The CD94/NKG2A–HLA-E Axis as a Target in Cancer Immunotherapy: A Critical Perspective

Abstract

Successful clinical development of PD-1/PD-L1 and CTLA4 immune checkpoint blockers intensified the search for other potential targets for cancer immunotherapy. Among them, the CD94/NKG2A inhibitory receptor displayed by NK cell and T-cell subsets, which specifically interacts with the nonclassic HLA-E class I molecule, has attracted special attention. Here, we provide an overview of basic concepts on the CD94/NKG2A–HLA-E axis biology relevant in the framework of ongoing cancer immunotherapy approaches in different scenarios. First, the effectiveness of blocking the NKG2A–HLA-E interaction in vitro and in preclinical models as well as the presence of infiltrating NKG2A+ CD8⁺ T cells in some solid tumors has led to the generation of clinical-grade NKG2A-specific monoclonal antibodies, pioneered by monalizumab, currently tested in clinical trials. Second, controlling NKG2A expression by genetic engineering constitutes a promising approach to improve advanced adoptive NK cell–based immunotherapies. Challenges include identifying predictive biomarkers of responsiveness, selecting appropriate clinical settings, and optimizing combinatorial regimens.

Introduction

Cancer immunotherapy remains a long-standing ambition which has progressed alongside the knowledge on the immune system and tumor biology, fueled by technological advances. Positive results and disappointments have uncovered the complex challenges faced for promoting an immune response against tumors to effectively improve the clinical outcome, complementing other therapeutic procedures. Among outstanding advances, evidence that antagonizing the function of inhibitory immune receptors (i.e., PD-1/PD-L1 and CTLA4) promotes clinically successful T-cell responses against some tumors boosted the search for other potential target regulatory molecules, generically designated checkpoints.

Natural killer (NK) cells are innate immune effectors involved in the response to pathogens, originally identified by their ability to kill tumor cells. Universal accessibility to monoclonal antibody (mAb) production technology and its use as a strategy for the characterization of human leukocyte differentiation antigens, coordinated by international workshops, gradually led to the identification of a variety of surface human NK cell receptors (NKR), revealing the complexity of this leukocyte lineage. Among them, discovery of inhibitory NKR (iNKR) specific for HLA class I (HLA-I) molecules explained the molecular basis of a key regulatory mechanism, anticipated by the “missing self” hypothesis and supported by observations in murine models (1). That set of iNKR include (i) members of the killer cell immunoglobulin-like receptor (KIR) family, specific for groups of HLA-I allotypes with shared structural motifs; (ii) the CD94/NKG2A C-type lectin-like heterodimer specific for HLA-E, and (iii) the ILT2 (LIR1; LILRB1; CD85j) receptor broadly reactive with HLA-I molecules (2–4). These iNKR are expressed, individually or in combination, by discrete NK cell and T-cell subsets as well as by other leukocyte lineages in the case of ILT2 (i.e., B and myelomonocytic cells). Moreover, additional checkpoints specific for different molecules have been identified which also contribute to regulate NK cell functions (5).

According to the established paradigm, the NK cell response against a potential target requires that signaling through activating receptors (aNKR) overcomes the control threshold imposed by iNKR. This explains the reactivity of NK cell subsets bearing iNKR which do not recognize HLA-I molecules on allogeneic cells, a situation particularly relevant in haploidentical hematopoietic stem cell transplantation for leukemia therapy (6, 7). Evidence that mAbs antagonizing iNKR–HLA-I interactions promote NK cell activation revealed their potential as checkpoint inhibitors. Here, we overview basic concepts on the CD94/NKG2A–HLA-E axis biology relevant in the framework of ongoing approaches for cancer immunotherapy.

Identification of CD94/NKG2 NKR: Early Lessons from Deceptive Appearances

Production of mAbs as a strategy to identify human NKR originally led to the obtention of reagents, selected based on functional assays and specifically binding to NK and T lymphocyte subsets, which immunoprecipitated a disulfide-linked dimer, reduced to a ¹²⁵I-labeled 43-kD band, termed Kp43 (8). The study of Kp43-specific mAbs at the fifth Human Leukocyte Differentiation Antigen (HLDA) Workshop (Boston 1993) allowed CD94 definition. Remarkably, mAb engagement of CD94 inhibited NK clones which expressed the 43-kD molecule, also recognized by a different mAb (Z199), but activated others displaying a smaller protein (Mr 39-kD; ref. 9). Gene cloning showed that CD94 corresponds to a type II C-type lectin assembled as a homodimer, yet which did not fully explain the biochemical/functional features of the NKR as detected in NK cells (10). The observation that an antiserum against the CD94 protein reacted with another molecule, unnoticed in ¹²⁵I-labeled immunoprecipitates, revealed that the NKR was assembled as a heterodimer (11). Subsequently, the 43- and 39-kDa proteins were identified as C-type lectins encoded by NKG2A and NKG2C genes respectively (12, 13), located within the NK-cell gene Complex (NKC) on chromosome 12 together with CD94 and other killer cell lectin-like receptors (KLR) genes (14), allowing to complete the characterization of CD94/NKG2A and CD94/NKG2C NKR. Similarly to other immune receptors, inhibitory signaling is mediated by tyrosine phosphorylation of NKG2A, leading to SHP-1 tyrosine phosphatase recruitment, whereas phosphorylation of the NKG2C-associated DAP-12 adapter triggers a tyrosine kinase-dependent activation pathway (2–4).

The hypothetical specificity of NKG2A for HLA-I was explored assessing the response of NKG2A+ NK cells against the HLA-I–defective 721.221 (.221) cell line transfected with different HLA-I molecules. The ability of some of them, including HLA-G, to selectively inhibit NKG2A+ NK cells was for some time misinterpreted as indirect evidence for their specific interaction with the NKR, despite the absence of common structural motifs. The conundrum was solved thanks to the observation that .221 cells constitutively synthesize HLA-E, a nonclassic HLA-I molecule whose surface expression depends on transporter associated with antigen processing (TAP)-dependent presentation of conserved nonapeptides from the leader sequences of other HLA-I molecules containing a Met but not Thr in p2 (Fig. 1; refs. 15, 16), allowing to formally prove that HLA-E constitutes a unique ligand for CD94/NKG2 NKR (17–19).

Figure 1.

CD94/NKG2A specifically interacts with HLA-E bound to peptides from other HLA class I molecules. HLA-E preferentially binds 9-amino acid peptides (VL9) derived from the signal sequence of classical HLA-A, -B, and -C molecules and HLA-G, containing methionine, but not threonine, in position 2 (p-21), allowing its stable surface expression and recognition by the CD94/NKG2A inhibitory receptor expressed by NK cell and T-cell subsets. (Adapted from an image created in BioRender. López-Botet, M. [2025] https://BioRender.com/9gm64x0.)

CD94/NKG2 NKR and HLA-E: Genetics, Expression, and Function

CD94, NKG2A, and NKG2C are respectively encoded by KLRD1, KLRC1, and KLRC2 genes, respectively (14); their orthologues have been identified in primates and rodents and the corresponding NKR respectively recognize MHC-E and murine Qa1b HLA-E homologs, consistent with their evolutionarily conserved roles. NKG2E and NKG2H are alternatively spliced products of the KLRC3 gene, structurally similar to NKG2C, whose function remains ill-defined. By contrast with the diversity of KIRs specific for classic HLA-A, -B, and -C molecules, in terms of gene copy-number variation and sequence polymorphism (20), KLRs were initially considered nonpolymorphic. Yet individuals lacking NKG2C expression have been identified, showing complete gene deletion homozygous in ∼5% of individuals, ∼30% being carriers (21). Moreover, we recently reported coding substitutions in NKG2C with three allelic forms (NKG2C*01, 02, and *03), associated with single-nucleotide polymorphisms (SNP) of the promoter and 3′-untranslated regions in haplotypes (22). On the other hand, the NKG2A amino acid sequence is well conserved, but expression and function are influenced by noncoding SNPs of NKG2A and other NKC genes in linkage disequilibrium, being also modulated by certain HLA polymorphisms (23, 24).

HLA-E coupled to β2m and bound to HLA-I leader sequence-derived nonapeptides with Met in p2 (p-21 of the leader sequence) is widely expressed in cells from different tissues, yet at lower surface levels than classic HLA-I molecules. This class Ib molecule shows a reduced polymorphism, with two major allotypes differing at a single amino acid (Arg-107 in HLA-E*01:01; Gly-107 in HLA-E*01:03) not directly involved in peptide binding nor in contact with CD94/NKG2 NKR; yet differences in their surface levels have been reported. Structural and biochemical analysis of NKR interaction with HLA-E/peptide complexes revealed a substantial contribution of CD94 (25) and greater affinity of CD94/NKG2A than CD94/NKG2C, influenced by the bound peptide sequences (26).

HLA-E has been reported to exceptionally present some pathogen-derived and synthetic peptides, occasionally engaging CD94/NKG2 NKR (27). Of special relevance among them is the presence within the UL40 human cytomegalovirus (HCMV) protein of a sequence which copies the endogenous HLA-I leader nonapeptides but binds in a TAP-independent manner to HLA-E, thus allowing its surface expression by HCMV-infected cells (28). CD8⁺ T cells that specifically recognize HLA-E bound to an allogeneic HLA-I–derived nonapeptide are identified in some individuals, presumably primed in response to infection by HCMV strains displaying that UL40 sequence (29). Moreover, HLA-E has been proposed to present peptides from tumor neoantigens promoting T cell–specific responses (30), likely an infrequent event considering its preferential binding to HLA-I leader sequence nonamers.

In summary, CD94/NKG2 NKR engagement is influenced by the (i) HLA-E genotype of interacting cells; (ii) constitutive HLA-E expression levels, which may vary among different cell lineages as well as on exposure to stimuli enhancing HLA-I expression (e.g., IFNs); and (iii) individual HLA-I leader peptides which affect the interaction affinity.

CD94/NKG2A and CD94/NKG2C are generally displayed by different peripheral blood NK cell subsets, except for a small fraction coexpressing both. NKG2A is the only HLA-I–specific iNKR detected on the minor CD56^bright NK cell population, an early maturation stage lacking cytotoxic capacity but producing cytokines. In addition, NKG2A is displayed by variable proportions of the major CD56^dim NK cell population, either alone or in combination with KIRs. By contrast, NKG2C is mainly coexpressed together with iKIR specific for self HLA-I molecules by a subset of NK cells designated “adaptive” or “memory-like,” which undergo differentiation and persistent expansion in response to HCMV infection efficiently mediating antibody-dependent effector functions. This process, originally reported by our laboratory, has been shown to be driven by a TAP-independent expression in infected cells of HLA-E bound to the UL40-derived viral peptide, which inhibits NKG2A+ NK cells while stimulating the adaptive NKG2C+ NK cell response (28, 31, 32). The magnitude of the effect varies in HCMV-seropositive individuals and, among other factors, presence of NKG2C on both chromosomes and the NKG2C*02 allele are associated with greater proportions of circulating adaptive NK cells (22, 33).

Importantly, both CD94/NKG2 NKR are also independently displayed by minor subsets of circulating and tissue-resident TcRαβ+ CD8⁺ and TcRγδ+ lymphocytes (34, 35); among the latter, NKG2A is mainly expressed by the Vδ2+Vγ9+ subset (36). Of note, NKG2A may be coexpressed on NK and/or CD8+ T cells together with iKIRs and other inhibitory checkpoints (e.g., PD-1, ILT2, TIGIT, CEACAM-1, Siglec-9…) and has been reported to be inducible in response to cytokines (e.g., IL12 and TGF-β; ref. 37) potentially acting on tumor-infiltrating lymphocytes.

Targeting NKG2A in Cancer Immunotherapy

Detection of infiltrating CD8⁺ NKG2A+ T cells has been reported in different types of tumors (38–41), and HLA-E expression has been associated with poor clinical outcomes (42). IL12 was confirmed to induce NKG2A expression on tumor-reactive CD8⁺ T cells (43), and CRISPR-based screenings supported that IFN-mediated upregulation of MHC class I molecules, including HLA-E/Qa-1b, may facilitate immune evasion (44). In murine models with tumor-infiltrating CD8⁺ T cells not expressing PD-1, knocking out Qa-1b enhanced the effects of therapeutic vaccines (45). Interestingly, the study by Salomé and colleagues (41) in bladder tumors highlighted detection of NKG2A+ PD-1+ CD8⁺ T cells able to react against HLA class I–deficient tumors through T cell receptor–independent innate-like mechanisms, whose activity was restored by NKG2A blockade.

Although NK cell infiltration in solid tumors is relatively scarce, it has been associated with improved clinical evolution in response to antibody-dependent activation (46), and evidence supporting that NK cells may indirectly contribute to promote an antitumor T-cell response has been obtained (47). However, in the context of anti–PD-1 immunotherapy, the role of NK cell infiltration remains controversial, as it was associated with improved responses (48) but has been recently reported to impair the efficacy of immune checkpoint blockade (49). It is uncertain which variables may explain the discrepancy between these studies, and no precise information on the expression of NKG2A was provided.

Several observations support the rationale for blocking NKG2A as a strategy to unleash the lymphocyte activation potential against tumor cells. As overviewed in the next sections, antagonistic NKG2A-specific mAbs have been developed and tested as checkpoint inhibitors in preclinical models and clinical trials (Fig. 2A). Their therapeutic effects rely on effectively blocking the interaction of tumor HLA-E with NKG2A+ lymphocyte subsets in which the iNKR plays a relevant regulatory role, either dominant or complementary to other checkpoints. Thus, besides general challenges in cancer immunotherapy, specific constraints determined by the NKG2A–HLA-E axis biology deserve attention and the development of suitable biomarkers is warranted to design and interpret clinical trials. An important caveat is that coexpression by NK or T cells of other checkpoints may attenuate the effects of selectively targeting NKG2A (50). Moreover, the NKG2A interaction with HLA-E is modulated by different factors influencing the HLA-E surface expression levels, which need to be precisely discerned by standardized techniques.

Figure 2.

Strategies for targeting the CD94/NKG2A–HLA-E axis in cancer immunotherapy. A and B, Antibody-based approaches: blockade of the inhibitory CD94/NKG2A receptor on NK and T cells with anti-NKG2A mAbs, alone or in combination with anti–PD-1 therapy; bispecific antibody platforms simultaneously blocking NKG2A and PD-1. C, Cell engineering approaches: genetic disruption of KLRC1 (NKG2A) in tumor-associated antigen (TAA)-directed CAR NK/T cells; expression of an activating NKG2A/C switch receptor eliminating NKG2A-expressing lymphocytes; and generation of allogeneic CAR T-cells targeting TAA and NKG2A, combined with triple knockout for TRAC/B2m/NKG2A, to prevent graft-versus-host disease (GvHD) and immune rejection. (Adapted from an image created in BioRender. López-Botet, M. [2025] https://BioRender.com/2tc7ol0.)

Anti-NKG2A Blocking mAbs in Clinical Development

Based on the original obtention of NKG2A-specific mAbs by Dr. A. Moretta (13), Innate Pharma in collaboration with AstraZeneca developed monalizumab, the first clinical-grade humanized anti-NKG2A mAb and the most advanced in clinical development (Supplementary Table S1), which antagonizes the NKG2A–HLA-E interaction favoring NKG2A+ NK and CD8⁺ T-cell activation. The design of its isotype (IgG4) impairs C’ activation and especially binding to FcgR on other cells, preventing the potential agonistic inhibitory effects triggered by NKG2A engagement.

Initial trials combined monalizumab with tumor-targeted antibodies which, among other effects, may trigger NK cell antibody-dependent cell cytotoxicity (ADCC). In a phase II study, monalizumab together with the anti-EGFR antibody cetuximab showed an encouraging objective response rate in previously treated patients with recurrent/metastatic head and neck squamous cell carcinoma, showing a relatively low incidence of tolerable side-effects (51). However, the subsequent phase III INTERLINK-1 trial failed to meet its primary endpoint (52), and monalizumab plus cetuximab did not improve overall survival compared with placebo plus cetuximab. Similarly, monalizumab was tested in the phase II MIMOSA trial in combination with the anti-HER2 antibody trastuzumab, with no objective responses but an acceptable safety profile (53). These findings suggest that NK cell dysfunction in heavily pretreated patients and/or limited immune infiltration in metastatic lesions (54, 55) may constrain monalizumab clinical efficacy in advanced disease; whether it might enhance the action of tumor-targeted antibodies at earlier disease stages deserves attention.

Preclinical data indicate that concurrent blockade of PD-1/PD-L1 and CD94/NKG2A–HLA-E axis can synergistically promote innate and adaptive antitumor immunity, particularly in tumors with heterogeneous HLA-I expression (Fig. 2B; ref. 56). Moreover, combination therapies involving monalizumab and PD-1/PD-L1 blockers have shown acceptable safety and immune activation in patients with advanced solid tumors (57). In non-small cell lung cancer (NSCLC), the NeoCOAST trial demonstrated improved pathologic complete response rates with neoadjuvant durvalumab plus monalizumab, associated with enhanced tumor immune infiltration, interferon responses, and systemic functional immune cell activation (58). The results were confirmed in the NeoCOAST-2 trial, which included a triplet regimen of perioperative monalizumab, durvalumab, and chemotherapy (59). In addition, the COAST trial showed clinical benefit of durvalumab and monalizumab combination in unresectable stage III NSCLC (60), leading to the ongoing phase III PACIFIC-9 trial (61). Other early clinical trials are exploring combinations of monalizumab and PD-1/PD-L1 blockers in BCG-unresponsive bladder cancer or BCG-exposed and mismatch repair–deficient tumors (Supplementary Table S1).

Finally, monalizumab has been safely administered as maintenance therapy following allogeneic HSCT, with trials assessing its ability to potentiate the graft-versus-tumor effects in acute myeloid leukemia /(AML) or MDS of the predominant CD94/NKG2A+ NK cell subset arising early after transplant, mainly corresponding to early maturation stages with limited cytotoxic potential (62). Yet, this approach is supported by preclinical studies in relapsed patients with AML showing that functional exhaustion of NKG2A+ NK cells, correlating with disease progression, can be reversed via NKG2A blockade or gene editing (63–65).

A Next Generation of NKG2A-Blocking Molecules

As summarized in Supplementary Table S1, a wave of anti-NKG2A agents is advancing into clinical development. S095029 is a human IgG1, engineered to prevent FcgR-binding, with enhanced affinity for NKG2A. Preclinical studies have shown superior NK and Tgd+ cell activation, both as monotherapy as well as in combination with tumor-targeted antibodies and/or PD-1–blocking agents (66). Phase I/II clinical trials are ongoing in patients with colorectal cancer, gastric cancer, and NSCLC (NCT05162755, NCT06116136, and NCT06162572; Supplementary Table S1). BRY805 is a human IgG4 which showed enhanced tumor growth control compared with monalizumab in a mouse xenograft model being currently evaluated in a phase I trial in advanced solid tumors (NCT06289894). Similarly, the HY-0102, a humanized IgG developed by Shanghai HyaMab Biotech, is undergoing clinical evaluation in the United States and China (NCT0491435 and NCT06094777). Interim results for the US dose-escalation cohort showed good tolerability and early signs of clinical activity. Finally, XB628 is a bispecific antibody targeting PD-L1 and NKG2A, which induced the colocalization of PD-L1+ tumor cells and NK cells promoting a greater NK cell–mediated tumor cell cytotoxicity than anti–PD-1 or anti-NKG2A antibodies alone or in combination (NCT06952010; ref. 67).

NKG2A Modulation in Adoptive Cellular Therapies

NKG2A modulation represents a promising approach to improve adoptive cell therapies (Fig. 2C) Preclinical studies show that CRISPR-mediated deletion or antibody blockade of NKG2A augments the antitumor activity of adoptively transferred NK or T cells, especially in high HLA-E–expressing tumors such as multiple myeloma (68). Indeed, simultaneous NKG2A/Qa1b blockade and intratumor delivery of activated NK cells in MC38 subcutaneous tumors significantly delayed tumor growth. This effect was partially attributed to CD8⁺ T-cell and cDC1 responses, which were further enhanced when combined with systemic anti–PD-1 therapy, resulting in partial abscopal efficacy on noninjected distant lesions (69).

Preclinical work has also validated the use of genetically engineered CAR-NK cells lacking NKG2A in AML and melanoma models (70, 71). Similarly, NKG2A-knockout primary (72) or iPSC-derived NK cells (73) exhibited superior activity against several solid and hematologic xenografts models and glioblastoma, respectively. Further innovations include engineering a chimeric CD94/NKG2A/C switch receptor enabling HLA-E recognition by the NKG2A ectodomain and DAP-12 activation through the NKG2C intracellular domain, whose expression in NK and T cells enhanced their response to HLA-E–overexpressing tumors (74). A caveat on this approach is the potential off-target toxicity against normal cells expressing HLA-E yet not perceived under the experimental conditions used.

The first-in-human application of this concept is underway with CT0590/CT0596 developed by CARSGEN, BCMA-targeted allogeneic CAR T-cell products targeting BCMA and NKG2A in combination with triple knockout for TRAC/B2m/NKG2A which allows them to avoid graft-versus-host disease and rejection, making them fratricide-resistant while enhancing their immune function. Early data from patients with relapsed/refractory multiple myeloma have demonstrated safety and preliminary efficacy (NCT06718270 and NCT06988059; ref. 75).

Conclusions and Future Directions

The blockade/inactivation of NKG2A represents an expanding area of cancer immunotherapy, with applications extending from mAb-based therapies of tumors to sophisticated genetic engineering of cellular therapies. Challenges include identifying predictive biomarkers of responsiveness, selecting appropriate clinical settings, and optimizing combinatorial regimens. Translational research and rigorously designed clinical trials will be critical to more precisely assess the potential of harnessing this piece of the immuno-oncology armory.

Supplementary Material

Supplementary Table S1

Summary of selected clinical trials assessing the safety/clinical efficacy of targeting NKG2A

Authors’ Disclosures

C. Vilches reports grants from Agencia Estatal de Investigación (AEI/FEDER, EU) during the conduct of the study. A. Muntasell reports grants from Instituto de Salud Carlos III/FEDER, Ministerio de Ciencia, Innovación y Universidades, and CIBERONC during the conduct of the study. No disclosures were reported by the other author.