Identification and characterization of a Ca2+-sensitive interaction of the vanilloid receptor TRPV1 with tubulin


Address correspondence and reprint requests to F. Hucho, Institut für Chemie-Biochemie, Freie Universität Berlin, Thielallee 63, 14195 Berlin, Germany. E-mail: or M. Dreger, University Laboratory of Physiology, Parks Road, Oxford OX1 3PT, UK.


The vanilloid receptor TRPV1 plays a well-established functional role in the detection of a range of chemical and thermal noxious stimuli, such as those associated with tissue inflammation and the resulting pain. TRPV1 activation results in membrane depolarization, but may also trigger intracellular Ca2+-signalling events. In a proteomic screen for proteins associated with the C-terminal sequence of TRPV1, we identified β-tubulin as a specific TRPV1-interacting protein. We demonstrate that the TRPV1 C-terminal tail is capable of binding tubulin dimers, as well as of binding polymerized microtubules. The interaction is Ca2+-sensitive, and affects microtubule properties, such as microtubule sensitivity towards low temperatures and nocodazole. Our data thus provide compelling evidence for the interaction of TRPV1 with the cytoskeleton. The Ca2+-sensitivity of this interaction suggests that the microtubule cytoskeleton at the cell membrane may be a downstream effector of TRPV1 activation.

Abbreviations used

bovine serum albumin




2,5-dihydroxybenzoic acid


fragile histidine triad protein


isopropyl thiogalactoside


matrix-assisted laser desorption ionization-mass spectrometry


maltose binding protein


TRPV1 C-terminal cytoplasmic portion MBP fusion protein


TRPV1 N-terminal cytoplasmic portion MBP fusion protein






amylose resin


nerve growth factor


phosphate-buffered saline


post-source decay


room temperature


sodium dodecyl sulphate polyacrylamide gel electrophoresis


transient receptor potential

There is growing evidence that ion channel proteins often occur in the cell membrane as parts of multiprotein complexes. These complexes, such as the NMDA receptor complex (Husi et al. 2000), are termed signalling complexes, because components of effector and modulator systems of ion channel function are concentrated by protein–protein interactions in close proximity to the ion channel proteins. Based on this concept of signalling complexes, considerable insight into signalling events triggered by ion channel function is expected to be gained from the identification of ion channel-interacting proteins.

The vanilloid receptor TRPV1 is of specific importance in thermosensation and detection of noxious stimuli associated with tissue inflammation (Tominaga et al. 1998; Caterina et al. 2000; Davis et al. 2000; Caterina and Julius 2001). This ion channel is a member of the transient receptor potential (TRP) family of non-selective ion channels and can be activated by noxious heat, low pH, capsaicin, resiniferatoxin and endogenous lipid-derived mediators, such as anandamide and N-arachidonoyl-dopamine (NADA) (Huang et al. 2002), partly in a synergistic manner. The phosphoinositide PIP2 (Prescott and Julius 2003) and direct phosphorylation of intracellular sites of TRPV1 (Bhave et al. 2002, 2003; Numazaki et al. 2002; Jung et al. 2003; Mohapatra and Nau 2003) contribute to the modulation of ion channel function. It is believed that TRPV1 channel activation leads to two lines of downstream events: membrane depolarization and Ca2+ influx. The Ca2+ influx itself modulates channel function by triggering channel desensitization (Koplas et al. 1997; Tominaga et al. 1998). The molecular basis of this, however, is not yet understood, though there is evidence that the phosphorylation status of TRPV1 and interaction with calmodulin (CAM) might be involved (Numazaki et al. 2003; Jung et al. 2004; Rosenbaum et al. 2004). In a hypothesis-driven approach based on the notion that nerve growth factor (NGF) signalling affects TRPV1 function, the high-affinity NGF receptor TrkA and phospholipase Cγ have been identified as candidate components of a TRPV1-signalling complex in vitro (Chuang et al. 2001).

Though no familiar protein domains have been detected within the C-terminal region of TRPV1, many functional features have been assigned to it, such as involvement in receptor desensitization (Vlachova et al. 2003), PIP2 binding (Prescott and Julius 2003), CAM binding (Numazaki et al. 2003), requirements for responses to heat (Vlachova et al. 2003) and capsaicin (Jung et al. 2002).

Here we report results based on a proteomic screen for proteins that interact with the C-terminal cytoplasmic domain of TRPV1. We demonstrate that tubulin dimers specifically bind to the TRPV1 C-terminus in vitro. This interaction also occurs with polymerized microtubules and is affected by the concentration of free Ca2+. This interaction exerts a stabilizing effect on microtubules (MT) when exposed to conditions that favour depolymerization. We conclude that Ca2+ signalling through TRPV1 may affect the structure of the MT network at the plasma membrane.

Materials and methods

Antibodies and reagents

The mouse monoclonal anti MBP antibody was purchased from New England Biolabs (Beverly, MD, USA), the rabbit polyclonal anti-N-terminal TRPV1 antibody and the respective blocking peptide (sequence M1EQRASLDSEESESPPQENSC21 which correspond to the first 21 amino acid residues of TRPV1) were from Affinity Bioreagents (Golden, CO, USA) and from Alexis Biochemicals (San Diego, CA, USA). The goat polyclonal anti-C-terminal TRPV1 antibody was from Santa Cruz Biotechnology (Santa Cruz, CA, USA). The sequence of the supplied blocking peptide for the goat polyconal anti-C-terminal TRPV1 antibody was found to contain the sequence stretch PEDAEVFKDSMV by matrix-assisted laser desorption ionization-mass spectrometry (MALDI-MS), which corresponds to the 823–834 amino acid residues of TRPV1. The mouse monoclonal anti α-tubulin antibody was from Oncogene (San Diego, CA, USA). The mouse monoclonal anti β-tubulin antibodies were from Boehringer (Mannheim, Germany), and also from Santa Cruz Biotechnolgy. Rat monoclonal anti tubulin antibody YL1/2 was purchased from Abcam Ltd (Cambridge, UK). The mouse monoclonal anti actin antibody was from Oncogene. Alexa fluor 594 phalloidin was from Molecular Probes (Eugene, OR, USA). Nocodazole and taxol were purchased from Sigma-Aldrich (Taufkirchen, Germany). EZ-Link Sulfo-NHS-LC-Biotin was purchased from Molecular Probes.

Expression and purification of TRPV1 fusion proteins

cDNA fragments corresponding to amino acid residues 1–431 (N-terminal cytoplasmic portion) of rat TRPV1 were amplified by PCR using primers 5′-GCGCGAATTCATGGAACAACGGGCTAG-3′ and 5′-GCGCTCAGATTACTTGACAAATCTG-3′. A cDNA fragment corresponding to amino acid residues 681–838 of rat TRPV1 (C-terminal cytoplasmic portion) was amplified by PCR using primers 5′-GCGCGAATTCATGGGTGAGACCGTCAA-3′ and 5′-GCGCTCTAGATTATTTCCCCTGGGAC-3′. Amplified fragments were subcloned into the EcoR1 and XbaI restriction sites of the pMAL-c2x vector (New England Biolabs). A stop codon was introduced into the pMAL-c2x vector in between EcoRI and HindIII sites using a linker made by 5′-AATTCGGTACCTGAA-3′ and 5′-AGCTTTCAGGTACCG-3′ to express only maltose-binding protein (MBP). All expression constructs were verified by automated nucleotide sequencing. Escherichia coli strain BL21DE3 was transformed by heat shock with plasmid coding for the TRPV1 C-terminal cytoplasmic portion MBP fusion protein (MBP-TRPV1-Ct), the TRPV1 N-terminal cytoplasmic portion MBP fusion protein (MBP-TRPV1-Nt) and MBP only. E. coli cells were induced to express the proteins by isopropyl thiogalactoside (IPTG) for 2 h. Thereafter, the cells were lysed by freezing and thawing cycles in lysis buffer (20 mm Tris–HCl, pH 7.4, 150 mm NaCl, 0.2 g/mL sucrose, lysozyme, benzonase and protease inhibitor cocktail). The lysed extracts were cleared by spinning at 100 000 g in a TFT 45 rotor for 2 h. The cleared lysate was applied to the amylose resin and washed throroughly. Bound protein was eluted with 10 mm maltose in elution buffer (50 mm PIPES, pH 6.8, 100 mm NaCl, 1 mm EGTA and 0.2 mm MgCl2). All protein determinations were performed according to Bradford (1976).

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, were followed by chromatography on phosphocellulose.

Preparation of high speed spinal cord extract

Rat spinal cord was homogenized in 20 mm HEPES (pH 7.4), 1 mm EGTA and 320 mm sucrose and protease inhibitor cocktail (Roche Molecular Biochemicals, Indianapolis, IN, USA). After homogenization, the extract was cleared by centrifugation at 48 000 g for 30 min at 4°C, followed by solubilization with 1% (w/v) Tween-20 for 2 h at room temperature (RT; 20°C). The solubilized extract was centrifuged at 100 000 g for 2 h at 25°C in fixed angle rotor TFT65.13. The top layers that contained myelin and the bottom layers that contained insoluble material were discarded. Myelin-free cleared supernatant from the middle of the gradient was used for the assay.

Pulldown assay

MBP alone, MBP-TRPV1-Ct and MBP-TRPV1-Nt were expressed in E. coli and the cleared cell lysates were applied to the amylose resin (NEB), and incubated for 1 h at RT followed by washing. Approximately 1 mL of amylose resin with the bound fusion protein was incubated with 10 mL of cleared spinal cord extract (1 mg/mL protein) overnight either in the presence or absence of Ca2+ (5 mm). This was followed by washing three times with 10 mL each and constant buffer conditions. The proteins were eluted by 10 mm maltose. Eluted samples were concentrated by TCA precipitation and analysed by 10% sodium dodecyl sulphate polyacrylamide gel electrophoresis (SDS–PAGE) according to Laemmli (1970).

In experiments aimed at determining if there is a direct interaction between tubulin and TRPV1, 40 µL of amylose beads bound to MBP-TRPV1-Ct, MBP-TRPV1-Nt or MBP alone were incubated with 50 µg of αβ-tubulin under buffer conditions as given in the figure legends.

Biotinylation of tubulin

Purified tubulin (1 mg/mL) in 1 mL PEM buffer (50 mm pH 6.8, 1 mm EGTA, 0.2 mm MgCl2) was labelled with 1 mg/mL of EZ-Link Sulfo-NHS-LC Biotin at 4°C for 2 h. The labelling reaction was terminated by adding Tris–HCl pH. 7.4 to the final concentration of 50 mm. Unreacted biotinylation reagent was separated from biotinylated tubulin by gel filtration.


The proteins of interest were digested in gel according to standard procedures (Shevchenko et al. 1996). Measurements were performed using 2,5-dihydroxybenzoic acid (DHB) as matrix substance. A Bruker Reflex (Bruker Daltonics, Bremen, Germany) mass spectrometer was used to acquire peptide mass fingerprint spectra as well as fragment ion spectra obtained from post-source decay (PSD) of selected precursor ions. PSD spectra were assembled using the FAST method (Bruker Daltonics). The search engines ProFound and PepFrag as available at were used to match peptide mass fingerprints and fragment ion patterns to NCB Inr database entries.

Immunoprecipitation assay and western blot analysis

F11 cells were transfected with the construct coding for full-length TRPV1 by lipofectamine (Invitrogen, Carlsbad, CA, USA). Two days after transfection, cells were extracted with 1% (w/v) dodecyl maltoside at RT for 2 h. The extract was cleared by centrifugation at 28 000 g for 30 min. The extraction buffer contained 20 mm PIPES pH 6.8, 1 mm EGTA, 1 mm CaCl2, 0.2 mm MgCl2, 1% (w/v) dodecyl maltoside and 150 mm NaCl along with the protease inhibitor cocktail and benzonase (Merck, Rahway, NJ, USA). Then, 600 µg of cleared cell extract at a protein concentration of 3 mg/mL were incubated with 2 µg of antibodies. Immunocomplexes were isolated with protein G sepharose (20 µL) and separated by SDS–PAGE. For immunoprecipitation of TRPV1 from spinal cord, 1 g (wet weight) of spinal cord tissue from rat were homogenized in the same buffer as used in the homogenization of F11 cells, followed by extraction with 1% (w/v) dodecyl maltoside and clearance of the extract by centrifugation. Then, 6 mL of cleared extract at a protein concentration of roughly 1 mg/mL were incubated with 20 µg of the N-terminus-specific anti-TRPV1 antibody, and immunocomplexes were collected with 25 µL protein G sepharose. For competition experiments using the blocking peptide, a roughly 300-fold molar excess of peptide over antibody was used. The proteins were separated by SDS–PAGE, and transferred to a nitrocellulose membrane by electroblotting. For western blot analysis, the membranes were blocked with 5% non-fat milk in TBS-T (20 mm Tris, 150 mm NaCl. 0.1% Tween-20) buffer followed by probing with the respective primary antibody, and horseradish peroxidase-conjugated antibodies were used as secondary antibodies. The ECL detection system (Amersham Biosciences, Freiburg, Germany) was used for the visualization of the immunoreactivity.

Co-sedimentation assay with taxol-stabilized MTs

A total of 140 µg of purified αβ-tubulin dimer in a total volume of 90 µL were incubated in modified PEM Buffer (20 mm PIPES, pH 6.8, 0.2 mm MgCl2 and 1 mm EGTA supplemented by 1 µm taxol, 5 mm GTP and 1 mm ATP) for 30 min at 37°C, to form MT. After MT formation, 5 µg of purified MBP-TRPV1-Ct, MBP-TRPV1-Nt and MBP were incubated with taxol-stabilized MT for 40 min at RT either in the presence or absence of 1 mm CaCl2, followed by centrifugal separation of pellet (MT) and supernatant (free dimer) at 70 000 g/30 min/35°C.

For MT formation under taxol-free conditions, 40 µg of tubulin dimer were used in PEM buffer with 1 mm CaCl2 (optional), 5 mm GTP, 1 mm ATP in the absence of taxol and incubated for 30 min at RT.

For cold-induced depolymerization of MT, first the MTs were allowed to form in PEM-S buffer (PEM buffer with 0.1 m NaCl) without taxol as described above. Ca2+ and MBP-TRPV1-Ct, or MBP-TRPV1-Nt, were added along with the tubulin dimer during the first cycle of polymerization. Polymerized MT (pellets) were isolated by centrifugation, resuspended in ice-cold PEM-S buffer by repeated pipet aspiration, and kept on ice for another 30 min. This was followed by the centrifugal separation of dimers (supernatant) and polymer MT (pellet) at 70 000 g (30 min, 4°C).


F11 cells were grown and transfected on glass coverslips. Two days after transfection, the cells were fixed with 2% paraformaldehyde for 10 min, permeabilized with 0.4% Triton X-100 in phosphate-buffered saline (PBS) for 10 min, followed by incubation with 100 mm glycin. The cells were blocked with 5% normal goat serum or bovine serum albumin (BSA). They were probed with the anti-N-terminal TRPV1 antibody (Affinity Bioreagents) at a 1 : 1000 dilution and rat monoclonal tubulin antibody YL1/2 (Abcam Ltd) at a 1 : 750 dilution. Anti-rat Alexa fluor 594-conjugated secondary and Cy2-labelled anti-rabbit secondary antibody were used to visualize the tubulin and TRPV1, respectively. Alexa Fluor 594 phalloidin (Molecular Probes) was used for visualizing actin fibres. The coverslips were mounted on glass slides with fluromount G. Images were taken by a confocal microscope with a 63× objective.


Construction and expression of MBP-TRPV1-Ct

The cytoplasmically located N- and C termini of TRPV1 are believed to play a role in protein–protein interactions. Recent studies on the proteomic screening for interaction partners of ligand-gated ion channels led us to search systematically for TRPV1-C terminus-interacting proteins by pulldown assays. As a bait, we used the cytoplasmic C-terminal portion of TRPV1 (amino acid residues 681–838) fused to the MBP, with the MBP at the fusion protein's N-terminus. The MBP portion served to immobilize the fusion protein to an amylose resin. The fusion protein was expressed in E. coli. The MBP-TRPV1-Ct appeared on SDS–PAGE as a single protein band at ∼65 kDa (Fig. S1a) as assessed using an anti C-terminal TRPV1 or anti-MBP western blot. Additional major degradation products or incompletely translated protein species were not observed (Fig. S1a). The bulk of fusion protein was readily soluble under native extraction conditions.

Detection of Ca2+-sensitive interactions of the TRPV1 C-terminal fragment

Ca2+ plays an important role in the desensitization of the TRPV1, and Ca2+ influx through TRPV1 may trigger intracellular signalling events. Assuming that at least some of the protein–protein interactions of the TRPV1 C-terminus might be Ca2+-sensitive, we conducted pulldown assays both in the presence and absence of Ca2+. The MBP-TRPV1-Ct fusion protein was for that purpose bound to an amylose resin, incubated with protein extracts, and the fusion protein along with interacting proteins was eluted from the resin with maltose. The soluble eluate was then precipitated with TCA and subjected to SDS–PAGE. We used extracts that predominantly contained soluble proteins from rat or pig spinal cord, or from the DRG-derived cell line F11, as sources for potential MBP-TRPV1-Ct-interacting proteins. In the presence of Ca2+, a number of spinal cord proteins were observed to be pulled down with MBP-TRPV1-Ct, while some of these proteins were not pulled down to the same extent in the absence of Ca2+ (Fig. 1a). None of these potential interacting proteins was observed in the absence of fusion protein in the fractions eluted from only amylose resin either in presence or in the absence of Ca2+, or when MBP alone was used as a bait (Fig. 1a). These observations were reproduced four times in independent experiments.

Figure 1.

Identification of α- and β- tubulin as a C-terminal of TRPV1-interacting partner. (a) MBP-TRPV1-Ct (indicated by arrow) interacts with a 55 kDa protein (indicated by asterisk). MBP-TRPV1-Ct immobilised on amylose beads (lane 2 and 3) or MBP alone immobilised on amylose beads (lane 4 and 5) were incubated with spinal cord extract over night in the presence of 5 mm Ca2+ or absence of Ca2+. Bound proteins were eluted with 10 mm maltose, precipitated with trichloroacetic acid, resuspended in SDS–PAGE sample buffer, and subjected to 10% SDS–PAGE. Proteins were visualised by silver stain, and subsequently subjected to in-gel digestion and MALDI-MS analysis. Lane 1, spinal cord extract, lane 2, pulldown eluate of proteins bound in the presence of Ca2+, lane 3, pulldown eluate in the absence of Ca2+, lanes 4 and 5 represent negative controls in the presence or absence of Ca2+, showing non-specific binding to MBP alone immobilized on amylose beads. (b) Western blot analysis reveals a Ca2+-sensitive interaction between both α- and β-tubulin with MBP-TRPV1-Ct. The proteins separated by SDS–AGE and transferred to nitrocellulose membranes were probed with monoclonal antibodies against α-tubulin (left panel) and β-tubulin (right panel) of spinal cord extract (lane 1), proteins bound to MBP-TRPV1-Ct in the presence of Ca2+ (lane 2) or in the absence of Ca2+ (lane 3), proteins bound to MBP alone with Ca2+ (lane 4) and without Ca2+ (lane 5).

By mass spectrometry, a protein pulled down by MBP-TRPV1-Ct in a Ca2+-sensitive manner, corresponding to a 55 kDa protein band observable on the gel, was identified as tubulin β chain 15 from rat (Fig. S2; a total of 16 of peptides covering 39% of the full protein sequence matched NCBI nr A25113). This result was confirmed by western blot analysis (Fig. 1b). Monoclonal antibodies specific for either α- or β-tubulin used for the western blot analysis revealed also the presence of α-tubulin. Both tubulins bound to the MBP-TRPV1-Ct predominantly in the presence of Ca2+, and α-tubulin was virtually only detected when the pulldown was conducted in the presence of Ca2+ (Fig. 1b). The pulldown of the tubulins by the MBP-TRPV1-Ct was observed from pig spinal cord extracts as well as from F11 cell extracts (data not shown).

The identification of other potential TRPV1-interacting proteins is ongoing. For the present study, we focussed on the characterization of the TRPV1–tubulin interaction.

Tubulin, but not actin, co-precipitates with full-length TRPV1 in anti-TRPV1 immunoprecipitation

To confirm the interaction between TRPV1 and tubulin by an independent method, we expressed full-length TRPV1 in F11 cells, and performed immunoprecipitation experiments. This cell line, due to its origin as a fusion cell line from mouse neuroblastoma cells and rat DRG neurones, is a favourable cellular system to mimick the TRPV1 molecular environment in the spinal cord (Jahnel et al. 2001). β-Tubulin co-precipitated with TRPV1-specific antibodies, both when N-terminus-specific anti-TRPV1 antibodies or C-terminus-specific anti-TRPV1 antibodies were used for immunoprecipitation (Fig. 2a, see also Fig. S3 for antibody specificity). Co-precipitation of tubulin from the same extract was not observed with non-specific antibodies, suggesting that the TRPV1 and tubulin interaction is specific. The anti-N-terminal TRPV1 antibody did not precipitate the tubulins from non-transfected cells (Fig. 2b), confirming further the tubulin–TRPV1 interaction observed.

Figure 2.

Tubulin can be co-immunoprecipitated with TRPV1. (a) Co-immunoprecipitation of β-tubulin and TRPV1 from TRPV1-transfected F11 cells. From extracts of TRPV1-transfected F11 cells (lane 1), proteins were immunoprecipitated by the anti-N-terminal TRPV1 antibody (lane 2), anti C-terminal TRPV1 antibody (lane 3) or a non-specific antibody (lane 4). Immunocomplexes were separated by 10% SDS–PAGE, followed by the detection of TRPV1 (left) and β-tubulin (right) by western blot. (b) Western blot analysis using antitubulin antibody (lane 1) failed to detect tubulin in the anti-N-terminal TRPV1 immunoprecipitate from non-transfected F11 cells, which do not express TRPV1. (c) Tubulin co-immunoprecipitates with TRPV1 from spinal cord extract. TRPV1 could be detected by western blots directly in the tissue extract in some preparations, not in others (see also Fig. S3e). However, detection of TRPV1 was possible after anti-TRPV1 immunoprecipitation. From detergent-solubilized extract of rat spinal cord (lane 1), proteins were immunoprecipitated by the anti-N-terminal TRPV1 antibody (lane 2), or a non-specific antibody (lane 3). Immunocomplexes were separated by 10% SDS–PAGE, followed by the detection of TRPV1 (left) and β-tubulin (right) by western blot. (d) Actin is not co-immunoprecipitated with TRPV1. Western blot analysis using an anti-actin antibody (left panel) and anti TRPV1 antibody (right panel) of an extract of F11 cells transiently expressing TRPV1 (lane 1), proteins that co-precipitate with TRPV1 from transfected F11 cells (lane 2), and proteins co-precipitated with a non-specific antibody (lane 3).

Tubulin was also co-precipitated with TRPV1 from spinal cord extracts, suggesting that the tubulin–TRPV1 interaction not only occurs in transfected cells, but also in native tissue (Fig. 2c).

We tested whether monoclonal antibodies against α-tubulin and β-tubulin are capable of co-precipitating TRPV1 with the tubulins. However, TRPV1 was not co-precipitated in those experiments (data not shown). We conclude from this observation that only a small subset of tubulin might interact with TRPV1, and TRPV1 was much less abundant in the preparation than tubulin.

Actin is another major constituent of cytoskeletal structures. In order to assess whether actin co-precipitates in anti-TRPV1 immunoprecipitation, we probed the anti-TRPV1 immunoprecipitate for the presence of actin. However, we did not detect any actin (Fig. 2d). Taken together, our results suggest that tubulin, but not actin, is a component of the TRPV1 complex, and links the receptor to the cytoskeleton.

TRPV1 co-localizes in the cell with microtubular structures

To assess the TRPV1–tubulin interaction at the cellular level, we performed co-localization experiments with indirect immunofluorescence. We transiently expressed full-length TRPV1 in F11 cells, and visualized TRPV1 and endogenous tubulin. We observed that TRPV1 was localized both in intracellular membranes and the plasma membrane. We observed co-localization of anti-TRPV1 immunoreactivity with anti tubulin immunoreactivity (Fig. 3a, see also Fig. S4a). In neurite-like extensions of F11 cells, we observed a dot-like anti-TRPV1 immunoreactivity. Many of these distinct dot-like TRPV1-immunoreactive structures also stained positive with anti tubulin antibodies. Interestingly, we did not observe such dot-like anti tubulin immunoreactivity pattern in non-transfected F11 cells (Fig. S4c). In contrast to what we observed with tubulin and TRPV1, we observed only minimal co-localization of TRPV1 with actin filaments and, in particular, no overlap in dot-like areas, when we stained F-actin with fluorescent phalloidin (Fig. 3b, see also Fig. S4b).

Figure 3.

TRPV1 co-localizes with tubulin in transfected F11 cells. (a) Confocal images (i–iv) and enlarged sections of the same images (v–viii) of F11 cells transiently expressing TRPV1. There is some distinct co-localization (yellow in the overlay) of TRPV1 (green) and tubulin (red) immunoreactivity. TRPV1 (i and v) and tubulin (ii and vi) were visualized by indirect immunofluorescence with an anti-TRPV1 antibody and an anti tubulin antibody. Regions of co-localization appear as yellow (iii and vii). Distinct spots of co-localization near the plasma membrane are marked with arrows. Phase-contrast pictures (iv and viii) of the same are shown. Scale bar: 20 µM. (b) Confocal images of F11 cells (i–iv) transiently expressing TRPV1 probed for TRPV1 (green; i) and actin fibres (red; ii) using the anti-TRPV1 antibody and Alexa Fluor 594 phalloidin. Scale bar: 20 µM. No significant overlap (iii) of the immunofluorescence signals was observed. Phase-contrast pictures (iv) of the same is shown. For serial Z-stacks views, see also Fig. S4.

Purified tubulin alone binds to the C-terminal sequence of TRPV1, but not with the N-terminal portion of TRPV1

In the MBP-TRPV1-Ct pulldown assay with proteins from spinal cord extracts we observed that tubulin interacts with the C-terminal domain of the TRPV1 along with other proteins. In order to test if the tubulin–TRPV1 interaction is direct and independent of other proteins, we prepared a soluble tubulin dimer fraction from porcine brain and performed the pulldown assay with this purified tubulin preparation. We observed that a considerable amount of purified tubulin was pulled down with MBP-TRPV1-Ct both in the presence and absence of Ca2+, but not with MBP alone. Purified tubulin was also not pulled down by MBP-TRPV1-Nt (Fig. 4a). We confirmed this by western blot analysis using antibodies directed against tubulin (Fig. 4b). In reverse pulldown experiments using biotinylated tubulin immobilized on streptavidin agarose as a bait, we found that MBP-TRPV1-Ct, but neither MBP-TRPV1-Nt nor MBP alone, bound to the tubulin (Figs 4c and d). This suggests that the TRPV1–tubulin interaction is direct and specific and is mediated by the TRPV1 C-terminal cytoplasmic portion, not by the N-terminal cytoplasmic sequence.

Figure 4.

Purified tubulin directly interacts with MBP-TRPV1-Ct in a Ca2+-sensitive way, but not with MBP-TRPV1-Nt. (a) MBP-TRPV1-Nt, MBP-TRPV1-Ct and MBP alone, all immobilized on amylose resin material, were incubated with purified tubulin dimers. Lane 1 MBP-TRPV1-Nt; lane 2, MBP-TRPV1-Ct; lane 3, MBP alone; lane 4, tubulin dimers. Tubulin dimers did not bind to the MBP-TRPV1-Nt (lanes 5 and 6) nor to the MBP alone (lanes 9 and 10), but bound to the MBP-TRPV1-Ct (lanes 7 and 8). Binding of tubulin dimers to MBP-TRPV1-Ct appeared to be slightly more in presence of Ca2+ (+) as compared to the situation in the absence of Ca2+ (–). The proteins were eluted from the amylose resin with 10 mm maltose and resolved by 10% SDS–PAGE. Proteins were stained with silver stain. (b) Western blot analysis of corresponding samples with anti tubulin antibdy YL1/2, confirming the presence of tubulin only in the MBP-TRPV1-Ct pulldown elutes. (c) MBP-TRPV1-Ct is pulled down by biotinylated tubulin immobilized with streptavidin agarose. Biotinylated tubulin immobilized on streptavidin agarose was incubated with MBP-TRPV1-Nt (lane 1–2), MBP-TRPV1-Ct (lane 3–4), or with MBP alone (lane 5–6) in the presence of Ca2+ (lane 1, 3 and 5) or in the absence of Ca2+ (lane 2, 4 and 6). Lanes 7–9 show the input of MBP-TRPV1-Nt, MBP-TRPV1-Ct and MBP, respectively. Only MBP-TRPV1-Ct was pulled down with the biotinylated tubulin. (d) Western blot analysis of samples corresponding to those in (c) probed with the anti-MBP antibody. Among all the pulldown samples, anti-MBP immunoreactivity shows up only in lanes 3 and 4, confirming that MBP-TRPV1-Ct, but not MBP-TRPV1-Nt or MBP alone, were pulled down with biotinylated tubulin.

MBP-TRPV1-Ct, but not MBP-TRPV1-Nt, interacts also with polymerized forms of tubulin

In order to assess if the C-terminal domain of TRPV1 also interacts with polymerized forms of tubulin, we performed a co-assembly/co-sedimentation experiment. MTs can be formed from preparations of soluble dimers by adding GTP. The addition of taxol favours MT formation and also stabilizes the MTs formed. MBP-TRPV1-Ct, MBP-TRPV1-Nt, or MBP were incubated with pre-formed MTs in the presence or absence of free Ca2+ at room temperature. The MTs that formed along with bound proteins were separated from soluble dimers and unbound proteins by centrifugation, and the corresponding supernatants and pellets were analysed by staining of proteins and western blots using anti-MBP antibodies. We observed that approximately 50% of the total MBP-TRPV1-Ct co-sedimented with the polymerized MTs (Fig. 5a). In contrast, MBP-TRPV1-Nt, and MBP alone, failed to co-sediment with polymerized MTs, and all of the MBP-TRPV1-Nt, or MBP, respectively, remained in the supernatants. This result was further confirmed by western blot analysis of the same samples using an anti-MBP antibody (Fig. 5b). These data suggest that MBP-TRPV1-Ct, but not MBP-TRPV1-Nt, is capable of interaction not only with tubulin dimers, but also with polymerized MTs. In the absence of polymerized MTs, MBP-TRPV1-Ct was not found in the pellet fraction. Instead, it remained in the supernatant after centrifugation under the same conditions (Fig. 5c). This supports our conclusion that the presence of the MBP-TRPV1-Ct in the pellet fraction is due to its direct interaction with polymerized MT. In this experiment, we observed no significant effect of Ca2+ on the interaction.

Figure 5.

MBP-TRPV1-Ct, but not MBP-TRPV1-Nt, or MBP alone, co-sediments with polymerized microtubules (MTs). Purified αβ-tubulin dimers were incubated with taxol and GTP to form MTs. Taxol-stabilized MTs were incubated with 5 µg of purified MBP alone, MBP-TRPV1-Ct, or of MBP-TRPV1-Nt, in the presence (+) or absence (–) of Ca2+ (1 mm). Supernatant (S) and pellet (P) were separated by centrifugation and analysed by 10% SDS–PAGE. (a) Distribution of the proteins between MT pellet (P) and supernatant (S) as visualized by silver staining of the proteins. MBP (left panel), MBP-TRPV1-Ct (middle panel), or MBP-TRPV1-Nt (right panel) were assayed for co-sedimentation with MT. Tubulin on the gel is indicated by an asterisk (*), and the positions of the MBP and fusion proteins are indicated by arrows. Only MBP-TRPV1-Ct co-sediments with MT. (b) Western blot analysis of samples corresponding to those shown in (a) using the anti-MBP antibody. Only when MBP-TRPV1-Ct was assayed for co-sedimentation with MT, anti-MBP immunoreactivity appeared in the pellet fractions (middle panel). In contrast, when MBP-TRPV1-Nt or MBP alone were assayed for cosedimentation with MT, anti MBP immunoreactivity was only detected in the supernatants. (c) MBP-TRPV1-Ct never appeared in the pellet fractions in the absence of MT as confirmed by silver stain (left panel) and by anti-MBP western blot (right panel).

Tubulin interaction with MBP-TRPV1-Ct, but not MBP-TRPV1-Nt, allows MT formation under conditions that otherwise destabilize MTs

To explore if MBP-TRPV1-Ct interaction with polymerized MT changes MT physico-chemical properties, especially their ‘stability as polymers’, we assessed MT formation from tubulin dimers in the absence of fusion proteins, or in the presence of MBP-TRPV1-Ct, MBP-TRPV1-Nt, under conditions which interfere with MT formation, i.e. in the presence of the MT-destabilizing agent nocodazole and Ca2+. In the presence of nocodazole or nocodazole/Ca2+, MT formation from tubulin dimers in the absence of fusion proteins was poor (Fig. 6, upper panel). Strikingly, when MBP-TRPV1-Ct was present, MT formation was considerably enhanced even in the presence of nocodazole or nocodazole/Ca2+, and a significant portion of MBP-TRPV1-Ct was found in the MT pellet (Fig. 6, middle panel). In contrast, in the presence of MBP-TRPV1-Nt, MT polymer formation was not altered as compared to the ‘tubulin-only’ sample, and no fusion protein was observed in the MT pellet. The entire amount of MBP-TRPV1-Nt remained in the supernatant (Fig. 6, bottom panel).

Figure 6.

MBP-TRPV1-Ct, but not MBP-TRPV1-Nt, allows MT polymerization in the presence of nocodazole: Comparison of tubulin distribution between supernatant (indicated by S), and pellet (corresponding to MT polymer), indicated by P, formed from αβ-tubulin dimer alone (upper panel), in the presence of MBP-TRPV1-Ct (middle panel), and in the presence of MBP-TRPV1-Nt (bottom panel), in buffer (lane 1), with nocodazole (lane 2), with Ca2+ (lane 3) and in presence of both nocodazole and Ca2+ (lane 4). Proteins were visualized by SDS–PAGE followed by silver staining of proteins in the supernatants and pellets, and anti tubulin western blot of the pellets. The asterisks (*) indicate the position of tubulin, MBP-TRPV1-Nt is indicated by the dotted arrow, and MBP-TRPV1-Ct by the solid arrow. Note that the amount of MT polymer formed in the presence of nocodazol or nocodazol/Ca2+ is significantly increased in the presence of MBP-TRPV1-Ct as compared to the ‘tubulin-only ‘control. In contrast, the presence of the same amount of MBP-TRPV1-Nt has no effect.

These data suggest that MBP-TRPV1-Ct is able to provide stability to the polymers even in the presence of nocodazole.

MBP-TRPV1-Ct, but not MBP-TRPV1-Nt, stabilizes MTs against cold-induced depolymerization in a Ca2+-sensitive manner

MTs are known to be susceptible to cold-induced depolymerization. In order to test if MBP-TRPV1-Ct affects this property, we performed tubulin polymerization from tubulin dimers in the absence of fusion proteins, or in the presence of MBP-TRPV1-Ct or MBP-TRPV1-Nt, with or without Ca2+, followed by depolymerization at low temperature (see Fig. 7a for a flow chart of the experiment). In the ‘tubulin-only’ sample (Fig. 7b, upper panel), a portion of tubulin dimers polymerized to form MT polymers at 37°C, the latter of which could be pelletted by centrifugation (Fig. 7b, lane 2). When subjected to ice-cold temperatures, the MT polymers were partially depolymerized under the experimental conditions used, and tubulin occured in the supernatant (Fig. 7b, lane 3) and a pellet (Fig. 7b, lane 4) fraction after another round of centrifugation. We observed that, after the first cycle of polymerization, the amount of formed MT polymer pellet was unaffected by the presence or absence of Ca2+. A wcomparable amount of MT polymers formed in the presence of MBP-TRPV1-Ct with, in agreement with the experiments shown above, a significant portion of the MBP-TRPV1-Ct was detectable in the MT pellet fraction (Fig. 7, middle panel). Strikingly, in the presence of Ca2+ and MBP-TRPV1-Ct, cold-induced MT depolymerization was dramatically reduced (Fig. 7, middle panel). The presence of MBP-TRPV1-Nt had no such effect. In fact, MBP-TRPV1-Nt was never observed in a MT polymer pellet, not even after the first round of MT polymerization from tubulin dimers (Fig. 7, bottom panel). This suggests a specific Ca2+-sensitive stabilizing effect of the TRPV1 C-terminus on MTs against cold-induced depolymerization.

Figure 7.

MBP-TRPV1-Ct provides stability to the MT towards depolymerization at low temperatures. (a) Flow chart of the experiment: First, MT were allowed to form from tubulin dimers, followed by separation of MT polymers from remaining dimers by centrifugation. Then, the MT polymer pellet was subjected to cold-temperature-induced depolymerization on ice. Subsequently, cold-stable MT polymers were separated from disassembled tubulin by another round of centrifugation. This sequential polymerization/depolymerization was performed with tubulin alone, in the presence of MBP-TRPV1-Ct, or MBP-TRPV1-Nt, in the presence or absence of Ca2+, respectively. The distribution of tubulin and added proteins between supernatant and pellet fractions was monitored. (b) αβ-Tubulin dimer alone (top panel) or in the presence of MBP-TRPV1-Ct (middle panel), or in the presence of MBP-TRPV1-Nt (bottom panel), were subjected to polymerization either in the absence of Ca2+ (–, left side) or presence of Ca2+ (+, right side). After the first cycle of polymerization, free dimers and unbound proteins (which remained in the supernatant, S, lane 1) were separated from MT polymers and associated proteins (pellet, P, lane 2) by centrifugation. The MT polymers were subsequently depolymerized by ice-cold buffer, followed by the centrifugal separation of disassembled tubulin dimers (S, lane 3) and cold-stable MTs (P, lane 4). Samples were anaylsed by 10% SDS–PAGE followed by silver staining. The position of tubulin is indicated by an asterisk (*), MBP-TRPV1-Ct is indicated by a solid arrow, and MBP-TRPV1-Nt by a dotted arrow. In contrast to MBP-TRPV1-Nt, which remained in the supernatant in step 1 (lanes 1, bottom panel), MBP-TRPV1-Ct co-sedimented with polymerized MT both in the presence or absence of Ca2+ (lanes 2, middle panel). Strikingly, the association of MBP-TRPV1-Ct with MT in the presence of Ca2+ results in an increased amount of cold-stable MT polymer (middle panel, right side, lanes 3 and 4).

In summary, there appears to be not only an interaction of the TRPV1 C-terminal portion with tubulin dimers and with polymerized MTs but – at least under some experimental conditions – also a Ca2+-sensitivity of this interaction. Furthermore, our data suggest that the C-terminus of TRPV1 affects the oligomerization status of tubulin.


This work is the first report on a direct interaction of the vanilloid receptor TRPV1 with the cytoskeleton. In a screen for cytoplasmic interaction partners of the vanilloid receptor TRPV1 by proteomic means, we identified β-tubulin as a protein which interacts with the TRPV1 C-terminus in a Ca2+-sensitive manner. This finding suggests that the MT cytoskeleton is a downstream effector of TRPV1-mediated Ca2+-signalling. Furthermore, we observed effects of the TRPV1 C-terminal fragment, but not with the TRPV1 N-terminal portion, on the tubulin oligomerization status. Remarkably, there was a Ca2+-sensitive stabilizing effect of the TRPV1 C-terminal fragment on MTs towards cold-induced depolymerization and a Ca2+-independent stabilization of MTs with respect to depolymerization by nocodazole.

How do these observations fit the current knowledge on interactions and functions of TRPV1? Though known sequence motifs are absent from the primary structure of the cytoplasmic C-terminal tail of TRPV1, there is a body of experimental evidence that attribute important functional properties to this region of the protein. The C-terminal sequence of TRPV1 was demonstrated to contain a binding site for the phosphoinositide PIP2, which negatively regulates the ion channel activity (Prescott and Julius 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, and usage of this site affects the TRPV1 desensitization characteristics in the presence of Ca2+ (Jung et al. 2004). Furthermore, Numazaki et al. (2003) detected a short sequence critical for receptor desensitization within the C-terminal tail, and this was suggested to carry a binding site for calmodulin. Vlachova et al. (2003) demonstrated that the TRPV1 C-terminal tail also carries structural determinants that take part in rendering the receptor sensitive to heat and capsaicin. Notably, several funtionally relevant structural determinants were all mapped to a certain sequence of the TRPV1 C-terminal tail, namely the range within the sequence stretch of amino acid residues 760–800 (based on the rat TRPV1 primary structure), which may represent a ‘hot spot’ for TRPV1 regulation. According to the structural model for the TRPV1 C-terminus suggested by Vlachova et al. (2003) this sequence stretch contains two beta strands. However, up to now we do not know whether the tubulin binding site is localized within this sequence stretch as well.

Interestingly, in the same article (Vlachova et al. 2003), a structural similarity between the TRPV1 C-terminal tail and the fragile histidine triad protein (FHIT), a tumour suppressor gene product, was reported. This was based on sequence alignments and molecular modelling of the putative 3D structure of the TRPV1 C-terminal tail based on the FHIT crystal structure. Remarkably, FHIT was reported to be a tubulin-binding protein (Chaudhuri et al. 1999).

However, with respect to the calmodulin binding, in our hands there was no detectable interaction between purified calmodulin and the C-terminal TRPV1-MBP fusion protein, regardless of the presence or absence of tubulin and/or Ca2+. At the present stage, we cannot explain the reason for this discrepancy to what has been reported by others (Numazaki et al. 2003). However, there was a recent report that rather the N-terminus than the C-terminus of TRPV1 might be involved in the interaction with calmodulin (Rosenbaum et al. 2004).

Tubulin carries a number of negetively charged amino acid residues on its surface, and most of the known MT-associated proteins contain many basic amino acid residues (Nogales 2001). Interestingly, fitting into this scheme, the C-terminal sequence of TRPV1 also contains several basic amino acid residues, and the calculated pI of the entire C-terminal sequences is as high as 9.2.

With regard to the identification of β-tubulin as a TRPV1-interactive partner in a Ca2+-sensitive screen, it is important to mention that tubulin has been shown to bind two molecules of Ca2+ at its C-terminal region (Hayashi and Matsumura 1975; Solomon 1977; Serrano et al. 1986), and a Ca2+-dependent conformational change of tubulin has been detected (Soto et al. 1996).

A number of membrane receptors have been shown to interact specifically with either α- or β-tubulin, such as isoforms of the metabotropic glutamate receptors mGluR1 and mGluR7 (Ciruela et al. 1999; Ciruela and McIlhinney 2001; Saugstad et al. 2002), or subunits of the NMDA receptor (van Rossum et al. 1999). Those isoforms that bind β-tubulin were reported to inhibit MT formation. A number of studies established a wide range of interactions between tubulin and different G proteins and pointed to functional implications of these interactions (Popova et al. 1997; Roychowdhury and Rasenick 1997; Roychowdhury et al. 1999; Sarma et al. 2003).

In this work we showed that the C-terminal sequence of TRPV1 not only interacts with the soluble tubulin dimer, but also enhances the resistance of the MT polymers to nocodazole, a potent drug often used for MT depolymerization. Interestingly, the polymers formed in presence of MBP-TRPV1-Ct show an enhanced stability against cold-induced depolymerization, and this process is Ca2+-dependent. This, in general terms, suggests that the MT cytoskeleton close to the plasma membrane is subject to signal-induced plastic changes. Within the context of nociception, there is emerging interest in the cytoskeleton because it has been demonstrated that certain forms of hyperalgesia depend on cytoskeleton integrity (Dina et al. 2003). Whether potential TRPV1-mediated effects on the cytoskeleton contribute to the modulation of nociceptive signalling, remains to be determined.


Technical assistance by Doris Krück and Hermann Bayer is gratefully acknowledged. This work was supported by grant no. 01 GG 9818/0, Molecular Pain Research; by the Deutsche Forschungsgemeinschaft, Sfb 515; and by the Fonds der Chemischen Industrie.

Supplementary material

The following material is available from:

  • Figure S1. Expression and purification of MBP-TRPV1-Ct and MBP-TRPV1-Nt.

  • Figure S2. Peptide mass fingerprint as obtained from MALDI-MS analysis of the peptide mixture derived from the protein corresponding to the 55 kDa band observed as a Ca2+-sensitive TRPV1 interaction partner. Peptides matching -tubulin are indicated.

  • Figures S3. Assessment of the specificity of the anti N-terminal and anti C-terminal TRPV1 antibodies.

  • Figure S4. Co-localisation of TRPV1 and tubulin in F11 cells.