Screening Anti-Cancer Drugs against Tubulin using Catch-and-Release Electrospray Ionization Mass Spectrometry
Tubulin, which is the building block of microtubules, plays an important role in cell division. This critical role makes tubulin an attractive target for the development of chemotherapeutic drugs to treat cancer. Currently, there is no general binding assay for tubulin–drug interactions. The present work describes the application of the catch-and-release electrospray ionization mass spectrometry (CaR-ESI-MS) assay to investigate the binding of colchicinoid drugs to αβ-tubulin dimers extracted from porcine brain. Proof-of-concept experiments using positive (ligands with known affinities) and negative (non-binders) controls were performed to establish the reliability of the assay. The assay was then used to screen a library of seven colchicinoid analogues to test their binding to tubulin and to rank their affinities.
KeywordsTubulin Drug Binding Affinity
The αβ-tubulin heterodimers are the building blocks of microtubules (MTs) , which are long, hollow, cylindrical protein polymers and dynamic structural components of the eukaryotic cytoskeleton [2, 3, 4, 5]. Both the α and β subunits exist as different isotypes in many organisms, with dissimilar relative amounts, e.g., there are six main isoforms of both α (αI-VI) and β (βI-VI) subunits in mammals, differing from each other in their amino acid sequences and encoded by different genes [6, 7]. Microtubules control cell shape and play important roles in mitosis, intracellular vesicle transport, organization, and positioning of membranous organelles [8, 9, 10]. The critical role of microtubules in cell division makes them an attractive target in cancer chemotherapy. Microtubule targeting agents (MTAs) modify the formation and function of MTs by altering microtubule dynamics . These MTAs interact with tubulin so as to stabilize or destabilize the polymeric microtubules, in each case leading to cell cycle arrest. In general, drugs that interact with microtubules are divided in two groups: (1) microtubule stabilizers, which include taxanes , epothilones , discodermolide , and peloruside A , and (2) microtubules destabilizers, including colchicinoids , vinca alkaloids , dolastatin , and hemiasterlins .
Of the known MTAs, taxol, colchicine, and vinblastine are among the best characterized and most widely used clinically . Colchicine binds at the interface of the α and β subunits , whereas taxol and vinblastine bind to β subunit but in different domains. Colchicine has been widely used to treat gout, familial Mediterranean fever, and auto-inflammatory diseases [22, 23]. It is an alkaloid found in Colchium autmnale and Gloriosa superba and is composed of three rings, a trimethoxy benzene ring (ring A), a methoxy tropone ring (ring C), and a seven-membered ring (ring B) . The administration of colchicine inhibits the cell division process through two proposed mechanisms. In the first mechanism, referred to as the end-poisoning mechanism, polymerization is inhibited through colchicine binding to the ends of the microtubules, thus preventing the growth of microtubules by sterically blocking further addition of the tubulin dimers. The other mechanism focuses on colchicine binding-induced conformational changes in the tubulin dimers, which leads to the inhibition of microtubule assembly or disassembly of the microtubules [9, 16, 24]. However, the high concentrations of colchicine required to effectively kill cancer cells also adversely affects non-tumoral cells, leading to significant dose-limiting side effects. Consequently, extensive research to design analogues of colchicine (e.g., combretastatin A4)  or colchicine derivatives with improved binding (to MTs), thereby lowering the required therapeutic does and widening the therapeutic window, has been underway in recent years .
Currently, no general binding assay exists for tubulin interactions with colchicinoid drugs as well as other classes of MTAs. The use of isothermal titration calorimetry (ITC), which represents the gold standard for quantifying protein–ligand interactions in vitro, is limited by the relatively long equilibration times required for these interactions (Figure S1, Supplementary Information). Radiolabeling assays have previously been used to study tubulin-colchicine binding. However, the radiolabeled compounds are expensive and require special handling, disposal, and detection. Fluorescence assays have also been used to quantify colchicine binding to tubulin [9, 32]. The main advantage of these methods is that neither requires the labeling of drugs nor the separation of free colchicine from the complex . However, some colchicine analogues (e.g., thiocolchicine) exhibit little or no fluorescence in such assays and cannot be studied using this approach [9, 33]. Therefore, there is a significant need for a general and robust assay to measure the affinities of colchicine analogues for tubulin. This would enable a reliable characterization and ranking of novel tubulin-binding ligands with greater specificity and selectivity for the targets, which could exhibit subtle but important structural differences such as found in tubulin isotypes and mutants.
Recently, the direct electrospray ionization mass spectrometry (ESI-MS) assay has emerged as a powerful tool for measuring the affinities of protein-ligand interactions in vitro [34, 35, 36, 37, 38, 39]. The assay is based on direct measurement of the relative abundances of the ligand-bound and free protein ions measured by ESI-MS. This assay has a number of strengths that make it a valuable technique for binding measurements, including simplicity (no labeling or immobilization is required), speed (measurements can be completed within a few minutes), and low sample consumption (typically <pmol per analysis). The ability to measure multiple binding equilibria simultaneously makes the ESI-MS assay well suited for screening libraries of compounds against target proteins [38, 39]. Where direct detection of the protein–ligand complexes by ESI-MS is not possible due to factors such as high molecular weight (MW) or protein heterogeneity, the catch-and-release (CaR) ESI-MS assay can be employed [40, 41, 42, 43, 44, 45, 46, 47]. In this assay, ligand binding is established by releasing (as ions) ligands bound to the protein in the gas phase using activation methods such as collision-induced dissociation (CID). Typically, ligands can be identified based on their measured MW; in some cases, fragmentation of the released corresponding ligand ion or ion mobility separation (IMS) may be required for positive identification. The CaR-ESI-MS assay has been used to screen carbohydrate, peptide, and small molecule libraries against target proteins [40, 41, 42, 43, 44, 45, 46, 47]. However, the reliability of the method has only been rigorously tested in the case of carbohydrate libraries [40, 41, 45, 46, 47].
Here, we describe the application of the CaR-ESI-MS assay to investigate the binding of several colchicinoid drugs to αβ-tubulin dimers extracted from porcine brain. First, proof-of-concept experiments, using positive (ligands with known affinities) and negative (non-binders) controls, were performed to establish the reliability of the assay for detecting tubulin-colchicinoid drugs interactions. The assay was then used to screen a small library of seven colchicinoid analogues to test their binding to tubulin and to rank their affinities.
αβ-Tubulin dimer (MW ~110 kDa), extracted from porcine brain, was purchased from Cytoskeleton, Inc. (Denver, CO, USA). Bovine serum albumin (BSA) was purchased from Sigma-Aldrich Canada (Oakville, ON, Canada). A stock solution of αβ-tubulin dimer was prepared by dissolving a known mass of protein in 500 μL of aqueous ammonium acetate (50 mM, pH 6.8, at 4 °C) and concentrated to 5 mg mL–1 (~40 μM) by ultracentrifugation using a Vivaspin, a 0.5 mL centrifugal filter (Sartorius Stedim Biotech, Gӧttingen, Germany) with MW cutoff of 30 kDa. The stock solution was then separated into aliquots, which were snap-frozen in liquid nitrogen and then stored at –80 °C until needed. The stock solution of BSA was prepared by dissolving a known mass of protein in deionized water, followed by buffer exchange into ammonium acetate (50 mM, pH 6.8) using a Vivaspin 0.5 mL centrifugal filter (Sartorius Stedim Biotech) with MW cutoff of 10 kDa. The stock solution, with a final concentration of 500 μM, was then stored at 4 °C.
All of the ESI-MS measurements were carried out in positive ion mode using a Waters Synapt G2S quadrupole-ion mobility separation-time of flight (Q-IMS-TOF) mass spectrometer (Manchester, UK) equipped with a nanoflow (nanoESI) source. A P-1000 micropipette puller was used to produce nanoESI tips from borosilicate capillaries (1.0 mm o.d., 0.68 mm i.d.) (Sutter Instruments, Novato, CA, USA). A voltage of ~1.3 kV was applied to platinum wire inserted into the nanoESI tip to carry out ESI. A source temperature of 60 °C and cone voltage 10 V were used for all measurements. Data acquisitions and processing were performed using MassLynx software ver. 4.1, Waters, Manchester, UK.
To carry out CaR-ESI-MS, the quadrupole mass filter was set to pass ions with a range of m/z values (4000–7000), unless otherwise noted. The quadrupole parameters were set to: mass 1, 2, and 3 at 3000, 4000, and 5000 m/z, respectively, with dwell times of 5% for mass 1 and 2, ramp time for m/z 3000 at 5%, and for m/z 4000 at 85%. Selected ions were then subjected to CID in the Trap region using collision energies ranging from 2 to 140 V and the released ligand ions identified based on the measured m/z. Where indicated, and following CID in the Trap region, the tubulin ions were subjected to another stage of CID in the Transfer region, using collision energies ranging from 1 to 120 V. Prior to the second stage of CID, the tubulin ions were separated from free ligand ions using IMS. For IMS, a wave height of 40 V and a wave velocity between 400 and 700 m s–1 were applied; helium and nitrogen (IMS gas) gas flow rates were 80 and 70 mL min–1, respectively. In all cases, data acquisition and processing were performed using MassLynx software ver. 4.1 (Waters, Manchester, UK).
Result and Discussion
Analysis of Tubulin-Colchicine Binding by ESI-MS
Shown in Figure 2c–f are the ESI mass spectra of aqueous ammonium acetate buffer (100 mM) solutions of αβ-tubulin (14 μM) and a range of colchicine (L1) concentrations (2–20 μM) following incubation at 37 °C for 1 h. While it was not possible to directly ascertain the presence of ions corresponding to L1-bound tubulin dimer from the mass spectra, the presence of L1 in solution resulted in a small but measurable increase in the m/z of the tubulin dimer ions and the magnitude of the effect increased with increasing L1 concentration. This observation is consistent with the presence of L1-bound tubulin dimer in solution. (Figure S2, Supplementary Information).
To conclusively establish the presence of tubulin-bound L1, the CaR-ESI-MS assay was employed. As a starting point, the quadrupole mass filter was set to pass all ions, with m/z between 4000 and 7000, produced from a solution of αβ-tubulin (14 μM); these were then subjected to CID in Trap region using collision energies ranging from 2 to 140 V (Figure S3a, Supplementary Information). It can be seen that the tubulin ions are stable and do not undergo any fragmentation, even at highest collision energies. In the same way, the Car-ESI-MS assay was applied to tubulin ions produced from a solution of αβ-tubulin (14 μM) and L1 (14 μM) (Figures S3b, Supplementary Information). It can be seen that CID resulted in the appearance of signal corresponding to protonated L1 (m/z 400.1) at all of the energies investigated. These results confirm that L1 was bound to the gaseous tubulin ions.
In order to rule out the possibility that the L1–tubulin interactions identified by CaR-ESI-MS arose during the ESI process (i.e., due to nonspecific binding), the measurements were repeated using a solution of αβ-tubulin (14 μM), L1 (14 μM), and BSA (14 μM) (Figure S4, Supplementary Information). To the best of our knowledge, BSA does not interact with L1 in solution and, therefore, served as a reference protein (Pref) to test for nonspecific binding. Isolation of the +18 charge state of the tubulin dimer, followed by CID, resulted in the appearance of protonated L1 (Figure S4, Supplementary Information). In contrast, isolation of the +16 charge state of BSA, followed by CID, failed to produce any signal corresponding to L1 (Figure S4, Supplementary Information). Taken together, these results suggest that the gaseous L1-bound tubulin ions are the result of specific binding in solution, with little or no contribution from nonspecific binding during the ESI process.
As described above, L1 binding to tubulin dimer is accompanied by a change in conformation. Therefore, it was of interest to investigate whether the binding of L1 to tubulin could be quantified based on differences in the collision cross sections of the free and ligand-bound tubulin dimer ions. To assess whether the ligand binding-induced conformational changes result in measurable differences in collision cross sections, IMS arrival time distributions (ATDs) were measured for tubulin dimer ions produced from solutions containing αβ-tubulin alone (14 μM) and in the presence of L1 in ammonium acetate solution (100 mM, pH 6.8) (Figure S5, Supplementary Information). Inspection of the ATDs measured for each charge state reveals no measurable differences for tubulin dimers produced in the absence or presence of L1, even at the highest concentration (28 μM) of L1 investigated.
Taken together, the aforementioned results reveal that with the current instrumentation, it is not possible to directly quantify tubulin–L1 interactions using ESI-MS. Consequently, subsequent efforts were directed towards the possibility of ranking ligand (drug) affinities for tubulin using CaR-ESI-MS. As a starting point, a series of control experiments were carried out to assess the reliability of the approach and to identify optimal instrumental conditions for its implementation. Important requirements when using CaR-ESI-MS to rank affinities based on the relative abundances of the released ligand ions are that the release and ionization efficiencies be similar for all of the bound ligands. Because of differences in the nature of the intermolecular interactions for different ligands in the gaseous protein-ligand complex ions, the dissociation (release) rate constants are expected to vary, possibly significantly, between ligands. Consequently, the condition of uniform release efficiencies can only reliably be achieved by fully releasing all bound ligands (i.e., complete dissociation). Based on the changes in the abundance of L1 relative to the d+18–d+21 ions, with collision energy (in the Trap region) measured in the mass spectra shown in Figure S6 (Supplementary Information), complete release appears to be achieved at energies ≥100 V. To further support this conclusion, the tubulin ions were subjected to a second stage of CID in the Transfer region. Shown in Figure S7a and b (Supplementary Information) are CID mass spectra acquired at Trap and Transfer voltages of 100 and 1 V, respectively (Figure S7a, Supplementary Information), and 1 and 100 V, respectively (Figure S7b, Supplementary Information). In both cases, abundant protonated L1 is detected. Shown in Figure S7c and d (Supplementary Information) are CID mass spectra acquired using Trap and Transfer voltages of 100 and 1 V, respectively (Figure S7c, Supplementary Information) and 100 and 100 V, respectively (Figure S7d, Supplementary Information). These data were acquired using IMS to exclude (based on arrival times) the contribution of any L1 ions released in the Trap region from the CID mass spectra. It can be seen that in both cases, no measurable signal corresponding to L1 ions, resulting from ligand release in the Transfer region, was detected. These results confirm that L1 is fully released from the tubulin ions at Trap voltages ≥100 V.
Analogous measurements were carried out on solutions of αβ-tubulin (14 μM), L1 (14 μM), vincristine (L9) (14 μM), and thiocolchicoside (L8) (14 μM). L8 does not bind to tubulin and served as a negative control. Shown in Figure S8a (Supplementary Information) is a representative ESI mass spectrum; the corresponding CID spectrum measured at a Trap voltage of 120 V is shown in Figure S8b (Supplementary Information). It can be seen that both protonated L1 and L9 are released from tubulin. However, signal corresponding to L8 was not detected at any of the collision energies investigated. Measurements carried out at different concentrations of L1, L8, and L9 produced similar results (data not shown). The results of these control experiments suggest that the CaR-ESI-MS assay can be used to detect ligand binding to tubulin in solution and, by screening libraries of compounds, to rank their affinities.
Taken together, the results of this study suggest that the CaR-ESI-MS assay could aid in detecting specific drug interactions with tubulin dimers in vitro and, when applied to libraries, to rank their affinities. To our knowledge, this is the first reported example of the application of the CaR-ESI-MS assay in the area of anti-cancer drug screening. Proof-of-concept experiments carried out on αβ-tubulin dimers extracted from porcine brain using positive (known ligands) and negative (non-binders) controls were performed to establish the reliability of the assay. The assay was then used to screen a library of seven colchicinoid analogues to test their binding to tubulin and to rank their affinities. The ultimate utility of this methodology lies in applying it as a means of selecting compounds that target specific isotypes of tubulin or specific mutants of tubulin. The reason for this is that tubulin is an abundant protein common to all cells in the body. However, the distribution of tubulin isotypes is different for all cell types and especially for cancer cells . Therefore, the idea of targeting an isotype of tubulin that is highly expressed in cancer cells and much less so in normal cells would lead to an optimization of treatment efficacy and minimization of side effects . A computational algorithm for such a selection process from a library of tubulin-binding ligands has recently been published . What is missing, however, is an experimental methodology whereby compounds can be ranked according to their affinity for tubulin isotypes, which would validate computational predictions. The present paper is intended to demonstrate that such a method is now emerging and, with future refinement, can be developed into a powerful platform for the drug discovery process.
The authors are grateful for financial support provided by the Natural Sciences and Engineering Research Council of Canada and the Alberta Glycomics Center.
- 8.Ballestrem, C., Magid, N., Zonis, J., Shtutman, M., Bershadsky, A.: Interplay between the actin cytoskeleton, focal adhesions and microtubules. In: Ridley, A., Peckham, M., Clark, P. (eds.) Cell Motility: From Molecules to Organisms, p. 75. Wiley, Chichester (2004)Google Scholar
- 13.Bollag, D.M., McQueney, P.A., Zhu, J., Hensens, O., Koupal, L., Liesch, J., Goetz, M., Lazarides, E., Woods, C.M.: Epothilones, a new class of microtubule-stabilizing agents with a taxol-like mechanism of action. Cancer Res. 55, 2325–2333 (1995)Google Scholar
- 14.Canales, A., Rodriguez-Salarichs, J., Trigili, C., Nieto, L., Coderch, C., Andreu, J.M., Paterson, I., Jiménez-Barbero, J., Díaz, J.F.: Insights into the interaction of discodermolide and docetaxel with tubulin. Mapping the binding sites of microtubule-stabilizing agents by using an integrated NMR and computational approach. ACS Chem. Biol. 6, 789–799 (2011)Google Scholar
- 17.Gigant, B., Wang, C., Ravelli, R.B.G., Roussi, F., Steinmetz, M.O., Curmi, P.A., Sobel, A., Knossow, M.: Structural basis for the regulation of tubulin by vinblastine. Nature 435, 519–522 (2005)Google Scholar
- 21.Ravelli, R.B.G., Gigant, B., Curmi, P.A., Jourdain, I., Lachkar, S., Sobel, A., Knossow, M.: Insight into tubulin regulation from a complex with colchicine and a stathmin-like domain. Nature 428, 198–202 (2004)Google Scholar
- 22.Van Echteld, I., Wechalekar, M.D., Schlesinger, N., Buchbinder, R., Aletaha, D.: Colchicine for acute gout. Cochrane Database Syst. Rev. 8, 1–41 (2014)Google Scholar
- 24.Dustin, P.: Microtubules (2nd ed.). Springer-Verlag Berlin Heidelberg, New York (1984)Google Scholar
- 25.Sun, L., Vasilevich, N.I., Fuselier, J.A., Coy, D.H.: Abilities of 3,4-diarylfuran-2-one analogs of combretastatin A-4 to inhibit both proliferation of tumor cell lines and growth of relevant tumors in nude mice. Anticancer Res. 24, 179–186 (2004)Google Scholar
- 27.Fernando Díaz, J., Andreu, J.M.: Kinetics of dissociation of the tubulin-colchicine complex. Complete reaction scheme and comparison to thermodynamic measurements. J. Biol. Chem. 266, 2890–2896 (1991)Google Scholar
- 28.Menéndez, M., Laynez, J., Medrano, F.J., Andreu, J.M.: A thermodynamic study of the interaction of tubulin with colchicine site ligands. J. Biol. Chem. 264, 16367–16371 (1989)Google Scholar
- 31.Banerjee, A., Luduena, R.F.: Kinetics of colchicine binding to purified beta-tubulin isotypes from bovine brain. J. Biol. Chem. 267, 13335–13339 (1992)Google Scholar
- 32.Tahir, S.K., Kovar, P., Rosenberg, S.H., Ng, S.C.: Rapid colchicine competition-binding scintillation proximity assay using biotin-labeled tubulin. Biotechniques 29, 156–160 (2000)Google Scholar
- 37.Gabelica, V., Rosu, F., De Pauw, E.: A simple method to determine electrospray response factors of noncovalent complexes. Anal. Chem. 81, 6708–6715 (2009)Google Scholar
- 43.Cheng, X., Chen, R., Bruce, J.E., Schwartz, B.L., Anderson, G.A., Hofstadler, S.A., Gale, D.C., Smith, R.D., Gao, J., Sigal, G.B., Mammen, M., Whitesides, G.M.: Using electrospray ionization FTICR mass spectrometry to study competitive binding of inhibitors to carbonic anhydrase. J. Am. Chem. Soc. 117, 8859–8860 (1995)Google Scholar
- 44.Gao, J., Cheng, X., Chen, R., Sigal, G.B., Bruce, J.E., Schwartz, B.L., Hofstadler, S.A., Anderson, G.A., Smith, R.D., Whitesides, G.M.: Screening derivatized peptide libraries for tight binding inhibitors to carbonic anhydrase II by electrospray ionization-mass spectrometry. J. Med. Chem. 39, 1949–1955 (1996)Google Scholar
- 46.Zhang, Y., Liu, L., Daneshfar, R., Kitova, E.N., Caishun, L., Feng, J., Cairo, C.W., Klassen, J.S.: Protein–glycosphingolipid interactions revealed using catch-and-release mass spectrometry. Anal. Chem. 84, 7618–7621 (2012)Google Scholar
- 50.Leandro-García, L.J., Leskelä, S., Landa, I., Montero-Conde, C., López-Jiménez, E., Letón, R., Cascón, A., Robledo, M., Rodríguez-Antona, C.: Tumoral and tissue-specific expression of the major human β-tubulin isotypes. Cytoskeleton 67, 214–223 (2010)Google Scholar