This study included 20 healthy donors (HDs) and 20 patients with psoriasis recruited at the Department of Dermatology and Allergy, Copenhagen University Hospital, Herlev and Gentofte with characteristics shown in Table 1 . None of the patients included were diagnosed with psoriatic arthritis. Collection of blood from both groups was approved by the Ethical Committee for the Capital Region of Denmark (nr. H-19031332) and written informed consent was obtained from all the participants. All patients were naïve for systemic psoriasis treatment, excluding one patient treated with methotrexate and brodalumab with a 6-week washout period prior to recruitment. All patients stopped any topical treatment two weeks prior to recruitment and did not receive phototherapy in the 3 months prior to sample collection.

Whole blood was collected in clot activator tubes (cat. 368815; BD Vacutainer®) and spun at 1200 x g for 10 minutes at 4°C. Serum was then aliquoted and stored at -80°C until further use. All aliquots were thawed only once before use.

Blood was collected in BD CPT™ tubes (cat. 362780; BD Vacutainer®) which contain both sodium heparin as anticoagulant and a liquid density medium that allows in-tube mononuclear blood cell separation. Subsequently, the tubes were spun at 1500 x g for 20 minutes at room temperature. After centrifugation, mononuclear cells (MNCs) were collected, resuspended in PBS + 2% fetal bovine serum (FBS) (v/v) and washed twice at 300 x g for 10 minutes. Cells were counted with a NucleoCounter® NC-100™ device (cat. 900-0004; Chemometec) and resuspended in RPMI 1640 medium (cat. #01-106-1a; Lonza) supplemented with 30% FBS (v/v) and 10% DMSO (v/v) for cryopreservation. Aliquots of 5x10⁶ cells were frozen overnight in CoolCell® LX boxes (cat. BCS-405PK; Biocision) at -80°C and transferred to liquid nitrogen tanks the next day.

Isolation of MNCs from one HD and one patient with psoriasis yielded less than 5x10⁶ cells; both were excluded from the experiments involving cell cultures.

Lyophilized native LL37 (LLGDFFRKSKEKIGKEFKRIVQRIKDFLRNLVPRTES) and citLL37 (LLGDFF[Cit]KSKEKIGKEFK[Cit]IVQ[Cit]IKDFL[Cit]NLVP[Cit]TES) (cat. SP-LL37-1 and SP-5380-1, Innovagen) were resuspended in sterile water at a concentration of 1 mg/ml. Aliquots were stored at -80°C. Lyophilized tetanus toxin (cat. 582243, Calbiochem) was resuspended in sterile water at a concentration of 50 μg/ml and stored at 4°C.

MNC aliquots were thawed and resuspended in RPMI 1640 supplemented with 10% (v/v) normal human serum (NHS) (cat. H4522, Sigma-Aldrich). Cells were then washed in PBS + 2% FBS (v/v) and counted using a NucleoCounter® NC-100™ device. 250,000 MNCs were added to each well in a final volume of 200 μl RPMI 1640 supplemented with 0.1% (v/v) gentamycin and 10% (v/v) NHS in 96-well, round-bottomed plates. Cells were cultured in the presence or absence of 5 μg/ml of native LL37, citLL37, or 0.5 μg/ml tetanus toxin for 29 hours. At that timepoint, cells were re-stimulated with the same concentrations of either native LL37, citLL37, or tetanus toxin used for initial stimulation. At 32 hours, a mix of brefeldin A + monensin (cat. 555029 and 554724, respectively, BD Biosciences) was added, diluted 1:1000 and 1:1500, respectively, to allow intracellular accumulation of cytokines for 16 hours. All samples were measured as duplicates during the optimization process to assess timepoints for cytokine measurement and eligible concentrations of LL37 and citLL37. Once the experimental conditions were established, the samples were run in singlets due to limitations in the amounts of cells available.

After 48 hours of culturing, plates were spun for 5 minutes at 300 x g, 4°C. MNCs were washed with 200 μl PBS + 2% FBS (FACS-PBS) and spun again as described above. Cells were resuspended in a mix of 1 μl immunoglobulin for intravenous use (IvIg, cat. 034401, CSL Behring) and 2 μl mouse serum (MS, cat. M5905, Sigma Aldrich) and subsequently, 20 μl of a mix of pre-titrated antibodies against extracellular markers was added as described in Table 2 for 30 minutes at 4°C. MNCs were then washed and resuspended in 100 μl of fixation/permeabilization solution (cat. 51.2090KZ, BD Biosciences). After 20 minutes of incubation at 4°C, cells were washed twice in 200 μl BD Perm/WashTM Buffer (cat. 51-2090KZ, BD Biosciences) and resuspended in 3 μl of MS + IvIg and 20 μl of a mix of pre-titrated antibodies against the cytokines captured inside the cells as shown in Table 2 . MNCs were re-incubated for 30 minutes at 4°C and washed in BD Perm/WashTM Buffer. Subsequently, they were resuspended in 200 μl PBS, and 80,000 cells were acquired using an AttuneTM NxT flow cytometer (Thermo Fisher Scientific). The gating strategy followed for the identification of cytokine-producing MNC populations is shown in Supplementary Figure 1 . The percentages of CD4⁺ T, CD8⁺ T, and B cells as well as the absolute number of cells in the gates in relation to the total amount of events acquired are presented in Supplementary Table 1 . Also shown in that table are the percentages of cytokine-positive cells in relation to their parental population as well as the absolute number of cells in the gates.

The concentrations of TNF-α (analyzed by means of bead and antibody set with cat. 171BA015M, Bio-Rad), IL-10 (171BA004M, Bio-Rad), IL-17 (171BA005M, Bio-Rad), IFN-γ (171BA013M, Bio-Rad), and IL-22 (171BA008M, Bio-Rad) in undiluted culture supernatants were analyzed by Luminex assays (reagent kit and standards; cat. 171304090M and 171DA0001, Bio-Rad) using a Bio-Plex 200 reader (Bio-Rad) according to the manufacturer’s protocol. The upper and lower limits of quantification (ULOQ and LLOQ, respectively) for each assay were: ULOQ/LLOQ (TNF-α): 4.68 and 0.3 pg/ml; ULOQ/LLOQ (IL-10): 12.92 and 3.2 pg/ml; ULOQ/LLOQ (IL-17): 25.92 and 3.1 pg/ml; ULOQ/LLOQ (IFN-γ): 11.38 and 0.7 pg/ml; and ULOQ/LLOQ (IL-22): 41.57 and 2.5 pg/ml.

Nunc amino immobilizer plates C8 (cat. 436023; Thermo Fisher) were coated for 1 hour shaking with 100 μl of a 2 μg/ml dilution of native LL37 or citLL37 in 100 mM sodium carbonate buffer, pH 9.6. Subsequently, wells were washed three times with PBST 0.05%, and 100 μl of 10 mM ethanolamine (cat. E9508, Sigma-Aldrich) diluted in coating buffer was added for 1 hour. After three further washes with PBST 0.05%, 100 μl of serum samples, diluted 1:25 in PBST 0.05%, was added. Plates were incubated for 1 hour, shaking. After three washes, 100 μl of an anti-human IgG antibody conjugated to horseradish peroxidase (HRP) (cat. P0214, Dako), diluted 1:15000 in PBST 0.05%, was added for one hour, shaking. Wells were washed three times more, and 100 μl of 1-Step Ultra TMB ELISA Substrate solution (cat. 34029, Thermo Fisher) was added. The colorimetric reaction was stopped with 0.5 M H₂SO₄, and optical density (OD) was measured at 450 and 540 nm using a SPECTROstar nano Microplate Reader (BMG LabTech).

All samples were run in duplicates and uncoated wells were used as negative controls. Net OD values were obtained after subtraction of the signal from blank wells.

Flat-bottomed, 96-well SH plates (cat. 269620, Thermo Fisher) were coated overnight at 4°C with 100 μl of an anti-MPO antibody (cat. 0400-0002, BioRad) at a final concentration of 2 μg/ml diluted in carbonate-bicarbonate buffer (15.7 mM and 34.3 mM, respectively), pH 9.6. The next day, wells were washed 3 times with 200 μl PBS + 0.2% (v/v) Tween-20 (PBST 0.2%) and blocked with 100 μl PBST 0.2% + 5% (v/v) milk (cat. M7409, Merck) for 2 hours. Wells were then washed 3 times with PBST 0.2% and 100 μl of serum was added to each well diluted 1:25 in PBST 0.2% + 1% probumin. Plates were incubated at 4°C overnight and the next day wells were washed 5 times as described above. Subsequently, 100 μl of an antibody against DNA conjugated to HRP (cat. 11544675001, Roche) was added at a 1:100 dilution in PBST 0.2% + 1% probumin.

For analysis of native LL37, citLL37 and IgG associated with NETs, ELISA plates were coated, blocked, and incubated with serum as described above but the detection antibody systems were those described in Table 3 , all in a total volume of 100 μl. Detection of human serum albumin in NETs was used as a negative control.

In all cases, after 2 hours of incubation with the HRP-conjugated antibody, wells were washed 5 times with PBST 0.2% and 100 μl of 1-Step Ultra TMB ELISA Substrate solution (cat. 34029, Thermo Fisher) was added. The colorimetric reaction was stopped with 0.5 M H₂SO₄, and OD was measured at 450 and 540 nm using a SPECTROstar nano Microplate Reader (BMG LabTech).

All samples were run in duplicates and uncoated wells lacking the anti-MPO antibody, were used as negative controls. Final OD values were obtained after subtraction of the signal from blank wells.

Cut-offs for the ELISA assays were calculated using a segmented regression model with breakpoint estimation using the R Studio software (1.4.1717, R Studio, 4.1.2); estimated breakpoints and confidence intervals for each breakpoint estimate are shown in Supplementary Figures 2 and 3 . Differences between patients and HDs in the frequencies of individuals positive for NETs and positive for NETs containing native LL37, citLL37 and IgG, as well as differences in the NET content between the two patient clusters, were calculated with Fisher’s exact test. Differences between unstimulated samples and native LL37- or citLL37-stimulated samples were evaluated using the Wilcoxon matched-pairs signed-rank test. Differences between the age of disease onset, body mass index, and psoriasis area and severity index of the two patient clusters were assessed using the Mann-Whitney U test. p-values <0.05 were considered statistically significant. The clustering of patients and controls was performed using the ComplexHeatmap package in R Studio applying Ward’s method for hierarchical clustering to the rows (representing the percentage of positive cells for each cytokine produced after citLL37 stimulation). The argument “k_row” was not specified to allow the clustering method to determine the optimal number of clusters in both groups. Graphical representation of the data and statistics were performed with R studio and/or GraphPad Prism 9 software (GraphPad Software, San Diego, CA, USA).

We examined the cytokine responses of CD4⁺ and CD8⁺ T cells to native LL37 and its citrullinated counterpart, citLL37 ( Figure 1 ). Tetanus toxin (TT) was used as positive control.

Neither native nor citLL37 induced IFN-γ-producing CD4⁺ T cells in MNC cultures from HDs, while citLL37 induced a significant increase in the proportion of IFN-γ-producing CD4⁺ and CD8⁺ T cells in cultures from patients with psoriasis compared to unstimulated- and native LL37-stimulated cultures ( Figures 1A, B, E, F ). CitLL37 also induced a significant increase in the proportion of IL-10-producing CD4⁺ T cells from patients with psoriasis compared to unstimulated cultures and cultures stimulated with native LL37, and a similar tendency was observed in the HD group ( Figures 1C, D ). We detected very low frequencies of IL-17⁺ CD4⁺ T cells after stimulation of patient and control MNCs with LL-37 or citLL37 ( Supplementary Figure 4 ), but citLL37 did induced a significant increase in the proportion IL-17 producing cells from patients with psoriasis compared to unstimulated and LL-37 stimulated cultures. We did not observe induction of IL-22- or TNF-α-producing CD4⁺ T cells in either group.

The proportion of IFN-γ-producing B cells in MNC cultures from patients increased upon stimulation with both forms of LL37, albeit only significantly after stimulation with citLL37 ( Figures 2A, B ). TT was used as a positive control. B cells producing IL-6, TNF-α, or IL-10 were not detected in either of the study groups.

Three HDs (15%) and four patients (20%) showed antibody reactivity against native LL37 ( Figure 2C ). Of these, all three HDs and two of the patients also tested positive for antibodies against citLL37, while one HD displayed antibody reactivity against the citrullinated peptide only ( Figure 2D ).

Native LL37 induced secretion of IL-10 by MNCs isolated from patients with psoriasis, and a similar trend was observed after stimulation with citLL37 ( Figure 3A ). Both forms of LL37 tended to induce secretion of TNF-α by MNCs from patients with psoriasis, but not by MNCs from HDs ( Figure 3B ). IFN-γ was only detected in some culture supernatants and did not show any distinct pattern ( Figure 3C ). TT was used as a positive control. The levels of IL-17 and IL-22 in the culture supernatants were constantly below the detection limit of the assays.

Having established that LL37 is capable of inducing cytokine production by T cells and B cells, we examined if circulating NETs contain LL37.

A cut-off OD value to differentiate NET-positive from NET-negative samples was calculated ( Supplementary Figure 2 ), and on this basis, 8 out of 20 patients with psoriasis (40%) and 11 out of 20 healthy donors (HDs) (55%) tested positive for circulating NETs ( Figure 4A ).

The NETs from three out of the eight NET-positive patients (38%) versus one out of the eleven NET-positive HDs (9%) showed attachment of IgG ( Figure 4B ), and the NETs from five patients with psoriasis (63%) versus three HDs (27%) contained native LL37 ( Figure 4C ). CitLL37 was only found in NETs from two individuals, who were both patients with psoriasis (25%) ( Figure 4D ).

Based on the T- and B-cell IFN-γ and IL-10 responses to citLL37 in patients with psoriasis, we identified two distinct clusters of patients: ten high-responders (cluster 1) and nine low-responders (cluster 2) ( Figure 5A ). No clear clustering was observed among HDs ( Figure 5B ). The majority of patients (70%) in cluster 1 had circulating NETs containing either native LL37, citLL37 and/or IgG, versus only one patient (11%) in cluster 2 (p=0.019).

Neither the cytokine responses to citLL37 nor the presence of circulating NETs correlated with the carriage of the risk allele HLA-C*06:02:01 in the patient group. HLA-C*06:02 is the strongest susceptibility allele regarding genetic risk factors (21), with roughly 50% of patients with psoriasis carrying it (22).

Clinically, patients in cluster 1 had lower age at psoriasis onset than patients in cluster 2 ( Figure 5D ). The BMI and PASI varied highly within both clusters and did not differ significantly between them ( Figures 5C, E ).