Residue-level determinants of the thermal stability of the extremophilic Ts2631 endolysin
Karolina Cieminska, Anna-Karina Kaczorowska, Lukasz Pawel Kozlowski, Marcin Gorniak, Olafur H. Fridjonsson, Gudmundur O. Hreggvidsson, Tadeusz Kaczorowski, Magdalena Plotka
TL;DR
This paper identifies specific amino acid residues that contribute to the thermal stability of a heat-resistant endolysin from a bacteriophage, offering insights for engineering more stable antimicrobial enzymes.
Contribution
The study identifies residue-level determinants of thermal stability in the endolysin Ts2631, revealing the role of aromatic and proline residues in thermostability.
Findings
Ts2631 endolysin has a melting point of up to 104.7°C, indicating high thermostability.
Residues R20, W102, W109, P140, and W145 are critical for thermal stability and peptidoglycan binding.
Buried aromatic residues significantly contribute to the thermophilic fold of Ts2631.
Abstract
In the face of the growing crisis of antibiotic resistance, peptidoglycan-degrading endolysins derived from bacteriophages offer a promising alternative to traditional antimicrobials. Thermostable variants, capable of maintaining their structure and activity under extreme conditions, are of particular interest. Here, we present a comprehensive analysis of endolysin Ts2631 from bacteriophage vB_Tsc2631, which infects Thermus scotoductus in Icelandic hot springs. This type 2 amidase exhibits exceptional thermostability, with a melting point ranging from 99.8 °C to 104.7 °C, depending on the solvent. Structural comparison with its mesophilic counterpart, T7 lysozyme, revealed shorter loops with lower B factors, suggesting reduced conformational flexibility. The Ts2631 sequence contains increased levels of tyrosine, proline, tryptophan, and arginine residues, amino acids commonly…
Genes, proteins, chemicals, diseases, species, mutations and cell lines named across the full text — each resolved to its canonical identifier and authoritative record.
Click any figure to enlarge with its caption.
Figure 1
Figure 2
Figure 3
Figure 4
Figure 5
Figure 6
Figure 7
Figure 8- —University of Gdansk Young Scientists grant
- —National Science Center (Kraków, Poland) OPUS 20 grant
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsEnzyme Production and Characterization · Bacterial Genetics and Biotechnology · Lipid Membrane Structure and Behavior
Introduction
The survival of bacteria under extreme conditions has fascinated biologists for a long time^1^. Thermophilic microorganisms optimally grow at temperatures between 60 °C and 80 °C^2^, but the archaeal isolate Pyrolobus fumarii doubles in ~ 8 h at 112 °C, and strain 121 grows at 121 °C^3^. Extremophilic bacteriophages, such as ΦIN93^1^, ΦYS40^2^, ΦTMA^3^ and vB_Tsc2631^4^, are also gaining attention because of their ability to thrive at high temperatures^5^. Although extremophilic bacteriophages (known as thermophages) are not able to replicate in and lyse mesophilic pathogens^1^, there are several reports of the successful use of their enzymes, endolysins, against mesophilic multidrug-resistant bacteria^6^. However, there is a lack of comprehensive studies on the thermostability of extremophilic endolysins.
On the basis of peptidoglycan cleavage sites, endolysins are categorized into five groups: N-acetyl-β-D-muramidases (also known as lysozymes or muramidases), N-acetyl-β-D-glucosamidases, lytic transglycosylases, endopeptidases, and N-acetylmuramoyl-L-alanine amidases^7^. Amidases cleave the linkage between the stem peptide and the lactyl moiety of the muramoyl residue of the glycan strand. Among thermophilic endolysins, amidases are particularly well represented. The extremophilic Ph2119 and Ts2631 endolysins from Thermus scotoductus phages^8,9^, LysBT1 from Brevibacillus thermoruber prophage BT1^10^ and PhiKo from Thermus thermophilus phage phiKo^11^ belong to the type 2 amidase class (pfam01510), whereas AmiP from Thermus parvatiensis prophage^12^ and GVE2 from Geobacillus phage GVE2^13^ represent type 3 amidases (pfam01520). Most thermophilic amidases are globular proteins with a catalytic domain only, but there are representatives whose amidase domain is one of the modules (e.g., Lys2972 from Streptococcus thermophilus phage with the Amidase_5 domain; pfam05382)^14^. Amidases differ mainly in size, structural organization, and level of characterization, with types 2 and 3 being well-studied zinc-dependent enzymes and type 5 remaining less understood and sometimes requiring additional factors such as choline for activity^15^. Given the potential of thermophilic amidases as antibacterial agents, it is essential to thoroughly investigate the molecular determinants underlying their thermostability. Recently, Li et al. (2025) revealed that among seven cysteine residues of LysBT1 endolysin, six are involved in stabilizing the enzyme, including the catalytic C156 coordinating the Zn^2+^ ion^10^. These results indicated that the catalytic Zn^2+^ plays an important role in the thermostability of the LysBT1 enzyme. Another amidase, PhiKo, is less stable in the presence of EDTA, which sequesters Zn^2+^^11^. Although important, the above studies were only part of the overall characterization of the new endolysins.
Other reported protein stabilizing features (but not experimentally studied for endolysins) include local (intrahelical) and tertiary (interhelical) salt bridges, which are connected with relatively high contents of charged residues^16^. An inverse correlation between thermostability and the occurrence of asparagine (N) and glutamine (Q) was observed since their deamination can lead to protein breakdown^16^. Notably, thermophilic proteins are shorter than their mesophilic counterparts and tend toward shorter loops to increase protein compactness and reduce flexibility^17^. However, Das & Gerstein argued that this phenomenon is more related to phylogenetic roots than to the harsh environment since yeasts and Caenorhabditis elegans have distinctly longer proteins than representatives of Archea and Bacteria^16^. In the study of Kumar et al., the importance of salt bridges was highlighted, but the authors did not observe a correlation between thermostability and protein hydrophobicity, compactness, oligomeric states, hydrogen bonds or polar or nonpolar contributions to surface areas^18^. Instead, they noted that arginine (R) and tyrosine (Y) were predominant, whereas cysteine (C) and serine (S) were less common in thermophilic proteins. Moreover, the importance of melting temperature measurements (the temperature at which 50% of the protein is in the unfolded state; T_m_) was emphasized by the authors as the best descriptor of protein thermodynamic stability^18^. While thermodynamic stability reflects the equilibrium between folded and unfolded states, kinetic stability describes the resistance of a protein to irreversible inactivation or denaturation over time under physiological and storage conditions^19^. Kinetic stability is commonly assessed by measuring the residual enzymatic activity after thermal challenge and may not directly reflect thermodynamic stability. For example, this decoupling has been observed for the streptococcal bacteriophage endolysin PlyC, where irreversible loss of activity occurs at temperatures well below the melting temperature of individual domains^20^.
As indicated above, many studies have revealed significant differences in the distributions of individual amino acids between thermophilic and mesophilic microorganisms. However, few studies have attempted to analyse whether these more frequent amino acids contribute equally to protein stability. The aim of our work is to investigate what factors make extremophilic endolysins more stable than their mesophilic counterparts what may be useful in future attempts to increase the stability of mesophilic proteins.
In our laboratory, the extremophilic Ts2631 endolysin derived from the Thermus scotoductus bacteriophage vB_Tsc2631 was characterized in terms of antibacterial activity and structure becoming a model protein for thermostable N-acetylmuramoyl-L-alanine type 2 amidases (EC: 3.5.1.28)^4^. The enzyme has a globular architecture with Zn^2+^ in the active center. Here, the distributions of the 20 amino acids in endolysin Ts2631 and its mesophilic homologue, T7 lysozyme, as well as in bacterial and viral proteins were compared. Structural features, such as protein length and loop regions, were also analysed. In addition, single-residue substitution mutants of the Ts2631 endolysin targeting conserved and/or overrepresented amino acids, with particular emphasis on proline, tryptophan, and arginine, were tested for thermal stability and lytic activity to identify determinants of protein thermal adaptation. The T_m_ measurements with activity assays performed before and after heat treatment were combined to differentiate between thermodynamic and kinetic stability of the studied mutants. To the best of our knowledge, this is the first study of this type.
Results and discussion
Sequence analysis of the Ts2631 endolysin
To highlight thermophilic and mesophilic enzymes related to the Ts2631 endolysin, a maximum likelihood phylogenetic tree was constructed using the amino acid sequences of Ts2631 endolysin (PDB 6FHG), N-acetylmuramoyl-L-alanine amidases (whole proteins or only amidase domains) and eukaryotic peptidoglycan recognition proteins (PGRPs) (Fig. 1).
Fig. 1. Phylogenetic tree. Maximum-likelihood phylogenetic tree constructed from amino acid sequences of the N-acetylmuramoyl-L-alanine amidase domain/peptidoglycan recognition proteins (PGRPs), inferred under the WAG+R4 model. Bootstrap support values are shown above the branches. The scale bar indicates substitutions per site. Proteins with known 3D structures are shown in bold. The upper (eukaryotic) clade is marked in navy blue, the T7 lysozyme is marked in pink, and thermophilic proteins most similar to endolysin Ts2631 are marked in green. All accession numbers are listed in Table S1.
The resulting topology revealed two well-supported major clades (bootstrap support, BS = 95) representing divergent evolutionary lineages (Fig. 1). The upper clade comprises eukaryotic taxa, whereas the lower clade includes a broad diversity of bacterial taxa, several archaeal representatives, and a subset of sequences of phage origin. Within both the eukaryotic and prokaryotic clades, internal relationships are generally poorly resolved, as evidenced by low bootstrap values. This suggests either rapid diversification or extensive amino acid variability within the amidase domain, likely reflecting functional divergence and/or adaptive evolution, which impedes the resolution of fine-scale relationships. In the lower (bacterial/archaeal/viral) clade, several key substructures are apparent. A strongly supported clade (BS = 100) encompasses thermophilic bacterial taxa, including Ts2631 endolysin from vB_Tsc2631 (PDB 6FHG), Ph2119 endolysin from vB_Ph2119 (PDB 6SU5), PhiKo endolysin from vB_phiKo, and amidase from Thermus sp. (MDM7323587.1) (Fig. 1, in green). A weakly supported subgroup within the bacterial cluster appears to group mesophilic bacteria, including sequences of the T7 lysozyme of the Escherichia coli bacteriophage T7 (PDB 1LBA) and LysC from Clostridium intestinale URNW (PDB 6SSC). The low support here may reflect high sequence divergence or lateral gene transfer events, which are common in genes coding for cell wall–degrading enzymes such as amidases.
For further comparative analysis, T7 lysozyme was selected from a group of mesophiles. Despite many years of research, it is still the best known N-acetylmuramoyl-L-alanine type 2 amidase with Zn^2+^ ion in the catalytic center^21^. The first step in revealing the determinants of the thermostability of Ts2631 was to analyse the spatial architectures and amino acid compositions of both proteins.
Comparative analysis of Ts2631 endolysin and T7 lysozyme
As revealed in previous studies, despite the low sequence similarity (31.2% identity, E = 2e-12, NCBI BLAST), the structures of Ts2631 endolysin and T7 lysozyme of the Escherichia coli bacteriophage T7 superimpose well (Fig. 2A)^4^.
Fig. 2. Structural alignment and amino acid composition of thermostable Ts2631 endolysin from the vB_Tsc2631 bacteriophage of Thermus scotoductus and T7 lysozyme from the T7 bacteriophage of Escherichia coli. (A) The T7 lysozyme (PDB 1LBA) is depicted in light blue, and the Ts2631 endolysin (PDB 6FHG) is depicted in orange (on the left) or is colored on the basis of the B factor (on the right); loops are marked by squares. A greater B factor results in flexible loops in mesophilic T7 lysozyme. Structures show a r.m.s.d. between 103 pruned atom pairs of 0.853 Å (across all 137 pairs: 4.425); (B) Pie charts of amino acid distribution in Ts2631 endolysin and T7 lysozyme sequences. Amino acids that are more or less abundant in the Ts2631 endolysin sequence than in the T7 lysozyme sequence are indicated by upwards or downwards arrows, respectively.
Both proteins contain five β-strands and three α-helices in the globular region, and Ts2631 endolysin differ in the presence of an additional α-helix in the protruding N-terminal region of the protein^4,21^. However, it was not noticed before that longer loops with higher B-factors can be observed in the structure of the T7 lysozyme than in that of the Ts2631 endolysin (Fig. 2A). This feature is not reflected in the significant difference in size of the T7 lysozyme vs. Ts2631 (151 aa and 156 aa, respectively). Studies on other than endolysins thermophilic proteins indicate that shorter loop regions with lower B-factors indicate greater rigidity of these regions^22^, and this rigidification might be achieved by the presence of proline, the most rigid, nonpolar residue^19^. However, some research groups argue that the rigidifying of the structure of thermophilic proteins seems to be a simplification, as enzymatic processes such as substrate binding and release require flexibility^23^. A delicate balance of molecular flexibility and rigidity manifested by entropic and enthalpic changes is generally observed, leading to a shift in the unfolding temperature (T_m_)^24^. Nevertheless, in the case of the tested endolysins, the difference in loop length is clearly visible.
A similar loop pattern is observed in the case of other thermophilic protein, the arginine repressor ArgR from Geobacillus stearothermophilus (PDB 1B4B) and its mesophilic orthologue ArgR from Escherichia coli K12 (PDB 1XXA). With the exception of low primary sequence identity (31%), the 3D structure of both proteins can be well superimposed and differ only in loop regions, which are shorter in thermophilic protein^24^.
Since protein stability is often correlated with the presence of particular amino acids, we examined the distribution of 20 amino acids in both proteins (Fig. 2B). The first difference was the significantly greater proline (P) content in the Ts2631 endolysin, accounting for 8.3% of the residues, than in the T7 lysozyme, where prolines constituted 2.7%. One can speculate that the higher amount of proline increases the rigidity of the endolysin (especially loop regions).
Additionally, the amount of tyrosine (Y) in the endolysin Ts2631 sequence (5.8%) is greater than that in the T7 lysozyme sequence (2.1%). Moreover, the high percentage of tryptophan (W) (4.5% vs. 2.7%) is observed. In many globular proteins, tryptophan residues are frequently found in the hydrophobic core^25^ and can participate in cation–π interactions, which may contribute to protein stability.
It is believed that thermophilic proteins present a relatively high frequency of Trp in a well-buried state^26^. However, Kumar et al. reported no differences in tryptophan distribution between mesophilic and thermophilic proteins (1.2 and 1.3%, respectively)^18^. Therefore, it is important to investigate the effect of the presence of tryptophan on the thermal stability of endolysin Ts2631.
Finally, the arginine content was greater in the Ts2631 sequence than in the T7 lysozyme sequence (9.0% vs. 6.2%, respectively). Arginine contains a 3-carbon aliphatic straight chain ending in a guanidino group. The guanidinium group of arginine allows for multiple electrostatic interactions, which can stabilize protein structures, especially in the outer part of the protein^27^. It has been reported that hyperthermophiles possess a greater proportion of charged residues than their mesophilic counterparts do^28^ and that electrostatic interactions are associated with an increase in the compactness of the unfolded state at elevated temperatures^23^. Lysines (K), rather than arginines (R), are responsible for this phenomenon, as lysines give rise to a much larger population of accessible rotamers^23,29^. However, the percentage of lysines in the Ts2631 sequence was lower than that in the T7 lysozyme sequence (5.1% vs. 7.5%, respectively). This also applies to the negatively charged amino acids aspartic acid (D) (2,6% vs. 6.2%) and glutamic acid (E) (5.8% vs. 6.2%, for Ts2631 and T7 lysozyme, respectively).
In addition to the presence of certain residues, such as those mentioned above, arginines or prolines, it can’t be forgotten, that also a lower content of thermolabile amino acids leads to greater protein stability^30^. In the sequence of Ts2631, a significantly lower percentage of serine residues is observed (1% vs. 6% for Ts2631 and T7 lysozyme, respectively). A similar trend was demonstrated in other studies^18,28^. Owing to its flexibility, serine can increase the conformational entropy of the protein in the unfolded state. The replacement of serine with more rigid residue may lower the entropy of the unfolded state and thus increase the thermostability of an enzyme^25^. Moreover, serine is known to interact well with water molecules surrounding proteins^31^. The release of water bound to Ser at high temperatures can disrupt the local structure of the protein, which can lead to protein instability.
It may be inferred that the amino acid composition of T7 lysozyme is distinct from that characteristic of viral and bacterial proteins. Therefore, we compared the amino acid frequencies in the Ts2631 sequence with those observed in 518,140 viral and 30,290,647 bacterial proteins (Figure S1). We observed the same pattern and greater percentages of prolines (P), tyrosines (Y), tryptophans (W) and arginines (R) with lower percentages of serines (S), aspartic acid (D) and glutamic acid (E) in the Ts2631 sequence than in the viral and bacterial proteins. The trend was inconsistent only for lysine (K). Its frequency in viral proteins was higher (6.2%) and that in bacterial proteins was slightly lower (4.2%) than that in the Ts2631 sequence (5.1%) (Figure S1).
Effects of conserved residues on Ts2631 endolysin stability
The next step was to verify whether the exchange of individual amino acids affects the thermal stability of the Ts2631 endolysin. In our studies, we aimed to focus mainly on non-conserved residues (important for future mutagenesis studies of mesophilic proteins to increase their stability), but in order to fully investigate the stability of endolysin Ts2631, we used previously prepared mutants of conserved residues (listed in Figure S2) and examined their thermal stability.
The 21 substitution mutants designed previously^4^ were freshly purified and tested here for thermal stability (Figure S2, Table 1). The melting temperature (T_m_) measurements were performed via nano Differential Scanning Fluorimetry (nanoDSF). The T_m_ values of the active site mutants, as well as those showing the largest changes in T_m_, are presented in Table 1; the T_m_ values of the remaining mutants are shown in Table S2.
Table 1. Melting temperature of Ts2631 endolysin and its mutants at conserved positions. The decreases in T_m_ of substitution mutants of catalytic residues coordinating the Zn^2+^ ion are bold; * for reference lytic activity measured previously by turbidity reduction assays is indicated by ‘+’ - more than 60% activity relative or ‘−‘ - less than 20% or no visible activity^17^.NoProteinLytic activityT_m_ [°C] ± SDΔT_m_1Ts2631_WT+104.73 ± 0.17**2H30N**−86.48 ± 0.08−** **18.253**Y58F***−103.98 ± 0.06−** 0.754Y60A−97.97 ± 0.07− 6.765Y69A+93.34 ± 0.10− 11.396L72A+93.33 ± 0.12− 11.407C80A−> 1108H131N−84.83 ± 0.39− 19.909T137K−94.96 ± 0.08− 9.7710****C139S−89.99 ± 0.06− 14.74
The main conclusion was the much lower thermal stability of the H30N, H131N, and C139S mutants coordinating the Zn^2+^ ion but not the Y58F (values in red; Table 1). Among the four listed residues, only Y58 coordinates the zinc ion not directly but through a water molecule (small red sphere in Fig. 4). It is possible that disruption of this weaker interaction abolishes the enzymatic activity of the protein but does not affect the thermodynamic stability of the enzyme. According to the literature, the binding of metal ions often serves a stabilizing function^32^, which aligns well with our results.
The Ts2631 endolysin sequence contains three conserved tyrosine residues, Y58, Y60, and Y69. Owing to its phenolic hydroxyl group, tyrosine can act as both a hydrogen bond donor and a weak acceptor. Furthermore, its aromatic ring allows participation in short-range hydrogen bonding as well as longer-range stabilizing interactions, such as π–π stacking and hydrophobic contacts.
The role of Y58 was discussed above. The T_m_ values of Y60 and Y69 were −6.76 °C and −11.39 °C lower than the T_m_ value of wild-type Ts2631, suggesting the role of these conserved residues in stabilizing the protein structure. Additionally, the L72 mutant has a lower T_m_ value (Table 1), which indicates the involvement of this hydrophobic residue in the stability of endolysin Ts2631, most likely through its contribution to a tightly packed hydrophobic core. However, since the overall number of leucine residues is similar in mesophilic proteins, we will not analyse the role of Leu in detail (Figure S1).
Another observation is the high T_m_ for the C80A mutant (> 110 °C). Substitution of C80 leads to inactivation of the protein and abolition of the disulfide bridge visible in the crystallographic structure between C80 and C90. The disulfide bonds can stabilize proteins by maintaining their fold^33^, but surprisingly, the T_m_ of the C80A mutant was even greater than the T_m_ of the wild-type protein (Table 1). Therefore, in this particular case, the lack of disulfide bond does not decrease but rather increases the thermodynamic stability of the protein, underscoring that conformational stabilization does not necessarily confer enhanced kinetic or functional stability. It can be speculated that breaking the disulfide bond relieves local energetic frustration, enabling the Ts2631 endolysin to adopt a lower-energy conformation. The higher T_m_ of the mutant lacking the disulfide bond indicates that the intact S-S bridge imposed a locally strained, higher-energy state.
Despite their destabilizing nature, some energetic conflicts in proteins may have been positively selected during evolution owing to their functional importance^34^. These conflicts are often found near residues involved in key functions, such as catalytic activity or ligand binding; indeed, C80 is located in the peptidoglycan binding cleft, and its substitution causes Ts2631 inactivation.
The observed effect of decoupling between thermodynamic and kinetic stability has been previously reported for gamma phage endolysin PlyG^35^, where the presence of a reducing agent (e.g. tris(2-carboxyethyl)phosphine (TCEP), dithiothreitol (DTT) or glutathione), significantly enhanced the kinetic stability of the enzyme, independently of equilibrium folding assessed by T_m_.
Proline mutants
Since the analysis of the amino acid distribution in the Ts2631 and T7 lysozyme strains revealed a significantly greater amount of proline in the Ts2631 sequence, we investigated the role of this amino acid in endolysin thermostability. By site-directed mutagenesis, 13 mutants of Ts2631 endolysin were prepared, in which single proline residues were substituted with alanine. All the mutants were purified (Figure S3), and their thermal stability and lytic activity were analysed (Fig. 3). The results of nano Differential scanning fluorimetry (nanoDSF) revealed a 21.6 °C reduction in the melting temperature of the P104A mutant relative to that of the wild-type enzyme (Fig. 3A). It is striking, that the substitution of just one residue can have a profound impact on protein thermal stability, as the difference in T_m_ for the other mutants did not exceed − 7.36 °C.
Fig. 3. Stability and activity analysis of the Ts2631 endolysin and its proline substitution mutants. (A) Melting temperatures of Ts2631 and its mutants assessed by nanoDSF. SD indicates standard deviation. The difference in Tₘ between Ts2631 and the tested mutant is shown in the last column; the highest value of ΔT_m_ is marked in red. (B) Lytic activity was measured via turbidity reduction assays. The samples were not heated (black bars) or heated before the experiment for 15 min at 90 °C (gray bars). The activity is expressed as a percentage relative to Ts2631 without heating. The error bars indicate standard deviations; *p < 0.01; **p < 0.001, ***p < 0.0001; Student’s t test; all the experiments were repeated at least three times; (C) the Ts2631 endolysin core with a bond between C139 and P140 in the cis position; (D) Sequence alignment of superimposed Ts2631 with 13 of the closest structural homologues from the PDB database searched with Foldseek^36^. 6FHG – Ts2631 endolysin; 6SU5 – Ph2119 endolysin; 6SSC – LysC; 1YCK – human PGRP-S; 5XZ4 - Bumblebee PGRP-SA; 4Z8I - BbtPGRP3; 2RKQ – Drosophila PGRP-SD; 7NSX – Drosophila PGRP-LB; 2APH – human PGRP-IαC; 2F2L – Drosophila melanogaster ectodomain complex of PGRP-LCa and PGRP-LCx; 5XZ3 - Apis mellifera PGRP-SA; 1LBA – T7 lysozyme; 7NT0 – Drosophila melanogaster PGRP-LB; 2EAX – human PGRP-IβC. The position of the conserved proline is marked with a red box.
To monitor the enzymatic activity of the studied mutants, a turbidity reduction assay (TRA) was performed at 60 °C and after heat treatment (15 min at 90 °C). Mutants were added to buffer suspensions of the Thermus thermophilus HB8 substrate, and the optical density at a wavelength of 600 nm was monitored over time. At 60 °C, the activity of all the mutants exceeded 64.3%, but after heat treatment, the functionality of P140A decreased to 37.6%, confirming its thermal instability (Fig. 3B, p < 0.0001).
In the sequence of the Ts2631 endolysin, four prolines, P6, P40, P86, and P107, are present in the α-helices, and the other nine P23, P35, P54, P73, P78, P103, P124, P136, and P140 are in loop regions. Our initial assumption was that P140 rigidifies the loop regions, but other prolines, which are also found in loops, do not contribute to the structural stability of the Ts2631 endolysin. However, P140 is located very close to the active center of the enzyme, as it is the next residue after C139 which is involved in zinc ion binding (Fig. 3C). The peptide bond occurs between proline and cysteine in the cis position, which allows C139 to interact with Zn^2+^. This proline is also highly conserved in the primary sequences of other amidases and PGRPs (Fig. 3D). In the P140A mutant, substitution of proline at position 140 with alanine forces the C139 - A140 peptide bond into a trans conformation. Its effect on thermal stability depends on the local structural context; in this case, the P140A mutation decreases protein stability by disrupting the interaction between C139 and the Zn² ion.
Tryptophan mutants
With respect to the 7 prepared tryptophan substitutions, the T_m_ of the 3 mutants was reduced by more than 14.0 °C relative to that of the wild-type protein. These were W102A, W109A and W145A, whose T_m_ values were 20.41 °C, 14.32 °C, and 24.13 °C lower, respectively (Fig. 4A). The T_m_ decrease for other tryptophan mutants did not exceed 7.7 °C.
Fig. 4. Analysis of tryptophan substitution mutants. (A) NanoDSF analysis of the T_m_ of tryptophan mutants in comparison with Ts2631_WT ± SD; the highest value of ΔT_m_ is marked in red; (B) Lytic activity measured by turbidity reduction assays; black bars - samples not heated, gray bars heated for 15 min at 90 °C. The activity is expressed as a percentage relative to Ts2631 without heating. The error bars indicate standard deviations; *p < 0.01; **p < 0.001, ***p < 0.0001; Student’s t test; all the experiments were repeated at least three times; (C) Ts2631 endolysin structure with W102, W109, W145 and P140 residues (in red and blue), whose substitutions caused the greatest decrease in the Ts2631 melting point, and W53 (in light blue), with decreased lytic activity at 60 °C. The catalytic residues H30, H131, and C139, which directly coordinate the zinc ion (gray sphere), and Y58, which binds zinc via a water molecule (small red sphere), are marked with pink and blue colors. With the exception of W53, all the mentioned residues are located in the peptidoglycan binding groove.
In the activity assays at 60 °C, the W102, W109A, and W145 mutants presented moderately reduced relative activities of 94.4%, 64.7%, and 99.6%, respectively. After heating, the relative activities of these mutants were significantly reduced and amounted to 55.9% for W102A, 12.3% for W109A, and 17.0% for W145A. The experimental results are more complex, as another mutant, W53A, exhibited a significant reduction in relative lytic activity to 18.7% prior to heating (Fig. 4B). Treatment at 90 °C caused a further decrease in activity to 14.9%. In the case of W53A, only a slight reduction in the T_m_ was observed (-3.8 °C) (Fig. 4A). The circular dichroism (CD) spectra did not reveal alterations in the secondary structure of the W53A mutant, indicating that the decrease in enzymatic activity was not caused by changes in the overall folding state (Figure S5). These results further underscore that thermodynamic stability alone does not reliably predict functional or kinetic performance. W53 may interact with R51, which is in close proximity, potentially leading to a π-stacking interaction that could further stabilize the protein’s structure. Cation–π interactions, along with π-π stacking or split electrostatic hydrogen bonding, may be responsible for the proper lytic function of the Ts2631 endolysin.
The positions of W53, W102, W109, and W145 with respect to P140 and the zinc-binding amino acids are shown in Fig. 4C. With the exception of W53, all of the listed residues are located in the substrate-binding groove and span the C-terminal part of the protein.
Arginine mutants
Fourteen arginine mutants were prepared and purified to homogeneity (Figure S3). The nanoDSF analysis revealed that none of the arginine mutants presented differences in T_m_ greater than 10 °C in comparison with Ts2631_WT (Fig. 5).
Fig. 5. Analysis of arginine substitution mutants. (A) NanoDSF analysis of the T_m_ of arginine mutants in comparison with Ts2631_WT ± SD; the highest value of ΔT_m_ is marked in red; (B) Lytic activity measured by TRA; black bars - samples not heated, gray bars heated for 15 min at 90 °C. The activity is expressed as a percentage relative to Ts2631 without heating. The error bars indicate standard deviations; *p < 0.01; **p < 0.001, ***p < 0.0001; Student’s t test. All the experiments were repeated at least three times; (C) Positions of the R51, W53 and R82 residues (in navy blue) in the endolysin Ts2631 structure and corresponding residues from 3 homologous structures of the PGRPs human PGRP-IαC (in blue; PDB 2APH), human PGRP-IβC (in dark green; PDB 2EAX), and Drosophila melanogaster ectodomain PGRP-LCx (in pink; PDB 2F2L). The Lys-type peptidoglycan is depicted in yellow, and the DAP-type peptidoglycan is depicted in dark green.
Among the 14 arginine mutants analysed, the greatest T_m_ decrease was recorded for the R20A mutant (-8.55 °C) (Fig. 5A). The activity of R20A slightly decreased to 82.4% at the 60 °C (Fig. 5B), but its activity further decreased to 49.1% after heat treatment, which supports the nanoDSF results. R20 is located in the N-terminal part of the protein, the deletion of which disrupts substrate binding^4^. Therefore, we propose its role in substrate anchoring.
Moreover, similar to the W53A mutant, which has low activity at 60 °C (18.7%), the two arginine mutants also exhibited lower activity at this temperature. The activities of R51A and R82A were reduced to 41.0% and 30.2%, respectively. The CD spectra for the R51A and R82A mutants overlapped with that of Ts2631_WT, indicating that the introduced substitutions did not affect the protein structure (Figure S5).
To determine whether residues R51 and R82 are involved in substrate binding, we closely examined the mechanism of PGN coordination among different PGRPs, which are structural homologues of the Ts2631 endolysin. We superimposed the structures of the Ts2631 endolysin and three PGRPs (Fig. 5C). Two of the PGRPs recognize Lys-type PGNs (in yellow; Fig. 5C), with the stem peptide L-Ala-D-iso-Gln-L-Lys-D-Ala-D-Ala^37^. These are human PGRP-IαC and the C-terminal PGN-binding domain PGRP-IβC. Third, PGRP-LCx from Drosophila recognizes the DAP-type PGN (green; Fig. 5C) with the stem peptide L-Ala-D-iso-Glu-m-DAP-D-Ala-D-Ala^38^.
The R51 residue overlaps with R235 of PGRP-IαC, L267 of PGRP-IβC, and W392 of PGRP-LCx, which are responsible for PGN binding (Fig. 5C). The same applies to R82 overlapping with Y266 of PGRP-IαC, P298 of PGRP-IβC, and K423 of PGRP-LCx interacting with the substrate. This may suggest the potential importance of the interaction of R51 and R82 (and closely located W53) with peptidoglycan. Therefore, we tested the ability of W53A, R51A, and R82A to bind peptidoglycan isolated from T. thermophilus HB8 (Fig. 6). Since 6 out of the 14 tested arginine residues are located in the N-terminal part of the Ts2631 endolysin and deletion of this 2–22 aa region impaired peptidoglycan binding^4^, we decided to include the R2A, R9A, R12A, R15A, R18A, and R20A mutants in the analysis.
Fig. 6. Peptidoglycan binding tests. (A) Ts2631 endolysin mutants were added to peptidoglycan isolated from T. thermophilus HB8, and after incubation, the samples were washed and centrifuged to separate the unbound (U) and bound (B) fractions. Wild-type Ts2631 served as a positive control, and carbonic anhydrase served as a negative control in these tests. The frames show proteins present predominantly in the unbound fraction, CA (negative control) and the R20A mutant. The samples were subjected to tricine SDS‒PAGE. Original gels are presented in Supplementary Figure S4. M – PageRuler prestained protein ladder (Thermo Fisher Scientific); (B) Quantitative analysis of protein band intensity performed using ImageJ. The error bars indicate standard deviations; (C) Structure of Ts2631 with flexible extension marked in yellow and residue R20 depicted in pink; zoom shows the location of R20 with respect to two other residues, K70 and Y60 (in pink), that are known to take part in peptidoglycan binding.
The peptidoglycan binding tests revealed that the R51A, W53A and R82A mutants are able to bind peptidoglycan and are predominantly present in the PGN-bound fraction (marked as B in Fig. 6). These results indicate the minimal role of these residues in PGN anchoring. However, among the arginine mutants, the binding of R20A was clearly impaired (Fig. 6A). Quantitative densitometric analysis indicated that, in the peptidoglycan binding assay, 55.7% of the R20A mutant was present in the PGN-unbound fraction (Fig. 6B), indicating the role of arginine R20 in substrate coordination. This arginine is located close to two other residues, Y60 and K70, which are known to interact with PGN (Fig. 6C).
Notably, arginine residues can form salt bridges, which are involved in the stabilization of protein structures. The cationic guanidinium moiety of arginine (R) can bind to the anionic carboxylate of either glutamic acid (E) or aspartic acid (D). These residues should be located close enough (≤ 4 Å) to experience electrostatic interactions^39^. Many studies indicate that the location of salt bridges (buried or at the surface) has a greater effect on protein stability than does their number^40^. For example, salt bridges rigidify the active site of the endo-mannanase ManB-1601 from Bacillus sp., contributing to its increased kinetic stability^41^. These buried salt bridges contribute more to protein thermostability than do solvent-accessible salt bridges, whose stabilizing effect can be significantly reduced by increasing the salt concentration in the environment^40^. Concerning arginine residues, out of the 4 potential salt bridges, only the E36–R64 salt bridge is located on opposite loops (Fig. 7A). Three other, R45-E49, R67-D65, and R110-E114, are located on the same structural elements at the surface of the protein (Fig. 7B). However, the T_m_ values of the four substitution mutants R45A, R64A, R67A, and R110A, which disrupt the formation of salt bridges, are 104.1 °C, 102.9 °C, 105.3 °C, and 101.8 °C, respectively (Fig. 5A). Therefore, the thermal stability of endolysin in the absence of salt bridges formed by arginine residues remains unchanged.
Fig. 7. Structure of the Ts2631 endolysin with an enlargement of four potential salt bridges formed by arginine. (A) Interlocal salt bridge E36-R64; (B) intralocal salt bridges R45-E49, R67-D65, and R110-E114.
Conclusions
In the Ts2631 endolysin sequence, in comparison with mesophilic viral and bacterial proteins, a greater proportion of aromatic tyrosine and tryptophan residues, rigid proline, and positively charged arginine was observed. Analysis of substitution mutants of individual residues revealed their unequal contributions to the stability of the Ts2631 endolysin. For example, out of the 14 arginine mutants analysed, only R20A had a significantly reduced melting point. The amino acids in the catalytic center that directly bind zinc ion play a major role in protein stabilization. The other amino acids are R20, W102, W109, P140, and W145, the last four of which are well buried in the substrate-binding groove. The disulfide bond between C80 and C90, as well as the salt bridges formed by arginine, appear to be of lesser importance.
The mechanism of the thermostability of Ts2631 may have broader implications with respect to other thermally stable type 2 amidases. Following the structural alignment of the homologues of the Ts2631 endolysin, there is an overlap between W102, W109, P140 and W145 of the Ts2631 endolysin with other thermophilic endolysins, Ph2119, PhiKo, and not characterized to date protein from Thermus sp. (accession no MDM7323587) (Fig. 8). The presence of tryptophans selectively in thermophilic proteins emphasizes the importance of Trp for the stability of thermophilic type 2 amidases. It is possible that changing the corresponding amino acid to tryptophan in mesophilic proteins will increase their thermal stability, which is important for the prolonged storage of endolysins with antibacterial potential. The results for the C80A and W53A mutants, where the proteins had significantly reduced activity at a high T_m_ value, indicate that measurements of T_m_, while informative about equilibrium folding stability, must be complemented with functional assays to assess kinetic or functional stability. We assume that substitution of non-conserved amino acids in mesophilic proteins with tryptophan should increase their thermodynamic stability (determined by measuring T_m_) while maintaining high antibacterial activity.
Fig. 8. Comparison of the sequence fragments after structural superposition of Ts2631 endolysin with its homologues. The gray rectangles indicate amino acids R20, W102, W109, P140 and W145 of the Ts2631 endolysin and the same amino acids in the sequences of the Ts2631 homologues.
Materials and methods
Bioinformatics analysis
Fourteen N-acetylmuramoyl-L-alanine amidase protein structures of amidases or peptidoglycan recognition proteins (PGRPs) (Table S1) were downloaded from the Protein Data Bank (RCSB PDB: Homepage). Structural alignments were performed via UCSF Chimera X^42^. In the next step, homologous sequences were identified via the BLASTp algorithm, followed by multiple sequence alignment via the Clustal Omega server (Clustal Omega < EMBL-EBI). On the basis of the structural alignment, manual correction of the alignment was performed via SeaView version 5.0.5^43^. Owing to the high variability observed in the N- and C-terminal regions of the analysed proteins, only the region encoding the N-acetylmuramoyl-L-alanine amidase domain (amidase_2) was used for further analysis. As a reference, a fragment of the Ts2631 endolysin from Thermus scotoductus phage vB_Tsc2631 (PDB Entry: 6FGH), spanning amino acid positions 20 to 142 (out of the full length of 1–156), was selected. Phylogenetic analysis (maximum likelihood) and evolutionary model selection were conducted via IQ-TREE software v. 2.2.2.6^44^ and visualized by iTOL: Interactive Tree Of Life. The best-fit model was selected according to the Bayesian information criterion (BIC) - WAG+R4. Node support was assessed via bootstrap analysis with 500 replicates. The effects of the mutation on the protein’s structure, the presence of salt bridges and structural alignments were analysed in Chimera 1.18^45^, as was the visualization of the endolysin Ts2631. A search for structural homologues of the endolysin Ts2631 from the PDB database was performed via Foldseek^36^. The sequence alignment was visualized via BioEdit (https://thalljiscience.github.io/). For statistical analysis, StatSoft Statistica 13 (https://www.statsoft.pl/programy/statistica/) was used.
Plasmids, bacterial strains and growth conditions
Escherichia coli DH5α and E. coli BL21(DE3) cells (Invitrogen) were used to propagate the pET15b expression vector harboring the ts2631 gene (pET15b_Ts2631) and for recombinant protein expression, respectively. The plasmid pRARE (Cm^R^) served as a source of tRNAs for E. coli rare codons (Novagen). Bacterial cells were routinely cultivated at 37 °C in Luria Bertani (LB) or Luria agar (LA) media. When necessary, the media were supplemented with 100 µg/mL ampicillin (Merck, Germany). Thermus thermophilus HB8 DSM 579 cells were grown at 60 °C in medium 273, also known as TM broth medium (recipe from the Japan Collection of Microorganisms; https://jcm.brc.riken.jp/en/).
Site-directed mutagenesis
The primers used for site-directed mutagenesis were designed via the QuikChange^®^ Primer Design Program (available via https://www.agilent.com/store/primerDesignProgram.jsp) and are listed in Supplementary Table S2. The pET15b_Ts2631 plasmid served as a template for the PCR. The entire procedure was performed according to the instruction manual of the QuikChange Site-Directed Mutagenesis Kit (Agilent Technologies). Thirty-one residues of the Ts2631 endolysin primary sequence were changed to alanine, and in each case, the correct mutant was confirmed by sequencing the corresponding plasmid DNA. Twenty-one additional mutants constructed during previous studies were used in this work^4,9^.
Protein overproduction and purification
E. coli BL21(DE3) bacterial cells carrying plasmids encoding wild-type Ts2631 endolysin or its substitution mutants were cultivated in 1 L of LB medium supplemented with ampicillin to an OD_600_ of 0.4. To induce protein overproduction, isopropyl-β-d-thiogalactopyranoside (IPTG) was added at a final concentration of 1 mM. The bacterial cultures were incubated for 4 h at 37 °C. Next, the cells were centrifuged (10,000 × g, 20 min, 4 °C), and the pellet was suspended in 3 mL of NP buffer (50 mM NaH_2_PO_4_ pH 8.0, 300 mM NaCl, 5% (vol/vol) glycerol) containing 10 mM imidazole and stored at -80 °C. For protein purification, the cell paste was suspended in 5 mL of NP buffer and sonicated (30 bursts of 10 s at an amplitude of 12 μm; Branson Ultrasonics Sonifier SFX550). The lysate was cleared by centrifugation (10,000 × g, 20 min, 4 °C), and the supernatant was injected into a 5 mL HiTrap TALON crude cobalt-based chromatography column (Cytiva). Immobilized metal affinity chromatography (IMAC) was performed via the ÄKTA pure system (Cytiva) according to the HiTrap TALON crude column instruction. The resin was washed with NP buffer containing 10 mM imidazole until the absorbance at 280 nm reached a steady baseline, and the proteins were eluted with NP buffer with 150 mM imidazole. SDS‒PAGE (15%) was used to analyse the purity of the eluted proteins. Fractions containing Ts2631 endolysin or its mutants were pooled and dialyzed against three different buffers depending on the subsequent analysis. Buffers A (20 mM HEPES, pH 7.4) and B (20 mM MES, pH 6.0) with 25 mM NaCl and 10% (vol/vol) glycerol were utilized for measuring the protein’s melting temperature (T_m_; the temperature at which 50% of the protein is unfolded). For these two buffers, buffer A was selected to conduct turbidity reduction assays (TRAs). Buffer C (10 mM potassium phosphate pH 8.0, 150 mM (NH_4_)2_SO_4) with 10% (vol/vol) glycerol was used for circular dichroism tests. The Bradford protein assay was performed to determine protein concentrations. All the proteins were stored at -80 °C until further analysis.
Preparation of the Thermus thermophilus HB8 substrate and turbidity reduction assays
Turbidity reduction assays (TRAs) were performed as described previously^17^. Briefly, Thermus thermophilus HB8 cells were cultivated in TM broth at 60 °C to an OD_600_ of 0.6 and then centrifuged (5,000 × g, 15 min, 4 °C). The pellet was suspended in chloroform/50 mM Tris-HCl, pH 7.7, to permeabilize the outer membrane. The suspension was left at room temperature for 45 min and then centrifuged as described above. The cells were washed with 10 mM potassium phosphate, pH 8.0, and suspended in the same buffer. Permeabilized bacteria were stored at -80 °C for further use. For TRA, the bacterial suspension was adjusted to an OD_600_ of 1.0 in buffer A. The protein concentration was adjusted to 0.1 µg/µL. In the test, the proteins were either heated at 90 °C for 15 min or left untreated on ice. Ten microlitres of protein and 190 µl of substrate were added to a flat bottom, nonbinding surface, polystyrene 96-well plate (Corning) in triplicate (the final protein concentration was 5 µg/mL). The assay was performed at 60 °C for 15 min at 1-minute intervals in an EnSpire multimode plate reader (PerkinElmer). The negative control consisted of protein storage buffer added to the substrate. The lytic activity was calculated via the formula [ΔOD_600_ sample (endolysin added) − ΔOD_600_ (buffer only)]/initial OD_600_].
Nano differential scanning fluorimetry
Ts2631 endolysin and its mutants at concentrations of approx. 0.5 mg/mL in buffer A or B were equilibrated to room temperature and centrifuged at 10,000 × g for 1 min to remove air bubbles. The samples were subsequently loaded into Prometheus Standard Capillaries, which were sealed with capillary sealing paste (NanoTemper Technologies) to prevent evaporation. The measurements were conducted at a temperature range from 20 °C to 110 °C with 1 °C intervals using a Prometheus NT.48 (NanoTemper Technologies). The fluorescence was excited at 280 nm, and the intensity of the tryptophan emission was detected at 330 nm and 350 nm to calculate the F350/F330 ratio for data analysis via the PR. Stability analysis software. This software allows for the determination of the protein’s T_m_ on the basis of changes in fluorescence intensity as a function of temperature. In preliminary experiments, the T_m_ of the enzyme was measured in two 20 mM buffers, namely, HEPES (pH 7.4) and MES (pH 6.0), supplemented with 25 mM NaCl and 10% glycerol (Figure S6). The T_m_ of the wild-type enzyme was above 100 °C in both buffers, 104.7 ± 0.2 °C in HEPES, and 101.6 ± 0.1 °C in MES. The difference in the Ts2631 melting point measured during previous studies in 20 mM MES, pH 6.0 (T_m_ = 99.8 °C)^9^vs 101.6 °C in 20 mM MES, pH 6.0, 25 mM NaCl, and 10% glycerol may be due to the stabilizing effect of the added salt and glycerol. For further analyses, 20 mM HEPES, pH 7.4, 25 mM NaCl and 10% glycerol buffer was selected. The same buffer was used to conduct the turbidity reduction assays.
Peptidoglycan isolation
T. thermophilus HB8 cells were grown at 60 °C to an OD_600_ of 0.7 in 1.5 L of TM broth medium (6 flasks of 250 mL). After centrifugation (5,000 × g, 18 °C, 10 min), each pellet (6 samples) was suspended in 3 mL of phosphate-buffered saline (PBS, Merck, Germany) and combined with 6 mL of 6% boiling SDS (Merck, Germany). The bacterial suspension was incubated in boiling water with shaking (187 rpm, 100 °C, 3 h). After boiling, the heat was turned off, and the sample was left overnight with shaking. The next day, the mixture was heated for 2 h to dissolve the precipitated SDS. The pooled samples, each 27 mL, were subsequently transferred to two 30 mL polypropylene copolymeric tubes (Thermo Scientific Nalgene) and centrifuged at 50,000 × g at 18 °C for 20 min (Beckman Optima XPN0199, rotor 70. Ti, Germany). The supernatant was gently removed, and the pellet was suspended in 20 mL of ultrapure water. Centrifugation was repeated two more times, and after the last wash, each sample was suspended in 2.7 mL of 10 mM Tris-HCl, pH 7.2, and 0.06% NaCl. The samples were transferred to 6 Eppendorf tubes in aliquots of 900 µL. Pronase E (Merck, Germany) at a concentration of 1 mg/mL was activated for 30 min at 60 °C. One hundred microliters of activated pronase E was added to each sample of peptidoglycan and incubated for 2 h at 60 °C. To stop the reaction, 200 µL of 6% SDS was added to the samples, which were subsequently incubated for 30 min at 100 °C. The samples were pooled again in one polypropylene copolymeric tube, ultracentrifuged, and washed with 20 mL of water as described above until the SDS was removed (with no air bubbles). After the last centrifugation, the sample was suspended in 10 mL of ultrapure water, aliquoted into 100 µL portions in 100 Eppendorf tubes and stored at -80 °C.
Peptidoglycan binding test
The PGN binding test was performed as previously described^4^. Briefly, for the PGN binding test, 3.6 µg of Ts2631 endolysin or its mutants were mixed with 100 µL of T-M buffer (10 mM Tris-maleate buffer, pH 6.5, 1 M NaCl) and 150 µL of PGN that had been washed three times with T-M buffer. Each mixture was incubated at 4 °C for 30 min. Next, the samples were centrifuged at 12,600 × g at 4 °C for 5 min. Twenty microliters of the supernatant was retained for further analysis, and the remainder was discarded. The pellet was washed three times with T-M buffer, and before the last centrifugation, it was transferred into a new tube. After the final centrifugation, 20 µl of a solution containing 62.5 mM Tris-HCl, pH 6.8, 2% (vol/vol) SDS, 19% (vol/vol) glycerol and 5% β-mercaptoethanol was added. To the fractions of the supernatant (which contained unbound protein) and the pellet (where peptidoglycan was sedimented), 6 µL of Laemmli buffer was added. Tricine-SDS‒PAGE was subsequently performed, and the gel was stained overnight with Coomassie brilliant blue R-250. Carbonic anhydrase (29 kDa, SERVA) was used as a negative control. Densitometric analysis of protein bands was performed using ImageJ software^46^.
Circular dichroism spectra
Proteins were dissolved in buffer C to a final concentration of 0.15 mg/mL (7.14 µM). A J-815 circular dichroism (CD) spectropolarimeter (JASCO) with a path length of 1 mm was used for data collection. The final circular dichroism spectrum is an average of six scans.
Supplementary Information
Below is the link to the electronic supplementary material.
Supplementary Material 1
The reference list from the paper itself. Each links out to its DOI / PubMed record.
- 1Liu, H., Kheirvari, M. & Tumban, E. Potential applications of thermophilic bacteriophages in one health. Int. J. Mol. Sci.24 (9), 8222 (2023).10.3390/ijms 24098222 PMC 1017906437175929 · doi ↗ · pubmed ↗
- 2Plotka, M., Kapusta, M., Dorawa, S., Kaczorowska, A. K. & Kaczorowski, T. Ts 2631 endolysin from the extremophilic Thermus scotoductus bacteriophage v B_Tsc 2631 as an antimicrobial agent against gram-negative multidrug-resistant bacteria. Viruses 11 (7), 657 (2019).10.3390/v 11070657 PMC 666986231323845 · doi ↗ · pubmed ↗
- 3Oechslin, F. et al. Fermentation practices select for thermostable endolysins in phages. Mol. Biol. Evol.41 (3), msae 055 (2024).10.1093/molbev/msae 055PMC 1098051738489607 · doi ↗ · pubmed ↗
- 4Hoopes, J. T., Heselpoth, R. D., Schwarz, F. P. & Nelson, D. C. Thermal characterization and interaction of the subunits from the multimeric bacteriophage endolysin Ply C. Biology (Basel)12 (10), 1277 (2023).10.3390/biology 12101277 PMC 1060420937886987 · doi ↗ · pubmed ↗
- 5Bechtel, T. J. & Weerapana, E. From structure to redox: the diverse functional roles of disulfides and implications in disease. Proteomics 17 (6), 10.1002 (2017).10.1002/pmic.201600391 PMC 536794228044432 · doi ↗ · pubmed ↗
- 6van Kempen,et al. Fast and accurate protein structure search with foldseek. Nat. Biotechnol.42 (2), 243–246 (2023).10.1038/s 41587-023-01773-0PMC 1086926937156916 · doi ↗ · pubmed ↗