weird bioanalyzer results

[newendophytologist]

Hey Folks,

I'm working on a metagenomics project looking at the ITS1 region (~300bp give or take a bit) of fungal endophytes. I extracted DNA from surface sterilized leaves and ran pcr using fungal-specific fusion primers. I took these pcr products and ran them on a gel, and found a very strong band around 150 bp, which I assumed to be adapter dimers. I excised the ITS1 band (~450bp with the 454 adapters) and ran a Qiaquick gel purification followed by an Ampure magnetic bead purification ( I used calibrated beads and an appropriate ratio of the beads to get rid of any remaining fragments of <150 bp).

*note* 9 of my 54 libraries showed no band when run on a gel, but they were excised around the same ~450 bp mark to attempt to recover what little DNA might be there even if it wasn't visible.

Samples were quantified via Qubit HS DsDNA kit. The 9 samples that failed to show a band showed concentrations at or below the lower detection level of the kit. These samples were concentrated in a speedvac and rerun on the qubit yielding more or less expected values given the initial read and how much I concentrated them.

I then diluted all samples to 10^8 molecules/uL. (I could not dilute to 10^7 as suggested in the 454 amplicon library prep manual because 2 of my libraries were below this concentration).

The samples were then pooled at this assumed concentration of 10^8 before qPCR. qPCR showed a concentration of 8.26E+07 copies/uL which I felt was within the realm of variability in the size of these ITS copies for various species and when accounting for minor pipetting error.

This qPCR product was then run on a bioanalyzer and yielded a bit of an odd result (the result is attached). Basically, the strongest peak was at 150bp and there were comparable, though smaller, peaks around 375 and 575 bp.

The higher peaks are odd, in that they are not correlating to the 450 bp band I saw when I ran the libraries on the gel against a 1kb ladder. But they could be artifacts of the 30cycle qPCR or maybe they were within the size range that I selected for when I excised the band, and merely appeared to be around 450bp.

I'm wondering if the 150bp products might from the 9 samples that showed no band at 450 and needed to be concentrated to get a qubit reading. They could possibly be adapter dimers that were present at low levels and are now showing up at significant quantities after concentrating and then running the qPCR.

Or, an alternative explanation is that they are plant DNA that had been amplified by the fungal specific primers mis-priming (merely due to the fact that plant DNA was presumably many times more numerous in the sample than fungal DNA) and perhaps these were present at low levels throughout the cleanup process and have been preferentially amplified in qPCR reaction.

Sorry for the long-winded message, but any thoughts/comments/interpretations/suggestions would be greatly appreciated.

Thanks,
Hank

[Susanne]

How does it look like if you turn off the analysis in the Bioanalyzer (stop button)? Maybe you could also share one of the initial gel images for comparison?

[pmiguel]

Hi Hank,

We are almost certainly in Bennetzen dictum ("Don't waste clean thoughts on dirty data") territory here. But nevertheless...

Did you clean up the qPCR reaction, or just run 1 ul on a bioanalyser high sensitivity chip? What type of qPCR was it?

Looks like the total concentration of your post qPCR sample is about 400 pg/ul or about 5x10^8 500 bp molecules/ul. So that makes no sense after 30 cycles that you would see so little amplification. 30 cycles would give you a theoretical maximum amplification of 2^30, or roughly 10^12-fold amplification. But according to the HS chip, you saw pretty much no amplification.

If you did not clean-up post-qPCR I am surprised you got a result at all -- the SYBR green would likely interact with the fluor in the HS bioanalyzer chip, and the salt in the PCR reaction would probably cause issues for electrophoresis on this chip type.

I wonder if your qPCR machine did a melt curve after performing qPCR? If so, then likely what you are seeing is the dregs of your PCR products that have not re-annealed to each other in aberrant ways -- mainly adapter-to-adapter. At high concentrations they may have formed very long concatamer "daisy chains" that never migrated past the detector of your bioanalyser chip because their molecular weight was too high.

If you ran the post-qPCR samples on an agarose gel, I guess you would see a big smear of high molecular weight stuff. But, to tell you the truth, I would be a little frightened to do so, since those would be high concentration amplicons and might contaminate subsequent experiments you did.

--
Phillip

[newendophytologist]

Just to be clear, I did not run the qPCR or bioanalyzer myself nor do I know much more than the basics of how they work. I did the library prep and pooling, and submitted the sample to our core labs which ran the qPCR and bioanalyzer. The guy who ran my samples has a great deal of experience with these procedures, so I don't doubt he took the necessary steps to get quality reads.

Since my original post I have done a second ampure cleanup on the pooled samples with the same calibrated ratio of beadsroduct (this time I used a much higher quality magnetic separator, which I'm certain allowed for a better cleanup). I also created a second pooled sample that left out the 9 suspected problem libraries, and ampure purified those a second time as well.

Our core labs manager had not informed me that there was a minimum concentration for the bioanalyzer, so I just submitted my samples in the same approximate 10^8 mol/uL dilution. He ran these samples (unamplified) anyways and got the attached results (unamp) which showed a loss or major reduction in the amplitude of the adapter dimer peak. The first two were run on 1uL volume and the second two were on 2uL of product volume. The qPCR was run after this and the full library had 5.08x10^7 mol/uL and the reduced had 3.01x10^mol/uL. These qPCR products were then run on the bioanalyzer and showed reduced adapter dimer (120-150bp) peak, but still showed this peak.

After talking to the Roche customer rep, we think that we are looking good enough to sequence. The rep thinks it will be safer to run the reduced library, which I don't doubt. I was really hoping I wouldn't have to throw out 9 of my 54 samples, and would like to run the full library even with the acknowledgement that much of those libraries could be useless data...it would still help to have something from those libraries. I'm likely oversampling my communities anyways, so some lost data would not be the end of the world. However, the issues with having too many short adapter dimer fragments polluting the rest of the read and essentially ruining the whole GS junior run are not something I wish to encounter. Any advice on how you would suggest to proceed would be helpful

[pmiguel]

Our core labs manager had not informed me that there was a minimum concentration for the bioanalyzer, so I just submitted my samples in the same approximate 10^8 mol/uL dilution. He ran these samples (unamplified) anyways and got the attached results (unamp) which showed a loss or major reduction in the amplitude of the adapter dimer peak. The first two were run on 1uL volume and the second two were on 2uL of product volume. The qPCR was run after this and the full library had 5.08x10^7 mol/uL and the reduced had 3.01x10^mol/uL. These qPCR products were then run on the bioanalyzer and showed reduced adapter dimer (120-150bp) peak, but still showed this peak.

And were the qPCR products subjected to a "melt curve" analysis prior to their loading on the bioanalyzer chip? That is, it is fairly common for qPCR programs to add a "melt curve" analysis after the PCR steps. Although this is pretty much useless for complex libraries, it doesn't hurt anything. However, if it were done the products would largely be expected to re-anneal ectopically because the kinetics of finding their original complement strand are not favorable. So various concatemers, etc, would be expected -- which is what you see.

Since these are ITS amplicons, it is also possible that the "150 bp" peak is a ssDNA full length amplicon that has folded enough to run compactly on the HS DNA chip. Although primer-dimers are always on the differential as well.

--
Phillip

[newendophytologist]

A melt curve was not performed as it was deemed useless. While there is a significant possibility that peak is composed of ss amplicons, it does seem strange that it would be proportionally higher in the full data set, suggesting there is a higher concentration of these 150bp fragmens in the 9 questionable libraries.

If in fact they are adapter dimers, do you think they would cause significant problems with the sequencing at the levels they are showing in the full library? Of course this is after a round of qPCR which would preferentially amplify these short fragments.

Having run 3 rounds of size selective cleanup, I'm shocked to see these fragments still showing up...

[pmiguel]

If in fact they are adapter dimers, do you think they would cause significant problems with the sequencing at the levels they are showing in the full library? Of course this is after a round of qPCR which would preferentially amplify these short fragments.

Who knows? PCR (in the qPCR reaction) would tend to favor the smaller amplicons. But probably so would emPCR. What ampure cut are you using?

Actually, if I recall correctly, you are using the ITS primers for qPCR? Not the 454 primers? That thing you are seeing may just be a primer dimer forming during qPCR, and have little to do with the performance of your amplicons.

--
Phillip

[newendophytologist]

I'm using the Ampure XP beads calibrated to remove anything under ~350bp. The qPCR was done using 454 primers to ensure that only amplicons with the adapters were being quantified, and this number was quite comparable to Qubit readings, suggesting that all (or virtually all) of the amplicons had the appropriate adapter sequences incorporated.

Thanks for all of your helpful suggestions! At this point I will probably go ahead and sequence the reduced library and hope for the best.

Cheers,
Hank

[pmiguel]

"Calibrated"? That would be 1:0.65 sample to ampure -- something similar to that?

--
Phillip

[newendophytologist]

Yes, 1:0.75. The particular batch of ampure beads was run against a standard ladder to "calibrate" particular size selection ratios.

[pmiguel]

Sure, we do that too. Never seen 1:0.75 remove 300 bp, though. Not that it matters, 150 should be gone. Which either means you sample doesn't have adapter dimers in it, what you are seeing are just being generated during qPCR. Or that you sample has adapter dimer strands annealed to your normal amplicons.

--
Phillip