Proteomic Scan for Tyrosinase Peptide Antigenic Pattern in Vitiligo and Melanoma: Role of Sequence Similarity and HLA-DR1 Affinity¹

Vitiligo is a skin disorder characterized by selective melanocyte destruction and concomitant appearance of depigmented macules that over time enlarge, coalesce, and form patches (1). It has been suggested that vitiligo is, at least in part, caused by autoimmune responses mediated by cytotoxic T cells against melanocytes, causing depigmentation. In addition, many patients with vitiligo produce autoantibodies against proteins expressed in melanocytes (2, 3, 4). The lysis of human melanocytes by vitiligo patients’ sera provides direct support for the autoimmune hypothesis of human vitiligo (5).

Interestingly, patients with metastatic melanoma may spontaneously develop vitiligo-like hypopigmentation. Some patients develop this clinical feature during immunotherapies (6). Moreover, in animal models of melanoma, humoral immune responses to pigment cells appear to correlate with tumor regression and hypopigmentation (7). Such observations have induced to hypothesize that the appearance of vitiligo-like depigmentation in some melanoma patients results from immune responses to Ags that are shared by normal melanocytes and melanoma cells (2).

Tyrosinase, the enzyme involved in melanin biosynthesis by both melanocytes and melanoma cells, is one main target of the autoantibodies found in the sera of patients with diffuse vitiligo and/or metastatic melanoma (8, 9, 10, 11). However, notwithstanding the number of studies and efforts in many laboratories, how anti-melanocyte Ab reactivity arises and causes vitiligo, and which is the molecular antigenic specificity of the immunological reactions in vitiligo and melanoma, remain two unanswered elusive questions (12, 13, 14). This lack of answers has its foundation in our ignorance of the mechanisms which dictate Ag (non)immunogenicity.

In this study, we have used a proteomics approach to define the repertoire of circulating anti-tyrosinase autoantibodies in vitiligo and melanoma patients following a hypothesis developed in our laboratories, i.e., that peptide immunogenicity might also be regulated by the sequence similarity level to the host’s proteome, in addition to the HLA-binding potential. The underlying scientific rationale is that immune system is allowed to respond only to rarely encountered/never seen antigenic sequences (15, 16, 17, 18, 19). To test our hypothesis in this study, synthetic peptides corresponding to low similarity sequences present in the tyrosinase autoantigen were immunoassayed by using vitiligo and melanoma sera. Synthetic peptides corresponding to sequences with high affinity to HLA class II molecules were used as peptide controls to better define the role of similarity level in peptide immunogenicity as compared with HLA II affinity contribution (20, 21). We report that, among the tyrosinase HLA-binding peptides we tested, only those hosting a sequence with low similarity to the human proteome exhibited immunoreactivity.

Analysis of the similarity level to the human proteome was conducted on the human tyrosinase sequence (SWISS-PROT, accession number: P14679, tyrosinase or monophenol monooxygenase or tumor rejection Ag AB). Tyrosinase sequence was dissected into pentamer motifs, that were probed for sequence similarity to human proteome by using the Protein Information Resource nonredundant protein database and peptide match program (〈pir.georgetown.edu/pirwww/search/nrefpeptide_demo.html〉) (22).

The SYFPEITHI program (〈www.syfpeithi.de/〉 or 〈http://syfpeithi.bmi-heidelberg.com/〉 or 〈www.uni-tuebingen.de/uni/kxi/〉) (20) was used as database of HLA ligands and peptide motifs in the epitope prediction study aimed to obtain the tyrosinase peptide scoring for HLA-DRB1 binding potential. The most common allelic subtype of a specific set of HLA-DR molecules comprising DR1, DR2, DR3, DR4, DR7, DR11 Ags were analyzed. More precisely, we analyzed tyrosinase peptide binding to HLA-DRB1*0101, HLA-DRB1*0301, HLA-DRB1*0401, HLA-DRB1*0701, HLA-DRB1*1101, and HLA-DRB1*1501. Following this analysis, the top-scoring 15-mer tyrosinase peptide for each HLA-DRB1 representative allele was selected to be tested as potentially immunogenic tyrosinase sequences in dot-blot immunoassays by using human sera.

Peptides were synthesized by standard F-moc-solid phase peptide synthesis (Primm). The purity of the peptides was >95% as assessed by HPLC. The molecular mass of purified peptides was confirmed by fast atom bombardment mass spectrometry.

The synthetic peptides used in dot-blot immunoassays were: tyrosinase_4–18 AVLYCLLWSFQTSAG, tyrosinase_95–104FMGFNCGNCK, tyrosinase_175–182LFVWMHYY, tyrosinase_176–190FVWMHYYVSMDALLG, tyrosinase_178–192 WMHYYVSMDALLGGY, tyrosinase_215–229LLRWEQEIQKLTGDE, tyrosinase_222–236 IQKLTGDENFTIPYW, tyrosinase_222–232IQKLTGDENFT, tyrosinase_233–242 IPYWDWRDAE, tyrosinase_233–247IPYWDWRDAEKCDIC, tyrosinase_238–247 WRDAEKCDIC, tyrosinase_349–363SPLTGIADASQSSMH, tyrosinase_423–438 ESYMVPFIPLYRNGD, tyrosinase_435–449NGDFFISSKDLGYDY, tyrosinase_472–486 SRIWSWLLGAAMVGA, HER-2/neu_27–41TGTDMKLRLPASPET, HER-2/neu_213–227 QSLTRTVCAGGCARC.

Biotinylated tyrosinase peptides identical to the above listed 15-mer peptides were used in gel-shift experiments. Exceptions were: tyrosinase_4–18AVLYCLLWSFQTSAG, tyrosinase_176–190FVWMHYYVSMDALLG, tyrosinase_472–486SRIWSWLLGAAMVGA, the solubility of which was low in the biotinylated form. To overcome the solubility problem, the following longer biotinylated peptides were synthetised: tyrosinase_4–23 AVLYCLLWSFQTSAGHFPRA, tyrosinase_176–193FVWMHYYVSMDALLGGSE, tyrosinase_469–486EQASRIWSWLLGAAMVGA (with the added amino acids given underlined).

Serum samples were obtained from a total of 14 patients with vitiligo, 23 patients with cutaneous melanoma, and 6 healthy subjects. Patients were diagnosed and treated in the Skin Cancer Unit at the Department of Dermatology, University Hospital Zürich (Zürich, Switzerland). The study was approved by the Hospital Institutional Review Board. We used only sera that otherwise had to be discarded. Sera were partially purified by precipitation with 40% saturated (NH4)₂SO₄ (2×). The precipitate was dissolved in PBS, dialyzed against PBS with several changes for 24 h at 4°C, then aliquoted and stored at −20°C until assay. Subject HLA typing was conducted at the Department of Dermatology, University Hospital Zürich. The HLA class II phenotype of the patients were determined on blood samples collected in acidic citrate dextrose tubes using a complement-mediated microlymphocytotoxicity test (Biotest). The test was performed following the manufacturer’s instructions.

Nitrocellulose membrane (0.2 μm pore size; Bio-Rad) was pretreated for 10 min with 1.0% glutaraldehyde. Peptides (5 μg) were spotted on the activated membranes and left to dry at room temperature. To block nonspecific binding sites, membranes were incubated for 1 h in PBS/0.05% (v/v) Tween 20 (PBST) containing 5% BSA and then with human polyclonal Ab (pAbs)³ (250 μg/10 ml incubation solution). Following 1 h incubation at room temperature (RT), membranes were washed for 10 min with PBST (3×) and further incubated in PBST/5% BSA for 1 h with HRP-conjugated goat anti-human polyclonal Abs (1/10000) obtained from Sigma-Aldrich. Membranes were washed in PBST for 5 min (3×), in PBS for 5 min (3×), and immunoblots were developed by ECL detection assay (Amersham Biotech).

Human THP-1 and SiHa cell lines were obtained from Interlab Cell Line Collection and grown according to the furnisher’s instructions. Cells were collected by centrifugation (10 min at 3,000 × g; RT), resuspended in 80% cell culture medium-10% FCS-10% DMSO-10 μg/ml caspase inhibitor, and incubated with biotinylated peptides at RT for 1 h. Approximately 2 × 10⁶ cells were incubated with 100 μg of biotinylated peptide. Following centrifugation (5 min at 10,000 × g; RT), cells were rapidly washed with the incubation medium, and then lysed for 30 min in nondenaturing lysis buffer (20 mM Tris-HCl (pH 7.5), 150 mM NaCl, 1 mM Na₂EDTA, 1 mM EGTA, 1% Triton, 1 mM β-glycerophosphate, 100 μg of pepstatin A/ml, 10 μg of leupeptin/ml, 200 U of RNase/ml, 200 U of DNase/ml), at RT.

PEMSA was conducted to investigate peptide binding to HLA-DR molecules. Following cell incubation with biotinylated peptides and nondenaturing lysis, cell lysate was electrophoresed (35 mA, 6 h, 16°C) through a 3.5% native polyacrylamide gel with 50 mM potassium HEPES (pH 8.0), 1 mM magnesium acetate, 0.01% Nonidet P-40 (v/v), as running buffer.

Gels were electroblotted onto a polyvinylidene difluoride (PVDF) nylon membrane (0,2 μm; Bio-Rad) for 30 min at 84 V by using 25 mM Tris, 190 mM glycine, as transfer buffer. Each sample was run in double: then, one PVDF membrane was assayed for biotinylated peptide shift by using ultrasensitive streptavidin-peroxidase polymer (Sigma-Aldrich). The second PVDF membrane was probed for the nature of the peptide-binding protein by using rabbit polyclonal anti-HLA-DRα IgGs (FL-254; Santa Cruz Biotechnology) at the dilution 1/80, v/v, followed by incubation with secondary peroxidase-conjugated anti-rabbit mAb (Sigma-Aldrich).

Signals were visualized by ECL reaction (ECL kit; Sigma-Aldrich). Kaleidoscope polypeptide standards (catalog 161-0325; Bio-Rad) were used as kilodalton marker.

Immunostaining of archival patient material was performed as detailed elsewhere (23). In brief, a representative tissue section was stained with H&E, and all sections were reviewed by expert dermatopathologists to confirm the diagnosis. A 3- to 5-μm adjacent section was used for immunohistochemistry with mAb (mAb) T311 (Novocastra Laboratories), and visualized by the alkaline phosphatase-anti-alkaline phosphatase method. The deparaffined sections were heated in a household microwave oven at 100 W (×3.5 min each) in 10 mM citric acid to enhance Ag detection. mAb T311 working solution was produced according to the manufacturer’s instructions, diluted 1/25 with RPMI 1640 (Invitrogen Life Technologies) and then applied for 60 min. The same procedure was applied to negative controls except for mAb T311 which was left out, and showed no unspecific immunoreactivity. In all specimens, non-neoplastic cells like normal epidermal cells, tumor-infiltrating lymphocytes, and fibroblasts, as well as other cells of the s.c. tissue, were present, thus serving as internal negative controls.

Tumor and non-neoplastic tissue specimens containing 1% or more immunostained cells were considered immunopositive. Intensity of staining was rated arbitrarily as negative, weak, moderate, strong, and very strong. Representative parts of the melanoma tissue sections were photographed and digitized by an Aperiotechnology ScanScope (Aperiotechnologies).

Fig. 1 illustrates the proteomic-based similarity analysis of the human tyrosinase protein sequence. In the histogram, the density of tyrosinase 5-mer motif matches along the human proteome measures the profile of the tyrosinase motif sharing. Scanning of the amino acid sequence of the melanocyte-associated-protein tyrosinase for regions with no or low similarity to the human proteome was conducted by using the following computational procedure. Because pentapeptide units can be antigenic sites sufficient to host minimal antigenic determinants (24, 25, 26), the human tyrosinase sequence was dissected into pentamers that were used against human proteome in similarity analyses by using exact peptide match program (15, 16, 17, 18, 19). Pentamers overlapped by four residues, i.e., MLLAV, LLAVL, LAVLY, AVLYC, VLYCL, etc., were sequentially used. Fig. 1 shows that, as found in other Ag analyses too (15, 16, 17, 18, 19), the level of sharing along the protein has an alternating behavior. Tyrosinase fragments have pentamers in common with a number of other human proteins and, at the same time, it is also evident that many pentamers in tyrosinase are peculiarly owned by the melanoma Ag. It is evident that tyrosinase is an ideal candidate to test our similarity hypothesis, given the general low level of sequence redundancy shared with other human proteins. In this preliminary approach, the sequences to be tested were chosen by selecting only peptide stretches with at least three consecutive zero matches pentamers. This arbitrary criterium was applied to obtain 7-mer fragments as minimal length because shorter peptides do not bind efficiently to the nitrocellulose membrane.

The selection produced two sequences, i.e., tyrosinase_95–104FMGFNCGNCK and tyrosinase_175–182LFVWMHYY, that appeared of relevance to the similarity analysis because all of their pentamers had no matches at all to the human proteome. According to the hypothesis pursued in our laboratories, the two sequences were promising immunogenic sites and the corresponding peptides were synthesized to be used as antigenic probes to investigate the immunoreactivity pattern of sera from vitiligo and melanoma patients.

The aim of this study was to analyze the tyrosinase autoepitope profile of melanoma/vitiligo patient sera by using sequence self-similarity as measurement unit. In the search for a control parameter, the binding potential of tyrosinase peptides to HLA class II molecules appeared suitable because specific binding of antigenic peptides to HLA II molecules is considered as a determining factor in the humoral immune response induction (20, 21, 27, 28). Therefore, it was essential to evaluate the immunoreactivity of low similarity peptide in the context of HLA affinity to distinguish the possible contribution to humoral immunogenicity due to similarity from that determined by high peptide affinity to HLA class II molecules. To this aim, the HLA II haplotype of the subject population under study was determined; the relative data are reported in Table I.

Then, we focused on HLA-DR molecules because: 1) many epitopes from melanoma Ags generate and maintain immune responses when presented in the context of HLA-DR molecules (29, 30, 31), and 2) specific HLA-DR molecules appear involved in autoimmunity diseases (32, 33). In particular, it is of interest that a tyrosinase peptide in the context of HLA-DR4 molecule was reported to be recognized by a T cell line from a patient with Vogt-Koyanagi-Harada’s disease, which is regarded as an autoimmune disorder in multiple organs containing melanocytes (34).

Analysis of HLA-DR Ags presented by the vitiligo/melanoma patients and the control subjects is reported in Table II. Specifically, the table illustrates the HLA-DR Ags presented by the subject population under study, the most common allelic subtype for each HLA-DR Ag, and their frequency in the subject group here studied as compared with their frequency in the world population (35, 36). As a first datum, it can be seen from Table II that the HLA-DR expression in the subject group reflects the estimated HLA-DR frequency in the global population, thus conferring general significance to the present study.

Then the tyrosinase peptide affinity toward the most common allelic subtype of each HLA-DR Ag was computationally calculated to select tyrosinase peptides with high affinity to the most common allelic subtype for each HLA-DR Ag. The entire tyrosinase sequence (aa 1–529) was analyzed for 15-mer peptides able to bind the allelic subtype of each HLA-DR Ag listed in Table II. The SYFPEITHI program (〈www.uni-tuebingen.de/uni/kxi/〉) was used as database of HLA ligands and peptide motifs. This algorithm uses motif matrices defined from pool sequencing approaches for scoring peptides and calculates the HLA binding potential score by giving the amino acids of a certain peptide a specific value (from 1 to 15) depending on whether they are anchor, auxiliary anchor or preferred residue. Amino acids that are regarded as having a negative effect on the binding ability are also evaluated by a negative value. The final score indicates the peptide affinity and probability to be presented to the immune system. Following SYFPEITHI analysis, we obtained a longest list of tyrosinase peptide sequences with binding score ranging from the highest (34) to negative (−5) values (not shown); only the first five top-scoring sequences for each HLA-DR1 subtype are reported in Table III. To our experimental aims, we selected the highest affinity peptide sequence for each HLA-DR1 subtype: these sequences are given bold in Table III. In addition, Table III reports the HLA-binding characterization of two additional tyrosinase peptides that were at hand and intended to be used as internal tyrosinase controls.

Fig. 2 is the detailed description of the two sets of tyrosinase peptide sequences, i.e., characterized by low similarity to human proteome and high potential binding to HLA-DR molecules, respectively, elected to be tested as antigenic fragments in dot-blot immunoassays using sera from vitiligo and melanoma patients. In addition, the two tyrosinase 15-mer peptides, available in the laboratory and intended to be used as controls, are also illustrated in Fig. 2.

As a second step, we synthesized the tyrosinase peptides corresponding to the sequences grouped as A and B in Fig. 2, and used them as Ags to analyze the tyrosinase peptide antigenic pattern in vitiligo and melanoma patients. The already available tyrosinase peptide set C was used as internal random control. As additional further control, two unrelated 15-mer peptides, i.e., HER-2/neu_27–41 TGTDMKLRLPASPET, and HER-2/neu_213–227QSLTRTVCAGGCARC, served as tyrosinase-unrelated, external controls (peptide set D). The four groups of peptides, named A to D and listed in the legend to Fig. 3, were tested in dot-blot immunoassays as shown in Fig. 3 and Table IV. Fig. 3 shows representative peptide dot-blot immunoassay analyses with vitiligo, melanoma or control serum. Table IV precisely details the antigenic pattern of the four peptide sets monitored by using sera from 14 vitiligo and 23 melanoma patients, and from 6 healthy controls.

Fig. 3 and Table IV indicate that the low similarity peptide 1, corresponding to the tyrosinase_95–104FMGFNCGNCK fragment, which carries a row of six pentamers with zero similarity to the human proteome, was recognized by the serum of all vitiligo/melanoma patients. The reactivity pattern with serum from vitiligo patients was similar to that obtained using melanoma sera. In addition, sera from healthy donors also reacted with the zero-similarity tyrosinase_95–104FMGFNCGNCK peptide by giving signals of intensity comparable to that obtained with vitiligo/melanoma pAbs. The reactivity pattern of the second zero similarity peptide, i.e., tyrosinase_175–182 LFVWMHYY, with the vitiligo/melanoma sera was lower both as signal intensity and number of sera when compared with the companion not-shared tyrosinase_95–104FMGFNCGNCK peptide.

Among the tyrosinase peptides characterized by high potential binding to HLA-DR molecules (i.e., peptide set B), no immunoreactivity was observed with any of the sera tested. There are only two exceptions: peptides 4 and 7. But it has to be noted that tyrosinase peptide 4, i.e., aa_176–190FVWMHYYVSMDALLG, contains a portion of the zero-similarity tyrosinase _175–182 LFVWMHYY, so that its reactivity appears referable to that of low-similarity peptide 2. In this regard, no reactivity was shown by sequence-related peptide 5, i.e., aa_178–192WMHYYVSMDALLGGY, the lack of immunoreactivity of which seems to indicate that the amino acid residue Val¹⁷⁷ might be decisive in shaping the antigenic determinant in peptides 2 and 4.

Likewise, also peptide 7, corresponding to tyrosinase_222–236IQKLTGDENFTIPYW sequence, appears to derive its immunogenicity by the sharing of the tetramer IPYW with peptide 11, corresponding to the internal control tyrosinase_233–247IPYWDWRDAEKCDIC peptide. It can be seen that peptide 11 was immunoreactive with vitiligo, melanoma, and control sera, even if endowed with a medium binding score to HLA DRB1*0101 (score = 19, Table III) and lowest binding score values to the other HLA DR1 molecules (data not shown).

To better define the contribution given by sequence similarity to the immunoreactivity of peptides 7 and 11, we synthetized three synthetic peptides encompassing the tyrosinase_222–247IQKLTGDENFTIPYWDWRDAEKCDIC sequence, i.e., the sequence comprehending peptides 7 and 11. The three peptides correspond, in the order, to the sequences: tyrosinase_222–232IQKLTGDENFT, tyrosinase_233–242IPYWDWRDAE, and tyrosinase_238–247WRDAEKCDIC and were assayed in dot-blot immunoassay by using serum from vitiligo, melanoma, or healthy subject. Fig. 4 is a cut confirmation of our similarity hypothesis by illustrating that the sera representative of the three cohorts produced a marked signal with the tyrosinase_222–247IPYWDWRDAE peptide, i.e., the tyrosinase peptide portion endowed with the lowest similarity level to human proteome. No reaction (or only extremely feeble signal) was observed with the two adjacent fragments, i.e., the NH₂-terminal region corresponding to tyrosinase_222–232IQKLTGDENFT or the COOH-terminal region represented by tyrosinase_238–247 WRDAEKCDIC.

With the exception of the above-discussed peptide no. 11, no reaction with any of the assayed sera was detected with control peptides, i.e., tyrosinase peptide control no. 12, or the two tyrosinase-unrelated Her-2/neu peptides (see Table IV).

Fig. 3 and Table IV show that the three autoantigenic tyrosinase peptides 1, 7, and 11 were responsive toward sera from healthy subjects too. Only peptide 2, i.e., tyrosinase_175–182LFVWMHYY peptide, and peptide 4, i.e., tyrosinase_176–190FVWMHYYVSMDALLG peptide, did not react with any of the healthy control sera. Interestingly, these two fragments host 1) the metal catalytic center with the copper binding His¹⁸⁰ and 2) the oculocutaneous albinism type I-A variant residue, with F176 as amino acid position of the variant (F→I) (37).

The results illustrated in Fig. 3 and Table IV were clearly indicative of a scarce contribution of HLA binding potential to the peptide immunogenicity. However, it was also possible that the peptides selected and synthetised for this study had unique stringent restriction toward HLA-DRB1* subclasses not represented in the subject population under analysis. Although unlikely, nonetheless it was essential to clarify this point. To this aim, the peptides used in this study were further analyzed for their capacity to bind DRB1* subclasses according to the anchor P1 position (20). Table V clearly illustrates that the peptides we had selected to test vitiligo and melanoma sera, are theoretically able to bind a broad range of HLA-DRB1* subtypes, so confirming the already reported high degeneracy level in peptide-DR interactions (38, 39). It is also evident from Table V that some peptides are able to bind in two different registers with identical score and, likewise, some peptides can bind by the same register and different score. We concluded that the ability of each peptide to bind numerous HLA-DRB1* subclasses by different registers, although with different affinity, as reported in Table V, ruled out the possibility that peptide immunogenicity had been prevented by the lack of adequate HLA presentation.

Still, it was also possible that the predicted theoretical binding scores shown in Tables III and V did not correspond to the actual HLA binding capacity. To experimentally verify this point, human THP-1 whole cells, which constitutively express DR1 and DR2 Ags (40, 41), were incubated with the biotinylated form of the synthetic peptides used in this study. Following careful cell lysis under nondenaturing conditions, cell lysate was analyzed for the interaction between HLA-DRα molecules and biotinylated peptides by gel shift electrophoresis and Western blot. Fig. 5,A, shows that the biotinylated peptides produced two signals, in correspondence of ∼29 and ∼16 kDa, respectively. When the membrane was probed with anti-HLA-DRα Abs (Fig. 5,B), a diffuse immunoreactivity pattern in the same area occupied by the shifted peptide signals was obtained. We concluded that the peptide mobility shift was due to the binding to HLA-DR1 and -DR2 molecules. Moreover, Fig. 5 demonstrates that all of the peptides used in this study are able to bind HLA-DR1 and HLA-DR2 molecules, although at different extent (see, for example, peptide nos. 6 and 7 in A, which required longer exposure time to produce evident signals). Therefore, the immunoreactivity pattern illustrated in Fig. 5 appears the experimental validation of the predicted binding data reported in Table V.

The retarded peptide mobility illustrated in Fig. 5,A was specific because no signal could be detected following incubation of the biotinylated peptides with SiHa cells, which do not express HLA-DR molecules (42) (Fig. 6).

However, e contrario, it could still be argued that the lack of immunoreactivity exhibited by the otherwise HLA-binding peptides was merely due to the lack of tyrosinase expression in the lesions of the subject population we studied in this work. i.e., if tyrosinase Ag is not expressed in the melanoma cells, tyrosinase peptides have no chance at all to raise any immune response.

This possibility is nullified by studies from one of our laboratories (23) that already demonstrated that 47 of 50 analyzed melanomas overall stain positive for tyrosinase (94%). As an addendum to these published data, immunohistochemistry for tyrosinase reactivity was conducted on patient material that was still available. The results are reported in Table VI and Fig. 7. Given the caveat of a not uniform/constant staining pattern, there was positive tyrosinase staining in many of the melanoma samples that could be examined by being available as archival material (Table VI). In particular, Fig. 7 clearly illustrates the tyrosinase immunoreactivity pattern of representative parts of the melanoma tissue sections from three melanoma patients whose sera were under analysis in this study, i.e., 32-FC, 37-MB, and 41-OM. The tumor sample of patient 41-OM did not present tyrosinase imunoreactivity. Cellular and the detected humoral immune responses possibly have contributed to select tyrosinase negative cells during disease progression in this patient.

Autoimmunity and cancer are strictly interwoven. Patients with various autoimmune diseases, including dermatomyositis, polymyositis, vasculitis, and scleroderma, have an increased risk for the development of cancers (43, 44, 45, 46). Likewise, malignancy is associated with the induction of autoimmunity and the consequent production of autoantibodies against a number of autoantigens (47, 48, 49, 50, 51), including self-Ags expressed in tumor cells (52, 53). However, although the list of known autoantigens is impressive, the molecular mechanisms by which an autoantigen becomes immunogenic are not clear.

Our laboratories have postulated that proteins harboring peptide motifs absent or scarcely represented in the host’s cellular proteins may evoke powerful immune responses and, vice versa, proteins with high and widespread similarity to several host proteins would be tolerogenic. Sharing of common motifs (molecular mimicry of self) might be one cause of limited immunogenicity of oncoprotein self-Ags and, conversely, presence of uniquely expressed sequences might possibly determine specific immune autoreactivity pattern in autoimmune diseases.

The immediate application of this hypothesis is, of course, the search for immunogenic epitopic peptide sequences which might evoke (in cancer) or block (in autoimmune diseases) specific therapeutic/pathogenic autoantibodies, respectively. In this context, analysis of the melanocyte-associated-protein tyrosinase is a privileged topic, by tyrosinase being the epitome of a paradigmatic model of the cancer-autoimmunity dualism (9, 12, 14, 53). By applying our hypothesis and following molecular mapping of low similarity sequences, we defined here the tyrosinase peptide antigenic pattern in vitiligo and melanoma HLA-DR1-typed patients as follows: 1) amino acid sequences uniquely represented in the tyrosinase protein are recognized by autoantibodies from patients with vitiligo or melanoma; 2) HLA-DR1 restriction does not play a determining role in the humoral response as compared with the non-self similarity. So, the first bona fide conclusion of this work is that low similarity seems to play a significant role in shaping tyrosinase peptide immunogenicity, while HLA binding ability does not appear a guarantee of peptide immunogenicity.

Pari passu, the here reported experimental analyses bring up the possible relationship between Ag sequence similarity and natural autoantibodies (NAAs). Indeed, the data reported in Fig. 3 and Table IV provide evidence that anti-tyrosinase autoantibodies are not specific for patients with melanoma/vitiligo and they may be found in healthy subjects too. This commonality appears of importance for the following reason. Circulating nonpathogenic autoantibodies (or NAAs) specific for self-proteins are a common feature of sera from normal healthy subjects (54). In particular, NAAs directed against tubulin, actin, thyroglobulin, myoglobin, fetuin, albumin, transferrin, collagen, and cytochrome-c have been detected in normal human serum (55, 56, 57, 58, 59) and no clinical signs related to autoimmune phenomena were observed. As yet, the relationship between natural and pathogenic autoantibodies remains elusive. Therefore, in this regard, this paper asks as a next logical question: what are the molecular characteristics/processes that render an immunogenic peptide pathogenic? In this study, it may be pertinent to observe that the immunoreactive peptide differentiating humoral response of vitiligo/melanoma patients from that of healthy subjects is only represented by a tyrosinase sequence crucial for tyrosinase activity, both for the presence of copper binding His¹⁸⁰ and oculocutaneous albinism I-A variant position F176 (37).

Elsewhere we already discussed the scientific-clinical implications of low similarity peptides in relation to the definition of the self-non-self (or, better, often-rarely encountered) discrimination concept, and their potential uselfulness as therapeutic agents to break the immunotolerance to the self-Ags without incurring into autoimmune cross-reactions (15, 16, 17, 18, 19). However, here it is opportune to underline that peptide immunogenicity in the similarity context needs to be further evaluated in relation to peptide interaction with paratope site(s) before achieving and defining a general rule. As a matter of fact, it has to be considered that a peptide fragment usually does not retain the conformation present in the folded protein and mostly represents only a part of a more complex epitope involved in the complex with the anti-antigenic protein Abs (60).

Given this caveat, wider mining of the tyrosinase low similarity peptidome and prospective studies aimed to establish the temporal profile of the tyrosinase peptide immunoreactive pattern in healthy population and vitiligo/melanoma patients can offer the basis for studying in molecular terms the physiopathological significance of anti-tyrosinase NAAs and, possibly, the switching mechanism(s) from normal healthy organisms toward vitiligo or melanoma (61).

We gratefully acknowledge Stefan Stefanovic for precious contributions in defining peptide binding potential, Maria Mastronardi for invaluable help in dot-blot immunoassays, and Guglielmo Lucchese for dedicating his summer weekends to gel-shift experiments.

A. Mittelman and D. Kanduc are inventors on patents for the method of identifying amino acid sequences in Ags of interest that are useful for evoking immune responses.