Designing nucleosomal force sensors

About three quarters of our DNA is wrapped into nucleosomes: DNA spools with a protein core. It is well known that the affinity of a given DNA stretch to be incorporated into a nucleosome depends on the geometry and elasticity of the basepair sequence involved, causing the positioning of nucleosomes. Here we show that DNA elasticity can have a much deeper effect on nucleosomes than just their positioning: it affects their “identities”. Employing a recently developed computational algorithm, the mutation Monte Carlo method, we design nucleosomes with surprising physical characteristics. Unlike any other nucleosomes studied so far, these nucleosomes are short-lived when put under mechanical tension whereas other physical properties are largely unaffected. This suggests that the nucleosome, the most abundant DNA-protein complex in our cells, might more properly be considered a class of complexes with a wide array of physical properties, and raises the possibility that evolution has shaped various nucleosome species according to their genomic context.

Figure 1

Top: two unwrapping states of the model nucleosome under tension, state $\left(0\right|5)$ (left) and state $\left(4\right|4)$ (right). Bottom left: energy landscape (in units of $k_{B}T$) of the nucleosome at position 826 of the YAL002W gene of S. cerevisiae under an external force of $14\mathrm{pN}$. Note that single wrapped states like $\left(0\right|5)$ are located in a metastable valley. Nucleosomes with just half a turn of wrapped DNA [e.g., $\left(4\right|4\left)\right]$ form a ridge in the landscape. Bottom right: designing a special nucleosome: result of a free MMC simulation on state $\left(4\right|4)$.

Figure 2

Energy along the ridge for sequences found using MMC on position 826 of the YAL002W gene of S. cerevisiae, held in unwrapping states $\left(4\right|4)$ (left) and $\left(1\right|7)$ (right). The solid line represents the original ridge. The dashed line is the ridge after free MMC and the dotted line after SynMMC.

Figure 3

Distributions along the nucleosome of AT-rich dinucleotides (AA, AT, TA, and TT, frequencies summed) from an ensemble of low-energy sequences of the fully wrapped nucleosome and of one in the $\left(4\right|4)$ unwrapping state. The central part, which is wrapped in both cases, is identical. A phase shift occurs in the perpendicularly bent unwrapped tails.

Figure 4

(a) Nucleosome energy landscapes in a small neighborhood of position 826 of the YAL002W gene, with the 826-sequence replaced by the sequences found through SynMMC at states $\left(1\right|7)$ and $\left(4\right|4)$. In each case, the replacement sequence still provides a local minimum. (b) Cyclical energy landscapes of sequences found through free MMC for states $\left(1\right|7)$ and $\left(4\right|4)$ compared to the sequence at position 826 of the YAL002W gene. There remains always a strong local minimum at position 0.

Figure 5

The height of the unwrapping barrier as a function of the tension applied, for the original genomic sequence, and the sequences we found through optimization for state $\left(4\right|4)$ at 14 pN, using both free and synonymous MMC. Though the sequences were optimized at a given force, they lower the barrier for the entire range of forces considered.