Mild hypoxia triggers transient blood-brain barrier disruption: a fundamental protective role for microglia

Abstract

We recently demonstrated that when mice are exposed to chronic mild hypoxia (CMH, 8% O₂), blood vessels in the spinal cord show transient vascular leak that is associated with clustering and activation of microglia around disrupted vessels. Importantly, microglial depletion profoundly increased hypoxia-induced vascular leak, implying that microglia play a critical role maintaining vascular integrity in the hypoxic spinal cord. The goal of the current study was to examine if microglia play a similar vasculo-protective function in the brain. Employing extravascular fibrinogen leak as an index of blood-brain barrier (BBB) disruption, we found that CMH provoked transient vascular leak in cerebral blood vessels that was associated with activation and aggregation of Mac-1-positive microglia around leaky vessels. Interestingly, CMH-induced vascular leak showed regional selectivity, being much more prevalent in the brainstem and olfactory bulb than the cerebral cortex and cerebellum. Pharmacological depletion of microglia with the colony stimulating factor-1 receptor inhibitor PLX5622, had no effect under normoxic conditions, but markedly increased hypoxia-induced cerebrovascular leak in all regions examined. As in the spinal cord, this was associated with endothelial induction of MECA-32, a marker of leaky CNS endothelium, and greater loss of endothelial tight junction proteins. Brain regions displaying the highest levels of hypoxic-induced vascular leak also showed the greatest levels of angiogenic remodeling, suggesting that transient BBB disruption may be an unwanted side-effect of hypoxic-induced angiogenic remodeling. As hypoxia is common to a multitude of human diseases including obstructive sleep apnea, lung disease, and age-related pulmonary, cardiac and cerebrovascular dysfunction, our findings have important translational implications. First, they point to a potential pathogenic role of chronic hypoxia in triggering BBB disruption and subsequent neurological dysfunction, and second, they demonstrate an important protective role for microglia in maintaining vascular integrity in the hypoxic brain.

Keywords: Angiogenesis; Blood vessels; Blood–brain barrier integrity; Brain; Chronic mild hypoxia; Endothelial proliferation; Fibrinogen; Microglia.

Fig. 1

Chronic mild hypoxia (CMH) triggers vascular leak in cerebral blood vessels associated with microglial clustering. Frozen brain sections taken from mice exposed to normoxia or 7 days hypoxia (8% O₂) were stained for the following markers: a the endothelial marker CD31 (AlexaFluor-488) and fibrinogen (Cy-3); d fibrinogen (Cy-3) and Mac-1 (AlexaFluor-488); e CD31 (AlexaFluor-488), fibrinogen [Cy-5 (blue)] and Mac-1 (Cy-3); f CD31 (AlexaFluor-488), IgG (Cy-3) and Tmem119 [Cy-5 (blue)]; and g, Mac-1 (AlexaFluor-488). All images were captured in the brainstem. Scale bars = 100 μm (a) and 50 μm (d–g). b, c Quantification of the number of leaky (fibrinogen-positive) blood vessels/FOV (b) or total area of vascular leak/FOV (c) in the brainstem after 4, 7, or 14 days hypoxia. h Quantification of the morphological categorization of microglia under different conditions. All results are expressed as the mean ± SEM (n = 4 mice/group). **p < 0.01 versus normoxic conditions. Note that CMH provoked transient vascular leak in brainstem blood vessels that was associated with wrapping of Mac-1/Tmem119-positive microglial processes around the damaged vessel and with a morphological switch from ramified to activated morphology (g)

Fig. 2

The extent of CMH-induced cerebrovascular leak is region-specific. a Frozen brain sections taken from mice exposed to 7 days hypoxia (8% O₂) were dual-labelled for CD31 (AlexaFluor-488) and fibrinogen (Cy-3). Images were captured in four different brain regions: brainstem (BS), olfactory bulb (OB), cerebral cortex (CX) and cerebellum (CB). Scale bar = 200 μm. b, c Quantification of the number of leaky (fibrinogen-positive) blood vessels/FOV (b) or total area of vascular leak/FOV (c) after 7 days hypoxia. Results are expressed as the mean ± SEM (n = 4 mice/group). *p < 0.05; **p < 0.01. Note that hypoxia-induced vascular leak was significantly higher in the brainstem and olfactory bulb versus the cerebral cortex and cerebellum

Fig. 3

Microglial depletion results in greater vascular leak during CMH. a Frozen brain sections taken from mice fed normal chow or PLX5622-containing chow and maintained under normoxic conditions for 7 days were stained for the microglial marker Mac-1 (AlexaFluor-488) and DAPI. b Quantification of microglial depletion after 7 days PLX5622 in the brainstem (BS), olfactory bulb (OB), cerebral cortex (CX) and cerebellum (CB). Results are expressed as the mean ± SEM (n = 4 mice/group). **p < 0.01. Note that 7 days PLX5622 reduced microglial density in all brain regions examined to less than 10% of untreated controls. c Images of brainstem taken from mice fed normal chow or PLX5622-containing chow and maintained under hypoxic conditions for 7 days stained for CD31 (AlexaFluor-488) and fibrinogen (Cy-3). d, f Quantification of the number of leaky vessels/FOV (d) or total area of vascular leak/FOV (f) in the brainstem (BS), olfactory bulb (OB), cerebral cortex (CX) and cerebellum (CB) in mice fed normal chow or PLX5622-containing chow and maintained under hypoxic conditions for 7 days. Results are expressed as the mean ± SEM (n = 4 mice/group). *p < 0.05. Note that all regions of brain examined in PLX5622-treated mice showed a much higher number of leaky blood vessels. e CD31/fibrinogen/Mac-1 triple-IF of control chow mice confirmed microglial clustering and elevated levels of Mac-1 expression by microglia surrounding the leaky vessel, but absence of microglial clustering in PLX5622-fed mice. Scale bars = 100 μm (a, c) and 50 μm (e)

Fig. 4

Under hypoxic conditions, absence of microglia results in increased MECA-32 expression and region-specific astrocyte-vascular uncoupling. a, c. Frozen brain sections (images captured in brainstem) taken from mice fed normal chow or PLX5622-containing chow and maintained under hypoxic conditions for 7 days were stained for CD31 (AlexaFluor-488) and AQP4 (Cy-3) in panel A or CD31 (AlexaFluor-488) and MECA-32 (Cy-3) in panel C. Scale bar = 50 μm. b, d. Quantification of the % of blood vessels expressing AQP4 (b) or number of blood vessels expressing MECA-32/FOV (d) in the brainstem (BS), olfactory bulb (OB), cerebral cortex (CX) and cerebellum (CB). Results are expressed as the mean ± SEM (n = 4 mice/group). *p < 0.05; **p < 0.01. Note that under hypoxic conditions, PLX5622-fed mice contained a significant number of cerebral blood vessels in the brainstem and olfactory bulb that lacked AQP4 expression (see arrows in a), though this AQP4 loss was not observed in the cerebral cortex or cerebellum (b). In addition, in mice fed PLX5622, all brain regions examined (brainstem, olfactory bulb, cerebral cortex and cerebellum) showed a higher number of vessels expressing MECA-32 compared with normal chow-fed controls

Fig. 5

Microglial depletion results in greater loss of endothelial tight junction protein expression during CMH. a, c Frozen brain sections (images captured in brainstem) taken from mice fed normal chow or PLX5622-containing chow and maintained under hypoxic conditions for 7 days were stained for CD31 (AlexaFluor-488) and ZO-1 (Cy-3) in a or CD31 (AlexaFluor-488) and occludin (Cy-3) in c. Scale bar = 50 μm. b, d. Quantification of endothelial expression of ZO-1 (b) or occludin (d) on blood vessels in the brainstem. Results are expressed as the mean ± SEM (n = 4 mice/group). *p < 0.05. Note that under hypoxic conditions, brains of PLX5622-fed showed focal areas in which blood vessels showed diminished expression of ZO-1 and occludin (see arrows in a, c)

Fig. 6

Levels of hypoxia-induced endothelial proliferation are region-specific. a Frozen brain sections taken from mice exposed to hypoxia (8% O₂) for 4 days were stained for CD31 (AlexaFluor-488) and the proliferation marker Ki67 (Cy-3). Images were captured in the brainstem, olfactory bulb, cerebral cortex, and cerebellum. Scale bar = 100 μm. b Quantification of the number of proliferating endothelial cells (CD31 + /Ki67 + cells)/FOV. Results are expressed as the mean ± SEM (n = 4 mice/group). *p < 0.05; **p < 0.01. Note that hypoxia-induced endothelial proliferation was significantly higher in the brainstem and olfactory bulb versus the cerebral cortex and cerebellum