Single-cell analysis of skin immune cells reveals an Angptl4-ifi20b axis that regulates monocyte differentiation during wound healing

The persistent inflammatory response at the wound site is a cardinal feature of nonhealing wounds. Prolonged neutrophil presence in the wound site due to failed clearance by reduced monocyte-derived macrophages delays the transition from the inflammatory to the proliferative phase of wound healing. Angiopoietin-like 4 protein (Angptl4) is a matricellular protein that has been implicated in many inflammatory diseases. However, its precise role in the immune cell response during wound healing remains unclear. Therefore, we performed flow cytometry and single-cell RNA sequencing to examine the immune cell landscape of excisional wounds from Angptl4+/+ and Angptl4−/− mice. Chemotactic immune cell recruitment and infiltration were not compromised due to Angptl4 deficiency. However, as wound healing progresses, Angptl4−/− wounds have a prolonged neutrophil presence and fewer monocyte-derived macrophages than Angptl4+/+ and Angptl4LysM−/− wounds. The underlying mechanism involves a novel Angptl4-interferon activated gene 202B (ifi202b) axis that regulates monocyte differentiation to macrophages, coordinating neutrophil removal and inflammation resolution. An unbiased kinase inhibitor screen revealed an Angptl4-mediated kinome signaling network involving S6K, JAK, and CDK, among others, that modulates the expression of ifi202b. Silencing ifi202b in Angptl4−/− monocytes, whose endogenous expression was elevated, rescued the impaired monocyte-to-macrophage transition in the in vitro reconstituted wound microenvironment using wound exudate. GSEA and IPA functional analyses revealed that ifi202b-associated canonical pathways and functions involved in the inflammatory response and monocyte cell fate were enriched. Together, we identified ifi202b as a key gatekeeper of monocyte differentiation. By modulating ifi202b expression, Angptl4 orchestrates the inflammatory state, innate immune landscape, and wound healing process.


Supplemental Figure legends
(A) Single-cell suspensions were obtained from Angptl4 +/+ and Angptl4 -/excisional wounds using optimized tissue dissociation protocols. 1-dpw wounds were pooled from three mice, and 5-dpw wounds were pooled from five mice. Cells were first gated using forward and side scatter to exclude debris and cell doublets, viable cells were gated as propidium iodide-negative cells, and immune cells were gated as CD45-positive cells.
(B) Quality control metrics of the collected viable cells. Single-cell libraries were prepared using 5000-6000 cells as input for each sample. After sequencing, alignment, and filtering, we obtained combined data from a total of 19579 cells.   Publicly deposited mouse DNAse-seq data available on ENCODE across all tissues and cell lines were studied. The datasets containing DNAse I hypersensitive sites on ifi202b are graphically shown here, using a genomic visualization tool by the UCSC Genomics Institute. The chromosomal location and intron-exon structure of ifi202b are displayed to scale. Each row shows the normalized fold change signal from DNAse-seq data. Dataset identifiers for the displayed ENCODE datasets are listed next to each row.    Figure S1A. The four populations were as follows: 1= Ly6C hi monocytes; 2=Ly6C int monocytes; 3=differentiating macrophages; and 4=macrophages.
(C) IPA analysis reveals enriched gene ontologies in inflammatory response, growth factor signaling, cellcell and cell-matrix interactions and cell signaling. Table S1: Differentially expressed genes between Angptl4 +/+ and Angptl4 -/monocytes and macrophages from wounds. Differential gene expression between Angptl4 +/+ and Angptl4 -/monocytes and macrophages was conducted using the nonparametric Wilcoxon rank sum test. All statistically significant differentially expressed genes are shown in this table. FC denotes the fold change in Angptl4 -/mice compared with Angptl4 +/+ mice.

Bioinformatic analysis of single-cell RNA sequencing data
The single-cell RNA sequencing data were analyzed using the R package Seurat (v3.0) (1). Datasets were first screened for quality control by excluding cells with fewer than 200 and more than 4500 detected genes to remove empty droplets and doublets. Cells with more than 5% mitochondrial genes were excluded from subsequent analysis to remove cells of poor viability. Genes detected in less than 3 cells were also excluded from further analysis. Gene expression data were then normalized using a regularized negative binomial regression method, and a subset of features that exhibited high cell-cell variation was calculated using a variance stabilizing transformation with SCTransform. These variable features were used for subsequent analyses.
Briefly, a shared gene correlation structure conserved between all datasets was learned using canonical correlation analysis (CCA). These datasets are then aligned into a conserved low-dimensional space using dynamic time warping algorithms to normalize for differences in feature scale. Finally, the integrated data were used for further downstream analysis and visualization.
Clustering is performed using a K-nearest neighbor (KNN) graph approach, followed by modularity optimization using the Louvain algorithm. Nonlinear dimensional reduction using UMAP was used to visualize the data. Cell type annotation was performed using the ImmGen database from celldex (v1.2.0) (2).
Differential gene expression testing was performed using the nonparametric Wilcoxon rank-sum test. For the identification of cluster biomarkers, differential expression testing was conducted between each cluster compared against all remaining clusters.

Pseudotime trajectory analysis
Pseudotime trajectory analysis was performed using Monocle3 (3). Preprocessed 10X genomics count barcode matrices were supplied as input to Monocle. Monocyte and macrophage populations were identified from the UMAP plots and subsets into a separate data file. A single-cell trajectory was then learned using Monocle's algorithm, with the monocyte cell cluster specified as the root node (pseudotime 0) of the trajectory. Cells were then ordered according to their calculated pseudotime.

RNA Extraction and Real-time Quantitative PCR
Total RNA was first extracted using TRIzol (Invitrogen, USA) reagent according to the manufacturer's instructions. After phase separation and precipitation, RNA was transferred to an RNA column-based kit (Research Instruments). Total RNA (1 μg) was quantified using a NanoDrop ND-1000 (Thermo Scientific) and reverse transcribed to cDNA using qScriptTM Reverse Transcription Supermix (Quanta Bio) according to the manufacturer's protocol. qPCR was performed using a Bio-Rad C1000 Thermal Cycler equipped with a CFX96 Real-Time System. Reactions were prepared using PerfeCTa SYBR Green Mix (Quanta Bio) according to the manufacturer's instructions. The primers used are outlined in Table S.

Wound fluid collection
Six-millimeter excisional wounds were created on the dorsal skin of Angptl4 +/+ or Angptl4 -/mice and covered with stacks of round absorbent papers. The wounds were then covered with parafilm and secured with Tegaderm TM Film Dressing. The absorbent papers were collected and replaced every 10-12 hours for up to two days. The volume of wound fluid absorbed by the papers was estimated based on weight change. Absorbent papers from the same genotype were pooled, and an equal volume of serumfree AIM-V medium with 2% penicillin/streptomycin was added. The mixture was sonicated for 15 minutes in an ice-cold water bath. Wound fluid was collected from the absorbent papers and sterile filtered. The protein concentration was determined with the Bradford assay.

Isolation of bone marrow and PBMC-derived monocytes
Immune cells were purified from PBMCs and bone marrow using Percoll at a 1.07 g/mL density. Positive selection of CD11b + cells was performed with CD11b MicroBeads (Miltenyi Biotec) according to the manufacturer's protocol. The CD11b-enriched fraction was used for in vitro monocyte culture.

RNA interference
Angptl4 silencing was performed using ON-TARGETplus SMARTpool siRNA (Horizon Discovery) targeting mouse Angptl4. RAW264.7 macrophage cells were transfected with siRNA using DharmaFECT 1 reagent (Horizon Discovery) according to the manufacturer's instructions. Transfection efficiency using DharmaFECT 1 reagent was optimized using siGLO reagent to achieve more than 80% transfection efficiency. Knockdown of ifi202b in bone marrow and PBMC-derived monocytes was achieved with Accell mouse ifi202b SMARTpool siRNA (Horizon Discovery) at a final concentration of 1 μM. Control cells were transfected with si-scrambled. Knockdown efficiency was determined by qPCR.

Western blot
Cells were lysed in ice-cold RIPA buffer and centrifuged at 12,000 g for 10 minutes. The supernatant was collected, and Laemmli loading buffer was added. Proteins were run on a 10% SDS-PAGE gel and transferred to a PVDF membrane. The membrane was blocked for 1 hour with Li-Cor Odyssey blocking buffer and incubated overnight with the respective primary antibodies at 4°C. The membrane was washed 3 times with TBST before incubation with secondary antibody for 1 hour at room temperature. The membrane was washed 3 times with TBST, dried, and scanned using an Odyssey CLx scanner (Licor).
Antibodies against ifi202b and β-tubulin were purchased from Santa Cruz and Developmental Studies Hybridoma Bank, respectively. Anti-mouse Angptl4 antibody was produced in-house as reported (4,5).

Cytokine assay
For the in vivo wound fluid cytokine assay, excisional wounds were performed on mice as described above. Sterile blotting sponges were placed at the wound incision site, and the wound area was protected with an occlusive Tegaderm dressing (3 M). Sponges were collected at the indicated time points, and wound fluid was extracted through centrifugation at 10,000 g for 5 minutes. The collected wound fluid was diluted 2-fold and analyzed using the MD31 31 31-plex mouse cytokine array (Eve Technologies).

Kinase Inhibitor Screening
Angptl4 -/-BMDMs were seeded on a 6-well plate prior to the experiment. Cells were treated with a panel of 95 kinase inhibitors from the SYN2103 kinase inhibitor library (SYNKinase) with 1-well of vehicle control. Cells were incubated with the inhibitors for 4 hours before total RNA was extracted, and the mRNA expression of ifi202b was quantified using qPCR. The kinase inhibitor library consisted of the following inhibitors, and cells were treated at IC50 concentrations (6) ( Table S4).

DNAse I Hypersensitivity Assay
Cells were detached with a 0.25% trypsin solution, pelleted, and resuspended in DNAse lysis buffer (10 mM Tris-HCl pH 7.4, 0.1% Triton X-100, 10 mM NaCl, and 3 mM MgCl2). The samples were treated with either 0 U, 0.6 U, 2.4 U, 6 U, 12 U, or 24 U of DNAse I in a 37°C water bath for 5 minutes to digest the chromatin. Two volumes of stop buffer (10 mM Tris-HCl pH 7.4, 10 mM NaCl, 10 mM EDTA, and 0.15% SDS) were added to halt the reaction. DNAse I hypersensitivity assay was performed (9). Digested chromatin was purified using the E.Z.N. A gel extraction kit (Omega Biotek). The digestion of DNase hypersensitive sites was quantified using qPCR primers against DNAse I hypersensitive sites in ifi202b.