The study was performed at the Burn Department, Ruijin Hospital, Shanghai Jiao Tong University School of Medicine, from February to December, 2021, and was approved by the Ethics Committee of Ruijin Hospital, Shanghai Jiao Tong University School of Medicine. All participants provided written informed consent.

Patients (2–61 years of age, male 19 and female 14) with burn-related HS, located on the limbs or trunks, which were featured by elevation, redness, and hardness, and needed scar excision and skin transplantation, were enrolled in this study. Meanwhile, 36 normal skin (NS) samples were harvested from these patients or other patients after skin transplantation, and all the donor sites were abdomen.

The patients with scar duration ranging from 6 to 24 months were included, none of the participants received any antibiotics (systemic or oral therapy) within recent 3 months, and patients with keloids, scar ulcers, diabetes, or cardiovascular disease were excluded.

Before surgery, the scars were assessed by Vancouver Scar Scale (VSS) score, and the indices of pigment, thickness, pliability, itching/pain, and vessel were assessed for each sample.

During the surgery, the process was strictly disinfected to avoid bacterial contamination.

After surgery, the tissue was quickly put into the sterile specimen box, which had been irradiated under a UV light for 90 min, then tissue samples were further irradiated with UV light for 30 min, and subsequently washed 3 times with sterile PBS.

The samples were then fixed in 10% buffered formalin, embedded in paraffin blocks, and cut into 6-μm-thick sections. The sections were subjected to hematoxylin and eosin (H&E) staining, Gram staining, electron microscopy, in situ hybridization, and immunohistochemistry.

H&E staining was performed according to a standard protocol. Briefly, after deparaffinization and rehydration, sections were stained with hematoxylin solution for 5 min followed by 5 dips in 1% acid ethanol (1% HCl in 70% ethanol) and then rinsed in distilled water. Then the sections were stained with eosin solution for 3 min and followed by dehydration with graded alcohol and clearing in xylene. The mounted slides were then examined and photographed using an Olympus BX53 fluorescence microscope (Tokyo, Japan). The experiment was repeated three times, the number of inflammation cells was calculated per view, and 8 samples were tested.

Gram staining was performed according to the protocol of the manufacturer (Beijing Solarbio Science& Technology, Beijing, China). Briefly, the heat-fixed smears on slides were flooded with 0.2% Victoria blue for 30 seconds and washed with tap water, smears were decolorized with 2% picric acid ethanol, and cells were counterstained with 0.004% fuchsin for 30 seconds and then washed with tap water. The positive staining was blue. The experiment was repeated three times, the area of gram-positive staining was calculated by Image J software, and 8 samples were tested.

Transmission Electron microscopy (TEM) was performed as previously described (22). First, tissue samples were fixed with 2.5% glutaraldehyde followed by 1% OsO4. Then, the samples underwent serial dehydration, soaking, embedding in epoxy resin, and sectioning into ultra-thin 60-nm sections. The sections were stained with a solution of uranyl acetate and lead citrate, and then a transmission electron microscope (HITACH 500, Hitachi, Ltd., Tokyo, Japan) was used at a voltage of 75 kV to observe the bacteria within the scar tissues. The experiment was repeated three times, the number of bacteria was calculated per view, and 5 samples were tested.

For fluorescence in situ hybridization, oligonucleotide probes (EUB338: 5′-GCTGCCTCCCGTAGGAGT-3′) conjugated with Alexa Flour 488 (Invitrogen, Carlsbad, CA, USA), targeting 16S rRNA, were used to label bacteria, and CD11b antibody conjugated with Alexa flour 594 (Invitrogen) was used to label the inflammatory cells. Briefly, sections were immediately fixed with RNase-free 4% formaldehyde and permeabilized, followed by dehydration with ethanol (50%, 75%, 100%). The sections were hybridized with EUB338 (5 ng/mL) overnight at 37°C, and the excess and non-specifically bound probe was washed with phosphate-buffered saline (PBS)×2. Then, sections were and incubated with rabbit anti-human CD11b primary antibody overnight at 4°C, washed with PBS×2, and incubated with Alexa Fluor 594-conjugated donkey anti-rabbit antibody at room temperature. Then the nuclei were stained with 4′,6-diamidino-2-phenylindole (DAPI; H-1200; Vector Laboratories, Burlingame, CA, USA). Images were captured using a con-focal laser scanning microscope (Carl Zeiss, Oberkochen, Germany). The experiment was repeated three times, the area of positive staining was calculated by Image J software, and 8 samples were tested.

VECTASTAIN Elite Avidin-Biotin Complex Kit (Maixin Biotech. Co, Fuzhou, China) was used for immunohistochemical staining of MCP-1, TNF-α, IL-6, and IL-8 according to the manufacturer’s instruction. The primary and secondary antibodies used are listed in Table 1 . Briefly, sections of the paraffin-embedded femurs were kept at 60°C for 24 h in the oven and then followed by deparaffinizing with xylene and hydrating with an ethanol gradient (100%–70%). After successively incubating with antigen retrieval solution (Shanghai Shunbai Biotechnology Company; Shanghai, China) and 3% H₂O₂ for 30 min, the slides were rinsed with water and incubated with the primary antibody MCP-1, IL-6, IL-8, and TNF-α overnight at 4°C(1:100 dilution; Santa Cruz Biotechnology Inc, Santa Cruz, CA). The next day, the slides were rinsed and incubated with the corresponding secondary antibody (1:100 dilution; Santa Cruz Biotechnology Inc)) for 30 min followed by 3,3′-diaminobenzidine (DAB) and hematoxylin staining, respectively. The slides were then examined and photographed using an Olympus BX53 fluorescence microscope (Tokyo, Japan). The experiment was repeated three times, the area of positive staining was calculated by Image J software, and 8 samples were tested.

As mentioned above, after the wash of the tissues, the epidermis of HS and NS were removed using a scalpel. Then the samples were stored at -80°C before genomic RNA extraction.

Tissue genomic DNA was extracted from 0.1 g frozen skin and scar samples using MP Fast DNA SPIN Kit for Soil according to the manufacturer’s protocol (23). Briefly, the DNA concentration and purification were measured, and the DNA quality was detected by electrophoresis. The 16S-rDNA gene were amplified using primers 341F: 5′-ACTCCTACGGGRSGCAGCAG-3′, and 806R: 5′-GGACTACVV GGGTATCTAATC-3′. PCR was performed. The reactions were performed on a thermocycler PCR system (GeneAmp 9700, ABI, USA). All PCR products were purified using an AxyPrep DNA Gel Extraction Kit (Axygen Biosciences, Union City, CA, USA) and quantified using QuantiFluor™-ST (Promega, USA). Purified and pooled amplicon libraries were paired-end sequenced (2×300) on the Illumina MiSeq platform (Illumina, San Diego, USA) according to the standard protocols by Majorbio Bio-Pharm Technology Co., Ltd. (Shanghai, China).

Raw sequence reads were demultiplexed, quality-filtered, merged, and clustered into OTUs with a 97% similarity cutoff using UPARSE (version 7.1, http://drive5.com/uparse/), and chimeric sequences were identified and removed using UCHIME. The taxonomy of the acquired OTUs was analyzed using the RDP Classifier Bayesian algorithm (http://rdp.cme.msu.edu/) against the SILVA database (version128) with a confidence threshold of 70%.

Data generated from H&E staining, Gram staining, Masson staining, TEM, in situ hybridization, and immunohistochemistry analyzed by Image J software, and the data was expressed as mean ± SD, and two groups were red by two-tailed Student’s t-test.

For the 16S sequencing data, all statistical analyses were performed using R packages (V·2·15·3) as follows.

Microbiota α-diversity, which represents microbial diversity within an individual group, was computed in QIIME through the whole tree phylogenetic diversity metric. And Shannon index was used.

Microbiota β-diversity, which indicates inter-variability of microbial diversity between groups, was examined through weighted UniFrac distances in QIIME and hierarchical clustering based on the unweighted pair group Method, and principal coordinates analysis (PCOA) was used.

To test the difference in microbiome composition between the two groups, the Kruskal-Wallis test was used, and phylum, genus, and species levels were selected for this analysis.

To evaluate the correlation between microbiota and clinical variables, the relationship was calculated through Spearman correlation.

For all statistical analyses, a 2-sided P <.05 was accepted as statistically significant.

In our study, 33 HS (male 22, female 11) and 36 NS (male 24, female 12) were involved in this study, and average ages were 19.67 ± 3.50 (HS) and 18.61 ± 3.27 (NS) respectively, mainly located at the neck, trunk, and upper and lower extremity. There was no significant difference between the 2 groups in sex (p=0.83), age (p=0.20), and sample location (p=0.92), but the indices of VSS score were significantly higher in HS than that in NS (Figure 1A; Table 1 ).

H&E staining showed almost no inflammatory cells in the NS with loose collagen arrangement, however, in HS tissue, many clusters of inflammatory cells with dense collagen fibers were detected ( Figure 1B ).

According to Gram staining, few bacteria colonized in the sub-epidermis of NS tissue, instead, a large number of gram-positive bacteria colonizing in HS tissue was observed. ( Figure 1C ).

Using TEM, we confirmed the presence of bacteria in HS tissues, whereas almost no signs of bacteria were detected in NS dermis. The individual bacteria in HS tissues appeared round or elliptical with wave-like membranes. Interestingly, it was found that a bacterium was swallowed by an inflammatory cell ( Figure 1D ).

In addition, 16S-rRNA and CD11b staining were used to label the bacteria and inflammation cells respectively. According to the result, the bacteria clusters could be seen in HS with inflammatory cells distributed around, both were increased significantly in HS than that in NS ( Figure 1E ).

Immunohistochemistry revealed that NS tissues showed extremely low expression of MCP-1, IL-6, IL-8, and TNF-α. In contrast, their expression level was significantly higher in HS( Supplement Figure ).

These results demonstrated that higher bacteria count and tissue inflammation were present in HS.

The 16S rRNA amplicon sequencing analysis was used to sequence the microbiome in 36 NS and 33 HS samples. In total, 11,222 OTUs, 1 domain, 1 Kingdom, 56 Phyla, 134 Classes, 370 Orders, 691 Families, 1,652 Genera, and 3,417 Species were obtained from all samples.

In order to compare the microbiota richness and evenness between HS and NS, the a-diversity was calculated using the Chao, Shannon, and Simpson index, however, no significant difference was found between the two groups ( Figure 2A ). (Chao index p=0·103, Shannon index p=0·28, and Simpson index p=0·176, respectively)

β-diversity reflects the intra-group variability of the microbiome composition between two groups. When the indices of PCoA were used, the results showed that there was a clear separation between HS and NS (R=0·4970, P=0·001, Figure 2B ), indicating a significantly different microbiota composition pattern between scar and NS.

The community abundance of each sample was counted at different taxonomic levels with R package, and the microbiome compositions of both groups were visualized by pie chart ( Figure 2C ). Then the microbiota differences between NS and HS were compared respectively at phylum, genus, and species levels.

At the phylum level, Actinobacteria, Firmicute, Proteobacteria, and Bacteroidetes were the major microbiota in HS and NS tissues, but the proportion varied. The Firmicute was significantly higher in HS than that in NS (53·2% & 5·52%, P=7·18×e^-11), whereas Actinobacteria in HS was significantly lower than NS (9·19% & 38·21%, P=1·735×e^-7). And the abundance of Proteobacteria and Bacteroidetes did not significantly differ between the two groups (data not shown in figure, P=0·065, P=0·59, respectively) ( Figure 3A ).

At the genus level, NS mainly harbored Cutibacterium (30·25%), Ralstonia (19·33%), Rhodococcus (4.96%), etc. Of note, HS tissues exhibited different proportions of the microbiome, named Staphylococcus (39·42%), Enterobacter (7·44%), Ralstonia (6·34%), Streptococcus (2·54%), etc. The abundances of Staphylococcus, Enterobacter, and Streptococcus in HS were significantly higher than that in the NS (39·42% & 1.22%, P=9·31×e-8; 7·44% & 0.24%, P=0·000006; 2·54% & 0.11%, P=0·0007 respectively). However, the prevalence of Cutibacterium and Ralstonia were significantly lower in HS tissues than that in NS (0.37% & 30·25%, P=1·14×e-7; 6.34% & 19·33%, P=0·0005 respectively) ( Figure 3B ).

At the species level, S. aureus had the highest prevalence in scar tissues, which was significantly greater than that in NS tissues (39·71% & 1·88%; P=1·13×e^-7). Secondly, unidentified Enterobacter and unclassified Streptococcus species were also significantly more prevalent in scar tissues than that in NS (7·44% & 0.22%, P=8·3×e^-5; 2·54% & 0.09%, P=3·42×e^-4, respectively), but the proportion was not high (data not shown in figure). However, Cutibacterium acnes, which was dominant in NS tissue, were significantly reduced in HS (28·89% & 0·33%, P=4·41×e^-8) ( Figure 3C ).

Normally, 16s sequencing could not identify species, but in our study, S.aureus and Cutibacterium acnes were identified. The above results indicate that microbiome dysbiosis occurred in HS, which was dominated by S.aureus.

In order to assess the correlation between the bacteria and the scar, the VSS indices including pigment, thickness, pliability, itching/pain, and vessel were used to evaluate HS clinical severity. For each HS patient, we calculated the Spearman correlation between the VSS scores and individual bacterial abundance, represented by a heatmap plot. Red indicates a positive correlation between bacterial content and clinical severity, and green represents a negative correlation. The results revealed that genus Staphylococcus, Enterobacter, and Streptococcus were positively correlated with clinical indices ( Figure 4 ). However, genus Cutibacterium, Ralstonia, etc. were negatively correlated with scar indices. For instance, g_Staphylococcus had a strong correlation with chroma (r=0.6230, p<0.001), vessel (r=0.5513, p<0.001), thickness (r=0.5743, p<0.001), pliability (r=0.5965, p<0.001), itching and pain (r=0.5940, p<0.001) ( Figure 4 ).

The spearman heatmap showed genus Staphylococcus, Enterobacter, and Streptococcus were positively correlated with these indices, and genus Cutibacterium, Ralstonia etc. were negatively correlated with scar indices. (*P<0.01, **P<0.001, ***P<0.0001).