Certain enzymes that synthesize antibiotics play a game of pass the parcel, handing biosynthetic intermediates from one active site to another. A study reveals the dynamic nature of interactions between the enzyme domains.
Although crystal structures provide vivid insights into the architecture of enzymes, they reinforce a static picture of the molecules, providing only a snapshot of what a protein looks like in one stable conformation. This can be misleading, because enzymes in solution are certainly not static. Some, such as the megasynthase family of enzymes, are more like molecular versions of factory assembly lines, with moving parts shuttling intermediates from one place to another. Two studies1,2 in this issue demonstrate the importance of dynamics in the catalytic cycle of a typical megasynthase that makes an antibiotic. Such insight could be used to produce variants of the enzyme that would make analogues of the naturally occurring antibiotic.
The megasynthase family embraces several subclasses of enzyme, but of particular interest to this work are polyketide synthases (PKSs) and non-ribosomal peptide synthetases (NRPSs), both of which produce antibiotics — including penicillin and vancomycin, the powerful 'drug of last resort'. PKS and NRPS enzymes are large and multifunctional, consisting of clusters of active sites known as modules. Starting with simple organic molecules, each module catalyses a series of reactions in which a new molecular building block is added to the growing chain of the antibiotic, and is then modified to install the chemical groups required in the final product. During this process, biosynthetic intermediates are tethered to carrier proteins, which shuttle them in sequence to designated active sites (known as client enzymes) in the module.
X-ray crystallography has revealed the structures of several PKSs3 and NRPSs4, and those of related enzymes known as fatty-acid synthases5. These snapshot pictures provide essential information about their likely mechanisms of action. But there is mounting evidence that protein dynamics has a vital role in enzyme catalysis — for example, the activity of adenylate kinase, a small enzyme often used as a model for biochemical investigations, critically depends on conformational changes that occur on several different timescales6,7. It seems likely that dynamic effects are also important in PKS and NRPS enzymes, but this cannot be determined from their crystal structures.
Previous work8 has revealed the existence of conformational flexibility in the carrier-protein domains of NRPS enzymes when they interact selectively with different client enzymes. Expanding on these findings, Frueh et al.1 (page 903) used nuclear magnetic resonance to determine the structure of a carrier protein in complex with a type I thioesterase — an enzyme that catalyses the final release of antibiotic molecules from PKS and NRPS proteins. The authors observed that part of the thioesterase acts as a lid, flipping between open and closed conformations that expose or conceal the enzyme's carrier-protein binding site. They went on to show that this lid movement is necessary for the tether of the carrier protein to gain access to the active site, thereby catalysing the transfer of the antibiotic to the thioesterase.
Similarly, Koglin et al.2 (page 907) report conformational sub-states of a thioesterase II enzyme — a class of protein that repairs the tether of PKS or NRPS enzymes if it becomes 'damaged' through attaching acetyl or related groups. The sub-states all shift towards a single conformation on interacting with an acetylated carrier protein. Taken together with Frueh and colleagues' observations1, these results2 suggest that at least some types of protein–protein interaction in megasynthases are selected from an ensemble of conformations that exist in dynamic equilibrium with each other. Such a model has previously been proposed9,10 to explain certain protein–protein interactions.
Now that protein dynamics has been shown to play a role in antibiotic biosynthesis by NRPS (and thus probably also PKS) enzymes, the next step is to understand which, if any, of the many dynamic modes are involved in controlling the specificity of interactions between proteins. The importance of specific protein–protein interactions in such biosyntheses is well established11,12, but a better understanding of them is essential if the enzymes are to be rationally re-engineered to synthesize new antibiotics. Thus far, attention has focused mainly on identifying static features that allow protein domains to dock selectively to each other, such as regions that have complementary shapes or electric charges (Fig. 1a). The assumption is that such features will be similar for evolutionarily related client enzymes, but will be different for other enzyme families. If so, one should be able to replace any individual client enzyme with a related enzyme, without perturbing the rest of the megasynthase.
But if protein–protein specificity is the result of favourable docking between selected equilibrium conformations of the proteins' domains (Fig. 1b), then rational design of megasynthases requires knowledge of an entirely different set of issues. For example, how similar are the conformational distributions of analogous domains in related proteins? Are the energy barriers for conformational switching comparable for analogous domains across an entire family of enzymes? And does selective docking always require the same pair of compatible conformations?
The principles of physical organic chemistry were founded on the study of small molecules, and have provided a strong basis for elucidating the roles of static features in protein–protein recognition. But these ideas are of limited use in determining the rules that govern protein dynamics. Time will tell whether these dynamics are pivotal to the behaviour of megasynthases. If we are ever to be able to modify such enzymes to prepare new antibiotics, we can only hope that evolution has been parsimonious in writing the design rules.
Frueh, D. P. et al. Nature 454, 903–906 (2008).
Koglin, A. et al. Nature 454, 907–911 (2008).
Tang, Y., Kim, C.-Y., Mathews, I. I., Cane, D. E. & Khosla, C. Proc. Natl Acad. Sci. USA 103, 11124–11129 (2006).
Tanovic, A. et al. Science 321, 659–663 (2008).
Jenni, S. et al. Science 316, 254–261 (2007).
Wolf-Watz, M. et al. Nature Struct. Mol. Biol. 11, 945–949 (2004).
Henzler-Wildman, K. A. et al. Nature 450, 913–916 (2007).
Koglin, A. et al. Science 312, 273–276 (2006).
Kumar, S. et al. Protein Sci. 9, 10–19 (2000).
Grünberg, R., Leckner, J. & Nilges, M. Structure 12, 2125–2136 (2004).
Khosla, C. et al. Annu. Rev. Biochem. 76, 195–221 (2007).
Lai, J. R., Koglin, A. & Walsh, C. T. Biochemistry 45, 14869–14879 (2006).
About this article
Chemistry & Biology (2010)
Conformational selection and induced fit mechanism underlie specificity in noncovalent interactions with ubiquitin
Proceedings of the National Academy of Sciences (2009)
Natural Product Reports (2009)