Protocol Online logo
Top : Forum Archives: : Real-Time PCR

Housekeeping gene validation - (Jul/16/2008 )

Hi all,

Was wondering if anyone had an opinion on my problem.

I'm trying to validate my housekeeping gene (currently using GAPDH).

i'm running qPCR on two cell lines (equine oral and limb fibroblasts) treated with various treatments. i have run all treatments for each cell line with my GAPDH primers and examination of the delta c(t) for each separate cell line indicates that GAPDH expression is not affected by treatment but when you put the data for both sets of cell lines together, there is a significant change in expression between the two. ie, expression of GAPDH is not altered by treatment in either of the cell lines, but it is expressed at different levels

I know i am therefore valid in using the GAPDH to compare within each cell line (ie I can compare all oral trts to oral GAPDH and all limb treatments to limb GAPDH) but is it valid for me to compare between the cell lines using the GAPDH (which is one of the main purpuses of the experiment).

a senior lecturer within the department felt that it was probably ok for me to use it, as you normalise everything within a cell line to the appropriate gapdh values before comparison takes place, but i wonder if anyone else had any opinions, or could refer me to any papers.

if you don't think its valid to compare between the two, does anyone have any suggestions for a housekeeping gene that is likely to be expressed at similar levels in both oral and limb fibroblasts?

many thanks for your help

-dreams579-

Hi,

well I'm not a senior lecturer, only a PhD student, but to me it sounds absolutely not ok. You want to compare the expression of gene XY between two cell lines and have normalized it within each cell line against GAPDH which is expressed at different levels between those cell lines. Am I right so far? Just imagine you would have been able to do this in absolute quantification:
Cell line A you measure gene XY with 15 copies and GAPDH with 3 copies.
Cell line B you measure gene XY with 15 copies and GAPDH with 5 copies.

What would you end up with? "Gene XY is higher expressed in cell line A than in cell line B". But it's not true, the GOI is expressed at the same level, just your reference gene is expressed different.

Or did I get something wrong?

Cheers!

-littleaxt-

QUOTE (littleaxt @ Jul 16 2008, 04:33 PM)
Hi,

well I'm not a senior lecturer, only a PhD student, but to me it sounds absolutely not ok. You want to compare the expression of gene XY between two cell lines and have normalized it within each cell line against GAPDH which is expressed at different levels between those cell lines. Am I right so far? Just imagine you would have been able to do this in absolute quantification:
Cell line A you measure gene XY with 15 copies and GAPDH with 3 copies.
Cell line B you measure gene XY with 15 copies and GAPDH with 5 copies.

What would you end up with? "Gene XY is higher expressed in cell line A than in cell line B". But it's not true, the GOI is expressed at the same level, just your reference gene is expressed different.

Or did I get something wrong?

Cheers!

Hi, thanks for your reply.

any yes, thats kinda what i was feeling, and why i asked for some other opinions on the matter.

putting numbers to the problem really helped see that actually, so thanks for that. i suppose the only thing you could then do is because you know the difference in the level of expression of the housekeeping gene between the 2 samples, is to add an extra step in the calculations to account for this difference eg in the example above, if you divided the cell line b by 1.66, it would effectively take this difference into consideration

am suffering wednesday afternoon brain fry, so i may be talking a bit pile of rubbish (in which case feel free to laugh hysterically at me!), and tis probably better to just go find another HKG.

-dreams579-

... i suppose the only thing you could then do is because you know the difference in the level of expression of the housekeeping gene between the 2 samples, is to add an extra step in the calculations to account for this difference eg in the example above, if you divided the cell line b by 1.66, it would effectively take this difference into consideration


[/quote]

Heiho!

Yeah, as you could see I haven't thought as far as you. I think you could do that. Allthough I must admit I would prefer a HKG that is equally expressed in both samples, because we allways measure an error and you would also divide your error by 1.66 and make it artificially smaler by this. I'm not that good in statistics but somehow it feels a bit strange to me. I think comparing the original data is more accurat than calculating to much with your data.

Just a thought....

Good luck!

-littleaxt-

So you are dealing with cell lines? If so, why dont you normalise your GOI to number of cells?

-Pallas-

QUOTE (Pallas @ Jul 22 2008, 03:10 PM)
So you are dealing with cell lines? If so, why dont you normalise your GOI to number of cells?

yes, this would have been a good idea - if i had counted the number of cells i had extracted the RNA from! I used the quiagen RNAprotect RNeasy kit to extract the cells, and they advised against trypsinising the cells first to minimise changes in gene expression. all treatments had the same amount of RNA transcribed to cDNA, so they should all be normalised in that respect

oh well, back to square 1!

-dreams579-

No it is not OK. Test out other housekeeping genes till you find some that are stable between your cell lines. There are a few articles around these days testing out different housekeeping genes, that'll be a good place to start. Otherwise you could order a geNorm kit.

-maset-

QUOTE (dreams579 @ Jul 24 2008, 08:31 PM)
QUOTE (Pallas @ Jul 22 2008, 03:10 PM)
So you are dealing with cell lines? If so, why dont you normalise your GOI to number of cells?

yes, this would have been a good idea - if i had counted the number of cells i had extracted the RNA from! I used the quiagen RNAprotect RNeasy kit to extract the cells, and they advised against trypsinising the cells first to minimise changes in gene expression. all treatments had the same amount of RNA transcribed to cDNA, so they should all be normalised in that respect

oh well, back to square 1!


Did you spec your RNA after DNase treatement? There is a non-linear relationship for cDNA generated from RNA template when total RNA isn't kept constant.

Cheers,
Maset.

-maset-