Brain endothelial LRP1 maintains blood–brain barrier integrity

Neuronal function requires tight regulation of the cerebral microenvironment, which is achieved through specialized brain barriers such as the blood–brain barrier (BBB) [1]. Dysfunction of these barriers lead to neuronal degeneration and cognitive decline [2]. A recent report demonstrated that global endothelial loss of the endocytic and cell signaling protein low-density lipoprotein receptor–related protein 1 (LRP1) utilizing a constitutive Tie2-Cre line results in increased brain penetration of blood-borne molecules such as IgG and fibrinogen, progressive neuronal damage and behavioral deficits in mice [3]. Nikolakopoulou and colleagues identified a cyclophilin A–matrix metalloproteinase (MMP)-9 pathway in the Lrp1-deficient endothelium underlying BBB impairment: Deletion of LRP1 elevates cyclophilin A levels, which increases metalloproteinase-9-mediated tight junction protein degradation which allows the paracellular brain penetration of blood proteins leading to neuronal damage. Notably, LRP1 gene therapy targeting the BBB partially reversed vascular leakage, neuronal damage and behavioral deficits in mice.

Whilst these findings have broad implications for understanding how loss of endothelial LRP1 contributes to brain pathology, some questions for the audience remain. It is incompletely described how transcellular passage of molecules contributes to brain leakage of blood-borne molecules. Tight junctions are not the only regulator of BBB permeability. The authors did not see any effects of Lrp1 deletion on pericyte coverage or endothelial MFSD2a and GLUT1 levels, known modulators of transcellular transport processes [4, 5]. However, endothelial solute and adenosine triphosphate binding cassette (ABC) efflux transporters such P-glycoprotein (P-gp, also known as ABCB1 or MDR1) limit the entry of many xenobiotics and endogenous molecules that might damage neuronal cells [6‐8]. ABC transporter expression is regulated by various modulators such as notch signaling, wnt/β-cantenin signaling, pregnane X receptor, or peroxisome proliferator-activated receptors (PPAR) signaling [8‐10]. Notably, it has been shown that endothelial LRP1 is a coactivator of the nuclear receptor PPARγ and directly participates in gene transcription [11]. If the PPAR signaling co-activator LRP1 is missing, ABC transporter levels could be altered. So, it is possible that, in addition to the described paracellular leakage described by Nikolakopoulou and colleagues, transcellular passage is altered due to a change in efflux transporters such a P-gp. Therefore, the increased permeability in Lrp1^lox/lox; Tie2-Cre described in the recent paper would only be the result of increased paracellular passage through a lack of tight junctional proteins.

The second question remains regards the specificity of the Cre mouse line that was used in the study: how does a constitutive and global deletion of Lrp1 in all endothelial cells contributes to the brain-related findings described in Nikolakopoulou et al.? In the study, the authors used a pan-endothelial expression of the Cre (Tie2-Cre) that targets all peripheral and CNS vasculature during development as well as adulthood. However, LRP1 is expressed in all endothelial cells throughout the organism and is involved in many endocytic and cell signaling events during development as well as angiogenesis [12, 13]. Therefore, Tie2-Cre-mediated Lrp1 deletion could have averse off targets effects that contribute to the described pathology. Interestingly, we collected data showing similar results regarding BBB leakage in mice using a conditional, tamoxifen-inducible Slco1c1-CreER^T2 not targeting peripheral vasculature and allowing the time-specific deletion of LRP1 in the vasculature of the brain [14].

The solute carrier organic anion transporter Slco1c1 is highly enriched in CNS vasculature over peripheral vasculature in mice [15]. Therefore, in Slco1c1-CreER^T2 mice, where Cre-mediated gene excision is controlled by the Slco1c1 promotor, genetic deletion of a gene of interest can be achieved by tamoxifen injection specifically in brain endothelium and choroid plexus epithelium [14]. Previously, we developed Lrp1^lox/lox; Slco1c1-CreER^T2 mice that allow the deletion of Lrp1 in brain endothelial cells in adult mice. This strategy reduces potential off target side effects compared to pan-endothelial Cre-lines like Tie2-Cre by leaving peripheral vasculature unaffected. Moreover, in contrast to a constitutive, global Tie2-Cre-driven promoter [16], temporal induction of gene deletion in adult Slco1c1-CreER^T2 mice can rule out any potential off target effects of Lrp1 deletion during development. Using Lrp1^lox/lox; Slco1c1-CreER^T2 mice, we sought to investigate whether a knockout of Lrp1 in CNS endothelial cells has any effect on barrier integrity of the BBB. When we cultured primary brain endothelial cells, we found that cells from Lrp1^lox/lox; Slco1c1-CreER^T2 mice showed lower levels of tight junctional proteins claudin-5 and zonula occludens-1 (ZO-1) compared to brain endothelial cells isolated from Lrp1^lox/lox littermates expressing LRP1 (Fig. 1A–C). Moreover, in freshly isolated brain endothelial cells of Lrp1^lox/lox; Slco1c1-CreER^T2 mice, claudin-5 and occludin protein levels were markedly reduced compared to littermate controls (Fig. 1D). At the same time, the levels of cyclophillin A, an activator of a MMP-mediated tight junction degradation pathway in endothelial cells [3], were significantly elevated (Fig. 1F). Utilizing primary brain endothelial cells from Lrp1^lox/lox; Slco1c1-CreER^T2 mice we found higher MMP activity (Fig. 2A) using a fluorogenic MMP substrate, lower transendothelial resistance (Fig. 2B) measured by impedance spectroscopy along with increased ¹⁴C-inulin permeability across an endothelial monolayer compared littermate control cells (Fig. 2C). These results suggested that elevated MMP activity in Lrp1^lox/lox; Slco1c1-CreER^T2 endothelial cells resulted in enhanced endothelial permeability due to tight junction degradation.

Disrupted tight junctions can lead to an extensive influx of hematogenous fluid and into the extravascular space leading to progressive elevation of brain water content and tissue swelling. Therefore, investigated whether the prolonged lack of LRP1 in brain endothelial cells has any effect on the brain penetration of endogenous IgG and water in aged mice. In 20-month-old mice, we detected substantially increased IgG levels in the cerebrospinal fluid (CSF) of Lrp1^lox/lox; Slco1c1-CreER^T2 mice compared to Lrp1^lox/lox littermates (Fig. 3A [17‐19]. Moreover, Lrp1^lox/lox; Slco1c1-CreER^T2 showed a higher brain water content (Fig. 3B) than their Lrp1^lox/lox littermates suggesting that there is increased influx of water into the brain due to an opening of the BBB [17, 18]. Of note, along with altered tight junctions, we found P-gp decreased (Fig. 1E, also reported in [20]). It remains to be determined by future studies what the functional consequences on the loss of P-gp are for brain penetration of xenobiotics in Lrp1^lox/lox; Slco1c1-CreER^T2 mice. However, a recent study shows a direct effect on brain uptake of P-gp substrate rhodamine123 upon changes in P-gp transcript levels [21]. Besides P-gp, other ABC transporter levels could be affected due to the lack of PPARγ coactivator LRP1. This will remain a subject for subsequent studies. Collectively, these data suggest that BBB permeability is increased by paracellular and potentially transcellular mechanisms when endothelial LRP1 is absent. In Lrp1^lox/lox; Slco1c1-CreER^T2 mice both body and brain weight were significantly reduced when the animals were housed on a constant tamoxifen-supplemented chow (Fig. 4A, B) suggesting that lack of endothelial LRP1 impairs homeostasis and metabolism as also suggested by earlier studies [11].

Unexpectedly, as reported earlier, we did not find any differences in BBB integrity in Lrp1^lox/lox; Slco1c1-CreER^T2 at 8 months [22] measured by endogenous IgG in the brain or brain penetration of intravenously-injected sodium-fluorescein brain. In these experiments, the mice were housed on a normal chow lacking tamoxifen after an initial 7-days treatment with tamoxifen at 8 weeks of age. Given the massive damage occurring to CNS vasculature reported here and in the study by Nikolakopoulou and colleagues [3], we are now questioning, whether a long-term brain endothelial LRP1 deletion will prevail over time in a conditional system in older mice when the driving Cre is not constitutively expressed and only a single treatment of tamoxifen early in adulthood is applied as it was done in the study. It has been shown that vascular damage recruits bone marrow-derived endothelial progenitor cells (sometimes also referred as circulating angiogenic cells) from the periphery for vascular repair [23‐25]. Different from a constitutive model as used by Nikolakopoulou and colleagues, a conditional model as used in our study could therefore regain gene expression over time by replacement of damaged endothelial cells with peripheral LRP1 expressing blood-circulating cells and therefore mask the initial effects of the initial knockout. To date, we have not tested whether long term tamoxifen deprivation after will result in a regain of LRP1 protein expression at the BBB. Further studies are needed to fully decipher the biological mechanisms underlying the effects seen in Lrp1^lox/lox; Slco1c1-CreER^T2mice.