Cerebral microhemorrhages (CMHs) are small focal bleeds in the brain that produce the magnetic resonance imaging signatures of cerebral microbleeds (CMBs) (Greenberg et al., 2009). CMBs increase with normal aging and under pathological conditions and are primarily hemorrhagic in origin (e.g., cerebrovascular disruption resulting in red blood cell (RBC) extravasation) (Martinez-Ramirez et al., 2014; Fisher et al., 2018; Lau et al., 2020). However, since CMBs are clinically identified as small magnetic resonance imaging signal voids resulting from the paramagnetic properties of iron in hemosiderin deposits, sources of CMBs that arise due to an increase in hemosiderin iron via alternate mechanisms have been suggested (Fisher, 2014; Janaway et al., 2014).

Under physiological conditions, the adhesion of RBCs to the vascular endothelium is negligible. However, an abnormal RBC-vascular endothelial adhesion has been observed in various pathological conditions (Bonomini et al., 2002; Nakagawa et al., 2011; Wautier and Wautier, 2013; Kucukal et al., 2018). Additionally erythrophagocytosis, a well-described feature of macrophages, by which injured/aged erythrocytes are ingested, digested, and cleared from the blood circulation, has been described in peripheral endothelial cells (Fens et al., 2010, 2012; Lee et al., 2011). A similar phagocytic phenotype has also been reported for the brain endothelium, and studies show that the brain endothelium can engulf and extrude emboli to the abluminal (brain) side via angiophagy, in vivo and in vitro (Grutzendler et al., 2014; van der Wijk et al., 2020). Similarly, our recent in vitro and in vivo work demonstrated an erythrophagocytic phenotype of the brain endothelium with subsequent passage of iron-rich hemoglobin and RBCs across the endothelial monolayer (Chang R. et al., 2018). The passage of iron-rich RBCs (or RBC degradation products) across an intact brain endothelium following erythrophagocytosis may present an alternate source of iron-rich CMH-like lesions (Fisher, 2014; Janaway et al., 2014; Chang R. et al., 2018).

Phosphatidylserine externalization on RBCs is well-recognized as an important signal that initiates their phagocytic removal from the circulation, and this mechanism is widely studied for macrophage-mediated erythrophagocytosis (de Back et al., 2014). Phosphatidylserine exposure on aged/injured RBCs is also a key trigger for erythrophagocytosis of oxidatively stressed or aged RBCs by peripheral endothelial cells (Fens et al., 2008, 2012). Similarly, our previous study reported a marked externalization of RBC phosphatidylserine by tert-butyl hydroperoxide (tBHP), an oxidative stressor used previously to trigger erythrophagocytosis by peripheral endothelial cells (Fens et al., 2012), and we found robust erythrophagocytosis of tBHP-RBCs by murine brain microvascular endothelial cells (Chang R. et al., 2018). Corroborating our findings, a recent study reported increased erythrophagocytosis of aged RBCs by the brain endothelium, which was associated with increased phosphatidylserine externalization and oxidative stress (Catan et al., 2019). These studies together show an association between RBC phosphatidylserine exposure and oxidative stress, and brain endothelial erythrophagocytosis, but a direct role of RBC phosphatidylserine externalization and oxidative stress in this process has not been reported.

The aim of the current study was to provide mechanistic insights into the cellular pathways involved in brain endothelial erythrophagocytosis. Murine RBCs were treated with tBHP to induce oxidative stress and phosphatidylserine externalization to study the role of these mediators in brain endothelial erythrophagocytosis in vitro using murine brain microvascular endothelial cells (bEnd.3). We further examined the impact of brain endothelial erythrophagocytosis on the iron-transport machinery of the bEnd.3 cells and the intracellular trafficking of phosphatidylserine exposing tBHP-RBCs by tracing their movement through early endosomes toward late endosomes, the latter being the presumed site of RBC degradation and release of iron-rich degradation products. Finally, we injected tBHP-RBCs into aged mice to provide histological evidence for increased iron-rich Prussian blue-positive lesion development in mice. Overall, our findings provide important mechanistic insights into the cellular pathways involved in, and the alterations to cellular iron homeostasis associated with, brain endothelial erythrophagocytosis.

Details about the source of the major resources used are in the Supplementary Material.

Our previous work showed that oxidatively stressed RBCs can undergo robust erythrophagocytosis by the brain endothelium, and this process was accompanied by an increase in the migration of an iron-rich RBC degradation product (hemoglobin) and RBCs across the brain endothelium, in vitro and in vivo, in the absence of vascular disruption (Chang R. et al., 2018). In the present report, we elucidate the cellular mechanisms that contribute to brain endothelial erythrophagocytosis. We show that phosphatidylserine cloaking and oxidative stress reduction significantly attenuate brain endothelial erythrophagocytosis in the in vitro murine brain microvascular endothelial cell model. Further, brain endothelial erythrophagocytosis alters endothelial iron homoeostasis and increases abluminal (brain) iron, without loss of the brain endothelial monolayer integrity, in vitro. Finally, we provide histological evidence for increased iron-rich Prussian blue-positive lesion load in mice injected with tBHP-RBCs, correlating with the in vitro increase in abluminal iron.

RBCs have a life span of 120 days in humans and about 45 days in mice (Gottlieb et al., 2012). ROS accumulation has been suggested to be a key RBC-life-span determinant by causing phosphatidylserine scrambling and externalization. Phosphatidylserine exposure contributes to the recognition of old/damaged RBCs by altering RBC adhesion to professional phagocytes (macrophages), which is followed by erythrophagocytosis and RBC clearance from the circulation (Franco, 2012; Gottlieb et al., 2012). Besides professional phagocytes, other cell types such as smooth muscle cells can phagocytose phosphatidylserine-exposing apoptotic bodies and aged RBCs (Garfield et al., 1975). Endothelial cells, which are in close contact with circulating RBCs, are also reported to have an erythrophagocytic phenotype. Fens et al. (2010) reported that peripheral endothelial cells can phagocytose RBCs subjected to increased oxidative stress and phosphatidylserine exposure. Similarly, a recent study showed that RBC aging and glycation increases phosphatidylserine externalization and oxidative stress in RBCs, which is associated with enhanced RBC phagocytosis by the vascular endothelium (Catan et al., 2019). Overall, the contribution of the endothelium to erythrophagocytosis and clearance of injured RBCs is now increasingly reported, as is the involvement of phosphatidylserine exposure and oxidative stress in endothelial erythrophagocytosis. Our recent work extended these findings to the brain endothelium and showed robust brain endothelial erythrophagocytosis of phosphatidylserine exposing tBHP-RBCs (Chang R. et al., 2018). To establish the definitive contribution of RBC phosphatidylserine exposure to brain endothelial erythrophagocytosis, herein we cloaked phosphatidylserine with annexin-V which has a high binding affinity for phosphatidylserine in a calcium-dependent manner (Figure 3A; Lizarbe et al., 2013). Cloaking of RBC phosphatidylserine with annexin-V significantly reduced erythrophagocytosis of tBHP-RBCs by the bEnd.3 cells (Figures 3B–F). Annexin-V also significantly reduced total and intracellular hemoglobin, an iron-rich RBC degradation product, in the bEnd.3 cells (Figure 3G), confirming the role of RBC surface phosphatidylserine exposure in brain endothelial erythrophagocytosis. Mechanistically, studies show that phosphatidylserine-dependent erythrophagocytosis by endothelial cells is mediated by the α_v integrins or phosphatidylserine-recognizing transmembrane receptors, stabilin-1 and stabilin-2 (Lee et al., 2011; Fens et al., 2012). Besides these, phosphatidylserine-recognizing receptors, including tyrosine kinase receptors, CD300, T-cell immunoglobulin, and mucin domain 1 and 4, are involved in the erythrophagocytosis of phosphatidylserine-exposing RBCs by macrophages, but the involvement of these receptors in endothelial erythrophagocytosis has not been reported and needs further investigation (Klei et al., 2017; Chang C. F. et al., 2018).

Besides RBC phosphatidylserine exposure, tBHP-RBCs showed a dose-dependent increase in intracellular ROS levels in the current study (Figure 1). Given that tBHP is a potent oxidative stressor (Fens et al., 2012; Chang R. et al., 2018), we investigated the involvement of oxidative stress in brain endothelial erythrophagocytosis. For this, we incubated tBHP-RBCs with vitamin C, a ROS scavenger. Vitamin C treatment resulted in a 90% reduction of ROS accumulation and phosphatidylserine exposure of tBHP-RBCs (Figures 4A,B). Accordingly, we observed a significant reduction in erythrophagocytosis of tBHP-RBCs, indicating that both ROS production and phosphatidylserine exposure play a vital role in the brain endothelial erythrophagocytosis (Figure 4). Notably, endothelial cell viability was not altered with brain endothelial erythrophagocytosis up to 48 h, which is consistent with the observations in human peripheral vascular endothelial cells (Catan et al., 2019). It should be noted that despite the almost complete reversal of phosphatidylserine exposure with phosphatidylserine masking with annexin-V (Figure 3) and the highest concentration of vitamin C (Figure 4), brain endothelial erythrophagocytosis was not completely abolished by these agents. These results suggest the involvement of alternate molecular mechanisms of brain endothelial erythrophagocytosis. In this regard, studies show the involvement of vascular cell adhesion molecule 1 (VCAM-1) and intercellular adhesion molecule 4 (ICAM-4) in erythrocyte adhesion to endothelial cells under pathological conditions (Wautier and Wautier, 2013). The expression of these adhesion molecules is increased by pro-inflammatory cytokines, including TNFα. Although we did not study the role of these adhesion molecules on brain endothelial erythrophagocytosis, treatment of brain endothelial cells and tBHP-RBCs with TNFα significantly enhanced brain endothelial erythrophagocytosis in a dose-dependent manner (Supplementary Figure 1), showing the role of inflammation in potentiating brain endothelial erythrophagocytosis, in vitro. Besides phosphatidylserine and adhesion molecules, CD47 on the RBC surface is a well-studied “eat me” or “don’t eat me” signal and regulates erythrophagocytosis through its interaction with signal-regulatory protein alpha that is expressed on phagocytes including endothelial cells (Burger et al., 2012). We found a significant 27% reduction in CD47 expression in tBHP-RBCs compared with PBS-RBCs (Supplementary Figure 4). Future work will be needed to determine the role of the CD47-signal-regulatory protein alpha pathway in brain endothelial erythrophagocytosis.

The intracellular processing of RBCs by macrophages following erythrophagocytosis has been extensively studied and involves engulfment of the damaged/injured RBCs into phagosomes via phagocytosis, fusion of the phagosome with the lysosome, lysosomal degradation of the RBC, and release of RBC degradation products into the cytosol (Korolnek and Hamza, 2015). Before fusing with the lysosomes, the phagosomes undergo a regulated sequence of fusion events, which involves fusion with the organelles in the endocytic pathway (Duclos et al., 2000). The endocytic pathway is composed of several distinct compartments based on their functions. For example, early endosomes play a primary role in sorting and recycling, while late endosomal compartments, such as lysosomes, serve as a final destination for proteolytic degradation (Hu et al., 2015). Our results demonstrate that a subset of internalized tBHP-RBCs is co-localized with both early and late endosome markers (Figure 5). To examine the fate of the remaining attached RBCs, tBHP- or PBS-RBCs were added to the bEnd.3 monolayer, incubated for 48 h, after which the bEnd.3 media were replaced with fresh media, and the attached RBCs were incubated with the bEnd.3 cells for another 72 h (all the free-floating RBCs were removed with the medium change). The tBHP-RBCs attached to the bEnd.3 cells after 72 h (RBCs appear as black spots in Supplementary Figure 2D) were reduced. The total cellular iron levels in the tBHP-RBC-treated bEnd.3 cells were not significantly different from the PBS-RBC-treated bEnd.3 cells, suggesting that the tBHP-RBCs were not present inside the bEnd.3 cells (data not shown). When we measured the iron in the media to determine if the RBCs had internalized and degraded and the resulting iron released to the media, we found a small but significantly higher iron level in the media collected from the bEnd.3 cells incubated with tBHP-RBCs compared with PBS-RBCs (Supplementary Figure 2E). These results suggest that with time, the engulfed RBCs are degraded within the lysosome, as implicated by RBCs co-localized with LAMP-1 (Figure 5), and the resulting iron released into the media. Contrary to this, a recent study showed that microspheres and emboli are retained in the brain endothelial cells without degradation, triggering their transmigration (van der Wijk et al., 2020). Our previous in vivo work showed a similar phenomenon of RBC migration across the brain endothelium (Chang R. et al., 2018), and some of the engulfed RBCs may eventually transmigrate across the brain endothelium. Overall, although the massive engulfment of tBHP-RBCs by some endothelial cells may overwhelm the abovementioned clearance mechanisms at some point, increase the intracellular iron pool, and trigger endothelial cell apoptosis, we did not observe overt endothelial cell death up to 5 days (48 + 72 h) following RBC incubation (Supplementary Figure 2D). However, we, cannot rule out apoptosis, although only a fraction of HUVEC cells showed signs of apoptosis post-erythrophagocytosis following 24 h of incubation (Fens et al., 2012).

We observed RBCs not only in the cytosol but also in the nucleus (Figure 5 and Supplementary Figure 2). Trafficking of cargo to the nucleus has been reported previously and been shown to impact cellular function and regulate gene expression (Olsnes et al., 2003; Chaumet et al., 2015). Further, cellular trafficking is dependent on the size of cargo, such that large cargos (3 μm) are directed to the nucleus for digestion (Keller et al., 2017). It is therefore conceivable that the large size of the RBC sorts them to the nucleus in the current study. This finding of RBC trafficking to the endothelial cell nucleus is novel, and its implication needs to be further investigated.

In the current study, brain endothelial erythrophagocytosis was accompanied with a significant increase in the intracellular iron pool (Figure 6), which is consistent with previous work showing an increase in cellular iron load with erythrophagocytosis by professional phagocytes (Knutson et al., 2003). Additionally, and more relevant from the standpoint of iron-rich CMH-like lesion development, we showed an increase in migration of iron across an intact endothelial cell monolayer, providing direct evidence for increased abluminal iron with brain endothelial erythrophagocytosis. These results build on our prior report that brain endothelial erythrophagocytosis was accompanied by passage of hemoglobin across the brain endothelial monolayer with unaltered brain endothelial monolayer integrity (Chang R. et al., 2018). Sources of abluminal iron in the current study could be hemoglobin or cytosolic iron (iron released after lysosomal degradation within the brain endothelial cells) which migrates across the brain endothelial monolayer. These results show that brain endothelial erythrophagocytosis is associated with an increase in abluminal (brain facing) iron load. Once in the brain parenchyma, microglial cells are well equipped to clear RBC and its degradation products, including hemoglobin, via scavenger receptors including CD36, CD47, CD163, and CD91 (Pan et al., 2020). Within the microglial cells, hemoglobin can be broken down into free heme followed by free iron that is stored as the insoluble hemosiderin-iron. Similarly, brain iron homeostasis is tightly regulated by the blood–brain barrier transferrin receptors, apotransferrin released by the oligodendrocytes, and iron transporters present on the ependymal surface. Transferrin-bound iron can also be taken up by astrocytes or neurons that express transferrin receptors. Any excess iron released from hemoglobin degradation may be stored as ferritin-iron (Wagner et al., 2003).

Increased cellular iron load after erythrophagocytosis is associated with alterations in iron transport proteins (Knutson et al., 2003). To elucidate the impact of brain endothelial erythrophagocytosis on the brain endothelial iron transporters, we measured levels of TfR, the major iron influx transporter, and ferroportin, the only known iron exporter at the blood–brain barrier (Ganz, 2005; Moos et al., 2007; McCarthy and Kosman, 2015). TfRs are localized to the luminal and abluminal surface of the brain microvasculature (Huwyler and Pardridge, 1998) and facilitate iron uptake into the brain microvascular endothelial cells and export into the brain via receptor-mediated transcytosis (Moos et al., 2007). Similarly, studies show that ferroportin is also localized to both the luminal and abluminal surfaces of the brain microvasculature to facilitate iron export from the brain endothelial cells (Wu et al., 2004; McCarthy and Kosman, 2015). In the current study, brain endothelial erythrophagocytosis was not associated with a change in TfR protein levels; however, ferroportin protein levels were significantly increased (Figure 6). This is consistent with an increase in ferroportin levels following erythrophagocytosis by macrophages (Knutson et al., 2003) and may be a compensatory response to maintain iron homeostasis and prevent iron overloading within the endothelial cells (Chiou et al., 2019). This also provides a potential explanation for why the brain endothelial cell viability and brain endothelial integrity are not compromised despite robust erythrophagocytosis and cellular iron increase (Figure 3 and Supplementary Figure 3).

Our previous work provided in vivo evidence for migration of tBHP-RBCs across an intact brain endothelium in Tie2-GFP mice, while the PBS-RBCs remained confined to the vasculature (Chang R. et al., 2018). These findings were recently corroborated by a study showing brain endothelial erythrophagocytosis of Plasmodium falciparum-infected RBCs in vivo (Adams et al., 2021). In the current study, we show that mice injected with tBHP-RBCs had a greater Prussian blue-positive CMH-like lesion load compared with mice injected with PBS-RBCs (Figure 7), supporting the in vitro findings of increased abluminal iron with brain endothelial erythrophagocytosis (Figure 6). Our previous work using ex vivo microscopy demonstrated an intact endothelium at the site of RBC extravasation (Chang R. et al., 2018) as opposed to a ruptured endothelium, which is seen with Prussian blue-positive lesions associated with blood–brain barrier damage (Sumbria et al., 2016). Further, we found no change in the blood–brain barrier tight junction protein ZO-1 seven days after tBHP-RBC injection (Figure 7). This is in line with our recent work showing no association between blood–brain barrier changes with CMH development in humans (Wadi et al., 2020). Accordingly, our in vitro data and other reports (Catan et al., 2019) show no change in brain endothelial cell viability and permeability with erythrophagocytosis (Figure 3 and Supplementary Figure 3). However, it is important to note that erythrophagocytosis of glycated RBCs is associated with alterations in endothelial function (e.g., reduction in migration and proliferation), and a small increase in apoptotic markers has been reported with endothelial erythrophagocytosis in vitro (Fens et al., 2012; Catan et al., 2019). Additionally, engulfment of microspheres by the brain endothelium and subsequent migration into the brain parenchyma is associated with temporary blood–brain barrier opening in a microembolization rat model (van der Wijk et al., 2020). Therefore, the increased RBC or iron load within the brain endothelial cells might trigger temporal endothelial remodeling or dysfunction resulting in blood–brain barrier changes. These changes may further increase the extravasation of RBCs and/or RBC degradation products into the brain, increasing the likelihood of iron-rich lesion development. Future work using intravital imaging will allow us to examine the causal and temporal relationship between brain endothelial erythrophagocytosis and blood–brain barrier permeability at the site of RBC extravasation using tracers followed by Prussian-blue staining to detect iron-rich lesions. These studies will also help elucidate how widespread brain endothelial erythrophagocytosis is, and if brain endothelial erythrophagocytosis-mediated abluminal iron increase is a focal or a diffuse event. Notably, our in vivo study showed focal iron deposits associated with tBHP-RBCs.

Overall, based on the results of the current study we propose the following sequence of events whereby brain endothelial erythrophagocytosis may result in iron-rich CMH-like lesion development (Figure 8): oxidative stress and phosphatidylserine exposure increase phagocytosis of RBCs by the brain endothelium. Other mechanisms including CD47 reduction may also contribute to this process. A subset of engulfed RBCs undergoes endosomal sorting and degradation. RBC engulfment and degradation increase intracellular pools of iron-rich RBC degradation products, hemoglobin and iron. Free iron may be effluxed from the brain endothelial cells via the iron exporter ferroportin. Additionally, hemoglobin may migrate across the brain endothelium (Chang R. et al., 2018). Both iron and hemoglobin migration across the brain endothelial cells increases brain abluminal iron, which results in Prussian blue-positive CMH-like lesion development.

The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding author.

The animal study was reviewed and approved by University of California-Irvine’s Institutional Animal Care and Use Committee, and carried out in compliance with University Laboratory Animal Resources regulations.

RKS and MJF conceptualized the study. JS, PV, RKS, and DHC assisted with the experimental design. JS, PV, SM, and RKS performed the experiments and data analysis. AP-H assisted with the data analysis. ACFN and WLL assisted with the in vivo study. JS and RKS wrote the first draft of the manuscript. RKS, DHC, and MJF acquired the funding. All authors discussed the results and their implications, and commented on the manuscript. All authors read and approved the final manuscript.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.