Two-pore channels (TPCs) are an ancient family of ion channels that are unusual because they are found, not at the cell surface, but spanning the membranes of acidic organelles such as endosomes and lysosomes. These organelles mediate biomolecule transport and breakdown, and serve as stores of calcium ions1 (Ca2+). TPCs are key for several organellar functions — releasing Ca2+ into the cytoplasm to control trafficking of material such as receptor proteins and viruses, for instance, and stabilizing junctions with other organelles1,2. They are increasingly being associated with disorders such as Parkinson’s disease, and are therefore emerging as potential therapeutic targets1. Detailed structural information is scant, but advances in cryo-electron microscopy are revolutionizing our ability to study ion channels. In a paper in Nature, She et al.3 use this technique to provide the first detailed view of an animal TPC.
Previous work4,5 has reported the atomic structure of a plant TPC. This consists of two subunits, each containing two similar transmembrane domains (6-TMI and 6-TMII) connected by a large cytoplasmic linker. 6-TMI and 6-TMII are in turn each made up of six membrane-spanning regions, dubbed S1–S6. The pore through which ions flow is formed by S5 and S6 from each transmembrane domain in each subunit.
She et al. resolved the structure of mouse TPC1. Their results revealed that the overall folding of this channel is, as expected, like that of plant TPC. Nonetheless, there is a surprising degree of structural conservation between the linkers, given that animal and plant TPCs have very different amino-acid sequences in this region.
There are some structural differences, however. In plant TPC, the linker binds Ca2+ to help open the channel4,5. But Ca2+ binding by mouse TPC1 is unlikely, because amino acids essential for this interaction are missing. And the authors show that the carboxy-terminal domain of mouse TPC1, which is longer than the equivalent domain in plant TPC, forms a horseshoe-shaped arrangement of four helices that makes direct contact with the linker. This animal-specific feature probably serves to fine-tune channel activity.
Activation of animal TPCs is complex and multifaceted. These channels were originally identified6,7 as the targets for a messenger molecule called NAADP, which releases Ca2+ from acidic organelles8. Subsequent work revealed9 that TPCs are also activated by the lipid PI(3,5)P2. In addition, TPC1 is regulated by changes in voltage across the organelle membrane10,11. She et al. demonstrated that both PI(3,5)P2 and voltage changes are required to open TPC1; neither alone is sufficient (they did not examine NAADP). The authors then resolved structures of TPC1 in both the absence and presence of PI(3,5)P2, giving insight into the structural transitions that occur during channel opening. This analysis produced two key findings.
First, the group pinpointed the PI(3,5)P2 binding site, which lies in 6-TMI (Fig. 1). Mutation of any one of several amino-acid residues in the network that forms this binding site can prevent TPC1 activation by PI(3,5)P2. Interestingly, two of these residues — arginines in a short linker between S4 and S5 — are also required12 for channel activation by NAADP. This suggests that PI(3,5)P2 probably acts as a cofactor for NAADP action. Comparison of the free and PI(3,5)P2-bound forms of TPC1 revealed that a single lysine residue in S6 transmits conformational changes to the pore in response to PI(3,5)P2 binding, thus directing the first stage of channel opening.
Second, the authors found that changes in voltage are sensed by arginine residues in 6-TMII (Fig. 1). Both 6-TMI and 6-TMII contain sequences in S1–S4 that are reminiscent of voltage sensors in other channels, but only 6-TMII has a specific helix in S4 that is required for voltage gating. The 6-TMII voltage sensor is in an upward, ‘activated’ form in both structures obtained by the authors — in this form, it can probably transmit changes to the pore, to which it is adjacent, completing opening of the channel.
She and colleagues’ work on TPCs from animals, together with analyses4,5 of plant TPCs, indicate that both 6-TMI and 6-TMII cooperate to open the channel. 6-TMII is a target for voltage changes in both proteins. By contrast, 6-TMI is targeted directly by PI(3,5)P2 in animal TPC1, and indirectly by Ca2+ in plant TPC. This is a prime example of how evolutionarily distant proteins have adapted to conserve a core function.
Which ions pass through animal TPCs once they open? Much research suggests that these channels are non-selective, like plant TPC, but some work indicates that they are selective for sodium ions1,9 (Na+). She et al. found that TPC1 was about 70 times more permeable to Na+ than to potassium ions (K+). Their structures reveal that the narrowest part of the pore through which ions are filtered is shaped like an oblong ‘coin slot’, constricted by specific asparagine residues. The authors provide evidence that these residues allow the small Na+ ions through, but not the larger K+ ions. This sieve effect is unlikely to explain the authors’ data indicating that TPC1 apparently selects for Na+ over Ca2+, because these ions are about the same size. However, the electrophysiological experiments used by the researchers to determine ion selectivity were performed under very different conditions from those in live cells, where the permeability of TPCs to Ca2+ is readily demonstrable13.
In sum, She and colleagues’ structures provide major insight into how TPCs work. They join recently reported structures14–16 for a related family of ion channels, the TRP mucolipins (TRPMLs). Like TPCs, TRPMLs reside in acidic organelles, are activated by PI(3,5)P2 and release Ca2+ to control cellular functions such as gene transcription17. The PI(3,5)P2 binding site in TRPMLs is probably in the protein’s amino-terminal region16,17 and is thus very different from that in TPCs, although it has yet to be directly observed.
These rapid advances in the structural biology of organellar ion channels will aid future attempts to rationally design drugs that modulate ion flux through the channels. This is pertinent as the number of diseases found to be associated with channel abnormalities grows. Mutations in TRPML1 cause a lysosomal storage disorder affecting children, and TPCs have been implicated in fatty liver disease, Ebola infection and several neurodegenerative disorders17,18. In this context, a human TPC structure would be most welcome.
Another challenge is to resolve the structure of TPC2. This protein is regulated by NAADP and PI(3,5)P2, but not by changes in voltage — begging the question of how conformational changes in one TM domain are transmitted to the other to allow channel opening. No doubt, TPCs will reveal further secrets through forthcoming structures.
Nature 556, 38-40 (2018)