Tau Interacts With Golgi Membranes andMediates Their Association With Microtubules
Carole Abi Farah,{{ Sebastien Perreault,{ Dalinda Liazoghli,{ Mylene Desjardins,Angela Anton, Michel Lauzon, Jacques Paiement, and Nicole Leclerc*
Departement de Pathologie et Biologie Cellulaire, Universite de Montreal,Montreal, Quebec, Canada
Tau, a microtubule-associated protein enriched in the axon, is known to stabilize andpromote the formation of microtubules during axonal outgrowth. Several studies havereported that tau was associated with membranes. In the present study, we furthercharacterized the interaction of tau with membranous elements by examining its dis-tribution in subfractions enriched in either Golgi or endoplasmic reticulum mem-branes isolated from rat brain. A subfraction enriched with markers of the medialGolgi compartment, MG160 and mannosidase II, presented a high tau content indi-cating that tau was associated with these membranes. Electron microscope morphom-etry confirmed the enrichment of this subfraction with Golgi membranes. Double-immunogold labeling experiments conducted on this subfraction showed the directassociation of tau with vesicles labeled with either an antibody directed againstMG160 or TGN38. The association of tau with the Golgi membranes was furtherconfirmed by immunoisolating Golgi membranes with an anti-tau antibody. Immuno-gold labeling confirmed the presence of tau on the Golgi membranes in neurons invivo. Overexpression of human tau in primary hippocampal neurons induced the for-mation of large Golgi vesicles that were found in close vicinity to tau-containingmicrotubules. This suggested that tau could serve as a link between Golgi membranesand microtubules. Such role for tau was demonstrated in an in vitro reconstitutionassay. Finally, our results showed that some tau isoforms present in the Golgi subfrac-tion were phosphorylated at the sites recognized by the phosphorylation-dependentantibodies PHF-1 and AT-8. Cell Motil. Cytoskeleton 63:710–724, 2006. ' 2006
Wiley-Liss, Inc.
Key words: tau membranous association; microtubules; Golgi membranes; subcellular fractionation;
immunogold; in vitro reconstitution assay
INTRODUCTION
Tau is a neuronal microtubule-associated protein(MAP) that is enriched in the axonal compartment [Bueeet al., 2000]. In vitro studies have shown that tau can stabi-lize microtubules by decreasing their dynamic instability[Garcia and Cleveland, 2001]. Consistently, the suppres-sion of tau expression decreases the pool of stable microtu-bules and impairs axonal differentiation in primary neuronalcultures [Caceres et al., 1991, 1992; Dawson et al., 2001].
Recently, tau was shown to be involved in thetransport of membranous organelles. Overexpression oftau in both neuronal and nonneuronal cells resulted inthe accumulation of mitochondria, Golgi membranes,
{These authors contributed equally to experiments in this study.
{Current address: Montreal Neurological Institute, McGill University,
3801 University Avenue, Montreal, Quebec, Canada H3A 2B4.
*Correspondence to: Dr. Nicole Leclerc, Departement de pathologie
et biologie cellulaire, Universite de Montreal, C.P.6128, Succ. Centre-
ville, Montreal, Quebec, Canada H3C 3J7.
E-mail: [email protected]
Contract grant sponsor: National Sciences and Engineering Research
Council of Canada; Contract grant sponsor: Canadian Institute of
Health Research; Contract grant numbers: MOP-53218, MOP-44022;
Contract grant sponsor: Alzheimer Society of Canada.
Received 2 May 2006; Accepted 26 July 2006
Published online 7 September 2006 in Wiley InterScience (www.
interscience.wiley. com).
DOI: 10.1002/cm.20157
' 2006 Wiley-Liss, Inc.
Cell Motility and the Cytoskeleton 63:710–724 (2006)
and peroxisomes in the perinuclear region and their lossat the cell periphery [Drewes et al., 1998; Stamer et al.,2002]. This redistribution of membranous organellescould be caused by impairment of the motor proteinbinding to microtubules by tau. Indeed, in vitro studiesshowed that kinesin binding to microtubules wasdecreased in the presence of tau [Heins et al., 1991]. Inastrocytes, the decrease of membranous organelle trans-port induced by the overexpression of tau was correlatedto a reduction of detyrosinated tubulin and a decrease ofkinesin protein level [Yoshiyama et al., 2003].
In Alzheimer’s disease (AD), tau becomes hyper-phosphorylated and aggregates in insoluble filamentscalled paired helical filaments (PHFs) [Lee et al., 2001].It is believed that hyperphosphorylated tau contributes toneurodegeneration by detaching from microtubules andthereby destabilizing the axonal microtubule network.
Besides interacting with microtubules, tau was alsofound to associate with the plasma membrane through aproline rich sequence located in the amino-terminal[Brandt et al., 1995]. This sequence mediates the associ-ation of tau with the SH3 domains of Fyn and src nonre-ceptor tyrosine kinases [Lee et al., 1998]. Interaction oftau with the plasma membrane was shown to be neces-sary for the formation of axonal-like processes in PC12cells [Brandt et al., 1995]. Tau was also found to be asso-ciated with the outer mitochondrial membrane [Junget al., 1993]. Collectively, the above observations indi-cate that tau could be involved in the interaction of mem-branous elements with microtubules.
Here, we describe the association of tau with Golgimembranes by using subcellular fractionation, immunoi-solation, and electron microscope immunocytochemistryin rat and mouse brain. Moreover, an in vitro reconstitu-tion assay was used to show that tau could serve as alinker between Golgi membranes and microtubules. Weand others have recently reported that the overexpressionof tau in neurons and astrocytes induced a fragmentationof the Golgi apparatus (GA) [Yoshiyama et al., 2003;Liazoghli et al., 2005]. The present results suggest thattau could be involved in regulating the structure of theGA through a direct association with Golgi membranesand/or by mediating their association with microtubules.
MATERIALS AND METHODS
Subcellular Fractionation
Adult Sprague–Dawley rats were purchased atCharles River (Charles River Laboratories Inc., Mon-treal, Quebec, Canada). The use of animals and all surgi-cal procedures described in this article were carried outaccording to The Guide to the Care and Use of Experi-mental Animals of the Canadian Council on Animal
Care. Brain was dissected from twenty adult Sprague–Dawley rats. Subcellular fractions were generated usingthe protocol previously described by Lavoie et al.[Lavoie et al., 1996] and illustrated in Farah et al. [Farahet al., 2005]. Briefly, total microsomes were isolated bydifferential centrifugation and ER and Golgi elementswere subsequently separated by ultracentrifugation in asucrose step-gradient.
Immunoblot Analysis
Protein assay was performed (Bio-Rad kit, Bio-Rad Laboratories Ltd., Mississauga, Ontario, Canada).Equal amounts of proteins were loaded in each lane andelectrophoresed on a 7.5% polyacrylamide gel. Follow-ing separation, proteins were electrophoretically trans-ferred to a nitrocellulose membrane. The nitrocellulosestrips were incubated with the primary antibodies for90 min at room temperature. They were then washedwith phosphate buffered saline (PBS) and incubated withthe peroxidase-conjugated secondary antibodies. Mem-branes were again washed and revealed by chemi-luminescence (Amersham Pharmacia Biotech, Quebec,Quebec, Canada). The intensity of the signal for eachband was measured on film by densitometry using theScion Image program (Scion Corp, Frederick, MA). Thefollowing primary antibodies were used: the monoclonalantibody anti-MAP2 (clone HM2, Sigma, Oakville, ON),the monoclonal antibodies anti-tau, Tau-1 directedagainst dephosphorylated tau (Oncogene Research Prod-ucts, San Diego, California), Tau-5 (Biosource Interna-tional, Medicorp, Montreal, Canada), and 49.2 (epitopeunknown) (kindly provided by Virginia M-Y Lee, Uni-versity of Pennsylvania, Philadelphia), a polyclonal anti-tau antibody (34–64 residues) (kindly provided byVirginia M-Y Lee, University of Pennsylvania, Philadel-phia), the monoclonal antibodies AT8 (Pierce Endogen)and PHF-1 (kindly provided by Dr. Peter Davies, AlbertEinstein Medical College, NY, NY) directed againstphosphorylated tau, the monoclonal antibody anti-a-tubulin (clone DM 1A, Sigma, Oakville, ON), a polyclo-nal antibody against ribophorin (kindly provided by Dr.G. Kreibich, New York University School of Medicine,New York, NY), a polyclonal antibody against calnexin(Stressgen Biotechnologies, Victoria, BC, Canada), apolyclonal antibody against mannosidase II (kindly pro-vided by Dr. M.G. Farquhar, University of California,San Diego), a polyclonal against NaK-ATPase (kindlyprovided by Dr. D. Fambrough, The Johns Hopkins Uni-versity, Maryland), a monoclonal directed againstMG160 (kindly provided by Dr. N. Gonatas, Universityof Pennsylvania, Philadelphia), GM130 (Oncogene,USA) and a polyclonal directed against TGN38 (Serotec,UK).
Tau Links Golgi Membranes to Microtubules 711
Immunoisolation
Total microsomes prepared from adult rat brainwere resuspended in 0.25 M sucrose. Magnetic beadseither covalently bound to sheep anti-mouse IgG or anti-rabbit IgG (Dynal, Biotech Inc., Lake Success, NY) wereeither coated with a primary antibody directed againsttau (polyclonal antibody, 1:500 or the monoclonal anti-tau antibody, Tau-1, 1:500) or with non-specific mouseor rabbit IgG (Jackson ImmunoResearch Laboratories,West Grove, PA) for 3 h at 48C. The beads were washedthree times to remove unbound antibodies. Then, theywere incubated with 250 lg of microsomes in the immu-noisolation buffer (IB) containing 50 mM Tris-HCl pH7.4, 0.2 M sucrose, 50 mM KCl, 10 mM MgCl2, 1.5%BSA, 1 mM NAF, 1 mM Na3VO4 and protease inhibitors(Roche Molecular Biochemicals, Mannheim, Germany)for 2 h at 48C. The beads containing tau-bound mem-branes were separated from the unbound material using amagnet. The beads were washed three times in IB. Theimmunoisolate was analyzed by western blot.
Electron Microscopy
Subfractions isolated from rat brain were fixedusing 2.5% glutaraldehyde, recovered onto Milliporemembranes by the random filtration technique ofBaudhuin et al. [Baudhuin et al., 1967] and processed forelectron microscopy as previously described [Lavoieet al., 1996].
Cell Culture and Transfection
Primary embryonic hippocampal cultures were pre-pared from rat embryos as previously described [Bankerand Goslin, 1998]. Hippocampus from 18-day-oldembryos were treated with trypsin (0.25% at 378C for 15min) then washed in Hank’s balanced solution and disso-ciated by several passages through a constricted Pasteurpipette. The cells were then plated on glass coverslipscoated with polylysine. Then, after 4 h to allow theattachment of the cells to the substrate, the hippocampalcells were inverted to face a monolayer of glial cells in aserum-free medium. Two day-old hippocampal neuronswere transfected with an expression vector containinghuman tau 3R fused to GFP tag as previously described[Farah et al., 2005; Liazoghli et al., 2005]. Twenty-fourhours after transfection, neurons were fixed and proc-essed for immunofluorescence.
Immunofluorescence
Neurons were fixed in 4% paraformaldehyde/PBSfor 30 min. The cells were then permeabilized with 0.2%Triton X-100 in PBS for 5 min. The GA was revealedusing a monoclonal antibody directed against GM130.To visualize microtubles, we used a rat monoclonal anti-
body directed against a-tubulin (abcam, Cambridge,UK). We used the following secondary antibodies: adonkey anti-mouse conjugated to Rhodamine (1:500)(Jackson Immunoresearch Laboratories, Bio/Cam,Mississauga, Ontario, Canada) and a donkey anti-ratconjugated to AMCA (1:200) (Jackson ImmunoresearchLaboratories, Bio/Cam, Mississauga, Ontario, Canada).All these antibodies were diluted in 5% BSA/PBS. Incu-bations were carried out at room temperature (RT) for 1h. After three washes in PBS, the coverslips weremounted in polyvinyl alcohol (Calbiochem, CA). Fluo-rescently labelled cells were visualized with a Zeiss Axi-ophot microscope (Carl Zeiss) and a Leica TCS-SP1confocal microscope using 633 and 1003 objectives.
Electron Microscope Immunocytochemistryon Spinal Cord Sections
Adult mice were deeply anesthetized with sodiumpentobarbital (somnotol, 65 mg/kg i.p.). They were thenperfused through the ascending aorta with 20 ml of 0.1M cacodylate buffer (CB, pH 7.4; RT), followed by 100ml of freshly prepared fixative containing 2.5% glutaral-dehyde (J.B.EM, Pointe Claire, Dorval, Quebec, Canada)and 1% paraformaldehyde (Fisher, NJ) in CB. The spinalcords were removed and kept in the same fixative for 1 hat RT after which they were washed in CB. The spinalcords were cut in sections of 100 lm with a vibratome.The sections were postfixed in 1% osmium tetroxide(J.B.EM, Pointe Claire, Dorval, Quebec, Canada) and1.5% potassium ferrocyanide (Sigma) in CB for 30 minat 48C. After an extensive wash in CB, the sections weredehydrated in a graded series of alcohols at 48C, thenwashed twice for 10 min in 100% ethanol at RT and em-bedded in LR white resin (J.B.EM). The resin was poly-merized under anaerobic conditions for 48 h at 568C.Ultrathin sections (*60 nm) were cut with an ultrami-crotome and placed on single slot nickel grids coatedwith Pioloform. These sections were then processed forpostembedding immunocytochemistry with an antibodydirected against tau (polyclonal antibody) as described inour previous study [Micheva et al., 1998].
Statistical Analysis
The statistical significance of the number of goldparticles in the Golgi compartments on spinal cord sec-tions was determined using an ANOVA one-way testfollowed by Tukey-Kramer multiple comparison test.Statistical significance was accepted if P < 0.05.
Binding of Tau to Rat Liver Golgi Membranes
Magnetic beads (2 3 107 DYNABEADS M-280Sheep anti-mouse IgG) from Dynal Biotech (Lake Suc-cess, NY) were washed according to the manufacturer.The beads were incubated with Tau-5 antibody in a solu-
712 Farah et al.
tion of 0.1% BSA/PBS overnight at 48C with agitation.The beads were washed in 0.1% BSA/PBS three timesand then incubated with 10 lg bovine tau obtained fromCytoskeleton (Denver, CO) in PEM buffer (80 mM PipespH 6.8, 1 mM EGTA and 2.5 mM MgCl2) for 1 h 30 minat 48C with agitation. This incubation was followed bythree washes of the beads in PEM. The beads were thenincubated with either rat liver Golgi (10 lg) or ER(10 lg) membranes for 1 h 30 min at 48C with agitationin GPEM (PEM and 1mM GTP). The beads were sedi-mented, resuspended in SDS sample buffer and boiledfor 5 min. The binding of membranes to tau was ana-lyzed by western blotting as described above.
To test the binding of MAP2 to rat liver Golgi andER membranes, the anti-MAP2 antibody, HM2 wasattached to the magnetic beads bound to an anti-mouseantibody (Dynal Biotech, Lake Success, NY). The beadswere then incubated with a brain MAP fraction obtainedfrom Cytoskeleton Inc. (Denver, CO) that containsMAP2. The beads containing MAP2 were washed andincubated with either rat liver Golgi or ER membranesas described above for tau.
In vitro Microtubule-Membrane ReconstitutionAssay
Bovine brain tubulin and bovine brain biotinylatedtubulin were purchased from Cytoskeleton (Denver,CO). Tubulin (47.5 lg) and biotinylated tubulin (2.5 lg)were incubated with or without 12 lg of bovine tau inGPEM for 20 min at 378C and then 20 lM taxol wasadded to the preparation for 15 min. Taxol stabilizedmicrotubules with or without tau were incubated withmagnetic beads coupled to streptavidin (5 3 106 DYNA-BEADS 280 Streptavidin) for 30 min at 378C with agita-tion. Then the beads covered with microtubules with orwithout tau were incubated with either Golgi or ERmembranes (10 lg) isolated from rat liver for 30 min at378C with agitation. Following this incubation, the beadswere sedimented, resuspended in SDS sample buffer andboiled for 5 min.
Negative Staining for Electron Microscopy
The microtubules with or without tau prepared asdescribed above were examined by electron microscopy.Five microliters of the microtubule preparation wereplaced on a carbon-coated Formvar-supported EM grid.After an incubation of 30 s, the grid was rinsed with dis-tilled water and stained with 1% (w/v) uranyl acetate for30 s. Samples were visualized with a Zeiss CM 100transmission electron microscope.
RESULTS
The Distribution of Tau in Subcellular FractionsPrepared From Adult Rat Brain
In the present study, we further characterized theinteraction of tau with membranous elements. To do this,subfractions enriched in ER and Golgi membranes werepurified from adult rat brain using the fractionation pro-tocol described and illustrated in our previous studies[Lavoie et al., 1996; Farah et al., 2005]. These subfrac-tions were characterized by using different membranousmarkers: ribophorin and calnexin as ER markers, manno-sidase II, MG160, GM130, and TGN38 as Golgi markersand Na-K-ATPase as a marker of the plasma membrane.As shown in Fig. 1, the RM subfraction was enriched inER markers. This enrichment was confirmed by an elec-tron microscope morphometric analysis in our previousstudy [Farah et al., 2005]. The Golgi markers wereenriched in the subfractions termed I2 and I3 each of
Fig. 1. Immunoblot analysis of adult rat brain subfractions. Subfrac-
tions obtained following subcellular fractionation of adult rat brains
were electrophoresed on a 7.5% polyacrylamide gel (30 lg/lane) andtransferred to a nitrocelllulose membrane as described in methods. A
monoclonal antibody directed against tau was used (clone tau5) as
well as antibodies directed against the Golgi markers mannosidase,
MG160, GM130 and TGN38, the plasma membrane marker Na-K-
ATPase, the endoplasmic reticulum markers, ribophorin and calnexin,
tubulin and MAP2 (HM2). E, cytoplasmic extract; P, total membrane
extract; S, cytosolic fraction; I, interface; RM, rough microsomes.
Tau Links Golgi Membranes to Microtubules 713
these fractions presenting an enrichment of Golgi pro-teins belonging to a distinct compartment. Based onstaining intensity, MG160 and mannosidase II, twomarkers of the medial Golgi compartment were enrichedin the I2 subfraction [Velasco et al., 1993; Torre andSteward, 1996; Gonatas et al., 1998]. TGN38 wasenriched in both I2 and I3 subfractions. GM130, amarker of the cis-Golgi was found in all the subfractions(RM, S, I1, I2, and I3) generated by our fractionationprocedure. However, this marker was present at a higherconcentration in the I3 than in the I2 subfraction.
In a previous study, we showed that MAP2, thedendritic MAP that shares sequence homology with tauin the microtubule-binding domain was highly concen-trated in the RM subfraction enriched in ER membranes[Farah et al., 2005]. The anti-tau antibody, Tau-5, thatrecognizes all tau isoforms regardless of post-transla-tional modifications was used to examine the distributionof tau in the subfractions. As shown in Fig. 1, Tau wasenriched in the cytosolic fraction (S) as noted for tubulin.Moreover, a significant amount of tau (*15%) was pres-ent in the I2 subfraction (Fig. 1). In contrast to MAP2that was enriched in the RM subfraction, tau was barelydetectable in this subfraction in comparison to theamount found in the I2 subfraction. Previous studiesreported that tau was associated with the plasma mem-brane [Brandt et al., 1995]. In the present subcellularfractionation conditions, the Na-K-ATPase stainingrevealed the presence of plasma membranes in the I2subfraction containing tau as well as in the I3 subfrac-tion. This indicated that tau could be associated withplasma membranes in the I2 subfraction. However, thesimilar distribution of tau and the medial Golgi compart-ment markers, MG160 and mannosidase II, in the I2 sub-fraction suggested that tau could be associated not onlywith plasma membranes but also with Golgi membranesin this subfraction.
Localization of Tau in the I2 Subfraction
To confirm the association of tau with the Golgimembranes, immunogold labeling of tau was performedin the I2 subfraction. Both vesicles and tubules withassociated fenestrations reminiscent of the Golgi mem-branes were found in this subfraction (Fig. 2). Quantifi-cation of the gold particle distribution from three differ-ent sets of experiments revealed that 82.1% of tau stain-ing was associated with membranes. To confirmnonequivocally the association of tau with Golgi mem-branes in the I2 subfraction, double electron microscopeimmunocytochemistry was carried out using eitherTGN38 or MG160 as Golgi markers and an anti-tau anti-body (Fig. 3). Tau immunolabeling was found on mem-branes labeled either with MG160 or TGN38. A quanti-tative analysis revealed on average 32% (n ¼ 2) and
54% (n ¼ 2) of the membranous labeling of tau on mem-branes containing MG160 and TGN38 respectively. Therest of tau staining was mainly found on non-labeledmembranes. Thus, from these data one could concludethat tau was associated with membranes of the medialGolgi compartment and the TGN in the I2 subfraction.The association of tau with the Golgi membranesappeared to be independent of microtubules since nomicrotubule was observed in the vicinity of tau-labeledmembranes in this subfraction. This was consistent withthe fact that microtubules were very rarely observed inthe I2 subfraction.
Immunoisolation of Golgi Membranes Withan Anti-Tau Antibody in Adult Rat Brain
The association of tau with the Golgi membraneswas further demonstrated by using an immunoisolationtechnique previously described by Taguchi et al.2003. Inthese experiments, an anti-tau antibody coupled to mag-netic beads was used to immunoisolate Golgi membranesfrom a fraction of total microsomes prepared from adultrat brains by a standard method [Farah et al., 2003]. Thepresence of Golgi membranes in tau immunoisolate wasanalyzed by western blotting. As shown in Fig. 4a, theprotein MG160 was present in tau immunoprecipitateindicating that tau was associated with Golgi mem-branes. GM130 was also found in tau immunoisolate butat lower level than MG160. When the anti-tau antibodywas replaced by non-specific IgG, the level of MG160
Fig. 2. Electron microscopy analysis of the I2 subfraction. The I2
subfraction was recovered onto a Millipore membrane by the random
filtration technique developed by Baudhuin et al. [1967]. Arrowheads
point to short flattened membranes typical of Golgi membranes. The
arrows point to small vesicles (less than 100 nm diameter). Scale bar:
500 nm.
714 Farah et al.
and GM130 attached to the magnetic beads was barelydetectable. The ER marker, calnexin was detected atbackground level in tau immunoisolate showing that, inthe present experimental conditions, tau was not associ-ated with ER membranes.
To verify whether the association of tau with theGolgi membranes was peripheral, the I2 subfraction wastreated with sodium carbonate buffer at pH 11.0, a proce-dure known to solubilize peripheral but not integral mem-brane proteins [Wiedenmann et al., 1985]. Under thistreatment, tau was not found in the membrane pellet aswas the integral membrane protein MG160 indicating thatit has been solubilized. From these results, one could con-clude that tau was a peripheral Golgi protein (Fig. 4b).
Localization of Tau at the GA by ElectronMicroscope Immunocytochemistryon Motoneurons
The localization of tau at the GA was examined inmotoneurons of mouse spinal cord by electron micro-scope immunocytochemistry. These neurons have theadvantage of presenting a large GA with multiple stacks
of Golgi saccules which appear discontinuous around thenucleus [Peters et al., 1976]. No gold particles werefound when the primary antibody directed against tauwas omitted as well as when this primary antibody waspreabsorbed with pure tau protein. Some staining waspresent in the cytoplasm not associated with any struc-ture. Remarkably, tau staining was observed on theGolgi membranes and appeared to be more concentratedto one side of the GA (Fig. 5). To confirm this, a quanti-tative analysis of the number of gold particles per micro-meter square of the different Golgi compartments, cis,medial and trans was carried out. Fenestration was usedto identify the cis side of the GA and clathrin-coatedvesicles to identify the trans Golgi side [Peters et al.,1976; Rambourg and Clermont, 1990; Marsh et al.,2001]. Moreover, on the trans side, ER adherent to thetrans-most cisternae (GERL) was observed as noted inmany mammalian secretory cells [Mogelsvang et al.,2004]. For the quantitative analysis, a total of 35 GA thatpresented a morphology allowing to distinguish the cisand trans side were selected from two animals. The sur-face of each Golgi compartment was directly measured
Fig. 3. Immunogold labeling of tau in the I2 subfraction. The I2 sub-
fraction was double-labeled with a monoclonal antibody directed
against tau (49.2) and a polyclonal antibody directed against either
TGN38 or MG160. A, B, and C) The monoclonal anti-tau antibody
49.2 and a polyclonal antibody directed against TGN38 were revealed
using a secondary anti-mouse antibody conjugated to 10 nm colloidal
gold particles and a secondary anti-sheep antibody conjugated to 5 nm
colloidal gold particles respectively. Tau and TGN38 were found on
the same short flattened membranes and vesicles. D, E, and F) The
monoclonal anti-tau antibody 49.2 and a polyclonal antibody directed
against MG160 were revealed using a secondary anti-mouse antibody
conjugated to 10 nm colloidal gold particles and a secondary anti-
rabbit antibody conjugated to 5 nm colloidal gold particles respec-
tively. Tau and MG160 were found on the same short flattened mem-
branes and vesicles. Scale bar: A, B, D, and E ¼ 150 nm; C and F ¼100 nm.
Tau Links Golgi Membranes to Microtubules 715
on enlarged electron negatives using the image analysissystem Northern Eclipse. Then, the number of gold par-ticles found in these compartments was counted (goldparticles/lm2). The quantitative analysis revealed that
14.3 6 2.1, 22.6 6 2.4, and 36.7 6 3.5 gold particles/lm2 were found on the cis, medial, and trans Golgi com-partments, respectively (Fig. 6). The number of gold par-ticles per square micrometer of Golgi membranes wassignificantly higher on the trans Golgi compartment thanthat on the cis and medial compartments.
Golgi Membranes Were Found in Close Vicinityto Tau-Containing Microtubules in PrimaryHippocampal Neurons
In a previous study, we showed that the overex-pression of human tau in primary hippocampal neuronsinduced a fragmentation of the GA [Liazoghli et al.,2005]. This event seems to occur in several steps. Inthree-day-old neurons, GA normally appears as a juxta-nuclear compact and clustered structure as revealed withan antibody directed against GM130 (Fig. 7). In tau-transfected neurons, the GA firstly becomes dispersed inlarge vesicles and then these vesicles are fragmented innumerous, small, round, and disconnected Golgi ele-ments distributed in the cell body [Liazoghli et al.,2005]. In hippocampal neurons, transfected human tauwas found in the cell body as well as in neurites (Fig. 7,GFP-Tau/GM130). The overexpression of human tau inhippocampal neurons induced the assembly of microtu-bule bundles as visualized with an anti-tubulin antibodythat co-localized with GFP-tau (Fig. 7, GFP-Tau/Tubu-lin). These bundles often emerged from the cell bodyand formed thin processes. In tau-transfected neuronspresenting a dispersion of the GA, large Golgi vesicleswere found in close vicinity to tau-containing microtu-bule bundles (Fig. 7, GM130/Tubulin). These resultssuggested that transfected human tau could induce a dis-persion of the GA by establishing a link between the
Fig. 4. Immunoisolation of Golgi membranes containing MG160
and GM130 with an anti-tau antibody. (A) In the left lane (Non-
specific IgG), the magnetic beads were incubated with a fraction of
total microsomes prepared from adult rat brain to detect any non-spe-
cific binding of membranes to the beads. In the right lane (Tau-anti-
body), the preparation of total microsomes was incubated with the
beads coupled to an anti-tau antibody (polyclonal anti-tau antibody or
Tau-1 antibody) for 2 h at 48C. Then, the beads were sedimented with a
magnet and washed three times. The proteins bound to the beads were
analyzed by SDS-PAGE. Membranes containing MG160 and GM130
were found attached to the magnetic beads when the anti-tau antibody
was added to the total microsome fraction. On the other hand, calnexin
was only detectable at the background level indicating that, in the pres-
ent experimental conditions, tau was not associated with ER mem-
branes. (B) Western blot analysis of the I2 fraction before and following
washing with sodium carbonate pH 11.0 as described in methods. No
detectable amount of tau was found in the I2 subfraction following the
washing step suggesting that the association of tau with the golgi mem-
branes is peripheral. As indicated by the MG160 staining, membrane re-
covery was not affected by sodium carbonate treatment.
716 Farah et al.
Golgi membranes and the newly formed microtubulebundles.
Tau: A Linker Between Golgi Membranesand Microtubules
We next used an in vitro reconstitution assay toinvestigate whether tau could act as a linker betweenGolgi membranes and microtubules. Membranesdeprived of tau were necessary to carry out this assay.To this purpose, ER and Golgi membranes were purified
from rat liver where tau is not detectable using a subcel-lular fractionation procedure previously described[Lavoie et al., 1996]. The ER subfraction was enrichedwith the ER marker, calnexin whereas the Golgi subfrac-tion contained the Golgi markers, GM130 and MG160 aswell as the ER marker, calnexin (Fig. 8a, lanes 1 and 2).We first verified whether tau could interact with thesemembranes. The anti-tau antibody, Tau-5 directedagainst bovine tau, was attached to magnetic beadsbound to an anti-mouse antibody. Then, bovine tau
Fig. 6. Quantitative analysis of tau
immunogold on the GA membranes.
The surface of each Golgi compart-
ment (cis, medial, and trans) was
measured using the image analysis
system Northern Eclipse and then
the number of Gold particles per
lm2 of membrane was calculated.
This analysis revealed that tau was
more concentrated on the trans side
of the GA than on the cis side.
Fig. 5. Immunogold labelling of tau on a section of a mouse spinal cord. Sections of mouse spinal cord
were processed for postembedding immunocytochemistry with a polyclonal antibody directed against
tau. Tau immunogold labeling was observed on the Golgi membranes and appeared to be more concen-
trated at the trans side of GA. Fenestration was used to identify the cis side of the GA and clathrin-coated
vesicles to identify the trans side. The borders of cis, medial and trans/GELR Golgi compartments were
defined by discontinuous lines. Scale bar, 500 nm.
Tau Links Golgi Membranes to Microtubules 717
Fig.7.
DistributionofGolgimem
branes
inprimaryhippocampal
neuronsoverexpressinghuman
tau.Primaryhippocampal
neuronsofthree-day-old
transfectedwithhuman
GFP-tau
weredouble-
stained
withan
anti-G
M130(red)andan
anti-tubulin(blue)
antibody.GFP-tau
was
presentin
thecellbodyandneuritesandcolocalizedwithtubulinstaining(G
FP-tau/Tubulin).Human
tauinducedthe
form
ationofmicrotubule
bundlesin
hippocampal
neurons.TransfectedneuronspresentedaGA
dispersedin
largevesicleswhereasGA
appearedas
acompactandclustered
structure
inuntransfected
neurons.ThelargeGolgivesicleswereoften
foundin
close
vicinityto
tau-containingmicrotubules(G
M130/Tubulin,inset,whitearrows).Scalebar,20lm
.
obtained commercially was incubated with the beads.Following this incubation, either Golgi or ER mem-branes isolated from rat liver were added to the beadscontaining tau. The beads were sedimented and boiled inSDS sample buffer. The specific binding of liver mem-branes to tau was examined by western blotting. Whenthe beads containing tau were incubated with the liverGolgi membranes, an enrichment of GM130 was notedindicating an interaction of tau with these membranes(Fig. 8a, lanes 3 and 4). On the other hand, when thebeads containing tau were incubated with the liver ERmembranes, no enrichment of calnexin was noted (Fig.8a, lanes 5 and 6). These results indicated that tau prefer-entially associated with Golgi membranes. We previ-ously showed that MAP2 binds to ER membranes [Farahet al., 2005]. Thus, MAP2 was used as a control to fur-ther test the specific binding of neuronal MAPs to livermembranes. In this case, the anti-MAP2 antibody, HM2,was attached to magnetic beads. Bovine MAP2 was ob-tained from a brain MAP fraction purchased from Cyto-skeleton (Denver, Colorado). MAP2-containing beadswere incubated with either Golgi or ER membranes iso-lated from rat liver. In contrast to tau, MAP2 was foundto be preferentially associated with ER membranes (Fig.8a, lanes 7 and 8). No binding of MAP2 to Golgi mem-branes isolated from rat liver was observed (data notshown). The above results confirmed that Golgi mem-branes purified from rat liver could be used to investigatethe role of tau as a linker between Golgi membranes andmicrotubules.
The in vitro reconstitution assay of the binding ofGolgi membranes to tau-containing microtubules wasperformed in two steps. First, microtubules that did notcontain or did contain tau were produced. The formationof microtubules was monitored by negative staining forelectron microscopy (Fig. 8b). The microtubules con-tained biotinylated tubulin (1 biotin-tubulin:20 non-bio-tin-tubulin) and therefore were attached to magneticbeads bound to streptavidin. Second, the microtubulesbound to the beads were incubated with either the Golgior ER subfraction purified from rat liver. Then, the beadswere sedimented and boiled in SDS sample buffer. Thebinding of membranes to microtubules was examined bywestern blotting (Fig. 8c). To make sure that an increaseof Golgi membrane binding to tau-containing microtu-bules was not merely caused by a higher amount ofmicrotubules in these preparations, the amount of tubulinattached to the magnetic beads was analyzed by immu-noblotting. A similar amount of tubulin was present inall our microtubule preparations (with or without tau)(Fig. 8c). The binding of Golgi membranes to microtu-bules was enhanced by tau as shown by the enrichmentof the Golgi markers, MG160 and GM130, but not thatof the ER marker, calnexin on microtubules containingtau (Fig. 8c, Golgi subfraction). Consistently, when
microtubules were incubated with the ER subfraction, noenrichment of calnexin was noted on microtubules con-taining tau (Fig. 8c, ER subfraction). These results werequantified by densitometry (Fig. 8d). The experimentswere repeated three times. To quantify the increase ofGolgi membrane binding to microtubules by tau, the ra-tio of the amount of Golgi markers found on microtu-bules that did contain and did not contain tau (Tau-Mts/Mts) was calculated (Fig. 8d). Data presents mean 6 SDof three experiments. In the case of calnexin, this ratiowas 1.11 6 0.11 indicating that tau faintly increased thebinding of ER membranes to microtubules. On the otherhand, this ratio was 6.61 6 2.33 and 4.01 6 1.14 for theGolgi markers GM130 and MG160 respectively, reveal-ing an important increase of the Golgi membrane bind-ing to microtubules in the presence of tau. From theabove experiments, one can conclude that tau can linkthe Golgi membranes to microtubules.
Association of Phosphorylated Tau With GolgiMembranes
It was previously shown that the pool of tau foundat the plasma membrane was not phosphorylated at thesites recognized by the phosphorylation-sensitive anti-bodies PHF-1, AT-8, and AT-180 and was dephospho-rylated at the Tau-1 antibody sites in PC12 cells [Maaset al., 2000]. The presence of phosphorylated tau at thesites recognized by the phosphorylation-sensitive anti-bodies PHF-1 and AT-8 was shown in normal adult ratbrain by western blot [Jicha et al., 1999]. To verifywhether phosphorylated tau was found on the Golgimembranes in adult rat brain, we investigated the phos-phorylation state of tau in the I2 Golgi subfraction. Incontrast to the pool of tau found at the plasma mem-brane, tau present in the I2 subfraction was phosphoryl-ated at the sites recognized by the antibody PHF-1(Ser396/Ser404) (Fig. 9). Four tau isoforms (45–65 kDa)were distinguishable with the antibody 49.2 in adult ratbrain I2 subfraction. A similar pattern was observed withPHF-1 antibody in this subfraction. The AT-8 antibodyonly recognized the tau isoform presenting the highestmolecular mass in the I2 subfraction indicating thatSer202 and Thr205 were phosphorylated in this isoform.In contrast, the antibody Tau-1 that recognizes tau de-phosphorylated at the region extending from 187 to 205residues moderately stained the top tau isoform butstrongly immunolabeled the other tau isoforms [Bueeet al., 2000]. Immunoreactive tau to PHF-1 is most likelyassociated with Golgi membranes in the I2 subfraction.This subfraction also contains membranes originatingfrom the plasma membrane and the ER. According to aprevious study, tau phosphorylated at the sites of PHF-1antibody was not found at the plasma membrane and inthe present study, we showed that very little tau is found
Tau Links Golgi Membranes to Microtubules 719
Figure
8.
in the RM subfraction enriched in ER membranes indi-cating that tau does not preferentially interact with thesemembranes. From the above results, one can concludethat different pools of phosphorylated tau exist within aneuron and these pools are most likely associated withdifferent membranous compartments. Finally, tau pres-ent in the cytosolic fraction (S100) was immunoreactiveto PHF-1 and tau-1 antibodies but not to AT-8 (Fig. 9).
DISCUSSION
In the present study, we demonstrated that tau wasassociated with the Golgi membranes. Electron micro-scope immunocytochemistry using double-labeling con-firmed that tau was associated with membranes labeledwith the Golgi markers GM160 and TGN38. The interac-tion of tau Golgi membranes was further confirmedby immunoisolating Golgi membranes with an anti-tauantibody. Electron microscope immunocytochemistryrevealed that tau association with the Golgi membraneswas found in all Golgi compartments but increased to-ward the trans side. We showed that tau could serve as alinker between the Golgi membranes and microtubulesby using an in vitro reconstitution assay. Finally, thepool of tau present in the Golgi fraction was phosphoryl-ated at the sites recognized by the phosphorylation-sensi-tive antibodies PHF-1 and AT-8.
Our data indicate that tau could mediate the inter-action of Golgi membranes with microtubules within aneuron. In this context, tau is not the sole microtubule-binding protein that could play such a role. Indeed, othermicrotubule-associated proteins including SCG10,GMAP-210, CLIPR-59, and Hook3, are also associatedwith the Golgi membranes [Bloom and Brashear, 1989;Lutjens et al., 2000; Walenta et al., 2001; Lallemand-Breitenbach et al., 2004; Rios et al., 2004]. As noted fortau in the present study, their association with the GAdoes not depend on microtubules. However, these pro-teins differ from each other and from tau on severalaspects. First, they do not share any sequence homologyin their binding domain to either Golgi membranes ormicrotubules. Second, they distinctly interact withmicrotubules and exert diverse effects on microtubulepolymerization [Bloom and Brashear, 1989; Lutjenset al., 2000; Lallemand-Breitenbach et al., 2004; Rioset al., 2004]. Third, they present a distinct distribution onthe Golgi membranes. Hook3 and GMAP-120 arelocated on the cis-Golgi side whereas CLIPR-59 andSCG10 are found on the trans side. In the case of tau,
Fig. 9. Immunoblot analysis of the phosphorylation state of tau in
the I2 subfraction. Fractions obtained following subcellular fractiona-
tion of adult rat brains were electrophoresed on a 7.5% polyacryl-
amide gel (15 lg/lane) and transferred to a nitrocelllulose membrane
as described in methods. The subcellular fractions were analysed by
using the tau phosphorylation-sensitive antibodies PHF-1 and AT-8
and the antibody Tau-1 directed against unphosphorylated tau. A
monoclonal antibody directed against tau (clone 49.2) was used to
reveal total tau. E, cytoplasmic extract; P, total membrane extract; S,
cytosolic fraction; I, interface; RM, rough microsomes.
Fig. 8. In vitro reconstitution assay of Golgi membrane binding to
tau-containing microtubules. (A) The ER (lane 1) and Golgi (lane 2)
subfractions isolated from rat liver were analyzed by immunoblotting
with an anti-calnexin antibody, an anti-GM130 antibody and an anti-
MG160 antibody. In the ER subfraction, a band corresponding to cal-
nexin (90 kDa) was noted. In the Golgi subfraction, a band was detected
with the three antibodies indicating that this subfraction contained both
ER and Golgi membranes. When tau-bound magnetic beads (lane 4)
were incubated with the Golgi subfraction, the level of GM130 staining
was higher than that of non-specific protein binding to unbound beads
(lane 3). Lanes 3 and 4 were also revealed with an anti-tau antibody to
show the presence of tau isoforms around 50 kDa in lane 4. No increase
of calnexin staining was detected when tau-bound beads (lane 6) were
incubated with the ER subfraction compared to unbound beads (lane 5).
In lane 6, 5 bovine tau isoforms are detected around 50 kDa. In lane 8,
an increase of calnexin staining was noted when MAP2-bound beads
were incubated with the ER subfraction compared to unbound beads
(lane 7). (B) Tau-bound and unbound microtubules were negatively
stained for electron microscopy to confirm the presence of microtu-
bules. Scale bar, 100 nM. (C) Tau-bound and unbound microtubules
were incubated with the Golgi subfraction isolated from rat liver. A
clear enrichment of the Golgi markers, MG160 and GM130 was noted
on tau-bound microtubules compared to unbound microtubules. No
enrichment of the ER calnexin marker was detected. A similar result
was obtained when tau-bound and unbound microtubules were incu-
bated with the ER subfraction. (D) Quantitative analysis of the mem-
brane binding to tau-bound microtubules and unbound microtubules.
Mean6 SD of three experiments is presented.
Tau Links Golgi Membranes to Microtubules 721
our results indicate that its concentration on the Golgimembranes increases towards the trans side of the Golgicomplex. Although it is believed that SCG10, GMAP-210, CLIPR-59, and Hook3 mediate the association ofthe Golgi membranes with microtubules and contributeto the positioning and morphology of the GA within acell, the respective function of these proteins on theGolgi membranes remains to be elucidated except forGMAP-210. GMAP-210 is found on cis-Golgi mem-branes and binds the minus end of microtubules located atthe centrosomes. In a recent study, it was shown thatGMAP-210 recruits g-tubulin complexes to the cis-Golgimembranes and this event is necessary for the pericentrio-lar position of the GA [Rios et al., 2004]. Alpha- and beta-tubulin were also found to be attached to the Golgi mem-branes [Yamaguchi and Fukada, 1995]. The proteinsinvolved in recruiting these tubulin isoforms to the Golgimembranes have not been identified yet. Our present dataindicates that tau could play such a role within a neuron.
A recent study showed that the overexpression oftau in astrocytes induced a fragmentation of the GA[Yoshiyama et al., 2003]. We have also observed a frag-mentation of the GA in primary hippocampal neuronstransfected with human tau forms and in the JNPL3 miceexpressing the mutant human form of tau P301L [Lia-zoghli et al., 2005]. An ultrastructural study also reporteda fragmentation of the GA in JNPL3 mice [Lin et al.,2003]. The mechanisms involved in this fragmentationof the GA by tau remain elusive. However, in previousstudies, a fragmentation of the GA was consistentlyobserved when the link between microtubules and theGA was perturbed [Lucocq and Warren, 1987; Lucocqet al., 1987; Cole et al., 1996]. For example, depolymer-ization of microtubules by nocodazole or colchicineresulted in a fragmentation of the GA [Boyd et al.,1982]. Overexpression of Hook3 or GMAP-210, twoproteins that mediate the interaction between microtu-bules and the Golgi membranes, also induced a disper-sion and fragmentation of the GA [Walenta et al., 2001;Pernet-Gallay et al., 2002]. The fragmentation of the GAinduced by the overexpression of tau in hippocampalneurons seems to occur in several steps that mightdepend or not on microtubules. Our present results indi-cate that the initial step characterized by the dispersionof Golgi membranes in large vesicles might depend onthe interaction of these membranes with microtubules.Tau would mediate such interaction. The subsequentfragmentation of the large Golgi vesicles in small struc-tures would involve the alteration of membrane transportand/or intracellular signalisation.
Most notably, tau was not uniformly distributed onthe Golgi membranes but an increase of its concentrationwas noted towards the trans side of the GA. The micro-tubule-destabilizing protein, SCG10 which is neuronal
specific is also enriched on the trans side of the GA [Lut-jens et al., 2000]. Interestingly, the association of SCG10to the Golgi membranes was shown to be necessary forits targeting to the growth cones. Thus, the association oftau to the GA might be involved in its axonal sorting. Onthe other hand, tau might serve as a tag on vesiclesemerging from the GA destined to the axonal process.Consistent with this possibility, an important percentageof tau immunogold labeling was found on small vesiclesin our subcellular fraction enriched in Golgi membranes.The trafficking pathways of the biosynthesis of integralmembrane proteins are well characterized. Their synthe-sis occurs in the rough endoplasmic reticulum and thenthey go through the Golgi complex where they are pack-aged into carrier vesicles that will be transported eitherto distinct cellular compartments to deliver their contentsto the plasma membrane [Van Vliet et al., 2003]. Sortingof vesicles destined to different cellular compartmentsoccurs at the TGN [Gleeson et al., 2004; Sytnyk et al.,2004]. Two mechanisms seem to regulate the targetingof integral membrane proteins to the axon [Sampo et al.,2003]. Some proteins are targeted both to the somato-dendritic and axonal compartment but are only retainedat the plasma membrane of the axon whereas other pro-teins are selectively delivered to the axonal plasma mem-brane. In this latter case, tau could act as an axonal local-ization signal for the delivery of the cargo proteins to theaxonal plasma membrane. Moreover, tau could remainassociated with the cargo proteins after their integrationto the plasma membrane since tau was found associatedwith this membrane.
As noted for the binding of tau to microtubules, theassociation of tau with membranes seems to be regulatedby its phosphorylation state. Indeed, the phosphorylationof tau abolished its binding to the plasma membrane inPC12 cells [Maas et al., 2000]. The pool of tau found inthe Golgi enriched subcellular fraction was phosphoryl-ated at the sites recognized by the phospho-dependentantibody PHF-1 whereas the pool of tau found at theplasma membrane was not [Maas et al., 2000]. Theamount of immuroreactive tau to AT8 antibody washigher in the I2 subfraction than in the cytosolic subfrac-tion (S100). In contrast, AT8 immunoreactive tau wasmainly in the cytosolic fraction in PC12 cells. This indi-cates that the distribution of phosphorylated tau couldalso vary from one cell type to another. PHF-1 immuno-reactive tau was detected in lipid rafts in transgenic miceoverexpressing human amyloid precursor proteinAPP695 with the ‘‘Swedish’’ mutation and in AD brain[Kawarabayashi et al., 2004]. A gradient of tau phospho-rylation was observed along the axon in primary hippo-campal cultures [Mandell and Banker, 1996]. From thesedata, it appears that different pools of phosphorylated tauexist along the axon which could preferentially interact
722 Farah et al.
with either the Golgi membranes or the plasma mem-brane. In degenerating neurons, tau becomes hyperphos-phorylated [Lee et al., 2001]. Therefore, an aberrantphosphorylation of tau might result in an alteration of itsassociation with membranes and this might contribute toits dysfunction.
The detection of tau on the GA by immunogoldlabeling showed that tau is present in the neuronal cellbody. Although most of the studies state that tau is anaxonal MAP, several observations suggest that tau is alsolocated in the cell body and dendrites. In previous stud-ies, tau was found in the somato-dendritic compartmentin adult rat brain [Papasozomenos and Binder, 1987].However, its phosphorylation state differed from thatin the axon. Similar results were reported in primaryhippocampal neurons [Mandell and Banker, 1996]. InDrosophila, tau is also distributed in all neuronal com-partments in the adult retina [Heidary and Fortini, 2001].The above observations indicate that different pools oftau exist within a neuron that present a different state ofphosphorylation and subcellular distribution. The func-tion of each of these pools remains to be elucidated. Thisis a crucial step to better understand the implication oftau in neuronal function and degeneration.
CONCLUSIONS
Our results show that tau is associated with Golgimembranes and can serve as a linker between thesemembranes and microtubules. Such association mightcontribute to the selective targeting of membranousprotein complexes to the axon.
ACKNOWLEDGMENTS
The authors would like to thank Dr. M. G. Farquharfor providing the anti-mannosidase II antibody, Dr. G.Kreibich for the anti-ribophorin antibody, Dr. N. Gonatasfor the anti-MG160 antibody, Dr. D. Fambrough for theNaK-ATPase antibody, Dr. Virginia Lee for the polyclonaland monoclonal anti-tau antibodies, and Dr. Peter Daviesfor the antibody PHF-1. The monoclonal antibodyE7 directed against b-tubulin developed by MichaelKlymkowsky was obtained from the Developmental Stud-ies Hybridoma Bank developed under the auspices of theNICHD and maintained by The University of Iowa,Department of Biological Sciences, Iowa City, IA 52242.We also thank Jean Leveille, Annie Vallee, and MirelaPascariu for their excellent technical support and Dr.Moise Bendayan and Diane Gingras for helpful discussion.N.L. is a scholar of Fonds de la recherche en sante du Que-bec (FRSQ) and C.A.F. and S.P. have a studentship fromCRSN and M.D. a studentship from NSERC.
REFERENCES
Banker G, Goslin K. 1998. Culturing Nerve Cells: A Bradford Book.
Cambridge, MA: MIT Press. 666 p.
Baudhuin P, Evrard P, Berthet J. 1967. Electron microscopic examina-
tion of subcellular fractions. I. The preparation of representa-
tive samples from suspensions of particles. J Cell Biol
32(1):181–191.
Bloom GS, Brashear TA. 1989. A novel 58-kDa protein associates
with the Golgi apparatus and microtubules. J Biol Chem
264(27):16083–16092.
Boyd A, Bolton W, Brinkley B. 1982. Microtubules and b cell func-
tion: Effect of colchicine on microtubules and insulin secretion
in vitro by mouse b cells. J Cell Biol 92:425–434.
Brandt R, Leger J, Lee G. 1995. Interaction of s with the neural
plasma membrane mediated by s’s amino-terminal projection
domain. J Cell Biol 131:1327–1340.
Buee L, Bussiere T, Buee Scherrer V, Delacourte A, Hof PR. 2000. sprotein isoforms, phosphorylation, and role in neurodegenera-
tive disorders. Brain Res Brain Res Rev 33(1):95–130.
Caceres A, Potrebic S, Kosik K. 1991. The effect of s antisense oligo-nucleotides on neurite formation of cultured cerebellar macro-
neurons. J Neurosci 11(6):1515–1523.
Caceres A, Mautino J, Kosik K. 1992. Suppression of MAP2 in cul-
tured cerebellar macroneurons inhibits minor neurite forma-
tion. Neuron 9(4):607–618.
Cole N, Sciaky N, Marotta A, Song J, Lippincott-Schwartz J. 1996.
Golgi dispersal during microtubule disruption: regeneration of
Golgi stacks at peripheral endoplasmic reticulum exit sites.
Mol Biol Cell 7:631–650.
Dawson H, Ferreira A, Eyster M, Ghoshal N, Binder L, Vitek M. 2001.
Inhibition of neuronal maturation in primary hippocampal neu-
rons from s deficient mice. J Cell Sci 114(6):1179–1187.
Drewes G, Ebneth A, Mandelkow E. 1998. MAPs, MARKs, and
microtubule dynamics. Trends Biochem Sci 23(8):307–311.
Farah C, Nguyen M, Julien J, Leclerc N. 2003. Altered levels and dis-
tribution of microtubule-associated proteins before disease
onset in a mouse model of amyotrophic lateral sclerosis. J Neu-
rochem 84(1):77–86.
Farah C, Liazoghli D, Perreault S, Desjardins M, Guimont A, Anton
A, Lauzon M, Kreibich G, Paiement J, Leclerc N. 2005. Inter-
action of microtubule-associated protein-2 and p63: A new link
between microtubules and rough endoplasmic reticulum mem-
branes in neurons. J Biol Chem 280(10):9439–9449.
Garcia ML, Cleveland DW. 2001. Going new places using an old
MAP: s, microtubules and human neurodegenerative disease.
Curr Opin Cell Biol 13(1):41–48.
Gleeson P, Lock J, Luke M, Stow J. 2004. Domains of the TGN:
Coats, tethers and G proteins. Traffic 5:315–326.
Gonatas J, Chen Y, Stieber A, Mourelatos Z, Gonatas N. 1998. Trunca-
tions of the C-terminal cytoplasmic domain ofMG160, a medial
Golgi sialoglycoprotein, result in its partial transport to the
plasma membrane and filopodia. J Cell Sci 111:249–260.
Heidary G, Fortini M. 2001. Identification and characterization of the
Drosophila s homolog. Mech Dev 108:171–178.
Heins S, Song Y, Wille H, Mandelkow E, Mandelkow E. 1991. Effect
of MAP2, MAP2c and s on kinesin-dependent microtubule
motility. J Cell Sci Suppl 14:121–124.
Jicha G, Weaver C, Lane E, Vianna C, Kress Y, Rockwood J, Davies
P. 1999. cAMP-dependent protein kinase phosphorylations on
s in Alzheimer’s disease. J Neurosci 19(17):7486–7494.
Jung D, Filliol D, Miehe M, Rendon A. 1993. Interaction of brain
mitochondria with microtubules reconstituted from brain
tubulin and MAP2 or s. Cell Motil Cytoskeleton 24(4):245–
255.
Tau Links Golgi Membranes to Microtubules 723
Kawarabayashi T, Shoji M, Youkin L, Wen-Lang L, Dickson D,
Murakami T, Matsubara E, Abe K, Ashe K, Younkin S. 2004.
Dimeric amyloid B protein rapidly accumulates in lipid rafts
followed by apoliprotein E and phosphorylated s accumulation
in the Tg2576 mouse model of Alzheimer’s disease. J Neurosci
24(15):3801–3809.
Lallemand-Breitenbach V, Quensnoit M, Braun V, El Marjou A, Pous
C, Goud B, Perez F. 2004. CLIPR-59 is a lipid raft-associated
protein containing a cytoskeleton-associated protein glycine-
rich domain (CAP-Gly) that perturbs microtubule dynamics.
J Biol Chem 279(39):41168–41178.
Lavoie C, Lanoix J, Kan FW, Paiement J. 1996. Cell-free assembly of
rough and smooth endoplasmic reticulum. J Cell Sci 109 (Part
6):1415–1425.
Lee G, Newman S, Gard D, Band H, Panchamoorthy G. 1998. s inter-acts with src-family non-receptor tyrosine kinases. J Cell Sci
111 (Part 21):3167–3177.
Lee V, Goedert M, Trojanowski J. 2001. Neurodegenerative tauopa-
thies. Annu Rev Neurosci 24:1121–1159.
Liazoghli D, Perreault S, Micheva K, Desjardins M, Leclerc N. 2005.
Fragmentation of the Golgi apparatus induced by the overex-
pression of wild-type and mutant human s forms in neurons.
Am J Pathol 166(5):1499–1514.
Lin W, Lewis J, Yen S, Hutton M, Dickson D. 2003. Ultrastructural
neuronal pathology in transgenic mice expressing mutant
(P301L) human tau. J Neurocytol 32:1091–1105.
Lucocq J, Pryde JG, Berger EG, Warren G. 1987. A mitotic form of
the Golgi apparatus in Hela cells. J Cell Biol 104(4):865–874.
Lucocq J, Warren G. 1987. Fragmentation and partitioning of the
Golgi apparatus during mitosis in Hela cells. EMBO J
6(11):3239–3246.
Lutjens R, Igarashi M, Pellier V, Blasey H, Di Paolo G, Ruchti E,
Pfulg C, Staple J, Catsicas S, Grenningloh G. 2000. Localiza-
tion and targeting of SCG10 to the trans-Golgi apparatus and
growth cone vesicles. Eur J Neurosci 12:2224–2234.
Maas T, Eidenmller J, Brandt R. 2000. Interaction of s with the neural
membrane cortex is regulated by phosphorylation at sites that
are modified in paired helical filaments. J Biol Chem
275(21):15733–15740.
Mandell J, Banker G. 1996. Microtubule-associated proteins, phos-
phorylation gradients, and the establishment of neuronal polar-
ity. Perspect Dev Neurobiol 4(2/3):125–135.
Marsh B, Mastonarde D, Buttle K, Howell K, McIntosh R. 2001.
Organellar relationships in the Golgi region of the pancreatic bcell line, HIT-115, visualized by high resolution electron to-
mography. Proc Natl Acad Sci USA 98:2399–2406.
Micheva K, Vallee A, Beaulieu C, Herman I, Leclerc N. 1998. bActinis confined to structures having high capacity of remodelling in
developing and adult rat cerebellum. Eur J Neurosci 10(12):
3785–3798.
Mogelsvang S, Marsh B, Ladeinsky M, Howell K. 2004. Predicting
function from structure: 3D structure studies of the mammalian
Golgi complex. Traffic 5:338–345.
Papasozomenos S, Binder L. 1987. Phosphorylation determines two
distinct species of s in the central nervous system. Cell Motil
Cytoskeleton 8:210–226.
Pernet-Gallay K, Antony C, Johannes L, Bornens M, Goud B, Rios R.
2002. The overexpression of GMAP-210 blocks anterograde
and retrograde transport between the ER and the Golgi appara-
tus. Traffic 3:822–832.
Peters A, Palay S, Webster H. 1976. The Fine Structure of the Nerv-
ous System: The Neurons and Supporting Cells. Philadelphia:
Saunders. 406 p.
Rambourg A, Clermont Y. 1990. Three-dimensional electron micros-
copy: Structure of the Golgi apparatus. Eur J Cell Biol 51:189–
200.
Rios R, Sanchis A, Tassin A, Fedriani C, Bornens M. 2004.
GMAP-210 recruits g-tubulin complexes to cis-Golgi mem-
branes and is required for Golgi ribbon formation. Cell
118:323–335.
Sampo B, Kaech S, Kunz S, Banker G. 2003. Two distinct mecha-
nisms target membrane proteins to the axonal surface. Neuron
37:611–624.
Stamer K, et al. 2002. s blocks traffic of organelles, neurofilaments
and APP vesicles in neurons and enhances oxidative stress.
J Cell Biol 156(6):1051–1063.
Sytnyk V, Leshchyns’ka I, Ditayatev A, Schachner M. 2004. Trans-
Golgi network delivery of synaptic proteins in synaptogenesis.
J Cell Sci 117:381–88.
Taguchi T, Pypaert M, Warren G. 2003. Biochemical sub-fractio-
nation of the mammalian Golgi apparatus. Traffic 4:344–
352.
Torre E, Steward O. 1996. Protein synthesis within dendrites: Glyco-
sylation of newly synthesized proteins in dendrites of hippo-
campal neurons in culture. J Neurosci 16:5967–5978.
Van Vliet C, et al. 2003. Intracellular sorting and transport of pro-
teins. Prog Biophys Mol Biol 83(1):1–45.
Velasco A, Hendricks L, Moremen K, Tulsani D, Touster O, Farquhar
M. 1993. Cell type-dependent variations in the subcellular
distribution of a-mannosidase I and II. J Cell Biol 122(1):
39–51.
Walenta J, Didier A, Liu X, Kramer H. 2001. The Golgi-associated
Hook3 protein is a member of a novel family of microtubule-
associated proteins. J Cell Biol 152:923–934.
Wiedenmann B, Lawley K, Grund C, Branton D. 1985. Solubilization
of proteins from bovine brain coated vesicles by protein pertur-
bants and triton X-100. J Cell Biol 101:12–18.
Yamaguchi N, Fukada M. 1995. Golgi retention mechanism of b-1–4-galactosyltransferase. Membrane-spanning domain-dependent
homodimerization and association with a- and b-tubulins.J Biol Chem 270:12170–12176.
Yoshiyama Y, Zhang B, Bruce J, Trojanowski J, Lee V. 2003. Reduc-
tion of detyrosinated microtubules and Golgi fragmentation are
linked to s-induced degeneration in astrocytes. J Neurosci
23:10662–10671.
724 Farah et al.
