• chaperone buffering;
  • compensatory mutations;
  • directed evolution;
  • kinetic stability;
  • protein engineering;
  • protein stability;
  • thermodynamic stability


  1. Top of page
  2. Abstract
  3. Introduction
  4. Protein functional expression and stability
  5. Promoting evolvability by modulating protein stability
  6. Methods to identify stabilizing/compensatory mutations
  7. Buffering strategies to modulate the stability threshold
  8. Combining compensatory mutations with buffering in directed evolution
  9. Perspectives
  10. Acknowledgements
  11. References

Protein engineering is widely used to generate proteins with novel or enhanced function. However, manipulating protein function in the laboratory can prove laborious, protracted and challenging. Recent developments in the understanding of protein evolutionary dynamics have unveiled the full extent by which the evolution of function is limited by protein stability - a revelation that may be applied to protein engineering on a whole. Thus, strategies that modulate protein stability and reduce its constraining effects may facilitate the engineering of protein function. A combinatorial approach involving the introduction of compensatory mutations and manipulation of the stability threshold by chaperone buffering during directed evolution can improve the functional adaptation of a protein, thereby fostering our ability to attain ever-more ambitious protein functions in the laboratory.


B-factor iterative test




tobacco etch virus


  1. Top of page
  2. Abstract
  3. Introduction
  4. Protein functional expression and stability
  5. Promoting evolvability by modulating protein stability
  6. Methods to identify stabilizing/compensatory mutations
  7. Buffering strategies to modulate the stability threshold
  8. Combining compensatory mutations with buffering in directed evolution
  9. Perspectives
  10. Acknowledgements
  11. References

The engineering of proteins with novel or enhanced function can provide many useful tools for biotechnology and medicine. A wide variety of methods have been developed, a number of successful examples have been reported, and many engineered proteins have entered into the commercial market [1, 2]. However, in most cases, protein engineering is still a challenging task that often results in dead ends [3]. Nature, with its multitude of complex functions, has been far more successful. What can be learned from nature that will improve the success of protein engineering? Understanding the evolutionary dynamics of proteins has helped protein engineers to develop new strategies that aim to efficiently alter protein function in the laboratory [3-5].

A recent finding revealed that modulating protein stability can help facilitate the modification of function [6-10]. In the most simplistic model, protein fitness, W, defines a protein's capacity for performing its functional role (i.e. the flux of an enzyme catalyzed reaction). Protein fitness is therefore reflective not just of the level of function (i.e. kcat/KM for an enzyme), f, but also the concentration of functional protein, [E]0:

  • display math(1)

The concentration of functional protein ([E]0) is largely influenced by that protein's stability. Although only a small fraction of mutations would change f (i.e. < 20% of residues comprise the active site of an enzyme and many mutations outside of it are largely neutral to function), any mutation has the potential to affect stability and thereby, [E]0. The role that stability plays in protein evolution is therefore critical. Indeed, many of the single nucleotide polymorphisms that cause genetic diseases appear to mostly affect protein stability, expression and aggregation, rather than actual function [11, 12].

The threshold model has been frequently used to describe protein stability in evolution (Fig. 1) [6, 8, 10, 13, 14]. To have functional expression in a cell, the stability of a protein must be maintained above a certain threshold level. Variants with stability lower than the threshold lose expression as a result of increased unfolding, misfolding, aggregation and degradation. Because most mutations compromise protein stability, a loss of stability is a major hurdle for the development of new function (Fig. 1A) [6, 15, 16]. A margin of stability from the threshold is therefore indicative of a protein's mutational tolerance and thereby, its evolvability [10]. In the directed evolution of a function, starting from a marginally stable protein, let alone a poorly expressed one, can be a major cause of evolutionary dead ends [7, 17]. Mechanisms by which protein stability is modulated and/or the stability threshold is manipulated may therefore comprise effective strategies for enhancing protein engineering.


Figure 1. Threshold model for protein stability in evolution [6, 10]. Black arrows indicate function-altering mutations that would be allowed to occur under standard conditions; red arrows indicate function-altering mutations that would be disallowed by the normal stability threshold; blue arrows indicate the acquisition of compensatory and stabilizing mutations. (A) Mutations that reduce the stability of a protein beyond a certain threshold are purged out because they cause loss of functional expression and result in significantly reduced fitness. (B) By accumulating compensatory/stabilizing mutations, previously inaccessible destabilizing mutations may be incorporated under standard conditions. (C) Chaperone buffering reduces the threshold by ensuring proper folding of proteins and allows for the incorporation of previously inaccessible mutations. (D) When used together, compensatory mutations and chaperone buffering allow for sustainable functional adaptation by modulating the protein's stability and reducing the negative effects of function-conferring mutations.

Download figure to PowerPoint

In the present review, we discuss strategies aiming to efficiently engineer and evolve soluble, stable proteins in the laboratory. We describe how soluble and functional expression is related to stability, summarize the recent developments in protein engineering for the modulation of protein stability and the stability threshold, and finally, we present a strategy that combines these methods to expedite long-term functional adaptation by directed evolution.

Protein functional expression and stability

  1. Top of page
  2. Abstract
  3. Introduction
  4. Protein functional expression and stability
  5. Promoting evolvability by modulating protein stability
  6. Methods to identify stabilizing/compensatory mutations
  7. Buffering strategies to modulate the stability threshold
  8. Combining compensatory mutations with buffering in directed evolution
  9. Perspectives
  10. Acknowledgements
  11. References

Stability plays a significant role in determining the level of a protein's functional and soluble expression in the cellular milieu. However, protein stability is a broad and ambiguous term. Thermodynamic stability [ΔGN – U; i.e. the difference in free energy between the native (N) and unfolded (U) states] is often used to describe the relationship between expression and stability in protein evolution (Fig. 2A) [6, 8, 10, 13, 14]. Indeed, it has been shown in some cases that thermodynamic stability does in fact dictate the level of soluble and functional expression in the cell [18, 19]. However, this may only apply to certain proteins, such as those that are small and fold quickly [20]. Growing evidence indicates that kinetic stability plays a significant role in determining the level of functional protein because it reflects the energy levels of folding intermediates between a protein's folded native state, unfolded states and/or misfolded states, which in turn can significantly control the rate of a protein's folding/unfolding, aggregation and degradation (Fig. 2B) [20-22]. Recent examples have shown that, for such proteins, thermodynamic stability does not necessarily correlate with the abundance of protein in the cell and/or with organismal fitness [23-25] (K. T. Wyganowski, M. Kaltenbach and N. Tokuriki, unpublished data). Therefore, for protein engineering, it is important to determine whether a target protein's expression is more related to thermodynamic or kinetic stability, as this will help determine the appropriate course of action for directed evolution (see below).


Figure 2. Simplistic schemes of protein folding in vitro and in vivo. (A) Thermodynamic stability is defined by the difference in free energy between the native (N), unfolded (U) states. The amount of native and folded proteins is simply dictated by the free energy difference between U and N. (B) The folding of a protein in vivo is considerably more complicated because the protein can access a number of possible fates (misfolding, degradation and aggregation). The level of functional protein is influenced by both thermodynamic and kinetic stability.

Download figure to PowerPoint

Promoting evolvability by modulating protein stability

  1. Top of page
  2. Abstract
  3. Introduction
  4. Protein functional expression and stability
  5. Promoting evolvability by modulating protein stability
  6. Methods to identify stabilizing/compensatory mutations
  7. Buffering strategies to modulate the stability threshold
  8. Combining compensatory mutations with buffering in directed evolution
  9. Perspectives
  10. Acknowledgements
  11. References

By increasing the inherent stability of a protein or by lowering the stability threshold, new evolutionary trajectories, which would normally be inaccessible as a result of a mutation's destabilizing effect, may be permitted. To date, two major strategies to modulate stability and access these pathways have been proposed.

Increasing protein stability with stabilizing/compensatory mutations

Mutations that increase the stability of proteins can compensate for the destabilizing effect of other mutations and widen the margin of stability from the threshold, thus allowing the protein to tolerate further destabilizing mutations that may lead to new or enhanced function (Fig. 1B). Using P450 as a model system, Bloom et al. [7] demonstrated that a highly thermostable protein could tolerate more destabilizing mutations and possessed higher evolvability than a marginally stable protein. The importance of compensatory mutations in evolution has been also demonstrated with a variety of other proteins [26-31].

Lowering the stability threshold by chaperone buffering

Overexpression of molecular chaperones, such as GroEL/ES, can buffer the destabilizing effect of mutations by lowering the stability threshold itself, thereby facilitating the folding of compromised proteins (Fig. 1C). We examined the buffering capability of GroEL/ES for several proteins, such as Escherichia coli glyceraldehyde 3-phosphate dehydrogenase and Pseudomonas diminuta phosphotriesterase (PTE). GroEL/ES overexpression doubled the number of mutations that could accumulate neutrally in terms of protein fitness and greatly increased the variability of accumulated mutations in a neutral drift experiment (i.e. mutation-accumulation experiments where selection is applied to maintain the enzyme's current expression levels and function) [9]. Moreover, the improvement of a new enzymatic activity for the enzyme, PTE, was facilitated under chaperone overexpression conditions: twice as many improved variants were obtained, and the new activity was determined to be more than 10-fold higher than in variants evolved in the absence of GroEL/ES overexpression [9].

Although both strategies were confirmed to be practical for promoting protein evolvability, the molecular mechanisms and evolutionary dynamics underlying each strategy differ fundamentally. Compensatory mutations create permanent effects because they intrinsically alter protein stability. They can affect thermodynamic and/or kinetic stability to increase functional expression in the cell at the same time as being global (i.e. compensate for a range of mutations throughout the protein) or local (i.e. specific for a given functional mutation). However, an excess of stability does not provide any advantage to the protein; it may even be deleterious in some cases [6]. Chaperone buffering, on the other hand, is temporal (i.e. buffering occurs only when chaperones are present) but can provide an immediate advantage by buffering the destabilizing effects of new mutations that would decrease the protein's stability below the threshold. Buffering directly affects kinetic stability by controlling protein folding, although it does not alter thermodynamic stability. It also has a limited capacity per protein (i.e. rescue is limited to a certain level of destabilization) and per cell (i.e. each cell can handle only a limited number of compromised protein molecules at a time).

For protein engineering, it is crucial to know which strategy to use, as well as how and when to incorporate them into the overall experimental process. Although each of the two mechanisms has specific advantages and limitations, they may be used to complement each other to minimize these limitations and optimize the success of the experiment.

Methods to identify stabilizing/compensatory mutations

  1. Top of page
  2. Abstract
  3. Introduction
  4. Protein functional expression and stability
  5. Promoting evolvability by modulating protein stability
  6. Methods to identify stabilizing/compensatory mutations
  7. Buffering strategies to modulate the stability threshold
  8. Combining compensatory mutations with buffering in directed evolution
  9. Perspectives
  10. Acknowledgements
  11. References

Both computational and experimental methods have been developed to identify compensatory mutations that stabilize a protein. Computational methods predict possible mutations based on the analysis of the ever-increasing amount of available sequence and structure data (Table 1). The experimental approaches generate mutagenized gene libraries from which variants can be screened and selected based on their ability to confer stability (Table 2). Both approaches are highly complementary; computational methods can predict stabilizing mutations with a reasonable error rate, which significantly reduces the library size required for experimental identification compared to traditional, randomly mutagenized libraries [32]. Thus, many recent successful examples have employed a combination of computational prediction and experimental identification [32, 33].

Table 1. Overview of recent examples where computational methods were employed to predict stabilizing mutations. [Table 1 was corrected on 28 June 2013 after original online publication]
Target proteinInputaMutagenesis methodbScreening methodcOverall resultReference
  1. a

     Initial data used in the method to predict stabilizing mutations. The number in brackets indicates the number of homologue sequences used.

  2. b

     Method by which mutations were introduced within the target protein's gene.

  3. c

     The screening method employed to experimentally validate the predicted stabilizing mutations (RA, residual activity after exposure to heat; HLI, half-life of inactivation at a given temperature/time; TD, thermal denaturation measured by CD).

Sequence-basedConsensus-guided mutagenesisCellobiohydrolase II (Phanerochaete chrysoporium)Homologue sequences (5)Inverse PCR methodRAA variant with all advantageous mutations combined could retain activity at 50 °C for > 72 h [46]
Pectate lyase (Xanthomonas campestris)Homologue sequences (4)Site-directed mutagenesisHLI (45 °C)A mutant with a 23-fold improvement of its half-life of inactivation at 45 °C was obtained [47]
Glucose dehydrogenase (Bacillus subtilis)Homologue sequences (12)Overlap extension PCRRAA triple-mutant with a half-life of ~ 20 min at 25 °C extended to ~ 3.5 days at 65 °C [44]
Penicillin G acylase (Escherichia coli)Homologue sequences (7)Site-directed mutagenesisHLI (50 °C)Two single-mutants were isolated with a three-fold longer half-life at 50 °C than wild-type [43]
B-lactamase (Enterobacter coacae)Homologue sequences (38)DNA shuffling with mutant oligonu-cleotidesTDA variant with eight mutations had a Tm 9.1 °C higher than wild-type [42]
Ancestral sequence methodB-amylase (Bacillus circulans)Homologue sequences (69)Site-directed mutagenesisTDOne variant with five mutations had a Tm 4.1 °C above wild-type [51]
Glycyl-tRNA synthetase (Thermus thermophilus)Homologue sequences (all available)Site-directed mutagenesisTDFive single-mutants and one double-mutant with Tm values 1.4–4.7 °C higher than the wild-type [50]
3-Isopropylmalate dehydrogenase (Thermus thermophilus)Homologue sequences (18)Site-directed mutagenesisRAA single-mutant with a half-denaturation time at 86 °C approximately three-fold longer than wild-type [49]
Isocitrate dehydrogenase (Caldococcus noboribetus)Homologue sequences (154)Site-directed mutagenesisRAA variant with five mutations had a 10 min half inactivation temperature 2.5 °C above the wild-type [48]
Structure-basedComputational algorithms (PoPMuSiC)Tobacco etch virus protein – proteaseProtein structure in PDB formatSite-directed mutagenesisTDA double-mutant was obtained that had a Tm 2.6 °C above wild-type and its solubility improved from 1.5 mg·mL−1 to > 40 mg·mL−1 [67]
B-FITLipase Lip2 (Yarrowia lipolytica)B-factor values calculated from X-ray structural information of enzymeSaturation mutagenesisRATwo single-mutants, one with a two-fold improvement and one with five-fold improvement in half-life at 50 °C over wild-type, were obtained [69]
Lip A (Bacillus subtilis)B-factor values calculated from X-ray structural information of enzymeSaturation mutagenesisRAAfter several more rounds of iterative saturation mutagenesis, two variants were obtained with half-lives at 55 °C of 905 and 980 min, compared to wild-type which was < 2 min [68]
CombinatorialComputational algorithms (foldx) and consensus guided methodCellobiohydrolase I Cel7A (engineered)Homologue sequences (41) and protein structureSite-directed mutagenesisHLI (10 min)A combination of identified mutations was used to raise T50 4.7 °C above wild-type [71]
Consensus guided method and B-FITα-amino ester hydrolase (Xanthomonas compestris)Homologue sequences (7) and X-ray structural information of enzymeSite-directed mutagenesisRAA quadruple-mutant with a temperature at which the half-life is 30 min that was 7 °C above wild-type [70]
Table 2. Overview of recent examples where stabilizing mutations were identified by experimental screening. IvAM, in vivo assembly of mutant libraries with different mutational spectra. GFP, green fluorescent protein. [Table 2 was corrected on 28 June 2013 after original online publication]
Target proteinMutagenesis methodaReporterbOverall resultReference
  1. a

     Method by which mutations were introduced within the target protein's gene.

  2. b

     Reporter protein used in reporter assay method to measure level of soluble expression of the target protein.

Screening native state stabilityMeasuring thermostabilityAlkaline protease (Bacillus gibsonii)Sequence saturation mutagenesisA recombined quadruple-mutant with a half-life at 15 °C 100-fold longer than wild-type [73]
Phytase (Yersinia mollaretii)Sequence saturation mutagenesis and error-prone PCROne variant (five mutations) with a Tm 1.5 °C above wild-type [74]
Microbial transglutaminase (Streptomyces mobaraensis)Saturation mutagenesis and DNA shufflingTriple-mutant with 12-fold higher half-life at 60 °C compared to wild-type [75]
Ligninolytic oxidoreductases (Saccharomyces cerevisiae)Mutagenic staggered extension process and DNA shuffling with IvAMBest high redox potential laccase had 10-fold imrpovement in maximal activity at 72 °C over wild-type and the best versatile peroxidase had three-fold improvement in maximal activity at 65 °C over wild-type [76]
ProsideTEM-1 β-lactamase (Escherichia coli)Error-prone PCRCombined several mutations found to create a variant with nine mutations and a Tm approximately 20 °C above wild-type [84]
β1 domain of protein G (Streptococcus sp.)Error-prone PCR and saturation mutagenesisCombined best mutations to create a variant with a Tm 35.1 °C above wild-type [83]
Bacterial cold shock protein B (Bacillus subtilis)Error-prone PCR and saturation mutagenesisCombined mutations to create a quadruple mutant with a Tm 31.2 °C above wild-type [82]
Screening for soluble expressionMeasuring cell lysate activityKemp eliminase (computationally designed)Error-prone PCR, DNA shuffling, site-directed mutagenesisA variant with eight mutations showed significantly increased catalytic efficiency (> 200-fold) [86]
Serum paraoxonase PON3DNA shuffling and error-prone PCRA variant with solubility increased from < 2 mg·mL−1 to approximately 25 mg·mL−1 [27]
Simvastatin synthase – lovDSite-directed mutagenesisA double-mutant that increased solubility approximately 50% [88]
Reporter assayβ1 domain of protein G (Streptococcus sp.)Overlap extension PCRFluorescence (split-GFP system)Variant with Tm increased by 12 °C over wild-type [96]
Glucocorticoid receptor ligand-binding domain (Homo sapiens)Error-prone PCRFluorescence (GFP-fusion)A final variant with 26-fold improvement in yield and 8 °C improvement in Tm over wild-type [97]
Immunity protein 7Error-prone PCRAntibiotic resistance (target protein inserted into TEM-1 β-lactamase loop)Obtained several variants with increased thermostability, resistance to proteolysis, solubility and slower unfolding [91]
N-carbamoyl-d-amino acid amidohydrolase (Burkholderia pickettii)Error-prone PCR and DNA shufflingColorimetric assay (structural complementation with β-galactosidase)Combined mutations into a triple-mutant that was three-fold more soluble than wild-type [98]

Computational methods for the prediction of stabilizing mutations

Sequence-driven prediction

Stabilizing mutations can be predicted by analyzing the frequency of amino acids that appear within a collection of extant homologues and/or predicted ancestral proteins. Several sequence-based methods that predict stabilizing mutations have been described [32], although the two most commonly employed techniques are the consensus method and the ancestral mutation method. In the consensus method, a single sequence composed of the most frequently occurring residue at each position amongst a collection of homologues is generated, residues that differ between the consensus and the target protein are identified, and the residues in the protein are mutated ‘back-to-consensus’ [34]. Alternatively, in the ancestral mutation method, a phylogenetic tree is constructed from extant homologues and probabilistic models are used to infer the sequences of its ancestors. Residues that differ from the ancestor and the target protein are then mutated back to their ancestral form [35]. In many cases, the prediction of back-to-consensus and ancestral mutations will give the same result, although it is also capable of yielding unique mutations. Both back-to-consensus and ancestral mutations have been shown to improve both thermodynamic and kinetic stability [30, 36, 37].

Both approaches may be useful in the prediction of stabilizing mutations, as shown by the many successful experiments reported for the consensus method [26, 38-47] and the ancestral mutation method [48-51] (Table 1). Consensus-guided mutagenesis was recently applied to cellobiohydrolase II using five related fungal enzyme sequences [46]. Of the 44 back-to-consensus mutations identified, 16 conferred higher thermostability while maintaining an approximately equivalent level of activity. By accumulating all 16 mutations in one gene, an enzyme was obtained with a T50120 value (the temperature required to reduce the initial activity by 50% in 120 min) 5.4 °C above the wild-type [46]. Yamashiro et al. [51] also recently used the ancestral mutation method to engineer a thermostable β-amylase. From the putative ancestral sequence, 22 predicted mutations were identified and seven were shown to have higher thermostability than the wild type, with the best variant having a two-fold increase in half-denaturation time at 60 °C [51].

However, there is no guarantee that either method will prove fruitful in every case. The elongation factor EF-Tu was recently used as a model to directly compare the different sequence-based methods currently available. Starting from the wild-type protein with a Tm of 39.1 °C, the best mutant derived from the consensus method had a Tm of 60.2 °C, whereas the ancestral mutation method construct yielded a variant with a Tm of 35.1 °C. This demonstrates how challenging predicting stabilizing mutations can be because not every method will work in any given situation and multiple methods may have to be attempted before a successful result is realized [32]. Even in the successful examples, < 40% of predicted mutations provided higher stability.

Structure-based prediction

Various prediction algorithms have been developed to compute the effect of mutations on thermodynamic stability (ΔΔG) based on three-dimensional protein structures [52-64]. In general, these methods can predict trends fairly well, although they tend to deviate on the specific ΔΔG value for each single mutation [64-66]. However, they still can prove to be a valuable asset for predicting stabilizing mutations.

One such algorithm, PoPMuSiC, was used to identify eight potential thermodynamically stabilizing mutations at five unique positions in TEV protease [67]. TEV protease was targeted for improvement because it is important in biotechnology for the cleavage of fusion proteins, although it proves to be poorly soluble (approximately 1 mg·mL−1), resulting in less than optimal application. To increase the chances that one of these predicted stabilizing mutations would also increase solubility, the most polar mutation at each of the five surface-exposed positions was considered and experimentally examined in the hope that the mutations would optimize favourable interactions with the solvent. Although four of these five mutations increased the Tm between 1 and 2 °C (with the other being approximately neutral), two were found to significantly improve the solubility of the protein to over 40 mg·mL−1 [67].

Another structure-based approach is the B-factor iterative test (B-FIT), which is based on the observation that flexible regions of proteins compromise overall stability [68]. Several flexible sites, defined by high B-factor values in the crystal structure of a protein, are subjected to iterative saturation mutagenesis to introduce rigidity and increase stability. B-FIT was used to identify two mutations in Lip2, a lipase from Yarrowia lipolytica, which raised the half-life of thermal inactivation at 50 °C from 1.97 min to 5.4 and 10.2 min, respectively [69]. When combined, they increased the half-life to 13.9 min. B-FIT has also been successfully employed to increase the thermostability of Lip A, a lipase from Bacillus subtilis [68], and an α-amino ester hydrolase [70] (Table 1).

Although structure-based predictions of ΔΔG are becoming more reliable, a potential pitfall of these methods is that they focus on stabilizing the native state. Therefore, the predictions do not address the effects on folding intermediates and as such, these methods may not be effective at predicting mutations that increase soluble expression for proteins requiring greater kinetic stabilization. Developing tools to accurately predict the effect of mutations on kinetic stability would be of great use. However, this currently remains a challenge.

Combined prediction strategies

Combinations of multiple tools can enhance the overall accuracy of the predictions. For example, the consensus method was used in conjunction with a structural-based algorithm, foldx, to identify mutations that would increase the thermostability of a fungal cellobiohydrolase I [71]. Although the consensus method alone proved to be 21% accurate (3 of 14 correctly identified stabilizing mutations) and foldx alone was 24% accurate (9 of 38), their combined accuracy was 33% (three of nine). Alternatively, different methods may produce unique results, which could lead to greater overall success. The consensus method was used with B-FIT to improve the thermostability of an α-amino ester hydrolase [70]. Both B-FIT and the consensus method predicted unique mutations that produced a variant with a half-life at 30 min, 7°C above the wild-type.

Experimental methods for the identification of stabilizing mutations

Experimental approaches for identifying stabilizing mutations mainly rely on the screening of large pools of diversified variants. They can be classified into two categories based on which property is being screened for. The first is screening for native state stability and is based on techniques that quantify the stability of proteins after they are expressed. The second category is screening for soluble expression level, which reflects the expression and proper folding of the targeted protein.

Screening for the stability of the native state

One method for experimentally identify potential compensatory mutations involves screening variants for their thermostability, which is reflective of their native-state stability. Variants in cell lysate are exposed to heat and the ratio of the residual activity to the initial activity is taken as a measure of each variant's thermostability. This method was used to identify mutations that raised the T5010 value (the temperature at which half the protein is denatured over 10 min) of a cytochrome P450 peroxygenase 15 °C above the initial enzyme [72]. Additionally, because this method can be easily applied to any enzyme with a measurable activity, it has been used to engineer more stable variants of an alkaline protease [73], phytase [74], transglutaminase [75], oxidoreductase [76] and esterase [77] (Table 2).

Alternatively, the ability of a protein to withstand proteolysis can also correlate with native-state stability. In Proside (protein stability increased by directed evolution), the protein of interest is inserted between two domains of a capsid protein required for phage propagation, subjected to heat or a denaturant, and exposed to proteases [78, 79]. The more stable variants can resist unfolding and subsequent proteolysis, and thereby maintain the phage's infectivity [80]. This method has been used to successfully increase the stability of several proteins, including a bovine pancreatic trypsin inhibitor [81], the cold shock protein CspB [82], the β1 domain of protein G [83] and TEM-1 β-lactamase [84] (Table 2).

Although there are many successful examples of screening for native-state stability, a caveat is that thermostability or resistance to proteolysis in some proteins may not necessarily correlate with the level of functional and soluble expression. Such an example was reported for a terpene synthase when selection for thermostability, including resistance to proteolysis, resulted in a variant that denatured at 83 °C, which is significantly higher than the wild-type enzyme at 38 °C. However, the ‘improved’ variants were expressed in inclusion bodies, whereas the original enzyme could be expressed in a soluble form [25]. Recently, studies of PTE also showed that there is no correlation between thermostability of variants and their solubility (K. T. Wyganowski, M. Kaltenbach and N. Tokuriki, unpublished data).

Screening for the soluble expression level of proteins

To ensure the identification of a variant with increased stability resulting in soluble expression, the activity of the variants in cell lysate can be measured. This is because cell lysate activity is a product of both specific activity (f) and the concentration of functional protein ([E]0). Although a mutation increasing f can occur, it is also highly likely that mutations increasing [E]0 will be found. Indeed, lysate screening of libraries from a poorly expressed protein often results in increasing solubility and functional expression level ([E]0) through the acquisition of stabilizing and compensatory mutations. However, this method requires an additional step in which the increased soluble expression level is verified by other methods such as SDS/PAGE. This method was performed with PON3, a serum paraoxanase, to improve its solubility from < 2 mg·mL−1 to approximately 25 mg·mL−1 with six stabilizing mutations [27]. It has also been used in several other cases [28, 85-88] (Table 2).

Alternatively, screening for soluble expression can also be carried out using a reporter protein. In these assays, the target protein's solubility directly relates to the function of a reporter protein, such as green fluorescent protein [89], β-galactosidase [90], dihydrofolate reductase [91] or an antibiotic resistance protein [92] (Table 2). Initially, the reporters were fused to the C-terminal end of the target protein, although a wide range of insertion and structural complementation methods are now available [90, 93-98]. For example, mutants of an immunity protein, Im7, were inserted into a loop region of TEM-1 β-lactamase and expressed in vivo. Those variants that confer higher soluble expression will also allow for greater β-lactamase activity, thereby increasing the host cell's resistance to antibiotics. Of 31 mutations obtained from the selection of variants with higher antibiotic resistance, 27 showed decreased unfolding rates, 26 had increased thermodynamic stability and two had greater resistance to degradation by proteinase K [93]. This method is particularly advantageous for those proteins that do not have a measurable function themselves and would be unavailable for direct screening.

Buffering strategies to modulate the stability threshold

  1. Top of page
  2. Abstract
  3. Introduction
  4. Protein functional expression and stability
  5. Promoting evolvability by modulating protein stability
  6. Methods to identify stabilizing/compensatory mutations
  7. Buffering strategies to modulate the stability threshold
  8. Combining compensatory mutations with buffering in directed evolution
  9. Perspectives
  10. Acknowledgements
  11. References

To date, the only chaperone system that has been employed to buffer destabilizing effects in directed evolution is GroEL/ES (see above) [9]. However, it should be noted that GroEL/ES recognizes only a certain set of proteins (10% of cytosolic proteins in E. coli) [99, 100]. The buffering capacity of GroEL/ES for nonsubstrate protein appears to be negligible [101]. Alternatively, other chaperones could be employed, such as DnaK/J (HSP70/40) [102, 103], chemical chaperones [104, 105] or any combination therein, because joint overexpression of a subset of chaperones can extensively expand the repertoire of assisted proteins [106]. Genetic fusion to a solubilizing tag protein, such as maltose-binding protein, glutathione S-transferase or thioredoxin, might also be an effective strategy for increasing the solubility of variants and buffer destabilizing mutations [107]. For additional buffering, favourable expression conditions, such as reduced temperature, weaker induction and attenuated promoter strength, can be combined with chaperone co-expression.

Combining compensatory mutations with buffering in directed evolution

  1. Top of page
  2. Abstract
  3. Introduction
  4. Protein functional expression and stability
  5. Promoting evolvability by modulating protein stability
  6. Methods to identify stabilizing/compensatory mutations
  7. Buffering strategies to modulate the stability threshold
  8. Combining compensatory mutations with buffering in directed evolution
  9. Perspectives
  10. Acknowledgements
  11. References

Compensatory and buffering strategies are highly complementary to each other. Combining both strategies can further enhance the evolvability of proteins and enable sustainable protein engineering through the avoidance of evolutionary dead ends (Fig. 1D). Recently, we have employed such a strategy for the directed evolution of PTE to yield higher arylesterase activity. By switching overexpression of GroEL/ES on or off depending on the starting variant's solubility, the screening environment was controlled to permit either the buffering of destabilizing mutations by chaperones or the acquisition of compensatory mutations (Fig. 3). To promote the gain of functional mutations, GroEL/ES overexpression was induced. Whenever the solubility of the variants became compromised, chaperone overexpression was stopped, which resulted in the PTE variants becoming aggregation-prone and expressed in inclusion bodies. Therefore, when screening for increased activity in this environment, the resulting selection pressure was aimed primarily at increasing the functional and soluble expression level in the cell. During 18 rounds of directed evolution, selection for compensatory mutations comprised rounds 3 and 9–11. Overall, the accumulation of 12 function-altering mutations and six compensatory mutations resulted in the arylesterase activity reaching a level comparable to wild-type PTE activity (> 104-fold increase), while maintaining soluble expression despite the accumulation of destabilizing, function-altering mutations.


Figure 3. The proposed combinatorial strategy. Soluble expression of the target protein is determined (1). If the protein has high soluble expression (red path), a mutagenized library is created (2) and screened for enhanced function with a buffering mechanism in place (3). Those variants with increased function are used as the templates for the next round (4). If the protein has low soluble expression (blue path), the mutagenized library (2) is then screened for increased soluble expression (3) and those variants with increased solubility are used as templates for the next round (4).

Download figure to PowerPoint

The crux of combining the two stability-modulating mechanisms is that the soluble expression level for the variants must be monitored. In this case, the activity ratio between the variants in the presence and absence of GroEL/ES overexpression served as an indicator for protein solubility [9] (K. T. Wyganowski, M. Kaltenbach and N. Tokuriki, unpublished data). It is crucial that the selection environment be switched according to the solubility of the starting gene (Fig. 3). One might be able to establish a switchable system by controlling expression temperature, induction protocol and/or the presence of a fusion partner, if any chaperone buffering is unavailable for a target protein.


  1. Top of page
  2. Abstract
  3. Introduction
  4. Protein functional expression and stability
  5. Promoting evolvability by modulating protein stability
  6. Methods to identify stabilizing/compensatory mutations
  7. Buffering strategies to modulate the stability threshold
  8. Combining compensatory mutations with buffering in directed evolution
  9. Perspectives
  10. Acknowledgements
  11. References

The modulation of protein stability during the engineering of new or optimized protein function can expedite the process and improve the overall outcome. Further development of such modulation strategies would pave the way for achieving more ambitious protein functions. However, for these methods to work effectively, it is important to identify whether the functional expression of a certain protein in vivo is correlated more to thermodynamic or kinetic stability. A reliable prediction tool for mutations that alter the functional and soluble expression level of proteins, particularly those that improve kinetic stability in the cellular milieu, would therefore be of great value. Moreover, versatile, effective and switchable buffering strategies will also be useful for overcoming the substrate limitations of GroEL/ES.


  1. Top of page
  2. Abstract
  3. Introduction
  4. Protein functional expression and stability
  5. Promoting evolvability by modulating protein stability
  6. Methods to identify stabilizing/compensatory mutations
  7. Buffering strategies to modulate the stability threshold
  8. Combining compensatory mutations with buffering in directed evolution
  9. Perspectives
  10. Acknowledgements
  11. References

We thank Miriam Kaltenbach for useful comments on the manuscript. This work was supported by the Natural Sciences and Engineering Research Council of Canada. N.T. is a Michael Smith Foundation for Health Research Scholar.


  1. Top of page
  2. Abstract
  3. Introduction
  4. Protein functional expression and stability
  5. Promoting evolvability by modulating protein stability
  6. Methods to identify stabilizing/compensatory mutations
  7. Buffering strategies to modulate the stability threshold
  8. Combining compensatory mutations with buffering in directed evolution
  9. Perspectives
  10. Acknowledgements
  11. References
  • 1
    Bornscheuer UT, Huisman GW, Kazlauskas RJ, Lutz S, Moore JC & Robins K (2012) Engineering the third wave of biocatalysis. Nature 485, 185194.
  • 2
    Turner NJ (2009) Directed evolution drives the next generation of biocatalysts. Nat Chem Biol 5, 567573.
  • 3
    Dalby PA (2011) Strategy and success for the directed evolution of enzymes. Curr Opin Struct Biol 21, 473480.
  • 4
    Kazlauskas RJ & Bornscheuer UT (2009) Finding better protein engineering strategies. Nat Chem Biol 5, 526529.
  • 5
    Bloom JD, Meyer MM, Meinhold P, Otey CR, MacMillan D & Arnold FH (2005) Evolving strategies for enzyme engineering. Curr Opin Struct Biol 15, 447452.
  • 6
    DePristo MA, Weinreich DM & Hartl DL (2005) Missense meanderings in sequence space: a biophysical view of protein evolution. Nat Rev Genet 6, 678687.
  • 7
    Bloom JD, Labthavikul ST, Otey CR & Arnold FH (2006) Protein stability promotes evolvability. Proc Natl Acad Sci USA 103, 58695874.
  • 8
    Zeldovich KB, Chen P & Shakhnovich EI (2007) Protein stability imposes limits on organism complexity and speed of molecular evolution. Proc Natl Acad Sci USA 104, 1615216157.
  • 9
    Tokuriki N & Tawfik DS (2009) Chaperonin overexpression promotes genetic variation and enzyme evolution. Nature 459, 668673.
  • 10
    Tokuriki N & Tawfik DS (2009) Stability effects of mutations and protein evolvability. Curr Opin Struct Biol 19, 596604.
  • 11
    Yue P, Li Z & Moult J (2005) Loss of protein structure stability as a major causative factor in monogenic disease. J Mol Biol 353, 459473.
  • 12
    Yue P & Moult J (2006) Identification and analysis of deleterious human SNPs. J Mol Biol 356, 12631274.
  • 13
    Bloom JD & Arnold FH (2009) In the light of directed evolution: pathways of adaptive protein evolution. Proc Natl Acad Sci USA 106, 999510000.
  • 14
    Bershtein S, Segal M, Bekerman R, Tokuriki N & Tawfik DS (2006) Robustness-epistasis link shapes the fitness landscape of a randomly drifting protein. Nature 444, 929932.
  • 15
    Tokuriki N, Stricher F, Schymkowitz J, Serrano L & Tawfik DS (2007) The stability effects of protein mutations appear to be universally distributed. J Mol Biol 369, 13181332.
  • 16
    Tokuriki N, Stricher F, Serrano L & Tawfik DS (2008) How protein stability and new functions trade off. PLoS Comput Biol 4, e1000002.
  • 17
    Woycechowsky KJ, Choutko A, Vamvaca K & Hilvert D (2008) Relative tolerance of an enzymatic molten globule and its thermostable counterpart to point mutation. Biochemistry 47, 1248913496.
  • 18
    Mayer S, Rüdiger S, Ang HC, Joerger AC & Fersht AR (2007) Correlation of levels of folded recombinant p53 in Escherichia coli with thermodynamic stability in vitro. J Mol Biol 372, 268276.
  • 19
    Randles LG, Lappalainen I, Fowler SB, Moore B, Hamill SJ & Clarke J (2006) Using model proteins to quantify the effects of pathogenic mutations in Ig-like proteins. J Biol Chem 281, 2421624226.
  • 20
    Sanchez-Ruiz JM (2010) Protein kinetic stability. Biophys Chem 148, 115.
  • 21
    Hartl FU & Hayer-Hartl M (2009) Converging concepts of protein folding in vitro and in vivo. Nat Struct Mol Biol 16, 574581.
  • 22
    Dobson CM (2003) Protein folding and misfolding. Nature 426, 884890.
  • 23
    Bershtein S, Mu W, Wu W & Shakhnovich EI (2012) Soluble oligomerization provides a beneficial fitness effect on destabilizing mutations. Proc Natl Acad Sci USA 109, 48574862.
  • 24
    Bershtein S, Mu W, Serohijos AWR, Zhou J & Shakhnovich EI (2013) Protein quality control acts on folding intermediates to shape the effects of mutations on organismal fitness. Mol Cell 49, 133144.
  • 25
    Diaz JE, Lin C-S, Kunishiro K, Feld BK, Avrantinis SK, Bronson J, Greaves J, Saven JG & Weiss GA (2011) Computational design and selections for an engineered, thermostable terpene synthase. Protein Sci 20, 15971606.
  • 26
    Rockah-Shmuel L & Tawfik DS (2012) Evolutionary transitions to new DNA methyltransferases through target site expansion and shrinkage. Nucleic Acids Res 40, 1162711637.
  • 27
    Khersonsky O, Rosenblat M, Toker L, Yacobson S, Hugenmatter A, Silman I, Sussman JL, Aviram M & Tawfik DS (2009) Directed evolution of serum paraoxonase PON3 by family shuffling and ancestor/consensus mutagenesis, and its biochemical characterization. Biochemistry 48, 66446654.
  • 28
    Khersonsky O, Kiss G, Röthlisberger D, Dym O, Albeck S, Houk KN, Baker D & Tawfik DS (2012) Bridging the gaps in design methodologies by evolutionary optimization of the stability and proficiency of designed Kemp eliminase KE59. Proc Natl Acad Sci USA 109, 1035810363.
  • 29
    Amar D, Berger I, Amara N, Tafa G, Meijler MM & Aharoni A (2012) The transition of human estrogen sulfotransferase from generalist to specialist using directed enzyme evolution. J Mol Biol 416, 2132.
  • 30
    Bershtein S, Goldin K & Tawfik DS (2008) Intense neutral drifts yield robust and evolvable consensus proteins. J Mol Biol 379, 10291044.
  • 31
    Fasan R, Meharenna YT, Snow CD, Poulos TL & Arnold FH (2008) Evolutionary history of a specialized p450 propane monooxygenase. J Mol Biol 383, 10691080.
  • 32
    Cole MF & Gaucher EA (2011) Utilizing natural diversity to evolve protein function: applications toward thermostability. Curr Opin Chem Biol 15, 399406.
  • 33
    Magliery TJ, Lavinder JJ & Sullivan BJ (2011) Protein stability by number: high-throughput and statistical approaches to one of protein science's most difficult problems. Curr Opin Chem Biol 15, 443451.
  • 34
    Steipe B, Schiller B, Plückthun A & Steinbacher S (1994) Sequence statistics reliably predict stabilizing mutations in a protein domain. J Mol Biol 240, 188192.
  • 35
    Miyazaki J, Nakaya S, Suzuki T, Tamakoshi M, Oshima T & Yamagishi A (2001) Ancestral residues stabilizing 3-isopropylmalate dehydrogenase of an extreme thermophile: experimental evidence supporting the thermophilic common ancestor hypothesis. J Biochem 129, 777782.
  • 36
    Godoy-Ruiz R, Ariza F, Rodriguez-Larrea D, Perez-Jimenez R, Ibarra-Molero B & Sanchez-Ruiz JM (2006) Natural selection for kinetic stability is a likely origin of correlations between mutational effects on protein energetics and frequencies of amino acid occurrences in sequence alignments. J Mol Biol 362, 966978.
  • 37
    Pey AL, Rodriguez-Larrea D, Bomke S, Dammers S, Godoy-Ruiz R, Garcia-Mira MM & Sanchez-Ruiz JM (2008) Engineering proteins with tunable thermodynamic and kinetic stabilities. Proteins 71, 165174.
  • 38
    Wang Q, Buckle AM, Foster NW, Johnson CM & Fersht AR (1999) Design of highly stable functional GroEL minichaperones. Protein Sci 8, 21862193.
  • 39
    Rath A & Davidson AR (2000) The design of a hyperstable mutant of the Abp1p SH3 domain by sequence alignment analysis. Protein Sci 9, 24572469.
  • 40
    Lehmann M, Pasamontes L, Lassen SF & Wyss M (2000) The consensus concept for thermostability engineering of proteins. Biochim Biophys Acta 1543, 408415.
  • 41
    Lehmann M, Loch C, Middendorf A, Studer D, Lassen SF, Pasamontes L, van Loon AP & Wyss M (2002) The consensus concept for thermostability engineering of proteins: further proof of concept. Protein Eng 15, 403411.
  • 42
    Amin N, Liu AD, Ramer S, Aehle W, Meijer D, Metin M, Wong S, Gualfetti P & Schellenberger V (2004) Construction of stabilized proteins by combinatorial consensus mutagenesis. Protein Eng Des Sel 17, 787793.
  • 43
    Polizzi KM, Chaparro-Riggers JF, Vázquez-Figueroa E & Bommarius AS (2006) Structure-guided consensus approach to create a more thermostable penicillin G acylase. Biotechnol J 1, 531536.
  • 44
    Vázquez-Figueroa E, Chaparro-Riggers J & Bommarius AS (2007) Development of a thermostable glucose dehydrogenase by a structure-guided consensus concept. ChemBioChem 8, 22952301.
  • 45
    Anbar M, Gul O, Lamed R, Sezerman UO & Bayer EA (2012) Improved thermostability of Clostridium thermocellum endoglucanase Cel8A by using consensus-guided mutagenesis. Appl Environ Microbiol 78, 34583464.
  • 46
    Ito Y, Ikeuchi A & Imamura C (2013) Advanced evolutionary molecular engineering to produce thermostable cellulase by using a small but efficient library. Protein Eng Sel Des 26, 7379.
  • 47
    Xiao Z, Bergeron H, Grosse S, Beauchemin M, Garron M-L, Shaya D, Sulea T, Cygler M & Lau PCK (2008) Improvement of the thermostability and activity of a pectate lyase by single amino acid substitutions, using a strategy based on melting-temperature-guided sequence alignment. Appl Environ Microbiol 74, 11831189.
  • 48
    Iwabata H, Watanabe K, Ohkuri T, Yokobori S-I & Yamagishi A (2005) Thermostability of ancestral mutants of Caldococcus noboribetusisocitrate dehydrogenase. FEMS Microbiol Lett 243, 393398.
  • 49
    Watanabe K, Ohkuri T, Yokobori S-I & Yamagishi A (2006) Designing thermostable proteins: ancestral mutants of 3-isopropylmalate dehydrogenase designed by using a phylogenetic tree. J Mol Biol 355, 664674.
  • 50
    Shimizu H, Yokobori S-I, Ohkuri T, Yokogawa T, Nishikawa K & Yamagishi A (2007) Extremely thermophilic translation system in the common ancestor commonote: ancestral mutants of Glycl-tRNA synthetase from the extreme thermophile Thermus thermophiles. J Mol Biol 369, 10601069.
  • 51
    Yamashiro K, Yokobori S-I, Koikeda S & Yamagishi A (2010) Improvement of Bacillus circulans beta-amylase activity attained using the ancestral mutation method. Protein Eng Des Sel 23, 519528.
  • 52
    Guerois R, Nielsen JE & Serrano L (2002) Predicting changes in the stability of proteins and protein complexes: a study of more than 1000 mutations. J Mol Biol 320, 369387.
  • 53
    Pokala N & Handel TM (2005) Energy functions for protein design: adjustment with protein-protein complex affinities, models for the unfolded state, and negative design of solubility and specificity. J Mol Biol 347, 203227.
  • 54
    Capriotti E, Fariselli P & Casadio R (2005) I-Mutant2.0: predicting stability changes upon mutation from the protein sequence or structure. Nucleic Acids Res 33, W306W310.
  • 55
    Parthiban V, Gromiha MM & Schomburg D (2006) CUPSAT: prediction of protein stability upon point mutations. Nucleic Acids Res 34, W239W242.
  • 56
    Lonquety M, Lacroix Z, Papandreou N & Chomilier J (2009) SPROUTS: a database for the evaluation of protein stability upon point mutation. Nucleic Acids Res 37, D374D379.
  • 57
    Benedix A, Becker CM, de Groot BL, Caflisch A & Böckmann RA (2009) Predicting free energy changes using structural ensembles. Nat Methods 6, 34.
  • 58
    Dehouck Y, Kwasigroch JM, Gilis D & Rooman M (2011) PoPMuSiC 2.1: a web server for the estimation of protein stability changes upon mutation and sequence optimality. BMC Bioinformatics 12, 151.
  • 59
    Lauck F, Smith CA, Friedland GF, Humphris EL & Kortemme T (2010) RosettaBackrub – a web server for flexible backbone protein structure modeling and design. Nucleic Acids Res 38, W569W575.
  • 60
    Worth CL, Preissner R & Blundell TL (2011) SDM – a server for predicting effects of mutations on protein stability and malfunction. Nucleic Acids Res 39, W215W222.
  • 61
    Li Y & Fang J (2012) PROTS-RF: a robust model for predicting mutation-induced protein stability changes. PLoS One 7, e47247.
  • 62
    Wickstrom L, Gallicchio E & Levy RM (2012) The linear interaction energy method for the prediction of protein stability changes upon mutation. Proteins 80, 111125.
  • 63
    De Baets G, Van Durme J, Reumers J, Maurer-Stroh S, Vanhee P, Dopazo J, Schymkowitz J & Rousseau F (2012) SNPeffect 4.0: on-line prediction of molecular and structural effects of protein-coding variants. Nucleic Acids Res 40, D935D939.
  • 64
    Khan S & Vihinen M (2010) Performance of protein stability predictors. Hum Mutat 31, 675684.
  • 65
    Thiltgen G & Goldstein RA (2012) Assessing predictors of changes in protein stability upon mutation using self-consistency. PLoS One 7, e46084.
  • 66
    Potapov V, Cohen M & Schreiber G (2009) Assessing computational methods for predicting protein stability upon mutation: good on average but not in the details. Protein Eng Des Sel 22, 553560.
  • 67
    Cabrita LD, Gilis D, Robertson AL, Dehouck Y, Rooman M & Bottomley SP (2007) Enhancing the stability and solubility of TEV protease using in silico design. Protein Sci 16, 23602367.
  • 68
    Reetz MT, Carballeira JD & Vogel A (2006) Iterative saturation mutagenesis on the basis of B factors as a strategy for increasing protein thermostability. Angew Chem Int Ed Engl 45, 77457751.
  • 69
    Wen S, Tan T & Zhao H (2012) Improving the thermostability of lipase Lip2 from Yarrowia lipolytica. J Biotechnol 164, 248253.
  • 70
    Blum JK, Ricketts MD & Bommarius AS (2012) Improved thermostability of AEH by combining B-FIT analysis and structure-guided consensus method. J Biotechnol 160, 214221.
  • 71
    Komor RS, Romero PA, Xie CB & Arnold FH (2012) Highly thermostable fungal cellobiohydrolase I (Cel7A) engineered using predictive methods. Protein Eng Des Sel 25, 827833.
  • 72
    Salazar O, Cirino PC & Arnold FH (2003) Thermostabilization of a cytochrome p450 peroxygenase. ChemBioChem 4, 891893.
  • 73
    Martinez R, Jakob F, Tu R, Siegert P, Maurer K-H & Schwaneberg U (2013) Increasing activity and thermal resistance of Bacillus gibsonii alkaline protease (BgAP) by directed evolution. Biotechnol Bioeng 110, 711720.
  • 74
    Shivange AV, Serwe A, Dennig A, Roccatano D, Haefner S & Schwaneberg U (2012) Directed evolution of a highly active Yersinia mollaretii phytase. Appl Microbiol Biotechnol 95, 405418.
  • 75
    Buettner K, Hertel TC & Pietzsch M (2012) Increased thermostability of microbial transglutaminase by combination of several hot spots evolved by random and saturation mutagenesis. Amino Acids 42, 987996.
  • 76
    García-Ruiz E, Maté D, Ballesteros A, Martinez AT & Alcalde M (2010) Evolving thermostability in mutant libraries of ligninolytic oxidoreductases expressed in yeast. Microb Cell Fact 9, 17.
  • 77
    Giver L, Gershenson A, Freskgard PO & Arnold FH (1998) Directed evolution of a thermostable esterase. Proc Natl Acad Sci USA 95, 1280912813.
  • 78
    Kristensen P & Winter G (1998) Proteolytic selection for protein folding using filamentous bacteriophages. Fold Des 3, 321328.
  • 79
    Sieber V, Plückthun A & Schmid FX (1998) Selecting proteins with improved stability by a phage-based method. Nat Biotechol 16, 955960.
  • 80
    Martin A, Schmid FX & Sieber V (2003) Proside: a phage-based method for selecting thermostable proteins. Methods Mol Biol 230, 5770.
  • 81
    Foit L, Mueller-Schickert A, Mamathambika BS, Gleiter S, Klaska CL, Ren G & Bardwell JCA (2011) Genetic selection for enhanced folding in vivo targets the Cys14-Cys38 disulfide bond in bovine pancreatic trypsin inhibitor. Antioxid Redox Signal 14, 973984.
  • 82
    Wunderlich M, Martin A & Schmid FX (2005) Stabilization of the cold shock protein CspB from Bacillus subtilis by evolutionary optimization of Coulombic interactions. J Mol Biol 347, 10631076.
  • 83
    Wunderlich M & Schmid FX (2006) In vitro evolution of a hyperstable Gbeta1 variant. J Mol Biol 363, 545557.
  • 84
    Kather I, Jakob RP, Dobbek H & Schmid FX (2008) Increased folding stability of TEM-1 beta-lactamase by in vitro selection. J Mol Biol 383, 238251.
  • 85
    Roodveldt C, Aharoni A & Tawfik DS (2005) Directed evolution of proteins for heterologous expression and stability. Curr Opin Struct Biol 15, 5056.
  • 86
    Khersonsky O, Röthlisberger D, Dym O, Albeck S, Jackson CJ, Baker D & Tawfik DS (2010) Evolutionary optimization of computationally designed enzymes: Kemp eliminases of the KE07 series. J Mol Biol 396, 10251042.
  • 87
    Lamla T, Hoerer S & Bauer MMT (2006) Screening for soluble expression constructs using cell-free protein synthesis. Int J Biol Macromol 39, 111121.
  • 88
    Xie X, Pashkov I, Gao X, Guerrero JL, Yeates TO & Tang Y (2009) Rational improvement of simvastatin synthase solubility in Escherichia coli leads to higher whole-cell biocatalytic activity. Biotechnol Bioeng 102, 2028.
  • 89
    Waldo GS, Standish BM, Berendzen J & Terwilliger TC (1999) Rapid protein-folding assay using green fluorescent protein. Nat Biotechol 17, 691695.
  • 90
    Wigley WC, Stidham RD, Smith NM, Hunt JF & Thomas PJ (2001) Protein solubility and folding monitored in vivo by structural complementation of a genetic marker protein. Nat Biotechnol 19, 131136.
  • 91
    Dyson MR, Perera RL, Shadbolt SP, Biderman L, Bromek K, Murzina NV & McCafferty J (2008) Identification of soluble protein fragments by gene fragmentation and genetic selection. Nucleic Acids Res 36, e51.
  • 92
    Maxwell KL, Mittermaier AK, Forman-Kay JD & Davidson AR (1999) A simple in vivo assay for increased protein solubility. Protein Sci 8, 19081911.
  • 93
    Foit L, Morgan GJ, Kern MJ, Steimer LR, von Hacht AA, Titchmarsh J, Warriner SL, Radford SE & Bardwell JCA (2009) Optimizing protein stability in vivo. Mol Cell 36, 861871.
  • 94
    Cabantous S, Terwilliger TC & Waldo GS (2005) Protein tagging and detection with engineered self-assembling fragments of green fluorescent protein. Nat Biotechnol 23, 102107.
  • 95
    Cabantous S, Rogers Y, Terwilliger TC & Waldo GS (2008) New molecular reporters for rapid protein folding assays. PLoS One 3, e2387.
  • 96
    Lindman S, Hernandez-Garcia A, Szczepankiewicz O, Frohm B & Linse S (2010) In vivo protein stabilization based on fragment complementation and a split GFP system. Proc Natl Acad Sci USA 107, 1982619831.
  • 97
    Seitz T, Thoma R, Schoch GA, Stihle M, Benz J, D'Arcy B, Wiget A, Ruf A, Hennig M & Sterner R (2010) Enhancing the stability and solubility of the glucocorticoid receptor ligand-binding domain by high-throughput library screening. J Mol Biol 403, 562577.
  • 98
    Jiang S, Li C, Zhang W, Cai Y, Yang Y, Yang S & Jiang W (2007) Directed evolution and structural analysis of N-carbamoyl-D-amino acid amidohydrolase provide insights into recombinant protein solubility in Escherichia coli. Biochem J 402, 429437.
  • 99
    Kerner MJ, Naylor DJ, Ishihama Y, Maier T, Chang H-C, Stines AP, Georgopoulos C, Frishman D, Hayer-Hartl M, Mann M et al. (2005) Proteome-wide analysis of chaperonin-dependent protein folding in Escherichia coli. Cell 122, 209220.
  • 100
    Fujiwara K, Ishihama Y, Nakahigashi K, Soga T & Taguchi H (2010) A systematic survey of in vivo obligate chaperonin-dependent substrates. EMBO J 29, 15521564.
  • 101
    Allali-Hassani A, Wasney GA, Chau I, Hong BS, Senisterra G, Loppnau P, Shi Z, Moult J, Edwards AM, Arrowsmith CH et al. (2009) A survey of proteins encoded by non-synonymous single nucleotide polymorphisms reveals a significant fraction with altered stability and activity. Biochem J 424, 1526.
  • 102
    Mogk A & Bukau B (2004) Molecular chaperones: structures of a protein disaggregase. Curr Biol 14, R78R80.
  • 103
    Sharma SK, Christen P & Goloubinoff P (2009) Disaggregating chaperones: an unfolding story. Curr Protein Pept Sci 10, 432446.
  • 104
    Bandyopadhyay A, Saxena K, Kasturia N, Dalal V, Bhatt N, Rajkumar A, Maity S, Sengupta S & Chakraborty K (2012) Chemical chaperones assist intracellular folding to buffer mutational variations. Nat Chem Biol 8, 238245.
  • 105
    Rajan RS, Tsumoto K, Tokunaga M, Tokunaga H, Kita Y & Arakawa T (2011) Chemical and pharmacological chaperones: application for recombinant protein production and protein folding diseases. Curr Med Chem 18, 115.
  • 106
    Niwa T, Kanamori T, Ueda T & Taguchi H (2012) Global analysis of chaperone effects using a reconstituted cell-free translation system. Proc Natl Acad Sci USA 109, 89378942.
  • 107
    Sun P, Tropea JE & Waugh DS (2011) Enhancing the solubility of recombinant proteins in Escherichia coli by using hexahistidine-tagged maltose-binding protein as a fusion partner. Methods Mol Biol 705, 259274.