Mice were maintained under specific pathogen-free conditions, and all procedures were performed according to the protocols approved by the Institutional Committee for Use and Care of Laboratory Animals of Tohoku University, which was granted by Tohoku University Ethics Review Board (2019AcA-003). For collection of tissue samples, mice were sacrificed by cervical dislocation, and all efforts were made to minimize suffering.

C57BL/6 mice (6 to 8-week-old females) were purchased from CLEA Japan. OT-I mice were purchased from Jackson Laboratories.

Palladium allergy was induced as described previously (23). In brief, mice were injected twice into both sides of the groin with 250 μl PBS containing 10 mM PdCl₂ and 10 μg/ml LPS at an interval of seven days. Seven days after the 2^nd injection, mice were challenged by intradermal injection of 25 μl of 10 mM PdCl₂ in PBS into both the left and right footpads.

Pd-responsive T cells were generated as described in a previous report (24). In brief, Pd allergy was induced in mice as described above. Twenty-four hours after challenge, inguinal and popliteal lymph nodes (LN) were collected. LN cells were cultured in the presence of 20 μM PdCl₂ for 5 days. After washing, these cells were further co-cultured with irradiated (20 Gy) splenocytes in the presence of 20 μM PdCl₂ and 100 U/ml recombinant IL-2 (Wako) for 5 days. Finally, LN cells were cultured in the presence of 100 U/ml IL-2 alone for 3 days. TCR repertoire analysis was performed as described previously (21).

The murine dendritic cell line DC2.4 (H-2K^b, H-2D^b) was purchased from Merck Millipore. Human CD8-expressing TG40 cells were a kind gifted from Dr. Kishi (Toyama University).

All antibodies used in flow cytometric analysis were purchased from Biolegend: FITC-conjugated anti-H-2K^bD^b antibody (clone 28-8-6), FITC-conjugated anti-H-2K^b antibody (clone AF6-88.5), APC-conjugated anti-H-2D^b antibody (clone KH95), APC-conjugated anti-CD8α antibody (clone 53-6.7), APC-conjugated antibody for H-2K^b bound to SIINFEKL (clone 25D1.16), FITC-conjugated anti-CD69 antibody (clone H1.2F3) and APC-conjugated anti-TCRβ antibody (clone H57-597).

DC2.4 (2 x 10⁵ cells) were cultured with 200 μM PdCl₂ in AIM-V medium (Gibco) at 37°C for 0 15, 30, 60, and 120 min. Cells were then washed with FACS buffer (0.5% BSA, 0.5 mM EDTA, 0.09% NaN₃/PBS) and stained with FITC-anti H-2K^bD^b antibody at 4°C for 20 min. All analyses were performed on FACS Canto II (BD Biosciences) with FlowJo software (TOMY digital biology). In some experiments, prior to staining cells were fixed and permeabilized with BD Fix/Perm Buffer according to the manufacturers’ instruction.

Mice were treated intravenously with 1 mg OVA protein (SIGMA) 50 μg OVA_257-264 (SIINFEKL) (Iwaki Bioservice). Twelve hours after injection, mice were administrated intraperitoneally with 300 μl of 10 mM PdCl₂. Twelve hours after PdCl₂ treatment, spleens were collected and lysed with ACK lysis buffer (155 mM NH₄Cl, 10 mM KHCO₃, 1 mM EDTA), and then washed. Splenocytes were irradiated (20 Gy) for use as antigen presenting cells. For preparation of responder cells, splenocytes was obtained from OT-I mice. These cells were labeled with 5 μM CFSE (Dojindo). Antigen-presenting cells and responder cells were then co-cultured for 48 hours. CFSE dilution in CD8α+ cells were evaluated by flow cytometric analysis to assess proliferation in response to OVA antigen recognition.

DC2.4 cells were pulsed with 5 ng/ml OVA_257-264 (SIINFEKL) (Iwaki Bioservice) in 10% fetal bovine serum (FBS)/RPMI for 1 hour at 26°C and washed with PBS. Then cells were treated with 200 μM PdCl₂ for 2 hours at 26°C. In some experiments, cells were washed and recovered in 10% FBS/RPMI for 1 hour. Finally, cells were stained with antibodies against H-2K^b or 25D1.16.

DC2.4 cells were treated with 25 U/ml recombinant mouse IFN-γ (Peprotech) for 48 hours. Cells were then treated with 100 μM PdCl₂ for 30 min followed by lysis with RIPA buffer (50 mM Tris-HCl (pH 7.4), 150 mM NaCl, 1% Triton X-100, 1% Sodium deoxycholate, 0.1% SDS) supplemented with protease inhibitor cocktail cOmplete™ (Roche) and kept on ice for 30 min. Lysates were centrifugated for 30 min at 4°C. Proteins in supernatant were quantified by MicroBCA protein assay kit (ThermoFicher Scientific). Supernatants were then used for SDS-PAGE, electrotransferred onto polyvinylidene difluoride membranes (Millipore), and membranes probed with the indicated primary antibodies for MHC class I, H-2K^b and H-2D^b (Abcam), followed by HRP-conjugated secondary antibodies. Membranes were then washed, and bands visualized with the enhanced chemiluminescence detection system (ECL) by Chemilumi One L (Nacalai Tesque). Band Intensity was quantified using Image J software.

DC2.4 cells were treated with 1 μg/ml SIINFEKL for 1 hour at 37°C and then cells were cultured in the presence or absence of PdCl₂ for 30 min. Cells were fixed with 1% paraformaldehyde/PBS for 10 min at 4°C, and then permeabilized with 0.1% Triton X100 for 10 min at room temperature. Cells were stained with antibody 25D1.16 prior to analysis with a TCS SP8 microscope (Leica).

DC2.4 cells were treated with and 10 ng/ml LPS with or without 50 μM PdCl₂ (depicted as (+) or (-) PdCl₂ in the figures, respectively) for 18 hours at 37°C. Cells were then washed with PBS and harvested using a cell scraper. Approximately 2 x10⁹ cells were used in the analysis. MHC ligands were isolated and sequenced using mass spectrometry as previously described (25). Briefly, peptide-MHC complexes in samples were captured by affinity chromatography using monoclonal antibodies (Y-3 for K^b and 28-14-8S for D^b). The MHC ligands were eluted, desalted, and injected into LC-MS/MS (Easy-nLC 1000 system and Q Exactive Plus, Thermo). In mass spectrometry, data were acquired with a data-dependent top 10 method. Survey scan spectra were acquired at a resolution of 70,000 at 200 m/z with an AGC target value of 3e6 ions and a maximum IT of 100 ms, ranging from 350 to 2,000 m/z with charge states between 1+ and 4+. MS/MS resolution was 17,500 at 200 m/z with an AGC target value of 1e5 ions and a maximum IT of 120 ms. MS/MS data were searched against the Swiss-Prot database using the Sequest HT along with the Percolator algorithm on the Proteome Discoverer platform (Thermo). The tolerance of precursor and fragment ions was set at 10 ppm and 0.02 Da, respectively, and no specific enzyme was selected for the search. Concatenated target-decoy selection was validated based on q-values, and a false discovery rate (FDR) of 0.05 was used in the Percolator node as a peptide detection threshold. Only the 8-11 mer (K^b) and 8-12 mer (D^b) peptides with IC50 (NetMHC-4.0) < 5,000 were counted as natural MHC ligands. Ligandome analysis were performed twice in each condition and whole peptides in the identified in both experiments were used for analysis. Complete list used in analysis are shown in Supplementary Tables 1 – 4 . Peptide analysis was performed by R venn-diagram package (ver. 3.5.2).

To generate a TCR α and β co-expression plasmid, the C-region of TCRα and β chain were first amplified from cDNA from wild type splenocytes and ligated into the pMX-IP vector (Cell Biolabs). Pd-reactive T cells were collected, RNA extracted using the RNeasy mini kit (QIAGEN), and then cDNA synthesized using Superscript III (ThermoFisher Scientific). The cDNA was then used to generate TCRα and β libraries as described previously (21). Each TCR library was mixed with a forward primer consisting of the annealing site of the adapter DNA (5’- AGCTAGTTAATTAAGGATCCTGATCACCGGACAGGAATTCC -3’) and 20 bps overlapping the pMX-IP vector and a reverse primer specific for the C-region of the TCR α and β (for TCRα 5’- TGGTACACAGCAGGTTCTGGGTTC-3’, for TCRβ 5’- CAAGGAGACCTTGGGTGGAGTCAC-3’). Next, TCR fragments were amplified by PCR. PCR fragments were ligated into the pMX-IP vector digested with BamHI and XhoI (TakaraBio) by mix with NEBuilder HiFi DNA Assembly Master Mix (New England Bio. Lab.) and incubation at 50°C for 15 min. The resulting vectors were transformed into NEB5α competent cells. Transformed cells were spread on LB agar plates and incubated for 16 hours at 37°C. Colonies were then collected, and mixed plasmids were purified using a QIAGEN miniprep kit.

Plat E cells (1 x10⁶ cells) were seeded on 6-well plates, then 1.25 μg of each TCRα and β/pMX-IP plasmid was transfected using polyethylene imine (Polysciences). Two days after transfection, supernatant was collected, and 4 μg/ml of polybrene was added to the supernatant. Then, TG40 cells (2x10⁵ cells) were mixed with supernatant and centrifuged at 1,800 g at 32°C for 60 min and seeded in 24 well plates. Two days after transduction, cells were selected by 1 μg/ml puromycin (Wako). After 6 days of selection, TCRβ-expressing cells were sorted by FACS Aria III (BD Biosciences).

DC2.4 cells (1x10⁴ cells) in 10% FBS/RPMI1640 (Wako) were seeded in 96-well flat bottom plates (FALCON). Six hours after cell seeding, cells were treated with 100 μM PdCl₂ and 25 U/ml recombinant mouse IFN-γ (Peprotech) for 18 hours. Then cells were washed with PBS, and co-cultured with TCR- transduced TG40 (2 x10⁴ cells) for 24 hours. Expression of CD69 on TCRβ positive cells was analyzed by flow cytometry to examine activation of TG40 cells in response to Pd. In some experiments, prior to co-culture with TG40 cells, peptides on MHC class I were stripped as described previously (26).

All data are presented as mean ± S.D. Significance of difference between two groups was determined using unpaired two-sided Student’s t-test ( Figures 4B – E and Supplementary Figure 3C ) and one-way ANOVA with Tukey’s multiple comparison ( Figures 1A, B , 2 , 3D , 4A and Supplementary Figure 1 , 2 , 4 , 5 ). * and ** denote p<0.05 and p<0.01 compared to the control. N.S., not significant. All analyses were performed using Graphpad Prism 8 (GraphPad).

First, we asked whether Pd-treatment affects MHC class I/peptide complexes. To examine this, mice were injected with ovalbumin (OVA) protein and 12 hours after injection, PdCl₂ was injected intraperitoneally. After 12 hours, splenocytes were then analyzed by using antibody clone 25D1.16, which recognizes MHC class I H-2K^b-loading SIINFEKL, an OVA-derived 8-mer peptide. Splenocytes from mice injected with OVA alone exhibited increased 25D1.16 binding as compared with non-treated mice. Interestingly, splenocytes from mice treated with OVA and PdCl₂ exhibited reduced binding of 25D1.16 ( Figure 1A , left panel). In contrast, the total H-2K^b level was comparable regardless of PdCl₂ treatment ( Figure 1A , right panel) with H-2K^b equally upregulated after OVA treatment in the presence and absence of PdCl₂. It has been reported that commercial OVA protein contains LPS (27) and thus, this H-2K^b upregulation is thought to be the result of LPS contamination. Therefore, we performed in vivo experiment using SIINFEKL peptide ( Supplementary Figure 1 ). When we used SIINFEKL peptide, we found that splenocytes from mice treated with SIINFEKL and PdCl₂ exhibited reduced binding of 25D1.16 as compared with group of SIINFEKL alone ( Supplementary Figure 1A , left panel). In contrast, the total H-2K^b level was comparable regardless of PdCl₂ treatment ( Supplementary Figure 1A , right panel). Next, we examined whether reduction of 25D1.16 binding was resulted in attenuation of CD8 T cell recognition. To test this, we used OT-I mice, which are OVA_257-264-specific TCR transgenic mice (28). Antigen-presenting cells (APC) were prepared from mice treated with OVA protein and PdCl₂ as described above. These irradiated APCs were then cocultured with CFSE-labeled OT-I cells. APCs obtained from mice treated with OVA protein alone induced robust OT-I proliferation ( Figure 1B ), while cells obtained from OVA and PdCl₂-treated mice exhibited reduced OT-I proliferation ( Figure 1B ). We also found that reduction of OT-I proliferation by splenocyte of SIINFEKL/PdCl₂-treated mice as compared with that of SIINFEKL alone treated mice ( Supplementary Figure 1 ). These results suggested three possibilities as follows; (1) PdCl₂ treatment reduced OVA peptide presentation on H-2K^b, (2) there is less of SIINFEKL-H-2K^b available for recognition by antibody/OT-I cells or (3) PdCl₂ changed SIINFEKL-H-2K^b (directly or indirectly) whilst SIINFEKL presentation is maintained.

Next, we also performed in vitro analysis using DC2.4 cells, a C57BL/6 (H-2K^b, H-2D^b) mouse-derived dendritic cell line ( Figure 2 ). DC2.4 cells were treated with SIINFEKL peptide, and then cultured in the presence of PdCl₂ for 1 hour or 2 hours. These cells were stained with 25D1.16 and antibody for H-2K^b. Consistent with Figure 1A , SIINFEKL-treated DC2.4 cells exhibited increased binding of 25D1.16 antibody, and this effect was reduced by PdCl₂ treatment for 1 hour or 2 hours ( Figure 2A, B ). However, in contrast with Figure 1A , total H-2K^b expression was also reduced after PdCl₂ treatment ( Figures 2A, B ). In addition, H-2D^b expression, which was not relevant to SIINFEKL loading, also reduced as well as H-2K^b after PdCl₂ treatment ( Figures 2A, B ). To examine the effect on antibody recognition by PdCl₂ treatment, DC2.4 cells were treated with PdCl₂ before SIINFEKL treatment, and then cells were stained with antibodies. As shown in Figure 2C , pretreatment with PdCl₂ did not affect 25D1.16 binding. This data suggested that PdCl₂ treatment did not affect recognition site of these antibodies in our experimental condition. To examine whether the reduction of 25D1.16 binding was the result due to MHC class I down-regulation, Pd-treated DC2.4 cells were cultured in the absence of PdCl₂ for an additional 60 min to aim to recover MHC class I expression. In this condition, we found that 25D1.16 binding was reduced whereas H-2K^b level was comparable regardless cells were pre-cultured with PdCl₂ or not ( Figure 2D ). These data collectively suggested that PdCl₂ treatment alters MHC class I/peptide complexes. Thus, to further analyze the mechanism through which Pd treatment alters peptide presentation by PdCl₂, we used DC2.4 cells.

Next, to examine the mechanism underlying the reduction of 25D1.16 antibody binding in DC2.4 cells as shown in Figure 2 , we focused on loading peptide on MHC class I molecules, H-2K^b and H-2D^b after PdCl₂ treatment. For these experiments we performed MHC class I ligandome analysis (25). As described in the Materials and Methods, DC2.4 cells were treated with PdCl₂ prior to immunoprecipitation of MHC class I (H-2K^b and H-2D^b) and identification of the presented peptides by LC-MS/MS ( Figure 3 and Supplementary Tables 1 – 4 ). Then, we compared peptide list on H-2K^b and H-2D^b between the presence or absence of PdCl₂, respectively (Comparison of “Sequence” column in Supplementary Tables ). When cells were treated with PdCl₂, 501 and 512 peptides were differentially loaded on H-2K^b and H-2D^b as compared with the absence of PdCl₂, respectively ( Figures 3A, B purple area), indicating that PdCl₂ treatment affected alternative peptide loading on H-2K^b and H-2D^b.

Furthermore, we analyzed source proteins from which the PdCl₂ treatment-induced peptides were derived (purple area in Figures 3A, B ). The analysis steps show as follow: (1) Protein name was referenced from peptide sequence by Swissprot. (2) We extracted proteins which was designated by above 501/512 peptides induced by PdCl_2. (3) We examined whether these proteins were found in the list without PdCl₂. We found that 379 and 346 source proteins in H-2K^b and H-2D^b, respectively, were not listed on the absence of PdCl₂, these results indicate that these peptides were derived from unique source protein ( Figure 3C , yellow bar). On the other hand, 109 and 164 proteins bound to H-2K^b and H-2D^b, respectively, were shared between presence and absence of PdCl₂ ( Figure 3C , red bar).

To demonstrate the importance of differential peptide loading with T cell stimulation, we performed a peptide stripping experiment as described previously, with some modifications (26). First, the effect of peptide stripping was examined using antibody 25D1.16 and OVA₂₅₇-pulsed DC2.4 cells ( Supplementary Figure 2 ). Binding of antibody 25D1.16 to SIINFEKL-pulsed DC2.4 cells was reduced by peptide stripping ( Supplementary Figure 2A ). Furthermore, CD69 expression on OT-I TCR-expressing TG40 cells was also reduced by coculture with peptide-stripped cells ( Supplementary Figure 2B ).

To examine whether the differential peptide presentation caused by PdCl₂ induces activation of Pd-responsive T cells, Pd-responsive T cells were established using LNs from mice sensitized and challenged with PdCl₂ as reported previously, with some modifications (24). After continuous and low dose Pd treatment, oligoclonal T cells were successfully established ( Supplementary Figures 3A, B ). In these cells, we found that highest frequent of TCR repertoire was TRAV16D/DV11-3-CAMRAYANKMIF-TRAJ47-03 and TRBV13-3-01-CASSDRTTNSDYTF-TRBJ1-2-01. Then RNA obtained from these cells was used to construct a TCR expression vector. These expression plasmids were then retrovirally transduced into TG40 cells (T cell hybridoma cell line lacking TCRα and TCRβ) and after selection with puromycin, TCRβ-expressing TG40 cells were sorted. Following co-culture with PdCl₂-pretreated DC2.4 for 24 hours, CD69 expression was analyzed as a marker of TG40 activation. We found that TG40 cells expressing the TCR library obtained from Pd-sensitized mice upregulated CD69 expression in response to PdCl₂-treated DC2.4 cells ( Figure 3D ). However, TG40 cells which express TCR library from lymph node of naïve mice did not respond to PdCl₂-treated DC2.4 cells ( Supplementary Figure 3C ).

Thus, we examined the effect of peptide stripping after PdCl₂ treatment and co-culture with Pd-responsive T cells. We confirmed that CD69 expression on Pd-responsive TG40 cells was reduced following co-culture with peptide stripped PdCl₂-treated DC2.4 cells ( Figure 3D , right 2 columns). These results suggest that the reason for the reduction in 25D1.16 binding following PdCl₂ treatment is due to alternative peptide loading on MHC class I.

As shown in Figure 2A , Pd-treatment temporally reduces cell-surface expression of H-2K^b. To explore the behavior of MHC class I in response to PdCl₂ treatment, we analyzed MHC class I expression on the cell surface in response to PdCl₂ treatment. DC2.4 cells were treated with PdCl₂ and cultured for the indicated times (0, 15, 30, 60, and 120 min), and subsequent staining for surface MHC class I. PdCl₂ treatment resulted in a reduction of surface MHC class I (H-2K^bD^b) within 15 min after which surface expression gradually recovered, albeit only partially, until 120 min ( Figure 4A ). Next, we assessed whether the reduction in MHC was due to degradation or internalization. To examine the level of MHC class I in PdCl₂-treated cells under permeabilizing conditions, DC2.4 cells were treated with PdCl₂ for 30 min, then cells were permeabilized and stained with antibodies for MHC class I. While surface MHC class I was reduced in the presence of PdCl₂ under non-permeabilizing conditions, total MHC expression in cells treated with PdCl₂ was comparable with that of non-treated cells ( Figure 4B ). In addition, western blot analysis confirmed that MHC class I levels of PdCl₂-treated cells are comparable with that of non-treated cells ( Figure 4C ). These data suggest that PdCl₂ treatment induces MHC class I internalization, and not degradation. Next, we followed the movement of MHC class I after PdCl₂ treatment. To this end, we used the 25D1.16 antibody to stain cells treated with SIINFEKL peptide prior to fixation and permeabilization. As shown in Figure 4D , surface staining of 25D1.16 was observed regardless of PdCl₂ treatment, but only following PdCl₂ treatment was staining also observed around the perinuclear area ( Figure 4D ), which has been reported as the recycling center (29). These results indicated that PdCl₂ treatment induced temporal internalization of MHC class I.

Finally, to examine the significance of MHC class I internalization on alternative peptide loading, DC2.4 cells were treated with PdCl₂ and LPS in the presence of sodium azide (NaN₃), which inhibits internalization of surface molecules. We found that NaN₃ treatment partially suppresses MHC class I internalization in response to PdCl₂ treatment ( Supplementary Figure 4 ). As a control experiment, we examined whether NaN₃ treatment affect antigenicity of exogenously added antigen. To this end, NaN₃-treated DC2.4 cells were pulsed with SIINFEKL, and then cells were co-cultured with OT-I TCR/TG40 and CD69 on TG40 cells was analyzed. As shown in Supplementary Figure 5 , activation of OT-1/TG40 by NaN₃-treated DC2.4 that had been exogenously added with SIINFEKL was comparable with non-NaN₃ treated SIINFEKL-loading DC2.4 cells. The result suggested that NaN₃ treatment affected inhibition of membrane movement rather than cell metabolism. Then, to examine the antigenicity of PdCl₂-treated DC2.4 cells in inhibition of membrane movement, DC2.4 cells were treated with or without PdCl₂ in the presence NaN₃, and then these cells were co-cultured with Pd-responsive TG40 cells. We found that upregulation of CD69 expression on Pd-responsive TG40 cells, as shown in Figure 3D , is suppressed by NaN₃-treated DC2.4 cells ( Figure 4E ). Thus, MHC class I internalization by PdCl₂ is crucial for alternative peptide loading on MHC class I.