Cryo‐EM structure of the Slo1 potassium channel with the auxiliary γ1 subunit suggests a mechanism for depolarization‐independent activation

Mammalian Ca2+‐dependent Slo K+ channels can stably associate with auxiliary γ subunits which fundamentally alter their behavior. By a so far unknown mechanism, the four γ subunits reduce the need for voltage‐dependent activation and, thereby, allow Slo to open independently of an action potential. Here, using cryo‐EM, we reveal how the transmembrane helix of γ1/LRRC26 binds and presumably stabilizes the activated voltage‐sensor domain of Slo1. The activation is further enhanced by an intracellular polybasic stretch which locally changes the charge gradient across the membrane. Our data provide a possible explanation for Slo1 regulation by the four γ subunits and also their different activation efficiencies. This suggests a novel activation mechanism of voltage‐gated ion channels by auxiliary subunits.

Slo channels are Ca 2+ -gated K + channels that are important for neuronal function, skeletal muscle contraction and maintaining smooth muscle tension in metazoa [1].Their characteristic feature is an extraordinarily high single-channel conductance that is, in excitable cells, initiated through synergistic activation by transmembrane voltage potential depolarization and increasing intracellular Ca 2+ levels [2].Slo channels, also called big potassium (BK) or maxiK channels, are ubiquitously expressed in many excitable cells, such as neurons and muscle cells, as well as non-excitable cells, such as epithelial cells [3][4][5][6].They are required for re-establishing the resting potential after an action potential, the release of neurotransmitters at synapses as well as preventing constant membrane depolarization on the postsynaptic side [4].Due to these important neuronal functions, Slo channels are the targets of many natural neurotoxins and candidate targets for novel insecticides [7].
Slo channels belong to the tetrameric voltage-gated ion channel superfamily in which each monomer provides minimally six transmembrane helices to the transmembrane domain: four helices S1-S4 comprise the voltage-sensor domain (VSD) and the two helices S5 and S6 stabilize the central selectivity filter hairpin.In vertebrate Slo1 and Slo3 as well as invertebrate Slo, S1-S6 are preceded by an additional N-terminal S0 helix running in a roughly antiparallel way along S3 that is absent in Slo2 and other voltage-gated ion channels [8][9][10][11][12][13].Helix S4 contains two arginine residues (R210 and R213 in rabbit Slo1) that act as gating charges.Upon depolarization, their positively charged side chains move toward the (negatively charged) outer leaflet of the membrane, inducing a rotational movement of the C-terminal half of S4 toward S1.This rearranges the position of the S4-S5 linker, making space for S6 to move toward its activated conformation which is equivalent with an opening of the intracellular gate, allowing potassium ions to access the selectivity filter [14,15].This rearrangement acts in concert with Ca 2+ -induced activation of the large, domain-swapped cytosolic gating ring containing a total of eight Ca 2+ -binding sites.Ca 2+ -binding induces a large outwards movement of the most N-terminal domain of the gating ring, straightening the preceding linker.This movement facilitates the displacement of the S4-S5 linker and pulls the S6 helix into the open conformation [8][9][10][11].Thus, the Ca 2+induced motion of the gating ring primes the VSDs by stabilizing the active conformation, and vice versa, the active conformation of the VSD assists Ca 2+ -gating through contacts between the S0-S1 and S4-S5 linkers of the VSD and the gating ring in the Ca 2+ -bound conformation [9].
Many native ion channels, such as voltage-gated sodium and potassium channels, include auxiliary subunits which bind to the transmembrane domain and modulate its conformational space and energy landscape [16][17][18][19].As such auxiliary subunits often have very specific tissue expression patterns, they allow the differential tuning of the activity of a ubiquitously expressed channel [20].Two families of auxiliary subunits of Slo termed b and c, each comprising four members, have been identified to date in vertebrates, but are notably absent in invertebrates [1,3,20,21].They are especially important for the action of Slo1 in non-excitable cells since they change the Ca 2+ -and voltage-gating behavior of Slo1 and allow the activation of the channel under conditions where the Slo1 core tetramer alone is inactive [20].
In smooth muscle, Slo1 exists exclusively in a complex with b1 which is crucial to control muscle tone by an increased Ca 2+ sensitivity [22,23].The other three b subunits are expressed in other tissues including chromaffin cells (b2) and in the brain (b4).b subunits increase the Ca 2+ sensitivity of Slo1, decrease the activation and deactivation kinetics and also alter the sensitivity toward pore-blocking spider toxins [3,24].They bind into the wide groove between two neighboring voltage-sensor domains of Slo1 in a 1 : 1 stoichiometry via their two transmembrane helices [10].The small interspersed extracellular domains of all four b4 molecules associate with each other to form a basket.The function of these extracellular domains is still largely unknown.
The first evidence of a second type of auxiliary subunit was the observation that the leucine-rich repeatcontaining protein 26 (LRRC26) induces a large À140 mV negative shift of voltage dependance of Slo1 in non-excitable prostate cancer cells [25].Three more members of this family with a similar architecture of one transmembrane helix and an extracellular leucine-rich repeat domain were identified later [3,20,21,26].c1-4 are expressed in a wide range of tissues including smooth muscle (LRRC26/c1), testis (LRRC52/c2), brain (LRRC55/c3), the adrenal gland (LRRC38/c4), and skeletal muscle (c2 and c4) [26][27][28].They have been termed c subunits as they differ not only in their molecular architecture from b subunits, but also in the effect they have on Slo1 (and Slo3 in the case of c2).They all induce negative shifts to different extents in voltage dependence and it was shown for c1 that it strongly enhances allosteric coupling of voltage sensing and pore opening, thus explaining how Slo1 can be activated in non-excitable cells in the absence of an action potential or strongly changing intracellular Ca 2+ levels [21,25,27,29,30].On the other hand, c subunits do not affect the Ca 2+ sensitivity.c and b subunits can even be present in the same Slo1 channel, increasing the possibilities for very fine tuning of K + translocation even further [31].
While the LRR domains do not contribute to the shift in voltage dependence induced by c subunits, it was shown that both the transmembrane helix and a directly adjacent short polybasic stretch on the C-terminal tails are the critical determinants for this shift, indicating a direct structural effect on the transmembrane domain of Slo.Furthermore, experiments with chimeric constructs have demonstrated that the transmembrane helices of c1 and c2 are very effective and contribute to voltage shifting much more than the transmembrane helices of c3 and c4.Conversely, the polybasic stretches of c1 and c3 are more potent in inducing voltage dependency shifts compared to c2 and c4 [32].
The mechanistic and structural understanding of Slo1 has deepened significantly in recent years.While the activation of the gating ring has been dissected already by x-ray crystallography [33][34][35], an increasing number of cryo-EM structures allow better understanding of the complexity of Slo regulation: Structures of Slo1 from Aplysia californica, human and fruitfly in the Ca 2+bound and -free conformations unveiled the coupling between gating ring and intracellular gate [8][9][10][11]36].Very recently, insights into gating charge movement and voltage activation were obtained by structure determination utilizing constitutively active and inactive mutations [14,15].Furthermore, our previous work uncovered how certain small molecule toxins can modulate the activity of Drosophila Slo, helping to identify some critical structural features in the pore domain [8].Lastly, structures of human Slo1 in complex with b4 in the Ca 2+ -bound and -free conformations helped to understand the mechanism by which b subunits regulate the channel [10].However, it remained unclear how c subunits bind to the transmembrane domain of Slo1 and what the molecular basis for the observed negative shifts in voltage dependence and the differences between the four c subunits is.
Here, we present a cryo-EM structure of c1 bound to a Slo1 tetramer.We show that the transmembrane helix of c1 binds to the distal face of the Slo1 VSD by perfect shape complementarity and that this arrangement is stabilized by a kink in the c1 transmembrane helix and an extracellular hook.Furthermore, in our structure, a polybasic stretch extends the c1 transmembrane helix into the cytosol and we propose that this polybasic stretch, by introducing immobile positive charges, helps the VSD to adopt its activated conformation even at resting potential.Sequence comparison of these structural elements between the c subunits helps to understand their different effectiveness in inducing voltage dependence shifts in Slo1.Finally, we show that simultaneous binding of b and c subunits is structurally possible under the condition that the extracellular LRR domain of c subunits, which form a flexible tetramer in the case of c1 in our structure, would open up and rearrange.
cDNAs encoding the four rabbit c subunits were ordered as synthetic, codon-optimized DNA fragments (Thermo Fisher Scientific, Regensburg, Germany) and also cloned into pEG BacMam vectors.The resulting fusion proteins carried HRV 3C-cleavable mCherry-His 10 tags.
Heterologous expression of Slo1 and c subunits in HEK293S GnTI-cells was performed using the BacMam method [38].Bacmids containing either Slo1 or c1-4 cDNA were produced in Escherichia coli EMBacY cells by transforming the respective pEG BacMam constructs.
Slo1 and c subunits were co-expressed in HEK293S GnTI-cells.Cells were cultured at 37 °C to a density of ~3 9 10 6 cellsÁmL À1 and infected with 7% (v:v) Slo1 baculovirus and 11% (v:v) c1 baculovirus (optimal concentrations of baculovirus had been determined previously by empirically testing combinations within a range of 3-15%).After 8 h, 10 mM sodium butyrate was added, and the temperature was reduced to 30 °C.~44 h post-induction, cells were collected by centrifugation and stored at À80 °C until further use.

Purification of the Slo1-c1 complex
Purification of the Slo1-c1 complex was performed according to [8,11].In order to ensure that purified Slo1 complexes carried at least one c1 protomer, we devised a dual-affinity where after the initial affinity purification via Slo1-Strep, we included a second affinity step utilizing the C-terminal His 10 tag on c1 (Fig. S2).
All purification steps were performed either on ice or at 4 °C.Cells were lysed by homogenization in hypotonic buffer containing 10 mM Tris-HCl pH 8.0 and 2 mM EDTA, supplemented with protease inhibitors (cOmplete TM Mini EDTA-free Protease Inhibitor Cocktail; Roche, Mannheim, Germany).The cell lysate was centrifuged for 15 min at 38 000 rpm and the membrane fraction was resuspended in hypotonic buffer.The lysate was centrifuged again for 15 min at 38 000 rpm and the membrane fraction was then resuspended in a basic buffer containing 20 mM Tris-HCl pH 7.6, 320 mM KCl, 10 mM CaCl 2 and 10 mM MgCl 2 , supplemented with protease inhibitors.Membranes were solubilized for 3 h using 1% (w/v) lauryl maltose neopentyl glycol (LMNG) (Neo Biotech, Nanterre, France) and 0.1% (w/v) cholesteryl hemisuccinate (CHS) (Sigma Aldrich, St. Louis, MO, USA).Subsequently, the sample was centrifuged at 38 000 rpm for 30 min.
The eluate of the Strep-Tactin beads was applied to Ni-NTA beads (QIAGEN, Hilden, Germany).After 1 h, the supernatant was removed and beads washed with basic buffer supplemented with 20 mM imidazole, 0.05% LMNG, and 0.005% CHS.The Slo1-c1 complex was eluted by a step gradient to basic buffer supplemented with 300 mM imidazole, 0.05% LMNG, and 0.005% CHS.Finally, the Ni-NTA eluate was applied to size exclusion chromatography on a Superose 6 Increase 5/150 column (Cytiva, Dreieich, Germany) in basic buffer supplemented with 0.003% LMNG and 0.0003% CHS.
The peak fractions corresponding to the Slo1-c1 complex were concentrated to 4.3 mgÁmL À1 and used for cryo-EM grid preparation.

Cryo-EM sample preparation and data acquisition
Grids were prepared using a Vitrobot Mark IV (Thermo Fisher Scientific, Eindhoven, Netherlands) at 13 °C and 100% humidity.4 lL of 4.3 mgÁmL À1 Slo1-c1 complex were applied to glow-discharged UltrAuFoil R2/2 200 grids (Quantifoil, Großl€ obichau, Germany).Any excess liquid was removed by blotting for 3.5 s at a blot force of À3, before the grids were vitrified by plunging them into liquid ethane.cryo-EM data were acquired on a Cs-corrected Titan Krios G3 electron microscope (Thermo Fisher Scientific, Eindhoven, Netherlands) equipped with a field emission gun.With a K3 camera (Gatan), a total of 16 802 movies were recorded in super-resolution mode at a nominal magnification of 105 0009.This resulted in a super-resolution pixel size of 0.34 A. Zero-loss filtration was performed using a Bioquantum post-column energy filter (Gatan, Pleasanton, CA, USA), with a slit width of 15 eV.Distributed over 60 frames, the total electron exposure was 52.3 e À Á A À2 .The dataset was collected using the automated data acquisition software EPU (Thermo Fisher Scientific, Eindhoven, Netherlands).Four acquisitions were acquired per hole using a defocus range of À1.2 to À2.4 lm.Details and statistics of data acquisition can be found in Table 1.

Data processing and model building
Data preprocessing in tranSPHIRE [39] included motion correction by MotionCor2 [40] and CTF estimation in CTFFIND4 [41].5 603 112 particles were picked using SPHIRE-crYOLO [42] and extracted using a box size of 256 9 256 pixels after 2-fold binning.2D classification was performed within cryoSPARC [43].3 284 471 particles that were assigned to well-resolved classes were used to create three ab initio models that were then used as references in a heterogeneous refinement; this and all further reconstruction steps were performed while applying C4 symmetry.A subsequent homogeneous refinement using the 2 196 730 particles assigned to the best class resulted in an initial 2.9 A reconstruction.These particles were then converted and further cleaned by 3D classification in RELION 3.1 [44]. 1 973 668 particles were subjected to particle polishing in RELION followed by non-uniform refinement including CTF refinement in CRYOSPARC, yielding a resolution of 2. 43 A. Two more rounds of ab initio modeling and heterogeneous refinement resulted in a subset comprising 826 667 particles which were refined by non-uniformed refinement to a slightly improved resolution of 2. 39 A The final reconstruction was post-processed in PHENIX [45] by applying a sharpening B factor of 73.7 A 2 .The density map shown in Fig. 1C has been post-processed using DEEPEMHANCER [46].
An initial molecular model of a Slo1 tetramer with bound c1 transmembrane and cytosolic region (residues 250-329) was created using ALPHAFOLD-MULTIMER [47], manually truncated and adjusted in COOT [48] and optimized using real space refinement against the PHENIX-sharpened real space map in PHENIX [45].All structural figures were created using UCSF CHIMERAX [49].
Details and statistics of data processing and model building can be found in Fig. S3 and Table 1.

Results
In order to gain structural and mechanistic insight into the regulation of Slo1 by auxiliary subunits, we coexpressed eGFP-Strep-tagged rabbit Slo1 with either of the four mCherry-His 10 -tagged c subunits (Fig. 1A and Fig. S1).We then purified the complex of Slo1 and c1/LRRC26 via a double affinity protocol (Fig. S2A).The complex eluted from the size exclusion chromatography column in a single symmetric peak and SDS-PAGE confirmed its purity (Fig. S2B,C).Using cryo-EM and single-particle analysis, we then determined a C4-symmetrized reconstruction of the purified Slo1-c1 complex at a resolution of 2.4 A (Fig. 1C, Fig. S3, Table 1) which allowed us to build a molecular model of almost full-length Slo1 and the transmembrane region (residues R251-Q298) of c1 (Fig. 1C,D and Fig. S3E).
Overall structure of the Slo1-c1 complex Rabbit Slo1, which has been a model system for molecular characterization of BK channels since the discovery of calcium-activated potassium flux more than 40 years ago [50], adopts a very similar overall structure and conformation to the previously published cryo-EM structures of Ca 2+ -bound human Slo1 (PDB 6V38) [10], Aplysia Slo1 (PDB 5TJI) [9,11] and Drosophila Slo (PDB 7PXE) [8], with RMSDs of 0.75, 3.07 and 1.79 A, respectively (Fig. S4A-C).The transmembrane module of Slo1 harbors the central selectivity filter in which four K + ions are stabilized by the conserved TVGYGD motif and the pore domain helices S5 and S6.This pore domain is surrounded by four voltage-sensor domains constructed of helices S0-S4 that extend to the four corners of a square when looking at the membrane from the extracellular side (Fig. 1B,D).Helix S6 connects via an extended linker to the first of two 'regulator of potassium conductance' (RCK) domains.The eight RCK domains of the Slo tetramer associate into a large but compact gating ring that makes up the majority of the mass of the channel.In agreement with the presence of 10 mM Ca 2+ during purification, all eight Ca 2+ -binding sites in the gating ring are occupied by ions and Slo1 adopts the active conformation characterized by an expanded gating ring, an extended S6 linker and an open intracellular gate [8,10,11].Similarly, the previously described Mg 2+ -binding sites between the gating ring and the VSDs are occupied [10,11].
The c1 transmembrane segment extends from the extracellular face of the VSDs, follows along their distal sides as transmembrane domain and protrudes into the cytosol as a helical polybasic stretch, after which there is no ordered density for the C-terminal tail.Furthermore, we observed a very diffuse density centrally above the selectivity filter (Fig. S5), which we attribute to the leucine-rich repeat domains of c1.This position is similar to the place where the b4 cage is localized in the structure of b4-bound human Slo1 (Fig. 3D,E) [10].However, the density was not ordered well enough to build a molecular model of the LRRs.

The kinked transmembrane helix of c1 binds the active VSD
The core of the interaction with Slo1 is mediated by the extended, kinked transmembrane helix of c1 (residues P259-C290) that binds along the distal side of the VSD (Fig. 1B-E).The interaction is predominantly hydrophobic and driven by perfect shape complementarity between c1 and the Ca 2+ -activated conformation of the VSD.The most prominent feature of the c1 helix is a kink at the outer leaflet of the lipid bilayer orchestrated by G269 which is invariable between c subunits and the subsequent P270 that is conserved only among some c1 homologs (Figs 1E and 2A-C and Fig. S1).The conformation of that kink running through a cleft along S0, S3 and S2 of the VSD is further stabilized by the flanking Y266 that caps the interaction toward the outside of S0 and F273 that inserts by hydrophobic and p-stacking interactions into a pocket spanned by L161 and F164 of helix S2 and S3 helix residues F187, V190, P191 and F194 (Figs 1E  and 2C).On the intracellular edge of the membrane, the S0-S1 connector helix which runs roughly parallel to the membrane plane serves as a lock for the Cterminal part of the c1 helix, effectively clamping it between its tryptophan residues W93 and W100, the latter touching the highly conserved G284 on c1 (which is an alanine in c4) (Fig. 1E).The c1 helix then continues at least three more turns into the cytosol until it almost touches the gating ring (Fig. 1C,D).
It was previously reported based on conductance measurements using chimeric c subunit constructs that the transmembrane helices and the C-terminal cytosolic tails are critical determinants of the observed differences in the ability of the c subunits to negatively shift the Slo1 voltage dependence (c1 > c2 > c3 > c4).Furthermore, when comparing the transmembrane helices, c1 and c2 are much more effective compared to c3 and c4 [32].While sequence conservation of key residues and structure predictions using ALPHAFOLD- MULTIMER [47] suggest that c2, c3 and c4 bind Slo1 in a very similar manner (Fig. 2B,F-H), sequence differences in several residues important for the interaction as well as slight differences in the ALPHAFOLD-MULTIMER predictions (Fig. 2B,F-H and Fig. S1) provide an explanation for the reported difference in effectiveness between the c subunit transmembrane helices.Specifically, the phenylalanine at the very center of the interaction with the VSD (F273 in c1 and F256 in c2) is a serine in c3 and c4 (Fig. 2B and Fig. S1), which weakens the overall interaction.Furthermore, P270 of c1 which is involved in kink stabilization, is a phenylalanine in c2 and c3 and an aliphatic residue in c4 (Fig. 2B and Fig. S1), and those larger side chains possibly force the c transmembrane helix in a different trajectory or lead to a locally different conformation of the VSD.

The c1 transmembrane domain is extended by structured features on both sides
The transmembrane helix of c1 is N-terminally connected to a hook comprising residues R251-P259 (Fig. 2A,D).This hook binds along the extracellular side of the VSD, thereby stabilizing the interaction of the transmembrane helix.The direction of the hook is defined by P259 and the invariable L257 might play an important role in stabilizing this conformation as well (Fig. 2B,D).Additionally, the side chain of the invariant C253 would be in a perfect distance and angle to form a cystine bond with C141 of the VSD (Fig. 2D).If such covalent anchoring would occur, it would directly strengthen the interaction of the kinked transmembrane domain and thereby help to stabilize the active conformation of the VSD.Cystine anchoring is known from the b2 and b4 subunits of mammalian voltage-gated sodium channels which also bind to the distal side of the VSD with the help of a single transmembrane helix and a disulfide bond and which also induce a (small) negative shift in voltage dependence [51,52].However, in our final reconstruction, we did not observe sufficient continuous density that would justify modeling such a covalent bond, and it remains unclear whether a cystine would form under native conditions.
At its C terminus, the c1 helix extends several turns beyond the membrane on the intracellular side.In rabbit c1, this R291-Q298 extension contains an impressive number of seven arginine residues (Fig. 3A,B,E).These positively charged Arg side chains are arranged around the c1 helix close to where the lipid head groups in a membrane would be and come into proximity of the RCK1 N-lobe without contacting it directly.Interestingly, the number of positively charged residues in this polybasic stretch differs between the c subunits proportional to their capability to induce negative shifts of the voltage dependence.It is highest in mammalian c1 with six or seven, while c2 contains three and c3 and c4 only two arginines or lysines (Fig. 2B and Fig. S1).Accordingly, the C-terminal tails of c1 and c3 are much more effective in shifting the voltage dependence of Slo1 compared to c2 and c4 and this depends for all but c2 entirely on the presence of the polybasic stretch [32].Thus, it can be said that the number of positive charges in the polybasic stretch correlates with the potency in Slo1 activation by the c subunits.

Simultaneous binding of b and c subunits
Earlier observations in electrophysiology experiments suggested that b and c subunits can exist in the same Slo1 channels.More specifically, overexpression of Slo1 with b2 and c1 resulted in a À120 mV shift in absence of Ca 2+ with respect to Slo1 apo, which is a characteristic of c1 regulation, while at the same time, a b2-specific complete inactivation occurs after activation [31].This observation is very much in line with the simultaneous presence of b2 and c1 and prompted us to ask whether simultaneous binding is also structurally possible and whether rearrangements would be necessary.
We generated a composite model by superposing the b4-bound human Slo1 (PDB 6V22) [10] onto our structure of c1-bound rabbit Slo1.Indeed, both types of regulatory subunits bind the transmembrane domain of Slo1 using completely different and nonoverlapping binding sites (Fig. 3A-C).While c1 binds to the distal side of the VSD with contacts to S0, S0-S1, S2 and S3 (Figs 1E and 3B), the two transmembrane domains of b4 fit in the cleft between two neighboring VSDs (Fig. 3A), and contacts include the cytosolic part of S0 of one and the extracellular S1-S2 loop of the other VSD (Fig. 3B); also lipid molecules mediate part of the interaction [10].The four ECDs of b4 are positioned at a short distance above the transmembrane helices in a cage-like tetrameric arrangement shielding the extracellular side of the Slo1 selectivity filter (Fig. 3A,C).No clashes between the TM helices of b4 and c1 are visible and the Slo1 VSD conformation is, apart from a minor shift of the anyway slightly mobile cytosolic part of S0, virtually identical between both structures (RMSD 0.825 A over 257 Ca atoms of the transmembrane domain of a single chain).Thus, simultaneous binding of both types of regulatory subunits is possible at least in the transmembrane portions when not taking into account the extracellular domains.

Discussion
In this study, we report the cryo-EM structure of the rabbit Slo1-c1 complex.c1 features a single kinked transmembrane helix that binds to the distal side of the voltage-sensor domain.The central kink is positioned such that c1 can follow the concave surface of the VSD, allowing tight binding.The transmembrane helix is capped on the extracellular side by an extended hook that further stabilizes the interaction and Cterminally extended by a helical polybasic stretch that extends into the cytosol.
After this manuscript was posted as a preprint and while it was under review, two other manuscripts have been published which report similar structures of the human Slo1-c1 complex in the Ca 2+ -free and bound state, respectively [15,53].These data confirm the binding mode of the c1 transmembrane domain that we observe (RMSD 0.79 A of rabbit vs. human Slo1-c1, PDB 7YO3; Fig. S4D) and further validate it by electrophysiology data.Thus, by looking these three studies together, we start to get a clear overall picture of how c1 activates the Slo1 VSD in non-excitable cells under physiological conditions, i.e. in the absence of an action potential-associated membrane depolarization (Fig. 4).
On the structural level, depolarization of the membrane potential, i.e. a reversal of the local charge gradient across the membrane, leads to a dragging force on the positive arginine gating charges in the S4 helices of ion channels toward the extracellular side [14,15,20,54].The following rotation of the C-terminal half of S4 and displacement of the S4-S5 linker primes Slo1 activation by allowing the S6 helix in the cytosolic gate to adopt the open conformation [14,15].Hence, a possible activation mechanism of c1 is to change the energy landscape for gating charge movement toward the extracellular side, which would make a VSD rearrangement and gate opening easier and more likely.
We propose that three structural elements act together to cause such a shift.The first is the shape complementarity of the c1 transmembrane helix with respect to the surface of the activated state of the VSD (Figs 1E and 2A).In particular, the central kink in the helix around G269-P270 seems important as the helix must change direction to be accommodated at the concave surface of the VSD (Fig. 2C).The perfect fit of c1 to the activated conformation of the VSD and the intense interactions especially with the S0-S1 connector helix that moves during voltage activation might indicate a preference for this state and a stabilization of the up conformation of the gating charges.Conversely, preferential binding of c subunits to the activated state might lower the energy cost of the voltage-driven rearrangement and thereby stabilize the active conformation.The kink is likely most pronounced in c1 due to the G269-P270 amino acid tandem, and the absence of a proline in the equivalent position, together with the substitution of the central F273 by serine, could be responsible for the lower effectiveness of activation by c3 and c4 transmembrane helices [32].Indeed, several Slo1 mutations which reduce the affinity of c1 to the voltage-sensor domain reduced the negative shift in voltage dependence induced by c1 [15,53].
The extracellular hook seems to stabilize the interaction with Slo1 further as it extends the interaction surface and contributes important contacts, thereby further locking the active conformation of the voltage sensor.Especially, a covalent cystine bridge between c1 C253 and Slo1 141 would strongly tether c1 in this position and help translate the stabilizing effect of the kink toward the extracellular side of the S4 helix, even though it remains unclear whether such a bond exists in native Slo1-c complexes.Intriguingly, Yamanouchi et al. [53] found that a Y198A in human Slo1 that should destabilize the conformation of the hook decreased the negative voltage shift induced by c1.
The mechanism by which the polybasic tail at the C terminus of the c1 transmembrane helix affects the voltage dependence shift is not yet completely clear.It has been observed that the À140 mV negative shift induced by c1 does not change the Ca 2+ sensitivity of Slo1 [25].Thus, its activity is described in the In the absence of c subunits, Slo1 is activated in excitable cells synergistically by membrane potential depolarization and intracellular calcium binding.In the resting state, a charge gradient over the membrane exists with negative net charges on the intracellular side and positive charges on the extracellular side.This distribution favors the resting state conformation of the voltage-sensor domains of Slo1 in which the two (positively charged) arginine gating charges reside close to the intracellular side of the membrane.Depolarization, i.e. charge gradient reversal, pulls the gating charges toward the extracellular side, leading to a rearrangement throughout the entire VSD.This is synergistically enhanced by binding of calcium ions to the gating ring, which induces an outwards movement of the S6 helix and opening of the intracellular gate.In non-excitable cells, where the resting state transmembrane potential is constant, binding of the kinked transmembrane helices of c subunits stabilizes the active conformation of the VSDs.Furthermore, we propose that the positive charges in the C-terminal region of the c subunits locally change the membrane potential and induce a similar movement of the Slo1 gating charges toward the extracellular side as observed during action potential.Thus, the need for the voltage depolarization signal is decreased or completely absent and only binding of calcium is necessary to open Slo1.
Horrigan-Aldrich model of Slo1 activation by 20-fold increase in the allosteric coupling of VSD activation and pore domain opening [2,25].In their recent publication, Yamanouchi et al. [53] challenged this by presenting data that suggests a direct coupling between this polybasic stretch and the close-by RCK1 N-lobe, which the authors claim would stabilize the active conformation of the gating ring.However, the proposed interaction between R295 of c1 and D326 and D370 of Hs Slo1 is not completely backed by their structural data as the cryo-EM reconstruction is very noisy in this region and no clear side chain density is visible.Even in our higher resolution structure, we did not observe any such ordered density for the R295 side chain despite very similar experimental conditions and overall virtually identical conformation.Yamanouchi et al. [53] additionally present electrophysiology data using R295A and D369A/D370A, respectively, which show a strong decrease in c1-mediated voltage dependence shift compared to wildtype c1.Thus, even though a direct interaction between the polybasic tail and the RCK1 N-lobe is at the moment not supported by structural data, it might contribute to the activity of c1.
On the other hand, the close proximity of the polybasic stretch to the VSD suggests another compelling explanation for its importance for the voltage dependence shift [32].During resting potential, the intracellular side of the plasma membrane is negatively charged, while the extracellular side is positively charged.Upon depolarization, this is reversed, causing the gating charges in S4 to follow the voltage change and move toward the extracellular side [54].The presence of several immobile positive charges on the intracellular side in close proximity to the VSD as in the case of the Slo1-c1 complex influences this charge distribution over the membrane, as already during resting potential there are positive charges on the cytosolic side, leading to a locally lower negative net charge surrounding the Slo1-c1 complex.Thus, the polybasic stretch might locally lower the resting state potential and repulse the gating charges, thereby reducing the energy to overcome for the VSD to transition to the active conformation.In particular, the activation of the VSD in excitable cells in the absence of c subunits requires the accumulation of positive charges on the intracellular side of the membrane close to the VSD (by influx of Na + ions through voltage-gated Na + channels), and the positive charges of the polybasic stretch of c1 might mimic this (Fig. 4).The electrophysiology result of the R295A mutant from Yamanouchi et al. [53] would also be in line with this model.Finally, it is worth mentioning that polybasic stretches in other proteins such as small GTPases have been shown to be important for association with lipid head groups and protein interactors [55], making a similar role of the c subunit polybasic stretch at least a possibility.The current data from this study and the complementary studies from Yamanouchi et al. and Kallure et al. do not allow a final conclusion on this question of the role of the polybasic stretch.Thus, further studies such as molecular dynamics simulations would be necessary to follow up on this.
Beyond this, it will be interesting to experimentally test whether simultaneous binding of b and c subunits is possible, especially regarding the necessary rearrangements within the extracellular domains (Fig. 3F).In particular, how the ECDs of b subunits and the LRRs of c subunits would arrange in case they are present on the same Slo1 tetramer is an open question, as the arrangement of c1 LRRs would clash with the cage formed by a b subunit tetramer.Thus, either the LRR tetramer would need to open up to prevent clashes with the b ECDs (Fig. 3F), or the b ECDs would need to.Also, future studies should investigate in more detail the molecular roles of the b and c extracellular domains which, due to their high conservation, must certainly be important for Slo1 function.As suggested previously for b subunits, the ECDs might bind to extracellular matrix proteins [10].
Besides Slo, also many other ion channels including voltage-gated sodium, potassium, and calcium channels as well as TRP channels contain auxiliary subunits which regulate their activation [16,18,19].While cystine bond anchoring has been shown at least for Na V b2 and b4 [51,52], other auxiliary subunits which induce local changes in transmembrane voltage potential by placing immobile charges close to the VSD have not been reported to the best of our knowledge.Thus, the activation mode by a polybasic stretch as we propose for Slo c subunits, if true, would be a novel mechanism that expands the repertoire of auxiliary subunits, and it will be very interesting to see whether similar modes of action are also applied by other proteins.

Fig. 1 .
Fig. 1.The transmembrane helix of c1 binds to the voltage-sensor domain of Slo1.(A) Schematic representation of c subunits.c1-c4 possess highly similar architectures comprising an N-terminal membrane-targeting signal sequence, an extracellular domain containing eight leucine-rich repeats, a transmembrane domain and a short C-terminal cytosolic tail.(B) Schematic of the Slo1 transmembrane domain topology and the c1 transmembrane helix which binds diagonally across helices S0, S3, S2 and the perpendicular S0-S1 connector helix, as observed in our Slo1-c1 structure.(C) Cryo-EM reconstruction of rabbit Slo1 in complex with c1 in two orientations.Clear density corresponding to the c1 transmembrane helix is visible at the four voltage-sensor domains.(D) Model of the c1-bound rabbit Slo1 in the same orientations as in panel (C).The c1 transmembrane helices bind to the four corners of the transmembrane domain tetramer, contacting the voltage-sensor domains.(E) The c1 transmembrane helix forms extensive hydrophobic contacts with helices S0, S0-S1, S2, and S3 of the Slo1 voltage-sensor domain and binds with perfect shape complementarity.

Fig. 2 .
Fig. 2. Three structural elements of c subunits modulate the function of Slo1.(A) Three structural elements in c1 likely contribute to VSD activation: a kink in the central transmembrane helix, a hook N-terminal of it and a polybasic stretch on the intracellular side.(B) Alignment of rabbit c subunit transmembrane region containing the hook, the kink, and the polybasic stretch.The cysteine in the hook which potentially forms a disulfide bond with Slo1 C141 is marked in red and the basic residues in the polybasic stretch in blue.(C) Close-up of the kink with important residues on both sides shown in stick representation.The half-transparent surface of the VSD illustrates the shape complementarity with c1.(D) Close-up of the hook region of c1 N-terminal of the transmembrane helix.The hook contacts the VSD via R255 and the invariant L257 and C253.(E) Close-up of the helical extension of the transmembrane helix that protrudes into the cytosol and harbors the polybasic stretch.In rabbit c1, the polybasic stretch contains seven arginine residues and thereby produces a strongly positively charged microenvironment close to the voltage-sensor domain.(F) Structure prediction of c2 binding to the Slo1 VSD.The models in panels (F-H) were created by ALPHAFOLD-MULTIMER, followed by careful manual adjustments.(G) Structure prediction of c3 binding to the Slo1 VSD.(H) Structure prediction of c4 binding to the Slo1 VSD.

Fig. 3 .
Fig. 3. Simultaneous binding of b and c subunits to Slo1 is structurally possible, but would require rearrangements in their extracellular modules.(A) Superposition of the human Slo1-b4 structure (PDB 6V22) onto our rabbit Slo1-c1 structure; the human Slo1 was omitted for clarity.The extracellular domains (ECDs) of the b subunits form a cage above the Slo1 transmembrane domain.(B) Close-up view on one of the VSDs of Slo1 with bound b and c subunit transmembrane helices as shown in panel (A).While the c transmembrane helix binds diagonally at the most distal side of the VSD, one b transmembrane helix pair contacts the intracellular part of helix S0, while another b subunit contacts the VSD from the other side mostly via the extracellular S1-S2 loop.(C) Cartoon model of Slo1 with bound b subunits and c subunit transmembrane helices.(D) At low threshold, weak density is visible centrally above the Slo1 transmembrane domains in a lowerresolution reconstruction (see also Fig. S5).Superposition of the human Slo1-c1 onto our rabbit structure places the c1 LRR domains (pink) into this weak density.(E) Cartoon model of Slo1 with full-length c subunits.The LRRs display some flexibility in all reported structures.(F) Cartoon of Slo1 with bound b subunits and full-length c subunits.A rearrangement (outwards movement) of c LRRs would be necessary to prevent clashes with b ECDs.

Fig. 4 .
Fig.4.Proposed model for Slo1 activation by c1.In the absence of c subunits, Slo1 is activated in excitable cells synergistically by membrane potential depolarization and intracellular calcium binding.In the resting state, a charge gradient over the membrane exists with negative net charges on the intracellular side and positive charges on the extracellular side.This distribution favors the resting state conformation of the voltage-sensor domains of Slo1 in which the two (positively charged) arginine gating charges reside close to the intracellular side of the membrane.Depolarization, i.e. charge gradient reversal, pulls the gating charges toward the extracellular side, leading to a rearrangement throughout the entire VSD.This is synergistically enhanced by binding of calcium ions to the gating ring, which induces an outwards movement of the S6 helix and opening of the intracellular gate.In non-excitable cells, where the resting state transmembrane potential is constant, binding of the kinked transmembrane helices of c subunits stabilizes the active conformation of the VSDs.Furthermore, we propose that the positive charges in the C-terminal region of the c subunits locally change the membrane potential and induce a similar movement of the Slo1 gating charges toward the extracellular side as observed during action potential.Thus, the need for the voltage depolarization signal is decreased or completely absent and only binding of calcium is necessary to open Slo1.