Factors Improving Thermostability: Hydrophobicity and Optimization of Charged Interactions
The addition of hydrophobic residues has shown to correlate with an increase in protein thermal stability [6]. This is achieved through the effects of hydrophobicity on the free energy of unfolding, a metric commonly used to determine protein stability. Through increasing the difference in energy between folded and unfolded states, the folded state in equilibrium will be favored due to the large energy of activation for the reverse reaction (unfolding). Additionally, hydrophobic bonds have demonstrated to increase in strength as temperature increases, possibly lending this type of bond as the dominant driver of protein stability at higher temperatures [6]. Studies on a group of Superoxide Dismutases (SODs) have demonstrated an increase in hydrophobicity and molecular weight of thermophilic variants relative to non-thermophilic proteins [7]. In addition to the SOD family of proteins, this trend has been demonstrated in a study by Spassov et al., where 183 proteins of documented tertiary structure obtained from the Protein Data Bank were analyzed based on protein-solvent interactions. As these proteins are non-homologous to one another, this proves the application of this property can be applied to a broader range of proteins [8]. In addition to hydrophobicity, interactions between charged residues can directly impact the thermal stability of proteins. The presence of close, charged amino acids in the 3D protein conformation can facilitate the formation of an ionic salt bridge, improving the stability of the region. Electrostatic interactions such as this may compensate for a lack of sufficient hydrophobic interactions, as demonstrated with the aforementioned group of 183 nonhomologous proteins [8]. On a molecular level, the improved intrinsic thermal stability is not based on post-translational modifications, but rather optimal redistribution of basic amino acids relative to what is found in mesophilic (moderate temperature preferring) proteins [9]. One frequent change found in thermophiles is the increase of buried hydrophobic residues, often characterized by an increase in leucine and proline and a decrease in glycine and methionine. Furthermore, the layer of thermophilic proteins most exposed to protein solvent interaction often demonstrated an increase in the amount of charged amino acids [10].
Comparison With Thermostable Variants
While PETase may lack in thermal stability, there exist multiple closely related enzymes showcasing improvements in this property, many of which are from the genus Thermobifida [4]. Thermobifida is often found in environments high in heated, natural organic polymers. It specializes in the production of thermostable enzymes to degrade said polymers, such as lignocellulose [11]. Many Thermobifida proteins, often cutinases, demonstrate off-target thermostable PET hydrolase ability [4]. Similarly, I. Sakeniesis was found in a waste facility specific to PET [3]. As previously mentioned, its PETase lacks the thermal stability of other related hydrolases, despite homology to Thermobifida [13, Figure 1]. This is commonly seen when comparing thermostable variants of enzymes as a stability-activity trade-off. When proteins adopt rigid conformations that favor thermostability, they lose some ability to undergo structural changes to properly fit the substrate, decreasing efficiency [12]. This property was demonstrated in a structurally well-documented enzyme, AmpC beta-lactamase, where many substitutions to the catalytic site correlated to an uptick in protein stability [14].
Following these findings, multiple variants were generated using a rational design process and diverse machine learning algorithms, whose methodology can be found through the following links:
Rational Design
Transfer Learning
Figure 1: Enzymatic activity comparison of iGEM Toronto variants (Tianyu BBa_K210002, Dimi 1 BBa_K2910003, Dimi 2 BBa_K2910004, Dimi3 BBa_K2910005, and Dimi4 BBa_K2910006) to Addgene mutant sequence W159 H S238F (BBa_K2910002) and wild-type IsPETase (WT). Assay followed p-nitrophenyl butyrate protocol to determine catalytic activity based on absorbance rates. All relations demonstrate a linear behavior as the assay was only run for 10 minutes. The mass spectrometer cannot read after 15 minutes of running assay. Therefore, the hyperbolic behavior that the degradation possesses is out of the interval shown in the graph.