The dual roles of natural killer cells in liver immunity and tolerance: Implications for health and disease.

INTRODUCTION
The liver is a central organ in both metabolic and immune regulation, serving as a key site for immune tolerance and surveillance. It filters pathogens and antigens from the bloodstream while maintaining tolerance to self-antigens and benign environmental molecules. This delicate balance between immune tolerance and surveillance is maintained within a specialized hepatic microenvironment that promotes tolerance yet retains the ability to eliminate pathogens and malignant cells. Central to this process are natural killer (NK) cells. NK cells, as a major lymphocyte subset, acquire tissue-specific phenotypes such as CD49a+ in mice and CD56bright in humans, which enable dynamic interactions with the hepatic microenvironment. In addition to pathogen defense, hepatic NK (H-NK) cells also support tissue homeostasis by resolving fibrosis, promoting regeneration, and modulating the immune response. However, in chronic liver diseases such as viral hepatitis, fibrosis, and hepatocellular carcinoma (HCC), NK cell function is often impaired, marked by reduced cytotoxicity, dysregulated receptor expression (including altered “missing-self” recognition), and aberrant cytokine production. These dysfunctions arise from microenvironmental disturbances and actively drive disease progression. Elucidating the context-specific regulation of NK cells is crucial for advancing immunotherapy to restore NK cell function and improve patient outcomes.
Emerging therapeutic approaches, including chimeric antigen receptor-engineered NK (CAR-NK) cells and cytokine-based therapies, provide potential strategies to enhance NK cell function in liver disease. These interventions aim to reactivate or reprogram NK cells under chronic conditions, potentially offering new avenues for managing and treating liver diseases.
This review explores the dual roles of NK cells in liver health and disease, focusing on their origin, phenotypic diversity, receptor interactions, and functional specialization across different pathological states. Furthermore, we discuss the therapeutic potential of NK cells in liver disease, highlighting recent advances and future directions in NK cell-based immunotherapy.

INTRODUCTION OF THE HEPATIC SYSTEM
The liver is anatomically divided into right and left lobes and further segmented for functional specialization. It possesses a dual blood system: ~75% of the flow comes from the nutrient-rich portal vein, whereas the rest is provided by the oxygenated hepatic artery.1 Blood converges in the portal tracts and flows through low-pressure hepatic sinusoids, maximizing contact with hepatocytes and non-parenchymal cells and facilitating immune surveillance and detoxification. This immunological capacity is further supported by extensive lymphatic networks (portal, sublobular, and capsular) that drain into regional lymph nodes, with portal lymphatic vessels constituting the principal pathway, accounting for ~80% of total lymphatic drainage.2 Within this structure, portal tract-associated lymphoid tissue (PALT) recruits dendritic cells (DCs) to modulate local T cell responses and regulate hepatic immune homeostasis and tolerance.
The liver microenvironment consists of hepatocytes, which represent approximately two-thirds of all liver cells, together with a diverse population of non-parenchymal cells, including sinusoidal endothelial cells, Kupffer cells, hepatic stellate cells (HSCs), and intrahepatic lymphocytes. Human hepatic lymphocytes are enriched in innate-like subsets such as NK cells, NKT cells, and γδ T cells, which can comprise up to 65% of the lymphocyte compartment and expand during infection or tissue injury to provide early immune defense.3

The liver tumor microenvironment (TME) constitutes a complex niche where immunosuppressive populations, including myeloid-derived suppressor cells (MDSCs), tumor-associated macrophages (TAMs), and regulatory T cells (Tregs), dominate and promote disease progression.4 MDSCs inhibit lymphocyte cytotoxicity through arginase-1 activity and reactive oxygen species while secreting IL-10 and TGF-β to promote Treg expansion.5 TAMs exhibit plasticity between M1 and M2 states and cooperate with MDSCs and Tregs to remodel chemokine landscapes and antigen presentation.6 Tregs directly suppress CD8+ T cell effector functions. Together, these mechanisms drive immune evasion, accelerating HCC progression and resistance to therapy.

NK CELLS: ORIGIN AND PHENOTYPIC CHARACTERISTICS
NK cells primarily develop in the bone marrow but can also mature in extramedullary sites like the thymus, acquiring tissue-specific phenotypes.7 Notably, thymus-derived NK cells exhibit distinct features compared with their bone marrow-derived counterparts, including higher expression of GATA-3 and CD127 (IL-7Rα) and dependency on IL-7 signaling for their maturation.8 (Figure 1).
In humans, NK cells are primarily classified into 2 subsets based on CD56 and CD16 expression: CD56brightCD16− and CD56dimCD16+. The CD56bright subset, enriched in secondary lymphoid organs (eg, spleen, lymph nodes), is functionally immature, produces abundant cytokines, and expresses homing markers such as CCR7 and CD62L. Conversely, the CD56dim subset, comprising ~90% of peripheral blood NK cells, is fully mature, highly cytotoxic, and mediates antibody-dependent cellular cytotoxicity (ADCC) via CD16.9 The CD56dimCD16+ NK cell subset can be further delineated based on the surface expression levels of markers, such as CD94/NKG2A, killer immunoglobulin-like receptors (KIRs), CD57, and CD62L.10 NK cells can be classified into cytokine-producing and cytotoxic subsets, with their maturation defined by CD11b and CD27 expression. These stages include an immature CD11b−CD27− population, 2 cytokine-producing CD11b−CD27+ and CD11b+CD27+ populations, and a fully cytotoxic CD11b+CD27− population.11

Liver-resident CD56bright NK cells in humans express tissue-specific markers (CCR5, CXCR6, and CD69) and are more abundant than in peripheral blood, whereas CD56dim NK cells maintain cytotoxic capacity but lack residency markers.12 Traditionally, murine H-NK cells have been subdivided into liver-resident NK cell (lr-NK, CD49a+DX5−) and conventional NK cell (cNK, CD49a−DX5+) subsets based on surface marker expression.13 The CD49a+DX5− population shares morphological and functional features (eg, cytokine production, sinusoid anchoring) with the “pit cells” first identified in rats, suggesting roles in immune surveillance and regeneration.14 However, the classification of innate lymphocytes within the liver is an evolving field, and recent advances have necessitated more precise delineation of these populations.
In light of contemporary innate lymphoid cell (ILC) biology, the traditional dichotomy requires significant refinement. It is now established that the murine hepatic CD49a+DX5− compartment is largely composed of type 1 innate lymphoid cells (ILC1s), a lineage distinct from cNKs, despite sharing functions such as IFN-γ production and T-bet dependence.15 Their fundamental distinctions include development, transcription factor dependency, phenotypic markers, and core functions. NK cells and ILC1s arise from a common innate lymphoid progenitor but subsequently follow distinct developmental programs. NK cells progress through the NK precursor stage and require eomesodermin (Eomes) for maturation, whereas ILC1s originate from an innate lymphoid cell precursor and depend on T-bet rather than Eomes.16 These developmental differences parallel their temporal and spatial distribution, as ILC1s appear prenatally and remain tissue-resident, whereas NK cells emerge after birth and circulate.17

Phenotypically, while markers can be context-dependent, ILC1s characteristically express CD49a and TNF-related apoptosis-inducing ligand (TRAIL). In mice, CD200R expression and the absence of DX5 (CD49b) help distinguish them from cNKs (CD49a−DX5+).18 In humans, ILC1s typically express CD127 (IL-7Rα), whereas CD56dim NK cells do not.19 However, CD56bright NK cells might complicate this distinction. The functional dichotomy is paramount: cNKs are dedicated cytotoxic effectors that express high levels of perforin and granzymes for direct killing. ILC1s, however, possess low cytotoxic potential and function primarily as cytokine producers, acting as early sentinels and immune regulators within tissues.20

The hepatic landscape also includes hybrid or intermediate populations (eg, CD49a+CD49b+) that coexpress markers and transcription factors of both lineages, underscoring their functional plasticity.21 Critically, this plasticity is dynamically regulated by the microenvironment, with key factors such as TGF-β and IL-12 reprogramming NK cells toward an ILC1-like phenotype, characterized by downregulation of Eomes, upregulation of CD49a and T-bet, and loss of potent cytotoxicity.2223 This conversion, observed in settings such as Toxoplasma gondii infection and the TME, represents a mechanism of immune evasion where potent anti-tumor NK cells are functionally attenuated into less cytotoxic ILC1-like cells.22 Similar plasticity exists for other ILCs; ILC3s and ILC2s can transdifferentiate into ILC1s under cytokine cues such as IL-12 and IL-1β, although the reverse conversion from ILC1s to NK cells has not been reported.24

NK CELL RECEPTORS
NK cell receptors can be categorized into activating and inhibitory classes. Activating receptors identify stress-induced ligands, such as MIC-A/B and UL16-binding proteins (ULBPs) on abnormal cells, whereas inhibitory receptors recognize self-MHC class I molecules to prevent attacks on healthy cells. This dual–receptor system enables NK cells to selectively eliminate threats while maintaining self-tolerance (Figure 2).

Inhibitory NK cell receptors
Human NK cells express 2 main types of HLA class I-specific inhibitory receptors: the C-type lectin family receptor (CD94/NKG2A) and KIRs, including KIR2DL1/L2/L3 and KIR3DL1.25 Both receptor classes feature immunoreceptor tyrosine-based inhibitory motifs (ITIMs) in their cytoplasmic tails for signal transmission.
Inhibitory KIRs primarily interact with HLA-I variants (HLA-A, HLA-B, and HLA-C), which are universally expressed in healthy somatic cells. KIRs convey inhibitory signals via ITIMs by recruiting phosphatases such as SHP-1/2 to dampen NK cell activation.26 The presence of MHC-I is crucial for these interactions, and its downregulation in cancerous or infected cells diminishes inhibitory signaling, thus enhancing NK cell activation. Human KIR gene composition varies among individuals, with inhibitory KIRs (2DL1 and 3DL1) predominating to prevent autoreactivity.27

NKG2A, a C-type lectin receptor, non-covalently pairs with CD94 to form a functional CD94/NKG2A receptor on the cell surface.28 CD94 enhances stability and ligand binding. By recognizing HLA-E loaded with peptides from MHC-I leaders, it delivers an inhibitory signal, ensuring immune tolerance toward healthy cells.29 Reduced or absent NKG2A engagement lifts inhibition, enabling activation of signals that initiate an immune response against abnormal cells.
NK cell subsets exhibit distinct receptor expression: CD56bright NK cells mainly express CD94/NKG2A, whereas CD56dim NK cells predominantly express KIRs.

Activating NK cell receptors
The activating NK cell receptor repertoire encompasses the natural cytotoxicity receptor (NCR) family, C-type lectin family receptors, and KIRs.30 Specific ligands detected by these receptors include MHC class I chain–related proteins A/B (MICA/B) and ULBPs, which are overexpressed during cellular stress, infection, or transformation.
NCRs are type I transmembrane proteins that contain extracellular immunoglobulin-like (Ig-like) domains for ligand binding. This receptor family comprises 3 members: NKp30 (also known as NCR3 or CD337), NKp44 (also known as NCR2 or CD336), and NKp46 (also known as NCR1 or CD335).31 NKp46 and NKp30 are constitutively expressed and associated with FcεRI-γ or CD3ζ signaling chains, whereas NKp44 binds to DAP12 and is expressed only upon NK cell activation, such as by interleukin-2 (IL-2).32

The C-type lectin superfamily comprises over 1000 proteins, each containing at least one C-type lectin-like domain (CTLD).33 NK cells can distinguish between normal and transformed cells by interacting with MHC-I, thereby protecting healthy cells from damage. NKG2D, a pivotal member of the C-type lectin family, is expressed on nearly all NK cells and certain T cell subsets. NKG2D does not form heterodimers with CD94 but instead signals via the adaptor molecule DAP10, triggering PI3K/Grb2 signaling through its YxxM motif.34 It recognizes stress-inducible ligands (MICA/B and ULBPs) that are upregulated on infected and tumor cells. Other receptors, such as NKG2C, CD226, and CD244, also enhance NK cell responses to tumors and infections.
The KIR family comprises 7 inhibitory receptors (2DL1–2DL3, 2DL5, and 3DL1–3DL3), 6 activating receptors (2DS1–2DS5 and 3DS1), and the dual-function receptor 2DL4.35 Most KIRs feature either 2 (KIR2D) or 3 (KIR3D) extracellular Ig-like domains for ligand binding, and recognize polymorphic HLA-A, HLA-B, and HLA-C molecules. Unlike inhibitory KIRs, activating KIRs have short cytoplasmic tails lacking ITIMs but instead pair with the DAP12 adaptor protein via a charged residue (KIR2DS1, KIR3DS1).36

NKG2D ligands
NKG2D ligands are stress-induced surface antigens that are upregulated in response to cellular damage. These molecules are crucial for activating the immunoreceptor NKG2D, which allows NK cells to detect and target diseased or stressed cells. In humans, they are categorized into 2 structurally distinct families: ULBP1-6 and MICA/MICB.37

MICA (PERB11.1) and MICB (PERB11.2) encode stress-inducible ligands of the MHC class I-related family. Among the 7 MIC genes (MICA–MICG), only MICA and MICB produce functional transcripts, while MICC, MICD, MICE, and MICG are pseudogenes. Most MICA/MICB alleles encode proteins with structural domains similar to classical HLA-I chains, including 3 extracellular domains (α1, α2, and α3), a transmembrane region, and a cytoplasmic segment.38 Under physiological conditions, these proteins are minimally expressed in gastrointestinal epithelial cells, possibly because of their interactions with the gut microbiota. However, their expression significantly increases in response to stressors such as heat shock, DNA damage, and viral infection.39

ULBPs, also known as RAET1 proteins, are a family of NKG2D ligands widely expressed in response to cellular stress. These proteins contain MHC class I-like α1 and α2 domains but lack the α3 domain; they do not bind β2-microglobulin or present peptides. Four variants, ULBP1 through ULBP4, have been identified: ULBP1–3 are GPI-anchored, while ULBP4 has a transmembrane domain.37

Functional integration of key receptor axes in the hepatic microenvironment
The functional fate of liver-resident cells is decisively shaped by the dynamic interplay between activating and inhibitory signals from the hepatic microenvironment. Among the most pivotal regulatory mechanisms are the NKG2D-activating axis and the NKG2A–HLA-E inhibitory axis, whose ligands are dynamically expressed on key hepatic cell types, including hepatocytes and HSCs.
The NKG2D-activating axis is central to the immunosurveillance of stressed cells and the intrinsic control of fibrosis. Its function relies on robust induction of NKG2D ligands on both senescent HSCs and compromised hepatocytes. During liver injury, senescent HSCs markedly upregulate ligands such as MICA/B and ULBPs.40 These ligands bind to NKG2D on liver-resident NK cells and trigger cytotoxic granule exocytosis, leading to the elimination of senescent HSCs and limiting fibrosis progression.41 Similarly, hepatocytes undergoing viral infection, DNA damage, or malignant transformation increase the expression of these “danger signals,” enabling NK cells to clear dysfunctional or pre-neoplastic cells and prevent disease progression.42 However, the effectiveness of this protective pathway declines as the liver disease progresses. A fibrotic or pro-tumorigenic microenvironment, enriched with suppressive cytokines, leads to the downregulation of NKG2D on NK cells, thereby reducing their cytolytic activity.43 As this axis becomes compromised, activated HSCs and abnormal hepatocytes evade immune clearance, thereby promoting fibrosis and carcinogenesis.
Conversely, the NKG2A–HLA-E inhibitory axis functions as a central regulator of hepatic immune tolerance, a mechanism that is exploited in disease to promote immune evasion and dysfunction. Under physiological conditions, constitutive HLA-E expression on hepatocytes and Kupffer cells engages the abundantly expressed NKG2A receptor on liver-resident NK cells, maintaining these cells in a restrained activation state and contributing to the liver’s inherently tolerogenic character.44 This axis is markedly amplified in chronic liver disease and HCC. Disease progression is associated with increased HLA-E expression on both parenchymal and immune cells, along with enhanced NKG2A expression on NK and CD8+ T cells, forming a reinforced inhibitory circuit.45 This interaction directly suppresses lymphocyte cytotoxicity, compromising antiviral responses and anti-tumor immunity. In HCC, coexpression of NKG2A on tumor-infiltrating lymphocytes and HLA-E on malignant cells is characteristic of an exhausted immune phenotype and predicts poor clinical outcomes.46 Notably, within the TME, HLA-E–NKG2A binding not only protects TAMs from NK cell-mediated lysis but also promotes the production of immunosuppressive cytokines, such as IL-10 and TGF-β, by NK cells. Tumors actively exploit this pathway; for example, secretion of factors such as granulin-epithelin precursor (GEP) upregulates HLA-E and NKG2A while downregulating activating NKG2D ligands, shifting the balance toward inhibition.4748 The therapeutic relevance of this checkpoint has been highlighted by studies demonstrating that blocking the NKG2A–HLA-E interaction with agents such as monalizumab can restore lymphocyte effector function. These findings provide a compelling rationale for combination strategies designed to counteract the immunosuppressive milieu characteristic of advanced liver disease.49

NK CELL FUNCTIONS

General functions of NK cells
NK cells serve as crucial innate immune effectors that respond rapidly to virus-infected and transformed cells. Their key functional roles include direct cytotoxicity, cytokine production, and immune regulation. NK cells exert cytotoxicity through 2 major mechanisms: (1) perforin/granzyme exocytosis and (2) death receptor-induced apoptosis via Fas ligand (FasL) and TRAIL signaling. In the first mechanism, NK cells release perforin to generate lytic channels in susceptible cell membranes, facilitating granzyme entry and subsequent activation of caspase cascades to execute apoptosis.50 Alternatively, NK cells induce programmed cell death in malignancies through death ligand–receptor engagement, utilizing FasL, TRAIL, and membrane-bound TNF to activate cognate receptors on neoplastic cells.51 These complementary cytotoxic mechanisms enable NK cells to efficiently clear aberrant cells, while preserving their immune surveillance capabilities (Figure 3).
Beyond cytotoxic roles, NK cells are instrumental in immune regulation by releasing cytokines, such as IFN-γ and TNF-α, which enhance antigen presentation and shape adaptive immune responses. NK cells also produce chemokines, such as CCL3, CCL4, and CCL5, which promote immune cell migration and amplify inflammatory responses. NK cell activity is finely tuned by a balance between activating and inhibitory signals. Activating receptors, including NKG2D, NKp30, and CD16, initiate cytotoxic actions, whereas inhibitory receptors, such as NKG2A and KIR2DL1, prevent activation by recognizing self-MHC-I complexes.52

NK cells serve as immunoregulatory effectors that modulate diverse immune populations, including macrophages, B cells, T cells, and DCs. Upon activation by soluble mediators, such as type I interferons, IL-15, and IL-18, NK cells enhance DC/macrophage maturation and T cell activation through dual mechanisms involving receptor–ligand interactions and cytokine signaling. For instance, activated NK cells license DCs to secrete IL-12, thereby promoting Th1-polarized adaptive immune responses. Conversely, DCs regulate NK cell function via IL-15 trans-presentation through membrane-bound IL-15Rα, augmenting IFN-γ production and cytotoxic activity. This bidirectional crosstalk establishes a critical feedforward loop for immune coordination. Furthermore, the direct interactions of NK cells with DCs and T cells highlight their dual roles in bridging the innate and adaptive immunity.53

NK cells demonstrate tissue-specific migration governed by chemokine receptors, such as CCR2, CXCR3, and CX3CR1. CXCR3 and CX3CR1 are essential for directing NK cell movement in tumors. Tumor-derived chemokines, including CXCL10 (which binds to CXCR3) and CCL5, facilitate NK cell recruitment to the tumor microenvironment, bolstering anti-tumor immunity.54

Functions of hepatic NK cells
The liver hosts a substantial population of NK cells, accounting for 50% of intrahepatic lymphocytes.55 H-NK cells exhibit specialized phenotypic and functional characteristics that enable them to maintain immune homeostasis while effectively responding to infections and malignancies. Compared with their circulating counterparts, H-NK cells are predominantly CD56bright, a phenotype associated with lower cytotoxicity but higher cytokine production, which contributes to immune regulation.
H-NK cells are vital for liver immunity and exhibit functional characteristics distinct from peripheral blood NK cells. These liver-resident NK cells demonstrate heightened cytolytic activity, particularly against targets lacking MHC-I molecules, and secrete abundant IFN-γ upon stimulation. Research in rodents and humans has revealed that liver-resident NK cells exhibit elevated baseline activation and enhanced cytotoxicity, partly attributable to the upregulation of genes encoding molecules such as NKG2D and granzyme B.56 Notably, H-NK cells respond more readily to IL-2 activation, which in turn induces the expression of TRAIL on their surface. TRAIL expression is a key mediator of cytotoxicity toward HCC cell lines and activated HSCs.57 This heightened reactivity makes H-NK cells crucial sentinels for hepatotropic viruses and HCC. However, their activation is tightly regulated within the liver microenvironment to prevent collateral damage. The primary tolerance mechanism involves engagement of NK cell inhibitory receptors (eg, KIR2DL1, KIR2DL2/3) with self-HLA-I molecules on hepatocytes.58 Furthermore, under inflammatory conditions, cytokine-induced upregulation of MHC-I expression on hepatocytes protects them from NK cell attack.
H-NK cells are not only effectors of innate immunity but also key regulators that shape local immune responses through dynamic crosstalk with resident and infiltrating immune cells. The nature of these interactions differs significantly between homeostatic and pathological conditions, influencing liver disease progression and clinical outcomes.
In the healthy liver, H-NK cells contribute to immune tolerance and tissue homeostasis through bidirectional communication with other immune cell populations. Kupffer cells, the liver-resident macrophages, produce IL-18 in response to microbial products, which primes H-NK cells for IFN-γ production.59 In contrast, IL-10 derived from Kupffer cells suppresses NK cell activity and IFN-γ expression, thereby preventing excessive inflammation and supporting a tolerogenic environment that allows the handling of gut-derived antigens without inducing autoimmunity. Resting HSCs express high levels of MHC class I, which engage inhibitory receptors on H-NK cells and protect HSCs from NK-mediated killing, thereby preventing unwarranted fibrogenesis.60 H-NK cells also engage in reciprocal interactions with DCs; DC-derived IL-12, IL-15, and IL-18 enhance NK cell cytotoxicity and cytokine production, while NK cell-derived IFN-γ and chemokines such as XCL1 and CCL5 promote DC maturation and recruitment, forming a positive feedback loop that supports immune surveillance without causing inflammation.61

In pathological settings, H-NK cell crosstalk is profoundly altered, potentially limiting or promoting disease progression. In viral hepatitis, H-NK cells initially display increased cytotoxicity and IFN-γ production. However, during chronic infection, sustained IFN-γ release and receptor-mediated interactions may contribute to T cell exhaustion and impaired viral clearance.62 NK cells can kill virus-infected hepatocytes through TRAIL and NKG2D; however, similar pathways may inadvertently activate HSCs and promote early fibrogenic responses.63 During liver fibrosis and cirrhosis, activated HSCs downregulate MHC class I and become susceptible to NK cell-mediated cytotoxicity through activating receptors. NK-derived IFN-γ inhibits HSC proliferation and induces senescence, whereas TGF-β produced by macrophages and Tregs in the fibrotic microenvironment suppresses NK cell function, establishing a dynamic balance that can favor either fibrosis progression or its resolution.64

In HCC, the tumor microenvironment rewires H-NK cell interactions toward immunosuppression. TAMs secrete TGF-β and IL-10, which dampen NK cell cytotoxicity, whereas M1-like macrophages enhance NK cell activity through IL-12, IL-15, and IFN-β. However, this axis is often attenuated in advanced HCC.65 MDSCs directly inhibit NK cell function through cell–cell contact and soluble mediators, impairing NKp30-dependent recognition of tumor cells.66 Tregs further suppress NK cell activity via membrane-bound TGF-β and consumption of IL-2, weakening anti-tumor responses.67 NK and DC crosstalk is also compromised in HCC; elevated α-fetoprotein (AFP) or soluble MICB reduces IL-12 production and DC maturation, thereby limiting NK cell priming and adaptive immune activation.68 In liver metastasis, hepatic macrophages can promote NK cell recruitment via CXCL9 and activate them through STING-dependent IL-18 release, enhancing NK-mediated elimination of circulating tumor cells.69 Conversely, metastatic cells may condition macrophages to inhibit NK cell function through CD48/2B4 interactions, facilitating immune escape.70 Through these context-dependent interactions, H-NK cells serve as key modulators of the hepatic immune microenvironment. Their crosstalk with other immune cells plays a decisive role in determining whether liver immunity remains protective or becomes pathogenic.

NK CELLS IN LIVER DISEASES

NK cells in HBV
Hepatitis B virus (HBV), a DNA-based pathogen, targets the liver and evades host immune destruction. It can cause hepatic disorders ranging from acute illness to chronic infection, potentially leading to cirrhosis and HCC. Globally, over 296 million people suffer from chronic HBV infection, making it a major health concern.71 HBV spreads via blood, perinatal exposure, or sexual contact, with chronic cases prevalent in sub-Saharan Africa and East Asia. The virus primarily infects hepatocytes and establishes a persistent presence through covalently closed circular DNA, thereby evading immune clearance. Unlike many viruses that cause direct cytopathic effects, HBV-induced liver damage is mainly driven by an immune response (Figure 4).
During acute HBV infection, activated CD56dim NK cells in the peripheral blood undergo rapid expansion, exhibiting enhanced cytotoxicity through the release of perforin and granzymes as well as TRAIL-mediated apoptosis of infected hepatocytes.72 These cells produce high levels of IFN-γ, which suppresses viral replication and primes adaptive immunity by promoting CD8+ T cell responses. HBV-induced IFN-α/β further amplifies NK cell activity, upregulating activating receptors such as NKG2D and TRAIL while downregulating inhibitory receptors like NKG2A.73

Chronic HBV infection remodels the hepatic microenvironment, creating conditions that promote viral persistence mainly by inducing NK cell dysfunction. This dysfunction arises from multiple interconnected mechanisms. First, a dominant immunosuppressive cytokine milieu emerges, characterized by elevated TGF-β and IL-10 secreted by activated HSCs, Tregs, and a subset of CD56bright NK cells.74 These cytokines directly suppress NK cell effector functions by downregulating key activating receptors (NKp46 and NKG2D) and upregulating inhibitory receptors (NKG2A and KLRG1), driving a state of functional exhaustion.75 TRAIL-expressing NK cells accumulate in the liver during disease flares, where IFN-α and IL-8 act cooperatively to enhance TRAIL-mediated hepatocyte apoptosis and exacerbate liver inflammation. This pathogenic role is further exacerbated by NK cell-mediated depletion of HBV-specific CD8+ T cells through TRAIL-R2 interactions, which undermine adaptive immunity and sustain viral persistence.72

In addition to cytokine-mediated suppression, chronic antigen exposure promotes upregulation of alternative immune checkpoint molecules on NK cells. Single-cell analyses revealed that as HBV infection progresses to cirrhosis and HCC, intrahepatic NK cells exhibit elevated expression of inhibitory markers such as LAG3, along with signatures of functional exhaustion, including reduced cytokine production and impaired cytotoxicity.76 This upregulation of immune checkpoints, similar to T cell exhaustion, further constrains NK cell activity within the immunosuppressive tumor microenvironment.
Moreover, metabolic dysregulation contributes to NK cell dysfunction in HBV. Specifically, hepatitis B surface antigen (HBsAg) competitively binds to IL‑15Rβ/CD122 on NK cells, disrupting normal IL‑15 signaling and subsequently inhibiting the activation of the mechanistic target of rapamycin (mTOR) pathway.77 As a central regulator of cellular metabolism and immune activation, mTOR exhibits impaired activity that leads to significant disruption of glycolytic programming. This is evidenced by the downregulation of key molecules, including hypoxia-inducible factor-1α (HIF-1α) and glucose transporter 1 (GLUT1), diminished glycolytic capacity, and increased reactive oxygen species (ROS) production. This metabolic paralysis deprives NK cells of the bioenergetic and biosynthetic precursors necessary for effector functions, resulting in reduced cytokine production (eg, IFN-γ, TNF-α), decreased expression of cytotoxic granules (perforin and granzyme B), and impaired differentiation, as reflected by the downregulation of T-bet.
NK cell subtypes play distinct roles in HBV progression. Single-cell profiling has identified a unique intrahepatic tissue-resident NK population characterized by markers such as CXCR6 and ITGA5, reflecting a liver-restricted surveillance system associated with inflammatory signaling and local homeostasis.78 During the cirrhosis-to-HCC transition, there is an expansion of highly cytotoxic GNLY-expressing NK subsets, but these cells also display functional impairments, including reduced GZMB expression and elevated immune checkpoint signatures, which are consistent with NK cell exhaustion induced by TME.7980

In summary, the progression from acute to chronic HBV infection is characterized by profound alterations in NK cell biology. This shift, from an initially activating environment to one dominated by immunosuppressive factors, such as inhibitory cytokines, checkpoint upregulation, and metabolic stress, systematically undermines NK cell function. The resulting mechanistic cascade not only promotes viral persistence but also contributes to immunopathology, fibrosis, and carcinogenesis.

NK cells in HCV
Globally, ~58 million people are infected with the hepatitis C virus (HCV), an RNA-based pathogen that primarily targets the liver. Chronic HCV infection frequently leads to progressive hepatic damage, including fibrosis, cirrhosis, and HCC.81 Unlike HBV, HCV lacks an effective vaccine because of its high genetic variability and rapid mutation rate. Although direct-acting antiviral (DAA) therapy has achieved remarkable cure rates exceeding 95%, HCV infection remains a significant public health concern owing to the persistent risks of reinfection and immune-mediated liver injury.82 HCV is non-cytopathic, and liver damage results primarily from the host immune response rather than direct viral toxicity. Among immune components, NK cells contribute to both antiviral defense and immunopathology in chronic HCV infections.
NK cells form the frontline of defense against infected hepatocytes via cytotoxicity and cytokine release. During acute HCV infection, NK cells produce IFN-γ, which suppresses viral replication and boosts CD8+ T cell responses, whereas cytotoxicity is facilitated by TRAIL and perforin/granzyme pathways.83 Genetic analyses have revealed that the KIR2DL3–HLA-C1 genotype diminishes NK cell suppression, correlating with higher rates of spontaneous viral clearance.84 Moreover, activating receptors such as NKp30 and NKG2D recognize stress-induced ligands on infected liver cells, thereby enhancing NK cell-mediated immunity.85

In chronic HCV infection, a key mechanism involves the manipulation of the cytokine and receptor environment. HCV infection induces an immunosuppressive shift, suppressing pro-inflammatory cytokines such as IFN-γ and TNF-α while enhancing the production of inhibitory cytokines like IL-10 and TGF-β.86 This altered milieu directly contributes to the downregulation of activating receptors (eg, NKp30, NKp46) and upregulation of inhibitory receptors like NKG2A, impairing the recognition and elimination of infected cells.87 Furthermore, HCV infection induces NK cell exhaustion, characterized by sustained expression of multiple immune checkpoint molecules. Recent single-cell RNA sequencing (scRNA-seq) studies have revealed that NK cells in chronic HCV infection exhibit upregulated expression of exhaustion markers such as TIGIT and Tim-3.78 While Tim-3 expression has been paradoxically linked to enhanced cytotoxicity in some contexts, its coexpression with other checkpoints like TIGIT within an immunosuppressive environment signifies a dysfunctional or regulated state. Inhibitory signaling through KLRG1 further suppresses NK cell function.8889 Reduced intrahepatic TRAIL expression weakens NK cell-mediated cytotoxicity, thereby aiding viral persistence. This exhausted phenotype is mechanistically linked to the observed functional suppression in certain NK subsets, including the reduced expression of key effector cytokines (eg, IFNG, CCL3L1).78 The dysfunctional NK cell compartment in chronic HCV also participates in pathogenic crosstalk. Similar to HBV, TRAIL-expressing NK cells accumulate in the liver, where they promote hepatocyte apoptosis and inflammation while also impairing adaptive immunity by targeting HCV-specific CD8+ T cells.85

These alterations demonstrate that HCV-mediated NK cell dysfunction is not merely a failure of activation but rather a complex, actively regulated state of exhaustion and altered functionality, driven by viral manipulation of the host immune environment. These mechanistic insights highlight pathways involving inhibitory cytokines (eg, TGF-β, IL-10) and immune checkpoints (eg, TIGIT, Tim-3) as potential therapeutic targets for restoring NK cell immunity in HCV-related liver disease.

NK cells in liver fibrosis
Progressive liver fibrosis is characterized by abnormal accumulation of extracellular matrix (ECM) proteins, particularly collagen, resulting from persistent hepatic injury. At the core of fibrogenesis is the activation of HSCs, which transdifferentiate into myofibroblast-like cells that produce ECM.90 Key mediators, such as transforming growth factor-β1 (TGF-β1), platelet-derived growth factor (PDGF), and vascular endothelial growth factor (VEGF), amplify fibrogenic signaling, whereas innate immune pathways, particularly TLR activation in HSCs, further exacerbate ECM production.91 Advanced fibrosis progresses to cirrhosis, a precancerous condition characterized by parenchymal nodularity and impaired liver function. Despite its complexity, fibrosis remains a dynamic and potentially reversible process, emphasizing the need to understand its cellular and molecular regulators.
NK cells are potent effectors of immunological surveillance of liver fibrosis. They target activated HSCs and secrete anti-fibrotic cytokines. Activated HSCs upregulate stress ligands such as retinoic acid early inducible 1 (RAE1), which binds to NKG2D and triggers cytotoxicity via perforin/granzyme release or TRAIL/FasL-mediated apoptosis. NK cells produce key cytokine IFN-γ, which enhances NKG2D and TRAIL expression, amplifies HSCs killing, and suppresses TGF-β/Smad3 signaling, thereby attenuating fibrogenesis.92

Experimental models show that TLR3 activation via poly (I:C) or IFN-γ enhances NK cell cytotoxicity against HSCs, mitigating fibrosis in carbon tetrachloride (CCl4) and thioacetamide (TAA) models.93 Conversely, NK cell depletion increased collagen deposition, underscoring their protective role. In advanced fibrosis, NK cell function declines due to reduced NKG2D/TRAIL expression, oxidative stress (eg, ethanol-induced), and HSC resistance mechanisms such as RAE1 loss and MHC class I downregulation.60 NK cells also secrete IL-10 and IL-22, promoting HSC senescence via the STAT3/p53 pathways, making these cells more susceptible to clearance.94

In murine models of schistosomiasis-induced liver fibrosis, scRNA-seq revealed that H-NK cells could be categorized into mature, immature, memory-like, and regulatory-like subsets, which were further divided into ten transcriptionally distinct clusters (C0–C9).95 Notably, the Thy1+ NK subset (C4) showed enhanced cytotoxic activity against HSCs, highlighting its potential role in limiting early fibrogenesis. In HCV- and HBV-related fibrosis, NK cells display marked exhaustion, characterized by upregulation of inhibitory receptors (eg, TIGIT, LAG3, TIM3) and downregulation of effector molecules (eg, IFNG, GZMB, PRF1).78 Another study identified that the prostaglandin E2 receptor EP3 promotes NK cell adhesion to and killing of activated HSCs via the integrin α4 (Itga4)–VCAM1 axis, a mechanism substantiated by scRNA-seq showing enriched adhesion pathways in EP3-expressing NK cells.96 These findings underscore that NK cells comprise functionally diverse subsets whose abundance, transcriptional state, and cellular crosstalk are dynamically reshaped during fibrosis progression. This technology has also pinpointed potential therapeutic targets, such as Thy1, EP3, and chemokine pathways, offering novel insights into NK cell-based interventions in liver fibrosis.

NK cells in MASLD
Metabolic dysfunction–associated steatotic liver disease (MASLD) is a prevalent chronic liver disease worldwide. MASLD ranges from simple hepatic steatosis to more inflammatory and progressive metabolic dysfunction–associated steatohepatitis (MASH). The transition from benign steatosis to MASH is a critical juncture driven by the complex interplay between metabolic stress, hepatocyte injury, and immune activation.97 Within this immunological framework, NK cells have emerged as crucial modulators of disease onset and early progression.
In experimental models of MASH, H-NK cells are rapidly activated, displaying a pro-inflammatory cytokine profile with increased IFN-γ levels and enhanced expression of activating receptors such as NKG2D. This early activation positions NK cells as the instigators of inflammation. Mechanistically, activated NK cells promote hepatocyte damage by secreting pro-inflammatory cytokines, such as IFN-γ, IL-1β, and CCL5, which trigger the JAK-STAT and NF-κB signaling pathways in hepatocytes, leading to oxidative stress and apoptosis.9899 In addition, NK cell-derived IFN-γ can polarize macrophages toward a pro-inflammatory (M1) phenotype, thereby amplifying the hepatic inflammatory milieu, a hallmark of early MASH.100 ScRNA-seq studies revealed that cNK cells display a pronounced pro-inflammatory phenotype characterized by high expression of chemokines (eg, CCL3, CCL4, XCL1), upregulation of IL-12/IL-18 receptors, and enrichment of ADCC pathways.78 Intervention studies further support these findings, showing that genetic deficiency and NK cell depletion using antibodies significantly reduce steatohepatitis, liver injury, and associated inflammatory signaling in murine models.101

However, the role of NK cells is dichotomous, with compelling data highlighting a protective and anti-fibrotic role in the early phase of the disease. Activated NK cells can directly eliminate activated HSCs by integrating several activating cues, including upregulation of NKG2D ligands and reduced MHC class I expression on HSCs, together with the engagement of activating receptors, such as NKG2D and NKp46.100 By perforin–granzyme release and TRAIL-dependent apoptosis, as well as IFN-γ–mediated growth arrest and senescence, this cytotoxic program functions as a physiological brake on early fibrogenesis.93102 Some studies have shown that the frequency and broad functional capacity of cNKs remain largely intact in patients with simple steatosis and early MASH.103 However, phenotypic alterations are evident, such as the upregulation of the activating receptor NKG2D on cNKs in patients with MASH compared with those with simple steatosis.
ScRNA-seq studies provided further mechanistic insights, showing that in MASH livers, NK cells exhibit increased expression of the exhaustion marker LAG3. Cell–cell interaction analysis also revealed enhanced signaling between peripheral blood-derived NK cells and HSCs via the LAMA2/LAMB1-CD44 axis, an extracellular matrix-mediated interaction that may facilitate NK cell retention within fibrotic tissue and potentially accelerate fibrogenesis.104 In contrast, other studies indicate a reduced frequency of peripheral NK cells in MASLD, particularly within the cytotoxic CD56dim subset, accompanied by impaired cytokine production, degranulation, and a higher proportion of dysfunctional PD-1+ NK cells.105 In the context of obesity and metabolic dysfunction, NK cells can undergo “metabolic paralysis,” characterized by excessive lipid uptake, mitochondrial dysfunction, and impaired effector functions, which may compromise their anti-fibrotic and tumor-surveillance capacities.106

In summary, NK cells act as early responders to metabolic stress, contributing to the inflammatory drive that characterizes the transition from steatosis to steatohepatitis while simultaneously exerting protective and anti-fibrotic effects. The effect on MASLD progression is likely determined by the balance between these opposing functions, which are influenced by local tissue microenvironment, disease stage, and metabolism. Future research on liver-resident NK cells and their metabolic reprogramming will be crucial to fully understand their role and therapeutic potential in MASLD pathogenesis.

NK cells in autoimmune liver diseases
Autoimmune liver disease (ALD) encompasses 3 distinct clinicopathological entities: primary biliary cholangitis (PBC), primary sclerosing cholangitis (PSC), and autoimmune hepatitis (AIH). Each of these conditions has characteristic histopathological features, unique serological profiles, and organ-specific autoimmune responses, which influence their clinical progression.107 In PBC, autoreactive T cells and anti-mitochondrial antibodies cause damage to the intrahepatic biliary tree, whereas in PSC, bile duct inflammation and fibrosis result from autoimmune-mediated injury to both intrahepatic and extrahepatic bile ducts.108 Conversely, AIH is characterized by hepatocellular inflammation and the appearance of self-reactive antibodies that target the liver antigens.
Autoimmune liver diseases commonly exhibit a loss of immune tolerance, influenced by genetic factors such as HLA alleles and KIR polymorphisms, alongside environmental triggers that provoke abnormal immune responses. Although adaptive immunity primarily drives ALD pathogenesis, recent evidence highlights the complex role of innate immune cells, notably NK cells, in disease modulation. In ALD, NK cells mediate context-dependent pathogenic and protective roles by engaging in both the innate and adaptive immune pathways.
PBC is a chronic cholestatic liver disease characterized by progressive inflammatory damage to small intrahepatic bile ducts. In developed nations, its occurrence varies between 1 in 5000 and 1 in 10,000, with a predominance in females and a typical diagnosis at approximately 50 years of age.109 Although the exact pathogenesis of PBC remains unclear, immune dysregulation is widely regarded as the fundamental driver. In PBC, NK cells infiltrate the liver and exhibit elevated levels of cytotoxic mediators such as perforin and TRAIL. They directly lyse biliary epithelial cells (BECs), primarily through the perforin–granzyme pathway. Moreover, TRAIL/DR5 interaction may further contribute to BEC injury or apoptosis, thereby exacerbating cholestatic damage. This cytotoxicity is amplified by TLR3/4 activation and IFN-α production from monocytes, with chemokine axes such as CXCR1/CXCL8 promoting NK cell recruitment.110 Paradoxically, when NK cells infiltrate at low densities relative to BECs, NK-derived IFN-γ upregulates HLA on BECs, protecting them from NK-mediated killing while enhancing susceptibility to autoreactive T cells, highlighting the context-dependent functions of NK cells. ScRNA-seq analyses in PBC further revealed that intercellular communication between cholangiocytes, HSCs, and immune populations, including NK cells, is mediated by inflammatory mediators such as CCL21, CX3CL1, and IL34, which shape a distinct inflammatory microenvironment.111

PSC is a chronic cholestatic disorder characterized by widespread biliary inflammation, leading to fibrosis and distinctive saccular dilatations in both the intrahepatic and extrahepatic ducts. Disease advancement is characterized by progressive multifocal obliteration of the biliary system culminating in biliary cirrhosis, portal hypertension, and eventual hepatic failure. PSC predominantly affects men and is more severe in Black individuals than in White individuals.112 Although strongly associated with specific HLA alleles, the precise pathogenic mechanisms remain largely unknown. Recent studies have linked genetic variations in the NKG2D receptor to an increased incidence of cholangiocarcinoma among PSC patients.113 In addition, NK cells are present in the blood and colonic mucosa during the progression of PSC. Tissue-resident NK cells dominate the hepatic NK compartment in PSC. These cells express high levels of GZMK and ALOX5AP, and exhibit upregulated anti-inflammatory pathways, such as PTEN and PPAR signaling, which may paradoxically contribute to fibrogenesis.106 H-NK cells also show reduced cytolytic function, likely due to elevated local TNF-α levels. This impaired cytotoxicity may be offset by NK cells’ capacity to destroy cholangiocytes through TRAIL-dependent pathways. In patients with PSC, cholangiocytes exhibit elevated levels of TRAIL receptor 5.114 In addition, activating NK cells have been demonstrated to trigger TRAIL expression on H-NK cells, suggesting a potential mechanism for NK cell-induced collapse of cholangiocytes. CCR7+ NK cells infiltrate the liver via CCL21, and their presence correlates with the severity of liver fibrosis. Conversely, liver-resident NK cells in TGF-β receptor II-knockout mice appear to exert an inhibitory effect on the activation of CD4+ T cells, highlighting the complex and dual immunomodulatory functions of NK cells.115

AIH, the most prevalent autoimmune disorder of the liver, is characterized by immune-mediated hepatocellular necrosis and inflammation, contributing to progressive liver parenchymal damage and potentially high mortality. The global incidence is about 1.28 per 100,000 annually, with a higher incidence in females.116 AIH manifests as 2 distinct forms (type I or II), which can be further differentiated by characteristic circulating autoantibodies. Type I AIH is marked by anti-nuclear or anti-smooth muscle antibodies, whereas type II AIH is defined by anti-liver–kidney microsomal type-1 or anti-liver–cytosol type-1 autoantibodies.117 Cell–cell communication analyses revealed strong interactions between NK and CD8+ T cells, primarily mediated via the CCL5-CCR1/CCR5 chemokine axis, highlighting the role of NK cells in amplifying intrahepatic inflammatory networks.118 Furthermore, NK cells in AIH display elevated expression of antigen-processing and cytotoxic genes, including granzyme B.119 The transcription factor PRDM1 is upregulated in both peripheral blood NK and CD8+ T cells, suggesting a common regulatory mechanism modulating granzyme B and chemokine expression.118 In AIH, NK cell subsets exhibit divergent roles; circulating CD56dim NK cells decline peripherally but accumulate in the liver, display activated phenotypes, and contribute to hepatocyte injury via IL-17C–driven pathways. Conversely, liver-resident CD49a+ subsets may suppress CD4+ T cell proliferation, suggesting their immunoregulatory potential. Murine models of AIH have revealed that CXCR3+ NK cells exacerbate hepatitis, whereas cNKs mitigate damage via IFN-γ–dependent antiviral responses, emphasizing the subset-specific functions of NK cells in the pathogenesis of AIH.120

NK cells in hepatocellular carcinoma
HCC is the most prevalent primary hepatic malignancy, primarily associated with chronic HBV/HCV infection or MASLD.121 Its pathogenesis is characterized by sustained inflammation, hepatocyte necrosis, and cirrhosis, which together create a microenvironment that promotes malignant transformation. The prognosis for HCC remains poor owing to frequent late-stage diagnoses and inherent resistance to therapy, highlighting the critical need to understand the immune mechanisms underlying tumor control and escape.
TME in HCC drives NK cell dysfunction via a complex immunosuppressive network. The immunosuppressive TME impairs NK cells via TGF-β and IL-10, as well as their interactions with Tregs, MDSCs, and TAMs.122 These factors inhibit NK function through contact-dependent mechanisms, such as TGF-β signaling and metabolic interference via indoleamine 2,3-dioxygenase (IDO) and adenosine pathways.66123 Similarly, cancer-associated fibroblasts (CAFs) and other immune cells actively interact with tumor cells through secreted factors such as WNT5A and HGF, collectively reinforcing an immunosuppressive milieu that inhibits NK activity.124

A critical feature of the HCC TME is metabolic dysregulation and oxidative stress, which directly compromise NK cell function. The rapid proliferation of tumor cells creates a nutrient-deprived, hypoxic, and metabolically altered environment characterized by increased levels of ROS.125 Oxidative stress severely impairs NK cell cytotoxicity by disrupting membrane integrity, reducing degranulation capacity, and diminishing IFN-γ production.126 This oxidative damage forces NK cells into a metabolically compromised state, where their energy production and biosynthetic pathways are disrupted, preventing them from meeting the high metabolic demands required for sustained activation and tumor elimination. ScRNA-seq studies have identified a specialized population of tumor-associated NK (TaNK) cells in HCC. These TaNK cells, enriched in tumor tissue, express high levels of stress-related genes (eg, DNAJB1, HSPA1A) and exhibit low-cytotoxicity signatures.127 Their abundance is associated with worse prognosis and reduced response to immunotherapy, highlighting a key mechanism of immune escape.
The hostile TME drives both phenotypic alterations and functional exhaustion of NK cells. Initially, NK cells exhibit strong cytotoxicity by targeting dysplastic hepatocytes expressing stress markers such as MICA/B and CD155 through activating receptors (eg, NKG2D, NKp30), releasing TNF-α and IFN-γ to suppress tumor onset.128 However, as tumors advance, cytotoxic CD56dim NK cells become depleted at tumor sites, whereas dysfunctional CD11b−CD27− NK cells accumulate, resulting in reduced cytokine production and killing capacity.129 NK cells also experience functional exhaustion, as evidenced by their reduced intratumoral presence and weakened cytolytic function. This exhaustion is associated with downregulation of activating receptors due to ligand shedding (eg, soluble MICA via ADAM17), receptor internalization, and inhibitory signaling via the KIR–HLA axis.130131 Inhibitory KIRs, such as KIR2DL1, interact with self-HLA alleles like HLA-C2 to suppress NK activation. ScRNA-seq studies have detailed this exhaustion, revealing a general downregulation of cytotoxicity-related genes (eg, GZMB, PRF1) and significant upregulation of inhibitory receptors (eg, TIGIT, HAVCR2, LAG3) on tumor-infiltrating NK cells.127

Collectively, the complex immunosuppressive network within the HCC TME highlights why simply increasing the NK cell numbers is insufficient. This emphasizes the need for combinatorial strategies that disrupt the hostile TME (eg, through ROS scavengers or TGF-β blockade) and directly rejuvenate NK cell function (eg, via checkpoint inhibition or metabolic support). Thus, restoring effective anti-tumor immunity in HCC requires dismantling this integrated circuit of suppression, offering a strong rationale for novel immunotherapeutic combinations.

NK CELL THERAPY

CAR-NK therapy
Engineered CAR-NK cells represent a significant breakthrough in cancer immunotherapy, particularly for HCC.132 Unlike CAR-T cells, NK cells rarely trigger graft-versus-host disease (GVHD), facilitating the development of “off-the-shelf” allogeneic CAR-NK treatments.133 Moreover, their natural cytotoxicity and cytokine release profile lower the risk of serious immunotherapy-related complications, such as cytokine release syndrome and neurotoxicity, often associated with CAR-T therapy.

CAR-NK cells are designed to express synthetic receptors that combine tumor-specific antigen-binding domains (eg, single-chain variable fragments) with intracellular signaling motifs (eg, CD3ζ, 4-1BB) to enhance cytotoxicity.134 For instance, second-generation CARs, which incorporate costimulatory domains like CD28 or 4-1BB, enhance cell activation and persistence. However, third-generation designs with dual costimulatory signals have not demonstrated consistent superiority.135 In HCC, a central challenge in exploiting CAR-NK cell therapy is the identification of tumor-specific antigens to avoid on-target/off-tumor toxicity. Glypican-3 (GPC3), AFP, mucin-1 (MUC1), and CD147 (Basigin) have emerged as key targets, with several CAR constructs demonstrating potent cytolytic activity against HCC cell lines and xenografts. A key advancement is the use of dual-antigen logic-gated CAR systems (eg, synNotch-GPC3-inducible CD147-CAR), which improve tumor selectivity while sparing healthy tissues.136 (Figure 5).
Despite these advancements, CAR-NK cell therapies for solid tumors face significant challenges. The HCC TME, characterized by TGF-β, hypoxia, and metabolic stressors like lactate and adenosine, suppresses NK cell cytotoxicity through downregulation of activating receptors (eg, NKG2D) and recruitment of inhibitory cells (Tregs, MDSCs).137 Tumor antigen heterogeneity and low expression of tumor-specific markers further complicate targeting.138 Strategies under investigation include coexpressing chemokine receptors, such as CXCR3, to enhance tumor infiltration and integrating CAR-NK cells with checkpoint inhibitors.139 For instance, NK-92 cells engineered with PD-L1-specific CARs and IL-2 showed increased perforin/granzyme release, effectively lysing PD-L1+ tumors.140 In addition, bispecific CARs targeting dual antigens like BCMA and GPRC5D or logic-gated systems such as synNotch-inducible CARs address antigen heterogeneity and escape, which are essential in HCC’s genetically diverse landscape.141 Production challenges, such as variable efficiency in viral transduction and electroporation, necessitate optimized protocols for clinical-scale expansion.142

While most clinical trials currently target hematologic malignancies, early preclinical studies on HCC indicate that CAR-NK cells may provide a safer, more versatile, and potentially more accessible alternative to CAR-T cell therapies.143 Advances in CAR design, transduction techniques, and combinatorial strategies are crucial for maximizing the therapeutic efficacy of CAR-NK cells in HCC.

Other NK-based therapies
Beyond CAR-NK engineering, innovative therapies harnessing NK cell biology are being developed to treat HCC and other liver diseases. Cytokine-based treatments, notably IL-15, demonstrate a better safety profile and greater efficacy than IL-2, promoting NK cell proliferation and cytotoxicity without inducing Treg expansion or activation-induced cell death (AICD).144 Chemoimmunotherapy agents, such as sorafenib and bortezomib, enhance NK cell-mediated tumor destruction by increasing membrane-bound MICA/B and decreasing soluble MICA, thus improving ligand availability for NKG2D activation.145146 In addition, histone deacetylase inhibitors (HDACis), such as SAHA, enhance NK cell cytotoxicity by epigenetically upregulating MICA/B expression and repressing MICA-targeting miRNAs via chromatin remodeling.147148

Reversing NK cell exhaustion via immune checkpoint blockade is promising. Monoclonal antibodies against NKG2A, TIGIT, and TIM-3 have been shown to restore IFN-γ secretion and cytotoxic function in preclinical HCC models.149 For instance, circUHRF1 derived from HCC exosomes induces TIM-3 overexpression on NK cells by degrading miR-449c-5p, thereby impairing anti-tumor immunity and driving anti-PD-1 resistance. This highlights TIM-3 as a compelling therapeutic target.150

Innovative strategies for NK cell delivery and enhancement have emerged along with systemic modulation. Magnetic nanocomplexes (HAPF) enable magneto-activation and MRI-guided intra-arterial delivery of NK cells, significantly enhancing infiltration and anti-tumor efficacy in HCC xenograft models.151 Epigenetic targeting via PROTAC-mediated degradation of BPTF, a chromatin remodeling factor, increases natural cytotoxicity receptor ligands on tumor cells, boosting NK-mediated cytolysis.152 Non-genetically engineered platforms, such as “Pin” technology, equip ex vivo NK cells with bispecific antibodies through CD16a, thereby mimicking the antigen specificity of CARs without genomic modification and enabling simultaneous targeting of multiple tumor antigens.153 iPSC-derived NK cells provide scalable and standardized cellular products. Engineered to resist TGF-β–mediated immunosuppression and express CARs against HCC-associated targets like GPC3 or AFP, iPSC-NK cells show potent anti-HCC effects even in hypoxic or TGF-β–rich environments.154 In addition, NK cell engagers, such as bispecific AFM13 targeting CD16 or NKp46 and tumor antigens like CD30, display CAR-like functionality in both hematologic and solid malignancies, including preclinical HCC models.154

Strategies to enhance NK cell trafficking have been employed to circumvent the spatial and metabolic barriers of solid tumors.155 These include chemokine receptor engineering (eg, CXCR3, CXCR4) and localized delivery of cytokine-loaded fusion proteins or radiation-induced chemokine release to create robust intratumoral gradients.156 Recent studies have indicated that restoring NK cell membrane protrusions by targeting sphingomyelin metabolism synergizes with immune checkpoint inhibitors, enhancing tumor recognition and killing.157 These complementary approaches highlight the versatility of NK cell-based therapies beyond CAR modification. By integrating cellular engineering, immune modulation, and targeted delivery, NK cells are being redefined as precise immunotherapeutic tools with significant potential for the treatment of HCC and other liver malignancies.

CONCLUSIONS
The liver’s unique anatomical and immunological features make it a critical site for immune surveillance and tolerance. NK cells, abundant in the hepatic environment, are essential for maintaining homeostasis by balancing signals from stimulatory and inhibitory receptors. These cells eliminate infected or transformed hepatocytes and contribute to anti-fibrotic responses. However, their function is context-dependent, with the potential to drive immunopathology in autoimmune diseases or to facilitate tumor immune evasion in HCC. In HCC, NK cell dysfunction driven by the immunosuppressive TME has fueled the development of innovative immunotherapies. Strategies such as CAR-NK cells, cytokine priming, checkpoint inhibition, and advanced delivery systems aim to restore NK cell cytotoxicity. However, challenges, including TME-mediated suppression, antigen heterogeneity, and manufacturing scalability, must be addressed, necessitating a deeper mechanistic understanding and more precise therapeutic design.
To address these barriers and advance this field, future research should embrace cutting-edge technological frameworks. A key avenue is the integration of single-cell and spatial multi-omics.158 While scRNA-seq has revealed NK cell heterogeneity and exhausted states, it fails to preserve essential spatial information due to its dissociation-based protocol. Combining spatial transcriptomics with multiplexed imaging (eg, MERFISH, CODEX) is essential for mapping NK cells within specific tissue microenvironments, such as the periportal zone, fibrotic septa, or tumor-stroma boundary.159 This approach will help us understand how localized cues influence their functional states and interactions with hepatocytes, HSCs, and other immune cells. In addition, integrating scRNA-seq with snRNA-seq can mitigate dissociation bias and facilitate the analysis of archived clinical samples, enriching cohort diversity and supporting longitudinal studies.
However, several challenges remain unaddressed. First, technical limitations in transcript and protein detection sensitivity can obscure rare but functionally important NK cell subsets. Solutions involve advancing multi-omics techniques (eg, CITE-seq, ATAC-seq) and computational tools that integrate transcriptomic, epigenomic, and proteomic data from the same cell, linking regulatory programs to phenotypic outcomes. Second, functional validation of computationally identified subsets, especially dysfunctional populations like TaNK cells, remains challenging. High-resolution spatial profiling can prioritize candidate interactions for validation using ex vivo coculture systems or patient-derived organoids. Third, translating findings from atlas-level data into clinically actionable insights requires overcoming the challenge of analyzing complex and often low-quality clinical samples. Developing spatial and snRNA-seq technologies that are compatible with formalin-fixed paraffin-embedded (FFPE) tissues is crucial for bridging this translational gap.160

Harnessing the complete therapeutic potential of NK cells requires a dual-path strategy that provides deep mechanistic insights into innovative engineering. The integrated use of scRNA-seq, spatial transcriptomics, and multi-omics will systematically decode the molecular circuits that govern NK cell recruitment, education, and exhaustion in a pathological liver environment. These insights will help to identify novel targets for reversing NK cell dysfunction, such as specific metabolic or epigenetic checkpoints. These discoveries provide a critical basis for the next generation of immunotherapies, such as CAR-NK cells with enhanced tissue-homing abilities, innovative payloads that target local immunosuppressive mechanisms through cell–cell interactions, and scalable “off-the-shelf” products like iPSC-derived NK cells engineered to overcome the suppressive effects of the TME. By integrating fundamental research and applied therapeutic engineering, we can translate our understanding of hepatic NK cell biology into precise and effective therapies, thereby accelerating the “bench to bedside” translation for patients with liver diseases.