Referee 1's Review (Domenico Mavilio, M.D., Ph.D.):

**********
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.
**********

In the present study, Daniel Scott-Algara et al. investigate on the role of CD85jpos NK cell subset as a potent suppressor of HIV-1 replication in an autologous in vitro NK/DC co-colture system. Authors also demonstrate that this phenomenon requires cell-to-cell interactions, it is not mediated by soluble factors or cytotoxicity and is mostly driven by a novel and still unidentified CD85J-ligand.

I found this work of great interest and it certainly advances our knowledge in the field. In line with the guidelines requested to publish a scientific article in PLoS ONE, the manuscript: 1) Reports experimental data that have not been published before; 2) Provides methodological approaches and analysis of high quality that justify all the conclusions drawn in the discussion paragraph; 3) Does not tend to overestimate the results and all experimental hypothesis are confirmed with solid data that also open new interesting questions for future investigations (as pointed out by authors in the discussion paragraph); 4) Provide a clear explanation of all techniques used; 5-6-7) It is well written, respects all applicable standards in regard to human experimentation and research conduct.

Nevertheless, before publication, I would better address some minor technical issues that, in the present form, are either incorrect or misleading.

a) Page 2 (Abstract), line 16: I suppose that rCD58j means recombinant CD58j. Authors need to specify this the first time.

b) Page 3, Second Paragraph, Line 2: the reference 12 does not include acute HIV-1 infection, but shows high frequencies of CD3neg/CD56neg/CD16pos NK among chronic HIV-1 infected viremic patients. In this cohort of individuals NK cell numbers are not showed as increased and CD3neg/CD56bright/CD16pos NK cells are not showed as depleted.

c) Page 5, Second Paragraph, Line 6: The word "co-culturecoculture" is misspelled.

d) Page 7, last line, and page 8, first line: "a modest decrease (I would rather write: "a not statistically significant decrease") of GFPpos MDCC induced by both CD58jpos and CD58jneg NK cells from co-coltures with uninfected MDCC" is clearly showed in Fig. 4c (not Fig. 3b) and is not "not shown" as written in the text.

e) Page 9, Line 12: Even though (as shown in Figure 6A) incubation with rCD58j alone undoubtedly reduced the CD85pos NK cell-mediated viral suppression to significant lower levels, it certainly did not abolish this viral suppression to levels similar to those obtained in control infected MDCC. If the figure is very clear, the text appears a little bit confusing. In this regard, authors should add another condition incubating rCD85j simultaneously with mAbs blocking all HLA A-B-C-E-G alleles. This last condition would likely abolish completely the viral suppression through the CD85j pathway and would further highlight the primary role of the unknown CD85j-ligand(s) and the secondary role of known CD85j-ligands (MHC-I molecules).

f) It seems unlikely from Figure 6b, Panel F, that 21.2% of HIVBal-infected MDCCs are positive for rCD85j after masking with HLA classI-PE abs. Moreover the legend of figure 6b is confusing. If the explanation in the text makes perfect sense, the figure legend is not that clear. Even the rCD85j on the x axes in Figure 6b is misleading given the fact panel B and E are stained (as written in the text) with HLA classI-PE abs and not with rCD85j.

g) Figures are not numbered

h) Figure 4 legend: line 7: I suppose authors mean "percentage of MDCCs (and not NK cells) expressing GFP.