I just had my first major set-back in grad school. For the past 4 months, I have been breeding transgenic mice to study the neurobiology of autism. I realized yesterday that I’ve been going about the breeding all wrong, and now I have to start over. It’s a subtle issue, so let me try to explain.
The issue is the genetic background of the mice. The mice generally used in biology experiments are completely inbred. For a given strain of mouse, every member of the strain has essentially the same DNA.There are many strains that people commonly use – one is C57 Black 6, another is FVB, another is 129; The names come from a random fact about the mouse when it was originally inbred. The name FVB, for example, comes from the fact that a mouse was found to be susceptible to Friend leukemia Virus strain B. All of these mice have been inbred with litter mates over dozens of generations until, statistically, they are homozygous at both alleles of each gene.
For the major inbred mouse lines, there are master strains managed by Jackson Labs, a huge government-funded mouse factory in Bar Harbor, Maine (If you want a mouse of a given strain, you can order it for $15-20 and have it shipped to your laboratory in 3 days).
When mice are mutated in some way (It used to be that mutant mice were generated by feeding mice mutagens and observing their offspring for strange traits and behaviors. The two most common approaches used these days are “knock out” by homologous recombination or gene insertion by classical transgenesis), a specific mouse strain must be chosen to make the mutation. FVB or 129 are good for classical transgenesis because they have large embryos, so it is easy to inject the exogenous mutant DNA into the embryo (which will then mysteriously integrate into the embryo’s genome and create a transgenic mouse). C57 is good because it’s behavior is well characterized and it has been used for lots of experiments for several decades, so you can directly compare your results to previous results.
There are significant differences in phenotype between different inbred mouse strains. Recessive alleles become homozygous during the inbreeding process and are a permanent trait of the mouse.FVB mice, for example, have an abnormal photoreceptor gene which causes retinal degeneration. C57 mice have good memory (for mice). 129 mice have large litters. And so on. If you mutate a specific gene, it may have different effects on different mouse lines, but this is unpredictable. So the important thing is to be internally consistent – to be sure that you’re doing all your work on one genetic background, or on one combination of backgrounds. That way it can be reproducible.
We have two major transgenic mouse lines in the lab that we use. The thy1-GFP mice express the Green Fluorescent Protein (GFP – a protein from jellyfish that glows in the dark) randomly in neurons. How they made the mouse: they took the promoter for thy1 (thy1=thymus cell antigen – a protein of unknown function which is expressed in neurons and the thymus), altered the promoter so that it would only cause expression in neurons, and placed GFP behind the promoter. They then injected this DNA fragment they made into C57 mouse embryos, which inserted into DNA at random spots.
Which cells in the body express the trangene depends on where the transgene inserts into the genome, and on the promoter. A gene’s promoter (aka enhancing sequence) determines which cells will express the protein, because it has binding sites for specific transcription factors (gene expression-regulating proteins). If a cell has the right combination of transcription factors, the gene will be expressed.
Because of thy1-GFP’s altered thy1 promoter, and because of where it inserted into the genome in the mouse line we use, a small subset of the mouse’s neurons glow in the dark. So if we make a hole in the mouse’s skull and put the brain under a microscope, we can see the mouse’s neurons (Actually, the fluorescent proteins only glow when excited – a laser must be scanned through the mouse’s brain to induce the neurons to glow.).By placing a glass coverslip into the hole in the skull, we can image the same neurons over several days, and see how they grow, respond to different experimental interventions, and form synapses with other cells.
The other mouse line we use is the mecp2-Tg1 mouse. This mouse overexpresses the gene methyl CpG binding Protein 2 (MECP2) at double normal levels. This same genetic defect causes a severe developmental syndrome in humans which most notably includes autism. The mecp2-Tg1 mouse also has weird behavior
If we cross this mecp2-tg1 mouse with the thy1-gfp mice, we can see the genetically-mutated neurons in the living animal, and they are structurally abnormal. Their dendrites are much more intricately branched than normal neurons, and they form fewer excitatory synapses, among other things. It’s interesting to characterize these mice because they may tell us something about the pathophysiology of autism
Okay, that was just context for the set back I had today.
The thy1-GFP mice were generated on a C57 background. The mecp2-Tg1 mice were generated on an FVB background. So when we cross these two mice (to allow us to visualize the neurons in the ‘autistic’ mouse [Saying a mouse has autism, or is autistic, or demosntrates autism-like behaviors, is ill-advised in science. Autism is a complex human disorder of unknown etiology, and its symptom presentation depends on uniquely human characteristics like social behavior and language. Although mice with these genetic abnormalities demonstrate changes in social behavior, and some even change the way they communicate with sound, it's a grand leap to label these phenotypic changes autistic.]), half of the offspring’s chromosomes are C57 and half are FVB. It’s not as ideal as a pure background, but at least it’s reproducible – you know that for every gene, one allele is C57 and one allele is FVB. This sort of cross generates F1 heterozygotes for experimental use.
(Random facts: Mice give birth 3 weeks after conception, are weaned at 3 weeks of age, and are fertile at 5 weeks of age. The standard breeding scheme is 1 male and 2 females. Females are fertile the day they give birth. Litter size is about 8-12 pups on average. So 1 mating cage (1 male + 2 females) can easily pump out 200 mice in a year. The American scientific enterprise ‘uses’ in excess of 6 million mice per year).
The problem in our lab is that when choosing the parents for the past year’s breedings, we only took transgene into account, not the background. So we crossed mice which already had a mixed C57xFVB background. In these crosses, the offspring all have different combinations of genes from each background. It may not have any effect on our measurements, but this can’t be proven (barring doing the experiments on many different backgrounds, which would be tedious and very time consuming).It is known that different mouse strains have different behavioral phenotypes, but traits like neuron structure are probably unchanged. It’s still a better practice where possible to stick to crossing pure C57 to pure FVB.
So we will have to reorganize the breedings for these mecp2-Tg1 x thy1-GFP mice. A major delay.
But I am not using these mice for my thesis project. My thesis in brief: Given that mecp2 overexpression causes autism in humans, and causes these structural changes in neurons in the brain, an important question is whether the neuron abnormalities are due to overexpression of mecp2 during development or during adolescence. It is also important to determine if resolving the genetic defect (returning mecp2 expression levels to normal) would resolve the abnormal neuron structure and behavioral changes caused by the defect.
To address these questions, I plan to use the tetracycline-inducible gene expression system (this is complicated, bear with me). This system, called “Tet-Off” for short, allows you to control the expression level of a mutant transgene by giving the mouse different doses of the antibiotic tetracycline. How it works is that you mutate the mouse to carry two new transgenes: One is a transcription factor from bacteria which drives the expression at a specific promoter (called the tetracycline-responsive element [TRE]) when not bound to tetracycline; the other is the gene whose expression you want to control placed behind a TRE promoter. For my project, the first gene is called tetracycline trans-activator (tTA); the second is called tetO-mecp2. If two transgenic mice, once carrying tTA and the other tetO-mecp2, are crossed together, their offspring will carry both transgenes, and they will overexpress mecp2 unless given tetracycline (in the food or water).
I also want to image the neurons of these mice, so I need to get thy1-GFP in addition to tTA and tetO-mecp2 on the same mouse, which takes multiple breeding crosses. tTA and thy1-GFP are on a C57 background; tetO-mecp2 is on FVB. My original plan was to breed both the thy1-GFP and tetO-mecp2 to homozygosity, then to cross the double homozygotes with a tTA mouse to generate experiment-ready offspring, half with all 3 alleles (thy1-GFP, tTA, & tetO-mecp2), half with 2 (thy1-GFP & tetO-mecp2 – without tTA, mecp2 will not express, making these mice littermate controls). Unfortunately, this strategy includes breeding steps in which I self-cross mixed C57xFVB mice, leading to random variability in allele distribution in the offspring. A better breeding strategy as far as genetic background goes would be to first homozygose the C57 thy1-GFP and tTA transgenes, and then cross this double homozygote (which is still C57) to the pure FVB biTetO-mecp2. This strategy would produce mice which are of determinate genetic background.
After meeting with a mouse breeding expert today, I decided that, instead, I’m going to cross heterozygous thy1-GFP/+;tTA/+ mice (which are on a pure C57 background) to tetO-mecp2/+ mice (which are on a pure FVB background). This strategy saves the cross to the FVB mouse for last and allows all experiments to be performed on F1 heterozygotes. The new strategy will generate a lot of extra mice which must be culled, but this is common according to the mouse expert I spoke with. Worst, it will add a month or two to my predicted breeding period. Best, it will deal with any problem of genetic background. and it will allow me to avoid having to homozygose the two transgenes (an arduous task which requires the technically-demanding quantitative polymerase chain reaction (qPCR – a method to quantify the number of copes if a specific gene) for genotyping (=determining if an individual harbors a specific gene or mutation).
To make matters worse, I found out that my tetO-mecp2 mice are not actually on a pure FVB background. After the above decision-making, I e-mailed the collaborator who provided the tetO-mecp2 mice to confirm that they are on a pure FVB background (as described in the paper which originally used the mouse).
“Should be FVB. Is their fur white?” She replied (FVB mice are white; C57 are black).
“No, two are black and one is brown,” I e-mailed.
“That means they are mixed galore. I recommend back-crossing to C57,” She said.
So I won’t be able to completely elucidate the genetic background of my mice unless I perform a back-cross, in which a transgenic mouse is iteratively crossed to a pure inbred strain over several generations. A speed back-cross to C57, using single-nucleotide polymorphism analysis kits developed by Rich Paylor here at BCM, will take 1 year, but is not too difficult. It uses PCR to pick out the mice that are most “C57-like” in each generation and cross those again to C57. It cuts the time to attain congenic status (Congenic means approximately inbred) from 2 years to 1 year.
The most reasonable plan I’ve come across so far is to use my mixed-background mice for the imaging experiments, as it is very unlikely that reviewers will think background is important for these measurements. In the background, proceed with the back cross, and in a year perform the behavioral tests on the mice after they are congenic for C57 (reviewers for behavioral experiments care much more about background).
This is the biggest setback I’ve had in graduate school thus far (I’m a bit over a year in). It could be much worse – a student in the class ahead of me had to change her experimental model from mouse to rat after doing 2 years of experiments on mice – the department of defense was funding her research on traumatic brain injury (TBI), and they decided that only the rat was a legitimate animal model for TBI.
The strange thing is that my biggest success in graduate school also occurred today. I was awarded the grant fellowship I applied to last summer (in which I proposed the above project). I hope they don’t read this, find out about all of these problems with the proposal I submitted, and change their mind.
. This can be rearranged to be 
or more generally 
, and as 
, we have
Further, as N(g1) = population size, gΩ, we then have gΩ=BΩ−1 or gΩ=B5 in this case as Ω=6. [I]t follows that the number of families in a population scales with the branching ratio as N(g2)=BΩ−2. As N(g2)=gΩ/g2, we can write the branching ratio as B=(gΩ/g2)1/Ω−2 where family size, g2, can be expressed in terms of the net reproductive rate, R, thus g2=2(R+1). Substituting this expression into the preceding equation, we then have
