My Sanger results can not be reproduced by other seq methods...?!

[Anke]

Hi everyone,

I am having troubles with proofing the results I obtained in Sanger sequencing by other methods.
At the beginning of my PhD I isolated RNA of human cancer cells, transcribed it into cDNA and amplified my gene of interest. I cloned it into the TOPO vector and transformed everything into TG1 cells. the DNA of the different clones was then checked by Sanger and I came up with several different versions of the protein. I repeated this quite often, I sequenced forward and reverse, I sequenced the PCR product directly and I used different polymerases and I always found these versions. I was also able to proof their existence with qRT-PCR and on protein level ( western+MALDI).
I then isolated DNA and came up with WT sequences. So we assumed that alternative splicing must take place.
My problem however is that so far any other sequencing method can' t repeat these results. I tried pyrosequencing, which didin ' t work for somehow the primers couldn't bind or for any other reason. Anyway pyrosequencing didn't give me sequences not even WT. I tried to use LNAs but I couldn't get the LNAs to differ between WT and "mutated" version. I tried digital droplet PCR which didn't work either. In the end we also performed RNA seq. The site of interest was only mapped by 3% however no mutated form was found.
We assume that the site of interest has a quite difficult struture and maybe hard to access but still this is really confusing. We already talked to several splicing experts however so far nobody has explanation for what mechanism might take place.
Is anyone having ideas how to explain these controversing data and how to proof the existence by any other method apart from Sanger? ??????????????

[Jessica_L]

Without knowing more about what you're doing, or what kind of troubleshooting you've already done, it's hard to say why your other methodologies aren't working. Since all you've said so far is that a lot of your methods didn't work, the best I can do is ask several questions about your process and maybe something will stick out. The first things I'd look for would be hardware failures, design flaws (i.e. an accidentally reverse complemented sequencing primer, high GC content but without any kind of additive like DMSO or betaine to prevent formation of secondary structure), or issues with the input DNA:

When you isolated DNA, did you check it on nanodrop? What were the 260/280 and 260/230 ratios? Did you run it on gel? Did everything look okay?

Have you tried running a gel of your PCR products prior to pyrosequencing? Did the PCR at least work as you expected? Have you re-checked your sequencing primer design? What instrument were you running on? Did you have any kind of positive controls on your run? Did those work?

What sequence did you design your digital PCR primers and probes against? Or are the primers the same as what you used for the original Sanger sequencing? Again, did you run any positive controls? Did those work? What instrument(s) are you using? What were your final droplet counts (positive, negative, and total)?

Getting into NGS is a whole different set of questions: how did the RNA-seq library prep look? Presumably you were able to quantify it via qPCR before you loaded the sequencer-- what was the concentration of your final library? Did you check it on the bioanalyzer, too? What kind of sequencer did you use? If it was an Illumina instrument, did the other lanes or did the PhiX control look okay?

For all I know you've already looked at all of these possible problems and ruled them out, but without knowing more, this is where I'd start. I hope that helps.

[Anke]

hi, thx for answering
- The sequence is indeed having a high GC content(>85%) however I am using betaine and so far that worked quite well.(DMSO is not helping it rather makes things worse)
- I checked the DNA on nanodrop and the values seemed ok, same for the gel, I also performed qRT PCR against the versions I found and this was also positiv, I even resequenced the RT PCR Product and it was the right sequence ( no WT)
-I did run a gel before pyrosequencing, this was ok, as positive control I used cells that are stable transfected with the version I am interested into. I also tried lots of different primers and changed all components in case of contaminations ,but it never worked.
- In the case of digital droplet (Bio-Rad QX100, Light cycler 480 software)), LNA and RNA seq I cooperated with people that should be experts .We tried several different primers, we again used stable transfected cell lines as positiv ctr as well as vector DNA ( linear ) but we came up with a high background ( also we added DNAse) and the probes (one probe against WT and one against the "Mutant") couldn't distinguish betwenn the 2 forms. Therefore we started RNA seq. Unfortunately this was done by another group and I don't know much about it. I prepared the RNA and gave it to them, they got rid of ribosomal RNA and used the high seq 2005 (illumina) with galaxy tools + topHAT. The concentration of the library was checked with the bioanalyzer however I don't know it. The result wasn't so good either, ingeneral was the mapping very low ( 3%) and the "mutant" I am interested into couldn't be found. However at the beginning I also found an already quite well known splice version of the protein ( with Sanger) and that one could be detected with the RNA seq.....
Right now I tried more or less everything I could think

[Jessica_L]

To be clear-- your initial experiments isolated mRNA from cells, which you reverse transcribed into cDNA and sanger sequenced. Now, for pyrosequencing, you tried isolating DNA straight from the cells to see if the splice variants you're observing were a result of RNA processing or a DNA mutation, right? Have I got that correct?

When you say the positive controls for pyrosequencing were stable cells, did you isolate DNA from those and use that as a positive control or did you reverse transcribe RNA from your gene of interest again? Also, it's not clear from your response if your positive controls worked properly. Did they? Also, I'm not at all surprised you'd be having trouble getting usable sequence data if the GC content is really >85%. Betaine, in my experience, works for some things, but not all. It's been a while since I've done any pyrosequencing myself, but you may want to make sure Betaine won't interfere with the sequencing chemistry. I'm sure the manufacturer of the instrument you're using has recommendations for how to deal with GC rich templates. At the very least, assuming everything was designed correctly, you should be seeing some sequence, even if it's not the answers you're hoping for.

For the digital PCR, I'm not sure what you mean by Bio-Rad QX100, Light Cycler 480 software-- those are different platforms for ddPCR and I'm not certain I understand what you're trying to say. Did you run your assay on two different platforms?

I'm also trying to understand the design of your experiment here-- you have a probe against the wt sequence, which would hybridize to DNA at the splice junction, assuming that sequence is unique and that the protein variant (mutant) would have a different sequence at the splice site? Or maybe your design is different...? It's hard to say why your ddPCR didn't work without more info. Based on what I think makes sense for your assay design, again, assuming everything was designed correctly, you should see hybridization on at least one of the probes, if not both. In any event, high background is something I've observed on a few custom assays. I think it may have to do with either the probe concentration or the type of reporter dye being used. It can be different between assays, though, so high background is not necessarily bad. Why would you add DNase, though? It's my understanding that digital PCR usually has a reverse transcription step when the starting material is RNA... Obviously, if you were working with collaborators, you may not have the answers to all the questions I might ask, either. Have you tried contacting BioRad (or whoever manufactured your instrument)? I've found their tech support and field scientists to be very helpful.


For the RNA-seq:
If you QC'd your RNA-seq library with the bioanalyzer and the results weren't very good, that might be a reason that the RNA-seq didn't work better. Assuming the HiSeq run itself was fine, the problem is probably with the library. For the library prep, what was your RIN value when you sent in your RNA? When you say mapping was 3%, do you mean you covered only 3% of your target gene or only 3% of your reads mapped to your reference sequence? What were the Q scores like for the run? What kinds of cluster counts were on the flow cell? In other words, how many reads did you get from your run on the HiSeq? I understand you were working with collaborators for the sequencing, so you may not have all this info, but the reason I'm asking is that low cluster counts might indicate library issues or a problem with template hybridization during the clustering process. That may or may not be related to the high GC content of your gene of interest. That's also why I was asking if there was a PhiX control run on the HiSeq-- that can help rule out reagent and clustering and hardware issues. Again, you may want to consider contacting Illumina tech support to help with some of this.

[Anke]

Thank you very much for all these suggestions and ideas. the problem is that I am really not having all these informations for everything was done in cooperation with people that should be experts in these methods and in a Core facility.
Right now we focus on the RNA seq cause pyroseq and LNA probes didn't work, so maybe we start by "solving" the problems there

Maybe I expressed myself a little bit confusing but I performed sanger with cDNA and RNA seq with the mRNA which I used for the RT into cDNA. So basiclly both probes had the same mRNA.

when I say mapping = 3% I refer to them only covering 3% of this region. (according to my collaborators they only got few reads at the site of interest ca 3%)...
So far I can't answer your other questions. However I will transfer them to my collaborators.