Phosphorylation of Tau Proteins to a State Like That in Alzheimer’s Brain Is Catalyzed by a Calcium/Calmodulin-dependent Kinase and Modulated by Phospholipids*

Calcium/calmodulin(CaM)-dependent protein ki- nases isolated from bovine and rat brains phosphorylate the microtubule-associated tau protein in the mode that shifts the mobility of tau in sodium dodecyl sulfate-polyacrylamide gel electrophoresis (mode I). This mode of tau phosphorylation is the one that occurs abnor-mally in Alzheimer’s lesions. Purified tau protein in solution can be phosphorylated by the Caa+/CaM kinases maximally to about 50% of the total tau protein. Incorporation of one phosphate group per mol of tau is sufficient to shift the protein to a slower migrating electrophoretic band. Additional phosphate incorporation into the shifted tau proteins can occur depending on protein kinase concentration. In the presence of phosphatidylserine, tau proteins were phosphorylated to an extent of 100% at a tau:phosphatidylserine ratio of 20. Phosphatidylethanolamine also stimulated tau phosphorylation by Ca2+/CaM kinase and phosphati- dylinositol was found to be a potent inhibitor of tau protein phosphorylation. The direct observation that tau proteins interact with phospholipids such as phosphatidylethanolamine and phosphatidylinositol, re- sulting in a smearing of the protein band on sodium dodecyl sulfate-gel electrophoresis, supports the pos- sibility that tau protein may interact with phospholipid membranes in vivo and that tau protein phosphoryla- tion


Phosphorylation of Tau Proteins to a State Like That in Alzheimer's Brain Is Catalyzed by a Calcium/Calmodulin-dependent
Calcium/calmodulin(CaM)-dependent protein kinases isolated from bovine and rat brains phosphorylate the microtubule-associated tau protein in the mode that shifts the mobility of tau in sodium dodecyl sulfatepolyacrylamide gel electrophoresis (mode I). This mode of tau phosphorylation is the one that occurs abnormally in Alzheimer's lesions. Purified tau protein in solution can be phosphorylated by the Caa+/CaM kinases maximally to about 50% of the total tau protein.
Incorporation of one phosphate group per mol of tau is sufficient to shift the protein to a slower migrating electrophoretic band. Additional phosphate incorporation into the shifted tau proteins can occur depending on protein kinase concentration. In the presence of phosphatidylserine, tau proteins were phosphorylated to an extent of 100% at a tau:phosphatidylserine ratio of 20. Phosphatidylethanolamine also stimulated tau phosphorylation by Ca2+/CaM kinase and phosphatidylinositol was found to be a potent inhibitor of tau protein phosphorylation. The direct observation that tau proteins interact with phospholipids such as phosphatidylethanolamine and phosphatidylinositol, resulting in a smearing of the protein band on sodium dodecyl sulfate-gel electrophoresis, supports the possibility that tau protein may interact with phospholipid membranes in vivo and that tau protein phosphorylation could be modulated by the phospholipid composition of the membranes with which tau interacts.
Microtubules are one of the major constituents of the neuronal cytoskeleton and play important roles in cell morphology and intracellular transport processes. In addition to tubulin and microtubule-associated protein-2, microtubules contain tau proteins (1, 2). Tau is a family of four closely related proteins of 55-68 kDa. We showed that tau can undergo two modes of phosphorylation, one which changes the electrophoretic mobilities of the tau's (mode I) and another, which does not (mode 11) (3,4). It was clear that in vivo brain tau contains a mixture of nonphosphorylated tau and tau phosphorylated in at least mode I. The phosphorylation of tau, moreover, partially inhibits its ability to promote tubulin polymerization in vitro (3) and thus might regulate microtubule formation. Still more recently, it was reported that mode I-phosphorylated tau proteins are major components of Alzheimer paired helical filaments, and this suggested that the phosphorylation of tau proteins, at least in mode I, might be a significant factor in cytoskeletal alterations that affect neurons in Alzheimer's disease (5,6). Therefore, there is considerable interest in understanding the factors that might regulate tau phosphorylation in vivo and thus probably their functions. The primary focus of this article is on the phosphorylation of tau by the Ca2+/CaM' protein kinase. This kinase catalyzes mode I phosphorylation of tau proteins in vitro, and its activity was modulated by the interaction of tau protein membrane phospholipids.

EXPERIMENTAL PROCEDURES
Materiak-[y3*P]ATP was supplied from ICN-Radiochemicals (catalog number 35020). Phosphatidylethanolamine, phosphatidylcholine from egg, and phosphatidylinositol from bovine liver were obtained from Avanti Polar-Lipids, Inc. Phosphatidic acid, phosphatidylglycerol from egg, and phosphatidylserine from bovine brain were from Sigma. Calmodulin was prepared as described in Ref. 7. Tau proteins were prepared as described in Ref. 8. The purified tau 1 and tau 2 protein species used in this study have been treated twice with alkaline phosphatase (40 units/ml; 5 h) to maximize their dephosphorylation.
Purification of Ca2+/Caldulin-dependenf Tau Protein Kinuse from Bovine Brain-One fresh bovine forebrain obtained from the local slaughterhouse was homogenized in 500 ml of Buffer A 50 mM Tris, 3 mM EGTA, 1 mM dithiothreitol, 0.5 mM phenylmethylsulfonyl fluoride, pH 7.5, for two 30-s bursts in a Waring blender at 4 "C. All subsequent steps were at 4°C. The homogenate was centrifuged at 9000 X g for 40 min. The supernatant was filtered on glass wool and brought to 45% ammonium sulfate. After 30 min of continuous gentle stirring, the protein solution was centrifuged at 9000 X g for 60 min. The pellet was resuspended in buffer A and dialyzed extensively overnight against 15 liters of buffer A. After dialysis, the solution was clarified by centrifugation and loaded on a DEAE-Sephadex A50 column (3 X 15 cm) previously equilibrated with buffer A. After washing the column with 2 volumes of buffer A, the bound proteins were eluted with buffer A containing successively 0.2 M NaCl, 0.4 M NaCI, and 0.8 M NaCl. The protein fractions were then tested for Ca2+/calmodulin-dependent kinase activities using tau proteins as substrate. Only the 0.2 M NaCl fractions showed significant tau kinase activity. The fractions that contained the maximal activity (13-20 ml) were pooled together and dialyzed against buffer B containing 50 mM Tris, pH 7.5, 0.2 mM dithiothreitol, 0.2 M NaCl. After dialysis, 1 mM Ca2+ was added to the enzyme solution and this was applied to a 1 X 7-cm column of calmodulin-Sepharose equilibrated in buffer B plus 1 mM CaCl,. The column was rinsed with 10 bed volumes of buffer B plus 1 mM CaCl,, and the tau kinase activity was eluted with buffer B containing 2 mM EGTA. Only the concentrated protein kinase fractions were pooled together, stored at -2O"C, and used within 1 month. Four different preparations gave similar results, and they were all characterized by a very low specific activity compared The abbreviations used are: CaM, calmodulin; EGTA, [ethylenebis(oxyethylenenitrilo)]tetraacetic acid SDS, Spdium dodecyl sulfate; PAGE, polyacrylamide gel electrophoresis. to rat kinase preparations (see below).
Purification of Tau Protein Kinuse from Rat Brain-The purification procedure used was essentially that of McGuinness et al. (9) except that buffers D and E contained 12% glycerol, Sephacryl S400 gel filtration was replaced by Sepharose CL-4B gel filtration column, and the last concentration step of the enzyme by dialysis against solid sucrose was omitted. The enzyme was assayed with crude tau (0.5 p~) as substrate. During all the purification steps the tau protein kinase activity was entirely dependent on Ca2+ and calmodulin and exhibited the same physicochemical characteristics as the Ca2+/CaMdependent microtubule-associated protein-2, synapsin, or casein protein kinase (9)(10)(11)(12)(13).
Protein Phosphorylation Assay-The standard protein assay for tau phosphorylation by the rat brain kinase contained 20 mM Tris-HCI, pH 7.5, 3 mM Mg+, 0.2 mM Ca'+, 50 p~ [y3'P]ATP. After preincubation for 30 sec at 30"C, the reaction was initiated by the addition of [y3'P]ATP and terminated by addition of an SDS-stop solution (5% SDS, 5% @-mercaptoethanol, 25% glycerol, 0.5 M Tris, pH 6.8) and heated for 2 min at 100 "C. The proteins were separated by 0.1% SDS. 10% polyacrylamide gel electrophoresis. The gels were dried and the phosphorylated protein detected by autoradiography. The phosphorylated proteins were cut out of the gels and then radioactivity was measured in a liquid scintillation counter. The stoichiometry of phosphorylation was calculated by use of the specific activity of the ATP and the amount of protein loaded on the gel.
Preparatwn of Phospholipid Solutions-Phospholipid solutions were prepared by slowly evaporating chloroform from commercially prepared phospholipid solutions in a stream of nitrogen gas. Dry phospholipids (1-2 mg/ml) were resuspended in 20 mM Tris buffer, pH 7.5, by vortex agitation and sonicated for 5-10 min. The solutions were stored a t 4 "C and were sonicated once more prior to each use.

din-dependent Protein Kinase Purified from Bovine Brain-
Tau protein directly extracted from brain tissue contains eight components which usually migrate as seven bands in SDS-PAGE (3,4). After treatment with alkaline phosphatase, they are converted to a four-band pattern that we call, respectively, T~ to 7, from the fastest to slowest (8). The change in electrophoretic mobility of the dephosphorylated tau proteins upon dephosphorylation must reflect conformational changes between the phosphorylated proteins and their dephosphorylated counterparts, since phosphorylation (in mode 11) is possible without change in mobility (8).
We searched among known protein kinases for one that might mimic the tau phosphorylation that was proven to occur in vivo; i.e. mode I that caused the conformational changes shifting taus to slower electrophoretic migrations. The calcium/phospholipid-dependent protein kinase (protein kinase C) proved able to phosphorylate tau protein in vitro, but phosphorylation of tau by this kinase failed to change the electrophoretic mobility (8) and therefore did not represent mode I. The CAMP-dependent protein kinase did not substantially phosphorylate tau protein in vitro. We turned our attention to Ca2+/calmodulin-dependent kinase activities, since Shulman (10) reported that tau proteins are good substrates for such a kinase purified from rat brain.
A Ca2+/calmodulin-dependent protein kinase was first purified from bovine brain using tau protein as exogenous substrate to monitor the kinase activity during the preparation procedure (see "Experimental Procedures"). The enzyme preparation obtained after affinity chromatography on calmodulin-Sepharose had a low specific activity when compared to rat brain preparation (see below) and was heavily contaminated by a protein of 68 kDa. A striking feature of the Ca2+/ CaM kinase is autophosphorylation that requires both calcium and calmodulin (12, 14, 15). Incubation of our enzyme preparation under conditions of Ca2+/CaM-dependent autophosphorylation resulted in phosphorylation of a single 60-kDa protein as revealed on the autoradiogram of the SDS-polyacrylamide gel used to study enzyme autophosphorylation (not shown). This single phosphorylated band might correspond to autophosphorylated enzyme since it has the same mobility as the subunit of rat enzyme (12). Phosphorylation of tau protein by the enzyme we prepared was almost totally dependent on the presence of calcium and calmodulin ( Fig.  1). There was no apparent preference of the kinase for one or another tau protein species (Fig. 2, lune 5). Furthermore, phosphorylation of tau by the CaM kinase induced shifts in the mobilities of the phosphorylated tau proteins that resembled those which occur in vivo (3,4) and more particularly, in degenerative neurons of Alzheimer's brain (5,6).
Although the phosphorylation of the mixture of four tau proteins shown in Fig. 2, lune 5, demonstrated that a fast set of four electrophoretic bands was shifted to a slow set of four, the pattern was too complicated to show whether or not there were multiple phosphorylations that produced intermediate shifts in mobility. Use of pairs or individual tau species has allowed the demonstration that each tau protein species considered independently was shifted to a single slower migrating electrophoretic band as a result of phosphorylation by this kinase (Fig. 2, lunes 1-4). The autoradiogram in Fig. 3 shows the time course of T~ phosphorylation by the bovine CaM kinase. These data confirmed that there was no intermediate phosphorylated tau form between the previously recognized nonphosphorylated and phosphorylated proteins. Since no phosphate was present in the unshifted band of protein, this observation indicated that incorporation of a single phosphate group per molecule was sufficient to induce the mobility change. Further studies on the stoichiometry of phosphate incorporation by bovine CaM kinase in the shifted T~ and   Fig. 1. A and B represent, respectively, the Coomassie Blue staining and autoradiography of the same gel.
polypeptide bands proved that no more than one phosphate group was incorporated per mol of tau (data not shown).
More surprising was the observation that even prolonged incubation of tau proteins with the kinase failed to shift all the tau protein to an apparent higher molecular weight, suggesting that a subpopulation (about half) of tau could not be phosphorylated (Figs. 2 and 3). Measurement of inorganic phosphate after ashing the protein (16) showed that the tau preparation used in this study was almost completely dephosphorylated, and thus the CaM kinase-resistant tau protein did not represent a phosphorylated subpopulation of tau. The other possibility was that the low specific activity of the bovine kinase preparation we used was not sufficient to allow complete tau phosphorylation. Therefore, since rat brain had a higher specific activity than bovine brain (12), we decided to examine the rat brain enzyme.

Phosphorylation of Bovine Brain Tau Protein by a Ca2+/ Calmodulin-dependent Protein Kinase Purified from Rat
Brain-The purification procedure used for the rat brain CaM kinase was essentially that of McGuinness et al. (9). Tau proteins were used as exogenous substrate to monitor the kinase activity. The enzyme migrated as two distinct polypeptide bands in SDS-polyacrylamide gel electrophoresis having apparent molecular masses of 50 and 60 kDa. Autophosphorylation of the enzyme resulted in a mobility shift of the 50and 60-kDa substrates to higher apparent molecular weights with a concomitant decrease in Coomassie Blue staining intensity of the polypeptide bands (12). The specific activity of the rat brain enzyme preparation was about 1000-fold that of the bovine preparation. Phosphoryhtwn- Fig. 4 shows the Coomassie Blue staining of the SDS-polyacrylamide gel and the corresponding autoradiogram used to study the effect of increasing rat Ca2+/ CaM kinase concentration on 7 2 phosphorylation. As observed with the bovine Ca2+/CaM kinase, phosphorylation of r2 by the rat enzyme induced a shift of the phosphorylated 7 2 species to an apparently higher molecular weight compared to its dephosphorylated counterpart. As the protein kinase concentration was increased from 0.6 to 13 pg/ml, there was a progressive increase in 32P incorporation into 7 2 protein, as revealed on the autoradiogram of the gel. However, the Coomassie Blue staining clearly shows that phosphate incorporation at high kinase concentrations was essentially all in the shifted phosphorylated T* species and that a constant subpopulation of 7 2 remained unphosphorylated. In addition to the mobility shift noted earlier, high levels of kinase produced superphosphorylated species of tau with progressively lower mobilities. This superphosphorylation of 7 2 was not observed with the bovine enzyme, probably because of the low specific activity of the preparation. The superphosphorylation of 7 2 protein results ultimately in a smear of the protein band on SDS-polyacrylamide gel. Fig. 5 shows the time course of r2 protein phosphorylation by the rat Ca2+/CaM kinase at two different enzyme concentrations, i.e. 0.9 and 9 pg/ml. As expected from the previous results, there were large differences in stoichiometry of 7 2 phosphorylation between the two kinase concentrations used. Since the percentages of unphosphorylated r2 protein as shown on the Coomassie Blue staining of the gel remain nearly identical whatever the kinase concentrations used (Fig.   a), the differences in stoichiometry mainly reflect differences in 7 2 superphosphorylation. At low kinase concentrations, after 5-min incubation, the stoichiometry was about 0.5 mol of 32P/mol of 7 2 protein and corresponds to the monophosphorylation and shift of about 50% of the 7 2 protein to slower mobility on the SDS-polyacrylamide gel (Fig. 5, inset). This percentage of shifted tau protein species did not significantly change as the incubation time was increased, and therefore the apparent linear increase in the stoichiometry of phosphorylation of 7 2 between 5 and 50 min might probably reflect occurrence of 7 2 superphosphorylation. Thus, as previously observed with bovine brain Ca2+/CaM kinase, the rat  enzyme was not able to fully phosphorylate the 72 protein even at very high enzyme concentration. Approximately 50% of the T2 protein remained resistant to phosphorylation. Similar behavior was observed for purified T~ protein (Fig. 6). With T~ protein, superphosphorylation also occurred shifting the primary phosphorylated tau protein species to progressively higher apparent molecular weights.

The Effect of Ca2+/CaM Kinase Concentration on Tau Protein
The Effect of Phosphatidylserine on Tau Protein Phosphorylation- Fig. 5, inset, shows the effect of phosphatidylserine on 7 2 phosphorylation monitored by its electrophoretic mobility shift in SDS-PAGE. The presence of phosphatidylserine with the enzyme in the incubation medium induced a 100% shift of 72 to the lower mobility that suggests a total phosphorylation of tau by the Ca2+/CaM kinase. (Phosphatidylserine alone had no effect on 7 2 mobility in SDS-PAGE; see below.) Indeed, the time course of 7 2 phosphorylation in the presence of phosphatidylserine and low enzyme concentration proved that a stoichiometry of 1 mol of 32p/mol of 7 2 protein was reached within 10 min, followed by a further moderate phosphorylation that is probably accounted for by tau superphosphorylation (Fig. 5). The addition of diacylglycerol to phosphatidylserine did not significantly change these results, and diacylglycerol alone did not affect tau phosphorylation. The effect of phosphatidylserine on T~ phosphorylation by the Ca2+/CaM kinase is shown in Fig. 6. As observed with 7 2 protein, phosphatidylserine stimulated T~ phosphorylation with a shift of all the protein to lower mobility on SDS-PAGE (Fig. 6, ZZA), in comparison to the phosphorylation of only half the T~ in the absence of this phospholipid (IA). As expected, the presence of phosphatidylserine doubled the amount of incorporated 32P as shown on the corresponding autoradiograms (ZB and ZIB).
The effect of phosphatidylserine on the rate and extent of phosphorylation of purified 7 2 protein was examined at various concentrations of the phospholipid and 7 2 protein. Fig. 7 shows that phosphatidylserine, in addition to stimulating 7 2 protein phosphorylation in a 72:Ps ratio-dependent manner, inhibited 7 2 phosphorylation at higher concentrations. The maximal rate of tau phosphorylation was a t a 72:Ps ratio of about 20. These results suggested an effect of phosphatidylserine on tau protein rather than on activation of the kinase.
The Effect of Other Phospholipids on Tau Protein Phosphorylation-Various phospholipids were compared to phosphatidylserine at similar concentrations (70 pg/ml) for their ability to stimulate the phosphorylation of purified T~ and 7 2 proteins by Ca2+/CaM kinase. Figs. 8 and 9 show the Coommassie Blue staining of SDS-polyacrylamide gels ( A ) and their corresponding autoradiogram ( B ) used to study the effects of various phospholipids on T~ and 7 2 phosphorylation. Figs. 8C and 9C show the effects of phospholipids on T~ and 7 2 mobility in the absence of kinase. In the conditions used here, phosphorylation of tau in the absence of phospholipids resulted in the phosphorylation of approximately 50% of tau protein.
As expected, in the presence of phosphatidylserine and diacylglycerol or phosphatidylserine alone, essentially 100% of tau protein was phosphorylated. The tau protein mobility in SDS-gel electrophoresis was not changed in the presence of these phospholipids if the kinase was omitted (Figs. 8C and   9C).
The presence of phosphatidylethanolamine shifted all the tau protein to lower mobility and like phosphatidylserine, increased 32P incorporation into tau protein. However, the phosphorylated tau bands appeared much more diffuse on the autoradiogram. The phosphorylated 7 2 protein even resolved into two distinct phosphoproteins when electrophoresis was extended (not shown). Since it was also observed that phosphatidylethanolamine in the absence of kinase significantly smeared the tau protein band on SDS-PAGE (Figs. 8C and  9C), whereas phosphatidylserine did not, it is impossible to say whether or not there were different modes of tau phosphorylation in the presence of the two phospholipids.
Phosphatidylinositol was found to be a potent inhibitor of tau protein phosphorylation and appeared to have the same effect as phosphatidylethanolamine on the mobility of tau in   presence ( A and B ) of rat brain Ca2+/CaM kinase (0.9 pg/ml) and subjected to 10% polyacrylamide, 0.1% SDS-gel electrophoresis. A and B represent, respectively, the Coomassie Blue staining and autoradiography of the same gel. PS, phosphatidylserine; PE, phosphatidylethanolamine; PC, phosphatidylcholine; PI, phosphatidylinositol.

SDS-PAGE.
There was a significant smearing of the tau protein band in the presence of phosphatidylinositol alone (Figs. 8C and 9C).
Phosphatidic acid at the concentration used slightly potentiated 7 2 phosphorylation but had the same effect as phospha- tidylinositol on the mobility of tau in SDS-gel electrophoresis in the presence and in the absence of the kinase. There was a significant smearing of the tau protein band (Fig. 9C).
In the presence or absence of kinase, phosphatidylcholine appeared to affect neither the incorporation of phosphate into tau nor the electrophoretic mobility of the protein. Similarly, phosphatidylglycerol had no significant effect on T protein phosphorylation nor on the electrophoretic mobility of the protein (not shown). Fig. 10 shows the effect of phospholipid concentrations on the extent of phosphorylation of purified 7 2 protein for three different membrane phospholipids. Phosphatidylethanolamine brought about the highest levels of phosphate incorporation of 7 2 and did not show an inhibitory effect at high concentration. Phosphatidylinositol, in contrast, had a significant inhibitory effect even at low concentration and did not stimulate tau protein phosphorylation at all. Phosphatidic acid had a pronounced stimulatory effect on 7 2 phosphorylation at low concentration, but it also drastically reduced r2 phosphorylation at high concentration.

DISCUSSION
We showed that purified tau protein in solution can be phosphorylated by Ca2'/CaM kinase maximally to about 50% of the total tau protein. Incorporation of one phosphate group/ mol of tau shifted the protein to a lower migrating electrophoretic band. For a number of reasons, we believe that there is physiological significance in the phosphorylation of tau by the Ca2'/CaM kinase. First, the mode of phosphorylation produced by this kinase is the one most clearly shown to exist in vivo (3,4). Second, the endogenous state of phosphorylation of tau in normal brain is a balance between nonphosphorylated and phosphorylated forms (3, 4). Third, the level of phosphorylation is known to affect the ability of tau to promote microtubule formation (3). Fourth, the phosphorylation by the Ca2'/calmodulin-dependent kinase appears to produce a conformational change in tau, and it seems likely that such a change would affect the microtubules into which those tau conformers were incorporated with regard to their stability or their interactions with other cytoskeletal components. Finally, the possibility of physiological significance in tau phosphorylation by Ca2+/CaM kinase is enhanced by the recent report that the tau proteins purified from paired helical filament of Alzheimer's brain are phosphorylated (5, 6) and have slower electrophoretic mobilities than their dephosphorylated counterparts. They are thus identical to tau species produced by phosphorylation of purified tau with the Ca2+/ CaM kinase.
Concerning this last point, a major challenge in understanding the biochemical alteration in Alzheimer's neurons is the elucidation of molecular events underlying such long lasting changes in the phosphorylation of tau proteins. The Ca"/ calmodulin-dependent kinase is one enzyme that could mediate such changes because it appears to act as a molecular switch that remains active long after an initial triggering event (14, 15). Ca2+ and calmodulin stimulate autophosphorylation of the enzyme and incorporation of 3-30 phosphate groups/holoenzyme (15). The phosphorylated form that results from autocatalysis becomes autonomous, being completely independent of Ca" and calmodulin. Such properties allow the enzyme to remain active long after the decay of an initial Ca2+ signal. Dephosphorylation by a distinct phosphatase is required to turn off the enzyme (15). An imbalance of the autophosphorylation-phosphatase system involving either an increase in kinase or a decrease in phosphatase activity that normally would control autophosphorylation of the en-zyme might be responsible for the abnormal phosphorylation of tau protein in Alzheimer's neurons. In this respect, it is interesting to note that Saitoh and Dobkins (17) reported the increase in phosphorylation of a M, 60,000 protein in brain from patients with Alzheimer's disease. This protein is a minor component of the cytosolic extract of Alzheimer's brain, and the authors suggested that it may correspond to a M, 60,000 protein kinase. Although it could well be coincidence, we note that our Ca2+/CaM kinase purified from bovine brain also migrated as a single M, 60,000 polypeptide upon autophosphorylation as well as the @ subunit of the rat enzyme. Comparison of the antigenic properties of the two proteins might be a first step in testing for a common identity of these proteins.
The finding that purified tau protein in solution could only be phosphorylated to a limit of approximately 50% suggests that some interaction between tau proteins can occur that prevents phosphorylation of half of it. Whether these interactions require phosphorylation of the first half of tau protein or are independent of previous phosphorylation is not known at present.
The fact that in the presence of phosphatidylserine, tau proteins were phosphorylated to an extent of 100% at a tau:phosphatidylserine ratio of 20 suggests that the effect of phosphatidylserine was probably through a breakdown of tautau interactions. The specific requirements for the primary structure at the phosphorylation site of protein substrates have been delineated for a number of protein kinases (18). In addition, substrate activity of protein may also be affected by the conformation of the protein (19,20). Results of the present study apparently confirm the latter point. Whether the purified tau proteins in solution or bound to phosphatidylserine liposomes assume their physiological conformations is not known, and further studies are required to clarify this question. However, different conformation states of tau protein in Alzheimer's brain compared to normal brain might also well account for the different extent of tau phosphorylation found in the two cases.
The direct observation that tau proteins can also interact with other phospholipids such as phosphatidylethanolamine and phosphatidylinositol, resulting in a smearing of the protein band on SDS-PAGE, supports strongly the possibility that tau protein may indeed interact with biological membrane in vivo. Furthermore, the fact that phosphatidylserine and phosphatidylethanolamine have effects opposite to that of phosphatidylinositol on tau phosphorylation brings out the interesting possibility that tau protein phosphorylation could be modulated by the phospholipid composition of the membranes with which tau interacts. Studies on the interactions and phosphorylation of tau with mixed phospholipid liposomes would not only illuminate the effect of the various phospholipids on tau protein phosphorylation but might shed new light on the biological function of tau protein in vivo.