E-Letter responses to:

reports: Chava Kimchi-Sarfaty, Jung Mi Oh, In-Wha Kim, Zuben E. Sauna, Anna Maria Calcagno, Suresh V. Ambudkar, and Michael M. Gottesman A "Silent" Polymorphism in the MDR1 Gene Changes Substrate Specificity Science 2006; 0: 1135308v1 [Abstract]

E-Letters: Submit a response to this article

Published E-Letter responses:

Re: Reply to Wolfgang Sadée et al.

Anton A. Komar   (13 November 2007)

Reply to Wolfgang Sadée et al.

Chava Kimchi-Sarfaty, Zuben E. Sauna, Suresh V. Ambudkar, Michael M. Gottesman   (13 November 2007)

The Silent Polymorphism 3435C>T of MDR1

Wolfgang Sadée, Danxin Wang, Andrew D. Johnson   (13 November 2007)

Alternate Folds with Alternate Functions--The Silent but Not Invisible SNPs

Eileen K. Jaffe   (10 April 2007)

Can Codon Context Explain the Effects of “Silent Mutations” on Protein Structure and Function?

She-pin Hung   (23 March 2007)

Re: Reply to Wolfgang Sadée et al. 13 November 2007
Anton A. Komar Department of Biological Sciences, Cleveland State University, Cleveland, OH 44115 Respond to this E-Letter: Re: Re: Reply to Wolfgang Sadée et al.	W. Sadée et al., in their comment on a recent paper by C. Kimchi-Sarfaty and co-authors (1), argue that while the silent 3435C>T polymorphism in MDR1 gene might indeed be associated with altered substrate selectivity of the P-glycoprotein (P-gp), the paper did not address directly the hypothesis that codon usage affects translation and nascent protein folding. Sadée also calls the hypothesis "provocative" and argues that other mechanisms can influence translation. Sadée argues that the silent 3435C>T polymorphism of MDR1 can affect the function of the expressed protein by various mechanisms. In particular, Sadée and colleagues claim that the 3435C>T SNP can also affect MDR1 mRNA folding, and therefore stability, leading to altered P-gp expression (2). I would not characterize the hypothesis of Kimchi-Sarfaty et al.—that the kinetics of protein translation might affect protein folding—as "provocative." This idea has been under consideration for almost 20 years (3). Also, recent in silico modeling experiments (which were largely based upon this hypothesis) demonstrated that kinetic control may indeed have a significant impact on the final conformation of a protein during co-translational folding (4). Yet, the hypothesis is very difficult to prove experimentally, since ultimate proof would require validation of every constituent step of the process. Firstly, one would need to prove (for a given protein) that a synonymous codon substitution did change the local rates of translation. Secondly, clear demonstration that a change in translation rate led to an altered sequence of co-translational folding events and overall folding (conformation) of a given protein molecule would be required. Finally, one would need to show that the primary sequence of the protein did indeed remain unaltered and exclude the possibility (admittedly quite rare) of a miscoding event that could, in principle, result from a silent SNP (5). Assuming that a cell-based system is employed in these studies—the most experimentally rigorous approach—it is quite obvious that it would be nearly impossible to address all constituents of the hypothesis with sufficient precision and confidence. Kimchi-Sarfaty et al. showed in their paper (and Sadée seems to agree with their conclusion) that the 3435C>T SNP contributes to the change in substrate selectivity of P-gp, without any significant change in the protein expression levels. The most reasonable conclusion is that the conformation of the protein has been altered. Therefore, at least one part of the hypothesis has solid support. Importantly, if a synonymous codon substitution at a given mRNA not only changes a codon to one of low-frequency usage, but also alters the folding of the mRNA, then which one of these two phenomenon would have a greater impact on the kinetics of protein translation? It was originally proposed that both the mRNA secondary structure (6) and codon usage (7 ,8) contribute to the non-uniformity of translational elongation rate. However, careful analysis has demonstrated that after translational initiation (particularly in eukaryotes) the ribosome can, in most cases, locally destabilize secondary structures and move along the message without any significant delays (9). This argues that codon usage is the primary cause of nonuniformity in translational elongation. Yet, very complex secondary structures (e.g. those involving pseudoknots) can stall elongating ribosomes, producing frameshifts or significant reductions in protein expression levels (10, 11). A reduction of protein expression levels was not observed by Kimchi-Sarfaty et al., which argues against the presence of complex mRNA structure capable of stalling ribosomes. Sadée argues that he and his colleagues "had shown that 3435C>T affects mRNA folding…"; however, they do not provide compelling evidence for this claim. In their 2005 study (2) the authors used the very popular mfold program (12) to predict putative RNA structures. However, this type of analysis cannot be considered proof that the predicted RNA structures actually form, nor that nucleotide substitutions would produce subsequent changes in RNA structure. To support their claim careful chemical and enzymatic probing, followed by mutational analysis, of the corresponding mRNAs would be necessary. It is well known that in silico predictions and experimental realities sometimes clash. For example, the structures of the 5’UTR of the mRNA for the arginine/lysine transporter cat-1 as predicted by mfold [Fig. 1C; in (13)] had very little in common with those obtained by chemical and enzymatic probing [Fig. 3C; in (14)]. There are numerous other examples that illustrate this point. In conclusion, the claim that Kimchi-Sarfaty et al. "fail to resolve the underlying mechanism…" has not been rigorously substantiated, particularly since the alternative mechanism (mRNA folding) proposed by Sadée et al. has yet little experimental evidence to support it. Anton A. Komar Department of Biological Sciences, Cleveland State University, 2121 Euclid Avenue, Cleveland, OH 44115, USA. References 1. C. Kimchi-Sarfaty et al., Science 315, 525 (2007). 2. D. Wang et al., Pharmacogenet Genomics 15, 693 (2005). 3. I. J. Purvis et al., J. Mol. Biol. 193, 413 (1987). 4. F. P. Huard, C. M. Deane, G. R. Wood, Bioinformatics 22, e203 (2006). 5. J. Precup, A. K. Ulrich, O. Roopnarine, J. Parker, Mol. Gen. Genet. 1218, 397 (1989). 6. W. G. Chaney, A. J. Morris, Arch. Biochem. Biophys. 194, 283 (1979). 7. S. Pedersen, EMBO J. 3, 2895 (1984). 8. F. Bonekamp, H. D. Andersen, T. Christensen, K. F. Jensen, Nucleic Acids Res. 13, 4113 (1985). 9. S. A. Liebhaber, F. E. Cash, S. H. Shakin, .J Biol. Chem. 259, 15597 (1984). 10. H. Kontos, S. Napthine, I. Brierley, Mol. Cell Biol. 21, 8657 (2001). 11. A. G. Nackley et al., Science 314, 1930 (2006). 12. M. Zuker, Nucleic Acids Res. 31, 3406 (2003). 13. J. Fernandez et al., J. Biol. Chem. 277, 2050 (2002). 14. J. Fernandez et al., Mol. Cell. 17, 405 (2005).
Reply to Wolfgang Sadée et al. 13 November 2007
Chava Kimchi-Sarfaty Laboratory of Cell Biology, National Cancer Institute, 37 Convent Drive, Bethesda, MD 20892, USA, Zuben E. Sauna, Suresh V. Ambudkar, Michael M. Gottesman Respond to this E-Letter: Re: Reply to Wolfgang Sadée et al.	We agree with W. Sadée et al. that the primary goal of this study (1) is to highlight the importance of synonymous SNPs contributing to changes in the function of P-glycoprotein. Although our results lead us to the conclusion that the observed phenotype may be a consequence of the local rate of translation being affected, as we note in our paper, this is by no means the only explanation. Understanding the underlying mechanisms of the phenomenon we have described is a difficult problem that we are currently addressing and trust that others will too. Sadée suggests that a major conclusion of our paper is that "3435C>T affects the rate of translation." On the contrary, we determined that there were no significant functional differences between the wild-type and the 3435C>T SNP. It was only when the 3435C>T SNP was studied in the context of the haplotype that we observed substantial functional differences. Testing each of the SNPs separately and the haplotype C1236T-G2677T revealed function similar to that of the wild-type. These results suggest that the SNP in position 3435 is necessary but not sufficient to alter the protein function, and that the non-synonymous SNP (G2677T) is neither necessary nor sufficient to produce this phenotype. While our paper does address the hypothesis that codon usage affects the final conformation of the protein by directly substituting a rare codon and demonstrating even more marked changes in inhibitor sensitivity, it does not explicitly test the hypothesis that codon usage affects the rate of translation. The vaccinia infection/transfection system that was used in our study results in high expression levels, but not higher than those seen in normal tissues such as adrenal cortex. However, cells can be analyzed only up to ~24-28 hours post-infection/transfection. We did not detect the small differences in mRNA levels described in Sadée's work, but the systems are quite different: i.) Wang et al. (2) studied the haplotype only in liver cells, whereas we used a transient expression system. ii.) When Wang et al. used a transient expression system (which was different from our own), they studied only the SNP 3435. iii.) Differences in mRNA expression by Wang et al. were relatively small, 1.5-fold. iv.) Most important, and certainly the key observation in our studies, is that we show qualitative differences in the function and substrate specificity of the haplotype that are independent of the level of expression. The expression levels of the wild-type, each of the SNPs (C1236T, G2677T and C3435T), as well as the haplotype combinations of these SNPs were very similar. The determination of similar P-gp levels for the various haplotypes included both cell surface expression using the specific cell surface monoclonal antibody, MRK16, and estimating total P-gp expression by immunoblots using the monoclonal antibody C219. Other recent studies (3–5) have shown that the mechanisms suggested by Sadée et al., including differential mRNA processing, mRNA folding and stability should all affect protein expression. Although our results don't prove that altered codon usage results directly in changes in nascent protein folding, two independent approaches (detection using a conformation-sensitive antibody and trypsin sensitivity) show clearly that the final conformation of the protein on the cell surface is altered. The functional consequence of this conformational change is accentuated when more DNA template is transfected, consistent with (but not proof of) the hypothesis that the rare codons substituted in the haplotype are limiting during cotranslational folding. We hope that these results stimulate others to explore a role for synonymous codons in altered phenotypes of proteins. Further research is clearly needed to elucidate the underlying mechanistic aspects. Chava Kimchi-Sarfaty, Zuben E. Sauna, Suresh V. Ambudkar, and Michael M. Gottesman Laboratory of Cell Biology, National Cancer Institute, 37 Convent Drive, Bethesda, MD 20892, USA. References 1. C. Kimchi-Sarfaty et al., Science 315, 525 (2007). 2. D. Wang et al., Pharmacogenetics and Genomics 15, 693 (2005). 3. A. D. Johnson, D. Wang, W. Sadée, Pharmacol. Ther. 106, 19 (2005). 4. M. Jia, Y. Li, FEBS Letters 579, 5333 (2005). 5. A. G. Nackley et al., Science 314, 1930 (2006).
The Silent Polymorphism 3435C>T of MDR1 13 November 2007
Wolfgang Sadée College of Medicine and Public Health, The Ohio State University, Columbus, OH 43210, USA, Danxin Wang, Andrew D. Johnson Respond to this E-Letter: Re: The Silent Polymorphism 3435C>T of MDR1	C. Kimchi-Sarvaty et al. (1) show that a 'silent' polymorphism (3435C>T) in the MDR1 gene unexpectedly changes substrate selectivity even though protein sequence is unchanged. They propose that 3435C>T affects the rate of translation because of changes in codon usage, resulting in altered folding of the nascent protein. To test this provocative hypothesis, the authors have measured substrate/inhibitor selectivity of wild-type MDR1 and variant haplotypes carrying 3435C>T embedded in a 3-SNP haplotype, including 1236C>T (also synonymous) and 2677G>T (nonsynonymous). While 3435C>T is associated with altered substrate selectivity in their experiments, no results directly address the hypothesis that codon usage affects translation and nascent protein folding. Additional mechanisms can influence translation, for example, changes in mRNA folding (2). We had shown that 3435C>T affects mRNA folding and alters mRNA turnover (3), an event not detectable in their experiments because mRNA expression was measured only at 24 hours after transfection. As to changes in nascent protein folding, Kimchi-Sarvaty et al. have analyzed functions of only mature protein after transfection, which appear to change with overall expression level (rather than rate of translation). Kimchi-Sarvaty et al. also show striking differences in protease digestion of MDR1 and the variant (at high expression levels); however, use of the 3-SNP haplotype with 1 nonsynonymous SNP confounds the interpretation that the synonymous C3435T is responsible. Numerous processes can affect protein conformation after the initial folding of the nascent protein—including saturation of protein binding partners. In summary, Kimchi-Sarvaty et al. highlight the potential importance of synonymous SNPs, often neglected in genetic studies, but fail to resolve the underlying mechanism. Moreover, evidence is lacking that the proposed mechanism contributes to an association between genotype and MDR1 activity in human subjects, under the native condition of gene expression in vivo. Wolfgang Sadée, Danxin Wang, and Andrew D. Johnson Department of Pharmacology, Program in Pharmacogenetics, College of Medicine and Public Health, The Ohio State University, 333 West 10th Avenue, Graves Hall, Columbus, Ohio 43210, USA. References 1. C. Kimchi-Sarfaty et al., Science, 315, 525 (2007). Published online 21 December 2006. 2. A .D. Johnson, D. Wang, W. Sadée, Pharmacol Ther. 106, 19 (2005). 3. D. Wang, A. D Johnson, A. C Papp, D. L. Kroetz, W. Sadée, Pharmacogen. Genomics 15, 693 (2005).
Alternate Folds with Alternate Functions--The Silent but Not Invisible SNPs 10 April 2007
Eileen K. Jaffe Fox Chase Cancer Center. Philadelphia, PA 19111 Respond to this E-Letter: Re: Alternate Folds with Alternate Functions--The Silent but Not Invisible SNPs	Anfinsen’s principle states that protein sequence provides the information necessary to specify the native three-dimensional protein structure (1). Implicit in this principle is the assumption that there is only one physiologically relevant native structure. Small structural variations are considered to be “conformational changes,” and dynamic conformational equilibria are widely appreciated as essential to protein function. Stable alternate conformations are generally associated with post-translational modifications. Hence, it was noteworthy that synonymous SNPs that cause altered translational kinetics can yield identical proteins with different folds and different functions (2, 3). The altered conformation reported for MDR1 can be considered to be an example of a metastable protein conformation, first predicted by Nickerson and Day (4). The reported haplotype MDR1 is metastable because it is long-lived, but can be readily drawn to the wild-type conformation by interaction with a drug substrate (3). We have recently described another metastable protein conformational phenomenon as “morpheeins” (5) and herein propose a potential relationship to synonymous SNPs. Morpheeins are homo-oligomeric proteins wherein alternate conformations of the monomer dictate different oligomeric assemblies with distinct functionalities. Interconversion between alternate morpheein forms requires dissociation, conformational change, and reassembly. For human porphobilinogen synthase, the morpheein forms are low-activity hexamers and high-activity octamers. The equilibration of these species is linked to substrate turnover, and the dysequilibrium is disease associated (6-8). For morpheeins to function in a physiologically relevant fashion, the alternate assemblies must be close in energy, and the equilibration of forms must respond to a small event, such as ligand binding, not unlike the response of the haplotype P-gps to drug binding. The crux is that the alternate metastable conformations are close in energy but not in a dynamic equilibrium under all conditions. Events like ligand binding, altered folding kinetics, altered pH, or catalysis can trigger the reequilibration of forms. I propose that there are many examples of native metastable protein conformations, some of which may be morpheeins, some of which may result from synonymous SNPs, and some of which may join both phenomena, thus helping explain how a single protein sequence can function differently in different environments or different individuals. References 1. C. B. Anfinsen, Science 181, 223 (1973). 2. A. A. Komar, Science 315, 466 (2007). 3. C. Kimchi-Sarfaty et al., Science 315, 525 (2007). 4. K. W. Nickerson, R. A. Day, Curr. Mod. Biol. 2, 303 (1969). 5. E. K. Jaffe, Trends Biochem. Sci. 30, 490 (2005). 6. L. Tang, L. Stith, E. K. Jaffe, J. Biol. Chem. 280, 15786 (2005). 7. L. Tang et al., J. Biol. Chem. 281, 6682 (2006). 8. E. K. Jaffe, L. Stith, Am. J. Hum. Genet. 80, 329 (2007).
Can Codon Context Explain the Effects of “Silent Mutations” on Protein Structure and Function? 23 March 2007
She-pin Hung, Director of Molecular Biology CODA Genomics, Inc., 26061 Merit Circle, Suite 101, Laguna Hills, CA 92653 Respond to this E-Letter: Re: Can Codon Context Explain the Effects of “Silent Mutations” on Protein Structure and Function?	In their Report, “A ‘silent’ polymorphism in the MDR1 gene changes substrate specificity” (27 Jan., p. 525), C. Kimchi-Sarfaty et al. suggest that the effects of a synonymous SNP (silent mutation) in the MDR1 gene on the structure and function of the P-glycoprotein (P-gp) are due to the generation of a slowly translated rare codon that affects the timing of co-translational folding and insertion of P-gp into the membrane. This translational slowing might also be explained by nearest neighbor codon context effects. Independent of codon usage, codon-pair utilization in all organisms examined is highly biased, reflecting over- and under-representation of many codon pairs compared with their random expectations (1). Observations that these codon pair biases are directional and restricted to nearest neighbors, and experimental evidence that tRNA-tRNA interactions on the surface of a translating ribosome influence translational elongation rates, have led many to suggest that codon pair biases reflect the compatibilities of adjacent tRNA isoacceptor molecules on the surface of a translating ribosome. Gutman and Hatfield (2) proposed that the frequency of one codon next to another co-evolved with the structure and abundance of tRNA isoacceptors in order to control translational step times without imposing constraints on amino acid sequences or protein structures. Later, Irwin et al. (3) demonstrated that silent mutations can change rapidly translated codon pairs into slowly translated codon pairs and vice versa. Now Kimchi-Sarfaty et al. have presented evidence that silent mutations affect co-translational protein folding (4). Thus it is likely that, in addition to codon usage, codon context also could explain the interesting effects of synonymous SNPs on the structural and functional properties of the MDR1 gene product. References 1. G. A. Gutman, G. W. Hatfield, Proc. Nat. Acad. Sci. U.S.A. 86, 3699 (1989). 2. G. W. Hatfield, G. A. Gutman, “Codon Pair Utilization Bias in Bacteria, Yeast and Mammals,” in Transfer RNA in Protein Synthesis D. L. Hatfield, B. J. Lee, R. M. Pirtle, Eds. (CRC Press, Boca Raton, FL, 1993). 3. B. Irwin, J. Heck , G. W. Hatfield, Biol. Chem. 270, 22801 (1995). 4. P. Cortazzo et al., Biochem. Biophys. Res. Commun. 293, 537 (2002).