Different amplicon Tm at different temperatures

[damonbio]

Hey guys,

I'm optimizing my qPCR assay, so I ran a qPCR with the same set of primers at different melting temperatures, in order to find the one in which the assay proceeds best.

I tested 54, 56, 58, 60, 62 and 64C for the PCR melting step and I added a melt curve analysis at the end of the qPCR run to test for multiple peaks (dimer primers, multiple amplicons, etc)

I got a single peak in the melt curve analysis for all reactions, however, the exact melting temperature of the amplicon was not the same for all of them, even though the template was the same and the primers were the same.

I got a Tm of about 82.4C for reactions at 54, 56,58 ,62 and 64, and of about 81.8 at 60C.

What could be the reason for such a shift? Is it significant?
I ran a gel and I saw a single band per reaction, with the same migration pattern.

Thanks!

-D

[mchlbrmn]

Your gel showed you that they all produce the same PCR product. The shift is probably caused by a slightly different salt concentration due to pipetting error, I guess. Did you use the same PCR reaction mix in all samples, or was there pipetting? I often (I admit) see shifts of that amount on different days with different mixes. I have seen some shift on later samples one day when my reaction mix wasn't mixed very well once (my enzyme mix is rather viscous, so usually I'm careful to make sure it is well mixed in).

[damonbio]

Thank you for the prompt reply.

Yes, the same mix was used for all samples, which was distributed with an electronic pipette. The 'weird sample' was in the middle of the plate, so I used the same mix for wells before that one and after.

[toxlab]

the salt concentration most strongly dictates the melting temp..but if I understand your post correctly you used the same master mix and the same sample to run different anealing temps for the primer set (so there should have been a homogenous salt conc., unless you diluted your samples differently across the test).

Regardless, the change of Tm is less than 1C and your product was verified on the gel--I would suggest the reaction is probably fine.

[mchlbrmn]

Suppose... below 62C a nonspecific PCR product can form, with a primer mismatch or two, and yet below 59C the template for this secondary PCR product folds into a secondary structure that is not accessible for PCR. Thus only at about 60C is this undesired PCR product produced. This alternate target is repeated in a high copy number (this contributes to the secondary structure) so more is produced than the desired target. Furthermore, the DNA sequence of this malicious PCR product contains inherent enzymatic activity, like a ribozyme or aptamer, that causes it to target and degrade the desired PCR product, so you only get the one spurious product in the melt curve with a 60C anneal temperature. (I think that it may also posses amphibious qualities that can enable it to crawl out of it’s well, across the lab, and to attack other experiments. I think this can explain most of your problems lately... and perhaps any missing colleagues.) :)
Either that or a bad well in the cycler.

[GBMEpiGuy]

Hello. A couple more notes perhaps worthy of thinking about. [Salt] and consistency of your reaction mix (thorough mixing before use) are both likely suspects. However, It's also important to remember that simple one-dimensional gels only inform you of significant variabilities to amplicon length. In the event that you're (un/)intentionally amplifying bi-allelic templates, variability in the degree of selection bias can easily effect reaction melting curve peaks, from reaction to reaction, by 0.5C, even across a very small number of SNPs. This is a phenomenon you'll never detect in a simple electrophoresis gel, as amplicon length isn't going to change and any variation in the electrical potential (from one SNP allele to the other) won't detectably (visually) effect the migration rates of the two clones. Also, it's really important to talk to the manufacturers of both your reaction mix and the qPCR platform you are using. At least one major manufacturer I've worked with will assure you that better than +/- 0.5C accuracy, regarding reproducibility of melt curve peaks, should not be anticipated with the available technology (which isn't to say this is the general case, hence, the need to investigate further behind the specific tech you're using). Hope any of this helps!