ERAP1-Dependent Antigen Cross-Presentation Determines Efﬁcacy of Adoptive T-Cell Therapy in Mice Karin Schmidt1, Christin Keller1, Anja A

Published OnlineFirst March 20, 2018; DOI: 10.1158/0008-5472.CAN-17-1946

Cancer Tumor Biology and Immunology Research

ERAP1-Dependent Antigen Cross-Presentation Determines Efﬁcacy of Adoptive T-cell Therapy in Mice Karin Schmidt1, Christin Keller1, Anja A. Kuhl€ 2, Ana Textor3, Ulrike Seifert1, Thomas Blankenstein3,4,5, Gerald Willimsky4,6, and Peter-Michael Kloetzel1,5

Abstract

Cytotoxic T lymphocytes can reject established tumors if their T cells in each case. ERAP1 expression by antigen cross-presenting target peptide is efficiently presented by MHC class I molecules cells of the ATT recipients was critical for expansion of therapeutic (pMHC-I) on the surface of cancerous cells. Therapeutic success monospecific T cells and correlated with tumor rejection. Specif- upon adoptive T-cell transfer (ATT), however, requires additional ically, lack of ERAP1 expression in the ATT recipient's noncan- cross-presentation of the same pMHC-I on noncancerous cells. cerous cells enabled progression of pMHC-I–positive, IFNg- Endoplasmic reticulum aminopeptidase 1 (ERAP1) is an enzyme responsive tumors, despite the presence of antigen-specific func- that customizes the N-terminus of proteasome-generated pep- tional cytotoxic T lymphocytes. These data reveal a decisive role tides so they can be loaded onto MHC-I molecules in the endo- for ERAP1 in T-cell–mediated tumor rejection and will enhance plasmic reticulum (ER). We show here that ERAP1 is critically the choice of MHC-I–restricted epitopes targeted by adoptive T- involved in the process of tumor rejection and assumes a dual role cell transfer. by independently operating on both sides. Direct presentation of Significance: This study demonstrates a role of ERAP1 in the two MHC-I–restricted epitopes of a cancer-driving transplanta- efficacy of adoptive T-cell transfer and has potential to improve tion rejection antigen through ERAP1 moderately affected tumor personalized T-cell therapy for solid tumors. Cancer Res; 78(12); rejection by adoptively transferred T-cell receptor gene–modified 3243–54. 2018 AACR.

Introduction Moreover, adequate T-cell persistence decides on the effective- ness of adoptive T-cell transfer (ATT), because the number of Oneofthemosteffectivetherapiesforpatientswithcanceris transferred T cells and the degree of their persistence in periph- the transfer of tumor-infiltrating T lymphocytes with objective eral blood correlates with cancer regression (1, 6). response rates >50% being achievable (1). Successful elimina- On the side of the target epitope, classical processing of MHC-I tion of cancer depends on both, efficient direct presentation ligands involves three consecutive steps assumed by proteasomes, of the targeted MHC-I epitope, and on the cross-presentation the transporter associated with antigen presentation (TAP), and of the same MHC-I epitope on noncancerous cells (2–5). endoplasmic reticulum aminopeptidase 1 (ERAP1). The majority of peptides presented on MHC-I molecules on the cell surface is generated by proteasomes and possesses a C-terminal residue 1Institute of Biochemistry, Charite—Universitatsmedizin€ Berlin, corporate member suitable to act as anchor for MHC-I binding (7). Those peptides of Freie Universitat€ Berlin, Humboldt-Universitat€ zu Berlin, and Berlin Institute of Health (BIH), Berlin, Germany. 2iPath.Berlin—Immunopathology for Experimental are transported into the endoplasmic reticulum (ER) by TAP (8). In Models, Charite—Universitatsmedizin€ Berlin, corporate member of Freie Uni- many cases the epitope-containing peptides are N-terminally versitat€ Berlin, Humboldt-Universitat€ zu Berlin, and Berlin Institute of Health elongated, requiring further optimization in the ER. Here, ERAP1 (BIH), Berlin, Germany. 3Max-Delbruck-Center€ for Molecular Medicine, Berlin, trims amino acid residues that flank the N-termini of antigenic 4 Germany. Institute of Immunology, Charite - Universitatsmedizin€ Berlin, precursors (9, 10). In the context of cancer immunotherapy, it € € corporate member of Freie Universitat Berlin, Humboldt-Universitat zu Berlin, was shown in mice that inhibition of ERAP1 caused increased and Berlin Institute of Health (BIH), Berlin, Germany. 5Berlin Institute of Health, Berlin, Germany. 6German Cancer Research Center (DKFZ), Heidelberg, Germany. tumor immunogenicity through direct presentation of a neo- epitope targetable by specific T cells (11). But when tumors evaded Note: Supplementary data for this article are available at Cancer Research therapy with T-cell receptor (TCR) gene-modified T cells, recog- Online (http://cancerres.aacrjournals.org/). nition of IFNg-resistant cancer variants by cytotoxic T lymphocytes G. Willimsky and P.-M. Kloetzel contributed equally to this article. (CTL) was reconstructible by overexpression of ERAP1 in vitro (12). Current address for U. Seifert: Friedrich Loeffler Institute for Medical Microbi- In contrast with the well-analyzed direct presentation of ology, University Medicine Greifswald, Greifswald 17475, Germany. antigens, the procedures of antigen cross-presentation remain Corresponding Authors: Karin Schmidt, Institute of Biochemistry, Chariteplatz enigmatic. Considering the necessity of N-terminal trimming 1, 10117 Berlin, Germany. Phone: 4930-4505-28395; Fax: 4930-4505-28921; of peptides, Saveanu and colleagues (13) demonstrated that E-mail: [email protected]; and Peter-Michael Kloetzel, Institute of Bio- insulin-responsive aminopeptidase is required for efficient cross- chemistry, Chariteplatz 1, Berlin 10117, Germany. Phone: 4930-4505-28071; presentation of endocytosed ovalbumin (OVA) and of phago- Fax: 4930-4505-28921; E-mail: [email protected] cytosed antigen. Information on the role of ERAP1 in antigen doi: 10.1158/0008-5472.CAN-17-1946 cross-presentation is barely available and is mainly restricted 2018 American Association for Cancer Research. to the analysis of the model antigen OVA (13–15). Here,

www.aacrjournals.org 3243

Schmidt et al.

we demonstrate the decisive role of ERAP1 for therapeutic eter and animals were excluded from analysis if they died from outcome, showing that for optimal efficacy of ATT, ERAP1 reasons unrelated to tumor burden. The experimenter was not must operate on both sides, direct antigen presentation in blinded for the treatment groups. cancerous cells and antigen cross-presentation in the ATT recipient's cells. Quantification and statistical analysis Comparison of two groups was done by the Mann–Whitney Materials and Methods test. Comparison of more 3 groups was done by the Kruskal– Wallis test. Two-way ANOVA was used followed by Bonferroni Mice post-test for multiple comparisons. All statistical analysis was LoxP-Tag and LoxP-Tag Alb-Cremiceweredescribed / done with GraphPad Prism software version 5.0 and considered previously (16, 17). Erap1 mice were provided by K. Rock significant at , P 0.05; , P 0.01; and , P 0.001. (18). Erap1 / mice were crossed to LoxP-Tag and Alb-Cre mice to obtain Erap1 LoxP-Tag Alb-Cre mice. SCID mice (C.B-17, strain code 236) were purchased from Charles River Results / tm1Fwa / Laboratories. Rag mice (B6.129S6-Rag2 ), Rag Two MHC-I epitopes differ in their dependence on ERAP1- / tm1Fwa tm1Wjl gc mice (B10;B6-Rag2 Il2rg ), and CD45.1 mice mediated N-terminal peptide trimming were purchased from Taconic and were bred in our animal The cancer-driving antigen SV40 T Antigen (TAg) contains two facilities at the FEM Charité-Universitätsmedizin Berlin. codominant MHC-I–restricted epitopes, the H2-Db-restricted / / / b Erap1 mice were crossed to Rag mice to obtain Erap1 10mer TAg – (SAINNYAQKL, TAg-I), and the H2-K –restrict- / / 206 215 Rag mice. P14 Rag mice were described previously ed 8mer TAg – (VVYDFLKC, TAg-IV; ref. 19). In Textor and 404 411 (12). C57BL/6 mice were provided by the FEM at Charite colleagues (12), we showed that both standard and immunopro- € Universitatsmedizin Berlin. Male or female mice ages 2 to teasomes generated TAg-I along with the potential epitope pre- 9 months were used in animal experiments. All mouse studies cursors TAg – (VSAINNYAQKL, 11mer) and TAg – € 205 215 204 215 were approved by the Landesamt fur Gesundheit und Soziales, (RVSAINNYAQKL, 12mer). When peptide translocation was Berlin, Germany. assessed, ATP-dependent transport was observed for TAg-I and the 11mer precursor peptide. The 12mer TAg-I precursor peptide Hepatocellular carcinoma cell lines did not translocate in an ATP-dependent fashion (Fig. 1A). TAP- þ þ þ To generate WT (Erap1 / )TAg hepatocellular carcinoma dependent transport was confirmed by reduced accumulation of þ (HCC) and Erap1 / TAg HCC lines, livers were removed ATP signal in the presence of a high-affinity competitor peptide from tumor-bearing Erap1 LoxP-Tag Alb-Cre mice. (Supplementary Fig. S1A). Upon analysis of N-terminal trimming Tumor tissue was cut into pieces with scalpels and was of the precursor peptides by recombinant mouse ERAP1 digested in RPMI supplemented with 10% heat-inactivated (rmERAP1), TAg-I was generated from the 12mer through the FCS (Biochrom), 1 penicillin/streptomycin (Biochrom), 11mer intermediate (Supplementary Fig. S1B; Fig. 1B and C). 1 mg/mL collagenase (Sigma), and 1 trypsin (Biochrom) Hence, though TAg-I is generated by rmERAP1, the consecutive for 4 hours at 37 Cinahumidified 5% CO2 incubator. To action of proteasomes and TAP may even provide sufficient prepare a single-cell suspension, digested tumor tissue was amounts of the 10mer. filtered through a 45-mm cell strainer (BD Biosciences) and In contrast, TAg-IV was difficult to detect after proteasomal washed with PBS twice. Single-cell suspensions were cultured cleavage, whereas the corresponding N-terminally extended epi- in RPMI supplemented with 10% heat-inactivated FCS, 1 tope precursor peptides TAg403–411 (SVVYDFLKC, 9mer) and penicillin/streptomycin, 2 mmol/L glutamine (Biochrom), TAg402–411 (DSVVYDFLKC, 10mer) were predominantly pro- and 50 mmol/L b-mercaptoethanol (AppliChem). Adherent duced (12). In peptide translocation experiments, all three HCC cells were detached from cell culture flasks with trypsin TAg-IV–containing proteasomal cleavage products were trans- (Biochrom) and were frozen in 90% culture medium sup- ported in dependence of ATP and TAP (Fig. 1D; Supplementary plemented with 10% dimethyl sulfoxide. Cells were culti- Fig. S1C). For the intermediate 9mer peptide precursor, translo- vated for a maximum of 10 passages after thawing. HCC cell cation efficiency did not exceed the experimental threshold set lines were regularly authenticated by Western blot analysis, by a control peptide (E5) not binding to TAP. When the TAg-IV– but were not tested for Mycoplasma contamination. containing N-terminal precursor peptides were subjected to rmERAP1-digestion, TAg-IV was generated from the 10mer pre- Tumor challenge and adoptive T-cell transfer cursor peptide through the 9mer intermediate, and thus may Age- and sex-matched mice were subcutaneously injected constitute an ERAP1-dependent epitope (Fig. 1E and F). into the right flank with 1 106 HCC cells in 200 mLPBS. / þ Tumor size was measured with a caliper and the average tumor Generation of primary wild-type and Erap1 TAg volume was determined from the measurements along three hepatocellular carcinoma with IFNg-inducible antigen- orthogonal axes (x, y, z). Tumor volumes were calculated processing machinery according to the formula V (mm3) ¼ (xyz)/2. On the day of Although both immunoproteasomes and TAP control the treatment mice received intravenous injections of either 1 106 quantity of peptides being available for MHC-I loading, ERAP1 þ polyclonal CD8 T cells, 5 104 gene-modified TCR-I T cells, regulates the quality of the presented peptides (20). To assess þ or 5 104 gene-modified TCR-IV T cells re-suspended in 200 mL the role of ERAP1 in vivo, TAg tumors genetically depleted for PBS. A small amount of blood was taken from the facial vein of Erap1 / were generated through crossing of Erap1 / mice to the mice one and/or 4 weeks after adoptive transfer. Animals LoxP-Tag Alb-Cre mice, which express TAg specifically in þ were sacrificed when the tumors reached 15-mm mean diam- hepatocytes and develop TAg HCC and cholangiolar carcinoma

3244 Cancer Res; 78(12) June 15, 2018 Cancer Research

ERAP1 Is Crucial for Peripheral T-cell Expansion

AD 120 ADP 120 ADP Figure 1. *** ATP *** ATP 100 100 Two MHC-I epitopes derived from the same tumor antigen show varyingly 80 80 strong dependence on two *** consecutive steps of antigen 60 60 processing. A, Transport assay for *** * 40 TAg-I and related N-terminally 40 *** *** extended precursor peptides. C4, high- 20 20 affinity peptide devoid of an N-core Peptide translocation (%) glycosylation site and labeled with 0 Peptide translocation (%) 0 fl 4 r 4 uorescein; E5, peptide not binding to -I e E5 IV er er C E5 C ST m TAP, including an N-core glycosylation NST 1m N 9m TAg 1 12mer TAg- 10 site and labeled with fluorescein; NST, reporter peptide, including an N-core glycosylation site and labeled with DSVVYDFLKC B RVSAINNYAQKL E (10mer) fluorescein. The experimental (12mer)

threshold (red dotted line) was set for VSAINNYAQKL min SVVYDFLKC min E5þATP (no TAP-binding). B and C, (11mer) 0 (9mer) 0 HPLC analysis of TAg-I-containing 30 30 N-terminally extended precursor SAINNYAQKL 60 60 (TAg-I) peptides trimmed by recombinant 120 120 mouse ERAP1 in vitro.3–6 ng rmERAP1 AINNYAQKL 240 240 were incubated with 50 mmol/L peptide at 37C and samples were 360 360 analyzed by HPLC after 0 to 360 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 minutes (label on the right). D, Retention time (min) Retention time (min) Transport assay for TAg-IV and related N-terminally extended precursor C F SVVYDFLKC peptides was performed as described VSAINNYAQKL (9mer) in A. E and F, HPLC analysis of TAg-IV- (11mer) min 0 containing N-terminally extended SAINNYAQKL min precursor peptides trimmed by (TAg-I) 30 0 recombinant mouse ERAP1 in vitro. The 30 VVYDFLKC 60 experiment was performed as (TAg-IV) described for B and C. A and D, Data 60 120 AINNYAQKL 120 are represented as mean SD, two- 240 way ANOVA with Bonferroni posttests 240 360 (, P < 0.01; and , P < 0.001). 360 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 12 13 14 15 16 17 18 19 20 21 22 23 24 Retention time (min) Retention time (min)

þ þ by the age of 3 to 4 months (Fig. 2A; refs. 17, 18). Erap1 / LoxP- (Fig. 2D). Upon stimulation with IFNg MHC-I expression Tag Alb-Cre double-transgenic offspring and Erap1 / increased in all HCC lines and was comparable for the WT þ LoxP-Tag Alb-Cre triple-transgenic offspring likewise and Erap1 / TAg HCC of each HCC pair. Those two sets of þ þ developed TAg HCC after about 3 to 4 months. Histological WT and Erap1 / TAg HCC lines with IFNg-inducible APM analysis of the primary HCC revealed a similar appearance of were used for further analysis. þ þ þ þ WT (Erap1 / ) TAg HCC and Erap1 / TAg HCC (Fig. 2B). þ WT and Erap1 / TAg HCC cell lines were subsequently ERAP1 regulates presentation of both IFNg-independent and established from primary HCC. Two cell lines of each WT IFNg-dependent tumor epitopes þ and Erap1 / TAg HCC were further characterized (referred Next, we examined the net effect of ERAP1 on the recognition of to as HCC pair 1 and HCC pair 2). Expression or lack of ex- the two MHC-I–restricted epitopes TAg-I and TAg-IV in depen- pression of ERAP1 was confirmed by Western blot (Fig. 2C). dence of IFNg. To do so, TCR gene-modified T cells recognizing Furthermore, IFNg-signaling was examined having a focus on either TAg-I (TCR-I T cells) or TAg-IV (TCR-IV T cells) were the expression of components of the antigen-processing generated by retroviral transduction of splenocytes from P14 x þ machinery (APM). WT and Erap1 / TAg HCC constitutively Rag / donor mice (transduction rates were between 70% and expressed JAK1 and downregulated JAK2 within 24 hours after 90%). When TCR-I T cells were cocultured with unstimulated þ þ stimulation with rmIFNg. A highly elevated expression of WT and Erap1 / TAg HCC lines, WT TAg HCC was recognized þ STAT1 and phosphorylation of STAT1 was noticed in WT and significantly better as compared with Erap1 / TAg HCC þ Erap1 / TAg HCC. Likewise, all HCC lines upregulated (Fig. 3A). In contrast, upon stimulation with rmIFNg before expression of APM components such as TAP1 and TAP2, and TCR-I T cell coculture, the effect of ERAP1 was negligible as the immunoproteasome subunits b1i, b5i, and b2i (Fig. 2C). shown by comparably good recognition of WT and Erap1 / þ Notably, WT HCC of pair 2 showed a higher basal expression of TAg HCC (Fig. 3B). These data reveal that TAg-I is IFNg- ERAP1, LMP7, and MECL1. This correlated with higher independent in the case of ERAP1-expression, but is IFNg- amounts of MHC-I in the absence of IFNg in this HCC line dependent in the absence of ERAP1.

www.aacrjournals.org Cancer Res; 78(12) June 15, 2018 3245

Schmidt et al.

A B H&E TAg Ki-67 X X

Erap1-/- LoxP-Tag Erap1-/- Alb-Cre WT WT

+/- +/- Erap1 x LoxP-Tag Erap1 x Alb-Cre -/- Erap1 Triple-transgenic mice developed -/- WT or Erap1 TAg+ HCC

C HCC pair 1 HCC pair 2 D HCC pair 1 -/- -/- WT Erap1-/- WT Erap1 MW WT Erap1 4 γ 104 10 IFN - + - + - + - + 171 137 100 kDa 10 3 12 TAg 103 10

2 ERAP1 102 10 100 kDa 1 101 10 JAK1 130 kDa 781 698 43 0 31 100 10 0 1 2 3 4 JAK2 130 kDa 100 101 102 103 104 10 10 10 10 10 100 kDa STAT1 HCC pair 2 -/- WT Erap1 4 100 kDa 104 10 pSTAT1 259 200 3 90 3 8 TAP1 10 10 70 kDa 2 102 10 TAP2 70 kDa 1 1 10 1653 10 1510 LMP2 (ß1i) 25 kDa b 342 0 19 100 10 0 1 2 3 4 0 1 2 3 4 25 kDa 10 10 10 10 10 10 10 10 10 10 LMP7 (ß5i) H2-D - IFNγ + isotype H2-Kb MECL1 (ß2i) + IFNγ + isotype 25 kDa - IFNγ + antibody β-Actin 40 kDa + IFNγ + antibody

Figure 2. / þ Primary WT and Erap1 TAg HCC with IFNg-inducible antigen-processing machinery was generated. A, Breeding strategy to obtain triple-transgenic Erap1þ/þ (WT) LoxP-Tag Alb-Cre and Erap1/ LoxP-Tag Alb-Cre mice. B, Histologic analysis of parafﬁn-embedded primary WT and Erap1/ TAgþ HCC of 3 to 4 months old LoxP-Tag Alb-Cre mice. One representative example is shown. C, Western blot analysis of IFNg-dependent expression of antigen processing machinery components in WT and Erap1/ TAgþ HCC. D, FACS analysis of MHC-I expression in WT and Erap1/ TAgþ HCC.

In contrast with TAg-I, TAg-IV was recognized by TCR-IV T cells plantation rejection antigen (16, 17). Therefore, the immunoge- þ þ on unstimulated WT TAg HCC, but not on unstimulated nicity of WT and Erap1 / TAg HCC was tested in immune- þ Erap1 / TAg HCC (Fig. 3C). IFNg stimulation before the competent mice. After subcutaneous transplantation into WT þ þ TCR-IV T-cell coculture increased the recognition of TAg-IV in (Erap1 / , C57BL/6) and Erap1 / mice (18), the recipients were the absence of ERAP1, but a significant difference was maintained observed for a period of 100 days. Of 3 to 4 mice injected per þ between WT and Erap1 / TAg HCC (Fig. 3D). Hence, optimal group, none developed a tumor, because those recipients are presentation and CTL recognition of TAg-IV critically required capable of priming a functional T-cell response toward TAg-I and joint action of IFNg and ERAP1. TAg-IV (Supplementary Fig. S2A–S2C). To establish full-grown þ tumors, WT and Erap1 / TAg HCC were grown in immune- Absence of ERAP1 accelerates tumor growth in immune- deficient Rag / mice (Fig. 4A; Supplementary S2A). Research by deficient recipients others proposes a role of natural killer (NK) cells in recognizing þ In LoxP-Tag Alb-Cre mice, WT and Erap1 / TAg HCC ERAP-deficient tumor cells (21, 22). Therefore, we compared þ primarily developed in a host that was tolerant for their trans- progression of WT versus Erap1 / TAg HCC in NK-cell–

3246 Cancer Res; 78(12) June 15, 2018 Cancer Research

ERAP1 Is Crucial for Peripheral T-cell Expansion

A -IFNγ/TCR-I C -IFNγ/TCR-IV HCC pair 1 HCC pair 2 HCC pair 1 HCC pair 2 25,000 25,000 20,000 20,000 ** WT *** WT WT *** WT 20,000 Erap1 -/- -/- -/- -/- 20,000 Erap1 15,000 Erap1 15,000 Erap1 ** 15,000 15,000 ** 10,000 10,000 10,000 10,000 IFN γ (pg/mL) IFN γ (pg/mL) ** IFN γ (pg/mL) 5,000 IFN γ (pg/mL) 5,000 5,000 5,000

0 0 0 0 1 10 100 1,000 1 10 100 1,000 1 10 100 1,000 1 10 100 1,000 E:T Ratio E:T Ratio E:T Ratio E:T Ratio

B +IFNγ/TCR-I D +IFNγ/TCR-IV HCC pair 1 HCC pair 2 HCC pair 1 HCC pair 2 25,000 25,000 20,000 20,000 WT WT WT WT -/- 20,000 -/- -/- *** -/- 20,000 Erap1 Erap1 15,000 Erap1 15,000 Erap1 15,000 15,000 ** 10,000 10,000 10,000 10,000 ** IFN γ (pg/mL) IFN γ (pg/mL) IFN γ (pg/mL) IFN γ (pg/mL) 5,000 5,000 ** 5,000 5,000

0 0 0 0 1 10 100 1,000 1 10 100 1,000 1 10 100 1,000 1 10 100 1,000 E:T Ratio E:T Ratio E:T Ratio E:T Ratio

Figure 3. ERAP1 regulates presentation of the subdominant epitope TAg-I and dominant epitope TAg-IV on TAg-driven primary HCC. A and B, Coculture of / þ TAg-I–speciﬁc TCR-I T cells with unstimulated (A)andIFNg-stimulated (B)WTandErap1 TAg HCC. C and D, Coculture of TAg-IV-speciﬁc TCR-IV / þ T cells with unstimulated (C)andIFNg-stimulated (D)WTandErap1 TAg HCC. A–D, Data are represented as mean SD, two-way ANOVA with Bonferroni posttests (, P < 0.01; , P < 0.001).

þ þ competent (Rag / ) recipients, progression of WT TAg HCC in efficiency, WT or Erap1 / TAg HCC were established in ERAP1- NK-cell–competent versus NK-cell–deficient (Rag / x gc / ) competent H2b recipients (Rag / , WT/H2b), ERAP1-competent þ recipients, and progression of Erap1 / TAg HCC in NK-cell– MHC-I–mismatched recipients (SCID, WT/H2d), and ERAP1- competent versus NK-cell–deficient recipients. No impact of NK deficient H2b recipients (Rag / , Erap1 / /H2b; Fig. 4A). At first, þ cells on tumor growth of TAg HCC could be seen in the majority we investigated whether ERAP1 specifically affected rejection of þ of mice (Supplementary Fig. S3A–S3C). One mouse showed TAg HCC through TCR-I T cells targeting TAg-I. The highest þ þ exceptional tumor outgrowth of Erap1 / TAg HCC starting rejection rate was observed, when WT TAg HCC were treated in þ 50 days after tumor cell inoculation. But NK-cell contribution at WT/H2b recipients with 100% and 88% of WT TAg of HCC pair 1 this late stage is unlikely. We concluded that NK-cell cytotoxicity is and 2 being rejected, respectively (Fig. 4B; Supplementary Table þ not relevant in the herein used model of transplanted TAg HCC. S1A). Notably, TCR-I gene-modified T cells were not capable of þ Interestingly, when the tumor size was compared in depen- rejecting WT TAg HCC in WT/H2d recipients, where the target þ dence of recipient's ERAP1 at the day of ATT, WT TAg HCC had a antigen cannot be cross-presented via MHC-I. Reduced rejection þ mean volume of 92 mm3 (WT!WT SD 63.5) in Rag / rates of 67% and 57% were also observed, when WT TAg HCC recipients while having a mean volume of 118 mm3 was treated with TCR-I T cells in Erap1 / /H2b recipients. In þ (WT!Erap1 / SD 57.9) in Erap1 / x Rag / recipients. contrast with WT TAg HCC, a reduced number of 70% and þ þ Similarly, Erap1 / TAg HCC were on average 143 mm3 in 80% Erap1 / TAg HCC was rejected in WT/H2b recipients, Rag / recipients (Erap1 / !WT SD 62.3) as compared with indicating some contribution of ERAP1 on the direct presentation þ þ Erap1 / TAg HCC having a mean volume of 233 mm3 of TAg-I in vivo. Although Erap1 / TAg HCC was not rejected in (Erap1 / !Erap1 / SD 123.3; Supplementary Fig. S3D). WT/H2d recipients, HCC pair 1 and 2 showed a huge difference In conclusion, lack of ERAP1 the noncancerous cells of with 33% and 100% being rejected, respectively, in the absence of immune-deficient recipients accelerated tumor growth after host ERAP1 (Fig. 4B; Supplementary Table S1A). Hence, although transplantation. expression of MHC-I and ERAP1 in ATT recipient's cells seemed to be required for rejection through TCR-I T cells, ERAP1-dependent þ ERAP1 facilitates efficient rejection of TAg HCC through TCR-I direct presentation of TAg-I exerted a moderate effect. T cells þ The use of TCR gene-modified T cells is a particularly effective Rejection of TAg HCC through TCR-IV T cells critically approach to target malignancies (23). To compare the impact of depends on ERAP1 expression in noncancerous cells ERAP1-dependent direct presentation, MHC-I–dependent cross- Adoptive transfers targeting TAg-IV through TCR-IV T cells were presentation, and ERAP1-dependent cross-presentation on ATT conducted in parallel. Here, in the group of WT/H2b recipients,

www.aacrjournals.org Cancer Res; 78(12) June 15, 2018 3247

Schmidt et al.

A T-cell transfer

Tumor Monitoring of transplantation Tumor growth tumor rejection

-/- WT or Erap1 ~50–100 days ~30–100 days TAg+ HCC Immune-deficient recipients

B C D

TCR-I TCR-IV Tumor Recipient's Erap1 Erap1 Recipient‘s Recipient‘s Tumor Recipient‘s haplo- HCC HCC Tumor Recipient‘s haplo- HCC HCC WT WT 93.8 Erap1 Erap1 type pair 1 pair 2 Erap1 Erap1 type pair 1 pair 2 WT -/- 61.9 WT WT H2b 100 87.5 WT WT H2b 100 100 -/- WT 75.0 d d WT WT H2 0.0 n.a. WT WT H2 0.0 n.a. -/- -/- 66.7

WT -/- H2b 66.7 57.1 WT -/- H2b 33.3 14.3 WT WT 100 -/- WT H2b 70.0 80.0 -/- WT H2b 70.0 83.3 * WT -/- 23.8 -/- WT H2d 0.0 n.a. -/- WT H2d 0.0 n.a. -/- WT 76.7 -/- -/- H2b 33.3 100 -/- -/- H2b 50.0 40.0 -/- -/- 45.0

0 20 40 60 80 100 0% Rejection 100% 0% Rejection 100% Rejection (%)

Figure 4. T-cell–mediated rejection of TAgþ HCC requires antigen cross-presentation and ERAP1. A, Schematic for tumor transplantation and rejection experiments. B, Graphical representation of rejection of WT and Erap1/ TAgþ HCC by TCR-I T cells in immune-deﬁcient recipients. For numerical summary, see Supplementary Table S1A. C, Graphical representation of rejection of WT and Erap1/ TAgþ HCC by TCR-IV T cells in immune-deﬁcient recipients. For numerical summary, see Supplementary Table S1B. B and C, Numbers represent the percentage of rejection. D, Summary of the percentage of rejection of HCC pair 1 and 2 in the H2b recipient groups is depicted in B and C. Numbers are the percentage of rejection. Data are represented as mean of HCC pair 1 and 2 SD, two-way ANOVA with Bonferroni posttests (, P < 0.05).

þ complete rejection was observed for WT TAg HCC (Fig. 4C; variations between the groups affected the efficiency of TCR-I þ Supplementary Table S1B). WT TAg HCC was not rejected if or TCR-IV T-cell therapy. At first, 57 mice treated with TCR-I antigen cross-presentation was lacking in WT/H2d mice and T cells were analyzed (Supplementary Fig. S3E). Mice were þ strikingly, WT TAg HCC rejection rates were also strongly grouped according to whether tumors were rejected or not reduced (33% and 14%) if ERAP1 was not expressed in the andthetumorsizeatthedayoftreatmentwasplottedfor recipient's cells. In addition, ERAP1 affected direct presentation both groups. In the case of TCR-I, both rejected and nonrejected of TAg-IV, as indicated by inferior rejection of 70% and 83% tumors had an average volume of 132 mm3 and 135 mm3, þ Erap1 / TAg HCC in WT/H2b recipients. The requirement of respectively (rejected SD 92.2; not rejected SD 95.4; antigen cross-presentation of TAg-IV was confirmed by lack of Fig. 5A). The same analysis was performed for 57 mice treated þ rejection of Erap1 / TAg HCC in WT/H2d mice, and by less than with TCR-IV T cells (Supplementary Fig. S3F). In this case, þ half (45%) of the Erap1 / TAg HCC being rejected in Erap1 / / rejected tumors had a mean volume of 135 mm3 on the day of H2b recipients (Fig. 4C; Supplementary Table S1B). In summary, ATT, whereas nonrejected tumors were on average 171 mm3 ERAP1-depedent direct presentation moderately affected tumor (rejected SD 83.9; not rejected SD 104.2; Fig. 5B). In rejection, whereas both, MHC-I–dependent antigen cross-presen- summary, the tumor size at the day of ATT did not affect tation and ERAP1 expression in noncancerous cells were critically rejection through TCR-I T cells or TCR-IV T cells. þ required for rejection of TAg HCC through TCR-IV gene-mod- In addition, T lymphocyte function was analyzed by in vivo ified cells (Fig. 4D). cytotoxicity analysis at the end of the experiments. Naive control mice did not respond to TAg-I, that is, they did not kill TAg-I þ Absence of ERAP1 enables escape of small TAg HCC despite (SAINNYAQKL)-loaded target cells, whereas the same peptide- the presence of functional CTL loaded target cells were efficiently killed by TAg-immunized mice Next, tumor volumes at the day of ATT were correlated with or TCR-I T-cell–treated mice (Fig. 5C). TAg (16.113)-immunized subsequent tumor rejection to rule out that the above described controls showed a mean of 75% specific cytotoxicity toward TAg-I

3248 Cancer Res; 78(12) June 15, 2018 Cancer Research

ERAP1 Is Crucial for Peripheral T-cell Expansion

A BC 600 600 Naïve control Immunized control 1K ) ) 48.4 3 ns 3 ns 800 51.6 68.7 31.3 400 400 600 1.3

400 Unloaded

SSC-H Unloaded Counts Counts TAg-I TAg-I 200 200 200 0 0 1 2 3 4 0 1 2 3 4 0 1 2 3 4 Tumor volume (mm Tumor volume (mm 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 CFSE CFSE CFSE 0 0 Not Not Rejected Not rejected Rejected rejected Rejected rejected 97.8 90.1 2.2 9.9 D TAg-I E TAg-IV Counts Counts P = 0.0004 P = 0.0002 100 100 Unloaded Unloaded TAg-I TAg-I 80 80 100 101 102 103 104 100 101 102 103 104 60 60 CFSE CFSE

40 40

20 20 Specific kill in vivo (%) Specific kill in vivo (%) 0 0 Naïve Immunized Not Naïve Immunized Not control control Rejected rejected control control Rejected rejected

Figure 5. Small TAgþ HCC recur in the presence of epitope-specific functional cytotoxic T lymphocytes. A, All mice depicted in Supplementary Fig. S3E were grouped according to whether TAgþ HCC was rejected or not and the tumor volume at the day of ATT was plotted, rejected (n ¼ 43), not rejected (n ¼ 14). B, All mice depicted in Supplementary Fig. S3F were grouped according to whether TAgþ HCC was rejected or not and the tumor volume at the day of ATT was plotted, rejected (n ¼ 37), not rejected (n ¼ 20). A and B, All data are represented with mean SD, Mann–Whitney test; , P < 0.05. C and D, In vivo cytotoxicity analysis for TAg-I. To detect CTL activity in vivo, TAg-I and/or TAg-IV–loaded spleen cells were labeled with carboxyfluorescein diacetate succinimidyl ester (CFSE) and were injected into the indicated mice. The ratio between different CFSE-labeled populations was determined and 18 hours later by flow cytometry. One representative example for TAg-I is shown in C and data of three experiments are shown in D.Na€ve control, n ¼ 6; immunized control, n ¼ 6; rejected, n ¼ 24; not rejected (n ¼ 6). E, In vivo cytotoxicity analysis for TAg-IV was performed as described in C.Dataof three experiments are shown. Na€ve control, n ¼ 8; immunized control, n ¼ 8; rejected, n ¼ 22; not rejected, n ¼ 8. D and E, All data are represented with mean SD, Kruskal–Wallis test.

(SD 18.0). For a determination of possible differences in the is required for efficient antitumor immunity (24, 25). Hence, the þ functionality of persistent CTLs after tumor rejection or recur- expansion of CD8 T cells (within white blood cells) was ana- rence, mice treated with TCR-I T cells were grouped on whether the lyzed and compared between the different recipient groups (Sup- þ þ TAg HCC were rejected or not. But while mice rejecting through plementary Fig. S4A–S4C). In the case of TCR-I T cells, CD8 cells TCR-I showed 87% cytotoxicity in vivo (SD 21.3), a mean of constituted on average 0.5% (SD 0.75) in WT/H2b recipients þ 81% TAg-I–specific cytotoxicity was detectable in mice with bearing WT TAg HCC one week after ATT. Similarly, in WT/H2b þ nonrejected tumors as well (SD 30.2). These data confirmed recipients bearing Erap1 / TAg HCC approximately 0.6% (SD þ that the functionality of TCR-I T cells was maintained throughout 0.89) were CD8 T cells (Fig. 6A). In contrast, TCR-I T cells did not the experiment, and failure of rejection was not caused by a loss of expand in WT/H2d recipients and were barely detectable in function of the transferred TCR-I T cells (Fig. 5D). Similarly, upon Erap1 / /H2b recipients regardless of whether they were bearing þ þ analysis of adoptively transferred TCR-IV T cells, naive control WT TAg HCC or Erap1 / TAg HCC (Fig. 6A). At this early time mice did not kill TAg-IV (VVYDFLKL)–loaded target cells, whereas point a considerable variation was observed between mice, and TAg (16.113)-immunized control mice presented on average MHC-I–independent proliferation of TCR-I T cells was noticed in þ 94% (SD 5.8) TAg-IV–specific cytotoxicity. In the sample the WT/H2d group bearing Erap1 / TAg HCC. Control experi- groups, on average 89% (SD 20.7) TAg-IV–specific cytotoxicity ments revealed that in vitro primed TCR-I gene–modified T cells þ was detectable in those mice in which TAg HCC were rejected, did not expand in tumor-free recipients in the absence of TAg, þ whereas mice with nonrejected TAg HCC showed a mean of whereas in vivo primed TAg-I–specificTE cells were capable of 88% (SD 21.6) TAg-IV–specific cytotoxicity (Fig. 5E). In sum- expanding antigen-independently (Supplementary Fig. S5A and þ mary, rejection of TAg HCC failed despite the presence of in vivo S5B). Four weeks after ATT, a sustained expansion was observed in þ cytotoxic TCR-IV T cells. all WT/H2b recipients, where CD8 T cells now constituted a mean of 9% (SD 6.57) and 7% (SD 3.90) in mice bearing WT þ þ Expansion of TCR gene-modified T cells requires ERAP1 TAg HCC or Erap1 / TAg HCC, respectively (Fig. 6B). Still, þ expression in ATT recipients CD8 T cells were not detectable in WT/H2d mice and were very For ATT, therapeutic T cells are transferred into lympho-deplet- low in Erap1 / /H2b recipients reaching a mean of 1% (SD ed individuals, because enhanced homeostatic T-cell proliferation 0.83) and 3% (SD 3.51; Fig. 6B). In summary, expansion of

www.aacrjournals.org Cancer Res; 78(12) June 15, 2018 3249

Schmidt et al.

ABTCR-I/1 week TCR-I/4 weeks 4 25 P < 0.01 P < 0.01 20 3 15 2 T cells (%) T cells (%) + + 10 CD8 1 CD8 5

0 0 Tumor Erap1 WT-/- WT -/- WT -/- Tumor Erap1 WT-/- WT -/- WT -/- Recipient's Erap1 WT WT WT WT -/- -/- Recipient's Erap1 WT WT WT WT -/- -/- Recipient's haplotype H2b H2b H2d H2d H2b H2b Recipient's haplotype H2b H2b H2d H2d H2b H2b

C TCR-IV/1 week D TCR-IV/4 weeks 8 25 P < 0.001 P < 0.001 20 6 15 4 T cells (%) T cells (%) + + 10 CD8 CD8 2 5

E TCR-I/1 week F TCR-I/4 weeks G TCR-IV/1 week H TCR-IV/4 weeks 4 25 8 25 * ns * * 20 20 3 6 15 15 2 4 T cells (%) T cells (%) T cells (%) T cells (%)

+ + 10 + + 10 CD8 CD8 CD8 CD8 1 2 5 5

0 0 0 0 Not Not Not Not Rejected rejected Rejected rejected Rejected rejected Rejected rejected

Figure 6. ERAP1 expression in noncancerous cells supports immediate expansion of epitope-speciﬁc TCR-I and TCR-IV T cells and decides on tumor rejection. A, FACS analysis of TCR-I T-cell expansion one week after ATT. WT!WT/H2b, n ¼ 16; WT!WT/H2d, n ¼ 3; WT!Erap1//H2b, n ¼ 11; Erap1/!WT/H2b, n ¼ 7; Erap1/!WT/H2d, n ¼ 4; Erap1/!Erap1//H2b, n ¼ 5. B, FACS analysis of TCR-I T cell expansion four weeks after ATT. WT!WT/H2b, n ¼ 6; WT!WT/H2d, n ¼ 2; WT!Erap1//H2b, n ¼ 3; Erap1/!WT/H2b, n ¼ 9; Erap1/!WT/H2d, n ¼ 4; Erap1/!Erap1//H2b, n ¼ 7. C, FACS analysis of TCR-IV T cells performed one week after ATT. WT!WT/H2b, n ¼ 16; WT!WT/H2d, n ¼ 4; WT!Erap1//H2b, n ¼ 11; Erap1/!WT/H2b, n ¼ 12; Erap1/!WT/H2d, n ¼ 3; Erap1/!Erap1//H2b, n ¼ 6. D, FACS analysis of TCR-IV T cells four weeks after ATT. WT!WT/H2b, n ¼ 6; WT!WT/H2d, n ¼ 3; WT!Erap1//H2b, n ¼ 4; Erap1/!WT/H2b, n ¼ 10; Erap1/!WT/H2d, n ¼ 3; Erap1/!Erap1//H2b, n ¼ 6. A–D, Data of n ¼ 3 experiments are shown with mean SD, Kruskal–Wallis test. E, All mice with haplotype H2b depicted in A were grouped according to whether TAgþ HCC was rejected or not and the percentage of CD8þ T cells was plotted, rejected (n ¼ 28), not rejected (n ¼ 11). F, All mice with haplotype H2b depicted in B were grouped according to whether TAgþ HCC was rejected or not and the percentage of CD8þ T cells was plotted, rejected (n ¼ 21), not rejected (n ¼ 4). G, All mice with haplotype H2b depicted in C were grouped according to whether TAgþ HCC was rejected or not and the percentage of CD8þ T cells was plotted, rejected (n ¼ 30), not rejected (n ¼ 15). H, All mice with haplotype H2b depicted in D were grouped according to whether TAgþ HCC was rejected or not and the percentage of CD8þ T cells was plotted, rejected (n ¼ 19), not rejected (n ¼ 7). E–H, Data are represented with mean SD, Mann–Whitney test (, P < 0.05).

þ adoptively transferred gene-modiﬁed TCR-I T cells required com- (SD 1.20) CD8 T cells were detectable in WT/H2b recipients þ patible MHC-I, and expression of ERAP1 in ATT recipients, bearing WT or Erap1 / TAg HCC (Fig. 6C). TCR-IV T cells whereas lack of ERAP1 in cancerous cells had not impact hereon. did not expand in WT/H2d mice and represented less than Expansion of TCR-IV T cells was analyzed at the same time. 0.3% in Erap1 / /H2b recipients (Fig. 6C). Similar to TCR-I T Here, too, a strong variation was observed between individual cells, TCR-IV T cells expanded within the following 3 weeks. mice one week after ATT. On average, 0.8% (SD 1.64) to 0.9% Accordingly, 4 weeks after ATT a mean of 5% (SD 1.57) and

3250 Cancer Res; 78(12) June 15, 2018 Cancer Research

ERAP1 Is Crucial for Peripheral T-cell Expansion

þ þ 8% (SD 8.11) CD8 T cells were detected in WT/H2b recipients of ERAP1-regulated epitope presentation. Six nonrejected TAg þ bearing WT or Erap1 / TAg HCC, respectively (Fig. 6D). In HCC were grown in vitro for further analysis. Among these, two WT þ þ WT/H2d recipients or Erap1 / /H2b recipients, CD8 T cells still TAg HCC were reisolated after TCR-I T-cell therapy in Erap1 / þ constituted no more than 0.01% to 0.8% (Fig. 6D). Although recipients (#377, #378), one WT TAg HCC was reisolated after TAg-I–specific T cells may also expand antigen independently, TCR-IV T-cell therapy in an Erap1 / recipient (#400), one þ TAg-IV–specific T cells expanded exclusively in the presence Erap1 / TAg HCC was reisolated after TCR-IV T-cell therapy þ of TAg (Supplementary Fig. S5C). Interestingly, if polyclonal in an Erap1 / recipient (#556), and two Erap1 / TAg HCC þ TAg-specific CD8 effector T (TE) cells [containing a mean of were reisolated after TCR-IV T cell therapy in ERAP1-competent 1.6% (SD 0.93, n ¼ 3) TAg-I–specificTE cells and 8.4% (SD recipients (#306, #282, Fig. 7A). First, we confirmed expression of 7.89, n ¼ 3) TAg-IV–specificTE cells] were isolated from the transplantation rejection antigen TAg by the nonrejected HCC þ TAg-immunized C57BL/6 mice, and were transferred into TAg (Fig. 7B). After stimulation with rmIFNg, JAK1 was constitutively þ HCC-bearing recipients, ERAP1 assumed a more prominent expressed in all reisolated TAg HCC lines in different amounts, role on tumor rejection as related to direct epitope presentation. but upregulation of STAT1, as well as phosphorylation thereof, However, overall expansion of the adoptively transferred was always detectable (Fig. 7B). In addition, TAP1 and TAP2 were þ þ polyclonal CD8 TE cells was not affected in ERAP1-deficient highly expressed in all six TAg HCC after IFNg stimulation, just as recipients, indicating that only antigen-specific T cells required the immunoproteasome subunits b1i, b5i, and b2i (Fig. 7B). recipient's ERAP1 to proliferate (Supplementary Fig. S6A–S6D). When MHC-I expression was analyzed, all reisolated tumors In conclusion, the recipients' ERAP1 status, but not the ERAP1 showed expression of both alleles, H2-Db and H2-Kb, albeit with þ status of the targeted TAg HCC was crucial for expansion of varying degree (Fig. 7C). epitope-specific TCR gene-modified T cells. Finally, in vitro recognition of the reisolated tumor cells by þ TCR-I and TCR-IV T cells was analyzed. All six TAg HCCs were þ Immediate expansion of antigen-specific T cells decides on recognized by TCR-I T cells (Fig. 7D). Likewise, all six TAg HCC tumor rejection were well recognized by TCR-IV T cells, with the exception þ Impaired expansion of antigen-specific T cells after ATT was a of #556, the Erap1 / TAg HCC that has progressed after remarkable symptom of ERAP1-deficient ATT recipients, in TCR-IV T-cell therapy in an Erap1 / recipient. This HCC line þ which TAg HCC was poorly rejected. To examine whether showed an IFNg-inducible, albeit very low, CTL response that was þ expansion of TCR-I T cells correlated with rejection of TAg in accordance with the rather low expression of MHC-I by this þ HCC, all H2b recipient mice that were analyzed according to HCC line in comparison with the other nonrejected TAg HCC þ their percentage of TCR-I CD8 T cells in the blood one week (Fig. 7E). If antigen cross-presentation was assessed in bone þ after ATT were grouped on whether mice had rejected the TAg marrow–derived dendritic cells (BMDC), a slightly impaired HCC or not. At this time point mice with later on rejected cross-presentation of TAg-I and TAg-IV as processed from full- þ tumors presented a mean of 0.5% (SD 0.73) CD8 Tcellsas length purified TAg protein was observed with Erap1 / BMDCs in compared with a significantly lower mean of 0.1% (SD 0.19) comparison with WT BMDCs (Supplementary Fig. S7A and S7B). þ CD8 T lymphocytes in mice that subsequently did not reject We concluded that the sole lack of ERAP1 in tumor cells or in the tumors (Fig. 6E). When the H2b TCR-I T-cell–treated mice were recipient's non-tumor cells determined the failure of ATT, even þ grouped according to the same parameters, CD8 T cells though further experimentation is required to elucidate the actual constituted on average 6% (SD 4.92) in mice with later on antigen cross-presentation pathway involving ERAP1. þ rejected tumors versus a mean of 4% (SD 6.60) CD8 T cells in those mice where tumors were not rejected (Fig. 6F). Hence, immediate expansion of TCR-I T cells one week after ATT Discussion correlated with subsequent tumor rejection. We showed that ERAP1 is critically required for the processing In those mice adoptively transferred with TCR-IV T cells, of MHC-I epitopes as targeted by ATT and critically affects suc- þ approximately 0.8% (SD 1.47) CD8 T lymphocytes were cessful eradication of cancer. In this study, rejection of transplant- þ detectable one week after ATT in H2b mice that rejected TAg able tumors was decreased by 25% and 23% for TCR-I and HCC later on. Opposite to that, a significantly lower percentage of TCR-IV, respectively, if cancer cells were deficient for ERAP1 but þ approximately 0.1% (SD 0.14) CD8 T cells was detected in expressed other IFNg-inducible APM components such as immu- þ mice with subsequently nonrejected TAg HCC (Fig. 6G). That noproteasomes and TAP. The previously reported inference of statistical connection was still observable 4 weeks after ATT, and TAg-I being ERAP1-independently processed was limited to direct þ on average 6% (SD 6.56) CD8 T cells were present in the blood antigen presentation in the presence of IFNg and with cross- of the mice with ultimately rejected tumors. In contrast, those presented TAg (12). The herein presented data imply that no mice that did not reject tumors had a mean of 0.8% (SD 0.79) other APM components are capable of completely substituting þ of CD8 T cells (Fig. 6H). These data confirm that both, ERAP1 the specialized function of ERAP1. function in ATT recipients and T-cell expansion and persistence, Previously, it was shown by others that NK cells can reject are required for successful tumor rejection after ATT. suspensions of ERAP1-silenced RMA lymphoma cells due to poor engagement of the Ly49C/I NK-cell–inhibitory receptor by altered Lack of ERAP1 enables escape of T-cell–recognizable IFNg- pMHC-I in the absence of N-terminal trimming of peptides (21). þ responsive TAg HCC And pharmacological inhibition of ERAP1 in human tumor cells Relating to earlier observations in Textor and colleagues (12) of induced an NK-cell response caused by the tumor cell's inability to IFNg–unresponsive cancer variants to evade T-cell recognition, engage inhibitory NK-cell receptors (22). Adoptive T-cell therapy we wanted to rule out that an acquired deficiency in IFNg- through transfer of TCR gene-modified T cells, however, is an þ signaling contributed to TAg HCC recurrence within the context approach to target solid tumors, that escaped the patient's

www.aacrjournals.org Cancer Res; 78(12) June 15, 2018 3251

Schmidt et al.

A C -/- Label #377 #378 #400 #556 #306 #282 WT Erap1 4 4 10 162 10 210 Therapy TCR-I TCR-I TCR-IV TCR-IV TCR-IV TCR-IV 34 18 103 103 Tumor Erap1 WT WT WT -/- -/- -/- 102 102 Recipient’s Erap1 -/- -/- -/- -/- WT WT 101 101 215 415 40 22 100 100 0 1 2 3 4 0 1 2 3 4 B 10 10 10 10 10 10 10 10 10 10 #377 #556 + 104 104 WT Recurred TAg HCC 409 68 41 11 (HCC 103 103 Label pair 2) #377 #378 #400 #556 #306 #282 MW 102 102 IFNγ - + - + - + - + - + - + - + 100 kDa 101 101 TAg 931 95 52 11 100 100 0 1 2 3 4 0 1 2 3 4 ERAP1 10 10 10 10 10 10 10 10 10 10 100 kDa #378 #306 JAK1 130 kDa 4 4 10 101 10 157 100 kDa 7 47 Stat1 103 103

pStat1 100 kDa 102 102 Short exposure 101 101 pStat1 100 kDa 186 189 9 37 Long exposure 100 100 100 101 102 103 104 100 101 102 103 104 TAP1 70 kDa #400 #282 4 4 TAP2 70 kDa 10 299 10 506 36 590 103 103 LMP2 (β1i) 25 kDa 102 102 LMP7 (β5i) 25 kDa 101 101 531 552

b 38 559 100 100 MECL1 (β2i) 100 101 102 103 104 100 101 102 103 104

25 kDa H2-D - IFNγ + isotype - IFNγ + antibody β b -Actin 40 kDa H2-K + IFNγ + isotype + IFNγ + antibody

D TCR-I E TCR-IV 75,000 75,000 -IFNγ +IFNγ -IFNγ +IFNγ

50,000 50,000

25,000

25,000 IFN γ (pg/mL) IFN γ (pg/mL) 700

0 -/- 0 -/- none WT #377 #378 #400 Erap1 #556 #306 #282 none WT #377 #378 #400 Erap1 #556 #306 #282 Target cells Target cells

Figure 7. þ / þ Lack of ERAP1 enables recurrence of TAg , TCR-recognizable HCC with IFNg-inducible antigen-processing machinery. A, Overview of WT or Erap1 TAg HCC / þ reisolated after TCR-I or TCR-IV therapy in WT/H2b or Erap1 /H2b recipients. B, Western blot analysis of IFNg-inducible APM components in reisolated TAg þ þ HCC. C, FACS analysis of IFNg-inducible MHC-I expression for reisolated TAg HCC. D, TCR-I T-cell recognition of reisolated TAg HCC. E, TCR-IV T-cell recognition of reisolated TAgþ HCC. D and E, Data are represented as mean SD.

spontaneous immune recognition and control experiments in this tion may induce cell surface presentation of protective tumor study did not conﬁrm an early role of NK cells, if Erap1 was antigens inducing functional T-cell responses (11). Those ﬁndings genetically depleted and surface expression of MHC-I was adjust- show that enhancing ERAP1 function is probably not an alterna- able by IFNg. Here, the therapeutic avenue of pharmacologically tive therapeutic avenue either, because ERAP1 has the potential to inhibiting ERAP1 is certainly a strategy that contrasts adoptive T- over-trim peptides and to destroy MHC-I epitopes. Our data are of cell transfer, if the latter one targets ERAP1-dependent MHC-I particular importance with regard to the various common SNPs in epitopes. Another study showed that attenuation of ERAP1 func- the human Erap1 homolog and the resulting naturally existing

3252 Cancer Res; 78(12) June 15, 2018 Cancer Research

ERAP1 Is Crucial for Peripheral T-cell Expansion

þ alleles of Erap1 encoding ERAP1 variants with different, some- proliferation of CD8 T cells after ATT, and that lack of ERAP1 times impaired, function (26–28). In this study, we used a mouse in ATT recipients consequently caused failure of adoptive T-cell model of genetically depleted Erap1, causing complete loss of therapy. function of the enzyme, and transferability of such findings to functionally impaired ERAP1 variants in humans remains to be Disclosure of Potential Conflicts of Interest seen. Nevertheless, personalized cancer immunotherapy of the No potential conflicts of interest were disclosed. future may consider processing of the targeted epitope as well Authors' Contributions as patient's Erap1 haplotype as an additional marker. Further Conception and design: K. Schmidt, U. Seifert, T. Blankenstein, G. Willimsky, investigations, including MHC-I epitopes presented by clini- P.-M. Kloetzel cally relevant tumor samples, would also show whether our Development of methodology: K. Schmidt, G. Willimsky findings, as obtained from a transplantable tumor model, can Acquisition of data (provided animals, acquired and managed patients, be translated to improve adoptive T-cell therapy approaches in provided facilities, etc.): K. Schmidt, A. Textor, G. Willimsky patients. Analysis and interpretation of data (e.g., statistical analysis, biostatistics, For the first time, we show that ERAP1 is certainly playing a role computational analysis): K. Schmidt, A. Textor Writing, review, and/or revision of the manuscript: K. Schmidt, A.A. Kuhl,€ in antigen cross-presentation of tumor epitopes as targeted by ATT A. Textor, T. Blankenstein, G. Willimsky, P.-M. Kloetzel and thereby supports rejection of established tumors. Previous Administrative, technical, or material support (i.e., reporting or organizing work by others reported that ERAP1 is required for cross-presen- data, constructing databases): C. Keller, A.A. Kuhl€ tation of the model antigen OVA (13–15). But if tumor antigens Study supervision: K. Schmidt, G. Willimsky, P.-M. Kloetzel are cross-presented in vivo, cell-associated antigens from dying Acknowledgments tumor cells may be the primary source of antigen. So far, only one The authors would like to thank F. Kirschner, A. Lehmann, and N. Albrecht- other study has examined ERAP1-dependent cross-presentation Kopke€ from the Institute of Biochemistry, Charite—Universit€atsmedizin Berlin of cell-associated antigens. The cross-presentation of cell-associ- for assistance with data acquisition. U. Seifert, T. Blankenstein, and G. Willimsky ated OVA and cell-associated male HY antigen was analyzed after received grants from the Deutsche Forschungsgemeinschaft (SFB-TR36). immunization of ERAP1-deficient mice with MHC-I–mismatched T. Blankenstein and P.-M. Kloetzel received grants from the Berlin Institute antigen-positive cells (15). Here, ERAP1-deficiency reduced pro- of Health (CRG-1). P.-M. Kloetzel received funding from the Wilhelm Sander- þ liferation of OVA-specific or HY antigen-specific CD8 T cells, Stiftung (2015.107.1). respectively. Our in vivo investigations with the tumor rejection antigen TAg confirm these earlier observations, but analysis of The costs of publication of this article were defrayed in part by the payment ERAP1-dependent cross-presentation of TAg-I and TAg-IV in of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. BMDCs in vitro requires improved experimental procedures. Importantly, our experiments showed that ERAP1 has a decisive Received June 29, 2017; revised February 13, 2018; accepted March 16, 2018; effect on tumor rejection by being critically required for the published first March 20, 2018.

References 1. Rosenberg SA, Yang JC, Sherry RM, Kammula US, Hughes MS, Phan GQ, 10. Serwold T, Gaw S, Shastri N. ER aminopeptidases generate a unique pool of et al. Durable complete responses in heavily pretreated patients with peptides for MHC class I molecules. Nat Immunol 2001;2:644–51. metastatic melanoma using T-cell transfer immunotherapy. Clin Cancer 11. James E, Bailey I, Sugiyarto G, Elliott T. Induction of protective antitumor Res 2011;17:4550–7. immunity through attenuation of ERAAP function. J Immunol 2013;190: 2. Spiotto MT, Yu P, Rowley DA, Nishimura MI, Meredith SC, Gajewski TF, 5839–46. et al. Increasing tumor antigen expression overcomes "ignorance" to solid 12. Textor A, Schmidt K, Kloetzel PM, Weißbrich B, Perez C, Charo J, et al. tumors via crosspresentation by bone marrow-derived stromal cells. Preventing tumor escape by targeting a post-proteasomal trimming inde- Immunity 2002;17:737–47. pendent epitope. J Exp Med 2016;213:2333–2348. 3. Spiotto MT, Rowley DA, Schreiber H. Bystander elimination of antigen loss 13. Saveanu L, Carroll O, Weimershaus M, Guermonprez P, Firat E, Lindo V, variants in established tumors. Nat Med 2004;10:294–8. et al. IRAP identifies an endosomal compartment required for MHC class I 4. Leisegang M, Engels B, Schreiber K, Yew PY, Kiyotani K, Idel C, et al. cross-presentation. Science 2009;325:213–7. Eradication of large solid tumors by gene therapy with a T-cell receptor 14. Yan J, Parekh VV, Mendez-Fernandez Y, Olivares-Villagomez D, Dragovic S, targeting a single cancer-specific point mutation. Clin Cancer Res 2016; Hill T, et al. In vivo role of ER-associated peptidase activity in tailoring 22:2734–43. peptides for presentation by MHC class Ia and class Ib molecules. J Exp Med 5. Engels B, Engelhard VH, Sidney J, Sette A, Binder DC, Liu RB, et al. Relapse 2006;203:647–59. or eradication of cancer is predicted by peptide-major histocompatibility 15. Firat E, Saveanu L, Aichele P, Staeheli P, Huai J, Gaedicke S, et al. The complex affinity. Cancer Cell 2013;23:516–26. role of endoplasmic reticulum-associated aminopeptidase 1 in immu- 6. Robbins PF, Dudley ME, Wunderlich J, El-Gamil M, Li YF, Zhou J, et al. nity to infection and in cross-presentation. J Immunol 2007;178: Cutting edge: persistence of transferred lymphocyte clonotypes correlates 2241–8. with cancer regression in patients receiving cell transfer therapy. J Immunol 16. Willimsky G, Czeh M, Loddenkemper C, Gellermann J, Schmidt K, Wust P, 2004;173:7125–30. et al. Immunogenicity of premalignant lesions is the primary cause of 7. Kloetzel PM. Antigen processing by the proteasome. Nat Rev Mol Cell Biol general cytotoxic T lymphocyte unresponsiveness. J Exp Med 2008;205: 2001;2:179–87. 1687–700. 8. Neefjes JJ, Momburg F, Hammerling GJ. Selective and ATP-dependent 17. Willimsky G, Blankenstein T. Sporadic immunogenic tumours avoid translocation of peptides by the MHC-encoded transporter. Science 1993; destruction by inducing T-cell tolerance. Nature 2005;437:141–6. 261:769–71. 18. York IA, Brehm MA, Zendzian S, Towne CF, Rock KL. Endoplasmic 9. Fruci D, Niedermann G, Butler RH, van Endert PM. Efficient MHC class I- reticulum aminopeptidase 1 (ERAP1) trims MHC class I-presented pep- independent amino-terminal trimming of epitope precursor peptides in tides in vivo and plays an important role in immunodominance. Proc Natl the endoplasmic reticulum. Immunity 2001;15:467–76. Acad Sci U S A 2006;103:9202–7.

www.aacrjournals.org Cancer Res; 78(12) June 15, 2018 3253

Schmidt et al.

19. Mylin LM, Schell TD, Roberts D, Epler M, Boesteanu A, Collins EJ, et al. more effectively target malignancies. Clin Cancer Res 2015;21: Quantitation of CD8(þ) T-lymphocyte responses to multiple epitopes 5191–7. from simian virus 40 (SV40) large T antigen in C57BL/6 mice immunized 24. Dummer W, Niethammer AG, Baccala R, Lawson BR, Wagner N, Reisfeld with SV40, SV40 T-antigen-transformed cells, or vaccinia virus recombi- RA, et al. T cell homeostatic proliferation elicits effective antitumor auto- nants expressing full-length T antigen or epitope minigenes. J Virol immunity. J Clin Invest 2002;110:185–92. 2000;74:6922–34. 25. Klebanoff CA, Khong HT, Antony PA, Palmer DC, Restifo NP. Sinks, 20. Hammer GE, Gonzalez F, Champsaur M, Cado D, Shastri N. The suppressors and antigen presenters: how lymphodepletion enhances aminopeptidase ERAAP shapes the peptide repertoire displayed by T cell-mediated tumor immunotherapy. Trends Immunol 2005;26: major histocompatibility complex class I molecules. Nat Immunol 111–7. 2006;7:103–12. 26. Reeves E, Colebatch-Bourn A, Elliott T, Edwards CJ, James E. Functionally 21. Cifaldi L, Lo Monaco E, Forloni M, Giorda E, Lorenzi S, Petrini S, et al. distinct ERAP1 allotype combinations distinguish individuals with anky- Natural killer cells efficiently reject lymphoma silenced for the endoplas- losing spondylitis. Proc Natl Acad Sci U S A 2014;111:17594–9. mic reticulum aminopeptidase associated with antigen processing. Cancer 27. Reeves E, Edwards CJ, Elliott T, James E. Naturally occurring ERAP1 Res 2011;71:1597–606. haplotypes encode functionally distinct alleles with fine substrate speci- 22. Cifaldi L, Romania P, Falco M, Lorenzi S, Meazza R, Petrini S, et al. ERAP1 ficity. J Immunol 2013;191:35–43. regulates natural killer cell function by controlling the engagement of 28. Roberts AR, Appleton LH, Cortes A, Vecellio M, Lau J, Watts L, et al. ERAP1 inhibitory receptors. Cancer Res 2015;75:824–34. association with ankylosing spondylitis is attributable to common geno- 23. Schmitt TM, Stromnes IM, Chapuis AG, Greenberg PD. New types rather than rare haplotype combinations. Proc Natl Acad Sci U S A strategies in engineering T-cell receptor gene-modified T cells to 2017;114:558–561.

3254 Cancer Res; 78(12) June 15, 2018 Cancer Research

ERAP1-Dependent Antigen Cross-Presentation Determines Efficacy of Adoptive T-cell Therapy in Mice

Karin Schmidt, Christin Keller, Anja A. Kühl, et al.

Cancer Res 2018;78:3243-3254. Published OnlineFirst March 20, 2018.

Updated version Access the most recent version of this article at: doi:10.1158/0008-5472.CAN-17-1946

Supplementary Access the most recent supplemental material at: Material http://cancerres.aacrjournals.org/content/suppl/2018/03/20/0008-5472.CAN-17-1946.DC1

Cited articles This article cites 28 articles, 18 of which you can access for free at: http://cancerres.aacrjournals.org/content/78/12/3243.full#ref-list-1

Citing articles This article has been cited by 2 HighWire-hosted articles. Access the articles at: http://cancerres.aacrjournals.org/content/78/12/3243.full#related-urls

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Department at Subscriptions [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://cancerres.aacrjournals.org/content/78/12/3243. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.