Distinct Fibroblast Lineages Give Rise to NG2+ Pericyte Populations in Mouse Skin Development and Repair

Abstract

We have examined the developmental origins of Ng2+ perivascular cell populations that adhere to the basement membrane of blood vessels, and their contribution to wound healing. Neural/glial antigen 2 (Ng2) labeled most perivascular cells (70-80%) in developing and adult mouse back skin, a higher proportion than expressed by other pericyte markers Tbx18, Nestin and Pdgfrβ. In adult mouse back skin Ng2+ perivascular cells could be categorized into 4 populations based on whether they expressed Pdgfrα and Pdgfrβ individually or in combination or were Pdgfr-negative. Lineage tracing demonstrated that although Ng2+ cells in embryonic and neonatal back skin contributed to multiple cell types they did not give rise to interfollicular fibroblasts within the dermis. Lineage tracing of distinct fibroblast populations during skin development showed that papillary fibroblasts (Lrig1+) gave rise to Ng2+ perivascular cells in the upper dermis, whilst Ng2+ perivascular cells in the lower dermis were primarily derived from reticular Dlk1+ fibroblasts. Following wounding of adult skin, Ng2+ dermal cells only give rise to Ng2+ blood vessel associated cells and did not contribute to other fibroblast lineages. The relative abundance of Ng2+ Pdgfrβ+ perivascular populations was comparable in wounded and non-wounded skin, indicating that perivascular heterogeneity was maintained during full thickness skin repair. In the wound bed Ng2+ perivascular populations were primarily derived from Lrig1+ papillary or Dlk1+ reticular fibroblast lineages, according to the location of the regenerating blood vessels. We conclude that Ng2+ perivascular cells represent a heterogeneous lineage restricted population that is primarily recruited from the papillary or reticular fibroblast lineages during tissue regeneration.

Keywords: blood vessels; dermis; fibroblast; lineage tracing; pericyte; skin; wound healing.

FIGURE 1

Expression of Tbx18, Nestin, Pdgfrβ, and Ng2 by perivascular cells in developing and adult skin. Sections of mouse back skin collected at the times shown were co-labelled with antibodies to Cd31 (green) and individual pericyte markers (red) and counterstained with DAPI (blue). White arrows highlight the presence of perivascular cells. Representative images of each time point are shown. The percentage of Cd31 positive blood vessels with each pericyte marker was calculated. (A,B) Tbx18, (C,D) Nestin, (E,F) Pdgfrβ, (G,H) Ng2. Error bars represent ±SD. *0.0152. N=3 for each marker and developmental stage. Scale bars 50 μm.

FIGURE 2

A population of perivascular cells expresses Ng2 and Pdgfrβ. (A) Flow cytometry of Ng2DsRed adult mouse back dermal cells labeled with Pdgfrβ at P3, P21, and P50. (B) The percentage of live dermal cells co-expressing Pdgfrβ and Ng2 at P3, P21, and P50. (C–E) Immunostaining of Ng2DsRed mouse back skin with anti-Pdgfrβ (green) and Cd31 (grey) at P3 (C), P21 (D), and P50 (E) showing significant co-expression of both markers (white arrows) by pericyte populations. N = 3. (F) Graph depicting the percentage of Cd31 positive blood vessel associated cells expressing both Ng2 and Pdgfrβ. Scale bars 50 μm. Flow cytometry N = 4. Statistical test: One-way ANOVA. Error bars represent ±SD.

FIGURE 3

Expression of Ng2 and Pdgfrα in dermal cell populations reveals a significant population of Ng2+Pdgfrα+ perivascular cells. (A) Pool of 4 flow cytometry plots of PdgfrαH2BGFP; Ng2DsRed double transgenic adult mouse back dermal cells. Approximately 6% of dermal cells expressed both Ng2 and Pdgfrα. (B) The percentage of live dermal cells expressing Pdgfrα and Ng2 individually or in (++). (C) Graph depicting the percentage of cells expressing Cd24, Sca1, and Cd26 within Ng2+ Pdgfrα+, Ng2+ Pdgfrα−, or Ng2− Pdgfrα+ dermal populations corresponding to (D–F), respectively. (D–F) Flow cytometry panels showing Ng2 and Pdgfrα expressing dermal cell populations (D) Ng2+ Pdgfra+ (red), (E) Ng2+ Pdgfra- (blue), (F) Ng2- Pdgfra+ (green) which co-expressed Cd24, Sca1, or Cd26. (G,H) Immunostaining of PdgfrαH2BGFP; Ng2DsRed double transgenic mice at P21 (G) and P50 (H) identifying a proportion of perivascular cells expressing both Ng2 and Pdgfrα (white arrows). Representative images shown. (I) Graph depicting the percentage of blood vessel associated cells which co-expressed Ng2 and Pdgfrα. *P≤0.05, **P≤0.01, ***P≤0.001, and ****P≤0.0001. N = 5 for each flow cytometry experiment and immunostaining experiment. Scale bars 100 μm. Error bars represent ±SD. Statistical test: One-way or two-way ANOVA.

FIGURE 4

Four distinct Ng2 perivascular populations are present in adult mouse back skin based on Pdgfrα and/or Pdgfrβ co-expression. (A) Pool of 3 flow cytometry plots of PdgfrαH2BGFP; Ng2DsRed double transgenic adult mouse back dermal cells at P21. Live dermal cells co-expressing Ng2 and Pdgfrα were assessed for Pdgfrβ expression. (B) Immunofluorescence of a dermal blood vessel from PdgfrαH2BGFP; Ng2DsRed double transgenic mouse labelled with anti-Pdgfrβ (gray). High magnification images revealed the presence of 4 Ng2 perivascular populations. Ng2+ Pdgfrα– Pdgfrβ– cells can be seen in all high magnification panels. Arrows show I, II: Ng2+ Pdgfrα+ Pdgfrβ+, III: Ng2+ Pdgfrα- Pdgfrβ+, IV: Ng2+ Pdgfrα+ Pdgfrβ– Representative images shown. (C) Graph depicting the percentage of Ng2+ perivascular cells which are Pdgfr-, Pdgfrα– Pdgfrβ+, Pdgfrα+ Pdgfrβ−, and Pdgfrα+ Pdgfrβ+ at P21. (D) Schematic illustrating the percentage contributions of each Ng2+ perivascular subpopulation to the total Ng2+ perivascular population present at P21. (E) Graph depicting the percentage of Ng2+ perivascular cells which express Pdgfrα– Pdgfrβ+, Pdgfrα– Pdgfrβ+ and Pdgfrα+ Pdgfrβ+ in papillary and reticular layers of the dermis at P21. *P≤0.05, **P≤0.01, ***P≤0.001, and ****P≤0.0001. N = 3 Scale bars 100 μm. Error bars represent ±SD. Statistical analysis conducted using One-way ANOVA or Multiple T-Tests using the Holm-Sidak method.

FIGURE 5

During skin development Ng2+ dermal cells contribute to Pdgfrα/β perivascular populations. (A,B) Ng2 DsRed mouse back skin at E12.5 labeled with Pdgfrα (A) or Pdgfrβ (B). (C) Graph depicting the percentage of Ng2+ perivascular cells expressing Pdgfrα/β at E12.5. (D) Ng2 DsRed mouse back skin at E18.5 labeled with Pdgfrα (green) and Cd31 (gray). Panels on the righthand side are higher magnifications of papillary (I) and reticular (II) blood vessels. (E) The percentage of Cd31+ blood vessels in the papillary and reticular layers of the dermis which express Ng2 Pdgfrα populations. (F) Ng2 DsRed back skin at E18.5 labeled with Pdgfrβ (green) and Cd31 (gray). Panels on the righthand side are higher magnifications of papillary (I) and reticular (II) blood vessels. (G) The percentage of Cd31+ papillary and reticular blood vessels with Ng2 Pdgfrβ populations. (H) Ng2CreER labeling strategy and timing of Tamoxifen labeling (T) and subsequent tissue isolation (I). Panels corresponding to labeling and isolation are highlighted. (I,K) Immunostaining of Ng2CreER; tdTomato mice with anti-Cd31 (gray) and DAPI nuclear counterstain (blue) labeled at E18.5 and isolated at P2 (I) and P28 (J). Ng2 tdTomato expression by perivascular cells are highlighted with white arrows. (K) Graph depicting the percentage of tdTomato labeled blood vessel associated cells at P2 and P28. (L,M) Immunostaining of Ng2CreER; tdTomato mice induced at E18.5 and isolated at P28 labeled with Pdgfrα (L) or Pdgfrβ (M) and Cd31. (N,O) Graphs depicting the percentage of Ng2 tdTomato labeled perivascular cells which co-express Pdgfrα (N) or Pdgfrβ (O) in papillary and reticular layers of the dermis. Statistical test: Paired T-Test. Error bars represent ±SD. **P≤0.01, ***P≤0.001, and ****P≤0.0001. N = 3 Scale bars 50 μm. Representative images of the immunostaining are shown.

FIGURE 6

Papillary and reticular fibroblast lineages contribute to Ng2+ perivascular populations. (A) Ng2 DsRed mouse back skin at E18.5 labeled with Lrig1 (green) and Cd31 (gray). Higher magnified panels on the righthand side show papillary (I) and reticular blood vessels (II). (B) The percentage of Cd31+ blood vessels with Ng2 Lrig1 positive populations. (C) Lrig1 lineage tracing mouse lines and labeling strategy. (D) Representative immunostaining of Lrig1CreER; tdTomato P21 back skin labeled with Ng2 (green). White arrows highlight Lrig1Cre Ng2+ cells on the blood vessels. (E) Representative images of Ng2 DsRed mouse back skin at E18.5 labeled with Dlk1 (green) and Cd31 (gray). Higher magnification panels on the right-hand side show papillary (I) and reticular (II) blood vessels. (F) The percentage of Cd31+ blood vessels in the papillary or reticular layer of the dermis with Ng2 and Dlk1-positive cell populations. (G) Dlk1 lineage tracing: mouse lines and labeling strategy. (H) Representative immunostaining from Dlk1CreER; tdTomato back skin at P21 labeled with Ng2. White arrows highlight Dlk1Cre Ng2+ cells on the blood vessels. (I) Schematic demonstrating that in adult mouse back skin Lrig1 and Dlk1 fibroblast lineages contribute to Ng2+ perivascular populations in their respective dermal layers. **P≤0.01 and ****P≤0.0001. Scale bars 50 μm. Paired T-test. Error bars represent ±SD.

FIGURE 7

Ng2CreER labelled perivascular cells only contribute to the regeneration of blood vessel associated cells during wounding. (A) Ng2CreER^t; tdTomato mouse line with labeling and wounding strategy. (B–G) Representative immunostaining of Ng2CreER^t; tdTomato adult mouse back skin with Tamoxifen induction at E18.5, tracing Ng2+ lineage expression to 4 (B), 7 (D), and 10 (F) days post wounding, with Cd31 labeled blood vessels (gray) and DAPI nuclear counterstain (blue). Panels on the right-hand side are higher magnification images of boxed regions on the left-hand side. White arrows indicate Ng2 lineage positive blood vessel associated cells. (C,E,G) Graphs depict the percentage of Cd31 positive blood vessels within the wound bed and outside the wound bed, with Ng2CreER^t tdTomato labeled cells. (H) Representative immunostaining of Ng2CreER^t; tdTomato adult mouse back skin with tamoxifen induction at E18.5 and isolation at 10 days post wounding labeled with Lrig1 (green) and Cd31 (gray). The higher magnification panel shows the presence of one Ng2 tdTomato+ Lrig1+ perivascular cell. (I) Graph depicting the percentage of Cd31 positive blood vessels inside and outside the wound bed with Ng2CreER^t tdTomato labeled perivascular cells expressing Lrig1. (J) Representative immunostaining of Ng2CreER^t; tdTomato adult mouse back skin with tamoxifen induction at E18.5 isolated at 10 days post wounding labeled with Pdgfrβ (green) and Cd31 (gray). The higher magnification panel includes white arrows highlighting Ng2CreER^t tdTomato+ Pdgfrβ+ perivascular populations. (K) Graph depicting the percentage of Cd31 positive blood vessels with Ng2 tdTomato labeled Pdgfrβ positive perivascular cells present in papillary and reticular layers of the dermis inside and outside the wound bed. Images are representative of N = 3 mice per time point. Scale bars 100 or 50 μm (higher magnification views). Statistical test: Paired T-test. Error bars represent ±SD.

FIGURE 8

Papillary and reticular fibroblast lineages give rise to Ng2+ perivascular populations during wound healing. (A) Lrig1CreER; tdTomato mouse line with labeling and wounding strategy. (B) Representative immunofluorescence images of skin from Lrig1CreER; tdTomato mice treated with tamoxifen at E18.5 and isolated at 10 days post wounding labeled with Cd31 (green), Ng2 (gray) and DAPI nuclear counterstain (blue). The panel on the right-hand side is a higher magnification image of part of boxed region on the left-hand side. White arrows indicate Lrig1CreER tdTomato blood vessel resident perivascular cells expressing Ng2. (C) The percentage of Cd31 positive blood vessels within and outside the wound bed with Lrig1 tdTomato positive Ng2 expressing cells at 10 days post wounding. (D) Graph showing no significant difference in the percentage of Cd31 positive blood vessels with Lrig1CreER tdTomato and Ng2 expression in papillary and reticular layers of the regenerating wound bed dermis. (E) Dlk1CreER; tdTomato mouse line with labeling and wounding strategy. (F) Representative immunofluorescence images of Dlk1CreER; tdTomato lineage traced mouse back skin treated at E18.5 and isolated at 10 days post wounding labeled with Cd31 (green), Ng2 (gray) and DAPI nuclear counterstain (blue). The panel located on the right-hand side is a higher magnification image of part of the boxed region on the left-hand side. White arrows indicate Dlk1CreER tdTomato blood vessel resident cells expressing Ng2. (G) Graph depicting the percentage of blood vessels within and outside the wound bed with Dlk1 tdTomato Ng2 expressing cells. (H) Graph showing significantly higher percentage of Cd31 positive blood vessels with Dlk1 tdTomato and Ng2 expression in the reticular layer of the regenerating wound bed dermis when compared to the papillary layer. Statistical analysis was paired T-test. Error bars represent ±SD. N = 3 mice (3 sections per mouse) were quantified. Scale bars 100 or 50 μm (higher magnification views).

FIGURE 9

Schematic summary showing origins of Ng2+ skin pericytes in unwounded and wounded skin. In unwounded skin, Lrig1 (green), and Ng2 (gray) lineages, labeled at E18.5, give rise to Ng2+ blood vessel resident pericytes in the papillary layer of the dermis whilst Dlk1 (red) and Ng2 lineages give rise to Ng2+ blood vessel resident pericytes in the reticular layer of the dermis. These NG2+ pericytes can differ in their Pdgfr status. In unwounded skin Lrig1 and Dlk1 fibroblast lineages contribute to Ng2+ pericyte populations in a spatially restricted manner. However, in wounded skin, the Lrig1 and NG2 lineages can contribute to pericytes in both layers of the dermis whilst the contribution of the Dlk1 lineage remains spatially restricted to the reticular layer.