The seminoma class of testicular germ cell tumor (TGCT) are characterized by a morphological resemblance to primordial germ cells (PGCs) or gonocytes, and chromosome duplications at 12p. Recently, it was determined that human embryonic stem cells (hESCs) express genes in common with PGCs, and that three of these genes, GDF3, STELLAR, and NANOG, are located on 12p. The current study was designed to identify whether expression of these 12p genes were elevated in seminoma relative to normal testis, and to determine whether elevated expression was unique to seminoma.
Real-time polymerase chain reaction (PCR) and immunohistochemistry were used to assess gene expression in seminoma samples relative to normal testis and endpoint PCR was used to identify the presence or absence of these genes in breast carcinoma.
GDF3 expression was increased in eight of nine seminomas compared with normal testis, whereas NANOG, OCT4, or both were expressed at the highest levels in seminoma compared with all other markers analyzed. In addition, the NANOG protein was expressed in the majority of seminoma cells. The adult meiotic germ cell markers BOULE and TEKT1 were undetectable in seminoma, whereas the embryonic and adult germ cell markers DAZL and VASA were significantly reduced. Analysis of these markers in breast carcinoma and the MCF7 breast carcinoma cell line revealed that a core hESC-transcriptional profile could be identified consisting of OCT4, NANOG, STELLAR, and GDF3 and that NANOG protein could be detected in breast carcinoma.
Testicular germ cell tumors (TGCT) of adolescents and adults are the most frequent carcinomas of Caucasian males ages 15–40.1 TGCTs are classified into either nonseminomas, which include embryonal carcinoma, yolk sac, teratomas, and choriocarcinoma or seminomas, which morphologically resemble transformed primordial germ cells (PGCs).2 Seminomas are the most common form of TGCT, accounting for 50% of cases, whereas nonseminomas are identified in 40% and mixed forms in 10% of cases (for review, see Ref. 3.)
Despite the common occurrence of TGCTs in young men, the underlying genetic mechanisms leading to their development are unknown. It is currently supposed that seminomas develop from carcinoma in situ (CIS) within the seminiferous tubules.4–6 CIS are microscopically distinct cells that reside on the basement membrane of the seminiferous tubules of the testis and have morphological features more similar to embryonic germ cells than spermatogonial stem cells. The evidence for CIS being the precursor of seminomas arises from the knowledge that they both histologically resemble PGCs and gonocytes and are characterized by positive staining for glycogen, stem cell factor receptor (c-KIT), and germ cell/placental alkaline phosphatase.7–10 However, at the genomic level seminomas are characterized by the presence of extra 12p genomic material, often appearing as an isochromosome (i12p) or as 12p11.2–p12.1 and 12p13.3 amplifications, whereas CIS does not have this characteristic amplification.11–13 Therefore, 12p duplications are hypothesized to be a later event in TGCT malignant transformation.14 Although it is clear that 12p duplications are a hallmark of invasive seminoma, to date no genes have been pathogenically linked to this malignancy.
We recently demonstrated that three genes on human chromosome 12p were elevated in two seminoma samples.15 These genes included STELLAR, also known as developmental pluripotency associated-3 (DPPA3), NANOG, and GDF3 (growth and differentiation factor 3). Interestingly, all three genes are normally expressed in fetal germ cells and human embryonic stem cells (hESCs),15 with expression of all three 12p genes decreasing with hESC differentiation.15 The finding that these three genes are located on human chromosome 12p suggests a potential association with the malignancy of seminoma relative to CIS. However, a systematic study of the expression of these three 12p genes has never been reported. In the current study, we examined expression of NANOG, STELLAR, and GDF3 in seminomas, as well as additional marker genes of hESCs, PGCs, gonocytes, and adult germ cells, to gain more information as to the identity of this carcinoma. Then, we also examined expression of the basic stem cell profile of OCT4, STELLAR, NANOG, and GDF3 in tumors other than TGCTs. The results suggest that stem cell genes may either play a direct role in different types of carcinoma progression or serve as valuable markers of tumorigenesis.
MATERIALS AND METHODS
Seminoma specimens (n = 9) were obtained from the tissue core of the Comprehensive Cancer Center at the University of California, San Francisco. Specimens were snap-frozen in liquid nitrogen at the time of radical orchiectomy. On histologic review for this study, the specimens contained 35–60% pure seminoma, with the remainder of the specimens consisting of variable amounts of connective tissue and lymphocytic infiltrates. No evidence of normal or arrested spermatogenesis was identified in any specimen. All tissues were obtained under protocols approved by the Committee on Human Research. Two panels of total RNA from adult human testis were prepared from 40 normal testes; total RNA from normal breast tissue was from US Biological (Swampscott, MA). Samples of the MCF7 breast carcinoma cell line were generously provided by Dr. Dale Leitman (University of California, San Francisco). Total RNA from an infiltrative ductal carcinoma staged using the TNM (Tumor, Node, Metastasis) system as Stage 3A T1N2M0 was purchased from ClonTech (Palo Alto, CA). The HSF-6 line (also known as UCO6) of hESCs was cultured as previously described.16 Immunofluorescence and Western blot analysis of hESCs were performed between passages 53 and 59 as previously described.16
Total RNA was isolated using RNeasy according to the manufacturer's instructions (Qiagen, Valencia, CA). RNA quality and quantity were determined by measuring absorbance at 260 and 280 nm. cDNA was constructed from 500 ng/ml of total RNA using 250 ng/ml random hexamers and Superscript II Reverse Transcriptase (Invitrogen Life Technologies, Carlsbad, CA) as previously described.16 Reverse transcriptase was omitted in negative controls.
Real-Time Polymerase Chain Reaction
Real-Time polymerase chain reaction (PCR) was performed in reaction mix containing 1× buffer, 4.5 mM MgCl2, 10 mM dATP, dGTP, dCTP, and dTTP, 2 μM primers (Table 1), and 0.4 units platinum Taq (Invitrogen), 1× SYBR green (Molecular Probes, Eugene, OR), 1× fluorescein (BioRad, Hercules, CA), and 2% DMSO. Each reaction was performed with 50 ng of the first-strand cDNA reaction described above. GAPDH was used as the internal control. Reactions were performed in duplicate and analyzed using iCycler iQ (BioRad) calibrated for use with SYBR green. Fold difference was calculated using the ΔΔCT method normalized to GAPDH and expressed as a fold difference compared with normal testis.17
Table 1. Primer Names and Sequences Used for Real Time PCR
Product size (bp)
bp: base pairs.
NANOG antibody was generated against the peptide ‘KTWFQNQRMKSKRWQK’ by the Invitrogen Custom Antibody Service. After injection of the antigen, rabbits were bled and serum tested by ELISA for increasing antibody titers. The 10th week postimmunization serum sample that had the highest titer of antibody compared with the preimmunization bleed was used to detect NANOG protein by Western blot analysis and immunohistochemistry. Seminomas were either snap-frozen in Tissue-Tek (Ted Pella, Redding, CA) OCT and sectioned at 5 μm before storage at –80 °C, or fixed in formaldehyde, processed to paraffin, and stored at room temperature. Breast carcinoma and normal breast samples as 5 μm paraffin-embedded sections were obtained from the tissue core facility at the UCSF Comprehensive Cancer Center. Frozen sections were brought to room temperature, immersed in phosphate-buffered saline (PBS), dried, and fixed for 2 minutes in 4% paraformaldehyde, pH 7.4, using a humid chamber at room temperature. Paraffin sections were dehydrated in descending concentrations of Et-OH to H2O before incubating in PBS. Sections were washed 3 × 5 minutes in PBS containing 3% bovine serum albumin (BSA) and 0.01% Tween 20 (PBS-BT) (Sigma-Aldrich, St. Louis, MO) before incubating in 10% H202 for 20 minutes followed by three washes in PBS-BT. Sections were then incubated for 5 minutes in 0.1% Triton X-100 before washing 3 × 10 minutes in PBS-BT. After washing, sections were incubated with 4% normal goat serum (Vector Laboratories, Burlingame, CA) for 30 minutes at room temperature and then NANOG immune sera or rabbit preimmune sera (1/500 dilution in 1% normal goat serum in PBS-BT). Sections were then incubated in a humid chamber at 4 °C overnight or room temperature for 1 hour. Sections were washed 3 × 10 minutes each in PBS-BT, then incubated with biotin-goat-antirabbit secondary antibody for 30 minutes at room temperature (1/200 dilution; Vector). After 3 10-minute washes in PBS-BT, sections were incubated with avidin-biotin ABC reagent (Vector). Sections were washed as above and the color reaction was developed with DAB reagent (Vector). Sections were washed in deionized water before counterstaining in hematoxylin for 10 seconds. Sections were dehydrated through ascending concentrations of ethanol to 100% ethanol before incubating overnight in xylene. Sections were mounted under glass coverslips using VectaMount (Vector).
Quantitative Gene Expression in Seminoma
We used real-time PCR on seminomas to analyze gene expression, as shown in Figure 1. We have previously shown that hESCs express OCT4, NANOG, STELLAR, and GDF3,15, 16 and that these genes are also expressed in PGCs, gonocytes, and spermatogonia.15, 18–21 In vertebrates, DAZL expression has been identified in migrating PGCs; however, expression of DAZL has been found to significantly increase once the PGCs have entered the genital ridges.22, 23 In humans, DAZL expression has been identified in gonocytes of fetal gonads, through spermatogonia, spermatocytes, and spermatids.24 Furthermore, DAZL expression has also been identified in hESCs at both the protein and RNA level, but not in the inner cell mass of human blastocysts.16VASA expression is first identified as PGCs enter the genital ridges and continues to be expressed in gonocytes, spermatogonia, and spermatocytes, but is down-regulated in spermatids.25 In contrast to DAZL, VASA is not expressed in undifferentiated hESCs.16BOULE is used as an adult marker of germ cell formation, as it is not expressed in hESCs, gonocytes, or spermatogonia, but is expressed in the adult testis in late-stage spermatogonia just before meiosis.16, 26 Finally, TEKT1 is only expressed in postmeiotic spermatids of the adult testis.27
By comparing the real-time PCR value of each gene normalized to GAPDH in seminomas relative to normal testis, we found 9 of 9 (100%) seminomas had lower levels of DAZL and VASA expression. Furthermore, BOULE and TEKT1 were undetectable in 9 of 9 (100%) seminomas (Table 2). With respect to genes known to be expressed in hESCs and germ cells, 8 of 9 (90%) seminomas expressed more than 1.5-fold higher levels of GDF3 compared with normal testis (Table 2), whereas OCT4 expression was elevated more than 1.5-fold in 5 of 9 (56%) seminomas and NANOG and STELLAR expression was elevated more than 1.5-fold in 3 of 9 (33%) seminomas (Table 2).
Table 2. Fold Change in Gene Expression Between Normal Testis and Seminoma
N/D: not detected.
Next, we compared the relative expression of the eight different markers within each specimen. In normal testis, DAZL and VASA were expressed at the highest levels, followed by BOULE, TEKT1, and NANOG (Fig. 2A). We also found that OCT4 and STELLAR were expressed at the lowest levels and GDF3 did not register above baseline for the amount of template used in our studies (Fig. 2A). Further examination of these expression profiles in nine seminomas revealed several interesting differences in gene expression patterns (Fig. 2B–J). In particular, DAZL and VASA were no longer the highest-expressed genes in each seminoma sample. Instead, OCT4 or NANOG were consistently the highest-expressed genes in each of the seminoma samples compared with the other genes in this study.
NANOG Protein Is Expressed in hESCs, Adult Testis, and Seminoma
Given that NANOG was one of the most highly expressed genes at the RNA level in each seminoma analyzed, we next explored the expression of NANOG protein. Given that the NANOG antibody used in the current study had not been previously characterized, we first tested the polyclonal antibody using immunofluorescence and Western blot analysis with hESCs as a positive control (Fig. 3A,B). We found that NANOG protein was expressed by cells of the hESC colony, but not of the CF1 feeder layer (Fig. 3A). Analysis of the hESC colony at higher magnification revealed that NANOG is localized to the nuclei of hESCs (arrow) (Fig. 3B). In contrast, the preimmune sera used as a negative control gave no positive signal (Fig. 3C). Next, we performed Western blot analysis on protein extracted from hESCs using both preimmune rabbit control sera and the NANOG immune sera (Fig. 3D). Analysis of the immune sera identified a specific band at 36–37 kDa (arrow), which is the expected size of NANOG. By using immunohistochemistry, we then examined NANOG expression in normal testis to determine its expression patterns within the seminiferous tubules. We found high levels of NANOG in spermatogonia (arrows) and lower levels in spermatocytes during meiosis and spermatids (arrows) (Fig. 3E). No specific staining was observed with negative control preimmune rabbit sera (Fig. 3F). Next, we examined NANOG protein expression in seminoma and observed that NANOG was expressed in the nucleus of the majority of seminoma cells (Fig. 4A–C); preimmune sera gave no specific signal in seminoma (Fig. 4D).
Stem Cell Gene Expression in Nongerm-Cell Carcinomas
Given that the stem cell genes NANOG, GDF3, STELLAR, and OCT4 were all expressed in premeiotic germ cells, and seminoma is reported to arise from transformed germ cells, we were interested in determining if these genes were also elevated in other nongerm-cell carcinomas. We found that normal breast tissue did not express detectable levels of any germ cell or hESC-specific genes (Fig. 5A). However, analysis of Stage 3 breast carcinoma revealed expression of all stem cell genes assayed (NANOG, OCT4, GDF3, and STELLAR) (Fig. 5B). We also analyzed the breast carcinoma cell line MCF7, which is an estrogen-responsive cell line derived from breast carcinoma (Fig. 5C); this cell line also expressed OCT4, GDF3, STELLAR, and NANOG. Figure 5E,F shows representative gels of the germ cell and germ-cell/stem-cell-specific genes in normal testis and seminoma, respectively. To further examine pluripotent gene expression in breast, we stained both normal breast and breast carcinoma sections with NANOG antibody. We did not find any specific staining for NANOG in sections of normal breast (results not shown). In comparison, we clearly observed NANOG protein expression in clusters of cells (Fig. 6), particularly in regions close to the adipocytes (Fig. 6B). NANOG protein was expressed in both the nucleus and cytoplasm of cells in the breast carcinoma samples (Fig. 6C–F).
We hypothesized that NANOG, STELLAR, and GDF3 would be up-regulated in seminomas given that these stem cell genes map to 12p and that all seminomas are reported to have amplifications at 12p being either isochromosomes i(12p) or 12p11.2–12p11.1 and 12p12.2–12p13.3 amplifications.13, 28 Our results demonstrated that NANOG and STELLAR were elevated in several seminoma samples; however, only GDF3 was elevated in almost all seminomas relative to normal testis. GDF3 is a secreted TGF-beta superfamily ligand that interacts with unknown receptors. In humans, GDF3 expression has been documented in human embryonic carcinoma cells (hECCs), hESCs, and fetal gonads; it is not expressed at high levels in other human somatic tissues.15, 25 Although this pattern of expression is intriguing, little is known of GDF3 function, including the identity of the receptors to which it binds.
In this study, we observed the highest expression of NANOG and OCT4 in seminoma compared with all other genes analyzed. Several recent studies have addressed gene expression of NANOG or OCT4 in both CIS and seminomas using gene chip expression profiling.28, 29 In particular, Almstrup et al.29 compared one CIS sample directly to normal testis and identified several genes that were up-regulated in CIS, including NANOG and OCT4. In comparison, Sperger et al.28 compared various pluripotent cell populations including hESCs, seminomas, nonseminomas, and hECCs to normal testis and somatic tumor cell lines. That study determined that OCT4 was highly up-regulated in hECCs, hESCs, and seminomas and that NANOG mRNA expression was highly associated with hECCs, but did not report the levels of NANOG expression in seminomas.28 In our study, elevated expression of NANOG and OCT4 were not identified in all seminomas compared with normal testis; however, these two genes were expressed at the highest levels in all seminomas compared with all the other genes in this analysis. Such variability in gene expression was also found by Almstrup et al.29 during PCR validation of CIS, therefore suggesting that variability in NANOG and OCT4 expression is found not only in the seminoma samples examined in our study, but also in the putative precursor CIS cells. Despite this variability, our data demonstrate that OCT4 and NANOG were consistently the highest-expressed mRNAs compared with all other genes analyzed in our study. Although more samples would reinforce our current data, we expect that this trend will continue in future studies.
With regard to protein, OCT4 expression has previously been examined in human seminomas using multitissue microarray immunohistochemistry.30 In this multitissue microarray study, 4 of 4 seminomas were positive for OCT4. NANOG protein expression has never been reported in seminomas. Therefore, in the current study we examined NANOG protein in both seminomas and hESCs. We found that NANOG protein was expressed exclusively in the nucleus of hESCs and seminomas, suggesting that in both cell types NANOG's main site of activity is in this cellular compartment. Furthermore, it is known that both mouse and human NANOG proteins contain a DNA recognition sequence, implying that NANOG may regulate transcription. However, the exact mechanism of NANOG activity in seminoma remains to be identified.
Although seminomas clearly express hESC and embryonic germ cell markers, we also examined the presence of additional germ cell markers in an attempt to further characterize the genetic identify of the seminoma cell with regard to temporal development of the germ cell lineage. Notably, the adult germ cell genes that mark meiotic and postmeiotic cells, BOULE and TEKT1, were not expressed in seminomas. This lends further genetic support to the finding that seminomas are not composed of meiotic or postmeiotic germ cells. Furthermore, given the complete absence of BOULE and TEKT1 in the seminomas, this also demonstrates that the carcinoma specimens were not contaminated with normal testis. Interestingly, both DAZL and VASA expression were decreased in all seminomas relative to normal testis. Previous work has examined DAZL protein expression in CIS and seminomas and shown that all CIS cells express DAZL,31 but DAZL expression in pure seminomas is rare and focal in nature.31 This is consistent with our findings that DAZL expression is not elevated in seminoma relative to normal testis.
Similar to DAZL expression, VASA expression in the current study was also uniformly low in seminomas. VASA protein expression has previously been examined in CIS and seminoma, being highly expressed in CIS cells, but having relatively weak expression in seminomas.32 These two genes, DAZL and VASA, are classic markers of gonocytes and spermatogonia23,33; therefore, their reduced expression in seminomas would suggest that seminomas are less similar to CIS and more closely resemble a different cell type that is yet to be fully characterized.
In addition to this work on seminomas, we also learned that hESC marker expression is not limited to germ cell tumors. We identified elevated expression levels of OCT4, together with NANOG, STELLAR, and GDF3 mRNA, in both breast carcinoma tissue and the breast carcinoma cell line MCF7. This suggests that breast carcinomas, and possibly other human malignancies, may contain cells reminiscent of embryonic-like stem cells. Furthermore, the absence of the germ cell-specific genes DAZL and VASA in the breast carcinoma would suggest that the OCT4-expressing breast carcinoma cells identified in the current study are distinct from germline stem cells that may have metastasized to the breast. In particular, a previous study using OCT4 protein alone as a marker of germ cell tumor metastasis to the breast argued that OCT4 is a useful marker of germ cell metastasis to extragonadal tissues.34 Our results would argue that OCT4 expression should be analyzed in the context of additional germ cell-specific markers including DAZL and/or VASA to identify germ cell metastatic origin versus an alternate origin of the OCT4-positive cells in this invasive carcinoma.
A model for the development of seminoma and reacquisition of markers of pluripotency in nongerm-cell tumors is illustrated in Figure 7. This model shows that PGCs are specified at a similar time to the somatic lineages from the epiblast of the embryo at the time of implantation. The specified PGCs subsequently retain the expression of many markers associated with pluripotency that are normally expressed by cells of the inner cell mass (ICM)16,18–21 (in red, Fig. 7). Somatic cells (green box, Fig. 7), by comparison, do not express pluripotent genes, including NANOG, OCT4, STELLAR, and GDF3.15, 18 The current hypothesis (pathway A, Fig. 7) suggests that seminoma is derived from CIS, and that CIS is derived from a PGC/gonocyte-type cell that accumulates chromosomal abnormalities.35 Our data supports this hypothesis, particularly through the elevated expression of the ESC- and PGC-expressed genes NANOG and OCT4. However, our data also reveal that in 100% of seminomas analyzed DAZL and VASA mRNA levels are reduced, whereas it has been reported that CIS express abundant levels of these genes.31, 32 This would suggest that in addition to chromosomal differences observed between CIS and seminoma, there is a reduction in germ cell marker expression in favor of a more robust pluripotent stem cell profile. Recently, a new cell type was identified in normal neonatal rodent testis that has hallmarks of embryonic stem cells as well as PGCs.36 This cell type is called a multipotent germline stem cell (mGSC) and, given its hallmark characteristics of pluripotency, it should now also be considered as a potential precursor to either CIS (B) or potentially seminoma directly (C) (Fig. 7). Furthermore, isolation of this cell type is enhanced in genetic backgrounds predisposed to germ cell tumors, suggesting a common link between this cell type and germ cell tumor.36 In addition, this cell type is also associated with elevated levels of NANOG.36 Although our data do not unequivocally support that seminomas arise from the mGSC, they do not refute this alternate pathway either, as currently there is no clear marker that would distinguish the mGSC from the PGC/gonocyte. Finally, our data also show that acquisition of markers at both the mRNA and protein level of pluripotency-associated genes such as NANOG can be identified in breast carcinomas but not normal breast tissue (red cells; Fig. 7). Therefore, acquisition of these pluripotent genes could be a marker of carcinoma progression from somatic tissues.
In conclusion, our results confirm and extend recent work suggesting that seminomas are not genetically like CIS with regard to the expression of the germ cell-specific genes DAZL and VASA. Our results also reveal that seminomas have elevated expression of GDF3 compared with normal testis, together with high expression levels of NANOG and OCT4 compared with additional germ cell and stem cell markers, in particular DAZL. We also identified a similar stem cell program (minus DAZL and VASA) and expression of NANOG protein in breast carcinoma, but not in normal breast tissue. This suggests that either amplification of a resident stem cell population or a reversion of somatic breast cells to a stem cell-like state may be associated with the progression of this and possibly other carcinomas. Certainly, the presence or activation of a stem cell-like program should be further investigated in these and other human malignancies.
The authors thank Michael Abeyta, Karen Chew, and Clarissa Bush for expert technical assistance, Dr. Aleksandra Cvoro for culture of MCF7 cells, and Dr. Jeff Simko for providing seminoma samples. The authors state that they have no competing financial interests.