Referee 2's Review (Marty Roop):

**********
N.B. These are the comments made by the referee when reviewing an earlier version of this paper. Prior to publication, the manuscript has been revised in light of these comments and to address other editorial requirements.
*********

O. R. Crasta et al.
Genome sequence of Brucella abortus vaccine strain S19 compared to virulent strains yields candidate virulence genes

The basis for the attenuation of S19 has been a mystery (or at least a point of controversy) for a long time. Hence, the comparative genomic data presented in this paper will be of great interest to those who work in the Brucella field. The data are clearly presented and the decision to use both strains 2308 and 9-941 as bases for comparison is a good one, since the progenitor of S19 no longer exists. Unfortunately, the authors' attempts to link their experimental findings to the biological characteristics of the strain S19 fall short. Specific points that need to be addressed by the authors are listed below.

1. One thing that is particularly striking is that this work did not uncover any overt mutations in genes that have previously linked to virulence in Brucella. Such genes include those involved in the biosynthesis of the LPS O-chain, Type IV secretion, purine or flagellar biosynthesis, to name just a few. The authors do note, however, that differences occurring in intergenic regions were also found between S19 and the other two strains. Did any of these particular intergenic differences occur upstream of established virulence genes? Both of these points should be clearly addressed in Discussion section of this paper.

2. On page 3, lines 4 and 5, the authors state "In 1956, S19 was further selected for its sensitivity to erythritol toxicity". What does this mean, and where are the supporting literature citations? It is this reviewer's understanding that the erythritol toxicity experienced by S19 is due to this strain's inability to completely utilize this compound because of the eryC and eryD mutations present in this strain (Sperry and Robertson. 1975. J. Bacteriol. 124:391-397). Unless this reviewer is mistaken, S19 was not exposed to erythritol in vitro and erythritol sensitive derivatives of S19 were not "selected" for. If this genetic selection did occur, the appropriate literature citation should be provided.

3. The work of Harry Smith and his colleagues should be cited (e.g. H. Smith et al. 1962. Nature 193:47-49) with regard to the proposal that erythritol plays an important role in the virulence of B. abortus strains in their natural ruminant host. The authors should also be careful to make sure that it is clear to the reader that the experiments that have been performed in mice DO NOT address the question of whether or not an eryCD mutation would attenuate B. abortus 2308 in a pregnant ruminant. A dhbC mutation that prevents siderophore production does not attenuate B. abortus 2308 in the mouse model, but the B. abortus dhbC mutant is extremely attenuated in pregnant cattle (Bellaire et al. 2003. Infect. Immun. 71:1794-1803).

4. Page 3, line 14 - is the smooth LPS phenotype of S19 the same as that of 2308 and 9-941? Literature citations to that effect should be provided, or data from the authors' own evaluations of these properties should be included because this is an important point of consideration with regard to the virulence properties of these strains.

5. Page 6, lines 17 through page 7, line 26 - it is unclear here what role that the authors are proposing that the deletion in the outer membrane protein encoded by BAB1_0069 in B. abortus 2308 has in the attenuation of B. abortus S19. Are they proposing a loss of host specificity or a more generalized lack of virulence? The fact that similar deletions are present in the corresponding genes in B. suis 1330 and B. melitensis 16M would argue against the latter as these strains are clearly virulent in their natural hosts and in humans. This section needs to be revised so that the authors' intended point is clearly presented.

6. Page 7, line 28 through page 8, line 21 - as noted previously, the authors need to clearly point out to the reader that the role of eryCD mutation in the attenuation of B. abortus S19 in ruminants is still an open question. The importance of the erythritol utilization genes in the virulence of B. abortus strains in pregnant ruminants has not yet been directly addressed. The authors should also take into account the fact that erythritol is toxic for S19 in their discussion of the eryF mutation identified in this strain. Specifically, how can a substance be toxic if the bacterium cannot transport it?

7. Page line 22 through page 10, line 26 - the authors' justifications for "highlighting" the genetic differences between S19 and 2308 and 9-941 discussed in these sections (e.g. mutations in BAB1_0019; BAB1_0237; BAB1_1188; BAB1_1354) is not at all clear. In fact, there are several other genes listed in Table 7 (e.g. clpX, uvrB) that would seem to be more likely candidates for genes that have been affected during the attenuation of S19 since mutations in these genes would predictably make this strain less resistant to the environmental conditions encountered in the phagosomal compartment of host macrophages (Köhler et al. 2003. Trends Microbiol. 11:215-219).

8. The abrupt ending of the Discussion section of this paper is disappointing and detracts from the presentation. A concise summary of the authors' experimental findings and a scholarly discussion of their potential biological relevance would seem to be in order. A specific point of discussion could be - can the results of this comparative genomics study provide a plausible explanation for the attenuation of B. abortus S19 in cattle?