Preclinical and clinical studies indicate that aging is a critical factor inducing endothelial dysfunction (3, 46, 47). In line, brain endothelial dysfunction is frequently found during healthy brain aging as well as in neurological disorders such as stroke and Alzheimer’s disease (48–51). However, the role of brain endothelial aging and senescence in BBB impairment remains largely unknown. In this study, we evaluated the consequences of ERCC1 deficiency, a model for accelerated aging, in brain ECs in vitro and in vivo. We show that ERCC1-deficient brain ECs display reduced BBB integrity, increased transporter expression and more endothelial sprouting. We validated our in vitro findings in EC-KO mice, which display a higher expression of angiogenic genes and more capillary ECs and pericytes, specifically in the WM. Furthermore, the WM of the EC-KO animals demonstrates IgG leakage and increased glial cell reactivity near the vasculature, which coincided with immune cell infiltration in the brain parenchyma. Together, our work highlights the effect of endothelial cell aging on BBB dysfunction, angiogenesis and local inflammation.
In this study, shERCC1 display classical hallmarks of DNA damage and cellular aging including H2AX phosphorylation and increased SASP component expression (i.e. IL-1B, IL-6, VEGFA). Conversely, expression of CDKN1A (P21) and CDKN2A (P16), known senescence markers (52), was unaffected in shERCC1 cells. SASP and P21/P16 elevation are not always concomitant (53, 54), and the presence of the latter is not a pre-requisite for cell aging as exemplified by studies in post-mitotic cardiomyocytes (55, 56). Furthermore, in our set-up, the high proliferative capacity of hCMEC/D3 cells combined with ERCC1 deficiency, may highlight cellular aging features directly associated to DNA damage-related cell stress, while overshadowing the P21/P16 expression present in the few senescent cells with halted cell cycle. Lastly, VEGF, highly expressed in the shERCC1 cells, is known to negatively regulate P21/P16 expression (57), which may explain the similarity with NTC cells. The cellular stress of shERCC1 cells is also accompanied by a reduced expression of Cldn5 and VE-cad, resulting in an impaired barrier integrity, as seen during aging in vivo and in vitro (58, 59). Together, these results substantiate the role of DNA damage in inducing cell aging, and highlight the effect of brain EC aging on BBB function.
In vivo, EC-KO mice displayed increased vascular P21 expression compared to WT animals. However, not all brain ECs were P21+, which aligns with previous reports on this model suggesting partial efficiency of the Cre-lox system in deleting Ercc1 (28, 60, 61). P21 expression is increased in the brain endothelium of elderly compared to young individuals (60, 62) and MRI studies positively associate age and enhanced BBB permeability in healthy elderly suggesting the possible effect of aging on BBB dysfunction (63–65). In our study, EC-KO mice displayed increased IgG leakage, possibly underlying reduced BBB resistance resulting from EC aging, which is in line with our in vitro findings on BBB dysfunction. Furthermore, we observed an increase in Cldn5 and Cdh5 expression in the whole brain homogenate of EC-KO animals compared to WT, but no changes in Cldn5 expression in the WM vessels. Interestingly, we show an increased number of brain EC in WM capillaries, which could partially explain the increase in junction mRNA expression. In sum, our findings present an in vivo model for accelerated vascular aging, which recapitulates some of the features, including reduced BBB integrity, observed during healthy cerebrovascular aging in humans.
In EC-KO mice, the increased number of ECs in WM brain capillaries was concomitant with an enhanced expression of angiogenic markers (Angpt2, Kdr), suggesting increased EC sprouting. Furthermore, we observed a general increase in vascular SNAI2 expression in the WM on EC-KO animals. High Snai2 expression has been previously shown to directly impair the Dll4-Notch1-axis, resulting in angiogenesis characterized by dysfunctional vessels (44). Similarly, shERCC1 cells displayed increased SNAI2 and reduced DLL4 and NOTCH1 expression together with increased sprout number. These findings may indicate a reactivation of the angiogenic program in the WM of EC-KO mice sustained by Snai2. In line with our findings, previous studies found increased angiogenic markers (i.e. Angpt2) in brain ECs isolated from the corpus callosum of aged mice compared to younger animals (66). Interestingly, in our study we found the major changes in the WM of EC-KO mice, while the CRTX and HC seemed to be less affected, suggesting a regional susceptibility to EC aging. In humans, the WM has been previously shown to be more susceptible to age-related pathologies including vascular dementia (67, 68). A possible explanation may lie in the inherent lower capillary density of the WM, which makes this area more sensitive to hypoxic insults, a known trigger for angiogenesis via different pathways including Snai2 upregulation (66, 67, 69–73). Eventually, studies in aged mice and elderly show vessel rarefaction and reduced vessel length, which may be the result of dysfunctional angiogenesis (11, 74–78). Together our data suggest that DNA-damage in brain ECs may sustain dysfunctional angiogenesis via dysregulated Dll4-Notch1 signalling, and that the WM is more susceptible for this process. However, more research is warranted to fully comprehend the mechanisms underlying brain vasculature maintenance and remodelling during aging.
With age, a decrease in BBB transporter expression is observed (8, 79). However, both our endothelial aging models show increased transporter (P-gp and Mfsd2a) expression, specifically in the WM of EC-KO mice. P-gp expression can be primarily regulated by inflammation and oxidative stress, as evidenced by increased P-gp levels in stroke and seizure studies (80–82). Further, other senescent mouse models showed higher P-gp brain vasculature expression, postulating a protective role for senescent cells in toxin efflux from the aging brain (83, 84). Similarly, the increase in Mfsd2a might also be protective. Mfsd2a limits vesicle-mediated transcytosis, which is crucial to maintain BBB integrity, as shown by barrier leakage in Mfsd2a KO mice (85, 86). Under homeostatic conditions, increased Mfsd2a expression induces characteristics of cellular aging, while Mfsd2a overexpression alleviates tissue damage after acute brain injury (23, 87, 88). These evidences highlight the multifaceted role of Mfsd2a in the regulation of endothelial cell fitness. It is plausible that the increase of P-gp and Mfsd2a levels in the ERCC1 models is an early protective response to the impaired BBB integrity to aid CNS homeostasis. However, further studies are needed to validate this hypothesis.
The observed changes of the BBB in the EC-KO mice is accompanied by IgG leakage and loss of homeostatic marker expression in microglia in the WM. The leakage of blood-derived components such as fibrinogen has been reported to activate microglia in elderly and AD subjects (89), and to contribute to neuroinflammation and cognitive decline. Indeed, we also observed an increase in ICAM1+ vessels and leukocyte infiltration, including CD8+ T cells, into the WM of EC-KO mice. ICAM1 is a crucial mediator of the immune cell migration cascade and its p53-induced overexpression has been previously found in senescent endothelial cells of atherosclerotic lesions (90). Moreover, the presence and function of infiltrating peripheral immune cells in the aging brain, AD subjects, and age-related mouse models is increasingly described (91–94). In the aging brain, CD8+ T cells are suspected to interact with CNS cells including microglia and neurons (95), but their role is still heavily debated (96). Together these results position endothelial cell aging as one potential starting point of BBB inflammation and immune cell infiltration in the aging brain.
Collectively, our data highlight the significance of brain EC aging in BBB dysfunction, a key contributor in the development and progression of several diseases of the CNS. Hence, targeting brain vascular aging may be a promising strategy to alleviate age-related vascular disorders and their neurological implications.