[CANCER RESEARCH 51, 3164-3170, June 15, I991|

Expression of HLA-A2 Antigen in Human Melanoma Cell Lines and its Role inT-Cell Recognition1

Franco Pandolfi,2 Lenora A. Boyle, Livio Trentin, James T. Kurnick, Kurt J. Isselbacher, andSebastiano Gattoni-Celli3

Department of Pathology ¡F.P., L. A. B., L. T., J. T. K.¡and Cancer Center [K. J. I., S. G-CJ, Massachusetts General Hospital and Harvard Medical School,Charlestown, Massachusetts 02129

ABSTRACT

Previous studies have suggested that, in human melanoma, expressionof HLA-A2 antigen is important for tumor cell recognition by autologousT-Iymphocytes. Because of the recent demonstration that expression ofH LA Class I antigens may be selectively lost in several human tumors,including melanoma, we derived pairs of tumor infiltrating lymphocytes(TIL) and melanoma cell lines from 4 human lymphocytic antigen (111 \ )-A2* patients with metastatic melanoma. We observed that, although all4 TIL cultures expressed HLA-A2 antigen, only 2 melanoma cell linesdid so. Melanoma cells derived from the other 2 patients showed neithersurface expression of the HLA-A2 antigen nor presence of the corresponding iiiKN'A. We also observed some correlation between loss of

HLA-A2 expression and level of c-myc transcription. TIL derived frompatients whose melanoma cell lines had normal expression of HLA-A2had a CDS phenotype and were capable of lysing autologous melanomacells. Melanoma cell killing was CD3 and major histocompatibilitycomplex Class I restricted in both cases, but HLA-A2 restricted in onlyone case. On the other hand, TIL derived from the 2 patients whosemelanoma cell lines had lost expression of HLA-A2 had a predominantCD4 phenotype and virtually no cytotoxic activity. Preincubation of theHLA-A2 negative melanoma cell lines with a- or 7-interferon did notinduce the re-expression of the HLA-A2 antigen. In an attempt to restoreHLA-A2 antigen expression in one of the melanoma cell lines that wereHLA-A2 negative, we transfected these cells with the HLA-A2 genesubcloned in the pSV2-neo vector. Four transfected clones, with highlevels of HLA-A2 antigen expression, were expanded and characterized.Proliferative and cytotoxic activities of TIL against the autologous transfected clones as well as the untransfected parental melanoma cell linewere measured and compared. ( '1)4* TIL showed no difference in the

proliferative response to autologous parental and HLA-A2 transfectedclones. However, we observed selective recognition of the HLA-A2expressing clones by autologous cultured peripheral blood lymphocytes(which contained CDS cells) as well as allogeneic CDS* TIL with a 111A-

A2 restricted pattern of recognition. In contrast, virtually no cytotoxicactivity was detected against either parental or HLA-A2 transfectedclones. Overall, our data suggest that selective down-regulation of 111A-A2 antigen expression in melanoma cells may represent one of themechanisms by which tumor cells escape immunological recognition.

INTRODUCTION

MHC-I4 antigens are integral membrane glycoproteins that

are expressed at varying levels on the surface of virtually allsomatic cells. MHC-I molecules appear to function by binding

Received 2/21/91; accepted 4/4/91.The costs of publication of this article were defrayed in part by the payment

of page charges. This article must therefore be hereby marked advertisement inaccordance with 18 U.S.C. Section. 1734 solely to indicate this fact.

1This work was supported in part by NIH Grants CA 44324. AR 39993, andDK 36350, American Cancer Society Institutional Grant 90-1008, and the NorthAtlantic Treaty Organizaiton (NATO).

2To whom requests for reprints should be addressed, at Massachusetts General

Hospital Pathology Research, 149 Thirteenth Street, Charlestown, MA 02129.3 Present address: Division of Radiation and Cancer Biology. New England

Medical Center, Box 824, 750 Washington Street, Boston. MA 02111.4The abbreviations used are: MHC-I, major histocompatibility complex Class

I; CTL, cytotoxic T lymphocytes; TIL, tumor infiltrating lymphocytes; IL-2,interleukin-2; mAb, monoclonal antibody; PHA. phytohemagglutinin: PBL, peripheral blood lymphocyte; FBS, fetal bovine serum; IFN. interferon; HLA,human lymphocyte antigen.

peptides from degraded endogenous proteins (processed antigens) and presenting these to T cell receptor specific for aparticular MHC-I antigen/peptide complex (combinatorial ep-itope) (1). Crystallographic studies have revealed that the a,and a2 domains are folded to form a cleft that is lined by theside chain of the polymorphic residues (2, 3). These studiesprovide a molecular explanation for how specific MHC-I alíelemay present some, but not all of the possible antigenic peptides.Thus, it may be predicted that down-regulation of a givenMHC-I alíelewould have a detectable effect on immune responses involving that allelic gene product.

Many human tumors, particularly those of epithelial origin,appear to express greatly reduced levels of MHC-I (HLA-I)antigens (4). More recently, selective changes in the expressionof the HLA-A2 and possibly other alíeleshave been reported;these changes, ranging from reduction to loss of expression,have been observed in several cancers, including human melanoma (5, 6). These observations have led to speculation thatmalignant progression could result from the loss of recognitionby CTL, allowing tumors to escape immune surveillance.

Wolfel et al. (7) have identified the HLA-A2 antigen as arestricting element for lysis of 3 different melanoma-associatedantigens. Similarly, Darrow et al. (8) observed that HLA-A2,and possibly other HLA-A alíeles,may function as restrictionelements for the recognition of shared melanoma-associatedantigens by CTL. This latter research group has also shownthat HLA-A2 represents a preferential restriction element forthe generation of CTL reactive against autologous melanomas(9). Taken together, these studies underscore the importance ofHLA-A2 antigen expression by human melanomas in targetinglysis by CTL.

We have derived pairs of tumor-infiltrating lymphocytes andautologous melanoma cell lines from 4 HLA-A2 patients withmalignant melanoma. HLA-A2 expression was lost in 2 of the4 melanoma lines and this loss correlated with lack of immunological recognition by autologous lymphocytes. To furtherstudy the role of HLA-A2 expression in the CTL responseagainst autologous melanomas, we have restored, by DNA-mediated gene transfer, the expression of HLA-A2 antigen inone of these 2 melanoma cell lines. Four transfected clones,expressing high levels of HLA-A2 antigen, were used as targetsin cytotoxicity and T cell activation assays. In our experimentalconditions, re-expression of HLA-A2 was associated with partial restoration of immunological recognition. This suggeststhat the down-regulation of HLA-A2 antigens might representan important mechanism by which melanoma cells could escapeimmune recognition. Additional factor(s) may also be neededto achieve effective lysis of autologous melanomas by CTL.

MATERIALS AND METHODS

Cell Lines. Four pairs of melanoma cell lines and TIL were derivedfrom parallel cultures originated from biopsies of HLA-A2 positivepatients with metastatic melanoma. Tissues were cut into 1-mm frag-

3164

on July 3, 2018. © 1991 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

HLA-A2 ANTIGEN IN HUMAN MELANOMA

merits and cultured in either RPMI supplemented with 10% FBS orRPMI supplemented with 5% human serum and IL-2 (100 units/ml).Several parallel cultures of melanoma cells and TIL were obtained (10,11). To confirm the melanoma origin of our cell lines, we used S-100and HMB-45 (12) mAb in indirect immunofluorescence assays on fixedcytocentrifuged samples. These data, in conjunction with morphological observations and melanin production in some cell lines, confirmedthe melanoma origin of the cells. TIL cultures were propagated in thepresence of IL-2 and restimulated with PHA and irradiated feeder cellsevery 2-4 weeks to enhance proliferation, as reported previously (10,11). Melanoma cell lines were labeled as M (for melanoma), followedby a patient identification code: 1 for Paient U. M., 2 for C. M., 3 forO. M., 4 for A. N., 5 for A. M. TIL cultures were labeled as TILfollowed by the patient identification number. For example, M l andTIL-1 indicate the melanoma cell line and the TIL cell line, respectively,derived from the same patient (U. M.). Long term cultures of IL-2-dependent cells were also established from PBL of Patient 2. Lymphocytes were isolated on a density gradient, activated with PHA and thenmaintained in IL-2 containing medium as described above.

Monoclonal Antibodies. Expression of surface antigen was studiedwith a panel of commercially available mAb, including anti-CD3, -4,-8 (Ortho Diagnostic System), -CD 16, and -CD57 (Becton Dickinson),W6/32 (specific for all MHC-I HLA heavy chains); BB7.2 (HLA-A2);MA2.1 (HLA-2 + HLA-B17); PA2.1 (HLA-A2 + HLA-A28) (fromAmerican Type Culture Collection). Monoclonal antibodies were usedin an indirect immunofluorescence test (13) and analyzed using a BectonDickinson FACScan II.

Proliferative Response of TIL to Melanoma Cell Lines. Autologousand allogeneic TIL cultures were washed and resuspended in RPMI1640 and added (5 x IO4 cells/well) to a 96-well fiat bottomed tissueculture dish in the presence of 5 units/ml of IL-2. Cesium-irradiated(12.5 x IO3 rad) melanoma cells were added at 1 x IO4cells/well and

used as stimulators. Experiments were performed in triplicate. Theplates were cultured for 5 to 6 days at 37°Cin a humidified atmospherecontaining 5% CO2, labeled with tritiated thymidine ('H, l ¿iCi/well)

overnight, and harvested onto paper filters using a Skatron cell harvester. 3H incorporation was measured with a ß-scintillationcounter.

Cytotoxicity. Cytotoxic activity was assessed by the lysis of s'Cr-

labeled target cells in a 4-h Cr release assay, as described previously(14). Target cells were represented by the autologous melanoma celllines or the natural killer cell-sensitive K562 cell line. Effector cellswere TIL kept in IL-2, 8-20 days after restimulation with PHA andirradiated feeder cells. Results are reported as percent of specific lysis.

Treatment of Melanoma Cells with IFN. To evaluate the effect of «-or 7-IFN on melanoma cells as targets in the cytotoxic assay, eachmelanoma cell line was incubated for 48 h with 1000 units/ml of «-IFN (Intron A; Shering Port Reading, NJ) or with 500 units/ml of 7-IFN (a gift from Genentech, San Francisco, CA) in RPMI with FBS at37°Cin a 5% CO2 atmosphere. These conditions were shown to produce

optimal enhancement of the expression of HLA-DR antigens on controlmelanoma cell lines. Melanoma cells incubated with IFN were washedand studied for expression of surface antigens as described above.

Inhibition of Cellular Cytotoxicity by Preincubation with MonoclonalAntibodies. TIL were incubated for l h at 37°Cin a CO2 atmosphere

with anti-CD3, anti-MHC-I, or anti-HLA-A2 mAb. A final dilution of10 Me/ml of purified antibodies (OKT3) or a 1:100 dilution of ascites(other mAb) were used. Cells incubated with the mAb were then testedin the Cytotoxicity assay as described. Excess antibody was presentthroughout the Cytotoxicity assay. As a control, an anti-HLA-DR mAbwas also used. Results are expressed as percent of inhibition of theCytotoxicity observed without a blocking antibody, at a given effectorto target ratio. Results are means of triplicate experiments.

DNA-Mediated Gene Transfer. The HLA-A2 gene subcloned in theEcoRl site of the vector pSV2-neo was generously provided to us byDr. M. Vega and Dr. J. Strominger (15). This plasmid, designatedpHLA-A2, contains the entire HLA-A2 genomic clone including theregulatory sequences and is capable of conferring the neomycin-resist-

ance phenotype to transfected cells. Exponentially growing cell cultureswere transfected with pHLA-A2 using the calcium phosphate precipi

tation technique as modified by Wiglere/a/. (16). Individual neomycin-resistant clones, selected in Dulbecco's modified Eagle's medium con

taining 10% FBS and the neomycin-analogue G418 (GIBCO) at 600Mg/ml, were picked 3 weeks after transfection and expanded for furthercharacterization. In control experiments, melanoma cell cultures weretransfected with plasmid pRSV-neo (17), carrying only the neomycin-resistance gene.

RNA Analysis and Probes. Total RNA was extracted from melanomacell cultures as described (18), and 10 ng/sample were used in RNaseprotection assays (19) to analyze the levels of HLA-A2 mRNA. Asprobes we have used 2 A2-specific Smal fragments, corresponding tothe «iand «2domains, subcloned into the riboprobe vector bluescript(Stratagene). For RNA blot hybridization, about 30 /jg/sample of totalRNA sample were size separated by denaturing gel electrophoresis andblot hybridized as described (20). All hybridizations were performed athigh stringency. As probes we used the following gel purified fragments,labeled with 32P by nick-translation: human c-myc (Xho\/EcoR\ DNAsegment encompassing exons 2 and 3 from plasmid pHSR-1) (21);Class I HLA (1.4 kilobase Pstl fragment corresponding to the entireHLA-B7 cDNA) (22); Class I HLA-B (358-base pair Pst\/Pvu\\ fragment encompassing the 3'-untranslated region of a HLA-B8 cDNA

clone) (23); and actin (500-base pair £coRIfragment of human skeletala-actin coding region, corresponding to amino acids 205-372, generously provided by Dr. Larry Kedes).

RESULTS

Expression of HLA-A2 on Human Melanoma Cell Lines. Weevaluated the surface expression of total HLA Class I andHLA-A2 antigens in 4 melanoma cell lines derived from HLA-A2+ patients. All melanoma cell lines and corresponding TILcultures had detectable expression of HLA-CIass I antigens asassessed by immunofluorescence staining with W6/32 mAb(not shown). Moreover, all 4 TIL cultures exhibited high levelsof HLA-A2 antigen expression. In contrast, only 2 (Ml andM5) of the 4 melanoma cell lines expressed normal levels ofHLA-A2 antigen on the cell surface, whereas in M2 and M3cultures HLA-A2 expression was undetectable (Fig. 1).

Loss of surface expression of HLA-A2 antigen in these 2melanoma cell lines was also confirmed at the mRNA level.Analysis of HLA-A2 transcription was carried out by RNaseprotection using 2 separate riboprobes specific for the a\ or «2domain of the HLA-A2 gene (Fig. 2). RNase protection is verysensitive and highly specific, preventing problems of cross-hybridization that are common with RNA blot analysis (seealso below). Of the 5 melanoma cell lines tested, 2 were positivefor HLA-A2 mRNA (Ml and M5). In contrast, the remaining3 samples were negative, either because of loss of expression(M2 and M3) or, in the case of the M4 cells, because this cellline was derived from a patient with a HLA-A2 negativehaplotype.

Treatment of M2 and M3 cells with a- or 7-IFN was unableto restore expression of HLA-A2 antigen (data not shown).

Phenotype of Lymphocyte Cultures. Virtually all TIL (>95%)were CD3 positive. As shown in Table 1, TIL derived fromPatients 1 and 5, whose autologous melanoma cell lines expressed normal levels of HLA-A2 antigen, were almost pureCDS cells, whereas TIL cultures derived from Patient 2 exhibited an almost pure CD4 phenotype, and those derived fromPatient 3 had a prevalence of CD4 cells. All cells were consistently negative for CD 16 and CD57 markers, suggesting thatour TIL cultures do not contain a significant proportion ofnatural killer cells and that MHC-I restricted Cytotoxicity (seebelow) did not correlate with CD57 expression. PBL culturedfrom Patient 2 were also analyzed and turned out to be a mixedpopulation with a prevalence (55%) of cells expressing CDS.

3165

on July 3, 2018. © 1991 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

HLA-A2 ANTIGEN IN HUMAN MELANOMA

A B

Fig. 1. Surface expression of HLA-A2 antigen in pairs of TIL and melanoma cell linesderived from the same patients. Numbers on theleft refer to patient numbers. Letters on top identify the columns as follows. A and B columnscorrespond to TIL staining (control and HLA-A2 staining, respectively); C and D columns represent the control and HLA-A2 staining of thecorresponding melanoma cells. HLA-A2 was expressed by all TIL cells. On the other hand,expression of this antigen was lost on melanomacells derived from Patients 2 and 3. Resultsshown were obtained with BB7.2 mAb (PatientsI, 3, and 5) and with PA2.1 (Patient 2). Stainingwith each anti-HLA-A2 mAb (including MA2.1)gave virtually identical results in all testedsamples.

mi jIMAIMV j1W

H4»>,„,„,,„.,IUIM¡MIM/\PU*i11

incoT-«NW W N

LU*oc

Table 1 Surface expression ofCD4 and CDS markers in TIL derived from HLA-A2-positive patients with metastatic melanoma

TIL-1 TIL-2 TIL-3 TIL-5

HLA-A2 expression on Yes No No Yesautologous melanomacells

CD4 <5° 100 71 6CDS 100 <5 35 96

•Results are expressed as % of positive cells.

protectedprobe — .281

-271

iMg. 2. RNase protection assay. Ten fig of total RNA extracted from eachmelanoma cell culture were reacted with a 32P-labeled riboprobe specific for the«2region of HLA-A2. Protection is indicated by the presence of a specific band(277 bases), whose intensity is proportional to the amount of specific mRNA inthe sample. Note that while M2 parental cells had undetectable level of transcript,the M2 transfectants (A2-5, -8, -10, -2) expressed high levels of the transcript.Cell lines MS and Ml were positive for HLA-A2 expression, whereas M3 andM4 were negative. Parallel RNase protection experiments with the «,riboprobegave identical results (data not shown). M4 is a melanoma cell line derived froma HLA-A2 negative patient, used as a control.

TIL-1TIL-2

TIL-3

TIL-5

EFFECTOR:TARQET RATIOS

Fig. 3. Cytotoxic activities of TIL derived from melanoma patients againstautologous melanoma cell lines. Results are expressed as percent of cytotoxicityat different effectontarget ratios. TIL-1 and TIL-5 had high levels of lytic capacity,while TIL-2 and TIL-3 showed virtually no cytotoxicity. HLA-A2 was expressedonly on Ml and MS melanoma cell lines, but was absent from M2 and M3 lines.

3166

on July 3, 2018. © 1991 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

HLA-A2 ANTIGEN IN HUMAN MELANOMA

MA2.1

HLA-DR

HLA-CLASS I

CD3

60

% INHIBITION

80 100

neomycin-resistance gene were negative for HLA-A2 mRNA.In contrast, all 4 independent M2 clones expressing the trans-fected pHLA-A2 plasmid exhibited a prominent signal specificfor HLA-A2 transcription (protected probe).

Flow-cytometry experiments run in parallel on transfectedclones showed surface expression of HLA-A2 antigen on all

TIL-5 transfected clones (Fig. 5). High levels of HLA-A2 antigen wereTin consistently detected in the transfected clones even after long-

term culture.Analysis of Proliferative and Cytotoxic Activities of Lympho

cytes Using the Transfected Clones as Target Cells. To investigate the role of HLA-A2 in tumor recognition mediated by T-lymphocytes, we have examined the proliferative response ofT-cells to irradiated melanoma M2 cells (HLA-A2 negative)and the 4 HLA-A2 transfected clones, in the presence of IL-2.In 3 separate experiments, autologous TIL (TIL-2) did not

Fig. 4. Effect of mAb to CD3, HLA-Class I, and HLA-A2 on the cytotoxicactivity of TIL against autologous melanoma cell lines. Results are expressed aspercent of inhibition of the autologous killing by the 2 highly cytotoxic TIL celllines (TIL-1 and TIL-5) after incubation with different mAb. TIL were preincu-bated with an anti-CD3 mAb, whereas melanoma cells were preincubated withW6/32 (HLA-Class I) or MA2.I (HLA-A2). Anti-CD3 and Class I mAb gave astrong inhibition in both cases. On the other hand, only TIL-1 cells were inhibitedby MA2. 1 mAb. Similar results were obtained with PA2. 1 and BB7.2 (anti-HLA-A2) mAb (data not shown).

Cytotoxic Activities of TIL against Autologous Melanoma CellLines. TIL were tested for their ability to lyse autologousmelanoma cell lines in a Cr release assay (Fig. 3), and acorrelation between the expression of HLA-A2 antigen onmelanoma cells and cytotoxicity by TIL was observed. In fact,TIL from Patients 1 and 5 effectively lysed the autologousmelanoma cells. In contrast, TIL-2 and 3, whose autologousmelanoma cells lacked HLA-A2 antigen expression, showed nolytic activity.

Inhibition of Cytotoxic Activity by Anti-HLA and Anti-CD3Antibodies. To further corroborate the role of the HLA-Class Imolecules in the autologous killing by TIL, we performedinhibition experiments preincubating the melanoma cell lineswith W6/32 or with anti-HLA-A2 mAb. As shown in Fig. 4,preincubation of the melanoma cells with W6/32 resulted in adramatic inhibition of the cytotoxic activity as did incubationof TIL with anti-CD3. No inhibition was observed with an anti-DR mAb. Using MA2.1 (anti-HLA-A2), we observed 48%inhibition with TIL-1, whereas with TIL-5 very little inhibitionwas observed. Similar results were also obtained with 2 additional anti-HLA-A2 mAb (BB7.2 and PA2.1) (data not shown).The fact that anti-HLA-A2 mAb were only partially effectiveat inhibiting cytotoxicity was not surprising since they recognizedeterminants present on the hypervariable extracellular domains (<*ior <*2),whereas it has been shown that the constantdomain (a3) of the Class I heavy chain expressed on target cellsis the one directly involved in recognition by CTL (24).

HLA-A2 restriction of the killing by TIL-1 cells was alsosuggested by allogeneic killing of the M5 (HLA-A2 positive)cell line, but not of M2 or M3 (HLA-A2 negative) cell lines(data not shown).

Transfection of M2 Cells with pHLA-A2. In an attempt torestore expression of HLA-A2 in M2 melanoma cells and toevaluate the role of this antigen lymphocyte recognition, exponentially growing M2 cultures were transfected with pHLA-A2by the calcium phosphate precipitation technique. IndividualG418-resistant clones were picked at about 3 weeks after trans-fection and expanded for extraction of total RNA and flow-cytometry assays. Fig. 2 shows that the RNA from the parentalcell line M2 as well as from M2 cells transfected with the

UL

A2.2

A2.5

1

A 2 -8

A 2.10 AFig. 5. Surface expression of HLA-A2 antigen on transfected clones derived

from the M2 cell line. Column A (control) and Column B (staining with BB7.2,anti-HLA-A2 mAb) of 4 clones (A2.2, A2.5, A2.8, and A2.10) transfected withthe plasmid pHLA-A2. All transfectants showed high expression of the HLA-A2antigen. M2 cells transfected with the plasmid pRSV-neo were used as controland showed no expression of HLA-A2. Other anti-HLA-A2 mAb (MA2.1 andPA2.1) gave similar results (data not shown).

3167

on July 3, 2018. © 1991 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

HLA-A2 ANTIGEN IN HUMAN MELANOMA

TIL-1

PBL-2

TIL-2

D M2-A2.10

Eä M2-A2.8

EI M2-A25

0 M2-A2.2

• M2

10000

CPM

20000

Fig. 6. Proliferative response of cultured lymphocytes derived from melanomapatients against irradiated stimulators (the M2 cell line and 4 HLA-A2 transfectedclones) in the presence of IL-2. TIL-2 cells (which were of the CD4 phenotype)did not proliferate in response to autologous M2 or transfected melanoma cells.On the other hand, cultured autologous PBL-2 (which contained CD4 as well asCDS cells) showed proliferation in response to HLA-A2 transfected clones butnot to M2 parental cells. Similar results were obtained using TIL-1 as effectorcells. These cells could also proliferate in response to the 4 transfected clones butnot to M2 cells. Three separate experiments were done and gave similar results.Results are expressed as cpm and represent means of triplicate wells.

• Ml

D M2-A2.10

0 M2-A28

EI M2-A2.5

0 M2-A22

• M2

% CYTOTOXICITY

Fig. 7. Autologous and allogeneic cytotoxic activities of cultured lymphocytesderived from melanoma patients against the M2 melanoma cell line and 4 M2-derived HLA-A2-transfected clones. TIL-1 had strong autologous lytic activity,but did not recognize the M2 cell line. TIL-2 and PBL-2 failed to show autologouskilling. PBL-2 had some allogeneic activity when Ml cells were used as targets.No significant lytic activity was observed when the 4 HLA-A2 transfected cloneswere used as targets.

recognize the transfected clones, whereas autologous culturedPBL (PBL-2) or allogeneic cells from Patient 1 (TIL-1) showeda consistent proliferative response (Fig. 6). On the other hand,cytotoxicity experiments (Fig. 7) showed no consistent recognition of the transfected clones by autologous or allogeneiceffector cells.

c-myc Expression of Melanoma Cell Lines. In parallel experiments, we have examined the relationship between c-myc transcription and MHC-I expression in our melanoma cell lines(including the transfected clones) by Northern blot hybridization. Fig. 8 (left and right panels) refers to a single filterhybridized sequentially with the indicated probes. The result ofblot hybridizations shown in Fig. 8 (right panel) appeared toconfirm the inverse correlation between c-myc and HLA expres

sion, most clearly in the case of the Ml cell line. Ml cells,which were HLA-A2 positive by RNase protection and flowcytometry, would also appear to be HLA-B positive by RNAblot hybridization. However, flow-cytometry analyses performed with locus-specific mAb have revealed that M l cells areHLA-B negative (data now shown). Therefore, we consideredthe possibility that the HLA-B-specific probe (a 358-base pairPstl/Pvull fragment spanning the 3'-untranslated region of the

HLA-B8 cDNA clone) (23) might still cross-hybridize withother members of the HLA family of Class I genes. The resultof blot hybridizations shown in Fig. 8 (left panel) providessupport for this interpretation, since the HLA-B probe hybridizes to the RNA of M2 clones expressing the transfected HLA-A2 gene in proportion to the amount of HLA-A2 transcript, asmeasured by RNase protection (see Fig. 2). We also noticedthat while the parental M2 cells and the neomycin-resistanttransfectants expressed equivalent amounts of c-myc mRNA,much less c-myc transcript was detectable in 3 of the transfectants expressing HLA-A2 antigen. Hybridization with the actinprobe confirmed that approximately equivalent amounts oftotal RNA were loaded per lane.

DISCUSSION

Although it is well established that malignant cells mayexpress low amounts of MHC-I antigens on the cell membrane(25), only recently has it been shown that a selective loss ofHLA alíelesmay also occur (6, 26). Previous studies haveindicated that the expression of the HLA-A2 antigen may beimportant in the recognition of melanoma cells by autologousT-lymphocytes (7, 9, 27). We have investigated the expressionof this antigen in 4 melanoma cell lines derived from HLA-A2"1"

patients and found that 2 of the 4 cell lines studied had neitherspecific mRNA nor surface expression of the HLA-A2 antigen.Autologous TIL corresponding to these 2 cell lines had apredominant CD4 phenotype and were unable to kill autologouscells. Although the loss of HLA-A2 in our cell lines could have

Oni

CM T^ 00

CM CM CM< < <

r- CO IO

2 5 5—c-myc —

HLA —

HLA-B— «•»——

—act i n —

Fig. 8. RNA blot hybridization. Total RNA (about 30 Mg/sample), derivedfrom melanoma cell lines and transfected clones, were run on denaturing agarosegel and transferred to nitrocellulose filter. Each of the 2 filters was sequentiallyhybridized with the indicated 32P-labeled probes. All the hybridizations were

performed in conditions of high stringency.

3168

on July 3, 2018. © 1991 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

HLA-A2 ANTIGEN IN HUMAN MELANOMA

occurred during in vitro propagation of the cells, it is worthnoting that efficient CTL were only derived from TIL obtainedfrom patients whose melanoma cell line expressed detectablelevels of HLA-A2.

Several lines of evidence suggest that the development andprogression of malignant melanoma may be at least partially-

controlled by the immune system. For instance, a significantcorrelation has been shown between the presence of TIL inprimary lesions and a favorable prognosis (28). In addition,immunotherapy with cells capable of lytic activities such aslymphokine-activated killer cells or TIL has proven effective insome case (29, 30). The observations reported in the presentstudy may have some important implications for better understanding the biology of lymphocytic response to melanoma cellsas well as for developing more efficient immunotherapeuticprotocols. In our cell lines, the expression of HLA-A2 antigenon melanoma cells correlated with the presence of CDS cells inthe TIL population. This observation raises the interestingpossibility that HLA-A2 expression may be needed for recruitment or proliferation of CDS lymphocytes and that loss ofClass I expression leads to a selective loss of CDS cells in TIL,with consequent impairment of tumor killing. Despite the finding of a significant correlation between clinical survival and thedegree of lymphocytic infiltration, data relating the clinicaloutcome of melanoma patients to the prevalence of either CD4or CDS cells in the tumor infiltrate are not available. AlthoughCD4 cells may exert cytotoxic activity against autologous melanoma cells (31, 32) in our experience as well as that of others(9), HLA-A2 restricted cytotoxicity was mostly confined toCDS cells and was T-cell receptor and MHC-I restricted asshown by inhibition experiments. Therefore, the hypothesis issuggested that down-regulation of the HLA-A2 antigen onmelanoma cells may have contributed to a decreased tumorsusceptibility to CTL because of a lack of CDS cells in the TIL.

To further understand the role of the HLA-A2 antigen intumor recognition by lymphoid cells, we have first attemptedto induce the expression of this antigen by incubating the HLA-A2-negative M2 and M3 cells with «-or 7-IFN. No expressionof HLA-A2 antigen could be induced at the mRNA or surfacemembrane level in these conditions. Therefore, we introducedthe HLA-A2 gene into M2 cells and obtained 4 independenttransfected clones expressing high levels of the HLA-A2 antigen. This finding rules out the possibility that M2 cells areinherently incapable of expressing HLA-A2; rather, the down-regulation appears to involve specifically the endogenous geneat the transcription level. However, we cannot rule out a mutation in the endogenous HLA-A2 gene in M2 cells.

Several experiments were performed to investigate the abilityof lymphocytes from different sources to recognize (by proliferation or cytotoxicity) the transfected clones. First, we observed that TIL-2 cells, tested in proliferative or cytotoxicityassays, did not appear to distinguish between M2 cell line orthe M2 derived transfectants expressing the HLA-A2 antigen.This was not surprising, considering the homogeneous CD4phenotype of TIL-2 cells, which would preclude MHC-I restricted recognition. To overcome this lack of autologous CDSeffectors with potential ability to recognize tumor-associatedantigens presented by the re-expressed HLA-A2 molecule, wederived a lymphocyte culture (PBL-2) from the peripheral bloodof the same patient. This cell line contained more than 50%CDS lymphocytes, and indeed these cells had some lytic capacity when tested against the allogeneic Ml (HLA-A2*) mela

noma target (Fig. 7). As an additional control, we used in the

same experiments a highly cytotoxic cell line (TIL-1) whosekilling is MHC-I and HLA-A2 restricted.

In the proliferation assay, both PBL-2 and TIL-1 showedactivation in response to all 4 M2-derived HLA-A2-transfectedclones, whereas no proliferation could be observed against theM2 parental cell line. These data suggest that the expression ofthe HLA-A2 molecule is relevant in presenting an antigenicdeterminant not recognized on the parental melanoma cell linethat is able to induce T-cell proliferation, although the natureof this antigenic determinant remains to be determined.

In the cytotoxicity assays, however, no significant lysis of thetransfected clones could be consistently shown. The negativeresults obtained in the cytotoxicity assay may be explained (atleast in the case of autologous PBL-2 cells) by the fact thatthese cells were not primed in vivo by the presence of theantigen(s). Therefore, we took advantage of a recently describedtechnique that allows the generation of autologous melanoma-specific CTL cells using HLA-A2-matched allogeneic melanoma cells (9). We incubated PBL-2 cells with irradiated HLA-A2+ positive Ml cells in the presence of 5 units/ml of IL-2,

and restimulated them with the same irradiated cells at 7-14-day intervals. This was done in the attempt to "educate" PBL-

2 cells in vitro to recognize the putative melanoma antigen(s)presented within the HLA-A2 complex. Once again, PBL-2cells, cultured as described, failed to kill autologous HLA-A2-transfected clones (data not shown). We cannot rule out thatthe priming of the cells in vitro may have not been effective ininducing recognition of the transfected cells. However, sinceboth PBL-2 and TIL-1 cells were able to lyse allogeneic (Mland M5) cells in a HLA-A2-restricted fashion, failure to lysethe M2-transfected clones suggests that additional factors, besides HLA-A2 antigen expression, are needed for effective lysis.The moderate antigenicity of the complex presented by theHLA-A2 molecule, which allows a proliferative response, maynot have been adequate to trigger cytotoxic abilities. An additional possibility is that expression of the HLA-A2 moleculeon the melanoma cells, although necessary for recognition, assuggested by our proliferation data, may not be sufficient perse to trigger the lytic process. If this is the case, other mole-cule(s), such as minor histocompatibility antigens, may play asignificant although yet not determined role. In fact, we havepreviously shown that lack of expression of these antigens onmelanoma cells may influence the immunogenicity of the tumor(33).

The mechanisms responsible for the selective loss of HLAalíelesin human melanoma are poorly understood. An inversecorrelation between the level of Class I HLA-B expression andthat of c-myc oncogene has been reported in human melanomacell lines (34). However, our observations do not support thesuggestion that up-regulation of c-myc preferentially suppressesClass I HLA-B locus. In fact, c-myc expression was very low inMl cells that do not express surface HLA-B, whereas they doexpress HLA-A2 antigen. Overall, the results of RNA blothybridization confirm the inverse correlation between c-mycand HLA expression, although the correlation was not perfectand certainly not locus specific.

In summary, our data lend support to the view that down-regulation of HLA-A2 in human melanoma may represent arelatively frequent event that could alter lymphocytic recognition of malignant cells. Taken together our studies, althoughsupporting an important role for HLA-A2 antigen in immunesurveillance, raise the possibility that continued expression ofthis Class I antigen may be necessary for recruitment and/or

3169

on July 3, 2018. © 1991 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

HLA-A2 ANTIGEN IN HUMAN MELANOMA

specific stimulation of CTL. These studies also suggest thatrestoration of HLA-A2 antigen expression in melanoma cellsmay not be sufficient per se to trigger an effective cytotoxicresponse. Additional studies are necessary to investigate thefrequency of down-regulation of HLA-A2 as well as other HLAalíelesin melanoma and other tumors and the mechanismsresponsible for this antigenic loss.

REFERENCES

1. Golch. F., Unihlunl. J., Howland, K., Townsend. A., and McMichael, A.Cytotoxic T lymphocytes recognize a fragment of influenza virus matrixprotein in association with HLA-A2. Nature (Lond.). 326: 881-882, 1987.

2. Bjorkman. P., J.. Saper. M. A., Samraoui. B.. Bennett. W. S.. Strominger.J. L.. and Wiley, D. C. The foreign antigen binding site and T cell recognitionregions of class I histocompatibility antigens. Nature (Lond.), 329: 512-518,1987.

3. Bjorkman, P. J., Saper. M. A.. Samraoui, B., Bennett, W. S.. Strominger, J.L., and Wiley, D. C. Structure of the human class I histocompatibilityantigen, HLA-A2. Nature (Lond.). 329: 506-511. 1987.

4. Tanaka, K.. Yoshioka. T., Bieberich. T.. and Jay, G. Role of the majorhistocompatibility complex class I antigens in tumor growth and metastasis.Annu. Rev. Immunol. 6: 359-380, 1988.

5. Natali, P. G., Nicotra, M. R., Bigotti, A., Venturo. !.. Marcenaro. L..Giacomini, P.. and Russo. C. Selective changes in the expression of HLAclass I polymorphic determinants in human solid tumors. Proc. Nati. Acad.Sci. USA, «6:6719-6723, 1989.

6. Smith. M. E. F., Marsh, S. G. E., Bodmer, J. G., Gelsthorpe, K., andMinima. W. F. Loss of HLA-A,B,C alíeleproducts and lymphocyte function-associated antigen 3 in colorectal neoplasia. Proc. Nati. Acad. Sci. USA, 86:5557-5561.1989.

7. Wolfel, T., Klehmann. E., Muller, C., Schutt, K. H., Meyer, Z. B. K. H., andKniHli. A. Lysis of human melanoma cells by autologous cytolytic T cellclones. Identification of human histocompatibility leukocyte antigen A2 as arestriction element for three different antigens. J. Exp. Med., 170: 797-810,1989.

8. Darrow, T. L., Slingluff, C. L. J.. and Seigler, H. F. The role of HLA class Iantigens in recognition of melanoma cells by tumor-specific cytotoxic Tlymphocytes. Evidence for shared tumor antigens. J. Immunol., 142: 3329-3335. 1989.

9. Crowley, N. J., Slingluff. C. L. J., Darrow, T. L., and Seigler, H. F.Generation of human autologous melanoma-specific cytotoxic T-cells usingHLA-A2-malched allogeneic melanomas. Cancer Res., 50: 492-498, 1990.

10. Mayer, T.. Fuller. T.. Lazarovits, A., Boyle. L.. and Kurnick. J. Characterization of in vivo activated allospecific T lymphocytes propagated fromhuman renal allograft biopsies undergoing rejection. J. Immunol.. 134: 258-264, 1985.

11. Kurnick, J.. Kradin. R.. Blumberg, R., Schneeberger, E., and Boyle, L.Functional characterization of T lymphocytes propagated from human lungcarcinomas. Clin. Immunol. Immunopathol., 38: 367-380. 1986.

12. Ordonez, N. G., Xialong, J. L, and Hickey, R. C. Comparison of HMB-45monoclonal antibody and S-100 protein in the immunohistochemical diagnosis of melanoma. Am. J. Clin. Pathol., 90: 385-390. 1988.

13. Pandolfi, F.. Strong, D. M., Slease. R. B.. and Bonnard. G. D. Serologicaland functional characterization of human T-cell subsets. Clin. Exp. Immunol., 43: 319-325. 1981.

14. Pandolfi. F., Strong. D. M., Slease, R. B.. Smith. M. L.. Ortaldo, J. R.. andHerberman. R. B. Characterization of a suppressor T-cell chronic lympho-cytic leukemia with ADCC but not NK activity. Blood. 56: 653-660, 1980.

15. Vega, M. A., and Strominger, J. L. Constitutive endocytosis of HLA class Iantigens requires a specific portion of the intracytoplasmic tail that sharesstructural features with other endocytosed molecules. Proc. Nati. Acad. Sci.USA, 86: 2688-2692. 1989.

16. Wigler, M.. Pellicer. A.. Silverstein, S., and Axel, R. Biochemical transfer ofsingle-copy eucaryotic genes using total cellular DNA as donor. Cell, 14:725-731, 1978.

17. Gorman, C., Padmanabhan, R.. and Howard, B. H. High efficiency DNA-mediated transformation of primate cells. Science (Washington DC), 221:551-553, 1983.

18. Cathala, G., Savouret, J. F.. Méndez,B., West, B. L., Karin, M., Martial, J.A., and Baxter, J. D. A method for isolation of intact, translationally activeribonucleic acid. DNA, 2: 329-335. 1983.

19. Sambrook, J., Fritsch, E. F., and Maniatis, T. Molecular Cloning. A Laboratory Manual. Cold Spring Harbor. NY: Cold Spring Harbor LaboratoryPress, 1989.

20. Gattoni-Celli, S., Willett, C. G.. Rhoads, D. B., Simon, B., Strauss, R. M.,Kirsch, K., and Isselbacher, K. J. Partial suppression of anchorage-independent growth and tumorigenicity in immunodeficient mice by transfection ofthe H-2 class I gene H-2Ld into a human colon cancer cell line (HCT). Proc.Nati. Acad. Sci. USA, 85: 8543-8547. 1988.

21. Alitalo, K.. Schwab, M., Lin, C. C. Varmus, H. E., and Bishop, J. M.Homogeneously staining chromosomal regions contain amplified copies ofan abundantly expressed cellular oncogene (c-myc) in malignant neuroendo-crine cells from a human colon carcinoma. Proc. Nati. Acad. Sci. USA, 80:1707-1711. 1983.

22. Sood. A. K., Pereira, D., and Weissman, S. M. Isolation and partial nucleo-tide sequence of a cDNA clone for human histocompatibility antigen HLA-B by use of an oligodeoxynucleotide primer. Proc. Nati. Acad. Sci. USA, 78:616-620, 1981.

23. Kohler. B. H.. Sidwell, B.. DeMars, R., and Orr, H. T. Isolation of HLAlocus specific DNA probes from the 3' untranslated region. Proc. Nati. Acad.Sci. USA, 81: 5175-5179, 1984.

24. Salter, R. D.. Benjamin, R. J., Wesley, P. K. Buxton. S. E., Carre«, T. P.J.,Clayberger, C., Krensky, A. M., Normen!, A. M., Littman, D. R., andParham, P. A binding site for the T-cell co-receptor CDS on the alpha 3domain of the HLA-A2. Nature (Lond.), 345:41-46, 1990.

25. Hammerling, G. J., Klar, J. D., Momburg, F., and Moldenhauer. G. Theinfluence of major histocompatibility complex class I antigens on tumorgrowth and metastasis. Biochim. Biophys. Acta, 907: 245, 1987.

26. Natali, P., Nicotra, M. R., Cavaliere, R., Bigotti, A., Romano, G., Temponi,M., and Ferrane, S. Differential expression of intercellular adhesion molecule1 in primary and metastatic melanoma lesions. Cancer Res., 50: 1271-1278,1990.

27. Knuth. A.. Wolfel, T., Klehmann, E., Boon, T., and Meyer, Z. B. K. H.Cytolytic T-cell clones against an autologous human melanoma: specificitystudy and definition of three antigens by immunoselection. Proc. Nati. Acad.Sci. USA, 86: 2804-2808. 1989.

28. Clark. W. H. J.. Elder, D. E., Guerry, D., Braitman, L. E., Track, B. J.,Schultz, D.. Synnestvedt, M., and Halpern. A. C. Model predicting survivalin stage I melanoma based on tumor progression. J. Nati. Cancer Inst., 81:1893-1904, 1989.

29. Kradin, R., Kurnick, J., Lazarus, L., Preffer, F., Dubinett, S., Pinto, C.,Gifford, J., Davidson, E.. Grove, B., Callaham, R., and Strauss, H. Tumor-infiltrating lymphocytes and interleukin-2 in the treatment of patients withadvanced cancer. Lancet. /: 577-580, 1989.

30. Rosenberg, S. A., Packard, B. S., Aebersold, P. M., Solomon, D., Topalian,S. L., Toy, S. T., Simon, P.. Lotze. M. T., Yang, J. C., Seipp, C. A., Simpson,C., Carter, C., Bock, S., Schwartzentruber, D., Wei, J. P., and White, D. E.Use of tumor-infiltrating lymphocytes and interleukin-2 in the immunother-apy of patients with metastatic melanoma. N. Engl. J. Med., 319: 1676-1680. 1988.

31. Miescher, S., Whiteside, T. L., Moretta, L., and Von Flinder, V. Clonal andfrequency analyses of tumor-infiltrating T lymphocytes from human solidtumors. J. Immunol., /38:4004-4011, 1987.

32. de Vries, J. E., and Spits. H. Cloned human cytotoxic T lymphocyte (CTL)lines reactive with autologous melanoma cells. I. In vitro generation, isolation,and analysis to phenotype and specificity. J. Immunol., 132: 510-519, 1984.

33. Goulmy, E.. Kurnick, J., Hawes, G.. Blókland,E.. Pool, J., Schrier, P., Vander Wiel, E.. Boyle, L., Pandolfi, F., and Gattoni Celli, S. Lack of minorhistocompatibility antigen expression on melanocytes and melanomas. In:Proceedings of the 16th Annual Meeting of the American Society for Histocompatibility and Immunogenetics, Human Immunology, p. 30. New York:Elsevier. 1990.

34. Versteeg, R., Kruse, W. K. M., Plomp. A. C., Van, L. A., Stam, N. J.,Ploegh, H. L., Ruiter, D. J., and Schrier, P. I. Suppression of class I humanhistocompatibility leukocyte antigen by c-myc is locus specific. J. Exp. Med.,/ 70: 621-635. 1989.

3170

on July 3, 2018. © 1991 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

1991;51:3164-3170. Cancer Res   Franco Pandolfi, Lenora A. Boyle, Livio Trentin, et al.   and its Role in T-Cell RecognitionExpression of HLA-A2 Antigen in Human Melanoma Cell Lines

  Updated version

  http://cancerres.aacrjournals.org/content/51/12/3164

Access the most recent version of this article at:

   

   

   

  E-mail alerts related to this article or journal.Sign up to receive free email-alerts

  Subscriptions

Reprints and

  .pubs@aacr.orgDepartment at

To order reprints of this article or to subscribe to the journal, contact the AACR Publications

  Permissions

  Rightslink site. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC)

.http://cancerres.aacrjournals.org/content/51/12/3164To request permission to re-use all or part of this article, use this link

on July 3, 2018. © 1991 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from