Welcome to Science with Shrike! Today we will look at one area of research that is sometimes controversial—using animals in research. This will be an overview of why, what types of animals are already used, and which ones might be best for learning something new.
First, some questions about animal work. If you’re already sold on animal work skip down to the next section where we discuss model choice.
Why use animals in research at all?
Can’t we just use computer modeling? Well, if the SARS-CoV2 pandemic predictions did not reveal how poorly computer modeling works to you, I’m not sure what to tell you. The models were completely wrong, even though we have a good idea of how coronaviruses spread and factors that influence airborne disease transmission.
But we didn’t know about SARS-CoV2 specifically! Exactly! This is the biggest reason we need to do experiments, including in animals. We don’t have the data to accurately model with computers how humans respond to a wide variety of challenges and substances. Where we can model, we do, but we usually need to check the work.
What about cell lines, organoids, etc? These are all useful tools. The problem is that cell lines do not have organ system or tissue organization. This limits what can be done to study problems of tissue and organ structure/function. Organoids are still a very long ways from being useful.
But some experiments in animals fail to translate to humans, so why not start with humans? Because it’s no longer the 1950s and you can’t give people lethal pathogens like Fransicella tulerensis even when you plan to cure them with antibiotics anymore. FWIW, the antibiotics approach worked; it was the other bioweapons we were developing that killed a few of the human test subjects. Human research has a wide range of ethical issues that are different from animal use, which has also evolved over time. We may cover that in a separate post. Suffice to say, current ethical constraints limit the types of experiments you are allowed to do in humans, and tissues you are able to collect. Also for drug development, you need two pre-clinical models (= different kinds of animals) before bringing an intervention into humans. Even without these standards, humans are large, and clinical grade pharmaceuticals are expensive. A 20 g mouse or 8 kg macaque need a way lower amount of drug than a 60 kg person.
But it’s cruel! Cruelty is inflicting needless suffering on something. Lab animals live just about their best life up until they are euthanized. A mouse in the wild is lucky to make it 3 months, and lives with plenty of stress. Mice in the lab make it 1-2 years, with all the food and water they can eat, plus enrichment. They live in the mouse equivalent of a 5-star hotel, and have full medical care. The downside may depend on protocol, which is still required to monitor for and minimize pain and distress wherever possible. An Institutional Animal Care and Use Committee (IACUC) monitors animal use to manage animal welfare. This committee is required to have community members on the panel. Animal facilities are also inspected by a third-party, AAALAC. Animal use needs to be justified (explain why a simpler model cannot be used, why you need the numbers you do, benefit of the research) AND conducted to minimize pain and distress. If anything, the laws require researchers to take better care of the animals than themselves.
But animals have emotion/feelings/consciousness! Since this topic is beyond the scope of today’s discussion, and the ‘no’ response resolves this objection, we’ll briefly consider stipulating this to illustrate the problems. We’ll also consider something higher up the food chain, like a cat or a pig to avoid controversies over nematode feelings. Let’s take cats. Feral cats murder billions of wild animals every year This isn’t just for food… cats, even well-fed house cats, traumatize and “play” with their prey prior to killing them. If we’ve stipulated that they have emotion, feelings or consciousness, this makes cats heinous murder-beasts. Anything we do to them, they deserve for the suffering they want to inflict on other animals. If it’s just in their nature to be heinous murder-beasts, why will we treat them better than human psychopaths, for whom it is also ‘in their nature’?
Ok… so what about something that isn’t a predator… like pigs. You mean like feral hogs? In contrast to cats, which are usually illegal to kill in the US, feral hogs are such a problem that you can gun them down from a helicopter or even a hot air balloon in Texas, with very few limitations. And with good reason. They are devastating to the environment, and drive out or kill local flora and fauna.
As it might be apparent from the above discussion, if we stipulate animals have feelings, we run into serious moral problems. Since the animals show zero behavioral traits associated with remorse when engaged in these activities, and repeatedly do it, it becomes clear that they enjoy doing awful things to other animals. This requires us to pronounce the moral judgment of evil against them because they engage in animal cruelty. Animal research becomes justice in this sense, especially considering the lack of cruelty, and the research improves the lives of others. This is why Shrike does not typically stipulate animal emotions/feelings/consciousness… it does not end well for the animals, even without getting into forced matings (some female ducks have labyrinthine reproductive organs to make it harder for male ducks to successfully fertilize their eggs if the female doesn’t assume the correct position). Instead, it is best to try to avoid anthropomorphizing animals. Deal with them as they are, and love your pets for what they are. For cats, this includes keeping them indoors.
Model Systems
So how do we model human health, and what makes a good model system? We have a wide range of options to choose from. How do we choose? Good model systems share certain traits. First, they need to be easy to grow and maintain. Second, they need to be genetically tractable. That means we can add and delete genes easily. We also need strict control on genetic variation. This means that every mouse we use is genetically the same as every other mouse we use, except for sex. Or every bacterium has the same genome as the others. This way we can tell that those genetic modifications we make are the cause of the change, instead of random differences in the population. Third, we need to have a lot of reagents for the model system in question. These reagents include things like validated PCR primers, and especially antibodies that recognize proteins in the model system. If you don’t have reagents, it is hard to use the model system. Coupled to the reagent problem, we also need validated assays that measure specific aspects of biology. Certain behavioral tests can measure memory in mice, but those tests might not work in frogs.
With all of these parameters in mind, we can cover the major model systems, what they are used for, and their limitations. We’ll start from organisms more phylogenetically distant from humans and then get closer to humans.
Bacteria
Bacteria need to do some of the same things humans do, and since these organisms are often causing disease, learning about them for their own sake is also valuable for human health. Given the wide range of pathogens, and ease of culture, many bacteria are studied as models. There are two major ones we’ll discuss. First, Escherichia coli (E. coli) is the best developed model. We have modified and worked with these bacteria to create strains that do almost everything we need. This means E. coli is foundational to biotechnology (enslaving the microbes to do our bidding), in addition to being a research tool for learning about cell growth and survival. Lab E. coli used for these purposes are not the pathogenic ones you read about in food poisoning cases or urinary tract infections. On the research end, Bacillus subtilis is one of the model Gram positive organisms. It also forms biofilms, and is non-pathogenic, which is why people like to use it as a model system. A lot of seminal gene regulation and metabolic discoveries were made in bacteria, and E. coli function is taught in undergraduate college genetics and microbiology courses. Bacteria genes have a three letter phenotypic description (shared by all genes in a cluster called an operon) followed by a letter. For example VanA is one gene in the Vancomycin-resistance operon. VanB is another. They are usually related by pathway instead of function. Thus, LacZ breaks down milk sugar, while LacY gets the milk sugar across the bacterial cell membrane.
Strengths of bacteria are the rapid growth, ease of transformation (sticking genes into them), and ability to select for inserted genes or plasmids using antibiotics. They are also easy and cheap to culture. One major drawback is that they are prokaryotes. This means they lack intracellular organelles, including a nucleus. They just have one long circular chromosome for their DNA (plus mitochondrial DNA). In contrast, human cells are eukaryotes, with organelles like a nucleus, and their genome is many linear strands of DNA (plus mitochondrial DNA). A second drawback is that they are very distant from humans, so even common tasks (duplicate DNA, make proteins) are done differently from humans. Finally, E. coli produce toxins that complicate their use for clinical purposes.
Yeast
Saccharomyces cerevisiae may truly be man’s best friend (sorry dogs). S. cerevisiae is baker’s and brewer’s yeast, so this organism has been making delicious carbs for us for millenia. We’ve also turned it into a model organism. Unlike bacteria, S. cerevisiae is a eukaryote, so we can use it to learn how cells organize their insides. While slower than bacteria, it is still very fast to grow. It is very genetically tractable, so genes can be added and deleted easily. You can tell what genes were identified in yeast because they have three letters standing for a phenotype, followed by a number and/or letter. For example, Vps4B is vacuolar protein sorting phenotype 4B. Kex2 comes from killer expression 2. These are roughly grouped by pathway, but it’s not hard and fast.
The strengths of yeast are similar to bacteria, except they are also eukaryotes, and do not make the primary toxin E. coli does. They do make other inflammatory things, however. While they have similar pathways to humans, they are still single celled organisms. This makes understanding specific cell types harder, let alone tissues and organs. They are also very small, making imaging them harder than mammalian cells.
Arabidopsis
Arabidopsis thaliana is the model plant. It’s small, grows relatively fast, and is genetically tractable. Plants also get very weird very fast, so it’s helpful to have something with more normal genetics. It’s not directly relevant to human health, but it’s a system you need to know if you get into plant biology.
Worms
Caenorhabditis elegans is a nematode that is one of the simplest model animals. It can reproduce in 3 days, it is genetically tractable, optically transparent, and it is a very simple multi-cellular organism with neuronal connections. The fate and development of every cell (1031 in males, 959 in adult hermaphrodites) in C. elegans is known. Similarly, the neuronal wiring is known for this organism. This makes it very easy to find defects in these developmental pathways and neuronal connections. Gene names are similar to yeast. For example, ced-6 is cell death protein 6.
The strengths of this system are fast generation time, knowing the fate/connection of all the cells, ease of genetic manipulation. Limited behavioral assays can be performed in them. The drawback is that some genes and pathways are absent in them. Other pathways get more complicated in humans as well.
Flies
Drosophila melanogaster, or the fruit fly is one of the most utilized model systems. They develop organs, are relatively easy to rear, and many phenotypes can be characterized from mutagenesis screens. Their development is well studied, which includes the complexities of limb, head and other organ development. They are genetically tractable, allowing for gene silencing or overexpression. They also have the most creative gene names. If it sounds like someone on CT named it, it was probably discovered in Drosophila. For example, bazooka, stardust, liquid facets, sonic hedgehog are all protein names. When their homologues are discovered in humans, they often retain the Drosophila name. For example, Toll-like receptors (named after Drosophila Toll protein), or human ether-a-go-go (mercifully abbreviated to hERG). “Ether-a-go-go” refers to the way flies swayed when they were knocked out with ether. It turns out this is a critical cardiac channel in humans, much to the human gene namooors’ dismay.
The strengths of fruit flies are their faster generation time and genetic tractability. They are cheap and easy to rear, which means they can more easily be used in undergraduate genetics labs. There is also over 100 years of work on them, so as a system they are very well understood. The primary weakness is that they are not mammals, so they lack genes and organ systems associated with them. They also lack an adaptive immune system.
Zebrafish
Danio rerio, the zebrafish, is another model system growing in popularity. The embryos are optically transparent, which makes observing their development easier. They are the simplest vertebrate system, yet advanced enough to have an immune system. They can be genetically modified, which makes them useful for understanding vertebrate development. Their generation time is relatively short, and some behavioral assays can be done with them. Their primary weaknesses are that they are not mammals, they are aquatic, and they had a gene duplication event in their history. This ‘teleost gene duplication event’ occurred in bony fishes (teleosts), and complicates identifying human homologues. For certain genes in zebrafish, there are two instead of one in humans and mice.
Mice
Along with fruit flies, mice are one of the best used model systems. Mice are the simplest mammals, so they have many similarities to humans that other model systems lack. Through the development of inbred mouse lines (every mouse in the line is genetically identical to the other of the same sex), and mouse genetics, they are genetically tractable organisms. Many behavioral assays can be performed, and they can be used to model many human diseases. Mice have been “humanized” in many ways to make them serve as better model systems without needing complex organisms. Mice are especially used for immunology because the immune system can be destroyed without killing the organism, so long as they are kept in pathogen-free conditions.
The strengths of mice is their widespread use and numerous tools and assays available. Their genes are similar to human, so discoveries in mice are more translatable than lower organisms. Genetic approaches are straightforward and widely applied. Their gestation time is the main comparison for other organisms (hence all the ones above were “fast”). They can be reared in reasonable numbers, and are susceptible to most of the same diseases humans are. There are a few notable exceptions, and/or differences (tuberculosis, influenza and atherosclerosis for example) that require consideration. The weakness is that they are not primates, and there are key differences in biology. They have a short lifespan, so diseases that take 10+ years to develop in humans are often poorly modeled in mice.
Non-human primates
Non-human primates are a model system used primarily for end phase therapeutic testing, and certain infectious diseases. They are extremely expensive, grow slowly, and are not genetically tractable. However, they are the closest we can get to humans without doing experiments in humans. This is their primary strength. There are few reagents for them (most people have to validate human reagents in the NHPs), but more than for cats and dogs (mostly due to cross-reactivity with human reagents). They are outbred. This means each animal is genetically distinct. This more closely mirrors the human population, but makes experiments messier.