Key Cell Functions are Modulated by Compression in an Animal Model of Hypertrophic Scar

Introduction. The value of compression studies and applications in hypertrophic scar (HTS) treatment is often undermined due to the lack of ideal controls, patient compliance, and clear action mechanisms. Objective. This study assesses the genome-wide compression effects on scars under well-controlled conditions. Materials and Methods. An automated pressure delivery system (APDS) applied controlled doses of pressure to scars in a red Duroc swine HTS model. Full-thickness wounds were created by a skin grafting instrument on each animal’s bilateral flanks and were observed through reepithelialization and scar development. On day 70, the APDSs were mounted on the developed scars; right flank scars received a pressure of 30 mm Hg, while left flank scars received APDSs with no pressure (sham) for 2 weeks. A genome-wide assessment of compression effect on transcription in scar specimens before (early), shortly after (mid), and long after (late) compression initiation were performed. Results. Analysis of early-phase biopsies showed similar transcriptome profiles, which diverged thereafter in gene numbers and functions between compression- and sham-treated scars in the mid phase. The majority of these changes persisted in the late-phase scar samples. Canonical pathway analysis of differentially regulated genes resulted in an almost identical list of pathways during the early phase prior to compression. In the mid and late phases after compression, many of the identified pathways shifted in significance, and new pathways such as calcium signaling and cholesterol synthesis emerged. Conclusions. Compression modulates transcription and affects multiple biological functions associated with an improved scar appearance.

Related Content

Establishment of an Uncomplicated Radiation-delayed Wound Healing Model Using Irradiation in Pigs

Preliminary Evaluation of a Collagen Matrix with PHMB Reducing MRSA Infections in a Porcine Model

Effects of a Topical Angiotensin-Converting Enzyme Inhibitor and a Selective COX-2 Inhibitor on the Prevention of Hypertrophic Scarring in the Skin of a Rabbit Ear

Hypertrophic scar (HTS) is a fibroproliferative tissue complication occurring after injury and characterized by distorted topography (height), pliability, pigmentation, and sensory augmentation. About 100 million cases of cutaneous HTS are estimated to occur annually in developed countries, many of which progress to produce unaesthetic or debilitating conditions.¹ Symptoms associated with HTS vary from mild itching to serious neuropathic pain that impairs daily activities, including sleep.² Contractures can be particularly debilitating and often require multiple reoperations to improve function.³

Hypertrophic scars are more common in wounds showing exorbitant expression of inflammatory mediators, prolonged and overlapping healing phases, and delayed periods of reepithelialization.⁴ Histological and molecular evaluation of scar tissue shows excessive deposition of extracellular matrix (ECM) molecules, such as collagen,⁵ elastin,⁶ and proteoglycans, in ratios different from those found in normal skin.^7,8 Changes to tissue cellularity and vascularity also are documented in HTS.⁹

The cellular and molecular mechanisms underlying impaired wound healing and subsequent scar formation are not completely elucidated.¹⁰ Selecting the most effective strategy to prevent or treat HTS is largely dependent upon wound type, location, and general patient health. For example, while mechanomodulatory approaches were reported to be effective in protection from scarring after incisional wounds,¹¹ it would be difficult to apply or treat a HTS from a large surface burn injury with mechanomodulatory methods because of different wound closure and healing mechanisms. Current treatment approaches for HTSs resulting from burn injuries range from invasive surgical excision with or without grafting, especially when contractures are involved,^12-14 to intralesional injections with interferon,¹⁵ bleomycin,¹⁶ or topical and intralesional corticosteroid administration.¹⁷ Silicone gel sheeting also has been reported to be beneficial,^18,19 and the recent use of laser therapy has gained increasing interest as a treatment option.^20,21 The most widely used measure in HTS prevention and treatment, both historically and presently, is compression application,^22-24 in spite of a lack of understanding of the mechanisms for scar mitigation and remission by compression. Different studies provide evidence of a role for prostaglandin E2 release,²⁵ apoptosis and cytokine release and regulation,^26,27 collagenase and matrix metalloproteinase expression and regulation of collagen turnover,^28,29 hypoxia and fibroblast survival and function,³⁰ decreased scar hydration and neovascularization due to mast cell stabilization,³⁰ and more recently, the activities and proliferation of keratinocytes and myofibroblasts.³¹ The multiplicity of hypotheses about the mechanism of compression has exacerbated the controversy about the value of compression in HTS treatment.

Studies of compression have been performed ex vivo or in vitro and have delivered unquantifiable or unstable pressure doses due to varying pressure application methodologies or unknown patient compliance.^10,32 Other studies^33-35 focused on the biological function of a specific molecule; hence, they reached different conclusions or provided seemingly contradicting results. Recently, the Firefighters’ Burn and Surgical Research Laboratory (Washington, DC) lab has tackled the inconsistent pressure delivery problem by developing a novel automated pressure delivery system (APDS), capable of delivering an adjustable steady pressure, and validating the system performance in a red Duroc scar model.^9,36 The effect of the APDS on scar appearance and key ECM components also was confirmed.^5,6

In the current report, a genome-wide approach was adopted to evaluate the effect of compression on the transcriptome of scar tissue. Compression effects were evaluated at the gene transcription level, and pathway analysis was used to decipher the biological processes involved in scar remission. The results confirmed previously reported findings^5,6,9,36 of molecular and gross histological modulations and revealed new mechanisms. Findings from this work resolve disagreements on compression effects and show that compression impacts the expression of multiple genes and biological processes.

Ethics

Juvenile, castrated, male Duroc swine 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 and Animal Welfare Assurance through the Public Health Service. All described animal work was reviewed and approved by the MedStar Health Research Institute’s Institutional Animal Care and Use Committee.

Animal model

Animals weighed 41.4 kg to 42.6 kg when wounds were created and 76.3 kg to 76.6 kg at the terminal time point, respectively.

After anesthesia using intramuscular injection of a combination of ketamine and xylazine, the 2 swine received 1 full-thickness wound (10.16 cm x 10.16 cm) on both the right and the left flanks. Wounds were created using a Zimmer Biomet Dermatome (Warsaw, IN) and excised over the lateral thorax to a full-thickness depth of 2.286 mm (3 passes each at 0.762 mm).

Mepilex Ag (Mölnlycke, Gothenburg, Sweden) was used to dress wounds and was changed on a weekly basis as per the manufacturer’s recommendations. At the end of each procedure, buprenorphine and fentanyl were used for analgesia and repeated after the weekly dressing change until the wound closed. Punch biopsies (3 mm) of uninjured skin pre- and post-excision were collected during weekly assessments. Biopsies used for transcriptome evaluation were stored in Allprotect Tissue Reagent (Qiagen, Valencia, CA) for RNA isolation. Excisional wounds were reepithelialized between days 35 and 49 post injury, and scars were formed by day 70.

An APDS was mounted on each wound on day 70 post injury. The left scar in each animal received APDS without pressure application (sham treatment: device present, no pressure) while the right scar received 30 mm Hg of pressure (compression-treated: device present, with pressure) for 2 weeks (days 70 to 84). Scar assessments and biopsy collection continued on a weekly basis until the end of the experiment on day 133 post injury (eFigure 1). Scars were evaluated using the Vancouver Scar Scale (VSS), and compression-treated scars were scored significantly lower, indicating a positive treatment effect based on the 4 metrics of the VSS (height, pliability, pigmentation, and vascularity).^5,9 Punch biopsy samples were collected during the weekly scar evaluation from day 49 (before APDS mounting), days 91 to 99, and days 119 to 133 (after APDS removal) and were grouped as early, mid, and late phase samples, respectively (eTable 1).

The APDS

The APDS system has been previously described.³⁶ Briefly, the APDS consists of a set of linear actuators for pressure delivery and force-sensitive resistors for delivery measurements. Real-time recordings of applied pressure and adjustments for animal activities were collected, and data were analyzed to ensure APDS proper performance.⁹

Messenger RNA purification

The reagent-preserved biopsies were processed to isolate messenger RNA using the RNA fibrous tissue kit based on the manufacturer’s protocol (Qiagen).

Microarray

A total of 28 biopsies were collected from the pigs, and the transcriptome changes in these biopsies were evaluated during the course of wound healing, scar formation, and pressure application. Four biopsies collected from the left and right flanks of the 2 pigs (1/flank) before wound creation (uninjured skin) were used as a baseline reference of transcription. For assessment of wound and scar progression and to evaluate differences based on sagittal planes alone, 8 biopsies were harvested on day 49 (4/animal: 2 left flank and 2 right flank) before pressure application (early phase). The effect of compression was evaluated from 8 biopsies harvested on days 91 to 99 (mid phase) and days 119 to 133 (late phase), respectively (4/animal: 2 compression-treated and 2 shams). Compression was applied between days 70 and 84 for a total of 2 weeks of continuous pressure. Microarray assays were performed using Agilent-026440 Sus scrofa (Pig) Oligo Microarray v2 expression array (Agilent Technologies, Inc, Santa Clara, CA) following the manufacturer’s protocol. Hybridized microarray slides were scanned using an Agilent Technologies Scanner G2505C US09493743 (Agilent Technologies, Inc). Images of scanned microarray slides were feature extracted and normalized using the company’s feature extraction software, version 10.7 or later, in the default setup (Agilent Technologies, Inc).

Data analysis

Feature-extracted data were filtered on flags to exclude probes with missing values in more than 1 sample and quantile normalized using the Limma Package of R (www.r-project.org). Normalized data from all conditions were analyzed and compared for scar and compression effects. Differentially regulated genes in each condition were identified using a moderated t test at P < .05 of the Limma Package of R.

Principal component analysis (PCA) was performed with Qlucore software (New York, NY), using the default parameters.

Ingenuity Pathway Analysis (canonical pathway enrichment analysis [CPEA]; Qiagen) for enrichment of significantly differentially expressed genes in known canonical pathways was performed using the CPEA library of canonical pathways. Multiple data cutoffs were used, including expression fold ratios ([Expr FR] < 1.5 or 1.8) alone or in combination with P < .01 or an adjusted P < .01. Analyses of complete data sets and subsets of genes differentially regulated exclusively in compression or sham conditions were reported using an adjusted P < .01. This cut-off was selected because it yielded no difference in gene transcription between the left and right wounds before compression application (early phase). Data sets contained 5923, 1618, and 2521 microarray elements with differential signals from all wounds at early-, mid-, and late-phase samples compared with baseline, respectively. Not all elements of interest mapped to a known human gene in the CPEA library. Analysis-ready genes numbered 1883, 584, and 882 in early-, mid-, and late-phase samples, respectively.

Wound healing and scar progression mechanisms dominated transcriptional changes

Global PCA of all the array elements showed clustering of samples from uninjured skin separate from the scar samples at all 3 scar stages regardless of compression treatment (eFigure 2). This suggests that changes in genome transcription were dominated by wound healing and scar progression. The 8 early-phase samples also formed a distinct cluster. The separation of early-phase scar biopsies from mid- and late-phase samples and from uninjured skin samples reflects the transcriptional and biological heterogeneity among the uninjured skin and other phases of scar samples. These observations highlight the capture of a compression-mediated effect on the transcriptomes of scar samples and validate the experimental approach and data robustness. A further PCA using a subset of ≈ 9400 elements with an adjusted P < .05 in an analysis of variance (ANOVA) showed the same pattern observed from the global PCA, but it did not resolve the samples based on the later scar phases (ie, mid and late) or compression. Interestingly, the spatial distribution of the samples in PCA suggested a gradual resolution of the transcriptome regulation from early to mid and late phases to resemble the original status in uninjured skin (eFigure 2). A heat map presentation of the top differentially regulated genes (ANOVA adjusted P value < 1^e-4; 522 elements) showed transitioning of most of the upregulated genes back to baseline expression levels in uninjured skin at a pace that was more rapid than downregulated genes (eFigure 3). This is in agreement with documented wound remodeling progression to simulate normal skin biology.³⁷

Compression changes the transcription of multiple genes   

In order to visualize differences between samples due to compression in a PCA, the investigators removed the early phase conditions that provided most of the variance in the initial PCA analysis and repeated the procedure with the sham- and compression-treated samples at the mid or late phase in the presence of control samples. Under these conditions and using elements meeting an adjusted P < .05 in ANOVA, samples clustered based on compression in mid (eFigure 4A) and late (eFigure 4B) phases, providing the first evidence for compression-induced transcriptional changes in multiple genes.

The number of elements showing statistically significant transcription changes relative to baseline from scars on the right and left animal flanks during the early phase (ie, prior to APDS mounting) was 5192 and 4921 array elements, respectively. The difference represented a small ratio (5.2%) of the total number of differentially regulated elements (adjusted P < .01), which demonstrates the high level of synchronicity among all 4 scars from the 2 animals during the early phase prior to compression application (eTable 2). In agreement with the anticipated tapering transcription during advanced healing stages, samples from mid and late phases showed decreases in the number of differentially transcribed elements in both compression- and sham-treated scars (eTable 2).

Typical decreases in the number of genes differentially transcribed during scar maturation are large and may mask the effect of compression. To address this hurdle, the differences in the numbers of modulated genes between left (sham) and right (compression) scars were tallied across phases and then compared to see if they remained proportional to the number of differentially regulated genes at each phase or if they exhibited a compression-mediated trend. Results showed an increase in the differences in gene number from 66 in the early phase to 123 in the mid phase immediately after removal of compression, which amounted to 25.8% of the total modulated genes in compression-treated scars. This represented almost a 6-fold difference compared with the gene numbers calculated in the early phase (4.15%) prior to compression (eTable 3). Some of these changes were temporal, as indicated by the lower number of modulated genes in compression relative to sham in the late phase (11.6% higher in sham/compression). A parallel assessment of the modulated genes (P < .01) in the right flank scars relative to the left flank scars at early, mid, and late phases showed a steady increase in the number of genes, further supporting a direct relationship between compression application and the observed dynamics of transcription differences (eTable 3).

Qualitative analysis of the distribution of modulated genes showed 70.7% of the sum of all array modulated elements during the early phase were common between left and right scars, and only 16.9% and 12.3% were exclusive to the right or left scars, respectively (eFigure 5). The percentage of common genes decreased from 70.7% to 42.6% in mid-phase scar samples immediately after compression. This sharp decrease was offset mainly by an increase in the percentage of genes exclusive to compression (from 16.9% to 37.6%) with only a 7.6% increase in sham biopsies (12.3% to 19.9%) (eFigure 5), thereby underscoring a compression response affecting a large number of genes and a divergent transcription profile between sham- and compression-treated scars.

Transitioning to the late phase, the percentage of common elements maintained similar percentages (41.6% relative to 42.6% in the mid phase), while elements unique to compression specimens decreased from 37.6% in the mid phase to 27.1%, suggesting some of the compression-mediated changes were reversible or transient. In the sham counterpart, the percentage of exclusive genes increased from 19.9% to 31.3% (eTables 2, 3; eFigure 5). Similar trends were observed when P values < .01 were selected as a threshold for data filtering (aforementioned analysis was based on adjusted P < .01) or when lists of significantly modulated genes (Expr FR < 1.3, 1.5, or 1.8) were used. This quantitative assessment of transcriptome dynamics revealed a clear compression-mediated effect on gene expression despite the complex background of scar development biology.

Examination of modulated gene lists before and after compression showed coverage of a similar wide range of functions. The largest skews were seen in G-protein coupled receptors (GPCR), transmembrane receptors, and peptidases. Ion channels showed a transient increase in the mid phase, and cytokines showed delayed decreases in late samples. A comparison of the top modulated gene lists among the designated 3 scar phases showed more variation across phases than between samples of compression- and sham-treated scar tissue (eTable 4), which supports the domination of healing and repair functions.

Canonical pathways enrichment analysis

Pathways analysis of total significantly (adjusted P < .01) modulated genes

Using data sets of genes passing an adjusted P value < .01 cutoff, CPEA showed a decreased number of pathways in mid and late phases after compression application compared with early-phase scars. The percentage of pathways unique to left flank or right flank scars in the early phase prior to compression applications was only 3.2% and 3.39% of the total number of identified pathways, respectively, while 93.41% of pathways were common among all scars (eTable 5). The very small and similar percentages of pathways exclusive to each animal flank relative to mutual pathways in the early phase underscore the similarity of the biological profiles in all scars despite the intrinsic variability in animal response. Although the total number of pathways identified by the differentially expressed genes was reduced in the mid phase, the reduction was not proportional to that seen in the number of genes; more importantly, although the percentage of common pathways was reduced from 93.41% to 82.82% after the application of compression, the number of pathways exclusive to compression-treated scars increased from 18 to 56 (3.1 fold) and was greater than the 17 to 28 change (1.6 fold) seen in sham-treated counterparts (eTable 5). These findings align with those from the differentially regulated genes, showing an immediate effect from compression. The number of common pathways between compression- and sham-treated scars in the late phase showed a steady decrease despite the increased number of genes in the same phase, further supporting a divergence in scar progression. The number of exclusive pathways identified in compression-treated scars in the late phase decreased from 56 (mid) to 49, suggesting not all compression-induced changes were permanent. Compression seems to have a long-term effect on multiple biological activities, as evidenced by the number of compression pathways in the late phase that remained smaller than the 66 exclusive pathways identified in the sham-treated counterparts (eTable 5).

The top pathways identified from an analysis of significantly modulated genes (P < .01) in increasing order of Fischer’s exact test P values (which determine the probability of random association between the genes in the datasets and the canonical pathway) showed the dissimilarities between sham- and compression-treated scars increased from early to mid and late phases, confirming a divergence in scar progression after compression treatment. The top 10 identified pathways for each phase of treatment are shown in eTable 6. Pathway lists show changes in the P values for each pathway due to the change in the number of modulated genes or level of regulation. No z score (a positive or negative value that predicts activation or inactivation of a given pathway) could be calculated for all the pathways at any time point. Similar results were found when analysis was performed using P < .01 as the data threshold.

Most of the top differentially regulated genes in terms of number, fold change ratios, and association with lowest P value were found in the early phase prior to compression. Those genes also were enriched in the mid and late phases after compression. As a result, the pathways identified in the analysis based on whole data sets (log P < .01 or adjusted P < .01) primarily represented healing and scar development biology rather than compression effect. To better identify the pathways affected most by compression in the presence of the dense background of scar and wound healing processes, 2 approaches were adopted. The first was to sort the pathways based on the activity differences between sham- and compression-treated scars (ie, pathway P values) at mid and late stages. This approach uncovered pathways exhibiting decreasing trends (compression-reduced) and pathways exhibiting increasing trends (compression-enhanced) after compression. The second approach was to identify genes that were exclusive to compression or sham in mid and late phases and then perform a pathway analysis.

Compression-mediated increases in the functions of multiple pathways

Results for compression-mediated activation suggested an effect on calcium transport, where calcium transport I and calcium signaling pathways were identified among the top 10 pathway list (eFigure 6A). The main genes involved in the modulation of Ca²⁺ transport were ATPase sarcoplasmic/endoplasmic reticulum Ca²⁺ transporting 2 (ATP2A2), which plays a role in the binding of calmodulin and calcium, and ATPase secretory pathway Ca²⁺ transporting 1 (ATP2C1), a member of the P-type cation transport ATPases family that serves as a magnesium-dependent enzyme that transports calcium ions using energy from adenosine triphosphate (ATP) hydrolysis. Other genes identified were involved in the calcium-signaling pathway, including protein phosphatase 3 catalytic subunit alpha (PPP3CA); calcium/calmodulin dependent protein kinase II delta (CAMK2D), an essential player in numerous other canonical pathways; and nuclear factor of activated T cells (NFAT) 3 (NFATC3), a member of the NFAT DNA-binding transcription complex. A parallel analysis using log Expr FR (FC > 1.5 and adjusted P < .01) showed increases in z scores for the Ca²⁺ signaling pathway in compression-treated scars.

In addition, the results introduced cholesterol biosynthesis as a major interface between compression and scar. The cholesterol biosynthesis I, cholesterol biosynthesis II (via 24, 25-dihydrolanosterol), cholesterol biosynthesis III (via desomosterol), superpathway of cholesterol biosynthesis, and bile acid biosynthesis neutral pathways were among the top pathways identified after compression treatment. The relevant zymosterol biosynthesis pathway was fourteenth in the list. These pathways share several genes, including methylsterol monooxygenase 1 (MSMO1), which encodes a sterol-C4-methyl oxidase-like protein l containing a set of putative metal binding motif similar to the yeast methyl sterol oxidase (ERG25) protein; 24-dehydrocholesterol reductase (DHCR24), which encodes a flavin adenine dinucleotide-dependent oxidoreductase catalyzing the reduction of the delta-24 double bond of sterol intermediates during cholesterol biosynthesis; cytochrome P450 family 51 subfamily A member 1 (CYP51A1), which is a member of the cytochrome P450 superfamily of enzymes serving as a monooxygenase and catalyzes many reactions involved in synthesis of not only cholesterol but steroids and other lipids; the aldo-keto reductase family 1 members C1 (AKR1C1) and C2 (AKR1C2); and the squalene epoxidase (SQLE). All also were upregulated with compression treatment. In addition, all of these genes localize to the endoplasmic reticulum membrane and many to the mitochondria.

The ras homolog gene family (RHO), member A (RHOA), signaling pathway was the seventh pathway to exhibit an upregulated trend of activity consistent with a compression effect. The main genes involved in this pathway included the rhophilin family of RHO-GTPase binding proteins (encoded by the RHPN2 gene), which binds both guanosine triphosphate (GTP)- and guanosine diphosphate (GDP)-bound RHOA and GTP-bound RHOB, and plays a role in the organization of actin cytoskeleton. Other important differentially regulated genes include protein phosphatase 1 regulatory subunit 12A (PPP1R12A), which is involved in myofiber contraction; and radixin (RDX) and ezrin (EZR), which are involved in cytoskeleton organization. Analysis for the z scores of pathways previously described showed positive increases in compression-treated scars relative to sham.

The cyclic adenosine monophosphate (cAMP)-mediated signaling pathway was among the top 10 pathways. The main genes modulated increasingly by compression include cAMP-dependent protein kinase inhibitor gamma (PKIG), A-kinase anchoring protein 9 (AKAP9), CAMK2D, and other genes shared with Ca²⁺ ion signaling and transport, and RHOA signaling pathways. The pleotropic E1A binding protein p300 (EP300) gene showed a regulation consistent with compression effect, especially during the late scar phase.

Compression-mediated decreases in the function of multiple pathways

Pathways meeting a comparable high activity in all wounds during the early phase, high activities in the sham mid and late phases, and low activities in the compression mid and late phases were identified and then sorted in decreasing order of compression-induced differences. This analysis identified the glutathione redox reaction I pathway on top of the list (eFigure 6B), which was mainly influenced by the transcription of the peroxiredoxin 6 (PRDX6) gene, a member of the thiol-specific antioxidant protein family that has a bifunctional activity due to its 2 distinct active sites. This enzyme is involved in redox regulation of the cell; it can reduce peroxides, short chain fatty acid, and phospholipid hydroperoxides. The glutathione S-transferase zeta 1 (GSTZ1) is another compression-modulated gene in the pathway. It is a member of the glutathione S-transferase (GSTs) superfamily of multifunctional enzymes involved in the detoxification of electrophilic molecules, including carcinogens, mutagens, and several therapeutic drugs by conjugation with glutathione. No z score was assigned to the glutathione redox reaction I pathway when data were analyzed based on expression fold change (-log Expr FC > 1.5, adjusted P < .01).

The acute phase response signaling pathway is usually activated in tissue injuries associated with adrenal hormone release, infections, pyrogens, and other factors. Signaling in this pathway leads to increasing levels of proinflammatory cytokines and induces several metabolic changes. This pathway was second on the list to exhibit lowered P values, driven mainly by the expression of conserved helix-loop-helix ubiquitous kinase (CHUK), serine protease inhibitor serpin family G member 1 (SERPING1), which is involved in complement cascade; nucleolar and coiled-body phosphoprotein 1 (NOPP140); pleiotropic proinflammatory interleukin (IL) 1 alpha (IL-1A) and IL-6; glycoprotein transferrin, which transports iron and participates in the removal of organic matter from serum; and cellular retinoic acid binding protein 1 (CRABP1), which encodes a specific binding protein that binds retinoic acid only at specific sites in the nucleus and affects vitamin A-directed differentiation in epithelial tissue.

The dendritic cell maturation pathway showed decreased activities in association with compression application (eFigure 6B). The main genes in the pathway to reflect this decrease were IL-1A and IL-6; signal transducer and activator of transcription 1 (STAT1), which respond to cytokines, interferons, and growth factors by activating the transcription of several genes; EP300; interferon alpha and beta receptor subunit 1 (IFNAR1); fragment crystallizable region of IgG receptor IIIa (FCGR3A) and IIIb (FCGR3B); collagen type I alpha (COL1A) 2 chain (COL1A2); and CHUK.

The androgen signaling pathway activity is negatively influenced by compression, which would inhibit multiple transcription factors and protein syntheses. The main genes showing changes by compression included receptors for activated C kinase 1 (RACK1), general transcription factor IIH subunit 4 (GTF2H4), EP300, RNA polymerase II subunit B (POLR2B), and CAMK4. Androgen signaling interferes with intracellular Ca²⁺ concentrations through a GPCR and regulation of calcium influx via nonvoltage gated Ca²⁺ channels.

Atherosclerosis signaling was introduced based on the regulation of IL-1A2, IL-6, and COL1A1, as found in other pathways. Other genes in this pathway include the chemokine (C-C motif chemokine ligand 2 [CCL2]) implicated in the pathogenesis of diseases involving monocytes infiltration, such as hypertrophy in the case of scar; golgi glycoprotein 1 (GLG1); and phospholipase B1 (PLB1). The same pathway was identified using other data analysis approaches, including the use of P < .01 or log Expr FR > 1.5 (ranked fourteenth in the FC1.5 analysis, and 311/375 in P < .01).

Interleukin 6 signaling showed reduced function in many of its genes upon compression application, including IL-6, which is a stimulator and a product on the pathway activation; cytochrome P450 family 19 subfamily A member 1 (CYP19A1); casein kinase 2 beta (CSNK2B); mitogen-activated protein kinase 10 (MAPK10); IL-1; and COL1A1. The z scores for IL-6 signaling suggest stronger inactivation in scar specimens of the mid phase immediately after compression removal, followed by increases in the late compression samples that remained lower than that in the late sham samples. Gene regulation of IL-6 contributed to the suggested roles of acute phase and inflammation responses pathways.

The appearance of the xenobiotic metabolism signaling pathway was due to the trends of IL-6, IL-1A, and EP300, as described earlier. Other genes showing the same trend included aldehyde dehydrogenase family 9 member A1 (ALDH9A1) enzyme, which catalyzes the dehydrogenation of gamma-aminobutyraldehyde to gamma-aminobutyric acid, and microsomal glutathione S-transferase 1 (MGST1), which is involved in the synthesis of the inflammation mediators leukotrienes and prostaglandin E and cell protection. Other enzymes in the pathway showed opposing expression trends that did not offset the general trend of the pathway; these included catalase, which plays an important role in protection against oxidative stress; protein phosphatase 2 catalytic subunit alpha (PPP2CA); and carboxylesterase 1 (CES1).

The γ-linolenate biosynthesis II pathway plays an important role in cholesterol synthesis regulation and transport as well as maintaining the level of skin hydration. The acyl-CoA synthetase long-chain (ACSL) family member 4 (ACSL4) and, to a lesser degree, ACSL1 were the 2 genes in this pathway to show reduced function by compression. Independent from compression effect, an increasing transcription trend was noted in the activity of the essential pathway enzyme cytochrome b5 type A (CYB5A), which acts downstream of ACSL1 and ACSL4, suggesting a role for this enzyme in γ-linolenate synthesis and scar development.

The lowered death receptor signaling pathway activation involved inactivation of tumor necrosis factor (TNF)-associated factor family member associated nuclear factor kappa B (NFKB) activator (TANK), poly (ADP-ribose) polymerase (PARP) family member 10 (PARP10) and PARP12, lamin A/C (LMNA), TNF superfamily member 10 (TNFSF10), and CHUK. The identification of the farnesoid X receptor (FXR)/retinoid X receptor (RXR) activation pathway (eFigure 6B) among the top 10 pathways showing reduced function associated with compression supports a metabolic change, which is consistent with the other findings herein involving cholesterol synthesis. The main genes contributing to this pathway trend were CYP19A1 and IL-1A.

CPEA of genes significantly transcribed exclusively in sham- or compression-treated scars

To improve the potential of identifying specific pathways responding to compression, significantly modulated genes exclusive to sham-treated or compression-treated scars in the mid and late phases (eFigure 5) relative to uninjured skin (adjusted P < .01) were analyzed independently in CPEA. Although the number of exclusive genes in the mid-phase sham (107), mid-phase compression (230), late-phase sham (296), and late-phase compression (228) used in the analysis is smaller than that in the complete lists (see Note in Acknowledgments), which would statistically underestimate the role of identified pathways (ie, lower Fisher’s exact test P values and ratios), this supplemental analysis reveals the scar pathways (sham) that were directly affected by compression. Results indicate the pathways showing the largest modulations at the mid phase immediately after compression removal involve cell migration and tissue infiltration, ECM deposition, metabolism, growth of epithelial cells, and signaling via the small GTPase proteins that affect a wide range of cell functions (eTable 7). The top late-phase pathways denote changes in calcium signaling and cholesterol and protein synthesis. It is important to note that cholesterol synthesis pathways in the late-phase scar treated with compression appeared among the top mid-phase sham pathways (eTable 7).

Analysis of exclusive genes at all 3 phases with or without compression identified 253 pathways that were affected in under at least 1 condition. Sorting pathways for a pattern of increasing (eFigure 7A) engagement likelihood under compression (ie, pathways showing the largest differences relative to sham) introduced the thyroid receptor (TR)/RXR activation pathway, which plays an important role in growth development and metabolism through regulating the expression of multiple genes. The WNT/Ca²⁺ pathway and the role of NFAT in regulation of the immune response pathway were the second and third pathways to show an increasing expression profile with compression application. Both pathways are dependent on calcium concentrations and involved in the regulation of basic developmental cell processes and immune response, respectively. The Rho GDP dissociation inhibitor (ARHGDI) signaling was the fourth to show significant increase in activity, probably owing to its involvement in multiple and basic cellular functions. The ARHGDI, which inhibits the Rho guanine nucleotide exchange factors (RhoGEF) and the GTPase activating proteins by simulating the GTP-hydrolytic reaction, and the small GTPase protein Rho serve as regulators of membrane trafficking, cytoskeletal reorganization, cell differentiation, cell migration, inflammatory responses, and apoptosis.

A key upregulated gene in compression-treated scars in the late phase is EP300. This pleotropic gene contributes to a myriad of cellular activities related to wound healing and the scar remodeling processes. Tight junction signaling showed significant but mixed regulation in many of its gene members, including several claudins (CLDN; eg, CLDN11, CLDN5, CLDN10, CLDN7, CLDN4, CLDN23) and occludins (OCLN) among other genes implicated in firm adhesions and epithelial cell polarity. A heat map showing the top 10 pathways indicating an upregulation in response to compression is provided in eFigure 7A. On the other hand, a decreasing pattern of several pathways activities associated with compression application included the NFKB signaling; acute phase response signaling; role of macrophage, fibroblasts, and endothelial cells in rheumatoid arthritis (inflammatory response in tissue); dendritic cell maturation; and peroxisome proliferator-activated receptor alpha (PPARA)/RXRA activation (eFigure 7B). Many of these pathways also were identified from the analysis of complete data sets (conclusive lists of significantly modulated gene; common to and exclusive).

Few pathways from analysis of the exclusive gene list of compression-treated scars yielded negative or positive z scores. Analysis of the 230 and 228 genes unique to compression-treated scars (adjusted P < 0.01) in mid and late phases identified 57 and 37 pathways (Fisher’s exact test P value < 1.3), respectively. The larger number of pathways identified in the mid phase, despite the use of an almost identical number of genes in the analysis, highlights the more diverse functions of genes modulated in the mid phase, which tend to decrease in the late phase, suggesting a transient, short-term effect for compression. Further examination of the pathways based on calculated z scores provided further support to this conclusion. Fourteen pathways in the mid phase and only 4 in the late phase yielded a quantifiable Z score (eFigure 8). Three pathways were common to both phases.

Pathway analysis of sham- or compression-exclusive genes simultaneously modulated relative to each other (P < .01) and to baseline (adjusted P < .01)

Data sets of significantly modulated genes (adjusted P < .01) relative to baseline in sham- and compression-treated scars of mid and late phases were crossed with the respective lists of significantly modulated genes in compression relative to sham (P < .01) to identify common genes (eFigure 9). This resulted in 11 and 12 genes unique to compression (relative to baseline, adjusted P < .01) and significantly different from sham (P < .01) at mid- and late-phase scar time points, respectively. The 11 mid-phase genes are involved in metabolic pathways of methionine cycling (biosynthesis and degradation) and related cysteine biosynthesis, retinol biosynthesis, and triacylglycerol degradation (eFigure 10). The latter 2 pathways are essential in lipid metabolism and free radical scavenging. The top 5 pathways identified by the 12 late-phase genes were components of cholesterol biosynthesis (eFigure 11). The common gene to all 5 pathways and contributing to cholesterol synthesis was hydroxysteroid 17-beta dehydrogenase 7 (HSD17B7), which also contributed to the presence of the estrogen biosynthesis and signaling pathways in the list (eFigure 11). Other pathways, such as the chemokine signaling, melatonin signaling, granulocyte-macrophage colony stimulating factor, and the crosstalk between dendritic cells and natural killer cells pathways, were brought about by the upregulated CAMK2D gene. The upregulated nuclear receptor coactivator 4 (NCOA4) gene introduced the role of WNT/GSK-3 beta signaling pathway in the pathogenesis of influenza pathway.

Although compression has been used extensively in scar prevention and treatment for more than 30 years,³⁸ the advantages of its use are still debated. Many explanations have been proposed on how compression modulates scarring, none of which are conclusive, and no standard protocol defining ideal pressure dose or treatment duration, initiation, and discontinuation is available for optimal outcome. Critics point to several caveats in studies advocating the use of compression, including the inability to apply steady and quantifiable pressure, patient compliance, and poor understanding of compression mechanism of action.²¹

The validated delivery of a steady compression dose and the significant modulations to HTS properties presented herein using the APDS in a porcine scar model^5,9,36 warranted the investigation of compression-induced changes at the molecular level. This work did not focus on specific genes or a single mechanism but rather adopted a systems biology approach using microarray technology to close this gap in HTS research. Results were confirmed using reverse transcription polymerase chain reaction for multiple genes,^5,6 and deep data mining was performed using multiple data cutoffs for robust findings.

The results showed changes (adjusted P < .01) in about 5% of the Sus scrofa genome (1526/30 364 genes) in scar tissue during the early phase. As expected, changes lessened to 1.2% and 2.1% during the mid and late phases, respectively. Compression-induced variances were less than those caused by scar progression, as evidenced by the lists of top differentially regulated molecules, the inability to identify any significantly modulated genes in compression relative to sham within each study phase (adjusted P < .01) while a large number of genes were identified in sham or compression in all phases relative to baseline, and results of PCA. Using a data cutoff of P < .01 introduced 363 and 562 elements of modulated genes in compression-treated in mid- and late-phase scars relative to sham in the same phases. One caveat to this cutoff selection is that 251 elements (89 genes, 0.29% of total genes) showed a significant change during the early phase where theoretically no difference should exist (no compression applied). These differences are mainly due to 2 types of variability; the first is introduced by individual responses among animals, and the second is that a more relaxed cutoff is more permissive to noise inherent in the assay. Despite these variations, a compression effect was evident and distinguishable from the dynamics of transcription across study phases quantitatively and qualitatively. For example, the number of microarray elements showing a difference in compression relative to sham (P < .01 cutoff) in mid and late phases increased steadily in contrast to the general decreasing trend in the number of modulated elements across phases resulting from all data analyses regardless of the used cutoff (ie, adjusted P < .01, P < .01, Expr FR > 1, 1.5, or 2) or reference (baseline or sham). This observation guided the selection of the adjusted P < .01, where no difference was detected in compression relative to sham at the early phase, as an ideal cutoff for the most conservative analysis and interpretation. Transcription at this cutoff showed a very small difference among the wounds at the early phase, which was followed by a large increase at the mid phase and a decrease in the late phase. Qualitatively, these changes were associated with a drop in the percentages of genes common to both animal flanks in the mid and late phases relative to the early phase, and an increase in genes specific to the compression-treated scars, especially during the mid phase. These dynamics of scar transcriptomes are solid evidence of the effect of compression at the molecular level, which could be distinguished even prior to deeper analysis.

The number of identified pathways in all phases did not fluctuate as much as the number of genes (adjusted P < .01), indicating the bulk of the changes were to pathways already active in the early phase (precompression). Pathways unique to compression-treated scars, however, increased more than 3 times of that seen in the early phase and more than 2 times the number in sham-treated scars. The decrease in the number of pathways in compression-treated scars in the late phase supports the presence of transient compression-induced pathways. Sham scar tissue showed a different profile in the number of pathways that increased steadily towards the late phase. The difference in the pathway profiles suggests compression might be accelerating in what otherwise would be the delayed processes seen in the sham. Again, different profiles of active pathways after compression support a compression effect involving a wide spectrum of biological activities and not a limited effect on a single gene or function, which has been proposed to date.

Analysis results confirmed almost all proposed mechanisms of compression action including hypoxia,^39,40 neovascularization,⁴¹ oxidative stress,⁴² and inhibition of metalloproteases,²⁸ prostaglandin E2 release,^25,43 and cytokines and apoptosis modulations.^26,27,44 Many of these mechanisms were detected at different levels in shams.

Advanced analysis based on common and unique gene lists identified novel compression-associated pathways involving calcium-based functions, such as calcium signaling and calcium transport, which in turn affect a myriad of important biological processes including cell signaling, molecular transport, and vitamin and mineral metabolism. A potentially modified gene pertaining to calcium signaling and transport, among multiple other functions, is ATP2C1, encoding a magnesium-dependent enzyme that transports calcium ions using ATP. Interestingly, defects in this gene cause Hailey-Hailey disease, an autosomal dominant skin or epidermal disorder. Another important gene in calcium transport is PPP3CA, which is a pleotropic gene involved in wound healing, new tissue regeneration, and hypertrophy.

Cholesterol has been proposed to play an important role in epithelialization and wound healing.⁴⁵ This work showed that multiple pathways critical to cholesterol synthesis are influenced by compression. A key gene, CYP51A1, which exhibits responses to compression, has been a target in new burn therapeutics. The identification of pathways regulating cholesterol biosynthesis, γ-linolenate biosynthesis, calcium transport, cAMP and ATP levels, and oxidative status among top pathways in this work is of special importance, because they create a network implicated in inflammation and hypertrophy and more recently in pain and itch.^46,47 Identification of the acute phase response signaling, the role of NFAT in regulation of immune response, NFKB signaling, and xenobiotic metabolism signaling in top modulated pathways suggest a strong influence of compression on local immune regulation, which seems to ameliorate hypertrophy, interfere with local bacterial presence, and modulate the presence of toxicants in scar. It is important to note that the effect of compression may vary based on the magnitude and duration of the pressure applied. Studying the effects of these 2 factors requires additional work that was beyond the scope of this study. Increasing the number of animals would improve results statistics; however, the cost of such a large animal model and its housekeeping is often a common limitation.

In conclusion, compression is not a mere fluids or circulation extrusion process but involves changes at the transcriptome level. These changes cover numerous genes and networks with diverse functions, some of which are transitional, but many others lasted for several weeks (at least) after compression removal.

This study was based on samples from 2 animals in which 2 scars underwent compression and 2 received APDSs without compression. A larger number of animals will reduce the potential for false positives. Although red Duroc porcine wound model provides best known resemblance of human skin, extrapolation of findings from this study should be validated using human scar samples. Optimal compression intensity and application duration to induce the largest biological changes need to be assessed independently. Samples from such a study would improve the accuracy of the findings.

Scar treatment with compression induced temporary and long-lasting transcriptional changes in a large number of genes with a wide-range of functions. These changes were associated with improved scar outcome in a red Duroc porcine model of HTS.

Note: Supplemental tables regarding genomic pathways are available upon request; contact the corresponding author for these.

Contributions: Scar preparation and biopsy collection were performed by Dr. Travis and Ms. Carney at the Firefighters’ Burn and Surgical Research Laboratory, MedStar Health Research Institute (Washington, DC), under the supervision of Dr. Shupp and Dr. Moffatt. Microarray and statistical analyses were performed by Dr. Muhie and Ms. Miller at the Integrative Systems Biology, US Army Center for Environmental Health Research (Frederick, MD), under the supervision of Dr. Hammamieh and Dr. Jett. Dr. Alkhalil performed the data analysis and wrote the manuscript.

Authors: Abdulnaser Alkhalil, PhD¹; Bonnie C. Carney, BS¹; Taryn E. Travis, MD^1,2; Seid Muhie, PhD³;Stacy-Ann Miller, BS³; Jessica C. Ramella-Roman, PhD⁴; Pejhman Ghassemi, PhD⁵; Rasha Hammamieh, PhD³; Marti Jett, PhD³; Lauren T. Moffatt, PhD¹; and Jeffrey W. Shupp, MD^1,2

Affiliations: ¹Firefighters’ Burn and Surgical Research Laboratory, MedStar Health Research Institute, Washington, DC; ²The Burn Center, Department of Surgery, MedStar Washington Hospital Center, Washington, DC; ³Integrative Systems Biology, The Geneva Foundation, US Army Center for Environmental Health Research, Fort Detrick, MD; ⁴Department of Biomedical Engineering, Florida International University, Miami, FL; and ⁵Center for Devices and Radiological Health, Office of Device Evaluation, U.S. Food and Drug Administration, Silver Spring, MD

Correspondence: Abdulnaser Alkhalil, PhD, Firefighters’ Burn and Surgical Research Laboratory, MedStar Health Research Institute, 110 Irving Street, NW, Washington DC 20010;
abdulnaser.alkhalil@medstar.net

Disclosure: This work was funded in part by the NIH grant No. 1R15EB013439 under the names of Dr. Shupp and Dr. Ramella-Roman. The views, opinions, and/or findings contained in this report are those of the authors and should not be construed as official Department of the Army position, policy, or decision, unless so designated by other official documentation. Citations of commercial organizations or trade names in this report do not constitute an official Department of the Army endorsement or approval of the products or services of these organizations. This research complied with the Animal Welfare Act and implementing Animal Welfare Regulations, the Public Health Service Policy on Humane Care and Use of Laboratory Animals, and adhered to the principles noted in The Guide for the Care and Use of Laboratory Animals (NRC, 2011). All described animal work was reviewed and approved by the MedStar Health Re-search Institute’s Institutional Animal Care and Use Committee.