Jun 292010
ResearchBlogging.orgWe all know that linear polymers of amino acids (proteins) adopt complex three-dimensional structures when they are dissolved in water. The process of forming these structures is called folding, and it is understood to occur because proteins are amphiphilic. Some parts of a protein chain like to interact with water (hydrophilic), while others are oily and want to get out of water (hydrophobic). Folding of the chain sticks all the oily parts together on the inside of the structure while the parts of the chain that have favorable interactions with water remain on the outside. An upcoming paper from the Journal of Physical Chemistry B suggests that sufficiently long alkanes might undergo a similar transition, even though they don’t have any chemical groups that like to interact with water.

Researchers from Purdue University performed a variety of simulations of linear alkanes, which are saturated hydrocarbon chains typically designated by the number of carbons they contain (C8 has 8 carbons). Because increasing the length of the chain just involves inserting an identical unit, one might expect that after a certain point the properties of these molecules would scale linearly with chain length. Previous experiments and simulations, however, indicated that the free energy of hydration did not match this predicted linear trend. Instead, the free energy of hydration (ΔG) remained flat or even decreased as chain length increased. This is because the linear extrapolation only holds for a chain that adopts a linear (all-trans) configuration. As chain length increases, however, the numerous additional degrees of freedom allow a chain to adopt a more compact conformation that decreases the penalty incurred by interacting with water.

Given this understanding, the authors asked whether a sufficiently long alkane might be hydrophilic. They established theoretical bounds for this question by examining two extreme possibilities. In the first, the alkane was assumed to be linear, and of course the ΔG never crossed zero. In the second, the alkane was assumed to collapse into a sphere; in this case the free energy becomes favorable after less than a dozen carbons. Presumably the reality lies somewhere in between. To get a more realistic view of the situation, they also simulated the behavior of alkane chains of various sizes, and found that the potential energy released by dissolving an alkane in water was correlated with the solvent-accessible surface area (SASA). From their findings they predicted that an alkane that was long enough would eventually cross over into hydrophilicity.

This is pretty interesting, but there are some significant weaknesses. The work here is purely theoretical and uses a molecular dynamics forcefield with an imperfect model of water behavior. The predictions of this simulation have been validated on real-world chemical samples, but only up to a comparatively modest chain length of C16. Because the surprising prediction lies very far from the zone of simulations that have been experimentally confirmed, one could argue that this is all just an extended discussion of a failure of the model. As they are aware of this shortcoming, the authors performed a number of simulations on collapsed (globular) C100 alkanes in order to determine the energy of the interaction between the chain and water, as well as the SASA. They found that, within error, the simulated values also indicated a negative ΔG of hydration.

As the authors note, this result doesn’t necessarily mean that you could dissolve a whole bunch of C100 in water. The ΔG calculated here is for transfer from the gas phase into water, and C100 is unlikely to be highly volatile, given that all the higher paraffins are solids. In addition, these are simulations of a single hydrocarbon chain in water, and so they don’t tell us about the energetics of lipid-lipid interactions. Oils segregate out of water because the oil-oil interaction is more favorable than the oil-water interaction. If this holds true for C100, even if dissolving in water is itself favorable, the alkane will still form oil droplets more readily than it will dissolve.

Using their models, the authors predict that the surface tension of an oil droplet will be negative at chain lengths greater than C50, thus tending to release oil into solution, but I’m a bit worried about this prediction. First, the molecular surface tensions are very far off from the macroscopic tensions, indicating that this calculation misses a great deal about the interaction. In addition, they perform a linear extrapolation from the approximately linear tail of what appears to be an exponential curve. It’s not clear why this extrapolation was used or what conclusions can be drawn from it; the molecular surface tension could just as easily be asymptotic with respect to the zero point.

Experimental verification of this prediction is unlikely to appear any time soon, if at all. Synthesizing very long alkanes is not trivial, especially in the quantities and purities required to put these simulations to the test. Ultimately the value of a paper like this is not in any practical application, but rather in the fact that it reminds us of the strength of the forces that guide the formation of higher-order chemical structures. Even in the absence of any group that has an intrinsically favorable interaction with water, the energy released by self-binding of hydrophobic groups may give rise to a “folded” structure for very long alkanes.

Underwood, R., Tomlinson-Phillips, J., & Ben-Amotz, D. (2010). Are Long-Chain Alkanes Hydrophilic? The Journal of Physical Chemistry B DOI: 10.1021/jp912089q

 Posted by at 12:26 AM
May 142008
ResearchBlogging.orgA recent episode of South Park featured a story in which two of the main characters got infected with HIV and discovered that the cure for AIDS is an injection of about $200,000. As any viewer of the series would expect, the episode is crude and vulgar, and it wobbles to and fro over the line between humor and offensiveness. Yet the episode might also turn out to be oddly prescient, if research described in an upcoming JACS article bears further fruit. As it turns out, researchers from UNC, the University of Colorado, and NC State have had some success in inhibiting HIV activity using drug-coated nanoparticles, made out of gold.

The approach Bowman et al. use is based on the idea of multivalence, which is the operating principle of Velcro. A single hook-loop interaction between two pieces of fabric usually isn’t enough to keep them fastened together. However, by having a large number of relatively weak interactions a strong connection can be made. Many biological systems make use of the same principle, using many weak interactions between repeating units to produce high overall affinity. The researchers set out to apply this idea to medicine, using many copies of a low-affinity drug attached to a nanoparticle.

The drug the authors used is based on a compound designated TAK-779, which is effective at preventing HIV virions from fusing with T cells, but also has some unpleasant properties for patients. The authors lopped off the part of the molecule that causes these problems, but doing so also removes most of its ability to fight HIV. So, they linked this new compound (called SDC-1721) to the gold particles at a ratio of about 12 molecules SDC-1721 per particle. In cultured cells, the nanoparticle-linked drug had an IC50 similar to TAK-779, even though SDC-1721 by itself was totally ineffective at preventing infection. Cutting the number of SDC-1721 molecules per particle to ~1 removed the inhibitory effect, proving that the multivalent approach was critical.

This is of course no demonstration of in vivo effectiveness, and there’s no telling whether the nanoparticle will have side-effects that are better or worse than TAK-779. However, if this initial success is borne out by further trials this may be a promising angle on treatments. One of the advantages to this approach is that it has some ability to counter resistance built into it because of the multivalent binding. Even if a virus evolves a lower affinity for the drug, the weak binding of many ligands, and the increased effective local concentration of those ligands, may be enough to rescue inhibitory activity. Injecting yourself with money is no way to cure anything, but it is possible that in the future we will attack viruses by injecting patients with (drug-laden) gold.

1. Bowman, M., Ballard, T.E., Ackerson, C.J., Feldheim, D.L., Margolis, D.M., Melander, C. (2008). Inhibition of HIV Fusion with Multivalent Gold Nanoparticles. Journal of the American Chemical Society DOI: 10.1021/ja710321g

 Posted by at 2:29 PM
Oct 252007
A clever little experiment appeared in this week’s JACS preprints, from an Italian group at the University of Bologna. Their efforts concern fluorescent nanoparticles, specifically ones in which the fluorophores have been encapsulated in a silica substance. The advantage of this sort of construct is twofold: it concentrates a large number of fluorescent molecules in the same spot, and also it sequesters them from ions in the solution that might quench their fluorescence. The latter effect could be maximized if you synthesized a nanoparticle that had a fluorophore-doped center and a fluorophore-free shell. Rampazzo et al. have accomplished just this, and demonstrated that the system can be tuned to produce some interesting effects.

Using a pyrene derivative that was only weakly fluorescent in oxygenated water, Rampazzo et al. constructed a reaction to create a doped nanoparticle. In an initial case, they added the dye to about 0.1%, and then grew nanoparticles up to a size of about 90 nanometers. The dye was almost completely incorporated into the particles, and the quantum yield correspondingly increased in an almost linear fashion. Similarly, the results of light-scattering experiments indicated that the particles were growing with a rate equal to that of the increase in quantum yield. The interesting feature here is that although these rates are equal, the fluorescence plateaus significantly before the apparent particle radius reaches its maximum. This result indicates that the fluorescent dye is incorporated relatively early but that the particle continues to grow after this supply is exhausted, thus creating precisely the kind of shell we wanted (A, B, C below):

This success seems to be based entirely on a fortuitous choice of dye concentration. What if that concentration is changed? Rampazzo et al. changed the concentration by an order of magnitude and produced another interesting effect. It turns out that when there are a number of these dye molecules close to one another, they form excimers with an emission maximum at a wavelength of light ~100 nm longer than the monomer. When the dye constitutes 1% of the reaction mixture, there is an initial decrease in emission from the monomer that occurs with roughly the same rate as an increase in excimer emission and in particle radius. Later the monomer emissions recover, and all three processes plateau at approximately the same time. This suggests the formation of a heterogeneous particle as shown in D, E, F above.

This time the dye is not sequestered entirely from the solution, but local concentration at the core of the particle is so high that an unusual fluorescent property is observed. It’s a conceptually simple little experiment, but it has an interesting result, and one that will have to be considered in future efforts to construct nanoparticles of this kind.

At the same time, if extreme local concentrations of some molecule have fortuitous or useful properties, this potential problem for nanoparticle construction might become an advantage. Consider if you have some molecule of interest that forms an excimer or exciplex with a known dye. Taking a sample containing traces of this molecule, one could use the encapsulation technique to construct a highly sensitive fluorescent detector.

 Posted by at 8:59 PM
Aug 292007
An interesting article appeared in JACS ASAP today from C.S. Hartley and J.S. Moore at UIUC on a clever way to direct the assembly of an asymmetric macrocycle. The approach relies on a simple idea to address a fundamental question about nanostructures, namely, how to make sure they assemble themselves as we would like. Given that protein design still isn’t quite far enough along to create any product we desire, we have to rely on more conventional chemistry. The basics of the approach should be obvious from the image on the right (shamelessly stolen from the paper). Note the reactive groups on the lower side of these compounds; even without knowing the relevant chemistry it should be clear what mechanism for directed assembly is implied. Given that the smallest geometrically stable assembly consists of three units, and that entropy favors the creation of small assemblies and enthalpy dislikes dangling functional groups, it should be obvious that dumping all these into a reactive pot should mostly produce units that have a composition of 1-1-1 or 1-2-3.

This is in fact what the authors observe when they perform the experiment, though they find (perhaps surprisingly) that in mixtures of all three components the 1-2-3 macrocycle predominates. Glancing at the product it’s clear that the macrocycles themselves will assemble into stacked arrays given the right conditions. If covalent chemistry is used to control this assembly, one has a rather simple method that could be used to construct fairly complex structures. Also, the yields from these reactions were encouragingly rich in the desired molecules, suggesting that only a little further optimization is necessary to produce effective scale-up.

Nothing in the paper is particularly earth-shattering, but then again, the cleverest answers often seem blazingly simple in hindsight. Hartley and Brown have come up with just such an answer here, in a commendably clear and readable paper.
 Posted by at 9:26 PM