Heparan sulfate proteoglycan in Alzheimer’s disease: aberrant expression and functions in molecular pathways related to amyloid-β metabolism

Abstract

Alzheimer’s disease (AD) is the most common form of dementia. Currently, there is no effective treatment for AD, as its etiology remains poorly understood. Mounting evidence suggests that the accumulation and aggregation of amyloid-β peptides (Aβ), which constitute amyloid plaques in the brain, is critical for initiating and accelerating AD pathogenesis. Considerable efforts have been dedicated to shedding light on the molecular basis and fundamental origins of the impaired Aβ metabolism in AD. Heparan sulfate (HS), a linear polysaccharide of the glycosaminoglycan family, co-deposits with Aβ in plaques in the AD brain, directly binds and accelerates Aβ aggregation, and mediates Aβ internalization and cytotoxicity. Mouse model studies demonstrate that HS regulates Aβ clearance and neuroinflammation in vivo. Previous reviews have extensively explored these discoveries. Here, this review focuses on the recent advancements in understanding abnormal HS expression in the AD brain, the structural aspects of HS-Aβ interaction, and the molecules involved in modulating Aβ metabolism through HS interaction. Furthermore, this review presents a perspective on the potential effects of abnormal HS expression on Aβ metabolism and AD pathogenesis. In addition, the review highlights the importance of conducting further research to differentiate the spatiotemporal components of HS structure and function in the brain and AD pathogenesis.

INTRODUCTION

Alzheimer’s disease (AD) is the most prevalent type of dementia, severely affecting patients’ quality of life (1). Currently, limited treatments are available to slow disease progression (2), but often lead to notable side effects (3). AD is marked by pathological lesions that include the accumulation of amyloid plaques in the cerebral parenchyma and vasculature (cerebral amyloid angiopathy, CAA), the formation of neurofibrillary tangles (NFT) in the neurons, and neuroinflammation (1, 4, 5). Mounting evidence supports the amyloid cascade hypothesis, which states that the accumulation and aggregation of amyloid-β peptides (Aβ) are crucial factors responsible for the initiation and acceleration of AD pathogenesis (6, 7). Compelling evidence suggests that amyloid plaques develop due to an imbalance between Aβ production and clearance from the brain. The equilibrium of Aβ production and clearance is determined by the expression and processing of the amyloid precursor protein (APP), the distribution and degradation of Aβ within the brain, and the elimination of Aβ via the cerebral blood vasculature (8). Therefore, a comprehensive understanding of the molecular mechanisms underlying amyloid plaque formation is essential for developing effective therapies to combat this devastating disease (9).

APP is a single-pass transmembrane protein that consists of a large N-terminal ectodomain and a shorter C-terminus that contains a crucial YENPTY sequence required for intracellular protein-sorting (Fig. 1) (10, 11). APP undergoes alternative splicing, generating eight isoforms, of which the 695 amino acid isoform is expressed predominantly in the central nervous system (CNS), whereas the 751 and 770 amino acid isoforms are more ubiquitously expressed (10, 11). APP is transported through secretory and endocytic pathways and undergoes sequential processing, leading to nonamyloidogenic and amyloidogenic end-products (Fig. 1). The nonamyloidogenic pathway begins with α-secretase cleaving the APP ectodomain to release soluble APPα (sAPPα) into the extracellular space. The remaining C-terminal fragment of 83 amino acids (CTF83) remains embedded in the plasma membrane. Cleavage of CTF83 by γ-secretase releases a small p3 fragment into the extracellular space and the APP intracellular domain (AICD) into the cytoplasm. The amyloidogenic pathway starts with β-secretase cleaving APP, releasing sAPPβ into the extracellular space. The remaining C-terminal fragment contains 99 amino acids (CTF99) and is further cleaved by γ-secretase to release Aβ into the extracellular space and AICD into the cytoplasm (10–14). The Aβ peptides with 40 or 42 residues in length (Aβ40 and Aβ42, respectively) are found in the most pathogenic plaques in AD. Aβ is primarily cleared through the vasculature via multiple pathways in the brain, including transport through the blood-brain barrier (BBB) to peripheral blood circulation, local degradation by cerebral vascular smooth muscle cells and pericytes, and diffuse drainage through perivascular space and the meningeal lymphatic system (15–22). In addition, Aβ is degraded locally in parenchyma by microglia, neurons, and extracellular proteolytic enzymes, including neprilysin, insulin-degrading enzyme, and matrix metalloproteinases 2 and 9 (23, 24). The brain coexists with different forms of Aβ, including nontoxic monomers, toxic and soluble oligomers, and aggregated amyloid fibrils (25). Nontoxic monomers and toxic soluble oligomers are present in a dynamic equilibrium, whereas aggregated fibrils are irreversible once formed (25). Excess Aβ production or disruption of Aβ clearance leads to the accumulation of toxic soluble Aβ oligomers and insoluble, aggregated Aβ fibrils, ultimately forming amyloid plaques in AD. The molecular mechanisms and root causes responsible for Aβ accumulation and deposition in AD are not entirely understood.

Figure 1.

Amyloid precursor protein (APP) cleavage by α-, β-, and γ-secretase. APP undergoes sequential cleavage following the nonamyloidogenic pathway (1) or the pro-amyloidogenic pathway (2). The nonamyloidogenic pathway is mediated by α- and γ-secretase, which produce the fragments soluble (s)APPα, P3, and amyloid precursor protein intracellular domain (AICD). The pro-amyloidogenic pathway is processed by β- and γ-secretase, which yields the fragments sAPPβ, amyloid β (Aβ), and AICD. The Aβ monomers aggregate to form oligomers and fibrils. Aβ, sAPPα and sAPPβ bind heparan sulfate (HS). The γ-secretase (beta-site APP cleaving enzyme 1, BACE1) binds HS and heparin, and this binding depends on 6S and 2S modifications and the level of N-acetylglucosamine (GlcNAc) residues. CTF, C-terminal fragment.

Heparan sulfate (HS) is a linear polysaccharide that belongs to the glycosaminoglycan (GAG) family (Fig. 2) (26, 27). HS is synthesized in the Golgi apparatus through a multiple-step process involving multiple enzymes (26, 27). The synthesis of HS starts with Exostosin-like 3 (EXTL3), which adds an N-acetylglucosamine residue (GlcNAc) to the glucuronic acid (GlcA) residue of a tetrasaccharide linker. This linker consists of GlcA-galactose (Gal)-Gal-xylose (Xyl) and attaches to protein cores to form heparan sulfate proteoglycans (HSPG). Subsequently, the heterodimer Exostosin-1/2 (EXT1/2) co-polymerases elongate the HS chain by adding GlcA and GlcNAc alternatively. The growing HS chain undergoes various modifications, including the initial N-sulfation (NS) of GlcNAc by N-deacetylase/N-sulfotransferase-1–4 (NDST1-4), and the following conversion of GlcA to iduronic acid (IdoA) by GlcA epimerase (GLCE), 2-O-sulfation (2S) of IdoA and, less frequently, GlcA by HS 2-O-sulfotransferase (HS2ST) and 6-O (6S) and 3-O (3S) sulfation of N-sulfated GlcNAc (GlcNS) by HS 6-O-sulfotransferase1-3 (HS6ST1-3) and HS 3-O-sulfotransferase1-6 (HS3ST1-6), respectively. After its biosynthesis, HS can undergo remodeling on the cell surface and in the extracellular matrix (ECM) by the 6-O-endosulfatases (SULF1-2), which remove 6S from GlcNS and causes changes in the structure of HS. HS can also be cleaved by heparanase (HPSE, not shown in Fig. 2), which breaks down the HS chain into smaller fragments (28–32). The abundance and structure of HS expression are determined by the enzymes responsible for its biosynthesis, remodeling, and degradation.

Figure 2.

Heparan sulfate proteoglycan (HSPG) and heparan sulfate (HS) biosynthesis and remodeling. HS is covalently attached to core proteins to form HSPG via the Xyl-Gal-Gal-GlcA tetrasaccharide linkage. HS biosynthesis includes chain initiation and elongation catalyzed by exostosin-like glycosyltransferase 3 (EXTL3) and exostosin glycosyltransferase 1/2 (EXT1/2), respectively, and concurrent chain modification catalyzed by N-deacetylase/N-sulfotransferase 1–4 (NDST1-4), glucuronic acid epimerase (GLCE), heparan sulfate 2-O-sulfotransferase (HS2ST), heparan sulfate 6-O-sulfotransferase 1–3 (HS6ST1-3), and heparan sulfate 3-O-sulfotransferase 1–6 (HS3ST1-6). Expressed HS can be remodeled through removal of 6S by 6-O-endosulfatase 1–2 (SULF1-2) and cleavage by heparanase (HPSE) (not shown).

The tetrasaccharide linker of HS covalently binds to selected serine residues of protein cores via O-glycosylation to form HSPGs that are expressed on the cell surfaces, in the ECM, and in intracellular secretory vesicles (26, 27). The HS chain of HSPGs interacts with numerous protein ligands, including growth factors, growth factor receptors, morphogens, cytokines, matrix proteins, and others. Through these interactions, HS modulates various physiological and pathological processes, such as organ development, leukocyte trafficking, tumorigenesis, and lipid metabolism (28, 33–40). HS binds protein ligands through electrostatic interactions imposed primarily by negatively charged sulfate residues and unique binding sites composed of varying sulfation and epimerization modifications in specific arrangements (29, 30, 32, 39, 41–44). In patients with AD and mouse models, HS co-deposits with Aβ in plaques in the cerebral parenchyma and vasculature (45–53). Biochemical and in vitro cell studies demonstrate that HS binds to Aβ, accelerating Aβ aggregation (54–59) and mediating Aβ cellular internalization and cytotoxicity (60–64). In AD mouse models, HS reduction through overexpression of HPSE, mainly in neurons and astrocytes, or neuronal HS ablation attenuates neuroinflammation and lowers Aβ burden in the brain, demonstrating that HS functionally promotes amyloidopathy in AD (24, 65, 66). Furthermore, the accumulation of HS in neurons at birth in a mouse model for a severe lysosomal storage disease called mucopolysaccharidosis type IIIA results in fibril formation and aggregation of Aβ as well as other amyloidogenic proteins, including α-synuclein, prion protein, and Tau, suggesting that HS may be a causative factor in Aβ fibrillization in AD (67). Therefore, inhibiting the Aβ-HS interaction may be a promising therapeutic strategy for AD treatment. These key findings have been discussed in detail in reviews by Snow et al. (52, 68), Ariga et al. (25), Zhang et al. (69), and Jin et al. (70). This review discusses recent studies reporting aberrant HS expression in patients with AD, the structural characteristics of HS-Aβ interaction, and the molecules with which HS interacts to modulate Aβ metabolism. Our perspective on the potential implications of aberrant HS expression in Aβ metabolism and AD pathogenesis is also provided.

ABERRANT HS EXPRESSION IN AD

Several studies documented clear aberrant expression of HS in the AD brain. Shimizu et al. (45) analyzed 25 patients with AD and found a 9.3- and 6.6-fold increase of HS in the hippocampus and the superior frontal gyrus, respectively. Compared with nondemented elderly subjects, HS in the AD brain is more densely concentrated in the thickened basement membrane adjacent to the endothelial cells of capillary vessels but not inside the endothelial cells and in the core of amyloid plaques (45, 59). These observations suggest that the marked HS increase in AD brains is mainly from the capillary basement membrane and senile plaques (45). Huynh et al. (46) and Wang et al. (71) confirmed the elevation of HS in hippocampus and frontal cortex of patients with AD, respectively. Despite an earlier study observing no alteration in HS structure in AD (72), Huynh et al. observed that GAGs from the AD hippocampus, including HS and chondroitin sulfate (CS), had altered growth factor binding capacities. AD GAGs showed decreased capacities to bind fibroblast growth factor-1 and -2 and vascular endothelial growth factor 165, and increased capabilities to bind heparin-binding epidermal growth factor-like growth factor, pleiotrophin, and tau, indicating a structural alteration in AD HS (46). Very recently, Wang et al. (71) analyzed the composition of AD HS and detected a significant increase in multiple sulfated disaccharides (ΔUA2S-GlcNS, ΔUA2S-GlcNAc, ΔUA-GlcNAc6S, and ΔUA2S-GlcNAc6S) and a tetrasaccharide with rare 3S (ΔUA-GlcNAc6S-GlcA-GlcNS3S6S) in human AD frontal cortex. These studies establish aberrant HS expression in the AD brain, affecting both HS expression levels and structure.

The composition, sulfation pattern, remodeling, and length of HS in tissue are determined by the expression profiles of HS biosynthetic and remodeling enzymes (Fig. 2). One study analyzed the postmortem hippocampus of 8 control and 8 patients with AD and found that multiple HS-regulated genes, including NDST2, GLCE, HS2ST, HS6ST1, HS6ST2, HS3ST2, HS3ST3A1, HS3ST3B1, and HPSE, were upregulated in AD (73). Another study examined 5 AD and 5 control human brains and observed similar results, reporting upregulated expression of NDST2, HS3ST2, HS3ST4, GLCE, and HPSE in the AD brains (46). Immunostaining of SULF1 and SULF2 in the brains of 20 patients with AD and 20 age-and gender-matched cognitively normal controls revealed that SULF2 expression in AD cases was significantly decreased in the parahippocampal gyrus and the frontal lobe gray matter (74). There were no differences in SULF1 expression in the parahippocampus or frontal lobe of AD cases and controls. In addition, there were no differences in SULF1 or SULF2 expression in the white matter of AD cases compared with cognitively normal controls (74). A recent study examined the transcriptomic expression of enzymes responsible for HS biosynthesis and modification in different brain regions in 18 patients with AD grouped at mild, moderate, or severe stages (6 per stage) and compared with six healthy individuals (75). The examined brain regions include the amygdala, anterior hippocampus, posterior hippocampus, claustrum, calcarine fissure, globus pallidus, and cerebellum. Several genes were downregulated across the three AD stages and in multiple brain regions, including EXT1, NDST2, HS2ST, HS6ST3, and SULF2. The number of gene alterations was higher in moderate AD compared with mild AD (Fig. 3A). In severe AD, there were fewer alterations of genes related to the early stages of HS biosynthesis. However, several genes involved in the late steps of HS biosynthesis, namely HS3STs, HS6STs, and SULF1, were upregulated (Fig. 3A). The alterations in HS gene expression correlated with progressive brain atrophy, with the cerebellum having the highest number of affected genes (Fig. 3B) (75). These results indicate that the alterations in HS biosynthetic and remodeling gene expression depend on the brain region and stage of AD pathology (Fig. 3, A and B). Interestingly, the altered transcriptional expression in the hippocampus reported by Perez-Lopez et al. differed significantly from what was observed earlier by Huynh et al. (Table 1). Despite this, these studies have consistently observed upregulated expression of HS3ST genes (Fig. 3 and Table 1) (46, 73, 75), which is in agreement with the elevated HS 3-O sulfation in AD reported by Wang et al. (46, 71, 75). Notably, a recent study utilizing multilayer analysis of patients with AD revealed that the proteopathic nature of AD is not reflected at an RNA level (77), which may explain discrepancies in studies using immunostaining versus transcript quantification. When grouped into 14 modules, the genes related to the cell-ECM interaction module were preserved at an RNA level. In contrast, the module containing matrisome-related genes showed an elevation only at the protein level in patients with AD. This change was observed more prominently in the temporal cortex than in the frontal cortex. As many proteins in the matrisome module bind HS or are HSPGs (77), suggesting that HS might critically regulate the functions of matrisome and its dysregulation could contribute to AD pathogenesis.

Figure 3.

Dysregulation of genes responsible for heparan sulfate (HS) synthesis and remodeling in human Alzheimer’s disease (AD) brains. Summarized from the study reported by Perez-Lopez et al. (75). The expression of HS biosynthesis and 6-O-endosulfatase (SULF) genes was altered in human AD brains, with the degree of dysregulation varying depending on the gene family, the disease stage, and the brain region. A: all HS-related gene families had greater dysregulation in moderate AD compared to mild AD. Genes responsible for early- and middle-stage-HS biosynthesis, including Exostosins (EXTs), N-deacetylase/N-sulfotransferases (NDSTs), glucuronic acid epimerase (GLCE), and heparan sulfate 2-O-sulfotransferase (HS2ST) families, were downregulated in subjects with AD, while genes involved in later-stage HS biosynthesis, such as heparan sulfate 6-O-sulfotransferases (HS6STs) and heparan sulfate 3-O-Sulfotransferases (HS3STs), and the remodeling gene SULFs, tend to be upregulated in AD. B: the cerebellum (CB) exhibited the highest number of gene perturbations, and the numbers of gene perturbations and the brain regions fit a linear regression following a spatiotemporal pattern of the amygdala (AM), anterior hippocampus (AH), posterior hippocampus (PH), claustrum (CL), calcarine fissure (CF) and globus pallidus (PG).

Table 1.

Dysregulated expression of HS biosynthesis genes in AD hippocampus and temporal cortex reported by Huynh et al., Perez-Lopez et al., and Rosenthal et al.

	Sepulveda-Diaz et al.	Huynh et al.	Perez-Lopez et al.						Rosenthal et al.
Gene	Hippocampus	Hippocampus	Anterior Hippocampus Mild Moderate Severe			Posterior Hippocampus Mild Moderate Severe			Temporal Cortex and Cerebellum
EXTL2							↓
EXTL3				↓			↓	↓
EXT1						↓		↓
EXT2				↓			↓
NDST1				↓			↓
NDST2	↑	↑		↓			↓
NDST3			—	—	—	—	—	—
NDST4							↑
GLCE	↑	↑			↑		↓
HS2ST	↑			↓			↓
HS6ST1	↑							↑
HS6ST2	↑		↑	↓	↓		↓
HS6ST3				↓		↓	↓	↓
HS3ST1				↓			↓		↓
HS3ST2	↑	↑					↓		↓
HS3ST3A1	↑								↑
HS3ST1B1	↑		↑		↑				↑
HS3ST4	↑	↑	↓						↓
HS3ST5			↑	↑	↑	↑	↑		↓
HS3ST6				↓		↓

Downward arrows indicate downregulation of gene expression; upward arrows indicate upregulation of gene expression; dashes indicate no detected transcripts. Sepulveda-Diaz et al. (73) analyzed the postmortem hippocampus from eight control and eight patients with AD (mean age = 76.8 ± 3.5 yr). Huynh et al. (46) analyzed the postmortem hippocampus from five control and five patients with AD (mean age = 72.3 ± 7.0 yr). Perez-Lopez et al. (75) analyzed the postmortem anterior and posterior hippocampus from six control and 18 patients with AD (6 mild, 6 moderate, and 6 severe AD) (mean age = 84 ± 11.0 yr). The gene transcript data from other brain regions are not included here (75). Rosenthal et al. (76) mapped the gene network landscape of AD by integrating genomics and transcriptomics data and observed altered HS gene expression in AD patients’ temporal cortex and cerebellum from a GWAS of 455,258 individuals with 71,880 proxy cases and 383,378 controls. EXT, exostosin glycosyltransferase; EXTL, exostosin like glycosyltransferase; GLCE, glucuronic acid epimerase; HS, heparan sulfate; HS2ST, heparan sulfate 2-O-sulfotransferase; HS3ST, heparan sulfate 3-O-sulfotransferase; HS6ST, heparan sulfate 6-O-sulfotransferase; NDST, N-deacetylase/N-sulfotransferase.

HEPARAN SULFATE-Aβ INTERACTION

Biochemical studies have shown that Aβ directly binds to HS and heparin, a highly sulfated HS analog. The major HS binding motif within Aβ has been identified as V12HHQKL17 (Fig. 4A) (49, 50, 68, 78–85). HS-Aβ binding depends on electrostatic interactions involving the imidazole side chain of the histidine residues of Aβ. This interaction is strengthened by increased sulfation and the size of HS and heparin (78, 86–89). Interestingly, HS binds to Aβ40 with higher affinity than Aβ42 (87, 90) and HS induces Aβ40, not Aβ42, to form maltese-cross spherical congophilic plaques identical to those seen in AD brain (68). Anti-HS phage display antibody staining has identified specific structures of HS that co-deposit with Aβ in the AD brain. The phage display antibodies EV3C3 and HS4C3 recognize fully N-sulfated motifs in the HS chain, whereas RB4EA12 and HS4E4 recognize HS regions with a lower degree of NS. The EV3C3 and HS4C3 antibodies stained strongly in both fibrillar and nonfibrillar Aβ plaques, whereas the RB4EA12 and HS4E4 antibodies stained fibrillar Aβ plaques only (Fig. 4B) (80). HS4E4 epitopes preferentially colocalized with Aβ40 in the cores of senile plaques and are absent from Aβ42-rich diffuse deposits in sporadic AD brains (Fig. 4B) (91). In agreement with the observations by Snow et al. (51), HSPGs were not found in diffuse deposits in the cerebellum but in the cores of plaque deposits in the hippocampus and cortex of AD brains. These results suggest that differently structured HS co-deposits with Aβ plaques. The HS structures that co-deposit with nonfibrillar Aβ plaques are more restricted, consisting of highly N-sulfated structures (80). In addition, the anti-HS phage display antibody RB4CD12 recognizes highly sulfated HS domains (92) and strongly stained diffuse and neurotic Aβ plaques in human AD and transgenic AD mouse models (93). The RB4CD12 staining diminished when the tissues were pretreated with extracellular sulfatases SULF1 and SULF2, suggesting that the Aβ-binding HS moieties contain 6-O-sulfation (Fig. 4A) (93). These immunostaining studies indicate that the Aβ-binding HS moieties within Aβ plaques. have common NS and 6S modifications in vivo.

Figure 4.

Interaction between heparan sulfate (HS) and amyloid beta (Aβ). A: the modifications highlighted in red within HS are necessary for high-affinity binding to Aβ, and the histidine residues within Aβ participate in the HS-Aβ interaction. B: the molecular interactions between HS and Aβ depend on both HS modifications and Aβ structure, as demonstrated by co-staining of Aβ with anti-HS phage display antibodies and analysis of affinity-purified Aβ-binding HS hexasaccharide from the human cerebral cortex. Additional data also indicate that heparin binding to Aβ aggregates depends primarily on sulfation level rather than a specific type of sulfate moiety (not shown here).

Biochemical and cell-based studies have gained additional insights into HS structural features critical for Aβ binding and reported conflicting findings. For instance, maltese-cross amyloid plaque core formation can be induced by N-desulfated, N-acetylated heparin, but not the completely desulfated heparin with or without N-acetylation, suggesting O-sulfation, not N-sulfation, is required for HS GAG to bind Aβ40 (68, 94). Analysis of HS oligosaccharides from the human cerebral cortex has determined that Aβ fibrils bind to N-sulfated hexasaccharide domains that contain critical IdoA2S residues, with GlcNS6S being an additional requirement for Aβ monomer binding (Fig. 4A) (54). A study with synthesized HS polysaccharides determined that NS and 6S are essential for Aβ binding beyond electrostatic attraction (Fig. 4A) (87). Mapping and molecular modeling of heparin oligosaccharides using two-dimensional (2-D) NMR have determined the tetrasaccharide HexA-GlcNS-IdoA2S-GlcNS6S as the Aβ-binding motif of heparin (Fig. 4A) (78). IdoA, NS, and 6S are essential modifications for high-affinity Aβ binding, while 3-O sulfation is not directly involved, although it may cause steric hindrance (87). Furthermore, a study utilizing a heparinoid-immobilized microarray chip assay determined that the binding of heparin to Aβ aggregates was not significantly altered after removing NS, 6S, or 2S (62). In the same study, heparin significantly inhibited the cellular uptake of Aβ aggregates. The inhibition was abolished when NS was removed but only moderately reduced upon removal of 6S or 2S, showing specifically NS and the overall level of sulfation are critical for HS to bind to aggregated Aβ (62). Taken together, current data indicate that the sulfation level and modification pattern of HS and the isoform and fibrillation state of Aβ determine the interactions between HS and Aβ. However, further clarification, particularly with respect to the related HS structures, is needed with advanced techniques.

THE MOLECULES THAT HS INTERACTS WITH TO MODULATE Aβ METABOLISM

HS plays a complex role in Aβ metabolism and related pathogenesis of AD. Besides direct binding to accelerate Aβ fibril and aggregate formation and internalization (54–56), HS also regulates Aβ generation and clearance via various molecular mechanisms. In the following sections, we discuss some of the most reported proteins that interact with HS and critically impact Aβ metabolism.

Secreted APPs

APP is an integral transmembrane protein and acts as a cell surface receptor to regulate various functions, including neurite growth, neuronal adhesion, axonogenesis, cell motility, and gene expression. APP undergoes amyloidogenic and nonamyloidogenic processing, which results in the generation of extracellular sAPP fragments (Fig. 1). Interestingly, sAPP is necessary and sufficient to rescue prominent abnormalities of APP-knockout mice, such as the impairment in spatial learning and long-term potentiation, indicating that the primary biological function of APP is mediated by its ectodomain (95–100). sAPP has been found to bind heparin (96) and cell membrane-anchored HSPGs, including glypicans and syndecans (101). It is unknown whether cell-surface HSPGs regulate APP functions, and if any alteration of this interaction contributes to AD pathogenesis. sAPP catalyzes the nitric oxide-dependent deaminative release of HS chains from the HSPG Glypican-1 (102). The Sortilin-related receptor 1 (SORT1) involves endosomal trafficking of APP and other targets in neurons and is genetically associated with AD (103). The extracellular HSPG perlecan is partly responsible for the ubiquitination and degradation of SORT1, suggesting that perlecan may inhibit SORT1 function to disturb APP processing, leading to APP accumulation in AD (103). Perlecan also binds to sAPP (104). Since perlecan accumulates before amyloid deposition, its interaction with sAPP and SORT1 may affect Aβ production (105, 106). In fact, perlecan was isolated and examined as a model for Aβ co-deposition and fibril formation before developing Aβ aggregation-driven mouse models of AD (107). Moreover, the interaction between sAPP and ECM is critical for signal transduction in neurons (96, 98). ECM-bound sAPP, but not free sAPP, stimulates neurite outgrowth in cultured chick sympathetic and mouse hippocampal neurons through HS binding (96, 98). One possible molecular mechanism underlying the accentuated neurite outgrowth may be attributed to binding to the ECM, which shields sAPP from degradation by proteases (99). Interestingly, not all forms of HSPG accentuate neurite outgrowth. HSPG purified from the postnatal day 3 mouse brain stimulated neurite outgrowth, whereas HSPGs from the embryonic day 10 mouse brain (a developmental stage before the major period of neurite outgrowth) or HSPGs isolated from the liver did not induce neurite outgrowth. These observations suggest that developmentally regulated HSPGs, most likely the specific HS structures, are required for sAPP to stimulate neurite outgrowth (96). Dysregulated HS structures in AD may disturb the vital neurite outgrowth-promoting function of sAPP, leading to the loss of synapses and neurites in AD. Furthermore, heparin binding to sAPP stimulates APP synthesis and secretion through the amyloidogenic pathway (108). Therefore, elevated HS levels in AD might promote APP expression to enhance Aβ production, contributing to AD pathogenesis. Other ECM proteins, specifically those associated with the basement membrane, such as collagen IV, laminin, fibronectin, and entactin, can also bind to sAPP (96, 104). Under physiological conditions, these interactions may function to form and maintain a healthy basement membrane. In AD, HS, perlecan, laminin, collagen IV, and fibronectin co-deposit with amyloid plaques and APP (109). The aberrant expression of HS in AD may disrupt the interaction between ECM HSPGs, sAPP, and other ECM proteins, destabilizing leading to basement membrane and, consequently, leading to basement membrane rupture and loss of neurons in AD (104).

β-Site APP Cleaving Enzyme 1

β-Site APP cleaving enzyme 1 (BACE1) is the primary β-secretase in the brain, responsible for cleaving and generating Aβ (14) (Fig. 1). Endogenous HS immunoprecipitated with BACE1 from cell lysates and colocalizes with BACE1 in the Golgi complex and on the cell surface, two of its putative sites of the enzyme action. HS and heparin bind BACE1 directly and inhibit BACE1 enzyme activity in cell culture, acting as an endogenous inhibitor of BACE1 to suppress Aβ generation (110–113). HS inhibition of BACE1 depends on HS size and specific structural characteristics; 6S is essential, 2S is moderately required, whereas additional GlcNAc residues enhance BACE1 activity. Oversulfation of HS chains reduces HS inhibition of BACE1 (110–113). These results indicate that the interaction of heparin and HS with BACE1 requires specific HS structures, not just sulfation levels. One study suggested that α-secretase functions by releasing sAPPα and HS to inhibit BACE1 and Aβ production (102). The downregulated expression of EXTL2, EXT1, EXT2, and HS6ST1-3 in AD (75) may reduce cell surface HS expression and the level of 6S, leading to attenuated HS suppression of BACE1 and a subsequent increase in Aβ production. In addition, reduced NDST expression in AD may increase GlcNAc residues, enhancing BACE1 activity to increase Aβ production. Therefore, aberrant HS expression may significantly increase BACE1 activity, representing an important mechanism to switch from a physiological nonamyloidogenic process to a pathological amyloidogenic process underlying enhanced Aβ production in AD. Furthermore, BACE1 influences HS expression as well. Pharmacological inhibition of BACE1 reduced HS6ST3 secretion and increased the cellular accumulation of 6-O sulfated HS (114), revealing a positive regulatory feedback mechanism. Therefore, BACE1 may promote HS6ST3 secretion and reduce cellular HS 6S to enhance its activity, thereby increasing Aβ production.

Apolipoprotein E

Apolipoprotein E (ApoE) is a secreted protein and is a critical player in lipid transport within the brain. It is abundantly expressed in astrocytes, microglia, vascular mural cells, and choroid plexus cells, and to a lesser extent in stressed neurons (115). There are three allelic variants of ApoE: ε2, ε3, and ε4. The ApoE ε4 variant is associated with an increased risk of AD, whereas the ApoE ε2 variant is associated with a lower AD risk compared with the more common ApoE ε3 variant (116–120). A growing body of evidence suggests that ApoE ε4 results in a higher AD risk by inhibiting Aβ clearance, promoting Aβ aggregation, and suppressing Aβ cellular uptake and metabolism, leading to extracellular Aβ accumulation (9, 121–125). On the other hand, ApoE ε2 decreases Aβ aggregation and promotes cellular uptake (126). In addition, ApoE ε4 is associated with the attenuated release of HS from Glypican-1 in endosomes, which leads to decreased suppression of BACE1 and increased Aβ production (102). ApoE stimulates neuronal APP transcription and Aβ production in a variant-dependent manner, with ε4 > ε3 > ε2 (127). Earlier studies suggested that the three ApoE variants bind HS with similar affinities (128–130). However, recent studies have determined that ApoE ε4 has a threefold higher affinity for HS than ε2 and ε3 (131, 132). Interestingly, the ApoE ε3 Christchurch R136S mutation impairs binding to HS and significantly delays the onset of cognitive impairment of a PSEN1 mutation carrier (133). Meanwhile, multiple approaches, including ¹³C nuclear magnetic resonance, molecular modeling, and testing with N-desulfated heparin, have determined that NS and 6S of glucosamine residues are required for high-affinity HS binding to ApoE (130). ApoE modulates HS-mediated Aβ cell surface binding and internalization, possibly in a lipidation-dependent manner. O’Callaghan et al. (134) observed that delipidated human ApoE increases cell association of Aβ40 through HS. On the contrary, Fu et al. (132) reported that lipidated ApoE, secreted from immortalized astrocytes, significantly suppresses Aβ binding to the cell surface and cellular Aβ uptake through cell surface HS, and this effect is ApoE variant-independent. Further studies with delipidated and lipidated ApoE variants in the same experimental setting may clarify this lipidation and ApoE variant dependence. Meanwhile, it would be interesting to test if the aberrant HS expression in AD alters HS-ApoE interaction contributing to dysregulated Aβ metabolism in AD, and whether the increased AD risk of ApoE ε4 variant is mediated by its higher binding affinity for HS.

Low-Density Lipoprotein Receptor-Associated Protein-1

Lipoprotein receptor-associated protein-1 (LRP1) is an endocytic receptor that has a repertoire of more than 40-ligand repertoire, including ApoE and Aβ. In the brain, LRP1 plays a major role in maintaining Aβ homeostasis and is expressed abundantly in various cell types, including neurons, vascular cells, and glial cells (124). LRP1-mediated endocytosis regulates cellular Aβ uptake by binding to Aβ directly or indirectly through its co-receptors or ligands. LRP1 and HSPG form an immunoprecipitated complex on the cell surface that mediates lipid metabolism (135). The HSPG and LRP1 complex partially mediates ApoE uptake (136). These findings suggest that HS and LRP1 have a cooperative function on the cell surface. Functional studies have shown that heparin and HS deficiency impair LRP1-mediated Aβ uptake in mouse neuronal cells and mouse embryonic fibroblasts, indicating that LRP1 requires cell surface HS to internalize Aβ (137). As Aβ directly binds to HS and LRP1, the current literature suggests that HS may act as a co-receptor to facilitate LRP1-mediated Aβ uptake. Similarly, the downregulation of EXTL2, EXT1, EXT2, NDST1-2, and HS6ST1-3 in AD may reduce cell surface HS expression, which could attenuate LRP1-mediated Aβ uptake. This may represent one important molecular mechanisms underlying increased Aβ accumulation in AD.

PERTURBATIONS IN HSPGs AND OTHER GLYCOSAMINOGLYCANS IN AD

In addition to alterations in HS chains in AD, changes in HSPG expression and function, including their core proteins, have also been reported (52). HSPGs include the ECM-associated perlecan, agrin, and collagen XVIII, the cell-membrane-anchored syndecans and glypicans, and serglycin in the secretory vesicle. First demonstrated by cationic dyes, including Ruthenium red and Cuprolinic blue (48), several human tissue staining studies have observed HSPGs in amyloid plaques in AD, which have been outlined in Table 2. In addition, with increasing neuroinflammation, immune cells also synthesize more HSPGs in AD (147).

Table 2.

Summary of HSPGs detected by immunolabeling in human patients with AD

	Immunopositively Labeled Areas					Antibody	Source of Tissue Sample	Subjects Analyzed Mixed Genders	References
	Normal Vessels	CAA Vessel	NP/SP	Primitive Plaque	Other
Syndecans	✓	✓	✓			10H4(SDC2 ectodomain)	F, P, T, O	7 AD/CAA, 3 controls	(138)
			✓			2E9 (SDC1,3 cytoplasmic domain)
			✓			1C7 (SDC3 ectodomain)
			✓		Fibrillar plaques only	10H4 (SDC2 ectodomain)	CB	8 AD, 3 controls	(139)
			✓			2E9 (SDC1,3 cytoplasmic domain)
			✓			1C7 (SDC3 ectodomain)
			✓		Classic plaques, not primitive plaques	10H4 (SDC2)	N/A	4 AD, 4 controls	(140)
			✓		Some neurons and NFTs	Anti-SDC1	TE, EH, PH, CL, CS, PL, CB	18 AD, 6 controls	(141)
		✓	✓			Anti-SDC4
Glypicans		✓	✓	✓		S1 (GPC1)	F, P, T, O	7 AD/CAA, 3 controls	(138)
		✓	✓	✓		S1 (GPC1)	CB, F	9 AD, 3 controls	(139)
		✓	✓	✓	fibrillar plaque only	S1 (GPC1)	N/A	4 AD, 4 controls	(140)
		✓	✓	✓	neurons and neuropils with NFTs	Anti-GPC4 Anti-GPC6	TE, EH, PH, CL, CS, PL, CB	18 AD, 6 controls	(141)
Agrin	✓	✓	✓	✓	highest of all HSPGs	JM72	F, T, H	11 AD, 3 controls	(142)
	✓	✓	✓	✓	Higher in AD, concentrated in plaques and neurons	Agrin antiserum	PFC, VC, T, H, A	16 AD, 11 control cases	(143)
	✓	✓	✓	✓	Higher in AD, in plaques and neurons	Agrin antiserum	PFC, H	22 AD, 12 controls	(144)
	✓	✓	✓	✓		Agrin antiserum	N/A	N/A	(145)
	✓	✓	✓	✓	The highest of all HSPGs	JMJ2 (Agrin peptide)	F, P, T, O	7 AD/CAA, 3 controls	(138)
						BI31 (Agrin core protein)
	✓	✓	✓	✓	The highest of all HSPGs	JMJ2 (Agrin peptide)	CB, F	9 AD, 3 controls	(139)
						BI31 (Agrin core protein)
Collagen XVIII	✓	✓	✓			QH48.18	H, F, O	11 AD/CAA, 5 AD only, 3 control cases	(139)
						QH1415.7
Perlecan	✓	✓			Not associated with plaques	1948P, 95J10,HK-102	F, T, H	11 AD, 3 control cases	(142)
	✓	✓			Not associated with plaques	1948P	F, P, T, O	7 AD/CAA, 3 controls	(138)
						95J10
	✓	✓			increased in areas with plaques in Braak >2	1948 P	CB, F	9 AD, 3 controls	(139)
						95J10
	✓	✓			Not associated with plaques	mAb (ZYMED)	N/A	4 AD, 4 controls	(140)
	✓	✓			Increased staining intensity with Braak stage	Abcam (ab2501)	SF, IT	8 AD, 10 mild AD 12 controls	(146)
Serglycin	✓	✓	✓	✓		Anti-SRGN	TE, EH, PH, CL, CS, PL, CB	18 AD, 6 controls	(141)

The table shows different immunopositively labeled areas, antibodies used, source of tissue samples, and subjects analyzed in various studies related to the investigation of heparan sulfate proteoglycans (HSPGs) in Alzheimer’s disease (AD) and cerebral amyloid angiopathy (CAA). The presence and distribution of these proteoglycans were investigated in different areas of the brain, including normal blood vessels, primitive plaques, fibrillar plaques, classic plaques, neurons, and neuropils with neurofibrillary tangles (NFTs). The number of subjects analyzed varied among the studies. GPC, glypican; NP/SP, neuritic plaque/senile plaque; SDC, syndecan. The areas of the brain are indicated by letters as follows: H, hippocampus; AH, anterior hippocampus; PH, posterior hippocampus; A, amygdala; CB, cerebellum; C, claustrum; CS, calcarine sulcus; F, frontal cortex; G, globus palludus; P, parietal cortex; PL, pallidum; T, temporal cortex; TE, trans entorhinal region; ST, superior temporal cortex; O, occipital cortex; PFC, prefrontal cortex.

Functional studies have indicated that perlecan is involved in AD pathology. Perlecan, along with Aβ40, is the only HSPG reported to induce congophilic maltese-cross formation of Aβ similar to that in clinical AD (94). Perlecan hinders Aβ from degradation for 2 wk in vitro (57) and causes fibrillar Aβ persistence in the brain (148). Perlecan is the primary HSPG in Matrigel (148), which was used in the “AD in a dish” model developed by Choi et al. (53). In this model, Matrigel was used to coat and grow neuronal cells, and plaque-like deposits were formed in a dish followed by tangle formation a few weeks later, suggesting that perlecan is required for the spontaneous development of Aβ plaques and Tau tangles in a neuronal cell context (53). In addition, HS induces the microtube-binding protein Tau to form paired helical filaments in neurons, another hallmark of AD pathology (4, 5, 43, 149, 150). These studies support the crucial role of HS and HSPGs in AD disease development as hypothesized by Snow et al. (68).

Alterations in HSPG expression in patients with AD have also been investigated, as outlined in Table 3. Notably, Lorente-Gea et al. (141) reported that HSPG expression in AD differs significantly with respect to the disease severity and spatiotemporal distribution. Recently, Johnson et al. (77) reported that the matrisome module was significantly elevated in patients with AD, and these proteomic changes are not reflected at the RNA level. The plaque and tangle-associated proteins with the highest eigenvalue correlations with the matrisome module were ApoE, APP, Syndecan 4, and Glypican 5 (77). In agreement with this finding, Lorente-Gea et al. (141) reported elevated Syndecan 5 and Serglycin expression throughout the AD brains. Further functional studies of the specific HSPGs are needed to better understand how they contribute to AD pathogenesis.

Table 3.

Summary of the expression profiles, impact on Aβ, and affected molecular pathways of HSPGs in patients with AD

HSPG	Expression in AD	Impact on Aβ	Affected Molecular Pathway
Perlecan (HSPG2)	Elevated expression in patients with AD (151)	Perlecan is the only HSPG reported to induce congophilic maltese-cross formation of Aβ in vitro (94) Slows Aβ degradation in vitro (57)	Ischemic conditions increase perlecan expression (105, 152) Perlecan accumulation precedes atherosclerotic lesions, suggesting that perlecan aggregation may contribute to endothelial cell disruption (106, 153) Domain V can inhibit a α2- and α5-β1-integrin mediated Aβ toxicity in vitro (152, 154–156) Senses flow through cation binding at the endothelial cell and blood interface (157)
Agrin (AGRN)	Abundant in AD (138, 142) and in insoluble Aβ plaques (139, 143) Increased soluble agrin with higher Braak stage diagnoses (144) and CSF levels can be used as a biomarker (161)	Increased Aβ results in accentuated agrin expression and sulfation in vivo (158, 159) AGRN binds to Aβ, accelerates its fibrillization and stability of formed Aβ fibrils (143, 145, 158)	Increased brain endothelial cell, but not neuronal, agrin expression is associated with a reduction in Aβ deposition and mediation of aquaporin-4 levels, demonstrating its role in the regulation of BBB permeability (160) Regulates synaptic assembly (162) and hippocampal neurogenesis in vitro (163)
Glypicans (GPC1–6)	All GPC1–6 are downregulated in the claustrum and some in the cerebellum, but upregulated in limbic areas of patients with AD (141)	Treatment with Aβ results in amplified GPC1 transcription and sulfation in vitro (159) and in proximal astrocytes (164) GPC1 promotes Aβ multimerization and internalization (166); GPC4 also promotes Aβ internalization, via LRP1, as well as toxicity (167)	Alterations in conserved loci that led to GPC4 reduction are associated with cognitive and social disorders (165) GPC1 senses flow for endothelial cell nitric oxide production and consecutive constriction in vitro (168)
Syndecans (SDC1-4)	SDC 3 and 4 are increased in patients with AD SDC 4, 3, and 2 to a lesser extent, were increased with AD stage, whereas SDC1 tended to decrease in moderate AD (141) Patients with AD had an increase in syndecan-binding protein, SDCBP, expression (171)	With high neuronal expression, SDC3 is critical for Aβ1–42 endocytosis. Internalized Aβ colocalizes with syndecans and flotillin, suggesting lipid rafts for syndecans-mediated internalization (169)	SDC3 is involved in hippocampal long-term potentiation in rats (170) Senses flow through cation binding at the endothelial cell and blood interface (157)
Collagen XVIII (COL18A1)	Upregulated expression in the entorhinal cortex and hippocampus of severe AD brains (141)		Extracellular matrix dysfunction is common in patients with AD and linked to COL18A1 in expression module (172)
Serglycin (SRGN)	Overexpressed in most brain areas in patients with AD (141)		Role as an interacellular and a secretory vesicle HSPG must be considered for its role
Testicans (SPOCK1-3)	SPOCK1 is elevated in the frontal, temporal, and entorhinal cortices of patients with AD (173)	SPOCK-1 accumulates and co-aggregates with Aβ plaques (173)	SPOCK2 is suggested to drive neuroinflammation in conjunction with Aβ (174).
	SPOCK2 is elevated in patients with AD (151) SPOCK2 is elevated in Tg-5xFAD animals >6 mo old (174)	APP affects SPOCK1 trafficking and SPOCK1 accumulation in Golgi reduces Aβ (40 and 42) production (173)

The table lists six heparan sulfate proteoglycans (HSPGs), namely, Perlecan (HSPG2), Agrin (AGRN), Glypicans (GPC1–6), Syndecans (SDC1–4), Collagen XVIII (COL18A1), and Serglycin (SRGN) and their different effects on Aβ aggregation and involvement in various molecular pathways. Perlecan is reported to induce congophilic maltese-cross formation of amyloid beta (Aβ) and slows Aβ degradation. Agrin, abundant in Alzheimer’s disease (AD) and insoluble Aβ plaques, binds to Aβ, accelerates its fibrillization and stabilizes formed Aβ fibrils. Glypicans promote Aβ multimerization, internalization, and toxicity, whereas syndecans are involved in hippocampal long-term potentiation and endocytosis of Aβ. Collagen XVIII and Serglycin are overexpressed in most brain areas in patients with AD, with their exact role still being explored. Finally, testicans (SPOCK1-3) are elevated in patients with AD and suggested to drive neuroinflammation in conjunction with Aβ production. APP, amyloid precursor protein; BBB, blood-brain barrier; CSF, cerebrospinal fluid; LRP1, low density lipoprotein-related protein 1; SPOCK, testican.

In addition, other GAGs, including CS and dermatan sulfate, are associated with Aβ plaques and NFT in AD (79, 175). Perineuronal nets are CS-containing matrices that enmesh neural networks involved in memory and cognition. CS sulfation patterns in perineural nets are altered in AD, and these alterations occur before the development of AD clinicopathology (176). Changing the CS sulfation pattern may contribute to the initiation and progression of the underlying degenerative processes, suggesting the alteration of CS sulfation pattern as a critical player in the development of AD clinicopathology (176). A newly discovered type of CS/HS proteoglycan, named testicans 1–3 (SPOCK1-3), was identified in patients with AD in proteomic analyses (173). Abundant in the ECM, SPOCK1 is trafficked by APP, affecting Aβ production (173), and SPOCK2 is elevated in patients with AD (151) and an transgenic AD mouse model (174), and it may drive neuroinflammation in conjunction with Aβ in AD pathogenesis (174). Furthermore, cerebral keratan sulfate is selectively lost in AD (177).

CONCLUSIONS AND PERSPECTIVE

Ample evidence has established that HS is aberrantly expressed in AD, including changes in its expression level, structure, and the enzymes responsible for its biosynthesis, remodeling, and degradation. However, the HS level and structural alterations detected in AD brains do not correlate well with changes in the expression of genes responsible for HS synthesis and remodeling (46, 71, 73, 75). The downregulation of HS genes responsible for its biosynthetic initiation and elongation include EXTL3, EXT1, and EXT2 (Table 1) (75), indicating that active HS biosynthesis is reduced in cells in the AD brain. However, biochemical analyses detected elevated levels of HS in AD brain (45, 46, 71). These observations suggest that the HS elevation in the AD brain is due to its accumulation rather than active HS biosynthesis. HS immunostaining studies have found HS to be more densely concentrated in the thickened basement membrane adjacent to the endothelial cells of capillary vessels (where HS co-deposition with Aβ is commonly seen) and amyloid plaques in AD brains (45). Meanwhile, Aβ binding has been shown to protect HS from degradation by HPSE (81), which may facilitate HS accumulation in AD brains. Taken together, these transcriptional, biochemical, and immunostaining data suggest that the elevation in HS detected in AD brains may reflect accumulated HS that co-deposits with Aβ, rather than newly biosynthesized HS, such as the cell surface HS. Cell surface HS essentially mediates Aβ binding and cellular uptake in AD. The downregulation of HS expression on cell surface may impair Aβ clearance, representing a major pathway failure contributing to Aβ accumulation in AD.

The gene transcription data from several studies conducted to date have been inconsistent (46, 73, 75). The discrepancy is likely due to the heterogeneity of AD etiology, as well as variations in disease severity, genetic background, brain regions and cell types analyzed, and differences in postmortem tissue collection. To improve our understanding of AD, new studies should be conducted with larger cohorts of patients with AD grouped based on systematic pathological analysis and with sufficient controls. In addition, these reported studies are based on tissue analysis in bulk. Single-cell RNA sequencing has revealed that the expression pattern of HS-related genes is unique for brain cell types (178), supporting the idea that the structure of HS is cell type-specific (179). Further studies using high-resolution single-cell and spatial multiomics techniques could advance our understanding and provide a clearer picture of the correlation between HS-related gene expression and HS structural alterations seen in AD. Specifically, these studies could answer how these changes might be associated with AD initiation and progression (77). Furthermore, a study that specially targeted HS expression in neurons in a mutant APP transgenic mouse model determined the roles of neuronal HS expression in Aβ metabolism and neuroinflammation (24). Similar studies are needed to determine the roles of HS expressed in other brain cell types in Aβ metabolism and AD-related pathology. Systematic in vivo studies could advance our understanding of the spatiotemporal roles of aberrant HS expression and its role in the disrupted Aβ metabolism as seen in AD. Moreover, these findings might facilitate the development of highly efficient HS-targeting therapeutics for AD treatment.

HS plays crucial role in regulating Aβ metabolism through direct and indirect mechanisms. The sulfation level and modification pattern of HS, along with the isoform and fibrillation state of Aβ, determine the binding of HS to Aβ. Direct binding of HS accelerates Aβ aggregation, fibrillization, and deposition in the brain. The indirect effects of HS function are mediated through interactions with distinct HS-binding proteins, including sAPP, BACE1, LRP1, and ApoE. These interactions may suppress Aβ production, enhance Aβ uptake, and modulate Aβ clearance (180). Downregulated expression of EXTL3, EXT1, EXT2, NDST2, and HS6STs suggests that brain cells reduce HS expression in AD (75). This reduction might disrupt the HS-modulated sAPP, BACE1, LRP1, and ApoE pathways, leading to increased Aβ production and suppressed Aβ clearance, thus exacerbating Aβ deposition and amyloid plaque formation in AD. Furthermore, HS may also affect other yet-to-be-revealed molecular pathways associated with Aβ metabolism, such as the triggering receptor expressed on myeloid cell 2 (TREM2) pathway. TREM2 is an extracellular innate immune receptor expressed on myeloid lineage cells, such as dendritic cells and resident tissue macrophages, including osteoclasts and microglia. TREM2 modulates microglial function, Aβ phagocytosis, and suppresses Aβ-induced inflammatory response (181–184). TREM2 binds to and inhibits Aβ polymerization (183, 185, 186). Cell surface HS binds to TREM2 likely to cluster or orient TREM2 for its ligand binding (182, 183). Genome-wide association studies linked TREM2 variants to late-onset AD, particularly the R47H and R62H variants (187–189). Interestingly, the heterozygous R47H variant carries roughly the same risk as a copy of the ApoE ε4 allele (190). The R47H and R62H variants have reduced binding to HS on the cell surface and to ligands such as phospholipids and lipid particles. There is the possibility that decreased binding of the TREM2 variants to HS may disrupt TREM2’s ability to recognize and signal in response to Aβ. Aberrant expression of HS in AD may alter this pathway to impair Aβ clearance, thus exacerbating Aβ deposition in the brain. Addressing these emerging questions in future studies will advance the mechanistic understanding of HS in Aβ metabolism.

GRANTS

This work was made possible through funding from several sources, including NIH Grants 7R01HL093339, 5U01CA225784, 1RF1AG074289, and 1RF1AG069039, and the Swedish Research Council Grant 2021-01094.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

I.O.M. and L.W. prepared figures; I.O.M. and L.W. drafted manuscript; I.O.M., J-P.L., and L.W. edited and revised manuscript; I.O.M., J-P.L., and L.W. approved final version of manuscript.