Solved the problem! It was the Dnase after all. Tried the RNase H suggestion with no luck, and started over with Life Tech Amp Grade DNase I, followed by RT and qPCR and the curves are beautiful. HOORRRAYYY!!!!

For future reference, the combination of a Invitrogen Turbo DNA-free kit with VILO superscript RT and Bio-Rad iQ SYBR green master mix is completely incompatible. I don't know why, but switching DNase fixed the problem.

Thank you greatly for all your help everyone.

Wow, that's weird :o

Are the melting curves ok? Also, I'd check if the passive reference was set correctly in the software, if there's any.

Hi

In my opinion you're using a lot of DNA and you're obtaining not conventional fluorescence patterns at low Ct. If you want to understand the nature of your problem you can try to evaluate the PCR efficiency of your samples. There exist different approaches, the most simplest is to put the fluorescence in log scale and evaluate the slope of the straight line with the standard slope.

It has happened to me before. With independence of your samples being too concentrated the curves of your standard also look weird (flatening, low efficency) which I attributed to low QPCR effciency most likely coming from the primers (bad design or many degenerancies). The only think I could recommend to fix the effciency and curve of teh standard is change the primers (if possible)....good luck

It could also be that your start cDNA sample has "contaminants" from the reverse transcription reaction (and/or also from your RNA extraction/DNAse digestion) and by diluting your cDNA you are getting rid of those "contaminants".

Have you tested primers before? I think primers design could be the problem.

It is not advisable to calculate the efficiency from such curves. It won't give you any information on your problem and the more, you won't get the true efficiency for this. I agree with Gerson: did you check the melting curve? Are the controls okay? How about the no RT?

It looks like you have very early amplification (high copy number of your target sequence) so you need to set different background settings than for other samples. The other solution is to dilute this sample.

I would check the controls. It seems to me that you have some contaminants or other things that interfere with the amplification. Good luck!

Cannot conclude before you have clean in silico data about your primers (GC content, self annealing, close Mt...) and clean melt curves. This is super important before troubleshooting anything else. If you are using the same kit as other primers, this makes it even more imperative.

To me, this could be one of the 3 following

1) Impurity in the template that inhibts the PCR reaction

2) Bad primers, this you can easily see when you run melting curve analysis...you should have only 1 peak, but for your primers I guess u see many peaks or no distict peak at all

3) Too high template/primer concentration. You could run the same sample on a serial diluted template, keeping the same concentration of primers. In this way you will not only evaluate primer effeciency but you will also get an idea on how much template should be used.

Good luck!

Hi Jason,

it seems to me a problem of fluorescence quenching. Normally SYBR green should emit fluorescence only when it binds double strand DNA and not when is free in solution. Have you prepared fresh primers for RT or for qPCR reaction? Perhaps you miss dilute the stock primer solutions and you over load one or both reactions. A lots of primer might give non specific amplification at early Ct. It seems that with high dilution the reaction works better, supporting the idea that primers may be wrong concentrated.

I agree with other comments... I have similar amplification plot during my first experiments on relative quantification of mRNA, when I used too much cDNA template. In particular those problems were also due to a bad design of assay (I was forced to design my assay on one single exon), with no optimization of primer concentrations and/or incomplete DNAse digestion. As already suggested by Vivian melting curve analysis, together with RT controls could help.

The highest concentrations in your dilution series show that you have way too much template (notable amplification before cycle 10 requires massive amounts of template copies). You either overloaded the reaction or your primers are amplifying multiple targets (As already mentioned: melting analysis information would be a great help).

The second set of curves are extremely random and seem not like specific amplification. Were these primers tested in silico or validated on a reference material? You might want to design a plasmid containing the target sequence (TOPO cloning should do) so you can set up a series of reactions with known target copies to validate your method's amplification and melting profiles.

I think, diluting cDNA samples will not solve the problem, since you showed us already the results of this in the serial dilution that you made. None of the dilutions is acceptable as you already wrote: the first ones look strange, the last one have too late Cts.

By writing this, there comes another idea in my mind: Maybe you overloaded the RT reaction by using too much RNA. However, I don't know if this is possible at all... ;)

Try running the qPCR products from each of the different concentrations on an agarose gell to first see whether you get a single unique product especially in samples with low dilution. Cyber green just measures ds products, not specific product. Melting profiles are also informative. Another test is to spike your different concentrations with a known non-archaea RNA or cDna template with corresponding unique primers and probe and/or cyber green at a concentration giving a Ct in the low 30's. (external PCR reaction control).

Hi Jason,

As mentioned by all the person before, the problem may be due to interference/inhibition in your PCR reactions either in the cDNA synthesis or the real-time stage or problem in the primer design. You mention using 100 ng of RNA, what is the volume of cDNA synthesised? Have you tried using less RNA in your cDNA synthesis reaction as using lot of RNA itself inhibit PCR reaction. Agree with Vivian.

Good Luck

Wow, thanks for all the ideas guys. I'll address you each individually below, but many of you want to see melt curves. I've attached an amplification plot and melt curves for 1:100 dilution of my cDNA with 39 samples.

@Gerson Kobayashi: The melting curves are perfect, even for the samples with very weird curves.

@Francesc Codony: According to my Qubit fluorimeter I have 200 ng/ul of cDNA in my undiluted samples. So my 1:1000, which still gives me a weird, unusable curve is only 0.2 ng/ul, and my 1:5000 which is the first usable dilution, is only 0.04 ng/ul. That's such a small amount!

I checked my PCR efficiency per your suggestion, but I don't think these ugly curves will give me true efficiency. If it did though, my efficiency is 21% over the range of dilutions tested, so pretty awful.

@Amin Talebi Bezmin Abadi: The same qPCR reaction mix was used in other qPCR reactions that worked just fine. The problem doesn't seem to be in the reagents, but in the samples.

@Chiranjeevi Sandi: I've tried serial dilutions, as I mentioned above. By the time I get useable curves my Ct is too low to be reliable.

@Laura Villanueva: I have used both sets of primers that are giving me this issue before with success and they each only produce one product (one band on PCR gel and single peak melt curve). They also give normal curves over a range of Ct's, from 12 up to 37. The only difference between the time they worked and this time is the first time I used a greater amount of RNA and diluted 1:10, and I used a different brand of Dnase.

@Manuel Sanguinetti: A good thought. Can these contaminants be removed somehow? Dilution clearly isn't good enough.

@Alberto de Marcos Serrano: Yes I've tested both sets of primers that are giving me this problem. No issues until now.

@Vivian Jürges: Melting curves look great, the correct product is being amplified. I did not do a no RT (we check the absence of DNA template prior to RT with regular PCR to avoid wasting expensive RT reagents), but there is no DNA contamination. A positive genomic DNA control produces a normal curve with a Ct of 20. A water negative control is either flat lined or very low Ct of 38-39.

@Marcin Schmidt: As mentioned above, dilution doesn't help because I get weird curves right up to the point where the Ct becomes unreliable.

@María Gonzalez: Favoring this contaminant idea as well. How to remove?

@Elia El Habre: Melt curves are clean and these primer sets have worked before.

@Inderjit Marjara: Impurity in the template seems likely, I've addressed your other concerns above.

@Angelo Ferraro: Very interesting idea, but unfortunately during my troubleshooting I have made up new primer dilutions twice. I doubt I would have made the same dilution mistake 3 times; only possibility is the company told me the wrong primer quantity, which also seems unlikely.

@Alessandro Benedetto: Melt curves and controls are fine.

@Antoon Lievens: Melt curves are fine. I've tested these primer sets before and they work.

@Vivian Jürges: I thought of overloaded RNA as well, but according to my Qubit fluorimeter every reaction received 100ng. The Qubit could be wrong, but why would a lot of RNA cause this in any case?

@Lester Shulman: The spiking idea is an interesting one. I am waiting on a collegue to run his qPCR today and then I may have to try that. If that reaction is inhibited too then there simply must be something in the sample that I need to get rid of.

@Manohursingh Runglall: I find that too much RNA is unlikely. We often use 2ug (20x more) from insects in our RT reactions without issue.

Hello Jason, did you do a RNAse digestion after the RT step? A lot of RNA can cause problems and inhibit PCR. Also, your melt profile does not look as clean as you are saying, there is a baseline increase between 68-75oC that could be due to non-specific amplification or primer-dimer. I would suggest starting from beginning with fresh reagents and running all controls: RT and no-RT (make sure to do RNAse A digest), primer combinations (cDNA alone; cDNA + primer-F; cDNA + primer-R; cDNA + primer-F + primer-R; and primer-F + primer-R). Run these reactions on a gel to check for non-specific amplicons, primer-dimer, and genomic DNA contamination. Based on your current data I think your most likely inhibitor/contaminant is RNA from the RT reaction.

The spiking experiment mentioned by Lester Shulman might help point at the problem indeed, since the dilutions didn't help.

You can even find commercial kit to do that.

Just make sure that the probe is compatible with your(s), and also include a 'control' with the spiked DNA only (none of your sample) in order to get its melting curve peak and check for good efficiency.

Good luck!

Hi Jason.

I agree with comment from Roberto Nussenzveig.

But I was wonderng, have you checked youre PCR instrument? Do you have an assay you know will work "everytime"?

It seems you have pretty good control on youre PCR generally, so I would have focused on the function on the PCR instument, is it working good? Does it have uniform temperature over the thermo block?

Well nothing more to add, good luck and remember: Sometimes PCR = Polymerase crap reaction.

Regards see

Are you using tubes/plates which match your qPCR instrument? Are your plates and the sealers compatible with each other? I have experienced such thing earlier. Most probably the tubes were not sealed properly which has caused the evaporation of the die and contamination of the detection system.

best wishes.

I think Steve solved the problem. This sounds logic to me :)

@Roberto Nussenzveig: I did not do an RNase digestion as my RT kit does not call for it (VILO superscript). Thing is it seems to me like I'm seeing early florescence, not inhibition of the reaction, unless you interpret these curves differently?

@Claire Perrin: My thinking now is that will tell me whether or not I have a contaminant, which seems almost a given now, but not what it is or how to get rid of it. My plan now is to use a PCR clean-up kit to get a 1:10 dilution of my cDNA in pure water and try again.

@Sten Erlandsen: I have other PCR reactions with different template that are working consistently. It doesn't seem to be a problem with the machine.

@Muntasir Alam: Good suggestion, but our plates are compatible and there is no visible evaporation after the run.

@Steve Baeyen & Vivian Jürges: I am using a Bio-rad CFX96 qPCR machine which does automatic background correction. I checked the settings and they are appropriate. Changing them manually per your suggestion did not help. Good thought though!

Did a clean-up of my cDNA and am running the qpcr again as we speak.

No luck after cDNA clean-up. Whatever is causing the problem must be bound to the DNA. Below are the amplification curves as the program is running on top, and after processing below. As you can see, the majority of the samples start in the negative RFU range and then plateau at zero. However, the melt curves are all the same, so presumably product is being made.

My labmate ran his qPCR and it didn't work either. His standards worked fine, but all his samples start out flat or negative and slowly increase near the end. The only thing in common between us was the Dnase kit and precipitation method we used. As I see it my options now are 1) Different Dnase kit, 2) Try RnaseH treatment, 3) New samples from a different time point, 4) some combination of the above.

Hello Jason, what does your NTC look like? Did you run an agarose gel of your reaction? You are correct regarding very early amplification. Did you try to run the primer combinations like I suggested above? I have seen curves like this before and they are usually associated with too much target and your reaction running out of substrate very quickly. The presence of contaminating RNA in your PCR step is very likely. My suggestion is the same as before: run all the primer controls (suggested above), the NTC, and also perform DNase I and RNase H treatments. Run the amplicons on a gel and post the results again. What are the primer concentrations in your PCR reaction? How are you doing the RT, are you using gene specific, oligo-dT, or random primers? Good luck and let us know how it goes.

Hello Jason, I forgot to mention that I do interpret your curves differently. They are biphasic with the first amplification occurring before 20 cycles and second afterwards. This can be typical of RNA contamination (RNA degrades at high temp) and it is slowly degrading at every cycle. RNA will be competing for primer binding, if you are dealing with a highly expressed target you can get the curves you are observing. Also, I did mention that your melt profile is not clean, you have an early bump, you need to run your reaction on a gel to look for non-specific products. You should also optimize your reaction by running a temperature gradient pcr and analyzing the results on a gel, post the results for more help.

Solved the problem! It was the Dnase after all. Tried the RNase H suggestion with no luck, and started over with Life Tech Amp Grade DNase I, followed by RT and qPCR and the curves are beautiful. HOORRRAYYY!!!!

For future reference, the combination of a Invitrogen Turbo DNA-free kit with VILO superscript RT and Bio-Rad iQ SYBR green master mix is completely incompatible. I don't know why, but switching DNase fixed the problem.

Thank you greatly for all your help everyone.

Congratulations, glad for you....I also have no idea about the strange curves.

Woohooo! Congrats!

This is awesome!

Is it possible that one company's reagents might be 'poisonous' to another company's reagents? It may be a high form of art amid the great landscape of corporate warfare. A good company gets people addicted to their reagents, and, one step further, it seems, would be to make sure that your company's reagents are incompatible with and/or destructive to all others. (I really hope this is not happening out there... but one never knows).

I do know that I have found it better to spin out the (TURBO DNase) inactivation reagent for 3 minutes at 10,000xg (rather than the 1 minute suggested in the kit literature). But, it sounds like the inactivation reagent may not have been the culprit here? Very interesting. In the past I noticed that a certain ABI One-step mastermix was incompatible with an Invitrogen RNA extraction/DNase treatment... this is the second time I've heard of this. I later think I identified the culprit as Mg++ chelators.

Some mixes can withstand more chelation than others... and it may be sheer amount of Mg++ added is all. Not sure though.

Hi every expert here. I am also having the same problem. Same GAPDH gene but in 2 different tissues. 3 negative RFU while 1 positive RFU. Melt curve is ok.

What are the reagents you are using?

I am using Qiagen Quantifast mastermix for real time PCR, use Invitrogen M-MLV for reverse transcription. Then I directly add 1uL of cDNA into 19uL of the PCR reagent and perform touchdown real time PCR in BIO-RAD I-Cylcer with duplicate tubes per gene.

Did you check the raw data (before baseline subtraction?) This seems more like a data processing artefact than a chemical or kinetic problem. Another question: how much target did you load to get amplification this early (Cq

What do you mean by data processing artifact? I load 1ul of undiluted cDNA to the mastermix. As the target gene expression is low, then if I diluted the cDNA, it will dilute my target gene expression.

By data processing artifact I believe he means that the readings are not accurate because of problems such as background/baseline/calibration and not the because of the reagents or reaction per se....

The Cq is indeed too low...do you have an image of the amplification curves of your target gene using these samples? If the Cqs are low too you can go ahead and make the dilutions.

What I mean is that, because you don' have any baseline, the algorithm for baseline subtraction uses cycles from the amplification or even the plateau to calculate the base fluorescence value. As a consequence it subtracts a value that is several orders of magnitude too high, thus yielding negative FU values. Since you're working on a biorad iCycler this is straightforward to check: select "Background Subtracted" from the 'analysis mode' dropdown menu when viewing the amplification curves. If all curves are now normal (and no longer negative) then the problem is caused by baseline subtraction.

You can also check which cycles were used for baseline calculation and adjust them manually by right-clicking on the curves and selecting "Baseline Threshold".

I'd also like to point out that you cannot expect correct curves with that many PCR target copies loaded. Cq 20 equals approximately 100.000 target copies, below Cq 5 you're loading more than 3.300.000.000 copies of your PCR target. There is little to gain from supercharging your reactions like that.

Are you multiplexing? Otherwise there is no reason why you cannot dilute your reference gene reaction while keeping your target reaction at the initial concentration. You can simply take the dilution factor into account when doing your calculations.

Dear Antoon, your suggestion worked after I changed to "Background Substracted". The curve is normal now. Okay. Thank you for your big help and advice. Appreciate it. :)

dilution was not proper

Hey Jason,

I have the same problem with my qPCR curves. In this case I am using genomic DNA for relative telomere length measurements. I run qPCRs of my target gene (telomere gen), and it worked fine, however when running the second one, i.e. the GAPDH gene, I got also negative values of the RFU for the most concentrated standards (150, and 30 ng/uL). I can understand weird values for the most concentrated one, actually is common to don´t include it in the final standard curve, but I would not expect it for the std 30.

Have anyone some suggestions about what I am doing wrong?

I am running a 2-step qPCR setting the cycles (45X) at 1 min 56°C and 1min at 95°C. A pre incubation at 95°C for 10 min was done.

Thank you in advance for your suggestions.

I solved my problem: I readjusted the time of each step for a one-step qPCR and also added less genomic DNA, that was saturating the polymerase reaction.