Friday, March 13, 2009

There’s more to life than sequence

I have been meaning for some time to write about an interesting paper in JACS by Naoki Sugimoto’s group in Kobe. It found its way into an article that I wrote this week for Nature’s online news. So I’ve decided to simply post this article here – it’s not all strictly relevant to water in biology, but hopefully is interesting stuff anyway. This is the version before editing, which has more detail.

Shape might be one of the key factors in the function of mysterious ‘non-coding’ DNA.

Everyone knows what DNA looks like. Its double helix decorates countless articles on genetics, has been celebrated in sculpture, and was even engraved on the Golden Record, our message to the cosmos on board the Voyager spacecraft.

The entwined strands, whose form was deduced in 1953 by James Watson and Francis Crick, are admired as much for their beauty as for the light they shed on the mechanism of inheritance: the complementarity between juxtaposed chemical building blocks on the two strands, held together by weak ‘hydrogen’ bonds like a zipper, immediately suggested to Crick and Watson how information encoded in the sequence of blocks could be transmitted to a new strand assembled on the template of an existing one.

With the structure of DNA ‘solved’, genetics switched its focus to the sequence of the four constituent units (called nucleotide bases). By using biotechnological methods to deduce this sequence, they claimed to be ‘reading the book of life’, with the implication that all the information needed to build an organism was held within this abstract linear code.

But beauty has a tendency to inhibit critical thinking. There is now increasing evidence that the molecular structure of DNA is not a delightfully ordered epiphenomenon of its function as a digital data bank but a crucial – and mutable – aspect of the way genomes work. A new study in Science [1] underlines that notion by showing that the precise shape of some genomic DNA has been determined by evolution. In other words, genetics is not simply about sequence, but about structure too.

The standard view – indeed, part of biology’s ‘central dogma’ – is that in its sequence of the four fundamental building blocks (called nucleotide bases) DNA encodes corresponding sequences of amino-acid units that are strung together to make a protein enzyme, with the protein’s compact folded shape (and thus its function) being uniquely determined by that sequence.

This is basically true enough. Yet as the human genome was unpicked nucleotide base by base, it became clear that most of the DNA doesn’t ‘code for’ proteins at all. Fully 98 percent of the human genome is non-coding. So what does it do?

We don’t really know, except to say that it’s clearly not all ‘junk’, as was once suspected – the detritus of evolution, like obsolete files clogging up a computer. Much of the non-coding DNA evidently has a role in cell function, since mutations (changes in nucleotide sequence) in some of these regions have observable (phenotypic) consequences for the organism. We don’t know, however, how the former leads to the latter.

This is the question that Elliott Margulies of the National Institutes of Health in Bethesda, Maryland, Tom Tullius of Boston University, and their coworkers set out to investigate. According to the standard picture, the function of non-coding regions, whatever it is, should be determined by their sequence. Indeed, one way of identifying important non-coding regions is to look for ones that are sensitive to sequence, with the implication that the sequence has been finely tuned by evolution.

But Margulies and colleagues wondered if the shape of non-coding DNA might also be important. As they point out, DNA isn’t simply a uniform double helix: it can be bent or kinked, and may have a helical pitch of varying width, for example. These differences depend on the sequence, but not in any straightforward manner. Two near-identical sequences can adopt quite different shapes, or two very different sequences can have a similar shape.

The researchers used a chemical method to deduce the relationship between sequence and shape. They then searched for shape similarities between analogous non-coding regions in the genomes of 36 different species. Such similarity implies that the shapes have been selected and preserved by evolution – in other words, that shape, rather than sequence per se, is what is important. They found twice as many evolutionarily constrained (and thus functionally important) parts of the non-coding genome than were evident from trans-species correspondences using only sequence data.

So in these non-coding regions, at least, sequence appears to be important only insofar as it specifies a certain molecular shape and not because if its intrinsic information content – a different sequence with the same shape might do just as well.

That doesn’t answer why shape matters to DNA. But it suggests that we are wrong to imagine that the double helix is the beginning and end of the story.

There are plenty of other good reasons to suspect that is true. For example, DNA can adopt structures quite different from Watson and Crick’s helix, called the B-form. It can, under particular conditions of saltiness or temperature, switch to at least two other double-helical structures, called the A and Z forms. It may also from triple- and quadruple-stranded variants, linked by different types of hydrogen-bonding matches between nucleotides. One such is called Hoogsteen base-pairing.

Biochemist Naoki Sugimoto and colleagues at Konan University in Kobe, Japan, have recently shown that, when DNA in solution is surrounded by large polymer molecules, mimicking the crowded conditions of a real cell, Watson-Crick base pairing seems to be less stable than it is in pure, dilute solution, while Hoogsteen base-pairing, which favours the formation of triple and quadruple helices, becomes more stable [2-4].

The researchers think that this is linked to the way water molecules surround the DNA in a ‘hydration shell’. Hoogsteen pairing demands less water in this shell, and so is promoted when molecular crowding makes water scarce.

Changes to the hydration shell, for example induced by ions, may alter DNA shape in a sequence-dependent manner, perhaps being responsible for the sequence-structure relationships studied by Margulies and his colleagues. After all, says Tullius, the method they use to probe structure is a measure of “the local exposure of the surface of DNA to the solvent.”

The importance of DNA’s water sheath on its structure and function is also revealed in work that uses small synthetic molecules as drugs that bind to DNA and alter its behaviour, perhaps switching certain genes on or off. It is conventionally assumed that these molecules must fit snugly into the screw-like groove of the double helix. But some small molecules seem able to bind and show useful therapeutic activity even without such a fit, apparently because they can exploit water molecules in the hydration shell as ‘bridges’ to the DNA itself [5]. So here there is a subtle and irreducible interplay between sequence, shape and ‘environment’.

Then there are mechanical effects too. Some proteins bend and deform DNA significantly when they dock, making the molecule’s stiffness (and its dependence on sequence) a central factor in that process. And the shape and mechanics of DNA can influence gene function at larger scales. For example, the packaging of DNA and associated proteins into a compact form, called chromatin, in cells can affect whether particular genes are active or not. Special ‘chromatin-remodelling’ enzymes are needed to manipulate its structure and enable processes such as gene expression of DNA repair.

None of this is yet well understood. But it feels reminiscent of the way early work on protein structure in the 1930s and 40s grasped for dimly sensed principles before an understanding of the factors governing shape and function transformed our view of life’s molecular machinery. Are studies like these, then, a hint at some forthcoming insight that will reveal gene sequence to be just one element in the logic of life?


1. Parker, S. C. J. et al., Science Express doi:10.1126/science.1169050 (2009). Paper here.
2. Miyoshi, D., Karimata, H. & Sugimoto, N. J. Am. Chem. Soc. 128, 7957-7963 (2006). Paper here.
3. Nakano, S. et al., J. Am. Chem. Soc. 126, 14330-14331 (2004). Paper here.
4. Miyoshi, D. et al., J. Am. Chem. Soc. doi:10.1021/ja805972a (2009). Paper here.
5. Nguyen, B., Neidle, S. & Wilson, W. D. Acc. Chem. Res. 42, 11-21 (2009). Paper here.

Tuesday, March 3, 2009

Making sense of solvent slaving

In my previous post I mentioned work by Pablo Debenedetti on ‘toy models’ of water. The places to look are: Buldyrev et al., PNAS 104, 20177 (2007) (here) for the solvation thermodynamics of ‘spherical’ water; and Patel et al., Biophys. J. 93, 4116 (2007) (here) and J. Chem. Phys. 128, 175102 (2008) (here) for water-explicit lattice models of proteins.

And in discussing recent work on the mechanism of urea-induced protein denaturation, I neglected to mention Bruce Berne’s PNAS paper from late last year with Ruhong Zhou, Dave Thirumalai and Lan Hua (105, 16928; paper here). That paper on MD simulations for lysozyme anticipated the more recent work showing that denaturation seems to be caused by direct urea-protein interactions: the urea displaced water from the first hydration shell and penetrates into the hydrophobic core to give a ‘dry globule’.

The notion that protein dynamics are ‘slaved’ to those of the hydration shell has been floating around for some time now. Hans Frauenfelder and colleagues have now brought considerable focus to the idea (PNAS doi:10.1073/pnas.0900336106; paper here) with dynamical measurements using dielectric spectroscopy, Mossbauer and neutron scattering. They find that large-scale protein motions follow the fluctuations of the solvent and are dependent on solvent viscosity. There are two classes of fluctuation in the solvent, alpha and beta, with different timescales. It seems that the former are ‘structural’ in nature and control protein shape; the latter are those to which the protein’s internal motions are slaved.

Jianxing Song at the National University of Singapore, who I met in Hangzhou, has sent me three papers on the intriguing solubilisation of ‘water-insoluble’ proteins in pure water. He and his coworkers found this effect in 2006 for a range of diverse proteins (M. Li et al., Protein Science 15, 1835 (2006) – paper here; M. Li et al., Biophys. J. 91, 4201 (2006) – paper here). They attributed it to the tendency of the ‘insoluble’ proteins to form partially folded states with many exposed hydrophobic residues, such that only a very low ionic strength is sufficient to screen out repulsive interactions and cause aggregation. In pure water, however, those electrostatic interactions remain sufficiently strong to suppress aggregation and precipitation. Jianxing has now provided an overview of this work, expanding on the importance of pH for this effect, in FEBS Letters (doi:10.1016/j.febslet.2009.02.022; paper here).

Roberto Righini at the University of Florence and colleagues have used IR spectroscopy to identify and quantify the various aqueous species that solvate the polar heads of phospholipids in bilayers (V. V. Volkov et al., J. Phys. Chem. B doi:10.1021/jp806650c; paper here). And Davide Donadio and coworkers in California have shown how electronic charge fluctuations show up in the IR spectra of water close to nonpolar surfaces (here graphite) (D. Donadio et al., J. Phys. Chem. B doi:10.1021/jp807709z; paper here). Still in that neck of the woods, Chuan-Shan Tian and Ron Shen have used sum-frequency-IR spectroscopy to sort out the nature of hydrogen-bonding at the air-water interface (JACS 131, 2791 (2009) – paper here). And Heather Allen and colleagues at Ohio State University have used this and other spectroscopic techniques to study hydration structure of the air-water interface for various divalent nitrates (M. Xu et al., J. Phys. Chem. B doi:10.1021/jp806565a; paper here).

Haiping Fang in Shanghai continues his exploration of how water and solutes can be manipulated within the confinement of carbon nanotubes. He and colleagues now show, using MD simulations, how a single charge outside a nanotube can be used to move a hydrated peptide inside it, regardless of whether the peptide itself is charged (P. Xiu et al., JACS 131, 2840 (2009) – paper here). This is due to the dipole-orientational ordering of the water molecules caused by confinement and interaction with the external charge.

There’s more, but later.