Eur. J. Biochem. 241, 765-771 (1996) 0 FEBS 1996

Phosphorylation and dephosphorylation in the proline-rich C-terminal domain of microtubule-associated protein 2 Carlos SANCHEZ’, Peter TOMPA2, KornClia SZUCS3, Peter FRIEDRICH’ and Jes6s AVILA’

I Centro de Biologia Molecular ‘Severo Ochoa’, Facultad de Ciencias, Universidad Aut6noma de Madrid, Spain * Institute of Enzymology, Biological Research Center, Hungarian Academy of Sciences, Hungary

Department of Medical Chemistry, University School of Medicine, Debrecen, Hungary

(Received 18 June 1996) - EJB 96 0894/3

The C-terminal domain of microtubule-associated protein 2 (MAP2) contains a proline-rich region and the tubulin-binding domain. We have generated antibodies to follow the phosphorylation state of the proline-rich domain. One of these antibodies (no. 305) has been raised against a synthetic peptide P (sequence RTPGTPGTPSY) phosphorylated at the threonine residues. This sequence is present in the proline-rich region of MAP2 and is phosphorylated in vitro by at least three different proline-directed protein kinases : p3PdC2, and GSK3 (glycogen-synthase kinase 3) alp. The MAP2 sites phosphory- lated by these kinases are different, although all of them phosphorylate the C-terminal domain of MAP2 as determined by Staphylococcus aureus V8 protease mapping. Nonphosphorylated peptide P can be phosphorylated in vitro by all three kinases studied with similar efficiency. In high-molecular-mass MAP2, this sequence is highly phosphorylated in vivo at the late stages of rat development. This motif can be rapidly dephosphorylated in vitro by protein-phosphatase 1 (PPl) and 2A (PP2A) catalytic subunits but not by PP2B.

Keywords : microtubule ; microtubule-associated protein 2 ; proline-directed protein kinase ; serinelthreo- nine protein phosphatase.

Neuronal microtubules are composed of the core protein tubulin and several microtubule-associated proteins. One of the most abundant microtubule-associated proteins within brain is microtubule-associated protein 2 (MAP2). This molecule has multiple isoforms generated from a single gene through alterna- tive RNA splicing. MAP2 isoforms are divided into high-molec- ular-mass MAP2 proteins, which include MAP2A (consisting of 1912 amino acids in rat, with an apparent molecular mass of 280 kDa) [I] and MAP2B (1830 amino acids and an apparent molecular mass of 270kDa) [2], and low- molecular-mass MAP2, MAP2C, and MAP2D (consisting of 467 and 498 amino acids, respectively, with apparent molecular masses in the range 70-75 kDa) [3,4]. Low-molecular-mass MAP2 contains the N- and C-terminal domains of high-molecular-mass MAP2 linked together, and lacks the middle intervening sequence. Recently, additional isoforms have been found [5, 61.

The location and developmental expression of these isoforms are different. MAP2B and MAP2C are expressed during early fetal development. MAP2C is down-regulated postnatally with a concomitant increase in the expression of MAP2A [7, 81. MAP2D expression increases at a later stage of development [3]. Whereas MAP2C is present in neuronal cell bodies, den- drites and axons, as well as in glial cells [9, 101, high-molecular-

Correspondence to C. Shnchez Martin, Cento de Biologia Molecular ‘Severo Ochoa’, Facultad de Ciencias, Universidad Aut6noma de Ma- drid, Cantoblanco, E-28049, Madrid, Spain

Fax: +34 1 3974799. Abbreviations. MAP, microtubule-associated protein ; MAPK, mito-

gen-activated-protein kinase; GSK3, glycogen-synthase kinase 3 ; PPI., catalytic subunit of protein phosphatase 1 ; PP2A,, catalytic subunit of protein phosphatase 2A; PP2B, protein phosphatase 2B ; PhMeSO’F, phenylmethylsulfonyl fluoride; CaM, calmodulin.

mass MAP2 is selectively localized in dendrites and neuronal cell bodies [ll-131. MAP2 has also been found localized with actin microfilaments and in the postsynaptic density in dendritic spines [14, 151.

MAP2 can be phosphorylated by several protein kinases [16-201 and dephosphorylated by several protein phosphatases [21, 221. The phosphorylation state of MAP2 may affect its binding to tubulin, both the rate and extent of microtubule poly- merization, and the motility driven by microtubule motors [19, 20, 23-26]. It can also control the association of MAP2 with actin microfilaments [27, 281 and the susceptibility of MAP2 to calpain attack [29, 301. Furthermore, correlations have been found between the phosphorylation state of MAP2 and several forms of neuronal plasticity [31-351.

MAP2 isoforms consist of two main regions: a C-terminal region, which contains the tubulin-binding motifs and a proline- rich region, and an N-terminal region, which contains the bind- ing site for the regulatory subunit of protein kinase A; in high- molecular-mass MAP2 there is also a large sequence connecting these two regions (Fig. 1). Recent work has shown that the bind- ing of MAP2 to tubulin and microtubule bundling are regulated not only by the tubulin-binding motifs, but also by flanking re- gions, such as the proline-rich regions [36]. This has also been demonstrated in the related z protein, which contains similar sequences [37].

Our aim has been to identify MAP2 sequences whose phos- phorylation state regulates MAP2 function in vivo. Recently, a MAP2 sequence has been found in the proline-rich region of the C-terminal domain which is phosphorylated in vivo in rat brain [38]. In the present work, we characterize the protein kinases and protein phosphatases that are likely to be involved in con- trolling the phosphorylation state of this sequence by the aid of

766 Sinchez et al. (ELK J. Biochem. 241)

an antibody which recognizes this epitope in its phosphorylated staie.

MATERIALS AND METHODS

Proteins and peptides. Brains were obtained from cows or Wistar rats of different ages (postnatal day 1, postnatal day 2 or adult rats of 3-4 months). To prepare MAP2, brains were homogenized in 0.1 M Mes, pH 6.5, 1 mM MgCL 2 mM ECTA, 0.1 M NaCI, I niM phenylmethylsulfonyl fluoride (PhMeSO,F), 10 pg/ml pepstatin, 10 pg/ml leupeptin, and 10 pg/ml aprotinin (in some cases, homogenization was per- formed in the presence of 5 pM okadaic acid, 50 mM NaF and 100 pM orthovanadate as phosphatase inhibitors). Homogenates were centrifuged at lOOOOOXg for 1 h at 20°C. The pellet was discarded and the supernatant (cytosolic fraction) was collected. The cytosol was brought to 1 M NaCl and 10 mM DL-dithio- threitol, boiled for 10 min and centrifuged at 100000Xg for 30 min at 2°C. Heat-re nt proteins in the resulting superna- tant were applied to a Sepharose CL-4B (Pharmacia) column (40 ml) equilibrated with 10 mM Pipes, pH 6.9, 0.2 M NaCl, and 1 mM DL-dithiothreitol. High-molecular-mass MAP2 was eluted as described by Herzog and Weber [39].

Microtubular proteins were prepared according to the pro- cedure of Karr et at. [401 from brain cytosol. The peptides RTPGTPGTPSY and SPQLATLAEDV (which comprise resi- dues 1616-1626 and 1809-1819, respectively, of MAP2 ac- cording to the sequence published by Lewis et al. [2]) (Fig. I ) were synthesized in their phosphorylated (RTPGT*PGT*PSY) and dephosphorylated forms i n an automatic solid-phase peptide synthesizer (type A430A, Applied Biosystems) and purified by reverse-phase HPLC on a Nova Pak C,, column (Waters).

Purified p42""" (ERK2) and p34"' (cdc2) were purchased from Upstate Biotechnology (catalogue no. 14-173 and 14-103, respectively. Enriched glycogen-synthase kinase 3 a/p (GSK3 dll) was prepared as described by Moreno et al. 1411.

The catalytic subunits of protein phosphatase 1 (PPI,) and 2A (PP2AJ as well as calcineurin (protein phosphatase 2B) were purified as described earlier [42]. Specific activities of the enzymes were 2350 U/mg for PPl,, 67 U/mg for PP2A,, and I S U/mg for PP2B. They were stored i n 60% glycerol at -20°C. PP2A, was dialysed against 50 mM Tris/HCl, pH 7.5, 0.1 mM EDTA, 10 mM uL-dithiothreitol and 1 mM PhMeSOzF prior to use.

Antibodies. Polyclonal antibodies 305, 972, and 291 were roised in rabbit against the synthetic peptides indicated above. Peptide coupling and immunization were performed as described by Ulloa et al. [42]. Immunoglobulins were purified from sera by ammonium sulfate precipitation.

Polyclonal antibody 30 was raised in mice iminunized with purified rat brain MAP2. It recognizes mainly high-molecular- mass MAP2 in an immunoblot assay.

Phosphorylation and dephosphorylation assays. The phosphorylation of purified MAP2, at 0.2-1 pM, with ERK2 (330 U/mg) and cdc2 (100 U/mg) was performed for 20 min or 4 h at 37°C in 20 mM Hepes, pH 7.4, 2 mM EGTA, 10 mM MgCI,, 1 mM PhMeSOZF and 1 mM ~~-dithiothreitol. Phos- phorylation by GSK3 (0.15 U/mg) was carried out in 25 mM Tris/HCl, pH 7.5, 1 mM EDTA, 1 mM EGTA, 10 mM MgCI, and 2 mM DL-dithiothreitol. The mixtures always contained 10 p g h l pepstatin. 10 pg/ml leupeptin, 10 pg/ml aprotinin, 5 pM okadaic acid, SO mM NaF, 100 pM orthovanadate and 50 pM/20 pM [1!-"P]ATP or 5 mM ATP (Amersham). Phosphor- ylation was stopped by the addition of SDS sample buffer (20 mM Tris/HCI, pH 6.8, 1% SDS, 10 mM DL-dithiothreitol,

5% glycerol, 0.05% Bromophenol Blue). Protein phosphoryla- tion was analyzed by SDS/PACE. Quantification was performed using an Image Plate autoradiography system (BAS) (Fuji Photo Film Co.) and data processing was performed by TINA software (raytest IsotopenmeBgerate GmbH).

The phosphorylation of peptide P (with the sequence RTPGTPGTPSY) was carried out at concentrations in the range 0.2 - 10 mM in the same buffer used for phosphorylating MAP2. Phosphorylation was stopped by boiling the samples for 2 min in the presence of 1 % trifluoroacetic acid. Phosphorylated pep- tide was isolated by reverse-phase HPLC on a Nova Pak C,, column (Waters) and the radioactivity associated with the eluted peptide was determined by Cerenkov radiation counting. In all cases, the K,,, values represent the average of three independent experiments 2 SD.

Dephosphorylation of MAP2 was performed at a final con- centration of 0.1 mg/ml. High-molecular-mass MAP2 was incu- bated at 30°C for 3 min in 50 mM Tris/HCI, pH 7.5, 0.1 mM EDTA and 10mM DL-dithiothreitol for PPI, and PP2A, treat- ment or in 50 mM TrislHCl, pH 7.0, 0.5 mM DL-dithiothreitol, 0.2 mM CaCI,, 10 pM MnC1, and 12 pM CaM (calmodulin) for PP2B. Dephosphorylation reactions were started by the addition of the enzyme at a final concentration of 28 U/ml (PPIJ, 8 U/ ml (PPZA,), or 4 U/ml (PP2B). All mixtures contained 10 pg/ml of aprotinin, leupeptin, and pepstatin and 1 mM PhMeS0,F as protease inhibitors. Aliquots of the reaction mixtures were with- drawn at 3, 8, and 20 min and the reaction was terminated by the addition of SDS sample buffer (with 40 mM EDTA for PP2B) and 5 min boiling.

Limited proteolysis of MAP2. After in vitro phosphoryla- tion by the kinases, MAP2 was purified by SDS/PAGE and par- tially digested with S. aureus V8 protease as previously de- scribed [32]. Phosphopeptides were analyzed by gel electropho- resis and autoradiography of dried gels onto Kodak X-Omat films. Alternatively, peptides derived from digested MAP2 were analyzed by western blotting.

Gel electrophoresis. SDSPACE was carried out according to the procedure of Laemmli [43] in 5 5% or 8 % polyacrylamide gels. Gels were stained with Coomassie brilliant blue. Prestained molecular-mass markers were purchased from BRL (reference no. 560-6040).

Western blot analysis was performed according to Towbin et al. [44]. To test for the antibody reaction, the ECL Amersham kit was used. Autoradiograms were quantified by densitometry performed in a Molecular Dynamics densitometer model 300 A equipped with Image Quant software version 3.0.

RESULTS

It has been previously shown that up to 46 sites can be phos- phorylated in MAP2 [25]. In this study, we concentrated on the phosphorylation of a specific MAP2 sequence, which showed characteristic changes of phosphorylation in vivo 1381. To this end, we developed several antibodies as markers (Fig. I ) . Rabbit polyclonal antibody 291 recognizes this sequence in a phosphor- ylation-independent manner, while antibodies 972 and 305 are phosphorylation dependent. Mouse polyclonal antibody 30 (not indicated in Fig. 1) was raised against the whole MAP2 and it mainly recognizes high-molecular-mass MAP2 independent of its state of phosphorylation.

I n vitro phosphorylation of MAP2 by proline-directed pro- tein kinase. We studied MAP2 phosphorylation in vitro at the sequence RTPGTPGTPSY by three kinases : p42"'pk (ERK2, MAPK), ~ 3 4 ' " ~ (cdc2), and GSK3 d f 2 . Western blot\ of phoq-

Sinchez et al. (Eur .I. Biochem. 241)

MAP2

767

1 6 0 1 5 1 2

r, /

1 ‘\ / 1828

I COOH

Ab 305 (*) Ab 972

Ab 291

Fig. 1. Location of the peptides used to generate antibodies for the MAP2 sequence. The mouse MAP2 sequence published by Lewis et al. [l] is shown. The tubulin-binding domain (TBD) and sequence 160- 1512, which is only present in high-molecular-mass MAP2 isoforms, are indicated. The sequences against which antibodies have been raised are highlighted. Antibody 291 (Ab 291) was generated against the C-termi- nal nonphosphorylated peptide 1809- 1819. Antibody 972 (Ab 972) rec- ognizes the nonphosphorylated sequence 1616- 1626, which is part of the proline-rich region, just before the tubulin-binding domain. Antibody 305 (Ab 305) recognizes the same sequence but only when it is phos- phorylated at the residues indicated (*).

Fig. 2. In vitro phosphorylation of MAP2. High-molecular-mass MAP2 was phosphoylated by p42*’pk (MAPK), p34”“ (CDC2), and GSK3 n//Y (GSK3) as described in the Materials and Methods section for 6 h in the presence of 5 mM ATP. The reaction was stopped by the addition of SDS sample electrophoresis buffer. Western blots were performed with samples incubated under the same conditions in the absence (-) or pres- ence (+) of kinases. Antibody 30 recognizes high-molecular-mass MAP2 in a phosphorylation-independent manner.

phorylated MAP2 developed with antibodies 30, 305, and 972 are presented i n Fig. 2. The results with antibody 305 show that this MAP2 sequence is phosphorylated by all three kinases. The same phosphorylation pattern was found with antibody 972, ex- cept for cdc2, where the results are less clear.

The Kn, values of MAP2 phosphorylation by MAPK and GSK3, 2.4 pM 2 0.5 and 2.8 pM 2 0.7, respectively, were very similar. Protein kinase cdc2 was also found to phosphorylate MAP2 in the same concentration range, around 6 pM.

MAP2 regions phosphorylated by proline-directed protein kinases in vitro. We compared the MAP2 regions phosphory- lated by each kinase. MAP2 phosphorylated in vitro in the pres- ence of [y-’’PIATP was treated with V8 protease and the pep-

Fig. 3. MAP2 regions phosphorylated in vitro. High-molecular-mass MAP2 was phosphorylated in v i m by ~42””’~ (MAPK), ~34‘“’’ (CDC2) and GSK3 n/p (CSK3) for 20 min in the presence of 50 pM ATP [y-”P] ATP. The reaction was stopped by addition of SDS sample electrophore- sis buffer. Samples were treated with S. aureus V8 protease as described in the Materials and Methods section. The autoradiography of the dried gels is shown here. Peptides located at the C-terminal domain of MAP2 are indicated with a horizontal line. Prestained molecular-mass markers are indicated on the left.

Fig. 4. MAP2 sequence recognized by antibody 305 is phosphory- lated in vivo during rat brain development. (A) Purified high-molecu- lar-mass MAP2 was phosphorylated in v i m as in Fig. 2. The samples were subjected to proteolysis with V8 protease. The blots of the samples incubated in the presence (+) or absence (-) of ~ 4 2 ’ ” ~ ~ (MAPK), p34‘“’’ (CDC2), and GSK3 cxlp (GSK3) were developed with antibodies 291 (Ab 291) and 305 (Ab 305). (B) High-molecular-mass MAP2 was puri- fied from rat brains at two developmental stages, postnatal day 1 (Pl) and postnatal day 21 (P21), in the absence (-) or presence (+) of phos- phatase inhibitors (PI) (50 mM NaF, 100 pM orthovanadate and 5 pM okadaic acid). The protein was subjected to proteolysis with VX protease and western blotted with antibodies 291 and 305. Prestained molecular- mass markers are indicated on the left.

tides generated were analyzed by SDS/PAGE. Different peptide patterns were obtained (Fig. 3). Nevertheless, all three kinases phosphorylated two peptides of 54 kDa and 42 kDa, which are located at the C-terminal domain of MAP2 1381.

Peptide P, which corresponds to the MAP2 sequence RTPGTPGTPSY, was also phosphorylated in vitro by MAPK, cdc2, and GSK3. The K,,, values were calculated for MAPK, 2 * 0.5 niM, for GSK3, 3 t 1.2 mM, and for cdc2, 0.6 mM.

768 Sknchez et al. (Eur. J. Biochem. 241)

PPl PPZA PPZB

'"1 '"1

Time (min)

Fig. 5. In vitro dephosphorylation of high-molecular-mass MAP2 by protein phosphatases at the sequence recognized by antibody 305. High-molecular-mass MAP2 was dephosphorylated in vitro by the cata- lytic subunits of protein phosphatase 1 (PPl) and protein phosphatase 2A (PP2A) as well as protein phosphatase 2B (PP2B), as described in the Materials and Methods section. Western blots were developed with antibodies 305 (0) and 30 (0). The relative values in the figure repre- sent the quantitative evaluation of autoradiograms obtained after devel- oping with the ECL Amersham kit. 100% corresponds to the density measured with samples without phosphatases. Averages 2 SD of three experiments are shown.

The sequence recognized by antibody 305 is also phosphory- lated in vivo in rat brain. Western blots of MAP2 phosphory- lated in v i m by each kinase and treated with V8 protease were visualized with antibodies 291 and 305 (Fig. 4A). Two peptides of 54 kDa and 42 kDa, which are possibly located at the C-ter- minal domain of MAP2 [38], are phosphorylated at the sequence recognized by antibody 305 in all cases.

To test if high-molecular-mass MAP2 was phosphorylated at the same sites in vivo, the protein was purified from brains of rats at two developmental stages (postnatal day 1 and day 21), in the presence or absence of phosphatase inhibitors. Samples were treated with V8 protease and western blotted with antibod- ies 291 and 305. As shown in Fig. 4B, the sequence recognized by antibody 305 is phosphorylated to a much higher degree in high-molecular-mass MAP2 of 21-day-old rat brains than in 1-day-old rats. This result correlates well with data obtained with antibody 972 1381.

In vitro dephosphorylation of the MAP2 sequence RTPGTPGTP by serinelthreonine protein phosphatases. We examined which phosphatases could dephosphorylate MAP2 in vitro at the site recognized by antibody 305. Both PPI, and PP2A, are able to dephosphorylate this sequence, although PP1. seems to be more efficient (Fig. 5) . We observed no dephosphor- ylation by PP2B treatment.

DISCUSSION

A MAP2 sequence that is phosphorylated in vivo i n rat brain has been previously described [38]. This sequence is recognized by antibodies 972 and 305 and is located in the proline-rich region of its C-terminal domain. It has been observed that low- molecular-mass MAP2 is highly phosphorylated at this sequence during early rat brain development, whereas it is essentially de- phosphorylated at this site by 21 days after birth. In contrast, high-molecular-mass MAP2 seems to be more phosphorylated at this site in the adult rat brain. In this work, we characterized the phosphorylation of this MAP2 sequence by proline-directed protein kinases MAPK, cdc2, and GSK3, and dephosphorylation by phosphatases PPI,, PP2A, and PP2B.

MAPKERK2 has a high affinity for microtubules [45] and MAP2 [46], localizes with MAP2 in dendrites [47], and might

be involved in synaptic potentiation [48]. This kinase could be a good candidate for regulating MAP2 phosphorylation. In con- trast, GSK3, which is very abundant in rat brain [49], phosphor- ylates the microtubule-associated protein T in a sequence similar to that recognized by antibody 305 in MAP2, and this phosphor- ylation is akin to that found in paired helical filament z [41, 50, 511. Thus, GSK3 may also phosphorylate MAP2. Recently, MAP2 was shown to be a substrate of cdc2 and was thus impli- cated in the morphological changes observed during the cell cy- cle [52]. A neuronal cdc2-like kinase, cdk5, phosphorylates se- quences that are suitable substrates for cdc2 and it phosphor- ylates z at sites shown to be phosphorylated in Alzheimer's dis- ease [53].

The expression of the kinases studied is differentially regu- lated during development. Boulton et al. [54] observed that ERKI and ERK2 expression was increased during rat brain de- velopment. However, Takahashi et al. [55] have recently shown that while GSK3 Q was detected at all rat brain developmental stages studied, GSK3 p expression decreased after postnatal day 8. The expression of cdk5 increases during development, being more abundant in postmitotic neurons at later develop- mental stages. Conversely, cdc2 is highly expressed in proliferat- ing cells, as a regulator of the cell cycle [56). While ERK and GSK3 can be found in cell bodies and dendrites where high- molecular-mass MAP2 is present, cdk5 is exclusively localized in axons. The action of cdk5 may be related to axonal growth, regulating axonal proteins such as z and MAPlB but not den- dritic proteins such as MAP2.

We developed antibody 305 against the phosphorylated MAP2 sequence RTPGTPGTPSY, which is recognized by anti- body 972 in its dephosphorylated stage [38]. Antibody 305 was more specific than antibody 972 in differentiating between the phosphorylated and dephosphorylated sequence. Another phos- phorylation-dependent antibody specific against MAP2 (API 8) has been characterized before [57]. We used these antibodies to monitor MAP2 phosphorylation by MAPK, GSK3, and cdc2 in vitro. While antibody 305 detected differences between phos- phorylated and dephosphorylated MAP2 in all three cases, anti- body 972 revealed no difference when MAP2 was phosphory- lated by cdc2. Possibly, this epitope is phosphorylated with a lower efficiency by cdc2 than by the other two kinases, leaving a larger population of MAP2 in its dephosphorylated form. A partial phosphorylation of MAP2 by cdc2 (only at one of the two threonine residues) could also explain the remaining immu- noreactivity of the antibody 972, or maybe changes in MAP2 conformation which would not allow an optimal phosphoryla- tion or accessibility for cdc2. The K,, obtained for the in vitro phosphorylation of MAP2 by MAPK is very similar to the one obtained for GSK3, and is in the range found by others [16]. GSK3 phosphorylates MAP2 more efficiently than proteins like myelin basic protein or inhibitor 2, with K,,, values of 59 pM and 16pM, respectively [58], although in these cases only the GSK3P isoform was used.

Previous results in our laboratory suggested that two pep- tides generated after digestion of MAP2 with V8 protease, with molecular masses of 42 kDa and 54 kDa, respectively, are lo- cated at the C-terminal domain of MAP2, including the tubulin- binding domain and proline-rich region [38]. This has been cor- roborated in the present work by the use of antibodies 291 and 305, which recognize this MAP2 domain. ERK2, cdc2, and GSK3 apparently phosphorylate different sites of MAP2 as shown by V 8 protease peptide patterns, but all three kinases phosphorylate the C-terminal domain of MAP2.

To test if the sequence recognized by the antibody 305 was a good substrate for the kinases, the synthetic peptide P was phosphorylated in v i fm . The K,, for the phosphorylation of this

Sinchez et al. (EUK J. Biochem. 241) 769

peptide by MAPK has been previously determined [38], and is very similar to those obtained for other synthetic peptides [59]. In contrast, previous work with GSK3 yielded K, values of 200 pM for a CAMP-responsive-element-binding protein (CREB) peptide [60] and 5 pM for the peptide GS-1 [61]. It should be mentioned, however, that with the latter peptides prephosphorylation with CAMP-dependent protein kinase or casein kinase 11, respectively, was necessary to obtain such low K, values.

The phosphorylation state of the antibody 305 epitope in high-molecular-mass MAP2 is increased during rat brain devel- opment, as it was also previously shown [38]. A similar result has been recently found for the MAP2 sequence recognized by the antibody AP18 during development of the visual cortex and cerebellum of the cat [62]. Low-molecular-mass MAP2 may be good substrate for GSK3 a and/or p at early developmental stages while high-molecular-mass MAP2 is a good substrate for ERK and/or GSK3 a at later developmental stages.

The approach used to study the dephosphorylation of the antibody 305 epitope, has already been applied for other MAP proteins. It has been previously shown that microtubule-associ- ated protein 1B (MAPlB) was dephosphorylated in vitro by PP2A and PP2B at sites phosphorylated by proline-directed pro- tein kinases [42]. In contrast, the microtubule-associated protein T could only be dephosphorylated by PP2A at the antibody z-1 site, which is similar to that recognized by antibody 305 [63]. It is noteworthy that in both of these cases, PP1, was ineffective in dephosphorylating consensus sequences for proline-directed protein kinases. In contrast, rapid dephosphorylation occurred with PP1, and PP2A, of the MAP2 sequence recognized by anti- body 305, but not with PP2B. This may suggest that a different conformation of proline-rich sequences in MAP proteins could result in a different susceptibility to these phosphatases. Several isoforms of PP1 and PP2A are highly concentrated in dendrites and dendritic spines [64, 651 and have an important role in syn- aptic transmission and potentiation [66-681. This may well in- clude the regulation of MAP2 function in these events. Never- theless, dephosphorylation of MAP2 by PP2B at sites different from that recognized by antibody 305 cannot be excluded and may also take place [30].

The in vivo relevance of all our phosphorylation studies is underlined by several lines of evidence. We have demonstrated that the sequence under study is indeed phosphorylated in vivo in MAP2 during rat brain development. Either way, the impor- tance of these changes is emphasized by their coincidence with the critical period of brain development when a pronounced ex- pression pattern and consolidation of the cytoskeletal lattice oc- cur [S]. This conclusion is further supported by recent reports that have shown an increase of MAP2 phosphorylation after treatment of different cultured neuronal cells with okadaic acid [69, 701. Okadaic acid is a specific inhibitor of PP1 and PP2A, and it can also activate MAPK. Furthermore, an increment has recently been observed in MAP2 phosphorylation in cultured rat cortical neurons and in a human neuroblastoma cell line after treatment with okadaic acid [71]. The increase in MAP2 phos- phorylation correlated with changes in the cytoskeleton and led to cell death. In all those cases, the increment in MAP2 phos- phorylation could be due to the inhibition of PP1 and PP2A and/ or activation of MAPK. Finally, preliminary results (Sgnchez, C., unpublished results) indicate an increase in MAP2 phosphor- ylation at the sequence recognized by the antibody 305 after treatment with okadaic acid in primary cultures of granule cells from cerebellum.

All these results suggest that the phosphorylation state of MAP2 can be influenced by ERK and/or GSK3 (although one cannot rule out a possible function of cyclin-dependent protein

kinases) and phosphatases PP1 and/or PP2A in neurons, at the sequence studied here. Alteration of the microtubular cytoskele- ton could in turn modify the arborization of neurons and the activity of synaptic contacts. The modification of the neuron dendritic structure could be responsible for changes in synaptic transmission and hence in synaptic plasticity.

The work of Carlos Sinchez was supported by a fellowship of Com- unidad de Madrid, by the Comisibn Interministerial de Cienciu y Tecno- logiu Exparlola (CICYT), and by an institutional grant of Funducidn Rumdn Areces. The work with phosphatases was supported by grants T12840, T6005, T6305, T17633, and F5363 from the Hungarian National Research Fund (OTKA). Peter Tompa was supported to stay in Madrid by an agreement between the Consejo Superior de Investigu- ciones CientGcas and the Hungarian Academy of Sciences. The authors are grateful to Drs Pal Gergely and Viktor Dombradi for providing phos- phatases and for helpful discussions.

REFERENCES 1. Chung, W., Kindler, S., Seidenbecher, C. & Garner, C. (1996)

MAP2a, an alternative spliced variant of microtubule-associated protein 2, J. Neurochem. 66, 1273-1281.

2. Lewis, S. A., Wang, D. & Cowan, N. J. (1988) Microtubule-associ- ated protein MAP2 shares a microtubule-binding motif with T protein, Science 242, 936-939.

3. Doll, T., Meichsner, M., Riederer, B. M., Honegger, P. & Matus, A. (1 993) An isoform of microtubule-associated protein 2 (MAP2) containing four repeats of the tubulin-binding motif, J. Cell Sci. 106, 633-640.

4. Kindler, S., Schulz, B., Goedert, M. & Garner, C. C. (1990) Molecu- lar structure of microtubule-associated protein 2b and protein 2c from rat brain, J. Biol. Chem. 265, 19679-19684.

5. Kalcheva, N., Albala, J., O’Guin, K., Rubino, H., Gamer, C. & Shafit-Zagardo, B. (1 995) Genomic structure of human microtu- bule-associated protein 2 (MAP2) and characterization of addi- tional MAP-2 isoforms, Proc. Natl Acad. Sci. USA 92, 10894- 10898.

6. Langkopf, A., Guilleminot, J. & Nunez, J. (1994) Two novel HMW MAP2 variants with four microtubule-binding repeats and dif- ferent projections domains, FEBS Lett. 354, 259-262.

7. Binder, L., Frankfurter, A,, Kin, H., Caceres, A,, Payne, M. R. & Rebhun, L. L. (1984) Heterogeneity of microtubule-associated protein 2 during rat brain development, Proc. Nutl Acad. Sci. USA 81, 5613-5617.

8. Riederer, B. & Matus, A. (1985) Differential expression of distinct microtubule-associated proteins during brain development, Proc. Natl Acad. Sci. USA 82, 6006-6009.

9. Tucker, R. P. (1990) The role of microtubule-associated proteins in brain morphogenesis: a review, Bruin Res. Rev. 15, 101 - 120.

10. Albala, J. S., Kress, Y., Liu, W., Weidenheim, K., Yen, S. C . & Shafit-Zagardo, B. (1995) Human microtubule-associated protein- 2c localizes to dendrites and axons in fetal spinal motor neurons, J . Neurochem. 64, 2480-2490.

1. De Camilli, P., Miller, P. E., Navone, F., Theurkauf, W. E. & Vallee, R. B. (1984) Distribution of microtubule-associated protein 2 in the nervous system of the rat studied by immunofluorescence, Neuroscience 11, 817-846.

2. Huber, G. & Matus, A. (1984) Differences in the cellular distribu- tions of two microtubule-associated proteins, MAP1 and MAP2, in rat brain, J. Neurosci. 4, 151-160.

3. Caceres, A,, Binder, L. I., Payne, M. R., Bender, P., Rebhun, L. & Stewart, 0. (1984) Differential subcellular localization of tubulin and microtubule-associated protein MAP2 in brain tissue as re- vealed by immunocytochemistry with monoclonal hybridoma an- tibodies, J. Neurosci. 4, 394-410.

14. Ciceres, A., Payne, M. R., Binder, L. I. & Stewart, 0. (1983) Immunocytochemical localization of actin and microtubule-asso- ciated protein MAP2 in dendritic spines, Proc. Nutl Acud. Sci.

15. Fifkova, E. & Morales, M. (1992) Actin matrix of dendritic spines, synaptic plasticity, and long-term potentiation, lnt. Rev. Cyml.

USA 80, 1738-1742.

139,261-307.

770 Sanchez et al. (Bur: J . Biochem. 241)

16. Hoshi, M., Ohta, K., Gotoh, Y., Mori, A,, Murofushi, H., Sakai, H. & Nishida, E. (1992) Mitogen-activated-protein-kinase-cata- lyzed phosphorylation of microtubule-associated proteins, micro- tubule-associated protein 2 and microtubule-associated protein 4, induces an alteration in their function, Eur: J. Biochem. 203, 43- 52.

17. Silliman, C. C. & Sturgill, T. W. (1989) Phosphorylation of microtu- bule-associated protein 2 by MAP kinase primarily involves the projection domain, Biochem. Biophys. Res. Commun. 160, 993 - 998.

18. Goldenring, J., Vallano, M. & DeLorenzo, R. (1985) Phosphoryla- tion of microtubule-associated protein 2 at distinct sites by calmo- dulin-dependent and cyclic-AMP-dependent kinases, J . Neu- rochem. 45, 900-905.

19. Ainsztein, A. & Purich, D. (1994) Stimulation of tubulin polymer- ization by MAP2, J . Bid. Chem. 26Y, 28465-28471.

20. Ldpez, L. & Sheetz, M. (1995) A microtubule-associated protein (MAP2) kinase restores microtubule motility in embryonic brain, J. Bid. Chem. 270, 12511-12517.

21. Goto, S., Yamamoto, H., Fukunaga, K., Iwasa, T., Matsukado, Y. & Miyamoto, E. (1 985) Dephosphorylation of microtubule-associ- ated 2, r factor, and tubulin by calcineurin, J . Neurochent 4, 276- 283.

22. Yamamoto, H., Saitoh, Y., Fukunaga, K., Nishimura, H. & Miya- moto, E. (1 988 b) Dephosphorylation of microtubule-proteins by brain protein phosphatases 1 and 2A, and its effect on microtubule assembly, J . Neurochem. 50, 1614- 1623.

23. Jameson, L.. Frey, T., Zeeberg, B., Dalldorf, F. & Caplow, M. (1980) Inhibition of microtubule assembly by phosphorylation of micro- tubule-associated proteins, Biochemistry 19, 2472-2479.

24. Murthy, A. S. N. & Flavin, M. (1983) Microtubule assembly using microtubule-associated protein MAP2 prepared in different states of phosphorylation with protein kinase and phosphatase, Eul: J. Biochem. 137, 37-46.

25. Tsuyama, S., Terayama, Y. & Matsuyama, S. (1987) Numerous phosphates of microtubule-associated protein 2 in living rat brain, J . Bid . Chem. 262, 10886-10892.

26. Hoshi, M., Akiyama, T., Shinohara, Y., Hiyata, Y., Ogawara, H., Nishida, E. & Sakai, H. (1988) Protein kinase C-catalyzed phos- phorylation of the microtubule-binding domain of MAP2 inhibits its ability to induce tubulin polymerization, Euc J. Biochem. 174, 225-230.

27. Sattilaro, W. (1986) Interaction of microtubule-associated protein 2 with actin filaments, Biochemistry 25, 2003 -2009.

28. Selden, S. & Pollard, T. D. (1983) Phosphorylation of microtubule- associated protein regulates their interaction with actin filaments, J. Bid. Chem. 258, 7064-7071.

29. Johnson, G. V. W. & Foley, V. G. (1993) Calpain-mediated proteoly- sis of microtubule-associated protein 2 (MAP2) is inhibited by phosphorylation by CAMP-dependent protein kinase, but not by Ca"/calmodulin-dependent protein kinase 11, J . Neurosci. Res.

30. Alexa, A., Tompa, P., Baki, A,, Vereb, G. & Friedrich, P. (1996) Mutual protection of microtubule-associated protein 2 (MAP2) and cyclic AMP-dependent protein kinase I1 against N-calpain, J. Neurosci. Rex 44, 438-445.

31. Johnson, G. V. W. & Jope, R. S. (1992) The role of microtubule- associated protein 2 (MAP-2) in neuronal growth, plasticity, and degeneration, J . Neurosci. Rex 33, 505-512.

32. Diaz Nido, J., Montoro, R., L6pez Barneo, J. & Avila, J. (1993) High external potassium induces an increase in the phosphoryla- tion of the cytoskeletal protein MAP2 in rat hippocampal slices, Eur: J. Neurosci. 5, 81 8-824.

33. Montoro, R . J . , Diaz Nido, J. , Avila, J. & Ldpez Barneo, J. (1993) N-methyl-D-aspartate stimulates the dephosphorylation of the microtubule-associated protein 2 and potentiates excitatory synaptic pathways in the rat hippocampus, Neuroscience 54,

34. Reid, S. & Daw, N. (3995) Dark-rearing changes dendritic microtu- bule-associated protein 2 (MAP2) but not subplate neurons in cat visual cortex, J. Cotnp. Neurol. 359, 38-47.

35. Quinlan, E. & Halpain, S. (1996) Postsynaptic mechanisms for bidi- rectional control of MAP2 phosphorylation by glutamate recep- tors, Ncuron 16, 357-368.

34, 642-647.

859- 871.

36. Ferralli, J., Doll, T. & Matus, A. (1994) Sequence analysis of MAP2 function in living cells, J. Cell Sci. 107, 3115-3125.

37. Gustke, N., Trinczek, B., Biernat, J., Mandelkow, E.-M. & Mandel- kow, E. (1994) Domains of r protein and interactions with micro- tubules, Biochemistry 33, 951 1-9522.

38. Sanchez, C., Diaz-Nido, J. & Avila, J . (1995) Variations in in vivo phosphorylation at the proline-rich domain of the microtubule- associated protein 2 (MAP2) during rat brain developmen, Bio- chem. J . 306, 481 -487.

39. Herzog, W. & Weber, K. (1978) Fractionation of brain microtubule- associated proteins. Isolation of two different proteins which stimulate tubulin polymerization in v i m , Eur 1. Biochem. 92, 1-8.

40. Karr, T. L., White, H. D. & Purich, D. L. (1979) Characterization of brain microtubule protein prepared by selective removal of mitochondria1 and synaptosomas components, .I. Biol. Chem. 254,

41. Moreno, F. J., Medina, M., Ptrez, M., Montejo, E. & Avila, J. (1995) Glycogen synthase kinase 3 phosphorylates recombinant human r protein at serine-262 in the presence of heparin (or tubulin), FEBS Lett. 372, 65-68.

42. Ulloa, L., Dombridi, V., Diaz-Nido, J . , Sziics, K., Gergely, P., Friedrich, P. & Avila, J. (1993) Dephosphorylation of distinct sites on microtubule-associated protein MAPlB by protein phospha- tases 1, 2A and 2B, FEBS Lett. 330, 85-89.

43. Laemmli, U. K. (1970) Cleavage of structural protein during assem- bly of the head of bacteriophage T4, Nuture 227, 680-685.

44. Towbin, H., Staehelin, T. & Gordon, J. (1979) Electrophoretic transfer of protein from polyacrylamide gels to nitrocellulose sheets: procedure and some application, Proc. Not/ Acad. Sci.

45. Reszka, A,, Seger, R., Diltz, C., Krebs, E. & Fischer, E. (1995) Association of mitogen-activated protein kinase with the micro- tubule cytoskeleton, Proc. Natl Acad. Sci. USA Y2, 8881 -8885.

46. Thomas, K. L. & Hunt, S. P. (1993) The regional distribution of extracellularly regulated kinase-l and -2 messenger RNA in the adult rat central nervous system, Neuroscience 56, 741 -757.

47. Fiore, R. S., Bayer, V. E., Pelech, S. L. & Baraban, J. M. (1993) p42 mitogen-activated protein kinase in brain : prominent local- ization in neuronal cell bodies and dendrites, Neuroscience 55, 463 -412.

48. Suzuki, T. (1993) Protein kinases involved in the expression of long- term potentiation, lnt. J. Biochem. 26, 735-744.

49. Takahashi, M., Tomizawa, K., Kato, R., Sato, K., Uchida, T., Fujita, S. & Imahori, K. (1994) Localization and developmental changes of r protein kinase Vglycogen synthase kinase-3/1' in rat brain, J . Neurochem. 63, 245-255.

50. Hanger, D., Hughes, K., Woodgett, J., Brion, J. & Anderton, B. (1992) Glycogen synthase kinase-3 induces Alzheimer's disease- like phosphorylation of r : generation of paired helical filament epitopes and neuronal localisation of the kinase, Neurosci. Lett.

51. Ishiguro, K., Shiratsuchi, A,, Sato, S., Omori, A,, Arioka, M., Kobayashi, S., Uchida, T. & Imahori, K. (1993) Glycogen syn- thase kinase 38 is identical to 7 protein kinase I generating several epitopes of paired helical filaments, FEBS Lett. 325, 167- 172.

52. Fellous, A,, Kubelka, M, Thibier, C., Taieb, F., Haccard, 0. & Jessus, C. (1994) Association of ~ 3 4 ' " ' ~ kinase and MAP kinase with microtubules during the meiotic maturation of Xenopus oo- cytes, Int. J. Dev. Bid. 38, 651-659.

53. Lew, J. & Wang, J. (1995) Neuronal cdc2-like kinase, Trends Bio- chem. Sci. 20, 33-37.

54. Boulton, T., Nye, S., Robbins, D., Ip, N., Radziejewska, E., Morgenbesser, S., DePinho, R., Panayotatos, N., Cobb, M. & Yancopoulos, G. (1991) ERKs: a family of protein-serine/threo- nine kinases that are activated and tyrosine phosphorylated in re- sponse to insulin and NGF, Cell 65, 663-675.

55. Takahashi, M., Tomizawa, K., Kato, R., Sato, K., Uchida, T., Fujita, S. & Imaliori, K. (1994) Localization and developmental changes of r protein kinase I/glycogen synthase kinase-3a in rat brain, J. Neurochem. 63, 245 -255.

56. Tsai, L., Takahashi, T., Caviness, V. & Harlow, E. (1993) Activity and expression pattern of cyclin-dependent kinase 5 in the embry- onic mouse nervous system, Development 119, 1029- 1040.

6107-6111.

USA 76,4350-4354.

147, 58-62.

Sanchez et al. (Eur: J . Biochem. 241) 771

57. Berling, B., Wille, H., Roll, B., Mandelkow, E.-M., Gamer, C. & Mandelkow, E. (1 994) Phosphorylation of microtubule-associated proteins MAP2a,b and MAP2c at Serl36 by proline-directed kinases in vivo and in vitro, EUI: J. Cell Biol. 64, 120-130.

58. Wang, Q., Fiol, C., DePaoli-Roach, A. &Roach, P. (1994) Glycogen synthase kinase-3P is a dual specificity kinase differentially regu- lated by tyrosine and serinelthreonine phosphorylation, J. Bid. Chem. 269, 14566-14574.

59. Clark-Lewis, I., Sanghera, J. & Pelech, S . (1991) Definition of a consensus sequence for peptide substrate recognition by p44"'pk, the meiosis-activated myelin basic protein kinase, J. Biol. Chem.

60. Wang, Q., Roach, P. & Fiol, C. (1994) Use of synthetic peptide as a selective substrate for glycogen synthase kinase 3, Anul. Bio- chem. 220, 397-402.

61. Woodgett, J. (1989) Use of peptide substrates for affinity puri- fication of protein-serine kinases, Anal. Biochem. 180, 237- 241.

62. Riederer, B. M., Draberova, E., Viklicky, V. & Draber, P. (1995) Changes of MAP2 phosphorylation during brain development, 1. Histochem. Cytochem. 43, 1269- 1284.

63. Sziics, K., Ledesma, M. D., Dombridi, V., Gergely, P., Avila, J. & Friedrich, P. (1994) Dephosphorylation of T protein from Alz- heimer's disease patients, Neurosci. Lett. 165, 175- 178.

266, 15 180-15 184.

64. Hashikawa, T., Nakazawa, K., Mikawa, S . , Shima, H. & Nagao, M. (1995) Immunohistochemical localization of protein phosphatase isoforms in the rat cerebellum, Neurosci. Res. 22, 133-136.

65. Quimet, C., Da Cruz e Silva, E. & Greengard, P. (1995) The u and y l isofoms of protein phosphatase 1 are highly and specifically concentrated in dendritic spines, Proc. Nail Acad. Sci. USA 92,

66. Nairn, A. & Shenolikar, S. (1992) The role of protein phosphatases in synaptic transmission, plasticity and neuronal development, Cur% Opin. Neurobiol. 2, 296-301,

67. Malenka, R. (1994) Synaptic plasticity in the hippocampus: LTP and LTD, Cell 78, 535-538.

68. Ajima, A. & Ito, M. (1995) A unique role of protein phosphatases in cerebellar long-term depression, Neuroreport 6, 297-300.

69. Gotoh, Y., Nishida, E. & Sakai, H. (1990) Okadaic acid activates microtubule-associated protein kinase in quiescent fibroblastic cells, Eur: J. Biochem. 193, 671 -674.

70. Miyasaka, T., Miyasaka, J. & Saltiel, A. (1990) Okadaic acid stimu- lates the activity of microtubule-associated protein kinase in PC- 12 pheochromocytoma cells, Biochem. Biophys. Res. Commun.

71. Arias, C., Sharma, N., Davies, P. & Shafit-Zagardo, B. (1993) Oka- daic acid induces early changes in microtubule-associated protein 2 and T phosphorylation prior to neurodegeneration in cultured cortical neurons, J . Neurochem. 61, 673-682.

3396- 3400.

168, 1237-1243.