Specialized pericyte subtypes in the pulmonary capillaries

Abstract

Pericytes are essential for capillary stability and homeostasis, with impaired pericyte function linked to diseases like pulmonary arterial hypertension. Investigating pericyte biology has been challenging due to the lack of specific markers, making it difficult to distinguish pericytes from other stromal cells. Using bioinformatic analysis and RNAscope, we identified Higd1b as a unique gene marker for pericytes and subsequently generated a knock-in mouse line, Higd1b-CreERT2, that accurately labels pericytes in the lung and heart. Single-cell RNA sequencing revealed two distinct Higd1b+ pericyte subtypes: while Type 1 pericytes support capillary homeostasis, Type 2 pericytes accumulate in arterioles, and co-express smooth muscle markers and higher levels of vimentin under hypoxic conditions. Lastly, healthy human lung pericytes with upregulation of vimentin exhibited increased adhesion, migration, and higher expression levels of the smooth muscle marker SM22 in vitro. These findings highlight the specialization of pulmonary pericytes and their contribution to vascular remodeling during hypoxia-induced pulmonary hypertension.

Keywords: Capillary; Higd1b; Pericytes; Pulmonary Hypertension; Single-cell RNA Sequence.

Figure 1. HIGD1B is identified as a unique and exclusive marker for human lung PCs.

(A) Dot, Violin, and Density plots show the expression of PC (CSPG4, PDGFRB, and HIGD1B) and SMC markers (ACTA2, CNN1, and TAGLN) across eleven selected cell types from the Human Lung Cell Atlas (HLCA). Note the exclusive expression of HIGD1B in PCs compared to other mural and vascular cells. On the right, a UMAP visualization illustrates PC distributions within lung tissues, using original UMAP coordinates and cell annotations from the ‘HLCA (core)’. The Density plot presents relative expression of selected PC and SMC markers within the PC and SMC populations. In the UMAP, Dot, and Violin plots, PCs were highlighted in light orange and SMCs in dark red. Comprehensive expression data across all annotated cell types were presented in Fig. EV1. (B) UMAP visualization of annotated cell type clusters using the ‘Xenium Human Lung Preview Data (Non-diseased Lung)’, derived from the spatial transcriptomic analysis. Dot and Violin plots show the expression of PC markers (CSPG4, PDGFRB, and HIGD1B) and SMC markers (CNN1 and MYH11) across thirteen annotated cell types. Consistent with the HLCA, HIGD1B had exclusive expression in PCs compared to other mural and vascular cells. ‘FindAllMarkers’ function of the Seurat pipeline was utilized to identify cell type markers for each cluster to annotate cell types. (C) One of three selected areas (yellow box) from the full lung tissue slide used for spatial transcriptomic quantifications. Four images (red boxes #1–4) were selected and inspected under higher magnification. PCs were annotated in yellow, SMCs in dark blue, and other cell types follow the color-code provided in the dot plot in (B). The expression of CSPG4 is shown as red dots (left), PDGFRB as blue dots (middle), and HIGD1B as purple dots (right). Comprehensive quantifications across all inspected areas are presented in Appendix Fig. S7 and Fig. EV2. (D) The table summarizes the PPV, NPV, Sensitivity, Specificity, and Accuracy of HIGD1B, CSPG4, and PDGFRB to distinguish PCs from SMCs in spatial transcriptomic images across all inspected areas (Fig. EV2). PPV positive predictive value, NPV negative predictive value. Comprehensive 2 × 2 contingency tables used for quantifications are presented in Fig. EV2.

Figure 2. Higd1b-CreERT2 knockin mouse line is constructed by CRISPR.

(A) PCLSs were obtained from Cspg4-CreERTM::Ai14 (Ng2-tdT) mice. Scale bar: 20 µm. (B) RNAscope of OCT sections from Ng2-tdT lungs show the coexpression of Higd1b mRNA (green and white) and the tdT protein expression (red). Nuclei were counterstained with DAPI (blue). Scale bar: 20 µm. (C) Schematic figure for Cre-ER insertion and PCR validation of insertion. a: Higd1b locus exon 1–4 and location of two gRNA and sequences in green, PAM sites in red and adjacent nucleotides in black after ATG. b: the Knock-in targeting construct: left (LHA) and right homology arms (RHA) flanked by P2A-CRE-ERT2, c. genomic structure after P2A-CRE-ERT2 knock-in. Primer pair LF + C1 identifies 5’ end, and C4 + RR identifies ERT2 and 3’ end, C2 + C3 Cre gene. Left: Two founder mice, #12 and #25, were identified by Cre-specific PCR using C2 + C3 (lane 3) primers LF is located outside of LHA, C1 is located 5’ of Cre, C4 is located at the 3’ of Cre, and RR is outside of RHA. Source data are available online for this figure.

Figure 3. Higd1b-Cre+ cells specifically label pulmonary and cardiac PCs.

(A) Representative images of lung tissue from Higd1b-tdT+/− mice stained for Sma (white), Cd31 (green), and DAPI (blue). PCs were labeled with endogenous tdT reporter (red). Scale bar: 50 µm. (B) Representative images of PCLSs from Higd1b-tdT+/− normoxic mice stained for Pdgfrβ (green, top) and Ng2 (green, bottom). Nuclei were stained with DAPI (Blue). White boxes indicate magnified areas shown in the right panels. Scale bar: 50 μm. (C) Quantification of tdT+ PCs from the lungs of Higd1b-tdT+/– mice expressing Pdgfrβ and Ng2 (N = 4 per group). Each symbol represents each lung. Each dot represents an individual image analyzed. (D) Representative images of lung tissue from Higd1b-tdT+/− mice stained for macrophages using MhcII (green) and Cd11c (magenta, left panel), neutrophils with Ly6g (green, middle panel) and epithelial cells with Cd326 (green, right panel). Scale bar: 50 μm. (E) Representative images of GFP+ PCs (green) from lung tissue of Higd1b-mTmG+/– mice stained for Sma (white, top panel), Cd31 (white, middle panel), Pdgfrβ (white, bottom panel), and DAPI (blue). Scale bar: 50 µm. (F) A precision cut heart slice from Higd1b-tdT+/− cleared and stained for RFP and Sma (right panels). Scale bar: 500 µm. The right panels show a magnified area where RFP-positive cells are negative for Sma (cyan). Scale bar: 25 µm. Source data are available online for this figure.

Figure 4. Lineage tracing reveals that Higd1b-Cre+ cells originating from the capillaries accumulate in muscularized distal arterioles over various durations of hypoxic exposure.

(A) Left: Images of PCLSs from Higd1b-tdT+/– mice in normoxia show the distribution and pattern of Sma (white) in bronchioles (Br, yellow), arteries (A, green), and veins (V, cyan). Scale bar: 100 μm. Middle: PCLSs from 3 wk Hx Higd1b-tdT+/– mice display Sma staining patterns in bronchioles (yellow), arteries (green), and veins (cyan). Blue arrowheads indicate the location of veins. Scale bar: 500 μm. Right: Represenative PCLSs showing tdT+ PCs (red) surrounding veins and arteries in Higd1b-tdT+/– mice after 3 wks of Hx. DAPI: blue. Scale bar: 50 μm. The bottom right shows the statistical analysis of tdT+ PC locations in the arteries (P = 0.0007), veins (P = 0.156), and capillaries (P = 0.0002) of Higd1b-tdT+/– mice in normoxia (N = 3) and 3 wks of Hx (N = 3). Each dot represents an individual image analyzed. Data is presented as mean ± SD with statistical analysis performed with an unpaired t-test. ***P < 0.001; indicates statistical significance. (B) PCLSs from Higd1b-tdT+/– mice exposed to 1 wk, 2 wk, and 3 wks of Hx show Sma (white) and DAPI (blue) staining. tdT+ PCs (red) accumulated in the distal vasculature after 1 wk of Hx and increased their coverage on arterioles after prolonged exposure to Hx. Yellow arrowheads indicate tdT+ PCs. Magnified areas are shown in adjacent panels. Scale bar: 25 µm. (C) PCLSs from Higd1b-mTmG+/– mice in normoxia and after exposure to 3 wks of Hx. GFP+ PCs on distal arterioles post-Hx. Sma: white. DAPI: blue. White boxes highlight magnified images in adjacent panels. Scale bar: 50 µm. (D) Lineage tracing of the recovery model in Higd1b-tdT+/– (top) and Higd1b-mTmG+/– (bottom) mice exposed to 3 wks of Hx followed by 3 wks of normoxia shows that the majority of reporter cells returned to the parenchyma with only a small number of PCs on the distal arterioles (yellow arrowheads). PCLSs were stained for Sma (white), and DAPI (blue). Scale bar: 50 µm. Source data are available online for this figure.

Figure 5. Two subtypes of PCs are identified using human lung scRNA-seq.

(A) UMAP visualization of five PC sub-populations within the annotated PC cluster from the ‘Human Lung Cell Atlas (core)’ which includes 50 or more annotated cell types. Utilizing Harmony and selecting data from studies with more than 400 PCs, we narrowed the PC sub-populations down to 2992 PCs to minimize the batch effects. In the middle, Donor distribution across PC sub-populations is shown. Data processing steps and the distribution of other potential confounders across the PC sub-populations are provided in Appendix Fig. S15. Dot, Violin and Density plots in the bottom panels illustrate the relative expression of PC markers (CSPG4, PDGFRB, and HIGD1B) across PC subclusters, highlighting the heterogeneity among these sub-clusters. Subcluster 0 and 3 were directly compared to identify the subcluster with the highest HIGD1B expression. (B) Differential expression (DE) analysis utilizing the Wilcoxon rank-sum test was conducted between subcluster 0 (highest HIGD1B expression among the subclusters) and subcluster 2 (lowest HIGD1B expression among the subclusters). The table above shows significantly higher HIGD1B expression and significantly lower PDGFRB expression in subcluster 0 compared to subcluster 2. Gene Set Enrichment Analysis (GSEA) comparing PC sub-Clusters 0 and 2 shows that sub-Cluster 0, marked by significantly higher expression of HIGD1B, is enriched in pathways related to metabolic activity and stress response. In contrast, sub-cluster 2, with a higher expression of PDGFRB, is associated with pathways involved in focal adhesion, cell cycle regulation, and movement. A comprehensive list of enriched pathways with an FDR below 0.1 can be found in Appendix Fig. S16. (C) Human IPAH and control mural and stromal cell subtypes were re-clustered from the previously published data. Sub-clusters 5 and 6 were identified as PC clusters enriched with HIGD1B expression. The Violin plot shows higher HIGD1B expression in sub-Cluster 6 compared to sub-Cluster 5, while PDGFRB is highly expressed in sub-Cluster 5. Pseudotime analysis reveals that Type 1 PCs (sub-Cluster 6, HIGD1B^high PDGFRB^low) are predominantly late lineages (yellow) whereas Type 2 PCs (sub-Cluster 5, HIGD1B^low PDGFRB^high) display a mix of early (purple) and middle (green) lineages. (D) The Violin plot shows VIMENTIN (VIM) expression is increased in IPAH Type 2 PCs from IPAH patients compared to healthy donor Type 2 PCs in sub-Cluster 5. (E) A re-analyzed UMAP plot of mural cells from murine lungs exposed to Hx vs. normoxia identifies PC cluster outlined by a dashed area. The UMAP of sample ID is also included. (F) The Volin plot demonstrates that Vimentin (Vim) expression is elevated in PCs from Hx-mice compared to normoxic PCs.

Figure 6. Type 2 PCs in Higd1b-tdT+/− lungs show upregulated Vimentin expression after exposure to Hx.

(A) High-resolution images show PCLSs obtained from Higd1b-tdT+/− lungs exposed to 3 wks of Hx stained for Sma (white) and DAPI (blue), highlighting the morphological differences between Type 1 and Type 2 PCs. Yellow boxes indicate Type 1 PCs, while white boxes highlight Type 2 PCs within the microvasculature. Panel a’ represents a Type 1 PC in normoxia, characterized by long and thin processes located in the capillaries, while panel a” shows a Type 2 PC near a partial Sma+ arteriole in normoxia. Panel b’ shows a Type 1 PC located in the capillaries in Hx with longer and thicker processes compared to normoxic conditions, while Panel b” shows a Type 2 PC in Hx accumulated and enwrapped around a distal arteriole, coexpressing Sma. Scale bar: 20 µm. (B) PCLSs from Higd1b-tdT+/− mice in normoxia (top panel) and after exposure to 3 wks of Hx (bottom panel) stained for Vim (green), Sma (white), and DAPI (blue). Scale bar: 50 µm. Magnified areas in boxes show Panel a’, depicting a Type 1 PC under normoxia and Panel a” depicting a Type 2 PC in normoxic conditions. Panel b’ highlights a Type 1 PC in Hx without coexpression of Vim, while Panel b” shows a Type 2 PC coexpressing Vim (green) and Sma (white) on an arteriole. Scale bar: 20 µm. (C) The proposed model illustrates the location of PC subtypes and arteriolar SMCs in the pulmonary vasculature under physiological conditions and in the development of Hx induced PH. After Hx, Type 2 PCs upregulate Vim, exhibit increased motility and lineage activity. Italic texts represent gene expression, while non-Italic texts denote protein expression. Source data are available online for this figure.

Figure 7. Overexpression of VIM promotes adhesion, migration, and SM22 expression in healthy human lung PCs after exposure to Hx.

(A) Cell adhesive assays performed on healthy human lung PCs under normoxic and Hx conditions, with and without overexpression of VIM, demonstrate increased PC adhesion in response to Vim. Black boxes highlight areas of increased magnification, showing close contact between VIM overexpression PCs and neighboring cells. Scale bar: 100 µm. The graph on the right quantifies the adhesion of PCs treated with VIM (green) compared to empty vector controls (black). Each dot represents data from three biological replicates, repeated three times. Data is presented as mean ± SEM with statistical analysis performed with one-way ANOVA. ****P < 0.0001 indicates statistical significance. (B) Migration assay indicates that VIM overexpression promotes cell migration of healthy PCs. After six hours, PCs treated with VIM plasmid demonstrated increased cell migration, visualized by F-actin (green) and DAPI (blue). The graph on the right quantifies the migration distance of PCs in normoxic and Hx conditions. Each dot represents an average distance of all PCs in a 100 µm × 200 µm area, using three biological replicates (N = 3 for each group). Data presented as mean ± SEM with statistical analysis performed with one-way ANOVA. **P < 0.01, ****P < 0.0001 indicates statistical significance. (C) Cell culture IF stainings in healthy human lung PCs with and without VIM overexpression for expression of SMC marker (SM22, SMA) in PCs (3G5), along with VIM (red). DAPI (blue). Graphs on the right display the percent of cells acquired from six images coexpressing VIM (top) and SM22 (bottom). Each experiment was repeated three times with three biological replicates. Data is presented as mean ± SEM with statistical analysis performed using one-way ANOVA. **P < 0.01, ****P < 0.0001 indicates statistical significance. Source data are available online for this figure.

Figure EV1. HIGD1B is expressed in human lung PCs.

(A) UMAP visualization of cell populations within human lung tissue using the original UMAP coordinates and cell type annotations from the ‘Human Lung Cell Atlas (core)’. Cell type annotations are color-coded, as shown in the dot plot. PC and SMC distributions within lung tissues, using original UMAP coordinates and cell annotations from the ‘HLCA (core)’ are highlighted in the bottom left. In UMAP, Dot, Violin and Heatmap plots, PCs were highlighted in light orange and SMCs in dark red. The Dot and Violin plots on the right illustrate the expression pattern of PC (CSPG4, PDGFRB, and HIGD1B) and SMC markers (ACTA2, CNN1, and TAGLN) across all annotated cell types. The Dot plot highlights the relative expression of markers, while the Violin plot depicts the distribution and intensity of expression within each cell type. (B) Differential expression (DE) analysis utilizing the Wilcoxon rank-sum test shows the expression profiles of PC markers (CSPG4, PDGFRB, and HIGD1B) and SMC markers (ACTA2, CNN1, and TAGLN) in annotated PCs and SMCs compared to all other cell populations within the ‘Human Lung Cell Atlas (core)’.

Figure EV2. Statistical analysis of spatial transcriptomic figures are collected from non-diseased lung tissue.

(A) The quantification of mural cell markers was performed using four selected areas from the spatial transcriptomic figure, derived from the three yellow boxes in Appendix Fig. S7. In these analyses, PCs are represented by yellow-colored cells, while SMCs are indicated by dark blue-colored cells. The expression of specific markers is depicted as follows: CSPG4 is represented by red dots (left), PDGFRB by blue dots (middle), and HIGD1B by purple dots (right). (B) The analysis results are summarized in 2 × 2 tables that display the presence of CSPG4, PDGFRB, and HIGD1B across cell populations. Detailed quantification processes are described in the Methods section. Each table is followed by the corresponding results of the analysis. (C) The bottom tables summarize the performance of each marker to distinguish PCs or SMCs from other cells (PC vs all other cells, SMCs vs all other cells, and PCs vs SMCs). TP true positive, FP false positive, TN true negative, PPV positive predictive value, NPV negative predictive value.

Figure EV3. tdT cells from Higd1b-tdT+/− label PCs in other organs in vivo.

(A) Stainings of brain, skeletal muscle, pancreas, heart, intestine, and connective tissue around the descending aorta from Higd1b-tdT+/− mice for Cd31 (green), Sma (white), and DAPI (blue). tdT endogenous reporter labeled PCs are in red. Note the presence of PCs found within multiple organ systems. Scale bar: 50 µm. (B) The kidney (top two panels) and retina (bottom panel) from Higd1b-tdT+/− mice were stained for Cd31 (green), Sma (white), and DAPI (blue). tdT+ PCs are in red. Scale bar: 50 µm. (C) Liver (top panel) and spleen (bottom panel) from Higd1b-tdT+/− mice stained for Cd31 (green), Sma (white), and DAPI (nuclei: blue) reveal an absence of tdT+ PCs (red) in both organs. Scale bar: 50 µm. (D) Quantification of tdT+ cells in various organs from Higd1b-tdT+/− mice in each image field inspected. Image area inspected = 429.5 μm². N = 3 for all organs except heart tissues (N = 2). Each dot represents an individual image analyzed. Data presented as mean ± SEM. (E) tdT+ cells from the kidney taken from Higd1b-tdT+/− mice (left) alongside binary immunofluorescence (middle). Scale bar: 50 µm. Quantification of tdT+ cells from the whole cell population identified with DAPI staining is displayed. N = 2, with each dot represents an image taken by an individual animal. (F) Percentage of tdT+, Sma+ cells per image field from the retina and kidney obtained from Higd1b-tdT+/− mice. Inspected image area = 429.5 μm². N = 3 for each group, with each dot representing an individual image analyzed. Data presented as mean ± SEM.

Figure EV4. Lineage tracing shows that tdT+ cells accumulate in muscularized distal arterioles by different hypoxic exposure times.

(A) tdT+ PCs from Higd1b-tdT+/− mice showing Sma negative tdT+ PCs(red) located in the parenchymal region and distal arterioles after 1, 2, and 3 wks of normoxia. Cd31 in green and Sma in white and DAPI in blue. (B) Representative images of PCLSs from Higd1b-tdT+/− mice show tdT+ PCs (red) on distal arterioles (Cd31, green) coexpressing Sma (white) after exposure to 1, 2, and 3 wks of Hx. Yellow arrowheads highlight the accumulation of PCs on remodeled distal arterioles that coexpress Sma after Hx exposure. Scale bar: 50 μm. (C) Accumulation of tdT+ PCs (red) in muscularized distal arterioles, with staining for SMC markers Smmhc (green) and Sma (white) after 1, 2, and 3 wks of Hx. DAPI: blue. Scale bar: 50 µm. Source data are available online for this figure.

Figure EV5. Chronic Hx results in upregulation of Vim in Type 2 PCs accumulated on distal arterioles.

PCLSs from three Higd1b-tdT+/− mice demonstrate the upregulation of Vim (green) in Type 2 PCs (red) accumulated on distal arterioles. Yellow arrowheads highlight the morphological changes of Type 2 PCs and the coexpression of Vim. Sma (white). DAPI (blue). Scale bar: 50 μm. The bottom figure shows the quantification of tdT+ cells coexpressing Vimentin in normoxic and Hx Higd1b-tdT+/− mice. N = 3 for each group. Each dot represents an individual image analyzed. Data is presented as mean ± SEM. Statistical analysis is performed with an unpaired t-test. ****P < 0.0001 indicates statistical significance. Source data are available online for this figure.