ICANCER RESEARCH56. 6-20. January t. 9961

Advances in Brief

MAGE-1-specific Precursor Cytotoxic T-Lymphocytes Present among Tumorinfiltrating Lymphocytes from a Patient with Breast Cancer:Characterization and Antigen-specific Activation'

John F. Toso, Coreen Oei, Farshid Oshidari, James Tartaglia, Enzo Paoletti, H. Kim Lyerly, Sohel Talib, andKent J. Weinhold2

Departments of Pathology fJ. F. T.J and Surgery fC. 0.. H. K. L, K. J. WI, Duke University Medical Center, Durham, North Carolina 27710: Virogenetics, Inc., Troy,New York 12180 Ii. T., E. P.1: and Applied Immune Sciences, Inc., Santa Clara, California 95054 [F. 0., 5. Ti

Abstract

A potential target for development of tumor-specffic immunotherapeutic strategies is the MAGE-1 gene. We have utilized a recently developedrecombinant canarypox (ALVAC) virus vector containing the MAGE-1gene (vCP23S) to activate CTLs from a breast cancer patient bearing aMAGE-1@tumor. Tumor-infiltrating lymphocytes (TILs) obtained fromthe tumor of a patient were stimulated in vitrowith irradiated autologousperipheral blood mononuclear cells acutely infected with the vCP23Sconstruct. These TILs preferentially expanded approximately 6-fold overa 16-day culture period and specifically recognized an allogeneic trans.formed B-cell line acutely infected with a vaccinia-MAGE-1 recombinant

targeting vector (vP1l88) in the context of HLA-A2 and/or B7. TCR Vfianalysis of in vitro expanded T cells by a quantitative multiprobe RNaseprotection assay revealed preferential expansion of TCR V@J63 andVfi6.4. In addition, homologous T-cell receptor fi CDR3joining sequenceswere found in the in vitro stimulated cultures. These results suggest thattumor antigen-specific, MHC-restricted CTLs may be derived from pre.

cursor CTLs present in TILs obtained from patients with MAGE-l@tumors by in vitro stimulation with recombinant avipox MAGE-1 virusinfected autologous cells. Collectively, these findings provide a rationalefor tumor-associated antigen-based immunization as a means of activatingprecursor CTLS residing in patients with tumors expressing defined hi.mor-associated antigens such as MAGE-1.

Introduction

Molecularly defined, conserved TAAs3 are potential immunotherapeutic targets for a number of human neoplasms. A potential vaccinefor inducing tumor-specific effectors could consist of a canarypoxvirus-based vector (ALVAC) expressing one or more defined tumorantigens. ALVAC would provide a safe and efficacious delivery ofTAAs because these recombinants do not replicate in mammaliancells but are known to elicit strong cellular immune responses againstencoded heterologous gene products (1, 2). Recently, the MAGE-1gene was inserted into an ALVAC vector. The MZ2-E and MZ2-Bbantigens encoded by the MAGE-1 gene are recognized by CTLs fromHLA-Al@ and HLA@Cw*l6Ol ±melanoma patients, respectively (3,4). MAGE-1 is a nonmutated gene not expressed in normal cells otherthan testis, but it is detected in approximately 40% of melanomas, as

Received 10/4/95; accepted I 1/14/95.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 in accordance with18 U.S.C. Section 1734 solely to indicate this fact.

I This work was supported by NIH Awards l-ROl-CA58005 and l-P50-CA68438

(Duke University SPORE in Breast Cancer) and NIH Training Grant 5-T32-A107392(J. F. T).

2 To whom requests for reprints should be addressed. at Department of Surgery, Box

2926, Duke University Medical Center, Durham, NC 27710-2926.3 The abbreviations used are: TAA, tumor-associated antigen; TIL, tumor-infiltrating

lymphocyte; PBMC, peripheral blood mononuclear blood cell; TCR, T-cell receptor;RT-PCR, reverse transcription-PCR; BLCL, B-lymphocyte cell line; CDR3, third complementary-determining region; IVS, in vitro stimulation; IL-2, interleukin 2; rIL-2,recombinant IL-2; SR. spontaneous release; MR. maximum release.

well as other tumors of nonmelanoma origin, notably breast carcinoma (5). With its relatively high frequency of expression in varioustumors and its proven immunogenicity, MAGE- 1 represents a potential target for immunotherapy. In this study, we demonstrate for thefirst time MAGE-l-specific CTL activity in ilLs obtained from aMAGE-l-expressing breast tumor. After in vitro stimulation withALVAC-MAGE- 1, the CTL recognized an HLA-A2- and/or B7-expressing target acutely infected with a vaccinia-MAGE-l recombinant. Also, analysis of TCR-Vp chain expression by the effector cellsrevealed an oligoclonal population of T cells (Vf36@) expressinghomologous TCR-f3 CDR3 junctional sequences. Collectively, thesedata suggest that antigen-specific CTLs can be readily generated fromClip residing within tumor tissue by in vitro stimulation with transient expression vectors expressing molecularly defined TAAs.

Materials and Methods

Patient History. The patientis a 60-year-oldfemale who was first diagnosed with breast cancer in 1993. In November 1993, she underwent a right

modified mastectomy, which revealed infiltrating ductal carcinoma histologi

cal grade III, with focal invasion of the lymphatics by carcinoma. Four of 15axillary lymph nodes were positive for metastatic disease. Subsequently, shereceived cyclophosphamide, 5-fluorouracil, and methotrexate for 4 months andlocal radiation therapy to the area of the tumor. One year from the developmentof her most recent cancer, she has no evidence of local or systemic disease.

Isolation of TILs. Tumor tissues were minced manually into 1-mm3 fragments and then enzymatically digested in 0.01% hyaluronidase (Sigma Chemical Co., St. Louis, MO), 0.002% DNase (Sigma), and 0.1% collagenase type

III (Atlanta Biological, Atlanta, GA) in RPM! 1640 (GIBCO) with 100 @.tg/m1gentamicin overnight on a magnetic stirrer. The cell suspension was washed inHBSS and passed through a metal mesh. CD3@TILs were isolated by attachment to anti-CD3 MicroCELLector T-25 flasks (Applied Immune Sciences,Santa Clara, CA) according to the manufacturer's recommendations. The cellswere cultured in the capture flasks for 2 days in RPM! 1640 containing 20%(v/v) FCS, recombinant IL-2 (100 units/ml; Hoffman-La Roche, Inc., Nutley,NJ), and gentamicin sulfate (100 @ag/ml).

ALVAC-MAGE-I-stimulated TIL Culture. Autologous PBMCs wereinfected with the ALVAC-MAGE-1 construct vCP235 (see description below)at a multiplicity of infection of S in RPMI 1640 containing 20% (v/v) FCS at

37°Cin 5% CO2. After 1.5 h, the acutely infected PBMCs were washed andirradiated with 6000 rads. For in vitro stimulation, CD3@TILs and acutelyinfected (vCP235), irradiated, autologous PBMCs were cultured at a respond

er:stimulator ratio of 5:1 in RPMI 1640 containing 20% FCS and 100 units/ml

recombinant IL-2 in the presence or absence of 100 ng/ml anti-CD28. Cellswere maintained at a concentration of 1 X 106cells/ml. An unstimulated CD3@TIL population was cultured in parallel in RPM! 1640containing 100 units/mIrIL-2. Both cultures were restimulated 2 weeks later with 100 ng/ml anti-CD3,100 ng/ml anti-CD28, and 1.5 X i05 cells/mi irradiated allogeneic PBMCs.

Cytotoxicity Assay. One million EBV-immortalized BLCLs were infectedfor 90 mm with either the control vaccinia targeting vector (vPl 170) or thevaccinia-MAGE-l construct (vPl 188) using a multiplicity of infection of 5,

and then labeled with 100—200 @aCiof sodium chromate (5tCr; DuPont,

16

Research. on November 24, 2020. © 1996 American Association for Cancercancerres.aacrjournals.org Downloaded from

MAGE-l-SPECIFIC CTLs IN A BREAST CANCER PATIENT

Wilmington, DE). The cells were incubated overnight in I ml of RPM! 1640

at 37°C in a 5% CO2 humidified incubator. After incubation, the BLCLs were

washed, counted, and plated at a concentration of 5 X l0@ viable cells/well in

a round-bottomed 96-well plate (Costar). Stimulated TIL populations wereused as effectors against the BLCL targets in a 4-h assay. SR and MR controlsconsisted of targets plus either medium (for SR) or 0.5% Triton X-lOO (forMR). Percentage of specific lysis was calculated as:

(cpm experimental release— cpm SR) >< l@

(cpm MR —cpm SR)

Construction of ALVAC-MAGE-1 Recombinants. The MAGE-l Sequences encoding the MZ2-E human melanoma rejection antigen (3) werederived from plasmid pTZ18RMAGE-l (provided by T. Boon, Ludwig Institute, Brussels, Belgium). MAGE- I coding sequences were fused to the vaccinia virus H6 early/late promoter element (6) and inserted into a generic

ALVAC (CPpp; Ref. 7) insertion plasmid. Standard in vitro recombinationassays (8) were performed using the resultant plasmid pMAWO36 as donor

DNA and ALVAC as rescue virus. MAGE-1 gene expression was confirmed in

vCP235-infected CEF by FACScan analysis using a MZ2-E-specific antibody

preparation (kindly provided by Dr. Lloyd Old, Ludwig Institute for CancerResearch, New York, NY).

RNA Extraction, cDNA Synthesis, and PCR. RNA of fresh-frozenhuman breast carcinoma samples was prepared using RNAzol B (CinnalBiotecx)according to the manufacturer's instructions. cDNA synthesis was performed

with 2 i@g of total RNA. PCR amplification was performed in a total volume

of 100 p@lcontaining one-fifth volume of the cDNA solution, 10 pJ of lOXPCR buffer [I X PCE buffer = 10 mM Tris-HC1 (pH 8.3)-SO mM KC!-l.5 mMMgCl2-O.OOl% (w/v) gelatin] (Perkin Elmer Cetus), 2 pAeach of 10 mM dNTP,1pi each of 80 @.LMsolutions of sense CHO-14 primer (CGGCCGAAGGAACCTGACCCAG) and antisense CHO-l2 primer (GCTGCIAACCCTCACTGGG1'TGCC) (5), and 2.5 units of Taq polymerase (Perkin Elmer Cetus).Amplification was performed at 94°C for 2 mm, followed by 32 cycles at 94°C

for 1 mm and 72°C for 3 mm. Amplification of 3-actin eDNA was also

performed to check integrity of the RNA sample. The PCR products were

resolved on an agarose gel and visualized by ethidium bromide staining.Multiprobe RNase Protection Assay. Preparationof the 29 humanV(3

antisense RNA probes (pooled into 3 probe sets) and RNase protection assay

were performed as described by Baccala et a!. (9). Hybridization of 1—2@.tgofRNA with radiolabeled Cj3probe and each of the probe sets was performed at56°Cfor 12—16h. Unhybridized probes and target RNA were digested withRNase A and RNase T. Purified protected probe-target mRNA duplexes were

electrophoresed in standard sequencing gels and analyzed by Phosphorlmager(Molecular Dynamics). The quantitative analysis of the Vf3and Cj3 transcriptlevels were performed using ImageQuant software (Molecular Dynamics).Results are expressed as the percentage of the sum of all 29 V@3.

Cloning and V@36,CDR3 Sequence Analysis. The PCR-amplifiedV@36products were cloned directly into pCRII vector (Invitrogen), and the inserts

were sequenced using the Taq DyeDexoy Terminator Cycle Sequencing kit(Applied Biosystems) using 3.2 pmol 17 or SP6 primer (Promega). Purified

reaction samples were run on a 373A DNA sequencer (Applied Biosystems),and the sequence data were analyzed by SeqEd (Applied Biosystems) andcompared with GenBank sequences.

Results

Presence of MAGE-1 mRNA in Primary Breast Carcinomas.Total cellular RNA was extracted from breast tumor tissue freshfrozen immediately after surgical resection. The presence of MAGE- 1mRNA was evaluated by RT-PCR with oligonucleotide primers thatare unique for the MAGE-l sequence (5). A panel of 56 fresh frozenbreast tumor samples, as well as the human melanoma cell line DM150, were tested for MAGE-1 expression. The DM 150 cell line(generously provided by Dr. H. Seigler, Duke University MedicalCenter) is an HLA-Al @,A2@ melanoma cell line that has been shownpreviously to express MAGE-l (10).

Fig. 1 shows the MAGE-l RT-PCR analysis of tumor cells fromtwo breast cancer patients (D. M. and L. A.), as well as the DM 150cell line. The presence of a 42l-bp fragment corresponding to the

4.—.--.616bp600

400

200

I 11 1

MAGE-1 ActinFig. 1. RT-PCR analysis of MAGE-l expression by the control DM150 melanoma cell

line and tumor tissue from breast cancer patients D. M. and L. A. (3-actin mRNA detectionwas used as a control to verify the integrity of the RNA samples.

17

@J- @# -

‘—QJ a) I—@

bp M123456

4.— 419bp

MAGE-l PCR product was detected in samples from the DM 150 cellline (Fig. 1, Lane 1) and patient D. M. (Fig. 1, Lane 2) but not patientL. A. (Fig. 1, Lane 3). A total of 11 (20%) breast tumor samples testedpositive for MAGE-l, a result which is consistent with that reportedpreviously (5). The integrity of the RNA samples used for RT-PCRwas ascertained by amplifying the (3-actin mRNA (Fig. 1, Lanes4—6).Because breast tumor cells from patient D. M. were shown tobe MAGE-l @,TILs isolated from the tumor tissue of the patient wereselected for further study.

In Vitro Stimulation of MAGE-1 CTL Activities from BreastTILs. We next examined the capacity of the recombinant ALVACMAGE-l vector construct (vCP23S) to stimulate CTL activity in theTIL population derived from a breast tumor expressing MAGE-1mRNA. Breast tumor ilLs from patient D. M. (see above) were

isolated and activated for 48 h by capture onto an anti-CD3 microCELLector flask. After 2 days of culture in the capture flask in thepresence of 100 units/ml of rIL-2, the activated TILs were collected,washed, and divided into 3 aliquots. One of the portions of ilLs wascultured for an additional 2 weeks in the absence of any antigenspecific stimulation. The remaining two aliquots were cocultured inthe presence of vCP235-infected, irradiated, autologous PBMCs at aresponder:stimulator ratio of 5: 1. One of the ALVAC-MAGE-l cultures was additionally supplemented with anti-CD28 monoclonal antibody (100 ng/ml, final concentration). All three cultures contained100 units/ml of rIL-2. Cellular proliferation was monitored by repeated determination of viable cell content throughout the 2-weekculture period. The results, presented in Fig. 2, revealed little or no

Research. on November 24, 2020. © 1996 American Association for Cancercancerres.aacrjournals.org Downloaded from

MAGE-I-SPECIFIC CTLs IN A BREAST CANCER PATIENT

—..-— Tll@ alone

......... TIL+ALVAC/MAGE-1

- - £- - TIL+ALVACJMAGE-l+a-CD28

4 6 8 10 12 14 16 18

Days in CultureFig. 2. Kinetics of antigen-specific expansion of TILs from patient D. M. CD3@ ilLs

were cultured with IL-2 (TILs alone) or in the presence of IL-2 and vCP235 (ALVACMAGE-l)-infected autologous PBMCs (±anti-CD28) at a responder:stimulator ratio of5:1.

expansion of viable TILs in the absence of antigen-specific stimulation(i.e. , TILs alone). Vector-stimulated cultures, however, showed substantial expansion of viable cells, resulting in nearly a 3-fold increase in thevCP235-stimulated culture and over a 6-fold increase in the culturecostimulated with vCP235 and anti-CD28 monoclonal antibody. Cellgrowth in the costimulated culture was progressive throughout the2-week period, whereas expansion in the vCP235 alone culture peaked byday 8 and remained at plateau levels thereafter.

At the end of the initial 2-week culture period, all 3 cultures werefurther expanded by stimulation with 100 ng/ml of anti-CD3 or 100ng/ml of anti-CD28 in the presence of 1.5 X l0@irradiated, allogeneicPBMCs/ml. On day 26, 10 days after the secondary costimulation,functional characteristics of the expanded TIL populations were cxamined. At this time point, the vCP235 plus anti-CD28 TIL cultureconsisted of 87% CD8@ T cells, 6.3% CD4@ T cells, and 1% CDl6@/CD564 natural killer cells (data not shown).

The functional activity of the ilL cultures was examined in thecontext of a series of cytotoxicity assays. Despite repeated attempts,an autologous breast tumor cell line could not be established. Toevaluate the cytotoxic reactivity present within the MAGE-1-stimulated TIL cultures, we used allogeneic transformed BLCLs sharingpartial HLA homology with the breast cancer patient (HLA-A2, 3; B7,16; Bw6; Cw7). These cells were acutely infected with either control(vPl 170) or MAGE-l-expressing (vPll88) targeting vectors. Asshown in Fig. 3, ilLs from the vCP23S plus anti-CD28 culturesexhibited lytic activity against donor 1 vPl188-infected BLCLs butfailed to lyse donor 2 BLCLs infected with the same vector. Althougha small amount of anti-vaccinia CTL reactivity was evident by thelysis of vPl 170-infected BLCLs, the increased level of lysis againstthe MAGE-l-expressing targets confirms the presence of MAGE-lspecific CTLs within the expanded TIL population. Unstimulated TILcultures failed to lyse the target cells of either donor (data not shown).The HLA alleles shared by the patient and donor 1 (and which are notexpressed by donor 2 cells) include HLA-A2 and HLA-B7, either ofwhich could act as restricting elements for the Cli activity depictedinFig.3.

TCR vp Gene Usage. To determinewhether a restrictedrepertoire of TCR genes was expressed in breast TILs after in vitrostimulation with the vCP235 vector, we quantitated the V@gene usageof ilLs (±ALVAC-MAGE-l stimulation) and PBMCs with a mul

tiprobe RNase protection assay. We found TCRs V/36.3 and VJ36.4 tobe dominant repertoires in cultured breast TILs as compared toPBMCs. This difference in Vf3 repertoire expression was furtherenhanced from day 7 to day 29. ilLs stimulated with vCP235 andanti-CD28 showed maximum expression of V@36.3and Vf36.4 (>80%of total repertoire; Fig. 4). To extend this analysis, we cloned andsequenced the PCR-amplified V@36cDNA of the day 29 stimulated(MAGE-l + anti-CD28) ilLs. Of 21 clones sequenced, 9, 8, and 4were V@6.1, V@36.3,and V@6.4, respectively (Table 1). Analysis ofV/NDN/J junctional region sequences indicates that for five differentsequences utilized by these clones, there exists two or more clonessharing identical sequences. Moreover, a number of these clones shareconserved amino acids (Leu at position 1 and Gly or Ala at position5) in the NDN region, indicating clonal expansion of T cells.

Discussion

The results presented in this communication represent the firstreport of MAGE-l-specific CU reactivities outside of the realm ofmalignant melanoma. Antitumor immune responses have been detected against a number of human cancers (1 1, 12) and may serve asa basis for the development of future therapeutic strategies. Amongthe cellular immune responses to human tumors are CTLs that recognize TAAs in association with HLA molecules on tumor cells (13,14). MAGE-1 is an example of a molecularly defined, nonmutatedgene that encodes tumor antigens recognized by CTL clones derivedfrom patients with malignant melanoma (3, 4). Because MAGE-lexpression has been detected in a variety of different human tumors(5) and has been shown to be immunogenic, MAGE-l may represent

the prototypic TAA for evaluation in the context of “cancervaccines.―Our working hypothesis for the present studies was that TIL pop

ulations derived from fresh tumor tissue expressing a defined set ofTAAs should contain immune effector reactivities, perhaps at theprecursor level, specific for those antigens. This hypothesis is basedon mounting evidence that patients do recognize and respond to TAAsexpressed by their own growing tumors. For example, Disis et a!. (15)have reported that women bearing HER-2/neu@ breast tumors havedemonstrable humoral and cellular anti-I-IER-2/neu reactivities. Likewise, a recent report from the same group documented circulatingCD4@ T-cell responses to substituted k-ras peptides in patients with

tumors expressing mutated forms of k-ras. (16). Cytotoxic reactivities, other than at the clonal level, have been more difficult to detect.This could be due in part to the relatively low frequency of Clip inthe peripheral circulation. We reasoned that if such activities were

—.-- BLCL

@ BLCL + Vac

-.4.-. BLCL + Vac/MAGE-1

20:1 iO@i 5:1 20:1 10:1 5:1

E:T Ratio

Fig. 3. Cytolysis of allogeneic BLCLs infected with either control (vPll7O) orMAGE-1(vPl188)recombinantvacciniatargetingvectors.Effectorsweretakenon day29 from Tft. cultures stimulated with vCP235-infected autologous PBMCs in the presenceof IL-2. BLCL target HLA homology with patient D. M. included: A2,3; B7; Cw7 fordonor I and A3; Bw6; Cw7 for donor 2.

__.3 £

k@ ‘

,/ ........,. ......

Donor 1 Targets40@ HLA: A2, A3, B7, CW7

30

20@

Donor 2 TargetsSharedHLA: A), BW6, CW7

0

(5

18

40

@ 100

Research. on November 24, 2020. © 1996 American Association for Cancercancerres.aacrjournals.org Downloaded from

Table 1 CDR3sequences ofbreast (patient D. M.) TILs stimulatMAGE-I and a-CD28 (29 days)ed

with ALVAC

No. of clonesV@ N-Do-NJf3V@fN-D@-N/J@1/21

1/2!1/211/213/212/21YLCASS

SGQAA f EAFFGQGYLCASS LVAGAV I TEAFFGQGYLCASS LIAGTG I TEAFFGQGYLCASS LTPSG@J EQYFGPGYLCASS LQGA@ ANVLTFGAGYLCASS QDTS@@@ TDTQYFGPG6.1/1.1/1.1

6. 1/1.1/1.16.1/1.1/1.16.1/2.1/2.76.1/1.1/2.6

6.1/2.1/2.32/21YLCASSLGAGAATEAFFGQG6.3/1.1/1.11/21YLCASSARTGAIQPQIIFGDG6.3/1.1/1.51/21YLCASSFGSSTNYGYTFGSG6.3/2.1/1.21/21YLCASSPGTVNEQYFGPG6.3/1.1/2.71/21YLCASSSQAGGYYEQYFGPG6.3/1.1/2.71/21YLCASSLGPLGHVPYEQYFGPG6.3/1.1/2.71/21YLCASSFSSGGALETQFGPG6.3/2.1/2.52/21

2/21YRCASSLDPP@ ELFFGEG

YRCASS LARD@1@] TEAFFGQG6.4/2.1/2.26.4/2.1/1.1

MAGE-I-SPECIFIC CTLs IN A BREAST CANCER PATIENT

45

40

35@ -

30@ -

25 - -

20 - -

15@

10@ -

5.

0-El i@ I I

I I I I I

Ir) @- ‘- c%J C@)cth@ c6c6@d

VB

(f) i

c@i@ iticth‘- .0 v•@ ‘-

‘- e- (f) ‘@ 10

(%J

c4J ‘@10 C')@ui ‘0 10 CD CD

Fig. 4. TCR-Vf3 gene expression in patient D. M. Relative gene usage in PBMCs (day 0), ilLs (day 7) and ilLs stimulated with vCP235 and anti-CD28 (day 29). The normalizedvalue of each V@ was expressed as the percentage of the sum of all 29 V@.

likely, a reflection of the relative inefficiency of vCP23S-infectedwhole PBMCs to serve as antigen-presenting cells. Although therecombinant poxvirus construct is pancytotropic and infects all populations comprising the PBMCs, the only infections that are relevantto Cli activation may be infection of “authentic―antigen-presentingcells such as blood monocytes, which represent a relatively minorsubpopulation of PBMCs. Other more abundant populations such asCD4@ and CD8@ lymphocytes are capable of processing and presenting specific antigens, but they generally fail to express requisite

costimulatory molecules such as CD8O and CD86 (17), which areligands for the CD28 molecule on effector T cells. The anti-CD28monoclonal antibody may supplant the need for CD8O/CD86 expression and significantly augment an otherwise suboptimal antigen

specific stimulation.

The determinants recognized by the vCP235-stimulated CTLs remain to be identified. The finding that lytic activity was restricted byHLA-A2 and/or HLA-B7 alleles strongly suggests that the targetepitopes are different from the MZ2-E determinants that have beenshown previously to be presented in the context of HLA-Al andHLA-Cwl6Ol (3, 4). Lack of absolute fidelity in peptide binding toHLA determinants has been demonstrated (18). Celis et a!. (19) haverecently described several peptide analogues of MAGE-l determinants that bound with significant affinity to multiple HLA moleculessuch as HLA-Al, HLA-A2.l, HLA-A3, HLA-Al 1, or HLA-A24 and,thereby, served as putative Cli epitopes. It is, therefore, reasonableto believe that multiple HLA elements may be operative in restrictingthe CTL reactivity against the numerous target determinants encodedby the MAGE-1 gene.

present in cancer patients, then their overall frequencies may be

greatest within the actual tumor bed by virtue of migration to and in

situ sensitization within the growing primary tumor. The MAGE-l

specific CTL reactivities detected in TILs after stimulation withvCP235-infected autologous PBMCs are consistent with the presenceof CTLp capable of recognizing as yet undefined MAGE-l determinants. It is highly unlikely that the day 26 CTL activities reflect a denovo CTL response because numerous attempts to elicit MAGE-1 -specific CTL responses from normal donor PBMCs, using the iden

tical stimulation strategy used in these studies, were uniformly

unsuccessful (data not shown).

The anti-CD28 effect seen in the in vitro expansion studies is, most

19

PBMCd.0

@ TILd.7

. TIL+MAGE1+antiCD28

Iillli-ft@riiiii iI@@f@4

Research. on November 24, 2020. © 1996 American Association for Cancercancerres.aacrjournals.org Downloaded from

MAGE-I-SPECIFIC CTLs IN A BREAST CANCER PATIENT

Recent studies have shown that ilLs can express a restrictedrepertoire of TCR-V/3 genes that recognize TAA determinants (20). Inthe present study, TCR V(3 analysis revealed an over representation ofV/36-restricted T cells in captured and expanded breast TILs ascompared to PBMCs. This oligoclonal T-cell response was furtherenhanced by costimulation with vCP235 vector and anti-CD28, whichresulted in the appearance of functional Cli activity. Sequenceanalysis of the VDJ junctional region revealed the presence of recurrent transcripts. Additionally, the overall length of the VDJ junction inthe majority of the cDNA clones was limited to five or six aminoacids, suggesting structural constraints on the V@3sequence. This isfurther supported by the presence of conserved amino acid residues(Leu or Gly) in the NDN region. Collectively, these observationsstrongly suggest antigen-specific expansion of T cells. Similarly,others have recently identified CDR3 length and sequence restrictionsin TCR Vj3 chains in the pigeon cytochrome C-specific T cells, as well

as CD8@ cells responsive to human immunodeficiency virus (21, 22).One caveat must be raised in conjunction with the present studies.

We have used a recombinant poxviruslMAGE-l vector (vCP235) toactivate Clip, the lytic reactivity of which was ultimately assessed onpartially HLA-matched, allogeneic BLCLs acutely infected with another recombinant poxviruslMAGE-1 targeting vector (vPll88). Although the overall HLA-restricted lysis of MAGE-l-expressingBLCLs was sufficiently above the anti-vaccinia background lysis ofcontrol targets to be indicative of MAGE-l-specific Cli reactivity(Fig. 3), a number of issues remain. The relatively small number ofTILs that were harvested from the anti-CD3 capture flask at day 2 didnot permit parallel stimulations with control ALVAC constructs.Thus, it is not clear what contribution the non-MAGE-l elements ofthe recombinant vector may have made to the overall cellular activation and expansion. Concomitant anti-vaccinia reactivities could conceivably facilitate the expansion of pre-existing anti-MAGE reactivities. Second, and more important, we did not have the opportunity totest our effector populations against bona fide tumor targets thatnaturally express MAGE-l determinants. Vaccinia-infected BLCLsmay not represent physiologically relevant tumor targets by virtue oftheir processing and overexpression of target antigens. Althoughautologous or partially matched breast tumor cell lines will mostlikely not be available for analysis, a panel of MAGE-l-expressingmelanoma cell lines is being assembled for these purposes. Suchanalyses represent an important aspect of future planned studies.

In conclusion, this study highlights an experimental approach thatcould have significant implications in the development and application of future cancer vaccine strategies. A clinical approach aimed atfirst defining the TAA expression by the tumor of an individualpatient, and selecting an antigen-specific intervention based on thatpattern of TAA expression, would represent an important first step inevaluating the effectiveness of immune-based therapies for cancer.Whether the present findings in which a patient with a MAGE-l@tumor also had TAA-specific Clip present within her TIL populationrepresent a paradigm for other human tumors remains to be examinedin the context of future preclinical studies. Such approaches would, ofnecessity, have to be combined with more conventional treatmentregimens, especially in cases in which there is an extensive tumorburden.

20

Acknowledgments

The authors thank Dr. Thierry Boon for providing the MAGE-l cDNAconstruct and for helpful discussion.

References

1. Taylor, J., Weinberg, R., Tartaglia, J., Richardson. C., Alkatib, G., Briedis, D., Appel,M., Norton, E., and Paoletti, E. Non replicating viral vectors as potential vaccines:Recombinant canarypox virus expressing measles virus fusion (F) and hemagglutinin(HA) glycoproteins. Virology, 187: 321—328,1992.

2. Cadoz M., Strady, A., Meigier, B., Taylor, J., Tartaglia. J., Paoletti, E., and Plotkin,S. Immunisation with canarypox virus expressing rabies glycoprotein. Lancet, 339:1429—1432,1992.

3. Traversed, C., van der Bruggen, P., Luescher. I. F.. Lurguin. C., Chomez, P., Van PcI,A., Dc Plaen E., Amar-Costesec, A., and Boon, T. A nonapeptide encoded by humangene MAGE-I is recognized an HLA-Al by Cii.. directed against tumor antigenMZ2-E. J. Exp. Med., 176: 1453-1457, 1992.

4. van der Bruggen, P., Szikora. J. P., Boel. P., Wildmann, C., Somville, M., Sensi, M.,and Boon, T. Autologous cytolytic T lymphocytes recognize a MAGE-l nonapeptideon melanomas expressing HLA@Cw*I60l*. Eur. J. Immunol., 24: 2134—2140, 1994.

5. Brasseur, F., Marchand, M., Vanwijck, R., Herin, M., Lethe, B., Chomez, P., andBoon, T. Human gene MAGE-I, which codes for a tumor rejection antigen, isexpressed by some breast tumors. mt. i. Cancer., 52: 839—841, 1992.

6. Perkus, M. E., Limbach, K., and Paoletti, E. Cloning and expression of foreign genes

in vaccinia virus using a host range selection system. J. Virol., 63: 3829—3836, 1989.7. Tartaglia, J., Perkus, M. E., Taylor, J., Norton, E. K., Audonney, J. C., Cox, W. I.,

Davis, S. W., Van Ocr Hoeven, J., Meignier, B., Riviere, M., Languet, B., andPaoletti, E. NYVAC. A highly attenuated strain of vaccinia virus. Virology, 188:217—232,1992.

8. Perkus, M. E., Kauffman, E. B., Taylor. J., Mercer, S., Smith, D., Van Der Hoeven,J., and Paoletti, E. Methodology of using vaccinia virus to express foreign genes intissue culture. J. Tiss. Cult. Methods, 15: 72—81,1993.

9. Baccala, R., Kono. D. H., Walker, S., Balderas, R. S., and Theofilopoulos, A. N.Genomically imposed and somatically modified human thymocyte V@ gene repertoire. Proc. Natl. Acad. Sci. USA, 88: 2908—2912, 1991.

10. Ding, M., Beck, R. J., Keller, C. J., and Fenton, R. G. Cloning and analysis ofMAGE-1 related genes. Biochem. Biophys. Res. Commun., 202: 549—555,1994.

I I. Yasumura, S., Hirabayashi, H., Schwartz, D. R., Toso, J. F., Johnson, J. T., Herberman, R. B., and Whiteside, T. L. Human cytotoxic T-cell lines with restrictedspecificity for squamous cell carcinoma of the head and neck. Cancer Res., 53:1461-1468,1993.

12. Topalian, S. L., Soloman, D., and Rosenberg, S. A. Tumor-specific cytolysis bylymphocytes infiltrating human melanoma. J. Immunol., 142: 3714—3725, 1989.

13. Darrow, T. L., Slingluff, C. L., and Seigler, H. F. The role of class I antigens inrecognition of melanoma cells by tumor-specific cytotoxic T lymphocytes: evidencefor shared tumor antigens. J. Immunol., 142: 3329—3335, 1989.

14. Kawakami, Y., Zakut, R., Topalian, S. L., Stotter, H., and Rosenberg, S. A. Sharedhuman melanoma antigens: recognition by tumor-infiltrating lymphocytes in HLAA2.l-transfected melanomas. J. Immunol., 148: 638-643, 1992.

15. Disis, M. L., Calenoff, E., McLaughlin, G., Murphy, A. E., Chen, W., Groner, B.,Jeshke, M., Lydon, N., McGlynn, E., Livingston, R. B., Moe, R., and Cheever, M. A.Existent T cell and antibody immunity to HER-2/neu protein in patients with breastcancer. Cancer Res., 54: 16—20,1994.

16. Qin, H., Chen, W., Takahashi, M., Disis, M. L., Byrd. D. R., McCahill, L., Bertram,K. A., Fenton, R. G., Peace, D. J., and Cheever, M. A. CD4@ T-cell immunity tomutated ras protein in pancreatic and colon cancer patients. Cancer Res., 55:2984—2987, 1995.

17. Harding, F. A., and Allison, J. P. CD28-B7 interactions allow the induction of CD8@cytotoxic T-lymphocytes in the absence of exogenous help. J. Exp. Med., 177:1791—1796,1993.

18. Townsend, A., Rothbard, J., Gotch, F., Bahadur, G., Writh, D., and McMichael, A.The epitope of influenza nucleoprotein recognized by cytotoxic T lymphocytes can bedefined with short synthetic peptides. Cell, 42: 959—968, 1986.

19. Celis, E., Fikes, J., Wentworth, P., Sidney, J., Southwood, S., Maewal, A., DelGuercio, M-F., Sette. A., and Livingston, B. Identification of potential CTL epitopesof tumor-associated antigen MAGE- I for five common HLA-A alleles. Mol. Immunol., 31: 1423—1430,1994.

20. Gaudin, C., Dietrich, P. Robache, S., Guillard, M., Escudier, B., Lacombe, M. T.,Kumar, A., Triebel, F., and Caignard, A. In vivo local expansion of clonal T cellsubpopulations in renal cell carcinoma. Cancer Res., 55: 685—690, 1995.

21. McHeyzer-Williams, M. G., and Davis, M. M. Antigen-specific development ofprimary and memory T cells in vivo. Science (Washington DC), 268: 106—111, 1995.

22. Pantaleo, G., Demarest, J. F., Soudeys, H., Graziose, V., Deals, F., Adelsberger,J. W., Borrow, P., Saag, M. S., Shaw, J. S., Sekaly, R. P., and Fauci, A. S. Majorexpansion of CD8@ T cells with a predominant V@usage during the primary immuneresponse to HIV. Nature (Lond.), 370: 463—467, 1994.

Research. on November 24, 2020. © 1996 American Association for Cancercancerres.aacrjournals.org Downloaded from

1996;56:16-20. Cancer Res   John F. Toso, Coreen Oei, Farshid Oshidari, et al.   Breast Cancer: Characterization and Antigen-specific Activationamong Tumor-infiltrating Lymphocytes from a Patient with MAGE-1-specific Precursor Cytotoxic T-Lymphocytes Present

  Updated version

  http://cancerres.aacrjournals.org/content/56/1/16

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/56/1/16To request permission to re-use all or part of this article, use this link

Research. on November 24, 2020. © 1996 American Association for Cancercancerres.aacrjournals.org Downloaded from