is it possible that the large band is a heteroduplex of either a polymorphism in the middle band or of the middle and small band. It would move slowly in a gel so look larger and if the design of the sequencing primer was unfortunate then only the one sequence would be sequenced....did you sequence from both directions and is there a polymorphism in the sequences?

Hi Patrick,

It is possible, how big is the amplicon and what cycling conditions are you using? If the bottom band is a lot more diffuse it is more likely that you are looking at double-stranded and single-stranded PCR product, I have seen this in the past.

Hi Steve, thank you for your answer. If it was the case of a single stranded versus double stranded product, wouldnt the lower band appear to be half the molecular weight?

My product is 1kb, I amplified for 35 cycles using Q5 polymerase.

I attached the image of my PCR products ... the lane next to the ladder is my control (wild-type) condition. sequencing results appear to indicate that the top two bands are identical (corresponding to full length mRNA), while the bottom band is a splice variant. However I should say that for the top and bottom bands I have only gotten quality sequencing for 90% of the product, the beginning and ends are missing.

Cheers!

is it possible that the large band is a heteroduplex of either a polymorphism in the middle band or of the middle and small band. It would move slowly in a gel so look larger and if the design of the sequencing primer was unfortunate then only the one sequence would be sequenced....did you sequence from both directions and is there a polymorphism in the sequences?

• You need the full length sequence to be sure, even if that means you clone and use universal primers > 30bp up stream and down stream to be sure
• If this is as hetero duplex as Paul has suggested or some sort of hair pin loop affect then denaturing the sample should eradicate that:
• In the old days to render double stranded DNA single stranded we would simply tear with 1mNaOH for 5 min and then column purify treated product; which you can then run on a gel
• Alternatively, you could think about how RNA samples were treated for agarose gels to eradicate secondary structure (which RNA is most prone to) but can also apply to DNA:
• That is denature the sample with running buffer + 10% formamide at 80C then snap cooll on ice then load and/or run your samples on a standard agarose gel but with 10-20% formaldehyde. That is an old style Northern gel designed to eradicate secondary structure and keep it eradicated during electrophoresis
• Preparing these gels however requires pouring formaldehyde into agarose in a fume hood in a water batch at 60C before casting the gel in fume hood and running as you would a standard agarose gel; except the samples migrate at about 1/3 speed @ 100v so allow 2-3 hours for separation
• This is fiddly and nasty however (brings back memories!) so I would start by denaturation of your sample in formamide running buffer which you then run on a standard 0.7% agarose gels and also clone fragment to obtain full length sequence. I think latter is crucial

Paul Rutland thank you for this suggestion. I am relatively inexperienced in DNA electrophoresis - I am wondering how the dsDNA could form a duplex and how a polymorphism would influence this?

I have sequenced in both directions, but of course I don't get reads for the full-length product... when I combine the sequences from both directions to get coverage of the entire product, they appear identical (no apparent polymorphism). Can you elaborate on what you mean by unfortunate sequencing primer? Thank you!!

Laurence Stuart Dawkins-Hall greatly appreciate all of these suggestions. I have actually just cloned the upper and middle band into a plasmid to try and get sequences for the entire 5' and 3' ends. I am curious what types of 5' or 3' changes I might see that could affect migration in this way?

The denaturing conditions are a nice idea, thank you... If I can observe just two bands in these assays that will be very helpful because when it comes down to it I am not sure I can publish these RT-PCR data and claim that the two upper bands are identical (if the sequences do end up aligning 100%).

• Clusters of G/C residues that could form hairpin loops at the ends
• Or as Paul has suggested

Laurence Stuart Dawkins-Hall thank you!

Dear Patrick

A heteroduplex typically forms from 2 alleles with a different snp base at a given position. Then the dna forms a duplex everywhere except where both strands have the same base so no Watson crick pairing is possible. The dna forms a bubble at this point and the strands open up locally so that the new heteroduplex runs slowly through agarose gels. I was suggesting that this is possible but also that if there is great homology between the long and short strands in your case then a different type of heteroduplex may form with all of the short product annealing with half of the long product with some single stranded unbound dna remaining of the lonf product. In my experience of SSCP gels single stranded longmers usually run slower than double stranded dna so I think that a ss/ds mixed molecule could run slower ( so looks larger) than the ds normal product

What I meant by unfortunate primer design is that it may have been possible to design ,say, a forward primer that sequenced both the long and short ( alternatively spliced) products but only annealed to the full length product reverse sequence and not to the short product then the reverse primer would only sequence the full length product giving the impression that the short product was not there.Possibly also one sequencing primer may have a snp near its 3' end so not sequence one allele at all again giving the effect that only one sequence exists while 2 may be present Alternatively if the amplimer is large then the primers may have been designed to sequence up to a position short of the SNP polymorphism so with an incomplete sequence it may look perfect but still have a polymorphism in the unsequenced region. It wil be very interesting to see your sequences. Meanwhile you could cut out the large band,denature it and renature with heat and see if it splits back into two bands which would indicate that either shape effects or het effects are present in this instance

Paul Rutland very interesting, thank you for this insight and the detailed feedback/explanation. I did not realize a snp could cause such a profound effect on electrophoresis, I will have to check more closely to make sure there are no polymorphisms. As far as the other heteroduplex possibility, the longer product corresponds to a 4 exon long mRNA that is identical to the lower band except the lower band is lacking exon 2. so they do have plenty of complementary sequences but based purely on size it seems like if the bottom band (~750kb) and middle band (~900kB) were forming a duplex it should run higher on the gel, as the top band is only 1kb.

ok Patrick Corbin Mullen so I will add the possibility that the middle band is the ss/ds heteroduplex to my answer running at intermediate size so if you have time to try isolating and denaturing then renaturing the bands to see if they split into 2 then you could include the middle band and see if this one could be a hybrid molecule

thank you Paul Rutland I will give that a try this week.

I had exactly the same issue. I asked to an expert and he confirmed those are heteroduplex resulted from the association of exon-including and exon-skipping PCR products. For unknown reasons when the products are run into 15% TBE polyacrylamide gel the 3rd band is not present anymore or run far from the expected bands.

Leila Alieh very interesting ... I am thinking of trying different gel systems to see if that alters the band pattern. however if you look at my gel image above, two bottom bands seem to big to produce a top band of that size. a duplex between the exon skipped product (~780bp) and the middle band (~850bp) seem like they should result in a higher molecular weight.

Hi Patrick.

I am not sure what the gel you posted represents, what did you load in the different wells? The kind of heteroduplex i was referring to are like the ones you can see in this paper,( figure 3 shows a regular agarose gel: the upper band in CERS6 is a heteroduplex)

https://www.nature.com/articles/s41559-019-0813-6/figures/3

Is that your case?

The size of the unexpected band does not tell much, since the 2 band products are only partially complementary, so a portion of the strand will loop out and cause an irregular run.

I also tried to denature my samples and run them in alkaline agarose gels (in NaOH), but although I could not see the extra band the resolution was much worse (critical factor in my case since the exons I was investigating were pretty small) and the process was laborious (not to mention that the alkaline conditions are bad for the electrophoresis chamber).

I also tried to extract the bands, clone them with the TA system and sequence them, but it all failed. If you are able to sequence your expected product that would help to confirm their identity, but I would not expect a successful sequencing for the heteroduplex.

Another strategy to confirm the identity of your bands is to perform a control digestion.

In any case I would not invest too much time on that. What is the purpose of your experiment? If it is a simple validation of sequencing and the unexpected band disappears or is shifted when you run in polyacrylamide gels you can safely conclude it is an artifact and just focus on the proportion of the 2 expected bands (actually you should measure the intensity of the heteroduplex and add 1/2 of it to the measure of each of your expected bands, assuming you are working on a single exon included/skipped).

Good luck!

Leila Alieh thank you very much for your thoughtful response, it is useful to see a publication where this issue was encountered and addressed.

The experiment I am doing is comparing RNA expression in control versus patient samples using RT-PCR. So I really cannot just ignore the top band without knowing exactly what it is.

I have isolated the top and middle bands and cloned them into a plasmid for sequencing. So far, the sequences look identical but I only have ~85% coverage of the sequences.

thanks again!