Identification of novel keloid biomarkers through Profiling of Tissue Biopsies versus Cell Cultures in Keloid Margin specimens Compared to adjacent Normal Skin

Patients and samples

Samples from 4 patients were used in this study. The mean age was 29 ± 4 years. Three patients were white, and the fourth patient was of white/black Caribbean ancestry (Table 1). Biopsies of normal skin and keloids margin were obtained (Fig 1).

Figure 1. Illustration of the lesional sites of keloids taken in this study.

Tissue culture

Primary tissue cultures were obtained by enzymatic digestion of biopsies. The collected samples were minced into small pieces with sterile scalpels and incubated in 0.25% to 5% collagenase A solution (Roche Diagnostics GmbH, Mannheim, Germany) at 37°C for 2.5 to 3 hours. The collagenase digestion was inhibited using fibroblast culturing media. The fibroblast culturing media consists of Dulbecco's modified Eagle's medium 3 (Lonza, Verviers, Belgium), supplemented with 10% heat-inactivated fetal bovine serum (Sigma, Gillingham, UK), 1% penicillin/streptomycin (Lonza), and 1% non–essential amino acids (Lonza).

After the digestion, each sample was spun at 1,200 rpm for 5 minutes, re-suspended in fibroblast culturing media, and seeded in 25-cm² culturing flask (Corning, Corning, NY). The cultures were maintained at 37°C and 5% CO₂, and the media was replaced every 48 hours. Passaging was carried out with trypsin-ethylene diamine tetraacetic acid (200 mg/L of ethylene diamine tetraacetic acid, 500 mg/L of trypsin; Lonza) when approximately 80% confluence was reached.

RNA extraction

Four pieces of 2-mm³ tissue were cut off from each biopsy sample that had been stored in RNAlater. Each piece of 2-mm³ tissue was finely diced and placed in 2-mL round-bottomed Eppendorf tubes with a flame-sterilized steel ball bearing and 1 mL of Trizol (Invitrogen, Carlsbad, Calif). The tissues in the Epppendorf tubes were homogenized by Qiagen tissue lyser (Qiagen, Hilden, Germany) that was set at 30 beats per second for 12 minutes. The homogenized tissue suspension was transferred to a sterile 1.5-mL Eppendorf tube and centrifuged at 13,000 rpm for 10 minutes to remove cell debris. After the centrifugation, the resulting supernatant were transferred to a new Eppendorf tube and mixed with 0.2 mL of chloroform/1 mL of Trizol (Invitrogen). Solutions in each tube were mixed well and left at room temperature for 2 minutes, after which the mixtures were centrifuged at 13,000 rpm for 15 minutes. The upper aqueous layer in each tube was transferred into a fresh Eppendorf tube and equal volume of 70% ethanol was added and mixed well.

The extracted RNA was further processed using RNeasy kit (Qiagen, UK) according the manufacturer's instructions, followed by DNase treatment with a DNAFree kit (Ambion, Austin, Tex) according to the manufacturer's protocol.

Complementary DNA synthesis

SuperScript II reverse transcriptase kit (Invitrogen) was used for synthesis of complementary DNA (cDNA). For each sample, 1000 ng of RNA, 1 µL of nucleotides mix (10 mM for each nucleotide) (Invitrogen), 375 ng of oligo-dT, 62.5 ng of random primers, and sterile nuclease-free water (Ambion) were mixed in a nuclease-free Eppendorf tube to make up a total volume of 12 µL. After incubation at 65°C for 5 minutes and rapid cooling on ice, 2 µL of 0.1 M of DTT (Invitrogen), 1 µL of RNaseOut (Invitrogen), and 4 µL of first-strand buffer (250 mM of Tris-hydrochloride, pH 8.3 at room temperature; 375 mM of potassium chloride; 15 mM of magnesium chloride; Invitrogen) were added to each tube. After incubation at 25°C for 2 minutes, 1 µL of SuperScript II reverse transcriptase (Invitrogen) was added and incubated at 25°C for a further 10 minutes, before being transferred to 42°C incubation for 50 minutes. Following this, the samples are incubated at 70°C for 10 minutes to inactivate the enzymes.

Selection for candidate genes

Bioinformatics methods were used to select candidate genes. Through literature searches, 7 non–pathway-specific microarray gene profiling studies were identified, including studies of Smith et al,¹² Seifert et al,⁷ Hu et al,¹³ Naitoh et al,¹⁴ Satish et al,¹⁵ and Chen et al,^16,17 and a list of genes from own unpublished data (Table 2). Full list of upregulated or downregulated genes were collected, and all given gene details were converted into the National Center for Biotechnology Information (NCBI) official gene symbol by DAVID Bioinformatics Resource 2008. Entries unrecognized by DAVID Bioinformatics Resource 2008 were converted manually by searching the NCBI database. Whether the reported dysregulation was upregulation or downregulation was also noted for each gene. Candidate genes were selected by the following criteria: (1) reported in 3 or more microarray studies; (2) reported in 2 or more microarray studies with agreeing upregulation or downregulation that were not further confirmed; (3) all genes that have more than 10 results from Scopus search term “keloid” and “gene name” were excluded from the study; (4) include the genes that have been reported in any microarray and; (5) a list of genes from our own unpublished data (Table 3).

In addition, candidate genes were also selected from linkage regions reported by Marneros et al.²⁰ Genes located between the markers D2S1328 and D2S2275 on chromosome 2 and between the markers D7S1818 and D7S4737 on chromosome 7 were identified using Ensembl (https://www.ensembl.org/index.html).

Thorough English language literature searches were carried out and genes likely to be involved in KD were selected. The selected genes were clustered through the tool, functional annotation clustering, provided in DAVID Bioinformatics Resources 2008 (https://david.abcc.ncifcrf.gov:8080/home.jsp). Annotation categories used for the functional clustering included all functional categories (COG_ONTOLOGY, PIR_SEQ_FEATURE, SP_COMMENT_TYPE, SP_PIR_ KEYWORDS, and UP_SEQ_FEATURE), all pathways (BBID, BIOCARTA, EC_NUMBER, KEGG_PATHWAY, and PANTHER_PATHWAY), all diseases (GENETIC_ASSOCIATION_DB_DISEASE, OMIM_DISEASE, and GENETIC_ ASSOCIATION_DB_DISEASE_CLASS), 3 protein domains (INTERPRO, PIR_ SUPERFAMILY, and SMART), and 3 gene oncology (GOTERM_BP_ALL, GOTERM_CC_ALL, and GOTERM_MF_ALL). The classification stringency was set to highest. Where possible, 1 candidate gene was selected from each cluster as a representative of the functional group.

Selection for reference genes

The selection of suitable reference genes was carried out using GeNorm.²¹ Seven reference genes were used to screen for the most stably expressed reference genes, including β-actin (ACT-β), β-2-microglobulin (β2M), glyceraldehyde-3-phosphate dehydrogenase (GAPDH), hydroxymethyl-bilane synthase (HMBS), hypoxanthine phosphoribosyl-transferase I (HPRT), ribosomal protein L32 (RPL32), and succinate dehydrogenase complex subunit A (SDHA).^21,22 The most stable reference genes across all biopsy, culture, normal, and margin samples were selected.

Reverse transcriptase-quantitative polymerase chain reaction

Reverse transcriptase-quantitative polymerase chain reactions (RT-qPCRs) were carried out using LightCycler 480 platform (Roche Diagnostics GmbH). Polymerase chain reactions were performed in 384 multiwell plates (Roche Diagnostics GmbH). Three replicates of each reaction were carried out. The reaction volume was composed of 4 µL of 1:20 diluted template cDNA, 5 µL of LightCycler 480 Probes Master (Roche Diagnostics GmbH), 0.2 µM of each primer (Metabion International AG, Martinsried, Germany) (Table 4), 0.1 µL of probe from Universal Probe Library (Roche Diagnostics GmbH), and nuclease-free water (Ambion) to make up to a total volume of 10 µL. Four microliters of nuclease-free water was used to substitute for the template cDNA for the no template controls. LightCycler 480 software, Version 1.2 (Roche Diagnostics, Burgess Hill, England) was used to specify the qPCR conditions and to calculate the threshold cycle number (C_T). The qPCR conditions were programmed as follows: First, there was 1 cycle at 95°C for 5 minutes for the activation of Hot Start Taq polymerase. Then, 45 amplification cycles were carried out; each cycle consisted of denaturation at 95°C for 10 seconds and annealing and extension at 60°C for 30 seconds. Finally, a cooling cycle was programmed at 40°C for 10 seconds. The reading of fluorescence level of the qPCRs in each amplification cycle was taken at the end of the 60°C step. The second derivative method was used for determining the C_T.

Gene expression–level analysis

The statistical significance of the difference in gene expression levels between the normal skin and margin was determined by the relative C_T method.^23,24 The Δ C_T for each gene was obtained by deducting the average C_T for the reference transcripts from the C_T for the candidate transcripts (equation 1).²⁴ Using a statistical method suggested by Yuan and Stewart,²⁴ the statistical significance of the Δ C_T of each gene in the normal and margin samples was compared with each other by paired t test with the software SPSS, Version 14.0 (SPSS, Inc).

a (1) Candidate gene C_T – reference gene C_T=ΔC_T

Because 2 copies of amplicon should be obtained during each cycle of the PCR, 2^{-Δ CT} was used to represent the relative gene expression levels in natural numbers for presenting the results in bar charts (equation 2).

(2) 2^{-Δ CT} = Relative gene expression for the candidate gene

The average fold change between normal and margin samples was then calculated averaging the 2^{-ΔΔ CT} for all patients (equations 3 and 4).²³

(3) Keloid margin ΔC_T – normal skin Δ C_T=Δ ΔC_T

(4) 2^{-ΔΔ CT} = Fold change of a candidate gene

A summary of steps taken to identify the potential biomarkers for KD is shown in Figure 2.

Figure 2. Flowchart summarizing steps taken and findings in this study. cDNA indicates complementary DNA; RPL32, ribosomal protein L32; RT-qPCR, reverse-transcription quantitative polymerase chain reaction; and SDHA, succinate dehydrogenase complex subunit A.

Candidate gene selection

A total of 47 genes have been found to be present in 2 or more microarray studies. Fifteen candidate genes, including alpha-2-macroglobulin (A2M), aggrecan (ACAN), annexin A-1 (ANXA1), asporin (ASPN), chromosome 5 open reading frame 13 (C5ORF13), epidermal growth factor receptor (EGFR), hepatoma-derived growth factor (HDGF), hypoxia-inducible factor 1, alpha subunit (HIF1A), insulin-like growth factor binding protein 5 (IGFBP5), insulin-like growth factor binding protein 7 (IGFBP7), inhibin, beta A (INHBA), lectin, galactoside-binding, soluble, 1 (LGALS1), pleiotrophin (PTN), serpin peptidase inhibitor, clade F (SERPINF1), and serpin peptidase inhibitor, clade H (SERPINH1), were selected following the selection criteria.

One hundred and sixty-eight genes were identified to be present in the 2q23 keloid linkage (between markers D2S410 and D2S1353) and 50 genes in the 7p11 keloid linkage (between markers D7S1818 and D7S473). The list of candidate genes was shortlisted to 9 genes through a comprehensive literature search for genes that directly or indirectly contribute to keloid etiopathogenesis. The criteria included relations to cell proliferation and migration, fibrosis, inflammation, MAP kinase signaling, tissue homeostasis apoptosis, tumor progression, and wound healing. The 9 genes included tumor necrosis factor-α inducible protein 6 (TNFAIP6), activin receptor IIA (ACVR2A), mitogen-activated protein kinase kinase kinase 2 (MAP3K2), myosin VIIB (MYO7B), LIM and senescent cell antigen-like domains 2 (LIMS2), dipeptidyl peptidase 10 (DPP10), bridging integrator 1 (BIN1), epidermal growth factor receptor (EGFR), and von Willebrand factor C domain containing 2 (VWC2). Eight of the 9 input genes were clustered using DAVID Bioinformatics Resources 2008 into 9 groups according to their functional categories. Five genes were then selected (Fig 3), with at least 1 gene chosen from each cluster. The selected candidate genes included TNFAIP6, LIMS2, MAP3K2, BIN1, and EGFR.

Figure 3. Functional clustering of the genes present within the keloid susceptibility loci. Selected genes present within keloid susceptibility loci, 7p11 and 2q23, have been functionally clustered using DAVID Bioinformatics Resources 2008 Functional Annotation Tool. The genes are separated into 9 separated category, and at least 1 gene from each categories has been selected for downstream quantitative polymerase chain reaction analysis. The selected genes are marked with a box around them. EGFR indicates epidermal growth factor receptor; BIN1, bridging integrator 1; LIMS2, LIM and senescent cell antigen-like domains 2; MAP3K2, mitogen-activated protein kinase kinase kinase 2; TNFAIP6, tumor necrosis factor-α inducible protein 6; ACVR2A, activin receptor IIA; DPP10, inactive dipeptidyl peptidase 10; and MYO7B, myosin VIIB.

Reference gene selection

RPL32 and SHDA were identified as most stably expressed reference genes across all samples. The pairwise variation in the 2 genes was less than 0.15, the threshold below which the inclusion of additional reference genes was not required. The 2 reference genes, RPL32 and SHDA, were therefore used as the internal control genes for the normalization in RT-qPCRs.

RNA quality

The values of RNA integrity number (RIN) for all samples were higher than 7.0; a RIN value of more than 5.0 indicated good total RNA quality, whereas a RIN value of more than 8.0 indicated best total RNA quality.²⁵

Gene expression levels of candidate genes

A statistically significant (P < .05) differential expression and a fold change of more than 2 were determined between keloid margin and internal control biopsy samples for 10 genes, including ACAN, ASPN, C5orf13, HIF1A, IGFBP7, INHBA, LGALS1, PTN, SERPINH1, and TNFAIP6 (Table 5). ACAN, ASPN, INHBA, TNFAIP6, and C5orf13 were the 5 candidate genes showing the highest average fold change in margin biopsies (fold change ≥ 9.9) (Fig 4). However, the significant high fold changes between normal and keloid margin samples was not observed in culture samples (Fig 4).

Figure 4. Relative gene expression levels in all samples for 5 genes that are highly upregulated in keloid margin. Significant upregulation have been observed in the following 5 genes in biopsies of keloid margin. However, this is not observed in the fibroblast culture equivalents. ACAN indicates aggrecan; ASPN, asporin; INHBA, inhibin, beta A; TNFAIP6, tumor necrosis factor-α inducible protein 6; and C5orf13, chromosome 5 open reading frame 13.