Cotranslational Protein Folding Kinetics and the Degeneracy of the Genetic Code
Jill Simard, Professor Crane, Pomona College, March 2007
Introduction
It wasn’t long after Beadle and Tatum published their “one gene, one enzyme”
hypothesis in 1941 that exceptions began to emerge (1). In particular, it was discovered
that RNA splicing and post-translational protein modifications can lead to different
proteins from a single transcribed sequence. However, until recently it was widely held
that a single mRNA encoding a specific amino acid sequence will be translated to give
identical protein products, assuming that the co- and post-translational modifications are
the same. That is, it was assumed that a protein’s overall structure and consequent
function was dictated solely by its amino acid sequence. This became known as
Anfinsen’s principle (2). While a large number of proteins continue to support Anfinsen’s
principle, in recent years we have also seen cases in which varying rates of translation
can affect protein cotranslational folding and therefore protein structure and behavior,
even if the primary sequences are held constant. Varying rates of translation in vivo can
be attributed to a wide range of factors, such as variations in tRNA and transcription
factor concentrations, as well as changes in codon usage (3). This last factor is
particularly interesting and will be the primary focus of this paper; it suggests that codons
previously thought to be synonymous can in fact have different effects on protein
conformation. Thus, “silent” mutations may not necessarily be truly silent, and the
degeneracy of the genetic code is more relevant to biological function than was
previously thought. This recent revelation has profound consequences both for
experimental protein expression and for the protein misfolding that has been shown to
lead to a wide range of conformational diseases.
Genetic code degeneracy (a review) and single nucleotide polymorphisms (SNP’s)
Because each codon contains three nucleotides and there are four possible
nucleotides, sixty-four different codons are possible. There are only twenty different
amino acids to encode, so in most cases multiple codons can encode a single amino acid.
As seen in Figure 1, the codons corresponding to a certain amino acid frequently differ by
only one nucleotide, making it possible for single nucleotide polymorphisms (SNPs) or
synonymous nucleotide substitutions to occur without changing the primary structure of a
protein. Cases involving a synonymous single nucleotide substitution are often called
“silent mutations,” and one popular explanation for the degeneracy of the genetic code
has been that such “silent” mutations decrease the prevalence of amino acid sequence
alterations, which could ultimately affect protein efficacy (4).
Figure 1. The genetic code and its degeneracy.
Cotranslational Protein Folding and Kinetics
Codon Usage
It has been shown that E. Coli and other organisms exhibit strong codon biases;
for example, if there exist four possible codons for a single amino acid, often one of those
codons will be present with significantly greater frequency. It has also been seen that
concentrations of cognate tRNAs are directly proportional to codon usage frequency (5).
Thus, it is often the case that the more rare the codon, the more slowly it is translated.
Furthermore, researchers have found a tendency for genes to contain clusters (of around
ten or less codons) of either frequent or rare codons (6), and the rare codon clusters tend
to encode turns, loops, and domain linkers within the translated protein (7). These
studies, along with supporting research, have lead to the now widely-held belief that the
genetic code’s degeneracy affects cotranslational protein folding by controlling the rate
of translation at certain regions. Such controlled translational rates can separate the
partial folding events that occur during cotranslational protein folding, which make
certain conformations in the final protein more probable.
Experimental support that codon changes can lead to changes in protein structure,
behavior, and function.
In the past decade a significant number of cases have been uncovered in which the
use of a non-wild-type synonymous codon (which would not change the protein’s
primary structure) alters the resulting protein’s normal folding. In many cases this altered
folding as been shown to also result in altered protein behavior and function. Studies
have been conducted on both prokaryotic and eukaryotic systems.
One recent study was performed using the EgPABP1 gene, which was expressed
and studied in E. coli. The resulting protein contains a short region of five codons
connecting two α helices, as indicated in Figure 2. In the wild-type, the five codons
present are rare; the researchers changed certain nucleotides to make seven variant
sequences of five frequent codons that were synonymous to the originals. They then
employed an in vivo stress response-induced protein-folding reporter to examine any
misfolding of the mutants, and also analyzed differences in solubility using SDS-PAGE
and Western blotting.
Figure 2. Two views of the EgPABP protein (rotated 90°). The arrows correspond to
a turn region between two α helices (amino acids 22-26). This region contains rare
codons in the wild-type; frequent codons were substituted.
In one of the seven variants (variant 6, sequence shown in Table 1), both
decreased solubility and aggregation were indicated, while it was confirmed that the
variant in fact had the same amino acid sequence as the wild-type. The researchers
hypothesize that in this particular variant, the accelerated translation rate altered the
initial folding steps of the protein such that post-translational protein folding was also
perturbed. The misfolding presumably exposed more hydrophobic residues, leading to a
change in solubility (8). It is unknown what factors in particular lead to a marked
difference in variant 6’s folding, while the behavior of the other variants remained
unchanged.
Table 1. Nucleotide sequences in the EgFABP1 wild-type and variants.
Variant
Wild
type
1
2
3
4
5
6
7
Sequence
F27
E21
R22
L23
G24
V25
D26
GAA
GAA
GAA
GAA
GAA
GAA
GAA
GAA
CGC
CGT
CGT
CGT
CGT
CGT
CGC
CGT
CTT
CTT
CTT
CTT
CTT
CTG
CTG
CTG
GGG
GGG
GGT
GGG
GGG
GGG
GGT
GGT
GTG
GTG
GTG
GTT
GTG
GTG
GTT
GTT
GAT
GAT
GAT
GAT
GAC
GAT
GAT
GAC
TTC
TTC
TTC
TTC
TTC
TTC
TTC
TTC
Komar and colleagues (9) have published research similar to that of the study just
described, replacing rare codons with frequent codons in a protein expressed in E. coli,
but this study in addition found evidence that the silent mutations affecting translational
rates also impacted protein activity. Here the CATIII (chloramphenicol aminotransferase
III) gene was used, and within a certain region 16 synonymous codon substitutions were
made, as shown in Figure 3A. To examine translational kinetics, the group used SDSPAGE analysis—the longer a certain mRNA region spends on the ribosome, the greater
the amount of “nascent peptide chains of the sizes corresponding to location of their
respective rare codons.” Therefore, the darker the band on the gel electrophoresis, the
slower that region was translated. Figure 3B shows not only that the kinetics of
translations in non-uniform, but also indicates that substitution with more frequent
codons in the mutant resulted in faster translation. This is particularly evident between
the 133-136 kDa region. By using published data on individual codons’ translation rates,
the researched had predicted that their synonymous substitutions would result in a twofold decrease in told total translation rate of their gene. In fact, the phosphorimager scan
data of the SDS gel (Figure 3C) indicates in 2.7-fold decreased total translation time in
the mutant.
Figure 3. A: Wild-type and mutant CATIII sequences. B: SDS gel electrophoresis
autoradiogram of CAT nascent chains. Column 1 is the mutant and column 2 is the
wild-type, both after 3 min. of translation. Molecular mass in kDa. C:
Phosphorimager scan of the products shown in B.
At different stages in translation, aliquots of the wild-type and the synonymous-codon
mutant (run under the same conditions) were removed in order to test for activity. As
Figure 4 shows, the mutant exhibits a 20% decreased activity.
Figure 4. Comparison of the specific activity of chloramphenicol acetyltransferase
wild-type and mutant enzymes in an E. coli S30 extract cell-free system. The results
are the averages of 5 trials. Specific activity is defined as the ratio of substrate
conversion to the total amount of the radioactively labeled substrate at any given
time.
Further evidence of codon usage’s importance in biological protein function has
been published just this year (10), this time involving synonymous SNPs in the Multidrug
Resistance 1 (MDR1) gene coding for P-glycoprotein, an ATP-driven efflux pump. The
protein plays an important role in drug pharmacokinetics and cancer cell drug resistance,
and extensive research has been conducted on its numerous polymorphisms. Using
multiple cell systems, researchers have shown that an SNP in certain MDR1 haplotypes
alters the substrate specificity of P-glycoprotein. Their results from FACS analysis
indicate that while the transport function does not change amongst the wild-type and
polymorphic P-glycoproteins, the efficacy of the P-glycoprotein inhibitors verapamil and
cyclosporine A decreases when the P-glycoprotein is a polymorph containing the SNP in
question. They also found that the differences in efficacy between the wild-type and the
haplotypes became more pronounced as more P-glycoprotein was translated. While they
have not yet experimentally explored why this occurs, it is thought that as certain tRNA’s
become more depleted (as translation goes on for a longer period of time), codon usage
becomes more important.
Treatment of both the haplotypes and the wild-type with trypsin indicates that the
altered function in the presence of inhibitors is related to a conformational change
amongst the variants. The wild-type and the haplotypes differed in their susceptibility to
trypsin—the haplotype required a 3.4 fold greater concentration for 50% degradation—
which suggests a discrepancy in tertiary structure. As in the other experimental cases, the
SNP studied involved a change between rare and frequent codons: the wild-type codon
GGC, exhibits 34% usage; the SNP produces the codon GGT, which has 16% usage. This
data, as before, points to the significance of translational kinetics in protein folding.
There are a few other ways in which synonymous codon usage could affect
protein folding that has not been thoroughly tested and supported, but are nevertheless
worth mentioning. One is the role of mRNA secondary structure. SNPs change the
nucleotide sequence and could therefore change the way in which the mRNA typically
folds, possibly altering the way it interacts with the ribosome during translation. Another
thought is that synonymous codon choice could alter structurally important interactions
between the charged tRNA’s and either the ribosome or the nascent peptide (11).
Relevance of the effect of the genetic code’s degeneracy on protein folding
Consequences for experimental work on proteins
Given that SNPs have been shown to alter protein structure and function, it would be
wise to pay more attention to the nucleotide sequence used for expression of a certain
protein. The environment in which the protein is translated would also seem to be
important, given what has recently been discovered about codon usages, tRNA
concentrations, and translation rates. Each species has its own codon usage frequency,
and the tRNA concentrations between species (and therefore codon translational rates)
vary accordingly. Thus, in cases of heterologous gene expression, misfolding can occur
due to the fact that the codon translational rates of the host are different from those in the
gene’s natural organism (3).
Consequences for pharmacotherapy
Kimchi-Sarfaty et al.’s work on synonymous SNPs in the MDR1 gene (10), which was
discussed earlier, highlights significant variations in the resulting protein’s response to
certain drugs. As the synonymous SNPs are occur naturally, it can safely be assumed that
there are significant variations in how people react to certain P-glycoprotein inhibitors.
As pharmacotherapy becomes increasingly personalized and tailored to an individual’s
genes, the effects of synonymous SNPs should be considered.
The potential role of “silent” mutations in protein misfolding and conformational
diseases
When a polypeptide folds into a non-functional conformation, there are several biological
pathways that exist to help the polypeptide achieve the functional conformation.
Competing with these pathways, however, are protein degradation and aggregation, both
of which can lead to a vast number of pathologies. Aggregation of misfolded proteins has
been implicated in Alzheimer’s disease, Creutzfeld-Jakob disease, and sickle cell anemia;
degradation has been linked to cystic fibrosis, familial hypercholesterolemia, and a vast
array of other diseases (12). Thus far many of the mutations known to cause these
pathologies are missense mutations. However, given the ability of synonymous codon
substitutions to affect protein folding, “silent” mutations should be considered, as well.
The research in this field is recent enough that few examples are available in which
synonymous single nucleotide substitutions have led to pathologies. Nevertheless, there
are multiple cases in which missense and nonsense mutations have failed to account for
many misfolding cases. For example, Kimchi-Sarfaty, et al., note that missense and
nonsense mutations account for only 60% of the cases of pseudoxanthoma elasticum, a
disease caused by a mutation in the MPR6 gene (10).