Modeling metabolic disease susceptibility and resilience in genetically diverse mice

Reviewer #1 (Public review):

Summary:

The authors performed an in-depth analysis of three mouse strains with different levels of susceptibility to metabolic disease. Transcriptomics analyses of relevant deep tissues revealed many strain-specific differences in response to diet. They used gene set enrichment analysis to highlight possible biological pathways that may be involved in obesity and its metabolic consequences. These results were then confirmed using public data in both mice and humans.

Strengths:

Overall, this is an interesting study into the biological basis of differing phenotypic outcomes in response to metabolic challenges. The findings uncover several pathways that may shed light on the etiology of obesity and the associated health risks, as well as offer potential therapeutic avenues to prevent them.

Weaknesses:

While the experimental design and analysis are mostly good, some aspects of the present paper could be improved.

(1) Most results are insufficiently described. P-values are almost entirely absent in the main text. Sometimes the significance is indicated in the figures, and other times it is missing. For example, strains are sometimes described as having a higher XYZ, something that is never shown in the plots, and no p-value is ever given.

(2) While the biological methods are meticulously described, statistical methods are barely mentioned in the methods section. For example, line 578, "multiple comparisons (...) were performed using the glht function of the multcomp package". What is this? What method does it use? And how was mediation analysis done? Line 575 mentions that models were compared, with no description of how this was done. Mentioning the package (or even function) is not sufficient. The package and function are an implementation; they are not the method. The actual method needs to be clearly mentioned and (at least minimally) described, in addition to having the reference for methods that are not ubiquitous (i.e., the Benjamin-Hochberg method is well-enough established to forgo this).

(3) The methods should also be briefly introduced in the results section before describing the results of those methods.

(4) The role of immune signaling pathways and associated phenotypes (e.g., monocyte fraction) is over-interpreted. While the differences shown are convincing, they do not convincingly show a role in either obesity or disease. The parsimonious explanation is that such changes happen as a consequence of dyslipidemia rather than a cause. It is possible that these pathways play a more direct role in this, but the authors do not present compelling evidence of this, and, failing this, the language in the text needs to be toned down.

Reviewer #2 (Public review):

This study investigated changes in metabolic health across three genetically diverse mouse strains (NZO/HlLtJ, C57BL/6J mice, CAST/EiJ) that were fed either control or high-fat high-sucrose diets. The strength of this study is the depth of metabolic phenotyping, the use of both male and female mice, and the multi-tissue metabolic analysis, including metabolic and gene expression analysis in pancreatic islets, kidney, muscle, heart, liver, and adipose tissue.

Weaknesses include that only three mouse strains were included in this comparison, particularly given that similar comparisons have been published in the past and that the Jax lab has access to a wide range of mouse strains with diverse genetic backgrounds. Why were CAST mice included over (for example) BALB/c mice that are more commonly used in metabolic studies and are well known to show protection against diet-induced metabolic disease? Furthermore, the feeding regime was limited to 9 weeks, which may not be sufficient to evoke pronounced metabolic remodelling.

NZO mice are well known to develop obesity. However, only approximately 50% develop type 2 diabetes and beta-cell dysfunction. How were these mice selected in the study? The results state 'Most of the male NZO mice and a few female mice displayed overt diabetes', suggesting that all mice were included irrespective of their diabetic phenotype. More information on the rationale for this is required.

The transcriptomics data are presented in a convoluted way. As a reader, the main interest would be to determine the differences in diet-induced adaptations within each strain (e.g., why are CAST mice resistant to diet-induced metabolic defects?). However, the way Figure 4 is currently presented does not allow for this. Instead, the data are 'compressed' by looking at general changes in metabolic pathways between tissues in all three mouse strains. In addition, Figure 4E does not show the directionality of the responses within each pathway. For example, are the metabolism and inflammation pathways suppressed or activated? While more data is shown for adipose tissue, this is not sufficient.

Currently, the metabolic cage data are separated by diet within the main figures. However, given that the diet effect is the major comparison, this needs to be rearranged, and strain differences within each diet could be shown within the supplement.

The graphs lack labelling throughout to specify which lines/bars represent which strains and diets. This is particularly the case in the metabolic cage analysis.

Reviewer #3 (Public review):

Summary:

Using three strains of mice that are founders of the Diversity Outbred Population of mice, this paper attempts to identify key genetic drivers of obesity and metabolic dysfunction. Through a series of in-depth phenotyping experiments, the authors describe substantial differences in the propensity of these strains to develop obesity and complications associated with obesity. The key here was the careful selection of these strains, as they mostly spanned the spectrum of minor susceptibility (C57BL/6J), major susceptibility (NZO/HILtJ), and complete resistance to diet-induced obesity (CAST/EiJ). This was done in the setting of both a normal diet and a high-fat diet. These studies identified that one of the most transcriptionally activated tissues in this setting across the strains was adipose tissue. Furthermore, a critical pathway in adipose tissue that inferred protection against obesity in the CAST strain was related to immune infiltration. Subsequently, the authors extended their studies into this phenotype using their existing access to the vast array of genetic information from the DO datasets. From this analysis, it was identified that a key region on Chr19 had a significant influence on this phenotype, and subsequent work investigated the potentially causal genes. Overall, this study encompasses an impressive amount of in vivo and genetic work and identifies some new gene regulators associated with obesity complications, which warrant further investigation.

Strengths:

This study engages multiple mouse lines with diet intervention, as well as powerful genetic mapping tools to isolate genetic drivers of various obesity related phenotypes. The animal studies are thorough and well performed, and they also include detailed omics analysis of several tissues. Subsequent genetic mapping uses some of the world's most powerful preclinical genetic approaches, and findings identify some novel genes associated with obesity.

Weaknesses:

These mouse lines and hybrid genetic screens in this paper have been used for some years now to map similar phenotypes, so in that sense, the approach is not overly novel. Moreover, the most compelling and exciting part of the study, in this reviewer's opinion, is the DO mapping of the immune phenotype in adipose tissue. In some ways, the authors could have potentially come to this same conclusion without the need to perform the mouse studies in the three different strains (other than the nice storytelling of finding the phenotype initially in CAST). Likewise, with this being the most novel aspect of the study, it was a shame that the genes identified at Chr19 were not investigated in more detail in the manuscript, other than just some associative outcomes in mice and humans. It would have been pleasing to see some attempt to validate one of these genes in a mouse model, linking it to either obesity or immune phenotypes in WAT.