Single-Cell RNA Sequencing Reveals the Interaction of Injected ADSCs with Lung-Originated Cells in Mouse Pulmonary Fibrosis

Pulmonary fibrosis (PF) is a severe chronic lung disease with little effective treatment options other than lung transplantation. Adipose-derived mesenchymal stem cells (ADSCs) have been shown to exert therapeutic effects on PF, but the underlying mechanisms remain to be further elucidated. Here, we show the interaction of ADSCs and lung-originated cells at the single-cell level, using bleomycin- (BLM-) induced mice PF model and green fluorescent protein– (GFP–) labeled mouse ADSCs. The intratracheally injected ADSCs were successfully recollected with flow cytometry and, together with lung-originated cells, were subjected to single-cell RNA sequencing (scRNA-seq). The ADSC treatment drastically changed the transcriptomic profile and composition of lung cells, especially macrophages. We explored the signal pathway interactions between ADSCs and lung-originated cells, showing potentially regulative pathways including NGR, ANNEXIN, HGF, and PERIOSTIN. Our data indicate that the injected ADSCs increased the number of Trem2+ antiinflammatory lung macrophages and lowered further inflammation and fibrosis in the lung. Our work realized the direct analysis of injected ADSCs to explore its in vivo interaction with the lung environment under PF and may provide critical information for future engineering of ADSCs to achieve better therapeutic effects in PF.

PF is traditionally considered to be an irreversible lung disease that is refractory to various treatments, resulting in high rates of mortality and morbidity [9]. Recently, the administration of mesenchymal stem cells (MSCs) has emerged as a promising therapy for PF [10,11]. MSCs have been isolated from various tissues such as the bone marrow, adipose tissue, and the stroma of organs. They exhibit selfrenewal ability and the potential to differentiate into a variety of cell lineages, including chondrocytes, osteoblasts, and adipocytes [12,13]. And it is reported that MSCs attenuate fibrotic diseases including PF, by immunomodulation, anti-apoptosis, antiinflammation, and antifibrosis functions [14][15][16][17][18]. Among several sources of MSCs, ADSCs have the advantages of easy accessibility and relative abundance, which make a practical cell source for cell therapy [19,20].
The effects of ADSCs have been widely investigated using both in vitro cultures and in vivo models. But on one hand, direct recovery and analysis of injected ADSCs were not performed yet, likely due to in adaptation and apoptosis after entering the pathological environment [21]; on the  Stem Cells International other hand, current characterization of ADSC-treated BLM lung remains on the tissue (bulk) level, while scRNA-seq can provide a higher resolution of the therapeutic effect, as well as enabling cell-cell interaction inference from the tran-scriptomic data. Therefore, in this study, we recollected the injected allogenic mouse ADSCs from BLM-treated mouse lungs and subjected recollected ADSCs and lungoriginated cells to scRNA-seq to analyze the crosstalk     Groups ii and iii mice were anesthetized with 2.5% tribromoethanol (0.8% NaCl, 1 mM Tris (pH 7.4), 0.25 mM EDTA (pH 7.4)) and administered 3.5 mg/kg bleomycin solution (Sigma B5507-15un) in 50 μL of phosphatebuffered saline (PBS) intratracheally on day 0 using a 1 mL syringe with a 25G needle. Control animals were treated with PBS instead of BLM [23,24].

Materials and Methods
The injection of ADSCs (at passages 3~6) was performed on day 3 post BLM treatment; 50 μL of a cell suspension containing 1 × 10 7 cells/mL in PBS (5 × 10 5 cells/mouse) was injected intratracheally using a 1 mL disposable syringe with a 25G needle. For the control group and BLM + PBS group, 50 μL PBS was injected instead. Before implantation, cells were washed three times to remove the culture medium. The duration of the study was set at four weeks, as the PF remodeling process in mice has been reported to be mostly complete (70~80%) within 3 weeks. All animals were weighed daily from day 0 to 20.   Alveolar macrophage-21  Stem Cells International ECM proteoglycans Immunoregulatory interactions between a lymphoid and a non-lymphoid cell Post-translational protein phosphorylation Requlation of insulin-like growth factor (IGF) transport and uptake by insulin-like growth factor binding proteins (IGFBPs) Platelet adhesion to exposed collagen  7 Stem Cells International lung tissue collection, mice were euthanized by CO 2 inhalation. The chest cavity was exposed, and the lungs were perfused and washed with ice-cold PBS. Then, the lungs were fixed with 4% (v/v) paraformaldehyde in PBS and then processed into 5-μm-thick paraffin sections mounted on glass slides.

Histological Evaluation of Lung Damage and Collagen
For histological assessment of fibrosis accumulation, slices were stained with hematoxylin and eosin (H&E) or Masson's trichrome stain, respectively. Axio Scan.Z1 (Zeiss, Germany) was used to scan the samples with the 20× objective. The images were processed with ZEN 2.3. The lung tissue parenchyma density (alveolar area ÷ parenchyma area×100%) and fibrosis level (fibrosis area ÷ total area×100%) were calculated with the Image-Pro premier 3D 9.2.0 using random and non-overlapping fields for grading the lung injury and fibrosis level [25,26].

Preparation of Single-Cell Suspensions.
For the preparation of single-cell suspensions of mouse lung tissue, mice were sacrificed on the 4th day following ADSCs treatment (namely day 7), and the lungs were perfused with PBS. All samples of BLM + ADSCs and BLM + PBS groups were digested using the Lung Dissociation Kit, mouse (130-095-927, Miltenyi Biotec, Germany) according to the manufacturer's instructions. Cells were passed through 70 μm and 40 μm strainers to remove large pieces, after which the red blood cell lysis solution (Sigma Aldrich, St. Louis, MO, USA) was used to eliminate erythrocytes, followed by flowassisted cell sorting of pulmonary cells (GFP -) and ADSCs (GFP + ), respectively.
2.7. scRNA-Seq Data Processing. The scRNA-seq library sequences were demultiplexed, mapped to the mouse genome, and counted to generate a gene expression matrix using CellRanger (10× Genomics). Further data processing was done using the R packages Seurat3 (UMAP visualization, clustering, and gene expression plots), ReactomePA (REACTOME pathway enrichment analysis), and CellChat (cell-cell ligand-receptor interaction analysis).

Immunofluorescence
Staining. For immunofluorescence staining of APOE and TREM2 in the (ii) BLM + PBS and (iii) BLM + ADSCs groups, mice were anesthetized and lungs were extracted on days 7 and 10. The left lung lobes were fixed with 4% (v/v) paraformaldehyde, paraffin-embedded and cut into 5μm-thick sections. Antibodies were purchased from Biolegend US and immunofluorescence were performed on the sections. The images were scanned using Axio Scan.Z1 and processed with ZEN 2.3.

Real-Time Quantitative PCR.
To assess the gene expression levels, mice were euthanized for lung tissue collection and the total RNA was extracted using TRIZOL reagent (Invitrogen, USA) and reverse-transcribed into cDNA using the SuperScript II First-Strand Synthesis System Kit (Gen-Star, A215, China). The RT-PCR reactions were conducted using 2 × RealStar Green Fast Mixture (GenStar, A304, China) with 3 technical repeats. GAPDH served as the internal reference gene and data were analyzed using the 2 -△△CT method. The specific primers were as follows: TNF-α (sense,

Statistical Analysis.
All the experiments were performed in at least three independent replicates, and data are presented as means ± SD and visualized using OriginPro or R software. The statistical significance of differences was assessed using two-way ANOVA followed by Tukey's multiple comparisons tests ( * P < 0:05; * * P < 0:01.

ADSCs Mitigates PF by Decreasing ECM Deposition and
Inflammation. In the BLM-induced PF mouse model, intratracheal injection of ADSCs significantly prolonged the survival of the mice in 45 days compared to the BLM + PBS
Similarly, quantitative real-time PCR showed that the proinflammatory cytokines tumor necrosis factor-alpha (TNF-α) and interleukin-1 beta (IL-1β) were significantly downregulated in the BLM + ADSCs group mouse lungs at days 7 and 14, while antiinflammatory cytokines IL-10 and IL-6 showed the opposite trend (Figure 1(f)). These observations demonstrate that ADSC treatment decreased lung ECM deposition, mitigated fibrosis, and inflammation in mouse PF model.

ADSCs Increased the Abundance of antiinflammatory
Trem2 + Macrophages in BLM-Treated Lung. To understand how ADSCs influence the PF environment, we sorted the GFP + and GFPcells from the BLM + ADSCs and BLM + PBS groups lung tissue cell suspensions (Figure 2(a)) and then compared single cell profiles of lung-originated (GFP -) cells between the BLM + PBS and BLM + ADSCs groups (Figure 2(b)), comprising 8181 and 6405 cells, respectively. A total of 27 cell subgroups were identified in the merged dataset of lung-originated cells (Supplementary Figures 1(a)and 1(b)), and we identified 12 major cell types for each group of lung cells using SingleR (Figures 2(b) and 2(c)).
Among the major cell types, granulocytes were the most abundant in the BLM+PBS group, while monocytes/macrophages were the dominant cell type in the BLM+ADSCs group (Figure 2(c)), indicating drastic inflammation-related changes.
Monocyte/macrophages contained 9 clusters with distinct transcriptomic profiles (Figures 3(a) and 3(b)). The Mo/Ma-1 cluster, characterized by marker genes MPG, COL1a1, and SPARC, was the most abundant population. The macrophage-4 cluster with high expression levels of APOE and TREM2 was the second most abundant (Figure 3(b)). They were both predominantly composed of BLM + ADSCs cells, indicating that they were the potential target and/or effector of with ADSC intervention.
We next examined the function of genes with significant changes of expression levels in the Mo/Ma-1 and macrophage-4 subgroups based on REACTOME pathway analysis. Mo/Ma-1 cluster was related to the regulation of the ECM and proteoglycans, nonintegrin membrane-ECM interactions, collagen chain trimerization, collagen degradation, and crosslinking of collagen fibrils (Figure 3(c)). Macrophage-4 cluster showed higher activity in platelet activation, signaling, aggregation and degranulation, MHC II antigen presentation, and regulation of complement cascade.
It was then determined as Trem2 + macrophages, whose abundance was confirmed to increase at day 7 in the BLM + ADSCs lung tissue, as indicated by the stronger fluorescence signals of TREM2 protein (Figure 3(d)). TREM2 expression is positively linked to the severity of lung diseases such as viral infection [27] and chronic obstructive pulmonary disease [28], but it is believed to suppress pro-inflammatory responses [27,29]. The macrophage-4 subgroup therefore might have alleviated the inflammation caused by BLM and ameliorated the subsequent fibrotic response.

ADSCs React to BLM-Treated Lung Environment with
Dampened Metabolic Activity. To investigate the potential therapeutic mechanism of intratracheally injected ADSCs in the BLM-treated inflammatory lung microenvironment (BLM + ADSCs group), we analyzed the single-cell transcriptomic profiles of recollected ADSCs from mouse lungs as well as ADSCs cultured in vitro. A total of 1 × 10 4 GFP + cells from three mice lungs (Figure 2(a)), namely, the recollected ADSCs (ADSC_rec), were sorted. Together with 5 × 10 4 cultured ADSCs (ADSC_cul), they were subjected to scRNA-seq, with libraries consisting of 417 cells and 2801 cells, respectively. Unlike the ADSC_ cul group, the ADSC_rec group does not show much heterogeneity, nor were they similar to any subgroup of ADSC_cul cells (Figures 4(a)-4(c)), which was likely the result of inadaptability of most cultured ADSCs upon contact with the inflammatory environment in BLM-treated lung. Notably, the ADSC_rec group showed high expression levels of ECM-related genes like COL6a1/2/3, TNC, and COL3a1, which were hardly expressed by ADSC_cul cells (Figure 4(d)). We further analyzed the functions of the genes with significant differences between ADSC_rec and ADSC_cul (log2 fold change >0.5 or < -0.5 and adjusted P value <0.05) using REACTOME pathway analysis. The upregulated genes of the recollected ADSCs were enriched to REACTOME pathway terms related to extracellular matrix remodeling and cell motility (Figure 4(e)), while their metabolic activity were downregulated as suggested by REACTOME enrichment, including glucose metabolism and RNA processing. (Figure 4(f)).
Recollected ADSCs showed elevated expression of ECMrelated genes, while interestingly, histological assessments suggested that ADSCs decreased fibrotic ECM deposition in the lung tissue. It is not likely that injected ADSCs, which only take up a small proportion of ADSC-treated lung cells, was the major contributor of lung ECM reconstruction. And this small population of ADSCs was possibly adhering and adapting to the lung tissue through elevated ECM production compared to their cultured state in vitro. In the meantime, the low recollection rate can be explained by the inhibited metabolic activity compared to their cultured counterpart and is another proof that the majority of injected ADSCs cannot permanently reside in the pulmonary tissue, which is unfavorable for the claim that ADSCs transdifferentiate into functional pulmonary cells. Also, the characteristics of recollected subpopulation may provide a hint for improving ADSC survival after injection.
14 Stem Cells International 3.4. ADSCs Crosstalk with Lung-Originated Cells, Especially the Trem2 + Macrophages. Due to the loss of most injected ADSCs, it is possible that they were gradually engulfed by recipient macrophages. To exclude this complication, we confirmed that the recollected ADSCs do not express Ptprc (CD45) or Cd68 and was the sole subgroup that highly expressed Egfp (Figures 5(a) and 5(b)).
To investigate the communication between the injected ADSCs and the lung-originated cells in the BLM-treated lung, we identified the cell-cell interaction with CellChat. The recollected ADSCs mainly received NRG signaling (NRG1-(ITGA-V+ITGB3)) from Mo/Ma-20 subgroup (Figures 5(c) and 5(d)). The NRG1-(ITGAV+ITGB3) binding is reported to be essential for NRG1-ERBB signaling [30], which may enhance ADSC migration and resistance to apoptosis [31].
In the meantime, ADSCs affect multiple lung-originated cell subgroups via paracrine signals (Figures 5(c) and 5(d)). The main receiver of ANGPTL, PERIOSTIN, and PTN was macrophage-4, with ANGPTL4-SDC4, POSTN-(ITGA-V+ITGB5), and PTN-NCL as the top or sole contributor of ligand-receptor pairs, respectively. ANGPTL4 secreted by MSCs inhibits macrophage proinflammatory polarization [32]. POSTN, however, though reported to promote viability, proliferation, and migration of implanted ADSCs in a hind limb ischemia mice model [33], is also a biomarker and promoter of patient idiopathic PF [34]. PTN pathway plays an important role in fetal lung development and promotes proliferation of type II alveolar cells [35], and PTN protein can induce monocyte-endothelial transdifferentiation [36], but as the top target subgroup of PTN pathway, macrophage-4 yet did not express known endothelial biomarkers such as Cd34 and Tie2. It is possible that these macrophages were also receiving other signals which counteracted or dampened the transdifferentiation-inducing effect of PTN signal.
ADSCs and macrophage-4 interacted with granulucytes-2/7, plasmacytoid dendritic cell (DC)-11, and some monocyte/macrophage subgroups via ANXA1-FPR2 pathway while also affecting macrophage-4 and plasmacytoid DC-11 via HGF. Annexin-A1 (Anxa1) counteracts BLMinduced PF [37] and mediates antiinflammatory effects in granulocytic differentiation [38]. HGF has been well studied for its proregenerative role in PF, whose effects include inhibition of epithelial/endothelial cell apoptosis and epithelialto-mesenchymal transition and facilitation of myofibroblast apoptosis [39]. The macrophage-4 subgroup was the major provider and receiver of HGF signal in ADSC-treated lungs. Macrophage-secreted HGF is reported to promote epithelial repair in Crohn's disease patients, and macrophages under the HGF signal will undergo transition into an anti-inflammatory phenotype in skeletal muscle regeneration [40]. Moreover, plasmocytic DC-11 subgroup is another strong receiver of the HGF signal, and DCs are found to be regulated by MSC-secreted HGF to be immunologically tolerated, which is supportive for alleviating acute lung injury [41]. In our case, these paracrine signals secreted by ADSCs may have played a proregenerative and antiinflammatory role in their encounter with cells in BLM-treated lung, increasing the abundance of Trem2 + macrophage-4 subgroup and further affecting other inflammation-related cell types including granulocytes and DCs, which orchestrated towards a less inflammatory environment.
Interestingly, our data also suggest that not everything that ADSCs secret may be beneficial for fibrosis resolution, such as POSTN. This could provide potential targets for future precise engineering of ADSCs to weaken their negative effects while enhancing positive ones. And though further explanations need to be made on how cell-cell communication between the injected ADSCs and the damaged lung contribute to mitigating PF, this study represents the first step towards the direct detection of injected cells and understanding their interaction with recipient cells in cell therapy.

Conclusions
This study is the first to recollect injected ADSCs in mouse PF model and apply scRNA-seq for analysis of interaction between ADSCs and lung-originated cell types. Intratracheally injected ADSCs can alleviate BLM-induced PF, but their number diminishes over time, likely due to dampened metabolic activity, which suggests that ADSCs did not contribute to PF alleviation through transdifferentiation into functional pulmonary cells. Although we only recovered a small fraction of injected ADSCs, and that they did not possess as much heterogeneity as expected, we successfully analyzed the ligand-receptor interactions likely going on in the BLM-treated lung. The remaining ADSCs which were affected by the local NRG signal, through the secretion of effectors like ANGPTL4, HGF, and ANXA1, might have altered macrophages phenotype and influenced DCs to diminish their proinflammatory response. The Trem2 + antiinflammatory macrophage-4 subgroup was significantly more abundant after ADSCs injection in BLM-treated lungs and was also a major target of multiple ADSC-secreted factors, which is likely the key mechanism through which ADSCs alleviated fibrosis. These results indicated that the injection of ADSCs reduced a proinflammative response via activating Trem2 + antiinflammatory macrophages in BLM-induced PF. Our work shows the prospect of utilizing scRNA-seq to achieve a perspective closer to the in vivo interaction events between injected stem cells and the various types of cells in the disease-challenged tissue, as well as a better understanding towards the mechanisms behind MSC's therapeutic effects.

Data Availability
The scRNA-seq data used to support the findings of this study are available from the corresponding author upon request.