Vol. 148, No. 3, 1987

November 13, 1987

BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS Pages 1453-1459

THE 28,000 M r MICROTUBULE-BINDING DOMAIN OF MICROTUBULE-ASSOCIATED PROTEIN-2

ALSO CONTAINS A NEUROFILAMENT-BINDING SITE*

GREGORY FLYNN +, JOHN C. JOLY, AND DANIEL L. PURICH

Department of Biochemistry and Molecular Biology, University of Florida College of Medicine,

Gainesville, Florida 32610

Received October 2, 1987

Summary: We have developed a thrombin proteolytic cleavage procedure to obtain higher yields of the M r 28,000 microtubule-binding and M r 240,000 microtubule-projection components of MAP-2. The former is a highly basic component, whereas the latter and intact MAP-2 are acidic polypeptides. Most notably, our studies reveal that this M r 28,000 fragment binds to neurofilaments, but the M r 240,000 projection domain fails to interact. These data indicate that microtubules and neurofilaments share a common binding site on high-molecular- weight MAP-2. ® 1987 Academic Press, Inc.

Neuronal cytoplasm is highly organized, and both microtubules and

neurofilaments run parallel with respect to the axon's longitudinal axis in a

manner suggesting microtubule-to-neurofilament cross-linking (1-3). In vitro

observations indicate that microtubules interact with neurofilaments (4-7), and

microtubule-associated proteins (MAPs) can enhance the attainment of high

solution viscosity and/or gelation. Aamodt and Williams (8) used falling-ball

viscometry to demonstrate the occurrence of an optimal MAP level in plots of

viscosity/gelation versus MAP concentration; they likened this MAP concentration

profile to that of bivalent antibody cross-linking in immunoprecipitin formation.

Our own studies (9) have traced this apparently optimal MAP profile to the

presence of endogenous GTPase activity, and the inhibition of cross-

linking/gelation at high MAPs can be eliminated with a GTP-regenerating system.

Nonetheless, high-molecular-weight MAPs do bind to microtubules (10-12) and

neurofilaments (4-7), and some interactions of the M r 280,000 neuronal MAP-2 have

been explored by limited proteolytic fragmentation. Tubule binding is restricted

*This research was supported in part by United States Public Health Service Grant GM-36149 from the National Institutes of Health.

+Predoctoral trainee, Department of Chemistry, University of California, Santa Barbara, California 93106.

Abbreviations: MAP, microtubule-associated protein; SDS, sodium dodecyl sulfate; NF, neurofilament; Tb, tubulin, NEPHGE, non-equilibrium pH gel electrophoresis.

1453

0006-291X/87 $1.50 Copyright © 1987 by Academic Press, Inc.

All rights of reproduction in any .)Corm reserved.

VoI. 148, No. 3, 1987 BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS

to a M r 30,000 tryptic or chymotryptic fragment of MAP-2, and the remaining M r

240,000 component corresponds to the lateral projections observed in electron

micrographs of microtubules decorated with MAP-2 (10,13).

During the course of studies on neurofilament-microtubule-MAP-2

interactions, we sought to localize the site(s) of neurofilament binding with

respect to the tubule-binding and -projection domains of MAP-2. A thrombin

cleavage technique was developed to obtain these MAP-2 fragments in higher yields

than that obtained with trypsin or chymotrypsin. Interestingly, the M r 28,000

tubule-binding domain was found to contain a neurofilament-binding site. Our

studies also suggest that this binding interaction has considerable ionic

character, as suggested by isoelectric point determinations of the MAP-2

fragments.

EXPERIMENTAL PROCEDURES

Materials: Bovine thrombin (catalog number, T 4648) and the catalytic subunit of cAMP-dependent protein kinase were purchased from Sigma. Ultrapure ammonium sulfate and urea were purchased from Schwarz-Mann, and Aquacide I from Cal-Biochem. [V-32p]ATP (sp. act. >7000 Ci/mmol) was an ICN product, and Ampholines were obtained from LKB. Assembly buffer (pH 6.8) contained 0.i M piperazine-N,NF-bis[2-ethanesulfonicacid], 1 mM ethyleneglycol-bis[~-amino-ethyl ether]-N,N,N',N'-tetracetic acid, and 1 mM magnesium sulfate.

Methods: Bovine brain microtubule protein was prepared by the procedure of Shelanski et al. (14). Neurofilaments were prepared from fresh bovine spinal cord by the method of Delacourte e_tt al. (15), as modified by Letterier e_!t a l. (7). Neurofilament triplet protein was prepared as described by Tokutake e!t al. (16), and the M r 68,000 neurofilament subunit was purified according to the method of Geisler and Weber (17).

MAP-2 was purified by the method of Herzog and Weber (18), concentrated by ammonium sulfate precipitation, and phosphorylated by the catalytic subunit of cAMP-dependent protein kinase in assembly buffer. Typically 500 units of kinase was dissolved in 0.025 ml dithiothreitol (50 mg/ml), incubated at room temperature for I0 min, and used immediately with 1.5 mCi [7-3zp]ATP, 20 #M unlabeled ATP, for 30 min at 37°C in the presence of approximately 40 mg heat- stable MAPs. The MAPs were separated on a BioGel A-I.5M (BioRad, Inc.) column with the MAP-2 fractions pooled and concentrated in a dialysis bag against dry Aquacide I at 4°C. This purified [32P]MAP-2 was clarified prior to use at 130,000 x g for 25 min in a Beckman Airfuge. Tubulin was separated from MAPs and tau by the phosphocellulose method of Weingarten et al. (19).

SDS gel electrophoresis was carried out as described by Laemmli (20), and "NEPHGE" gels were prepared and electrophoresed using the method of Roberts e_!t al. (21). Gels were stained with Coomassie Blue, destained, dried under vacuum, and exposed to Kodak X-AR5 film at -80°C.

RESULTS

In view of the high molecular weight of MAP-2, proteolytic fragmentation by

trypsin or chymotrypsin has proven to be useful in defining the MAP-2 domain(s)

interacting with other cytoskeletal components (22). We have found, however,

that the stability of such proteolytic fragments is quite limited, leading to the

loss of the initially cleaved domains and the ability of these fragments to

stimulate microtubule assembly. In a survey of the action of other proteases, we

1454

Vol. 148, No. 3, 1987 BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS

(A) NF No NF Protein Tubulin or Tb

M ~ + + - - - - + + - - - M

(B) NF No NF Protein Tubulin or Tb

\s p s p A s p s pA s p s p / \ s p s p A s p s p A s p s p/ I 2 3 4 5 6 7 8 9 10 II 12 I 2 3 4 5 6 7 8 9 10 II 12

Fig. i. Binding of MAP-2 or MAP-2 Fragments to Neurofilaments and Tubulin. Panel A, Protein Staining Pattern; Panel B, Autoradiogram of [32p]-Labeled Proteins. Radiolabeled MAP-2 was incubated with (+) or without (-) thrombin or assembly buffer alone as described earlier, quenched with i mM phenylmethanesulfonyl fluoride and incubated with 2 mg/ml neurofilament protein at 4°C for i0 min or i mg/ml taxol-stabilized tubulin (phosphocellulose purified) and i mM GTP at 37°C for 30 min. The samples were layered over 20% sucrose (w/v) in assembly buffer and pelleted at 130,000 x g for 20 min. Supernatants were removed and the pellets were washed with I mg/ml bovine serum albumin and 0.01% Triton X-100 and resuspended in 8M urea. Aliquots of each were counted by liquid scintillation spectrophotometry, and equal counts were loaded on SDS 7-17% polyacrylamide gels and treated as described under "Methods". (Note: The faintly visible band at M r 67,000 in the tubulin lanes is bovine serum albumin and not neurofilament protein.) Molecular weight standards are shown in Lane M of Panel A, and their corresponding values are given in Panel B.

observed that thrombin predominantly yielded MAP-2 fragments of M r 28,000 and

240,000 (vide infra). Indeed, phenylmethanesulfonyl fluoride (I.0 mM final

concentration) blocked any further degradation over a 5-7 day period at 4°C.

This development has allowed us to probe with much greater ease the

interactions of MAP-2 fragments with neurofilaments and microtubules. The basic

approach is to determine which proteins or protein fragments co-sediment with

assembled microtubules or neurofilaments using ultracentrifugation and subsequent

electrophoretic analysis. We first enzymatically phosphorylated MAP-2 as

described under "Experimental Procedures", and we digested the labeled

microtubule-associated protein with thrombin. Next, we incubated the thrombin

digested MAP-2 fragments (indicated by the plus sign) or undigested MAP-2

(indicated by the minus sign) with microtubules or neurofilaments under the

conditions shown in Fig. i. We then separated the polymerized and unpolymerized

eytoskeletal proteins into pellet [p] and supernatant [s] fractions by

ultracentrifugation. Because the projection and tubule-binding domains do not

contain identical phosphorylation sites, we applied a constant total amount of

radioactivity for each electrophoretic sample. Lanes 1-4 of Fig. IA and the

1455

Vol. 148, No. 3, 1987 BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS

(A) NF NF Triplet Low NoNF

M I++ --\I’++ --V++ - -\ M

(B) NF NF Triplet Low

\SP SPAS P SP/\SP SP/ 12 34 56 78 9 10 11 12

\s P s PAS P s P/\s P s P/ 1234 56 78 9 10 11 12

m. 2: MAp-2 Binding to Purified Neurofilament Triplet and the LOW- Molecular-WeiPht Subunit of Neurofilaments. Panel A, Protein Staining Pattern; Panel B, Autoradiogram of [32P]-labeled Proteins. Radiolabeled MAP-Z was incubated with thrombin or assembly buffer as described earlier, quenched with 1 mM phenylmethanesulfonyl fluoride and incubated with either 2.2 mg/ml neuro- filament triplet protein or 1.1 mg/ml low-molecular-weight subunit at 4°C for 10 min. Samples were pelleted and analyzed as described in Fig. 1.

corresponding lanes 1-4 of Fig. 1B demonstrate our new finding that only the M,

28,000 MAP-2 fragment binds to neurofilaments. The next four lanes demonstrate

that this thrombin fragment behaves as the so-called microtubule-binding domain

of MAP-2. In the absence of neurofilaments or microtubules (Lanes 9-12),

however, the M, 28,000 thrombin cleavage product remains in the supernatant

fraction even after ultracentrifugation. Indeed, the entire pellet fraction was

used for the electrophoretic analysis in Lanes 10 and 12, and virtually no high-

molecular-weight or fragmented MAP-2 co-sediments without neurofilaments present.

These observations verify that the fragment is only sedimentable as a result of

interactions with either neurofilaments or microtubules.

Next, we sought to determine the neurofilament protein(s) interacting with

MAP-2 or the M, 28,000 fragment. Earlier work by Miyata et al. (23) and Heimann

et al. (24) demonstrated MAP bind.ng to the low-molecular-weight neurofilament

component (i.e., M, 68,000 polypeptide). We conducted a second series of binding

assays (Fig. 2), and lanes 3, 4 and 7, 8 of Panel B verified that uncleaved MAP-2

binds to the neurofilament triplet and to the M, 68,000 component. The data

shown in lanes 1, 2 and 5, 6 extend the earlier observations by clearly

demonstrating the binding of the M, 28,000 fragment to the triplet and M, 68,000 neurofilament protein components. It should be noted that only trace levels of

tubulin are evident in the 50,000 to 55,000 molecular weight range in the gels

1456

Vol. 148, No. 3, 1987 BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS

LIJ

i

a ( 83

5

~2

,4

NEPHGE Fig. ~: Autoradiogram from Non-Equilibrium DH Gradient Electro-

phoresis/Sodium Dodecyl Sulfate Polyacrylamide Ge__~l Electrophoresis ("NEPHGE"- SD___SS PAGE). Radiolabeled MAP-2 was digested with 4 units/ml bovine thrombin, and the digestion was quenched by the addition of an equal volume of 9.3 M urea, 0.59 (w/v) dithiothreitol and 29 (v/v) Nonidet NP-40 in 5 mM potassium carbonate. Samples were loaded onto "NEPHGE" tube gels with I~ (v/v) Ampholine (pH range of 9-11) and 29 (v/v) Ampholine (pH range of 3.5-10), and run for 2000 Volt-hours. The arrow designates the M r 28,000 fragment. Protein molecular weight/iso- electric point reference standards were: i) phosphorylase b; 2) bovine serum albumin; 3) carbonic anhydrase; 4) soybean trypsin inhibitor; 5) ribonuclease; and 6) lysozyme. Ribonuclease (pl = 9.3) and lysozyme (pl = i0.5-ii.0) provided "NEPHGE" reference standards over the basic region of the gel. The second dimension was an SDS 7-179 polyacrylamide gel.

shown in Fig. 2. Much higher levels of tubulin are required for binding of MAP-2

to assembled tubules. This suggests that binding depends on presence of

neurofilaments and does not require tubulin or assembled microtubules for

binding.

Tubulin and neurofilament proteins contain negatively charged regions that

may be important in MAP binding (22). The data presented above demonstrate that

the M r 28,000 fragment binds to both cytoskeletal organelles. Accordingly, we

used "NEPHGE"-SDS PAGE to estimate the isoelectric point of this M r 28,000

protein species. This method yielded a pl value of about I0, indicating the

strongly basic nature of this fragment. As shown in Fig. 3, this radiolabeled

species migrates in the 28,000 molecular weight range (near carbonic anhydrase,

M r 30,000). The comigration of the trace-labeled fragment in both dimensions

also indicates that the phosphorylation conditions do not lead to significant

heterogeneity in overall ionic charge or molecular weight. Furthermore, the

dephosphorylated form will necessarily display an even higher isoelectric point.

DISCUSSION

The findings presented in this report indicate that neurofilaments, a

special class of intermediate filaments, interact with MAP-2 in the region

1457

VoI. 148, No. 3, 1987 BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS

corresponding to the M r 28,000 microtubule-binding fragment. In earlier work

with aetin, Sattilaro (25) reached a similar conclusion about the binding of MAP-

2 to polymerized actin. Likewise, this same sub-fragment of MAP-2 constitutes

the mierotubule-binding domain (10,13). Thus, all three major cytoskeletal

classes (~.e., microfilaments, intermediate filaments, and microtubules) interact

with a common structural region of MAP-2. In view of our estimated pI value of

I0 for the M r 28,000 component, all of these MAP interactions must be strongly

influenced by electrostatic charge. Vallee (26) has clearly demonstrated that

0.35 M sodium chloride can remove MAPs from taxol-stabilized microtubules, and

this observation is also in harmony with the notion of ionic interactions between

tubules and MAPs.

Our use of trace labeling of MAPs with [?-32p]ATP and protein kinase

reinforces the general utility of this method as first applied by Heimann e!t a l.

(24). The radiolabeled low-molecular-weight thrombin fragment of MAP-2 is more

stable than that obtained with trypsin or chymotrypsin, and this stability may

also facilitate studies of the stoichiometry and dissociation constants for

fragment binding to microtubules, neurofilaments, or actin. Other more

approximate methods using intact MAPs and either densitometry or pelleting of

assembled cytoskeletal elements (23,24) still require refinement and/or

verification.

Finally, the thrombin results underscore the facility and generality of

serine protease cleavage of MAP-2 into low- and high-molecular-weight fragments.

This behavior is reminiscent of proteolytic action on myosin. Furthermore,

intact MAP-2 has an isoelectric point of 5.4 (27), whereas we found that the

tubule/filament-binding domain has a value of I0. This suggests that there must

be significant acidic and basic charge localization in the high- and low-

molecular-weight fragments, respectively, of MAP-2. Nonetheless, the biological

significance of this protein design feature awaits further understanding of the

role of MAP-2 in the cytomatrix.

REFERENCES

I. Wuerker, R. B., and Palay, S. L. (1969) Tissue Cell i, 387-402. 2. Ellisman, M. H., and Porter, K. R. (1980) 2. Cell Biol., 87, 464-479. 3. Hirokawa, N. (1982) 2. Cell Biol., 94, 129-142. 4. Runge, M. S., Laue, T. M., Yphantis, D. A., Lifsics, M. R., Saito, A., Altin,

M., Reinke, K., and Williams, R. C., Jr. (1981) Proc. Natl. Acad. Sci. U.S.A., 78, 1431-1435.

5. Aamodt, E. J., and Williams, R. C., Jr. (1984) Biochemistry, 23, 6031-6035. 6. Minami, Y., and Sakai, H. (1983) 2. Biochem., 94, 2023-2033. 7. Leterrier, J. F., Liem, R. K. H., and Shelanski, M. L. (1982) ~. Cell Biol.,

95, 982-986. 8. Aamodt, E. J., and Williams, R. C., Jr. (1984) Biochemistry, 23, 6023-6031. 9. Flynn, G., and Purich, D. L. (1987) ~. Biol. Chem., in press.

i0. Kim, H., Binder, L. I., and Rosenbaum, J. L. (1979) ~. Cell Biol., 80, 266-276.

ii. Vallee, R. B. (1982) ~. Cell Biol., 92, 435-442.

1458

Vol. 148, No. 3, 1987 BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS

12. Purich, D. L., and Kristofferson, D. (1984) Adv. Prot. Chem., 36, 133-212. 13. Vallee, R. B., Borisy, G. G. (1977) 2. Biol. Chem., 252, 377-382. 14. Shelanski, M. L., Gaskin, F., Cantor, C. R. (1973) Proc. Natl. Acad. Sei.

U.S.A., 70, 765-768. 15. Delacourte, A., Filliatreau, G., Boutteau, F., Biserte, G., Schrevel, J.

(1980) Biochem. 2., 191, 543-546. 16. Tokutake, S., Hutchison, S. B., Pachter, J. S., Liem, R. K. H. (1983) Anal.

Biochem., 135, 102-105. 17. Geisler, N., Weber, K. (1981) 2. Mol. Biol., 151, 565-571. 18. Herzog, W., Weber, K. (1978) Eur. 2. Biochem., 92, 1-8. 19. Weingarten, M. D., Lockwood, A. H., Hwo, S., and Kirshner, M. W. (1975) Proc.

Natl. Acad. Sci. U.S.A. 72, 1858-1862. 20. Laemmli, U. K. (1970) Nature, 227, 680-685. 21. Roberts, R. M., Baumbach, G. A., Buhi, W. C., Denny, J. B., Fitzgerald, L. A.,

Bebelyn, S. F., Horst, M. N. (1984) "Molecular and Chemical Characterizaiton of Membrane Receptors" (J. C. Venter, and L. C. Harrison, eds.), pp. 61-113.

22. Olmsted, J. B. (1986) Ann. Rev. Cell Biol., ~, 421-457. 23. Miyata, Y., Hoshi, M., Nishida, E., Minami, Y., Sakai, H. (1986) 2. Biol.

Chem., 261, 13026-13030. 24. Heimann, R., Shelanski, M. L., Liem, R. K. H. (1985) 2. Biol. Chem., 260,

12160-12166. 25. Sattilaro, R. F. (1986) Biochemistry, 25, 2003-2009. 26. Vallee, R. B. (1982) 2- Cell Biol., 92, 435-442. 27. Berkowitz, S. A., Katagiri, J., Binder, H. K., Williams, R. C., Jr. (1977)

Biochemistry, 16, 5610-5617.

1459