probe fluorescence curve nose-dives after peak - (rather than plateau-ing: image attached) (May/12/2008 )

hey,

i'm optimising a duplex real time PCR on the SmartCycler System for a bacterial pathogen and a b-globin internal control. The pathogen PCR works beautifully as a uniplex AND duplex, with wonderful peaks, but when i run the b-globin, the peak gradually nose-dives (about uhm, a 30deg angle?) rather than plataeu-ing. The curves Ct are about 22-26 cycles.

i ran a a titration of primers (100-400nM) and probes (100-300nM) for b-globin, and ALL the curves are exhibiting the same behaviour. please see attached image. (i think the off-the-charts curve is a strange artefact, i have yet to duplicate run)

the reaction has NaSO4 in it, so a MgCl2 titration has no effect, but i'm wondering if it is also responsible for this behaviour in the b-globin?? or perhaps it is an annealing temp issue? i have optimised a couple of RT PCRs before, but have never seen this effect. this is also the first time i've had to work with NaSO4 in the mix (comes in the commercial MM i am using).

my pathogen probe is labelled with FAM/BHQ1, whilst my b-globin is labelled with Cy3/BHQ2. hopefully, that won't be the problem, but i thought i'd mention it anyway. i didn't design the primers, this project was put in my lap b/c the person developing has gone on annual leave.

pathogen:
primer 1 Tm: 66.41
primer 2 Tm: 65.68
probe Tm: 73.79

b-globin:
primer 1 Tm: 60.28 (could be issue?)
primer 2 Tm: 65.88
probe Tm: 75.22

annealing temp: 60
extension temp: 72

any ideas? do i need to provide other info?


cheers,
agnieszka.

AFAIK you don't have problems with a "diving" curves, but your raw data baseline substraction. Software takes a range of initial cycles (usually 2-5) as a horizontal background, and remodulates the graph according to that.
In case the fluorescence reads in these cycles are not linear but slightly increasing, reformating makes your curves look like they dive.

If you can look at the raw data chart, you will see if I'm right. Anyway, if your software doesn't have problem computing a Ct value from a curve, you don't need to worry about that.