A "Silent" Polymorphism in the MDR1 Gene Changes Substrate Specificity

See allHide authors and affiliations

Science  26 Jan 2007:
Vol. 315, Issue 5811, pp. 525-528
DOI: 10.1126/science.1135308


Synonymous single-nucleotide polymorphisms (SNPs) do not produce altered coding sequences, and therefore they are not expected to change the function of the protein in which they occur. We report that a synonymous SNP in the Multidrug Resistance 1 (MDR1) gene, part of a haplotype previously linked to altered function of the MDR1 gene product P-glycoprotein (P-gp), nonetheless results in P-gp with altered drug and inhibitor interactions. Similar mRNA and protein levels, but altered conformations, were found for wild-type and polymorphic P-gp. We hypothesize that the presence of a rare codon, marked by the synonymous polymorphism, affects the timing of cotranslational folding and insertion of P-gp into the membrane, thereby altering the structure of substrate and inhibitor interaction sites.

The MDR1 gene product, the adenosine triphosphate (ATP)–binding cassette (ABC) transporter ABCB1 or P-gp, is an ATP-driven efflux pump contributing to the pharmacokinetics of drugs that are P-gp substrates and to the multidrug resistance of cancer cells (1, 2). To date, more than 50 single-nucleotide polymorphisms (SNPs) have been reported for MDR1 ( One of these, a synonymous SNP in exon 26 (C3435T), was sometimes found to be associated with altered P-gp activity (36) and, when it appears in a haplotype, with reduced functionality (7). This association may be explained in different ways. Perhaps it is because C3435T is in linkage disequilibrium with other common functional nonsynonymous polymorphisms such as G2677T. In fact, the C1236T (a synonymous SNP), G2677T, and C3435T polymorphisms are part of a common haplotype (8, 9). Another possible explanation is that allele-specific differences in mRNA folding could influence splicing, processing, or translational control and regulation (10, 11). A third possibility is that the effect of the C3435T polymorphism on the levels of cell surface P-gp activity or its function is rather modest or drug-specific. Finally, numerous environmental factors are known to affect the expression and phenotypic activity of P-gp (12).

To determine whether the C3435T polymorphism actually does affect P-gp activity, we expressed wild-type and polymorphic P-gps in HeLa cells with the use of a transient expression system (13). The same experiments were carried out on BSC-1 (epithelial cells of African green monkey kidney origin), Vero-76 (monkey kidney cells), and 12E1 (CEM human cells) cell lines (14), with similar results, indicating that this phenomenon is not specific to HeLa cells.

Assays for P-gp's transport function with the fluorescent substrates Rhodamine 123 (Rh123), bodipy-FL-paclitaxel, bodipy-FL-verapamil, daunorubicin, bodipy-FL-vinblastine, and calcein-AM (14, 15) were performed on HeLa cells expressing the MDR1 wild-type; polymorphisms at C1236T, G2677T, or C3435T; and haplotypes consisting of these polymorphic variant combinations: C1236T-G2677T, C1236T-C3435T, G2677T-C3435T, and C1236T-G2677T-C3435T. The functions of P-gp for all single-polymorphism plasmids as well as for wild-type MDR1, as measured by intracellular accumulation or by efflux of fluorescent compounds, were not distinguishable under standard conditions (14). HeLa cells expressing double- and triple-haplotype mutants also revealed results similar to those for the single mutants (Fig. 1, A to C). However, the P-gp inhibitors cyclosporin A (CsA) and verapamil (fig. S1) were less effective against all the substrates in cells expressing the double or triple haplotypes carrying C3435T relative to the wild type, the SNPs, or the haplotype that does not carry C3435T. Thus, it is not the presence of the nonsynonymous polymorphism G2677T that results in the phenotype, but rather the presence of C3435T in combination with one or two of the other polymorphisms.

Fig. 1.

Drug transport function of wild-type MDR1 and seven MDR1 SNPs and haplotypes. The drug efflux from vaccinia virus infected/transfected HeLa cells was determined by FACS analysis. Cells were transfected with pTM1 (control; purple), pTM1-MDR1 (wild-type P-gp; green), C1236T (pink), G2677T (lavender), C3435T (orange), C1236T-G2677T (blue), C1236T-C3435T (yellow), G2677T-C3435T (light blue), and C1236T-G2677T-C3435T (red). (A) 0.5 μM Rh123; (B) 0.1 μM bodipy-FL-paclitaxel; (C) 0.5 μM bodipy-FL-verapamil. (D to F) Effect of plasmid DNA concentration during infection/transfection on Rh123 efflux (0.5 μM) in the presence of an inhibitor, 10 μM CsA; infected/transfected DNA, (D) 3 μg, (E) 10 μg, (F) 15 μg.

We next tested to see whether these differences correlated with the concentrations of transduced plasmid DNA. The expression and function of all transduced cells were measured by fluorescence-activated cell sorting (FACS) with MRK16 monoclonal antibody (mAb) staining and by Rh123 in the presence of CsA, respectively (14). The differences in inhibition by CsA and Rh123 between the cells expressing wild-type MDR1 and the haplotype C1236T-G2677T-C3435T were more distinct as the concentration of the DNA increased (Fig. 1, D to F). These data suggest that the differences were more pronounced at higher levels of mRNA where more P-gp was being translated in the cells. The expression levels of P-gp from the vaccinia infection/transfection system and cells of normal human adrenal glands were found to be comparable (fig. S2).

Figure S1, C to E, shows that the haplotypes including C3435T had altered susceptibility to verapamil, but not to rapamycin (fig. S1F) (14). When the cells were incubated with the inhibitors before adding the fluorescent substrates, as opposed to simultaneous incubation with the drugs, the same pattern was observed. Bodipy-FL-verapamil, wild-type P-gp, and the haplotype (C1236T-G2677T-C3435T) exhibited different accumulations in a concentration-dependent manner, suggesting a change in affinity (fig. S3).

Synonymous SNPs or mutations can cause inactivation of the native splicing donor site, which results in a premature stop codon (16) or exon skipping, yielding a shorter mRNA. A previous report indicated that the polymorphism C3435T resulted in decreased levels of mRNA expression (17). We therefore compared mRNA levels (14) in the wild-type and haplotype (C1236T-G2677T-C3435T) with the use of real-time quantitative reverse transcription polymerase chain reaction (RT-PCR), which revealed equivalent mRNA levels (Fig. 2A). Western blots using C219 mAb showed comparable total cell expression of P-gp, and the infected/transfected cells expressed the same levels of P-gp as determined by FACS assays and immunohistochemical staining with MRK16 mAb (Fig. 2, B and C). This result was reproduced in different cell lines including BSC-1, Vero-76, and 12E1 (fig. S4). The complete amino acid sequence of MDR1 haplotype C1236T-G2677T-C3435T protein was identical to the predicted sequence.

Fig. 2.

mRNA levels and P-gp expression in the vaccinia expression system. (A) Analysis of pTM1 only, wild-type MDR1, and the haplotype C1236T-G2677T-C3435T (3X) with real-time quantitative RT-PCR. Crossing-point values for the graph on the left are plotted in the histogram. (B) Assessment of cell surface expression, using MRK16 mAb of all nine constructs as described in Fig. 1. (C) Confocal assessment of MDR1 expression, using MRK16 mAb with fluorescein isothiocyanate–conjugated secondary antibody of pTM1 (control; left panel), pTM1-MDR1 (wild-type P-gp; middle panel), and C1236T-G2677T-C3435T (right panel). (D) Immunoblot analysis of pTM1 only, wild-type MDR1, and the haplotype C1236T-G2677T-C3435T (2 μg protein/lane) with C219 mAb (14). The mature fully glycosylated (∼170 kD) andimmatureP-gpbands (∼150kD) are marked by arrows (19).

We hypothesized that a conformation difference between wild-type and haplotype P-gp might explain these results. Indeed, UIC2 (14), a conformation-sensitive mAb, alone or in combination with CsA or vinblastine at 37°C, revealed pronounced differences in binding consistent with altered conformations in the haplotype (fig. S5) (18). To determine whether there are subtle differences in the folding of wild-type and haplotype P-gp, we compared their relative susceptibility to trypsin. Figure 3 shows the disappearance of the P-gp band as a function of trypsin concentration. The concentration required for 50% degradation (IC50, here expressed as μg trypsin) was greater for haplotype P-gp than for wild-type P-gp by a factor of about 3.4; this result implies that the two have slightly different tertiary structures. Both wild-type and haplotype P-gps had comparable IC50 (μg trypsin) values in the presence of verapamil, which suggests that the altered conformation can be corrected by drug interaction with P-gp. The immature, coreglycosylated form of P-gp (150-kD band) was more sensitive to trypsin than the mature, glycosylated form, consistent with (19), by a factor of 5. However, the ratios (wild-type:haplotype) of the IC50 (μg trypsin) values were comparable for the mature and immature bands (3.86 μg versus 4.4 μg). Thus, it is unlikely that altered glycosylation is responsible for the functional differences observed.

Fig. 3.

Determining the sensitivity of wild-type and the haplotype C1236T-G2677T-C3435T P-gp to trypsin. Crude membranes prepared from vTF7-3 infected/transfected HeLa cells expressing wild-type MDR1 or the haplotype C1236T-G2677T-C3435T were treated with increasing concentrations of trypsin and the disappearance of the P-gp band was quantified as described above. (A) Experiment performed in the absence of verapamil; IC50 =2.1 μg (wild type), 7.1 μg (C1236T-G2677T-C3435T). The mature (170 kD) and immature (150 kD) P-gp bands were also analyzed separately; IC50 = 0.68 μg (wild-type immature), 2.9 μg (haplotype immature), 2.8 μg (wild-type mature), 10.8 μg (haplo typemature). (B) Same experiment in the presence of 30 μM verapamil; IC50 = 3.7 μg (wild type), 3.3 μg (C1236T-G2677T-C3435T). Values for the mature and immature P-gp bands: IC50 = 2.5 μg (wild-type immature), 2.5 μg (haplotype immature), 3.6 μg (wild-typemature), 3.2 μg (haplotype mature). Immunoblots with C219 mAb are shown at the bottom.

The use of rare codons appears to influence the translation rate, which in turn affects protein folding (2025), with the third base in the codon having the largest effect (26). We hypothesize that as the cell produces more P-gp (Fig. 1, D to F), the role of codon usage may become more critical when certain tRNA species become depleted. The codon usage for the SNP at position 12/1236 with GGC changed to GGT (both encode Gly) changes from 34% [relative synonymous codon usage (RSCU), 22.4] to 16% (RSCU, 10.8). The SNP at position 21/2677 that changes GCT (Ala) to TCT (Ser) also uses a less common codon (26% to 18%; RSCU values change from 18.5 to 15.1). The SNP at position 26/3435 that changes the codon from ATC (Ile) to ATT (Ile) reduces the codon usage from 47% to 35% (RSCU values change from 20.9 to 15.8). Clusters of rare codon usage (table S1) occur both upstream and downstream of each of these SNPs. Codon usage rates are similar in humans and monkeys, which explains the similarity in the results with all transduced cells (27).

To test whether codon usage compromises P-gp function, we introduced C3435A (isoleucine codon usage for ATA is 18%, RSCU 7.4) to produce the haplotype C1236T-G2677T-C3435A. Functional assays using bodipy-verapamil or Rh123 in the presence of digoxin (Fig. 4, A and B) showed even larger decreases in inhibitor effects between this haplotype and the common haplotype C1236T-G2677T-C3435T. Moreover, use of Rh123 in the presence of fexofenadine revealed median fluorescence of 26.9 for the wild type, 24.3 for C1236T-G2677T-C3435T, and 20.3 for C1236T-G2677T-C3435A. The median fluorescence in the presence of paclitaxel and fexofenadine was 38.2 for the wild type, 28.6 for C1236T-G2677T-C3435T, and 22.9 for C1236T-G2677T-C3435A.

Fig. 4.

Drug transport function of wild-type and two MDR1 haplotypes. The drug efflux of vaccinia infected/transfected HeLa cells was determined by FACS analysis (14). Cells were transfected with pTM1 (control; purple), MDR1, (wild-type P-gp; green), C1236T-G2677T-C3435T (red), and C1236T-G2677T-C3435A (brown). (A) 0.5 μM bodipy-FL-verapamil in the presence of 500 μM digoxin; (B) 0.5 μM Rh123 in the presence of 150 μM digoxin.

The amino acid sequence of proteins is generally believed to determine protein expression, folding, and function; mutations that alter the primary structure of a protein can affect these properties. The important question addressed by this study is the role of silent mutations (i.e., those that do not affect amino acid sequence) in protein folding and function. Recent theoretical studies have suggested that codon usage is not random, and experimental studies in prokaryotes suggest that this may be so (28). Here we show that a silent mutation in a complex, mammalian membrane transport protein alters the substrate specificity. We hypothesize that when frequent codons are changed to rare codons in a cluster of infrequently used codons, the timing of cotranslational folding is affected (29) and may result in altered function. This finding may be clinically important. For example, mutations in the MRP6 (ABCC6) gene cause the disease pseudoxanthoma elasticum, but missense and nonsense mutations are found in only about 60% of cases (30), raising the possibility that mutations that do not change coding sequence may contribute to disease by a similar mechanism.

Supporting Online Material

Materials and Methods

Figs. S1 to S5

Table S1


References and Notes

View Abstract

Navigate This Article