Background
Pericytes are vascular mural cells embedded within the basement membrane of capillaries originally discovered by Rouget in 1873 [
1]. In the central nervous system (CNS), pericytes are widely believed to be integral, multifunctional members of the neurovascular unit at the capillary level [
2‐
5]. Pericytes are seen to ensheathe microvascular endothelial cells forming multiple synapse-like "peg-socket" contacts with adjacent endothelial cells in brain capillaries suggesting the possibility of tightly regulated signaling and functional coupling between these two cell-types [
4,
6,
7]. Although it has been known that brain capillaries have much greater pericyte coverage than peripheral vascular beds, the presence and functional responsibilities of CNS pericytes have largely been neglected until the past two decades [
6,
7].
Much of the recently gained insight into pericyte biology arose from the analysis of pericyte deficient transgenic mice with disrupted platelet derived growth factor B (PDGF-B)/platelet derived growth factor receptor beta (PDGFRβ) signaling [
8‐
13]. During development of brain capillaries PDGFRβ is exclusively expressed in perivascular pericytes [
10,
14]. In the embryonic neural tube, endothelial-secreted PDGF-B binds to the PDGFRβ receptor located on the pericyte plasma membrane resulting in dimerization of PDGFRβ, subsequent autophosphorylation of cytoplasmic tyrosine residues and binding of SH2 domain containing proteins which in turn initiate a multitude of signal transduction pathways ultimately stimulating the proliferation, migration, and recruitment of pericytes to the vascular wall of newly formed blood vessels [
10,
14,
15]. Complete knockout of
Pdgfb or
Pdgfrβ results in a complete lack of pericytes and embryonic lethality [
8,
9]. Normal PDGF-B/PDGFRβ interactions and corresponding pericyte recruitment have been demonstrated to play a pivotal role in the regulation of the cerebral microcirculation, including regulating angiogenesis, vascular stability, and integrity of the blood-brain barrier during embryonic development [
2,
7,
16]. Although it has been speculated that brain pericytes might fulfill similar roles in the adult brain, there is limited
in vivo experimental evidence to support such claims. Therefore, the functional roles of brain pericytes in the adult and aging brain are relatively less well understood in part due to a lack of adequate and/or properly characterized experimental models.
To address whether genetic disruption of PDGFRβ signaling would result in a pericyte-specific insult in adult mice and may therefore be used to study the roles of brain pericytes in the adult and aging brain, we sought to characterize the pattern of expression of PDGFRβ in the adult mouse brain in both control mice and viable F7 mice with two hypomorphic
Pdgfrβ alleles on a 129S1/SvlmJ genetic background [
12]. The F7 mice were generated by seven point mutations in which multiple cytoplasmic tyrosine residues at positions 578, 715, 739, 750, 770 and 1008 were substituted with phenylalanine and the tyrosine at position 1020 was replaced with isoleucine. As a result of these mutations, Src-, Grb2-, PI3K-, RasGAP-, SHP-2 and PLC
γ-dependent downstream signaling of the PDGFRβ receptor is disrupted. Impaired PDGFRβ downstream signaling in turn results in diminished pericyte recruitment to the vessel wall leading to a 55-75% reduction in the number of pericytes in the embryonic CNS at day E14.5 as previously shown [
12]. However, the effects of such alterations in PDGFRβ signaling in the adult mouse brain remain unknown.
Through dual fluorescent in situ hybridization (FISH), triple immunofluorescent staining for different cell types in the neurovascular unit, and a fluorescent in situ proximity ligation assay (PLA) to visualize molecular PDGF-B/PDGFRβ interactions on brain tissue sections, we show for the first time that PDGFRβ is exclusively expressed in pericytes in several brain regions in the adult control 129S1/SvlmJ mouse model, and confirm a similar pattern of pericyte-specific PDGFRβ expression in the adult brain in F7 mutants on the same genetic background. We demonstrate that F7 mutants are viable, but exhibit substantial pericyte reductions in different brain regions offering a prospective model system appropriate for examining the role of pericytes in the adult and aging brain.
Discussion
As our results demonstrate, PDGFRβ is exclusively expressed in brain pericytes and not in neurons, astrocytes or endothelial cells in adult, viable mouse models with normal and deficient PDGFRβ signaling, such as the F7 mutants, on a 129S1/SvlmJ genetic background. Significantly, this pericyte-specific PDGFRβ expression is observed at both the mRNA and protein level in multiple brain regions. Furthermore, PDGFRβ expression is consistently restricted to brain pericytes with use of multiple fixatives and tissue processing techniques during immunofluorescent staining (see
Methods). Previous studies utilizing mice on a 129Sv/C57BL6 genetic background have also found a similar pericyte-restricted pattern of PDGFRβ expression in mouse brain capillaries during embryonic development [
10,
14]. Moreover, gene expression profiling on acutely isolated neurons from C57BL6 mice and S100β (S100 calcium-binding protein B) transgenic mice expressing enhanced green fluorescent protein on a hybrid C57BL6/DBA background failed to detect expression of
Pdgfrβ in neuronal cell populations [
27,
28].
Intriguingly, recent
in vitro studies utilizing primary cortical neurons isolated from C57BL6 mice during postnatal day 1 and 6 day old cultures have claimed PDGFRβ expression in cultured neurons after 6 days [
29] in contrast to the previous report on acutely isolated non-cultured neurons from C57BL6 mice or mice on C57BL6/DBA genetic background showing a lack of
Pdgfrb mRNA expression [
27,
28]. It remains unclear as to whether this discrepancy is the result of alterations in gene expression in response to the growth factors and/or other constituents in the medium during culturing [
29], and may therefore not accurately represent the transcriptional profile of neuronal cell populations
in vivo, as suggested by other studies [
30‐
32] or after an acute isolation of neurons [
27,
28].
Other studies utilizing adult C57BL6 mice have further claimed that PDGFRβ is expressed in mouse cortical neurons through
ex vivo immunofluorescent staining and low resolution epi-fluorescent imaging [
23,
33]. Although these reports demonstrate PDGFRβ labeling consistent with neurons in size, shape and distribution, no colocalization studies were conducted and there is no concluding evidence against or in favor of neuronal PDGFRβ expression. In Ishii et al. 2006 [
33], the authors further claim to have deleted PDGFRβ specifically from neurons utilizing a
nestin-Cre transgenic system and demonstrate no neuronal phenotype but an exacerbated response to cryogenic injury of
nestin-Cre
+
PDGFRβ
flox/flox
transgenic mice. The interpretation of Ishii et al. experiments [
33] is, however, complicated by recent findings demonstrating that brain pericytes express
nestin [
34] and therefore PDGFRβ expression in pericytes can be affected by
nestin-Cre genetic manipulation. Since pericyte number and/or coverage have not been determined in the
nestin-Cre
+
PDGFRβ
flox/flox
transgenic mouse model [
33], it remains uncertain to what extent the observed susceptibility to cryogenic brain injury in this mouse model is due to brain pericyte loss as opposed to primary neuronal insult.
The same group utilizing organotypic brainstem slices from mice on a C57BL6 background in conjunction with single cell patch clamp neuronal recordings has suggested that neuronal PDGFRβ mediates changes in the excitatory postsynaptic currents following application of human recombinant PDGF-BB onto the slices [
23,
24]. It is of note that PDGF-BB binds to other transmembrane receptors such as platelet derived growth factor receptor alpha (PDGFRα) which is abundantly expressed on neurons in the murine CNS [
35‐
39] and can contribute to the observed changes in postsynaptic currents. As a result, it remains unclear as to whether the observed effects are mediated by binding of recombinant PDGF-BB to neuronal PDGFRα at the concentration utilized and/or whether accompanying PDGF-B/PDGFRβ signaling on brain pericytes may indeed influence neuronal electrophysiological responses further complicating the interpretation of these experiments [
24]. In support of our argument, using an
in situ proximity ligation assay capable of detecting single molecular events, we provide evidence demonstrating that detectable PDGF-B/PDGRβ interactions on tissue sections occurs exclusively on microvascular pericytes following exogenous application of equivalent concentrations (150 ng/mL) of recombinant murine PDGF-BB as in a previous study [
24].
The source of the above contradictions between reports demonstrating a lack of neuronal PDGFRβ expression
in vivo in mouse models on a 129S1/SvlmJ background in the present study and 129SV/C57BL6 background [
10,
14] and reports suggesting the presence of neuronal PDGFRβ expression in mouse models on a C57BL6 background, including the
nestin-Cre
+
PDGFRβ
flox/flox
transgenic mouse model [
23,
24,
33] remains unclear at present time. It is possible, however, that these conflicting results may be due to subtle genetic differences between mouse strains with different genetic backgrounds, as reported by others showing significant strain-specific responses to injury and strain-specific gene expression mapping in the adult mouse brain [
40‐
43]. Intriguingly, previous reports have demonstrated neuronal PDGFRβ expression in the rat CNS [
44], whereas our work in progress in the adult human CNS and Alzheimer's disease patients shows a pericyte-restricted pattern of PDGFRβ expression [Winkler EA, Bell RD, Zlokovic BV, unpublished results] indicating that species-specific differences in PDGFRβ expression may indeed exist and that the present mouse models on a 129S1/SvlmJ background could be potentially used as correlates for human studies on brain pericytes.
Conclusions
Here we have demonstrated that PDGFRβ is exclusively expressed in brain pericytes and may serve as an abundant pericyte-specific marker in the adult 129S1/Sv1mJ and F7 mouse brain. Moreover, these findings suggest that genetic disruption of PDGFRβ signaling in adult, viable mice with this genetic background, such as the F7 mice, will result in a pericyte-specific insult and will not exert a primary effect on neurons because PDGFRβ is not expressed in neurons of the adult 129S1/SvlmJ and F7 mouse brain. Therefore, mouse models with normal and deficient PDGFRβ signaling on a 129S1/SvlmJ genetic background may effectively be used to deduce the specific roles of pericytes in maintaining the cerebral microcirculation and BBB integrity within the neurovascular unit in the adult and aging brain as well as during neurodegenerative and vascular brain disorders.
Methods
Animals
Transgenic F7 mutants and their littermate controls on a 129S1/SvlmJ genetic background were provided by Dr. Philippe Soriano (Fred Hutchinson Cancer Res. Cntr., Seattle, WA). The F7 mutants were generated by point mutations in which multiple cytoplasmic tyrosine residues at positions 578, 715, 739 and 750, 770, 1008 were substituted with phenylalanine and the tyrosine at position 1020 was replaced with isoleucine, thereby disrupting Src-, Grb2-, PI3K-, RasGAP-, SHP-2 and PLC
γ-dependent signal transduction, respectively, as previously described [
12]. This results in a substantial pericyte loss in the embryonic CNS, as shown at day E14.5 [
12].
xLacZ4 mice were provided by Dr. Michelle Tallquist (UT Southwestern, Dallas, TX). Mice were cared for in accordance to the University of Rochester Medical Center Vivarium and Division of Laboratory Animal Medicine guidelines. All procedures were approved by the Institutional Animal Care and Use Committee at the University of Rochester using National Institute of Health guidelines. Mice were anesthetized with a single intraperitoneal injection of ketamine (100 mg/kg) and xylazine (10 mg/kg) and then transcardially perfused with 50 mL PBS containing 5 U/ml heparin. Brain sections were prepared as described below.
Hematoxylin and Eosin Histologic Staining
This was performed to provide a detailed view of cerebral cytoarchitecture and orientation for fluorescent imaging. Mice were anesthetized, euthanized, and perfused as described above (see Animals). Tissue sections were subsequently fixed overnight in 4% paraformaldehyde (PFA.) Staining was conducted utilizing the FD hematoxylin solution and FD eosin Y solution (FD Neurotechnologies; Catonsville, MD) as described by the manufacturer.
Digoxigenein (DIG) labeled oligonucleotides
A 129S1/SvlmJ mouse brain cDNA library was created from freshly dissected mouse cortex. A 555 bp fragment of Pdgfrβ cDNA was selectively amplified using the following PCR primer combination: 5'- ACCTAGTCGACCACCTTTGTTCTG ACCTGCTC-3' and 5'- TTCGTGGATCCATGGTGATGCTCTCGCCCT-3'. To facilitate directional cloning, primers were designed to include a 5' overhang containing a 5 nucleotide spacer sequence and either a SalI or BamHI restriction site in the forward and reverse primer, respectively. The resulting Pdgfrβ amplicon was subsequently ligated into the pBLUESCRIPT II SK vector (Stratagene; Cedar Creek, TX) containing both T3 and T7 RNA polymerase sequences. The plasmid was sent to ACGT Inc. (Wheeling, IL) for sequencing to verify proper insertion. The inserted sequence subsequently aligned to Mus Musculus Pdgfrβ mRNA. Both sense and antisense DIG-labeled riboprobes were synthesized with the DIG RNA labeling kit (Roche Applied Science; Indianapolis, IN) utilizing T7 and T3 RNA polymerase, respectively. The DIG-labeled riboprobes were then purified using G-50 sephadex quick spin columns (Roche Applied Science) and diluted to 100 ng/μL in THE RNA storage solution (Ambion; Austin, TX). All riboprobes were aliquoted and stored at -80°C.
Dual fluorescent in situhybridization
The brains of 6-8 month old 129S1/SvlmJ control mice were studied to determine the pattern of
Pdgfrβ mRNA expression in the adult brain utilizing dual fluorescent
in situ hybridization and immunofluorescent staining as previously described [
45,
46]. Mice were anesthetized and transcardially perfused as described above (see Animals). Snap frozen brains were cryosectioned at 20 μm and stored at -80°C. The following steps were performed at 25°C unless otherwise indicated. All solutions were made using nuclease-free water and glassware was pre-cleaned using RNase Zap reagent (Ambion). In brief, sections were allowed to dry for 30 minutes and then immediately fixed with 4% PFA for 20 minutes. Sections were then incubated with 0.1 M triethanolamine containing 0.25% (v/v) acetic anhydride to minimize non-specific binding of the riboprobe for 10 minutes and were treated with 10 μg/mL proteinase K (Ambion) diluted in PBS for 10 minutes to facilitate tissue penetrance. Sections were then incubated in 5× SSC (0.75 M NaCl, 0.75 M Na-citrate) for 15 min and then prehybridized with prehybridization buffer (5× SSC, 50% formamide, pH to 7.5 with HCl, 100 μg/ml sheared fish sperm DNA) for 2 hours at 65°C. Sections were placed in a humidified chamber and incubated with 300 ng/ml DIG-labeled probe diluted in prehybridization buffer for 12-16 hours at 65°C. Sections were hybridized with either antisense or sense riboprobe as a negative control. Following hybridization, slides were washed with 2× SSC and 0.1× SSC for 1 hour at 70°C. Sections were rinsed with PBS and incubated in 3% H
2O
2 for 1 hour to block endogenous peroxidase activity. Tyramide signal amplification (Invitrogen; Carlsbad, CA) was then conducted as described by the manufacturer. In brief, sections were rinsed with PBS and incubated in 10% normal swine serum (Vector Laboratories; Burlingame, CA) for 1 hour. Sections were then incubated with horseradish peroxidase (HRP)-conjugated sheep anti-DIG Fab fragments (Roche Applied Science) for 1 hour and rinsed with PBS. Oregon Green 488-conjugated tyramide diluted in amplification buffer/0.0015% H
2O
2 was applied to the sections for 10 minutes. Sections were rinsed with PBS and diluted Oregon Green 488-conjugated tyramide was reapplied for 10 minutes. Sections were then incubated with Alexa Fluor 488-conjugated goat anti-fluorescein/Oregon Green IgG (Invitrogen) overnight at 4°C. To visualize brain endothelial cells, sections were incubated with mouse anti-mouse glucose transporter 1 (GLUT1) IgG (Abcam; Cambridge, MA) overnight at 4°C. To detect GLUT1, sections were incubated with Alexa Fluor 680-conjugated donkey anti-mouse IgG (Invitrogen) for 1 hour.
Immunodetection of PDGFRβ, Desmin, Chondroitin Sulfate Proteoglycan NG2 and the xLacZ4transgenic reporter
The brains of 6-8 month old 129S1/SvlmJ mice were studied for co-localization of PDGFRβ and the well-established pericyte marker Chondroitin Sulfate Proteoglycan NG2 (NG2) [
7,
17‐
19]. Mice were anesthetized and transcardially perfused as described above (see Animals). Snap frozen brain sections were cyrosectioned at 14 μm and subsequently fixed in acetone for 10 minutes at -20°C. Sections were incubated in 10% normal swine serum (Vector Laboratories) for 1 hour at room temperature. Sections were then incubated with goat anti-mouse PDGFRβ IgG (R&D systems; Minneapolis, MN) and rabbit anti-rat NG2 IgG (Millipore; Billerica, MA) which cross-reacts with mouse NG2 overnight at 4°C. For PDGFRβ and NG2 detection sections were incubated with Alexa Fluor 546-conjugated donkey anti-goat IgG (Invitrogen) and Alexa Fluor 488-conjugated donkey anti-rabbit IgG (Invitrogen), respectively, for 1 hour at room temperature. To visualize brain microvascular endothelial cells, sections were incubated with biotin-conjugated
Lycopersicon esculentum (tomato) lectin (Vector Laboratories) for 1 hour at room temperature followed by incubation with AMCA-conjugated streptavidin (Vector Laboratories) for 30 minutes at 37°C.
To confirm colocalization of PDGFRβ with the
xLacZ4 transgenic reporter, a marker of committed and differentiated, non-proliferating pericytes/vascular smooth muscle cells [
20], snap frozen 2 month old
xLacZ4 transgenic mouse brain sections were cryosectioned at 14 μm and subsequently fixed in a 1:1 solution of acetone and methanol for 10 minutes at room temperature. Sections were blocked and incubated with goat anti- mouse PDGFRβ IgG (R&D systems) and mouse anti-β-galactosidase IgG (Cell Signaling; Danvers, MA) as described above. For PDGFRβ and LacZ detection sections were incubated with Alexa Fluor 488-conjugated donkey anti-goat IgG (Invitrogen; Carlsbad, CA) and Alexa Fluor 546-conjugated donkey anti-mouse IgG (Invitrogen), respectively, as described above. Brain microvascular endothelial cells were visualized with fluorescent lectin staining as described above.
To study PDGFRβ-positive or desmin-positive pericyte coverage, tissue sections were prepared as described above. Sections were then incubated with goat anti-mouse PDGFRβ IgG (R&D systems) or mouse anti-human desmin IgG (Dako USA; Carpinteria, CA) which cross-reacts with mouse desmin overnight at 4°C. For PDGFRβ and desmin detection sections were incubated with Alexa Fluor 546-conjugated donkey anti-goat IgG (Invitrogen; Carlsbad, CA) or Alexa Fluor 546-conjugated donkey anti-mouse IgG (Invitrogen), respectively, for 1 hour at room temperature. Brain microvascular endothelial cells were visualized with fluorescent lectin staining as described above.
Immunodetection of neuronal-specific and astrocyte-specific markers
The brains of 6-8 month old 129S1/SvlmJ littermate controls and 6-8 month old F7 mice were studied for colocalization of PDGFRβ with neuronal neurofilament-H (SMI-32), neuronal nuclear antigen A60 (NeuN), or astrocyte glial fibrillary acidic protein (GFAP). Mice were anesthetized as described above and transcardially perfused with 10 mL PBS containing 5 U/ml heparin followed by 30 mL 4% PFA. Brains were carefully removed from the skull and immersion fixed in 4% PFA overnight at 4°C. Forty μm vibratome-sectioned brain sections were incubated in target antigen retrieval solution, pH 9 (Dako USA) for 15 minutes in a 80°C water bath and then blocked and incubated with the following primary antibodies as described above: goat anti-mouse PDGFRβ (R&D systems), mouse anti-mouse SMI-32 (Abcam), mouse anti-mouse NeuN (Millipore), and mouse anti-mouse GFAP (Cell Signaling). PDGFRβ was detected as described above. To detect NeuN, SMI-32, and GFAP sections were incubated with Alexa Fluor 488-conjugated donkey anti-mouse IgG for 1 hour at room temperature. Brain microvascular endothelial cells were visualized with fluorescent lectin staining as described above.
Quantification of PDGFRβ-positive and Desmin-positive pericyte coverage of lectin-positive brain microvessels
All images were acquired by a blinded investigator using a custom-built Zeiss LSM 510 meta confocal laser scanning microscope (see below) using a Zeiss LD LCI Plan-Apochromat 25×/0.8 Imm Korr DIC water immersion objective. In each mouse, four different 200 μm × 200 μm fields were analyzed per brain region in six nonadjacent sections approximately 100 μm apart. PDGFRβ-postive or desmin-positive immunofluorescent and lectin-positive fluorescent signals from brain microvessels < 6 μm in diameter were individually subjected to threshold processing and the areas occupied by their respective signals were quantified by a blinded investigator using the NIH Image J software Area measurement tool. The percent of PDGFRβ pericyte coverage of lectin-positive microvessels was then deteremined by normalizing the PDGFRβ-positive signal to the lectin positive-brain microvessel signal. We used Graph-Pad Prism software for statistical calculations. One-way analysis of variance followed by a Tukey posthoc test were used to determine statistically significant differences in PDGFRβ-positive pericyte coverage amongst different brain regions in both 129S1/SvlmJ mice and F7 mice. Mean values are reported plus or minus the standard error of the mean (SEM).
Fluorescent in situPDGF-B/PDGFRβ proximity ligation assay
The brains of 6-8 month old 129S1/SvlmJ mice were studied to determine the localization of PDGF-B/PDGFRβ interactions in situ. Mice were anesthetized and transcardially perfused as described above (see Animals). Frozen brain sections were cryosectioned at 14 μm. Sections were rehydrated with PBS and blocked with 10% normal swine serum (Vector Laboratories) for 1 hour at room temperature. Sections were then incubated with 150 ng/mL mouse recombinant PDGF-BB (GenWay Biotech, Inc.; San Diego, CA) for 1 hour at room temperature. Sections were washed with PBS containing 0.05% triton X-100 and subsequently fixed with 4% PFA for 10 minutes at room temperature. Sections were incubated with goat anti-mouse PDGFRβ IgG (R&D systems) and rabbit anti-human PDGF-B IgG (Thermo Scientific; Rockford, IL) which cross reacts with mouse PDGF-B overnight at 4°C. Proximity ligation was then conducted in situ as described by the manufacturer (Olink Bioscience; Uppsala, SE) utilizing the Duolink II PLA probe anti-goat PLUS, Duolink II PLA probe anti-rabbit MINUS, and Duolink II detection reagents orange to visualize PDGF-B/PDGFRβ interactions. Following serial SSC washes, sections were rinsed with PBS were incubated with biotin-conjugated Lycopersicon esculentum lectin (Vector Laboratories) overnight at 4°C. To visualize brain microvascular endothelial cells and PDGFRβ, sections were incubated with Dylight 649-conjugated streptavidin (Vector Laboratories) and Alexa Fluor 488-conguated donkey anti-goat IgG (Invitrogen), respectively, for 1 hour at room temperature.
Laser scanning confocal microscopy imaging
All images were acquired using a custom-built Zeiss LSM 510 meta confocal laser scanning microscope with a Zeiss LD LCI Plan-Apochromat 25×/0.8 Imm Korr DIC, C-Apochromat 40X or 63X water immersion objective (Carl Zeiss Microimaging Inc.; Thronwood, NY). A 488 nm argon laser was used to excite Alexa Fluor 488 and the emission was collected through a 500-550 band-pass filter. A 543 HeNe laser was used to excite Alexa Fluor 546 and the emission was collected through a 560-615 band-pass filter. A 633 HeNe laser was used to excite Alexa Fluor 680 and Dylight 649 and the emission was collected through a 650-710 band-pass filter. An 800 nm mode-locked femtosecond pulsed DeepSee Ti:sapphire laser (Spectra Physics; Santa Clara, CA) was used for AMCA excitation and emission was collected using 405-450 nm band-pass filter.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
EAW conducted fluorescent in situ hybridization and proximity ligation experiments and helped to draft the manuscript. RBD conducted all immunostaining experiments, quantified pericyte coverage, and helped to draft the manuscript. BVZ designed and supervised all experiments and wrote the manuscript. All authors read and approved the final manuscript.