Nature reviews. Cancer.
`. 14, NO. 2 (Feb. 2014)
`~©"8ral Collection
`“VT NA839F
`4-02-14 06:38:28
`
`KEVIEWS
`
`reuruaiy 2014 volume 14 no.2
`“w.nature.com/reviews
`
`CANCER
`
`
`
`
`PROPERTY OF THE
`NATIONAL
`
`‘
`fF
`LIBRARY OF
`
`wich}~~MEDICINE cS
`WANOMAL teIT
`
`
`| Genome & Co.v. Univ.of Chicago
`vc
`040574
`
`NATURE REVIEWS CANCER
`NPGR2019-00002
`
`UNIV. CHICAGO EX. 2027
`
`ig?
`=SoHETmnt
`iv
`z
`
`
`
`
`bt246np frog nding the basics
`a; SREDOSE
`
`
`
`
`
`

`

` CONTENTS
`
`ID parade, p77
`
`REVIEWS
`
`The ID proteins: master regulators of cancer
`stem cells and tumour aggressiveness
`Anna Lasorella, Robert Benezra and Antonio lavarone
`Inhibitor of DNA binding(ID) proteins are
`transcriptional regulators that control the timing of
`cell fate determinationand differentiation in stem
`and progenitorcells. The ability of ID proteins to
`functionas central ‘hubs’for the coordinationof
`multiple cancer hallmarksis establishing them as
`therapeutic targets and biomarkersin specific types
`| of human tumours.
`
`| Pheochromocytomaand paraganglioma
`_ pathogenesis: learning from genetic
`heterogeneity
`Patricia L. M. Dahia
`Patricia L. M. Dahia gives an overviewof insights
`learned from the study of pheochromocytomas and
`-
`| paragangliomas, whichcarry the highest degree of
`| heritability of all human tumours.
`
`ARTICLE
`
`FEATURED
`
`Paediatric and adult glioblastoma:
`multiform (epi)genomic culprits emerge
`Dominik Sturm, Sebastian Bender, David T. W. Jones,
`PeterLichter, Jacques Grill, Oren Becher,
`Cynthia Hawkins, Jacek Majewski, Chris Jones,
`Joseph F. Costello, Antonio lavarone, Kenneth Aldape,
`Cameron W. Brennan, Nada Jabado and
`Stefan M.Pfister
`Thelatest large-scale genomic and epigenomic
`profiling studies have yielded an unprecedented
`abundanceof novel data andprovided deeper
`insights into gliomagenesis across all age groups.
`Thesestudies have highlighted key distinctions, but
`also some commonalities, which are discussed in
`this Review,
`
`fs Germline mutation|
`(a Somatic mutation |
`
`
`li, i Ball. sg
`
`|
`
`and readers to
`
`@© SERIES ON CLINICAL INSIGHTS
`On the web www.nature.com/reviews/cancer
`mentarticles that focuses on thought-provoking
`Aspecial series of Com
`
`found at http://www.nature.com/nre/series/clinin
`Ady
`i
`clinicalfindings can be
`
`: an online publication
`oe ve an advance online publication (AQP)service for authors
`Links to further information
`The full text of articles includes author biographies,links to glossary terms and
`
`Fo.
`atest articles publishedonline aheadofprint.
`links to websites and databases with relevantinformation.
`rthcomingarticles:
`
`he
`ces
`;
`*rmal ablari
`
`ity Visraee lation of tumours: biologicalmechanisms and advan
`E-alert table of contents
`he content of thisjournal — sent FREE to your
`Get monthly e-mail alertsto t
`“tring FChyChu and Damian E
`
`p to receive the latest content as an
`inbox — by registering online. Or signu
`Lymph
`>. Dupuy
`iogenesis
`<
`¥
`ae
`:
`RSS newsfeedbyvisiting www.nature.com/reviews/cancer
`t
`Hogenesis and lymphatic vessel remodelling in cancer
`even A. Stacker
`S
`:
`andMare Cee P Williams, Tara Karnezis, Ramin Shayan, Stephen B, Fox
`Twitter Follow Nature Reviews Cancer on Twitterfor the latest news from the
`
`
`journal and beyond: @NatureRevCancer
`
`nicer(ISSN 1474-1753) is
`aabliching Group
`divisionof Macmillan
`aT »)ispublished montht by Nature Publishing Group.ann
`
`Online correspondence xX
`
`
`4.
`The MacmillanBuilding, 4 Crinan §
`retgndon NL 9XW, UK.Nopart of thispublication Mey
`
`Hretrieval system, or transmitted in any form jelectnonic or otherwise)without prior
`
`Commentsfrom readers aboutourarticles. All correspondence will be
`nature.com.US
`NY. andadditional mailing
`
`sdicals postagepaidat Jamaica,
`;
`‘
`.
`5
`highlighted on the table ofcontents and one-mail alerts.See page 146.
`MonMASTER; Sendaddress
`vostoNaturePublishingGroup.AirBusinessLtd,e/oWorldnet
`
`o. PING Ine,
`4 Maemil
`ine ae 16thAvenue, 2ndFloor,
`Jamaica, NY 11434, USA.
`Ublishers Limited, All rights reserved, Printed in United Kingdom.
`
`NATupr
`TURE REVIEWS |CANCER
`i
`
`This material was copied
`atthe NLM and may be
`Subject US
`i
`
`VOLUME14 | FEBRUARY 2014
`
`

`

`nature
`REVIEWS
`
`CANCER
`
`Sane
`a” vey
`Yee,
`
`
`» COVER: ‘Not a BB)’ by Lara Crow, inspired by the
`Review on p77 and the Timeline on p135.
`
`CONTENTS
`
`COMMENT
`Spontaneous regressionof metastatic cancer: learning from neuroblastoma
`© Scott J. Diede
`RESEARCH HIGHLIGHTS
`Selections from the recent scientific literature
`
`PERSPECTIVES
`OPINION
`Beyond E-cadherin: roles of other cadherin superfamily
`members in cancer
`
`Frans van Roy
`The cadherin superfamily includes many proteins other than E-cadherin.
`These cadherinsare very diverse in both structure andfunction, and
`their mutualinteractions seemto influence cancer development and
`progressionin complex andversatile ways.
`
` o
`
`2%
`ipin tO e
`a
`I LFTI
`ILLES
`LESTI SVID SIS
`
`TIMELINE
`Tumour antigens recognized by T lymphocytes:
`pedJooceosecencsocencecnenaneccekl|Josonosaseneacoesesoa%co00000
`at the core of cancer immunotherapy
`Pierre G. Coulie, Benoit J, Van den Eynde, Pierre van der Bruggen and
`Thierry Boon
`In this Timeline, the authors discussthe identification of tumourantigens
`that are recognizedby T lymphocytes andhowthesefindings can be
`effectively andsafely transferredto the clinic.
`
`
`
`Finding tumourantigens, p135
`
`Links between oestrogen receptoractivation and proteolysis:
`relevance to hormone-regulated cancer therapy (corrigendum)
`WenZhou andJoyce M. Slingerland
`
`EDITORS
`
` NICOLA MCCARTHY
`&GEMMA ALDERTON
`nDARREN BURGESS ISABEL LOKODY
`
`
`aa
`
`|
`SARAH SETON-ROGERS
`
`EDITORIAL & PRODUCTION OFFICE
`The Maconillarn Building,
`1 Crinan Street, London NL OXW, UK
`Tek 444 (0)/0 $844 3620
`Lnwil NatureReviews@inatune.com
`wiuwaluire.con/reviews/© anc er
`
`CHIEF EDITOR Nicola McCarthy
`SENIOR EDITORS Gemma Alderton, Sarah Seton Rogers
`ASSISTANT EDITORS Darren Furqess, Isabel Lokody
`COPY EDITOR Simon Neil
`ART EDITOR Lara Crow
`COPY EDITING MANAGER © atriona Rouwell
`SENIOR COPY EDITOR Mariam farurgi
`EDITORIAL ASSISTANTS | mily Finn, Rebecca Cromie
`MANAGING PRODUCTION EDITOR Judith Shadwell
`SENIOR PRODUCTION EDITOR Sinn Fenwick
`ART CONTROLLER Susanne [artis
`SENIOR ART EDITORS Patrick Morgan,
`Vieky Sumunershy
`PRODUCTION CONTROLLER Natalie South
`PUBLISHER Alison Mitchell
`
`CUSTOMER SERVICES www.nature.com/help
`PENSONAL SUBSCRIPTIONS subscriptionsainature.com
`REPRINTS weew.nature.com/reprints
`ADVERTISING & SPONSORSHIP
`www nature com/advertising
`SITE LICENCES
`weewnature.com/libraries/site_licenses/indes. html
`PRESS OFFICE press@onature.com
`MARKETING marketingiinature.cam
`
`MANAGEMENTOFFICES
`NPG LONDON [he Macmillan Building, 4 Crinan Street
`London N1 OXW, UK, Tel: 644 (0)20 7833 4000
`NPG NEW YORK 75 Varick Street, 9th floor NewYork,
`NY 10019-1917, USA. Tel: «1 212 726 9200
`NPG ASIA-PACIFIC Chiyoda Building 2-37 lehigayatamachi,
`Shinjuku Ku, Tokyo 16200544, Japan, Tel +1 212 726 4200
`Alist of other offices can be found at wwwnature.com
`npg _/contact/oltices.html
`Copyright @ 7014 Nature Publishing Group
`Printed in Wales by Cambrian Printers on acid-free paper
`
`NATURE REVIEWS | CANCER
`
`This material was copied
`at the NLM and may be
`Subject US Copyright Laws
`
`VOLUME 14] FEBRUARY 2014
`
`

`

`
` This material may beprotected by Copyrightlaw (Title 17 U.S. Code)
`
`
`TIMELINE
`
`Tumour antigens recognized by
`T lymphocytes: at the core of
`cancer immunotherapy
`
`pierre G. Coulie, Benoit J. Van den Eynde, Pierre van der Bruggen
`and Thierry Boon
`
`Abstract | In this Timeline, we describe the characteristics of tumour antigensthat
`are recognized by spontaneousT cell responsesin cancerpatients andthe paths
`thatledto their identification. We explain on what genetic basis most, butnotall,
`of these antigens are tumourspecific: thatis, present on tumourcells but not on
`normal cells. We also discuss how strategies that target these tumour-specific
`antigens can lead either to tumour-specific or to crossreactive T cell responses,
`whichis an issue that has important safety implications in immunotherapy.
`Thesesafety issues are even more of a concernforstrategies targeting antigens
`that are not knownto induce spontaneousT cell responsesin patients.
`
`Cancer immunotherapy that involves the
`deliberate use ofthe adaptive immunesystem
`to reject tumours or to prevent their recur-
`renceis gaining momentum. Interesting
`clinical results have been obtained using
`cancer vaccines, adoptive I’ cell therapies and
`antibodies that stimulate the activity of
`‘T lymphocytes. Moreover, increasing evi-
`dence suggests that adaptive immunity con-
`tributes to the long-termclinical benefits of
`anticancertreatments such as chemotherapy
`andradiotherapy. At the core oftheseclini-
`cal developments lies thefact that cancer
`patients can produce T lymphocytesthat
`recognize Lumour-specific antigens. Thefirst
`human tumour-specific antigens that were
`recognizedby T’ cells were discovered about
`20 years ago(FIG.
`| (TIMELINE). Considering
`the increasing number of clinical studies
`that rely on the presence of tumour-specific
`antigens that are recognized by Tcells, it is
`worth summarizing the key steps that led to
`their identification, andit is worth deserib-
`ing the genetic processes that result in their
`presence on tumourcells. A proper under-
`standing ofthefactors that affect the degree
`ofspecificity ofthe T lymphocyte response
`against tumour antigens is essential to aid
`the design of immunotherapystrategies that
`are not only efficient but alsofree of adverse
`side effects.
`
`Identification of mouse antigens
`Initial controversy about the existence of
`tumourrejection antigens. From 1940 to
`1960, the study of mouse tumours that were
`
`NATURE REVIEWS | CANCER
`
`induced with oncogenic viruses showed
`that the immune system could reject these
`tumours following the recognitionofviral
`antigens’. The first evidence that mouse
`tumours that were not induced byviruses
`could also be recognized by the immune
`system was obtained by Gross andcolleagues
`in 1943 (FIG.
`| (TIMELINE). They induced
`tumours in mice through the use of chemi-
`cal carcinogensandthen resected these
`tumours. These mice wereable to reject the
`sametumourcells on subsequent exposure’.
`Mice that were immunizedwith lethally
`irradiated tumourcells were similarly pro-
`tected. Theseresults were confirmedby
`other groups’, andin the 1960sit became
`widely accepted that mouse tumourcells and
`therefore possibly humancancercells could
`be recognizedbythe immunesystem.
`In sharp contrast, in 1976, Hewitt!
`reportedthat a similar analysis carried out
`with spontaneous tumours that developed
`in micefailed to produce anyevidence of
`immunecontrol. He concludedthat mouse
`tumourantigens were artefacts that were
`inducedbythe chemical treatment used
`to induce experimental tumours and were
`therefore unlikelyto be present on human
`tumours.
`
`In the 1970s, we treated a mouse terato-
`carcinomacell line in vitro with a strong
`mutagen, and we showedthat manycell
`clones that were derived from the mutated
`population wereincapableofforming
`progressive tumours wheninjectedinto
`syngeneic mice”. These ‘tum’ variants were
`
`This material was copied
`atthe NLM and may be
`Subject US Copyright Laws
`
`PERSPECTIVES
`
`rejected by an immune response directed
`against new antigensthat weredifferent for
`everyVariant (tum antigens). Remarkably,
`mice that had rejected tum variants were
`also protectedagainst a subsequentinjection
`of the parental tumourcells’, even though
`this teratocarcinoma was non-immunogenic,
`similarto the tumours that were described
`by Hewitt. We concludedthat anefficacious
`response against the tum antigens had an
`additional effect: it triggered a response
`against antigens that were present on the
`original tumour but that were apparently
`non-immunogenicon their own. In col-
`laboration with Hewitt, we treatedcells from
`spontaneous tumours with multagensto
`obtain tum) variants, and we observedthat
`these variants were also capable of induc-
`ing immuneprotection against the parental
`tumours®. This showedthat spontaneous
`mouse tumours do express tumouranti-
`gens,albeit poorly immunogenic ones. We
`became convinced that human tumours
`might also be susceptible to immunological
`treatment andthat we shouldfirst identity
`the nature ofthe rejection antigens that were
`observed on the mouse tum variants and
`their parental tumour.
`
`Molecular identification ofantigens recog-
`nized by T lymphocytes on mousetumours.
`Afterthe discovery of T lymphocytesin the
`1960s", their essential role in graft rejection
`and tumourrejection was soonrealized!”
`Inthe tum: system, we observed that adop
`tive transfer of T cells, which werecollected
`frommice following rejection of a tum
`variant, protected irradiated mice against
`the growthofthe samevariant. This clearly
`indicated the involvement of 'T lymphocytes
`in the tum) phenomenon. Accordingly,for
`several years, we attempted to obtain specific
`cytolytic T cells that weredirected against
`tum. variants; this was unsuccessful. We
`eventually turned to the P815 mastocytoma
`cell line, which proved to be remarkably easy
`to cultivate andto clone becauseit prolif-
`erated in suspension. ‘lum variants were
`readily obtainedforthis cell line''. Moreover,
`excellent cytotoxic 'T lymphocyte (CTL)
`responses were obtained that showedclear
`specificity for each tum: variant".
`Wethenbenefited from a major advance
`inthe CTL field: microcultures could be
`derived from a single CTL by repeated
`stimulations with irradiated target cells
`in the presence of a Tcell growthfactor
`that was later identifiedas interleukin-2
`(IL-2)'. These clonal CTL cultures could
`be expandedtolarge numbers and could be
`maintained indefinitely. These stable CTL
`
`VOLUME 14] FEBRUARY 2014 | 135
`
`

`

`PERSPECTIVES
`
`clones that weredirected against a single
`antigen provedto be crucial for a rigorous
`analysis and dissection ofthe antigens
`recognized by I cells on several target cells.
`With great help from Cerrotini andhis
`group, who hadhad a prominent role in
`these developments, we obtainedstable
`CTL clonesthat killed the stimulatory
`tum variant but not the other tum variants
`nor the parental tumour cells'®. These CTL
`clones clearly recognized a tum antigen
`that was induced by the mutagentreat-
`ment. Other CTL clones killed both the
`tum andparentalcells, evidently recog-
`nizing an antigen that was present on the
`original P815 tumourcells, That these
`antigens were genuine rejection antigens
`was shown by the i vivo observation that
`some tumours progressed, then nearly com-
`pletely regressed, then progressedagain,
`These ‘escaping’ tumours hadinvariably
`lost the antigen that was recognized by one
`of the CTL clones". This wastrue not only
`for tum: antigens but also for antigens that
`were present onthe parental tumour'’. In
`fact, these andotherstudies that were car-
`ried out in the early 1980s formally showed
`the reality of tumour immunesurveillance
`andthe occurrence of tumour escape after
`immuneselection", whichis a process that
`wasrecently renamed‘immunoediting’
`(REF 20). Although immunoselected tumour
`variants wereresistant to some CTL clones,
`they werestill sensitive to others. A detailed
`analysis of a panel ofsuch variants led to the
`conclusion that CTLs recognized several
`(typicallyless than ten) distinct antigens on
`a given tumour!"
`
`Thenext step was to define the molecular
`nature oftheseantigens. The onlyavailable
`tools werethe stable CTL clones. The exact
`molecular nature of the antigens that were
`recognized by CTLs was unknownatthat
`time. However, the notion that antigensare
`recognized by T lymphocytes inassocia-
`tion with major histocompatibility complex
`(MHC; humanleukocyte antigen (HLA) in
`humans) molecules had been known fora
`decade*'. In 1986, ‘Townsend showedthat
`antiviral CTLs recognized small peptides of
`eight to ten aminoacids, which were derived
`fromaviral protein andpresentedat the
`surface ofinfectedcells in association with
`MHC class | molecules”. Soon thereafter, an
`excellent crystallographic study showedthat
`MHC class I molecules present small pep-
`tides in a groovethatis locatedat the surface
`of the molecule*'. We nowknowthat these
`peptides are producedbypartial digestion
`ofthe parental protein, mainly through the
`proteasome machinery (BOX |). These pep-
`tides then becomeassociated with the MHC
`class | molecule andaredisplayedat thecell
`surface following a process knownas the
`‘antigen processing pathway’(FG. 2).
`‘Toidentify our antigens, we useda genetic
`approach that aimed to clone the gene encod-
`ing the antigen, Once again, the P815 cells
`were invaluable, as we wereable to select a
`highly transfectable variant named P1.-HTR".
`We transfected P1.HTRwith a gene library
`derivedfromcells that expressed a given tum
`antigen and, using the relevant CTL clone,
`weidentifieda transfectantthat expressed
`the antigen. The encoding gene wasretrieved
`fromthe transfectant and sequenced.
`
`The first gene that encoded a tum: anti-
`gen was cloned in 1988 (REF 25). Ibencoded
`a ubiquitous protein of unknownfunction.
`Crucially, the coding region containeda
`mutation that changed one aminoacid
`in the protein. Small peptides that contained
`the mutated residue were shown to sensitize
`parental P815 cells to CTL-inducedcell
`death, whereas corresponding wild-type
`peptides did not’. We concluded that the
`antigen was a complex between the mutated
`peptide and the presenting MHC class |
`molecule.
`The identification of two other tum
`antigens’indicatedthat each of themalso
`resulted from a point mutation in a ubiqui-
`tously expressed gene. Each mutation cre-
`ated a newantigenic peptide. In some cases,
`the mutation enabled the peptide to bind
`to the groove ofthe presenting MHC mol-
`ecule. In other cases, the mutation created
`a new epilope ina peptidethat wasalready
`bound to MHC,but thewild-type peptide
`was not recognized by IT cells because of
`central tolerance (FIG, 4a). Even though
`tum antigens wereartificially induced by
`mutagen treatment, their identification
`establishedthe principle that rejection anti-
`gens can result from mutations in ubiqui-
`tously expressedgenes.‘Theseresults showed
`for thefirst time the occurrenceofa process
`of immunesurveillanceof genomeintegrity.
`Wethenset out to identify the tumour
`rejection antigen that was present on the
`parental mouse tumour P815. This time,
`the identified antigen, which was named
`PLA, did not result from a mutation. The
`antigenic peptide correspondedto the normal
`
`Timeline | Milestones in the discovery of tumour rejection antigens
`
`|
`|
`
`Mouse tumoursinduced
`Mutagenic treatment of mouse
`| CTL clones raised
`with methylcholanthrene
`| Discovery of
`| tumour cells produces a high
`
`Stable CTL clones
`induce protective
`| frequency of tumour variants
`| cytolytic
`| against mouse tum |
`permit dissection of
`I lymphocytes and
`immunity against the same |
`antigens are used to
`| that are rejectedfollowing a
`
`
`first use of
`select loss variants
`Generation
`tumour, indicating that
`|
`tumour antigens
`[cell immuneresponse. These
`chromium release
`
`
`anddissect tumour
`present on human
`tumours have specific
`of long-term
`‘tum“variants express new
`
`melanomalines"
`CTLelones"
`assay!"
`rejection antigens’
`antigens”
`antigens" "
`
`
`
`
`
`|
`
`Discovery of
`| Tlymphocytes
`
`Identification of MHC
`| restriction of cytolytic
`| T lymphocytes’!
`
`| CTLs raised against
`| viral antigens on
`) tumours!”
`
`| Irradiated tumour cells from spontaneous mouse |
`tumours do not induce any protective immunity,
`suggesting that tumour antigens are an artefact’
`
`Micethat have rejected tum
`variants are protected against the
`original tumour, even when this
`tumour is spontaneous and
`non-immunogenic, indicating that
`tumour antigens exist and might be
`| present on human tumours’”
`
`|
`|
`
`The antigens
`recognized by T cells
`are small peptides
`bound to MHC
`molecules’!
`
`|
`
`Black boxesrefer to discoveriesthat are related to mouse tumours; red boxesrefer to discoveries that are related to human tumours, Observations and discoveries that
`are relatedtoviral antigens are not included. BAGE, B melanoma antigen; CTAG, cancer/testis antigen; CTL, cytotoxic T lymphocyte: GAGE, G antigen; MAGEA1,
`melanoma antigen family A, 1; MHC, major histocompatibility complex; WT1, Wilms’ tumour protein,
`
`This material was copied
`www.nature.com/reviews/cancer
`136 | FEBRUARY 2014|VOLUME I4
`at the NLM and may be
`Subject US Copyright Laws
`
`

`

`PERSPECTIVES
`
`sequence ofa gene of unknownfunction
`that was named Trapla*’. The antigen
`was recognizedby I’ cells on the tumour
`because ofthe complete lack of expression
`ofthe gene in normaladulttissues, which
`preventedthe establishment of immune
`tolerance. The onlycell types in which the
`gene was expressed were spermatogonia
`andplacental trophoblasts, which are two
`cell types that do not express MHC class |
`molecules on their surface andthere-
`fore cannot presentthe antigento 'Tcells
`(FG. 4a). ‘Tumourantigen PIA is therefore
`clearly tumour-specific, even thoughit is
`not mutated,
`Trap la was expressed in several mouse
`tumours ofdifferent histological types".
`P1A was theretore the first example ofa
`tumour-specific antigen that was shared
`amongdistinct tumours. Vaccination of mice
`against PLA inducedprotective responses
`that led to the rejection of P815 tumours,
`which further validated this type ofantigen
`as a genuine tumourrejection antigen”.
`As expected,vaccination did not induce any
`deleterious immune response against normal
`organs, including testes. Trapla sharesits
`characteristic expression protile andloca-
`tion on the X chromosome with human
`‘cancer-germline’ genes (discussed below),
`‘Thus, the work that was carried out with
`mouse tumours provided a method ofiden-
`tification of tumourantigens, and it identi-
`fied the two main genetic mechanisms that
`produce tumour-specific antigens recog-
`nized by Tcells: namely, gene mutation and
`activation of genes that are silent in normal
`tissues.
`
`Identification of human antigens
`In the early 1980s, several groups began
`to stimulate in vitro blood’ or tumour-
`infiltrating lymphocytes” that were iso-
`lated fromcancer patients with autologous
`tumourcells killed by irradiation. This
`produced CTLs withhigherlytic activity
`towards autologous tumourcells than con-
`trol cells. However, such Tcell populations
`always had somelytic activity against normal
`cells, which made their degree of tumour
`specificity difficult to establish,
`Once again, the productionofstable CTL
`clones was crucial. In the late 1980s, using
`Tcells froma patient with melanoma, we
`obtained stable anti-melanoma CTLclones
`that were completely inactive against a wide
`range of normal cells**. Such CTLclones
`were used in immunoselection experiments
`to dissect the various antigens that were
`present on the autologous tumour, and the
`results indicatedthe presenceofat least six
`distinct antigens”.
`
`Genetic approach. ‘lo identify these anti-
`gens, we used a CTL clone anda DNA
`library that was derived fromthe autologous
`melanoma; this was the same strategy that
`we had used with the mouse tumour cells.
`In 1991, this led to the identification ofthe
`first human genethat codedfor a tumour-
`specific antigen recognized byTcells".
`This new gene, which was named mela-
`nomaantigen family A,
`1 (MAGEA1), was
`expressed in many human tumoursofdif-
`ferent histological types. No expression was
`found in normaltissues, with the exception
`ofmale germline cells andtrophoblasticcells.
`
`In humans,as in mice,these twocell types
`donot produce MHCmolecules and
`therefore cannot present antigens to Tcells
`(FIG, 4a)". Thus, the expressionprofile of
`MAGEA| was similar to that of mouse
`Trapla. The antigenic MAGEA1 peptide,
`whichis presented to CTLs by HLA-Al
`molecules, wasidentified by transfecting
`short DNA fragments, thereby narrowing
`downthe peptide-encoding region until
`candidate peptides could be synthesized
`andtested for CTL recognition”.
`MAGEAI provedto be a member of
`a large newgene family!’ ". The MAGE
`family comprises 25 cancer-germline
`genes with a similar pattern of expres-
`sion. Several other cancer-germline gene
`families were identified in the following
`years” — our procedure having been
`updated bythe use of CDNA instead of
`genomiclibraries.
`Other antitumour CTLs were shownto
`recognize peptides that were encoded by
`mutated genes”. Thefirst was caused by
`a mutation in a ubiquitously expressed gene
`that encodeda protein of unknownfunc-
`tion”; the second was caused by a mutation
`in cyclin-dependent kinase 4 (CDK4)".
`In addition, to our surprise, CTLs
`frompatients with melanomawere found
`to recognize peptides derived from
`melanocyte-specific proteins. Thefirst iden-
`tified peptide was derivedfromtyrosinase,
`whichis present in normal melanocytes and
`in most melanomas”, Interestingly, a pep-
`tide of tyrosinase was also the first tumour
`antigen that was found to be recognized by
`CD4" Tcells”.
`
`
`
`lumour-infiltratinglymphocytes are
`present in human tumours and can be
`reactivatedin vitro tokill autologous
`
`cultured tumour cells"!
`|
`
`Peptides presented by MHC
`molecules at the cell surface
`can be eluted by acidic
`treatment and characterized"
`
`
`* Identification of several additional human
`families of cancer-germline genes, suchas
`BAGE, GAGE and CTAGM8"
`
`* Identification of a peptide encodedby the
`Identification
` gene ERBB? asa valid antigen
`
`ationts will)
`The generation of
`melanoma makeT cell
`some tumour-specitic
`of a peptide
`
`
`of WT1 asan
` antigens depends on
`responses against
`
`
`differentiation
`antigenic
`Identification of a large number of human
`thetype of
`antigens, such as
`tumour-specific antigens resulting from point
`peptide in
`proteasome present
`
`
`
`tyrosinaseEnna!
`mutations in ubiquitously expressed genes” "'"
`leukaemias!”
`in the tumourcell!"
`
`
`
`
`|
`
`
`|
`
`Identification of the
`Identification of several
`\dentification of a gene encoding a
`mouse tumour-specific antigen.
`cancer germline genes
`point mutations that
`result in the new
`homologous to MAGEA1.
`| This geneis silent in normaltissues
`ours”
`The MAGE family
`Hi jat
`antigenic peptides
`| andactivatedin sometun
`present on tum variants,
`comprises #5 genes
`This indicates that
`immunosurveillance of
`the integrity of the
`mammalian genomeisa
`reality!"
`
`
`
`CTL cells on human tumours
`
`Identification of thefirst human
`lumour-specific antigen, encoded
`Ay
`by MAGEA1, that is recognized by
`
`
`
`
`
`
`
`
`Some tumour-specitic
`nt fraction of
`Animporta
`antigenic peptides
`tumour-specific antigenic
`peptidesar
`containjuxtaposed
`© produced by
`anomalous
`distant protein
`genetic processes
`such as intr
`sequences, This results
`‘onic transcription,
`from thesplicing of
`antisense transcription and
`bei
`protein fragments inside
`post-translational
`modification
`the proteasome!!!"
`
`
`NATUREREVIEWS | CANCER
`
`This material was copied
`at the NLM and may be
`Subject US Copyright Laws
`
`VOLUME 14 [FEBRUARY 2014 | 137
`
`

`

`PERSPECTIVES
`
`Box 1 | Proteasome-generated antigenic peptides —
`In the steady state, most cells contain the standard proteasome, which has three catalytic subunits
`called $1, 62 and 85. Under inflammatory conditions, these standard catalytic subunits are replaced
`by their interferon-inducible counterparts, [}1i, 52i and (55i. The resulting ‘immunoproteasome’ has
`slightly different catalytic activities. In addition, some normaltissues and tumoural tissues contain
`intermediate proteasomes that comprise a mixed assortmentofcatalytic subunits (81-$2-[5i or
`61i-B2-$5i)'". Several tumour antigens are produced only by some types of proteasome. For
`example,the peptide Melan-A.,,,, is produced only by the standard proteasome!", the peptide
`MAGEA3., ,, ,,, is produced by the immunoproteasomeand the intermediate proteasomes'"*'°, and
`aT
`the peptides MAGEA3
`and MAGEC2_
`vw, are exclusively produced by one and both of the
`intermediate proteasomes,respectively'**, Therefore, the expression of a given tumour antigen
`doesnotalwaysparallel that of the parental protein and major histocompatibility complex (MHC)
`class|, but it also depends on the proteasome contentofthe cell.
`Proteasomescan also splice peptide fragments that are located at distance from each otherin
`the parental protein''”'"*, Peptide fragments canbespliced either in the same order or in the
`reverse orderto that in which they occur in the parental protein’’, Therefore, the sequence of an
`antigenic peptide cannot alwaysbedirectly deducedfrom thatof the encoding gene.
`
`In 1995, another approachtoidentify
`genes that were preferentially expressed
`in tumours madeuse not of CTLs but of
`antibodies from cancerpatients. Using a
`new methodology, which was developed by
`Pfreundschuhandcolleagues” and named
`serological analysis of recornbinant CDNAexpres
`sion libraries (SEREX), libraries of tumour
`cDNAthat was expressedin bacteria were
`screened using serum samples from cancer
`patients, with the expectation that the serum
`containedantibodies that would bindto sur-
`face or intracellular proteins that were specif-
`ically or preferentially present in tumours”.
`Thisledtothe identificationofseveral genes,
`a fewof which turned out to be newcancer-
`germline genes, such as synovial sarcoma
`X breakpoint 1 (SSX 1), SSX2 (REF 5/) and
`cancer/testis antigen 1A (CTAGIA; also
`known as LAGE2 and NYESO1)™. A reper-
`toire of genes encoding proteins that elicit an
`antibodyresponsein cancer patients can be
`foundin the SEREX database”,
`
`Biochemical approach. Following pioneering
`work bythe group of Rammenseeonthe
`acid elution ofantigenic peptides boundto
`MHC molecules, Hunt andcolleagues"! used
`this approachtoidentify a new melanoma
`antigen. Theidentified antigenic peptide was
`derived fromthe melanocytic differentiation
`protein GP100 (also known as PMEL and
`PMEL17)". Relatively few tumour-specific
`antigens have beenidentified by this approach,
`whichis technically very demanding,
`However, biochemicalanalysis of eluted pep-
`tides wasessential to show that some peptides
`have undergonepost-translational modifica-
`tions”**. For example, the deamidation ofan
`asparagine into anaspartic acid residue was
`observedfora tyrosinase peptide, andthis
`change wasessential for the efficient T cell
`recognitionofthis particular peptide”.
`
`138 | FEBRUARY 2014 | VOLUME 14
`
`Reverse immunology. ‘The numerous genes
`ofthe cancer-germlinefamilies are expected
`to be sources ofa vast number of antigenic
`peptidesthat bind toa wide range of HLA
`molecules. Another possible sourceof
`tumour-specific antigens is genes that are
`mutatedin many tumours, such as KRAS,
`TPS3 or the BCR-ABLI fusion gene. Thus,
`thereis a needtoidentify new antigenic pep-
`tides onthe basis of gene sequences andin
`the absence ofaI’ cell that is directed against
`these antigens.
`‘To achieve this,the first step is to iden-
`tify candidate peptides that bind to a given
`HLA molecule, This is carried out by using
`computer-generated algorithms"thatselect
`peptides within a protein thatarelikely to
`bind to an HLA molecule. Theresulting
`
`Glossary
`Adoptive transfer
`the Intision inbe patient
`In Cancer iminunotherapy,
`of aun
`Lis
`anblumnoau | cells that have boon
`
`iffO. The: Winphocytes can also be
`
`Wh
`retroviral expression vectors in onder
`
`
`to
`fiven T cell receptor or other pene
`produets.
`
`|
`
`|
`
`Anerey
`Ol
`Hy pOTeSPONSIVETIOSS OF LESPOrSsivETioss
`TWeiphocytes
`alter recognition of their antigen
`
`Central tolerance
`autoreactive
`| The deletion or inactivation of immature
`eels and Peels ofthe prunary Winphoid orpans
`The bone marrow (cells) and the thyrius (T eells)
`The
`renaining mature autoreactive B cells and T
`
`aire che
`wilh by The mechanisms
`
`
`cell
`
`}
`
`
`
`| Deamidation
`InN elycosyviated
`mile group.
`) The sermval of
`ar
`In Wapatagine by the
` elyCanase @CNerailNSs
`an
`aspartate
`by dearnidation This can result
`in
`an antigen
`
`This material was copied
`at the NLM and maybe
`Subject US Copyright Laws
`
`candidate peptides of about nine amino
`acids are synthesized, andtheir bind-
`ing to HLAis tested in vitro, Cells pulsed
`with peptides that most efficiently bind to
`HLA are usedto stimulate T lymphocytes
`in orderto derive populations or, prefer-
`ably, clones of Icells that recognize cells
`expressing the appropriate HLA pulsed
`with the peptide. It is essential to verify
`that these CTLs also recognize unpulsed
`tumourcells that express the protein from
`whichthe peptide is derived, because many
`peptides against which Tcells can beraised
`are not producedbythe antigen-processing
`machinery(FIG. 2). Ina related approach,
`Tlymphocytes are stimulated with dendritic
`cells that are loaded with a recombinant
`proteinor that are infected with a recom-
`binant virus containing a tumour-specific
`DNA sequence”.
`Several antigenic peptides that are recog-
`nized by CD4’ or CD8" Tl lymphocytes have
`been identified using these approaches” ©.
`‘The peptides that are expressed by tumour
`cells are listed in a database, whichis
`regularly updated”.
`
`Humanantigens:classes
`Antigensof high tumoural specificity, Three
`types of tumourantigens have the potential
`to elicit immune responses t