Endogenous Retrovirus Expression Is Required for Murine Melanoma Tumor Growth In vivo

Tumor development is a multistep process in which both genetic and epigenetic events cooperate for the emergence of a malignant clone. The possibility that endogenous retroviruses promote the expansion of a neoplastic clone by subverting immune surveillance has been proposed, but remained elusive. Here we show that knocking down—by RNA interference—an endogenous retrovirus spontaneously induced in the B16 murine melanoma results in the rejection of the tumor cells in immunocompetent mice, under conditions where control melanoma cells grow into lethal tumors. The knockdown does not modify the transformed phenotype of the cells, as measured both in vitro by a soft agar assay and in vivo by tumor cell proliferation in immunoincompetent (Xirradiated and severe combined immunodeficiency) mice. Tumor rejection can be reverted upon adoptive transfer of regulatory T cells from control melanoma-engrafted mice, as well as upon reexpression of the sole envelope gene of the endogenous retrovirus in the knocked down cells. These results show that endogenous retroviruses can be essential for a regulatory T-cell–mediated subversion of immune surveillance and could be relevant to human tumors where such elements—and especially their envelope gene—are induced. (Cancer Res 2005; 65(7): 2588-91)


Introduction
Tumors are the final outcome of a series of genetic events, which include not only cell transformation per se but also modifications of the interactions between the transformed cells and the host.In particular, it is now recognized that developing tumors arise from the fraction of transformed cells having acquired the capacity to escape immune surveillance (reviewed in refs.[1][2][3].The molecular bases for the acquisition of an immune-resistant phenotype are not fully understood, although several mechanisms have been described, including down-regulation of MHC class I antigens (2) and up-regulation of apoptosis-inhibiting factors (3).Along this line, we have previously shown that introduction of an expression vector for the envelope (env) gene of an infectious murine retrovirus can induce escape from immune surveillance, in cells otherwise rejected after engraftment into immunocompetent mice (4).Similar effects were observed with the env gene of a human endogenous retrovirus (ERV; ref. 5).ERVs, which occupy a large fraction of mammalian genomes, are the genomic traces of ancient retroviral infections that reached the germ line, with the integrated proviruses then being transmitted in a Mendelian manner (reviewed in refs.6, 7).Most of these sequences have accumulated point mutations and deletions, but there are still hundreds of fulllength proviral elements in the human and mouse genomes, with some of them still containing coding genes responsible, for instance, for the observation of viral-like particles in some tumors (6,8).Accordingly, we hypothesized that expression of endogenous retroviruses could be involved in tumor progression in vivo, by subverting the immune response.In this report, we used a spontaneous murine melanoma tumor cell line and show that an ERV, the melanoma-associated retrovirus (MelARV), whose expression is spontaneously induced in melanoma of C57Bl/6 origin (9), is necessary for a regulatory T-cell-mediated immune escape of the tumor cells in vivo.

Materials and Methods
Mice and cell lines.C57BL/6 and severe combined immunodeficiency mice, 8 to 12 weeks old, were obtained from Janvier (Le Genest-St-Isle, France).B16 (murine melanoma cell line of C57BL/6 origin, European Collection of Cell Cultures, Salisbury, United Kingdom) and 293T (human embryonic kidney cells, American Type Culture Collection, Manassas, VA) were maintained in DMEM supplemented with 10% heat-inactivated FCS and antibiotics.
Constructions.We constructed plncxH1 expression vectors derived from the plncx (10) and the pSUPER (11) vectors to generate short transcripts directed against MelARV (targeted to the genomic transcript within the gag sequence; nucleotide positions 1220-1238 from the start codon), or against the green fluorescent protein transcript (nucleotide positions 215-233 from the start codon) as a control.They were obtained by first inserting annealed 64-mer oligonucleotides (sequences in Fig. 1) into pSUPER opened at the BglII and HindIII sites, followed by introduction of the BamHI-HindIII fragment from these constructs into plncx opened at the corresponding sites.The expression vector for the MelARV envelope (pDFG MelARVenv) and the control (pDFG none) were constructed by introducing (or not) a reverse transcription-PCR product, generated from the MelARV viral RNA using an AgeI-containing primer at the envelope 5V-end and a XhoI-containing primer at the envelope 3V-end, into a hygromycincontaining pDFG vector (4) opened at the same sites.
Establishment of ERV KD B16 tumor cells.293T cells (7.5 Â 10 5 ) were cotransfected with the plncxH1 vector (1.75 Ag) and expression vectors for the MLV proteins (0.55 Ag for the amphotropic MLV envelope vector and 1.75 Ag for the MLV gag and pol vector; see ref. 12).Thirty-six hours after transfection, viral supernatants were collected for infection of the B16 tumor cells (2.5 mL supernatant for 5 Â 10 5 cells, with 8 Ag/mL polybrene).Cells were maintained in selective medium (1 mg/mL neomycin) for 3 weeks.In some experiments, the pDFG MelARVenv expression vector (or control pDFG none) was additionally introduced into the cells using the same protocol and infected cells were selected with 300 units/mL hygromycin.
Expression of MelARV proteins.Analysis of MelARV expression was done by Western blot analyses.The supernatants of 10 7 cells were collected, centrifuged for 10 minutes at 100 Â g, filtered, and concentrated by ultracentrifugation in a SW41 Beckman (Fullerton, CA) rotor (150,000 Â g, 1 hour, 4jC).Pellets were resuspended in lysis buffer, submitted to SDS-PAGE, blotted, and revealed with an anti-Env monoclonal antibody ( 13) and an anti-Gag goat serum (Viromed Biosafety Labs, Minnetonka, MN).
In vitro transformation assay.Both control and ERV KD -B16 cells were plated in soft agar to determine the efficiency of anchorage-independent growth.Cells (2 Â 10 3 or 2 Â 10 4 ) were plated in 5 mL 0.33% agar in DMEM with 10% fetal bovine serum overlaid onto a solid layer of 0.5% agar in DMEM supplemented with 10% fetal bovine serum.The culture was maintained for 4 weeks; the colonies were stained with INT solution (Sigma-Aldrich, St. Louis, MO) and then counted.
Tumor progression in vivo.For in vivo assays, tumor cells were washed thrice with PBS, scrapped without trypsination, and s.c.inoculated in the shaved area of the right flank of the mice.Tumor establishment was determined by palpation and tumor area was determined by measuring perpendicular tumor diameters.
CD4 + + + CD25 + + + T-cell purification and adoptive transfer in syngenic C57BL/6 mice.CD4 + CD25 + cells were freshly isolated from spleens of C57BL/6 mice engrafted with 2 Â 10 5 B16 cells 17 days before.Cells were purified by a two-step procedure of negative and positive selections, using MACS magnetic beads (mouse regulatory T-cell isolation kit, Miltenyi Biotech, Auburn, CA), according to the manufacturer's instructions.Fifty thousand purified lymphocytes were transferred i.v.into naive C57BL/6 mice.Recipient mice were challenged the same day with 2 Â 10 5 control-or ERV KD -B16 cells in the right flank.

Results and Discussion
Knocking down endogenous retrovirus expression does not modify the transformed phenotype of B16 melanoma cells.We used an RNA interference approach based on stable vectors producing short double-stranded RNA directed against the viral genome of the MelARV element and the irrelevant gfp gene as a control.The rationale of the procedure and the structure of the plasmids used are illustrated in Fig. 1A and B. Figure 1C clearly shows that the ERV-specific double-stranded RNA vector almost completely abolished ERV expression in the transduced B16 cells (ERV KD B16 cells) with a >10-fold reduction in the amount of both the Env and Gag viral proteins compared with the control transduced cells (control B16 cells).As a next step, the transformed phenotype of the ERV KD and control B16 cells was assayed both in vitro and in vivo.In vitro, we measured the anchorageindependent growth rate after plating in semisolid media (soft agar assay).As illustrated in Fig. 2A, the ERV KD B16 cell line gave rise to a similar number of colonies as the control B16 cells.In vivo, we analyzed the growth rate of the two cell populations after engrafting into X-irradiated or severe combined immunodeficiency mice.As illustrated in Fig. 2B, both cell populations have a transformed phenotype with similar growth rates.Altogether, these results show that knocking down the MelARV endogenous retrovirus has no effect on the transformed state of the melanoma cells.
Knocking down endogenous retrovirus expression inhibits B16 tumor cell growth in vivo and increases survival of immunocompetent hosts.To investigate whether tumor cells may overwhelm the antitumor response in vivo through an ERVdependent mechanism, we explored the impact of the knocking down of MelARV on tumor progression by injecting C57BL/6 immunocompetent mice with the control and the ERV KD B16 cells.As illustrated in Fig. 3A, growth of control B16 cells, as expected, led to large tumors in most of the animals, whereas the ERV KD B16 cells yielded tumors of a limited size and in only a small number of engrafted mice.The difference in tumor cell growth is also clearly substantiated by the extent of animal survival (Fig. 3B): As soon as day 70, 90% of the mice engrafted with the control B16 cells had been killed by their tumor, whereas 80% of mice engrafted with ERV KD B16 cells were alive and tumor-free (and still so at day 130).
Cancer Res 2005; 65: (7).April 1, 2005 Figure 1.Knockdown procedure and rationale of the assay.A, to knock down ERV expression, we constructed a plncx-derived vector making use of the pSUPER vector to generate, under control of the H1-RNA promoter, short double-stranded transcripts for RNA interference.B16 cells were transduced with these expression vectors, submitted to G418 selection, and the resulting ERV KD and control B16 cells were injected s.c.into the flank of the mice, whose tumor growth was monitored.B, predicted structure of the double-stranded RNA generated by the ERV and control (gfp ) vectors; numbers, nucleotide positions within the respective targeted sequences (see Materials and Methods).C, Western blot analysis of Gag and Env expression in the supernatant of ERV-knocked down (ERV KD ) and control cells.
In an attempt to identify the MelARV genes involved in the observed effects, we introduced back into the ERV KD B16 cells an expression vector (lacking the double-stranded RNA-targeted sequence) for the sole MelARV env gene.The resulting doubletransduced ERV KD + env (or control) B16 cells were then engrafted into C57BL/6 mice.As illustrated in Fig. 3C, this resulted in partial reversion of the knockdown effect, with already 50% of the mice engrafted with the Env-expressing cells dead by day 70.This reversion indicates that the env gene is, at least in part, responsible for tumor immune escape.The partial effect of the reversion is most likely explained by the lower expression (Supplementary Fig. S1) of the Env protein when expressed by the exogenous vector.
Adoptive transfer of CD4 + + + CD25 + + + regulatory T lymphocytes restores tumor progression of ERV KD B16 cells.It has been reported that B16 tumor growth in vivo requires the presence of regulatory T cells (14,15).On the other hand, murine leukemia viruses closely related to MelARV have also been shown to induce regulatory T cells (16,17).One can, therefore, hypothesize that ERV expression by the B16 melanoma is directly responsible for regulatory T cell induction, thus resulting in tumor progression.
To test this issue, we attempted to complement the reduced tumor progression of the ERV KD B16 cells by adoptive transfer of the CD4 + CD25 + regulatory T cells isolated from C57BL/6 mice engrafted with wild-type B16 cells.As shown in Fig. 4, the transfer of these CD4 + CD25 + cells had no effect on tumor progression of engrafted control B16 cells (Fig. 4A, left, and B), but enhanced ERV KD B16 tumor cell growth up to levels similar to those of the control B16 cells (Fig. 4A, right, and B).These results, therefore, strongly suggest that regulatory T cells are the necessary intermediates of the ERV-mediated effect.
Conclusion and perspectives.The present data show that tumors are able to overwhelm the immune system by expressing the envelope of an ERV.Blocking ERV expression resulted in enhanced tumor rejection, and expression of the sole ERV envelope protein partially restored the initial extent of tumor progression.We also show that regulatory T cells are sufficient to allow a tumor www.aacrjournals.org Figure 2. Knocked down cells have conserved a transformed phenotype.A, in vitro analysis of the transformed phenotype using soft agar assay.ERV KD and control B16 cells (2 Â 10 3 or 2 Â 10 4 ) were plated onto a semisolid layer for 4 weeks, and then colonies were counted.B, assay for the transformed phenotype in vivo using immunoincompetent mice.ERV KD and control B16 cells (2 Â 10 5 ) were injected s.c.into the flank of either X-irradiated (5 Gy) C57Bl/6 or severe combined immunodeficiency mice (two to five independent experiments with five mice per group) and tumor growth was determined by measuring tumor area.  1) and related to immunosurveillance as initially described by Burnet and Thomas (18).Indeed, we illustrate that tumor development not only involves mutations within classic proto-oncogenes or antioncogenes, but also involves active processes targeted to unanticipated genes that have no effect on the transformation state per se, such as the presently identified MelARV gene.In combination with classic chemotherapy or radiotherapy, targeting such genes by any available means, including the injection of unvectorized small interfering RNA, may contribute to tumor regression by restoring the natural ability of the immune system to reject tumor cells and, therefore, contribute to cancer treatment.Along this line, it is of interest that a first series of experiments using synthetic small interfering RNA targeted to MelARV, and injected i.p. 12 days after engraftment of B16 cells into immunocompetent mice, actually resulted in a one-third inhibition of tumor growth compared with mice injected with control small interfering RNA and, as illustrated in Supplementary Fig. S2, in a moderate, but reproducible, increase in survival delay.Finally, it is noteworthy that, in humans, the expression of ERV env genes, mainly restricted to placenta and testis in normal tissues, can be observed in several tumor types, such as seminomas and melanomas (8,19,20).Moreover, several of the fully coding env genes that can be identified in the human genome are immunosuppressive in vivo in a mouse model assay (5), and thus might act as the MelARV env gene to promote tumor escape.Future work should now be aimed at the identification of a possible role of such genes in human tumorigenesis.

Figure 3 .
Figure 3. Inhibition of tumor cell growth and increased mouse survival upon ERV knockdown.A, tumor cell growth of control and ERV KD B16 cells engrafted into immunocompetent C57Bl/6 mice (22 mice per group; same experimental conditions as in Fig. 2B).B, percentage of survivors among the control and ERV KD B16 cells engrafted mice (10 mice per group).C, percentage of survivors (10 mice per group) among MelARV env -and control-transduced ERV KD B16 cells.
28, 2017.© 2005 American Association for Cancer cancerres.aacrjournals.orgDownloaded from cell line devoid of ERV expression to progress into a tumor.More generally, the results also suggest that tumor cells are likely subject to immunoediting, a concept developed by Dunn et al. (