How Much Control is Enough?
The purpose of a control is to allow you to fully interpret all of
your results. What often happens is an experiment is designed and successfully
executed only to fall short due to inadequate controls. These results go
into your notebook but will never be seen in public. You are too embarrassed
to admit you left out the one control condition that would have made your
experiment the brilliant example of the scientific method you originally
intended. However, it is possible to have too many controls; there is no
need to control for every single variable (color of your socks, for example),
only the variables which have the potential to influence the interpretations
of your results.
Below you will see some results that are just short of being publishable
(i.e. wasted effort). Your mission, should you decide to pass this course,
is to interpret the results fully. This means you should give all reasonable
explanations of the results (no need to mention phase of the moon etc.).
Then, redesign the experiment to include the controls necessary to distinguish
between all the explanations you listed for the original data. Situation
number 5 is a bit different, so you will be asked to do a different task.
- You have done a ligation of a 2.3 kb fragment into a plasmid. You know
that the gel purification worked because you ran some purified DNA on a
gel before you attempted the ligation. After the ligation, you transformed
the ligated DNA into some E. coli and plated the cells onto a LB
(standard growth medium) plate and you see a lawn of cells. There are 2
possible explanations and at least one very important control.
One possibility is that all of these cells took up the DNA (not likely).
The other is that since you plated the cells on LB, you did not have a
basis for positive selection of transformants. Therefore, the control you
should have run is a mock transformation - no plasmid DNA. You would expect
no colonies to form on this plate.
- You have just spent the last 5 days working on a Southern blot and
you very excitedly go into the dark room to develop the X-ray film. Loaded
onto the gel was one lane of molecular weight markers and five lanes of
genomic DNA from five endangered species, so you were very careful with
the samples. These species had never had their DNA probed (with a fly cDNA)
before and you were the first in the world to do it. You process the film
and stagger backwards at what you see - a blank film. You were so excited
about this blot that you had scheduled a meeting with your thesis advisor
and now you have a lot of explaining to do - get busy. Give all probable
interpretations and redesign your experiment. (The second moral of the
story is don't schedule the meeting before you see the blot.)
Since you did not have a positive control, it is impossible to determine
if the probe was good or not. So, you cannot interpret these results and
now the World Wildlife Fund is angry at you. So, next time, put some fly
genomic DNA on the blot too. This will serve as a positive control.
- You have cloned the single gene that could cure all cancers (congratulations)
and you decide to call the gene FAR (Fame And Riches). You want to show
that FAR is not expressed in normal cells but is expressed in all cancer
cells. You transfect this FAR DNA into a human cell line and immunofluorescently
label the expressed protein (you made a monoclonal antibody against FAR
in your spare time). What you see is bright labeling in the cytoplasm.
Then you look at some immunofluorescently labeled sections from normal
and cancerous tissues and you see that all cells are brightly labeled in
the cytoplasm. Interpret these results. What controls should you have included
with 1) the transfected cells experiment and 2) the tissues experiment?
You should have done a negative control when you tested you antibody
on the transfected cells - cells that had not been transfected. If this
had been labeled with your mAb, you would have realized earlier that your
mAb is not specific for FAR. Therefore, you cannot fully interpret these
result. Perhaps FAR is expressed in all cells but is mutated in cancerous
cells but not in healthy tissue (similar to ras ). Alternatively,
your antibody is not specific to FAR and binds to another common protein.
- You want to clone the mammalian gene that encodes the first enzyme
in the pathway that allows us to metabolize glucose. To do this, you decide
to utilize a yeast strain that is leu2- and is defective for this
same glucose catabolic enzyme (this mutant strain grows fine on fructose
+ leucine but cannot live on a glucose + leucine medium). You isolate human
genomic DNA and ligate it into a 2 µ yeast shuttle vector that also
has the LUE2 marker. You electroporate the ligated plasmids into
yeast and plate the cells onto a glucose plate, but no cells live. What
went wrong?! This should have worked, you know humans encode this
protein too. Explain these results and repeat the experiment with good
You forgot to do this in stages. You should have plated the electroporated
cells onto fructose plates to select for transformants, and then replica
plated these transfromed cells onto glucose plates. As it is now, you do
not know if the transformation worked or not. By doing it in steps, you
can verify that the transformation worked, and then go looking for the
gene by complimentation.
- Judging from the four examples above, it has been a difficult year
for you. You have switched fields 4 times because you have been a little
careless. This time you will get it right. With your usual good luck, you
think you have cloned a gene that will prevent the common cold (this time
you call it RID, Rolling In Dough). You cloned the yeast RID gene that
prevents yeast from catching colds. You then decide to clone the human
homolog by using the yeast cDNA as a probe to screen a human genomic library.
You plate the library, screen it and find one positive plaque. You rescreen
this phage and it is positive again. You are very satisfied with yourself,
lean back, open a brew, and reach for the phone to call the New York Times.
Abruptly, you hang up. You decide that you have not done one very important
experiment. What should you do before taking dancing lessons so you can
dance with the queen at the Nobel awards ceremony?
You have not determined if this gene is functionally responsible for
the phenotype you want. You should express this human gene in cells that
can catch colds (rid- yeast cells in this fictious world). If this
human homolog can protect the yeast cells, then you know you have gotten
it. But always verify your clones with a functional assay if possible.
Return to Controls
Molecular Biology Main Page
Biology Main Page
© Copyright 2002 Department of Biology, Davidson College, Davidson, NC 28036
Send comments, questions, and suggestions to: firstname.lastname@example.org