Peloruside- and Laulimalide-Resistant Human Ovarian Carcinoma Cells Have βI-Tubulin Mutations and Altered Expression of βII- and βIII-Tubulin Isotypes

Abstract

Peloruside A and laulimalide are potent microtubule-stabilizing natural products with a mechanism of action similar to that of paclitaxel. However, the binding site of peloruside A and laulimalide on tubulin remains poorly understood. Drug resistance in anticancer treatment is a serious problem. We developed peloruside A- and laulimalide-resistant cell lines by selecting 1A9 human ovarian carcinoma cells that were able to grow in the presence of one of these agents. The 1A9-laulimalide resistant cells (L4) were 39-fold resistant to the selecting agent and 39-fold cross-resistant to peloruside A, whereas the 1A9-peloruside A resistant cells (R1) were 6-fold resistant to the selecting agent while they remained sensitive to laulimalide. Neither cell line showed resistance to paclitaxel or other drugs that bind to the taxoid site on β-tubulin nor was there resistance to microtubule-destabilizing drugs. The resistant cells exhibited impaired peloruside A/laulimalide-induced tubulin polymerization and impaired mitotic arrest. Tubulin mutations were found in the βI-tubulin isotype, R306H or R306C for L4 and A296T for R1 cells. This is the first cell-based evidence to support a β-tubulin–binding site for peloruside A and laulimalide. To determine whether the different resistance phenotypes of the cells were attributable to any other tubulin alterations, the β-tubulin isotype composition of the cells was examined. Increased expression of βII- and βIII-tubulin was observed in L4 cells only. These results provide insight into how alterations in tubulin lead to unique resistance profiles for two drugs, peloruside A and laulimalide, that have a similar mode of action.

Introduction

Microtubules are major cytoskeletal polymers composed of heterodimers of α- and β-tubulin subunits (1). They are vital for several cellular functions including mitosis, intracellular trafficking, and maintenance of cell shape. The importance of microtubules in mitotic spindle formation and chromosome movement during cell division makes them a major target for chemotherapeutic drugs to halt the uncontrolled division of cancer cells (2).

Peloruside A and laulimalide (see Supplementary Fig. S1 for structures) are 2 microtubule-targeting agents isolated from different marine sponges Mycale hentscheli and Cacospongia mycofijiensis, respectively (3, 4), although other sources of laulimalide have also been described. The compounds have a similar mechanism of action to paclitaxel and are cytotoxic to a number of mammalian cancer cell lines (5–7). Peloruside A and laulimalide bind to and stabilize the polymerized form of microtubules, inhibiting microtubule dynamics. This interferes with the function of the mitotic spindle and promotes mitotic arrest and apoptosis. Although having a similar mechanism of action to paclitaxel, peloruside A and laulimalide have a number of unique features that make them potentially useful second-generation microtubule-stabilizing drugs for anticancer therapeutics.

Peloruside A and laulimalide share a similar or overlapping binding site on the tubulin dimer that is distinct from the taxoid site (6–9); thus, peloruside A and laulimalide retain their efficacy in paclitaxel- and epothilone-resistant cell lines that are mutated in the taxoid site (6, 7). Another possible advantage of the compounds is that, based on their chemical structure, they may have a greater polarity than paclitaxel. This greater polarity may reduce the need to use a vehicle like Cremophor EL for clinical delivery, as required for paclitaxel (10). Although these features may enhance the potential of using drugs like peloruside A or laulimalide for cancer chemotherapy, as with most of the anticancer drugs in clinical use, the possibility of tumors developing resistance to the drugs cannot be ignored.

Resistance by tumor cells to antimicrotubule agents stems from, but is not limited to, overexpression of the P-glycoprotein (Pgp) drug efflux pump (11), elevated levels of microtubule-destabilizing factors (12), and changes in the target molecule of drugs, tubulin (13–15). Tubulin alterations include tubulin mutations (14), altered expression of β-tubulin isotypes (13, 15), and increased microtubule dynamics (16). Because peloruside A and laulimalide are poor substrates for the Pgp drug efflux pump (6, 7), resistance to these compounds is unlikely to be dependent on development of an MDR phenotype.

Several cellular studies have indicated a role for βItubulin mutations in the development of resistance to paclitaxel and other taxoid site microtubule-stabilizing agents (13, 14, 17–19). Mutations at the taxoid-binding site on βI-tubulin impair drug–tubulin interactions (17); however, the clinical occurrence of such β-tubulin mutations in drug-resistant cancer patients seems to be rare, although it has yet to be thoroughly investigated in patients with acquired resistance (20, 21). Differential expression of specific β-tubulin isotypes has also been associated with resistance to microtubule-targeting agents in cell lines and in patients (13, 15). The aim of this study was, therefore, to generate cell lines resistant to peloruside A and laulimalide and to determine the mechanisms of resistance by testing for tubulin mutations and changes in tubulin isotype expression.

Materials and Methods

Drugs

Natural peloruside A and laulimalide were isolated and purified from the marine sponges M. hentscheli (New Zealand) and C. mycofijiensis (Tonga), respectively and stored as 1 mmol/L stock solutions in absolute ethanol at −80°C. Synthetic laulimalide was a gift from Arun Ghosh, Purdue University, West Lafayette, IN, and synthetic discodermolide was a gift from Ian Paterson, University of Cambridge, Cambridge, United Kingdom. Paclitaxel, epothilone A, vinblastine, vincristine, and colchicine were purchased from Sigma Chemical Co. Epothilone B was from Novartis Pharmaceuticals Corp. and 2-methoxy-estradiol was from Sigma or EntreMed, Inc.

Cell culture

The human ovarian carcinoma cell line 1A9, a derivative of the A2780 cell line, was obtained from the NIH (17). Cells were cultured in RPMI 1640 medium (Invitrogen) supplemented with 10% fetal calf serum (Invitrogen), 0.25 units/mL insulin (Sigma), 100 units/mL penicillin, and 100 units/mL streptomycin (Invitrogen). The cells were maintained at 37°C in a humidified air atmosphere containing 5% CO₂.

Generation of peloruside A- and laulimalide-resistant cell lines

A laulimalide-resistant cell line 1A9-L4 (L4) was obtained by exposing 1A9 cells to increasing concentrations of laulimalide (10–150 nmol/L) over a period of 18 months. The L4 cells that grew in the presence of 150 nmol/L laulimalide exhibited the greatest resistance and were selected for further analysis. The stability of the resistant phenotype in L4 cells was assessed by incubating the cells in the presence of drug-free medium for more than 6 months with no reduction in the relative resistance to laulimalide.

A peloruside A-resistant cell line 1A9-R1 (R1) was derived from 1A9 human ovarian cancer cells grown for a number of months in medium with peloruside A at a concentration just above the IC₅₀ value (25 nmol/L). Verapamil (10 µmol/L) was included throughout to prevent selection of cells with the MDR phenotype, as peloruside A has been shown to be a substrate, although a poor one, for the Pgp pump (6). Over a period of 6 months, the concentration of peloruside A was gradually increased in steps to 30, 40, 50, 60, 80, 100, and 150 nmol/L. The cells were typically exposed to each dose of peloruside A for 3 to 5 days and then allowed to recover in drug-free medium. Once confluent, cells were treated with the next highest concentration of peloruside A until cells were able to grow at the higher peloruside A concentrations. The cells were then cloned to ensure that resistant cell stocks were homogeneous. The resistant phenotype in R1 cells was stable in drug-free medium with no reduction in the relative resistance to peloruside A observed over many months and passages.

Cell proliferation assay

To determine the IC₅₀ value for growth inhibition, we used either a sulforhodamine B protein staining assay or an MTT cell proliferation assay as previously described (5, 17). IC₅₀ values of the resistant cell lines were always compared with parental 1A9 cells in the same (paired) experiment. Cells were tested for resistance against laulimalide, peloruside A, 3 taxoid-binding site drugs (paclitaxel, epothilone A, and discodermolide), and 3 microtubule-destabilizing agents (VLB, colchicine, and 2-methoxy-estradiol).

Analysis of drug effects on tubulin polymerization and function

Intracellular polymerization of tubulin

Drug-induced tubulin polymerization in cells was measured by SDS-PAGE electrophoresis of soluble and pelleted fractions from centrifuged cell lysates (14,400 × g for 10 minutes) and immunoblotting for tubulin. Cells were treated with different concentrations of peloruside A or laulimalide for 16 hours and then processed as previously described (17). For L4 cells, mouse monoclonal α-tubulin primary antibody (1:1,000; Sigma) was used in conjunction with an anti-mouse Alexa Fluor 680 secondary antibody (Invitrogen). Immunoreactivity was detected in a LI-COR Odyssey infrared imaging system (LI-COR Biosciences) and quantified by densitometry with ImageJ (NIH). For Western blotting of R1 cells, a rabbit polyclonal primary antibody to α-tubulin (1:1,000; Abcam) was used with a Cy5-conjugated goat anti-rabbit secondary antibody (1:2,500; Amersham). The electrophoresed proteins were transferred to an Immobilon FL membrane (Millipore Corp), and fluorescence was measured using a Fujifilm FLA-5100 imaging system (Fuji Photo Film Co.). The percentage of polymerized tubulin was calculated from the band densities of soluble and pelleted tubulin.

Immunocytochemistry and confocal microscopy

Immunocytochemistry and confocal microscopy of the 1A9, L4, and R1 cells was carried out as previously described (22, 23).

Cell-cycle analysis

Cell-cycle analysis was conducted by flow cytometry following staining of the DNA with propidium iodide as previously described (5).

Analysis of tubulin structural alterations and isotype expression patterns

Sequence analysis of human α- and β-tubulin genes

Total RNA was isolated from the cells using a Qiagen RNeasy kit, and reverse transcriptase-PCR was carried out using a Protoscript First Strand cDNA Synthesis Kit (New England Biolabs). PCR amplification of cDNA of βI-tubulin (HM40/TUBB gene; RNA accession NM_178014) was carried out using the following primers purchased from Invitrogen: 5′-CTTGCCCCATACATACCTT-3′ and 5′-GTAAGACGGCTAAGGGAACTG-3′.

PCR products were then direct sequenced using the following 7 primers:

5′-TCTGGGGCAGGTAACAACT-3′; 5′-AGTTGTTACCTGCCCCAGA-3′; 5′-CTCCGCAAGTTGGCAGTCAAC-3′; 5′-TGGCCTCCAGATGGCAGTC-3′; 5′-GGGG ATCCATTCCACAAAGTA-3′; 5′-GGACCATGTTGACT GCCAAC-3′; and 5′-GACTGCCATCTTGAGGCCAC-3′.

Subcloning of the cDNA PCR-amplified products was conducted using the TOPO TA cloning system (Invitrogen), followed by direct sequencing with the primers listed above. A minimum of 20 clones were analyzed for each sample. All PCR reactions were carried out in 50 µL reaction mixtures, as previously described (24). Multiple sequence alignment of β-tubulin isotypes was conducted using the Clustal W method and the MegAlign program (Lasergene, DNASTAR). All α-tubulin primer sequences for the predominant α-tubulin isotype (Kα1) were obtained from Poruchynsky and colleagues (25). Primers were purchased from Invitrogen.

Quantitative real-time PCR

mRNA expression levels of the 7 β-tubulin isotypes and a house-keeping gene were determined using a quantitative SYBR green PCR method. Several common house-keeping genes were tested, but only 18S rRNA showed similar expression in the parental and the resistant cells. Total RNA was extracted from frozen cells using an RNeasy Protect Cell Mini kit (Qiagen) following the manufacturer’s instructions. Extracted total RNA (1 µg) was treated with DNase I (amplification grade; Invitrogen) to remove any genomic DNA, and cDNA was synthesized using SuperScript III First-Strand Synthesis Supermix (Invitrogen) in accordance with the manufacturer’s instructions. Excluding a previously described primer set for the βIII-tubulin gene (26), all primers were designed using Beacon Designer (Premier Biosoft International) and manufactured by Invitrogen and are listed in the Supplementary Table S1. For quantification of expression levels of each of the 7 β-tubulin isotypes, and 18S rRNA, singleplex reaction mixes were prepared containing a single set of isotype-specific or 18S rRNA primers at validated concentrations (200 nmol/L) and reagents supplied in the SYBR GreenER qPCR SuperMix Universal Kit (Invitrogen) according to the manufacturer’s instructions. Samples were prepared in duplicate by adding the cDNA sample (200 ng) into the prepared reaction mix (total volume of 52 µL), and then transferring two 25-µL aliquots containing 100 ng of cDNA/aliquot into 0.2 mL optical PCR tubes. The amplification reaction was run on an iCycler real-time PCR detection system (Bio-Rad) under the following conditions: 1 cycle of 50°C for 2 minutes, 1 cycle of 95°C for 10 minutes, 40 cycles of 95°C for 15 seconds, and 60°C for 60 seconds. Controls included reaction mixtures that did not include template and samples that underwent reverse transcription PCR with the exclusion of SuperIII/RNaseOUT Enzyme Mix to check the effectiveness of the DNase treatment. The PCR amplification efficiency for each set of primers, evaluated using the slope of a serially diluted (1:1–1:64) cDNA sample, was above 96% (Supplementary Table S2). The primer specificity was confirmed by sequencing the PCR product. The 1A9 parental cell line was used as a control for the resistant cell lines and quantification of samples was determined by the 2^(−ΔΔC_t) method (27).

One-dimensional gel electrophoresis and Western blotting

Cell lysates were prepared using a cell lysis buffer containing 30 mmol/L Tris-HCl, 7 mol/L urea, 2 mol/L thiourea, and 4% (w/v) 3-[(3-cholamidopropyl) dimethylammonio]-1-propanesulfonate (CHAPS), pH 8.5. Total protein was quantified with a protein assay dye reagent kit (Bio-Rad Laboratories). Lysate containing 25 µg of protein was electrophoresed on a 10% SDS-PAGE gel at 120 V for 1 hour. Western blot transfer to an Immobilon FL membrane was carried out for 17 hours at 20 V in a wet transfer unit. Immunoblotting was carried out with mouse monoclonal βII-tubulin (1:1,000; MMS-422P; Covance), βIII-tubulin (3:1,000; T8660; Sigma), total β-tubulin (3:1,000; T4026; Sigma), β-actin (1:3,000; A2228; Sigma), and rabbit polyclonal α-tubulin (1:1,000; ab18251; Abcam). Primary antibodies were incubated at room temperature with the membrane for 2 hours with rocking. A Cy5-conjugated anti-mouse (1:2,500; PA45010V; Amersham) or anti-rabbit (1:2,500; PA45011V; Amersham) secondary antibody was used, and bands were detected with a Fujifilm FLA-5100 imaging system. Immunoblots were analyzed with ImageJ software.

Two-dimensional gel immunoblot

The cell lysate preparation and protein quantification were carried out as described in the Western blotting procedure. A sample containing 80 µg of protein was mixed with rehydration buffer [2 mol/L thiourea, 7 mol/L urea, 2% IPG buffer, 2% dithiothreitol (DTT), 4% CHAPS], followed by overnight incubation with immobilized pH gradient strips (pH = 3–5.6). Isoelectric focusing was conducted in an Ettan IPGphor Isoelectric Focusing Unit (GE Healthcare) with the following settings: (i) step and hold 300 V for 30 minutes; (ii) gradient 1,000 V for 30 minutes; (iii) gradient 5,000 V for 90 minutes; and (iv) step and hold 5,000 V for 25 minutes. After the isoelectric focusing, the strips were equilibrated in equilibration buffer (50 mmol/L Tris, 6 mol/L urea, 30% glycerol, and 2% SDS) containing 1% DTT for 10 minutes, then in equilibration buffer containing 2.5% iodoacetamide for 10 minutes. The second dimension was conducted on a 4% to 12% gradient NuPage Bis-Tris gel (Invitrogen) for 55 minutes at 200 V. The separated proteins were transferred to an Immobilon FL membrane and probed with mouse monoclonal primary antibodies to βII-tubulin (1:1,000; MMS-422P; Covance), βIII-tubulin (3:1,000; T8660; Sigma), and β-actin (1:3,000; A2228; Sigma). The protein bands were visualized with a Cy5-conjugated anti-mouse secondary antibody (1:2,500; PA45010V; Amersham). The blots were scanned with the Fujifilm FLA-5100 imaging system, and the densities of the protein spots were quantified with ImageJ software and normalized to the β-actin band density.

Data analysis

Tests for significant differences used ANOVA or Student’s t tests. P < 0.05 was taken as significant, and all data are presented as mean ± SEM.

Results

Resistance profile of L4 and R1 cells

To gain a better understanding of the specific effects and interactions of peloruside A and laulimalide with the microtubule cytoskeleton, we selected 1A9 ovarian carcinoma cells with increasing concentrations of either peloruside A or laulimalide and generated 2 stable resistant cell lines named R1 and L4, respectively. The ability of drugs to inhibit growth of these resistant cells is shown in Table 1. The L4 cell line exhibited 39-fold resistance to laulimalide compared with the parental 1A9 cells but showed no significant cross-resistance to microtubule-targeting agents known to bind at the taxoid site (paclitaxel, epothilone A, and epothilone B), the vinca site (VLB and vincristine), or the colchicine site (colchicine and 2-methoxy-estradiol). L4 cells showed a slightly increased resistance to paclitaxel and epothilone A, but the magnitudes of the effects were very small and unlikely to be biologically significant. Interestingly, the L4 cells were highly cross-resistant to peloruside A, confirming earlier evidence that the 2 drugs have an identical or overlapping binding site (6). The R1 cells were 5.6-fold resistant to peloruside A but remained sensitive to the other microtubule-targeting drugs tested (Table 1). Interestingly, unlike the case of L4 cells that were resistant to both peloruside A and laulimalide, there was no significant resistance of R1 cells to laulimalide. This suggests that the binding sites for peloruside A and laulimalide on tubulin might not be identical, but overlapping.

Table 1.

Resistance of cells to various microtubule-targeting agents in parental 1A9 and mutant L4 and R1 cells

	Giannakakou laboratory			Miller laboratory
	1A9	L4	Fold change	1A9	R1	Fold change
LAU	3.6 ± 0.2	141 ± 6a	39.2	14.9 ± 2.9	20.1 ± 3.4	1.3
PLA	10.1 ± 0.8	396 ± 23a	39.2	12.9 ± 1.5	72.4 ± 3.7a	5.6
PTX	1.3 ± 0.1	1.8 ± 0.1b	1.4	14.4 ± 5.1	12.1 ± 4.8	0.8
EPOA	1.3 ± 0.1	1.6 ± 0.1b	1.2	8.1 ± 0.2	9.0 ± 0.8	1.1
EPOB	0.17 ± 0.07	0.47 ± 0.12	2.8
DISC				48.1 ± 10.8	31.6 ± 7.01	0.7
2ME	317 ± 60	233 ± 33	0.7
VBL	0.40 ± 0.06	0.33 ± 0.09	0.8	0.84 ± 0.11	1.00 ± 0.13	1.2
VCR	6.2 ± 0.5	6.7 ± 0.2	1.1
COL				5.4 ± 1.1	4.7 ± 0.7	0.9

NOTE: Growth inhibition by various microtubule-targeting agents was measured in 1A9, L4, and R1 cells. Data are the average IC₅₀ values for growth inhibition in nmol/L. Data are presented as mean ± SEM.

Abbreviations: LAU, laulimalide; PLA, peloruside A; PTX, paclitaxel; EPOA, epothilone A; EPOB, epothilone B; DISC, discodermolide; 2ME, 2-methoxy-estradiol; VBL, vinblastine; VCR, vincristine; COL, colchicine.

^aP < 0.0001 (unpaired Student's t test; n = 3–7 biological replicates). Data on left of table (1A9 and L4) are from PG lab, except for PLA.

^bP < 0.05.

Impaired drug-induced tubulin polymerization

To determine whether the drug-resistant cells displayed altered drug–tubulin interactions, we tested the ability of the drugs to induce tubulin polymerization using a cell-based assay. Treatment of 1A9 cells with as little as 10 nmol/L laulimalide resulted in near-maximum polymerization (90% P) compared with untreated cells in which no polymerized tubulin was detected (Fig. 1A). In contrast, laulimalide treatment of the L4 cells had only a minimal effect at concentrations as high as 100 nmol/L (24% P), consistent with the drug-resistant phenotype of these cells. Paclitaxel-induced polymerization in both cell lines, however, was not compromised. With peloruside A (Fig. 1B), 100 nmol/L peloruside A induced 61% tubulin polymerization in 1A9 cells; whereas, in L4 and R1 cells, 100 nmol/L peloruside A only induced 13% and 22% polymerization, respectively. These results for peloruside A and laulimalide mirror the antiproliferative assays summarized in Table 1.

Figure 1.

Impaired ability of peloruside A and laulimalide to induce tubulin polymerization in L4 and R1 cells. An in situ cellular assay was used to quantify drug-induced tubulin polymerization by SDS-PAGE electrophoresis of soluble (S) and pelleted (P) fractions of centrifuged cell lysates, as described in Materials and Methods. Cells were treated with laulimalide and paclitaxel (A) or peloruside A (B) for 16 hours. The percentage soluble or pelleted tubulin is given below each lane. LAU, laulimalide; PTX, paclitaxel; PLA, peloruside A.

Using confocal microscopy, we showed that the microtubule-stabilizing effects of peloruside A and laulimalide were significantly impaired in the drug-resistant cell lines (Fig. 2). Whereas treatment of 1A9 cells with as little as 10 nmol/L laulimalide induced microtubule stabilization (bundling; Fig. 2A), this effect was only seen at much higher drug concentrations in L4 and R1 cells [≥300 nmol/L laulimalide for L4 (data not shown) and ≥50 nmol/L for R1]. With peloruside A, microtubule bundling in interphase cells and multiple asters in mitotic cells were observed at a peloruside A concentration of 40 nmol/L in 1A9 cells, but only at much higher concentrations in L4 and R1 cells (≥500 nmol/L or ≥200 nmol/L peloruside A, respectively; Fig. 2B). Paclitaxel induced microtubule aberrations at low concentrations in all 3 cell lines. Quantitative counts of cells with aberrant microtubule morphologies after treatment with microtubule-stabilizing agents are presented in the Supplementary Material (Supplementary Table S3). Taken together, these results corroborate and extend our cytotoxicity and cell-based tubulin polymerization assays and suggest that impaired drug–tubulin interaction mediates the resistance phenotype.

Figure 2.

Microtubule aberrations in 1A9, L4, and R1 cells treated with laulimalide, peloruside A, or paclitaxel. A, immunofluorescent staining of tubulin and DNA is presented in fixed cells following overnight treatment with laulimalide or paclitaxel. Arrows point to microtubule bundles. B, 1A9, L4, and R1 cells were treated with peloruside A and paclitaxel for 12 hours. The cells were fixed and stained with anti-α-tubulin antibody and 4′,6-diamidino-2-phenylindole nuclear stain. Staining was visualized by confocal microscopy. Arrows point to microtubule bundles. Both interphase and mitotic cells illustrate microtubule bundles and multiple asters, respectively. The images are representative of 4 independent experiments at each drug concentration.

We also evaluated the ability of peloruside A and laulimalide to induce cell cycle arrest in the resistant cells (Fig. 3). Again, we found that the resistant cells required higher concentrations of peloruside A and laulimalide, but not paclitaxel, to obtain a given level of G₂–M block.

Figure 3.

G₂–M block of 1A9, L4, and R1 cells by microtubule-stabilizing agents. Cells were treated with peloruside A, laulimalide, or paclitaxel for 16 hours at 37°C. Cell-cycle analysis was carried out by flow cytometry. Values represent the mean ± SEM of 3 to 6 independent preparations. PLA, peloruside A; LAU, laulimalide; PTX, paclitaxel.

Distinct tubulin mutations are identified in peloruside A- and laulimalide-resistant cell lines

The higher IC₅₀ values of peloruside A and laulimalide in the R1 and L4 cells prompted us to investigate whether a sequence alteration in the tubulin gene itself (13, 14) might account for the resistance of the cells. In addition, we reasoned that Pgp overexpression was an unlikely mechanism of resistance, as we did not observe any significant cross-resistance to paclitaxel, which is an excellent Pgp substrate (Table 1).

Given that the exact binding site for peloruside A or laulimalide on tubulin remains unknown and in light of modeling studies that suggest that peloruside A and laulimalide bind to α-tubulin (28–30), we first sequenced the predominant α-tubulin isotype in these cells, Kα1. No mutations in this gene were identified in either resistant cell line. Sequencing the predominant β-tubulin isotype, βI-tubulin, revealed distinct acquired mutations in both cell lines. L4 cells had acquired 2 single-point mutations, both of them at amino acid 306 that changed the wild-type R306 to either H306 or C306 (Supplementary Fig. S2A and B). Residues are numbered according to their positioning in the sequence of the human β2a tubulin isoform (UniProt accession number Q13885). RNA was extracted from the L4 cells and subcloned into bacteria. Individual bacterial clones were then picked and sequenced. This revealed that approximately 67% of the clones expressed histidine at position 306, whereas 33% expressed cysteine at 306. No trace of wild-type 306 was found. The R1 cell line harbored a single-point mutation, A296T, relative to the 1A9 parental cells (Supplementary Fig. S3). As with L4 cells, the R1 cells showed no expression of the wild-type β-tubulin allele at position 296, nor was there any evidence of a mutation at position 306 in the βI-tubulin of R1 cells.

Alterations in β-tubulin isotype composition

β-Tubulin isotype mRNA levels in 1A9, L4, and R1 cells

Next, we sought to determine the relative mRNA expression of β-tubulin isotypes in parental and resistant cells. The tubulin isotype levels were normalized to 18S rRNA, and their expression relative to parental cells was determined. In L4 cells, mRNAexpression of βII (class II)- and βIII (class III)-tubulin isotypes increased 16.8 ± 1.6-fold and 5.7 ± 0.8-fold, respectively (Fig. 4A and Supplementary Table S4). In contrast, the expression levels of these genes in R1 cells were similar to that in parental cells. The mRNA expression levels of the other isotypes, βI (class I), βV (class V), and βVI (class VI), were similar in the parental and the resistant cells. We were unable to amplify βIVa- and βIVb-tubulin despite trying various primer sets and different amplification conditions for the PCR. There are 2 likely reasons for this problem, low mRNA abundance or nonspecific primers.

Figure 4.

β-Tubulin isotype mRNA and protein expression in parental and resistant cells. A, mRNA expression of β-tubulin isotypes. The mRNA expression profile of β-tubulin isotypes in peloruside A and laulimalide resistant cells was analyzed using qRT-PCR. The relative fold expression of mRNA for each isotype is compared with the parental 1A9 cells (dashed line). Data are the mean ± SEM of 5 to 7 experiments. *, P < 0.05; **, P < 0.001, Student's t test. B, protein abundance in parental and resistant cells by immunoblotting. Total cellular protein (25 µg) was resolved on 10% SDS-PAGE and analyzed by Western blotting with βII-, βIII-, total β- and α-tubulin antibodies. β-Actin was used as a loading control. The image is a representative blot from 5 independent experiments. C, summary of the protein abundance data. Quantitation of immunoblot density in Western blots was determined relative to the actin loading control and presented as the fold expression relative to 1A9 cells. *, P < 0.05; **, P < 0.01, Student's t test. Data are the mean ± SEM (n = 5 experiments).

Tubulin protein levels in 1A9, L4, and R1 cells

The protein expression levels of the 2 β-tubulin isotypes, βII and βIII, and the 2 microtubule subunits, α- and β-tubulin, in the parental and the resistant cells were investigated by Western blotting (Fig. 4B and C). The total αand β-tubulin protein levels were similar in all 3 cell lines, but βII- and βIII-tubulin isotype protein expression was increased by 7.4 ± 1.2-fold and 5.6 ± 0.4-fold, respectively, in the resistant L4 cells (Fig. 4C). Conversely, βII- and βIII-tubulin isotype protein levels in R1 cells were similar to that in parental cells. The increased expression of βII and βIII isotypes in L4 cells was confirmed by immunocytochemistry (Supplementary Fig. S4).

Evidence for posttranslational modifications of tubulin isotypes

Posttranslational modifications of the βII- and βIII-tubulin isotypes in the resistant cells are illustrated in Fig. 5. In L4 cells, 1 additional protein spot of βII-tubulin (black arrow) and 2 additional protein spots for βIII-tubulin (white arrows), which were not present in the parental 1A9 cells, were detected. One faint additional protein spot for each of the βII- and βIII-tubulin isotypes was found in R1 cells, compared with 1A9 cells. These additional spots indicate posttranslational modifications of the tubulin isotypes in the resistant cells.

Figure 5.

Two-dimensional gel immunoblot analysis of β-tubulin isotypes. Protein lysates (80 µg) from 1A9, L4, and R1 cells were isoelectrofocused on a pH gradient strip (pH 3–5.6) and separated in the second dimension on a 4% to 12% PAGE. The proteins were transferred and probed with βII- and βIII-tubulin isotype–specific antibodies. The βII- and βIII-tubulin region (denoted by a rectangle on the gel image near the middle) is enlarged in the top right inset. Posttranslational modification of βII-tubulin (black arrows) and βIII-tubulin (white arrows) is indicated. Membranes were reprobed with β-actin to confirm equal protein loading. The image is a representative example from 3 experiments.

Discussion

βI-Tubulin structural alterations

To verify that the resistance of the L4 and R1 cells was due to altered interactions of peloruside A or laulimalide with tubulin, we examined the ability of peloruside A and laulimalide to induce tubulin polymerization in the resistant cells, form microtubule bundles or multiple asters, or block cells in G₂–M of the cell cycle. We then showed that L4 and R1 cells have a single-nucleotide mutation at amino acid positions 306 and 296, respectively, in the βI-tubulin gene. No mutations in the α-tubulin gene were found. Recent studies have shown that peloruside A and laulimalide bind to an exterior site on β-tubulin (8, 31, 32) and not on α-tubulin as originally proposed from computer modeling studies (28–30). For peloruside A-binding site modeling, Huzil and colleagues (8) used hydrogen–deuterium exchange mass spectrometry (HDX-MS) and found that there was a reduction in labeling on peptides β294–301 (H9–H9′ loop), β302–314 (H9′–S8), and β332–340 (H10 loop), suggesting that these regions of β-tubulin are involved in peloruside A binding. Importantly, the peptide β302–314 (H9′–S8) had a large reduction in deuteration and thus represents a strong candidate for residing in the peloruside A-binding site. This was further confirmed by data-directed molecular docking simulations. Modeling laulimalide binding, Bennett and colleagues (31) used mass shift perturbation analysis and data-directed docking, proposing that laulimalide binding promotes a reorganization of the R306 residue. This reorganization causes stabilization of the loops in this region and generates a polar contact with the oxygen of the dihydropyran side chain of laulimalide. Recently, Nguyen and colleagues (32) modeled both peloruside A and laulimalide into the β-tubulin site identified by Huzil and colleagues (8). They showed that the peptide containing amino acid β296 provided a favorable binding surface for the compounds. These results (8, 31, 32) show that the above amino acid residues are necessary for the stable binding of peloruside A and laulimalide to β-tubulin, but that they have no effect on taxoid site drug binding. Therefore, amutation in 306 (as seen in L4 cells) and 296 (as seen in R1 cells) could cause impaired drug-induced tubulin polymerization and, therefore, confer resistance to the drugs. Notably, amino acid R306 seems to be of major importance for peloruside A and laulimalide binding. This might explain why the L4 cell line shows a very rare mutation event in which 2 point mutations have occurred at the same location in both alleles. Alternatively, however, it is possible that L4 may comprise 2 cell populations, 1 with theH306 and 1 with the C306 mutation. In any case, the fact that there was no wild-type 306 left suggests that indeed this residue is important for laulimalide binding and that the cells had to mutate it to adapt to the drug selection pressure. The fact that L4 cells are also resistant to peloruside A provides strong support that peloruside A and laulimalide share the same binding site. Mutation of amino acid A296 has a selective effect on peloruside A activity with no impact on laulimalide activity, but the drug–amino acid interactions have not been specifically modeled at this stage, although alanine 296, like arginine 306, is located in the proposed laulimalide/peloruside A site.

These results are the first to provide cell-based support for a β-tubulin–binding site of peloruside A and laulimalide. All other evidence is based on computer-docking simulations and HDX-MS (8, 31, 32). The clinical relevance of βI-tubulin mutations in drug resistance is not clear, as such mutations are rarely seen in tumors (20, 21). This led us to search for other tubulin-related mechanisms that contribute to a cell’s distinct resistance profiles to peloruside A and laulimalide.

β-Tubulin isotype alterations

β-Tubulin in humans exists as 7 different isotypes: βI (class I), βII (class II), βIII (class III), βIVa (class IVa), βIVb (class IVb), βV (class V), and βVI (class VI; ref. 33). Some of these isotypes exhibit tissue-specific expression; whereas, other isotypes are constitutively expressed in all tissues (34, 35). The present study investigated mRNA expression of 5 of these isotypes in the resistant L4 and R1 cells. βII- and βIII-tubulin mRNAs were found to be increased in L4 cells but not R1 cells. The protein abundance of the isotypes in the cells mirrored the mRNA results, but there were no significant alterations in the total β-tubulin and total α-tubulin protein levels. Based on the qRT-PCR results, in 1A9 parental cells, βII and βIII mRNA levels were each 7% of the total mRNA (excluding βIVa/b) and in the resistant L4 cells, βII and βIII mRNA levels were increased to 55% and 19% of the total mRNA, respectively. Comparative protein levels in the cells could not be determined from the Western blots, as different antibodies were used for each isotype; however, it is expected that the βI isotype will still be the predominant β-isotype in the L4 cells. Overexpression of βII- and βIII-tubulin has been reported to be a predictive marker for drug resistance in cancer patients (13, 15). βIII-Tubulin is a multifunctional protein, and its upregulation is associated with tumor progression and resistance to tubulin-binding and DNA-damaging agents (36, 37). There is now a clear indication that βIII-tubulin plays a vital role in microtubule dynamic instability and in opposing the ability of tubulin-binding agents to suppress spindle dynamics. A recent study by Gan and colleagues (38) showed that βIII-tubulin knockdown in non–small cell lung cancer cells enhanced the suppression of microtubule dynamics at low concentrations of paclitaxel and vincristine. Given that microtubule polymerization and stabilization are the major modes of action of peloruside A and laulimalide, increased expression of βIII-tubulin presumably counteracts the action of these 2 drugs. The functional significance of βII-tubulin expression with regard to drug sensitivity differs from βIII-tubulin (13). It is not clear at this point what the role in resistance of βII-tubulin isotype overexpression is in L4 cells. Because the cells also have a mutation in the βI-tubulin gene, the resistance is likely to be due to a combination of decreased binding of peloruside A and laulimalide as a result of the βI-tubulin structural mutation and the effects of βII- and/or βIII-tubulin isotype overexpression.

Two-dimensional gel immunoblots revealed posttranslational modification of βII- and βIII-tubulin in L4 cells. Increased levels of tyrosinated α-, βIII-, and βIV-tubulin have been associated with paclitaxel resistance in MCF-7 breast cancer cells (39). The additional protein spots observed in the L4 and R1 cells, not seen in the 1A9 cells (Fig. 5), were in the low pH region of the isoelectric focusing but were at a similar molecular weight (50 kDa) as the main βII- and βIII-tubulin isotypes, indicating that the additional spots were likely to be due to posttranslational modifications such as phosphorylation or glutamylation. A study by Khan and Luduena (40) has shown that the phosphorylation of βIII-tubulin can regulate microtubule assembly in vivo. Phosphorylation of βIII-tubulin occurs at serine or tyrosine residues near the C-terminus (40). Tubulin modification at the C-terminus can affect the conformation of the tubulin protein, preventing the binding of a drug to tubulin (41). It would be of great interest to determine the precise posttranslational modifications of βII- and βIII-tubulin in L4 cells and their contribution to the resistance phenotype.

Overall, our results provide the first cell-based evidence in support of a β-tubulin–binding site for peloruside A and laulimalide that involves R306 for both compounds and A296 for peloruside A. We also showed that resistance to peloruside A and laulimalide may arise from altered drug–tubulin interactions as a result of not only βI-tubulin structural mutations but also increased βII- and βIII-tubulin isotype expression and posttranslational modification. Future directions of study will be to elucidate the functional significance of these β-tubulin mutations and isotype alterations in the resistant cells. Understanding the role of multiple β-tubulin alterations will help improve targeting of anticancer drugs whose mechanisms of action involve interactions with microtubules.