|Home | About | Journals | Submit | Contact Us | Français|
Alzheimer’s disease and other tauopathies are characterized by the intracellular accumulation of insoluble filaments of the microtubule-associated protein tau. The six canonical tau isoforms in the adult brain consist of an N-terminal “projection” domain followed by a proline-rich region, a microtubule-binding repeat region, and a C-terminal tail. However, alternative splicing in exon 6 produces an additional set of tau isoforms, termed 6D and 6P, which contain only the N-terminus and part of the proline-rich region. We have previously shown that constructs representing N-terminal fragments of tau, which resemble the naturally occurring 6P and 6D isoforms, inhibit polymerization of the full-length protein in an in vitro filament formation assay, and traced the inhibitory activity to amino acids 18-42. Here we report that 6P and 6D tau isoforms inhibit polymerization of full-length tau (hTau40) in a similar manner, likely by stabilizing full-length tau in a soluble conformation. The absence of exons 2 and 3 decreased the effectiveness of the 6D isoforms, but not the 6P variants or the N-terminal tau fragments from our previous study, indicating that the 18-42 region is not the sole determinant of inhibitory ability. Finally, this paper demonstrates that inhibition is blocked by pseudophosphorylation of tyrosines 18 and 29, providing a potential link between tyrosine phosphorylation and disease progression. Taken together, these results indicate that the 6P/6D isoforms are potential endogenous inhibitors of tau filament formation, and suggest a mechanism by which this ability may be disrupted in disease.
The microtubule-associated protein tau forms intracellular, filamentous aggregates in Alzheimer’s disease (AD) and other tauopathies. The formation of tau pathology is thought to be intimately linked to neurodegeneration, in part because the appearance of tau pathology in AD follows a spatial and temporal progression through the anatomical regions that underlie the clinical symptoms. Filamentous tau deposition is first found in areas associated with learning and memory, and later spreads through much of the cerebral cortex (reviewed in (1)). Yet even in severe AD some areas, including the cerebellum and primary sensory-motor areas, remain relatively free of tau pathology (2-4). Identifying the differences between susceptible and protected neuronal populations, including endogenous triggers and inhibitors of tau aggregation, may be crucial to understanding selective cellular vulnerability in AD.
Although tau was historically considered to be an unstructured protein in solution, recent evidence indicates that soluble tau occupies a number of folded states, and that these conformations influence tau’s propensity to form filaments. The amino terminus of tau has emerged as an important regulator of tau folding. The association of the extreme amino terminus (5-15) with the microtubule binding repeat region (MTBR), recognized by the conformational antibody Alz50, is thought to underlie the transition from soluble to filamentous tau (reviewed in (5)). The amino terminus is also an important negative regulator of tau aggregation. When present in molar excess, constructs containing only the N-terminus inhibit polymerization of the full-length protein in an in vitro filament formation assay. This inhibitory effect requires amino acids 18-42 in the N-terminal fragments and residues 392-421 in the C-terminus of full-length tau (5). The N-terminal fragments may act by stabilizing a “paperclip” conformation of tau (5), in which the C-terminus associates with the MTBR region and the N-terminus comes into close proximity to the C-terminus. This conformation is observed when tau is a soluble monomer, and may discourage polymerization (6-8).
The majority of tau studies to date focus on six canonical isoforms in the adult CNS produced by alternative splicing in the amino terminus (exons 2 and 3) and the MTBR region (exon 10) of the protein (reviewed in (9)). This picture grew more complicated with the discovery of additional tau isoforms that contain all or part of the sequence encoded by exon 6, which is absent from the canonical isoforms. Inclusion of the entire exon (6+) extends the proline-rich region of tau, and results in isoforms that are otherwise identical to their canonical counterparts. The 6+ isoforms may play a regulatory role in neurite elongation (10). In addition to the 3′ splice site at the end of exon 6, there are two alternate 3′ splice sites within exon 6 itself, termed “6P” and “6D” according to location (Fig. 1). Use of either of these splice sites results in a frame shift that introduces a unique amino acid sequence, PCCVPRATFLS (6P) or FWSKGDETQGG (6D), followed by a stop codon. As a result, tau 6P and 6D isoforms are truncated, lacking part of the proline-rich region, the MTBR region, and C-terminus (11, 12). Of these, the MTBR region is absolutely essential to tau-microtubule binding (13-15) and tau aggregation (16, 17).
The expression of 6P/6D tau isoforms is spatially and developmentally regulated. At the mRNA level, 6P levels are similar in fetal and adult brain, while 6D is more prominent in fetal brain (18). In adult brain, 6P and 6D mRNA levels are highest in cerebellum and spinal cord, but are detectable in all CNS tissues examined, including the cerebral cortex and the hippocampus (11). Using an antibody to the C-terminus of 6D, Luo et al. compared the expression of 6D to and full-length isoforms in various human tissues. 6D levels were comparable to full-length isoforms in cerebellum, and levels were reduced but detectable in the cerebral cortex, hippocampus. It is notable that 6D protein expression is particularly high in the cerebellum, which is not affected by tau lesions in AD, and lowest in tangle-prone areas (hippocampus, cerebral cortex). Even within AD hippocampus, anti-6D labeling does not colocalize with an antibody that recognizes neurofibrillary tangles (Tau-5; (19)).
In light of the polymerization suppressive properties of N-terminal tau fragments and the expression pattern in human brain, we asked whether 6P/6D isoforms might act as endogenous inhibitors of tau aggregation. To begin to address this question, we assessed the effects of 6P and 6D isoforms on the polymerization of full-length tau in an in vitro filament formation assay (20, 21). We report that 6P and 6D isoforms inhibit polymerization of full-length tau, and that the efficacy of each isoform depends on whether it is of the 6D or 6P variety, and on the presence or absence of N-terminal exons 2 and 3. We also demonstrate that inhibition is abolished by mutations in the crucial 18-42 region that mimic phosphorylation. Collectively, our results indicate that the 6P/6D isoforms have the potential to act as endogenous regulators of tau filament assembly, and suggest a basis for the disruption of this regulatory ability by disease-related phosphorylation events.
The six canonical tau isoforms contain zero, one, or two alternatively spliced N-terminal inserts (designated 0N, 1N, and 2N, respectively), and either three or four MTBRs (3R or 4R). The full-length tau used in this study (hTau40) is the longest canonical isoform in the human central nervous system, containing 441 amino acids, including both alternatively spliced N-terminal exons (e2 and e3) and four microtubule binding repeats (m1-m4; Fig. 2A). The various 6D and 6P isoforms were generated by restriction digestion and ligation of cDNA constructs described previously (19) and hTau40 (2N4R) (22), hTau23 (0N3R) and hTau37 (1N3R) (23). 6DY18/29F and 6DY18/29E were created by site-directed mutagenesis (Stratagene) on the 2N6D background. The constructs 1-196 has been described previously (5). All constructs were verified by sequencing prior to protein purification. Proteins were expressed in E. coli and purified by means of an N-terminal poly-histidine tag (16, 22). Protein concentrations were determined by the Lowry assay (24).
Arachidonic acid (AA) was obtained from Cayman Chemical (Ann Arbor, MI) and stored at −20°C. Working solutions were prepared in 100% ethanol immediately prior to use. Tau polymerization was induced by arachidonic acid as previously described (21). Briefly, tau protein (4 μM) was incubated at room temperature in reaction buffer (10 mM HEPES, pH 7.6, 100 mM NaCl, 0.1 mM EGTA, 5 mM DTT) in the presence of 75 μM AA The final volume of ethanol in these reactions was 3.8%, and this volume was added to control reactions in the absence of AA. Unless otherwise noted, 6P/6D constructs and 1-196 were added at a concentration of 8 μM to the polymerization reaction mixture prior to the addition of arachidonic acid. Polymerization was monitored at room temperature by the intensity of right angle laser-light scattering (is; (21)). End-point (t = 300 min) laser light scatter data from at least three independent experiments were analyzed. Statistical significance was determined by comparing polymerization in the presence and absence of N-terminal constructs by a student’s two tailed t-test. Time course data were fit with curves using GraphPad Prism 3.0 software. Error bars in all figures represent plus and minus one standard error of the mean.
Polymerization reactions were allowed to proceed at least five hours, fixed with 2% glutaraldehyde (Electron Microscopy Sciences, EMS, Hatfield, PA), spotted onto 300 mesh formvar/carbon coated copper grids (EMS), and negatively stained with 2% uranyl acetate (EMS) as previously described (20). Grids were examined using a JEOL JEM-1220 electron microscope at 60kV and 12,000x magnification, and photographed using a MegaScan 794/20 digital camera and DigitalMicrograph software version 3.9.3 (Gatan). Optimas 6.0 imaging software (Media Cybernetics) was used to automatically identify and measure filaments (defined as objects > 20 nm in length). At least five separate fields from each grid were randomly chosen under low illumination to prevent bias. In experiments where different constructs were analyzed for effects on hTau40, data from each experiment were normalized to control reactions containing hTau40 alone and expressed as percent of the hTau40 value. Results from at least three independent experiments were analyzed by student’s two-tailed t tests to determine if polymerization was significantly different from controls (GraphPad Prism 3.0 software).
Polymerization reaction mixtures were incubated for 5 h in the presence of arachidonic acid. Following assembly, a pretreatment sample was removed and the remainder of the reaction mixture was centrifuged at 100,000 x g for 20 minutes at 25° C over a 40% glycerol cushion. Samples of the starting material and supernatants were diluted in 2X Laemmli buffer and boiled. Pellets were resuspended in an amount of polymerization buffer equal to the starting volume, prior to addition of 2X Laemmli buffer. Proteins were separated by SDS-PAGE, transferred to nitrocellulose membranes, and probed with the amino-terminal antibody Tau-12 (4 ng/mL) (25) to visualize all tau constructs, and the carboxy-terminal antibody Tau-7 (40 ng/mL) (5), which recognizes full-length hTau40 but not the 6D/6P isoforms. HRP-conjugated goat anti-mouse secondary antibody (Vector Laboratories, Burlingame, CA) and ECL (GE Healthcare, Amersham, UK) were used to detect proteins.
To examine the effects of the 6D and 6P isoforms on the polymerization of full-length tau (hTau40), several protein constructs were created and purified (Fig. 2A-B). The 6P and 6D constructs, like the previously described N-terminal tau construct 1-196, lack the MTBR region of tau known to be necessary for polymerization (16, 17). Alternative splicing of exons 2 and 3 produces canonical tau isoforms containing zero, one, or two N-terminal inserts (designated 0N, 1N, and 2N, respectively). Hence, constructs containing zero, one or two alternately spliced N-terminal exons were created on the background of the 6P and 6D isoforms. An additional tau protein construct, 1-196, that was previously shown to inhibit hTau40 polymerization (5) was included as a positive control. All constructs contain the specific amino acid region (18-42) previously identified as crucial for the inhibition of hTau40 polymerization (5).
To determine whether 6P and 6D isoforms influence tau polymerization, hTau40 (4 μM) was incubated in the presence or absence of a twice-molar excess of 2N6D, 2N6P, or 1-196. Polymerization was induced by the addition of arachidonic acid, and right angle laser-light scattering (LLS) was used to monitor filament formation (Fig. 3A). After five hours of polymerization, a similar degree of inhibition was apparent in the presence of 2N6P (28.63 ± 6.27%, P < 0.05) and 1-196 (33.40 ± 6.59%, P < 0.05). 2N6D also significantly inhibited hTau40 polymerization, albeit to a lesser extent (20.20 ± 2.85%, P < 0.05).
To verify the results of the LLS experiments and to further characterize the effects of 2N6P and 2N6D on hTau40 polymerization, we performed quantitative electron microscopy (EM) on filaments formed under each experimental condition (Fig. 3B). Previous work from our lab demonstrated that incubation with N-terminal tau fragments reduced the overall mass of hTau40 filaments formed, with the primary effect being a reduction in the number of filaments per field (5). Quantitative analysis revealed similar effects in the present study (Fig. 4). Polymerization of hTau40 (4 μM) in the presence of 2N6P (8 μM) caused a significant reduction in the number of filaments per field (36.26 ± 2.36%; P < 0.001), and in the overall mass of polymerized material per field (42.80 ± 4.64%; P < 0.001). Incubation with 2N6D also reduced filament number (25.20 ± 4.62%; p < 0.001), and overall polymer mass (30.12 ± 3.23%; P < 0.001). In agreement with our previous report (5), incubation with 1-196 reduced both the number and overall mass of filaments (45.07 ± 3.86% and 51.63 ± 3.27%, respectively; P < 0.001). These results are consistent with the reduction in polymer mass as observed by LLS. Taken together, the results of the LLS and EM experiments indicate that, like the N-terminal fragments previously examined, 2N6P and 2N6D inhibit polymerization of full-length tau.
Alternative splicing of exons 2 and 3 on the 6D/6P background potentially gives rise to three isoforms apiece (see Fig. 2A). To determine whether alternatively spliced 6P and 6D differentially affect hTau40 polymerization, we assayed inhibition by each isoform using quantitative EM (Fig. 4). Filament number was significantly reduced by 0N6P and 1N6P in a manner similar to 2N6P (42.72 ± 2.94% and 36.08 ± 2.45%, respectively; p < 0.01), as was filament mass (43.95 ± 3.22% and 40.38 ± 4.30%, respectively; p < 0.01, p < 0.05). This is in agreement with previous results demonstrating that exons 2 and 3 had no effect on the inhibitory abilities of N-terminal fragments (5). In contrast, while 2N6D significantly reduced filament number and mass, 6D constructs containing zero or one N-terminal exons (0N6D and 1N6D) failed to significantly inhibit hTau40 polymerization. These experiments indicate that although alternative splicing has no effect on the ability of 6P isoforms to inhibit hTau40 polymerization, the absence of exons 2 and/or 3 reduces the effectiveness of 6D isoforms.
In a previous study, we demonstrated that N-terminal tau fragments do not associate with hTau40 filaments in a co-sedimentation assay (5). This result suggested that the fragments act in the soluble fraction of the polymerization mixture to produce their effects. To determine whether 6D and 6P isoforms inhibit polymerization through a similar mechanism, we assembled hTau40 (4 μM) in the presence or absence of 2N6P, 2N6D, or 1-196 (8 μM) and then separated the mixtures into soluble and filamentous fractions through ultracentrifugation. The resulting supernatants and pellets were processed by SDS-PAGE gel electrophoresis, transferred to nitrocellulose membranes, and probed with antibodies against tau (Fig. 5). All three constructs, 2N6P, 2N6D and Tau196, remained in the supernatant fraction and were absent from the filamentous pellet, even at longer exposures (data not shown). These results indicate that 2N6P and 2N6D act in the soluble fraction to inhibit hTau40 polymerization, as did the N-terminal tau fragments previously studied.
We have previously demonstrated that the inhibitory effect of N-terminal tau fragments requires amino acids 18-42 in the amino terminus (5). Since 6D and 6P isoforms have similar effects on full-length tau polymerization (i.e. reduce filament number, remain in the soluble fraction), it is likely that amino acids 18-42 are crucial for their effect as well. This sequence contains tyrosine residues at positions 18 and 29, which are potential targets for kinase activity in AD (reviewed in (26)). Therefore, we asked whether phosphorylation at these tyrosine residues affects the ability of 6D and 6P isoforms to inhibit polymerization of the full-length protein.
In order to overcome technical problems associated with in vitro phosphorylation reactions (i.e. inefficient phosphorylation, phosphorylation at unintended secondary sites), we used pseudophosphorylation, a well-established strategy that employs site-directed mutagenesis to replace a phosphorylatable residue with glutamic acid (27). We modeled the effects of tyrosine phosphorylation on the ability of 6D and 6P isoforms to inhibit tau polymerization by generating a construct based on 2N6D in which both tyrosine residues were mutated to glutamic acid (Y18/29E). As a control, we generated a construct in which both residues were mutated to phenylalanine (Y18/29F) (see Table 1, Fig 6A).
We polymerized hTau40 (4 μM) in the presence or absence of these constructs (8 μM), and measured the effects on filament number, average length, and mass by quantitative EM. Pseudophosphorylation at both tyrosine residues (Y18/29E) blocked the ability of 2N6D to inhibit hTau40 (Figure 6B-D). In contrast, the effects of the control construct containing mutation to phenanlynine (Y18/29F) were not significantly different than wild-type 2N6D. These results indicate that tyrosine phosphorylation may block the inhibitory effects of the N-terminal isoforms, and that the residues18-24 identified in our earlier work using N-terminal hTau40 fragments (5), is also important for the inhibitory effects of the 6D and 6P isoforms.
Alternative splicing of tau exon 6 produces isoforms that contain the amino terminus of the canonical protein, but lack the proline rich region, MTBR region and the C-terminal tail. The 6D/6P isoforms bear a striking resemblance to N-terminal tau fragments previously shown to inhibit in vitro polymerization of full-length hTau40 (5). Based in part on this physical resemblance, we assayed 6D and 6P isoforms for effects on hTau40 filament formation. The results presented here demonstrate that 6P/6D isoforms inhibit in vitro polymerization of full-length tau. Like the N-terminal fragments previously studied (5), the 6D/6P isoforms reduce the number of filaments formed and the overall filament mass, and remain in the soluble fraction of the polymerization mixture, consistent with a model in which these isoforms stabilize a conformation of full-length tau that discourages polymerization (Fig. 7).
Surprisingly, although all 6P/6D isoforms contain the region necessary for inhibition (a.a. 18-42), we observe differences in their effectiveness. In general, 6P isoforms are more potent inhibitors than 6D isoforms. Additionally, the absence of exons 2 and 3 decreased the effectiveness of the 6D isoforms, but not the 6P variants or the N-terminal tau fragments previously tested (5). A simple interpretation is that the C-terminal 6D sequence encourages a protein folding event that limits accessibility of the 18-42 sequence, and that this effect is counteracted by the presence of exons 2 and 3. Future insights into the structure of these short isoforms, as well as their interaction with full-length tau, should distinguish between this and other possible explanations.
Our results demonstrate that 6P/6D isoforms are potential endogenous inhibitors of filament formation, but the question of whether they do so in the brain is still unanswered. A twice- molar excess of the 6D/6P isoforms was necessary to inhibit in vitro polymerization, but with the exception of the cerebellum, 6D/6P levels are low relative to canonical tau isoforms in adult brain (11, 18, 19, 28). In this in vitro assay system, however, the 6D/6P isoforms (8 μM) must compete with the arachidonic acid (75 μM) that drives polymerization, presumably by inducing hTau40 to adopt an Alz50-like conformation. Lower levels of the 6D/6P isoforms may be required in the context of a cellular system. It is also possible that enrichment of 6D/6P isoforms in certain parts of the cell results in increased local levels relative to canonical tau. Supporting this idea, anti-6D immunoreactivity in cultured cells and in brain is punctuate, and enriched in the proximal section and distal tips of neuronal processes (19). A candidate site for 6D/6P enrichment is the plasma membrane, which interacts with the amino terminus of tau (29). Because these truncated isoforms lack the MTBR region, they may even be more likely to interact with the plasma membrane than full-length isoforms. Future studies in cellular or animal models are necessary to determine whether 6D/6P isoforms are endogenous inhibitors of tau aggregation.
Future studies of 6P/6D expression levels and patterns in human brain may reveal more about the potential of these isoforms as endogenous regulators of tau polymerization. Although we propose that expression of 6D/6P may protect the cerebellum in AD, this region develops fibrillar pathology in other tauopathies (30, 31). This indicates either that 6D/6P is not protective against tau aggregation, or that this ability is compromised in these diseases. It is not known at present whether 6D/6P expression or phosphorylation (discussed below) is disrupted in these tauopathies, although brain-specific changes in the expression of these isoforms were detected in myotonic dystrophy type 1, a disease marked by tau aggregation (28). Comparative neuropathological studies of 6D/6P expression in various tauopathies may inform this issue.
In the present study, we demonstrate that pseudophosphorylation of tyrosines 18 and 29 blocks the inhibitory ability of the 6D/6P isoforms. Pseudophosphorylation may disrupt the interaction of 18-42 with full-length tau directly or through conformational effects that reduce the availability of this region. Phosphorylation at tyrosine 18 is elevated in AD (32, 33), and may be targeted by Fyn or other members of the src family of tyrosine kinases (reviewed in (26)). Affinity of src tyrosine kinases for the 6D/6P isoforms might be reduced though, because they are truncated prior to an SH3-binding motif (233PKSP236) that binds src family tyrosine kinases (34, 35). In addition to phosphorylation, other factors with the potential to affect the inhibitory abilities of the 6D/6P isoforms include nitration at tyrosines 18 and 29 (36, 37), the mutations R5L (38) and R5H (39), which cause frontotemporal dementia and Parkinsonism linked to chromosome 17 (FTDP-17), and removal of the amino terminus, which occurs early in AD progression and may be mediated by caspase 6 (40) or puromycin-sensitive aminopeptidase (41). If the 6D/6P isoforms represent endogenous inhibitors of filament formation, these changes in the amino terminus could contribute to disease progression by removing a barrier to tau aggregation.
Recent studies indicate a role for the N-terminus of tau, and therefore the 6P/6D isoforms, in intracellular signaling. The N-terminus of tau may be part of the cellular response to beta-amyloid (Aβ) (42-45). Expression of N-terminal fragments of tau in cell culture can be neuroprotective (46) or toxic, depending on the length, and can induce cell death through a mechanism involving N-methyl-D-aspartate (NMDA) receptors and calpain (46, 47). In isolated axoplasm, 6P/6D isoforms selectively inhibit kinesin-based axonal transport by activating a signaling cascade that results in phosphorylation of kinesin light chains (48). Finally, the unique C-terminal 6D and 6P sequences may confer isoform-specific cellular roles to these proteins. Further work is necessary to reveal the role of these truncated tau isoforms in human brain.
†Supported by NIH Awards NS049834 (N.L.), AG09466 (L.I.B).