Comparative transcriptomic analysis reveals translationally relevant processes in mouse models of malaria
Curation statements for this article:-
Curated by eLife
Evaluation Summary:
In this work, the authors use independent public datasets to perform an unbiased investigation of the similarities and differences of mouse models to human malarial disease using comparative transcriptomics. Whilst the data cannot convincingly identify which mouse models are best suited for studying specific human malaria phenotypes, the comparative analyses do indicate that these models reflect the broad diversity of human disease.
(This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #1 agreed to share their name with the authors.)
This article has been Reviewed by the following groups
Listed in
- Evaluated articles (eLife)
Abstract
Recent initiatives to improve translation of findings from animal models to human disease have focussed on reproducibility but quantifying the relevance of animal models remains a challenge. Here, we use comparative transcriptomics of blood to evaluate the systemic host response and its concordance between humans with different clinical manifestations of malaria and five commonly used mouse models. Plasmodium yoelii 17XL infection of mice most closely reproduces the profile of gene expression changes seen in the major human severe malaria syndromes, accompanied by high parasite biomass, severe anemia, hyperlactatemia, and cerebral microvascular pathology. However, there is also considerable discordance of changes in gene expression between the different host species and across all models, indicating that the relevance of biological mechanisms of interest in each model should be assessed before conducting experiments. These data will aid the selection of appropriate models for translational malaria research, and the approach is generalizable to other disease models.
Article activity feed
-
-
Evaluation Summary:
In this work, the authors use independent public datasets to perform an unbiased investigation of the similarities and differences of mouse models to human malarial disease using comparative transcriptomics. Whilst the data cannot convincingly identify which mouse models are best suited for studying specific human malaria phenotypes, the comparative analyses do indicate that these models reflect the broad diversity of human disease.
(This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #1 agreed to share their name with the authors.)
-
Reviewer #1 (Public Review):
Georgiadou et al have compared the changes in gene expression in the leukocytes in mice infected with different rodent malaria parasites with that of humans infected with P. falciparum. Their goal is to provide a scientific rationale for the use of rodent malaria models. This is an important and well-written study that has the merit to go beyond pre- or ill-conceived opinions on the use of models. The experiments are mostly well-designed, performed, and analyzed. My only major concern is that the conclusion that P . yoelii 17XL is the closest model to uncomplicated and severe malaria is a bit premature. The limits of the study should be kept in mind, to allow a more balanced view of their results.
-
Reviewer #2 (Public Review):
Georgiadou et al compare transcriptomic changes in malaria-infected rodents versus humans to determine which/whether mouse malaria model(s) correspond to human malaria responses. The authors also provide new data on lactate levels in infected mice. Given the controversy over the relevance of mouse models, the underlying purpose of the study is valuable.
The authors should more clearly inform readers of fundamental differences between the mouse models and human malaria infections they study that will contribute to differences, including:
• Humans but not mice are the natural host for their corresponding parasites
• Human infection is initiated by sporozoite inoculation vs blood stage in their models
• Human Pf densities are far below the values seen in the Py17XL mouse model
• Humans here are …
Reviewer #2 (Public Review):
Georgiadou et al compare transcriptomic changes in malaria-infected rodents versus humans to determine which/whether mouse malaria model(s) correspond to human malaria responses. The authors also provide new data on lactate levels in infected mice. Given the controversy over the relevance of mouse models, the underlying purpose of the study is valuable.
The authors should more clearly inform readers of fundamental differences between the mouse models and human malaria infections they study that will contribute to differences, including:
• Humans but not mice are the natural host for their corresponding parasites
• Human infection is initiated by sporozoite inoculation vs blood stage in their models
• Human Pf densities are far below the values seen in the Py17XL mouse model
• Humans here are malaria-experienced while mice are malaria-naïve in this study
Despite the inherent differences between their models and the human condition, the finding that Py17XNL infection of mice were concordant in their differential transcriptomic profiles to those of different severe malaria syndromes will be of interest to readers. The inconsistent findings between the PCA analysis and those of the 20 most differentially expressed human genes will leave some questions as to the robustness of these findings.
-
Reviewer #3 (Public Review):
Georgiadou, et al. combined public human transcriptomic datasets with new mouse model experiments to determine if high-throughput gene expression comparisons provide higher resolution, compared to clinical phenotypes, for matching five different mouse models of malaria to human malarial syndromes. A notable strength of the analysis is that the authors have attempted to correct for differential abundance of immune-cell population subsets in their differential expression (DE) calculations. This is a perennial bias in peripheral-blood transcriptomic studies. However, a potential concern is that the affect of this correction was not directly compared to uncorrected models and/or direct modeling of these cell proportions to the disease outcomes in each experiment. This is relevant because the cell populations for …
Reviewer #3 (Public Review):
Georgiadou, et al. combined public human transcriptomic datasets with new mouse model experiments to determine if high-throughput gene expression comparisons provide higher resolution, compared to clinical phenotypes, for matching five different mouse models of malaria to human malarial syndromes. A notable strength of the analysis is that the authors have attempted to correct for differential abundance of immune-cell population subsets in their differential expression (DE) calculations. This is a perennial bias in peripheral-blood transcriptomic studies. However, a potential concern is that the affect of this correction was not directly compared to uncorrected models and/or direct modeling of these cell proportions to the disease outcomes in each experiment. This is relevant because the cell populations for the mouse experiments were assessed via flow cytometry, whereas they were inferred from the transcriptomics data in the human studies. It would also be useful to see a comparison of proportions measured by flow with those inferred from transcriptomics on the same samples.
The authors have done a commendable job of using independent public datasets for validation and to achieve their aim of performing an unbiased investigation of the similarities and differences of mouse models to human malarial disease using comparative transcriptomics. As presented, the results do not convincingly identify which mouse models are best suited to use to study specific human malaria phenotypes. On the other hand, the span of the mouse models with one another compared to human in the PCA plots is consistent with the claim that a strength of mouse models is that they reflect the broad diversity of human disease.
Readers should be aware that early vs. late timepoints were used for DE calculations in the mouse studies to recapitulate uncomplicated vs. severe disease in humans. Early timepoints from animals that go on to develop severe disease may already show some transcriptomic patterns of SM, which could thus be lost in this comparison.
-
