Hi Aaron,

If you add 400 ng RNA/20 uL RT reaction, and assume 100% conversion to cDNA, your final cDNA samples are 20 ng/uL and not 40 ng/uL. RNA does not participate in the qPCR (but left over RNA could sequester your reverse primer). The RNA is either degraded (by the RNase H activity of the RT enzyme) during RT, and/or considered a non-entity after RT. cDNA is single stranded. RT is considered "first-strand synthesis."

You are starting at 1 ng cDNA/uL in your most concentrated qPCReactions - which is a perfect place to start (in my opinion).

On your melt curves: your GA3PDH looks great - especially when template is present. And your GOI has primer dimers that form the more and more you dilute your template. You could avoid the dimer either by redesigning another primer pair (or pairs) and or, only work within the dynamic dilution range within which no primer dimers form. Other than that, I think you are doing everything correctly and already have a good compass of what you should do. Your work looks nice.

It is not possible to get meaningful NanoDrop readings of cDNA for the reasons you stated: the dNTPs, left over RNA, degraded RNA nucleotides, and cDNA all contribute to the reading. In order to measure cDNA a Qubit and oli-green or oligogreen or RiboGreen would have to be used (fluorometric methods that measure only single-stranded nucleic acids...)

So normalizing RNA before reverse transcription is the way to go -- and use the same amount on RNA per each RT reaction. Between 30 and 50 ng/uL is usually the sweet range for that; but, whatever you choose, use the same for all.

Concerning your dCq ("Cq" is now used instead of "Ct"; connoting quantification cycle "Cq"): what do the separate standard curves look like? Stating dCq as the comparison is a little confusing as one cannot tell which curve has gone askew (perhaps due to template inhibition at the start and/or primer dimers at the dilute end). Ideally, yes, the two 1:10 curves would exhibit points that are exaclty 3.3219 cycles apart (e.g. log base2 of serial dilution factor of 10 = 3.3219...), and thus your dCq values would all be some constant value ...but, rarely are things always ideal...

Both GOI and HK (reference gene) efficiencies should be stated before more advice could be given here.

Sometimes 1:10 serial dilutions are too aggressive to generate good standard curves... so exploring serial 1:4 or serial 1:5 curves could be helpful.

Using too much cDNA at the start of the curve will often inhibit the PCR, and too little near the ends of the curves can invite primer dimer or lack of, or near lack of signal (due to Poisson noise of 1-copy reactions) thus sporadic Cq values there.

So, there is a sweet dynamic dilution range within which to work for each target. The hope is that this sweet range could apply to all targets in your study so you don't have to differentially dilute all samples and standards to address each different target's sweet range. Most often, HK (ref gene) reactions have to be doted on differently that other targets, given their usual inordinate abundance or even hyper-abundance (e.g. 16S and 18S ribosomal targets).

Hi Jack, and thank you for the response. I have only performed the standard curve reaction once, and with duplicates, not triplicates.

The equations for respective curves with R^2 are as follows;

HK gene - (GAPDH)

sample 1

y = -2.9905x + 23.868

R² = 0.98114

Sample 2

y = -2.6502x + 25.669

R² = 0.96446

GOI

Sample 1

y = -3.2217x + 23.872

R² = 0.95421

Sample 2

y = -2.8855x + 25.952

R² = 0.97939

I used 10 uL of 40 ng/uL RNA input to make cDNA, so total input was 400 ng. I then made 6 - 1:10 serial dilutions to generate the curves. The original cDNA sample was included, so concentration range was 1 (highest) to 10^-5 (lowest).

I believe my next step should be to perform the standard curve with less steep serial dilutions, like you said, maybe 1:4. Would I gain any information from this? The only reason I question the results is the discrepancy in dCq seen at different concentrations.

Another question - When you make cDNA, do you perform the standard reaction volume or do you make extra (say x2 or x3) and aliquot it? Do you generally dilute it prior to the Real Time reaction to conserve it?

Thanks for your help.

Aaron

Hi Aaron,

Some of the odd dCq's you were seeing could be due to not weighing water with every pipet adjustment you make. Every pipet (no matter how well or recently calibrated) is off. E.g., 10 uL of purified water should weigh ~0.01 g on an analytical scale. Doing triplicates is usually advised -- so that is a good idea until you become a pipet expert and are absolutely certain you have confidence in performing perfect technical duplicates eventually.

Melt curve analysis is key too... if you are using SYBR Green-based qPCR, be sure to check whether or not you are amplifying primer dimers or splice variants or pseudogenes along with your targets. GA3PDH has a pseudogene in mice and if DNase treatment was not sufficient you could be amplifying its gDNA as well..

You mention 400 ng input RNA into RT -- but in what volume of RT reaction?

20 uL? 50 uL? 10 uL? That is a good thing to know here as well, as dilutions then can be directly related to "ng cDNA/uL" in your final qPCR reactions. And yes, making 2X or 3X more of cDNA is a good idea always -- and there is an almost absolute necessity to dilute at least 1:5 or 1:10 before it is suitable to use in qPCR as the starting (most concentrated) dilution. Some cDNA is less inhibitory than others though -- and you will always notice this [inhibition] if the Cq is inordinately high in your most/more concentrated standard point(s). If no inhibition, but you have a good dynamic dilution range (> or = 3 log range for targets and 5 to 6 log range for robust housekeepers/ref genes), you can use as concentrated as you want - but then you'll run out of samples more quickly and this could limit the # of targets you can interrogate.

Your efficiencies of:

115.97% GA3PDH Sample 1

138.41% GA3PDH Sample 2

104.36% GOI Sample 1

122.11% GOI Sample 2

using Efficiency = [10^(-1/m)] -1

I assume were each generated by doing serial 1:10 dilutions of each sample to create a standard curve out of each one. The lower 2 efficiencies are most reasonable, whereas the higher 2 efficiencies seem to be problematic and could be the result of primer dimer, splice variant and/or psuedogene gDNA signal additions to your intended reactions. What do your melt-curve analyses show? One nice peak, or multiple peaks?

Another ghost to haunt your mind with, is that if RNA is not quantified carefully before RT, adding too much (and different amounts) of RNA to RT reactions can differentially skew and/or inhibit the reverse transcriptase enzyme reaction itself - and cause different yields of cDNA in each reaction. Hopefully this is not a problem you are dealing with. But it is not possible to tell whether it is or not given only "400 ng input" - without knowing the final volume of RT rxn used. But, even 400 ng RNA per 20 uL RT rxn should be OK as it would give you cDNA that is theoretically 20 ng/uL -- which is very safe (if highly pure, non-inhibitory RNA goes in). I'll wait til you reply again. In the meantime, weighing water to prove pipet settings and checking your melt curves is what I would advise. Looks like you are doing pretty well already for the most part. But, qPCR is a beast. Best to you.

Hi again Jack,

In response to your question, I perform 20 uL reactions, each containing 1 uL of cDNA solution. The cDNA was diluted 10x each time. We use Applied Biosystems Fast SYBR green Master mix, which states, "For optimal performance, use up to 20 ng of cDNA per 20- μL reaction, plus RNase-free water." If I use 1 uL of starting concentration of cDNA, it would theoretically be a 1:1 ratio of mRNA:cDNA. The mass ratio is quite close to 1, so this would contain ~40 ng/uL cDNA, so 40 ng in 20 uL reaction. If I dilute this 1:1 with pure water prior to performing the reaction, I could use 1 uL and be right in the sweet spot. Does this seem reasonable, or do I have to determine experimentally the proper volume to use?

As for melt curve, the HK gene GAPDH melt curve is attached. Each sample is represented by one line. In general, when I perform this, the no RNA input sample has a double peak, and the other samples have a single peak.

It seems my next steps will be to check the precision and accuracy of my pipettes.

Should I be concerned with the variation in Cq values at each cDNA input level? I would expect these to be more similar.

Thanks for your insight,

Aaron

Here is the GOI melt curve. Seems to be more variable in peaks and a consistent second peak.

Hi Aaron,

If you add 400 ng RNA/20 uL RT reaction, and assume 100% conversion to cDNA, your final cDNA samples are 20 ng/uL and not 40 ng/uL. RNA does not participate in the qPCR (but left over RNA could sequester your reverse primer). The RNA is either degraded (by the RNase H activity of the RT enzyme) during RT, and/or considered a non-entity after RT. cDNA is single stranded. RT is considered "first-strand synthesis."

You are starting at 1 ng cDNA/uL in your most concentrated qPCReactions - which is a perfect place to start (in my opinion).

On your melt curves: your GA3PDH looks great - especially when template is present. And your GOI has primer dimers that form the more and more you dilute your template. You could avoid the dimer either by redesigning another primer pair (or pairs) and or, only work within the dynamic dilution range within which no primer dimers form. Other than that, I think you are doing everything correctly and already have a good compass of what you should do. Your work looks nice.

Hi Jack,

I have performed the RT reaction again and got the following melt curves. My GOI melt curve looks much better and this was using 1 uL of the 1 ng/uL cDNA (no dilution).

Do you think I need to optimize primer concentration, or does this look OK?

And thanks again for your help. I'm still learning a lot. This RT seemed like a huge undertaking, but I feel like I'm finally getting a better hold on it.

GAPDH Melt Curve

Looks great.

And for SYBR Green qPCR anywhere between 100-600 nM seems to work well. I personally prefer ~400 to 500 nM of each primer in the final reactions. This is something you would have to determine empirically. In general, adding more primer is like turning up the volume knob on a stereo -- but you also run the risk of bringing up the noise floor as well (enticing dimers and/or other non-specifics to occur). Your stuff looks pretty clean and loud right now... but, there's always room for improvement I suppose. Nice melt curves!

One last thing - be sure to read the following by Dr. Stephen A. Bustin and crew:

http://www.ncbi.nlm.nih.gov/pubmed/19246619

(The MIQE Guidelines: Minimum Information for Publication of Quantitative Real-Time PCR Experiments)

Best to you Aaron~

hello, 

one question: when you are talking about RNA amount (400 ng for example) is that measured with nanodrop? is it possible nanodrop gives false results of RNA concentration??

thanks a lot

If DNA contamination is present, the ND could indeed give an inaccurate reading for RNA.  That is why some opt for fluorescent measurements of nucleic acid concentrations - either specific for RNA or DNA; single-stranded or double-stranded.  Qubit is one such system; but one needs to be very careful in using those methods as well (appropriately warming up reagents for the standard curves, etc.). RiboGreen, OliGreen (OligoGreen), PicoGreen etc. ... just a few of the advertised names of the fluors used for the fluorescent approach.

what amount and concentration of cDNA should be used for Real Time -PCR while working with zebrafish model.

See the attached...

The general, overarching ideas in the publication still apply nonetheless. We narrowly missed a big grant for building a platform-independent GUI for the program in 2009. Took me untold 1000s of hours to build the Excel-based program, so it was ripe for further development (with letters of endorsement from some of the top people in the qPCR world); but the fiscal climate in 2009 was non-conducive to federal science grant funding for all of us ...

Hi Alessandro, the different working ranges will not matter in the final calculations but, if you are still concerned about that - see this other post here:

https://www.researchgate.net/post/How_can_I_take_into_account_the_dilutions_made_in_qPCR_experiment_which_formula_ties_the_Ct_to_the_relative_concentration

The important thing is to work within the valid range for each target, for only then, can one 'trust' the Cq values in the first place. The relative nature of the equations involved cancel the working ranges out of consideration; working through the math numerous times will demonstrate this to you.  E.g., perhaps your standard curves for targets 1-3 may go from 1:25 to 1:2500 (or 1:25,000 or more), while target 4's stnd curve range may go from 1:125 to 1:12,500 (or 1:125,000 or more). Using a representative mixture of samples as the standard curve material should show you what you are dealing with for each target, and once that is empirically viewed, one then chooses a singular point within the good proven range at which to assess each target in each sample against the backdrop of the standard curve constructed over an appropriate dynamic dilution range relevant to the target being assessed. The PQ program merely makes it very easy to find these ranges and then calculates all of the dilutions necessary for any range needed and makes sure one does not run out of any sample or samples in particular throughout the course of the entire study. It also calculates all RT and qPCR mastermix formulations, primer dilutions and so on. It makes a day without enough coffee still land squarely on its feet when the circles under the eyes are dragging across the floor and one is wishing for Santa to fly a liquid handling robot to the lab ;]