Human samples
Participants were included after written informed consent, and no compensation was provided. Sex and gender was not considered in the study design, and no sex- or gender-based analyses were performed, as they were determined not relevant within this study setting. Sex was reported as self-indicated in clinical reports. Gender identity was not specifically inquired within this study, as no clinical relevance of this was apparent within this study setting. Clinical routine data and blood samples for platelet isolation were collected in the in- or outpatient clinic of the Department of Dermatology in Heidelberg, Germany. Clinically proven psoriasis or AD patients, diagnosed by board-certified Dermatologists, were included. Patients were eligible for systemic anti-inflammatory therapy according to national and European guidelines33,34. Eligibility was defined as either being naïve to systemic therapy or having undergone a 12-week washout of prior therapy. Exclusion criteria consisted of infectious/malignant/additional autoimmune diseases, age <18 years, pregnancy as well as use of anti-platelet or anti-coagulant drugs. Clinical information and blood samples were collected at baseline (before therapy), at one and four months post-initiation of systemic therapy. Therapies such as ixekizumab, secukinumab, brodalumab, guselkumab, and others were administered according to drug data sheets and guideline recommendations, considering patient-specific characteristics and decisions made jointly by the patient and physician. Healthy donors were defined as individuals without inflammatory skin diseases and without meeting any of the exclusion criteria.
Platelet isolation
Platelets were isolated according to a previously published protocol for platelet RNA sequencing from whole blood collected in EDTA-coated tubes (cat. no. 02.1066.001, Sarstedt, Nuembrecht, Germany; 3 ml for high dimensional flow cytometry, 9 ml for proteomic analyses), yielding highly purified platelets35. In short, within max. 3 h post sample collection, platelet-rich-plasma was collected after a first centrifugation step (120 G, 20 min, room temperature (RT)). For spectral flow cytometry, after a further centrifugation step (120 G, 20 min, RT), the platelet pellet was resuspended in the patient’s own plasma containing 5% DMSO (cat. no. D4540-100ml, Merck KGaA, Darmstadt, Germany), and transferred to −80 °C in an isopropanol freezing container (Mr. Frosty, cat. no. 5100-0001, Thermo Scientific, Massachusetts, USA). For total platelet proteomics, after a further centrifugation step (120 G, 20 min, RT), the platelet pellet was resuspended in RIPA buffer (cat. no. 9806S, Cell Signaling Technology, Massachusetts, USA) with 1 tablet PHOSSTOP (cat. no. 4906837001 Roche Products, Basel, Switzerland), and cOmplete, Mini, EDTA-free Protease (cat. no. 4693159001, Roche Products, Basel, Switzerland), and directly transferred to −80 °C. Random assessment of platelet purity was performed by microscopy (Axio Vert. A1, Carl Zeiss, Oberkochen, Germany) and estimating the number of nucleated cells per 10 million platelets in a Neubauer counting chamber (Cat. no. 717805, BRAND GMBH + CO KG, Wertheim, Gemany) (Supplementary Fig 4a–c; ZEN microsopy software 3.11, Carl Zeiss, Oberkochen, Germany). For flow cytometry analyses, purity was controlled by additional gating on leukocyte markers (see Supplementary Fig. 4d).
Proteomics sample preparation
Proteins (10 µg) were run for 0.5 cm into an SDS-PAGE and the entire piece was cut out and digested using trypsin according to Shevchenko et al.36 adapted for the DigestPro MSi robotic system (INTAVIS Bioanalytical Instruments AG, Tuebingen, Germany).
Flow-cytometry analysis
Shortly prior to analysis, samples were taken out of −80 °C storage and briefly thawed in a waterbath (37 °C). They were then resuspended in 1 ml of wash buffer (WB; Tyrode’s buffer (cat. no T2397-1L, Merck KGaA, Darmstadt, Germany) + 100 mg BSA (cat. no. 9048-46-8, Carl Roth GmbH + Co. KG, Karlsruhe, Germany)) while working carefully to avoiding bubbles and mechanical stress. Cells were spun at 400 G for 5 min at RT. After straining through a 70uM filter, they were then stained with an antibody cocktail (see Supplementary Table 2) for 20 min at RT in the dark. After an additional wash step with WB, platelets were fixed using the cytofix/cytoperm kit (cat. No. 554714, BD Biosciences, California, USA) and additionally stained with antibodies coupled to intracellular markers (see Supplementary Table 2). All reagents were tested and titrated prior to use in experiments. Titrations were repeated for new commercial lots where appropriate. All antibody incubations were carried out in Tyrode’s buffer supplemented with BSA and 20% Brilliant Stain Buffer (cat. no. 563794, BD Biosciences, California, USA). The samples were washed once in Perm/Wash buffer (eBioscience, California, USA) and centrifuged to pellet the cells before resuspension in Tyrode’s Buffer prior to flow-cytometry acquisition. Samples were acquired the same day, on aCytek Aurora spectral analyser (Cytek Biosciences, California, USA) using SpectroFlo Software (version 2.0) following daily quality control procedures as instructed by the manufacturer.
MS method Orbitrap Exploris 480
A liquid chromatography- tandem mass spectrometry (LC-MS/MS) analysis was carried out on an Ultimate 3000 Ultra-High Performance Liquid Chromatography (UPLC) system (Thermo Fisher Scientific) directly connected to an Orbitrap Exploris 480 mass spectrometer for a total of 120 min. Peptides were online desalted on a trapping cartridge (cat. no. 174500; Acclaim PepMap300 C18, 5 µm, 300 Å wide pore; Thermo Fisher Scientific) for 3 min using 30 µl/min flow of 0.05%v/v trifluoroacetic acid (TFA; cat. no. 00202341A8BS; Biosolve) in water. The analytical multistep gradient (300 nl/min) was performed using a nanoEase MZ Peptide analytical column (cat. no. 186008794; 300 Å, 1.7 µm, 75 µm x 200 mm, Waters) using solvent A (0.1%v/v formic acid in water; cat. no. 00069141A8BS, Biosolve and cat. no. W6-4, Thermo Fisher Scientific) and solvent B (0.1%v/v formic acid in acetonitrile; cat. no. A955-212; Thermo Fisher Scientific). For 102 min the concentration of B was linearly ramped from 4% to 30%, followed by a quick ramp to 78%. After two minutes the concentration of B was lowered to 2% and a 10 min equilibration step appended. Eluting peptides were analyzed in the mass spectrometer using data-independent acquisition (DIA) mode. A full scan at 120k resolution (380-1400 m/z, 300% automatic gain control (AGC) target, 45 ms maximum injection time (maxIT)) was followed by 47 MS2 (DIA) windows covering the mass range from 400 to 1000 m/z with variable width (30k resolution, AGC target 1000%, maxIT 54 ms, with 1 m/z overlap, 28% higher-energy collisional dissociation (HCD) collision energy). System performance was constantly monitored via an internal tool using regular (approx. every two days) injections of a quality control (QC) sample.
Data analysis, spectral flow cytometry
From the raw data acquired by spectral flow-cytometry, using FlowJo software version 10.6.2 and 10.7.1 (TreeStar, BD, Oregon, USA), live platelets were identified using manual gating on FSC versus SSC. For flow-cytometry data, the compensation matrix was corrected in FlowJo (TreeStar, BD, Oregon, USA) by pre-gating on live platelets, and the total fraction of live, singlet platelets was exported. Data were then transformed with an inverse hyperbolic sine (arcsinh) function (co-factors ranging between 5 and 18000) and imported into the R environment (version 3.6.1) for subsequent analysis37.
The high dimensional analysis was carried out using the R environment, based on the workflow described previously by Hartmann et al.38 Briefly, UMAPs were generated using the package umap version 0.2.7.012, and FlowSOM clustering was overlaid on the dimensionality reduction maps13. Frequency plots were generated using the ggplot2 package version 3.3.5, and heatmaps were generated using the pheatmap package version 1.0.12.
Principal Component Analyses (PCA) were carried out using the PCATest R package. Statistical significance testing of the principal components was performed using permutation-based methods within PCATest (Supplementary Table 3).
Data analysis, proteomics
Analysis of DIA RAW files was performed with Spectronaut (Biognosys, version 17.1.221229.55965) in directDIA+ (deep) library-free mode. Default settings were applied with the following adaptions. Within the Pulsar Search in Result Filters the m/z Max was set to 1800 and Min to 300, the Relative Intensity was set to 5%. Within DIA Analysis under Quantification the Protein LFQ Method was set to MaxLFQ. The data was searched against the human proteome from Uniprot (human canonical reference database, containing 81,837 unique entries from 26.10.2022).
Downstream statistical analysis was then performed in R (version 4.2.2). For this, the Spectronaut output was imported into R using the package protti (version 0.6.0). Imported data was then further processed using the R package proteus (version 0.2.16). As part of the analysis with proteus, data was log2 transformed, followed by the Exploratory Data Analysis which included visualization of sample grouping (PCA, Hierarchical clustering); protein expression (Heatmaps) and protein expression sets and their intersections (Upset Plots). Although proteus offers wrappers around Limma functions, differential expression (DE) analysis was performed with the R package Limma (version 3.54.1) outside proteus to have more flexibility in linear modeling (see Supplementary Table 4–7). For graphical visualization of DE expression analysis results volcano plots were produced in R to show statistical significance -log10(adj. P. Val) versus log2 FC, and log10 (adj. P. Val) versus mean expression of total platelet proteins (see Supplementary Fig. 2−3).
Proteomics difference in skin diseases: In order to identify differentially expressed proteins between the skin diseases at the different post therapy timepoints and the baseline (prior to therapy) and the healthy volunteers, respectively, a simple linear model was fitted using Limma to each protein consisting of a fixed effect for a combined factor of the two main experimental factors skin disease (factor levels: Pso (psoriasis) and AD (atopic dermatitis)) and therapy time (factor levels: baseline=prior to therapy, 1 month = 1 month post systemic therapy, 4 months = 4 months post systemic therapy initiation). Using this approach separate coefficient for each of these factor combinations is built in the linear model fit, which then allows to extract comparisons of interest (e.g. Pso-4month vs Healthy controls) as contrasts. The final results include statistics for each protein including empirical Bayes moderated nominal p values which were adjusted for multiple testing by controlling the false discovery rate (FDR) using the Benjamini–Hochberg procedure39. As threshold, a protein was called DE with a multiple adjusted p ≤ 0.05%. No log fold change filter criterion was applied for statistical assessment to find DE proteins.
Statistics
Descriptive statistics as frequency, mean, range and standard deviation were used to analyse the human cohort. Kolmogorov-Smirnov test was used to investigate normality of the data. Where appropriate, Spearman correlations were applied. Multiple linear models were fitted for confounders and residuals visualized the using boxplots. Statistical analysis was carried out using the R package rstatix v.0.7.0. Briefly, unpaired t tests provided p values, which were then adjusted for multiple comparisons using a Benjamini–Hochberg (BH) test. P values of less than 0.05 were considered significant and are indicated by an asterisk (*) or the numerical value on the respective graphs39.
Ethics
This study was conducted according to the Declaration of Helsinki. Patients and donors provided written and informed consent, and the study was approved by the ethics committee of Heidelberg (approval no. S-834/2020).
Reporting summary
Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.