Mechanistic insights into phosphoactivation of SLAC1 in guard cell signaling

Significance

Environmental signals, such as darkness or drought, control the stomatal apertures in plant leaves through phosphorylations in the cytoplasmic regulatory domains (CRDs) of SLAC1 anion channels—phosphoactivation opens SLAC1 pores and consequent anion flux closes stomata. Electrophysiological analyses of structure-inspired CRD mutants of Arabidopsis SLAC1 detail a mechanism of phosphoactivation that entails the release of inhibition, establishment of a transient pores-opened active state of the trimer, and ultimate accommodation by CRD dissociation to yield a pores-closed state.

Abstract

Stomata in leaves regulate gas (carbon dioxide and water vapor) exchange and water transpiration between plants and the atmosphere. SLow Anion Channel 1 (SLAC1) mediates anion efflux from guard cells and plays a crucial role in controlling stomatal aperture. It serves as a central hub for multiple signaling pathways in response to environmental stimuli, with its activity regulated through phosphorylation via various plant protein kinases. However, the molecular mechanism underlying SLAC1 phosphoactivation has remained elusive. Through a combination of protein sequence analyses, AlphaFold-based modeling and electrophysiological studies, we unveiled that the highly conserved motifs on the N- and C-terminal segments of SLAC1 form a cytosolic regulatory domain (CRD) that interacts with the transmembrane domain(TMD), thereby maintaining the channel in an autoinhibited state. Mutations in these conserved motifs destabilize the CRD, releasing autoinhibition in SLAC1 and enabling its transition into an activated state. Our further studies demonstrated that SLAC1 activation undergoes an autoinhibition-release process and subsequent structural changes in the pore helices. These findings provide mechanistic insights into the activation mechanism of SLAC1 and shed light on understanding how SLAC1 controls stomatal closure in response to environmental stimuli.

Tight regulation of stomata is essential for plants to adapt to various environments (1, 2). The stomatal pore, formed by paired guard cells on the surface of leaves, controls the exchange of gases (carbon dioxide and water vapor) and water with the surrounding atmosphere. Guard cells can integrate a wide range of environmental signals, such as darkness, high carbon dioxide levels, ozone, low air humidity, and drought, and convert them into turgor pressure changes to regulate the stomatal aperture (3–10). The phytohormone abscisic acid (ABA) is critical in signal transduction from these stimuli (11–14). Upon entering the guard cell, ABA binds to its receptor (e.g., PYR1) (15–18). This hormone–receptor complex interacts with and inactivates the protein phosphatase PP2C (e.g., ABI1) (19–21), leading to the release of SnRK2 protein kinases (e.g., OST1) to phosphorylate downstream effectors (e.g., SLAC1). The activation of SLAC1 causes anion efflux from guard cells, resulting in membrane depolarization (22–24). Subsequently, the GORK channel is activated to mediate K⁺ efflux, driving water out of the cell, decreasing guard cell turgor, and inducing stomatal closure (25, 26).

The SLAC1 was initially identified through the genetic screens for ozone and carbon dioxide sensitivity in Arabidopsis (27, 28). The SLAC1 of Arabidopsis thaliana (AtSLAC1) is 556 amino acids in length, with a central pore-forming TMD (residues 182 to 502) flanked by N-terminal and C-terminal tails (NT and CT, respectively). The cryogenic electron microscopy (cryo-EM) structure of Brachypodium distachyon SLAC1 (BdSLAC1, ~63% sequence identity with AtSLAC1) mainly resolved the TMD portion as assembled into a symmetric trimer with each of its three pores gated by a highly conserved phenylalanine residue (F460) (29). The NT and CT extensions are disordered in this initial structure, but they are critical for channel regulation (29), and portions of them are ordered in subsequent cryo-EM structures of AtSLAC1 (30, 31). Several studies have shown that phosphorylation of SLAC1 leads to channel activation, and critical phosphorylation sites at the N terminus have been identified (29, 32–37). However, the molecular mechanism by which phosphorylation of SLAC1 leads to channel opening remains largely enigmatic.

In this study, we investigated the molecular mechanism underlying SLAC1 activation using a multidisciplinary approach that included protein sequence analysis, electrophysiology, and AlphaFold modeling. Our electrophysiological analyses on SLAC1 truncation mutants revealed that the NT and CT segments are vital for channel regulation. An intact, unphosphorylated NT is required for autoinhibition; however, whereas the entire CT is expendable for autoinhibition, the C-terminal half of CT is indispensable for maintaining an active, channel-opened state. AlphaFold-inspired mutational tests revealed that the NT and CT segments of SLAC1 interact with one another to constitute a CRD, which interacts with the TMD to regulate channel activity. Mutations in the conserved NT and CT motifs destabilize the CRD, which results in the release of autoinhibition in SLAC1, thereby allowing its transition into an activable state. Our further studies show that SLAC1 activation undergoes a complex multistep process that enables the pore-helix reorientation for channel opening.

Most of the mutational tests conducted in this study were inspired by AlphaFold models of the CRD, which was disordered in the structure of styrene-maleic acid (SMA) polymer-extracted BdSLAC1 (29). A subsequent structure of detergent-extracted AtSLAC1 included a partially ordered CRD (30), and a following AtSLAC1 structure included an additional portion of the CRD (31). Ordered portions of the CRDs in these cryo-EM structures of AtSLAC1 corroborate our AlphaFold modeling.

Sequence Analyses of SLAC1 and SLAHs

Apart from SLAC1, four SLAC1 homologs (namely, SLAH1-SLAH4) were found in A. thaliana (Fig. 1A) (27). While SLAH3 overlaps with SLAC1 in guard cells (38), it is also expressed in roots and facilitates the transport of NO₃⁻ and Cl⁻ from root to shoot (39–41). SLAH2 is mainly expressed in root stele cells and is crucial in nitrate acquisition within the vascular system (42, 43). The physiological functions of SLAH1 and SLAH4 remain unclear. To gain a deeper understanding of how SLAC1/SLAHs are represented in plants, we clustered 930 SLAC1-related sequences into one superfamily at the E ≤ 10⁻³ level and further divided them into two families (SLAC1/SLAH2/SLAH3 and SLAH1/SLAH4) at the threshold of E ≤ 10⁻¹⁴⁰. Subsequently, at the threshold of E ≤ 10⁻²⁴⁰, the SLAH2/SLAH3 subfamily was separated from the SLAC1 subfamily (Fig. 1B and SI Appendix, Table S1). The SLAH1/SLAH4 family is distinguished by the absence of the large NT and CT segments of the SLAC1/SLAH2/SLAH3 family (Fig. 1A).

Fig. 1.

Phylogenetic analysis of SLAC/SLAH family proteins. (A) Comparison of Arabidopsis SLAC1/SLAHs. SLAH1/SLAH4 only has the TMD, while SLAC1 and SLAH2/SLAH3 have both the TMD and substantial N and C termini essential for channel regulation. The conserved core regions in the SLAC1/SLAH2/SLAH3 family are highlighted in a black box. Highly conserved motifs are shown in green. (B) The family tree of SLAC/SLAH proteins. A total of 930 plant SLAC1-related sequences were clustered into one superfamily at the E ≤ 10⁻³ level and were divided into two families (SLAC1/SLAH2/SLAH3 and SLAH1/SLAH4) at the threshold of E ≤ 10⁻¹⁴⁰. At the threshold of E ≤ 10⁻²⁴⁰, the SLAH2/SLAH3 subfamily was separated from the SLAC1 subfamily. (C) The protein sequence of AtSLAC1 is shown, with superior coils defining helical segments and red triangles indicating pore-lining residues. Highly conserved motifs (motif-1 to motif-7) are indicated in boxes. The phosphorylation residues (S59, S86, S113, T114, S116, S120, and T513) are shown in orange, and the gating phenylalanine residues are highlighted in magenta (F276 and F450). The N-terminal negatively charged patches, 3DE (E164/D166/E167) and 5DE (D170/E172/D173/E179/E180), are shown in red, and the C-terminal positively charged patches, 8KR (K514/R515/K516/R519/K521/K522/K525/R526) and 5KR (K531/R532/K535/K540 /K541), are shown in blue. The residues involved in ionic interaction at the CRD–TMD interface are highlighted in green (E164/Y448, D166/R263, and R263/D507). The conserved residues mutated in the current study are highlighted in cyan.

Sequence alignments for the representative members of the SLAC1 subfamily are shown in SI Appendix, Figs. S1 and S2. These sequences feature a highly conserved core (residues 151 to 514 in AtSLAC1), encompassing the TMD region and the adjacent extensions connected to TM₁ and TM₁₀. Further analyses revealed that highly conserved motifs are presented in the cytosolic regulatory tails, including motif-1 (⁵⁴FSRQVSLETG⁶³), motif-2 (⁸⁰LPRSGRSFG⁸⁸), motif-3 (¹⁰⁵DFSMFRTKSTLSKQKS¹²⁰), motif-4 (¹⁵¹VSAGRYFAALRGPELDEV¹⁶⁸), motif-5 (¹⁷⁰DNEDILLPKEEQW¹⁸²), motif-6 (⁵⁰⁶NDLAIAITKRKLTR⁵¹⁹), and motif-7 (⁵³⁰LKRW⁵³³) (Fig. 1C). Previously identified phosphorylation sites are present in motif-1 (S59), motif-2 (S86), motif-3 (S113, T114, S116, and S120), and motif-6 (T513). Nevertheless, the functional roles of these motifs are still unsettled.

Validation of SLAC1 Phosphorylation Sites in Arabidopsis Transgenic Plants

In a previous study, we used mass spectrometry to identify 13 potential phosphorylation sites on the NT of AtSLAC1 as actual targets for the OST1 kinase, and by electrophysiology, we determined that six of them (S59, S86, S113, T114, S116, and S120) play crucial roles in mediating SLAC1 activation (29). Substitution of these critical residues with phosphomimetic aspartates in AtSLAC1-6D (S59D, S86D, S113D, T114D, S116D, and S120D) transformed this mutant into a constitutively active channel, independent of kinase phosphorylation, as evidenced by the stimulation of currents at the phosphoactivated level in Xenopus oocytes. To further evaluate their functional role in regulating stomatal closure in vivo in planta, we transformed slac1-4 plants with the mutated SLAC1 genes, including AtSLAC1-13A, AtSLAC1-13A-6S, AtSLAC1-6A, and AtSLAC1-6D (Fig. 2A), and conducted ABA-induced stomatal closure experiments on the resultant transgenic plants (Fig. 2 B and C).

Fig. 2.

Functional analyses of AtSLAC1 phosphorylation sites in mediating stomatal closure. (A) Diagram of mutant slac1-4 and potential phosphorylation sites at the NT of AtSLAC1. slac1-4 (Salk_137265) is a knockout mutant in Arabidopsis. In this mutant, the first exon of AtSLAC1 is broken via inserting a T-DNA sequence. Phosphorylation sites were identified in mass spectrometry by using SLAC1::OST1. Thirteen phosphorylated residues at the NT are indicated, six of them play key roles and are highlighted in red. (B) The phenotype analysis of ABA-induced stomatal closure in Arabidopsis wild-type, slac1-4 mutant, and the phosphorylation site mutants including slac1-4:AtSLAC1-13A, slac1-4:AtSLAC1-13A-6S, slac1-4:AtSLAC1-6A, and slac1-4: AtSLAC1-6D. The AtSLAC1-13A indicates AtSLAC1-S6A/S38A/S59A/S86A/S113A/T114A/S116A/S120A/S134A/T137A/T142A/S146A/S152A, with all 13 potential phosphorylation sites changed to alanine residues, and the AtSLAC1-13A-6S indicates AtSLAC1-13A A59S/A86S/A113S/A114T/A116S/A120S, with 6 of the 13 phosphorylation sites reverted to serine residues on the background of AtSLAC1-13A. The AtSLAC1-6A and AtSLAC1-6D indicate AtSLAC1-S59A/S86A/S113A/T114A/S116A/S120A and AtSLAC1-S59D/S86D/S113D/T114D/S116D/S120D, in which the six phosphorylation sites were replaced with alanine residues and phosphomimetic aspartates, respectively. The rosette leaves from 5-wk-old plants incubated in a stomatal-opening buffer were exposed to the high-light condition for 3 h to ensure the opening of all stomata. ABA (0 or 10 µM) was added to the samples, and stomatal closure was observed after 3 h of treatment. The representative photographs of stomata are shown. (Scale bar, 10 μm.) (C) Statistics of the stomatal aperture in B. The data were obtained from at least 20 stomata for each point. Significant differences between 0 µM (–) and 10 µM (+) ABA-treated samples were determined by multiple t tests (*P < 0.01).

As expected, wild-type Arabidopsis leaves close their stomatal pores in response to ABA treatment, while the slac1-4 mutant leaves remain insensitive to ABA and keep their stomata open. Transgenic lines of AtSLAC1-13A, where all 13 potential phosphorylation sites are replaced with alanine residues, exhibited the same phenotype as slac1-4, indicating that sites for mediating phosphorylation have indeed been removed, whereby they cannot initiate stomatal closure. The restoration of six critical residues (59S, 86S, 113S, 114T, 116S, and 120S) in AtSLAC1-13A-6S reestablished sensitivity to ABA, rendering it comparable to wild type. Consistent with the above observation, the stomata in AtSLAC1-6A, where only the same six phosphorylation sites have been deactivated by alanine replacement, remain open upon ABA application. Conversely, AtSLAC1-6D, with phosphomimetic aspartates in place of the phosphorylation sites, keeps the stomata closed, irrespective of ABA treatment. These results, consistent with the activity observed in Xenopus oocytes (29), indicate that these six critical residues alone are sufficient to mediate ABA-induced SLAC1 phosphorylation and activation for the control of stomatal closure in leaves.

Role of the SLAC1 NT in Autoinhibition and in Activation after Inhibition Release

The cytosolic NT segment of AtSLAC1 (residues 1 to 181) carries the key phosphorylation sites critical for channel activation; however, due to the lack of structural information, it is unclear how it regulates SLAC1 activity. To investigate the functional role of NT, we designed a series of N-terminally truncated AtSLAC1 constructs (Fig. 3A) and measured channel conductance by two-electrode voltage clamp (TEVC) after injecting their cRNAs into Xenopus oocytes.

Fig. 3.

The N-terminal truncations of SLAC1s lead to kinase-independent channel activation. (A) The cartoon drawing of AtSLAC1 is shown, with domain boundary, phosphorylation sites, and conserved motifs as indicated. The conserved N-terminal negatively charged patches (3DE and 5DE) and the C-terminal positively charged patches (8KR and 5KR) are marked as red star and blue star, respectively. The truncated positions are indicated below. (B and C) The N-terminal truncation analyses of AtSLAC1. Representative current traces at different voltages (from −110 mV to +70 mV in 20 mV increment) are shown in B. (D and E) The N-terminal truncation analyses of BdSLAC1. Representative current traces at different voltages (from −110 mV to +70 mV in 20 mV increment) are shown in D. Currents at –90 mV from the TEVC recordings are used to measure the channel activities. Data are mean ± SEM, n ≥ 8.

Wild-type AtSLAC1 exhibited nearly null currents, representing the basal activity for SLAC1 alone expressed in oocytes due to the absence of kinase phosphorylation in the Xenopus oocyte system. Small but slightly increased channel activity was observed among several truncated mutants, including Δ1-45, Δ1-74, and Δ1-98. Surprisingly, the truncated mutants of Δ1-123 and Δ1-157 displayed substantially enhanced currents, akin to those of the phosphomimetic mutant AtSLAC1-6D (29) (Fig. 3 B and C and SI Appendix, Fig. S3 A and B). The kinase-independent channel activity observed in TEVC recordings of these truncated mutants strongly implies that SLAC1 has been activated by the removal of autoinhibitory elements. The results unequivocally demonstrate that the elements responsible for maintaining the unphosphorylated channel in a closed state reside in the region encompassing residues G99 to F157 (bearing motif-3 and -4a). Amazingly, the further truncations of mutants Δ1-169 and Δ1-177 returned the channel activity to basal levels (Fig. 3 B and C). This finding implies that the juxtamembrane portion of NT (residues 158 to 177), comprising motif-4b and -5 in AtSLAC1, is essential for maintaining an active open channel.

To verify whether the kinase-independent activity observed in NT truncations of AtSLAC1 is a common feature among other SLAC1 members, we also designed a similar set of truncated mutants for Brachypodium BdSLAC1 (Fig. 3 D and E and SI Appendix, Fig. S3 C and D). Despite differences in anion permeability, we observed a similar pattern of kinase-independent channel activity in NT truncations of BdSLAC1. Specifically, both Δ1-126 and Δ1-166 mutants (corresponding to Δ1-123 and Δ1-157 in AtSLAC1, respectively) exhibit substantial nitrate currents. Altogether, this truncational analysis indicates that the conserved region in BdSLAC1 (residues 99 to 166) likewise serves as an autoinhibitory element in regulating channel activation. This finding may not be surprising, given that the monocotyledonous BdSLAC1 and the dicotyledonous AtSLAC1 share a high sequence identity and conserved regulatory motifs. In summary, our findings collectively indicate that the NT segments of the SLAC1 channel govern its resting state and subsequent activation processes through a common mechanism, owing to the presence of conserved phosphorylation sites and regulatory motifs within their core regions (SI Appendix, Fig. S1).

Role of the SLAC1 CT in Maintaining an Activated, Channel-Opened State

The cytosolic CT segment of SLAC1 (residues 503 to 556) contains two conserved motifs (motif-6 and -7). Unlike the NT truncations, none of the CT truncations resulted in a noticeable increase in oocyte currents (Fig. 4A and SI Appendix, Fig. S4 A and B). This implies that CT does not play a similar autoinhibitory function as does NT in regulating SLAC1 activity.

Fig. 4.

The C terminus plays an essential role in SLAC1 channel function. (A) The C-terminal truncation analyses of AtSLAC1 wild type and constitutively active mutants (Δ1-123 and AtSLAC1-6D). (B) Mutational analyses of the conserved residues of motif-7 in the constitutively active mutant AtSLAC1-6D. (C) Conservation analysis of N-terminal extension, including motif-7 and positive charged K/R residues, as indicated by red triangle. Currents at –90 mV from the TEVC recordings are used to measure the channel activities. Data are mean ± SEM, n ≥ 8.

To further examine the functional role of CT, we analyzed these truncations in the constitutively active constructs of AtSLAC1 Δ1-123 and AtSLAC1-6D. When compared to the large currents observed in Δ1-123, its CT truncation Δ1-123^Δ540-556 (residues 540 to 556 deleted) resulted in reduced currents (approximately 50%), and further truncations (Δ528-556 and Δ503-556) led to nearly null currents in oocytes (Fig. 4A). Similar results were obtained from CT truncations of the AtSLAC1-6D phosphomimetic (Fig. 4A). Quite similarly, AtSLAC1-6D^Δ540-556 truncation showed only slightly reduced currents (retaining ~95%). In comparison, Δ528-556 significantly reduced currents to about 10% of those from full-length AtSLAC1-6D, and Δ503-556 led to nearly null currents. These observations show that while the CT segment of SLAC1 is expendable for autoinhibition, the region of residues 529 to 540 (including motif-7, see Fig. 1C) within CT is indispensable for maintaining an active, channel-opened state.

Next, we dissected the conservation islands within the critical motif-7 (⁵³⁰LKRW⁵³³) in the constitutively active AtSLAC1-6D. Each individual alanine mutation caused a moderate to significant reduction of the channel currents of AtSLAC1-6D (Fig. 4 B and C and SI Appendix, Fig. S4 C and D), with the greatest impact from replacements of the bulky hydrophobic side chains in L530A and W533A (reduction to ~30%). The effects were not additive, however; thus, the combined quadruple mutation did not further reduce channel currents. These observations suggest that the AtSLAC1-6D phosphomimetic is in a metastable state, and motif-7, particularly L530 and W533, is essential for maintaining SLAC1 in a fully activated conformation once autoinhibition is relieved.

AlphaFold-Based Structural Modeling of the SLAC1 Channel

Our previous studies on plant BdSLAC1 (29) and the bacterial homolog Haemophilus influenzae TehA (HiTehA) (44) provided profound insights into SLAC1 structure and gating of SLAC1; however, the disorder of cytosolic portions of BdSLAC1 frustrates the design and interpretation of mutational studies. To address this issue, we used AlphaFold (AF)-based modeling to construct predicted models of AtSLAC1 (hereinafter referred to as the AF model, Fig. 5 A and B). The top five AF models are highly similar, with large portions at high confidence as measured by per residue predicted local-distance deviation test (pLDDT) scores (45). Apart from one loop, the entire TMD (183 to 502) has pLDDT > 70, and the AF model closely resembles the cryo-EM structure of BdSLAC1 (PDB 7en0), with a rmsd of 1.12 Å for 912 Cα atoms (SI Appendix, Fig. S5A). The predicted reliability is generally lower and more varied for the CRD; however, NT (residues 105 to 112 and 150 to 182) and CT (residues 503 to 517) have pLDDT > 50.

Fig. 5.

The AlphaFold-based modeling of AtSLAC1. (A and B) Ribbon drawings of the AF model of AtSLAC1, trimer (A) and protomer (B). The regions in the AF model with a pLDDT (predicted lDDT-Cα) score >50 are drawn (except for the region of residues 113 to 120, which contains four phosphorylation sites). The N- and C-terminal extensions (purple and cyan, respectively) establish a structural connection between motif-3 (green) and motif-4 (salmon), thus forming a CRD that directly interacts with the TMD (white). The Bottom view of the trimer is shown in A. Side views of the protomer are shown in B, with coloration based on the pLDDT score [blue (90 to 100), light blue (70 to 90), yellow (60 to 70), orange (50 to 60), and red (30 to 50); Top] and a zoomed view of the CRD in motif coloring (Bottom). (C) Conservation analysis of N- and C-terminal extension outside the TMD, with motif-5 and motif-6 indicated by underlines. Conserved D/E or R/K residues in 5DE and 8KR patches are indicated by red triangles (D) Mutational analyses of the CRD, including ¹⁷²EDI¹⁷⁴ in motif-5 and ⁵⁰⁹AIA⁵¹¹ and ⁵¹²ITK⁵¹⁴ in motif-6. The designated residues are replaced by consecutive glycine in each case. Currents at –90 mV from TEVC recordings are used to measure the channel activities. Data are mean ± SEM, n ≥ 8.

The SLAC1 TMD consists of 10-TM helices (TM_1-10) organized into five pairs of helical hairpins, with the TM_odd helices forming a channel pore gated by a highly conserved phenylalanine residue (F450). The well-predicted core of the CRD is formed by two oppositely arranged cytoplasmic extensions directly linked to the neighboring TM₁ and TM₁₀ helices (Fig. 5B). The NT portion encompasses residues 150 to 182, including motif-4 and motif-5, while the CT portion comprises residues 503 to 514, including motif-6 (Fig. 5C). The additionally well-predicted NT segment (residues 105 to 110), which forms a short helix in the AF model, corresponds to motif-3a and is bound to motif-5 of the core CRD (Fig. 5B). A stretch of phosphorylation sites (residues 113 to 120) is less reliably predicted (pLDDTs of 30 to 40) and projects laterally from the trimeric channel. This AF model was then used to design further mutational tests reported below. The cryo-EM structures of AtSLAC1, reported after we completed these tests, have corroborated the AF model. The AtSLAC1 structure of Li et al. (30) has core CRD elements ordered similarly as in the AF model, and more recent structures of Lee et al. (31) also have partially ordered CRD elements, but now also including motif-3b (SI Appendix, Fig. S5 B–D). Further comparison will be elaborated in the Discussion section. With the AF model in hand, we next dissected the roles of the conserved CRD motifs in SLAC1 activation.

The CRD Maintains the Unphosphorylated SLAC1 Channel in a Closed State

The pore-forming assembly of ten TM helices in SLAC1 has TM₁ and TM₁₀ in proximity, such that their extensions into the cytosol align in an antiparallel manner with NT motif-5 and CT motif-6 in contact (Fig. 5B). Motif-6 contains a short helix (⁵⁰⁹AIA⁵¹¹) and an extended strand (⁵¹²ITK⁵¹⁴) arranged antiparallel to another extended strand (¹⁷²EDI¹⁷⁴) from motif-5. We tested these interactions here by mutation. A previously identified phosphomimetic mutation in motif-6, T513D, converts SLAC1 into a constitutively active channel (32), and we replicated this mutation with similar results. We also observed substantial currents after replacing ⁵⁰⁹AIA⁵¹¹ of the short helix with ⁵⁰⁹GGG⁵¹¹. Disruption of the predicted β-strand interactions with consecutive glycines, giving ¹⁷²GGG¹⁷⁴ or ⁵¹²GGG⁵¹⁴, resulted in even larger channel currents, even similar to those of the constitutively active AtSLAC1-6D (Fig. 5D and SI Appendix, Fig. S6 A and B). These observations suggest that interactions between membrane-proximal NT and CT elements are important for maintaining the unphosphorylated channel in an autoinhibited state; such disruption can lead to full kinase-independent activation of SLAC1 (as for ⁵¹²GGG⁵¹⁴) or partial activation as for T513D. This is not a simple mechanism of inhibition-release, however, since the complete removal of CRD elements or even of NT alone (Δ1-177) is not at all activating (Fig. 3 B–E).

Motif-4 Serves as a Stopper to Maintain the Channel in an Autoinhibited State

The AF model and AtSLAC1 structures reveal that the motif-4 segment comprises a short helix (motif-4a) and a plug-like structure (motif-4b), which interact directly with the TMD. The plug-like structure forms two ionic pairs (E164-Y448 and D166-R263) with the TMD, while the motif-4a helix interacts with intracellular TMD loops and with other CRD elements (Fig. 6 A and B). Mutations of conserved residues of the motif-4 helix transformed the SLAC1 into a constitutively active channel, independent of kinase phosphorylation (Fig. 6C and SI Appendix, Fig. S7 A and B). Single mutants (Y156A, F157A, and L160A) produced moderate to large currents in oocytes, and their combination (Y156A/F157A and Y156A/F157A/L160A) resulted in even larger currents, comparable to the constitutively active mutants of Δ1-123 and phosphomimetic AtSLAC1-6D (Fig. 6C). Removal of the bulky hydrophobic side chains in mutants of Y156, F157, and L160 is expected to weaken the interaction of motif-4 helix with the TMD, leading to the release of autoinhibition and stimulation of channel activity in SLAC1. In support of this notion, mutations such as Y156P, A159Y, or A159R in the motif-4 helix, from which steric hindrance would be expected, also stimulate SLAC1 activity (Fig. 6C). These results suggest that the motif-4 helix in the CRD acts like a stopper, interacting with the TMD and maintaining SLAC1 in an autoinhibited resting state.

Fig. 6.

Motif-4 serves as a wedge to inhibit the channel in its closed state. (A) Close-up view of motif-4 and its interaction with TMD. Motif-4 (salmon) interacts with the TM connecting loops of TM4-TM5 and TM6-TM7 (white). (B) Conservation analysis of the motif-4 sequence, with E164, D166, and E167 within the 3DE patch indicated in the red triangles. (C–E) Mutational analyses of the conserved residues in motif-4 (C) and its potential interacting site in the TMD, the TM connecting Loop^TM4-TM5 (D) and Loop^TM6-TM7 (E). Currents at –90 mV from TEVC recordings are used to measure the channel activities. Data are mean ± SEM, n ≥ 8.

We also performed mutational tests on helix-connecting TMD loops contacted by the motif-4 helix, namely loop^TM4-TM5 and loop^TM6-TM7 (Fig. 6A). Loop^TM4-TM5 encompasses a sequence of ³²¹RRLCKVAN³²⁸ enriched with conserved positive charged R/K residues, while loop^TM6-TM7 comprises a sequence of ³⁷⁵RLPTSEALPK³⁸⁴, corresponding to a disordered region (residues 385 to 394) in the cryo-EM structure of BdSLAC1 (29). We performed deletions and mutations within these two regions, most of which failed to show significantly increased channel activity, highlighting the complexities associated with rearrangements in the pore-helices. Nevertheless, single or double mutations of C324E and K325E in loop^TM4-TM5 did result in moderate kinase-independent channel activity, possibly caused by a partial destabilization of the autoinhibited state (Fig. 6D and SI Appendix, Fig. S7 C and D). The deletion of Δ379-381, the substitution of ³⁷⁷GGSGG³⁸¹, or mutations P377A and L376A/L382A in loop^TM6-TM7 showed no significant current in TEVC measurements (Fig. 6E and SI Appendix, Fig. S7 E and F). However, mutations of P377A and L376A/L382A impaired the activity elicited in constitutively active Δ1-123. These observations indicate their critical roles in maintaining the activated state.

Motif-3 Functions as an Anchor for Autoinhibition and a Trigger for Its Release

The highly conserved motif-3 can be divided into motif-3a (residues 105 to 110) and motif-3b (residues 112 to 120). In the AF model, motif-3a interacts directly with the core CRD, while motif-3b contains a stretch of critical phosphorylation sites essential for SLAC1 activation (Fig. 7 A and B). Deleting motif-3b (Δ113-120) alone did not cause apparent channel activation, whereas deletion of motif-3a (Δ104-110) led to significantly enhanced kinase-independent activity. Furthermore, removal of the entire motif-3 (Δ103-123) resulted in even larger currents, comparable to the level observed in the constitutively active mutant Δ1-123 (Fig. 7C and SI Appendix, Fig. S8 A and B).

Fig. 7.

Motif-3 acts as a trigger for relieving autoinhibition. (A) Close-up view of motif-3 and its interaction with the N- and C-terminal conserved regions connected to the TMD. Motif-3a interacts with a binding pocket formed by the N- and C-terminal regions (purple and cyan, respectively) directly connected to TM1 and TM10 and the amphiphilic helix H2,3 connecting TM2 and TM3. (B) Conservation analysis of the motif-3 sequence, with phosphorylation sites marked as red stars. (C–E) Mutational analyses of the conserved residues in motif-3 (C), its binding sites (D), and T111 (E). (F) Close-up view of the T111 and its interaction with backbones of residues of F249-Y250-F251. Currents at –90 mV from TEVC recordings are used to measure the channel activities. Data are mean ± SEM, n ≥ 8.

From the model, two highly conserved hydrophobic residues from motif-3a, F106 and F109, mediate interactions of this element with the core CRD at its junction with the TMD (Fig. 7A), and mutational analysis supports this picture. Single mutants of F106A or F109A produced a moderate enhancement of oocyte currents, while the double mutant of F106A/F109A elicited even greater channel activities, akin to the Δ1-123 truncation (Fig. 7C). Again, by the model, the binding pocket for the motif-3a phenylalanines is alongside amphiphilic helix H_2,3 at the juncture of NT with the TMD (Fig. 7A). Alanine replacements of highly conserved residues within the pocket (L176A, P177A, W182A, L186A, and Y250A) each resulted in moderate to large currents in TEVC measurements, most notably for W182A (Fig. 7D and SI Appendix, Fig. S8 C and D). Overall, this mutational study supports the importance of motif-3a and its juxtamembrane binding site as an anchor for autoinhibition. Significantly, the 176 to 182 segment of this binding pocket is adjacent to NT segment 172 to 174 and CT segment 512 to 514, which are also critical for maintaining the autoinhibited state (Fig. 5).

Another highly conserved motif-3 residue of interest is T111, located at the boundary of the 3a and 3b submotifs (Fig. 7 A and B). Previous work showed that the T111D phosphomimetic mutation in AtSLAC1 resulted in moderate kinase-independent channel activity (29), which puzzled us since T111 was not identified as a phosphorylation site. Interestingly, the Δ1-111 and Δ111-120 mutants resulted in much greater channel activity than neighboring Δ1-109 or Δ113-120 mutants (Fig. 7E and SI Appendix, Fig. S8 E and F). Substitutions of T111 with various residues (A, S, V, and W) each resulted in moderate channel activity, similar to the T111D mutant (Fig. 7E). T111 is positioned in the AF model at the juncture of TM₂ with helix H_2,3 where its side-chain OH group hydrogen bonds with the backbone atoms of F249-Y250-F251 (Fig. 7F).

In its bipartite composition, motif-3 is critical both for the autoinhibition of channel activity and for triggering the release from this inhibition. Motif-3a participates with motif-4 in keeping the channel closed. Motif-3b is likely disordered and flexibly associated in the unphosphorylated channel (Fig. 7A). Since Δ1-123 SLAC1 (from which the entirety of motif-3 and all relevant phosphorylation sites have been deleted) is as active as fully phosphorylated SLAC1 (29) (Fig. 3C), we expect that motif-3b remains disordered in activated SLAC1. How can NT phosphorylations or the phosphomimetic substitutions of AtSLAC1-6D trigger inhibition release and activation despite such a disorder? Perhaps electrostatic repulsion from the concentration of eight carboxylates in a span of 17 residues in neighboring motif-4b and motif-5 is involved.

Tests of Charge Interactions at the CRD–TMD Interface

Based on our AF model, corroborated by AtSLAC1 structures (30, 31), there are several CRD-to-TMD contacts mediated by hydrogen bonds with charged side chains (Fig. 8A), and we have tested these residues by mutation. Mutations E164A, D166A, R263A, D507A, and R263A/D507A each resulted in significant but weak to moderate (R263A) channel currents (Fig. 8B and SI Appendix, Fig. S9 A and B). The R256A mutation gave a more substantial effect at ~45% of the fully activated AtSLAC1-6D (Fig. 8B). R256 in H_2,3 hydrogen bonds to I510 in ⁵⁰⁹AIA⁵¹¹ of motif-6, which too is activating when disrupted (Fig. 5D). We additionally tested these gain-of-function mutations in the constitutively active phosphomimetic AtSLAC1-6D. The R256A and R263A mutations had little or no effect, whereas the E164A and D507A showed significantly reduced channel activity (Fig. 8B).

Fig. 8.

The interactions at the CRD–TMD interface regulate channel function. (A) Close-up view of the interactions at the CRD–TMD interface. (B) Mutational analyses of the ionic interactions between the CRD and TMD, including E164-Y448, D166-R263, and R263-D507 pairs. (C) Mutational analyses of N-terminal negatively charged patches (3DE and 5DE, see Fig. 1C) of AtSLAC1 in the wild type and the constitutively active mutant AtSLAC1-6D. The 3DE and 5DE patches include the conserved E164/D166/E167 and D170/E172/D173/E179/E180, respectively. Mutations of 3DE->3A and 5DE->5A represent the replacements of D/E to alanine in 3DE and 5DE, respectively. (D) Mutational analyses of the C-terminal positively charged patches (8KR and 5KR, see Fig. 1C). The 8KR and 5KR patches include the conserved K514/R515/ K516/R519/K521/K522/K525/R526 and K531/R532/K535/K540/K541, respectively. Mutations of 8KR->8M and 5KR->5M represent the replacements of K/R to methionine in 8KR and 5KR, respectively. (E) Charge-swapped mutational analyses in AtSLAC1-6D. Mutation of 5DE->5K represents the replacements of D/E to lysine in 5DE. Mutations of 8KR->8E and 5KR->5E represent the replacements of K/R to glutamate in 8KR and 5KR, respectively. The combination of charge-swapped mutations includes 5K/8E (5DE->5K plus 8KR->8E) and 5K/5E (5DE->5K plus 5KR->5E). In the cartoons, the negative charges and positive charges are indicated as red and blue spheres, respectively. Currents at –90 mV from TEVC recordings are used to measure the channel activities.Data are mean ± SEM, n ≥ 8.

While this mutational analysis clearly implicates interfacial charged groups at the CRD–TMD interface in maintaining the autoinhibited state of SLAC1, none of these mutations completely eliminated channel inhibition. On the other hand, when introduced into AtSLAC1-6D, both the NT mutation E164A and the CT mutation D507A strongly reduced currents (to 23% and 25% of the activated level, respectively) from the activated channel (Fig. 8B), whereas, on their own, each relieved inhibition only slightly (to 25% and 12% of activated, respectively). Similar effects are seen from CT truncations in AtSLAC1-6D (Fig. 4A) and from mutations in loop^TM6-TM7 when added to Δ1-123 (Fig. 6E). It is evident that the CRD adopts an alternative conformation after the release of inhibition, one which then supports the TMD in an opened-channel conformation.

Tests of Conserved CRD Charged Patches in Channel Regulation

The protein sequence analysis of the SLAC1 family revealed two negatively charged (D or E) patches, 3DE (E164/D166/E167) and 5DE (D170/E172/D173/E179/E180), in the juxtamembrane portion of NT and two positively charged (R or K) patches, 8KR (K514/R515/K516/R519/K521/K522/K525/R526) and 5KR (K531/R532/K535/K540/K541), in CT (Fig. 1C). We first tested charge neutralizing mutations in each of these patches. Alanine substitutions for the acidic patches (3DE->3A and 5DE->5A) each significantly stimulated activity in wild-type SLAC1 (Fig. 8C and SI Appendix, Fig. S9 C and D), suggesting roles in the maintenance of the autoinhibited state. Substitution of these same mutations into the activated AtSLAC1-6D channel, resulted in an impaired channel (decreased to approximately 65%) for the quintuple mutant (5DE->5A) but no significant effect of the triple replacement (3DE->3A), which suggests that the 5DE patch is crucial for maintaining the channel in an opened state.

We had already shown the entire CT segment is expendable for autoinhibition of the unphosphorylated channel (Fig. 4A). In contrast, however, tests of the neutralized positive CT patches in the constitutively active Δ1-123 or AtSLAC1-6D showed strong involvement of both patches in supporting the open-channel state (Fig. 8D and SI Appendix, Fig. S9 E and F), and essentiality of the 5KR patch. These results are consistent with and extend the conclusions from CT truncations of the constitutively active channels (Fig. 4A) that CT termini are essential for the opened-channel state of SLAC1.

Given the involvement of both positively and negatively charged residues for maintaining SLAC1 in an active state, we next performed charge-swapped mutagenesis of these conserved charged patches in AtSLAC1-6D. Each of the three patches that had shown effects when neutralized in the context of AtSLAC1-6D was also tested as charge swapped (5DE->5K, 8KR->8E, and 5KR->5E). Each retained only 20 to 25% of the AtSLAC1-6D level of activity (Fig. 8E and SI Appendix, Fig. S9 G and H). Next, we checked for possible compensation from the doubly swapped, charge-reversal mutations 5K/8E (5DE->5K plus 8KR->8E) and 5K/5E (5DE->5K plus 5KR->5E) (Fig. 8E). Although 5K/8E was little changed from the individual charge swaps, the 5K/5E charge reversal showed dramatic recovery of activity, with complementation from ~20% for the individual charge swaps to ~67% of the AtSLAC1-6D level when combined.

These findings decisively implicate an electrostatic interaction of the 5KR patch (residues 531 to 541) of CT with the 5DE patch (residues 170 to 180) of NT in maintaining an active opened-channel state. Clearly, the opened-channel state does not derive from a mere release of autoinhibition; the complete removal of either NT or CT from the CRD eliminates activity (Fig. 4A), and cryo-EM structures with completely disordered CRD regions have pores completely blocked to ion permeation (29, 31). Instead, from the introduction of certain mutations into constitutively active constructs Δ-123 or AtSLAC1-6D (Figs. 4B, 6E, and 8 B–E), we find that an active, opened-pore state is established through a requisite reordering of the CRD. We presume that such reordering proceeds concomitantly with the release of inhibition upon phosphorylation, with the inhibitory CRD reinforcing a closed disposition of TM helices and the alternative CRD conformation reinforcing an opened-pore disposition of these helices.

Discussion

SLAC1, with a central TMD and a CRD composed of substantial NT and CT segments, is localized to the plasma membrane of guard cells, where it mediates anion efflux to induce stomatal closure in response to environmental stimuli (46). While it is well established that NT phosphorylation activates the channel, the molecular mechanism underlying this activation has remained undetermined. We began our research by analyzing the protein sequence and the AlphaFold predicted structure, and we used these analyses in designing mutations in the regulatory NT and CT segments, which we tested in electrophysiological experiments conducted in AtSLAC1. As we progressed with our studies and compiled data, the cryo-EM structures of Arabidopsis SLAC1 were published (30, 31). These studies validate the AlphaFold model and they assist in integrating our electrophysiological with structural studies toward elucidation of the mechanism for SLAC1 activation; however, we also note major discrepancies between our results and conclusions drawn from the most recent structures (31).

Relevance of Recent AtSLAC1 Structures Considering Functional Analyses.

Lee et al. (31) chose to study a construct of AtSLAC1 designed by Chan et al. (37) and named SLAC1-6D (T62D/S65D/S107D/S124D/S146D/S152D), but which we identify as AtSLAC1-6D′ to distinguish it from our construct named AtSLAC1-6D (S59D/S86D/S113D/T114D/S116D/S120D). The design of AtSLAC1-6D′ was based on identified sites of phosphorylation by SIF2, a kinase associated with stomatal immunity, whereas we based the AtSLAC1-6D construct on a subset of 13 identified OST1 phosphorylation sites with particular importance for channel activity (29). No sites of 6D′ are in common with those of 6D, and only two of the sites of 6D′ are even among the 13 identified OST1 sites. That AtSLAC1-6D is an authentic mimic of phosphoactivated SLAC1 is verified by our in vivo analysis showing ABA-independent stomatal closure by slac1-4:AtSLAC1-6D (Fig. 2).

We have compared the activities of AtSLAC1-6D′ with that of AtSLAC1-6D in our Xenopus oocyte assay (SI Appendix, Fig. S10), and we find that activation by 6D′ is barely noticeable compared to that by 6D, even though significant compared to that of unphosphorylated wild-type SLAC1 (i.e., no OST1). We conclude that AtSLAC1-6D′ is not an appropriate model for activated AtSLAC1. As noted by authors of both AtSLAC1 structure papers (30, 31), their preparations produced in insect cells or HEK293F cells may have been phosphorylated by endogenous kinases; however, the phosphorylation state was not evaluated in either case.

Lee et al. (31) present structures of wild-type AtSLAC1, AtSLAC1-6D′, and multiphosphomimetic mutant 8D (6D′ + S59D/S86D). They describe two conformational states, one being closed as for the previous structures of BdSLAC1 (29) and of S59A AtSLAC1 (30) and the other as being “open”. Although these “open” structures show reorganization of loops at the cytoplasmic pore openings, “open” is certainly a misnomer for the channel as a whole. As shown in Fig. 3B of ref. 31, the minimal pore radius remains essentially unchanged at 1.1 Å in this new state, far smaller than expected radii shown for a Cl⁻ ion, either when hydrated (3.3 Å) or dehydrated (1.8 Å), let alone for the physiologically important NO₃⁻ ion. We discuss the conformational states of SLAC1 below, but we conclude that these newly reported structures remain closed.

Li et al. (30) described the inactivated, closed-state structure of S59A AtSLAC1 and provided compelling evidence for the involvement of interactions between CRD and TMD residues in maintaining this closed state. Residue E164 appears to be an intracellular plug on the inner side of the ion pore, and disruptions from Y156A, E164A, and D166A of the CRD and from R256A and R263A of the TMD are strongly activating. On the other hand, the Y448A mutation at the structural partner of E164 had little effect, and other mutations well away from the plug interface, such as F106A/F109A (Fig. 7C), W182A (Fig. 7D), and ⁵¹²ITK⁵¹⁴-3G (Fig. 5D), are similarly activating. Thus, it may be that the observed activation arises from CRD rearrangement rather than from dissociation of a “plug”. Although the E164 is positioned as if plugging the pore, that pore is already blocked (notably by F450). The process of activation cannot simply be the release of inhibition since AtSLAC1 with either NT deleted (Fig. 3C) or CT deleted (Fig. 4A) and BdSLAC1 with the CRD entirely disordered (29) all remain nonconductive. Moreover, intact CT segments are essential for maintaining the activated state of Δ1-123 AtSLAC1 and AtSLAC1-6D (Figs. 4A and 8D). This alternatively ordered disposition of the CRD must be required to reinforce an opened-pore conformation, as determined by redisposition of the inner TM_odd helices, which in turn must provoke alternative orientations of the outer TM_even helices and, thereby, alter the overall character of the trimer. Pore opening through alternative rotameric states of pore-lining residues (F276 and F450) does not seem achievable.

Implications of Functional Studies for CRD Regulation of SLAC1 Activity.

Here, as in our previous studies of SLAC1 function (29), we have conducted electrophysiological analyses in Xenopus oocytes. These oocytes are ideal here since they lack kinases that can phosphorylate and activate SLAC1, such as occur in HEK293T cells and other alternative systems and can confound the analysis (30, 31). SLAC1 alone in oocytes is electrically silent, but its activity is stimulated when coexpressed or fused with an appropriate kinase. Using the SLAC1::OST1 fusion, we identified the sites of mediating phosphorylation on SLAC1 and showed that the AtSLAC1-6D phosphomimetic has activity similar to that of fully phosphorylated SLAC1.

In probing the roles of NT and CT segments of the CRD by truncation, we unexpectedly found that certain NT truncations, notably Δ1-123, also stimulated SLAC1 activity, similarly as from full phosphorylation; however, NT elimination as in Δ1-177 also eliminated activation (Fig. 3C). To rule out defective trafficking, we directly fused a green fluorescent protein (GFP) tag to Δ1-177 and observed similar surface expression as for the wild type (SI Appendix, Fig. S11 A and B; other null mutants in current work were also examined). Despite differences in anion permeability between the dicotyledon AtSLAC1 and the monocotyledon BdSLAC1, both exhibit similar “bell-shaped” patterns of kinase-independent channel activity upon progressive N-terminal truncation (Fig. 3 B–E). In contrast, truncations at the terminal third of CT (through Δ540-556) had no effect on channel activity; however, these same truncations were severely deleterious when introduced into the highly activated Δ1-123 or AtSLAC1-6D channels (Fig. 4A). These results indicate that the NT segment is crucial for inhibiting the channel activity of unphosphorylated SLAC1 and that the terminus of CT is expendable for this inhibition but indispensable for maintaining an opened-channel state.

It is inescapable from these results alone that SLAC1 activation is not merely a matter of inhibition release after phosphorylation. The CRD, both NT and CT, must interact with the TMD to generate an active state, which is in metastable and readily subject to disruption.

Further mutational tests proceeded from the conserved sequence motifs and the AlphaFold model, assuming that the AF model represents the unphosphorylated, inactive state of the channel since its pores are closed just as in BdSLAC1 and AtSLAC1 structures. We consistently observed strong gain-of-function in mutations in conserved motifs (Figs. 5D, 6C, 7 C–E, and 8B). As for the CT terminus, we also identified many mutations that either impaired or completely abolished currents from constitutively active Δ1-123 or AtSLAC1-6D when introduced into these constructs (Figs. 4B and 8B, D, E). These kinase-independent channel activities further support the notion that SLAC1 is maintained in an autoinhibited state at rest and that CRD disruptions release this autoinhibition to activate the channel into an alternatively ordered CRD conformation that establishes an opened-pore TMD.

Conformational States of SLAC1 in Light of the Structures and Electrophysiology.

How do the reported SLAC1 structures relate to the functional findings, and what conformational states are implicated by structure and function combined? What seems clear and consistent is that CRD portions ordered as in the CRD-associated AtSLAC1 structures (30, 31) and in the AF model for SLAC1 represent SLAC1 at rest as unphosphorylated. The new CRD-associated AtSLAC1 structures (PDBs 8gw6 and 8j0j) are very similar to one another (0.57Å rmsd for 1,077 trimer Cα atoms) and also to the previous AtSLAC1 structure (PDB 7wnq; 0.51 Å and 0.53 Å Cα rmsd for 1,083 and 1,098 Cα atoms, respectively) and also highly similar to the AF model.

The structure of the activated, opened-pores trimer remains unknown; however, two sources of evidence help to inform a picture of this state: 1) Structure-inspired functional analyses suggest that activation requires substantial TMD reorganization, and 2) mutational analyses on the highly activated Δ1-123 AtSLAC1 and AtSLAC1-6D proteins provide clues to the CRD conformation that maintains this activated state.

The structures of HiTehA (44) and BdSLAC1 (29) showed that their gating phenylalanine residues, F262 and F460 respectively, are in high-energy conformations that cannot be relieved without redisposition of the pore-lining, TM_odd helices; and, while the pore-opening F450A mutation of the corresponding AtSLAC1 residue gives substantial currents and phosphorylation of F450A increased these currents 1.8-fold, an intact F450 is required for the 9.3-fold increase of wild-type AtSLAC1 (29). Incidentally, rearrangement of TM_odd helices will necessarily entail rearrangement of TM_even helices, whereby the quaternary structure of the trimer will be changed as well.

Several structural inferences can be deduced from the effects of mutation on activated SLAC1 proteins. Most notably, NT residues 1 to 123 are dispensable and most likely disordered in this phosphoactivated state; NT residue E164 and CT residue D507 must make critical interactions with the TMD; the CRD must reinforce TMD Loop^TM6-TM7; juxtamembrane NT residues 170 to 180 of the 5DE patch must interact with CT residues 531 to 541 of the 5KR patch; and neighboring hydrophobic CT residues L530 and W533 must make critical interactions, perhaps with the juxtamembrane pocket occupied by F106 and F109 in the unphosphorylated, CRD-engaged state.

The CRD-stabilized, open-state SLAC1 conformation appears to be metastable and is readily destabilized, such that the CRD modules dissociate from the TMD. When this happens, the trimer returns to a closed TMD conformation that is very similar to that in the unphosphorylated, CRD-engaged state. This seems to be the case for the BdSLAC1 structure, for which we suspect the CRD was disrupted by the SMA polymer used in the extraction, whereas the detergent extractions used for the AtSLAC1 structures have preserved intact CRDs associated with closed TMD trimers. Similarly, entire deletions of either NT or CT generate closed AtSLAC1 channels. Moreover, the bacterial homolog HiTehA trimer has a default structure that is very similar to that of the TMDs in SLAC1 trimers, and we suspect that the SLAH1/SLAH4 (ΔNT/ΔCT) proteins will follow this pattern.

As noted above, despite assertions of “open” in identifying names, the recent CRD-dissociated AtSLAC1 structures (PDBs 8gw7 and 8j1e) are decidedly closed in structure and AtSLAC1-6D′ is inactive affirmed by in vivo physiology (SI Appendix, Fig. S10). These two structures are very similar (0.77 Å Cα rmsd for 933 trimer Cα atoms), and they do differ appreciably from the TMD portions of the CRD-associated AtSLAC1 structures (1.17 to 1.32 Å Cα rmsd for 841 to 871 trimer Cα atoms) and the BdSLAC1 structure (PDB 7en0, for 1.40 Å Cα rmsd for 912 trimer Cα atoms). The distinctions include differences in the “downward” disposition of TMD Y448 as hydrogen-bonded to E164 in the CRD-associated state of AtSLAC1 as compared with an “upward” rotamer in the CRD-free conformation of the analogous Y458 in BdSLAC1 structure (SI Appendix, Fig. S12). Such change may be significant since Y448 is nearby to the pore-blocking F450 residue in TM9.

The conformation of the newly described CRD-dissociated state of AtSLAC1 (PDBs 8gw7 and 8j1e) is significantly different from that of the CRD-engaged AtSLAC1 trimers (PDBs 8gw6 and 8j0j) such that the intracellular vestibule to the ion-conducting pore is widened (31). The two states coexist in cryo-EM preparations, attesting to the metastability of the CRD–TMD association, and making it difficult to attribute mechanistic significance to this state. An analogous metastability is found in ligand-gated ion channels, where desensitization can ultimately lead to pore closure despite the persistent presence of the ligand (47). The time constant for desensitization is often very short for neuronal channels; however, it could understandably be quite long for a process such as stomatal closure. Indeed, when at an imposed membrane potential like that of guard cells, currents recorded from opened SLAC1 channels decay rapidly (on the plant time scale) toward much-reduced asymptotic levels (Fig. 3 B and D).

The Activation Mechanism for SLAC1.

Based on our findings in the above structural and functional analyses, we propose a working model for SLAC1 phosphoactivation: Under basal conditions, for example, as in daytime, the SLAC1 channel is unphosphorylated and closed. This closure is maintained by autoinhibition, but not by direct blockage of the TMD ion pores; rather, the inhibiting CRD (CRD_i) conformation is one that resists the transition to an activating CRD (CRD_a) conformation. When environmental cues trigger guard cell signaling, appropriate phosphorylation ensues, which prompts reordering that generates the CRD_a conformation. CRD_i reinforces the inhibited TMD (TMD_i) conformation, in which the pores are blocked by phenylalanine side chains (and incidentally by an intracellular plug as well), and CRD_a reinforces the activated TMD (TMD_a) conformation. We imagine that the protomers of TMD_a must be changed substantially from those of TMD_i to produce open pores and that, to accommodate, the trimer must also change in its quaternary structure. The activated SLAC1 conformation is metastable, and it decays to a desensitized conformation in which the CRDs are disordered and the TMDs are closed. In the desensitized state, by equilibrium, the CT and phospho-NT segments may reorder to restore the CRD_a conformation. Then, with a certain probability that depends on the duration of phosphorylation and the existing membrane potential, the activated conformation may again be adopted. Thereby, homeostatic stabilization of stomatal closure can be maintained until basal environmental conditions are reestablished. Dephosphorylation then ensues and the inhibited state returns.

This comprehensive understanding of the molecular processes that govern the phosphoactivation of SLAC1 channels opens avenues for developing innovative strategies to engineer drought-resistant or water-saving crops, providing valuable insights into agricultural production in global climate change and growing water scarcity challenges (48).

Materials and Methods

Bioinformatics Analysis of SLAC1 Proteins.

SLAC1-related sequences were clustered into superfamily/family using position-specific iterated BLAST (PSI-BLAST) (49), and the sequence logos of conserved motifs were generated using WebLogo (50).

Stomatal Closure Assay.

For ABA-induced stomatal closure assays, we used the rosette leaves of AtSLAC1 and selected mutant variants from 5-wk-old plants and followed the protocol of Chen et al. (51).

Electrophysiology.

Channel properties were analyzed for AtSLAC1 and selected mutant variants after injection of corresponding complementary RNAs into Xenopus oocytes as previously described (29, 52).

AlphaFold-Based Modeling of SLAC1.

AlphaFold-multimer (2.1.1) was used to predict the trimeric structures of AtSLAC1, and the top-scored AF model was used for further structural analysis.

Fluorescence and Confocal Imaging.

Oocytes expressing GFP-tagged constructs of AtSLAC1 and selected mutant variants were imaged using a confocal laser microscope (Zeiss, LSM 980), and the fluorescence intensities were measured with software ImageJ.

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

Previously published data were used for this work (PDB 7en0) (29). All study data are included in the article and/or SI Appendix.