HS3ST2 expression is critical for theabnormal phosphorylation of tau inAlzheimer’s disease-related tau pathology

Julia Elisa Sepulveda-Diaz,1,2,† Seyedeh Maryam Alavi Naini,3,4,† Minh Bao Huynh,1,†

Mohand Ouidir Ouidja,1,2 Constantin Yanicostas,3,4 Sandrine Chantepie,1 Joao Villares,5

Foudil Lamari,6 Estelle Jospin,1 Toin H. van Kuppevelt,7 Ayikoe Guy Mensah-Nyagan,8

Rita Raisman-Vozari,2,* Nadia Soussi-Yanicostas3,4,* and Dulce Papy-Garcia1,*

†,*These authors contributed equally to this work.

Heparan sulphate (glucosamine) 3-O-sulphotransferase 2 (HS3ST2, also known as 3OST2) is an enzyme predominantly expressed

in neurons wherein it generates rare 3-O-sulphated domains of unknown functions in heparan sulphates. In Alzheimer’s disease,

heparan sulphates accumulate at the intracellular level in disease neurons where they co-localize with the neurofibrillary pathology,

while they persist at the neuronal cell membrane in normal brain. However, it is unknown whether HS3ST2 and its 3-O-sulphated

heparan sulphate products are involved in the mechanisms leading to the abnormal phosphorylation of tau in Alzheimer’s disease

and related tauopathies. Here, we first measured the transcript levels of all human heparan sulphate sulphotransferases in hippo-

campus of Alzheimer’s disease (n = 8; 76.8 � 3.5 years old) and found increased expression of HS3ST2 (P50.001) compared with

control brain (n = 8; 67.8 � 2.9 years old). Then, to investigate whether the membrane-associated 3-O-sulphated heparan sulphates

translocate to the intracellular level under pathological conditions, we used two cell models of tauopathy in neuro-differentiated

SH-SY5Y cells: a tau mutation-dependent model in cells expressing human tau carrying the P301L mutation hTauP301L, and a tau

mutation-independent model in where tau hyperphosphorylation is induced by oxidative stress. Confocal microscopy, fluorescence

resonance energy transfer, and western blot analyses showed that 3-O-sulphated heparan sulphates can be internalized into cells

where they interact with tau, promoting its abnormal phosphorylation, but not that of p38 or NF-�B p65. We showed, in vitro,

that the 3-O-sulphated heparan sulphates bind to tau, but not to GSK3B, protein kinase A or protein phosphatase 2, inducing its

abnormal phosphorylation. Finally, we demonstrated in a zebrafish model of tauopathy expressing the hTauP301L, that inhibiting

hs3st2 (also known as 3ost2) expression results in a strong inhibition of the abnormally phosphorylated tau epitopes in brain and

in spinal cord, leading to a complete recovery of motor neuronal axons length (n = 25; P5 0.005) and of the animal motor

response to touching stimuli (n = 150; P5 0.005). Our findings indicate that HS3ST2 centrally participates to the molecular

mechanisms leading the abnormal phosphorylation of tau. By interacting with tau at the intracellular level, the 3-O-sulphated

heparan sulphates produced by HS3ST2 might act as molecular chaperones allowing the abnormal phosphorylation of tau. We

propose HS3ST2 as a novel therapeutic target for Alzheimer’s disease.

1 Laboratory Cell Growth, Tissue Repair and Regeneration (CRRET), Centre National de la Recherche Scientifique (CNRS) EAUPEC 4397/ERL CNRS 9215, Universite Paris Est Creteil, Universite Paris Est, F-94000, Creteil, France

2 Sorbonne Universite UPMC UM75 INSERM U1127, CNRS UMR 7225, Institut du Cerveau et de la Moelle Epiniere, Paris, France3 INSERM UMR 1141, Hopital Robert Debre, 75019 Paris, France4 Universite Paris Diderot, Sorbonne Paris Cite, Paris, France5 Aging and Neurodegenerative Diseases Brain Bank Investigation Laboratory, Universidade Federal de Sao Paulo, Sao Paulo,

04023-062, Brazil

doi:10.1093/brain/awv056 BRAIN 2015: Page 1 of 16 | 1

Received April 25, 2014. Revised December 21, 2014. Accepted January 2, 2015.

� The Author (2015). Published by Oxford University Press on behalf of the Guarantors of Brain. All rights reserved.

For Permissions, please email: journals.permissions@oup.com

Brain Advance Access published April 4, 2015by guest on A

pril 5, 2015D

ownloaded from

6 Biochimie des Maladies Neuro-metaboliques, Groupe Hospitalier Pitie-Salpetriere, 75013 Paris, France7 Radboud University Medical Center, Radboud Institute for Molecular Life Sciences, Nijmegen, The Netherlands8 INSERM U1119, FMTS, Universite de Strasbourg, 67000 Strasbourg, France

Correspondence to: Dulce Papy-Garcia,

Laboratory CRRET,

EA UPEC 4397/ERL CNRS 9215,

Universite Paris Est Creteil,

61, Av. du General de Gaulle,

F-94000, Creteil, France

E-mail: papy@u-pec.fr

Correspondence may also be addressed to: Nadia Soussi-Yanicostas, INSERM U 1141, Hopital Robert Debre, 75019 Paris, France,

E-mail: nadia.soussi@inserm.fr

Keywords: HS3ST2/3OST2; heparan sulphates; Alzheimer’s disease; tauopathy; 3-O-sulphation; Zebrafish

Abbreviations: dpf = days post fertilization; FRET = fluorescence resonance energy transfer; HSPG = heparan sulphate proteogly-can; PKA = protein kinase A; PP2A = protein phosphatase 2A

IntroductionHeparan sulphate (glucosamine) 3-O-sulphotransferase 2

(HS3ST2, also known as 3OST2) is an enzyme predomin-

antly expressed in brain (Shworak et al., 1999; Yabe et al.,

2005), where it generates rare 3-O-sulphated domains of

still unrevealed physiological roles in heparan sulphates

(Lawrence et al., 2007; Mochizuki et al., 2008; Thacker

et al., 2014). During adulthood, HS3ST2 is expressed in

brainstem, trigeminal ganglia (Lawrence et al., 2007), cere-

bral cortex (Mochizuki et al., 2008) and hippocampus

(Huynh et al., 2012), brain regions vulnerable to tau path-

ology in Alzheimer’s disease (Wenk, 2003; Braak and Del

Tredici, 2012). In contrast, it is almost absent in cerebellum

(Lawrence et al., 2007; Mochizuki et al., 2008; Thacker

et al., 2014), a brain region of low disease vulnerability

(Wu et al., 2005). Under physiological conditions heparan

sulphates are typically located at the cell surface (Kreuger

and Kjellen, 2012), whereas in Alzheimer’s disease they

accumulate at the intracellular level long before the detec-

tion of the tau pathology (Snow et al., 1990) and co-

localize with neurofibrillary tangles when the pathology is

detected (Snow et al., 1990; Goedert et al., 1996;

Hernandez et al., 2002). This suggests the implication of

heparan sulphates in the pathophysiological mechanisms

leading to tauopathy. The abnormal hyperphosphorylation

of the microtubule-associated protein tau is a critical event

characterizing both sporadic and inherited tauopathies

(Iqbal et al. 2010; Wang et al., 2013), including

Alzheimer’s disease, Pick’s disease, progressive supranuclear

palsy, corticobasal degeneration, frontotemporal dementia

with parkinsonism linked to chromosome-17 (FTDP-17)

and related disorders (Williams, 2006; Spillantini and

Goedert, 2013). Interestingly, the abnormal phosphoryl-

ation of tau in the disease brain is assumed to be mediated

by the same kinases that phosphorylate tau in normal brain

(Hashiguchi and Hashiguchi, 2013; Wang et al., 2013);

however, in vitro, these kinases can only induce the abnor-

mal phosphorylation of tau at disease-specific sites only if

the enzymatic reaction takes place in the presence of poly-

anions such as heparin (Hasegawa et al., 1997; Zheng-

Fischhofer et al., 1998; Paudel and Li, 1999; Sibille et al.,

2006). Because heparin is a heparan sulphate analogue

carrying high levels of 3-O-sulphation in the sugar back-

bone (Kreuger and Kjellen, 2012; Shriver et al., 2012), this

raises the question whether 3-O-sulphated heparan sul-

phates, products of neuronal 3-O-sulphotransferases, are

involved in the molecular and cellular mechanisms leading

to the abnormal phosphorylation of tau. Here, we used

biochemical, cellular, and animal strategies to investigate

the implication of HS3ST2 in the molecular and cellular

mechanisms leading to the abnormal phosphorylation of

both wild-type tau and of tau carrying the mutation

P301L (hTauP301L) responsible of FTDP-17 (Alonso Adel

et al., 2004; Williams, 2006). We show that the chemical

inhibition of heparan sulphate sulphation, or the down-

regulation of HS3ST2 expression in cells, strongly avoids

the hyperphosphorylation of tau induced by oxidative

stress or by the hTauP301L mutation, and that inhibiting

the expression of hs3st2 in a FTDP-17 zebrafish model of

tauopathy (Paquet et al., 2009) avoids the abnormal phos-

phorylation of tau at several Alzheimer’s disease-related

epitopes resulting in tauopathy arrest and animal func-

tional recovery.

Materials and methods

Human tissues

Human hippocampus samples were obtained from the Agingand Neurodegenerative Diseases Brain Bank InvestigationLaboratory of the Universidade Federal de Sao Paulo, Brazil.Protocols were approved by the local ethics committee(No.285/04). Two experimental groups were included in the

2 | BRAIN 2015: Page 2 of 16 J. E. Sepulveda-Diaz et al.

by guest on April 5, 2015

Dow

nloaded from

study (Supplementary Table 1), a control group (n = 8) and anAlzheimer’s disease group (n = 8) with subjects’ ages rangingfrom 60 to 98 years old with a mean age of 73.4 � 10.7 years.Post-mortem intervals averaged 14 h 59 min � 5 h 41 min.Neuropathological changes in brains were investigated usingBraak and Braak and Consortium to Establish a Registry forAlzheimer’s Disease (CERAD) guidelines. Senile plaques andneurofibrillary tangles were determined on Bielschowsky-stained sections of middle frontal gyrus, middle temporalgyrus, inferior parietal lobule, occipital pole, hippocampalCA1 and enthorinal cortex. Senile plaques were countedusing a �10 objective and neurofibrillary tangles were countedwith a �20 objective. An arithmetic mean (mean � standarderror of the mean) was calculated from the counts of six fieldsfor senile plaques by mm2 and neurofibrillary tangles by mm2

for each region. Neuropathological diagnosis was then madeusing the guidelines proposed by Braak and Braak criteria(Supplementary Table 1). Brain samples were powdered inliquid nitrogen and stored at �80�C until use.

Quantitative polymerase chainreaction

Total RNA was extracted from frozen hippocampus CA1 sam-ples as described (Huynh et al., 2012). The amount of RNA inthe extracts was measured spectrophotometrically at 260 and280 nm. The quality of RNA was confirmed suitable deter-mined by gel electrophoresis. RNA quality was additionallyconfirmed by RIN (RNA integrity number) determination(Schroeder et al., 2006) obtained by the ‘Standardization ofRNA Quality Control Protocol’ in a 2100 Bioanalyzer(Agilent). RIN average was 47.0 for all samples. ExtractedRNA was used to synthesize cDNA using reverse transcriptaseas previously reported (Huynh et al., 2012). Genes of interestwere analysed in template cDNA by quantitative real-timePCR using primers (Eurofins) designed by the Primer3outputprogram. PCR was performed according to the LightCycler�

FastStart DNA Master SYBR� Green kit manufacturer’s in-structions (Roche). Relative gene expression quantificationwas performed using the comparative ��CT method (Schefeet al., 2006). Two reference genes (TUBA1A and TBP) wereused as endogenous controls. Normalization of gene expres-sion was accomplished with the Genorm program(Vandesompele et al., 2002).

Cells models of tauopathy

SH-SY5Y and SH-SY5Y/hTauP301L (SH/hTauP301L) cells (a giftfrom J. Gotz, Zurich, Switzerland) were maintained and cul-tured as previously described (Ferrari et al., 2003). Briefly, SH-SY5Y cells were grown in DMEM-GlutaMAXTM (Invitrogen)supplemented with 10% foetal bovine serum (Invitrogen) and1% penicillin/streptomycin (Invitrogen). For SH-SY5Y/hTauP301L cells, 300mg/ml geneticin (Invitrogen) and pyruvatewere added to the medium. For all experiments, cells weredifferentiated 24 h after plating by treatment with 10 mM all-trans-retinoic acid (Sigma) in complete growth medium for 7days. Media were changed every 2 days. For oxidative stressstudies, differentiated SH-SY5Y cells were treated with H2O2

(500mM final concentration in medium) for 30 min as previ-ously described (Zhang et al., 2013). In assays involving

NaClO3 (Keller et al., 2008), the salt was added to cell cul-tures (25, 50 or 75 mM final concentration) at the third or fifthday of differentiation, as indicated, and maintained until theend of the experiment. For enzymatic treatments, differentiatedcells were treated either with a heparitinase mix containingheparitinase I (2 U/ml), heparitinase II (0.2 U/ml), and hepar-itinase III (0.2 U/ml) (all from Sigma), or with chondroitinaseABC (0.1 U/ml) (Sigma) and incubated at 37�C for 1.5 or 6 h(as indicated). After enzymatic treatment, medium was chan-ged for new fresh medium and cells were subjected to thespecified corresponding experimental conditions.

Cell transfections, plasmids andlentivirus

Two lentiviral vectors pTrip-CMV-Egfp-miRNA-HS3ST2 de-signed to inhibit HS3ST2 expression were purchased fromthe vectorology platform of the Brain and SpinalCord Institute (Paris, France). Briefly, four plasmids containingdifferent miRNAs targeting the HS3ST2 gene coding sequencewere designed with the following sequences: dsOligo-miRNA1:50-TGCTGCCTTCTTCACGCCCACAATGAGTTTTGGCCACTGACTGACTCATTGTGCGTCAAGAAAG-30, 30-CGGAAGAAGTGCGGGTGTTACTCAAAACCGGTGACTGACTGAGTAACACGCAGTTCTTTCGTCC-50; dsOligo-miRNA2:50-TGCTGGACTCGCCCCATCTCGCCGGCGTTTTGGCCACTGACTGACGCTGGTGATGGGTCGTATT-30, 30-CCTGAGCGGGGTAGAGCGGCCGCAAAACCGGTGACTGACTGCGACCACTACCCAGCATAAGTCC-50. A sequence of theGFP was also introduced in the vector for monitoring simul-taneous expression. The lentiviral vector pTrip-CMV-eGFP-miRneg was used as control. For silencing experiments, SH/hTauP301L cells (300 000) were suspended and transduced witha 1:20 dilution of a concentrated suspension of the two lenti-viral vectors mixture containing 0.1 � 109 physical particles/ml.Cells were then plated in 24-well plates (Nunc) and incubatedfor 72 h under standard conditions. Cell lysates were immuno-blotted as specified above with antibodies directed to HS3ST2,pSer396, Tau-5, and anti-b-actin (Life Technologies).

Immunofluorescence

For immunofluorescence studies, cells were plated in 35-mmPetri dishes (m-Dish, Ibidi) at a density of 40 000 cells/dishfollowing manufacturer’s instructions. Differentiation, stressand any other treatments were performed as described above.For analysis, cells were washed, fixed in 100% methanol at�20�C and washed again before been incubated with 2%donkey serum in phosphate-buffered saline (PBS) for 20 minat room temperature. Cells were then incubated for 1 h withthe corresponding primary antibody at room temperature,washed and incubated with the appropriate secondary anti-body for 30 min at room temperature. The antibodies usedwere HS4C3 (Tem Dam et al., 2006) (TH v Kuppevelt isauthor of the work; 1:100), tau K9JA (Dako; 1:500) or tauV-20 (Santa Cruz; 1:100). Secondary antibodies were Fluo488donkey anti-rabbit (Interchim; 1:200), Fluo488 donkey anti-goat (Interchim; 1:200), and Cy3 goat anti-mouse (Sigma;1:200). The phage display HS4C3 antibody was revealed byan anti-VSV antibody (Sigma; 1:100). Stack images were ob-tained with the software CellSens from a spinning disk

HS3ST2 is critical for tau pathology BRAIN 2015: Page 3 of 16 | 3

by guest on April 5, 2015

Dow

nloaded from

inverted confocal microscope (IX81 DSU Olympus, �60 N.A.1.35) coupled to an Orca Hamamatsu RCCD camera. Imageswere processed with the ImageJ software (W. Rasband,National Institute of Health).

Western blotting

For protein analysis by western blot, powdered frozen tissue orcells were suspended (100 mg of tissue/ml or 1.8 � 105 cells/ml) in an extraction buffer (50 mM Tris-HCl, 150 mM NaCl,1% TritonTM X-100, 10 mM NaF, 1 mM Na3VO4, pH 8.0)supplemented with protease and phosphatase inhibitors(Pierce). Tissue homogenates were centrifuged at 12 000 g for15 min and the protein content in the supernatants was deter-mined using BCA Protein Assay kit (Pierce). Samples contain-ing 20 mg proteins were subjected to SDS-PAGE in 10%acrylamide gel. Primary antibodies for phosphorylated tauwere pSer199, pSer199/pSer202, pThr231, pSer262, pSer396(Life Technologies), AT8, AT270, PHF1, AT180, AT100(Pierce, Thermo Scientific), and MC1 (a gift from PeterDavies); anti-total tau antibody was Tau-5 (Millipore). Otherantibodies used were against HS3ST2, HS3ST4 (Clinisciences),p38, pp38, NF-�B p65, pNF-�B p65 (Cell Signaling technol-ogy), and anti-b-actin (Life Technologies). Corresponding sec-ondary antibodies diluted in phosphate-buffered salinecontaining 1% milk were incubated for 1 h at room tempera-ture. Blots were developed either with Immobilon WesternChemiluminiscent HRP Substrate, Luminata Forte, LuminataCrescendo (Millipore), or SuperSignal West Dura (ThermoScientific) following manufacturer’s instructions.Densitometric quantification of immunoreactivity was per-formed by using the ImageJ software. Hyperphosphorylatedtau signals were normalized to total tau detected with theanti-Tau-5 antibody. b-Actin was used to standardize totalprotein load.

Fluorescence resonance energytransfer assay

Fluorescence resonance energy transfer (FRET) assays wereperformed as in Li et al. (2012) by using a cryptate-d2system (CISBIO Bioassays). Briefly, cell lysates (10 ml at2.6 mg/ml of protein) were mixed with HS4C3 antibody in384-well plates and incubated for 20 min at room temperature.Next, cryptate-conjugated anti-VSV and d2-conjugated anti-tau K9JA were added to the reaction mixture following themanufacturer’s instructions. After 20 min of incubation, energytransfer was measured using a TECAN Infinite 1000 spec-trometer, data were processed according to manufacturer’sinstructions.

Tau and tauP301L phosphorylationassay

Tau phosphorylation reactions were carried out according toHasegawa et al. (1997) with some modifications. Briefly, fourdifferent solutions containing tau at 200mg/ml, GSK3B(Promega) at 20 mg/ml, respectively, heparin (Sigma) at 60 mg/ml, and ATP at 250mM, were prepared in a reaction buffer(40 mM Tris-HCl, 20 mM MgCl2, 0.1 mg/ml BSA, pH 7.5).The four solutions were mixed, and when necessary diluted

in reaction buffer, to set reaction mixtures containing 1 mg ofthe tau substrate, 100 ng of the enzyme, and 50 mM ATP, inthe presence or absence of 0.3 mg heparin (20 ml final reactionvolume). When indicated, protein kinase A (PKA) (Sigma) wasincluded in the reaction mixture at 3 mg/ml final concentration.Tau was either full-length recombinant human tau (Millipore)or full-length recombinant tauP301L (Interchim). The amount oftau, GSK3B and PKA used in the reactions was selected fromdose response experiments with 0.5, 1.0, or 2.0mg of tau and50, 100, or 200 ng of enzyme. Reactions were carried out at30�C for 3 h and stopped by cooling at �20�C. Reaction mix-tures were resolved by western blot as described above, withone or several of the following antibodies: pSer396, pSer199/202, AT270, PHF1, AT8, AT100, and MC1.

Kinase and phosphatase activities

The effect of heparin on the kinase and phosphatase enzymaticactivities was assessed with kits for GSK3B (Promega), proteinphosphatase 2A (PP2A, Promega), and PA (Abcam) by follow-ing the manufacturer’s instructions. The enzymatic reactionswere performed in the absence or presence of heparin at sev-eral concentrations (from 0.1 ng/ml to 100mg/ml).

Enzyme-linked immunosorbentassays

Binding of tau, tauP301L, GSK3B, PKA, or PP2A to heparinwas evaluated by ELISA as in Huynh et al. (2012). Briefly,ELISA-type 96-well plates were coated with a 2mg/ml bovineserum albumin (BSA) heparin conjugate solution. After wash-ing the wells with PBS containing 0.05% Tween-20, wells weresaturated with 3% BSA in PBS. Then, the tested protein (fulllength human tau, tauP301L, GSK3B, PKA or PP2A) was addedto the plate at various concentrations and plates were incu-bated for 12 h at 4�C. After washing, the protein bond toheparin was revealed by the corresponding specific antibodyTau-5, anti-GSK3B (Thermofisher), anti-PKA, or anti-PP2A(Sigma) followed by a peroxidase-labelled secondary antibody.Peroxidase activity was measured by the TMB detection kit(Pierce) following fabricant instructions. For the ELISA com-petition assays, heparin or heparan sulphate were added to thewells at concentrations ranging from 0.01 ng/ml to 10 mg/mlfollowing addition of the corresponding protein (tau,tauP301L, GSK3B, PKA, or PP2A) to the wells.

Zebrafish

Zebrafish were maintained in a standard zebrafish facility(Aquatic Habitats, Apopka). Developmental stages were deter-mined as hours post fertilization (hpf) or as days post fertil-ization (dpf), as previously described (Kimmel et al., 1995).Wild-type embryos were from the AB and TL strains. TheTg[HuC::hTauP301L;DsRed] transgenic line has been previouslydescribed (Paquet et al., 2009). All procedures involvinganimal handling complied with the guidelines of the FrenchAnimal Ethics Committee and were approved by the samecommittee under the ethics statement N�: 2012-15/676-0069.

4 | BRAIN 2015: Page 4 of 16 J. E. Sepulveda-Diaz et al.

by guest on April 5, 2015

Dow

nloaded from

Morpholino oligonucleotides

Morpholinos MO-HS3ST2AUG: 50-GGCTTGACAGGAACCTATATGCCAT-30, MO-HS3ST2SPL: 50-AAAATTAACCTTACCTGTACCAGTC-30 and mmMO-HS3ST2AUG: 50-GGCTTGACATGAAGATATATGGTAT-30, were obtained from GeneTools. For morpholino-mediated zebrafish hs3st2 transcript in-activation, 2 nl of a 1.5 mM solution, corresponding to 3 pmolof each morpholino were injected in 1- to 2-cell stage embryosusing standard protocols. Antibodies directed against patho-logic tau epitopes were AT180, MC1 and AT270. AT8 andPHF1 for zebrafish experiments were kindly provided byD. Paquet. Anti-human tau antibody was purchased fromDakoCytomation. Primary motor neuron morphology wasanalysed using the mouse monoclonal anti-synaptotagmin 1antibody (anti-Syt1; 1/300, Hybridoma Bank).

Analysis of zebrafish behaviour

Touch-induced escape response was assessed by gently touch-ing the tail of embryos with the tip of a fine plastic rod (Paquetet al., 2009). Embryos were then classified as responders ornon-responders.

Statistical analyses

All data are expressed as the mean � standard error of themean (SEM) or standard deviation (SD) as specified. P-valueswere generated using non-parametric t-test, Mann Whitney U-test, or ANOVA, as specified. Asterisks indicate significant dif-ferences (*P50.05, **P5 0.005, ***P50.001); n.s. indi-cates non-significant differences.

Results

HS3ST2 is overexpressed inAlzheimer’s disease brain

In a previous work, we showed that HS3ST1, HS3ST2,

HS3ST3A, HS3ST3B, HS3ST4, and HS3ST5 are expressed

in human elderly hippocampus (Huynh et al., 2012).

However, it is unknown whether the expression of these

enzymes is altered in brain of Alzheimer’s disease. To first

investigate this possibility, we compared the transcript

levels of all human sulphotransferases (Kreuger and

Kjellen, 2012) from Alzheimer’s disease hippocampus

(CA1) to those from age-matched controls. We used quan-

titative PCR to determine levels of mRNA encoding for the

four heparan sulphate N-deacetylase/N-sulphotransferases

(NDST1, NDST2, NDST3 and NDST4) and for the differ-

ent sulphotransferases including HS2ST, HS3STs (1, 2, 3A,

3B, 4, 5, and 6), and HS6STs (1, 2 mRNA variants 2L and

2S, and 3), as well as C-5 epimerase (GLCE) and HPSE

(heparanase), two other enzymes implicated in heparan sul-

phate metabolism. Among the 18 analysed transcripts, ex-

pression levels of most sulphotransferases were slightly

(P50.05) increased in the Alzheimer’s brains compared

to age-matched-individuals. However, only HS3ST2 and

HS3ST4 were expressed at higher significant levels

(P5 0.001), as analysed by the Mann Whitney U-test

(Table 1). Remarkably, western blot analysis of the hippo-

campal protein extracts showed a higher HS3ST2 expres-

sion, compared to HS3ST4, in the Alzheimer’s tissue

(Supplementary Fig. 1). This suggests that in Alzheimer’s

disease, 3-O-sulphated domains in heparan sulphates are

increased in brain regions in where these enzymes are ex-

pressed (Lawrence et al., 2007; Mochizuki et al., 2008).

3-O-sulphated heparan sulphatesinteract with tau in cell models oftauopathy

In Alzheimer’s disease, heparan sulphates accumulate in the

intracellular space of neurons before the apparition of the

tau pathology (Snow et al., 1990) and strongly co-stain

with tau in intracellular neurofibrillary tangles during dis-

ease (Snow et al., 1990; Goedert et al., 1996; Hernandez

et al., 2002). However, under physiological conditions,

these sulphated polysaccharides are located at the extracel-

lular space, in where they exert their known biological

functions (Kreuger and Kjellen, 2012; Thacker et al.,

2014). Thus, we first thought to investigate whether 3-O-

sulphated heparan sulphate internalization and interaction

with tau can be provoked under tauopathy-associated

Table 1 Expression of heparan sulphate metabolic

enzyme mRNA in hippocampus from control and

Alzheimer’s disease

Enzymes Isoforms Control group AD group

(AU � SD)

NDST1 0.33 � 0.06 0.38 � 0.06

NDST2 0.59 � 0.38 1.38 � 0.42*

NDST3 n.d. n.d.

NDST4 n.d. n.d.

HS2ST 0.89 � 0.07 1.03 � 0.13*

HS3ST1 0.75 � 0.09 0.96 � 0.20

HS3ST2 0.37 � 0.07 2.06 � 0.39***

HS3ST3A1 0.35 � 0.08 2.17 � 1.69*

HS3ST3B1 0.58 � 0.21 1.76 � 0.90*

HS3ST4 1.33 � 0.35 5.04 � 0.34***

HS3ST5 0.10 � 0.04 0.08 � 0.02

HS3ST6 n.d. n.d.

HS6ST1 0.45 � 0.57 1.35 � 0.77*

HS6ST2(var L) 0.35 � 0.30 1.06 � 0.55*

HS6ST2(var S) 0.47 � 0.48 0.64 � 0.22

HS6ST3 2.41 � 0.77 2.96 � 0.82

GLCE 0.35 � 0.21 1.13 � 0.62*

HPSE 0.23 � 0.03 0.38 � 0.06*

GLUL 0.49 � 0.08 1.23 � 0.18**

CXCR4 0.73 � 0.26 0.68 � 0.14

*Significant changes in transcript expression as defined by the Mann Whitney U-test

(n = 8).

n.d. = the transcript was not detected.

AU � SD = arbitrary units � SD.

HS3ST2 is critical for tau pathology BRAIN 2015: Page 5 of 16 | 5

by guest on April 5, 2015

Dow

nloaded from

conditions, such as oxidative stress or expression of a

pathologic tau mutation. We used two cell models: (i) an

oxidative stress (SH/ox) model in wild-type SH-SY5Y cells

(Zhang et al., 2013); and (ii) a tau-mutation-dependent

model in SH-SY5Y cells overexpressing the hTauP301L

mutant (SH/hTauP301L) (Ferrari et al., 2003), which under-

lies FTDP-17 (Alonso Adel et al., 2004). By using a 3-O-

sulphated heparan sulphate specific antibody (HS4C3) (Ten

Dam et al., 2006), we confirmed by confocal microscopy

that the sulphated polysaccharide accumulates at the cell

surface in wild-type and control cells (Fig. 1A and C, and

Supplementary Fig. 2), whereas a strong intracellular

immunostaining was observed in both SH/ox (Fig. 1B and

Supplementary Fig. 3) and SH/hTauP301L cells (Fig. 1D and

Supplementary Fig. 4A). In SH/ox cells we observed a

strong 3-O-sulphated heparan sulphate staining near the

cell membrane at 10 min after the stress pulse

(Supplementary Fig. 3A), whereas at 6 h after stress the

membrane staining was strongly decreased and highly pro-

nounced at the intracellular space (Supplementary Fig. 3B),

suggesting a stress-driven time course internalization of the

membrane-associated polysaccharides. In contrast, in

SH/hTauP301L cells, the 3-O-sulphated polysaccharide was

permanently and highly accumulated in the form of intra-

cellular inclusions showing different levels of overlap with

tau (Supplementary Fig. 4A), but nearly absent at the cell

membrane. No differences in the level or distribution of

chondroitin sulphates were observed in either of the cell

models (not shown).

The accumulation of intracellular 3-O-sulphated heparan

sulphates in both SH/ox and SH/hTauP301L cells raised the

question whether these sulphated polysaccharides were of

membrane origin. To address this issue, we cultured SH/ox

and SH/hTauP301L cells in the presence of a mixture of

heparitinase I, II and III, an enzymatic cocktail that digests

the heparan sulphate glycanic chains of membrane-bound

heparan sulphate proteoglycan (HSPG). As expected, the

heparitinases treatment strongly decreased the staining of

membrane-associated heparan sulphate in control cells

(Fig. 2A), and, surprisingly, nearly suppressed the staining

of the intracellular heparan sulphate inclusions in the two

cell models (Fig. 2B–D and Supplementary Fig. 4B), suggest-

ing that the intracellular 3-O-sulphated heparan sulphates

observed in both SH/ox and SH/TauP301L cells existed as

HSPG in cell membranes before being internalized within

the cells. As confocal microscopy co-localization cannot be

Figure 1 Intracellular 3-O-sulphated heparan sulphates and tau co-staining in two cell models of tauopathy. (A) Immunostaining

of cell membrane-associated 3-O-sulphated heparan sulphates (red) and intracellular tau (green) in control cells (optical section). (B) 3-O-

sulphated heparan sulphate immunostaining increases in cell membranes early after stress to then decrease in a time dependent manner with

concomitant increase of intracellular staining. (C) Empty vector (EV) transfected cells. (D) Intracellular 3-O-sulphated heparan sulphate inclusions

accumulate and co-stain with tau at different extents in SH/hTauP301L cells. Pictures represent observations on cells from three independent

experiments. Scale bars = 10mm.

6 | BRAIN 2015: Page 6 of 16 J. E. Sepulveda-Diaz et al.

by guest on April 5, 2015

Dow

nloaded from

considered as an evidence of molecular interactions, we fur-

ther investigated whether the 3-O-sulphated heparan sul-

phates interact with tau at the molecular level by

performing FRET experiments (Li et al., 2012) in SH/ox

cells. We incubated cell lysates with a europium-labelled sec-

ondary antibody directed against the HS4C3 antibody and a

d2-labelled anti-tau primary antibody. The FRET signal

observed in the stressed cells as early as 10 min and at 6 h

after the stress pulse was nearly abolished when cells were

cultured in the presence of heparitinase (Fig. 2E), but not of

chondroitinase (not shown), indicating a close interaction of

3-O-sulphated heparan sulphates with tau.

To investigate the importance of 3-O-sulphates in the

interaction of heparan sulphates with tau, we performed

an ELISA-based competition essay to compare the capacity

of tau to bind to highly 3-O-sulphated heparan sulphates,

here represented by heparin, compared to less sulphated

heparan sulphates (Shriver et al., 2012). As expected, we

Figure 2 3-O-sulphated heparan sulphates interact with tau under pathological conditions. Heparitinase (Hep-ase) treatment

decreases membrane-associated and intracellular 3-O-sulphated heparan sulphates (red) staining and co-localization with tau. Z-projections are

depicted for (A) control cells; (B) SH/ox cells treated with heparitinase before the oxidative stress pulse; and (C) SH/hTauP301L cells.

(D) Quantification of heparan sulphates (HS) intracellular inclusions in SH/hTauP301L cells treated with hep-ase during 1.5 or 6 h. (E) FRET-based

measurement of the molecular interaction between 3-O-sulphated heparan sulphates and tau in SH/ox cell lysates at different time points after the

stress pulse. Calculated delta F (�F) shows a time-dependent molecular interaction abolished when cells were cultured in the presence of

heparitinase I, II, III cocktail (Hep-ase). (F) ELISA competition assay showing the higher capacity of heparin, used as a prototype of 3-O-sulphate

heparan sulphates, to inhibit the binding of tau to immobilized heparin compared to heparan sulphates. Data represents the mean of three

different experiments performed in triplicate each time; error bars show SEM.

HS3ST2 is critical for tau pathology BRAIN 2015: Page 7 of 16 | 7

by guest on April 5, 2015

Dow

nloaded from

observed that tau binds to heparin better than to

heparan sulphates (Fig. 2F), suggesting that increasing

3-O-sulphation levels in heparan sulphate could result in

heparan sulphate increased capacity to interact with tau.

Together, these results suggest that under pathological con-

ditions, 3-O-sulphated heparan sulphates can interact with

tau inside cells in where they could participate to the cel-

lular mechanisms inducing the pathological phosphoryl-

ation of the protein.

Pathological phosphorylation of taurequires 3-O-sulphated heparansulphates

In an attempt to understand the role of the 3-O-sulphated

heparan sulphates in the abnormal hyperphosphorylation

of tau, we sought to investigate whether these polysacchar-

ides are required for tau hyperphosphorylation in both

SH/ox and SH/hTauP301L cells. As western blot analysis

using the Tau-5 antibody showed a higher level of total

tau in the SH/hTauP301L cells compared to SH/ox cells

(Supplementary Fig. 5A), all phosphorylated tau levels

were further normalized against total tau (Tau-5). First,

we confirmed that, compared to control cells, SH/ox and

SH/hTauP301L cells accumulate higher levels of tau phos-

phorylated at Ser199 (pSer199), Thr231 (pThr231),

Ser262 (pSer262), and Ser396 (pSer396), characteristic

hyperphosphorylated residues in Alzheimer’s disease

(Steinhilb et al., 2007; Duka et al., 2013). In the SH/ox

cells, maximum pSer199, pThr231, and pSer262 levels

were observed 10 min after the stress pulse, while max-

imum pSer396 was detected 6 h later (Supplementary Fig.

5B). To investigate whether heparan sulphates are required

for tau hyperphosphorylation at these sites, we treated cells

with sodium chlorate (NaClO3), an inhibitor of the synthe-

sis of PAPS (30-phosphoadenosine 50-phosphosulphate), the

sulphate donor in glycosaminoglycan biosynthesis

(Venkatachalam, 2003). High NaClO3 concentrations

have shown to inhibit the formation of highly sulphated

heparan sulphates under classic cell culture conditions

(Safaiyan et al., 1999; Keller et al., 2008). Here, following

treatment with NaClO3, we observed a significant decrease

in hyperphosphorylated tau accumulation, notably in SH/

ox cells (Fig. 3A). Although this decrease was less promin-

ently in SH/hTauP301L cells (Fig. 3B), which required higher

NaClO3 concentrations, this suggests the involvement of

sulphated glycosaminoglycans in the hyperphosphorylation

event. To determine which glycosaminoglycan species are

involved in this process, we incubated SH/ox and

SH/TauP301L cells with either chondroitinase ABC or hepar-

itinase. While heparitinase digestion significantly decreased

the accumulation of hyperphosphorylated tau in both SH/

ox and SH/TauP301L cells (Fig. 3C–E), chondroitinase treat-

ment showed no effect (Fig. 3C and D), indicating that

highly sulphated heparan sulphates, but not chondroitin

sulphates, are required for tau hyperphosphorylation in

both cell models. To investigate whether inhibiting the cel-

lular heparan sulphates can affect the phosphorylation of

other cytosolic proteins, we evaluated the phosphorylation

of the subunit p65 of NF-�B and of p38 in both models.

No differences in the phosphorylation of these proteins

were observed (Fig. 3F). This result, as well as the 3-O-

sulphated heparan sulphates interaction with tau observed

by FRET, suggests a specific effect of 3-O-sulphated

heparan sulphates on tau. However, this effect could also

be driven by altered kinases or phosphatases activities

through indirect mechanisms that should be further studied.

Thus, by using an ELISA assay, we compared the capacities

of tau, tauP301L, GSK3B, PKA and PP2A, to bind to immo-

bilized heparin as a prototype of 3-O-sulphated heparan

sulphates. Our results show that tau and tauP301L, but

not GSK3B, PKA, and PP2A, could efficiently bind to the

sulphated polysaccharide (Fig. 3G).

Together, our results suggest that highly sulphated

heparan sulphates favour the abnormal phosphorylation

of tau in cells in where 3-O-sulphated heparan sulphates

interact with tau. To examine if 3-O-sulphated heparan

sulphates directly participate to the abnormal phosphoryl-

ation of wild-type tau, we used an in vitro GSK3B-depend-

ent tau phosphorylation reaction and examined the

occurrence of phosphorylation at residues Thr181

(AT270 antibody), Ser396/Ser404 (PHF1 antibody),

Ser202/Thr205 (AT8 antibody), Ser212/Ser214 (AT100),

and the conformational anti-tau MC1 antibody, specific

of Alzheimer’s disease and other tauopathies (Cripps

et al., 2006; Yoshida and Goedert 2006; Steinhilb et al.,

2007; Duka et al., 2013). We observed that the GSK3B-

dependent phosphorylation at AT270, PHF1, and AT8

epitopes requires the presence of heparin in the reaction

mixture (Fig. 4A). Interestingly, the effect of heparin at

the AT100 epitope required the introduction of PKA into

the GSK3B reaction mixture. Under these conditions, we

observed that the abnormal phosphorylation of tau at the

AT100 epitope is also heparin dependent (Fig. 4B).

However, when we used the MC1 antibody, no differences

were observed in western blots of reactions performed in

the presence or absence of heparin with and without add-

ition of PKA (not shown). This can be possibly due to a

blotting condition-dependent loss of the tau conformation

required for recognition of tau by this antibody, most

classically used for histological applications. We then ana-

lysed the effect of heparin in the GSK3B-dependent phos-

phorylation of tau at sites Ser396 and Ser199/202, found

phosphorylated at some extent in normal brain but hyper-

phosphorylated in case of tauopathy. We observed an

increased level of phosphorylation at these sites in reactions

carried out in the presence of heparin (Fig. 4C). We then

carried out the phosphorylation reaction using tauP301L in-

stead of the wild-type tau as substrate of the phosphoryl-

ation reaction. Heparin favoured the tauP301L abnormal

phosphorylation (Fig. 4D), in agreement with the role of

heparan sulphates in the abnormal phosphorylation of tau

in the HS/hTauP301L model of tauopathy. Interestingly,

8 | BRAIN 2015: Page 8 of 16 J. E. Sepulveda-Diaz et al.

by guest on April 5, 2015

Dow

nloaded from

Figure 3 Sulphated heparan sulphates are required for tau hyperphosphorylation in two cell models of tauopathy. Tau hyper-

phosphorylated at Ser199, Thr231, Ser262, and Ser396, total tau (Tau-5), and b-actin levels in cells treated with NaClO3 in (A) SH/ox cells, and in

(B) SH/hTauP301L cells. (C) Hyperphosphorylated tau at Ser199 and Ser396 in SH/hTauP301L cells treated or not with heparitinases I, II, and III

cocktail (Hep-ase) or chondroitinase ABC (Cse). (D) Tau hyperphosphorylated at Ser199 in SH/ox cells after a stress pulse treated or not with

heparitinase, Cse or NaClO3. (E) Tau hyperphosphorylated at Ser396 in SH/ox cells treated or not with heparitinase prior to the oxidative stress

pulse. Data represent the mean of three different experiments performed in duplicate each time; error bars show SEM. (F) Western blots

showing the phosphorylation of NF-kB p65 and p38 proteins in SH/ox (left) and in SH/hTauP301L (right) cells treated or not with the heparitinases

cocktail or with NaClO3. Total levels of each protein, as well as that of b-actin, were used as loading controls. (G) ELISA assay showing the binding

of heparin to tau (filled circles, solid line), TauP301L (open circles, dashed line), GSK3B (upward triangles, spotted line), PP2A (downward triangles,

spotted line), and PKA (diamonds, solid line). Data in histograms were made from the average signal obtained from the densitometric analysis of

duplicates carried in each individual western blot, or by the average of two different western blots when duplicates were not performed.

HS3ST2 is critical for tau pathology BRAIN 2015: Page 9 of 16 | 9

by guest on April 5, 2015

Dow

nloaded from

phosphorylation reaction showed to be more efficient for

tauP301L than for wild-type tau, suggesting a higher suscep-

tibility to tau pathology when the mutation is present.

Our results suggest that implication of heparan sulphates

in the abnormal phosphorylation reaction is mediated by

their direct interaction with tau, and less probably by their

direct interaction with kinases or phosphatases (Fig. 3F and

G). To reinforce this hypothesis, we investigated if heparin

can modulate the enzymatic activity of GSK3B, PP2A, and

AP in vitro. By using commercial kits we measured the

activity of these enzymes and found that the addition of

heparin at various concentrations has no modulatory effect

in any of the tested enzymes (not shown), weakening the

assumption of a putative role of the sulphated polysacchar-

ides as kinase activators or phosphatase inhibitors.

Nevertheless, it remains to be explored if the activities of

kinases and phosphatases are altered in cells by mechan-

isms independent of their direct interaction with the

sulphated sugars.

Together, the in vitro and the intracellular interaction of

tau with heparin or heparan sulphates, as well as the effect

of these interactions on the abnormal phosphorylation of

tau, support the idea that intracellular 3-O-sulphated

heparan sulphates could display a tau-chaperone activity

critical for the abnormal phosphorylation of tau observed

under pathological conditions. However, the implication of

the specific 3-O-sulphation of heparan sulphates remains

unclear. To explore this possibility, we investigated the

effect of specifically silencing HS3ST2 expression in patho-

logical tau phosphorylation in SH/hTauP301L cells. We used

lentiviral vectors carrying miRNA designed to suppress

HS3ST2 expression. Cells were transfected and monitored

for pSer396 as a marker of tau hyperphosphorylation.

Inhibition of HS3ST2 expression resulted in a strong de-

crease of pSer396 accumulation (Fig. 4E), supporting the

implication of 3-O-sulphated heparan sulphates in the tau

phosphorylation event.

HS3ST2 is critical for the abnormalphosphorylation of tau in vivo

We then investigated whether HS3ST2 is critical for abnor-

mal phosphorylation of tau in vivo. To address this issue,

we used the Tg[HuC::hTauP301L;DsRed] zebrafish trans-

genic line (Paquet et al., 2009) and morpholino-mediated

inactivation of the zebrafish hs3st2. Tg[HuC::

hTauP301L;DsRed] transgenic embryos express the human

mutant TauP301L protein under the control of the pan-neur-

onal HuC promoter and reproduce several characteristic

features of tauopathies, including an increased accumula-

tion of hyperphosphorylated tau in brain and spinal cord

neurons, neuronal abnormalities, neuronal death and im-

paired motility (Paquet et al., 2009). As expected, we de-

tected high levels of abnormally phosphorylated tau in a

large number of neurons scattered in several brain regions

of 5 dpf Tg[HuC::hTauP301L;DsRed] larvae (n = 100)

(Supplementary Fig. 6). By BLAST analysis of the Zv9

version of the zebrafish genome sequence, we identified

a single hs3st2 gene in this species (ENSDARG000000

59616). We confirmed that zebrafish hs3st2 is composed

of two exons separated by a 452 kb intron, allowing the

synthesis of both translation- and splice-blocking morpho-

linos, which are referred to below as MO-hs3st2AUG and

MO-hs3st2SLP, respectively (Supplementary Fig. 7A). RT-

PCR analysis demonstrated efficient zebrafish hs3st2 spli-

cing inhibition in Tg[HuC::hTauP301L;DsRed] embryos

after injection of 1.5 mM MO-hs3st2SPL (Supplementary

Fig. 7B). As control, we used a five mismatch-containing

derivative of MO-hs3st2AUG (mmMO-hs3st2AUG).

Tg[HuC::hTauP301L;DsRed] embryos injected with 1.5 mM

MO-hs3st2SPL (n = 250) or MO-hs3st2AUG (n = 250) did

not show increased lethality or any visible phenotype

including developmental delay when compared to non-in-

jected controls or embryos that received 1.5 mM mmMO-

hs3st2AUG (n = 250) (Supplementary Fig. 7C and D). We

Figure 4 Phosphorylation of tau and tauP301L at disease-

related sites is induced by heparin in vitro and by 3-O-

sulphated heparan sulphates in vitro and in cells. (A–C)

GSK3B-mediated abnormal phosphorylation of wild-type tau and

tauP301L is dependent on the presence of heparin in the reaction

mixture. (A) GSK3B-mediated abnormal phosphorylation of tau

at AT270, PHF1 and AT8. (B) GSK3B/PKA-mediated abnormal

phosphorylation of tau at the AT100 site. (C) GSK3B-mediated tau

hyperphosphorylation at sites pSer396 and pSer199/202.

(D) GSK3B-mediated abnormal phosphorylation of tau and tauP301L

at epitopes AT270 and PHF1. (E) Lentiviral vector miR-HS3ST2-

mediated down regulation of HS3ST2 shows a decreased accumu-

lation of hyperphosphorylated tau in SH/hTauP301L cells. All

experiments were performed at least three times.

10 | BRAIN 2015: Page 10 of 16 J. E. Sepulveda-Diaz et al.

by guest on April 5, 2015

Dow

nloaded from

then quantified abnormally phosphorylated tau at Thr181

(pThr181) and observed a nearly 60% decrease in the

pThr181 accumulation in Tg[HuC::hTauP301L;DsRed] em-

bryos (n = 300) following either morpholino-mediated zeb-

rafish hs3st2 inactivation, or treatment with LiCl, an

inhibitor of tau phosphorylation used as control treatment

(Supplementary Fig. 7E). Using fluorescence microscopy we

quantified the accumulation of abnormally phosphorylated

tau with antibodies AT270, AT180 (pThr231/pSer235),

and AT8 in brains of 5 dpf Tg[HuC::hTauP301L;DsRed]

larvae injected or not with MO-hs3st2SPL (n = 150). We

observed a significant fluorescence decrease in all the

assayed sites in the brain in zebrafish hs3st2 depleted em-

bryos (n = 150) (Fig. 5). Accordingly, we observed a strong

decreased accumulation of AT270, AT180, PHF1, and AT8

epitopes (Fig. 6) and of the conformational antibody MC1

(Supplementary Fig. 8) in the spinal cord neurons of the 2

dpf zebrafish hs3st2-depleted Tg[HuC::hTauP301L;DsRed]

larvae injected with MO-sh3st2SPL (n = 150). These results

confirm that 3-O-sulphated heparan sulphates play a key

role in abnormal phosphorylation of tau in tauopathy

in vivo.

Hs3st2 depletion rescues TauP301L

induced functional defects

Because 2 dpf Tg[HuC::hTauP301L;DsRed] larvae display

reduced motor neuronal axon branching and elongation

and, as a likely consequence, an impaired escape response

to touch stimuli (Paquet et al., 2009), this line provides us

with a tool to further investigate in vivo the relationship

between 3-O-sulphated heparan sulphates, tau abnormal

phosphorylation and neuronal pathophysiology. First, we

analysed motor neuronal axon morphology using the anti-

syt1, antibody that specifically recognizes synaptotagmin

1, a synaptic protein expressed in primary motor neurons

(Trevarrow et al., 1990; Fox and Sanes, 2007). As previ-

ously described (Paquet et al., 2009), we observed that

motor neuronal axons were morphologically altered in

the Tg[HuC::hTauP301L;DsRed] embryos, whereas morph-

ology was partially recovered after zebrafish hs3st2 deple-

tion (Fig. 7A). We then measured the length of

outgrowing axons of the four caudal primary motor

neurons immediately anterior to the end of the yolk

extension. As expected, the mean length of motor

neuronal axons was markedly reduced in 48 hpf

Tg[HuC::hTauP301L;DsRed] embryos when compared to

that observed in their non-transgenic siblings (Paquet

et al., 2009) However, after zebrafish hs3st2 depletion,

the mean length of motor neuronal axons was significantly

increased in Tg[HuC::hTauP301L;DsRed] embryos to

nearly normal length (n = 25; P50.005) (Fig. 7B). We

next sought to investigate whether zebrafish hs3st2 deple-

tion can rescue, at least partially, the motility defects

observed in Tg[HuC::hTauP301L;DsRed] zebrafish larvae

(Paquet et al., 2009). Following injection of 1.5 mM

MO-hs3st2SPL, the escape response to touch stimulus in

Tg[HuC::hTauP301L;DsRed] larvae was significantly

increased to near levels measured in non-transgenic sib-

lings (n = 150; P5 0.005) (Fig. 7C). Collectively, these

results provide functional evidence supporting the essen-

tial involvement of 3-O-sulphated heparan sulphates in

tau neuropathology in vivo.

DiscussionWe show that HS3ST2 is overexpressed in the hippocam-

pus of patients with Alzheimer’s disease and that intracel-

lular 3-O-sulphated heparan sulphates are critical for the

abnormal phosphorylation of both wild-type tau and tau

carrying the P301L mutation in cells and in a zebrafish

model of tauopathy. Although the occurrence of the intra-

cellular accumulation of heparan sulphates long before the

development of the tau pathology was known in neurons of

patients with Alzheimer’s disease and those with Down’s

syndrome for several years (Snow et al., 1990; Goedert

et al., 1996), the mechanisms, biological significance, and

pathophysiological consequences of this event remain un-

known. Similarly, although HS3ST2 expression occurs in

neurons of brain regions vulnerable to Alzheimer’s disease,

the involvement of neuronal 3-O-sulphated heparan sul-

phates and of their 3-O-sulphotransferase synthetizing en-

zymes in the mechanisms leading to tau pathology had

never been investigated. Here, we show that under path-

ology-associated conditions, such as hTauP301L expression

or oxidative stress, membrane-associated heparan sulphate

uptake occurs concomitantly with tau hyperphosphoryla-

tion, delineating a link between membrane biology,

heparan sulphate biology, and tau pathology. This link is

supported by several observations associating HSPG with

altered-membrane biology in Alzheimer’s disease, for in-

stance: (i) the intracellular co-staining of membrane-asso-

ciated HSPG, but not of secreted HSPGs, with

neurofibrillary tangles in Alzheimer’s disease (van Horssen

et al., 2003); (ii) the heparan sulphate-dependent endocyto-

sis of amyloid-b oligomers (Sandwall et al., 2010), tau fi-

brils (Holmes et al., 2013), apolipoprotein-E (Wilsie et al.,

2006), and other proteins for which heparan sulphates are

cell-surface endocytic receptors (Christianson and Belting,

2014); (iii) the relevance in lipid rafts of HSPG, which act

as molecular mediators of raft-dependent endocytosis

(Chen and Williams, 2013); (iv) the heparan sulphates cap-

acity to interact and regulate the activity of membrane-

associated enzymes involved in Alzheimer’s disease such

as the raftophilic b-secretase (Patey et al., 2006).

Moreover, the 3-O-sulphated heparan sulphates require-

ment for HSV1 uptake by neuronal cells (O’Donnell

et al., 2006; Antoine et al., 2014) supports the existence

of the proposed synergy between HSV1 infection and

Alzheimer’s disease pathology (Carter, 2008). Finally, the

central importance of an altered neuronal membrane biol-

ogy in Alzheimer’s disease (Williamson and Sutherland,

HS3ST2 is critical for tau pathology BRAIN 2015: Page 11 of 16 | 11

by guest on April 5, 2015

Dow

nloaded from

Figure 5 Hs3st2-mediated HSPG sulphation is required for abnormal phosphorylation of tau in zebrafish brain. Double

immunostaining for total tau (red) and phosphorylated tau (green) at residues AT270 (A1-A2”), AT180 (B1-B2”), and AT8 (C1-C2”) in 5 dpf

dissected brains from Tg[HuC::hTauP301L] (Hu::Tau) embryos following morpholino-mediated depletion of zebrafish hs3st2 (Hu::Tau + MO-

hs3st2) (n = 150). Scale bar = 50 mm. The percentage of cells expressing phospho-tau epitopes recognized by antibodies AT270 (left), AT180

(middle) and AT8 (right) compared to cells immunolabelled for total tau is shown (n = 150).

12 | BRAIN 2015: Page 12 of 16 J. E. Sepulveda-Diaz et al.

by guest on April 5, 2015

Dow

nloaded from

Figure 6 HS3ST2-mediated sulphation of heparan sulphates is required for the pathological phosphorylation of tau in spinal

cord neurons in zebra fish. (A) Double immunostaining for total tau (red) and abnormally phosphorylated tau (green) with anti-tau antibodies

AT180, AT270, PHF1, and AT8 in spinal cord neurons of Tg[HuC::hTauP301L] (Hu::Tau) embryos following morpholino-mediated depletion of

zebrafish hs3st2 (Hu::Tau + MO-hs3st2). (B) Immunostaining quantification of antibodies AT180, AT270, PHF1, and AT8 signals in spinal cord

neurons. Panels in A represent observations (n = 150) of 5 dpf larvae from two independent experiments. Panels in B represents means of

fluorescence analyses on 150 larvae; error bars show SEM. Scale bar = 50 mm.

HS3ST2 is critical for tau pathology BRAIN 2015: Page 13 of 16 | 13

by guest on April 5, 2015

Dow

nloaded from

2011) is strongly supported by the striking observation that

most, if not all, Alzheimer’s disease associated loci identified

by genome-wide association studies (GWAS) including

ABCA7, MS4A6A/MS4A4E, EPHA1, CD33, CD2AP, CLU,

PICALM, CR1, BIN1, and SNORA7A (Hollingworth et al.,

2011), are most of them, as APP, PSEN1, and PSEN2, genetic

variants of membrane-associated proteins. Whether these

genetic variants, as well as an altered tau biology (Stamer

et al, 2002; Ke et al, 2012), can provoke cell-membrane

alterations allowing increased heparan sulphate intern-

alization, are important points that remain to be explored.

Concerning the implication of intracellular heparan

sulphates in the molecular mechanisms leading to the ab-

normal phosphorylation of tau, our data suggest that the

3-O-sulphated glycosaminoglycans bind to tau in a molecu-

lar chaperon-like manner allowing the tau abnormal phos-

phorylation. This hypothesis is supported by previous 1H

NMR (proton nuclear magnetic resonance) and biochem-

ical cross-linking studies showing that heparin readily in-

duces conformational changes in tau allowing the access of

different kinases to epitopes otherwise inaccessible for

phosphorylation (Paudel and Li, 1999; Sibille et al.,

2006), and further supported by our immunostaining,

FRET and phosphorylation data, as well as by the intra-

cellular accumulation of heparan sulphates observed previ-

ous to the apparition of the tau pathology in neuronal cells

of Alzheimer’s disease and Down’s syndrome (Snow el al.,

1990; Goedert et al., 1996). Furthermore, as the 3-O-

sulphated heparan sulphates prompt the abnormal phos-

phorylation of both tau and tauP301L, the enhanced effect

observed on tauP301L could be associated with the presence

of the tau mutation (Alonso Adel et al., 2004). Thus, the

higher susceptibility of tauP301L to undergo heparin-dependent

abnormal phosphorylation can explain the effect of inhibiting

HS3ST2 expression in SH/hTauP301L cells and of hs3st2 in

Tg[HuC::hTauP301L; DsRed] zebrafish embryos, in which the

levels of hs3st2 transcripts were found unchanged compared

to those observed in control embryo (not shown).

Whether the central role of the abnormal phosphoryl-

ation of tau in the Alzheimer’s disease neurodegenerative

process is still controverted, the possibility of a correlation

between genetic variability in membrane associated-pro-

teins, or between altered tau biology, with heparan sul-

phate internalization and interaction with tau, requires to

be considered. This additionally raises questions concerning

the intracellular association of different species of glycosa-

minoglycans with other intracellular heparin-binding pro-

teins, such as �-synuclein and superoxide dismutase,

proteins involved in important neurodegenerative diseases.

Indeed, as for neurofibrillary tangles, sulphated glycosami-

noglycans generally co-localize with amyloidosis lesions,

including �-synucleinopathies and prion diseases (Zhang

and Li, 2010), suggesting the existence of common events

in the mechanisms involving the sulphated sugars in pro-

teinopathies. In conclusion, our results position intracellu-

lar 3-O-sulphated heparan sulphates, and the enzymes

responsible of their high sulphation in neurons, as central

modulators of tau abnormal phosphorylation before it

occurs. This opens a wide area of research in the field of

glyco-neurobiology, positioning HS3ST2 and its products

as potential key players in the development and evolution

of the tau pathology, with the therapeutic consequences

that this implies.

Figure 7 Partial depletion of zebrafish hs3st2 rescues both motor neuron axon outgrowth defects and behavioural abnorm-

alities in Tg[HuC::hTauP301L] embryos. (A) Spinal cord neurons lateral views in yolk extensions, visualized by anti-Syt1, in wild-type

embryos (Control), in Tg[HuC::hTauP301L] embryos that express the hTauP301L mutation (Hu::tau), and in Tg[HuC::hTauP301L] embryos injected

with MO-hs3st2 (Hu::tau + MO-hs3st2). (B) Motor neuronal axons length mean (n = 25) in control wild-type embryos (WT), in

Tg[HuC::hTauP301L] embryos (Hu::Tau), and in Tg[HuC::hTauP301L] embryos following depletion of hs3st2 (Hu::Tau + MO-hs3st2). (C) Motor

response to touching stimuli determined by the percentage of responding larvae in Tg[HuC::hTauP301L] embryos (Hu::Tau) and in

Tg[HuC::hTauP301L] embryos following depletion of hs3st2 (Hu::Tau + MO-hs3st2) (n = 150); error bars show SEM.

14 | BRAIN 2015: Page 14 of 16 J. E. Sepulveda-Diaz et al.

by guest on April 5, 2015

Dow

nloaded from

AcknowledgementsWe thank to C. Haass, B. Schmid, and D. Paquet (DZNE,

Munich, Germany) for providing the

Tg[HuC::hTauP301L;DsRed] transgenic line, to A. Eckert

(University of Basel, Switzerland) and J. Gotz (University

of Queensland-Brisbane, Australia) for providing the initial

vials of SH-SY5Y cells overexpressing mutant hTauP301L,

and to D. Biard (CEA, Fontenay-aux-Roses, France) for his

help in gene silencing experiments. We also thank to G.

Carpentier for assistance in microscopy experiments, to

A. Boiret, S. Christiaanse (UPEC), and P. Claus (INSERM

U676) for technical assistance and to C. Romain for help

mantaining zebrafish facility. We thank to B. Jacquet, O.

Stettler (UPEC), and C. Duyckaerts (P. Salpetriere Hospital

Brain Bank) for interesting discussions.

FundingAuthors thank to the Association France Alzheimer &

Maladies Apparentees and to the SATT Idf Innov for sup-

porting this work. J.E. Sepulveda-Diaz received the schol-

arship No. 308978 from CONACyT, Mexico. M.B. Huynh

received scholarship from the French Ministry of Higher

Education and Research. S.M. Alavi Naini received a

grant from « Institute de Recherche Servier » independently

of the work in this article.

Supplementary materialSupplementary material is available at Brain online.

ReferencesAlonso Adel C, Mederlyova A, Novak M, Grundke-Iqbal I, Iqbal K.

Promotion of hyperphosphorylation by frontotemporal dementia tau

mutations. J Biol Chem 2004; 279: 34873–81.

Antoine TE, Yakoub A, Maus E, Shukla D, Tiwari V. Zebrafish 3-O-

sulfotransferase-4 generated heparan sulfate mediates HSV1 entry

and spread. PLoS One 2014; 9: e87302.

Braak H, Del Tredici K. Where, when, and in what form does spor-

adic Alzheimer’s disease begin? [Review]. Curr Opin Neurol 2012;

25: 708–14.Carter CJ. Interactions between the products of the Herpes simplex

genome and Alzheimer’s disease susceptibility genes: relevance to

pathological-signalling cascades [Review]. Neurochem Int 2008;

52: 920–34.

Chen K, Williams KJ. Molecular mediators for raft-dependent endo-

cytosis of syndecan-1, a highly conserved, multifunctional receptor. J

Biol Chem 2013; 288: 13988–99.Christianson HC, Belting M. Heparan sulfate proteoglycan as a cell-

surface endocytosis receptor [Review]. Matrix Biol 2014; 35:

51–510.1016/j.matbio.2013.10.004.Cripps D, Thomas SN, Jeng Y, Yang F, Davies P, Yang AJ. Alzheimer

disease-specific conformation of hyperphosphorylated paired helical

filament-Tau is polyubiquitinated through Lys-48, Lys-11, and Lys-6

ubiquitin conjugation. J Biol Chem 2006; 281: 10825–38.

Duka V, Lee JH, Credle J, Wills J, Oaks A, Smolinsky C, et al.

Identification of the sites of tau hyperphosphorylation and activation

of tau kinases in synucleinopathies and Alzheimer’s diseases. PloS

One 2013; 8: e75025.

Ferrari A, Hoerndli F, Baechi T, Nitsch RM, Gotz J. beta-Amyloid

induces paired helical filament-like tau filaments in tissue culture. J

Biol Chem 2003; 278: 40162–8.

Fox MA, Sanes JR. Synaptotagmin I and II are present in distinct

subsets of central synapses. J Comp Neurol 2007; 503: 280–96.

Goedert M, Jakes R, Spillantini MG, Hasegawa M, Smith MJ,

Crowther RA. Assembly of microtubule-associated protein tau into

Alzheimer-like filaments induced by sulphated glycosaminoglycans.

Nature 1996; 383: 550–3.Hasegawa M, Crowther RA, Jakes R, Goedert M. Alzheimer-like

changes in microtubule-associated protein Tau induced by sulfated

glycosaminoglycans. Inhibition of microtubule binding, stimulation

of phosphorylation, and filament assembly depend on the degree of

sulfation. J Biol Chem 1997; 272: 33118–24.

Hashiguchi M, Hashiguchi T. Kinase-kinase interaction and modula-

tion of tau phosphorylation [Review]. Int Rev Cell Mol Biol 2013;

300: 121–60.

Hernandez F, Perez M, Lucas JJ, Avila J. Sulfo-glycosaminoglycan

content affects PHF-tau solubility and allows the identification of

different types of PHFs. Brain Res 2002; 935: 65–72.

Holmes BB, DeVos SL, Kfoury N, Li M, Jacks R, Yanamandra K,

Ouidja MO, et al. Heparan sulfate proteoglycans mediate internal-

ization and propagation of specific proteopathic seeds. Proc Natl

Acad Sci USA 2013; 110: E3138–47.Hollingworth P, Harold D, Sims R, Gerrish A, Lambert JC,

Carrasquillo MM, et al. Common variants at ABCA7, MS4A6A/

MS4A4E, EPHA1, CD33 and CD2AP are associated with

Alzheimer’s disease. Nat Genet 2011; 43: 429–35.

Huynh MB, Villares J, Dıaz JE, Christiaans S, Carpentier G,

Ouidja MO, et al. Glycosaminoglycans from aged human hippo-

campus have altered capacities to regulate trophic factors activities

but not Abeta42 peptide toxicity. Neurobiol Aging 2012; 33:

1005e11–22.

Iqbal K, Liu F, Gong CX, Grundke-Iqbal I. Tau in Alzheimer disease

and related tauopathies [Review]. Curr Alzheimer Res 2010; 7:

656–64.Ke YD, Suchowerska AK, van der Hoven J, De Silva DM, Wu CW,

van Eersel J, et al. [Review]. Lessons from tau-deficient mice. Int J

Alzheimers Dis 2012; 2012: 873270.Keller KE, Bradley JM, Kelley MJ, Acott TS. Effects of modifiers of

glycosaminoglycan biosynthesis on outflow facility in perfusion cul-

ture. Invest Ophthalmol Vis Sci 2008; 49: 2495–505.Kimmel CB, Ballard WW, Kimmel SR, Ullmann B, Schilling TF. Stages

of embryonic development of the zebrafish. Dev Dyn 1995; 203:

253–310.

Kreuger J, Kjellen L. Heparan sulfate biosynthesis: regulation and vari-

ability. [Review]. J Histochem Cytochem 2012; 60: 898–907.

Lawrence R, Yabe T, Hajmohammadi S, Rhodes J, McNeely M, Liu J,

et al. The principal neuronal gD-type 3-O-sulfotransferases and their

products in central and peripheral nervous system tissues. Matrix

Biol 2007; 26: 442–55.

Li M, Chen X, Ye QZ, Vogt A, Yin XM. A high-throughput FRET-

based assay for determination of Atg4 activity. Autophagy 2012; 8:

401–12.

Mochizuki H, Yoshida K, Shibata Y, Kimata K. Tetrasulfated disac-

charide unit in heparan sulfate: enzymatic formation and tissue dis-

tribution. J Biol Chem 2008; 283: 31237–45.

O’Donnell CD, Tiwari V, Oh MJ, Shukla D. A role for heparan sulfate

3-O-sulfotransferase isoform 2 in herpes simplex virus type 1 entry

and spread. Virology 2006; 346: 452–9.

Paquet D, Bhat R, Sydow A, Mandelkow EM, Berg S, Hellberg S,

et al. A zebrafish model of tauopathy allows in vivo imaging of

neuronal cell death and drug evaluation. J Clin Invest 2009; 119:

1382–95.

HS3ST2 is critical for tau pathology BRAIN 2015: Page 15 of 16 | 15

by guest on April 5, 2015

Dow

nloaded from

Patey SJ, Edwards EA, Yates EA, Turnbull JE. Heparin derivatives asinhibitors of BACE-1, the Alzheimer’s beta-secretase, with reduced

activity against factor Xa and other proteases. J Med Chem 2006;

49: 6129–32.

Paudel HK, Li W. Heparin-induced conformational change in micro-tubule-associated protein Tau as detected by chemical cross-linking

and phosphopeptide mapping. J Biol Chem 1999; 274: 8029–38.

Safaiyan F, Kolset SO, Prydz K, Gottfridsson E, Lindahl U,

Salmivirta M. Selective effects of sodium chlorate treatment on thesulfation of heparan sulfate. J Biol Chem 1999; 274: 36267–73.

Sandwall E, O’Callaghan P, Zhang X, Lindahl U, Lannfelt L, Li JP.

Heparan sulfate mediates amyloid-beta internalization and cytotox-icity. Glycobiology 2010; 20: 533–41.

Shriver Z, Capila I, Venkataraman G, Sasisekharan R. Heparin and

heparan sulfate: analyzing structure and microheterogeneity

[Review]. Handb Exp Pharmacol 2012; 207: 159–76.Schefe JH, Lehmann KE, Buschmann IR, Unger T, Funke-Kaiser H.

Quantitative real-time RT-PCR data analysis: current concepts and

the novel “gene expression’s CT difference” formula [Review].

J Mol Med 2006; 84: 901–10.Schroeder A, Mueller O, Stocker S, Salowsky R, Leiber M,

Gassmann M, et al. The RIN: an RNA integrity number for

assigning integrity values to RNA measurements. BMC Mol Biol

2006; 7: 3.Shworak NW, Liu J, Petros LM, Zhang L, Kobayashi M,

Copeland NG, et al. Multiple isoforms of heparan sulphate

D-glucosaminyl 3-O-sulfotransferase. Isolation, characterization,and expression of human cDNAs and identification of distinct gen-

omic loci. J Biol Chem 1999; 274: 5170–84.

Sibille N, Sillen A, Leroy A, Wieruszeski JM, Mulloy B, Landrieu I,

et al. Structural impact of heparin binding to full-length Tauas studied by NMR spectroscopy. Biochemistry 2006; 45:

12560–72.

Snow AD, Mar H, Nochlin D, Sekiguchi RT, Kimata K, Koike Y,

et al. Early accumulation of heparan sulfate in neurons and in thebeta-amyloid protein-containing lesions of Alzheimer’s disease and

Down’s syndrome. Am J Pathol 1990; 137: 1253–70.

Spillantini MG, Goedert M. Tau pathology and neurodegeneration[Review]. Lancet Neurol 2013; 12: 609–22.

Stamer K, Vogel R, Thies E, Mandelkow E, Mandelkow EM. Tau

blocks traffic of organelles, neurofilaments, and APP vesicles in

neurons and enhances oxidative stress. J Cell Biol 2002; 156:1051–63.

Steinhilb ML, Dias-Santagata D, Fulga TA, Felch DL, Feany MB. Tau

phosphorylation sites work in concert to promote neurotoxicity

in vivo. Mol Biol Cell 2007; 18: 5060–8.Ten Dam GB, Kurup S, van de Westerlo EM, Versteeg EM, Lindahl U,

Spillmann D, et al. 3-O-sulfated oligosaccharide structures are recog-

nized by anti-heparan sulfate antibody HS4C3. J Biol Chem 2006;281: 4654–62.

Thacker BE, Xu D, Lawrence R, Esko JD. Heparan sulfate 3-O-sulfa-

tion: a rare modification in search of a function [Review]. Matrix

Biol 2014; 35: 60–7210.1016/j.matbio.2013.12.001.

Trevarrow B, Marks DL, Kimmel CB. Organization of hindbrain seg-ments in the zebrafish embryo. Neuron 1990; 4: 669–79.

van Horssen J, Wesseling P, van den Heuvel LP, de Waal RM,

Verbeek MM. Heparan sulphate proteoglycans in Alzheimer’s dis-

ease and amyloid-related disorders [Review]. Lancet Neurol 2003; 2:482–92.

Vandesompele J, De Preter K, Pattyn F, Poppe B, Van Roy N, De

Paepe A, et al. Accurate normalization of real-time quantitative

RT-PCR data by geometric averaging of multiple internal controlgenes. Genome Biol 2002; 3:RESEARCH0034.

Venkatachalam KV. Human 3’-phosphoadenosine 5’-phosphosulfate

(PAPS) synthase: biochemistry, molecular biology and genetic defi-ciency [Review]. IUBMB Life 2003; 55: 1–11.

Wang JZ, Xia YY, Grundke-Iqbal I, Iqbal K. Abnormal hyperpho-

sphorylation of tau: sites, regulation, and molecular mechanism of

neurofibrillary degeneration. [Review]. J Alzheimers Dis 2013; 33(Suppl 1): S123–39.

Wenk GL. Neuropathologic changes in Alzheimer’s disease [Review].

J Clin Psychiatry 2003; 64 (Suppl 9): 7–10.

Wilsie LC, Gonzales AM, Orlando RA. Syndecan-1 mediates internal-ization of apoE-VLDL through a low density lipoprotein receptor-

related protein (LRP)-independent, non-clathrin-mediated pathway.

Lipids Health Dis 2006; 5: 23.

Williams DR. Tauopathies: classification and clinical update on neu-rodegenerative diseases associated with microtubule-associated pro-

tein tau [Review]. Intern Med J 2006; 36: 652–60.

Williamson R, Sutherland C. Neuronal membranes are key to thepathogenesis of Alzheimer’s disease: the role of both raft and non-

raft membrane domains [Review]. Curr Alzheimer Res 2011; 8:

213–21.

Wu X, Jiang X, Marini AM, Lipsky RH. Delineating and understand-ing cerebellar neuroprotective pathways: potential implication for

protecting the cortex [Review]. Ann N Y Acad Sci 2005; 1053:

39–47.

Yabe T, Hata T, He J, Maeda N. Developmental and regional expres-sion of heparan sulphate sulfotransferase genes in the mouse brain.

Glycobiology 2005; 15: 982–993.

Yoshida H, Goedert M. Sequential phosphorylation of tau protein bycAMP-dependent protein kinase and SAPK4/p38delta or JNK2 in

the presence of heparin generates the AT100 epitope. J

Neurochem 2006; 99: 154–64.

Zheng-Fischhofer Q, Biernat J, Mandelkow EM, Illenberger S,Godemann R, Mandelkow E. Sequential phosphorylation of Tau

by glycogen synthase kinase-3beta and protein kinase A at Thr212

and Ser214 generates the Alzheimer-specific epitope of antibody

AT100 and requires a paired-helical-filament-like conformation.Eur J Biochem 1998; 252: 542–552.

Zhang X, Li JP. Heparan sulfate proteoglycans in amyloidosis

[Review]. Progr Mol Biol Transl Sci 2010; 93: 309–34.Zhang G, Morin C, Zhu X, Bao Huynh M, Ouidir Ouidja M,

Sepulveda-Diaz JE, et al. Self-evolving oxidative stress with identifi-

able pre- and postmitochondrial phases in PC12 cells. J Neurosci

Res 2013; 91: 273–84.

16 | BRAIN 2015: Page 16 of 16 J. E. Sepulveda-Diaz et al.

by guest on April 5, 2015

Dow

nloaded from