Clinics in Dermatology (2013) 31, 156–165

Immunology of melanomaBijay Mukherji, MD⁎

Department of Medicine, University of Connecticut School of Medicine, 263 Farmington Avenue, Farmington, CT 06030, USA

Abstract It can be safely said that human melanomas are immunogenic. Virtually all the majorprinciples of “tumor immunology” have been experimentally established in this model. It is now amplyclear that melanoma cells display multiple antigens and peptide epitopes that are targetable by the hostimmune system and that patients with melanoma are capable of responding to these antigens andepitopes serologically as well as through the cell-mediated mechanisms. The immune responses againstmelanoma are, however, subject to regulation by the regulatory processes within the immune systemitself and melanoma cells can resort to overt evasive activities. Indeed, the intrinsic as well as theextrinsic mechanisms within the immune system that are designed to control the magnitude as well asthe duration of immune responses at times act as constraints against generating a robust and long-lastingantimelanoma response and melanoma cells are capable of using all the tricks (eg, downregulation oftargetable molecules, elaboration of immunosuppressive cytokines) available to living organisms so asto evade immune recognition and destruction. As a result, the immune system often fails to protect thehost against melanoma development and progression. The cumulative knowledge over the years onmelanoma-associated antigens and epitopes, on methods of immunization, and on technologies forgenerating melanoma antigen-specific T cells, natural or engineered, have led to the development ofimmunotherapeutic strategies with “melanoma vaccines” and with T-cell–based adoptive immunother-apy for melanoma. Although these strategies have not been uniformly successful in all cases, durablecomplete regressions of metastatic melanoma can at times be obtained with active specific immunizationor adoptive cell therapy. There is reason for hope that continued research in the field is likely to improvethe outcome of melanoma immunotherapy: the ultimate goal of tumor immunology.© 2013 Published by Elsevier Inc.

Introduction

Human melanoma is possibly the best example of an“immunogenic” tumor. Human melanomas exhibit a set ofunique features strongly suggestive of “footprints” of hostimmune responses and virtually all the tenets of tumorimmunity have now been experimentally established, in thelaboratory as well as through translational human clinical

⁎ Corresponding author. Tel.: +1 860 679 4236; fax: +1 860 679 1228.E-mail address: mukherji@nso2.uchc.edu.

0738-081X/$ – see front matter © 2013 Published by Elsevier Inc.http://dx.doi.org/10.1016/j.clindermatol.2012.08.017

trials, in this disease model. Herein, I point out the clinicalfeatures that are uniquely associated with melanoma and aresuggestive of some form of host immune responses againstthis disease and then move the discussion to the experimentalevidence that supports this contention.

Clinical features suggestive of an immunogenicnature of melanoma

That melanoma might be an immunogenic tumor has longbeen appreciated from the following well-recognized facts:

157Immunology of melanoma

1. Primary melanomas at times spontaneously undergopartial or complete regression.

2. Primary melanomas often exhibit strong lymphocyticinfiltrations.

3. Nevi at times show a ring of depigmentation (“halo”)around them.

4. Primary melanomas also often show areas ofdepigmentation.

5. Development of vitiligo carries a good prognosis inpatients with melanoma.

Although these features, by themselves, do not prove thatspontaneous regression, halos, depigmentation, or a rela-tionship between vitiligo and good prognosis representunequivocal evidence of host immune responses, the ideathat human melanoma is an immunogenic tumor getsconsiderable support from laboratory observations thathave shown (1) the infiltrating lymphocytes are mostly α/βT cells, custodians on cell-mediated immunity; (2) regressingmelanomas show evidence of clonal amplification of T cells,in situ; and (3) T cells isolated from regressing melanomasexhibit cytolytic activity against autologous melanomas.1-3

With this preamble, the discussion will now be moved to thereview of the evidence supporting the contention thatpatients with melanoma indeed respond to their melanomaswith immunological tools, to the topic of the nature of the“melanoma antigens” that are targeted by the immuneapparatus, and then to an examination of how some of thelaboratory findings have been translated to the clinics. Butbefore moving into these subjects, it is useful to briefly pointout the critical elements of host immunity and to provide ahistorical overview of seminal pieces of work that led to thesearch for immune response against tumor, in general, andagainst melanoma, in particular.

Nature of the host immune responseto melanoma

Host immune responses encompass 2 discreet butinterrelated immunological systems: (1) innate system and(2) adaptive system.

Multicellular organisms acquired an innate system asprotection against invasion of pathogens quite early inontogeny. The adaptive immune system was grafted in byevolution later to improve the system and to provide precisionor “specificity”: the hallmark of adaptive “immunity.”Although this preamble seems to suggest that the immunesystemwas engrafted to protect us from invasion by pathogens,Paul Ehrlich, the visionary thinker as well as experimentalist,envisioned that the system might also have a role in protectionagainst cancer.4 He carried out a series of experiments usingimmunization with necrotic tumor tissues followed bytransplantation of viable tumor as an experimental techniqueand showed that such immunization could protect nearly 50%of the immunized mice from tumor growth4 and firmly planted

the seed of our current belief that a host is capable of mountingan immune response against cancer. Ehrlich also envisionedthe existence of both innate and acquired immunity againstcancer. Unfortunately, he could not establish the critical issueof specificity, as his experiments were done in wild-type mice;inbred mice were not available in his time. Accordingly, hisexperiments could not rule out the possibility that tumor graftrejections in his mice could have simply resulted from“allogeneic response” and not from a tumor-specific immuneresponse. Interestingly, almost half a century later, usingchemically induced tumors and using the same immunizationand tumor transplantation technique in mice, but this time ininbred mice, tumor immunologists finally provided thedefinitive proof of “specificity” in tumor rejection by showingthat chemically induced tumors in mice harbored individuallyspecific transplantation antigens unique for a given tumor.5

The very idea of “tumor immunity” was, however, soonquestioned byHewitt et al,6 who found no evidence of immuneresponse in mice when experiments were performed in inbredmice using “spontaneously” grown mouse tumors. Hewett etal's6 questioning of the existence of tumor immunity, however,did not deter investigators from looking for true tumorimmunity and from moving the issue from mice to humans,as our understanding of the immune system improved and thetools and techniques became available for critical experimentsin human systems. A number of important developments in thefield (serological techniques for careful examination of thequestion of whether or not a host does synthesize a relevantclass of antibody that would bind to and react against the host'sown cancer cells; technology for addressing the same questionwith T cells [ie, does a patient have T cells that wouldrecognize the patient's own tumor cells with a modicum ofspecificity]; recognition of natural killer [NK] cells as animportant custodian of cell-mediated innate immunity, and soforth) allowed careful examination of all facets of hostimmunity against human tumors. For reasons not entirelyclear but perhaps because melanoma cells could be culturedwith relative ease, a good deal of the work on tumor immunitywas conducted in the melanoma model. In the followingsections, I address the following 3 basic issues using humanmelanoma as the model human cancer:

Do NK cells play any role in antimelanoma immunity?Do patients mount a serological response against melanoma?Do melanoma patients bear T cells capable of recognizingpatients’ melanoma cells?

After dealingwith these 3 issues, I briefly address the obviousquestion: If the answer to these questions is yes, why does ourimmune system, then, fail to protect us from melanoma?

Innate immunity against melanoma

Although there are cells other than NK cells thatparticipate in innate immunity, the present discussion is

158 B. Mukherji

limited to NK cell–mediated immune response to melanomaand the role of NK cells in surveillance against melanoma.

Historically, NK cells were defined by their ability to lysea variety of tumor cell lines, in general, and certain lymphoidcell lines that did not express MHC class-1 molecules, inparticular, without requiring prior activation.7,8 NK cells donot posses activation receptors, such as through which T cellsare activated and with which T cells recognize their targets incontext to MHC molecules: MHC class I for CD8+ T cellsand MHC class II for CD4+ T cells. Early efforts atunderstanding the NK cell biology and mechanism ofactivities led to 2 sets of understanding: (1) that NK cellssomehow recognize the lack of MHC protein expression ontarget cell surface and that it is this “missing self” thattriggers them9; and (2) they have a role in immunesurveillance especially against cancer. Subsequent work onNK cells revealed that (1) NK cells express both MHC classI–specific “inhibitory” receptors, such as KIR, NKJG2A,and so forth,10,11 as well as “activation” receptors, such asNKp46, NKp30, NK p44, and so forth, but NK cell receptorsare not like α/β or γ/δ receptors on T cells; and (2) thesereceptors are not uniquely expressed by NK cells as someT cells also express them, including expressing NKG2D, andas such some T cells share properties of NK cells.10,11

Unfortunately, although the nature of the ligands for a T-cellreceptor (TCR) is well defined (ie, TCRs recognize shortpeptides displayed on a given MHCmolecule) the ligands forNK cell receptors (NCR) to date remain undefined with theexception of the ligands for activation receptor, NKG2D,expressed by NK cells as well as NK T cells that happen to bethe stress-inducible proteins MICA/B and ULBPs.10,11

Although the ligands for most of the NK cell–determinedactivation remains undefined, it is widely believed that NKcell receptors primarily recognize “stressed” cells (eg, tumorcells or virus-infected cells). Interestingly, it is from thisfinding and from a number of other pieces of evidence, a rolefor NK cells in tumor immunity is envisioned. Finally,regarding the biology of NK cells, although the descriptionof NK cell as “naturally” cytolytic for tumor cells continuesand the description of inhibitory receptors and certain HLAallele-based inhibition of NK cells remain well established, itshould be mentioned that NK cells can also be “unlicensed”to kill as well as “licensed” to kill through a set of poorlyunderstood mechanisms independent of HLA class Iexpression or “HLA-noninhibited killing.”10-12 Becauseour topic is the immunology of melanoma, the rest of thissection will be devoted to examining the role for NK cells inimmune responses against melanoma, specifically againsthuman melanoma.

NK cell activities against human melanoma

Historically, evidence for NK cell activities againstmelanoma were established mostly in cell-mediated micro-cytotoxicity (CMC) assays using NK cells “purified” from

peripheral blood of patients with melanoma, as large granularlymphocytes (LGL) and tested primarily against melanomacell lines as targets along with the favorite NK target lineK-562. These experiments showed that purified NK cells areindeed capable of killing K-562 cells, as well as humanmelanoma cells particularly that have lost MHC expressionor those that express low levels of MHC molecules.10-12

Given that human tumor cell lines, including melanoma celllines, often express lower levels of MHC molecules,selectively or even globally, whether resulting from thelack of expression of beta 2 microglobulin or fromnonexpression of the genes coding for specific MHCmolecules, the concept that NK cells have a role in immunityagainst melanoma was solidified. This conclusion gotsupport from other studies that showed that NK cells wereable to lyse MHC class I–expressing autologous melanomacells,12 that they could prevent growths of human melanomagrowths in SCID mice,13 and that NK call-mediatedactivities could “influence” certain features of melanomacells, such as growth, levels of invasion, metastatic spread,and so forth, all providing indirect evidence that NK cellsmight have a role in melanoma immunity.12,13 Althoughthese studies were not indisputable evidence for a role forNK cells in melanoma immunity, they were enough toprovide a rationale for clinical trials of adoptive cell therapy(ACT) with NK cells in patients with melanoma. The earlyresults, however, were disappointing. It should be pointedout that the lack of clinical results from the early NK cell–based ACT does not rule out a role for NK cells in melanomaimmunity. A number of factors could explain failure toinduce a therapeutic effect from NK cell–based ACT. Muchmore work will be needed to establish a role for NK cell–based ACT. In fact, such efforts are under way. Time will tellwhether ACT with NK cells might be able to elicit clinicalactivities similar to what has been observed from ACT withT cells.

Serological responses to melanoma

The idea that patients with cancer are capable of mountingserological responses by making antitumor antibodies is anold idea. Indeed as soon as technologies, whether immuno-flurescence-based microscopy, complement fixation assay,and immune adherence assay, became available, investiga-tors began looking for antimelanoma antibody in sera frompatients with cancers including melanoma. Not unexpected-ly, early experiments did reveal circulating antibodies againstmelanoma cells in patients with melanoma.14 Unfortunately,as a great majority of the initial studies were performed withallogeneic cell lines and despite extensive “absorption” of thesera with nontumor tissues, the “specificity” of the reactioncould not be convincingly established. To get around theproblems inherent in doing serology in allogeneic set ups,Lloyd Old's group initiated a rigorous examination ofserological responses in cancer patients using autologous

159Immunology of melanoma

tumor cells and extensive absorption with autologous normaltissues, a technique they termed autologous typing. With thisstrategy, his group was able to establish the following points:

1. Patients with cancer, including patients with melano-ma, do produce immunoglobulin (Ig)M as well as IgGantibodies against cell surface–associated antigens onautologous tumor cells including melanoma.

2. Although the specificities were often found to beshared, between tumor cells of different types as wellas between tumor cells and normal cells, individual or“unique” specificity for autologous tumors, especiallyin the melanoma model, also existed.

Unfortunately, although autologous typing revealedspecificity in serological response in a few cases, the natureof the antigens recognized remained mostly undefined with acouple of exceptions. Autologous typing led to therecognition of a ganglioside as cell surface antigens onmelanoma cells15 and also defined a 90,000-Da cell-surfaceglycoprotein as a potentially unique melanoma antigen.16

Around the same time, autologous typing and serologicalstrategies with polyclonal and monoclonal antibodies raisedagainst tumor cells led to the recognition of a number ofantigens, especially that of a series of ganglioside antigens.

The field of serology against tumor antigen witnessed aremarkable development with the advent of SEREXtechnology.17 Soon the SEREX technology defined a hostof antigens in a variety of human tumors includingmelanoma. Although many SEREX-defined antigens, in avariety of tumors, including melanoma, turned out to benonspecific or shared, “unique” antigens were also defined.18

One of the most important findings that emerged fromSEREX-based work is that a fair number of antigens thatwere recognized by T cells were also recognized serologi-cally and that the serological response included IgG-typeresponses. Because the generation of IgG response requiresCD4+ T-cell help, this methodology provided the evidencethat CD4+ T cells are also involved in immune responsesagainst melanoma! This aspect of dual serological as well asT-cell–based host immune responses against melanoma isdiscussed in the following section.

Cell-mediated immune responses in melanoma

Ehrlich, the father of tumor immunology, predicted thatthe host immune responses against tumors are likely to bemediated by cells rather than by antibodies.4 It is worthremembering that at Ehrlich's time, neither NK cells norT cells were identified as components of the immune systemand the remarkable intricacies involved in the workings ofthe system were not known. Nonetheless, his seminal workhelped establish the foundation for transplantation biologiststo address the issue of tumor immunity in mice withchemically induced tumors. A series of seminal work from a

number of investigators revealed tumor-specific immuneresponse in mice immunized by and then grafted with thesame chemically induced tumors. The phenomenon wastermed tumor-specific transplantation immunity and theantigen evoking the response was called tumor-specifictransplantation antigen or TSTA.5,14 Interestingly, althoughHewitt et al's subsequent observations with spontaneouslygrown mouse tumors6 questioned the very foundation ofantitumor immunity, search for cell-mediated immuneresponses continued in mice as well as in humans. Becausetransplantation experiments similar to that could be done inmice but could not be performed in humans, the experimentalapproaches involved mostly in vitro assays. Although anumber of assays were used to monitor cell-mediatedimmune responses against cancer cells, the advent of cell-mediated microcytotoxicity (CMC) assay led to a great dealof attention on finding the evidence of cellular reactivityagainst tumor cells in CMC assays. As pointed out earlier,perhaps because melanoma cell lines could be fairly easilyestablished, a good deal of work on cell-mediated immunityin humans was done and continued to be done in themelanoma model.

Using CMC assay, several groups showed that T cellsisolated from patients with cancer, including patients withmelanoma (from blood as well as from excised melanomatissues), were capable of killing allogeneic as well asautologous tumors.19 Here again, just as the task ofestablishing specificity was a problem with early serologicalanalyses, the question of specificity in cell-based immuneresponses, observed in CMC assay, or in any other assay,using polyclonal T cells at the population level remainedproblematic. However, as interleukin (IL)-2 became avail-able, 2 parallel strategies, one involving growing tumor-infiltrating lymphocytes (TILs), championed by Rosenberg'sgroup at the National Cancer Institute (NCI),20 and anotherinvolving establishing monoclonal T-cell lines from patientswith melanoma,21-23 produced much better results. Indeedcareful functional analyses with TILs and T-cell clonesderived from blood or from tumor tissues carried out againsta panel of autologous as well as allogeneic target cellsaffirmed the core principle that patients with melanoma doindeed harbor T cells capable of recognizing autologousmelanoma cells as potential immunological targets.24 Offurther interest, the strategy also soon led to the identificationof the nature of the antigens recognized by such T-cell lines,as well as led to figuring out the molecular nature of theT-cell receptors (TCRs) involved in such recognition. Moreimportantly, both strategies also led to the development ofadoptive cell therapy (ACT) for melanoma.

CD8+ cytolytic T-lymphocyte responseagainst melanoma

A great deal of interest on T-cell responses againstmelanoma has been predominantly restricted to CD8+ T-cell

160 B. Mukherji

response, as most melanomas, like other nonlymphoidtumors, do not express MHC class II molecules and expressonly MHC class I. Thus, one of the major tasks for tumorimmunologists was to show that tumor-bearing hosts doindeed harbor CD8+ T cells capable of recognizingautologous tumor cells with a measure of specificity. Asmentioned earlier, several laboratories were able to establishmonoclonal populations of CD8+ T cells derived from singleT cells and were able to demonstrate that some of the CD8+T cells selectively recognized autologous melanoma cells inmicrocytotoxicity assays. Although continued work withcloned CD8+ cytolytic T lymphocytes (CTL) would expandour understanding of selective versus shared specificity, theseexperiments provided the proof that patients with melanomado harbor monoclonal T cells capable of recognizingautologous melanoma cells.24 Figure 1 shows a visualdepiction of killing of melanoma cells by autologous CTLs.

The technique of T-cell cloning, whether from peripheralblood, from draining lymph nodes, or from tumor tissuesthemselves, allowed investigators to address a number ofrelated issues such as function (other than cytotoxicity),specificity, and, most importantly, to address the nature ofthe antigen recognized by such T cells. Apart fromcytotoxicity, melanoma-reactive CTLs elaborate a numberof cytokines, such as tumor necrosis factor α, interferon γ,granulocyte-macrophase colony-stimulating factor, and IL-2.The specificity is not always solely restricted to autologousmelanoma cells, as many cloned CTL lines also recognizeallogeneic melanoma cells in a strict MHC-restricted mannergiving rise to the notion that some melanoma antigens are notuniquely restricted to a given melanoma and may representshared antigens.24

CD4+ T-cell response against melanoma

As already noted, the interest in CTL response inmelanoma was because most melanoma cells express onlyMHC class I molecules; however, some melanomas express

Fig. 1 T lymphocytes melanoma cell interaction. T lymphocytes clonedwith autologous melanoma cells. Panel A shows multiple T cells formingshows a T–melanoma cell conjugate stained with trypan blue 4 hours aftnegative (ie, alive), the melanoma cells are trypan blue positive (ie, dead

MHC class II molecules and CD4+ T cells can surelyrecognize a melanoma-associated peptide epitope processedand presented by antigen presenting cells (APCs). That this isindeed the case can be established from the fact thatserological host response through IgG antibodies tomelanoma antigens exist; as the IgG response requiresCD4+ T cell help, one can envision CD4+ T-cell response tocertain melanoma antigens; however, long before SEREXtechnology provided support for the existence of CD4+T-cell–mediated response against melanoma, several labo-ratories were able to demonstrate CD4+ T-cell responses inthis tumor model.25,26 The concept of CD4+ T-cell–mediated responses in melanoma received more solidsupport as the T-cell cloning technology led to the generationof CD4+ T-cell lines that recognized melanoma targets in anMHC class II–restricted manner showing both helper as wellas regulatory function.25 Indeed, these studies revealed forthe first time that CD4+ T cells can “suppress” CD8+ T-cellresponses in human systems.25 Interestingly, despite thecanonical view on the separation of function between CD8+and CD4+ T cells (ie, CD8+ T cells are cytolytic whereasCD4+ T cells are “helper” cells), recent work with TCR-engineering of CD4+ T cells has revealed that CD4+ T cellscan be made to perform multiple functions, including killing.

Although the literature provides ample support for CD4+T-cell–mediated immune response against a number ofmelanoma-associated antigens, it should be pointed out thatthe issue of CD4+ T-cell response against melanoma, ingeneral, has been conceptually somewhat difficult toenvision, as most melanoma cells, like other nonlymphoidtumor cells, do not express MHC class II molecules. Indeed,this is precisely why the task of engaging CD4+ T cells inmelanoma immunity has been a difficult proposition. Lately,given the interest in engaging CD4+ T cells in tumorimmunity, several laboratories have taken the strategy ofgenerating large numbers of “monoclonal CD4+ T cells”specific for MHC class I–restricted epitopes by engineeringthem to express a set theα/β chains of a TCR specific for suchMHC class I–restricted epitopes. The strategy led to novel

from circulatory T cells from a melanoma patient were co-culturedconjugates with pigmented melanoma cells in the center. Panel B

er incubation. Although the T cells in the periphery are trypan blue).

161Immunology of melanoma

insights into the biology of such TCR-engineered CD4+T cells. For example, CD4+ T cells engineered to express theHLA-A2.1 restricted MART-26-35 peptide-specific TCR(cloned from aCD8+T cell line) were found to be remarkablymultifunctional. They were found to elaborate a number ofinflammatory cytokines, helped the expansion of the epitope-specific CD8+ T cells, and exhibited cytolytic activities, all inan epitope-specific fashion (ie, recognizing the epitope on theappropriate MHC class I molecules just as CD8+ T cellsdo).27–29 Interestingly, contrary to the conventional beliefthat CD8+ T cells are cytolytic cells and CD4+ T cells arehelper cells, this set of observations clearly suggest thatCD4+T cells can bemade to perform all the functions that areusually ascribed to CD8+ T cells and more (Figure 2)! Thesefindings have revealed a potentially novel mechanism forengaging CD4+ T cells in tumor immunity and clinical trialswith such TCReng CD4+ T cells have begun.

T-cell receptors recognizing melanoma antigens

As soon as melanoma antigen-specific T-cell clonesbecame available, work began on defining the nature ofTCRs that recognize melanoma antigens. Initial expectationwas that these receptors might exhibit common structural

Fig. 2 Functional analyses of MHC class I-restricted MART-1 epitopeT cells. Panel A shows cytokine synthesis by F5 CD4+ T cells, and pansynthesis by F5 CD8+ T cells, and panel D shows cytolytic activity ofmelanoma cell targets, A375 is an HLA-A2–positive but MART-1–negatanti-class I antibody; A375-M1 represents A 375 cells engineered to exp

properties. Interestingly, collective evidence on the subjectrevealed that whereas TCRs for a given epitope mightexhibit some preferential use of certain variable regions,the usage of the various β chain and sequences isremarkably broad.30,31 So much so that a given patientswith melanoma can harbor T-cell clones expressingmultiple segments and sequences for a single melanomaepitope. Nonetheless, from a therapeutic viewpoint,considerable interest now exists to find the most avidTCR specific for an epitope of choice for designing therapywith engineered T cells.

Nature of melanoma antigens

Long before the advent of T-cell clones available forprobing the nature of immunologically relevant melanomaantigens, efforts were under way to define melanomaantigens. Using primarily serological approaches, whetherwith sera from patients with melanoma or with monoclonalantibodies, several groups of investigators have attempted todefine melanoma antigen in molecular terms. Although theseefforts led to finding a variety of “melanoma antigens, thediscoveries of ganglioside antigens were an importantmilestone on our search for melanoma antigens.32,33

-specific TCR-engineered CD4+ (F5CD4+) and CD4+ (F5 CD8+)el B shows cytolytic activity of the same; panel C shows cytokinethe same (PT-M and M202 are HLA-A2– and MART-1–positiveive melanoma target, PT-M+ anti-class I represents PT-M cells withress the MART-1 gene).

162 B. Mukherji

Ganglioside antigens on melanoma cells

Gangliosides are glycophosphingolipids containing neur-aminic acid. This class of tumor-associated antigens wasfirst discovered as oncofetal antigens using antibody madefrom a lymphoblastoid B-cell line.32 The oncofetal antigenI was soon identified as a ganglioside antigen33 andsubsequent studies by several groups defined a number ofdifferent types of ganglioside (eg, GM2, GD2, GD3) onhuman tumor cell surfaces. Soon it was found that patientswith cancer, including patients with melanomas, makehumoral immune responses to ganglioside.15,34,35 Theseobservations led to the recognition that gangliosides dohave therapeutic potential and thereafter clinical trials withganglioside vaccines revealed their immunogenic as well astherapeutic potential.36 A detail discussion of variousclinical trials with ganglioside antigens is beyond thescope of this paper; however, it should be mentioned thatganglioside antigen-based vaccines in melanoma as well asin other human cancers continue to be pursued withconsiderable interest.

Table 1 Examples of melanoma-associated antigens,peptides, and MHC molecules presenting those peptides

Proteins Peptides PresentingMHC

MAGE-A1 a EADPTGHSY HLA-A1 & B37MAGE-A1 a TSCILESLFRAVITK HLA-DP4MAGE-A1 a EYVIKVSARVRF HLA-DR15MAGE-A3 a EVDPIGHLY HLA-A1MAGE-A3 a FLWGPRALV HLA-A2MAGE-A3 a VIFSKASSSLQL HLA-DR4NY-ESO-1 a SLLMWITQC HLA-A2NY-ESO-1 a MPFATPMEA HLA-B51NY-ESO-1 a EFYLAMPFATPM HLA-DR1Melan-A/MART-1 b

(E)AAGIGILTV HLA A2

Melan-A/MART-1 b

EAAGIGILTV HLAB35

Melan-A/MART-1 b

ILTVILGVL HLA-A2

Melan-A/MART-1 b

AAGIGILTVILGVL HLA-DR1

Tyrosinase b MLLAVLYCL HLA-A2Tyrosinase b SSDYVIPIGTY HLA-A1Tyrosinase b SYLQDSDPDSFQD HLA-DR4gp100/pmel17 b KTWGQYWQV HLA-A2gp100/pmel17 b LIYRRRLMK HLA-A3gp100/pmel1 b GRAMLGTHTMEVTVY HLA-DQ6CDK4 c ACDPHSGHFV HLA-A2beta-catenin c SYLDSGIHF HLA-A24N-ras c ILDTAGREEY HLA-A1

a Shared antigens.b Differentiation (CT) antigens.c Mutated antigens.

T-cell–determined melanoma antigens

The search for the nature of the antigens recognized bysuch CTL clones began shortly after the T-cell cloningtechnology led to the development of melanoma-reactiveCTL clones. The work in this area was pioneered by Boon'sgroup37 and shortly thereafter by the Rosenberg group atNCI.38 Using a CTL clone generated from a patient withmelanoma and using a genetic approach at defining the genethat codes for the antigen recognized by the CTL, Van derBruggen et al37 from Brussels defined the first CTL-determined melanoma antigen, which they called melanomaantigen-1 (MAGE-1). Shortly thereafter, they defined theHLA-A1 restricted peptide epitope that was the ligand for theTCR on the corresponding CTL.39 Most interestingly,neither the sequence of the MAGE-1 gene nor the aminoacid sequence of the peptide ligand for the CTL revealed anymutation! Subsequently, a number of other melanoma-associated antigens and peptide ligands for melanoma-reactive CTL were identified and most of them turned outto be “self” epitopes; later studies revealed unique CTL-determined melanoma epitopes resulting from gene muta-tion. In time, the compendium of CTL-determined melano-ma antigens and peptide epitopes became quite large and thecollective body of information allowed some categorizationbased on the patterns of their expression. Table 1 shows apartial list.

As can be seen, a great majority of these epitopes turnedout to be “self ” epitopes, although “unique” epitopesresulting from mutations (such as β-catenin, N-ras) were alsoidentified. Among the “self” epitopes, some were clearlyassociated with melanocytic differentiation (hence referred

to as differentiation antigens and epitopes) and these epitopesare found on melanoma cells as well as on normalmelanocytes. Melan-A/MART-1, Tyrosinase, gp 100, andTRP are the best examples of differentiation antigens.Another set of “self” antigen/epitopes is found on melanomacells (as well as on other types of malignant cells) plus ontesticular cells. These antigens (eg, MAGE-1, MAGE-3,other family members, such as BAGE, GAGE, NY-ESO-1),are coded by the X chromosome and are also expressed bytestis, hence the name Cancer Testes or CT antigens. CTantigens, however, can be expressed by a variety of tumors.For example, NY-ESO-1, originally discovered in a patientwith esophageal cancer, is also expressed by melanoma andby other types of neoplasms.

As mentioned earlier, a fair number of melanoma antigensand MHC class II–restricted epitopes that are recognized byCD4+ T cells have also been defined. Of interest, as shown inTable 1, several antigens that provide forMHCclass-I restrictedepitopes also provide for MHC class-II restricted epitopes.

Regarding strategies leading to the discoveries of CTL-determined epitopes in melanoma, it should be mentionedthat such epitopes have not been only defined using T-cellclones and genetic approaches, T cell–determined peptides

163Immunology of melanoma

have also been eluted from melanoma cells, providing thestrongest evidence that melanoma cells do naturally displaypeptide epitopes for melanoma antigen-specific T cells.40

Clearly, these observations have been the strongest evidencesupporting the notion that melanoma cells display targetableas well as immunogenic peptide epitopes.

Needless to mention, the list of melanoma antigens forCD8+ T cells as well as for CD4+ T cells continues toexpand at a steady pace along with continued interest inmelanoma immunotherapy. Melanoma immunology is nolonger simply restricted to studying the topic at the bench.Virtually all the observations made at the bench have nowbeen translated in the clinic and intense effort is under way tomake cancer immunotherapy, in general, and melanomaimmunotherapy, in particular, more effective.

Immune evasion by melanoma

Although the preceding discussion makes it amply clearthat a host with melanoma (and any other tumor for thematter) harbors all the tools needed for mounting immuneresponses against his or her melanoma cells. How domelanoma cells (and any other tumor cells), then, escapeimmunological destruction? The topic of immune evasion bymelanoma cells is worthy of a separate paper and it is beyondour present scope; however, a short synopsis of thisimportant issue is provided here.

This topic needs to be addressed from 2 separateviewpoints: (1) immune escape during early growth phase(ie, in the early period of tumor development), and (2)immune evasion during continued growth, metastasis, and soforth. No matter how many processes contribute to evasionduring both phases, it requires no leap of imagination torecognize that the ultimate result (ie, escape from immunity)has to result from:

1. Intrinsic control of the immune system from reacting against“self.” Tumors arise in vivo (within), and as such, mostcancer cells are also “self”; they do not threaten the individualimmediately, as invasion from exogenous infectious agentsdo, and as nature has engrafted a mechanism for tolerating allthings “self,” tumor cells also seem to get “tolerized.”

2. Tumor cells orchestrating a series of evasive tactics.

On tolerance for tumors, one point is clear. It is thattolerance induction for in vivo–grown tumor cells is not dueto central deletion of T cells capable of recognizing antigenson melanoma cells. Research on tumor immunology hasprovided convincing proof against the textbook descriptionthat “all self-reactive T cells are centrally deleted.”Melanoma antigen-specific T cells do exist but apparentlythey exist in a tolerant state, unable to prevent tumor cellgrowth, during inception and subsequently after initiation. IfT cells with TCRs capable of recognizing targetable epitopeson melanoma cells do exist, why can't they do their job?

Work over several decades on this issue suggests severalreasons. First, the TCRs on the T cells bearing TCRs formelanoma antigens that gain access to the periphery are oflow affinity, which is the reason for not being centrallydeleted. Under physiological conditions, they do not getactivated. Even if they are activated, they do not seem toreach proper amplification and/or to acquire the proper“fitness” to become efficient effector cells. Although thesearguments seem attractive, recent observations on “immuneediting” during T-cell–mediated surveillance seem tosuggest that the reason for escape is dictated by the immuneresponse itself. Indeed, considerable evidence for the ideathat immune response leads tumor cells to “edit” theirimmunological make up has emerged.41 The phenomenon of“immune editing” will be brought up later in conjunctionwith the discussion on tactics tumor cells use to evadeimmune responses.

Returning to “intrinsic” factors within the immune systemresponsible for immune escape by tumor cells, severalinherent conditions within the biology of T-cell functionseem to work against effective removal of tumor cells.Among the various conditions potentially contributing tofailure of T cells to effectively eradicate tumors, 3 conditionsmerit discussion. First, the very process of T-cell activationand expansion following exposure to antigens underappropriate physiological condition (ie, the antigen beingpresented by professional antigen-presenting cells), takesplace under strict internal regulatory processes. For example,following priming, not only do the kinetics of T-cell responsefollow a bell curve (amplification followed by contractionunder a strict set of intrinsic genetic rules); a fraction of fullyactive T cells undergoes apoptosis following antigenexposure at the effector phase, referred to as activation-induced cell death. Further, following activation, T cellsdisplay the cell surface molecule, cytolytic T-lymphocyteactivation 4 (CTLA4) that competes with CD28 forcostimulatory molecules and sends a negative signal onligation with the costimulatory molecules on APC. Addi-tionally, all immune responses undergo “contraction”following amplification. During contraction, the expandedT cells undergo apoptosis as a form of programmed cell death.Interestingly, all these mechanisms are exquisitely regulatedby specific proteins intrinsic to the T cells; these functions aregenetically controlled as checks and balances againstcontinued T-cell expansion and consequent unwarrantedside effects of continued inflammation. Although these are allvery good and proper in immune response against infectiousagents, they surely do not help the cause of tumor immunity.

T cells often develop a condition known as “exhaus-tion” after repetitive antigen exposure. Several molecules(eg, PD1, TIM3) have been identified as markers for“exhaustion.” Exhaustion in adoptive T-cell–based immu-notherapy for melanoma has also been encountered.Understandably, efforts are under way to elucidate theprecise mechanism underlying exhaustion and to intervenewith it in melanoma immunotherapy.

164 B. Mukherji

Additionally, the immune system itself has also providedanother layer of control for continued immune response andconsequential side effects. Indeed all immune responses, ingeneral, and all immune responses against “self,” inparticular, are subject to regulation by regulatory mecha-nisms (eg, by T regulatory cells, myeloid suppressor cells).Several lines of investigation in mice and humans haveshown T-cell–mediated immune responses against cancerare also regulated. Of the various populations of cellsinvolved in regulation, T-regulatory cells (Treg) and myeloidsuppressor cells have assumed considerable importance. Alarge literature on Tregs (natural or inducible), on myeloidsuppressor cells, and on their mechanisms of action inantitumor as well as in antimelanoma immunity exists42,43

and intense interest has been generated for developing waysto mitigate the negative effects of Tregs and myeloidsuppressor cells in antitumor immune responses.

Finally, it should be mentioned that melanoma cells (aswell as other tumor cells for that matter) use a number oftactics to evade immune responses. This subject has alsoreceived extensive scrutiny and studies in this area haverevealed several mechanisms that melanoma cells use toavoid immune recognition and destruction.44,45 In general,melanoma cells, like other neoplastic cells, use 2 basicstrategies: (1) to change their make up so as to makethemselves invisible to the immune system, and (2) toelaborate a host of “immunosuppressive” cytokines to thwartthe immune-based effector activities. As far as changing theirmakeup, the phenomenon of immune editing, mentionedearlier, essentially represents a form of “self-sculpting” bythe melanoma cells primarily to avoid continued immuneattack. Although the phenomenon of immune editing hasreceived considerable attention as an underlying mechanismof immune evasion during the inception phase of tumorgrowth, melanoma cells continue to sculpt themselves inseveral different ways through their lifespan. For example,melanoma cells have been known to downregulate MHCmolecules, either selectively, sequentially, or totally to avoiddisplaying targetable epitopes. Similarly, they also have beenfound to downregulate the expression of gene coding for agiven antigen. And, at times, they have been found to doboth, all to avoid immune destruction. Melanoma cells arecapable of elaborating a number of immunosuppressivecytokines, such as IL-10 and transforming growth factor β.Thus, by not displaying targetable markers and byelaborating immunosuppressive cytokines, melanomas alsoco-opt these 2 fundamental tricks often used by invadingpathogens when subjected to selective pressure.

Translational immunology in melanoma

Since the topic of immunotherapy for melanoma is withelsewhere in this issue, this topic is not discussed in anylength here; however, because the ultimate goal of tumorimmunology is to develop immunotherapeutic strategies, a

few brief remarks are appropriate. In earlier times, seriousimmunotherapeutic strategies for melanoma included,among other things, BCG, interferon, and then IL-2.Although all these agents are approved therapies for cancerand have shown some degrees of effectiveness, their overalleffectiveness in melanoma has been somewhat modest.Recent immunotherapeutic strategy, to a great extent, isbased on the concept of specificity (ie, active specificimmunization and adoptive T-cell–based therapy). In thiscontext, as CTL-determined melanoma-related peptideepitopes got defined, translational therapy with immuniza-tion with peptide/APC or peptide alone in melanomabegan.46,47 Similarly, as in vitro culture technique forgenerating large numbers of melanoma-reactive T cellswere worked out, adoptive cell therapy for melanomaentered into translational medicine.48 Both strategies areexperimental, both have shown some promise, but morework is needed to improve their effectiveness. Fortunately,intense efforts are under way in pursuance of that goal.There is reason to believe that continued research is likely tomake them more effective.

Acknowledgment

The author thanks Arvind Chhabra for constructive helpin preparation of the paper and regrets that many importantcontributions in the field by many investigators could notbe cited.

Supported by Public Health Service grant 2 R01 CA88059 (B.M.) and by CA 1P01CA132681 (David Baltimore,PI), UCHC Sub Award: 21B-108893 (B.M.).

References

1. Mackensen A, Ferradini I, Carcelain IG, et al. Evidence for in situamplification of cytolytic T-lymphocytes with antitumor activity in ahuman regressing melanoma. Cancer Res 1993;53:3569-73.

2. Ferradini I, Mackensen A, Geneve C, et al. Analysis of T cell receptorvariability in tumor-infiltrating lymphocytes from a regressive mela-noma. J Clin Invest 1993;91:1183-90.

3. Mackensen A, Carcelain S, Viels S, et al. Direct evidence to support theimmunosurveillance concept in a regressive melanoma. J Clin Invest1994;93:1397-402.

4. Ehrlich P. Ueber den jeitzigne stand der karzinomforschung. Neder-landsch Tijdschrift voor Geneeskunde 2011;5:273-90.

5. Klein G. Tumor antigens. Annu Rev Microbiol 1996;20:223-52.6. Hewitt HB, Blake ER, Walder AS. A critique of the evidence for active

host defense against cancer, based on personal studies of 27 murinetumours of spontaneous origin. Br J Cancer 1976;33:241-59.

7. Herberman R, Ortaldo J. Natural killer cells: their role in defenseagainst disease. Science 1981;214:24-9.

8. Trinchieri G. Biology of natural killer cells. Adv Immunol 1989;47:225-74.

9. Karre K. Natural killer cell recognition of missing self. Nat Immunol2008;9:477-80.

10. Lanier LL. NK cell recognition. Annu Rev Immunol 2005;23:225-74.

165Immunology of melanoma

11. Moretta A, Bottino C, Marcenaro E, et al. Activating receptors andcoreceptors involved in human natural killer cell-mediated cytolysis.Annu Rev Immunol 2001;19:197-223.

12. Carrega P, Pezzino G, Queirolo P. Susceptibility of human melanomacells tom autologous natural killer (NK) cell killing: HLA-relatedeffector mechanisms and role of unlicensed NK cells. PLoS One 2009:4e8132.

13. Lozupone F, Pende D, Burgio VL, et al. Effect of human natural killerand gamma delta T cells on the growth of human autologous melanomaxenografts in DCID mice. Cancer Res 2004;64:378-85.

14. Old LJ. Cancer immunology: the search for specificity. G. H. A. ClowesMemorial lecture. Cancer Res 1981;41:361-75.

15. Lloyd KO, Old LJ. Human monoclonal antibodies to glycolipids andother carbohydrate antigens: dissection of the humoral immuneresponse in cancer patients. Cancer Res 1989;49:3441-51.

16. Real FX, Jules Mattes M, Houghton A, et al. Class 1 (Unique) tumorantigens of human melanoma. J Exp Med 1984;160:1219-33.

17. Pfreundscuh M. Exploitation of the B cell repertoire for theidentification of human tumor antigens. Cancer Chemother Pharmacol2000;46:3-7.

18. Sahin U, Tureci O, Schmitt H, et al. Human neoplasms elicit multiplespecific immune responses in the autologous host. Proc Natl Acad SciU S A 1995;92:11810-3.

19. Hellstrom KE, Hellstrom I. Cellular immunity against tumor antigens.Adv Cancer Res 1969;12:167-214.

20. Rosenberg SA, Spies P, Lafreniere R, et al. A new approach to theadoptive immunotherapy of cancer with tumor infiltrating lymphocytes.Science 1986;233:1318-21.

21. Mukherji B, MacAllister TJ. Clonal analysis of cytotoxic T cellresponse against human melanoma. J Exp Med 1983;158:240-5.

22. Knuth A, Danowski B, Oettgen HF, et al. T cell mediatedcytotoxicity against autologous malignant melanoma: analysis withinterleukin-2 dependent T cell cultures. Proc Natl Acad Sci U S A1984;81:3511-5.

23. Kawakami Y, Suzuki Y, Shofuda T, et al. T cell immune responsesagainst melanoma and melanocytes in cancer and autoimmunity.Pigment Cell Res 2000;13:163-9.

24. Boon T, Coulie PG, Van den Eynde BJ. Human T cell responses againstmelanoma. Annu Rev Immunol 2006;24:175-208.

25. Mukherji B, Guha A, Chakraborty NG, et al. Clonal analysis ofcytotoxic and regulatory T cell responses against human melanoma.J Exp Med 1989;169:1961-76.

26. Topalian SL. MHC class II restricted tumor antigens and the role ofCD4+ T cells in cancer immunotherapy. Curr Opin Immunol 1994;91:741-5.

27. Clay TM, Custer MC, Sachs J, et al. Efficient transfer of a tumorantigen-reactive TCR to human peripheral blood lymphocytes conferantitumor reactivity. J Immunol 1999;163:507-13.

28. Chhabra A, Yang P, Wang B, et al. CD4+CD25- T cells transduced toexpress MHC class-I restricted epitope specific TCR synthesize TH1cytokines and exhibit class-I restricted cytolytic effector function in ahuman melanoma model. J Immunol 2008;181:1063-70.

29. Ray S, Chhabra A, Chakraborty NG, et al. MHC-I-restricted melanomaantigen specific TCR-engineered human CD4+ T cells exhibit multi-functional effector and helper response, in vitro. Clin Immunol 2011;136:338-47.

30. Sensi M, Salvi S, Casteli C, et al. T cell receptor (TCR) structure ofautologous melanoma reactive cytotoxic T lymphocytes (CTL) clones:tumor infiltrating lymphocytes over express in vivo the TCRβ chainsequence used by an HLA-A2-restricted and melanocyte lineagespecific CTL clone. J Exp Med 1993;178:1231-46.

31. Shilyansky J, Nishimura M, Yanelli J, et al. T cell receptor usage bymelanoma specific clonal and highly oligoclonal tumor infiltratinglymphocyte lines. Proc Natl Acad Sci U S A 1994;91:2829-33.

32. Irie RF, Sze LL, Saxton RE. Human antibody to OFA-1, a tumorantigen produced in vitro by Epstein-Barr virus-transformed humanB-lymphoblastoid cell lines. ProcNatl Acad SciU SA 1982;79:5666-70.

33. Cahan LD, Irie RF, Sing R, et al. Identification of a humanneuroectodermal tumor antigen (OFA-I-2) as ganglioside GD2. ProcNatl Acad Sci U S A 1982;79:7629-33.

34. Tsuchida T, Saxton RE, Morton DL, et al. Gangliosides of humanmelanoma. J Natl Cancer Inst 1987;78:45-54.

35. Furukawa K, Lloyd KO. Gangliosides in melanoma. In: Ferrone S,editor. Human Melanoma: Immunology, Diagnostics, and Therapy.Heidelberg: Springer-Verlag; 1990. p. 15-30.

36. Livingston PO. Approaches to augmenting the immunogenicity ofmelanoma gangliosides: from whole melanoma cells to ganglioside-KLH conjugate vaccines. Immunol Rev 2006;145:147-66.

37. Van der Bruggen P, Traversari C, Chomez P, et al. A gene encoding anantigen recognized by cytolytic T lymphocytes on a human melanoma.Science 1991;254:1643-7.

38. Kawakami Y, Rosenberg SA. Human tumor antigens recognized byT cells. Immunol Rev 1997;16:313-39.

39. Traversari C,Van derBruggenP, Luescher IF, et al.Anonapeptide encodedby humanMAGE-1 is recognized on HLA-A1 by cytolytic T lymphocytesdirected against tumor antigen MZ2-E. J Exp Med 1992;176:1453-7.

40. Hunt DF, Henderson RA, Shabanowitz J, et al. Characterization ofpeptides bound to the class I MHC molecule HLA-A2.1 by massspectrometry. Science 1992;225:1261-9.

41. Dunn GP, Bruce AT, Ikeda H, et al. Cancer immunoediting: fromimmunosurveillance to tumor escape. Nat Immunol 2002;3:991-8.

42. Sakaguchi S. Naturally arising CD4+ regulatory T cells for immuno-logic self-tolerance and negative control of immune responses. AnnuRev Immunol 2004;22:531-49.

43. Nagaraj S, Grabrilovich D. Myeloid-derived suppressor cells. Adv ExpMed Biol 2007;601:213-23.

44. Rabinovich GA, Gabrilovich D, Sotomayor EM. Immunosuppressivestrategies that are mediated by tumor cells. Annu Rev Immunol 2007;5:267-96.

45. Ferrone S, Whiteside TL. Tumor microenvironment and immuneescape. Surg Oncol Clin N Am 2007;16:755-74.

46. Mukherji B, Chakraborty NG, Yamasaki S, et al. Induction of antigen-specific cytolytic T cells in situ in human melanoma by immunizationwith synthetic peptide-pulsed autologous antigen presenting cells. ProcNatl Acad Sci U S A 1995;92:8079-82.

47. Marchand M, van Baren N, Weynants P, et al. Tumor regressionsobserved in patients with metastatic melanoma treated with an antigenicpeptide encoded by gene MAGE- 3 and presented by HLA-A1. Int JCancer 1999;80:219-30.

48. Rosenberg SA, Restifo NP, Yang JC, et al. Adoptive cell transfer: aclinical path to effective cancer immunotherapy. Nat Rev Cancer 2008;8:299-308.