Dyspigmented Hypertrophic Scars: beyond Skin Color

Abstract

Although pigment synthesis is well understood, relevant mechanisms of psychologically debilitating dyspigmentation in nascent tissue after cutaneous injuries are still unknown. Here, differences in genomic transcription of hyper- and hypopigmented tissue relative to uninjured skin were investigated using a red Duroc swine scar model. Transcription profiles differed based on pigmentation phenotypes with a trend of more upregulation or downregulation in hyper- or hypopigmented scars, respectively. Ingenuity Pathway Analysis (IPA) of significantly modulated genes in both pigmentation phenotypes showed pathways related to redox, metabolic, and inflammatory responses were more present in hypopigmented samples, while those related to stem cell development differentiation were found mainly in hyperpigmented samples. Cell-cell and cell-extracellular matrix interactions and inflammation responses were predicted (z-score) active in hyperpigmented and inactive in hypopigmented. The proinflammatory high-mobility group box 1 (HMGB1) pathway showed the opposite trend. Analysis of differentially regulated mutually exclusive genes showed an extensive presence of metabolic, proinflammatory, and oxidative stress pathways in hypopigmented scars, while melanin synthesis, glycosaminoglycans (GAGs) biosynthesis and cell differentiation pathways were predominant in hyperpigmented scar. Several potential therapeutic gene targets have been identified.

1. Introduction:

Increases in the number of cases of hypertrophic scars due to improved burn care (Pereira et al., 2006) have made post-burn complications, including hypertrophic scars (HTS) (Bombaro et al., 2003) and dyspigmentation, the next challenge in burn injury rehabilitation and recovery (Fidel-Kinori et al., 2016). Skin color is constituted of a mix of four pigments: oxygenated hemoglobin, deoxyhemoglobin, various carotenes (Britton, 1995), and melanin (Lin & Fisher, 2007). The latter, which is the strongest color determinant, has two main types: a brown-black eumelanin pigment and a pink-red pheomelanin pigment (Liu et al., 2005). Secondary types of melanin include the low molecular weight trichochromes and the neuromelanin pigments produced extensively by catecholaminergic neurons of the human brain in quantities higher than in all other primates (Double et al., 2002; Fedorow et al., 2005). Skin color is affected by the redox status of hemoglobin and blood perfusion (Stamatas & Kollias, 2004) which produces a flushed or pale skin appearance when increased or decreased, respectively. It is affected also by carotene, which is an orange-yellow pigment that accumulates in epidermal cells and fatty tissue of the dermis upon consumption of carotene-rich food such as carrots and oranges (Pezdirc et al., 2016).

Melanin is produced primarily by melanocytes in a specialized organelle called melanosomes that protect the cell from the toxic oxidation environment during synthesis (Marrot, Belaidi, Meunier, Perez, & Agapakis-Causse, 1999). The process is mediated by several enzymes, namely, tyrosinase (TYR), tyrosinase related protein-1 (TYRP-1), and TYRP-2, or DOPAchrome tautomerase (DCT) (Plonka et al., 2009). The initial building blocks of melanin pigments, phenylalanine and tyrosine, are hydroxylated by TYR to produce dihydroxyphenylalanine (DOPA), then oxidized by the same enzyme to DOPAquinone (Fitzpatrick, Miyamoto, & Ishikawa, 1967). This is the rate-limiting step in melanin synthesis because all subsequent steps including cyclization to give DOPAchrome, tautomerization to 5,6-dihyroxyindole-2-carboxilic acid (DHICA) by DCT, and oxidation by TYRP-1 to produce eumelanin or the decarboxylation and oxidation by TYR are almost spontaneous. Alternatively, DOPAquinone could bind cysteine to yield cysteinyldopa, which transforms to the bicyclic molecule alanyl-hydroxy-benzothiazine and finally produces pheomelanin (Hearing, 2011; Simon, Peles, Wakamatsu, & Ito, 2009). The mechanisms of switching synthesis between these two melanin types remain unknown; however, the G-coupled receptor melanocortin 1 receptor (MC1R) has been implicated (Abdel-Malek et al., 2000). Activation of MC1R is initiated by the adrenocorticotropic hormone (ACTH) and the alpha melanocyte-stimulating hormone (α-MSH) produced by keratinocytes in response to DNA stress (Costin & Hearing, 2007; Plonka et al., 2009). Typical stressors of keratinocytes, such as UV light, ionizing radiation, systemic or topical exposures to chemicals, replication errors, and metabolic abnormalities, cause DNA damage and activation of DNA repair and cell survival and protection mechanisms. If the damage is severe enough, cells succumb to apoptosis (Haupt, Berger, Goldberg, & Haupt, 2003; McGill et al., 2002). Among multiple activities, tumor suppressor protein 53 (TP53) gene is expressed, and the TP53 protein is produced to eventually lead to the synthesis of pro-opiomelanocortin (POMC), which is the precursor of ACTH and α-MSH (Cui et al., 2007). Activation of MC1R by these hormones leads to increased cyclic adenosine monophosphate (cAMP) by increasing adenylyl cyclase (Rouzaud, Kadekaro, Abdel-Malek, & Hearing, 2005). Protein kinase A (PKA) is activated by c-AMP to phosphorylate the c-AMP response element binding protein (CREB) in the nucleus, which then binds the CREB responsive element (CRE) to produce a key activator of melanin synthesis and production of TYR, TYRP1, and TYRP2, microphthalmia transcription factor (MITF) (Miyamura et al., 2007; Slominski, Tobin, Shibahara, & Wortsman, 2004).

Melanin synthesis can be inhibited by antagonists of MC1R such as the agouti signaling protein (ASP) (Abdel-Malek et al., 2000; Lu et al., 1994); however, other melanocyte receptors may signal inadvertent melanin synthesis. For example, MITF can be activated (i.e., phosphorylated) by mitogen activated protein kinase (MAPK) secondary to the activation of tyrosine protein kinase mast/stem cell growth factor receptor (c-kit or CD117) by stem cell factor (SCF) (Costin & Hearing, 2007). Similarly, altering the activity of melanin synthesis enzymes affects melanin production. For example, increasing adenylcyclase activity by epinephrine binding to the β-adrenergic receptor (PDEI4) (Grando, Pittelkow, & Schallreuter, 2006; Kauser, Schallreuter, Thody, Gummer, & Tobin, 2003) or inhibiting its enzymatic function upon the activation of the bone matrix protein receptors (BMPR) by UV light results in cAMP concentration changes that ultimately change melanin synthesis (Park et al., 2009). Modulating the stability of cAMP also by phosphodiesterase 4D3 inhibitors (PDEI4), which inhibit cAMP breakdown, leads to increased melanin synthesis (Alvarez et al., 1995). Many dyspigmentation diseases have been linked to mutated genes encoding proteins or enzymes involved in pigment synthesis (Yamaguchi & Hearing, 2014).

Although pigment synthesis is fully elucidated, little is known about dyspigmentation after burns, and the underlying biology remains elusive (Acikel, Ulkur, & Guler, 2000; Kahn & Cohen, 1996; Mulekar, Issa, & Eisa, 2011). The multifaceted nature of the melanin synthesis pathway involving multiple cell types and a large number of polymorphic and pleiotropic genes, in addition to the difficulties emanating from activities of wound healing processes, makes studying this particular health burden challenging.

In previous work, dyspigmented hypertrophic scars showing hyper- and hypopigmented regions were produced in red Duroc swine (Alkhalil et al., 2015; Carney et al., 2017; Tejiram et al., 2016), and pigmentation phenotypes were shown to be due to increased or absent melanin synthesis (Travis, Ghassemi, et al., 2014), respectively. Melanocytes were also shown to be present in comparable densities both in dyspigmented and in normal pigmentation phenotypes (Travis, Ghassemi, et al., 2014). Here, the biological differences between these phenotypes were investigated for differences in transcriptome profiles using microarrays. Pathway analysis was performed using stringent selection criteria to assess the underlying biological processes causing dyspigmentation.

2. Methods:

2.1. Animal Model

The red Duroc swine wound model was selected because of many similarities its skin shares with that of human, which made this animal models the prime choice for human wound healing and burn studies (Sullivan, Eaglstein, Davis, & Mertz, 2001; Zhu et al., 2003). Juvenile castrated male Duroc swine were used to minimize potential interference from the estrogen cycle and to reduce animal aggressiveness (Michael Swindle & Smith, 2008). Animals were handled according to facility standard operating procedures under the animal care and use program accredited by the Association for Assessment and Accreditation of Laboratory Animal Care International (AAALAC) and Animal Welfare Assurance through the Public Health Service (PHS). All described animal work was reviewed and approved by the MedStar Health Research Institute’s Institutional Animal Care and Use Committee (IACUC) under protocol number MHRI-IACUC 2014–009.

Each animal received one full thickness 4 inch x 4 inch (10.16 cm x 10.16 cm) wound on each flank. Wounds were created using a Zimmer dermatome (Zimmer, Ltd, Swindon, UK) and excised over the lateral thorax to a full thickness depth of 0.090 inches (3 passes each at 0.030 in). Details related to anesthesia, animal care, wound healing, and scar formation have been previously described (Alkhalil et al., 2015; Carney et al., 2017; Ghassemi, Travis, Moffatt, Shupp, & Ramella-Roman, 2014; Tejiram et al., 2016; Travis et al., 2015; Travis, Mino, et al., 2014).

Punch biopsies (3 mm) were collected of un-injured skin pre-excision and post-excision during weekly assessments. After wounds were completely reepithelialized, and areas of both hypopigmentation and hyperpigmentation were formed, biopsies were taken of these distinct areas during weekly assessments. These assessments continued through 133 days post-excision. Biopsies used for mRNA isolation were stored in Allprotect Tissue Reagent (Qiagen, Valencia, CA).

2.2. mRNA purification

Allprotect-preserved biopsies were used to isolate mRNA using the RNeasy Fibrous Tissue Kit based on the manufacturer’s protocol (Qiagen).

2.3. Microarray

A total of 28 biopsies were collected from two red Duroc pigs, and the transcriptome changes in these biopsies were evaluated during the course of wound healing and scar formation in the context of dyspigmentation. Four biopsies collected from the left and right flanks of two pigs before wound creation (un-injured skin) were used as a baseline reference of transcription. Eight biopsies were harvested on day 49 (4 per animal, 2 hyperpigmented and 2 hypopigmented). To evaluate the potential changes in dyspigmentation over time, eight biopsies were harvested on days 91–99 (mid-stage), and days 119–133 (late-stage) (4 per animal, 2 hyperpigmented and 2 hypopigmented). Microarray assays were performed using Agilent-026440 S. scrofa (Pig) Oligo Microarray v2 expression array (GE 4×44K v2 two color microarray) (Agilent Technologies, Inc., Santa Clara, CA) following the manufacturer’s protocol. Hybridized microarray slides were scanned using an Agilent Technologies Scanner G2505C US09493743. For feature extraction and normalization, images of scanned microarray slides were feature-extracted and normalized using Agilent’s feature extraction software, version 10.7 or later, in the default setup (Agilent Technologies, Inc.).

2.4. Data analysis

Feature-extracted data were filtered on flags to exclude probes with missing values in more than one sample and quantile normalized using the Limma Package (v3.28.21) of R v3.3.3 (www.r-project.org). Normalized data from all conditions were analyzed and compared for scar pigmentation phenotype. Initial lists of differentially regulated genes in each condition were identified using Moderated T-test at p < 0.05 of the Limma package of R.

Principal component analysis (PCA) was performed using prcomp function and ggplot2 package (v2.2.1) of R.

Ingenuity Pathway Analysis (IPA, Q4–2017) for enrichment of significantly differentially expressed genes in known canonical pathways was performed using the IPA library of canonical pathways. The analysis of complete data sets used differentially transcribed genes in mid and late time points of hyper- and hypopigmented tissues (8 samples each). Genes in these lists met expression fold change > 1.3 and p-values < 0.01. The pathways resulting from hypo- and hyperpigmented tissues were compared by sorting them into three separate groups. The first and second groups included pathways that were exclusive to either of the tissue phenotypes. The third group included pathways commonly found in both tissue phenotypes. Significant pathways in the last group were considered when showing at least one log difference and twice the p-value in one tissue relative to the other. The same analysis was used for assessments based on z-score (i.e., positive and negative z-score suggests activated or inactivated pathway, respectively). Again pathways were sorted into three groups. The first and second included pathways exclusively identified with a z-score value in one and not the other. The third group included pathways showing a z-score with a similar sign (i.e., positive or negative). Levels of activity in the latter group were considered non-significant if the difference was less than a single z-score unit or less than twice the z-score value for a given pathway in one tissue phenotype relative to the other. Pathways showing staggered z-score values were excluded from meeting the second selection criterion (i.e., an absolute value in one tissue phenotype twice that in the other).

Another IPA analysis was performed using data subsets of genes that were exclusive to either hyper- or hypopigmented tissues and passing an expression fold ratio of 1.3 (FC) in one of the two tissue phenotypes and p-value <0.01). Pathways resulting from this analysis were compared by sorting them into three groups of unique pathways to each tissue phenol type, and common pathways in both that passed double the p-value in one relative to the other and a difference of at least one log p-value. For analysis based on z-scores, lists of exclusive genes with p-value < 0.01 were used.

2.5. En face staining of epidermal sheets

Hyperpigmented and hypopigmented epidermal sheets were stained with a melanocyte marker, S100 calcium binding protein B (S100β), and a marker of melanocyte activation and melanin synthesis, tyrosinase related protein 1 (TYRP1) by an en face staining technique adapted from a previously described method(Millonig, Niederegger, & Wick, 2001).On day 98, hypertrophic scars were excised from the animal. In a laminar flow hood, the scar was cut into regions of hyperpigmentation and hypopigmentation. The small scar pieces were then incubated in 1X dispase solution (CELLnTEC, Switzerland) with 10 µg/mL gentamycin and 0.25 µg/mL amphotericin B overnight at 4^oC. The next day, the scar pieces were removed from dispase, rinsed with PBS, and epidermal sheets were obtained by peeling the epidermis from the dermis. Epidermal sheets were fixed in ice-cold methanol for 10 minutes. The sheets were then rehydrated in PBS for 30 minutes, followed by blocking in 3% non-fat dry milk in PBS for 1 hour. Sheets were then incubated with primary antibody diluted in 3% milk (S100β at 1:50 and TYRP1 at 1:250) (Abcam, Cambridge, MA). The following day, sheets were rinsed with 1% BSA in PBS for 1 hour. The secondary antibody was then applied (anti-mouse-CY5 and anti-rabbit-CY3 at 1:100) (Abcam) and incubated for an hour. The sheets were subsequently washed again, and stained for 10 minutes with DAPI (1 µg/mL), rinsed in distilled water for 5 minutes, and lastly, the sheets were mounted on slides with the basal side of the sheet facing upwards and fluoroshield (Sigma Aldrich, St. Louis, MO) was added. No positive staining was identified in sheets stained in absence of primary antibody. Sheets were viewed under DAPI, CY3, and CY5 fluorescent filters to create a composite image using a Zeiss microscope with Zen software (Zeiss, Oberkochen, Germany).

3. Results

Wounds were completely reepithelialized by day 42. Heterogeneity in the pigmentation of newly generated tissue was noted at the periphery of the wound by day 21 and before wound closure. Abnormal pigmentation started as faint grayish spots by day 14, then gradually radiated towards the center of the wound at a pace slower than the leading reepithelialization edge (Figure 1). Spots became increasingly distinct by day 28, resulting in more intense contrast with the rest of the newly generated skin devoid of melanin that was colored mainly by the perfusing blood. The tone of the spots developed to closely match that of the uninjured skin, albeit the intensity was clearly darker by day 42. Spots started to coalesce at decreasing rates, reaching near plateau by 10–13 weeks post-injury where little changes to size, color tone, and intensity of hyperpigmented areas were observed thereafter. Rarely, dark colored islands appeared within the hypopigmented skin and followed similar color changes of hyperpigmented areas surrounding the wound (Figure 2A). The progression of these islands of pigmentation and synchronicity with the hyperpigmented areas on wound circumference suggest a shared origin. Retained bits of the hair follicles at 0.1 inch or deeper contain differentiated melanocytes and melanoblasts that would potentially give rise to such hyperpigmented islands. The color of hypopigmented scars was determined mainly by underlying blood flow and oxidated/reduced hemoglobin balance. In general, scar segments closer to the center were hypopigmented, and the majority of hyperpigmented scar areas were on the periphery (Figure 2B).

Figure 1.

Schematic Representation of Study Design and Biopsy Collection. Biopsies from hyper- and hypopigmented scar tissues were interrogated for transcriptomic differences using microarrays.

Figure 2.

Dyspigmentation in a red Duroc Model of Hypertrophic Scarring. From left to right, normal skin pre-injury, through day 126 post-injury (A). Completely reepithelialized wound at day 42 showing streaks of dyspigmented nascent tissue prior to hypertrophic scar development (B).

Principal component analysis captured sufficient transcriptomic differences to sort scars according to pigmentation phenotypes

Principal component analysis of transcriptomes from matured dyspigmented tissues (days 91–133) separated samples based on pigmentation phenotypes (Figure 3A). Biopsies collected as early as day 49, right after completion of reepithelialization, showed similar pigmentation-based separation, confirming that causes of dyspigmentation occur early during healing and prior to scar development and associated disorders in extracellular matrix composition (Figure 3A). Heat maps showed variations in regulation of scores of genes leading to samples clustering in accordance with pigmentation phenotype. Heat map analysis detected distinctive regulation patterns for multiple genes in hyper and hypopigmented samples (Figure 3B). The regulation patterns were stable throughout the experiment, indicating that dyspigmentation is not transient (Figure 3B).

Figure 3.

Transcriptome Profiles Separated Biopsies Based on Pigmentation Phenotypes. (A) Principal component analysis (PCA) and (B) heatmap of top differentially regulated genes in dyspigmented scar tissues at different post-injury phases.

Note: If the upper complete integer number occurred, it was not included in the bin range

These results validated the quality and suitability of the microarrays data for further analysis to assess the biological events underlying pigment synthesis disruption in scars. Because dyspigmentation was observed early before HTS development and the tone of dyspigmentation stabilized during mid and late experiment phases, analysis focused on the latter time points, and samples were pooled into hyper- or hypopigmented groups regardless of any changes to scars.

Scar hypo- and hyperpigmentations are associated with relative increases in downregulation and upregulation, respectively

The number of array elements normalized to uninjured skin (FC > 1.3 and p-values < 0.01) showed a small difference of 106 elements representing 1.47% and 1.49% of total differentially transcribed ones in hyper- and hypopigmented tissue (7115 in hypopigmented and 7221 in hyperpigmented biopsies). Bigger differences were observed in the regulation mode of these elements where a histogram (a single log₂ FC bin size) showed that the number of downregulated elements in hypopigmented biopsies was 11.19% more compared to those in hyperpigmented (2504 vs. 2252, respectively), while upregulated elements were 7.76% less in hypopigmented biopsies (4611 vs. 4969). The log₂ (FC) values were proportionally distributed across all bins where (consistently higher bars for downregulated and lower bars for upregulated elements in hypopigmented biopsies, Figure 4). The cumulative percentages of value frequencies through all bins in the histogram showed a higher curve for hypopigmented elements in the downregulated side of the figure that crosses the hyperpigmented curve in the upregulated figure side (Figure 4). Elements exhibiting extreme differential regulation were found enriched in hypopigmented biopsies (Figure 3, inset table).

Figure 4.

Quantitative Assessment of Gene Regulation Distribution in Hyper- and Hypopigmented Scars

A total of 840 differentially transcribed elements were identified in hyper- and hypopigmented (FC >1.3, p < 0.0089), about 20.95% of which were exclusive to hypopigmented biopsies, 14.76% were exclusive to hyperpigmented, and 64.28% were common to both tissue phenotypes (Figure 5A). The top ten genes exclusively transcribed in hypo- or hyperpigmented scar tissues are listed in Table 1.

Figure 5.

Distribution of Significantly (p-value < 0.0089) Modulated Genes in Hyper- and Hypopigmented Scar Tissues. Expression log₂ ratio >1.3 (A) or fold change >1.3 or < −1.3 (B) in at least one of the pigment phenotype tissue.

Table 1.

Top Differentially Transcribed Genes Exclusive in Hyper- or Hypopigmented Scar Biopsies.

Gene Name	Symbol	p value	Fold Change
Hyperpigmented tissue
tenomodulin	TNMD	4.99E-03	−6.12
chemokine (C-C motif) ligand 28	CCL28	3.85E-04	−5.54
sulfotransferase family 1E, estrogen-preferring, member 1	SULT1E1	1.24E-03	−5.05
serpin peptidase inhibitor, clade B (ovalbumin), member 7	SERPINB7	7.56E-04	−5.04
selectin E	SELE	4.87E-03	−5.04
SIX homeobox 4	SIX4	1.99E-04	6.63
mesoderm specific transcript homolog (mouse)	MEST	5.42E-03	6.64
C1q and tumor necrosis factor related protein 3	C1QTNF3	5.99E-03	7.28
heat shock 27kDa protein 3	HSPB3	3.48E-03	7.29
ATPase, class I, type 8B, member 4	ATP8B4	2.96E-03	7.91
Hypopigmented tissue
ADAM-like, decysin 1	ADAMDEC1	5.03E-04	7.27
phosphatase and actin regulator 3	PHACTR3	5.25E-03	5.85
endoplasmic reticulum protein 27	ERP27	6.72E-03	5.59
DNA-damage regulated autophagy modulator 1	DRAM1	5.26E-04	5.15
thymocyte selection-associated high mobility group box	TOX	7.76E-03	4.77
parathyroid hormone-like hormone	PTHLH	8.69E-05	−7.46
cell death-inducing DFFA-like effector c	CIDEC	6.39E-03	−7.54
myocilin, trabecular meshwork inducible glucocorticoid response	MYOC	8.16E-05	−8.20
lymphocyte antigen 6 complex, locus G6F	LY6G6F	3.73E-09	−11.35
tyrosinase-related protein 1	TYRP1	2.30E-13	−76.18

Identification of Pathways Involved in Pigment Perturbation

Significantly modulated genes were identified using the basal transcription levels in uninjured normally pigmented skin as a reference; therefore, these lists included genes involved in wound healing and scar development. Two approaches were adopted to focus analysis on identifying pathways more relevant to differences in skin pigmentation. The first was to analyze complete data sets from hypo- and hyperpigmented scar tissues, then compare the resultant pathways using specific criteria to accept a defined pathway being related to dyspigmentation. The basis for this approach is that the present variations in wound creation and animal individual responses are minimal because of identical wound creation, treatment, animal diet, animal housing, and controlled environmental factors. The synchronous healing and scar progression support the use of this approach. The second approach was based on creating subset lists for genes that were exclusively differentially transcribed in hypo- or hyperpigmented scar tissue, then performing separate pathway analysis (Figure 5). Cross examining findings from both approaches enhanced the robustness of interpretations.

Pathway analysis of complete data set of differentially transcribed genes

Pathways identified from analysis of complete lists of genes (FC > 1.3 and p-values < 0.0089) for hypo- and hyperpigmented scar samples were produced using IPA. Complete lists of pathways using p-values and z-scores are provided in supplementary Tables 1 and 2, respectively. Comparison of the identified pathways based on p-values showed that only two pathways were unique to hyperpigmented samples. Those were the embryonic stem cell differentiation into cardiac lineages and the role of OCT4 in mammalian embryonic stem cell pluripotency pathways (Table 2). The transcriptional regulatory network in embryonic stem cells, which is functionally related to the prior two exclusive pathways to hyperpigmented tissue, the nicotine degradation II, the phenylalanine degradation I (aerobic), the pyruvate fermentation to Lactate, the caveolar-mediated endocytosis signaling, and the VDR/RXR activation pathways were identified at significantly greater log p-values in hyperpigmented scars (Table 2). No pathways were unique to hypopigmented scars; however, a larger number of pathways showed significantly greater -log (p-values) relative to that in hyperpigmented scars (Table 2). The biggest differences were seen in the mitochondrial dysfunction, arginine degradation I (arginase pathway), and regulation of the epithelial-mesenchymal transition pathways (-log p-value difference > 1.5). Pathways related to redox status, metabolic modulations, and inflammatory responses were more present in hypopigmented samples, while pathways related to stem cell development and differentiation were found mainly in hyperpigmented ones.

Table 2.

Comparison of Pathways Identified from Ingenuity Enrichment Analysis. Pathways were identified using complete data sets of significantly modulated genes in hyper- and hypopigmented scars. (pathways are listed in decreasing order of the hyper/hypo p-value ratio for hyperpigmented and increasing ratio of hypo/hyper in hypopigmented scar. Hyperpigmented scar pathways on top).

Canonical Pathway*	Hyperpigmented(-log p-value)	Hypopigmented(-log p-value)
Embryonic Stem Cell Differentiation into Cardiac Lineages	1.50	Not present
Role of Oct4 in Mammalian Embryonic Stem Cell Pluripotency	1.37	Not present
Transcriptional Regulatory Network in Embryonic Stem Cells	1.57	0.45
Nicotine Degradation II	2.31	0.72
Phenylalanine Degradation I (Aerobic)	1.42	0.53
Pyruvate Fermentation to Lactate	1.50	0.57
Caveolar-mediated Endocytosis Signaling	1.74	0.72
VDR/RXR Activation	2.12	1.01
Citrulline Biosynthesis	1.43	2.97
Glucocorticoid Receptor Signaling	0.86	1.99
Sertoli Cell-Sertoli Cell Junction Signaling	0.78	1.87
Arginine Degradation VI (Arginase 2 Pathway)	1.12	2.91
Urea Cycle	0.95	2.52
HMGB1 Signaling	0.74	2.07
Nitric Oxide Signaling in the Cardiovascular System	0.53	1.78
Regulation of the Epithelial-Mesenchymal Transition Pathway	0.52	2.21
Arginine Degradation I (Arginase Pathway)	0.50	2.17
Mitochondrial Dysfunction	0.34	2.22

^*Pathways in bold are unique to hyper-pigmented scars, those in regular font are significantly (> 2-fold P values and difference of one log or more) higher in hyper-pigmented, and those in italic are significantly greater in hypo-pigmented scar tissue

Examination of pathways showing opposite activity status based on z-score values (negative or positive score values predict an inactivated or activated pathway, respectively) indicated that signaling through actin cytoskeleton, ILK, integrin, paxillin, and VEGF is active in hyperpigmented and inactive in hypopigmented scar tissue (Figure 6), while HMGB1 signaling was the only pathway to be active in hypopigmented and inactive in hyperpigmented scars (Figure 6). Fibroblast growth factor receptor 2, 4 (FGFR2, FGFR4), protein tyrosine phosphatase none-receptor type 11 (PTPN11), and klotho (KL) genes were common to all six pathways. Twelve genes were common to two or more of the five pathways active in hyperpigmented tissue.

Figure 6.

Pathways Showing Opposite Activation Predictions Based on z-scores in Hyper- and Hypopigmented Scars.

Pathway analysis of differentially modulated genes exclusive to hyper- OR hypopigmented scar tissue

The second approach to identifying pathways related directly to pigmentation phenotypes used data subsets of the significantly modulated genes (expression fold change > 1.3 or < −1.3 and p < 0.01) found exclusively in hyper- or hypopigmented scar tissue relative to uninjured normally pigmented skin (supplementary Table 3). A list of pathways unique to hypo- or hyperpigmented scar tissues (log p-value > 1.3) is reported in Tables 3 and 4. The list of hypopigmented scar pathways showed an extensive presence of metabolic pathways where 18 out of the 35 pathways from this analysis involve metabolic functions. Most of these pathways are involved in the degradation of amino acids such as arginine, alanine, cysteine, methionine, and most importantly the essential building block in melanin synthesis, phenylalanine (Table 3). Other pathways impacting melanin synthesis via hormonal signaling or regulation of cAMP/PKA signaling cascade were represented by the α-adrenergic signaling and glucocorticoid receptor signaling, and the dopamine-DARPP32 feedback in cAMP signaling pathways. Pathways suggestive of changes in the redox homeostasis in hypopigmented tissue such as the oxidative phosphorylation and fatty acid β-oxidation I pathway (Table 3) were identified. Pathways involved in innate immune response and inflammation including the TGF-β signaling, Th2 pathway, and rheumatoid arthritis pathogenesis were also present. Fewer pathways exclusive to hyperpigmented scar tissue showed significant modulation and implicated mainly functions of cell development and differentiation, glycosaminoglycan (GAGs) biosynthesis (chondroitin and dermatan sulfate), nicotine and phenylalanine degradation, and pigment synthesis represented by the eumelanin biosynthesis pathways (Table 4).

Table 3.

Exclusive Pathways to Hypopigmented Scars (-log p-value >1.3).

Canonical Pathway*	Hypopigmented(-log p-value)
Arginine Degradation I (Arginase Pathway)	3.39
Arginine Degradation VI (Arginase 2 Pathway)	3.00
Urea Cycle	3.00
Citrulline Biosynthesis	2.73
Oxidative Phosphorylation	2.67
Regulation of the Epithelial-Mesenchymal Transition Pathway	2.30
Phenylalanine Degradation IV (Mammalian, via Side Chain)	2.24
Superpathway of Citrulline Metabolism	2.24
α-Adrenergic Signaling	2.23
Gap Junction Signaling	1.86
Nitric Oxide Signaling in the Cardiovascular System	1.84
Role of Macrophages, Fibroblasts and Endothelial Cells in Rheumatoid Arthritis	1.83
β-alanine Degradation I	1.78
L-cysteine Degradation III	1.78
Putrescine Biosynthesis III	1.78
Glutamate Degradation II	1.61
Aspartate Biosynthesis	1.61
HMGB1 Signaling	1.60
Superpathway of Methionine Degradation	1.54
Fatty Acid β-oxidation I	1.54
Glucocorticoid Receptor Signaling	1.49
Methylmalonyl Pathway	1.49
L-cysteine Degradation I	1.49
Melatonin Degradation II	1.49
Glycerol-3-phosphate Shuttle	1.49
TGF-β Signaling	1.45
Th2 Pathway	1.44
Creatine-phosphate Biosynthesis	1.39
2-oxobutanoate Degradation I	1.39
Factors Promoting Cardiogenesis in Vertebrates	1.39
Role of Osteoblasts, Osteoclasts and Chondrocytes in Rheumatoid Arthritis	1.35
Dopamine-DARPP32 Feedback in cAMP Signaling	1.34
TR/RXR Activation	1.32
Glycogen Biosynthesis II (from UDP-D-Glucose)	1.31
Pyrimidine Ribonucleotides Interconversion	1.31

Table 4.

Exclusive Pathways to Hyperpigmented Scar Tissue (-log p-value >1.3).

Canonical Pathway	Hypopigmented (-log p value)
Embryonic Stem Cell Differentiation into Cardiac Lineages	2.84
Nicotine Degradation II	2.23
Transcriptional Regulatory Network in Embryonic Stem Cells	1.65
Thyroid Hormone Metabolism II (via Conjugation and/or Degradation)	1.65
Phenylalanine Degradation I (Aerobic)	1.64
Eumelanin Biosynthesis	1.54
Chondroitin Sulfate Biosynthesis (Late Stages)	1.47
Pyruvate Fermentation to Lactate	1.47
Thioredoxin Pathway	1.47
Chondroitin and Dermatan Biosynthesis	1.47
Amyloid Processing	1.45
Nicotine Degradation III	1.41
Chondroitin Sulfate Biosynthesis	1.35
Dermatan Sulfate Biosynthesis	1.32

Although the subsets of genes used in this analysis were mutually exclusive by default, they identified 30 pathways common to both pigmentation phenotypes and maintained significant p-values < 0.05 and a ratio of more than 2-fold in one tissue phenotype relative to the other. The top common pathways (p < 0.01 and 2x ratio) are reported in Table 5. The biggest difference was in the mitochondrial dysfunction pathway, along with the NRF2-mediated oxidative stress response pathways (Table 5). These pathways support the earlier suggestion, from the analysis of complete data sets, that abnormal redox status is exclusive to hypopigmented scar tissue. The Wnt/β-catenin signaling pathway shows more than 5 folds the p-values of hyperpigmented scar. This pathway affects CREB and pigment synthesis via the DAG complex/PKC activation.

Table 5.

Pathways Showing Large p-values Differences Between Hyper- and Hypopigmented Scars. Pathways were identified from canonical pathways enrichment analysis (IPA) of mutually exclusive lists of differentially modulated genes in the two pigmentation phenotypes.

Canonical Pathway*	− log(p) Hypo	− log(p) Hyper	Hyper/Hypo	fold difference
Mitochondrial Dysfunction	4.03	0.20	0.05	20.17
Wnt/β-catenin Signaling	3.27	0.59	0.18	5.54
Agranulocyte Adhesion and Diapedesis	2.99	1.01	0.34	2.95
Granulocyte Adhesion and Diapedesis	2.44	0.56	0.23	4.33
Sertoli Cell-Sertoli Cell Junction Signaling	2.43	0.56	0.23	4.34
Hepatic Fibrosis / Hepatic Stellate Cell Activation	2.37	0.54	0.23	4.36
NRF2-mediated Oxidative Stress Response	2.26	0.99	0.44	2.28
Role of Oct4 in Mammalian Embryonic Stem Cell Pluripotency	0.50	3.84	7.72	7.72
Glioma Invasiveness Signaling	0.35	2.11	5.94	5.94
Caveolar-mediated Endocytosis Signaling	0.35	2.09	5.97	5.97

^*Pathways met -log (P) > 2 in either of the scar pigmentation phenotype and 2-fold ratio in one relative to the other.

Analysis of z-score values used a larger subset of exclusive genes (fold change > 1.3 and p-value < 0.01). This resulted in 335 and 396 genes exclusively significantly differentially transcribed in hyper- or hypopigmented scars (Figure 5B). Analysis showed only three pathways exhibiting staggered score values (Figure 7). The biggest differences in z-scores were in the HMGB1 signaling and acute phase response pathways, which were predicted to be activated in hypopigmented tissue and inactivated in hyperpigmented, contrary to the trend in the melanocyte development and pigmentation signaling pathway, which was activated in hyperpigmented and inactivated in hypopigmented scar tissue in agreement with the tissue phenotypes. Interestingly, the three pathways shared three common genes in hyperpigmented scars, and no gene was common with the melanocyte development and pigmentation signaling pathway genes in hypopigmented scars. This observation supports different mechanisms for dyspigmentation in hyper- and hypopigmented scars. The cardiac hypertrophy signaling, IL-8 signaling, production of nitric oxide and reactive oxygen species in macrophages, and the Wnt/Ca+ pathways showed positive z-scores >1 in hypo- but not hyperpigmented scars. Two pathways met the same criteria in hyperpigmented scars; the type II diabetes mellitus signaling and the phosphatase and tensin homolog (PTEN) signaling pathways.

Figure 7.

Pathways Showing Staggered z-scores from the Analysis of Exclusively Differentially Regulated Genes Lists in Hyper- or Hypopigmented Scars.

4. Discussion

Melanin plays an important role in removing reactive oxygen species and protecting cells from oxidative damage and radiation. Adequate melanin synthesis and deposition levels in response to harmful environmental stressors are indicators of healthy skin, and failure in one or more of the multiple processes or enzymes of melanin synthesis and homogeneous deposition is often associated with serious co-morbidities. Among other symptoms, pigmentation synthesis abnormality is a hallmark of several diseases and often is the mildest. Several diseases showing pigment synthesis abnormality as a symptom among many others were investigated (Passeron, Mantoux, & Ortonne, 2005; Yamaguchi & Hearing, 2014). The accumulated results from these studies characterized multiple genes and cellular development and differentiation stages that are helpful in understanding dyspigmentation in the context of wounds and burns. The density of melanocytes was shown to be homogenous in dyspigmented scars, suggesting that the causes for dyspigmentation are either the melanocytes’ stimulation factors or differences in the response of the melanocytes in hypo- and hyperpigmented scar to stimulants. Signs of dyspigmentation observed as early as the third week post-injury include an increased melanin synthesis at the wound periphery. The blue-gray color surrounding the wound at this stage is reminiscent of the color described in Mongolian spots (or dermal melanocytosis), which is typical of trapped melanin-carrying cells in dermis layers during migration and development of nascent tissue. Upon maturation and arrival of cells to the right position in tissue the color turns to normal skin tone; however, in the case of wounds, extra melanin quantities are produced on wound edges and darker-colored, new skin is formed prior to complete wound closure. The two conditions of melanocytes capable of melanin production and the presence of persistent stimulants seem to be met at that part of the wound. Melanin synthesis as judged by skin color gradually declines later after complete reepithelialization and complete resolution of the wound inflammatory phase at wound periphery. The chronology of skin color and melanin synthesis intensity changes in the hyperpigmented areas is similar to that described in post-inflammatory hyperpigmentation. The melanin levels in hypopigmented tissue were much less than the basal levels produced in non-stimulatory conditions of normal skin. This indicates that melanocytes of hypopigmented tissue lacked the function of melanin production or response to stimulants and excludes the presence of stimulants as a possible reason for dyspigmentation. This was supported further by the fact that dyspigmentation was detected during the early wound healing phase and before scar formation with all associated changes in ECM composition, which is the milieu for stimulants exchange. Differences in cellular activities and gene transcription between hyper- and hypopigmented tissues were captured by PCA and heat map analysis, which distinguished samples based on pigmentation phenotypes. General assessment of gene modulation showed more differentially transcribed genes in hypopigmented scar relative to hyperpigmented, and that frequency of upregulation was higher in hyperpigmented tissue, while downregulation in hypopigmented scars was more notable in number and intensity. Pathway enrichment analysis identified the embryonic stem cell differentiation into cardiac lineages and the role of OCT4 in mammalian embryonic stem cell pluripotency to be exclusively involved in hyperpigmented scars. The closely related transcription regulatory network in the embryonic stem cell pathway was also identified in hyperpigmented at much lower p-values relative to hypopigmented scars. The SRY (sex determining region Y)-box 2 (SOX2) gene was common to all three pathways, and was downregulated in hypopigmented scar. The product of this gene is a transcription factor involved in the regulation of embryonic development and determination of cell fate. The mesoderm posterior bHLH transcription factor 1 (MESP1), which plays an essential role in epithelial mesenchymal transition (EMT), and the myocyte enhancer factor 2A (MEF2A), which is a transcription factor that activates stress-induced genes, were upregulated in the first and second pathways exclusively identified to hyperpigmented tissue, respectively. Although no z-score was assigned for both pathways, the regulation of these transcription factors and genes is suggestive of more active self-renewal promoted by OCT4 and EMT with a suppressed differentiation of mesenchymal stem cells in hyperpigmented scars, which is in concord with the hypothesis of melanocytes and other cells of the hyperpigmented tissue being produced mainly from the proliferation of cells of the wound edges. Most of the pathways that meet one log difference and at least a two-fold p-value in hypopigmented relative to hyperpigmented were related to redox, immune response and inflammation, and catabolism functions. Analysis of z-score predicted five pathways to be active in hyperpigmented and inactive in hypopigmented tissues. Those were the actin cytoskeleton signaling, ILK signaling, integrin signaling, paxillin signaling, and VEGF signaling pathways. All five pathways are critical to tight junction development and inflammatory response dampening. Interestingly, the pro-inflammatory high group box 1 (HMGB1) pathway was the only predicted pathway to be active in hypopigmented and inactive in hyperpigmented tissue. All six pathways shared four significantly (p < 0.0089, FC > 1.3) differentially regulated genes, namely, fibroblast growth factor receptor 2 and 4 (FGFR2, FGFR4), klotho (KL), and protein tyrosine phosphatase (PTN11). The alpha actinin 1 and 2 (ACTA1, ACTA2) genes were common to the five pathways active in hyperpigmented tissue, and several genes were common in two or more of the same five pathways, including myosin heavy chain 1, 2, 4, and 7 (MYH1, 2, 4, 7), myosin light chain 1 (MYL1), filaments-actin binding protein (FLNA), gelsolin(GSN), and (HIF1A). These results collectively indicate a stronger inflammatory and high oxidative environment with higher presence of reactive oxygen species (ROS) in hypopigmented tissue. Results from the analysis of unique genes to each tissue phenotype reiterated the same findings. Pathways related to oxidative stress, metabolism and amino acids degradation, cell extravasation and differentiation, and maintenance of tissue energy homeostasis were enriched in the list of pathways unique to hypopigmented tissue. Pathways of glycosaminoglycans synthesis such as chondroitin sulfate and dermatan sulfate biosynthesis were uniquely identified to hyperpigmented tissue. Production of GAGs has an anti-inflammatory effect contrasting the pro-inflammatory environment invoked in hypopigmented tissue. Embryonic stem cell (ESC) differentiation and transcriptional regulation networks of ESC pathways were uniquely identified in hyperpigmented tissue. The gene regulation in these pathways favors suppression of ESC development or differentiation. The eumelanin biosynthesis pathway was identified uniquely in hyperpigmented tissue. Evaluating pathways based on z-scores predicted staggered activation status in three pathways. Consistent with a prolonged inflammatory phase in hypopigmented tissue, the HMGB1 signaling and acute phase response signaling pathways were active in hypopigmented and inactive in hyperpigmented tissues, and more importantly, melanocyte development and pigmentation signaling were inactive in hypopigmented and active in hyperpigmented tissues. Although this report capitalize on the strength of microarrays in exploring genome wide mechanisms and biological processes potentially contributing to dyspigmentation, post-transcriptional processes ultimately yielding functional proteins need to be considered in making final conclusions from such reports. To acknowledge this intrinsic variable to all transcriptomic studies, protein based assays were initiated to assess protein-transcript variance. For example, tyrosinase-like protein 1 (TYRP-1), a key enzyme in melanin synthesis cascade showing significantly dysregulation in microarrays data, was interrogated using immunohistochemistry. Results from this protein based assay were concordant with findings from microarrays (Figure 8), and TYRP-1 was expressed in hypopigmented tissues at much lower level relative to that observed in hyperpigmented tissues. Further investigations are underway to assess other proteins identified in this work as contributors to dyspigmentation.

Figure 8.

Regions of Hyperpigmentation Express Higher Levels of TYRP1 Compared to Hypopigmented Regions of Scar. Hyperpigmented (A) and hypopigmented (B) epidermal sheets from Day 98 were stained with S100β (green), TYRP1 (red), and DAPI (blue). White arrows (TYRP1 positive

5. Conclusion

The types of pathways dominating biological activities in hyper- and hypopigmented tissue invoke a model in which healing in large wounds involves two cell pools; the first originates by self-renewal and expansion from wound edges and vestiges of salvaged basal layer after injury, and the second from pluripotent cells recruited from systemic sources. The first pool maintains the function of responding to a short inflammatory phase during early wound healing by synthesizing and depositing melanin in excess quantities to regain homeostatic skin environment, giving rise to hyperpigmented tissue. The second pool contains cells seemingly differentiating under the suboptimal wound conditions of chronic or extended inflammation, oxidative stressors, and exogenous bacterial or viral presence that ultimately result in cells, particularly melanocytes, lacking pigment synthesis capability which exacerbates stress from ROS and immune responses. It is important to note that this findings reported here are based mainly on transcriptomics, and changes at levels of proteins functions are frequently introduce at translational and post-translational levels. Further cell biology-based and protein-based approaches are in progress to validate this hypothesis.

cells).

Supplementary Material

Supplementary table1 1.

(-log p-values)

Supplementary table1 2.

(z-scores)

Supp TableS3

(-log p-values)

Significance:

Scar dyspigmentation after wound and burn injuries frequently leads to psychosocial distress affecting patients’ reintegration and recovery, increase risk of skin cancer, and accelerate photoaging. This report is the first to explore biological processes involved in scar dyspigmentation using a systems biology approach. Results indicated that cells in the peripheral hyperpigmented scar region exhibit different redox balance, interaction with surrounding cells and extracellular matrix, and inflammation status relative to those in the central predominantly hypopigmented area of the scar. The work identified several genes involved in dyspigmentation, and highlighted the role of the varied wound microenvironments in the functions of newly differentiated or propagated cells in the scar.