Activation of PDGF-CC by tissue plasminogen activator impairs blood-brain barrier integrity during ischemic stroke

Abstract

Thrombolytic treatment of ischemic stroke with tissue plasminogen activator (tPA) is markedly limited owing to concerns about hemorrhagic complications and the requirement that tPA be administered within 3 h of symptoms. Here we report that tPA activation of latent platelet-derived growth factor-CC (PDGF-CC) may explain these limitations. Intraventricular injection of tPA or active PDGF-CC, in the absence of ischemia, leads to significant increases in cerebrovascular permeability. In contrast, co-injection of neutralizing antibodies to PDGF-CC with tPA blocks this increased permeability, indicating that PDGF-CC is a downstream substrate of tPA within the neurovascular unit. These effects are mediated through activation of PDGF-alpha receptors (PDGFR-alpha) on perivascular astrocytes, and treatment of mice with the PDGFR-alpha antagonist imatinib after ischemic stroke reduces both cerebrovascular permeability and hemorrhagic complications associated with late administration of thrombolytic tPA. These data demonstrate that PDGF signaling regulates blood-brain barrier permeability and suggest potential new strategies for stroke treatment.

Figure 1

Active PDGF-CC mediates tPA-induced cerebrovascular permeability. (a) Comparison of EB extravasation 1 h following either intravenous injection (IV) with 10 mg/kg of tPA as a bolus (~250 µg/mouse) or intraventricular injection (CSF) with 585 ng of tPA. (b) EB extravasation 1 h after intraventricular injection of PBS (PBS), tPA (tPA), active PDGF-CC (PDGF-CC), or active PDGF-CC together with tPA (PDGF-CC+tPA). (c) EB extravasation 1 h after intraventricular injection of either PBS (PBS), active tPA (tPA), tPA with blocking antibodies to PDGF-CC (tPA+anti-PDGF-CC), or tPA together with control IgG (tPA+IgG). (d) EB extravasation 1 h after intraventricular injection of PBS (PBS), PDGF-AA (PDGF-AA), PDGF-BB (PDGF-BB), or active PDGF-CC (PDGF-CC). (e) EB extravasation 1 h after intraventricular injection with either PBS (PBS), active PDGF-CC (PDGF-CC), or active PDGF-CC together with the LRP antagonist RAP (PDGF-CC+RAP). For all injections into the CSF, 3 µl of 3 µM protein was used except for antibodies which were 0.4 mg/ml. Each group n = 8–10 and errors represent S.E.M. Single asterisks indicate p < 0.01 vs. PBS, and ** indicates p < 0.05 vs the IgG control. (f) PDGF-CC cleavage by tPA is impaired in Lrp^−/− cells (LRP KO). Serum free medium from Lrp^−/− MEFs and wild-type cells demonstrates that both cell lines express the 48 kDa full length PDGF-C, while the addition of exogenous tPA to the cells only generates the 22 kDa PDGF-C species in the presence of the wild type cells but not in Lrp^−/− cells.

Figure 2

TPA and PDGF-CC induce similar morphological changes in brain vasculature. (a–c) Light microscopy images of cerebral sections prepared for EM analysis but stained with toluene blue. (d–f) High magnification micrographs from electron microscopy of cerebral arterioles. Brains were harvested 1 h after intraventricular injection of either PBS (a and d), tPA (b and e) or PDGF-CC (c and f). Scale bar is 50 µM in a–c and 2 µM in d–f.

Figure 3

PDGF-CC, tPA and the PDGFR-α are expressed in the NVU. (a) Sections from normal mouse brains stained with antibodies to PDGF-CC show PDGF-CC staining in patches associated with arterioles (arrows), but not with capillaries (arrowheads). (b) Sections from normal mouse brains stained with antibodies to tPA show mainly perivascular tPA staining in association with an arteriole (arrows). (c–h) Micrographs showing PDGFR-α expression in mouse brain using a GFP reporter. (c) Sections of double heterozygous Pdgfrα^+/GFP/Pdgfc^+/lacZ mouse brains stained with the PDGF-C reporter Xgal and antibodies to the endothelial cell marker platelet/endothelial cell adhesion molecule-1 (PECAM red). The arrows indicate vessel associated PDGF-C expression and arrowheads indicate non-vessel associated expression. GFP and PECAM staining was visualized using fluorescence, whereas Xgal staining was viewed in bright field. (d,e) Whole mount immunofluorescence staining of GFP-positive vessel fragments stained for SMA (red) (d) confirmed that the GFP-positive vessel fragments are arterioles. The GFP-positive nuclei are mainly localized outside of the SMA-positive cells. In contrast, co-staining with GFAP (red) (e), an astrocyte marker, suggests that the GFP-positive cells are astrocytes. (f–h) Immunofluorescence staining of brain sections from Pdgfrα^+/GFP mice stained with markers for astrocytes, GFAP (red) (f), vascular smooth muscle cells, SMA (red) (g), and endothelial cells, PECAM (red) (h). Similar to the isolated vessel fragments, co-localization of GFP expression with GFAP was abundant in the stained brain sections and produced a yellow color (g). Scale bar is 50 µm in a–b, d–h, and 20 µm in c.

Figure 4

PDGF-CC is expressed by astrocytes in culture. (a–c) Immunoblot analysis of PDGFCC in astrocytes. (a) A 48 kDa band corresponding to full length PDGF-C is seen in control cell media whereas addition of tPA (tPA) to the cells prior to collection of the medium induces release of the active 22 kDa PDGF-C species. PDGFR-α expressed by astrocytes in culture is stimulated by addition of active PDGF-CC (b) or tPA (c) to the cells. Receptor tyrosine phosphorylation is determined by immunoblotting (IB) of cell lysates for phosphotyrosine. Immunoblotting calnexin is also shown as a loading control.

Figure 5

Blocking PDGFR-α activation reduces cerebrovascular permeability and stroke volume after MCAO. (a) PDGFR-α activation following MCAO in wild-type mice 6 h after MCAO. Brains were divided into ipsilateral and contralateral hemispheres, detergent-solubilized and the lysates immunoprecipitated with PDGFR-α-specific antibodies followed by immunoblotting with phosphotyrosine-specific antibodies. As a loading control, the precipitated protein was visualized with PDGFR-α-specific antibodies. (b) Quantitation of PDGFR-α activation following MCAO in wild-type or tPA KO mice (n = 2 in each group and errors represent S.E.M). (c,d) Representative images of the ipsilateral hemisphere 24 h after MCAO from control (c) and Imatinib-treated mice (d) injected with EB 1 h before euthanasia. (e) Quantification of the EB extravasation 24 h after MCAO (n = 11 for each group and errors represent S.E.M.). (f) Quantification of the EB extravasation 24 h after MCAO in mice treated with intraventricular PBS, preimmune IgG or PDGF-CC-specific antibodies followed immediately by MCAO (n = 10 for each group and errors represent S.E.M). (g) Quantification of infarct size 72 h after MCAO in mice treated with either Imatinib or vehicle (n = 9 for control group and n = 11 for Imatinib group and errors represent S.E.M.). For all Imatinib studies mice were treated by gavage with either vehicle or 200 mg/kg Imatinib 1 h and 8 h after MCAO and twice daily for the duration of the experiment. In panels e–g the asterisks indicate p < 0.05 vs. control animals. Scale bar is 2 mm in c and d.

Figure 6

Blocking the PDGF-CC/PDGFR-α pathway reduces intracerebral hemorrhage after MCAO. (a,b) Representative images of the intracerebral hemorrhage in the ipsilateral hemisphere 24 h after MCAO of animals treated with tPA 5 h after MCAO. Mice were treated by gavage with either vehicle (a) or 200 mg/kg Imatinib (b) 1 h and 8 h after MCAO. (c) Quantification of hemoglobin content in the ischemic hemisphere 24 h after MCAO (n = 10 for control group and n = 12 for Imatinib group and errors represent S.E.M). The asterisk indicate p < 0.05 vs. control animals. Scale bar is 2 mm in a and b.