ChIP-Seq analysis is used to study the control of gene expression by identifying DNA regions that are associated with chromatin at the time of cellular analysis. The ChIP-Seq procedure typically involves cross-linking DNA to the associated chromatin proteins, sonication to fragment the chromatin, and immunoprecipitation of chromatin using an antibody to DNA Polymerase II or a transcription factor. The DNA regions associated with chromatin are identified by DNA sequence analysis. To prepare the DNA for sequencing one must dissociate and purify the DNA away from chromatin, create blunt ends, add poly(A) tails, and ligate adaptors for PCR amplification.
Importance of size selection with high yield
A DNA size selection step is done to improve the quality of the input sample for sequencing. PCR amplification is performed for a limited number of cycles (18-20) to increase the amount of material available for sequence analysis. Using the Pippin Prep for DNA size selection, one can get a tight size distribution and a high yield (50-80%) in the selected size range. The high yield is particularly important because it means that fewer PCR cycles are needed, which in turn means that the amplified sample will have less sequence bias and be more representative of the original DNA sample.
Appearance of an apparent ‘secondary’ larger band after DNA size selection
Because of the very small amounts of DNA present in ChIP-Seq samples, amplification is sometimes performed before rather than after the size selection step. In these cases some users have been surprised to find a secondary band of larger size when the size-selected and PCR-amplified DNA is analyzed on an e-Gel (Life Technologies) or FlashGel (Lonza) system. What could be leading to this observation?
One plausible explanation for this phenomenon is the presence of some single-stranded DNA (ssDNA) in the amplified DNA. A certain amount of ssDNA can be generated among the dsDNA reaction products of PCR if too many cycles of amplification are performed or if the primer concentration is too low (1). Below we present our interpretation of this phenomenon. We also welcome you to comment on this blog post and share your personal experiences that may be pertinent.
How can over-amplification lead to ssDNA in the sample?
A ChIP-Seq library contains a tiny amount of immunoprecipitated DNA, so it is tempting to over-amplify to be sure of getting sufficient material for sequence analysis. Unlike traditional PCR in which a single-copy gene is amplified from a small amount of DNA, all of the ChIP-Seq DNA contains adaptors and is eligible for amplification using the universal primers. Thus the primers are quickly consumed. As the primer concentration becomes limiting, one of the two PCR primers may be better able to anneal than the other, and this can lead to accumulation of some ssDNA within the population of amplified dsDNA (2).
Furthermore, in the later cycles the amplification products can accumulate as denatured ssDNA strands. (The two strands can fail to reanneal because the amplicons are diverse in insert sequence even though they have the same primer binding sites. Therefore most or all of the library amplicons may not be present at sufficient concentration to reanneal during the lower temperature steps in the PCR cycle.
In a preparative agarose gel system (with a low voltage gradient) ssDNA fragments could potentially co-migrate with dsDNA fragments. When a portion of the size-selected sample is then run on an analytical agarose gel (with a high voltage gradient) ssDNA fragments could potentially migrate differently than dsDNA fragments. Personal communications from some users tell us that this explanation is plausible. When a ‘secondary’ DNA band of apparently larger size was re-amplifiied and analyzed on an analytical gel, only the smaller, primary band was generated unless the sample was over-amplified. In this case, sequence analysis also verified that the sequence of the upper and lower bands were indistinguishable.
In general, limit the number of PCR cycles to 18-20. If ‘secondary’ DNA bands are seen in the size selected sample, try reducing the number of PCR cycles. This may have the added benefit of reducing PCR-induced sequence bias.
Contribute your experiences! As mentioned above, we encourage you to comment on this blog post and share any experiences that may be pertinent to the discussion.
- Roux, K.H. Optimization and troubleshooting in PCR. Genome Res. 1995 4: S185-S194 Link
- PCR Troubleshooting web site published by The PCRCore Facility at Children’s Hospital of Philadelphia Link
Follow @DNAsizing on Twitter