Is there a genomic code for nucleosome positioning?

Question

What does a genomic code for nucleosome positioning in eukaryotes actually mean? By the code is it right to think that specific DNA sequences favour nucleosomes and others don't? I see that there for and against arguments on this topic. What is the current view on this topic?

Answer 1

The genome is the complete set of DNA in an organism, including genes and non-gene sequences of base pairs (bp).1 Each codon of three base pairs in a DNA sequence specifies one of twenty different amino acids. There are four available bases in DNA; Adenine (A), Thymine (T), Guanine (G), and Cytosine (C). Four letters taken three at a time (where order matters and repetition of a letter is permitted) gives 64 possible codes. There are only 20 amino acids necessary in most life on earth, so only 20 codes are needed from the 64 possible. One codon is also used as a start codon (start translation) and three codons are used as stop codons (stop translation). So there are 41 extra codons available. As it turns out, for most of the 20 amino acids several different codons can specify the same amino acid, but each codon only specifies a single, unique amino acid, so there is no ambiguity. The code is therefore said to be redundant or degenerate. This degeneracy allows the same protein (composed of amino acids) to be coded with different base pair sequences, which is important if one base pair sequence is better than another in its mechanical properties of bending DNA around a histone octamer to form a nucleosome.

If the DNA molecules in a typical human cell were lined up head to tail, they would form a string about 2 meters long. The diameter of a cell nucleus is only 5 μm (0.000005 meter), so the DNA really must be packaged or organized to fit. The packing solution eukaryotes developed is to coil DNA around histone proteins to form beadlike units called nucleosomes (packing the DNA also has implications for gene expression, but we will ignore that here).

Each nucleosome constains a 147 bp (base pair) stretch of DNA, which is sharply bent and tightly wrapped in a 1 and 3/4 turn (a left-handed superhelical turn to be exact) around the histone protein octamer. There are many of these nucleosome beads on each strand of DNA. Human chromosome 19, for example, has 58.6 million base pairs, so that makes for a lot of 147 base pair beads, even with streches of unwrapped linker DNA between the beads--about 75 – 90% of the base pairs end up wrapped in nucleosomes.

The bending of the DNA around the histone is made easier or more difficult depending on the geometrical and mechanical properties of the specific two base pairs or dinucleotides at the bend. This is not a trivial mechanical property: Some sequences may have a more than 1000 fold capacity to bend around the histone octamer compared with other sequences; we will say these sequences have an affinity for nucleosome positioning since they make it easy for the DNA to wrap. Conversely, some sequences are very "stiff" and are much more difficult to bend, e.g. poly dA:dT sequences. On average, high affinity sequences have more AA, TT and TA steps at positions where the minor groove faces inward towards the histone octamer, i.e., when more bendable di-nucleotides like AT and TA occur on the face of the helical repeat where it can directly interact with the histones (the helical repeat is about every 10 bp; see below).

What do we mean by "grooves?" Remember that DNA is a double stranded helix (as if you were twisting a flexible ladder) coiling two chains of sugar-phosphate "backbones" around the outside of the helix with the nitrogenous base pairs that connect the strands (with hydrogen bonds between base pairs) pointing toward the center of the helix (if you were twisting a flexible ladder into a helix the rungs would be the connected base pairs beween the two backbones). If you held up a screw you would see the threads as "grooves" in this similar helical structure. The backbones of the two strands are closer together on one side ("minor groove") of the double helix than the other ("major groove"). In undeformed β-DNA, the complementary base pairs of the "rungs" rise about 0.34 nm and twist ~36 degrees around the center axis of the "ladder" from each base pair rung to the next. A "helical repeat" will therefore occur every 10 bp's (10 x 36 = 360 degrees) as the helix completes a full turn around its axis. Don't confuse this helix with the wrapping of the DNA around a histone core; the DNA is a double stranded helix and that is in turn wound around the histone in a nucleosome.

But do nucleosomes in actual living organisms ("in vivo") actually tend to position themselves on sequences of DNA that wrap more easily around the histone core of the nucleosome, i.e., do nucleosomes in vivo actually show an affinity for specific DNA sequences? If they do, then that could be considered to be a "genomic code for nucleosome positioning." There are other factors (e.g., remodelling complexes which can move nucleosome positions) and related questions, e.g., do stiff sequences like poly dA:dT inhibit the formation of nucleosomes (and are nucleosome inhibitory sequences more common in DNA sequences like transcription start sites that should be exposed to facilitate interaction with trans proteins), but we will just consider simply the basic question.

By analogy, suppose I had a printer that could only print strips of paper a sentence high (vertical height of the strip sufficient to contain a 10 pitch character, say 0.1 inch high) but of arbitrary length. Imagine the printer is fed with a spool of this strip paper, like an old time printing calculator (or ATM receipt printer). If I printed this answer text page, out would come English sentences (at least I intend this to be recognizable English) on a long, thin strip. This sounds rather like a phylactery (tefillin) printer, now that I think of it, but that is a topic for comparative religion. If there are 66 lines of 7 inch long sentences on my page, then that is 38 feet of 0.1 inch paper strip to carry around for every page of text. I might find it more convenient to wind up portions of that long strip on little wooden dowels (like empty sewing thread spools for example) and then unwind them only when I wanted to read my article again (which I can never do enough). If DNA is kind of like this analogy (it is far more complex actually), then the spools are "histones" and when wrapped with paper strip become "nucleosomes." Does it makes sense that the letters I chose to construct the words in the sentences of my article were chosen to some extent based on some hypothetical capacity for some sequences of alphabetic characters to bend and coil around a spool (say some sequences had less cumulative stiff ink drying on the paper strip)? That gives you one angle on the controversy involving the idea of nucleosome positioning by genomic code. However, consider that, as in the case with degenerate DNA codons, I can choose different sequences of characters, i.e., words, to mean the same thing, so possibly I could still choose my words for meaning, but select synonyms that hypothetically bend easier, making it easier on average to coil my sentence strip up on spools.

A recent study, Multiplexing Genetic and Nucleosome Positioning Codes: A Computational Approach, demonstrated that the local minima (the easy bending sequences) of a real gene could be repositioned (in computer simulation), that is, the nucleosome moved to another place in a real DNA gene sequence, by substituting equivalent base pairs (remember that a single amino acid can typically be defined by several different codons) to preserve the same protein encoding (a protein being a specific sequence of amino acids). This suggests that nucleosomes can at least theoretically be positioned anywhere on top of a gene, multiplexing the genetic information with the mechanical information as it were.

But do real genomes show any evidence of multiplexing genetic and mechanical information as described above? Experimentally mapped nucleosomes on top of genes do show strong signals in the probability for occurrence of mechanically favorable dinucleotides, but that does not prove the case (since shifting nucleosomes on random sequences may also demonstrate similar signals). The researchers in the study just cited then reasoned that if multiplexing is actually occuring, that is, if actual genomes have to balance the energy benefit of placing nucleosomes on DNA sequences where it is easier to bend the DNA into a coil around the histone, then it must be easier (and a more frequent occurrence therefore) to do that in non-coding sequences of DNA where only the "mechanical preference of the amino acids" need be considered (and not the necessity to maintain a specific sequence of amino acids to encode a particular protein).

Accordingly, they looked at the probability distribution for the amino acid threonine codons along nucleosomes on top of coding and non-coding regions of the S. cerevisiae and S. pombe genomes (high resolution nucleosome maps exist for both organisms). Both spectra showed a peak at the 10 bp (remember our earlier discussion of helical repeat), indicating that the codons for threonine display an overall rotational preference with respect to DNA bending within nucleosomes, but the non-coding peak was significantly higher for both organisms. When the 10 bp periodicity was plotted for all 20 amino acids for coding vs non-coding regions in both organisms the majority of points exhibited higher amplitude outside of genes, i.e., in non-coding regions where only the position preferences were a factor.

These findings (that the local minima, the easy bending sequences, of a real gene could be repositioned using degeneracy; that strong signals in the probability for occurrence of mechanically favorable dinucleotides are found in real genomes; and that the amplitude of 10 bp periodicity is higher in non-coding regions of actual genomes) at least suggest the possibility that nucleosome positions are the product of a mechanical evolution of DNA molecules, i.e., that there is a genomic code for nucleosome positioning in eukaryotes.

Answer 2

There are definitely genomic DNA sequences that nucleosomes preferentially package in vitro. They are A/T rich and have a periodic structure that facilitates bending. However such sequences would typically not be found on exons in vivo (because their protein-coding potential is reduced). Based on the accumulating literature, active promoters (and regions that will soon be activated) are usually Nucleosome-free, but I have yet to see evidence that this correlates with sequences that do not favour nucleosomes binding. These open areas are sensitive to DNase I, and the first nucleosomes downstream are phased. For Epigenetic codes, people focus on the many post-translational modifications on the Histone tails.