NexteraXT - how is it possible to dual barcode by PCR?

[M4TTN]

I understand the principle of multiplexing a single molecule by placing two different indexes (A on one end, B on the other). What I don't understand is how the Nextera and NexteraXT kits manage to achieve this (via PCR).

i.e. How does the kit manage to create molecules that only have A and B barcodes on each respective end (and not molecules with barcode A on both ends, and others with barcode B on both ends).

Illumina tech support responded and said it is due to the NPM buffer containing "suppression PCR primers". Which are what exactly?

Does anybody here have any insight?

[SNPsaurus]

If a fragment has A adapters on both ends, during PCR the strand will hairpin since you have the A adapter on one side and its complement on the other side. This will reduce the effectiveness of primer binding and amplification. I think that is what is going on.

[creeves]

Actually, I think AA and BB fragments are present (25% of each plus 50% of AB fragments), but cause no trouble because bridge amplification will only occur for fragments that are AB. An AA or BB fragment can hybridize initially to the flow cell, but the other end cannot hybridize to the other oligo on the flow cell, leaving one single strand, which is not much of a cluster.

[SNPsaurus]

That helps as well. But if half the library is non-cluster forming fragments, that would affect the calculation of what to load after checking the library by qPCR since AA fragments would happily amplify as well. I don't see a difference in calculating the loading concentration between Nextera and protocols that enforce A-B fragments by other means.

See this: http://www.evrogen.com/img/PS-effect-large.png

[M4TTN]

Further response - directly from the keyboard of the Illumina tech support:

"This is accomplished by the presence of 'mosaic ends' in the primers. The Index Primers (from the Nextera Index Kit) extend the adapters in PCR to allow the template to hybridize to the flowcell. The primers in the PCR Primer Cocktail (PPC) amplify the template, but the mosaic ends have a higher annealing temperature than these primers. As the temperature decreases during PCR, those template molecules with the same adapter on both ends will preferentially snap shut (forming a pan handle-like structure). The correctly formed libraries, on the other hand, will be amplified."

So, yes, it appears that prior to primer annealing, molecules with identical ends anneal to each other creating hairpins that prevent annealing by the amplification/indexing primers.

I'm surprised it works given that even nonidentical ends still have the same Read1/2 primer sequence within them (which I understand to be identical). And since annealing in PCR is at 55˚C, I would have thought these regions would anneal, and thus prevent binding by the indexed primers.

But I think there is more going on. The tech does mention "primer cocktail" and in an earlier email to me:
"The NPM not only includes amplification master mix, but it also includes suppression PCR primers to selectively amplify only those molecules with adapters in the correct position/orientation.
So, in reality this is a four step PCR reaction with suppression primers provided in the NPM and the indexes added separately."

But they are unable to give more detailed (proprietary) information.