Transposon mutagenesis in Mycobacterium abscessus identifies an essential penicillin-binding protein involved in septal peptidoglycan synthesis and antibiotic sensitivity

Decision letter after peer review:

Thank you for submitting your article "High-density transposon mutagenesis in Mycobacterium abscessus identifies an essential penicillin-binding lipo-protein (PBP-lipo) involved in septal peptidoglycan synthesis and antibiotic sensitivity" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, and the evaluation has been overseen by Bavesh Kana as the Reviewing and Senior Editor. The reviewers have opted to remain anonymous.

The reviewers have discussed their reviews with one another, and the Reviewing Editor has drafted this to help you prepare a revised submission.

Essential revisions:

1. There is no evidence that PBP-lipo is a lipoprotein although the authors presented this as a certainty (lines 143-144). Several PBPs have a cysteine residue within or near the end of the N-terminal TM region, but these are not lipoproteins. For example, E. coli PBP3 has a cysteine at position 30 shortly after the TM region, but it does not appear to be a lipoprotein. The authors should either prove that PBP-lipo is a lipoprotein, which has been achieved for lipoproteins in other species by mass spectrometry and/or labelling with radioactive palmitate. If such evidence cannot be provided, they should rename the enzyme, removing 'lipo' from the name, and refrain from presenting the protein as a lipoprotein.

2. There appears to be a problem with the mRFP-PBP-lipo construct in which mRFP is tagged at the N-terminus of PBP-lipo (line 210). If PBP-lipo is indeed a lipoprotein, as assumed by the authors and perhaps confirmed in the above-mentioned point, then this N-terminal tagging may not be the best approach. Lipoproteins are typically cleaved at the N-terminus of the cysteine residue, which would remove the mRFP tag from the protein and make the cellular localization experiments meaningless. Considering these concerns, the authors should:

a. Prove that the construct is functional and can complement cells depleted of WT PBP-lipo, and

b. Subject cell extracts with Bocillin labelling and confirm the presence of an mRFP-PBP-lipo band or

c. Detect mRFP-PBP-lipo in cell extracts with an antibody against mRFP.

(A is essential and either b or c would be acceptable)

Also, in line 211 they should state whether mRFP-PBP-lipo was expressed in the presence (or not) of WT PBP-lipo.

3. Reviewers appreciated the experiments to investigate the PG networks in Mab and other mycobacteria which revealed, for example, the genetic interactions between the PBP-lipo gene and the dacB1, pbpB and MAB_0519 genes. However, to support the appending conclusions, the authors need to confirm that DacB1 and PBPB were depleted by the CRISPR technology to similar extent in Mab and Msm, to exclude the possibility that differences in genetic interactions were due to different efficiency in the CRISPR gene depletion in the two species. This is particularly important for Msm, as there were no genetic interactions (lines 270-2760). Hence, they should perform Bocillin assays and quantify the depletion of DacB1 and PBPB in both species.

4. Control experiments to show that the DacB1-GFPmut3 fusion is functional should be added.

5. To strengthen their model, the authors should co-localize PBPlipo with another divisome component.

6. The authors should Colocalise mRFP-PBP-lipo (if functional) with DacB1-GFP (if functional).

7. It is known that several bacteria have homologous PBPs that cannot fully replace each other and have specific roles under different growth conditions. The authors don't present or discuss such examples in their Discussion. The discussion should be expanded beyond mycobacteria to include examples from other species, such as:

– Salmonella has a dedicated class B PBP3 homologue (called PBP3-SAL) next to its 'normal' PBP3 required for cell division. PBP3-SAL is specifically required for cell division in acidified phagosomes (mBio. 2017 Dec 19;8(6):e01685-17. doi: 10.1128/mBio.01685-17).

– E. coli uses a specialised DD-CPase PBP6B to maintain cell morphology when growing under acidic conditions (mBio. 2016 Jun 21;7(3):e00819-16. doi: 10.1128/mBio.00819-16.).

Such examples can use used to discuss whether PBP-lipo and its homologue might not be essential under certain growth conditions at which the homologue is highly active.

Reviewer #1 (Recommendations for the authors):

This manuscript was a pleasure to read and easy to evaluate. It is interesting and exceptionally well done.

Two questions that occurred to me are: (1) is the localisation of PBP-lipo affected in the presence of AMX or AMP, and( 2) is it possible to find a β-lactam that would sensitize WT Mab to AMX or AMP to the same degree as the PBP-lipo knock-down Mab strains.

The title refers to PBP-lipo as a lipoprotein, but no such evidence that it is indeed a lipoprotein has been provided.

Reviewer #2 (Recommendations for the authors):

I would like to ask the authors to address the following points:

1. There is no evidence that PBP-lipo is a lipoprotein although they presented this as a certainty (lines 143-144). To my knowledge, several PBPs have a cysteine residue within or near the end of the N-terminal TM region, but these are not lipoproteins. For example, E. coli PBP3 has a cysteine at position 30 shortly after the TM region, but it does not appear to be a lipoprotein. They should prove that PBP-lipo is a lipoprotein, which has been achieved for lipoproteins in other species by mass spectrometry and/or labelling with radioactive palmitate. If they cannot provide such evidence, they should rename the enzyme, removing 'lipo' from the name, and refrain from presenting the PBP as a lipoprotein.

2. It is known that several bacteria have homologous PBPs that cannot fully replace each other and have specific roles under different growth conditions. Unfortunately, the authors don't present or discuss such examples in their Discussion. I suggest that they expand their discussion beyond mycobacteria to include examples from other species, for example:

– Salmonella has a dedicated class B PBP3 homologue (called PBP3-SAL) next to its 'normal' PBP3 required for cell division. PBP3-SAL is specifically required for cell division in acidified phagosomes (mBio. 2017 Dec 19;8(6):e01685-17. doi: 10.1128/mBio.01685-17).

– E. coli uses a specialised DD-CPase PBP6B to maintain cell morphology when growing under acidic conditions (mBio. 2016 Jun 21;7(3):e00819-16. doi: 10.1128/mBio.00819-16.).

They could include such examples and discuss whether PBP-lipo and its homologue might not be essential under certain growth conditions at which the homologue is highly active? In my view they cannot exclude such a possibility, as they did not assay growth phenotypes at different growth conditions.

3. I see a main problem with the mRFP-PBP-lipo construct in which mRFP is tagged to the N-terminus of PBP-lipo (line 210). If PBP-lipo is indeed a lipoprotein, as assumed by the authors and perhaps confirmed in additional experiments (my point 1 above), then this N-terminal tagging cannot work. Lipoproteins are typically cleaved at the N-terminus of the cysteine residue, which would remove the mRFP tag from the protein and make the cellular localization experiments meaningless. Because of these doubts regarding the mRFP-PBP-lipo construct they should:

– prove that the construct is functional and can complement cells depleted of WT PBP-lipo,

– subject cell extracts with Bocillin labelling and confirm the presence of an mRFP-PBP-lipo band and

– detect mRFP-PBP-lipo in cell extracts with an antibody against mRFP.

Also, in line 211 they should state whether mRFP-PBP-lipo was expressed in the presence (or not) of WT PBP-lipo.

4. I appreciate the experiments to investigate the PG networks in Mab and other mycobacteria which revealed, for example, the genetic interactions between the PBP-lipo gene and the dacB1, pbpB and MAB_0519 genes. However, in my view they should confirm that DacB1 and PBPB were depleted by the CRISPR technology to similar extent in Mab and Msm, to exclude the possibility that differences in genetic interactions were due to different efficiency in the CRISPR gene depletion in the two species. This is particularly important for Msm, as there were no genetic interactions (lines 270-2760). Hence, they should perform Bocillin assays and quantify the depletion of DacB1 and PBPB in both species.

5. Figure 10B. They should add control experiments to show that the DacB1-GFPmut3 fusion is functional (Bocillin labelling of extracts; Western Blot with anti-Gfp).

Reviewer #3 (Recommendations for the authors):

The experiments largely support the conclusion derived. However, I feel the following experiments/clarifications will further strengthen some of the conclusions in the manuscript:

(i) Colocalization of PBP-lipo with at least one of the divisome components may strengthen the argument that PBP-lipo localizes to the division septa.

(ii) Similarly the idea that PBP-lipo interacts with PbpB and/or DacB1 can be further strengthened by at least colocalising mRFP-PBP-lipo with DacB1-GFP.

(iii) Is the localization of PBP-lipo affected in PbpB or DacB1 or MAB_0519 depleted cells? Are the localization of these proteins interdependent?

(iv) Does depletion of pbpB or dacB1 or MAB_0519 make Mab cells susceptible to antibiotics as during PBP-lipo knockdown?

[Editors’ note: further revisions were suggested prior to acceptance, as described below.]

Thank you for resubmitting your work entitled "High-density transposon mutagenesis in Mycobacterium abscessus identifies an essential penicillin-binding lipo-protein (PBP-lipo) involved in septal peptidoglycan synthesis and antibiotic sensitivity" for further consideration by eLife. Your revised article has been evaluated by Bavesh Kana (Senior Editor) and a Reviewing Editor.

The manuscript has been improved but there is one major concern that needs to be addressed, as outlined below.

Reviewers remain concerns about the experiments supporting the annotation of PBP-Lipo as a lipoprotein:

1. Please provide sufficient detail on the lipoprotein extraction procedure. The information is not available in Ref 26, which also appears to focus on Vibrio, rather than Mycobacteria.

2. Results of the attempts to verify that PBP-lipo has a lipid modification are shown in Figure 3 —figure supplement 3B. However, this figure shows that the majority of PBP-lipo (>90%?) seems to segregate into the non-LP fraction. Quantification is not possible because the non-LP lane is overloaded but the results suggest the absence of a lipoprotein modification. The Methods section lacks sufficient information about this experiment (how were the LP and non-LP fractions obtained?), and there is no positive or negative control of a verified lipoprotein or non-lipoprotein. As a result, reviewers indicate that this experiment does not address the issue of the Lipo modification. You should either rename the enzyme, removing "lipo" from the name, or state very clearly in the manuscript that the lipo modification is hypothetical.

Other concerns:

L. 273: insert "genetic" in front of "PG network" in the heading.

L. 309: format problem.

Figure 6 —figure supplement 5A. the y-axis should read "log (or ln?) fold change mRNA expression" instead of "fold change mRNA expression".

L507: PbpB is 4-3 crosslinking enzyme. It does not necessarily follow that PBP-lipo must have the same activity. The ectopic complementation could be due to other crosslinks that also stabilize PG. Please reconsider this statement.