HLA Class-I APM Upregulation by Anti-Cancer Therapies in the Era of Checkpoint Inhibitors: A Review

Abstract

Importance:

Although typically impressive, objective responses to immune checkpoint inhibitors (ICIs) occur in only 12.5% of advanced cancer patients. The majority do not respond due to cell-intrinsic resistance mechanisms, including, but not restricted to HLA class-I antigen processing machinery (APM) defects. The latter, which have a negative impact on neoantigen presentation to cytotoxic T-lymphocytes (CTLs), are present in the majority of malignant tumors. They are caused by structural alterations in <25% of cases and by dysregulated signaling and/or epigenetic changes in the remaining cases, making them frequently correctable. In this review, we summarize the growing clinical evidence that chemotherapy, targeted therapies, and less so radiotherapy, can correct HLA class-I APM defects in cancer cells and improve responses to ICIs.

Observations:

Most chemotherapeutics enhance HLA class-I APM component expression/function in cancer cells, tumor CTL infiltration, and responses to ICIs in preclinical and clinical models. Despite preclinical evidence, radiotherapy does not appear to upregulate HLA class-I expression in patients and does not enhance the efficacy of ICIs in clinical settings. The latter findings underscore the need to optimize the dose/schedule of radiation and timing of ICI administration to maximize their immunogenic synergy. By increasing DNA/chromatin accessibility, epigenetic agents (histone deacetylase-, DNA methyltransferase-, and EZH2-inhibitors) upregulate HLA class-I APM component expression/function in many cancer types, a crucial contributor to their synergy with ICIs in patients. Furthermore, EGFR- and BRAF/MEK-inhibitors are effective at enhancing HLA class-I expression in EGFR- and BRAF-mutant tumors, respectively; these changes may contribute to the impressive clinical responses induced by these inhibitors in combination with ICIs.

Conclusions and Relevance:

Chemotherapy and targeted therapies are effective at enhancing HLA class-I APM component expression/function in cancer cells. Their resulting increased immunogenicity and recognition/elimination by cognate CTLs contributes to the anti-tumor activity of these therapies, as well as their synergy with ICIs.

Introduction

The impressive clinical responses to immune checkpoint inhibitor (ICI)-based therapy have convincingly demonstrated that a patient’s immune system can recognize and eliminate malignant cells.^1–3 However, only 12.5% of patients, on average, across several cancer types display objective responses to ICI-based therapy.⁴ A number of genomics studies, almost exclusively focusing on structural alterations within cancer cells, have identified subsets of patients who are resistant to ICI-based therapy. These patients include those with a low tumor mutation burden (reducing the likelihood cognate cytotoxic T lymphocytes [CTLs] recognize a suitable neoepitope target expressed by cancer cells), interferon-γ pathway abnormalities (resulting in impaired killing of cancer cells by cognate CTLs), or defects in human leukocyte antigen (HLA) class-I heavy chain or beta 2-microglobulin (β₂m), the two HLA class-I subunits mediating tumor antigen presentation (resulting in escape of cancer cells from elimination by cognate CTLs).⁵ Because of their structural nature, the latter defects are only correctable via gene therapy, which cannot be routinely applied in a clinical setting. However, structural alterations in HLA class-I heavy chain, β₂m, and other antigen processing machinery (APM) components represent at most 25% of the identified HLA class I APM defects.⁶ The majority of these defects which are frequently present in most, if not all cancer types (Figure 1, Table S1), are caused by dysregulated signaling within cancer cells and/or epigenetic mechanisms. Therefore, they are potentially correctable by clinically relevant strategies which counteract the underlying mechanism(s). HLA class-I APM component defects affect a key step in the sequence of events leading to the recognition and elimination of cancer cells by cognate CTLs and therefore are likely to play a major role in their resistance to T cell-mediated elimination.⁶ Growing evidence, summarized in this paper, suggests that correction of these defects can greatly enhance the efficacy of T cell-based immunotherapies, including ICI-based therapy.

Fig. 1 Legend. Frequency of HLA class-I APM Component Defects by Cancer Type.

HLA class-I APM component defect frequency by cancer type was assessed via systematic review of the literature.⁶ A total of 237 papers were identified that analyzed immunohistochemical HLA class-I APM component expression in cancer tissues. Only cancer types in which HLA class-I APM component expression was assessed in at least 100 cases have been included in this heatmap. Combined defect frequency amongst all cancers is shown in the bottom row. Exact proportions of cancers with HLA class-I APM component defects are available in Supplementary Table 1.

Role of HLA Class-I APM Components in the Response to Immune Checkpoint Inhibitor-Based Therapy

Presentation of tumor antigen (TA)-derived peptides by class-I major histocompatibility complex (MHC) antigens is crucial for the recognition of malignant cells by cognate CTLs, which are the major players in immune editing.⁷ This process involves degradation of mostly, but not exclusively endogenous proteins by the proteasome or immunoproteasome, transport of the resulting peptides into the endoplasmic reticulum lumen through the heterodimeric TAP1/TAP2 complex and loading of the peptides onto MHC class-I heavy chain-β₂m dimers. The last step involves chaperones calreticulin, tapasin, and ER resident protein 57 [ERp57]. The resulting MHC class-I heavy chain-β₂m-peptide trimolecular complexes (MHC class-I in every species, HLA class-I in humans) travel to the cell membrane and are recognized by the T cell receptor (TCR) expressed by cognate CTLs (Figure 2).⁶

Fig. 2 Legend. Function of HLA Class-I Antigen Processing Machinery Components in Cancer Cells.

Proteins are degraded by the proteasome into peptides with the correct length and sequence to fit the HLA class-I heavy chain/β₂m dimer groove. The catalytic β1 (Delta), β2 (Zeta) and β5 (MB1) proteasome subunits are involved in this process and are exchangeable with IFNγ-inducible LMP2, LMP10, LMP7 immunoproteasome subunits, respectively. The latter constitute the more efficient immunoproteasome. During degradation of proteins, suboptimal peptides are also generated. Both optimal and suboptimal peptides are shuttled into the endoplasmic reticulum lumen via the heterodimeric transporter associated with antigen processing complex consisting of TAP1 and TAP2. ER aminopeptidases ERAP1 and ERAP2 refine the length of the suboptimal peptides. Furthermore, newly synthesized HLA class-I heavy chain is stabilized by calnexin in the ER, associates with β₂m, and then a processed peptide is loaded onto the dimer with the help of the chaperone molecules calreticulin, ER resident protein 57 (ERp57), and tapasin. The resulting trimolecular complexes are transported to the plasma membrane via the Golgi apparatus and are recognized by the T cell receptor of cognate cytotoxic T cells. Neolacto-series glycosphingolipids synthesized by glycosyltransferase B3GNT5 sterically impede the interactions of HLA class-I trimolecular complexes with the T cell receptor, impairing the recognition of cancer cells by cognate cytotoxic T lymphocytes. B3GNT5 activity is inhibited by protease SPPL3. Defects in HLA class-I antigen processing machinery component expression may compromise expression of HLA class-I antigen/tumor antigen-derived peptide complexes, negatively impacting recognition/elimination of cancer cells by cognate cytotoxic T lymphocytes, and potentially providing cancer cells with an escape mechanism from immune surveillance.

Collectively referred to as HLA class-I APM components, HLA class-I heavy chain, β₂m, and proteins involved in the assembly and cell surface expression of HLA class-I heavy chain-β₂m-peptide complexes play a crucial role in generating an optimal response to the T cell-based immunity triggered by therapies such as vaccines and ICIs. Indeed, B2m knockouts cause resistance to anti-PD-1 monoclonal antibody (mAb) therapy in mouse models of lung cancer. Moreover, in melanoma and lung cancer patients, mutations in β₂m are associated with acquired resistance to ICIs.^8,9 Lastly, non-structural alterations such as transcriptional repression of HLA-A or HLA-B are associated with acquired resistance to ICIs in Merkel cell carcinoma patients.¹⁰

In contrast, higher HLA class-I APM component expression levels may enhance TA-derived peptide presentation, improving immunogenicity as well as recognition and elimination of cancer cells by cognate CTLs. Consistent with this notion, high HLA class-I APM component expression in cancer cells is associated with improved prognosis, including longer disease-specific, progression-free, and overall survival in cancer patients (Table 1). Melanoma patients with >30% of malignant cells in their tumors stained by HLA class-I-specific mAbs, displayed a significantly lower frequency of progressive disease on week 13 of treatment with ipilimumab, compared to patients with <30% of malignant cells stained by HLA class-I-specific mAbs.¹¹ Likewise, melanoma patients displaying high HLA class-I APM component mRNA expression in their tumors demonstrated prolonged overall- and progression free-survival in response to CTLA-4 blockade, compared to patients with low HLA class-I APM component mRNA expression.¹²

Table 1.

Association of Higher HLA Class-I APM Component Expression Level in Cancer Cells with Better Survival Outcomes.

Cancer Type	Component	Survival Outcome	Reference1
Bladder	Calreticulin	OS	Cathro et al. (2010)
Breast	HLA class-I	DSS	Sik Park et al. (2019)
Cervical	HLA-A, HLA class-I, LMP7, TAP1, ERAP1	OS	Mehta et al. (2008)
Cervical	HLA-A, TAP1, ERAP1	DFS	Mehta et al. (2008)
Clear Cell RCC	HLA class-I	OS, RFS	Sekar et al. (2016)
Colon	HLA-A, HLA-C	DFS	Benevolo et al. (2007)
Colon	HLA-C	OS	Benevolo et al. (2007)
Colon	Calreticulin	OS	Peng et al. (2010)
Colorectal	HLA class-I	DFS	Iwayama et al. (2015)
Colorectal	HLA-A	DFS	Sandel et al. (2005)
Colorectal	HLA class-I	DSS	Simpson et al. (2010)
Colorectal	Tapasin	OS	Sokol et al. (2015)
Colorectal	HLA-B/C+B2M co-expression	DSS	Watson et al. (2006)
Endometrial	HLA-B/C	PFS, OS	Yakabe et al. (2015)
ESFT	HLA class-I	OS	Yabe et al. (2011)
Esophageal	HLA class-I	RFS	Hosch et al. (1997)
Esophageal	HLA class-I, B2M, TAP1	OS	Tanaka et al. (2012)
Esophageal SCC	HLA class-I	OS	Mizukami et al. (2008)
Esophageal SCC	HLA class-I	OS	Zhang et al. (2013)
Gastric	HLA class-I	OS	Ishigami et al. (2008)
Gastric	ERp57	OS	Leys et al. (2007)
HCC	HLA-C	OS	Wang et al. (2019)
HNSCC	HLA-B/C	DFS	Bandoh et al. (2010)
HR-proficient ovarian	HLA class-I	PFS	Matsushita et al. (2020)
Laryngeal SCC	HLA-B/C, B2M, LMP2	DFS, CSS	Ogino et al. (2006)
Maxillary Sinus SCC	Tapasin, HLA-B/C	OS	Ogino et al. (2003)
Melanoma	TAP1, TAP2, HLA-B/C	PFS, OS	Kageshita et al. (1999)
NSCLC	HLA class-I	OS	Ichinokawa et al. (2019)
NSCLC	HLA class-I	OS	Kikuchi et al. (2007)
Osteosarcoma	HLA class-I	OS, EFS	Tsukahara et al. (2006)
PDAC	HLA-B/C	OS	Michelakos et al. (2021)
Penile	HLA-A	OS	Djajadiningrat et al. (2015)
Prostate	HLA class-I	OS	Levin et al. (1994)
Rectal	HLA class-I	DFS	Reimers et al. (2014)
Rectal	HLA class-I	OS, DFS	Speetjens et al. (2008)

¹Citations provided in Supplementary Information.

Abbreviations: B2M, beta 2-microglobulin; CSS, cause-specific survival; DFS, disease-free survival; DSS, disease-specific survival; EFS, event-free survival; ERAP, endoplasmic reticulum aminopeptidase; ERp57, endoplasmic reticulum resident protein 57; ESFT, Ewing sarcoma family of tumors; HCC, hepatocellular carcinoma; HLA, human leukocyte antigen; HNSCC, head and neck squamous cell carcinoma; HR, homologous recombination; NSCLC, non-small-cell lung cancer; OS, overall survival; PDAC, pancreatic ductal adenocarcinoma; PFS, progression-free survival; RCC, renal cell carcinoma; RFS, relapse-free survival; SCC, squamous cell carcinoma; TAP, transporter associated with antigen processing.

Thus, HLA class-I APM component upregulation in cancer cells may be a promising strategy to enhance the efficacy of ICI-based therapy. Herein, we present the growing clinical evidence that chemotherapy, certain targeted therapies, and less so radiotherapy can restore/enhance HLA class-I APM component expression in cancer cells. These changes lead to increased CD8+ T cell tumor infiltration and are associated with improved clinical responses to ICI-based therapy.

HLA Class-I APM Component Upregulation by Chemotherapy

Virtually all classes of chemotherapeutics upregulate HLA class-I APM component expression in various cancer cell lines, cultured in vitro and/or grafted in mice, and enhance their recognition/elimination by cognate CTLs.^13–18 For many drugs, this evidence has been corroborated in the clinical setting. In ovarian cancer patients, both carboplatin and paclitaxel monotherapies significantly upregulate HLA-B expression in malignant cells. The functional implication of this change is indicated by its association with increased intratumoral CD8+ T cell infiltration in ovarian cancer patients treated with taxanes or platinum-based alkylating agents.¹⁹ Likewise, the 5-fluorouracil plus cisplatin combination enhances both HLA class-I expression and stromal CD8+ T cell infiltration in esophageal squamous cell carcinoma patients.²⁰

In addition, chemotherapy-induced MHC class-I antigen upregulation is associated with better responses to ICI-based therapy. In a murine breast cancer cell line displaying MHC class-I antigen downregulation and poor response to anti-PD-L1 mAbs, irinotecan treatment significantly enhanced MHC class-I antigen expression and sensitized the breast cancer cells to PD-L1 inhibitors.²¹ These data have been corroborated by the results from clinical trials which also suggest an increased therapeutic efficacy of ICIs when combined with chemotherapeutics.^22,23 Doxorubicin induction therapy doubled the objective response rate to nivolumab in a phase II trial of metastatic triple-negative breast cancer (TNBC) (35% ORR [6/17 patients] in the doxorubicin induction group vs. 17% ORR [2/12 patients] in the no induction group).²² Likewise, phase III trials in advanced non-small-cell lung cancer patients have shown that the chemotherapy plus ICI combination significantly prolongs patients’ survival compared to chemotherapy alone.^24–26 Although the role of other variables cannot be excluded, chemotherapy-induced HLA class-I APM component upregulation in cancer cells may be an important contributor to the observed synergies.^18,27–29

The mechanisms underlying chemotherapy-induced HLA class-I APM component upregulation in cancer cells include type I interferon (IFNα/β) secretion and subsequent autocrine/paracrine signaling, as well as NF-κB signaling activation (Figure S1, Figure S2).^17,18 These processes are inducible by a number of agents that appear to synergize with ICIs in the clinical setting, possibly reflecting the important role played by HLA class-I expression in ICI response. In this regard, activation of stimulator of interferon genes (STING) by synthetic dinucleotides such as MK-1454 promotes type I interferon secretion by cancer cells. In phase I trials, the MK-1454 plus pembrolizumab combination yielded a 24% objective response rate (6/25 patients) in various advanced cancer types, while MK-1454 monotherapy induced no objective responses (0/26 patients).³⁰ Moreover, retinoid-induced type I IFN secretion through the activation of retinoic-acid inducible gene I (RIG-I),³¹ can account for the MHC class-I antigen upregulation by all-trans retinoic acid on melanoma cells and enhancement of their eradication by CTLs both in vitro and in syngeneic mouse tumor models.³² An ongoing phase II trial investigating the efficacy of the all-trans retinoic acid and ipilimumab combination in stage III/IV melanoma patients (NCT02403778) will assess the clinical relevance of this combinatorial strategy. Furthermore, the RIG-I and STING signaling pathways are activated during the antiviral response³³. Therefore, oncolytic viruses represent an attractive method to enhance HLA class-I APM component expression by cancer cells. Specifically, RIG-I- and STING-driven type I interferon secretion by infected cells and subsequent paracrine signaling are expected to enhance HLA class-I APM component expression in nearby cancer cells. Clinical evidence demonstrates a synergy of oncolytic viruses with ICIs. In a phase II trial of stage IIIB-IV melanoma patients, the talimogene laherparepvec plus ipilimumab combination induced a 39% objective response rate (38/98 patients), significantly higher than that induced by ipilimumab alone (ORR=18% [18/100 patients]).³⁴

HLA Class-I APM Component Upregulation by Radiation Therapy

Radiation upregulates HLA class-I APM component expression in cancer cells both in vitro and in animal models utilizing the same molecular mechanisms as chemotherapy (Figure S1).^17,35–38 At variance with chemotherapy, radiation-induced HLA class-I upregulation has rarely been described in patients. Compared to uveal melanoma patients receiving no neoadjuvant therapy, those receiving neoadjuvant radiotherapy did not display any significant change in either HLA class-I or β₂m expression level.³⁹ Similarly, HLA-A and HLA-B/C expression levels in irradiated pancreatic ductal adenocarcinoma (PDAC) patients were not significantly different from patients receiving no neoadjuvant therapy.⁴⁰ Likewise, HLA class-I expression level on rectal cancer cells was not significantly different between 476 irradiated and 532 non-irradiated patients.⁴¹ It is unclear whether the lack of effect of radiation reflects differences between preclinical systems and clinical settings or flaws in the analysis of the clinical data. We believe the latter to be the case. The conclusions of the clinical studies are not derived from analyses of sequential samples obtained from the same patient but are instead derived from comparisons among tumors surgically removed from irradiated and non-irradiated patients. Patients receiving neoadjuvant radiotherapy are likely to harbor more advanced cancers with greater HLA class-I APM component downregulation, due to selective pressure acting over a longer timeframe.⁴² Therefore, it is possible that radiation-induced HLA class-I upregulation may not be appreciated, since the patients used for comparison are in an earlier stage of the disease and therefore, display a lower frequency of HLA class-I defects. Our hypothesis is supported by the results obtained in sarcoma patients for whom sequential biopsies could be analyzed; in 60% (9/15) of the patients receiving radiation, HLA class-I expression level was significantly higher post-treatment.⁴³ However, other factors including differences in the dose/schedule of radiation among studies cannot be excluded.

Many trials combining radiation with ICIs are ongoing. Results from a phase II trial in metastatic head and neck squamous cell carcinoma patients testing the stereotactic body radiotherapy (SBRT) plus nivolumab combination versus nivolumab alone indicate no significant differences in the objective response rate between the two groups (nivolumab plus SBRT ORR: 29.0% [9/31 patients; 95% CI, 16.1% to 46.6%], nivolumab ORR: 34.5% [10/29 patients; 95% CI, 19.9% to 52.7%], p = 0.86).⁴⁴ Likewise, metastatic TNBC patients receiving nivolumab plus induction radiotherapy in a phase II trial displayed a slightly lower objective response rate than patients receiving nivolumab alone (8% ORR [1/12 patients] in the induction radiotherapy group vs. 17% ORR [2/12 patients] in the no induction group).²² These findings emphasize the need to optimize the dose/schedule of radiation when it is used as an immunological adjuvant, perhaps not with the primary goal of maximizing radiation-induced cytotoxicity and with greater emphasis on its potential impact on a patient’s immune system.^45–47 In the context of antigen presentation, this optimization is especially important as radiation and chemotherapy transiently induce the release from cancer cells of antigens which are loaded on tumor stromal cells and sensitize them to elimination by cognate CTLs. The time-dependence of this process stresses the importance of dose/schedule optimization of anti-cancer therapies in the adjuvant setting.⁴⁸

HLA Class-I APM Component Upregulation by Targeted Therapy

Correcting Epigenetic Repression of HLA Class-I APM Component Transcription

Several epigenetic regulators of HLA class-I APM component expression have been identified including DNA methyltransferases (DNMTs), histone deacetylases (HDACs), and polycomb repressive complex 2 (PRC2) enzymatic component enhancer of zeste homolog 2 (EZH2) (Figure S1). They operate by reducing transcription factor accessibility of HLA class-I APM component chromatin/DNA, thereby resulting in their downregulation.^49–52 The reduced HLA class-I APM component transcription caused by high DNA methylation levels in melanoma cells⁵³ has provided the rationale for a phase II clinical trial testing the DNA methyltransferase inhibitor guadecitabine plus ipilimumab combination; in stage III/IV melanoma patients, a 26% objective response rate was reported (5/19 patients).⁵⁴ DNMT inhibition may also activate endogenous retroviral genes (ERVs) and trigger production of type III interferons.⁵⁵ This mechanism underlies MHC class-I APM component upregulation by CDK4/6 inhibitors, which downregulate DNMT1 expression, and account for the synergy between CDK4/6 inhibitors and anti-PD-L1 mAbs described in murine breast cancer models.⁵⁶ Moreover, HDACs have been shown to repress HLA class-I APM component transcription in Merkel cell carcinoma cells in vitro,⁵⁷ providing the rationale to combine the HDAC inhibitor panobinostat with avelumab in the treatment of a metastatic Merkel cell carcinoma patient refractory to ICI-based therapy due to lack of HLA class-I expression. Panobinostat treatment restored HLA class-I expression by malignant cells, increased intratumoral CD8+ T cell infiltration, and in combination with avelumab, stabilized the disease for 3 months.⁵⁸ Lastly, murine oral squamous cell carcinoma (OSCC) cells have been shown to overexpress EZH2 and downregulate MHC class-I antigens. The restoration of their expression by the EZH2 inhibitor GSK126 sensitizes the OSCC cells to anti-PD-1 mAbs.⁵⁹ A similar finding has been reported in murine malignant prostate tumors.⁶⁰ Overall, DNMT, HDAC, and EZH2 inhibitors are effective at enhancing HLA class-I expression across a wide range of malignancies (Table 2). These inhibitors have the potential to improve the efficacy of ICI-based therapy in cancer types which display HLA class-I APM component expression defects caused by epigenetic mechanisms.

Table 2.

Upregulation of MHC Class-I APM Component Expression by Targeted Therapies in Distinct Cancer Types.

Targeted Molecule/Pathway	Cancer Type	References1
Histone deacetylases (HDACs)	Merkel cell carcinoma2	Ugurel et al. (2019), Ritter et al. (2017)
	Breast cancer	Gameiro et al. (2016)
	Cervical cancer	Mora-García et al. (2006)
	Colon cancer	Khan et al. (2008)
	Gastric cancer	Shan et al. (2011)
	Glioma	Yang et al. (2020)
	Melanoma	Komatsu et al. (1998), Khan et al. (2007), Setiadi et al. (2008), Woan et al. (2015)
	Prostate cancer	Kortenhorst et al. (2013), Kitamura et al. (2007), Setiadi et al. (2008)
DNA methyltransferases (DNMTs)	Melanoma2	Fonsatti et al. (2007), Fazio et al. (2018), Rodríguez et al. (2007), Coral et al. (1999), Fonsatti et al. (2003), Coral et al. (2006), Gasparollo et al. (2001), Serrano et al. (2001), Coral et al. (2013)
	Bladder cancer	Santamaria et al. (1988)
	Breast cancer	Carlow et al. (1985), Sultan et al. (2018)
	Cervical cancer	Ye et al. (2010), Jiang et al. (2010)
	Esophageal SCC	Nie et al. (2001)
	Fibrosarcoma	Bonal et al. (1986), Aboud et al. (1991), Ananthaswamy (1988)
	Glioma	Natsume et al. (2008), Konkankit et al. (2011)
	Non-small-cell lung cancer	Tripathi et al. (2016)
	Oral SCC	Jiang et al. (2010)
	Ovarian cancer	Adair et al. (2009)
	Teratoma	Cremisi (1983)
Enhancer of zeste homolog 2 (EZH2)	Oral SCC3	Zhou et al. (2020)
	Merkel cell carcinoma	Burr et al. (2019)
	Neuroblastoma	Burr et al. (2019)
	Small cell lung cancer	Burr et al. (2019)
Epidermal growth factor receptor (EGFR)	Head and neck cancer4	Srivastava et al. (2015)
	Lung cancer4	Watanabe et al. (2019), He et al. (2013), Garrido et al. (2017)
	Breast cancer	Garrido et al. (2017)
	Colon cancer	Garrido et al. (2017)
	Epidermoid carcinoma	Garrido et al. (2017), Pollack et al. (2011)
	Esophageal SCC	Mimura et al. (2013)
	Gastric cancer	Mimura et al. (2013)
	Melanoma	Garrido et al. (2017)
BRAF	Melanoma4	Sabbatino et al. (2016), Frazao et al. (2017), Sottile et al. (2016), Kakavand et al. (2017)
MEK	Breast cancer	Inoue et al. (2012)
	Colorectal cancer	Gravett et al. (2018), Sers et al. (2009)
	Esophageal SCC	Mimura et al. (2013)
	Gastric cancer	Mimura et al. (2013)
	Non-small-cell lung cancer	Watanabe et al. (2019)
	Papillary thyroid cancer	Angell et al. (2014)
SHP2	Colorectal cancer5	Quintana et al. (2020)
Hedgehog signaling pathway	EBV-associated gastric cancer	Deb Pal et al. (2015)
Lysosome	PDAC5	Yamamoto et al. (2020)
	Lung cancer	Yamamoto et al. (2020)
	Melanoma	Li et al. (2010)
Autophagy	Melanoma	Li et al. (2010)

¹Citations provided in Supplementary Information

²Evidence in patients in combination with checkpoint inhibitors

³Evidence in syngeneic mouse tumor models in combination with checkpoint inhibitors

⁴Evidence in patients as monotherapy

⁵Evidence in syngeneic mouse tumor models as monotherapy.

Abbreviations: EBV, Epstein-Barr virus; PDAC, pancreatic ductal adenocarcinoma; SCC, squamous cell carcinoma.

Correcting HLA Class-I APM Component Downregulation Caused by Dysregulated Signaling Pathways

Dysregulated signaling, including activation of the hedgehog signaling pathway and the receptor-tyrosine kinase (RTK)/mitogen-activated protein kinase (MAPK) pathway, may also downregulate HLA class-I APM component expression in cancer cells (Figure S1).^61–63 Hedgehog-mediated HLA class-I downregulation appears to be restricted to Epstein-Barr virus (EBV)-associated cancers that express latent membrane protein 2A (Table 2).⁶¹ On the other hand, activating mutations or amplifications in the RTK/MAPK pathway (e.g. alterations in EGFR, RAF, RAS, MEK, and/or SHP2) result in HLA class-I APM component downregulation in many cancer types. This effect is mediated by MAPK-driven inactivation of STAT1 and STAT3, which play a major role in HLA class-I APM component transcription.^64,65 In these instances, HLA class-I APM component expression is usually restored by MAPK inhibitors (Table 2). In the clinical setting, treatment of EGFR-mutated non-small-cell lung cancer patients with EGFR inhibitors markedly increased HLA class-I expression by malignant cells.⁶⁶ Moreover, in head and neck cancer patients treated with cetuximab, both HLA-A and HLA-B/C were significantly upregulated in the responding tumors as determined by a volume reduction.⁶⁷ Similarly, in BRAF-mutated melanoma patients treated with BRAF inhibitors ± MEK inhibitors, HLA-A expression post-therapy was enhanced or stable in cases of tumor shrinkage, but was selectively downregulated in progressing tumors.⁶⁸

Synergy has been observed between RTK/MAPK inhibitors and ICIs in preclinical and clinical settings. In mice grafted with syngeneic head and neck squamous cell carcinoma cells displaying high levels of MAPK activation, the MEK inhibitor trametinib inhibited ERK phosphorylation, enhanced MHC class-I antigen expression, and in combination with anti-PD-L1 mAbs, overcame resistance to monotherapy and significantly controlled tumor growth.⁶⁵ In phase I trials, EGFR tyrosine kinase inhibitor plus ICI combinations induced objective response rates of up to 73% in advanced renal cell carcinoma patients (38/52 patients treated with the axitinib plus pembrolizumab combination) and up to 80% in EGFR-mutant non-small-cell lung cancer patients (8/10 patients treated with the gefitinib plus durvalumab combination).^69,70 Furthermore, the cobimetinib, vemurafenib, and atezolizumab combination induced a 93% objective response rate (13/14 patients) in BRAF-mutant melanoma patients.⁷¹ HLA class I upregulation is likely to play an important role in the described exceptional clinical responses.^{62,63,66,67,72,73}

Correcting Abnormalities in HLA Class-I Trafficking

Abnormalities in trafficking of MHC class-I heavy chain-β₂m-peptide trimolecular complexes may cause MHC class-I downregulation in cancer cells (Figure S1). This mechanism underlies HLA class I defects in pancreatic cancer cells, which are characterized by elevated autophagic flux: the autophagy receptor NBR1 mediates the trafficking of HLA class-I trimolecular complexes to the lysosomes. Treatment of cells with a lysosome inhibitor such as chloroquine prevents HLA class-I trimolecular complex degradation and restores their expression on the cell surface.⁷⁴ Guided by this finding, chloroquine was shown to enhance MHC class-I antigen expression on murine PDAC tumors, sensitizing them to dual immune checkpoint inhibition with anti-PD-1 and anti-CTLA4 mAbs. A concomitant increase in intratumoral CD8+ T cells and PD1+TIM3- (non-exhausted) T cell infiltration was observed.⁷⁴ Interestingly, in a phase II trial in PDAC patients, the hydroxychloroquine plus gemcitabine/nab-paclitaxel combination induced a significant increase in tumor immune cell infiltration compared to patients treated with chemotherapy-only.⁷⁵ This finding is consistent with the ability of lysosome inhibitors to induce HLA class-I upregulation on PDAC cells and supports the notion that they will synergize with ICIs in patients. Autophagy-lysosomal pathway-dependent degradation of MHC class-I trimolecular complexes is not restricted to PDAC cells (Table 2). Autophagy inhibition with 3-methyladenine or wortmannin, and lysosome inhibition with chloroquine or bafilomycin A1 in murine melanoma cells upregulated MHC class-I antigen expression in vitro.⁷⁶

Lastly, proprotein convertase subtilisin/kexin type 9 (PCSK9), a protein involved in cholesterol metabolism, can also promote lysosomal trafficking of MHC class-I antigens (Table 2).⁷⁷ Specifically, PCSK9 is secreted by cancer cells and binds to the extracellular α1 region of murine H2-K1 (ortholog of HLA class-I heavy chain). The resulting PCSK9-heavy chain-β₂m-peptide complex is endocytosed and degraded through the endosomal-lysosomal pathway.⁷⁷ Two inhibitors binding to extracellular PCSK9 (alirocumab, evolocumab) have already been approved for the treatment of hypercholesterolemia. These inhibitors enhance MHC class-I antigen expression and show synergy with ICIs in the treatment of murine colorectal cancer cells grafted in syngeneic, immunocompetent mice.⁷⁷ Clinical trials testing anti-PCSK9 mAb plus ICI combinations are warranted.

Conclusion

HLA class-I APM component expression/function are frequently defective in cancer cells; in most cases, expression/function of these components can be restored or enhanced by chemotherapy, targeted therapies, and less so by radiotherapy. The correction of these defects appears to have clinical relevance, as it is associated with increased CD8+ T cell tumor infiltration, enhanced susceptibility of cancer cells to T cell-mediated immunity triggered by ICI-based therapy, and an improved clinical course of the disease. These results provide a convincing rationale to combine chemo-, radio-, and targeted therapies with ICI-based therapy. This strategy is likely to require changes in the dosing and timing of chemo-, radio-, and targeted therapies to maximize their immunogenic effects. Lastly, as HLA class-I APM component defects are associated with acquired resistance to ICI-based therapy, it will be critical to assess their value as biomarkers of ICI response, as well as identify the mechanisms underlying their downregulation. The latter information will be crucial for the rational design of individualized strategies to counteract these defects.