Phage resistance profiling identifies new genes required for biogenesis and modification of the corynebacterial cell envelope

  1. Amelia C McKitterick
  2. Thomas G Bernhardt

  • Curated by eLife

    eLife logo

    eLife Assessment:

    The authors perform a Transposon-Sequencing screen to determine bacterial factors (including receptors) important for infection by two phages in the model bacterium Corynebacterium glutamicum. Using their established high-density transposon library, they identify genes required for infection with the phages Cog and CL31. They also identified a spontaneous phage-resistant mutant that led to the discovery of a gene involved in mycolic acid synthesis. Overall, the work is of broad interest to scientists in the field of cell wall biogenesis, phage infection, and bacterial cell biology.

    (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #1, Reviewer #2 and Reviewer #3 agreed to share their name with the authors.)

Reviewed by

Read the full article See related articles

Listed in

Log in to save this article

Abstract

Bacteria of the order Corynebacteriales including pathogens such as Mycobacterium tuberculosis and Corynebacterium diphtheriae are characterized by their complex, multi-layered envelope. In addition to a peptidoglycan layer, these organisms possess an additional polysaccharide layer made of arabinogalactan and an outer membrane layer composed predominantly of long-chain fatty acids called mycolic acids. This so-called mycolata envelope structure is both a potent barrier against antibiotic entry into cells and a target of several antibacterial therapeutics. A better understanding of the mechanisms underlying mycolata envelope assembly therefore promises to reveal new ways of disrupting this unique structure for the development of antibiotics and antibiotic potentiators. Because they engage with receptors on the cell surface during infection, bacteriophages have long been used as tools to uncover important aspects of host envelope assembly. However, surprisingly little is known about the interactions between Corynebacteriales phages and their hosts. We therefore made use of the phages Cog and CL31 that infect Corynebacterium glutamicum ( Cglu ), a model member of the Corynebacteriales, to discover host factors important for phage infection. A high-density transposon library of Cglu was challenged with these phages followed by transposon sequencing to identify resistance loci. The analysis identified an important role for mycomembrane proteins in phage infection as well as components of the arabinogalactan and mycolic acid synthesis pathways. Importantly, the approach also implicated a new gene ( cgp_0396 ) in the process of arabinogalactan modification and identified a conserved new factor (AhfA, Cpg_0475) required for mycolic acid synthesis in Cglu .

Article activity feed

  1. Version published to 10.7554/elife.79981 on eLife
  2. eLife

    Author Response

    Reviewer #1 (Public Review):

    This paper primarily assessed the host/phage interactions for bacteria in the order of Cornyebacteriales to identify novel host factors necessary for phage infection, in regards to genes responsible for bacterial envelope assembly. Bacteria in this order, such as Mycobacterium tuberculosis and Corynebacterium diphtheriae have unique, complex envelopes composed of peptidoglycan, arabinogalactan, and mycolic acids. This barrier is a potent protector against the therapeutic effects of antibiotics. Phages can be used to discover novel aspects of this bacterial envelope assembly because they engage with cell surface receptors. To uncover new factors, the researchers challenged a high-density transposon library of Corynebacterium glutamicum (called Cglu in the paper) with phages, Cog, and CL31. Results by transposon sequencing identified loci that were interrupted, leading to phage resistance. This study implicated the importance of Cglu genes, ppgS, cgp_0658, cgp_0391, and cgp_0393. They also identified a new gene called cgp_0396 necessary for arabinogalactan modification and recognized a conserved host factor called Ahfa (Cpg_0475) that plays a crucial role in Cglu mycolic acid synthesis. Ultimately, this work implicated the importance of mycomembrane porins, arabinogalactan, and mycolic acid synthesis pathways in the assembly of the Cornyebacteriales envelope.

    Strengths of the research:

    • Language choice: A major strength of the paper is that this could easily be given to an undergraduate student with introductory knowledge of biology and they would still be able to get the gist of this paper. The language is written in a clear, concise fashion with explanations of terms not everyone would immediately know unless they worked in the field specifically.
    • These figures are generally explained in a direct manner, clearly stating the major conclusions the reader should get after carefully analyzing the presented data

    We thank the reviewer for the enthusiasm for our work and our description of it.

    How the research could be strengthened:

    • It could be worthwhile to describe some of your results mathematically. For example, the differences you see in your phage infections relating to the differences in logs, etc. Bar graphs also should be described in mathematical terms, when "something is lower compared to the WT," how much is lower, etc?

    To keep the text streamlined, we refrained from adding descriptions of the results mathematically in the text. The reader can refer to the figures to get the magnitudes of any changes observed.

    • There were no p values relating to the statistical significance of any of the data presented, which should be changed for the final manuscript implicating the importance of this work.

    We added the p-values as requested.

    • Figure 8 was not entirely supported by the data, especially Figure 8A which either could be improved with better images that support the author's claims, etc.

    We do not understand why the reviewer believes that Figure 8A does not support our conclusions. The mutant cells do not label with the 6-TMR-Tre dye whereas the WT control does. The dye labels mycolic acid such that our conclusion that AhfA is involved in mycolic acid synthesis is valid. In any case, we have included an additional supplementary source data file of the uncropped image of the 6-TMR-Tre treated cells to show a larger number of mutant cells that fail to stain, further supporting our conclusion.

    Reviewer #2 (Public Review):

    In this manuscript, McKitterick and Bernhardt use genetic approaches to investigate genes in Corynebacterium glutamicum that are required for efficient phage infection. They make use of a high-density transposon library that was generated in the Bernhardt lab recently. They challenged the library with two phages, CL31 and Cog. Importantly, they elegantly adapted the phages to the laboratory strain MB001 before. The MB001 strain is ideal for genetic experiments since all prophage elements were removed in this strain. The evolved phages are likely a very useful tool for further investigations aiming to understand host/virus interactions in this model. The phage-infected libraries were plated and the collected colonies were sequenced. Genes involved in efficient phage infection had multiple transposon insertions. Using this method the authors identified specific genes required for infection with Cog and CL31. The Cog phage needs apparently the porin proteins in the mycolic acid membrane for efficient infection and the authors speculate that the porins may act as auxiliary receptors for phage adsorption. Furthermore, genes involved in putative arabinogalactan modification were found to be important. Mutants in these genes did not abolish phage adsorption and thus play a role in viral genome injection. For phage CL31 the authors show that in particular genes involved in mycolic acid synthesis are essential. The genes identified include one coding for a protein involved in protein mycoloylation. A candidate for such a lipidation is the porin protein complex PorAH. The trehalose-6-phosphate synthase OtsA was also identified as important for phage infection. Also strictly required for the establishment of the myco membrane, otsA deletions are viable in C. glutamicum. As part of their analysis, they also identified an unknown factor in mycolic acid synthesis in C. glutamicum. Analysis of a spontaneous resistant mutant to CL31 revealed a mutation in cg_0475 (renamed ahfA). Deletion of ahfA drastically reduced mycolic acid production. This was proven by thin layer chromatography and fluorescent staining. Interestingly, deletion of ahfA also results in a cell morphology defect, indicating the importance of a correct mycolic acid layer for cell shape.

    In summary, the authors provide an excellent paper that is clearly written and experiments are conducted nicely.

    We thank the reviewer for their kind words and enthusiasm for the work.

    Reviewer #3 (Public Review):

    In their manuscript, McKitterick and Bernhardt perform a screen to determine host factors, such as receptors, which are important for bacterial viruses (phages) to infect Corynebacterium glutamicum., an organism that shares the unique membrane of mycobacteria (mycomembrane), with M. tuberculosis. To do so, they challenge a previously described Tn-seq library with a high MOI of 2 phages - Cgl and Cog. The surviving strains are those in which genes important for phage infection (such as receptors) are disrupted. The authors' screen is successful, and the authors identify and validate several factors important for the infection of each phage, providing the first such screen in Corynebacterium. Moreover, the authors perform a suppressor screen to identify additional factors and experimentally follow up several genes of interest. Finally, the authors use the newly determined host specificity of te phages to implicate new genes in mycolic acid synthesis. As a whole, this is a strong work that paves the way to a deeper understanding of Corynebacterial and (by extension) Mycobacterial phages and should be of broad interest.

    Below, we suggest additional analyses, context, and elaboration that will help the ms. elaboration to fully realize its impact.

    Major points:

    1. Although the authors' experimental design is fundamentally sound, I am worried about the possibility of "jackpotting" in shaping their results, particularly in the uninfected control experiment. If the authors' Tn-seq library is ~200,000 strains, and they don't plate at least 10-100x times that many colonies then any given strain (regardless of its phenotype) may or may not be represented in the output of the experiment, causing false phenotypes to be ascribed to genes based on chance. This is particularly a problem for the uninfected control, where the authors choose to dilute the culture 1000fold to mimic the number of colonies that survive infection. They may be better served by plating the whole culture on the plates, to ensure adequate representation of the library. Part of the reason for this concern is that an overwhelming majority of statistically significant hits (something like 80-90%) appear to confer susceptibility rather than resistance (source data Fig 2) - something the authors' experimental design should not be able to measure. The lack of accurate representation of distributions of strains in the starting culture also calls into question the quantitative differences they present in the results

    We thank the reviewer for their thorough analysis of our experimental design. The Tn-Seq experiments were repeated with the uninfected controls plated at a density that maintains the representation of the original library. The overall results are largely unchanged because we maintain our focus on hits that become greatly enriched following phage infection not those that become depleted. The vast majority of these hits were validated for their involvement by constructing mutant strains, indicating the robustness of the current and previous analyses. With respect to the depletion of insertion mutants, we mentioned in the original submission that they are unlikely to be biologically meaningful.

    a. L138. Where the authors describe their initial experimental design it would be helpful to add more details. What is the size of the Tn library? What is the coverage in their experiment? Approximately how many colonies are recovered on the plates after phage infection and in the uninfected control?

    This information has been added (Fig. 2 table supplement 1).

    b. it is important to know how the number of colonies on the plates compares to the number of reads in the experiment. In the analysis of most HT screens, one implicitly assumes that each read corresponds to 1 cell, hence each read can be treated as statistically independent. This assumption is critical to the statistical methods used to analyze this data. By scraping a plate of colonies (which may be required for efficient phage infection), the authors potentially violate this assumption (since the number of cells → number of colonies, which are the actual statistically independent entities in the experiment). Does this assumption hold (or approximately hold) for the screen? If not, a different statistical method should be used to determine p-values.

    We respectfully disagree with the reviewer on this point. In our view, a slurry of colonies from a plate is no different than a culture. Both contain a mixture of cells containing an array of different transposon mutants each represented multiple times in the population due to replication of the original mutant. We do not think there is any meaningful difference to the analysis whether this replication occurs in liquid or on a plate. In both cases, a read corresponds to a single cell/molecule of purified genomic DNA from the population.

    1. The authors' Tn-seq methodology is different from previously published HT-phage screens (e.g. Mutalik et al., 2020 and Rousset et al., 2018). Based on my knowledge of classical phage biology, I agree that plating the infected cells has advantages. However, the rationale will not be clear for most people performing such experiments. Please explain the rationale for the experimental protocol.

    Although the authors in the Mutalik et al paper did do competition experiments in liquid over several infection cycles, they also made use of a solid platebased assay in which they adsorbed their phages to the library cells for 15 minutes before plating. These plates were incubated overnight and resistant colonies were scraped, pelleted, and DNA prepped in a similar manner to the approach we took.

    We prefer plating over liquid growth because colony formation is an easy way to ensure that the mutant population has undergone numerous rounds of doubling under a given condition before the analysis is performed.

    a. Why did the authors plate the cultures after initial phage absorption instead of remaining in liquid?

    We were concerned that some potential receptor-related mutants would be less fit and would therefore be lost in a competition experiment. As such, plating after phage adsorption would decrease the competition between phage survivors. Furthermore, we thought that plating would additionally ensure that the bacteria that are sequenced are true survivors and not just reflect remnant DNA in the culture.

    b. How reproducible are the authors' Tn-seq results? The SRA ascension shows multiple replicates but this is not described in the manuscript nor reflected in the supplementary data. Given the potential for bottleneck and jackpotting effects in this assay, some measure of reproducibility is important for interpreting the results (see point 1).

    We performed completely new Tn-seq experiments for each phage in duplicate. The hit lists remained largely unchanged from our initial analysis and those that were investigated further were enriched for insertions in both new data sets. Thus, the results are highly reproducible.

    c. L587 "Significant hits with fewer than 10 insertions on each strand were manually removed." Why did the authors choose this criterion? Almost all of the genes they removed have very asymmetric distributions (e.g. in the Cog experiment, cgp3051 has 47853 fwd reads and 6 rev reads. Asymmetric distribution of insertions suggests that overexpression of downstream genes has an important (positive or negative) effect. This is a worthwhile pursuit, and many automated analysis pipelines can disambiguate these effects, including those developed in the Walker Lab (e.g. doi: 10.1038/s41589018-0041-4). These genes shouldn't be thrown away when they are arguably some of the most informative hits!

    We have updated the criteria we used for selecting the most impactful insertion enrichments. Our concern in this report was to investigate mutants that affect phage infection when inactivated. We will pursue genes that affect phage infection when overexpressed (as indicated by asymmetric insertion orientation distributions) in a follow-on study. We think such a study would best be carried out with a different transposon containing a strong outward facing promoter.

    1. There is a somewhat extensive phylogeny of M. smegmatis phages (phagesdb.org). Are the phages that the authors work on related to any of these phages? If so, what cluster do they map to? What is the host range of other phages in that cluster? If not, may be worthwhile to mention that these are quite distinct from other studied phages.

    We agree that the phylogenetic history of corynephages is quite interesting. Very few phages that infect Cglu have been isolated and sequenced, let alone studied. Neither Cog nor CL31 share significant nucleotide identity with other sequenced phages, thus they do not have assigned clusters at the moment.

    1. Given that cgp_0475 was a strong hit in the Tn-seq, why was it not identified in the previous chemical genomics experiments from the lab (https://doi.org/10.7554/eLife.54761) ?

    We appreciate the reviewer’s interest in previous work from the lab. In the prior phenotypic analysis, cgp_0475 was identified as having severe fitness defects across many conditions. However, it was not possible to correlate its phenotype with other genes involved in mycolic acid synthesis like pks and fadD2 because they were found to be so sick in the phenotypic outgrowth that they were classified as essential.

    1. Is there any relationship between the growth-rate of the mutants and their phage susceptibility? This can be analyzed using the authors' previous studies of this library.

    While some of the phage resistant mutants are associated with poor fitness (namely those involved in mycolic acid synthesis), not all were associated with decreased growth. For example, there were minimal fitness defects associated with deletions of either porAH or the genes involved GalN decoration. However, loss of these genes greatly inhibited the ability of Cog to infect.

  3. eLife

    eLife Assessment:

    The authors perform a Transposon-Sequencing screen to determine bacterial factors (including receptors) important for infection by two phages in the model bacterium Corynebacterium glutamicum. Using their established high-density transposon library, they identify genes required for infection with the phages Cog and CL31. They also identified a spontaneous phage-resistant mutant that led to the discovery of a gene involved in mycolic acid synthesis. Overall, the work is of broad interest to scientists in the field of cell wall biogenesis, phage infection, and bacterial cell biology.

    (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #1, Reviewer #2 and Reviewer #3 agreed to share their name with the authors.)

  4. eLife

    Reviewer #1 (Public Review):

    This paper primarily assessed the host/phage interactions for bacteria in the order of Cornyebacteriales to identify novel host factors necessary for phage infection, in regards to genes responsible for bacterial envelope assembly. Bacteria in this order, such as Mycobacterium tuberculosis and Corynebacterium diphtheriae have unique, complex envelopes composed of peptidoglycan, arabinogalactan, and mycolic acids. This barrier is a potent protector against the therapeutic effects of antibiotics. Phages can be used to discover novel aspects of this bacterial envelope assembly because they engage with cell surface receptors. To uncover new factors, the researchers challenged a high-density transposon library of Corynebacterium glutamicum (called Cglu in the paper) with phages, Cog, and CL31. Results by transposon sequencing identified loci that were interrupted, leading to phage resistance. This study implicated the importance of Cglu genes, ppgS, cgp_0658, cgp_0391, and cgp_0393. They also identified a new gene called cgp_0396 necessary for arabinogalactan modification and recognized a conserved host factor called Ahfa (Cpg_0475) that plays a crucial role in Cglu mycolic acid synthesis. Ultimately, this work implicated the importance of mycomembrane porins, arabinogalactan, and mycolic acid synthesis pathways in the assembly of the Cornyebacteriales envelope.

    Strengths of the research:
    - Language choice: A major strength of the paper is that this could easily be given to an undergraduate student with introductory knowledge of biology and they would still be able to get the gist of this paper. The language is written in a clear, concise fashion with explanations of terms not everyone would immediately know unless they worked in the field specifically.
    - These figures are generally explained in a direct manner, clearly stating the major conclusions the reader should get after carefully analyzing the presented data

    How the research could be strengthened:
    - It could be worthwhile to describe some of your results mathematically. For example, the differences you see in your phage infections relating to the differences in logs, etc. Bar graphs also should be described in mathematical terms, when "something is lower compared to the WT," how much is lower, etc?
    - There were no p values relating to the statistical significance of any of the data presented, which should be changed for the final manuscript implicating the importance of this work.
    - Figure 8 was not entirely supported by the data, especially Figure 8A which either could be improved with better images that support the author's claims, etc.

  5. eLife

    Reviewer #2 (Public Review):

    In this manuscript, McKitterick and Bernhardt use genetic approaches to investigate genes in Corynebacterium glutamicum that are required for efficient phage infection. They make use of a high-density transposon library that was generated in the Bernhardt lab recently. They challenged the library with two phages, CL31 and Cog. Importantly, they elegantly adapted the phages to the laboratory strain MB001 before. The MB001 strain is ideal for genetic experiments since all prophage elements were removed in this strain. The evolved phages are likely a very useful tool for further investigations aiming to understand host/virus interactions in this model. The phage-infected libraries were plated and the collected colonies were sequenced. Genes involved in efficient phage infection had multiple transposon insertions. Using this method the authors identified specific genes required for infection with Cog and CL31. The Cog phage needs apparently the porin proteins in the mycolic acid membrane for efficient infection and the authors speculate that the porins may act as auxiliary receptors for phage adsorption. Furthermore, genes involved in putative arabinogalactan modification were found to be important. Mutants in these genes did not abolish phage adsorption and thus play a role in viral genome injection. For phage CL31 the authors show that in particular genes involved in mycolic acid synthesis are essential. The genes identified include one coding for a protein involved in protein mycoloylation. A candidate for such a lipidation is the porin protein complex PorAH. The trehalose-6-phosphate synthase OtsA was also identified as important for phage infection. Also strictly required for the establishment of the myco membrane, otsA deletions are viable in C. glutamicum. As part of their analysis, they also identified an unknown factor in mycolic acid synthesis in C. glutamicum. Analysis of a spontaneous resistant mutant to CL31 revealed a mutation in cg_0475 (renamed ahfA). Deletion of ahfA drastically reduced mycolic acid production. This was proven by thin layer chromatography and fluorescent staining. Interestingly, deletion of ahfA also results in a cell morphology defect, indicating the importance of a correct mycolic acid layer for cell shape.

    In summary, the authors provide an excellent paper that is clearly written and experiments are conducted nicely.

  6. eLife

    Reviewer #3 (Public Review):

    In their manuscript, McKitterick and Bernhardt perform a screen to determine host factors, such as receptors, which are important for bacterial viruses (phages) to infect Corynebacterium glutamicum., an organism that shares the unique membrane of mycobacteria (mycomembrane), with M. tuberculosis. To do so, they challenge a previously described Tn-seq library with a high MOI of 2 phages - Cgl and Cog. The surviving strains are those in which genes important for phage infection (such as receptors) are disrupted. The authors' screen is successful, and the authors identify and validate several factors important for the infection of each phage, providing the first such screen in Corynebacterium. Moreover, the authors perform a suppressor screen to identify additional factors and experimentally follow up several genes of interest. Finally, the authors use the newly determined host specificity of te phages to implicate new genes in mycolic acid synthesis. As a whole, this is a strong work that paves the way to a deeper understanding of Corynebacterial and (by extension) Mycobacterial phages and should be of broad interest.

    Below, we suggest additional analyses, context, and elaboration that will help the ms. elaboration to fully realize its impact.

    Major points:

    1. Although the authors' experimental design is fundamentally sound, I am worried about the possibility of "jackpotting" in shaping their results, particularly in the uninfected control experiment. If the authors' Tn-seq library is ~200,000 strains, and they don't plate at least 10-100x times that many colonies then any given strain (regardless of its phenotype) may or may not be represented in the output of the experiment, causing false phenotypes to be ascribed to genes based on chance. This is particularly a problem for the uninfected control, where the authors choose to dilute the culture 1000-fold to mimic the number of colonies that survive infection. They may be better served by plating the whole culture on the plates, to ensure adequate representation of the library. Part of the reason for this concern is that an overwhelming majority of statistically significant hits (something like 80-90%) appear to confer susceptibility rather than resistance (source data Fig 2) - something the authors' experimental design should not be able to measure. The lack of accurate representation of distributions of strains in the starting culture also calls into question the quantitative differences they present in the results

    a. L138. Where the authors describe their initial experimental design it would be helpful to add more details. What is the size of the Tn library? What is the coverage in their experiment? Approximately how many colonies are recovered on the plates after phage infection and in the uninfected control?

    b. it is important to know how the number of colonies on the plates compares to the number of reads in the experiment. In the analysis of most HT screens, one implicitly assumes that each read corresponds to 1 cell, hence each read can be treated as statistically independent. This assumption is critical to the statistical methods used to analyze this data. By scraping a plate of colonies (which may be required for efficient phage infection), the authors potentially violate this assumption (since the number of cells → number of colonies, which are the actual statistically independent entities in the experiment). Does this assumption hold (or approximately hold) for the screen? If not, a different statistical method should be used to determine p-values.

    2. The authors' Tn-seq methodology is different from previously published HT-phage screens (e.g. Mutalik et al., 2020 and Rousset et al., 2018). Based on my knowledge of classical phage biology, I agree that plating the infected cells has advantages. However, the rationale will not be clear for most people performing such experiments. Please explain the rationale for the experimental protocol.

    a. Why did the authors plate the cultures after initial phage absorption instead of remaining in liquid?

    b. How reproducible are the authors' Tn-seq results? The SRA ascension shows multiple replicates but this is not described in the manuscript nor reflected in the supplementary data. Given the potential for bottleneck and jackpotting effects in this assay, some measure of reproducibility is important for interpreting the results (see point 1).

    c. L587 "Significant hits with fewer than 10 insertions on each strand were manually removed." Why did the authors choose this criterion? Almost all of the genes they removed have very asymmetric distributions (e.g. in the Cog experiment, cgp3051 has 47853 fwd reads and 6 rev reads. Asymmetric distribution of insertions suggests that overexpression of downstream genes has an important (positive or negative) effect. This is a worthwhile pursuit, and many automated analysis pipelines can disambiguate these effects, including those developed in the Walker Lab (e.g. doi: 10.1038/s41589-018-0041-4). These genes shouldn't be thrown away when they are arguably some of the most informative hits!

    3. There is a somewhat extensive phylogeny of M. smegmatis phages (phagesdb.org). Are the phages that the authors work on related to any of these phages? If so, what cluster do they map to? What is the host range of other phages in that cluster? If not, may be worthwhile to mention that these are quite distinct from other studied phages.

    4. Given that cgp_0475 was a strong hit in the Tn-seq, why was it not identified in the previous chemical genomics experiments from the lab (https://doi.org/10.7554/eLife.54761) ?

    5. Is there any relationship between the growth-rate of the mutants and their phage susceptibility? This can be analyzed using the authors' previous studies of this library.

  7. Version published to 10.1101/2022.05.16.492151v1 on bioRxiv