The Mouse In Biomedical Research

Why We Still Use Mice Despite Knowing They're Terrible Models

I've spent twelve years working with murine models across pharmacology and immunology projects. The first thing I should tell you is that mice are not good models for human biology. They share about 90% of their genes with us, but the remaining 10% includes most of the immune system, metabolic pathways, and cancer susceptibility genes that actually matter when you're trying to translate findings into clinical applications. We use them anyway because they're cheap, they breed fast, and we have decades of baseline data on them. That's the honest answer to why The Mouse In Biomedical Research remains the default organism despite every good reason to switch. If you're new to this and trying to design your first study, start with the strain selection because that will determine everything else about your experiment. C57BL/6 is the standard lab mouse and it's the default for about 70% of published work. It's inbred, so you get consistency between litters. But it's also hypersensitive to certain infections, runs prone to spontaneous tumors after 18 months, and has an immune response that skews Th1-heavy, which makes it a poor choice for vaccine studies or allergy research. For those applications you'd want BALB/c instead, or maybe NOD/SCID if you're doing xenograft work because it lacks functional T and B cells entirely. The real problem people run into isn't choosing a strain, it's underestimating how much genetic drift matters. If you're ordering mice from one vendor and your collaborator is ordering from another, even within the same strain designation, you can get different immune phenotypes and metabolic rates. I had a project once where we reproduced a published drug efficacy study exactly and got completely opposite results. Turned out the commercial vendor had been maintaining their C57BL/6 colony on a high-fat chow diet without telling anyone, and the mice had subtly different liver enzyme expression. We switched to feeding our own mice the same specialized diet and the data matched the paper. You need to document everything about your husbandry conditions, and if you're publishing, state the vendor, cage density, light cycle, and diet explicitly because reviewers will skip over that section and it will bite you later.

What Nobody Tells You About Mouse Work

Here's something that comes up constantly and almost nobody discusses in methods sections. The circadian rhythm of your mice affects virtually every physiological measurement you'll take. Corticosterone peaks in the early light phase, glucose tolerance varies by time of day, and immune cell counts shift significantly between zt6 and zt18. I've seen entire grant proposals rejected because the reviewer noticed that treatment and control groups were weighed at different times of day. Just keep the clock consistent across all procedures and note the Zeitgeber time in your methods, even if your field doesn't care about it yet. It will matter when someone tries to replicate your work five years from now. Another thing that trips people up is the difference between substrate and bedding. Pine and cedar shavings contain aromatic hydrocarbons that cause respiratory irritation and elevate baseline inflammatory markers. Use aspen or recycled paper-based bedding instead. I switched our facility from pine to kiln-dried aspen and saw our baseline IL-6 levels drop by about 40% across the board. That single change resolved what we thought was a contamination problem in our cell culture work for three months. When you're doing injections, the route matters more than the volume. Intraperitoneal injections are the easiest route and most people default to them, but absorption is unpredictable and you can hit the bladder or bowel if the mouse isn't positioned correctly. Subcutaneous injections into the dorsal scruff are more reliable for controlled release studies. For IV work, the tail vein is the standard approach and it takes about six weeks of practice before you stop missing the vein and injecting into the surrounding tissue. The saphenous vein under brief isoflurane anesthesia is actually faster once you get past the learning curve because the vein is more superficial and you can see it better. I switched my lab from tail vein to saphenous and cut our successful injection rate from about 60% to 90% within a month.

The Limitations That Will Break Your Study

Mice are not miniature humans. This sounds obvious but I've read too many papers where the authors make claims that go well beyond what the data actually support. The drug metabolism difference between mice and humans is substantial because the cytochrome P450 isoforms are distributed differently. A compound that shows clearance half-life of four hours in mice might have a half-life of twenty-four hours in humans, or vice versa. If you're doing pharmacokinetics, you need human-relevant dosing calculations, not simple allometric scaling from body surface area. The FDA guidance documents cover this but most researchers don't read them. Immunology work with mice has its own set of problems. The mouse Major Histocompatibility Complex, the H-2 complex, is far less diverse than the human HLA system. This means that when you're testing a vaccine or therapeutic antibody, you're essentially testing it against one or two haplotypes depending on your strain. A drug that works in C57BL/6 might fail completely in a human population because the epitope isn't presented by common HLA alleles. I've seen this happen with a cancer immunotherapy candidate that showed strong tumor rejection in mice and then failed phase II because the target antigen wasn't immunogenic in the patient population being studied. Xenograft models have their own issues. Human tumor cells grown in immunodeficient mice will grow, but they adapt to the murine microenvironment over time. Passages beyond P3 tend to show drift in growth kinetics and gene expression profiles. If you're doing a long-term study, keep your passage number low and freeze early-passage stocks. I recommend not going beyond P2 for any experiment where you're measuring tumor response to treatment because the adaptation can look like drug effect if you're not careful.

There's also the issue of microbiome variation between facilities. The gut microbiome affects immune development, drug metabolism, and even behavior in ways that are still not fully characterized. If you move mice from one facility to another, expect physiological parameters to shift over the first two weeks. I had a colleague who shipped mice from a university core facility to his lab and his baseline inflammatory markers jumped 3x for ten days before stabilizing. He attributed it to stress initially but it turned out to be the microbiome difference. Quarantine and acclimatization periods matter more than most protocols account for.

Alternatives Worth Considering

If your research question doesn't specifically require a mammalian system with adaptive immunity, zebrafish larvae offer a compelling alternative for high-throughput screening. They're transparent, they have adaptive immunity by about 7 days post-fertilization, and you can do whole-organism drug screens at costs that are a fraction of mouse work. For studies focused on human-specific immune responses, organoid systems are improving rapidly and some groups are now producing lymphoid organoids that mirror human germinal center reactions. They won't replace mice for systemic studies, but for certain questions they're actually more relevant than mouse models. The computational approaches to replacing animal models are advancing but they're nowhere near ready for predictive toxicology or complex disease modeling. The EU has been pushing for reduction and replacement under REACH for years and while there's been progress on in vitro methods, regulatory agencies still require in vivo data for most new chemical entities. That's not going to change soon. The best approach right now is to design mouse studies that maximize the translational value of each animal used, which means proper power calculations, randomization, blinding, and reporting that meets ARRIVE guidelines. Most papers still don't meet those standards and that's a problem for the field as a whole. If you want to learn more about current practices and get access to standardized protocols, the Jackson Laboratory website has detailed strain descriptions and husbandry guidelines that are freely available. The Mouse Phenotype Database at MGI is also useful for understanding what's already known about specific gene knockouts. These resources are maintained by people who actually work with these animals daily and the information is generally more reliable than what you'll find in review articles, which tend to oversimplify the practical challenges.