Characterization of the Major Histocompatibility Complex Class I Deficiencies in B16 Melanoma Cells¹

B16 melanoma cells have deficient MHC class I surface expression and demonstrate weak immunogenicity. Despite the same origin, several B16 melanoma sublines with different metastatic potentials and behaviors have been established (1, 2). In vivostudies demonstrated that immunization of C57BL/6 mice with B16 melanoma cells generated humoral rather than T-cell responses, which resulted in the development of numerous mAbs.³These mAbs have been effectively used in the eradication of melanoma,lung, and liver metastases attributable to antibody-dependent,cell-mediated cytotoxicity (3, 4). In addition, H-2 gene transfer enhances the immunogenic phenotype of B16 melanoma cells and causes their rejection in immunocompetent hosts (5, 6), which is associated with a complete loss of melanoma-associated antigen expression, reduced substrate adherence,the inhibition of melanogenesis, and the loss of metastatic capacity(5, 6, 7).

Although B16 melanoma cells fail to induce strong T cell-mediated immune responses, some B16 melanoma antigen-specific T cells have also been described (8, 9). Adoptively transferred surrogate marker OVA-specific CD8⁺ T cell populations in mice bearing established OVA-transfected B16 melanoma lung metastases mediate a reduction of tumor growth and, subsequently, a prolonged survival (10). In addition, the poorly immunogenic phenotype of B16 cells could be altered by the fusion of dendritic cells with syngeneic B16 melanoma cells (11).

Despite these reports, the use of the H-2-deficient B16 melanoma model as a prototype for T cell-based immunotherapies independent of cross-priming has to be reconsidered (12). Abnormalities of MHC class I surface antigens are often associated with an immune escape of tumor cells. The reduction or loss of MHC class I surface expression in human and murine tumors of distinct histologies could be attributable to structural alterations and/or dysregulation of various components of the MHC class I APM (13, 14, 15, 16). The complexity of the MHC class I antigen pathway has been well defined in the last decade. The pathway includes four major components:(a) the multicatalytic proteasome, in particular the LMPs 2,7, and 10, and its PAs, PA28α and -β; (b) TAP;(c) numerous chaperones; and (d) the MHC class I molecules (17, 18, 19, 20, 21, 22). Deficient expression of LMP, PA28, and TAP alters both the quality and/or quantity of the peptide repertoire presented in the context of MHC class I molecules (19). In addition, down-regulated or a lack of tapasin expression also affects MHC class I expression (23, 24).

The underlying mechanism of the impaired MHC class I surface expression in B16 melanoma cells has not yet been identified. We now present evidence that the deficient H-2 expression of B16 melanoma cells is attributable to a coordinate suppression of multiple components of the MHC class I APM. These defects could be corrected by IFN-γadministration, which transcriptionally induces the expression of various APM components, thereby also enhancing MHC class I surface expression in B16 cells. Thus, our data have important implications for the use of the B16 melanoma model in immunotherapeutical studies and the interpretation of their results as well as for the identification of the underlying mechanisms of the impaired expression of the antigen-processing genes.

Different B16 melanoma sublines with a distinct metastatic potential were established from C57BL/6 melanoma. B16F1 exhibiting a low metastatic potential was established from metastatic foci in the lung of i.v. injected B16 cells (2). In contrast, B16F10 cells,established by 10 successive selections for lung metastases after i.v. injection, are highly metastatic. The two B16 subclones, kindly provided by I. John Fidler (MD Anderson Cancer Center, Houston,TX), RMA cells, and the respective TAP-deficient RMA-S cells,both derived from a Rauscher virus-induced T-cell lymphoma, were grown in RPMI 1640 (Seromed, Berlin, Germany) containing 10% (v/v)FCS, 2 mm l-glutamine, penicillin (100 units/ml), and streptomycin (100 μg/ml; Refs. 1 and25). The murine renal cell carcinoma cell line Renca was maintained in Dulbecco’s modified medium supplemented with 10% FCS, 2 mm glutamine, and respective antibiotics. For IFN-γ stimulation, tumor cells were incubated in the presence of 20 ng/ml murine recombinant IFN-γ (Roche Diagnostics, Mannheim, Germany)for 24–48 h at 37°C (26).

The mAbs used in this study were antimouse H-2K^bD^b (Cedarlane, Hornby,Canada) and the FITC-conjugated goat antimouse immunoglobulin(Beckman/Coulter, Krefeld, Germany) as the secondary mAb. For staining of MHC class I antigens, cells were incubated with the primary mAb for 30 min on ice, washed twice in PBS and incubated further with a FITC-conjugated goat antimouse immunoglobulin (Coulter/Beckman,Krefeld, Germany) as a secondary mAb for an additional 30 min. After washing in PBS/1% heat-inactivated FCS, cells were analyzed on a flow cytometer (Coulter Epics XL MCL; Beckman/Coulter, Krefeld,Germany).

Peptide-binding assays were performed using the H-2K^b restricted peptide SIINFEKL (OVA pos 257–264) and the human HLA-A2-restricted nonsense peptide (HIV pos 476–484), as described recently by Wölfel et al.(27). Briefly, 3 × 10⁵ cells/well were incubated for 16 h in serum-free RPMI 1640 (Biochrom, Berlin, Germany) in the presence or absence of the respective peptides (100 μm,dissolved in PBS containing 5% DMSO) with or without 2.5 μg/ml humanβ ₂-microglobulin (Sigma, Deisenhofen,Germany), before MHC class I surface expression was determined by flow cytometry.

The TAP1 and TAP2 expression vectors⁴carrying the human TAP1 or TAP2 cDNA, respectively, under the control of the cytomegalo virus-promoter and the neo^R gene as a selection marker, as well as the control vector P46 carrying the neo^R gene alone, were used for stable gene transfers. Transfections were performed with 2.5 × 10⁶ cells in a volume of 500 μl of MEM containing 1% heat-inactivated FCS and 10 μg of linearized plasmid DNA using electroporation (450V, 1200 μF, 2 μs; EPI 2500; Fischer, Heidelberg, Germany). After pulsing, the cells were plated in complete medium on Petri dishes and 48 h later,neo^R cells were selected in complete RPMI 1640 containing 600 μg/ml G418 (Seromed, Berlin, Germany). Four weeks later, neo^R cells were analyzed for the expression of the transgene as well as for MHC class I surface antigens.

All oligonucleotides used for PCR amplifications were purchased from Biometra (Göttingen, Germany) and are listed in Table 1. For conventional RT-PCR analysis, reverse transcription was performed using 1 μg of total cellular RNA extracted by guanidinium isothiocyanate/cesium chloride preparation (28) in 22-μl reaction volume in the presence of a hexanucleotide mixture (Roche Diagnostics, Mannheim, Germany) and 10 units of Moloney leukemia virus reverse transcriptase (United States Biochemical, Cleveland, OH). For amplification, 2-μl aliquots of cDNA were used as a template in 50μl of reaction buffer containing the respective concentration (pmol)of each primer, 1.5 mm MgCl₂, 200μ m deoxynucleotide triphosphate (Perkin-Elmer,Weiterstadt, Germany), the appropriate amount of 10× PCR-buffer (Roche Diagnostics, Mannheim, Germany), and 1.5 units Taq DNA Polymerase(Roche Diagnostics, Mannheim, Germany) or Ampli-Taq-Gold(Perkin-Elmer). One-step PCR was performed as described recently(29) using 200 ng of total RNA. The cDNA amplifications were performed with multiple primer sets in a thermocycler (Biometra,Göttingen, Germany) with 24–35 cycles of denaturation (1 min;95°C), specific annealing (1 min; 56°C-63°C), and elongation (1 min, 72°C). The amplification products were separated on 1–2% agarose gels, stained with ethidium bromide, and photographed under UV light. Specificity of amplification reactions was confirmed by Southern blot.

Protein extracts were prepared from 1–5 × 10⁶ cells by the standard procedure and electrophoresed (20 μg protein/lane) on 10 or 15% SDS-PAGE. Proteins were electrophoretically transferred onto nitrocellulose membranes(Schleicher & Schüll, Dassel, Germany) using a tank blotter (Bio-Rad, Munich, Germany). and visualized by Ponceau S. The filters were incubated with the mAb antimouse TAP1 (1:750), antimouse TAP2 (1:250), antimouse LMP2 (1:250), and antimouse LMP7 (1:750),kindly provided by Dr. Klaus Früh (Howard Johnsson, La Jolla, CA), as well as with antimouse LMP10 (1:1000), antimouse PA28α (1:2000), and antimouse PA28β mAb (1:1000; Affinity, Exeter,United Kingdom) at 4°C overnight and then incubated with horseradish peroxidase goat antirabbit immunoglobulin (1:1250; DAKO,Hamburg, Germany) for 60 min at room temperature. Filters were then processed using the chemiluminescence kit (ECL; Amersham,Braunschweig, Germany) and exposed to X-OMAT Blue films (Kodak,Rochester, NY). The experiments were performed at least twice.

Cells (2.5 × 10⁶) per incubation were harvested and washed with the incubation buffer[130 mm KCl, 10 mm NaCl, 1 mmCaCl₂, 2 mm EGTA, 2 mmMgCl₂, and 5 mm HEPES (pH 7.3); Ref.30]. Cells were permeabilized with 2 IU streptolysin O/ml (Welcome Reagent Ltd., Beckenham, United Kingdom) in 50μl of incubation buffer for 10 min at 37°C before the addition of 10 μl ATP (10 mm; La Roche Diagnostics,Mannheim, Germany), 2.5 μl of radioiodinated model peptides (peptide no. 63, RYWANATRSI; peptide no. 67, RYWANATRSF; and peptide no. 600,TNKTRIDGQY) and incubation buffer to a final volume of 100 μl. Peptide translocation was performed routinely for 15 min at 37°C. The permeabilized cells were lysed with 1 ml of NP-40 lysis mix. The glycosylated peptides within the endoplasmic reticulum were recovered with ConA-Sepharose (Pharmacia, Uppsala, Sweden) as described previously (30).

Limited information is available about the mechanisms underlying the H-2-deficient phenotype of B16 melanoma cells. Because MHC class I surface expression of TAP-deficient cells can be increased and stabilized by culture at low, unphysiological temperatures or in the presence of specific MHC class I-binding peptides, we determined whether the level of MHC class I expression of two B16 cell variants, i.e., B16F1 and B16F10, could be enhanced under these culture conditions. The T-cell lymphoma cell line RMA, its antigen-processing mutant RMA-S, and the renal cell carcinoma cell line Renca served as controls. The different cell lines were incubated for 16 h in parallel at 37°C in the absence or presence of H-2-binding peptides or at 26°C before MHC class I surface expression was examined by flow cytometry. The results demonstrated that both B16 melanoma cell lines and RMA-S cells expressed barely detectable levels of H-2 antigens, whereas RMA and Renca cells expressed high levels of MHC class I surface antigens. H-2 surface expression was enhanced by neither incubation at low temperatures nor by incubation at 37°C in the presence of specific H-2-binding peptides in both B16 subclones,RMA, and Renca cells. In contrast, an increase of MHC class I antigen expression under these culture conditions was observed in the TAP2-deficient RMA-S cells (Table 2). These data suggest that decreased levels of MHC class I surface antigens in B16 melanoma cells appear to be caused not only by an impaired peptide supply in the endoplasmic reticulum but rather to be attributable to a more complex process.

Indeed, RT-PCR analysis revealed a significant coordinated down-regulation, or even a total inhibition, of mRNA transcription of the major components of the antigen-processing family, including the proteasome subunits LMP2, LMP7, and LMP10, the PAs PA28α and -β,the peptide transporter TAP1, and tapasin in both B16 subclones when compared with RMA control cells (Fig. 1,A). Interestingly, the steady-state mRNA levels of TAP2,calnexin, calreticulin, ER60, and protein disulfide isomerase were either unaltered or only marginally down-regulated in both B16 cell lines (Fig. 1 A; data not shown).

As representatively shown for B16F10 cells, Western blot analysis demonstrated a coordinated lack of protein expression of the major APM components, such as both peptide transporter subunits, LMP2, LMP7,LMP10, and PA28α and -β (Fig. 1,B). The results also suggest a differential regulation of the various APM components. The presence of normal levels of TAP2 mRNA (Fig. 1,A) coupled with low TAP2 protein levels (Fig. 1 B) appears to be attributable to either posttranscriptional down-regulation, a decrease in TAP2 translation, or TAP2 protein stability. In contrast, the lack of or low levels of steady-state mRNA and proteins for the other APM components, such as TAP1, tapasin, the LMP subunits, and PA28, may account for the transcriptional control of these components.

To examine whether the impaired TAP expression in B16 cells was directly associated with a strong suppression of the peptide transport rate, peptide translocation assays were performed. Streptolysin O-permeabilized B16 cells and respective controls (RMA,RMA-S, and Renca) were incubated in the presence and absence of ATP with three different iodinated model peptides (nos. 63, 67, and 600)carrying a glycosylation sequence, respectively. As demonstrated representatively in Fig. 2, for the peptides no. 63 and no. 67, TAP function was nearly totally inhibited in the B16 subclones in the absence and presence of ATP. As expected, similar results were obtained for the TAP2-deficient RMA-S cells (data not shown). In contrast, Renca cells used as positive controls showed high levels of peptide transport in the presence of ATP(Fig. 2), which was comparable with that of RMA cells (data not shown). Similar results were obtained in translocation assays using peptide no. 600 (data not shown).

To investigate whether IFN-γ was able to revert the deficient expression of MHC class I APM genes, the steady-state mRNA and protein levels of the various APM components in the two B16 subclones were analyzed after 24 and 48 h of IFN-γ treatment,respectively. As shown in Fig. 1, the mRNA and protein expression of the peptide transporter, the LMP subunits, the PAs, and/or tapasin was stimulated in the B16F1 and B16 F10 cell lines upon IFN-γadministration. The IFN-γ-mediated induction of these APM components was accompanied by an increase in MHC class I surface expression in both B16 subclones and in RMA and Renca cells but not in TAP2-deficient RMA-S cells (Table 1).

To investigate whether TAP down-regulation is the major component affecting the MHC class I surface expression, TAP1 or TAP2 genes alone and in combination were stably introduced into both B16 melanoma subclones. After selection in G-418,neo^R clones were analyzed for integration and expression of the transgenes by genomic and RT-PCR analysis,respectively. All neo^R B16 transfectants showed both integration and expression of the TAP subunits. However, flow cytometry revealed that overexpression of TAP1 and TAP2 alone or in combination was not accompanied by an induction of H-2 antigen expression (data not shown). Because the impaired expression of MHC class I antigens could not be reconstituted by TAP gene transfer, the coordinated down-regulation of multiple APM components seems to account for the deficient levels of MHC class I antigens on B16 melanoma cells and their immune escape phenotype. The underlying mechanisms of such concordant dysregulation of APM components still have to be elucidated.

The present experiments identify a combined down-regulation of multiple APM components as a strategy for the immune evasion of B16 melanoma cells. This is in line with results in human tumor cells and virus-infected cells, as well as in in vitro models of oncogenic transformation, describing LMP, PA28, TAP, and/or tapasin down-regulation combined with altered MHC class I surface expression(13, 14, 15, 16, 23, 26, 31).

A coordinated suppression of LMP, PA28, TAP, tapasin, and MHC class I heavy chain was observed in the B16 subclones (Fig. 1, Table 2), but both the extent as well as the level of down-regulation of these components appear to be distinct. With the exception of TAP2 and a number of chaperones, low levels of steady-state mRNA as well as protein was found for various APM components. Transcriptional down-regulation or low mRNA stability may account for this defect. In contrast, the normal levels of TAP2 mRNA associated with low TAP2 protein levels may be attributable to a posttranscriptional or translational down-regulation or may occur at the posttranslational level (Fig. 1).

The simultaneous down-regulation involved genes not only clustered in the MHC class II locus, suggesting a general mechanism of APM-component suppression in the B16 melanoma system. Thus, the B16 melanoma model serves as an example of how tumor cells can and do avoid the host immune system. These results are in accordance with those of Johnsen et al., (32) demonstrating an impaired expression of various APM molecules in tumors of distinct origin. On the other hand, these data are in contrast with oncogene- and virus-transformed murine fibroblasts, in which a concordant lack of a few specific APM components was described, whereas the expression of other APM genes was only marginally affected or unaltered(23, 26, 31). Because of the extensive defects in the B16 system, more than known previously, the generation of T-cell responses may require either the repair of the dysfunctional mechanisms, which is difficult unless the precise lesion(s) is (are) identified, or cross-priming (12).

The underlying mechanisms of the described concordant dysregulation of APM components in B16 cells still have to be elucidated. The immune escape phenotype of B16 cells could be abolished by cytokine treatment,but not by gene transfer of one of the major APM components, the peptide transporter TAP (data not shown), as it has been shown for human tumor cells (33, 34, 35, 36). Because the decreased expression of proteasome subunits is not likely to decrease MHC class I expression (15), one might speculate that the defect described in B16 subclones is attributable to impaired tapasin expression.

Because of the IFN-γ-mediated induction of APM components in B16 cells, the deficient expression of these genes appears to be attributable to regulatory mechanisms rather than to structural alterations (Fig. 1; Table 2). At least for TAP1 and LMP2, a concordant expression has been implemented in both human and murine cells because of a shared bidirectional promoter (37). Although the TAP1/LMP2, LMP7, LMP10, and PA28α promoters contain different elements, the observed simultaneous down-regulation of these APM components argues for a common regulatory mechanism in B16 melanoma cells, which is presently under investigation. In addition to the IFN-γ-mediated stimulation of various APM components, MHC class I surface expression was also increased upon IFN-γ treatment of B16 subclones, but to a lesser extent (Table 2). These data suggest that additional regulatory factors, which are important for efficient MHC class I antigen processing and presentation, may exist. This hypothesis is also strengthened (1) by the fact that the MHC class I surface expression of B16 subclones could be enhanced by neither the incubation of cells at low temperatures nor by the addition of exogeneous MHC class I-binding peptides (Table 2; Ref.2) and by the partial reconstitution of the MHC class I surface expression in adenovirus 12-transformed fibroblasts after the gene transfer of deficient APM components (31). In addition, the induction of APM component expression upon IFN-γtreatment may affect the H-2-deficient phenotype of B16 cells by altering the quality and quantity of generated antigenic peptides. Immunization with B16 cells stably transfected with the IFN-γ gene alone, or particularly in combination with H-2 genes, induced cytotoxic T-cell responses, a longer survival, and a significant inhibition of metastasis formation(5, 6, 38). However, these CD8⁺cytotoxic T lymphocytes may not be directed solely against the peptide repertoire presented by unmodified B16 cells.

Because of the extensive defects, more than known previously, T cell-based immunotherapy in the B16 system is much more difficult than expected. The generation of T-cell responses and an effective antitumor immunity may require the correction of the underlying dysfunction, e.g., by IFN-γ. Thus, given the presence of the deficiencies in the antigen-processing pathway, the B16 melanoma cells should not be used as a prototype for vaccination and immunization strategies, or, at least, investigators using this model must be aware of its demanding nature. A complete understanding of the processes involved in APM suppression of B16 cells will provide new insights into the molecular mechanisms of immune escape as well as into the possibilities of how to repair those pathways. In this respect, B16 melanoma cells represent an excellent model system to study the regulation of APM components, to identify new factors involved in this process, and to optimize the effectivity of immunotherapy in correcting these defects.

We thank Drs. Thomas Spies for the TAP1 and TAP2 plasmids, Klaus Früh for antisera against mouse TAP1/2, LMP2, and LMP7, and Dirk Jung for constructing the TAP expression vectors.