Introduction

Psoriasis is a chronic and inflammatory disorder that mainly affects the skin, with an estimated prevalence of 2–4% worldwide1,2. In recent years, psoriasis is now more than beyond skin and has important systemic manifestations such as psoriatic arthritis, metabolic syndrome, and cardiovascular disease3.

Epidemiologic studies have shown a pathophysiological correlation between psoriasis and cardiovascular disease (CVD) based on the concept of systemic inflammation1,4,5. These studies suggest that possible mechanisms such as insulin resistance, oxidative stress, and common inflammatory cytokines, can be used to explain the link between psoriasis and CVD1,3,6. Risk factors of CVD including hypertension, diabetes mellitus, dyslipidemia, smoking, alcohol intake, and obesity are found more frequently in patients with psoriasis than in healthy people7. As the prevalence of CVD risk factors is increasing, life expectancy has been shown to decrease in patients with psoriasis, emphasizing that a more accurate assessment of CVD risks can allow for making earlier treatments and prevention of clinical events and death in patients with psoriasis.

During the last few decades, proteomics technology has been extensively employed to investigate several aspects of psoriasis and CVD. Additionally, it has been widely used to expand the pathogenesis and pathophysiology of diseases, and discovering novel biomarkers for diagnosis, prognosis, and evaluation of treatment efficacy8. Although biomarker candidates associated with psoriasis or CVD, respectively, have been suggested2,9,10, they are limited in reflecting CVD occurrence in patients with psoriasis. Thus, the discovery of novel protein biomarkers specific to CVD occurrence in patients with psoriasis is expected to improve the risk assessment for CVD in these patients. Therefore, this study aimed to identify novel biomarker candidates for predicting the risk of CVD in patients with psoriasis and tandem mass tag (TMT)-based quantitative proteomic analysis allowing for simultaneous quantification of multiple samples was pursued using serum samples composed of psoriasis groups with and without CVD risk factors and healthy controls.

Methods

Ethics statement

The study protocol was approved by the Institutional Review Board of Hanyang University Hospital (HYU 2018-05-017), and all of the research was carried out in accordance with the applicable rules and regulations. All participants provided written informed consent prior to enrollment.

Participants and serum samples

The study participants were divided into three groups (Group A, B, and C) recruited from the dermatology outpatients at the Hanyang University Hospital. Group A (n = 7) included psoriasis patients with CVD risk factors. The patients had been diagnosed with psoriasis and had at least one underlying condition including hypertension, diabetes mellitus, and dyslipidemia. Group B (n = 4) included psoriasis patients who had no underlying conditions as described above. Group C (n = 3) included healthy patients as the control group. We excluded participants with underlying diseases other than CVD risk factors that could affect the systemic inflammatory responses. We collected serum samples from the participants and selected a total of 14 age- and sex-matched individuals for proteomic analysis, as shown in Table 1. Blood samples from each participant was collected in a plain tube (6 mL Vacutainer (BD, Franklin Lakes, NJ, USA)). The blood samples were allowed to clot for 15–30 min at room temperature. Finally, serum from peripheral blood was prepared by centrifugation (1300×g, 10 min) and stored in a deep freezer (at < − 70 ℃) until analysis.

Table 1 The demographic data of participants in three groups.
Full size table

Immunoaffinity depletion of top 14 high-abundance proteins and in-solution tryptic digestion

A protease inhibitor cocktail (Roche Diagnostics, Mannheim, Germany) was added to the 14 serum samples from the psoriasis with and without CVD risk factors (n = 7 and n = 4, respectively) and healthy control (n = 3) groups that were selected for proteomic analysis. The protein concentrations of each serum samples were determined using a bicinchoninic acid (BCA) protein assay kit (Thermo Fisher Scientific, Bremen, Germany). Each serum sample was then depleted of the top 14 high-abundance proteins (albumin, alpha-1 acid glycoprotein, alpha-1 antitrypsin, alpha-2 macroglobulin, apolipoprotein A-I, apolipoprotein A-II, complement C3, fibrinogen, haptoglobin, immunoglobulin A, immunoglobulin G, immunoglobulin M, serotransferrin, and transthyretin) using a human multiple affinity removal system (MARS)-14 column (4.6 × 50 mm, Agilent, Santa Clara, CA, USA) according to the manufacturer’s instructions. 1100 HPLC system (Agilent, Santa Clara, CA, USA) with a fraction collector was used to collect the flow-through fraction. The flow-through fraction not bound to the MARS-14 column in each serum sample was concentrated using Amicon Ultracel-3 centrifugal filter devices (3 kDa cutoff, Millipore, Billerica, MA, USA) and the protein concentrations of serum samples after the removal of the top 14 high-abundance proteins were determined using the BCA protein assay kit. The concentrated samples were resuspended in 8 M urea buffer/25 mM ammonium bicarbonate, and subsequently reduced and alkylated using 5 mM tris (2-carboxyethyl) phosphine hydrochloride and 10 mM iodoacetamide, respectively. Afterwards, 25 mM ammonium bicarbonate was added to the serum samples to decrease the urea concentration to less than 1 M and digested with lysyl endopeptidaseR (Lys-C, Fujifilm Wako Pure Chemical Corporation, Osaka, Japan) in an enzyme/substrate ratio of 1 mAU Lys-C per 50 μg of total protein at 25 °C for 2 h11. Trypsin (Promega, Madison, WI, USA) was then added to the samples at a protease/substrate ratio of 1:50 (wt/wt) and incubated at 37 °C overnight. Digested peptide samples were acidified and desalted using a Sep-Pak tC18 cartridge (Waters Corporation, Milford, MA, USA). The desalted peptide samples were then dried in a miVAC vacuum concentrator (Genevac Ltd., Ipswich, UK).

Tandem mass tags (TMT) labeling

The desalted peptides from each serum samples were resuspended in 30 μL of 100 mM triethyl ammonium bicarbonate (TEAB) and the peptide concentrations were determined using a quantitative colorimetric peptide assay kit (Thermo Fisher Scientific, Rockford, IL, USA). After TMTpro 16plex label reagents (0.5 mg per vial, Thermo Fisher Scientific, Rockford, IL, USA, VI306840 (lot number)) were resuspended in 60 μL of anhydrous acetonitrile (ACN), 30 μL of each label reagent was added to 30 μg of each peptide samples for labeling. In addition to the 14 serum samples used for proteomic analysis, two serum samples obtained from patients diagnosed as psoriatic arthritis were also labeled because a total of 16 TMT label reagents were available; however, the two serum samples were not used for TMT-based quantitative analysis (labels used 126: psoriasis patient 1 with CVD risk factors; 127N: psoriasis patient 2 with CVD risk factors; 127C: psoriasis patient 3 with CVD risk factors; 128N: psoriasis patient 1 without CVD risk factors; 128C: psoriasis patient 2 without CVD risk factors; 129N: psoriasis patient 3 without CVD risk factors; 129C: psoriasis patient 4 without CVD risk factors; 130N: control 1; 130C: control 2; 131N: control 3; 131C: psoriasis patient 4 with CVD risk factors; 132N: psoriasis patient 5 with CVD risk factors; 132C: psoriasis patient 6 with CVD risk factors; 133N: patient 1 with psoriatic arthritis; 133C: patient 2 with psoriatic arthritis; 134N: psoriasis patient 7 with CVD risk factors). After the labeling reaction was made for 60 min at room temperature, 5 μL of 5% hydroxylamine in 100 mM TEAB was added to each peptide sample and incubated for 15 min to quench the labeling reactions. Equal amounts of TMT-labeled samples were combined and desalted using a Sep-Pak tC18 cartridge and dried in an miVAC vacuum concentrator.

Basic pH reversed‑phase liquid chromatography

The TMT-labeled peptide samples were resuspended in 10 mM ammonium formate and the peptide concentrations were determined using a quantitative colorimetric peptide assay kit. Afterwards, 292 μg of each peptide sample was loaded onto an X-Bridge peptide BEH C18 column (4.6 mm i.d. × 250 mm length; pore size 130 Å; particle size 3.5 μM, Waters Corporation, Milford, MA, USA) and fractionated by basic pH reversed-phase liquid chromatography using an Agilent 1290 Infinity liquid chromatography (LC) system (Agilent Technology, Santa Clara, CA). The peptides were separated at a flow rate of 0.5 mL/min with the following gradient conditions: 0 min 100% buffer A (10 mM ammonium formate, pH 10) and 0% buffer B (10 mM ammonium formate (pH 10) in 90% acetonitrile); 0–10 min, 0–5% B; 10–48.5 min, 5–40% B; 48.5–62.5 min, 40–70% B; 62.5–72.5 min, 70% B; 72.5–82.5 min, 70–5% B; and 82.5–92.5 min, 5% B. Fractionation was pursued by collecting 96 wells (1 well/0.8 min, Restek corporation, Bellefonte, PA, USA) during the chromatographic run (from 10 to 82.5 min). The resultant 96 fractions were pooled into 24 concatenated fractions using the following rule: A set of arithmetic sequences with a common difference of 24 was pooled into one concatenated fraction; for example, fractions 1, 25, 49, and 63 were pooled into the first concatenated fraction. The 24 peptide fractions were dried and subsequently resuspended in 12.16 μL of 0.4% acetic acid.

Liquid chromatography and tandem mass spectrometry (LC–MS/MS) analysis

3 µg of the fractionated peptide sample was injected onto a trap column (2 cm × 75 µm i.d., 100 Å, 3 µm) and separated on a reversed-phase Acclaim PepMap RSLC C18 column (50 cm × 75 µm i.d., 100 Å, 2 µm) using an UltiMate 3000 RSLCnano System (Thermo Fisher Scientific, Bremen, Germany). The column temperature was constantly set to 50 ℃ by using a column heater. The operating flow rate was 300 nL/min with the following gradient conditions: 0 min 95% buffer A (100% water with 0.1% formic acid) and 5% buffer B (100% acetonitrile with 0.1% formic acid); 0–4 min, 5% B; 4–13 min, 5–10% B; 13–150 min, 10–25% B; 150–155 min, 25–28% B; 155–160 min, 28–40% B; 160–165 min, 40–80% B; 165–170 min, 80% B; 170–170.1 min, 80–5% B; and 170.1–180 min, 5–0% B. The nano UHPLC system was coupled to an Orbitrap Eclipse Tribrid mass spectrometer (Thermo Fisher Scientific, Bremen, Germany). MS1 data were collected using an Orbitrap (120,000 resolution; scan range of 350–2000 m/z; maximum injection time 50 ms; AGC 4 × 105). Determined charge states between 2 and 6 were required for sequencing and a 30 s dynamic exclusion window was used. Data-dependent ‘top 20’ MS2 scans were performed in 0.5 Da isolation window of the ion trap with collision-induced dissociation (CID) fragmentation (NCE 35%; maximum injection time 35 ms; AGC 1 × 104). MS3 quantification scans were performed using the multi-notch MS3-based TMT method (ten synchronous precursor selection (SPS) ions; resolution, 50,000; NCE, 55% for higher-energy collisional dissociation (HCD); the maximum injection time was 130 ms; AGC 1.5 × 105)12.

Data processing for protein identification and quantification

MS raw files were searched against the SwissProt human database (May 2020) with 20,329 entries using Proteome Discoverer software (version 2.4, Thermo Fisher Scientific). The search criteria were set to a mass tolerance of 10 ppm for MS data and 0.6 Da for MS/MS data with fixed modifications of carbamidomethylation of cysteine (+ 57.021 Da) and TMT on lysine residues and peptide N termini (+ 304.207 Da) and variable modification of methionine oxidation (+ 15.995 Da). The false discovery rate (FDR) was set at 0.01 for the identification of peptides and proteins. All the proteins were identified by two or more unique peptides. Reporter ion isotopic distribution correction was pursued in order to correct the factors for natural carbon isotopes and incomplete isotope incorporation. The reporter ion isotopic distributions obtained from the product data sheet (VI306840) of the TMTpro reagent were used as isotope correction factors in TMTpro 16plex method template in Proteome Discoverer software (version 2.4). Reporter ion quantification was performed with a mass tolerance of 20 ppm. While signal-to-noise ratio values of reporter ions were used for quantification of peptides, only spectra with an average reporter signal-to-noise ratio threshold of ≥ 10 across 16 TMTpro 16plex channels were considered for the quantification. The signal-to-noise ratio values of each reporter ion channel were summed across all quantified proteins and normalized so that the summed signal-to-noise ratio values of each channel could be equal across all 16 channels, which were called normalized abundance values. The normalized abundance values were first log-transformed and missing values were then replaced using values computed from the normal distribution with a width of 0.3 and a downshift of 1.8. Proteins exhibiting statistical significances among the psoriasis groups with and without CVD risk factors and a healthy control group were detected using one-way analysis of variance (ANOVA) comparison of the log2(normalized abundance) values using Perseus software (1.6.14.0)13. The proteins with a P-value < 0.05, from ANOVA analysis, were then further subjected to a Scheffe’s post-hoc analysis to identify those proteins that exhibited statistically significant changes between the psoriasis group with CVD risk factors and the psoriasis group without CVD risk factors (P-value < 0.05). For hierarchical clustering of the proteins with statistically significant changes (P-value < 0.05, > 2.0-fold) between psoriasis groups with and without CVD risk factors, log2(normalized abundance) values were first normalized using z-score and then clustering of both columns and rows was pursued based on Euclidean distance using the average linkage method using Perseus (1.6.14.0).

Gene Ontology (GO) enrichment analysis and ingenuity pathway analysis (IPA)

To identify functional implications and signaling pathways involved in proteins showing significant increases in psoriatic patients with CVD risk factors, we carried out Gene Ontology (GO) and Ingenuity Pathway Analysis (IPA). GO functional classifications of the proteins showing statistically significant increases (P-value < 0.05, > 2.0-fold) in the psoriasis group with CVD risk factors were analyzed using the DAVID software (http://david.abcc.ncifcrf.gov) to identify GO terms that were significantly enriched in the proteins. Additionally, IPA software (data version 76765844; QIAGEN, Redwood City, CA, USA) was used to analyze molecular and cellular functions, canonical pathways, and the associated network functions of the proteins exhibiting statistically significant increases in the psoriasis group with CVD risk factors.

Enzyme-linked immunosorbent assay (ELISA)

From the TMT-based proteomic data, caldesmon (CALD1), LIM and SH3 domain protein 1 (LASP1), myeloid cell nuclear differentiation antigen (MNDA), and zyxin (ZYX) showed statistically significant increases in the psoriasis group with CVD risk factors in compared to the group without CVD risk factors, and were selected for verification of ELISA using 42 serum samples consisting of 22 psoriasis samples from participants with CVD risk factors and 20 psoriasis samples from participants without CVD risk factors (see Supplementary Table S1 online for the validation cohort). The concentrations of CALD1 (Novus Biologicals, Centennial, CO, USA), LASP1 (Cusabio, Houston, TX, USA), MNDA (Mybiosource, Inc., San Diego, CA, USA), and ZYX (Cusabio, Houston, TX, USA) were determined using ELISA kits. The serum samples were diluted 1:3, 1:3, 1:2, and 1:2 for measurement of concentrations of CALD1, LASP1, MNDA, and ZYX, respectively (see Supplementary Fig. S1 online for standard curves for the measurement of concentrations of each protein).

Statistical analysis

Data were analyzed with using SPSS software version 25 (SPSS, Chicago, IL, USA). Differences in demographic data and baseline clinical findings among the three groups were assessed using the Kruskal–Wallis test. The results of proteomic analysis using LC/MS were compared using one-way ANOVA and post-hoc analysis was performed using pairwise multiple comparison procedures (Scheffe’s method) to evaluate differences between the psoriasis groups with and without CVD risk factors. Subsequent verification with ELISA were analyzed using the independent t-test and Mann–Whitney U-test.

Results

Baseline characteristics

Fourteen age- and sex-matched individuals were included in this proteomic analysis. The demographic and descriptive variables are shown in Table 1. All groups consisted of male participants with a mean age of 37.5 ± 6.43 years. The mean of Psoriasis Area Severity Index (PASI) score was not significantly different among the three groups (P-value = 0.23). In addition, the differences in body mass index (BMI) and disease duration among the three groups were not statistically significant (P-value = 0.16, P-value = 0.99, respectively).

Proteomic analysis of TMT-labeled samples

From the MS analysis of the TMT-labeled serum samples, a total of 1079 proteins were identified by two or more unique peptides and 1041 proteins of the identified proteins were quantifiable (see Supplementary Table S2 online). From one-way ANOVA analysis of the three groups consisting of the psoriasis groups with and without CVD risk factors and the healthy control group, 184 proteins were found to show statistical significance (P-value < 0.05) (see Supplementary Fig. S2online for hierarchical clustering analysis of the 184 proteins). Then, the 184 proteins with a P-value < 0.05 from the ANOVA analysis were subsequently subjected to Scheffe’s post-hoc analyses, respectively, for a psoriasis group with CVD risk factors and a healthy control group, a psoriasis group without CVD risk factors and a healthy control group, and the psoriasis groups with and without CVD risk factors to evaluate differences between the two groups (see Supplementary Table S3 online). From the results, 73, 3, and 143 proteins were, respectively, statistically significant (P-value < 0.05) from pairwise multiple comparisons between the psoriasis group with CVD risk factors and the healthy control group, between the psoriasis group without CVD risk factors and the healthy control group, and between the psoriasis groups with and without CVD risk factors. These results showed some similarity with the heat map exhibiting hierarchical clustering analysis of the 184 proteins as seen in Supplementary Fig. S2 because the psoriasis group without CVD risk factors and the healthy group did not reveal distinct expression patterns from the hierarchical clustering analysis. Since we originally attempted to identify those proteins that exhibited statistically significant changes between psoriasis groups with and without CVD risk factors, we primarily focused on the psoriasis groups with and without CVD risk factors for further analysis. While 98 proteins showed statistically significant changes (> 2.0-fold, P-value < 0.05) between the two groups, a hierarchical clustering analysis of the 98 proteins showed that most of the proteins with statistically significant changes increased in psoriasis group with CVD risk factors in comparison to that without CVD risk factors (Fig. 1). Actually, 90 proteins among 98 proteins showed significant increases in psoriasis group with CVD risk factors compared to that without CVD risk factors (Table 2).

Figure 1
figure 1

Heat map exhibiting hierarchical clustering of 98 proteins with statistically significant changes (> 2.0-fold, P-value < 0.05 from ANOVA followed by Scheffe’s post-hoc analysis) between psoriasis groups with and without cardio vascular disease (CVD) risk factors. The rows represent each protein and the columns show four and seven biological replicates of psoriasis serum samples without and with CVD risk factors, respectively. Hierarchical clustering of the 98 proteins was performed in Perseus software (1.6.14.0) on log-transformed normalized signal-to-noise ratio values after z-score normalization of the data.

Full size image
Table 2 Proteins showing statistically significant increases (> 2.0-fold, P-value < 0.05) from TMT-labeling based quantitative analyses of psoriasis groups with and without CVD risk factors.
Full size table

Over-represented functional processes and ingenuity network analysis

From the GO enrichment analysis of the 90 proteins showing statistically significant increases (> 2.0-fold) in the psoriasis group with CVD risk factors, platelet aggregation and platelet activation terms were enriched in the top five biological process terms (Fig. 2A). In addition, actin binding and integrin-binding terms were found in the top five molecular functions (Fig. 2B). In addition to GO enrichment analysis, integrin signaling was also found to be the top canonical pathway from the IPA analysis of the 90 proteins exhibiting significant increases in the psoriasis patients with CVD risk factors (see Supplementary Table S4 online). Actin cytoskeleton signaling that was included as one of the top five canonical pathways for the 90 proteins from the IPA analysis has been reported to be implicated in inflammation in cardiovascular disease14. The top five molecular and cellular functions in order of significance were; ‘cell to cell signaling and interaction, ‘cellular movement’, ‘cellular assembly and organization’, ‘cellular function and maintenance’, and ‘cellular morphology’. The top five canonical pathways were ‘integrin signaling, ‘remodeling of epithelial adherens junctions’, ‘GP6 signaling pathway’, ‘axonal guidance signaling’, and ‘actin cytoskeleton signaling’. The top three networks, respectively, containing 19, 18, and 18 proteins contained extracellular signal-regulated kinase (ERK), integrin/ERK1/2, and actin as the major hubs (Fig. 3).

Figure 2
figure 2

Top 5 biological process (BP) and molecular function (MF) terms (a and b), respectively, over-represented in proteins showing statistically significant increases (P-value < 0.05, > 2.0-fold) by Gene Enrichment (GO) analysis.

Full size image
Figure 3
figure 3

Top three interaction networks generated by ingenuity pathway analysis (IPA) software (data version 76765844) are displayed, respectively, in (a)–(c). Proteins showing statistically significant increases (P-value < 0.05, > 2.0-fold) in psoriasis patients with CVD risk factors are in red. ERK, integrin/ERK1/2, and actin are, respectively, are seen as the major hubs in each network. Straight lines indicate direct interactions, whereas dashed lines indicate indirect ones.

Full size image

Selection of biomarker candidates and subsequent verification using ELISA

To identify potential biomarker candidates for further verification, we used both mass spectrometric data and existing evidence in the current literature on biomarkers related to CVD. Regarding proteins that showed statistically significant increases (Scheffe’s test P-value < 0.05, > 2.0-fold) in the psoriasis group with CVD risk factors compared to those without CVD risk factors, we attempted to select novel biomarker candidates, which were studied in relation to CVD but their roles in psoriasis were not previously reported from literature search. Finally, four biomarker candidates including CALD1, LASP1, MNDA, and ZYX were selected for verification based on previous literature reviews15,16,17,18,19,20,21,22,23,24,25, the availability of suitable ELISA kits, and available sample volumes.

While all of four proteins (CALD1, LASP1, MNDA, and ZYX) were differentially expressed between psoriatic patients with and without CVD risk factors (Fig. 4 and Supplementary Table S5 online for raw data of ELISA), the concentrations of CALD1, MNDA, and ZYX were significantly higher in the psoriasis group with CVD risk factors than in the psoriasis group without CVD risk factors (P-value < 0.05).

Figure 4
figure 4

Boxplots of biomarker candidates verified by ELISA. The order of groups is identical in all charts as indicated on the bottom (CVD risk_O: psoriasis groups with CVD risk factors, CVD risk_X: psoriasis groups without CVD risk factors). The groups that show significant differences of the respective protein are indicated by brackets. Statistical significance was defined as *P < 0.05. (a) CALD1: Caldesmon, (b) LASP1: LIM and SH3 domain protein, (c) MNDA: Myeloid cell nuclear differentiation antigen, (d) ZYX: Zyxin.

Full size image

Discussion

Psoriasis is a common immune-mediated inflammatory disorder with an increased risk of CVD, which implies that chronic cutaneous inflammation expands to the systemic comorbidities26,27. In fact, systemic inflammatory processes could cause endothelial dysfunction, increased arterial thickness, vascular inflammation, and insulin resistance, which could lead to CVD in patients with psoriasis28,29,30,31. Early studies of biomarkers for psoriasis focused on underlying pathophysiologic processes and various circulating biomarkers for diagnosis, severity, and treatment response of in patients with psoriasis have been reported2,6,8,32. For example, serum levels of inflammatory and pro-inflammatory markers including C-reactive protein (CRP), complement 3 (C3), platelet P-selectin, tumor necrosis factor (TNF)-α, interleukin (IL)-6, IL-8, IL-12, and IL-18 were increased in patients with psoriasis9,10. Furthermore, S100 proteins such as S100A7, S100A8, and S100A9 are involved in abnormal keratinocyte differentiation and are upregulated in psoriatic skin lesion as well as in the blood2,33. Moreover, cytoskeletal and Ca2+-binding proteins such as thymosin-γ4, profilin-1, and talin-1 were upregulated in patients with psoriasis34. From our TMT-labeled quantitative analysis, 56 proteins were found to show statistically significant increases (P-value < 0.05, > 2.0-fold) in psoriasis group with/without CVD risk factors in comparison to healthy control group (see Supplementary Table S3 online). Among these proteins, well-known circulating markers for psoriasis such as P-selectin (P16109), S100A8 (P05109), S100A9 (P06702), and talin-1 (Q9Y490) were found to be upregulated in patients with psoriasis compared to healthy control.

Various potential biomarkers for CVD have been suggested with regard to myocardial injury or stress (cardiac troponins and natriuretic peptides), plaque instability (lipoprotein-associated phospholipase A2, matrix metalloproteinase-9), and adhesion molecules (VCAM-1, ICAM-1, and E-selectin)35,36,37,38,39. Also, general inflammatory biomarkers such as CRP, TNF-α, IL-6, and IL-8, systemic stress, and calcium homeostasis were regarded as important pathophysiological processes in CVD40,41. Additionally, many biomarkers associated with oxidative stress including isoprostanes, lipoprotein-associated phospholipase A2, coenzyme Q10, and glutathione, have been found40. However, there have been no practical biomarkers available to early predict the risk of CVD in psoriasis patients in real clinical settings. This study investigated a set of serum proteins related to CVD risk factors to identify those associated with subclinical cardiovascular events in the setting of psoriasis. Oxidative stress related proteins such as Peroxiredoxin-5 (P30044) and thioredoxin (P10599) were more upregulated in psoriasis with CVD risk factors than in psoriasis without CVD risk factors in the current study. The effects of oxidative stress and elevated antioxidant levels have already been reported in CVD-like atherosclerosis42,43.

Through TMT as one of isobaric labeling tags-based quantitative proteomic analysis and subsequent verification using ELISA, we especially identified three novel biomarkers associated with CVD risks in patients with psoriasis; CALD1, ZYX, and MNDA. All showed statistically significant differences between psoriatic patients with and those without CVD risk factors in ELISA experiments. In addition, GO and IPA analyses of the proteins showing statistically significant increases in psoriasis group with CVD risk factors in comparison to that without CVD risk factors revealed that biological processes of platelet aggregation/activation and molecular processes of integrin and actin bindings were overrepresented. Furthermore, integrin and actin were as the major hubs of signaling network analysis using IPA. As a group of cytoskeletal proteins, the main function of CALD1 and ZYX is actually a vital mediator in many biological pathways from cell to cell adhesion and migration to cellular proliferation44. Specifically, CALD1 is a calcium-calmodulin and actin-binding protein found most abundantly in smooth muscles, which is involved in stabilizing actin filaments, cellular proliferation, and migration15,45. Some studies have shown the possible relationship between CALD1 expression in tissue or blood and CVD15,16,17. ZYX is a focal adhesion protein that regulates actin cytoskeleton and shows intracellular distribution where actin filaments are highly concentrated46. As a stretch-sensitive regulator, ZYX is known to be stimulated by prolonged hypertension, an important predisposing factor in the development of endothelial dysfunctions, which subsequently induces a broad range of inflammatory responses23,24. Regarding molecular functions, ZYX transmits external forces like hypertension to focal adhesion proteins, enhances actin assembly, and subsequently interacts with integrins, which are associated with multiple adaptor proteins in various signaling pathways25. The functions and the associations of the cytoskeletal proteins with CVD risks in psoriatic patients have not been clarified; however, those proteins could be associated with CVD risks in psoriatic patients based on our current findings that molecular functions and interaction networks related to actin/integrin binding were overrepresented in psoriatic patients with CVD risk factors. In particular, CALD1 was closely related to actin, the main hub of the third network in our IPA results, which corresponds to the main function of CALD1.

Besides, platelet was included in the main biological processes associated with CVD in patients with psoriasis by GO analysis. In both psoriasis and CVD, platelets are actually known to play a critical role in the pathogenesis. In coronary artery diseases, increased platelet activation or aggregation is known to be associated with the risk factors including smoking, hypertension, and hypercholesterolemia47,48,49. Furthermore, platelets also act as immune cells that initiate and regulate inflammatory processes, which could be related to increased risks of cardiovascular events48. The hemostatic imbalance as well as immunologic factors are important in prothrombotic circumstances, which might be induced by platelet activation or aggregation50. While integrin binding was included in the top five molecular function terms, integrin signaling was found to be related to cardiovascular diseases such as atherosclerosis via MAP kinase signaling and cardiac hypertrophy via Rho GTPase signaling51,52. Specifically, a dimer of integrin alpha-IIb (P08514)/integrin beta-3 (P05106), which showed significant increases (> 2.0-fold) in psoriasis patient with CVD risk factors in our current study (Table 2), is known to be activated through binding of fibrinogen followed by platelet aggregation51,53. ERK signaling was also reported to be associated with the cardiac hypertrophy54 and the relationship ERK and MAP kinase was identified in the first network in our IPA results.

MNDA has been proposed to be a transcriptional factor in myeloid cells, and it plays a role in both differentiation or apoptosis of myeloid cells and inflammatory pathways55,56. In particular, MNDA is related to the neutrophil death program and proper elimination of neutrophils is required for the resolution of inflammation55. In psoriatic skin, abundant neutrophils including Munro’s microabscesses are important histopathologic hallmarks. Circulating neutrophils are collected in skin lesions and activate subsequent inflammatory responses such as reactive oxygen species (ROS) production, T cell stimulations, and angiogenesis57. Also, the neutrophil-to-lymphocyte ratio was increased in patients with psoriasis58. Similar to psoriasis, neutrophils have been recognized as important contributors in inflammatory processes to inflammatory processes in CVD. Previous studies suggested there was an association between neutrophils and the risk of CVD21,22,59,60,61. Moreover, it was reported that neutrophils are involved in the early atherosclerotic plaque formation62. Taken together, neutrophils may be involved in early damage of both the epidermis and the endothelium. Also, a prior proteomic study on the spondyloarthropathy implicated MNDA as a meaningful biomarker to detect inflammatory responses earlier than other markers such as IL-6 and CRP63. Elevated MNDA upregulation could implicate the development of inflammatory processes in various disorders where neutrophils especially play an important role.

Our study has some limitations. First, we had a limited sample size, which could preclude the generalizability of our results. While we attempted to include a similar number of serum samples in the three groups, there was a limit to collecting an equal number of samples when we selected the samples for proteomic analysis. Despite these limitations, our study has several strengths. First, we attempted to divide the patients with psoriasis into two groups according to the strict diagnosis of cardiovascular risk factors and attempted to analyze the differences in protein expression levels of each serum samples. Second, isobaric labeling mass tags-based quantitative analysis in the current study allowed for simultaneous quantification of several samples due to multiplexing capability, resulting in reduction of instrument time and inter-sample variation compared to a label-free quantification method64,65. Third, further verification using ELISA for biomarker candidates specific to psoriatic patients with CVD risk factors was performed.

So far, there are no independent biomarkers to predict cardiovascular risk in patients with psoriasis. In this study, we found that differences in the expression of CALD1, ZYX, and MNDA between psoriasis with cardiovascular risk factors and those without cardiovascular risk factors were significant. We expect these biomarker candidates can be used to predict CVD risk in psoriasis in future clinical settings. The results require further external validation and may currently be viewed as a hypothesis requiring future research. The next steps are to confirm the value of the identified proteins in a large size of subjects.

In conclusion, we found that CALD1, ZYX, and MNDA showed significant discrimination between psoriatic patients with and without CVD risk factors. Several pathways have been associated with CVD risk factors in patients with psoriasis, including ‘integrin signaling’ and ‘actin cytoskeleton signaling’. The role of integrin signaling could be related to actin cytoskeleton signaling via ZYX, with regard to mechanoreceptor and subsequent gene regulation. Additionally, platelet aggregation with integrin signaling may be associated with prothrombotic events or inflammatory processes in CVD. Further studies should be performed to evaluate using larger sizes of subjects whether these biomarkers will have practical clinical applications for the early diagnosis and management of psoriasis in patients with CVD risk factors.