LncRNA expression profiling and functional screening in human skin wound healing
To probe gene expression during human skin wound healing, we created full-thickness excisional wounds on the skin of healthy volunteers (n = 26) and then collected wound-edge tissues from the same donors at each healing stage, i.e., inflammation (day 1 wounds, NW1), proliferation (day 7 wounds, NW7), and remodeling (day 30 wounds, NW30) (Fig. 1A and Table S1). We conducted whole-transcriptome analysis using polyadenylation-independent total RNA sequencing (RNA-seq) on matched skin, NW1, and NW7 samples from five donors, identifying 342 lncRNAs with significantly altered expression during wound repair (one-way ANOVA, p ≤ 0.005) (Fig. 1A and Table S2). We focused on 20 lncRNAs conserved in humans and rodents15, 16, as evolutionary conservation suggests crucial functional roles17. After transfection of Lincode siRNAs targeting each of these lncRNAs, we assessed their impact on the inflammatory response, proliferation, and migration of human keratinocyte progenitors (Fig. 1B). Silencing SNHG26, RMRP, MALAT1, CRNDE, or SNHG6 increased TNFa-induced production of the inflammatory cytokines IL6 and CCL20. Moreover, downregulating SNHG26, SNHG6, or LOC101929709 decreased cell proliferation, while silencing SNHG26, WDFY3-AS2, CRNDE, SOX9-AS1, or GAS5 reduced keratinocyte migration. Notably, among these 20 conserved lncRNAs, knockdown of SNHG26 exhibited the most pronounced proinflammatory, antiproliferative, and antimigratory effects on keratinocytes.
Characterization of SNHG26 expression in skin wound healing
SNHG26 is an intergenic lncRNA encoded on human chromosome 7 (GRCh37/hg19, chr7: 22,893,745 − 22,901,055) and mouse chromosome 5 (GRCm38/mm10, chr5:23,850,597 − 23,855,038, annotated as 2700038G22Rik) (Fig. 1C, Figure S1A and B). Both the human and mouse SNHG26 genes are located between the TOMM7 and FAM126A coding genes, suggesting interspecies syntenic conservation (Fig. 1C, Figure S1A and B).
In both humans and mice, SNHG26 expression temporarily increased during the inflammatory and proliferative phases of wound healing, returning to baseline levels during remodeling phase (Fig. 1D, E). To identify the primary cell type(s) expressing SNHG26 in human skin and wounds, we isolated epidermal CD45 − cells (primarily composed of keratinocytes) and CD45 + cells (leukocytes), dermal CD90 + cells (fibroblasts), CD14 + cells (macrophages), and CD3 + cells (T cells) from matched skin and day 7 wounds of five healthy donors (Table S1). SNHG26 was found in various skin cell types, but keratinocytes showed the most significant upregulation during wound healing (Fig. 1F). This observation was further validated by spatial transcriptomics (Fig. 1G, Figure S1C, D) and fluorescence in situ hybridization analysis (FISH, Fig. 1H) of human acute wounds, indicating that the injury-induced transient upregulation of SNHG26 expression predominantly occurred in wound-edge basal keratinocytes. Notably, the SNHG26 gene locus also contains a small nucleolar RNA SNORD93 in both human and mice (Fig. 1C, Figure S1A, B), but only SNHG26 showed increased expression in keratinocytes during wound repair (Figure S1E, F). Moreover, SNHG26 expression was reduced in human diabetic foot ulcers and diabetic mouse wounds compared to acute wounds (Figure S1G-I), highlighting its relevance in wound healing and chronic wound pathology.
We further characterized the molecular features of SNHG26 in keratinocyte progenitors. The Cap Analysis of Gene Expression (CAGE) project18 previously located the transcription start site (TSS) of SNHG26 at GRCh37/hg19 chr7:22893816–22893881 (Fig. 1C). Our 3′ rapid amplification of complementary DNA ends (RACE) analysis indicated that the SNHG26 sequence matched the transcript ENST00000415611.9 from Ensembl, with a length of 3100 nt (Fig. 1C). This finding was further validated through Northern blot analysis (Fig. 1I). Furthermore, FISH and nucleus-cytoplasm fractionation assay revealed the nuclear localization of SNHG26 in both cultured keratinocyte progenitors and basal keratinocytes in human skin (Fig. 1H, J, K). Moreover, by separating the poly(A)+ fraction from the poly(A)− RNA fraction, we found SNHG26 to be a polyadenylated RNA (Figure S1J). With the Coding Potential Calculator 2 algorithm19, we confirmed the non-coding potential of SNHG26 (Figure S1K). Additionally, we determined that the half-life of SNHG26 in keratinocyte progenitors was 2.2 hours (Figure S1L). In summary, SNHG26 is a conserved, polyadenylated, nuclear lncRNA transiently upregulated in wound-edge basal keratinocytes.
Impaired wound healing in Snhg26-deficient mice
Considering the potential role of SNHG26 in regulating keratinocyte inflammation, proliferation, and migration (Fig. 1B), we proceeded to investigate whether SNHG26 upregulation is indispensable for the wound healing process. To examine this, we generated Snhg26-knockout (KO) mice by deleting the 8 kb genomic locus (Figure S2A and B). These KO mice exhibited a generally normal phenotype with a slightly thinner epidermis (Figure S2C). However, upon skin injury, we observed a significant 40% delay (p = 0.0001) in wound healing in the Snhg26-KO mice compared to control mice (Fig. 2A). Recognizing that compensatory mechanisms might be at play in this model with constitutive Snhg26 deletion, we also intradermally injected Snhg26 antisense oligos (Snhg26-ASOs) at wound edges in wild-type (WT) mice (Figure S2D-F). These ASOs were designed to block the injury-induced upregulation of Snhg26 by facilitating RNaseH1-dependent cleavage of Snhg26 transcripts20. Notably, these ASOs effectively reduced SNHG26 expression in the epidermal compartment, while the dermal compartment remained unchanged (Figure S2F). Similar to the Snhg26-KO mice, mice treated with Snhg26-ASOs exhibited a 37.5% delay (p = 0.003) in wound healing compared to those receiving scrambled ASOs, emphasizing the crucial role of Snhg26 transcripts in wound repair (Fig. 2B).
To uncover the gene expression changes linked to reduced healing in Snhg26-KO mice, we compared epidermal gene expression between KO (n = 3) and WT mice (n = 3). Microarray analysis identified 253 and 431 differentially expressed genes (DEGs; fold-change ≥ 2 or ≤ -2, p value < 0.05) in the Snhg26-KO mice skin and day 3 wound edge, respectively (Figure S2G). Gene Ontology (GO) analysis revealed that the downregulated genes in Snhg26-KO skin epidermis were mainly associated with epidermal differentiation, while no enriched GO term was found among the upregulated DEGs (Figure S2H). In the wound-edge epidermis, Snhg26-KO mice exhibited upregulated inflammation-related genes and downregulated genes linked to skin development and cell migration (Fig. 2C). Gene set enrichment analysis (GSEA) showed that upregulated genes were associated with leukocyte migration, while downregulated genes were linked to epithelial cell proliferation and migration in the wound-edge epidermis of Snhg26-KO mice (Fig. 2D)
We also noted reduced expression of Krt16 and Sprr1b in the wound-edge epidermis of Snhg26-KO mice compared to controls, as confirmed by qRT‒PCR analysis of additional skin and wound epidermal samples (Fig. 2E). Both Krt16 and Sprr1b are known for their roles in keratinocyte proliferation and migration21–23. The decreased expression of these genes could potentially contribute to delayed re-epithelialization, as supported by histomorphometry analysis of newly formed epithelial tongues (Fig. 2F). Furthermore, we observed significant upregulation of proinflammatory cytokines, including Il6 and Il1b, in the wound-edge epidermis of Snhg26-KO mice (Fig. 2G) and increased macrophages (CD68+) in the dermal wound bed of Snhg26-KO mice (Fig. 2H). Similarly, mice treated with Snhg26-ASOs exhibited elevated levels of inflammatory cytokines (Il6, Ccl20, Ccl2, and Ccl5) in the wounds (Figure S2I-L).
In summary, our study, utilizing Snhg26-KO mice and mice with transient Snhg26 inhibition at wound edges, provides conclusive evidence of Snhg26's in vivo role in wound repair. Its deficiency not only delayed re-epithelialization but also exacerbated inflammatory response.
Single-cell transcriptome analysis of inhibited wound repair in Snhg26-KO mice
To gain deeper insights into the inhibition of wound repair in Snhg26-KO mice, we conducted single-cell RNA sequencing (scRNA-seq) analysis on skin and day 3-wound tissues from Snhg26-KO mice (n = 3) and control littermates (n = 3). Unsupervised clustering identified 23 cell clusters, including keratinocytes (basal stem and progenitors: C1-C4; spinous: C5; granular: C6; hair follicle keratinocytes: C7-C10), melanocytes (C11), fibroblasts (C12-C14), sebaceous gland cells (C15), vascular endothelial cells (C16-C17), muscle cells (C18-C19), macrophages (C20), T helper cells (C21), γδ T cells (C22), and Langerhans cells (C23) (Fig. 3A, Figure S3A). Each of these cell clusters contained cells from all the investigated samples (Figure S3B). Notably, our analysis revealed that Snhg26 was predominantly expressed in keratinocytes (Fig. 3B, C). While the cellular composition remained similar between the KO and WT mouse skin samples, notable differences were observed in the basal keratinocytes and immune cells in the wounds of Snhg26-KO mice compared to WT mice (Fig. 3D). We also calculated the number of DEGs in each cell cluster between Snhg26-KO and WT mice and found that keratinocytes showed the highest number of DEGs between the two groups (Fig. 3E), suggesting the critical role of SNHG26 in keratinocytes during wound healing.
Consistent with a recent study on mouse wounds using scRNA-seq24, our results identified four basal stem and progenitor cell clusters: C1 expressed Col17a1, a marker enriched in epidermal stem cells; C2 consisted of proliferative basal cells expressing mitosis-related genes, such as Pcna and Mki67; C3 showed enrichment in glycolysis-related genes, important for keratinocyte migration25, and specifically expressed the keratinocyte activation markers Krt6b and Krt1626; and C4 expressed genes promoting cell cycle arrest (e.g., Ovol1) and inflammation (e.g., the TFs Fosl1 and Rel)27, 28 (Fig. 3F, Figure S3C-D, Table S3). Functional characteristics of these keratinocyte progenitor clusters, including the proliferative C2, migratory C3, and inflammatory C4 progenitors, were more evident when we compared overall gene expression scores for these processes (Fig. 3G, Table S4). Analyzing the functional scores of individual keratinocytes, we observed fewer proliferative or migratory keratinocytes but a higher number of inflammatory keratinocytes in the wounds of Snhg26-KO mice compared to WT mice (Fig. 3H, I). This observation was supported by GO analysis of the DEGs in keratinocyte progenitors between the Snhg26-KO and WT mice (Figure S4A-C). Furthermore, a cell‒cell crosstalk analysis29 highlighted enhanced CCL and CXCL chemokine signaling from basal keratinocytes to macrophages and T cells in Snhg26-KO wounds (Fig. 3J, Figure S4D). These findings align with the more numbers of macrophages and Th cells detected in these wounds by scRNA-seq (Fig. 3D) and IF staining (Fig. 2H).
Single-cell analysis allowed us to precisely discern the impact of Snhg26-KO in various skin cell types. In dermal fibroblasts, which play critical roles in skin wound healing, especially in modulating the inflammatory response30, we observed 198 DEGs (Fig. 3E). The downregulated genes in the Shng26-KO fibroblasts were associated with the TGF-β regulation of extracellular matrix, while the upregulated genes were related to TNF-α signaling via NF-kB pathway (Figure S4E). Cell-cell crosstalk analysis further confirmed the role of fibroblasts in wound inflammation. Specifically, fibroblasts (Fb1 and Fb2) were identified as the primary cell types sending CCL and CXCL signals in mouse wounds (Fig. 3J and Figure S4D). In Snhg26-KO wounds, fibroblasts exhibited heightened CCL signals directed towards keratinocytes (Fig. 3J). Additionally, there was a decreased laminin signal from keratinocytes to fibroblasts in the Snhg26-KO mice (Figure S4F).
In summary, Snhg26 disruption skewed keratinocyte progenitors towards an inflammatory state and away from proliferation or migration. While Snhg26 had a lesser impact on dermal fibroblasts, the changes in keratinocyte-fibroblast communication collectively contributed to delayed wound healing.
SNHG26 inhibits the human keratinocyte inflammatory response and promotes re-epithelialization
Next, we explored the physiological relevance of SNHG26 in human keratinocytes and skin wound healing. We silenced SNHG26 expression in human keratinocyte progenitors by transfecting SNHG26-ASOs followed by TNFa treatment to trigger an inflammatory response31. This ASO was designed to target the exons of SNHG26, and it led to a significant reduction in SNHG26 expression while leaving SNORD93 unaffected (Figure S5A, B). Microarray analysis of these cells revealed gene expression changes (ANOVA, P < 0.001) grouped into five patterns (Fig. 4A). Module 2 (M2) contained 898 upregulated genes related to inflammation. M3 comprised 123 TNFα-induced genes that were further upregulated by SNHG26 knockdown. M4 and M5 included 1157 and 918 downregulated genes related to mitosis.
We confirmed expression changes in genes associated with inflammation, including IL6, IL8, and CCL20, which are induced in response to skin injury and play crucial roles in the inflammatory process32, 33. Silencing SNHG26 increased their expression under both basal and TNF-α-triggered inflammatory conditions (Fig. 4B). Conversely, nuclear overexpression of SNHG26 using a pZW1-snoVector reduced TNF-α-induced IL6, IL8, and CCL20 expression (Fig. 4C, Figure S5C)34. However, cytoplasmic overexpression of SNHG26 using a pcDNA3.1 vector had no discernible effect on these inflammatory genes (Figure S5D-G), highlighting the importance of subcellular localization in SNHG26's function.
Furthermore, we substantiated our findings by confirming that the silencing of SNHG26 reduced, whereas overexpression of SNHG26 increased the migration and proliferation of human keratinocyte progenitors. This was evident in scratch wound assays (Fig. 4D, F), cell proliferation assays (Fig. 4E), and colony formation assays (Fig. 4G, Figure S5H). Next, we extended our analysis to human ex vivo wound closure, which is an in vivo-like and clinically relevant model for analyzing human wound re-epithelization35, 36. We topically applied SNHG26-ASOs (Fig. 4H, I, Figure S5I, J) or its overexpression plasmid sno-SNHG26 (Fig. 4J, K) mixed in a lipid-based transfection reagent to the wounds. This intervention specifically modulated epidermal SNHG26 levels, leaving dermal levels unaffected (Figure S5K). We found that SNHG26-ASOs reduced wound re-epithelialization, while overexpression of SNHG26 enhanced this process.
In addition, we conducted experiments to distinguish the roles of SNHG26 and SNORD93, both sharing the same genomic locus (Fig. 1C). Using CRISPR-Cas9 technology, we generated a SNHG26 knockout in hTERT-immortalized human keratinocyte cell line (Ker-CT) (Figure S5L-M). Subsequently, we reinstated the expression of either SNORD93 or SNHG26 through the transfection of their respective overexpressing plasmids (Figure S5N-O). Notably, SNORD93 had no impact on inflammatory responses or migration, whereas SNHG26 reintroduction exhibited anti-inflammatory and pro-migratory effects, confirming the specific role of SNHG26 (Figure S5P-R).
Collectively, our mouse and human data jointly demonstrated the evolutionary conservation of SNHG26 in both expression and function: it is an injury-induced lncRNA in keratinocyte progenitors and plays pivotal roles in inhibiting cell inflammatory response while promoting cell proliferation and migration and wound re-epithelialization.
SNHG26 hijacks the transcription factor ILF2 from inflammatory genomic loci
To elucidate SNHG26's molecular mechanism, we performed RNA pull-down assay in human keratinocyte progenitors (Fig. 5A). An SDS‒PAGE analysis revealed a protein band at ~ 45 kDa specifically copurified with SNHG26 (Fig. 5B). A mass spectrometry (MS) analysis identified this band represented interleukin enhance-binding factor 2 (ILF2) protein, which was confirmed by Western blotting (Fig. 5C). The reciprocal binding of SNHG26 and ILF2 protein was validated via RNA immunoprecipitation (RIP) using the anti-ILF2 antibody, which showed that SNHG26, but not GAPDH or 18S rRNA, coprecipitated with ILF2 protein in human keratinocyte progenitors (Fig. 5D, E).
Using established methodologies37–39, our quantitative estimations revealed that each human keratinocyte progenitor cell contains approximately 1.5-2 copies of SNHG26 transcripts and about 104,125 ILF2 protein molecules (Figure S6 A-C). During wound healing, SNHG26 expression increases 27.5-fold (Fig. 1F), resulting in an estimated 41–55 copies of SNHG26 RNA per wound keratinocyte cell. Notably, the stoichiometry of SNHG26 RNA to ILF2 protein remains skewed, ranging from 1:1893 to 1:2540. In line with these findings, co-staining analysis showed SNHG26 RNA and ILF2 protein co-localized within a condensate-like structure in the cell nucleus, but the abundance of ILF2 proteins greatly exceeded that of SNHG26 RNA (Figure S5D, E).
ILF2 is a constitutively expressed nuclear protein that interacts with chromatin40 and is a TF for both mitotic and inflammatory genes41. It is highly expressed in the human epidermis, as shown in the Human Protein Atlas database42 (Figure S6F). ILF2 expression was upregulated in the inflammatory phase (NW1) of human skin wound healing (Fig. 5F). To study the effects of SNHG26 knockdown on ILF2 genomic occupancy, we performed ILF2 chromatin immunoprecipitation (ChIP)-sequencing in human keratinocyte progenitors. We identified 670 peaks (p value < 0.05) enriched in keratinocytes treated with Ctrl-ASO and 593 peaks in cells with SNHG26-ASO (Fig. 5G, Table S5). Focusing on ILF2 binding sites near gene promoters, we found that genes with more ILF2 binding in the Ctrl-ASO group were associated with cell growth pathways (Figure S6G, Table S5), while those with more ILF2 binding in the SNHG26-ASO group were linked to the MAPK signaling pathway, which mediates the inflammatory response in keratinocytes43 (Figure S6H, Table S5). Overall, the ILF2 ChIP-seq analysis underscores the impact of SNHG26 perturbation on ILF2 binding site dynamics, shifting the focus from cell growth to inflammation-related processes within keratinocytes.
Particularly noteworthy is the prominent ILF2 binding locus at the JUN promoter, located within a 1-kilobase proximity to the TSS, which ranked as one of the top binding sites in cells with SNHG26 knockdown (Fig. 5G, Figure S6I, Table S5). This result was further corroborated through ILF2 ChIP-qPCR analysis (Fig. 5H). Additionally, a ChIP‒qPCR analysis revealed that TNFa treatment triggered ILF2 binding to the JUN promoter within 30 minutes (Fig. 5I). However, knockdown of ILF2 completely abolished TNFα-induced JUN expression, emphasizing the crucial role played by ILF2 in driving JUN transcription (Fig. 5J, Figure S6J).
Jun/AP1 is an immediate-early gene and a master TF that controls the expression of a wide range of inflammatory genes in epidermal cells44. Upon examining the scRNA-seq data of mouse wounds, we found that not only was Jun expression increased (Fig. 5K), but the genes it regulated were also enriched among the upregulated genes in the wound-edge basal keratinocytes of the Snhg26-KO mice (Fig. 5L). Moreover, Jun emerged as a central hub connected to multiple genes involved inflammatory signaling, e.g., Rel, Nfkbia, Mapk9, and Map3k845–47 in wound keratinocytes of Snhg26-KO mice (Figure S6K). Additionally, in the mouse wound-edge epidermis (Fig. 2C and D), c-Jun was identified as the Snhg26-regulated TF that controls the most genes with aberrant expression after Snhg26-KO (Fig. 5M). It's worth noting that, unlike keratinocytes, JUN expression remained unaltered both in human fibroblasts transfected with SNHG26-ASO (Figure S6L) and in Snhg26-KO mice wounds (Figure S6M), indicating that SNHG26 likely exerts distinct functional mechanisms in fibroblasts compared to keratinocytes. Additionally, TNF-α also induced ILF2 binding to the promoters of IL6, IL8, and CCL20 (Fig. 5N). As a result, silencing ILF2 led to a reduction in their expression in keratinocytes (Fig. 5O). Intriguingly, silencing SNHG26 significantly increased the binding of ILF2 to the promoters of IL6, IL8, and CCL20 (Fig. 5P).
Thus, we concluded that SNHG26 interacts with ILF2 and prevents it from binding to inflammatory genomic loci, including the master TF JUN and several key cytokines and chemokines, consequently suppressing the inflammatory response of keratinocyte progenitors.
SNHG26 directs ILF2 protein to the LAMB3 genomic locus
As SNHG26 hijacks ILF2 from the promoters of inflammatory genes, we asked whether SNHG26 may interact with chromatin and translocate ILF2 to another genomic locus. To assess this possibility, we map SNHG26 occupancy genome wide by Chromatin Isolation by RNA Purification sequencing (ChIRP-seq) in human keratinocyte progenitors48. This method allowed us to isolate approximately 22% of SNHG26 RNA in human keratinocyte progenitors (Fig. 6A, B). Importantly, the SNHG26 probes did not retrieve GAPDH, nor did the LacZ probes retrieve SNHG26, confirming the method's specificity (Fig. 6B). Through ChIRP-seq, we identified 43 SNHG26 occupancy sites across the genome (FDR < 0.05) (Fig. 6C, Table S6). These sites were notably enriched within genic regions, particularly in regions annotated as promoters and introns (Fig. 6D). Notably, one of the most prominent SNHG26 RNA occupancy sites was located within its own genic region, further confirming the specificity of this ChIRP experiment (Fig. 6C, Figure S6N). Further analysis of the ChIRP-seq data using Multiple EM for Motif Elicitation (MEME)49 revealed that SNHG26 exhibited a preference for binding to a GA-rich polypurine DNA motif (Fig. 6E), indicating that SNHG26 accesses the genome through specific DNA sequences.
We focused on the LAMB3 gene due to its high-ranking among the top SNHG26 ChIRP-seq peaks (Fig. 6C, F). Silencing SNHG26 reduced ILF2 binding to the LAMB3 gene, indicating that SNHG26 directs ILF2 to the LAMB3 locus by interacting with chromatin (Fig. 6G). Additionally, we demonstrated that ILF2 was crucial for driving LAMB3 expression, as silencing ILF2 significantly decreased the LAMB3 mRNA level in keratinocytes (Fig. 6H). LAMB3 encodes subunit beta 3 of Laminin 332, which is required for keratinocyte attachment to the basement membrane in the epidermis and crucial for keratinocyte proliferation and migration50. Consistent with previous research50, silencing LAMB3 significantly impaired the migration, proliferation, and long-term growth of human keratinocyte progenitors (Fig. 6I-M, Figure S6O). Furthermore, LAMB3 expression increased at the proliferative phase of human skin wound healing (NW7), supporting its role in wound re-epithelialization (Fig. 6N). Our scRNA-seq analysis of mouse wounds indicated that Lamb3 was primarily expressed in basal keratinocytes and upregulated in wounds compared to the skin (Fig. 6O, P). However, Snhg26-KO mice exhibited significantly lower Lamb3 expression in wound-edge basal keratinocytes compared to WT mice (Fig. 6P, Q). This finding aligns with the delayed re-epithelialization observed in Snhg26-KO mice (Fig. 2E, F).
In light of this multifaceted evidence, we propose a novel mechanism for the nuclear lncRNA SNHG26: it interacts with ILF2 and redirects it from inflammatory genomic loci, such as JUN, IL6, IL8, and CCL20, to the genomic locus of LAMB3, reshaping the gene expression program to facilitate the transition of keratinocyte progenitors from an inflammatory to a proliferative state.