Alzheimer's disease (AD) is the most common neurodegenerative disease, accounting for 60%−80% of all cases of dementia (Sonkusare et al., 2005). AD is characterized by various pathological markers in the brain, such as the accumulation of beta-amyloid peptide (Aβ), which can form senile plaques, intracellular accumulation of neurofibrillary tangles formed by hyperphosphorylated tau protein, and progressive loss of nerve cells (Scheltens et al., 2016). Most cases of AD are sporadic and aging is considered a major risk factor for AD, but the pathways through which aging triggers the development of the disease are still unclear. It has been suggested that aging may induce post-translational modifications of Aβ (PTMs), which enhance its pathogenic properties (Moro et al., 2018). Thus, Aβ is capable of undergoing various PTMs that are triggered by enzymes or low molecular weight substances, as well as spontaneously (Barykin et al., 2017). Some of these modifications are isomerization of the aspartic acid residue at position 7 (iso-Aβ) and phosphorylation at serine 8 (pS8-Aβ). These modifications are located in the metal-binding domain of Aβ, which regulates its zinc-dependent oligomerization (Zirah et al., 2006; Barykin et al., 2018) and interaction with receptors (Barykin et al., 2018; Forest et al., 2018). In amyloid plaques, iso-Aβ was found to constitute more than 50% of all Aβ molecules (Mukherjee et al., 2021). Iso-Aβ has an increased ability to oligomerize (Shimizu et al., 2005), is more toxic (Mitkevich et al., 2013) and demonstrates resistance to proteolysis (Kummer and Heneka, 2014). At the same time, the level of iso-Aβ increases with age and in patients with AD (Moro et al., 2018). PS8-Aβ was detected in brain tissue of both patients with AD and AD model mice. It is localized both in amyloid plaques and in the cytoplasm of neurons, and compared to unmodified Aβ has increased neurotoxicity in vitro (Jamasbi et al., 2017) and higher resistance to degradation by an insulin-degrading enzyme (Kummer and Heneka, 2014). Thus, pS8-Aβ and iso-Aβ are important isoforms that differ significantly in properties from intact Aβ. The changes in the homeostasis of these isoforms may trigger pathological events contributing to development of AD.

Numerous studies have shown that AD is accompanied by a disruption of the blood-brain barrier (BBB), which occurs at an early stage of the disease (Nation et al., 2019; Barisano et al., 2022). The BBB controls the entry of Aβ from plasma into the brain via the RAGE receptor, as well as the clearance of Aβ from the brain into the peripheral circulation via the LRP-1 receptor (Zenaro et al., 2017). Disruption of these BBB functions can lead to pathological accumulation of Aβ in the brain and manifestation of AD symptoms. Increasing evidence indicates that Aβ from blood can enter the brain and serve as a trigger for the disease (Bu et al., 2018; Sun et al., 2021). Interestingly, peripheral injection of synthetic Aβ₄₂ into the bloodstream did not lead to the formation of amyloid plaques in the brains of mouse models of AD. However, intravenous injections of modified forms of Aβ altered the pathology of AD: the injection of iso-Aβ accelerated the amyloidogenesis (Kozin et al., 2013), while injection of pS8-Aβ reduced the number of amyloid plaques in the brain of transgenic mice (Barykin et al., 2018). This evidence suggests that pathogenic isoforms of Aβ may arise in the circulatory system, after which they penetrate the brain and contribute to AD pathology (Kozin and Makarov, 2019). However, the transport of modified forms of Aβ across the BBB has not been previously studied.

In this work, we compared the efficiency of transport of Aβ₄₂, pS8-Aβ₄₂, and iso-Aβ₄₂ through a monolayer of BBB endothelial cells, and also established the contribution of clathrin- and caveolin-dependent mechanisms to this process. It was also determined how modifications of Aβ affect its affinity for RAGE.

It is known that Aβ is expressed not only in the brain, but also in cells of other organs and tissues: kidneys and adrenal glands, heart, liver, spleen, pancreas, as well as in muscles, blood cells and endothelium (Roher et al., 2009). Significant amounts of Aβ have been found in human red blood cells, and the Aβ₄₂/Aβ₄₀ ratio in red blood cells is higher than in plasma (Kiko et al., 2012). Platelets also express large amounts of amyloid precursor protein (APP) and release beta-amyloid. Once activated, for example by bleeding, platelets can secrete significant amounts of Aβ into the blood (Humpel, 2017; Carbone et al., 2021). Several studies have shown that platelets from people with AD may have a greater tendency to become activated and therefore release Aβ into the blood (Carbone et al., 2021).

Increasing evidence indicates that peripheral Aβ can penetrate into the brain and play a significant role in the pathogenesis of AD. Thus, peripheral inoculation of brain extracts containing Aβ led to amyloidosis in the brain of mice (Eisele et al., 2010, 2014; Burwinkel et al., 2018). It has also been shown that increasing the concentration of peripheral Aβ significantly reduces its removal from the brain (Marques et al., 2009). Inhibition of RAGE-ligand interaction suppressed brain Aβ accumulation in a transgenic mouse model (Deane et al., 2003). The important role of peripheral Aβ and its ability to enter the brain and trigger AD pathology was further highlighted in a parabiosis model in which the circulatory systems of a transgenic mouse with AD-like pathology and a wild-type mouse were connected. Using this model, the researchers demonstrated that human Aβ derived from a transgenic animal entered the wild-type mouse brain and initiated AD-like pathology, including tau hyperphosphorylation, neurodegeneration, neuroinflammation, impaired hippocampal long-term potentiation, and amyloid plaque formation (Bu et al., 2018). Another study demonstrated the contribution of Aβ produced by blood cells to the pathogenesis of AD: when bone marrow was transplanted from transgenic mice to wild-type mice, the latter showed signs of AD pathology (Sun et al., 2021). A number of data indicate that induction of AD requires not just an increase in the concentration of Aβ₄₂, but the appearance of pathogenic forms carrying post-translational modifications (Kummer and Heneka, 2014; Barykin et al., 2017). Thus, intravenous administration of iso-Aβ₄₂ accelerates amyloidogenesis in the brain of transgenic mice modeling AD (Kozin et al., 2013), and introduction of pS8-Aβ₄₂ into the blood, on the contrary, reduces the number of amyloid plaques (Barykin et al., 2018). At the same time, intravenous administration of the unmodified peptide does not affect the formation of amyloid plaques in the brain of model mice. It is possible that modified forms of Aβ arise in the circulatory system, after which they enter the brain and contribute to AD pathology (Kozin and Makarov, 2019).

In this work, we compared the efficiency of transport of Aβ isoforms in an in vitro model of the BBB, and also determined the contribution of different mechanisms of endocytosis to the passage of Aβ₄₂, pS8-Aβ₄₂ and iso-Aβ₄₂ through the endothelium. It was found that pS8-Aβ₄₂ and iso-Aβ₄₂ are better transported by BBB endothelial cells than Aβ₄₂ (Figure 1), which may be one of the factors determining the ability of modified forms of Aβ to influence cerebral amyloidogenesis when administered intravenously (Kozin et al., 2013; Barykin et al., 2018).

The main mechanism of transport of Aβ from the bloodstream to the brain is caveolin-dependent endocytosis (Zhu et al., 2018). Indeed, the inhibitor of this form of endocytosis, filipin, suppressed the transport of Aβ₄₂, pS8-Aβ₄₂, and iso-Aβ₄₂ from the upper to lower compartment to the same extent (Figure 4D). Strikingly, the addition of chlorpromazine, which is an inhibitor of clathrin-dependent endocytosis, significantly suppressed the transport of iso-Aβ₄₂ (Figure 5D). Thus, the transport of iso-Aβ₄₂ may also be dependent on clathrin endocytosis. Also, the contribution of clathrin endocytosis was found for Aβ₄₂ and pS8-Aβ₄₂, but less pronounced than for iso-Aβ₄₂. The involvement of clathrin-dependent endocytosis in transport of proteins from the bloodstream to the brain was previously shown for transferrin and insulin receptors (Roberts et al., 1992; Goulatis and Shusta, 2017; Ayloo and Gu, 2019; Pemberton et al., 2022), but not for beta-amyloid peptides. There is also evidence that LRP-1 can mediate Aβ transport in both directions (Pflanzner et al., 2011). It is possible that in the bEnd.3 cell line some part of the molecules of this receptor is present on the luminal side, which could explain the slight effect of the inhibitor on the transport of Aβ₄₂ and pS8-Aβ₄₂.

It is assumed that RAGE plays a major role in the transfer of Aβ from the circulatory system to the brain. It was previously shown that in cells expressing RAGE an inhibitor of this receptor, FPS-ZM1, prevented oxidative stress induced by Aβ₄₀ and Aβ₄₂ (Deane et al., 2012). However, the effect of FPS-ZM1 on the transport of Aβ and its isoforms across the BBB endothelium in vitro has not been studied. We found that FPS-ZM1 reduced the passage of Aβ₄₂ through the endothelium of the BBB (Figure 6), which correlates well with data obtained previously for Aβ₄₂ in vivo (Deane et al., 2003, 2012). FPS-ZM1 also inhibited the transport of pS8-Aβ₄₂ and iso-Aβ₄₂, but the effect of this inhibitor on the passage of pS8-Aβ₄₂ was less pronounced than for other isoforms (Figure 6D). Thus, it appears that RAGE is the major receptor in the transport of both Aβ and its modified forms across the BBB.

Since Aβ₄₂, pS8-Aβ₄₂, and iso-Aβ₄₂ differed in their ability to penetrate the cell monolayer, we decided to compare the ability of these isoforms to interact with RAGE. There is relatively little data in the literature on the interaction parameters of Aβ with RAGE. Thus, in cell cultures, the dissociation constants of RAGE with Aβ₄₀ and Aβ₄₂ were 75 ± 5 nM (Deane et al., 2012) and 92 ± 40 nM (Chellappa et al., 2021), respectively. For purified RAGE, a dissociation constant with Aβ₄₀ was shown to be 57 ± 14 nM (Yan et al., 1998). Using the surface plasmon resonance method, it was revealed that sRAGE binds Aβ₄₂ oligomers with a Kd of 17 nM (Chen et al., 2007), and the Kd for endogenous soluble RAGE (esRAGE) and Aβ₄₂ was 44.9 nM (Sugihara et al., 2012). Thus, direct measurements of the interaction of Aβ₄₂ monomers and its isoforms with RAGE have not been previously carried out. The interaction constants obtained for Aβ₄₂ are an order of magnitude higher compared to constants estimated in other systems. This may be due to the fact that in our experiments stabilizing agents and other additives that are far from physiological were used, which could affect the obtained constants. Nevertheless, this model allowed us to compare the binding of different isoforms with RAGE in the same conditions. Across the three Aβ₄₂ isoforms, we found that RAGE demonstrates the highest affinity to Aβ₄₂ and the lowest to iso-Aβ₄₂ (Figure 7). These data are in good agreement with the results of computer modeling that we obtained earlier, according to which sRAGE has the lowest calculated Kd value with Aβ₄₂ and the highest with iso-Aβ₄₂ (Tolstova et al., 2022). The obtained Kd values correlate with the accumulation of amyloid peptides inside cells (Figure 9). We also found that RAGE interacts with Aβ_{17 − 42}, but not with Aβ_{1 − 16}, and the binding constant of the receptor with Aβ_{17 − 42} was an order of magnitude smaller than the binding constant with the full-length Aβ₄₂ peptide. Previously, we observed a similar pattern in the interaction of Aβ with Na⁺/K⁺-ATPase: binding to the enzyme was detected for Aβ_{17 − 42}, but not for Aβ_{1 − 16} (Barykin et al., 2018). Probably, the hydrophobic C-terminal fragment Aβ_{17 − 42} makes a major contribution to the binding of Aβ₄₂ to various protein molecules, while Aβ_{1 − 16} modulates this interaction.

Apparently, the high affinity of Aβ₄₂ for RAGE is the reason for its accumulation in cells and lower transport efficiency compared to other isoforms, while the isoforms with lower affinity for the receptor are more easily transported across the endothelial cell and are able to dissociate from the receptor on the abluminal side. This mechanism was previously shown for the passage of antibodies to the transferrin receptor across the BBB (Yu et al., 2011; Goulatis and Shusta, 2017). High-affinity antibodies against the transferrin receptor cause the antibody-receptor complex to be mainly directed to lysosomes, and those that undergo transcytosis remain associated with the receptor on the abluminal side. Low-affinity antibodies undergo transcytosis and dissociate on the abluminal side to a greater extent (Yu et al., 2011; Goulatis and Shusta, 2017). Similar studies focusing on drug delivery to the brain showed that transferrin-containing nanoparticles with high avidity for the transferrin receptor remained tightly associated with endothelial cells, whereas low avidity nanoparticles dissociated from the receptor after transcytosis (Wiley et al., 2013). In bEnd.3 cells, it was shown that strong binding of ligands to LRP-1 triggers internalization leading to endo-lysosomal sorting and degradation of ligand-receptor complex, while ligands with moderate binding strength to the receptor were transported across the endothelium (Tian et al., 2020). Thus, the stronger binding of Aβ₄₂ to RAGE may be the reason for its lowest transport efficiency of all isoforms across the bEnd.3 cell monolayer. Another factor influencing Aβ transport across the BBB may be different degrees of enzymatic degradation of Aβ isoforms. Thus, isomerization of the aspartate residue in Aβ has been shown to prevent its proteolysis in lysosomes (Lambeth et al., 2019). PS8-Aβ is resistant to degradation by insulin degrading enzyme, unlike unmodified Aβ (Kummer and Heneka, 2014). We found different affinities of Aβ isoforms for RAGE, which may affect enzymatic degradation.

Once Aβ enters the brain, it exerts multiple effects on its neuronal and glial targets (Mroczko et al., 2018). Phosphorylated and isomerized isoforms of Aβ act differently; as such, iso-Aβ is likely more toxic to cholinergic neurons bearing certain receptor types, such as alpha7 nicotinic acetylcholine receptor (Barykin et al., 2019), and pS8-Aβ is less prone to inhibit Na,K-ATPase (Barykin et al., 2018). Together with different transport rates of these isoforms, a complex interaction emerges. Aβ₄₂ has also been shown to bind to pyramidal neurons after administration into the blood (Clifford et al., 2007). However, the distribution of blood-derived Aβ isoforms in the brain is a subject for future research.

Aging may cause the appearance of modified Aβ isoforms (Moro et al., 2018). Thus, with age, isomerized and deaminated proteins accumulate, and the balance of phosphorylation/dephosphorylation is disrupted (Barykin et al., 2017). Pathogenic forms of Aβ carrying post-translational modifications can arise in the blood and then penetrate the brain, induce aggregation of endogenous beta-amyloid and cause AD pathology. Thus, the appearance of modified forms may precede the formation of plaques and occur in the early stages of the disease. We hypothesize that PTMs are more relevant to sporadic Alzheimer's disease than to familial Alzheimer's disease. However, genetic mutations can also lead to disruption of the PTM process if these mutations affect Aβ-modifying enzymes.

According to our data, phosphorylated and isomerized Aβ are transported more efficiently across the endothelium of the BBB than the unmodified peptide. The RAGE receptor was found to be essential for the transport of both Aβ₄₂ and its isoforms across the BBB. Differences in the transport of Aβ₄₂, pS8-Aβ₄₂, and iso-Aβ₄₂ may be due to different mechanisms of endocytosis or different affinities of these isoforms for the RAGE receptor. The mechanisms of transport of Aβ₄₂, pS8-Aβ₄₂ and iso-Aβ₄₂ across the BBB should be taken into account when developing agents for the treatment of AD. Thus, the data obtained may contribute to understanding the causes of the disease, as well as to the search for new drugs that prevent the accumulation of pathogenic Aβ isoforms in the brain.

Limitations: The use of higher than physiological concentrations of Aβ (1 μM) to study its transport is a limitation of this article. However, no saturation transport occurs at 1 μM (Supplementary Figure 1), which justifies the use of this concentration in our study. There are also a number of studies that use high concentrations of Aβ to study its transport across the BBB (Shackleton et al., 2016; Dal Magro et al., 2019; Shubbar and Penny, 2020; Zinchenko et al., 2020). Last, our study is the first to measure all isoforms of Aβ using single ELISA, and the sensitivity of our assay does not allow to measure lower concentrations.

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

KV: Formal analysis, Investigation, Methodology, Writing – original draft. IP: Conceptualization, Investigation, Methodology, Supervision, Writing – review & editing. VM: Conceptualization, Funding acquisition, Project administration, Resources, Supervision, Validation, Writing – review & editing. EB: Conceptualization, Formal analysis, Investigation, Methodology, Writing – original draft. AM: Conceptualization, Funding acquisition, Project administration, Resources, Supervision, Writing – review & editing.