Very complete and good description of your problem! Thank you, it´s rare on RG. One thing that should have been mentioned: RT (rand hex or oligo dT) and qPCR system (sybr-Green I assume).

- all you do is tip-top so from my experience the best thing to do - if you don´t want to invest time to investigate what technical curiosity produced this problem- is to generate an alternative set of primers and just test those. Sure, you can put a lot of effort into primer design to reduce the risk for unspecific products, but in the end it´s all theory and you can end up with something that´s not easy to explain (as in your case).

->one thing: running your RNA on a gel will only help you to determine RNA integrity/degradation and to check if you have a lot gDNA contamination (depending on the detection limit of your dye/documentation system). Hence, DNA contamination cannot be excluded. What is usually recommended to check for DNA-contamination is -RTase controls and running those alongside your +RTase samples. This can also give you an indication in qPCR if any product that you may get is actually enough to change your result significantly. Say, you get a CT for your -RT of 30 and a CT for your +RT of 21, then you have a difference of 9 cycles. 3 cycles difference is roughly a factor of 10x, so 9 cycles is 1000x. Hence only 1/1000 of your +RT-signal is from spurious reactions and I bet that pipetting errors and variability between biological samples is are in the range of %. Of course, if your -RT and +RT are only apart 3 cycles, it´s unacceptable.

-Coming back to gDNA contamination: as also Mr Hall pointed out, it cannot be excluded. If the band sizes you observe cannot be explained by current gene models, you may also have novel splice forms in your sample; if that is of any interest to you, you may wish to check for that or at least keep in in the back of your mind.

you do PCR or qPCR? because for qPCR u can use probe also so it makes it more specific and might help u to get the better result....in the top u mention qPCR but your question is about PCR. Also primer designing for these two is different.

I am doing q PCR and I am following the primer design protocols for this method. I've kept the question fairly broad to try and get the most help possible! Yes I know probes can help, and will use them in future.  I just don't understand why I'm getting these larger non specific amplicons, as my primer design and pcr prep is so tightly controlled!

Hello Sherina,

your Primerdesign looks pretty decent, so i can't see any obvious failors. Have you tried to change your reagents (Primer, Enzyme, Buffer, Water etc.) to be sure, that there is no contamination in one agent?

What kind of sample material do you use? Is there a possibility, that you got a mix with fungi or bacteria?

If you want to have more information on your PCR product you could clone with a TA or Blunt Kit and sequence it. Maybe this will give you some good answers.

Maybe you could try the melting temperature, i.e. from 55-95 degrees to 70 (look at your product band and decide this temperature)-95 degrees. I had the same problem, a professor in the lab told me this, and it worked

Have u tried standrad curves to find the best DNA concentration?

Dear Sherina. It does sound like you are doing everything correctly. Can I ask if your other primer pairs which you can optimise to produce one product are 2 different genes ?

• Regarding temp gradient optimisation given that your specific product is strong have you tried annealing at the actual primer (s) Tm ?  When I perform a temp gradient I routinely go from Tm-3C to Tm +2C 
• As you say exon boundary primers should preclude the possibility of those larger products being genomic DNA by products
• If you have indeed gone to at least the Tm of your primers in terms of annealing temp then it isn't impossible that you could be dealing with alternative splice forms although to be honest those bands do look non specific compared to your string specific bands (?)
• Thus all I can suggest is repeating the PCR but increasing your annealing temp by 2C and adding 5% DMSO to your PCR reaction to see if you can eradicate those larger bands
• Failing that I might try another set of primers and if the bands persist extract and sequence ( just to be sure you are not dealing with alternative splice forms specific to your tissue; this can happen)

Very complete and good description of your problem! Thank you, it´s rare on RG. One thing that should have been mentioned: RT (rand hex or oligo dT) and qPCR system (sybr-Green I assume).

- all you do is tip-top so from my experience the best thing to do - if you don´t want to invest time to investigate what technical curiosity produced this problem- is to generate an alternative set of primers and just test those. Sure, you can put a lot of effort into primer design to reduce the risk for unspecific products, but in the end it´s all theory and you can end up with something that´s not easy to explain (as in your case).

->one thing: running your RNA on a gel will only help you to determine RNA integrity/degradation and to check if you have a lot gDNA contamination (depending on the detection limit of your dye/documentation system). Hence, DNA contamination cannot be excluded. What is usually recommended to check for DNA-contamination is -RTase controls and running those alongside your +RTase samples. This can also give you an indication in qPCR if any product that you may get is actually enough to change your result significantly. Say, you get a CT for your -RT of 30 and a CT for your +RT of 21, then you have a difference of 9 cycles. 3 cycles difference is roughly a factor of 10x, so 9 cycles is 1000x. Hence only 1/1000 of your +RT-signal is from spurious reactions and I bet that pipetting errors and variability between biological samples is are in the range of %. Of course, if your -RT and +RT are only apart 3 cycles, it´s unacceptable.

-Coming back to gDNA contamination: as also Mr Hall pointed out, it cannot be excluded. If the band sizes you observe cannot be explained by current gene models, you may also have novel splice forms in your sample; if that is of any interest to you, you may wish to check for that or at least keep in in the back of your mind.

One more thing: for qPCR I usually use only primer3 online software for finding primers (exon spanning if possible) and then I just try how they work. If they are not good I just get other primers instead of trying to finding optimal conditions for each primer pair. This allows me to run any primer pair in parallel in the same run, which in the end saves a lot of time if many genes have to be checked on the same set of samples.

Hi

Many thanks to everyone who has responded, I appreciate your time and effort.

Dear Wolfgang, yes i have systematically changed my reagents (one at a time) to see if this was the problem. I even changed my plasticware (using fresh unopened packages) to see if this might help. Overall it hasn't improved things although switching the brand of Taq helped slightly. I am using a primary culture (human immune cells) so bacteria is a possibilty, although I have blasted my primer sequences and can't see any obvious matches.

Dear Lawrence, yes, the primers that i have optimised correctly to produce one product are from a different gene. I run temperature gradients from -5 -+5 tm, I ran these PCRs at +4 tm because the gradient looked the best at this point. I will definitely repeat the PCR with DMSO, many thanks for that advice. I plan on redesigning the illustrated primers today. Regarding alternative splice forms - I design the primers (to the best of my knowledge) to cover the specific isoform(s) of interest using the information provided on the NCBI database. Could these genes have other isoforms that are not listed? (I am not a geneticist so please excuse my ignorance)

Dear Christophe. I have been using random hexamers because my cells have a short half life and i was concerned there might be a slight risk of degraded RNA within my sample. I'd read that random hexamers are better in this circumstance. What are your thoughts on this? Thank you for your advice regarding +RT/-RT control, I will do this from now on. Likewise i will have a look at primer 3 today. I have been persistant with trying to figure out the cause of this problem because i was concerned i had made a mistake somewhere/might have an underlying issue that could affect my results. I feel a bit better now so thank-you!

Sincere thanks

Sherina

In my experience, spurious bands of larger than expected size in addition to the expected product often occur when template concentration is too high. Try diluting your cDNA more and see if this helps. If it doesn't design new primer pairs as others have suggested.  

My experience with random hexamers and oligo-dT is most likely limited. I tried both, I compared (for macrophages) both and got quite similar results so I´d say it matters not. Perhaps the risk of detecting pre-mRNAs is a bit higher with random hexamers, but I´ve never heard anyone having problems with this so I think it´s a theoretical problem.

Philips point is a good one. Not only for reducing spurious bands (perhaps through reducing their abundance below detection limits and partly by reducing the risk for spurious reactions) but also for economic reasons. I always use my cDNA 10x (minimum) diluted in the qPCR reaction (normally I dilute cDNA 50x and then use 5 uL of this in a 20-25 uL qPCR reaction). There was an old article checking amplification efficiency and it was lower with undiluted cDNA and I think 5x dilution was ok to solve the issue. However, as long as you get CTs in acceptable range, diluting cDNA more than 5x does not hurt but will give you more volume to run reactions with, saving your PIs money and your time (for preparing enough cells, RNA and cDNA) ;-)

Thank you both, Annemieke and Christophe.  I have designed new primers but will take your advice on board (re cDNA dilution and annealing times) when optimising them.  Best wishes, Sherina

In my opinion you have pretty much narrowed down the problem to your sample.  You could try labeling your primers and use them to probe the original sample in a northern blot.  If you detect a larger band then it is some strange splice variant.  Remember to run the proper controls 

All great suggestions thus far. I have three of my own from my experience running many qPCRs.

1. Use primers with sequences obtained from the PrimerBank (see link). Most if not all mouse primers have been experimentally validated. This is not the case for the human, but all of the human primers I have used based on the sequences in this database have worked for me.

2. Don't use primer concentrations greater than 125nM. While I experimentally determined that this was the ideal concentration to minimize or avoid primer dimers, reducing primer concentration is also another way to reduce non-specific priming.

3. Similar to what others have suggested and along the same line of reasoning for why you wouldn't want to have high primer concentration, is to ensure that your template concentration is not too high. I use 1.25ng of the original (100ng) input RNA used for first-strand synthesis (w/ polyA enrichment). I have found this to be sufficient to detect low abundant transcripts as well (down to approximately 0.1-1.0 RPKM if you are validating RNA-seq data).

4. (Bonus) As others have suggested, in most instances you may save yourself significant amount of time and resources if you just discard your current troublesome primers, and design and order new ones.

In the rare case that a suitable primer pair is not available from PrimerBank I design new ones using NCBI Primer3 (see link).

By the way, I work with human genes and use the KAPA FAST 2X SYBR Green mix. I perform the qPCRs in 10uL volumes in 384-well plate and run them on the Roche LightCycler 480 System.

Best.

https://pga.mgh.harvard.edu/primerbank/index.html

https://www.ncbi.nlm.nih.gov/tools/primer-blast/

Sherina,

I find ensembl.org easier to handle than NCBI for finding alternative splice forms.

One suggestion that I have is that I would reduce the number of cycles in the PCR and reduce your input cDNA ie use a higher dilution. Do not reduce primer concentration to below 500 nM (you do not want primer conc. to be a limiting factor in qPCR as this would reduce optimal PCR efficiency and therefore invalidate quantification!)

With PCR you are going to amplify something sooner or later as random stuff like primer concatemers, abortive transcript priming, template switching and so on will kick in at some point, particularly if there is no proper template. It is however quite unusual to get non-specific larger amplicons (usually they are smaller).

This could be alternative isoforms, but the pattern does seem to go from ~130 specific, 260 and 390 unspecific. Also, the unspecific band is in both the ARG and BAX.

So I am wondering if you are doing a Melt curve after your reaction and only then use the samples on the gel? If so, your amplicons may contain some parts of sequence homology that would lead to self-concatemer formation after the PCR and during the Melt curve? A simple solution would be to run the samples on a gel without the Melt curve or the re-denature samples and cool on ice.