Abstract

   Keloids, characterized by skin fibrosis and excessive accumulation of
   extracellular matrix, remain a therapeutic challenge. In this study, we
   systematically capture the cellular composition of keloids by the
   single-cell RNA sequencing technique. Our results indicated that there
   are significant differences in most cell types present between 12 pairs
   of keloid and adjacent normal tissue. We found that fibroblasts,
   endothelial cells, mast cells, mural cells, and Schwann cells increased
   significantly in keloid. The proportion of mesenchymal fibroblast
   subpopulations in keloids was markedly higher than those in the
   surrounding normal skin tissue. Furthermore, we found that the immune
   profiles between two groups varied significantly. The proportion of
   macrophages in the keloid was significantly elevated compared to the
   surrounding normal tissue, while cDC2 cells significantly decreased.
   Hotspot and pseudotime trajectory analysis indicated two modules of
   macrophage cells (Module2: highly expresses RNASE1, C1QA, CD163, CD14,
   C1QC, FCGRT, MS4A7; Module10: highly expresses APOC1, CTSB, CTSL,
   TYROBP), which exhibited the characteristics of tumor-associated
   macrophages, were upregulated in more-advanced keloid cells.
   Subsequently, the analysis of cellular communication networks suggested
   that a macrophage-centered communication regulatory network may exist
   in keloids and that fibroblasts in keloids may facilitate the
   transition and proliferation of M2 macrophages, which contributes to
   further comprehension of the immunological features of keloids.
   Overall, we delineate the immunology landscape of keloids and present
   new insights into the mechanisms involved in its formation in this
   study.

   Keywords: keloid, single-cell RNA sequencing, fibroblast, macrophage,
   immunology profiles

Introduction

   Keloid (K) is the result of abnormal tissue repair caused by a variety
   of factors ([47]1, [48]2). Because of its clinical characteristics of
   continuous growth, no self-limitation, easy recurrence after resection,
   and invasive growth to normal skin, it has brought serious damage and
   psychological burden to patients ([49]3). Keloid tissue is composed of
   cellular components and extracellular matrix components (ECMs) ([50]4,
   [51]5). The immune cells play a decisive role in the biological
   behavior of keloids. Keloid fibroblasts proliferate abnormally and
   secrete too much extracellular matrix, which have a strong
   proliferation ability ([52]6). Keloid fibroblasts have always been
   considered to be the cause of the vigorous proliferation of keloids,
   but the influence of the surrounding immune microenvironment has been
   ignored. Keloid fibroblasts and immune cells interact with each other
   and develop synergistically and the complex regulation of their
   molecules constitutes the immune microenvironment of keloids ([53]4).
   Studying the changes of the immune microenvironment and its regulatory
   mechanism will benefit keloid treatment by regulating important immune
   cells.

   Single-cell transcriptome sequencing is a novel technique for
   sequencing the transcriptome of a single cell as a specific research
   object and performing bioinformatics statistical analysis on the
   obtained data ([54]7). Based on technologies such as microfluidics and
   barcode labeling, a huge number of large cells can be rapidly and
   efficiently labeled, sequenced, and analyzed, and gene expression
   profiles and differences at the single-cell level can be obtained
   ([55]7, [56]8). Several previous studies have applied single-cell
   sequencing techniques in the field of keloids, but they mainly focus on
   the study of extracellular matrix and fibroblasts ([57]9–[58]11). This
   study focuses on the study of the immune microenvironment of keloids,
   revealing immunology profiles in human keloids.

Methods

Patient

   This study was approved by the Medical Ethics Committee of Peking Union
   Medical College Hospital (JS-2907). All participants provided written
   informed consent. From October 2019 to January 2022, 12 patients with
   keloids were included in this study. Surgically excised skin tissue was
   collected from the normal skin tissue of cosmetic surgery patients.
   Keloid skin (group K) and the surrounding normal skin (group N) served
   as controls for single-cell sequencing.

Tissue dissociation and preparation

   Fresh keloid and surrounding normal skin tissue was obtained and stored
   in GEXSCOPETM Tissue Preservation Solution and transferred to the
   laboratory frozen immediately. All specimens were cleansed thrice with
   Hanks Balanced Salt Solution (HBSS) and prepared into 1–2-mm pieces,
   which were subsequently digested in a GEXSCOPETM tissue dissociation
   solution (Singleron Bio Com, Nanjing, China). After the samples were
   filtered and centrifuged and the supernatant discarded, the sediment
   was resuspended in phosphate buffered saline (PBS) (HyClone, China).
   Specimens with excess erythrocytes were treated with GEXSCOPETM
   Erythrocyte Lysate (Singleron Bio Com, Nanjing, China) for
   de-erythrocyte removal. Samples were stained with trypan blue (Sigma,
   United States) and evaluated microscopically.

Library construction and single-cell sequencing

   The cell concentration of the single-cell suspension was maintained at
   1 × 10^5 cells/ml and loaded onto a microfluidic chip. Single-cell
   RNA-seq (scRNA-seq) libraries were constructed using the GEXSCOPETM
   Single Cell RNA Library Kit (Singleron Biotechnologies) ([59]12).
   Individual libraries were diluted to 4 nM and sequencing was completed
   on the Illumina HiSeq X sequencing platform, selecting the 150-bp
   double-end mode.

Contamination removal

   The decontx algorithm in the celda package (v.1.3.8) performed
   environmental RNA removal using raw UMI counts and cell clusters
   calculated by the FindClusters function. A default random seed was
   applied throughout the analysis to ensure reproducibility.

scRNA-seq quantifications

   Raw data, which were processed after filtering out reads1 without poly
   T, screening out reads2 (fastp V1) with connectors and poly A tails,
   and comparing with the reference genome from the ensembl database (STAR
   2.5.3a and featureCounts 1.6.2), were generated as a gene expression
   matrix ([60]13). The data were grouped according to cell barcodes, and
   the number of UMIs was calculated per gene in each cell. With the
   Seurat package ([61]http://satijalab.org/seurat/, R package, v.3.0.1),
   gene expression matrices were created as Seurat objects to perform cell
   type annotation and downscaling clustering analysis ([62]14). The
   FindCluster function was set to a resolution of 0.6 to cluster the
   mixed cells, and the FindMarker function was applied to identify
   differentially expressed genes in different clusters of cells. The
   functional annotation of gene sets was implemented with the
   clusterProfiler package ([63]15).

Quality control, dimension-reduction, and clustering (Scanpy)

   For each sample dataset, the expression matrix was filtered through the
   Scanpy package v1.8.2 ([64]16) based on the following exclusion
   criteria: 1) cells with the maximum 2% of UMI counts; 2) cells with the
   maximum 2% of gene counts; 3) cells with fewer than 200 genes; 4) genes
   expressed in less than five cells; and 5) cells with >50% mitochondrial
   gene content. For the screened raw count matrix, normalization was
   performed by dividing by the total counts per cell and calculating the
   natural logarithm. The first 2,000 variable genes were selected, and
   the matrix was subjected to principal component analysis, with the
   first 20 principal components serving for subsequent clustering and
   dimensionality reduction analysis. The removal of batch effects between
   samples was implemented by Harmony v1.0 ([65]17). The clustering of
   cells was achieved by the Louvain algorithm, setting the resolution
   parameter to 1.2, and visualized by Uniform Manifold Approximation and
   Projection (UMAP).

Differentially expressed gene analysis

   Differentially expressed genes were identified by the FindMarker
   function based on the Wilcox likelihood ratio test with the
   qualification that they are expressed in more than 10% of cells in one
   cluster and with an average log (fold change) greater than 0.25.

Cell type annotation

   The cell types of each cluster were determined by the expression of the
   identified differential genes and typical cellular markers reported in
   the previous literature, and the visualization of typical differential
   genes in cell clusters was achieved using heatmaps.

Pathway enrichment analysis

   Gene ontology (GO), Kyoto Encyclopedia of Genes and Genomes (KEGG)
   analysis, and GSVA pathway enrichment analysis were used to predict the
   biological function, cellular composition, and possible pathways
   involved in differentially expressed genes, with an adjusted P-value of
   less than 0.05 being considered statistically significant ([66]18).

Trajectory analysis

   Monocle2 was employed for pseudo-temporal trajectory analysis, and the
   DDRTree method was adopted for dimensionality reduction to map the
   differentiation of macrophage subtypes in keloid. The results were
   visualized by the “plot_cell_trajectory” function ([67]19).

RNA velocity

   The velocyto package was utilized to analyze BAM files containing
   target cells and genomic GRCh38 (hg38) for RNA rate analysis, and the
   results were projected onto UMAP plots for Seurat clustering analysis
   for visual consistency ([68]20, [69]21).

Single-cell entropy analysis

   SLICE (version 0.99.0) was used to evaluate the stemness of cells by
   the entropy of gene expression based on single-cell expression
   profiles. After removing and ribosomal genes, a SLICE object was
   created to perform a bootstrap calculation of single-cell gene entropy
   values by the getEntropy function.

Cell–cell interaction analysis (CellPhoneDB)

   Cellphone DB v2.1.0 was equipped with a detailed receptor–ligand
   database to predict cell–cell interactions to identify cell pairs in
   the presence of reciprocal regulation or communication. The number of
   ineffective distributions of reciprocal exchanges was set to 1,000. The
   predicted interaction pairs with a p-value <0.05 and of average log
   expression >0.1 were considered as significant ([70]22).

Transcription factor regulatory network analysis (scenic)

   The SCENIC R toolkit was used to construct transcription factor
   networks on scRNA expression matrices. The GENIE3 package inferred
   coexpression modules between transcription factors and candidate target
   genes. The RcisTarget package performed cis-regulatory motif analysis
   for each coexpression module, constituting a gene regulatory network
   module containing both TF and target genes. The AUCell package
   calculated the AUC values of this genome to assess the activation of
   the gene regulatory network modules in cells.

Expression program analysis

   The transcriptional programs were extracted by the cNMF algorithm, the
   top 50 genes were taken as the meta-signature, and the score of each
   program per cell was computed by the meta-signature. The meta-program
   was calculated and there was hierarchical clustering according to the
   Pearson correlation between each program.

Immunohistochemical staining

   Immunohistochemical experiments were performed according to the
   streptavidin–biotin–peroxidase complex procedure. The tissue sections
   of keloid and surrounding skin were dewaxed and rehydrated with xylene
   and different concentrations of ethanol, kept in 3% hydrogen peroxide
   solution for 20 min, and subsequently rinsed with distilled water for
   5 min. Samples were treated with primary antibody working solutions
   against CD8 (ProteinTech, #66868-1-Ig,1:400), CD4 (ProteinTech,
   #67786-1-Ig, 1:450), CD68 (ProteinTech, #66231-2-Ig,1:1,000), iNOS
   (ProteinTech, #22226-1-AP,1:500), IL-10 (ProteinTech,
   #60629-2-Ig,1:500), and IL-12 (Abcam, #ab9992, 1:1,000), respectively.
   The samples were incubated with biotin-labeled secondary antibodies for
   30 min at 37°C. The sections were added with an appropriate amount of
   an alkaline phosphatase–labeled streptavidin working solution and
   incubated at 37°C for 30 min. After rinsing with PBS, the chromogenic
   agent was treated for 15 min. The slices were restained, dehydrated,
   and sealed.

Results

Characteristics of the overall cell types in human keloids by single-cell
RNA-seq

   To explore the immunological profile of keloids and to determine
   cellular heterogeneity, we used single-cell RNA sequencing to compare
   keloid skin tissue (CASE) with matched adjacent normal tissue (CTRL)
   from 12 patients ([71]Supplementary Table 1). The 24 samples were
   dissociated into single cells and subjected to single-cell RNA-seq
   ([72]Figure 1A). The transcriptomes of 427171 number cells (CASE:
   213105; CTRL: 214066) were yielded after quality control. Based on
   hierarchical clustering and established lineage-specific marker genes
   ([73]Supplementary Table 2), we assigned these clusters to 11 cell
   lineages ([74]Figures 1B–D). The keratinocyte lineage (marked by KRT1,
   KRT10, KRT14, KRT5) ([75]23, [76]24) constituted the maximum percentage
   (71.7%) of cells. Endothelial cells (ECs) (marked by PECAM1, CDH5, VWF,
   and KDR) ([77]25) represented the second biggest proportion (5.7%) of
   cells, while the fibroblast lineage made up the third biggest
   proportion (4.4%) of cells (marked by DCN, COL1A1, and COL1A2) ([78]25,
   [79]26). Additionally, other classic skin lineages were identified,
   like sweat gland cells (labeled by DCD and AQP5) ([80]24), mural cells
   (marked by RGS5, ACTA2, TAGLN, and MYLK) ([81]26, [82]27), Langerhans
   cells (marked by CD207 and CD1A) ([83]27), melanocytes (marked by
   MLANA, PMEL, DCT, and TYRP1) ([84]28), and Schwann cells (marked by
   S100B, PMP22, and PLP1) ([85]29, [86]30). These clusters revealed
   diverse molecular profiles ([87]Figure 1E), capturing the cellular
   variety and heterogeneity of keloid tissue.

Figure 1.

   [88]Figure 1
   [89]Open in a new tab

   Single-cell RNA-seq (scRNA-seq) reveals the cellular diversity and
   heterogeneity of keloid skin tissue. (A) Schematic representation of
   the experimental procedure. Keloids and adjacent normal skin tissues
   were harvested separately at the time of surgery (n=24). (B) Uniform
   Manifold Approximation and Projection (UMAP) of unbiased clustering of
   427,171 cells reveals 11 cellular clusters. Clusters are distinguished
   by different colors. The number in the parenthesis indicates the cell
   count. (C) Hierarchical clustering of the clusters based on the average
   expression of the 2,000 most variable genes. (D) Comparison of the
   ratio of cells in keloids and adjacent normal skin tissues. (E)
   Representative molecular signatures for each cell cluster. SGCs, sweat
   gland cells; ECs, vascular endothelial cells; MPs, mononuclear
   phagocytes.

Characteristics of fibroblasts in keloids

   To account for the significant changes in fibroblasts during the
   fibrotic process in keloids ([90]Figures 1C, D) and the dominance of
   fibroblasts in the pathogenesis of fibrosis, we proceeded to perform an
   unsupervised clustering analysis of fibroblasts in all keloids and
   normal skin. A previous study proved that normal human dermal
   fibroblasts can be divided into four subpopulations:
   secretory-papillary (SPF), secretory-reticular (SRF), mesenchymal (MF),
   and pro-inflammatory (PF) ([91]31). Hierarchical cluster analysis
   suggested that fibroblasts from keloid and normal tissue in our study
   could also be divided into these four subpopulations ([92]Figures 2A–C)
   ([93]Supplementary Table 2). Moreover, we found that the proportion of
   MF increased significantly in keloids compared to normal tissue, while
   SPF and SRF decreased ([94]Figure 2B). We further explored
   differentially expressed genes between these subpopulations in keloid.
   We found that genes associated with matrix synthesis and fibroblast and
   osteoblast differentiation, including ASPN, POSTN, CTHRC1, FN1, and
   COL11A1, and secretory proteins, such as SPARC, SFRP4, COMP, COL11A1,
   and COL12A1, were enriched in the mesenchymal subpopulation
   ([95]Figure 2D).

Figure 2.

   [96]Figure 2
   [97]Open in a new tab

   Fibroblasts of keloid and normal skin tissue subcluster into distinct
   cell populations. (A) Subclustering of keloid and normal tissue
   fibroblasts identified four distinct subtypes: mesenchymal (MF),
   pro-inflammatory (PF), secretory-papillary (SPF), and
   secretory-reticular (SRF). (B) Demonstration of the ratio of the four
   fibroblast subpopulations in keloids and adjacent normal tissue. (C)
   Color-coded UMAP plot is shown, and each fibroblast subcluster is
   defined. (D) Heatmap plots showing representative differentially
   expressed genes between the four fibroblasts in the keloids. (E) Gene
   ontology (GO) biological process enrichment analysis of differentially
   expressed genes in the four fibroblasts between keloid and normal
   scars. (F) GO enrichment analysis of differentially expressed genes in
   mesenchymal fibroblasts.

   scRNA-seq analysis demonstrated that the ratio of keloid mesenchymal
   fibroblasts was markedly elevated in comparison to normal skin. Thus,
   our following work concentrated on this fibroblast subpopulation. The
   heatmap indicated a marked difference in gene expression between the
   mesenchymal subpopulation and other subpopulations ([98]Figure 2D). The
   keloid and normal skin varied in the expression of a variety of genes
   among fibroblast subtypes, and this differentiation was particularly
   evident in mesenchymal fibroblasts. Genes involved in extracellular
   matrix formation including COL1A1, COL3A1, POSTN, COL1A2, FN1, and ASPN
   were significantly upregulated in keloid compared to the normal skins
   ([99]Supplementary Figure 1A). GO analysis revealed that differentially
   expressed genes in the mesenchymal subpopulation were associated with
   protein secretion, oxidative phosphorylation, cholesterol homeostasis,
   MYC, and E2F ([100]Figures 2E, F), whereas SPF was correlated with
   TGF-β, hypoxia, epithelial mesenchymal transition, angiogenesis,
   inflammatory response, apoptosis, and IL2 ([101]Figure 2E).
   Furthermore, GO analysis indicate that angiogenesis, protein secretion,
   EMT, and the PI3K/AKT pathway enriched in all of the four types of
   keloid fibroblast ([102]Figure 2E).

Characteristics of immunology profiles in human keloids

   The scRNA-seq analysis revealed that the immune profiles between the
   keloids and adjacent tissues also varied significantly, especially the
   mononuclear phagocytic system (MPS). The proportion of macrophages was
   considerably elevated in keloids compared to normal tissue, while the
   proportion of cDC2 dendritic cells was reduced ([103]Figures 3A–C). The
   change of other myeloid cells type is small ([104]Figure 3B). The
   heatmap indicated that gene expression was markedly diverse among MPS
   ([105]Figure 3D), with increased genes in macrophages including some
   RNASE1, SELENOP, C1QC, C1QB, C1QA, FOLR2, SLC40A1, APOE, LGMN, and CTSD
   ([106]Figure 3D). The increased genes in the cDC2 included CD1C, FCENA,
   CD1E, CD52, INSIG1, CLEC10A, IL1R2, HSPAa, and GOS2. Furthermore, it
   indicated that the ratio of M2 macrophage subpopulations was higher
   than M1 macrophage subpopulations in keloid ([107]Figures 3E, F), and
   the heatmap showing representative differentially expressed genes in M1
   and M2 macrophage subpopulations ([108]Figure 3G). The expression of
   genes that participate in mitochondrial oxidative phosphorylation
   processes such as MT-CO1, MT-CYB, MT-ND5, MT-CO3, and genes involved in
   extracellular matrix binding and formation, like COL1A1 and SPARC, were
   elevated in M2 macrophages in keloids, while the expression of
   extracellular matrix catabolic proteins as MMP9 was diminished
   ([109]Supplementary Figure 1B). In addition, the hotspot analysis of
   macrophage cells, in which cells are grouped according to the degree of
   similarity in their transcriptomes, shows 13 modules
   ([110]Figures 3H–J). Among them, Module2 (highly express RNASE1, C1QA,
   CD163, CD14, C1QC, FCGRT, and MS4A7) and Module10 (highly express
   APOC1, CTSB, CTSL, and TYROBP) are more likely to display the
   characteristics of tumor-associated macrophages (TAMs).

Figure 3.

   [111]Figure 3
   [112]Open in a new tab

   Characteristics of immunology profile in human keloids. (A) Color-coded
   UMAP plot is shown, and each immune cell subpopulation is defined as:
   macrophages, conventional type 1 dendritic cells (cDC1s), conventional
   type 2 dendritic cells (cDC2s), mature dendritic cells (MatureDCs),
   proliferating mononuclear phagocytes (ProliferatingMPs), plasmacytoid
   dendritic cells (pDCs), mast cells, neutrophils, and monocytes. (B, C)
   The proportions of immune cell subpopulations in keloid and adjacent
   normal tissue were shown. (D) Heatmap plots showing representative
   differentially expressed genes between the different types of immune
   cells in the keloids. (E, F) Demonstration of the ratio of M1 and M2
   macrophage subpopulations in keloid and adjacent normal tissues. (G)
   Heatmap showing representative differentially expressed genes in M1 and
   M2 macrophage subpopulations. (H, I) Demonstration of representative
   modules of macrophages in keloid. (J) Pseudo-trajectory and cell source
   transition of macrophage cells between keloid and adjacent normal
   tissue.

   Considering the essential role of T cells in immunity, we proceeded to
   perform a descending clustering of T-cell populations and identified
   six cell types, including CD4+naïve T cells, CD4+ effector memory T
   cells, CD8+ effector T cells, CD8+ mucosa–associated invariant T cells,
   CD4+ regulatory T cells, and proliferative T cells, with the proportion
   of CD8+ effector T cells predominating in keloids ([113]Figures 4A–C).
   The ratio of CD4+naïve T cells and proliferating T cells was decreased
   in keloids compared with adjacent normal skin tissue ([114]Figure 4B),
   and Augur analysis suggested that proliferative T cells were the cell
   type with the greatest variation ([115]Figure 4D). Enrichment analysis
   suggested distinct differences in T cell subtypes between the two
   groups in terms of angiogenesis, myogenesis, interferon alpha response,
   and interferon gamma response ([116]Figure 4E). The ratio of immune
   cells was verified by immunohistochemistry, and the results indicated
   that there were more CD8+ and CD68+ cells and fewer CD4+ cells in the
   keloid compared to the surrounding normal skin tissue ([117]Figure 5).
   The expression of some inflammatory factors also differed between the
   two groups. The expression of iNOS, IL-10, and IL-12 were higher in
   keloids ([118]Figure 5). The results of the image mass spectrometry of
   an extensive immune panel including 25 markers indicated that CD68,
   CD14, and CD16 positive macrophages and CD11c-positive DC cells were
   enriched in keloids, which is consistent with the results of scRNA-seq
   ([119]Supplementary Figure 2).

Figure 4.

   [120]Figure 4
   [121]Open in a new tab

   T cells of keloid and normal skin tissues were subdivided into
   different cell subsets. (A) Six different subsets of T cells are
   identified in keloid and normal tissues: CD4+ naïve T cells
   (CD4NaiveT), CD4+ effector memory T cells (CD4Tem), CD8+ effector T
   cells (CD8Teff), CD8+ mucosa-associated invariant T cells (CD8MAIT),
   CD4+ regulatory T cells (Treg), and proliferating T cells. (B) The
   ratio of six T-cell subsets in keloid and adjacent normal tissues is
   shown. (C) Heatmap illustrating representative differentially expressed
   genes among the six T-cell subsets. (D) Augur analysis indicates that
   proliferative T cells are the cell type with the greatest variation
   among subsets. (E) Enrichment analysis of differentially expressed
   genes in the six T-cell subpopulations.

Figure 5.

   [122]Figure 5
   [123]Open in a new tab

   Typical images of immunohistochemical staining in the keloid (group K)
   and surrounding normal skin tissue (group N). Gross appearance: ×400.

Cell communications

   We investigated cell–cell communication with CellPhoneDB, and the
   results demonstrated different types of cellular communication between
   keloids and adjacent normal tissues. An intensive communication network
   between fibroblasts and other cells was observed in normal skin tissue
   and keloid ([124]Figures 6A, B). The most abundant interactions in both
   cases appeared between the four fibroblast subpopulations, indicating
   the importance of fibroblast interaction signals in the dermis.
   Furthermore, we present the observation of a dense communication
   network between macrophages, DC cells, and other monocytes, which also
   indicated the importance of these cells in skin. In addition, we found
   that the cell interactions among keloids are much more strong than
   normal tissues.

Figure 6.

   [125]Figure 6
   [126]Open in a new tab

   Potential ligand–receptor interactions analysis in fibroblast
   subpopulations. (A) Heatmap showing the numbers of interpopulation
   communications with each other in keloid and adjacent normal tissue.
   (B) The ligand–receptor pairs exhibit significant changes in
   specificity between any one of the populations and any one type of
   fibroblast in adjacent normal tissue versus keloid. (C) Putative and
   relative signaling or transcriptional factors (TFs) within fibroblasts
   and other cell populations’ keloid and adjacent normal tissue.

   Since macrophages may be the major immune cells contributing to
   keloids, we evaluated specific interactions among macrophages and
   additional cell types ([127]Figure 6A). The interaction of macrophages
   with fibroblasts was strikingly noted ([128]Figure 6A). Fibroblasts in
   keloids may specifically interact with macrophages and other MPs via
   TGFB2–TGFBR2, TGFB3– TGFBR3, CCL2-CCR1, and CCL2-CCR5 binding
   ([129]Figure 6B), which play inhibitory roles on the target cells. We
   also found that the interactions between macrophage cells with the T
   cell and macrophage itself were remarkable in keloid. Furthermore, we
   found that the fibroblasts may activate the transcription factors ATF4
   and RB1 in MPs ([130]Figure 6C). Since the downstream genes of RB1 are
   related with M2 macrophages, for example: C1QC,C1QB,MRC1, fibroblasts
   in keloid may be beneficial for the transition and proliferation of M2
   macrophages.

Discussion

   Keloid scars are the consequence of an overexpansion of the
   extracellular matrix during wound healing. Multiple factors, consisting
   of limited autophagy-mediated degeneration, the chronic inflammatory
   status, and the reorientation of fibroblasts into myofibroblasts,
   contribute to excessive skin fibrosis ([131]32–[132]34). A multitude of
   immune cells and cytokines are involved in the formation and
   progression of the above pathological processes. Despite extensive
   research on keloids, the key mechanisms of keloids have not been
   thoroughly elucidated, especially the role of immune cells including
   macrophage cells, dendritic cells, and T cells in keloid pathology is
   poorly understood, and methods for early identification, prevention,
   and treatment of keloids are still scarce ([133]35).

   With the increasing maturity of high-throughput sequencing technology,
   transcriptome sequencing has been widely adopted in the fields of
   biological physiological mechanism research, genetic breeding, and
   pathological analysis. Traditional bulk RNA-seq can merely reflect the
   gene expression of a vast collection of cells, while for some tissues
   with high cell heterogeneity, bulk RNA-seq does not fully reveal the
   fine-scale cellular distribution, and it is essential to investigate
   the mechanisms of gene expression variations and cellular
   differentiation at the level of single cells ([134]36). Therefore, the
   aim of this paper is to depict the cellular landscape of keloids and
   highlight the types of immune cell composition in keloids. These
   findings facilitate a better understanding of the immune
   microenvironment of keloids, which may provide new targets for the
   immunotherapy of keloids.

   Keloids are composed of abundant, swirling, irregular collagen fibers
   that are partially infiltrated with inflammatory cells and chemokines,
   forming their unique histological structure. Fibroblasts are recognized
   as the central cells of keloids, but the immune infiltrative profile
   around keloids remains underevaluated ([135]37). This study revealed a
   significant increase in the proportion of mesenchymal fibroblasts in
   keloids, which are involved in the expression of genes related to
   matrix synthesis and fibroblast and osteoblast differentiation. The
   identification of the cell subpopulations of mononuclear phagocyte
   clusters implied a decreased proportion of cDC2 in keloids compared to
   normal tissues. As key antigen-presenting cells that initiate and shape
   the immune response, dendritic cells are critical in recognizing
   pathogens, promoting naïve T-cell activation, and maintaining
   peripheral T-cell homeostasis ([136]38, [137]39). A recent study by
   Mayer et al. revealed that interleukin (IL)-13, produced by type 2
   innate lymphocytes, is an essential molecule for maintaining cDC2
   homeostasis and suppressing inappropriate inflammatory responses
   through the IL-13-STAT6 axis ([138]40). IL13 downregulates the
   macrophage activity and inhibits the production of pro-inflammatory
   cytokines and chemokines ([139]41, [140]42). Thus, the reduction of
   cDC2 in keloids may be associated with the downregulation of IL-13
   expression, and the possible inappropriate inflammatory response
   associated with the downregulation of IL-13 expression is consistent
   with the pathogenesis of keloids.

   Macrophages can polarize into M1-type and M2-type macrophages under
   different microenvironments in addition to their original ability to
   phagocytose bacteria and antigen presentation, and M2-type macrophages
   are critical for the induction of skin fibrosis during wound healing
   ([141]43). In our study, it indicated that the ratio of M2 macrophage
   subpopulations was higher than M1 macrophage subpopulations in keloid.
   Furthermore, two modules of macrophages (Module2: highly expresses
   RNASE1, C1QA, CD163, CD14, C1QC, FCGRT, and MS4A7; Module10: highly
   expresses APOC1, CTSB, CTSL, and TYROBP), which exhibited the
   characteristics of TAMs, increased significantly in keloid. Since TAMs
   usually refer to the versatile macrophage M2 phenotype, M2 cells
   perform their multiple tasks by interacting with Th2 cells, regulatory
   T cells (Tregs), and fibroblasts ([142]44, [143]45). Their ability to
   scavenge receptors, promote angiogenesis, and express anti-inflammatory
   cytokines is superior to that of M1-type macrophages ([144]45,
   [145]46). M2 macrophages in fibrotic diseases maintain fibroblast
   proliferation, enhance the expression of the tissue inhibitors of
   metalloproteinases, induce fibroblast differentiation into
   myofibroblasts by secreting growth factors like the transforming growth
   factor (TGF-β1) and platelet- derived growth factor (PDGF), and thus
   directly or indirectly promote collagen synthesis ([146]47, [147]48).
   Furthermore, TAMs secrete factors like IL-10 that induce the expression
   of monocyte costimulatory molecules and cooperate with regulatory T
   cells by producing CCL17 and CCL22 to suppress the function of
   cytotoxic T lymphocytes and participate in the formation of a local
   immunosuppressive environment ([148]49).

   Mononuclear phagocytes and T cells presented distinct quantitative
   differences between the keloids and the surrounding normal skin
   tissues. Further dimensionality reduction clustering revealed that T
   cells were subdivided into six cell types, including naïve T cells,
   CD4+ effector memory T cells, CD8+ effector T cells, CD8+
   mucosal-associated invariant T cells, CD4+ regulatory T cells, and
   proliferating T cells, of which the proportion of CD8+ effector T cells
   was predominant in keloids. The results of functional enrichment
   analysis suggested that T-cell subtypes in keloid and normal tissues
   were associated with epithelial mesenchymal transition and
   angiogenesis. The naïve T cells are enriched in homeostatic skin after
   receiving antigen presentation from Langerhans cells and dendritic
   cells, coordinated by specific chemokine receptors like CCR4, and
   gradually start clonal proliferation and cytokine release ([149]50,
   [150]51). The decreased ratio of naïve T cells and proliferating T
   cells and the increased ratio of effector T cells are consistent with
   the previous knowledge that keloids are closely associated with a
   regional chronic inflammatory response.

   The findings of cellular communication implicated that the macrophages
   may interact with fibroblasts through TGFB2-TGFBR2, TGFB3-TGFBR3,
   CCL2-CCR1, and CCL2-CCR5 binding. Previous studies have suggested that
   in normal tissues, a symbiotic relationship exists between fibroblasts
   and macrophages in a two-cell circuit, interacting with each other for
   mutually required growth factor signals. This means that fibroblasts
   maintain macrophage proliferation by providing CSF1, while fibroblast
   survival is dependent on PDGF ligands provided by macrophages
   ([151]52). In fibrotic diseases where fibroblasts are the driver cells,
   fibroblasts provide signals for macrophage chemotaxis, resident, and
   activation by maintaining the expression of CSF1, interleukin 6, and
   the macrophage chemokine CCL2 ([152]52, [153]53). Activated macrophages
   secrete multiple factors such as TGF-β, which individually or
   synergistically drive fibroblast proliferation and activation. Previous
   studies have reported that activated macrophages are key cells involved
   in extracellular matrix degradation and collagen breakdown ([154]54);
   therefore, the dual role and polarization state of macrophages and
   their interaction with fibroblasts influence the extent of fibrosis.

   A keloid is the paradigm of cutaneous fibrosis, and many keloid-based
   findings have been confirmed in other diseases ([155]2). Pro-fibrotic
   macrophages have been identified in the single cell profiles of
   idiopathic pulmonary fibrosis, which upregulate extracellular
   matrix–related gene expression like SPARC and MMP9 and can construct
   autocrine feedback loops by expressing CSF1 ([156]55). Aran et al.
   identified macrophage heterogeneity in a model of lung fibrosis and
   identified the subpopulations of macrophages that promote mesenchymal
   formation in the fibrotic fraction and characterized specific markers
   with therapeutic implications, including CX3CR1 and PDGF-AA ([157]56).
   In liver fibrosis, macrophages are involved in the liver inflammation,
   metabolism, and activation of hepatic stellate cells to promote
   fibrosis and can facilitate the decomposition of the extracellular
   matrix during the degenerative phase, and clinical trials have shown
   that the use of autologous macrophage infusion can improve the level of
   cirrhosis ([158]57, [159]58). CD163+ macrophages express chemokines
   associated with perivascular inflammation, and they recruit
   inflammatory cells into the perivascular environment in scleroderma
   ([160]59). The heterogeneity and different activation states of
   macrophages may have different effects on fibrosis in scleroderma
   ([161]60). This fits well with our findings of macrophage changes in
   keloid formation and their close association with fibroblasts.
   Furthermore, studies in recent years have gradually identified
   mesenchymal fibroblasts as one of the most contributing cells in
   keloids. This phenomenon may be of general significance, as the
   expression of the markers POSTN and COMP of mesenchymal fibroblasts has
   also been observed in scleroderma fibroblasts ([162]9, [163]61).

   In conclusion, our study was based on single-cell sequencing technology
   to deeply explore the immune cell infiltration in keloid, and we
   identified differences in the infiltration of macrophages, mast cells,
   and dendritic cells between normal tissues and keloids, especially the
   close interaction between macrophages and fibroblasts in keloids, which
   could, to some extent, contribute to our understanding of the
   pathogenesis of keloids.

Data availability statement

   The datasets presented in this study can be found in online
   repositories. The name of the repository and accession number can be
   found below: NCBI Sequence Read Archive; PRJNA844167.

Ethics statement

   The studies involving human participants were reviewed and approved by
   the Medical Ethics Committee of Peking Union Medical College Hospital
   (Beijing, China). The patients/participants provided their written
   informed consent to participate in this study.

Author contributions

   CF, MS, YX, and YongH performed the data analysis and were major
   contributors to the preparation of the manuscript. KH, YouW, and ZZ
   provided technical support in the methodology of the raw letter
   analysis. CF, KS, TM, HL, ZL, YingW, and YanH made significant
   contributions to the design of the study. CF, YouW, and YongH confirmed
   the authenticity of all raw data. All authors read and approved the
   final manuscript.

Funding

   This study was supported by The National Natural Science Foundation of
   China (81871538) and Beijing Municipal Commission of Science and
   Technology (Z191100006619009).

Acknowledgments