Interplay between glypican-1, amyloid-β and tau phosphorylation in human neural stem cells

Introduction: Alzheimer ’ s disease (AD) is characterized by accumulation of amyloid beta (A β ) and hyper-phosphorylated tau (Tau-P) in the brain. A β enhances the activity of kinases involved in the formation of Tau-P. Phosphorylation at Thr 181 determines the propagation of multiple tau phosphorylations. A β is derived from the amyloid precursor protein (APP). Cleavage of APP by β -secretase also initiates release of heparan sulfate (HS) from the proteoglycan glypican-1 (GPC1). Objectives: In this study, we have explored possible connections between GPC1 expression, HS release, APP processing and Tau-P formation in human neural stem cells. Methods: GPC1 formation was suppressed by using CRISPR/Cas9 and increased by using a vector encoding GPC1. HS release from GPC1 was increased by growing cells in medium containing Arg and ascorbate. Effects were monitored by immunofluorescence microscopy and slot immunoblotting using antibodies/antisera recognizing A β , GPC1, HS released from GPC1, total Tau, and Tau phosphorylated at Thr-181, 217 or 231. The latter have been used as blood biomarkers for AD. Results: Suppression of GPC1 expression resulted in increased phosphorylation at Thr 181 and Thr 217. When GPC1 was overexpressed, phosphorylation at Thr 217 decreased. Stimulation of HS release from GPC1 diminished tau phosphorylation at all of the three Thr positions, while expression of GPC1 was unaffected. Simultaneous stimulation of HS release and APP processing by the cytokine TNF-α also suppressed tau phosphorylation. Conclusion: The increased release of GPC1-derived HS may interfere with A β formation and/or A β interaction with tau.


Introduction
In the brain of Alzheimer's disease (AD) patients, there is excessive formation of amyloid beta (Aβ) peptides, which accumulate extraneuronally as senile plaques in early stages of the disease.This is followed by intraneuronal accumulation of hyperphosphorylated tau (Tau-P) which forms neurofibrillary tangles.The occurrence of tangles correlates better with clinical symptoms than do Aβ plaques.Aβ accelerates tau hyperphosphorylation, but there could also be upstream modulators (for recent reviews, see e.g.Busche and Hyman 2020;Lauretti et al., 2020;Wegmann et al., 2021;Zhang et al., 2021;Roda et al., 2022).Age and the presence of the ε4 allele of apolipoprotein E (ApoE4) are two of the most influential risk factors for sporadic, late onset AD (Geula et al., 1998; for recent reviews, see e.g.Williams et al., 2020;Uddin et al., 2021).
In tau, which is present in the cytosol, there are numerous phosphorylation sites in the N-terminal and mid-regions and also in the microtubule-binding region (Fig. 1A).A moderate level of phosphorylation is required for its normal function.However, Aβ enhances the activity of kinases involved in the phosphorylation of multiple Ser/Thr sites (Zhang et al., 2021).Hyperphosphorylated tau (Tau-P), which aggregates into insoluble filaments, is partly degraded by proteolysis generating aggregation-prone regions that form the neurofibrillary tangles and a soluble fraction, portions of which are released into the cerebrospinal fluid and subsequently enter the blood.Tau-P phosphorylated at Thr-181, Thr-217 or Thr-231 (indicated in Fig. 1A) have been successfully utilized as blood biomarkers for AD (for reviews, see Ossenkoppele et al., 2022;Karikari et al., 2022;Li et al., 2024).
In this study, we have explored possible connections between GPC1 expression, HS-anMan release and tau phosphorylation in human neural stem cells (NSC).We find that stimulation of HS release from GPC1 suppresses tau phosphorylation at Thr-181, Thr-217 and Thr-231.Fig. 1.A) Processing of APP, GPC1 and Tau.Aβ, amyloid beta; anMan/AM, anhydromannose; Asc, ascorbate; β-CTF, C-terminal fragment of APP generated by β-secretase cleavage; β-NTF, N-terminal fragment of APP generated by β-secretase cleavage; SH, thiol; SNO, S-nitrosothiol; HS is denoted by blue rods; GPI is denoted by the blue oval; microtubule-binding region in tau is denoted by red squares; P, phosphate groups at Thr-181, -217 and -231 in Tau-P, respectively.B) Representative immunofluorescence images at low and high magnifications of human neural stem cells (NSC) after knockout of GPC1.Sparse cultures were transfected with either a non-specific control CRISPR/Cas9 plasmid (Control) or a GPC1-targeted knockout plasmid (CRISPR/Cas9) and grown to near confluence.Staining was performed with GPC1 antiserum (red, 1:1000) or mAb AM (for HS-anMan, red, 1:250), and DAPI (for nuclei, blue).Exposure time was the same in both cases.Insets, fluorescence intensities in arbitrary unit ± SE of GPC1 and HS-anMan (AM) staining, respectively, divided by the DAPI intensity; n = 5.P-values 0.00000000487 and 0.000000004826 for GPC1 and AM, respectively.

Materials and reagents
Human NSCs from the cortex region of human brain and human neural stem cell growth medium were purchased from AcceGen Biotech (Cat.# ABM-SM0105 and Cat.# ABM-SM0105 respectively).Accutase and Matrigel were obtained from Fisher Scientific (Cat. # 11599686 and Cat. # 11573560, respectively).A rabbit antiserum against human GPC1 was obtained after immunization with 6-His tagged recombinant GPC1 core protein comprising the sequence Ile 54 to Pro 519.The creation of the vector, the expression of the protein in E. coli M15 and its subsequent purification have been described in detail elsewhere (Svensson and Mani 2009;Svensson et al., 2009).The Aβ antibody was pAb Aβ40 (PA3-16760) and the tau antibodies were pAb Tau (PA5-29610), pAb Tau-P Thr 217 (PA5-37639), pAb Tau-P Thr 231 (44-746G), and mAb Tau-P Thr 181 (AT270), all from Invitrogen.mAb AM is specific for HS/ heparin tetrasaccharide or larger fragments that are generated by partial deaminative cleavage at N-unsubstituted glucosamines and therefore terminate with anhydromannose (anMan) at the reducing end (Pejler et al., 1988;Cheng et al., 2012).It has been characterized in previous studies (Ding et al., 2002;Mani et al., 2004).The secondary antibodies were: FITC-tagged goat anti-mouse Ig from Sigma-Aldrich, Alexa-Fluor 594-tagged donkey anti-rabbit IgG from Invitrogen and horseradish peroxidase-conjugated anti-rabbit and anti-mouse IgG from Bio-Rad.The commercial antibodies were used as recommended by the manufacturers.The DNA staining compound 4,6-diaminido-2-phenylindole (DAPI), Arg and ascorbate were obtained from Sigma-Aldrich and TNF-α from Alomone labs.Proteinase inhibitor kit (cOmplete mini) was from Roche, and BCA Protein Assay Kit from Thermo Fisher and used according to the manufacturer's instructions.

Cell culture
Human Neural Stem Cells were grown and maintained in human neural stem cell growth medium.The cells were cultured in a 37 • C, 5 % CO 2 humidified incubator.Every 2-3 days, half of the culture medium was replaced.For dissociation, cells were treated with neural stem cell dissociation solution, Accutase, for 2-3 min at room temperature.Two volumes of human neural stem cell growth medium were added to the cells, which were then gently detached by pipetting up and down.The cells were subsequently centrifuged at 200 g for 3 min at room temperature.The supernatant was then removed and the cells were resuspended in human neural stem cell growth medium and plated in Matrigel coated plates.

CRISPR/Cas9 targeting of GPC1
NSC were transfected with either a GFP-tagged, human GPC1 targeted CRISPR/Cas9 knockout plasmid (sc-402002-NIC; Santa Cruz Biotechnology) or a non-specific, GFP-tagged CRISPR/Cas9 plasmid not targeting any known gene (sc-437281; Santa Cruz Biotechnology).Ten μl of UltraCruz Transfection reagent (sc-395739) was incubated with 1 μg of plasmid DNA in 50 μl of Plasmid Transfection Medium (sc-108062) for 5 min at room temperature.Prior to transfection, the culture media were replaced with 700 μl of fresh antibiotic-free Plasmid Transfection Medium.Then, 50 μl of the Plasmid DNA/UltraCruz Transfection Reagent Complex was added dropwise to each well and the cells were incubated for 24-72 h.Successful transfections were monitored by immunofluorescence microscopy (green channel).

Overexpression of GPC1
NSC were transiently transfected with a vector encoding GFP-tagged GPC1-protein for 72 h using Promega standard protocol for transfection with FuGENE 6 Transfection Reagent (E2691; Promega Biotech AB).The GPC1-coding sequence included an N-terminal signal peptide, the GFP with a disrupted start codon, followed by the GPC1 core protein including the C-terminal signal peptide for membrane attachment.The procedure to create the vector has been described in detail elsewhere (Cheng et al., 2011).Successful transfections were monitored by immunofluorescence microscopy (green channel).

Deconvolution immunofluorescence microscopy
NSC were examined by immunofluorescence microscopy as described previously (Cheng et al., 2014).In detail, cells cultured in 4 well slides were fixed in acetone for 2 min at room temperature in order to retain cellular and subcellular structures and to ensure the preservation of carbohydrates.The fixed cells were first pre-coated with 10 % anti-mouse total Ig and then exposed to primary antibodies overnight.The secondary antibodies used were FITC-tagged goat anti-mouse Ig when the primary antibody was a mouse monoclonal and Alexa Fluor 594-tagged goat anti-rabbit IgG when the primary antibody was a polyclonal.Controls omitting the primary antibody was used to check for nonspecific binding of the secondary antibodies.DNA staining with DAPI, as well as staining with commercial antibodies was performed as recommended by the manufacturers.The fluorescent images were analyzed by using a Carl Zeiss AxioObserver inverted fluorescence microscope with deconvolution technique and equipped with objective EC "Plan-Neofluar" 63x/1.25 Oil M27 and AxioCam MRm Rev Camera.Identical exposure settings and times were used for all images.During microscopy, the entire slides were scanned and immunofluorescence images at 20x and 100x magnifications were captured.The low magnification images were used to identify representative locations for high magnification images and for intensity measurements.Immunofluorescence signal in 5 randomly chosen low-magnification areas were quantified by densitometry using Zeiss ZEN 3.5 pro blue edition software and presented relative to DAPI staining.For co-localization analysis, entire images were also examined at 2,5D.Data analysis was performed using Zeiss AxioVision Release 4.8 software.

Slot immunoblot
NSC were extracted with radio-immunoprecipitation assay (RIPA) buffer (0.1 % w/v SDS, 0.5 % v/v Triton X-100, and 0.5 % w/v sodium deoxycholate in PBS) supplemented with proteinase inhibitors (cOmplete mini) and shaken for 10 min at 4 • C. Protein concentrations were determined (BCA assay kit from Pierce), concentrations were normalized to 1 mg/ml and equal amounts of protein were loaded on the PVDF membranes using slot blot; n = 5 in each experiment.The membranes were then incubated with various antibodies, washed extensively with PBS containing 0.5 % Tween-20, treated with horseradish peroxidaseconjugated IgG (anti-mouse for mAbs and anti-rabbit for pAbs) and developed by chemiluminescence (Pierce fast western blot kit) using Amersham ImageQuant 500 detector from Cytiva.Staining intensities were recorded by densitometry using GelAnalyzer 19.1.No signal was observed when the primary antibody was omitted.The data points are shown as the means ± SE.

Statistical analysis
The data points in the graphs are shown as the means ± SE (error bars) as indicated in the figure legends; n = 5 in each experiment.For statistical analysis, 2 group comparisons were performed using unpaired 2-tailed Student t-test and unequal variances data analysis.Error probabilities of P < 0.05 were considered to be statistically significant.Indication of P-value summarizes: ns (not significant) P ≥ 0.05, *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001.

GPC1 is the major source of HS-anMan
To show that HS-anMan was derived from the GPC1 proteoglycan, we used CRISPR/Cas9 to knockout the GPC1 gene (CPC1) in human neural stem cells (NSC).Cells were examined by immunofluorescence microscopy after staining with antiserum against human GPC1 and a monoclonal antibody (mAb AM) that detects HS-anMan.In the control, there was expression of GPC1 and also formation of HS-anMan (Fig. 1B, Control).Knockout of GPC1 resulted in an 89 % reduction in HS-anMan formation compared to the control (Fig. 1B, CRISPR/Cas9; insets, intensity measurements, white bars).

Stimulation of HS-anMan release from GPC1 suppresses tau phosphorylation
Deaminative cleavage of HS is catalysed by Cys-SNO located in the stem region of the GPC1 core protein near the HS attachment sites (Fig. 1A) (Cheng et al., 2012;Awad et al., 2015).As NO is consumed in the deaminative process (to N 2 ) (Cheng et al., 2012), Cys-SH has to be re-S-nitrosylated during recycling of GPC1 in order to restore the conditions for HS release (Cheng et al., 2002;Ding et al., 2002).NO is generated by NO-synthases from arginine (Arg) and molecular oxygen (O 2 ) (Stamler et al., 1992).When the capacity to release HS-anMan is exhausted, supplementation with Arg restores the capacity (Cheng et al., 2022).To examine if stimulated release of HS-anMan from GPC1 could affect tau phosphorylation, NSC were grown in regular medium or in medium containing both 1 mM Arg and 1 mM ascorbate.Cell extracts were analysed by slot immunoblotting by using GPC1 antiserum, pAb Tau (for total Tau) as well as antibodies specific for Thr-P 181, 217 and 231 in Tau-P, respectively.Expression of GPC1 was unaffected (Fig. 4A), while phosphorylation at all of the three Thr positions diminished (Fig. 4B-D); the greatest effect (approx.50 %) was at position 181 (Fig. 4B).

Stimulation of HS-anMan release from GPC1 increases interaction between HS-anMan and Aβ
NSC were grown to confluence in regular medium or in medium containing both 1 mM Arg and 1 mM ascorbate and then subjected to immunofluorescence microscopy.Cells were co-stained with mAb AM (for HS-anMan) and pAb Aβ40, and with mAb AM (for HS-anMan) and pAb Tau (for total Tau).As expected, there was an increase in HS-anMan formation (Fig. 5B and D, AM; 2D images), and an overlap of HS-anMan and Aβ staining (Fig. 5B, yellow in merged; 2D and 2.5D images).Overlap between HS-anMan and tau staining appeared less pronounced (Fig. 5D, merged; 2D and 2.5D images).

Tumour necrosis factor-α (TNF-α) suppresses tau phosphorylation
TNF-α upregulates expression of β-secretase (Deng et al., 2017).This leads to increased cleavage of APP into β-NTF and β-CTF and, indirectly, to increased release of HS-anMan from GPC1 (Fig. 1A).We have previously shown that TNF-α induces clustered co-accumulation of HS-anMan and β-CTF in NSC and N2a neuroblastoma cells (Cheng et al., 2020).To examine if this could affect tau phosphorylation, NSC were grown in regular medium or in medium containing TNF-α.To monitor tau phosphorylation, cell extracts were analysed by slot immunoblot using pAb Tau (for total Tau) as well as antibodies specific for Thr-P 181, 217 and 231 in Tau-P, respectively.The extent of tau phosphorylation at all of the three Thr positions diminished; Thr 181 to 49 % of untreated (Fig. 6A), Thr 217 to 28 % of untreated (Fig. 6B) and Thr 231 to 21 % of untreated (Fig. 6C).

Discussion
The tau protein contains numerous Thr and Ser phosphorylation sites.Clusters of phosphorylated Thr/Ser, including Thr 181, 217 and 231, appear in tau obtained from cerebrospinal fluid of AD patients (Barthélemy et al., 2022).A mechanistic link between an initial sitespecific phosphorylation and subsequent multi-site phosphorylations has been described.Thr 181 is one of three master sites that determine the propagation of multiple phosphorylations (Stefanoska et al., 2022).As shown here, increased release of HS-anMan from GPC1 was obtained by including Arg and ascorbate in the NSC culture medium.This resulted in decreased phosphorylation of Thr 181 as well as of Thr 217 and Thr 231, while the expression level of GPC1 was unchanged.We conclude that the effect on tau phosphorylation was mediated by the released HS-anMan rather than the GPC1 core protein.Modulation of GPC1 expression affects the capacity to generate HS-anMan.In this regard, suppression of GPC1 expression was more efficient than overexpression of GPC1.Stimulation of HS-anMan release from GPC1 appeared to induce increased interaction between HS-anMan and Aβ.Reduced tau phosphorylation was also observed when NSC was exposed to TNF-α.As shown previously, this cytokine induces formation of SDS-stable complexes between HS-anMan and β-CTF which ultimately accumulate in autophagosomes of neuronal cells (Cheng et al., 2020).This may preclude generation of Aβ when APP processing is augmented.The GPC1 protein contains the nuclear localization signal KRRRGK and portions of non-glycanated GPC1 could potentially return to the nucleus and regulate the expression of kinases involved in tau phosphorylation.Indeed, presence of GPC1 in the nuclei of neurons has been observed (Liang et al. 1997).However, we found no evidence for a nuclear localization of GPC1, nor HS-anMan, in NSC (Fig. 1B, Control; GPC-1/DAPI, AM/DAPI, high magnification).The GPC1-derived HS is also a potential modulator of α-synuclein aggregation and a possible vehicle for the transport of both α-synuclein and APP degradation products to autophagosomes (Cheng et al. 2020(Cheng et al. -2023)).
Humans cannot synthesize ascorbate due to lack of gulonolactone oxidase (GulO) and are therefore dependent on dietary supply of vitamin C.There are two bio-available forms of vitamin C, ascorbic acid and its oxidized form dehydroascorbic acid.The former is taken up by the Na +dependent vitamin C transporter SVCT1 (for review, see Bȕrzle et al., 2013).Dehydroascorbic acid uptake is mediated by the facilitative hexose transporters GLUT2 and GLUT8 (Corpe et al., 2013).A competition between glucose and dehydroascorbic acid uptake is therefore anticipated.In the enterocytes, dehydroascorbic acid is efficiently converted to ascorbate, which can then be transferred to the blood via anion channels and then traverse the blood-brain barrier via SVCT2 (for review, see Lykkesfeldt and Tveden-Nyborg, 2019).
Mice can synthesize ascorbate.However, most familial AD mouse models do not take this into account.A mouse model of sporadic human AD is not yet available.In a familial AD mouse model unable to synthesize ascorbate, supplementation of vitamin C reduced amyloid plaque burden (Kook et al. 2014).In another familial AD mouse model with SVTC2 +/− resulting in decreased brain vitamin C, there was accelerated amyloid deposition and cognitive impairment (Dixit et al., 2015).
It has been debated whether vitamin C supplementation could be of prophylactic value in human AD.Clinical trials have given inconsistent results, which may be due to variation in the modes of dietary intake (see e.g.Monacelli et al., 2017).In a population-based prospective study of aged individuals, women with ApoE4 and a high blood vitamin C level had a significantly reduced risk of cognitive decline (Noguchi-Shinohara et al., 2018).A meta-analysis of 12 studies quantifying the vitamin C plasma levels in AD and control subjects showed a significant decrease in the plasma vitamin C levels of AD patients as compared to healthy controls (Hamid et al., 2022).

F
.Cheng et al.

F
.Cheng et al.

Fig. 5 .
Fig.5.Growth in the presence of arginine and ascorbate increases overlap between HS-anMan and Aβ staining.Representative immunofluorescence images of NSC that were grown to confluence in regular medium (UT) or in medium containing 1 mM arginine and 1 mM ascorbate (Arg + Asc).Staining was performed with mAb AM (for HS-anMan, green, 1:250), pAb Aβ40 (for Aβ, red), pAb Tau (for total Tau, red), and DAPI (for nuclei, blue).Exposure time was the same in all cases.Images are presented both as 2D and 2.5D.UT = untreated.

Fig. 6 .
Fig. 6.Growth in the presence of TNF-α decreases the extent of phosphorylation at Thr-181, Thr-217 and Thr-231.Immunoblots of RIPA extracts of NSC grown to confluence in regular medium (UT) or in medium containing 100 pg/ml TNF-α.The extracts were blotted on PVDF membranes and probed with pAb Tau (for total Tau) and (A) mAb Tau P-181, (B) pAb Tau P-217, or (C) pAb Tau P-231 and expressed as Tau-P/total Tau.UT, untreated; the error bars indicate band intensities in arbitrary unit ± SE; n = 5.P-values: Tau P-181, 0.022; Tau P-217, 0.008; and Tau P-231, 0.004.