The Best-Laid Schemes of Mice and Men.

We’ve domesticated animals for as long as we can remember. When you think of domestication, you probably think of companion animals like dogs and cats, work animals like horses and oxen, and meat animals like chickens and cows. Each of these species can be found in many shapes and sizes, due to millennia of selective breeding that strengthened the traits we as humans found useful or desirable.

Today's discussion, though, focuses not on companion, work or meat animals, but on the house mouse, Mus musculus, the most widely-used mammalian genetic model organism. A model organism is a species widely studied due to its easiness to maintain in lab settings and, in some cases, biological quirks that facilitate investigation. While mice have been bred as pets since at least 1,100 BC, genetic studies using mice began at the turn of the 20th century, and the first full-time laboratory mouse repository, The Jackson Laboratory, opened its doors in 1929. Why, you might wonder, would scientists want to work with mice? Mice are a nice compromise between ease of use and relevance to human physiology: they reproduce more quickly than, say, primates, decreasing the time and costs required to do experiments, but feature biological processes more analogous to those in humans than, say, fruit flies.

More importantly, however, over the past few decades, new genetic technologies have allowed neuroscientists to generate mice that express marker proteins or lack genes in specific cell types. These technologies allow us to ask how specific genes, and the molecular signaling pathways they operate, impact neural function and behavior, uncovering precise mechanistic insights into the workings of the nervous system. Additionally, as scientists, we want to control as many variables as possible when setting up experiments. Some mouse “strains” have been inbred under laboratory conditions for so long that individual members of a given inbred strain have almost identical genomes. Scientists thus often choose to study a single strain, allowing them to more equitably compare results obtained by different teams at different institutions with different animal housing facilities.

However, precisely because mice have been bred in laboratories for so long, there are some experimental caveats to consider. There are at least dozens of widely-used inbred strains, each with their own behavioral eccentricities. Some strains are very aggressive; some are very docile. Some are more attentive parents; others consistently cannibalize their pups. This raises a critical question: Can we compare results obtained using different strains? And if not, how can we hope to find generalizable results that would, down the line, reliably apply to humans?

Additionally, some strains have been bred for over 110 years (about 600 generations!) in lab environments, which are very different than the environments of wild mice. Lab mice have all the food and water they want with no need to forage and no predators to fear. Furthermore, human caretakers participate in selection processes, for example, by culling overly-aggressive mice, thus weeding out aggressiveness over the years.  However, we’re ultimately interested in how biology operates outside of the lab—the lab simply gives us a relatively controlled environment in which to isolate effects of specific variables on specific readouts. Can inbred lab strains effectively model naturalistic behaviors? The two papers we discuss here today address the two bolded questions.

Sittig et al. addressed the first question we posed: Does the strain of a lab mouse affect conclusions drawn while studying it? Specifically, do studies using different inbred strains produce consistent results? This question is especially important in disease modeling, where disruption of one gene can lead to multiple downstream observable effects called phenotypes. To address this question, they studied the effects of two “disease” genes: Cacna1c, which has mutations implicated in bipolar disorder and schizophrenia; and Tcf7l2, which has mutations associated with type 2 diabetes, bipolar disorder and schizophrenia. The scientists started with male mice of the B6 strain, a widely-used inbred strain, lacking a functional copy of Cacna1c or Tcf712. To obtain disease gene-bearing mice of different genetic backgrounds, they crossed these males to females from a total of 30 different inbred strains. Each cross produced either heterozygous (“HET”) offspring with a “null” copy of Cacna1c or Tcf712 from its father plus a functional copy from its mother, or wildtype (“WT”) offspring with functional copies from both parents. Crucially, each of these mice had a 50% B6 genetic background from dad and a 50% background of some other type from mom (Fig. 1). The scientists then recorded the effects of each disease gene by measuring several characteristics like anxiety and blood sugar levels in HET progeny and their WT littermates, and, most importantly, asked whether these differences were found in all genetic backgrounds.

Their findings are a sober reminder that comparing across strains can be like comparing apples to oranges. They found that genetic strain strongly impacted the effects of Cacna1c and Tcf7l2 gene deletions. In some cases, differences between WT and HET siblings were present in some strains but not others. In a minority of even more puzzling cases, gene effects on specific phenotypes were entirely reversed between different strains, indicating that the strain of a mouse can affect what symptoms a “disease gene” causes. Finally, as the mice studied were on 50-50 mixed backgrounds, one can only imagine the enhanced variability that might have been observed had all studies been conducted on mice with pure genetic backgrounds.

Figure 1. Comparing the effects of Cacna1c and Tcf7l2 gene deletions on multiple inbred lab strain genetic backgrounds. From Sittig et al., 2016.


So now we know that different inbred strains may react differently to loss of the same gene. Let’s address the second question: How reliably do lab strains behave like wild mice? To approach this question, Chalfin et al. repeatedly mated the offspring of wild mice caught in Idaho to lab mice lacking TrpC2, a gene necessary for the sensory function of most pheromone-detecting olfactory sensory neurons. This process yielded mice with a wild mouse background but a lab mouse-derived genetic deletion (Fig. 2). As pheromone chemicals mediate intra- and inter-species communication, you’d expect TrpC2-deficient mice to display specific defects in social behaviors. Indeed, inbred lab mice lacking TrpC2 are reported to display such defects. But what about wild mice lacking TrpC2? Additionally, are there any baseline behavioral differences between wild and inbred lab mice?

Interestingly, Chalfin et al. observed large differences in phenotypes between wild female and lab female mice, but not between wild male and lab male mice. This finding could suggest that behavior of female mice is less genetically hard-wired than that of male mice and/or that laboratory selection pressures just happen to exert more force on female behavior. For example, at baseline, wild females displayed both female-female and female-pup aggression, traits almost never observed in lab strains. Deletion of TrpC2 reduced both female-female and female-pup aggression, indicating that the pheromone system plays a role in driving these behaviors in wild mice. However, because inbred lab strains do not display this behavior to begin with, its circuit and molecular bases have not been studied at all. How many such traits exist in the wild but have been ignored due to the almost exclusive study of inbred lab strains?

Figure 2. Generating TrpC2 knockouts on a wild mouse genetic background. From Chalfin et al., 2014.


There are a few lessons to take home. Findings from Sitting et al. demonstrate that it can be dangerous to make conclusions based on studies performed on single inbred strains. It may be safer, especially when studying complex behavioral traits, to perform studies in parallel on multiple strains. Findings from Chalfin et al. demonstrate that the behavior of lab mice has indeed been strongly impacted by many generations of laboratory selection. As wild and inbred lab mice feature distinct behavioral repertoires, then perhaps the neurobiology and behavioral repertoires of wild mice should be studied more often. If neuroscientists studying mice take some of these lessons and caveats into consideration, then maybe one day we will have a better understanding of both mice and men.