gp100/pmel17 and tyrosinase encode multiple epitopes recognized by Th1-type CD4+T cells

CD4+ T cells modulate the magnitude and durability of CTL responses in vivo, and may serve as effector cells in the tumour microenvironment. In order to identify the tumour epitopes recognized by tumour-reactive human CD4+ T cells, we combined the use of an HLA-DR4/peptide binding algorithm with an IFN-γ ELISPOT assay. Two known and three novel CD4+ T cell epitopes derived from the gp 100/pmel17 and tyrosinase melanocyte-associated antigens were confirmed or identified. Of major interest, we determined that freshly-isolated PBMC frequencies of Th1-type CD4+ T recognizing these peptides are frequently elevated in HLA-DR4+ melanoma patients (but not normal donors) that are currently disease-free as a result of therapeutic intervention. Epitope-specific CD4+ T cells from normal DR4+ donors could be induced, however, after in vitro stimulation with autologous dendritic cell pulsed with antigens (peptides or antigen-positive melanoma lysates) or infected with recombinant vaccinia virus encoding the relevant antigen. Peptide-reactive CD4+ T cells also recognized HLA-DR4+ melanoma cell lines that constitutively express the relevant antigen. Based on these data, these epitopes may serve as potent vaccine components to promote clinically-relevant Th1-type CD4+ T cell effector function in situ. http://www.bjcancer.com © 2001 Cancer Research Campaign


Peptide selection and synthesis
The protein sequences of the melanoma associated antigens gp 100/pmel17 and tyrosinase were obtained from GENBANK and analysed for HLA-DR*0401 binding peptides using a neural network algorithm (Honeyman et al, 1998). High scoring nine amino acid long sequences were typically extended by three amino acids on either flank using the genomic corresponding sequences. Alternatively, if high scoring nine amino acid long sequences were found to overlap, the longer overlapping sequences were synthesized, with amino acid extensions added to the end(s) of putative DR4-binding sequence(s). Overall, peptides were 15-23 amino acids in length (Table 1) and were synthesized by FMOC chemistry by the University of Pittsburgh Cancer Institute's (UPCI) Peptide Synthesis Facility (Shared Resource). Peptides were > 90% pure based on HPLC profile and MS/MS mass spectrometric analysis performed by the UPCI Protein Sequencing Facility (Shared Resource).

Peptide-binding assay
The peptide-binding determinations were performed using a solid-state competitive-binding assay as previously described (Southwood et al, 1998).

Isolation of patient and normal donor PBMC-Derived T cells
Forty-to-one hundred ml of patient or normal donor heparinized blood was obtained with informed consent under an IRB-approved protocol and diluted 1:2 with HBSS, applied to ficoll-hypaque gradients (LSM, Organon-Teknika, Durham, NC) per the manufacturer's instructions, and centrifuged at 550 × g for 25 min at RT. Patient and normal donor information is provided in Table 2. Lymphocytes at the buoyant interface were recovered and washed No normal donors displayed significant CD4+ T cell reactivity against any of these peptides prior to in vitro stimulation (see Figure 1C). Patients were considered positive responders if their calculated frequency of peptide-specific spots exceeded the mean + 3 standard deviation values (see Figure 1 legend) obtained from 10 normal HLA-DR4+ donors against that indicated peptide. Known CTL epitopes embedded in these sequences are indicated by shading, while positive T cell responses are in bold and underlined (Kittlesen et al, 1998;Skipper et al, 1996;Ranieri et al, 2000;Brinckerhoff et al, 2000). Peptides indicated with ** represent previously identified HLA-DR4-presented epitopes. Binding data is reported as IC 50 in nM and represents the concentration of the indicated peptide to block 50% of the binding of a radio-labeled reference peptide to purified HLA-DR4 complexes. IC50 values ≥ 10 000 nM are indicated as-. NT = Not Tested.
twice with HBSS to remove residual platelets and ficoll-hypaque. Cells were frozen in 90% FCS containing 10% DMSO (Sigma Chemical Co., St Louis, MO) at 10 7 lymphocytes/vial using controlled-rate freezing technique. On the day of ELISPOT assay, cells were thawed and washed twice with HBSS. CD4+ T cells were first isolated using MACS ™ (Miltenyi) anti-human CD4 beads and MiniMACS ™ columns per the manufacturer's protocol. CD4+ T cell yields were typically 25-35% of starting PBMC numbers loaded, with purity exceeding 97% as assessed by flow cytometry.

Induction of antitumour T effector lymphocytes
Autologous dendritic cells (10 7 ) were prepared as previously described by 7-day culture of plastic-adherent PBMC in GM-CSF and IL-4 (Tueting et al, 1998). Harvested, non-adherent DC were then loaded with antigen in one of three ways. DC were either: (1) pulsed with 10 uM synthetic peptides for 3-4 h at 37˚C; (2) pulsed with 250 µg/ml Mel 526 (gp100/pmel17+ tyrosinase+) lysate for 24 h at 37˚C; or (3) infected with recombinant vaccinia virus encoding the gp100/pmel17 or tyrosinase antigens (Kittlesen et al, 1998) at an MOI of 50 for 24 h at 37˚C. The resulting 'antigenloaded' DC were washed, irradiated (3000 rad) and used to stimulate purified CD4+ T cells at a 50:1 responder-to-stimulator ratio. Primary in vitro cultures were performed in AIM-V medium containing 5% HuAB serum and 1 ng/ml of both rhIL-1 and rhIL-7 (Genzyme). Cultures were restimulated with the residual 1/2 of the DC-tumour stimulator preparation (cryopreserved for 1 week) on day 7 in AIM-V medium containing 5% HuAB serum and 10 IU/ml rhIL-2 (kind gift of Chiron Corporation, Emeryville, CA). These stimulated T cells were harvested on day 14-17 and analyzed for peptide/tumour specificity in ELISPOT assays.

IFN-γ ELISPOT and capture of peptide-reactive CD4+ T cells
ELISPOT was performed essentially as previously described (Herr et al, 1998). Spots were imaged using the Zeiss AutoImager and statistical comparisons determined using a Student's two-tailed t-test analysis. The data are represented as IFN-γ spots per 100 000 CD4+ T cells analysed. HLA-restriction of CD4+ T cell responses was demonstrated by addition of blocking mAb (5 ug/well) directed against HLA-DR4 (clone 359-13F10, IgG, kindly provided by Dr Janice Blum, Indiana University School of Medicine, Indianapolis, IN).

PCR analysis
PCR analyses were performed to determine patient HLA-DR4 genotype using a commercial PCR panel according to the manufacturer's instructions (Dynal, Oslo, Norway). RT-PCR analysis was also used to determine tumor expression of gp100/pmel17 and tyrosinase mRNA .

Selection and screening of candidate DR4-binding peptides derived from gp100/pmel17 and tyrosinase
In order to identify a series of candidate peptides capable of being recognized by CD4+ T cells in HLA-DRA+ patients who have, or have had melanoma, we subjected the cDNA sequences of gp100/pmel17 and tyrosinase to analysis using a computer algorithm designed to predict HLA-DR*0401-binding peptides (Honeyman et al, 1998). Nine amino acid-long 'core' sequences were evaluated and scored from 0-10, with the highest scoring sequences taken to represent peptides most likely to bind to HLA-DR*0401. A total of 18 peptides were produced (Table 1). One gp100/pmel17 (gp100/pmel17 44-59 ) and two tyrosinase peptides (tyrosinase 56-70 and tyrosinase 448-462 ), previously identified to be recognized by DR4-restricted CD4+ TIL , were included as 'positive controls' among the panel used in this study. Based on the predictive algorithm used, these 'control' peptides would have been produced for analysis due to their inclusion of core epitopes yielding algorithm scores ≥ 4 (Table 1). These selected synthetic peptides were then assessed for their comparative abilities to bind to purified HLA-DR4 molecules using a solid-state competitive-binding assay (Southwood et al, 1998). The data reported as the dose of peptide capable of inhibiting 50% of the binding of a reference DR4-binding peptide (IC 50 in nM) are listed in Table 1. Most of the peptides evaluated bound both HLA-DR4 alleles, albeit over a wide range of observed IC 50 values. The strongest HLA-DR4 (both HLA-DRB1*0401 and -DRB1*0404) binding peptides for each antigen tested were gp100 615-633 (identical to pmel17 622-640 ) and tyrosinase 365-381D , respectively (Table 1).

Immunoreactivity of CD4+ T cells against predicted HLA-DR4-binding tumour peptides in HLA-DR4+ melanoma patients
In a preliminary screen of the immunogenicity of these peptides, we evaluated the ability of CD4+ T cells isolated directly from the peripheral blood of 16 HLA-DR4+ patients treated for melanoma (Table 2) to recognize these putative peptide epitopes using the IFN-γ ELISPOT assay. Eleven of these individuals (SLM1-SLM11) were rendered disease-free as a result of surgery and/or immunotherapy and we hypothesized that their lack of current disease might, in part, be attributed to circulating anti-melanoma CD4+ T cells. As indicated in Tables 1 and 3 and Figure 1A, a number of these disease-free patients displayed detectable frequencies of circulating Th0/Th1-type (i.e. IFN-γ secretors) CD4+ T cells that recognized a sub-set of the peptides selected for analysis (i.e. gp100/pmel17 6-26 , gp100/pmel17 44-59 , gp 100 615-633 /pmel17 622-640 , tyr. 56-70 , tyr. 156-175 , tyr. 365-381D and tyr. 365-381N ). Interestingly, with the exception of a single HLA-DR4+ patient with ocular melanoma (SLM12), four additional patients (SLM13-SLM16) with active stage III or IV disease exhibited minimal or no reactivity to any of the peptides analysed in IFN-γ ELISPOT assays ( Figure 1B). These same peptides failed to be recognized, or were recognized extremely poorly by freshly-isolated CD4+ T cells harvested from a series of 10 normal HLA-DR4+ donors ( Figure 1C).

CD4+ T cells stimulated in vitro with autologous DC infected with vaccinia encoding gp100/pmel17 or tyrosinase or pulsed with melanoma lysates recognize DR4-presented melanoma peptides
We next investigated whether autologous HLA-DR4+ DC that have acquired gp100/pmel17 or tyrosinase as a result of infection with a recombinant vaccinia virus or via exogenous loading with a freeze-thaw lysate derived from the allogeneic DR4-negative, gp100/pmel17+, tyrosinase+ Mel 526 promoted the activation and expansion of epitope-specific normal donor CD4+ T cells in vitro.

DISCUSSION
Using HLA-DR4+ patient's PBMC, a peptide-binding algorithm, and a high-sensitivity IFN-γ ELISPOT assay, we have identified three novel, non-overlapping epitopes derived from the gp100/pmel17 and tyrosinase melanoma-associated antigens (MAA) that are functionally recognized by Th1-type CD4+ T cells. In addition, two of three previously identified CD4+ epitopes derived from gp100/pmel17 and tyrosinase (gp100/pmel17 44-59 and tyrosinase 56-70 but not tyrosinase [448][449][450][451][452][453][454][455][456][457][458][459][460][461][462] were also confirmed in these analyses (Topalian et al, 1996;Storkus and Zarour, 2000;Touloukian et al, 2000). Overall, 7/18 peptides chosen for evaluation (based on the algorithm analysis) appeared to promote Th1-type responses from purified CD4+ T cells obtained from melanoma patients. CD4+ T cells from normal HLA-DR4+ donors were poorly reactive against epitopes and, with the exception of the dramatic anti-tyrosinase responses noted for a single patient with ocular melanoma (i.e. SLM12), CD4+ T cells from patients with active disease also tended to display muted peptide-specific responses. In marked contrast, CD4+ T cells isolated from the PBMC of patients that have been effectively treated in the clinic and that remain disease-free, displayed significantly elevated responses to a number of these putative epitopes. This suggests that long-term 'memory' CD4+ T cells may circulate at high frequency in the periphery of these 'protected' patients.
Alternatively or additionally, the T cell repertoires of these patients may have been periodically restimulated (by clinically undetectable, recurrent micrometastatic disease) by either class II+ tumours themselves or via 'cross-presentation' by tumour-associated antigen presenting cells (Ossendorp et al, 1998;Albert et al, 1998;Lotze et al, 1997). Our data provided in Figure 3 and support the ability of CD4+ T cells to recognize processed epitopes 'crosspresented' by autologous DC. In particular, novel epitopes within  (18) and then used to stimulate autologous CD4+ T cells at a responder-to-stimulator ratio of 20:1. After being restimulated with identically prepared DC on day 7 of culture, CD4+ T cells were harvested and evaluated for their reactivity against peptide-pulsed T2.DR4 target cells or a series of melanoma cell lines in 20 h IFNγ ELISPOT assays one week later (open symbol). The mean of background T cell responses against the T2.DR4 target (without peptide) have been subtracted in all cases of peptide testing, but were 23 spots and 16 spots per 100,000 CD4+ T cells in the Vacc-gp100 and Vacc-tyrosinase cultures, respectively. T cells responses to melanoma target cell lines were not corrected and represent actual spot numbers. Peptide CS is an HLA-DR4-presented epitope derived from the malarial circumsporozooite protein that serves as a negative biological control. Melanoma cell line phenotypes are as follows: SLM2 is HLA-DR4+, gp100/pmel17+, tyrosinase+; M21 is HLA-DRdim+, gp100/pmel17+, tyrosinase+; COLO38 is HLA-DR4+, gp100/pmel17+, tyrosinase+; and Mel1011 is HLA-DR4+, gp100/pmel17+/-, tyrosinase+ (see Materials and Methods). The DR4-restricted nature of T cell reactivity was demonstrated by addition of blocking anti-HLA-DR4 mAb 359-13F10 (filled symbol). Data represent the mean +/-SD of triplicate determinations from 1 of 3 donors evaluated. All donors displayed similar reactivity patterns.
Of the three novel, non-overlapping epitopes defined in this study, one was derived from the gp100/pmel17 protein (gp 100 615-633 (identical to pmel17 622-640 )) and two were derived from the tyrosinase protein (tyrosinase 156-175 and tyrosinase 365-381D (overlaps with tyrosinase 365-381N )). While the gp100/pmel17 6-26 peptide was reacted against by CD4+ T cells obtained from melanoma patients free of disease and normal donors stimulated with peptides in vitro (Table 1), these responses were very modest and were not observed in cultures stimulated by Vac-gp100-infected autologous DC (Figure 3). This could reflect the differential processing of this gp100 epitope by melanomas versus transduced DC, given variance in processing observed amongst antigenpresenting cells (Vidard et al, 1992), or it could reflect T cell crossreactivity against another naturally-processed sequence expressed in the melanoma setting. Given these uncertainties, we do not classify this peptide as containing a naturally-processed epitope at this time.
Interestingly, one of the HLA-DR4 presented tyrosinase epitopes (i.e. tyrosinase 365-381 ) encompasses a glycosylation site known to be post-translationally-modified in melanoma cells. Thus, the genomically-encoded tyrosinase contains an asparagine (N, single letter designation) residue at position 370 that is modified into an aspartic acid (D, single letter designation) as a result of enzymatic processing. Skipper et al (1996) have previously demonstrated that this modification yields the naturally-processed and HLA-A2-presented tyrosinase 368-376 (YMDGTMSQV) epitope. We have now similarly shown that a 'post-translationally modified' tyrosinase 365-381D , but not the genomically-encoded tyrosinase 365-381N peptide, serves as a strong HLA-DR4-presented immunogen recognized by melanoma-reactive CD4+ T cells. The observed enhanced immunoreactivity of the tyrosinase 365-381D peptide can be attributed, in part, to its approximately 100-fold better binding 'affinity' to HLA-DRB1*0401 (vs the tyrosinase 365-381N peptide, Table 1). The lack of cross-reactivity of responder CD4+ T cells against these two epitopes, however, may also argue for non-overlapping peptide-specific T cell repertoires.
The immunogenic gp100/pmel17 and tyrosinase epitopes identified in this study are presented by HLA-DR4, which is expressed by 18-20% of the population afflicted with melanoma (Gjertson and Lee, 1998). The clinical utility of these sequences would be enhanced greatly if they could be presented in the context of additional HLA-DR alleles to melanoma-reactive CD4+ T cells. Our preliminary analysis suggests that a number of these epitopes bind to a broad range of MHC class II alleles (data not shown). For instance, the gp100/pmel17 44-59 peptide binds to (at least) the HLA-DR1, -DR3, -DR4, -DR7, -DR13 and -DRw53 (i.e. -DRB4*0401) alleles that cover in excess of one-half of the American population (Gjertson and Lee, 1998). If naturally presented by these non-HLA-DR4 alleles, these pan-DR binding peptides may prove immunogenic to a broad range of patients, a possibility that we are currently evaluating.
While we have identified a series of biologically-active epitopes in this study, we have little conclusive data (save for confirmation of the gp100/pmel17 44-59 sequence, recently identified as a natural epitope Touloukian et al, 2001) to support that these sequences are 'naturally-presented' in the exact format as tested. We are currently performing mass spectrometric analyses of peptides derived from affinity-purified HLA-DR4 complexes obtained from MHC class II+ melanoma cell lines in order to determine a series of natural tumour-processed 'minimal core epitopes'. From a practical perspective however, the identification of a 'minimal core epitope' may be a subordinate issue in the context of the potential clinical applications of these peptides, which include immune monitoring and vaccine construction. Long peptides (i.e. 12-16 amino acids) containing a core 9-mer epitope have been previously shown to elicit protective immunity mediated by CD8+ T cells in the in vivo vaccine setting (Kast et al, 1993) or in vitro after direct binding of peptide to the MHC molecule (Kittlesen et al, 1998 and unpublished observations). Similar mechanisms would likely allow for 'optimal' loading of HLA-DR complexes with relevant epitopes that are typically not as constrained for length as are MHC class I-presented peptides (Falk et al, 1994). This might be best analysed in preclinical in vivo systems such as the DR-IE transgenic mouse and SCID-Hu vaccine models that have been recently used to corroborate or define the gp100 44-59 or gp100 231-243 epitopes (Touloukian et al, 2000;Cochlovius et al, 2000). The latter model (Cochlovius et al, 2000) in particular, could provide strong support for the clinical utility of vaccines incorporating HLA-DR presented epitopes since DC-peptide vaccines may inhibit the growth of melanoma cell lines expressing the relevant antigen (i.e., from which the peptide derived) in situ.
A potentially added benefit to the clinical use of certain of the identified gp100/pmel17 and tyrosinase peptides is that they contain embedded CTL epitopes (Table 1). Hence, a single synthetic peptide may yield both CD4+ and CD8+ T cell recognized determinants when appropriately processed in situ, by at least some patients (bearing the appropriate MHC class I alleles for CTL epitope presentation). Such combination vaccines containing 'helper' T and CTL epitopes may prove particularly therapeutic (Ranieri et al, 2000).