Recent experiments show that the microtubule-associated protein (MAP) 1B is a major phosphorylation substrate for the serine/threonine kinase glycogen synthase kinase-3β (GSK-3β) in differentiating neurons. GSK-3β phosphorylation of MAP1B appears to act as a molecular switch regulating the control that MAP1B exerts on microtubule dynamics in growing axons and growth cones. Maintaining a population of dynamically unstable microtubules in growth cones is important for axon growth and growth cone pathfinding. We have mapped two GSK-3β phosphorylation sites on mouse MAP1B to Ser1260 and Thr1265 using site-directed point mutagenesis of recombinant MAP1B proteins, in vitro kinase assays and phospho-specific antibodies. We raised phospho-specific polyclonal antibodies to these two sites and used them to show that MAP1B is phosphorylated by GSK-3β at Ser1260 and Thr1265 in vivo. We also showed that in the developing nervous system of rat embryos, the expression of GSK-3β phosphorylated MAP1B is spatially restricted to growing axons, in a gradient that is highest distally, despite the expression of MAP1B and GSK-3β throughout the entire neuron. This suggests that there is a mechanism that spatially regulates the GSK-3β phosphorylation of MAP1B in differentiating neurons. Heterologous cell transfection experiments with full-length MAP1B, in which either phosphorylation site was separately mutated to a valine or, in a double mutant, in which both sites were mutated, showed that these GSK-3β phosphorylation sites contribute to the regulation of microtubule dynamics by MAP1B.
Microtubule-associated protein 1B (MAP1B) is a developmentally regulated phosphoprotein that is expressed at high levels in differentiating neurons and in regions of the adult nervous system that show neuronal plasticity or regenerate after injury (Gordon-Weeks and Fischer, 2000). Although it is clear that MAP1B plays an important role in neurite growth and growth cone function, the underlying molecular mechanisms are unknown (Gonzalez-Billault et al., 2004).
The first clue to the molecular function of MAP1B came from experiments in which it was established that MAP1B influences microtubule dynamics by controlling microtubule stability and that this function is regulated by the serine/threonine kinase, glycogen synthase kinase-3β (GSK-3β) (Goold et al., 1999; Owen and Gordon-Weeks, 2003), which phosphorylates MAP1B (Lucas et al., 1998). Expression of GSK-3β phosphorylated MAP1B in non-neuronal cells increases the population of unstable microtubules at the expense of the stable microtubules (Goold et al., 1999). These findings predicted that inhibition of GSK-3β, or a reduction of MAP1B expression, would lead to an increase in stable microtubules in differentiating neurons. This prediction has been borne out by both experimental approaches. Inhibition of GSK-3β by lithium, a non-competitive inhibitor of GSK-3β (Klein and Melton, 1996; Stambolic et al., 1996) or SB-216763, a potent and specific competitive inhibitor of GSK-3β (Coghlan et al., 2000), causes a striking increase in the population of stable microtubules in growing axons and growth cones (Goold et al., 1999; Hall et al., 2002; Owen and Gordon-Weeks, 2003). The reduction of MAP1B phosphorylation produced by inhibition of GSK-3β is associated with a reduction in axon growth and a dramatic increase in growth cone size (Lucas et al., 1998; Goold et al., 1999; Takahashi et al., 1999; Hall et al., 2000; Hall et al., 2002; Goold and Gordon-Weeks, 2001; Williams et al., 2002; Owen and Gordon-Weeks, 2003; Goold and Gordon-Weeks, 2003). These effects are seen with either lithium or SB-216763 and, since these compounds inhibit GSK-3β by different mechanisms (Coghlan et al., 2000; Ryves and Harwood, 2001), this strongly suggests that GSK-3β mediates these effects, rather than some other enzyme (see Williams et al., 2002). Furthermore, signalling through the Wnt pathway, in which GSK-3β is also inhibited, phenocopies the effects of lithium (Lucas et al., 1998; Hall et al., 2000; Ciani et al., 2004). However, GSK-3β has a number of other phosphorylation targets in neurons and so it is not clear from these experiments that the loss of MAP1B phosphorylation is causal to the morphological changes. That this is probably the case, however, is shown by the finding that cultured neurons taken from transgenic mice in which MAP1B expression is absent or greatly reduced also show striking increases in stable microtubules in growth cones, and this is associated with a reduction in axon growth and an increase in growth cone size (Takei et al., 2000; Gonzalez-Billault et al., 2001; Gonzalez-Billault et al., 2002; Teng et al., 2001). Recent experiments in neuronal migration and axon guidance in vivo have confirmed the correlation between GSK-3β phosphorylated MAP1B and microtubule dynamics (Kawauchi et al., 2003; Del Rio et al., 2004). Collectively, these studies reveal a role for MAP1B in controlling microtubule dynamics in growing axons and growth cones and show that this function is regulated by GSK-3β phosphorylation of MAP1B. The significance of these findings is that maintaining a population of dynamic, unstable microtubules in differentiating neurons is an important requirement for axon growth and growth cone turning, a key event during growth cone pathfinding (Tanaka et al., 1995; Williamson et al., 1996; Challacombe et al., 1997) (reviewed by Gordon-Weeks, 2004).
An important first step to understanding the molecular mechanism by which MAP1B regulates microtubule dynamics is to map the site(s) at which GSK-3β phosphorylates MAP1B. We report here the identification of two GSK-3β phosphorylation sites on mouse MAP1B, at Ser1260 and Thr1265, identified using site-directed point mutagenesis of recombinant MAP1B proteins, in vitro kinase assays and phospho-specific antibodies. We raised phospho-specific polyclonal antibodies to phospho-Ser1260 and phospho-Thr1265 and showed that they recognise GSK-3β phosphorylated MAP1B in the developing nervous system. In differentiating neurons in rat embryos, GSK-3β phosphorylated MAP1B is expressed exclusively in growing axons in a gradient that is highest distally, despite the expression of both MAP1B and GSK-3β throughout all compartments of the neuron. This suggests that there is some underlying mechanism that spatially restricts the GSK-3β-phosphorylation of MAP1B in differentiating neurons to growing axons and growth cones. In addition, the hypothesis that GSK-3β phosphorylation of MAP1B regulates microtubule dynamic control by MAP1B was tested by investigating the effects of mutating Ser1260 or Thr1265, separately or together, in full-length MAP1B on microtubule populations in transfected, non-neuronal cells.
Materials and Methods
Production of SP and SPΔ plasmids
The SPAKS motif is a 21 amino acid sequence in mouse MAP1B between amino acids 1244 and 1264 (Fig. 1A). Two cDNA sequences, SP (nucleotides 3781-4643 of the mouse MAP1B sequence), which contains the SPAKS motif, and SPΔ (nucleotides 3845-4643) which does not, were produced by PCR amplification of mouse MAP1B (GenBank accession number X51396,) cloned into the pSVsport vector (Invitrogen; a kind gift from Nicholas Cowan) (Noble et al., 1989). A common reverse primer (5′-GGA TAC GGA TCC CAC AGG TGT GGC TGA CTT GT-3′) and two forward primers, one for SP (5′-TAC GAA GGA TCC GAG AGA CTT AGC CCA GCC AAG A-3′) and one for SPΔ (5′-AAG CTC GGA TCC ACT CCC CTG GGT GAA CGT A-3′) were used. All three primers were designed to incorporate BamHI restriction sites at the end of the coding region. The PCR products were subcloned into the prokaryotic expression vector pGEX-2T (Amersham Biosciences). Clones were tested for correct insertion in the appropriate orientation by asymmetric digests using Bsu361 restriction enzyme. Positive clones were characterized definitively by direct sequencing using fluorescent chain terminators on an ABI 310 capillary autosequencer.
Point mutations in recombinant proteins
Oligonucleotide-directed mutagenesis was used in conjunction with PCR and cloning to introduce point mutations into the SP sequence. The majority of point mutations (SP3, 4, 5 and all double point mutants except SP1/5 and SP2/5, see Table 1) were produced using a two PCR reaction system based on a published method (Ho et al., 1989). Using the SP plasmid DNA as a template for single point mutations and the SP5 plasmid DNA for double point mutations, the PCR reactions were designed to copy the entire plasmid DNA of the pGEX-2T vector as well as the SP/SP5 insert DNA. In each PCR reaction, either the 5′ or the 3′ primer specified the desired point mutation, along with a separate silent mutation that introduced a new restriction site in the DNA sequence to allow easy identification of mutants. The PCR products were produced in the same manner as described for the SP and SPΔ recombinants but with a specialised buffer for `long PCR reactions' (New England Biolabs). PCR products were size purified and then mixed together, heated to 95°C for 15 minutes to melt the DNA, and then incubated at room temperature for 2 hours to allow annealing. This permitted the formation of circular heterodimers owing to the overlap between the two PCR products. E. coli XL-1 Blue cells were transformed and plated on Luria-Bertani agar dishes containing 100 μg/ml ampicillin and 0.4% w/v glucose (De Bellis and Schwartz, 1990). Clones were screened for the presence of the novel restriction site. The presence of the desired point mutation was determined by direct sequencing as above.
The recombinants SP1 and SP2 and double point mutations SP1/5 and SP2/5 (see Table 1) were made with only one PCR reaction, using the 5′ primers to introduce the point mutation, while simultaneously copying the entire plasmid. However, both primers were also used in exchanging the BamHI site at the 5′ end of the SP coding sequence for an XhoI site. This produced a PCR product that had an XhoI restriction site at either terminus.
PCR products were purified with a Wizard PCR Product Purification Kit (Promega), digested with XhoI (Promega) and ligated to form a circular plasmid. E. coli XL-1 Blue cells were transformed with the DNA and transfected clones identified by testing for the presence of the XhoI enzyme restriction site and characterized definitively by direct sequencing as above.
Point mutations in full-length MAP1B at Ser1260 and Thr1265
Oligonucleotide-directed mutagenesis was used in conjunction with PCR and cloning to introduce point mutations into the mouse MAP1B sequence in the pSVsport vector. Point mutations were introduced at either residue S1260, residue T1265 or both, exchanging the serine/threonine for a valine. The S1260V and the double mutations were made using QuikChange XL (Stratagene). Individual forward and reverse primers were specifically designed for each mutation (S1260V forward, 5′-CCT GAG TCC TTC TCC GCC GGT ACC CAT AGA GAA GAC TCC-3′; S1260V reverse, 5′-GGA GTC TTC TCT ATG GGT ACC GGC GGA GAA GGA CTC AGG-3′; S1260V/T1265V forward, 5′-TTC TCC GCC AGT CCC CAT AGA GAA GGT ACC CCT GGG TGA AC-3′; S1260V/T1265V reverse, 5′-GTT CAC CCA GGG GTA CCT TCT CTA TGG GGA CTG GCG GAG AA-3′). For the T1265V mutation, a complementary pair of primers (forward, 5′-CCATAGAGAAGGTACCCCTGGGTGAA-3′; reverse, 5′-TTCACCCAGGGGTACCTTCTCTATGG-3′) was designed with a valine at codon 1265 introducing a KpnI restriction site (GGTACC). The mutagenic forward primer was used with an anti-sense primer from codons 1560-1566 (5′-ACGTCATCTGTGGTTGCCTC-3′) to amplify MAP1B sequence downstream of the mutation in a Pfu polymerase-mediated PCR reaction. A separate reaction to amplify sequence upstream of codon 1265 was used with the reverse mutagenic primer and a forward primer from codons 628-635 (5′-CAAACCCAAAGTCACCAAGT-3′). The resulting PCR products, 0.9 and 1.9 kbp respectively, were digested with restriction enzyme KpnI and then ligated in an approximately 1:1 molar ratio. This established the mutation in a 2.8 kbp segment of MAP1B coding sequence. To substitute the mutated sequence in pSVsport-MAP1B, the fragment was digested with XhoI and BstXI, which cleave at codons 646 and 1529 respectively. This fragment, and a BStXI fragment of 1.4 kbp that results from a second site in MAP1B digestion were ligated to the 8 kbp XhoI-BStXI fragment that comprises the remainder of pSV-Sport-MAP1B.
All mutations were identified by the presence of a KpnI restriction site. Recombinants were checked for size and the presence of the KpnI site and sequenced as above to establish that no unplanned mutations had arisen in handling.
Expression and purification of recombinant proteins
E. coli XL-1 Blue cells were transformed with SP, SPΔ or SP1-SP5 plasmids. Cells were grown in Luria-Bertani medium supplemented with 0.4% (w/v) glucose throughout (De Bellis and Schwartz, 1990). Cells were induced to express protein by the addition of 1 mM isopropyl β-D-thiogalactopyranoside (Sigma). Cultures were incubated for a further 2 hours and harvested by centrifugation at 5000 g for 15 minutes. The pellet was resuspended in lysis buffer [1 mg/ml lysozyme (Sigma) and 1 mM phenylmethyl sulfonylfluoride in Tris-buffered saline] and incubated at room temperature for 30 minutes, after which 1% Triton X-100 and 1 mM EDTA were added. The bacterial lysate was then sonicated to lyse the cells and centrifuged at 15,000 g for 15 minutes to pellet the cellular debris, which was then discarded. Recombinant proteins were purified from the supernatant by affinity chromatography using glutathione-agarose beads (GSH-linked agarose, Sigma).
Protein kinase assay
An in vitro kinase assay was used essentially as described (Johnstone et al., 1997; Lucas et al., 1998). Briefly, MAP1B-GST fusion proteins attached to GSH-linked agarose beads (Sigma) were incubated overnight at 37°C with a high-speed supernatant (S1) from neonatal (postnatal day 1-6) TO (Theiler's original) mouse brain in kinase buffer. Protein samples from the kinase assay were subjected to gel electrophoresis and immunoblotting as described below. In some experiments, lithium chloride (20 mM) or sodium chloride (20 mM) was added to S1 before addition of recombinant protein and in other experiments baculovirus expressed GSK-3β (Bax et al., 2001) was substituted for S1.
Gel electrophoresis and immunoblotting
Protein samples were subjected to sodium dodecyl sulphate polyacrylamide gel electrophoresis (SDS PAGE) using gradient (9%) gels and Laemmli buffer (Laemmli, 1970) and transferred onto nitrocellulose membranes (Towbin et al., 1979). To block non-specific binding, blots were incubated in blocking buffer [5% (w/v) non-fat milk solids in PBS] at room temperature for 1 hour or at 4°C overnight. The blots were then incubated in the appropriate monoclonal (mAb) or polyclonal antibody (pAb) in blocking buffer for 2 hours at room temperature. Blots were then washed five times in PBS/0.1% Tween 20 and then incubated in the appropriate peroxidase-conjugated secondary antibody (Sigma) diluted 1:500 in blocking buffer for 2 hours at room temperature. The blots were then washed as above, developed with enhanced chemiluminescent kits (Perbio Science) and scanned using a flatbed scanner (Visioneer OneTouch).
Polyclonal antibody production
Two phosphopeptides, with the sequence CSPSPPSPIEKT and CSPIEKTPLGER, in which the underlined amino acids were phosphorylated, were synthesised and conjugated to keyhole limpet haemocyanin and antisera, pAb BUGS and pAb SuperBUGS respectively, raised in rabbits (Eurogentec). Both antisera were affinity purified using a Pierce Sulfolink kit (Perbio Science). Unphosphorylated peptides with the same sequence were also synthesised for peptide inhibition studies (Eurogentec).
Preparation of tissue for immunoblotting and immunostaining
Wistar rat pregnancies were dated by counting the day of a vaginal plug as embryonic (E) day 0. E12 rat embryos were dissected from pregnant females anaesthetised with an intraperitoneal injection of sodium pentobarbitone (Sagatal) and, after decapitation, either chopped into small pieces and homogenised in gel electrophoresis buffer (see above) or fixed by immersion for 2 hours either in a solution of 3% (w/v) formaldehyde and 0.2% (v/v) glutaraldehyde in PBS at 37°C or in methacarn (Leroy and Brion, 1999).
Immunostaining of tissue sections
All tissues were embedded in a gelatine (5 mg/ml), albumin (0.3 g/ml) and sucrose (0.2 mg/ml) solution, polymerised with glutaraldehyde (2.5% v/v), sectioned at 50 μm with a Vibroslice (Camden Instruments) and the free-floating sections placed into PBS. To neutralise endogenous peroxidase activity, sections were incubated in a 4:1 methanol and 0.3% hydrogen peroxide solution for 20 minutes. Following extensive washing with PBS, the sections were left overnight at 4°C in blocking buffer [BB; 5% (v/v) normal horse serum, either 5% (v/v) normal goat serum or 5% (v/v) normal rabbit serum, 50 mM L-lysine and 0.2% (v/v) Triton X-100 in PBS]. Immunolabelling was performed by incubating sections overnight at 4°C with pAb BUGS (1:200), pAb SuperBUGS (1:900), pAb MAP1B (N-19) against all forms of MAP1B (1:1000; Santa Cruz) or pAb GSK-3β (1:200; Cell Signalling) diluted in BB, followed by a peroxidase-conjugated goat anti-rabbit or rabbit anti-goat antibody diluted in BB (1:100, DAKO) for 2 hours. Sections were extensively washed between each step with PBS and the antibody complex was visualised with diaminobenzidine (Sigma) at 0.5 mg/ml in Tris-buffered saline containing 0.01% H2O2. Controls, in which the primary antibody was replaced with BB were developed in parallel, and were negative. After washing with PBS, the sections were air-dried onto 1% gelatine-coated microscope slides, dehydrated in ethanol, cleared in xylene, and mounted in DPX. All sections were viewed with an Olympus BH2 microscope equipped with Nomarski optics and images captured with a DP70 colour CCD camera (Olympus).
Cerebral cortical cultures
Cerebral cortices were dissected from E16 mouse embryos in Hank's balanced salt solution and dissociated using trypsin (0.25%, Gibco BRL) followed by trituration with a fire-polished glass pipette. Cells (1-5×105) were plated on 13 mm glass coverslips coated with poly-D-lysine (100 μg/ml, Sigma) and laminin (10 μg/ml, Sigma). Cells were incubated at 37°C in 5% CO2 in humidified air in Neurobasal medium (Gibco BRL) supplemented with 2% (v/v) B27 supplement (Gibco BRL), 2 mM glutamine, 100 IU/ml penicillin, 100 IU/ml streptomycin (Sigma) and 0.45% D(+)-glucose for 48 hours.
Immunofluorescence staining of neuronal cultures
Cultures were washed once with PBS at 37°C followed by fixation with 3% (w/v) formaldehyde and 0.2% (v/v) glutaraldehyde in PBS for 10 minutes at 37°C. Fixed cultures were treated with BB for a minimum of 20 minutes before addition of primary antibody. Cultures were incubated for at least 1 hour with pAb BUGS (1:100) or pAb SuperBUGS (1:200), pAb MAP1B (N-19) against all forms of MAP1B (1:100; Santa Cruz), and mAb YL 1/2 (SeroTec, 1:100), which recognises tyrosinated α-tubulin (Kilmartin et al., 1982), diluted in BB, followed by five washes with PBS and then incubation with the appropriate Alexa-conjugated secondary antibodies and Alexa-conjugated phalloidin (Molecular Probes) in BB for at least 30 minutes. To control for non-specific binding, primary antibodies were excluded, and to control for cross-reactivity of secondary antibodies, combinations of inappropriate primary and secondary antibodies were assessed. All controls were negative or showed low background staining. Finally, cultures were washed five times with PBS and mounted onto microscope slides in FluorSave (Calbiochem) and viewed with phase-contrast or fluorescence optics using an Olympus BM2 microscope or a Leica TCS confocal microscope equipped with argon, krypton and HeNe lasers. In the confocal microscope, cells were imaged with ×40/1.00 PL Fluotar or ×100/1.4 PL APO oil-immersion objectives and recorded at 1024×1024 pixels per image. Switching off the appropriate laser line using the acousto-optical transmission filter (AOTF) in the confocal microscope showed that there was negligible `bleed-through' between channels. Complete z series optical sections were collected and projected onto a single plane using Leica TCS software. Fluorescent images in TIFF format were manipulated using Adobe Photoshop.
Peptide inhibition studies
Phosphorylated peptides used to raise pAb BUGS and SuperBUGS (see above) or their unphosphorylated forms were incubated at concentrations of 75 μM (SuperBUGS peptides) or 80 μM (BUGS peptides) in BB with the corresponding antibody at room temperature for 1 hour before the antibodies were used for immunoblotting or immunolabelling. For controls, antibodies were similarly treated but omitting the peptide.
Transfection of COS-7 cells
COS-7 cells were transfected with either full-length mouse MAP1B cDNA (wild type or point-mutated at Ser1260, Thr1265 or at both sites) cloned into the pSVsport vector (Invitrogen) (Noble et al., 1989) or human haemagglutinin-tagged GSK-3β cDNA cloned into the pMT-2 vector (Lovestone et al., 1994) or both, as described previously (Goold et al., 1999). Plasmids were purified on caesium chloride gradients and used to express protein in COS-7 cells grown in DMEM (Gibco BRL) containing 10% foetal bovine serum (Gibco BRL) supplemented with 2 mM glutamine, 100 IU/ml penicillin, 100 IU/ml streptomycin (Sigma). Cells (1×105) were either plated onto 13 mm glass coverslips in 35 mm petri dishes for immunofluorescence microscopy or directly onto plastic petri dishes for biochemical analysis. Cells were transfected by lipofection using Lipofectamine 2000 reagent (Gibco BRL) and 2 μg DNA per dish according to the manufacturer's protocol. For double transfection experiments, an equal quantity of each plasmid was added simultaneously. For controls, cells were treated as for transfections except that DNA was omitted. 24 hours after transfection was initiated the transfection medium was replaced with complete DMEM and the cells were harvested or fixed 24 hours later. At this time point, cultures were ∼80% confluent.
Immunofluorescence staining of COS-7 cells
Cells were washed once with PBS at 37°C and then fixed in methanol at –20°C for 5 minutes and re-hydrated in PBS. Cells were stained as previously described (Goold et al., 1999) using the following antibodies in combination: a goat pAb against all forms of MAP1B (pAb MAP1B-N19, Cell Signalling, diluted 1:100), pAb BUGS (1:100), pAb SuperBUGS (1:100), a rat mAb against the haemagglutinin epitope tag, to recognise haemagglutinin-tagged GSK-3β (clone 3F10, 1:100, Boehringer Mannheim) and a pAb against detyrosinated α-tubulin (pAb SUP GLU, diluted 1:1000) (Gundersen et al., 1984). Triple labelling was done with appropriate Alexa-conjugated secondary antibodies. To control for non-specific binding, primary antibodies were excluded, and to control for crossreactivity of secondary antibodies, combinations of inappropriate primary and secondary antibodies were assessed. All controls were negative or showed low background staining. Cultures were washed in PBS and mounted in FluorSave on microscope slides and viewed with fluorescence optics using an Olympus BM2 microscope or a Leica TCS confocal microscope (see above).
Quantification of stable microtubules in transfected cells
To determine the concentration of stable microtubules in COS-7 cells, cultures were triple-labelled with: pAb SUP GLU, to identify stable microtubules, pAb MAP1B-N19, to identify cells expressing MAP1B and a rat mAb against the haemagglutinin epitope tag, to identify cells expressing haemagglutinin-tagged GSK-3β. Measurements were made of the intensity of fluorescence due to pAb SUP GLU labelling. Complete z series (0.5 μm separation) were collected from fields containing both transfected and non-transfected cells using ×40/1.00 PL Fluotar oil-immersion objective and the series projected onto a single plane. The fluorescence intensity of pAb SUP GLU labelling was then measured using the Leica TCS NT software and, at the same time, the area of the cell was measured so that the fluorescence intensity per unit area could be calculated. Ten cells were measured from each of three separate experiments for both transfected and non-transfected cells.
GSK-3β phosphorylates mouse MAP1B on Ser1260 and Thr1265
In previous studies, we mapped a glycogen synthase kinase 3β (GSK-3β) phosphorylation site on mouse microtubule-associated protein 1B (MAP1B) to a region between amino acids 1244 and 1264 (Fig. 1A) (Johnstone et al., 1997). This 21 amino acid stretch is referred to as the SPAKS sequence, because of a motif contained within it, and has five serine/proline motifs that are candidates for GSK-3β phosphorylation (Fig. 1A). We recognised the GSK-3β site using a monoclonal antibody (mAb) SMI-31 that we have previously shown binds to a GSK-3β phosphorylation epitope on MAP1B (Lucas et al., 1998). To confirm the location of the GSK-3β site in the SPAKS sequence we made two glutathione S-transferase (GST)-MAP1B recombinant proteins that differed only in the presence (SP) or absence (SPΔ) of the SPAKS sequence. When these two recombinant proteins were assessed in an in vitro protein kinase assay only the SP recombinant became immunoreactive for mAb SMI-31 (Fig. 1B). This result provides independent confirmation of our earlier work mapping the mAb SMI-31 epitope generated by GSK-3β to the SPAKS region (Johnstone et al., 1997).
To precisely locate the GSK-3β phospho-acceptor amino acid(s), we made a panel of recombinant proteins derived from the SP recombinant, in which point mutations were made separately in all five serine/proline (SP) motifs in the SPAKS sequence by changing serine to valine, so that it cannot be phosphorylated, and tested each in the kinase assay (Table 1, Fig. 2). Only the SP5 mutant showed a reduction in GSK-3β phosphorylation, strongly suggesting that Ser1260 is the GSK-3β phospho-acceptor amino acid in the SPAKS site (Fig. 2). Although the phosphorylation was markedly reduced it was not abolished. The antibody that we used to detect GSK-3β phosphorylated MAP1B, mAb SMI-31, also recognises phosphorylation epitopes on the microtubule-associated protein tau (Lichtenberg-Kraag et al., 1992). Mapping studies of the mAb SMI-31 sites in tau have shown that maximum binding of mAb SMI-31 requires two phosphorylated amino acid residues and that if either residue is mutated so that it cannot be phosphorylated, the binding is reduced by about one half (Lichtenberg-Kraag et al., 1992). We reasoned by analogy that a similar requirement for the binding of mAb SMI-31 to MAP1B may apply and so we prepared a panel of double mutated recombinant proteins in which Ser1260 and one of the other serines in the SPAKS site were changed to valines (Table 1, Fig. 3). In addition, as there are GSK-3β sites that require prior `priming' by phosphorylation of an amino acid four residues downstream of the GSK-3β phospho-acceptor amino acid (Doble and Woodgett, 2003), we mutated Thr1265 to a valine, even though it is five amino acids downstream of Ser1260 (Fig. 3). When we tested the double mutants in the kinase assay, recombinant proteins in which serines 1247, 1251, 1253 or 1255 had been changed to valine, in addition to Ser1260, were immunoreactive to mAb SMI-31, whereas recombinant proteins in which Ser1257 or Thr1265 had been changed to valine, in addition to Ser1260, were completely unreactive, or showed greatly reduced immunoreactivity to mAb SMI-31 (Fig. 3). This suggests that mAb SMI-31 requires the phosphorylation of serines 1260 and 1257 and Thr1265 for maximum binding. Finally, we produced a single point-mutated recombinant protein (SPT, Table 1) in which Thr1265 was changed to valine in the SP recombinant. Interestingly, this mutation abolished the binding of mAb SMI-31, suggesting that the phosphorylation of Thr1265 is essential for binding of mAb SMI-31 (Fig. 4). This occurred despite the fact that the SPΔ mutant (see above, Fig. 1), which retains Thr1265, does not become reactive to mAb SMI-31 after phosphorylation in the kinase assay, perhaps because it lacks any upstream sequence.
Polyclonal antibodies to phospho-Ser1260 or phospho-Thr1265 recognise MAP1B in vivo
We raised rabbit polyclonal antibodies to two phosphopeptides and then affinity purified them; one corresponded to the SPAKS sequence in which Ser1260 was phosphorylated (pAb BUGS) and one in which Thr1265 was phosphorylated (pAb SuperBUGS; see Materials and Methods). Polyclonal antibodies BUGS and SuperBUGS recognised the SP recombinant protein only when it had been phosphorylated in an in vitro protein kinase assay (Fig. 4). Furthermore, the binding of both antibodies was considerably reduced when lithium, a GSK-3 inhibitor (Klein and Melton, 1996; Stambolic et al., 1996), was included in the kinase assay buffer (Fig. 5). Polyclonal antibody BUGS failed to recognise the SP5 recombinant protein, in which Ser1260 had been mutated to valine, after incubation with S1 in kinase buffer (Fig. 5), or the double mutant SPT/5 (Fig. 4), but continued to recognise recombinant proteins SPT, in which Thr1265 had been mutated to valine (Fig. 4). In contrast, polyclonal antibody SuperBUGS recognised the SP5 but not the SPT recombinant protein, and, like pAb BUGS, failed to recognise the double mutant SPT/5 (Fig. 4). Finally, incubation of the SP recombinant protein with recombinant GSK-3β in kinase buffer generated the pAb BUGS and pAb SuperBUGS epitopes (Fig. 5). These results show that pAb BUGS is specific for GSK-3β phosphorylated MAP1B at Ser1260 and pAb SuperBUGS is specific for GSK-3β phosphorylated MAP1B at Thr1265.
Recombinant proteins SP and SP5 undergo gel-shifts when phosphorylated in the kinase assay, as indicated by Coomassie Blue staining (not shown) and anti-GST blots (Fig. 4). Interestingly, the SP recombinant gel-shift is reduced, but not abolished, by lithium treatment (Fig. 5), and thus inhibition of GSK-3β, suggesting that there are additional kinases that phosphorylate SP in the kinase assay. The SP5 recombinant lacks the Ser1260 GSK-3β site but still undergoes a partial inhibition of the gel-shift after lithium treatment (not shown), consistent with a second GSK-3β site, Thr1265, in the SP recombinant. A similar partial inhibition of the gel-shift after lithium is also seen with the SPT recombinant (not shown).
MAP1B phosphorylated at Ser1260 and Thr1265 by GSK-3β is expressed in a proximo-distal gradient in growing axons in rat embryos
A number of studies have shown that developmentally downregulated phosphorylated isoforms of MAP1B are exclusively expressed in growing axons and that in some cases there is a proximo-distal gradient of expression that is highest at the growth cone (reviewed by Gordon-Weeks and Fischer, 2000). However, in all of these studies the kinase responsible and the phosphorylation sites were uncharacterised. It was of interest therefore, to establish the expression pattern of MAP1B phosphorylated at Ser1260 and Thr1265 by GSK-3β using pAb BUGS and pAb SuperBUGS. We chose to investigate this in the E12 embryonic rat spinal cord where the trajectories of growing axons from a small number of well-characterised neurons are readily distinguishable.
Immunoblot analysis of E12 rat embryos (Fig. 6) showed that pAb BUGS and pAb SuperBUGS exclusively recognise a protein that co-migrates with native MAP1B, as judged by probing with mAb AA6, which recognises all forms of MAP1B (Riederer et al., 1986), and SMI-31, which recognises a GSK-3β phosphorylated form of MAP1B (Lucas et al., 1998). Peptide inhibition studies showed that pre-incubation of pAb BUGS or pAb SuperBUGS with the phosphopeptide to which they were raised (see Materials and Methods), but not the unphosphorylated peptide, abolished antibody binding to MAP1B in immunoblots (not shown).
Polyclonal antibodies BUGS and SuperBUGS specifically stained growing axons in sections of the spinal cord of E12 rat embryos (Fig. 7A,B). Commissural axons and axons in longitudinal tracks were stained by both antibodies, as were axons in the peripheral nervous system, including those of sensory neurons in the dorsal root ganglion and axons of motor neurons (Fig. 7A,B). We noticed that there was often a staining gap in the proximal regions of axons near the parent cell body, particularly with pAb SuperBUGS; for example, there was little proximal staining of motor neuron axons in the cord and there was often a distinct gap discernable between the border of the spinal cord and the stained axons in the ventral root (Fig. 7B). Similarly, the axons of sensory neurons in the dorsal root ganglia were only stained at the poles of the ganglia. Neither antibody stained neuronal cell bodies. In contrast, antibodies that recognise all forms of MAP1B stained neurons in their entirety, including neuronal cell bodies and axons throughout their length (Fig. 7C). For example, axons could be seen traversing the entire length of the dorsal root ganglion. This suggests that the GSK-3β phosphorylated form of MAP1B recognised by these antibodies is expressed in a gradient that is highest distally in growing axons. Consistent with this idea, when pAb BUGS and pAb SuperBUGS were diluted out there was a progressive loss of proximal axonal staining until, at very low dilutions, only the most distal regions of axons were stained (not shown). In contrast to the staining pattern of pAb BUGS and pAb SuperBUGS, antibodies to GSK-3β stained the entire axon and the cell bodies of differentiating neurons in the embryonic spinal cord, including primary sensory neurons and motor neurons (Fig. 7D), as previously reported for differentiating neurons in the brain (Takahashi et al., 1994; Leroy and Brion, 1999).
In addition to axonal staining, pAb SuperBUGS, but not pAb BUGS, also stained mitotic cells throughout the embryo (Fig. 7B). Most commonly, this staining was associated with the mitotic spindle in cells in metaphase (Fig. 7B, inset). This staining is reminiscent of the staining of mitotic cells in the developing cat cerebellum seen with mAb 150, which recognises an uncharacterised phosphorylation epitope on MAP1B (Riederer et al., 1993) and is probably due to phosphorylated MAP1B in the spindle (Tombes et al., 1991). Peptide inhibition studies showed that pre-incubation of pAb BUGS or pAb SuperBUGS with the phosphopeptide to which they were raised (see Materials and Methods), but not the unphosphorylated peptide, abolished all staining of these antibodies (not shown).
Embryonic cerebral cortical neurons in culture express GSK-3β-phosphorylated MAP1B predominantly in growing axons
After 48 hours in culture, many cerebral cortical neurons had extended a single, long process, which develops into an axon, and a number of short processes, which will become dendrites (Fig. 8). Polyclonal antibody SuperBUGS only labelled long processes and the labelling was invariably present in the form of a gradient, which was highest in intensity at the growth cone (Fig. 8). This labelling pattern is similar to that seen with some other phosphospecific antibodies against MAP1B (Mansfield et al., 1991; Bush et al., 1996). In contrast, although pAb BUGS labelling was occasionally restricted to the long process and expressed as a gradient, more commonly, pAb BUGS labelled the long process along its entire length and also labelled the neuronal cell body and proximal regions of the short processes, although less intensely than the long process (not shown). Labelling within the axonal growth cone with pAb SuperBUGS (Fig. 8D-F) and pAb BUGS (Fig. 9G-I) was not uniform and, in addition to filamentous labelling, which colocalised with microtubules, there was some labelling that occasionally overlapped with the actin filaments (Fig. 8). This distribution is entirely consistent with a role for these MAP1B phosphorylation sites in the regulation of microtubule dynamics in growth cones. Non-neuronal cells in these cultures were not labelled by either antibody (not shown). As expected, phosphate-independent antibodies to MAP1B labelled the entire neuron including all processes (Fig. 8).
Transfection of non-neuronal cells with MAP1B and GSK-3β generates the pAb BUGS and pAb SuperBUGS epitopes on MAP1B
In previous experiments, we have shown that transient transfection of MAP1B and GSK-3β into non-neuronal cells (COS-7 and CHO) results in the phosphorylation of MAP1B by GSK-3β and the appearance of the mAb SMI-31 epitope on MAP1B (Goold et al., 1999). We repeated these experiments with COS-7 cells and looked for the appearance of immunoreactivity for pAb BUGS and pAb SuperBUGS in transfected cells. When double-transfected cultures were immunolabelled with pAb BUGS or pAb SuperBUGS, cells that expressed MAP1B and GSK-3β were immunopositive with these antibodies (Fig. 9A-F). In contrast, when cells were transfected with GSK-3β and point-mutated MAP1B, in which Ser1260 or Thr1265 or both had been changed to a valine by site-directed mutagenesis, although full-length MAP1B was expressed in these cells, as confirmed by immunoblotting, the epitopes recognised by pAb BUGS or pAb SuperBUGS were not generated (not shown). In cells transfected with either MAP1B or GSK-3β alone, no staining with either antibody was seen (not shown). These experiments confirm the immunoblotting data with recombinant MAP1B in showing that pAb BUGS recognises phospho-Ser1260 and pAb SuperBUGS recognises phospho-Thr1265 on MAP1B and that these phosphorylation sites are generated on MAP1B by GSK-3β.
Transfection of non-neuronal cells with GSK-3β and MAP1B, mutated at Ser1260, Thr1265 or double mutants, partially rescues the loss of stable microtubules
We have shown previously that transient expression of GSK-3β phosphorylated MAP1B in non-neuronal cells results in the loss of stable microtubules from these cells (Goold et al., 1999). It was of interest, therefore, to see whether full-length MAP1B in which either Ser1260 or Thr1265 or both had been mutated to a valine, which precludes phosphorylation by GSK-3β at these sites, would also cause a loss of stable microtubules when transfected into non-neuronal cells. When we transfected GSK-3β and full-length wild-type MAP1B into non-neuronal cells, MAP1B became phosphorylated and there was a significant reduction in the proportion of cells containing stable microtubules, as judged by immunostaining with pAb SUP GLU, which recognises stable microtubules (Gundersen et al., 1984), confirming our previous findings (Fig. 9G-I) (Goold et al., 1999). However, when we transfected COS-7 cells with full-length MAP1B in which Ser1260, Thr1265 or both had been mutated to a valine, although this also resulted in a considerable reduction of stable microtubules, many double-transfected cells still contained a few stable microtubules (Fig. 9J-L). When we measured the amount of fluorescence due to pAb SUP GLU staining and expressed it as a concentration, we found that mutating Ser1260, Thr1265 or both to valine in MAP1B resulted in a lower reduction of stable microtubule fluorescence than that seen with wild-type MAP1B (Fig. 9M). Interestingly, the double mutation did not improve the degree of rescue over that seen with the single point mutations suggesting that both sites must be phosphorylated to contribute to the regulation of microtubule dynamics. Thus, mutating Ser1260, Thr1265 or both to valine in MAP1B, so that these sites cannot be phosphorylated by GSK-3β, partially rescues the loss of stable microtubules seen with wild-type MAP1B. As previous studies have shown that the loss of stable microtubules is entirely dependent on GSK-3β phosphorylation of MAP1B (Goold et al., 1999), the present results imply that other GSK-3β sites within MAP1B contribute to this regulation.
MAP1B is phosphorylated by GSK-3β on Ser1260 and Thr1265
MAP1B is an important downstream target of GSK-3β in differentiating neurons (Lucas et al., 1998; Goold et al., 1999; Goold and Gordon-Weeks, 2001; Goold and Gordon-Weeks, 2003; Goold and Gordon-Weeks, 2005; Hall et al., 2002; Kawauchi et al., 2003; Owen and Gordon-Weeks, 2003). Phosphorylation of MAP1B by GSK-3β appears to regulate the control that MAP1B exerts on microtubule dynamics in growing axons and growth cones, without altering the affinity of MAP1B for microtubules (Goold et al., 1999; Owen and Gordon-Weeks, 2002). Identification of the site(s) on MAP1B phosphorylated by GSK-3β is an important prerequisite to determining their physiological effect on MAP1B function. We have mapped two GSK-3β phosphorylation sites on mouse MAP1B to Ser1260 and Thr1265 by combining site-directed mutagenesis with an in vitro kinase assay based on mAb SMI-31, which recognises GSK-3β phosphorylated MAP1B (Johnstone et al., 1997; Lucas et al., 1998). The serine and threonine are conserved in human, mouse and rat MAP1B and the serine in the Drosophila homologue, Futsch (Hummel et al., 2000; Roos et al., 2000), which strongly suggests that they have an important function. Whether these amino acid residues are phosphorylated in all of these species has yet to be determined.
To confirm the identification of Ser1260 and Thr1265 in MAP1B as phospho-acceptor residues for GSK-3β, we raised two polyclonal antibodies, pAb BUGS and pAb SuperBUGS, against synthetic phosphopeptides containing the MAP1B sequence around amino acid residues 1260 and 1265 respectively. Using recombinant MAP1B proteins and site-directed point mutagenesis in conjunction with in vitro kinase assays, we showed that pAb BUGS selectively and specifically recognises phospho-Ser1260 in recombinant MAP1B proteins and that pAb SuperBUGS selectively and specifically recognises phospho-Ser1265. Furthermore, both phosphorylation sites are generated by GSK-3β since they can be produced in vitro by recombinant GSK-3β and in vivo by heterologous cell transfection and are sensitivity to lithium, an inhibitor of GSK-3β (Klein and Melton, 1996; Stambolic et al., 1996).
Ser1260 and Thr1265 are not primed sites
There are two classes of substrates phosphorylated by GSK-3β, one that requires prior phosphorylation of the substrate by another kinase four amino acids downstream of the GSK-3β phospho-acceptor amino acid, a so-called `priming' site, and one that requires no priming and tends to occur at serine/threonine-proline motifs (Doble and Woodgett, 2003). Previous in vitro experiments had shown that recombinant GSK-3β can phosphorylate an unprimed recombinant MAP1B protein at the site recognised by mAb SMI-31 (Lucas et al., 1998), confirmed in the present experiments, and failed to find biochemical evidence for priming (Goold and Gordon-Weeks, 2001), suggesting that the GSK-3β phosphorylation site on MAP1B recognised by mAb SMI-31 is not a primed site. Although there is no phosphorylatable amino acid four residues downstream of Ser1260 in MAP1B, there is a threonine five amino acid residues downstream. However, although mutation of this threonine to valine did have an effect on residual mAb SMI-31 binding, it did not affect pAb BUGS binding, confirming that there is no priming site for Ser1260. The nearest serine/threonine downstream of Thr1265 in mouse MAP1B is Ser1271 and therefore this site is also a non-primed one. However, four amino acids downstream of either Ser1260 or Thr1265 are charged amino acid residues; lysine and glutamic acid respectively, and the latter is often used to mimic phosphorylation in point substitution analyses. In the consensus motif for GSK-3β primed sites there are three amino acids separating the phosphorylated amino acids (Doble and Woodgett, 2003) whereas there are four amino acids separating the two GSK-3β phosphorylation sites that we have mapped on MAP1B. Recent mapping experiments of the GSK-3β primed phosphorylation sites on the protein CRMP have revealed that two of these sites are separated by four amino acids (Cole et al., 2004). We will therefore have to relax the consensus motif to allow for this variation in amino acid spacing between the phospho-acceptor amino acids.
The epitope recognised by mAb SMI-31 in tau protein requires two phosphorylated serine residues that contribute equally to the binding of the antibody (Lichtenberg-Kraag et al., 1992). Both serines are phosphorylated by GSK-3β and neither site is primed (Cho and Johnson, 2003). We show here that mAb SMI-31 also requires more than one phospho-amino acid for full binding to MAP1B, although the contribution that each makes to binding is not equal. When Ser1260 was mutated to valine in MAP1B recombinant proteins, mAb SMI-31 binding was almost completely lost but was only totally abolished when certain neighbouring serines/threonines were also mutated. This suggests that there is a conformational component to mAb SMI-31 binding to MAP1B, which is the case with mAb SMI-31 binding to tau (Lichtenberg-Kraag et al., 1992).
The expression of GSK-3β-phosphorylated MAP1B is restricted to growing axons in differentiating neurons
We showed that pAb BUGS and pAb SuperBUGS only recognise native MAP1B in immunoblots of rat embryo extracts and peptide inhibition studies confirmed that the MAP1B epitopes are phosphorylated and so we used these antibodies to determine the expression pattern of MAP1B phosphorylated at Ser1260 and 1265 in sections of rat embryos. Both antibodies exclusively stained growing axons, thus confirming that Ser1260 and Thr1265 in MAP1B are physiologically phosphorylated by GSK-3β in this compartment of differentiating neurons. This pattern of expression is seen with a number of other antibodies against phospho-specific epitopes on MAP1B that are developmentally downregulated (reviewed by Fischer and Gordon-Weeks, 2002). However, in embryonic cerebral cortical neurons in culture, although pAb BUGS and pAb SuperBUGS labelling was neuron-specific, the labelling of pAb SuperBUGS was restricted to axons and present in the form of a proximo-distal gradient, whereas pAb BUGS labelling was usually uniform along the axon and there was some labelling of neuronal cell bodies and minor (dendritic) processes. These neurons are regenerating axons in culture and this difference, and the fact that their differentiation is no longer under embryonic influence, may underlie the cell body staining. However, there are other explanations for the difference in the spatial expression of the two phosphorylated forms of MAP1B. For example, it is possible that there are other kinases that can phosphorylate these sites in vivo and that there is differential kinase usage between the two sites. Previously, GSK-3β phosphorylated MAP1B has been recognised using mAb SMI-31 but this antibody has a number of disadvantages compared to pAb BUGS and pAb SuperBUGS, including crossreactivity towards neurofilament and tau phospho-epitopes and an uncharacterised nuclear phospho-epitope. Polyclonal antibodies BUGS and SuperBUGS will prove more useful tools than mAb SMI-31 with which to investigate GSK-3β phosphorylation of MAP1B.
We also showed that both MAP1B and GSK-3β are expressed throughout differentiating neurons, including the cell body, and, therefore, the staining pattern seen with pAb BUGS and pAb SuperBUGS implies that there is a local regulation of GSK-3β-mediated phosphorylation in growing axons that restricts the localisation of MAP1B phosphorylation. A similar phenomenon has been seen in differentiating PC12 cells (Goold and Gordon-Weeks, 2000). The mechanism for restricting the phosphorylation of MAP1B by GSK-3β to growing axons is unknown. It could conceivably involve an inhibition of phosphorylation in the cell body or the activation of phosphorylation in the axon, by a process associated with axonogenesis, for example the targeting of some cellular factor to the axon or the interaction of the axon with its environment in the embryo. This latter mechanism would locally activate GSK-3β phosphorylation of MAP1B. In support of such a mechanism we have recently found that GSK-3β must be activated in order to phosphorylate MAP1B and that activation probably involves a post-translational modification of GSK-3β such as phosphorylation (Goold and Gordon-Weeks, 2005). Finally, it could be that MAP1B is phosphorylated by GSK-3β in the cell body but that the phosphorylation epitope is masked, for example by a binding protein, so that the antibody cannot access it (Bush et al., 1996).
Mutation of Ser1260, Thr1265 or both to unphosphorylatable residues partially rescues the loss of stable microtubules
In previous experiments we have shown that phosphorylation of MAP1B by GSK-3β regulates the control of microtubule dynamics that MAP1B exerts in growing axons and growth cones (Goold et al., 1999; Owen and Gordon-Weeks, 2003) (see also Kawauchi et al., 2003). Identifying GSK-3β sites in MAP1B provided the opportunity to test the role of these sites in regulating MAP1B function. We found that mutated MAP1B, in which Ser1260, Thr1265 or both were changed to a valine, so that they could not be phosphorylated by GSK-3β at these sites, only partially rescued the loss of stable microtubules that GSK-3β phosphorylated MAP1B causes when transfected into non-neuronal cells. Although confirming that GSK-3β sites on MAP1B are involved in the regulation of microtubule dynamics, this finding suggests that other GSK-3β sites in MAP1B also contribute to the regulation of microtubule stability. Tau, another microtubule binding protein, is phosphorylated by GSK-3β at more than one site and, interestingly, both primed and unprimed sites are used (Cho and Johnson, 2003). In future studies we will search for other GSK-3β phosphorylation sites on MAP1B and characterise and test their function.
We thank Chlöe Bulinski for pAb SUP GLU, Nicholas Cowan for MAP1B cDNA, Jim Woodgett for GSK-3β cDNA, GlaxoSmithKline for baculovirus-expressed GSK-3β and the Gordon-Weeks lab for comments on the manuscript. N.T. was supported by an MRC postgraduate studentship, A.W.-K. was supported by a GlaxoSmithKline studentship and the work was funded by a grant from The Wellcome Trust.
↵* Present address: Centre for Molecular Neurobiology and Department of Neuroscience, The Ohio State University, Rightmire Hall, 1060 Carmack Road, Columbus, OH 43210, USA
- Accepted December 30, 2004.
- © The Company of Biologists Limited 2005