To evaluate GzmK levels in human psoriasis tissues, GzmK expression was assessed using microarray data available through the Gene Expression Omnibus (GEO, NCBI) database (32). According to dataset GDS4602 (33–35), GzmK mRNA expression is elevated in lesional psoriasis skin compared to both healthy control and non-lesional psoriasis skin ( Figure 1A ). Similar results were obtained using another dataset, GSD4600 (36, 37) ( Supplementary Figure 1A ). To further assess the presence of GzmK in psoriatic skin, a database query for GzmK was performed on skin mRNA microarray datasets publicly available on SEEK (Princeton University) (38). According to this search, the majority of prioritized datasets were related to psoriasis, underscoring the significance of GzmK to psoriasis pathology ( Supplementary Table 1 ). Importantly, dataset enrichment analysis revealed ‘psoriasis’ as the disease category with the highest number of relevant datasets retrieved, further emphasizing its significance in the context of GzmK ( Supplementary Table 2 ).

To examine GzmK at the protein level and its distribution in human psoriasis tissue, formalin-fixed paraffin-embedded sections of healthy control skin and lesional psoriasis skin were evaluated. In agreement with the GzmK mRNA expression data, all psoriasis skin samples exhibited a significant increase in GzmK protein levels compared with healthy control skin ( Figures 1B, C ). This increased GzmK detection occurred in all psoriasis samples despite reported differences in clinical appearance, sex, and age of individuals (patient data listed in Supplementary Table 3 ). Within psoriasis lesions, GzmK-positivity was also observed in the extracellular milieu surrounding the GzmK-positive cells compared with healthy controls ( Supplementary Figure 1B ), indicating potential extracellular secretion.

In all stained psoriasis lesions, GzmK-positive cells were primarily located within the inflammatory cell infiltrate of the papillary dermis and adjacent to small blood vessels ( Figure 1B ). In healthy skin, GzmK was detected almost exclusively in mast cells (Toluidine Blue O (TBO)⁺ and tryptase⁺), as previously reported (22). GzmK-positive mast cells were observed in psoriatic skin as well ( Supplementary Figures 1C–E ); however, there was a decrease in the percentage of GzmK-positive mast cells. Serial immunostaining and co-immunofluorescence of human psoriasis lesions further revealed a population of CD68⁺ cells (monocytes/macrophages) positive for GzmK ( Figure 1D , Supplementary Figure 2A ). We also detected a minor population of CD3⁺ cells (both CD4⁺ and CD8⁺ T cells) and Neutrophil Elastase (NE)⁺ cells (neutrophils) positive for GzmK ( Supplementary Figures 2A , 3A ). No GzmK was detected in CD56⁺ (natural killer cells), CD1a⁺ (Langerhans cells), nor CD11c⁺ cells (dermal dendritic cells) despite each of these cell populations being elevated in human psoriasis lesions ( Supplementary Figures 2A , 3A ). Further, human THP-1 monocytes and macrophages were confirmed as a source of GzmK in vitro ( Figure 1E ), corresponding with previous observations (21).

To investigate whether GzmK exerts a pathologic role in psoriasis, GZMK knockout (GzmK KO) mice were utilized. In comparative sequence analysis, human and mouse GzmK exhibited a 71.82% amino-acid sequence identity and full conservation of the catalytic triad (His, Asp, Ser) ( Supplementary Figure 4A ), underscoring the relevance of the mouse model for investigating the mechanistic role of GzmK in the context of psoriasis. Using the established murine model of imiquimod (IMQ)-induced psoriasis-like skin inflammation (39), a psoriasis-like phenotype was induced in both wild-type (WT) and GzmK KO mice ( Figure 2A ). WT and GzmK KO mice showed no visible signs of distress, and their body weights remained stable throughout the study ( Supplementary Figure 4B ).

Clinically recognizable lesions developed earlier in IMQ-treated WT mice (day 5) compared to GzmK KO mice (day 6) ( Figure 2B ). Erythema and desquamation severity were reduced in GzmK KO mice, as determined using the modified psoriasis area and severity index (PASI) scoring system (scoring outline listed in Supplementary Table 4 ) (40). Both sets of individual scores were significantly reduced in GzmK KO mice at days 6 and/or 7 ( Figure 2C ). Accordingly, GzmK KO mice exhibited a reduced cumulative severity score (defined as combined erythema and desquamation severity scores) for the duration of the study and a significant decrease in cumulative severity score at day 7 compared to WT mice ( Figure 2C ). Other phenotypes observed in GzmK KO mice included reduced inflammatory infiltrate in the upper dermis and the absence of parakeratosis and vascular dilation in the papillary dermis compared to WT mice ( Figure 2D , Supplementary Table 5 ). Altogether, these results indicate that GzmK deficiency attenuates disease severity in the murine model of IMQ-induced psoriasis.

Compared with IMQ-treated WT mice, GzmK KO mice displayed an attenuated inflammatory response, characterized by a decreased leukocyte infiltration ( Figures 3A, B ). T cells (CD3⁺) were most abundant in both IMQ-treated WT and GzmK KO mice compared to other cell types, similar to human psoriatic tissue (41). However, there was a decrease in the number of CD3⁺ per area in IMQ-treated lesions of GzmK KO mice compared to WT mice ( Supplementary Figures 5A, B ). Additionally, a decrease in microvascular density (CD31⁺) was observed in IMQ-treated GzmK KO mice compared to WT mice ( Supplementary Figures 5C, D ). These findings highlight the potential importance of GzmK in modulating inflammatory responses during disease development.

In IMQ-treated GzmK KO mice, IL-17 and IL-23 tissue levels were reduced compared to IMQ-treated WT mice, whereas IL-12p70 levels were increased in GzmK KO mice ( Figure 3C ). To examine the lesional tissue distribution of IL-23, immunohistochemistry was performed. IL-23A staining of murine tissues identified that the majority of IL-23⁺ cells were located in the dermis within the vicinity of blood vessels and exhibited a morphology characteristic of mononuclear cells ( Figure 3D ). In contrast to GzmK KO mice, infiltration of macrophages (F4/80⁺) (CD11c⁺), two major sources of IL-23 in psoriasis (42), were observed in IMQ-treated WT mice ( Figure 3E ). The role of GzmK in inducing IL-23 release from macrophages was therefore investigated. THP-1-derived M0 and M1 macrophages were exposed to increasing concentrations of active recombinant human GzmK. GzmK increased IL-23 mRNA expression in a dose-dependent manner and promoted IL-23 secretion from both M0 and M1 macrophages ( Figures 3F , G ). These results support a pro-inflammatory function for GzmK in psoriasis through the stimulation of IL-23 production.

The contribution of GzmK to epidermal hyperplasia, a hallmark of psoriasis, was assessed in the murine model of IMQ-induced psoriasis-like skin inflammation. IMQ-treated GzmK KO mice exhibited a reduction in epidermal thickness compared with WT mice at day 7 ( Figures 4A, B ). In IMQ-treated GzmK KO mice and untreated controls, the proliferation marker Ki67 was restricted to basal keratinocytes, whereas in IMQ-treated WT mice, a large number of suprabasal keratinocytes also exhibited positive staining for this marker ( Figure 4C ). Comparatively, GzmK KO mice exhibited a decrease in the mean number of Ki67⁺ cells per µm² compared with WT mice ( Figure 4D ). In IMQ-treated GzmK KO mice, a similar decrease in proliferating cell nuclear antigen (PCNA) was observed compared to IMQ-treated WT ( Supplementary Figure 6A ). Further, when both primary (NHEKs) and immortalized (HaCaTs) keratinocytes were cultured in the presence of GzmK, increased levels of PCNA were observed ( Supplementary Figure 6B ). Consequently, GzmK may contribute to psoriasis epidermal hyperplasia through the stimulation of keratinocyte proliferation.

As documented previously, both normal human epidermal keratinocytes (NHEKs) and keratinocyte cell lines (HaCaTs) express PAR-1 (43, 44). GzmK is known to activate PAR-1 through cleavage of its tethered ligand, initiating receptor activation and cellular responses from various cell types (21, 24, 26). We examined whether extracellular GzmK was inducing proliferation through the cleavage and subsequent activation of PAR-1. To address this, we used 2 siRNA sequences designed against human F2R (PAR-1). Both siRNAs were transfected at a final concentration of 40 nM in HaCaTs, which achieved an 87% (siRNA #1) and 100% (siRNA #2) reduction at 72 h post-transfection and 100% (siRNA #1 and #2) at 120 hours post-transfection ( Figures 5A, B ). Those HaCaTs transfected with PAR-1 siRNA demonstrated a marked reduction of GzmK-mediated proliferation as visualized by Ki67 immunocytochemistry ( Figure 5C ), supporting our hypothesis that GzmK contributes to keratinocyte proliferation through a PAR-1 dependent extracellular mechanism.

As activation of the MAPK signaling pathway has been previously linked to keratinocyte proliferation (45–47), we next assessed their potential downstream activation by PAR-1 in keratinocytes. Western blot analysis revealed a robust phosphorylation of p38 (MAPK14) and p44/42 (ERK1/2), but not p46/54 (SAPK/JNK) MAPK, in keratinocytes stimulated by GzmK for up to 3 hours ( Supplementary Figures 7A–C ).

We next examined the involvement of PAR-1 in GzmK-dependent MAPK signaling. When keratinocytes were transfected with F2R (PAR-1) siRNA as above, GzmK-mediated phosphorylation of both p38 and p44/42 MAPK was inhibited ( Figures 6A, B ). Taken together, the findings suggest that GzmK activation of the PAR-1/MAPK pathway contributes to the induction of keratinocyte proliferation.

To further understand the downstream signaling pathways involved in GzmK/PAR-1/MAPK-mediated keratinocyte proliferation, STAT3 phosphorylation was assessed. Upon GzmK stimulation of human keratinocytes, we observed a marked increase in STAT3 phosphorylation at Ser727 and to a lesser extent at Tyr705, both peaking at 15 mins and maintaining elevated levels up to 3 hours-post stimulation ( Supplementary Figure 7A ). This finding suggests a sustained activation of STAT3 signaling in response to GzmK.

We subsequently examined whether the observed GzmK-mediated STAT3 signaling was PAR-1 and/or MAPK pathway-dependent. Keratinocytes transfected with F2R (PAR-1) siRNA showed inhibited GzmK-mediated phosphorylation of STAT3 at both Ser727 and Tyr705 ( Figure 6B ). To determine whether this effect occurs downstream of MAPK, keratinocytes were pre-treated with U0126, a p44/42 MAPK inhibitor. Following this treatment, the phosphorylation of STAT3 Ser727 was notably abrogated ( Figure 6C ), indicating that STAT3 phosphorylation occurs downstream of PAR-1-mediated p44/42 activation.

Gene Expression Omnibus (GEO) (https://www.ncbi.nlm.nih.gov/geo/) was used to retrieve raw scRNA-seq gene expression profiles from healthy control subjects, non-lesional and lesional psoriasis patients (accession numbers: GSE13355, GSE30999). The GSE13355 dataset (GDS4602) was generated using human skin punch biopsies obtained from 58 psoriasis patients that sampled both lesional and non-lesional skin and 64 healthy control subjects (33–35), while the GSE30999 dataset (GDS4600) included lesional skin collected from 85 psoriasis patients along with matched biopsies of non-lesional psoriasis skin (36, 37). The resulting datasets were queried for GZMK normalized gene expression using the GEO Profiles database. The expression values in the table are after adjustment of RMA expression values (on the log scale) to account for batch and sex effects as appropriate.

SEEK (http://seek.princeton.edu/) is a search engine used to identify/query gene(s) co-expression with other genes or datasets. After defining the search range to (non-cancer) skin-related datasets, available datasets were ordered by their relevance to/enrichment for GZMK (see Supplementary Tables 1, 2 ).

Paraffin-embedded psoriasis skin was obtained retroactively from patients diagnosed with psoriasis (N > 3). Psoriasis patients had not received any treatments prior to biopsy (additional patient data is available in Supplementary Table 3 ). Paraffin-embedded healthy control skin was obtained from subjects undergoing elective abdominoplasty (N ≥ 3).

All animal ethics and procedures were approved and performed in accordance with the guidelines for animal experimentation as per the University of British Columbia (UBC) Animal Care Committee. C57BL/6 mice (listed as WT in text) were purchased from Jackson Laboratories (Bar Harbor, ME). GzmK knockout mice (C57BL/6 background) (listed as GzmK KO in text) were kindly gifted from Dr. Phillip Bird (Department of Biochemistry and Molecular Biology, Monash University, Australia) and are routinely backcrossed with C57BL/6 mice. Both WT and GzmK KO mice were bred and housed at the International Collaboration on Repair Discoveries (ICORD) vivarium. Six female mice (aged 6-8 weeks) were used per genotype.

The IMQ-induced mouse model of psoriasis was performed as previously described (39). Daily topical administration of IMQ on shaved dorsal skin induced erythematous, scaly lesions resembling human psoriasis. On day 8 of the experiment, mice were euthanized, and tissue collected for analysis.

Five-micrometer-thick, vertical sections were prepared from all formalin-fixed, paraffin-embedded skin samples and processed for staining using Hematoxylin and Eosin for morphological analysis, Toluidine Blue O (TBO) for mast cells, or specific antibodies for immunohistology as listed in Supplementary Table 6 . All histological and histochemical stains were imaged with the Aperio CS2 slide scanner (Leica Biosystems). The resulting images were prepared using the Aperio ImageScope Viewer (Leica Biosystems) or QuPath (University of Edinburgh) (81).

The density of positive cells and/or intensity of staining was determined using the formula: (number of positive cells or total intensity of strong positive staining)/(total skin area examined), which was measured using the Aperio ImageScope Viewer (Leica Biosystems) or by evaluating H-Score or Positive Cell Detection using QuPath (University of Edinburgh). Epidermal thickness was determined using the formula: (tissue epidermal area)/(tissue length), which was measured using the Aperio ImageScope Viewer (Leica Biosystems). Areas of skin layers (dermis and subcutaneous fat), hair follicles and glands were carefully excluded as appropriate.

Evaluations of disease severity were performed daily using a modified version of the Psoriasis Area and Severity Index as previously described ( Supplementary Table 4 ) (40). IMQ-induced psoriasis-like skin severity was graded independently by two dermatology researchers. The two scores were averaged.

Biopsies from the dorsal region were harvested and placed in formalin for subsequent paraffin-embedding or snap-frozen in liquid nitrogen.

Human keratinocytes (HaCaTs) were cultured in DMEM (Sigma-Aldrich) containing 10% (volume/volume) fetal bovine serum and 1% (volume/volume) penicillin/streptomycin. Normal human keratinocytes (NHEKs) were cultured in Keratinocyte Basal Medium 2 with supplements (Promo Cell) and 1% (volume/volume) penicillin/streptomycin. Human THP-1 monocytes (ATCC^® TIB202™) were cultured and differentiated into M0 macrophages, and polarized into M1 macrophages, as previously described (82). Cellular phenotypes of cells transformed from THP-1 monocytes were validated using flow cytometry ( Supplementary Figure S8A ). All cells were cultured in serum-free media conditions for 24 hours before and during each experiment. Where pertinent, cells were treated with rhGzmK protein (BonOpus, non-commercial supply).

HaCaT cells were seeded into 6-well plates at a density of 0.25 x10⁶ cells/well approximately 24 h before transfection. Two siRNA targeting the F2R gene (PAR-1) and a negative control siRNA with no complementary target sequences were used (Qiagen). Cells were transfected with 10-40 nM siRNA using HiPerFect Transfection Reagent (Qiagen), as per supplier instructions. The efficiency of gene knockdown was evaluated by Western blot after incubation in normal cell culture conditions for 72 and 120 h.

Coefficient of Variation (CV) was equal to an average of 1%. CV ≤ 5% was defined a-priori as the acceptable limit. Data was analyzed using Graphpad Software for Windows. Data was analyzed for normal (Gaussian) distribution using the Shapiro-Wilk test (alpha=0.05). Statistical significance was determined by t-tests or ANOVA with post-hoc test for multiple comparisons, as appropriate (particulars of statistical tests specified in figure legends). ^* P ≤ 0.05 was defined as statistically significant.

This study was carried out in accordance with the recommendations of institutional guidelines of the University of British Columbia. All human studies were approved by the University of British Columbia Human Research Ethics Committee. All animal experiments were approved by the Animal Care Committee.