Granule-secreted enzymes (granzymes) are a family of serine proteases identified to mediate cell death by natural killer cells and cytotoxic T lymphocytes (1–5). There are five human granzymes, comprising tryptases granzyme A (GzmA) and GzmK, aspartase GzmB, chymase GzmH, and metase GzmM. Despite sharing structural sequence homology and a conserved secondary structure (6, 7), granzymes exhibit distinct substrates and varied roles in both healthy tissues and pathologic ones in multiple disease modalities. GzmA and GzmB are the most extensively studied granzymes, while the others are less well elucidated, thus referred to as ‘orphan’ granzymes. In recent years, several emerging studies have focused on GzmK, revealing significant implications in various diseases and offering new insights into mechanisms of action. As a result, investigating GzmK has become an exciting area of active research.

GZMK, the human GzmK gene (EC: 3.4.21) is located on chromosome 5.q11.2 and encodes a 264 amino acid protein. Also known as granzyme-3, fragmentin-3, or NK-tryptase-2, GzmK is synthesized in the rough endoplasmic reticulum as a zymogen precursor and then stored in granules, where it is associated with the proteoglycan serglycin. To become proteolytically active, the proteinase cathepsin C (also known as dipeptidyl peptidase I) performs NH₂-terminal processing (8). As a highly cationic tryptase-like protease, GzmK cleaves after basic amino acids, preferentially after positions 6 and 9 but also after positions 7 and 8 (9). Since both GzmA and GzmK are tryptases, have closely related three-dimensional structures (9) and share some common substrates, GzmK was long considered a redundant enzyme to GzmA. GZMK is located near GZMA on chromosome 5, likely due to gene duplication. However, the idea that GzmK is merely redundant to GzmA is now rejected, as GzmK has unique substrates and functions that distinguish it from GzmA. GzmK and GzmA also shows wide structural variation around the active subsites (10).

GzmK detection in plasma and tissues is low in healthy conditions but becomes elevated in response to disease/tissue injury, specifically accumulating in regions of inflammation ( Table 1 ). GzmK is elevated in bronchoalveolar lavage fluid from acute bronchopneumonia patients and allergic asthma subject’s post-allergen challenge, but not mild chronic obstructive pulmonary disease (11). Plasma GzmK is elevated in patients experiencing sepsis (12) and Dengue fever (13). In renal transplant patients with immunosuppressive therapy and suffering from cytomegalovirus infection, plasma GzmK is elevated and associated directly with the infection (13). GzmK is also transiently elevated in circulation following lipopolysaccharide (LPS) administration (14). GzmK is only released upon stimulation with Pseudomonas aeruginosa, but not Escherichia coli BL21, and Neisseria meningitidis, suggesting upregulation is pathogen specific. Tissue GzmK levels predict chronic rhinosinusitis-associated nasal polyp recurrence and asthma comorbidity (15). GzmK is also elevated in skin damaged by acute burn injury (16), and lesions in the inflammatory skin diseases, psoriasis (17) and atopic dermatitis (18).

Specific immune cell populations, and in particular T cells, have augmented GzmK expression in response to certain disease states, including rheumatoid arthritis, amyotrophic lateral sclerosis, and aging. A subset of GzmK⁺ exhausted memory T cells (Taa) has been identified to accumulate with age in the lung, liver, peritoneal cavity, and spleen (19). Separately, GzmK⁺ CD8⁺ T cells were found to be higher in the plasma of older adults (20). Humans seropositive for cytomegalovirus exhibit higher GzmK⁺ CD8⁺ T cells. GzmK⁺ NK cell frequency is inversely correlated with antibody titers pre-and post-influenza vaccination. GzmK⁺ T cells are increased in both cancer and inflammaging, including squamous cell carcinoma (21), melanoma (22), Alzheimer’s disease (23), and atherosclerosis (24, 25). In Sjögren’s syndrome patients, there is an increased proportion of CXCR6⁺GzmK⁺CD8⁺ T cells in the peripheral blood, with these displaying an activated phenotype (26). In rheumatoid arthritis, GzmK⁺ CD8 T cells are enriched, with these greater than 10% of all live cells in inflamed RA synovium (27). GzmK⁺ cytotoxic T cells were also found to be a major CD8⁺ T cell population in gut samples from Crohn’s disease patients and bronchoalveolar lavage fluid samples from COVID-19 patients, with these enriched in diseased tissue but also found in circulation (27). Finally, there is a higher proportion of CD8⁺GzmK^hi effector memory T cells in the cerebrospinal fluid of patients with amyotrophic lateral sclerosis (28).

The development of a GzmK knockout (GzmK^-/-) mouse (29) has allowed elucidation of the biological role of this protease in a variety of disease states. Comparing GzmK^-/- and GzmA^-/- mice has allowed confirmation that there is the lack of overlap between the functions of GzmK and GzmA. GzmK^-/- mice exposed to Chikungunya virus infection displayed reduced foot swelling, although this is less than observed in GzmA^-/- mice (30). Sepsis scores are also reduced in GzmK^-/- mice compared to WT mice, however, only GzmA^-/- mice have improved survival (31).

In acute burn injury, GzmK^-/- mice resolved inflammation faster, and improved wound closure, quality of healing, and scar strength compared to wild-type mice (16). Separately, in oxazolone-dermatitis (18) and imiquimod-psoriasis (17) models of inflammatory skin disease, severity is reduced in GzmK^-/- mice. In the dermatitis mice, GzmK^-/- mice display reduced scaling, erosions and erythema, with an associated improvement in angiogenesis and decreased microvascular damage. In the psoriasis mice, GzmK^-/- mice have reduced plaque formation, less erythema, and decreased epidermal thickening. Using a different GzmK^-/- mouse, this time with skin exposed to imiquimod to induce skin inflammation, there is decreased erythema, scaling and skin thickness (15). In mouse asthma models, GzmK knockdown or pharmacological inhibition decreased tissue pathology and restored lung function (15). GzmK^-/- mice display reduced arthritis severity and dermatitis with reduced complement activation.

Historically, all granzymes were believed to mediate cytotoxic lymphocyte-mediated cell death. Upon target cell engagement, granules release their granzyme payload into the immunological synapse. The pore-forming protein perforin is released in conjunction with these granzymes and facilitates granzymes entry into the target cells. Once internalized, granzymes induce cell death through caspase-independent or -dependent mechanisms (reviewed in (32)). The specific details related to these mechanisms remain unclear and have not been independently confirmed. As such, this remains an area of controversial area, especially the idea that GzmK contributes to cell killing, and has been questioned by several independent studies (33).

There are three main key indicators that GzmK is released from cells and into the extracellular space: 1/leakage from the immunological synapse, 2/secretion from non-cytotoxic and possibly non-immune cells, and 3/interaction with extracellular substrates. Notably, the GzmK⁺ CD8 T cells found to have a relatively increased expression in multiple disease states minimally express cytotoxic markers (27), suggesting extracellular roles may be especially important in disease.

Human GzmK has been demonstrated to bind to both purified LPS and LPS on Gram-negative bacterial cell walls (41). GzmK modulates toll-like receptor 4 (TLR4) signaling in immune cells, leading to increased pro-inflammatory cytokine expression, including TNF-α from monocytes and IL-1β from macrophages. Together, extracellular GzmK contributes directly to the immune response to bacterial infections. However, based on studies in GzmK^-/- mice, the contribution of GzmK to overall disease severity in response to infection appears to be less than other immune-secreted proteases (i.e., GzmA) (29, 30).

PARs, a subfamily of G protein-coupled receptors, mediate the cellular effects of proteinases. Comprising PAR1, 2, 3 and 4, they have unique but sometimes overlapping roles in inflammation, hemostasis, and thrombosis (45). Multiple studies confirm GzmK to cleave and activate PAR1 (40, 44). This leads to increase pro-inflammatory cytokine secretion and has been observed in multiple cell types, including peritoneal macrophages and cultured M1 macrophages (IL-1β) (16, 46), lung fibroblasts (IL-6, IL-8) (40), keratinocytes and skin fibroblasts (IL-6) (16), and endothelial cells (IL-6) (44). These observations are supported in vivo, where GzmK^-/- mice with acute burns display decreased IL-1β and IL-6 compared to WT mice (16). GzmK-mediated PAR1 activation in endothelial cells also increases the expression of intercellular adhesion molecule (ICAM)-1, vascular cell adhesion molecule (VCAM)-1, and monocyte chemotactic protein (MCP)-1 (44). This led to increased monocyte attachment to endothelial cells, suggesting GzmK as an immune cell attractant. In support, thermal injured GzmK^-/- mice wounds display reduced ICAM-1, VCAM-1, and MCP-1 expression in combination with lower macrophage detection (16).

More recently, GzmK is reported to cleave and activate PAR2 (18, 43). Similarly to PAR1, GzmK-mediated PAR2 activation increases pro-inflammatory cytokine expression (IL-6 and IL-8), which is observed in epithelial cells (43). GzmK activation of PAR2 is separately demonstrated through detection of cleavage on the surface of a reporter cell (nLuc–PAR–eYFP CHO) (18). Mechanistically, GzmK-mediated PAR2 cleavage led to recruitment of β-arrestin and phosphorylation of ERK (43). Notably, both GzmK and trypsin cleave PAR2 at the same location, however, GzmK is unable to induce a classical Ca²⁺ flux.

Multiple naturally expressed proteases have also been described to cleave PAR1 and/or PAR2, including trypsin, plasmin, kallikreins, neutrophil elastase, mast cell tryptase, tissue factor/factor VIIa/factor Xa, activated protein C, and matrix metalloproteinase-1 (47). Some of these proteases may be dysregulated in disease, whilst others are tightly regulated. As an example, although matrix metalloproteinase-1 is elevated in response to tissue injury, it is tightly regulated by tissue inhibitors of metalloproteinases (TIMPs), thereby limiting its ability to act uncontrolled. To better understand the role of GzmK in disease, future studies must therefore elucidate the relative contributions these proteases play in PAR activation and under what conditions does GzmK have the greatest impact. This includes identifying whether GzmK accumulates and increases its proteolytic activity in response to sustained inflammation.

Lymphocyte-derived GzmK is emerging as having a key role in complement activation (38). GzmK mediates activation through the cleavage of C2 and C4. Ultimately, this results in the generation of C3a, C3b, C4b, and C5a, which are key effectors of complement. This has been observed in vivo in rheumatoid arthritis patients, where regions of complement activation correspond to increased GzmK detection (38). Moreover, in arthritis and dermatitis mice, GzmK-mediated complement activation reportedly contributes to disease progression.

Together, it is now clear GzmK mediates a pro-inflammatory phenotype, occurring through multiple and distinct pathways. GzmK will therefore likely have important pathologic roles in multiple disease modalities characterized by inflammation.