lncRNA-H19 in Fibroblasts Promotes Wound Healing in Diabetes

Cutaneous wound healing is a dynamic and complex process, in which the intricate cross talk among plenty of highly organized mediators is required to heal injured skin toward healthy barrier (1). However, the well-organized sequence of events in this process may undergo disorder at various steps along the pathway, underlying the states of numerous diseases states, which causes chronic wound healing and results in significantly heavy burden to both the patient and medical system globally. For instance, diabetic foot ulcers (DFU) in patients with diabetes are frequently accompanied by a high risk of infection and even amputation, which raise the ratio of 5-year mortality up to 80% (1,2). Generally, the process of wound healing consists of four main continuous phases over time, including hemostasis, inflammation, and proliferative and maturity phases (3). During these phases, numerous cell types (i.e., immune cells, fibroblasts, keratocytes, etc.) are involved and exert the intricate interplay via directly interacting with or releasing a number of cytokines, chemokines, and growth factors (4).

As one of the main cell types in the dermis, dermal fibroblasts have been thought to be responsible for the synthesis and remodeling of extracellular matrix (i.e., collagen). However, a growing body of studies reveal the crucial roles for fibroblasts in regulating ongoing cell proliferation and inflammation by producing a variety of signaling molecules that modulate fibroblasts themselves, as well as immune cells, keratinocytes, endothelial cells, etc., via direct interaction or autocrine/paracrine communications (5). Most fibroblasts present in chronic wounds exhibit abnormal morphologies and impaired migratory/proliferative capacities, which slow down the progression of wound healing (6,7). In line with impaired proliferative capacity, fibroblasts in chronic wounds exhibit less responsiveness to growth factors, caused by reduced receptors for the growth factors and downstream components of the signaling cascades in fibroblasts (8). However, the underlying mechanisms for impaired fibroblasts proliferation in DFU still remain unknown.

Long noncoding RNAs (lncRNAs) are a class of noncoding transcripts with >200 nucleotides, which can regulate gene expression at various levels such as chromatin modification and the processing of transcription and posttranscription (9,10). While lncRNAs have primarily been implicated in various contexts of genomic imprinting, developmental processes, and various diseases, the critical roles for lncRNAs have been uncovered in the development of many chronic diseases, including diabetes and its complications (i.e., DFU) (11). lncRNA-H19 (hereafter called lncH19 or H19), a multifunctional lncRNA in both nucleus and cytoplasm (12), has been implicated in controlling hepatic gluconeogenesis in diabetic hyperglycemia (13) and also in regulating osteogenic differentiation of mesenchymal stem cells (MSCs) during the regeneration process of tooth tissue (14). Recently, lncH19 derived from bone marrow mesenchymal cells was reported to exert beneficial effects on wound healing in DFU of a type 1 diabetic animal model (15), suggesting its crucial role in regulating diabetic chronic wound healing.

The endogenous cell source and the molecular mechanism(s) for lncH19 underlying enhanced diabetic wound healing remain largely unknown, which dampens the application of lncH19 in DFU clinical treatment. Here, we uncovered that less lncH19 of PDGFRα⁺ cells in injured skin tissues causes impairment of wound healing in patients with diabetes or animals with diet-induced type 2 diabetes (T2D). Replenishing lncH19 to injured sites via exosomes derived from adipocyte progenitor cells (APCs) gives rise to the proliferative capability of dermal fibroblasts and inflammatory macrophage recruitment and subsequently restores the wound healing in T2D animals.

The injured tissues from patients without diabetes (n = 10: 4 female and 6 male) and patients with type 2 diabetes (n = 11: 3 female and 8 male) were collected at the Shanghai Eighth People’s Hospital. The skin tissues of the individuals were collected at 3 or 4 days postinjury. Samples of a normal control group were collected from the five healthy individuals (three female and two male) at the medical examination center of the hospital. None of the individuals had any serious lower-extremity arterial diseases, cardiocerebrovascular diseases, or other vascular-associated complications at the time of tissue collection. This study was approved by the Institutional Review Board of the Shanghai Eighth People’s Hospital in accordance with the Declaration of Helsinki. Informed, written consent for this study was obtained from each participant.

C57BL/6J male wild-type mice were obtained from Shanghai Lab Animals Center (SLAC). The Pdgfrα-CreERT2 mouse was obtained from The Jackson Laboratory (stock no. 032770). The lncH19 whole-body knockout (KO) mouse (cat. no. NM-KO-18045) was obtained from Shanghai Model Organisms Co. Ltd. H19 floxed mice (H19^loxp/loxp) were generated with the CRISPR/Cas9 system in which exons 1–5 of murine lncH19 gene were flanked with two loxp sequences. All of the mice were maintained on a 12-h dark/12-h light cycle and housed in groups of three to five mice with free access to food and water. All mice in the experiments of this study were male.

For minimizing the discomforts of animals, the mice were well anesthetized with 2% isoflurane prior to tissue harvest or ketamine/xylazine cocktail before surgery. For developing inducible Pdgfrα-specific H19 KO mice, one Pdgfrα-CreER^T2 allele was bred into the mice carrying two homozygous alleles of H19^loxp/loxp to generate Pα-H19KO mice (Pdgfrα-CreER^T2; H19^loxp/loxp). For inducing H19 gene depletion in PDGFRα⁺ cells, tamoxifen (no. T5648; Sigma-Aldrich) was prepared as a 20 mg/mL stock in corn oil and 7-week-old male mice were injected with tamoxifen at the dosage of 100 mg/kg body wt for 5 consecutive days and rested for 1 week before further experiments.

For inducing T2D, 8-week-old male mice were maintained on a high-fat diet (60% of calories from lipids) (no. D12492i; Research Diets) for 12 weeks. For glucose tolerance test, mice were subjected to a 16-h fast prior to the one-time injection of glucose (no. G7021; Sigma-Aldrich) at a dosage of 1 g/kg body wt i.p. For insulin tolerance test, mice were administered insulin (no. 0002751001; Eli Lilly) at a dosage of 0.75 units/kg body wt i.p. right after a 6 h fast. Blood glucose concentrations were measured with Bayer CONTOUR glucometers. Serum levels of insulin were measured in the serum obtained from mice fasted for 4 h with a mouse insulin ELISA kit (no. 90080; Crystal Chem) according to the manufacturer’s instructions.

For wound healing experiments, the dorsal skin at the midline of the mouse shoulder was excised for a full-thickness wound extending through the panniculus carnosus with a 6 mm biopsy punch (Miltex). Then, the mice were caged individually, and each injured site was covered with a sterile circular silicone splint with Krazy Glue and then fixed with 5-0 polyglycolic acid sutures. The wounds were applied with a cocktail of antibiotics (including bacitracin, polymyxin B, neomycin) and dressed with 3M Tegaderm Transparent Film Dressing daily during the first 3 days. The sizes of the wounds were measured at the time indicated.

For isolation of dermal PDGFRα⁺ cells, mice were anesthetized and shaved to depilate the dorsal skin. Then, the skin was dissected and transferred immediately into a petri dish for mincing into small pieces (∼1 mm³ in size). The minced tissues were then transferred into a 50-mL tube with previously added 10 mL digestion buffer (Hanks’ balanced salt solution with 1.8 units/mL Liberase TL (no. 05401020001; Roche), 1.5% BSA (no. A7906; Sigma-Aldrich), and 100 IU/mL penicillin-streptomycin (no. 15140; Gibco). For digestion, the tissues were incubated in a 37°C water bath and kept shaking for 1 h. APCs were sorted from adipose tissue as previously described (16). In brief, fat depots were dissected and thoroughly minced prior to incubation with digestion buffer (Hanks’ balanced salt solution supplemented with 1.5% BSA and 1 mg/mL Collagenase D [no. 11088882001; Roche]) in a 37°C shaking water bath for 1.0–1.5 h. Then, the digestion was gently pipetted and filtered through a 100-µm cell strainer into a new 50-mL tube with previously added Dulbecco’s PBS (DPBS) with 2% FBS. Cells were spun down at 600g for 10 min. After removal of red blood cells with use of 1X RBC Lysis Buffer (no. 00-4333-57; eBioscience), the cells were resuspended and filtered through a 40 µm cell strainer before 600g spin for 5 min. The obtained cells were resuspended in medium for further culture or other assays.

For sorting PDGFRα⁺ cells, the obtained cell mixture was resuspended in DPBS with 2% FBS and incubated with anti-mouse CD16/CD32 Fc Block (1:200) for 5 min. Then, primary conjugated antibodies were added to the cells in blocking buffer for another 15 min incubation at 4°C. After one wash, the cells were resuspended in 2% FBS/DPBS for sorting with a BD Biosciences FACSAria cytometer. The primary antibodies worked in the following concentrations: PerCP/Cyanine5.5 anti-CD45, 1:300 (clone 30-F11, no. 103132; BioLegend); PerCP/Cyanine5.5 anti-CD31, 1:300 (clone 390, no. 102420; BioLegend); PDGFRα-FITC, 1:200 (clone APA5, no. 135905; BioLegend); and SCA1-APC, 1:200 (clone D7, no. 108112; BioLegend).

For evaluating fibroblast proliferation in vivo, stromal cell suspensions were prepared from the skin tissues as described above. After incubation with Fc Block and primary flow antibodies for staining of PDGFRα⁺ fibroblasts, the cell pellets were added to cold 70% ethanol drop by drop before being incubated 1 h at −20°C. After washing with Cell Staining Buffer (no. 420201; BioLegend), the cell suspension was incubated with anti–Ki-67–FITC antibodies, 1:100 (no. 556026; BD Biosciences), at 4°C for 1 h before flow cytometry analysis with a BD Biosciences LSRFortessa.

For the analysis of macrophages in cutaneous tissue, the obtained cell mixture was incubated with primary antibodies for 1 h at 4°C. After one wash with 2% FBS/PBS, cells were fixed by incubation with BD Cytofix (no. 554655; BD Biosciences) for 20 min before flow cytometry analysis with LSRFortessa. The primary antibodies and the working concentrations are as follows: PerCP/Cyanine5.5 anti-CD45, 1:300 (clone 30-F11, no. 103132; BioLegend); Pacific Blue anti-CD11b, 1:200 (clone M1/70, no. 101224; BioLegend); phycoerythrin anti-F4/80, 1:200 (clone BM8, no. 123110; BioLegend); APC anti–Ly-6C, 1:400 (clone HK1.4, no. 128015; BioLegend); and Alexa Fluor 647 anti-CD192 (CCR2), 1:200 (clone SA203G11, no. 150603; BioLegend). All the flow cytometric results were analyzed with FlowJo, version 10.

Freshly isolated subcutaneous white adipose tissue (scWAT) PDGFRα⁺SCA1⁺ APCs were cultured in the growth media with DMEM/F12 supplemented with 10% FBS, penicillin-streptomycin, and gentamicin. For in vitro differentiation, APCs were cultured in 10% CO₂ at 37°C until confluence. Confluent cultures were treated with adipogenic induction media (growth media supplemented with 5 μg/mL insulin, 1 μmol/L dexamethasone, and 0.5 mmol/L 3-isobutyl-1-methylxanthine) for 48 h. Subsequently, cells were maintained in growth media supplemented with 5 μg ⋅ mL⁻¹ insulin (maintenance media) for 4 days.

Murine primary dermal fibroblasts were prepared and cultured as previously described (17). In brief, the dermal tissue of newborn mice was dissected and digested in the solution of 0.1% Collagenase D (no. 11088858001; Roche) for isolating individual fibroblasts. Then, the cells were cultured in DMEM (4.5 mmol/L glucose, no. D0422; Sigma-Aldrich) supplemented with 2 mmol/L l-glutamine, 100 IU/mL penicillin-streptomycin, and 10% heat-inactivated FBS (Gibco) in the chamber with 5% CO₂ at 37. The primary fibroblasts used in this study were never passaged over five times.

For lentivirus transduction of primary cells, lentivirus was firstly packaged in phoenix cells. For lncH19 expression, pLJM1-EGFP plasmids (no. 19319; Addgene) were inserted with full-length lncH19 (pMSCV-H19) or H19Δ123 (pMSCV-H19Δ123). For p53 guide RNA, the open-access software program CRISPR (https://crispr.mit.edu/) was used to design guide RNAs targeted to Trp53 exon 5. Annealed oligonucleotides were ligated into the lentiCRISPR v2 (no. 52961; Addgene) plasmid backbone. For lentivirus production, 10 μg lentiviral plasmids were transfected with Lipofectamine 2000 Transfection Reagent (no. 11668019; Thermo Fisher Scientific) into Phoenix packaging cells along with 5 μg pMD2.G (no. 12259; Addgene) and 5 μg psPAX2 (no. 12260; Addgene). Viral supernatants were harvested 48 h after transfection. Then, PDGFRα⁺ cells or dermal fibroblasts were transduced with diluted virus (1:10 ratio) in the growth media in the presence of 8 μg/mL polybrene (no. H9268; Sigma-Aldrich). The cells were then replaced in fresh growth media following 24-h incubation with lentiviruses as indicated prior to further experiments.

Bone marrow–derived macrophages (BMDMs) used in this study were differentiated from bone marrow cells (BMCs), which were isolated from the femurs and tibias of male mice as previously described (18). Briefly, bone marrow cells were maintained in differentiation medium (DMEM [1 mmol/L glucose, no. D6046; Sigma-Aldrich] supplemented with 10% FBS and 20 ng/mL M-CSF [no. 315-02; PeproTech]) for 7 days before further experiments. In vitro chemotaxis assay was performed with a Transwell plate (no. 3428; Corning). BMDMs (1 × 10⁵ cells/100 μL/well) were loaded into the upper chamber with an 8-μm polycarbonate membrane, whereas freshly sorted PDGFRα⁺ cells were seeded in the lower chamber. After 2-h incubation, migrated cells were fixed for methanol and counterstained with DAPI. DAPI⁺ cell numbers were counted on six randomly selected ×20 magnification images of stained membrane.

Dissected tissues were fixed in freshly prepared 4% paraformaldehyde for 48 h and then maintained with 50% ethanol solution. Paraffin embedding, sectioning, and hematoxylin-eosin staining were conducted by the Pathological Department of the Shanghai Eighth People’s Hospital. For immunofluorescence staining, paraffin sections were dewaxed and hydrated via incubation in xylene, ethanol, and double-distilled H₂O subsequently. Then, the slides were placed in 1% R-Buffer A (pH 6.0) solution and incubated for ∼2.5 h in Antigen Retriever 2100 (Electron Microscopy Sciences) for antigen retrieval. The slides were incubated with PBS containing 10% normal goat serum (no. 01-6201; Invitrogen) for 40 min at room temperature before being incubated with primary antibodies overnight at 4°C. Following several washings by PBS, the slides were incubated with secondary antibodies for 2 h at room temperature before being mounted with Prolong antifade mounting medium containing DAPI (no. P36941; Invitrogen). Antibodies and concentrations used for immunofluorescence include the following: rabbit anti-mouse CD31, 1:1,000 (no. ab222783; Abcam); mouse anti-mouse Ki67, 1:500 (no. ab279653; Abcam); goat anti-rabbit Alexa Flour 647, 1:200 (no. 21244; Invitrogen); and goat anti-mouse Alexa Flour 488, 1:200 (no. SA5-10150; Invitrogen).

For exosome preparation, APCs (PDGFRα⁺SCA1⁺ cells) freshly sorted from scWAT of mice were seeded in the dishes to grow to reach the confluency of 70–80% prior to changing of the growth media. The conditioned medium was collected after another 24-h incubation and then centrifuged several times to remove the debris as follows: 300g, 5 min; 1,200g, 20 min; 1,200g, 20 min;10,000g, 30 min; and 10,000g, 30 min. Then the exosomes were extracted and enriched according to a magnetic bead–based isolation protocol previously described (19).

Transmission electron microscopy was applied to characterize isolated exosomes. Briefly, 30 μL resuspension of exosomes was placed dropwise onto copper mesh followed by 1 min rest. Subsequently, the exosome sample was counterstained with 30 μL phosphotungstic acid solution (pH 6.8) at room temperature for 10 min. Images of the exosomes were captured with transmission electron microscopy.

For luciferase assays, NIH-3T3 cells (no. CRL-1658; ATCC) were maintained in 10% FBS high-glucose DMEM (Sigma-Aldrich) supplemented with 10% FBS, 100 μg/mL Normocin, 10 μg/mL Blasticidin, and 50 μg/mL Hygromycin B Gold. Cells were placed in 24-well plates and transfected with Lipofectamine LTX with p53 activity luciferase reporter and Renilla internal control vector, in combination with constructs expressing full-length H19 (pMSCV-H19) or H19 mutant (pMSCV-H19Δ123), or empty vectors. The luciferase reporter construct with p53-responsible element was obtained from Promega (no. E3651). After 24 h incubation, luciferase activity of cell extract was evaluated with Dual-Luciferase Reporter Assay System (no. E1910; Promega). The values of Renilla were used for internal normalization.

For the cell growth assays, primary dermal fibroblasts were sorted out and seeded into a 96-well plate (∼1 × 10³ each well). The optical density at 450 nm values of the cells were evaluated at 0, 2, 3, and 4 days after culture with use of the Cell Counting Kit-8 (cat. no. CK04; Dojindo Molecular Technologies) according to the manual. In brief, a mixture of 10 μL reagent and 90 μL cell medium was added into each well after the discard of the original cell medium. After 2 h incubation, OD450 values were determined by plate reader for the evaluation of cell number.

Total RNA from tissues and cells was extracted and purified with TRIzol reagent (no. 15596026; Invitrogen), and total RNA from freshly FACS-sorted cells was extracted and purified with use of the RNAqueous-Micro Total RNA Isolation Kit (no. AM1931; Thermo Fisher Scientific). For reverse transcription, cDNA was synthesized with use of random hexamer primers (no. N8080127; Thermo Fisher Scientific) and M-MLV Reverse Transcriptase (no. 28025013; Thermo Fisher Scientific). Relative levels of mRNAs were determined by quantitative real-time PCR with the SYBR Green PCR system (Applied Biosystems), and housekeeping gene Rps18 was used as an internal control for calculation with use of the ΔΔ−Ct method. All primers sequences used in this study are listed in Supplementary Table 1.

Proteins were extracted from cells or tissues by homogenization in radioimmunoprecipitation assay lysis buffer (no. 89900; Thermo Fisher Scientific). The supernatant was collected for centrifuging prior to incubation at 100°C with the addition of SDS-PAGE loading buffer. After the proteins in the samples were primarily separated by 10% SDS-PAGE, the proteins were then transferred onto polyvinylidene fluoride membrane (no. IPVH00010, Millipore). After incubation with the indicated primary antibodies at 4°C overnight, the blots were incubated with horseradish peroxidase–conjugated secondary antibodies. The bands of proteins were detected by enhanced chemiluminescence assay. The primary antibodies and diluted ratio are as follows: anti-p53, 1:1,000 (no. 2524; Cell Signaling Technology) and anti-PTEN, 1:1,000 (no. 9188; Cell Signaling Technology).

RNA-binding protein immunoprecipitation (RIP) was conducted with the EZ-Magna RIP Kit (no. 17-701; Millipore) according to the manufacturer’s instructions. Briefly, cells were lysed with use of RIP lysis buffer. Subsequently, the cell extracts were coimmunoprecipitated with anti-p53 (no. 2524; Cell Signaling Technology) and anti-PTEN (no. 9188; Cell Signaling Technology), or mouse IgG as negative control. The retrieved RNA was applied to quantitative RT-PCR, and U1 was used as a nonspecific control target.

For RNA pull-down assays, the fragments of H19 and mutants were firstly generated by in vitro transcription assays with the MEGAscript T7 Transcription Kit (no. AM1334; Life Technologies) according to the manufacturer’s instructions. The RNA of H19 and mutants were labeled by desthiobiotinylation with use of the Pierce RNA 3′ End Desthiobiotinylation Kit (no. 20163; Thermo Fisher Scientific). Subsequently, RNA pull-down assays were conducted with the Magnetic RNA-Protein Pull-Down Kit (no. 20164; Thermo Fisher Scientific) according to the manufacturer’s instructions. The proteins were detected with immunoblotting assays.

Primary cells were sorted out and seeded into a 96-well plate (∼1 × 10³ each well). At 1 and 4 days after the culture, the cells were collected for determination of the collagen contents with use of Soluble Collagen Assay Kit (Fluorometric) (no. K532-100; BioVision) according to the manual.

All the data are presented as mean ± SD. Statistical significance was set as P value < 0.05, with analysis by the software GraphPad Prism, version 7.0, using the two-tailed unpaired Student t test for two groups or one-way or two-way ANOVA for more than two groups. All of the experiments presented in this article were repeated two times or more.

All the data and resources in this study are available on reasonable request to corresponding author Q.H.

Given that lncH19 is involved in DFU in animals (15), we asked whether lncH19 is implicated in the progression of wound healing of skin tissues. To this end, we collected injured skin tissues at the time points of 0, 1, 3, 6, 9, and 13 days post–wound injury from the back of wild-type mice. lncH19 expression in the injured tissues changes rapidly during the time and reaches the peak ∼3 days after injury (Fig. 1A), reflecting the implication of lncH19 in regulating wound healing, particularly in the early phases. To further characterize the physiological functions of lncH19 in this process, we introduced global lncH19 KO mice (H19-KO), which displayed body weight comparable with that of control animals (Supplementary Fig. 1A and B). The absence of lncH19 apparently delayed the wound healing process in the animals (Fig. 1B), indicating the requirement of lncH19 for the wound healing. To explore the role for lncH19 in wound healing in T2D, we analyzed lncH19 expression in injured cutaneous tissues collected from individuals without diabetes and T2D patients with obesity, as well as normal skin tissues from individuals without diabetes. Compared with those of the normal control group, the wound tissues of individuals without diabetes exhibited much higher levels of lncH19 transcripts (Fig. 1C). However, cutaneous lncH19 expression was much lower in injured tissues of T2D patients compared with control subjects without diabetes (Fig. 1C), suggesting to us that decreased lncH19 may contribute to the impaired tissue repair in diabetes.

To further evaluate the function of lncH19 in chronic wound healing in diabetes, we developed a type 2 diabetic animal model by feeding wild-type mice with high-fat diet (HFD) (60% of calories from fat). After being maintained on HFD for >12 weeks, the mice displayed extreme obesity, hyperinsulinemia, impaired glucose disposition capability, and also systemic insulin resistance (Supplementary Fig. 1C–F), indicating the successful induction of T2D in the mice. Next, these T2D mice accompanied by chow diet–feeding controls were subjected to wound injury on the back (Fig. 1D). Consistently, lncH19 expression levels were upregulated in the injured tissues of chow diet–feeding control animals (Fig. 1E). Meanwhile, injury-induced upregulation of lncH19 expression was reduced in the injured sites of diabetic mice (Fig. 1E). In parallel with this, the wound healing of T2D mice exhibited a much slower rate compared with that of the control group (Fig. 1F).

To identify the precise cell contributors of lncH19 in cutaneous tissue, we fractionated skin tissues and separated them into four main cell subsets according to the surface markers by FACS, including CD45⁺ hematopoietic immune cells, CD31⁺ endothelial cells, PDGFRα⁺ fibroblasts, and nondefined cells (negative). lncH19 RNA was highly enriched in PDGFRα⁺ cells compared with the other three subpopulations (Fig. 2A), reflecting that this fibroblast subset acts as the main source for lncH19 in skin. Linking these with data in Fig. 1A, we wanted to determine whether the increased lncH19 contents in wound sites is due to upregulated expression in dermal fibroblasts per se or increased fibroblasts in number. Relative to the control group, the freshly sorted PDGFRα⁺ cells from skin tissues 3 days postinjury exhibited much higher levels of lncH19 transcript (Fig. 2B). Additionally, wound tissues displayed more PDGFRα⁺ cells compared with control (Fig. 2C). Moreover, lncH19 expression levels were apparently lower in PDGFRα⁺ cells isolated from injured tissues of T2D mice relative to wild-type controls (Fig. 2D).

To further investigate the physiological roles for PDGFRα⁺ cell–expressed lncH19 in wound healing, we generated an inducible PDGFRα⁺ cell–specific lncH19 deficiency mouse model (Pα-H19KO) (Fig. 2E). After the tamoxifen treatment, relative to control animals, Pα-H19KO mice exhibited much lower lncH19 levels in both isolated PDGFRα⁺ cells and the skin tissues (Fig. 2F), but no changes were observed in other tissues (white adipose tissue, liver, skeletal muscle, and colon) (Supplementary Fig. 2A). Next, Pα-H19KO mice and control counterparts were subjected to skin injury. For calculating the area of wound sites, we measured the diameters of each wound site in three different directions using a vernier caliper to get the average value. The averaged value was then normalized to the diameter of the original wound site at day 0 (labeled as the circle of Fig. 2H). All of the area not completely closed and covered with scars was considered as the wound. At day 9 postinjury, >50% of the wound sites in Pα-H19KO mice were still healing compared with only ∼30% in control animals (Fig. 2G and H). lncH19 deficiency in PDGFRα⁺ cells obviously delayed the formation of new layer in the wound site (Fig. 2I). These results are in line with previous data regarding the requirement of lncH19 for wound healing.

Given that fibroblasts accumulation in proliferative phase is essential for wound healing (20), we tested whether the deficiency of lncH19 impacts on the proliferation of dermal fibroblasts per se. First, in vitro assays demonstrated that the growth rate of dermal fibroblasts from Pα-H19KO mice was decreased in comparison with that of control animals (Fig. 3A). In line with the reduced cells, collagen production in the cultured H19KO cells was also less than in control cells (Fig. 3B). Moreover, the frequency of Ki67-expressing PDGFRα⁺ cells was markedly reduced in wound tissues of Pα-H19KO animals (Fig. 3C and D). Consequently, Pα-H19KO mice displayed much fewer PDGFRα⁺ cells in injured skin tissues (Fig. 3E) and lower expression levels of fibrogenic genes in injured tissues (Fig. 3F) compared with control animals.

The infiltration of immune cells into the injured site, particularly macrophages and neutrophils, in the early phase (∼3 days postinjury) is essential for skin repair for their critical functions in getting rid of debris and dead cells and also promoting angiogenesis (20,21). To explore whether lncH19 influences the infiltration of immune cells, we analyzed macrophage contents in skin tissues 2 days postinjury. Pα-H19KO mice exhibited an obviously lower proportion of CD45⁺CD11b⁺F4/80⁺ macrophages (Fig. 3G) and also less mRNA of macrophage marker genes (Adgre1, Cd11b, Cd68) than the control animals (Fig. 3H), indicating that lncH19 is required for the recruitment of immune cells in injured skin. Furthermore, without any significant impacts on inflammatory genes expression in PDGFRα⁺ cells per se (Supplementary Fig. 2B), the abrogation of PDGFRα⁺ cell lncH19 resulted in obviously reduced inflammatory genes (Ccl2, Il6, Tnfa, Il1b) expression and upregulated expression of anti-inflammatory genes (Mrc1, Il10) (Fig. 3I). Moreover, we observed that lncH19 deficiency caused much less contents of LY6C⁺CCR2⁺ macrophages in the injured sites, which were reported as important players in promoting angiogenesis during wound healing (21). Consistent with these findings, Pα-H19KO animals exhibited lower expression levels of Vegfa and Cd31 genes and also less contents of CD31⁺ blood vessel cells relative to control mice (Fig. 3K and L). Together, our data demonstrate that lncH19 expressed in PDGFRα⁺ fibroblasts accelerates mouse cutaneous wound healing via regulating dermal fibroblast proliferation and immune cell recruitment to promote extracellular matrix distribution and angiogenesis.

To further explore molecular mechanism(s) underlying lncH19-regulated fibroblasts proliferation, we firstly checked the cell cycle of fibroblasts by using flow cytometry. The cell cycle was obviously arrested in lncH19-deficient fibroblasts, which is in line with the reduction of proliferative capacity of Pα-H19KO fibroblasts (Supplementary Fig. 3A). In previous studies, it has been appreciated that p53, the critical mediator in cell cycle arrest, is implicated in regulating fibroblast proliferation and wound healing (22,23). We hypothesized that p53 is involved in lncH19-regulated fibroblast proliferation. Consistently, mRNA levels of genes associated with cell cycle arrest (Cdkn1a, Ccne1, Ccna2, and Ccnb1), of which the transcription is controlled by p53, were obviously heightened in dermal fibroblasts isolated from Pα-H19KO animals (Fig. 4A) and decreased in the cells with lncH19 overexpression (Supplementary Fig. 3B) in comparisons with their control groups. lncH19-deficient mice displayed significantly increased p53 expression in injured skin tissues (Fig. 4B). Additionally, the patterns of p53 expression in injured skin tissues of diabetic animals and patients with diabetes (Fig. 4C and D) were the inverse of these of lncH19 transcripts (Fig. 1C and E), suggesting a closely negative correlation between lncH19 and p53. To determine whether the impaired proliferative capability of lncH19-deficient fibroblasts is p53 dependent, we deleted p53 gene in primary fibroblasts using CRISPR/Cas9 (Supplementary Fig. 3C). The slower proliferation rate of lncH19-deficient fibroblasts was restored when p53 gene was abrogated (Fig. 4E). In addition, the deficiency of p53 slightly but insignificantly augmented the increase of fibroblast proliferation caused by lncH19 overexpression (Supplementary Fig. 2D).

To test whether lncH19 influences the transcription activity of p53, we introduced a reporter construct in which the transcription of luciferase gene is controlled by the p53 response element. Overexpression of lncH19 efficiently repressed the luciferase activity, which was augmented by p53 overexpression in NIH-3T3 cells (Fig. 4F). Next, we hypothesized that lncH19 could bind directly to p53 proteins on the basis of the result predicted from a bioinformatics algorithm (https://pridb.gdcb.iastate.edu/RPISeq/index.html) (Supplementary Fig. 3E). To this end, we conducted RIP assays using the antibodies specifically against p53 proteins (α-p53 Ab). We also used antibodies against PTEN proteins (α-pTEN Ab) as a negative control, as PTEN was reported to be implicated in fibroblast proliferation during wound healing (15) but not predicted to be an lncH19-binding target according to the database (Supplementary Fig. 3E). The use of α-p53 Ab immunoprecipitated >20-fold lncH19 RNA compared with the groups of IgG control or α-PTEN Ab, and moreover, the enrichment of RNA by α-p53 Ab was specific to lncH19 RNA rather than U1 controls (Fig. 4G), indicating the direct interaction between p53 protein and lncH19 RNA. To further support these findings, we performed RNA pull-down assays in cell lysates of NIH-3T3 fibroblasts that were previously transfected with biotinylated fragments of lncH19 sense as well as its antisense. More highly enriched p53 proteins in the pull down were observed in the group of lncH19 sense fragment rather than antisense fragment (Fig. 4H). To further identify the p53-interacting region within lncH19, we generated four new biotinylated fragments of lncH19 (1–570Δ1, 571–1140Δ2, 1141–1710Δ3 and 1711–2288Δ4) (Fig. 4I). The p53 proteins were visibly detected only in the pull down by 1711–2288Δ4 (Fig. 4I), indicating p53 protein interacts with lncH19 RNA within the region of 1711–2288 base pairs. Furthermore, overexpressed lncH19 with truncation in the region of 1711-2288 (H19Δ123OE) obviously repressed the effects of lncH19 overexpression on promoting fibroblast proliferation (Fig. 4J). Collectively, our results indicate that lncH19 RNA interacts with p53 protein directly to repress the transcription activity of p53 and enhance dermal fibroblasts proliferation.

Growth and differentiation factor 15 (GDF15), also known as microphage inhibitory factor 1 (MIC-1) and also a target gene of p53 (24), exerts a critical role in repressing macrophage infiltration and inflammatory activation (25,26). Given that GDF15 is elevated in patients with diabetes and even considered as a biomarker for T2D (27,28), we hypothesized that GDF15 is involved in regulating macrophage accumulation in diabetic injured sites. To this end, we evaluated the expression levels of Gdf15 in the skin tissues in wound healing. The wound cutaneous tissues from both T2D patients and mice displayed dramatically elevated lncH19 transcripts in comparison with control (Fig. 5A and B). Notably, Gdf15 mRNA levels in skin tissues of lean mice were significantly reduced on injury (Fig. 5C), but these changes were almost abolished when lncH19 was deleted from PDGFRα⁺ cells (Fig. 5C). It is worth noting that lncH19 deficiency in PDGFRα⁺ cells slightly heightened Gdf15 expression in skin tissues (Fig. 5C) and also in sorted fibroblasts from injured sites (Fig. 5D), suggesting the implication of lncH19 in regulating GDF15 expression. In line with this, heightened Gdf15 expression caused by lncH19 deficiency was significantly compromised in the fibroblasts without p53 (Supplementary Fig. 4A). Additionally, Gdf15 expression was dramatically repressed in fibroblasts with overexpression of full-length lncH19 (H19OE) but not lncH19 truncated fragments (H19Δ123OE). Overexpression of p53 restored Gdf15 expression, which was repressed in lncH19-overexpressed cells (Fig. 5E). These data together indicate that lncH19 suppresses Gdf15 expression in fibroblasts in a p53-depedent manner.

To explore the role for GDF15 in lncH19-deficient wound healing, we applied neutralized antibodies against GDF15 (α-GDF15 Ab) to get rid of secreted GDF15 proteins. α-GDF15 Ab efficiently accelerated the progression of wound healing in T2D mice (Supplementary Fig. 4B) and Pα-H19KO mice (Fig. 5F). The application of α-GDF15 Ab apparently elevated macrophage contents in wound tissues of Pα-H19KO mice (Fig. 5G and Supplementary Fig. 4C). Additionally, the expression of proinflammatory genes in injured tissues increased on addition of α-GDF15 Ab (Fig. 5H). To evaluate the infiltrated capability of macrophages under the induction of fibroblasts, we separately cocultured BMDMs and freshly isolated PDGFRα⁺ fibroblasts from tissues in wound healing in a Transwell system (Fig. 5I). Relative to the control group, the number of infiltrated macrophages was significantly compromised in the presence of lncH19-deficient fibroblasts and mostly restored after the addition of α-GDF15 Ab (Fig. 5I). These data demonstrate that GDF15 upregulated in diabetic or lncH19-deficient fibroblasts inhibited inflammatory macrophage accumulation during wound healing.

A growing number of studies reported that lncH19 in bone marrow MSCs (BM-MSCs) could promote wound healing via the delivery by MSC-derived exosomes (15), leading us to hypothesize that low lncH19 contents in diabetic skin could be replenished by the adding of lncH19-carried exosomes. To this end, we isolated CD45⁻CD31⁻PDGFRα⁺SCA1⁺ MSCs, which are considered as the APCs (29), from various adipose tissue depots including epididymal white adipose tissue (epWAT), mesenteric white adipose tissue, scWAT, and brown adipose tissue using FACS (Fig. 6A). APCs isolated from scWAT displayed much higher lncH19 expression relative to other depots (Fig. 6B). Moreover, these cells had highly proliferative capability, were able to be passaged several times, and also displayed an elongated irregular granular polygon shape (Supplementary Fig. 5A). Some of the cells spontaneously differentiated into adipocytes even without induction, and most of the cells displayed perfect phenotype as adipocytes with numerous lipid droplets after being induced to differentiation in vitro (Supplementary Fig. 5A), indicating the high adipogenic capacity of these APCs. To ascertain that lncH19 could be secreted extracellularly by APCs in the manner of exosomes, we prepared the extracellular vesicles (EVs) from the cell culture (Fig. 6A). Most of the EVs from APCs were 80–120 nm in size and exhibited high levels of exosome markers (CD61 and CD81), indicating that these EVs were exosomes (Supplementary Fig. 5B and C). We also found that lncH19 was extremely enriched in isolated exosomes rather than the nonexosome supernatant (NES) (Fig. 6C). To test whether APC-derived exosomes could promote wound healing in diabetic mice, we locally injected the isolated exosomes to the injured sites of T2D animals. The application of exosomes efficiently improved the wound healing in the mice relative to that of NES (Fig. 6D). To further define whether the improvement caused by exosomes was contributed by the PDGFRα⁺ cell–expressed lncH19, we treated the T2D mice with exosomes from Pα-H19KO mice in which lncH19 contents were very low (Fig. 6E). Compared with the control group, lncH19 deficiency significantly slowed down wound healing procession, suggesting that lncH19 is required for the beneficial effects of APC-derived exosomes (Fig. 6F). Additionally, we administrated T2D mice with exosomes from lncH19-overexpressed APCs and found significant acceleration in wound healing relative to exosome from control cells (Fig. 6G). Compared with NES, the presence of exosomes raised the growth rate of fibroblasts in vitro, implying that exosomes exert the promoting function on wound healing through enhancing fibroblast proliferation (Fig. 6H). These impacts of exosomes were significantly impaired with lncH19 deficiency (Fig. 6H) but enhanced with lncH19 overexpression (Supplementary Fig. 5D). Meanwhile, the expression of p53 target genes as well as Gdf15 in fibroblasts changed in parallel with the cell growth rate (Fig. 6I and Supplementary Fig. 5E). Together, these results indicate that exosomes derived from scWAT APCs improve wound healing of T2D mice via delivering lncH19 to fibroblasts.

lncH19 plays important roles in embryo development and growth control and is associated with human genetic disorders. Recently, a growing body of studies implied the critical functions for lncH19 in regulating the development of chronic metabolic syndromes, including diabetes (13); cardiovascular disease (30,31); and DFU (15). Our current study demonstrates that lncH19, particularly enriched in PDGFRα⁺ dermal fibroblasts, promotes fibroblast proliferation by controlling p53-regulated cell cycle and regulates immune cell infiltration via repressing fibroblast-derived GDF15. Lower contents of lncH19 in diabetic wound tissue confer the impaired wound healing of patients and animals with T2D due to the attenuated fibroblast growth and macrophage recruitment in injured tissues. Exosomes derived from APCs deliver lncH19 RNA to injured sites of T2D mice to improve the impaired wound healing (Fig. 6J).

The alterations of lncH19 levels are well appreciated in multiple tissues of patients with diabetes and diabetic animals, although these results are controversial (13,32,33). In our study, while no significant differences were seen in murine normal tissues, lncH19 expression levels were obviously decreased in the wound-healing cutaneous tissues from patients with diabetes and diabetic animals relative to their individual control counterparts (Fig. 1C and E). However, the molecular mechanism for regulating lncH19 expression is still unclear. It is worth noting that cutaneous lncH19 levels are robustly induced on injury and significant differences in lncH19 levels between nondiabetic and diabetic mice can only be observed in the setting of wound healing (Fig. 1), which imply that lncH19 may act as a stress-responsive molecule. That is in line with previous studies where it was reported that lncH19 expression is highly regulated in various settings of stress, including oxidative stress (34), hypoxia (35), chemotherapy (36), and abnormal proliferation-induced cellular stress (37). Linking to the extreme stress environment inside wound-healing tissues, it is rational to hypothesize that the increase of lncH19 expression is one of the stress responses on injury. Thus, many more studies are needed to elucidate the molecular mechanism in lncH19 expression regulation in future.

In a recent study it was reported that lncH19 delivered by MSC-derived exosomes promoted wound healing of DFU in type 1 diabetes via regulating fibroblast growth and migration (15). Even far from clear in mechanism, the findings of this study support our findings that endogenous lncH19 in PDGFRα⁺ dermal fibroblasts is required for fibroblast proliferation (Fig. 3) and replenishing lncH19 to diabetic dermal fibroblasts by APC-derived exosomes efficiently restores the impaired proliferative capability and subsequently wound healing (Fig. 6). Compared with the streptozotocin-induced type 1 diabetes animal model used in that study (15), the pathological model in our study is the mouse with HFD-induced obesity and T2D, which is more physiological and similar to T2D patients. Moreover, we uncovered a novel mechanism for lncH19-promoted fibroblast proliferation and macrophage recruitment in which lncH19 interacts directly with p53 protein to inhibit its binding to the promoter of p53-targeted genes (i.e., cell cycle genes and Gdf15) so as to relieve the cell cycle arrest and repress GDF15 release in injured sites (Fig. 4). Previous study revealed that lncH19 is specifically recruited to the promoter region of Notch1 gene to occupy the binding site for p53 to impede the transcription of Notch1 (38). In contrast to this, we found a large number of p53-target genes changing in dermal fibroblasts with lnvH19 alterations (Fig. 4A and Supplementary Fig. 3B), supporting that fibroblastic lncH19 exerts its function in repressing p53 activity in trans during wound healing.

As a critical mediator in the development of obesity and T2D, GDF15 could serve as a novel biomarker (27,28) and a promising therapeutic target for clinical treatment of diabetes and obesity (39,40). As far as we know, our current study is the first to uncover the critical role of GDF15 in diabetic wound healing. In previous study (24), the transcription of fibroblastic Gdf15 was also controlled by p53 via promoter region binding. We found that disrupting of this binding by lncH19 efficiently represses GDF15 expression. In contrast to regulating fibroblast proliferation, elevated GDF15 from dermal fibroblasts in diabetic mice reduces immune cell infiltration into injured tissues at the inflammatory phase of wound healing (Fig. 5G). Linking to the fact that GDF15 exerts a critical role in repressing macrophage infiltration and inflammatory activation (25,26), we also demonstrate that Gdf15 expression in fibroblasts modulated by lncH19 is inversely correlated with macrophage contents of injured tissue in vivo and macrophage infiltration in vitro (Fig. 5). In support of this, the application of α-GDF15 Ab in the injured tissues efficiently accelerates the progression of wound healing in T2D mice (Fig. 5F). Although we have not evaluated the changes of other immune cells (i.e., neutrophils) in this process, we can justly conclude that elevated GDF15 confers the delayed tissue repair in patients with diabetes partially through repressing macrophage infiltration in the early phase of wound healing. Notably, immunotherapy with use of α-GDF15 Ab may serve as a new alternative method for clinical treatment in diabetic wound healing (i.e., DFU).

MSC-derived exosomes are widely appreciated for application in cell-free regeneration medicine. Exosomes produced by BM-MSCs are generally preferred in treating wound healing (41). In our current studies, we used adipose progenitor cells instead of BM-MSCs as the source for exosomes to deliver lncH19 mainly for the following two reasons: 1) as a kind of mesenchymal-like stem cell, APCs are also the main contributors for exosome production and also widely applied in wound healing study (41–43), and 2) compared with BM-MSCs, APCs are higher in abundance, less immunogenic, more genetically stable, and easer to obtain without invasive procedures, especially for humans (30,44). PDGFRα⁺SCA1⁺ cells within adipose stromal cells are acknowledged as the adipocyte progenitors of white adipose tissue in humans and mice (45,46), and isolated scWAT APCs can fully differentiate into adipocytes in vitro (Supplementary Fig. 3A). Although we have not identified the precise target cell type(s) in the injured tissue of host, our data indicate that the lncH19-carrying exosomes efficiently accelerate the delayed wound healing process in diabetic mice partially via targeting dermal fibroblasts (Fig. 6H–I and Supplementary Fig. 5D and E). Our current study suggests the high potential of application of this kind of APC-derived exosome in cell-free regeneration medicine in future.

In summary, our study reveals a new role for lncH19 in promoting cutaneous tissue repair in both normal and diabetes contexts. This work provides novel insight into the development of chronic skin ulcers in diabetes and a new promising target for developing clinical therapy in the treatment of diabetes.

Funding. This work was partially supported by the Xuhui District Medical Research Project of Shanghai (grants SHXH201802 and SHXH201606) and the Shanghai Municipal Key Clinical Specialty (shslczdzk00901).

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. P.Y., W.C., and Q.H. conceived and designed the studies. P.Y., Y.J., and Q.H. conducted most of the experiments and analyzed the data. J.G. and X.S. conducted some of the cell experiments. J.L., N.X., and W.C. helped conduct some of the animal experiments. P.Y., Y.J., and Q.H. wrote the manuscript. Q.H. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.