Experimental Design – Part 3 – Controls

Don’t like the narration? Scroll down for the slideshare version

or, read the text form here

Reach part 4 via this link

You are working on a newly discovered protein Ogg1 that is thought to be crucial for cell migration. Your first experiment plan involves knocking-down expression of this protein using short-interfering RNA, then introducing a scratch wound and measuring the rate at which the wound closes. For each of the types of control below, can you identify what you might use? Think of an answer before clicking the button
Experimental controls
Here you need things that show that your experiment worked the way you want it to. So, you will need to check you have knockdown of your protein (probably a western blot). In addition you will want a completely untreated cell population to show that without any treatment you get effective wound closure and something that will inhibit wound closure, likely the same as the biological control
Biological controls
For these you will want – something that definitely inhibits migration, ideally knock down of another protein so that you are using the same mechanism to have the effect you are studying. You will also want something that increases migration in case you get the surprising finding of seeing more rapid wound closure.
Calibration control
You may not need one of these here. However, it might be useful to say that different levels of knockdown cause different amounts of scratch closure inhibition. Therefore a dose response curve might be something to consider (this would change the stats approach but could give you tighter data)
Interpretation control
Here you will want something to control for the process of knocking down the protein such as a scrambled or non-targeting siRNA sequence. You will also want to control for off target effects (i.e. is it the knockdown of your target protein or the accidental knockdown of something else that is causing the issue), for this the best answer is usually a rescue i.e. knockdown your protein, add back your protein. If the rescue doesn’t at least partly restore wound closure rate you would interpret the data as being due to an off-target effect.
The connective tissue growth factor (CTGF) protein is thought to be a key mediator of fibrotic responses. You have access to mouse models where CTGF expression can be increased through injection with tamoxifen. You are designing a series of experiments where you will induce pulmonary fibrosis in these animals and compare them to wild-type mice to test whether pulmonary fibrosis progresses more rapidly in animals where CTGF expression is increased. Which of the following controls would you need, and what purpose would they serve?
Mice with the CTGF transgene injected with the solution using to dissolve the tamoxifen but no tamoxifen
Yes, this provides a interpretation control. This ensures that it isn’t any other aspect of the drug injection that is causing the effects you will be studying
mice without the transgene injected with tamoxifen
Yes, this will control for whether it is the tamoxifen injection that causes the effect or the induction of the transgene. Usually these will be the littermate controls of your experimental animals
Another transgenic line that has already been demonstrated to worsen fibrosis progression
Yes, this would show you that it is possible using this system for worsening of pulmonary fibrosis progression to occur. IF your data didn’t show any worsening in the CTGF over-expressing animals but your other model did demonstrate worsening then you would interpret the data as refuting your hypothesis. Be aware that as you are planning an experiment that involves making animals sick, you may not be allowed to include this control, you may have to rely on published results. If you were allowed to include this, likely it would be a small number of animals to verify the model system in your hands.
wild-type animals injected with with the solution using to dissolve the tamoxifen but no tamoxifen 
What does is add? It would show what effect the injection has on wild-type animals but you will already get these data from some of your other experimental conditions. So, do you need this control or is it just wasted animals? In the lecture I talked about how you would need lots of controls, but when it comes down to it, you have to be able to justify their inclusion and the value that the extra animals used will yield.

Reach part 4 via this link


%d bloggers like this: