Activation of the STING pathway in mice has been established as a model for the initiation of SS. This pathway is normally activated following exposure to foreign nucleic acids and is thought to mimic exposure of cells to DNA/RNA from viruses and bacteria. Activation of this pathway in salivary glands is characterized by initiation of a type-1 interferon response, mild immune cell infiltration, and a marked loss of saliva secretion without obvious morphological damage and, therefore, mimics the early clinical manifestations of SS disease. Thus, to investigate the early cellular events in acinar cells during the initiation of SS in mice, we chose to pharmacologically activate this pathway using DMXAA, a STING pathway agonist. As described in Methods, DMXAA (or control solution) was administered on day 0 and day 21 of the experimental timeline (Figure 1A). Immunofluorescent staining in sliced SMG tissue indicated that STING protein was increased in SMG in the DMXAA-treated mouse on day 28, seven days after the final DMXAA administration (Figure 1—figure supplement 1) confirming the activation of the STING pathway. Whole saliva production from the major salivary glands was evaluated on day 28 following systemic stimulation with the muscarinic receptor agonist, pilocarpine. To avoid potential weight-related variations in saliva secretion, the total saliva output was normalized to the individual mouse’s body weight. Notably, the average saliva production was reduced from 130.1±48.96 mg in vehicle-treated mice to 63.71±30.41 mg in DMXAA-treated mice, a reduction in saliva production of 48.97% (Figure 1B). Consequently, DMXAA treatment resulted in 51.99% saliva production compared to vehicle-treated mice (Figure 1C). Moreover, H&E staining indicated that mild immune infiltration was observed in the DMXAA-treated mice with no overt changes to the morphology of the gland (Figure 1D). The mass of the SMG was also not significantly different in DMXAA vs. vehicle-treated animals, consistent with no loss of secretory tissue (Figure 1E). Collectively, these results suggest that DMXAA-treated mice exhibit characteristics of early-stage SS and could be a useful model for investigating the pathophysiological mechanisms underlying secretory dysfunction and advancing our understanding of the disease’s progression.

To further investigate the individual relative contribution of the SMG and PG to the decrease in total saliva secretion using a more physiological stimulation paradigm, we performed experiments where the nerve bundle innervating a particular gland was electrically stimulated and saliva secretion quantitated. Previous research in our lab has established the range and parameters for physiological stimulation of secretion (Takano et al., 2021). The production of saliva was significantly diminished in DMXAA-treated animals compared with vehicle controls at stimulation frequencies of 7 and 10 Hz in SMG (Figure 1F) and at 5, 7, and 10 Hz in the PG (Figure 1G). These findings confirm that activation of the STING pathway reduces the function of both SMG and PG, consistent with the diminished production of whole saliva. Notably, the reduction in function of PG, the gland responsible for the majority of stimulated saliva secretion, was relatively greater than in SMG.

An increase in intracellular Ca²⁺ plays a central role in regulating the cellular machinery underlying the fluid secretion mechanism. In particular, as noted, an increase in Ca²⁺ is important for the activation of ion channels localized in particular domains of the polarized acinar cell which play a central role in saliva secretion (Takano et al., 2021). Given that the precise spatiotemporal characteristics of the Ca²⁺ signal in salivary acinar cells are thought to be fundamental to the appropriate activation of the fluid secretion machinery, we evaluated whether hypofunction following DMXAA treatment resulted from dysregulation of the stimulated Ca²⁺ signal. Previous research in our lab developed a platform to study Ca²⁺ signaling in vivo using Multiphoton (MP) imaging in transgenic mice engineered to express a genetic-encoded Ca²⁺ indicator, GCaMP6f, specifically in acinar cells (Takano et al., 2021). These mice were generated by crossing homozygous mice expressing the fast genetically encoded Ca²⁺ indicator GCaMP6f (B6J.Cg-Gt(ROSA)26Sor^{tm95.1(CAG-GCaMP6f)Hze}/MwarJ) floxed with a STOP cassette with heterozygous tamoxifen-inducible Mist1 Cre mice (B6.129-Bhlha15^{tm3(cre/ERT2)Skz}/J). The protocol for STING pathway induction was applied to the mice expressing GCamp6f (Figure 2—figure supplement 1B). Salivary gland function was assessed by the amount of pilocarpine-induced saliva secretion. The secretion deficiency observed in wild-type mice was recapitulated in these mice from a different genetic background (Figure 2—figure supplement 1B). We reasoned that decreased fluid secretion could result from reduced or dysregulated Ca²⁺ signaling in DMXAA-treated mice. Therefore, next, we compared the spatially averaged Ca²⁺ signal evoked by direct nerve stimulation in DMXAA-treated vs. vehicle control animals in vivo in SMG. SMG were stimulated at frequencies optimum for fluid secretion (1–10 Hz) for 10 s and the Ca²⁺ signals were recorded. Figure 2A shows standard deviation (SD)-projection images visualizing the spatial and amplitude changes in Ca²⁺ throughout the field of view during the entire period of stimulation at the indicated frequencies. In vehicle-treated mice, the Ca²⁺ signals at low stimulus strengths occurred in a minority of the cells and predominantly propagated below the apical PM. As the stimulation strength increased, Ca²⁺ signals became more pronounced, and more acinar cells responded. Strikingly, acinar cells in DMXAA-treated mice demonstrated enhanced sensitivity to stimulation. At lower stimulation strengths, a larger number of acinar cells responded, and the spatially averaged Ca²⁺ signals in these cells were notably larger when compared to those in the control group. Figure 2B shows a time series of images following 7 Hz stimulation. This augmented response was manifested as an elevated maximum peak [Ca²⁺] (Figure 2D), shorter latency (Figure 2E), and larger area under the curve (AUC) during stimulation in DMXAA-treated animals (Figure 2F).

In addition to the absolute magnitude of the spatially averaged Ca²⁺ signal, the subcellular spatial characteristics of the stimulated Ca²⁺ rise are also important for appropriate stimulation of fluid secretion (Takano et al., 2021). Ca²⁺ signals stimulated following nervous stimulation in SMG are invariably initiated in the extreme apical pole of acinar cells and subsequently establish a standing gradient that dissipates rapidly to result in apically confined signals that do not substantially propagate to the basal aspects of the cell following physiological stimulation (Takano et al., 2021). We, therefore, investigated if the spatial characteristics of stimulated Ca²⁺ signals were altered in DMXAA-treated animals. SD image projections generated during the period of stimulation demonstrated that the [Ca²⁺]_i increase was tightly localized below the apical PM within the acinar cells in the vehicle-treated animals (Figure 3A). However, in the DMXAA-treated animals, the [Ca²⁺]_i exhibited a more global distribution through the entire cell cytoplasm (Figure 3B). The [Ca²⁺]_i was visualized via line-scan plots revealing the temporal alterations along a line extending from the apical PM to the basolateral PM, traversing the nucleus, over time within an acinar cell. A significant [Ca²⁺]_i elevation was evident at the basolateral aspects of the acinar cell in the DMXAA-treated animals (Figure 3D) when compared to the vehicle-treated control (Figure 3C). As shown in the kinetic plots,the [Ca²⁺]_i in the apical region is greater in DMXAA-treated mice compared to the vehicle-treated mice (Figure 3E). Moreover, a significant Ca²⁺ signal was observed in the extreme basal region of the cell in DMXAA but not in vehicle-treated animals (Figure 3F). The comparison of Ca²⁺ signal ratios at the apical versus basolateral PM indicated the most significant globalization of the Ca²⁺ signal occurred at 10 Hz stimulation (Figure 3G) which corresponds to the stimulation strength that results in maximal fluid secretion (Takano et al., 2021). In total, these data demonstrate that the magnitude of the spatially averaged Ca²⁺ signal, together with the spatiotemporal characteristics of the signal are altered in the DMXAA-treated animals, but that these changes can not readily account for the reduction in fluid secretion observed.

The rate-limiting step for the secretion of fluid is the activation of the Ca²⁺-activated Cl^- channel, TMEM16a. We considered that a reduction in fluid secretion could conceptually occur by a reduction or mislocalization of TMEM16a protein, or by compromised muscarinic receptor-stimulated activation of the channel. Western blotting indicated the TMEM16a protein expression was comparable between the vehicle and SS mouse models (Figure 4A and B). In addition, immunolocalization using confocal microscopy demonstrated that TMEM16a localization remained largely unchanged in the SS mouse model (Figure 4C). Thus, a decrease in protein expression or mislocalization of the protein does not result in a reduction in stimulated saliva secretion. Similarly, the expression and localization of Aquaporin 5 were also not altered (Figure 4—figure supplement 1A–C). We next investigated whether the activation of this ion channel was compromised in DMXAA-treated animals using whole-cell patch clamp electrophysiology. In the absence of stimulation, no Cl^- currents were observed in either vehicle or DMXAA animals following either depolarizing or hyperpolarizing voltage steps from a holding potential of –50 mV (Figure 4D). In the presence of 1 μM of the muscarinic agonist Carbachol (CCh), robust Cl^- currents were measured in acini prepared from vehicle-treated animals (Figure 4E), which were greatly reduced in DMXAA-treated animals (Figure 4D and F). Reduced CCh-stimulated Cl^- currents could potentially occur because of altered Ca²⁺ regulation of TMEM16a following disruption of the spatial characteristics of the stimulated Ca²⁺ signal.Theoretically,it is also possible that the [Ca²⁺]_i in the immediate vicinity of TMEM16a was disrupted, despite the augmented spatially averaged peak response. We, therefore, next tested whether TMEM16a activity stimulated directly by 0.5, 1 or, 5 μM Ca²⁺ in the pipette solution (and thus globally in the cytoplasm) was altered in DMXAA-treated animals. Surprisingly, TMEM16a was activated to a similar extent by Ca²⁺ in the SS mouse model (Figure 5A and B). In total, our data suggest that TMEM16a abundance, localization, or activity per se are not altered and thus do not explain the significant reduction in saliva secretion in the DMXAA-treated model.

An alternative mechanism could be that the microdomain between the apical ER Ca²⁺ release sites and the apical PM TMEM16a is disrupted in the disease model, resulting in compromised local coupling between the ER and PM channels. Therefore, we investigated how the activation of TMEM16a was affected by buffering the CCh-stimulated cytosolic Ca²⁺ with slow and fast Ca²⁺ chelators. EGTA is a high-affinity Ca²⁺ buffer with slow kinetics, while BAPTA has much more rapid kinetics. Experimentally, BAPTA can quickly buffer Ca²⁺ changes close to Ca²⁺ release sites and limit local activation of effectors within 20 nm. In contrast, EGTA has been shown to attenuate the rise in the bulk cytosol but is too slow to buffer local Ca²⁺ in a restricted microdomain (Eisner et al., 2023). In the presence of BAPTA, no CI^- currents were detected in either vehicle-treated or DMXAA-treated animals (Figure 6A). However, in the presence of the slow Ca²⁺ chelator, EGTA, CCH-stimulated Cl^- currents were observed in acinar cells from vehicle-treated animals but were absent in the DMXAA-treated animals (Figure 6A and B). These data suggest that physiologically TMEM16a is activated by local changes in Ca²⁺ signaling as a result of IP₃-induced Ca²⁺ release in the apical domain of the acinar cell.

We noted that the abundance of IP₃R2/3 was not significantly altered in DMXAA-treated animals (Figure 7—figure supplement 1A–C), nevertheless next, we employed STED super-resolution microscopy to closely examine the spatial relationship between apical PM TMEM16a and IP₃R3 on the apical ER (Figure 7A). Despite the cell-cell contact distance remaining consistent in the disease model, as indicated by the distance between TMEM16a on the PM of adjacent acinar cells (Figure 7F), a notable increase in distance between the apical TMEM16a and IP₃R3 expressed on apical ER compared to the control group was observed (Figure 7E). In the control mice, the distance between TMEM16a and IP₃R3 was on average 84±17 nm, versus 155±20 nm in the SS disease mice. Similarly, the distance between IP₃R3 in adjacent cells was increased from 505±34 nm to 689±68 nm (Figure 7C and D). In total, these observations support the conclusion that the reduced activity of the TMEM16a channel is attributable to the disruption of the microdomain between TMEM16a and IP₃R3, such that the Ca²⁺ flux through the IP₃R is not communicated appropriately to its effector, TMEM16a.

Ca²⁺ modulates cellular metabolism by the intricate bidirectional interaction between the ER and mitochondria. Ca²⁺ transfer between ER and mitochondria is essential for optimal bioenergetics, and dysregulated [Ca²⁺]_i can be deleterious to mitochondrial function and alter morphology (Duchen, 2000; Csordás et al., 2006; Ye et al., 2021; Katona et al., 2022). The transfer of Ca²⁺ between ER and mitochondria is dependent on the intimate physical localization of the organelles (Katona et al., 2022). Notably, aberrant mitochondrial morphology has been reported in the salivary glands of SS patients (Barrera et al., 2021). We first investigated mitochondrial abundance and morphology by immunofluorescence staining with antibodies directed against ATP5A, a component of the ATP synthesis machinery to visualize mitochondria, and Na⁺/K⁺ ATPase to localize the plasma membrane (Figure 8A). Using previously published methodologies (Valente et al., 2017; Harwig et al., 2018), quantification revealed a 22.16%±4.95 reduction in mitochondrial numbers in the SS mouse model relative to the vehicle-treated control (Figure 8B). Consistent with reduced mitochondrial numbers, less area was occupied by mitochondria in DMXAA-treated acinar cells (Figure 8C). Mitochondria morphology is intricately linked to their bioenergetic status We next evaluated mitochondrial morphology by their ‘so-called’ aspect ratio (AR) and form factor (FF) in DMXAA and vehicle-treated animals.The AR, the length of the major over minor axes of mitochondria documents the degree of fragmentation or elongation of individual mitochondria (Katona et al., 2022; Harwig et al., 2018). Mitochondria exhibited an 18.35%±4.62 decrease in mitochondrial elongation (Figure 8D) and a 20.7%±7.78 decrease in mitochondrial branching (Figure 7E) in the disease model compared to the vehicle-treated control condition. Importantly, these changes in mitochondrial number and morphology were not exclusive to the SMG as similar patterns were observed in the PG mitochondria, again marked by reduced mitochondrial count, increased fragmentation, and decreased branching (Figure 8—figure supplement 1A–E).

Next, we utilized electron microscopy (EM) to investigate mitochondrial ultrastructure. At low magnification, acinar cells from control mice contained defined mitochondria and well-formed ER stacks (Figure 9A. blue arrow). In contrast, the ER structure was disrupted in the SS disease model (Figure 9A’). At higher magnification,the coordinated ER structure was largely absent in diseased mice (Figure 9B and B’), and the proximity between ER and mitochondria was disrupted (Figure 9H and I). Moreover, we also observed scattered mitochondrial cristae at the highest magnification (Figure 9C’ , and G). Consistent with immunofluorescence studies, quantification of EM micrographs revealed that mitochondria were smaller, more fragmented (Figure 9D and E), and rounder (Figure 9F) in shape. In summary, our results collectively indicate significant morphological alterations in mitochondria in the SS disease model.

Mitochondrial morphology is a dynamic process that is intimately associated with mitochondrial bioenergetics and alterations in both, occur in response to changes in cellular status (Hogan et al., 2005; Martínez et al., 2020; Navaratnarajah et al., 2021; Schaefer et al., 2019; Zorova et al., 2018) Therefore, we investigated if changes in morphology might be associated with the disrupted function of mitochondria in the disease model. We measured mitochondrial membrane potential (ΔΨ_m), established by the electrochemical H⁺ gradient, which is the driving force of ATP production. Isolated SMG acinar cells were loaded with TMRE, a ΔΨ_m-specific dye, and MitoTracker Green, to confirm mitochondrial localization and to facilitate the normalization of indicator loading. The maximal z-stacks projection images taken by confocal microscopy revealed colocalization of TMRE with MitoTracker Green (Figure 10A). The basal TMRE fluorescence was reduced in cells from DMXAA vs. vehicle-treated animals (Figure 10—figure supplement 1). To assess ΔΨ_m, we quantified the relative maximum dissipation of ΔΨ_m in DMXAA and vehicle-treated acinar cellsby the mitochondrial uncoupler, FCCP (Figure 10B). Consistent with the reduction in basal TMRE fluorescence, the change in TMRE fluorescence normalized to mitochondrial content revealed a marked reduction in ΔΨ_m in the acinar cells from the SS disease model (Figure 10C).

An appropriate ΔΨ_m mitochondrial membrane potential is vital for maintaining bioenergetics (Zorova et al., 2018). Given that mitochondrial ΔΨ_m was significantly depolarized in DMXAA-treated animals, we next evaluated the OCR, a key metric of mitochondrial bioenergetic function in isolated SMG acinar cells. We employed sequential exposure to agents that target the function of the mitochondrial electron transport chain (ETC) using Seahorse technology (Figure 10D). Our results revealed a 25% reduction in basal OCR in the SS model compared to the control animals (at –25.25±7.89 pmol/min; Figure 10E). While ATP-linked respiration showed no significant difference in post-oligomycin-induced ETC Complex V blockade in both conditions (Figure 10F). Intriguingly, the FCCP-provoked maximal respiration rate, an indicator of stress tolerance, remarkably declined by 47%±9.19 after FCCP treatment in the SS model (Figure 10G). These data indicate impaired mitochondrial function and stress responses in the SS mouse model.

The murine model of Sjögren’s syndrome was established through the induction of the STING pathway (Bagavant and Deshmukh, 2020). Briefly, 8–10 weeks old female C57BL/6 J wild-type (WT) mice (Jackson Laboratory; Jax 000664) received subcutaneous injections of DMXAA (Vadimezan; GC16280) at a concentration of 25 mg/kg of body weight on both day 0 and day 21 of the experimental timeline (see Figure 1). The control mouse received vehicle (5% sodium bicarbonate; Sigma-Aldrich; S8761), the DMXAA solvent at the corresponding time points. Experiments were performed on day 28 of the experimental timeline. Animals had access to water and were fed ad libitum. All animal procedures were approved by the University of Rochester Committee on Animal Resources (UCAR-2001-214E).

The mice were fasted for 2 hr prior to the evaluation of saliva production. The mice were anesthetized with a solution containing Ketamine (10 mg/mL) and Xylazine (1 mg/ml) by intraperitoneal injection (IP) at a dose of 7 μl/gm body weight over 2 min. The mouse was placed on a heating pad at 37℃ during experimentation. A Salimetrics Children’s swab (Salimetrics; Cat. no. 5001.05) was placed within the oral cavity of each mouse. The mice were administered the muscarinic agonist pilocarpine (0.375 mg/kg body weight; Millipore Sigma; P6503) by IP injection. Two minutes after the pilocarpine injection, saliva was collected for the following 15 min. The saliva absorbed was subsequently separated from the moist swab through centrifugation at 10,000 rpm for 1 min. The measurement of saliva weight served as a quantitative evaluation of the efficacy of whole saliva secretion. To measure neurotransmitter-stimulated saliva secretion more directly, the mouse was anesthetized as previously described (Takano et al., 2021) and a surgical incision was made in the skin to expose the submandibular gland (SMG). The surrounding connective tissue was excised to facilitate positioning within a custom-made 3D-printed gland holder. A pair of stimulation electrodes were attached to the duct bundle and the SMG. The pre-weighed filter paper was positioned within the oral cavity of the mouse to capture saliva secretion. Secretion was initiated by electrical stimulation sequences generated by a stimulus isolator (Iso-flex, A.M.P.I.) set at 5 mA, 200 ms, at frequencies of 1, 3, 5, 7, and 10 Hz with train frequency and duration (typically 1 min) controlled by a train generator (DG2A, Warner Instruments). The interval between each stimulus was 3 min. After stimulation, the filter paper was removed and weighed. The difference between the weight of filter paper before and after the electrode stimulation represented the saliva produced by the respective salivary gland during the given stimulation period.

To generate mice expressing a fluorescent Ca²⁺ indicator in exocrine acinar cells, homozygous mice expressing the fast genetically encoded Ca²⁺ indicator GCaMP6f (B6J.Cg-Gt(ROSA)26Sor^{tm95.1(CAG-GCaMP6f)Hze}/MwarJ) floxed by a STOP cassette, were crossed with heterozygous tamoxifen-inducible Mist1 Cre mice (B6.129-Bhlha15^{tm3(cre/ERT2)Skz}/J)A week before the DMXAA or 5% sodium bicarbonate injections, tamoxifen (Sigma-Aldrich; T5648) was given to the mice via oral gavage at a dose of 0.25 mg/g of body weight for three consecutive days to excise the loxP sites flanking the STOP codon allowing expression of the Ca²⁺ indicator within salivary glands. The mice were anesthetized and gland-exposed, as described previously (Takano et al., 2021; Takano and Yule, 2022). The immobilized gland was secured within the holder using a cover glass and maintained in Hank’s salt solution (HBSS). Ca²⁺ imaging was conducted in vivo via two-photon microscopy using an Olympus FVMPE-RS system equipped with an Insight X3 pulsed laser (Spectra-Physics) utilizing a heated (OKOLab COL2532) 25 x water immersion lens (Olympus XLPlan N 1.05 W MP). GCaMP6f was excited at 950 nm and emission collected between 495–540 nm, with images captured at 0.5 s intervals following stimulation for 10 s with 3 min between stimulation periods. Statistical analyses were performed with two-way ANOVA with multiple comparisons using Prism (GraphPad) as indicated in the figure legends.

Following verification of decreased saliva secretion in mice, glands were processed for immunocytochemistry. Briefly, the isolated salivary glands were fixed in 4% paraformaldehyde at 4℃ overnight. The fixed gland was processed, embedded in paraffin, and subsequently sliced into 5 μm thick sections. Two temperature-induced antigen retrieval protocols were used either based on HIER buffer (10  mM Tris-base, 1  mM EDTA-dehydrate, pH 9.2) or sodium citrate buffer (10 mM sodium citrate, 0.05% Tween 20, pH 6.0). Gland sections were blocked with the 10% donkey serum in 0.2% PBSA (PBS + BSA) at room temperature (RT) for 1 hr. Sections were incubated with the primary antibody at 4℃ overnight (TMEM16a (Millipore Sigma; P6593; 1:250), Na⁺/K⁺ ATPase (Abcam; ab2872; 1:250), ATP5A (Abcam; ab14748; 1:500), AQP5 (Abcam; ab239904; 1:500), STING (Cell signaling Technology, Cat. 13647; 1:500)). Following washing, the sections were then incubated with the secondary antibody at RT for 1 hr (Donkey anti-rabbit Alexa 488 (Thermo Fisher Scientific; A-21206; 1:500), Donkey anti-mouse Alexa 594 (Thermo Fisher Scientific; A-21203; 1:500)). Nuclei were identified by incubation in DAPI (Thermo Fisher Scientific; Cat. 62248; 1:1000) at RT for 5 min. Tissue sections were mounted using Immu-Mount solution on a slide and then sealed under a coverslip. Images were acquired by Olympus FV1000MP confocal microscopy employing an Olympus UPlanSApo 60 x oil immersion objective. The analysis of images was performed using FIJI software. Statistical analyses were performed with a t-test using Prism (GraphPad) as indicated in the figure legends.

Acinar cells were allowed to adhere to Cell-Tak-coated glass coverslips for 15 min before experimentation. Coverslips were transferred to a chamber containing extracellular bath solution (155 mM tetraethylammonium chloride to block K⁺ channels, 2 mM CaCl₂, 1 mM MgCl₂, 10 mM HEPES, pH 7.2). Cl^- currents in individual cells were measured in the whole cell patch clamp configuration using pClamp 9 and an Axopatch 200B amplifier (Molecular Devices). Recordings were sampled at 2 kHz and filtered at 1 kHz. Pipette resistances were 3–5 MΩ, and seal resistances were greater than 1 GΩ. Pipette solutions (pH 7.2) contained 60 mM tetraethylammonium chloride, 90 mM tetraethylammonium glutamate, 10 mM HEPES, 1 mM HEDTA (N-(2-hydroxyethyl) ethylenediamine-N, N’, N’-triacetic acid) and 20 μM CaCl₂ were used to mimic physiological buffering and basal [Ca²⁺]_i conditions (~100 nM Ca²⁺). Free [Ca²⁺] was estimated using Maxchelator freeware. Agonists were directly perfused onto individual cells using a multibarrel perfusion pipette. The pipette solution for the increased basal [Ca²⁺]_i contained hEDTA and a free [Ca²⁺]_i of 500 nM, 1 mM, or 5 mM to induce calcium-activated Cl^- currents without the addition of any agonists. Experiments comparing EGTA, BAPTA, and HEDTA effects upon chloride currents induced by CCh stimulation contained 5 mM free concentrations of the chelator and 100 nM free [Ca²⁺] in the patch pipette. Chloride currents following agonist application were monitored with a single voltage step to 80 mV from a holding potential of –80 mV every second until current magnitudes reached a plateau. Current-voltage relationships were obtained by 20 mV incremental steps between –80 mv and 120 mV from a holding potential of –50 mV.

3D STED microscopy was performed using an Abberior Instruments Expert Line STED microscope equipped with an Olympus UPLSAPO 100x/1.4NA oil immersion objective. Briefly, lobules <1 mm were isolated following injection of saline beneath the capsule with a 29-gauge needle. The connecting tissue was digested in 0.1 mg/ml collagenase containing image buffer at 37℃ for 5 min. Then isolated lobules were fixed in 100% methanone at –20℃ for 5 min, and subsequently were blocked with 10% BSA in 0.1% PBST (PBS + 0.1% Tween20) at RT for 1 hr with gentle shaking. Isolated lobules were incubated with primary antibodies overnight at 4 °C (TMEM16a (Millipore Sigma; P6593; 1:300), IP₃R3 (BD Transduction Laboratory; Cat. 610313; 1:200)). After being washed with 0.1% PBST, the membranes were incubated with secondary antibodies at RT for 1 hr (STAR RED, goat anti-rabbit IgG secondary antibody (Abberior, Cat#STRED-1001–500 UG; 1:1000), Alexa Fluor 594 anti-rabbit IgG secondary antibody (Molecular Probes Cat#A-11037; 1:1000)). The tissue was mounted on the slides with Prolong Gold antifade reagent (Invitrogen; Cat. P36930). Sequential confocal and STED images were obtained following excitation of Alexa Fluor 594 and STAR RED by 594 and 640 nm lasers, respectively. Both fluorophores were depleted in three dimensions with a 775 nm pulsed STED laser. Z-stacks were obtained by collecting images at 50 nm intervals using the 3D STED mode. Rescue STED was employed to minimize the light dosage. Blend mode depth projection images were generated and fluorophore volumes and interfaces between these volumes were analyzed using FIJI.

Isolated SMGs were finely minced and subsequently resuspended in a solution composed of 0.5% bovine serum albumin (BSA) in Hank’s balanced salt solution (HBSS). To isolate acinar cells, the minced tissue was incubated in 0.5% BSA/HBSS containing 0.2 mg/ml of collagenase type II (Worthington; LS004204) for 30 min. Following this incubation, the suspension of cells was centrifuged at 500 rpm for 1 min and the cellular pellet was then resuspended in 40 μg/ml of Trypsin inhibitor (Millipore; Cat. 65035) to terminate further digestion. The function of mitochondria was assessed in isolated acinar cells by measurement of oxygen consumption rate (OCR) employing a Seahorse XF Cell Mito Stress Test system (Agilent, USA). Briefly, Equal sized SMG cell pellets were suspended in buffer and 10 μl of the acinar cell suspension was seeded into individual wells of Seahorse cell culture microplates coated with 10 uL of Cell-Tak (0.25 mg/ml) and the OCR was determined utilizing the Seahorse XFe96 extracellular flux analyzer following sequential exposure to 4 μg/ml oligomycin (Millipore Sigma; O4876), 4 µM carbonyl cyanide-4 (trifluoromethoxy)phenylhydrazone (FCCP; Millipore Sigma; C2920), and 0.5 µM rotenone/antimycin (Millipore Sigma; R8875; A8674) to measure the quantification of basal respiration, ATP-linked respiration, and maximum respiration rate, respectively. Statistical analyses were performed with a t-test using Prism (GraphPad) as indicated in the figure legends.

Isolated SMG acinar cells were loaded with 20 nM Tetramethylrhodamine, Ethyl Ester (TMRE; ThermoFisher Scientific: T669), and 1 μM of MitoTracker Green (Invitrogen; M7514). Fluorescence of both TMRE and MitoTracker Green was captured simultaneously using an inverted epifluorescence Nikon microscope with a 40 x oil immersion objective. The TMRE fluorescence was excited at 560 nm and emitted light collected at 574 nm; MitoTracker Green was excited at 488 nm and emitted light collected at 530 nm. Images were obtained every 1 s with an exposure of 20 ms and 4 × 4 binning using a digital camera controlled by TILL Photonics, TILLvision software. The acinar cells were exposed to 4 μM FCCP for 3 mins by perfusion to rapidly dissipate the membrane potential. Mitochondrial membrane potential was quantified as the change in the ratio of TMRE/MitoTracker Green fluorescence before and after the administration of FCCP. Statistical analyses were performed with a t-test using Prism (GraphPad) as indicated in the figure legends.

Finely minced salivary glands were homogenized in a lysis buffer supplemented with protease inhibitor cocktail (Complete mini; Roche Diagnostics) for 16–20 strokes. After incubating on ice for 30 min, solubilized proteins were separated by centrifugation at 13,000 rpm at 4℃ for 30 min. 10 μg of protein lysate was loaded on 7.5–12% SDS- polyacrylamide gels. Subsequently, the proteins were transferred to PVDF membranes at a voltage of 35 V at 4 °C overnight. The membrane was blocked with 5% non-fat skimmed milk in TBST (50 mM Tris-HCl, pH 7.5 with 0.1% Tween20) at RT for 1 hr and subsequently incubated with primary antibodies overnight at 4 °C (Actin (Millipore Sigma; A2228; 1:10000), IP₃R2 (Antibody Research Corporation; 1:1000), IP₃R3 (BD Transduction Laboratory; Cat. 610313; 1:1000), TMEM16a (Abcam; ab84115; 1:1000)). After being washed with 0.1% TBST, the membranes were incubated with secondary antibodies at RT for 1 hr (Goat anti-rabbit IgG (H&L) (Invitrogen; SA535571; 1:10000), Goat anti-mouse IgG (H&L) (Invitrogen; SA535521; 1:10000)). Protein band intensity from western blotting was quantified by FIJI. The relative ratio of DMXAA-treated/vehicle control was calculated in Excel. Lastly, graphical generation and statistics were performed with a t-test using Prism (GraphPad) as indicated in the figure legends.