Toward a postbiotic era of microbiome science: Opportunities to advance immunotherapies for hepatocellular carcinoma

Introduction

Treatment strategies for hepatocellular carcinoma (HCC) have rapidly evolved over the past two decades. In tackling the considerable resistance of HCC to a broad range of cytotoxic chemotherapeutic drugs, much efforts have been devoted to the development and approval of multi-tyrosine kinase inhibitors with the goal of improving prognoses worldwide. Yet progressive HCC is prevalent and importantly, causally linked with dampened antitumor surveillance by the adaptive and innate immune system. Central to this pathophysiological mechanism is immune evasion by tumor-stromal cells that express co-inhibitory ligands, such as PDL1, to prevent activation of effector lymphocytes in the tumor microenvironment (TME).^{1, 2} Therefore, one can imagine that a potential HCC immunotherapy would effectively make use of monoclonal antibodies to block the binding of immune checkpoint ligands, more specifically PDL1, with PD1 receptor primarily expressed by tumor-infiltrating T cells to reverse their dysfunction–exhaustion phenotype.^{1, 2} Another active area of interests concerns the moderately low genomic mutational burden in HCC.³ Whether neoepitope presentation by HCC can be targeted by vaccine therapies to orthogonally influence such tumor-specific T cell responses has not yet been studied in detail.

Given that liver harbors one of the most diverse populations of antigen presenting cells in the human body and that HCC is non-sterile,⁴ it is plausible that anti-drug immunity develops and that cancer drugs bioaccumulate in intratumor bacteria,^{5, 6} facilitating therapeutic resistance. There are likely many other unmet challenges to overcoming resistance to and toxicities with first-line and second-line cancer therapies involving tyrosine kinase inhibitors, histone deacetylase inhibitors, immunologic agents, and anti-angiogenics for solid tumors,^7-10 let alone HCC. To gain novel insights into mechanisms underlying an individual's responsiveness to molecular cancer therapeutics for HCC, the scientific community has adopted the microbiota model of liver carcinogenesis.

Obesity-T2DM-NAFLD-HCC epidemic is a modern immune system-microbiota conundrum that represents an opportunity for immunotherapy research

Exploiting liver immunity for cancer precision therapeutics has garnered intensive clinical-basic research interests in light of data supporting the immunomodulatory roles of gut bacteria. This paradigm shift is exemplified by multiple arms of oncology trials targeting neoplastically advanced cancers of extrahepatic origins.^11-13 In these studies of immune checkpoint inhibitors (ICI), the intestinal bacterial community explicates measurable RECIST responses. Similar results have been observed in sorafenib-resistant advanced human HCC subset in which anti-PD-1 response is associated with differential gut bacterial commensal structure.¹⁴ While microbiome analysis has only begun to provide microbial insights into such treatment outcome, the field remains cautiously optimistic about the adaptations of suitable interventional preclinical HCC models, such as MUP-uPA and DIAMOND models of NASH-HCC, to better recapitulate multiple stages and molecular heterogeneity of human chronic liver diseases (CLD).¹⁵

Disentangling genetic and environmental contributors to anti-HCC immunity is a new interdisciplinary challenge posed by its extremely heterogenous background in humans. HCC immunobiology encompasses a broad spectrum of metabolic disorders associated with CLD. In CLD, a central dogma of host-bacterial metabolic maladaptation is the leaky gut model wherein a reduced bacteria-impermeable inner mucus layer has been postulated.¹⁶ Given the enormous challenge of calibrating immunological responses under steady-state and inflammatory conditions in the liver sinusoidal area, this model is instrumental to the study of how gut bacteria-derived enterohepatic secretome interweaves parenchymal and non-parenchymal cells in an immunoregulatory cross-talk that sets the tone for hepatocarcinogenesis as early as obesity.¹⁷

Gut bacterial correlates of human obesity are taxonomically indiscernible yet functionally distinct from their physiological counterparts.^{17, 18} In mouse models of defective microbial sensing mediated by T cell-specific Myd88 TLR adaptor signaling, dysregulated intestinal IgA response promotes dysbiosis of the members of Clostridia, resulting in the expansion of Desulfovibrio;^{19, 20} interestingly, these taxonomic shifts constitute a feature that mirrors CLD-linked microbiome and bacteremia.^21-23 While obesity is considered strategically treatable by various weight reduction measures, albeit, with a predisposition to recurrence; there is a consensus that obesity, if left untreated, induces glucose intolerance and chronic hyperglycemia, which precipitate T2DM and related risk factors for CLD. In classic obesity mouse models LepR^db and Lep^ob, the integrity of intestinal epithelia is compromised due in part to enhanced glucose transport by intestinal epithelial Glut2.²⁴ Such intestinal barrier dysfunction is an imminent risk for medically unrecognized enteric infections and bacterial invasion into portal veins, resulting in the systemic dissemination of microbial PRR ligands²⁴—potent activators of hepatic dendritic cells, which may prime a chronic pro-inflammatory T cell phenotype in the liver. Analogous to such long-lived T cell activation is the concept of gut bacterial memory-like signatures that appear to mediate mechanisms of rapid weight regain cycles in formerly obese individuals.²⁵ A healthy tolerogenic liver may process persistent influx of gut-derived immunometabolic signals in a homeostatic cytokine milieu. Such function, however, is likely perturbed in pre-CLD along the NAFLD-HCC sequence.

Therefore, thinking of HCC immunity from a TME-centric standpoint as in other models of human malignancies, such as colon cancer, undoubtedly underestimates the complexity of immune-escape mechanism in HCC as has been recently demonstrated.^{26, 27} An ideal immunotherapeutic approach for advanced HCC should consider a holistic treatment that not only resensitizes immune cells to clear cancers in conjunction with locoregional interventions²⁸ but also ameliorates CLD conditions in a synergistic long-lasting manner. Uncovering gut microbial enzymatic pathways that complement liver functions will strongly augment our understanding of CLD etiology and consequently what amplifies or mitigates responsiveness to HCC therapeutics.

Metabolic niches of the commensal microbiota in liver cancer: immunosuppressive or immunopermissive?

One of the most outstanding challenges in studying the microbiota in cancer is arguably the identification of immunomodulatory microbes and their anticancer bioactives. In the foreseeable years of oncology trials for ICIs administered either as monotherapy or in combination, disparate tumor-immune responses are to be expected. Data indicating microbiota contribution to anti-NAFLD-HCC therapy are still limited, and a consensus has yet to be reached whether multi-cohort-based microbiome signatures predict HCC in retrospect and prospect, leaving virtually no clear bacterial targets that can be prioritized for research.

Nonetheless, what's encouraging is there are now evidences where function, rather than structure, of the commensal-symbiont compartment of intestinal microbiota defines hepatic malignancy. In a demographic-controlled three-armed cohort, fecal microbiota filtrates from subjects diagnosed with NAFLD-HCC specifically promote and attenuate the expansion of CD3⁺CD4⁺CD25⁺Foxp3⁺ regulatory T cells (Tregs) and CD3⁺CD8⁺CCR7⁻CD45RO⁻ cytotoxic T cells, respectively, in an ex vivo human peripheral blood mononuclear cell differentiation culture.²⁹ In a mouse model of primary sclerosing cholangitis-cholangiocarcinoma induced by hydrodynamic injection of activated PKT/YAP and subsequent bile duct ligation, such immunosuppressive phenotype is supported by the accumulation of CD11b⁺Ly6G⁺Ly6C^loCXCR2⁺ polymorphonuclear myeloid derived suppressor cells in the liver via CXCL1/LPS/TLR4 hepatocyte-dependent mechanism.³⁰ This hepatic recruitment of polymorphonuclear myeloid derived suppressor cells is abrogated by neomycin treatment that targets gram-negative commensals.³⁰ Short-chain fatty acids (SCFA) are one of many classes of microbiota-derived messengers that enable such modulations of liver tumor immunity. Inulin, a soluble fermentable fiber, when supplemented with high-fat diet, promotes cholestatic HCC in TLR5-deficient mice.³¹ Selective depletion of gram-positive bacteria by vancomycin rescues this phenotype.³² Circulatory SCFAs, bacterial end-products of fiber fermentation, diminish subcutaneous syngeneic tumor regression by anti-CTLA4 blockade in immunocompetent MC38/CT26/MCA101_OVA models.³³ Further, SCFAs inhibit cross-priming activities of bone marrow-derived dendritic cells to restrict tumor infiltration by CD8⁺IFNγ⁺ T cells.³⁴ In human NAFLD-HCC, whether inhibitory Treg interaction with CD8⁺ T cells is due to increased fecal/systemic SCFAs and correlates with a poor prognosis is still a charged area of scientific debate as broad-spectrum antibiotic treatment for respiratory/urinary tract infections appears to reduce clinical efficacy of ICIs.^35-38

In health and pre-CLD, the universe of gut microbial metabolisms extends far beyond immunosuppressive mechanisms. Unlike in the context of cancer immunology, SCFAs are widely regarded as pleotropic and health-promoting. Butyrate promotes differentiation of activated CD8⁺ T cells into a KLRG-1^hiCD127^lo memory phenotype for long-term survival.³⁹ Monocyte differentiation into enhanced microbicidal intestinal macrophage phenotype is crucial for first-line defense mechanism, which is mediated by butyrate via HDAC3 inhibition, high LC3-II turnover and NOX2 activities downstream in activated macrophages.⁴⁰ In portal circulation, lithocholic acid, a pro-carcinogenic secondary bile acid—another class of immuno-active metabolite derived from primary bile acids by Clostridium cluster XIV—controls liver CXCL16 expression to regulate the accumulation of hepatic CXCR6⁺ NKT cells, which are implicated in anti-tumor immunity as well as autoimmune diseases.^{41, 42} In obesity, deoxycholic acid fosters a pre-HCC microenvironment by inducing senescence-like phenotype in hepatic stellate cells.¹⁷ In deoxycholic acid-induced senescent hepatic stellate cells, lipoteichoic acid—a major Gram-positive bacterial cell wall component—engages innate immune receptor TLR2 to co-instigate a COX2/PGE2/PTGER4-dependent pro-tumorigenic signaling.⁴³ By contrast, expansion of gram-negative Bacteroides commensals, such as Bacteroides thetaiotaomicron, Bacteroides fragilis, and Bacteroides rodentium, which are core members of the gut microbiota in humans and mice, specifically benefits anti-CTLA-4 therapy.^{13, 44} Not surprisingly, bacterial-ICI co-therapeutic effect converges from multiple lineages of microbiome species. In mesenteric lymph nodes, intestinal barrier translocation of Bifidobacterium pseudolongum-derived inosine, a bioenergetic source for CD8⁺ T cell metabolism and function in tumors,⁴⁵ enhances anti-CTLA-4 therapy via the induction of T-bet⁺IFN-γ⁺CD4⁺/CD8⁺ effector T cells in a similar Th1/Tc1 response for anti-PD1 blockade mediated by CCR9⁺CXCR3⁺CD4⁺ T cell recruitment when bi-colonized with Akkermansia muciniphila and Enterococcus hirae.^{46, 47} Importantly, a significant next step would be leveraging appropriate human HCC models to determine: (i) whether a cocktail of clinical human commensal Clostridiales/Bacteroides isolates that are underrepresented in tumors and known to elicit a natural CD8⁺ T cell-mediated anticancer immunity is useful for a range of ICI therapies^{48, 49} and (ii) how a defined immuno-active postbiotics mix derived from such commensal microbes, as a promising superior alternative to existing probiotics formulation,⁵⁰ performs in precision oncology trials for HCC.

Microbiota modulation of ICB for NAFLD-HCC: considerations for future adjuvant and neoadjuvant therapies

Mitigating risks associated with immune-related adverse events for cancer therapeutics, including but not limited to HCC, is of critical importance to maximizing long-term survivals. Given the background of NAFLD-NASH in metabolic syndrome, obesity, and T2DM as alluded above, there is increasing awareness for scientific consensus that these medical conditions inevitably complicate ICI therapies, leading to a new spectrum of hard-to-diagnose immune-related adverse events, such as autoimmune hepatitis as just one example.^{51, 52}

Microbiota-targeted therapeutics may address some of these emerging challenges. Metformin, a common anti-T2DM prescription drug, alters microbiome composition to promote the production of glycoursodeoxycholic acid, a bacterial secondary bile acid that inhibits intestinal FXR signaling to alleviate obesity and hyperglycemia.^53-57 A thermostable outer membrane protein from A. muciniphila antagonizes obesity via TLR2-mediated tight junction signaling by downregulating intestinal Cnr1 and reduces activities of pro-T2DM hepatic enzymes glutamyltransferase and aspartate aminotransferase.^{58, 59} A similar anti-diabetic response in a randomized clinical study of T2DM is supported by the dietary fiber-induced expansion of Bifidobacterium pseudocatenulatum, which exerts a postprandial glycemic control.⁶⁰ A diet rich in fibers may have a therapeutic value beyond CLD risk factor management; a more recent preclinical study demonstrates the promising use of such diet to promote the intestinal outgrowth of A. muciniphila, which produces cyclic di-adenosine monophosphate, a bacterial STING agonist that stimulates IFN-I signaling in intra-tumoral NK cells via upregulations of Xcl1 and Ccl5, enhancing anti-PD1/PD-L1 therapies for tumor models of various immune TMEs, such as MC38, BRAF^V600E/PTEN^−/− melanoma, and TUBO mammary carcinoma.⁶¹

More broadly, an individual's long-term metabolic health is shaped by gut–liver homeostasis. Recovery from environmental insults, such as colitis, is a mechanism driven by functional redundancy in the microbiome. A variety of pangenomes spanning across Actinobacteria, Bacteroidetes, and Firmicutes phyla harbor biosynthetic gene clusters that encode 3α/β-hydroxysteroid dehydrogenases and 5α/β-reductases to convert secondary bile acid 3-oxolithocholic acid to isoalloLCA,⁶² which is required for NR4A1-dependent differentiation of total Tregs,⁶³ and possibly, the induction of RORγ⁺Helios⁻ colonic Treg subset via other families of nuclear receptor bile acid sensors for immunotolerance and longevity.⁶⁴

Whether such microbial metabolites confer a unifying therapeutic benefit for NAFLD-HCC checkpoint immunotherapy warrants future investigations as commensal microbiome signatures are shared by CLDs and strain-resolved or genome-loci-resolved data underpinning such taxonomic alterations, not to mention microbiota metabolic shifts as a result of chronic use of antibiotics-like drugs for CLDs,⁶⁵ are yet lacking.