PRDM1/BLIMP1 induces cancer immune evasion by modulating the USP22-SPI1-PD-L1 axis in hepatocellular carcinoma cells

Most patients with hepatocellular carcinoma are unresponsive to PD-1/PD-L1 axis blockade.

“However, most patients, especially patients with HCC, are unresponsive to the blockade of the PD-1/PD-L1 axis5,6.”

• cites [5] Sangro, B. et al. Association of inflammatory biomarkers with clinical outcomes in nivolumab-treated patients with advanced hepatocellular carcinoma. J. Hepatol. 73, 1460–1469 (2020). sourceSupported
• cites [6] Yang, W. et al. A selective HDAC8 inhibitor potentiates antitumor immunity and efficacy of immune checkpoint blockade in hepatocellular carcinoma. Sci. Transl. Med. 13, eaaz6804 (2021). sourceNot checked — no full text of the source

Tumoral PD-L1 expression alone is not a reliable predictive biomarker of immunotherapy in most patients with HCC.

“In addition, tumoral PD-L1 expression alone is not a reliable predictive biomarker of immunotherapy in most patients with HCC5.”

• cites [5] Sangro, B. et al. Association of inflammatory biomarkers with clinical outcomes in nivolumab-treated patients with advanced hepatocellular carcinoma. J. Hepatol. 73, 1460–1469 (2020). sourcePartly supported

  The Discussion reports responses and changes in target-lesion size regardless of tumour PD-L1 expression, concluding that PD-L1 expression alone may not be an adequate biomarker. That is tentative evidence from nivolumab-treated patients with advanced HCC; it does not establish the claim's definitive conclusion about immunotherapy in most patients with HCC. (LLM, conf=0.86)

  Experimental — accuracy not yet measured

PRDM1 upregulation impaired CD8+ T cell activation (CD8+GZMB+ T cells) in co-culture.

“PRDM1 upregulation impaired the CD8+ T cell activation (CD8+GZMB+ T cells) and T cell-mediated tumor cell killing activity (CD8+TNFα+ T cells) in the co-culture”

Among the identified regulators, PRDM1 showed the highest correlation with PD-L1 expression.

“Among these, PRDM1 showed the highest correlation with PD-L1 expression (Supplementary Fig. 1a).”

PRDM1 knockout had contrasting effects on CD8+ T cell activation and killing activity in co-culture.

“whereas PRDM1 knockout had contrasting effects (Fig. 1g, h).”

PRDM1 overexpression upregulated PD-L1 at mRNA and total and cell surface protein levels in Hep3B cells with or without IFN-γ stimulation.

“PD-L1 upregulation at both mRNA and total and cell surface protein levels after PRDM1 overexpression was confirmed in PRDM1-overexpressing Hep3B cells, with or without IFN-γ stimulation”

PRDM1 knockout decreased PD-L1 at mRNA and total and cell surface protein levels in Huh7 cells.

“and the opposite effect was observed in PRDM1-knockout Huh7 cells (Fig. 1a–e).”

Correlation analysis using the GEPIA database identified PRDM1, PRDM2, PRDM3, PRDM4, PRDM5, PRDM6, PRDM8, PRDM10, PRDM11, and PRDM13 as PD-L1 expression regulators.

“PRDM1, PRDM2, PRDM3, PRDM4, PRDM5, PRDM6, PRDM8, PRDM10, PRDM11, and PRDM13 were identified as PD-L1 expression regulators.”

PRDM1 upregulation impaired T cell-mediated tumor cell killing activity (CD8+TNFα+ T cells) in co-culture.

“PRDM1 upregulation impaired the CD8+ T cell activation (CD8+GZMB+ T cells) and T cell-mediated tumor cell killing activity (CD8+TNFα+ T cells) in the co-culture”

Following CHX treatment, USP22 overexpression significantly attenuated SPI1 degradation.

“following CHX treatment, USP22 overexpression significantly attenuated SPI1 degradation”

The C-terminal ubiquitin-specific peptidase domain of USP22 binds SPI1 strongly.

“The C-terminal ubiquitin-specific peptidase domain of USP22 binds SPI1 strongly, whereas the N-terminal zinc finger domain binds SPI1 weakly (Fig. 5d).”

GEPIA database indicated a positive relationship between SPI1 and PD-L1 expression in HCC.

“HCC data mining of the GEPIA database indicated a positive relationship between SPI1 and PD-L1 expression, and T cell exhaustion markers, providing proof for SPI1-induced PD-L1 upregulation (Supplementary Fig. 5a–h).”

PRDM1 overexpression increased the mRNA and protein levels of USP22.

“PRDM1 overexpression only increased the mRNA and protein levels of USP22, whereas PRDM1 knockout decreased the mRNA and protein levels of USP22 (Fig. 4g, h and Supplementary Fig. 4g).”

SPI1 overexpression significantly upregulated PD-L1 mRNA and protein levels following incubation with or without IFN-γ.

“SPI1 overexpression significantly upregulated PD-L1 mRNA and protein levels following incubation with or without IFN-γ, whereas SPI1 knockdown had the opposite effect (Fig. 3e and Supplementary Fig. 4c).”

Four putative SPI1-binding sites were observed in the genomic region (NM_017798) of the PD-L1 promoter.

“Four putative SBS were observed in the genomic region (NM_017798) (Fig. 3f).”

PRDM1-overexpressing Hep3B cells showed significantly lower polyubiquitination levels compared to control cells.

“PRDM1-overexpressing Hep3B cells showed significantly lower polyubiquitination levels compared to control cells”

The N-terminal zinc finger domain of USP22 binds SPI1 weakly.

“The C-terminal ubiquitin-specific peptidase domain of USP22 binds SPI1 strongly, whereas the N-terminal zinc finger domain binds SPI1 weakly (Fig. 5d).”

In immunodeficient mice, Prdm1 overexpression reduced cell proliferation.

“Nevertheless, Prdm1 overexpression reduced cell proliferation, while Prdm1 knockout promoted cell proliferation in immunodeficient mice (Supplementary Fig. 2b, c).”

In immunodeficient mice, shortened overall survival was observed in the sgPrdm1 groups.

“Further, prolonged and shortened OS were observed in the Prdm1 and sgPrdm1 groups, respectively.”

In subcutaneous transplanted immune-competent model, Prdm1 groups had decreased activity (GZMB+) of infiltrated CD8+ T cells.

“flow cytometry revealed reduction of infiltrated CD8+ T cells and decreased activity (GZMB+) of infiltrated CD8+ T cells in the Prdm1 groups in the subcutaneous transplanted model.”

The C-terminal ubiquitin-specific peptidase domain of USP22 containing a point mutation could bind SPI1 weakly.

“the C-terminal ubiquitin-specific peptidase domain of USP22 containing a point mutation could bind SPI1 weakly, whereas the N-terminal zinc finger domain of USP22 containing a point mutation could bind SPI1 more strongly (Fig. 5g).”

In immunodeficient mice, tumor proliferation was distinctly slower in the Prdm1 group than in corresponding controls.

“tumor proliferation was distinctly slower in the Prdm1 group and was faster in the sgPrdm1 groups than in the corresponding controls.”

No obvious differences were observed between Prdm1 or sgPrdm1 groups and controls in Ki67 staining in immunocompetent mice.

“Moreover, no obvious differences were observed between the Prdm1 or sgPrdm1 groups and their corresponding controls in terms of Ki67 staining in immunocompetent mice.”

In immunodeficient orthotopic models, sgPrdm1 groups showed opposite effects (larger tumors).

“Immunodeficient orthotopic models also confirmed that Prdm1 overexpression resulted in smaller tumors, whereas the opposite effects were observed in the sgPrdm1 groups (Fig. 2j–o).”

Myc-tagged USP22 (the C-terminus ubiquitin-specific peptidase) and His-tagged SPI1 D3 were mutually immunoprecipitated.

“Additionally, Myc-tagged USP22 (the C-terminus ubiquitin-specific peptidase) and His-tagged SPI1 D3 were mutually immunoprecipitated (Fig. 5l).”

The C-terminus of USP22 containing a point mutation did not affect SPI1 ubiquitination.

“Further, the C-terminus of USP22 containing a point mutation did not affect SPI1 ubiquitination (Fig. 5h).”

Similar results were obtained in luciferase assays and ChIP assays of Huh7 cells with SPI1 knockdown.

“Similar results were also obtained in luciferase assays and ChIP assays of Huh7 cells with SPI1 knockdown (Fig. 3g–j).”

PRDM1-knockout Huh7 cells displayed dramatically higher polyubiquitination levels compared to sgCtrl cells.

“whereas PRDM1-knockout Huh7 cells displayed dramatically higher polyubiquitination levels compared to sgCtrl cells (Fig. 4d, e).”

PD-1 mAb treatment dramatically inhibited tumor proliferation compared to IgG in Hepa1-6 cells.

“Nevertheless, PD-1 mAb treatment dramatically inhibited tumor proliferation and extended survival time compared to the IgG group in Hepa1-6 cells.”

USP22-knockdown Huh7 cells displayed significantly higher polyubiquitination levels compared to control cells.

“whereas USP22-knockdown Huh7 cells displayed significantly higher polyubiquitination levels compared to the control cells (Fig. 4l, m).”

Luciferase assays combined with site-deletion or site-directed mutagenesis suggested that SBS 3 and 4 in the PD-L1 promoter induced SPI1-enhanced promoter activity in PRDM1-overexpressing Hep3B cells.

“Luciferase assays combined with site-deletion or site-directed mutagenesis suggested that SBS 3 and 4 in the PD-L1 promoter-induced SPI1-enhanced promoter activity in PRDM1-overexpressing Hep3B cells.”

In immunodeficient mice, Prdm1 knockout promoted cell proliferation.

“Nevertheless, Prdm1 overexpression reduced cell proliferation, while Prdm1 knockout promoted cell proliferation in immunodeficient mice (Supplementary Fig. 2b, c).”

Prdm1 overexpression group showed no visible difference in tumor proliferation and survival time compared with control group.

“The Prdm1 overexpression group showed no visible difference in tumor proliferation and survival time compared with the control group.”

PRDM1 knockout diminished PD1 binding to HCC cells.

“PRDM1 knockout-mediated decreased PD-L1 expression also diminished PD1 binding to HCC cells (Fig. 1i, j).”

Tumors from Prdm1 groups in C57BL/6 mice showed significant upregulation of PD-L1 protein and mRNA expression.

“Significant upregulation of PD-L1 protein and mRNA expression was found in tumors from Prdm1 groups, related to an obvious restraint in the proportions of infiltrated CD8+ T cells.”

In Pd-l1−/− Hepa1-6 and H22 cells inoculated into immune-competent mice, Prdm1 knockout contributed to tumor proliferation.

“The results revealed that Prdm1 overexpression in immune-competent mice inhibited tumor proliferation and that Prdm1 knockout contributed to tumor proliferation (Supplementary Fig. 3i–n).”

A potential BLIMP1-binding site (BBS) in the USP22 promoter was determined using Jaspar.

“Intriguingly, a potential BLIMP1-binding site (BBS) in the USP22 promoter was determined using Jaspar (Fig. 5m).”

BLIMP1 protein expression was inversely associated with the activity (GZMB+) of CD8+ T cells (R = −0.4882, P = 0.0014) in infiltrating CD45+ cells in HCC cohort 1.

“BLIMP1 protein expression was inversely associated with the proportion of CD8+ T cells (R = −0.4877, P = 0.0014) and the activity (GZMB+) of CD8+ T cells (R = −0.4882, P = 0.0014) in infiltrating CD45+ cells.”

GEPIA datasets validated that T cell exhaustion markers were positively correlated with PRDM1 levels in HCC.

“GEPIA datasets further validated that T cell exhaustion markers were positively correlated with PRDM1 levels in HCC, indicating the TILs regulation of PRDM1 expression (Supplementary Fig. 1b).”

ChIP assays validated BLIMP1 occupancy at the USP22 promoters in Hep3B and Huh7 cells.

“ChIP assays validated BLIMP1 occupancy at the USP22 promoters in Hep3B and Huh7 cells (Fig. 5p, q).”

In immune-competent mice, Prdm1 overexpression and Prdm1 knockout showed no obvious differences in tumor size and overall survival compared with their corresponding controls.

“However, no obvious differences were observed between the Prdm1 or sgPrdm1 groups and their corresponding controls in terms of tumor size and OS.”

GEPIA database indicated a positive relationship between SPI1 and T cell exhaustion markers in HCC.

“HCC data mining of the GEPIA database indicated a positive relationship between SPI1 and PD-L1 expression, and T cell exhaustion markers, providing proof for SPI1-induced PD-L1 upregulation (Supplementary Fig. 5a–h).”

The USP22 knockdown-mediated decrease in SPI1 expression was abolished by MG132.

“The USP22 knockdown-mediated decrease in SPI1 expression was abolished by MG132 (Fig. 4k and Supplementary Fig. 4j).”

BLIMP1 protein expression was inversely associated with the proportion of CD8+ T cells (R = −0.4877, P = 0.0014) in infiltrating CD45+ cells in HCC cohort 1.

“BLIMP1 protein expression was inversely associated with the proportion of CD8+ T cells (R = −0.4877, P = 0.0014) and the activity (GZMB+) of CD8+ T cells (R = −0.4882, P = 0.0014) in infiltrating CD45+ cells.”

In subcutaneous transplanted immune-competent model, Prdm1 groups had reduced infiltrated CD8+ T cells.

“flow cytometry revealed reduction of infiltrated CD8+ T cells and decreased activity (GZMB+) of infiltrated CD8+ T cells in the Prdm1 groups in the subcutaneous transplanted model.”

PRDM1-induced PD-L1 upregulation was reversed by SPI1 knockdown in Hep3B cells with or without IFN-γ stimulation.

“Moreover, PRDM1-induced PD-L1 upregulation was reversed by SPI1 knockdown in Hep3B cells with or without IFN-γ stimulation (Fig. 3k).”

CCK8 assays verified that PRDM1 suppressed proliferation in both human and murine HCC cell lines.

“CCK8 assays verified that PRDM1 suppressed proliferation in both human and murine HCC cell lines (Supplementary Fig. 2d).”

PRDM1 knockout decreased USP22 promoter activity in Huh7 cells.

“The luciferase assay indicated that PRDM1 overexpression induced higher promoter activity in Hep3B cells, whereas the opposite effect was observed in PRDM1-knockout Huh7 cells.”

In sgPrdm1 groups, infiltrated CD8+ T cell exhaustion was significantly reduced.

“Infiltrated CD8+ T cells and the activity (GZMB+) of infiltrated CD8+ T cells was obviously increased, and infiltrated CD8+ T cell exhaustion was significantly reduced in the sgPrdm1 groups (Fig. 2g–i).”

Combined treatment with Prdm1 overexpression and PD-1 mAb further impaired tumor proliferation compared with Prdm1 overexpression or PD-1 mAb alone.

“More importantly, combined treatment with Prdm1 overexpression and PD-1 mAb further impaired tumor proliferation and prolonged survival time compared with that in Prdm1 overexpression or PD-1 mAb treatment alone (Fig. 6a, c, e, f).”

In immunodeficient orthotopic models, Prdm1 overexpression resulted in smaller tumors.

“Immunodeficient orthotopic models also confirmed that Prdm1 overexpression resulted in smaller tumors, whereas the opposite effects were observed in the sgPrdm1 groups (Fig. 2j–o).”

The N-terminal zinc finger domain of USP22 was not involved in SPI1 ubiquitination.

“Moreover, the N-terminal zinc finger domain of USP22 was not involved in SPI1 ubiquitination (Fig. 5e).”

PRDM1 overexpression increased PD1 binding to HCC cells.

“With PRDM1-induced increased PD-L1 expression, PRDM1 overexpression increased PD1 binding to HCC cells.”

Combined treatment with Prdm1 overexpression and PD-1 mAb prolonged survival time compared with Prdm1 overexpression or PD-1 mAb alone.

“More importantly, combined treatment with Prdm1 overexpression and PD-1 mAb further impaired tumor proliferation and prolonged survival time compared with that in Prdm1 overexpression or PD-1 mAb treatment alone (Fig. 6a, c, e, f).”

IP/MS results validated USP22 as an SPI1-interacting protein.

“The immunoprecipitation and mass spectrometry (IP/MS) results validated USP22 as an SPI1-interacting protein (Supplementary Fig. 6a, b).”

Proteomics analysis using Hep3B cells stably overexpressing PRDM1 and control cells identified SPI1 as the most likely candidate based on fold change.

“SPI1, a transcription factor for PD-L1, was the most likely candidate based on fold change (Fig. 3b).”

• AssumptionRests on an assumption: SPI1 is a transcription factor for PD-L1.

SPI1 mRNA levels were unchanged in HCC cells with differential PRDM1 expression despite differences in SPI1 protein levels.

“Despite differences in SPI1 protein levels, its mRNA levels were unchanged in HCC cells with differential PRDM1 expression, suggesting that PRDM1 may affect SPI1 protein stability (Fig. 3c, d and Supplementary Fig. 4b).”

The N-terminal zinc finger domain of USP22 containing a point mutation could bind SPI1 more strongly.

“the C-terminal ubiquitin-specific peptidase domain of USP22 containing a point mutation could bind SPI1 weakly, whereas the N-terminal zinc finger domain of USP22 containing a point mutation could bind SPI1 more strongly (Fig. 5g).”

PRDM1 overexpression significantly attenuated SPI1 degradation in HCC cells treated with cycloheximide.

“PRDM1 overexpression significantly attenuated SPI1 degradation.”

Only SPI1 mutants containing the C-terminal (aa 170-270) region, but not the N-terminal region (aa 1-85) and the middle region (aa 86-169), were found to interact with USP22.

“Only SPI1 mutants containing the C-terminal (aa 170–270) region, but not the N-terminal region (aa 1–85) and the middle region (aa 86–169), were found to interact with USP22 (Fig. 5k).”

Tumors from Prdm1 groups in C57BL/6 mice showed an obvious restraint in the proportions of infiltrated CD8+ T cells.

“Significant upregulation of PD-L1 protein and mRNA expression was found in tumors from Prdm1 groups, related to an obvious restraint in the proportions of infiltrated CD8+ T cells.”

The C-terminal ubiquitin-specific peptidase domain of USP22 containing a point mutation did not affect the degradation of SPI1 following CHX treatment.

“The C-terminal ubiquitin-specific peptidase domain of USP22 containing a point mutation did not affect the degradation of SPI1 following CHX treatment (Fig. 5i).”

In T cell-mediated tumor cell-killing assays, PRDM1 overexpressing HCC cells showed increased resistance to activated primary CD45+CD3+ T cells isolated from PBMCs in vitro.

“In T cell-mediated tumor cell-killing assays, PRDM1 overexpressing HCC cells showed increased resistance to activated primary CD45+CD3+ T cells isolated from PBMCs in vitro (Fig. 2p, q).”

The binding of IFN-γ pre-treated HCC cells with PD1/Fc protein indicated that the PD-L1-PD1 interaction steadily increased over 12 h.

“The binding of IFN-γ pre-treated HCC cells (500 IU/ml) with PD1/Fc protein (green fluorescence) indicated that the PD-L1-PD1 interaction steadily increased over 12 h.”

IF assays revealed co-localization of USP22 and SPI1.

“Immunofluorescence (IF) assays also revealed the co-localization of USP22 and SPI1 (Fig. 5b).”

PRDM1 knockout decreased the mRNA and protein levels of USP22.

“PRDM1 overexpression only increased the mRNA and protein levels of USP22, whereas PRDM1 knockout decreased the mRNA and protein levels of USP22 (Fig. 4g, h and Supplementary Fig. 4g).”

SPI1 protein was upregulated in PRDM1-overexpressing HCC cells.

“SPI1 protein upregulation and downregulation were confirmed in PRDM1-overexpressing and PRDM1-knockout HCC cells, respectively (Fig. 3c, d and Supplementary Fig. 4b).”

MG132 treatment recovered SPI1 levels in sgPRDM1 or sgCtrl Huh7 cells.

“MG132 treatment recovered the SPI1 levels in sgPRDM1 or sgCtrl Huh7 cells”

BLIMP1 protein expression was positively correlated with the proportion of exhausted CD8+ T cells (PD1+) (R = 0.3722, P = 0.0181) in CD8+ TILs in HCC cohort 1.

“BLIMP1 protein expression was positively correlated with the proportion of exhausted CD8+ T cells (PD1+) (R = 0.3722, P = 0.0181) in CD8+ TILs (Fig. 1k, l).”

In immunodeficient mice, tumor proliferation was faster in the sgPrdm1 groups than in corresponding controls.

“tumor proliferation was distinctly slower in the Prdm1 group and was faster in the sgPrdm1 groups than in the corresponding controls.”

Infiltrated CD8+ T cell exhaustion assessed by PD-1+ staining was significantly increased in Prdm1 groups.

“Further, infiltrated CD8+ T cell exhaustion assessed by PD-1+ staining was significantly increased in Prdm1 groups.”

Hep3B cells with USP22 overexpression showed markedly lower polyubiquitination levels compared to control cells.

“Hep3B cells with USP22 overexpression showed markedly lower polyubiquitination levels compared to the control cells”

Co-IP assays verified the interaction between USP22 and SPI1.

“Co-IP assays were then performed to verify the interaction between USP22 and SPI1. USP22 was immunoprecipitated with an anti-SPI1 antibody and vice versa (Fig. 5a).”

ChIP results indicated that SPI1 is recruited only to promoter regions containing SBS 3 and 4 in SPI1-overexpressing Hep3B cells.

“ChIP results also indicated that SPI1 is recruited only to promoter regions containing SBS 3 and 4 in SPI1-overexpressing Hep3B cells.”

PD-L1 levels were comparable in Flag-PD-L1-overexpressing HCC cells with or without PRDM1 overexpression/knockout.

“However, PD-L1 levels were comparable in the Flag-PD-L1 cells with or without PRDM1 overexpression/knockout, suggesting that PRDM1 impacts PD-L1 expression through its endogenous promoter (Fig. 3a and Supplementary Fig. 4a).”

USP22 knockdown strikingly facilitated SPI1 degradation.

“which in contrast, was strikingly facilitated in USP22 knockdown cells (Fig. 4i, j and Supplementary Fig. 4h, i).”

SPI1 degradation was strikingly facilitated in PRDM1-knockout cells.

“In contrast, SPI1 degradation was strikingly facilitated in PRDM1-knockout cells (Fig. 4a, b and Supplementary Fig. 4d, e).”

Downregulation of only two types of DUBs (USP22 and USP33) decreased SPI1 protein levels by more than 50% among 98 known human DUBs.

“Among the 98 currently known human DUBs (Supplementary Table 1), downregulation of only two types of DUBs (USP22 and USP33) decreased the SPI1 protein levels by more than 50% (Fig. 4f).”

• AssumptionRests on an assumption: Deubiquitinating enzymes control ubiquitin-dependent pathways by separating protein-ubiquitin bonds.

SPI1 protein was downregulated in PRDM1-knockout HCC cells.

“SPI1 protein upregulation and downregulation were confirmed in PRDM1-overexpressing and PRDM1-knockout HCC cells, respectively (Fig. 3c, d and Supplementary Fig. 4b).”

PD-1 mAb treatment extended survival time compared to IgG in Hepa1-6 cells.

“Nevertheless, PD-1 mAb treatment dramatically inhibited tumor proliferation and extended survival time compared to the IgG group in Hepa1-6 cells.”

Prdm1 knockout group showed similar tumor proliferation and survival time comparable to sgCtrl group.

“Similarly, the Prdm1 knockout group showed similar tumor proliferation and survival time comparable to the sgCtrl group (Fig. 6a–h).”

PRDM1 overexpression induced higher USP22 promoter activity in Hep3B cells.

“The luciferase assay indicated that PRDM1 overexpression induced higher promoter activity in Hep3B cells, whereas the opposite effect was observed in PRDM1-knockout Huh7 cells.”

In immunodeficient mice, prolonged overall survival was observed in the Prdm1 group.

“Further, prolonged and shortened OS were observed in the Prdm1 and sgPrdm1 groups, respectively.”

Co-treatment with Prdm1 overexpression and PD-1 mAb significantly increased infiltration of CD8+ T cells compared with Prdm1 overexpression group.

“Flow cytometry analysis and IF staining revealed that co-treatment with Prdm1 overexpression and PD-1 mAb significantly increased the infiltration of CD8+ T cells and the activity (GZMB+) of infiltrated CD8+ T cells, and reduced the exhaustion of infiltrated CD8+ T cells as reflected by PD-1+ staining compared with that in the Prdm1 overexpression group (Fig. 6i and Supplementary Fig. 7a–c).”

SPI1 knockdown had the opposite effect on PD-L1 mRNA and protein levels.

“SPI1 overexpression significantly upregulated PD-L1 mRNA and protein levels following incubation with or without IFN-γ, whereas SPI1 knockdown had the opposite effect (Fig. 3e and Supplementary Fig. 4c).”

SPI1 mRNA levels showed no obvious differences in HCC cells with PRDM1 overexpression or knockout and their corresponding controls.

“with no obvious differences in SPI1 mRNA levels in HCC cells with PRDM1 overexpression or knockout and their corresponding controls.”

Co-treatment with Prdm1 knockout and PD-1 mAb had a limited effect on tumor growth and survival time compared with Prdm1 knockout alone.

“However, co-treatment with Prdm1 knockout and PD-1 mAb had a limited effect on tumor growth and survival time compared with Prdm1 knockout alone (Fig. 6b, d, g, h).”

Co-treatment with Prdm1 overexpression and PD-1 mAb significantly reduced exhaustion of infiltrated CD8+ T cells compared with Prdm1 overexpression group.

“Flow cytometry analysis and IF staining revealed that co-treatment with Prdm1 overexpression and PD-1 mAb significantly increased the infiltration of CD8+ T cells and the activity (GZMB+) of infiltrated CD8+ T cells, and reduced the exhaustion of infiltrated CD8+ T cells as reflected by PD-1+ staining compared with that in the Prdm1 overexpression group (Fig. 6i and Supplementary Fig. 7a–c).”

Tumoral PRDM1/BLIMP1 overexpression inhibits cell-intrinsic cell growth.

“PRDM1/BLIMP1 overexpression inhibits cell-intrinsic cell growth while promoting tumor cell immune evasion by up-regulating PD-L1 and dampening CD8+ T cell anti-tumor immune response simultaneously.”

In the DEN/CCl4-induced HCC mouse model, combined treatment with Prdm1 overexpression and PD-1 mAb resulted in less advanced liver lesions, as assessed by gross appearance, H&E staining, tumor numbers, and maximum tumor diameter.

“In this model, we further validated that combined treatment could result in less advanced liver lesions, as assessed by gross appearance (Fig. 6l), H&E staining (Fig. 6m), tumor numbers (Fig. 6n), and maximum tumor diameter (Fig. 6o).”

In HCC biopsies before PD-1 mAb-based therapies, BLIMP1 protein levels were negatively correlated with CD8+ T cells and the activity (GZMB+) of infiltrated CD8+ T cells.

“As expected, BLIMP1 protein levels were negatively correlated with CD8+ T cells and the activity (GZMB+) of infiltrated CD8+ T cells (Supplementary Fig. 9a, b).”

PRDM1 enhances the transcription of PD-L1.

“In our study, based on screening using the online tool GEPIA and establishment of CMV-driven Flag-PD-L1 overexpressing HCC cell lines, we identified that PRDM1 enhances the transcription of PD-L1.”

Tumoral PRDM1/BLIMP1 overexpression promotes tumor cell immune evasion by up-regulating PD-L1 and dampening CD8+ T cell anti-tumor immune response.

“PRDM1/BLIMP1 overexpression inhibits cell-intrinsic cell growth while promoting tumor cell immune evasion by up-regulating PD-L1 and dampening CD8+ T cell anti-tumor immune response simultaneously.”

BLIMP1 was identified as a vital transcription factor of USP22.

“Interestingly, BLIMP1 was then identified as a vital transcription factor of USP22 in our study.”

In immunocompetent orthotopic models, co-treatment with Prdm1 knockout and PD-1 mAb had a limited effect on tumor growth compared with Prdm1 knockout alone.

“Moreover, immunocompetent orthotopic models also confirmed that co-treatment with Prdm1 overexpression and PD-1 mAb further impaired tumor proliferation compared with that in Prdm1 overexpression or PD-1 mAb treatment alone, whereas co-treatment with Prdm1 knockout and PD-1 mAb had a limited effect on tumor growth compared with that in Prdm1 knockout alone (Fig. 6c, d).”

USP22 could reduce SPI1 protein degradation through deubiquitination.

“Here, by utilizing the DUB siRNA library, we found that USP22 could reduce SPI1 protein degradation through deubiquitination.”

In HCC tissues, the protein levels of BLIMP1, USP22, and SPI1 were positively associated with PD-L1 protein levels.

“Our TMA results revealed that the protein levels of BLIMP1, USP22, and SPI1 were positively associated with the PD-L1 protein levels and were negatively associated with CD8+ T cell infiltration in HCC tissues (Fig. 7a, b).”

• Also stated in: § PRDM1-USP22-SPI1 axis regulates PD-L1 levels in patients with HCC · plain

  “Multi-color IHC revealed that BLIMP1, USP22, and SPI1 levels had a positive relationship with PD-L1 expression (Fig. 7c).”

• Also stated in: § Discussion · plain

  “In HCC samples, we determined that BLIMP1, USP22, and SPI1 were positively associated with PD-L1 expression.”

In immunocompetent orthotopic models, co-treatment with Prdm1 overexpression and PD-1 mAb further impaired tumor proliferation compared with Prdm1 overexpression or PD-1 mAb alone.

“Moreover, immunocompetent orthotopic models also confirmed that co-treatment with Prdm1 overexpression and PD-1 mAb further impaired tumor proliferation compared with that in Prdm1 overexpression or PD-1 mAb treatment alone, whereas co-treatment with Prdm1 knockout and PD-1 mAb had a limited effect on tumor growth compared with that in Prdm1 knockout alone (Fig. 6c, d).”

In scRNA-seq of HCC biopsies, patient 2 expressed higher levels of PRDM1 in tumor cells than patient 1.

“The sub-clusters indicated that patient 2 expressed higher levels of PRDM1 (Fig. 8c).”

PRDM1 overexpression in HCC cells dampens T cell-induced cytotoxicity by upregulating tumoral PD-L1 expression.

“Functionally, T cell-mediated cancer cell-killing assays and subcutaneous and orthotopic models demonstrated that PRDM1 overexpression in HCC cells dampens T cell-induced cytotoxicity by upregulating tumoral PD-L1 expression.”

In the DEN-induced HCC mouse model, combined treatment with Prdm1 overexpression and PD-1 mAb maximized the survival of HCC-bearing mice.

“Meanwhile, the combined treatment maximized the survival of HCC-bearing mice (Fig. 6p).”

CD8α mAb cotreatment aggravated tumor burden by eliminating CD8+ T cells in both Hepa1-6 and H22 subcutaneous and orthotopic tumor models.

“We found that CD8α mAb cotreatment aggravated tumor burden by eliminating CD8+ T cells in both Hepa1-6 and H22 subcutaneous and orthotopic tumor models (Supplementary Fig. 8a–h).”

Co-treatment with Prdm1 overexpression and PD-1 mAb significantly increased activity (GZMB+) of infiltrated CD8+ T cells compared with Prdm1 overexpression group.

“Flow cytometry analysis and IF staining revealed that co-treatment with Prdm1 overexpression and PD-1 mAb significantly increased the infiltration of CD8+ T cells and the activity (GZMB+) of infiltrated CD8+ T cells, and reduced the exhaustion of infiltrated CD8+ T cells as reflected by PD-1+ staining compared with that in the Prdm1 overexpression group (Fig. 6i and Supplementary Fig. 7a–c).”

USP22-induced deubiquitination of SPI1 dominantly enhanced PD-L1 transcription.

“Herein, we found that USP22-induced deubiquitination of SPI1 dominantly enhanced PD-L1 transcription.”

In HCC tissues, the protein levels of BLIMP1, USP22, and SPI1 were negatively associated with CD8+ T cell infiltration.

“Our TMA results revealed that the protein levels of BLIMP1, USP22, and SPI1 were positively associated with the PD-L1 protein levels and were negatively associated with CD8+ T cell infiltration in HCC tissues (Fig. 7a, b).”

PRDM1 stabilizes SPI1 via USP22.

“Thus, PRDM1 was inferred to stabilize SPI1 via USP22.”

• AssumptionRests on an assumption: The ubiquitin-proteasome system enhances proteasomal degradation of its target proteins.
• AssumptionRests on an assumption: Deubiquitinating enzymes control ubiquitin-dependent pathways by separating protein-ubiquitin bonds.

T cell-mediated immune evasion impedes the anti-tumor effect of PRDM1 overexpression.

“Thus, T cell-mediated immune evasion impedes the anti-tumor effect of PRDM1 overexpression.”

PRDM1 impacts PD-L1 expression through its endogenous promoter.

“suggesting that PRDM1 impacts PD-L1 expression through its endogenous promoter”

PRDM1 enhances USP22 transcription.

“Mechanistically, PRDM1 enhances USP22 transcription, thus reducing SPI1 protein degradation through deubiquitination, which enhances PD-L1 transcription.”

• AssumptionRests on an assumption: The role of the PRDM family in regulating anti-tumor immunity and immune molecules in HCC cells remains largely unknown.

Prdm1 might promote immune evasion by upregulating PD-L1.

“Thus, Prdm1 might promote immune evasion by upregulating PD-L1.”

• AssumptionRests on an assumption: T cell exhaustion due to PD-L1/PD1 inhibitory receptor signaling impairs T cell proliferation and effector functions.

PRDM1 could modulate SPI1 protein stability through a proteasome-dependent pathway.

“thus, PRDM1 could modulate SPI1 protein stability through a proteasome-dependent pathway (Fig. 4c and Supplementary Fig. 4f).”

PRDM1 may affect SPI1 protein stability.

“suggesting that PRDM1 may affect SPI1 protein stability”

Binding sites 3 and 4 are critical for activating PD-L1 transcription by SPI1.

“Thus, binding sites 3 and 4 are critical for activating PD-L1 transcription by SPI1.”

PRDM1-induced PD-L1 upregulation occurred at the transcriptional level.

“These data confirmed that PRDM1-induced PD-L1 upregulation occurred at the transcriptional level.”

The C-terminal ubiquitin-specific peptidase domain of USP22 is the dominant regulator of binding between USP22 and SPI1.

“Thus, the C-terminal ubiquitin-specific peptidase domain of USP22 is the dominant regulator of binding between USP22 and SPI1.”

PRDM1 restrains SPI1 polyubiquitination and proteasomal degradation.

“These results suggest that PRDM1 restrains SPI1 polyubiquitination and proteasomal degradation.”

• AssumptionRests on an assumption: The ubiquitin-proteasome system enhances proteasomal degradation of its target proteins.
• AssumptionRests on an assumption: Deubiquitinating enzymes control ubiquitin-dependent pathways by separating protein-ubiquitin bonds.

PRDM1 overexpression potentiates T cell exhaustion.

“We noticed that CD8+ cytotoxic T cell subpopulations were predominant in patient 1, who showed low expression of PRDM1 in tumor cells, whereas patient 2 showed a higher percentage of CD8+LAG3+ exhausted T cells and CD4+FOXP3+ Tregs, supporting the notion that PRDM1 overexpression potentiates T cell exhaustion (Fig. 8e).”

SPI1 may serve as a pivotal transcriptional factor that enhances PD-L1 expression by acting as a downstream effector of PRDM1.

“Proteomics analysis revealed that SPI1 may serve as a pivotal transcriptional factor that enhances PD-L1 expression by acting as a downstream effector of PRDM1.”

PRDM1/BLIMP1 overexpression combined with PD-(L)1 mAb treatment provides a therapeutic strategy for the treatment of patients with HCC.

“Furthermore, PRDM1/BLIMP1 overexpression combined with PD-(L)1 mAb treatment provides a therapeutic strategy for the treatment of patients with HCC (Fig. 7e).”

• AssumptionRests on an assumption: Only a fraction of patients with HCC respond to PD-(L)1 mAb therapy.

Tumoral Prdm1 expression in HCC might be used as a predictive biomarker for superior immunotherapeutic efficacy.

“Collectively, tumoral Prdm1 expression in HCC might be used as a predictive biomarker for superior immunotherapeutic efficacy.”

Prdm1 overexpression contributes to the increased efficacy of PD-1 mAb therapy.

“Collectively, these data confirm that Prdm1 overexpression shapes an immunosuppressive microenvironment by upregulating PD-L1 expression, thus contributing to the increased efficacy of PD-1 mAb therapy.”

Prdm1 overexpression shapes an immunosuppressive microenvironment by upregulating PD-L1 expression.

“Collectively, these data confirm that Prdm1 overexpression shapes an immunosuppressive microenvironment by upregulating PD-L1 expression, thus contributing to the increased efficacy of PD-1 mAb therapy.”

PD-(L)1 blockade conquers HCC resistance to PRDM1 overexpression by overcoming immune surveillance.

“We thus propose that PD-(L)1 blockade conquers HCC resistance to PRDM1 overexpression by overcoming immune surveillance.”

This study uncovers a regulatory mechanism for tumoral PD-L1 expression through the PRDM1-USP22-SPI1 axis.

“Thus, this study uncovers a regulatory mechanism for tumoral PD-L1 expression through the PRDM1-USP22-SPI1 axis.”

PRDM1 overexpression combined with PD-(L)1 mAb treatment provides a therapeutic strategy for HCC treatment.

“Furthermore, PRDM1 overexpression combined with PD-(L)1 mAb treatment provides a therapeutic strategy for HCC treatment.”

CD8+ T cells are indispensable for the synergistic effect of co-treatment with Prdm1 overexpression and PD-1 mAb.

“Thus, CD8+ T cells are indispensable for the synergistic effect of co-treatment with Prdm1 overexpression and PD-1 mAb.”

PRDM1 might suppress the proliferation ability of HCC cell lines by mediating tumor cell death.

“Immune regulatory parameters such as positive regulation of immune response, response to interferon-gamma, and lymphocyte activation, were significantly enriched. Regulation of the cell death pathway was also significantly enriched, indicating that PRDM1 might suppress the proliferation ability of HCC cell lines by mediating tumor cell death (Fig. 8f).”

The results confirmed that PRDM1 overexpression contributes to the therapeutic effects of PD-1 mAb therapy.

“The results confirmed our hypothesis that PRDM1 overexpression contributes to the therapeutic effects of PD-1 mAb therapy.”