Paclitaxel-resistant cells have a mutation in the paclitaxel-binding region of β-tubulin (Asp²⁶Glu) and less stable microtubules

KB-15-PTX/099 cells are resistant to paclitaxel but retain sensitivity to HTI-286 in vivo. KB-3-1 (open symbols/dotted lines) or KB-15-PTX/099 (closed symbols/solid lines) tumor cells were implanted s.c. into athymic mice. Animals bearing established tumors were treated i.v. with saline (diamonds), 1.25 mg/kg HTI-286 (squares), or 60 mg/kg paclitaxel (triangles). Vehicle and drugs were administered weekly for three cycles. Points, mean of tumor size (mg) at each time point; bars, SD. *, P < 0.01, significant difference between treatments versus vehicle is indicated at the end of the treatment periods.

Analysis of P-glycoprotein levels and accumulation of paclitaxel in KB-3-1 parental and resistant cell lines. A, the indicated cell lines were analyzed for levels of P-glycoprotein (P-gp). Protein levels were determined by SDS-PAGE/immunoblot analyses. Positive control cell lines are included for P-glycoprotein expression (KB-15-PTX and KB-8.5). B, the cell lines were analyzed for mRNA expression by quantitative reverse transcription-PCR, with relative ratio of P-glycoprotein/actin set to 1 for KB-3-1 cells. C, the indicated cell lines were incubated for 2 h with [¹⁴C]paclitaxel ([¹⁴C]PTX) and levels of radioactive drug remaining in cells quantified by liquid scintillation counting. Representative experiment. Columns, percentage of drug remaining in resistant cells compared with parental cells (KB-3-1) set at 100%; bars, SD. *, P < 0.01, significant difference from KB-3-1.

Dependence of KB-15-PTX/099 cells on microtubule-stabilizing agents. KB-15-PTX/099 cells were grown in the indicated tubulin agents, and growth was measured after 3 d. One hundred percent growth of untreated cells (dotted line). Microtubule-stabilizing agents (closed symbols). Microtubule-destabilizing agents (open symbols).

Comparison of microtubule stability in parental and paclitaxel resistant cells. A, cells were grown in the presence (white columns) or absence (black columns) of 15 nmol/L paclitaxel, and the amount of tubulin in microtubules was measured as described in Materials and Methods. The intensity of the tubulin band in the immunoblot was quantified. Columns, average of three measurements; bars, SD. *, P < 0.01, significant difference between indicated cell line and treatments versus KB-3-1 in the absence of paclitaxel. B, cells were grown in the absence of paclitaxel, and the amount of acetylated tubulin was measured by immunoblotting with an acetylated tubulin–specific antibody. Columns, average percentage of acetylated-tubulin from three experiments normalized to actin and KB-3-1 set to 100%; bars, SE. *, P < 0.05.

Paclitaxel (“T-Taxol”) docked and optimized in β-tubulin. Model (from ref. 39) shows several protein side chains interacting with the bound ligand.

Model of paclitaxel and docetaxel in wild-type and mutant tubulin. Amber94/MOE energy optimized models of paclitaxel (magenta) and docetaxel (teal) in the binding site of wild-type and mutant β-tubulin; protein carbons (white). A, in the Asp²⁶ wild type, the protein atoms of each model are almost completely superimposable, with only a small displacement of His²²⁷ (<0.1Å). The hydrophobic C-3′ NHR end group of each ligand packs against the Van der Waal surfaces of both the Asp²⁶ CH₂ (heavy atom distances, 3.5 Å) and the Val²³ side chains. The OtBu of docetaxel allows positioning of its C-3′ N-H within hydrogen bonding distance (2.0 Å) of the Asp²⁶ carboxylate, whereas the additional bulk of the paclitaxel phenyl keeps its N-H ≥2.8Å away from the carboxylate. The corresponding reduction in noncovalent attraction may account for the lower activity of paclitaxel in wild-type systems. B, the Glu²⁶ mutants were derived from the wild-type models and potential rotamer positions analyzed (only paclitaxel shown). Three possible rotamer states (R1–R3) were identified for each ligand complex [cf. text and Table S1 (available at http://mct.aacrjournals.org)] to provide starting structures for further force field optimization.