INTRODUCTION

The neuronal microtubule-associated protein tau consists of a family of closely related phosphoproteins that are produced from a single gene by alternative splicing and posttranslational modification (for reviews, see Refs. 1, 2). Tau isolated from brain is phosphorylated at several sites and is a substrate for many kinases in vitro (for review, see Ref. 3). Although it was initially shown that dephosphorylation of tau isolated from brain increased its ability to promote microtubule assembly (4), it has since become apparent that the extent to which the activities of tau are modulated depends critically on the identity of the site phosphorylated (5-8). Phosphorylation of tau may also change its conformation, resulting in an increased "stiffness" and a decreased electrophoretic mobility (9-11).

Phosphorylation of tau by PKA¹ may be a useful model for studying phosphorylation-induced changes in tau structure and function because PKA phosphorylates sites in the microtubule-binding domain and the two flanking regions (11). Phosphorylation at these sites induces shifts in electrophoretic mobility indicative of conformational changes and decreases microtubule binding and assembly (11-13). Phosphorylation of tau by PKA could also be of functional importance for neuronal development because PKA is present in high levels in the brain and within neurons (14).

In addition to its potential role during development, phosphorylation of tau has also been implicated in neurodegenerative disorders. Abnormal tau phosphorylation is thought to contribute to its aggregation into paired helical filaments (PHFs), which are a major constituent of the neurofibrillary tangles in the brain of patients with Alzheimer's disease (15, 16). Tau protein isolated from Alzheimer's disease patients shows a decreased electrophoretic mobility on SDS gels (17-19) and is less active in promoting microtubule assembly than tau isolated from control brains (20). Interestingly, all of the major phosphorylation sites characteristic of PHF-tau are clustered in two regions that flank the microtubule-binding domain of tau (21), suggesting that phosphorylation events in these domains are involved in the changes in the structure and function during neurodegeneration.

Previously, we have provided evidence that phosphorylation of two of the five residues that are phosphorylated by PKA in vitro are critical for the conformation of tau and its microtubule assembly activities (13). These two sites are located in two regions that amino-terminally (serine 156) and carboxyl-terminally (serine 327) flank the microtubule-binding domain of tau (numbering refers to the fetal specific human tau isoform containing 352 residues (22)). However, because kinase reactions yield a mixture of differentially phosphorylated tau species, the function of individual phosphorylation sites is difficult to assess. In other systems, it has been shown that the effect of phosphorylation can be imitated by introducing negatively charged residues (23, 24). In this study, we have used this method to analyze the effect of individual phosphorylation sites on the structure and function of tau. A set of mutated tau isoforms was prepared where individual residues have been changed to either alanine or aspartate. Structural consequences of the mutations were analyzed by SDS-polyacrylamide gel electrophoresis (PAGE). Microtubule growth and nucleation assays were performed to assess the functional consequences of the mutations on the activity of tau. In addition, a cellular test system was used to assess the activity of the mutants to promote process formation in living neural cells.

EXPERIMENTAL PROCEDURES

All reagents, unless otherwise specified, were obtained from Sigma (Deisenhofen, Germany).

Prokaryotic expression plasmids were constructed in pET-3d as described previously using p19tau cDNA (25). Mutations at position 327 were constructed using M13 in vitro mutagenesis as described by Kunkel et al. (26). The mutation at position 156 was introduced using polymerase chain reaction (vent polymerase, New England Biolabs Inc., Beverly, MA), with one primer introducing a NcoI site on the amino-terminal end and the other converting the target site to either alanine or aspartate. The resulting DNA was cut with NcoI and SmaI. This piece was introduced in the NcoI and SmaI sites of pET-Tau(Ala 327) or pET-Tau(Asp 327). Proteins were expressed and purified as described previously (25).

For expression in eukaryotic cells, inserts from wild-type tau and tau carrying alanine or aspartate double mutations were prepared from the pET constructs by polymerase chain reaction using primers introducing a ClaI site at the amino-terminal side and an ApaI site at the carboxyl side. These inserts were introduced into the ClaI and ApaI sites of the fpRC/CMV vector, a modification of pRC/CMV (Invitrogen, San Diego, CA). The modification resulted in the expression of proteins with the sequence MDKDDDDK (FLAG (27, 28)) fused to the amino-terminal end as an epitope tag.

Constructs for prokaryotic and eukaryotic expression were verified by dideoxy sequencing using Sequenase (United States Biochemical, Cleveland, OH). Restriction enzymes and T4 DNA ligase were purchased from NEB, Inc. (Beverly, MA).

pET-3d tau plasmids were transformed into Escherichia coli BL21(DE3)pLysS cells for expression (29). Cells were grown, induced, and harvested as described previously (29). Tau was purified from the bacterial pellet as described previously (25) but concentrated using polyethylene glycol instead of microconcentrators. Protein concentrations were determined by densitometry of Coomassie Blue-stained gels using bovine serum albumine as a standard. Densitometry employed a LKB Ultroscan XL Laser Densitometer.

Tubulin was isolated from bovine brain by two assembly-disassembly cycles and phosphocellulose chromatography as described previously (25). Tubulin concentrations were determined by the method of Bradford (30) using bovine serum albumine as a standard.

Microtubule assembly assays were performed as described previously (25) with tau constructs and incubation times as specified. After fixation, 0.1% (v/v) or 0.5% (v/v) aliquots were collected by centrifugation onto polylysine-treated coverslips and prepared for anti-tubulin immunofluorescence as described previously (25). Fluorescence microscopy employed a Zeiss Axioskop Neofluar × 100 lens. The number and lengths of microtubules were determined as described previously (25).

Taxol-stabilized microtubules were prepared from purified tubulin that had been precleared by centrifugation for 30 min at 100,000 × g (Beckman 100.2 rotor). The precleared tubulin was brought to 50 µM and polymerized by stepwise addition of 1/100 volume of 10, 100, and 1 mM taxol with 5-min incubations at 37 °C after each addition. The incubation mixture contained in 50 µl of BRB80 (80 mM K-PIPES, 1 mM EGTA, 1 mM MgCl₂, pH 6.8) 15 µl of taxol-stabilized microtubules (final concentration, 15 µM), 1 mM GTP, 10 µM taxol, and 5 µg of tau constructs or the control (myoglobin). The mixture was incubated for 10 min at 37 °C, subsequently loaded onto 100 µl of room-temperature 30% (w/v) sucrose in BRB80 containing 1 mM GTP and 10 µM taxol, and spun for 1 h at 100,000 × g at 20 °C. For tau-containing samples, it was necessary to remove the tubulin, which would run at a similar molecular weight as tau. This was done by heat denaturation, in which the microtubule pellet was dissolved in 50 µl of boiling buffer (50 mM PIPES/KOH, pH 6.8, 1 mM EGTA, 0.2 mM MgCl₂, 0.5 M NaCl, 5 mM dithiothreitol), boiled for 10 min, and then centrifuged for 10 min at 4 °C at 35,000 × g (Beckman TLA45), thereby pelleting denatured tubulin. The supernatants were adjusted with SDS-sample buffer and separated side by side by SDS-PAGE on 10% polyacrylamide. Pellets from samples containing myoglobin were directly dissolved in SDS-sample buffer after cosedimentation with the microtubules and separated by SDS-PAGE on 20% polyacrylamide.

PC12 cells were grown, transfected using Lipofectin (Life Technologies, Inc.), and treated with cytochalasin B as described previously (31). The average transfection efficiency was 11%, with a range of 5-18%. Cells were fixed 2 days after transfection and stained with monoclonal antibody M5 against the epitope tag FLAG (Kodak, New Haven, CT) and rhodamine-labeled donkey anti-mouse secondary antibody (Jackson ImmunoResearch, West Grove, PA) as described previously (31). Cells were mounted, photographed, and assessed for process extension as described previously (31).

SDS-polyacrylamide gel electrophoresis and immunoblots were performed as described previously (25, 32).

RESULTS

Human fetal tau cDNAs were constructed in which serine residues at position 156 and 327 were mutated to an alanine or aspartate (Fig. 1A). The constructs were expressed in E. coli and purified as described under "Experimental Procedures." Fig. 1B shows that the proteins were more than 95% pure according to SDS-PAGE and had apparent molecular masses of between 45 and 48 kDa. Bands at molecular masses lower than the full-length proteins were due to limited proteolysis during expression, as confirmed by immunoreaction.

It has been previously shown that phosphorylation of tau may result in a conformational change that is reflected by a decreased electrophoretic mobility during SDS-PAGE (4, 33). Fig. 1B (left) shows that a serine to aspartate mutation at position 327 was sufficient to cause a decreased electrophoretic mobility of the protein. A similar shift in the electrophoretic mobility had been previously observed after phosphorylation with calmodulin kinase II, which phosphorylates serine 327 exclusively (10). This indicates that aspartate at position 327 is capable of mimicking a phosphorylation-induced change in the conformation of tau. An additional mutation at position 156 did not cause a further shift in the electrophoretic mobility of tau, although a region containing residues 154-172, with serine 156 being the only PKA site within this sequence, was previously found to be required for a phosphorylation-induced supershift in the mobility of tau (13).

To test whether the decreased electrophoretic mobility was due to a different charge or to a conformational change, SDS-PAGE was performed in the presence of 6 M urea. Under these conditions wild-type tau and mutants had the same electrophoretic mobility (Fig. 1B, middle). This indicates that the mobility shifts were due to conformational changes that produced unfolded and SDS-resistant domains in tau protein (34).

All constructs were reactive to the phosphorylation-sensitive monoclonal antibody Tau-1, indicating that a mutated position 156 is not sufficient to abolish Tau-1 immunoreactivity (Fig. 1B, right).

It has been reported previously that five sites in recombinant tau are phosphorylated in vitro by PKA with a total phosphate to tau stoichiometry of 1.5-2.6 (11, 13). This indicates a substoichiometric phosphorylation of several sites. To determine the extent of phosphorylation at the mutated residues, wild-type tau and tau(Ala, Ala) were phosphorylated by PKA in the presence of [-³²P]ATP as described previously (13). Conversion of serines 156 and 327 to alanine reduced the stoichiometry of phosphorylation by 40% (±6%; n = 3), indicating a major contribution of these sites to PKA-dependent tau phosphorylation. This decrease was significant (p < 0.05).

To analyze whether serine 156 and serine 327 were required and sufficient for PKA-induced changes in the conformation of tau, wild-type and mutant tau were subjected to in vitro phosphorylation reactions. Because we had previously shown that substrate modulation affected the phosphorylation state of tau and its conformation (13), assays were performed in the presence or absence of heparin (Fig. 2). In the presence of heparin, PKA induced a dramatic shift ("supershift") in the electrophoretic mobility of wild-type tau from a species with an apparent molecular mass of 45 kDa to a species of about 52 kDa. In the absence of heparin, proteins with no shift and an intermediate shift (48 kDa) were observed. Mutation of serine 327 to aspartate induced a shift similar to the intermediate shift. No intermediate shift was observed with the alanine 327 mutant. In the absence of heparin, 38-43% of each construct was phosphorylated to yield the 52-kDa species. After phosphorylation in the presence of heparin, a supershift was induced in all mutants, yielding a species of about 52 kDa. Heparin induced a complete phosphorylation to this species. The data indicate that (i) phosphorylation of serine 327 is required and sufficient to induce a shift in electrophoretic mobility from a 45-kDa species to a species with an apparent molecular weight of about 48 kDa; and (ii) phosphorylation of serine 156 is not required for and does not interfere with inducing a supershift.

Taken together, the data suggest that serine to aspartate mutations are a useful system for analyzing the contribution of individual sites to phosphorylation-induced changes in tau conformation.

To determine the activity of the mutants in promoting microtubule polymerization, assembly reactions were performed at conditions similar to those of previously published experiments (13, 25). In this assay, 15 µM purified tubulin was incubated with various tau concentrations in assembly buffer containing GTP. The assembly reactions were terminated by glutaraldehyde fixation after 10 min of incubation, and the polymerization products were analyzed using immunofluorescence microscopy following anti-tubulin staining. It should be noted that microtubule assembly as measured in our assay reflects the behavior of a population of microtubules that is the product of both elongation and nucleation processes. Tau(Ala, Ala) and tau(Ala, Asp) promoted the assembly of microtubules similarly to wild-type tau; a significant number of microtubules were polymerized at a concentration as low as 2 µM tau (Fig. 3A). In contrast, only a few microtubules were assembled in the presence of up to 10 µM tau(Asp, Asp), indicating a much lower activity of this fragment to promote microtubule assembly. Although the absolute number of microtubules assembled by tau(Asp, Asp) was lower compared with the other constructs (at 2.2 µM tau), the kinetics of nucleation was similar to wild-type tau (Fig. 3B). No microtubules were observed in the absence of tau or in the presence of corresponding concentrations of a non microtubule-binding control (myoglobin) (not shown).

To test for the activity of the constructs to promote microtubule growth, the lengths of the assembled microtubules after different incubation times were determined. Because we have previously shown that microtubule growth is suppressed under conditions of high nucleation activity (35), a tau:tubulin ratio was used where the number of microtubules was low (2.2 µM tau; see Fig. 3A). All constructs, including tau(Asp, Asp), promoted the assembly of a sufficient number of microtubules to allow length measurements, making it unnecessary to use centrosomes or exogeneously added microtubule fragments to seed assembly. Fig. 3C shows that all constructs effectively promoted microtubule growth. Mean microtubule growth rates, as estimated from interpolating the linear portion of the curve (0-10 min of incubation), were in the range of 1.3-2.2 µm/min for all mutants, suggesting that the low activity of tau(Asp, Asp) in promoting microtubule assembly is primarily caused by a low nucleation activity rather than a low activity to promote growth of existing microtubules.

The balance between the microtubule growth and nucleation activities of tau also depends on the total amount of tubulin in the assembly reaction, with microtubule assembly requiring much less tau with increasing amounts of tubulin (35). To test for the effect of an increased tubulin concentration, assembly reactions were performed at 30 µM tubulin. Tau(wt), tau(Ala, Ala) and tau(Ala, Asp) promoted significant microtubule assembly at tau concentrations as low as 1.5 µM (Fig. 4). Again, tau(Asp, Asp) was less active than the other constructs in promoting microtubule assembly, and only a few microtubules were assembled up to 10 µM (the highest concentration tested). Interestingly, these microtubules had the longest mean length. Thus, at the conditions of increased microtubule nucleation as induced by the high tubulin concentration, tau(Asp, Asp) was most effective in promoting microtubule growth. This is consistent with results obtained previously using a tau deletion mutant that was defective in nucleation activity (35). The results provide evidence that phosphorylation of an individual residue can selectively suppress the microtubule nucleation activity of tau without having a major effect on its growth-promoting activity.

To determine whether the tau mutants differed in the interaction with polymeric tubulin, cosedimentation assays of taxol-stabilized microtubules with the mutated tau proteins were performed. Under the conditions employed, all tau proteins almost completely (>90%) cosedimented with microtubules, whereas a control protein (myoglobin) remained in the supernatant (Fig. 5). No tau was detected in the pellet in the absence of microtubules (not shown). The results are consistent with the finding that all proteins had similar activities in promoting microtubule growth because this activity is thought to reflect the affinity of tau for microtubules. To test for interaction with dimeric tubulin, ligand blotting experiments were performed as described previously (25). All constructs were capable of interacting with tubulin under the conditions employed (1 µg of immobilized tau mutants, 5 µg/ml tubulin; data not shown). Taken together, the data suggest that the changes in conformation and function induced by aspartate mutations are not reflected by a major difference in the interactions of the proteins with tubulin.

It has previously been demonstrated that cytochalasin treatment of cells transfected with microtubule-associated proteins induces microtubule-dependent process formation (36). Using this system, we showed that specific tau sequences were required for process outgrowth in PC12 cells (31). To test the involvement of phosphorylation in this system, pRC/CMV vectors containing epitope-tagged (FLAG) wild-type tau and constructs containing alanine and aspartate double mutations were prepared and transiently expressed in PC12 cells. All tau constructs associated with the cytoskeleton, as judged from the staining after using a combined fixation-extraction protocol indicative of cytoskeletal association (37, 38) (Fig. 6, A and B). We did not observe an obvious difference in the transfection efficiency and the staining intensities of the constructs as judged by visual inspection of the cells, which suggests that all constructs were expressed to a similar extent. Although only few nontransfected cells established processes after cytochalasin treatment, many of the cells expressing the constructs established long and thin cellular processes. Quantitation showed that in the absence of cytochalasin, expression of tau generally did not result in process formation, whereas in the presence of cytochalasin, about 40% of cells transfected with tau(wt) or tau(Ala, Ala) established processes (Fig. 6C). When the cells were transfected with tau(Asp, Asp), this percentage was slightly lower (34%) but not significantly different (p > 0.05). The amount of cells with more than one process was similar for all constructs (59, 47, and 50% of all process bearing cells after transfection with tau(wt), tau(Ala, Ala) and tau(Asp, Asp), respectively), suggesting that there was no major effect of the particular construct on the number of processes per cell.

To analyze the activity of each construct to promote process growth, the mean lengths of the extended processes were determined. Fig. 6D shows that processes from cells that had been transfected with wild-type tau or tau(Ala, Ala) were about two-thirds longer than processes in control cells. Tau(Asp, Asp) promoted the formation of processes with about a 20% longer mean length than the processes formed with wild-type tau or tau(Ala, Ala). This difference was significant for the total number of the processes from all experiments, as well as for the mean of the individual experiments (p < 0.05). Because tau(Asp, Asp) had a lower nucleation activity, this may indicate that in tau(Asp, Asp)-transfected cells, a higher ratio of tau is available for promoting microtubule growth rather than microtubule nucleation, thus resulting in the establishment of longer processes.

DISCUSSION

Conformational changes, as reflected by a decreased electrophoretic mobility on SDS gels, occur after certain phosphorylation events and are characteristic of tau isolated from PHFs from patients with Alzheimer's disease (17-19). Although phosphorylation of tau by PKA appears not to be directly involved in neurodegeneration, it may serve as a useful model system for studying phosphorylation-induced changes in conformation and function. We have previously shown that phosphorylation of tau by PKA induces several discrete shifts in electrophoretic mobility, which correlate with the degree of phosphorylation and differentially affect the ability of tau to promote microtubule growth and nucleation (13). Indirect evidence obtained from in vitro phosphorylation experiments using a panel of truncated tau proteins points toward two of the five phosphorylation sites (Ser-156 and Ser-327) as being of particular importance for inducing a conformational change and a change in the activity of tau.

Our data show that serine to alanine mutations at these positions are neutral toward the conformation of tau and that aspartate mutations at Ser-327 introduce conformational changes similar to those induced by phosphorylation of this residue. The results indicate that phosphorylation of Ser-327 is required and sufficient for one of the conformational changes induced by PKA. Previous experiments using fragments of tau have shown that the presence of residues 154-172 is required for a further shift in mobility. Such a supershift had previously been observed when tau was phosphorylated by PKA. Interestingly, introduction of an additional aspartate mutation at residue 156 did not cause a supershift in mobility on SDS gels. This indicates a complex relationship between phosphorylation events and sequence requirements modulating the conformation of tau. Mutations of Ser-156 did not change the immunoreactivity with the phosphorylation-sensitive antibody Tau-1, whose epitope has been mapped between residues 131-149 (39-41).

Neither the alanine mutations at positions 156 and 327 nor the conformational change induced by the mutation of Ser-327 to aspartate affected the activity of tau in our microtubule assembly assays. However, when Ser-156 is mutated to aspartate, a marked decrease in the activity of tau to promote microtubule nucleation is observed. This confirms previous results where tau, phosphorylated by PKA in the presence of heparin and containing almost completely phosphorylated Ser-156, showed a decreased nucleation activity (13). It is unlikely that dynamic instability would account for the low nucleation activity because under our conditions, microtubules exhibit net growth at both ends even in the absence of any microtubule-associated proteins (42). The results indicate that the nucleation activity of tau is not correlated with its conformation as reflected by its electrophoretic mobility. The fact that Ser-156 is located in the region that has previously been found to be required for the nucleation activity of tau (residues 154-172 (13)) confirms the importance of this domain in regulating the nucleation activity of tau. In addition, the experiments show that a shift in electrophoretic mobility is only a poor indicator for the functional activity of tau.

With the exception of phosphorylation of Ser-262 (5), most of the phoshorylation events of tau have only moderate effects on microtubule binding. In agreement, our results show that all mutated isoforms interact with microtubules to a similar extent despite the large difference in nucleation activity of tau(Asp, Asp). This may indicate that the nucleation activity of tau has additional features beyond the simple binding of tubulin. For instance, during nucleation, tau may undergo conformational changes that bring tubulin heterodimers together. These interactions may be finely regulated by additional sequence and/or conformational requirements. This is consistent with previous results showing that the activities of tau to nucleate microtubules are mechanistically distinct and involve primary structure elements not required for microtubule binding (25). Many of the potential phosphorylation sites of tau, including sites that have been found to be specifically modified in PHF-tau, are localized within a region adjacent to the microtubule binding domain of tau (residues 154-197). It will be interesting to evaluate whether all phosphorylation events within this region have similar effects on the activity of tau or whether nucleation can be regulated by the phosphorylation of specific sites.

Serine to aspartate or alanine mutations allow assessment of the function of certain phosphorylation events in a cellular context. We have previously shown that tau induces microtubule-dependent process formation in actin-depolymerized PC12 cells (31). Interestingly, in this system a lack of process formation was observed following transfection with a tau construct missing residues 154-173. This could be explained by a lack of nucleation activity, by a lower affinity to microtubules, or by a decreased microtubule-stabilizing activity of this construct. Because our results indicate that phosphorylation of Ser-156 may modulate the nucleation activity of tau in vitro, the effect of transfections with the mutant proteins was determined. PC12 cells similarly elaborated processes whether transfected with wild-type tau or with alanine or aspartate mutations, suggesting that phosphorylation of serine 156 and serine 327 is not critical for the interaction of tau with cellular microtubules and the promotion of process outgrowth. The results suggest that the activity of the transfected constructs to nucleate microtubules is not critical for the formation of processes in this system. The low levels of endogenous tau may provide adequate nucleation activity in this assay.

Taken together, our data indicate that serine to aspartate mutations are a useful system to test the effect of single phosphorylated residues on the structure and function of tau. It will be interesting to test the effect of various phosphorylation events that may be correlated with the formation of PHF-tau on the function of tau both in vitro and in cells.