How to study the dynamics of lamina associated domains

Lamina associated domains – or LADs in short – are large, mostly inactive DNA domains frequently associating with the nuclear lamina. In the lab of Bas van Steensel, there is a long history of studying these domains. This has resulted in genome-wide maps of lamina association in various cell types (Peric-Hupkes, 2010), tools for visualization of LADs (Kind, 2013) and even single-cell measurements (Kind, 2015).

All of these observations are based on the DamID method. In contrast to other tools for probing protein-DNA interactions, DamID has proven very successful to determine heterochromatic interactions. Briefly, a protein of interest (i.e. a lamina protein) is fused to the bacterial methyltransferase Dam, resulting in m6A methylation of nearby GATC sequences. This DNA methylation is virtually absent in eukaryotic DNA (see footnote) and can be specifically amplified and sequenced (Vogel, 2007). It is important to note that once methylation is deposited by Dam, it is not actively removed. In other words, DNA that loses its lamina interaction maintains the methylation, at least until the next replication event. This provides us with great opportunities for visualization (again, see Kind, 2013). However, the persistence of Dam methylation limits the possibilities to study any short-time dynamics of lamina association.

Previous work used DamID to map lamina interactions in various stages during differentiation (Peric-Hupkes, 2010). This work showed that LADs are dynamic and correlate with changes in gene expression. Still, it remains interesting to look into the order of events. Is there a specific sequence to reshuffling LADs, and is the lamina association lost before, at the same time or after transcription starts? Fundamentally simpler questions can be asked as well. During cell division, our DNA is replicated, condensed and split between daughter cells. How do these processes affect lamina association, and which LADs are the first to reestablish their lamina association after mitosis?

To answer these questions, we needed a new strategy. We decided to combine the possibilities of DamID with the versatility of CUT&RUN (Skene, 2018). The working name of this hybrid method is proteinA-DamID, or pA-DamID. We start with cell permeabilization followed by antibody-based localization of the Dam enzyme. In turn, Dam activation leads to swift deposition of methylation in the vicinity of the lamina (Fig. 1).

Figure 1. Introducing pA-DamID.
In pA-DamID, cells are permeabilized with low concentrations of digitonin to allow primary antibody access but without compromising nuclear integrity. In turn, the primary antibody is bound by a fusion of proteinA and Dam (pA-Dam). Unbound pA-Dam is thoroughly washed away. All of these steps take place on ice to minimize chromatin movements. Finally, S-adenosylmethionine – the methyl-donor of Dam – is added and nuclei are incubated for 30 minutes at 37C to deposit targeted methylation. Resulting nuclei can be processed for pA-DamID mapping profiles or visualization.

As with regular DamID, the resolution is limited to GATC sequences in the genome. To account for biases in accessibility, we normalize the specifically deposited methylation over in-situ methylated nuclei using freely diffusing Dam enzyme (Fig. 2). 

Figure 2. Example pA-DamID data.
Example data tracks for free Dam enzyme and two targeted pA-DamID experiments. To account for differences in accessibility and amplification bias, data tracks are (log2) normalized over the Dam-only control.

We are still actively developing and validating pA-DamID. Regardless, coming back to the question of LAD reshuffling during differentiation, we have generated initial pA-DamID data at various time points in mouse neural progenitor to astrocyte differentiation. In the resulting data, we can see that most movements take place in the first hours (Fig. 3). We hope to extend this data set soon and hopefully shed more light on the questions raised.

Figure 3. LAD dynamics during differentiation.
Example of normalized pA-DamID data tracks during mouse neural progenitor to astrocyte differentiation. These tracks show clear changes in lamina association in the first hours of differentiation (red arrows). Additional data generation and more detailed analyses are taking place at this very moment.

As a final note, I wish to highlight again that this new method uses antibody-based localization of the pA-DamID fusion, and thus does not require transformed cells. This means that should be easily adapted for primary material. Amongst others, this opens possibilities to study the effect on genome organization in laminopathies – genetic disorders caused by mutations in lamina proteins – and hopefully, better understand these diseases.

Tom van Schaik
– PhD student in the lab

Footnote: Or is it? Low levels of m6A methylation have been reported for human DNA in various papers with various methods (i.e. Xiao, 2018). The mechanism and function remain elusive. For now, it should be enough to note that the extent of endogenous methylation is very low and GATC-independent, making it irrelevant compared to the deposited Dam methylation.