The Major Histocompatibility Complex (MHC) gene region encodes proteins that are foundational to immune surveillance and self/non-self discrimination. These proteins are broadly categorized into classical (MHC-Ia) and non-classical (MHC-Ib) molecules, which possess fundamentally divergent functions (1).

Classical MHC-I molecules, namely HLA-A, HLA-B, and HLA-C in humans, are defined by their high degree of polymorphism and their ubiquitous expression across almost all nucleated cells. Their primary and canonical function is to present a vast repertoire of intracellularly derived antigenic peptides to the T-cell receptors (TCRs) of CD8+ cytotoxic T-lymphocytes (CTLs) (2). This interaction is the principal mechanism by which the adaptive immune system recognizes and eliminates virus-infected or malignantly transformed cells.

In contrast, the non-classical MHC-Ib molecules, which include HLA-G, HLA-E, and HLA-F, represent a more functionally diverse and specialized group (3). They are characterized by limited polymorphism, restricted and often inducible tissue expression, and a broader range of immunoregulatory roles that extend beyond conventional antigen presentation (4). A key functional divergence is that while MHC-Ia molecules are the primary ligands for T-cell activation, MHC-Ib molecules like HLA-G serve to mediate inhibitory or, in some cases, activating stimuli, particularly for Natural Killer (NK) cells of the innate immune system (5). HLA-G, the focus of this review, epitomizes this immunoregulatory role, acting as a potent ligand for inhibitory receptors that bridge both innate and adaptive immunity (6).

HLA-G possesses several unique molecular features that distinguish it from its classical counterparts and define its specialized function. Firstly, the HLA-G gene exhibits remarkably low polymorphism in its coding region (7). This evolutionary conservation is in stark contrast to the hypervariable nature of classical HLA genes and is critical for HLA-G to function as a “universal” inhibitory ligand, capable of engaging its receptors on the immune cells of genetically diverse individuals.

Secondly, the HLA-G primary transcript undergoes complex alternative splicing to generate at least seven distinct protein isoforms, a level of diversity not seen in classical HLA molecules (8). These isoforms are structurally and functionally heterogeneous, broadly classified into four membrane-bound (mHLA-G) isoforms (HLA-G1, -G2, -G3, and -G4) and three soluble (sHLA-G) isoforms (HLA-G5, -G6, and -G7) (9). The full-length HLA-G1 protein and its soluble counterpart, HLA-G5, are the most extensively studied (10).

Thirdly, and most central to its function, HLA-G exerts its profound immunomodulatory effects by serving as a ligand for a specific set of inhibitory receptors expressed on immune effector cells (11). The most critical of these are the Immunoglobulin-Like Transcript (ILT) receptors, now standardized as Leukocyte Immunoglobulin-Like Receptors (LILRs) (11). The most critical of these are: LILRB1 (ILT2/CD85j): This receptor is broadly expressed on B cells, monocytes, dendritic cells (DCs), and distinct subsets of T cells and NK cells (8). It recognizes both classical MHC-I molecules and HLA-G, though it binds HLA-G with significantly higher affinity (12, 13). LILRB2 (ILT4/CD85d): The expression of this receptor is more restricted to the myeloid lineage, including monocytes, macrophages, and dendritic cells (8). Unlike LILRB1, LILRB2 demonstrates a preferential binding to HLA-G over classical MHC-I molecules (12).

The signaling axis formed by the interaction of HLA-G with LILRB1 and LILRB2 is the central molecular pathway for HLA-G-mediated immune suppression and its function as a non-classical immune checkpoint (14).

The highly restricted expression pattern of HLA-G under normal physiological conditions—confined to sites like the placenta, thymus, and cornea—points to its dedicated role as a guardian of immune privilege (15). In these “sanctuary” tissues, its function is not only beneficial but essential for maintaining immune tolerance and preventing autoimmune or allo-immune-mediated pathology (16).

However, this potent tolerogenic function possesses a pathological “dark side.” HLA-G is frequently neo-expressed or aberrantly upregulated in a wide array of pathological conditions, most notably in solid and hematological malignancies and during chronic viral infections (17). In these disease states, HLA-G expression is no longer a protective physiological mechanism but is instead hijacked as a sophisticated tool for immune evasion (18). This aberrant expression allows tumor cells and virus-infected cells to “disarm” the host immune response by engaging inhibitory receptors, leading directly to disease progression, metastasis, and poor clinical outcomes (19).

This review is structured around this central thesis: HLA-G is a “dual-role” molecule, acting as a molecular master switch that can be physiologically engaged to promote tolerance or pathologically exploited by disease to drive evasion (Figure 1).

In accordance with the central thesis, this review will first dissect the fundamental molecular biology of HLA-G, including its complex isoform structure and the regulatory networks that govern its expression (11). It will then explore the two faces of its dual function: its essential role in establishing and maintaining physiological tolerance in pregnancy, transplantation, and immune-privileged sites, followed by its detrimental role in pathological immune evasion in cancer and chronic infections. Finally, we will survey the rapidly advancing translational frontiers, critically evaluating HLA-G as a diagnostic biomarker and as a next-generation therapeutic target, before concluding with a discussion of the major challenges and future perspectives for the field.

The functional versatility of HLA-G is encoded in its molecular architecture, which is defined by the seven distinct protein isoforms generated via alternative splicing of the HLA-G primary transcript (20) (Figure 2).

Membrane-bound (mHLA-G) Isoforms (Table 1):

HLA-G1: This is the canonical, full-length MHC-I protein, structurally homologous to classical HLA molecules. It consists of the α1, α2, and α3 extracellular domains, a transmembrane domain (encoded by exon 5), and a cytoplasmic tail (encoded by exons 6-8). It non-covalently associates with β2-microglobulin β2m (9). HLA-G2: A truncated isoform generated by the splicing (skipping) of exon 3, which encodes the α2 domain (20). HLA-G3: A further truncated isoform lacking both the α2 and α3 domains, resulting from the splicing of both exon 3 and exon 4 (20). HLA-G4: An isoform that retains the α1 and α2 domains but lacks the α3 domain due to the skipping of exon 4 (20).

Soluble (sHLA-G) Isoforms:

HLA-G5: This is the primary soluble isoform and is structurally analogous to HLA-G1. It is generated when intron 4 is retained in the mature mRNA. This intron contains a premature stop codon, which prevents the translation of the downstream transmembrane domain (23). This process results in a protein containing the full α1-α3 extracellular domain associated with β2m, but with a unique 21-amino-acid C-terminal tail that confers solubility (26).

HLA-G6: The soluble counterpart to HLA-G2, this isoform lacks the α2 domain and is also rendered soluble by the retention of intron 4 (22). HLA-G7: A highly truncated soluble isoform consisting only of the α1 domain, resulting from the retention of intron 2 (27).

These structural variations dictate receptor engagement. Not all isoforms engage receptors equally. For instance, dimeric forms of HLA-G (linked via Cys42-Cys42 disulfide bonds) exhibit a significantly higher affinity for LILRB1 and LILRB2 compared to their monomeric counterparts (28, 29). Specifically, the dimerized HLA-G molecule presents an accessible binding site that favors LILRB1 interaction, effectively increasing the avidity of the immune checkpoint signal (28). Furthermore, truncated isoforms lacking the α3 domain (such as HLA-G4) may fail to bind LILRB2 efficiently, as the α3 domain is a critical contact site for this receptor (30). Thus, the specific isoform profile of a tissue directly shapes its immunomodulatory potential.

The restricted, “on-demand” expression of HLA-G is maintained by a multi-layered and highly sophisticated regulatory network. While its coding region is conserved, the HLA-G gene’s regulatory regions—specifically the 5’ upstream regulatory region (URR) and the 3’ untranslated region (UTR)—are, in contrast, highly polymorphic. This suggests that evolutionary selective pressure has favored the ability to modulate expression levels of HLA-G rather than alter its core inhibitory function (7, 31, 32).

Transcriptional and Epigenetic Control: In most normal, healthy tissues, the HLA-G gene is transcriptionally silent. This silencing is often enforced by epigenetic mechanisms, primarily DNA hypermethylation at CpG islands within the promoter region (33). In pathological contexts, such as cancer, the de novo expression of HLA-G is frequently linked to the loss of this silencing, often through promoter demethylation. Furthermore, polymorphisms within the 5’ URR can alter the binding affinity of key transcription factors, thereby dictating the basal and inducible levels of HLA-G transcription (20, 34).

Post-Transcriptional (miRNA) Control: The 3’ UTR of the HLA-G mRNA is a critical hub for post-transcriptional regulation, acting as a target for non-coding microRNAs (miRNAs) (35).

miR-148a and miR-152: A significant body of evidence has identified miR-148a and miR-152 as key negative regulators of HLA-G. These miRNAs bind to specific sites in the HLA-G 3’ UTR, triggering mRNA degradation and/or translational repression, thereby silencing HLA-G protein expression (36).

Tissue-Specific Regulation: This mechanism is fundamental to HLA-G’s tissue-specific expression. The physiologically high expression of HLA-G in the placenta, for instance, is not only due to transcriptional activation but also to the fact that the placenta lacks high-level expression of miR-148a and miR-152. This low-miRNA environment permits the HLA-G mRNA to remain stable and be efficiently translated, which is essential for materno-fetal tolerance (32).

Impact of Regulatory Polymorphisms: Genetic variations within these regulatory regions directly impact the level of HLA-G protein expression, linking an individual’s genotype to their immune-regulatory potential. The most extensively studied example is the 14-base pair (bp) insertion/deletion polymorphism (rs371194629) located in the 3’ UTR (31, 37). The presence of the 14-bp insertion allele has been functionally linked to lower mRNA stability and, consequently, reduced HLA-G protein production. This is thought to occur because the insertion sequence alters the secondary structure of the mRNA, potentially affecting its stability or its processing via alternative splicing of the 3’ UTR itself (Figure 3) (38).

The interaction between HLA-G and its receptors triggers distinct intracellular signaling cascades depending on the cell type involved, effectively reprogramming the immune microenvironment (11).

Inhibition of Cytotoxicity (via LILRB1): In effector cells such as NK cells and CD8+ T cells, the engagement of LILRB1 by HLA-G initiates a direct inhibitory program. Upon ligand binding, the Immunoreceptor Tyrosine-based Inhibitory Motifs (ITIMs) located within the cytoplasmic tail of LILRB1 undergo phosphorylation (12). This modification creates a docking site for the protein tyrosine phosphatases SHP-1 and SHP-2. Once recruited, these phosphatases dephosphorylate key signaling molecules involved in the activation pathways (such as ZAP70 or Syk), effectively terminating the “activation” signal (12, 39). This process directly suppresses granule exocytosis, cytokine production, and cell proliferation, thereby neutralizing the cytotoxic function of the immune effector cell (40).

Reprogramming of Antigen-Presenting Cells (via LILRB2): In myeloid APCs, the engagement of LILRB2 by HLA-G induces a more complex functional shift known as “tolerogenic reprogramming” (41). Unlike the canonical inhibitory pathway, the recruitment of SHP-2 by LILRB2 in dendritic cells and macrophages has been linked to the paradoxical activation of the NF-κB signaling pathway (42).This activation drives the transcription and secretion of Interleukin-6 (IL-6) (43).The secreted IL-6 then acts in an autocrine or paracrine manner to bind IL-6 receptors on the APC surface, triggering the phosphorylation of STAT3 (44).Constitutive STAT3 activation locks the APC into a tolerogenic state, characterized by the downregulation of MHC Class II and costimulatory molecules (CD80/86) and resistance to maturation signals (45). Consequently, these “reprogrammed” APCs not only fail to prime effector T cells but actively promote the differentiation of inducible regulatory T cells (iTregs) (46). Thus, the HLA-G/LILRB2 axis acts as a sophisticated switch to convert antigen-presenting cells from immunogenic to tolerogenic, actively shaping a suppressive microenvironment (Figure 4, Table 2).

The canonical and most well-understood physiological role of HLA-G is in human pregnancy, where it is the central mediator of materno-fetal tolerance (21). The fetus is a semi-allograft, expressing paternal antigens that should, by all immunological rules, be recognized as “non-self” and rejected by the maternal immune system. This rejection is actively prevented.

The mechanism for this tolerance is localized at the materno-fetal interface. Fetal-derived extravillous trophoblast (EVT) cells, which are invasive and migrate into the uterine wall (decidua), come into direct contact with maternal immune cells. These EVT cells have a unique and specialized MHC profile: they do not express the classical, highly polymorphic HLA-A and HLA-B molecules, which would present paternal peptides and trigger a CTL response. Instead, they uniquely and abundantly express HLA-G (21, 49).

This trophoblast-expressed HLA-G (both membrane-bound and soluble) interacts with the array of inhibitory receptors (LILRB1, LILRB2, and KIR2DL4) on the large population of maternal immune cells in the decidua, particularly decidual NK (dNK) cells and macrophages (48). This engagement neutralizes the cytotoxic potential of dNK cells, modulates macrophages towards a supportive (M2) phenotype, and protects the semi-allogeneic fetus from immune attack (15). This process actively establishes the placenta as a site of profound, HLA-G-mediated immune privilege.

The clinical significance of this tolerance mechanism is distinctively underscored by its disruption in pathological pregnancies. Emerging evidence suggests that the dysregulation of the HLA-G inhibitory axis is a critical driver of recurrent pregnancy loss (RPL) and infertility. Specifically, recent clinical reviews have highlighted that ‘immunological mismatch’ and defects in maternal-fetal compatibility are central to the pathogenesis of RPL (50). This pathology is often mediated by functional imbalances in dNK cells and T cell subsets (51), as well as specific alloimmune mechanisms that disrupt the delicate pro- and anti-inflammatory balance required for implantation (52). At the molecular level, the regulatory function of HLA-G in early pregnancy is exerted through complex signaling interactions with receptors on immune cells, particularly NK cells (53). Crucially, mechanistic studies—including those investigating tumor resistance models—have further elucidated how HLA-G engagement with receptors like KIR2DL4 can orchestrate immune escape and cytokine signaling (e.g., via JAK2/STAT1 pathways), offering translational insights that are highly relevant to understanding immune privilege at the fetal interface (54).

The “placental model” of tolerance has been directly and successfully applied to the field of solid organ transplantation (55). In this context, HLA-G expression is not a source of pathology but a sign of induced tolerance.

A large body of clinical evidence has demonstrated a strong positive correlation between the expression of HLA-G and improved allograft acceptance (56). This has been observed in heart, kidney, liver, and lung transplant recipients. Higher levels of HLA-G, detected either within the graft tissue via biopsy or, non-invasively, as soluble HLA-G (sHLA-G) in the recipient’s circulation, are associated with a significantly reduced incidence of both acute and chronic rejection episodes (24, 25).

This strong correlation has positioned HLA-G as a promising non-invasive biomarker for monitoring the immune status of transplant patients. Elevated sHLA-G levels may signify a state of operational tolerance and predict better long-term graft survival, potentially allowing for the careful tailoring of immunosuppressive drug regimens (57). Furthermore, these findings have inspired the development of novel therapeutic strategies (discussed in section 5.2.3) aimed at administering HLA-G agonists or inducing its expression to actively promote graft-specific tolerance (58).

Beyond the temporary immune-privileged site of the placenta, physiological HLA-G expression is found in a small number of other permanent “sanctuary” tissues that require constitutive protection from immune-mediated damage (59).

The Eye (Cornea): The anterior chamber of the eye and the cornea are classic examples of immune-privileged sites. This privilege is critical for preserving vision, as an inflammatory response in the cornea can lead to scarring and blindness. Studies have confirmed the presence of both HLA-G protein and its transcripts in healthy human corneal tissue, specifically in keratocytes, corneal epithelial cells, and endothelial cells (60). It is widely postulated that this constitutive HLA-G expression, similar to its role in the placenta, is a key factor in maintaining ocular immune tolerance and is a major reason for the high success rate of corneal allografts, which are often performed without HLA matching (59).

The Testis: The testis is another well-established immune-privileged site, essential for protecting developing germ cells from autoimmune attack (61). Sperm-specific antigens appear at puberty, long after central immune tolerance has been established, meaning they are “auto-antigenic.” The blood-testis barrier, formed by Sertoli cells, provides a physical barrier, but an active immunological barrier is also required. HLA-G has been identified in the testis, where its expression by cells like Sertoli and Leydig cells is believed to contribute to the local immunosuppressive microenvironment that actively tolerizes the immune system to the presence of these germ cell antigens (61, 62).

The most intensively studied pathological role of HLA-G is its aberrant expression in cancer (66). While absent in the vast majority of healthy adult tissues, de novo HLA-G expression is detected in a wide variety of human malignancies, including renal cell carcinoma, urothelial bladder carcinoma, melanoma, glioblastoma, colorectal cancer, gastric cancer, and breast cancer (67, 68).

The role of HLA-G in autoimmune diseases is substantially more complex and less clear-cut than in cancer. Here, the pathology is not one of simple upregulation but rather a dysregulation or imbalance of its tolerogenic function (72). Theoretically, higher levels of the inhibitory HLA-G molecule should be protective against autoimmunity.

However, clinical data are conflicting. Many studies associate the risk of autoimmune diseases like Systemic Lupus Erythematosus (SLE), Rheumatoid Arthritis (RA), and Multiple Sclerosis (MS) not with expression levels per se, but with genetic polymorphisms in the HLA-G regulatory regions (5’ URR and 3’ UTR) (60). This suggests that a genetically-determined “set point” for HLA-G expression may predispose an individual to an imbalanced immune response.

Studies measuring sHLA-G levels in patients have yielded inconsistent results (60). For example, in MS, some research suggests sHLA-G levels in the cerebrospinal fluid (CSF), but not necessarily in the serum, may be a useful marker for monitoring disease activity, while other studies find no association (73). Similarly, for RA, while the HLA-G gene may be upregulated in some patient cells, studies of the key 14-bp polymorphism have been inconclusive (74). This complexity suggests that the role of HLA-G in autoimmunity is highly context-dependent, and a simple “more tolerance is better” model does not fully capture the biology.

Similar to the strategy employed by tumors, many chronic viruses actively exploit the HLA-G pathway to evade immune clearance and establish a state of persistent infection (75).

The mechanism involves the induction of HLA-G expression on virus-infected host cells. This de novo expression, which is not present on the healthy cell, serves as a protective shield. By engaging inhibitory receptors on antiviral NK cells and CTLs, the induced HLA-G suppresses the immune response directed at the infected cell, allowing the virus to replicate and persist (16, 75).

This phenomenon has been documented in:

Human Immunodeficiency Virus (HIV): Elevated levels of sHLA-G are commonly found in the serum of untreated HIV patients. Furthermore, specific HLA-G polymorphisms, such as the G*010108 allele, have been linked to an increased risk of HIV-1 infection, suggesting a genetic predisposition to this evasion mechanism (76).

Hepatitis C Virus (HCV): Patients with chronic HCV infection also show elevated plasma sHLA-G levels (77). In the context of HCV/HIV co-infection, HLA-G expression within the liver is associated with more severe histopathological stages of disease and more rapid progression of liver fibrosis (78). This indicates that HLA-G is not just a marker of infection but an active contributor to its pathology.

Beyond oncology and virology, HLA-G expression plays a nuanced role in parasitic infections and allergic conditions. In parasitic diseases such as Malaria, HLA-G expression is often upregulated as a host-protective mechanism to dampen excessive inflammatory responses that could damage tissues (79). However, similar to the tumor context, parasites may exploit this tolerance to ensure their own survival. For instance, increased soluble HLA-G levels have been observed in patients with Plasmodium falciparum infection, correlating with higher parasitic load but reduced severity of cerebral symptoms (80).

Conversely, in allergic disorders like Asthma and Allergic Rhinitis, where the immune system is hyper-reactive, HLA-G deficiency or downregulation is often observed. Studies suggest that soluble HLA-G levels are significantly lower in asthmatic patients compared to healthy controls, implying that a lack of HLA-G-mediated suppression contributes to the sustained Th2-type inflammation and airway hyper-responsiveness (81). Thus, HLA-G appears to act as a “rheostat” for immune sensitivity, with its absence predisposing individuals to hypersensitivity reactions.

The strong correlation between HLA-G expression and disease status has driven significant research into its use as a clinical biomarker.

The identification of the HLA-G/ILT axis as a major, non-redundant immune checkpoint has triggered a wave of drug development, positioning it as a highly promising, next-generation therapeutic target (71).

Despite decades of progress, the single greatest hurdle impeding the clinical maturation of HLA-G is the crisis in detection and standardization (101).

To unlock the full potential of HLA-G, the field must pivot from its current exploratory phase to one defined by rigorous standardization and mechanistic precision.

The immediate and urgent priority is the development and validation of a new generation of isoform-specific monoclonal antibodies and standardized, quantitative assays (11). These tools must be capable of accurately and reproducibly quantifying each of the key isoforms (e.g., G1, G5) and, critically, distinguishing between free sHLA-G and HLA-Gev in liquid biopsies.

With these tools, future research can definitively dissect the in vivo function of each isoform and the true clinical impact of the various regulatory polymorphisms. Alternative therapeutic strategies, such as small molecules that prevent HLA-G dimerization or target its peptide-binding groove, should also be explored (104).

The ultimate goal is to move HLA-G from a general prognostic marker to a predictive biomarker that guides precision medicine. A future clinical workflow might involve:

Genotyping the patient’s HLA-G 5’ URR and 3’ UTR to determine their baseline expression potential.

Quantifying isoform-specific HLA-G, particularly HLA-Gev, in their plasma as a real-time measure of tumor immune evasion.

Profiling the TME for co-expression of LILRB1, LILRB2, and PD-L1.

Using this multi-modal signature, clinicians could one day stratify patients to select the optimal, individualized therapy: anti-HLA-G (TTX-080) for one patient, anti-LILRB1 (BND-22) for another, a bispecific T-cell engager for a third, or a combination blockade with anti-PD-1 for a fourth. Only by resolving the fundamental complexities of HLA-G biology can we fully realize its potential as a transformative target in immunology.