Why should polymer physicists study biopolymers?


For many years now, the fields of synthetic and natural polymers have been clearly divided, in terms of conferences, journals, and even nomenclature used. Although the end uses of the two classes of polymers may be very different, the physics underlying their behavior and the techniques appropriate for their study are often broadly similar (this is of course in contrast to the differences in their chemistry). Thus, the somewhat artificial delineation between the two classes of polymers is perhaps unfortunate. Recently, there has been an upsurge in interest in natural polymers—call them biomacromolecules or biopolymers as you will—among polymer physicists for a wide range of reasons, but there is plenty of room for more researchers to join the fold bringing their particular skills to bear on the multiplicity of problems.

Two years ago, Jones in a Viewpoint article in this journal, highlighted taking the science in the opposite direction,1 learning from single molecule biophysics and the working of biological molecular machines to design synthetic molecules to perform analogous tasks. This of course in itself requires that the understanding of biomacromolecules is advanced, and provides one motivation for their study. One can identify a wealth of additional reasons and situations in which a study of these polymers might be of interest both for reasons of pure intellectual curiosity, and also for the consequences a better understanding of these polymers might lead to.

Biomacromolecules come in a variety of classes: proteins and polysaccharides are perhaps the most obvious and familiar, but in addition one can identify glycoproteins and proteoglycans (proteins with sugar sidechains, in which the carbohydrate content and length of the sidechains are much higher in the latter than the former) as well as the nucleic acids, DNA, and RNA. A great diversity of architecture and response is therefore available for study. Proteins are the major biomacromolecules in our bodies and which determine so much of how our bodies function (or don't). Yet, for over half a century their study has been seen as the preserve of biochemists and molecular biologists rather than polymer physicists, although one can argue that molecular biology largely grew out of the development of X-ray crystallography by physicists. Study has frequently been directed at understanding their crystal structure and sequence rather than at their generic behavior. With the advent of single molecule techniques (see refs.2 and3 for early examples of this for the extremely large protein titin) this distinction between biochemistry and physics has begun to blur. This is increasingly true as the opportunities for creating specific mutations enable designer molecules to be made for specific studies of the role of precise groups at precise points along the chain e.g., to explore the effect of a particular residue on folding/unfolding and the consequent mechanical properties, which can be tested by pulling with an AFM tip, for instance (see examples in refs.4–6, but there are many other papers in this area).

Other trends are also seeing an increased interest in protein conformation, structure, and aggregation by physical scientists. One clear example of this is the activity around folding, unfolding and misfolding, and the impact on aggregation. The problem of folding and unfolding is one of the questions to which AFM techniques, alluded to above, are contributing significantly. However, a further issue has been highlighted by Dobson,7 namely, that it appears that the aggregation of misfolded protein molecules into so-called amyloid fibrils is generic; in other words, if appropriate conditions can be found to promote the (partial) unfolding then it would appear that any protein is capable of forming amyloid fibrils. Appropriate conditions often involve heat or denaturing solvent for instance, in other words, for some proteins the conditions relate to states far from a physiological environment. Nevertheless, the immense interest in amyloid fibrils stems from their implication in many of the diseases of old age which beset the population, such as Alzheimer's disease and CJD (Creutzfeldt-Jakob disease), as well as some of the more recent entrants including v-CJD associated in the UK with “mad-cow disease” (more properly known as bovine spongiform encephalitis). Clearly some proteins—prion proteins in the case of CJD and v-CJD, and A-β in the case of Alzheimer's disease—are capable of misfolding under physiological conditions, with lethal consequences. The very fact that it appears the ability to form amyloid fibrils is not sequence-specific, since it is generic, (even if the ease with which they form clearly relates to the details of the protein) indicates that physical scientists, as well as biochemists, have a role to play in understanding their nature and formation.

More recently still, our own work has identified two other forms of protein aggregation which also appear to be generic. Amyloid fibrils form under conditions far away from the protein's isoelectric point (iep), when the molecules possess significant charge. But, it transpires they do not need to form as isolated fibrils, although that is how methods such as AFM and transmission electron microscopy (TEM) typically reveal them, because they need to image rather dilute dispersions. In practice, suprafibrillar aggregates can also form, which in the polarising light microscope have exactly the same appearance as the spherulites seen in synthetic polymers. The nature of these spherulites has been studied by a variety of approaches8, 9 and it appears that they form by radial growth of the fibrils templated onto some sort of (currently unidentified) nucleus. Thus they do not form from the aggregation of preformed fibrils, but the fibrillation occurs simultaneously with the growth of the fibrils.10 As a biologist once remarked to me, perhaps these spherulites are the “crud” that tends to sit at the bottom of the test-tube when experiments on fibril growth are carried out, but because TEM can only work with electron-thin samples which comprise diluted dispersions subsequently dried down, the spherulites never show up in such experiments. Nevertheless they have been observed in different proteins including insulin9 and β-lactoglobulin11 in the laboratory, and these same structures have also been observed in sections taken from diseased animals,12–14 confirming that the spherulites are not simply an in vitro curiosity but may well be a convenient way for the body to try to sequester the misfolded protein.

The food industry has long utilized the ability of proteins such as β-lactoglobulin (BLG), a major protein component of milk, to form fibrillar structures to texture foods such as yogurt by forming gels. It has only rather recently been recognized that these fibrillar structures are indeed identical to amyloid fibrils,15, 16 and therefore that the fibrils are themselves not necessarily toxic as implicated in the diseases mentioned earlier. Whereas insulin forms the suprafibrillar-spherulitic aggregates rather readily and fast, BLG is much more reluctant to form spherulites, and they only appear to form over a rather narrow range of pH.11 This shows that although the broad behavior is generic, the detail may be specific to each different protein. The perfection of packing, the actual detailed mechanism by which they form, and the range of conditions which are appropriate for their formation may all depend sensitively on the sequence, charge state, and overall size of the molecule. I will return to this point later.

BLG can also form gels of a totally different nature after heating to denature the protein around its isoelectric point (at a pH of ∼ 5.13), something else the food industry uses to its advantage. These gels, known as particulate gels, have been extensively studied and theories proposed for their formation relating to the specific details of the BLG molecule.17–19 But why should BLG be unique? The gels form by a rather monodisperse set of particles forming, which abut to give a connected structure in the gel. The particles are amorphous, and each consists of millions of protein molecules aggregated in a rather loose way. Clearly the lack of net charge on the protein facilitates this, although around the iep there are still charges distributed along the chain, just that there are approximately equal numbers of positive and negative charges. So, after heating to denature the protein, there are a large number of approximately neutral chains which are able to stick together and grow (again one can ask what the nucleation event is, but there is little evidence about this yet) to form spherical particles. This sketchy mechanism offers no suggestion that BLG should differ from any other protein, and recent work demonstrates that very different proteins exhibit broadly speaking similar behavior,20 with comparable particulate gels forming from five other proteins when heated around their isoelectric point. Of course, again, the details of the response may differ: the temperature at which the denaturation occurs, the size of the particles, and so on, but once more we see a broadly generic response.

Thus for proteins we can identify three generic forms of aggregation: particulate gels forming around the iep, isolated amyloid fibrils, and spherulites well away from the iep (it is worth distinguishing these two forms, since their roles in vivo may be substantially different). Once biological control is lost by heat—or other effects such as solvent—so that the native protein conformation is destroyed wholly or partially, polymer physics rather than biochemistry may be the most useful toolset to apply. Tackling proteins in this way is a rather young activity, but one which offers great promise. Having identified quasi-universal mechanisms and structures, now is the time to identify in much more detail what is specific to each protein which confers the precise form/size of the structures, the ease with which they form, the conditions which apply, and so on. In the case of the growth of insulin spherulites, an early model10 has explored the transition between growth which is “reaction” limited, i.e., limited by the time it takes for approaching molecules to adopt the right conformation to join on to a growing fibril, to “diffusion” limited, when supply is the rate-limiting state—analogous to what is known to occur in colloidal aggregation.21 However, early indications are that this model which works for insulin, may not work for BLG. One can speculate that this might be because BLG is a much larger molecule, with much higher charge state, but this example is simply used to illustrate the scope for polymer physicists to apply their particular brand of skills, modeling, and techniques to advance the field beyond these early stages of development.

I have written at length about the case of proteins because they offer a clear example utilizing approaches familiar to readers of this journal. Perhaps strangely, polysaccharides have actually always been more studied by the polymer community. In part, this is probably because polysaccharides have long had a major role to play in many commodity products such as cotton and paper, both of which are made predominantly of cellulose, and in adhesives and paper coatings, for which starch has been a major ingredient for many years. The biological roots of the polymers have not formed much of a focus of this research, and by the time the industrial processes have been implemented—for instance unpleasant solvents are typically required to solubilize cellulose—biology seems a long way away. Nevertheless biology—plus nutritional aspects—should not be forgotten. With the increasing demand for “greener” products, we can expect additional uses of these bulk polymers to be found in biodegradable packaging, for instance, in novel compatibilizers and in biofuels to give some specific examples, but in many of these new applications it will not simply be sufficient to identify “starch” as “starch” and equivalently for other polysaccharides. Each plant produces a subtly different variant of the molecules, and these variations may have profound implications for utilization.

Starch actually comprises two main polysaccharides—linear amylose and highly branched amylopectin—and basically contains chains of sugar rings. Both polysaccharides are extremely high molecular weight (running up to many millions), but the molecular weight is not well defined in the way it is for a protein. Rather, as with polymers produced in a conventional factory, there is a fairly broad molecular weight distribution. Likewise, the branching of the amylopectin molecule is not precisely defined and indeed is very hard to determine.22 It is, for instance, not clear whether the branching is random or bunched nonrandomly along the chain, because methods for characterizing the molecule usually require debranching and hence this information is lost. So, in many ways, starch and other polysaccharides look a lot more like synthetic polymers than proteins do, since they are so much less well-defined. But their biology matters. If one looks at the molecular weight distribution, different sources of starch have different distributions. The average side chain lengths of the branched amylopectin are likewise source-dependent.23 And these effects can have many consequences ranging from an impact on the viscosity of their solutions (important when trying to apply paper coatings at speed in a factory) to the ease with which recrystallization occurs in their solutions and dryer products, a factor which influences how fast bread stales and the freeze-thaw stability of starch containing products. All this is still rather far from biology.

However, with the development of genetically modified plants (or indeed, if this worries the reader, just understanding the plant biochemistry better so that modifications can be made to the plants using standard classical breeding but with insight rather than empiricism), a new challenge for the polymer physicist becomes apparent. Now changing something upstream in the plant “factory” can modify the molecules produced—so what changes might be desirable? Agronomy has been driven for years by the desire for high yields, but now the opportunity exists in principle to tailor the molecular products of the plant, if the complete pathway of biosynthesis to end use can be mapped out. Rather little work has so far been done on that, enlightened empiricism still being largely the route taken rather than anything more fundamental. Yet the opportunity exists for the polymer physicist to make significant contributions.

Again let me take a parochial example from my own work on starch to illustrate the point I want to make, that the polymer physicist has a role to play in demonstrating what modifications might be desirable when deciding which plants are “best” to develop and plant. The starch molecules are not produced and deposited randomly within a plant. They are produced within an entity known as a starch granule. Granules act as a store of energy and may be found within leaves (where they are turned over diurnally), within tubers (in the case of potatoes and tapioca/cassava for instance), and grain within the (wheat, maize, etc). The details of the granule are once again source-dependent. Leaf starch granules in all plants are almost invariably very small, 1–2 μm in size, whereas potato starch granules from a tuber are typically much larger, up to ∼ 80 μm in diameter. Some granules are approximately spherical but wheat-starch, for instance, has a lenticular shape with a well-defined central groove, and ginger starch is flat and almost a disc. The granules are very organized structures with a hierarchy of structural elements present. Surprisingly, to a polymer physicist's mind, it is not the linear amylose molecule which is involved in the crystals present in the granule, but the side-chain branches of the amylopectin which form double helices which align to give small crystals a few nanometer in thickness.24 For systems in which the amylopectin side chain branch distribution contains a significant number of very short branches, crystallinity is significantly impeded, and here we see an example of where the biological route of production can affect the end products. The granules of many genetically modified starches are found to be misshapen and cracked (compared with the equivalent native species), and changes in the molecular composition are clearly a key ingredient in these changes.25

At the very short lengthscale within the granule, we therefore have a situation in which the amylopectin molecules are involved in crystallization. These crystals are gathered together into a semicrystalline growth ring, each of these rings being separated by an amorphous growth ring where (presumably) the amylose sits. The growth rings can be revealed by etching with α-amylase, an enzyme which preferentially attacks the amorphous regions, the packing of the crystals within the semicrystalline growth ring can be explored using small angle X-ray scattering (SAXS). This shows that, despite all the variations in chain architecture and molecular weight distribution between starches, there is a fairly consistent and universal 9-nm repeat for them all,26 and this cannot be attributed to something specific about the enzymes involved in producing the starches since these vary so much between different species. So here we see how physics can come into play: not only does it provide the tool to study the packing (SAXS is not a very familiar technique to biologists) but we must also use physical insight to attempt to understand this universal 9-nm repeat. Our current best suggestion for this universality comes from considering the details of the molecular packing. The sidechains must be attached to the backbone of the amylopectin, and there is thus a compromise between the way in which the sidechains can accommodate themselves in the double helices within the crystals, and the entropy of the backbone. Using a side chain liquid crystalline polymer analogue,27 one can relate this to the need for a sufficiently flexible “flexible spacer” by which the part of the sidechain involved in the double helix formation is decoupled from the backbone. Short segments in a double helix will lead to rather unstable crystals, which will need a long flexible spacer to provide sufficient decoupling and conversely, long double helices will lead to thicker crystals needing a shorter flexible spacer to enable stabilization; these opposing effects together lead to an “optimum” long period repeat of 9 nm. This remains only a hypothesis, but one which is so far consistent with the evidence.

So here is one example of how the physicist can provide insight into the biological situation for starch. The second example I will quote briefly, is what happens when the granule is actually used in practice: in general it is broken down for industrial use (or in cooking) by heat, a process known as gelatinization. How the granule breaks down will depend on many factors including concentration and temperature, as well as of course the detailed granule structure, and its breakdown can be followed by in situ experiments in a synchrotron using combined SAXS/WAXS/DSC28, 29 during the gelatinization process, and also—to follow the details of the water ingress—via SANS.30, 31 Thus if one understands this structure, understands how it may be affected by the starch biosynthesis route, then in principle one might be able to “design” starches which break down in a particular desired way (e.g., over a specific temperature range). This goal has not yet to my knowledge been realized, but it should serve to illustrate how bringing polymer physics approaches to bear on this complex biological entity can reveal new insight.

One could give many further examples of where polymer physics has much to offer the field of biopolymer research; to conclude this article I will just give a couple more for brief “tasters”. Recently, an increasing interest in cellular mechanics has been evinced by the community. Microrheology, developed for the study of complex fluids, has been extensively applied to solutions of the cytoskeletal polymer actin32 and more recently still to cell contents.33–35 Theoretical ideas on entangled semiflexible polymers can be applied to analyze the anticipated behavior of the cytoskeleton under different conditions,36 and coupled with microrheological measurements to compare with real cells. When cells adhere to surfaces they produce a complex glue of polymers known as the extracellular matrix. Some of the response will undoubtedly arise from specific interactions between ligands and receptors, but some of it can probably be treated using ideas familiar from synthetic aqueous gels.

Many naturally occurring polymers are long and stiff, exactly the properties which confer liquid crystallinity on main chain synthetic liquid crystalline polymers. DNA, for instance, is well known to form liquid crystalline phases in the test-tube.37 This may or may not have biological significance, since the concentrations of DNA within a cell are rarely so high as that required to create the liquid crystalline phase in vitro, but for other biopolymers the situation is less clear. Actin is another example of a rather stiff polymer which is known to form liquid crystalline phases.38 Does this matter as cytoskeletal rearrangements occur within the cell, and the actin filaments successively polymerize and depolymerise as the cell advances?

The time is ripe for more polymer physicists to bring their expertise garnered on synthetic polymers to bear on the natural world; to ask questions from a very different perspective than that of the biochemist who customarily works with these natural polymers; to look for similarities in response rather than the unique difference each particular sequence along a chain might bring; and to take theories developed on simple chains containing only one or two monomers and develop them to the point that they can enlighten us about the way our bodies function. As a community we have the tools and theories. We now need to apply them to situations where the biopolymers remain in their native, collective state to elucidate biological function in all its richness and complexity.