A Novel Gene from the Human Endogenous Retrovirus K Expressed in Transformed Cells¹

Retroviral proviruses were inserted into the germ line of human ancestors 40 million years ago (1) and are the origin of the present HERV³ sequences. Data provided by the Human Genome Sequencing Project show that up to 8% of the human genome consists of retroviral sequences (2), most of which are defective because of the accumulation of mutations. The only HERV family still encoding all essential retroviral proteins is HERV-K (HML-2). Although most of the ∼30 copies are mutated, this family contains the most intact HERV discovered thus far, the HERV-K (HML-2.HOM) provirus localized on chromosome 7 (3). Recently we provided evidence for a link between HERV-K expression and the development of GCTs in humans (4). Individuals with GCTs produce serum antibodies against the viral Gag and Env proteins, and the gag and env genes are expressed in these tumors, but not in other tumor types (5). When studying the transforming potential of HERV-K sequences, we found that cORF, a protein encoded by the central open reading frame within the env-gene (6), can support tumor growth in nude mice and associates with PLZF (7). PLZF has been documented to be critical for spermatogenesis in mice. Furthermore, abnormal spermatogenesis is thought to predispose humans for the development of GCT. In this context, it was interesting to ask whether other splice variants of the HERV-K env sequences may be associated with tumors. Here we describe the identification of a novel HERV-K gene, termed np9, that is transcribed exclusively from the HERV-K type 1 provirus because of a type 1-specific splice donor site. In contrast, cORF is transcribed exclusively from type 2 proviruses and includes a 292-bp sequence that is deleted in type 1 (8). In this report, we show that np9 is significantly expressed in tumor tissues and transformed cell lines, but not in normal cells.

For RT-PCR, total RNA from tissues or cell lines was extracted with RNA clean (Hybaids-AGS, Heidelberg, Germany), according to the manufacturer’s protocol. To remove residual DNA, a DNase I treatment was carried out. The RT with Superscript II (Life Technologies, Inc.) was performed on 5 μg RNA and 25 pmol of random primers. To amplify env reading-frame transcripts, the following primers were used: A, 5′-ATGAACCCATCGGAGATGCAA-3′; and B, 5′-ACAGAATCTCAAGGCAGAAG-3′ (underlined in Fig. 1 A). For gag fragments, the following primers were used: C, 5′-GGCCATCAGAGTCTAAACCACG-3′; and D, 5′-GCAGCCCTATTTCTTCGGACC-3′. As a control, a 192-bp fragment from the GAPDH cDNA was amplified with the following primers: E, 5′-AGTCCAGTGAGCTTCCCGTTCAGCA-3′; and F, 5′-TGGTATCGTGGAAGGACTCATGAC-3′. PCR cycling conditions were as follows: 3 min at 94°C; 30 cycles of 50 s at 94°C, 50 s at 58°C, and 3 min at 72°C; and 10 min at 72°C.

The plasmid pGEM-np9 was constructed with the pGEM-T vector kit (Promega) for Taq polymerase amplified PCR products. The np9 fragment was amplified with primers A and B and ligated into the pGEM-T vector. Plasmids pCEP4-np9, pLRNL-np9, and pEGFP-np9 were constructed by amplification of np9 with the following primers: G, 5′-CGCGCggatccATGAACCCATCGGAGATGCAA-3′; and H, 5′-CGCGCggatccAACAGAATCTCAAGGCAGAAG-3′. Primers G and H were derived from primers A and B (see Fig. 1 A) to provide BamH1 restriction sites, represented by lowercase letters in primers G and H. The amplified fragment was digested with BamH1 and ligated into the BamH1 sites of pCEP4 (Invitrogen), pLRNL (7), and pEGFP-C1 (Clontech), respectively. For the generation of the pATH-np9–2e construct, the second exon of np9 was amplified from pCEP4-np9 with the following primers: I, 5′-CGCGCggatccGAGATGTCTGCAGGTGTAC-3′; and K, 5′-AACAggatccCAAGGCAGAAGAATTTTTC-3′. The second exon of np9 was then ligated into the BamH1 site of the pATH-10 vector (9). The cORF plasmids pCEP4-cORF and pEGFP-wtcORF have been described previously (7, 10).

All cells were maintained at 37°C and a 5% CO₂ atmosphere. The generation of stable cORF expressing Raji cell line (RajiwtcORF) has been described (7). The stable np9 expressing and a matching vector-only expressing cell line (Raji-np9 and Raji-pCEP-4, respectively) were developed by transfecting Raji cells with the pCEP-4-np9 or pCEP-4 vector and by subsequent selection with 300 μg/ml hygromycin B. Cell lines 293gp-np9 and 293gp-pLRNL were generated by stable transfection with plasmids pLRNL-np9 or pLRNL, followed by selection with 400 μg/ml G418. The production of the retroviral vectors and infection of Rat-1 cells was performed as described elsewhere (7). Rat-1-np9 and Rat-1-pLRNL express the np9 transgene and the pLRNL empty vector, respectively.

To raise polyclonal antibodies, the second exon of np9 was bacterially expressed from the pATH-np9–2e construct. Rabbits were immunized with the protein as described previously (11). Raji-np9 cells, Raji-wtcORF cells, and untransfected Raji cells were treated with 20 μg/ml TPA for 48 h to induce the expression of the recombinant proteins. Western blot analysis was performed with the cellular extracts following standard procedures (7). Immunofluorescence studies were carried out as described elsewhere (12).

293gp cells were grown to ∼20% density on glass cover slips and were then incubated with a calcium phosphate coprecipitate that included 2 μg of either the empty vector or the pEGFP-np9 or the pEGFP-wtcORF vector construct. Cells were fixated in paraformaldehyde (4% in PBS) at 24 h after transfection and the DNA was stained with 200 ng/ml DAPI in methanol to visualize nuclei. Intracellular localization of EGFP fusion proteins was studied by fluorescence microscopy.

The plasmid pGEM-np9 was sequenced with the SequiTherm EXCEL II kit (Biozym) with T7 and SP6 primers from the pGEM-T vector kit (Promega) according to the manufacture’s protocol. GenBank searches were performed with the National Center for Biotechnology Information program for Nucleotide-GenBank research,⁴ the alignment of np9 and cORF was performed by the BCM Search Launcher. For further sequence analysis, the research programs PSORT II from Swiss Prot and Scan Prosite were used.

Tumor biopsies were collected from four institutions (Women and Children’s Clinic University of Saarland Medical School, Homburg; the Gerhard-DomagkInstitute of Pathology, University of Muenster, Muenster; the Urology Clinic of the Philipps-University Marburg, Marburg; and the Institute for Pathology of the Clinic Bamberg, Bamberg, Germany).

During further investigations on the role of HERV-K cORF in GCT development, the RNA expression patterns of the env gene was analyzed by RT-PCR in tumor biopsy samples and in transformed cell lines. Apart from the 473-bp fragment specific for the cORF transcript and a variety of PCR products not consistently detected in all samples, we found a 256-bp fragment in three of three seminoma biopsies, of which one is documented in Fig. 2. This prompted us to study this variant in more detail.

GenBank searches exhibited 100% homology of the 256-bp fragment with the HERV-K env sequence from HERV-K101 published by Barbulescu et al. (1). This fragment, shown in Fig. 1,A, is used for all following studies and is designated np9. Further RT-PCR studies revealed the presence of two other variants of the np9 sequence that differ from np9 in up to four nt positions, giving rise to three amino acid changes (Fig. 1,A). The variants show 100% homology to proviruses located on chromosome 3q13 [HERV-KII (Chr. 3q13); see Ref 13] and 22q11 [HERV-K (Chr. 22q11)], respectively (Fig. 1,A). Importantly, all np9 genes originate from HERV-K sequences of type 1, which are not able to express intact cORF because of the type 1-specific deletion of the 292 bp (see Ref. 8; Fig. 1,B). The expression of np9 from provirus type 1 is the result of the presence of an alternative splice donor site, produced through the replacement at nt positions 6495 and 6496 of HERV-K101 of A and G, present in type 2 viruses, for a T and A in type 1 (see Ref. 14; Fig. 1,C). Further differences between cORF and the 256 bp np9 sequences are outlined in Fig. 1,A. Translation reveals that the first exon of np9 encodes the N-terminal on aa 15 of the 87 aa that constitute the first exon of cORF. The second exon of np9, beginning at nt position 8118 in the HERV-K101 sequence, contains 178 bp of coding sequence. The COOH-terminal 59 aa of Np9 are derived from the third (non-env, non-cORF) reading frame. The deduced Np9 protein has thus a predicted size of 74 aa and a predicted molecular mass of 8.7 kDa. By motif analysis, three putative NLSs were identified (see Fig. 1,A, light boxes). The putative NLSs I and III were identified by the PSORT II search program from Swiss Prot; NLS II was predicted on the basis of a minimal consensus sequence, defined by Garcia-Bustos et al. (15). The presence of these motifs suggested a nuclear localization of the HERV-K protein Np9. Furthermore, a putative casein kinase II-phosphorylation site was identified between nt positions 8460 and 8475 by the Scan Prosite research program (Fig. 1 A).

To show expression of the recombinant Np9 protein in a human cell system, a stable np9 expressing Raji-cell-line (Raji-np9) with the pCEP4-np9 construct was generated. Raji cells were chosen because they express recombinant HERV-K proteins at high levels (see Ref. 7 and unpublished observations). Furthermore, the levels of transgene expression from pCEP4 can be enhanced by TPA (16). Western blot analysis showed expression of a protein of ∼9 kDa in cell extracts from Raji-np9 cells (Fig. 3,A). Furthermore, we were able to detect the recombinant Np9 protein in TPA-treated Raji-np9 cells by indirect immunofluorescence (Fig. 3,B). No signal was detected in Raji-cORF cells transfected to stably produce recombinant cORF (see Ref. 7; Fig. 3 A), or in Raji-np9 without TPA-induction (data not shown), indicating that Np9 is normally expressed at very low levels.

We then wanted to study the subcellular localization of recombinant Np9. In a first set of experiments, a hybrid protein consisting of full-length Np9 and the EGFP fused to the NH₂ terminus of Np9 was used in transient transfections of human 293gp cells. The fact that EGFP-Np9 is located predominantly in the nucleus, and that it preferentially stains structures of unknown identity within the nucleus, is shown in Fig. 3,C. A similar nuclear-staining pattern is generated by EGFP-cORF. In contrast, EGFP alone stains the cell uniformly. Thus, EGFP-Np9 is a nuclear protein. The nuclear localization of recombinant Np9 was confirmed by immunofluorescence on Raji-np9 cells with the polyclonal anti-Np9 antibody (Fig. 3 B).

Stably control-transfected 293gp or Raji cells expressing either Np9 or control vector were subjected to growth analysis. There was no significant difference in the population doubling time between the cell lines expressing Np9 or the control plasmid over a time course of 5 days (Fig. 4,A). Furthermore, to study whether the presence of Np9 has a biological relevance in proliferation and survival, standard colony formation assays were carried out with infected Rat-1 cells. Likewise, no differences were observed between the colony formation capability of cells expressing Np9 or the empty vector (Fig. 4 B), suggesting that Np9 expression confers no simple growth and/or survival advantage on cells in vitro.

To study the expression of np9 in human and primate cells, a panel of established cell lines, tumor biopsies, and lymphocytes from leukemic and normal individuals, as well as normal, nontransformed cells were screened for np9 mRNA expression by RT-PCR analysis. Examples for the obtained results are shown in Fig. 2. Transcripts from the gag and cORF reading frames were analyzed in parallel (Fig. 1,A). RNA quality and amplification efficiency were monitored by amplification of a fragment from the housekeeping gene GAPDH. The HERV-K-negative cell line B95–8 (a B-cell line) derived from a New World monkey served as negative control. The results of these studies are summarized in Table 1. Np9 expressed from HERV-K type 1 proviruses was detected in >90% of all transformed cell lines (n = 15). Likewise, gag was produced in >90% of the cell lines. In contrast, cORF expressed from HERV-K type 2 was detected in only 13% of the cell lines; it was present in only one teratocarcinoma and one mammary carcinoma line. Remarkable differences between the presence of np9 and cORF were observed in primary tumor samples. Although np9 could be detected in >45% of the samples, the presence of the transcript was not equally distributed among the tumor types. The transcript was most frequently found in mammary carcinomas (52%; n = 21), whereas it was less frequently present in germ-cell tumors (37%; n = 8) and leukemia blood lymphocytes (33%; n = 6). Both ovarian carcinomas tested negative. Of further interest might be the relative frequencies of cORF and np9 in the tumor biopsies. Although similar numbers of germ-cell tumors express cORF and np9, mammary carcinomas expressing np9 are more frequent than mammary carcinomas expressing cORF by a factor of five to six. This is in accord with findings by Etkind et al. (17), demonstrating a 10-fold increase in the quantity of full-length transcripts from provirus type 1 relative to type 2 in the breast carcinoma cell line T47D, and by Wang-Johanning et al. (18), detecting exclusively type 1-specific env transcripts in breast carcinomas. Grading of the mammary carcinomas according to the Tumor-Node-Metastasis classification (Union International Contre Cancer) failed to show any correlation between tumor grade and np9 expression (data not shown). Finally and most importantly, none of the tested healthy tissues (n = 14; lymphocytes and fibroblasts, and samples from the gut, placenta, and stomach) expressed np9 at detectable quantities. The normal diploid human fibroblasts KH5109 and nontransformed derivatives designed to express the dominant-negative mutant 175H of the p53 tumor suppressor also failed to test positive, as did immortal nontransformed human 041 fibroblasts lacking wild-type p53 and their 175H-transduced derivatives. Like np9, cORF was not detectably produced in nontransformed cells and tissues. In stark contrast, the gag transcript was detected not only in the transformed cells and tumor samples but also in >80% of the normal, nontransformed cells. Thus, np9 expression is closely correlated with transformation, more closely than cORF expression, and might serve as a marker for transformation.

HERV sequences are frequently transcribed in several normal tissues (19). Recent work has indicated that syncytin, a product of the env gene of the defective HERV-W, may have an important function in human placental morphogenesis through the mediation of the fusion of the placental cytotrophoblasts (20, 21). Several cellular genes are specifically regulated by HERV-LTRs, although thus far no LTR of the HERV-K class could be linked to the transcription initiation of cellular sequences. One reason for the conspicuous absence of HERV-K-regulated cellular genes may be that HERV-K-LTRs seem both to be regulated highly specifically by cell-type and to be dependent on the state of differentiation (22).

Besides affecting normal cell physiology, endogenous retroviral genes may contribute to tumorigenesis. Early hints pointing in this direction came from the observation that certain infectious retroviral proteins and related polypeptides may be immunosuppressive and anti-inflammatory. For example, the retroviral Env p15E protein can inhibit interleukin 1-mediated signal transduction and protein kinase C activity (23, 24) and was documented to be able to negatively affect monocyte chemotactic responses and lymphocyte blastogenesis in human cells in vitro (25). Env p15E is homologous to the transmembrane region of the Env protein against which serum antibodies are produced highly specifically in patients with germ-cell tumors. The env region of HERV-K (HLM-2.HOM) may thus positively influence the maintenance and progression of such tumors via the inhibition of an effective immune response. HERV-K message, proteins and antibodies directed against them are very frequently observed in patients with GCT (4, 12) and occasionally in breast cancer patients and in HIV- or cytomegalovirus-infected individuals (26), but not or only very rarely in healthy individuals.

Further hints at the implication of endogenous retroviral genes and proteins in tumorigenesis were found in the murine system: MCA205 tumor cells proved incompetent in producing neoplasias in immunocompetent mice but escaped immune rejection and produced tumors in such mice when designed to express either the full-length Env protein or the Env transmembrane region from the Moloney murine leukemia virus (27). Other recent work has shown that CORF, the 15-kDa protein translated from the COOH-terminal open reading frame within the env gene of HERV-K, can act as a tumor-susceptibility factor, support transformation of immortal cells in immunocompromised mice and, moreover, can physically interact with PLZF (7). PLZF has been implicated in leukemogenesis in humans and, important in this context, also in spermatogenesis in mice. It is precisely the disruption of spermatogenesis of the sort produced by the functional impairment of PLZF that may prepare the ground for the precursor lesions associated with an increased risk for testicular germ-cell cancers (28). CORF and Env share the N-terminal 87 aa residues, and both share their N-terminal 15 aa with the novel HERV-K protein Np9 reported here. However, the domain in CORF that makes contact with PLZF (aa 21–87) is not present in Np9. It is thus unlikely that Np9 interacts with PLZF.

Although it has been reported that breast carcinomas contain exclusively HERV-K type 1-specific env transcripts (18), we were not able to find sequence motifs in the LTRs that were specific for type 1 or type 2. However, sequence differences in LTRs have indeed been linked to tissue specificity of the expression of HERV-K (13). They could not, however, be associated with a particular type of provirus. This suggests that sequences in the proximity of individual LTRs, and thus the position of LTRs in the genome, play an important role for the tissue-specific expression of the provirus types. Such sequences may constitute regulatory motifs for transcription factors or methylation signals (29, 30). The sensitive regulation of the LTRs may also explain why HERV-K viral transcripts are detectable with PCR methods, whereas, in contrast, the respective proteins including the Np9 protein described here are difficult to detect. As of yet, we were not able to detect native Np9 protein in any cell type, neither by Western blotting nor by immunofluorescence or immunoprecipitation; however, even the native Env and cORF proteins have thus far been detected in only one cell type (6, 31). It is furthermore conceivable that posttranscriptional regulatory mechanisms exist that influence the HERV-K protein levels. Whether Np9 has a function in transformed mammary epithelial cells is the subject of further investigations. At present, there are no known sequence motifs in Np9, apart from the nuclear localization signals and CKII phosphorylation site, that would allow the identification of putative protein domains and speculations about the function of this protein.

We are grateful to H. Herbst, M. Holländer, T. Kälble, and M. Seitz for providing tumor material and to J. Mayer for critically reading the manuscript.