Comprehensive Proteomic Analysis Reveals Intermediate Stage of Non-Lesional Psoriatic Skin and Points out the Importance of Proteins Outside this Trend

To better understand the pathomechanism of psoriasis, a comparative proteomic analysis was performed with non-lesional and lesional skin from psoriasis patients and skin from healthy individuals. Strikingly, 79.9% of the proteins that were differentially expressed in lesional and healthy skin exhibited expression levels in non-lesional skin that were within twofold of the levels observed in healthy and lesional skin, suggesting that non-lesional skin represents an intermediate stage. Proteins outside this trend were categorized into three groups: I. proteins in non-lesional skin exhibiting expression similar to lesional skin, which might be predisposing factors (i.e., CSE1L, GART, MYO18A and UGDH); II. proteins that were differentially expressed in non-lesional and lesional skin but not in healthy and lesional skin, which might be non-lesional characteristic alteration (i.e., CHCHD6, CHMP5, FLOT2, ITGA7, LEMD2, NOP56, PLVAP and RRAS); and III. proteins with contrasting differential expression in non-lesional and lesional skin compared to healthy skin, which might contribute to maintaining the non-lesional state (i.e., ITGA7, ITGA8, PLVAP, PSAPL1, SMARCA5 and XP32). Finally, proteins differentially expressed in lesions may indicate increased sensitivity to stimuli, peripheral nervous system alterations, furthermore MYBBP1A and PRKDC were identified as potential regulators of key pathomechanisms, including stress and immune response, proliferation and differentiation.

To date, all therapies available for psoriasis only manage symptoms. Understanding alterations that cause the disease is highly important for developing new therapies to better manage the disease.
Our skin connects, and at the same time separates internal the external environment. It is constantly subjected to many different stimuli that requires proper response, through which the skin can influences the function of other organs, like the brain and the endocrine system in a mutual way 1,2 . In psoriasis, the macroscopically healthy looking non-lesional skin harbors alterations that might cause symptoms 3 . One of the most characteristic properties of non-lesional skin is an altered response to mechanical stress or injury 4 leading to barrier disruption 5 , which leads to an elevated innate immune response 6,7 . Alterations in non-lesional skin are not restricted to keratinocytes. Angiogenesis is also among those mechanisms that is already affected in non-lesional skin, resulting in altered quantity and quality of microvessels 8 . In addition, it is becoming clear that some adaptive immune responses are also altered 9 . Abnormalities in the dermal extracellular matrix composition -such as elevated expression of the oncofetal splice variant of fibronectin 10 , due to altered splicing events 11 indicate the involvement of dermal fibroblasts 3 . Several matrix metalloproteinases (MMPs), such as MMP-9, previously thought to be increased only in lesions, are now known to be elevated in non-lesional skin compared to healthy skin 12 . There is also evidence for mechanisms in non-lesional skin that contribute to the maintenance of its state. The PRINS long non-coding RNA is induced by stress and nucleic acids, and it is anticipated to have a protective

Results
Proteomic workflow and information on involved donors. Three biological replicas of our proteomic approach were performed following sequential protein extraction of total skin biopsies. Each proteomic replica contained samples from three healthy donors as well as non-lesional and lesional biopsies from three psoriatic patients. The schematic overview of the applied proteomic strategy is summarized in Fig. 1 (also see Supplementary Information: Materials and Methods), and basic demographic and clinical characteristics of psoriatic patients and healthy donors are listed in Table 1 Biological processes associated with differential expression in healthy and lesional skin. As an initial step, proteomic results of lesional and healthy skin samples were compared and the relative abundance of 249 proteins was found to be different ( Fig. 2a and Supplementary Table 1). A protein-protein interaction-based enrichment analysis was performed with these proteins. We screened for interaction networks and biological processes related to the observed differences in expression using Gene Ontology (GO) analysis of the STRING database (version 10.5). Based on the GO nomenclature and protein composition, the identified biological processes could be classified into the following categories: development, proliferation, regulation of expression and response to stimulus related processes. The ten most significantly different biological processes of each category are listed in Fig. 2b,c and Supplementary Table 2.
Since the major characteristics of psoriatic alterations include altered stress and immune responses as well as dysregulation of proliferation and differentiation, we screened among proteins expressed differentially in lesion compared to healthy skin for central regulators participating in all four of these mechanisms (Fig. 2d). As a result, four central proteins -MYBBP1A, PML, PRKDC and STAT1 -were identified (Fig. 2e).
Differential protein expression in non-lesional and lesional skin and the biological processes associated with these proteins. Comparison of non-lesional and lesional skin proteomes led to the identification of 56 proteins exhibiting at least 2-fold differences in relative abundances. Of these proteins, 32 exhibited higher protein abundance in non-lesional skin compared to lesions, whereas 24 exhibited lower abundance (Fig. 3a and Supplementary Table 3). Functional enrichment analysis of these 56 proteins revealed several biological processes identified in psoriasis pathomechanism, including development, and response to stimulus (Fig. 3b  and Supplementary Table 4).
We also found a subset of proteins to be differentially expressed in non-lesional and lesional skin that were not differentially expressed in healthy skin and lesions (Fig. 3c).
The levels of eight proteins were greater in non-lesional skin and lower in lesional skin compared to the levels in healthy skin (non-lesional < healthy < lesional), and one protein exhibited the opposite trend (non-lesional > healthy > lesional). Although the non-lesional and lesional differences in the abundance of these proteins were not statistically significant when compared to healthy skin, the difference in abundance between non-lesional and lesional samples differed significantly by more than two-fold (Fig. 3d).
We also identified 44 proteins that had altered expression only in the comparison of lesional skin to either non-lesional or healthy skin; it is anticipated that these proteins play a role in manifestation and/or maintenance of lesions. The results of a computer-aided, keyword-based literature search suggests that, of these 44 proteins, 23 are already associated with the disease (Fig. 3e), whereas 21 have not yet been associated with psoriasis pathogenesis (Fig. 3f).
Proteins without previous association to psoriasis that exhibited decreased expression in lesions compared to expression in to both non-lesional and healthy skin include modulators of apoptosis, signaling, endothelial cell ( Fig. 4a). Among these, the relative amount of four proteins (GART, CSE1L, GBP1 and UGDH) was similar in the non-lesional and lesional skin samples. Out of the eight proteins that are differentially expressed in non-lesional skin compared to healthy GBP1, KLK10 and S100A7 have already been associated with psoriasis pathogenesis; the other five are potential novel, early markers of the disease.
To verify our proteomic results, immunofluorescent staining was performed to gain additional information regarding protein localization, deposition and distribution. UGDH had the largest expression differences in non-lesional and healthy skin. As UGDH has not been linked to psoriasis previously, this protein was chosen for further analysis. UGDH staining showed similar epidermal distribution in all three sample types, with the highest protein levels detected in basal keratinocytes (n = 10 different individuals in each group, listed in Supplementary  Table 6). Despite the similarities in the UGDH localization pattern, clear differences in staining intensities were observed. The non-lesional and lesional psoriatic samples displayed more robust intensities compared to that of healthy samples, confirming our proteomic results (Fig. 4b,c).
To determine which lesional alterations and to what extent are manifest in non-lesional skin, we selected the 249 proteins that exhibited differential expression in healthy and lesional skin and their expression levels was compared to those in non-lesional skin. In non-lesional skin, the expression of 199 (79.9%) of the proteins differed from the expression in healthy and lesional skin by less than two-fold. Therefore, this category was termed as intermediate, as they may represent a discrete step in the healthy-to-lesional transition (Supplementary Table 7). psoriatic biomarkers, biological functions, canonical pathways and annotation of diseases associated with the detected alterations in protein amounts. To examine the validity of our experimental approach, we further screened our proteomic dataset (lesional vs. healthy) for known, major biomarkers characteristic for psoriasis. These were identified previously in by large scale genomic, transcriptomic and/or proteomic studies. Out of these biomarkers AKR1B10, CSTA, FABP5, PI3, SCCA2, STAT1, STAT3 and members of the S100 family, including S100A7, S100A8, S100A9 were also found in our study. These molecules exhibited elevated expression levels in psoriatic lesions, compared to healthy control skin ( Table 2).
Further analysis was performed to identify the cellular mechanisms that may be associated with the proteins that were detected in altered amounts in a proteomic approach, using the Ingenuity Pathway Analysis software (IPA). Diseases annotation revealed 'psoriasis' as the first hit when lesional and healthy (Table 3), or lesional and non-lesional differences (Table 3) were compared.
Annotation of biological functions by IPA highlighted 'initiation of protein translation' (Table 3) and 'killing of Staphylococcus aureus' as the main functions likely to be affected, respectively. Ingenuity canonical pathway screening identified the 'role of IL-17A in psoriasis' among the top ten most significant canonical pathways, when either lesional, or non-lesional protein expression was compared to healthy samples (Table 4). In addition, several cancer, neurological and neuromuscular canonical pathways were also highlighted.

Discussion
To expand knowledge about the pathomechanism of psoriasis, many extensive, large-scale comparative proteomic approaches have been performed 17,24,26 . However, the comparison of healthy, non-lesional and lesional skin at the proteomic level has been missing from these studies. To fill this gap, our comparative proteomic analysis included healthy skin as well as non-lesional and lesional psoriatic samples.      www.nature.com/scientificreports www.nature.com/scientificreports/ In order to check the validity of our proteomic approach we compared major known psoriatic biomarkers published in previous genetic (genome-wide association studies) 21 , transcriptomic 19,21,23 and proteomic studies 17,18,20,22 with our proteomic dataset. Known psoriatic lesional biomarkers also found in our study includes AKR1B10 16 , CSTA 17 , FABP5 18 , PI3 19 , SCCA2 20 , STAT1 16 , STAT3 21 , S100A7 18-20 , S100A8 19,22 and S100A9 19,20,22,23 . Moreover, annotation of diseases resulted in the identification of psoriasis with the strongest correlation based on differentially expressed proteins in either lesional vs. healthy or in lesional vs. non-lesional comparison. Canonical pathway analysis of non-lesional differences compared to healthy skin resulted in the identification of 'Role of IL-17A in Psoriasis' . However, these annotations also highlighted cancer, neurological, neuromuscular or muscular disease-related mechanism, suggesting their potential involvement in disease pathomechanism, or some similarities between these diseases.
Since our proteomic and in silico analysis cannot distinguish between cell-types, and provide information whether mechanistically linked alterations are taking place within the same, or different cell types, further experiments are required in this direction to clarify the exact relevance of these predicted connections to psoriasis pathomechanism.
We performed literature a search for known functions of proteins found to be altered in amounts in our study to suggest mechanism through which they may potentially participate in the pathomechanism of the disease. The detected differences in the expression of proteins in healthy and lesional skin highlighted involvement in psoriasis of cell proliferation 28 , development 29 , response to stimulus 30 , expression 31 related processes. In the comparison of non-lesional and lesional skin, we identified 56 proteins with differential expression, which represents only 22.5% of the number of proteins which showed altered expression in the comparison of healthy and lesional skin (56 vs. 249). This highlights the importance of studying healthy skin in comparisons using patient samples for pinpointing disease-associated alterations. Qualitative literature-based analysis of these 56 proteins led to the identification of several mechanism for which association with psoriasis has already been described, including processes related to development 29 , response to stimulus 26 and expression 31 .
Further analysis focused on gaining insight about the extent to which alterations are manifest in lesions and in non-lesional skin. Strikingly, nearly 80% of the 249 proteins exhibiting differential expression in lesional and healthy skin exhibited an intermediate expression level in the non-lesional skin, suggesting the possible presence of early, lesional-like alterations in non-lesional skin. Divergence from this trend was only observed for two small protein groups. Ten proteins -CHCHD6, CHMP5, COLEC12, FLOT2, ITGA7, LEMD2, NOP56, PLVAP, RRAS and SMARCA5 -differed in relative protein amounts in non-lesional and lesional skin, but the amounts of these proteins were similar in healthy and lesional samples. These ten proteins are likely to represent a group of non-lesional characteristic alteration. For nine proteins -CD207, COLEC12, CTSV, ITGA7, ITGA8, PLVAP, PSAPL1, SMARCA5 and XP32 -the direction of the expressional changes was different in non-lesional and lesional samples compared to healthy skin and might represent proteins that contribute to maintaining the non-lesional state.
Next, with the proteins in these two groups, we performed an extensive literature search to suggest potential mechanisms by which they may influence disease pathogenesis. Interestingly, all the identified proteins may play a role in signaling at different levels starting from the cell surface all away to the nucleus or mitochondria. The identified cell surface receptors include two integrins (ITGA7 and ITGA8) that are important in external signal recognition. Decreased ITGA7 levels -as observed in lesional vs. non-lesional skin -could be associated with delayed autophagy 32 , differentiation 33 and increased migration 34 , all known to be affected in psoriatic lesions. In contrast, elevation of ITGA7 may induce growth suppression 35 . However, ITGA7 is characteristically expressed mainly by smooth muscle cells 36 in the skin, suggesting that its involvement in keratinocyte related events are at least limited, or none. Instead, may suggest alterations in (vascular) smooth muscle cell adhesion-related processes 37 . Alternatively, ITGA7 may influence neurite outgrowth 38 . Therefore, further studies are required to confirm the observed TGA7 expression alteration and to identify the cell-types of source. Another identified cell-surface molecule, MYO18A, through recognizing microorganism lipopolysaccharides, may increase innate immune responses 39 promoting cytokine production towards Th1 direction 40 ; known to be important in psoriasis 41 .  43 and COLEC12 44 participate in endocytic pathways, regulating the levels of cell-surface receptors and, thereby, signaling. FLOT2 is a known component of the raft microdomain complex that represents the major unit regulating STAT signaling pathways according to the raft-STAT signaling hypothesis 45 . The alteration of FLOT2 expression could suggest a high relevance since STAT3 is a key regulator in psoriasis 46 . Reduction of the scavenger receptor COLEC12 could trigger psoriasis by trastuzumab treatment 47 . Moreover, COLEC12 may influence the mitochondrial respiratory chain 48 , and this property is in agreement with the decreased level of mitochondrial MICOS complex subunit CHCHD6 49 observed in lesions compared to non-lesional skin. CHCHD6 regulates oxygen consumption and thereby may influence cell growth 50 . Reduced levels of CHCHD6 were shown previously to lead to a shift from oxidative metabolism to glycolytic metabolism 51 that negatively influences keratinocyte differentiation 52 , and both types of mechanisms are known to be affected in psoriasis 53,54 .
The altered expression of LEMD2 may suggest that signal transduction is also altered at the level of the nucleus. LEMD2, located in the inner nuclear membrane, regulates nuclear import/export processes 55 and thereby intranuclear signaling 50 . During this regulation, LEMD2 is associated with the same complex as CHMP5, which was also identified in our studies. In the nucleus, the STAT-regulated protein NOP56 56 , a core protein of the box C/D small nucleolar ribonucleoprotein (snoRNP) complex, participates in the biogenesis of rRNAs 57 . Increased rRNA biogenesis is suggested to be necessary for high proliferation rate 58 , a process that is crucial for the development of psoriatic lesions.
Abnormal proliferation 28 , differentiation 52 and, thereby, skin barrier function are key processes during psoriatic plaque formation. SMARCA5 is a component of the nucleosome remodeling factor complex 59 . Decreasing SMARCA5 levels are required for basal keratinocytes to shift form proliferation toward differentiation 60 . XP32 is also a component of the epidermal differentiation complex 61 and associated with skin barrier function 62 . The observed contrasting expressional differences of these two proteins in the non-lesional and lesional skin may contribute to our understanding of lesion formation and how non-lesional skin maintains its state.
Overall, our results indicate that dysregulation of cellular signaling -from signal detection, through endocytosis of receptors and transduction of signal from the cell surface to the nucleus -may be affected during the disease. The alteration of these systems is likely to lead to increased reaction to external signals that could contribute to the maintenance of psoriatic plaques.
By comparing non-lesional and healthy skin, differential expression was observed for eight proteins (CSE1L, GART, GBP1, KLK10, MYO18A, S1007A, SMARCA5 and UGDH). Four of these proteins (CSE1L, GART, GBP1, UGDH) might be predisposing factors, as their expression was similar in non-lesional and lesional skin, and their significance would have been missed in comparisons for which healthy samples were not included. Of these, UGDH was detected with the highest relative difference. UGDH has not been highlighted previously in association with psoriasis. We therefore decided to analyze it further. Immunohistochemical analysis confirmed our proteomic results: higher UGDH levels were found in non-lesional and lesional skin compared to healthy skin, that was mainly associated with keratinocytes. Elevated UGDH levels may increase chondrocytes proliferation indirectly, likely through increased hyaluronan production that binds different cytokines 63 . However, in vitro downregulation of UGDH and consequently decreased hyaluronan amounts did not influence keratinocyte proliferation 64 . These results are in line with our observation, suggesting that elevated UGDH levels observed in non-lesional keratinocytes are not sufficient to modify their proliferation.

Categories (L vs. H)
Diseases or disease related processes p-value Predicted Activation State Activation z-score www.nature.com/scientificreports www.nature.com/scientificreports/ Proteins for which expression was affected only in lesions are often considered "trigger" proteins, as changes in the expression of these proteins are linked to the shift of the disease state. The proteins that have not previously been associated with psoriasis were categorized into two groups. The first group of proteins might contribute to the mechanosensitivity of the tissue (SGCD 65 , SYNM 66 , MYH11 67 , ATP1B1 68 ). The second group functions within the nervous system (MPZ, PRX 69 , CSPG4 70 , CNTN1 and ITGA8 71 , ATP1B1 72 ), could suggest the involvement of the peripheral nerve system in psoriasis 73 .

Biological function annotation of protein expressional differences between non-lesional (NL) and lesional (L) skin
Finally, we searched for potentially central proteins in disease pathogenesis participating in key mechanisms of psoriasis including regulation of stress and immune response, proliferation and differentiation. Some of the identified proteins, such as PML 74 and STAT1 75 , have already been linked to psoriasis. We also identified two proteins, PRKDC and MYBBP1A, which have not previously been highlighted in context with the disease. The PRKDC may plays a role in the detection and repair of breaks in double-stranded DNA 76 and mediates the phosphorylation of c-MYC 77 and p53 78 suggesting a potentially important role in psoriasis. The suggested altered expression by our results of the transcription factor MYBBP1A may also be among the potentially important proteins in psoriasis implicated in the pathogenesis since it functions as a co-repressor of NF-κB that may regulate responses to stress and cytokines 79 .
Taken together, our comparative proteomic approach of healthy, non-lesional and lesional skin led to the identification of various proteins which may function in psoriasis pathogenesis, providing a strong base for future studies. Proteins exhibiting opposite expression changes in lesional and non-lesional samples compared to healthy skin may function in the maintenance of the non-lesional stage and may represent future targets for therapeutic purposes.  criteria for inclusion of patients in the study and skin sample collection. To identify alterations that are general in chronic plaque psoriasis and keep the number of volunteers for skin biopsy collection to a minimum, for our proteomic approach, we randomly engaged individuals (I) of different age to minimize possible age-related differences; (II) with various Psoriasis Area Severity Index (PASI) scores between 5 and 25, since the score for an individual patient varies over time and with relapse; (III) of both genders to avoid possible gender-associated differences; and (IV) with both early and late onset. A total of 9 (3 × 3) patients suffering from chronic plaque psoriasis and the same number of healthy donors were involved in our study. The data of individuals involved in the study are summarized in Table 1. All psoriatic patients had not received any kind of treatment for the condition for at least 6 months. The 6 mm skin punch biopsies containing the epidermis and the dermis were collected from an area of the upper-middle gluteal region that is not exposed to sunlight. Both lesional and non-lesional samples were collected from patients. Non-lesional samples were taken at least 7 cm from the edge of the lesion subjected for biopsy. The presence of psoriasis was clinically verified for all patients, and clinical as well as demographic data of the donors are presented in Table 1.
comprehensive and comparative proteomics of healthy, non-lesional and lesional skin. Sample preparation from skin biopsy and sequential protein extraction. Samples were cut with a razor blade. Skin proteins were extracted sequentially in four consecutive solubility-based extraction steps. Extraction buffers were used in increasing order of their solubilizing properties for a better separation of proteins. Samples were initially incubated in extraction buffer I. (0.15 M NaCl, 50 mM Tris-HCl, pH 7.4) for 24 h at 4 °C in the presence of protease inhibitors. Protein extracts were then clarified by centrifugation and separated from the pellet. This step was repeated by resuspending the pellet in extraction buffer II, which contained 1 M NaCl, 25 mM EDTA, 50 mM Tris-HCl, pH 7.4. Following extraction with 250 mM SDS-containing extraction buffer III (8 h at room temperature), guanidine hydrochloride containing extraction buffer IV (4 M GuHCl, 10 mM EDTA, 50 mM Tris-HCl, pH 7.4) was applied for 48 h at 4 °C. The same protein extracts of the three donors were pooled in each investigated group (healthy, non-lesional, lesional). Extraction procedure was carried out three times and each contained extracted proteins of three donors following the pooling of the samples which were than subjected for downstream proteomic analysis.
Protein identification by 2D LC-MSMS. A total 35 µg protein from each sample was applied for mass spectrometry analysis. A modified filter-aided sample preparation method was used for tryptic digestion of the protein extracts 80 . High-pH reversed-phase chromatography was performed on a C18 column (Phenomenex, Kinetex 5 µ EVO C18 100 A, 2.1 × 100 mm; cat. no. 00D-4622-AN, flow rate: 150 µl/min). Forty-eight fractions were collected from 1 to 25 minutes (half minute/fraction) and 4-4 fractions were combined (1,13,24,37; 2,14,25,38 and so on) to get 12 final fractions. Each fraction was subjected to nano LC-MSMS analysis on an Orbitrap Elite hybrid mass spectrometer (Thermo) coupled with a Waters nanoAcquity UPLC system, using a gradient elution after trapping the samples onto the trap column. Data-dependent analyses were applied; the 20 most intense peaks were selected for ion-trap collision-induced dissociation after each survey scan measured in the Orbitrap. Proteome Discoverer (ver.: 1.3) was used to generate MS/MS peak-list files and our in-cloud ProteinProspector (ver.: 5.16.0) database search engine was used for protein identification against the human sequences from the UniProtKB.2015.12.14. random.concat (149781/55820795 entries searched) database. Detailed protocols and applied counting for semiquantitative analysis is described as Supplementary Information.
Immunofluorescence staining of skin sections for UGDH. For immunofluorescence analysis, 5 µm sections of frozen embedded skin biopsies from psoriatic patients (non-lesional and lesional skin) and healthy individuals were used. After fixation and permeabilization (Foxp3 staining buffer set, fixation/permeabilization kit, Miltenyi Biotec, used according to the description of the manufacturer), samples were blocked in Tris-buffered saline (TBS) containing 1% bovine serum albumin (BSA) and 1% normal goat serum (NGS) for 1 h at room temperature. Samples were incubated overnight at 4 °C in TBS with 1% NGS and primary antibodies against UGDH (rabbit polyclonal antibody, ab155005, Abcam), diluted to 1:100. Following washing in TBS, AF546 secondary antibodies (Goat anti-Rabbit IgG (H + L) Highly Cross-Adsorbed Secondary Antibody, Alexa Fluor 546, A-11035, Invitrogen), diluted in TBS containing 1% NGS to 1:500, were applied for 1 h at room temperature.
Literature search to identify novel psoriasis-associated proteins. To identify proteins not yet linked with the pathomechanism of psoriasis, literature mining was carried out using protein names or the encoding gene's HUGO Gene Nomenclature Committee (HGNC) symbol(s), applying the following strategy: each protein or gene name was searched together with "psoriasis" as a keyword using the RISmed R package.

Statistical analysis.
To compare protein abundance from healthy, lesional and non-lesional skin extracts, significant differences were determined based on relative peptide ion chromatograms and spectrum counting and evaluated using two different approaches: (1) modified t-test (limma) and (2) rank product test (as described by Schwämmle et al. 81 ) following t-test. We considered a protein amount to be different between two samples if at least one of the three tests were significant (test <0.05) and the absolute fold change was at least two or higher.