Blockade in the Cell's Waterway

The ubiquitous class of channel proteins known as aquaporins provides a path for water to cross cell membranes while blocking the passage of protons. When the first structural pictures of aquaporins were obtained a few years ago, an elegant mechanism emerged for how these water-filled channels keep protons out. Recent theoretical studies, however, have called this mechanism into question.

Aquaporins play a key role in cellular water homeostasis in humans, animals, and plants (C&EN, Nov. 3, 2003, page 35). These water channel proteins associate as tetramers. Each aquaporin contains a single 28-Å-long cylindrical pore that supports a string of nine hydrogen-bonded water molecules in single file.

"Normally, continuous, single-file columns of water like these conduct protons," notes Robert M. Stroud, a professor of biochemistry and biophysics at the University of California School of Medicine, San Francisco. He points out that protons can easily hop along a similar single-file water column in the water pore of gramicidin A. But unlike gramicidin A, an antibiotic peptide that kills bacteria by forming pores in their cell membranes, aquaporins must not leak protons. Doing so would disturb the delicate balance of charge across cell membranes that underlies cellular function.

So how do aquaporins manage to do what gramicidin A can't? The water column in aquaporins does have some distinguishing features. Near the center of the aquaporin pore, hydrogen bonds from a pair of asparagine residues (as well as the pull of nearby -helix dipoles) reorient the central water molecule, preventing it from accepting a proton from nearby water molecules.

From these structural observations, it was suggested that the forced reorientation of this central water molecule might break the continuous chain of hydrogen-bonded waters and possibly prevent protons from "hopping" along the hydrogen-bonded water "wire." This defect in the water wire was later confirmed by molecular dynamics simulations of water moving through the pore performed by a team led by biophysicists Klaus Schulten and Emad Tajkhorshid of the University of Illinois, Urbana-Champaign [Science, 296, 525 (2002)].

 EARLIER, biophysicists Bert L. de Groot and Helmut Grubmüller of the Max Planck Institute for Biophysical Chemistry in Göttingen, Germany, had independently performed similar water simulations, also concluding that the main barrier to protons hopping along the water chain is an interruption of the hydrogen-bonded water chain. But they observed the water defect at a different spot, where the channel becomes most narrow [Science, 294, 2353 (2001)].

"Until recently, all we knew about the proton blockage mechanism in aquaporins came from indirect evidence," notes Grubmüller. Both Schulten's and his own conventional force-field-based molecular dynamics simulations followed only water movement through the channel and did not directly study proton transfer, he points out.

A flurry of recently published studies attempts to probe more directly how aquaporins exclude protons.

The first, by Grubmüller, de Groot, and their colleagues, used a molecular dynamics method to simulate proton transfer within aquaporins [J. Mol. Biol., 333, 279 (2003)]. Their computer simulations show that the barrier to proton transfer is highest in the part of the aquaporin pore that boasts the highest electrostatic potential: the region centered around the two central asparagines and the partially positively charged -helix dipoles, known as the NPA region.

The results indicate that the main barrier to proton transfer is not caused by interruption of the hydrogen-bonded water chain, as had previously been speculated, but rather by an electrostatic field created by the -helix dipoles in the NPA region, Grubmüller and de Groot conclude. They estimate that the barrier to proton transfer in aquaporins is 6–7 kcal per mole--about the same as estimates of the barrier to proton translocation across lipid membranes, Grubmüller says.

In the same vein, chemistry professor Arieh Warshel of the University of Southern California tells C&EN that his many years of simulating proton transport in other proteins have shown that such processes are always controlled by electrostatic effects, not water orientation. He claims that a recent study of his suggests that the same is true for aquaporins.

Warshel and graduate student Anton Burykin used a computer simulation to estimate the free energies of different steps in proton transport through an aquaporin pore [Biophys. J., 85, 3696 (2003)]. They found that the barrier to proton transport is a whopping 15 kcal per mole--much larger than the measly 2 kcal per mole it costs to send a water molecule through the channel, Warshel points out.

THIS LARGE BARRIER kicks in around the same part of the pore as that observed by Grubmüller and de Groot in their most recent paper: the NPA region, where the disruption in the water wire was observed in structural studies. But Warshel is adamant that attempts to attribute the barrier to proton transport to a structural motif in the protein are misleading.

"The main source of the barrier comes from the electrostatic penalty of moving a charged proton from bulk solution onto a water molecule inside the protein channel," Warshel maintains. He contends that the energetic cost of rotating water molecules to mend a break in the water wire is trivial compared with the energetic cost of desolvating a proton and sticking it in an oily protein pore.

And the barrier to proton transport has nothing to do with the pair of asparagines or the -helix dipoles, he stresses. "We have repeated our calculations on an aquaporin whose asparagines can't donate hydrogen bonds to the central water, on an aquaporin lacking the asparagines, and on an aquaporin lacking the helix dipoles," he tells C&EN. "Each of them displays a very high barrier to proton transport."

The NPA region, Warshel says, simply happens to be located near the center of the pore, about halfway through the lipid bilayer, where the desolvation penalty for passing a positively charged proton through a protein pore is expected to be largest. As further proof, he notes that he and Burykin observe the same large barrier to proton translocation in a model protein pore lined entirely with nonpolar amino acids.

But Schulten and Tajkhorshid argue that their detailed analysis of the electrostatic field inside the channel [Biophys. J., 85, 2884 (2003)] shows that protons actually interact with the NPA region quite strongly. And biochemists Benoit Roux of Cornell University's Weill Medical College and Régis Pomès of the University of Toronto point out that Warshel's purely electrostatic argument doesn't jibe with their own experimental and theoretical studies with the gramicidin A channel, a narrow membrane pore through which protons rapidly travel along a water wire [Biophys. J., 71, 19 (1996)].

ROUX AND POMÈS NOTE that their own recently published study of proton blockage in aquaporins confirms that the protein stringently controls the orientation of water in the NPA region, breaking the hydrogen-bonded water chain [Structure, 12, 1 (2004)]. But they conclude that the approximately 10–12-kcal-per-mole barrier to proton translocation that they find peaking at the NPA region results from a combination of factors. These include not only the orientational control of water molecules but also desolvation penalties and electrostatic effects caused by the charge distribution in the channel.

University of Utah chemistry professor Gregory A. Voth--whose own computer simulations of proton translocation in aquaporins, in collaboration with Illinois' Schulten, will soon appear in Proteins: Structure, Function & Bioinformatics--has calculated a barrier to proton transport that's similar in magnitude to that found by Warshel and Burykin. Voth and postdoctoral associate Boaz Ilan have calculated that the barrier to proton transport is about 18 kcal per mole, and it's centered at the same place: the NPA region. Voth goes on to say that the 6–7-kcal-per-mole barrier estimated by Grubmüller's team is likely too small to prevent proton translocation in vivo. Instead, Voth argues that a barrier height on the order of what his team and Warshel's team both calculate is more physiologically relevant.

But unlike Warshel, Voth says it's too early to conclude that this structural motif isn't playing a role in blocking proton transport in aquaporins. He tells C&EN that his team is now using its method--which Voth asserts contains the essential physics of proton transport through water chains--to do some of the same computer experiments that Warshel and Burykin did, such as repeating the computer experiment without the NPA motif.

"I have no doubt that the electrostatic environment in the channel plays a critical role in defining the behavior of proton transport," Voth says. "But if the hydrogen-bonding defect in the water wire doesn't participate in blocking protons, then it's hard for me to imagine why nature has put it there at all." The NPA region that gives rise to this hydrogen-bonding defect is in fact present in all aquaporins.

Each of these teams uses different methods to come to its respective conclusions about the magnitude of the barrier to proton translocation and the origin of that barrier. And each has its own opinion about whose method is more accurate. "But calculations, from which these controversies spring, are all based on different assumptions--so it's not surprising that they reach different conclusions," UC San Francisco's Stroud notes. "But they provide great insight into the dynamic properties of the channel. Ultimately, the marriage of experiment and theory is an important and increasingly reachable goal," he adds.

Stroud also views these calculations as valuable signposts for where he and his colleagues should be addressing experimental tests. He and graduate student David F. Savage and their coworkers recently unveiled a 2.5-Å structure of a recombinantly expressed water-selective aquaporin from Escherichia coli [PLoS Biol., 1, 334 (2003)]. They are now working to make versions of the channel with mutations in the NPA region and elsewhere. Such mutations, Stroud thinks, are key to answering experimentally the question of what controls the insulation of the channel toward protons.

PROTON TRANSFER

Model System Allows Experimental Study Of Proton Conduction Along Hydrogen-Bonded Wires

Direct, molecular-level experimental observation of proton transfer along a hydrogen-bonded wire is hard to come by. That's because protons shuttle rapidly across microscopic distances through a constantly fluctuating hydrogen-bonded network of solvent molecules.

"Most of the literature on single-file proton conduction has been based on computer simulations," notes Samuel Leutwyler, a professor of physical chemistry at the University of Bern, in Switzerland. "Little has been done in the way of real experiments."

Leutwyler is trying to change that. With graduate student Christian Tanner and postdoc Carine Manca, he recently unveiled an elegant small-molecule model system for studying proton hopping along hydrogen-bonded wires [Science, 302, 1736 (2003)].

Their simple model is constructed around an aromatic scaffold molecule, 7-hydroxyquinoline. A hydrogen-bonded wire of three ammonia molecules is stretched between the scaffold's hydroxyl group and its quinolinic nitrogen atom.

No protons are transferred along the ammonia wire when 7-hydroxyquinoline is in the ground state. But when electronically excited with a laser, its hydroxyl group becomes much more acidic and injects a proton into the hydrogen-bonded ammonia wire. Protons shuttle along the length of the wire, eventually protonating the quinolinic nitrogen and yielding 7-ketoquinoline. Leutwyler's team tracks the proton transfer by following the distinct UV spectra of 7-hydroxyquinoline and 7-ketoquinoline.

In parallel, Leutwyler's team managed to calculate a potential energy surface for proton transfer. Or, more accurately, for proton-electron-coupled transfer: Ab initio calculations show that an electron hops along the wire with the proton in this system, Leutwyler notes.

The model cluster also allows spectroscopic probing of the effect of the solvent molecules' vibrational motions on the rate of proton transfer. More recently, the team has tried substituting various triamines for the ammonia molecules in the current cluster to study the effect of solvent preorganization on the rate of proton transfer.

Leutwyler hopes that lessons learned with their proton-conducting ammonia-wire model system will help them design proton-conducting water-wire versions. He tells C&EN that when the ammonia wire is replaced with a water wire in this particular aromatic scaffold, no proton conduction takes place. His team is now working to design larger, more acidic scaffolds that might support a proton-conducting water wire.