Address correspondence and reprint requests to F. Hucho, Freie Universität Berlin, Institut für Chemie und Biochemie, Thielallee 63, 14195 Berlin, Germany. E-mail: firstname.lastname@example.org
Previously, we reported that TRPV1, the vanilloid receptor, interacts with soluble αβ-tubulin dimers as well as microtubules via its C-terminal cytoplasmic domain. The interacting region of TRPV1, however, has not been defined. We found that the TRPV1 C-terminus preferably interacts with β-tubulin and less with α-tubulin. Using a systematic deletion approach and biotinylated-peptides we identified two tubulin-binding sites present in TRPV1. These two sequence stretches are highly conserved in all known mammalian TRPV1 orthologues and partially conserved in some of the TRPV1 homologues. As these sequence stretches are not similar to any known tubulin-binding sequences, we conclude that TRPV1 interacts with tubulin and microtubule through two novel tubulin-binding motifs.
Like other transient receptor potential (TRP) channels, TRPV1 is a non-selective cation channel (Caterina et al. 1997). Both N-terminal and C-terminal sequences of TRPV1 form cytoplasmic domains. Previously, we identified αβ-tubulin as TRPV1 interacting partner (Goswami et al. 2004). We demonstrated that the C-terminus of TRPV1 is sufficient and interacts directly with microtubules (Goswami et al. 2004). It provides stability to microtubules both in vitro and in vivo (Goswami et al. 2004, 2006). Interestingly, tubulin interaction is observed also for other members of the TRP super family. Interaction of β-tubulin with TRPC1 has been reported recently (Bollimuntha et al. 2005). Two other members, namely TRPC5 and TRPC6, contain tubulin as constituent of its ‘signalplex’ (Goel et al. 2005). Very recently, it has been shown that Polycystin-2 type TRP channels are regulated by microtubular structures in primary cilia of renal epithelial cells (Li et al. 2006). This suggests that tubulin interaction might be common for many of the TRP ion channels.
In spite of the functional implication of the interaction of tubulin with several transmembrane receptors and ion channels, very little is known about the binding structure/s that underlie these interactions. Therefore, we set out to identify the exact tubulin-binding region of TRPV1 and further characterised the interacting structures.
Materials and methods
Reagents and antibodies
The microtubule stabilising drug Taxol® (paclitaxel), the cross-linker DMS and purified actin, were purchased from Sigma–Aldrich (Taufkirchen, Germany). Biotinylated-peptides (KSFLKCMRKAFRSGKLLQVGF-K-Biotin and KRTLSFSLRSGRVSGRNWKNF-K-Biotin) were synthesised at Biosynthan (Berlin, Germany). Mouse monoclonal α-tubulin antibodies (clone DM1A), mouse monoclonal β-tubulin antibodies (clone D66), mouse monoclonal tyrosinated tubulin antibodies (clone TUB1A2), mouse monoclonal polyglutamylated tubulin antibodies (clone B3), mouse monoclonal acetylated tubulin antibodies (clone 611-B-1), mouse monoclonal phosphoserine antibodies (Clone PSR-45) and mouse monoclonal anti-β-tubulin sub type III (clone SDL.3D10) were purchased from Sigma–Aldrich. Mouse monoclonal neurofilament 200 kDa antibodies (clone RT97) and rabbit polyclonal detyrosinated tubulin antibodies were purchased from Chemicon (Chandlers Ford, UK). Mouse monoclonal actin antibodies (clone JLA20) was purchased from Oncogene (Cambridge, MA, USA). Mouse monoclonal anti-maltose-binding protein (MBP) antibodies and amylose resin were purchased from New England Biolab (Beverly, MD, USA). Enriched neurofilament fraction was a kind gift from O. Bogen (Bogen et al. 2005). Subtilisin-digested tubulin and control tubulin were kindly provided by Linda Amos (Cambridge, UK). For the detection of subtilisin-digested tubulin and control tubulin by western-blot analysis, we used mouse monoclonal anti-β-tubulin (clone D10, Santa Cruz Biotechnology, Heidelberg, Germany).
Expression and purification of TRPV1 fusion proteins
Expression and purification of MBP-TRPV1-Nt (N-terminal cytoplasmic domain of TRPV1 fused with MBP) and MBP-TRPV1-Ct (C-terminal cytoplasmic domain of TRPV1 fused with MBP) were described in Goswami et al. (2004). The cDNA fragments of TRPV1-Ct (see Fig. 1) were amplified by PCR using specific primers (Table 1). All amplified DNA fragments were subcloned into the EcoR1 and Hind III restriction sites of the pMAL-c2x vector (New England Biolabs, Beverly, MA, USA). A stop codon was introduced in each construct at the C-terminus of the coding sequences. All expression constructs were verified by automated nucleotide sequencing. Escherichia coli (E. coli) strain BL21DE3 was transformed by heat shock with the plasmid coding for the TRPV1 cytoplasmic domains and fragments fused with MBP protein. E. coli cells were induced to express the proteins by isopropyl thiogalactoside (IPTG) for 2 h. The cells were lysed by repeated freeze-thaw cycles in lysis buffer (20 mmol/L Tris–HCl, pH 7.4, 150 mmol/L NaCl, 0.1% Tween 20, lysozyme, benzonase and protease inhibitor cocktail). The lysed extracts were cleared by centrifugation (100 000 g in a TFT 45 rotor for 2 h). The cleared lysate was applied to amylose resin and washed thoroughly. Bound protein was eluted with 10 mmol/L maltose in elution buffer (50 mmol/L PIPES, pH 6.8, 100 mmol/L NaCl, 1 mmol/L EGTA and 0.2 mmol/L MgCl2). Protein concentration was determined according to method described by Bradford (1976).
Table 1. Primers used for making the TRPV1-Ct deletions and fragments
F, forward primer; R, reverse primer; Underlines indicate the presence of stop codon.
F: 5′ CCGGAATTCCTCATGGGTGAGACCGTCAAC 3′
R: 5′ CCCAAGCTTTTAGCTTGCATCCCTCAGAAGGGG 3′
F: 5′ CCGGAATTCCTCATGGGTGAGACCGTCAAC 3′
R: 5′ CCCAAGCTTTTAGTTGATGATACCCACATTGGT 3′
F: 5′ CCGGAATTCCTCATGGGTGAGACCGTCAAC 3′
R: 5′ CCCAAGCTTTTAGAACCCCACCTGCAGCAGCTT 3′
F: 5′ CCGGAATTCCTCATGGGTGAGACCGTCAAC 3′
R: 5′ CCCAAGCTTTTACTCTGTATCCAGGATGGTGAT 3′
F: 5′ CCGGAATTCAAGAGCTTCCTGAAGTGCATG 3′
R: 5′ CCCAAGCTTTTAGAACCCCACCTGCAGCAGCTT 3′
F: 5′ CCGGAATTCACTCCTGACGGCAAGGATGAC 3′
R: 5′ CCCAAGCTT TTAGACGCCCTCACAGTTGCCTGG 3′
F: 5′ CCGGAATTCAAGCGCACCCTGAGCTTCTCC 3′
R: 5′ CCCAAGCTTTTACCTCAGAAGGGGAACCAGGGC 3′
F: 5′ CCGGAATTCACTCGAGATAGACATGCCACC 3′
R: 5′ CCCAAGCTTTTATTTCTCCCCTGGGACCATGGA 3′
F: 5′ CCGGAATTCGAGGACCCAGGCAACTGTGAG 3′
R: 5′ CCCAAGCTTTTATTTCTCCCCTGGGACCATGGA 3′
F: 5′ CCGGAATTCACTCCTGACGGCAAGGATGAC 3′
R: 5′ CCCAAGCTTTTATTTCTCCCCTGGGACCATGGA 3′
F: 5′ CCGGAATTCAAGAGCTTCCTGAAGTGCATG 3′
R: 5′ CCCAAGCTTTTACCTCAGAAGGGGAACCAGGGC 3′
Purification of tubulin
αβ-tubulin dimers were purified from porcine brain according to Shelanski et al. (1973). In brief, two cycles of assembly from soluble brain extract in the presence of glycerol and GTP and disassembly by cold temperature (ice-cold) were followed by chromatography on phosphocellulose.
MBP-LacZ, MBP-TRPV1-Ct, different MBP-TRPV1-Ct fragments and deletion constructs (see Fig. 1) were expressed in E. coli, the cleared cell lysates were applied to amylose resin (NEB), and incubated for 1 h at 25°C followed by washing. The amylose resin with bound proteins were re-suspended in PEM-S buffer (50 mmol/L PIPES, pH 6.8, 100 mmol/L NaCl, 1 mmol/L EGTA and 0.2 mmol/L MgCl2). Approximately, 50 μL of amylose resin with the bound fusion protein was incubated with 50 μL of soluble tubulin (1 mg/mL protein) for 1 h at 25°C either in the presence or absence of Ca2+ (2 mmol/L). This was followed by three washes with 200 mL each time and constant buffer conditions. The proteins were eluted by 10 mmol/L maltose in 100 μL solution. Eluted samples were analysed by 10% sodium dodecyl sulphate polyacrylamide gel electrophoresis (SDS-PAGE) according to Laemmli (1970).
For experiments determining the binding of subtilisin-digested tubulin with TRPV1-Ct, MBP-TRPV1-Ct immobilised on amylose resin (20 μL) in PEM-S buffer were incubated with 2 μg of subtilisin-digested tubulin or same amount of control tubulin for 1 h at 25°C. After three washes in PEM-S buffer, the bound protein complexes were eluted and analysed further.
To identify if there is a direct interaction between tubulin and short peptides carrying TRPV1 sequences, biotinylated-peptides were incubated with avidin agarose (Sigma–Aldrich) at 25°C for 1 h, washed extensively with PEM-S-T (50 mmol/L PIPES, pH 6.8, 1 mmol/L EGTA, 0.2 mmol/L MgCl2, 150 mmol/L NaCl and 0.1% Tween 20) buffer thrice, incubated with the soluble tubulin dimer (40 μg in 100 μL) for 1 h. Avidin–agarose resins were washed thrice with PEM-S-T buffer and finally taken in to Laemmli sample buffer and analysed by western-blot analysis for bound proteins.
Cross-linking of proteins
A protein mixture (1 mg/mL) of equal amounts of αβ-tubulin dimer and MBP-TRPV1-Ct, in PEM buffer was adjusted to 0.2 mol/L triethanolamine (pH 8.1) buffer for cross-linking with dimethyl suberimidate (DMS, Sigma, 1 mg/mL). The reaction was carried out at 25°C for 1 min to 1 h and stopped by adding Tris–HCl (pH 6.8) to a final concentration of 50 mmol/L. Samples were subjected to SDS-PAGE separation and western-blot analysis.
To perform western-blot analysis, the proteins were separated by SDS-PAGE, and transferred either to a nitrocellulose membrane or PVDF (Millipore, Schwalbach, Germany) by semidry electro blotting. The membranes were blocked with 5% non-fat milk in TBS-T (20 mmol/L Tris, 150 mmol/L NaCl. 0.1% Tween-20) buffer followed by incubation with the respective primary antibody for 1 h at 25°C, washed thrice times with TBS-T buffer. Subsequently, the membranes were incubated with horseradish peroxidase-conjugated secondary antibody for 1 h at 25°C. For the detection of biotinylated-peptides, PVDF membranes containing peptide spots were probed with HRP-conjugated avidin (Sigma–Aldrich, 1 : 1000 dilution). The membranes were washed thoroughly with TBST. The ECL detection system (Amersham Biosciences, Freiburg, Germany) was used for the visualisation of the immunoreactivity.
Co-sedimentation assay with taxol-stabilised microtubules (MT)
Approximately 100 µg of purified αβ-tubulin dimer in a total volume of 100 µL were incubated in modified PEM buffer (20 mmol/L PIPES, pH 6.8, 0.2 mmol/L MgCl2 and 1 mmol/L EGTA supplemented by 1 µmol/L taxol and 5 mmol/L GTP) for 30 min at 37°C, to form MT. After MT formation, 5 µg of purified proteins representing different MBP-fusion proteins were incubated with taxol-stabilised MT for 40 min at 37°C followed by centrifugal separation of pellet (MT) and supernatant (free dimer) at 70 000 g/30 min/37°C. In a similar manner biotinylated-peptides (approximately 1 μg each in 100 μL) were first centrifuged at 25 000 g for 5 min to remove all aggregates. Clear supernatant containing soluble peptides were further used for microtubule co-sedimentation assay. Corresponding pellet and supernatant fractions were further spotted on a PVDF membrane (Amersham) by a dot-blot apparatus (Bio-Rad, Munich, Germany) and analysed for bound peptides. For MT formation under taxol-free conditions, 100 µg of tubulin dimer were used in PEM buffer with 5 mmol/L GTP in the absence of taxol and incubated for 30 min at 37°C.
To carry out overlay experiments, either native tubulin dimer were spotted directly or denatured and SDS-PAGE-separated proteins (tubulin dimer) were transferred from gels to nitrocellulose or PVDF membrane. Membranes were blocked for 1 h with 5% fat-free milk in PBST buffer. Subsequently, the membranes were washed thrice. Membranes were air-dried and incubated with MBP-TRPV1-Ct or MBP alone (protein concentration 0.2 μg/mL, with 5% fat-free milk in PBST buffer) for 1 h at 25°C. In blot overlay experiment with biotinylated-peptides, PEM-S-T (50 mmol/L PIPES; pH 6.8, 1 mmol/L EGTA, 0.2 mmol/L MgCl2, 150 mmol/L NaCl and 0.1% Tween 20) buffer was used. Peptides were used at a concentration of approximately 5 ng/mL. After incubation, the membranes were washed thrice (each time for 10 min) and incubated with 0.1% formaldehyde for 30 min to cross-link the bound proteins. Finally, the membranes were quenched with 100 mmol/L glycine in TBS buffer and processed for western-blot analysis with MBP antibodies or HRP-labelled avidin to detect the bound proteins or peptides respectively.
TRPV1 interacts with soluble tubulin, but neither with soluble actin nor with soluble neurofilaments
Previously, we observed that the C-terminus of TRPV1 binds to tubulin and stabilises microtubules (Goswami et al. 2004). To understand if the C-terminus of TRPV1 can also interact with cytoskeleton components other than tubulin, we performed pull-down experiments with purified soluble actin and enriched soluble neurofilament preparations. The C-terminal cytoplasmic domain of TRPV1 fused with MBP (MBP-TRPV1-Ct) was used as bait. However, we could not observe any significant direct interaction of actin or neurofilaments with the MBP-TRPV1-Ct (Fig. 2a). As activation of TRPV1 results in influx of Ca2+, we tested if actin and neurofilaments interact with MBP-TRPV1-Ct in the presence of high Ca2+. We observed no interaction even in the presence of Ca2+. Under the same conditions a significant amount of soluble tubulin interacts with the MBP-TRPV1-Ct. In agreement with our previous results (Goswami et al. 2004), this interaction was slightly stronger in the presence of Ca2+. In similar experiments, MBP-TRPV1-Ct, but not MBP-LacZ pulls down neuron-specific beta III tubulin as well as other post-translationally modified tubulins, i.e. acetylated tubulin, polyglutamylated tubulin, phosphotubulin (phosphoserine), tyrosinated tubulin and de-tyrosinated tubulin (Fig. 2b). These results indicate that the C-terminus of TRPV1 interacts specifically with the constituents of both dynamic and stable microtubules.
β-tubulin, but not α-tubulin preferentially interacts with TRPV1
Tubulin preparations from brain tissue consist predominantly of αβ-tubulin heterodimers. In order to analyse, which subunit of the tubulin dimer interacts with the C-terminal domain of TRPV1, we performed a cross-linking experiment (Fig. 3a). A mixture of αβ-tubulin dimer and MBP-TRPV1-Ct was cross-linked by using dimethyl suberimidate (DMS), a homobifunctional cross-linking agent, which reacts with amino groups. The cross-linked products of tubulin dimers and MBP-TRPV1-Ct were subsequently analysed by gel electrophoresis and western-blot analysis with the appropriate antibodies. We observed that cross-linking occurred fast and the entire amount of MBP-TRPV1-Ct appeared on the gel as a high-molecular weight complex after only 1 min of cross-linking. All β-tubulin in the reaction mixture also appeared in the same high-molecular weight complex. Also the α-tubulin was found in that complex, but approximately 50% of the α-tubulin did not react. Even after 60 min of reaction, the non-cross-linked α-tubulin population remained at its monomeric molecular weight on the SDS-PAGE. The high-molecular weight complex was not observed when we used purified MBP instead of MBP-TRPV1-Ct for cross-linking experiments with αβ-tubulin dimer (data not shown). From these data we conclude that the MBP-TRPV1-Ct interacts with the tubulin dimer predominantly via β-tubulin.
The C-terminus of TRPV1 interacts with blotted denatured tubulin
Three-dimensional crystal structures and EM pictures reveal that the C-terminal tail of the tubulin dimer does not integrate into the core of the microtubule filaments, but remains outside (Nogales et al. 1999; Lowe et al. 2001). These exposed C-terminal over-hanging regions of both α-tubulin and β-tubulin are strongly negatively charged and unstructured too (Lowe et al. 2001; Nogales 2001). Most of the known microtubule-binding proteins interact with microtubules and tubulins through these regions (see Fig. 4a, see also Discussion).
To analyse if the TRPV1 C-terminus also interacts with tubulin via this acidic unstructured region, we separated purified αβ-tubulin dimer on SDS-PAGE and performed a blot overlay on the denatured tubulin with purified MBP-TRPV1-Ct. We subsequently probed for bound protein by western-blot analysis with MBP antibodies. MBP immunoreactivity was detected at 55 kDa (the position of tubulin) on the membrane, but only if the blot overlay was performed with MBP-TRPV1-Ct and not with MBP alone (Fig. 3b). This suggests that the C-terminus of TRPV1 interacts even with unstructured tubulin polypeptides, possibly at the C-terminal negatively charged overhanging region.
AA 681-730 is sufficient for tubulin binding
To identify the tubulin-binding region in the C-terminus of TRPV1, we undertook a systematic deletion approach. We designed three MBP-fused C-terminal deletion constructs along with the complete C-terminus, namely MBP-TRPV1-Ct-Δ1 (aa 681–800) and MBP-TRPV1-Ct-Δ2 (aa 681–760) as well as MBP-TRPV1-Ct-Δ3 (aa 681–730), each shorter by few amino acids at the C-terminal end respectively (Fig. 1). All these fusion proteins were expressed in E.coli, purified and confirmed by western-blot analysis (data not shown). MBP-pull-down experiments were performed with these deletion-proteins. Pulled-down eluate samples were first analysed by SDS-PAGE and silver staining and then probed for bound tubulin by western-blot analysis.
We observed that all three MBP-TRPV1-Ct deletion proteins pulled down tubulin when used as baits (Fig. 5). As expected, no tubulin was observed in pull-down samples where an unrelated construct (MBP-LacZ) was used as bait (data not shown, a similar experiment is shown in Fig. 2). However, the amount of tubulin pulled down by different deletion-proteins varies, indicating that different regions of the C-terminus of TRPV1 influence the tubulin interaction to a different extent (Fig. 5). Tubulin binding with MBP-TRPV1-Δ1 was slightly higher when compared with MBP-TRPV1-Ct. This indicates that the amino acid sequence 800–838 is less important for the interaction. Deletion of this region may even enhance the tubulin interaction. Similarly, a stronger amount of tubulin binding was observed with MBP-TRPV1-Δ3, which contains amino acids 681–730 of TRPV1. Significantly less tubulin binding was observed with MBP-TRPV1-CtΔ2 (amino acid residues 681–760).
These results indicate that a stretch of 50 amino acids (position 681–730) is sufficient for the interaction with tubulin. This also suggests that the sequences amino acids 731–760 and 800–838 of TRPV1 reduce tubulin binding, while the sequence residues 760–800 enhance it.
Two small basic sequences located in the C-terminal cytoplasmic domain of TRPV1 have potentiality to modulate the interaction
Many tubulin-binding proteins and microtubule-binding proteins contain more than one short basic-repeat sequences which mediates the interaction with tubulin and/or microtubules (see Discussion). To explore, if the C-terminal cytoplasmic domain of TRPV1 also contains basic-sequence stretches for tubulin binding, the theoretical pI values of the entire C-terminus of TRPV1 as well as of short sequence stretches were calculated (see Fig. 1). Two short sequence stretches that contain basic amino-acid residues and thus have a higher pI were identified at residues 710–730 (calculated pI; 11.17) and 770–797 (calculated pI; 12.6). These two sequences contain a number of positively charged amino acids, but are devoid of any negative charges. However, the two sequences are flanked by sequence stretches that consist of negatively charged amino acids. Therefore, these motifs (short basic sequences flanked by acidic amino acids) may act as tubulin-binding structures.
Interestingly, a helical plot of these two short sequences shows that all the basic amino acids are located on one side of the α-helix (discussed later, see Fig. 4b). This may indicate that these basic amino acids are engaged in the interaction with negatively charged sequences, either of neighbouring sequences of the same polypeptide (with negatively charged residues) or of sequences from other interacting proteins, like the C-terminal overhanging acidic region of tubulin. The latter explanation is supported by our tubulin pull-down experiments. Beside the MBP-TRPV1-Ct, only MBP-TRPV1-CtΔ1 and MBP-TRPV1-CtΔ3, but not MBP-TRPV1-CtΔ2 binds tubulin.
TRPV1-Ct amino acid sequences 710–730 and 770–797 influence TRPV1/tubulin interaction
To further narrow down the interacting regions, which mediate the tubulin interaction, the C-terminus of TRPV1 was subdivided into five short segments. These short fragments expressed as N-terminal MBP fusion proteins and referred as MBP-TRPV1-Ct-fr1 to MBP-TRPV1-Ct-fr5 (see Fig. 1). Fragments 1, 3 and 5 contain negatively charged amino acids (with low pI), while fragments 2 and 4 represent basic stretch 1 and basic stretch 2 respectively. Two more fragments were prepared, in which the basic stretch 2 has an acidic stretch to its right side (named MBP-TRPV1-Ct-Fr6, see Fig. 1) or with acidic sequences on both sides (referred to as MBP-TRPV1-Ct-Fr7, see Fig. 1). One further fragment was prepared which contained both the basic sequence stretches connected by the acidic stretch in between. This fragment is referred to as MBP-TRPV1-Ctfr8 (see Fig. 1).
All these deletion constructs and fragments of the C-terminus of TRPV1 were created as fusion proteins with an N-terminal MBP and expressed in E.coli. The whole set covers most of all possible combinations, i.e. either two short basic sequences are present in combination, or alone, or are excluded completely. The expressed fusion proteins were used for pull-down assays and also for microtubule-binding assays. These fusion proteins were additionally tested for their binding to SDS-PAGE-separated denatured tubulin dimer or to native tubulin dimers spotted on membranes in a blot-overlay experiment (data not shown).
In the MBP-pull-down assay, it was observed that MBP-TRPV1-Ct-fr 2 and MBP-TRPV1-Ct-fr 4 bind tubulin strongly (Fig. 6). Notably, these two fragments represent basic stretch sequence 1 and 2 respectively. In contrast, MBP-TRPV1-Ct-fr1 and MBP-TRPV1-Ct-fr 5 reveal no binding (Fig. 6). We observed some weak binding for MBP-TRPV1-Ct-fr3. Comparison of tubulin-binding and microtubule-binding to MBP-TRPV1-Ct-fr6, MBP-TRPV1-Ct-fr7 and MBP-TRPV1-Ct-fr8 reveals that MBP-TRPV1-Ct-fr7 has the lowest affinity to both soluble tubulin and polymerised microtubules while MBP-TRPV1-Ct-fr8 showed the strongest binding for both. MBP-TRPV1-Ct-fr6 also reveals significant binding to both tubulin and microtubules (Fig. 7). Similar results were obtained, when these fragments were used in a blot-overlay experiment with SDS-PAGE-separated tubulin (data not show). These results support a positive modulation of tubulin binding by the presence of the two short basic amino acid stretches.
Biotinylated-peptides corresponding to these two short basic stretches bind to soluble tubulin and co-sediment with the polymerised microtubules
To confirm these findings by a different method, we synthesised biotinylated-peptides (each peptides containing 21 amino acid, amino acid 710–730 and 770–790). Therefore, these two peptides represent the basic amino acid stretches. We observed that the two peptides co-sediment with taxol-stabilised microtubules (Fig. 8a). Avidin–agarose coupled to either of these two peptides pulled down a significant part of the purified soluble tubulin (Fig. 8b). In contrast, avidin–agarose beads alone did not pull down any tubulin. We observed that these two peptides can detect SDS-PAGE-separated denatured tubulin on membranes in the blot overlay assay in nanomolar concentration (Fig. 8c). This not only confirms that these two peptides can interact with tubulin and microtubules specifically, but also shows that the interaction does not need a native tubulin structure.
The C-terminal overhanging region of tubulin is important for TRPV1-Ct binding
As the C-terminal overhanging regions of both α-tubulin and β-tubulin are very acidic and involved in the interaction with most of the microtubule-binding proteins (see Discussion), we tested if this is the region where the C-terminal cytoplasmic domain of TRPV1 binds. To test this, we analysed if subtilisin-digested tubulin (see Fig. S1) binds with MBP-TRPV1-Ct. We performed MBP-pull-down assay with control tubulin and subtilisin-digested tubulin using MBP-TRPV1-Ct, and analysed the eluates by SDS-PAGE. Although undigested control tubulin binds significantly to the MBP-TRPV1-Ct, we could not detect any binding of subtilisin-digested tubulin with MBP-TRPV1-Ct (Fig. 9a). To confirm this result by a more sensitive method, we used mouse monoclonal anti-β-tubulin antibody, clone D10 (Santa Cruz). This antibody detects control tubulin as well as subtilisin-digested tubulin, both in SDS-PAGE-separated denatured and in native form (Fig. 9b). By western-blot analysis, we could detect a significant binding of control tubulin, but we failed to find any binding of subtilisin-digested tubulin (Fig. 9a). This indicates that the C-terminal acidic region of tubulin is important for TRPV1-Ct binding.
Recently, we demonstrated that the C-terminus of TRPV1 not only interacts but also stabilises microtubules in vitro (Goswami et al. 2004), and when over-expressed, this results in the formation of more stable and bundled microtubules in vivo (Goswami et al. 2006). In a reverse manner, activation of TRPV1 results in rapid disassembly of dynamic microtubules (Goswami et al. 2006). This demonstrates that TRPV1 is physically linked to microtbules, and also suggests that the microtubule cytoskeleton is a downstream effector of TRPV1 activation. However, the exact tubulin-binding sequence was not determined. In this study, we characterised the TRPV1/tubulin interaction and also demonstrate that the C-terminus of TRPV1 neither interacts with soluble actin nor with soluble neurofilaments, but specifically interacts with the components of microtubule cytoskeleton.
We demonstrate that the C-terminus of TRPV1 preferably interacts with β-tubulin rather than α-tubulin. In a similar manner, two groups independently reported that β-tubulin interacts with members of TRPC channels (Bollimuntha et al. 2005; Goel et al. 2005). Microtubule plus ends terminate their protofilaments with a β-tubulin at the end. The higher preference of the TRPV1-Ct for β-tubulin and the ability to stabilise microtubules therefore indicates that the observed interaction may exert its effect at the microtubule plus end rather than at the minus end. This accords well with a recent observation that Xenopus TRPN1, another member of the TRP channel super family is localised at the plus ends of the microtubule-based cilia structures (Shin et al. 2005).
In contrast to the C-terminus, the N-terminus of TRPV1 neither interacts with soluble tubulin nor with polymerised microtubules (Goswami et al. 2004). Similarly, the N-terminus of TRPV1 fails to modify the properties of microtubules in vitro (Goswami et al. 2004). The N-terminal domain of TRPV1 does not form any specific high-molecular weight complex when cross-linked with tubulin dimers (data not shown). From all these observations it seems that the N-terminus of TRPV1 is not involved with the tubulin interaction.
Because of the fact that both α-tubulin and β-tubulin are coded by several genes, and are subjected to different post-translational modifications, much heterogeneity among the tubulin monomers as well as dimers exists. The C-terminal sequence of both α-tubulin and β-tubulin is unstructured and contains highly negatively charged residues (Nogales et al. 1998; Lowe et al. 2001). Despite great heterogeneity within the C-terminal tails of different tubulins, the negatively charged residues are highly conserved in both α-tubulin and β-tubulin (Fig. 4a). Due to the presence of these highly negatively charged residues, the calculated and measured isoelectric points of different α-tubulin and β-tubulin monomers are very low and range from 4.8 to 5.2 (Towbin et al. 2001; Stracke et al. 2002; Verdier-Pinard et al. 2003). The measured pI around 4.2 for the αβ-tubulin dimer (the predominant form in which both monomers exist) is even lower than the pI of the monomers alone (Stracke et al. 2002).
The majority of the microtubule-binding proteins interact with microtubules and tubulin via these acidic C-terminal tail sequences. Our study points at two small sequence stretches within the C-terminus of TRPV1 (amino acid 710–730 for stretch 1 and amino acid 770–797 for stretch 2, respectively), each of which can modulate the tubulin interaction. These sequences have a basic pI value of 11.17 and 12.6, respectively. The binding capability of smaller- and deletion-fragments of TRPV1-Ct correlates well with two factors: the presence or absence of these two short basic stretches and the overall pI of the protein. This is in agreement with many observations showing that the majority of tubulin interacting proteins contain short imperfect repeat sequences composed of basic amino acids. For example, isoforms of microtubule-binding protein tau contain three to four 18-amino acid long imperfect repeats separated by 13–14-amino acid inter-repeat sequences (Himmler et al. 1989; Goode et al. 2000). Doublecortin (Dcx), another microtubule-binding protein, contains two repeat sequences with high pI (9.7 and 9.9) by which the microtubule interaction is mediated (Taylor et al. 2000). Several microtubule-associated proteins (MAPs) also contain repeat sequences with basic pI important for microtubule binding (Lewis et al. 1988; Noble et al. 1989; Al-Bassam et al. 2002). Microtubule interaction of the kinesin motor domains is attributed to the basic amino acids located at the surface (Woehlke et al. 1997). A monomeric kinesin (KIF1A) has much higher affinity for microtubules and an enhanced processivity along the microtubules due to the presence of six extra lysine residues within the ‘K-loop’ of the motor domain (Okada and Hirokawa 1999). The microtubule-binding domain of Stu2p contains highly basic amino acids with a predicted pI of 10.7 (Wang and Huffaker 1997). Notably, the microtubule-binding domain of Stu2p contains two imperfect repeat sequences. A short peptide (KKKKKSKTKCVIM) with multiple lysine residues representing the C-terminus of K-Ras has been reported to interact with microtubules (Thissen et al. 1997; Chen et al. 2000).
Our conclusion that these two short basic sequences are important for tubulin binding is furter strengthened by the fact that positively charged amino acids located within these two stretches are distributed on one side of a putative helix (Fig. 4). More interestingly, the stretch 1 contains several hydrophobic amino acids distributed on this putative helix just opposite to the charged basic amino acids (Fig. 4). Therefore, it is possible that this portion of the C-terminus is partially embedded in the plasma membrane while the other exposed surface is involved in tubulin binding. It is important to mention that the α-tubulin-binding site of mGluR7 also forms a putative helix with positively charged residues located towards one direction (Saugstad et al. 2002).
In absence of X-ray or NMR data, the structure of the C-terminal cytoplasmic domain of TRPV1 has been modelled by two groups (Vlachováet al. 2003; García-Sanz et al. 2004). According to the model proposed by García-Sanz et al., the first stretch sequence is exposed and therefore easily accessible to interact (García-Sanz et al. 2004). In this study, we did not measure the ‘Kd’ value of the interaction, but the binding of two different biotinylated-peptides to the tubulin at nanomolar concentration and observed interaction between MBP-TRPV1-Ct with tubulin even at the high salt condition (0.5 mol/L, data not shown) indicates for a stable and strong binding. Though our results strongly suggest that the C-terminal cytoplasmic domain of TRPV1 contains two tubulin-binding sites, from our study, it is not clear how many tubulin (dimer/monomer?) binds per receptor. It is possible that these two sequences bind tubulin independent of each other and there is a cooperative binding. It is also possible that in native condition, the tubulin binding by the two short stretches are masked by the presence of short neighbouring sequences or other interacting proteins that contain negative charges. From our study it is also not clear whether all the positively charged residues present within each stretch are important for tubulin binding. But our experiments show that subtilisin-digested tubulin does not bind to the MBP-TRPV1-Ct. As subtilisin-digested tubulin retain its core structure but loose the C-terminal over-hanging acidic region (see Fig. S1), it indicates that the TRPV1-Ct-binding ability of tubulin is most likely located within the C-terminal overhanging regions.
The C-terminal sequence of TRPV1 is shown to be involved in several other interactions and functions and may represent a ‘hot spot’ for TRPV1 regulation (Fig. 10). For example, the region comprising amino acid residues 684–721 has been shown to be important for tetramerisation (García-Sanz et al. 2004). Amino acid 761 has been shown to be important for the capsaicin-sensitivity (Jung et al. 2002). Vlachováet al. demonstrated that the C-terminal tail of TRPV1 carries structural determinants rendering the receptor sensitive to heat and capsaicin. They could demonstrate that deletion of the distal 72 amino acids results in a decline of the capsaicin-, pH-, and heat-sensitivity of the receptor (Vlachováet al. 2003). Moreover, the C-terminal tail contains phosphorylation sites used by protein kinase C (Numazaki et al. 2002) and Ca2+/calmodulin-dependent protein kinase II. Phosphorylation of these sites affects the TRPV1 desensitisation in the presence of Ca2+ (Jung et al. 2004). Two serine residues (at position 774 and 820) located within the C-terminus of TRPV1 are phosphorylated by PKA in vitro (Bhave et al. 2003; Mohapatra and Nau 2003). Furthermore, Numazaki et al. detected a short sequence within the C-terminal tail that is critical for receptor desensitisation (Numazaki et al. 2003). They could also prove that this sequence contains a binding site for calmodulin. Eferin, an EF-hands-containing Rab11/25-interacting protein was reported to bind at the C-terminal cytoplasmic domain of TRPV1 (Lee 2005). Furthermore, the C-terminal sequence of TRPV1 contains a PIP2 binding site, which negatively regulates the ion channel activity (Prescott and Julius 2003). Remarkably tubulin-binding sites detected in this study are located close to the regions that are important for receptor tetramerisation, PIP2 binding or PKC phosphorylation (Fig. 10). Hence it is possible that tubulin binding to the C-terminus of TRPV1 may alters these interactions.
The two short basic stretches present in TRPV1 are highly conserved in most of the reported mammalian orthologous sequences (Fig. 11). Among the members of the TRPV subfamily, similar sequences including positively charged amino-acids within the first sequence stretch are conserved to a certain extent, especially among TRPV1, TRPV2, TRPV4, and TRPV3 (not so much in TRPV5 and TRPV6) (Fig. 11). The second sequence stretch described above is much less conserved (Fig. 11). In agreement with the significance of these short sequences, a small sequence of the C-terminal cytoplasmic domain of TRPV4 has been shown to interact with MAP7, a microtubule binding protein (Suzuki et al. 2003). Additionally, TRPV4 has been shown to be important in mechanical hyperalgesia induced by taxol, a microtubule cytoskeleton-regulating drug (Alessandri-Haber et al. 2004). In agreement with that notion, the involvement of an intact microtubule cytoskeleton in the second messenger signalling for inflammatory pain has also been reported (Bhave and Gereau 2003; Dina et al. 2003).
In summary, our work identifies two short basic-amino acid stretches within the C-terminus of TRPV1 that interacts with tubulin. These stretches represent novel tubulin-binding motifs that are also conserved to a certain extent in some other members of the TRPV subfamily.
We thank Mark Hartman, Linda Stewani and Shu Liu for preparing the TRPV1 constructs. We thank Oliver Bogen for providing enriched neurofilament preparation. We are thankful to Dr Linda Amos (Cambridge, UK) for providing subtilisin-digested tubulin and related control reagents. We acknowledge the kind support provided by Prof H.H. Ropers. Financial support by Max Plank Institute of Molecular Genetics (Berlin), BMBF, Deutsche Forschungsgemeinschaft, Sfb 515, and Fonds der Chemischen Industrie is acknowledged.