- Email: [email protected]
Deciphering psoriasis. A bioinformatic approach
G Model DESC 3284 No. of Pages 7
Journal of Dermatological Science xxx (2017) xxx–xxx
Contents lists available at ScienceDirect
Journal of Dermatological Science journal homepage: www.jdsjournal.com
Deciphering psoriasis. A bioinformatic approach Juan L. Melero, Sergi Andrades, Lluís Arola, Antoni Romeu* Department of Biochemistry and Biotechnology, Rovira i Virgili University, Tarragona, Spain
A R T I C L E I N F O
A B S T R A C T
Article history: Received 21 July 2017 Received in revised form 25 September 2017 Accepted 18 November 2017
Background: Psoriasis is an immune-mediated, inflammatory and hyperproliferative disease of the skin and joints. The cause of psoriasis is still unknown. The fundamental feature of the disease is the hyperproliferation of keratinocytes and the recruitment of cells from the immune system in the region of the affected skin, which leads to deregulation of many well-known gene expressions. Objective: Based on data mining and bioinformatic scripting, here we show a new dimension of the effect of psoriasis at the genomic level. Methods: Using our own pipeline of scripts in Perl and MySql and based on the freely available NCBI Gene Expression Omnibus (GEO) database: DataSet Record GDS4602 (Series GSE13355), we explore the extent of the effect of psoriasis on gene expression in the affected tissue. Results: We give greater insight into the effects of psoriasis on the up-regulation of some genes in the cell cycle (CCNB1, CCNA2, CCNE2, CDK1) or the dynamin system (GBPs, MXs, MFN1), as well as the downregulation of typical antioxidant genes (catalase, CAT; superoxide dismutases, SOD1-3; and glutathione reductase, GSR). We also provide a complete list of the human genes and how they respond in a state of psoriasis. Conclusion: Our results show that psoriasis affects all chromosomes and many biological functions. If we further consider the stable and mitotically inheritable character of the psoriasis phenotype, and the influence of environmental factors, then it seems that psoriasis has an epigenetic origin. This fit well with the strong hereditary character of the disease as well as its complex genetic background. © 2017 Japanese Society for Investigative Dermatology. Published by Elsevier Ireland Ltd. All rights reserved.
Keywords: Psoriasis Keratinocyte Gene expression Epigenetics Bioinformatics
What’s already known about this topic? Psoriasis influences a wide variety of genes with a wide range of biological functions. Psoriasis is also known to have a strong hereditary character and a complex genetic basis.
Combining the strong hereditary character of the psoriasis with the results we show here, it appears that psoriasis could be a disease of epigenetic origin.
1. Introduction What does this study add? Cyclins B1, A2, E2, CDK1, and B-type lamin genes are highly expressed in psoriatic tissue. Dynamin system is up-regulated in psoriasis. Catalase, superoxide dismutase and glutathione reductase a down-regulated in psoriasis. Psoriasis is associated with kallikreins. The effect of psoriasis extends across all regions of the chromosomes of the human genome. This effect on gene expression affects not only a single gene but also a gene cluster.
* Corresponding author. E-mail address: [email protected] (A. Romeu).
Psoriasis is a chronic autoimmune skin disease [1,2]. Although the cause of the disease is still unknown, psoriasis is a phenotype of skin biochemical and immune disorder. Some hereditary factors predispose to psoriasis, which has been linked with at least nine chromosomal loci (PSORS1 through PSORS9). The molecular mechanisms responsible for psoriasis still have to be fully elucidated. The disease is multifactorial and noteworthy involves the hyperproliferation of keratinocytes in the epidermis. Various reviews have been published of the advances in our understanding of psoriasis [3–5]. Gene expression, although representing a specific site of regulation, is only one step in the complex cascade from an initial stimulus to a final phenotype. In this context, we were keen to explore the extent of the effect of psoriasis on gene expression in
https://doi.org/10.1016/j.jdermsci.2017.11.010 0923-1811/ © 2017 Japanese Society for Investigative Dermatology. Published by Elsevier Ireland Ltd. All rights reserved.
Please cite this article in press as: J.L. Melero, et al., Deciphering psoriasis. A bioinformatic approach, J Dermatol Sci (2017), https://doi.org/ 10.1016/j.jdermsci.2017.11.010
G Model DESC 3284 No. of Pages 7
2
J.L. Melero et al. / Journal of Dermatological Science xxx (2017) xxx–xxx
the affected tissue. First, at a quantitative level, that is, how many genes there are (or can be) that in a healthy state are silenced and awakened by the disease. And, vice versa, how many genes are highly expressed in a healthy state but totally silenced in psoriasis. Second, at a more qualitative level, we were interested in studying the extent of psoriasis in functional genomics, as well as the regions of the genome that are affected. To do this, we used freely available data from the NCBI-Gene Expression Omnibus (GEO) database of whole human genome arrays with paired lesional and non-lesional skins from psoriasis patients. Using this data, we made an analysis of variance to identify differences in gene expression profiles among groups. From this standpoint, then, this study uses a bioinformatic approach to data mining. 2. Methodology The sources of information were the current version of the Human Genome (NCBI, GRCh37.p13) and the freely available NCBI Gene Expression Omnibus (GEO) database: DataSet Record GDS4602 (Series GSE13355) based on an analysis of skin punch biopsies from lesional skin and non-lesional skin from a group of 58 psoriatic patients. The platform used in these whole Human Genome arrays was the Affymetrix GPL570 (HG-U133_Plus_2, Affymetrix Human Genome U133 Plus 2.0 Array). For this study, we created our own pipeline of scripts in Perl and MySql for data management. The rationale of this work is based on the following tasks: 2.1. Task 1. Samples download In accordance with the experimental design of the GDS4602 record, gene expression profiles in uninvolved and involved skin from 58 psoriasis affected individuals were characterised. We used these two array groups. Each array (or sample) record consists of a text file of 54,675 rows, in which each “probe” has its “probe-RNA hybridization signal”. These values are considered arbitrary units. We downloaded the 116 array (sample) records. 2.2. Task 2. Probe selection The GPL570 platform has a 54,675 Affymetrix probe set. However, we only selected the probes whose associated Entrez_id number and gene symbol match those annotated in the current Human Genome release. A total of 46,002 probes were selected. Some of the probes in this platform are designed to recognize only transcripts (ending with ‘_at’) and while others are designed to recognize multiple transcripts from the same gene family (ending with ‘_s_at’ or ‘_a_at’). In order to further adjust the analysis of gene expression, here we have only considered the transcript-specific probes. On the other hand, with respect to the genes in the GPL570 platform, 19,873 had a match in the Human Genome release. Of these, 12,927 genes are associated with a single probe, 4387 are associated to two probes, 1611 genes are associated to three probes, 599 genes are associated to four probes, etc. 2.3. Task 3. From “probe signal value” to “gene signal value” Our goal was to assign each gene from each array (sample) with a “gene signal value” based on the signal values of the probes associated with the genes. When a gene had one, and only one, associated probe, the signal value of that probe was the signal value of the gene. When a gene had more than one associated probe, we performed a descriptive statistical analysis of the signal values of the associated probes. If the standard error was less than ten percent of the mean value, we considered that this mean value of
the probe signal values was consistent enough to be accepted as the corresponding gene signal value. However, if the standard error was greater than ten percent of the mean, we removed an extreme value (the minimum or maximum depending on the relative values of the mean and the median), which were expected to be outliers, and repeated the analysis with the remaining values. We used the same criteria to accept the “gene signal value”. If the error value was still large after this second analysis, no gene signal value was accepted and the gene in question was not included in the gene list of the array (sample) analysed. We took each sample record individually, so not every sample had the same number of genes because we discarded genes from some samples for being ambiguous. Once we had calculated the “gene signal value” for each array and each gene, and to facilitate the data analysis, we also calculated the “gene rank order” of signal measurements. All the gene signal values of one array were sorted, and the percentage calculated. This gave us an indication of where the signal of one gene fell with respect to all the other genes in one array. 2.4. Task 4. Descriptive statistics of gene signal values from each experimental group We had two experimental groups of whole human genome arrays (N = 58): skin biopsies from lesional skin and non-lesional skin. So, for each group, the data set consists of 58 variables (each array or sample), and the cases were the genes (about 20,000 rows). The units of the variables were the “gene signal values”. For each group, we used a Perl script to make an automatic descriptive statistical analysis and calculate the signal value for each gene. 2.5. Task 5. Statistical significance of mean gene signal values between groups We were interested in gathering information about these groups so that we could compare them. We focused on the genes common to both groups. For each gene, we automatically used the test of significance for two known means, with known standard deviations and “N” (number of variables). The null hypothesis was that the means of the gene expression value were equal between groups. We calculated the two-sample z statistic. The significance level we considered was a = 0.01. Therefore, there were significant differences if z < 2.58 or z > 2.58. All the results were analysed in detail using a local MySQL database. 3. Results and discussion The discussion is based on the statistical analysis of specific patterns of mRNA expression values from gene expression profiling by arrays of healthy and lesional skin biopsies of psoriasis. Because environmental factors are important in psoriasis [6], we avoided mixing data from different array experiments from the NCBI-GEO series because they involved different population groups. According to the experimental design, the cell population in both types of skin biopsies (control and lesional) is notoriously different. In healthy skin biopsies, the cell population is typical of the epidermis, dermis and hypodermis, in which keratinocytes are the majority [7]. However, in lesional biopsies there is also a population of immune cells [1]. Thus, in the present data analysis of the gene array signal of probe-RNA hybridization, it should be borne in mind that this is not the analysis of a single cell but of a cluster of cells. Table S1 shows the complete list of human genes in both experimental groups (control and lesional skin biopsies), which have been through the filtering procedures described above. Table S1 shows the mean group values of the probe-RNA signals for
Please cite this article in press as: J.L. Melero, et al., Deciphering psoriasis. A bioinformatic approach, J Dermatol Sci (2017), https://doi.org/ 10.1016/j.jdermsci.2017.11.010
G Model DESC 3284 No. of Pages 7
J.L. Melero et al. / Journal of Dermatological Science xxx (2017) xxx–xxx
each gene, the statistical significance of the differences, and the chromosome locations. The data in Table S1 is also shown in Fig. 1, which shows the mean values of both experimental groups. Fig. 1 clearly shows that genes with high signal values in one group have very low values in the other, and vice versa. This reveals that psoriasis has a significant general effect on gene expression profile in the cells of the affected skin tissue. Of course, not all human genes are expressed in all cells, and bioinformatics cannot define a threshold of the “array signal value” above which a given gene can be considered to be expressed. This would require a laboratory protein analysis. However, genes with a high signal value of probe-RNA hybridization may be expressed in the biological system under study. In this context, the genes we consider to be positively affected by psoriasis have signal values that range from very low to very high (more than four times the low value). This suggests that these genes are not expressed in the group of healthy cells, and are highly expressed in the group of psoriatic cells. Likewise, negatively affected genes are highlyexpressed in healthy cells and not expressed in psoriasis. We divide the section below into two subsections. First, we discuss changes in the expression of genes which are associated with processes that are already known in the pathogenesis of psoriasis. This first analysis will serve to verify the consistency of our data. In the second subsection, we point out the significant effect of psoriasis on some specific genes, which as far as we know, has not previously been associated with psoriasis.
3
Table S1 show that array signals of genes encoding K6, K16 and K17 increase about 10-fold in lesional psoriatic skin tissue in comparison with healthy skin.
3.1. Results that are consistent with well-known facts about psoriasis
3.1.2. Late cornified cell envelope The late cornified cell envelope is the outermost layer of the epidermis. It consists of dead cells (corneocytes) with no nuclei or cell organelles. The interior of the corneocytes consists mainly of keratin filaments bound by filaggrin protein [11]. In agreement with the downregulating filaggrin expression in lesional psoriatic skin [12], here we show (Table S1) that filaggrin gene array signals (FLG, FLG2) in psoriatic biopsies were much lower (7-fold) than in normal skin samples. In the cell envelope that protects corneocytes in the epidermis, involucrin (IVL) and loricrin (LOR) are the protein components found in terminally differentiated epidermal cells [13]. In psoriasis, involucrin, as an early keratinocyte differentiation marker, is upregulated, whereas loricrin, as a late keratinocyte differentiation marker and a major cornified cell envelope component, is down regulated. We report that the involucrin gene array signals increase 4-fold in the psoriatic group. In contrast with involucrin, we found that the loricrin gene array signal also decreases about 4-fold in psoriatic skin tissue (Table S1). Also in this group of structural proteins, the small proline-rich proteins (SPRRs) are involved in the assembly of the cornified cell envelope [14]. SPRRs are early keratinocyte differentiation markers, and are up-regulated in psoriasis [15]. We show an extreme increase (about 10-fold) of the SPRR gene signal array in the psoriatic group (Table S1).
3.1.1. Keratins Keratinocytes express the keratins that allow the skin to perform its barrier function [8]. Two clusters of keratin genes are located on chromosomes 12 and 17, encoding basic type II and acid type I, respectively. The keratin filament assembly requires both keratin types [9]. Psoriatic keratinocytes overexpress type II keratin 6 (KRT6A, KRT6B, KRT6C), and type I keratin 16 and 17 (KRT16, KRT17). Keratins 16 and 17 are described as markers of keratinocyte hyperproliferation in psoriasis [10]. The data in
3.1.3. Cytokines Psoriatic skin inflammation depends on the presence of immune cells and their cytokines (interleukins, tumour necrosis factor, interferons) [3]. The data in Table S1 shows the dramatic increase in the hybridization signal in the arrays of the psoriasis group compared to the control group of several cytokines. The IL1B gene increased 16-fold. This cytokine is produced by activated macrophages as a proprotein, which is proteolytically processed to its active form by caspase 1 (CASP1) [16]. We also found a
Fig. 1. Gene signal values of probe-RNA hybridization. Gene signal values of skin punch biopsies from non-lesional skin (control group) and lesional skin (uninvolved group) from psoriatic patients. The x-axis represents the 20,000 genes (approximately) considered in the present study. The y-axis is the mean value of the gene signal of the probe-RNA hybridization in arbitrary units. The genes were sorted from highest to lowest probe-RNA hybridization signals of the control group. The gene signal profile of the control group is denoted in brown; The gene signal profile of the psoriatic group is denoted in blue. Data are taken from Table S1.
Please cite this article in press as: J.L. Melero, et al., Deciphering psoriasis. A bioinformatic approach, J Dermatol Sci (2017), https://doi.org/ 10.1016/j.jdermsci.2017.11.010
G Model DESC 3284 No. of Pages 7
4
J.L. Melero et al. / Journal of Dermatological Science xxx (2017) xxx–xxx
significant 6-fold increase in the CASP1 gene signal in psoriatic tissue. Another interleukin whose gene signal increases significantly is the IL36G. Interferon-gamma, tumour necrosis factoralpha and interleukin 1 beta (IL1B) are reported to stimulate the expression of this cytokine in keratinocytes [17]. Table S1 shows that in psoriatic tissue, the gene array signal of IL37 dramatically decreases and the gene hybridization signal in the arrays of the STAT3 gene significantly increases. Accordingly, the impaired function of regulatory T cells in patients with psoriasis is mediated by the phosphorylation of STAT3. 3.1.4. S100 family proteins S100 proteins are a family of low-molecular-weight proteins characterized by two calcium-binding sites. They are involved in the inflammatory process, and are part of the epidermal differentiation complex [18]. S100A7 (psoriasin) is overexpressed in inflammatory diseases, and this overexpression contributes to dysregulated differentiation in psoriasis [19]. Likewise, we found that not only S100A7, but also S100A2, S100A9, S100A11, S100A12, S100A14 and S100A16 showed very significantly higher gene array values in psoriasis biopsies than in controls (Table S1). 3.1.5. Chemokines In psoriasis, keratinocytes and immune cells produce a broad repertoire of chemokines, most of which are upregulated [20,21]. Table S1 shows the significant, 10-fold up-regulation of CXCL9, CXCL10 and CXCR4 (from the CXC chemokine family), and CCL20, CCL22 and CCR7 (from the CC chemokine family) in psoriatic tissue. We also show a psoriatic decrease of 16-fold in the gene signal array of CCL27, which agrees with the differential downexpression of the chemokine CCL27 gene in psoriasis [22]. 3.1.6. The major histocompatibility complex (MHC) locus The major genetic determinant of psoriasis is found in the MHC locus (PSORS1, chromosome 6). However, within the MHC locus, we found a significant increase (4-fold) of the TAP2 gene signal in psoriatic arrays (Table S1). The membrane-associated protein encoded by TAP2 is a member of the superfamily of ATP-binding cassette (ABC) transporters, which is involved in antigen presentation [23]. The association of TAP alleles with psoriasis has been described [24,25]. 3.1.7. Antimicrobial peptides (AMPs) The induction of AMPs is markedly increased in psoriasis [26– 28]. Cationic cathelicidins are increased in psoriasis [1]. The case of cathelicidin LL-37 (coded by the CAMP gene) is well described [29]. However, we found just a moderate increase of the CAMP gene signal in psoriatic arrays. Lipocalins base their AMP function on binding to bacterial siderophores [30] and are consistently upregulated in psoriasis [31]. Here we show that lipocalin-2 (LCN2) is 14-fold higher in the psoriatic groups than in controls. Defensins are small cysteine-rich cationic proteins which are active against bacteria, fungi and many enveloped and nonenveloped viruses [28]. Psoriasis is associated with defensin upregulation [32,33]. Here we report a high increase in the gene array signal of the beta defensins DEFB4A,B (about 19-fold) and DEFB103A,B (about 4-fold), due to psoriasis. At this point, we also point out the dramatic increase (about 20-fold) in the gene expression in the psoriatic group in comparison to the control group of genes encoding lactotransferrin (LTF) and lysozyme (LYZ). 3.2. Psoriasis. What else? 3.2.1. Cell cycle We were interested in the effect of psoriasis on the expression of cyclins and cyclin-dependent kinase coding genes. Table S1
shows the increase and decrease in hybridization array signals of genes encoding these proteins. We point out the signal gene values of the cyclins B1, A2 and E2 (CCNB1, CCNA2, CCNE2, respectively), and the cyclin-dependent kinase (CDK1), which were much more highly expressed in psoriatic tissue (about 10-fold). With the dysregulation of these cyclin coding genes, it is plausible to think that psoriasis affects strategic points of the cell cycle. Cyclin E2 plays a role in the G1/S portion of the cell cycle. Cyclin A2 is synthesized at the onset of the S phase and is localized in the nucleus. And cyclin B1 is a regulatory protein involved in mitosis (provided by NCBI, RefSeq). Interestingly, we found that the effect of psoriasis on the cyclin B/Cdk1 kinase complex is also linked to the effect on the two B-type lamin proteins (LMNB1 and LMNB2) (Table S1), which are also extremely up-regulated. Lamins are components of the fibrillar network proteins inside the nucleus (nuclear lamina), attached to the inner nuclear membrane, and the cyclin B/Cdk1 protein kinase complex initiates lamin disassembly (provided by NCBI, RefSeq). Interestingly, we have also observed a significant increase in MYC gene expression in arrays of psoriatic tissue. This agrees with the effect of the proto-oncogene MYC, which promotes differentiation and cell cycle deregulation in human keratinocytes [34]. In fact, uncontrolled keratinocyte proliferation is a hallmark of psoriatic skin lesions [1,2]. 3.2.2. Dynamin Psoriatic skin has long been known to contain elevated levels of interferon types [21]. So, we focused on the interferon-induced dynamins. Dynamins are a superfamily of large multidomain mechanochemical GTPases (EC 3.6.5.5), which are coordinately involved in the fission and fusion of membrane structures [35]. The myxovirus resistance proteins (MX) of the dynamin family are induced by type I interferons and block the replication of a broad spectrum of viruses at various sites in the cell [36]. We found (Table S1) that the two MX coding genes (MX1 and MX2) were upregulated (about 10-fold) in psoriasis biopsy samples in comparison with controls. Similarly, the dynamin family guanylate-binding proteins (GBPs) are induced by type II interferons and antiinflammatory cytokines [37]. We also found (Table S1) that all GBP genes (GBP1-6) were extremely highly regulated in psoriatic skin, and reached values up to 15-fold higher than controls. As of the interferon-inducible GTPase family, both MX and GBP dynamins have an antiviral effect and play a role in intracellular defence [38,39]. We also note that the effect of the disease on the dynamin system also reaches other dynamin families, covering such fundamental processes as mitochondrial dynamics [40]. We emphasize the 7-fold increase of the array signal value of the mitofusin MFN1 gene (Table S1), which is involved in mitochondrial fusion. Further evidence of the consistent association between the dynamin system and the pathogenesis of psoriasis is the significant increase in the expression of the dynamin-2 gene (DNM2). Because this study is based on multicellular data (biopsies from healthy controls and disease tissue) but not a single type of cells, and it is known that the dynamin system acts in keratinocytes in a different scenario of psoriasis, such as viral infection [41,42], we suggest that the effects of psoriasis on the expression of dynamin-related genes take place at the keratinocyte level. 3.2.3. Antioxidants The role of oxidative stress in various stages of psoriasis has been described [43,44]. Oxidative stress involves not only ROS, but also the antioxidants that control their presence. In this context, we found a significant down-regulation (about 7-fold) of genes encoding key antioxidant enzymes (that is, catalase, CAT; superoxide dismutases, SOD1-3; and glutathione reductase, GSR) (Table S1). The significant negative effect of psoriasis on catalase, superoxide dismutase and glutathione reductase gene
Please cite this article in press as: J.L. Melero, et al., Deciphering psoriasis. A bioinformatic approach, J Dermatol Sci (2017), https://doi.org/ 10.1016/j.jdermsci.2017.11.010
G Model DESC 3284 No. of Pages 7
J.L. Melero et al. / Journal of Dermatological Science xxx (2017) xxx–xxx
5
Fig. 2. Location of selected gene clusters in the human genome. Distribution of gene clusters in the human genome, containing genes whose expression is significantly affected by psoriasis in the affected areas of the skin. These genes have been commented on in the text. Schemes of the chromosomes have been taken from the NCBI (Human Genome Resources). The cytogenetic location is used. For each gene cluster, denoted in red, we give a reference coordinate and the name of the protein or the protein family encoded by the genes of the cluster.
Please cite this article in press as: J.L. Melero, et al., Deciphering psoriasis. A bioinformatic approach, J Dermatol Sci (2017), https://doi.org/ 10.1016/j.jdermsci.2017.11.010
G Model DESC 3284 No. of Pages 7
6
J.L. Melero et al. / Journal of Dermatological Science xxx (2017) xxx–xxx
expressions is a typical oxidative stress scenario, which may be an additional factor in the pathogenesis of the disease. Psoriasin (S100A7) increases the expression of ROS in keratinocytes [45]. This correlates with the significant high values of S100A7 gene expression found in lesional samples, and is consistent with the down-regulation of the key antioxidant enzymes (Table S1). 3.2.4. Kallikreins The detailed analysis of gene expression differences between both healthy and psoriatic experimental groups, on the one hand and the positions occupied by highly expressed or silenced genes in the genome, on the other, led us to discover that kallikreins are involved in psoriatic pathogenicity. The subgroup of serine proteases is involved in several physiological processes, including normal skin desquamation [46–48]. The link between kallikreins and psoriasis has not yet been described. Here we show that in psoriasis there is a significant up-regulation (about 9-fold) of a set of kallikrein genes (KLK1-KLK13) in the cells of the psoriaticinjured epithelial tissue. In agreement with our results, there is some evidence to support the kallikreins-psoriasis association. For example, kallikrein-mediated proteolysis regulates the antimicrobial effects of cathelicidins in skin [49]. Also, kallikrein-related peptidase-8 (KLK8) is an active serine protease in human epidermis and sweat and is involved in a skin barrier proteolytic cascade [50]. And finally, kallikrein 5 mediated inflammation in rosacea [51]. Psoriasis has an effect on a wide variety of genes with a wide range of biological functions. As we have just seen, these genes range from those that are well known as mediators of the pathogenesis of the disease to genes whose biological function has not yet been associated with psoriasis. A detailed analysis of Table S1 will enable the list of genes whose expression may be affected by psoriasis to be extended. From the information in Table S1, we observe that the effect of psoriasis extends across all regions of the chromosomes of the human genome. Psoriasis is also known to have a strong hereditary character and a complex genetic basis [1–3]. In fact, “heredity” and “genetic predisposition” are different issues. If we add the sensitivity of psoriasis to environmental factors [6], and combine everything with the results we show here, it appears that psoriasis could be a disease of epigenetic origin. In psoriatic patients, the phenotype of the lesional skin tissue is stable and mitotically heritable. We need to consider that healthy and affected cells share the same genome. These are the credentials of an epigenetic context [52]. This epigenetic view is supported by the fact that in many cases, the effect of psoriasis on gene expression affects not only a single gene but also a gene cluster. Fig. 2 illustrates this phenomenology with genes discussed in this section. Other gene clusters can be discovered by a detailed analysis of Table S1. These patterns of changes in gene expression within chromosome regions seem to reflect what happens in imprinted chromosomal regions, in their “control regions” [53]. However, unlike in imprinting, in psoriasis the two homologous chromosomes are affected. We suggest that the genome of a psoriasis patient contains some epigenetic imprints (e.g. differentially methylated regions, histone modifications, non-coding RNAs, insulators, and so on) that coincide in the epidermis and, in conjunction with environmental factors, lead to the manifestation of psoriasis. Other authors, C. M. Nguyen and W. Liao [54], have reviewed the possible evidence of the epigenètic phenomenon “imprinting” in psoriasis. However, this is not what we are suggesting. Of the environmental factors (so important in epigenetics) the skin’s bacterial flora could play an important role. This may explain why psoriasis is present in some regions of the skin, and not in others. This epigenetic view of psoriasis fits well with the heritable nature of the disease. In this context,
various reviews have been published about the role of epigenetics in psoriasis [55–60]. Finally, we believe that the complex pathogenesis of psoriasis can be studied using an epigenetic approach. Conflict of interest statement The authors declare that there are no conflicts of interest. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at https://doi.org/10.1016/j. jdermsci.2017.11.010. References [1] G.K. Perera, P. DiMeglio, F.O. Nestle, Psoriasis, Annu. Rev. Pathol. 7 (2012) 385– 422. [2] F.C. Eberle, J. Brück, J. Holstein, K. Hirahara, K. Ghoreschi, Recent advances in understanding psoriasis, F1000Res. (2016) 5. [3] J. Baliwag, D.H. Barnes, A. Johnston, Cytokines in psoriasis, Cytokine 73 (2015) 342–350. [4] P. Di Meglio, F. Villanova, F.O. Nestle, Psoriasis, Cold Spring Harb. Perspect. Med. 4 (8) (2014) (pii: a015354). [5] H. Alexander, F.O. Nestle, Pathogenesis and immunotherapy in cutanis psoriasis: what can rheumatologists learn? Curr. Opin. Rheumatol. 29 (2017) 71–78. [6] E. Dika, F. Bardazzi, R. Balestri, H.I. Maibach, Environmental factors and psoriasis, Curr. Probl. Dermatol. 35 (2007) 118–135. [7] E. Proksch, J.M. Brandner, J.M. Jensen, The skin: an indispensable barrier, Exp. Dermatol. 17 (2008) 1063–1072. [8] E. Fuchs, Keratins and the skin, Annu. Rev. Cell. Dev. Biol. 11 (1995) 123–153. [9] L. Ramms, G. Fabris, R. Windoffer, N. Schwarz, R. Springer, C. Zhou, et al., Keratins as the main component for the mechanical integrity of keratinocytes, Proc. Natl. Acad. Sci. U. S. A. 110 (2013) 18513–18518. [10] I.M. Leigh, H. Navsaria, P.E. Purkis, I.A. McKay, P.E. Bowden, P.N. Riddle, Keratins (K16 and K17) as markers of keratinocyte hyperproliferation in psoriasis in vivo and in vitro, Br. J. Dermatol. 133 (1995) 501–511. [11] S. Kezic, I. Jakasa, Filaggrin and skin barrier function, Curr. Probl. Dermatol. 49 (2016) 1–7. [12] T. Makino, M. Mizawa, T. Yamakoshi, M. Takaishi, T. Shimizu, Expression of filaggrin-2 protein in the epidermis of human skin diseases: a comparative analysis with filaggrin, Biochem. Biophys. Res. Commun. 449 (2014) 100–106. [13] E. Candi, R. Schmidt, G. Melino, The cornified envelope: a model of cell death in the skin, Nat. Rev. Mol. Cell. Biol. 6 (2005) 328–340. [14] P.M. Steinert, E. Candi, T. Kartasova, L. Marekov, Small proline-rich proteins are cross-bridging proteins in the cornified cell envelopes of stratified squamous epithelia, J. Struct. Biol. 122 (1998) 76–85. [15] H. Iizuka, H. Takahashi, M. Honma, A. Ishida-Yamamoto, Unique keratinization process in psoriasis: late differentiation markers are abolished because of the premature cell death, J. Dermatol. 31 (2004) 271–276. [16] N.M. de Vasconcelos, N. Van Opdenbosch, M. Lamkanfi, Inflammasomes as polyvalent cell death platforms, Cell. Mol. Life. Sci. 73 (2016) 2335–2347. [17] K. Takahashi, A. Nishida, M. Shioya, H. Imaeda, S. Bamba, O. Inatomi, et al., Interleukin (IL)-1? Is a strong inducer of IL-36? Expression in human colonic myofibroblasts, PLoS One 10 (11) (2015) e0138423. [18] M. Kypriotou, M. Huber, D. Hohl, The human epidermal differentiation complex: cornified envelope precursors, S100 proteins and the ‘fused genes' family, Exp. Dermatol. 21 (2012) 643–649. [19] A.K. Ekman, J. Vegfors, C.B. Eding, C. Enerbäck, Overexpression of psoriasin (S100A7) contributes to dysregulated differentiation in psoriasis, Acta Derm. Venereol. 97 (2017) 441–448. [20] M. Sugaya, Chemokines and skin diseases, Arch. Immunol. Ther. Exp. (Warsz) 63 (2015) 109–115. [21] T. Cao, S. Shao, B. Li, L. Jin, J. Lei, H. Qiao, et al., Up-regulation of Interferoninducible protein 16 contributes to psoriasis by modulating chemokine production in keratinocytes, Sci. Rep. 6 (2016) 25381. [22] J.L. Riis, C. Johansen, C. Vestergaard, R. Bech, K. Kragballe, L. Iversen, Kinetics and differential expression of the skin-related chemokines CCL27 and CCL17 in psoriasis, atopic dermatitis and allergic contact dermatitis, Exp. Dermatol. 20 (October (10)) (2011) 789–794. [23] M. Naderi, M. Hashemi, S. Amininia, Association of TAP1 and TAP2 gene polymorphisms with susceptibility to pulmonary tuberculosis, Iran. J. Allergy Asthma Immunol. 15 (2016) 62–68. [24] H. Saeki, S. Kuwata, H. Nakagawa, A. Asahina, K. Tamaki, Y. Shibata, et al., Analysis of HLA class II and TAP alleles in Japanese patients with psoriasis vulgaris, Hum. Immunol. 59 (1998) 503–511. [25] C.W. Pyo, S.S. Hur, Y.K. Kim, T.Y. Kim, T.G. Kim, Association of TAP and HLA-DM genes with psoriasis in Koreans, J. Invest. Dermatol. 120 (2003) 616–622.
Please cite this article in press as: J.L. Melero, et al., Deciphering psoriasis. A bioinformatic approach, J Dermatol Sci (2017), https://doi.org/ 10.1016/j.jdermsci.2017.11.010
G Model DESC 3284 No. of Pages 7
J.L. Melero et al. / Journal of Dermatological Science xxx (2017) xxx–xxx [26] G. Wang, Human antimicrobial peptides and proteins, Pharmaceuticals (Basel) 7 (2014) 545–594. [27] M. Marcinkiewicz, S. Majewski, The role of antimicrobial peptides in chronic inflammatory skin diseases, Postepy Dermatol. Alergol. 33 (2016) 6–12. [28] T. Takahashi, R.L. Gallo, The critical and multifunctional roles of antimicrobial peptides in dermatology, Dermatol. Clin. 35 (2017) 39–50. [29] R. Lande, E. Botti, C. Jandus, D. Dojcinovic, G. Fanelli, C. Conrad, et al., The antimicrobial peptide LL37 is a T-cell autoantigen in psoriasis, Nat. Commun. 5 (2014) 5621. [30] M. Jung, C. Mertens, R. Bauer, C. Rehwald, B. Brüne, Lipocalin-2 and iron trafficking in the tumor microenvironment, Pharmacol. Res. 120 (2017) 146– 156. [31] S. Shao, T. Cao, L. Jin, B. Li, H. Fang, J. Zhang, et al., Increased lipocalin-2 contributes to the pathogenesis of psoriasis by modulating neutrophil chemotaxis and cytokine secretion, J. Invest. Dermatol. 136 (2016) 1418–1428. [32] W.H. Boehncke, The alpha-defensins HNP-1 and HNP-2 are dominant selfpeptides presented by HLA class-II molecules in lesional psoriatic skin, Eur. J. Dermatol. 14 (2004) 142–145. [33] P.E. Stuart, U. Hüffmeier, R.P. Nair, R. Palla, T. Tejasvi, J. Schalkwijk, et al., Association of?-defensin copy number and psoriasis in three cohorts of European origin, J. Invest. Dermatol. 132 (2012) 2407–2413. [34] A. Gandarillas, The mysterious human epidermal cell cycle or an oncogeneinduced differentiation checkpoint, ABBV Cell Cycle 11 (2012) 4507–4516. [35] G.J. Praefcke, H.T. McMahon, The dynamin superfamily: universal membrane tubulation and fission molecules? Nat. Rev. Mol. Cell Biol. 5 (2004) 133–147. [36] O. Haller, G. Kochs, Interferon-induced mx proteins: dynamin-like GTPases with antiviral activity, Traffic 10 (2002) 710–717. [37] D. Pilla-Moffett, M.F. Barber, G.A. Taylor, J. Coers, Interferon-inducible GTPases in host resistance, inflammation and disease, J. Mol. Biol. 428 (2016) 3495– 3513. [38] D.P. Barlow, M.S. Bartolomei, Genomic imprinting in mammals, Cold Spring Harb. Perspect. Biol. 6 (2014) (pii: a018382). [39] O. Haller, P. Staeheli, M. Schwemmle, G. Kochs, Mx GTPases: dynamin-like antiviral machines of innate immunity, Trends Microbiol. 23 (2015) 154–163. [40] H. Lee, Y. Yoon, Mitochondrial fission and fusion, Biochem. Soc. Trans. 44 (2016) 1725–1735. [41] E. Rahn, P. Petermann, M.J. Hsu, F.J. Rixon, D. Knebel-Mörsdorf, Entry pathways of herpes simplex virus type 1 into human keratinocytes are dynamin- and cholesterol-dependent, PLoS One 10 (2011) e25464. [42] J.L. Smith, S.K. Campos, M.A. Ozbun, Human papillomavirus type 31 uses a caveolin 1- and dynamin 2-mediated entry pathway for infection of human keratinocytes, J. Virol. 81 (2007) 9922–9931. [43] X. Lin, T. Huang, Oxidative stress in psoriasis and potential therapeutic use of antioxidants, Free Radic. Res. 50 (2016) 585–595. [44] S. Srivastava, D. Singh, S. Patel, M.R. Singh, Role of enzymatic free radical scavengers in management of oxidative stress in autoimmune disorders, Int. J. Biol. Macromol. 101 (2017) 502–517.
7
[45] E. Shubbar, J. Vegfors, M. Carlström, S. Petersson, C. Enerbäck, Psoriasin (S100A7) increases the expression of ROS and VEGF and acts through RAGE to promote endothelial cell proliferation, Breast Cancer Res. Treat. 134 (2012) 71– 80. [46] A. Eissa, E.P. Diamandis, Human tissue kallikreins as promiscuous modulators of homeostatic skin barrier funcions, Biol. Chem. 389 (2008) 669–680. [47] M. Kishibe, Y. Bando, R. Terayama, K. Namikawa, H. Takahashi, Y. Hashimoto, et al., Kallikrein 8 is involved in skin desquamation in cooperation with other kallikreins, J. Biol. Chem. 282 (2007) 5834–5841. [48] A.C. Stefanini, B.R. daCunha, T. Henrique, E.H. Tajara, Involvement of kallikreinrelated peptidases in normal and pathologic processes, Dis. Markers (2015) 946572. [49] K. Yamasaki, J. Schauber, A. Coda, H. Lin, R.A. Dorschner, N.M. Schechter, et al., Kallikrein-mediated proteolysis regulates the antimicrobial effects of cathelicidins in skin, FASEB J. 20 (2006) 2068–2080. [50] A. Eissa, V. Amodeo, C.R. Smith, E.P. Diamandis, Kallikrein-related peptidase-8 (KLK8) is an active serine protease in human epidermis and sweat and is involved in a skin barrier proteolytic cascade, J. Biol. Chem. 286 (2011) 687– 706. [51] A.M. Two, J.Q. Del Rosso, Kallikrein 5-mediated inflammation in rosacea: clinically relevant correlations with acute and chronic manifestations in rosacea and how individual treatments may provide therapeutic benefit, J. Clin. Aesthet. Dermatol. 7 (2014) 20–25. [52] S.L. Berger, T. Kouzarides, R. Shiekhattar, A. Shilatifard, An operational definition of epigenètics, Genes. Dev. 23 (2009) 781–783. [53] D.P. Barlow, M.S. Bartolomei, Genomic imprinting in mammals, Cold Spring Harb. Perspect. Biol. 6 (2014) (pii: a018382). [54] C.M. Nguyen, W. Liao, Genomic imprinting in psoriasis and atopic dermatitis: a review, J. Dermatol. Sci. 80 (2015) 89–93. [55] P. Zhang, Y. Su, Q. Lu, Epigenetics and psoriasis, J. Eur. Acad. Dermatol. Venereol. 26 (2012) 399–403. [56] J.E. Gudjonsson, G. Krueger, A role for epigenetics in psoriasis: methylated cytosine-guanine sites differentiate lesional from nonlesional skin and from normal skin, J. Invest. Dermatol. 132 (2012) 506–508. [57] P. Zhang, M. Zhao, G. Liang, G. Yin, D. Huang, F. Su, et al., Whole-genome DNA methylation in skin lesions from patients with psoriasis vulgaris, J. Autoimmun. 41 (2013) 17–24. [58] R.M. Trowbridge, M.R. Pittelkow, Epigenetics in the pathogenesis and pathophysiology of psoriasis vulgaris, J. Drugs Dermatol. 13 (2014) 111–118. [59] F. Zhou, C. Shen, J. Xu, J. Gao, X. Zheng, R. Ko, et al., Epigenome-wide association data implicates DNA methylation-mediated genetic risk in psoriasis, Clin. Epigenet. 8 (2016) 131. [60] R.A. Pollock, F. Abji, D.D. Gladman, Epigenetics of psoriatic disease: a systematic review and critical appraisal, J. Autoimmun. 78 (2017) 29–38.
Please cite this article in press as: J.L. Melero, et al., Deciphering psoriasis. A bioinformatic approach, J Dermatol Sci (2017), https://doi.org/ 10.1016/j.jdermsci.2017.11.010
Journal of Dermatological Science xxx (2017) xxx–xxx
Contents lists available at ScienceDirect
Journal of Dermatological Science journal homepage: www.jdsjournal.com
Deciphering psoriasis. A bioinformatic approach Juan L. Melero, Sergi Andrades, Lluís Arola, Antoni Romeu* Department of Biochemistry and Biotechnology, Rovira i Virgili University, Tarragona, Spain
A R T I C L E I N F O
A B S T R A C T
Article history: Received 21 July 2017 Received in revised form 25 September 2017 Accepted 18 November 2017
Background: Psoriasis is an immune-mediated, inflammatory and hyperproliferative disease of the skin and joints. The cause of psoriasis is still unknown. The fundamental feature of the disease is the hyperproliferation of keratinocytes and the recruitment of cells from the immune system in the region of the affected skin, which leads to deregulation of many well-known gene expressions. Objective: Based on data mining and bioinformatic scripting, here we show a new dimension of the effect of psoriasis at the genomic level. Methods: Using our own pipeline of scripts in Perl and MySql and based on the freely available NCBI Gene Expression Omnibus (GEO) database: DataSet Record GDS4602 (Series GSE13355), we explore the extent of the effect of psoriasis on gene expression in the affected tissue. Results: We give greater insight into the effects of psoriasis on the up-regulation of some genes in the cell cycle (CCNB1, CCNA2, CCNE2, CDK1) or the dynamin system (GBPs, MXs, MFN1), as well as the downregulation of typical antioxidant genes (catalase, CAT; superoxide dismutases, SOD1-3; and glutathione reductase, GSR). We also provide a complete list of the human genes and how they respond in a state of psoriasis. Conclusion: Our results show that psoriasis affects all chromosomes and many biological functions. If we further consider the stable and mitotically inheritable character of the psoriasis phenotype, and the influence of environmental factors, then it seems that psoriasis has an epigenetic origin. This fit well with the strong hereditary character of the disease as well as its complex genetic background. © 2017 Japanese Society for Investigative Dermatology. Published by Elsevier Ireland Ltd. All rights reserved.
Keywords: Psoriasis Keratinocyte Gene expression Epigenetics Bioinformatics
What’s already known about this topic? Psoriasis influences a wide variety of genes with a wide range of biological functions. Psoriasis is also known to have a strong hereditary character and a complex genetic basis.
Combining the strong hereditary character of the psoriasis with the results we show here, it appears that psoriasis could be a disease of epigenetic origin.
1. Introduction What does this study add? Cyclins B1, A2, E2, CDK1, and B-type lamin genes are highly expressed in psoriatic tissue. Dynamin system is up-regulated in psoriasis. Catalase, superoxide dismutase and glutathione reductase a down-regulated in psoriasis. Psoriasis is associated with kallikreins. The effect of psoriasis extends across all regions of the chromosomes of the human genome. This effect on gene expression affects not only a single gene but also a gene cluster.
* Corresponding author. E-mail address: [email protected] (A. Romeu).
Psoriasis is a chronic autoimmune skin disease [1,2]. Although the cause of the disease is still unknown, psoriasis is a phenotype of skin biochemical and immune disorder. Some hereditary factors predispose to psoriasis, which has been linked with at least nine chromosomal loci (PSORS1 through PSORS9). The molecular mechanisms responsible for psoriasis still have to be fully elucidated. The disease is multifactorial and noteworthy involves the hyperproliferation of keratinocytes in the epidermis. Various reviews have been published of the advances in our understanding of psoriasis [3–5]. Gene expression, although representing a specific site of regulation, is only one step in the complex cascade from an initial stimulus to a final phenotype. In this context, we were keen to explore the extent of the effect of psoriasis on gene expression in
https://doi.org/10.1016/j.jdermsci.2017.11.010 0923-1811/ © 2017 Japanese Society for Investigative Dermatology. Published by Elsevier Ireland Ltd. All rights reserved.
Please cite this article in press as: J.L. Melero, et al., Deciphering psoriasis. A bioinformatic approach, J Dermatol Sci (2017), https://doi.org/ 10.1016/j.jdermsci.2017.11.010
G Model DESC 3284 No. of Pages 7
2
J.L. Melero et al. / Journal of Dermatological Science xxx (2017) xxx–xxx
the affected tissue. First, at a quantitative level, that is, how many genes there are (or can be) that in a healthy state are silenced and awakened by the disease. And, vice versa, how many genes are highly expressed in a healthy state but totally silenced in psoriasis. Second, at a more qualitative level, we were interested in studying the extent of psoriasis in functional genomics, as well as the regions of the genome that are affected. To do this, we used freely available data from the NCBI-Gene Expression Omnibus (GEO) database of whole human genome arrays with paired lesional and non-lesional skins from psoriasis patients. Using this data, we made an analysis of variance to identify differences in gene expression profiles among groups. From this standpoint, then, this study uses a bioinformatic approach to data mining. 2. Methodology The sources of information were the current version of the Human Genome (NCBI, GRCh37.p13) and the freely available NCBI Gene Expression Omnibus (GEO) database: DataSet Record GDS4602 (Series GSE13355) based on an analysis of skin punch biopsies from lesional skin and non-lesional skin from a group of 58 psoriatic patients. The platform used in these whole Human Genome arrays was the Affymetrix GPL570 (HG-U133_Plus_2, Affymetrix Human Genome U133 Plus 2.0 Array). For this study, we created our own pipeline of scripts in Perl and MySql for data management. The rationale of this work is based on the following tasks: 2.1. Task 1. Samples download In accordance with the experimental design of the GDS4602 record, gene expression profiles in uninvolved and involved skin from 58 psoriasis affected individuals were characterised. We used these two array groups. Each array (or sample) record consists of a text file of 54,675 rows, in which each “probe” has its “probe-RNA hybridization signal”. These values are considered arbitrary units. We downloaded the 116 array (sample) records. 2.2. Task 2. Probe selection The GPL570 platform has a 54,675 Affymetrix probe set. However, we only selected the probes whose associated Entrez_id number and gene symbol match those annotated in the current Human Genome release. A total of 46,002 probes were selected. Some of the probes in this platform are designed to recognize only transcripts (ending with ‘_at’) and while others are designed to recognize multiple transcripts from the same gene family (ending with ‘_s_at’ or ‘_a_at’). In order to further adjust the analysis of gene expression, here we have only considered the transcript-specific probes. On the other hand, with respect to the genes in the GPL570 platform, 19,873 had a match in the Human Genome release. Of these, 12,927 genes are associated with a single probe, 4387 are associated to two probes, 1611 genes are associated to three probes, 599 genes are associated to four probes, etc. 2.3. Task 3. From “probe signal value” to “gene signal value” Our goal was to assign each gene from each array (sample) with a “gene signal value” based on the signal values of the probes associated with the genes. When a gene had one, and only one, associated probe, the signal value of that probe was the signal value of the gene. When a gene had more than one associated probe, we performed a descriptive statistical analysis of the signal values of the associated probes. If the standard error was less than ten percent of the mean value, we considered that this mean value of
the probe signal values was consistent enough to be accepted as the corresponding gene signal value. However, if the standard error was greater than ten percent of the mean, we removed an extreme value (the minimum or maximum depending on the relative values of the mean and the median), which were expected to be outliers, and repeated the analysis with the remaining values. We used the same criteria to accept the “gene signal value”. If the error value was still large after this second analysis, no gene signal value was accepted and the gene in question was not included in the gene list of the array (sample) analysed. We took each sample record individually, so not every sample had the same number of genes because we discarded genes from some samples for being ambiguous. Once we had calculated the “gene signal value” for each array and each gene, and to facilitate the data analysis, we also calculated the “gene rank order” of signal measurements. All the gene signal values of one array were sorted, and the percentage calculated. This gave us an indication of where the signal of one gene fell with respect to all the other genes in one array. 2.4. Task 4. Descriptive statistics of gene signal values from each experimental group We had two experimental groups of whole human genome arrays (N = 58): skin biopsies from lesional skin and non-lesional skin. So, for each group, the data set consists of 58 variables (each array or sample), and the cases were the genes (about 20,000 rows). The units of the variables were the “gene signal values”. For each group, we used a Perl script to make an automatic descriptive statistical analysis and calculate the signal value for each gene. 2.5. Task 5. Statistical significance of mean gene signal values between groups We were interested in gathering information about these groups so that we could compare them. We focused on the genes common to both groups. For each gene, we automatically used the test of significance for two known means, with known standard deviations and “N” (number of variables). The null hypothesis was that the means of the gene expression value were equal between groups. We calculated the two-sample z statistic. The significance level we considered was a = 0.01. Therefore, there were significant differences if z < 2.58 or z > 2.58. All the results were analysed in detail using a local MySQL database. 3. Results and discussion The discussion is based on the statistical analysis of specific patterns of mRNA expression values from gene expression profiling by arrays of healthy and lesional skin biopsies of psoriasis. Because environmental factors are important in psoriasis [6], we avoided mixing data from different array experiments from the NCBI-GEO series because they involved different population groups. According to the experimental design, the cell population in both types of skin biopsies (control and lesional) is notoriously different. In healthy skin biopsies, the cell population is typical of the epidermis, dermis and hypodermis, in which keratinocytes are the majority [7]. However, in lesional biopsies there is also a population of immune cells [1]. Thus, in the present data analysis of the gene array signal of probe-RNA hybridization, it should be borne in mind that this is not the analysis of a single cell but of a cluster of cells. Table S1 shows the complete list of human genes in both experimental groups (control and lesional skin biopsies), which have been through the filtering procedures described above. Table S1 shows the mean group values of the probe-RNA signals for
Please cite this article in press as: J.L. Melero, et al., Deciphering psoriasis. A bioinformatic approach, J Dermatol Sci (2017), https://doi.org/ 10.1016/j.jdermsci.2017.11.010
G Model DESC 3284 No. of Pages 7
J.L. Melero et al. / Journal of Dermatological Science xxx (2017) xxx–xxx
each gene, the statistical significance of the differences, and the chromosome locations. The data in Table S1 is also shown in Fig. 1, which shows the mean values of both experimental groups. Fig. 1 clearly shows that genes with high signal values in one group have very low values in the other, and vice versa. This reveals that psoriasis has a significant general effect on gene expression profile in the cells of the affected skin tissue. Of course, not all human genes are expressed in all cells, and bioinformatics cannot define a threshold of the “array signal value” above which a given gene can be considered to be expressed. This would require a laboratory protein analysis. However, genes with a high signal value of probe-RNA hybridization may be expressed in the biological system under study. In this context, the genes we consider to be positively affected by psoriasis have signal values that range from very low to very high (more than four times the low value). This suggests that these genes are not expressed in the group of healthy cells, and are highly expressed in the group of psoriatic cells. Likewise, negatively affected genes are highlyexpressed in healthy cells and not expressed in psoriasis. We divide the section below into two subsections. First, we discuss changes in the expression of genes which are associated with processes that are already known in the pathogenesis of psoriasis. This first analysis will serve to verify the consistency of our data. In the second subsection, we point out the significant effect of psoriasis on some specific genes, which as far as we know, has not previously been associated with psoriasis.
3
Table S1 show that array signals of genes encoding K6, K16 and K17 increase about 10-fold in lesional psoriatic skin tissue in comparison with healthy skin.
3.1. Results that are consistent with well-known facts about psoriasis
3.1.2. Late cornified cell envelope The late cornified cell envelope is the outermost layer of the epidermis. It consists of dead cells (corneocytes) with no nuclei or cell organelles. The interior of the corneocytes consists mainly of keratin filaments bound by filaggrin protein [11]. In agreement with the downregulating filaggrin expression in lesional psoriatic skin [12], here we show (Table S1) that filaggrin gene array signals (FLG, FLG2) in psoriatic biopsies were much lower (7-fold) than in normal skin samples. In the cell envelope that protects corneocytes in the epidermis, involucrin (IVL) and loricrin (LOR) are the protein components found in terminally differentiated epidermal cells [13]. In psoriasis, involucrin, as an early keratinocyte differentiation marker, is upregulated, whereas loricrin, as a late keratinocyte differentiation marker and a major cornified cell envelope component, is down regulated. We report that the involucrin gene array signals increase 4-fold in the psoriatic group. In contrast with involucrin, we found that the loricrin gene array signal also decreases about 4-fold in psoriatic skin tissue (Table S1). Also in this group of structural proteins, the small proline-rich proteins (SPRRs) are involved in the assembly of the cornified cell envelope [14]. SPRRs are early keratinocyte differentiation markers, and are up-regulated in psoriasis [15]. We show an extreme increase (about 10-fold) of the SPRR gene signal array in the psoriatic group (Table S1).
3.1.1. Keratins Keratinocytes express the keratins that allow the skin to perform its barrier function [8]. Two clusters of keratin genes are located on chromosomes 12 and 17, encoding basic type II and acid type I, respectively. The keratin filament assembly requires both keratin types [9]. Psoriatic keratinocytes overexpress type II keratin 6 (KRT6A, KRT6B, KRT6C), and type I keratin 16 and 17 (KRT16, KRT17). Keratins 16 and 17 are described as markers of keratinocyte hyperproliferation in psoriasis [10]. The data in
3.1.3. Cytokines Psoriatic skin inflammation depends on the presence of immune cells and their cytokines (interleukins, tumour necrosis factor, interferons) [3]. The data in Table S1 shows the dramatic increase in the hybridization signal in the arrays of the psoriasis group compared to the control group of several cytokines. The IL1B gene increased 16-fold. This cytokine is produced by activated macrophages as a proprotein, which is proteolytically processed to its active form by caspase 1 (CASP1) [16]. We also found a
Fig. 1. Gene signal values of probe-RNA hybridization. Gene signal values of skin punch biopsies from non-lesional skin (control group) and lesional skin (uninvolved group) from psoriatic patients. The x-axis represents the 20,000 genes (approximately) considered in the present study. The y-axis is the mean value of the gene signal of the probe-RNA hybridization in arbitrary units. The genes were sorted from highest to lowest probe-RNA hybridization signals of the control group. The gene signal profile of the control group is denoted in brown; The gene signal profile of the psoriatic group is denoted in blue. Data are taken from Table S1.
Please cite this article in press as: J.L. Melero, et al., Deciphering psoriasis. A bioinformatic approach, J Dermatol Sci (2017), https://doi.org/ 10.1016/j.jdermsci.2017.11.010
G Model DESC 3284 No. of Pages 7
4
J.L. Melero et al. / Journal of Dermatological Science xxx (2017) xxx–xxx
significant 6-fold increase in the CASP1 gene signal in psoriatic tissue. Another interleukin whose gene signal increases significantly is the IL36G. Interferon-gamma, tumour necrosis factoralpha and interleukin 1 beta (IL1B) are reported to stimulate the expression of this cytokine in keratinocytes [17]. Table S1 shows that in psoriatic tissue, the gene array signal of IL37 dramatically decreases and the gene hybridization signal in the arrays of the STAT3 gene significantly increases. Accordingly, the impaired function of regulatory T cells in patients with psoriasis is mediated by the phosphorylation of STAT3. 3.1.4. S100 family proteins S100 proteins are a family of low-molecular-weight proteins characterized by two calcium-binding sites. They are involved in the inflammatory process, and are part of the epidermal differentiation complex [18]. S100A7 (psoriasin) is overexpressed in inflammatory diseases, and this overexpression contributes to dysregulated differentiation in psoriasis [19]. Likewise, we found that not only S100A7, but also S100A2, S100A9, S100A11, S100A12, S100A14 and S100A16 showed very significantly higher gene array values in psoriasis biopsies than in controls (Table S1). 3.1.5. Chemokines In psoriasis, keratinocytes and immune cells produce a broad repertoire of chemokines, most of which are upregulated [20,21]. Table S1 shows the significant, 10-fold up-regulation of CXCL9, CXCL10 and CXCR4 (from the CXC chemokine family), and CCL20, CCL22 and CCR7 (from the CC chemokine family) in psoriatic tissue. We also show a psoriatic decrease of 16-fold in the gene signal array of CCL27, which agrees with the differential downexpression of the chemokine CCL27 gene in psoriasis [22]. 3.1.6. The major histocompatibility complex (MHC) locus The major genetic determinant of psoriasis is found in the MHC locus (PSORS1, chromosome 6). However, within the MHC locus, we found a significant increase (4-fold) of the TAP2 gene signal in psoriatic arrays (Table S1). The membrane-associated protein encoded by TAP2 is a member of the superfamily of ATP-binding cassette (ABC) transporters, which is involved in antigen presentation [23]. The association of TAP alleles with psoriasis has been described [24,25]. 3.1.7. Antimicrobial peptides (AMPs) The induction of AMPs is markedly increased in psoriasis [26– 28]. Cationic cathelicidins are increased in psoriasis [1]. The case of cathelicidin LL-37 (coded by the CAMP gene) is well described [29]. However, we found just a moderate increase of the CAMP gene signal in psoriatic arrays. Lipocalins base their AMP function on binding to bacterial siderophores [30] and are consistently upregulated in psoriasis [31]. Here we show that lipocalin-2 (LCN2) is 14-fold higher in the psoriatic groups than in controls. Defensins are small cysteine-rich cationic proteins which are active against bacteria, fungi and many enveloped and nonenveloped viruses [28]. Psoriasis is associated with defensin upregulation [32,33]. Here we report a high increase in the gene array signal of the beta defensins DEFB4A,B (about 19-fold) and DEFB103A,B (about 4-fold), due to psoriasis. At this point, we also point out the dramatic increase (about 20-fold) in the gene expression in the psoriatic group in comparison to the control group of genes encoding lactotransferrin (LTF) and lysozyme (LYZ). 3.2. Psoriasis. What else? 3.2.1. Cell cycle We were interested in the effect of psoriasis on the expression of cyclins and cyclin-dependent kinase coding genes. Table S1
shows the increase and decrease in hybridization array signals of genes encoding these proteins. We point out the signal gene values of the cyclins B1, A2 and E2 (CCNB1, CCNA2, CCNE2, respectively), and the cyclin-dependent kinase (CDK1), which were much more highly expressed in psoriatic tissue (about 10-fold). With the dysregulation of these cyclin coding genes, it is plausible to think that psoriasis affects strategic points of the cell cycle. Cyclin E2 plays a role in the G1/S portion of the cell cycle. Cyclin A2 is synthesized at the onset of the S phase and is localized in the nucleus. And cyclin B1 is a regulatory protein involved in mitosis (provided by NCBI, RefSeq). Interestingly, we found that the effect of psoriasis on the cyclin B/Cdk1 kinase complex is also linked to the effect on the two B-type lamin proteins (LMNB1 and LMNB2) (Table S1), which are also extremely up-regulated. Lamins are components of the fibrillar network proteins inside the nucleus (nuclear lamina), attached to the inner nuclear membrane, and the cyclin B/Cdk1 protein kinase complex initiates lamin disassembly (provided by NCBI, RefSeq). Interestingly, we have also observed a significant increase in MYC gene expression in arrays of psoriatic tissue. This agrees with the effect of the proto-oncogene MYC, which promotes differentiation and cell cycle deregulation in human keratinocytes [34]. In fact, uncontrolled keratinocyte proliferation is a hallmark of psoriatic skin lesions [1,2]. 3.2.2. Dynamin Psoriatic skin has long been known to contain elevated levels of interferon types [21]. So, we focused on the interferon-induced dynamins. Dynamins are a superfamily of large multidomain mechanochemical GTPases (EC 3.6.5.5), which are coordinately involved in the fission and fusion of membrane structures [35]. The myxovirus resistance proteins (MX) of the dynamin family are induced by type I interferons and block the replication of a broad spectrum of viruses at various sites in the cell [36]. We found (Table S1) that the two MX coding genes (MX1 and MX2) were upregulated (about 10-fold) in psoriasis biopsy samples in comparison with controls. Similarly, the dynamin family guanylate-binding proteins (GBPs) are induced by type II interferons and antiinflammatory cytokines [37]. We also found (Table S1) that all GBP genes (GBP1-6) were extremely highly regulated in psoriatic skin, and reached values up to 15-fold higher than controls. As of the interferon-inducible GTPase family, both MX and GBP dynamins have an antiviral effect and play a role in intracellular defence [38,39]. We also note that the effect of the disease on the dynamin system also reaches other dynamin families, covering such fundamental processes as mitochondrial dynamics [40]. We emphasize the 7-fold increase of the array signal value of the mitofusin MFN1 gene (Table S1), which is involved in mitochondrial fusion. Further evidence of the consistent association between the dynamin system and the pathogenesis of psoriasis is the significant increase in the expression of the dynamin-2 gene (DNM2). Because this study is based on multicellular data (biopsies from healthy controls and disease tissue) but not a single type of cells, and it is known that the dynamin system acts in keratinocytes in a different scenario of psoriasis, such as viral infection [41,42], we suggest that the effects of psoriasis on the expression of dynamin-related genes take place at the keratinocyte level. 3.2.3. Antioxidants The role of oxidative stress in various stages of psoriasis has been described [43,44]. Oxidative stress involves not only ROS, but also the antioxidants that control their presence. In this context, we found a significant down-regulation (about 7-fold) of genes encoding key antioxidant enzymes (that is, catalase, CAT; superoxide dismutases, SOD1-3; and glutathione reductase, GSR) (Table S1). The significant negative effect of psoriasis on catalase, superoxide dismutase and glutathione reductase gene
Please cite this article in press as: J.L. Melero, et al., Deciphering psoriasis. A bioinformatic approach, J Dermatol Sci (2017), https://doi.org/ 10.1016/j.jdermsci.2017.11.010
G Model DESC 3284 No. of Pages 7
J.L. Melero et al. / Journal of Dermatological Science xxx (2017) xxx–xxx
5
Fig. 2. Location of selected gene clusters in the human genome. Distribution of gene clusters in the human genome, containing genes whose expression is significantly affected by psoriasis in the affected areas of the skin. These genes have been commented on in the text. Schemes of the chromosomes have been taken from the NCBI (Human Genome Resources). The cytogenetic location is used. For each gene cluster, denoted in red, we give a reference coordinate and the name of the protein or the protein family encoded by the genes of the cluster.
Please cite this article in press as: J.L. Melero, et al., Deciphering psoriasis. A bioinformatic approach, J Dermatol Sci (2017), https://doi.org/ 10.1016/j.jdermsci.2017.11.010
G Model DESC 3284 No. of Pages 7
6
J.L. Melero et al. / Journal of Dermatological Science xxx (2017) xxx–xxx
expressions is a typical oxidative stress scenario, which may be an additional factor in the pathogenesis of the disease. Psoriasin (S100A7) increases the expression of ROS in keratinocytes [45]. This correlates with the significant high values of S100A7 gene expression found in lesional samples, and is consistent with the down-regulation of the key antioxidant enzymes (Table S1). 3.2.4. Kallikreins The detailed analysis of gene expression differences between both healthy and psoriatic experimental groups, on the one hand and the positions occupied by highly expressed or silenced genes in the genome, on the other, led us to discover that kallikreins are involved in psoriatic pathogenicity. The subgroup of serine proteases is involved in several physiological processes, including normal skin desquamation [46–48]. The link between kallikreins and psoriasis has not yet been described. Here we show that in psoriasis there is a significant up-regulation (about 9-fold) of a set of kallikrein genes (KLK1-KLK13) in the cells of the psoriaticinjured epithelial tissue. In agreement with our results, there is some evidence to support the kallikreins-psoriasis association. For example, kallikrein-mediated proteolysis regulates the antimicrobial effects of cathelicidins in skin [49]. Also, kallikrein-related peptidase-8 (KLK8) is an active serine protease in human epidermis and sweat and is involved in a skin barrier proteolytic cascade [50]. And finally, kallikrein 5 mediated inflammation in rosacea [51]. Psoriasis has an effect on a wide variety of genes with a wide range of biological functions. As we have just seen, these genes range from those that are well known as mediators of the pathogenesis of the disease to genes whose biological function has not yet been associated with psoriasis. A detailed analysis of Table S1 will enable the list of genes whose expression may be affected by psoriasis to be extended. From the information in Table S1, we observe that the effect of psoriasis extends across all regions of the chromosomes of the human genome. Psoriasis is also known to have a strong hereditary character and a complex genetic basis [1–3]. In fact, “heredity” and “genetic predisposition” are different issues. If we add the sensitivity of psoriasis to environmental factors [6], and combine everything with the results we show here, it appears that psoriasis could be a disease of epigenetic origin. In psoriatic patients, the phenotype of the lesional skin tissue is stable and mitotically heritable. We need to consider that healthy and affected cells share the same genome. These are the credentials of an epigenetic context [52]. This epigenetic view is supported by the fact that in many cases, the effect of psoriasis on gene expression affects not only a single gene but also a gene cluster. Fig. 2 illustrates this phenomenology with genes discussed in this section. Other gene clusters can be discovered by a detailed analysis of Table S1. These patterns of changes in gene expression within chromosome regions seem to reflect what happens in imprinted chromosomal regions, in their “control regions” [53]. However, unlike in imprinting, in psoriasis the two homologous chromosomes are affected. We suggest that the genome of a psoriasis patient contains some epigenetic imprints (e.g. differentially methylated regions, histone modifications, non-coding RNAs, insulators, and so on) that coincide in the epidermis and, in conjunction with environmental factors, lead to the manifestation of psoriasis. Other authors, C. M. Nguyen and W. Liao [54], have reviewed the possible evidence of the epigenètic phenomenon “imprinting” in psoriasis. However, this is not what we are suggesting. Of the environmental factors (so important in epigenetics) the skin’s bacterial flora could play an important role. This may explain why psoriasis is present in some regions of the skin, and not in others. This epigenetic view of psoriasis fits well with the heritable nature of the disease. In this context,
various reviews have been published about the role of epigenetics in psoriasis [55–60]. Finally, we believe that the complex pathogenesis of psoriasis can be studied using an epigenetic approach. Conflict of interest statement The authors declare that there are no conflicts of interest. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at https://doi.org/10.1016/j. jdermsci.2017.11.010. References [1] G.K. Perera, P. DiMeglio, F.O. Nestle, Psoriasis, Annu. Rev. Pathol. 7 (2012) 385– 422. [2] F.C. Eberle, J. Brück, J. Holstein, K. Hirahara, K. Ghoreschi, Recent advances in understanding psoriasis, F1000Res. (2016) 5. [3] J. Baliwag, D.H. Barnes, A. Johnston, Cytokines in psoriasis, Cytokine 73 (2015) 342–350. [4] P. Di Meglio, F. Villanova, F.O. Nestle, Psoriasis, Cold Spring Harb. Perspect. Med. 4 (8) (2014) (pii: a015354). [5] H. Alexander, F.O. Nestle, Pathogenesis and immunotherapy in cutanis psoriasis: what can rheumatologists learn? Curr. Opin. Rheumatol. 29 (2017) 71–78. [6] E. Dika, F. Bardazzi, R. Balestri, H.I. Maibach, Environmental factors and psoriasis, Curr. Probl. Dermatol. 35 (2007) 118–135. [7] E. Proksch, J.M. Brandner, J.M. Jensen, The skin: an indispensable barrier, Exp. Dermatol. 17 (2008) 1063–1072. [8] E. Fuchs, Keratins and the skin, Annu. Rev. Cell. Dev. Biol. 11 (1995) 123–153. [9] L. Ramms, G. Fabris, R. Windoffer, N. Schwarz, R. Springer, C. Zhou, et al., Keratins as the main component for the mechanical integrity of keratinocytes, Proc. Natl. Acad. Sci. U. S. A. 110 (2013) 18513–18518. [10] I.M. Leigh, H. Navsaria, P.E. Purkis, I.A. McKay, P.E. Bowden, P.N. Riddle, Keratins (K16 and K17) as markers of keratinocyte hyperproliferation in psoriasis in vivo and in vitro, Br. J. Dermatol. 133 (1995) 501–511. [11] S. Kezic, I. Jakasa, Filaggrin and skin barrier function, Curr. Probl. Dermatol. 49 (2016) 1–7. [12] T. Makino, M. Mizawa, T. Yamakoshi, M. Takaishi, T. Shimizu, Expression of filaggrin-2 protein in the epidermis of human skin diseases: a comparative analysis with filaggrin, Biochem. Biophys. Res. Commun. 449 (2014) 100–106. [13] E. Candi, R. Schmidt, G. Melino, The cornified envelope: a model of cell death in the skin, Nat. Rev. Mol. Cell. Biol. 6 (2005) 328–340. [14] P.M. Steinert, E. Candi, T. Kartasova, L. Marekov, Small proline-rich proteins are cross-bridging proteins in the cornified cell envelopes of stratified squamous epithelia, J. Struct. Biol. 122 (1998) 76–85. [15] H. Iizuka, H. Takahashi, M. Honma, A. Ishida-Yamamoto, Unique keratinization process in psoriasis: late differentiation markers are abolished because of the premature cell death, J. Dermatol. 31 (2004) 271–276. [16] N.M. de Vasconcelos, N. Van Opdenbosch, M. Lamkanfi, Inflammasomes as polyvalent cell death platforms, Cell. Mol. Life. Sci. 73 (2016) 2335–2347. [17] K. Takahashi, A. Nishida, M. Shioya, H. Imaeda, S. Bamba, O. Inatomi, et al., Interleukin (IL)-1? Is a strong inducer of IL-36? Expression in human colonic myofibroblasts, PLoS One 10 (11) (2015) e0138423. [18] M. Kypriotou, M. Huber, D. Hohl, The human epidermal differentiation complex: cornified envelope precursors, S100 proteins and the ‘fused genes' family, Exp. Dermatol. 21 (2012) 643–649. [19] A.K. Ekman, J. Vegfors, C.B. Eding, C. Enerbäck, Overexpression of psoriasin (S100A7) contributes to dysregulated differentiation in psoriasis, Acta Derm. Venereol. 97 (2017) 441–448. [20] M. Sugaya, Chemokines and skin diseases, Arch. Immunol. Ther. Exp. (Warsz) 63 (2015) 109–115. [21] T. Cao, S. Shao, B. Li, L. Jin, J. Lei, H. Qiao, et al., Up-regulation of Interferoninducible protein 16 contributes to psoriasis by modulating chemokine production in keratinocytes, Sci. Rep. 6 (2016) 25381. [22] J.L. Riis, C. Johansen, C. Vestergaard, R. Bech, K. Kragballe, L. Iversen, Kinetics and differential expression of the skin-related chemokines CCL27 and CCL17 in psoriasis, atopic dermatitis and allergic contact dermatitis, Exp. Dermatol. 20 (October (10)) (2011) 789–794. [23] M. Naderi, M. Hashemi, S. Amininia, Association of TAP1 and TAP2 gene polymorphisms with susceptibility to pulmonary tuberculosis, Iran. J. Allergy Asthma Immunol. 15 (2016) 62–68. [24] H. Saeki, S. Kuwata, H. Nakagawa, A. Asahina, K. Tamaki, Y. Shibata, et al., Analysis of HLA class II and TAP alleles in Japanese patients with psoriasis vulgaris, Hum. Immunol. 59 (1998) 503–511. [25] C.W. Pyo, S.S. Hur, Y.K. Kim, T.Y. Kim, T.G. Kim, Association of TAP and HLA-DM genes with psoriasis in Koreans, J. Invest. Dermatol. 120 (2003) 616–622.
Please cite this article in press as: J.L. Melero, et al., Deciphering psoriasis. A bioinformatic approach, J Dermatol Sci (2017), https://doi.org/ 10.1016/j.jdermsci.2017.11.010
G Model DESC 3284 No. of Pages 7
J.L. Melero et al. / Journal of Dermatological Science xxx (2017) xxx–xxx [26] G. Wang, Human antimicrobial peptides and proteins, Pharmaceuticals (Basel) 7 (2014) 545–594. [27] M. Marcinkiewicz, S. Majewski, The role of antimicrobial peptides in chronic inflammatory skin diseases, Postepy Dermatol. Alergol. 33 (2016) 6–12. [28] T. Takahashi, R.L. Gallo, The critical and multifunctional roles of antimicrobial peptides in dermatology, Dermatol. Clin. 35 (2017) 39–50. [29] R. Lande, E. Botti, C. Jandus, D. Dojcinovic, G. Fanelli, C. Conrad, et al., The antimicrobial peptide LL37 is a T-cell autoantigen in psoriasis, Nat. Commun. 5 (2014) 5621. [30] M. Jung, C. Mertens, R. Bauer, C. Rehwald, B. Brüne, Lipocalin-2 and iron trafficking in the tumor microenvironment, Pharmacol. Res. 120 (2017) 146– 156. [31] S. Shao, T. Cao, L. Jin, B. Li, H. Fang, J. Zhang, et al., Increased lipocalin-2 contributes to the pathogenesis of psoriasis by modulating neutrophil chemotaxis and cytokine secretion, J. Invest. Dermatol. 136 (2016) 1418–1428. [32] W.H. Boehncke, The alpha-defensins HNP-1 and HNP-2 are dominant selfpeptides presented by HLA class-II molecules in lesional psoriatic skin, Eur. J. Dermatol. 14 (2004) 142–145. [33] P.E. Stuart, U. Hüffmeier, R.P. Nair, R. Palla, T. Tejasvi, J. Schalkwijk, et al., Association of?-defensin copy number and psoriasis in three cohorts of European origin, J. Invest. Dermatol. 132 (2012) 2407–2413. [34] A. Gandarillas, The mysterious human epidermal cell cycle or an oncogeneinduced differentiation checkpoint, ABBV Cell Cycle 11 (2012) 4507–4516. [35] G.J. Praefcke, H.T. McMahon, The dynamin superfamily: universal membrane tubulation and fission molecules? Nat. Rev. Mol. Cell Biol. 5 (2004) 133–147. [36] O. Haller, G. Kochs, Interferon-induced mx proteins: dynamin-like GTPases with antiviral activity, Traffic 10 (2002) 710–717. [37] D. Pilla-Moffett, M.F. Barber, G.A. Taylor, J. Coers, Interferon-inducible GTPases in host resistance, inflammation and disease, J. Mol. Biol. 428 (2016) 3495– 3513. [38] D.P. Barlow, M.S. Bartolomei, Genomic imprinting in mammals, Cold Spring Harb. Perspect. Biol. 6 (2014) (pii: a018382). [39] O. Haller, P. Staeheli, M. Schwemmle, G. Kochs, Mx GTPases: dynamin-like antiviral machines of innate immunity, Trends Microbiol. 23 (2015) 154–163. [40] H. Lee, Y. Yoon, Mitochondrial fission and fusion, Biochem. Soc. Trans. 44 (2016) 1725–1735. [41] E. Rahn, P. Petermann, M.J. Hsu, F.J. Rixon, D. Knebel-Mörsdorf, Entry pathways of herpes simplex virus type 1 into human keratinocytes are dynamin- and cholesterol-dependent, PLoS One 10 (2011) e25464. [42] J.L. Smith, S.K. Campos, M.A. Ozbun, Human papillomavirus type 31 uses a caveolin 1- and dynamin 2-mediated entry pathway for infection of human keratinocytes, J. Virol. 81 (2007) 9922–9931. [43] X. Lin, T. Huang, Oxidative stress in psoriasis and potential therapeutic use of antioxidants, Free Radic. Res. 50 (2016) 585–595. [44] S. Srivastava, D. Singh, S. Patel, M.R. Singh, Role of enzymatic free radical scavengers in management of oxidative stress in autoimmune disorders, Int. J. Biol. Macromol. 101 (2017) 502–517.
7
[45] E. Shubbar, J. Vegfors, M. Carlström, S. Petersson, C. Enerbäck, Psoriasin (S100A7) increases the expression of ROS and VEGF and acts through RAGE to promote endothelial cell proliferation, Breast Cancer Res. Treat. 134 (2012) 71– 80. [46] A. Eissa, E.P. Diamandis, Human tissue kallikreins as promiscuous modulators of homeostatic skin barrier funcions, Biol. Chem. 389 (2008) 669–680. [47] M. Kishibe, Y. Bando, R. Terayama, K. Namikawa, H. Takahashi, Y. Hashimoto, et al., Kallikrein 8 is involved in skin desquamation in cooperation with other kallikreins, J. Biol. Chem. 282 (2007) 5834–5841. [48] A.C. Stefanini, B.R. daCunha, T. Henrique, E.H. Tajara, Involvement of kallikreinrelated peptidases in normal and pathologic processes, Dis. Markers (2015) 946572. [49] K. Yamasaki, J. Schauber, A. Coda, H. Lin, R.A. Dorschner, N.M. Schechter, et al., Kallikrein-mediated proteolysis regulates the antimicrobial effects of cathelicidins in skin, FASEB J. 20 (2006) 2068–2080. [50] A. Eissa, V. Amodeo, C.R. Smith, E.P. Diamandis, Kallikrein-related peptidase-8 (KLK8) is an active serine protease in human epidermis and sweat and is involved in a skin barrier proteolytic cascade, J. Biol. Chem. 286 (2011) 687– 706. [51] A.M. Two, J.Q. Del Rosso, Kallikrein 5-mediated inflammation in rosacea: clinically relevant correlations with acute and chronic manifestations in rosacea and how individual treatments may provide therapeutic benefit, J. Clin. Aesthet. Dermatol. 7 (2014) 20–25. [52] S.L. Berger, T. Kouzarides, R. Shiekhattar, A. Shilatifard, An operational definition of epigenètics, Genes. Dev. 23 (2009) 781–783. [53] D.P. Barlow, M.S. Bartolomei, Genomic imprinting in mammals, Cold Spring Harb. Perspect. Biol. 6 (2014) (pii: a018382). [54] C.M. Nguyen, W. Liao, Genomic imprinting in psoriasis and atopic dermatitis: a review, J. Dermatol. Sci. 80 (2015) 89–93. [55] P. Zhang, Y. Su, Q. Lu, Epigenetics and psoriasis, J. Eur. Acad. Dermatol. Venereol. 26 (2012) 399–403. [56] J.E. Gudjonsson, G. Krueger, A role for epigenetics in psoriasis: methylated cytosine-guanine sites differentiate lesional from nonlesional skin and from normal skin, J. Invest. Dermatol. 132 (2012) 506–508. [57] P. Zhang, M. Zhao, G. Liang, G. Yin, D. Huang, F. Su, et al., Whole-genome DNA methylation in skin lesions from patients with psoriasis vulgaris, J. Autoimmun. 41 (2013) 17–24. [58] R.M. Trowbridge, M.R. Pittelkow, Epigenetics in the pathogenesis and pathophysiology of psoriasis vulgaris, J. Drugs Dermatol. 13 (2014) 111–118. [59] F. Zhou, C. Shen, J. Xu, J. Gao, X. Zheng, R. Ko, et al., Epigenome-wide association data implicates DNA methylation-mediated genetic risk in psoriasis, Clin. Epigenet. 8 (2016) 131. [60] R.A. Pollock, F. Abji, D.D. Gladman, Epigenetics of psoriatic disease: a systematic review and critical appraisal, J. Autoimmun. 78 (2017) 29–38.
Please cite this article in press as: J.L. Melero, et al., Deciphering psoriasis. A bioinformatic approach, J Dermatol Sci (2017), https://doi.org/ 10.1016/j.jdermsci.2017.11.010