Introduction

Ischemic stroke is a clinical emergency caused by sudden interruption of cerebral blood flow to the brain. The only FDA-approved treatment for ischemic stroke is tissue plasminogen activator. This treatment, however, can only help ~ 10% of patients due to a short therapeutic window [1, 2]. Thus, alternative treatments are urgently needed. Recent studies suggest that pericytes may be targeted in the treatment of ischemic stroke.

In addition to their well-characterized functions in angiogenesis [3, 4], vascular integrity [5,6,7], and cerebral blood flow [6, 8,9,10], pericytes also demonstrate stem cell properties after ischemic stroke [11,12,13,14,15]. For example, both mouse and human brain pericytes have been found to differentiate into cells of neural and vascular lineages under ischemic/hypoxic conditions [11]. In addition, PDGFRβ+ cells isolated from ischemic but not non-ischemic mouse brains express both stem cell markers and microglial markers (e.g. Iba1 and CD11b) and differentiate into microglia-like cells with phagocytic capability [12, 13]. Similarly, PDGFRβ+ cells become Iba1+ cells after ischemic injury in a permanent MCAO model [12]. More importantly, human brain pericytes adopt a microglial phenotype and up-regulate microglial markers under hypoxic conditions, and activated pericytes express microglial markers in human stroke brains [15]. In contrast to these reports, one study found that Tbx18+ mural cells maintained their identity and failed to differentiate into other cell types in aging and diverse pathological conditions [14]. It should be noted that ischemic stroke was not one of the pathological conditions examined in this study. Together, these findings suggest that brain pericytes may differentiate into microglia-like cells after ischemic stroke.

One major weakness of these previous studies is that the “pericyte” population contains contaminating cells, including vascular smooth muscle cells (vSMCs) and fibroblasts. Currently, there are no pericyte-specific markers. The most widely used pericyte marker PDGFRβ also labels vSMCs and fibroblasts [16, 17], which exert important functions in injury resolution [18, 19]. Therefore, it remains unclear whether pericytes, vSMCs, or fibroblasts acquire “stemness” after ischemic stroke. Distinguishing these populations has important implications in cell-based therapies for ischemic stroke. Another weakness is that immunohistochemistry rather than lineage-tracing was used in most previous studies. The former method stains cells at a specific stage, while the latter approach labels cells and all their progenies.

To address these issues, we performed lineage-tracing studies using various cell-specific Cre lines, including PDGFRβ-Cre, PDGFRβ-CreERT2, SM22α-Cre, Myh11-CreERT2, and Col1α1-Cre lines. Our results showed that SMAlow/undetectable pericytes and fibroblasts but not vSMCs substantially proliferated at the subacute phase after injury, and that SMAlow/undetectable pericytes rather than vSMCs or fibroblasts differentiated into both microglia-like and macrophage-like cells after ischemic stroke.

Materials and methods

Animals

PDGFRβ-Cre (a gift from Dr. Volkhard Lindner), PDGFRβ-CreERT2 (Jax: 029,684), SM22α-Cre (Jax: 004,746), Myh11-CreERT2 (Jax: 019,079), and Col1α1-Cre (Riken BRC: RBRC05603) mice were crossed with the Ai14 reporter line (JAX:007,914), which contains a floxed STOP sequence before reporter gene tdtomato, to generate Ai14:PDGFRβ-Cre, Ai14:PDGFRβ-CreERT2, Ai14:SM22α-Cre, Ai14:Myh11-CreERT2, and Ai14:Col1α1-Cre mice. Adult (2–3 months) mice of both genders (body weight 20–28 g) and 4–5 animals per group were used in this study. A formal size assessment was not conducted, since this study does not involve treatment. Mice were maintained on a 12/12 h dark/light cycle in the animal facility. All procedures were approved by the Institutional Animal Care and Use Committee (IACUC) in accordance with the National Institutes of Health Guidelines for the Care and Use of Laboratory Animals. The Animal Research: Reporting In Vivo Experiments (ARRIVE) guidelines for reporting experiments involving animals were followed.

Tamoxifen injection

Tamoxifen (T5648; Sigma, St. Louis, MO) dissolved in sunflower oil was injected into Ai14: PDGFRβ-CreERT2 and Ai14:Myh11-CreERT2 mice intraperitoneally at 75 μg/g body weight once a day for three consecutive days. Ten days after the first injection of Tamoxifen, mice were subjected to ischemic stroke.

Middle cerebral artery occlusion (MCAO)

Tamoxifen-injected Ai14:PDGFRβ-Cre, Ai14:PDGFRβ-CreERT2, Ai14:SM22α-Cre, Ai14:Myh11-CreERT2, and Ai14:Col1α1-Cre mice were randomly divided into stroke and sham groups. In stroke group, mice were subjected to 30 min of MCAO produced by transient intraluminal occlusion of the middle cerebral artery using a filament as described previously [20, 21]. Briefly, mice were anesthetized with 2,2,2-tribromoethyl alcohol (250 mg/kg, i.p.). A midline neck incision was made and the common carotid artery (CCA), external carotid artery (ECA), and internal carotid artery (ICA) on the right side were carefully isolated. The ECA and CCA were ligated distal to the carotid bifurcation. The ICA was clipped temporarily. A 6–0 silicone monofilament suture (Doccol) with a 0.21-mm diameter was introduced into the CCA via an incision, advanced 9 mm distal to the carotid bifurcation, and secured in place. Successful occlusion of the middle cerebral artery was confirmed with the PeriCam PSI HR system (Perimed) based on laser speckle contrast analysis technology. Animals showing diminished blood flow of at least 80% during occlusion with at least 75% recovery of blood flow after reperfusion were used for experimentation. In sham group, mice were subjected to the same procedures except MCAO. The body temperature was maintained at 37.0 ± 0.50C during the surgery using a heating pad. Animals had free access to food and water throughout the reperfusion period.

Infarct volume

Serial sectioning was used in this study. Briefly, 20 μm-thick serial sections were cut with Cryostat (Micro HM 550, Thermo Scientific), and eight sections evenly distributed along the rostral-to-caudal axis were collected from each brain. Brain infarct volume was quantified as infarct volume percentage (%) as described previously [22,23,24]. Briefly, cresyl violet-stained brain sections were imaged using the Nikon Eclipse Ti microscope. The areas of the contralateral hemisphere (Ci), ipsilateral hemisphere (Ii), and ipsilateral non-ischemic region (Ni) were determined using the Image J software (NIH), and the infarct volume (%) was calculated as follows:

$$\mathrm{Infarct\; volume\; }\left(\mathrm{\%}\right)=\left(\frac{\sum_{i}\left(\left(\frac{{I}_{i}-{N}_{i}}{I\dot{i}}\right){C}_{i}\right)}{2\sum_{i}{C}_{i}}\right)\times 100.$$

Immunohistochemistry analyses

Immunohistochemical analyses were performed according to standard protocols. Briefly, brain sections were fixed in 4% PFA for 15 min at room temperature and washed in PBS three times. Next, the sections were blocked in blocking buffer (5% normal donkey serum in PBS + 1% BSA + 0.3% Triton X-100) for 2 h at room temperature, followed by incubation with rabbit anti-Ki67 (GeneTex: GTX16667), mouse anti-Caspase-3 (BD Bioscience: 611049), rabbit anti-Iba1 (Wako: 019-19741), rat anti-CD31 (BD Bioscience: 553370), goat anti-ER-TR7 (Santa Cruz Biotechnology: sc-73355), rabbit anti-Collagen-1 (Millipore: AB765P), goat anti-PDGFRα (R&D Systems: AF1062), mouse anti-S100a4 (R&D: MAB4137), rabbit anti-RALDH2 (Sigma: HPA010022), goat anti-mCherry (Scigen: AB0081-200), rabbit anti-PDGFRβ (Cell Signaling Technology: 3169S), mouse anti-smooth muscle actin-α (SMA, MilliporeSigma: F3777), rabbit anti-Olig2 (Novus: NBP1-28667), and rat anti-CD13-FITC (BD Biosciences: 558744) overnight at 4ºC. After extensive washes in PBS, the sections were incubated with appropriate secondary antibodies for 2 h at room temperature. Then, the sections were washed in PBS three times and mounted in Fluoromount-G with DAPI.

Images were taken from both ischemic core and infarct periphery using a Nikon Eclipse Ti microscope or LSM710 confocal microscope. Ischemic core and infarct periphery were determined using a well-established morphological method [25,26,27]. Briefly, ischemic core was identified as the region where the majority of DAPI-stained nuclei were shrunken; whereas infarct periphery was identified as a 400 ~ 500 μm-thick region surrounding ischemic core, where the nuclei had normal morphology. Image processing was performed using ImageJ and Adobe Photoshop.

Image analyses

SMAlow/undetectable pericytes, vSMCs, and fibroblast-like cells were expressed as the mean number of tdtomato+ cells per field as described previously [6]. Similarly, proliferating and apoptotic cells were quantified as the percentages of Ki67+tdtomato+ and Cas3+tdtomato+ cells over total tdtomato+ cells, respectively. SMAlow/undetectable pericyte-, vSMC-, and fibroblast-derived microglia/macrophages were quantified as the percentages of Iba1+tdtomato+ cells over total Iba1+ and/or total tdtomato+ cells. For quantifications, at least three random fields from each section, eight serial sections along the rostral-to-caudal axis for each brain, and five animals were used. All data analyses were performed on z-projection images.

Imaging flow cytometry

Brains were dissected and cut into the contralateral and ipsilateral hemispheres. Single-cell suspension was prepared using a well-established protocol [28]. Briefly, brain tissue was minced with a blade and incubated in 0.2% Collagenase/Dispase (Roche: 11097113001) and 25 μg/ml DNase I (Sigma: D5319) at 37 °C for 40 min with rotation. Tissue suspension was triturated with 1 ml pipette tips and passed through a 100-μm cell strainer (Fisher Scientific, Pittsburgh, PA, USA). Single cells and myelin/debris were collected by centrifugation (1500 rpm, 10 min). Myelin/debris was removed using 22% Percoll (GE Healthcare: 17–0891-02) by centrifugation (1750 rpm, 10 min). Single cells in the pellets were resuspended in RBC lysis buffer and spin down at 1500 rpm for 6 min.

Next, single cells were fixed in 4% PFA for 15 min and washed with sorting buffer. The cells were stained with rat anti-CD45-FITC (Biolegend, 103108, 1:100), rat anti-CD11b-APC (Biolegend, 101212, 1:50), rat anti-CD11b-FITC (Biolegend, 101206, 1:100), mouse anti-CD36-APC (BD Bioscience, 562744, 1:50), rat anti-Ly6G-PE/Cyanine7 (Biolegend, 127618, 1:100), and rabbit anti-TMEM119 (Abcam, 209064, 1:100), followed by Alexa Fluor 405-conjugated anti-rabbit IgG (Abcam, 175652, 1:100) on ice. Samples were analyzed using an ImageStreamX Mark II Imaging Flow Cytometer (Luminex corporation). Single color controls were used to set compensation and FMO controls were used to set gating boundaries. Microglia were gated as CD11b+CD45intLy6G− or CD11b+TMEM119+Ly6G− cells, while macrophages were gated as CD11b+CD45highLy6G− or CD11b+CD36+Ly6G− cells. TdT expression in these populations was determined and quantified. Data analysis was performed using IDEAS software (Millipore).

Statistical analysis

All statistical analyses were performed using the GraphPad Prism 8 software. The non-parametric Mann–Whitney U test was used to determine statistical significance between two independent groups. Significance was set at p < 0.05. Data were presented as mean ± SD.

Results

Validation of Cre lines

Consistent with the single-cell RNAseq study, where Pdgfrβ is expressed in mural cells (pericytes and vSMCs) and fibroblasts (Suppl. Fig. 1a) [17], tdtomato co-localized with the following: (1) mural cell markers (PDGFRβ and CD13) in both large penetrating blood vessels and capillaries and (2) fibroblast markers (PDGFRα, Col1, and ER-TR7) in both meninges and large penetrating blood vessels in Ai14:PDGFRβ-Cre (Fig. 1a) and Ai14:PDGFRβ-CreERT2 brains (Fig. 1b). These findings indicate that pericytes, vSMCs, and fibroblasts are all labeled with tdtomato in these mice.

Fig. 1
figure 1

Validation of Cre lines. a and b Representative images showing the expression of tdtomato (endogenous, red), mural cell markers [PDGFRβ (IHC, green) and CD13 (IHC, green)], fibroblast markers [PDGFRα (IHC, magenta), Col1 (IHC, green), and ER-TR7 (IHC, magenta)] and endothelial cell marker [CD31 (IHC, magenta)] in Ai14:PDGFRβ-Cre (a) and Ai14:PDGFRβ-CreERT2 (b) brains (coronal sections). c and d Representative images showing the expression of tdtomato (endogenous, red), vSMC marker [SMA (IHC, green)] and mural cell marker [PDGFRβ (IHC, magenta)] in Ai14:SM22α-Cre (c) and Ai14:Myh11-CreERT2 (d) brains (coronal sections). e Representative images showing the expression of tdtomato (endogenous, red), fibroblast markers [Col1 (IHC, green), PDGFRα (IHC, magenta), ER-TR7 (IHC, magenta), S100a4 (IHC, green), and RALDH2 (IHC, green)], oligodendrocyte marker [Olig2 (IHC, green)] and endothelial cell marker [CD31 (IHC, magenta)] in Ai14:Col1α1-Cre brains (coronal sections). Tdt tdtomato, cPC Capillary pericyte, M Meninges, PBV Penetrating blood vessel, IHC immunohistochemistry

Full size image

Both SM22α-Cre and Myh11-CreERT2 lines were used to label vSMCs and their progenies. Consistent with previous reports [29, 30], tdtomato merged with vSMC marker SMA and co-localized with PDGFRβ in large but not small vessels in both Ai14:SM22α-Cre (Fig. 1c) and Ai14:Myh11-CreERT2 brains (Fig. 1d), indicating that vSMCs but not SMAlow/undetectable pericytes are labeled with tdtomato in these mice.

Given that Col1α1 mRNA is exclusively expressed in fibroblasts in the CNS (Suppl. Fig. 1b) [17], Col1α1-Cre mice were used to genetically mark fibroblasts. In Ai14:Col1α1-Cre mouse brains, all tdtomato+ cells expressed Col1 and ER-TR7, although not all Col1+ or ER-TR7+ cells expressed tdtomato (Fig. 1e). In addition, tdtomato was also found in PDGFRα+Olig2−, S100a4+, and RALDH2+ fibroblast-like cells (Fig. 1e). Together, these findings suggest that Col1α1-Cre exclusively labels a subpopulation of fibroblasts.

Biphasic change of SMAlow/undetectable pericytes, vSMCs and fibroblasts after ischemic stroke

Next, these mice were subjected to transient MCAO and analyzed at 2 and 7 days post-injury (dpi). All mice developed significant and comparable infarct size (Fig. 2), indicating successful induction of ischemic injury to the same level in these mice. Compared to sham controls, tdtomato+ cells were substantially reduced at the injury site in all mouse lines at 2 dpi (Fig. 3 and Suppl. Fig. 2). Like tdtomato, cellular markers for mural cells (PDGFRβ and CD13), vSMCs (SMA), and fibroblasts (ER-TR7) were also dramatically decreased at the injury site at 2 dpi (Suppl. Fig. 2), suggesting that these perivascular cells are lost probably by cell death at this time point after MCAO. At 7 dpi, tdtomato+ cells were substantially increased at the injury site in Ai14:PDGFRβ-Cre, Ai14:PDGFRβ-CreERT2, and Ai14:Col1α1-Cre mice (Fig. 3). In Ai14:SM22α-Cre and Ai14:Myh11-CreERT2 mice, however, tdtomato+ cells exhibited minimal repopulation at 7 dpi (Fig. 3). These findings suggest that SMAlow/undetectable pericytes and fibroblasts but not vSMCs are the major cell types that substantially increase in number in the subacute phase after ischemic stroke.

Fig. 2
figure 2

Successful induction of ischemic stroke in various reporter mouse lines. a Representative images of cresyl violet-stained coronal brain sections from Ai14:PDGFRβ-Cre, Ai14:PDGFRβ-CreERT2, Ai14:SM22α-Cre, Ai14:Myh11-CreERT2, and Ai14:Col1α1-Cre mice. b Quantification of infarct volume (%) in these mice at day 2 and day 7 after injury. Data are shown as mean ± SD. n = 4. D, day

Full size image
Fig. 3
figure 3

Low-magnification view of PDGFRβ+, SM22α+/Myh11+, and Col1α1+ cells after stroke. a Representative images of tdtomato (endogenous, red) expression in Ai14:PDGFRβ-Cre, Ai14:PDGFRβ-CreERT2, Ai14:SM22α-Cre, Ai14:Myh11-CreERT2, and Ai14:Col1α1-Cre mouse brains (coronal sections) in sham controls and at 2 & 7 days after ischemic stroke. b Quantification of tdtomato+ cells in Ai14:PDGFRβ-Cre, Ai14:PDGFRβ-CreERT2, Ai14:SM22α-Cre, Ai14:Myh11-CreERT2, and Ai14:Col1α1-Cre mouse brains in sham controls and at 2 and 7 days after ischemic stroke. n = 5, **p = 0.0052 and ***p = 0.0006 by Mann–Whitney U test compared to sham controls. Data are shown as mean ± SD. Tdt tdtomato, C ischemic core, P infarct peripheral

Full size image

SMAlow/undetectable pericytes and fibroblasts dramatically proliferate in the subacute phase after ischemic stroke

To investigate if these changes are due to proliferation, co-staining with Ki67 was performed. Although barely observed in sham controls and at 2 dpi, Ki67+ cells were substantially increased at 7 dpi in all mice (Fig. 4a). Comparable numbers of total Ki67+ cells were observed among different mouse lines (Suppl. Fig. 3), again suggesting successful induction of ischemic injury to the same level in these mice. Quantification revealed that 68.9% and 69.6% of tdtomato+ cells expressed Ki67 at 7 dpi in ischemic core and infarct peripheral in Ai14:PDGFRβ-Cre mice, respectively (Fig. 4b). In Ai14:PDGFRβ-CreERT2 mice, 67.9% and 68.1% of tdtomato+ cells were Ki67+ at 7 dpi in ischemic core and infarct peripheral, respectively (Fig. 4b). Similarly, 44.7% and 25.9% of tdtomato+ cells were Ki67+ at 7 dpi in ischemic core and infarct peripheral in Ai14:Col1α1-Cre mice, respectively (Fig. 4b). Interestingly, much less pronounced changes were observed in Ai14:SM22α-Cre and Ai14:Myh11-CreERT2 mice. Specifically, 11.1% and 4.2% of tdtomato+ cells expressed Ki67 at 7 dpi in ischemic core and infarct peripheral in Ai14:SM22α-Cre mice, while 4.3% and 5.4% of tdtomato+ cells were Ki67+ at 7 dpi in ischemic core and infarct peripheral in Ai14:Myh11-CreERT2 mice (Fig. 4b). These findings suggest that SMAlow/undetectable pericytes and fibroblasts but not vSMCs dramatically proliferate at 7 dpi, which contributes to their repopulation in the subacute phase after ischemic stroke.

Fig. 4
figure 4

SMAlow/undetectable pericytes and fibroblasts but not vSMCs dramatically proliferate in the subacute phase after ischemic stroke. a Representative images of tdtomato (endogenous, red) and Ki67 (IHC, green) expression in Ai14:PDGFRβ-Cre, Ai14:PDGFRβ-CreERT2, Ai14:SM22α-Cre, Ai14:Myh11-CreERT2, and Ai14:Col1α1-Cre mouse brains (coronal sections) in sham controls and at 2 & 7 days after ischemic stroke. b Quantification of the percentages of Ki67+tdtomato+ cells over total tdtomato+ cells in Ai14:PDGFRβ-Cre, Ai14:PDGFRβ-CreERT2, Ai14:SM22α-Cre, Ai14:Myh11-CreERT2, and Ai14:Col1α1-Cre mouse brains in sham controls and at 2 and 7 days after ischemic stroke. n = 5, *p = 0.0286 by Mann–Whitney U test compared to sham controls. Data are shown as mean ± SD. Tdt tdtomato, C ischemic core, P infarct peripheral, IHC immunohistochemistry

Full size image

In addition, we also examined apoptosis after ischemic stroke in these mice. Caspase-3 (Cas3) was undetected in sham controls but dramatically induced after ischemic stroke, especially at 2 dpi, in all mouse lines (Suppl. Fig. 4a). Comparable numbers of total Cas3+ cells were observed among different mouse lines (Suppl. Fig. 4b), again suggesting similar level of ischemic injury in these mice. Quantification revealed that the percentage of Cas3+tdtomato+ cells over total tdtomato+ cells is < 3% in Ai14:PDGFRβ-Cre and Ai14:PDGFRβ-CreERT2 mice, < 8% in Ai14:SM22α-Cre mice, < 7% in Ai14:Myh11-CreERT2 mice, and < 7% in Ai14:Col1α1-Cre mice after ischemic stroke (Suppl. Fig. 4c). These results suggest a minimal role of apoptosis in tdtomato+ cell number alteration in these lines after ischemic stroke.

SMAlow/undetectable pericytes differentiate into both microglia- and macrophage-like cells after stroke

To determine which cell type differentiates into microglia/macrophage-like cells after stroke, we first examined Iba1 expression in tdtomato+ cells by immunohistochemistry. Although absent in sham controls or at 2 dpi, Iba1+tdtomato+ cells were observed in Ai14:PDGFRβ-Cre (Fig. 5a–c) and Ai14:PDGFRβ-CreERT2 mice (Fig. 5d–f) at 7 dpi. Quantification showed that 40.4 ~ 64.5% and 52.1 ~ 69.5% of tdtomato+ cells expressed Iba1 at 7 dpi in ischemic core and infarct peripheral, respectively (Fig. 5b, e). In addition, 25.1 ~ 30.3% and 43.5 ~ 45.6% of Iba1+ cells were tdtomato+ at 7 dpi in ischemic core and infarct peripheral, respectively (Fig. 5c, f). In sharp contrast, very few Iba1+tdtomato+ cells were detected in Ai14:SM22α-Cre, Ai14:Myh11-CreERT2, and Ai14:Col1α1-Cre mice after ischemic stroke (Suppl. Fig. 5). These findings indicate that SMAlow/undetectable pericytes rather than vSMCs or fibroblasts differentiate into microglia/macrophage-like cells after ischemic stroke.

Fig. 5
figure 5

SMAlow/undetectable pericytes differentiate into Iba1+ cells after stroke. a and d Representative low- and high-magnification images of tdtomato (endogenous, red) and Iba1 (IHC, green) expression in Ai14:PDGFRβ-Cre (a) and Ai14:PDGFRβ-CreERT2 (d) mouse brains (coronal sections) in sham controls and at 2 & 7 days after ischemic stroke. b and c Quantification of the percentages of Iba1+tdtomato+ cells over total tdtomato+ cells (b) and total Iba1+ cells (c) in Ai14:PDGFRβ-Cre mouse brains in sham controls and at 2 and 7 days after ischemic stroke. e and f Quantification of the percentages of Iba1+tdtomato+ cells over total tdtomato+ cells (e) and total Iba1+ cells (f) in Ai14:PDGFRβ-CreERT2 mouse brains in sham controls and at 2 and 7 days after ischemic stroke. n = 5, *p = 0.0159 and **p = 0.0079 by Mann–Whitney U test compared to sham controls. Data are shown as mean ± SD. Tdt tdtomato, IHC immunohistochemistry

Full size image

To determine if SMAlow/undetectable pericytes differentiate into microglia-like and/or macrophage-like cells after stroke, we performed imaging flow cytometry analysis. First, microglia and macrophages were separated based on CD11b and CD45 expression. We identified tdtomato+ cells in both populations from Ai14:PDGFRβ-Cre and Ai14:PDGFRβ-CreERT2 mice at 7 dpi after ischemic stroke (Fig. 6a, b). Quantification revealed that tdtomato+ cells accounted for 12.1 ~ 12.5% of CD45highCD11b+ macrophages (Fig. 6c) and 10.3 ~ 10.6% of CD45intCD11b+ microglia (Fig. 6d) in these mice at 7 dpi after ischemic stroke. Similar results were observed when microglia were gated based on CD11b and TMEM119, a microglia-specific antibody [31, 32] (Fig. 6a, b). Specifically, 13.8% and 13.4% of CD11b+TMEM119+ microglia were tdtomato+ in Ai14:PDGFRβ-Cre and Ai14:PDGFRβ-CreERT2 mice at 7 dpi after ischemic stroke, respectively (Fig. 6e). In addition, we also gated macrophages based on CD11b and CD36 expression, and found similar results (Fig. 6a, b). Specifically, 17% and 17.3% of CD11b+CD36+ macrophages were tdtomato+ in Ai14:PDGFRβ-Cre and Ai14:PDGFRβ-CreERT2 mice at 7 dpi after ischemic stroke, respectively (Fig. 6f). In sham controls, however, capillary pericytes failed to differentiate into microglia or macrophages in Ai14:PDGFRβ-Cre and Ai14:PDGFRβ-CreERT2 mice (Suppl. Fig. 6). These results suggest that SMAlow/undetectable pericytes differentiate into both microglia-like and macrophage-like cells after ischemic stroke.

Fig. 6
figure 6

SMAlow/undetectable pericytes differentiate into both microglia-like and macrophage-like cells at day 7 after stroke. a Illustration of gating strategies for microglia and macrophages. b Representative single-cell images of tdtomato+ and tdtomato− microglia and macrophages from Ai14:PDGFRβ-Cre and Ai14:PDGFRβ-CreERT2 mouse brains at day 7 after ischemic stroke. c–e Quantification of the percentages of tdtomato-expressing CD45highCD11b+ macrophages (c), CD45intCD11b+ microglia (d), TMEM119+CD11b+ microglia (e), and CD36+CD11b+ macrophages (f) in Ai14:PDGFRβ-Cre and Ai14:PDGFRβ-CreERT2 mouse brains at day 7 after ischemic stroke. Data are shown as mean ± SD. n = 4. Tdt tdtomato

Full size image

SMAlow/undetectable pericyte, vSMC and fibroblast changes show no gender-specific effects

To determine any gender-specific effects, changes in male and female mice were shown separately. We found no differences in infarct volume (Suppl. Fig. 7a), total tdtomato+ cells (Suppl. Fig. 7b), total Ki67+ cells (Suppl. Fig. 7c), the percentage of Ki67+tdtomato+ cells (Suppl. Fig. 7d), total Cas3+ cells (Suppl. Fig. 7e), and the percentage of Cas3+tdtomato+ cells (Suppl. Fig. 7f) between male and females in Ai14:PDGFRβ-Cre, Ai14:PDGFRβ-CreERT2, Ai14:SM22α-Cre, Ai14:Myh11-CreERT2, and Ai14:Col1α1-Cre mice. In addition, comparable percentages of Iba1+tdtomato+ cells were observed between males and females in these mice (Suppl. Fig. 8a–g). Consistent with these results, male and female mice from Ai14:PDGFRβ-Cre and Ai14:PDGFRβ-CreERT2 lines demonstrated similar percentages of tdtomato+ microglia and macrophages in both sham controls and at 7 days after ischemic injury (Suppl. Fig. 8 h–o). Together, these findings suggest no gender-specific effects in the changes of SMAlow/undetectable pericytes, vSMCs and fibroblasts after ischemic stroke.

Discussion

Using lineage-tracing technique, this study investigated the fates of SMAlow/undetectable pericytes, vSMCs, and Col1α1+ fibroblasts after ischemic stroke. We found that SMAlow/undetectable pericytes were dramatically reduced in the acute phase (day 2) after ischemia–reperfusion injury. This observation is consistent with previous reports that ischemia induces pericyte degeneration and loss [9, 33, 34]. In contrast to our data, one study reported increased Rgs5+ pericytes at day 1 after ischemic injury [15]. This controversy is probably due to distinct pericyte markers and different ischemic models. A transient MCAO model that involves both ischemia and reperfusion was used in our study, while a permanent MCAO that involves only ischemic injury was used in the previous study.

Consistent with previous findings that pericytes proliferate in the subacute phase after stroke in both rodents and humans [33, 35], SMAlow/undetectable pericytes showed a high (67.29–71.18%) proliferation rate at day 7 after ischemic stroke. Interestingly, Col1α1+ fibroblasts exhibited a moderate (26.42–45.99%) proliferation rate, while SM22α+/Myh11+ vSMCs showed a low (3.78–11.01%) proliferation rate at day 7 after ischemic stroke. These results suggest that SMAlow/undetectable pericytes and fibroblasts are involved in the pathogenesis of ischemic stroke. In support of this speculation, pericytes have been found to regulate cerebral blood flow, blood–brain barrier integrity, angiogenesis, immunological properties, and scar formation after ischemic stroke [36]. Similarly, Col1α1+ fibroblasts have been shown to regulate stroke recovery via acting as a source of retinoic acid following stroke [37].

Consistent with our finding that Col1α1+ fibroblasts increase in the subacute phase after ischemic stroke, elevated fibroblast number has been reported in various neurological disorders. For example, using Col1α1-GFP mice, it has been shown that fibroblasts expand and form fibrotic scar after ischemia–reperfusion injury and spinal cord injury [37, 38]. Our unpublished results show that Col1α1+ fibroblasts also accumulate in the peri-hematoma region in the subacute phase after hemorrhagic stroke. In addition, using Col1α2-CreERT2 and Col1α1-GFP mice, it has been reported that fibroblast number dramatically increases in the spinal cord after EAE [39]. These findings suggest that fibroblast proliferation/activation may be a common mechanism for various neurological diseases.

Our lineage-tracing and immunohistochemical analyses showed that SMAlow/undetectable pericytes rather than vSMCs or fibroblasts differentiated into microglia/macrophage-like cells after ischemic injury. This is, to our knowledge, the first in vivo evidence that distinguishes brain SMAlow/undetectable pericytes from vSMCs and fibroblasts. Compared to a previous study [15], we found slightly higher percentages of Iba1-expressing SMAlow/undetectable pericytes (40.2–69.3% vs. 44%) and SMAlow/undetectable pericyte-derived Iba1+ cells (25.4–45.1% vs. 16%) at day 7 after stroke. The slightly increased numbers may be explained by different ischemic models. Specifically, a permanent MCAO model was used in the previous study [15], while a transient MCAO model was used in our study. Since the transient MCAO model also involves reperfusion injury, it usually induces more severe inflammation and brain damage [40], which may lead to increased differentiation of SMAlow/undetectable pericytes into microglia/macrophage-like cells. In addition, different transgenic lines may also be responsible for the observed difference. It is possible that Rgs5 marks a subpopulation of pericytes, and that Rgs5− pericytes can also differentiate into Iba1+ cells after ischemic stroke.

Using imaging flow cytometry analysis, we further demonstrated that brain SMAlow/undetectable pericytes differentiated into both microglia-like and macrophage-like cells after ischemic stroke. This is the first study to distinguish SMAlow/undetectable pericyte-derived microglia-like and macrophage-like cells. Interestingly, the percentages of SMAlow/undetectable pericyte-derived microglia/macrophages were lower in imaging flow cytometry analysis compared to immunohistochemistry. This is probably due to different brain volumes used for analysis (infarct peripheral/ischemic core in immunohistochemistry versus the whole ipsilateral hemisphere in imaging flow cytometry). The larger sample volume reduces the percentages of SMAlow/undetectable pericyte-derived microglia/macrophages.

The current study has a few limitations. First, although there is a debate on whether pericytes express SMA, the controversy mainly lies in the definition/location of pericytes (e.g. whether mural cells in pre-capillary arterioles should be called pericytes or vSMCs). In the brain, it is generally believed that pericytes in the middle of the capillary bed express low or undetectable levels of SMA, while pericytes in the arteriole end of the capillary bed are SMA-positive [18, 41,42,43]. These findings clearly show that both SMA+ and SMAlow/undetectable pericytes exist in the brain. In this study, we focus on SMAlow/undetectable pericytes.

Next, fibroblasts are a heterogeneous population with many subtypes. Molecular characterization of these cells is challenging due to the lack of pan/subtype-specific fibroblast markers in the brain. The Col1α1-Cre line was selected in this study because it is specific for brain fibroblasts (it does not label other cells in the brain). Our data showed that all tdtomato+ cells expressed Col1 and ER-TR7 in the Ai14:Col1α1-Cre mouse brains, but not all Col1+ or ER-TR7+ cells expressed tdtomato. This may be explained by the relatively low expression efficiency of the Col1α1 promoter. In this case, not all Col1α1+ fibroblasts are labeled in our study. An alternative explanation for the imperfect colocalization of tdtomato with Col1/ER-TR7 is that tdtomato is located intracellularly, whereas Col1 and ER-TR7 are extracellular matrix (ECM) proteins. Since fibroblasts usually upregulate ECM protein expression after injury, enhanced Col1/ER-TR7 levels do not always mean increased fibroblast number. Therefore, non-ECM markers should be used to mark fibroblasts. Unfortunately, almost all good fibroblast markers are ECM proteins.

Furthermore, 30-min ischemia was induced using size-6.0 monofilament with long coating in this study. These conditions were chosen for two reasons. First, they produce the standard infarct volume of 50 mm3, which is routinely used to evaluate stroke outcomes [44,45,46,47,48,49,50]. Second, they generate reproducible ischemia (> 85% reduction of perfusion in MCA territory) and consistent & significant behavioral outcomes with less mortality in our hands. These conditions, however, induced ischemia in the hippocampus, a region outside the MCA territory. This may be caused by the following three reasons: (1) it has been reported that 90% of C57BL/6 mice display defects in posterior communicating arteries (PComAs) [51], which supply blood to the hippocampus. Since mice used in this study are in C57BL/6 background, the defects in PComAs may contribute to ischemia in the hippocampus during MCAO. (2) the anterior choroidal artery (AChA), which originates from the ICA and supplies blood to the hippocampus, may be blocked by the monofilament with long coating used in this study. (3) a variety of pathophysiological mechanisms, including excitotoxicity, spreading depression, reactive oxygen species, inflammation, and apoptosis, may lead to collateral damage in the regions outside the MCA territory [52]. It should be noted that many studies have reported hippocampal ischemia following MCAO [51, 53,54,55].

In summary, we show that SMAlow/undetectable pericytes and Col1α1+ fibroblasts but not vSMCs substantially proliferate in the subacute phase after ischemic stroke, and that SMAlow/undetectable pericytes rather than vSMCs or Col1α1+ fibroblasts differentiate into both microglia-like and macrophage-like cells after ischemic stroke (Table 1). The underlying molecular mechanisms and the functions of SMAlow/undetectable pericyte-derived microglia/macrophages in stroke pathogenesis will be explored in future research.

Table 1 Proliferation, apoptosis and microglia/macrophage differentiation of tdtomato+ cells in each reporter line at day 7 after stroke
Full size table