A conserved microtubule-binding region in Xanthomonas XopL is indispensable for induced plant cell death reactions

Pathogenic Xanthomonas bacteria cause disease on more than 400 plant species. These Gram-negative bacteria utilize the type III secretion system to inject type III effector proteins (T3Es) directly into the plant cell cytosol where they can manipulate plant pathways to promote virulence. The host range of a given Xanthomonas species is limited, and T3E repertoires are specialized during interactions with specific plant species. Some effectors, however, are retained across most strains, such as Xanthomonas Outer Protein L (XopL). As an ‘ancestral’ effector, XopL contributes to the virulence of multiple xanthomonads, infecting diverse plant species. XopL homologs harbor a combination of a leucine-rich-repeat (LRR) domain and an XL-box which has E3 ligase activity. Despite similar domain structure there is evidence to suggest that XopL function has diverged, exemplified by the finding that XopLs expressed in plants often display bacterial species-dependent differences in their sub-cellular localization and plant cell death reactions. We found that XopL from X. euvesicatoria (XopLXe) directly associates with plant microtubules (MTs) and causes strong cell death in agroinfection assays in N. benthamiana. Localization of XopLXe homologs from three additional Xanthomonas species, of diverse infection strategy and plant host, revealed that the distantly related X. campestris pv. campestris harbors a XopL (XopLXcc) that fails to localize to MTs and to cause plant cell death. Comparative sequence analyses of MT-binding XopLs and XopLXcc identified a proline-rich-region (PRR)/α-helical region important for MT localization. Functional analyses of XopLXe truncations and amino acid exchanges within the PRR suggest that MT-localized XopL activity is required for plant cell death reactions. This study exemplifies how the study of a T3E within the context of a genus rather than a single species can shed light on how effector localization is linked to biochemical activity.

Reviewer #2: This interesting and relevant study does a deep dive into the molecular mechanisms that underpin T3SS effector function in plants. Previously, this group demonstrated that the E3 ligase, XopL, from Xanthamonas spp. associated with the microtubule (MT) cytoskeleton in the absence of E3 ligase activity and may is involved in chloroplast stromule dynamics. Here, by exploring XopL activities, sequence conservation, and domain architecture across Xanthamonas species, they find that XopLxcc from Xanthamonas campestris campestris is unable to bind associate with microtubules. Through deletion analysis, mutagenesis, and domain swap experiments, they demonstrate that MT binding is conferred by a proline-rich tract and the NT region it is embedded within. Further, the authors show that in planta, microtubules are disrupted by XopLxe and that this perturbation requires E3 ligase activity. Cell death also correlates with the combination of MT-binding and destruction and E3 ligase activity; however, disruption of MTs by overexpression of MAPs is not sufficient to initiate cell death. In a simple pulldown experiment, the authors demonstrate the ability to bind MT in vitro; however, the molecular mechanism of MT destabilization in planta is likely indirect and requires further exploration. I find this body of work to be impressive, experiments carefully conducted and adequately controlled, and the results critically interpreted.
Reviewer #3: The manuscript (PPATHOGENS-D-23-00407)， "A conserved microtubule-binding region in Xanthomonas XopL is indispensable for induced plant cell death reactions."， written by Ortmann et al., described their novel finding that the ancestral effector XopL displays bacterial species dependent differences in their sub-cellular localization and plant cell death reactions. It is very interesting that XopL from X. euvesicatoria (XopLXe) directly associates with plant microtubules (MTs) and causes strong cell death in agroinfection assays in N. benthamiana. This event is not associated with XopLXcc which come from X. campestris pv. campestris that fails to localize to MTs and to cause plant cell death. Thus，the authors further confirmed that a proline-rich-region (PRR)/α-helical region is important for MT localization. The MT-localized XopL activity is required for plant cell death reactions. Best to my knowledge, this is the first report that a T3E within the context of a genus rather than a single species can shed light on how effector localization is linked to biochemical activity. The data are sufficient to support the statement, the manuscript is well organized and written. I would like it accept for publication in this lovely journal.

Major Changes:
Reviewer #1: A main concern (and to generalize the authors findings) is the cell death-inducing activity of XopLXoo and XopLXac. The comparison of XopLXe with these two MT-binding variants could be used to further support the NTalpha region features required for MT-binding (in chimera with Xcc LRR and XL domains) and to clarify the link with cell-death inducing activity of MT-binding XopL variants.
As the reviewer suggested it would provide even more support for the link between microtubule binding and cell death if we could have properly evaluated cell death in XopLXoo and XopLXac inoculations. However, since these strains were not available to us, we had the genes synthesized and codon optimized for N. benthamiana, but this led to poor/unpredictable expression (Fig 4). Prompted by the reviewers we have sought to optimize the expression and/or stability of these proteins by re-cloning these XopLs with GFP tags (in our experience stability of proteins is sometimes higher when they are tagged with GFP) and tried higher Agrobacterium titers to see if we could create a scenario where expression was better and comparable. We also tried to improve expression with a silencing inhibitor p19. Unfortunately, these approaches did not result in comparable expression between XopL proteins, and, likely as a result of variable expression levels, did not show consistent macroscopic phenotypes when compared to the non-codon optimized versions used for all other experiments. Therefore, we are still not comfortable drawing any conclusions about cell death from this data and have left this aspect of the manuscript as it was. The figure below depicts Westerns as well as leaf pictures that demonstrate variable XopL expression and examples of leaf phenotypes.
As we mention in the text of the manuscript, we were reluctant to continue with the codon optimized versions of these genes because of this inconsistency. Despite these results, we would like to point out that we have shown sufficient evidence for our claim that the MT-association and cell death are strongly correlated for XopLXe, a finding that is backed by testing of 10 independent XopL variants and quantified in Figure 6 and 8. We would like to acknowledge a mistake we made on p. 20 line 14 and 15 where we wrote: 'Comparative analyses of the subcellular localization of XopLs after agroinfection revealed that XopL from different Xanthomonas species (Xe, Xac and Xoo) associate with microtubules (MTs) in planta, triggering their breakdown and ultimately inducing plant cell death.' This has now been changed to read: Comparative analyses of the subcellular localization of XopLs after agroinfection revealed that XopL from different Xanthomonas species (Xe, Xac and Xoo) associate with microtubules (MTs) in planta, ultimately triggering their breakdown. In the case of XopLXe we were also able to show that MT-association is strongly correlated with plant cell death.
We would like to note that the domain swapping between XopLXe and XopLXcc was added as complement to the detailed analysis we did for XopLXe where we tested amino acid exchanges and truncations extending in either direction to identify the importance of this the NTαLRR for MT-association, and subsequent consequences to cell death caused by the WT protein. Without this level of analysis for the other XopLs (and indeed without knowing for sure whether the WT versions cause cell death in our conditions) we do not want to rely on domain swapping alone to draw conclusions about XopLXac and XopLXoo.
Reviewer #2: A couple of suggestions for further improvement of these studies: 1) The in vitro microtubule binding studies shown in Fig. 6A should be conducted with a dose series and the data for XopL in the pellet curve fit to estimate a Kd value. Ideally, this would be done with a known MT side-binding protein as a positive control. Finally, binding of the truncated and/or mutated forms of XopL should be examined to demonstrate that the PRR in NT is necessary and sufficient for MT binding.
The in vitro microtubule binding assay was performed to determine if XopL is capable of direct microtubule binding as well as to generally confirm microtubule association in a second context. We find the point raised by the reviewer very interesting and indeed have done a few dose-dependent experiments with XopL to start to investigate the nature of direct binding, which is consistent, but appears to be transient in vitro (see figure below). Meaning that there is an equilibrium between MTbound and unbound XopL protein. However, as you can see and as we mentioned in the manuscript, the low solubility of XopL (as shown by the gray points in the graphs below) presents a significant problem, particularly at higher XopL concentrations.

Figure.
The left panel shows the total amount of XopL protein added to the in vitro MT spin-down assays (x-axis) plotted against the amount retained on the MTs (pink line). The amount of XopL that sediments in the absence of MTs is also plotted (gray). The right panel plots the same data as a percentage of total protein.
While this question is worth exploring, as well as questions related to this, such as how XopL influences microtubule dynamics, and if it binds monomers etc. the focus of this manuscript was primarily on effector-microtubule interactions in planta, rather than a detailed biochemical dissection of tubulin binding properties. We hope that the reviewer will understand that to fully address these questions is beyond the scope of this manuscript.
In our eyes current in vitro data combined with the MAP-like architecture of XopL provides strong evidence for the direct interaction with MT. As the three reviewers have kindly pointed out, we were careful in our interpretation and description of the in planta data. We purposefully employed the term 'microtubule association' whenever possible, in place of 'binding', but we welcome the fact that our data encourage these ideas and see this as confirmation of the relevance of our findings.
2) They have established nice methods to quantify XopL-MT colocalization as well as the effects on microtubule numbers. The former should be applied to the experiments shown in Fig. 7 with the XopL truncations and domain swap constructs to further strengthen the conclusions derived from the observations. We found this a very reasonable request and we have quantified the association of the different truncations with MTs as suggested (Fig 7H). It nicely complements the microscopy images and further supports our claims. This was a very nice addition to the manuscript. Since the domain swap was performed only to check whether the construct associates with MT we believe that the microscopy image was sufficient to demonstrate this. I would also suggest to move the excellent correlation analysis shown in Fig. S8 moved into the main text.
We are very happy to hear that Reviewer 2 liked our analysis and we would have liked to include it in the main text. However, since this data is best suited to figure 6, and this figure is quite full, we found it better to include this analysis in the supplement.

Part III -Minor Issues: Editorial and Data Presentation Modifications
Please use this section for editorial suggestions as well as relatively minor modifications of existing data that would enhance clarity.
Reviewer #1: As a general note about the presentation of the results, I recommend the authors to shorten and merge certain parts of the result sections. For example: the 4 XopL alleles E3 ligase activity; the 4 XopL alleles localization; the 4 XopL alleles induced cell-death… Although confocal images are space consuming, splitting controls and samples in different mains and supplemental figures equally splits the attention of readers.
Our rationale for presenting the results in this way and in this order was to tell our story in a series of logical steps that we chose to take based on where the data took us. By merging some of these data for the purposes of compaction we think that the logic behind some of the steps we took would not be clear. I hope the reviewer will forgive us in this case, since we are very satisfied with how the manuscript reads in this order and got positive feedback on this arrangement from Reviewer 3.
We can see the point that some experiments are split between figures. As mentioned by the reviewer, in some cases microscopy pictures take up a great deal of space. At times it was simply not possible to fit all images related to one experiment into a single figure and simultaneously maintain resolution at a level to allow for visualization of fine cellular features like microtubules. However, in case of Fig S1, we have moved this to the main text (into Fig 2) since there was enough space to display the entire experiment there. In general, we tried to divide experiments only when necessary and keep key findings and controls in the main text. I hope this is enough to satisfy the reviewer.
P6l2: hints -> hint (variability and difference in cell death) done P6l12-22: move to introduction or discussion We would like to leave this where it is because it is necessary to understand our approach (so it should not be left until the discussion) and yet it does not belong in the introduction because it would dilute our message related to the biological question.
We apologize here, the statement made about the cell death was reworded to reflect that the cell death was only correlated with binding for Xe XopL. This was simply overlooked in the first version; we were very open about the fact that our codon optimized XopL constructs did not express well and therefore it was not possible to evaluate cell death with confidence.
A reference from XopLXoo-induced cell death in N. benthamiana (17) is already mentioned in the introduction.  On the Zeiss 780 this is reported as % laser power. This percentage can be a bit misleading however, since the maximum laser power ever used on our microscope is not more than 20-25%. The increase from 10% to 16.8% needed to visualize XopL in the low expressing cells is therefore quite substantial. Gain was kept the same for all images.   A good suggestion, and honestly, this was our original intention, but the TUA6-GFP blots had a lot of background signals, and we were therefore not confident that the TUA signal was discernable. Combined with the fact that we also wanted to blot for β-tubulin, we then switched to using wild-type plants and blotting for native tubulin proteins. Figure 4 A and B + legend: the boxes are outlined in black not in white (l6) and difficult to make out. In our PDF this was not the case, and perhaps this happens during the upload. We will try to make sure this is not the case in the final version.  Yes, there was no loading control included. The signal of the CTD is around the size of rubisco and this signal was still visible on the membrane after development and was apparent in the amido black stain. For this reason, we chose not include the loading control. Since we did not want to make any comparisons of protein level between the different samples and only used the western to show the proteins are present and at the right size we thought this was appropriate. However, we are fine with including it at the reviewers request and have now done so.  Figure S6 and S11: change Xcv to Xe for the PRR box We would like to thank the reviewer for catching this. It has now been corrected. Figure S10 I: add y-axis title (Cell death or mean gray value) We would like to thank the reviewer for catching this. It has now been corrected.
Reviewer #2: N/A Reviewer #3: There are no minor issues for revision.