Oct 232008
ResearchBlogging.orgThe practical aim of the investigation of allostery is the manipulation of this property as a means to aid human health and industry. We already have in hand the sequences of numerous enzymes that carry out unique and useful chemical reactions, and recent advances suggest that we will in the near future be able to design man-made enzymes that efficiently carry out completely novel reactions. Making the fullest use of these abilities demands that we be able to regulate the enzymatic activity of interest. Fortunately, just as nature possesses a rich array of enzymatic activities, it also holds a number of binding proteins, so we don’t have to start from scratch. The trick is that there must be some way to communicate the binding event from one domain into the active site of the other domain. In last week’s edition of Science researchers from the University of Pennsylvania and the University of Texas Southwestern Medical Center claim to have achieved just that.

The core idea of their approach is disarmingly simple. Identify a protein that has a long-range conformational response to a binding event, and locate the distal region on its surface where this response gets read out. Then, on an enzyme of interest, find a region of the surface that has an energetic connection to the active site. Join the proteins at these surfaces and voila! Now you have a regulatory switch for your enzyme!

The reality, of course, is likely to be trickier. In order for efficient communication between sites to occur via these pathways, the structural dynamics of allostery at the points of attachment must be compatible. For nearly all proteins, the precise nature of the conformational reactions that drive communication are essentially unknown. Thermodynamic mutant cycle analysis cannot provide detailed mechanistic information, and structural dynamics experiments from NMR and other techniques can provide only general information about what occurs on these pathways. Only molecular dynamics simulations are likely to give us the information we need to tune the allosteric control precisely. Absent that, all you can do is just stick things together and hope for the best, which is essentially what Lee et al. did.

Of course, they didn’t go in totally blind. Lee et al. used the statistical coupling analysis (SCA) technique pioneered by Dr. Ranganathan to identify distal surface sites linked to the light sensitivity of a PAS domain and the enzymatic activity of a bacterial dihydrofolate reductase (DHFR). I’ve mentioned this technique before in connection with Ranganathan’s research on the PDZ domain. The SCA results indicated that a surface loop of DHFR was energetically linked to its active site. The analysis also indicated that a region encompassing the N- and C- termini of the LOV2 PAS domain was likely to be a readout for its detection of light. These results accorded with existing knowledge about these proteins. The result with the PAS domain was particularly convenient. Because the N- and C- termini are adjacent, it meant that the PAS domain could simply be inserted at a loop site. Also conveniently, the surface identified for DHFR was a loop.

Thus, Lee et al. inserted the PAS domain at two sites in DHFR. One was the loop identified by SCA, and the other was a control site equally distant from the active site but not predicted to be linked. Figure 3 of the paper shows the key result: insertion of the PAS domain at the SCA-identified site (A site), but not the control location (B site), resulted in a modest light-dependence of the hydride transfer rate for DHFR. All of the A site chimeras had substantially reduced DHFR activity, similar to the effect of a G121V mutation. Interestingly, shifting the insertion site by even a single residue completely abolished the light-dependence of the activity.

Granted, the light dependence is less than twofold at room temperature; this approach did not generate a genuine light-dependent on/off switch for DHFR. However, for the reasons I mentioned before, a perfect switch is hardly something that could have been expected. What this experiment does do is prove that this approach is workable. Conceivably, with further tuning the hybrid PAS-DHFR can be made to carry out its catalytic function exclusively in the presence (or absence) of light. Since PAS domains bind a wide array of ligands, the approach can probably be adapted for various chemical triggers.

On a more fundamental level, the authors claim that this result supports the view that specific surface locations in many domains may be evolutionarily-conserved loci for allosteric control. This does not mean that every PAS domain (or PDZ domain, or DHFR) actually possess allosteric properties, but it does imply that all of them have the potential to exert or receive allosteric influences. If this is true, then it may be possible to adapt a wide array of binding modules as allosteric regulators for natural and designed enzymes. As our understanding of intradomain signaling improves, our ability to make use of these approaches will only increase.

J. Lee, M. Natarajan, V. C. Nashine, M. Socolich, T. Vo, W. P. Russ, S. J. Benkovic, R. Ranganathan (2008). Surface Sites for Engineering Allosteric Control in Proteins Science, 322 (5900), 438-442 DOI: 10.1126/science.1159052

  3 Responses to “Making a molecular switch”

  1. Site A doesn't look all that distant from the active site. Could the PAS insert be doing something simpler like blocking substrate binding or release?

  2. I think the solution might (or would in the future) be protein interface design. Creating a chimera of the two proteins is to constraining. Inserting mutations into the PAS domain that would make it bind upon light sensing might do the trick…

  3. I agree, nir. The future is definitely in designing interfaces that match up in such a way that information gets communicated efficiently, whether the different modules are part of a chimera or not. The challenge will be in figuring out how to make sure the communication works. Just designing a stable interface may not be enough; it's likely we'll need to design interfaces that transduce conformational fluctuations in a predictable way. That's a pretty tall order.

Sorry, the comment form is closed at this time.