Crystal Structure of aMammalian Voltage-DependentShaker Family KþChannelStephen B. Long, Ernest B. Campbell, Roderick MacKinnon*Voltage-dependent potassium ion (Kþ) channels (Kv channels) conduct Kþions across the cell membrane in response to changes in the membranevoltage, thereby regulating neuronal excitability by modulating the shape andfrequency of action potentials. Here we report the crystal structure, at aresolution of 2.9 angstroms, of a mammalian Kv channel, Kv1.2, which is amember of the Shaker Kþchannel family. This structure is in complex with anoxido-reductase b subunit of the kind that can regulate mammalian Kvchannels in their native cell environment. The activation gate of the pore isopen. Large side portals communicate between the pore and the cytoplasm.Electrostatic properties of the side portals and positions of the T1 domain andb subunit are consistent with electrophysiological studies of inactivationgating and with the possibility of Kþchannel regulation by the b subunit.Voltage-dependent Kþ(Kv) channels aremembers of the voltage-dependent cationchannel family, which includes the voltage-dependent Kþ,Naþ,andCa2þchannels (1).These channels are present in all the majorkingdoms of life. In eukaryotic cells, they workin concert with other ion channels to produceand modulate the electrical activity of the cell.This electrical activity is important for manyprocesses in electrically excitable cells such asneurons and muscle, as well as in nonexcitablecells (1). In the quintessential excitable cell, theneuron, Kv channels return the membrane volt-age to its negative resting value after an actionpotential, modulate the shape of action poten-tials, and set the action potential firing rate (1).Most of our knowledge of Kv channelfunction comes from studies of the Shaker Kþchannel from Drosophila melanogaster and itsfamily members from mammalian cells (2).Shaker family channels have been extensivelystudied with electrophysiology, because theycan easily be expressed in Xenopus laevisoocytes and in other cells. In contrast, nearlyall of our knowledge of Kþchannel structureis based on studies of prokaryotic Kþchannels,because they are more easily expressed at highlevels in Escherichia coli. Such studies havetaught us much about their pores, selectivityfilters, and gates (3).Eukaryotic Kv channels in many respectsare very similar to their prokaryotic counter-parts. The selectivity filter sequence is so con-served that we expect its structure to beessentially the same in all Kþchannels. Thepore_s Binverted teepee[ arrangement of innerhelices, which holds the selectivity filter in itswider half near the extracellular surface, is alsoexpected to be a conserved feature (4). How-ever, beyond their conserved pore and certaindomains that regulate the opening of the pore_sgate, eukaryotic Kv channels have certainunique features. For example, in the S6 innerhelix (on the intracellular side of the selectivityfilter), a highly conserved triplet sequence,Pro-X-Pro (where X is any amino acid), ispresent in Shaker family Kv channels but notin prokaryotic Kv channels. Mutations showthat this sequence is very important for gating,but the reason why has yet to be determined(5, 6). There has been speculation that thisregion of the pore (the inner pore), which islined by the S6 inner helices, is different inShaker Kv channels (5, 7) than in prokaryotes.Shaker family Kv channels have an adap-tation that, as far as we know, does not exist inprokaryotic Kv channels, and apparently allowsthem to carry out tasks that are unique toeukaryotic cells. Preceding the first membrane-spanning helix, S1, the N terminus forms a T1domain inside the cell (8–11). Four T1 do-mains, one from each of the four Shaker Kvchannel subunits, come together to form a tet-rameric assembly at the intracellular mem-brane surface. This domain is located directlyover the pore entryway to the cytoplasm,which means that the transmembrane poremust communicate with the cytoplasm throughside portals in order to allow Kþions to flowfreely between the cell and the transmembranepore (12–14). These portals not only permitKþions, but they also must be large enough toallow the entry of a polypeptide chain from thechannel_s N terminus, which functions as aninactivation gate in some Shaker family Kvchannels (15, 16).The T1 domain in eukaryotic Kv chan-nels is a docking platform for the b subunit(12, 17, 18). The b subunit forms a tetramer ofproteins related to aldo-keto reductase enzymes,which are oxido-reductases dependent onNADPH (the reduced form of nicotinamideadenine dinucleotide phosphate), with a-bbarrel structures (19–24). Crystal structures ofa b subunit tetramer and of a b-T1 domaincomplex showed that the enzyme_sactivesitecontains an NADPþcofactor and catalyticresidues for hydride transfer (12, 24). The bsubunit active site is mysterious because itsfunction is still unknown. Is it an enzyme thatcan be regulated by a Kv channel or does itserve as a sensor for the Kv channel, allowingthe redox state of a cell to influence elec-trical activity at the membrane? Nearly allShaker family Kv channels in the mamma-lian nervous system are associated with theseoxido-reductase b subunits (25).Experimental evidence suggests that voltage-dependent gating is fundamentally similar inprokaryotic and eukaryotic Kv channels andthat their voltage sensors are structurallysimilar (26–29). But structural studies withShaker family Kv channels have the potentialto be more informative. The structures of theprokaryotic Kv channel KvAP suggest thatan intact lipid membrane is required to keepthe voltage sensors correctly oriented withrespect to the pore (30) Esee also Protein DataBank (PDB) ID 2A0L^. Because the mem-brane is removed during isolation of the chan-nel, this requirement has so far obscured allattempts to deduce the structural basis forelectro-mechanical coupling between the volt-age sensor and the pore. The voltage sensor inShaker family Kv channels is constrained byits attachment to a cytoplasmic T1 domain, andtherefore may maintain its tenuous but impor-tant connections to the pore.Structure of the Kv1.2–b subunit com-plex. The Kv1.2 Kþchannel from rat brain(31) was coexpressed with the b2Kþchannelb subunit from rat brain (32) in the yeastPichia pastoris as described in (33) with mod-ifications (34). The protein was purified andcrystallized (34). Throughout purificationand crystallization, it was necessary to