SSC migration, also decreased significantly in Cldn11 KO testes. Cxcl5 was most significantly decreased in Cldn11 KO mouse testes, but there were no significant changes in the rest of the genes tested. These results suggested that Cldn11 deficiency reduced SSC chemokine gene expression and that neither Cxcl12 nor Gdnf, both of which were shown to directly influence SSC homing in vivo (24), was responsible for the enhanced colonization of SSCs in Cldn11 KO mice.

Restoration of Spermatogenesis in Cldn11 KO Mouse Testes.

Through analysis of Cldn11 KO recipient testes, we noted that the donor WT cells formed colonies with multiple layers of germ cells (Fig. 5A), which suggested that SSCs differentiated into haploid cells despite the defective environment. To confirm the degree of differentiation, immunostaining of Cldn11 KO recipient testes was carried out. While the germ cells in intact Cldn11 KO testes did not progress beyond meiosis, we noted many tubules in Cldn11 KO recipient testes with PNA⁺ haploid cells (Fig. 5B). We also confirmed the lack of CLDN11 in Cldn11 KO recipient testes (Fig. 5C). Although no infiltrating CD4⁺ or CD8⁺ lymphocytes were found in these testes (SI Appendix, Fig. S4A), real-time PCR analysis revealed a significant increase in Ccl2 and Tnf expression levels (SI Appendix, Fig. S4B). These results suggested that spermatogenesis occurred in the absence of the BTB despite inflammation.

Fig. 5.

Regeneration of spermatogenesis in Cldn11 KO mouse testes after spermatogonial transplantation. (A) Appearance of EGFP-expressing WT or Cldn11 KO testis cell colonies in busulfan-treated Cldn11 KO testes showing tubules with adluminal donor cell differentiation. (B) Lectin (PNA) immunostaining of recipient testes. (C) CLDN11 expression in recipient testes after EGFP-expressing donor cell colonization. (D) Offspring born from Cldn11 KO recipients after microinsemination using germ cells from Venus-labeled WT GS cells (Left) or Cldn11 KO testis cells (Right). (E) Histological appearance of Cldn11 KO testis after double depletion of Cldn3 and Cldn5. Arrows indicate elongated spermatids. (F) Offspring born after microinsemination using round spermatids developed in Cldn11 KO mouse after double depletion of Cldn3 and Cldn5. (G) A hypothetical model of suppressed spermatogenesis in Cldn11 KO mice. Although CLDN3 expression is enhanced in untreated Cldn11 KO mice, it disappears after busulfan treatment. However, germ cell transplantation initiates CLDN3 expression. This probably starts in the center of the colonies in which most differentiated cells are found. A germ cell-induced increase in CLDN3 expression would create regions with normal levels of CLDN3, which may allow regeneration of spermatogenesis. (Scale bars: A, 1 mm; B and C, 20 μm; E, 50 μm.) Stain: Hoechst 33342 (B and C) and hematoxylin & eosin (E).

Based on these observations, autologous transplantation of Cldn11 KO testis cells was performed. Because Cldn11 KO testis cells could undergo spermatogenesis in busulfan-treated WT recipients, we deduced that homologous interaction of CLDN11 between germ cells and Sertoli cells is dispensable for spermatogenesis progression at least after transplantation. In our initial experiments, we collected the right testis cells and stored for 4 wk to allow for busulfan treatment of the left testis prior to transplantation. Histological sections showed development of PNA⁺ haploid cells in all three recipients (SI Appendix, Fig. S5). Because regeneration of endogenous spermatogenesis cannot be excluded without a marker, we used green Cldn11 KO testis cells into busulfan-treated Cldn11 KO mouse testes in the next experiments, which confirmed donor cell-derived germ cell colonies (Fig. 5A). Immunohistochemistry showed differentiation of germ cells into PNA⁺ haploid cells (Fig. 5B).

Although PNA⁺ cells were found in Cldn11 KO testes regardless of the donor genotype, it was possible that they lacked fertilization activity because they developed in the absence of the BTB. Microinsemination experiments were conducted to test whether these haploid germ cells were functionally normal. In these experiments, we used Cldn11 KO recipient testes transplanted with WT GS cells with a Venus transgene (lentivirus) or Cldn11 KO testis cells with an Egfp transgene (green mouse). Seminiferous tubules with spermatogenic colonies were dissociated into a single cell suspension. Elongated spermatids or spermatozoa were microinjected into WT oocytes for fertilization (SI Appendix, Table S2). Both donor cell types produced normal offspring. While 26 offspring were born when WT donor GS cells were used, 11 offspring were born from Cldn11 KO donor testis cells (Fig. 5D). The donor cell origin was confirmed by PCR using WT GS cells because Venus expression was not clear under a UV light (SI Appendix, Fig. S6A). These results suggested that SSCs can mature into functional gametes in Cldn11 KO mouse testes regardless of their genotype.

Regeneration of Endogenous Spermatogenesis in Cldn11 KO Mice.

To understand the mechanism of spermatogenesis regeneration after autologous transplantation, immunostaining of other claudins was carried out. Although CLDN3 expression in Sertoli cells was lost after busulfan treatment in Cldn11 KO mice (Fig. 1E), it was restored after spermatogonial transplantation (SI Appendix, Fig. S7A), suggesting that germ cells induce CLDN3 expression in Cldn11 KO mice. However, CLDN5 expression was enhanced in Cldn11 KO mice after busulfan treatment (Fig. 1F). However, CLDN5 expression decreased after spermatogonial transplantation into Cldn11 KO mice (SI Appendix, Fig. S7B). When untreated Cldn11 KO mice were examined, intensity of CLDN3 and CLDN5 staining was stronger (SI Appendix, Fig. S7 C–F). Therefore, we reasoned that strong CLDN3 and CLDN5 expression in Sertoli cells might inhibit spermatogenesis in untreated Cldn11 KO mice.

To test this hypothesis, we microinjected lentivirus expressing shRNA against Cldn3 and/or Cldn5 into untreated Cldn11 KO mouse testes. Two months after microinjection, we noted PNA⁺ haploid cells in Cldn11 KO testes (Fig. 5E and SI Appendix, Fig. S8A). Although no significant differences were found among the three groups (Cldn3, Cldn5, and Cldn3+Cldn5 KD), spermatogenesis occurred most extensively after double KD of Cldn3 and Cldn5 (SI Appendix, Fig. S8B). The same treatment did not show apparent impact in WT testes (SI Appendix, Fig. S8A). To confirm the fertility of the germ cells, the seminiferous tubules of Cldn11 KO testes were dissociated ∼2 mo after double KD of Cldn3 and Cldn5 and used for microinsemination. Although elongated spermatids were not recovered after cell dissociation, a total of 49 eggs were constructed using round spermatids and four offspring were born (Fig. 5F and SI Appendix, Table S2). As expected, all of the offspring were heterozygous for the Cldn11 KO allele because oocytes from WT females were used (SI Appendix, Fig. S6B). PCR analysis confirmed the lack of transgenes in the offspring (SI Appendix, Fig. S6B).

Discussion

To understand the impact of the BTB on spermatogenesis and SSC homing, we used Cldn11 KO mice. Our initial experiments showed that premeiotic Cldn11 KO germ cells that are outside of the BTB are not completely normal. Moreover, transplantation studies showed impaired SSC colonization when Cldn11 KO GS cells were transplanted into WT testes, suggesting that CLDN11 positively promotes SSC colonization through the BTB. Therefore, a lack of CLDN11 in the germ cells may prevent passage through the BTB during normal spermatogenesis. However, CLDN11 in the Sertoli cells inhibits SSC colonization because WT SSCs colonized more efficiently in Cldn11 KO testes. These results showed that the function of CLDN11 differs between germ cells and Sertoli cells in SSC homing.

In the course of our analyses of recipient testes, we found haploid cells in Cldn11 KO mice. Surprisingly, these haploid cells were capable of producing offspring. These results were unexpected because a lack of a BTB would change the composition of the seminiferous tubule fluid and disrupt the normal cytokine expression pattern. Moreover, although postmeiotic cell development is supported by mitotically quiescent, polarized Sertoli cells under physiological conditions, our results suggested that neither the cell cycle status nor the polarity of Sertoli cells is critical for completing functional spermatogenesis. Because inflammatory cytokines were up-regulated in recipient testes, production of haploid cells caused immune response. However, unlike after allogeneic transplantation (25), we did not find infiltration of lymphocytes and haploid cells were not rejected. Because conflicting observations on the BTB function are reported for busulfan-treated mice and W mice (20–22), more extensive studies need to be carried out to confirm the role of the BTB in normal spermatogenesis. Nevertheless, our results based on nonphysiological conditions revealed a remarkable flexibility of spermatogenesis. Perhaps, the BTB may simply amplify spermatogenesis efficiency.

There were significant changes in claudin expression patterns after transplantation in both the germ cells and Sertoli cells. In particular, CLDN3 expression depended on the germ cells because its expression disappeared after busulfan treatment. Unlike untreated Cldn11 KO mice that contain germ cells in all areas of the seminiferous tubules, spermatogenesis from transplanted SSCs occurs in restricted areas because only a limited number of SSCs can colonize the recipient testes (26). Given the lack of CLDN3 in busulfan-treated Cldn11 KO Sertoli cells, only CLDN5 was expressed in Cldn11 KO Sertoli cells. If CLDN3 is induced by germ cells, CLDN3 expression would only occur when Sertoli cells interact with appropriate numbers of germ cells. In germ cell transplantation, colonies tend to differentiate in the center (26). This would create a gradient of CLDN3 expression in developing colonies (Fig. 5G). However, such heterogeneity of CLDN3 expression may not occur during spermatogenesis in intact Cldn11 KO testes because all areas of the seminiferous tubules are filled with germ cells to the same degree. We reasoned that this potential heterogeneity of claudin expression in germ cell colonies of busulfan-treated testes might have allowed some germ cells to pass through the BTB.

To test this hypothesis, we carried out in vivo KD experiments. Because both Cldn3 and Cldn5 KD induced spermatogenesis in intact Cldn11 KO testis, both claudins are involved in suppression of endogenous spermatogenesis. Therefore, a balance in claudin expression levels appears to be critical for the completion of spermatogenesis. However, we still do not know how these claudins blocked spermatogenesis. Because transgenes were not found in the offspring and lentivirus cannot infect endogenous germ cells (27), claudin expression in Sertoli cells are probably responsible for regeneration after in vivo KD. One possibility is that increased expression of CLDN3 and CLDN5 in Sertoli cells enhanced the adhesiveness between Sertoli cells, which may physically block germ cell migration. Although Cldn3 KO mice does not have apparent phenotype (28), increased Cldn3 expression might physically block spermatogenesis. However, this is unlikely because Sertoli cells are mitotically active in Cldn11 KO mice (14). It is difficult to conceive that proliferating Sertoli cells adhere more strongly to each other and block spermatogenesis. We rather think that increased claudin expression in Sertoli cells might have directly interacted with claudins on germ cells. Because Cldn11 KO germ cells lacked CLDN5 expression, this interaction may occur via CLDN3. Perhaps CLDN3 on Sertoli cells may trap CLDN3-expressing spermatocytes and inhibit their differentiation into haploid cells. Given the success with Cldn5 KD, however, CLDN5 may indirectly influence CLDN3 expression. Further investigations into the type of germ cells that induce CLDN3 expression and the nature of claudin interaction between germ cells and Sertoli cells are warranted.

SSC transplantation has been widely used for functional analysis of SSCs and dissection of germ cell–Sertoli cell interaction. However, the possibility of autologous transplantation has been overlooked. The current study demonstrated the involvement of Cldn11 in SSC homing but also showed that the BTB is dispensable for spermatogenesis. In addition, lack of immunological rejection for haploid cells suggests that it is not the BTB per se, but rather the immunosuppressive properties of Sertoli cells are responsible for the immune privilege of regenerated germ cells. Our study also provides a possibility in stem cell transplantation therapy and suggests that other forms of congenital male infertility can be rescued by SSC transplantation or slight modification of testicular environment.

Materials and Methods

Animals and Microinjection Procedure.

Cldn11 KO mice were provided by S. Tsukita (Osaka University, Suita, Japan) (13). We crossed Cldn11 KO mice with green mice (gift from M. Okabe, Osaka University, Suita, Japan) to introduce a donor cell marker. For the transplantation, 4-wk-old C57BL/6 (B6) or B6 × DBA/2 F1 (BDF1) mice were injected intraperitoneally with busulfan (44 mg/kg). For the transplantation experiments, busulfan was administered when the animals were 7 wk old. At least 1 mo following the busulfan treatment, the animals were used for transplantation. In some experiments, we also used 4- to 6-wk-old W mice for microinjection of Cldn11 KD lentivirus particles (1.2 × 10⁸/mL with polybrene [333 μg/mL]) 1 wk before transplantation (Japan Shizuoka Laboratory Animal Center). For the microinjection, dissociated single-cell suspensions or virus particles were inserted into seminiferous tubules via the efferent duct (29). Each injection filled 75–85% of the seminiferous tubules. Where indicated, we administered the GnRH agonist leuprorelin acetate (0.19 mg per mouse; Takeda Pharmaceutical Co.) by subcutaneous injection 9 and 4 wk before transplantation, as described previously (19). The Institutional Animal Care and Use Committee of Kyoto University approved all of the animal experimentation protocols.

Statistical Analyses.

Results are presented as the means ± SEM. Data were analyzed using Student’s t tests. Multiple comparison analyses were performed using ANOVA followed by Tukey’s honestly significant difference (HSD) test.

SI Appendix.

Additional data discussed in the paper and full details of methods are provided in SI Appendix.