Expression of high molecular weight tau in the central and peripheral nervous systems

Microtubule associated proteins (MAPs) are a group of proteins that associate with microtubules in vivo and in vitro, and are involved in regulating microtubule assembly and function. The two major classes of MAPs are the microtubule associated motors (e.g. kinesin, dynein and dynamin) and the microtubule assembly promoters. This latter class of MAPs can be further subdivided into heat stable MAPs (MAP2, MAP4 and tau) and heat labile MAPs (MAP1a and MAP1b) (for reviews see Vallee, 1990; Bloom, 1992; Chapin and Bulinski, 1992). The heat stable MAPs contain highly conserved microtubule-binding domains near their C-termini, composed of three or four homologous 18 amino acid repeats separated by domains of 13 to 14 amino acids (Lee et al., 1988; Lewis et al.,1988; Chapin and Bulinski, 1991) and N-terminal regions of variable lengths and unknown functions. A single repeat of the microtubule binding domain can promote microtubule assembly (Ennulat et al., 1989) and the affinity of these proteins for microtubules increases with the number of repeats (Butner and Kirschner, 1991).

Tau proteins serve in promoting process formation in neurons, microtubule assembly in vivo and process stabilization in PC12 cells (Drubin et al., 1984; Caceres and Kosik, 1990; Henemaaijer and Ginzburg, 1991; Knops et al., 1991).

Tau proteins consist of two groups: the low molecular weight taus (LMW tau), which range in apparent molecular mass (Mapp) from 50-68 kDa on SDS-PAGE (Cleveland et al., 1979; Binder et al., 1985; Butler and Shelanski, 1986) and the more recently discovered high molecular weight tau (HMW tau) which has a Mapp of ∼110 kDa (Oblinger et al., 1991; Georgieff et al., 1991). LMW taus are highly expressed in central nervous system (CNS) neurons and are preferentially targeted to the axon (Binder et al., 1985; Shiomura and Hirokawa, 1987; Migheli et al., 1988). During development of the CNS, there is initial expression of a limited number of tau isoforms (Mareck et al., 1980; Couchie and Nunez, 1985), which increase with age until adulthood, when as many as thirty different isoforms can be resolved on two-dimensional gels (Butler and Shelanski, 1986). These forms are generated by a combination of alternative splicing of the tau message and differential phosphorylation of the resulting proteins (Lindwall and Cole, 1984; Baudier and Cole, 1987; Lee et al., 1988; Himmler, 1989; Kosik et al., 1989).

High molecular weight tau is preferentially expressed in the adult peripheral nervous system (PNS) (Georgieff et al., 1991; Oblinger et al., 1991). In the newborn rat dorsal root ganglion (DRG) both high and low molecular weight proteins are seen, while in the adult DRG only HMW tau is present (Georgieff et al., 1991; Oblinger et al., 1991). Small amounts of HMW tau are observed in the CNS, and have been thought to be confined to afferent sensory fibers arising in the periphery (Georgieff et al., 1991). HMW tau cDNAs have recently been cloned from the rat PC12 pheochromocytoma and mouse N115 neuroblastoma tumor cell lines (Couchie et al., 1992; Goedert et al., 1992). Each of these contains a novel region of over 700 bp inserted between exons 4 and 5 of LMW tau.

In the studies that follow, we have used a novel, threestep polymerase chain reaction (PCR) to clone a fully encoding cDNA for HMW tau from rat dorsal root ganglia and used this cDNA to generate specific probes and antibodies to study the developmental expression, distribution and heterogeneity of HMW tau.

Cloning of HMW tau was carried out by polymerase chain reaction (PCR) using total RNA isolated from adult rat dorsal root ganglia. Specifically, first strand cDNA synthesis was performed with random hexamer primers using a RNA PCR kit (Perkin-Elmer Cetus, Norwalk, CT). Amplification was carried out in two steps according to the manufacturer’s protocol (Perkin-Elmer Cetus). PCR product 1 was generated with primers to exon 1 (ex1) and the junction of exon 7 and 9 (ex7/9) of low molecular weight tau:

PCR product 2 was generated with primers to exon 7 (ex7) and exon13 (ex13) of low molecular weight tau:

Each of these PCR products was digested at the EcoRI and HindIII restriction sites that were present in the primers. The digested inserts were gel purified and subcloned into pGEM3Z, which had been previously digested with EcoRI and HindIII. The fully encoding HMW tau cDNA was generated by amplifying PCR products 1 and 2, using primers ex1 and ex 13 for 35 cycles (94°C for 60 s; 60°C for 120 s; 72°C for 180 s). In the first round of amplification products 1 and 2 serve as the primers and extension occurs at their 3′ ends. In subsequent amplification steps, primers ex1 and ex13 are used. The final PCR product was subcloned into pGEM3Z (pDRGTAU) in the same manner as described above.

Sequencing fragments were generated by unidirectional deletions with the Erase-a-Base System (Promega, Madison, WI). Both strands of all clones were sequenced by the dideoxy chain termination technique (Sanger et al., 1977) using the Sequenase Version 2.0 kit (USB Corp., Cleveland, OH). Sequence analysis was carried out with IBI/Pustell sequence analysis software (IBI, New Haven, CT).

RNA transcripts of HMW tau were synthesized in vitro using either the SP6 or the T7 promoter of the pGEM3Z plasmid containing the 2.1 kb cDNA after linearization with ScaI. Transcribed RNA was translated in a rabbit reticulocyte lysate system (NEN, Wilmington, DE) containing [35S]methionine. Reactions were mixed with sample buffer and separated on 7.5% SDS-polyacrylamide gels, which were subsequently dried and exposed to X-ray film.

Total RNA was isolated from rat tissues according to the method of Chomczynski and Sacchi (1987). RNA was electrophoresed on 1% or 0.8% agarose-formaldehyde gels and transferred to nitrocellulose membranes. Products generated by PCR were electrophoresed on 1% agarose gels and transferred to nitrocellulose membranes for Southern blot analysis. All northern and Southern blots were hybridized with cDNA probes labeled with 32P using a DNA labeling kit (randomly primed) from Boehringer-Manheim (Indianapolis, IN). Two probes were used: an exon 4a probe, which was obtained by amplifying this region by PCR (pEX4a); and a BamHI fragment corresponding to the common 3′ end of LMW and HMW tau.

An exon 4a-specific antibody was generated against a synthetic peptide spanning amino acids 360 to 374 of the exon 4a insert. The synthetic peptide and the antibody were produced by Immuno-Dynamics Inc. (La Jolla, CA). This antibody was affinity-purified on a fusion protein containing exon 4a, which was prepared in the following manner: oligonucleotide primers corresponding to the ends of exon 4a were used to amplify this region by PCR. The resulting 762 bp fragment was cloned into the pGEX-3X expression vector (Smith and Johnson, 1988). Following induction with 0.5 mM isopropyl-β-D-thiogalactopyranoside, the glutathione-S-transferase fusion protein was purified by affinity chromatography on glutathione-agarose (Guan and Dixon, 1991). The purified fusion protein was coupled to CNBr-activated Sepharose 4B (Sigma, St. Louis, Mo) and used for antibody purification (Kaplan et al., 1991).

Rat brain and spinal cord samples were homogenized in 1 volume of ice-cold reassembly buffer (0.1 M MES, 1 mM EDTA, 0.5 mM MgCl₂, 10 mM NaF and 1 mM PMSF, pH 6.8). After centrifugation at 100,000 g, the supernatants were recovered and protein concentrations were determined (Lowry et al., 1951). Dorsal root ganglia, trigeminal and sciatic nerves were homogenized directly in sample buffer. Proteins were electrophoresed on 7.5% SDS-polyacrylamide gels according to the method of Laemmli (1970). Proteins were transferred to nitrocellulose membrane and incubated overnight with either a polyclonal anti-tau antibody (Sigma, St. Louis, MO) or the anti-4a antibody. Immunoreaction products were visualized with 125I-labeled Protein A (ICN Biochemicals, Costa Mesa, CA), followed by exposure to X-ray film.

Rats were anesthetized and perfused with 4% paraformaldehyde in 0.1 M phosphate buffer (pH 7.4). Tissues were removed, postfixed for 4 h, and then transferred to 30% sucrose for 12 h. Sections were cut with a cryostat and stained with either the anti-tau (Sigma) or the anti-4a antibody followed by an anti-rabbit peroxidase antibody. Detection of immunoreactivity was with 3,3-diaminobenzidine (Sigma) and H₂O₂.

Frozen tissue sections were treated with 1 mg/ml proteinase K in 0.1% DEPC-treated dH₂O for 2 min, washed in DEPC-dH₂O, and incubated for 10 min each in 0.1 M triethanolamine and 0.25% acetic anhydride in 0.1 M triethanolamine. Slides were washed, dehydrated through a series of graded ethanols and air dried. Slides were hybridized overnight at 50°C in hybridization mix (0.6 M NaCl, 10 mM Tris, pH 7.5, 0.02% Ficoll, 0.02% polyvinylpyrroli-done, 0.02% BSA, 1 mM EDTA, 50% formamide, 10% dextran sulfate, 0.005% yeast tRNA, and 0.05% herring sperm DNA) containing 35S-labeled (500,000 c.p.m./section) riboprobes in the sense or antisense orientation specific for exon 4a, which were generated by SP6 or T7 polymerase using pEX4a as template. After hybridization, sections were rinsed in 0.5× SSC for 10 min and then digested with 20 μg/ml RNase A for 30 min at room temperature. Sections were washed twice in 2× SSC for 1 h at 50°C, and once in 0.2× SSC for 2 h at 50°C, dehydrated through a series of graded ethanols and air-dried. Slides were coated with NTB2 emulsion (Kodak) and stored for 1-6 weeks.

The experimental strategy we employed to isolate a cDNA clone for HMW tau from rat dorsal root ganglia was based on preliminary data suggesting that HMW tau may differ from LMW tau by an insertion of unknown length between exons 4 and 5 of LMW tau. Cloning of HMW tau cDNA was done by PCR using total RNA isolated from adult rat dorsal root ganglia. In view of the large coding region expected for HMW tau, we decided on a strategy in which we generated PCR products covering overlapping regions, each of which was equal to approximately half of the molecule. Four oligonucleotide primers were designed matching sequences within exons 1, 7, 9, and 13 of rat LMW tau (Fig. 1). First strand cDNA synthesis was carried out using random hexamer primers (see Materials and Methods). PCR amplification was accomplished in the following manner: PCR reaction with primers ex1 and ex7/9 (at the junction of exons 7 and 9) generated product 1 with a size of 1.2 kb, 0.7 kb longer than the predicted product for LMW tau. Primers ex7 and ex13 yielded PCR product 2, with a size of 0.9 kb, the predicted size for exons 7 to 13 of LMW tau. To obtain the complete DRG tau, PCR products 1 and 2 were then reamplified using primers ex1 and ex13. As shown in Fig. 1, there is an initial extension of PCR products 1 and 2 at their 3′ ends to form full length complementary strands that are then amplified using primers ex1 and ex13. The final 2.1 kb cDNA was subcloned into pGEM3Z for further analysis (pDRGTAU).

To determine whether this additional 0.7 kb insert in PCR product 1 was specific to DRGs, a similar amplification was carried out using brain mRNA with primers ex1 and ex7/9, yielding a 0.5 kb product, which is the expected size for this region of LMW tau.

Sequencing pDRGTAU revealed a 2080 bp cDNA with extended sequence homology to LMW tau (Fig. 2). A 762 bp insert (exon 4a) between exons 4 and 5 demonstrated no homology to any LMW tau sequences, but showed a high homology to a similar region in HMW tau from N115 cells (Couchie et al., 1992) and is identical with this region in PC12 cells (Goedert et al., 1992). Sequences corresponding to exons 1, 2, 3, 4, 5, 7, 9, 10, 11, 12 and 13 of LMW tau are present. Exon 6, which is present in N115 HMW tau was not found in our cDNA, nor was it detectable by direct PCR analysis of DRG mRNA. The 2080 bp sequence has an open reading frame that codes for a protein of 686 amino acids with a calculated molecular mass of 71,783.

In vitro transcription/translation of pDRGTAU in a rabbit reticulocyte lysate system gave a protein migrating as a single band at 110 kDa on an SDS-polyacrylamide gel, in the same range as HMW tau protein from adult rat sciatic nerve (Fig. 3). This result confirms that pDRGTAU encodes the entire DRG HMW tau molecule and that no frame shift artifacts were introduced during the PCR cloning.

Neonatal DRGs have been reported to contain both LMW and HMW tau, as well as a prominent 8 kb tau mRNA (Oblinger et al., 1991). The presence of an 8 kb tau mRNA, as opposed to the 6 kb mRNA in brain, has been correlated with the presence of higher molecular weight polypeptides immunoreactive with tau antibodies (Drubin et al., 1988; Georgieff et al., 1991; Oblinger et al., 1991). Using cDNA probes specific to exon 4a, the inserted region in HMW tau, and to a region common to all tau molecules, we have performed northern blots on RNA isolated from neonatal and adult rat DRGs. On the first postnatal day (P0), tau mRNAs of approximately 6 kb and 8 kb were detected in DRGs using the common probe, while only the 8 kb mRNA was detected with pEX4a (Fig. 4). At P0, the 6 kb mRNA is more abundant than the 8 kb mRNA (Fig. 4A, lane 2). At P4, P7 and in the adult, a single 8 kb mRNA was found in DRGs that hybridized with both the common probe and pEX4a (Fig. 4A and B, lanes 2 and 4). No 8 kb mRNA was detected in any part of the CNS by northern blots (Fig. 4B, lanes 1 and 3).

Examination of total RNA from adult rat cerebral hemispheres, hindbrain, cerebellum, and dorsal root ganglia by northern blot analysis using pEX4a showed hybridization only to an 8 kb transcript in dorsal root ganglia (Fig. 5B). Reprobing with a cDNA insert common to both HMW and LMW tau revealed hybridization of the probe to a 6 kb mRNA species in all adult central nervous system tissues (Fig. 5A).

To increase the sensitivity of detection, PCR analysis was performed on mRNA from cerebral hemispheres, retina and DRGs using primers ex1 and ex7/9. To identify these PCR generated products with certainty, Southern blot analysis with pEX4a and the common tau probe were carried out. In DRGs the common tau probe hybridized to several fragments ranging in size from 394 bp to 650 bp, as well as a 900 bp and a 1200 bp fragment (Fig. 6A). Surprisingly pEX4a hybridized not only to the expected 1200 bp fragment, but to the 900 bp product (Fig. 6B, lane 2). In retina, PCR products hybridizing with pEX4a were slightly smaller than in the DRGs, with species at 800 bp and 1050 bp hybridizing (Fig. 6B, lane 3), while the common tau probe detected all products (Fig. 6A, lane 3). None of the brain-derived PCR products hybridized with pEX4a (Fig. 6B, lane 1), but several of the bands between 394 bp and 650 bp hybridized to the common tau probe (Fig. 6A, lane 1).

A more detailed study of CNS tissue was performed using the same PCR approach with mRNA from adult cerebellum, brainstem and spinal cord. Cerebellum and brainstem analysis gave identical results as with cerebral hemispheres (Fig. 7A and B). The surprising finding was in spinal cord, where pEX4a hybridized to two PCR products, at 900 bp and 1200 bp, the same size as in the DRGs (Fig. 7B). This strongly suggests the presence of a HMW tau message in spinal cord, previously undetected on northern blots. As expected, the common probe hybridized to all PCR products including the fragments between 394 bp and 650 bp present in brain, cerebellum, brainstem and DRGs (Fig. 7A).

In situ hybridization on adult dorsal root ganglia sections with a radiolabeled antisense RNA probe specific to exon 4a produced a hybridization signal in ganglionic neurons but not in non-neuronal cells (Fig. 8A). A probe in the sense orientation did not show specific hybridization to any DRG structures (Fig. 8B).

Next we used an antibody raised against exon 4a to localize HMW tau in dorsal root ganglia, spinal cord and brain. Sections of DRGs showed cytoplasmic staining of most neurons as well as the dorsal roots (Fig. 8C). In the spinal cord, immunoreactivity was primarily in the gray matter, with relatively weak staining in the white matter (Fig. 9A). The most pronounced staining was found in large motor neurons in the ventral horn (Fig. 9C). There was also staining in intermediate sized motor neurons and in secondary sensory neurons throughout the gray matter, including sensory neurons in the dorsal horns (Fig. 9A and B). There was also low level staining in the neuropil of the gray matter which was stronger in the dorsal horns, especially the dorsolateral fasciculus of Lissauer where the peripheral sensory fibers enter the spinal cord (Fig. 9A). Much more intense staining was seen in the gray matter of the spinal cord when an antibody against all tau forms was used (Fig. 9D). Sections of adult rat brain showed no detectable immunostaining with the antibody against HMW tau, but showed the characteristic tau staining pattern with an antitau antibody (Fig. 8D).

On western blots anti-4a antibody recognized only the HMW tau band in extracts of adult rat spinal cord and adult trigeminal nerve (Fig. 10A), in addition to the bacterial fusion product (Fig. 10B). Although, some LMW tau protein is seen in the trigeminal nerve, the predominant form of tau in this tissue appears to be HMW tau.

In common with other neuronal cells of neural crest origin in both rat and mouse, the dorsal root ganglia of the rat express a high molecular weight tau isoform (Drubin et al., 1988; Georgieff et al., 1991; Oblinger et al., 1991). This is the sole isoform of tau present in adult DRGs, and is present together with LMW tau in the neonatal DRGs. We have cloned the cDNA encoding the HMW tau molecule from DRG using PCR and have demonstrated that its greater size compared to LMW tau cDNAs is due to the insertion of a novel coding region between exons 4 and 5 of LMW tau (exon 4a). This inserted region is identical to the insert in HMW tau cDNA from rat PC12 cells (Goedert et al.,1992), demonstrating that there is no alteration in this protein between a tumor cell line and normal peripheral ganglia. Rat HMW tau cDNA is larger by 51 bp than mouse HMW tau cDNA, which we described previously (Couchie et al., 1992). It also differs by the absence of exon 6. The inserted region is rich in acidic amino acids and would be predicted to lead to an even more elongated tau molecule (Hagestedt et al., 1989). While the calculated molecular weight of the protein encoded by this cDNA is 71,783 Da, in vitro transcription/translation yields a protein of Mapp ≈110 kDa, which migrates in the same range as HMW tau from sciatic nerve on SDS-PAGE. This retarded migration is in accord with data on LMW tau, which also migrates with an apparent molecular weight 1.3× to 1.5× its calculated molecular weight (Himmler et al., 1989). This anomalous migration could be due to phosphorylation or other post-translational modifications as well as to the relatively asymmetric conformation of tau (Lichtenberg et al., 1988).

High molecular weight tau protein is most abundant in the peripheral nervous system. It is found in decreasing quantities in the spinal cord, brainstem and cerebral hemispheres and is also found in the optic nerve (Georgieff et al., 1991; Taleghany and Oblinger, 1992). Based on this gradient of protein concentration and the fact that HMW tau is far richer in the dorsal columns of the spinal cord than in other regions, we have previously proposed that HMW tau is synthesized predominantly in the peripheral nervous system and is carried centrally by the axons of peripheral neurons (Georgieff et al., 1991). This idea is supported by experiments that show that the 8 kb mRNA in the PNS contains the 4a insert and that neither an 8 kb mRNA nor any other mRNA containing exon 4a is detectable in the cerebral hemispheres, cerebellum or brainstem by northern blot or by PCR analyses. Therefore, it is clear that any HMW tau in these structures must be produced at another site and carried in by the axons projecting to them.

However, new data presented here indicate that HMW tau is not as strictly limited to the PNS as we suggested previously. Using a highly sensitive combination of PCR and Southern blot analyses we have been able to demonstrate that PCR products derived from spinal cord and retina hybridize to pEX4a, suggesting the presence of mRNAs in these CNS regions that contain exon 4a. In spinal cord, as well as DRGs we obtain two products of 1200 bp and 900 bp. While the larger fragment is identical to that used in our cloning studies, the finding of an additional smaller fragment in both spinal cord and in DRGs was unexpected. Assuming that there are no alterations in the 3′ portion of the mRNA, the smaller mRNA would be expected to produce a protein with predicted migration on SDS-polyacrylamide gels of 85 kDa to 95 kDa. A more rapidly migrating minor band is often seen in DRGs and spinal cord, which could be this product, but we cannot rule out the possibility that it is a degradation product of HMW tau (Georgieff et al., 1991). The data from retina show even greater complexity, with pEX4a hybridizing to fragments at 800 bp and 1050 bp. The larger of these would encode a protein approximately 1600 Da smaller than DRG HMW tau and would be predicted to migrate proportionally faster during SDS-PAGE. This is in agreement with the behavior of HMW tau from optic nerve and suggests that this protein is produced in the retinal ganglion cells rather than in peripheral neurons (Taleghany and Oblinger, 1992). Although HMW tau mRNA appears as a single 8 kb band on northern blots, our PCR studies demonstrate that there is heterogeneity in this mRNA.

Our developmental studies strongly suggest that the 6 kb mRNA encodes only LMW tau protein. During the immediate postnatal period, DRGs have both 6 kb and 8 kb tau mRNAs and express both LMW and HMW tau. By 4 days after birth, the 6 kb mRNA, which lacks exon 4a, disappears, although the LMW tau protein that it encodes persists up to postnatal day 7 (Oblinger et al., 1991). In the adult DRG there is only the exon 4a-containing 8 kb tau mRNA and its protein product, HMW tau. In situ hybridization with an antisense probe to exon 4a and immunostaining with an anti-4a antibody show restricted localization in the adult DRG. All neurons are positive and the immunostaining is seen in the nerve fibers as well as cell bodies. No staining of non-neuronal cells is seen.

Although we can detect HMW tau mRNA and protein in the CNS, it does not appear that the expression is nearly as abundant as in the PNS. The necessity of using a combination of PCR and Southern blot to detect the mRNA suggests either that a restricted group of neurons express HMW tau, or that HMW tau is expressed by many neurons at very low levels. Our immunohistochemical data demonstrate restriction to certain neuronal groups at low levels of expression.

In summary, HMW tau from normal rat peripheral neurons is identical to that in rat PC12 tumor cells but differs from HMW tau from mouse N115 cells by a small difference in the sequence of the inserted exon 4a region, and by the absence of exon 6. HMW tau mRNA can be detected specifically by DNA probes complementary to the inserted exon 4a region and HMW tau protein by antibodies raised against the exon 4a region of the protein. HMW tau is made predominantly in peripheral neurons but this protein is also found in limited areas of the CNS. A related protein and mRNA are present in retinal ganglion cells and the optic nerve. However, even by our most sensitive PCR assay, HMW tau mRNA cannot be detected in cerebral hemispheres, cerebellum or brainstem.

LMW tau expressed in fibroblasts confers an intermicrotubule spacing of 15 nm (Hirokawa et al., 1988). One possible role of HMW tau would be the maintenance of a wider spacing between microtubules in peripheral axons. A review of existing data demonstrates that, on average, spacing between microtubules in axons of the CNS is 1 to 1.5 microtubule diameters, in contrast with PNS axons where the distance increases to 2 to 2.5 microtubule diameters (Friede and Samorajski, 1970; Peters et al., 1976; Bray and Bunge, 1981). Although these data suggest that the microtubule spacing distance between the peripheral and central nervous systems varies, the challenge will be to differentiate the spacing related to tau from that due to other factors.

We thank Kristy Brown and Mike Kaplan for extensive technical assistance in performing the in situ hybridization and immunohistochemistry. This work was supported by grants from the NINDS and the Alzheimer’s Disease Research center. Dr Irene Georgieff is supported by a postdoctoral fellowship from the National Institute on Aging.