Strange PCR efficiencies

[hulklogan]

I was wondering if anyone can help me with this issue I've been experiencing with my PCR results. I am working with the Chromo4 RT-PCR machine, SYBR green primers, and use Opticon Monitor and Gene Expression Macro (Bio-Rad) to analyze my work. I don't typically do standard curves because my results are compared back to a control. The issue is, regardless of RNA stock used, the individual pipetting, or the primer used, the PCR efficiencies are horrendous. In fact, within the same triplicate I typically get values of 10%, 250% and 50%... all over the place. My melting curves, however, look quite tight. I have no idea how I can fix this problem and after repeating the experiment 1x10^12 times I am going insane.
Please help.

[danfive]

I used the Gene Expression Macro from Bio-Rad last year. It didn't work so I dumped it and created my own spreadsheet to normalize gene expression. For some reason macros get corrupted so you may need to dump it occasionally and download a new version.

The following quote includes the references to the formulas I used.

Originally Posted by danfive

I read Michael Pfaffl's papers on comparative gene expression and found them to be the most useful. I think you will find his formulas very straightforward.

Here are two articles I use as references:
A new mathematical model for relative quantification in real-time RT-PCR.
Relative expression software tool (REST) for group-wise comparison and statistical analysis of relative expression results in real-time PCR.

[hulklogan]

Thanks for your response.
The PCR efficiencies are determined by the Opticon Monitor program and when I import them into the Gene Expression Macro program it ignores the efficiency data and assumes 100% efficiency. I tried the REST software but I found it terribly complicated and importing data seemed impossible. Also, I think I could change the efficiency per gene but not per sample (ie. for each of the triplicates). Would normalizing the PCR data to 100% produce valid results despite the huge deviations (ie. 250% and 10%)?
Is there just a simple program/formula to convert my Ct values based on 100% efficiency and then import them into the Gene Expression Macro?
Thanks again.

[hulklogan]

I looked at the paper you suggested "A new mathematical model for relative quantification in real-time RT-PCR" and equation 1 looks like the most useful for my predicament. However, the problem with the formula is that I can only input the efficiency of the target gene but the efficiency of the control is ignored. Is there a way to input the efficiency of each the control and the sample in order to determine the ratio?
Thanks.

[danfive]

Say if your gene efficiency is 10% and the norm gene is 100%, you shouldn't try normalizing with this pair, preferably your two gene efficiencies are much closer.
For the relative quantity qPCR, you can also see Livak, Analysis of Relative Gene Expression Data Using Real-time Quantitative PCR and the 2 ^-delta delta Ct Method.

[danfive]

Here are my notes from a previous project:

Relative Gene Expression: Quantitative PCR will require using a housekeeping gene to normalize the expression levels of a target gene; this is expressed in a ratio, followed by comparison of changes in ratio against either a normal control or untreated sample.
Several ratios abound for the relative quantitation of gene expression such as 2-(Delta Delta C(T) method, PCR kinetics quantitation method, and target gene: reference gene normalization from Pfaffl. Currently the Pfaffl method is my method of choice as it normalizes target gene to reference gene expression per sample and then normalizes that with the target to gene expression in a control. The equation follows:

R = (Etarget)^[ΔCP target (control - sample)]
(Ereference)^[ΔCP reference (control - sample)]

Another method for quantitative PCR is to establish a recombinant vector (such as a plasmid) and use a standard curve method for quantitation against number copies of target gene mRNA. This works well for DNA because the vector is extracted and purified in the form of DNA, and you get one copy per vector. Obviously that is not the case for mRNA where one vector gives you many copies of mRNA; an alternative is to use a RNA viral vector. Overall this is usually seen as an artificial and flawed method because normalization to different expression levels (inherently present in all samples) does not take place.

[hulklogan]

I found out that the single well efficiencies provided by opticon monitor are useless... at least according to BioRad they should not be used the way I was proposing.

Originally Posted by danfive

R = (Etarget)ΔCP target (control – sample)
(Ereference)ΔCP reference (control – sample)

So, using this formula, if I were to compare the gene expression over different treatments, I would find the R value for each condition normalized to the B-actin and plot that data?

Thanks agai.n

[danfive]

Originally Posted by hulklogan

So, using this formula, if I were to compare the gene expression over different treatments, I would find the R value for each condition normalized to the B-actin and plot that data?

Thanks agai.n

Essentially, YES. If you use the prescribed method and calculate the R-value (relative expression ratio) you can then plot the different conditions/samples against each other.

But I have to warn you that the prescribed method means reproducing the Table 2 data (in "A new mathematical model...") for your PCR assay.
That entails:
1. Calculating PCR efficiencies for target and reference gene in a low-high cDNA range; eg 0.40-50ng cDNA input.
2. Calculating the test precision and test variability for each gene--this means performing each PCR assay on 20ng control sample cDNA, a) in triplicate for each gene for intra-assay variation, then on b) three separate runs, (three separate days) for interassay variation.
3. Perform your experiment at the high and/or low cDNA input level, eg 20ng and 4ng. All reactions in triplicate (so n=3).
4. Calculate mean CP, CV, delta CP at each cDNA input. This is exactly what I mean by reproducing that "Table 2 data."
5. Calculate the R (relative expression ratio) for the input cDNA used in the experiment.

[hulklogan]

I think I am finally understanding PCR efficiency. I just have a few final questions . . .

Do I have to do a standard curve for each pcr plate that I run? Can I simply do a bunch over the course of several days and find an average and use that efficiency value in my Plaffl equations?

I often use multiple cell lines when I do PCR . . . would there be any reason why the efficiency would be different amongst the (similar) cell lines? In other words, would I have to find the efficiency of the primer for each cell line?

Finally, can I simply do serial dilutions of total RNA used in my experiment to determine slope, or is it necessary to create cDNA first and then perform the standard curve?

Thanks. . . you have been a great resource.

[danfive]

Originally Posted by hulklogan

I think I am finally understanding PCR efficiency. I just have a few final questions . . .

Do I have to do a standard curve for each pcr plate that I run? Can I simply do a bunch over the course of several days and find an average and use that efficiency value in my Plaffl equations?.

The only standard curve I can think of, is the one used to determine PCR efficiency, by having a cDNA dilution series (fig 1 in article). You can do this beforehand, separately. And doing the series in duplicate/triplicate will allow you to calculate CV for PCR efficiency. That way you can say PCR efficiency for gene1 is 2.09 (CV=0.06%). Meaning it is basically constant as long as the major PCR components stay the same.
From there you can run samples and controls in the qPCR, and optionally include another cDNA dilution series--for further documentation of PCR efficiency.

Originally Posted by hulklogan

I often use multiple cell lines when I do PCR . . . would there be any reason why the efficiency would be different amongst the (similar) cell lines? In other words, would I have to find the efficiency of the primer for each cell line?.

Simple answer is your PCR efficiency is calculated from your control, not the samples (different cell lines). So you find the efficiency for each primer related to the control. The final RNA/DNA from cell lines or samples or controls should be equivalent in concentration and quality.

Originally Posted by hulklogan

Finally, can I simply do serial dilutions of total RNA used in my experiment to determine slope, or is it necessary to create cDNA first and then perform the standard curve?.

The article uses a two step RT-qPCR, so RT is done with random hexamer primers and a quantified input RNA (line 2, 4, 5...ng), then 1 microliter is used as template for qPCR rxn.
But if you mean do a 1-step qRT-PCR, you should be fine. As a matter of fact I did the comparison between 1 and 2-step protocols....found very similar results between them, so I settled on the 1-step qRT-PCR (Biorad iScript 1-step RT-PCR kit).