beginners guide to qPCR for differences in gene expression - Help (Jun/01/2010 )
Hello,
I am this week starting qPCR on a Roche light cycler 480. I am looking at a gene in barley and want to see if its expression changes between treatments. I have good primers which I know work well, and I have primers for a couple of house-keeping genes as well.
The problem is I have read lots about qPCR and gene expression but I have no idea how to practically tackle an actual run. I would like to use the standard curve method to calculate relative expression, but am not really sure how to set-up a plate. What should I use to make a standard curve in this context? Could I use one of my samples serially diluted? Am also confused as to exactly how I use the standard curve and house-keeping genes to determine whether expression is different between my samples. I am throughly confused, and on the verge of giving up now. No-one seems to be able to help me and I am an absolute beginner. There seems to be loads of different approaches in the literature and I find myself very confused. Does anyone have a step-by step guide or protocol which could help me? Can anyone explain clearly how to use the standard curve approach to determine relative differences in gene expression. How for instance do I 'normalise' the expression of my samples to a gene of interest?
Any help or guidance would be much appreciated, if I just had a clear explanation I feel sure I could proceed with a run and at least grasp some of what was going on!
First think to do is check your primers (probes?) on program you choose, if you hadn't done it already. They may work on classic PCR but it may be good to check if they work well on the real-time machine and if you're running SYBR, check melting curve if you have a single product and no primer dimers.
Then I would personaly reconsider using standard curve approach as I think it's a more complicated calculation that usualy used Pffafl efficiency corrected method (actually delta delta Ct plus efficiency correction). But you can do both on LC480 and they both need same initial setting, the standard curve method then requires having standard curve in each run, Pffafl not. Both methods should have similar results.
In every case you need to perform a serial dilution curve experiment, take some cDNA you have in excess and you know it contains your genes of interest and housekeeping genes in a high numbers. You just dilute your sample (usualy 10x, 100x, 1000x) and run in replicates (triplicates) on one plate. As 10 fold dilution increases the Ct value about 3 cycles, it's important to have all Ct values within 15-30 cycles to construct a good standard curve (that's why it's good to use a sample with high abundancy of your genes for dilution). You need to do it just once, set the samples like standards in the program, with relative concentrations like "1, 0.1, 0.01, 0.001" and analyse using Absolute quantification module. That will give you Ct's and assay efficiency for Pffafl equation or concentration for the standard curve method.
So you make a plate with replicates of dilution curves for all your genes (if they fit in the plate, otherwise do some of the genes on the next plate, but all the dilutions of a single genes must be ran together).
Good thing to do if you don't know what housekeeping gene to use, is to choose the most stable ones.
You should gene samples from all your various treatments and run them on all of your housekeeping genes. You can then calculate the most stable ones using a software like geNorm. I usualy use two most stable and calculate relative quantity as described in the geNorm manual. For the calculation you will need the gene efficiencies obtained earlier. This is a third method how to calculate the relative quantity, using the mean of several stable housekeeping gene quantities for normalisation, instead of just selecting one and using the Pffafl equation or standard curve method.
Also, LC480 has an option to buy a Advanced relative quatification module, that can calculate the fold changes for you, normalizing to one or more housekeeping genes. It uses eqations mathematicaly equal to Pffafl.
Hope it's at least marginaly comprehensible, and feel free to ask any further questions.
Hello,
Yes, that's clear thank-you am sure I'll get the hang of it as I go along. I do have a few quick questions:
The concentrations of cDNA in my samples is likely to be quite different as I did not normalise the RNA concentrations I used before DNA synthesis as I was told this didn't matter. Does this matter? There may be much more cDNA in one sample than another, but am thinking as it is normalised against a house-keeping gene anyway this shouldn't matter right??
I have 6 cDNA samples to test for my gene of interest. Is it OK to just take one of these samples and dilute down to make a standard. Someone in the lab. has suggested taking a small amount from each sample to be tested and then pooling these samples together. This combined sample can be diluted down futher to create a standard curve. Does it matter how I create the standard curve??
Am going to set-up some standard curves today for house-keeping genes and genes of interest. Is it Ok to run them all on one plate:
gene of interest undiluted, 1:10, 1:100, 1:1000, 1:10000, negative control (4 replicates of this),
unknown samples 1, 2 ,3 ,4
House-keeping gene undiluted, 1:10, 1:100, 1:100, 1:10000, negative control (4 replicates etc.)
unknown samples 1,2,3,4
Thanks
phosphate girl on Jun 8 2010, 12:42 PM said:
I'm affraid the usual setup is to reverse transcribe same amount of RNA for every sample, (usualy starting with 1 ug). It's true, that if you normalise to a good housekeeping the differences should diminish, but I wouldn't recoment it. First, you can't test if the housekeeping gene is good with different cDNA concentrations and second, the range of your Ct will be wider and some of them may not be in usable range.
phosphate girl on Jun 8 2010, 12:42 PM said:
It's best to use the (expected) most concentrated or most abundant sample to make a dillution curve. By pooling samples you will use just a bit of all the samples if you don't have one in excess, but for the calculation of efficiency it's not realy needed I would say. Assuming all the samples are prepared and isolated in the same conditions and there will be no variation in the amount of inhibitors for example.
phosphate girl on Jun 8 2010, 12:42 PM said:
gene of interest undiluted, 1:10, 1:100, 1:1000, 1:10000, negative control (4 replicates of this),
unknown samples 1, 2 ,3 ,4
House-keeping gene undiluted, 1:10, 1:100, 1:100, 1:10000, negative control (4 replicates etc.)
unknown samples 1,2,3,4
If you ran your unknown samples as 4 replicates too, and I'm counting right, that should be 80 wells so it's OK.
Hello Again,
Many Thanks for your response very helpful. It's a bit annoying that I have cDNA that is all now different concentrations but a least I can use this to test my primers. I'll have to make some more cDNA when I come to testing my housekeeping genes and I have somewhere to start now!