Puromycin-sensitive aminopeptidase acts as an inhibitory auxiliary subunit of volume-regulated anion channels and regulates cGAMP transport

Abstract

Volume-regulated anion channels (VRACs) are large-pore channels expressed in most vertebrate cells and are critical for cell volume regulation and autocrine/paracrine signaling. Here, we identify the ubiquitously expressed puromycin-sensitive aminopeptidase (PSA) as a binding partner of the obligatory VRAC subunit SWELL1 (also known as LRRC8A) and determine the cryo-electron microscopy structure of the SWELL1–PSA complex. Three PSA molecules bind a single SWELL1 hexamer, coupling adjacent leucine-rich repeat (LRR) domains into local dimers. Functionally, PSA overexpression suppresses VRAC activation, whereas PSA deletion dramatically elevates basal channel activity. Notably, PSA’s modulation of VRACs requires physical binding but not aminopeptidase activity, indicating a structural mechanism. Our findings identify PSA as an auxiliary subunit of VRACs, highlight the role of intracellular LRR domains in allosteric channel gating, and suggest a strategy to tune VRAC function in diverse physiological contexts, including cGAMP transport and downstream STING signaling.

Graphical Abstract

eTOC Blurb

Zheng et al. show that a ubiquitous enzyme, puromycin-sensitive aminopeptidase, functions as an inhibitory auxiliary subunit of volume-regulated anion channels. By binding the essential pore-forming subunit SWELL1, it allosterically suppresses channel activity, tuning VRAC function in diverse physiological settings.

INTRODUCTION

Volume-regulated anion channels (VRACs) are activated by cell swelling to facilitate the efflux of chloride ions and organic osmolytes, playing a central role in the cellular response to osmotic stress^1-3. Beyond their canonical function in cell volume regulation, VRACs mediate the transport of diverse signaling molecules—including 2’3’-cyclic-GMP-AMP (cGAMP), glutamate, GABA, and ATP—as well as drugs such as cisplatin^4-10. By doing so, VRACs influence a wide array of physiological and pathological processes, including immune responses^4,11, neuron-glia interactions^5,6, metabolism^12,13, and cancer⁸.

SWELL1 (LRRC8A), a member of the leucine-rich repeat-containing family 8 (LRRC8) proteins, is the main pore-forming subunit of VRACs^14,15, essential for their localization to the plasma membrane. It assembles into hetero-hexameric channel complexes with one or more closely related paralogues (LRRC8B–E)^14,16. The specific subunit composition determines the channels’ biophysical properties, including conductance and substrate selectivity^9,16. While many ion channels also stably interact with non-pore-forming auxiliary subunits to modulate their trafficking and gating properties^17-19, none have been identified for VRACs. However, the cytoplasmic leucine-rich repeat (LRR) domains of VRACs are well-established scaffolds for protein-protein interactions. Notably, several synthetic nanobodies (sybodies) have been generated to bind distinct regions of the LRR domains and modulate VRAC activity²⁰, supporting the possibility that native proteins may similarly regulate the channels through allosteric mechanisms. However, to date, no endogenous VRAC auxiliary subunits have been identified.

To investigate how VRACs are regulated under native conditions, we performed proteomics analyses and identified puromycin-sensitive aminopeptidase (PSA; gene name: NPEPPS) as a binding partner of SWELL1. Cryo-electron microscopy (cryo-EM) revealed that a single PSA molecule engages two adjacent leucine-rich repeat (LRR) domains of SWELL1, stabilizing a dimeric configuration. Electrophysiology demonstrated that PSA suppresses VRAC activity, and functional assays showed that PSA correspondingly modulates cGAMP transport. While this manuscript was in preparation, an independent study reported that PSA/NPEPPS interacts with VRACs to regulate cisplatin import and sensitivity across multiple tumor types²¹, underscoring the broader physiological relevance of this interaction. By establishing PSA as an inhibitory auxiliary subunit of VRACs, our work provides a structural and mechanistic framework that advances fundamental understanding of VRAC regulation.

RESULTS

Identification of PSA as a SWELL1-binding protein

Since SWELL1 is the principal component of VRACs, we overexpressed SWELL1-Flag in HEK293 cells and performed affinity purification using an anti-Flag antibody followed by gel electrophoresis and mass spectrometry (MS) to identify SWELL1-interacting partners. In addition to SWELL1 itself, one of the most prominent binding proteins identified was puromycin-sensitive aminopeptidase (PSA, also known as NPEPPS), a ubiquitously expressed cytosolic M1 family metallopeptidase that cleaves N-terminal amino acids from peptides (Figure 1A). Among the VRAC subunits, only LRRC8C was detected, likely due to the low abundance of native LRRC8 proteins. To further validate this interaction, we used HeLa cells with a Flag tag knocked into the endogenous SWELL1 locus (SWELL1-Flag KI)¹⁶, thereby circumventing the lack of a suitable immunoprecipitation (IP)-grade anti-SWELL1 antibody. IP followed by immunoblotting revealed that the anti-Flag antibody co-precipitated PSA in the KI cells, but not in wild-type (WT) control cells (Figure 1B). Furthermore, reciprocal IP using an anti-PSA antibody pulled down SWELL1 in WT HeLa cells (Figure 1C). These results indicate that SWELL1 and PSA interact with each other in native cells. To assess whether the SWELL1–PSA interaction is regulated by cell swelling, SWELL1-Flag KI HeLa cells were exposed to hypotonic solution and then subjected to chemical crosslinking prior to lysis to preserve protein associations. Comparable levels of PSA were co-immunoprecipitated with SWELL1 under isotonic conditions and after varying durations of hypotonic treatment (Figure 1D), indicating that their association remains stable during cell swelling-induced VRAC activation.

Figure 1. Identification of PSA as a VRAC-binding partner.

(A) Unique peptide counts of common hits (≥3 distinct peptides absent from control samples) identified in both replicates by mass spectrometry of affinity-purified samples from SWELL1-Flag overexpressing cells. Three top hits are annotated.

(B) Immunoprecipitation (IP) followed by immunoblotting showing that anti-Flag antibodies pulled down PSA in SWELL1-Flag knock-in HeLa cells, whereas WT HeLa cells served as negative controls.

(C) IP followed by immunoblotting showing that anti-PSA antibodies pulled down SWELL1 in WT HeLa cells. IgG was used as a negative control.

(D) The co-IP levels of PSA were not changed during cell swelling. SWELL1-Flag knock-in HeLa cells were treated with hypotonic solution (230 mOsm/kg) for 0, 5, or 30 min prior to cell lysis, with WT HeLa cells as negative controls. IP/input ratios were normalized to 1 in the KI cells at the 0-min time point.

(E-I) IP followed by immunoblotting with the anti-Flag antibody showing that PSA co-precipitated with SWELL1 (E), LRRC8B (F), LRRC8C (G), LRRC8D (H), and LRRC8E (I) when both were overexpressed in LRRC8^−/− HEK293 cells.

All bars represent mean ± SEM for the number of cells indicated.

The main mRNA transcript of PSA contains two AUG start codons²². Although the first AUG is often presumed to serve as the translation initiation site, previous studies based on protein sequencing and antibody recognition have demonstrated that translation instead initiates from the second AUG^23,24. Consistently, the shorter PSA cDNA construct initiating at the second AUG localized diffusely to the cytosol (Figure S1), in line with its role as a cytosolic enzyme and its ability to interact with SWELL1 at the plasma membrane. In contrast, the longer PSA isoform, which contains an additional 44 N-terminal amino acids translated from the first AUG, was predominantly mislocalized to mitochondria (Figure S1). Therefore, all subsequent experiments were performed using the shorter PSA cDNA construct initiating at the second AUG.

To exclude the possibility that the SWELL1–PSA interaction is mediated indirectly through other VRAC subunits, we co-expressed both proteins in LRRC8⁻/⁻ HEK293 cells, in which all five LRRC8 genes are disrupted¹⁴. co-IP revealed a robust interaction between SWELL1 and PSA (Figure 1E), indicating that their association occurs independently of other LRRC8 subunits. Given the homology among LRRC8 proteins, we also individually expressed LRRC8B–E together with PSA and detected varying degrees of interaction (Figure 1F-I): similar as SWELL1, LRRC8D and LRRC8E showed higher pull-down efficiency, whereas LRRC8B and LRRC8C exhibited much weaker interactions.

Cryo-EM structure of the SWELL1–PSA complex

To gain structural insight into their interaction, we first purified human PSA and confirmed its binding to detergent-solubilized human SWELL1 using fluorescence-detection size exclusion chromatography (FSEC)²⁵ (Figure S2A). We then purified SWELL1, combined it with purified PSA, and subject the mixture for single-particle cryo-EM analysis (Figure S2B and S2C). The three-dimensional (3D) structure of the SWELL1–PSA complex was reconstructed to an overall resolution of 3.19 Å with C3 symmetry imposed (Figure 2A and Figure S3A-D). 3D classification revealed that the majority of particles contained three PSA molecules associated with one hexameric SWELL1 complex, although some of PSA densities were only partially aligned, likely reflecting heterogeneous binding mode of PSA and/or flexibility of the LRR domains (Figure S3E). This heterogeneity appears to account for the low local resolution observed in the maps (Figure S4A). To improve the local resolution of the PSA-bound SWELL1 LRR domains, symmetry expansion and focused refinement were applied, yielding a 3.74 Å map (Figure S3 and S4; Table S1). The quality of the map enabled building of a structural model, with the exception of the N-terminus and two small, disordered regions within the first extracellular and intracellular loops of SWELL1 (Figure S4J-M).

Figure 2. Cryo-EM structure of the SWELL1-PSA complex.

(A) Ribbon model of the SWELL1–PSA complex, shown in side (top) and bottom (bottom) views. SWELL1 and PSA are colored magenta and blue, respectively, with the two SWELL1 subunits in each dimer distinguished by light and dark. Gray bars indicate the approximate boundaries of the lipid membrane. NTD: N-terminal domain; CD: catalytic domain; LD: linker domain; CTD: C-terminal domain.

(B and C) Binding site of PSA and the LRR domains, viewed from NTD side (B) and CTD side (C) The number of repeats contacted by PSA are labeled.

(D) Surface potentials of PSA and the LRR domains.

(E) Close-up view of the interface between NTD of PSA and LRR-r, viewed from the side (top) and from the top (bottom). LRRs contacted by PSA and key interacting amino acids are labeled.

(F) Pore radii of PSA-bound SWELL1 (purple) and apo SWELL1 channels (gray, PDB ID: 5ZSU). The radius of dehydrated chloride ion is indicated as the black dashed line.

PSA adopts a V-shaped architecture in the complex structure, with minimal conformational changes relative to the previously reported apo form²⁶ (Protein Data Bank (PDB) ID: 8SW0) (Figure 2A). It simultaneously engages the convex side of two adjacent LRR domains of SWELL1, interacting tightly with the right LRR (LRR-r) at repeats 6–9 via the N-terminal domain (NTD), burying a surface area of ~700 Ų (Figure 2B), and more loosely with the left LRR (LRR-l) at repeats 7–8 via the C-terminal domain (CTD), burying ~193 Ų (Figure 2C). 3D classification of the locally refined density maps revealed exclusively dimer classes of the LRR domains (Figure S3), suggesting that PSA binding stabilizes the LRR domains into a dimeric configuration and suppresses their flexibility^27-30. The interacting surfaces of PSA and the LRR domains exhibit complementary electrostatic features, with PSA presenting predominantly negatively charged surfaces and the LRR domains displaying positively charged regions at the binding interfaces (Figure 2D). Consistently, at the interface between the NTD of PSA and LRR-r, three acidic side chains—D55, E81, and E107—are positioned opposite two polar residues from LRR-r (N585 and H614), likely forming hydrogen bonds (Figure 2E). Notably, N585 is fully conserved among all LRRC8 paralogues, whereas H614 is not (Figure S5A). This difference may, at least in part, underlie the variable co-IP efficiencies with PSA observed among LRRC8 family members.

In the complex structure, the overall architecture of the homo-hexameric SWELL1 also closely resembles the previously reported C3-symmetric apo structures^27-31, which are thought to represent a closed channel conformation. The root-mean-square deviation (RMSD) is less than 3 Å compared to one of the SWELL1 apo structures (PDB ID: 5ZSU) determined under the same 150 mM NaCl condition²⁹ (Figure S5B). Notably, PSA binding slightly narrows the ion conduction pore at its most constricted site, formed by arginine 103 (R103), reducing the pore radius to less than that of a chloride ion (Figure 2F and S5C). Additionally, the helical regions of the intracellular loops lining the pore also shift modestly inward (Figure 2F and S5B), which may influence the propagation of conformational changes from the LRR domains to the pore region³². Nevertheless, the significance of these conformational changes in channel activation remains unclear considering previous structural studies of SWELL1 in complex with inhibitory and potentiating sybodies^2,20.

PSA broadly inhibits VRAC activation triggered by diverse stimuli

The flexibility of the LRR domains have been shown to couple to VRAC activation^2,3. Given its unique binding mode, we next investigated whether PSA modulates VRAC gating. To this end, we transfected PSA into HEK293 cells and performed whole-cell patch-clamp recordings to measure endogenous VRAC currents activated by either extracellular hypotonicity-induced cell swelling or intracellular low ionic strength. Remarkably, PSA overexpression completely suppressed VRAC currents in response to both stimuli (Figure 3A-3C). To ensure that this inhibition was not due to impaired channel trafficking, we performed surface biotinylation assays and observed comparable levels of SWELL1 at the plasma membrane in PSA-overexpressing and control cells (Figure 3A). Beyond cell swelling and low ionic strength, VRACs can also be activated—albeit to a lesser extent—by other more physiologically relevant stimuli, such as sphingosine-1-phosphate (S1P)^7,33, an important inflammatory mediator. Although the activation mechanisms likely differ³⁴, PSA overexpression similarly suppressed S1P-induced VRAC currents in mouse BV2 microglial cells (Figure 3D). Together, these findings indicate that PSA broadly inhibits VRAC activation downstream of diverse stimuli.

Figure 3. PSA broadly inhibits VRAC activation triggered by diverse stimuli.

(A) Immunoblotting confirming PSA overexpression in HEK293 cells with no change in surface SWELL1 protein levels.

(B) Time course and quantification of hypotonicity (230 mOsm/kg, applied at the arrow)-activated VRAC currents at +100 mV in HEK293 cells transfected with vector or PSA. Two-tailed t-test, p ** < 0.01.

(C) Time course and quantification of low intracellular ionic strength (50 mM)-activated VRAC currents at +100 mV in HEK293 cells transfected with vector or PSA. Two-tailed t-test, p *** < 0.001.

(D) Time course and quantification of 100 nM S1P-activated VRAC currents at +100 mV (background-subtracted) in BV2 microglial cells transfected with vector or PSA. Two-tailed t-test, p* < 0.05.

(E) Low intracellular ionic strength (50 mM)-activated VRAC currents at +100 mV in HeLa cells transfected with vector, WT PSA or various PSA mutants. One-way ANOVA and post hoc test, p* < 0.05, p ** < 0.01, p **** < 0.0001, n.s., not significant.

(F) IP followed by immunoblotting showing that unlike WT PSA, the E81A/E107A double mutant failed to interact with SWELL1 when both were overexpressed in LRRC8^−/− HEK293 cells.

All bars represent mean ± SEM for the number of cells indicated.

To determine whether the aminopeptidase activity of PSA is required for VRAC inhibition, we treated HeLa cells with tosedostat, a potent inhibitor of M1 family aminopeptidases, including PSA (Figure S6A). Tosedostat treatment had no effect on the ability of PSA overexpression to suppress VRAC currents (Figure S6B), indicating that PSA inhibits VRACs independently of its enzymatic activity. To further test this, we expressed a catalytically inactive PSA mutant in which the active-site glutamate was replaced with valine³⁵ (E309V; also denoted as E353V in some publications where the longer PSA isoform was used) (Figure S6C and S6D). Like the WT protein, the E309V mutant robustly suppressed VRAC currents activated by low intracellular ionic strength (Figure 3E), further supporting that PSA’s enzymatic activity is dispensable for its inhibitory effect on VRACs. In contrast, deletion of the PSA NTD—the primary region interacting with SWELL1—completely abolished its ability to inhibit VRAC activity (Figure 3E), highlighting the essential role of the NTD in channel regulation. Consistent with this, the NTD alone was sufficient to suppress VRAC activity (Figure 3E), although it was less potent than the WT protein. This suggests that other domains, such as the CTD, also contribute to stabilizing the SWELL1–PSA interaction and enhancing channel inhibition.

To further dissect the SWELL1–PSA interface, we individually substituted three negatively charged residues in the NTD (D55, E81, and E107) with alanine. While the D55A mutant retained its inhibitory effect, both E81A and E107A were less effective at suppressing VRAC currents (Figure 3E). Strikingly, despite being expressed at levels comparable to WT PSA (Figure S6D), the E81A/E107A double mutant completely lost the ability to inhibit VRAC activity (Figure 3E). Biochemically, it also exhibited a markedly diminished co-IP signal (Figure 3F), indicating impaired interaction with SWELL1. Taken together with the structural data, these functional results support a model in which PSA inhibits VRAC gating by interacting with and stabilizing of the SWELL1 LRR domains.

Deletion of PSA enhances basal VRAC activity

Given its ubiquitous expression, endogenous PSA may also play a regulatory role in VRAC activation. To test this hypothesis, we generated PSA knockout (KO) HeLa cells using CRISPR-Cas9 and performed whole-cell patch-clamp recordings (Figure 4A). Interestingly, PSA KO cells exhibited substantial basal currents under isotonic conditions (Figure 4B-D). These basal currents displayed hallmark features of VRACs, including outward rectification and sensitivity to dicumarol, a known VRAC inhibitor (Figure 4B, S7A and S7B)⁷. Additional deletion of SWELL1 in PSA KO cells completely abolished the basal currents (Figure 4B-D), confirming that they are indeed mediated by VRACs. Similar basal VRAC currents were also observed in PSA KO human Jurkat T cells (Figure 4E and 4F), supporting a physiological role for PSA in maintaining VRACs in a closed state under resting conditions. Notably, hypotonicity-induced cell swelling further activated VRAC currents in PSA KO cells to levels comparable to those in control cells (Figure 4D and 4F), indicating that PSA deletion does not enhance maximal VRAC activity. Moreover, knocking out PSA did not alter VRACs’ anion selectivity for iodide or glutamate as determined by reversal potential measurements (Figure S7C and S7D). These results are consistent with the model that cytosolic PSA functions as an auxiliary, but not a pore-forming, subunit of VRACs.

Figure 4. Deletion of PSA enhances basal VRAC activity.

(A) Immunoblotting of protein expression in control, SWELL1 KO, PSA KO, and SWELL1/PSA double KO (dKO) HeLa cells.

(B) Representative whole-cell currents recorded using a voltage ramp protocol for various HeLa cell lines in isotonic solution (300 mOsm/kg, left) and following perfusion of hypotonic solution (230 mOsm/kg, right).

(C and D) Time course (C) and quantification (D) of whole-cell currents at +100 mV in various HeLa cell lines in isotonic solution and following perfusion of hypotonic solution.

(E) Immunoblotting of protein expression in control, SWELL1 knockdown (KD), PSA KO, and SWELL1 KD/PSA KO (dKD/KO) Jurkat cells.

(F) Quantification of whole-cell currents at +100 mV in various Jurkat cell lines in isotonic solution and following perfusion of hypotonic solution.

All bars represent mean ± SEM for the number of cells indicated. One-way ANOVA and post hoc test, p ** < 0.01, n.s., not significant.

To quantitatively assess the effect of PSA on VRAC gating, we systematically varied intracellular ionic strength under isotonic conditions and recorded VRAC currents³⁶. As expected, PSA KO HeLa cells exhibited robust basal currents at the physiological ionic strength of 150 mM, which progressively declined at 175 mM and became undetectable at 200 mM (Figure 5A and 5B). In contrast, lowering the ionic strength to 125 mM failed to activate VRACs in control WT cells but elicited additional currents in PSA KO cells (Figure 5A and 5B). Further reduction to 100 mM initiated VRAC activation in control cells and induced even larger currents in PSA-deficient cells (Figure 5A and 5B). At ionic strengths below 75 mM, both control and KO cells displayed maximal VRAC activation; however, the kinetics of current development were markedly faster in the absence of PSA (Figure 5A-C). PSA deletion significantly increased the sensitivity of VRACs to ionic strength, reducing the half-maximal effective concentration (EC₅₀) from 113 mM in control WT cells to 89 mM in PSA KO cells (Figure 5B). Together, these data suggest that endogenous PSA binding attenuates VRAC sensitivity to intracellular ionic strength, thereby finetuning channel activity and preventing excessive activation under physiological conditions.

Figure 5. PSA reduces VRAC sensitivity to intracellular ionic strength.

(A) Time course of low intracellular ionic strength-activated VRAC currents at +100 mV in control and PSA KO HeLa cells.

(B) VRAC current-to-intracellular ionic strength relationship in control and PSA KO HeLa cells.

(C) Time constant (tau) of the rising of VRAC currents activated by various low intracellular ionic strength in control and PSA KO HeLa cells. Two-way ANOVA with post hoc test, p * < 0.05, p **** < 0.0001.

All bars represent mean ± SEM for the number of cells indicated.

PSA regulates cGAMP transport by modulating VRAC activity

In addition to chloride ions, VRACs also permeate various small signaling molecules, including the immunomodulator cGAMP^4,10, which is synthesized by the enzyme cGAS in response to cytosolic DNA. Extracellular cGAMP, released from damaged or diseased cells, can enter neighboring host cells through VRACs under physiological conditions, where it activates the innate immune STING pathway^4,37. Given PSA’s role in modulating VRAC activity, we next examined whether PSA regulates cGAMP transport and downstream signaling. To this end, we treated Jurkat cells with cGAMP and assessed STING phosphorylation (p-STING), the earliest detectable event following cGAMP import and binding, as well as the downstream phosphorylation of TBK1 (p-TBK1). As expected, knockdown of SWELL1 reduced p-STING and p-TBK1 levels (Figure 6A), consistent with VRAC’s role in cGAMP import. In contrast, loss of PSA markedly increased p-STING and p-TBK1 levels (Figure 6A), in line with the elevated basal VRAC activity observed in PSA KO cells (Figure 4E and 4F). Notably, this enhancement was completely dependent on SWELL1 (Figure 6A), confirming that it occurs through the VRAC pathway. Similar results were also observed in TIME cells (Figure 6B), a telomerase-immortalized human microvascular endothelial (HMVEC) line, and HeLa cells (Figure S8), indicating that PSA negatively regulates cGAMP transport through VRAC inhibition across diverse cell types.

Figure 6. PSA regulates cGAMP transport by modulating VRAC activity.

(A) Immunoblotting (left) and quantification (right) of STING signaling in control, SWELL1 KD, PSA KO, and SWELL1 KD/PSA KO (dKD/KO) Jurkat cells treated with cGAMP (10 μM, 1 hr).

(B) Immunoblotting (left) and quantification (right) of STING signaling in control, SWELL1 KO, PSA KO, and SWELL1 KO/PSA KO (dKO) TIME cells treated with cGAMP (20 μM, 1 hr).

(C) Immunoblotting (left) and quantification (right) of STING signaling in lentiviral vector or WT PSA-transduced HeLa cells treated with cGAMP (50 μM, 1 hr) and with or without tosdedostat (10 μM, 48 hrs).

(D) Immunoblotting (left) and quantification (right) of STING signaling in lentiviral vector, WT or the E81A/E107A mutant PSA-transduced HeLa cells treated with cGAMP (50 μM, 1 hr).

All bars represent mean ± SEM for the number of experiments indicated. p-STING/STING ratios were normalized to 1 in control cells.

To further investigate how PSA regulates cGAMP transport, we overexpressed PSA in WT HeLa cells and observed reduced p-STING levels upon cGAMP treatment (Figure 6C). Tosedostat neither enhanced cGAMP uptake in control cells nor reversed the suppression by PSA overexpression (Figure 6C), indicating that PSA’s aminopeptidase activity is dispensable. Consistently, PSA overexpression dampened cGAMP-induced STING activation in PSA KO cells, whereas the E81A/E107A mutant, defective in SWELL1 binding and VRAC inhibition, had no effect (Figure 6D). These findings demonstrate that PSA regulates cGAMP transport through interaction with SWELL1 rather than its enzymatic activity, thereby revealing a previously unrecognized role for PSA in modulating cGAMP transport and STING signaling.

DISCUSSION

Through affinity purification and mass spectrometry, we have identified the ubiquitously expressed puromycin-sensitive aminopeptidase (PSA) as an auxiliary subunit that allosterically regulates VRAC activation. PSA binds to the convex surface of the SWELL1 LRR domains, reminiscent of the binding mode of the previously reported inhibitory sybodies²⁰. Unlike these sybodies, which bind individual domains, the V-shaped PSA simultaneously engages two adjacent domains—an elegant arrangement that likely enhances its ability to stabilize the dimeric conformation. Multiple lines of evidence support a link between LRR domain conformation and VRAC gating: sybodies targeting the concave surface of the LRR domains increase their mobility and potentiate channel activity²⁰; bulky fluorescent proteins fused to the C-termini of LRRC8 proteins drive constitutive channel activity^38,39; and LRR mobility has been observed during hypotonicity-induced cell swelling³⁹. Our finding that PSA inhibits VRAC activity thus reinforces the model that conformational dynamics of the LRR domains are tightly coupled to channel gating^2,3.

Although SWELL1 homo-hexameric channel exhibits poor activation properties and may not form under physiological conditions²⁷, its structure is more homogenous than those of SWELL1/LRRC8 heteromeric channels, making it more amenable to structural interpretation of PSA binding. Importantly, the trimer-of-dimers architecture observed in a subset of SWELL1 homomers appears to be conserved in certain SWELL1/LRRC8C and SWELL1/LRRC8D heteromers, where the predominant configuration consists of two SWELL1 dimers and one LRRC8 dimer^40-42. While the binding affinities of PSA to different LRRC8 subunits remain to be fully determined, and whether and how it associates with these subunits in endogenous heteromeric channels is still unclear, the binding mode observed in SWELL1 homomers is likely conserved in heteromeric assemblies. Beyond allosterically modulating VRAC activity, PSA may also facilitate channel assembly by stabilizing LRRC8 dimers as discrete structural units during biogenesis. These possibilities will be addressed in future structural and functional studies, for which the present work provides an important foundation.

Our results support a model in which some pore-forming subunits of VRACs (SWELL1 and/or other LRRC8 proteins) are bound and stabilized by endogenous PSA in native cells (Figure 7). This interaction reduces the sensitivity of the channels to ionic strength and likely other stimuli, thereby preventing excessive basal VRAC activity under physiological conditions. When PSA is overexpressed or upregulated, it saturates all available binding sites, maximally stabilization of the LRR domains and rendering the channels largely unresponsive to diverse stimuli. An outstanding question is whether PSA dissociates from VRACs during activation as part of the gating mechanism. Our data do not support such a regulatory mechanism: the SWELL1–PSA interaction was not detectably altered by cell swelling in co-IP assays. Furthermore, VRAC currents are readily reversible upon transitions between isotonic and hypotonic solutions, consistent with gating that does not require complete PSA dissociation. Finally, because the PSA–LRR interface is primarily electrostatic, it would be expected to strengthen, rather than weaken, under the lower ionic strength accompanying swelling. That said, we cannot rule out the possibility that PSA binding undergoes subtle changes that contribute to channel activation.

Figure 7. Model for allosteric modulation of VRACs by PSA.

Schematic illustrating how PSA binding to the intracellular LRR domains of SWELL1/LRRC8 stabilizes a dimeric configuration, restricting their mobility and thereby inhibiting VRAC activation. This allosteric regulation reduces the channel’s sensitivity to various stimuli, making VRACs less active under physiological conditions (middle). Deletion of PSA increases LRR domain flexible and elevates basal VRAC activity (left), whereas PSA overexpression fully stabilizes the LRR domains and suppresses channel activation upon stimulation (right).

Our results indicate that PSA acts as a ubiquitous regulator of VRAC activity, in contrast to typical auxiliary subunits, which are expressed in a tissue-specific manner to modulate channel function in select cell types^17-19. Accordingly, beyond its canonical function in proteolysis, PSA may also participate in the diverse physiological and pathological processes regulated by VRACs. Indeed, we demonstrate that PSA modulates cGAMP import across multiple cell types. Thus, it may represent a therapeutic target for enhancing or restricting extracellular cGAMP uptake and STING signaling, with potential to boost anti-tumor immunity or mitigate pathological inflammation in a range of diseases⁴³. Interestingly, a recent study identified PSA as a driver of cisplatin resistance in human bladder cancer cells by regulating intracellular cisplatin levels⁴⁴. Depletion of PSA sensitized resistant cells to cisplatin, whereas its overexpression in sensitive cells conferred increased resistance. In follow-up work, the authors showed that PSA interacts with VRACs and regulates cisplatin import in a SWELL1-dependent manner²¹. Through pharmacological and mutagenesis approaches, they concluded that PSA’s enzymatic activity is essential for mediating cisplatin resistance and proposed its pharmacological inhibition as a potential strategy to improve patient responses to cisplatin-based chemotherapy^21,44. Our work complements their findings by providing the structural basis of the SWELL1–PSA interaction and elucidating the molecular mechanism by which PSA regulates VRAC activity. However, our study indicates that PSA’s enzymatic activity is dispensable for this regulatory function. The reason for this discrepancy remains unclear. One possibility is that tosedostat may target proteins other than PSA in bladder cancer cells, contributing to the observed cisplatin sensitization. Additionally, the catalytically inactive mutant used in their study was generated in the longer PSA isoform, which in our experiments predominantly mislocalizes to mitochondria and may therefore behave differently from the corresponding mutant expressed from the shorter isoform used in our study. Clarifying this discrepancy will be important, as our findings suggest that disrupting the interaction between SWELL1 and PSA, rather than inhibiting PSA’s enzymatic activity, may represent a more effective therapeutic strategy to enhance VRAC activity and, in turn, increase sensitivity to cisplatin-based chemotherapy or cGAMP-mediated anti-tumor immunity.

Limitations of the study

Our study establishes PSA as an inhibitory auxiliary subunit of VRACs and provides structural and mechanistic insight into its regulation. While we primarily characterized the interaction between PSA and SWELL1, the main VRAC subunit, PSA can also bind to other LRRC8 proteins. The binding affinities of PSA for these different subunits have not yet been determined, and how PSA engages with SWELL1 and specific LRRC8 subunits within endogenous heteromeric VRAC complexes remains to be clarified. Addressing these questions will be important for a more complete understanding of the regulatory mechanism. In addition, our analyses of cGAMP transport and STING signaling were carried out in cell lines, and further in vivo studies will be needed to assess the physiological and pathological relevance of PSA-mediated VRAC regulation in cGAMP transport and STING signaling.

STAR★METHODS

EXPERIMENTAL MODEL AND STUDY PARTICIPANT DETAILS

Mammalian tissue culture

HeLa cells, HEK293 cells, and BV2 microglial cells were maintained in Dulbecco’s Modified Eagle’s Medium (DMEM) supplemented with 10% fetal bovine serum (FBS) and 1% penicillin/streptomycin. SWELL1-Flag knock-in HeLa cells¹⁶, SWELL1 KO HeLa cells⁵, and SWELL1 KO BV2 cells⁷ were generated previously. LRRC8⁻/⁻ HEK293 cells were generously provided by T. Jentsch¹⁴. ExpiSf9 cells were cultured in ExpiSf CD medium (Gibco) at 27°C. Jurkat T cells were grown in Roswell Park Memorial Institute (RPMI) 1640 medium containing GlutaMAX, supplemented with 10% FBS. SWELL1 knockdown Jurkat cells were generated using lentiviruses expressing SWELL1-specific shRNA (GGUACAACCACAUCGCCUA) as previously reported¹⁵. Control and SWELL1 KO telomerase-immortalized human microvascular endothelial (TIME) cells were cultured in vascular cell basal media supplemented with a microvascular endothelial cell growth kit-VEGF (ATCC) and 1% penicillin/streptomycin. HeLa cells, HEK293 cells and ExpiSf9 cells are female. BV2 microglial cells, Jurkat T cells and TIME cells are male. All cells (except ExpiSF9 cells) were cultured at 37°C in a humidified 5% CO2 incubator.

METHOD DETAILS

Mass spectrometry

HEK293 cells were transfected with either pCMV vector or pCMV-SWELL1-Flag (OriGene, RC208632) using Lipofectamine 2000 transfection agent (Life Technologies) for 2 days prior to lysis in radioimmunoprecipitation assay (RIPA) buffer (50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 2.5% deoxycholic acid, 10% NP-40, 10 mM EDTA) containing 1% protease inhibitor cocktail (Roche). Lysates were rotated at 4°C for 30 min and clarified by centrifugation at 13,800 × g for 30 min. The resulting supernatants were incubated with anti-Flag M2 magnetic beads (Sigma) overnight at 4°C. The beads were washed four times for 5 min each and eluted with 50 μg/ml 3× Flag peptide (Sigma) in a buffer containing 20 mM Tris, pH7.5, 150 mM NaCl, 0.05% DDM and a cocktail of protease inhibitors. The affinity-purified samples were separated by SDS–PAGE gel electrophoresis and visualized by Coomassie blue staining. To increase proteomics coverage, the entire gel lane from samples purified from control and SWELL1-Flag-expressing cells was excised into 20 evenly sized slices, each subjected individually to mass spectrometry before pooling for data analysis. Gel bands were de-stained, reduced with dithiothreitol (DTT), alkylated with iodoacetamide, digested with trypsin overnight, and dried under vacuum. Dried samples were reconstituted in 0.1% formic acid and analyzed on a Q-Exactive Orbitrap Mass Spectrometer as previously reported¹⁶. Data files were imported and automatically aligned using Mascot. Peptide false discovery rate (FDR) was set to 1%, and peptide identification was performed against a human slice of the UniProt database. Proteins with more than 3 unique peptides detected in affinity-purified SWELL1-Flag samples but absent from controls were selected as positive hits.

Cloning

The coding sequence for the longer human PSA isoform (NM_006310) with Myc and Flag tags in pCMV vector was obtained from OriGene (RC209037). The coding sequence starting from the second ATG (2,625 bp) was PCR (Vazyme) amplified and sub-cloned into pIRES2-EGFP (Clontech) and pLenti-EF1a (Addgene). The PSA point mutations and truncations were introduced using Q5 Site-Directed Mutagenesis Kit (NEB). Myc-Flag-tagged cDNAs of human SWELL1 (NM_019594), LRRC8B (NM_015350), LRRC8C (NM_032270), LRRC8E (NM_025061) in pCMV vector were obtained from OriGene (RC208632, RC205553, RC222603, RC209849). The coding sequences without a tag were PCR amplified and sub-cloned into pIRES2-EGFP vector (Clontech). The cDNA of human SWELL1 were also sub-cloned into the pFastBac vector (Gibco). The cDNA of human LRRC8D (NM_018130) in pCMV6-Entry vector was obtained from OriGene (SC319580).

Immunoblotting and antibodies

Homogenates of cultured cells were prepared in RIPA buffer containing 1% protease inhibitor cocktail (Roche). Lysates were rotated at 4°C for 30 min and clarified by centrifugation at 13,800 × g for 30 min. The total protein levels were determined by BCA protein assay kit (Thermo scientific). 6 × Laemmli SDS-sample buffer (Boston bioproduct) was added into the supernatants then followed by boiling at 95°C for 10 min. Samples were resolved on 4–20% Bis-Tris SDS-PAGE gels (Life Technologies) and transferred to Immobilon-P PVDF membranes (Millipore). Membranes were blocked in Tris-buffered saline (TBS) containing 0.05% Tween-20 (TBST) and 5% fat-free milk for 1 hr at room temperature (RT) before incubation with primary antibodies. Membranes were probed overnight at 4°C with the following primary antibodies: anti-PSA (Invitrogen, #MA5-47084), anti-SWELL1⁵, anti-LRRC8B (Sigma-Aldrich, # HPA017950), anti-LRRC8C (Sigma-Aldrich, # HPA029347), anti-LRRC8D (Sigma-Aldrich, # HPA014745), anti-LRRC8E (Sigma-Aldrich, # HPA HPA020466), anti-phospho-STING (Ser366) (Cell Signaling Technology, #19781), anti-STING (Cell Signaling Technology, #13647), phospho-TBK1/NAK (Cell Signaling Technology, #5483), TBK1/NAK (Cell Signaling Technology, #3504), anti-Myc (Cell Signaling Technology, # 5605), anti-alpha-Tubulin (Cell Signaling Technology, #3873) and anti-beta-Actin (Cell Signaling Technology, #4967). After primary antibody incubation, membranes were washed by TBST for three times at RT and incubated with horseradish peroxidase (HRP)-conjugated secondary antibodies: Goat Anti-Rabbit IgG (H+L) (Jackson ImmunoResearch Laboratories, #111-035-144) or Goat Anti-Mouse IgG (H+L) (Jackson ImmunoResearch Laboratories, #115-035-166) diluted in 5% fat-free milk for 1 hr at RT. The membranes were washed by TBST for three times and 10 min each at RT. Immunoreactive bands were visualized using enhanced chemiluminescence (ECL) (Vazyme). Films were scanned with the AI-680 system (GE) and analyzed using ImageJ.

Immunoprecipitation

For IP of endogenous proteins, HeLa SWELL1-Flag knock-in cells were crosslinked with 0.5% formaldehyde in washing buffer (20 mM HEPES, pH 7.6, 10 mM KCl, 1.5 mM MgCl₂) for 10 min on ice⁵⁴. The reaction was quenched by adding one-fourth volume of 1.25 M glycine for 10 min. The cells were centrifuged at 1,000 × g for 5 min at 4°C and resuspended with washing buffer. After sonication with 10-sec pulses for 2 min on ice, one-ninth volume of 5 M NaCl was added to the extract. The lysates were centrifuged at 13,800 × g for 10 min at 4°C, and the resulting supernatants were incubated overnight at 4°C with anti-PSA polyclonal antibody (Invitrogen, #PA5-83788) with gentle mixing. For IP of endogenous SWELL1, HeLa wild-type cells were prepared in RIPA buffer containing 1% protease inhibitor cocktail (Roche). Lysates were rotated at 4°C for 30 min and clarified by centrifugation at 13,800 × g for 30 min at 4°C, and the resulting supernatants were incubated overnight at 4°C with anti-Flag monoclonal antibody (Sigma, F1804). After incubated with antibodies, the protein G magnetic beads were pre-washed with TBS buffer and were added into the supernatant-antibody mixes followed with gently rotating for 2 hr at 4°C. Precipitated proteins were washed using TBS buffer for four times. To elute the proteins from beads, 2× Laemmli SDS-sample buffer (Boston bioproduct) was added followed by boiling at 95°C for 10 min. The affinity-purified samples were analyzed via immunoblot as described above. IgG isotype was used a negative control, and whole-cell lysates served as input controls. For some experiments, cells were treated with hypotonic solution (230 mOsm kg⁻¹, as described in the electrophysiological recordings) for 0, 5, or 30 min prior to IP.

For IP of overexpressed proteins, PSA which has a Flag tag was individually co-transfected with SWELL1, LRRC8B, LRRC8C, LRRC8D and LRRC8E in LRRC8⁻/⁻ HEK293 cells using Lipofectamine 2000 transfection reagent (Life Technologies). After 2 days, cells were lysed in RIPA buffer containing 1% protease inhibitor cocktail. Lysates were clarified by centrifugation at 13,800 × g for 30 min at 4°C. IP was performed using anti-Flag-M2 magnetic beads (Sigma) with overnight incubation at 4°C under gentle mixing. Precipitated proteins were washed, eluted, and analyzed via immunoblot as described above.

Surface biotinylation

Surface biotinylation was performed as previously described⁵⁵. One day after PSA transfection, HEK293 cells were rinsed once with ice-cold PBSCM (PBS containing 0.1 mM CaCl₂, 1 mM MgCl₂, pH 8.0), then incubated with Sulfo-NHS-SS-biotin (1 mg ml⁻¹; Thermo Scientific) for 20 min at 4°C. Unreacted biotin was quenched by washing cells twice for 5 min with 20 mM glycine in PBSCM. Cells were lysed in buffer containing PBS, 50 mM NaF, 5 mM sodium pyrophosphate, 1% NP-40, 1% sodium deoxycholate, 0.02% SDS, and the protease inhibitor cocktail. Equal amounts of protein were incubated overnight at 4 °C with NeutrAvidin agarose beads (Thermo Scientific), followed by four washes with lysis buffer. Biotinylated proteins were eluted in 2× SDS loading buffer by boiling at 70 °C for 10 min, then analyzed by SDS-PAGE and immunoblot with anti-SWELL1 antibody and anti-Actin as a negative control.

Generation of PSA knockout cells

To generate PSA knockout (KO) cells, we used CRISPR-Cas9 technology⁵⁶. The guide RNA (5’-GGCCAAACTAAAAATTCTAA-3’, sense strand) targeting PSA was cloned into PX458-mCherry (Addgene) and transfected in HeLa cells or SWELL1 KO HeLa cells using Lipofectamine 2000. 2 days post-transfection, single mCherry-positive cells were FACS-sorted into 96-well plates. After 2-3 weeks, single KO colonies were isolated based on immunoblot and target-site-specific PCR followed by Sanger-sequencing. For hard-to-transfect cells (i.e., Jurkat and TIME cells), the same guide RNA was cloned into lentiCRISPR-v2-Blast vector (Addgene). Lentiviruses were produced by co-transfecting it with packaging vectors (pVSV-G, pMDL, and pRSV) in HEK293 cells following Addgene’s instruction. Jurkat and TIME cells (both control and SWELL1 knockdown or knockout) were transduced in the presence of 8 μg/mL polybrene, selected with blasticidin (2.5-10 μg/mL) 3 days post-transduction for 2-3 weeks. The panel of KO cells was validated by immunoblot.

Live cell imaging

HeLa cells were plated on glass-bottom dishes and transfected with PSA plasmids tagged with mCherry on the C-terminus for one day before imaging. For the mitochondria co-localization study, cells were co-transfected with Su9-EGFP (Addgene). Live cells were imaged on a Zeiss LSM900 confocal microscope 24 hours post-transfection. Images were processed using custom macros in ImageJ FIJI.

Protein expression and purification

The shorter human PSA isoform was PCR amplified from pCMV-PSA and inserted into the pFastBac vector (Gibco), with a thrombin cleavage site followed by mCherry and Strep-tag II at C-terminus. Baculoviruses were generated in Spodoptera frugiperda ExpiSf9 cells (Gibco), cultured in ExpiSf CD medium (Gibco) at 27°C. P2 baculovirus was used to infect ExpiSf9 cells at a density of approximately 3.5-5.0 × 10⁶ cells ml⁻¹. After 84-96 h, the cells were collected by centrifugation (5,000 × g, 10 min, 4°C), frozen in liquid nitrogen, and stored at −80 °C until further purification. The cell pellet was disrupted by sonication in lysis buffer (20 mM Tris, pH 7.5, 150 mM NaCl, 10% glycerol). The cell debris was removed by centrifugation (2,000 × g, 10 min, 4°C) and the membrane fraction was removed by ultracentrifugation (142,000 × g, 1 h, 4°C). The supernatant was incubated with Strep-Tactin Superflow beads (IBA Lifesciences) for 1.5 h at 4°C. The protein-bound beads were poured into an open column and washed with 10 column volumes (CV) of lysis buffer. The proteins were eluted with elution buffer (100 mM Tris, pH 8.0, 150 mM NaCl, 2.5 mM desthiobiotin). The eluate was dialyzed in lysis buffer overnight with thrombin to digest the mCherry-Strep tag II. The dialyzed proteins were concentrated with a centrifugal filter unit (Merck Millipore, 50 kDa molecular weight cutoff) and purified by anion exchange chromatography on a HiTrap Q HP column (Cytiva), equilibrated with AE buffer (20 mM Tris, pH 7.5, 75 mM NaCl). The peak fractions were collected and dialyzed in dialysis-P buffer (20 mM Tris, pH 7.5, 150 mM NaCl) overnight. The dialyzed proteins were concentrated to 30 mg ml⁻¹ with a centrifugal filter unit (Merck Millipore, 100 kDa molecular weight cutoff), ultracentrifuged at 98,000 × g for 25 min to remove the aggregation, frozen in liquid nitrogen, and stored at −80 °C until further use.

The cDNA of human SWELL1 was PCR amplified from pCMV-SWELL1 and inserted into the pFastBac vector, with a thrombin cleavage site followed by GFP and Strep-tag II at C-terminus. Baculoviruses were generated in Spodoptera frugiperda ExpiSf9 cells (Gibco), cultured in ExpiSf CD medium (Gibco) at 27°C. P2 baculovirus was used to infect ExpiSf9 cells at a density of approximately 3.5-5.0 × 10⁶ cells ml⁻¹. After 84-96 h, the cells were collected by centrifugation (5,000 × g, 10 min, 4°C), frozen in liquid nitrogen, and stored at −80 °C until further purification. The cell pellet was resuspended and solubilized for 1 h and 4°C in solubilization buffer (20 mM Tris, pH 7.5, 150 mM NaCl, 10% glycerol, 1% n-dodecyl-β-D-maltoside (DDM; Anatrace), 0.2% cholesteryl hemisuccinate Tris salt (CHS; Anatrace)). Insoluble materials were removed by ultracentrifugation (186,000 × g, 1 h, 4°C). The detergent-soluble fraction was incubated with Strep-Tactin XT 4Flow beads (IBA Lifesciences) for 2.5 h at 4°C. The protein-bound beads were poured into an open column and washed with wash buffer (20 mM Tris, pH 7.5, 150 mM NaCl, 10% glycerol, 0.03% glyco-diosgenin (GDN; Anatrace)). The proteins were eluted with elution-B buffer (20 mM Tris, pH 7.5, 150 mM NaCl, 10% glycerol, 0.03% glyco-diosgenin (GDN; Anatrace), 50 mM biotin). The eluate was dialyzed in wash buffer overnight with thrombin to digest the mCherry-Strep tag II. The dialyzed proteins were concentrated with a centrifugal filter unit (Merck Millipore, 100 kDa molecular weight cutoff) and incubated with purified PSA in the ratio of 1:1 (w/w) for 1 h. The sample was then purified by size-exclusion chromatography (SEC) on a Superose 6 Increase 10/300 GL column (Cytiva), equilibrated with SEC buffer (20 mM Tris, pH 7.5, 150 mM NaCl, 0.03% GDN). The peak fractions were collected, concentrated to 3.5-4.0 mg ml⁻¹ with a centrifugal filter unit (Merck Millipore, 100 kDa molecular weight cutoff), ultracentrifuged at 98,000 × g for 25 min to remove the aggregation, and immediately used for cryo-EM grid preparation.

Cryo-EM image acquisition and data processing

C-flat (CF-1.2/1.3-2Au-50, Electron Microscopy Sciences) grids were coated with 50 nm Au by Sputter Coater Leica EM ACE600 and plasma cleaned with Ar/O₂ by Tergeo Plasma Cleaner (Pie Scientific) to make holey 1.2/1.3 gold grids with gold mesh based on published methods^57,58. A 3 μl portion of the protein sample was applied to a glow-discharged homemade gold grid, blotted using a Vitrobot Mark IV (FEI) under 4°C and 100% humidity conditions, and then frozen in liquid ethane. The grid images were obtained with a Glacios microscope (Thermo Fisher Scientific) operated at 200 kV and recorded by a Falcon 4i direct electron detector (Thermo Fisher Scientific). A total of 11,663 movies were obtained in the electron counting mode, with a physical pixel size of 1.1896 Å pixel⁻¹. The data set was acquired with the EPU software, with a defocus range of −0.6 to −1.6 μm. Each image was dose-fractionated to 40 frames to accumulate a total dose of 40 e⁻ Â⁻².

CryoSPARC (version 4.6.0) was preliminarily used for all aspects of data processing⁴⁷. 3,117,293 particles were selected by blob-based auto-picking and extracted in 1.1896 Å pixel⁻¹ from motion-corrected and dose-weighted micrographs. After iterative two-dimensional classification, 240,320 good particles were selected. Well-aligned 79,371 particles were selected from them and used to generate an initial 3D map. Initially selected 240,320 particles were subsequently refined using non-uniform refinement⁴⁷ with C3 symmetry, resulting in an overall map at 3.19 Å. Focus masks for the channel core region (extracellular, transmembrane, and intracellular regions) and for LRR-PSA region (two LRRs and one PSA) are created separately and used for local refinement. Channel core region is refined with C3 symmetry, resulting in a local map at 3.16 Å. For the local refinement of LRR-PSA region, the particle stack was expanded by symmetry expansion with C3 symmetry. Expanded particles then undergo particle subtraction with the inverted mask of LRR-PSA region and 3D classification with the focus mask. 221,271 well-aligned particles are used for the local refinement, resulting in a local map at 3.74 Å (Figure S3-S5).

Model building

The cryo-EM structure of homo-hexameric human SWELL1 (PDB: 7XZH) and the crystal structure of human PSA (PDB: 8SW0) were used as initial model and fitted into density map in COOT⁴⁸. The model was refined using PHENIX with secondary structure restraints⁴⁹ and modified manually in COOT iteratively. The electrostatic potential calculation was performed by the program APBS⁵⁰. The van der Waals radii of the ion pathway were calculated using the HOLE program⁵¹. The figures depicting the molecular structures were prepared with UCSF ChimeraX⁵² and CueMol (http://www.cuemol.org/).

Patch clamp electrophysiology

Patch clamp electrophysiology was performed as previously described^5,7. Cells were seeded onto 12-mm poly-L-lysine–coated glass coverslips one day prior to whole-cell patch clamp recordings. In select experiments, cells were transfected with pIRES2-EGFP expressing WT PSA or various PSA mutants, for 24-36 h before recordings. For hypotonicity-activated VRAC current recordings, whole-cell patch clamp configuration was established in an isotonic bath solution containing (in mM): 105 NaCl, 2 KCl, 1 MgCl₂, 1 CaCl₂, 10 HEPES, and 80 mannitol (pH 7.3 adjusted with NaOH; osmolality ~310 mOsm kg⁻¹), and then a hypotonic solution that has the same ionic composition but without mannitol (osmolality ~230 mOsm kg⁻¹) was perfused. Recording electrodes (2–4 MΩ) were filled with an intracellular solution containing (in mM): 145 CsCl, 10 HEPES, 4 Mg-ATP, 0.5 Na₃-GTP, and 5 EGTA (pH 7.3 adjusted with CsOH; osmolality ~300 mOsm kg⁻¹). For S1P-activated VRAC current recordings in BV2 cells, everything was the same except that the cells were perfused with isotonic solution with 100 nM S1P.

For low intracellular ionic strength-activated VRAC current recordings, whole-cell patch clamp configuration was established in the same isotonic bath solution as above. Intracellular solutions contained (in mM) 25–200 CsCl in addition to 10 HEPES, 4 Mg-ATP, 0.5 Na₃-GTP, and 5 EGTA (pH 7.3 adjusted with CsOH). For intracellular solutions containing 25–125 mM CsCl, mannitol was added to adjust the osmolality to ~300 mOsm kg⁻¹; for intracellular solutions with 175 or 200 mM CsCl, mannitol was added to the bath solution to match its osmolality to that of the intracellular solution.

To determine VRAC permeability to Cl⁻, I⁻, and glutamate, HeLa cells were patched with Cl⁻-based intercellular solution containing (in mM): 120 NaCl, 10 HEPES, and 100 mannitol. Cells were bathed initially in an isotonic solution containing (in mM): 120 NaCl, 10 HEPES, and 100 mannitol, followed by perfusion with three different hypotonic solutions sequentially containing (in mM): 120 NaX, 10 HEPES, and 50 mannitol (where X = Cl⁻, I⁻, or glutamate). Solutions were adjusted to pH 7.3 with NaOH.

Recordings were performed using a MultiClamp 700B amplifier and a Digidata 1440A digitizer (Molecular Devices). Constant voltage ramps (every 5 s, 500 ms duration) were applied from a holding potential of 0 mV to ±100 mV. Data acquisition was carried out with Clampex 10.7 software (Molecular Devices).

Aminopeptidase activity assay

HEK293 cells were lysed for 10 min on ice in digitonin-based lysis buffer containing (in mM): 20 HEPES (pH 7.6), 10 KCl, 1.5 MgCl₂, and 0.025% digitonin (Sigma). Lysates were then centrifuged at 1,000 × g for 10 min at 4°C. The supernatant was collected, and protein concentration was determined using the BCA Protein Assay Kit (Thermo scientific). To measure PSA aminopeptidase activity, 5 μg of total protein from each sample was incubated with the fluorescent substrate Leu-AMC (Bachem) at a final concentration of 100 μM in 96-well plates at 37°C for 1 h. Liberated AMC fluorescence was measured using an Infinite M Plex Microplate Reader (TECAN) with excitation/emission wavelengths of 380/460 nm. The lysis buffer alone served as the blank control. The specificity of the assay was validated by using lysates from PSA KO HEK293 cells and through treatment with the PSA inhibitor tosedostat (10 μM).

cGAMP treatment

Jurkat cells were seeded in 24-well plates at 2 × 10⁶ cells per well one day before treatment. Following removal of the old media by centrifugation (350 × g for 5 min), the cells were resuspended in the fresh media containing 10 μM cGAMP (Invivogen). HeLa cells and TIME cells were seeded in 6-well plates at 3 × 10⁵ cells per well one day before treatment. The old media was gently removed and replaced with fresh media containing cGAMP at a final concentration of 50 μM for HeLa cells and 20 μM for TIME cells. The cells were treated with cGAMP for 1 hr prior to lysis and subsequent immunoblot analysis. The relative levels of phospho-STING versus total STING served as an indicator of cGAMP import. HeLa cells stably overexpressing PSA were generated by transducing lentiviruses packaged from pLenti-EF1a-PSA-P2A-GFP and pLenti-EF1a-PSA-E81A/E107A-P2A-GFP. 2 days post-transduction, GFP-positive cells were FACS-sorted. PSA overexpression in the stable cells were confirmed by immunoblotting.

QUANTIFICATION AND STATISTICAL ANALYSIS

All statistical analyses were performed using GraphPad Prism 9.1 software. Parametric tests, including paired or unpaired two-tailed t-tests, were used for comparisons between two groups with normally distributed data. For comparisons among more than two groups, analysis of variance (ANOVA) and post hoc tests was applied. Data are presented as means ± SEM. Statistical significance was defined as P < 0.05. The number of cells or independent experiments for each analysis is indicated in the corresponding figure and figure legend.

KEY RESOURCES TABLE

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
PSA (HL1531) Rabbit mAb	Thermo Scientific	MA5-47084; RRID: AB_2938156
PSA Rabbit Ab	Thermo Scientific	PA5-83788; RRID: AB_2790940
SWELL1 Rabbit Ab	Yang, J. et al.	N/A
LRRC8B Rabbit Ab	Sigma-Aldrich	HPA017950, RRID: AB_1853346
LRRC8C Rabbit Ab	Sigma-Aldrich	HPA029347, RRID: AB_10601544
LRRC8D Rabbit Ab	Sigma-Aldrich	HPA014745, RRID: AB_1853348
LRRC8E Rabbit Ab	Sigma-Aldrich	HPA020466, RRID: AB_1853352
Phospho-STING (Ser366) (D7C35S) Rabbit mAb	Cell Signaling Technology	19781, RRID: AB_2737062
STING (D2P2F) Rabbit mAb	Cell Signaling Technology	13647, RRID: AB_2732796
Phospho-TBK1/NAK (Ser172) (D52C2) Rabbit mAb	Cell Signaling Technology	5483, RRID: AB_10693472
TBK1/NAK (D1B4) Rabbit mAb	Cell Signaling Technology	3504, RRID: AB_2255663
c-Myc (D84C12) Rabbit mAb	Cell Signaling Technology	5605, RRID: AB_1903938
Monoclonal ANTI-FLAG M2 Mouse mAb	Sigma-Aldrich	F1804, RRID: AB_262044
α-Tubulin (DM1A) Mouse mAb	Cell Signaling Technology	3873, RRID: AB_1904178
β-Actin Rabbit Ab	Cell Signaling Technology	4967, RRID: AB_330288
Peroxidase AffiniPure Goat Anti-Rabbit IgG (H+L)	Jackson ImmunoResearch Labs	111-035-144, RRID: AB_2307391
Peroxidase AffiniPure Goat Anti-Mouse IgG (H+L)	Jackson ImmunoResearch Labs	115-035-166, RRID: AB_2338511
Chemicals, peptides, and recombinant proteins
2’3’-cGAMP (cyclic [G(2’,5’)pA(3’,5’)p])	Invivogen	tlrl-nacga23-5
H-Leu-AMC	Bachem	4001570
S1P	Cayman	62570
3 × FLAG Peptide	Sigma-Aldrich	F4799
Tosedostat	Sigma-Aldrich	SML2303
Dicumarol	Acros Organics	AC204120050
Adenosine 5′-triphosphate magnesium salt	Sigma-Aldrich	A9187
Guanosine 5′-triphosphate sodium salt hydrate	Sigma-Aldrich	H3375
L-Glutamic acid monosodium	Sigma-Aldrich	G1626
Sodium chloride	Sigma-Aldrich	S9888
Sodium iodide	Sigma-Aldrich	383112
n-dodecyl-β-D-maltoside	Anatrace	D310S
Cholesteryl hemisuccinate Tris salt	Anatrace	CH210
Glyco-diosgenin	Anatrace	GDN101
Critical commercial assays
Pierce BCA Protein Assay Kit	Thermo Scientific	23227
Q5 Site-Directed Mutagenesis Kit	New England Biolabs	E0554S
Phanta Max Super-Fidelity DNA Polymerase	Vazyme	P505
Microvascular Endothelial Cell Growth Kit-VEGF	ATCC	PCS-110-041
Lipofectamine 2000 Transfection Reagent	Invitrogen	11668019
SuperFemto ECL Chemiluminescence Kit	Vazyme	E423-01
EZ-Link Sulfo-NHS-SS-Biotin	Thermo Scientific	A39258
Pierce NeutrAvidi Agarose	Thermo Scientific	29200
Pierce Protein G Magnetic Beads	Thermo Scientific	88847
Anti-FLAG M2 Magnetic Beads	Sigma-Aldrich	M8823
Strep-Tactin Superflow resin	IBA Lifesciences	2-1206-025
Strep-Tactin XT 4Flow resin	IBA Lifesciences	2-5010-025
Deposited data
SWELL1-PSA heterocomplex coordinates	This paper	PDB: 9O5K
SWELL1-PSA heterocomplex EM map	This paper	EMDB: EMD-70413
Uncropped western blots	This paper	Mendeley data: http://dx.doi.org/10.17632/vybv2h96pd.1
Experimental models: Cell lines
HeLa cell	ATCC	CCL-2; RRID: CVCL_0030
HEK293 cell	ATCC	CRL-3216; RRID: CVCL_0063
BV2 cell	Gift from Tony Wyss-Coray45,46	N/A
Jurkat T cell	ATCC	TIB-152 RRID: CVCL_0367
TIME cell	Gift from Lingyin Li10	CRL-4025; RRID: CVCL_0047
ExpiSf9 cell	Gibco	A35243
Oligonucleotides
See Table S2 for all oligonucleotides	This paper	N/A
Recombinant DNA
pCMV6 Vector	OriGene	PS100001
pCMV6-SWELL1	OriGene	RC208632
pCMV6-PSA(long)	OriGene	RC209037
pIRES2-EGFP Vector	Clontech	PT3743-5
pIRES2-PSA	This paper	N/A
pLenti-EF1a-P2A-Puro Vector	Origene	PS100142
pLenti-EF1a-PSA	This paper	N/A
pLenti-EF1a-PSA-E81A/E107A	This paper	N/A
Su9-EGFP	Addgene	23214
pCMV6-LRRC8B	OriGene	RC205553
pCMV6-LRRC8C	OriGene	RC222603
pCMV6-LRRC8E	OriGene	RC209849
pIRES2-SWELL1	Qiu, Z. et al.15	N/A
pIRES2-LRRC8B	This paper	N/A
pIRES2-LRRC8C	This paper	N/A
pIRES2-LRRC8E	This paper	N/A
pCMV6-LRRC8D	OriGgene	SC319580
PX458-mCherry Vector	Addgene	161974
PX458-PSA-sgRNA	This paper	N/A
lentiCRISPR-v2-Blast Vector	Addgene	52961
lentiCRISPR-PSA-sgRNA	This paper	N/A
pFastBac Vector	Gibco	10360014
pFastBac-PSA	This paper	N/A
pFastBac-SWELL1	This paper	N/A
Software and algorithms
ImageJ	NIH	https://imagej.nih.gov/ij/download.html; RRID: SCR_003070
GraphPad Prism	GraphPad Software	https://www.graphpad.com/; RRID:SCR_002798
ZEN	Zeiss	https://www.zeiss.com/microscopy/en/products/software/zeiss-zen-lite.html
pCLAMP 10.7	Molecular Devices	http://www.moleculardevices.com/products/software/pclamp.html; RRID:SCR_011323
Clampfit 10.7	Molecular Devices	N/A
EPU	Thermo Fisher Scientific	https://www.thermofisher.com/us/en/home/electron-microscopy/products/software-em-3d-vis/epu-software.html
CryoSPARC v.4.6.0	Punjani, A. et al.47	https://cryosparc.com/
COOT	Emsley, P. et al.48	https://www2.mrc-lmb.cam.ac.uk/personal/pemsley/coot/
PHENIX	Adams, P.D. et al.49	https://phenix-online.org/download
APBS	Jurrus, E. et al.50	https://server.poissonboltzmann.org/
HOLE	Smart, O.S. et al.51	https://www.holeprogram.org/
UCSF ChimeraX	Pettersen, E.F. et al.52	https://www.cgl.ucsf.edu/chimerax/
Clustal Omega	Madeira. F. et al.53	https://www.ebi.ac.uk/jdispatcher
ESPript 3	Robert, X. et al.46	http://espript.ibcp.fr/
CueMol	N/A	http://www.cuemol.org/
Other
CF-1.2/1.3-2Au	Electron Microscopy Sciences	CF213-50-Au

Highlights.

• Puromycin-sensitive aminopeptidase binds to the cytosolic LRR domains of SWELL1.

• PSA overexpression inhibits VRAC activation; deletion elevates basal activity.

• PSA’s modulation of VRAC requires physical binding but not aminopeptidase activity.

• PSA regulates cGAMP transport by modulating VRAC activity.