(C) PLOS One
This story was originally published by PLOS One and is unaltered.
 . . . . . . . . . .



Phage endolysins are adapted to specific hosts and are evolutionarily dynamic [1]

['Frank Oechslin', 'Département De Biochimie', 'De Microbiologie', 'Et De Bio-Informatique', 'Faculté Des Sciences Et De Génie', 'Université Laval', 'Québec City', 'Groupe De Recherche En Écologie Buccale', 'Faculté De Médecine Dentaire', 'Xiaojun Zhu']

Date: 2022-08 

Endolysins are produced by (bacterio)phages to rapidly degrade the bacterial cell wall and release new viral particles. Despite sharing a common function, endolysins present in phages that infect a specific bacterial species can be highly diverse and vary in types, number, and organization of their catalytic and cell wall binding domains. While much is now known about the biochemistry of phage endolysins, far less is known about the implication of their diversity on phage–host adaptation and evolution. Using CRISPR-Cas9 genome editing, we could genetically exchange a subset of different endolysin genes into distinct lactococcal phage genomes. Regardless of the type and biochemical properties of these endolysins, fitness costs associated to their genetic exchange were marginal if both recipient and donor phages were infecting the same bacterial strain, but gradually increased when taking place between phage that infect different strains or bacterial species. From an evolutionary perspective, we observed that endolysins could be naturally exchanged by homologous recombination between phages coinfecting a same bacterial strain. Furthermore, phage endolysins could adapt to their new phage/host environment by acquiring adaptative mutations. These observations highlight the remarkable ability of phage lytic systems to recombine and adapt and, therefore, explain their large diversity and mosaicism. It also indicates that evolution should be considered to act on functional modules rather than on bacteriophages themselves. Furthermore, the extensive degree of evolvability observed for phage endolysins offers new perspectives for their engineering as antimicrobial agents.

Funding: F.O. was supported by the Schweizerischer Nationalfonds zur Förderung der Wissenschaftlichen Forschung under grants P400PB_191059 and P2LAP3_181297. X.Z. was supported by graduate scholarships from Fonds de Recherche du Québec - Nature et Technologies (1C-203754). M.B.D. is recipient of a graduate scholarship from Fonds de Recherche du Québec - Nature et Technologies (259257). S.M. acknowledges funding from the Natural Sciences and Engineering Research Council of Canada (RGPIN/06705-2019), Canada Research Chair (950-232136) and Fonds de Recherche du Québec - Nature et Technologies (188158). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

In this study, we used phages that infect the gram-positive Lactococcus lactis as a model to study the roles of endolysin diversity on phage biology, host adaptation, and evolution. Analysis of 253 lactococcal phage genomes revealed 10 different types of endolysins, which were specific to phage groups. Although different from a biochemical perspective, genes coding for these endolysins and additional ones from phages infecting other bacterial species could be exchanged in different lactococcal phage genomes. The fitness costs were marginal if both recipient and donor phages were infecting the same bacterial strain, but gradually increased between phages that infect different strains or bacterial species. We also observed that phage lytic modules can be naturally exchanged between virulent phages and prophages. Finally, we showed that phage endolysins can rapidly adapt to their new phage/host environment by acquiring adaptative mutations.

The biological relevance for such diversity is not well understood. A possible explanation might be related to the evolutionary pressure that is put on phages to adapt their lysis system to a diversified and changeable bacterial cell wall [ 12 ]. It has also been proposed that the coevolution of phages along with their bacterial hosts selects for endolysin domains that target cell wall associated factors that are essential for host viability [ 23 ]. The presence of a specific type of endolysin could also be the result of an adaptation related to the phage’s preferred host. The PlyG endolysin only lyses specific Bacillus anthracis strains, which closely matches the host range of the phage [ 24 ]. Similarly, the CBD of Listeria phage endolysins Ply118 and Ply500 were shown to have a ligand binding specificity at the serovar level, like their respective phages [ 13 ]. Finally, a more provocative hypothesis would be that the observed endolysin diversity has no direct implication for the phage biology and is only a consequence of active domain exchanges between phages. Thus, one would expect that phage endolysins are interchangeable even between genetically unrelated phages.

The endolysin diversity is illustrated in mycobacteriophages as 26 endolysin structures were observed through various combinations of 15 domains, even if the 220 analyzed phages were infecting the same Mycobacterium smegmatis mc 2 155 host [ 21 ]. One of the first metanalysis studies that described the genetic diversity of endolysins reported 89 types of structures from phages infecting 64 bacterial genera [ 12 ]. More recently, a database of 2,182 endolysin sequences was analyzed for possible correlations between domain families and bacterial hosts [ 22 ]. Remarkably, a clear differential distribution was observed between phages that infect gram-positive or gram-negative bacteria, except for amidase CDs, which were found in phages that infect both Gram types. In the case of gram-positive bacteria, amidase CD or LysM CBD were widely distributed. Conversely, other domains like PSA CBD or CPL1 CBD were restricted to phages infecting Listeria or streptococci. Importantly, no bacterial genus was associated with just 1 endolysin architectural composition.

In addition to the CD, endolysins that target gram-positive bacteria usually have an additional C-terminal cell wall–binding domain (CBD) connected by a flexible linker [ 11 ]. Endolysins from phages that infect gram-negative bacteria rarely exhibit this modular organization and usually have only a CD [ 12 ]. CBDs are known to provide specificity for certain types of molecules present or associated with the peptidoglycan and can noncovalently attach to them [ 13 ]. Several CBDs have been described and include, among others, LysM domains. These domains interact with the sugar backbone of the peptidoglycan and are reported to be the most common [ 14 , 15 ]. The structure of these enzymes can also include more than 2 modules. Endolysins with 2 CDs and 1 CBD at the C-terminal [ 16 , 17 ] or central position [ 18 , 19 ] were reported from staphylococcal and streptococcal phages. Some endolysins can even be multimeric and composed of 2 gene products, as with the PlyC endolysin [ 20 ].

Peptidoglycan, the main component of the bacterial cell wall, provides mechanical resistance for cell integrity. It is composed of a complex meshwork of N-acetylglucosamine (GlcNAc)–N-acetylmuramic acid (MurNAc) glycan strands that are cross-linked by short stem peptides attached to MurNac residues [ 8 ]. Variation in the composition of the peptidoglycan has been observed between bacterial species, with approximately 100 types described to date [ 9 ]. Consequently, a wide variety of catalytic domains (CDs) has been observed among phage endolysins, which can cleave the glycosidic bonds between the sugar moieties (lysozymes/muramidases), the glycan–peptide linkage (amidases) and the stem peptide or its cross-bridge (endopeptidase) [ 10 ].

For most dsDNA phages, host lysis at the end of the replication cycle is due to the coordinated actions of 2 proteins. Holins are proteins that control the timing of lysis by permeabilizing the inner membrane of the host to allow the diffusion of the lytic enzymes, namely, the endolysins. The latter then gains access and degrades the cell wall peptidoglycan to induce lysis [ 5 ]. Destabilization of the outer membrane with the help of a third type of proteins called spanins is also required for phages that infect gram-negative bacteria [ 6 ]. In addition, some endolysins do not rely on holins but instead use signals to interact with the general host secretion pathway [ 7 ].

Bacteriophages (phages) exhibit exceptional structural and genetic diversity [ 1 ]. A key feature of their genome organization is its mosaic gene composition, which results in the absence of universal genes. Still, individual genes or genetic regions can be shared between unrelated phage genomes [ 2 ]. Horizontal gene transfer between nonidentical ancestors is a major mediator of phage evolution, and phages that infect the same host may exhibit considerable diversity [ 3 ]. Phages are found in all studied biomes and are estimated to kill half of the global bacterial population every 48 h [ 4 ]. For this reason, phage-induced lysis is perhaps the most common fate for bacteria after cell division [ 5 ].

Results

Endolysins of lactococcal phages can be grouped into 11 types First, we investigated the diversity of endolysins in virulent and temperate phages that infect L. lactis. A set of 253 complete lactococcal phage genomes, ranging from 21,562 bp (phage 50504) to 132,949 bp (phage AM4) in length, were obtained from NCBI and analyzed for the presence of endolysins, conserved domains, and phylogenetic relatedness (Fig 1 and S1 Data). Four types of CDs (group A: amidase_2, group B: CHAP, group C: GH25_Cpl1-like, group D: lysozyme-like muramidase) were predicted based on HHPred and BLASTP. PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 1. Phylogenetic relationship of endolysins of phages infecting Lactococcus lactis. The diversity of endolysins from 253 complete lactococcal phage genomes was investigated. We used ClustalW (v2.1) to perform multiple alignments and generate a phylogenetic tree (S1 Data). Conserved CDs and CBDs were determined based on HHPred and BLASTP predictions. The 253 endolysins were classified into 4 groups according to their CDs and are represented by different colors. Each phage is followed by the accession number of its respective endolysin. The phage taxonomic status is indicated when more than 1 species/genus is represented. CBD, cell wall–binding domain; CD, catalytic domain. https://doi.org/10.1371/journal.pbio.3001740.g001 Endolysins of group A have a deduced average molecular weight of 27.1 ± 2.8 kDa and are the most abundant as they were observed in 81% (n = 205) of these genomes. They all have a predicted N-terminal amidase_2 CD but are linked to diverse CBDs. Their C-terminal region shares homology with either the CBD of the Enterococcus phage endolysin LysIME-EF1 (phage 1706 and members of the Skunavirus genera) or the SH3b domain of lysostaphin (phage P087 and Skunavirus phages) or an unrecognized CBD (phages 1706, KSY1, and 949 as well as a few Skunavirus). Endolysins of group B (7.5%, n = 19) have a molecular weight of 28.4 ± 3.1 kDa and a CHAP CD (Cysteine, Histidine-dependent Amidohydrolases/Peptidases). The cysteine and histidine residues, which are the hallmarks of CHAP domains [25], are conserved in all of them (cysteine at position 30 ± 3 amino acids and histidine at position 86 ± 5 aa; S1 Fig). Except for the endolysins from phage 1358, which had a SH3b-type CBD, the rest of the group B endolysins had C-terminal regions having no homology with known CBDs. Endolysins of groups C and D were predicted to be muramidases. In group C (6.7%, n = 17), the GH25_Cpl1-like muramidase domain is associated with a predicted N-terminal transmembrane domain (TMD) of approximately 20 aa, followed by a signal peptidase I (SPaseI) cleavage site (CS) and 2 C-terminal LysM CBDs (46.3 ± 0.1 kDa) (S2 Fig). The only exception is the endolysin of phage 4268 (P335 group), which has a C-terminal sharing homology with the CBD of LysIME-EF1. In group D, the lysozyme-like muramidase CD (4.7%, 24.8 ± 0.7 kDa, n = 12) is always associated to a C-terminal region that shares homology with the CBD of LysIME-EF1. Taken altogether, the combination of these CD and CBD domains resulted in 11 types of endolysins. In general, the phylogeny of the endolysins followed the taxonomy of lactococcal phages, including for the 3 most common groups (Skunavirus, Ceduovirus, and P335). Phages belonging to the Skunavirus genus have endolysins with amidase_2 CDs. Phages from the P335 group have endolysins with GH25_Cpl1-like and CHAP CDs, and Ceduovirus phages only had lysozyme-like muramidases. As host cell lysis by lactococcal phages requires the presence of holins, we also investigated holin diversity and whether there is any correlation with the endolysin type. According to the number of TMDs (TMHMM tool), phylogenetic relatedness, and homology, holins were classified into at least 17 groups (S3 and S4 Figs and S7 Data). The group A endolysins found in Skunavirus phages was either associated with class II holins or class I holins with 3 putative TMDs. The other lactococcal phage groups that possessed group A endolysins either has class I holins (Q54), class II holins (P087), or class III holins with 1 TMD (KSY1, 1706, and 949). Group B endolysins were either associated with class II holins in P335 phages or type I holins in the 1358 and P034 groups. Endolysins from Group C were associated with class III holins or class II holins that belong to the phage_holin_1 and Dp1 superfamily. Group D muramidases were always associated with class I holins with 3 TMDs. Overall, we noticed that some types of holins were always associated with 1 type of endolysin, while other endolysins were associated with multiple types of holins.

Endolysins from lactococcal phages are biochemically different We characterized 1 representative endolysin from each of the 4 CD groups. The genes coding for the endolysins of the virulent phages P008 (Skunavirus, LysP008, group A, amidase_2 CD and IMEEF1 CBD), 1358 (Lys1358, group B, CHAP endopeptidase CD and SH3b CBD), and c2 (Ceduovirus, Lysc2, group D, lysozyme-like CD and IMEEF1 CBD) were cloned into the expression vector pET28, introducing 6-His at the C-terminal position of the enzyme (Fig 2A). The endolysin gene from the virulent phage P335 (LysP335, group C, GH25_Cpl1-like muramidase CD and 2 LysM CBD) was cloned into the expression vector pETG20a to add a TRX-tag at the N-terminal position. This tag was necessary to improve the expression of the protein, which was also cloned both with and without its TMD (Fig 2A). The endolysins were overexpressed in Escherichia coli and purified using Ni-NTA affinity chromatography. The purity and molecular weight of each of the purified protein were verified on 4% to 12% BisTris gels (S5 Fig). PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 2. Biochemical characterization of the endolysins of the lactococcal phages P008, c2, 1358, and P335. (A) The endolysin of phage P008 (LysP008) is composed of an amidase_2 CD that is known to hydrolyse the glycan–peptide linkage and a C-terminal region with homology to the CBD of the endolysin found in the Enterococcus phage IMEEF1 [26]. The endolysin of phage c2 (Lysc2) has a lysozyme-like CD that can hydrolyse the glycosidic bonds between the sugar moieties and a C-terminal region that also shares homology with the CBD of phage IMEEF1 endolysin [26]. The endolysin of phage 1358 (Lys1358) has a CHAP CD and a predicted SH3 CBD. CHAP domains are endopeptidases that can hydrolyse the peptidoglycan crossbridge [25]. These 3 endolysins are associated with the class I holins and could be purified by affinity chromatography using a his-tag at the C-terminal of the enzyme. The endolysin from phage P335 (LysP335) was associated with a class II holin and has a more complex structure. LysP335 has an N-terminal predicted secretion signal composed of a TMD followed by a GH25_Cpl1-like domain with predicted glucosamidase activity and 2 C-terminal LysM CBDs. A predicted cleavage sequence was also observed after the TMD, which suggests that the LysP335 is exported to the periplasm in a holin-independent manner and stays at the membrane until it is released and activated through cleavage of its TMD. LysP335 and a construct without its TMD (LysP335ΔTMD) were also purified using affinity chromatography but with the addition of a TRX tag that was necessary for expression. (B) The lytic activity of the purified endolysins was characterized by following the decrease in turbidity of L. lactis IL1403 cells in exponential growth phase. The concentrations used corresponded to their specific activities, which were defined as the amount of enzyme needed to decrease the absorbance by 50% in 15 min (S1 Data). (C) As no lytic activity could be measured on L. lactis cells for the 2 purified LysP335 recombinant proteins, their enzymatic activity was tested on purified L. lactis IL1403 CW. Because LysP335 has a predicted glucosamidase activity, a Park–Johnson assay was used to measure glycan hydrolysis through the quantification of reducing groups, expressed in mg/ml of glucose equivalents released in the incubation mixture. The different endolysins (10 μM) were incubated with 100 mg/ml of cell wall. Solutions composed of the LysP335 endolysin, CW, or only buffer were used as controls (S1 Data). D) The host range of the endolysins LysP008, Lysc2, and Lys1358 was determined on several L. lactis strains. The activity of the endolysins was normalized for comparison and according to their specific activity measured on strain IL1403. The color gradient indicates the percent of decrease in absorbance measured over time. Values are means and standard deviations from triplicates (S1 Data). CBD, cell wall–binding domain; CD, catalytic domain; CW, cell wall; TMD, transmembrane domain. https://doi.org/10.1371/journal.pbio.3001740.g002 As previously mentioned, LysP008 and Lysc2 have a CBD with homology to the one present in the LysIME-EF1 endolysin [26]. Due to the presence of an alternative start codon before the CBD of LysIME-EF1, the endolysin was previously observed to form a tetramer composed of the full-length enzyme and 3 additional CBDs. We also observe that LysP008, Lysc2, and related endolysins appear to have an alternative start codon before their CBDs (S6 Fig). However, expression of the LysP008 and Lysc2 genes did not resulted in the production of 2 polypeptides (S6 Fig and S1 Data), even without codon optimization (S7 Fig). The enzymatic activity of the endolysins was tested by monitoring the turbidity decrease of a suspension of L. lactis IL1403 cells over time (Fig 2B and S1 Data). Our data clearly indicate that the addition of purified LysP008, Lys1358, and Lysc2 led to cell lysis, but LysP335 did not (with and without TMD). Although LysP335 could not lyse lactococcal cells, catalytic activity could still be measured when the enzyme was incubated with purified lactococcal cell walls (Fig 2C and S1 Data). Indeed, the hydrolysis of the glycan part of the peptidoglycan was observed with the LysP335ΔTMD construct only, and at a rate similar to Lysc2, which is consistent with their glucosamidase activity. The absence of activity with the full LysP335 suggest that the TMD must be removed to activate the enzyme. This might be achieved through the general secretion pathways and the action of signal peptidase I [27] as a SPaseI CS was observed between the TMD and the rest of the protein (S2 Fig). Moreover, LysP335 had to be purified as a membrane protein, which might indicate its location prior to its release and activation by proteolytic cleavage (Fig 2A). LysP008, Lysc2, and Lys1358 also had different specific activities and host ranges. Lysc2 was the most active, as a concentration of 0.025 μM was enough to achieve a turbidity decrease of 50% in 15 min compared to 0.31 μM for LysP008 and 2.5 μM for Lys1358. To further compare their activity, each enzyme was normalized to a concentration needed to achieve a 50% turbidity decrease in 15 min using L. lactis IL1403. Next, these enzymes were tested on 18 L. lactis strains. Interestingly, variation in lytic activities was observed according to the type of endolysins and strains used (Fig 2D and S1 Data). Indeed, a turbidity decrease of at least 50% was observed after 30 min for 11 out of 18 (61%) tested strains for LysP008, 12 of 18 (67%) for Lysc2, and 16 of 18 (89%) for Lys1358.

Endolysin genes can also be exchanged by homologous recombination between virulent phages and prophages An interesting feature of the P335 endolysin (muramidase) was the presence of 2 LysM CBDs. We set up a CRISPR-Cas9 assay to remove these 2 domains and determine if they were essential for LysP335 activity. In situ deletion of the LysM domains was attempted using a strain carrying a Cas9-plasmid targeting the lysP335 gene and a second plasmid (repair template) with lysP335 without the sequence coding for the LysM domains. Infection of the recombinant strain with the virulent phage P335 resulted in an expected phage titer reduction (>5 logs, Fig 5A and S1 Data) due to Cas9 interference activity. Unexpectedly, we did not observe the anticipated deletion within the endolysin gene in the 16 recovered phages analyzed. Instead, these phages that had either a mutation in the protospacer adjacent motif (PAM) (4/16) or had acquired a different endolysin gene (12/16). Genome sequencing of 4 of these LysP335-negative mutants showed that they had acquired an endolysin gene from a known prophage (bIL286) of L. lactis IL1403. We estimated the recombination frequency at 2.5 × 10−6 (Fig 5B). Analyses of the recombinant phages (named P335/bIL286) indicated that a genomic region encompassing genes coding for a neck passage structure, holin, and type B endolysin was exchanged (Fig 5C). This exchange was likely due to homologous recombination, as the flanking genomic regions share homology between both the virulent phage P335 and the prophage bIL286 (Fig 5C). It also suggested that the deletion of the 2 LysM domains was detrimental, leading to the selection of these functional recombinant phages. PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 5. Exchange of endolysin-coding genes by homologous recombination between different phages. (A) Phage P335 was titered on the host strain L. lactis IL1403 and on the phage-resistant L. lactis IL1403 + pL2Cas9 LysP335 + pNZ123 LysΔM1–2 (S1 Data). (B) Graphical representation of the prophages integrated in the genome of L. lactis IL1403 and the recombination event that took place between the virulent phage P335 and prophage bIL286. (C) Graphical representation of the genomic region exchanged between P335 and bIL286 and includes part of the gene coding for the neck passage structure and the entire lysis cassette (holin and endolysin genes). (D) Phylogenetic analyses based on the genomes of lactococcal virulent phages and prophages (S1 Data). Sequence accession numbers are shown for virulent phages. For prophages, the positions of the start and end of the genomes are indicated after the accession number of the L. lactis accession number from which it was extracted. The type of endolysin CD and phage taxonomy is indicated by each color. CBD, cell wall–binding domain; CD, catalytic domain; TMD, transmembrane domain. https://doi.org/10.1371/journal.pbio.3001740.g005 Due to this endolysin–gene swapping between virulent phages and prophages, we further investigated the genetic relatedness of our 253 phages and a set of 54 prophages retrieved from the genome of 26 L. lactis strains (Fig 5D and S1 Data). The analysis yielded results that were like the endolysin-based phylogeny, as phages were observed to cluster according to their endolysin types. However, the clustering was less obvious in prophages that belong to the P335 group, as some of them shared genetic similarities and were found to have either type B or C CDs. This was exemplified by the virulent phages ul36.k1t1 and ul36.t1 (86% coverage, 99.8% identity), phage 38502 and prophage NZ_CP015908_1004577–1046548 (46% coverage and 99.2% identity), or prophages NC_013656_1063356–1118507 and CP015898_964690–1021768 (54% coverage, 94.7% identity). The analysis further supported the possibility that different types of endolysins can be exchanged between related virulent phages as well as prophages.

Adaptation was observed during experimental evolution of the different phage endolysin mutants As shown above, the exchange of endolysin gene may affect, in some cases, phage growth. Thus, we explored if these less efficient phage mutants would rapidly adapt to their new endolysins. We set up a short-term experimental evolution assay, in which the P335 and P008 phage mutants were amplified in 20 serial transfers (Fig 7A). Filtration of the infected cultures was performed between each of the transfers to ensure that only phages were allowed to carry over and evolve. First, we measured the plaque size of the phages after 1 (T1), 10 (T10), and 20 (T20) transfers to assess whether adaptation occurs during serial amplification. All phage mutants produced significantly larger plaques at T20 compared to T1 (p < 0.001, Welch two-sample t test), except for P008>Lysc2 (p = 0.0215) (Fig 7B–7D and S1 Data). When looking at the increase of plaque sizes from T1 to T20, larger differences were particularly observed among phage mutants with endolysins from phages infecting other bacterial species (Fig 7E and S1 Data). These data suggested that the phage mutants indeed adapted to their new endolysin gene, leading to an increase in their plaque size. PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 7. Experimental evolution of the phage mutants generated during the study. (A) Endolysin mutants from phages P335 and P008 were successively amplified for 20 transfers. L. lactis IL6288 cells in the exponential growth phase (OD 600nm 0.2) were infected at an initial MOI of 10. After an incubation period of 12 h at 30°C, phage lysates were filtered and diluted before another round of amplification was started. Dilutions of 1/1,000 were used for phages P335, P335>LysP008, P335>Lysc2, P335>Lys1358, P008>LysEfaS, P008>LysFL3B, and P008>LysLfeSau and dilutions of 1/10,000 for phages P008, P008>LysP335, P008>Lysc2, and P008>Lys1358. (B, C, and D) Size of the lytic plaques of the different phages after 1, 10, and 20 transfers (S1 Data). (E) plaque-size increased from T1 to T20 (S1 Data). Measures were done in triplicates. https://doi.org/10.1371/journal.pbio.3001740.g007

[END]
---
[1] Url: https://journals.plos.org/plosbiology/article?id=10.1371/journal.pbio.3001740

Published and (C) by PLOS One
Content appears here under this condition or license: Creative Commons - Attribution BY 4.0.

via Magical.Fish Gopher News Feeds:
gopher://magical.fish/1/feeds/news/plosone/