Short-term starvation inhibits CD36 N-glycosylation and downregulates USP7 UFMylation to alleviate RBPJ-maintained T cell exhaustion in liver cancer

Abstract

Rationale: Short-term starvation (STS) has been shown to enhance the sensitivity of tumors to chemotherapy while concurrently safeguarding normal cells from its detrimental side effects. Nonetheless, the extent to which STS relies on the anti-tumor immune response to impede the progression of hepatocellular carcinoma (HCC) remains uncertain. Methods: In this study, we employed mass cytometry, flow cytometry, immunoprecipitation, immunoblotting, CUT&Tag, RT-qPCR, and DNA pull-down assays to evaluate the relationship between STS and T-cell antitumor immunity in HCC. Results: We demonstrated that STS alleviated T cell exhaustion in HCC. This study elucidated the mechanism by which STS blocked CD36 N-glycosylation, leading to the upregulation of AMPK phosphorylation and the downregulation of USP7 UFMylation, thus enhancing ubiquitination and destabilized USP7. Consequently, diminished USP7 levels facilitated the ubiquitination and subsequent degradation of RBPJ, thereby inhibiting T cell exhaustion through the IRF4/TNFRSF1B axis. From a therapeutic standpoint, STS not only suppressed the growth of patient-derived orthotopic xenografts but also enhanced their sensitivity to immunotherapy. Conclusions: These findings uncovered a novel mechanism by which N-glycosylation participated in UFMylation/ubiquitination to regulate T cell exhaustion, and we underscored the potential of targeting USP7 and RBPJ in anti-tumor immunotherapy strategies.

Keywords: HCC; RBPJ; STS; T cell exhaustion; UFMylation; USP7; ubiquitination.

Figure 1

STS alleviated T cell exhaustion. (A-C) Effects of STS on subcutaneous tumor growth (n = 6). (A) Representative. (B) Growth curve. (C) Tumor weight. (D, E) Effects of STS on primary cancer growth (n = 6). (D) Representative. (E) Survival curve. (F) Effects of STS on T cell infiltration and its PD1 expression in primary cancer (n = 6). (G-I) Effects of STS and neutralizing CD3⁺ T cells on subcutaneous tumor growth (n = 6). Anti-mouse CD3ε antibody was injected into the tail vein to clear T cells. (G) Representative. (H) Growth curve. (I) Tumor weight. (J, K) Effects of STS and neutralizing CD3⁺ T cells on primary cancer growth (n = 6). (J) Representative. (K) Survival curve. (L) Non-metric Multidimensional Scaling analysis compared the similarity of T cell characteristic antigen expression before and after STS (n = 6). (M) Heatmap showed the median expression of the antigen used to generate SOM (n = 6). (N) SOM was superimposed on mass cytometry data of primary carcinoma-infiltrating T cells (n = 6). (O) Mosaic of single T cells (n = 6). (P) Heatmap showed the proportion of T cell subsets (n = 6). (B), (C), (F), (H), (I), and (P) represented mean ± SD analyzed by unpaired t test, (E) and (K) were analyzed by Log-rank test, (M) was analyzed by Euclidean Distance Clustering Algorithm. *P <0.05, **P <0.01. SOM, self-organizing map; STS, short-term starvation.

Figure 2

STS alleviated T cell exhaustion by inhibiting USP7. (A-C) Effects of STS on USP7 expression in primary carcinoma-infiltrating CD3⁺ T cells were analyzed by mass spectrometry (A), immunoblotting (B) and RT-qPCR (C) (n = 6). (D) Effects of KO or OE of Usp7 on PD1 expression in CD3⁺ T cells (n = 6). (E-G) Effects of Usp7-cKO on subcutaneous tumor growth (n = 6). (E) Representative. (F) Growth curve. (G) Tumor weight. (H, I) Effects of Usp7-cKO on primary cancer growth (n = 6). (H) Representative. (I) Survival curve. (J) Non-metric Multidimensional Scaling analysis compared the similarity of T cell characteristic antigen expression before and after Usp7-cKO (n = 6). (K) Heatmap showed the median expression of the antigen used to generate SOM (n = 6). (L) SOM was superimposed on mass cytometry data of primary carcinoma-infiltrating T cells (n = 6). (M) Mosaic of single T cells (n = 6). (N) Heatmap showed the proportion of T cell subsets (n = 6). (O) Median expression of T cell characteristic antigen (n = 6). (P-R) Effects of STS and Usp7-cKI on subcutaneous tumor growth (n = 6). (P) Representative. (Q) Growth curve. (R) Tumor weight. (S, T) Effects of STS and Usp7-cKI on primary carcinoma growth (n = 6). (S) Representative. (T) Survival curve. (A-D), (F), (G), (N), (O), (Q), and (R) represented mean ± SD analyzed by unpaired t test, (I) and (T) were analyzed by Log-rank test, (K) was analyzed by Euclidean Distance Clustering Algorithm. *P <0.05, **P <0.01. cKI, conditional knock-in; cKO, conditional knockout; OE, overexpression; SOM, self-organizing map; STS, short-term starvation.

Figure 3

USP7 inhibited RBPJ ubiquitination. (A) Identification of proteins bound by USP7 in primary carcinoma-infiltrating CD3⁺ T cells by immunoprecipitation-mass spectrometry assays. (B) PXD039633 dataset showed the binding of USP7 and RBPJ (n =10). (C, D) Effects of OE or KO of Usp7 on the expression of RBPJ protein (C) and transcript (D) in CD3⁺ T cells (n = 3). (E, F) Effects of STS on the expression of RBPJ protein (E) and transcript (F) in CD3⁺ T cells (n = 3). (G) Effects of STS and Usp7-OE on the RBPJ protein levels in CD3⁺ T cells (n = 3). (H) CD3⁺ T cell lysate was treated with anti-IgG control and -USP7 (left) or -RBPJ (right) antibodies, with 5% lysate as input control (n = 3). (I) Surface plot presented docking models and interface residues between USP7 (sky blue) and RBPJ (green) proteins, hydrogen bonds highlighted with dashed lines. (J, K) Schematic plot of domain-deficient mutants for USP7 (J) and RBPJ (K). (L, M) Immunoprecipitation identified key domains of USP7 and RBPJ binding (n = 3). CD3⁺ T cells were co-transfected with WT or domain-deficient mutants (L) of Myc-labeled USP7 and His-labeled RBPJ, and vice versa (M). (N, O) Effects of OE (N) or KO (O) of Usp7 on RBPJ ubiquitination in CD3⁺ T cells (n = 3). (P, Q) Effects of the NTR (P) and AEEPPAHAP (Q) of RBPJ on its ubiquitination in CD3⁺ T cells (n = 3). (R, S) Effect of the UBL domain (R) and the DUB domain (S) of USP7 on RBPJ ubiquitination in CD3⁺ T cells (n = 3). (T) CD3⁺ T cells with His-labeled RBPJ were treated with Cycloheximide, and their expression was evaluated (n = 3). (B-G), and (T) represented mean ± SD analyzed by unpaired t test. *P <0.05, **P <0.01. KO, knockout; OE, overexpression; STS, short-term starvation; WT, wild-type.

Figure 4

USP7 aggravated T cell exhaustion by up-regulating RBPJ. (A-C) Effects of Rbpj-cKO on subcutaneous tumor growth (n = 6). (A) Representative. (B) Growth curve. (C) Tumor weight. (D, E) Effects of Rbpj-cKO on primary cancer growth (n = 6). (D) Representative. (E) Survival curve. (F) Non-metric Multidimensional Scaling analysis compared the similarity of T cell characteristic antigen expression before and after Rbpj-cKO (n = 6). (G) SOM was superimposed on mass cytometry data of primary carcinoma-infiltrating T cells (n = 6). (H) Mosaic map of single T cells (n = 6). (I) Heatmap showed the proportion of T cell subsets (n = 6). (J-L) Effects of Rbpj-cKO and Usp7-cKI on subcutaneous tumor growth (n = 6). (J) Representative. (K) Growth curve. (L) Tumor weight. (M, N) Effects of Rbpj-cKO and Usp7-cKI on primary cancer growth (n = 6). (M) Representative. (N) Survival curve. (O-Q) Effects of STS and Rbpj-cKI on subcutaneous tumor growth (n = 6). (O) Representative. (P) Growth curve. (Q) Tumor weight. (R, S) Effects of STS and Rbpj-cKI on primary cancer growth (n = 6). (R) Representative. (S) Survival curve. (B), (C), (I), (K), (L), (P), and (Q) represented mean ± SD analyzed by unpaired t test, (E), (N), and (S) were analyzed by Log-rank test. *P <0.05, **P <0.01. cKI, conditional knock-in; cKO, conditional knockout; SOM, self-organizing map.

Figure 5

RBPJ enhanced transcription of exhaustion genes. (A, B) The cluster plot showed the DNA fragments that RBPJ binds to (A), and their distribution across the genome (B) in primary carcinoma-infiltrating CD3⁺ T cells (n = 3). (C) Pathway enrichment analysis of genes mapped to RBPJ-bound DNA fragments compared to IgG controls in CD3⁺ T cells (n = 3). Red indicated T cell- or HCC-related items. (D) A four-quadrant plot presented the distribution of genes with significant alterations in both the DNA fragment bound to RBPJ and the mRNA levels in CD3⁺ T cells (n = 3). (E, F) Snapshot plots showed explicit transcription expression of Irf4 (E) and Tnfrsf1b (F) and the enrichment signal of RBPJ on their promoter in CD3⁺ T cells (n = 3). Gray indicated the differential signal. (G) DNA pull-down experiment demonstrated the binding of RBPJ to the promoter of Irf4 or Tnfrsf1b in CD3⁺ T cells (n = 3). (H) Effects of loss of TGGGAA of Irf4 promoter or TTACCA of Tnfrsf1b promoter on DNA affinity of RBPJ in CD3⁺ T cells (n = 3). (I, J) Effects of Rbpj-cKO on transcription (I) and protein (J) expression of IRF4 or TNFRSF1B in CD3⁺ T cells (n = 3). (K) Flow cytometry demonstrated inhibitory receptor expression in CD3⁺ T cells (n = 6). (L) ELISA demonstrated 22 histone H3 modification levels in CD3⁺ T cells (n = 4). (M, N) Immunofluorescence (M) and immunoblotting (N) demonstrated the expression of H3S10P in CD3⁺ T cells (n = 3). (O) Flow cytometry presented the cell cycle of CD3⁺ T cells (n = 3). (I-O) represented mean ± SD analyzed by unpaired t test. *P <0.05, **P <0.01. cKO, conditional knockout.

Figure 6

USP7 underwent UFMylation. (A) PXD039633 dataset showed the binding of USP7 and UFL1 (n =10). (B) Effects of OE or KO of Ufl1 or Ufsp2 on Usp7 transcript expression in CD3⁺ T cells (n = 3). (C-F) Effects of OE (C, E) or KO (D, F) of Ufl1 (C, D) or Ufsp2 (E, F) on USP7 protein expression in CD3⁺ T cells (n = 3). (G) CD3⁺ T cell lysate was treated with IgG control and UFL1 (left) or USP7 (right) antibodies, and 5% lysate was used as input control (n =3). (H) CD3⁺ T cell lysate was treated with anti-IgG controls and -UFM1 (left) or -USP7 (right) antibodies (n =3). (I) CD3⁺ T cell lysate was treated with anti-IgG controls and -UFSP2 (left) or -USP7 (right) antibodies (n =3). (J-L) Surface plot presented docking models and interface residues between USP7 (sky blue) and UFL1 (purple) (J)/UFM1 (orange) (K)/UFSP2 (green) (L) proteins, with hydrogen bonds highlighted by dashed lines. (M-O) Schematic plot of the domain-deficient mutants of UFL1 (M), UFM1 (N), and UFSP2 (O). (P, Q) Immunoprecipitation identified key domains of USP7 and UFL1 binding (n =3). CD3⁺ T cells were co-transfected with WT or domain-deficient mutants of Flag-labeled UFL1 and Myc-labeled USP7 (P) and vice versa (Q). (R, S) Immunoprecipitation identified key domains of USP7 and UFM1 binding (n =3). CD3⁺ T cells were co-transfected with WT or domain-deficient mutants of GFP-labeled UFM1 and Myc-labeled USP7 (R) and vice versa (S). (T, U) Immunoprecipitation identified key domains of USP7 and UFSP2 binding (n =3). CD3⁺ T cells were co-transfected with Flag-labeled UFSP2 and Myc-labeled WT or domain-deficient mutants of USP7 (T) and vice versa (U). (A-F) represented mean ± SD analyzed by unpaired t test. *P <0.05, **P <0.01. KO, knockout; OE, overexpression; WT, wild-type.

Figure 7

STS inhibited USP7 UFMylation thus increasing its ubiquitination. (A-C) Effects of short-term starvation (A), refeed (B), and deletion of the last three amino acid residues (83Gly-Ser-Cys85, ΔC3) of UFM1 (C) on USP7 UFMylation (n = 3). CD3⁺ T cells were co-transfected with Myc-labeled USP7, Flag-labeled UFL1, and GFP-labeled WT or deletion mutants of UFM1. (D-G) Effects of OE (D, F) or KO (E, G) of Ufl1 (D, E) or Ufsp2 (F, G) on USP7 UFMylation in CD3⁺ T cells (n = 3). (H) Polyubiquitination of six Myc-labeled USP7 lysine mutants in CD3⁺ T cells (n = 3). (I) Cross-species conservation of K1097 of USP7. Red indicated lysine site linked to ubiquitin. Underline indicated the same amino acid sequence. (J) CD3⁺ T cells with Myc-labeled USP7 were treated with Cycloheximide and USP7 protein expression was evaluated by immunoblotting (n = 3). (K) Ubiquitination chain analysis of USP7 in CD3⁺ T cells (n = 3). (L) Effects of UFL1 on K48-linked USP7 ubiquitination in CD3⁺ T cells (n = 3). (J) represented mean ± SD analyzed by unpaired t test. *P <0.05, **P <0.01. KO, knockout; OE, overexpression; WT, wild-type.

Figure 8

STS inhibited CD36 N-glycosylation and thus preventing USP7 UFMylation. (A) AMPK pathway-associated proteins that might undergo N-glycosylation. (B, C) Effects of OE (B) or KO (C) of Cd36 on AMPK phosphorylation in CD3⁺ T cells (n = 3). (D) Effects of Tunicamycin, PNGase F, Swainsonine, and glucose starvation on CD36 N-glycosylation in CD3⁺ T cells (n = 3). (E) Protein structure and N-glycosylation site of CD36. (F) Effects of three possible N-glycosylated asparagine mutations to glutamine on CD36 expression in CD3⁺ T cells (n = 3). (G) Influence of mutations of three N-glycosylation sites on the membrane localization of CD36 on non-permeable CD3⁺ T cells (n = 3). (H) Effects of three N-glycosylation site mutations on AMPK phosphorylation in CD3⁺ T cells (n = 3). (I-M) Effects of AMPK inhibitor Dorsomophine (I), AMPK agonist A-769662 (J), Cd36-OE (K), Cd36-KO (L), or mutations of three N-glycosylation sites (M) on the binding of USP7 and UFL1 in CD3⁺ T cells (n = 3). (N) Influence of mutations of three N-glycosylation sites and Dorsomophine on USP7 UFMylation in CD3⁺ T cells (n = 3). (O) Mechanism diagram. (B-D) and (F-H) represented mean ± SD analyzed by unpaired t test. *P <0.05, **P <0.01. KO, knockout; OE, overexpression.