Adipose-derived stem cells (ADSCs) can differentiate into vascular lineages and participate in vascular remodeling. Perivascular ADSCs (PV-ADSCs) draw attention because of their unique location. The heterogeneity of subcutaneous (SUB) and abdominal ADSCs were well addressed, but PV-ADSCs’ heterogeneity has not been investigated. In this study, we applied single-cell analysis to compare SUB-ADSCs and PV-ADSCs regarding their subpopulations, functions, and cell fates. We uncovered four subpopulations of PV-ADSCs (Dpp4+, Col4a2+/Icam1+, Clec11a+/Cpe+, and Sult1e1+ cells), among which the Clec11a+ subpopulation potentially participated in and regulated PV-ADSC differentiation toward a smooth muscle cell (SMC) phenotype. Distinct characteristics between PV-ADSCs and SUB-ADSCs were revealed.

Adipose tissue (AT) is initially linked to body weight, lipid storage, and energy metabolism. Study findings indicating AT releases adipokines confirmed AT’s contribution to vascular remodeling (1,2). Mammals have two types of AT, brown and white (3), which are biologically and functionally distinctive (4). However, comparison of AT among different depots is rarely mentioned. Transplantation of perivascular (PV) AT to wire-injured carotid arteries accelerated neointimal hyperplasia, adventitial macrophage infiltration, and adventitial angiogenesis compared with subcutaneous (SUB) AT(1).

Adipose-derived stem cells (ADSCs) were isolated first from subcutaneous AT by Zuk et al. (5) in 2001. ADSCs’ surface markers include SCA-1, PDGFRA, CD34, and CD29. ADSCs could differentiate into vascular lineage and participate in various vascular remodeling model (6). Several groups focused on the difference of SUB-ADSCs and visceral ADSCs regarding cell proliferation (7), differentiation (8), and metabolism (9), demonstrating that ADSCs varied from region to region. Compared with SUB-ADSCs, PV-ADSCs do not appear to have specific cell markers. Therefore, investigating PV-ADSCs at the subpopulation level is urgently needed for ADSCs’ study.

According to traditional theory, smooth muscle cells (SMCs), macrophages, and adventitial progenitor cells contribute to vascular hyperplasia (1012); however, adjacent PV-ADSCs are seldom considered cellular stakeholders (13). Gu et al. (14) demonstrated in a in murine vein graft model that PV-ADSCs could migrate to neointima and differentiate into SMCs. In summary, differentiation from PV-ADSCs to SMCs is a possible participant in vascular remodeling.

Single-cell RNA (scRNA) sequencing enables more detailed classification, differentially expressed genes (DEGs), and functional states of ADSCs at the single-cell level (15). In this study, we compared SUB-ADSCs and PV-ADSCs from the perspective of cell heterogeneity, representative markers, and cell-fate trajectory, providing a more comprehensive profile of PV-ADSC characteristics.

Experimental Animals

All animal experiments were approved by the Animal Ethics Committee of Zhejiang University in accordance with the Guide for the Care and Use of Laboratory Animals. C57BLKS/J mice were purchased from Shanghai Model Organisms Center (catalog no. SCXK (HU) 2017–0010). All animals were fed a chow diet in a 12-h light/dark environment at 25°C.

Isolation of ADSCs

The protocol of single-cell suspension of ADSCs was previously described (14). Briefly, AT was digested by 2 mg/mL collagenase type I (catalog no. 17100017; Gibco) and 1 mg/mL dispase II (catalog no. D4693; Sigma) in Hanks’ balanced salt solution at 37°C for 30–45 min. The digestion was subsequently passed through 100-μm and 70-μm filters, followed by a centrifugation at 300g for 5 min. The stromal vascular fraction (SVF) suspension was mixed with red blood cell lysis buffer for 5 min on ice. The cells were then centrifuged and resuspended in FACS buffer.

FACS

SVF was resuspended in FACS buffer for incubation with the following antibodies for 30 min at 4°C: anti–mouse CD45–FITC (5 μL (0.5 μg)/test; catalog no. AM04501; MultiSciences), Live/Dead Fixable Near-IR (APC/Cy7 channel) Dead Cell Stain Kit (1:1,000; catalog no. L34975; Invitrogen), and Hoechst 33342-DAPI (1:1,000; catalog no. H3570; Invitrogen). The cells were washed three times in FACS buffer and then passed through a BD FACSAria flow cytometer (BD Biosciences). Compensation was made for each single color. The cells without any staining served as blanks.

scRNA Sequencing

The SVF was flow sorted to obtain live, single-cell suspension without debris, red blood cells, dead cells, or CD45+ cells. The sorted cells underwent a standard 10× Genomics’ single-cell procedure performed by Novogene.

scRNA Analysis

Each sample was sequenced, yielding 8,669 and 8,194 cells (Supplementary Fig. 1A and D). The quality control was performed using Seurat package, version 3, in R Studio, version 4.0.0. Cells with expressed gene counts >6,500 or <200 were excluded. Sequencing saturation, mitochondrial gene percentage, and hemoglobin percentage were strictly controlled (Supplementary Fig.1B and E). Seurat, version 3 was used to perform single-cell analysis including quality control, unique molecular identified counting, cell clustering, DEGs analysis, and data visualization. Cluster was defined by dimensionality reduction of t-distributed stochastic neighbor embedding (t-SNE) according to the result of an elbow plot (Supplementary Fig. 1C and F). The Monocle2 R package was applied for pseudo-time analysis.

Statistical Methods

All the results are presented as mean ± SD. Data were put in graphic form with GraphPad Prism, version 9.0, and R Studio, version 4.0. Significance was considered at P < 0.05. Cell DEGs were identified with a minimum log-fold change of 0.25.

Data and Code Availability

Raw sequencing data from scRNA sequencing were uploaded to the GEO database (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE172336). The code or scripts for generating the data are available from the corresponding author on reasonable request.

Single-Cell Atlas of SUB-ADSCs

SUB- or PV AT was isolated with CD45, Live/Dead Fixable Near-IR, and DAPI staining to exclude CD45+, dead cells and erythrocytes. 10× Genomics’ scRNA sequencing was applied (Fig. 1A). For SUB-SVF, 43.7% of the single-cell suspension were live cells, among which 64.7% were CD45− cells (Fig. 1B). Cells from PV-SVF were sorted using similar methods.

SUB-SVF had nine distinct clusters within two ADSC subpopulations (Fig. 1C), according to t-SNE. Both ADSC subpopulations expressed canonical mesenchymal stem cell (MSC) markers such as Ly6a, Cd34, and Pdgfra (Supplementary Fig. 2A). The top 10 DEGs between two ADSC clusters were selected to generate a heatmap (Fig. 1D). The top DEGs among nine clusters were selected to create feature plots (Fig. 1E and F; Supplementary Fig. 2B).

The Dpp4+ subpopulation exclusively expressed Pi16, Anxa3, and Wnt2, which were previously identified as “interstitial progenitor” markers. The Col4a2/Icam1+ subpopulation particularly expressed Fabp4, Apoe, and Lpl, which were previously identified as committed adipocytes or preadipocytes. Our findings were similar with those of Merrick et al. (16) in murine and human SUB-ADSCs and indicate the robustness of the scRNA sequencing technologies and analytical strategies we used.

ScRNA Profiling of PV-ADSCs

The role of PV-ADSCs in vascular remodeling has been realized recently (14,17), but the characteristics of PV-ADSCs remain unclear. The t-SNE map shows 10 clusters with four ADSC subpopulations (Fig. 2A), respectively marked by Dpp4, Col4a2, Clec11a/Cpe, and Sult1e1. All ADSCs clusters expressed canonical MSC markers (Fig. 2B). The top 3 marker genes relative to other subpopulations are shown on feature plots (Fig. 2C) and violin plots (Fig. 2D). Dpp4 (16,18,19), Col4a2 (16,18,20), and Sult1e1 (18) were reported to contribute to adipogenesis. CD142/Clec11a+ cells were defined as adipogenesis-regulatory cells (Aregs) that inhibited adipogenesis. However, inconsistent with previous results from work on SUB-ADSCs, Clec11a did not co-express Cd142 in PV-ADSCs.

SUB-ADSCs and PV-ADSCs Are Distinctive

To compare ADSCs from different depots, the aligned and unaligned t-SNE visualizations of SUB-ADSCs and PV-ADSCs are shown in Fig. 3A and B. Cell ratios for each cluster are displayed (Fig. 3C). PV-ADSCs possessed a large portion of the Clec11a+ subpopulation, whereas SUB-ADSCs did not (Fig. 3B and C). The top 3 DEGs among subpopulations were selected to generate violin plots (Fig. 3DG). The heatmap of the top 10 DEGs in each cluster shows an obvious boundary between two groups of ADSCs (Supplementary Fig. 3A). Gene ontology analysis inferred that PV-ADSC–enriched genes were associated with cell migration (Supplementary Fig. 3B). Our in vitro wound-healing assays (Supplementary Fig. 3G and H) showed a more robust migration in PV-ADSCs.

For the in vivo migration assay, we performed guide-wire injury in murine femoral arteries with the adventitial transplantation of red fluorescent protein–labeled ADSCs. One day after surgery, confocal stack immunofluorescent images showed that more PV-ADSCs (Supplementary Fig. 3J and K) than SUB-ADSCs (Supplementary Fig. 3I and K) migrated into the injured endothelial layer. Last, our in vitro 5-ethynyl-2′-deoxyuridine and CCK-8 assays revealed a less significantly intense cell proliferation in PV-ADSCs compared with the difference in cell migration (Supplementary Fig. 3E and F). The positive results of migration assays could be partly attributed to cell proliferation.

Because the Clec11a+ cluster was the main difference between SUB-ADSCs and PV-ADSCs, the Clec11a+ subpopulation was further analyzed. Gene ontology analysis results suggested that Clec11a+ subpopulation–enriched genes were related to the receptor protein serine/threonine kinase signaling pathway and its ligand TGF-β (Supplementary Fig. 3C). Panther pathway analysis inferred that the TGF-β signaling pathway was among the top 5 signaling pathways in the Clec11a+ subpopulation (Supplementary Fig. 3D). These results suggested that Clec11a+ was relatively unique for PV-ADSCs. Its biological functions might be involved in TGF-β signaling pathways.

PV-ADSCs Could Develop Into an SMC Phenotype Through the Clec11a+ Subpopulation

To elucidate the role of the TGF-β pathway in Clec11a+ subpopulation, we sketched the pseudotime trajectory for PV-ADSCs (Fig. 4A and B). The Clec11a+ subpopulation (Fig. 4C) lay at the end of the pseudotime trajectory along cell differentiation. Pseudotime analysis indicated that smooth muscle–related markers significantly upregulated and reached their peak in the Clec11a+ subpopulation along the trajectory (Fig. 4E) with the decreasing expression of MSC markers (Fig. 4D). In addition, endothelial cell markers also slightly increased in Clec11a+ subpopulation along the trajectory (Fig. 4F). Canonical markers of MSCs (Fig. 4G), fibroblasts (Fig. 4H), SMCs (Fig. 4I), and endothelial cells (Fig. 4J) were plotted along the trajectory for visualization. Given the involvement of the TGF-β signaling pathway and markedly enhanced SMCs markers, we hypothesized that Clec11a+ ADSCs might give rise to smooth muscle–like cells via TGF-β pathways.

In the present study, we analyzed SUB-ADSCs and PV-ADSCs by scRNA sequencing technology. Our approach identified two major subpopulations of SUB-ADSCs, similar to previous findings published by Merrick et al. (16). Burl et al. (20) reported two main clusters of visceral and subcutaneous AT, which also largely coincided with our subpopulations. High similarities among independent studies illustrated the rigor of ADSCs’ biological characteristics and of single-cell sequencing technologies, and analytical strategies.

Few studies have compared ADSCs from different origins. Raajendiran et al. (21) compared the characteristics of ADSCs from omentum, the SUB abdominal region, and SUB gluteofemoral region, which had significant differences in CD34 expression. Burl et al. (20) identified significant depot-selective differences in ADSCs’ extracellular matrix remodeling and adipogenesis abilities between visceral ADSCs and SUB-ADSCs. The comparison between SUB- and PV-ADSCs is absent. Our data suggested that PV-ADSCs uniquely occupied a high percentage of Clec11a+ cells, which drove us to further study the role of the Clec11a subpopulation.

Recently, Clec11a+ cells have received more attention in ADSCs’ study. Using scRNA sequencing, Angueira et al. (22) discovered three fibroblastic subpopulations between SMCs and adipocytes in murine stromal vascular cells of digested thoracic aortas at 3 days after birth), as follows: Pparg+/Pdgfra+ preadipocytes at the periphery of AT lobes, Bace2+/Clec11a+ intermediate cells at the outer wall of aorta, and Pi16+/Sca-1+ progenitor cells at the aortic adventitia. They confirmed that Bace2+/Clec11a+ intermediate cells could not differentiate into adipocytes (22). In SUB-ADSCs, Merrick et al. (16) claimed that Dpp4+ interstitial progenitors could give rise to Icam1+ committed preadipocytes and Cd142+/Clec11a+ Aregs. Analogous to the study of Merrick et al. (16), Schwalie et al. (19) also identified Cd142+ Aregs, which inhibited adipogenic capacities of other stromal cells. Because several studies confirmed that Clec11a+ cells that expressed MSC markers could not differentiate into adipocytes, we hence studied Clec11a cells’ fate.

In PV AT, Clec11a+ cells are approximate to SMCs, expressing lower levels of Acta2 (22). Using dual recombinase-mediated lineage-tracing technology, Tang et al. (23) discovered that aortic adventitial Sca-1+/Pdgfra+ cells, not Sca-1+/Pdgfrb+ cells, differentiated into SMCs. However, they did not distinguish adipocyte progenitor cells and ADSCs, because ADSCs also expressed Sca-1 and Pdgfra. Unlike the finding of Merrick et al. (16) in murine SUB-ADSCs, PV-ADSCs contained a distinct population of Clec11a+ ADSCs, whereas Cd142 broadly was co-expressed with Icam1 and Sult1e1 in the adipogenic cells (Supplementary Fig. 4A). The Clec11a+ subpopulation was prevalent in PV-ADSCs; the ratios of Pdgfra to Pdgfrb are shown in Supplementary Fig. 4B. Pseudotime trajectory revealed that Clec11a was potentially related to the differentiation from ADSCs to smooth muscle–like cells. However, whether PV-ADSCs can differentiate into SMCs requires additional investigation, including Clec11a lineage-tracing mice, which we have already constructed for more study.

In conclusion, we defined a transcriptional profiling of murine PV-ADSCs in detail. ADSCs vary from region to region. Clec11a and Clec11a+ subpopulations in PV-ADSCs were crucial for SMC phenotype development at the transcriptional level. We provide useful information to ADSCs’ study with respect to PV-ADSCs’ role in SMC phenotype development.

Y.X., Y.J., and Y. Lu contributed equally to this work. Y.X. and M.X. supervised the work equally.

This article contains supplementary material online at https://doi.org/10.2337/figshare.16930936.

Acknowledgments. We convey our thanks to Prof. Qingbo Xu for his guidance and strategies regarding the scRNA analysis. Many thanks to Dr. Yanhua Hu for her help with the murine model of femoral artery injury. We also thank Dr. Wenduo Gu for her support on strategies of cell sorting.

Funding. This research was supported by the National Natural Science Foundation of China (grants 81800427 to Y.X. and 81870203 to M.X.).

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

Author Contributions. Y.X. contributed to the study design, acquisition and analysis of the single-cell data set, the in vivo assays, and manuscript writing. Y.J. contributed to acquisition of the single-cell data set, and to the in vitro assays. Y. Lu contributed to single-cell analysis and critical revision of the manuscript. Y.M. and H.N. contributed to in vitro assays. J.S., H.M., and Y. Lin contributed to data analysis, figure layout, and manuscript revision. C.J. contributed to manuscript revision. Y.C. contributed to language polishing. M.X. contributed to study design, experimental instruction, and manuscript revision. Y.X. is the guarantor of this work and takes responsibility for the integrity and accuracy of all the work in this research.

1
Manka
D
,
Chatterjee
TK
,
Stoll
LL
, et al
.
Transplanted perivascular adipose tissue accelerates injury-induced neointimal hyperplasia: role of monocyte chemoattractant protein-1
.
Arterioscler Thromb Vasc Biol
2014
;
34
:
1723
1730
2
Mancio
J
,
Oikonomou
EK
,
Antoniades
C
.
Perivascular adipose tissue and coronary atherosclerosis
.
Heart
2018
;
104
:
1654
1662
3
Lazar
MA
.
Developmental biology. How now, brown fat?
Science
2008
;
321
:
1048
1049
4
Baglioni
S
,
Cantini
G
,
Poli
G
, et al
.
Functional differences in visceral and subcutaneous fat pads originate from differences in the adipose stem cell
.
PLoS One
2012
;
7
:
e36569
5
Zuk
PA
,
Zhu
M
,
Mizuno
H
, et al
.
Multilineage cells from human adipose tissue: implications for cell-based therapies
.
Tissue Eng
2001
;
7
:
211
228
6
Ni
H
,
Zhao
Y
,
Ji
Y
,
Shen
J
,
Xiang
M
,
Xie
Y
.
Adipose-derived stem cells contribute to cardiovascular remodeling
.
Aging (Albany NY)
2019
;
11
:
11756
11769
7
Russo
V
,
Yu
C
,
Belliveau
P
,
Hamilton
A
,
Flynn
LE
.
Comparison of human adipose-derived stem cells isolated from subcutaneous, omental, and intrathoracic adipose tissue depots for regenerative applications
.
Stem Cells Transl Med
2014
;
3
:
206
217
8
Tang
Y
,
Pan
ZY
,
Zou
Y
, et al
.
A comparative assessment of adipose-derived stem cells from subcutaneous and visceral fat as a potential cell source for knee osteoarthritis treatment
.
J Cell Mol Med
2017
;
21
:
2153
2162
9
Baglioni
S
,
Francalanci
M
,
Squecco
R
, et al
.
Characterization of human adult stem-cell populations isolated from visceral and subcutaneous adipose tissue
.
FASEB J
2009
;
23
:
3494
3505
10
Allahverdian
S
,
Chaabane
C
,
Boukais
K
,
Francis
GA
,
Bochaton-Piallat
ML
.
Smooth muscle cell fate and plasticity in atherosclerosis
.
Cardiovasc Res
2018
;
114
:
540
550
11
Tabas
I
,
Bornfeldt
KE
.
Macrophage phenotype and function in different stages of atherosclerosis
.
Circ Res
2016
;
118
:
653
667
12
Zhang
L
,
Issa Bhaloo
S
,
Chen
T
,
Zhou
B
,
Xu
Q
.
Role of resident stem cells in vessel formation and arteriosclerosis
.
Circ Res
2018
;
122
:
1608
1624
13
Minteer
D
,
Marra
KG
,
Rubin
JP
.
Adipose-derived mesenchymal stem cells: biology and potential applications
.
Adv Biochem Eng Biotechnol
2013
;
129
:
59
71
14
Gu
W
,
Nowak
WN
,
Xie
Y
, et al
.
Single-cell RNA-sequencing and metabolomics analyses reveal the contribution of perivascular adipose tissue stem cells to vascular remodeling
.
Arterioscler Thromb Vasc Biol
2019
;
39
:
2049
2066
15
Shapiro
E
,
Biezuner
T
,
Linnarsson
S
.
Single-cell sequencing-based technologies will revolutionize whole-organism science
.
Nat Rev Genet
2013
;
14
:
618
630
16
Merrick
D
,
Sakers
A
,
Irgebay
Z
, et al
.
Identification of a mesenchymal progenitor cell hierarchy in adipose tissue
.
Science
2019
;
364
:
eaav2501
17
Pan
XX
,
Ruan
CC
,
Liu
XY
, et al
.
Perivascular adipose tissue-derived stromal cells contribute to vascular remodeling during aging
.
Aging Cell
2019
;
18
:
e12969
18
Cho
DS
,
Lee
B
,
Doles
JD
.
Refining the adipose progenitor cell landscape in healthy and obese visceral adipose tissue using single-cell gene expression profiling
.
Life Sci Alliance
2019
;
2
:
e201900561
19
Schwalie
PC
,
Dong
H
,
Zachara
M
, et al
.
A stromal cell population that inhibits adipogenesis in mammalian fat depots
.
Nature
2018
;
559
:
103
108
20
Burl
RB
,
Ramseyer
VD
,
Rondini
EA
,
Pique-Regi
R
,
Lee
YH
,
Granneman
JG
.
Deconstructing adipogenesis induced by β3-adrenergic receptor activation with single-cell expression profiling
.
Cell Metab
2018
;
28
:
300
309.e4
21
Raajendiran
A
,
Ooi
G
,
Bayliss
J
, et al
.
Identification of metabolically distinct adipocyte progenitor cells in human adipose tissues
.
Cell Rep
2019
;
27
:
1528
1540.e7
22
Angueira
AR
,
Sakers
AP
,
Holman
CD
, et al
.
Defining the lineage of thermogenic perivascular adipose tissue
.
Nat Metab
2021
;
3
:
469
484
23
Tang
J
,
Wang
H
,
Huang
X
, et al
.
Arterial Sca1+ vascular stem cells generate de novo smooth muscle for artery repair and regeneration
.
Cell Stem Cell
2020
;
26
:
81
96.e4
Readers may use this article as long as the work is properly cited, the use is educational and not for profit, and the work is not altered. More information is available at https://www.diabetesjournals.org/journals/pages/license.