Differential regulation of spermatogenic process by Lkb1 isoforms in mouse testis

Liver serine/threonine kinase B1 (LKB1) is a tumor suppressor associated with the pathogenesis of Peutz-Jeghers syndrome. Affected males are at increased risk of developing Sertoli cell tumors and display defective spermatogenesis. Male mice lacking the short isoform (Lkb1S) of Lkb1 were sterile and exhibited abnormal spermiogenesis. In addition to the short isoform, the long isoform of Lkb1 (Lkb1L) is also expressed in testis; however, the requirement of the long isoform for fertility and the functional difference between the isoforms remain unknown. Herein, different from the spermiation failure reported in Lkb1S knockout mice, conditional deletion (cKO) of both isoforms of Lkb1 in germ cells resulted in male sterility stemming from defects in acrosome formation, as well as nuclear elongation and condensation during spermatid differentiation. Additionally, cKO mice showed a progressive germ cell loss that was never reported in mice with Lkb1S deletion. Further experiments revealed that the defect resulted from the failure of spermatogonial stem/progenitor cells (SPCs) maintenance. Although increased mTORC1 activity in postnatal cKO testes was consistent with a tendency toward germline stem cell differentiation, in vivo inhibition of the pathway by rapamycin treatment failed to rescue the phenotype. Concurrently, we detected a significant reduction of mitochondrial activity in Lkb1deficient SPCs. The results suggest that the regulation of LKB1 on SPCs’ maintenance is associated with mitochondrial functions but not through the mTOR signaling pathway. In summary, our study supports different roles of Lkb1 isoforms in spermatogenesis with Lkb1L directing SPCs maintenance, and Lkb1L and Lkb1S coordinately regulating spermatid differentiation.

Liver kinase B1 (Lkb1), also known as Serine/Threonine protein kinase 11 (Stk11), encodes an evolutionarily conserved serine/threonine kinase required in many physiological processes. 1 The well-known downstream substrates of LKB1 are AMP-activated protein kinase (AMPK), and 12 other AMPK-related kinases. Tremendous progress over the last few years revealed that LKB1 negatively regulates mammalian target of rapamycin complex 1 (mTORC1) activity through its substrate AMPK, and the loss of LKB1 leads to the aberrant activation of mTORC1 in a variety of tissues. 2 In brief, the LKB1/AMPK/mTORC1 cassette constitutes a canonical signaling pathway that integrates information on the metabolic and nutrient status to regulate many biological processes including cell proliferation, cell survival and autophagy.
Many studies have shown the tissue-specific functions of Lkb1 in pancreas, skin, muscle and gastrointestinal tissues. [3][4][5] In testis, the function of LKB1 was not clearly described until knockout of an alternate splice variant was reported to cause sterility in mice. 6 The newly identified isoform was termed LKB1 short form (LKB1 S ) as opposed to the previously reported long form (LKB1 L ). 6 The two proteins are identical except for the C-terminus, which is encoded by different exons, 9A versus 9B for LKB1 L and LKB1 S , respectively and thus differ in length.
Although both isoforms are widely expressed in rodent and human tissues, Lkb1 S is particularly abundant in haploid spermatids. 7 Male mice that cannot generate Lkb1 S are sterile with abnormal release of mature spermatids from the seminiferous epithelium. 7,8 Spermatogenesis is a unique physiological process, during which diploid spermatogonial stem/progenitor cells (SPCs) undergo a series of differentiated steps, followed by meiosis and spermatid differentiation, and culminate in the production of mature haploid sperm cells. Although Lkb1 S is the dominant isoform in germ cells (round spermatids), 8 it remains unknown if the Lkb1 L isoform is expressed in germ cells and if the two isoforms function redundantly or indispensably during the process of spermatogenesis.
Herein, we report the differential expression of the two isoforms of Lkb1 in different spermatogenic cell types with Lkb1 S mainly expressed in spermatids and Lkb1 L in SPCs, respectively. After conditional deletion (cKO) of Lkb1 in male germ cells, mice were sterile as evidenced by severe defects on round spermatid differentiation and flagellum biogenesis. Meanwhile, these cKO mice also suffered a progressive loss of germ cells, resulting in a Sertoli-cell-only phenotype. Further experiments demonstrated the essential roles of Lkb1 in SPCs maintenance and spermatid differentiation.

Results
Distinct expression of Lkb1 L and Lkb1 S in different spermatogenic cell types. To delineate the roles of LKB1 in male germ cells during spermatogenesis, we first examined the expression of the two isoforms of Lkb1 in the different germ cell stages including spermatogonial stem cells (SSC), spermatocytes (SP), round spermatids (RS) and    Figure  1a-c). 9-11 Lkb1 S was predominantly expressed in differentiated germ cells (SP, RS and ES) and a marked increase was observed in RS and ES. Lkb1 L was expressed in these cell types but maintained a stable expression level. Unlike Lkb1 S , Lkb1 L was detected at a considerable level in SSCs, albeit its expression level was lower than that in other cell types ( Figure 1d). Western blot analysis further demonstrated that LKB1 S is the main isoform expressed in testis, while LKB1 L is the main isoform in SSCs (Figure 1e). The developmentally differential expression pattern of two Lkb1 isoforms points to their potentially discrepant regulatory functions in spermatogenesis.
Defects of spermiogenesis in Lkb1 cKO mice. To study the functions of Lkb1 in germ cells, we conditionally deleted Lkb1 in the male germ line by means of Vasa-Cre (Ddx4-Cre) and the flox'd allele Lkb1 loxp . 12 The result of western blot confirmed the mutation in cKO testis ( Figure 2a). As   (Figure 2b). Moreover, the sperm heads were invariably round with most displaying multi-nucleation and/or multi-flagella ( Figure 2e). Upon histological evaluation, in cKO mice the epididymal lumen was filled with sloughed germ cells or cell debris (Figure 2d). Transmission electron microscopy further demonstrated that this debris consisted of degraded multinucleated germ cells (Figure 2f). Testis sections from 5-week old control and cKO mice were used to analyze the first wave of spermatogenesis.
In control mice, mature spermatozoa observed at the luminal interface. By contrast, multinucleated giant round cells were frequently observed towards the center of the tubule in cKO mice ( Figure 2c). These results suggest that the dramatic reduction of spermatozoa in the epidydimis stemmed from a failure in the process of spermiogenesis.
Lack of acrosome formation in Lkb1 cKO mice. In mammals, spermiogenesis is characterized by a massive wave of transcriptional activity and can be accurately divided into continuous steps on the basis of nuclear elongation and acrosome morphology. 13 To further assess potential defects in spermiogenesis stemming from the Lkb1 cKO phenotype, we first compared spermatid nuclear elongation between similarly staged seminiferous tubules from control and cKO mice at 6 weeks of age. At stages from I to VIII, round spermatids with a spherical, central nucleus were commonly observed in both control and cKO mice. However, in stages IX and X, progressive elongation and condensation of spermatid nucleidid did not occur in cKO tubules ( Figure 3a). We next used fluorescently tagged peanut agglutinin to label and follow acrosome development. In rodents, acrosome development begins early in the spermiogenic process and can be divided into four phases (Golgi, Cap, Acrosome and Maturation) comprising the transformation from round spermatids to mature spermatozoa. 14 The Golgi and Cap phases, which encompass the formation of a large acrosome granule (Golgi) and its growth and flattening over the nucleus (Cap phase), were observed in stages II-VIII round spermatids from control mice (Figure 3b, upper panel). However, only occasionally, two pre-acrosomal vesicles could be observed associated with the nucleus in stages II-III round spermatids from cKO mice (Figure 3b, lower panel). Moreover, Cap flattening was seldom observed in stages IV-VIII cKO tubules and, instead, pre-acrosomal granules failed to fuse and appeared scattered over the spermatid nucleus (Figure 3b, lower panel). TEM analysis revealed the typical acrosomal granule and cap structure over round spermatids at Golgi phases in control mice ( Figure 3c). Then, cap flattening continued through acrosome-acroplaxome complex formation and culminating in acrosome development. Although the closer leaflets of nuclear envelope could be identified at the beginning of the acrosome formation in cKO mice ( Figure 3c, black arrows), it lacked a typical large acrosomal granule; furthermore, it failed to form the cap structure and instead displayed small granules scattered and attached over the nuclear membrane in Cap phase ( Figure 3c, red arrowheads, cKO versus control).
Additionally, localization of the centriole distally to begin formation of the flagellum as observed by Acrosome phase in control tubules was not observed in spermatids from cKO spermatids ( Figure 3c, white arrow). Taken together, the failure of acrosome and flagellum genesis observed in round spermatids impaired nuclear elongation and remodeling and ultimately lead to abnormal spermatozoal maturation in cKO mice. At the round spermatid stage, a large number of testisspecific genes are expressed under the control of a limited number of key regulator genes, including Crem, Trf2, Rnf17, Piwil1, Boule, Tpap, Ddx25 and Rfx2. [15][16][17][18][19][20][21][22][23][24] To investigate the potential role of Lkb1 in spermiogenesis regulation, we compared the mRNA expression of these key regulator genes in adult testes of control and cKO mice. Notably, in cKO testes there was a global significant decrease in the expression of all genes examined except Piwil1 (Figure 3d). Then, to rule out whether the decreased gene expression levels stemmed directly from faulty transcriptional regulation by Lkb1 rather than the altered ratio of cell types in adult testis, we repeated these experiments in testes from 24-day-old mice, tubules from both control and cKO mice presented a similar histomorphology (Supplementary Figures S1a,b). In agreement, at this age, cKO mice also showed a significant downregulation of most genes examined ( Figure 3e). Altogether, these results suggest that loss of Lkb1 may disrupt spermatid gene transcription.
Progressive loss of germ cells in cKO mice. While a link between Lkb1 S KO and male sterility resulting from spermiogenetic defects was previously reported, no phenotypes were reported on the other stages of spermatogenesis. 8 In this study, we recorded a gradual decrease in testis weight with advancing age (Figures 4a and b). Histology analysis also showed a progressive loss of germ cells in seminiferous tubules that finally resulted in Sertoli-only tubules ( Figure 4c). The loss of spermatogenic cells began after 5 weeks of age and, by 13 weeks, tubules undergoing various degrees of germ cell loss were widespread in cKO testis' sections ( Figure 4c; Supplementary Figures S1c-g). By 40 weeks, cKO testes became completely atrophic and nearly 90% were Sertoli-only tubules ( Figure 4d). These results are consistent with a normal onset of spermatogenesis in juvenile cKO mice followed by an age-dependent progressive deletion of spermatogenic cells from the seminiferous tubules.
Depletion of SPCs in Lkb1 cKO mice. To investigate the underlying cause for the scarcity of spermatogenic cells observed in cKO seminiferous tubules, we used the TUNEL assay to assess apoptosis in 5-and 13-week-old control and cKO testes. Although apoptotic cell counts increased in 13week cKO testis, numbers were still low (~1.2 apoptotic cells per tubule) and there was no preference for any specific cell type ( Figure 4e). These results suggested that the testicular atrophy observed in cKO mice was not caused by induction of apoptosis. We then used stage-specific spermatogenic cells markers to assess at which developmental stage germ cell loss was happening. These markers included: PCNA, a proliferation marker confined to the nuclei of spermatogonia and early spermatocytes; 25 γH2AX, in spermatocytes in meiotic prophase I; 26 and Plzf, a SPC marker. 27 In control Intriguingly, PCNA and γH2AX staining was comparable to that of controls in 5-week-old cKO mice (Figure 5a), but both of them decreased significantly by 13 weeks in cKO testes ( Figure 5b). Strikingly, in opposition to the observations with PCNA and γH2AX labeling, the number of Plzf-positive cells was reduced as early as 5 weeks in cKO mice (Figures 5a  and b). These results suggest that progressive depletion of SPCs is at the origin of germ cell loss in cKO mice.
The indispensable role of Lkb1 in SPC self-renewal and maintenance. A major advance in the study of male germline biology was the development of a culture system allowing long-term maintenance and expansion of mouse SSCs by in vitro culture. 28 To investigate the effects of Lkb1 deletion on SSC survival and maintenance, we enriched SPCs using Thy-1 microbeads according to previous reports. 29 Similar numbers of cells were separated and seeded onto STO feeder cells for culture. To our surprise, the Lkb1 cKO germ cells failed to form typical SSC colonies in culture (Figure 6b). After two passages, the Lkb1 cKO germ cells underwent complete depletion (Figure 6c). Freshly enriched SPCs from control and cKO testis were then used to evaluate the expression of SSC self-renewal and differentiation-related genes by RT-PCR. As shown in Figure 6d, key regulators of SSC self-renewal including Oct4, Bcl6b, Nanos3, Id4, and Gfrα1 displayed decreased expression in Lkb1 cKO germ cells. [30][31][32] Conversely, the expression of the SSC differentiation markers, Stra8 and c-kit was significantly upregulated in this population. 33,34 The depletion of SPCs in cKO mice is independent of the mTOR signaling pathway but associated with mitochondria dysfunction. It has been reported in a variety of tissues that the inhibition of LKB1 signaling induced the aberrant activation of the mTOR signaling pathway. [35][36][37] As a central signaling pathway in stem cell homeostasis, the aberrant activation of mTORC1 was recently shown to promote SSC differentiation in both mouse and Drosophila. 35,38,39 Since RT-PCR revealed an increased expression of differentiation markers Stra8 and c-kit in freshly isolated SSCs from cKO mice, we assessed whether the mTOR signaling pathway was functioning downstream of LKB1 in regulating SSCs maintenance and differentiation in our model. Testes were collected at different age and the highest mTORC1 activity was detected at P8 (Figure 7a). We then used P8 testes to compare mTORC1 signaling between control and cKO mice. As shown in Figure 7b, Lkb1-deficient germ cells induced a significant increase in phosphorylated S6K1 and rpS6, but no effect on p-Akt, a marker of the PI3K signaling pathway. Additionally, Lin28 and Plzf showed similar expression in both control and cKO P8 testes at mRNA and protein levels (Figures 6d and 7b). To investigate the origin of the aberrant activation of the mTOR signaling pathway, we then checked the localization of p-rpS6 on sections of control P8 testis using VASA as a germ cell marker, GATA4 as a Sertoli cell marker and Lin28a, as a SPC marker. From the localization of VASA, p-rpS6 and GATA4, it was confirmed that the activation of mTORC1 was restricted to germ cells and not all VASA-positive germ cells are p-rpS6 positive (Figure 7c). Then we used Lin28a to label SPCs in the tubules and found p-rpS6 and Lin28a seldom co-localized in the same germ cell when serial sections were used for staining. Moreover, even when they were expressed in the same germ cell, the signal for both p-rpS6 and Lin28a was always lower than that of when expressed separately (Figure 7d, arrows). Because Lin28a expressed in both undifferentiated and differentiating spermatogonia, 40 we then used Plzf to label undifferentiated germ cells and Plzf-and p-rpS6-positive cells were counted (Figure 7e). While there was only a little decrease of Plzfpositive germ cells in P8 cKO testes, p-rpS6-positive cells increased significantly in cKO testes with nearly twice as many as those in controls (Figure 7f). These results suggest that the abnormal higher activity of the mTOR signaling pathway may be related to the acceleration of spermatogonial differentiation in cKO testis. Next, we assess whether rapamycin (the mTORC1 inhibitor) could reverse the loss of SSCs in cKO mice. We use testes just after injection to test Rapamycin effect and collect testes at 5 weeks of age for analysis. As expected, rapamycin treatment caused a significant decrease of testis per body weight in both control and cKO mice compared to that in vehicle-control mice (Figures 8a and b). The level of p-rpS6 was dramatically reduced after injection and recovered to normal at 5 weeks of age (Figures 8c and d). In control mice of the rapamycintreated group, the seminiferous tubules displayed a reduced diameter and a blockade of germ cell development at stages earlier than spermiogenesis (Figure 8e, control+Rapa). Moreover, the number of Plzf-positive germ cells increased~1.4fold when compared with control vehicle-treated control testes (Figure 8f). However, rapamycin treatment in cKO mice did not have a rescue effect on SPCs accumulation in cKO mice (Figure 8e, cKO+Rapa). In fact, the number of Plzf-positive cells was slightly below that of vehicle-treated cKO testes (Figure 8f). We infer from these results that although the deletion of Lkb1 in germ cells will yield an increase in the number of differentiating spermatogonia via upregulation of mTORC1 activity, the regulation of Lkb1 on SSC self-renewal or maintenance is mTORC1-independent.
Since mitochondria dysfunction is associated with the failure of stem cell maintenance in Lkb1-deficient hematopoietic stem cells (HSCs), 41 we next test the mitochondrial membrane potential (ΔΨm), a sensitive and reliable indicator of mitochondrial activity, in Lkb1-deficient SPCs by JC-1 staining. 42 After a ratiometric analysis of JC-1 staining, a significant reduction in ΔΨm was observed in SPCs of Lkb1 cKO mice (Supplementary Figures S2a,b). Thus, instead of mTOR signaling pathway, the regulation of LKB1 on mitochondrial function may be essential in SSC maintenance and survival.

Discussion
Herein we showed that Lkb1 L and Lkb1 S functioned coordinately in regulating spermiogenesis, while Lkb1 L also served as the key regulator for SSC maintenance. Previous studies showed that Lkb1 S KO male mice were infertile with defects in the release of mature spermatids. 8 Moreover, Lkb1 S -deficient spermatozoa were immotile and presented head and tail morphological abnormalities. 7 In our study, deletion of both Lkb1 isoforms from germ cells also resulted in sterility but more severe defects in spermiogenesis. Firstly, in addition to the coiled tails or round heads observed in spermatozoa from Lkb1s KO mice, most abnormal spermatozoa found within the epididymis of Lkb1 cKO mice were multinuclear and multiflagellar, something seldom seen in Lkb1 S KO mice. Secondly, the most striking difference between Lkb1 S KO and Lkb1 cKO testes was that the latter lacked elongated and condensing spermatids in their seminiferous tubules at stages VIII-XII. Further experiments showed that the severe defects on acrosome formation happened as early as the beginning of the Golgi phase, which likely precluded the establishment of nuclear polarity and ultimately resulted in the failure on nucleus elongation and flagellar formation. 13,43 There is growing evidence that LKB1 and AMPK play pivotal roles in the establishment of cell polarity in epithelial cells, [44][45][46][47] intestinal paneth and goblet cells, 4 and pancreatic βcells. 48,49 Recent studies show that a cytoplasmic linker protein of 170 kDa, CLIP-170, functions downstream of AMPK to affect cell polarity by shifting its binding sites to the more distal ends of microtubules. 50 In mice, CLIP-170 has been identified to associate with the sperm manchette, and CLIP-170 knockout mice were subfertile and produced sperm with abnormal heads. 51 Therefore, it is feasible that the same regulatory mechanism functions for LKB1-mediated nuclear polarization during spermatid differentiation. Despite tissuespecific differences, both Lkb1 L and Lkb1 S can activate AMPK and AMPK-related kinases in vitro. 7 However, given the Spermatogenesisthe differentiation of male germ cellsis a specialized developmental process, which is precisely regulated at the transcriptional, posttranscriptional, and translational levels. 52,53 Following meiosis, the beginning of spermiogenesis is characterized by a massive wave of transcription in round spermatids and subsequent expression of genes are required for morphological and biochemical reprogramming. 54 To date, only a limited number of genes were identified as key regulators of spermiogenesis, including Crem, Rnf17, Tpap, Ddx25, Trf2, Rfx2, Piwil and Boule. [15][16][17][18][19][20][21][22][23][24] Accordingly, our results suggest a selective downregulation of these key regulators during spermiogenesis in cKO testis. However, the decreased gene expression levels stemming from Lkb1 cKO testis provided some but not sufficient support for the completion of spermiogenesis.
The detrimental effects of aberrant activation of mTORC1 in stem cell maintenance have been extensively reported. 35,39,55 Moreover, the important roles of mTORC1 in SSC differentiation have emerged in recent years. [56][57][58][59] As the major upstream Different roles of Lkb1 isoforms on spermatogenesis F Kong et al kinase to phosphorylate AMPK, LKB1 is best known to inhibit mTORC1 activity through activation of AMPK. 2 We inferred aberrant activation of mTORC1 results in accelerating SSC differentiation. However, our following experiments using in vivo rapamycin treatment failed to rescue the defects on SSC maintenance. Furthermore, to some extent, this treatment slightly accelerated germ cell depletion in Lkb1 cKO mice. Therefore, based upon these results, the effects on SSC differentiation stemming from Lkb1 deletion cannot be solely due to a downstream effect on the mTOR signaling pathway, and other mechanisms must be in place in the regulation of LKB1 on SSC maintenance and survival.
As an evolutionarily conserved regulator of cellular energy metabolism, the involvement of LKB1 in stem cell maintenance has been only reported in HSCs. 41 Inactivation of Lkb1 in mice caused progressive depletion of HSCs and eventual pancytopenia. Although the deletion of Lkb1 was associated with predictable loss of p-AMPKα and an increase in p-rpS6, it was shown that Lkb1 regulation of HSC maintenance occurred in an AMPK-and mTORC1independent manner. 41,60,61 Concurrently, the Lkb1-deficient HSCs exhibited mitochondrial defects, aneuploidy and ATP depletion. 41,60,61 In our study, survival and maintenance of SSCs was also in an mTORC1-independent manner. And, a significant decrease of mitochondrial activity was observed in Lkb1-deficient SPCs. A stem cell niche was essential for maintenance of stem cell populations. 62 Although this niche may be essentially different between HSCs and SSCs, similar phenotypes suggest that a common mechanism exists in the regulation of Lkb1 in adult stem cell homeostasis. It would be very beneficial to define the common and specific effectors when comparing the role of LKB1 in these two stem cell types. Since Lkb1 L is the major isoform expressed in SSCs, our results revealed its indispensable role on maintenance of the stem potential in SSCs.
In summary, by using Lkb1 cKO mice, we demonstrated the differential regulation mediated by Lkb1 isoforms in spermatogenesis. Although Lkb1 S is the dominant isoform expressed in testis and plays crucial roles in spermiogenesis, the Lkb1 L was also required, with both isoforms functioning independently or coordinately at different stages of spermatogenesis. Our data suggest an indispensible role of Lkb1 L in SSC maintenance and the cooperative regulations of both Lkb1 isoforms in spermatid differentiation.

Materials and Methods
Mouse genetics, husbandry and treatment. Mice carrying floxed Lkb1 alleles (FVB; 129S6-Stk11tm1Rdp/Nci) were obtained from NCI Mouse repository. Vasa-Cre mice were purchased from Model Animal Research Center of Nanjing University (Nanjing, Jiangsu, China). 63 The floxed Lkb1 mouse strain was backcrossed onto a C57BL/6 background for at least nine times. The two strains were kept on the same genetic background (C57BL/6) and were bred by standard husbandry techniques to obtain conditional knockout mice with deletion Lkb1 gene in germ cells only (Lkb1 cKO mice). The primers to detect wt and flox Lkb1 alleles are R1: 5′-CTGTGCTGCCTAATCTGTCG-3′, F2: 5′-TTCACCATCCCTTGTGACTG -3′ and F4: 5′-ATCGGAATGTGATCCAGCTT-3′. 60 The mice were housed in the animal facility at Nanjing Medical University and all animal protocols were approved by the Committee on the Ethics of Animal Experiments of Nanjing Medical University. Testes at different developmental stages (P8, P24, 5 W, 6 W, 13 W, 20 W and 40 W) were collected to evaluate morphology and other analyses.
To test the effect of rapamycin on SSC maintenance and differentiation, mice at P7 from control and Lkb1 cKO mice were treated daily with rapamycin (4 mg/kg, LC Laboratories, Woburn, MA, USA) or vehicle for 7 days as described 39 and then raised until 5 weeks of age for testes collection.
Germ cell isolation and SSC culture. Germ cell isolation and SSC culture were established according to previous study. 28 Briefly, germ cells were isolated from P8 of control and cKO mice and Thy-1-positive cells were enriched by using magnetic activated cell sorting (Miltenyi Biotech, Bergisch Gladbach, Germany). Cells were plated at a density of 1.5 to 2 × 10 5 per well on 12-well plates with mitotically inactivated STO feeder layers (ATCC, Manassas, VA, USA). Medium was replaced every 2-3 days and were passaged at a ratio of 1 : 2 every 5-6 days. Cell numbers were counted at each passage in order to plot the cell growth curve. Cultures were maintained at 37°C in an incubator with humidified 5% CO 2 and 95% air atmosphere.
Sta-put velocity sedimentation. Spermatogenic cell fractionation was performed by sedimentation of cells prepared from adult mouse testes through a BSA gradient as previously described with a modification. 64 At least 30 testes were pooled from male mice at 5 weeks of age and incubated with 1 mg/ml collagenase IV in DMEM (Life Technology, Waltham, MA, USA) for 15-20 min at 37°C in water bath and gently shook into seminiferous tubules. The cords were then washed twice in DMEM by sedimentation at unit gravity and decanting the supernatant. Further digestion were then performed with 1 mg/ml DNase I in 0.25% trypsin for 5-15 min at 37°C. Pellet cells at 500 g for 5 min and re-suspend pellet with wash buffer (10% FBS, 0.5% BSA, 200 μg/ul DNase I). Monodisperse cell suspension was made after filtering the cells with 40 μm cell strainer. The dispersed cells were layered over a linear gradient (2-4%) of BSA in DMEM, then allow to sediment at unit gravity for a total period of 2.5-3 h. The cells were collected, identified and pooled into quasihomogeneous populations. The isolated cell types were characterized on the expression of marker genes by RT-PCR.
Sperm counts and sperm staining. Adult wild-type and Lkb1 cKO mice (13 weeks of age) were killed and one caudal epididymidis was dissected from each mouse to collect spermatoza. The extruded spermatoza were incubated in 37°C PBS for 15 min and the suspension was diluted 1 : 100 for sperm counting using a hemocytometer. After cell counting, 10 μl of the diluted suspension was collected to draw a sperm smear. Then the smears were stained with eosin for morphological examination.
Acrosome staining. Testes at 6 weeks of age were collected in both control and cKO mice. Tissues were fixed in Hartman's Fixative overnight, then incubated in 5 and 30% sucrose, embedded in optimum cutting temperature compound (OCT, Tissue-Tek, Torrance, CA, USA) and cut into 8 μm sections using a microtome- Transmission electron microscopy (TEM). Adult caudal epididymidis (13 weeks) or testes at 6 weeks of age were fixed with 2.5% glutaraldehyde in 0.2 M cacodylate buffer overnight and cut into small pieces (1 mm 3 ) after washing in 0.2 M cacodylate buffer. Small blocks were dehydrated through a graded ethanol series and embedded in resin. Ultrathin sections were cut on an ultramicrotome and observed with a transmission electron microscope (Tecnai G2 Spirit Bio TWIN).
Quantitative RT-PCR analysis. RNA was extracted from testes or germ cells using TRIzol reagent (Invitrogen) according to the manufacturer's instructions. Reverse transcription was performed on 500 ngRNA using HiScript II qRT super mix (Vazyme, Nanjing, China). Quantitative real-time PCR was performed using Eva green master mix (Applied Biological Materials Inc, Vancouver, Canada) on an ABI StepOne™ Real-Time PCR System (Applied Biosystems, Foster City, CA, USA). The relative expression of each gene was calculated by a 2-ΔCT method with the expression of β-actin as internal control. All the primers for RT-PCR were seen in Supplementary  Table S1.
Measurement of mitochondrial membrane potential (ΔΨm). To measure the mitochondrial membrane potential, SPCs were incubated for 20 min at 37°C with 5% CO 2 in SSC culture medium supplemented with 2 μM JC-1 probe (Beyotime Institute of Biotechnology, Beijing, China), in accordance with the manufacturer's protocol. After washing twice, SPCs were analyzed immediately by confocal microscopy (LSM 700, Zeiss, Germany). JC-1 is a cationic dye that produces two fluorescence emission peaks, indicated by a fluorescence emission shift from green (JC-1 monomers) to red (JC-1 aggregates). Consequently, mitochondrial depolarization was indicated by a decrease in the red/green fluorescence intensity ratio.
Statistical analysis. All data are presented as the mean ± S.E.M. and the statistical significance of the difference between control and cKO mice were examined using Student's t-test with a paired two-tailed distribution. The data were considered significance when Po0.05 (*) or 0.01(**). Values were calculated by using Prism 5.0 for Macintosh (GraphPad Software, Inc., La Jolla, CA, USA).