Chimera generation across transposon: how to avoid?

[pag]

I am attempting to use PCR followed by Sanger sequencing on genomic DNA (derived from a single colony, albeit grown to large volume liquid culture) to attempt to bridge gaps in the (454-generated) genome using the ends of predicted pairs of contigs. As I was not able to generate PCR products for all predicted gap bridges (only one failed) I tried re-pairing combinations of the primers to allow for alternate genome arrangements. Unfortunately when I did this with other combinations of primers that themselves generated sequence-able products (i.e. I had paired A+B and C+D before, now I pair A+D and B+C where A,B,C, and D are all single-copy), I generated products as well. Sequencing the gaps revealed a transposable element (I'm not sure if it's an "active" transposon or not), so I think the transposon is actually acting as a better PCR primer than my primer itself after the first cycle (see attached crude drawing).

The rub is I still haven't found ANY partner on one end of the unclosed gap, although a TAIL-PCR indicates that I likely have a copy of the transposon just "upstream" of that contig. The other end of the gap without corresponding PCR product appears to have a copy of the transposon followed by copy of another repeated contig.

The transposon appears to have a GC content of approximately 54%. My genome as a whole appears to have a slightly higher GC content (56%). The Tm of my primers as calculated by consed are 58-60 degrees. By IDT's formula with 50mM NaCl, the Tms are a bit lower (~52-55). The length of the transposon is slightly less than 1KB.

Is there a method of increasing my stringency and avoiding the chimeric product generation? Should I instead sheer my genomic DNA and generate a paired end library?

My current reaction conditions (which reduces many of the supposed "chimeras" to a weaker band as seen an agarose gel, while producing strong products for MANY of my supposed "correct pairings") include an annealing at 65 degrees for all cycles. At 55.5 degrees, the "chimeras" had about equal intensity to the "correct pairs."

volumes in microliters
Rxn volume: 50
10x Buffer (no Mg): 5
20mM MgSO4: 5
10mM Ea dNTPs: 1
Taq polymerase (5U/uL): 0.25 (1.25U)
Primers (5 micromolar): 2 each

cycle conditions: 30 cycles
denature at 92 for 45 seconds
anneal at 65 for 45 seconds
extend at 68 for 4.5 minutes
(plus about a 5 minute initial denaturation and a final extension of 5 minutes at 72)


The following steps have been suggested: 1) decrease template concentration drastically, 2) attempt a gradient PCR from 50-70 annealing for a "known positive" involving the chimera. I've also thought of 3) decrease cycles to 25 4) some form of competitive binding: have three to four primers in the reaction and hope that the "best" product is generated most frequently.

My issue with suggestion #2 is I'm not 100% certain of ANY of my products that span the transposon right now. How do I know that I'm not actually favoring creation of a chimera, given the annealing temperature is far in excess of the calculated Tm for the primers? But I would think that I would generate even more non-specific products if I decrease the annealing temperature (basically, I was always under the assumption that to increase stringency you increase temperature to the point where all bands except your target product drop away). #4 seems likely to backfire and produce even more complicated chimeras or multiple same-sized amplicons that are sequenced with the same primer.

Clone library preparation from the PCR products isn't really a solution: the chimeric products have already been generated. Is there a prescribed method of avoiding chimera creation while still using a PCR-based method? Or should I sheer my genomic DNA and ligate it either into a vector or an illumina-style paired-end oligo to generate a library? I fear this will add months and/or great expense onto my project.

Note: based on coverage estimates from the 454 data, the transposon is predicted to be present in 30 copies in this genome (prokaryotic, presumed circular). Thusfar I've identified 15 likely locations for this transposon in the genome. A related species (same taxonomic Family) has 3 copies of a homologous transposon.

[krobison]

One trick that can solve difficult templates is to use a more processive polymerase, especially one of the ones engineered with a clamp (Pfu-UltraII for example). Chimaeras may be forming if you are having incomplete extension on a cycle, and that is priming somewhere else.

Did you look through your 454 data for anything resembling the transposon? Perhaps a primer inside the transposon might work when paired with one of your outside primers

[pag]

thanks for the tip about the processivity of the enzyme. Do any polymerases require the 3' end of the primer to be "clamped" before they will extend, or do they only care about the previous base (or previous turn of the helix)?

If what I illustrated is what is going on, I'm not entirely sure that a more processive enzyme will help, but it couldn't hurt (other than being slightly more expensive per reaction)

Yes, I do have primers designed from the middle of the transposon pointing each way. I've only, to date, been using this for sequencing rather than amplification as I fear that with an average of one transposon per 100KB, I'll probably have quite a few self double-priming products (I know of one likely location for this off the top of my head). But that's what single primer negative control reactions are for - better to test than assume. Also, if I get a 400-600bp product, it's probably the right thing.

The transposon is in my 454 data (hence the 30-copy estimate) in the form of three smaller contigs that had been filtered from my original data set (I think our collaborator assumed them to be virus or other extragenomic, non-plasmid DNA).