The Effect of Tau Antisense Oligonucleotides on Neurite Formation of Cultured Cerebellar Macroneurons

The Journal of Neuroscience, June 1991, 1 I(6): 1515-l 523

Alfred0 Caceres, Sonja Potrebic, and Kenneth S. Kosik Department of Neurology (Neuroscience), Harvard Medical School, and Center for Neurologic Diseases, Department of Medicine (Division of Neurology), Brigham and Women’s Hospital, Boston, Massachusetts 02115

Tau, a microtubule-associated protein (MAP) enriched in ax- by the direction that they move organellesin relation to the ons, may have a role in the generation and maintenance of asymmetry of the microtubule, the contrasting orientations of an axonal morphology. Neurons from embryonic day 15 rat microtubules within axons and dendrites could give rise to the cerebellum in culture elaborate two morphologically distinct distinct subcellular components that contribute to the ultra- neurite populations-one with nontapering, elongated ax- structural identity of thesestructures (Black and Baas, 1989). onlike neurites and the other with tapered dendritelike neu- When neuronsfrom various siteswithin the developing ner- rites that branch frequently and are selectively stained with vous system are cultured, they elaborate net&es, which may antibodies to MAPP. Tau antisense oligonucleotides were over time approximate someaspects oftheir in situ morphology. utilized in two ways: (1) continuous application of antisense Neurites from several neuronal culture systems,including hip- every 24 hr for variable periods of time or (2) application of pocampus,cerebellum, sympathetic ganglia,and cerebrocortex, antisense that was delayed until neurite differentiation was develop the morphological characteristics of axons and den- underway. In both cases, 24 hr after the administration of drites (Bartlett and Banker, 1984; Peng et al., 1986; Kosik and the antisense, tau protein was not detected immunocyto- Finch, 1987; Ferreira and Caceres,1989). The axon- and den- chemically. When the antisense was given continuously di- dritelike structures of cultured hippocampal neuronshave been rectly after plating, the neurites persisted as simple minor characterizedin the most detail. Ultrastructural analysesofthese outgrowths. When antisense was added 72 hr after plating, two classesof neurites revealed the presenceof ribosomesand axonlike neurites were lost, while the remaining neurites postsynaptic elementsin dendritelike processesand the absence continued to grow and increase in complexity. We concluded of ribosomesand presynaptic elementsin axonlike structures that the initial establishment of an elongated axonlike neurite (Bartlett and Banker, 1984). Tritiated uridine is transported is a prerequisite for further neurite maturation; however, once selectively into cultured hippocampal dendrites (Davis et al., the axon is established, the remaining neurites are able to 1987). grow independently and assume a tapered dendritelike ap- Two microtubule-associatedproteins (MAPS) can also serve pearance. as markers of axons or dendrites. The MAP family is a highly diverse group of proteins that can promote microtubule poly- Axons are collectively defined by morphological, biochemical, merization and codistribute with microtubules in cells. Segre- and physiological criteria. A key morphological feature of an gation of the somatodendritic marker MAP2 has been shown axon is its nontapering, elongated structure. Becausethe micro- in hippocampal (Cacereset al., 1984), cerebellar(Ferreira et al., tubule system representsa link to structure, it is reasonableto 1989), cerebrocortical (Kosik and Finch, 1987),and sympathetic consider how microtubules contribute to the generation of an cultures (Peng et al., 1986). Tau protein has been shown to elongatedaxonlike structure. A related question is how micro- segregateto the axon in cerebellar (Ferreira et al., 1989) cere- tubules contribute to the generation of two morphologically brocortical (Kosik and Finch, 1987), and sympathetic cultures distinct setsof neurites-axons and dendrites. The orientation (Peng et al., 1986). The principal site of synthesisof tau in the of the plus and minus endsof microtubules servesto distinguish brain is in neurons,on a population of mRNAs in the neuronal axons and dendrites (Burton and Paige, 1981; Filliatreau and cell body and within the most proximal portion of the dendrite DiGiamberdino, 1981; Heidemann et al., 1981; Baas et al., (Kosik et al., 1989a). 1988). In axons, the plus ends are aligned distally, whereasin In PC 12 cells, neurite outgrowth correlateswith the induction dendrites, microtubules have a mixed orientation, particularly of tau protein (Drubin et al., 1985) and tau mRNA (Drubin et in the midportion of the dendrite. Becausethe two microtubule- al., 1988); therefore, a suggestedfunction of tau involves neurite basedmotor molecules,kinesin and dynein, are defined in part elongation. When infected into the foreign host cell (39 cells from the moth ovary) via a baculovirus, tau induced lengthy Received Sept. 14, 1990; revised Dec. 21, 1990; accepted Jan. 3, 1991. processoutgrowths (Kosik et al., 1990). Tau may also have a We are indebted to G. Hall. G. Lee. A. Andreadis. and A. Ferreira. for their role in the generation of asymmetric neuronal morphologies, critical reviews of the manuscript and to R. Maciewicz for use of his Eutectics Neuron Tracing System. The studies were supported by NIH Grants AGO6601 becausetau antisenseexposure in neuronal cell culture specif- and AGO6 172. ically inhibits the elaboration of an elongated, nontapering ax- Correspondence should be addressed to Kenneth S. Kosik, M.D., Center for onlike structure (Caceresand Kosik, 1990). Neurologic Diseases, Brigham and Women’s Hospital, 75 Francis Street, Boston, MA 02115. Shortly after neuronsare plated, the minor neuritesbegin to Copyright 0 199 1 Society for Neuroscience 0270-6474/9 l/l 115 15-09$03.00/O differentiate, the first into an axon, and then the others into 1516 Caceres et al. l Tau and Neurites in Culture

dendrites. A minor neurite becomes defined as an axon in hip- al., 1988b) and tau 1 (Binder et al., 1985); both monoclonals were used pocampal neurons once it has elongated 10 pm longer than any at 1:20 and l:lOO, respectively. The MAP2 antibodies were 5F9, a monoclonal(Kosik et al., 1988a)used at 1:100, and R4, a MAP2 poly- other neurite (Goslin and Banker, 1989). Uniformly aligned clonal(Kosik and Finch, 1987)used at 1500. Tubulinwas labeled with microtubules are present in minor neurites (Baas et al., 1989) TuJl to total tubulin (Geisert and Frankfurter. 1989). Tvrosinated tu- and therefore cannot be used to ascertain the onset of volaritv. bulin and MAPla antibodies were obtained from Sihma’Chemical Co. To identify those moleculesimportant in the early definition of (St. Louis, MO). The secondary antibodies used w&e the mouse ex- an axon, one strategy has been to localize moleculesimmuno- travidin staining kit (Sigma) with diaminobenzidine as the chromagen, rabbit anti-mouse IgG biotin conjugate (Sigma), goat anti-mouse IgG cytochemically within the neurite that is transforming to an rhodamine conjugate (Calbiochem), extravidin fluorescein isothiocyan- elongated axonlike structure. The molecules reported to meet ate (FITC) conjugate (Sigma), and goat anti-rabbit IgG rhobdamine these criteria are GAP43 (Goslin et al., 1988), synapsin, and conjugate (Sigma). synaptophysin(Fletcher et al., 1989). However, thesemolecules For immunocytochemistry, cells were fixed with either cold methanol (-20°C) or 4% paraformaldehyde in PBS with 0.12 M sucrose at 37°C. do not provide a direct link to the microtubules, the structure Paraformaldehyde-fixed cells were permeabilized in 0.2% Triton X-100 required for elongation and possiblyfor the spatial configuration in PBS for 5 min. For double-labeling immunofluorescence, the cells of an axonlike structure. were incubated with polyclonal and monoclonal antibodies simulta- To study the interdependenceof axonal and dendritic growth, neously, washed, and incubated with the secondary antibodies goat anti- we administered tau antisenseoligonucleotides in a sustained rabbit TRITC conjugate, rabbit anti-mouse IgG biotin conjugate, and extravidin FITC conjugate together. In all experiments, the coverslips manner from the time of plating and in a delayed manner after were preincubated with 5% bovine serum albumin. Photomicrographs neurite differentiation was underway. Antisense oligonucleo- were taken on an Carl Zeiss microscope equipped with epifluorescence tides can be taken up by cells from the media in a saturable, illumination throuah 40 x or 63 x obiectives usina TMAX 100 or 400 size-dependentmanner consistent with receptor-mediated en- ASA film (EastmanKodak Co.). ” Analysisof cultures.Developing cerebellar neurons were staged ac- docytosis(Loke et al., 1989). We have shown that neuronswith cording to Ferreira et al. (1989) and are illustrated in Figure 1. Stage I a sustainedexposure to tau antisensefrom the time of plating cells are either rounded or have a lamellipodial veil. Stage II cells have remain locked as an array of minor net&es, while failing to a fairly symmetric array of short, unbranched neurites (Fig. IA). The develop the morphologically distinct neurites characteristic of transition from stage II to stage III represents the acquisition of mor- axons or dendrites. On the other hand, the addition of tau an- phological polarity in that a single neurite begins to elongate (Fig. 1B). At stage III, one neurite has a relatively uniform diameter over most tisenseoligonucleotides to cultures that have establishedelon- of its length and thus has the appearance of an axon (Fig. 1C). Stage IV gated axonlike structures results in the apparent regressionof occurs when the minor neurites develop a tapered, highly branched the most elongatedneurite and the continued elaboration of the appearance, similar to that of dendrites (Fig. l&E). Analogous stages other neurites. of maturation have been described for cultured hippocampal neurons (Dotti et al., 1988). Some cultures were drawn using a 63 x ocular on the Eutectics Neuron Materials and Methods Tracing System (Johnson and Capowski, 1983) and analyzed according to its software. Camera lucida drawings were made, including a set with Cell cultures.Dissociated cultures of cerebellar macroneurons, prepared from embryonic day 15 rat embryos, were dissociated and plated onto a 100 x oil-immersion objective to determine neurite caliber at the tip. All drawings were made from slides stained with the tubulin or MAP2 polylysine-coated coverslips at a uniform density of 1O,OOO-20,000/cmZ (Ferreira et al., 1989). After 1 hr in serum-containing medium, cells were antibody. Bright-field drawings were checked by phase-contrast mi- maintained in D-minimum essential medium/Ham’s F12 suuplement- croscopy to be sure all neurites were drawn. ed with the N2 mixture of Bottenstein and Sato (1979). All the cultures were grown in a 37°C incubator with 5% CO,. Two nonoverlapping tau antisense oliaonucleotides designated RT 1 1- 14 and RT23 were used at 50 PM in all-the experiments k show that the effects were specific to Results the tau sequence. The degree to which these oligonucleotides entered Efect of prolonged exposure to tau antisense the cells and the response to varying doses were described previously Tau antisenseoligonucleotides were added to the media of cer- (Caceres and Kosik, 1990). Antisense RT 1 1- 14 is the rat tau nucleotide GGTTCAGCCATGCTGCTTCAAAGCC, which corresponds to the ebellar macroneurons in culture from the time of plating until inverse complement of nucleotides - 14 to + 11 (Kosik et al., 1989b). 48 hr to determine the effects of sustainedantisense exposure Antisense RT23 is the nucleotide TGATAATCGACAGGAGGCGAG- on morphological development. The abolition of detectable tau GACA, which corresponds to the inverse complement of nucleotides protein in antisense-treatedcultures was demonstrated previ- - 15 to -25 (Kosik et al., 1989b). The rat tau sequences selected represent adjacent regions, one that spans the intiator AUG and the other ously by dot immunobinding (Caceresand Kosik, 1990) and is immediately upstream. The sequences selected were not found in the shown here by immunocytochemistry (Figs. 2, 3). Staining by data base in any other known sequence. Adjacent control wells were two tau monoclonal antibodies (5E2 and tau 1) revealed back- treated with the same concentration of the corresponding sense-strand round levels of reaction product over the entire treated culture. oligonucleotide. The oligonucleotides were synthesized on an Applied Both antisenseoligonucleotides (RTl l-14 and RT23) were Biosystems 380B synthesizer, purified over an NAP5 column (Phar- macia), ethanol precipitated, and taken up in medium. equally effective in inhibiting tau labeling. In all of the para- Antisense oligonucleotides were either administered continuously from digms,the sense-and antisense-treatedcultures synthesized oth- the time of plating or delayed until after the onset of polarity. When er proteins as determined by their reactivity with tubulin, administered continuously, the tau oligonucleotides were added 1 hr MAPla, and MAP2 antibodies (Figs. 2, 3).With a sustained after plating, and again at 24-hr intervals to 48 hr, and fixed for immunocytochemistry at 72 hr. When administered in a delayed fashion, dose of antisensefrom the time of plating, cultured neurons a single dose of tau antisense was added after the cells were in culture extended initial exploratory neurites and maintained a sym- for 72 hr and then fixed 24 hr later. One additional protocol was utilized metric array of minor neurites (Fig. 2). Control sense-treated in which antisense oligonucleotides were added with a delay of only 24 cultures, on the other hand, not only extended initial neurites, hr, and in some cases, a second dose was administered at 48 hr. These but also elongated a single neurite within the first 24 hr that cultures were fixed 24 hr after the last dose of antisense. For each experiment, several plates were not fixed at the determined time point, developed the appearanceof an axon. Neurons could be kept but were allowed to recover from the effects of the antisense. locked at this stageby administration of tau antisenseevery 24 Immunocytochemistry.The tau antibodies used were 5E2 (Kosik et hr. At 72 hr in culture, cells treated according to the sustained The Journal of Neuroscience, June 1991, 7 7(6) 1517

Figure I. Development of cerebellar macroneurons labeled with tau monoclonal antibody tau 1 (A-C) or MAP2 (R4) antibody (0, E). A, Stage II cell with minor neurites (arrow) after 12 hr in culture. B, Transition to stage III after 24 hr in culture; the arrow shows an elongating neurite. C, Stage III cell with a single elongated neurite (arrow) and very-fine-caliber remaining minor neurites after 24 hr in culture. At this stage, the remaining minor neurites become thinner. D, Stage IV neuron after 4 d in culture, stained with MAP2 antibody, at which time the caliber of the minor neurites increases and many neurites appear tapered (small arrows); the large arrows point to a bundle of axons, which at this time point are still reactive with the MAP2 antibody. E, Seven days after plating, only the dendritelike neurites are labeled with the MAP2 antibody.The arrows point to the trajectory of an axon apparent only by phase contrast. Scale bar, 10 pm. exposure protocol markedly contrasted with control cells, which a single elongated neurite (Fig. 20), followed by the develop- had elongatedaxonlike structures (cf. Figs. 1, 2). The effect of ment of branched and tapered neurites (Fig. 2E). Releasefrom antisensetreatment was very different from a complete inter- the phenotypic effectsof tau antisenseappeared to occur within ference with protein synthesis as determined by the adminis- 24-48 hr of the cessationof antisenseadministration. This time tration of cycloheximide at comparable times. This treatment period, which must approximate the half-life of the antisense, resulted in many nonviable neurons and the complete inhibition was the basis for selectinga 24-hr interval betweendoses. The of any neurite outgrowth (data not shown). cells were able to recover from the sustainedexposure paradigm When the daily administration of antisensewas discontinued and resume development from the point at which they were after 72 hr and the cells allowed to recover, they acquired po- arrested. Antisense treatment did not irreversibly impair the larity and continued to differentiate their neurites within a time viability or the ability of the neuronsto differentiate. However, frame identical to cells plated at the time antisensewas discon- over the period observed, neurite growth failed to catch up with tinued (Fig. 2D,E). Cells fixed at 120 hr (Fig. 20) and at 144 the controls and maintained a lag commensuratewith the time hr (Fig. 2E) after plating and antisensetreatment first formed in antisense. 1518 Caceres et al. - Tau and Neurites in Culture

Figure 2. Continuousexposure to tau antisensegiven 1,24, and48 hr afterplating and fixed at 72hr. A andB aredouble-labeled neurons stained with MAP2 antibodyR4 (A) and tau antibodytau 1 (B). Only tracetau immunoreactivityis detectable.C is an identicallytreated plate of cells stainedwith tyrosinatedtubulin antibody.The appearanceof thesecells is similarto the stageII cell shownin Figure 1A. D andE are recovering cells stainedwith tau 1, 2 (D) and 3 (E) d following the cessation of antisense administration. Scale bar, 15 pm.

branching in cerebellar macroneuronscultured for 7 d (Fig. 4). E$ect of tau antisense administration after the onset of Neurites fell into two populations-those that tapered and those polarity that did not. For nearly all cells, a single nontapered neurite Tau antisenseoligonucleotides were administeredafter the onset could be identified. Measurementsof the dendritic diameterjust of polarity to determine their effect on cells that had already proximal to any terminal expansion gave a range of diameters formed an axonlike neurite. Preliminary to determining the ef- from 0.8 to 1.2 pm. The terminal diameter of axonlike neurites fects of tau antisenseon neurites, we first establishedthat two was 0.5-0.8bm. These values may also distinguish axons and morphologically distinguishable populations of neurites were dendrites (Banker and Waxman, 1988). present in this culture system. Becausetapering of neurites is At 72 hr after plating, cerebellar cultures were treated with considered a close correlate of dendritic identity (Baas et al., tau antisensefor 24 hr and then fixed for immunocytochemistry 1989), the degreeof tapering was quantitated among primary (Fig. 3). Tau immunoreactivity wasabolished; however, tubulin branchesand secondary branchesthat did not undergo further and MAP2 immunoreactivity were identical to the controls. The Journal of Neuroscience, June 1991, 7 7(6) 1519

Figure 3. Delayedexposure of cerebellarmacroneurons to tau oligonucleotidesgiven as a singledose 72 hr after plating.Cells were fixed 24 hr later anddouble labeled with tau antibody5E2 (1:20) and MAP2 polyclonalantibody R4 (1:500).A and B representthe tau antisense,and C, the sensecontrol. Tau immunoreactivityis demonstratedin B and C, andMAP2 immunoreactivity,in A. Scalebar, 10pm.

Likewise, MAP 1a immunoreactivity was identical to controls pering over the first 10 Km. Thickness versus distance mea- (data not shown). Double labeling of treated cells with tau and surementsof the neurites from cultures treated with delayed MAP2 antibodies demonstrated that cells with intense MAP2 administration of antisensehad a correlation coefficient iden- immunoreactivity were not stained by the tau antibody under tical to that of the dendritelike population of neurites from antisenseconditions (Fig. 3A,B). The most dramatic difference untreated cultures (Fig. 4). When thesecells remained in culture between the appearanceof the antisenseand control cells was beyond the 24-hr period between antisenseadministration and the nearly complete lack of neurites that resembledaxons. Be- fixation, they recovered normally. causetapering distinguishedneurite populations, a plot of thick- Thirty cells treated with delayed administration of oligonu- nessversus distance from the cell body for primary and sec- cleotideswere labeled with the tubulin antibody, and randomly ondary neurites was used to demonstrate that the remaining selectedfields of the sense-and antisense-treatedcultures were population of neuritesafter antisensetreatment had a dendrite- drawn with the Eutectics Neuron Tracing System(Johnson and like morphology (Fig. 4). In untreated cultures, there was a Capowski, 1983). Representative cells are shown in Figure 5. highly significant inverse correlation for the population of den- While an axonlike neurite was identifiable in nearly all of the dritelike net&es; for the population of neurites arising from the sense-treatedneurons, only 6 of 30 antisense-treatedcells had cell body with an axonlike appearance,the caliber did not change a similar axonlike neurite that could be distinguishedfrom the as a function of distance from the cell body beyond the initial other neurites on the basis of length and lack of tapering. Al- 5 pm. Some axons arose from the proximal portion of den- though these six cells appeared to have escapedthe morpho- dritelike processes,in which there was occasionally some ta- logical effects of the antisense,the lack of tau immunoreactivity 1520 Caceres et al. l Tau and Neurites in Culture

B 6 5 C I

07 20 40 60 80 160 20 40 50 80 100 0 20 40 60 80 100 Distance Distance Distance

D E

Figure 4. Relationship of the neurite diameter to the distance from the cell body for dendritelike processes after 7 d in culture (A), axonlike processes after 7 d in culture (B), minor neurites after 1 d in culture (C), neurites after 4 d in culture (O), and neurites treated with tau antisense given at 72 hr after plating and fixed 24 hr later(E). Diameters are in microns, and the distance from the cell body is standardized for the variable lengths of cultures A-E by setting the neurite tip at 100 and measuring the diameter at three points that represent percentages of the total length. The correlation coefficients (r) with length are A, 0.92; B, 0.24; C, 0.24; D, 0.87; and E, 0.92. The analysis utilized primary net&es that did not undergo further branching. Twenty cells with an average of 2.5 neurites per cell were drawn for each graph. washomogeneous over the entire culture. Ten antisense-treated 4.06 f 1.29; antisense,4.33 f 1.83). Therefore, the increasein neurons had a single neurite with unusual morphological fea- branch complexity is due to higher-order sprouting. Becausean tures, such as multiple branchesemanating from a common axon could not be definitively identified in many of the anti- origin or unusual branch angles.The loss of a single elongated sense-treatedcells, thesecalculations were done in several ways: or axonlike neurite is quantitated in Figure 6, which showsthat (1) inclusion of all neurites without assumptionsregarding den- the length of the longestneurite per cell was significantly greater dritic or axonal identity; (2) exclusion of the axon from the under the control conditions and that antisense-treatedneurons sense-treatedcultures; (3) exclusion of the axon from the sense- had increasednumbers of shorter net&es. Comparison of all treated cultures and the systematicexclusion of a single neurite the neurites from the antisense-treatedpopulation revealed a from each of the antisense-treatedcells; and (4) exclusion of any significantly shorter total summed neurite length, thicker pri- axonal-appearing structure from both groups. All methods of mary and secondaryneurites, and neuritic volumes equal to the calculation gave significantly increasedbranching in the antisense-treatedcells (Table 1). sensecultures; the most conservative values (calculation 3) are Neurite complexity was analyzed by the probability of presented. branching from each segment(Caceres and Steward, 1983). The The presenceof tapered neurites that continued to increase averagenumber of branchesper cell wasgreater in the antisense- in complexity while exposedto tau antisensesuggested that this treated cultures (antisense,17.6 + 10.8; sense,11.5 + 7.8; p < population of neurites was resistant to the effects of the anti- 0.01). There were 3.93 f 0.25 second-ordersegments in the sense.Together the resultsfrom the two protocols (summarized sensecultures and 5.86 + 0.10 in the antisensecultures (p < in Fig. 7) suggestedthat growth of highly branched tapered 0.00 1). There were 1.93 + 0.10 third-order branchesin the sense neurites could occur in the absenceof an elongated axonlike cultures and 3.44 f 0.16 in the antisensecultures (p < 0.001). neurite, but only after somedegree of neurite differentiation had The likelihood of second-ordersegments branching further was already begun to occur. Cellswith early and sustainedexposure 0.32 for the antisenseand 0.22 for the sense.The likelihood of to antisensedid not develop complex branched neurites. An third-order segmentsbranching further was 0.28 for the anti- additional intermediate time point of antisenseadministration senseand 0.15 for the sense.The number of first-order segments was tried to demonstrate that the responseto antisenseat 72 hr arisingfrom the somadid not differ in the two conditions (sense, was developmentally distinct from that observed with contin- The Journal of Neuroscience, June 1991, 17 (6) 1521

2 Anti-se rise

Figure 5. Representative cells from sense- (A) and antisense-treated (I?) cultures, stained with anti-tubulin (TuJ l), and traced with the Eu- tectics system under 63 x oil immersion lens directly from the micro- ::::::: scope stage. Thickness values were entered as the cells were traced. Scale ::::::: bar, 15 pm. ::::::: ::::::: ::::::: ::::::: uous application from the time of plating. Neurons were treated .::::::::::::: with antisense oligonucleotides after 24 hr in culture, a time ::::::: ::::::: when the population of neurites appears to consist only of simple ::::::: ::::::: minor neurites and a single elongated neurite on many cells. .::::::::::::: Some cultures were given a second dose of tau antisense at 48 :::::::.:::::: ~ ::::::: I -.I hr after plating. Fixation 24 hr after the last dose of antisense 300 600 700 revealed cells identical to those treated with continuous application of antisense from the time of plating. Therefore, cerebellar Let lgth (I neurons must develop for at least 24 hr in culture before their minor neurites are able to maintain sustained growth in the Figure 6. The left shift to shorterneurites is shownin the antisense- treatedpopulation. The meanvalue for the longestsense-treated neurite presence of tau antisense. is 302.0pm, with a rangeof 45.7-551.3pm. The meanvalue for the Finally, when the application of tau antisense was delayed longestantisense-treated neurite is 130.5,with a rangeof 27.1-298.3 until 96 hr after plating and the cells fixed 24 hr later, there was pm. Thisdifference was highly significant 0, < 0.0003)using the Mann- no detectable effect on tau immunoreactivity or on the mor- Whitney U test for the nonparametricdistribution of the histogram. phology. Whether continuous application of tau antisense be- ginning at 96 hr could affect neurite growth was not tested, but might be effective if the half-life of tau has increased with neu- ciably changed. At 4 d in culture, the minor neurites show ronal maturity. tapering and the complex branching pattern typical of dendrites (Fig. 1D). Discussion The continuous exposure to tau antisensepermitted the hy- The analysis of IO-d-old cerebellar neurons revealed two pop- pothesisthat the prior development of an elongated axonlike ulations of neurites-those that tapered and branched at rela- neurite was necessaryfor the conversion of minor neuritesinto tively short distances, and those that did not taper and had an dendrites. Once the cell passedthe stageof axonal formation, elongated appearance. As neurites are elaborated from cerebellar a distinct population of neurites emergedthat resembledden- neurons, they follow a highly reproducible pattern of maturation drites. With application of the antisensedelayed until 72 hr, (Fig. 1). Within the first 12 hr of plating, the neurons elaborate thesedendritelike neurites were capableof continued growth in a symmetric array of minor neurites that are not tapered (Fig. the absenceof an identifiable axon. Therefore, the transition 1A). By 24 hr, most of the cells have elaborated an axonlike from minor neuritesto dendritelike processesrequires a chrono- structure, while the remaining minor neurites have not appre- logically prior stageof axonal initiation. Becausethis observa-

Table 1. Effects of tau antisense on neuronal shape parameters

Total Total Thickness/ Thickness/ length(pm) volume(pm)) 1” segment 2” segment Somaarea bm2) Sense 420.7 167.9 1.1 + 0.07 0.5 + 0.05 105.2k 4.11 (75.6-691.8) (27.5-482.3) Antisense 252.6 188.5 1.4 rfr 0.11 0.7 k 0.05 106.8k 4.87 (63.8439.2) (16.7-737.8) Significance p < 0.005 p > 0.1 p < 0.05 p < 0.01 p > 0.2 The parameterslisted were compared for cultures treated with either sense or antisense oligonucleotides added at 72 hr after plating and fixed 24 hr later. For the total length and total volume, the significance was determined by the Mann- Whitney U test because the distribution of values was skewed. For the remaining determinations, a two-tailed Student’s t test was used. All values were generated with n = 30, and errors refer to SEM. The numbers in parentheses are ranges. 1522 Caceres et al. * Tau and Neurites in Culture

Figure 7. Camera lucida drawings of anti-tubulin-labeled neurons treated with either continuous or delayed exposure to tau antisense oligonucleotides for the number of days in culture in- dicated along the top. Row A cells were continuously exposed to tau antisense from the time of plating (see Materials and Methods). Row B cells received a single dose of tau antisense 72 hr after plating and were fixed 24 hr later. At 3 B d with daily exposure to antisense since the time of plating, only short minor neurites are evident, whereas untreated cells Crow B) at 3 d show an elonaated process. With the cessation of daily antisense treatment, the cells begin to show recovery at 4 d, whereas delayed antisense administration results in the loss of the axonlike neurite and continued growth of the other net&es. By 5 d, cells that were initially exposed to daily antisense administration continue to recover and look similar to untreated cells at day 3. With a single dose of antisense at day 3, recovery of the axon by day 5 is apparent. Arrows correspond to. neurites that extend beyond the figure.

tion suggesteda developmental threshold, an additional inter- inhibit other minor neurites from becoming axons. In hippo- mediate time point of antisense application was studied. campal neurons,the transection of an axon closeto the cell body Oligonucleotidesgiven at 24 hr in culture, a time when polarity at early time points releasesthe inhibition on elongation and is just becoming evident, resulted in morphologies indistin- permits the conversion of one of the other early neurites to an guishablefrom continuous antisenseapplication from the time axon (Goslin and Banker, 1989). To perform a unique role of plating. Thus, at 24 hr, neurites still require tau synthesisfor within only one of the neurites, either tau is distinct within a any differentiation beyond minor net&es. At 72 hr, only the singleneurite undergoingelongation, or, asan effector molecule, elongation of the axonlike neurite is dependent upon continued it acts differentially upon a substratethat is distinct within the tau synthesis. incipient axon. Although tau does undergo complex develop- The presenceof neurites with only minimal axonal features mentally regulated splicing events (Kosik et al., 1989b), there is when antisensewas administered at 72 hr suggestedneurite no changein the electrophoretic migration of the tau isoforms regressionand a role for tau in sustaining an axon. Dendritic during the time course of these experiments (Ferreira et al., branching complexity hasbeen described as a function of plating 1989). A more likely neurite-specific alteration of tau is in its density (Banker and Waxman, 1988); however, in this study, relationship to the microtubule. The binding of tau to micro- under conditions of decreasedcell-cell contact due to the loss tubules selectively within a single neurite may occur stochas- of the long axon, branching complexity was increasedin com- tically and serve as a critical step in the formation of the axon. parison to controls. In many systems,particularly in vivo sys- Once tau binds, it may engendercooperative phenomenathat tems, when a neuron is deprived of its axon, sprouting occurs. enhanceadditional binding and lead to neurite elongation. Den- This responseusually occursfrom the axon stump, but can occur dritic differentiation can proceed in the absenceof tau, once ectopically from dendritic tips after a closeaxotomy (Hall et al., neurons have entered stageIII. The MAP2 molecule, with its 1989). It is possible that the enhanced branching complexity ability to displacetau from microtubules (Sandoval and Van- after delayedtau antisenseadministration representeda reactive dekerckhove, 198l), representsa good candidate for a comple- responseto the loss of an axonlike process. mentary role in dendritic differentiation. How neurites gain distinct identities as axons and dendrites When the 72-hr cultures were fixed after only 16 hr in anti- must entail a cascadeof molecular events that lead to a mature sense,considerable tau immunoreactivity remained. Becausea morphology. One aspect of axonal identity is the distinct mor- 24-hr exposureto the antisenseresulted in clear effectson morphology attained by a single elongated, nontapering structure phology and tau immunoreactivity, the briefer 16-hr interval among a population of neurites. While tau protein probably has may representinsufficient time for the turnover of the tau pro- a role in the determination of the axon’s unique morphology tein pool synthesizedbefore the administration of the antisense. (Caceresand Kosik, 1990), neuronalpolarity doesnot arise from Fixation after 48 hr in the antisenserevealed the reappearance the selective segregationof tau protein to a single neurite. Tau of tau immunoreactivity, probably due to exhaustion of the is present in all neurites, including minor neurites at the time antisenseoligonucleotide. When tau antisensewas added after of axonal formation. Therefore, a mechanism must exist to 96 hr in culture and analyzed 24 hr later, there wasonly minimal The Journal of Neuroscience, June 1991, 7 I(6) 1523 loss of the axonlike neurite. The failure to see a decrement in Drubin DG, Kobayashi S, Kellogg D, Kirschner MW (1988) Regu- tau protein after 24 hr, when the antisense was added at 96 hr lation of microtubule protein levels during cellular morphogenesis in nerve growth factor-treatedPC 12 cells. J Cell Biol 106:1583-l 59 1. rather than 72 hr in culture, could be explained by a lengthening Ferreira A, Caceres A (1989) The expression of acetylated microtu- of the tau protein half-life between these time points. The ob- bules during axonal and dendritic growth in cerebellar macroneurons served decrement in tau mRNA levels with maturity (Mangin which develop in vitro. Dev Brain Res 49:207-213. et al., 1989) is consistent with a concomitant lengthening of its Ferreira A, Busiglio J, Caceres A (1989) Microtubule formation and neurite growth in cerebellar macroneurons which develop in vitro: half-life. Tau could become less vulnerable to proteolysis by evidence for the involvement of the microtubule-associated proteins binding to microtubules and proteolytically eliminated when MAP- la, HMW-MAP2, and tau. Dev Brain Res 49:2 13-228. free of microtubules (Okabe and Hirokawa, 1989). Thus, bind- Filliatreau G, DiGiamberdino L (198 1) Microtubule polarity in my- ing of tau to microtubules may represent a critical step in the elinated axons as studied after decoration with tubulin. Biol Cell 42: transformation of an initial exploratory neurite to an axon. In- 69-72. Fletcher TL, De Camilli P, Banker GA (1989) Synaptogenesis in hip- direct support for the importance of microtubule stability in the pocampal cultures: axons and dendrites become competent to form elaboration of an axonlike structure stems from the finding that, synapses at different stages of development. Sot Neurosci Abstr 15: in cultured, cerebellar neurons, microtubules are first acetylated 1388. in the axon (Ferreira and Caceres, 1989). Furthermore, the ad- Geisert EE Jr, Frankfurter A (1989) The neuronal response to injury dition of tau antisense just after plating results in the failure to as visualized by immunostaining of class III fl-tubulin in the rat. Neurosci Lett 102: 137-14 1. detect acetylated microtubules, despite the persistence of tubulin Goslin K, Banker G (1989) Experimental observations on the devel- reactivity (A. Caceres and K. S. Kosik, unpublished observa- opment of polarity by hippocampal neurons in culture. J Cell Biol tions). Because the phosphorylation state of tau is known to 108:1507-1516. affect its binding to microtubules (Lindwall and Cole, 1984), a Goslin K, Schreyer DJ, Skene JHP, Banker G (1988) Development of neuronal polarity: GAP-43 distinguishes axonal from dendritic balance between cellular kinases and phosphatases may regulate growth cones. Nature 336~672-674. tau binding to microtubules and thus operate in the generation Hall GF, Poulos A, Cohen MJ (1989) Sprouts emerging from the of polarity. dendrites of axotomized lamprey central neurons have axonlike ul- trastructure. J Neurosci 9:588-599. References Heidemann SR. Landers JM. Hambora MA (198 1) Polaritv orientation of axonal microtubules. J Cell I%01 91:661-665. - Baas PW, Black MM, Banker GA (1989) Changes in microtubule Johnson EM, Capowski JJ (1983) A system for the three-dimensional polarity orientation during the development of hippocampal neurons reconstruction of biological structures. Comput Biomed Res 16:79. in culture. J Cell Biol 109:3085-3094. Kosik KS, Finch EA (1987) MAP2 and tau segregate into axonal and Baas PW, Deitch JS, Black MM, Banker GA (1988) Polarity orien- dendritic domains after the elaboration of morphologically distinct tation of microtubules in hippocampal neurons: uniformity in the neurites: an immunocytochemical study of cultured rat cerebrum. J axon and nonuniformitv in the dendrite. Proc Nat1 Acad Sci USA Neurosci 7:3 142-3153. 85:8335-8339. - Kosik KS, Orecchio LD, Bakalis S, DulTy L, Neve RL (1988a) Partial Banker GA, Waxman AB (1988) Hippocampal neurons generate nat- sequence of MAP2 in the region of a shared epitope with Alzheimer ural shapes in cell culture. In: Intrinsic determinants ofneuronal form neurofibrillarv tandes. J Neurochem 5 1:587-598. and function (Lasek RJ, Black MM, eds), pp 61-82. New York: Liss. Kosik KS, Orecchio-LD, Binder L, Trojanowski JQ, Lee V, Lee G Bartlett WP, Banker GA (1984) An electron microscopic study of the (1988b) Epitopes that span the tau molecule are shared with neu- development of axons and dendrites by hippocampal neurons in cul- rofibrillary tangles. Neuron 1:8 17-825. ture. J Neurosci 4:1944-1953. Kosik KS, Crandall JE, Mufson E, Neve RL (1989a) Tau in situ Binder LI, Frankfurter A, Rebhun LI (1985) The distribution of tau hybridization in normal and Alzheimer brain: a predominant local- in the mammalian central nervous system. J Cell Biol 101: 1371- ization in the neuronal somatodendritic compartment. Ann Neurol 1385. 26:352-361. Black MM, Baas PW (1989) The basis of polarity in neurons. Trends Kosik KS, Orecchio LD, Bakalis S, Neve RL (1989b) Developmentally Neurosci 12:211-214. regulated expression of specific tau sequences. Neuron 2: 1389-l 397. Bottenstein JE, Sato GH (1979) Growth of a rat neuroblastoma cell Kosik KS, Caceres A, Pardee J, Cohen-Gould L, Knops J, McConlogue line in serum-free supplemented media. Proc Nat1 Acad Sci USA 76: L (1990) A role for tau in neural morphogenesis. J Cell Biol 111: 514-517. 436a. Burton PR, Paige JL (198 1) Polarity of axoplasmic microtubules in Lindwall G, Cole RD (1984) Phosphorylation affects the ability of tau the olfactory nerve of the frog. Proc Nat1 Acad Sci USA 78:3269- protein to promote microtubule assembly. J Biol Chem 259:5301- 3273. - 5305. Caceres A, Kosik KS (1990) Inhibition of neurite polarity by tau Loke SL, Stein CA, Zhang XH, Mori K, Nakanishi M, Subasinghe C, antisense olinonucleotides in primary_ _cerebellar neurons. Nature 343: Cohen JS, Neckers LM (1989) Characterization of oligonucleotide 461-463. - transport into living cells. Proc Nat1 Acad Sci USA 86:3474-3478. Caceres A, Steward 0 (1983) Dendritic reorganization in the dener- Mangin G, Couchie D, Chaniere-Bertrand C, Nunez J (1989) Timing vated dentate gyrus of the rat following entorhinal cortical lesions: a of expression of 7 and its encoding mRNAs in the developing cerebral Golgi and electron microscopic study. J Comp Neurol214:387403. neocortex and cerebellum of the mouse. J Neurochem 53:45-50. Caceres A, Banker GA, Steward 0, Binder L, Payne M (1984) MAP2 Okabe S, Hirokawa N (1989) Rapid turnover of microtubule-asso- is localized to the dendrites of hippocampal neurons which develop ciated protein MAP2 in the axon revealed by microinjection of bio- in culture. Dev Brain Res 13:3 14-3 18. tinylated MAP2 into cultured neurons. Proc Nat1 Acad Sci USA 86: Davis L, Banker G, Steward 0 (1987) Selective dendritic transport of 41274131. RNA.in hippocampal neurons in culture. Nature 330:477-479. Peng A, Binder LI, Black MM (1986) Biochemical and immunological Dotti CG. Sullivan CA. Banker GA (1988) The establishment of po- analyses of cytoskeletal domains of neurons. J Cell Biol102:252-262. larity by hippocampal neurons in culture. J Neurosci 8: 1454-1468. Sandoval IV, Vandekerckhove JS (198 1) A comparative study of the Drubin DG, Feinstein SC, Shooter EM, Kirschner MW (1985) Nerve in vitro polymerization of tubulin in the presence of microtubule- growth factor-induced neurite outgrowth in PC 12 cells involves the associated proteins MAP2 and tau. J Biol Chem 256:8795-8800. coordinate induction of microtubule assembly and assembly pro- moting factors. J Cell Biol 10 1: 1799-l 807.