Binding of microtubule-associated protein 2 and tau to the intermediate filament reassembled from neurofilament 70-kDa subunit protein. Its regulation by calmodulin.

Two major brain microtubule-associated proteins (MAPs), MAP2 and tau, were found to bind to the intermediate filaments reassembled from neurofilament 70-kDa subunit protein (= 70-kDa filaments). The binding was saturable. The apparent dissociation constant (KD) for the binding of MAP2 to the 70-kDa filaments was estimated to be 4.8 X 10(-7) M, and the maximum binding reached 1 mol of MAP2/approximately 30 mol of 70-kDa protein. The apparent KD for the tau binding was 1.6 X 10(-6) M, and the maximum binding was 1 mol of tau/approximately 3 mol of 70-kDa protein. It was also found that MAP2 and tau did not compete with each other for binding to the 70-kDa filaments. Most interestingly, calmodulin, a ubiquitous Ca2+-binding protein in eukaryotic cells, was found to inhibit the binding of MAP2 and tau to the 70-kDa filaments. The inhibition by calmodulin was regulated by changes in Ca2+ concentration around 10(-6) M, and was canceled by trifluoperazine, a calmodulin inhibitor.

filaments (17)(18)(19). Interestingly, a recent report has suggested that one of the high molecular weight MAPs, MAP1, might be associated with actin-containing stress fibers rather than microtubules in cultured mammalian cells (20). In addition, it has been reported that another high molecular weight MAP, MAP2, is found in association with intermediate filaments in cultured brain cells (21). A high molecular weight protein similar to MAPs, called plectin, was also reported to be associated with intermediate filaments (22).
In view of these facts, we thought it very important to study interactions of MAPs with actin filaments and intermediate filaments. Previously, we characterized interactions of MAPS with actin filaments in detail (23)(24)(25), and found that the interactions are inhibited by phosphorylation of MAPs (23,24) and by Ca2+-calmodulin (25). In this study, we report our finding that brain MAPs, MAP2 and tau, bind to the intermediate filaments reassembled from neurofilament 70-kDa subunit protein. We also present evidence that the binding is inhibited by calmodulin in a Ca2+-dependent manner.
Previous studies from our laboratory and others demonstrated the in uitro interaction of microtubules with neurofilaments (26,27). We further demonstrated that one of the neurofilament subunit proteins, 200-kDa protein, promotes tubulin polymerization, and suspected that the 200-kDa protein mediates, at least in part, microtubule-neurofilament interactions in neuron (28)(29). On the other hand, Leterrier et al. (30) showed the binding of high molecular weight MAPs to neurofilaments and thereby suggested a possibility of MAPS-mediated interactions. Aamodt and Williams (31) also presented evidence for the MAPs-mediated interactions of microtubules with neurofilaments. The present study extends these studies and provides some important implications to our understanding about the interactions between microtubules and neurofilaments at the molecular level.

EXPERIMENTAL PROCEDURES
Purification of MAPs, MAP2, and Tau-Microtubule protein was isolated from porcine brains by three cycles of temperature-dependent assembly and disassembly (32, 33). MAPs were prepared from microtubule protein by heat treatment (100 "C, 5 min) as previously described (33). MAPs consisted of mainly MAP2 and tau. MAP2 and tau were purified from the MAPs fraction by gel filtration on Bio-Gel A-15m and DE--cellulose chromatography, as described previously (24,25). SDS-polyacrylamide gels of the purified MAP2 and tau were shown in Fig. 1 a and b. porcine brains by gel filtration on Sepharose C1-2B and centrifuga-Purification of Neurofilaments-Neurofilaments were isolated from tion, and then purified by hydroxyapatite column chromatography in the presence of 6 M urea as described previously (28). The fractions containing neurofilament triplet were concentrated by ultrafiltration and then dialyzed against PM solution (100 mM PIPES, 2 mM EGTA, and 1 mM MgS04, pH 6.7) to reconstitute neurofilaments. The reconstituted neurofilaments were washed and suspended in the PM solution and stored a t -80 "C until use. The purity of the purified neurofilaments was shown in our previous paper (28).
Preparation of 70-kDa Filaments-The neurofilament triplet proteins (200,000, 150,000 and 70,000) were separated into each component by DEAE-cellulose chromatography in the presence of 6 M urea according to the method of Geisler and Weber (34). The fractions containing 70-kDa protein were combined, and rechromatographed on DEAE-cellulose to obtain highly purified preparations of 70-kDa protein. Then, they were concentrated by ultrafiltration and dialyzed against the PM solution to reconstitute intermediate filaments, as described previously (28). The filaments were composed of 70-kDa subunit protein with slightly contaminating 150-kDa subunit protein (less than 1% of total protein), as analyzed by SDS-polyacrylamide gel electrophoresis (see Fig. IC) (28). The filaments were referred to as 70-kDa filaments in this paper.
Purification of Calmodulin-Calmodulin was purified from porcine brains as described previously (33).
Pelleting Assay-Binding of MAPs, MAP2, or tau to 70-kDa filaments or neurofilaments was examined by the pelleting assay. Filaments were mixed with the MAPs fraction, and incubated at 0-2 or 35 "C for about 10 min, and then the mixture (150 pl) was centrifuged a t 35,000 X g for 40 min a t 0-2 or 35 "C. The resulting supernatant and pellet fractions were analyzed by SDS-polyacrylamide gel electrophoresis. Gels were stained with Coomassie Blue, and the intensity of the stained band was determined by scanning the gels with a densitometer (Beckman model R-112).
Other Procedures-SDS-polyacrylamide gel electrophoresis was carried out by the method of Laemmli (35) with 10% acrylamide slab gels. Protein concentration was determined by the method of Lowry et al. (36), using bovine serum albumin as a standard.

RESULTS
Binding of MAPs to the 70-kDa Filaments-Intermediatesized filaments composed of the neurofilament 70-kDa subunit protein were prepared as described under "Experimental Procedures." The filaments (= 70-kDa filaments) showed typical characteristics of 10-nm filaments when negatively stained specimens were viewed with an electron microscope, as described in our previous paper (29).
The binding of MAPs to the 70-kDa filaments was examined by the pelleting assay. Under the conditions used, more than 95% of the 70-kDa filaments sedimented. When MAPs (composed mainly of MAP2 and tau) were mixed with the 70-kDa filaments and then centrifuged at 0-2 "C, a portion of MAPs sedimented with the filaments (Fig. 2, O), while in the absence of the 70-kDa filaments MAPs did not sediment at all (Fig. 2, 0). When the binding assay was performed a t 35 "C, the amount of MAPs that sedimented with the 70-kDa filaments (Fig. 2, A) was increased slightly as compared with that at 0-2 "C ( Fig. 2,O). However, at 35 "C a relatively large portion of MAPs sedimented in the absence of the filaments (Fig. 2, A) because MAPs tend to aggregate at higher temperatures. Considering these results, we carried out the binding assay at 0-2 "C hereafter.
Binding of MAP2 and Tau to the 70-kDa Filaments-MAP2 and tau were purified from MAPs fraction, and the binding of each to the 70-kDa filaments was examined by the pelleting assay. As shown in Fig. 1, d and e, both purified MAP2 and purified tau were found to sediment with the 70-kDa filaments. Bovine serum albumin did not sediment with the 70-kDa filaments, and the amount of the cosedimeaed MAP2 or tau did not change irrespective of the presence of bovine serum albumin. This may indicate the specificity of the MAP2 or tau binding as well as the reliability of this assay procedure. When increasing concentrations of MAP2 were mixed with a fixed concentration of the 70-kDa filaments, the amount of bound MAP2 increased and seemed to attain a plateau level (Fig. 3). The binding curve seems to be hyperbolic and a reciprocal plot of the data was linear. An apparent dissociation constant (KO) for the binding and a binding capacity at saturation were estimated by extrapolating the line (Fig. 3, inset). The KD was 4.8 X M and the maximum binding reached 1 mol of MAP2/30 mol of 70-kDa protein, assuming that the molecular weight of MAP2 is 300,000.
The same kind of experiment was carried out for the binding of tau to the 70-kDa filament (Fig. 4). A reciprocal plot of the data was linear (Fig. 4, inset). The apparent dissociation constant was determined to be 1.6 X M, and the maximum binding was 1 mol of tau/-3 mol of 70-kDa protein, assuming that the average molecular weight of tau is 60,000.
Electron microscopic observations showed that MAP2 70-kDa-filament mixture and tau 70-kDa filament mixture were indistinguishable from 70-kDa filament alone when negatively stained specimens were viewed (data not shown). There was no sign of increased cross-linking or amorphous aggregate in the presence of MAP2 or tau. The effect of tau on the binding of MAP2 to the 70-kDa filaments was examined. As shown in Fig. 5A, increasing Concentrations of tau did not appreciably inhibit the binding of MAP2 to the 70-kDa filaments. The effect of MAP2 on the binding of tau to the filaments was also examined (Fig. 5B). The result clearly showed that the binding curve of tau in the presence of MAP2 (0) roughly coincided with that in the absence of MAP2 (0). These results suggest that MAP2 and tau bind to different sites on the 70-kDa filaments.
Calmodulin Inhibits the Binding of MAP2 and Tau to the 70-kDa Filaments in a ea2+-dependent Manner-We previously demonstrated that calmodulin inhibits both microtubule assembly and MAPS-actin interactions in a Ca2+-dependent manner (25,33). Thus, it was interesting to examine the effect of calmodulin and Ca2+ on the binding of MAPS to the 70-kDa filaments. In preliminary experiments we found that calmodulin inhibits the binding of MAP2 and tau to the 70-kDa filaments only in the presence of Ca2+. Fig. 6 shows that the inhibition by calmodulin in the presence of Ca2+ was dependent on the calmodulin concentration. For the halfmaximal inhibition about 5-and 14-fold molar excess of calmodulin over the tau and MAP2 were required, respectively (Fig. 6, A and B). As shown in Fig. 7, trifluoperazine, a calmodulin inhibitor, canceled the calmodulin-induced inhibition of the binding of tau (A) and MAP2 (B) to the 70-kDa filaments. This indicates that the observed inhibition was indeed due to calmodulin and not to any possible contaminants. mg/ml), and calmodulin (0.80 mg/ml) were incubated at 0-2 "C in a buffer solution consisting of 50 mM ME$, pH 6.8, 33 mM KC1, 0.7 mM MgC12, 0.9 mM EGTA, and various concentrations of CaC1,. Free Ca" concentrations were calculated from the apparent association constant of EGTA with Ca'+ (K-) = 5 X lo6 M" at pH 6.8 (33). These samples were then processed for the measurement of cosedimented MAP2 as described in Fig. 2. The amount of bound MAP2 in the absence of Ca2+ was regarded as 100%.
either (data not shown). These data may suggest that MAPs (MAP2 or tau) complexed with Ca2+-calmodulin are unable to bind to the 70-kDa filaments.
Binding of MAPs to Neurofilaments-Neurofilaments were composed of 70-, 150-,and 200-kDa subunit proteins. The binding of increasing concentrations of MAPs to a fixed concentration of neurofilaments was examined and compared with the binding to the 70-kDa filaments. In these experiments the concentrations of neurofilaments and the 70-kDa filaments were adjusted such that both the reaction mixtures contained the same amount of 70-kDa protein, and results were expressed as bound MAPsf70-kDa protein. As clearly seen in Table I, neurofilaments which contained 200-and 150-kDa subunit proteins in addition to 70-kDa protein showed a lower affinity for MAPS than the 70-kDa filaments.

DISCUSSION
In this study, we have found that two major brain MAPs, MAP2 and tau, are capable of binding to the 70-kDa fila-

Binding of increaaing comentratwns of MAPs to the 70-kDa filaments or neurofilaments
The 70-kDa filaments (0.40 mg/ml) or neurofilaments (0.80 mg/ ml) were incubated with increasing concentrations of MAPS at 0-2 "C (because the 70-kDa protein forms about 50% of neurofilament proteins, the neurofilaments at a concentration twice as high as the 70-kDa filaments have about the same amount of the 70-kDa protein as the 70-kDa filaments), and then processed as described in Fig. 2 MAPs were previously regarded as proteins which specifically bind to microtubules (1-7), but recent immunocytological studies (20)(21)(22) have raised the possibility that MAPs bind to other cytoskeIetons as described under the "Introduction." Biochemical studies have also revealed interactions of MAPs with actin filaments in vitro (17)(18)(19)(23)(24)(25).
Our results reported here may imply interaction of MAPs with intermediate filaments in cells. It is well known that although intermediate filaments are universally present in almost all types of cells and tissues, their subunit proteins are respectively specific to each type of cell (38,39). Therefore, a general argument that MAPs will be capable of binding to any type of intermediate filaments cannot be vigorously asserted on the basis of the present experiment alone which used neurofilament 70-kDa protein as a filament source. However, since various types of intermediate filaments are structurally very similar (38,39) and since their subunit proteins have homologous regions in amino acid sequence in common (40- Previously, Letterrier et al. (30) reported the finding that high molecular weight MAPs bind to neurofilaments composed of 200-, 150-, and 70-kDa subunit proteins. The 200and 150-kDa subunit proteins are supposed to protrude from the neurofilament core (43-45), although a question of whether these subunit proteins copolymerize with the 70-kDa subunit protein to participate in the formation of neurofilament core or if they are peripherally attached to core filament made of 70-kDa subunit protein has not been conclusively answered. Leterrier et al. speculated that MAPs bind to protruding portions of neurofilaments, i e . , 200-or 150-kDa subunit proteins, and further suggested that the interactions between neurofilaments and microtubules may be represented as arm-to-arm binding rather than arm-to-core (30). However, our data together with the data reported by Heimann et al. (37) indicate that MAPs bind to the 70-kDa subunit protein in the neurofilaments, since the 70-kDa filaments have a higher affinity for MAPs than the neurofilaments composed of all of the triplet proteins. Thus, if MAPs mediate the interactions between microtubules and neurofilaments, the interactions may be represented as arm-to-core binding. We previously suggested another possibility that the 200-kDa subunit protein mediates the neurofilament-microtubule interactions based on the finding that both the purified 200-kDa protein alone and the 200-kDa protein incorporated into the filaments are capable of promoting tubulin polymerization, thereby causing network formation (28,29). A number of electron microscopic observations have revealed thin, filamentous cross-bridging structures between microtubules and neurofilaments in neuron (16,46). Whether these structures are composed of MAPs or the 200-kDa subunit protein or both will be elucidated in the future work.
In this study, we further found that calmodulin inhibits the binding of MAP2 and tau to the 70-kDa filaments in a Ca2+dependent manner. Since MAPs, both MAP2 and tau, are reported to hind calmodulin in a Ca2+-dependent manner (47)(48)(49), and since we did not obtain any evidence for Ca2+dependent binding of calmodulin to the 70-kDa filaments, it is reasonably inferred that Ca2+-calmodulin inactivates the ability of MAPs to bind to the 70-kDa filaments by forming a complex of Ca'+-calmodulin-MAPS. We recently demonstrated that calmodulin inhibits the interactions of MAPs with actin filaments (25). Moreover, it is widely known that calmodulin regulates microtubule assembly in a Ca'+-dependent manner (33,50). Thus, calmodulin is capable of regulating the interactions of MAPs with three types of cytoskeleton in uitro. Considering the significance of calmodulin as a Ca'+ receptor together with the possible significance of MAPs as regulators of cytoskeleton, it is possible that calmodulin plays some roles in the Ca'+ regulation of cytoskeletal networks in living cells by controlling the function of MAPs.