Cytotoxic T lymphocytes can reject established tumors if their target peptide is efficiently presented by MHC class I molecules (pMHC-I) on the surface of cancerous cells. Therapeutic success upon adoptive T-cell transfer (ATT), however, requires additional cross-presentation of the same pMHC-I on noncancerous cells. Endoplasmic reticulum aminopeptidase 1 (ERAP1) is an enzyme that customizes the N-terminus of proteasome-generated peptides so they can be loaded onto MHC-I molecules in the endoplasmic reticulum (ER). We show here that ERAP1 is critically involved in the process of tumor rejection and assumes a dual role by independently operating on both sides. Direct presentation of two MHC-I–restricted epitopes of a cancer-driving transplantation rejection antigen through ERAP1 moderately affected tumor rejection by adoptively transferred T-cell receptor gene–modified T cells in each case. ERAP1 expression by antigen cross-presenting cells of the ATT recipients was critical for expansion of therapeutic monospecific T cells and correlated with tumor rejection. Specifically, lack of ERAP1 expression in the ATT recipient's noncancerous cells enabled progression of pMHC-I–positive, IFNγ-responsive tumors, despite the presence of antigen-specific functional cytotoxic T lymphocytes. These data reveal a decisive role for ERAP1 in T-cell–mediated tumor rejection and will enhance the choice of MHC-I–restricted epitopes targeted by adoptive T-cell transfer.

Significance: This study demonstrates a role of ERAP1 in the efficacy of adoptive T-cell transfer and has potential to improve personalized T-cell therapy for solid tumors. Cancer Res; 78(12); 3243–54. ©2018 AACR.

One of the most effective therapies for patients with cancer is the transfer of tumor-infiltrating T lymphocytes with objective response rates >50% being achievable (1). Successful elimination of cancer depends on both, efficient direct presentation of the targeted MHC-I epitope, and on the cross-presentation of the same MHC-I epitope on noncancerous cells (2–5). Moreover, adequate T-cell persistence decides on the effectiveness of adoptive T-cell transfer (ATT), because the number of transferred T cells and the degree of their persistence in peripheral blood correlates with cancer regression (1, 6).

On the side of the target epitope, classical processing of MHC-I ligands involves three consecutive steps assumed by proteasomes, the transporter associated with antigen presentation (TAP), and 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 suitable to act as anchor for MHC-I binding (7). Those peptides are transported into the endoplasmic reticulum (ER) by TAP (8). In many cases the epitope-containing peptides are N-terminally elongated, requiring further optimization in the ER. Here, ERAP1 trims amino acid residues that flank the N-termini of antigenic precursors (9, 10). In the context of cancer immunotherapy, it was shown in mice that inhibition of ERAP1 caused increased tumor immunogenicity through direct presentation of a neo-epitope targetable by specific T cells (11). But when tumors evaded therapy with T-cell receptor (TCR) gene-modified T cells, recognition of IFNγ-resistant cancer variants by cytotoxic T lymphocytes (CTL) was reconstructible by overexpression of ERAP1 in vitro (12).

In contrast with the well-analyzed direct presentation of antigens, the procedures of antigen cross-presentation remain enigmatic. Considering the necessity of N-terminal trimming of peptides, Saveanu and colleagues (13) demonstrated that insulin-responsive aminopeptidase is required for efficient cross-presentation of endocytosed ovalbumin (OVA) and of phagocytosed antigen. Information on the role of ERAP1 in antigen cross-presentation is barely available and is mainly restricted to the analysis of the model antigen OVA (13–15). Here, we demonstrate the decisive role of ERAP1 for therapeutic outcome, showing that for optimal efficacy of ATT, ERAP1 must operate on both sides, direct antigen presentation in cancerous cells and antigen cross-presentation in the ATT recipient's cells.

Mice

LoxP-Tag and LoxP-Tag × Alb-Cre mice were described previously (16, 17). Erap1−/− mice were provided by K. Rock (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 Laboratories. Rag−/− mice (B6.129S6-Rag2tm1Fwa), Rag−/− × gc−/− mice (B10;B6-Rag2tm1FwaIl2rgtm1Wjl), and CD45.1 mice were purchased from Taconic and were bred in our animal facilities at the FEM Charité-Universitätsmedizin Berlin. Erap1−/− mice were crossed to Rag−/− mice to obtain Erap1−/− × Rag−/− mice. P14 × Rag−/− mice were described previously (12). C57BL/6 mice were provided by the FEM at Charité Universitätsmedizin Berlin. Male or female mice ages 2 to 9 months were used in animal experiments. All mouse studies were approved by the Landesamt für Gesundheit und Soziales, Berlin, Germany.

Hepatocellular carcinoma cell lines

To generate WT (Erap1+/+) TAg+ hepatocellular carcinoma (HCC) and Erap1−/− TAg+ HCC lines, livers were removed from tumor-bearing Erap1 × LoxP-Tag × Alb-Cre mice. Tumor tissue was cut into pieces with scalpels and was digested in RPMI supplemented with 10% heat-inactivated FCS (Biochrom), 1× penicillin/streptomycin (Biochrom), 1 mg/mL collagenase (Sigma), and 1× trypsin (Biochrom) for 4 hours at 37°C in a humidified 5% CO2 incubator. To prepare a single-cell suspension, digested tumor tissue was filtered through a 45-μm cell strainer (BD Biosciences) and washed with PBS twice. Single-cell suspensions were cultured in RPMI supplemented with 10% heat-inactivated FCS, 1× penicillin/streptomycin, 2 mmol/L glutamine (Biochrom), and 50 μmol/L β-mercaptoethanol (AppliChem). Adherent HCC cells were detached from cell culture flasks with trypsin (Biochrom) and were frozen in 90% culture medium supplemented with 10% dimethyl sulfoxide. Cells were cultivated for a maximum of 10 passages after thawing. HCC cell lines were regularly authenticated by Western blot analysis, but were not tested for Mycoplasma contamination.

Tumor challenge and adoptive T-cell transfer

Age- and sex-matched mice were subcutaneously injected into the right flank with 1 × 106 HCC cells in 200 μL PBS. Tumor size was measured with a caliper and the average tumor volume was determined from the measurements along three orthogonal axes (x, y, z). Tumor volumes were calculated according to the formula V (mm3) = (x y z)/2. On the day of treatment mice received intravenous injections of either 1 × 106 polyclonal CD8+ T cells, 5 × 104 gene-modified TCR-I T cells, or 5 × 104 gene-modified TCR-IV T cells re-suspended in 200 μL PBS. A small amount of blood was taken from the facial vein of the mice one and/or 4 weeks after adoptive transfer. Animals were sacrificed when the tumors reached 15-mm mean diameter and animals were excluded from analysis if they died from reasons unrelated to tumor burden. The experimenter was not blinded for the treatment groups.

Quantification and statistical analysis

Comparison of two groups was done by the Mann–Whitney test. Comparison of more ≥3 groups was done by the Kruskal–Wallis test. Two-way ANOVA was used followed by Bonferroni post-test for multiple comparisons. All statistical analysis was done with GraphPad Prism software version 5.0 and considered significant at *, P ≤ 0.05; **, P ≤ 0.01; and ***, P ≤ 0.001.

Two MHC-I epitopes differ in their dependence on ERAP1-mediated N-terminal peptide trimming

The cancer-driving antigen SV40 T Antigen (TAg) contains two codominant MHC-I–restricted epitopes, the H2-Db-restricted 10mer TAg206–215 (SAINNYAQKL, TAg-I), and the H2-Kb–restricted 8mer TAg404–411 (VVYDFLKC, TAg-IV; ref. 19). In Textor and colleagues (12), we showed that both standard and immunoproteasomes generated TAg-I along with the potential epitope precursors TAg205–215 (VSAINNYAQKL, 11mer) and TAg204–215 (RVSAINNYAQKL, 12mer). When peptide translocation was assessed, ATP-dependent transport was observed for TAg-I and the 11mer precursor peptide. The 12mer TAg-I precursor peptide did not translocate in an ATP-dependent fashion (Fig. 1A). TAP-dependent transport was confirmed by reduced accumulation of ATP signal in the presence of a high-affinity competitor peptide (Supplementary Fig. S1A). Upon analysis of N-terminal trimming of the precursor peptides by recombinant mouse ERAP1 (rmERAP1), TAg-I was generated from the 12mer through the 11mer intermediate (Supplementary Fig. S1B; Fig. 1B and C). Hence, though TAg-I is generated by rmERAP1, the consecutive action of proteasomes and TAP may even provide sufficient amounts of the 10mer.

Figure 1.

Two MHC-I epitopes derived from the same tumor antigen show varyingly strong dependence on two consecutive steps of antigen processing. A, Transport assay for TAg-I and related N-terminally extended precursor peptides. C4, high-affinity peptide devoid of an N-core glycosylation site and labeled with fluorescein; E5, peptide not binding to TAP, including an N-core glycosylation site and labeled with fluorescein; NST, reporter peptide, including an N-core glycosylation site and labeled with fluorescein. The experimental threshold (red dotted line) was set for E5+ATP (no TAP-binding). B and C, HPLC analysis of TAg-I-containing N-terminally extended precursor peptides trimmed by recombinant mouse ERAP1 in vitro. 3–6 ng rmERAP1 were incubated with 50 μmol/L peptide at 37°C and samples were analyzed by HPLC after 0 to 360 minutes (label on the right). D, Transport assay for TAg-IV and related N-terminally extended precursor peptides was performed as described in A. E and F, HPLC analysis of TAg-IV-containing N-terminally extended precursor peptides trimmed by recombinant mouse ERAP1 in vitro. The experiment was performed as described for B and C. A and D, Data are represented as mean ± SD, two-way ANOVA with Bonferroni posttests (**, P < 0.01; and ***, P < 0.001).

Figure 1.

Two MHC-I epitopes derived from the same tumor antigen show varyingly strong dependence on two consecutive steps of antigen processing. A, Transport assay for TAg-I and related N-terminally extended precursor peptides. C4, high-affinity peptide devoid of an N-core glycosylation site and labeled with fluorescein; E5, peptide not binding to TAP, including an N-core glycosylation site and labeled with fluorescein; NST, reporter peptide, including an N-core glycosylation site and labeled with fluorescein. The experimental threshold (red dotted line) was set for E5+ATP (no TAP-binding). B and C, HPLC analysis of TAg-I-containing N-terminally extended precursor peptides trimmed by recombinant mouse ERAP1 in vitro. 3–6 ng rmERAP1 were incubated with 50 μmol/L peptide at 37°C and samples were analyzed by HPLC after 0 to 360 minutes (label on the right). D, Transport assay for TAg-IV and related N-terminally extended precursor peptides was performed as described in A. E and F, HPLC analysis of TAg-IV-containing N-terminally extended precursor peptides trimmed by recombinant mouse ERAP1 in vitro. The experiment was performed as described for B and C. A and D, Data are represented as mean ± SD, two-way ANOVA with Bonferroni posttests (**, P < 0.01; and ***, P < 0.001).

Close modal

In contrast, TAg-IV was difficult to detect after proteasomal cleavage, whereas the corresponding N-terminally extended epitope precursor peptides TAg403–411 (SVVYDFLKC, 9mer) and TAg402–411 (DSVVYDFLKC, 10mer) were predominantly produced (12). In peptide translocation experiments, all three TAg-IV–containing proteasomal cleavage products were transported in dependence of ATP and TAP (Fig. 1D; Supplementary Fig. S1C). For the intermediate 9mer peptide precursor, translocation efficiency did not exceed the experimental threshold set by a control peptide (E5) not binding to TAP. When the TAg-IV–containing N-terminal precursor peptides were subjected to rmERAP1-digestion, TAg-IV was generated from the 10mer precursor peptide through the 9mer intermediate, and thus may constitute an ERAP1-dependent epitope (Fig. 1E and F).

Generation of primary wild-type and Erap1−/− TAg+ hepatocellular carcinoma with IFNγ-inducible antigen-processing machinery

Although both immunoproteasomes and TAP control the quantity of peptides being available for MHC-I loading, ERAP1 regulates the quality of the presented peptides (20). To assess the role of ERAP1 in vivo, TAg+ tumors genetically depleted for Erap1−/− were generated through crossing of Erap1−/− mice to LoxP-Tag × Alb-Cre mice, which express TAg specifically in hepatocytes and develop TAg+ HCC and cholangiolar carcinoma by the age of 3 to 4 months (Fig. 2A; refs. 17, 18). Erap1+/+ × LoxP-Tag × Alb-Cre double-transgenic offspring and Erap1−/− × LoxP-Tag × Alb-Cre triple-transgenic offspring likewise developed TAg+ HCC after about 3 to 4 months. Histological analysis of the primary HCC revealed a similar appearance of WT (Erap1+/+) TAg+ HCC and Erap1−/− TAg+ HCC (Fig. 2B). WT and Erap1−/− TAg+ HCC cell lines were subsequently established from primary HCC. Two cell lines of each WT and Erap1−/− TAg+ HCC were further characterized (referred to as HCC pair 1 and HCC pair 2). Expression or lack of expression of ERAP1 was confirmed by Western blot (Fig. 2C). Furthermore, IFNγ-signaling was examined having a focus on the expression of components of the antigen-processing machinery (APM). WT and Erap1−/− TAg+ HCC constitutively expressed JAK1 and downregulated JAK2 within 24 hours after stimulation with rmIFNγ. A highly elevated expression of STAT1 and phosphorylation of STAT1 was noticed in WT and Erap1−/− TAg+ HCC. Likewise, all HCC lines upregulated expression of APM components such as TAP1 and TAP2, and the immunoproteasome subunits β1i, β5i, and β2i (Fig. 2C). Notably, WT HCC of pair 2 showed a higher basal expression of ERAP1, LMP7, and MECL1. This correlated with higher amounts of MHC-I in the absence of IFNγ in this HCC line (Fig. 2D). Upon stimulation with IFNγ MHC-I expression increased in all HCC lines and was comparable for the WT and Erap1−/− TAg+ HCC of each HCC pair. Those two sets of WT and Erap1−/− TAg+ HCC lines with IFNγ-inducible APM were used for further analysis.

Figure 2.

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

Figure 2.

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

Close modal

ERAP1 regulates presentation of both IFNγ-independent and IFNγ-dependent tumor epitopes

Next, we examined the net effect of ERAP1 on the recognition of the two MHC-I–restricted epitopes TAg-I and TAg-IV in dependence of IFNγ. To do so, TCR gene-modified T cells recognizing either TAg-I (TCR-I T cells) or TAg-IV (TCR-IV T cells) were generated by retroviral transduction of splenocytes from P14 x Rag−/− donor mice (transduction rates were between 70% and 90%). When TCR-I T cells were cocultured with unstimulated WT and Erap1−/− TAg+ HCC lines, WT TAg+ HCC was recognized significantly better as compared with Erap1−/− TAg+ HCC (Fig. 3A). In contrast, upon stimulation with rmIFNγ before TCR-I T cell coculture, the effect of ERAP1 was negligible as shown by comparably good recognition of WT and Erap1−/− TAg+ HCC (Fig. 3B). These data reveal that TAg-I is IFNγ-independent in the case of ERAP1-expression, but is IFNγ-dependent in the absence of ERAP1.

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 TAg-I–specific TCR-I T cells with unstimulated (A) and IFNγ-stimulated (B) WT and Erap1−/− TAg+ HCC. C and D, Coculture of TAg-IV-specific TCR-IV T cells with unstimulated (C) and IFNγ-stimulated (D) WT and Erap1−/− TAg+ HCC. AD, Data are represented as mean ± SD, two-way ANOVA with Bonferroni posttests (**, P < 0.01; ***, P < 0.001).

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 TAg-I–specific TCR-I T cells with unstimulated (A) and IFNγ-stimulated (B) WT and Erap1−/− TAg+ HCC. C and D, Coculture of TAg-IV-specific TCR-IV T cells with unstimulated (C) and IFNγ-stimulated (D) WT and Erap1−/− TAg+ HCC. AD, Data are represented as mean ± SD, two-way ANOVA with Bonferroni posttests (**, P < 0.01; ***, P < 0.001).

Close modal

In contrast with TAg-I, TAg-IV was recognized by TCR-IV T cells on unstimulated WT TAg+ HCC, but not on unstimulated Erap1−/− TAg+ HCC (Fig. 3C). IFNγ stimulation before the TCR-IV T-cell coculture increased the recognition of TAg-IV in the absence of ERAP1, but a significant difference was maintained between WT and Erap1−/− TAg+ HCC (Fig. 3D). Hence, optimal presentation and CTL recognition of TAg-IV critically required joint action of IFNγ and ERAP1.

Absence of ERAP1 accelerates tumor growth in immune-deficient recipients

In LoxP-Tag × Alb-Cre mice, WT and Erap1−/− TAg+ HCC primarily developed in a host that was tolerant for their transplantation rejection antigen (16, 17). Therefore, the immunogenicity of WT and Erap1−/− TAg+ HCC was tested in immune-competent mice. After subcutaneous transplantation into WT (Erap1+/+, C57BL/6) and Erap1−/− mice (18), the recipients were observed for a period of 100 days. Of 3 to 4 mice injected per group, none developed a tumor, because those recipients are capable of priming a functional T-cell response toward TAg-I and TAg-IV (Supplementary Fig. S2A–S2C). To establish full-grown tumors, WT and Erap1−/− TAg+ HCC were grown in immune-deficient Rag−/− mice (Fig. 4A; Supplementary S2A). Research by others proposes a role of natural killer (NK) cells in recognizing ERAP-deficient tumor cells (21, 22). Therefore, we compared progression of WT versus Erap1−/− TAg+ HCC in NK-cell–competent (Rag−/−) recipients, progression of WT TAg+ HCC in NK-cell–competent versus NK-cell–deficient (Rag−/− x gc−/−) recipients, and progression of Erap1−/− TAg+ HCC in NK-cell–competent versus NK-cell–deficient recipients. No impact of NK cells on tumor growth of TAg+ HCC could be seen in the majority of mice (Supplementary Fig. S3A–S3C). One mouse showed exceptional tumor outgrowth of Erap1−/− TAg+ HCC starting 50 days after tumor cell inoculation. But NK-cell contribution at this late stage is unlikely. We concluded that NK-cell cytotoxicity is not relevant in the herein used model of transplanted TAg+ HCC.

Figure 4.

T-cell–mediated rejection of TAg+ HCC requires antigen cross-presentation and ERAP1. A, Schematic for tumor transplantation and rejection experiments. B, Graphical representation of rejection of WT and Erap1−/− TAg+ 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−/− TAg+ 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).

Figure 4.

T-cell–mediated rejection of TAg+ HCC requires antigen cross-presentation and ERAP1. A, Schematic for tumor transplantation and rejection experiments. B, Graphical representation of rejection of WT and Erap1−/− TAg+ 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−/− TAg+ 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).

Close modal

Interestingly, when the tumor size was compared in dependence of recipient's ERAP1 at the day of ATT, WT TAg+ HCC had a mean volume of 92 mm3 (WT→WT SD ± 63.5) in Rag−/− recipients while having a mean volume of 118 mm3 (WT→Erap1−/− SD ± 57.9) in Erap1−/− x Rag−/− recipients. Similarly, Erap1−/− TAg+ HCC were on average 143 mm3 in Rag−/− recipients (Erap1−/−→WT SD ± 62.3) as compared with Erap1−/− TAg+ HCC having a mean volume of 233 mm3 (Erap1−/−Erap1−/− SD ± 123.3; Supplementary Fig. S3D). In conclusion, lack of ERAP1 the noncancerous cells of immune-deficient recipients accelerated tumor growth after transplantation.

ERAP1 facilitates efficient rejection of TAg+ HCC through TCR-I T cells

The use of TCR gene-modified T cells is a particularly effective approach to target malignancies (23). To compare the impact of ERAP1-dependent direct presentation, MHC-I–dependent cross-presentation, and ERAP1-dependent cross-presentation on ATT efficiency, WT or Erap1−/− TAg+ HCC were established in ERAP1-competent H2b recipients (Rag−/−, WT/H2b), ERAP1-competent MHC-I–mismatched recipients (SCID, WT/H2d), and ERAP1-deficient H2b recipients (Rag−/−, Erap1−/−/H2b; Fig. 4A). At first, we investigated whether ERAP1 specifically affected rejection of TAg+ HCC through TCR-I T cells targeting TAg-I. The highest rejection rate was observed, when WT TAg+ HCC were treated in WT/H2b recipients with 100% and 88% of WT TAg+ of HCC pair 1 and 2 being rejected, respectively (Fig. 4B; Supplementary Table S1A). Notably, TCR-I gene-modified T cells were not capable of rejecting WT TAg+ HCC in WT/H2d recipients, where the target antigen cannot be cross-presented via MHC-I. Reduced rejection rates of 67% and 57% were also observed, when WT TAg+ HCC was treated with TCR-I T cells in Erap1−/−/H2b recipients. In contrast with WT TAg+ HCC, a reduced number of 70% and 80% Erap1−/− TAg+ HCC was rejected in WT/H2b recipients, indicating some contribution of ERAP1 on the direct presentation of TAg-I in vivo. Although Erap1−/− TAg+ HCC was not rejected in WT/H2d recipients, HCC pair 1 and 2 showed a huge difference with 33% and 100% being rejected, respectively, in the absence of host ERAP1 (Fig. 4B; Supplementary Table S1A). Hence, although expression of MHC-I and ERAP1 in ATT recipient's cells seemed to be required for rejection through TCR-I T cells, ERAP1-dependent direct presentation of TAg-I exerted a moderate effect.

Rejection of TAg+ HCC through TCR-IV T cells critically depends on ERAP1 expression in noncancerous cells

Adoptive transfers targeting TAg-IV through TCR-IV T cells were conducted in parallel. Here, in the group of WT/H2b recipients, complete rejection was observed for WT TAg+ HCC (Fig. 4C; Supplementary Table S1B). WT TAg+ HCC was not rejected if antigen cross-presentation was lacking in WT/H2d mice and strikingly, WT TAg+ HCC rejection rates were also strongly reduced (33% and 14%) if ERAP1 was not expressed in the recipient's cells. In addition, ERAP1 affected direct presentation of TAg-IV, as indicated by inferior rejection of 70% and 83% Erap1−/− TAg+ HCC in WT/H2b recipients. The requirement of antigen cross-presentation of TAg-IV was confirmed by lack of rejection of Erap1−/− TAg+ HCC in WT/H2d mice, and by less than half (45%) of the Erap1−/− TAg+ HCC being rejected in Erap1−/−/H2b recipients (Fig. 4C; Supplementary Table S1B). In summary, ERAP1-depedent direct presentation moderately affected tumor rejection, whereas both, MHC-I–dependent antigen cross-presentation and ERAP1 expression in noncancerous cells were critically required for rejection of TAg+ HCC through TCR-IV gene-modified cells (Fig. 4D).

Absence of ERAP1 enables escape of small TAg+ HCC despite the presence of functional CTL

Next, tumor volumes at the day of ATT were correlated with subsequent tumor rejection to rule out that the above described variations between the groups affected the efficiency of TCR-I or TCR-IV T-cell therapy. At first, 57 mice treated with TCR-I T cells were analyzed (Supplementary Fig. S3E). Mice were grouped according to whether tumors were rejected or not and the tumor size at the day of treatment was plotted for both groups. In the case of TCR-I, both rejected and nonrejected tumors had an average volume of 132 mm3 and 135 mm3, respectively (rejected SD ± 92.2; not rejected SD ± 95.4; Fig. 5A). The same analysis was performed for 57 mice treated with TCR-IV T cells (Supplementary Fig. S3F). In this case, rejected tumors had a mean volume of 135 mm3 on the day of ATT, whereas nonrejected tumors were on average 171 mm3 (rejected SD ± 83.9; not rejected SD ± 104.2; Fig. 5B). In summary, the tumor size at the day of ATT did not affect rejection through TCR-I T cells or TCR-IV T cells.

Figure 5.

Small TAg+ 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 TAg+ HCC was rejected or not and the tumor volume at the day of ATT was plotted, rejected (n = 43), not rejected (n = 14). B, All mice depicted in Supplementary Fig. S3F were grouped according to whether TAg+ HCC was rejected or not and the tumor volume at the day of ATT was plotted, rejected (n = 37), not rejected (n = 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. Naïve control, n = 6; immunized control, n = 6; rejected, n = 24; not rejected (n = 6). E,In vivo cytotoxicity analysis for TAg-IV was performed as described in C. Data of three experiments are shown. Naïve control, n = 8; immunized control, n = 8; rejected, n = 22; not rejected, n = 8. D and E, All data are represented with mean ± SD, Kruskal–Wallis test.

Figure 5.

Small TAg+ 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 TAg+ HCC was rejected or not and the tumor volume at the day of ATT was plotted, rejected (n = 43), not rejected (n = 14). B, All mice depicted in Supplementary Fig. S3F were grouped according to whether TAg+ HCC was rejected or not and the tumor volume at the day of ATT was plotted, rejected (n = 37), not rejected (n = 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. Naïve control, n = 6; immunized control, n = 6; rejected, n = 24; not rejected (n = 6). E,In vivo cytotoxicity analysis for TAg-IV was performed as described in C. Data of three experiments are shown. Naïve control, n = 8; immunized control, n = 8; rejected, n = 22; not rejected, n = 8. D and E, All data are represented with mean ± SD, Kruskal–Wallis test.

Close modal

In addition, T lymphocyte function was analyzed by in vivo 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 (SAINNYAQKL)-loaded target cells, whereas the same peptide-loaded target cells were efficiently killed by TAg-immunized mice or TCR-I T-cell–treated mice (Fig. 5C). TAg (16.113)-immunized controls showed a mean of 75% specific cytotoxicity toward TAg-I (SD ± 18.0). For a determination of possible differences in the functionality of persistent CTLs after tumor rejection or recurrence, mice treated with TCR-I T cells were grouped on whether the TAg+ HCC were rejected or not. But while mice rejecting through TCR-I showed 87% cytotoxicity in vivo (SD ± 21.3), a mean of 81% TAg-I–specific cytotoxicity was detectable in mice with nonrejected tumors as well (SD ± 30.2). These data confirmed that the functionality of TCR-I T cells was maintained throughout the experiment, and failure of rejection was not caused by a loss of function of the transferred TCR-I T cells (Fig. 5D). Similarly, upon analysis of adoptively transferred TCR-IV T cells, naive control mice did not kill TAg-IV (VVYDFLKL)–loaded target cells, whereas TAg (16.113)-immunized control mice presented on average 94% (SD ± 5.8) TAg-IV–specific cytotoxicity. In the sample groups, on average 89% (SD ± 20.7) TAg-IV–specific cytotoxicity was detectable in those mice in which TAg+ HCC were rejected, whereas mice with nonrejected TAg+ HCC showed a mean of 88% (SD ± 21.6) TAg-IV–specific cytotoxicity (Fig. 5E). In summary, rejection of TAg+ HCC failed despite the presence of in vivo cytotoxic TCR-IV T cells.

Expansion of TCR gene-modified T cells requires ERAP1 expression in ATT recipients

For ATT, therapeutic T cells are transferred into lympho-depleted individuals, because enhanced homeostatic T-cell proliferation is required for efficient antitumor immunity (24, 25). Hence, the expansion of CD8+ T cells (within white blood cells) was analyzed and compared between the different recipient groups (Supplementary Fig. S4A–S4C). In the case of TCR-I T cells, CD8+ cells constituted on average 0.5% (SD ± 0.75) in WT/H2b recipients bearing WT TAg+ HCC one week after ATT. Similarly, in WT/H2b recipients bearing Erap1−/− TAg+ HCC approximately 0.6% (SD ± 0.89) were CD8+ T cells (Fig. 6A). In contrast, TCR-I T cells did not expand in WT/H2d recipients and were barely detectable in Erap1−/−/H2b recipients regardless of whether they were bearing WT TAg+ HCC or Erap1−/− TAg+ HCC (Fig. 6A). At this early time point a considerable variation was observed between mice, and MHC-I–independent proliferation of TCR-I T cells was noticed in the WT/H2d group bearing Erap1−/− TAg+ HCC. Control experiments revealed that in vitro primed TCR-I gene–modified T cells did not expand in tumor-free recipients in the absence of TAg, whereas in vivo primed TAg-I–specific TE cells were capable of expanding antigen-independently (Supplementary Fig. S5A and S5B). Four weeks after ATT, a sustained expansion was observed in 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 TAg+ HCC or Erap1−/− TAg+ HCC, respectively (Fig. 6B). Still, CD8+ T cells were not detectable in WT/H2d mice and were very low in Erap1−/−/H2b recipients reaching a mean of 1% (SD ± 0.83) and 3% (SD ± 3.51; Fig. 6B). In summary, expansion of adoptively transferred gene-modified TCR-I T cells required compatible MHC-I, and expression of ERAP1 in ATT recipients, whereas lack of ERAP1 in cancerous cells had not impact hereon.

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 = 16; WT→WT/H2d, n = 3; WT→Erap1−/−/H2b, n = 11; Erap1−/−→WT/H2b, n = 7; Erap1−/−→WT/H2d, n = 4; Erap1−/−Erap1−/−/H2b, n = 5. B, FACS analysis of TCR-I T cell expansion four weeks after ATT. WT→WT/H2b, n = 6; WT→WT/H2d, n = 2; WT→Erap1−/−/H2b, n = 3; Erap1−/−→WT/H2b, n = 9; Erap1−/−→WT/H2d, n = 4; Erap1−/−Erap1−/−/H2b, n = 7. C, FACS analysis of TCR-IV T cells performed one week after ATT. WT→WT/H2b, n = 16; WT→WT/H2d, n = 4; WT→Erap1−/−/H2b, n = 11; Erap1−/−→WT/H2b, n = 12; Erap1−/−→WT/H2d, n = 3; Erap1−/−Erap1−/−/H2b, n = 6. D, FACS analysis of TCR-IV T cells four weeks after ATT. WT→WT/H2b, n = 6; WT→WT/H2d, n = 3; WT→Erap1−/−/H2b, n = 4; Erap1−/−→WT/H2b, n = 10; Erap1−/−→WT/H2d, n = 3; Erap1−/−Erap1−/−/H2b, n = 6. A–D, Data of n = 3 experiments are shown with mean ± SD, Kruskal–Wallis test. E, All mice with haplotype H2b depicted in A were grouped according to whether TAg+ HCC was rejected or not and the percentage of CD8+ T cells was plotted, rejected (n = 28), not rejected (n = 11). F, All mice with haplotype H2b depicted in B were grouped according to whether TAg+ HCC was rejected or not and the percentage of CD8+ T cells was plotted, rejected (n = 21), not rejected (n = 4). G, All mice with haplotype H2b depicted in C were grouped according to whether TAg+ HCC was rejected or not and the percentage of CD8+ T cells was plotted, rejected (n = 30), not rejected (n = 15). H, All mice with haplotype H2b depicted in D were grouped according to whether TAg+ HCC was rejected or not and the percentage of CD8+ T cells was plotted, rejected (n = 19), not rejected (n = 7). E–H, Data are represented with mean ± SD, Mann–Whitney test (*, P < 0.05).

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 = 16; WT→WT/H2d, n = 3; WT→Erap1−/−/H2b, n = 11; Erap1−/−→WT/H2b, n = 7; Erap1−/−→WT/H2d, n = 4; Erap1−/−Erap1−/−/H2b, n = 5. B, FACS analysis of TCR-I T cell expansion four weeks after ATT. WT→WT/H2b, n = 6; WT→WT/H2d, n = 2; WT→Erap1−/−/H2b, n = 3; Erap1−/−→WT/H2b, n = 9; Erap1−/−→WT/H2d, n = 4; Erap1−/−Erap1−/−/H2b, n = 7. C, FACS analysis of TCR-IV T cells performed one week after ATT. WT→WT/H2b, n = 16; WT→WT/H2d, n = 4; WT→Erap1−/−/H2b, n = 11; Erap1−/−→WT/H2b, n = 12; Erap1−/−→WT/H2d, n = 3; Erap1−/−Erap1−/−/H2b, n = 6. D, FACS analysis of TCR-IV T cells four weeks after ATT. WT→WT/H2b, n = 6; WT→WT/H2d, n = 3; WT→Erap1−/−/H2b, n = 4; Erap1−/−→WT/H2b, n = 10; Erap1−/−→WT/H2d, n = 3; Erap1−/−Erap1−/−/H2b, n = 6. A–D, Data of n = 3 experiments are shown with mean ± SD, Kruskal–Wallis test. E, All mice with haplotype H2b depicted in A were grouped according to whether TAg+ HCC was rejected or not and the percentage of CD8+ T cells was plotted, rejected (n = 28), not rejected (n = 11). F, All mice with haplotype H2b depicted in B were grouped according to whether TAg+ HCC was rejected or not and the percentage of CD8+ T cells was plotted, rejected (n = 21), not rejected (n = 4). G, All mice with haplotype H2b depicted in C were grouped according to whether TAg+ HCC was rejected or not and the percentage of CD8+ T cells was plotted, rejected (n = 30), not rejected (n = 15). H, All mice with haplotype H2b depicted in D were grouped according to whether TAg+ HCC was rejected or not and the percentage of CD8+ T cells was plotted, rejected (n = 19), not rejected (n = 7). E–H, Data are represented with mean ± SD, Mann–Whitney test (*, P < 0.05).

Close modal

Expansion of TCR-IV T cells was analyzed at the same time. Here, too, a strong variation was observed between individual mice one week after ATT. On average, 0.8% (SD ± 1.64) to 0.9% (SD ± 1.20) CD8+ T cells were detectable in WT/H2b recipients bearing WT or Erap1−/− TAg+ HCC (Fig. 6C). TCR-IV T cells did not expand in WT/H2d mice and represented less than 0.3% in Erap1−/−/H2b recipients (Fig. 6C). Similar to TCR-I T cells, TCR-IV T cells expanded within the following 3 weeks. Accordingly, 4 weeks after ATT a mean of 5% (SD ± 1.57) and 8% (SD ± 8.11) CD8+ T cells were detected in WT/H2b recipients bearing WT or Erap1−/− TAg+ HCC, respectively (Fig. 6D). In WT/H2d recipients or Erap1−/−/H2b recipients, CD8+ T cells still constituted no more than 0.01% to 0.8% (Fig. 6D). Although TAg-I–specific T cells may also expand antigen independently, TAg-IV–specific T cells expanded exclusively in the presence of TAg (Supplementary Fig. S5C). Interestingly, if polyclonal TAg-specific CD8+ effector T (TE) cells [containing a mean of 1.6% (SD ± 0.93, n = 3) TAg-I–specific TE cells and 8.4% (SD ± 7.89, n = 3) TAg-IV–specific TE cells] were isolated from TAg-immunized C57BL/6 mice, and were transferred into TAg+ HCC-bearing recipients, ERAP1 assumed a more prominent role on tumor rejection as related to direct epitope presentation. However, overall expansion of the adoptively transferred polyclonal CD8+ TE cells was not affected in ERAP1-deficient recipients, indicating that only antigen-specific T cells required recipient's ERAP1 to proliferate (Supplementary Fig. S6A–S6D). In conclusion, the recipients' ERAP1 status, but not the ERAP1 status of the targeted TAg+ HCC was crucial for expansion of epitope-specific TCR gene-modified T cells.

Immediate expansion of antigen-specific T cells decides on tumor rejection

Impaired expansion of antigen-specific T cells after ATT was a remarkable symptom of ERAP1-deficient ATT recipients, in which TAg+ HCC was poorly rejected. To examine whether expansion of TCR-I T cells correlated with rejection of TAg+ HCC, all H2b recipient mice that were analyzed according to their percentage of TCR-I CD8+ T cells in the blood one week after ATT were grouped on whether mice had rejected the TAg+ HCC or not. At this time point mice with later on rejected tumors presented a mean of 0.5% (SD ± 0.73) CD8+ T cells as compared with a significantly lower mean of 0.1% (SD ± 0.19) CD8+ T lymphocytes in mice that subsequently did not reject tumors (Fig. 6E). When the H2b TCR-I T-cell–treated mice were grouped according to the same parameters, CD8+ T cells constituted on average 6% (SD ± 4.92) in mice with later on 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 correlated with subsequent tumor rejection.

In those mice adoptively transferred with TCR-IV T cells, approximately 0.8% (SD ± 1.47) CD8+ T lymphocytes were detectable one week after ATT in H2b mice that rejected TAg+ HCC later on. Opposite to that, a significantly lower percentage of approximately 0.1% (SD ± 0.14) CD8+ T cells was detected in mice with subsequently nonrejected TAg+ HCC (Fig. 6G). That statistical connection was still observable 4 weeks after ATT, and on average 6% (SD ± 6.56) CD8+ T cells were present in the blood of the mice with ultimately rejected tumors. In contrast, those mice that did not reject tumors had a mean of 0.8% (SD ± 0.79) of CD8+ T cells (Fig. 6H). These data confirm that both, ERAP1 function in ATT recipients and T-cell expansion and persistence, are required for successful tumor rejection after ATT.

Lack of ERAP1 enables escape of T-cell–recognizable IFNγ-responsive TAg+ HCC

Relating to earlier observations in Textor and colleagues (12) of IFNγ–unresponsive cancer variants to evade T-cell recognition, we wanted to rule out that an acquired deficiency in IFNγ-signaling contributed to TAg+ HCC recurrence within the context of ERAP1-regulated epitope presentation. Six nonrejected TAg+ HCC were grown in vitro for further analysis. Among these, two WT TAg+ HCC were reisolated after TCR-I T-cell therapy in Erap1−/− recipients (#377, #378), one WT TAg+ HCC was reisolated after TCR-IV T-cell therapy in an Erap1−/− recipient (#400), one Erap1−/− TAg+ HCC was reisolated after TCR-IV T-cell therapy in an Erap1−/− recipient (#556), and two Erap1−/− TAg+ HCC were reisolated after TCR-IV T cell therapy in ERAP1-competent recipients (#306, #282, Fig. 7A). First, we confirmed expression of the transplantation rejection antigen TAg by the nonrejected HCC (Fig. 7B). After stimulation with rmIFNγ, JAK1 was constitutively expressed in all reisolated TAg+ HCC lines in different amounts, but upregulation of STAT1, as well as phosphorylation thereof, was always detectable (Fig. 7B). In addition, TAP1 and TAP2 were highly expressed in all six TAg+ HCC after IFNγ stimulation, just as the immunoproteasome subunits β1i, β5i, and β2i (Fig. 7B). When MHC-I expression was analyzed, all reisolated tumors showed expression of both alleles, H2-Db and H2-Kb, albeit with varying degree (Fig. 7C).

Figure 7.

Lack of ERAP1 enables recurrence of TAg+, TCR-recognizable HCC with IFNγ-inducible antigen-processing machinery. A, Overview of WT or Erap1−/− TAg+ HCC reisolated after TCR-I or TCR-IV therapy in WT/H2b or Erap1−/−/H2b recipients. B, Western blot analysis of IFNγ-inducible APM components in reisolated TAg+ HCC. C, FACS analysis of IFNγ-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 TAg+ HCC. D and E, Data are represented as mean ± SD.

Figure 7.

Lack of ERAP1 enables recurrence of TAg+, TCR-recognizable HCC with IFNγ-inducible antigen-processing machinery. A, Overview of WT or Erap1−/− TAg+ HCC reisolated after TCR-I or TCR-IV therapy in WT/H2b or Erap1−/−/H2b recipients. B, Western blot analysis of IFNγ-inducible APM components in reisolated TAg+ HCC. C, FACS analysis of IFNγ-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 TAg+ HCC. D and E, Data are represented as mean ± SD.

Close modal

Finally, in vitro recognition of the reisolated tumor cells by TCR-I and TCR-IV T cells was analyzed. All six TAg+ HCCs were recognized by TCR-I T cells (Fig. 7D). Likewise, all six TAg+ HCC were well recognized by TCR-IV T cells, with the exception of #556, the Erap1−/− TAg+ HCC that has progressed after TCR-IV T-cell therapy in an Erap1−/− recipient. This HCC line showed an IFNγ-inducible, albeit very low, CTL response that was in accordance with the rather low expression of MHC-I by this HCC line in comparison with the other nonrejected TAg+ HCC (Fig. 7E). If antigen cross-presentation was assessed in bone marrow–derived dendritic cells (BMDC), a slightly impaired cross-presentation of TAg-I and TAg-IV as processed from full-length purified TAg protein was observed with Erap1−/− BMDCs in comparison with WT BMDCs (Supplementary Fig. S7A and S7B). We concluded that the sole lack of ERAP1 in tumor cells or in the recipient's non-tumor cells determined the failure of ATT, even though further experimentation is required to elucidate the actual antigen cross-presentation pathway involving ERAP1.

We showed that ERAP1 is critically required for the processing of MHC-I epitopes as targeted by ATT and critically affects successful eradication of cancer. In this study, rejection of transplantable tumors was decreased by 25% and 23% for TCR-I and TCR-IV, respectively, if cancer cells were deficient for ERAP1 but expressed other IFNγ-inducible APM components such as immunoproteasomes and TAP. The previously reported inference of TAg-I being ERAP1-independently processed was limited to direct antigen presentation in the presence of IFNγ and with cross-presented TAg (12). The herein presented data imply that no other APM components are capable of completely substituting the specialized function of ERAP1.

Previously, it was shown by others that NK cells can reject suspensions of ERAP1-silenced RMA lymphoma cells due to poor engagement of the Ly49C/I NK-cell–inhibitory receptor by altered pMHC-I in the absence of N-terminal trimming of peptides (21). And pharmacological inhibition of ERAP1 in human tumor cells induced an NK-cell response caused by the tumor cell's inability to engage inhibitory NK-cell receptors (22). Adoptive T-cell therapy through transfer of TCR gene-modified T cells, however, is an approach to target solid tumors, that escaped the patient's spontaneous immune recognition and control experiments in this study did not confirm an early role of NK cells, if Erap1 was genetically depleted and surface expression of MHC-I was adjustable by IFNγ. Here, the therapeutic avenue of pharmacologically inhibiting ERAP1 is certainly a strategy that contrasts adoptive T-cell transfer, if the latter one targets ERAP1-dependent MHC-I epitopes. Another study showed that attenuation of ERAP1 function may induce cell surface presentation of protective tumor antigens inducing functional T-cell responses (11). Those findings show that enhancing ERAP1 function is probably not an alternative therapeutic avenue either, because ERAP1 has the potential to over-trim peptides and to destroy MHC-I epitopes. Our data are of particular importance with regard to the various common SNPs in the human Erap1 homolog and the resulting naturally existing alleles of Erap1 encoding ERAP1 variants with different, sometimes impaired, function (26–28). In this study, we used a mouse model of genetically depleted Erap1, causing complete loss of function of the enzyme, and transferability of such findings to functionally impaired ERAP1 variants in humans remains to be seen. Nevertheless, personalized cancer immunotherapy of the future may consider processing of the targeted epitope as well as patient's Erap1 haplotype as an additional marker. Further investigations, including MHC-I epitopes presented by clinically relevant tumor samples, would also show whether our findings, as obtained from a transplantable tumor model, can be translated to improve adoptive T-cell therapy approaches in patients.

For the first time, we show that ERAP1 is certainly playing a role in antigen cross-presentation of tumor epitopes as targeted by ATT and thereby supports rejection of established tumors. Previous work by others reported that ERAP1 is required for cross-presentation of the model antigen OVA (13–15). But if tumor antigens are cross-presented in vivo, cell-associated antigens from dying tumor cells may be the primary source of antigen. So far, only one other study has examined ERAP1-dependent cross-presentation of cell-associated antigens. The cross-presentation of cell-associated OVA and cell-associated male HY antigen was analyzed after immunization of ERAP1-deficient mice with MHC-I–mismatched antigen-positive cells (15). Here, ERAP1-deficiency reduced proliferation of OVA-specific or HY antigen-specific CD8+ T cells, respectively. Our in vivo investigations with the tumor rejection antigen TAg confirm these earlier observations, but analysis of ERAP1-dependent cross-presentation of TAg-I and TAg-IV in BMDCs in vitro requires improved experimental procedures. Importantly, our experiments showed that ERAP1 has a decisive effect on tumor rejection by being critically required for the proliferation of CD8+ T cells after ATT, and that lack of ERAP1 in ATT recipients consequently caused failure of adoptive T-cell therapy.

No potential conflicts of interest were disclosed.

Conception and design: K. Schmidt, U. Seifert, T. Blankenstein, G. Willimsky, P.-M. Kloetzel

Development of methodology: K. Schmidt, G. Willimsky

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): K. Schmidt, A. Textor, G. Willimsky

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): K. Schmidt, A. Textor

Writing, review, and/or revision of the manuscript: K. Schmidt, A.A. Kühl, A. Textor, T. Blankenstein, G. Willimsky, P.-M. Kloetzel

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): C. Keller, A.A. Kühl

Study supervision: K. Schmidt, G. Willimsky, P.-M. Kloetzel

The authors would like to thank F. Kirschner, A. Lehmann, and N. Albrecht-Köpke from the Institute of Biochemistry, Charité—Universitätsmedizin Berlin for assistance with data acquisition. U. Seifert, T. Blankenstein, and G. Willimsky received grants from the Deutsche Forschungsgemeinschaft (SFB-TR36). T. Blankenstein and P.-M. Kloetzel received grants from the Berlin Institute of Health (CRG-1). P.-M. Kloetzel received funding from the Wilhelm Sander-Stiftung (2015.107.1).

The costs of publication of this article were defrayed in part by the payment 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.

1.
Rosenberg
SA
,
Yang
JC
,
Sherry
RM
,
Kammula
US
,
Hughes
MS
,
Phan
GQ
, et al
Durable complete responses in heavily pretreated patients with metastatic melanoma using T-cell transfer immunotherapy
.
Clin Cancer Res
2011
;
17
:
4550
7
.
2.
Spiotto
MT
,
Yu
P
,
Rowley
DA
,
Nishimura
MI
,
Meredith
SC
,
Gajewski
TF
, et al
Increasing tumor antigen expression overcomes "ignorance" to solid tumors via crosspresentation by bone marrow-derived stromal cells
.
Immunity
2002
;
17
:
737
47
.
3.
Spiotto
MT
,
Rowley
DA
,
Schreiber
H
. 
Bystander elimination of antigen loss variants in established tumors
.
Nat Med
2004
;
10
:
294
8
.
4.
Leisegang
M
,
Engels
B
,
Schreiber
K
,
Yew
PY
,
Kiyotani
K
,
Idel
C
, et al
Eradication of large solid tumors by gene therapy with a T-cell receptor targeting a single cancer-specific point mutation
.
Clin Cancer Res
2016
;
22
:
2734
43
.
5.
Engels
B
,
Engelhard
VH
,
Sidney
J
,
Sette
A
,
Binder
DC
,
Liu
RB
, et al
Relapse or eradication of cancer is predicted by peptide-major histocompatibility complex affinity
.
Cancer Cell
2013
;
23
:
516
26
.
6.
Robbins
PF
,
Dudley
ME
,
Wunderlich
J
,
El-Gamil
M
,
Li
YF
,
Zhou
J
, et al
Cutting edge: persistence of transferred lymphocyte clonotypes correlates with cancer regression in patients receiving cell transfer therapy
.
J Immunol
2004
;
173
:
7125
30
.
7.
Kloetzel
PM
. 
Antigen processing by the proteasome
.
Nat Rev Mol Cell Biol
2001
;
2
:
179
87
.
8.
Neefjes
JJ
,
Momburg
F
,
Hammerling
GJ
. 
Selective and ATP-dependent translocation of peptides by the MHC-encoded transporter
.
Science
1993
;
261
:
769
71
.
9.
Fruci
D
,
Niedermann
G
,
Butler
RH
,
van Endert
PM
. 
Efficient MHC class I-independent amino-terminal trimming of epitope precursor peptides in the endoplasmic reticulum
.
Immunity
2001
;
15
:
467
76
.
10.
Serwold
T
,
Gaw
S
,
Shastri
N
. 
ER aminopeptidases generate a unique pool of peptides for MHC class I molecules
.
Nat Immunol
2001
;
2
:
644
51
.
11.
James
E
,
Bailey
I
,
Sugiyarto
G
,
Elliott
T
. 
Induction of protective antitumor immunity through attenuation of ERAAP function
.
J Immunol
2013
;
190
:
5839
46
.
12.
Textor
A
,
Schmidt
K
,
Kloetzel
PM
,
Weißbrich
B
,
Perez
C
,
Charo
J
, et al
Preventing tumor escape by targeting a post-proteasomal trimming independent epitope
.
J Exp Med
2016
;
213
:
2333
2348
.
13.
Saveanu
L
,
Carroll
O
,
Weimershaus
M
,
Guermonprez
P
,
Firat
E
,
Lindo
V
, et al
IRAP identifies an endosomal compartment required for MHC class I cross-presentation
.
Science
2009
;
325
:
213
7
.
14.
Yan
J
,
Parekh
VV
,
Mendez-Fernandez
Y
,
Olivares-Villagómez
D
,
Dragovic
S
,
Hill
T
, et al
In vivo role of ER-associated peptidase activity in tailoring peptides for presentation by MHC class Ia and class Ib molecules
.
J Exp Med
2006
;
203
:
647
59
.
15.
Firat
E
,
Saveanu
L
,
Aichele
P
,
Staeheli
P
,
Huai
J
,
Gaedicke
S
, et al
The role of endoplasmic reticulum-associated aminopeptidase 1 in immunity to infection and in cross-presentation
.
J Immunol
2007
;
178
:
2241
8
.
16.
Willimsky
G
,
Czéh
M
,
Loddenkemper
C
,
Gellermann
J
,
Schmidt
K
,
Wust
P
, et al
Immunogenicity of premalignant lesions is the primary cause of general cytotoxic T lymphocyte unresponsiveness
.
J Exp Med
2008
;
205
:
1687
700
.
17.
Willimsky
G
,
Blankenstein
T
. 
Sporadic immunogenic tumours avoid destruction by inducing T-cell tolerance
.
Nature
2005
;
437
:
141
6
.
18.
York
IA
,
Brehm
MA
,
Zendzian
S
,
Towne
CF
,
Rock
KL
. 
Endoplasmic reticulum aminopeptidase 1 (ERAP1) trims MHC class I-presented peptides in vivo and plays an important role in immunodominance
.
Proc Natl Acad Sci U S A
2006
;
103
:
9202
7
.
19.
Mylin
LM
,
Schell
TD
,
Roberts
D
,
Epler
M
,
Boesteanu
A
,
Collins
EJ
, et al
Quantitation of CD8(+) T-lymphocyte responses to multiple epitopes from simian virus 40 (SV40) large T antigen in C57BL/6 mice immunized with SV40, SV40 T-antigen-transformed cells, or vaccinia virus recombinants expressing full-length T antigen or epitope minigenes
.
J Virol
2000
;
74
:
6922
34
.
20.
Hammer
GE
,
Gonzalez
F
,
Champsaur
M
,
Cado
D
,
Shastri
N
. 
The aminopeptidase ERAAP shapes the peptide repertoire displayed by major histocompatibility complex class I molecules
.
Nat Immunol
2006
;
7
:
103
12
.
21.
Cifaldi
L
,
Lo Monaco
E
,
Forloni
M
,
Giorda
E
,
Lorenzi
S
,
Petrini
S
, et al
Natural killer cells efficiently reject lymphoma silenced for the endoplasmic reticulum aminopeptidase associated with antigen processing
.
Cancer Res
2011
;
71
:
1597
606
.
22.
Cifaldi
L
,
Romania
P
,
Falco
M
,
Lorenzi
S
,
Meazza
R
,
Petrini
S
, et al
ERAP1 regulates natural killer cell function by controlling the engagement of inhibitory receptors
.
Cancer Res
2015
;
75
:
824
34
.
23.
Schmitt
TM
,
Stromnes
IM
,
Chapuis
AG
,
Greenberg
PD
. 
New strategies in engineering T-cell receptor gene-modified T cells to more effectively target malignancies
.
Clin Cancer Res
2015
;
21
:
5191
7
.
24.
Dummer
W
,
Niethammer
AG
,
Baccala
R
,
Lawson
BR
,
Wagner
N
,
Reisfeld
RA
, et al
T cell homeostatic proliferation elicits effective antitumor autoimmunity
.
J Clin Invest
2002
;
110
:
185
92
.
25.
Klebanoff
CA
,
Khong
HT
,
Antony
PA
,
Palmer
DC
,
Restifo
NP
. 
Sinks, suppressors and antigen presenters: how lymphodepletion enhances T cell-mediated tumor immunotherapy
.
Trends Immunol
2005
;
26
:
111
7
.
26.
Reeves
E
,
Colebatch-Bourn
A
,
Elliott
T
,
Edwards
CJ
,
James
E
. 
Functionally distinct ERAP1 allotype combinations distinguish individuals with ankylosing spondylitis
.
Proc Natl Acad Sci U S A
2014
;
111
:
17594
9
.
27.
Reeves
E
,
Edwards
CJ
,
Elliott
T
,
James
E
. 
Naturally occurring ERAP1 haplotypes encode functionally distinct alleles with fine substrate specificity
.
J Immunol
2013
;
191
:
35
43
.
28.
Roberts
AR
,
Appleton
LH
,
Cortes
A
,
Vecellio
M
,
Lau
J
,
Watts
L
, et al
ERAP1 association with ankylosing spondylitis is attributable to common genotypes rather than rare haplotype combinations
.
Proc Natl Acad Sci U S A
2017
;
114
:
558
561
.