1. IntroductionTauopathies are a heterogenous family of progressive neurodegenerative diseases featured with the deposition of abnormally folded species of the microtubule-associated protein tau (tau) mainly in neurons, glia, and extracellular space [1]. There are 26 tauopathies identified, including Alzheimer’s disease (AD), frontotemporal dementia with parkinsonism-17 (FTDP-17), corticobasal degeneration (CBD), progressive supranuclear palsy (PSP), argyrophilic grain disease (AGD), Pick’s disease (PiD), and other diseases where tau plays a major role. Based on the major tau isoforms contained in the tau deposits, tauopathies are classified into 3 repeat (R) tauopathies, 4R tauopathies, and 3R+4R tauopathies (with approximately an equal ratio of 3R tau and 4R tau) [1,2,3]. The diseases are also classified into primary and secondary tauopathies [1,3]. The primary tauopathies are a group of neurodegenerative diseases in which tau is believed to be the major contributing factor of the neurodegenerative process, such as PiD, PSP, CBD, and AGD. The secondary tauopathies are featured with tau aggregation as a response to other pathological proteins or events, like amyloid beta (Aβ) in AD and repetitive brain injury in the chronic traumatic encephalopathy [1]. In this review, we provide a brief overview of the structure and expression of tau and its physiological and pathological functions in tauopathy and then focus on the interaction of tau with heparan sulfate proteoglycan and related pathological processes. 2. The Tau ProteinTau protein belongs to the microtubule-associated proteins family [4]. Tau is found predominantly in the axon of the adult neurons and at low levels in the cell body, dendritic spines, and axonal presynaptic terminal [4]. The subcellular distribution of the tau is developmentally and environmentally regulated and isoform-dependent [5,6,7,8]. In addition, tau is detected at low levels in glial cells and outside cells [2,7,9]. The various subcellular, extracellular, and cell-type expressions indicate that tau may play various functions under different circumstances. Human tau is encoded by a single gene, microtubule-associated protein tau (Mapt, located on chromosome 17q21. Mapt gene comprises 16 exons (Figure 1. By alternative splicing of exon 2 (E2), exon 3 (E3), and exon 10 (E10), six isoforms of tau are expressed in the normal adult human brain [2]. The largest isoform contains 441 amino acid residues, including an N-terminus projection domain with two inserts (N1 and N2), a proline-rich domain, and a microtubule-binding repeat domain composed of four repeat motifs (R1–R4) that mediate microtubule-binding and tau aggregation [2]. The isoforms differ in the absence or presence of N1, N2, and R2 domains leading to the following 6 isoforms: 2N4R, 1N4R, 0N4R, 2N3R, 1N3R, and 0N3R (Figure 1) [1,3,10]. The N1 and N2 are encoded by E2 and E3, respectively, and R2 is encoded by E10. Besides the six tau isoforms, researchers also widely used another two truncated tau proteins, the K18 and K19, which contain only the four and three microtubule-binding domains, respectively (Figure 1). Given the differential distribution of tau in different cell compartments and various tau isoforms, tau likely plays different functions in different environments. 3. Tau in Physiological States As a microtubule-associated protein, Tau promotes axonal outgrowth through stabilizing neuronal microtubules [2]. Early studies indeed showed that tau stabilizes the axonal microtubules, promotes the assembly of microtubules, and regulates the dynamic instability of the microtubules (Figure 2A) [11,12,13], suggesting that tau is critical for developing a healthy neuron. However, knockout of mouse Mapt, the gene encoding tau protein, does not lead to a severe developmental defect or overt abnormalities at young ages [14,15,16]. Additionally, in neuronal cell line studies, tau was shown to inhibit axonal transport through multiple mechanisms, including competing with kinesin or dynein for binding to microtubules [17], competing with other cargos for binding to kinesin [18,19], reducing the number of cargo-associated kinesin motors [20], and releasing cargos from the kinesin chains [21] (Figure 2A). However, knockout or overexpression of tau does not alter axonal transport in cultured primary neurons [22,23,24]. These observations suggest that tau might be dispensable for microtubule assembly, stability, and axonal transport. The lack of the expected microtubule and axonal transport defects in the Mapt−/− mice is likely due to redundancy in function among tau and compensatory effect from other microtubule-associated proteins for the loss of tau [15,25]. Tau has been suggested to regulate synaptic physiology and plasticity, although it is expressed at a low level in the dendrites (Figure 2A) [26,27,28]. Tetsuya et al. observed a selective deficit in long-term depression (LTD) but not long-term potential (LTP) in Mapt−/− mice [27]. A study reported by Ahmed et al. observed a severe deficit in LTP, but no change in LTD, in another Mapt−/− mice [28]. Tau is also expressed in the nuclei and is thought to maintain the integrity of the genomic DNA (Figure 2A) [29]. Recent studies have emerged that tau may generally act as a scaffolding protein interacting with various kinases and phosphatases to regulate phosphorylation-based signaling pathways, functioning as a signaling hub protein within cells [30].The Mapt−/− mice studies have uncovered that tau functionally involves neurogenesis, locomotor function, and the learning and memory [31]. In neurogenesis studies, various lines of Mapt−/− mice were reported with contradictory results, showing that tau deficiency either decreased [16,32] or increased [14,33,34] neurogenesis reflected by the neuroD and DCX positive cells in the mouse brain. In behavioral studies, Mapt−/− mice displayed impairments to motor function, learning, and memory at 10–11 weeks old [35] and marked motor deficits at old age (12–20 months) [36,37]. However, other studies reported normal learning, memory, and motor function in middle-aged Mapt−/− mice [14,38,39,40]. Although the observations are inconsistent, these studies do demonstrate that tau critically plays vital physiological functions in the CNS. 5. Heparan Sulfate ProteoglycansHeparan sulfate proteoglycans (HSPGs) are macromolecules ubiquitously expressed in mammalian tissues. They are comprised of a core protein to which one or more HS glycosaminoglycan (GAG) chains are covalently attached [87,88]. The HSPGs are classified based on the location of their core proteins (Figure 3A). Syndecans (SDCs) and glypicans are the two major membrane-bound PGs that are linked to the plasma membrane by a transmembrane domain or a glycosylphosphatidylinositol (GPI) linker, respectively. In addition, there are three “part-time” cell-surface PGs, betaglycans, neuropilins, and CD44v3, which do not always have an HS chain moiety and are located on the cell surface through their transmembrane domains. Other HSPGs include agrin, perlecan, type XVIII collagen in the extracellular matrix (ECM), and serglycin in the intracellular secretory vesicles (Figure 3A) [87,88].Most of the biological functions of HSPGs are mediated by their HS chains [87]. The HS chain is a linear polysaccharide containing 50–200 disaccharide repeats composed of uronic acid (either glucuronic acid (GlcA) or iduronic acid (IdoA)) and N-acetylglucosamine (GlcNAc) (Figure 3B). The biosynthesis of HS occurs in the endoplasmic reticulum and the Golgi apparatus and involves various enzymes. HS biosynthesis occurs in three major steps: chain initiation, elongation, and modification [88]. Before HS biosynthesis, the xylose (Xyl) residue of a tetrasaccharide linker, GlcA-galactose (Gal)-Gal-Xyl, is covalently linked to a selected serine residue in the core protein [89]. HS biosynthesis is initiated by exostosin-like glycosyltransferase 3 (Extl3), which attaches the first GlcNAc residue to the GlcA residue of the linker to form the first GlcNAc-GlcA disaccharide repeat, followed by Ext1/Ext2 that alternately adds GlcA and GlcNAc to extend the HS chain. Meanwhile, the nascent HS chain undergoes a series of modifications, including the replacement of the N-acetyl groups in GlcNAc residues with N-sulfo groups by N-deacetylase-N-sulfotransferases (Ndst), the epimerization of GlcA to IdoA by D-glucuronyl C5-epimerase (Glce), and the addition of sulfate groups at the C2 position of adjacent IdoA residues by 2-O-sulfotransferases (Hs2st), the C6 position of GlcNAc residues by 6-O-sulfotransferases (Hs6st), and the C3 position of the GlcNS residues by 3-O-sulfotransferases (Hs3st) (Figure 3B) [90]. In addition, the mature HS is subjected to remodeling on the cell surface by HS 6-O-endosulfatases (Sulf) that selectively remove 6-O-sulfo groups from glucosamine residues [90]. Mature HS in the ECM can also be truncated by heparanases secreted by immune cells, which selectively target a variety of trisaccharide sequences and hydrolyze the β-O-linkage between the GlcA and glucosamine residues [91].Due to substrate specificity and incompletion of the modification by the enzymes, the modifications tend to occur in clusters and generate tremendous structural heterogeneity. The modification patterns form binding sites for many protein ligands, including growth factors, growth factor receptors, and tauopathy-related proteins such as tau, Aβ, and α-syn [92,93,94,95]. In addition, the HS structures are cell-type/tissue/developmental stage-specific, indicating that HS may interact selectively with a fraction of protein ligands to play spatiotemporal regulatory roles under different biological conditions [87,96]. 6. Heparan Sulfate-Tau Interaction: The Related StructuresHS and heparin, a highly sulfated form of HS, directly bind to tau protein [97,98]. Snow et al. and Su et al. observed by ultrastructural immunolocalization that HS co-localizes with NFTs in neurons in the brains of AD patients [99,100], suggesting that HS interacts with tau in the AD brain. Like other heparin-protein interactions [101,102], the tau/HS interface is dominated by electrostatic interaction mediated through the highly positively charged residues/domain within the tau protein and the highly negatively charged sulfate residues within HS. Further studies determined that the hexapeptides 275VQIINK280 in R2 and 306VQIVYK311 in R3 of tau bind to HS [103,104,105,106,107]. In NMR titrations of heparin to K18, L282 displayed the most chemical shift perturbation, suggesting that the two lysines near 275VQIINK280 in R2, K280 and K281, are likely crucial to heparin and HS binding. A recent study further showed that Arixtra, a sulfated pentamer analog of heparin, bound proline-rich region II (PRR2) with 4 times stronger binding affinity than R2 [108]. These studies indicate that the major HS binding sites within tau protein are in the R2 and PRR2 domains (Figure 4A).In parallel, several studies examined the HS structural features involved in tau binding. Hasegawa et al. suggested that the overall sulfation level of different GAGs determines their binding affinity to tau, such as heparin having a higher overall sulfation level than HS and showing a higher binding affinity to tau [109]. However, further studies with chemically modified heparins revealed that the binding affinity to tau also depends on heparin’s delicate structure. Removal of N- and 6-O-sulfation significantly reduced tau-heparin binding, while the removal of 2-O-sulfation had less impact [95,106].Evaluating the interaction of tau with 52 pure and defined HS oligosaccharides on a glycan microarray revealed a striking 3-O-sulfation dependence; the binding of tau was increased 1.5- to -5-fold by adding a single 3-O-sulfation group on a septamer or dodecamer HS oligosaccharide [93]. Sepulveda-Diaz et al. reported that 3-O-sulfated HS interacts with tau and promotes tau phosphorylation [110]. Our recent study also showed that introducing a 3-O-sulfate significantly increased the HS binding affinity to tau [93]. These biochemical studies strongly prove that the binding of HS/heparin to tau depends on N-, 6-O-, and 3-O-sulfations. 8. Aberrant HSPG Expression in AD and Other TauopathiesThe proteoglycan cores of HSPGs are differentially expressed in AD compared with normal human brains, which has been shown to impact tau internalization [132]. The impact of Alzheimer’s disease on membrane-bound HSPG agrin distribution has been documented since 1999; in normal brains, agrin is in the soluble fraction of detergent extracted samples, while in AD brains, agrin shifts to the insoluble fraction [133]. One study found that genes associated with syndecan-4 (SDC4), followed by SDC3, SDC2, agrin, and serglycin (the dominant intracellular PG in immune cells) [134] were consistently overexpressed in AD brains compared with controls [135]. SDC3 has the highest affinity for tau monomer, but interestingly, overexpression of SDC3 and -4 in K562 cells (lymphoblast cells with low endogenous HS) did not lead to a greater internalization of tau, but increased fibrillation and accumulation of fibrils on the cell surface [132]. These studies emphasize the impact of tauopathies on the specific core proteins of HSPGs.Several studies have documented altered HS expression in AD patients (Table 1). Su et al. examined 7 AD patients and 4 age-matched controls and observed that the number and intensity of the HS co-staining with PHFs were denser in AD than in control brains [99]. Shimizu et al. examined 25 AD patients with 10 non-demented elderly patient controls. They detected a 9.3-fold HS increase in the hippocampus and a 6.6-fold increase in gyrus frontalis superior in the AD patients. They also observed that HS is most abundantly expressed in the basement membrane of capillary endothelial cells [136]. Other groups’ studies confirmed the abnormal HS expression in the AD brain [92,137].AD HS showed significantly altered interaction with heparin-binding proteins. Compared to the control brains, GAGs isolated from the AD brains showed decreased binding to growth factors, such as fibroblast growth factor 2 (FGF-2), brain-derived neurotrophic factor and vascular endothelial growth factor (VEGF), and increased binding to tau, heparin-binding EGF-like growth factor and pleiotrophin [92], reflecting GAG structural alteration in AD brain. This has been supported by a recent study showing multiple sulfated disaccharides (ΔUA2S-GlcNS, ΔUA2S-GlcNAc, ΔUA-GlcNAc6S, ΔUA2S-GlcNAc6S) and a tetrasaccharide with rare 3S (ΔUA-GlcNAc6S-GlcA-GlcNS3S6S) were increased in AD [137]. These increased di- and tetrasaccharides are rich in N-, 6-O-, and 3-O-sulfation that bind tau [137] and may increase tau’s propensity for aggregation.Consistent with these immunostaining and biochemical analyses, the transcripts of several HS-related genes were up-regulated in the AD brain, including Hs3st2 in Sepulveda-Diaz, J.E. et al. study [110], Ndst2, Hs3st2, Hs3st4 and Glce in Huynh et al. study, [92], Hspe and Hspe2 in Garcia et al. study [140], and Extl3, Hs6st1, Hs3st1, Hs3st2, Hs3st3A1, Hs3stB1, Hs6st5 and Hs6st6 in severe AD in Pérez-López et al. study [139], and down-regulated, including HS 6-O-endosulfatase-2 (Sulf2) in Roberts et al. study [138] and Sepulveda-Diaz, J.E. et al. study [110]. The Pérez-López et al. study has shown, so far, the most comprehensive HS gene expression profile in AD study, analyzing all HS biosynthesis and remodeling/degradation genes expression in different AD stages and different brain regions [139]. Overall, the results in Pérez-López’s study correlate HS gene expression with AD pathology. The positive or negative correlation depends on the disease’s severity, the area of the brain regions, and the gene function in HS biosynthesis. The most obvious is the upregulation of Extl3, Hs6st1, and six of the seven members of the Hs3st family in severe AD. These findings revealed that the aberrant HS gene expression might generate more tau-binding sites to enhance HS-facilitated tau aggregation, thereby exacerbating the tauopathy [139]. Furthermore, upregulation of HSPE and HSPE2 in AD brains indicates an active role of HS restructuring during disease progression, as the increased expression correlated closely with each concurrent Braak stage of AD [141].Although most tauopathy studies focus on AD, several studies examined HS expression in other tauopathies. HS co-deposits with NFTs in PiD, Niemann-Pick disease type C, subacute sclerosing panencephalitis, myotonic dystrophy, and motor neuron disease [105,142], although it remains unknown if HS expression is altered in these tauopathies. 9. Future Studies from the HSPG Aspect

Much remains to be investigated to understand the pathogenesis of tauopathies better. From the HSPG perspective, most of our knowledge has been gained from AD studies compared to other tauopathies. We have learned that HS critically regulates tau protein secretion, internalization, aggregation, and phosphorylation. We also learned some structural features of the HS motifs that interact with tau protein. However, we are far from clearly defining the roles of HSPGs in all tauopathies and several questions remain, including:

(1). Cell-type specific roles of HS in tauopathies. The structure of HS is cell-type/developmental/disease-stage dependent, and the biological functions of HSPGs are also location dependent, such as cell surface-anchored vs. in the extracellular matrix. HS on the surfaces of different CNS cell types may also have different roles; e.g., one may hypothesize that HS on neurons may promote the transcellular spread of tau. In contrast, HS on glial cells may promote tau degradation and clearance. It will be essential to understand the spatial and temporal regulatory processes and roles of HS in the pathogenesis of tauopathies, including in tau secretion, internalization, aggregation/deposition, posttranslational modification, and pathological prion-like propagation.

(2). The delicate HS structures that bind tau protein. It has been demonstrated that the tau-binding HS motifs contain N-, 6-O-, and 3-O-sulfation with 2-O-sulfation in debate. However, their chemical composition, and more importantly, their delicate modification patterns, are unknown. Successful delineation of the tau-binding site fine structure will open the door to a better understanding of the structure-function relations of HS in interaction with tau proteins, including its six normal isoforms and truncated forms with and without posttranslational modifications. This new information will significantly aid drug design to treat the various tauopathies.

(3). Testing if pharmacological inhibition of HS-tau interaction will ameliorate tauopathy. An early study reported that the treatment with low-molecular-weight heparin prevented abnormal tau protein formation in rat hippocampus [143]. Other studies have observed that heparin, heparin-like molecules (heparinoids and oligosaccharides) competitively inhibit cellular tau uptake in vitro and in vivo [95,111,144] and decrease tau-induced cell toxicity [145]. These findings revealed HS as a promising therapeutic target to inhibit the progression of tauopathies.Several strategies have been studied targeting HSPGs for therapeutic development, focusing on cancer treatment [146,147]. These include anti-HSPG antibodies, HS antagonists, HS mimetics, and synthetic xylosides. The human monoclonal HS-specific antibody HS20 blocks the activation of the HS-dependent hepatocyte growth factor (HGF)/Met pathway. Consequently, it inhibits HGF-induced hepatocellular carcinoma cell migration, motility, spheroid formation, and liver tumor growth in vivo [148]. Synstatin, a mimetic peptide, inhibits the signaling complex formation between SDC-1, IGF1R, and integrin αvβ3 and attenuates HS-dependent angiogenic VEGF and FGF2 signaling, and blocks tumor angiogenesis in vivo [149,150]. Surfen, bis-2-methyl-4-amino-quinolyl-6-carbamide, had been previously identified as a small molecule antagonist of HS [151]. It neutralizes the anticoagulant activity of unfractionated and low molecular weight heparins and blocks HS-dependent angiogenic FGF2 and VEGF signaling in cultured endothelial cells. Recently, surfen was reported to reduce tumorigenicity of glioblastoma cells in the rat brain [152] and of Ewing sarcoma cells in a zebrafish model [153]. M402 is a rationally engineered, non-cytotoxic HS mimetic and effectively inhibits murine melanoma cell seeding to the lung, a process potentially facilitated by HS, in an experimental metastasis model [154].Xylosides compete with proteoglycan for HS biosynthetic enzymes and prime GAG chains secreted into the extracellular environment to compete with endogenous proteoglycan-linked GAGs for different binding ligands [155]. By these two mechanisms, xylosides act to block HSPG functions. Xylosides have been shown to inhibit glioblastoma cell viability [156], glioma cell invasion [157], tumor angiogenesis in vitro [158], and various tumor cell line growth in vitro and human bladder carcinoma growth in vivo [159]. These anti-HSPG strategies and available agents give great potential to future tauopathy treatments, exemplified by a recent study from Naini et al. showing that surfen and oxalyl surfen decreased tau hyperphosphorylation and mitigated neuron deficits in vivo in a zebrafish model of tauopathy [160].

In summary, ample evidence supports that HSPGs are critically involved in tauopathy. Heparan sulfate proteoglycans directly bind to co-deposit with tau and modulate tau secretion, internalization, and aggregation. Meanwhile, the expression of heparan sulfate proteoglycans is dysregulated in the disease, and this dysregulation may exacerbate tauopathies. These understandings merge a great potential of targeting heparan sulfate proteoglycan-tau interaction as a novel therapeutic option for the disease.