Introduction
Blood vessels form hierarchically branched networks that develop organotypically in a highly unique stereotyped fashion. Once differentiated, maturation and maintenance of this vascular network are provided by stabilizing mural cells (MCs): pericytes (PCs) that envelop small caliber vessels, the capillaries, and vascular smooth muscle cells (vSMCs) covering big caliber vessels, the arteries and veins [
1,
2]. Blood flow-induced physiological shear stress is an essential contributor to vessel stabilization in part by regulating the expression of endothelial ligands required for recruitment and maintenance of MCs, e.g.
Pdgfb (encoding platelet-derived growth factor-B, PDGFB),
Jag1 (encoding Jagged1) or
Tgf-β1 (encoding transforming growth factor β1, TGF-β1) [
3,
4]. In turn, the direct contact of MCs with endothelial cells (ECs) allows the MCs to regulate blood flow and the vascular tone within the vascular network [
5]. PCs are also important regulators of endothelial angiogenic behavior by limiting VEGF [
6] and Angpt2-Tie2 signalling [
7], maintaining capillary zonation [
8] and suppressing inflammatory responses in the endothelium [
9]. Recently, it has been proposed that contractile junctional PCs also dynamically control of blood flow directionality within the capillary network [
10].
The recruitment of PCs during angiogenesis is mediated by the angiocrine PDGFB produced by the ECs which paracrine signals through its corresponding receptor PDGF receptor-β (PDGFRβ), exclusively expressed by PCs. Disrupted PDGFB/PDGFRβ signalling pathway results in strongly reduced PC number and various vascular defects, including vascular hyperplasia, microaneurysm, leakage, diabetic retinopathy and impairment in the formation of the blood–brain and blood–retina barrier (BBB, BRB) [
7,
9,
11‐
16].
Impaired PC coverage has also been reported in sporadic human brain arterial-venous malformations (AVMs) with a more pronounced loss in ruptured AVMs [
17]. Also, PC dropout is a hallmark of murine Human Hemorrhagic Telangiectasia (HHT)-like AVMs caused by loss of function (LOF) of canonical BMP9 and BMP10 signaling [
18‐
21]. Whether the localized reduction in PC coverage is a consequence of pathological flow, a contributing factor, or a causal determinant of the high flow-AVMs is still not clear. Much less is known about the angiocrine-paracrine signaling pathways whose disruption contributes to PC dysfunction and AVM formation, or whether flow-induced PDGFB activation is implicated.
NOTCH signalling acting upstream of PDGFB, also promotes PC recruitment and adhesion to microvessels during vascular morphogenesis. Loss of canonical NOTCH in perivascular cells [
22] or gain of NOTCH signaling in ECs [
23] has been associated with PC reduction and AVM pathogenesis. A similar phenotype was observed when
Srf (Serum response factor) transcription factor, acting downstream of PDGFB-PDGFRβ, is depleted from MCs [
24].
In the present study, we evaluated the role of PDGFB signaling-mediated PC recruitment and maintenance in AVM pathogenesis. Our data show that disruption of endothelial Pdgfb resulting in PC loss in developing vessels leads to capillary enlargement giving rise to organotypic arterio-venous (AV) shunting containing non-proliferative hyperplastic, hypertrophic and miss-oriented ECs with an altered AV capillary zonation. Mechanistically, loss of Pdgfb is associated with an increase in Krüppel like factor 4 (KLF4) expression-meditated excessive Bone morphogenic protein (BMP), TGF-β and NOTCH activation in ECs. Collectively, the data here suggest that PDGFB-mediated PC recruitment and maintenance on developing vessels restrict capillary EC size and caliber to prevent hemodynamic changes and AV shunting. Furthermore, our study identifies novel targets with the potential to prevent vascular lesions.
Discussion
MCs and ECs act in concert to regulate angiogenesis and to maintain vascular homeostasis [
2]. On the one hand, angiocrine factors produced by the endothelium promote PC recruitment and thereafter their maintenance on the endothelial layer. In turn, MCs provide dynamic control of blood perfusion within the vascular networks and thus, maintain an appropriate vascular tone. Yet, despite recent advances, how dysfunctional ECs-MCs crosstalk contributes to the progression of many diseases, including high-flow mediated-AVM pathogenesis, still remains poorly understood. AVMs are direct connections between an artery and a vein with a loss of capillary bed. Sporadic or inherited, these vascular lesions show a localized drop of PCs. Whether PC loss from developing capillaries triggers AVM formation or it is secondary to pathological hemodynamics-mediated AVMs remains an open question in the field. Furthermore, which angiocrine-paracrine signals are essential for maintaining functional PCs to stabilize the developing endothelium remain largely unknown.
Recent advances in the field identified key signaling components in MCs required to protect the endothelium against forming AVMs. One such example is NOTCH signaling which acts upstream of PDGFRβ to regulate PC survival and proliferation [
22]. Downstream of PDGFRβ, ablation of
Srf transcription factor also leads to AVM formation due to inadequate PC migration towards the endothelium and contractile SMC-mediated loss of vascular tone [
24].
On the other hand, gain of function of NOTCH [
23] or LOF of canonical BMP9/10 signaling in ECs leads to AVM formation [
3,
19‐
21], and both pathways were shown to regulate
Pdgfb expression [
3,
53]. Interestingly the two pathways converge in regulating many common downstream genes [
54], including N-cadherin, a critical player in mediating EC-PC communications, downstream of PDGFB-PDGFRβ signaling [
55]. Physiological flow regulates NOCTH activation in ECs [
56] and BMP9 is indispensable for flow-induced expression of master regulators of MC recruitment in ECs, including PDGFB [
3], thus emphasizing intertwined upstream regulatory mechanisms of angiocrine PDGFB production to maintain vascular homeostasis.
Yet, disrupting PDGFB/PDGFRβ signaling either in endothelium or MCs leads to a plethora of vascular defects, but this angiocrine-paracrine signaling axis has never been interrogated in the context of inducing AVM pathogenesis. Interestingly, overexpression of PDGFB reduces hemorrhage in bAVMs and HHT patients [
18,
57]. Yet, to our knowledge, our study is the first to provide a genetic link between
Pdgfb ablation-mediated PC loss and AVM development in mice.
Herein, employing inducible model of Pdgfb ablation, we demonstrate that EC-derived Pdgfb is indispensable for PC recruitment, but also for PC maintenance on developing vessels. Depletion of Pdgfb either before or after the vasculature differentiates, results in PC loss and capillary enlargement across multiple vascular beds and AV shunting.
Yet, the resulting AVM-like structures are not identical to the inherited HHT-like AVMs. Upon
Pdgfb ablation, AVMs-like structures form at the vascular front in low flow regions, suggesting distinct cell events whose disruption initiate and contribute to these lesions. Indeed, the resulting AVMs-like structures are characterized by enlarged capillaries containing non-proliferative, hyperplastic and hypertrophic ECs with un-organized junctions. Similarly to
Pdgfbret/ret brain capillaries [
35], retinal capillary ECs in
PdgfbiΔEC showed a venous skewing, emphasizing a conserved role for PDGFB signaling in maintaining capillary zonation. Yet, specifically and exclusively, the AV shunts displayed gain of arterial identity with some arterial ECs present in the veins, emphasizing either an arterial origin, or disruption in flow-migration coupling as the triggering event for capillary enlargement. While we failed to find an arterial origin for the AVMs-like structures upon
Pdgfb loss, instead similarly to
Alk1 or
Eng depletion [
30,
31,
58], we identified an altered EC migration among the hallmarks that could precipitate AV shunting. This underscores an important role for capillary and/or venous PDGFB-PDGFRβ signaling in flow-migration coupling.
If the perturbed migration is a consequence of defective hemodynamics within the hyperplastic and hypertrophic capillaries or it is due to altered attraction-repulsive mechanisms that further precipitate AV shunting, remains to be further investigated. Indeed,
Pdgfb ablation resulted in activation of the EphrinB2-EphB4 signaling mediated arterial-venous EC repulsive-attraction [
59] and interestingly also of Apelin/Apj signaling pathway. Yet, loss of Apelin, a BMP suppressed target gene [
46], leads to a defective EC migration against the bloodstream [
60]. One possible explanation for the discrepant results is that increased
Apelin expression upon
Pdgfb loss is mediated through another upstream regulator, such as NOTCH [
60].
Abundant KLF4 within the enlarged capillaries engaged in AV shunts indicates perturbed hemodynamics with a disrupted vascular tone. Yet, it may also suggest an increased sensitivity of ECs to FSS upon
Pdgfb loss, as recently reported in AVMs upon
Smad4 depletion [
61]. Interestingly, hypertensive (αSMA positive) veins at the expense of hypotensive arteries (αSMA negative) seen in the
Pdgfb LOF retinas and in human AVMs [
17], further confirm an altered hemodynamics. Surprisingly, in vitro we could show that
PDGFB knock down potently augments flow-induced morphological events. Whether PDGFB cell autonomously restrains FSS-mediated EC responses to maintain capillary EC size, caliber and zonation by restricting KLF4-mediated hemodynamic changes or these cell events occur due to ensheathing PCs-mediated intrinsic and/or extrinsic signaling is so far unclear. At this point, we can only speculate that vascular lesions observed in these mutant mice are rather secondary effects of PC coverage defects. Yet, the possibility for an autocrine contribution to AVM pathogenesis cannot fully be excluded. Whether the venous or capillary PDGFB protects the endothelium against AVMs also needs to be investigated further and will require generation of capillary- or vein-specific Cre driver lines.
Based on our findings, we propose a model in which EC Pdgfb ablation-mediated loss of PCs results in capillary enlargement leading to perturbed hemodynamics and an increase of EC sensitivity to flow, which will result in KLF4 overactivation. Excessive KLF4 then triggers changes in arterial-venous identity regulatory programs: activation of NOTCH and TGF-β1-Smad3 mediated arterial identity in the same time with BMP6-mediated venous identity through Smad1/5 activation. This altered EC fate pattern together with loss of repulsion-attraction regulatory mechanisms then triggers a confusion in the migration of ECs against the bloodstream that leads to an accumulation of ECs in capillaries, further increasing capillary caliber and ultimately giving rise to AVMs.
Materials and methods
Animal experiments
Deletion of endothelial Pdgfb (PdgfbiΔEC) was achieved by crossing Pdgfb fl/fl with Tx inducible Cdh5-CreERT2 mice. Deletion of Pdgfb in the arterial ECs (PdgfbiΔBMX) was achieved by crossing Pdgfb fl/fl to the Tx inducible Bmx-CreERT2;mTmG mice. Gene deletion was achieved by i.p injections of 100 µg Tx (Sigma, T5648) into PdgfbiΔEC or PdgfbiΔBMX at postnatal days (P0–P2) or (P5–P7). Tx-injected Cre-negative littermates (fl/fl) were used as controls. Mice were maintained under standard specific pathogen-free conditions, and animal procedures were approved by the animal welfare commission of the Regierungspräsidium Karlsruhe (Karlsruhe, Germany) and The Comité Institutionnel des Bonnes Pratiques Animales en Recherche (CIBPAR), Canada.
Latex red-dye injection
P7 or P12 pups were anaesthetized and perfused with 2 ml of PBS. Latex dye (Connecticut Valley Biological Supply Company) was slowly injected through left ventricle with an 1 ml insulin syringe. To visualize pulmonary arteries, latex was injected into the right ventricle. Tissues were fixed in 4% PFA at 4 °C overnight and washed in PBS the following day. The dissected organs were imaged under a dissection microscope.
Reagents and antibodies
For immunodetection: anti-NG2 (#AB5320, 1:200, millipore), Isolectin B4 (IB4, #121412, 10 μg/ml, Life Technologies), anti-GOLPH4 (#ab28049; 1:200, Abcam), anti-ERG (#92513; 1:200, Abcam), anti-KLF4 (#AF3158, 1:200, R&D systems), anti-SOX17 (#AF1924, 1:200, R&D systems), anti-VE-cadherin (#555289, 1:400, BD), anti-Jag1 (#AF599, 1:200, R&D systems), anti-Endomucin (#sc-65495, 1:200, Santa Cruz), anti-Dll4 (#AF1389, 1:200, R&D systems), anti-phospho-SMAD3 (#ab52903, 1:100, Abcam), anti-phospho-SMAD1/5 (#13820S, 1:100, Cell Signalling), anti-ID1 (#AF4377, 1:100, R&D systems), anti-Endoglin (#AF1320, 1:100, R&D systems), anti-GM130 (#610823; 1:600 BD Bioscience), anti-BMP6 (#ab15640, 1:100, Abcam), anti-TGF-β1 (#MAB7666, 1:100, R&D systems), anti-αSMA (#ab184675,1:600, Abcam), anti-ICAM2 (#553326, 1:200, BD), anti-Collagen IV (#ab6586, 1:200, Abcam) and anti-P21 from CNIO-Centro Nacional de Investigaciones Oncológicas).
For WB: anti-PPDGFB (#ab23914, 1:1,000, Abcam), anti-SMAD1 (#6944S, 1:1,000, Cell Signalling), anti-SMAD2/3 (#8685S; 1:1000; Cell Signalling), anti-phospho-SMAD1/5/8 (#13820S, 1:1000, Cell Signalling), anti-phospho-SMAD3 (#ab52903, 1:1000, Abcam), anti-TGF-β1 (#sc52893, 1:500, Santa cruz) and anti-VE-cadherin (#sc9989, 1:200, Santa cruz).
Appropriate secondary antibodies were fluorescently labelled (Alexa Fluor donkey anti-rabbit, #R37118, Alexa Fluor donkey anti-goat 555, #A-21432, 1:250, Thermo Fisher) or conjugated to horseradish peroxidase for WB (Anti-Rabbit #PI-1000-1 and Anti-mouse #PI-2000-1 IgG (H + L), 1:5000, Vector Laboratories).
Proliferation assay in vivo
For analysis of cell proliferation in the retinas, pups were injected i.p with 5-ethynyl-2-deoxyuridine (EdU, 100 mg/Kg; Thermo Fischer Scientific) 4 h before dissection. Retinas were collected and EdU labelling was detected with the Click-it EdU Alexa Fluor-488 Imaging Kit (C10337, Life Technologies) according to the manufacturer’s instructions.
Isolation of mLECs
Mouse lung ECs were isolated using MACS (Miltenyi Biotec). Mice were sacrificed and lungs were harvested immediately. Lungs were cut into small pieces and digested with collagenase I at 37 °C for 45 min. Tissue suspension was passed through 70 μm cell strainer, incubated with red blood cell lysis Buffer (11814389001, Sigma) and washed several times with PEB buffer (0.5% BSA, 2 mM EDTA in PBS). Cell suspension was incubated with CD45 MicroBeads (1:10) for 15 min at 4 °C and passed through the MS columns. The unlabeled cells were collected and centrifuged at 1000 rpm at 4 °C for 10 min. Cell pellets was resuspended with PEB buffer and incubated with CD31 MicroBeads (1:10) for 15 min at 4 °C and applied onto MS columns, eluted from the columns with PEB buffer and directly used for RNA or protein extraction.
Quantitative PCR
RNA extraction from mLECs were performed using RNeasy-kit (74106, Qiagen) according to the manufacturer’s instructions. The RNA was reverse transcribed using High-Capacity cDNA Reverse Transcription Kit (4368813, Thermo Fisher) and quantitative PCR assays were carried out using PowerUP SYBR Green Master Mix (A25778, Thermo Fisher) with a QuantStudio 3 (Thermo Fisher) according to the manufactures protocol. The following primers were used for mLECs: Eng (Forward: AGGGGTGAGGTGACGTTTAC, Reverse: GTGCCATTTTGCTTGGATGC), Thbs1 (Forward: CCTGCCAGGGAAGCAACAA, Reverse: ACAGTCTATGTAGAGTTGAGCCC), Fn1 (Forward: ATGTGGACCCCTCCTGATAGT, Reverse: GCCCAGTGATTTCAGCAAAGG), Dll4 (Forward: TTCCAGGCAACCTTCTCCGA, Reverse: ACTGCCGCTATTCTTGTCCC), Notch1 (Forward: GATGGCCTCAATGGGTACAAG, Reverse: TCGTTGTTGTTGATGTCACAGT), Notch4 (Forward: GAACGCGACATCAACGAGTG, Reverse: GGAACCCAAGGTGTTATGGCA), Hey1 (Forward: CCGACGAGACCGAATCAATAAC, Reverse: TCAGGTGATCCACAGTCATCTG), Hey2 (Forward: CGCCCTTGTGAGGAAACGA, Reverse: CCCAGGGTAATTGTTCTCGCT), Hes1 (Forward: TCAACACGACACCGGACAAAC, Reverse: ATGCCGGGAGCTATCTTTCTT), Klf4 (Forward: GGCGAGTCTGACATGGCTG, Reverse: GCTGGACGCAGTGTCTTCTC), Bmp6 (Forward: GCGGGAGATGCAAAAGGAGAT, Reverse: ATTGGACAGGGCGTTGTAGAG), Tgf-β1 (Forward: CCACCTGCAAGACCATCGAC, Reverse: CTGGCGAGCCTTAGTTTGGAC), Pdgfb (Forward: CATCCGCTCCTTTGATGATCTT, Reverse: GTGCTCGGGTCATGTTCAAGT), Ephb4 (Forward: GGAAACGGCGGATCTGAAATG, Reverse: TGGACGCTTCATGTCGCAC), Nrp2 (Forward: GCTGGCTACATCACTTCCCC, Reverse: GGGCGTAGACAATCCACTCA), Nr2f2 (Forward: CATCGAGAACATTTGCGAACTG, Reverse: GTCGGCTGACATGGGTGAAG), Aplnr (Forward: CCAGTCTGAATGCGACTACG, Reverse: CTCCCGGTAGGTATAAGTGGC), Sox17 (Forward: GATGCGGGATACGCCAGTG, Reverse: CCACCTCGCCTTTCACCTTTA), Nrp1 (Forward: ACCTCACATCTCCCGGTTACC, Reverse: AAGGTGCAATCTTCCCACAGA), Unc5b (Forward: CGGGACGCTACTTGACTCC, Reverse: GGTGGCTTTTAGGGTCGTTTAG), Efnb2 (Forward: TTGCCCCAAAGTGGACTCTAA, Reverse: GCAGCGGGGTATTCTCCTTC), Tgfbr1 (Forward: AAAACAGGGGCAGTTACTACAAC, Reverse: TGGCAGATATAGACCATCAGCA), Acvr1 (Forward: ATGGTCGATGGAGTAATGATCCT, Reverse: TGCTCATAAACCTGAAAGCAGC).
Retina isolation and immunostaining
The eyes from P7 or P12 pups were fixed in 4% PFA for 17 min at room temperature (rt). Post dissection, the retinas were washed 3 times with PBS and then incubated in blocking buffer (1% fetal bovine serum, 3% BSA, 0.5% Triton X-100, 0.01% sodium deoxycholate, 0.02% sodium azide in PBS at pH 7.4) for 15 min at rt. Post-blocking, retinas were incubated with specific antibodies diluted in blocking buffer overnight, at 4 °C. The next day, retinas were washed and incubated with anti-IB4 together with the corresponding secondary antibody in PBLEC buffer (1 mM CaCl2, 1 mM MgCl2, 1 mM MnCl2 and 0,25% Triton X-100 in PBS) for 1 h at rt, postfixed for 20 min with 4% PFA in rt, washed and mounted in fluorescent mounting medium (RotiMount FluorCare #HP19.1, CarlRoth). Whole-mount retina images were acquired using Zeiss LSM800 confocal microscope with Airyscan Detector and the Zeiss ZEN software. Quantification of retinal vasculature was analyzed using Fiji.
Western blotting
Total protein from the mLECS were lysed with Laemmli buffer (1610747, Biorad). Samples were separated on 10% SDS-PAGE gels and transferred on 0.2 µm nitrocellulose membranes (10600004, GE Healthcare). Western blots were developed with the Clarity Western ECL Substrate (1705061, Biorad) on a Luminescent image Analyzer, Fusion FX (Vilber). Bands’ intensity were quantified using ImageJ.
Cell culture, siRNA transfection
Human umbilical vein endothelial cells (HUVECs), from Lonza (#C2519A), were grown in culture using Endothelial Cell Growth Medium MV2 supplemented with a mix of additives (#C-22022, PromoCell) and 1% Penicillin/Streptomycin solution (#P4333, Sigma-Aldrich). These cells were cultured in an incubator set at 37 °C, with a 5% CO2 atmosphere and 100% humidity. Deletion of PDGFB was carried out by transfecting 25 pmol of PDGFB siRNA (ON-TARGETplus Human PDGFB siRNA Smart Pool, #L-011749-00-0005) using Lipofectamine RNAiMax (Invitrogen) in OPTI-MEM. The experiments were conducted within the window of 48 to 72 h post-transfection, and the results were compared to HUVECs transfected with siRNA CTRL (ON-TARGETplus Non-Targeting Pool D-001810-10-05).
Exposure of ECs to FSS
siRNA tansfected HUVECs were placed onto a µ-Slide VI0.4 (Ibidi, #80601) and exposed to laminar fluid shear stress levels of 1 and 12 DYNES/cm2 for 24 h using the Ibidi pump system (Ibidi, #10902).
Image analysis
PC and αSMA coverage were assessed by quantifying the area of NG2-positive or αSMA positive area normalised to the area of IB4-positive vasculature. The vascular area was quantified by determining the threshold value of EC area relative to the total field area. The radial length was determined by measuring the distance from the optic nerve to the outer edge of the vascular front for each leaflet. The vessel diameter was calculated by averaging of six-eight measurements per retina. The number of side branches (intersections in major vessels) was calculated in 3–4 major vessels per retina. The number of tip cells in sprouting vascular fronts was measured in four pictures per retina.
Statistical analysis
All data are presented as mean ± standard error of the mean (SEM). Samples with equal variances were tested using Mann–Whitney U test or two-tailed Student’s t test between groups using GraphPad Prism (GraphPad Software). P value < 0.05 was considered to be statistically significant. Statistical analyses were performed for all quantitative data using Prism 9.0 (Graph Pad).
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.