Reviewer 1, Pushpa Pandiyan
In the manuscript titled "IgG3 Deficiency Extends Lifespan and Attenuates Progression of Glomerulonephritis in MRL/lpr Mice", Greenspan et al., have studied the role of IgG3 antibody in the autoimmunity associated with MRL/lpr mice. Although IgG3 antibodies are implicated in autoimmunity, their definitive role in causing immunopathology is unclear. These authors have generated IgG3 deficient mice and have shown that the loss of IgG3 protects mice against glomerulonephritis-associated morbidity and mortality. The authors deserve credit for the well-controlled experiments using MRL/lpr mice with all three genotypes derived from the same litters of crosses between heterozygotes. They also deserve credit for presenting their data clearly. The authors have shown that although IgG3 deficient MRL/lpr mice suffer renal damage, the process of progression to end stage renal disease is retarded or arrested, thereby increasing their survival, compared to normal MRL/lpr mice. While these evidence imply that IgG3 antibodies play a pathogenic role in MRL/lpr mice, they are not conclusive. To further strengthen the studies in elucidating the role of IgG3, they should experimentally demonstrate that passive transfer of IgG3 antibodies derived from MRL/lpr mice or injection of hybridomas secreting IgG3 antibodies causes pathology or at least decreases the survival in IgG3 deficient mice.
Minor concerns and questions:
• It is known that IgG3-deficient mice in BALB/c background are more susceptibleto pneumococcal infection than wild-type BALB/c mice. In this study, did the authors observe any infection by commensal bacteria or any evidence of inflammatory bowel disease in IgG3 deficient mice?
• In Figure , the signs " +/+, +/- and -/-" are not aligned with the gel lanes properly.
The manuscript can be accepted for publication in "Biology Direct" provided that the authors perform the suggested experiment.
Authors' responses to reviewer 1 (Pandiyan)
The experiments requested by the reviewer were done many years ago and reported in multiple publications, two of which are cited as references 13 and 14 in our manuscript.
In the present study, we did not investigate infections or the possible presence of inflammatory bowel disease. Such matters are beyond the scope of the present study.
Comments on once-revised manuscript - reviewer 1 (Pandiyan)
Antibodies of the IgG3 subclass are previously known to play a role in the pathogenesis of the spontaneous glomerulonephritis in MRL/MpJ-Tnfrsf6lpr (MRL/lpr) mice. However, as put forward by the authors, the rationale of the current study is to provide a direct and a more definitive evidence that IgG3 antibodies are important in the spontaneous renal disease of MRL/lpr mice, using mice with genetic deficiency in the capacity to synthesize IgG3 antibodies. In order to validate the direct pathological role of IgG3, the authors must perform the passive transfer of the antibodies in their mouse model. I agree that it has been previously shown that passive transfer of IgG3 monoclonal antibodies can elicit renal pathology analogous to the damage that develops spontaneously in MRL/lpr mice. However as authors argued, the serum levels of such antibodies required for those effects may surpass those that are physiologically relevant in the spontaneous disease. Moreover those experiments were not performed in mice that were specifically defective in IgG3 subclass antibodies in MRL/lpr background. Therefore, the authors must demonstrate a direct role of this subclass of antibodies in the pathology in mice defective in IgG3 subclass antibodies in MRL/lpr background, which was not done in the previous studies, cited as references 13 and 14. It is necessary to show that such a passive transfer of physiological levels of IgG3 antibodies reverses and reduces the longer life span, observed in the IgG3-deficient MRL/lpr mice in their model. The manuscript can be accepted for publication in "Biology Direct" provided that the authors perform this experiment.
Authors' responses to reviewer 1 (Pandiyan) regarding initially revised manuscript
While we agree with reviewer 1 that it would potentially be useful to perform passive transfer experiments with IgG3 antibodies in the IgG3-deficient MRL/lpr mouse model, such experiments are not guaranteed to recreate the in vivo conditions that obtain during the evolution of spontaneous glomerulonephritis. It would take substantial effort to produce the relevant reagents (and we do not know what, if any, antigen specificities these antibodies would need to express), define the necessary experimental parameters (e.g, the amounts of antibodies and the duration of the period of administration), and carry out the experiments. These efforts potentially represent an amount of work sufficient for another manuscript. Furthermore, neither of the other two reviewers commented on the need for passive transfer experiments to justify the publication of the present manuscript. Therefore, we respectfully disagree with the reviewer on this point.
Reviewer 2, Irun Cohen
The aim of the study was to provide direct evidence that IgG3 isotype antibodies are critical in the glomerulitis that develops in MRL/lpr mice. The method was to backcross a 'defective' gamma3 heavy chain into MRL/lpr mice and observe serum anti-dsDNA antibodies particularly of the IgG3 isotype; kidney lesions; IgG and IgG3 eluted from the kidneys; kidney function and survival. The mice with the 'defective' gamma3 (termed -/-) demonstrated a slight decrease in general serum IgG anti-dsDNA with essentially no IgG3; less IgG and much less IgG3 eluted from the kidneys; modified histologic kidney pathology; and delayed death compared to wild type (termed +/+) and heterozygous mice (termed +/-).
I have several problems with the conclusions one may draw from these results:
1. The so-called -/- mice are transgenic for a defective gamma3 heavy chain; they are not gene knockouts, and, in fact, IgG3 was eluted from their kidneys. Therefore, they do produce at least some IgG3 antibody. So the observations and results cannot be attributed to the absence of IgG3 antibodies.
2. We are not given any details about the actual defect in the transgene, or how expression of the defective gene might have affected expression of the wild type gene that was left, or ideas about the possible role of IgG3 in the recognition of dsDNA, or the likely relationship between the mouse IgG3 isotype and the equivalent human IgG isotype (IgG2?) and their possible roles in SLE in humans vs mice.
3. The mice with the defective IgG3 heavy chain gene died, albeit later than their counterparts; was there any effect of the transgene on the general lymphoproliferative disease of the mice? Did the transgene affect only the kidneys? What actually killed the mice? I enjoyed reading the paper; I found it to be thought provoking and raised the issue of how a single gene product might affect the pathophysiology of a complex immune system and organ system disease.
Authors' responses to reviewer 2 (Cohen)
The original gamma 3 heavy chain-deficient mice were obtained by homologous recombination of the targeting vector at the gamma 3 heavy chain locus (ref. 30 in the manuscript). The SNP analysis, included for reviewers only, strongly suggests that the defective gamma 3 heavy chain allele inserted as intended eliminating the wild-type allele and making the genetically-manipulated mice true knockouts.
The targeting vector, as described in ref. 30, contained a neomycin-resistance gene just upstream of the sequence coding for CH1. This gene structure was designed to be unexpressed.
The Western blot analysis of sera from mice possessing the defective gamma 3 heavy chain allele in both ref. 27 and the present manuscript support the inference that if there is any IgG3 antibody in the sera of these mice it is at a concentration below the level of detection of this assay. In addition, in ref. 27 it was shown by ELISPOT analysis that there were no spleen cells spontaneously secreting IgG3 (sensitivity approximately one cell per million) in mice of the BALB/c background that were genotyped as gamma 3 -/-. In contrast, spleens from wild-type BALB/c mice contained numerous cells secreting IgG3.
Mouse IgG3 and human IgG2 are not fully 'equivalent,' although both isotypes are associated with responses to polysaccharide immunogens. Human IgG2 has not been described to dynamically self-associate non-covalently like murine IgG3, but Yoo et al. (J Immunol. 2003 Mar 15;170(6):3134-8. PubMed PMID: 12626570) have described IgG2 dimers that form due to disulfide bonds between Fc regions of otherwise separate IgG monomers.
Reviewer 3, Etienne Joly
The manuscript by Greenspan et al. describes that, when an inactivated IgG3 locus is introgressed in MLR/lpr mice those become partially protected against glomerulomephritis.
On the positive side, I found the observations reported convincing, and those could clearly be relevant to deciphering the pathogenesis of kidney damage in lupus, and other antibody-mediated diseases.
On the other hand, although the overall quality of the English was more than adequate, I also found that many aspects of the work were very poorly presented, and this was particularly true for the material and methods and the results sections. I would therefore strongly suggest that many modifications should be made to the manuscript before it is published. Below, I have provided a non exhaustive list of suggestions, but I would also advise the authors to seek advice from willing colleagues inside and outside of their immediate field of study to help them improve the overall structure and contents of their manuscript.
2) In the abstract, I suggest replacing the sentence "gamma3 -/- mice exhibited minimal titers of these auto-antibodies " by "gamma3 -/- mice exhibited baseline levels of these auto-antibodies"; Indeed, 'minimal titers' suggests that there are some IgG3 still present, and according to McLay et al. 2002, the IgG3 KO mice have a complete absence of IgG3 synthesis. Any signal detected should be due either to background noise, or cross reactivity of the reagents on other isotypes.
3) In the introduction, in the sentence "In terms of IgG subclasses, there is evidence implicating both IgG2a and IgG3 antibodies in the pathogenesis of renal disease [6
]" you should specify whether this is true for humans, mice, or both
4) further down, I would change 'Subsequent reports revealed that many IgG3 antibodies self-associate in antigen-independent [16
]' to 'Subsequent reports revealed that a specific particularity of many IgG3 antibodies is to self-associate in antigen-independent [16
5) In the section on genotypes of mice in the M&M, there should be a much more thorough description of the IgG3 +/- genotyping protocol, with a reference to panel A of Figure . In fact, it may be best to describe the whole of Figure in the M&M. The SNP genotyping should then be presented as a separate paragraph (or section even) Since the Fas locus is very important in the MRL/lpr phenotype, it would also be good to underline that this region is indeed correctly from the MRL origin. There is also a typo on the last line: table, not tabel.
6) In the section entitled, Measure of auto-antibodies specific for double-stranded (ds) DNA and g-actinin, I am not exactly sure of what the purpose of the passage of dsDNA through a 0,45 micron filter is. It is possibly to sterilize it, and/or to shear the DNA, but certainly not to purify it !
7) In the section "Elution of antibodies from kidneys": the buffer(s) used, and the precise conditions should be specified ( volumes, temperature, time, shaking or not...). Also, I suggest changing 'The total eluted IgG or IgG3 in each group was determined ...' to 'The amounts of eluted IgG or IgG3 in each group were determined ...'. On this same subject, in the results section, the amounts of antibodies eluted are expressed as concentrations, which really does not mean anything if one does not know the final volume used for elution. I would strongly suggest expressing those as total amounts of antibody per kidney, or per mouse.
8) Evaluation of renal structure: Upon sacrifice (at specific ages selected to provide insight into ( not to) the progression
9) I suggest that the first paragraph of the results really belongs in the M&M section. The layout of Figure has also gone wrong( lane legends are nto aligned with picture), but I suspect this has happened during the submission process. As a whole, however, I found that the layout of figures was really awkward, and not bearing in mind that those would be printed as a very small fraction of a page in the final pdf. Blank areas should thus be kept to a minimum, and important information presented in a way that can sustain shrinking (such as the picture of the gel on Figure )
10) On Figure , I find it really bizarre not to have the percentages of normal Balb/C mice. As it is, how do we know that the low percentage of DN is not an effect of the IgG3-/- specific to the Balb/c background?
11) In the text describing Figure , I would replace 'minimal levels' by 'baseline levels'. The last sentence of that sections is completely out of place here: you have not made any previous mention of differences in renal function in the -/- mice yet !!
12) In the section on eluted antibodies: as mentioned earlier, these need to be expressed as total amounts (in micrograms) per kidney or mouse. You also need to describe how the ratios between total IgG and IgG3 were established. The most disturbing thing to me, however, was that these data were not presented as a figure. Is this because this was done only once or twice (which would explain why there is not deviation provided)? If it is the case, you should really repeat this a few more times instead of providing data in this shoddy format which is both difficult to apprehend, and does not correspond to data which had been documented sufficiently thoroughly.
13) Regarding figure plots, I would strongly encourage you to convert them to a smaller format with white backgrounds. Figures , , and the data on eluted antibodies could then all be presented on a single figure with 4 small panels.
14) I would like to express my very strong disagreement (not to say outrage) regarding the following sentences: ""Serum creatinine in both the +/+ and +/- mice at 26 weeks was significantly (t > 3.6, p < 0.01) higher than the level in the same genotype at 18 weeks, whereas the difference in serum creatinine between 18 and 26 week old -/- mice was not significant (t = 2.2). These data indicate that there was progressive renal insufficiency (and presumably nephron loss) in the mice capable of producing IgG3, but not in -/- mice.". Indeed, just because a difference does not reach statistical significance, it does not mean that there is no difference. I would thus suggest the following wording: "Serum creatinine in both the +/+ and +/- mice at 26 weeks was significantly (t > 3.6, p < 0.01) higher than the level in the same genotype at 18 weeks, whereas the difference in serum creatinine between 18 and 26 week old -/- mice was less pronounced and did not reach statistical significance (t = 2.2). These data thus suggest that there was progressive renal insufficiency (and presumably nephron loss) in the mice capable of producing IgG3, but that this was attenuated in -/- mice."
15) I found the whole section on histopathology extremely difficult to read and understand. Indeed, whereas, in other sections of the manuscript a clear effort is made to be didactic for a non specialist reader ( which is commendable since Biology Direct is not a specialty journal), I found that this section was incredibly difficult to decipher. I would actually suggest pooling Figures and , and placing the actual titles next to each of the histology panels rather than letters, to make it clear that those are provided as examples, and not representative of mice of one or another genotype. And once the manuscript has been modified, I would strongly urge you to seek the assistance of colleagues outside the field to help make this section into something that is more easy to understand for people like me.
16) Regarding Figure , it took me a while to convince myself that the data for all three groups of mice had been pooled. This is simply not an appropriate procedure ! I suspect that, once again, this was due to the fact that you did not have sufficient numbers of mice in each group to plot the survival curves independently for each group, and looking at the curves, it seems to have been due to a lack of females, most probably because those had been needed to breed to maintain the colony and generate more mice. In this kind of situation, where you have made the observation over and over again, but do not have the right numbers of mice in each group, I would suggest to take Figure out, and simply stating in the text that, as in the MRL/lpr parental strain, you did not detect any gender effect in any of the backcrossed groups.
17) In the discussion, the idea about an IgG3 rheumatoid factor is interesting, but it would be helpful to define clearly what a rheumatoid factor is (i.e. an antibody that can bind to the Fc portion of other antibodies).
18) Further down, replace "with the notion that elimination of IgG3 antibodies attenuates the irreversible renal damage associated with the MRL/lpr mouse strain." by "with the notion that the absence of IgG3 antibodies attenuates the irreversible renal damage associated with the MRL/lpr mouse strain."
Authors' responses to reviewer 3 (Joly)
1) The abstract now notes that the MRL/lpr strain has been widely studied as a model of human systemic lupus erythematosus.
2) The requested wording change in the abstract has been implemented.
3) The relevant sentence (p. 4) now indicates that the reference to Ig3 and IgG2a antibodies applies to mice. Humans do not have a subclass referred to as "IgG2a."
4) The relevant passage (Introduction, paragraph 3) has been modified to read as follows: "Subsequent reports revealed that a notable characteristic of many IgG3 antibodies is to self-associate in antigen-independent [16
] or antigen-dependent [21
5) In the M&M section on "Genotypes of mice," reference is now made to Figure . As suggested, we have also moved the sentences about the SNP typing to a separate section with a sub-heading, and the implications of the SNP analysis for the allelic origins of the genes at the Fas locus on chromosome 19 are now explicitly noted in the appropriate subsection in the Results (p. 9). The typographical error in the last line of what is now the "Genomic SNP analysis" section has been fixed.
6) Passage through the 0.45 micron filter is to remove the single stranded DNA, thus enriching for dsDNA. This is a standard method which has been used in all of our (CP) previous relevant publications, and it has never been questioned.
7) The wording changes requested have been implemented in the M&M section on the methods for elution of IgG antibodies from kidneys. As requested, details of the methods have also been inserted.
8) The requested revision in the text has been implemented in the M&M section on "Evaluation of renal structure."
9) For the gel photograph in Figure , we have substituted single-letter abbreviations (w - wild-type; k - knockout; h - heterozygote; m - markers) which we hope will better preserve alignment with the correct lanes throughout the submission and publication process.
10) We have reduced Figure to one panel focused on T-cell subsets. The data presented are from an experiment in which spleen cells from wild-type C3H mice were used for comparison to the spleen cells from MRL/lpr mice of all three γ3 genotypes (+/+, +/-, -/-). Similar results were obtained in a repeat experiment with spleen cells from BALB/c γ3 -/- mice.
11) The requested revision has been made (Results, paragraph 4; "minimal levels" replaced by "baseline levels"). In addition, we have modified the wording of the final sentence of the section to read: "The greater concentration of the IgG3 auto-antibodies in the IgG3 producing mice could contribute to any observed γ3 genotype-associated differences in renal function, renal histopathology, and survival (see below)."
12) See response to comment 7.
13) We acknowledge the reviewer's comment but do not believe it is necessary to consolidate the figures or create a figure for the limited kidney elution data.
14) We have implemented requested change in wording in paragraph 7 of Results.
15) As suggested, the first paragraph of the section has been significantly revised.
16) As suggested, we have eliminated Figure and now refer to the corresponding results in the text.
17) A description of rheumatoid factors has been inserted, as requested, in paragraph 4 of the Discussion.
18) We have replaced "elimination" with "the absence" in paragraph 5 of the Discussion.
Comments on once-revised manuscript - reviewer 3 (Joly)
The manuscript by Greenspan et al. describes that, when an inactivated IgG3 locus is introgressed in MLR/lpr mice those become partially protected against glomerulomephritis. I found the observations reported convincing, and those could clearly be relevant to deciphering the pathogenesis of kidney damage in lupus, and other antibody-mediated diseases.
Compared to the initial version, I find that the modifications introduced following the suggestions of the three referees have resulted in a very significant improvement of the manuscript.
I do, however, remain thoroughly unsatisfied with the aspects dealing with elution of immunoglobulins from kidneys, for the following reasons:
- The authors only provide the concentrations of antibodies recovered, but not the total volumes of those eluates. There is not even an indication that the same volumes were used for all samples. What should really be provided, and compared, are amounts of antibodies per kidney (or per mg of tissue).
- From the way the data is presented in the text, I have come to suspect very strongly that these data were obtained only once, with each value corresponding to one eluate obtained from the pooled half-kidneys in each of the cohorts. I can understand that this may be an unavoidable limitation of the approach, but this should have been made very clear, and justified, in the text.
- In this paragraph, the authors state that "< 12% of the IgG in eluates from -/- mice was IgG3.", suggesting that there are some IgG3 to be found in IgG3 KO mice. According to McLay et al. 2002 (ref 27), however, the IgG3 KO mice have a complete absence of IgG3 synthesis. Any signal detected would thus be due either to background noise, or cross reactivity of the reagents on other isotypes.
- And what to say of the +/- mice, for which 'IgG3 represented ~10% of the total IgG eluted', i.e. roughly the same levels as those from IgG3 KO mice?
- All those considerations are particularly relevant since, contrary to what the authors state, the differences in total IgG concentrations between +/+ and -/- mice are actually rather minor: 16.8 vs 6 μg/ml at 18 weeks, and 86 vs 66 μg/ml at 21 weeks. From what I can make out from the data in the form in which it is provided, the IgG deposits are mostly of the IgG3 subtype in +/+ mice, and not so much less, but of other subtypes in -/- mice, and very surprisingly also in +/- mice. This last point is completely brushed over by the authors. This observation in itself underlines the need for those experiments to be repeated on additional mice, or cohorts of mice, before those results can be considered as sound and reliable.
I was also sorry to see that the authors had chosen to ignore my suggestions on two much less important points:
- In the paragraph of the M&M section about anti dsDNA ELISA, I thought that it would be helpful to clarify that the purpose of passing the DNA through a 0.45 micron filter was to eliminate single stranded DNA.
- I also suggested that the layout of figures should be modified. As they are now, I suspect that those figures will become extremely difficult to read once they have been reduced to printing sizes, and will become completely useless to people if printed in black and white (as well as for color-blind people).
Authors' responses to reviewer 3 (Joly) regarding initially revised manuscript
We now explicitly state in the relevant section of Materials and Methods that the kidney elutions were only performed once. As for the methodology, the identical volumes of fluid were used for eluting each kidney. We believe that the apparent IgG3 concentrations eluted from the γ3 -/- kidneys most likely arose from the crossreactivity of the detection reagent for the ELISA.
As requested, we have added the rationale for the filtration of the DNA to the section on the ELISA for autoantibodies specific for dsDNA in Materials and Methods.
Although we appreciate the reviewer's substantial efforts in reviewing the manuscript and take seriously his suggestions (as demonstrated by implementing almost all of them), we are not convinced that it will be a net benefit to modify the figures.