Strange MID reactions, short sequence reads

[npercy]

I've been doing a bit of 454 sequencing in my laboratory have had successfully used the machine in the past with good reads (well 40,000-60,000) and good lengths using Unidirectional amplicon sequencing with various MIDs.

However, I recently designed and ordered new primers from IDT (with HPLC purification), and have had some issues with them. I obtained a set of 14 MID labelled primers (for forward primer), and found that two of them had issues when PCRing, with failed reactions. If this template was used with another MID the PCR worked fine.

The ones that did work were purified, and checked for primer carryover after the purification and there was none visible upon the gel check (unfortanetly I do not have access to a bioanalyser). However when sequencing, I only got around 1,000 good length reads (300 bases, expected size), with the majority of them being lost due to short read lengths?

If I look at the actual sequences of the primers there are a few multiple bases in the primer (mainly TTT and AAA), which are then incorrectly called in the sequencing for a lot of them. What I'm wondering could there be an issue with the primers, and if so how would I prove this to be the case? I have used two different reverse primers, and have seen the same example both times, so my intuituon suggests the forward (MID labelled) primers are the cause, but I can't prove it.

Thanks in advance.

[ajthomas]

There probably is something wrong with the two primers that didn't work. I've had that happen as well, more than once. You had them HPLC purified, so they should be okay, but the fact that they didn't work would suggest that there is some sort of problem with them. When I had that problem, I didn't know whether it had something to do with the MID sequence or if the primer was just made bad, so I ordered a new one, but with a different MID, just to be sure.

Are you using Lib-L or Lib-A chemistry? Because of the way you talk about forward and reverse primers, I'll make the assumption you're using Lib-L. If so, did you reduce the amount of amplification primer you used in the emPCR? If not, that could be the source of the miscalls in the sequence. (If using Lib-A or if you already knew that, you can skip the rest of this paragraph.) When sequencing amplicons with the Lib-L kit, it helps to reduce the amount of amplification primer to 25-50% of the amount called for in the protocol. The reason has to do with the difficulty of sequencing amplicons. When there are many wells in the PTP producing a signal at the same time, as would happen when sequencing amplicons, the camera can be saturated by the amount of light produced, especially when sequencing through homopolymers, and especially in the early cycles of the sequencing run. Reducing the amount of amplification primer limits the number of DNA molecules on the beads and reduces the overall signal strength, thus avoiding the problem. The result of using the full amount of amplification primer when using Lib-L chemistry for amplicons is miscalls in homopolymer stretches, especially in the primer region and 5' end of the read, which is what you describe.

So, I would suggest reordering the two primers that didn't work in PCR, perhaps substituting the MID sequence with some other MID sequence. Also, if indeed you are using Lib-L chemistry, try reducing the primer concentration.

[npercy]

Thanks for the suggestion about the Lib-L and the emPCR amplification primer, I did not know that so have never reduced it in any of the amplifications I have done (usually I have gotten 60,000 sequences with previous primer sets, but there do seem to be a lot of short fails). I'll have to try that next time.

I do not even have confidence in the primers that did work, due to the short sequence lengths. Unfortanetly the costs of ordering new ones for another failed run would put me in the doghouse, so I'm trying to prove the issue whether it is the primers or not, but not sure how to do it.

I did notice one thing, when looking at the sequences after a Shotgun selection (i.e. allowing short lengths), 5 of the 12 MIDs grouped together with errors in the primer sequence (using Qiime for analysis of this). I need to sit down and look at what these errors are, and whether they are consistent, but this suggests to me something strange has occured (never had errors in my primers like this before)

[ajthomas]

When I've ordered primers for this, I haven't had them purified in any special way; I have always ordered them just desalted and it has always worked fine. I can't say it will work for you, but in my experience, desalted is good enough.

If you want to check your primers, you could try running them out on a polyacrylamide gel and see if you get a single band. That might use a lot of what you have, so perhaps you could try cloning your PCR products and sequencing a few of the clones. Alternatively, you might try just sequencing the PCR product by Sanger sequencing using a primer matching the first 20 or so bases of the 454 sequencing adapter. If there are a lot of errors in the primers, you will see dirty sequence in that portion of the sequence.