Point by point reply to the concerns of the reviewer (s)
Comments: Reviewer #1 (Remarks for the Author):

Comment: “….there is no methods of determining the directionality of the protein insertion into the liposome. If this is the case, then it does not seem possible to definitively determine the orientation of the proteins within the membrane using this method.”
Our Response:
We agree with the reviewer that there is no definitive method of determining the directionality of the protein insertion within the liposome. We, therefore, standardized the method for protein reconstitution within the liposomes using different combinations of phospholipids to get the optimum transport in the conditions described in the manuscript. It is reasonable to argue that when both ATP and carboxymycobactin (cMyco) are encapsulated within the liposome (LP-48), transport is only possible if both SBD (that binds siderophores) and ATPase are located within the liposome. These results are indeed consistent with a well documented mechanism of reconstitution in which the hydrophilic regions of ABC transporters prevent a random insertion of the protein into the membrane (KM Standifer et al., 1993) and thus support the basis of our observations. We have, nonetheless downplayed the statement about the directionality of the protein insertion in the liposome in the first part of our result section.

Comment: “Two protected domains are apparent at 26 and 29kDa, but if there are two populations of protein in the liposome, some SBD and some ATPase domain would be available for cleavage, even if they were on opposite sides of the membrane.”
Our response:
We appreciate the reviewers’ concern but, as is evident from the liposomal transport assay (fig 7B), both the domains are towards the same face of the membrane, for the transport to proceed, when both ATP and carboxymycobactin are encapsulated within the liposome.

Comment: “Also, why does the remaining full length protein visible in lane 2 not show up on the western blot?”
Our response:
We thank the reviewer to drawing our attention to this. The experiment reported was carried out with the low antibody titre to avoid the back ground noise which probably masked the band. However, when the western blotting experiment was repeated with a relatively higher antibody titre, the presence of full length protein in lane 2 is indeed visible. The corresponding blot is now represented in the revised manuscript as fig 1E.

Comment: “In the RT-PCR showing induction of genes by iron, the quantitation does not appear correct. The low iron bands are much more intense than the iron replete, yet are listed in the histogram as '8%' increased: what is this compared to? If time=0 under low iron conditions gave a similar band intensity to iron replete conditions, then the induction looks to be several fold i.e. >>100%.”
Our response:
To address the reviewer’s concern, we have reanalyzed the intensity of the bands (corresponding to the transcription of genes) using two softwares, Image Quant (used earlier) and ImageJ (available online at NIH website). In the revised Fig.4, the intensity of the RT-PCR bands, quantified additionally using Image J, the fold increase in the expression of the genes under low iron conditions is evident when compared with normal. Increase is easier to comprehend in terms of fold increase instead of %age increase and these have been incorporated in the revised version.


Comment: “Also, if the experiments have been performed in triplicate, some estimate of errors should be possible. This is most important for Rv2895c, where the difference in calculated Kd values is quite modest and the binding curves look very similar.”
Our response:
We have incorporated the error estimates in the revised version. Fig 5A and B, now, depict a mean of at least three readings with error analysis instead of one representative experiment out of three as previously mentioned. The close similarity in the Kd values have been explained in the revised version.

Comment: “The comment about 'biphasic binding curves' is also unclear: the red line in Fig 5A does not appear to be biphasic.”
Our response:
The figure ‘5A’ has been redrawn taking the mean of the three readings at each data point that make the biphasic kinetics more pronounced and easy to comprehend.

Comment: “The plotting of phosphate generation vs fluorescence is a non-intuitive way of presenting this time-dependent data. It would be better to plot both of these measurements vs time. Also, it would be very helpful to show the control data that has been described but not shown, i.e. the lack of activity without intra lysosomal ATP, the inability of Rv1349 to import Fe-cMyco without Rv2895c etc.”
Our response:
As per the reviewer’s suggestion, we have replotted the graphs separately representing the ATPase activity and export or import measured by fluorescence as a function of time. The controls, not shown previously, have also been added in the current graphs.

Comment: “ Is msmeg_6554 the orthologue of Rv1348? Are there any other candidates in the M. smegmatis genome?”
Our response:
Yes, on the basis of similarity searches done by us using M. smegmatis genome, msmeg_6554 was identified as an orthologue of Rv1348. No other candidate in M. smegmatis shares the significant homology to qualify as an orthologue of Rv1348.


Reviewer #2:

Comment: Three areas for improvement include:

Comment: “Probably the greatest weakness of the paper is its failure to adequately address J. Bact. 2006, 188, 424-430 by Rodriguez and Smith (reference 11). That paper identified Rv1348 and Rv1349 as components of an ABC transporter (i.e., carboxymycobactin importer), and suggested that they formed a heterodimer. This putative transporter was in turn compared to YbtP and YbtQ from Yersinia pestis (Mol. Microbiol. 1999, 32, 289-299). Because the authors appear to contradict the published account of Rodriguez and Smith, it is critical that they summarize the principal results from this paper and give more detailed commentary on how their own data add to the emerging model.”
Our response:
In the present paper, we have demonstrated an interplay among three iron regulated Mycobacterium tuberculosis (M. tb) proteins, namely, Rv1348 (IrtA), Rv1349 (IrtB) and Rv2895c in export and import of Mtb siderophores across the membrane and consequent iron uptake. The J. Bact. paper (Rodigruez and Smith, 2006) identified Rv1348 and Rv1349 as components of an ABC transporter while our manuscript elucidates the mechanistic details of the two ABC transporters in addition to the identification of a previously unknown importer component, Rv2895c. As advised by the referee, we have now adequately discussed our results in the background of the J. Bact. paper (Plz. see revised manuscript pages 20 and 21).

Comment: “The constants measured for binding of ferrated/deferrated siderophores to the two siderophore binding proteins are fairly similar. It may be an overstatement to say that Rv1349 has true specificity for ferrated siderophores, and Rv2895c for deferrated siderophores.”
Our response:
The Kd values of SBD highlight a critical two fold difference between the affinity towards non ferrated (cMyco) and ferrated (Fe-cMyco) siderophores which appears to be reasonably significant for SBD to preferentially bind to only non ferrated siderophores.
In addition, the inability of LP-48 to export Fe-cMyco (fig. 7A) suggests a possible mechanism whereby only the binding of cMyco is required for exporter function of IrtA, probably due to the need for subsequent structural changes necessary to facilitate the export. Nonetheless, it did not escape our notice that the intracellular concentration of non ferrated siderophores is always more than the ferrated ones which upon entry inside the cell, is tagged for degradation upon removal of iron. This warrants detailed kinetic analysis of the binding of the two siderophores subsequent to structural changes that bring about the transport which will require a substantive additional study beyond the scope and preview of the present work. We have accordingly moderated the conclusions about specificity of IrtB/Rv2895c for ferrated and non ferrated siderophores.

Comment: “Also, I may have missed the appropriate data in the paper, but were the following controls done? If so, perhaps a word or two about the results would be helpful, as I missed them”

(1): “pull-down assays for IrtB/rRv2895c in the absence of siderophore, and in the presence of deferrated siderophore.”
Our response:
Yes, the assays have been carried out to reveal that the binding of Rv2895c to Rv1349 does not occur in the absence of siderophore. Furthermore, liposome based import assay (as presented in the revised version, Fig. 7C 1 and 2) clearly points out that incubation of Rv2895c alone with LP-49 is unable to activate its ATPase activity. Also, incubation of cMyco bound Rv2895c does not bring about its import or activates the ATPase. These observations sufficiently indicate the specificity of interaction and import by Fe-cMyco bound.

(2): “..liposome reconstitution experiments with only rRv1349 and only rRv2895c, in addition to the rRv1349-rRv2895c combination.”
Our response: Both the experiments have now been included in our experiment as shown in Fig. 7C in the revised version.

Comment: “…Finally, the authors analyzed the interior of their liposomes for carboxymycobactin content via the fluorescence emission of the lysed pellet fraction. Similar transport assays are often done by monitoring the flux of radioactive Fe. Can the authors comment on the error analysis involved in their procedure?”
Our response: The procedure followed by us is an established one and is a widely documented method to carry out these studies. In fact this procedure has an advantage over regular transport assays carried by monitoring the flux of radioactive Fe because of ease of experimentation. The error analysis has been calculated and shown in each figure for both the import (Fig. 7B) and the export (Fig. 7C).

Comment: “Were control experiments done? For example, how much apparent uptake/efflux occurs in the absence of one or more of the proteins?”
Our response:
Yes, indeed the control experiments carried out involving encapsulation of only ATP and only cMyco in liposomes ruled out the passive efflux of these components. Also, incubation of liposomes with Fe-cMyco did not show any influx when the intrasliposomal contents were analyzed. These controls are shown in Fig. 7B and 7C of the revised manuscript.