Research organism

elife

eLife

2050-084X

eLife Sciences Publications, Ltd

54670

10.7554/eLife.54670

Research Article

Neuroscience

Structural Biology and Molecular Biophysics

The Cryo-EM structure of pannexin 1 reveals unique motifs for ion selection and inhibition

Michalski

Kevin

1†Syrjanen

Johanna L

2†Henze

Erik

1Kumpf

Julia

http://orcid.org/0000-0002-3813-1255

1Furukawa

Hiro

furukawa@cshl.edu2Kawate

Toshimitsu

https://orcid.org/0000-0002-5005-2031

toshi.kawate@cornell.edu11Department of Molecular Medicine, Cornell University

Ithaca

United States2WM Keck Structural Biology Laboratory, Cold Spring Harbor Laboratory

Cold Spring Harbor

United States

Swartz

Kenton J

Senior EditorNational Institute of Neurological Disorders and Stroke, National Institutes of HealthUnited StatesSwartz

Kenton J

Reviewing EditorNational Institute of Neurological Disorders and Stroke, National Institutes of HealthUnited States

†

These authors contributed equally to this work

12022020

2020

e54670

2112201911022020

2020

Michalski et al

http://creativecommons.org/licenses/by/4.0/

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Pannexins are large-pore forming channels responsible for ATP release under a variety of physiological and pathological conditions. Although predicted to share similar membrane topology with other large-pore forming proteins such as connexins, innexins, and LRRC8, pannexins have minimal sequence similarity to these protein families. Here, we present the cryo-EM structure of a frog pannexin 1 (Panx1) channel at 3.0 Å. We find that Panx1 protomers harbor four transmembrane helices similar in arrangement to other large-pore forming proteins but assemble as a heptameric channel with a unique constriction formed by Trp74 in the first extracellular loop. Mutating Trp74 or the nearby Arg75 disrupt ion selectivity, whereas altering residues in the hydrophobic groove formed by the two extracellular loops abrogates channel inhibition by carbenoxolone. Our structural and functional study establishes the extracellular loops as important structural motifs for ion selectivity and channel inhibition in Panx1.

pannexinheptameric channelATP releaseextracellular loopion selectivitycarbenoxolone

Research organismHumanXenopus

http://dx.doi.org/10.13039/100000002

National Institutes of Health

GM114379

Kawate

Toshimitsu

http://dx.doi.org/10.13039/100000002

National Institutes of Health

NS113632

Furukawa

Hiro

http://dx.doi.org/10.13039/100000002

National Institutes of Health

GM008267

Michalski

Kevin

Henze

Erik

Kumpf

Julia

http://dx.doi.org/10.13039/100013356

Cold Spring Harbor Laboratory

Robertson funds

Furukawa

Hiro

Doug Fox Alzheimer's fund

Furukawa

Hiro

Austin's Purpose

Furukawa

Hiro

Heartfelt WingAlzheimer's Fund

Furukawa

Hiro

http://dx.doi.org/10.13039/100001312

Charles H. Revson Foundation

Senior Fellowship in Biomedical Science

Syrjanen

Johanna L

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Author impact statement

The structure of a pannexin 1 uncovers the extracellular loops of pannexin 1 as important structural motifs for ion selectivity and channel inhibition.

Introduction

Large-pore forming channels play important roles in cell-to-cell communication by responding to diverse stimuli and releasing signaling molecules like ATP and amino acids (Giaume et al., 2013; Ma et al., 2016; Okada et al., 2018; Osei-Owusu et al., 2018). Pannexins are a family of ubiquitously expressed large-pore forming channels which regulate nucleotide release during apoptosis (Chekeni et al., 2010), blood pressure (Billaud et al., 2011; Billaud et al., 2015), and neuropathic pain (Bravo et al., 2014; Weaver et al., 2017; Mousseau et al., 2018). While pannexins have limited sequence identity with innexins (~15% identity), they have virtually no sequence similarity to other large-pore forming channels (Panchin et al., 2000). Among the pannexin family, pannexin 1 (Panx1) has garnered the most attention for its role as a large-pore forming channel responsible for ATP release from a variety of cell types (Bao et al., 2004; Dahl, 2015). Different kinds of stimuli have been reported to activate Panx1 including voltage, membrane stretch, increased intracellular calcium levels, and positive membrane potentials (Bruzzone et al., 2003; Bao et al., 2004; Locovei et al., 2006; Wang et al., 2014; Chiu et al., 2018). Panx1 is also targeted by signaling effectors, such as proteases and kinases, to permanently or temporarily stimulate channel activity (Pelegrin and Surprenant, 2006; Thompson et al., 2008; Sandilos et al., 2012; Billaud et al., 2015; Lohman et al., 2015). The above evidence suggests that Panx1 has a capacity to integrate distinct stimuli into channel activation leading to ATP release. Despite playing critical roles in a variety of biological processes, a mechanistic understanding of pannexin function has been largely limited due to the lack of a high-resolution structure. Here, we show the cryo-EM structure of Panx1, which reveals the pattern of heptameric assembly, pore lining residues, important residues for ion selection, and a putative carbenoxolone binding site.

ResultsStructure determination and functional characterization

To identify a pannexin channel suitable for structure determination, we screened 34 pannexin orthologues using Fluorescence Size Exclusion Chromatography (FSEC)(Kawate and Gouaux, 2006). Frog Panx1 (frPanx1; 66% identical to human, Figure 1—figure supplement 1) displayed high expression levels and remained monodisperse when solubilized in detergent, suggesting high biochemical integrity. We further stabilized frPanx1 by truncating the C-terminus by 71 amino acids and by removing 24 amino acids from the intracellular loop between transmembrane helices 2 and 3 (Figure 1—figure supplement 1). This construct, dubbed ‘frPanx1-ΔLC’, displayed high stability in detergents and could be purified to homogeneity (Figure 1—figure supplement 2a and b). We verified that frPanx1 forms a functional pannexin channel by whole-cell patch clamp electrophysiology (Figure 1a and b; Figure 1—figure supplement 2e and f). Purified frPanx1-ΔLC was reconstituted into nanodiscs composed of MSP2N2 (an engineered derivative of apolipoprotein) and soybean polar lipids, and subjected to cryo-electron microscopy (cryo-EM) and single-particle analysis (Figure 1—figure supplement 2c and d). We used a total of 90,185 selected particles for 3D reconstruction at 3.0 Å resolution (Figure 1—figure supplement 3). The map quality was sufficient for de novo model building for the majority of frPanx1-ΔLC with the exception of disordered segments of the N-terminus (residues 1–10), ECL1 (88–100), and ICL1 (157–194) (Figure 1c; Figure 1—figure supplement 4, Video 1, and Table 1).

Figure 1.frPanx1 forms a heptameric ion channel.

(a) Whole-cell patch clamp recordings from HEK 293 cells expressing hPanx1, frPanx1, and frPanx1-ΔLC. Cells were clamped at −60 mV and stepped from −100 mV to +100 mV for 1 s in 20 mV increments. To facilitate electrophysiological studies, we inserted a Gly-Ser motif immediately after the start Met to enhance Panx1 channel opening as we previously described (Michalski et al., 2018). CBX (100 μM) was applied through a rapid solution exchanger. (b) Current-voltage plot of the same channels shown in (a). Recordings performed in normal external buffer are shown as circles, and those performed during CBX (100 μM) application are shown as squares. Each point represents the mean of at least three different recordings, and error bars represent the SEM. (c) EM map of frPanx1-ΔLC shown from within the plane of the membrane. Each protomer is colored differently, with the extracellular side designated as ‘out’ and the intracellular side as ‘in.’ (d) Overall structure of frPanx1-ΔLC viewed from within the lipid bilayer. (e) Structure of frPanx1 viewed from the extracellular face.

Figure 1—figure supplement 1.Sequence alignment and structural features.

Amino acid sequence of frPanx1 compared to various Panx1 orthologues. Amino acids highlighted in black are completely conserved, in gray are similar, and in white are not conserved. Disulfide-forming cysteines are highlighted in yellow. Secondary structure features are shown above the sequence alignment. Green boxes depict the transmembrane helix boundaries, and orange boxes show deleted or truncated regions to create frPanx1-ΔLC.

Figure 1—figure supplement 2.Characterization of frPanx1-ΔLC.

(a) Size exclusion chromatogram of frPanx1-ΔLC. Concentrated protein was injected onto a Superose 6 10/300 column equilibrated with 150 mM NaCl, 10 mM Tris pH 8.0, 1 mM EDTA, 0.5 mM DDM. (b) SDS-PAGE analysis of peak fractions collected from (a). (c) Size exclusion chromatogram of frPanx1-ΔLC after reconstitution into nanodiscs. The running buffer contained 150 mM NaCl, 10 mM Tris pH 8.0, 1 mM EDTA. (d) SDS-PAGE analysis of peak fractions collected from (c). (e) Whole-cell recordings of wild-type (no Gly-Ser) hPanx1, frPanx1, and frPanx1-ΔLC. Whole-cell patches from transfected HEK 293 T cells were obtained, held at −60 mV, and stepped between −100 mV and +160 mV for 1 s. CBX (100 μM) was applied through a rapid solution exchanger. (f) Current-voltage plot of wild-type pannexin recordings shown in (e). Each circle represents the mean current at a particular voltage, with squares depicting the same current when treated with 100 μM CBX. N = 3–11.

Figure 1—figure supplement 3.Cryo-EM image processing workflow for single particle analysis of frPanx1-ΔLC.

(a) A representative micrograph (scale bar = 50 nm), representative 2D class averages, and the 3D classification workflow are shown. (b) The FSC plots of the two half maps (top) and the map vs model (bottom) are shown. (c) The angular distribution plot for class 3. (d) Local resolutions of class three were calculated using ResMap (Kucukelbir et al., 2014).

Figure 1—figure supplement 4.Representative cryo-EM density of frPanx1-ΔLC.

Each domain is shown as stick representation and fit into the corresponding density contoured at σ = 3.0.

Table 1.Cryo-EM data collection, refinement and validation statistics.

	frPanx- ΔLC (EMD-21150) (PDB: 6VD7)
Data collection and processing
Magnification	130,000
Voltage (kV)	300
Electron exposure (e–/Å²)	57.2
Defocus range (μm)	1.2–2.8
Pixel size (Å)	1.07
Symmetry imposed	C7
Initial particle images (no.)	297374
Final particle images (no.)	90185
Map resolution (Å) FSC threshold	3.02 0.143
Refinement
Initial model used (PDB code)	de novo
Model resolution (Å) FSC threshold	3.29 0.5
Model resolution range (Å)	3–6
Map sharpening B factor (Å²)	−90
Model composition Non-hydrogen atoms Protein residues Ligands	16506 2079 0
CC map vs. model (%)	0.85
R.m.s. deviations Bond lengths (Å) Bond angles (°)	0.008 0.759
Validation MolProbity score Clashscore Poor rotamers (%)	1.92 5.96 0.78
Ramachandran plot Favored (%) Allowed (%) Disallowed (%)	88.32 11.68 0

Video 1.Cryo-EM density of frPanx1-ΔLC.

The model is shown as wire representation and fit into the corresponding density contoured at σ = 3.0. Each domain is colored differently and Tryp74 and Arg75 are labeled in the close-up view.

Overall structure and protomer features

The frPanx1-ΔLC structure revealed a heptameric assembly, which is unique among the known eukaryotic channels (Figure 1d and e). Other large-pore forming channels include hexameric connexins (Maeda et al., 2009) and LRRC8s (Deneka et al., 2018; Kasuya et al., 2018; Kefauver et al., 2018), and the octameric innexins (Oshima et al., 2016) and calcium homeostasis modulator1 (CALHM1) (Syrjanen et al., 2020; Figure 2—figure supplement 1). Our result differs from previous studies that suggest hexameric assembly of pannexin based on single channel recordings on concatemeric channels and negative stain electron microscopy (Boassa et al., 2007; Wang et al., 2014; Chiu et al., 2017). The heptameric assembly observed in the current study is unlikely to be caused by the carboxy-terminal truncation or intracellular loop deletion because cryo-EM images of the full-length frPanx1 also display clear seven-fold symmetry in the 2D class averages (Figure 2—figure supplement 2a). Furthermore, 2D class averages of hPanx1 display a heptameric assembly, but not other oligomeric states (Figure 2—figure supplement 2b). Thus, overall, our data suggests that the major oligomeric state of Panx1 is a heptamer. This unique heptameric assembly is established by inter-subunit interactions at three locations: 1) ECL1s and the loop between β2 and β3; 2) TM1-TM1 and TM2-TM4 interfaces; and 3) α9 helix and the surrounding α3 and α4 helices, and the N-terminal loop from the neighboring subunit (Figure 2—figure supplement 3). Notably, the majority of residues mediating these interactions are highly conserved (e.g. Phe67 and Tyr111; Figure 1—figure supplement 1).

The overall protomer structure of Panx1 resembles that of other large-pore forming channels including connexin, innexins, and LRRC8. Like other large-pore forming channels, each Panx1 protomer harbors four transmembrane helices (TM1-4), two extracellular loops (ECL1 and 2), two intracellular loops (ICL1 and 2), and an amino (N)-terminal loop (Figure 2a and b). The transmembrane helices of Panx1 are assembled as a bundle in which the overall helix lengths, angles, and positions strongly resemble the transmembrane arrangements observed in other large-pore channels (Figure 2c). In contrast, Panx1 has no similarity in transmembrane arrangement to another group of large-pore channels, CALHMs whose protomers also contain four transmembrane helices (Choi et al., 2019; Syrjanen et al., 2020; Figure 2—figure supplement 1). Structural features in the Panx1 ECL1 and ECL2 domains are conserved among large-pore channels despite limited sequence similarity (Figure 2d–g; Figure 2—figure supplement 1). For example, the Panx1 ECL1 and ECL2 are joined together by two conserved disulfide bonds (Cys66 with Cys267, Cys84 with Cys248) in addition to several β-strands. ECL1 also contains an alpha-helix that extends towards the central pore and forms an extracellular constriction of the permeation pathway. While much of the transmembrane domains and extracellular loops show similarities to other large-pore forming channels, the Panx1 intracellular domains are structurally unique (Figure 2—figure supplement 1). ICL1 and ICL2, for example, together form a bundle of helices that make contact with the N-terminus. The N-terminal loop of Panx1 forms a constriction of the permeation pathway and extends towards the intracellular region. The first ~10 amino acids of the N-terminus are disordered in our structure, but these residues might play a role in ion permeation or ion selectivity (Wang and Dahl, 2010).

Figure 2.Subunit architecture of frPanx1.

(a) Structure of the frPanx1 protomer. Each domain is colored according to the cartoon scheme presented in (b). (c) Superimposition of the transmembrane helices from frPanx1 (red), connexin-26 (green), innexin-6 (orange), and LRRC8 (blue) shown top-down from the extracellular side (top) or from within the plane of the membrane (bottom). (d-g) Cartoon representation of the extracellular loops of large pore forming channels. ECL1 is colored in light blue, and ECL2 is colored in dark blue, and disulfide bridges are shown as yellow spheres. These domains are viewed from the same angle (from top) as shown in the top panel in (c).

Figure 2—figure supplement 1.Comparison of frPanx1 with other large pore channels.

The structures of connexin-26 (PDB: 2ZW3), LRRC8 (PDB: 6NZW), Panx1, innexin-6 (PDB: 5H1Q), and CALHM1 (PDB: 6VAM) are shown from within the plane of the membrane (top) and viewed from the extracellular side (bottom). One subunit of each channel is shown colored with transmembrane domains in orange/yellow, ECL1 is colored in light blue, ECL2 in dark blue, ICL1 in gray, ICL2 in green, and NTL in red, as shown in Figure 2a.

Figure 2—figure supplement 2.2D classes of full-length frog and human pannexin 1.

(a) 2D classes of the full-length frog pannexin1 in nanodiscs showing top (left) and side (right) views. All seven protomers are labeled with numbers in the 2D class of the top view. (b) 2D classes of the full-length human pannexin one in DDM in the similar orientation to panel a. The top view shows a heptameric assembly (numbered). No symmetry was imposed in 2D classification.

Figure 2—figure supplement 3.Inter-subunit interactions.

Three major inter-subunit interfaces between two neighboring subunits (blue and red) are highlighted in orange ovals (left). Close-up views (right) show the highly-conserved residues mediating the inter-subunit interactions.

Ion permeation pathway and selectivity

The Panx1 permeation pathway spans a length of 104 Å, with constrictions formed by the N-terminal loop, Ile58, and Trp74 (Figure 3a and b). The narrowest constriction is surrounded by Trp74 located on ECL1 (Figure 3c). Trp74 is highly conserved among species including hPanx1 (Figure 1—figure supplement 1). Because Panx1 has been previously characterized as an anion selective channel (Ma et al., 2012; Romanov et al., 2012; Chiu et al., 2014), we wondered if positively charged amino acids around the narrowest constriction formed by Trp74 may contribute to anion selectivity of the channel. Interestingly, Arg75 is situated nearest to the tightest constriction of the permeation pathway (Figure 3d). We hypothesized that Arg75 might be a major determinant of anion selectivity of Panx1 channels in the open state. To assess whether Arg75 contributes to anion selectivity, we generated a series of point mutations at this position on hPanx1 and compared their reversal potentials (Erev) in asymmetric solutions using whole-cell patch clamp electrophysiology (Figure 3e and Figure 3—figure supplement 1). We kept sodium chloride (NaCl) constant in the pipette solution while varying the extracellular solution. When treated with the large anion, gluconate (Gluc^-), Erev shifted to +26 mV, suggesting the channel is more permeable to Cl^- than to Gluc^-. When exposed to the large cation, N-methyl-D-glucamine (NMDG⁺), Erev remained close to 0 mV, suggesting that Na⁺ and NMDG⁺ equally (or do not) permeate Panx1. These results are consistent with Panx1 being an anion-selective channel. The Arg75Lys mutant maintains the positive charge of this position, and displayed Erev values comparable to WT. Removing the positive charge at this position, as shown by the Arg75Ala mutant, diminished Cl^- selectivity as the Erev in NaGluc remained near 0 mV. Interestingly, the Erev in NMDGCl shifted to −22 mV, suggesting the channel had lost anion selectivity and Na⁺ became more permeable than NMDG⁺. A charge reversal mutant, Arg75Glu, shifted the Erev in NaGluc to −16 mV and in NMDGCl to −45 mV, indicating that Gluc^- became more permeable to Cl^-. Overall, these results support the idea that the positively charged Arg75 plays a role in anion selectivity of Panx1.

Figure 3.Permeation and ion selectivity of Panx1 channels.

(a) HOLE (Smart et al., 1996) diagram demonstrating constrictions along the permeation pathway. NTL; N-terminal loop. (b) Surface representation of the internal space along the molecular 7-fold axis running through the center of frPanx1. The surface was generate using HOLE. (c and d) Top view facing the extracellular side (c) or side view (d) of frPanx1, with ECL1 shown in light blue and ECL2 in dark blue. Trp74 and Arg75 are shown as sticks. (e) Reversal potentials of various hPanx1 ion selectivity mutants. Each point represents the Erev measured in NaCl (black), NaGluc (red), or NMDGCl (blue), and bars represent the mean values. I-V curves were obtained by a ramp protocol from −80 mV to +80 mV. (f) Close-up view of the Trp74-Arg75 interaction at the interface of protomer A (blue) and B (red).

Figure 3—figure supplement 1.Representative traces of the ramp recordings.

HEK293 cells were held at −60 mV and ramped between −100 mV and + 100 mV over 3 s duration. The numbers indicate the reversal potentials in mV.

Figure 3—figure supplement 2.Electrostatic surface potential of the ion permeation pathway.

A coronal section of frPanx1-ΔLC shows the surface of the ion permeation pathway. Electrostatic surface potential was calculated using APBS plugin (Baker et al., 2001) in Pymol (Schrodinger, LLC, 2015) and presented in the range between −15 kT/e (red: acidic) and +15 kT/e (blue: basic).

We next wondered if introducing a charge at position 74 might alter ion selectivity of Panx1 channels. Interestingly, both Trp74Arg and Trp74Glu mutants become less selective to anions and more permeable to Na⁺ (Figure 3e). These results suggest that introducing a charge at this position disrupts the natural ion selectivity of Panx1 channels but that position 74 itself does not control ion selectivity. We observed that the distance between the guanidino group of Arg75 and the benzene ring of Trp74 from an adjacent subunit is ~4 Å, suggesting that these two residues likely participate in an inter-subunit cation-π interaction key to Panx1 ion selectivity (Figure 3f). To test this hypothesis, we generated Trp74Ala and Trp74Phe mutations and measured Erev potentials. Trp74Ala showed a marked decrease in Cl^- permeability and an increase in Na⁺ permeability, despite preservation of the positive charge at Arg75. A more conservative mutation, Trp74Phe, still disrupted ion selectivity, suggesting that proper positioning of the benzene ring at position 74 is important for anion selection. Altogether, our data suggests that anion selectivity is only achieved when Trp74 and Arg75 form a cation-π interaction. Given that our structure has disordered and truncated regions in the N-terminus, ICL1, and ICL2, it is possible that additional ion selectivity or gating regions exist in the full-length channel. For example, the N-termini of LRRC8 and connexins perform an important role in ion selectivity (Kyle et al., 2008; Kronengold et al., 2012; Kefauver et al., 2018). It is possible that the N-terminus of Panx1 is mobile and may further constrict the permeation pathway. Another possibility is that the electrostatic potential along the pore pathway contributes to the ion selectivity. Interestingly, both cytoplasmic and extracellular entrances of the permeation pathway are mostly basic, suggesting that non-permeant cations may be excluded from the pore (Figure 3—figure supplement 2). In contrast, the region underneath the W74 constriction is highly acidic, supporting the idea that anions may be selected around this area.

CBX action mechanism

We have previously demonstrated that CBX, a potent nonselective inhibitor of Panx1, likely acts through a mechanism involving ECL1 (Michalski and Kawate, 2016). In these experiments, mutations at a number of residues in ECL1 rendered Panx1 less sensitive to CBX-mediated channel inhibition. Mapping such residues in the Panx1 structure revealed that they are clustered proximal to the extracellular constriction by Trp74, in a groove formed between ECL1 and ECL2 (Figure 4a and b). This supports our previous speculation that CBX is an allosteric inhibitor, not a channel blocker (Michalski and Kawate, 2016).

Figure 4.CBX action requires residues from both ECL1 and ECL2.

(a and b) Surface (a) and cartoon (b) representations of the frPanx1 ECL1 (light blue) and ECL2 (dark blue), with potential CBX-interacting residues shown in orange. (c) Quantification of whole-cell currents from hPanx1 mutants when treated with CBX (100 μM). Mutants are numbered according to the hPanx1 sequence while the mutants in parenthesis are the corresponding residues in frPanx1. Recordings were performed by stepping to +100 mV in the absence or presence of CBX, and each point represents the normalized current amplitude during the CBX application. Bars represent the mean value from each mutant. Asterisks indicate significance of p<0.05 determined by one-way ANOVA followed by Dunnett’s test comparing WT to each mutant (F262C: p=0.0007; I247C: p=0.0471; V258C: p=0.0363).

Given that this hydrophobic groove is formed also by residues in ECL2, we wondered if residues in ECL2 might also play a role in CBX-mediated inhibition. We mutated selected residues in ECL2 of hPanx1 to cysteines and measured channel activity before and after CBX application. We found that mutations at Ile247, Val258, and Phe262 (hPanx1 numbering) diminished CBX-sensitivity (Figure 4c). These data suggest that both ECL1 and ECL2 play important roles in inhibition of Panx1 by CBX. Although we do not have a cryo-EM structure complexed to CBX at this point, we speculate that CBX inhibits Panx channels by binding between ECL1 and ECL2 and ‘locking’ the conformation of gate forming ECL1 in favor of channel closure.

Discussion

The frPanx1-ΔLC structure uncovered a unique heptameric assembly of a large-pore channel that harbors an extracellular constriction formed by Trp74 and Arg75. These residues are located on ECL1 and face toward the central pore of the channel and thus, are situated to regulate channel function. Mutagenesis studies at these positions revealed that both residues play pivotal roles in ion selection. Unlike the LRRC8A anion channel, however, the positively charged Arg75 does not seem to form a canonical selectivity filter. Instead, the guanidino group of Arg75 likely mediates a cation-π interaction with Trp74 in the neighboring subunit, which seems to control ion selection. One possible ion selection mechanism is that this cation-π interaction stabilizes the inter-subunit interactions, which in turn creates an electrostatic environment that favors anion permeation. Another possibility is that tight inter-subunit interactions in the extracellular domain are necessary to form an ion selectivity filter in the missing region in our current model (e.g. N-terminus or C-terminal domain).

Which functional state does our model represent? Based on the lack of channel activity at 0 mV (Figure 1—figure supplement 2e and f), our current structure may represent a closed conformation. This is supported by the existence of a highly acidic region near Trp74 (Figure 3—figure supplement 2), which may serve as a barrier for anions to permeate. However, given that the narrowest constriction at Trp74 is ~10 Å wide, it is possible that the structure actually represents an open conformation. Indeed, the +GS version of frPanx1-ΔLC shows larger leak currents (Figure 1a and b), suggesting that the C-terminal truncation may promote channel opening while lack of the N-terminal modification renders it closed. If the conformation of the N-terminus in frPanx1-ΔLC is somehow compromised during purification or reconstitution into nanodiscs, it is possible that our structure may actually look closer to the +GS version. While further studies are necessary to define the functional state of our current structure, the weak EM density in the N-terminal region leaves the possibility that frPanx1-ΔLC may be representing an open state.

We found that ECL1 and ECL2 interact to each other and form a potential CBX binding pocket. Both ECL1 and ECL2 may undergo movement based on conformational alterations of the TMDs and cytoplasmic domains. For example, it is conceivable that movement of the TMDs caused by membrane stretch or voltage, or changes in the cytoplasmic domain triggered by caspase cleavage may be coupled to conformational rearrangements in the extracellular domain. The major role of the extracellular domain in pannexin function is strongly supported by our experimental results demonstrating that mutating Trp74 and Arg75, as well as surrounding residues in ECL1 and ECL2, alter channel properties including ion selectivity. Furthermore, we previously demonstrated that application of CBX to mutants at Trp74 (e.g. to Ala, Ile, Lys) potentiates voltage-dependent channel activity (Michalski and Kawate, 2016), which indicates that CBX likely acts as an allosteric inhibitor rather than a channel blocker.

In contrast to the extracellular domain, roles of the intracellular domain remain elusive. While the C-terminal domain has been demonstrated to play important roles in Panx1 channel gating (Sandilos et al., 2012), our study neither confirms or refutes this mechanism as half of this domain is missing in our current structure. Likewise, the first 10 residues in the N-terminus are disordered, making it challenging to understand how these residues tune the activity of Panx1 channel (Michalski et al., 2018). Given their important roles in channel gating, it is possible that the unmodeled N-terminal region may interact with the deleted region of the C-terminal domain. It is also possible that these domains may form a channel gate. In contrast to these domains, the deleted residues in ICL-1 (between Gly171 and Lys194) seems to play a minimal role in channel gating. We surveyed 23 different deletion constructs (in which each variant harbored a different deletion length and position) and among these, all deletions constructs showed voltage-dependent channel activity via whole-cell patch clamp, with the exception of a construct in which the entire region between Lys155 and Lys194 was removed. We also tested these deletion constructs using FSEC and found that all functional constructs were properly assembled into heptamers. The above evidence indicates that the deleted region in ILC-1 plays an insignificant role in channel gating. The EM density in this region was weak and could not be modeled, indicating a high degree of conformational flexibility.

In conclusion, our frPanx1-ΔLC structure provides an important atomic blueprint for dissecting functional mechanisms of Panx1. While we did not observe a gate-like structure in the current cryo-EM map, the missing domains, especially the N-terminal loop and the C-terminal domain, may serve as a channel gate on the intracellular side of the channel. Further structure-based experiments such as cysteine accessibility and molecular dynamics simulations will facilitate our understanding of how this unique large-pore channel functions.

Materials and methods

Key resources table

Reagent type (species) or resource	Designation	Source or reference	Identifiers	Additional information
Gene (Xenopus tropicalis)	frPanx1	Synthesized by Genscript	NCBI Reference Sequence: NP_001123728.1	Frog pannexin-1 gene sequence
Gene (Homo sapiens)	hPanx1	Synthesized by Genscript	NCBI Reference Sequence: NP_056183.2	Human pannexin-1 gene sequence
Cell line (Homo sapiens)	HEK293T cells	ATCC	Cat#: CRL-3216, RRID: CVCL_0045
Cell line (Spodoptera frugiperda)	Sf9 cells	ATCC	Cat#: CRL-1711, RRID: CVCL_0549
Recombinant DNA reagent	pIE2 hPanx1	doi: 10.1085/jgp.201711804		Mammalian expression vector for electrophysiology presented in Figure 1—figure supplement 1 and 2
Recombinant DNA reagent	pIE2 hPanx1 +GS	doi: 10.1085/jgp.201711804		Mammalian expression vector for electrophysiology presented in Figures 1, 3 and 4
Recombinant DNA reagent	pIE2 frPanx1	This paper		Mammalian expression vector for electrophysiology presented in Figure 1—figure supplement 1 and 2
Recombinant DNA reagent	pIE2 frPanx1 +GS	This paper		Mammalian expression vector for electrophysiology presented in Figure 1
Recombinant DNA reagent	pIE2 frPanx1-ΔLC	This paper		Mammalian expression vector for electrophysiology presented in Figure 1—figure supplement 1 and 2
Recombinant DNA reagent	pIE2 frPanx1-ΔLC +GS	This paper		Mammalian expression vector for electrophysiology presented in Figure 1
Recombinant DNA reagent	pC-NG-FB7 frPanx1-ΔLC	This paper		Insect cell/baculovirus expression construct
Recombinant DNA reagent	pC-NG-FB7 frPanx1	This paper		Insect cell/baculovirus expression construct
Recombinant DNA reagent	pC-NG-FB7 hPanx1	This paper		Insect cell/baculovirus expression construct
Peptide, recombinant protein	MSP2N2	doi: 10.1016/S0076-6879(09)64011–8		nanodisc expression construct
Commercial assay or kit	Fugene 6	Promega	Cat#: E2691
Chemical compound, drug	Carbenoxolone	Sigma	Cat#: C4790
Chemical compound, drug	C12E8	Anatrace	Cat#: APO128
Chemical compound, drug	DDM	Anatrace	Cat#: D310
Chemical compound, drug	Soybean polar lipid extract	Avanti	Cat#: 541602
Software, algorithm	cisTEM	doi: 10.7554/eLife.35383	RRID: SCR_016502
Software, algorithm	Warp	doi: 10.1038/s41592-019-0580-y
Software, algorithm	Coot	doi: 10.1107/S0907444904019158	RRID: SCR_014222
Software, algorithm	PHENIX	doi: 10.1107/S09074449052925	RRID: SCR_014224
Software, algorithm	Axon pClamp 10.5	Axon (Molecular Devices)	RRID: SCR_011323

Cell line generation

HEK293 (CRL-1573) cell lines were purchased from the American Type Culture Collection (ATCC, Manassas, VA), and therefore were not further authenticated. The mycoplasma contamination test was confirmed to be negative at ATCC.

Purification of frPanx1-ΔLC

frPanx1 (NP_001123728.1) was synthesized (Genscript) and cloned into the BamHI/XhoI sites of pCNG-FB7 vector containing a C-terminal Strep-tag II (WSHPQFEK). Amino acids from the IL1 and IL2 were removed by standard PCR strategies, and the BamHI site was also removed by quickchange mutagenesis. The full length frPanx1 and hPanx1 (NP_056183.2; synthesized by Genscript) were also subcloned into pCNG-FB7 vectors by standard PCR. Sf9 cells were infected with high titer baculovirus (20–25 mL P2 virus/L cells) at a cell density of 2.5–3.0 × 10⁶ cells/ mL and cultured at 27°C for 48 hr. Cells were collected by centrifugation, washed once with PBS, and lysed by nitrogen cavitation (4635 cell disruption vessel; Parr Instruments) at 600 psi in PBS containing leupeptin (0.5 μg/mL), aprotinin (2 μg/mL), pepstatin A (0.5 μg/mL), and phenylmethylsulfonyl fluoride (0.5 mM). Broken cells were centrifuged at 12,000 x g for 10 min, and membranes were collected by ultracentrifugation at 185,000 x g for 40 min. Membranes were suspended and solubilized in PBS containing 1% C12E8 (Anatrace) for 40 min, followed by ultracentrifugation at 185,000 x g for 40 min. Solubilized material was incubated with StrepTactin Sepharose High-Performance resin (GE Healthcare) for 40 min in batch. Resin was collected onto a gravity column (Bio-Rad), washed with 10 column volumes of wash buffer (150 mM NaCl, 100 mM Tris-HCl pH 8.0, 1 mM EDTA, 0.5 mM C12E8), and eluted with five column volumes of wash buffer supplemented with 2.5 mM desthiobiotin. Eluted protein was concentrated and further purified on a Superose 6 10/300 Increase column (GE Healthcare) with 150 mM NaCl, 10 mM Tris pH 8.0, 0.5 mM DDM as the running buffer. Peak fractions were collected and pooled. All steps were performed at 4°C or on ice.

Reconstitution into nanodiscs

MSP2N2 apolipoprotein was expressed and purified as described previously (Ritchie et al., 2009), and the N-terminal His tag was cleaved off using TEV protease prior to use. To incorporate frPanx1 into nanodiscs, soybean polar extract, MSP2N2 and frPanx were mixed at final concentrations of 0.75, 0.3 and 0.3 mg/ml, respectively. The mixture was incubated end-over-end for 1 hr at 4°C, followed by detergent removal by SM2 Bio-Beads (Bio-Rad). The supernatant and wash fractions were collected after an overnight incubation (~12 hr) and further purified by size exclusion chromatography using a Superose 6 10/300 column in 20 mM Tris-HCl pH 8.0, 150 mM NaCl, 1 mM EDTA. Peak fractions were pooled and concentrated to 3 mg/mL.

Cryo-EM sample preparation and image collection

frPanx1 in nanodiscs or hPanx1 in n-Dodecyl-β-D-Maltopyranoside (DDM; Anatrace) were applied to glow-discharged lacey carbon-coated copper grids (Electron Microscopy Services). The grids were blotted for 4 s with blot force 7 at 85% humidity at 15°C, and plunge frozen into liquid ethane using a Vitrobot Mark IV (Thermo Fisher). All data were collected on a FEI Titan Krios (Thermo Fisher) operated at an acceleration voltage of 300 keV. For frPanx1-ΔLC, a total of 2034 images were collected at 130 k magnification with a pixel size of 1.07 Å in electron counting mode. Each micrograph was composed of 32 frames collected over 4 s at a dose of 1.79 e / Å²/frame and a total exposure per micrograph of 57.3 e / Å². Data were collected using EPU software (FEI). For full-length frPanx1 in nanodiscs, a total of 574 images were collected at 130 k magnification with a pixel size of 1.06 Å in electron counting mode. Each micrograph was composed of 50 frames collected over 10 s at a dose of 1.4 e / Å²/frame. The total exposure per micrograph was 70 e / Å². Data were collected using SerialEM (Schorb et al., 2019). Data for full-length hPanx1 in DDM were collected in a similar fashion.

Cryo-EM image processing and single particle analysis

Warp was used for aligning movies, estimating the CTF and particle picking for frPanx1-ΔLC and full-length hPanx1. For full-length frPanx1, movie alignment and CTF estimation were performed using the program Unblur and CTFFind, respectively, within the cisTEM package (Grant et al., 2018). 2D classification, ab-initio 3D map generation, 3D refinement, 3D classification, per particle CTF refinement and B-factor sharpening were performed using the program cisTEM (Grant et al., 2018). The single particle analysis workflow for frPanx1-ΔLC is shown in Figure 1—figure supplement 3. De novo modeling was performed manually in Coot (Emsley and Cowtan, 2004). The final model was refined against the cryo-EM map using PHENIX real space refinement with secondary structure and Ramachandran restraints (Adams et al., 2010). The FSCs were calculated by phenix.mtriage. Data collection and refinement statistics are summarized in Extended data Table 1.

Electrophysiology

HEK293 cells were plated onto 12 mm glass coverslips (VWR) in wells of a six-well plate and transfected 24 hr later with 500–800 ng plasmid DNA using FUGENE 6 (Promega) according to the manufacturer’s instructions. Recordings were performed ~16–24 hr later using borosilicate glass micropipettes (Harvard Apparatus) pulled and polished to a final resistance of 2–5 MΩ. Pipettes were backfilled with (in mM) 147 NaCl, 10 EGTA, 10 HEPES pH 7.0 with NaOH. Patches were obtained in external buffer composed of (in mM) 147 NaCl, 10 HEPES pH 7.3 with NaOH, 13 glucose, 2 KCl, 2 CaCl₂, 1 MgCl₂. A rapid solution exchange system (RSC-200; Bio-Logic) was used to perfuse cells with CBX or various salt solutions. Currents were recorded using an Axopatch 200B amplifier (Axon Instruments), filtered at 2 kHz (Frequency Devices), digitized with a Digidata 1440A (Axon Instruments) with a sampling frequency of 10 kHz, and analyzed with the pClamp 10.5 software (Axon Instruments). For voltage step recordings, Panx1-expressing cells were held at −60 mV and stepped to various voltage potentials for 1 s in 20 mV increments before returning to −60 mV. For ramp recordings, cells were held at −60 mV, and ramped between −100 mV and + 100 mV over 3 s duration.

Acknowledgements

We thank the members of the Kawate and the Furukawa lab for discussions. We also thank D Thomas and M Wang for managing the cryo-EM facility and the computing facility at Cold Spring Harbor Laboratory, respectively. This work was supported by the National Institutes of Health (GM114379 to TK; NS113632 to HF; GM008267 to KM, EK, and JK; GM008267 to KM), Robertson funds at Cold Spring Harbor Laboratory, Doug Fox Alzheimer’s fund, Austin’s purpose, and Heartfelt Wing Alzheimer’s fund (to HF). JLS is supported by the Charles H Revson Senior Fellowship in Biomedical Science.

Additional information

Competing interests

No competing interests declared

Author contributions

Data curation, Formal analysis, Validation, Investigation, Methodology, Writing - review and editing

Data curation, Software, Formal analysis, Investigation, Visualization, Methodology, Writing - review and editing

Data curation, Formal analysis, Investigation, Writing - review and editing

Data curation, Formal analysis, Investigation, Methodology

Resources, Data curation, Software, Formal analysis, Supervision, Validation, Investigation, Methodology, Writing - review and editing

Conceptualization, Resources, Data curation, Formal analysis, Supervision, Funding acquisition, Validation, Investigation, Visualization, Methodology, Writing - original draft, Project administration, Writing - review and editing

Additional files

Transparent reporting form

Data availability

Cryo EM data and the pannexin model has been deposited in PDB under the accession code 6VD7.

The following dataset was generated:

SyrjanenJLMichalskiKFurukawaHKawateT2020Cryo-EM structure of Xenopus tropicalis pannexin 1 channelRCSB Protein Data Bank6VD7

References

Adams

Afonine

Bunkóczi

Chen

Davis

Echols

Headd

Hung

Kapral

Grosse-Kunstleve

McCoy

Moriarty

Oeffner

Read

Richardson

Terwilliger

Zwart

2010

PHENIX: a comprehensive Python-based system for macromolecular structure solution

Acta Crystallographica Section D Biological Crystallography66213221

10.1107/S0907444909052925

20124702

Baker

Sept

Joseph

Holst

McCammon

2001

Electrostatics of nanosystems: application to microtubules and the ribosome

PNAS981003710041

10.1073/pnas.181342398

11517324

Bao

Locovei

Dahl

2004

Pannexin membrane channels are mechanosensitive conduits for ATP

FEBS Letters5726568

10.1016/j.febslet.2004.07.009

15304325

Billaud

Lohman

Straub

Looft-Wilson

Johnstone

Araj

Best

Chekeni

Ravichandran

Penuela

Laird

Isakson

2011

Pannexin1 regulates α1-adrenergic receptor- mediated vasoconstriction

Circulation Research1098085

10.1161/CIRCRESAHA.110.237594

21546608

Billaud

Chiu

Lohman

Parpaite

Butcher

Mutchler

DeLalio

Artamonov

Sandilos

Best

Somlyo

Thompson

Ravichandran

Bayliss

Isakson

2015

A molecular signature in the pannexin1 intracellular loop confers channel activation by the α1 adrenoreceptor in smooth muscle cells

Science Signaling8

ra17

10.1126/scisignal.2005824

25690012

Boassa

Ambrosi

Qiu

Dahl

Gaietta

Sosinsky

2007

Pannexin1 channels contain a glycosylation site that targets the hexamer to the plasma membrane

Journal of Biological Chemistry2823173331743

10.1074/jbc.M702422200

17715132

Bravo

Ibarra

Retamal

Pelissier

Laurido

Hernandez

Constandil

2014

Pannexin 1: a novel participant in neuropathic pain signaling in the rat spinal cord

Pain15521082115

10.1016/j.pain.2014.07.024

25102401

Bruzzone

Hormuzdi

Barbe

Herb

Monyer

2003

Pannexins, a family of gap junction proteins expressed in brain

PNAS1001364413649

10.1073/pnas.2233464100

14597722

Chekeni

Elliott

Sandilos

Walk

Kinchen

Lazarowski

Armstrong

Penuela

Laird

Salvesen

Isakson

Bayliss

Ravichandran

2010

Pannexin 1 channels mediate 'find-me' signal release and membrane permeability during apoptosis

Nature467863867

10.1038/nature09413

20944749

Chiu

Ravichandran

Bayliss

2014

Intrinsic properties and regulation of pannexin 1 channel

Channels8103109

10.4161/chan.27545

24419036

Chiu

Jin

Medina

Leonhardt

Kiessling

Bennett

Shu

Tamm

Yeager

Ravichandran

Bayliss

2017

A quantized mechanism for activation of pannexin channels

Nature Communications8

14324

10.1038/ncomms14324

28134257

Chiu

Schappe

Desai

Bayliss

2018

Revisiting multimodal activation and channel properties of pannexin 1

The Journal of General Physiology1501939

10.1085/jgp.201711888

29233884

Choi

Clemente

Sun

2019

The structures and gating mechanism of human calcium homeostasis modulator 2

Nature576163167

10.1038/s41586-019-1781-3

31776515

Dahl

2015

ATP release through pannexon channels

Philosophical Transactions of the Royal Society B: Biological Sciences370

20140191

10.1098/rstb.2014.0191

26009770

Deneka

Sawicka

Lam

AKM

Paulino

Dutzler

2018

Structure of a volume-regulated anion channel of the LRRC8 family

Nature558254259

10.1038/s41586-018-0134-y

29769723

Emsley

Cowtan

2004

Coot: model-building tools for molecular graphics

Acta Crystallographica. Section D, Biological Crystallography6021262132

10.1107/S0907444904019158

15572765

Giaume

Leybaert

Naus

Sáez

2013

Connexin and pannexin hemichannels in brain glial cells: properties, pharmacology, and roles

Frontiers in Pharmacology4

10.3389/fphar.2013.00088

23882216

Grant

Rohou

Grigorieff

2018

cisTEM, user-friendly software for single-particle image processing

eLife7

e35383

10.7554/eLife.35383

29513216

Kasuya

Nakane

Yokoyama

Jia

Inoue

Watanabe

Nakamura

Nishizawa

Kusakizako

Tsutsumi

Yanagisawa

Dohmae

Hattori

Ichijo

Yan

Kikkawa

Shirouzu

Ishitani

Nureki

2018

Cryo-EM structures of the human volume-regulated anion channel LRRC8

Nature Structural & Molecular Biology25797804

10.1038/s41594-018-0109-6

Kawate

Gouaux

2006

Fluorescence-Detection Size-Exclusion chromatography for precrystallization screening of integral membrane proteins

Structure14673681

10.1016/j.str.2006.01.013

Kefauver

Saotome

Dubin

Pallesen

Cottrell

Cahalan

Qiu

Hong

Crowley

Whitwam

Lee

Ward

Patapoutian

2018

Structure of the human volume regulated anion channel

eLife7

e38461

10.7554/eLife.38461

30095067

Kronengold

Srinivas

Verselis

2012

The N-terminal half of the connexin protein contains the core elements of the pore and voltage gates

The Journal of Membrane Biology245453463

10.1007/s00232-012-9457-z

22825713

Kucukelbir

Sigworth

Tagare

2014

Quantifying the local resolution of cryo-EM density maps

Nature Methods116365

10.1038/nmeth.2727

24213166

Kyle

Minogue

Thomas

Domowicz

DAL

Berthoud

Hanck

Beyer

2008

An intact connexin N-terminus is required for function but not gap junction formation

Journal of Cell Science12127442750

10.1242/jcs.032482

Locovei

Wang

Dahl

2006

Activation of pannexin 1 channels by ATP through P2Y receptors and by cytoplasmic calcium

FEBS Letters580239244

10.1016/j.febslet.2005.12.004

16364313

Lohman

Leskov

Butcher

Johnstone

Stokes

Begandt

DeLalio

Best

Penuela

Leitinger

Ravichandran

Stokes

Isakson

2015

Pannexin 1 channels regulate leukocyte emigration through the venous endothelium during acute inflammation

Nature Communications6

7965

10.1038/ncomms8965

Compan

Zheng

Martin

North

Verkhratsky

Surprenant

2012

Pannexin 1 forms an anion-selective channel

Pflügers Archiv - European Journal of Physiology463585592

10.1007/s00424-012-1077-z

Tanis

Taruno

Foskett

2016

Calcium homeostasis modulator (CALHM) ion channels

Pflügers Archiv - European Journal of Physiology468395403

10.1007/s00424-015-1757-6

Maeda

Nakagawa

Suga

Yamashita

Oshima

Fujiyoshi

Tsukihara

2009

Structure of the connexin 26 gap junction channel at 3.5 Å resolution

Nature458597602

10.1038/nature07869

Michalski

Henze

Nguyen

Lynch

Kawate

2018

The weak voltage dependence of pannexin 1 channels can be tuned by N-terminal modifications

The Journal of General Physiology15017581768

10.1085/jgp.201711804

30377218

Michalski

Kawate

2016

Carbenoxolone inhibits Pannexin1 channels through interactions in the first extracellular loop

The Journal of General Physiology147165174

10.1085/jgp.201511505

26755773

Mousseau

Burma

Lee

Leduc-Pessah

Kwok

CHT

Reid

O'Brien

Sagalajev

Stratton

Patrick

Stemkowski

Biernaskie

Zamponi

Salo

McDougall

Prescott

Matyas

Trang

2018

Microglial pannexin-1 channel activation is a spinal determinant of joint pain

Science Advances4

eaas9846

10.1126/sciadv.aas9846

30101191

Okada

Islam

Sabirov

2018

Molecular identities and ATP release activities of two types of Volume-Regulatory anion channels, VSOR and Maxi-Cl

Current Topics in Membranes81125176

10.1016/bs.ctm.2018.07.004

30243431

Osei-Owusu

Yang

Vitery

MDC

Qiu

2018

Molecular biology and physiology of Volume-Regulated anion channel (VRAC)

Current Topics in Membranes81177203

10.1016/bs.ctm.2018.07.005

30243432

Oshima

Matsuzawa

Murata

Tani

Fujiyoshi

2016

Hexadecameric structure of an invertebrate gap junction channel

Journal of Molecular Biology42812271236

10.1016/j.jmb.2016.02.011

26883891

Panchin

Kelmanson

Matz

Lukyanov

Usman

Lukyanov

2000

A ubiquitous family of putative gap junction molecules

Current Biology10R473R474

10.1016/S0960-9822(00)00576-5

10898987

Pelegrin

Surprenant

2006

Pannexin-1 mediates large pore formation and interleukin-1beta release by the ATP-gated P2X7 receptor

The EMBO Journal2550715082

10.1038/sj.emboj.7601378

17036048

Ritchie

Grinkova

Bayburt

Denisov

Zolnerciks

Atkins

Sligar

2009

Chapter 11 - Reconstitution of membrane proteins in phospholipid bilayer nanodiscs

Methods in Enzymology464211231

10.1016/S0076-6879(09)64011-8

19903557

Romanov

Bystrova

Rogachevskaya

Sadovnikov

Shestopalov

Kolesnikov

2012

The ATP permeability of pannexin 1 channels in a heterologous system and in mammalian taste cells is dispensable

Journal of Cell Science12555145523

10.1242/jcs.111062

22956545

Sandilos

Chiu

Chekeni

Armstrong

Walk

Ravichandran

Bayliss

2012

Pannexin 1, an ATP release channel, is activated by caspase cleavage of its pore-associated C-terminal autoinhibitory region

Journal of Biological Chemistry2871130311311

10.1074/jbc.M111.323378

22311983

Schorb

Haberbosch

Hagen

WJH

Schwab

Mastronarde

2019

Software tools for automated transmission electron microscopy

Nature Methods16471477

10.1038/s41592-019-0396-9

31086343

Schrodinger, LLC

2015

The PyMOL Molecular Graphics System

Smart

Neduvelil

Wang

Wallace

Sansom

1996

HOLE: a program for the analysis of the pore dimensions of ion channel structural models

Journal of Molecular Graphics14354360

10.1016/S0263-7855(97)00009-X

9195488

Syrjanen

Michalski

Chou

Grant

Rao

Simorowski

Tucker

Grigorieff

Furukawa

2020

Structure and assembly of calcium homeostasis modulator proteins

Nature Structural & Molecular Biology27150159

10.1038/s41594-019-0369-9

31988524

Thompson

Jackson

Olah

Rungta

Hines

Beazely

MacDonald

MacVicar

2008

Activation of pannexin-1 hemichannels augments aberrant bursting in the Hippocampus

Science32215551559

10.1126/science.1165209

19056988

Wang

Ambrosi

Qiu

Jackson

Sosinsky

Dahl

2014

The membrane protein Pannexin1 forms two open-channel conformations depending on the mode of activation

Science Signaling7

ra69

10.1126/scisignal.2005431

25056878

Wang

Dahl

2010

SCAM analysis of Panx1 suggests a peculiar pore structure

The Journal of General Physiology136515527

10.1085/jgp.201010440

20937692

Weaver

Arandjelovic

Brown

K Mendu

S Schappe

Buckley

Chiu

Shu

Kim

Chung

Krupa

Jevtovic-Todorovic

Desai

Ravichandran

Bayliss

2017

Hematopoietic pannexin 1 function is critical for neuropathic pain

Scientific Reports7

42550

10.1038/srep42550

28195232

10.7554/eLife.54670.sa1

Decision letter

Swartz

Kenton J

Reviewing EditorNational Institute of Neurological Disorders and Stroke, National Institutes of HealthUnited States

Dutzler

Raimund

ReviewerUniversity of ZürichSwitzerlandZhou

Ming

ReviewerBaylor College of MedicineUnited States

In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.

Acceptance summary:

This is an important manuscript describing the first near atomic resolution structure of a pannexin channel, a family of large pore channels thought to be important for the movement of ions and small molecules such as ATP across cellular membranes. The mechanisms by which the activity of pannexin channels are regulated are poorly understood, yet these channels are thought to play important roles in innate immunity, neuronal signaling, apoptosis, cancer and ischemia. Having a structure of the channel will enable further mechanistic studies on how this protein plays important biological roles. By combining data from cryo-EM and electrophysiology, this manuscript describes the first structure of a pannexin channel at near-atomic resolution and it sheds light on the determinants of anion selectivity, gating and pharmacology. The data appears to be of high quality and the major conclusions of the manuscript are sound. The reviewers agreed that the work is a strong candidate for publication in eLife, and the authors have done an excellent job of revising the manuscript to address the concerns of the reviewers.

Decision letter after peer review:

Thank you for submitting your article "The Cryo-EM Structure of a Pannexin 1 Channel Reveals an Extracellular Gating Mechanism" for consideration by eLife. Your article has been reviewed by Kenton Swartz as the Senior Editor, a Reviewing Editor, and three reviewers. The following individuals involved in review of your submission have agreed to reveal their identity: Raimund Dutzler (Reviewer #2); Ming Zhou (Reviewer #3).

The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.

Summary:

This is an important manuscript describing the first near atomic resolution structure of a pannexin channel, a family of large pore channels thought to be important for the movement of ions and small molecules (e.g. ATP) across cellular membranes. The reviewers thought this was an excellent manuscript that is appropriate for publication in eLife. As is spelled out in the reviews below, all three reviewers have suggestions for improving the manuscript in revision. Most of the issues they raise can be readily addressed by careful revision of the text to openly discuss unresolved questions or to tone down some conclusions (e.g. the potential CBX binding site). We ask that you prepare a revised version that addresses the issues raised to the best of your ability.

Reviewer #1:

This manuscript presents for the first time the structure of a pannexin channel. This is a timely and very important finding for the field that will serve as a basis for understanding and studying pannexin-1 channel function.

I think it is important to note that in the current (and frenetic) race of structural biologists to determine channel structures, the main concern for physiologists is whether these structures really represent physiological conformations of the channel. I believe the authors prove, at least partially, that this is the case with the frog pannexin-1 channel. The structural interpretations that the extracellular loop 1, in particular, residues at positions W74 and R75 regulate ionic selectivity are nicely supported by their functional data. The functional data supporting inhibition by carbenoxolone at extracellular region are complementary, but less demonstrative.

The major concern with this work is with respect to some forced interpretations that do not properly address the limitations of the pannexin-1 structure and require proper clarification. This refers mainly to the lack of reliable structural data related for the cytosolic domains N-terminus (NT), intracellular loops (ICL) and C-terminus (CT), which have been shown to be crucial in gating and permeation of these channels. This needs to be properly acknowledged in the manuscript.

Another manuscript describing the pannexin-1 structures from frog and human was published a few days after the current manuscript in BioRxiv. They share similar structural results, and both lack structural information for the N-terminal and C-terminal domain.

Essential revisions:

1) As stated by the authors, the NT region has been shown to regulate gating and permeation in pannexin-1 channels. In this presented pannexin-1 structure the first 10 residues of the NT region are not resolved. The remaining residues are lining the pore as shown for other large pore channels including connexin, innexin and LRRC8 channels. Importantly, the intracellular loops in the aforementioned structures are critical for stabilizing the N-terminal region. This has been more recently demonstrated for connexin hemichannels. To solve the pannexin-1 structure, the authors had to remove 24 amino acids from the first ICL. How the deletion of this segment could alter the interactions between ICL1, ICL2, and NT, and the potential implications in channel gating or selectivity need to be discussed.

2) The structure does not contain a significant part of the C-terminal region, which has been shown to play an important role in the gating of human, rat and mouse Pannexin 1 channels. The lack of the CT in pannexin-1 leads to constitutively open channels in these species, even at negative potential. How is therefore explained that the frog pannexin-1 structure lacking the CT region corresponds to a closed channel conformation as the authors propose?

3) The authors suggest that the motif formed by the residues W74 and R75 is an extracellular gate in pannexin-1 channels. They referred to this motif in the text as a unique mechanism of gating, which gives the impression of a unique gate. If W74 is the only gate, one would predict that mutations in W74 (such as W74A) will disrupt the gate and produce a leaky channel, due to the substitution of an amino acid with a short side chain. This does not seem to be observed by the authors. If the mutations do not lead to a leak current, it is likely that there is a second gate for atomic ions.

4) Pannexin channel currents for different constructs shown in Figure 1B and Figure 1—figure supplement 2F are not normalized by channel levels. This implies that currents of frPanx1, hPanx1, and frPanx1-ΔLC are not comparable.

5) Figure 1—figure supplement 2F suggests that the channels used to determine the cryo-EM structure (i.e., those formed by frPanx1-ΔLC) are not functional. As noted, the authors needed to add a GS motif in the NT region to increase voltage-sensitivity. This reinforces the idea of a role for the NT in gating.

6) The Discussion section is incomplete. I think that many of the concerns in this review need to be addressed by the authors in this section.

7) The authors conclude that "our structural and functional study establishes the extracellular loops as the unique structural determinants for channel gating and inhibition in Panx 1…". Considering the above issues this sentence is incorrect. It must be also noted that the inhibition mechanism proposed by the authors is only ascribable to CBX. I think that modeling/docking of CBX with the pannexin-1 structure will reinforce the experimental data and provide a more solid conclusion about the mechanism of blockade by CBX.

Reviewer #2:

By combining data from cryo-EM and electrophysiology, the work describes the first structure of a Pannexin channel at high resolution and it sheds light on the determinants of anion selectivity. This is an excellent manuscript that provides important insight into an important ion channel. The data appears to be of high quality and the major conclusions of the manuscript are sound. I thus think that the work is a strong candidate for publication in eLife.

Essential revisions:

– The cryo-EM data is of high quality and it supports the structural claims made in the manuscript. The ring of aromatic residues at the extracellular constriction and the close-by arginine residues are particularly interesting, and the proposed cation-π electron interaction is supported by the data (considering the limited resolution of the map). Whereas the role of both residues for shaping the selectivity properties of the protein have been nicely demonstrated in patch clamp recordings, the evidence for the same region to act as gate is to some degree vague and speculative since it has not been demonstrated that the same residues would provide that largest energy-barrier that obstructs permeation in the closed conformation.

– Given the proposal that different region contribute to anion selectivity it would be interesting to know more about the electrostatics within the pore region.

– I wonder whether there is evidence that the described Arg residues contribute to the observed voltage dependence of currents. Although the authors have shown traces of recordings of the mutants to demonstrate shifts in the reversal potential, possible changes in the voltage dependence of gating could be better documented. I expect in their structure, the positively charged Arg would be located outside the electric field.

– I wonder whether the authors have additional evidence to exclude a potential role of CBX as pore blocker. Is the effect of CBX reversible?

– Are small currents after addition of CBX reflect incomplete occupancy of the blocker or residual conduction of the CBX-bound channel and are currents in mutants a consequence of decreased potency or increased conductance of the blocked channel (given the large amount of data presented in the study such experiments might be focus of a future study).

– In the electrophysiology section, it would be important to show some comparative data from mock-transfected cells.

– Was the change in the junction potential in asymmetric solutions in the selectivity experiments calculated to be significant and was it corrected?

Reviewer #3:

The manuscript from the Kawate lab titled "The cryo-EM structure of a pannexin 1 channel reveals an extracellular gating mechanism" reports the structure of a pannexin 1 from frog at about 3.0 Å resolution and functional studies that tested structure-based hypotheses on ion selectivity and inhibitor binding site. The work is significant in that this is the first pannexin structure, and that the structure revealed a heptameric assembly different from other large-pore forming channels such as connexins, innexins and LRRC8. The initial FSEC and excellent biochemical work led to identification of a stable construct suitable for structural studies. The procedures for cryo-EM seem straightforward, and the quality of the density map seems consistent with the claimed resolution.

I feel that the experiments are all solid and interpretations are sound. I have two comments.

First, the study on CBX is a very good follow up on a previous JGP paper by the same lab, and it provides further evidence that extracellular loop 1 harbors the inhibitor. However, short of a CBX-pannexin complex, one could argue that the inhibitor may bind elsewhere and induce a propagated effect that is mediated by the extracellular loop.

Second, I see utility of using ion permeation as a read out for testing certain aspects of pannexin, but I do not quite understand the motivation of studying ion selectivity in pannexin. If the main function of a pannexin is to form a large pore to allow release of ATP, would it be more relevant to study permeation of ATP and its gating on a proteoliposome assay? Is there any indication that anion permeation is relevant in physiology?

10.7554/eLife.54670.sa2

Author response

Summary:

This is an important manuscript describing the first near atomic resolution structure of a pannexin channel, a family of large pore channels thought to be important for the movement of ions and small molecules (e.g. ATP) across cellular membranes. The mechanisms by which the activity of pannexin channels are regulated are unknown, yet these channels are thought to play important roles in innate immunity, neuronal signaling, apoptosis, cancer and ischemia. Having a structure of the channel will enable further mechanistic studies on how this protein plays important biological roles. The reviewers thought this was an excellent manuscript that is appropriate for publication in eLife. As is spelled out in the reviews below, all three reviewers have suggestions for improving the manuscript in revision. Most of the issues they raise can be readily addressed by careful revision of the text to openly discuss unresolved questions or to tone down some conclusions (e.g. the potential CBX binding site). We ask that you prepare a revised version that addresses the issues raised to the best of your ability.

We thank the reviewers for the valuable comments. As suggested, we have toned down uncertain conclusions and added more discussion throughout the manuscript, especially on the limitations of our current study. We believe our revised and retitled manuscript addresses the concerns raised by the reviewers. Below we summarize our point-by-point answers/comments to the suggestions made by the reviewers.

Reviewer #1:

[…]

Essential revisions:

We created 23 different deletions in ICL1 between K155 and K194. Except for the construct that lacks this entire region, all other constructs showed voltage-dependent channel activity in our whole-cell patch clamp experiments. We also investigated these deletion constructs using FSEC and found that the functional constructs were properly assembled into heptamers. Based on these results, we think the deleted region in the loop has insignificant contribution in channel gating. Indeed, the EM density in this region was weak and we could not model the residues between K155 and K194. The NT region on the other hand does seem to play a crucial role in channel gating, as we previously demonstrated (Michalski et al., 2018). It is possible that the unmodeled NT region may interact with the deleted region of the CTD. We included this information in the Discussion section.

In our hands, CT truncated Panx1 (1-355 in human) does not open constitutively. This is independent of species or intracellular loop deletion. Because frPanx1-ΔLC is closed at 0 mV (Figure 1—figure supplement 2), we initially thought that the current structure represents a closed state. However, the +GS version of frPanx1-ΔLC shows larger leak currents (Figure 1), leaving a possibility that the CT truncation does promote channel opening but lack of the NT modification somehow renders it closed. If the conformation of the NT in frPanx1-ΔLC is compromised during purification or reconstitution into nanodiscs, it is possible that our structure may actually look closer to the +GS version, which may then represent an open state. We included this discussion in our revised manuscript.

We agree with the reviewer. Because we do not have sufficient data to confirm W74's role as a channel gate, we removed the phrase "extracellular gate" from our manuscript.

That is true. We do not intend to compare those constructs quantitatively, but aim to qualitatively demonstrate that all three constructs present voltage-dependent channel activity when the N-termini are modified with the addition of Gly-Ser.

We agree that the NT region plays essential roles in pannexin channel gating. Unfortunately, our current EM structure falls short of proposing a possible mechanism, as the first 10 amino acids are not modelled. As suggested by the reviewers, we expanded the Discussion section to clarify the limitation of our studies.

6) The Discussion section is incomplete. I think that many of the concerns in this review need to be addressed by the authors in this section.

Thank you for the comments. Our revised Discussion section should address the concerns raised by the reviewers.

We used "unique" in a sense that it is unusual for an ion channel to have tryptophan residues surrounding the most constricted region of the pore. However, we agree with the reviewer that the sentence is misleading. We therefore toned down its potential role as a gating machinery throughout the manuscript. We also agree that further studies are necessary to clarify the action mechanism of CBX.

Reviewer #2:

[…]

Essential revisions:

Thank you for the comments, which we agree. As described above, we have toned down the potential extracellular gating mechanism.

– Given the proposal that different region contribute to anion selectivity it would be interesting to know more about the electrostatics within the pore region.

Thank you for the suggestion. The electrostatic potential map revealed that the ion permeation pathway is mostly basic except for a highly acidic region near the constriction formed by W74. This charge distribution is consistent with an anion channel with a selectivity machinery located near W74. We now include an extra figure (Figure 3—figure supplement 2) and discussion in the text (subsection “Ion permeation pathway and selectivity”).

We did observe various degrees of voltage dependence from different R75 mutants. However, it is challenging to assess the contribution of the positively-charged side chain in voltage gating, as the interaction with W74 also seems to affect voltage dependence. One clear thing is that the positive charge at this position is not necessary for voltage-dependent channel activity, as R75A or even R75E give rise to voltage-dependent currents.

– I wonder whether the authors have additional evidence to exclude a potential role of CBX as pore blocker. Is the effect of CBX reversible?

Yes, the effect of CBX is reversible. We previously demonstrated that voltage-dependent channel activity of W74A, W74I, or W74K is potentiated by CBX (Michalski et al., 2016). In combination with our systematic mutagenesis studies, CBX is more likely to be an allosteric inhibitor than a channel blocker.

This is an interesting point. Our previous single channel recordings suggest that CBX reduces the channel open probability without changing the conductance (Michalski et al., 2018). This suggests that the residual currents after CBX application reflect incomplete occupancy. But we agree that further experiments are necessary to clarify this.

– In the electrophysiology section, it would be important to show some comparative data from mock-transfected cells.

We have already published several responses from mock-transfected cells that confirm the channel activities in our lab indeed come from Panx1 (Michalski et al., 2016 and Michalski et al., 2018).

– Was the change in the junction potential in asymmetric solutions in the selectivity experiments calculated to be significant and was it corrected?

We corrected junction potentials experimentally using an agar bridge.

Reviewer #3:

[…]

I feel that the experiments are all solid and interpretations are sound. I have two comments.

We appreciate the comments. We agree with the reviewer that CBX may bind outside of the tested residues. And yes, CBX-pannexin complex structure will be very helpful.

While ATP release has been widely studied, anion permeation through pannexin channels has also been implicated in physiology (e.g. Thompson et al., 2008). The suggested in vitro ATP-release assay would be interesting.