relative quantification - things to do before the actual qPCR - (May/17/2016 )

Hope someone can help me with planning of the qPCR assay.

 

I need to compare gene expression levels in a wild type strain and in a strain expressing a heterologous protein.

I have 4 samples for each strain (different timepoints) and in each sample I'm analyzing the expression of 7 genes (housekeeping genes, genes involved in the secretory pathway and for one of the strains - a heterologous gene). So that's 168 qPCR reactions (3 technical replicates for each sample/gene combination).

 

I've synthesized cDNA's but not really sure what to do next.

According to the literature, I need to do serial dilutions of the cDNA and determine the concentration that will give me Ct between 15-25. Is it a Ct of the housekeeping gene or do I need to do it for each gene in my experiment? Do I need to do it for each sample or just for the one of them and assume that the rest will behave in the similar fashion?

 

I'm planning to use delta-delta Ct method; so, I need to determine the efficiencies of the normalizer gene and the target gene.Do I have to do it for each sample and for each target gene? Most of the primers were designed by my colleagues and have been used before, so I know that they have similar efficiency.

 

Does anyone have any ideas about simplifying my experiment? Now it looks like I have to spend more time and reagents on the preliminary reactions than on the actual qPCR.

Serial dilutions are not very useful if you run multiple genes. You should however check if your Ct is in that level, and adjust primer concentrations perhaps, but keep in mind that 3,3 Ct is around 10-fold difference. Usually when the genes are very low abundant, you may not get to those values unless you have a very efficient RT enzyme, if your protocol tells you, you should always dilute cDNA, then it is likely one of those.

 

Housekeeping genes are mostly quite abundant, especially 18S (but it depends, your housekeeping gene should be close in abundancy to your target genes), and you expect them to be kind of similar for all your samples, so testing housekeeping is a good start.

 

But my workflow would be different. First I would test all primers/probes/whatever are working, on their own (you can skip that, since they worked for your colleagues, but I double check everything I get from someone, several times I was pretty surprised, not in good). You can use any cDNA that you expect to express the gene in question. Usually, for the tests, you make more of some cDNA to do testing, so it won't disturb the set of RNAs you isolated together (they should not differ in treatment up to the reaction, idealy).

 

Then I would check the efficiency of each gene. If you know the efficiencies from some other experiments you can skip that too, this doesn't change that much. As far as I know, in delta-delta most people use just 2 anyway. But knowing efficiency is generaly good thing, you can again use any cDNA that amplifies for each gene.

 

Then finding the dilution.. if you did the prior steps, at his point you roughly know how much to dilute. You keep the dilution same for reference (normalizer) gene and target genes, so just check all are in.

If not, use one and 

 

These things you can do in duplicates, or less, they don't need to be that precise.

The other really depends on how big you expect the differences to be, but logically if your wt is 16 say, and the experimental sample runs 27 on the same gene, then you just can't move concentrations to fit all the span. And sometimes some genes are negative (not expressed at all). The heterologous protein you have, not sure what you plan to compare if it possibly is not expressed at all in wild-type? 

 

If you have a cDNA prep, that is high-efficiency and thus is always diluted, you just need to find the right dilution. If you want to save reagents, maybe consider checking data on previous uses, what Ct those housekeeping genes had, if some of your genes are particulary high a very little abundant, and try to fit those most outlying into right range. This is diffucult to estimate since it differs in each experiment, but try to select say two samples that you expect to be most differingin all genes, and run them in duplicates for full set of genes (in dilution you already decided from previous tests or just guessed). That can tell you most. If all fits, in both extremes (housekeeping shoud generaly be the same) the rest would be possibly fine.

 

But more important I see planning your experiment. 
What do you want to compare with what? All samples compared with each other shoud fin in one run. But not all genes need to be in the same run, just all the samples. In your case all 8 samples+negatives, so if I count right you can fit 3 genes for all samples on one 96well. 

One sample in each set is a calibrator, that others are normalized to in each set. In such setting of timepoints and two strains, you may have multiple normalizations variants, depending on what you need. You can have same timepoint of wt strain normalized to second strain - changes between strains regardless time (4 sets with 4 calibrators), or zero timepoint of each strain as a calibrator to the rest - progression in time in each strain (2 sets, 2 calibrators), or you set zero of wt as a calibrator for them all to see changes. You can do this all, this doesn't matter, as long as all samples fit in one run you can do that (having different genes on different runs doesn't matter, their inter-run differences will be normalized by the reference gene). 
This is often overlooked problem of initial design, that can't be fixed after you run it. 

One important thing about calibrator in any set. Due to the nature of the delta-delta calculation, calibrator cannot have zero expresion (or some very bad one, high Ct), say you may have a set of timepoints gradually increasing expression of a gene, but in zero timepoint it is expressed very low. this sample you should not select as calibrator, it would skew the other samples fold changes, if the Cts are extreme, they may be less precise, but only for that extreme value. If the extreme value is in calibrator, it's less precise for all.

Thanks a lot for your reply! Things are getting a bit clearer now :)

 

One goal is to compare the heterologous gene expression in a mutant strain between different timepoints. So that would be plate 1 - housekeeping genes and heterologous gene in 4 samples.

 

The second goal is to compare the secretory pathway genes in wt and the mutant strain (the theory is that the overexpression of the heterologous protein affects the secretory pathway). So it would be all genes for the same timepoints on one plate or all samples/3 genes on one plate and the rest on another plate.

I was actually thinking about using 384 well plates to fit all samples in one or two plates so that I don't have to duplicate some reactions.

 

If I'm getting it right, it's ok to have different dilutions for the different samples as long as Ct values for the normalizer genes are within the 20 - 30 range?

I see. There is no point in running the heterologue on the wt stain then, it would not amplify.

 

Fot the rest, if my calculation is corrct, once you can fit all samples per gene on one plate, you don't duplicate anything.

At most you have just 8 (+neg) samples x 3 = 27.  That is three full genes on 96-well. Next three genes you put on a second plate, and if my calculation is correct, the heterologue will only run on 4 (+1) samples x 3, which is 15 and that just fills up our second plate.
No need to do a 384-well.

As I said, all genes do not need to fit in the same plate, inter-gene differences are normalized anyway by the housekeepings, inter-sample differences cannot be, if they are in different runs.

 

If you have a set of say a 36 samples+neg (different strains etc. but you want to compare them together), then they would not even fit on the 96 plate/per gene, in that case, calibrator sample needs to be run twice, on first plate with some of the samples and also on the second one with the rest of the samples (considering this for a just single gene). All samples on plate 1 will be normalized to calibrator 1 and all samples on the plate 2 with calibrator 2. But since both calibrators are identical samples, the ratios you can then compare among all these samples.
I hope it is more clear now.

 

But maybe I would consider starting (if this is your first qPCR) separately with the heterologous gene experiment, and then the other. That would of course need to duplicate the housekeepings for your mutant strain, but if you for example have no prior experience with pipetting so large set of samples (and one plate just after the another) I would probably start will a small-scale. I would not personally go for 384-well unless I have a robot, but with my attention-deficit it is probably a personal preference ;)

 

Yes it is OK, to have different dilutions if you use the same dilutions for both your housekeepings and target genes, if you need to keep them in the same level. But that is unlikely to happen. RNA amount (measured after DNase treatment) to RT-PCR reaction should be same for all samples to minimize differences in RT efficiency and then you should put the same amount (same dilution) of cDNA to each reaction, again to minimize variability. It's not good at all to have e.g. 100x dilution for target gene and 1000x dilution for a reference (housekeeping), you don't compare the same sample anymore, different original dilutions are never accurate enough to keep the identical ratio. Then the housekeeping genes should normalize for the rest and for slight error is concentration measurement/inhibition of PCR/... (still, if your selection of houseekings was good, they should be all very close in Ct).

So in this case if you kept concentration identical, it is not really possible that housekeeping Cts would differ accross samples. So unlikely you would need to have e.g. 100x dilution in sample 1 and 50x in sample 2. But you may need to adust all the dilutions depending on your most/least abundant gene. Ct 15-30 is usually a gold standard.