HRS mediates tumor immune evasion by regulating proteostasis-associated interferon pathway activation

SUMMARY

By sorting receptor tyrosine kinases into endolysosomes, the endosomal sorting complexes required for transport (ESCRTs) are thought to attenuate oncogenic signaling in tumor cells. Paradoxically, ESCRT members are upregulated in tumors. Here, we show that disruption of hepatocyte growth factor-regulated tyrosine kinase substrate (HRS), a pivotal ESCRT component, inhibited tumor growth by promoting CD8⁺ T cell infiltration in melanoma and colon cancer mouse models. HRS ablation led to misfolded protein accumulation and triggered endoplasmic reticulum (ER) stress, resulting in the activation of the type I interferon pathway in an inositol-requiring enzyme-1α (IRE1α)/X-box binding protein 1 (XBP1)-dependent manner. HRS was upregulated in tumor cells with high tumor mutational burden (TMB). HRS expression associates with the response to PD-L1/PD-1 blockade therapy in melanoma patients with high TMB tumors. HRS ablation sensitized anti-PD-1 treatment in mouse melanoma models. Our study shows a mechanism by which tumor cells with high TMB evade immune surveillance and suggests HRS as a promising target to improve immunotherapy.

Graphical abstract

In brief

Zhang et al. show that HRS protects tumor cells from anti-tumor immunity. HRS ablation leads to the accumulation of misfolded proteins and ER stress in tumor cells, which results in activation of the type I interferon pathway and anti-tumor immunity. HRS depletion stimulates anti-tumor immunity and sensitizes tumors to anti-PD-1 therapy.

INTRODUCTION

In cells, proteostasis is maintained through two major quality control systems: molecular chaperones that facilitate protein folding and refolding and proteasomes and endolysosomes that degrade terminally misfolded proteins.^1,2 Error-prone protein synthesis is exacerbated in cancer cells with mutations and rapid protein translation, as well as in response to numerous intrinsic and extrinsic stresses.³ Enhanced capacity to degrade misfolded proteins is essential for tumor cell survival and oncogenic transformation.³

The endosomal sorting complexes required for transport (ESCRTs) mediate ubiquitin-dependent sorting of membrane proteins into the endolysosomes.^4–6 The ESCRT machinery consists of four distinct complexes, ESCRT-0, ESCRT-I, ESCRT-II, and ESCRT-III, and several accessory components.^5,7 These complexes are sequentially recruited from cytoplasm to the surface of endosomes, where they sort protein cargo by invagination of limiting membranes toward the lumen, followed by subsequent fission to form intraluminal vesicles (ILVs). The ILVs are either transported to lysosomes for degradation or are released at the plasma membrane as exosomes.^8,9 The ESCRT machinery was thought to be tumor suppressive, as it mediates the lysosomal degradation of activated receptor tyrosine kinases (RTKs) such as epidermal growth factor receptor (EGFR), thus attenuating oncogenic signaling.^10–14 Surprisingly, studies in human cancer tissues showed that members of the ESCRTs such as hepatocyte growth factor-regulated tyrosine kinase substrate (HRS, also known as HGS) and tumor susceptibility gene 101 (TSG101) were mostly upregulated in tumors.^15–18

Here, we investigated the function of HRS, which is a key component of the ESCRT that is upregulated in a wide variety of tumor types. Ablation of HRS in tumor cells led to the accumulation of destabilized/misfolded protein, disrupted proteostasis, endoplasmic reticulum (ER) stress, and unfolded protein response (UPR), which in turn activated the type I interferon (IFN) pathway. HRS was upregulated in tumor cells with high mutational burden and was associated with anti-PD-1 therapeutic response in the patients with high tumor mutational burden (TMB) melanomas. In mouse models, disruption of HRS resulted in increased T cell infiltration and anti-tumor immunity and enhanced the response to anti-PD-1 treatment. Our study demonstrates a role of HRS in tumor immune evasion and suggests that targeting HRS in tumors with high TMB helps to improve anti-PD-1 treatment.

RESULTS

Ablation of HRS in tumor cells elicits anti-tumor immunity

To understand the role of ESCRT machinery in cancer, we analyzed the mRNA expression of 29 ESCRT and ESCRT-associated genes in various cancer types using The Cancer Genome Atlas (TCGA) patient datasets. Among the 19 pairs of tumors and matching normal tissues we examined, the mRNA levels of 7 ESCRT-related genes were significantly upregulated in more than 9 cancer types (Figure S1A). In particular, HRS was upregulated in 15 different cancer types and only downregulated in glioblastoma multiforme. We further analyzed the Pearson correlation between the ESCRTs and their associated genes and CD8A in 33 cancer types. Negative correlations were found between CD8A and many ESCRT genes. Among them, VPS37D, CHMP3, and HRS were negatively correlated with CD8A in more than 15 cancer types (Figure S1B), suggesting that ESCRT function relates to anti-tumor CD8 T cell function.

HRS is a pivotal component of ESCRT-0. We first knocked out HRS in the melanoma cell line B16F10 (Figure S2A) and established B16F10 syngeneic tumor models in C56BL/6J wild-type mice. Knockout (KO) of HRS in B16 cells significantly inhibited tumor growth (Figures 1A and 1B). In contrast, there was no significant growth difference between HRS KO and control B16 tumors in athymic nude mice (Figures 1C and 1D). We also generated HRS knockdown (KD) B16F10 (Figure S2B) and HRS KO Braf^V600E Pten^−/− (‘‘Braf Pten’’ henceforth) cells (Figure S2C). Similar observations were made for models with HRS KD B16F10 (Figures S2D–S2G) and HRS KO Braf Pten tumors (Figures S2H–S2K). These in vivo studies suggest that HRS ablation in tumor cells elicits anti-tumor immune response in mice.

Figure 1. HRS depletion inhibits tumor growth in a CD8⁺ T cell-dependent manner.

(A and B) Tumor growth (A) and survival curves (B) of C57BL/6J mice inoculated with control or HRS KO B16F10 melanoma cells.

(C and D) Tumor growth (C) and survival curves (D) of athymic nude mice inoculated with control or HRS KO B16F10 melanoma cells.

(E and F) Tumor growth (E) and survival curves (F) of C57BL/6J mice inoculated with control or HRS KD MC38 colon cancer cells.

(G and H) Tumor growth (G) and survival curves (H) of athymic nude mice inoculated with control or HRS KD MC38 colon cancer cells.

(I and J) Tumor growth (I) and survival curves (J) of C57BL/6J mice inoculated with control or HRS KO B16F10 cells, and the mice were intraperitoneally injected with isotype or anti-CD8 antibodies.

(K and L) Tumor growth (K) and survival curves (L) of C57BL/6J mice inoculated with control, HRS KO B16, B2M KO, or HRS B2M double KO (dKO) B16F10 cells. Data represent the mean ± SEM; n = 7 or 8 immunocompetent C57BL/6J mice per group, n = 5 or 7 athymic nude mice per group. The p values were determined by two-way ANOVA (A, C, E, G, I, and K) or log-rank test (B, D, F, H, J, and L).

See also Figure S2 and S3.

B16 and Braf Pten tumors are immunologically ‘‘cold’’ with low immune cell infiltration.¹⁹ The mouse colon carcinoma model MC38 with more T cell infiltration was considered a ‘‘hot’’ tumor.¹⁹ We generated HRS KD MC38 tumors (Figure S2L) and inoculated them into C56BL/6J mice. While HRS KD significantly inhibited MC38 tumor growth (Figures 1E and 1F), no difference between HRS KD and control MC38 tumors was found in athymic nude mice (Figures 1G and 1H), further suggesting the involvement of HRS in anti-tumor immunity.

We depleted CD8⁺ T cells in immunocompetent mice by intraperitoneally injecting anti-CD8 antibodies (Figures S3A and S3B). Depletion of CD8⁺ T cells partially restored HRS KO B16 tumor growth in C57BL/6J WT mice (Figures 1I and 1J). It is likely that, in addition to CD8⁺ T cells, CD4⁺ T cells and other immune cells (e.g., natural killer cells) also contribute to the HRS-depletion-mediated anti-tumor immunity. We also knocked out B2M in HRS KO B16 cells to block cell surface expression of major histocompatibility complex (MHC) class I (Figure S3C). Similar tumor growth and mouse survival were observed between HRS B2M double KO (dKO) and control B16 tumors in C57BL/6J mice (Figures 1K and 1L). This result suggests that the HRS-ablation-induced anti-tumor immunity was dependent on MHC class I recognition in tumor cells and provides further support for the close relationship between HRS and anti-tumor immunity.

Suppression of HRS increases T cell infiltration into tumors

Immunohistochemistry (IHC) staining showed a significant increase in CD45⁺ immune cells and CD8⁺ T cells in HRS KO B16F10 tumors (Figures 2A and 2B). Flow cytometry analysis (Figure S4) showed increased CD4⁺ and CD8⁺ T cell infiltration in HRS KO B16 melanoma tumors (Figure 2C). By analyzing the expression of the proliferation marker Ki67 and cytotoxic factor granzyme B (GzmB), we found higher levels of Ki67⁺ and GzmB⁺ CD8⁺ tumor-infiltrating lymphocytes (TILs) in HRS KO B16 tumors (Figure 2D). A similar observation was made in HRS KD MC38 colon tumors. Increased CD4⁺ and CD8⁺ T cell infiltration and a higher level of GzmB⁺ CD8⁺ TILs were noticed in HRS KD MC38 colon tumors (Figures 2E and 2F). Together, our results suggest that HRS depletion in tumor cells increases T cell infiltration and enhances anti-tumor activities of TILs in the tumors.

Figure 2. HRS depletion promotes T cell infiltration into tumors.

(A and B) Immunohistochemistry (A) with quantification (B) of CD45⁺ immune cells (left) and CD8⁺ T cells (right) in control and HRS KO B16F10 tumors.

(C) The infiltration of CD4⁺ (left) and CD8⁺ (right) lymphocytes (tumor-infiltrating lymphocytes [TILs]) from control and HRS KO B16F10 tumors in C57BL/6J mice.

(D) The expression of Ki-67 (left) and GzmB (right) by CD8⁺ TILs from control and HRS KO B16F10 tumors.

(E) The infiltration of CD4⁺ (left) and CD8⁺ (right) TILs from control and HRS KD MC38 colon tumors in C57BL/6J mice.

(F) The expression of Ki-67 (left) and GzmB (right) by CD8⁺ TILs from control and HRS KD MC38 colon tumors.

Data represent the mean ± SD; n = 7 or 8 mice per group. The p values were determined by unpaired Student’s t test (B–F).

See also Figure S4.

HRS ablation activates type I IFN pathway in tumor cells

We next performed RNA sequencing (RNA-seq) study on HRS KD and control B16 cells with Gene Ontology (GO) enrichment analysis. Immune-related processes including innate immune response, cellular response to IFN-β, response to virus, and defense response to virus were ranked among top 10 GO terms of biological processes (Figure 3A), indicating enhanced type I IFN signal in HRS KD tumors. In addition, genes associated with hallmark IFN-α response were enriched in HRS KD cells and homograft tumors as analyzed by gene set enrichment analysis (GSEA) (Figures S5A and S5B).

Figure 3. HRS depletion activates type I IFN signaling in tumor cells.

(A) Control and HRS KD B16F10 cells were subjected to RNA-seq analysis. Gene Ontology (GO) analysis of upregulated genes (log2 fold change [log2(FC)] > 1 and false discovery rate [FDR] < 0.05) in HRS KD vs. control B16 cells was performed. Top 10 terms related to biological process (BP) are shown in a dot map.

(B and C) mRNA expression of IFN-β and ISGs in control and HRS KO B16F10 (B) and Braf Pten (C) cells.

(D) The expression of phospho-TBK1, TBK1, RIG-I, MDA5, and ISG15 in control and HRS KO B16 (left) and Braf Pten (right) cells.

(E) The expression of phospho-TBK1, TBK1, RIG-I, MDA5, and ISG15 in control and HRS KO B16 cells treated with isotype or anti-IFNAR antibodies.

(F) mRNA expression of CXCL10 and ISG15 in control and HRS KO B16 cells with isotype or anti-IFNAR antibody treatment.

(G) mRNA expression of IFN-β, CXCL10, and ISG15 in control and HRS KO B16 cells with or without BX-795, a specific inhibitor of TBK1.

(H and I) KD of RIG-I (top) and MDA5 (bottom) was performed in control and HRS KO B16 cells (H). mRNA expression of IFN-β and ISGs was examined (I).

(J) Tumor growth in C57BL/6J mice inoculated with control or HRS KD B16 cells and treated with anti-IFNAR antibodies intraperitoneally.

(K) The infiltration of CD4⁺ (left) and CD8⁺ (right) TILs in transplanted B16 tumors (n = 6 or 7 per group) harvested from C57BL/6J mice.

(L) The expression of Ki-67 (left) and GzmB (right) by CD8⁺ TILs in (K).

Data represent the mean ± SD from three independent experiments. Data from animal study represent the mean ± SEM; n = 6 or 7 mice per group. The p values were determined by unpaired Student’s t test (B and C), one-way ANOVA with Sidak’s multiple comparison tests (F, G, I, K, and L), or two-way ANOVA with Dunnett’s multiple comparison tests (J).

See also Figures S5 and S6.

Increased mRNA expression of IFN-β and downstream genes, including ISG15, CXCL10, IFIH1 (a.k.a. MDA5), and DDX58 (a.k.a. RIG-I), was found in HRS KO B16 and Braf Pten cells (Figures 3B and 3C). Type I IFN production is induced after the sensing of nucleic acids by pattern-recognition receptors (PRRs), such as Toll-like receptors (TLRs), RIG-I-like family of receptors (RLRs), and cyclic GMP-AMP synthase (cGAS).²⁰ TBK1 is a converging point for signaling of all three pathways. We found increased levels of phosphorylated (phospho)-TBK1, MDA5, RIG-I, and ISG15 (Figure 3D), but not TLR3, cGAS, or phospho-STING, in HRS KO cells (Figure S5C). Blocking the type I IFN pathway decreased phospho-TBK1 expression and MDA5, RIG-I, and ISG15 protein expression (Figure 3E), as well as the mRNA expression of ISG15 and CXCL10 in HRS KO B16 cells (Figure 3F). TBK1 inhibitor BX-795 downregulated the mRNA levels of IFN-β, ISG15, and CXCL10 in HRS KO cells (Figure 3G). The same observations were made using HRS KO Braf Pten cells (Figures S5D–S5F). KD of RIG-I or MDA5 (Figure 3H) significantly decreased mRNA expression of IFN-β, ISG15, and CXCL10 in HRS KO B16 (Figure 3I) and Braf Pten cells (Figures S5G and S5H). Consistent with the results, HRS KO B16 cells were more sensitive to IFN-β treatment (Figures S6A and S6B) and exhibited more potent growth inhibition upon IFN-β treatment (Figures S6C and S6D). IFNAR1 blocking also restored the growth of HRS KD or HRS KO B16 tumors (Figures 3J and S6E), which was accompanied by decreased T cell infiltration (Figure 3K). Furthermore, the amounts of Ki67⁺ and GzmB⁺ TILs were also decreased (Figure 3L). The results suggest that enhanced anti-tumor T cell immunity upon HRS suppression is mediated by the activation of type I IFN signaling in tumor cells.

IRE1α/XBP1 is required for the activation of type I IFN pathway in HRS-depleted tumor cells

RNA-seq analysis of HRS KD B16 cells showed that genes engaged in the UPR were ranked as the top GO term on molecular function (Figure 4A). In addition, active RNA splicing and mRNA and rRNA processing may be associated with RIG-I/MDA5 activation in HRS KO cells, as misprocessed host RNAs can be recognized by RIG-I like receptors.²¹ Genes associated with UPR were enriched in HRS KD B16 cells by GSEA (Figure S7A), with the heatmap showing increased UPR target genes in HRS KD B16 cells and tumors (Figures S7B and S7C). These results were further verified in HRS KO B16 and Braf Pten cells (Figures 4B and 4C). Moreover, IHC staining showed elevated expression of BiP (a.k.a. HSPA5 or GRP78) in HRS KO B16 tumors (Figure S7D).

Figure 4. HRS ablation induced the unfolded protein response (UPR) required for type I IFN pathway activation.

(A) HRS KD and control B16 cells were subjected to RNA-seq analysis. GO analysis of upregulated genes (log2(FC) > 1 and FDR < 0.05) in HRS KD vs. control B16 cells was performed. Top 10 terms related to molecular function (MF) are shown in a dot map.

(B and C) mRNA expression of UPR-related genes in control and HRS KO B16F10 (B) and Braf Pten (C) cells.

(D) The expression of BiP, phospho-IRE1α, IRE1α, XBP1s, phospho-eIF2α, eIF2α, ATF4, and ATF6 in control and HRS KO B16F10 (left) and Braf Pten (right) cells.

(E) Immunofluorescence of ATF4 and ATF6 in control and HRS KO B16F10 and Braf Pten cells. Cell nuclei were stained with DAPI. Scale bar, 50 μm.

(F) Percentage of cells with ATF4 detected in nuclei in B16F10 (left) and Braf Pten (right) cells as shown in (E).

(G) The levels of BiP, XBP1s, MDA5, RIG-I, and ISG15 in control and HRS KO B16F10 cells with STF-083010, an inhibitor blocking IRE1α/XBP1 pathway.

(H) mRNA expression of IFN-β, ISG15, and CXCL10 in control and HRS KO B16F10 cells, which were treated with STF-083010.

(I) The levels of HRS, IRE1α, BiP, XBP1s, MDA5, RIG-I, and ISG15 in control or HRS-ablated B16F10 cells with IRE1 or XBP1 knocked out.

(J) mRNA expression of IFN-β, ISG15, and CXCL10 in control and HRS KO B16F10 cells in which IRE1 or XBP1 was knocked out.

(K) Tumor growth in C57BL/6J mice inoculated with IRE1 KO vs. HRS IRE1 dKO B16F10 cells (left) and XBP1 KO vs. HRS XBP1 dKO B16F10 cells (right).

(L) Immunohistochemical staining of CD45 and CD8 in HRS KO, HRS IRE1 dKO, and HRS XBP1 dKO B16F10 tumors.

(M) Quantification of CD45⁺ immune cells (left) and CD8⁺ T cells (right) in the B16 tumors as described in (L).

(N) The infiltration of CD4⁺ (left) and CD8⁺ (right) TILs from the transplanted B16 tumors in C57BL/6J mice.

(O) The expression of Ki-67 (left) and GzmB (right) by CD8⁺ TILs in (N).

Data represent the mean ± SD from three independent experiments. Data from animal study represent the mean ± SEM; n = 7 or 8 per group. The p values were determined by unpaired Student’s t test (B, C, and F), one-way ANOVA with Sidak’s multiple comparison tests (H, J, and M–O), and two-way ANOVA with Dunnett’s multiple comparison tests (K).

See also Figure S7.

The mammalian UPR consists of three signaling branches: the inositol-requiring enzyme-1α (IRE1α)/X-box binding protein 1 (XBP1) pathway, the protein kinase RNA-like ER kinase (PERK)/activating transcription factor 4 (ATF4) pathway, and the ATF6a pathway.²² BiP was increased in HRS KO B16 and Braf Pten cells. Moreover, phospho-IRE1α (p-IRE1α), spliced XBP1 (XPB1s), and phospho-a subunit of translation initiation factor eIF2 (p-eIF2α) were increased in these cells (Figure 4D). Immunofluorescence imaging also showed a marked increase of ATF4, but not ATF6, in the nuclei (Figures 4E and 4F). The data suggest that HRS ablation in tumor cells induced ER stress and subsequent UPR, and the IRE1α/XBP1 and PERK/ATF4 signaling branches were activated.

Increased endonuclease activity of IRE1α under ER stress leads to increased cleavage of mRNA to reduce the translation burden, a process known as regulated IRE1-dependent decay (RIDD).²³ RIG-I/MDA5 sense small RNA fragments generated by RIDD, leading to their activation and, if the cleaved mRNA cannot be degraded completely and timely, triggering innate immunity.²⁴ Treatment of cells with STF-083010, an IRE1α/XBP1 pathway inhibitor, led to downregulation of MDA5, RIG-I, and ISG15 protein expression in HRS KO B16 (Figure 4G) and Braf Pten cells (Figure S7E) and inhibited IFN-β, ISG15, and CXCL10 mRNA expression in both cell lines (Figures 4H and S7F). Another IRE1α RNase inhibitor, 4μ8C, that blocked Xbp1 splicing also showed such an inhibitory effect (Figures S7G and S7H). We further knocked out IRE1 and XBP1 in HRS KO tumor cells, respectively. Protein levels of MDA5, RIG-I, and ISG15 and mRNA expression levels of IFN-β, ISG15, and CXCL10 in HRS XBP1 and HRS IRE1 dKO tumor cells were much more reduced than those in HRS KO tumor cells (Figures 4I and 4J). Similar results were obtained using MC38 and human melanoma cells (WM9 and A375) (Figures S7I–S7L). Re-expression of hemagglutinin (HA)-tagged HRS in HRS KO B16 cells attenuated type I IFN response (Figure S7M). p-IRE1α and XPB1s were downregulated in the rescued cells. Together, these results suggest that HRS-ablation-induced ER stress triggers the type I IFN signal in an IRE1α/XBP1-dependent manner. KO of IRE1 or XBP1 suppressed tumor growth in immunocompetent mice (Figure S7N). IRE1 KO B16 tumors showed slower tumor growth than XBP1 KO tumors, which might be due to IRE1α not only mediating XBP1 splicing but also regulating the MAPK/JNK pathway and the RIDD process.^25,26 Compared with single KO of IRE1 or XBP1, additional HRS KO did not cause significant growth inhibition of B16 tumors (Figure 4K). IHC staining showed an increase in CD45⁺ immune cell and CD8⁺ T cell infiltration in HRS KO tumors, which was significantly decreased in HRS IRE1 dKO and HRS XBP1 dKO B16 tumors (Figures 4L and 4M). Flow cytometry analysis demonstrated that CD4⁺ and CD8⁺ T cell infiltration in HRS KO B16 tumors was much attenuated in either HRS IRE1 dKO or HRS XBP1 dKO tumors (Figure 4N). Moreover, a decrease of Ki67⁺ and GzmB⁺ CD8⁺ T cell populations was found in HRS IRE1 dKO and HRS XBP1 dKO B16 tumors compared with HRS KO B16 tumors (Figure 4O). Taken together, these results suggest that the IRE1α/XPB1 pathway is required for the anti-tumor immunity elicited by HRS depletion in cancer cells.

Ablation of HRS promotes the accumulation of misfolded proteins in tumor cells

HRS recognizes and sorts protein cargo into ILVs for degradation in lysosomes or secretion through exosomes.⁷ HRS ablation also disrupts the autophagic and lysosomal pathway function^27,28 and therefore may lead to the accumulation of misfolded proteins. Consistently, HRS KO induced the accumulation of both soluble and precipitated ubiquitinated proteins in B16 and Braf Pten cells, which was further exacerbated with the treatment of thapsigargin (Tg), a commonly used ER stress inducer²⁹ (Figures 5A and 5B). The accumulation of ubiquitinated proteins was mitigated in the HRS-rescued B16 cells (Figure S8A). We further induced protein misfolding stress with Tg (5 nM), the proteasome inhibitor MG132 (2 μM), or the HSP70 inhibitor VER-155008 (20 μM) and found that significantly more ubiquitin-positive puncta were induced by MG132 and VER-155008 in HRS KO cells (Figures 5C and 5D). For Tg treatment, while the numbers of ubiquitin-positive puncta were similar in individual cells for the control and HRS KO B16 groups, the size of the puncta is larger for the HRS KO cells (Figures 5D, S8B, and S8C). Similarly, in control Braf Pten cells, MG132 did not induce ubiquitin-positive puncta, whereas it significantly increased the ubiquitin-positive puncta in HRS KO tumor cells (Figures S8D and S8E).

Figure 5. HRS ablation promotes misfolded protein accumulation in tumor cells.

(A and B) Control or HRS KO B16F10 (A) and Braf Pten (B) cells were treated with thapsigargin (Tg). Cell lysates were fractionated into soluble (‘‘S’’) and precipitated (‘‘P’’) fractions. Ubiquitinated proteins were probed with anti-ubiquitin antibodies. Quantification of ubiquitinated proteins is shown to the right. The blots of soluble and precipitated fractions in control groups were normalized to 1, respectively.

(C) Immunofluorescence of ubiquitin in control and HRS KO B16 cells treated with Tg, MG132, and VER-155008, respectively. Cell nuclei were stained with DAPI. Scale bar, 20 μm.

(D) Quantification of the ubiquitin-positive puncta in control and HRS KO B16 cells with indicated treatment as shown in (C).

(E) Ubiquitinated proteins in the same amounts of exosomes from control or HRS KO B16 and Braf Pten cells were analyzed by immunoblotting. CD63 was used as an exosome marker.

(F) Quantification of ubiquitinated proteins in exosomes from B16 (left) and Braf Pten (right). The levels of ubiquitinated proteins in exosomes from control groups were normalized to 1.

(G) The human A375 melanoma cells were transfected with firefly luciferase (Fluc)-GFP or the mutant FlucDM-GFP. The co-localization of Fluc-GFP or FlucDM-GFP (green) with HRS (red) in A375 cells with MG132 treatment was analyzed. Nuclei were stained with DAPI. Scale bar, 20 μm.

(H) HRS was knocked down in A375 cells with stable expression of FlucDM-GFP. The levels of soluble and precipitated FlucDM-GFP were analyzed.

(I) Quantification of Fluc-GFP or FlucDM-GFP in the soluble (left) and precipitated (right) fractions from (H). GADPH was used as a control for soluble fraction loading. The blots of soluble and precipitated fractions in control groups were normalized to 1, respectively.

(J) Specific activities of Fluc in the soluble extracts from A375 cells that were stably transfected with Fluc-GFP or its mutant. Activities of Fluc in control A375 cells were set as 100%.

Data represent the mean ± SD from three or four independent experiments. The p values were determined by one-way ANOVA with Sidak’s multiple comparison tests (A and B), one-sample t test (F and I), and unpaired Student’s t test (D and J).

See also Figure S8 and S9.

PROTEOSTAT dye becomes highly fluorescent upon binding to protein aggregates and has been used to detect aggresomes in vitro and in vivo.³⁰ Using PROTEOSTAT dye, we observed the ubiquitin-positive aggresomes in HRS KO B16 cells but not in control cells. These aggresomes were further increased in HRS KO cells with Tg, MG132, or VER-155008 treatment (Figure S8F).

HRS also functions in exosome biogenesis.⁸ We collected the exosomes from control and HRS KO B16 cells. Nanoparticle tracking analysis (NTA) indicates that HRS KO reduced the number of exosomes released from B16 cells (Figures S8G and S8H). Moreover, the secretion of ubiquitinated proteins through these exosomes was also significantly decreased for HRS KO cells (Figures 5E and 5F).

To further examine protein misfolding in cells, we took advantage of a structurally destabilized mutant firefly luciferase (FlucDM-GFP) as a reporter.³¹ We observed aggregates of FlucDM-GFP (Figure S8I) and a co-localization of FlucDM-GFP puncta and HRS in MG132-treated human and mouse melanoma cells, whereas the stable protein control Fluc-GFP showed mostly diffuse cytosolic distribution (Figures 5G and S9). The observed co-localization supports the role of HRS in the clearance of misfolded proteins, although we cannot totally rule out the possibility that some of the HRS was ‘‘trapped’’ by these protein aggregates. HRS KD led to the accumulation of both soluble and precipitated FlucDM-GFP in human melanoma A375 cells (Figures 5H and 5I). Depletion of HRS in cells transfected with stable Fluc caused increased Fluc activity. However, Fluc activity was similar between control and HRS KD A375 cells transfected with FlucDM-GFP (Figure 5J), suggesting that these accumulated proteins were dysfunctional and probably aggregated. These results suggest that depletion of HRS affects proteostasis and leads to increased protein aggregation in tumor cells.

Tumor cells with high mutational burden were more sensitive to HRS depletion

Tumors with high TMB are more responsive to immune check-point blockade (ICB), as these tumors are more likely to present neoantigens recognizable by the host immune system. However, many patients with high-TMB tumors do not benefit from ICB treatment.^32–34 A recent study showed that the genome-wide mutations led by the defects in DNA mismatch repair (MMR) result in abundant misfolded proteins in cells.³⁵ We speculate that tumors with deficient MMR (dMMR) are more dependent on HRS for the clearance of unfolded/misfolded proteins. MSH2 is a member of the MutSα complex responsible for repairing single base-base mismatches and is frequently lost in cancer cells.^36–38 Therefore, we generated MSH2-deficient B16F10 and Braf Pten cells by CRISPR and passaged these cells to allow the accumulation of mutations as previously described.³⁹ Western blotting showed that the ubiquitinated proteins accumulated in MSH2 KO tumor cells (Figures 6A and 6B). Increased levels of both soluble and precipitated HRS were detected in MSH2 KO cells, suggesting that more HRS interacted with aggregated proteins in MSH2 KO cells (Figures 6C and 6D). We also performed experiments taking advantage of a panel of human endometrial and colorectal microsatellite-instable (MSI) (HCT-116, MFE-296, and RKO) and microsatellite-stable (MSS) (SW948, MFE-280, and KLE) cells as used in a previous study.³⁵ More ubiquitinated proteins were found in MSI cells than in MSS cells (Figures S10A and S10B). Furthermore, higher levels of HRS in the precipitated fractions were found in the MSI cancer cell lines than in MSS cells (Figures 6E and 6F), suggesting that HRS is adaptively upregulated in dMMR cancer cells.

Figure 6. Tumor cells with high mutational burden are more sensitive to HRS depletion.

(A and B) The expression of soluble and precipitated ubiquitinated proteins in control or MSH2 KO B16 and Braf Pten cells (A). Quantification of ubiquitinated proteins from B16 (left) and Braf Pten (right) cells is shown in (B). The blots of soluble and precipitated ubiquitinated proteins in control groups were normalized to 1, respectively.

(C and D) The expression (C) of HRS in the soluble and precipitated protein fractions from control and MSH2 KO B16 and Braf Pten cells. GADPH was used as a loading control for the soluble proteins. Quantifications are shown in (D). The expressions of soluble and precipitated HRS in control groups were normalized to 1, respectively.

(E and F) The expression (E) of HRS in the soluble and precipitated protein fractions from a panel of human MSS and MSI cell lines. GADPH was used as a loading control for the soluble protein parts. Quantifications are shown in (F). The blots of soluble and precipitated HRS in MFE280 cells were normalized to 1, respectively.

(G and H) The expression (G) of HRS in the soluble (left) and precipitated (right) fractions from YUMM1.7 and YUMMER1.7 cells. GADPH was used as a loading control for the soluble proteins. Quantification is shown in (H). The blots of soluble and precipitated HRS in control groups were normalized to 1, respectively.

(I and J) The expression of soluble and precipitated ubiquitinated proteins in HRS KO, MSH2 KO, and MSH2 HRS dKO Braf Pten cells. GADPH was used as a loading control for soluble proteins (I). Quantifications of soluble (left) and precipitated (right) fractions of ubiquitinated proteins are shown in (J).

(K and L) The expression of soluble and precipitated ubiquitinated proteins in YUMM1.7 and YUMMER1.7 cells. GADPH was used as a loading control for the soluble protein (K). Quantifications of soluble (left) and precipitated (right) fractions of ubiquitinated proteins are shown in (L).

(M) The expression of XBP1s, RIG-I, MDA5, and ISG15 in control and HRS KO, MSH2 KO, and MSH2 HRS dKO Braf Pten cells.

(N) mRNA expression of IFN-β, ISG15, and CXCL10 in control and HRS KO, MSH2 KO, and MSH2 HRS dKO Braf Pten cells.

(O) The expression of BiP, XBP1s, RIG-I, MDA5, and ISG15 in YUMM1.7 and YUMMER1.7 cells with or without HRS KD. GADPH was used as a loading control.

(P) mRNA expression of IFN-β, ISG15, and CXCL10 in control and HRS KD YUMM1.7, as well as YUMMER1.7 cells, as analyzed by quantitative real-time PCR.

(Q and R) Tumor growth (Q) and survival curves (R) of C57BL/6J mice inoculated with control, MSH2 KO, HRS KO, or MSH2 HRS dKO Braf Pten cells. Tumor growth curves in the control group were ended at day 16 due to the death/euthanasia of mice.

Data represent the mean ± SD from at least three independent experiments. Data from animal study represent the mean ± SEM; n = 7 mice per group. The p values were determined by one-sample t test (B, D, H, J, and L), nested t test (F), one-way ANOVA with Sidak’s multiple comparison tests (N and P), two-way ANOVA with Dunnett’s multiple comparison tests (Q), or log-rank test (R).

See also Figure S10.

YUMM1.7 is a cell line designated from the genetically engineered Braf Pten mouse melanoma model, and YUMMER1.7 was generated from YUMM1.7 after ultraviolet B radiation.^40,41 In comparison to the parental YUMM1.7 cells, YUMMER1.7 was shown to have a significant increase in mutational load and elicited a robust immune response.⁴⁰ We observed increased HRS expression in YUMMER1.7 cells when compared to YUMM1.7 cells (Figures 6G and 6H). Tg treatment, which induced the accumulation of unfolded proteins, also increased HRS expression in melanoma cells (Figures S10C and S10D).

HRS KO led to an increased accumulation of ubiquitinated proteins in MSH2 KO Braf Pten cells (Figures 6I and 6J) and YUMMER1.7 cells (Figures 6K and 6L). Consistently, higher XBP1s expression was observed in these cells (Figures 6M and 6O), suggesting increased ER stress. HRS KO led to increased expression of MDA5, RIG-I, and ISG15 (Figure 6M) and increased mRNA expression of IFN-β, ISG15, and CXCL10 in MSH2 HRS dKO cells (Figure 6N). Similar results were found in YUMMER1.7 cells (Figures 6O and 6P).

To test whether HRS depletion is more efficient in inhibiting the growth of high-TMB tumors, we established mouse tumor models by subcutaneously inoculating MSH2 HRS dKO Braf Pten melanoma cells into immunocompetent C56BL/6J mice. As expected, MSH2 deficiency in Braf Pten cells slowed tumor growth. HRS KO in these cells further suppressed tumor growth (Figures 6Q and 6R). Together, these results suggest that tumor cells with a higher mutational burden accumulated more destabilized/misfolded proteins and were more dependent on HRS for their clearance, and more type I IFN signals were induced upon HRS KO in these tumors.

HRS expression is associated with response to anti-PD-1 therapy in patients with melanoma, and depletion of HRS enhances anti-PD-1 treatment

Riaz et al. for the first time reported the relationship between the gene expression programs and clinical response of patients with melanoma with different TMBs.⁴² We analyzed HRS expression using the dataset from this cohort of patients. HRS levels positively correlated with tumor mutational load (Figure 7A). Moreover, patients with melanoma with low HRS expression had a better overall survival (OS) probability than those with high HRS expression in the same cohort (Figure 7B). For patients with high tumor mutation load, higher HRS expression was found in non-responders to anti-PD-1 treatment (patients with stable disease [SD] or progressive disease [PD] after therapy) than responders (patients with complete response [CR] or partial response [PR] after therapy) (Figure 7C), suggesting an association of HRS expression with response to anti-PD-1 therapy in patients with high-TMB melanoma.

Figure 7. HRS level is associated with response to anti-PD-1 therapy in patients with melanoma, and depletion of HRS enhanced anti-PD-1 treatment in murine melanoma models.

(A) Pearson correlation of the HRS mRNA expression to melanoma mutational load.

(B) Kaplan-Meier curves for overall survival for patients with melanoma with low and high HRS mRNA expression.

(C) The HRS mRNA expression levels of the non-responders (‘‘NR’’) and responders (‘‘R’’) with low (left) and high (right) mutational load (‘‘ML’’).

(D and E) Tumor growth (D) and survival curves (E) of C57BL/6J mice inoculated with control or HRS KO Braf Pten cells and treated with isotype or anti-PD-1 antibodies. Tumor growth curves in control and PD-1 blockade (‘‘αPD-1’’) treatment groups were ended at day 16 due to the death/euthanasia of the mice.

(F and G) Tumor growth (F) and survival curves (G) of C57BL/6J mice inoculated with MSH2 KO or MSH2 HRS dKO Braf Pten cells and treated with isotype or anti-PD-1 antibodies. Tumor growth curves in control and αPD-1 therapy groups were ended at day 23 due to the death/euthanasia of the mice.

Data represent the mean ± SEM; n = 7 or 8 mice per group. The p values were determined by Pearson correlation analysis (A), log-rank test (B, E, and G), unpaired Welch’s t test (C), or two-way ANOVA (D and F).

Next, we examined whether HRS KO enhances PD-1 blockade treatment in mouse melanoma models. Anti-PD-1 antibody alone barely had any inhibitory effect on Braf Pten tumor growth. However, in the HRS KO strain, anti-PD-1 antibodies markedly reduced tumor growth (Figures 7D and 7E). Tumor growth inhibition was further acerbated in MSH2 HRS dKO tumors with PD-1 blockade treatment (Figures 7F and 7G). In fact, upon anti-PD-1 treatment, three out of the seven MSH2 HRS dKO strains showed no tumor growth during the 2 months of observation. The data suggest that HRS suppression sensitizes tumors to anti-PD-1 treatment and improves the efficacy of anti-PD-1 antibodies in tumors with high TMB.

DISCUSSION

Studies in cultured cells showed that the ESCRT machinery mediates the degradation of cell surface signaling molecules such as EGFR in endolysosomes, thereby attenuating oncogenic signaling in tumor cells.^12,43 Surprisingly, HRS, together with many other ESCRT proteins, were upregulated in various types of cancers,^10,15 suggesting a pro-tumorigenic role in vivo. Our data demonstrate that HRS ablation in tumor cells disrupts protein degradation in endolysosomes,^27,28 leading to the accumulation of ubiquitinated proteins, and triggers ER stress and UPR. ER stress and UPR, in turn, contribute to the activation of the type I IFN signaling pathway in an IRE1α/XBP-dependent manner, promoting anti-tumor T cell immunity in both melanoma and colon cancer mouse models.

Patients with highly TMB tumors are thought to be more likely to respond to ICB, as the tumors produce more neoantigens that can be recognized by the host immune system, allowing for tumor clearance by cytotoxic lymphocytes.^44,45 Indeed, cancer types that have higher TMB (e.g., melanoma and lung cancer) are, in general, more responsive to anti-PD-1 treatment.^46,47 However, a significant portion of patients with a high TMB do not benefit from anti-PD-1 treatment.^48,49 Using a generalized linear mixed model (GLMM), Tilk and colleagues found increased proteostatic stress in the high-mutational-load tumors.⁵⁰ Meanwhile, the machinery related to translation regulation, protein folding, and protein degradation is upregulated to mitigate and prevent protein misfolding. Consistently, our data show the accumulation of ubiquitinated proteins in MSI colon carcinoma and melanoma cells, as well as ultraviolet B (UVB)-treated melanoma cells. Notably, HRS was upregulated in high-TMB tumor cells with an increased level of ubiquitinated proteins. Analysis of clinical samples shows a positive correlation between HRS expression and TMB in patients with melanoma, further supporting that upregulation of HRS is an adaptive mechanism in response to elevated proteotoxic stress.

Interestingly, neddylation-mediated clearance of misfolded proteins was reported in MSI cancers; inhibition of neddylation induced immunogenic cell death.³⁵ This study and our results reported here suggest that, in addition to the generation of neoantigens, the accumulated destabilized/misfolded proteins in high-TMB tumor cells also affect the immunotherapy response. Our data show that loss of HRS effectively converted tumors that were resistant to PD-1 blockade to ones that are responsive in mouse melanoma models. In further support of this notion, we found that HRS depletion modestly prolonged survival of mice bearing poorly immunogenic B16F10 tumors but significantly extended animal survival in the MC38 colon cancer model, which carried a high TMB. Similarly, HRS depletion also caused significant tumor growth inhibition and prolonged animal survival in the MSH2 KO Braf Pten tumors, in which numerous mutations are developed due to the defects on DNA repair.

UPR is induced in response to proteotoxic stress. It was reported that the UPR in tumor-associated immune cells, including CD8⁺ T cells,⁵¹ dendritic cells (DCs),⁵² and myeloid-derived suppressor cells (MDSCs),⁵³ impaired their anti-tumor functions and facilitated tumor immune evasion. However, the relationship between the UPR and anti-tumor immunity in tumor cells is unclear. Here, we show that ablation of HRS triggered a strong anti-tumor immune response in different mouse models, and the effects were regulated by the IRE1α/XBP1 pathway, the most conserved branch of the UPR. Interestingly, a lack of IRE1α/XBP1 pathway activation was reported to promote pancreatic cancer cell survival with negative expression of MHC class I in immunocompetent mice, and forced expression of active XBP1 stimulated the outgrowth of macrometastatic lesions, but only in mice with CD8⁺ T cell depletion, suggesting that the IRE1α/XBP1 pathway is associated with tumor immunogenicity.⁵⁴ It was also reported that IRE1α activation in low-protein diet-induced colon cancer cells enhanced CD8⁺ T cell-mediated tumor killing, depending on the RIDD-mediated innate immune response.⁵⁵ Here, we found that XBP1 is also necessary for the activation of the type I IFN pathway upon ER stress. It is possible that XBP1 acts as an enhancer of ifnb1 promoter activity by physically associating with p300 along with IRF3 and CREB-binding protein.^56–59 Bioinformatic analysis in glioblastoma and melanoma using a gene signature reflecting IRE1α activation showed that high IRE1α activities correlated with increased levels of T cell markers.⁵⁵ A positive relationship between higher UPR scores and CR was noticed in an analysis of patients with MSI cancer treated with pembrolizumab.³⁵ In our study, we also noticed the activation of PERK/eIF2α/ATF4 in HRS KO melanoma cells. However, we did not observe a clear effect of PERK inhibition on the IFN pathway. On the other hand, a recent study reported that PERK elimination in melanoma cells caused paraptosis-mediated immunogenic cell death, which induced type I IFN activation in monocyte-derived DCs (MoDCs) and triggered anti-tumor immunity.⁶⁰

Our study suggests that, beyond the neoantigen generation in high-TMB tumors, the maintenance of protein homeostasis through HRS is also importantfor tumor cells to evade immune surveillance and reduce immunotherapy response. In addition, ESCRT proteins have also been reported to regulate lysosomal degradation of MHC class I.⁶¹ Our data also show that HRS may negatively regulate cell surface expression of MHC class I through endosomal trafficking and through the proteostasis-associated type I IFN pathway, as shown in our current work. All these results make HRS a promising therapeutic target to boost ICB response.

Limitations of the study

While we demonstrated the co-localization of HRS with ubiquitinated protein aggregates, the impact of HRS’s presence on the fate of these protein aggregates remains unknown and warrants further investigation. Our study focused on melanoma and colon cancer cells, but exploring the implications in other cancer types would also be of great interest.

STAR★METHODS

RESOURCE AVAILABILITY

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Wei Guo (guowei@sas.upenn.edu).

Materials availability

Plasmids and cell lines generated for this work are available upon request.

Data and code availability

• All data needed to evaluate the conclusions in the paper are present in the paper and/or the Supplementary Materials.

• Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

• This paper does not report original code.

EXPERIMENTAL MODEL AND SUBJECT PARTICIPANT DETAILS

Mammalian cell lines

B16F10, MC38, SW948, RKO, and 293T cells were cultured in DMEM supplemented with 10% FBS and 1% penicillin/streptomycin. Braf^V600E/Pten^−/− (‘‘Braf Pten’’), KLE and HCT116 cells were cultured in 1640/RPMI supplemented with 10% FBS and 1% penicillin/streptomycin. MFE280 and MFE296 cells were cultured in 1640/RPMI supplemented with 10% FBS, 10 μg/mL insulin, and 1% penicillin/streptomycin. YUMM1.7 and YUMMER1.7 cells were cultured in DMEM/F-12 supplemented with 10% FBS, 1% Non-Essential Amino Acids (NEAA), and 1% penicillin/streptomycin. All cells were cultured in a 5% CO₂ incubator at 37°C.

Mice

6 to 8-week-old female mice were used for all experiments. The wild type C57BL/6J mice and nude/J mice were purchased from The Jackson Laboratory. Prior to all experiments, purchased mice were allowed one week to acclimate to housing conditions at The University of Pennsylvania Perelman School of Medicine Animal Facility. All experimental mice were housed in specific pathogen–free conditions and used in accordance with animal care guidelines from the University Laboratory Animal Resources (ULAR), University of Pennsylvania.

METHOD DETAILS

Cell culture

All cell lines were cultured in a 5% CO₂ incubator at 37°C and passaged every 2–3 days. One day before compound treatment, cells were seeded in 6-well plates and then treated with indicated compounds in triplicates for indicated times.

Gene knockdown by shRNA and ectopic expression

The shRNA oligos cloned in a pLKO.1-Puromycin⁺ (Puro) were obtained from the High-Throughput Screening Core at the University of Pennsylvania and Children’s Hospital of Philadelphia, and sequences for their respective target genes are listed in Key resources table. The shRNAs and scrambled shRNA (Addgene #1864) were packaged into lentiviral particles using HEK293T cells co-transfected with viral packaging plasmids. Lentiviral supernatants were harvested 48–72 h after transfection by passing through a 0.45 μm filter. Collected lentivirus was used to infect cells in the presence of 8 μg/mL polybrene (Santa Cruz, cat#134220). Infected cells were selected and expanded with puromycin (Gold Biotechnology, cat#P-600-500) at 2 μg/mL or blasticidin (Sigma-Aldrich, cat#15205) at 10 μg/mL for 5 days before being used for subsequent experiments. Knockdown was verified by western blotting.

KEY RESOURCES TABLE.

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Rabbit monoclonal anti-HRS (clone D7T5N)	Cell Signaling Technology	Cat#15087; RRID:AB_2798700
Rabbit monoclonal anti-MDA5 (clone D74E4)	Cell Signaling Technology	Cat#5321; RRID:AB_10694490
Rabbit monoclonal anti-RIG-I (clone D14G6)	Cell Signaling Technology	Cat#3743; RRID:AB_2269233
Rabbit monoclonal anti-p-TBK1/NAK (Ser172) (clone D52C2)	Cell Signaling Technology	Cat#5483; RRID:AB_10693472
Rabbit monoclonal anti-TBK1/NAK (clone D1B4)	Cell Signaling Technology	Cat#3504; RRID:AB_2255663
Rabbit polyclonal anti-ISG15	Cell Signaling Technology	Cat#2743; RRID:AB_2126201
Rabbit monoclonal anti-GAPDH (clone D16H11)	Cell Signaling Technology	Cat#5174; RRID:AB_10622025
Mouse polyclonal anti-TLR3	Proteintech	Cat#40952
Rabbit monoclonal anti-cGAS (clone D3O8O)	Cell Signaling Technology	Cat#31659; RRID:AB_2799008
Rabbit monoclonal anti-p-STING (Ser366) (clone E9A9K)	Cell Signaling Technology	Cat#50907; RRID:AB_2827656
Rabbit monoclonal anti-STING (clone D2P2F)	Cell Signaling Technology	Cat#13647; RRID:AB_2732796
Rabbit monoclonal anti-BiP (clone C50B12)	Cell Signaling Technology	Cat#3177; RRID:AB_2119845
Rabbit polyclonal anti-p-IRE1α (Ser724)	Novus Biologicals	Cat# NB100-2323
Rabbit monoclonal anti-IRE1α (clone 14C10)	Cell Signaling Technology	Cat#3294; RRID:AB_823545
Rabbit monoclonal anti-XBP-1 (clone EPR22004)	Abcam	Cat# ab220783; RRID:AB_2920809
Mouse monoclonal anti-XBP-1 (clone F-4)	Santa Cruz	Cat# sc-8015; RRID:AB_628449
Rabbit monoclonal anti-XBP-1s (clone E8Y5F)	Cell Signaling Technology	Cat#82914
Rabbit monoclonal anti-p-eIF2α (Ser51) (clone D9G8)	Cell Signaling Technology	Cat#3398; RRID:AB_2096481
Rabbit monoclonal anti-eIF2α (clone D7D3)	Cell Signaling Technology	Cat#5324; RRID:AB_10692650
Rabbit monoclonal anti-ATF-4 (clone D4B8)	Cell Signaling Technology	Cat#11815; RRID:AB_2616025
Rabbit monoclonal anti-ATF-6 (clone D4Z8V)	Cell Signaling Technology	Cat#65880; RRID:AB_2799696
Rabbit monoclonal anti-Ubiquitin (clone E4I2J)	Cell Signaling Technology	Cat#43124; RRID:AB_2799235
Mouse monoclonal anti-Ubiquitin (clone P4D1)	Cell Signaling Technology	Cat# 3936; RRID:AB_331292
Rabbit monoclonal anti-CD63 (clone ERP21151)	Abcam	Cat# ab217345; RRID:AB_2754982
Mouse monoclonal anti-GFP (clone B34)	BioLegend	Cat#902605; RRID:AB_2734671
Rabbit monoclonal anti-HA Tag (clone C29F4)	Cell Signaling Technology	Cat #3724; RRID:AB_1549585
Rabbit polyclonal anti-Tsg101	Bethyl Laboratories	Cat# A303-507A
Rabbit monoclonal anti-CD45 (clone D3F8Q)	Cell Signaling Technology	Cat#70257; RRID:AB_2799780
Rabbit monoclonal anti-CD8α (clone D4W2Z)	Cell Signaling Technology	Cat#98941; RRID:AB_2756376
CD16/32 (clone 93)	BioLegend	Cat#101301; RRID:AB_312800
CD45, APC/Cyanine7 (clone 30-F11)	BioLegend	Cat#103116; RRID:AB_312981
CD3, FITC (clone 17A2)	BioLegend	Cat#100451; RRID:AB_2564591
CD4, Brilliant Violet 605 (clone GK1.5)	BioLegend	Cat#100203; RRID:AB_312660
CD8b, PerCP/Cyanine5.5 (clone YTS156.7.7)	BioLegend	Cat#126610; RRID:AB_2260149
Ki67, Alexa Fluor™ 700 (clone SolA15)	eBioscience	Cat#56-5698-82
GzmB, PE (clone NGZB)	eBioscience	Cat#12-8898-82
Rat IgG isotype (clone 2A3)	BioXCell	Cat# BP0089
InVivoMAb anti-mouse CD8 (clone YTS 169.4)	BioXCell	Cat# BP0117
InVivoMAb anti-mouse PD-1 (clone 29F.1A12)	BioXCell	Cat# BP0273
InVivoMAb anti-mouse IFNAR-1 (clone MAR1-5A3)	BioXCell	Cat# BE0241
Bacterial and virus strains
Stellar™ Competent Cells	Takara	Cat# 636766
Lentivirus	This paper	N/A
Chemicals, peptides, and recombinant proteins
BX-795	Selleckchem	#S1274
STF-083010	Selleckchem	#S7771
4μ8C	Selleckchem	#S7272
Thapsigargin (Tg)	Selleckchem	#S7895
MG132	Cell Signaling Tech	#2194
VER155008	Selleckchem	#S7751
AUY-922	Selleckchem	#S1069
Polybrene	Santa Cruz	cat#134220
Puromycin	Gold Biotechnology	cat#P-600-500
Blasticidin	Sigma-Aldrich	cat#15205
Recombinant Murine IFN-g	PeproTech	Cat# 315-05
Recombinant Murine IFN-b	Novus Biologicals	Cat# 8234-MB-010
Critical commercial assays
PureLink™ RNA Mini Kit	Thermo Fisher Scientific	Cat# 12183018A
PrimeScript RT Reagent Kit with gDNA Eraser	Takarabio	Cat# RR047B
LightCycler 480 SYBR Green I Master	Roche Diagnostics	Cat# 50-720-3233
Live/Dead Fixable Aqua Dead Cell Stain Kit	Life technologies	Cat# L34957
Fixation/Permeabilization Kit	eBioscience	Cat# 00-5523-00
PROTEOSTAT® Aggresome detection kit	Enzo Life Sciences	ENZ-51035-0025
Deposited data
RNA-seq data	This paper	GSE244678
Experimental models: Cell lines
Mouse cell line: B16-F10	ATCC	CRL-6475; RRID:CVCL_0159
Mouse cell line: BrafV600E/Pten−/− (Braf Pten)	Xiaowei Xu, Penn Perelman School of Medicine	N/A
Mouse cell line: MC38	Serge Y. Fuchs, Penn Veterinary School of Medicine	PMID: 32807917; RRID:CVCL_B288
Human cell line: WM9	Meenhard Herlyn, The Wistar Institute	PMID: 30089911; RRID:CVCL_6806
Human cell line: A375	ATCC	CRL-1619; RRID:CVCL_0132
Human cell line: MFE280	Shiaw-Yih Lin, MD Anderson Cancer Center	PMID: 32109374; RRID:CVCL_1405
Human cell line: MFE296	Shiaw-Yih Lin, MD Anderson Cancer Center	PMID: 32109374; RRID:CVCL_1406
Human cell line: KLE	Shiaw-Yih Lin, MD Anderson Cancer Center	PMID: 32109374; RRID:CVCL_1329
Human cell line: HCT116	Shiaw-Yih Lin, MD Anderson Cancer Center	PMID: 32109374; RRID:CVCL_0291
Human cell line: SW948	Shiaw-Yih Lin, MD Anderson Cancer Center	PMID: 32109374; RRID:CVCL_0632
Human cell line: RKO	Shiaw-Yih Lin, MD Anderson Cancer Center	PMID: 32109374; RRID:CVCL_0504
Human cell line: HEK293T	ATCC	CRL-3216; RRID:CVCL_0063
Experimental models: Organisms/strains
C57BL/6J	The Jackson Laboratory	Cat#000664; RRID:IMSR_JAX:000664
NU/J	The Jackson Laboratory	Cat#002019; RRID:IMSR_JAX:002019
Oligonucleotides
shRNA against HRS (human) #1 5’-GCACGTCTTTCCAGAATTCAA-3’	Alissa Weaver, Vanderbilt University School of Medicine	PMID: 25968605
shRNA against HRS (human) #2 5’-GCATGAAGAGTAACCACAGC-3’	Alissa Weaver, Vanderbilt University School of Medicine	PMID: 25968605
shRNA against Hrs (mouse) #1 5’-ATGATGAAGTGGCCAACAAAC-3’	Perelman school of medicine, University of Pennsylvania	TRCN0000313946
shRNA against Hrs (mouse) #2 5’-GGGCTACCAGCCGTACAATAT-3’	Perelman school of medicine, University of Pennsylvania	TRCN0000314016
shRNA against B2m (mouse) 5’-GCCGAACATACTGAACTGCTA-3’	Perelman school of medicine, University of Pennsylvania	TRCN0000288438
shRNA against Ddx58 (mouse) 5’-GCCATGCAACATATCTGTAAA-3’	Perelman school of medicine, University of Pennsylvania	TRCN0000103888
shRNA against Ifih1 (mouse) 5’-CCTACAAATCAACGACACGAT-3’	Perelman school of medicine, University of Pennsylvania	TRCN0000103648
sgRNA against Hrs (mouse) #1 5’-TCCTGCTCCACAGAGGCAAG-3’	This paper	N/A
sgRNA against Hrs (mouse) #2 5’-ATCTGCGACCTGATCCGTC-3’	This paper	N/A
sgRNA against Ern2 (mouse) 5’-CAGGGTCGAGACAAACAAC-3’	This paper	N/A
sgRNA against Xbp1 (mouse) 5’-GGAGAAAACTCACGGCCTTG-3’	This paper	N/A
sgRNA against Msh2 (mouse) 5’-CGTGATCAAGTACATGGGGC-3’	This paper	N/A
Primer: Ifnb1 F 5’-GGTGGAATGAGACTATTGTTG-3’	This paper	N/A
Primer: Ifnb1 R 5’-AGGACATCTCCCACGTC-3’	This paper	N/A
Primer: Isg15 F 5’-GGTGTCCGTGACTAACTCCAT-3’	This paper	N/A
Primer: Isg15 R 5’-TGGAAAGGGTAAGACCGTCCT-3’	This paper	N/A
Primer: Cxcl10 F 5’-ATCATCCCTGCGAGCCTATCCT-3’	This paper	N/A
Primer: Cxcl10 R 5’-GACCTTTTTTGGCTAAACGCTTTC-3’	This paper	N/A
Primer: Ifih1 (Mda5) F 5’-CTACGCACTTTCCCAGTGGAT-3’	This paper	N/A
Primer: Ifih1 (Mda5) R 5’-TGTTCAGTCTGAGTCATGGGC-3’	This paper	N/A
Primer: Ddx58 (Rig-I) F 5’-AAGAGCCAGAGTGTCAGAATCT-3’	This paper	N/A
Primer: Ddx58 (Rig-I) R 5’-AGCTCCAGTTGGTAATTTCTTGG-3’	This paper	N/A
Primer: Oasl F 5’-CAGGAGCTGTACGGCTTCC-3’	This paper	N/A
Primer: Osal R 5’-CCTACCTTGAGTACCTTGAGCAC-3’	This paper	N/A
Primer: Il28 b F 5’-AGCTGCAGGTCCAAGAGCG-3’	This paper	N/A
Primer: Il28 b R 5’-GGTGGTCAGGGCTGAGTCATT-3’	This paper	N/A
Primer: IFNB1 F 5’-GCCATCAGTCACTTAAACAGC-3’	This paper	N/A
Primer: IFNB1 R 5’-GAAACTGAAGATCTCCTAGCCT-3’	This paper	N/A
Primer: ISG15 F 5’-CCTTCAGCTCTGACACC-3’	This paper	N/A
Primer: ISG15 R 5’-CGAACTCATCTTTGCCAGTACA-3’	This paper	N/A
Primer: CXCL10 F 5’-GGTGAGAAGAGATGTCTGAATCC-3’	This paper	N/A
Primer: CXCL10 R 5’-GTCCATCCTTGGAAGCACTGCA-3’	This paper	N/A
Primer: Ecro1L F 5’-GAATGTGAGCAAGCTGAGCG-3’	This paper	N/A
Primer: Ecro1L R 5’-CATACTCAGCATCGGGGGAC-3’	This paper	N/A
Primer: Calr F 5’-CCTGAATACTCCCCCGATGC-3’	This paper	N/A
Primer: Calr R 5’-CCCACGTCTCATTGCCAAAC-3’	This paper	N/A
Primer: Pdia4 F 5’-CCTGATTGGACACCTCCACC-3’	This paper	N/A
Primer: Pdia4 R 5’-GGGGCAAGTTTCTTGCAGTG-3’	This paper	N/A
Primer: Pdia5 F 5’-TGGGGGATAACTTCCGGGAT-3’	This paper	N/A
Primer: Pdia5 R 5’-CAGTGAAGTGGGGGATGACC-3’	This paper	N/A
Primer: Sec61a1 F 5’-TCTGCAAAAAGGGTACGGCT-3’	This paper	N/A
Primer: Sec61a1 R 5’-GCGGTAGAATGCCTCTCGAA-3’	This paper	N/A
Primer: Sec24d F 5’-GCGTGCAGAGCAGGGTTATT-3’	This paper	N/A
Primer: Sec24d R 5’-GAAGGCCCCAATGGCTTCAT-3’	This paper	N/A
Primer: Dnajc3 (p58/Ipk) F 5’-CACAGTTTCACGCTGCAGTT-3’	This paper	N/A
Primer: Dnajc3 (p58/Ipk) R 5’-CTCTGTAGTCTTGCGGCAGT-3’	This paper	N/A
Primer: Actin (Actb) F 5’-CATCGTACTCCTGCTTGCTG-3’	This paper	N/A
Primer: Actin (Actb) R 5’-AGCGCAAGTACTCTGTGTGG-3’	This paper	N/A
Primer: GAPDH F 5’-CAACGGATTTGGTCGTATTG-3’	This paper	N/A
Primer: GAPDH R 5’-GCAACAATATCCACTTTACCAGAGTTAA-3’	This paper	N/A
Primer: Xbp1 F (Xbp1 splice) 5’-AGTTAAGAACACGCTTGGGAAT-3’	This paper	N/A
Primer: Xbp1 R (Xbp1 splice) 5’-AAGATGTTCTGGGGAGGTGAC-3’	This paper	N/A
Recombinant DNA
Plasmid: HA-HRS	This paper	N/A
Plasmid: LentiCas9-Blast	Addgene	#52962; RRID:Addgene_52962
Plasmid: LentiGuide-Puro	Addgene	#52963; RRID:Addgene_52963
Plasmid: LentiGuide-Hygro	Addgene	#160090; RRID:Addgene_160090
Plasmid: pCI-neo Fluc EGFP (Fluc-EGFP)	Addgene	#90170; RRID:Addgene_90170
Plasmid: pCI-neo FlucDM EGFP (FlucDM-EGFP)	Addgene	#90172; RRID:Addgene_90172
Software and algorithms
CRISPR Design	Broad Institute	http://crispr.mit.edu
shRNA Design	Broad Institute	https://portals.broadinstitute.org/gpp/public/help
GraphPad Prism 6	GraphPad Software	https://www.graphpad.com
ImageJ	NIH	https://imagej.nih.gov/ij/
PennSCAP-T Pipeline	Junhyong Kim, Penn School of Arts & Sciences	https://kim.bio.upenn.edu/software/ngs_pipeline/home.shtml

For double knockdown, B16 cells were first transduced with lentiviral pLKO-scramble-Puro or pLKO-shB2M-Puro and selected with puromycin for 5 days. Then cells were transfected with lentiviral pLKO-scramble-Bsd or pLKO-shHrs-Bsd and selected with both puromycin and blasticidin for 5 days. The double knockdown was verified by western blotting and flow cytometry.

To establish stable cell lines expressing Fluc-EGFP or FlucDM-EGFP, human melanoma cell line A375 was transfected by electroporation. A375 cells in late-log phase were trypsinized and resuspended at 4×10⁶ cells/ml in ice-cold BTXpress electroporation buffer (BTX, cat#45-0803). 400 mL A375 cell suspension were transferred into electroporation cuvettes mixture with 12 μg desired plasmid and incubated for 5 min on ice. The cuvette was then placed in BTX ECM-600 Electro Cell Manipulator with a voltage of 220 V, R4 resistance and capacitance of 800 mF. After electroporation, cells were returned to full media without antibiotics and then selected by 400 μg/mL G418 after two days for at least one week.

Gene deletion using CRISPR-Cas9

The sequences of gRNA oligos for their respective target are genes listed in Key resources table. The oligos were annealed and cloned into the LentiGuide-Hygro (for HRS) or LentiGuide-Puro (for the other genes) vector. B16 and Braf Pten cells were transfected with the lentivirus containing LentiCas9-Blast and selected with 5 μg/mL Blasticidine for 5 days. Then the cells were transfected with the lentivirus containing LentiGudie-Hygro/Puro carrying respective gRNA and selected with 200 μg/mL hygromycin B or 2 μg/mL puromycin for 5 days. Cells were then resuspended at super-low density and seeded into 96-well plates to allow colony formation from single cells. Colonies were then picked and expanded for validation by immunoblot. For double knockout, HRS KO B16 cells and Braf Pten cells were used to delete the second target genes as described. For the rescue experiment, the homologous human HRS cDNA with an HA-tag was transfected into HRS KO B16 cells.

RNA extraction and RT-PCR

All reagents, buffers, and containers used for RNA extraction and reverse transcription were RNase-free grade to eliminate RNase contaminants. For total RNA extraction, cells in culture were harvested and washed by PBS before lysis and extraction using the PureLink RNA Mini Kit (Thermo Fisher Scientific, cat#12183018A). RNA extraction was performed according to manufacturer’s instructions. 1 μg extracted RNA was reversely transcribed into cDNA using the PrimeScript RT Reagent Kit with gDNA Eraser (Takarabio, cat#RR047B). The obtained cDNA samples were diluted with RNase-free water and used for real-time quantitative PCR (RT-qPCR). SYBR green (Roche Diagnostics, Cat#50-720-3233) and gene specific primers with sequences listed in Key resources table were used for PCR amplification and detection on an Applied Biosystems 7500 Fast Real-Time PCR System. The RT-qPCR data were normalized to β-actin (for mouse cell lines) or GAPDH (for human cell lines) and presented as fold changes of gene expression in the test samples compared to the control.

Soluble and insoluble protein extraction and immunoblot analysis

Cultured cells were harvested and washed with ice-cold PBS to completely remove the media. The cells were lysed on ice for 10 min using Cell Lysis Buffer (Cell Signaling Technology, cat#9803) supplemented with protease inhibitors (Roche, cat#04693132001) and phosphatase inhibitor cocktail (Bimake, cat#B15002). Cell lysates were then cleared by centrifugation at 14,000xg at 4°C. The pellets were washed three times with ice-cold PBS supplemented with 0.1% Triton X-100 and protease inhibitors and then dissolved in urea lysis buffer (8 M urea, 150 mM 2-Mercaptoethanol and 25 mM Tris, pH 7.5). The proteins in soluble supernatant were quantified using Bio-Rad protein assay (Bio-rad, cat#5000006). Then 6xSDS loading buffer was added to the samples followed by boiling at 95°C for 5 min 30 μg protein for each sample (equal volume of insoluble proteins compared to their soluble supernatants) was separated using 12% sodium dodecylsulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and transferred to a nitrocellulose membrane (Cell Signaling Technology, cat#12369). After being blocked with 5% skim milk at room temperature for 1 h, the membranes were incubated with appropriate primary antibodies at 4°C overnight. The membranes were then washed with TBS-T three times and probed with horseradish peroxidase-conjugated secondary antibodies at room temperature for 1 h. The membranes were washed again with TBS-T three times and developed with ECL Western Blotting Substrate (Pierce). Information about the primary antibodies is included in Key resources table.

Fluorescence microscopy

Cells were cultured on glass coverslips in 6-well plates with indicated treatments for 24 h and washed twice with PBS to remove residue medium. The cells were fixed in 4% paraformaldehyde for 15 min and permeabilized within 0.1% Triton X-100 in PBS for 20 min. After being blocked with 10% normal goat serum for 20 min, the coverslips were incubated with appropriate primary antibodies at 4°C overnight. On the next day, the samples were washed with cold PBS three times and incubated with fluorescence labeled secondary antibodies (Life Technologies) for 40 min. After washed with ice-cold PBS three times, the samples were mounted with ProLong Gold Antifade Reagent with DAPI (Cell Signaling Technology, cat#8961). Samples were observed using an Eclipse TE2000-U inverted microscope (Nikon) equipped with a PLAN APO × 100 1.3 NA objective and Cascade 512B CCD camera (Photometrics) driven by Metamorph imaging software (Molecular Devices). The images were analyzed using NIS-Elements Advanced Research software (Nikon; version 4.50). See Key resources table for the information of antibodies.

For quantification of ubiquitin positive puncta in the control and HRS KO B16 cells, 10 images were taken under microscope and the numbers of cells with puncta inside and total cells were counted. Meanwhile, the number of puncta in positive cells were also counted. The percentage of cells with ubiquitin positive puncta and the number of ubiquitin positive puncta per cell were calculated, respectively. The ubiquitin puncta index was calculated according to the formula: index = [the percentage of cells with ubiquitin positive puncta] × [the number of ubiquitin positive puncta per cell].

XBP1 splicing assay

Cells in culture were treated with 4μ8C for 24 h and harvested for RNA extraction. RNA extraction and reverse transcription were performed as described above. cDNA was amplified using the primers (listed in Key resources table) designed to bracket the mouse Xbp1 spliced sequence to amplify both spliced and un-spliced forms. The PCR products were electrophoresed on 12% polyacrylamide/TBE (Tris-Borate-EDTA) gel and stained with ethidium bromide.

Aggresome detection

The detection of aggresome in cells were performed according to the manufacturer’s instructions. Cells were cultured on coverslips in 6-well plates with indicated treatments for 24 h and washed twice with PBS buffer to remove residue medium, then fixed in 4% paraformaldehyde for 15 min and permeabilized within Permeabilizing Solution for 30 min. The samples were blocked with 10% normal goat serum for 20 min and incubated with anti-ubiquitin antibodies at 4°C overnight. On the next day, the samples were washed three times with PBS and incubated with Alexa Fluor 647 conjugated secondary antibodies (Life Technologies) for 40 min. After washed twice with PBS, the samples were incubated with Dual Detection Reagent for 30 min at room temperature. The samples were mounted with ProLong Gold Antifade Reagent with DAPI after being washed with PBS. Images were captured and analyzed as described above.

Purification of extracellular vesicles (EVs)

For the collection of EVs, cells were cultured in media supplemented with 10% exosome-depleted FBS, in which EVs were removed by overnight centrifugation at 100,000xg. Supernatants were then collected 48–72 h later for EV purification. Briefly, culture supernatants were centrifuged at 2,000xg for 20 min at 4°C to remove cell debris and dead cells (Beckman Coulter, Allegra X-14R). Supernatants were obtained and microvesicles were pelleted after centrifugation at 16,500 g for 45 min at 4°C (Beckman Coulter, J2-HS) with PBS. Supernatants were further centrifuged at 100,000xg for 2 h at 4°C (Beckman Coulter, Optima XPN-100). After the pelleted exosomes were suspended in PBS, the exosomes were further purified by ultracentrifugation at 100,000xg for 2 h. The harvested exosomes were diluted in filtered PBS, and the size and concentration of the exosomes were assessed using a NanoSight NS300 (Malvern Instruments).

Animal studies

Mice were anesthetized with Avertin (2.5%), shaved at the back. 200,000 or 250,000 B16F10 (control, HRS KO, scramble, HRS KD, HRS IRE1 dKO, HRS XBP1 dKO), or 250,000 Braf Pten (control, HRS KO), or 1,000,000 MC38 (scramble, HRS KD) tumor cells were injected in the back subcutaneously. Tumors were measured every 2–3 days once palpable (long diameter and short diameter) with a digital caliper. Tumor volumes were calculated by the formula: length x (width)²/2. The mice were euthanized before the longest dimension of the tumors reached 2.0 cm or upon ulceration/bleeding.

For CD4⁺ or CD8⁺ T cell depletion, mice were given 100 μg isotype or anti-CD8 (clone YTS 169.4, BioXCell, cat# BP0117) antibodies by intraperitoneal injection on days 0, 4, 8, 12 and 16 post tumor injection. For verification of CD4⁺ or CD8⁺ T cell depletion, the blood samples were collected from the tail veins of the mice. To block type I IFN pathway in vivo, mice were given 100 μg anti-IFNAR1 antibodies (clone MAR1-5A3) by intraperitoneal injection on days 0, 4, 8, 12 and 16 post tumor injection.

For anti-PD-1 antibody treatments, each mouse was given 100 μg antibodies by intraperitoneal injection on Day 8, 11 and 14 post tumor injection using the following antibodies: anti-PD-1 (clone 29F.1A12, BioXCell, cat# BP0273) or isotype (clone 2A3, BioXCell, cat# BP0089). The mice were randomized such that treatment groups had similar average tumor volumes before treatment initiation.

Tumor-infiltrating lymphocyte flow cytometry

Tumors were excised on Day 14 post tumor injection and cut into 2 mm sized pieces in collagenase, hyaluronidase and DNase. Samples were incubated at 37°C for 40 min and passed through a 70 μm filter. Red blood cells were lysed by incubating samples in a homemade ammonium chloride hemolysis buffer on ice for 10 min. The cells were washed with PBS and counted in a Countess II Automated Cell Counter (Invitrogen) and transferred to tubes at 10⁶ cells/tube. The samples were blocked with anti-mouse CD16/32 antibody in FACS buffer (PBS containing 0.2% BSA) for 20 min. Live/Dead Fixable Aqua Dead Cell Stain Kit (Life technologies, cat# L34957) together with surface antibody cocktails were added into each sample with incubation on ice for 20 min. Then the samples were fixed and permeabilized using Fixation/Permeabilization Kit (eBioscience, cat# 00-5523-00) and stained with an intercellular antibody cocktail. The samples were analyzed on the BD LSR II system. Information about the flow cytometery antibodies is included in Key resources table.

Immunohistochemistry

Immunohistochemical (IHC) staining was performed at The Wistar Institute Histotechnology Facility using a Leica Bond automated staining platform. The B16 tumors were harvested 14 days post-injection and fixed in 4% paraformaldehyde immediately. The samples were paraffin-embedded and sectioned into slides. Antigen retrieval was performed by steaming the slides in citrate buffer (pH 6.0) for 5 min. After washing with PBS, the slides were blocked by 10% normal goat serum for 20 min and incubated with anti-CD45 and anti-CD8α antibodies overnight at 4°C. After washing with PBS three times, the slides were incubated with a biotinylated secondary antibody (Jackson ImmunoResearch, cat#) for 30 min. Detection was performed using Nova Red (Fisher Scientific, cat#). Slides were visualized using QuPath-0.2.3 software. CD45⁺ and CD8⁺ cells that stained with strong membranous positivity were enumerated in five separate areas at 10x magnification in a blinded fashion for each slide.

RNA sequencing

Total RNA was extracted from cultured control and HRS KD B16 cells, or the control and HRS KD B16 tumors harvested from tumor-bearing immunocompetent mice. The concentration of RNA was determined with Qubit RNA Assay Kit in Qubit 2.0 Fluorometer (Life Technologies). RNA integrity was assessed with RNA Nano 6000 Assay Kit in the Bioanalyzer 2100 system (Agilent Technologies). 3 μg of RNA per sample were used to construct RNA-seq libraries. Sequencing libraries were generated using the NEBNext Ultra RNA Library Prep Kit for Illumina (New England Biolabs, cat#E7530L) and sequenced on an Illumina Hiseq platform, and 125/150 bp paired-end reads were generated (Novogene Bioinformatics Technology). Three biological repeats for B16 cells and five biological repeats for B16 tumors were performed for the RNA-Seq experiment.

QUANTIFICATION AND STATISTICAL ANALYSIS

RNA-seq analysis

Raw reads were processed by PennSCAP-T Next Generation Sequencing Pipeline (https://github.com/safisher/ngs). Basically, Illumina adapters were searched and trimmed from raw reads. The processed reads were aligned to reference genome mm10 using STAR (v 2.4.0h2)⁶² with the parameters ‘‘–outFilterScoreMin 0 –outFilterScoreMinOverLread 0 –outFilterMatchNmin 30 –outFilter-MatchNminOverLread 0.4 –outFilterMismatchNmax 100 –outFilterMismatchNoverLmax 0.3’’ and multiple alignments were filtered. VERSE (v0.1.4) (Zhu et al. 2016) was used for count qualification with the parameters ‘-t “exon; mito; intron; intergenic” -z 3 –non-emptyModified’. Exonic counts were normalized by DESeq (Anders and Huber 2010). Genes with average counts <1 were further filtered out. The remaining count data were modeled by DESeq2 (Love, Huber, and Anders 2014, 2), and the differential genes were called using the thresholds BH-adjusted p value ≤0.1 and foldchange ≥2 or ≤0.5. R package clusterProfiler (v3.10.1)⁶³ was used to perform gene enrichment analysis on differential genes using annotation databases Gene Ontology.⁶⁴ In addition, Gene Set Enrichment Analysis (GSEA) was performed with gsea.jar (v2.0.10).⁶⁵

TCGA data analysis

TCGA RNA-seq count tables (v16.0) were downloaded from Genomic Data Commons (GDC) using GDC Data Transfer Tool. Samples labeled ‘‘Primary Tumor’’ and ‘‘Solid Tissue Normal’’ were selected and those noted with technical errors or concerns were excluded. In total, 7893 samples remained, encompassing 19 tumor types. For each tumor type, data normalization and tumor vs. normal differential analysis were done by DESeq2.⁶⁶ The correlation of CD8A with ESCRT-related genes was calculated as Pearson Correlation Coefficient (PCC). To test whether an ESCRT gene is up-regulated in a tumor type, we used the cutoff log2FC > 0 and Benjamini-Hochberg⁶⁷ corrected p value <0.05. To test whether an ESCRT gene is negatively correlated with CD8A expression, we used the cutoff PCC <0 and p value <0.05.

Public dataset analysis

The clinical characteristics data from the Riaz et al. study⁴² were downloaded from the Supplemental Information mmc2.xlsx. The RNA-seq metadata were downloaded from ==============https://cvws.icloud-content.com/B/Ae889JGiOnyPE8rDD4e-RftcHTh4AYHkWvJm_INChgzNI83lL2pOAuf_/SampleInfo.txt?o=AlFUUlx-ZWRA_R5PAzsjri3tkEtMcGdiJeAPKan2pMRc&v=1&x=3&a=CAogLXDDozxqmTeTMvIhNAAzNNwx3ORDuGw76WYOWTLWYDgSehCB75fiwC8Ygf-StsovIgEAKgkC6AMA_z4wEZpSBFwdOHhaBE4C5_9qJ_eWuyb4PiVTWxsFo4nSSlsQf3U8kjRb5qRbSqekpJURQ45CPKr1wXInulFrM3RJPZ8-dRp2C6UYl5aOfuh9O40Xh8cQyweHq-TmWaCoIRsa&e=1634885484&fl=&r=238F5DDF-BF6B-4DCC-AA0A-C3CBAE141875-1&k=H05xfmVCN_tx015DoVmP2w&ckc=com.apple.largeattachment&ckz=6CBFF55A-A0FB-46C8-A6A0-8B5ACFFF764F&p=58&s=b7tt7qTMFxRE9PI5RCztsIdGS7g&teh=1&%20=ea4a0f2e-b139-458b-99fa-e902dd6ff1c9. The RNA-seq TPM matrix was kindly shared by the authors.⁴² We then selected pre-treatment data points for analysis in the present study. For patients with multiple RNA-seq samples, we averaged HRS expression values throughout 7A-C. For survival analysis (Figure 7B), we used the median of HRS expression to define High/Low HRS levels. For Figure 7C. Statistical significance was obtained by unpaired Welch’s t test.

Statistical analyses

Statistical analyses were performed using GraphPad Prism v.8.0 software and a two-tailed value of p < 0.05 was considered statistically significant. Normality of distribution was determined by D’Agostino-Pearson omnibus normality test and the equal variance assumption between groups was assessed by Brown-Forsythe test. For equal variance data, the significance of mean differences was determined using unpaired Student’s t-test (two groups) or one-way ANOVA with appropriate post-hoc tests (more than two groups); for groups that differed in variance, unpaired t test with Welch’s correction (two groups) or Welch’s ANOVA with appropriate post-hoc tests (more than two groups) was performed. two-way ANOVA was used to compare mouse volume data among different groups. For comparing survival curves, a Log rank (Mantel-Cox) test was performed.

For RNA-seq data, statistical analyses were performed with R (version 3.4.0) unless otherwise specified. Student’s t-test was used to determine whether a significant shift in mean occurred for all comparisons unless otherwise specified.

Highlights.

• HRS ablation leads to misfolded protein accumulation and ER stress

• HRS ablation activates type I interferon pathway through IRE1α/XBP1

• HRS loss in tumor cells stimulates anti-tumor immunity

• HRS depletion sensitizes tumor with high mutational burden to anti-PD-1 therapy