Ae EAEIEE
`
`SS +"
`\
`\
`Vs
`\
`s 8
`NY
`8
`2
`2 8
`Ss
`NodNowe SS Sow SOA
`ob 10.1048/naturez3003
`
`
`Personalized RNA mutanome vaccines mobilize
`poly-specilic therapeutic immunity against cancer
`Evelyria Derhovanessian!, Matthias Miler! , BySrri- t
`Philip Kioke!, Fetra Simon’, Martin Lower’, Valesca Bukurt
`Ugur Sahin!
`Arbel D. Tadmor?,
`Ulrich Luxerntnirger!, Barbara Schrérs*, Tana Chr
`nokoko!, Mathias Vormehr'“, Christian Albrecht®,
`tag Miller’, Inga €“44
`janina Byick!
`Anna Parazynaki!, Andreas N. Kabnl,
`Sandra Heesch!, Katharina H. Schreeh, elicit
`
`Richard Rae“, Andrea Breithreuz’, Chaudia Tolliver
`isabel Vogler’, Eva Godehard?, Sebastian Artig??
`net,
`tict, Alexander Hohberger’, PatrickSerr?flan ivekmann',
`
`Goran|
`i Janko Ciesia?, Olga Waksrna
`
`Oebermek, Barbara K
`
`
`Aridree |
`i
`
`: aOger, 212d BricKENMeike Wie
`
`
`
`eraano-, David Langer,
`
`+? Stephan Grabbe?,
`, Christoffer
`-, Roroina Nemec
`
`# & ‘Sol
`, Garraen Loqu
`
`setier?,
`
`' Martin Suchar’,
`
`Tcells directed against mutant neo-cpitapes drive cancer immmunity.
`However, spontaneous immune recognition of mutations is
`inefficient. We recently introduced the concept of individualized
`mutanomevaccines and implemented an RIWA-based poly-neo-
`epitope approsch io mebilize immunity against a spectrum ofcancer
`muntations’*, Here we repart thefirst-in-heman application ofthis
`concept in melunenia. We set up a process comprising coraprehensive
`identification of individual mutations, computational prediction
`of neo-epitopes, and design and manufacturing of a vaccine unique
`for each patient. All patients developed T cell responses against
`multiple vaccine neo-epitepes at ap to high single-digit percentages.
`Jaccine-induced T cell infiltration and neo-epitope-specific killing
`of gutelogous tumour cells were shown ba past-vaccination resected
`metastases from twe patients.Fhe cumulative rate of metastatic
`events was highly significantly reduced after the start of vaccination,
`resulting in a sustained progression-free survival. Twa of the five
`patients with metastatic disease experienced vaccine-related
`objective responses. One of these patients had a late relapse owing
`to outgrowth of 42-microglobulin-deficient melanoma cells as an
`acgyied resistance mechaniom.A third patient developed a complete
`response to vaccination in combination with PD-i blockade therapy.
`Our study demonstrates that individual nvutations can be exploited,
`thereby opening a path to personalized inmuunctherapy for patients
`with cancer.
`Cancer nustations can form neo-epitopes recognized by T cells
`on HLA molecules, which contributes to the clinical success of
`
`
`iyommanotherapy?”. Cmby a srnail fraction of mutations
`induce
`span-
`taneous imonune responses in the tumour-bearinghost, which lirsits
`irummunotherapyefficacy to tumours with a high mutational load? 19.
`
`In mouse vacciaation models, a substantia] fraction af cancer
`
`nintations is immunogenic and preferably recognized by CD4>
`T cells” Vaccines composed of predicted HLAclass H? and class f
`
`neo-epitopes!!* have previously been shown to induce rejection
`of mouse tamonrs. Altogetber, these findings created enthusiaam
`
`neo-epitope vaccines'>, As the vast majority of cancer muta-
`ons are uniqueto the individual patient, personalized approaches
`are needed. For the first-in-humantesting of such an approach, we
`developed a process cormpliant with regulatory requirements(Fig. la}.
`Now-symopnynious mutations expressed by thirteen patients with
`stage Hi and TV melanoma were identified by comparative exome
`
`and RNA sequencing of routine turmenr biopsies and healthy blood
`cells. Mutations were ranked according te: (1) predicted high-affinity
`binding to autelogous HLAclass iT and high expression of the
`miutation-encoding RNA®, and (2) predicted HLA class I binding.
`Ten selected mutations per patient (five for patient POS) were engi-
`neeéred into two synthetic RNAs, each encodingfive linker-connected
`27merpeptides with the nautation in position 14 (pentatope RNAs}
`
`
`(Fig. 1b, Supplementary Table 1). RNA was produced within 68 days
`
`(tange 49 to 102 days} according to good manufacturing Practice
`
`with a success rate of 100% (Fig.
`ic}. Analytical testing extended the
`
`median time from selection of mutationsto vaccinerelease to 103 days
`
`(range 89 to 160 davs). Patients with NY-ESO-1- and/or tyrasinase-
`positive melanoma received am RNA vaccine encodingthese
`shared
`
`
`turmonr-associated self-antigens until release of their nec-epitope
`vaccine. At least cight doses of the nea-epitope vaccine were injected
`percutaneouslyinto inguinal lyreph nodes under ultrasound control
`(Fig. id). Previously, in mouse models, we showedefficient uptake,
`
`transfation of RNA-encoded antigens by Fymaph-node-resident
`dendritic cells (DCs), and intrinsic adjuvantactivity’®
`AH patients completed treatment with a maximam of20 nea-epitape
`vaccine doses (Extended Data Table 1}, which they tolerated well
`
`without related serious adverse events. The immunogenicity ofeach
`ofthe 125 mutations administered inthis study was ari:alysedby TEN-y
`
`ELISpot inb4? and CD8° Fcells in pre-aadpast-vaccination blood
`samples (SaypplementayTable2).Responsesweredetectedagainst60%
`
`
`ofthe px
`itopes (Fig, le, Bxtended Data Fable 2), and each
`patient developed cells againstat least three rautations. Pre-existing
`
`weakresponses agaiust one-third of the immunogenic BeG-epitopes
`were augriented upon vaccination (Fig. le, f, Extended Data Fig. 2a).
`The other two-thirds were de neve responses. immunogenic matations
`
`were overlydistributed across the five positions of the pentatope RNA
`(Fig. 1g}, indicating Lack of positional bias. As previously observed
`in mice+, the majority of neo-epitopes mounted exclusively ¢cpa?
`responses (Fig. le, Extended Data Fig. 2b). A smaller fraction was
`
`recognized by CIO87 cytotoxic lymphocytes (CTLs) only. One-quarter
`showed concomitant CD47 and CD8t responses, recognizing different
`regions of the mutated 27mer sequence (Fig. le, Extended Data
`
`
`Tmmunetoferancei
`xpected to9Serpress.
`Coell| responses against
`
`y of neo-epitope-
`broadly expre
`d proteins. In fact, the miajori
`
`: Spitatge
`velberg Univer
`Im Neuen
`
`
`
`far Ind
`
`OO MONTH 2017 | VOL Goo | NATURE |
`
`
`1
`
`

`

`AySeyRERSERRENBS
`
`
`
`PEERS
`SRR«|
`
`Saraple
`aguiskion
`
`Mutation
`discovery
`
`Target
`selection
`
`Template DNA
`design and
`production
`
`GMP ANA
`
`
`Cap*
`
`analogue
`
`© bo Marker RNA
`
`
`250:
`
`ion, dose: S00 or 1,000 ug
`
`
` GMP. GcLP, GCP-coimoliant process
`
`WO
`B§cp4
`CD4 and CDa
`@EDtrmunogenic
`
`
`EX Non-imrounogenic
`xisting ECDs
`
`PPPLEPPOPPY
`
`
`
`
`Fluorescence 3oO
`Reference
`
`
`
`CLL NY KML MD NY
`Numberofmutations Total = 125
`
`
`f CD8*Tcell response CD8* T cell response
`
`=CD4
`P06 KIF26B(N256S)
`P19 UTP6(H137Y)
`
`* CD8
`Pre-vaccination Post-vaccination Pre-vaccination Post-vaccination
`3
`s
`ge 008
`ss
`“
`Sw
`
`8 =
`400
`bo
`
`Bg
`ag2 200
`
`DC+OLP4 DC
`z
`
`DC+OLP1
`
`DC+OLP2
`DC+OLP8
`
`800
`
`a“
`
`8
`
`a
`
`Soo
`et
`ae
`
`0
`
`600
`400
`200
`Pre-vaccination
`IFNy spots per 5 x 104 cells
`
`Figure 1 | Broad mobilization of mutation-specific immunity by
`vaccination, a, Vaccine design, manufacturing and clinical study
`
`
`procedure. TAA, tumour-associated self-antigen. b, Two
`synthetic
`
`pharmacologically optimized RNA molecules per patient, each with five
`mutations (pentatepe RNA). c, Denaturing agarose gel and microfluidic
`capillary electrophoresis of RNA. d, UHrasound-guided percutaneous
`vaccine injection inte lymph nodes. ¢-g, Characterization of CD4* and
`a
`CD8* 'fcells after in vitro stimulation with pentatope-RNA-transfected
`autologous irradiated CD8/CD4-depleted PRMCs in ELISpot assays read-
`
`vaccine-induced responses did not or only weaklycross-reacted with
`autologous DCs transfected with the RNA-encoded wild-type epitope
`(Fig. 1g, Extended Data Fig. 2e}. Characterization of stronger wild-
`type cross-reactive responses indicated recognitionofartificial epitopes
`not presented by normal cells (Extended Data Fig. 2f-g). Moreover,
`we found that wild-type cross-reactive responses‘represented mixed
`populations, including T cells, which recognize exclusively the nvutated
`epitope (Extended Data Figs 2h, 3g).
`Responses against one-fifth of the immunogenic mutations were
`detectable in blood without in vitro stimulation (Fi‘ig. 2a}. In patients
`vaccinated with nea-epitapes and shared tamour-associated self-
`
`antigens, neG-epitope responses werestronger (Fig. 2h), probably owing
`to the lack of central immunetolerance.
`For molecular characterization, we cloned neo-epitope-specific
`
`
`Tceil receptor (TCR) a/3 chains fromsingle cells and co-transfected
`therinto Tcells for functional characterization.
`
`Moreover, for 13 of
`the reactivities we determinedthe actual recognized HLAclass [ligands
`within the mutated sequence stretch to enable HLA multimerstudies
`(Extended Data Table 3). TCR-3 sequences ofeight TCRs cloned from
`neo-epitope-specific CD47 and CD8"T cells of two patients (Fig. 2c,
`Extended Data Pig. 3, Extended Data Table 4) were not detectable in
`
`
`TCR deep-sequencing data of pre-vaccination blood saniples
`from
`the patients,|but were abundant in post-vaccination samples (Pig. 2d),
`
`confirming de nove priming of CTLs.
`o
`
`2| NATURE | VOL GO
`
`2017
`| 00 MONTH
`7 Macmillan Pu
`
`
`
`
`
`out on autologous DCs. e, T cell responsestatistics for al} 125 mutations
`on cohort level (pie charts} or per individual patient (bar charts).
`i, Pre- ver
`st-vaccination responses to DCs loaded with one ofthe
`four overlapping l5mer peptides (OLPs) of the respective neo-epitope
`for 10 patients (left) or examples(right). g. Positions of immunogenic
`
`
`eo-epitopes within the pentatope RMA. Control RNA, luciferase; UTR,
`
`
`
`un ransiated region; ORE, open reading frame; SP,
`signal peptide; MITD,
`MHCclass | trafficking domain.
`
`Neo-epitope-specific C8" T cells expanded within 2-4 weeksto
`up to high single-digit percentages, as shownin several patients by
`ex vive HLA multimer blood analysis (Fig. 2e). T cells had a weakly
`PD-1, central- or effector-memory phenotype and were fully func-
`tional with concomitant expression of IPN-y and TNFo(Pig. 2f) on
`antigenstimulation.
`‘The patients bad a recent history of recurrent disease and a high
`risk ofrelapse (Fig. 3a, top, Extended Data Table 5). Comparison of
`documented melanoma recurrencesin all patients before and after
`neo-epitope vaccination(Fig. 3a, bottomleft) showed a highly signift-
`cant reduction of longitudinal cumulative recurrent metastatic events
`(P< 0.0001), translating inte sustained progression-free survival
`(Fig. 3a, bottomright).
`Fight patients had noradiologically detectable lesions at start of
`neo-epitope vaccination (Fig. 3a, top). These patients mountedstrong
`immune responses (Fig. le, Extended Data Fig. 4a) and remained
`recurrence-free for the whole follow-up period (12 to 23 months).
`The other five patients experienced melanomarelapses shortly after
`inclusion and, despite initiation of standard treatment, had progressing
`metastases at start of their neo-epitope vaccination. Ofthese, patient 03
`(P03) and P04 developed neo-epitope-vaccine-related objective clinical
`responses. PO3 exhibited a complete response of multiiple progressing
`
`metastases unresponsiveto local radiotherapy and CTLA-4 blockade,
`and remained relapse-free for 26 months. P04 bad a vaccine-related
`
`
`
` ishers Limited, part of Springer Nature. All rights reserved.
`
`

`

`Eeao
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`g
`=
`
`= 0.04g
`5 0.03
`;
`g 00?
`©
`= 0.01
`0.00
`a
`es
`oe
`
`0.08
`0.04
`0.00
`
`&
`
`0.008
`
`0.004
`0.000
`ge
`e
`
`e
`
`
`© S88 day 11
`Day 15
`Day 22
`Day 20
`pay 43
`bay 50
`302
`,
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`{ nekSs
`
`vuao@
`
`bkd
`
`bkd
`
`
`
`
`
`
`
`
`
`
`
`
`
`3£Ss
`
`nnn OD9 mnnnnnnnnnnnnnnnnnne,
`
`CD197+ CD45RA-
`— CD279--->EI CD197- CD45RA-
`CD197- CD45RA+
`
`partial response ofabdominal lymph node metastases. PO2 hada slowh
`progressing multi-metastatic disease under BRAFinhibitor treatment
`and developed a mixed response upon adding neo-epitepe vaccine. P17
`had an axles lymph node metastasis that remained stableafterthe start
`ofneo-epitope vaccination. it was removed after four vaccine injections
`and used to generate tumourinfiltrating lymphocytes (TILs) and a
`melanomacell line. PO7 developed complete response in combination
`with PD-1 blockade.
`‘three patients provided furtherinsights into neo-epitope vaccine
`ere
`effects. PO7 experienced multiple
`relapses with shortening relapse-
`free intervals and progressing metastases at the start of neo-epitape
`vaccination. Owingto fast disease progression despite a strong response
`against six neo-antigens (five measurable ex viva, Figs Le, 2a}, vacci-
`nation was discontinued and P07 entered a compassionate anti-PD-1
`(pembrolizamab) treatment program. The patient experienced 80%
`reductionin size of multiple melanomalesions within two months
`and eventuallyaa complete response (Fig. 3b). Vaccine-induced Tcells
`
`persisted for up to 9 monthsafter the end ofvaccination (Fig. 3c).
`P1? had responses against all ten vaccine-encoded mutations in
`post-vaccine TILs and in peripheral blood mononuclearcells (PBMCs)
`(Figs le, 3d). In the TILs, we confirmed reactivity against retinol
`gaturase RETSAT(P546S} andidentified the HLA-A*68G1-restricted
`
`ligand derived from this mutation (Fig. 3c}. Characterization ofthe
`
`RETSAT(P5465S)-specific TCR#S identified from TILs by single-cell
`cloning (Fig. 3£»g) unexpectedby showed HLA-B*3701-restricted
`
`
`
`
`recognition ofa different ligand derived from the same mutation
`
`(Fig. 3g, h, Extended Data Fig. 4d-f). TCR#8 conferred efficient killing
`of the autologous melanoma cell line MZ-1-017 derived from the
`post-vaccination sample, but not of autologous APCs (Fig.31).
`P04 was regarded as tumour-free on the basis of magnetic reso-
`nance imaging (MRT scans conducted beforeinitiation of neo-epitope
`vaccination. Follow-up MRI scans showed abdominal lymph node
`metastases missed at the baseline. As PO4 had mounted responses
`against eight mutations and was stable (Fig. le, Extended Data
`Fig. 5a}, vaccination was continued. After 12 applications, residual
`netastases were
`resected and pathologically diagnosed as almost com-
`pletely necrotic. ont"'tcell infiltration, PD-L1 staining and expres-
`sion of immune activation and inflammation markers was increased,
`as comparedto the pre-vaccination melanoma specimen (Pig. 4a,b,
`Extended Data Fig. Sd, e}. Two months after surgery, M RE scans
`revealed new abdominal! and liver metastases. Anb-PD-1 Grivohumab)
`treatment was initiated, but the disease progressed rapidly and the
`patient died.
`‘To understand this unexpected treatment failure, we revisited
`the immunological data. T cefls transfected with TCRs from CD8t
`ec
`TiLs directed against the mutated neo-epitopes FLNA(P639L) and
`
`7LL
`CDC37L1(P183L) recognized the nutated sequences with high
`
`sensitivity (Extended Datta Fig. 5b, c}), whereas MZ-GaBa-G18, the
`autologous melanomacell line from the post-vaccination resectate,
`was not recognized (Fig. 4e).
`MZ~GaBa-018 cells expressed HLA class 1, but showed no con-
`stitutive or IEN+-inducible HLA class I surface expression (Fig. 4c).
`
`it
`2017 Macmillan Pubtishers Limited, pa
`
`ry
`
`t of Springer Nature. All right
`
`VOL G00 | NATURE |
`
`Ex vivo responses
`ws
`
`SOLES
`ass
`P19 IMPDH2(P23.4h)
`P19 CLNTI (14721)
`P19 UITPAILI 37%)
`P09 GSTK1(G238R)
`POS MAN A2(E3250)
`PitGOKAnOdL
`Pt borAVALOINS&ON}
`porAbrsize)
`POTEraeaaroy
`Poa CDG©eel)
`PO1 NARFL(E62k}
`
`Ex vivo responses
`
`Oo
`
`1,500
`1,000
`500
`IFNy spots per 10° cells
`
` cv
`600
`” 500
`3
`o
`é
`x, 00
`a
`300
`2 >00
`&
`~& 100
`
`“
`
`a
`aa
`a
`ane
`a
`Aaga
`-epi
`Shared Neo-epitope
`
`e
`°
`
`Mutated
`iid
`typ
`wie
`
`tm Cpa
`
`CD8*T cells
`
`(Ron, PO NARFL(E62K) 5
`4
`BR00% Su.ghch
`Ser oS
`wee
`
`GD8*T cells
`CRopgPO1 NARFL(E62K)5
`
`Figure 2 | Rapid expansion of neo-epitepe-
`specific Tcells with central and effector
` phenotvpes by vaccinatl
`memory phenotypes by vaccination,
`b, PBMCs read-out without
`previous
`a,
`3
`.
`:
`.
`TAG
`MCs read-out without previcus
`a,
`dD,
`
`expansion against neo-epitope-RNA-loaded
`_
`autologous DCs in ELISpot.Mann-Whitney
`
`:
`test (FFP == 0.0017). c, Activation-induced [1Ny
`-secretion-basedsingle-cell sorting of CD38"
`wn
`oy
`:
`TP cells after in vitro stimulation with neo-
`ae
`or
`epitope-RNA-loaded autelogous DCs for TCR
`cloning. Control, luciferase RNA. Healthy-
`5
`>
`.
`Ss
`ome
`5
`
`donor-derived CD8* T cells co-transfected
`
`CRopg P02 PPFIA4(S709N)—TCRop,P01 NARFL(IE62K) TCRopg P02 HPN(G71R) with RNAs encoding the identified TCR-o
`
`
`
`ome
`one
`one
`.
`2
`/Q chains tested on peptide-pulsed R562cells
`0.05:
`0.012
`S88 Pre-vacc. |
`:
`.
`.
`—_
`t
`0.12
`ic 40
`expressing the patient’s individual HLAalleles.
`2 a0.
`
`d, Frequencies af neo-antigen-specitic TCRs.-
`not detectable; TOR control, irrelevant-
`2 20
`
`8
`3 sequence. e, Kinetics and
`&
`i
`5
`
`31°
`phenotype
`elected Cps* T cell responses.
`
`
`
`
`DCs +
`0
`PEMCsstained with neo-epitepe-specific. a a 1 _ ~ 4
`
`
`
`
`
`multimers (left). See also Extended Data
`Fig. 4b, c,
`& for specificity of selected dextramers.
`BRAC
`etal
`Fo OT
`PBMCsstained for CD279 (PD-1} within the
`multimer-positive (red) and -negrative Cola“AS
`cell population Guid}. MemoryTcell subse
`based on CD45RA and CDi97expression
`(night). €
`intracellular cytokine stainingof cDs*
`¥ cells co-cultured with RNA-transfected DCs.
`Control, DCs without RNA.
`
`S
`B
`<
`8
`<
`
`&
`
`

`

`Figure 3 | Disease control by vaccination in melanoma patients with
`high risk of relapse. a, Recurrence and treatment (top) and progression-
`free survival (bottom, right) of patients. Cumulative sum of metastatic
`events per month before (prey) or after (green) neo-epitepe RNA
`
`vaccination (bottom, left}, Patient no., number of monitored patients.
`Fisher's exact test comparingthe cura ulative observationtime without
`a metastatic event to the umber of months with an event (P < 0.0001,
`
`
` lesions and vaccine-(bottom, left). b, <, Computer tomographyof target
`
`induced
`ex vivo responses by ELISpotfor P07. d-i, Vaccine-induced T cell
`
`responsesofP17. TILs and the tumourcell line MZ-I-17 derived from 4
`lymph node metastasis resected after four RNA neo-epitope vaccinations.
` Qo
`d, TIL reactivity against individual neo-epitopes e, HLA multimer staining
`>
`-
`Ie
`ofHLA-A*6801-restricted CD8* TILs recognizing a RETSAT(P5465)
`Ipilimumab
`|
`R Radiotherapy
`V Vemurafinib
`+ Death
`Y Metastasis (resected)
`
`minimal epitope. £, LEN--secretion-basedsingle-cell sorting of CD87 'TELs
`
`D Dabrafenib
`P Pembrolizumab
`L Lostto follow-up a Interferona
`Y Metastasis
`for TCRcloning after co-culture with RNA-transtected| Cs. Contrel,
`
`~ Neo-epitope RNA vaccination+Initial diagnosis|* Measurable lesion at start of RNA vaccination
`
`
`eGEP RNA. ah CD8t T cells expressing TCR#8 cloned from single
`thik
`
`
`TILs weretested for recognition of peptide-pulsed K562 cells transiected
`with individual HLAalleles of the patient by ELISpot (g) or forkilling of
`
`
`
`Pi7-derived target cells by caspase 3/7 or luciferase cytotoxicity assays:
`(i), Controls, CDs* T cells without TCR RNA (left); autologous CD!i4t
`
`cells + OLBs8 Grighb. Results oftriplicates (mean +s.d.). h, Nes-epitope
`
`
`~&Pre neo-epitope RNA vaccination
`30 +> Post neo-epitope RNA vaccination
`ee
`aan SS
`se
`2
`s
`
`:
`#
`a
`
`8 £
`a3
`Se
`oo 20
`g
`ag
`E® 10
`GE
`
`So
`aa
`BEsaanganth SETSSOTATST
`Months 13 6
`9
`12
`15
`18
`21
`24
`wy
`ou
`Down
`Mao
`Patient No.
`Cumulative
`observation
`time (month)
`
`
`
`SSN|
`
`+346 +9 412 415 418 +21 +24
`
`-6
`
`-3
`
`-24 -21 -18 -15 -12 -9
`
`Stage Patient
`Mo
`POF eect
`
`
`NV
`PO2*
`
`Vv
`PO3* <-
`
`VV
`PO4*
`Ilb
`P05 <
`
`
`AyCR a
`
`
`75
`
`gs 100
`2
`=
`z
`o
`Z
`7
`£ 50
`5
`@
`Boos
`=
`*
`
`0
`
`0
`
`21 24 27
`18
`15
`12
`9
`6
`3
`Time to progression (months)
`
`Ycells ¢
`
`
`
`sample ofP04 stained
`5g). Whereas the pre-treatment tumour
`Extended Data Fig
`almont homogenously for B2M,all tumourcells
`ere B2M-negative (Fig. 4P).
`in the post-vaccine resectate
`
`all, ourfindings Hlustrate that an unex:pectedlybrroad repertoire
`Ov
`of fTcells is recruited }by reo-epitope vaccination. Every patient,
`including individuals with relatively low mutational load, raised
`poly-specific immune responses. We found that one mutation may
`give rise to neo-epitopes either recognized concomitantlyby cp4t
`and CD&” T cells or presented on different HLA class I restriction
`elements to different CD8* Tcells. Likewise, the same neo-epitope-
`HLA combination was foundto be
`gnized by neo -epitope-specific
`
`T cells with different TCR clonotypes
`(Extended Data Table 4).
`Onlya fraction ofpredicted high-affinity HLA class | binding routant
`peptides are expected to be a naturallypresented andjimmunogenic
`HLA class { ligand!*!*_ Accordingly, we observed CISresponses
`against only 20 of 69 (29%) of predicted high-affinity class I binders.
`
`sx New lesion subcutis chest
`= Index lesion subcutis chest
`--:- New lesion subcutis gluteal
`
`4004
`
`b
`
`_
`E
`£
`8
`a
`8
`an
`4
`
`c > P07 ARF3(R99Q)
`RHRPOF Mill4(L2569F)
`« 1,000
`o
`6
`& 100
`oa
`2
`g
`g
`°°
`e
`=Zz
`wo
`
`10
`1
`
`& ce
`
`+ POY ZNF287(H470Y)
`-x-- POF HYAL2(N380H)
`->- POY GPX1(S95F)
`
`gS
`
`We
`
`s
`°—_—_——
`Anti-PD-1
`
`
`ff SEN
`d Ss
`
`HECVMASLR _(A°6801)
`PI7
`P17 RETSAT(P546S) Fran
`7
`-
`P17 RETSAT(P5468)
`Control
`RETSAT(P5465)
`EiaguaeaisocdL|
`°
`t 0.03 |
`OF
`P17 MYO1E(H284
`
`3
`.
`7
`P17 GPNMB(P382F|
`E
`.
`Pentatope B
`P17 GAK(A315Vj
`=
`Control
`P17 GBP1(P86F}
`=
`i
`P17 RWDD3(S259F)
`P17 GYPC(AI 14) me CD8 =>
`P17 TPR(SS30F,
`Pentatope A
`Control
`
`‘
`
`=
`z
`=
`=.
`|
`CDB
`\
`2
`ae
`&
`A
`A
`
`pee? poyol ‘eBE?
`
` ©CD8* T cells
`TCRopg P17 RETSAT(P5468)8
`
`100 200 300 400 500
`0
`IFNy spots per 1 x 10° TILs
`
`
`
`
`
`
`CD8* T cells + TORopg P17 RETSAT(P5468) 8 a
`
`
`
`However, 12 afae© roinimal CD8" neo-epitopeswe‘identifiedhavea
`good predicted bindingaffinity (Extended Data Table 3, Supplementary
`fables 1, 2). Thus,despite its relatively low specificity, HLAclass Pneo-
`epitope prediction enablesselectionof strongly immunogenic ligands.
`Likewise, HLA class Ifbinding§prediction correlated well with
`immunogenicity: 21 of 30 (70%), 30 of 67 (45%) and 9 of 26 (34%) of
`mutated sequences with a HLAclass I bindingscore of <1, 1-10 or
`> 10, respectively, induced CD47 responses. About 20%of the
`responses were raised against neo-e pitopes with poor predicted HLA-
`
`binding scores, indicating the need for improvementofthe sensitivity
`
`eT
`ET
`ofour algorithms!*!". Post-vaccination biopsies from patients Pod
`3
`100 7ae 61
`§ 800 gan 25:;
`and P17 enabled us to evaluate theeffects of vaccine-induced immune
`=
`2 80
`= 700
`N
`2
`G
`responses on the patient's tumour. We confirmed infiltration of
`gy
`cy
`= 60
`= 600
`vaccine-induced neo-epitope- specific Tcells in the respective tumours.
`
`N—& 500 ° 40
`2
`B
`9
`We showedrecognition andefficient killing of autologous melanoma
`2
`8 400
`20
`6
`8
`:
`S 300
`0
`cells but not autologous monocytes mediated by neo-epltope-specific
`Aw 8
`-
`z
`TOR +
`8
`oe
`:
`TCRs. Collectively, these findings provide the complete chain of
`
`
`
`Tumourcell +0+N MZ-47: ay eo x
`=
`SS
`abSe
`evidence proving that the neo-epitope vaccine executes its mode of
`action iM vivo.
`Ourobservations indicate that neo-epitope vaccines alone may
`prevent recurrent disease in highrisk patients. Morcover, they provide
`a rationale for combining the vaccine with PD-1/PID-Li blockade’®?”
`Neo-epitope-specific T cell subsets were PD-11? and of memory
`phenotype, and post-vaccine lesions were shownto upregulate PD-L1.
`Upon anti-PD-1] treatment after neo-epitope vaccination, FO? rapidly
`developed a complete response, the likelihood of which is otherwise
`reported to be below10%?
`
`Poly-neo-epitopic immunity reduces therisk ofoutgrowth of single
`neo-antigen loss variants’, bat completeloss of HLAclass I presentation
`
`
`
`
`
`HLAclass I transcription was verified by next-generation sequencing
`ONGS) of MZ-GaBa-018 (RPKMvalues: HLA-A, 279; HLA-B, 207;
`HLA-C, 292), but not §2-microgiobulin transcripts (RPKM== 0).
`Whole-exome sequencing revealed loss of both alleles of B2M and
`TRIMS9 genes by a deletion-inversion event (Fig. 4d, Extended Data
`Fig. 50). Transtection of MZ-GaBa-018 with Bai RNAreconstituted
`
`
`HLA class [
`surface expression (Pig. 4c} and killing by Tcells repro-
`grammedwith the cloned neo-epitope-specific TCRs (Fig. 4e), as well
`
`as by autologous neo-epitope-RNA-stimulated blood-derived CD8*
`
`4 | NATURE
`
`VOL 000 } 60
`
`

`

`.PNK1Re
`
`2.
`3.
`A.
`
`
`es drive therapeutic immune
`
`
`ninessensitivity to PD-1 blockade
`
`|
`ce 348, 124-128 (2015),
`A of acGS
`:
`is for clinical
`response to CTLA-4 biockade in
`
`
`
`melanoma,NEEngl I Med, 372, 2189-2199 (2014)
`
`5.
`VanAller
`il, Genomic conreiates of resporise te CTLA-4 blockadein
`metastatic &matanoma. Science 350, 2
`id 201 5
`
`
`
`6. MegGranahan, N. ef al. Clonal neoan
`is
`elicit T cell
`immunoreactivity and
`ity to immune che
`;
`ckade. Science 351, 1463-1469
`
`
`
`arpeting Mutant KRASin cancer. N. Enel J
`requent
`an melanoma, Nat, Med.
`
`7.
`&.
`
`3.
`
`10.
`
`
`T-ceill-dependen
`
`ations in hur
`
`gastrointestinal
`
`
` &. Chen, B.S “g Malin an, |. Elements of cancer imrnunity and the cancer-
`
`evant
`
`
`come analysis reveals a
`sfNature482, 400-404(201"
`
`
`cancers. Sciencco 350, 1387-12
`
`
`L. Yadav, M. etal Predicting im
`ric turnour mutations by cornbining mass
`spectrom \
`ome 8
`Neture 535, 572-57 62O14)
`
`
`, MLM. et afl Checkpointt block cancer immunotherapytargets
`[2,
`specific mutant anttigen. Nature 515, 577-581 (2014),
`
`
`cer immunotherapy. A dendritic cell vaccine
`ena, B. M. etal
`3.
`increases
`
`
`
`a diversity of f
`nelanoma neoantigen-specitic T cells. Science
`248, 803-808 (2015),
`th arapy. Meo aoproaches
`14. Delamarre, L, Mellman, |. & Yadav, M. Cancer im
`
`to cancer vaccines. Science 348, 760--761 (015).
`
`5. KwalesSS.
`7
`anodal vaccirnation with rakedweenane
`
`tent pr
`
` Sternb
`
`
`
`
`
`
`necepitopes pres
`Oy
`e human melarnoma tis
`
`
`spectrometry. Nai
`rimun. 7,
`13404 (2015).
`
`as$ specirometry profiling of HLMasso iate
`AbeliniJ. G. etal Ma
`
`enables rmiore
`accurate epitope prediction. inneriunily 46,
`in mono-~aliere celis
`
`
`
`S41, 321-230 (2017).
`mune set point Nature
`
`Tumeh, © G.
`etal, ve bieckade induces responses by inhibi
`lature 515, 568-571 (
`
`sab versus ipilimumab in advanced melanoma
`
`2832 (2015).
`
`
`J antigen expression by cultured
`i
`ane expression, J Clin. favest.
`
`
`
`20.
`
`2i
`
`
`
`23. Melero, Le
`
`
`ia combat caricer.Nat.Ro Canneer 15, 457-47
`
`
`rer, LG
`and 3. Voer,
`
`Ecker, M. Lechsctnt 8
`S$Weessel, C.
`
`
`
`6S. WOl, C
`Barea Roldan, C. Wa
`i
`M. Drude §Pets MLM ler, L
`
`2,0. Britten,on
`Castle and B. Pless for techs ‘cal suoport project
`
`nc
`aclvice. We thank A. Tuttenberg for
`port with a figure. We
`
`
`jJelamarre and G. Fine for critical readingof the rnanuscript.
`
`
`
`ar her advice. Th
`tudly was supported by the Ci3 cluster
`eral Ministry of E
`ation and Research (BMBF).
`Author Contributions US. conceptualized the work and strategy. E.D., Pe.Si,
`TO. OSL, SA, AR. and BK planned and analysed experiments. £.G, RR,
`
`AB. C.T and AH. dic experiments. MA
`CA. APand PaSo,
`
`performed and
`sed NG
`Ss. A.
`nanufactured the RNA
`
`
`
`
`vaccines, G.W., MLW
`Z
`ormed a
`nee. VM. BK,
`§.&SH.
`
`Pier.ae AK-B.. DL and3.8
`
`
`
`i
`ini sees. s. CL. is the pri ci
`
`a
`cal grace
`
`.
`interpreted data and wrotethe:manuscript. A
`
`
`nofthe manuscript.
`
`nformationis available at
`
`peeS
`
`re competing financial interests:
`
`per. Readers
`are welcome to
`
`
`:
`tur
`regard ta wurist
`
`
`
`ons. Correspondence2and re
`insti
`s should be
`
`adaresse
`dito UW,
`S. Sahin@uni-rnainz.de).
`Reviewer Information Natu
`anks C. Metiet and the other ano
`
`
`
`ionto the peer review of this work.
`reviewer(s) for their conirisu
`
` A
`
`Ml Post-
`vacc.
`
`vacc.
`
`a 250
`&
`E 200
`2
`g
`3
`= 150
`5 100
`5
`a
`@ 50
`8
`
`0
`
`5
`
`9a
`2
`6
`3
`5
`=
`oa
`a
`
`By
`%
`& &
`
`MZ-GaBa-0718
`
`Untreated
`IFNytreated
`82M transfer
`71
`4 90.8aes
`
`
`c
`
`x=
`<
`
`=
`
`
`
`: aSECON.
`
`SLAY
`
`
`MZ-GaBa-078.
`
`
`f
`
`Pre-vaccination
`Seen
`
`
`
`
`
`
`
`
`
`
`
`
`
`TILspost-vacc.
`
`
`
`
`
`
`
` 100
`
`a
`
`Pre-vaccination
`
`Post-vaccination
`
`SS Necrosis
`Whitespace
`
`S
`Scalabar: 2
`SS Tumourtissue
`8 Normaltissue
`s Artefact
`jackground
`HIAKS! FEV
`b riusomyiv
`(A*0201)
`(A*0201)
`GILGFVFTL
`Po4
`po4
`(A*0201)
`
`CDC37L1(P186L) FLNA(P369L)
`
`Influenza M1.
`0.0:
`0.01
`(0.117
`
`
`
`odGo
`Ssoo
`o6
`
`27
`<
`:
`ue
`i =, [tS
`2
`KSI:
`(0.03
`
`0.54
`
`=
`=
`S
`
`e SSS
`
`GDS* T celis
`TCRopg P04 GDG37L1(P186L) 2
`TCRopg P04 FLNA(P369L)
`<
`&
`100
`ES:
`80
`=
`x
`:
`s
`=
`. x
`+
`$F
`
`80
`
`SB
`= 60
`é
`5
`10
`bok.
`20
`o
`TCR: -
`BME +
`-
`MZ-GaBa-O16: +
`
`JES
`
`x.
`=
`A
`iN
`MS
`AE
`ES: x aS
`+
`+
`+
`=
`+
`+
`
`-
`-
`+4
`
`100
`80
`
`60
`40
`20
`0
`
`-
`t+
`+
`
`=
`o-
`oF
`
`-
`=
`Ee
`=
`60
`.
`S
`40
`
`=
`20
`7 =
`0
`+ Pw> A ON
`
`
`= oe Oo
`x
`+
`
`Figure 4 | Neo-epitope-induced CTL responses associated with
`immune escape by outgrowth of B2M-deficient melanoma cells in PO4.
`
`a, Composition of metastases by computerized visualization and image-
`
`
`analysis-based quantifi
`nm (right, mean-+s.d. of triplicates).
`
`b, Frequency of CD8* Tcells against
`two nea-epitopes in blood and
`TLLs of a post-vaccination lesion detected by HLA multimers. Control,
`influenza M1 (A*0201)}. ¢, HLA surface expression of MZ-GaBa-0158 under
`Gifferent conditions. d, Genomic mapping of the deletion and inversion
`event leading to B2Mloss. e, Specific lysis of PG4-derived. target cells
`by neo- epitope TCR-transfected cpat Tcells measured by luciferase
`ytotoxicity assay. Controls: MZ-GaBa-018 cells without B2Mtransfer,
`[8+ T cells without TCR RNA Ueft and middie}; autelogous CD14*
`cells + FLNA-P269L OLP (right). Results of triplicates (mean + s.d.).
`f, B2Mstainingeof melanomacells in pre- and post-vaccination metastases.
`
`
`remains an effective escape mechanism’. The outgrowth of B2M-
`deficient tumour cells in the presence of poly-neo-epitope-specific
`inamunityin PG4 indicates that neo-epitope vaccinesare potent snoast
`to evoke the same resistance mechanisms as checkpoint blockade and
`adoptiveT cell transfer®?*. This risk can he mitigated by combining
`mutanome vaccines with immunotherapies that do not rely on intact
`HLAclass I presentation’. In summary, our study demonstrates the
`clinical feasibility, safety and antitumouractivity oftargeting individual
`cancer routations by RNA neo-epitope vaccines, thereby supporting the
`case for making individuallytailored medicines accessible to a wider
`rangeof patients.
`
`Gniline Content Methods,
`along with anyadditional Extended Data display
`
`
`Source Daia, are available in the online version of the paper: referencesurnig.ue tG
`
`these sections agpear onily in the online oager
`
`Received 30 January; accepted 6 June 2017,
`Published online 5 July 2027,
`
`
`
`
`nome for tumor vaccination. Cancer Re
`
`1
`
`2017 Ma ian Puoishers Limit
`
`cO MONTH
`ed, part of Springer Nature. All rights reserved,
`
`

`

`
`
`Data reporting. No statistical methods were used to predetermine sample size.
`The experiments were not randomized and the investigators were not blinded to
`
`allocation during experiments
`and outcome assessment.
`Study design. The main objectives of this multicentre phase I study
`(NCT02025956) were to assess safety of the vaccine and vaccine-induced
`m
`antigen-specific immune responses.
`accordance with the Declaration of Helsinki
`
`The study was carried out
`
`and good clinical practice guidelines and with approval bythe institutional
`
`review board or independent ethics committee of each participating site
`and the
`competent regulatory authorities. All patients provided written informed consent.
`
`Eligible patients were > 18 years old, and had malignant melanomastage HLA-C
` 2
`
`or FV C(AJCC2009 melanoma classification) in complete remission, partial remis-
`sionotstable disease at anystage of treatment. Patients with metastases were eli-
`
`gible if they could be treated with an active compound until availability of their
`individualized vaccine. P
`
`ents required adequate haematological and end-organ
`function. Key exch
`
`sion criteria were clinically relevant autoimmune disease, HIV,
`HBV, HOVand acute EBV or CMVinfections and brain metastases.
`Regular treatment was eight injections within 43 days; continued treatment
`was left to the investigators’ discretion. The RNA pentatopes were diluted in
`9
`10mg mi! Ringer's solution Rotexmedica ov BAG Healthcare) and injected into
`separate inguinal lymph nodes. Ten patients were administered 500i1g and three
`
`patients 1,000 ug per treatment to explore two different dose ranges.
`
`Keystudy assessments. Leukaphereses for immunogenicity testing were
`
`
`
`formed before the first (visit 12, referred to as ‘pre-waccination’) and after the
`
`&th vaccine injection (visit 20; referred to as ‘post-
`vaccination).
`Imaging of thorax, abdomen, brain by CT scans and MRI were performed
`
`at baseline (visit 1}, pre-vaccination (visit 12), day 99 (visit 21) and at endof
`continued treatment (visit 26) according to the local imaging guidelines and
`RECISTversion 1.1 and the immune-related responsecriteria GrRC} guideline”
`Satety was characterized according to CTCAE v4.03 from grade