Those of you who follow me on Twitter may have sensed a disturbance in the Force earlier in the week. The lab I work in does a great deal of in vivo work using an array of different transgenic and knockout animals. Typically we have a tech who maintains strains–mostly controls–that multiple people in the lab use. Other strains are the responsibility of the trainee who will be the primary user. About a year ago, we obtained a strain from a biotech/pharma company for a new direction of our work. A fellow in the lab was given responsibility of the strain, which was then transferred to another fellow, and ultimately transferred to me a few months ago. Coming from a biochemistry lab, I had absolutely no experience working with mice or maintaining strains, which I made clear to all parties involved and asked for help along the way. It was the understanding of the tech and other postdoc that the deletion was stable, the breeders were homozygous deficient, and thus there was no need for genotyping. So I proceeded under this presumption.
Perhaps you see where this is going.
Someone in the lab wanted to use these mice for another project. At this point, we needed to check the genetic background for the strain. There is a handful of inbred lines that are commonly used for manipulating the genome and maintaining strains. Different inbred lines are not genetically identical; in fact, the same strain inbred at different companies may not even be genetically identical. Genetic differences between strains can translate into widely varying responses in your disease model of choice. So it’s kind of important to know your background. When I asked for some sort of documentation on this strain, our tech produced a health report sent from the company, and on the report, the background was listed as FVB/N. So we bought this strain and have been breeding them for about 6 months or so.
When I started doing in vivo work a few months ago, I realized something was a little odd. FVB/N mice are white and fluffy. The knockouts were sleek and black. I brought this up with another fellow and essentially was told not to worry about it. Sometimes these things happen–maybe it’s an artifact from the backcross. Blah, blah, whatever. Moving on.
I was a planning a pretty big experiment with these knockouts this week. But that little niggling doubt just wouldn’t go away. So I emailed the fellow that initially received the mice–he had nothing. Then I started tracking back through our breeding records–and dead ended at the mice from the company. I took a closer look at the health report that had been produced previously and realized that (a) the mice on it were way too old to have been the ones we initially received and (b) they were all female. Where was the info for the males? Clearly there were males in my breeding cages because they were producing litters. With some help from our trusty lab admin, I was able to contact the company’s vet manager, who emailed me the info they had on record within an hour–informing me that the strain was on a B6 background and that they had sent us 2 males and 2 females… all heterozygous for the mutation.
That was the point that my head almost exploded. Not only did we have the background wrong, we had the genotype wrong. I promptly requested a genotyping protocol, received one almost immediately, and got to work sorting out this little clusterfuck. Fortunately Mendelian inheritance seemed to prevail, and even though we no longer had the original breeders, I still had a mix of +/+, +/-, and -/- mice. All was not lost as I had feared.
Even so, all this meant:
- I spent approximately 1.7 days sorting out this mess instead of doing other experiments I had planned for the week.
- Results from the handful of experiments this strain had been used for have to be thrown out since we have no idea what genotype those mice were.
- We bred and maintained a strain of mice for 6 months that we never needed.
- I will be waiting a minimum of 8 weeks to the experiment planned for the knockout strain.
So what’s the lesson here?
Be careful what information you trust. And be sure that you have solid documentation of what you have. Documentation and records management is something that, in my experience, most academic labs really suck at. Don’t be afraid to push a little to get documentation, even if multiple people relay the information verbally–they might know or they might simply be repeating what someone else has said. You might think that as a grad student or postdoc, you’re here to do research–it’s not really your job to track down records. Make it part of your job. It will save you tremendous headaches and lost time down the road.
Ever on & on 
Things like this happen all the time. I work only with computers, but I still work on messy biological problems. I was told “oh this doesn’t change anything.” After working on an entirely new approach to the problem, it turns out that changes things. Significantly. Of course, I wasn’t lied to, the other people involved just never bothered to check, or never bothered to really drill down and figure it out, since it made their lives easier to just say “this doesn’t change anything”.
I trusted no one in my Ph.D. It went very well. So far, I trusted a couple people during my postdoc. All that trust has been misplaced. I’m now carefully going over everything involved. Good thing I started side-projects.
I also spent all week defending one of my side-projects from being taken because a student needed to graduate. I’m sorry, I didn’t do a postdoc to give other people my hard work so they can graduate.
I know a lab that studies a particular organ that is enlarged in their transgenics, this presence of an enlarged organ is what they use to genotype when they do flow experiments. They looked at me like I was fucking crazy when I asked them do they ever go back and take some tail DNA to genotype. I guess some people are either lazy or dumb and don’t think they need to find out for themselves. Rule #1, check everything. Plasmid comes into the lab, don’t rely on their vector map, sequence it. Receive a cell line, its better be sent off for STR fingerprinting and mycoplasma checks. I trust no one but the boss, and barely because he is as paranoid about controls as I am.
Life is made extraordinarily difficult if you trust no one. But you have to learn who you can trust. And even then it’s good to find the paper trail. With what appears to be increasing scientific misconduct, research is being more closely scrutinized, making clear, reliable, organized records as important as ever.
It’s not only about *who* you trust; it’s about *what* you trust and *when* you trust. If the consequences of something not being what it is supposed to be are minor in terms of lost effort and resources, and the fact that it is not what it is supposed to be will manifest itself eventually anyway, then it is reasonable to trust. If the consequences are severe, and it might not become apparent for a long time, then it is not reasonable to trust. Essentially, you have to weigh the relative costs and benefits for trusting versus verifying.
In the case of genotyping mice, the cost of verifying is negligible: a couple hours snipping tails, setting up some PCRs, and running a gel. And the cost of a fuck-up–as you discovered–can be huge in wasted time, effort, and resources, as well as delay.
You got off easy, and learned something important. Shit like this can lose you months or years or — if you really screw the pooch — a retraction that tarnishes your reputation and pisses off anyone who ever sent you a *good* reagent.
Trust, but verify.
It was a pain in the ass, but you’re right that I got off easy. The potential of lost time, money, and resources to unreliable results was exactly why I followed up on that gut feeling that something was off. I should have done it earlier, but my inexperience led me to rely too much on other people.
Spiny & PhysioProf–you both make excellent points about who and what to trust in research. I consider it a lesson learned and hope this represents a mistake I won’t make again… at least for a while.
It’s sad how often this happens. I agree with CPP, though. Anytime I use someone else’s strain for my own studies, I spend a couple of days verifying it myself. No matter how trustworthy the person is/was, no matter how good the record-keeping was. I’ve seen too many people get screwed after years (literally) of dissertation/postdoctoral work to take that chance myself.
A similar thing happened in our lab. The tech whose job it was to take care of the mice MESSED UP and left and for awhile we were under the assumption that what we were using was to correct line. Since we were doing generic genotyping we never picked up on the problem. Not until we and our collaborator realized the experiments were not working, we went over everything reagents, protocol, how drugs were being administered etc. Still nothing was working until one day I suggested that maybe we’re not using the right mice. Long story short, 3 lines got mixed up years ago and no one picked up on it and so I spent 4 months out of my graduate student life sorting out this mess. Sadly many mice were sacrificed and we’ve bounced back and I’ve established very rigorous monitoring and genotyping of all our lines.
Excellent post. Should be required reading for all n00bs at any level. Record Everything. My personal approach is to be like Howard Hughes…
“Show me all the blueprints, Show me all the blueprints, Show me all the blueprints, Show me all the blueprints, Show me all the blueprints…”
OCD works folks.