Tom Anderson via (by ucgatan from
Fri Feb 1 16:09:00 EST 2008

On Thu, 31 Jan 2008, Yvonne Couch wrote:

> I have another (relatively stupid) question.  This is definitely not my
> education but my own brain not being able to understand the problem.
> I've just started using the MTT assay to look at the toxicity of two
> different compounds.  I get how the assay works, mitochondrial reduction
> of the salt to formazan which forms purple crystals, more crystals =
> more mitochondrial activity.  But I apparently need to run a standard
> curve with increasing cell numbers in each well.  I'm not really sure
> why.  I've run it and I get more purple as my cell number increases but
> surely this is just because there are MORE cells and therefore the very
> nice line graph I have doesn't really tell me anything except that more
> cells make more crystals?

Exactly. I've never done an MTT assay, but i think it's just a really
roundabout way of counting cells: if you know how much purple colour you
get from ten thousand cells, then when you measure the amount of purple
you get from your experimented-on cells, you can work out how many there
are. It's a bit like counting a crowd of people by taking off their socks
and weighing them.

Except you can't actually come up with a real number, because you don't
actually know how many cells you have in the control, because you seed
them one day, and measure two days later. All you can say is that there
are X% as many cells as in the control. Or rather, X% as much
mitochondrial reductase activity, which could correspond to a change
proliferation rate (affecting cell number), growth rate (affecting cell,
and so mitochondrial, mass), regulation of mitochondrial replication or
growth (affecting the mass of mitochondria per unit of cell mass), or the
level of expression or activity of the reductases. I think if i had access
to a Coulter counter, i'd be using that instead, and getting direct
measurements of cell number and volume.

Or is the point to make a measurement which involves mitochondrial
parameters, and so reflects cell health? Or is this just for people who
have plate readers and are determined to use them?

> The proliferation rate of the cells, no matter what their number, should
> essentially be the same?

Yes, until they get confluent, or use up their nutrients, or soil their
medium or whatever and start growing more slowly.

> My next query is that after running my standard curve I'm supposed to
> pick a cell number and use that in all the assays.  I have picked 10k
> cells per well since it was nicely in the middle of my graph but on
> different days it has not given the same reading when untreated (0.3
> absorbance one day 0.6 the next)

Yep, that sounds like cells to me.

> so if I treat the cells with a compound how am I supposed to know if
> it's doing anything to their rate of proliferation?

You need a standard every time you do the experiment. I'm not sure if you
need a whole standard curve, or just a single control well, where you
haven't added a test compound. The latter should be fine if the assay's
linear. Maybe three control wells, to be sure.


Tom Anderson, MRC Laboratory for Molecular Cell Biology, UCL, London WC1E 6BT
(t) +44 (20) 76797264   (f) +44 (20) 76797805   (e) thomas.anderson from

More information about the Methods mailing list