Hi,

The differences in the procedures starting from RNA isolation till cDNA synthesis, which can affect the quality and quantity of the final template (First std CDNA) will finally affect the Cq values. The results can be comparable if procedure followed is carried out exactly. The unexpected difference in Cq of 18S RNA may be more likely due to the procedure followed for total RNA extraction. If the rRNA pool is low right from the beginning, the Cq values are bound to be more.

Disclaimer: my experience is limited chiefly to mammalian tissues/cells, rather than plants, but....

Honestly, I think your protocol sounds more robust than the cited values: 1/10 dilution of cDNA is certainly safer than 1/5, and using less of this diluted cDNA is better than more, and allows more assays: they're effectively using 1ul of neat cDNA per reaction, while you're only using 0.2ul (frankly, I don't see the point in doing a 1/5 dilution if you're going to use 5ul anyway: might as well use 1ul of undiluted stock).

Values of 9-12 are what I would expect for 18S in mammalian samples, and I can't imagine plants have wildly divergent demands for protein synthesis: no matter what kingdom you're from, you're going to need a lot of ribosomes. I would, therefore, be inclined to trust your 18S values.

I have no experience with TUBB, but in my hands GAPDH tends to hover around 15-18, so your values are a bit more concerning here. If your 18S is coming up as higher expression (lower Cq), then you'd expect your other references to follow suit.

How are you preparing your cDNA? I assume random priming is involved (because you have 18S signal), but did you also include oligodT?

Do your reference genes agree, regardless of actual Cq values (i.e.if a sample is low for 18S, is it also comparably low for TUBB and GAPDH)?

Also, how are you determining your Cq values? qPCR software usually defaults to using threshold methods ("which cycle did the traces cross an imaginary line?"), but in my experience I get better data using regression methods ("at which cycle was the peak of the second derivative of the amplification trace?"): this latter metric also typically gives lower Cq values (generally by ~2 cycles), which could play a role here.

Your safest bet is to perform a dilution series, to establish over which range your data is linear. Since you're already diluting your cDNA a lot, you have scope to go both up and down the concentration scale. A 2-fold dilution series, starting with, say, 4ul of your 1/10 stock (so 4, 2, 1, 0.5 etc -make dilutions of the actual cDNA for more dilute samples) should give you more information. If all your traces show clear linear relationships (2 fold dilution = 1 cycle increase) over the range you're using, then you know that your data is reliable: regardless of any other possible problems, you will know you are at least accurately measuring what you have in your cDNA. And this would be a good thing.

If you can safely establish that, then we can troubleshoot further (if we need to: remember, it's not impossible that the published values are...wrong).

Your numbers are reasonable to work with since you can have values higher and lower than those Cq values. Remember, you are calculating either absolute copy number (or relative copy numbers) for gene expression in YOUR samples. It doesn't matter what the exact published values were, what matters is the analysis. The folks in the paper you refer to most likely just started with a more concentrated RNA sample.

HI, if you are getting ct values of 22-25 for GAPDH ( I don't have much experience with TUBB and I work with mammalian cell lines, not plants. That simply means your cDNA is extra diluted and there is less of the reference gene available for amplification. Generally, the values we obtain are around 15-18. But if you are getting above 20 that doesn't necessarily mean you have a problem you can still compare the amplification with your test gene and get an idea of how the expression is varying in your test conditions. But you should repeat your experiment with less diluted cDNA perhaps. Also, you can check the melt curve to find out if the primers/probes are working fine. Also, you can select a reference gene which has ct values in the range of your target gene. Hope your problem gets resolved soon.

John Hildyard

Thank you very much for the informative answer!

Indeed, I'm also a bit more concerned about the fact that my reference genes aren't uniformely more or less expressed than the ones from the paper. But that 18s has lower Cq values while the other two have higher Cq-values...

My cDNA is prepared using a mix of both random hexamers and oligo dT(mainly because my virus has no polyA-tail)

To answer your second question. Well, some agree, some have slight changes (not more than 1 Cq-value however). I use qbase+ to analyse my samples and according to this programme 18s rRNA and GAPDH are the most stable reference genes (TUBB is a bit less stable).

My Cq-values are indeed chosen by the software itself! How do you calculate these values using regression methods? In a seperate program?

Well, I didn't mention this before but I performed a dilution series to check my primer efficiencies. I'll post them underneath.

Shivani Yadav ,

Thank you very much for the answer!

I would like to try using more cDNA in my qPCR reaction. However, next to the problem that indeed my Cq-values are higher for GAPDH and TUBB compared to that paper. I do have much lower Cq-values for 18s rRNA... So when I will add more cDNA, this values will even lower more and that will not change my problem in general because the ratio of 18s rRNA vs GAPDH-TUBB will not change I guess?

Laura De Keukelaere : thanks for the extra detail, and sorry about the delay (everything takes longer in a pandemic).

Your GAPDH amp curves look flatter than I'd expect: after maybe two cycles post-detection limit, they essentially look linear (rather than exponential). This doesn't imply the reaction is less efficient at earlier cycles, but does imply that it reaches reaction saturation/limiting conditions very quickly.

You don't specify what you dilution series was, nor what efficiency results you got, but my guess is a four-fold dilution? All your data seems to be about two cycles apart between dilutions.

Note that the threshold line for GAPDH crosses the traces in the range where it's already likely suffering from reaction limitations, which might lead to artificially higher Cq values (though might not affect your efficiency calcs). You can also click and drag the threshold line up and down, which gives you an idea of how arbitrary the concept of Ct can be.

http://users.stlcc.edu/departments/fvbio/thermal_cycler_mxpro_manual.pdf

In the case of your GAPDH data, I would simply drop the line a little to make sure it's clearly in the exponential phase. According to the manual (for the software I am guessing you're using, based on screenshots), it doesn't look like regression thresholding is a feature, which is a shame: ostensibly it's the peak of the second derivative (where the rate of change of the rate of change of fluorescence is the greatest)

I suspect you _could_ calculate it yourself, but I'm not sure it's necessary, if your efficiencies are holding up ok. If your data is solid, holds up to dilution conditions, and appears to be consistent and you get comparatively good agreement between references, I'd trust it.

Article A new real-time PCR method to overcome significant quantitat...

Article Improved real-time RT-PCR method for high-throughput measure...

John Hildyard

Ofcourse no problem, I’m very thankfull allready for all the information!

I did also notice that my GAPDH is quite flat... however I have no idea to solve that. Since it’s clearly not a “contamination” issue( for example from the cDNA synthesis reaction).

Can I fix this? When you say it reaches the limiting conditions fast it sound like I should just add more mastermix?

Yes, forgot about that! My dilution series is 1:5. So I start with 1:4 then, 1:20, 1:100 and 1:500. my efficiencies are indeed pretty good! 18s rRNA and TUBB have an efficiency of 100% While GAPDH had an efficiency of 98%.

Indeed I don’t think the software has this possibility, but I will look for it!

Calculating will probably take a lot of time... thinking that in the future I want to test at least 170 samples... Because I only want to investigate the relative proportion of the virus in different samples I was also thinking that somehow the ratio is more important than the exact values.

ps. Also thanks for the papers, I’ll check them

Excessively quick reaction saturation can happen for a host of reasons, but a dilution curve is empirical evidence that it doesn't matter: if you can take a sample, dilute it 5-fold four times, and still get Cq values consistent with that dilution, then your data is solid.

Based on your responses, I would trust all three of your genes.

Do the values agree with published data? Maybe not, maybe so, but...if they're trustworthy in your hands, then...trust them. There are (as noted) a lot of ways measured Cq (or Ct) values can vary in absolute terms, but as long as the relationship between them holds up, they're still giving you meaningful information.

If your ultimate goal is to _detect_ viral infection, rather than to _quantify infection in very exacting detail_, then arguably qPCR is pointless anyway: with a viral genome it's either there or it's not, and if it's there, it usually REALLY there, so for much of your analysis you may be looking at pretty much "yes/no" questions. If you also want to look at downstream effects on cellular transcription patterns, then yeah: this is all important stuff to check.

My personal take is that your data looks solid. If I were reviewing this, you'd get bonus points for

• using multiple reference genes
• establishing PCR efficiencies

And you'd lose no points just because your specific Cq values are slightly different from published values. If your 18S values were like, 27 or so, I'd be very suspicious, but as it stands: this all looks good to me.