We cloned Lin28 (previously known as Tex17) from mouse spermatogonia in a cDNA subtraction screen 7. Semi-quantitative RT-PCR analysis using enriched germ cell populations showed that the expression of Lin28 in testis is restricted to spermatogonia 8. Western blot analysis of a panel of adult mouse tissues revealed that the LIN28 protein is abundantly expressed in testis but not in other tissues examined (Fig. 1A). The testis of XX^Y* male mice completely lack germ cells but contain somatic cells such as Sertoli and Leydig cells 14. The absence of LIN28 in XX^Y* testes demonstrated that LIN28 was germ cell-specific (Fig. 1B), in agreement with previous studies 7. As controls, LIN28 protein was abundant in mouse embryonic stem cells but absent in fibroblast feeder cells (Fig. 1B). We examined the relative protein level of LIN28 in juvenile testes. LIN28 was detectable in testes right after birth, increased its abundance with age, and was most abundant around puberty (day 12 – 18) (Fig. 1C).

We next examined the expression of LIN28 by immunostaining of juvenile testis sections (Additional file 1). LIN28 was expressed in gonocytes from postnatal day 1-old mice and in spermatogonia from day 6- and day 14-old mice. LIN28 was predominantly cytoplasmic with punctate nuclear staining. Immunostaining analysis of adult testis sections revealed that LIN28-expressing cells were sparsely located at the periphery of seminiferous tubules (close to the basement membrane), suggesting that LIN28 is expressed in a subset of spermatogonia but not in meiotic and post-meiotic germ cells (Fig. 2A and Additional file 2). We also found that the number of LIN28-positive spermatogonia per tubule varied among the stages of seminiferous tubules (Fig. 2B). The seminiferous epithelium of mice is divided into twelve stages, each of which is defined by a unique association of differentiating germ cells 15. For example, undifferentiated A_al spermatogonia become differentiating A1 spermatogonia during stages VII-VIII 3. Our results showed that the number of LIN28-positive spermatogonia peaked at stage VIII but decreased sharply at stage IX, indicating that LIN28 might be expressed in undifferentiated spermatogonia (Fig. 2B).

To address whether LIN28-expressing cells are indeed undifferentiated spermatogonia, we performed whole-mount immunofluorescent studies on dissected seminiferous tubules from adult testes. In these studies, undifferentiated spermatogonia can be definitively identified as A_s, A_pr, or A_al. The A_pr and A_al spermatogonia are connected by intercellular cytoplasmic bridges as a chain of 2ⁿ cells. We found that the expression of LIN28 was restricted to A_s and the chained 2ⁿ cells (1, 2, 4, 8, 16, or 32 cells) (Fig. 3A). A chain of 32-interconnected cells was very rare (Fig. 3). Chains with a non-2ⁿ number of LIN28-positive cells were also observed at a low frequency (Fig. 3B). GFRA1 (GDNF receptor) is a marker of undifferentiated spermatogonia 1617. As expected, LIN28 was expressed in GFRA1-positive spermatogonia (Additional file 3). These whole-mount analyses demonstrated that LIN28 marks the undifferentiated spermatogonia.

Spermatogonial stem cells (SSCs) are believed to be a subset of A_s cells 3. Currently, there are no cytological markers that could distinguish SSCs from "non-stem" A_s cells. To examine whether LIN28 is expressed in SSCs, we performed double immunostaining of cultured spermatogonia highly enriched for SSCs with anti-LIN28 and anti-PLZF or anti-GFRA1 antibodies. PLZF is required for maintenance of SSCs 1819. We found that LIN28 was expressed in cultured SSCs, but the abundance of LIN28 in SSCs was not uniform, suggesting the heterogeneity of in vitro cultured SSCs (Fig. 4A and Additional file 2).

In an attempt to determine the role of Lin28 in the maintenance of SSCs, we treated SSCs with Lin28 siRNAs. The siRNA knockdown decreased the level of Lin28 mRNA by 60% and consequently reduced the abundance of LIN28 protein by nearly 60% (Fig. 4B, C). However, siRNA treatment did not causes a change in the total number of cultured cells (Fig. 4D), suggesting that the remaining LIN28 protein might be sufficient for maintaining SSC or that LIN28 is dispensable for the survival of SSCs.

Several recent studies have demonstrated that LIN28 is a negative regulator of let-7 microRNA biogenesis in embryonic stem cells and other stem cells 2021222324. Specifically, LIN28 prevents Dicer from processing let-7 microRNAs by mediating the terminal uridylation of let-7 microRNA precursors 21. In agreement with these studies, siRNA knockdown of LIN28 in cultured SSCs led to an increased level of mature let-7g miRNA (Fig. 4E).

Ngn3 is specifically expressed in undifferentiated spermatogonia (A_s to A_al) 25. To determine if Ngn3 and Lin28 mark the same population of undifferentiated spermatogonia, we made use of Ngn3-GFP mice, in which GFP was inserted into the Ngn3 locus by gene replacement 26. We performed whole-mount immunostaining of Ngn3-GFP seminiferous tubules with anti-LIN28 and anti-GFP antibodies. This analysis revealed that only a subpopulation of LIN28-positive spermatogonia was GFP-positive (Fig. 5A). Overall, ~40% of LIN28-positive A_s spermatogonia were GFP-positive, supporting that the population of A_s cells were not homogeneous. The A_pr and A_al spermatogonia were either all GFP-positive or all GFP-negative (Fig. 5A) except a few as described later (Fig. 6). Interestingly, the percentage of Ngn3-GFP-positive spermatogonia increased dramatically as spermatogonia develop from A_s to A_al (16 cells) (Fig. 5B). While ~40% of A_s cells were GFP-positive, nearly all A_al (16-cell) spermatogonia were GFP-positive. As the number of chained cells increases, spermatogonia become more and more committed to differentiation. Taken together, our data suggested that Ngn3 delineates a more committed subpopulation of undifferentiated spermatogonia, in contrast, the LIN28-positive but Ngn3-GFP-negative spermatogonia are more primitive.

We observed heterogeneity of Ngn3-GFP expression among A_pr and A_al spermatogonia (Fig. 6). In A_pr spermatogonia, one cell was GFP-positive and the other was GFP-negative (Fig. 6A). In an 8-cell chain of A_al spermatogonia, seven cells were GFP-positive but one was GFP-negative (Fig. 6B). In a 16-cell chain of A_al spermatogonia, two cells in the middle of the chain were GFP-negative (Fig. 6C). Twelve out of 710 clusters examined (1.7%) were found to contain both GFP-positive and GFP-negative cells in the same chain (one A_pr, two 4-cell A_al, six 8-cell A_al, and three 16-cell A_al spermatogonia). The presence of Ngn3-GFP-negative cells in a chain of GFP-positive spermatogonia suggested that the GFP-negative cells might have de-differentiated and thus reverted to a more primitive (stem cell) fate.

Mouse tissues were homogenized using a glass homogenizer in the extraction buffer (62.5 mM Tris-HCl, pH 6.8, 3% SDS, 10% glycerol, 5% β-mercaptoethanol). Protein lysate (20 μg) was separated on 12% SDS-PAGE gels and electro-blotted onto PVDF membranes. Western blotting was performed using the following antibodies: goat anti-LIN28 antibody (1:100, Cat# AF3757, R&D Systems) and anti-β-actin monoclonal antibody (1:2,500, Cat# A5441, Sigma-Aldrich). HRP-conjugated secondary antibodies were used (Sigma-Aldrich).

To prepare frozen sections, testes from C57BL/6J mice of postnatal day 1, 6, 14 or 2-month (adult) were fixed in 4% paraformaldehyde (PFA) at 4°C for 8 hours and were dehydrated in 30% (w/v) sucrose overnight. Testes were embedded with Neg 50 tissue freezing solution (Cat# 6502, Thermo Scientific) and frozen in dry ice/ethanol. Sections (8 μm) were cut using a Reichert-Jung cryo-microtome and then post-fixed in 4% PFA at room temperature for 10 minutes prior to immunostaining.

For whole-mount analysis, seminiferous tubules from adult (2-month-old) C57BL/6J mice were prepared as previously described with modifications 41. Briefly, testis tubules were washed once with PBS, fixed in 5 ml 4% PFA for 3 hours, and incubated sequentially with 5 ml of 25%, 50%, 75% and 100% TBST (1×TBS containing 0.1% Tween 20) at 4°C each for 30 minutes. Testis tubules were frozen in 1×TBS at -20°C. Immunostaining of testis sections, whole mounts of seminiferous tubules, or SSCs was performed with the following primary antibodies: goat anti-LIN28 (1:100), guinea pig anti-ACRV1 (1:500, gift from PP Reddi) 42, rabbit anti-GFP (1:500, Cat# Ab6556, Abcam), anti-PLZF (1:200, Cat# OP128L, Calbiochem), and anti-GFRA1 (1:20, Cat# sc-10716, Santa Cruz Biotech). Texas red or FITC-conjugated secondary antibodies were used (Vector Laboratories). Nuclear DNA was stained with DAPI provided in mounting medium. Samples were visualized under a Zeiss Axioskop 40 fluorescence microscope. Images were captured with an Evolution QEi digital camera (MediaCybernetics) and processed with the Image-Pro software (Phase 3 Imaging Systems).

The derivation of Ngn3-GFP mice has been described previously 26. In Ngn3-GFP mice, the enhanced green fluorescent protein (eGFP) substitutes the Ngn3 coding region through gene replacement; thus GFP is under the transcriptional control of all endogenous Ngn3 regulatory elements. Adult (2-month-old) Ngn3-GFP heterozygous mice on a mixed (129/C57BL/6) genetic background were used, because homozygous (Ngn3^-/-) mice die by postnatal day 3. XX^Y* mice were generated by breeding XY* males with wild type females 14. The care and use of mice were within standard ethical guidelines and were approved by the Institutional Animal Care and Use Committee at the University of Pennsylvania.

Mouse spermatogonia highly enriched for spermatogonial stem cells (SSCs) were prepared and cultured as previously described 4344. Briefly, single-cell suspensions were prepared from eight testes from post-natal day 6~8 C57BL/6 pups by digestion with Trypsin-EDTA (0.25%, Invitrogen) and DNase I (7 mg/ml, Sigma). Cell suspensions were layered on top of a 30% Percoll solution and were centrifuged to enrich germ cells. After resuspension, SSCs were isolated by magnetic activated cell sorting (MACS) using Thy1.2 antibody-conjugated microbeads (Cat#130-049-101, Miltenyi Biotec). Thy1⁺ cells were seeded at a density of 0.5 – 1.0 × 10⁵ cells per well on 12-well culture plates with mitomycin C-treated STO feeders. Self-renewing SSCs were cultured in a chemically defined serum-free MEMα medium (Invitrogen) containing 0.2% BSA, 10 μg/ml Transferrin, 7.6 μeq/L free fatty acids, 3 × 10^-8 M Na₂SeO₃, 50 μM β-ME, 5 μg/ml Insulin, 60 μM Putrescine, 2 mM L-glutamine, and 10 mM HEPES), 20 ng/ml GDNF (R&D Systems), 150 ng/ml soluble GFRα1 (R&D Systems), and 1 ng/ml bFGF (BD Biosciences). The medium was changed every 2–3 days. All cultures were maintained at 37°C in a humidified 5% CO₂ incubator. Cells were passaged at 7-day intervals at 1:2–3 dilution.

Mouse Lin28 siRNAs (On-target plus Smartpool, Cat# L-050153, Thermo Scientific Dharmacon) was used. Silence^® siRNA served as a negative control (Cat# AM4611, Ambion). After trypsin digestion and washing, SSCs were plated into wells of a 12-well dish without feeders in the antibiotic-free culture medium at a density of 2 × 10⁵ cells/well. Cells were allowed to settle for 2–3 hours prior to siRNA treatment. For each well, 75 pmol of siRNA and 2 μl of LipofectamineTM RNAiMAX reagent (Invitrogen) were mixed with 200 μl of OptiMEM (Invitrogen). After the 30-hour incubation, total RNA and proteins were prepared for qPCR and western blotting. Quantitative RT-PCR (qPCR) analysis was performed using SYBR green on an ABI 7300 sequence detection system with the following Lin28 primers: AGACCAACCATTTGGAGTGC and AATCGAAACCCGTGAGACAC. Level of mature let-7g miRNA was measured by using specific TaqMan probes per the manufacturer's instructions (Applied Biosystems). Quantification of Lin28 and mature let-7g transcript levels was normalized to Rps2 (ribosomal protein S2) within the log phase of the amplification curve. For SSC count, 1 × 10⁵ cells/well after 30-hour siRNA treatment were plated onto fresh feeders, cultured in a defined serum-free media with 20 ng/ml GDNF, 150 ng/ml GFRα1 and 1 ng/ml bFGF for 7 days. Each experiment was performed on three independent SSC lines.