Blood clotting, olfaction and vision are just a few of the many physiological processes regulated by proteins called calcium-activated chloride channels and lipid scramblases. These proteins reside in the cell membrane and control the transport of chloride ions and lipids across this otherwise impermeable barrier. The machinery underlying these activities remained unknown for decades until research by several groups over the past 12 years showed that they are at least partially comprised of the TMEM16 and bestrophin families of proteins1,2,3,4,5,6,7. In two papers published in this issue, Brunner et al.8 (page 207) and Kane Dickson et al.9 (page 213) report the breakthrough determination of the three-dimensional structures of two of these proteins.
Calcium is a ubiquitous signalling ion. Certain events, such as the activation of cell-surface receptors by hormones or neurotransmitter molecules, can lead to a rapid increase in the intracellular calcium concentration. The extra calcium ions can then interact with and regulate different types of protein, including ion channels and transporters.
When calcium-activated chloride channels (CaCCs) bind calcium, they open to let chloride ions flow through the cell membrane. Moving ions across the membrane changes the electrical properties of a cell, and this can in turn affect other cellular functions. For example, CaCCs have been shown to affect the firing rate of neurons and to modulate the sensation of odorants in olfactory cells10.
Bestrophins and some of the TMEM16 proteins have been classified1,2,3,4as CaCCs, but other members of the TMEM16 family are not ion channels. Instead, they are lipid scramblases5,6 — proteins that let phospholipids flip from one side of the cell membrane to the other. This is crucial in many physiological processes, with one of the most established roles being that of exposing a lipid called phosphatidylserine to the surface of platelets to initiate blood clotting.
The structures reported by Brunner et al. and Kane Dickson et al. start to address the basic mechanisms of how bestrophins and TMEM16 proteins work. Namely, how does calcium binding lead to opening of a chloride-selective channel or activation of a lipid scramblase? And, in the case of the TMEM16 family, how can proteins with similar sequences (and thus structures) transport such different substrates?
Brunner et al. solved the structure of nhTMEM16, a calcium-activated lipid scramblase from the fungus Nectria haematococca. Around 40% of the amino-acid sequence of nhTMEM16 is identical to those of its mammalian counterparts, so it probably shares the same basic structure and mechanism of action. They find that the protein is organized as a dimer of two identical subunits (Fig. 1a). Each subunit has a region the authors call the subunit cavity, a narrow crevice that spans the membrane.
The cavity is lined by hydrophilic amino-acid residues, which is remarkable because it is exposed to the hydrophobic lipid environment of the cell membrane. The authors propose that this design facilitates lipid scrambling by providing a conduit for the hydrophilic lipid headgroups, while letting the hydrophobic lipid tails remain in the membrane. Several of the amino acids that line the subunit cavity have been implicated in ion conduction and selectivity in TMEM16 chloride channels5,11, and so the authors propose that this cavity is also the pathway for chloride ions in those proteins.
Within each subunit of nhTMEM16, at a point that corresponds to roughly one-third of the way into the cell membrane from the cytoplasm, just behind the subunit cavity, lies a binding site for two calcium ions. The location of the binding site within the membrane explains the voltage-dependence of TMEM16 proteins' calcium activation: more-positive voltages across the membrane make it easier for calcium ions to partially traverse the transmembrane electric field.
Many of the amino acids that bind the calcium ions are evolutionarily conserved throughout the TMEM16 family. When the researchers mutated these amino acids in mTMEM16A, a mouse chloride channel, they observed loss of activity. This suggests that the calcium-binding site probably regulates the activity of all TMEM16 family members.
Kane Dickson et al.9 solved the structure of chicken bestrophin 1 (BEST1), which is 74% identical to human BEST1. The structure reveals a completely different architecture from that of nhTMEM16. Instead of a dimer of subunits that seem to function independently, the BEST1 channel is a pentamer in which the assembled subunits create a pore for chloride ions to pass through the middle of the protein complex (Fig. 1b). This pentameric assembly seems to be a versatile platform for proteins, because a similar architecture was observed in the recently solved structure12 of KpBest — a bacterial (Klebsiella pneumoniae) protein that is distantly related to bestrophins in eukaryotes (organisms that include plants, animals and fungi), but which is a cation channel and is not activated by calcium ions.
The authors propose several ways in which structures along the pore allow BEST1 channels to conduct only ions that have a single negative charge, such as chloride ions. First, a region called the outer entryway on the extracellular side of the protein is overall negatively charged. This will repel most anions, especially doubly charged ones. However, ten weakly positively charged pockets within this region are sufficient to draw in singly charged anions.
Further along the pore is a narrow region called the neck, which is mostly lined by hydrophobic amino-acid residues. This would exclude both anions and cations, were it not for a ring of phenylalanine residues at the narrowest part. The phenylalanines are arranged such that the small positive charge localized on the edge of their benzene rings points to the middle of the pore. This facilitates the passage of small anions, but blocks cation movement. Finally, the inner cavity of the pore, which resides on the cytoplasmic side of the protein, is highly positively charged, to help to attract anions from inside the cell. A narrow aperture just below this cavity may prevent the entry of larger anions such as proteins or nucleic acids, which would block the pore.
Kane Dickson and colleagues report that each BEST1 subunit has a calcium-binding site, termed the Ca2+ clasp. The site is located within the intracellular part of the channel, close to the neck region. Because of this proximity, the authors suggest that the neck might be closed when calcium ions are not bound at the site, but that calcium binding induces conformational changes in the protein that leads to opening of the neck.
For both BEST1 and nhTMEM16, structures of the calcium-free states will be necessary to understand how calcium binding is transduced into mechanical work (to open the channel in BEST1 or activate the lipid transporter in nhTMEM16). In both proteins, the calcium-binding site is completely buried by protein, suggesting that the calcium-free state must have an appreciably different conformation to allow calcium ions to enter the site from inside the cell. It will also be important to obtain a structure of a TMEM16 chloride channel to understand the structural basis for the subunit cavity's dichotomous nature — its ability to serve as either a lipid- or a chloride-ion conduit.