The TRIP12 complementary DNA (cDNA) (NM_001284214.2) was a generous gift from Jiahuai Han’s lab (Xiamen University, China). TRIP12 cDNA was cloned into pCMV-flag or pCMV-myc vector. TRIP12 mutants were generated by Fast Mutagenesis Kit (Vazyme, Nanjing, China) following the manufacturer’s instructions. The pENTER-NFATc1-flag plasmid was purchased from Vigene Biosciences (Shandong, China) and NFATc1 (NM_172390.3) was subcloned into pCMV-myc vector. The pCMV-ubiquitin (Ub)-HA, pcDNA3.1-WARS-flag, pcDNA3.1-SIRT1/2/3/4/5/6/7-HA plasmids were constructed previously in our laboratory. Tryptophanyl-tRNA synthetase (WARS) (NM_173701.2) cDNA was cloned into pCMV-flag or pCMV-myc vector to obtain WARS expression plasmids. The sgRNAs for WARS/TRIP12 KO were inserted into pSpCas9(BB)-2A-Puro vector and the sgRNAs for TRIP12 knockin were inserted into pSpCas9(BB)-2A-GFP. All constructs were confirmed by sequencing.

HEK293T, Jurkat T, CT26, Lewis lung carcinoma (LLC), and B16F10 cell lines were originally purchased from the Cell Bank of the Chinese Academy of Sciences.

HEK293T, LLC, and CT26 cells were cultured in complete Dulbecco’s Modified Eagle Medium medium (Hyclone, Logan, Utah, USA) supplemented with 10% FBS (Bioind, Israel) and antibiotics (100 µg/mL streptomycin and 100 units/mL penicillin, Gibco, Paisley, UK), Jurkat T cells were cultured in RPMI 1640 medium (Hyclone) containing 10% FBS and streptomycin/penicillin. Mouse CD8⁺ T cells were cultured in complete RPMI 1640 medium containing 10% FBS, antibiotics and interleukin (IL)-2 (10 ng/mL). All cells were maintained at 37℃ in humidified 5% CO₂.

Plasmids were transfected into adherent cells using polyethylenemine (PEI) (1 μg/μL, Polysciences, Warrington, USA), small interfering RNAs (siRNAs) were transfected into adherent cells by HilyMax (Dojindo Molecular Technologies, Kumamoto, Japan). Transfection of Jurkat T was performed through electroporation by Lonza Amaxa Nucleofector 2B (Lonza, Basel, Switzerland) with program S-018.

Cell culture plates were pre-coated with anti-CD3 and anti-CD28 antibodies at 4°C overnight, and then washed 3 times with RPMI 1640 medium. Jurkat T or mouse-naïve CD8⁺ T cells were activated with the pre-coated plates for 2 days.

Proliferation was measured by flow cytometry using carboxyfluorescein diacetate succinimidyl ester (CFSE) (eBioscience). Briefly, Jurkat cell suspensions (1×10⁷ cells) were prepared and labeled with 0.5 µM CFSE at 37°C for 10 min, washed with RPMI 1640 medium containing 50% FBS, and then resuspended in pre-warmed cell culture medium containing 10% FBS. Forty-eight hours later, dividing cells were detected by flow cytometry and cell proliferation was quantified by CFSE dilution.

Total RNA for reverse transcription quantitative PCR (RT-qPCR) was extracted using Trizol reagent. Following the manufacturer’s instruction, RT was performed with PrimeScript RT reagent kit (Takara, Tokyo, Japan) and qPCR was performed using AceQ qPCR SYBR Green Master Mix (Vazyme). Quantitative analysis of qPCR was carried out by 2^−△△Ct method used actin as the internal reference indicator.

Cells were lysed with sodium dodecyl sulfate (SDS) protein sample loading buffer or lysed in 0.5% NP-40 lysis buffer (50 mmol/L Tris-HCl, 150 mmol/L NaCl, 0.5% NP-40, pH=7.2–7.4, protease inhibitors cocktail tablets (Roche, Indianapolis, Indiana, USA)) at 4℃ for 30 min. Protein was separated using SDS-PAGE in appropriate gel concentration by electrophoresis according to the molecular weight of the target protein, and then transferred onto nitrocellulose (NC) membranes. Membranes were blocked with 5% non-fat milk in triethanolamine buffered saline (TBS) containing 0.05% Tween-20, and incubated overnight at 4℃ with primary antibodies (actin, 1:10,000; NFATc1, 1:200; TRIP12, 1:1000; WARS, 1:1000; IDO1, 1:1000). The blots were washed, incubated in peroxidase-coupled secondary antibodies against rabbit or mouse IgG (1:5000) at room temperature, then washed and developed using Super Signal Substrate (CLiNX, Shanghai, China). The Myc-tagged or Flag-tagged or HA-tagged proteins were detected with HRP-labeled primary antibodies (anti-HA-HRP, 1:2000; anti-flag-HRP, 1:2000; anti-myc-HRP, 1:2000). Signal intensity was visualized by ChemiScope 3300 Mini Imager Systems (CLiNX).

The ELISA kit for testing TNF-α, IFN-γ was purchased from Biolegend. Cell culture supernatants were collected and the cytokines were assessed according to the manufacturer’s instructions. Capture antibody solution was added to microplate wells and incubated overnight (16–18 hours) between 2℃ and 8℃. Assay diluent buffer was used to block non-specific binding. Serially diluted protein standard or test sample were incubated for 2 hours at room temperature with shaking. The trapped antibody was then added and detected by substrate solution. Optical density of each well was determined within 20 min using a microplate reader set to 450 nm.

The ELISA kit for ovalbumin (OVA)-sIgG was purchased from Jining Industrial (Shanghai, China). Blood samples were collected from the orbital venous plexus of mouse, and were left to stand for 1 hour at room temperature, then centrifuged at 1500 rpm for 20 min to separate serum. The OVA-sIgG was measured according to the manufacturer’s instructions.

Beads used in the immunoprecipitation were washed and equilibrated in 0.1% NP-40 lysis buffer. Briefly, cell lysates were prepared using immunoprecipitation assay buffer (50 mmol/L Tris-HCl, 150 mmol/L NaCl, 0.1% NP-40, pH=7.2–7.4, protease inhibitors cocktail tablets (Roche)) for 30 min at 4℃, then incubated with anti-flag beads (Sigma) to enrich the target protein overnight at 4℃. For immunoprecipitaion of TRIP12-myc protein, the cell lysates were first incubated with anti-myc antibody overnight at 4℃, followed by 4 hours incubation with protein A/G plus-agarose beads. The beads were then washed vigorously with immunoprecipitation lysis buffer for 6–8 times. The protein was eluted by boiling the beads with SDS loading buffer for 5 min, and then subjected to western blot analysis.

Plasmids ubiquitin-HA, NFATc1-flag, and TRIP12-WT-myc, or TRIP12-K1136R/W-myc, or TRIP12-ΔHECT-myc were co-transfected into HEK293T cells. The HEK293T cells were treated with 20 µM MG-132 for 6 hours before harvest. NFATc1 was purified by co-immunoprecipitation using anti-flag-conjugated agarose beads (Sigma), followed by western blot analysis. To examine the effect of WARS on ubiquitination of NFATc1, siWARS were co-transfected with ubiquitin-HA, NFATc1-flag, and TRIP12-WT-myc in HEK293T cells. Alternatively, WARS were reintroduced into HEK293T-WARS-KO cells by co-transfection with ubiquitin-HA, NFATc1-flag, and TRIP12-WT-myc.

Female C57BL/6 mice aged 4 weeks were purchased from SLAC Laboratory Animal (Shanghai, China). The mice were housed in a standard environment and fed with a standard diet during a 1-week adaptation period. The LLC cells or B16F10 cells of logarithmic phase were harvested, and single cell suspensions were prepared. Healthy female C57BL/6 mice, weighing 18–20 g, were subcutaneously inoculated in the left axillae with 5×10⁵ LLC cells or B16F10 cells. The day after inoculation, the mice were randomly divided into four groups, 10 mice in each group. For LLC-inoculated mice, the Me-Trp group received an intraperitoneal injection of Me-tryptophan solution (≈150 mg/kg, dissolved in PBS), and indoximod group received an intraperitoneal injection of indoximod suspension liquid (≈100 mg/kg, dissolved in NaOH solution, pH adjusted to 7.2–7.4). The Me-Trp and indoximod group were intraperitoneally injected with both Me-Trp (≈150 mg/kg, dissolved in PBS) and indoximod (≈100 mg/kg). The corresponding control mice received an equivalent volume of PBS. All the injections were performed twice per day (09:00 and 17:00 hours). For B16F10-inoculated mice, Me-Trp and control group were the same as that in LLC-inoculated mice. Anti-PD-1 (10 mg/kg) were administered by intraperitoneal injections on days 3, 6, 9, and 12 after inoculation in anti-PD-1 group. The anti-PD-1 and Me-Trp group was injected with both anti-PD-1 and Me-Trp. Tumor sizes were measured 2 or 3 times a week by randomizing the order, with each mouse tested within an hour on the same day. The tumor volume was calculated by the following formula: tumor volume=0.5×tumor length×tumor width×tumor width. Mice were excluded if they died or if their tumors ruptured, or the tumor volume exceeded 1 cm³. The tumor-bearing mice were sacrificed by cervical dislocation. Tumors and spleen were carefully removed, and tumors were weighed and photographed.

Tumors removed from sacrificed mice were fixed for 3 days with 4% paraformaldehyde at 4℃, and then placed in 30% sucrose solution for 2 days for dehydration. The samples and cryostat microtome were prechilled to −20℃, and then tumor sections (45 µm) were cut on a cryostat microtome, blocked with 5% goat serums in PBS containing 0.1% Triton X-100 for 2 hours, and then incubated with primary antibodies (CD4, 1:150; CD8, 1:150) in blocking buffer overnight. Subsequent to the excess primary antibody being washed off, sections were incubated with Alexa Fluor 647-labeled goat antirat IgG (Jackson ImmunoResearch, 1:200) and DAPI for 3 hours. The sections were sealed with antifluorescence quencher. After standing for a few days, fluorescence was visualized and images along the z-axis (z-axis interval: 2 µm) were captured by Olympus FV3000 inverted confocal microscope (Centre Valley, Pennsylvania, USA) equipped with a 40× objective lens. The cell number of positive signals per cubic millimeter was calculated.

For specific cytotoxicity assays, CD8⁺ T cells isolated from OVA-immuned BALB/c mice were used as effector cells. Emulsified OVA protein was injected into BALB/c mice by three subcutaneous injections at 1-week or 2-week intervals. CD8⁺ T cells were obtained from spleens by flow cytometric sorting with surface markers CD3 and CD8. Then CD8⁺ T cells were treated with tryptophan in different concentrations. Target mouse cancer cell lines LLC and CT26, which were placed in the U-type 96-well plate in advance, were soaked with 10 µM OVA peptide (SIINFEKL) at 37℃ for 1 hour. Excess OVA peptide was washed away by RPMI 1640 medium, then CD8⁺ T cells were added and co-cultured with target cancer cells at E:T ratios of 10:1. Subsequently, floating CD8⁺ T cells from the supernatant were washed off by PBS, the remaining adherent cancer cells were digested by 0.25% trypsin, washed by PBS for 3 times, and stained using an annexin V-FITC/propidium iodide apoptosis kit (TaKaRa, Dalian, China). Apoptosis of target cancer cells was measured with Beckman Coulter Gallios flow cytometer to reflect the cytotoxicity ability of CD8⁺ T cells. A CD3 staining was used to distinguish CD3⁻ cancer cells and the remaining CD3⁺ CD8⁺ T cells.

For adoptive T-cell transfer, CD8⁺ T cells were isolated from spleens and lymph nodes of ovalbumin-transgenic (OT-1) mice (kindly provided by Dr Xiaofei Yu of Fudan University) with mouse CD8⁺ T cells isolation kit (Biolegend, San Diego, California, USA), and then transferred into recipient C57BL/6 WT mice (2×10⁶ cells/mouse) by intravenous injection. The next day, each mouse was immunized by subcutaneous multiple-point injections with 20 µg OVA (Sigma-Aldrich) diluted in 100 µL PBS that mixed with 100 µL complete Freund’s adjuvant (Sigma-Aldrich). Me-Trp (150 mg/kg) or indoximod (100 mg/kg) was administered by intraperitoneal injections twice a day. NFAT inhibitor VIVIT peptide (10 mg/kg) (MCE, Monmouth Junction, New Jersey, USA) was intraperitoneally injected once a day. All the treatments were started 1 day before adoptive transfer (day 0) until sacrifice.

Day 8 after adoptive transfer, splenocytes were obtained and stain with zombie aqua (Biolegend), anti-CD8 (Biolegend), anti-TCR-Va2 (Biolegend), anti-KLRG1 (Biolegend), and anti-CD127 (Biolegend). Short-lived effector CD8⁺ T cells and memory precursor effector cells were distinguished by KLRG1 and CD127 staining.

Whole-body SIRT1 knockout (SIRT1-KO) mice and the corresponding control mice were were generated by the Cre/loxP recombination system. SIRT1-KO mice (Cre⁺ SIRT1^flox/flox) were generated by crossing homologous SIRT1^flox/flox (Southern Model Organisms, Shanghai, China) with UBC-Cre-ERT2 mice (Southern Model Organisms). The control mice (wild type (WT)) were SIRT1^flox/flox (Cre negative). Activation of Cre recombinase was induced by intraperitoneal injection of tamoxifen (Sigma). Tamoxifen was dissolved in corn oil containing 10% ethanol (20 mg/mL) at 37℃, stored at −20℃ and protected from light. Each mouse was administered 100 µL of tamoxifen solution per day for seven consecutive days by intraperitoneal injection. Mice were tested after a recovery period of at least 1 week after the final tamoxifen injection to permit deletion of the SIRT1 gene and degradation of residual SIRT1 and SIRT1 messenger RNA (mRNA). PCR and western blot analysis were used to confirm the efficiency of Cre-loxP-mediated SIRT1 deletion in splenic tissue. The DNA was extracted from splenic tissue using Genomic DNA Extraction Kit (TIANGEN, Shanghai, China) according to the manufacturer’s protocol, and then PCR was performed to identify mice with Cre recombinase and SIRT1 loxP sites.

Oligonucleotides

siRNA TRIP12-A: GACAAAGACUCAUACAAUAUU

siRNA TRIP12-B: GAACACAGAUGGUGCGAUAUU

siRNA NFATc1-A: CCCGUUCACGUCAGUUUCUACGUCU

siRNA NFATc1-B: CGAGCCGUCAUUGACUGUGCCGGAA

siRNA WARS-A: CCAGGAUCCUUACUUUAGAdTdT

siRNA WARS-B: GGUUCUUGAUGCCUAUGAAdTdT

CRISPR/Cas9-mediated TRIP12 gene KO sgRNA-A: CACCGCAACCACAAGACGACTCAAT

CRISPR/Cas9-mediated TRIP12 gene KO sgRNA-B: CACCGCTGGAACAACACTCTTGAC

CRISPR/Cas9-mediated WARS gene KO sgRNA: CACCGAACTGCCCAGCGTGACCAG

CRISPR/Cas9-mediated TRIP12-K1136R gene knockin sgRNA-A: CACCGTAAAGGTTGGATTAAGGAGC

CRISPR/Cas9-mediated TRIP12-K1136R gene knockin sgRNA-B: CACCGTCTTGTAGAGAAAAAATTAA

Q-PCR NFATc1-F: AAAACTGACCGGGACCTGTG

Q-PCR NFATc1-R: GCTCATGTTCACGGCTTACG

Q-PCR PD-1-F: CCAGGATGGTTCTTAGACTCCC

Q-PCR PD-1-R: TTTAGCACGAAGCTCTCCGAT

Q-PCR actin-F: CCTCTCCCAAGTCCACACAG

Q-PCR actin-R: GGGCACGAAGGCTCATCATT

Statistical analyses and graphs were undertaken using GraphPad Prism V.8.0 (GraphPad Software, San Diego, USA). The mean values±SD of independent experiments triplicates and every data point were shown in graph. Unpaired two-sided Student’s t-test was used to reflect the difference between two groups.

Results infigures 1B, online supplemental figure S1A–H, S2B-D, S2F, S2G, S3A, S3C-G, S3J, S4G-H, S6A–C, S6E, 7A–B, 7I are representative data of three independent repeats. And there were similar results in three independent repeats.

In immortalized anti-CD3 and anti-CD28 antibodies co-stimulated Jurkat cells (online supplemental figure S1A), L-methyl-tryptophan ester (Me-Trp) administration in culture media at 73.5 µM increased intracellular tryptophan by 446.1% (figure 1A), and decreased the cell surface PD-1 by 37.00% (figure 1B). In contrast, depriving the culture media of tryptophan upregulated PD-1 by 50.93% (figure 1C, online supplemental figure S1B). These results suggest that tryptophan or its catabolites negatively regulate PD-1 expression in Jurkat cells. Since tryptophan can modify lysine residues in proteins and regulate protein functions,19 we investigated whether tryptophan inhibited PD-1 expression through lysine tryptophanylation. Overexpression of WARS, the proven lysine tryptophanyl transferase,19 phenocopied tryptophan to decrease cell surface PD-1 in Jurkat T cells (figure 1D, online supplemental figure S1C). Conversely, silencing WARS with siRNAs increased surface PD-1 level (figure 1E, online supplemental figure S1D). Administration of cells with general aminoacylation inhibitor pyrophosphate19 20 resulted in PD-1 increasing on Me-Trp supplemental (figure 1F, online supplemental figure S1E) and Trp deprivation (figure 1G, online supplemental figure S1F). These findings suggest that WARS-mediated lysine tryptophanylation likely repressed PD-1 expression.

Me-Trp decreased anti-CD3 and anti-CD28 antibodies activated the cell surface PD-1 in mouse CD8⁺ T cells (figure 1H, online supplemental figure S1G and S1H. Notably, when WARS was silenced by siRNAs, the PD-1 altering potencies of Me-Trp administration (figure 1B) and tryptophan deprivation (figure 1C and online supplemental figure 1B) on surface PD-1 expression were reduced, further supporting that the effects of Trp on PD-1 are mediated by WARS.

The fact that T lymphocytes are extremely sensitive to tryptophan shortage and IDO sensitizes T cells to apoptosis14 promoted us to examine T cells such as T lymphocyte Jurkat cells express IDO1, as previous studies obtained negative results.22 23 After activated with anti-CD3 and anti-CD28 antibodies, the levels of IDO1 mRNA increased by >80 folds and the expression of IDO1 was detectable by western blot analysis (online supplemental figure S2A), consistent with a report that Jurkat cells actually expressed IDO124 and explained, at least partially, that T cells sensitive to tryptophan levels. As PD-1 expression is restricted to immune cells.25 We examined a hypothesis that IDO inhibitors may decrease PD-1 expression by elevating intracellular tryptophan levels. Using the IDO inhibitors, namely indoximod, epacadostat, and NLG919 analog at appropriate concentrations increased tryptophan levels and left kynurenine levels negligibly affected in Jurkat T cells (figure 2A), phenocopied Me-Trp supplementation (figure 1B) to decrease the cell surface PD-1 expression in the anti-CD3-activated and anti-CD28-activated Jurkat T cells (figure 2B, online supplemental figure S2B). These, together with that kynurenine had a negligible effect on surface PD-1 expression in Jurkat T cells (figure 2C and online supplemental figure S2C), suggested that tryptophan rather than its catabolites, negatively regulates surface PD-1 expression of Jurkat T cells. Moreover, indoximod has previously been shown to affect T cell function through affecting mammalian target of rapamycin complex 1 (mTORC1) signaling.26 We tested whether indoximod effects on PD-1 can be independent of mTORC1. Indoximod still able to regulate PD-1 expression (online supplemental figure S2D) when mTORC1 signaling was inhibited by rapamycin in Jurkat cells (online supplemental figure S2E), suggested indoximod may regulate PD-1 in mTORC1-independent mechanisms.

The cell surface PD-1 expression decreasing effects of IDO inhibitors were recaptured in naïve CD8⁺ T cells, isolated from the spleens of BALB/c mice, that were co-treated with anti-CD3 and anti-CD28 antibodies for 48 hours (figure 2D and online supplemental figure S2F). These results suggest that tryptophan decreases surface PD-1 levels through a mechanism other than tryptophan catabolism.

IDO inhibitors increased the levels of IL-2, IFN-γ, and TNF-α, indicators of T cell activation,27 28 in activated Jurkat cells (figure 2E and online supplemental figure S2G) and in mouse CD8⁺ T cells (figure 2F). These results were consistent with the decrease in PD-1 expression by immunosuppressive tryptophan catabolites.

RT-qPCR revealed that the mRNA level of PD-1 was decreased by both Me-Trp and IDO inhibitors treatments in Jurkat cells (figure 3A), suggesting that Me-Trp and IDO inhibitors decrease PD-1 expression level, at least in part, by altering PD-1 transcription. We tried to identify transcription factors of PD-1 that may be regulated by Me-Trp and IDO inhibitors. Protein levels (figure 3B), but not mRNA levels (figure 3C), of NFATc1, a reported transcriptional factor for PD-129 that was confirmed by us (online supplemental figure S3A), were downregulated by Me-Trp supplementation in the culture media for activated Jurkat T cells. Furthermore, tryptophan deprivation of culture media of activated Jurkat T cells increased the protein levels (figure 3D), but not mRNA levels (figure 3E), of NFATc1 of Jurkat cells. These, together with that IDO inhibitors decreased NFATc1 (online supplemental figure S3B) and the cell surface PD-1 expression (see figure 2B), indicate that NFATc1 is regulated by tryptophan and IDO inhibitors, for decreasing cell surface PD-1.

Notably, WARS overexpression, a manipulation that may increase protein tryptophanylation,19 decreased NFATc1 protein levels (figure 3F), but not NFATc1 mRNA levels (figure 3G), of Jurkat cells. Moreover, WARS knockdown by siRNAs increased NFATc1 protein levels (figure 3H), but not NFATc1 mRNA levels (figure 3I), of Jurkat cells. These results suggest that tryptophanylation regulates NFATc1 at post-transcription levels. Furthermore, NFATc1 knockdown, which decreased surface PD-1 expression of Jurkat cells (online supplemental figure S3A), abrogated Me-Trp supplementation (figure 3J, online supplemental figure S3C) to decrease surface PD-1 in Jurkat T cells, indicating that tryptophanylation regulates surface PD-1 expression through NFATc1. These, together with that NFATc1 knockdown abrogated WARS overexpression and IDO inhibitors to decrease surface PD-1 in Jurkat T cells (figure 1D and online supplemental figure 1C figure 2B and online supplemental figure 2B), confirmed that decreased NFATc1 is the major reason for these manipulations to downregulate surface PD-1 expression.

Compared with that inhibiting NFAT with VIVIT peptide,30 31 which had no effect on T cell proliferation (online supplemental figure S3D), inhibited IL-2, TNF-α, and IFN-γ production (online supplemental figure S3F), tryptophan supplemental increased IL-2, TNF-α, and IFN-γ production and proliferation (online supplemental figure S3E, S3G) in Jurkat cells. The differences between NFAT inactivation and tryptophan treatment was likely due to that besides decreasing NFATc1 and PD-1 expression to activate T cells, tryptophan supplemental may alter IDO-mediated immunosuppression, which was consistent with the finding that elevation in circulating tryptophan levels overcome IDO1-mediated immunosuppression in the tumor, along with the IFN-γ mRNA increasing.32 However, as activated T cell express high levels of IL-2, IFN-γ, and TNF-α,27 28 which were confirmed in Jurkat (figure 2E and online supplemental figure S2G) and in mouse CD8⁺ T cells (figure 2F), the increased cytokines in IDO inhibitors may be consequences of T cell activation, a notion that was supported by the fact that both indoximod and tryptophan increased cytokine production (online supplemental figure S3H and S3I) and increased the number of short-lived effector cells (KLRG⁺CD127⁻⁾ in C57B6/L mice adoptively transferred with CD8⁺ isolated from OT-1 mice (online supplemental figure S3H and S3J).

Proteasome inhibitor MG132 treatment on anti-CD3 and anti-CD28 antibodies co-stimulated Jurkat T cells resulted in elevated NFATc1 protein levels (figure 4A), suggesting that NFATc1 is subject to proteasomal degradation. Moreover, as MG132 supplementation blocked tryptophan supplementation (online supplemental figure S4A), WARS overexpression (online supplemental figure S4B) and IDO inhibitors (online supplemental figure S4C) decreased NFATc1 levels in Jurkat T cells, suggesting that tryptophanylation regulates NFATc1 levels by altering its proteasomal degradation.

Searching the tryptic peptide library of human liver cancer tissue revealed that lysine 1136 of TRIP12, an E3 ligase, was tryptophanylated (figure 4B), suggesting that TRIP12 may be regulated by tryptophanylation. TRIP12 co-immunoprecipitated with NFATc1, when they were co-expressed in HEK293T cells (figure 4C,D). The half-life of endogenous NFATc1 was prolonged by knockdown of TRIP12 (figure 4E). Deleting TRIP12 from HEK293T cells (online supplemental figure S4D) decreased the ubiquitination of ectopically expressed NFATc1 (figure 4F). Knockdown of TRIP12 in Jurkat T cells increased endogenous levels of NFATc1 protein (figure 4E). Furthermore, overexpressing TRIP12, but not a catalytically dead TRIP12 with its HECT domain deleted, increased the ubiquitination of NFATc1 (figure 4G) in HEK293T cells, consistent with reduced expression of NFATc1 (online supplemental figure 4E). Furthermore, TRIP12 KO increased NFATc1 protein levels in HEK293T cells (online supplemental figure S4F). Moreover, knockdown of TRIP12 increased PD-1 expression in Jurkat cells (figure 4H and online supplemental figure S4G). These results collectively suggest that TRIP12 is an E3 ligase of NFATc1, which was further substantiated by the knockdown of TRIP12 in Jurkat T cells elevating cell PD-1 mRNA (figure 4I) and surface PD-1 protein (figure 4J and online supplemental figure S4H), whose levels are positively regulated by NFATc1.

Co-immunoprecipitation revealed that WARS interacted with TRIP12 (online supplemental figure S5A). Moreover, on probing with an antitryptophanylation antibody (online supplemental figures S5B and S5C), WARS overexpression was found to increase tryptophanylation of TRIP12, but not a mutant TRIP12 with K1136 switched to arginine (TRIP12^K1136R), which had very low tryptophanylation signal (figure 5A), suggesting that K1136 is the main tryptophanylation site of TRIP12, and allowed us to probe TRIP12 tryptophanylation with the pan-tryptophanylation antibody. Conversely, taking the advantage that HEK293T cells WARS KO (online supplemental figure S5D) was viable as KO other tRNA synthetases such as methionine tRNA synthetase and histidyl-tRNA synthetase 2, we were able to observe that WARS KO resulted in decreased NFATc1 ubiquitination (figure 5B) and increased NFATc1 protein levels in HEK293T cells (figure 5C). Tryptamine, a WARS inhibitor,33 which lowered TRIP12 tryptophanylation levels (online supplemental figure S5E), also decreased NFATc1 ubiquitination in HEK293T cells (online supplemental figure S5F) and increased NFATc1 protein levels in Jurkat T cells (online supplemental figure S5G). These results collectively support that TRIP12 lysine1136 tryptophanylation (Trp-K1136) activated its E3 activity toward NFATc1.

The levels of NFATc1 in HEK293T cells, but not in TRIP12^K1136R knockin HEK293T cells (online supplemental figure S5H), were reduced by tryptophan supplementation (figure 5D) or WARS overexpression (figure 5E), suggesting that Trp-K1136 is the major underlying mechanism connecting tryptophan and WARS to PD-1 expression. K1136 is located between the WWE and HECT domains of TRIP12 (online supplemental figure S5I). We, therefore, hypothesized that K1136Trp enhanced the ability of TRIP12 to degrade NFATc1 by enhancing the ability of TRIP12 to recruit NFATc1. Recombinant tryptophanylation-mimetic TRIP12, which switch lysine 1136 of TRIP12 into tryptophan (TRIP12^K1136W), had higher affinity than the WT TRIP12 to bind NFATc1 (figure 5F). Moreover, the affinity between TRIP12 and WARS was enhanced by Me-Trp supplementation (figure 5G), and the affinity between TRIP12 and NFATc1 was decreased by tryptophan deprivation (figure 5H), consistent with Me-Trp supplementation resulting in a dose-dependent increase in ubiquitination of NFATc1 (figure 5I), and tryptophan starvation leading to lower ubiquitination of NFATc1 (figure 5J). WARS knockdown decreased the affinity between TRIP12 and NFATc1 (figure 5K), and resulted in lower ubiquitination of NFATc1 (figure 5L). Conversely, WARS overexpression in HEK293T-WARS-KO cells increased the affinity between TRIP12 and NFATc1, but not between TRIP12^K1136R and NFATc1 (figure 5M), and resulted in higher ubiquitination of NFATc1 (figure 5N). These results showed that K1136Trp positively regulated the E3 activity of TRIP12 toward NFATc1 by increasing TRIP12-NFATc1 interaction.

We further investigated whether K1136Trp is dynamically regulated. NAD⁺-dependent sirtuins (SIRTs), namely SIRT1-SIRT7,34 can function as deaminoacylases.19 Pan-sirtuin inhibitor NAM decreased the cell surface PD-1 and PD-1 mRNA levels in anti-CD3 and anti-CD28 co-stimulated Jurkat T cells (figure 6A,B and online supplemental figure S6A. The cell surface PD-1 decreasing effects of NAM were recaptured in stimulated mouse CD8⁺ T cells (figure 6C and online supplemental figure S6B). These results supported the possibility that K1136Trp is reversed by SIRTs. To pinpoint which SIRT possesses de-K1136Trp activity, we screened the abilities of SIRTs to reverse K1136Trp. When all SIRTs, namely SIRT1-7, were individually co-expressed with TRIP12 in HEK293T cells, SIRT1, SIRT6, and SIRT7 showed interaction with TRIP12 (figure 6D). However, only SIRT1 overexpression significantly decreased the tryptophanylation of isolated TRIP12 (figure 6E), and only SIRT1 overexpression clearly increased the surface PD-1 expression in Jurkat T cells (figure 6F and online supplemental figure S6C). These results suggested that SIRT1 is a de-tryptophanylase of K1136Trp. This notion was confirmed in SIRT1 KO C57BL/6 mice, in which decreased PD-1 expression in CD8⁺ T cells was observed along with abrogated NAM ability to decreased surface PD-1 expression in CD8⁺ T cells (figure 6G and online supplemental figure S6D, S6E). These results also suggest that SIRT1 is the major de-tryptophanylase of K1136Trp that regulates surface PD-1 expression.

Tryptamine, an analog of tryptophan that competitively inhibits WARS,33 reduced K1136Trp (online supplemental figure S5E) and induced the expression of cell surface PD-1 (figure 7A, online supplemental figure S7A) and PD-1 mRNA (figure 7B) in Jurkat T cells and induced surface PD-1 expression in stimulated mouse CD8⁺ T cells (figure 7C, online supplemental figure S7B), consistent with potentiation of T cells by tryptophanylation.

As CD8⁺ T cells are the major effector cells in eliminating cancer cells,35 we examined whether tryptophan treatment could increase the cytotoxic activity of CD8⁺ T cells, since tryptophan (figure 1H) or IDO inhibitor (figure 2D) treatment decreased the expression of PD-1 in mouse CD8⁺ T cells. CD8⁺ T cells were isolated from OVA-stimulated mouse according to a previously reported protocol.36 CD8⁺ T cells were sorted from the OVA-immunized BALB/c splenocytes by surface CD3 and CD8 (online supplemental figure S7C and S7D). The ability of tryptophan to increase the toxicity of CD8⁺ T cells was tested in LLC and mouse CT26 colorectal carcinoma cells. Tryptophan resulted in a dose-dependent increase in the apoptotic rate of LLC cells that were loaded with the OVA peptide (SIINFEKL) and co-cultured with CD8⁺ T cells at 1:10. Conversely, it had a negligible effect on the apoptotic rate of LLC that were cultured alone (figure 7D and online supplemental figure S7E). Enhancement in the toxicity of CD8⁺ T cells by tryptophan was also observed in mouse CT26 colorectal carcinoma cells (figure 7E and online supplemental figure S7F). These results, together with that kynurenine had negligible effects on TrpK and expression of NFATc1 (revised online supplemental figure S7G) and PD-1 (see figure 2C), IL-2, IFN-γ, and TNF-α (online supplemental figure S7H, I), showed that tryptophan potentiated the toxicity of CD8⁺ T cells by enhancing tryptophanylation toward cancer cells.

Consistent with the results of cultured cells, when LLC were injected subcutaneously into C57BL/6 mice, intraperitoneal administration of either Me-Trp or indoximod slowed down tumor growth and the combined treatment of Me-Trp and indoximod resulted in synergistic effect on tumor growth inhibition (figure 7F,G). Two observations were made in tumor grafts following Me-Trp and indoximod treatments. First, both Me-Trp and indoximod decreased the PD-1 levels of the CD8⁺ T cells of mice spleens (figure 7H, online supplemental figure S7J) and tumor-infiltrating lymphocyte (TIL) CD8⁺ T cells, while combined treatment of Me-Trp and indoximod led to greater decrease of PD-1 (figure 7I, online supplemental figure S7K). Second, the number of CD4⁺ and CD8⁺ T cells in the tumor mass of Me-Trp-treated and indoximod-treated mice was more than double that of the tumors of untreated mice (figure 7J), Me-Trp and indoximod co-treatment showed a combined effect to enhance the number of CD4⁺ and CD8⁺ T cells, suggesting that these treatments enhanced CD8⁺ T cells survival and their tumor mass filtration ability.

To generalize the tumor growth inhibition effects of tryptophanylation by downregulating PD-1 and activating T cells, we examined tryptophan-supplemental effects on tumor growth of B16F10 melanoma cells under with and without PD-1 antibody injection. Tryptophan decreased the growth of B16F10 tumor cells (figure 7K) and decreased surface PD-1 levels in both spleen and TIL CD8⁺ T cells (figure 7L,M), suggested that tryptophan supplemental inhibits growth of cancer cells regardless of their origins. Moreover, although both tryptophan and PD-1 antibody caused tumor regression, tryptophan exerted a potency to inhibit tumor growth in mice treated with PD-1 antibody (figure 7K-M). The synergistic effects of PD-1 antibody and tryptophan supported that tryptophan exerts its antitumor effects through inhibiting PD-1.