大家好，今天推送的文章是2024年4月发表在JAFC上的“Computation-Guided Rational Design of Cysteine-Less Protein Variants in Engineered hCGL”，通讯作者为中国医学科学院北京协和医学院药物生物技术研究所的金媛媛副研究员和杨兆勇研究员。
Cysteine is highly reactive and uniquely nucleophilic due to its low S-H bond dissociation energy and significant polarizability of sulfur atoms. Intramolecular disulfide bonds give proteins structural stability, but the additional intermolecular disulfide bonds resulting from the very close proximity of cysteine residues between molecules may lead to protein aggregation and subsequent precipitation. Studies have shown that cysteine mutations to remove disulfide bonds can greatly improve protein solubility in vitro and also prevent protein aggregation due to random disulfide bond bridging. In this paper, the authors' team used an unconventional approach to enhance the pharmaceutical potential of engineered hCGLs by systematically redesigning all cysteine residues to obtain mutant variants that exhibit high activity or enhanced stability, while identifying the critical role of cysteine residues in engineered hCGLs and analyzing potential modification sites.
1
Reconstituted identification of cysteine residues in engineered hCGL
The crystal structure of hCGL and the engineered hCGL E59T-E339V complexes with L-Cys has been resolved. The authors' team mutalogized all cysteine residues in engineered hCGL to explore their effects on activity and stability. Engineered hCGL proteins contain 10 cysteine residues, C70, C84, C109, C137, C172, C229, C252, C255, C307, and C310, within each monomeric unit (Figure 1). C70, C172, C252, and C255 are in α helix. C84 and C109 are located within the β fold and are deeply embedded inside the protein. The cysteine residues at positions 137, 229, 307, and 310 are located within the ring region and on the surface of the protein. These 10 unbound cysteine moieties do not establish disulfide bonds within a single monomeric structure or in different subunits (Figure 1).
The authors' team used the Cartesian_ddG module of Rosetta and ABACUS to perform virtual saturation mutagenesis of all cysteine residues within the engineered hCGL, with the aim of assessing the effects of mutations in terms of folding energy and statistical energy. The results showed that there were a total of 8 potentially beneficial mutants (ΔΔG less than -1.5 or B less than -3.0): C84A, C109V, C137V, C172V, C229D, C255L, C307F, and C310V. However, this rational design approach ignores the possible modifications of hCGL cysteine residues, such as S-sulfide, S-nitrosation, and polysulfidation.
Figure 1
2
Characterization and enzymatic properties of mutant activity, thermostability, and stability
The authors' team purified wild-type and mutants using affinity chromatography, and the substitution of free cysteine at a single site did not affect protein expression levels compared to wild-type. The specific activities of WT and mutants were determined using l-Cys and CSSC, respectively, under the 37°C reaction (Figure 2). Mutants C109V, C137V, C172V, C229D, and C255L have increased activity against both substrates compared to WT. The activity of mutant C109V against l-Cys and CSSC was 1.9 and 1.8 times higher than that of WT, respectively, and the activity of the other four mutants C137V, C172V, C229D and C255L increased by 77%, 64%, 18% and 27%, respectively, and the activity of CSSC as substrate increased by 47%, 15%, 55% and 51%, respectively. Mutants C84A and C310V showed varying degrees of reduced activity, while mutant C307F decreased by 28% with L-Cys as substrate and increased by 20% with CSSC. When cysteine, located at position 84 on the surface of a protein, mutates to hydrophobic alanine, it causes protein instability, resulting in reduced stability of the C84A mutant. When l-Cys is used as substrate, the activity of C109S and C307F is lower than that of WT. In addition to this, the supplementary material provides evidence that cysteine substitution in mutants alters the oligomeric form of proteins.
Figure 2
As shown in Figure 3A, the authors' team evaluated the optimal reaction temperature of the mutant enzyme, and the mutants C109V, C137V, and C229D exhibited optimal reaction temperatures consistent with WT, while the optimal reaction temperature of the mutants C84A, C172V, C255L, C307F, and C310V was significantly reduced by 5 °C compared to WT. However, some mutants are characterized by a lower optimal reaction temperature and exhibit higher residual activity at 55 °C. Notably, the mutant C229D exhibited the highest residual activity at 80 °C among all the mutants.
To study the thermal stability of the enzyme, the authors' team measured the Tm values of the wild-type and its mutants. As shown in Table 2, the two mutants exhibited enhanced Tm values, with an increase of approximately 2 °C in mutant C229D and an increase of more than 8 °C in mutant C109V compared to WT. In addition, the catalytic activity of mutants C109V and C229D showed varying degrees of improvement under both substrates, and all mutants were incubated at 60 °C for different periods of time, and their specific activity was subsequently determined (Fig. 3CD), and after 80 minutes of incubation at 60°C, mutant C229D retained 90% of the initial activity, and its specific activity was 1.63 and 1.41 times higher than that of WT with l-Cys and CSSC as substrates, respectively. Most mutants exhibit a similar trend of reduced activity to WT, with the exception of mutant C84A, which exhibits a significant loss of activity after 60 minutes of incubation.
Table 2
Figure 3
The authors' team determined the optimal response pH of the mutants on a pH range of 5.6–11.0. Figure 3B shows that WT, C84A, C137V, and C307V exhibit the same optimal reaction pH at pH 7.5 when CSSC is used as substrate, while C255L exhibits its peak activity at a slightly higher pH of 8.2. Notably, C109V, C172V, C229D, and C310V exhibit high activity over a wide range of pH 7.5–9.0. When L-Cys is used as a substrate, the optimal reaction pH range for mutants is pH 8.2-9.0.
In order to further study the properties of these mutants, the authors' team conducted kinetic experiments using l-Cys and CSSC as substrates, and the results showed that WT and its mutants were more biased towards substrate CSSC rather than substrate l-Cys in terms of substrate selectivity. Compared with the substrate L-Cys, both WT and its mutants exhibited higher kcat values in catalyzing CSSC. When L-Cys was used as substrate, the KM values of mutants C84A, C109V, C137V and C172V decreased compared with WT, while the KM values of mutants C307F did not change much, while the mutants C229D, C255L and C310V increased to varying degrees. The kcat of mutants C109V, C137V, C172V, C229D, and C255L increased, with C109V increasing nearly 2-fold. The catalytic efficiency (kcat/KM) calculation showed that the kinetic efficiency of the mutants C109V and C137V was increased by more than 2 times, and the kinetic efficiency of the mutant C172V was increased by about 1.7 times. When CSSC is used as a substrate, all mutants except mutant C172V exhibit increased KM values. The kcat of the mutant C109V showed an increase of about 1.8-fold, and the kcat of the mutants C137V, C172V, C229D, and C255L showed an increase of about 1.3-fold. The kcat/KM efficiency of the mutants C109V and C137V was about the same as that of WT, while the kinetic efficiency of the mutant C172V was increased by 1.3 times.
3
Interaction analysis based on protein structure
As can be seen from the above results, the mutants C109V and C229D exhibit amazing performance. The mutant C109V with both substrates had an approximately 2-fold increase in specific activity, while Tm increased by approximately 8°C. Although its catalytic efficiency for CSSC substrates is comparable to that of WT, the catalytic efficiency for L-Cys substrates is increased by more than 2-fold. Compared with WT, the Tm value of the C229D mutant was increased by nearly 2 °C, and its catalytic activity against L-Cys and CSSC substrates was increased by 18% and 55%, respectively.
In order to elucidate the reasons for the significant improvement in heat tolerance of the two mutants, the authors used AlphaFold2 to construct the structures of the above mutants (Figure 4). In the C109V mutant, the residue at position 109 is located inside the protein and in a hydrophobic environment. The cysteine mutation to hydrophobic valine significantly enhanced the hydrophobic interaction with surrounding amino acids (V113, F121, I132, F134, and W155) (Figure 4AB). To verify that enhancing the hydrophobicity of this region is beneficial for increased activity, the authors' team performed a saturation mutation at position 109, as shown in Figure 5, and the substitution of the hydrophobic amino acids leucine, isoleucine, and methionine also significantly increased activity. Interestingly, when the mutation at position 109 was mutated to aspartic acid, glutamic acid, histidine, or lysine, the activity of the mutant decreased dramatically. In the case of the C229D mutant, replacing cysteine at position 229 with aspartic acid not only retains the pre-existing hydrogen bonds, but also introduces two additional hydrogen bonds between D229 and E230 and S231 (Figure 4CD). As a result of the production of two supplemental hydrogen bonds, the Tm of the mutant C229D increased by 1.8 °C relative to WT. This mutant exhibits excellent thermal stability at 60 °C compared to all other mutants.
Figure 4
Figure 5
4
Computational analysis of the dynamic behavior of WT and mutants
In order to elucidate the structural basis of the different activities and properties of WT and its mutants C109V, C137V, and C172V, the authors' team performed several 100 ns molecular dynamics (MD) simulations at 310 K (Figure 6A). Both hCGL and MGL are members of PLP-dependent γ lyase and catalyze similar reactions. Based on the catalytic mechanism of MGL, the distance between O and N of Tyr114 (OTyr) and N of L-Cys (NCys) and C and NCys of PLP was determined. Reflects the stability of the near-attack conformation (NAC) and can be used to assess catalytic activity. Figure 6B illustrates the two distances between WT, C109V, C137V, and C172V observed during the MD simulation. The distance scattering distribution of C109V, C137V, and C172V mutants was more compact than that of WT, and the distance scattering of C109V and C137V mutants was more concentrated in the lower left corner than WT, indicating that the OTyr-NCys and CPLP-NCys in C109V and C137V mutants were less distanced, so its NAC was more favorable than WT, and it was speculated that the improvement of NAC stability was the root cause of the enhanced activity of mutants C109V, C137V, and C172V.
Figure 6
Engineered human cystathion-γ-lyase (hCGL) can enhance activity against cysteine and cystine, revealing potentially potent antitumor activity. However, the presence of cysteine residues has the potential to induce oligomeric or incorrect disulfide bonds, which may reduce the bioavailability of the drug. The authors' team obtained a panel of potentially beneficial mutants through a well-designed virtual screen targeting cysteine residues in hCGL using Rosetta and ABACUS. Experimental measurements showed increased activity against substrates L-Cys and CSSC in most mutants. In addition, the Tm values of the mutants C109V and C229D increased by 8.2 and 1.8 °C, respectively. After 80 minutes of incubation at 60°C, the mutant C229D still maintained high residual activity. Surprisingly, the mutant C109V showed approximately 2-fold higher activity against both substrates than wild-type (WT), but exhibited disappointing instability in plasma, suggesting that further consideration is still needed for computational design. Analysis of their structural and molecular dynamics (MD) simulations revealed the effects of hydrophobic interactions, hydrogen bonding, and near-attack conformational (NAC) stability on activity and stability.
Article Information:
https://doi.org/10.1021/acs.jafc.3c06821