Genetic dysregulation of immunologic and oncogenic signaling pathways associated with tumor-intrinsic immune resistance: a molecular basis for combination targeted therapy-immunotherapy for cancer

Tumor-intrinsic defects in type I and type II interferon signaling

The significance of tumor-intrinsic type II interferon (IFN) signaling to the control of cancer has been appreciated since the mid to late 1990s, when Robert Schreiber’s group published seminal studies documenting the direct action of IFNγ on cancer cells as a requirement for functional tumor immunosurveillance [10, 11]. Subsequent work by several groups has since highlighted various anti-tumor actions of both IFNγ and type I IFN (IFNα/β) on cancer cells, including suppression of cell cycle progression, induction of (or sensitization to) apoptosis, and potentiation of tumor antigen presentation to CTL [12, 13]. Perhaps not surprisingly, defects in type I and type II IFN signaling pathways have emerged in recent years as critical determinants of tumor resistance to immunotherapy [14]. James Allison’s group was the first to report this link, highlighting in independent melanoma patient cohorts that IFN pathway gene copy number alterations and single nucleotide variants were both enriched in patients who failed to respond to anti-CTLA-4 ICB therapy [15]. Others have since reported genome-level aberrations in IFN pathway genes as correlates of primary and acquired resistance to PD-1 blockade [16–19]. Collectively, these studies have documented in tumors of non-responding patients a variety of IFN pathway-associated genetic aberrations, from amplification of negative regulator genes (i.e., SOCS1 and PIAS4) to inactivating mutations or copy loss events among genes relevant to all aspects of the IFN pathway, including IFN receptors, kinases associated with IFN signaling (i.e., JAK and TYK kinase genes), transcription factors (i.e., STAT and IRF genes), and IFN-stimulated gene (ISG) targets. Consistent with their role as effectors of anti-tumor immunity, the collection of ISGs found to harbor genetic defects in these studies includes those encoding pro-apoptotic factors and antigen processing and presentation machinery (discussed in more depth below). In addition, recent data from experimental studies employing CAR-T cells have shown that ISGs with cell adhesion function are also critical to therapeutic efficacy against solid tumors. Using a genome-wide CRISPR-Cas9 screening approach to identify genes associated with CAR-T resistance, Larson et al. found that loss of IFNγ pathway genes, including IFNGR1, JAK1, or JAK2, was associated with CAR-T resistance in multiple solid tumor models and that impaired IFNγ responsiveness reduced the duration and avidity of CAR-T binding to tumor cells [20]. Although this mechanism did not impact CAR-T efficacy against liquid cancers, it likely accounts for some of the challenges that have plagued CAR-T therapy for non-hematologic malignancies to date.

In addition to genomic defects that compromise IFN pathway activity, tumor-intrinsic loss of sensitivity to IFN signaling can also be achieved by transcriptional repression of genomically intact IFN pathway genes. Indeed, several studies have documented the prognostic significance of IFN pathway gene expression signatures in tumor tissue as correlates of response to immunotherapy [21–25]. Although many of these studies report outcomes of whole transcriptome RNA-sequencing analyses that do not distinguish gene expression levels in cancer versus non-cancer cell populations in biopsied tissue, experimental work in murine models has provided compelling evidence that intrinsic suppression of IFN pathway genes in cancer cells limits the efficacy of anti-tumor immunity. In a comparative study of PD-1/PD-L1 inhibitor responsive CMT167 tumors versus the non-responsive Lewis lung carcinoma (LLC), Bullock et al. purified GFP-negative cancer cells growing in lungs of GFP-positive transgenic mice and found that IFNγ pathway gene expression was significantly repressed in LLC cells [26]. Using an alternate approach, Sehgal et al. employed single cell RNA-sequencing technology to evaluate gene expression signatures in residual MC38 colon adenocarcinoma cells that persist following anti-PD-1 treatment of murine-derived organotypic tumor spheroids (MDOTS). Following separation of MC38 cells from IgG or anti-PD-1-treated spheroids, transcriptionally distinct cell states in the anti-PD-1-treated group included two dominant clusters that were each marked by significant downregulation of type I and type II IFN gene expression signatures [27]. While expression of IFN pathway genes in non-cancer cell populations in the TME also plays an important role in the outcome of anti-tumor immunity (reviewed recently in [14]), these studies demonstrate the significance of tumor-intrinsic transcriptional regulation of these genes as a critical determinant of immune resistance versus susceptibility.

Among the ways in which tumor cells reduce expression of IFN pathway genes, epigenetic suppression is well-documented, with several mechanisms of silencing IFN pathway gene expression having been linked to tumor-intrinsic immune resistance and cancer progression. Owen et al. recently demonstrated in a murine model of castration-resistant prostate cancer that IFN pathway gene expression is retained in dormant cells, but lost in actively proliferating cells, from progressing bone metastatic lesions [28]. Consistent with the loss of IFN signaling and immunogenicity that has been observed in bone metastases of prostate cancer patients, this phenomenon was attributed to shifts in histone acetylation, as the reduced expression of IFN pathway genes in activated, proliferating cells isolated from tumor-bearing animals could be restored by histone deacetylase (HDAC) inhibition. Importantly, this intervention also impeded outgrowth of bone metastases in vivo and enhanced the efficacy of systemic immunotherapy, suggesting that therapeutic reversal of HDAC-mediated IFN pathway gene silencing in cancer cells may increase their susceptibility to anti-tumor immune responses in patients as well.

Epigenetic imprinting in the form of aberrant methylation patterns has also been found to influence IFN pathway gene expression in the context of cancer. In this regard, histone dimethylation of IFN pathway gene bodies limits ISG expression specifically in cancer cells [29], and DNA hypermethylation and repression of IFN pathway genes has been reported in several cancer types, including Epstein Barr virus-associated Burkitt’s lymphoma and gastric carcinoma [30] as well as NSCLC [25]. Of note, this latter study also demonstrated that promoter hypermethylation of IFN pathway genes in pre-treatment tumor tissue correlated with failed response to ICB therapy. These reports underscore the significance of other studies demonstrating the utility of DNA methyltransferase inhibitors such as 5-Aza-2’-deoxycytidine in promoting ISG expression in tumor tissue. Indeed, this epigenetic modifier has been found to synergize with both type I IFN cytokine therapy as well as combination therapy involving type I IFN and a DC-targeting cancer vaccine in a murine melanoma model [31, 32].

Finally, the influence of chromatin dynamics on IFN pathway activity and immune sensitivity of cancer cells has further been highlighted by a recent study on clear cell renal cell carcinoma (ccRCC). In this study, loss-of-function mutations in the PBRM1 gene that encodes a subunit of the SWI/SNF complex correlated with IFN pathway gene expression and improved clinical response to ICB therapy [33]. Though genomic aberrations in SWI-SNF complex components are not consistently associated with ICB outcomes across diverse cancer types [34], these data highlight a role for chromatin dynamics as a determinant of cancer cell sensitivity to IFN and immune responsiveness in certain contexts.

Similar to the diverse mechanisms by which epigenetic modifiers regulate IFN pathway gene expression in the context of cancer, a number of post-translational regulators can also interfere with IFN pathway activity in cancer cells. Many of these regulators have been identified in murine tumor models with CRISPR-Cas9 screens designed to detect immune resistance genes. Using such an approach, the Apelin receptor (a positive regulator of IFNγ signaling) was found to support the efficacy of ICB and ACT therapeutic regimens for melanoma, a finding that supported clinical observations of multiple loss-of-function mutations in the APLNR gene in tumors refractory to immunotherapy [35]. Likewise, CRISPR screens as well as gene silencing studies have revealed direct immune evasion roles for genes that encode negative regulators of IFN pathway signaling. For instance, silencing or depletion of SOCS1, PTPN2, ADAR1, or FOXA1 can sensitize tumor cells to IFNγ signaling and immunotherapy in various murine models [26, 36–40]. Because many of these IFN pathway inhibitors are upregulated in human cancers, therapeutically targeting these post-translational mechanisms of IFN pathway suppression may be a useful strategy to overcome tumor-intrinsic resistance to immunotherapy.

Based on the diverse mechanisms by which cancer cells evade the anti-tumor actions of type I and type II IFN, several strategies that aim to restore IFN pathway activity and overcome tumor-intrinsic immune resistance have been explored as part of combinatorial regimens with immunotherapy [14]. As alluded to above, epigenetic modifiers including HDAC and DNA methyltransferase (DNMT) inhibitors can restore IFN pathway activity directly in cancer cells, not only by activating expression of silenced IFN pathway genes themselves but also by promoting expression of endogenous retroviral RNA elements which in turn stimulate innate sensors that promote IFN signaling. Indeed, epigenetic reactivation of previously dormant retroviral transposons is particularly well-documented as a driver of ISG expression in both cancer and non-cancer settings in which DNMT are genetically disrupted or targeted therapeutically, and the resulting IFN pathway activation has been shown to depend specifically on innate sensing by TLR3 and MDA5 in some cases [41–45]. Interestingly, it has also been documented that cancer cells often suppress IFN responsiveness by actively silencing retrotransposons [46]. Based on these studies and the prevalence of transposons in the human genome, these foreign genetic elements likely play an underappreciated role as regulators of cancer susceptibility to immunotherapy, particularly in the context of combination therapies involving epigenetic modifiers. As such agents continue to be investigated in the context of combinatorial regimens going forward, it is also important to recognize that because HDAC- and DNMT-associated chromatin modifications impact expression of an array of target genes, therapeutics targeting these enzymes are likely to influence expression of non-IFN pathway genes in cancer cells as well, and they may also act on other cell populations within the TME in both IFN-dependent and IFN-independent ways. Though the broad consequences of such epigenetic interventions will need to be studied in more depth in the years ahead, it is worth noting that such strategies have already been found to potentiate anti-tumor immune responses when combined with ICB therapy, and in some cases these synergistic effects were directly related to increased IFN pathway activity in cancer cells themselves [47–50].

In addition to approaches for activating expression of genomically intact IFN pathway genes, it is also possible to bypass canonical type I and type II IFN signaling pathway components that might be mutated in cancer cells by targeting innate RNA or DNA sensors that operate independently of IFNγ and IFNα/β receptors. Such approaches also have the potential to bypass negative regulators of upstream IFN pathway signaling molecules that may be hyperactive in cancer cells. Activation of these IFN salvage pathways has been successfully achieved in cancer cells treated with agonists for several RNA sensors, including RIG-I, PKR, and TLR3, and these maneuvers have been shown to overcome tumor-intrinsic resistance to both ICB and ACT therapies [51, 52]. Similarly, stimulation of the cytoplasmic DNA sensor STING can also support IFN pathway activation and anti-tumor immunity. In elegant studies in a murine model of pancreatic ductal adenocarcinoma, Vonderhaar et al. found that the immunostimulatory effects of STING agonism were not mediated by tumor-extrinsic type I IFN signaling, as the anti-tumor efficacy achieved by STING activation was the same in wild-type and IFNAR1^–/– mice. Rather, STING-mediated type I IFN signaling in tumor cells promoted expression of CXCL9 and CXCL10 chemokines, which in turn enhanced tumor infiltration by CXCR3+ T lymphocytes that were necessary to achieve therapeutic benefit [53]. Interestingly, though STING agonism produced similar therapeutic benefit in the B16-F10 murine melanoma model, its efficacy was dependent on type I IFN signaling in host cells, rather than transplanted tumor cells [53, 54]. Similar tumor-extrinsic requirements for STING activation have been described in tumor challenge studies with a mouse oral squamous cell carcinoma cell line in STING-deficient mice [55]. Differences in tumor- versus host-targeted effects in these studies may be inherent to the specific tumor models themselves or, alternatively, to the use of different STING agonists. Nevertheless, data from these and related reports highlight both tumor-intrinsic and -extrinsic mechanisms by which STING agonism can enhance anti-tumor immune responses, particularly when coupled with ICB therapy [55–57]. Finally, it is worth noting that agonists of endosomal TLR7/8/9 DNA sensors have also been shown to mediate IFN-associated synergism with immunotherapy, though these effects have largely been attributed to tumor-extrinsic influences of the agonists on other cell populations in the TME, including stromal cells and tumor-infiltrating DC [58–62]. Regardless of the mechanism, early reports from clinical trials have highlighted the potential of such agonists to support anti-tumor immunity when used in combination with ICB, in many cases overcoming primary and acquired resistance to prior ICB monotherapy [63, 64]. As these strategies continue to be translated into clinical interventions, the challenge going forward will be to determine optimal ways of achieving the desirable anti-tumor functions of IFN activity without driving immune resistance mechanisms that also accompany chronic type I and type II IFN signaling in cancer cells. In this regard, it is worth noting that despite the many pro-immune functions of type I and II IFN as outlined above, in certain contexts IFN signaling in cancer cells has been shown to be detrimental to anti-tumor immune reactivity, driving expression of several immune checkpoint ligands, limiting tumor mutational burden through activation of the DNA damage response, and driving cancer stemness [65–68]. Interventions that aim to epigenetically rewire expression of particular IFN pathway genes, or specific genetic programs activated by IFN pathways, may indeed enable context-dependent manipulation of the broad downstream targets of IFN signaling in a way that ultimately favors the anti-tumor, rather than pro-tumor, functions of these cytokines.

Tumor-intrinsic defects in antigen processing and presentation

One of the major immunologic benefits of IFN signaling in cancer cells is the ability of both type I and type II IFN pathways to induce expression of genes encoding MHC molecules and related antigen processing and presentation machinery. Collectively, expression of these gene products enables tumor antigen recognition by effector CTL. However, even in tumor cells that maintain functional IFN pathway activity, tumor-intrinsic defects in antigen processing and presentation pathways themselves can confer resistance to CTL-mediated immunity [69]. Indeed, mutations and loss of heterozygosity in alleles encoding HLA molecules are well-described in studies of tumor immune escape [70–72]. Likewise, point mutations, deletions, and loss of heterozygosity in the gene encoding β2-microglobulin, which is needed for stable expression of cell surface HLA/antigen complexes, have been shown to emerge over the course of cancer progression and often correlate with resistance to immunotherapy [17, 73, 74]. Other genomic aberrations that compromise the function of transport proteins (TAP1/TAP2) and chaperones (tapasin, ERAP1, etc.) involved in antigen loading onto HLA molecules are also associated with tumor progression and recurrence following immunotherapy [75, 76].

As with IFN pathway signaling components, transcriptional repression can also inhibit the expression and function of antigen processing and presentation machinery, thereby interfering with tumor antigen presentation to T lymphocytes. Across diverse cancer types, several studies have shown that HLA expression is diminished or lost when genes encoding these molecules or antigen processing/presentation pathway components have hypermethylated promoters [69, 75, 77, 78]. Of note, transcriptional repression mediated by such promoter methylation has been reported in patient tumors exhibiting acquired resistance to ACT + ICB combination immunotherapy, highlighting the significance of this epigenetic regulation as a mechanism of tumor evolution and immune escape [79]. Others have shown that altered histone methylation/acetylation patterns at loci relevant to the antigen processing and presentation pathway also suppress HLA expression and tumor antigen presentation by cancer cells [80, 81]. Importantly, the identification of epigenetic regulators that control DNA/histone modifications and other determinants of gene activity has provided crucial molecular insights into the dysregulation of tumor antigen presentation in recent years. In this regard, a broad survey of 16 solid tumor types recently revealed a high frequency of promoter methylation, copy number loss, and missense mutations in the gene encoding NLRC5, a critical component of the CITA enhanceosome that transactivates expression of genes encoding class I MHC molecules and other antigen processing machinery [82]. Not surprisingly, loss of NLRC5 expression/function in tumor tissue correlates negatively with expression of effector CTL markers and is a poor prognostic factor for overall patient outcome as well as specific response to CTLA-4 or PD-1 checkpoint blockade [82, 83]. More recently, Ren et al. also identified a specific role for one of NLRC5’s binding partners, the WHSC1 histone dimethyltransferase, in the regulation of class I MHC-restricted anti-tumor immunity. In this study, it was found that reduced expression of WHSC1 in colorectal cancer specimens is associated with poor T cell infiltration and patient outcome, and mechanistic studies in a murine model demonstrated that loss of this protein led to transcriptionally repressed chromatin signatures at class I MHC-related genes [84]. The SAFB scaffold attachment factor was also recently found to drive transcriptional repression of MHC class I antigen processing machinery in cancer cells. In this study, MYC-induced SUMOylation of SAFB was found to stabilize this protein, which in turn was responsible for downregulation of several genes essential to tumor antigen presentation, including HLA-A, HLA-B, B2M, TAP1, LMP2, and LMP7 [85].

Post-transcriptional and translational mechanisms of regulating antigen processing and presentation in tumor cells have also come to light in recent years. At both levels of regulation, non-coding RNAs are emerging as important oncogenic factors that limit tumor antigen presentation. For example, miR-27a directly targets a sequence in the 3’ UTR of calreticulin mRNA, thereby diminishing translation of a key chaperone needed to ensure proper folding and peptide loading of class I MHC molecules. Of note, this miRNA is upregulated in advanced colon cancers and correlates negatively with HLA expression and CD8+ T cell infiltration of tumors [86]. Similar immunophenotypes have been reported in triple negative breast cancers overexpressing the lncRNA LINK-A, which is a poor prognostic factor for response to PD-1 checkpoint blockade. Mechanistically, LINK-A inhibits PKA-driven phosphorylation of the E3 ubiquitin ligase TRIM71, in turn promoting its ubiquitin-mediated degradation of several components of the peptide-loading complex [87]. While these post-translational mechanisms of disrupting tumor antigen presentation to CD8+ T cells act prior to delivery of HLA/antigen complexes to the cell surface, others function to limit recognition of these complexes following their transport to this destination. One such mechanism involves shielding of HLA/antigen complexes by aberrantly expressed glycosphingolipids on cancer cells. Cell surface glycosphingolipid expression is controlled by the SPPL3 protease, which typically cleaves and inactivates the glycosyltransferase enzyme B3GNT5. Downregulation of SPPL3 gene expression in low-grade gliomas was found to disrupt this axis, leading to an accumulation of cell surface neolacto-series glycosphingolipids, which sterically interfere with HLA molecule recognition by CD8+ T cells. Interestingly, in vitro studies with pharmacologic inhibitors of glycosphingolipid synthesis could restore T cell recognition of otherwise shielded tumor cells, highlighting the potential of these agents to overcome this mechanism of tumor immune escape [88]. Another study has described a role for MAL2 in the loss of cell surface HLA expression on breast cancer cells. This transmembrane protein is overexpressed in aggressive breast cancers and was found to promote endocytosis and degradation of HLA/antigen complexes, thereby limiting anti-tumor T cell reactivity [89].

In light of the diverse mechanisms by which cancer cells may disrupt tumor antigen processing and presentation, a number of strategies for restoring the integrity of this pathway have been investigated. In cases where pathway defects arise from loss-of-function mutations, immunotherapeutic options are indeed limited. Outside of gene therapy, whose long-term efficacy would require restoration of gene function in all tumor cells harboring a given mutation in order to prevent simply selecting for outgrowth of unrepaired cells (a challenge that continues to plague this approach), CAR-T-based therapy that bypasses self MHC restriction and targets non-restricted cell surface antigens is currently the only viable immunotherapeutic option for overcoming genome-level defects in antigen processing and presentation pathway components. Where defects in this pathway arise from transcriptional repression, however, there are indeed attractive options for targeted interventions to support traditional T cell-based immunotherapies. Pharmacologic inhibition of various components of the PRC2 polycomb repressive complex, such as EED and EZH2, blocks transcriptionally repressive histone methylation and enhances the expression of HLA molecules and several antigen processing proteins in cancer cells, making them susceptible to CD8+ T cell reactivity [81]. Likewise, DNA methyltransferase and histone deacetylase inhibition can each epigenetically reprogram chromatin to activate expression of antigen processing and presentation machinery, in many cases potentiating the efficacy of ICB therapy in preclinical models [90–93]. Targeting overexpressed non-coding RNAs and regulatory proteins that mediate post-transcriptional or post-translational suppression of antigen processing and presentation pathway components also offers opportunities to support T cell based immunotherapies. Even in the absence of active pathway suppression, pharmacologic agents that augment the activity of antigen processing components can also increase tumor antigen presentation to T cells. Such an outcome was recently achieved with the herbal medicine-derived small-molecule compound atractylenolide, which overcomes inherent limitations in the catalytic activity of the immunoproteasome, enhancing its antigen processing capacity in a way that synergized with ICB therapy to promote tumor control and survival [94]. Finally, though genomic aberrations in TAP transporters have been linked with tumor progression and resistance to immunotherapy due to their impact on the presentation of traditional T cell epitopes, there is accumulating evidence that TAP-independent self peptides (which are not presented to developing T cells during thymic selection) can be exploited as immunogenic neoantigens in tumors that lose TAP function as a mechanism of immune escape [95, 96]. Immunization against such neoantigens in patients harboring TAP-deficient tumors may therefore increase the breadth of the anti-tumor T cell repertoire and improve immune reactivity against tumors that have evolved to escape responses directed against TAP-dependent antigens. Actively targeting functional TAP in tumor-specific ways can also promote TAP-independent neoantigen presentation and enhance the efficacy of immune-based therapies. Such an approach has already been achieved preclinically with TAP-targeted siRNA conjugated to a nucleolin aptamer, and this intervention was found to improve the efficacy of PD-1 checkpoint blockade [97]. Because tumor neoantigens derived from mutated oncoproteins become biased toward peptides with low binding affinity for class I MHC molecules over the course of tumor progression [98], strategies that increase neoantigen load by eliciting T cell responses against non-mutated, TAP-independent neoantigens may be particularly promising adjuncts to immunotherapy.

Tumor-intrinsic resistance to immunologic cell death pathways

CTL-mediated killing of cancer cells depends not only on tumor antigen recognition but also on the functional operation of various immunologic cell death pathways. These pathways can be mediated by direct CTL engagement of death receptors on tumor cells, such as FAS and TRAIL-R, as well as by granzymes that enter CTL-engaged tumor cells via perforin pores. In both cases, cell death is typically triggered by activation of a caspase cascade that culminates in the induction of tumor cell apoptosis. In the case of the perforin pathway, granzymes that enter target cells function as serine proteases that cleave inactive procaspases to yield active caspases with apoptosis-inducing enzymatic activity. To overcome this mechanism of immune-mediated cell death, several cancer types have been found to upregulate expression of the serine protease inhibitor serpinB9 (also known as PI-9), which suppresses granzyme B activity and confers tumor resistance to CTL-mediated killing [99–102]. Of note, it has been demonstrated in a murine melanoma model of acquired resistance to anti-CTLA-4 ICB therapy that expression of the Serpinb9 gene is significantly upregulated in resistant versus parental tumors. Similarly, elevated expression of this gene is a poor prognostic factor for progression-free and overall survival of melanoma patients treated with CTLA-4 blockade [103]. In keeping with these findings, a recent study reported immune potentiating effects of a serpinB9 small molecule inhibitor. This drug, which reduced serpinB9/granzyme B complex formation and enhanced tumor-intrinsic caspase-3 activity in melanoma cells, also promoted remodeling of the TME, with a shift in the immune compartment from immunosuppressive populations to effector T cell populations [104]. Collectively, these data suggest that serpinB9 inhibition has the potential to enhance immunotherapeutic efficacy through a combination of tumor-intrinsic and tumor-extrinsic mechanisms of action.

Intrinsic defects in death receptor signaling also contribute significantly to tumor immune evasion [105]. Tumor cells may escape the action of these pathways by mutating or downregulating expression of death receptors themselves [106], adaptor proteins such as FADD that mediate signaling through functional death receptors [107], and caspases that act as the effectors of death receptor signaling [108, 109]. Increased expression of proteins that negatively regulate this pathway, such as c-FLIP, can also circumvent cell death signals in tumor cells and promote immune evasion [110, 111]. Likewise, dysregulation of alternative death receptor signaling pathways, including the mitochondrial apoptotic pathway and the necroptotic cell death pathway, can confer anti-tumor immune resistance.

Death receptor activation of the mitochondrial cell death pathway is controlled by a variety of tumor-intrinsic properties, including metabolic and genetic factors. In terms of tumor metabolism, glucose addiction is a vulnerability that can be targeted to promote mitochondrial death via death receptor signaling in many cancers. The glycolytic inhibitor 2-deoxy-d-glucose, as well as glucose deprivation, can sensitize cancer cells to apoptosis mediated by TNF-family death receptors [112–114], suggesting that strategies for targeted interference with glucose metabolism in cancer cells have the potential to support immunotherapies that rely on death receptor signaling to achieve anti-tumor efficacy [115]. In addition to metabolic regulation of the mitochondrial cell death pathway, anti-apoptotic factors such as BCL-2 also play a particularly important role in escape from mitochondria-associated cell death. Often expressed at elevated levels in cancer cells, BCL-2 and related anti-apoptotic family members confer resistance to T cell-mediated killing, the significance of which is underscored by studies documenting BCL-2 inhibitors as sensitizing agents that improve the efficacy of ICB regimens [116–118]. Apoptosis regulators and other death receptor pathway genes are also key determinants of CAR-T cytolotyic activity, as evidenced by recent in vitro studies assessing CD19-redirected CAR-T cell killing of acute lymphblastic leukemia (ALL) cells. Not only did these studies reveal roles for pro-apoptotic signaling molecules and anti-apoptotic molecules as determinants of tumor cell susceptibility to CAR-T-mediated killing via FasL or TRAIL, they also found that tumor resistance to death receptor pathway signaling led to progressive CAR-T dysfunction in the form of an exhausted phenotype, which developed as a result of persistent antigenic signaling from resistant tumor cells. These data offer an explanation for clinical observations that reduced expression of death receptor pathway genes in pre-treatment leukemia-infiltrated bone marrow correlates with CAR-T dysfunction and poor survival of ALL patients [119].

Resistance to immunotherapy has also been attributed to tumor-intrinsic dysregulation of the necroptotic cell death pathway. RIPK1 and RIPK3, key regulators of this pathway, have emerged as particularly interesting determinants of ICB resistance. Though RIPK1 mediates necroptosis by promoting TNF family receptor signaling through TRADD, cancer cells can divert TNF receptor signaling away from this cell death pathway and toward tumor-promoting, pro-inflammatory signaling. Collectively, these effects confer RIPK1-associated resistance to ICB therapy, whereas disrupting Ripk1 gene function enhances ICB efficacy in a T cell- and caspase 8-dependent manner [120]. At the same time, others have shown that intratumoral delivery of necroptotic cells induced to die by transduction with a constitutively active version of RIPK3, a binding partner of RIPK1 in the necrosome, synergizes with ICB therapy to promote control of established tumors [121]. Future insights into the factors that direct RIPK1/RIPK3-mediated signaling toward desirable necroptotic cell death outcomes and away from pro-inflammatory outcomes will be important for developing strategies to manipulate this pathway in ways that favor pro-immune and anti-tumor signaling. Collectively, these and related interventions that aim to restore the integrity of immunologic cell death pathways in immune-resistant tumors have the potential to significantly increase the impact of cancer immunotherapy.

Tumor-intrinsic activation of the RAS/MAPK pathway and immune resistance

Comprehensive genomic analyses of all cancer types catalogued in The Cancer Genome Atlas have revealed the RAS/MAPK pathway as the most frequently altered cell signaling pathway in cancer cells [122]. Gain-of-function mutations as well as copy number and epigenetic alterations in genes encoding receptor tyrosine kinases, RAS GTPases, and MAP kinases such as BRAF and MEK can all promote RAS/MAPK pathway activity, leading to uncontrolled cell division, enhanced cell survival, and cellular dedifferentiation [123]. Additionally, activation of this pathway has also been shown to mediate immune evasion by a number of mechanisms, including those associated with T lymphocyte recruitment to the TME, suppression of T lymphocyte effector function within the TME, and cancer cell susceptibility to CTL-mediated killing.

In terms of T lymphocyte recruitment, genomic/transcriptomic activation of RAS/MAPK signaling has been correlated with reduced numbers of tumor-infiltrating lymphocytes (TIL) in murine models as well as cancer patients [124–126]. These effects have been reported in the context of both endogenous anti-tumor responses as well as in adoptive transfer settings. One study found that MAPK pathway activation in BRAF-mutant melanomas reduced T cell infiltration of tumors by activating VEGF secretion [124], likely contributing to vascular abnormalities that preclude efficient T cell influx into tumors. Another study using KRAS-driven models of lung cancer found that this pathway limits cancer cell secretion of the T cell chemoattractant CXCL10 [126].

Multiple mechanisms by which RAS/MAPK pathway activation suppresses the activity of T lymphocytes that do infiltrate tumor tissue have also been described. Several studies have reported that this pathway promotes expression of the PD-L1 checkpoint ligand on cancer cells [125, 127, 128]. In tumor models engineered to enable inducible expression of constitutively active KRAS^G12V expression, oncogene activation triggers mitochondrial dysfunction and an increase in reactive oxygen species, which in turn promote PD-L1 expression by enhancing EGFR and FGFR1 growth factor receptor activity [127]. Work in a KRAS^G12D mutant tumor model has also shown that constitutive KRAS activation drives eIF4-mediated translation of ARF6 and AMAP1, which cooperate to recycle intracellular PD-L1 for re-expression on the cell surface [128]. Another cell surface inhibitory molecule whose expression has been linked with activating NRAS or BRAF mutations is the ectoenzyme CD73, which catabolizes extracellular ATP into the immunosuppressive oncometabolite adenosine [129]. In addition to these cell surface immunoregulators, tumor cell secretion of immunosuppressive molecules, including IL-6 and IL-10, is enhanced by MAPK pathway signaling as well, further contributing to the immunologically hostile nature of TMEs in which this pathway is hyperactive [130].

Finally, pathways associated with the recognition and response of anti-tumor T lymphocytes are also dysregulated by oncogenic RAS/MAPK signaling in cancer cells. Oncogenic signaling through BRAF^V600E has been shown to limit tumor cell expression of class I and class II MHC molecules via multiple mechanisms. In the case of class I MHC, BRAF^V600E signaling triggers rapid internalization and intracellular sequestration of class I MHC molecules within endolysosomal compartments [131]. At the same time, signaling mediated by this oncoprotein also blocks transcription of genes encoding class I and class II MHC [132, 133]. This latter mechanism is at least in part attributable to the fact that RAS/MAPK pathway activity also limits tumor cell responsiveness to both type I and type II IFN, blocking expression of several IFN pathway target genes with important anti-tumor functions [125, 132, 134–136].

Although small molecule inhibitors targeting the RAS/MAPK pathway have shown efficacy against a range of cancer types, tumor-acquired resistance and disease relapse occur in a majority of patients [123]. Importantly, following discontinuation of targeted therapy, these patients also respond poorly to subsequent immunotherapy [137–140]. The development of such cross-resistance results from several factors associated with tumor evolution in response to prior MAPK pathway inhibition. In vitro studies with human melanoma cell lines harboring BRAF^V600 mutations have demonstrated that prolonged treatment with BRAF inhibitors leads to loss of both shared melanocyte differentiation antigens as well as a unique neoantigen, compromising tumor recognition by autologous TIL and antigen-specific CTL clones [141]. Recent work in preclinical murine models has also shown that acquired resistance to BRAF inhibition, mediated by constitutive reactivation of MEK, triggers a change in the chromatin landscape and transcriptional output of tumor cells that confers resistance to both ICB and ACT therapies. This unique genetic program, not present in targeted therapy-naïve tumors with hyperactive MAPK signaling, involves a downregulation of type I IFN pathway genes and upregulation of immune evasion genes and could be reversed by subsequent inhibition of MEK, which in turn synergized with concomitant immunotherapy to promote durable tumor control [136]. An alternate mechanism for targeted therapy-induced cross-resistance to immunotherapy has also recently been reported by Terp et al., who found that sustained activation of p38 MAPK signaling compensated for MEK inhibition, leading to upregulation of CD73 in several murine tumor models. Targeting CD73 in conjunction with MEK also improved tumor control in this setting [142]. These findings likely explain the synergistic anti-tumor effects that have previously been observed when BRAF/MEK inhibition is combined with targeted inhibition of the A2A adenosine receptor in a murine melanoma model [143]. Finally, others have shown that prolonged exposure and resistance to BRAF or MEK1/2 inhibitors are associated with tumor-extrinsic shifts in infiltrating immune cell populations, marked by a notable accumulation of MDSC in the TME [144, 145]. This shift in the TME after prolonged exposure to targeted therapy coincides with a lack of response to anti-PD-1 treatment, which is otherwise efficacious when administered shortly after targeted therapy, at a time when tumor-infiltrating MDSC are reduced by MAPK pathway inhibition [145].

Based on the collective ways in which hyperactive RAS/MAPK signaling in cancer cells confers immune resistance, there is indeed strong rationale for coupling pathway-targeting interventions with immune-based treatment regimens. In considering these types of coupled approaches, insights into the immune cross-resistance that accompanies resistance to MAPK pathway inhibitors underscores the significance of therapeutic scheduling in the design of such strategies, arguing for combinatorial, rather than sequential, treatment regimens. Indeed, several preclinical studies have shown that MAPK pathway inhibitors support a range of immunotherapies, including ICB, ACT, and CAR-T regimens, when administered in conjunction with the particular immune-based therapy under investigation [125, 133, 136, 146, 147]. Early results from some of the clinical trials investigating combination front-line therapy with MAPK pathway inhibitors and ICB for BRAF^V600 mutant melanoma have shown similar promise, though treatment-related adverse events have led to discontinuation of therapy in some patients [148, 149]. To maximize the safety and efficacy of such combinatorial regimens going forward, it will therefore be important to: (1) define the optimal duration of combination therapy administration and (2) identify prognostic biomarkers that predict those patients most likely to benefit from such regimens. Finally, in light of the suppressive effects that constitutive MAPK pathway signaling has on IFN responsiveness in cancer cells, it is also interesting to consider how oncolytic viruses might be leveraged as an alternative to targeted therapy for use as an immune-supporting agent against RAS/MAPK pathway hyperactive tumors. In this regard, loss of IFN responsiveness in such tumors increases their susceptibility to oncolytic viruses [134], which can ultimately trigger immunogenic cell death to support both natural and therapy-associated anti-tumor immune responses. Collectively, these approaches to target the tumor-intrinsic vulnerabilities manifested by hyperactive RAS/MAPK signaling have the potential to improve the reach of immunotherapy in the large percentage of cancer patients whose tumors are driven by this pathway.

Immunologic consequences of tumor-intrinsic WNT signaling

Another oncogenic pathway that has frequently been linked with tumor immune resistance is the WNT/β-catenin pathway. Hyperactivation of this pathway in cancer cells may be achieved in a variety of ways, including overexpression of autocrine-acting WNT family growth factors, silencing of genes encoding WNT antagonists, gain-of-function mutations in β-catenin, and loss-of-function mutations in the APC tumor suppressor or other components of the β-catenin destruction complex that otherwise limit pathway activity by targeting β-catenin for proteasomal degradation in the absence of WNT signaling [150, 151]. In addition to these and other genetic aberrations that activate mediators of canonical WNT signaling, genetic dysregulation of non-canonical, β-catenin-independent WNT pathways also contribute to tumor progression and immune evasion [152]. Importantly, these effects are mediated not only by tumor-intrinsic consequences of WNT pathway activation in cancer cells themselves but also by extrinsic changes to the TME that are triggered by cancer cells in which WNT pathways are dysregulated.

As with dysregulated RAS/MAPK signaling, tumor-intrinsic WNT signaling can promote immune evasion in several ways. Non-canonical signaling mediated by WNT5A has been shown to reduce CTL recognition of cancer cells by activating PKC- and STAT3-dependent downregulation of tumor-associated antigen gene expression [153]. At the same time, activation of the canonical WNT/β-catenin pathway induces expression of immune checkpoint ligands, including PD-L1 and CD47, the latter of which negatively regulates innate immunity by transmitting a so-called “don’t eat me” signal through SIRPα expressed on phagocytes [154–156]. While these negative regulators are easily targeted with checkpoint inhibitors, several studies have shown that WNT signaling also confers resistance to ICB therapy [157–163], highlighting checkpoint-independent immune evasion functions for this pathway as well. Indeed, the majority of work on WNT pathway activity in the context of cancer immune evasion has revealed that WNT signaling promotes tumor immune exclusion [164], a phenomenon that of course precludes responsiveness to checkpoint inhibitors whose efficacy relies on tumor-infiltration by T lymphocytes. This mechanism of WNT-associated immune evasion was first reported by Spranger et al., who found that WNT/β-catenin signaling blocked tumor cell secretion of CCL4 and CXCL1, impeding tumor-infiltration by CD103+ DC needed to prime anti-tumor CD8+ T cell responses [157]. Others have since shown that tumor-intrinsic β-catenin also dampens expression of CXCL9, CXCL10, and CXCL11, directly interfering with recruitment of CXCR3+ T lymphocytes into the TME [165]. Not only does this effect have consequences for endogenous anti-tumor T cells that might be supported by ICB therapy, but it likely also contributes to the exclusion of adoptively transferred T cells that has been observed in ACT therapy-resistant tumors [166].

In addition to driving immune exclusion, WNT signaling can act in both autocrine and paracrine manners to promote immunosuppressive TME dynamics. For example, WNT/β-catenin pathway activity compromises the immunogenicity of DC that do infiltrate the TME. Xenograft studies with human melanoma cell lines expressing a hyperactive β-catenin mutant have shown that the β-catenin/TCF complex directly drives expression of IL-10, which suppresses the immunostimulatory capacity of both tumor-infiltrating and splenic DC [167]. Tumor-derived WNT5A can also promote β-catenin-dependent metabolic reprogramming in DC to support other immunoregulatory, tolerogenic functions. Specifically, this pathway increases lipid uptake and fatty acid oxidation by DC, leading to IDO production and tryptophan metabolism within the TME. Elevated levels of the kynurenine byproduct of tryptophan catabolism in turn supports intratumoral Treg accumulation and activity [168, 169]. Tumor-intrinsic WNT signaling can also activate CXCL5-dependent MDSC recruitment to the TME, driving tumor-acquired resistance to ICB therapy. This phenomenon was recently reported by Theivanthiran et al., who demonstrated that CD8+ T cell activation following PD-1 blockade drives IFNγ- and PD-L1-dependent activation of the NLRP3 inflammasome, leading to HSP70 release and autocrine signaling through TLR4. This axis promoted expression of WTN5A, which in turn activated non-canonical, YAP-dependent CXCL5 expression, MDSC recruitment to the TME, and tumor progression [170].

Several strategies for targeting WNT signaling pathways have been explored in the context of combinatorial regimens with cancer immunotherapy. In an autochthonous melanoma model, WNT ligand antagonism synergized with PD-1 blockade to suppress both primary tumor growth and metastatic progression. In this same study, preliminary data from a Phase I study of the PORCN inhibitor ETC-159 also demonstrated that this WNT antagonist could remodel the tumor immune microenvironment of distinct solid tumors, suggesting that this intervention is likely to support a more favorable response to ICB therapy in patients as well [159]. Likewise, a pharmacologic inhibitor of β-catenin (iCRT14) that interferes with TCF complex formation and transcriptional activity augmented the efficacy of a tumor vaccine in a murine model of colorectal cancer [165]. Another β-catenin inhibitor, RX-5902, was similarly shown to potentiate ICB therapies directed against either CTLA4 or PD-1 in a mouse model of triple negative breast cancer [162]. Biomimetic nanoclusters that target β-catenin for degradation have been used as an alternative approach to enhance PD-1/PD-L1 blockade in experimental models as well [171]. Finally, in addition to these direct methods of WNT pathway inhibition, interventions to disrupt immunosuppressive networks that arise in WNT pathway-hyperactive tumors also have the potential to reshape the TME, creating T cell-inflamed conditions that are more conducive to immunotherapy [169].

Dysregulated PI3K/AKT/mTOR signaling and tumor immune resistance

First recognized as a pro-survival pathway in cancer cells, the PI3K/AKT/mTOR pathway is now appreciated as a driver of several hallmarks of cancer progression, including immune evasion. This pathway may become hyperactive in cancer cells as a result of excessive hormone and growth factor receptor signaling or by various genetic aberrations in its various pathway components, which have been found to occur at a high frequency across a wide range of cancer types. Indeed, one study of nearly 20,000 solid tumors reported that 30% exhibited loss of the PTEN tumor suppressor gene that negatively regulates this pathway, and another 6% were found to harbor PTEN mutations, including those known to confer loss of function. Activating mutations in the gene encoding the PIK3CA catalytic subunit of PI3K were also reported at a high frequency, as were mutations in AKT1, though these were less common [172]. Altogether, these aberrations mark the PI3K/AKT/mTOR pathway as one of the most significant oncogenic drivers of cancer progression.

Loss of the PTEN tumor suppressor gene was first linked with immune evasion when it was shown to enable PI3K/AKT-mediated induction of PD-L1 on glioma cells, conferring resistance to T cell-mediated lysis [173]. Shortly thereafter, experimental work in models of immune susceptible versus immune resistant tumor cell lines found that AKT activation could also confer resistance to T cell killing [174]. More recently, AKT and its upstream activator, PIK3CA, were shown to reduce tumor immunogenicity by downregulating expression of class I MHC and costimulatory molecules on cancer cells [175]. Much like tumor-intrinsic WNT/β-catenin pathway activation, in addition to these mechanisms of immune escape and suppression, PI3K/AKT hyperactivity in cancer cells also correlates with: (1) T cell exclusion from the TME [176, 177] and (2) an increase in the tumor-infiltrating immunosuppressive myeloid compartment [178].

As a result of these diverse immunologic consequences, genetic aberrations in PI3K/AKT pathway components, particularly PTEN, have emerged as significant correlates of immune therapy resistance, underscoring the relevance of this pathway to tumor immune evasion [158, 174, 176, 178–181]. Specifically, clinical analyses of triple negative breast cancer patients have shown that PTEN alterations correlate negatively with response to PD-1 or PD-L1 checkpoint inhibitors [181], and PTEN loss similarly correlates with poor response of melanoma patients to ICB therapy [180]. The lack of response to PD-1-targeting checkpoint inhibitors (nivolumab or pembrolizumab) observed in this latter study is likely explained by reduced CD8+ T cell infiltration into tumors exhibiting PTEN loss. Of note, although CD8+ TIL could still be recovered at low levels from PTEN-deficient tumors, TIL expansion from these tumors was also compromised, a finding that has important implications for ACT-based therapies as well. Biallelic PTEN loss has also been observed in longitudinal analyses of tumor biopsy specimens from patients whose cancers developed acquired resistance to either anti-PD-1 monotherapy [179] or anti-CTLA-4/anti-PD-1 combination therapy [158], further highlighting the role of this pathway in tumor immune escape.

Several strategies for manipulating PI3K/AKT pathway activity have been investigated for their potential to enhance the outcome of cancer immunotherapy. In murine models of Pten-mutant or Pten-null tumors, immune resistance can be overcome by administering PTEN-encoding mRNA-encased polymeric nanoparticles to restore PTEN activity in tumors [182]. In addition to promoting immunogenic tumor cell death, restoration of PTEN function in this way enhanced tumor infiltration by both T_H1 helper T cells and CD8+ T cells while decreasing Treg and MDSC infiltration. Importantly, this maneuver also synergized with PD-1 blockade to enhance tumor immune control. Alternatively, the availability of small molecule inhibitors targeting PI3K/AKT pathway components has also spearheaded efforts to pharmacologically disrupt this pathway in cancer cells. While several pathway inhibitors have been developed, the role of PI3K/AKT signaling in the activation and function of T lymphocytes and other immune cell populations is a complicating factor that must be considered when implementing such strategies as anti-cancer regimens. One way to limit deleterious effects of PI3K/AKT pathway inhibition in T lymphocytes while still achieving anti-tumor activity is to administer drugs targeting this pathway on an intermittent, rather than a daily, dosing schedule [183]. In this regard, ongoing use of the PI3Kα/β/δ inhibitor BAY1082439 was recently found to reduce the number of both CD4+ and CD8+ TIL in Pten-null prostate tumors, whereas intermittent drug cycling had no effect on the number of intratumoral CD4+ T cells and actually increased the number of CD8+ TIL. This drug also exhibited tumor cell-intrinsic effects associated with improved immunogenicity, including activation of type I and type II IFN pathway genes that promote antigen presentation (B2M) and T cell recruitment (CCL5 and CXCL10), and its intermittent use drove intratumoral expansion of CD8+ T cells as well. By creating a T cell-inflamed TME, this intervention also overcame primary resistance to PD-1 blockade [183]. Another attractive approach to avoid potential negative immunologic consequences of PI3K pathway inhibition is to selectively target specific PI3K isoforms that drive cancer progression, rather than administering pan-PI3K inhibitors. A dual-targeting PI3Kα/δ inhibitor targeting isoforms known to activate tumor-intrinsic signaling (PI3Kα) as well as tumor-extrinsic signaling in Tregs (PI3Kδ) has been shown to enhance the anti-tumor activity of CD8+ T cells [184]. Interestingly, a selective PI3Kδ inhibitor has also been shown to restrain CD8+ T cell differentiation ex vivo, improving engraftment and anti-tumor efficacy of adoptively transferred T cells [185]. These data suggest that inhibitors targeting this particular PI3K isoform might also prevent terminal differentiation and exhaustion of endogenous CD8+ T cells, further contributing to the anti-tumor and pro-immune effects of such regimens. Specific targeting of PI3K isoforms has indeed led to synergism with various forms of ICB therapy in murine models [180, 186], and promising results have been reported from an early phase clinical trial testing a PI3Kδ inhibitor in combination with pembrolizumab [187]. Finally, targeting downstream mediators of the PI3K pathway such as AKT and mTOR has also been shown to enhance the efficacy of immunotherapy in murine systems [174, 188, 189], highlighting the potential for these combinatorial strategies to improve therapeutic outcomes in clinical settings as well.

Cell cycle pathway dysregulation and tumor-intrinsic immune resistance

While loss of cell cycle regulation has long been recognized as a core tenet of the uncontrolled proliferation that is a hallmark of cancer progression, data have emerged in recent years demonstrating that cell cycle regulators can also confer tumor immune resistance. In particular, aberrations in several genes encoding cyclin and cyclin-dependent kinase family members have been linked with tumor resistance to immunotherapy. In lung adenocarcinoma, elevated expression of CDC25C is associated with reduced CD8+ T cell infiltration, increased Treg infiltration, and poor response to PD-1 checkpoint blockade [190]. Likewise, CDK4 gene amplification, CCND1 gene amplification, and copy number loss of the CDKN2A gene encoding a cyclin-dependent kinase inhibitor all correlate with primary resistance to PD-1 inhibition in patients with various melanoma subtypes [191]. Similar observations have been made for other solid tumor types harboring CDKN2A deletions or loss of function mutations as well [192, 193]. In this regard, it is worth noting that a recent gene set enrichment analysis of urothelial carcinomas harboring loss-of-function alterations in CDKN2A revealed that expression signatures for antigen processing, IFNγ signaling, and T cell receptor signaling pathways were all reduced in CDKN2A-altered tumors as compared with CDKN2A wild type tumors [192], effects that may at least be partially explained by concomitant loss of the chromosomally proximal JAK2 gene in cases of genomic deletions [194]. At the same time, however, experimental gene rescue of CDKN2A-null melanoma cells enhances tumor cell secretion of T cell-attracting chemokines [195], suggesting that direct influences of CDKN2A on the tumor immune microenvironment are also likely. Moreover, IFNγ pathway response genes are downregulated in tumors with CDK4 copy gain, an anomaly that is completely independent of a situation in which JAK deletion might potentially be coupled to loss of a nearby cell cycle-regulating gene [191]. Together with studies demonstrating that CDK inhibitors can: (1) convert immunologically “cold” tumors to T cell-inflamed tumors [196, 197] and (2) enhance CD8+ T cell effector function [198], these data provide convincing evidence that cell cycle regulators play active roles in immune evasion.

In light of these observations, several studies have investigated therapeutic strategies for targeting cell cycle regulators in the context of cancer immunotherapy. Importantly, the pleiotropic effects of CDK4/6 inhibition have been found to support anti-tumor immunity in a number of ways. First, by activating tumor-intrinsic expression of endogenous retroviral elements, CDK4/6 inhibitors promote dsRNA sensing, which in turn stimulates type III IFN production and tumor antigen presentation [196]. Second, by inducing metabolic stress in cancer cells, CDK4/6 inhibitors trigger expression of chemokines such as CCL5 and CXCL9/10/11 that recruit T lymphocytes to the TME [199]. These inhibitors also suppress Treg proliferation, an effect that correlates with reduced circulating and intratumoral Tregs in inhibitor-treated animals [196]. While CDK4/6 inhibitors also impact CD8+ T cell proliferation, this tumor-extrinsic action is offset by improved T cell activation, memory formation, and infiltration into tumor-bearing lung tissue [197, 199, 200]. Importantly, in these and related studies, the immune-potentiating effects of CDK4/6 inhibition have been shown to augment anti-tumor responses achieved by ACT/CAR-T therapy, OX-40/4-1BB agonist therapy, and ICB therapy [191, 196, 197, 199–202]. Similar results have also recently been reported in studies of ICB therapy in combination with CDK7 inhibitors [198, 203], a CDK12/13 inhibitor [204], and a broad-spectrum CDK1/2/5/9/12 inhibitor [205]. Collectively, these studies highlight the potential of targeting cell cycle regulators in the context of immunotherapy, providing strong rationale for the pursuit of such combination regimens in clinical settings.

Conflict of interest

The author has no relevant financial or non-financial interests to disclose.

Ethical approval

Not applicable, no approvals were necessary for the preparation of this review article.

Consent to participate

Not applicable, no research on human subjects was performed.

Consent to publish

Not applicable for this review article.