Background: The organization of eukaryotic DNA into chromatin has a strong influence on the accessibility and regulation of genetic information. The locations and occupancies of a principle component of chromatin, nucleosomes, are typically assayed through use of enzymatic digestion with micrococcal nuclease (MNase). MNase is an endo-exo nuclease that preferentially digests naked DNA and the DNA in linkers between nucleosomes, thus enriching for nucleosome-associated DNA. To determine nucleosome organization genome-wide, DNA remaining from MNase digestion is sequenced using high-throughput sequencing technologies (MNase-seq). Unfortunately, the results of MNase-seq can vary dramatically due to technical differences and this confounds comparisons between MNase-seq experiments, such as examining condition-dependent chromatin organizations.
Results: In this study we use MNase digestion simulations to demonstrate how MNase-seq signals can vary for different nucleosome configuration when experiments are performed with different extents of MNase digestion. Signal variation in these simulations reveals an important DNA sampling bias that results from a neighborhood effect of MNase digestion techniques. The presence of this neighborhood effect ultimately confounds comparisons between different MNase-seq experiments. To address this issue we present a standardized chromatin preparation which controls for technical variance between MNase-based chromatin preparations and enables the collection of similarly sampled (matched) chromatin populations. Standardized preparation of chromatin includes a normalization step for DNA input into MNase digestions and close matching of the extent of digestion between each chromatin preparation using gel densitometry analysis. The protocol also includes directions for successful pairing with multiplex sequencing reactions.
Conclusions: We validated our method by comparing the experiment-to-experiment variation between biological replicates of chromatin preparations from S. cerevisiae. Results from our matched preparation consistently produced MNase-seq datasets that were more closely correlated than other unstandardized approaches. Additionally, we validated the ability of our approach at enabling accurate downstream comparisons of chromatin structures, by comparing the specificity of detecting Tup1-dependent chromatin remodeling events in comparisons between matched and un-matched wild-type and tup1Δ MNase-seq datasets. Our matched MNase-seq datasets demonstrated a significant reduction in non-specific (technical) differences between experiments and were able to maximize the detection of biologically-relevant (Tup1-dependent) changes in chromatin structure.