Characterization of a PC12 cell sub-clone (PC12-c41) with enhanced neurite outgrowth capacity: implications for a modulatory role of high molecular weight tau in neuritogenesis

Neuritogenesis is an important event during nervous system development whereby immature nerve cells acquire unique neuronal morphology by extending axons and/or dendrites. Previous studies have identified a number of molecules that play crucial roles in neurite outgrowth (reviewed by Bixby and Harris, 1991). Amongst these are a number of neuronspecific cytoskeletal elements, including microtubule-associated proteins (MAPs; Nunez, 1986; Olmsted, 1986; Matus, 1988; Tucker, 1990). Although many of the functional aspects of such molecules have been determined, it is less clear how their relative levels of expression may contribute to defining overall neuronal and neuritic morphology (cf. Bunge, 1986; Matus, 1988).

The rat pheochromocytoma PC12 cell line (Greene and Tischler, 1976, 1982) has been an instructive model system for studying the underlying mechanisms of neurite outgrowth (reviewed by Aletta et al., 1990). PC12 cells proliferate in serum-containing culture medium and resemble their non-neoplastic adrenal pheochromoblast counterparts. When exposed to nerve growth factor (NGF; see Levi-Montalcini and Angeletti, 1968; Levi and Alemá, 1991, for review), these cells cease division and gradually develop many of the characteristics of mature sympathetic neurons, including the outgrowth of long, branching neurites. This process is accompanied by changes in levels and post-translational modification of a variety of cytoskeletal proteins (Aletta et al., 1990) as well as of other neural markers (reviewed by Greene and Tischler, 1982; Levi and Alemá, 1991). As a replenishable cell line, PC12 cells offer the advantages of providing ample material for biochemical analyses and are amenable to genetic manipulation (Greene et al., 1991). The latter is a particularly powerful approach for understanding complex biological phenomena.

Previous studies have documented a number of PC12 mutant/variant cell lines that in various ways differ from wild-type PC12 cells (Bothwell et al., 1980; Van Buskirk et al., 1985; Green et al., 1986; Katoh-Semba et al., 1987; Baetge and Hammang, 1991; Shoji-Kasai et al., 1992) and some of these lines have proven valuable for studying aspects of the neurotrophin signal transduction pathway (Green et al., 1986; Loeb et al., 1991). The analysis of PC12 variants with altered neurite outgrowth behaviour should likewise further our understanding of the molecular mechanisms of neuritogenesis and of neuronal morphogenesis. Here we report the isolation of a PC12 cell sub-clone, designated PC12-C41, that has significantly enhanced neuritogenic potential when compared with the wild-type parental line. To determine the possible cause of this difference, we have analyzed and compared the expression and post-translational modification of a number of cytoskeletal proteins in the two lines. Our data point in particular to the potential role of tau MAPs (reviewed by Lee, 1990) in generating morphological diversity of PC12 cells and raise the possibility of a similar morphoregulatory function for tau in the nervous system.

A portion of this study has previously appeared in abstract form (Teng and Greene, 1991).

NGF was prepared from adult mouse submaxillary glands as described (Mobley et al., 1976) and, unless stated otherwise, was used at a concentration of 50 ng/ml. Rat spinal cords were dissected from adult animals and the samples were prepared by immediate solubilization in 1 × SDS sample buffer (Laemmli, 1970). Dorsal root ganglia (DRG) were dissected and extracted with perchloric acid as described (Georgieff et al., 1991). The retroviral vector pZIP.Neo SVX (I) was propagated in the Ψ-2 cell line as described (Mann et al., 1983; Cepko et al., 1984) and was a gift from Dr Matthew Lo (ICI Americas, Wilmington, DE). Monoclonal antiserum 1B-4 raised against bovine MAP 1B (Bloom et al., 1985) was kindly provided by Dr Richard Vallee (Worcester Foundation for Experimental Biology, Worcester, MA). The neomycin analogue G418 (50% purity) was purchased from Sigma Chemical Co., St. Louis, MO, and was used at 1 mg/ml. Nocodazole was obtained from Aldrich Chemical Co., Milwaukee, WI, and was stored as a 50 mM stock (1000 ×) in DMSO. All radiolabeling reagents were obtained from Du Pont-NEN, Boston, MA, except Trans³⁵S-label™, which was purchased from ICN Biomedical, Inc., Irvine, CA. Protein A-Sepharose was purchased from Pharmacia LKB Biotechnology Inc., Piscataway, NJ. Protein concentrations were assayed by the method of Brad-ford (1976) with reagents purchased from Bio-Rad Laboratories, Richmond, CA, using bovine serum albumin (BSA) as the stan-dard.

To obtain clonal PC12 variants, wild-type PC12 cells were infected with the retroviral vector pZIP.Neo SVX(I) as previously described in detail (Kadan and Lo, 1990; Greene et al., 1991). The infected PC12 cells were selected in G418-containing medium for 2 weeks. A typical infection experiment (infection efficiency of 1 in 1000) produced isolated G418-resistant PC12 cells, which were allowed to grow into small colonies. These isolates were then treated with 50 ng/ml NGF for 1-2 weeks (in complete medium, while they were still attached to the original tissue cul-ture dishes) followed by visual identification of potential ones with altered neurite morphologies. These colonies were then picked, expanded into sub-clones and re-screened in the presence of 50 ng/ml NGF for alteration in neurite growth properties. Northern blotting analysis indicated the constitutive expression of the neomycin resistance gene in various lines of PC12 sub-clone but not in wild-type PC12 line from which they were derived. The PC12-C41 sub-clone was selected for further analysis on the basis of its heightened neuritogenic potential and increased neurite abundance after long-term NGF exposure.

PC12 and PC12-C41 cells were cultured on collagen-coated dishes in RPMI 1640 medium supplemented with 10% heat-inactivated horse serum and 5% fetal bovine serum (complete medium; Greene and Tischler, 1976; Greene et al., 1987). For NGF treat-ment, cultures were seeded at low density (5 ×10⁵ cells per 35 mm dish or 1 ×10⁶ to 2 ×10⁶ cells per 100 mm dish) in RPMI 1640 medium containing 1% horse serum or complete medium and NGF. Suspension cultures of PC12 and PC12-C41 cells were maintained as previously described, in uncoated plastic Petri dishes (Greene et al., 1987). In all experiments, comparisons between PC12 and PC12-C41 cells were carried out with parallel sets of cultures that received identical treatments.

Neurite-bearing cells were scored as previously described (Burstein and Greene, 1978). Only processes greater than 2 cell body diameters in length (i.e. about 20 μm) were considered to be neurites. For each culture condition, at least 100 cells from ran-domly chosen fields were scored.

Neurite stability after NGF withdrawal was assayed on long-term (> 10 days) NGF-treated PC12 and PC12-C41 cells by wash-ing-out NGF (by rinsing 3 × with NGF-free medium) and re-feed-ing the cultures with complete medium without NGF. The percentage of neurite-bearing cells was measured at various times as described above. The effect of a microtubule (MT) depoly-merizing agent on neurites was measured on long-term NGF-primed PC12 and PC12-C41 cells by adding 50 μM nocodazole (final concentration) to the cultures. After 2 days, the proportion of neurite-bearing cells was measured as described. NGF was present throughout the entire time of the experiment.

Neurite lengths were determined by the use of an eyepiece equipped with a calibrated micrometer. Twenty random neurites were assessed per culture condition.

Neurite regeneration assays were carried out with NGF-primed cells that had been mechanically divested of neurites and were scored as previously described (Burstein and Greene, 1978).

Photomicrographs were taken of random fields within PC12 and PC12-C41 cell cultures at either high (×200) or low (×100) mag-nification. Neurite densities were determined from enlarged prints (5 in × 7 in) by first drawing a line across the photographed field and then counting the numbers of neurites that intersect this line. The linear neurite density is defined as the number of intersections per unit length of the line (corrected for intervals occupied by cell bodies).

Metabolic radiolabeling of PC12 and PC12-C41 cell cultures with phosphate was performed in 35 mm dishes, as previously described, in 1 ml of modified Krebs-Ringer, HEPES-buffered solution by incubation for 2 hours at 37°C with 50-300 μCi/ml [³²P]orthophosphate (Black et al., 1986). Where applicable, the labeling medium contained NGF.

Whole cell lysates were prepared as described (Aletta and Greene, 1987) in SDS sample buffer and placed immediately in a boiling water bath for 10 minutes before analysis by SDS-PAGE. To harvest samples in suspension cultures, cells were first washed 3 × with PBS by centrifugation at 500 g for 5 minutes. The final pellets were then vortexed briefly to loosen cell clumps followed by gradual addition of 1 × sample buffer with continuous vortex-ing to ensure complete sample solubilization. The samples were boiled for 10 minutes before analysis by SDS-PAGE.

Preparations enriched in MTs and chartin MAPs were gener-ated from cells attached to 35 mm culture dishes as described (Aletta and Greene, 1987; Teng and Greene, 1993). Triton/glyc-erol (MT-stabilized) and calcium (MT-enriched) extraction buffers were used at final volumes of 500 μl. The extractions were ter-minated by adding the extracted material to 167 μl of 4 × SDS sample buffer and boiling immediately for 10 minutes.

Discontinuous slab gel electrophoresis (SDS-PAGE) was car-ried out on either 5% or 7.5% polyacrylamide. Where applicable, gradient gel electrophoresis was performed on 5% to 10% poly-acrylamide. Purified bovine MTs/MAPs (generously provided by Dr Wilfredo Mellado, Columbia University, New York, NY) were included with molecular mass standards (Bio-Rad Laboratories) to indicate the positions of tubulins and MAP 1B/1.2. The gels containing ³²P-labeled proteins were fixed, stained, dried and then exposed to Kodak XAR-5 X-ray film.

Samples separated by SDS-PAGE were electroblotted (50 mA, overnight) onto supported nitrocellulose (S & S, Keene, NH) as described by Towbin et al. (1979). Equal amounts of material were compared in each experiment. The uniformity of sample loading was verified by staining the blots in 2% Ponceau S/2% trichloroacetic acid (TCA) following by several rinses in distilled water. Blots were then blocked with 5% non-fat milk in PBS for at least 2 hours before further processing.

Monoclonal antibodies against bovine MAP 1B (1B-4; Bloom et al., 1985) were used at a 1:2000 dilution in PBS containing 1% BSA. β-Tubulin was detected with monoclonal antibody Tu-9B (Ferreira et al., 1987; 1:500 in 1% BSA/PBS). Peripherin protein was detected with a rabbit polyclonal antiserum (1:2000 dilution in 1% BSA/PBS) prepared against a synthetic peptide corre-sponding to the C-terminal portion of the protein (Aletta et al., 1988a). Polyclonal antiserum that recognizes both HMW and LMW taus (B19-1; Gache et al., 1990) was used at 1:4000 in 1%BSA/PBS.

For MAP 1B/1.2 and β-tubulin detections, blots were devel-oped with 350 nCi of ¹²⁵I-labeled goat anti-mouse IgG (Du Pont-NEN)/ml of PBS containing 1% BSA and were washed 5 × in PBS. Blots for peripherin and taus were developed for 60-90 min-utes with ¹²⁵I-labeled Protein A (Du Pont-NEN) at a final concentration of 350 nCi/ml of PBS containing 5% BSA, followed by 3 washes (20 minutes each) in 0.05% Triton X-100/PBS. After the washing steps, blots were exposed to XAR-5 X-ray film at −70°C with an enhancing screen.

Immunoprecipitation of ³²P-or ³⁵S-labeled taus from PC12 and PC12-C41 cells was performed as follows. The cultures were labeled either in 300 μCi/ml [³²P]orthophosphate as described above or in 100 μCi/ml Trans ³⁵S-label™ (L-[³⁵S]methionine: L-[³⁵S]cysteine) in methionine and cysteine-free RPMI 1640 medium (ICN Biomedical, Inc.) for 4 hours at 37°C. Heat-stable cellular extracts were prepared by harvesting and boiling the sam-ples in 1% SDS/PBS (100 μl) for 10 minutes. These were then diluted in 900 μl IP diluent (150 mM NaCl, 50 mM Tris, pH 8.0, 25 mM NaF, 1% Triton X-100, 5 mM EDTA, 5 mM EGTA, 2 mM PMSF, 100 units Aprotinin). The samples were incubated on ice for 5-10 minutes. All subsequent steps were carried out at 4°C. After centrifugation (10,000 g) for 15 minutes, the supernatants were pre-cleared in Protein A-Sepharose (6 mg/sample) for 45 minutes. Equal TCA-precipitable counts from each sample (adjusted to equal final volume) were subjected to immunopre-cipitation by overnight incubation with antiserum B19-1 (1:400 dilution). A 6 mg sample of Protein A-Sepharose was then mixed with each sample for 1-2 hours. The immunoprecipitates were washed 3 × with IP diluent (1 ml/sample). After a final rinse in IP diluent without Triton X-100 (1 ml/sample), the samples were sol-ublized and boiled in 1 × sample buffer. Equal volumes of each sample were analyzed by SDS-PAGE. Autoradiography for ³²P-labeled tau was performed on fixed, dried gels. Fluorograms of ³⁵S-labeled immunoprecipitates were obtained from salicylate-treated gels as previously described (Chamberlain, 1979).

Total cellular RNA was isolated from PC12 and PC12-C41 cells according to the method of Chomczynski and Sacchi (1987). Except for the detection of tau transcripts, RNA was subjected to electrophoresis at 150 V on 1% agarose-formaldehyde gels for 1 hour. To ensure adequate resolution of the closely migrating HMW and LMW tau transcripts, RNA samples were allowed to enter the gel slowly (1% agarose-formaldehyde) at 30 V (2-3 hours) and were then resolved at 60-100 V for an additional 2-3 hours.

RNA was transferred to GeneScreen Plus membrane according to the manufacturer’s protocol (DuPont-NEN). ³²P-labeled probes for c-fos (Curran et al., 1982), ODC (Volonté and Greene, 1990), HMW and LMW tau (BamHI fragment of the 3 ′ end of LMW and HMW tau; Georgieff et al., 1993) or HMW tau (exon 4a probe; Georgieff et al., 1993) were prepared by random priming reactions (Boehringer-Manheim, Indianapolis, IN). Blots were hybridized overnight at 37°C with radiolabeled probes and were washed at high stringency as described for GeneScreen Plus mem-brane. The equality of sample loading was verified by ethidium bromide staining of the gel. Sample normalizations were per-formed on stripped blots (according to manufacturer’s protocol) by re-hybridization to ³²P-labeled β-actin probe (clone 40; Leonard et al., 1987).

Densitometric scanning of autoradiograms was performed on a Molecular Dynamics model 300 Series Computing Densitometer (Molecular Dynamics, Sunnyvale, CA) or on an Apple OneScan-ner with Ofoto scanning software (Apple Computer Inc., Cuper-tino, CA) coupled to the NIH Image 1.43 gel analysis program (National Institutes of Health, Bethesda, MD). Only autoradi-ograms that were neither over-nor under-exposed were used for quantitation. The absorbance (within the linear range detectable by the scanner) of each band was expressed as total numbers of pixels in the area covered by that band. For quantification of 64 kDa chartin (Fig. 5C) and β-tubulin (Fig. 5D) phosphorylations, an invariant band (≈ 85 kDa) in each lane was employed for nor-malization. Where applicable, statistical analyses were performed using Student’s two-tailed t-test. The difference between two means was considered significant if P ≤ 0.05.

To characterize cellular components that may play regulatory roles in neurite outgrowth, we sought to establish PC12 cell sub-clones with altered neuritogenic responses to NGF. However, because the inherent tendency of PC12 cells to aggregate complicates conventional sub-cloning proce-dures, we employed the following approach to generate clonally derived PC12 sublines. Wild-type PC12 cultures were infected with the retroviral vector PZIP.NeoSVX (I) under conditions that produced an infection efficiency of 1 in 1000 (Kadan and Lo, 1990; Greene et al., 1991). Since the retroviral vector contains the neomycin resistance gene driven by the Moloney murine leukemia virus long termi-nal repeat promoter (Cepko et al., 1984), isolated PC12 cells appeared after selection in G418-containing medium. Individual cells were allowed to grow into small colonies, which were then exposed to NGF for 1-2 weeks. A total of approximately 150,000 colonies (average 5 ×103 colonies per experiment; 30 independent infection experiments) were visually inspected. Amongst them, 50 sub-clones appeared to exhibit unusual neuritic growth in response to NGF and were therefore isolated and expanded into cell lines for further analysis. After an additional round of re-screening in NGF, one sub-clone, designated PC12-C41, was found consistently to display a denser neurite network than the parental PC12 line. Southern blotting analysis of genomic DNA from PC12-C41 cells revealed that they are clonal and contain a single retroviral integration site within their genome (data not shown).

Fig. 1 compares the morphologies of PC12 and PC12-C41 cells before and after exposure to NGF. Without NGF, PC12-C41 cells closely resemble the parental PC12 line; no cytoplasmic extensions or neurites are present. With NGF, both lines began to extend neurites within 2-3 days, and by 2 weeks of treatment, both displayed dense neurite networks. However, the neurite network of NGF-treated PC12-C41 cells is consistently of greater density (compare Fig. 1C and D). To quantify this difference directly, we counted the number of neurites that intersected arbitrary lines drawn across photomicrographs of randomly chosen fields of cultures that had been treated with NGF for 2-3 weeks (see Materials and Methods). As shown in Table 1, NGF-treated PC12-C41 cell cultures consistently display a 2-to 3-fold (average 2.5 (± 0.5)-fold; P < 0.02) increase in neurite density in comparison with PC12 cell cultures exposed to NGF for comparable periods of time. This dif-ference in neurite density in turn gives rise to the apparent variation in morphology between cultures of NGF-primed PC12 and PC12-C41 cells.

We next examined the time courses of NGF-elicited neurite outgrowth in PC12 and PC12-C41 cell cultures. Fig. 2A shows that after an initial lag, PC12-C41 cells generate neurites more rapidly than PC12 cells. For instance, the times for NGF to induce 50% of PC12 and PC12-C41 cells to extend neurites are 8 and 6 days, respectively. Interest-ingly, as shown in Fig. 2B, there is no significant differ-ence in the mean neurite length in PC12 and PC12-C41 cell cultures. Data from 5 independent experiments further revealed that the mean rates for neurite elongation are 42.3 (± 6) μm and 45.3 (± 7) μm per day for PC12 and PC12-C41 cell cultures, respectively.

To determine if additional neuritic properties differ between the two cell types, we compared the stability of neurites in NGF-primed PC12 and PC12-C41 cell cultures. Fig. 2C shows that the neurites of PC12-C41 cells are substantially more resistant to loss after NGF withdrawal. In this exper-iment, while 90% of the neurites in long-term NGF pre-treated PC12 cell cultures disintegrated after 4 days of NGF deprivation, neurites of PC12-C41 cells that received the same treatment showed no sign of loss until 10-12 days of NGF withdrawal. We also examined the effect of the MT depolymerizing agent nocodazole on PC12 and PC12-C41 neurites. As shown in Fig. 2D, after 2 days of treatment with 50 μM nocodazole only 12% of the NGF-primed PC12 cells were neurite-bearing and these remaining neurites dis-played marked signs of deterioration. In contrast, the neu-rites of PC12-C41 cells were more resistant to this treat-ment. More than 90% of the neurites in PC12-C41 cultures remained intact after 2 days of exposure to nocodazole, although some morphological defects in the neurite network were apparent.

Since enhanced responsiveness to NGF may lead to an increase in the rate of neurite production (Hempstead et al., 1992), we examined whether the signal transduction path-way of PC12-C41 cells is significantly more sensitive than that of PC12 cells. First we measured the dose responsive-ness to NGF in a neurite regeneration assay. As illustrated in Fig. 3, the dose-response curves for both lines are indis-tinguishable. Furthermore, naive PC12 and PC12-C41 cells show a similar NGF concentration dependence for neurite initiation (data not shown). Consistent with this, the gp140^prototrk NGF receptor (Kaplan et al., 1991) exhibits comparable levels of NGF-induced autophosphorylation in both lines (data not shown).

We next examined induction of the c-fos and ornithine decarboxylase (ODC) genes, two representative examples of early PC12 cellular responses to NGF (Volonté and Greene, 1990; Batistatou et al., 1992). Previous studies have established that induction of c-fos mRNA is independent of macromolecular synthesis (Milbrandt, 1986) while that of ODC requires active protein synthesis (Feinstein et al., 1985). The data in Fig. 4 reveal that both the kinetics and magnitudes of c-fos (Fig. 4A) and ODC (Fig. 4B) mRNA induction are comparable in PC12 and PC12-C41 cultures.

Previous studies have suggested that the interaction amongst various cytoskeletal elements plays an important role in NGF-induced neuritogenesis and that alteration in their composition is likely to affect the properties of grow-ing neurites (Nunez, 1986; Matus, 1988; Aletta et al., 1990). We therefore sought to determine whether PC12-C41 cells differ in any respect from PC12 cells regarding cytoskele-tal composition.

MAP 1.2 (also known as MAP 1B: Bloom et al., 1985; and MAP 5: Brugg and Matus, 1988) is a high molecular weight MAP whose protein levels and phosphorylation states in PC12 cells are regulated by NGF (Greene et al., 1983; Drubin et al., 1985; Aletta et al., 1988b). Fig. 5A shows the relative abundance of MAP 1.2 protein in PC12 and PC12-C41 cells (with or without NGF treatment) as determined by western immunoblotting. As previously reported for PC12 cells (Aletta et al., 1988b), long-term NGF treatment of PC12-C41 cells leads to a substantial increase in MAP 1.2 protein levels. Quantification indicated that there is a 30% greater increase in MAP 1.2 protein in NGF-primed PC12-C41 cells, but that this change does not reach statistical significance (P > 0.5). The relative levels of phospho-MAP 1.2 in PC12 and PC12-C41 cells before and after various times of NGF treatment are presented in Fig. 5B (n = 4). Consistent with prior reports (Greene et al., 1983; Black et al., 1986; Aletta et al., 1988b; Teng and Greene, 1993), there was a large increase in this parameter after NGF treatment. While there was no difference (P < 0.2) in the phospho-MAP 1.2 level in long-term (> 13 days) NGF-treated PC12-C41 cells as compared to that of NGF-treated PC12 cells, statistical analysis indicates that the rel-ative phosphorylation of MAP 1.2 between the two cell lines differs significantly at earlier times of NGF exposure (2-4 days, P < 0.02; 7 days, P < 0.05).

Previous studies have shown that long-term NGF treatment leads to increases in β-tubulin and chartin MAP phos-phorylation in both PC12 and PC12-C41 cells (Black et al., 1986; Aletta and Greene, 1987; Teng and Greene, 1993). The data in Fig. 5C indicate that there is no significant dif-ference between the two lines in phosphorylation of the 64 kDa chartin, in either the basal or NGF-induced states. While the 72 and 80 kDa chartin MAPs did not resolve adequately to permit reliable quantification by densitome-try, their phosphorylation levels in NGF-treated PC12-C41 cells also do not appear to differ significantly from those in NGF-treated PC12 cells (data not shown). Examination of phospho-β-tubulin in NGF-primed PC12 and PC12-C41 cells (Fig. 5D) reveals a small (35%), but non-statistical (P > 0.5), difference in its relative phosphorylation. As shown in Fig. 5E, western immunoblotting analysis of PC12 and PC12-C41 cellular extracts also shows a minor (20%), but not statistically significant, increase (n = 5; P < 0.2) in the β-tubulin protein level in NGF-treated PC12-C41 cells as compared to that of PC12 cells.

Peripherin is the major intermediate filament of PC12 cells and is induced by NGF (Leonard et al., 1988; Aletta et al., 1989). Whole-cell extracts from naive and long-term NGF-treated PC12 and PC12-C41 cells were compared for peripherin levels by western immunoblotting analysis. Although the results in Fig. 5F show an average 50% increase in peripherin expression in NGF-primed PC12-C41 cells as compared to NGF-treated PC12 cells, statistical analysis revealed that this difference is not significant (P < 0.2).

Previous studies have demonstrated the presence of both HMW and LMW taus in PC12 cells and their induction by NGF (Drubin et al., 1985, 1988; Hanemaaijer and Ginzburg, 1991). Since tau proteins appear to affect MT formation, stability and cross-linking (reviewed by Lee, 1990), we therefore characterized and compared their expression in PC12 and PC12-C41 cells.

Fig. 6A illustrates the time-course of HMW and LMW tau protein induction by NGF in PC12 and PC12-C41 cells as revealed by western immunoblotting analysis with a polyclonal antibody (B19-1) that recognizes HMW and LMW taus equally well (Georgieff et al., 1991). The data indicate that naive PC12 and PC12-C41 cells express both HMW and LMW taus. LMW tau MAPs in both cell types consist of 2 major (M_r ≈ 59,000 and 52,000) and at least 1 minor (M_r ≈ 49,000) isoform (referred to collectively as LMW tau). This is in accord with earlier findings that PC12 cells express multiple LMW tau isoforms (Drubin et al., 1985, 1988). The HMW taus in PC12 and PC12-C41 cells also exhibit heterogeneity with 2 prominent isoforms (M_r≈ 113,000 and 103,000; referred to collectively as HMW tau). Both relative molecular mass forms of tau exhibit a gradual NGF-dependent up-regulation, which generally parallels the rate of neurite outgrowth. Note that HMW tau expression in PC12-C41 cells is higher than that of PC12 cells at all times, while the converse holds true for the LMW tau isoforms.

Fig. 6B compares the expression of tau isoforms in adult rat spinal cord and DRG with the PC12-C41 cell line. Inter-estingly, the patterns are distinct for both HMW and LMW taus. Particularly notable is that of the HMW tau; only a single species of M_r ≈ 110,000 is detected in the spinal cord and DRG, while PC12 and PC12-C41 cells possess two dis-tinct isoforms. In addition, the two cell lines appear to express multiple LMW tau variants different from those of the adult spinal cord.

Although PC12 and PC12-C41 cells exhibit a qualita-tively similar pattern of tau expression, several quantitative differences were apparent. As shown in Fig. 7A, the rela-tive abundances of LMW tau in naive PC12 cells and PC12-C41 cells are similar (P < 0.1), although the level averages 30% lower in the latter line. Consistent with previous reports (Drubin et al., 1985, 1988; Hanemaaijer and Ginzburg, 1991), NGF treatment results in an up-regulation of LMW tau in PC12 cells. An increase is also exhibited by PC12-C41 cells. However, while long-term NGF expo-sure yields a 5-fold increase in the relative abundance of LMW tau in PC12 cells, a more moderate induction (aver-age 2.8-fold) was observed for PC12-C41 cells. As a result of this difference, the relative levels of LMW tau in NGF-primed PC12 cells are about 2-to 3-fold (P < 0.05) higher than those in NGF-primed PC12-C41 cells.

The data in Fig. 7B show the relative levels of HMW tau in naive and NGF-treated PC12 and PC12-C41 cell cul-tures. Only a small, non-significant (P < 0.2) difference in HMW tau levels was observed between the two lines in their naive states. Moreover, NGF treatment does not appear to alter the expression of HMW tau in PC12 cells (P > 0.5). In contrast, there is a significant induction of HMW tau proteins by NGF in PC12-C41 cells, so that the relative abundance of HMW tau in NGF-primed PC12-C41 cells is 3.3-fold (P < 0.005) higher than that of compara-bly treated PC12 cells. As in the case of LMW tau, no dif-ferential modulation of HMW isoforms was noted in either cell line.

We also compared the induction of total cellular tau pro-tein, i.e. HMW tau plus LMW tau, by NGF in PC12 and PC12-C41 cells. Fig. 7C indicates that NGF treatment results in a similar increase in total tau levels in the two cell types (averaging 2.2- and 2.4-fold for PC12 and PC12-C41 cells, respectively). In addition, the data show no sta-tistically significant difference in the total tau abundance for either naive (P < 0.5) or NGF-primed (P < 0.5) PC12 and PC12-C41 cells. Our data thus reveal that although NGF does not cause a significant difference in total tau con-tent of PC12 and PC12-C41 cells, it does differentially affect the induction of LMW and HMW taus in the two lines. This is illustrated more clearly in Fig. 7D, which shows the ratio of HMW tau to total tau protein in the two cell lines before and after NGF treatment. While the rela-tive ratio of HMW tau to LMW tau remains unchanged in PC12-C41 cultures (± NGF), that in PC12 cells falls by 2-fold (P < 0.001) after exposure to NGF. The differential NGF-dependent modulation of tau MAPs represents the most prominent difference that we have observed between the cytoskeletal protein composition of PC12-C41 cells and the parental PC12 cell line.

To determine whether the increase in HMW tau abundance in PC12-C41 cells might be the consequence, rather than the possible cause, of the heightened neuritogenic response to NGF, both these and PC12 cells were exposed to NGF in suspension, i.e. under conditions non-permissive for neu-rite outgrowth. Western immunoblotting analysis of taus in these cultures as well as in replicate cultures that were maintained on a collagen substratum indicates that both HMW and LMW taus are induced by NGF even when neu-rite outgrowth is prevented (Fig. 8). The experiment also shows that the differences in HMW and LMW tau expression between PC12 and PC12-C41 cells persist in suspension (± NGF). This suggests that the high degree of HMW tau induction in PC12-C41 cells by NGF is an intrin-sic property of this line and not a secondary effect due to higher neurite density.

It has been reported that both HMW tau and LMW tau are phosphoproteins (Drubin et al., 1984) and that phosphory-lation may affect the biochemical properties of at least LMW tau (Lindwall and Cole, 1984). We therefore addressed the issue of whether long-term NGF treatment affects the phosphorylation of HMW and LMW taus in the two cell types. PC12 and PC12-C41 cells (before and after NGF exposure) were labeled with [³²P]orthophosphate or [³⁵S]methionine, [³⁵S]cysteine, cellular extracts were sub-jected to immunoprecipitation with anti-tau antiserum (B19-1; Gache et al., 1990) and the immunoprecipitates were analyzed by SDS-PAGE. The results shown in Fig. 9 indicate that both HMW and LMW taus are phosphorylated in the two cell lines and to a degree reflecting their rela-tive abundances and rates of synthesis (compare Fig. 6 with 9). Moreover, the data suggest that NGF does not substan-tially or differentially affect the rate of phosphate incorpo-ration into HMW taus or LMW taus in either PC12 or PC12-C41 cells.

Northern blotting analysis was performed to characterize the tau transcripts in PC12 and PC12-C41 cells and to determine the effects of NGF on their expression. The results obtained with a generic tau probe are shown in Fig. 10A. Two messages are detected of apparent sizes 6 and 8 kb. These correspond to the previously reported sizes of tran-scripts encoding LMW and HMW taus, respectively (Drubin et al., 1988; Couchie et al., 1992; Goedert et al., 1992; Georgieff et al., 1993). The results from this and an additional northern blot are quantified in Fig. 10C, D. That the 8 kb transcript represents a distinct HMW tau-encod-ing mRNA was corroborated by blots performed with a probe that specifically recognizes this transcript (Fig. 10B). The quantification of two independent experiments is shown in Fig. 10. These data reveal moderate up-regula-tion (average 2-to 4-fold) of the 6 kb LMW tau message by NGF in both PC12 and PC12-C41 cells (Fig. 10C). NGF also promoted a several-fold increase in 8 kb HMW tau mRNA (Fig. 10D, E). Note that the differences observed above with respect to LMW and HMW tau proteins (±NGF) in PC12 and PC12-C41 cells (Fig. 6) were gener-ally reflected by the relative levels of 6 and 8 kb tau mRNA (Fig. 10).

It has long been recognized that neuronal cytoskeletal pro-teins play key roles in the morphogenesis of neurons and, in particular, in the generation and subsequent maintenance of axonal and/or dendritic networks (Matus, 1988; Nunez, 1988; Ginzburg, 1991). However, the nervous system con-tains many different neuron types, each endowed with unique morphological characteristics. Paradoxically, the limited numbers of neuronal MAPs thus far identified appear to be present in diverse classes of neurons (Matus and Riederer, 1986; Tucker et al., 1988). This raises the question of how the nervous system achieves its immense morphological diversity with little apparent specificity of expression of these cellular components. One possible means of accomplishing this is by differential variation in the relative abundances of cytoskeletal proteins that mod-ulate various parameters of neurite growth.

Cell culture models, including rat PC12 cells, have proven suitable for studying potential roles of cytoskeletal proteins in regulating neuritogenesis. For example, previous studies have gathered valuable information by either selectively over-(Baas et al., 1991; Knops et al., 1991; Lewis et al., 1989) or under-expressing (Teichman-Weinberg et al., 1988; Caceres and Kosik, 1990; Caceres et al., 1991, 1992; Dinsmore and Solomon, 1991; Hanemaaijer and Ginzburg, 1991; Troy et al., 1992; Brugg et al., 1993) specific cytoskeletal proteins in appropriate cell types. An alterna-tive, but complementary, methodology is to isolate natu-rally arising morphological variants, which are then ana-lyzed to identify underlying cellular changes in the cytoskeleton. A potential advantage of this approach is that it is based on physiological levels of expression and does not depend on pre-conceived notions of which molecules might be important for regulating neuritogenesis. We there-fore have isolated PC12 sub-clones with altered neuritic growth properties and have begun systematically to com-pare these variants with parental PC12 cells in order to reveal determinants of neuronal form. The PC12-C41 line represents one such example.

The most prominent morphological difference between PC12 and PC12-C41 cells is a 2-to 3-fold increase in neu-rite abundance in the latter after comparable periods of NGF exposure. This does not appear to be the result of overall increased responsiveness to NGF or overt alteration of the NGF signaling mechanism. Although PC12-C41 cells dis-play a somewhat faster rate of neurite initiation, the rates of neurite extension as well as the final proportion of neu-rite-bearing cells are similar in PC12 and PC12-C41 cell cultures. Therefore, the difference in neurite initiation between the two cell lines should play little or no part in regulating neurite density. In addition, light and electron microscopic observations (K. Brown, K. K. Teng and J. M. Aletta, not shown) reveal no apparent difference in neurite fasciculation that might cause a distinction in neurite mor-phology.

It is of possible relevance to the morphology issue that the neurite network in PC12-C41 cultures is substantially more resistant to NGF withdrawal and to nocodazole. Pre-liminary analysis of the nocodazole-treated PC12-C41 cells indicates that their cytoskeletons are more resistant to MT depolymerization because, in contrast to PC12 cells (Aletta and Greene, 1987), a substantial portion of phosphorylated β-tubulin and chartins are recoverable in the cytoskeletal fraction (K. K. Teng, data not shown). Our data therefore suggest that neurites of PC12-C41 cells are more stable than those of PC12 cells, which in turn correlates with the greater neuritogenic response of the former. Increased neurite sta-bility could permit the formation and maintenance of enhanced numbers of neurites and of neurite branches (Black and Greene, 1982; Ferreira et al., 1989, 1990; Cac-eres and Kosik, 1990; Ferreira and Caceres, 1991; Caceres et al., 1991, 1992).

Because of the potential roles of cytoskeletal proteins in regulating neurite initiation and stabilization, a panel of NGF-regulated cytoskeletal elements was examined to determine if the expression of any of these is differentially affected in PC12-C41 cells. Our results show that most of these molecules exhibit no significant difference in protein and/or phosphorylation levels between the two cell types, either before or after long-term NGF treatment. Interest-ingly, there is a small but statistically significant increase in MAP 1.2 phosphorylation in PC12-C41 cells (as com-pared to PC12 cells) between 2 and 7 days of NGF expo-sure. Previous studies have suggested that phosphorylation of MAP 1.2 may be an important regulatory step during neuritogenesis (Greene et al., 1983; Aletta et al., 1990). It is therefore noteworthy that the difference in MAP 1.2 phosphorylation temporally parallels the difference in rate of neurite initiation for the two cell lines. A recent report (Brugg et al., 1993) that treatment of PC12 cells with MAP 1.2/1B antisense oligonucleotides abolishes NGF-induced neurite initiation is consistent with a neuritogenic role for MAP 1.2.

Initial characterization of tau from the central nervous system (CNS) revealed that it consists of a group of 55-70 kDa polypeptides (reviewed by Lee, 1990). Drubin et al. (1984) first reported the presence of a HMW tau-immunore-active species in PC12 cells. Recent immunological (Georgieff et al., 1991) and molecular cloning analyses (Couchie et al., 1992; Goedert et al., 1992; Georgieff et al., 1993) have established that this protein is indeed a HMW form of tau that is encoded by the same gene as LMW tau isoforms, but that differs from the LMW species in the inclusion of a 762 bp coding exon near the 5 ′ portion of its mRNA. Additional studies (Peng et al., 1986; Georgieff et al., 1991, 1993) have shown that HMW tau appears to be localized to the peripheral nervous system (PNS) and that although both HMW and LMW taus are present there during development, the mature PNS expresses the HMW form almost exclusively (Georgieff et al., 1991; Oblinger et al., 1991). The C terminus of LMW tau contains three to four homologous repeats that function as MT-binding and assembly-promoting sites (Ennulat et al., 1989; Lee et al., 1989; Lewis et al., 1989). These repeats are present in the deduced sequence of HMW tau (Couchie et al., 1992; Goedert et al., 1992; Georgieff et al., 1993) and antibody raised against them also recognizes HMW tau in PC12 and PC12-C41 cells (K. K. Teng, unpublished data). In addition, the MT cross-linking/bundling domain identified in the N-terminal half of LMW tau (Kanai et al., 1992) is present in the HMW tau molecule. These findings thus suggest that HMW tau shares similar MT-directed activities with the LMW tau isoforms. In agreement with this, overexpression of either LMW tau (Knops et al., 1991; Baas et al., 1991) or HMW tau (T. F. Frappier, I. S. Georgieff, K. Brown and M. L. Shelanski, unpublished data) in non-neuronal sf9 cells leads to the formation of MT-filled, neurite-like cytoplas-mic extensions. Such observations imply that both LMW and HMW taus can play critical roles in neurite outgrowth. This is strongly supported by findings that antisense oligonucleotides that should affect the synthesis of both HMW and LMW taus inhibit the establishment of long, stable neuritic processes in PC12 cells (Hanemaaijer and Ginzburg, 1991) and cultured cerebellar neurons (Caceres et al., 1991).

To assess the possible role of taus in the differential neu-rite-generating capacities of PC12 and PC12-C41 cells, we quantified HMW, LMW and total tau levels before and after long-term treatment with NGF. Interestingly, we found con-trasting actions of NGF on LMW and HMW tau in the two cell lines. Thus, NGF promoted a 3-fold greater specific activity of HMW tau in PC12-C41 cells and a comparably greater induction of LMW tau in the parental line. The net effect of NGF is therefore to bring about a profound dif-ference in the ratios of HMW and LMW taus in the two cell types. Prior to NGF exposure, 70-80% of the total tau protein in both cell lines is in the HMW form; after long-term NGF treatment, the tau content of PC12 cells switches to 70% LMW tau whereas in PC12-C41 cells 80% remains in the HMW form. It is of interest to note that NGF-treated PC12 cells express a ratio of HMW to LMW taus that is more characteristic of immature peripheral neurons, while in comparably treated PC12-C41 cultures, this ratio is closer to that in the mature PNS (see Georgieff et al., 1991; Oblinger et al., 1991, for examples).

One interpretation raised by our data is that the con-trasting ratios of HMW to LMW taus in PC12 and PC12-C41 cells at least partly contribute to their distinct degrees of neuritogenic capacity. The fact that this difference in tau expression may be a possible cause rather than a conse-quence of increased neurite outgrowth is indicated by the enhancement of HMW tau protein (Fig. 8) and mRNA (unpublished observations) levels in PC12-C41 cells even under NGF treatment conditions in which neurite outgrowth is prevented. Since both HMW and LMW taus are encoded by the same gene (Himmler, 1989; Kosik et al., 1989), our results evoke the possibility that an intrinsic difference between the PC12 and PC12-C41 cells may lie at the gen-eration (or the regulation thereof) of differentially spliced transcripts for HMW and LMW taus.

Potential mechanisms by which elevated HMW tau may lead to changes in neuritogenic capacity would include effects on MT assembly, cross-linking and stabilization (Horio and Hotani, 1986; Kirschner and Mitchison, 1986). Although there is currently no information that points to functional differences between the two molecular weight forms of tau, evidence has recently been obtained that the spaces between adjacent MTs in HMW tau-expressing sf9 cells are significantly greater than those that express the LMW isoforms (T. F. Frappier, I. S. Georgieff, K. Brown and M. L. Shelanski, unpublished). Also, it is interesting to note that Kanai et al. (1992) have shown that the MT-bundling ability of LMW tau increases as the length of the N-terminal half of the molecule is extended. Since the deduced sequence of HMW tau predicts an even longer N terminus extension, HMW tau might be more effective than the LMW isoforms in cross-linking MTs and consequently may play a more effective role in neurite stabilization. Such a possibility is in accord with our observations that PC12-C41 cell neurites are more resistant to NGF removal as well as to nocodazole. The observation that the ratio of LMW to HMW tau decreases during development of the PNS (Georgieff et al., 1991; Oblinger et al., 1991) lends further support to this notion, since the adult axonal cytoskeleton is generally considered to be of greater stability. In this regard, Oblinger et al. (1991) reported that axotomy of rat DRG neurons is accompanied by a decrease in HMW tau abundance and suggested that this protein has a role in sta-bilizing the mature axonal cytoskeleton.

There have been several studies on tau induction in PC12 cells. Hanemaaijer and Ginzburg (1991) showed that NGF treatment of PC12 cell cultures leads to a transition from immature to mature LMW tau isoforms. However, by west-ern immunoblotting analysis, we were unable to detect such changes. In other studies, Drubin and co-workers reported 10-(Drubin et al., 1985) and 30-to 40-fold (Drubin et al.,1984) increases in the specific levels of total PC12 cell tau after NGF treatment (caused predominantly by enhance-ment of LMW isoforms). In contrast, we find that tau induc-tion by NGF is substantially less (≈ 2-fold for total tau and 5-fold for LMW tau). A possible reason for this discrep-ancy may lie in the relative specificity of the tau antibod-ies used in the two studies. The antiserum employed here was generated against a synthetic peptide corresponding to the N-terminal portion of the molecule (Gache et al., 1990) and appears to recognize all tau forms equally well (Georgi-eff et al., 1991). The reagent used by Drubin et al. (1985, 1988) was obtained from affinity-purified antiserum against taus purified from bovine brain (Pfeffer et al., 1983). It is conceivable that this antiserum shows differential recogni-tion of tau in NGF-treated vs naive PC12 cells due to post-translational modification, such as phosphorylation (Cole-man and Anderton, 1990; Ueda et al., 1990).

The issue of the degree to which NGF induces tau pro-tein is an important one, since it bears on the question of the role of tau in neurite outgrowth. In contrast to the con-clusion drawn by Drubin et al. (1985, 1988), our findings suggest that promotion of neurite growth does not require a substantial increase in the specific cellular levels of tau proteins and that relatively small changes in tau may yield significant alterations in neuronal morphology.

Our findings confirm the observations of Drubin et al. (1984) that all major tau isoforms in NGF-treated PC12 cells are phosphorylated. We have also extended this work by examining tau phosphorylation in naive PC12 cells and in naive and NGF-treated PC12-C41 cells. These experi-ments revealed that in contrast to its effects on MAP 1.2/1B (Aletta et al., 1988b) and chartins (Aletta and Greene, 1987), NGF does not promote major changes in the phos-phorylation of taus. That is, in all cases examined, the major tau isoforms were phosphorylated to extents that reflected their protein levels. However, our data do not rule out the possibility that NGF exposure may lead to changes in the specific sites at which taus are phosphorylated.

Recent observations have established that HMW and LMW taus are encoded by distinct mRNAs (6 and 8 kb, respectively; Couchie et al., 1992; Goedert et al., 1992; Georgi-eff et al., 1993). We found that NGF produced relatively small (2-to 4-fold) changes in the relative levels of these messages in both PC12 and PC12-C41 cells. The magnitude of these changes appears to be comparable to the range observed by Drubin et al. (1988). Within the limits of the variation of our data, we found a reasonably good correla-tion between the relative steady state protein levels, incor-poration of [³⁵S]methionine, [³⁵S]cysteine during a short-term incubation, and levels of corresponding mRNAs for both the HMW and LMW taus. Nevertheless, a relatively minor exception was noted among these parameters; although NGF appeared to increase the level of HMW tau mRNA in PC12 cells, the levels of this protein did not change. One plausible explanation for this discrepancy may be that tau protein levels are subject to post-translational as well as transcriptional regulation (Drubin et al., 1988; Char-rière-Bertrand and Nunez, 1992).

In conclusion, we have described a PC12 sub-clone, namely PC12-C41, that extends a significantly more abun-dant and more stable neuritic network in response to NGF. Analysis of various properties of this sub-clone after NGF treatment indicates few changes other than an altered ratio of HMW to LMW taus. This has led us to speculate that HMW tau may be more efficacious in promoting neurite stability and thereby in permitting an increase in neurite density. Thus, subtle variation among cytoskeletal compo-nents may be a means of generating neuronal morphologi-cal diversity. In view of the observation that HMW tau pro-tein in PC12-C41 cells is significantly elevated upon NGF treatment, this cell line should be a useful alternative to the parental PC12 cell line for future investigation on HMW tau functions, its regulation by NGF and relationship to neu-ritogenesis. Finally, our findings indicate that promotion of neurite outgrowth by NGF neither requires nor is accom-panied by massive induction of tau expression.

We thank Drs Richard Vallee and Wilfredo Mellado for gen-erously providing antiserum 1B-4 and bovine MAPs, respectively. We also thank Dr Douglas Baird for suggestions on measurement of neurite density, and Dr Matthew Lo for gifts of retroviral vector and packaging cell line, and advice on retroviral infection proce-dure. This work was supported by grants from NIH-NINDS.