I subscribe to several Nature Publishing Group (NPG) journals. Anyone serious about science probably reads Nature, but NPG publishes around 100 other journals. I subscribe to Nature Reviews Drug Discovery and Nature Reviews Genetics because a) they review the current literature and thinking in the field and b) our position is based on genetics and illustrated by drug development. So those two journals provide me with valuable information. I also read other NPG journals including Nature Medicine and Nature Biotechnology and find them all very informative. Moreover, I have found the NPG journals far more likely than Science to accept letters and manuscripts that criticize the science of using animals as predictive models for humans. Nevertheless, NPG is a very large publisher of animal model data.
In the August issue of Nature Reviews Genetics, several scientists(1) comment on the future of model organisms in light of all the genetic studies ongoing in humans. Most have a vested interest in model organisms in general so their comments are more or less what one would expect. But some are nonetheless interesting so I will examine a few of them below. (The article is available here but you will probably have to pay to access it.)
Proponents of using genetically modified animals in research endeavoring to find information relevant to human disease and drug targets argue that there have been genes located in animals that perform more or less the same function in humans.(1) This is to be expected. We have never said that animals and humans are so dissimilar that there will be no overlap between them in terms of gene function even in terms of the role of genes in specific diseases. There are two issues. 1. Can gene function in disease states be determined using humans, a more reliable model for humans? 2. How many times do the animal models get the wrong answer? In other words what is the sensitivity, specificity, positive and negative predictive value for genetically modified animal models. Given the fact that humans are complex systems, the differences among species in how genes are used, gene networks, and genome composition, and the empirical data revealing that human diseases like cancer are very different from the mouse versions, it is reasonable to conclude that studying genetically modified animals will be far less rewarding than studying humans.
Michael O. Hengartner states: “Even in years past, the main benefit of model organisms was not to facilitate the identification of human disease genes or even to identify the specific function of disease genes but rather to put these disease genes into a biological context. We study model organisms to understand life.”(1) I bet that mouse modelers did not say that on their grant applications. Compare this to the following statement from Trudy F. C. Mackay in the same article: "Despite the advances in identifying genetic variants associated with human disease, we must remain cognizant that genetic research in humans is a descriptive, hypothesis-generating endeavour. The resolution of human GWASs is limited by the scale of linkage disequilibrium (LD) — any locus within a large LD block associated with a trait could potentially be causal, and identifying causal variants within a gene cannot be accomplished using population data. Determining causality can sometimes be inferred from functional studies in cell lines, but more often it requires studies of the effects of mutations or human transgenes in model organisms."(1)
Derek L. Stemple, in the same article stated: “Model organism research will contribute important validation of the molecular basis of human genetic disease in both rare and common diseases.”(1) (Emphasis added.)
We have frequently pointed out that animal models can be used to formulate a hypothesis that must then be tested in humans (if one is claiming the research is human-related). We have gone on to state that animal modelers claim just the opposite—that human data must be confirmed in animals. The thinking being that if a human reacts adversely to a drug but no animal model can be found that reproduces the reaction, well, the human data must have been wrong. I have had animal-based researchers actually tell me this verbatim. When we would relate this notion, people, especially physicians would scoff, saying that no one could be that stupid (their words not mine). In the future, I will refer those who dismiss our reporting to Trudy F. C. Mackay, William Neal Reynolds Distinguished University Professor of Genetics at North Carolina State University and Derek L. Stemple of the Wellcome Trust Sanger Institute.
However, gene studies on less complex organisms, like yeast and C. elegans, have yielded many insights into human physiology. This is in part due to the fact that yeast are not being used to study diseases like cancer, for the most part, but instead are used to study conserved processes. Studying conserved processes in less complex organisms should be a fruitful endeavor because the variables are easier to isolate and the module under study is amenable to study by reductionism—not a lot of other components influencing the process.
Clearly, if one wishes to learn more about life in general, or about a specific species, then studying that specific species will be useful. That is not the same as making the claims we see above. The article ends with the following. I had not seen this in print but have been waiting for it as I knew it was coming, Stemple: "Human genetic variation can profoundly affect the prognosis of a disease or the outcome of a drug treatment. In many respects, different humans — and different human populations — can vary so much that the humans are not necessarily the best models for other humans. It is possible to ‘humanize’ model organisms. For example, researchers can generate mice that carry specific human variants of cytochrome P450 genes for testing the dynamics of drug metabolism."(1)
Needless to say, inserting one or more genes into a mouse is not going to turn that mouse into you. The entire statement from Stemple assumes animals are simple systems and that all their traits are amenable to study by reductionism. But I must admit I admire the audacity of someone that says mice are better models for humans than humans themselves. I don’t admire his character—but unmitigated guile is impressive.
1. Aitman TJ, Boone C, Churchill GA, Hengartner MO, Mackay TFC, Stemple DL. The future of model organisms in human disease research. Nat Rev Genet. [10.1038/nrg3047]. 2011;12(8):575-82.