Keloid tissue analysis discredits a role for myofibroblasts in disease pathogenesis

Myofibroblasts, renowned for their contractility and extracellular matrix production, are widely considered the key effector cells for nearly all scars resulting from tissue repair processes, ranging from normal scars to extreme fibrosis. For example, it is often assumed that myofibroblasts underpin the characteristics of keloid scars, which are debilitating pathological skin scars lacking effective treatments because of a poor understanding of the disease mechanisms. Here, we present primary and published transcriptional and histological evidence that myofibroblasts are not consistently present in primary keloid lesions, and when alpha‐smooth muscle actin (αSMA)‐positive cells are detected, they are not greater in number or expressing more αSMA than in normal or hypertrophic scars. In conclusion, keloid scars do not appear to require αSMA‐positive myofibroblasts; continuing to consider keloids on a quantitative spectrum with normal or hypertrophic scars, with αSMA serving as a biomarker of disease severity, is hindering advancement of understanding and therapy development.


| INTRODUCTION
Keloids are pathological scars that can grow very large and beyond the original wound margin. This is an important medical condition due to their prevalence (up to 16% of certain ethnic populations including Afro-Caribbean and Japanese), their painful nature, lack of understanding about their aetiology and the paucity of effective treatment options.
Although keloids can develop spontaneously without any obvious tissue injury, generally they develop after an insult or wound to the skin. During normal skin wound repair, local fibroblasts and potentially other cell populations respond to the plethora of biochemical and mechanical cues (e.g., growth factors such as TGFβ1 and change in tissue tension), which trigger them to migrate into the wound bed and differentiate into myofibroblasts. 1 Myofibroblasts are renowned for their contractility (due to alpha-smooth muscle actin (αSMA) expression; gene: ACTA2) and extracellular matrix (ECM) production; with these functions, they make important contributions to the repair process, but are also thought to be the cell type responsible for scarring. 2 Consistent with this, there is a near-universal assumption that myofibroblasts are the culprit cells causing keloid development and growth. 3 However, it is still unclear whether keloids fall on a quantitative spectrum with normal or hypertrophic scars, with the same wound-associated myofibroblasts persisting and causing the pathology. Confusingly, the fibroblasts in keloids have been described as everything from de-differentiated myofibroblasts to "end-stage" myofibroblasts. Despite this, the keloid cell behaviours deemed pathological in in vitro studies are widely described as myofibroblastic, with αSMA ubiquitously used as a biomarker, along with Collagen I and Fibronectin expression. Nearly all cell-based research, admirably striving to develop novel treatment strategies for this disease, interpret a reduction in αSMA expression as a readout of success, but unfortunately this has never translated into a successful clinical therapy. Thus, it is imperative to revisit the tissue-based evidence about whether myofibroblasts are effectors of keloid scarring. This study interrogates published as well as new transcriptional and histological data for a myofibroblast presence in keloid scars. We report that a significant proportion of keloids are devoid of αSMA expression, and when present, its intensity is less than other scar types (e.g., hypertrophic scars) and therefore it does not distinguish this unique fibrotic condition.

| Searches
Literature searches were performed in PubMed and Google Scholar for "keloid AND myofibroblast" as well as "keloid AND smooth muscle actin (SMA)". NBCI Gene Expression Omnibus (GEO) was used to find RNA analyses of keloid tissue. The following data sets were interrogated: GSE2945, 4 GSE90051, 5 GSE92566 6 and GSE158395. 7 Graphs were created and statistical analysis was performed using Gra-phPad Prism 8.   Table 1). The largest study used immunohistochemistry to investigate αSMA protein expression in 40 keloid patient samples; 45% (18 of 45) were positive. 8     Myofibroblasts can also be identified by ultrastructural features observed by electron microscopy (e.g., prominent rough endoplasmic reticulum, stress fibres). Although these cellular features have been observed in keloid tissue, 14 they were less notable in keloids than many other scar types. 15 To summarise, although myofibroblast presence (based on αSMA positivity) has been reported in numerous keloid scar studies, there is no evidence that there is a greater abundance of these cells or a more extreme cellular phenotype in this pathological context compared to other scar types (i.e., normotrophic, hypertrophic). In fact, there are abundant data that many keloid lesions are entirely devoid of myofibroblasts.

| Transcriptional profiling of keloid tissue does not endorse a myofibroblast composition
Our second approach to evaluate whether myofibroblasts are a key cell type constituting keloid scars was to mine publicly available transcriptional profiling data of keloid tissue. There were four studies that compared gene expression between lesional and patient-matched non-lesional skin whose results were interrogated specifically for ACTA2 gene expression (three microarray and one RNAseq; Table 1). [4][5][6][7] There were remarkably small differences in

| DISCUSSION
The keloid research community has extrapolated the understanding that myofibroblasts are key effector cells in skin wound repair and scarring, 1 as well as in fibrosis of many organ systems, and assumed that myofibroblasts are also likely the causative cell driving keloid development, growth and persistence. However, with few publications on the keloid scar tissue itself, and most research relying on cell culture techniques known to trigger myofibroblast differentiation, it was important to revisit the evidence.
We first considered published histological analyses of keloid tissue for αSMA-positivity [8][9][10][11][12][13] ; although the findings were highly F I G U R E 1 RNA analysis of keloid tissue discounts significant myofibroblast composition. (A) Publicly available microarray and RNAsequencing data (GSE2945, 4 GSE90051, 5 GSE92566 6 and GSE158395 7 ) were mined for ACTA2 expression. The differential expression values (log2 fold-change, lesional/non-lesional) are plotted (mean ± SD). (B) ACTA2 expression was compared between keloid and normal skin tissue in GSE158395. 7 Normalised expression values (FPKM) are shown. (C) ACTA2 expression was analysed by qRT-PCR in 4 normal skin and keloid scar tissue samples (in technical duplicate); expression values were established against a standard curve and normalised to a reference gene (P4HB). Mean ± SD is plotted; *p < 0.05, **p < 0.01 by non-parametric Mann-Whitney test variable, from 0 to 100%, the results demonstrate that myofibroblast presence is not a requirement of disease. Also, when αSMA was detected in keloids, its intensity was not greater than other scar types, reinforcing that this is not a distinguishing feature of keloids. 8 Publications with transcriptional profiling of keloid tissue were also examined for ACTA2 gene expression. [4][5][6][7] The data were consistent in revealing very little difference in ACTA2 transcript abundance in keloid lesions compared to non-lesional tissue, and indeed slightly less expression in all cases. In line with this, Wu et al 7  did not show elevated ACTA2, 16 countering the idea that keloid fibroblasts may initially have a myofibroblast phenotype, which then progresses, regresses, or changes entirely.
Collectively, the published work regarding αSMA-positive myofibroblasts in keloids demonstrates that this is not an overabundant or overzealous cell type in these lesions. This is consistent with the clinical observations that keloids are generally not associated with contractures, whereas hypertrophic scars are. 12 Distinct constituent cells between the two scar types may also account for the histological difference (e.g., the ECM). 12,17 These observations highlight numerous unanswered questions warranting additional research: • Are αSMA-positive and αSMA-negative keloids distinct? Could and should they be stratified on this feature?
• If myofibroblasts are not forming keloids, what is the identity of the cells that are and where do they fall on a differentiation spectrum (from undifferentiated and potentially plastic 3 through to highly differentiated and stable)?
• TGFβ1, its regulators and its downstream signalling pathways have been widely implicated in keloid pathogenesis; what are their roles in this context, if not triggering myofibroblast differentiation?
• Is contractility a functionally significant feature of keloid cells? If yes, what machinery is driving this, if not αSMA?
Considering the tissue-based evidence presented herein that keloids are not masses of overactive contractile myofibroblasts, it is essential to recognise that keloids may not fall on a quantitative spectrum with normal or hypertrophic scarring but rather could be qualitatively distinct. 6,17 Appropriate biomarkers of keloid disease are urgently needed, as is sufficient understanding of the disease processes in order to establish the pathological cellular features important to target clinically.