Author contributions: K.K.: conception and design, collection and assembly of data, data analysis and interpretation and manuscript writing; P.R., R.K.S., M.J.A.-B., D.W.H., B.G., and J.K.: collection and assembly of data; N.T.: conception and design, data analysis and interpretation, and manuscript writing; S.K.: provision of study material or patient biopsies; M.Z.: data analysis and interpretation; H.R.S.: conception and design, data analysis and interpretation, manuscript writing, and final manuscript approval. K.K., P.R., and N.T. contributed equally to this work.
Disclosure of potential conflicts of interest is found at the end of this article.
First published online in STEM CELLS EXPRESS June 7, 2011.
Human adult germline stem cells (haGSCs) were established from human testicular biopsies and were claimed to be pluripotent. Recently, the gene expression profile of haGSCs demonstrated that these cells presented with a fibroblast rather than a pluripotent identity. Nevertheless, haGSCs were reported to generate teratomas. In this report, we address this discrepancy. Instead of using haGSCs, which are no longer available for the stem cell community, we used a human testicular fibroblastic cell (hTFC) line that presents with a gene expression profile highly similar to that of haGSCs. Indeed, as shown by microarray analysis, the similarity between hTFCs and haGSCs is comparable to human embryonic stem cell (hESC) lines derived by different laboratories. We argue that the almost identical gene expression profile of hTFCs and haGSCs should result in a very similar if not identical differentiation potential. Strikingly, hTFCs were not able to generate teratomas after injection into nude mice. Instead, they formed a mesenchymal lesion that morphologically resembled the putative haGSC-derived teratomas reported previously. We conclude that haGSCs, which exhibit a profile similar to that of fibroblasts and could not generate teratomas, are not pluripotent. Future work will have to show if pluripotent cells can be derived from human testicular biopsies. Mouse work and certain testicular germ cell tumors indicate that this will be possible. STEM CELLS 2011;29:1304–1309
Several reports have demonstrated that mouse unipotent spermatogonial stem cells (SSC) can be dedifferentiated to a pluripotent state under specific culture conditions without any genetic modification [1–4]. Reproducing these results in human SSC would represent a source of patient-specific pluripotent cells for regenerative medicine purposes. In 2008, Conrad et al.  reported that pluripotent cells, named human adult germline stem cells (haGSCs), could be derived from human testis, thus emulating the mouse reports. Subsequently, other groups also claimed that pluripotent embryonic stem cell (ESC)-like cells similar to haGSCs could be obtained during the culture of human testicular cells [6–8].
Following the NIH guidelines, human pluripotency is defined by a pluripotent global gene expression profile and by the ability to generate teratomas . Recently, the haGSC microarray analysis reported by Conrad et al. was reanalyzed, demonstrating that haGSCs are not pluripotent but represent fibroblast-like cells . Unfortunately, the gene profile of the pluripotent ESC-like cells reported by the other groups could not been analyzed, as they did not publish the microarray data [6–8]. Strikingly, in contrast to the ESC-like cells reported by the other groups, only haGSCs were able to generate typical human embryonic stem cell (hESC) teratomas in nude mice. For human pluripotent cells, the teratoma assay is defined as the gold standard for determining pluripotency. After subcutaneous, intramuscular, or intratesticular injection into immunodeficient mice, the pluripotent cells spontaneously form cystic and benign teratoma-like masses containing ectoderm (as nerve and skin), mesoderm (as bone, cartilage, and muscle), and endoderm (as liver and gut) tissues. Therefore, formation of differentiated cells from the three somatic germ layers is taken as the best indicator for demonstrating human pluripotency .
In this study, we intend to resolve the incongruity presented by haGSC, an ability to generate teratomas despite their nonpluripotent gene profile. For this purpose, we established a human testicular fibroblastic cell (hTFC) line that highly resembles haGSC, as shown by global gene expression profile, and we analyzed its ability to generate teratomas. Our study demonstrates that haGSC cannot form teratomas but a mesenchymal lesion that is in accordance to their fibroblastic gene profile.
MATERIALS AND METHODS
Derivation of hTFCs
hTFC isolation was performed as described previously .
Generation of iPS Cells from hTFCs
Induced pluripotent stem (iPS) cells were generated as reported previously [12, 13].
Teratoma Histology and Immunohistochemistry
Tissues were fixed with 4% paraformaldehyde, embedded in paraffin, and sectioned (3 μm thick). Primary antibodies from DAKO (Glostrup, Denmark) and Sigma-Aldrich (St. Louis, MO) are described in (Table 1).
Table 1. Antibodies used for immunohistochemical analysis of the teratoma sections
Characterization of the Generated iPS Cells
Bisulfite sequencing, immunofluorescence, quantitative reverse transcription polymerase chain reaction (qRT-PCR) for viral detection, microarray analysis, and teratoma assay were performed as described previously . The hTFC microarray data is available from the Gene Expression Omnibus (GEO) website under the GSE17772 accession number. The rest of the microarray data were downloaded from the GEO database with accession numbers GSE11350 (haGSCs and hESCs [H1])  and GSE15148 (hESC lines H1L, H7, H9 H13B, and H14A) . The arrays were globally normalized with the Robust Multiarray Average algorithm implemented in R-Bioconductor .
RESULTS AND DISCUSSION
hTFCs were established from human testicular biopsies . hTFCs formed clusters in culture that morphologically resembled haGSCs . These fibroblast aggregates could either be trypsinized to give rise to a fibroblast monolayer (Fig. 1A), which subsequently produced clusters (Fig. 1B), or mechanically dissociated and propagated, never losing the aggregate morphology. hTFCs and haGSCs were very similar not only in morphology (Fig. 1C), but also in gene expression profile, as shown by scatter plot analysis . The distance between haGSCs and hTFCs was 0.058, measured as one minus the Pearson's correlation coefficient, which is in the same distance range (0.010–0.060) of six hESC lines derived by two different laboratories [5, 15] (Fig. 1D), with the lowest number representing the highest similarity. In contrast, the distance between Conrad et al.'s H1 hESCs and haGSCs is reflected by a coefficient of 0.11, comparable to the distance between H1 hESCs and hTFCs, with a coefficient of 0.17. These findings demonstrate that haGSCs, which Conrad et al. claimed to be pluripotent, actually resemble fibroblasts and not hESCs .
Several groups have reported the isolation of ESC-like cells from human testis; however, only Conrad et al. claimed that they were able to generate hESCs typical teratomas. To date, Conrad et al. have not distributed the haGSCs to any independent research group. In fact, these haGSCs have already been discarded, as the patient authorization form in the study of Conrad et al. specified the destruction of these cells 3 years after isolation from the patient. As shown by microarray analysis, the similarity between hTFCs and haGSCs is comparable to hESCs cultured in different laboratories. Such a high similarity in gene expression profile should reflect a similar pluripotent potential. Therefore, we evaluated the teratoma ability of haGSCs by using hTFCs. Clumps of hTFCs were injected into nude mice to assess for teratoma formation. iPS cells generated from two independent hTFC cell lines, and termed human testicular induced pluripotent stem (hTiPS) cells, were used as the teratoma positive control. Three hTiPS cells lines were established for each hTFC cell line and their pluripotent profile was demonstrated by hESC colony morphology (Fig. 2A), alkaline phosphatase staining (Fig. 2B), immunofluorescence for pluripotent markers (Fig. 2C), retroviral silencing (Fig. 3A), methylation analysis for the OCT4 and NANOG promoters (Fig. 3B), and global gene expression analysis (Fig. 3C–3D).
hTiPS cells generated a typical cystic benign teratoma containing mature tissues of all three germ layers (Fig. 4A–4C). Immunohistochemical staining was performed for quantitative assessment of the teratoma tissues (Table 1). We detected positive stainings for: vimentin and α-smooth muscle actin (SMA), markers for mesoderm and smooth muscle cells, respectively (Fig. 4D, 4E); caldesmon and desmin, markers for smooth and skeletal muscle differentiation (Fig. 4F, 4G); CD34, a marker for hematopoietic progenitors and endothelial cells (Fig. 4H); pan-cytokeratin (CK), a marker for both endodermal and ectodermal differentiation (Fig. 4I); CD117 (c-kit), a marker for immature endothelia (Fig. 4M); α-fetoprotein, a marker for hepatocytes (Fig. 4N); neuron-specific enolase (NSE), CD56, and S100, markers for neuronal tissues (Fig. 4O–4Q); and CK 7 (Fig. 4J). In addition, a negative staining for the tumor suppressor gene p53 provided a clear sign of terminal differentiation (Fig. 4K). Moreover, only cells within neural rosettes, representing neural stem cells, showed proliferation marker Ki67+ activity, underlining the terminal differentiation of the tissues (Fig. 4L). The proliferation index of the whole teratoma was <3%.
In contrast, the injected hTFCs were characterized by a storiformous growth pattern resembling a nodular lesion of mesenchymal origin that was neither encapsulated nor discernible from the mouse host tissue (Fig. 4A′–4C′). Turnbull-blue-positive siderophages marked the injection side of the mouse. In contrast to the hTiPS teratoma, this nodular lesion stained positive only for the mesodermal markers vimentin and α-SMA (Fig. 4D′, 4E′), an expression pattern typical of (myo)fibroblasts. Neither mature muscle, epithelial, nor endodermal markers were expressed (Fig. 4F′–4N′). Although the lesion resembled peripheral nerves, the injected cells stained negative for the neural markers NSE, S100, and CD56 (Fig. 4O′–4Q′). The proliferation index determined by Ki67 staining was <1%, indicating a low-proliferative cell fraction. Noteworthy, only hTFC clumps but not dissociated hTFCs were able to generate these mesenchymal lesions, suggesting that the clump structure prevented their elimination from the mouse host. These observations, together with the small nodule size, suggest that the nodules correspond to the initial injected clumps that managed to survive without proliferating. In conclusion, hTFCs, which highly correlate with haGSCs in gene expression profile, are not able to generate teratomas, but only a nodular mesenchymal lesion that resembles the putative teratoma reported by Conrad et al. .
As is the case for hTFCs, the haGSC teratoma did not meet the typical characteristics/requirements for a well-defined, encapsulated, cystic tumor, and presented with an intramuscular localization, despite the subcutaneous injection of the haGSCs. Furthermore, it is difficult to comprehend how the teratoma tissues could be distinguished from the host mouse tissue, particularly considering that only three antibodies were used for the immunohistochemical analysis and their mouse cross-reactivity was not excluded. In addition, the subcutaneous and intramuscular aggregates of the putative teratoma resemble necrotic tissue, which might actually reflect the accumulation of the 10 million injected cells. Finally, the teratoma presented in the supplemental data by Conrad et al.  showed no clearly discernible tissues at either low or high magnification and the blue part of the teratoma, evident in the H&E staining overview, does not correspond to the tissues shown at higher magnification. From a histopathological standpoint, a more detailed analysis is required before identifying the structure as a teratoma. In summary, based on our evaluation, formation of a typical teratoma was not demonstrated after injection of haGSCs  and hTFCs into nude mice.
Several reports recommend that researchers investigating teratoma formation conduct immunohistochemical analysis of the putative teratoma, together with histological analysis [17, 18], to corroborate the presence of mature differentiated cells of all three germ layers. In addition, if antibodies specific to human genes—i.e., without cross-reactivity to the mouse counterparts—are available, it is preferable that they would be used to avoid misinterpretations. These requirements should be met by all studies presenting teratoma formation as proof for pluripotency, as the teratoma assay remains the most rigorous test for ascertaining the cellular pluripotency of human cells. We consider our study important in proving the erroneous attribution of pluripotency to haGSCs and in guiding future related studies. In summary, haGSCs are not pluripotent due to their previously reported fibroblastic-like gene expression profile  and to the inability to generate teratomas that has been demonstrated in this study. Future work will have to show if pluripotent cells can be derived from human testicular biopsies as has been suggested by mouse work  and certain testicular germ cell tumors .
We thank Boris Burr for technical assistance. This work was supported by the Max Planck Society, the German Research Foundation (DFG) grant SPP 1356 “Pluripotency and Cellular Reprogramming” (SCHO 340/5-1), and the Federal Ministry of Education and Research (BMBF) grant “Disease-specific iPS cells” (FKZ 01GN0811).
DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
The authors indicate no potential conflicts of interest.