Salivary fluid secretion is vital to the health of the oral cavity (Pedersen et al., 2002). This is most clearly appreciated in disease states which result in hypofunction where patients experience difficulty chewing and swallowing food, speaking and are susceptible to oral infections that result in dental caries (Melvin, 1991). Fluid secretion from the salivary gland requires the trans epithelial movement of Cl^- across polarized acinar cells (Foskett, 1990; Melvin, 1999; Melvin et al., 2005; Turner and Sugiya, 2002). The polarization of exocrine cells is a consequence of the presence of tight junction protein complexes which functionally separate the transport machinery in the apical PM from those in the functionally equivalent, basal and lateral PM (collectively termed basolateral PM). This structural specialization allows distinct proteins to be localized in each membrane to promote vectoral ion and fluid movement (Melvin et al., 2005). The current, widely accepted model for fluid secretion posits that Cl^- are transported against their electrochemical gradient by the Na⁺/K⁺/2Cl^- cotransporter, NKCC1 at the basolateral plasma membrane (PM) (Evans et al., 2000) and subsequently enters the lumen of the acinus across the apical PM through TMEM16a chloride channels (Romanenko et al., 2010a). Cl^- accumulation imparts a negative potential to the lumen relative to the basolateral PM, which draws Na⁺ paracellularly through tight junctions (Melvin et al., 2005). Subsequently, water follows osmotically to generate the primary saliva which is modified by the ductal epithelia. An elevation in cytosolic [Ca²⁺] following neurotransmitter release plays a central role in the stimulation of fluid secretion from salivary gland acinar cells (Ambudkar, 2011; Ambudkar, 2016; Gallacher and Petersen, 1983; Putney, 1982; Weiss et al., 1982). Acetylcholine (ACh) is the primary neurotransmitter released following parasympathetic input to the major salivary glands after gustatory and olfactory stimulation (Matsui et al., 2000). ACh results in activation of muscarinic receptors, G-protein coupled stimulation of inositol 1,4,5 trisphosphate (IP₃) production and subsequent Ca²⁺ release from endoplasmic reticulum stores via IP₃ receptors (IP₃R) (Futatsugi et al., 2005; Lee et al., 1997; Pages et al., 2019). The ER stores are subsequently refilled by store operated Ca²⁺ entry across the PM, which is required for sustained salivary secretion (Cheng et al., 2011a; Cheng et al., 2011b; Melvin et al., 1991; Takemura et al., 1989). To stimulate saliva flow, the primary effector of the increase in [Ca²⁺] are TMEM16a channels localized in the apical PM (Romanenko et al., 2010a; Arreola et al., 1996a; Arreola et al., 1996b; Begenisich and Melvin, 1998). As Cl^- moves across the apical PM, the membrane potential approaches the equilibrium potential for Cl^-; however, Ca²⁺ also increases the activity of K channels, conventionally thought to be localized predominantly at the basolateral PM, to maintain the membrane potential to favor continued Cl^- exit and secretion (Maruyama et al., 1983; Nakamoto et al., 2008; Nakamoto et al., 2007; Romanenko et al., 2007; Romanenko et al., 2010b).

Over the past several decades, the spatiotemporal properties of the agonist-stimulated Ca²⁺ signal have been described in preparations of enzymatically isolated preparations of salivary gland acinar cells or lobules (Park et al., 2012; Warner et al., 2008) loaded with fluorescent Ca²⁺ indicators and imaged by time resolved wide field or confocal microscopy. In total, these studies indicate that the agonist stimulated Ca²⁺ signals exhibit characteristics which appear ideally tuned to activate these spatially separated Ca²⁺-dependent ion channels required for secretion. For example, stimulation with muscarinic agonists, or more directly by photolysis of caged-IP₃, invariably results in the Ca²⁺ signal initiating in the extreme apical ER followed by a rapid Ca²⁺ wave which results in globalization of the signal throughout the cytoplasm (Bruce et al., 2004; Bruce et al., 2002; Giovannucci et al., 2002; Larina and Thorn, 2005; Tojyo et al., 1997; Won et al., 2007). The initiation sites, termed the ‘trigger zone’, correspond to the localization of the majority of type-2/3 IP₃R (Lee et al., 1997; Pages et al., 2019), in ER in close apposition to TMEM16a in the PM. The intimate association of apical ER and apical PM is emphasized by the apparent co-localization of IP₃R and TMEM16a by confocal microscopy, despite their localization in different membranes. Super-resolution STED imaging does resolve distinct distributions of the two proteins, such that the distance between apical ER and PM is estimated to be ~50–100 nm (Sneyd et al., 2021) and thus constitutes a region of privileged communication between messenger and effectors. The subsequent global spread is consistent with Ca²⁺-induced Ca²⁺ release through IP₃R and ryanodine receptors (RyR) (Won et al., 2007) to allow the Ca²⁺ signal to efficiently activate the majority of K channels present on the basolateral PM. At low agonist concentrations, the cycle of apical Ca²⁺ release, global spread and subsequent Ca²⁺ sequestration is repeated, such that temporally, the Ca²⁺ signal is manifested as a series of oscillations with a relatively slow period (5–8/min) (Bruce et al., 2002). At higher concentrations of agonist, the initial globalization is maintained to produce a ‘peak and plateau’ type temporal response.

While in vitro studies of Ca²⁺ signaling have provided a wealth of information to greatly inform our models of salivary secretion, there are general caveats which apply to these data. Primarily, the cells are studied out of a physiological context with no extracellular matrix, basement membrane and with probable disruption of cellular junctions as a result of the enzymatic isolation. A further complication is that isolated cells are typically exposed for a period of time to a constant concentration of agonist which in many cases is a non-hydrolysable analog, as opposed to the neural delivery of transmitter occurring in a pulsatile fashion that is subject in vivo to enzymatic degradation. Whether these factors influence the spatiotemporal characteristics of Ca²⁺ signals in salivary glands is unknown. Additionally, since only indirect measurements of fluid secretion inferred from volume changes can be made in isolated acinar cell preparations (Teos et al., 2016), it is unclear how the concentration-dependent characteristics of stimulated Ca²⁺ signals in vitro relate to the physiological stimulation of fluid secretion in vivo.

In this study, we generated mice expressing a genetically encoded Ca²⁺ indicator specifically in exocrine acinar cells and imaged the Ca²⁺ signals in submandibular glands in living mice by intravital multiphoton (MP) microscopy. Notably, stimulation of the endogenous neural input to the gland induces Ca²⁺ signals with major spatiotemporal properties which are distinct from those observed in isolated acinar cells. Specifically, following moderate stimulation, apically localized Ca²⁺ signals are generated which did not propagate globally. Furthermore, the apical signals consist of rapid oscillations. At stimulation intensities optimum for secretion, a significant apical-basal standing Ca²⁺ gradient was also established. These striking disparities suggest a modification of our previous model is necessary (Pages et al., 2019; Sneyd et al., 2014) to incorporate the characteristics of physiological Ca²⁺ signals in vivo and thus provide a revised framework to better understand salivary fluid secretion.

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) (Chen et al., 2013) floxed by a STOP cassette, were crossed with heterozygous tamoxifen-inducible Mist1 Cre mice (B6.129-Bhlha15^{tm3(cre/ERT2)Skz}/J) (Maruyama et al., 2016). The STOP codon was subsequently excised following tamoxifen gavage (Figure 1A). The expression of GCaMP6f was investigated in vivo in anaesthetized mice following surgery to expose the submandibular salivary glands (SMG). One SMG was lifted and placed on a small platform with a coverslip placed on top of the gland (Figure 1B). Bipolar stimulation electrodes were inserted to the duct bundle connecting to the imaged SMG tissue, in order to stimulate nerves innervating the SMG. A cover glass was placed on top of the SMG to gently hold the tissue and to maintain a saline solution bathing the gland. MP imaging in serial z sections through the gland revealed expression of GCaMP6f exclusively in acinar cells. GCaMP6f expression was uniformly observed in essentially all acinar cells while GCaMP6f expression was not evident in any other cell type such as blood vessels, nerves, myoepithelial cells or ductal cells. As expected by the size of GCaMP6f, the probe was largely excluded from nuclei. Large voids with lack of expression of GCaMP6f were evident in z planes distal from the surface of the gland likely representing the localization of ductal structures (Figure 1C).

Figure 1

Download asset Open asset

We next monitored changes in GCaMP6f fluorescence by intravital microscopy in live animals using MP excitation. Images were acquired at 30 frames per second (fps) as described in Materials and methods (each image a three frame average to result in acquisition at 10 fps). Under basal conditions, prior to stimulation, the cells were quiescent (Figure 2B). Electrical stimulation at various stimulus strengths delivered to nerves innervating the SMG (1–100 Hz, 12 s initiated at 10 s) induced increases in intracellular [Ca²⁺] in acinar cells (Figure 2A,B). Following 1 Hz stimulation, a minority of the cells exhibited small, rapid changes in fluorescence, (n = 4 fields from four animals, Figure 2A,B), while most cells were refractory. Following higher frequency stimulation (3–10 Hz), the majority of the cells in the field responded and the amplitude of the [Ca²⁺] signal increased as the stimulation strength was increased (Figure 2A,B). During the 12 s of stimulation, overall fluorescence gradually increased and immediately returned to baseline after the termination of the stimulation (Figure 2A,B). The maximum [Ca²⁺] signal evoked during the 12 s stimulation progressively increased and was maintained from 1 to 10 Hz throughout the period of stimulation, but this trend did not hold following higher frequency stimulation (n = 4 fields from four animals, Figure 2C). Following stimulation at 30–100 Hz, the [Ca²⁺] increase was not maintained for the duration of stimulation. Under these conditions, the [Ca²⁺] rapidly peaked in the initial 1–3 s then decreased thereafter (Figure 2A,B), probably as a result of exceeding the maximal firing rate of the neurons innervating the SMG. Of note, the stimulation likely led to increased neural activity specifically, rather than stimulating acinar cells directly, as a consequence of current leak through the saline depolarizing nerve endings, as the contralateral SMG were unresponsive to even the highest 100 Hz nerve bundle stimulation of the ipsilateral gland (p > 0.1 compared to without stimulation, n=4 fields from four animals, Figure 2C).

Figure 2

Download asset Open asset

Figure 2—source data 1

Source data associated with Figure 2B and C.

https://cdn.elifesciences.org/articles/66170/elife-66170-fig2-data1-v2.csv

Download elife-66170-fig2-data1-v2.csv

To further confirm the physiological consequence of nerve stimulation, we monitored saliva secretion under identical stimulus conditions. There was no significant secretion activated by 1–3 Hz nerve stimulation for 1 min (p > 0.05 compared to the absence of nerve stimulation, n = four animals), but saliva secretion progressively increased following 5–10 Hz nerve stimulation (p < 0.05 compared to the absence of stimulation, n = four animals) (Figure 3A). Higher frequency stimulation did not result in augmented secretion, likely reflecting failure to sustain the rise in [Ca²⁺] (Figure 2A). At moderate stimulation intensity (0–10 Hz), there was a linear correlation between the peak [Ca²⁺] increases by stimulation and saliva secretion (r² = 0.995, n = four animals, Figure 3B). These stimulus paradigms (1–10 Hz), therefore likely represent the range of physiological stimulation of both evoked Ca²⁺ signals and the resulting saliva secretion and thus characterization of these signals, represent the focus of the remainder of the study.

Figure 3

Download asset Open asset

Figure 3—source data 1

Source data associated with Figure 3A and B.

https://cdn.elifesciences.org/articles/66170/elife-66170-fig3-data1-v2.csv

Download elife-66170-fig3-data1-v2.csv

Sustained saliva secretion requires both Ca²⁺ release and subsequent Ca²⁺ influx (Ambudkar, 2011; Cheng et al., 2011b; Ambudkar et al., 2017). While the former is dependent on IP₃-induced Ca²⁺ release, the latter phase is ultimately dependent on Orai based channels (Hong et al., 2011). To provide evidence that similar mechanisms operate in vivo, experiments were performed stimulating glands at 10 Hz for a more extended period of time (~5 min) and evaluating the effects of an Orai channel blocker, GSK7975A (Zhang et al., 2020) on the evoked Ca²⁺ signals. Stimulation resulted in an initial peak in [Ca²⁺]_i which decayed to a new plateau level which was maintained throughout the 5 min of stimulation (Figure 4A, n = four animals). Animals were subsequently injected with GSK7975A (20 mg/kg ip) and restimulated 20–70 mins later. No significant effect was observed on the magnitude of the initial peak, presumably because sufficient refilling of the ER Ca²⁺ store had occurred during exposure to blocker. However, the plateau measured following 5 min of stimulation was significantly attenuated in drug injected animals vs. vehicle (Figure 4A, pooled data in Figure 4B/C) suggesting a requirement for Orai channel dependent influx for sustained Ca²⁺ signals and secretion.

Figure 4

Download asset Open asset

We next investigated the characteristics of Ca²⁺ signals evoked by stimulation over this physiological range. Following nerve stimulation at low stimulation strengths (1–5 Hz), we observed both the recruitment of responding cells and an increase in the amplitude of the [Ca²⁺] response (Figure 5). To determine more closely how the increased stimulation intensity impacts the characteristics of Ca²⁺ responses to result in increased saliva secretion, a software-based work-flow was designed to post-process the image series. Imaged fields were sub-divided into 8 by 8 equal quadrants to yield 64 grid squares (Figure 5C,D). Each grid was 32 x 32 µm, and typically included a portion of an acinus consisting of 1–5 cells and was used to approximate the spatial and temporal behavior of individual cells or lobules in the field. In order to analyze the activity in each region, the image was then processed to show percent ΔF/F for each time-lapse frame in the sub-divided regions (Figure 5D,E). Using this general scheme, we compared information obtained from 8 x 8 grids, 16 x 16 grids and randomly manually selected regions of interest representing either acinar clusters or single acinar cells. Each scheme produced essentially equivalent data, ultimately validating the use of the 8 x 8 grid (Figure 5—figure supplement 1). At 1 Hz stimulation, the activity was sporadic, with the responding regions tending to give multiple responses while non-responding regions remained quiescent throughout the stimulus period (Figure 5F). Overall, 34.0 ± 5.1% (n = 19 fields from eight animals) of the sub-regions showed significant responses during the stimulation (Figure 5G). Moreover, the increase in [Ca²⁺] in the responding regions exhibited a prolonged latency prior to the initial rise following stimulation (6.72 ± 0.79 s, n = 19, Figure 5H). Stimulation at higher frequencies progressively recruited more grid regions and shortened the latency to the initial response (Figure 5G,H). At 10 Hz, virtually all regions responded to the stimulation (98.5 ± 0.6%, n = 19) with minimal latency (1.70 ± 0.27 s, n = 19) (Figure 5G,H). These data suggest that the increase in overall [Ca²⁺] and secretion by rising stimulation strength was at least in part due to the recruitment of responding cells and a faster time for cells to respond. Further, the amplitude of the maximal increase in [Ca²⁺] in the subdivided grids also increased as the strength of the stimulation was augmented (Figure 5I), indicating that enhancement of individual cellular [Ca²⁺] responses also contribute to the overall increase of [Ca²⁺] amplitudes.

Figure 5 with 1 supplement see all

Download asset Open asset

Figure 5—source data 1

Source data associated with Figure 5G–I.

https://cdn.elifesciences.org/articles/66170/elife-66170-fig5-data1-v2.csv

Download elife-66170-fig5-data1-v2.csv

We observed that increased stimulus strength resulted in both a higher number of responding regions and a higher peak [Ca²⁺] amplitude in individual grids. We next examined whether grids which responded at lower stimulus intensity also strongly responded to higher frequency stimulation. Grids in an imaging field were ranked based on the peak [Ca²⁺] evoked by 5 Hz stimulation and then re-sorted based on progressively increasing amplitude (Figure 6A,B). We subsequently sorted the data obtained from 1, 3, and 10 Hz stimulation from the identical field in the grid rank order established for 5 Hz stimulation (Figure 6C). This analysis demonstrated a clear pattern whereby, at each stimulus strength, particular regions of the field of view were most susceptible to stimulation, such that grids giving the largest amplitude response at 1 Hz similarly responded by displaying the largest amplitude signals at 3, 5 and 10 Hz stimulation. (n = 5 fields from four animals) (Figure 6D).

Figure 6

Download asset Open asset

Figure 6—source data 1

Source data associated with Figure 6D–G.

https://cdn.elifesciences.org/articles/66170/elife-66170-fig6-data1-v2.csv

Download elife-66170-fig6-data1-v2.csv

The extent of the increase in [Ca²⁺] following greater strengths of stimulation was not uniform throughout individual grids. High-responding grid regions showed linearly enhanced increases following stimulation (R² = 0.9197, n = 4 fields from four animals), while the increase in Ca²⁺ signal following increased stimulation in less responsive regions was proportionately smaller (Figure 6E). This trend was not limited to the peak amplitude attained. During the 12 s stimulation, grid regions exhibited [Ca²⁺] fluctuations superimposed upon an elevated plateau (Figure 6F). We evaluated the extent of this fluctuation by comparing the standard deviation of the ΔF/F signal occurring during the stimulation period. With this analysis, a higher standard deviation reflects greater variability in the stimulated change in [Ca²⁺]. When the 5 Hz data was sorted according to the peak [Ca²⁺] ranking as in Figure 6D, the variation of grids with 5 Hz stimulation also demonstrated that low-responding grids had smaller variability than high-responding grids (n = 4 fields from four animals) (Figure 6F). Likewise, the same trend was exhibited at each stimulation strength, with a linearly increasing rate of enhancement as intensity increased (R² = 0.8664, n = 4) (Figure 6G). Therefore, an increased fluctuation in the response is also a property of sensitive, highly responsive regions.

Stimulation of salivary fluid secretion is predominantly mediated through parasympathetic input as a result of the action of ACh released from nerve endings acting on muscarinic M3 receptors on acinar cells (Matsui et al., 2000). To gain direct evidence that neural release of ACh was responsible for the changes in [Ca²⁺] observed, we systemically administered a cholinesterase inhibitor to block the clearance of ACh and effectively increase the local [ACh]. Physostigmine (0.1 mg/kg body weight IP) enhanced the peak amplitude of the Ca²⁺ signals following stimulation (Figure 7A–D), indicating that acinar cell [Ca²⁺] changes reflect the extracellular availability of acetylcholine released by stimulation. The augmentation in the Ca²⁺ signal was not statistically significant at 1 Hz stimulation (p > 0.05, n = 5 fields from four animals), but became more prominent at 3 Hz with shorter latency, an increase in the number of responding grids and peak amplitude (p < 0.05 in % response and peak ΔF/F, n = 5) and following 5 Hz stimulation (p < 0.05 in latency and peak ΔF/F, n = 5 fields from four animals) (Figure 7E–G). At 10 Hz stimulation, no further increase in the grid responsiveness or latency was observed, reflecting the fact that all grids were active and exhibited minimal latency at this stimulus strength, however the peak ΔF/F showed a robust enhancement (p < 0.01, n = 5) (Figure 7E–G). While these data do not rule out the involvement of additional neurotransmitters, they confirm a prominent role for ACh in mediating the increase in [Ca²⁺] observed following neural stimulation.

Figure 7

Download asset Open asset

Figure 7—source data 1

Source data associated with Figure 7E–G.

https://cdn.elifesciences.org/articles/66170/elife-66170-fig7-data1-v2.csv

Download elife-66170-fig7-data1-v2.csv

While the previous grid-based analysis yielded important information relating to the general properties of the evoked signals within an imaged field, this approach was not suited to provide spatial information on a sub-acinus or lobule scale. To interrogate the subcellular spatial characteristics of the Ca²⁺ signals evoked by neural stimulation, we generated standard deviation (SD) images from the stimulated portion of the image series. This analysis revealed that at all physiological stimulus strengths the increase in [Ca²⁺]_i was highly heterogeneous, with the largest increases in signal occurring in the apical aspects of acinar clusters, directly below the apical PM (Figure 8—figure supplement 1). Figure 8A illustrates a field in which an acinar cluster is highlighted prior to neural stimulation at 5 Hz (red box and image series in Figure 8B). In marked contrast to in vitro experiments, analysis of the image series demonstrated that the signal consisted of brief (< 2 s in duration), repetitive increases in [Ca²⁺] (Figure 8B). The signals were highly localized, being initiated in, and almost entirely confined to the apical portions of individual cells, within a few μms of the apical PM (Figure 8B–E), with only very rare, small elevations in the basal aspects of the cell. This localization is illustrated by SD images of the whole field (Figure 8D) and a topographical SD representation of the highlighted cluster (Figure 8D) together with analysis of the fluorescent changes in time from regions of interest manually assigned to an apical Region of Interest (ROI) (red box), cytoplasmic ROI (blue box), and a basal ROI (green box) (Figure 8E). Consistent with tight coupling between acinar cells by gap junctions, signals often appeared to invade the apical region of neighboring cells (see images time stamped between 30.8 and 31.8 s in Figure 8B).

Figure 8 with 1 supplement see all

Download asset Open asset

At 10 Hz stimulation, a rise in [Ca²⁺] was also rapidly initiated in the extreme apical ER, close to the apical PM (imaging field in Figure 9A with red box showing the highlighted cluster depicted in Figure 9B). In contrast to lower stimulus strengths, Ca²⁺ signals were observed in the basal aspects of the cell (Figure 9B–E), but of considerably lower amplitude, approximately 20–30% of that observed in the apical domain. Effectively a standing apical-basal [Ca²⁺] gradient was established, as shown in the SD image of the field and the topographical representation of the SD of the highlighted region (Figure 9C/D), together with the analysis of manually positioned representative apical (red ROI), cytoplasmic (blue ROI) and basal regions (green ROI) (Figure 9E) to illustrate the magnitude of the gradient.

Figure 9

Download asset Open asset

To further investigate this spatial heterogenicity, we performed MP line scans following stimulation at 3–10 Hz stimulation where the scanned line was orientated to interrogate changes in fluorescence originating in the apical region extending through the nucleus to the base of an individual cell. Figure 10A shows a representative field with the position of the scanned line. Figure 10BI,CI and DI show the changes in line fluorescence over time evoked by 3, 5, 10 Hz stimulation, respectively. Consistent with the previous data, following 3 and 5 Hz stimulation the changes in fluorescence were largely confined to regions close to the apical PM of the cell (defined as a 3 μm line in the kinetic from apical, cytoplasmic, and basal ROIs shown in Figure 10BII and CII and pooled data in Figure 10E/F). Stimulation at 10 Hz again resulted in large amplitude apical signals which diminished rapidly, such that the magnitude of the signal was ~25% of the apical signal in the extreme basal aspects of the cell distal to the nucleus (Figure 10DI, kinetic in 10 D_II and pooled data from 11 line scans from 3 animals in 10E/F). The apical-basal gradient was further analyzed by plotting the amplitude of the maximum fluorescent change across the line profile which further reinforces that at each stimulus intensity the signal rapidly decays away from the apical initiation sites to the nucleus and is ~25% of the apical signal in the extreme base of the cell (Figure 10. B_III, C_III and D_III). In total, these data demonstrate that in contrast to experiments performed in in vitro preparations, physiological stimulation results in prominent, apical Ca²⁺ signals which do not propagate and globalize to result in a homogeneous [Ca²⁺]_i throughout the cell, such that steep apical-basal gradients are established.

Figure 10

Download asset Open asset

Figure 10—source data 1

Source data associated with Figure 10E–F.

https://cdn.elifesciences.org/articles/66170/elife-66170-fig10-data1-v2.csv

Download elife-66170-fig10-data1-v2.csv

To further analyze the neurally evoked apical Ca²⁺ signals, Python software scripts were written in the Jupyter lab notebook environment to automate the designation of apical ROIs from masks generated from average or SD images. These apical ROIs were then subsequently used to quantitate the magnitude and frequency of signals within these regions (see Materials and methods for details). Figure 11A shows an imaging field depicting the stimulated-basal average image for the field generated during the 12 s of stimulation at 5 Hz from which 79 apical ROIs were generated from masks shown in Figure 11B/C. Figure 11D shows representative responses from 4 of the 79 ROIs to stimulation over the range of physiological stimulus strengths. Consistent with the previous grid analysis, few apical regions responded to 1 Hz stimulation while the majority of ROIs responded to stimulation at 5–10 Hz (see Videos 1–4). At both 3 and 5 Hz stimulation, individual apical ROIs typically responded by generating a rapid train of oscillations. A peak counting algorithm was used to analyze the frequency of the oscillations for each ROI. This demonstrated that the frequency of oscillations was markedly faster (~1 Hz) when measured in vivo than previously described in isolated cells (~0.2 Hz, this study, Figure 12) and further modestly increased between 3 and 5 Hz stimulation (Figure 11E; 79 ROIs experiment shown in Figure 11A–C and pooled data from 13 individual fields from five animals (Figure 11F)). In contrast to the robust correlation between saliva secretion and the magnitude of the evoked Ca²⁺ signal, the relationship between secretion and the frequency of Ca²⁺ oscillation was less strong (R² = 0.824, Figure 11G).

Figure 11

Download asset Open asset

Figure 11—source data 1

Source data associated with Figure 11E and F.

https://cdn.elifesciences.org/articles/66170/elife-66170-fig11-data1-v2.csv

Download elife-66170-fig11-data1-v2.csv

Figure 12 with 1 supplement see all

Download asset Open asset

Video 1

Download asset

Video 2

Download asset

Video 3

Download asset

Video 4

Download asset

Although GCaMP6f and fura-2 have similar affinities (Badura et al., 2014; Lewis and Cahalan, 1989), a possibility exists that the marked differences in the spatiotemporal properties of Ca²⁺ signals observed in vivo vs. reported in vitro are a function of the genetically encoded indicator or alternatively because the vitro studies were largely performed at room temperature. We therefore prepared small acinar clumps from Mist1^+/– x GCaMP6f^+/– animals by enzymatic digestion and monitored Ca²⁺ signals stimulated by the muscarinic agonist carbachol (CCh) by wide field imaging at 37°C. Individual acinar cells failed to respond to 5 nM CCh; however, at 25 nM CCh single cells within an acinus responded by evoking repetitive Ca²⁺ transients (Figure 12) with a frequency ~0.2 Hz. The period of the transients tended to increase with increasing CCh concentrations (Figure 12B). In response to increasing concentrations of CCh, the initial peak increased (Figure 12C, n=14–32 cells, three animals) and the oscillations dampened to generate an elevated plateau. Oscillations evoked even by threshold concentrations of CCh were initiated below the apical PM and invariably globalized with all CCh concentrations (Figure 12D–F and Video 5). In experiments performed at room temperature in GCaMP6f expressing cells (~23°C), the spatiotemporal characteristics of the signal were identical to those previously reported with chemical indicators (Figure 12—figure supplement 1A–C). The peak [Ca²⁺]_i evoked by CCh stimulation was similar at 37°C (Figure 12—figure supplement 1D), while the oscillation frequency was significantly faster than observed at room temperature (Figure 12—figure supplement 1E). In total, the spatiotemporal profile of the Ca²⁺ responses evoked in acutely isolated GCaMP6f expressing acini were strikingly similar to numerous reports documenting agonist-evoked Ca²⁺ responses with conventional chemical Ca²⁺ indicators and thus the properties of the genetically encoded indicator at physiological temperature per se do not explain the differences observed.

Video 5

Download asset

We have previously constructed a series of mathematical models to help understand and explain our experimental results and those from other groups (Pages et al., 2019; Sneyd et al., 2014; Almássy et al., 2018; Sneyd et al., 2017a; Sneyd et al., 2017b; Vera-Sigüenza et al., 2019). Briefly, these multiscale models were based on cyclical global increases in [Ca²⁺] activating spatially separated ion channels required for sustained secretion in a 3D collection of seven acinar cells forming an acinus. In these older models, apical initiation of the Ca²⁺ signal resulted in activation of Cl^- flux through TMEM16a Cl^- channels and the Ca²⁺ signal was subsequently transmitted across the cell cytoplasm by a process of Ca²⁺-induced Ca²⁺ release (via RyR or IP₃R). This led to periodic increases of [Ca²⁺] in the basal regions of the cell and hence activation of basal Ca²⁺ activated K⁺ (KCa) channels. Activation of the KCa channels is critical for maintenance of the electrochemical potential driving the efflux of Cl^- and thus secretion. However, the spatiotemporal information from the current in vivo data is obviously not compatible with these older models. We have therefore constructed a new model, as shown in Figure 13. The apical region is defined as the domain within 3 μm of the apical PM, which is itself demarcated by the tight junctions. This region now includes all the components necessary for saliva secretion, with KCa channels and Na/K ATPases in addition to the TMEM16a channels (Figure 13) which are poised to respond to Ca²⁺ released from the apical ER, from IP₃R that are situated in close proximity to the TMEM16A (Pages et al., 2019). We emphasise that, in the model, KCa channels and Na/K-ATPases are also present in the basal lateral membranes, and therefore the secretory machinery is not restricted to the apical PM. A more detailed description of the model can be found in methods and in Pages et al., 2019; Vera-Sigüenza et al., 2019; Vera-Sigüenza et al., 2020. The structure of the three-dimensional cell upon which our model is solved is shown in Figure 14—figure supplement 1, from where it can be seen that the apical membrane is a fingered structure that partially wraps around the cell, whereas the basal membrane (including the lateral membrane) extends right to the edge of the apical membrane.

Figure 13

Download asset Open asset

Figure 14A shows the average apical [Ca²⁺] for a range of values of V_PLC in the new model. V_PLC controls the rate of production of IP₃, and is thus a proxy for frequency of neural stimulation. Simulations with a pulsatile V_PLC (to mimic oscillatory neural stimulation) give identical qualitative results. As V_PLC increases, the [Ca²⁺] oscillation frequency increases and the latency decreases. At high stimulation levels, the [Ca²⁺] oscillation frequency is approximately 0.5–1 Hz, and the oscillations are superimposed on an increasing baseline, which is due to a decreasing cell volume as a result of saliva secretion. These features all agree well with experimental data. One aspect of the data not reproduced by the model is the appearance of very low frequency baseline [Ca²⁺] spikes at the lowest stimulation frequencies. These baseline spikes are highly likely to be due to stochastic properties of the IP₃R, as described by Falcke and others (Thurley et al., 2012; Thurley et al., 2014; Thurley et al., 2011). Our model contains no stochastic components and is therefore, a priori, unable to reproduce such low-frequency baseline spiking.

Figure 14 with 1 supplement see all

Download asset Open asset

Figure 14, panels B-D compare the new model with two older versions in which there were no apical KCa channels, but a Ca²⁺ wave was propagated from the apical ER, throughout the cell, thus allowing for activation of basal KCa. Figure 14B shows saliva secretion (with VPLC = 0.008 μM/s) in the new model (red curve) compared with the old model of Vera-Sigüenza et al., 2020 (green curve), which has no apical PM KCa but does exhibit a propagated [Ca²⁺] signal from apical ER throughout the cell, and also compared to a version of the old model (blue curve) which has neither apical KCa nor a propagated [Ca²⁺] wave. Note that the old model with no apical KCa is quantitatively different from the new model with no apical KCa. Fluid secretion in the new model with no apical KCa is slightly greater than that in the old model with no KCa, and is shown in panel F. These simulations show that absence of both a propagated [Ca²⁺] wave and apical KCa channels significantly decreases saliva secretion. The associated [Ca²⁺] traces, averaged from the apical and basal regions, as well as from the cytoplasmic region of the cell (representing a domain between these regions), are shown in Figure 14, C, D, using the same color scheme as in Figure 14B. In the new model (Figure 14C), basal [Ca²⁺] is between 25% and 35% of apical [Ca²⁺], as observed experimentally, while in the model of Vera-Sigüenza et al., 2020; Figure 14D basal [Ca²⁺] is a significantly greater fraction of apical [Ca²⁺]. The [Ca²⁺] traces for the model with neither apical KCa nor a propagated wave are very similar to those shown in Figure 14C, and so are not shown explicitly. The intracellular spatial gradient of [Ca²⁺] is illustrated by the model line scan shown in Figure 14E, which also agrees well with the experimental line scans, in that Ca²⁺ is largely restricted to the apical region of the cell. The experimental data show more spatial inhomogeneity (due to the presence of the nucleus and the mitochondria), features which are not included in the model.

It is important to note that, in the model, of the basal region, includes the lateral PM and extends right to the edge of the apical region, which is a thin, fingered region. Thus, even in the absence of a propagated [Ca²⁺] wave, diffusion of Ca²⁺ from the apical region will still activate a fraction of the KCa channels. This is illustrated in Figure 14F, in the new model and for a range of stimulation levels, which shows how saliva secretion depends on the relative density of KCa channels in the apical PM. Even when there are no apical KCa channels (i.e. a relative density of 0), diffusion of Ca²⁺ from the apical region can activate those basal KCa channels which are situated close to the apical membrane, and this is sufficient to cause significant fluid flow. Furthermore, consistently with previous theoretical work (Palk et al., 2012) there is an optimal density of apical KCa channels; if the apical membrane contains too high a fraction of the total number of KCa channels then fluid flow is hampered. Optimal fluid secretion occurs when the apical and basal KCa densities are approximately equal.

An increase in [Ca²⁺] following neurotransmitter release from parasympathetic neurons is fundamentally important for the stimulation of salivary fluid secretion. It is also recognized that the spatiotemporal properties of intracellular Ca²⁺ signals in exocrine cells are pivotal for appropriately activating ion channels necessary for the underlying process (Almassy et al., 2012; Kasai and Augustine, 1990; Kidd and Thorn, 2000). Despite a wealth of information documenting agonist-stimulated Ca²⁺ signals, recorded in dissociated single cells, isolated acini or excised lobules, the properties of physiological Ca²⁺ signals in vivo following endogenous neural stimulation have not been reported. We therefore generated transgenic mice that express a genetically encoded Ca²⁺ indicator specifically in exocrine acinar cells and used MP microscopy in live mice to document the Ca²⁺ signals generated following stimulation of the submandibular nerve which innervates the SMG. By establishing stimulation parameters optimum for fluid secretion, we have inferred the spatiotemporal properties of Ca²⁺ signals that ultimately underlie the physiological stimulation of fluid secretion at the lobule level and subsequently at subcellular resolution.

Following threshold stimulation, a small proportion of the imaged field responded after a prolonged latency by eliciting a single, or a few brief transients. These Ca²⁺ spikes were initiated in the extreme apical ER, consistent with the localization of the majority of IP₃R and previous in vitro data (Lee et al., 1997; Pages et al., 2019; Won et al., 2007). In stark contrast however, these signals did not globalize and were confined to a few μms distant from the apical PM. This threshold stimulation resulted in minor measurable fluid secretion. At greater stimulus strengths, where secretion was readily evident, an increasingly larger proportion of the field was recruited with shorter latency to yield multiple Ca²⁺ transients with increasing peak magnitude. The increase reflecting an [Ca²⁺]_i rise in all cells within individual responding acini. Strikingly, these signals were again predominantly localized to regions immediately juxtaposed to the apical PM with limited propagation toward the basal aspects of the cell. Upon optimal stimulation for secretion, essentially the entire field was recruited to produce a sustained increase with periodic fluctuations superimposed on the plateau after a minimal latent period. Notably, while the affinity of the indicator may have masked a greater magnitude of apical Ca²⁺ rise, or more prominent fluctuations in the signal at these higher stimulus strengths, a prominent apical to basal gradient was readily evident, such that the signal in the basal region reached only ~25–30% of that in the apical region. Further increases in the intensity of the stimulation resulted in a rapid peak followed by a gradual waning of the signal and consistent with the absence of a sustained Ca²⁺ signal, the rate of fluid secretion was diminished. This reduced secretion likely results from a ‘tetanic’ stimulus where either neurons fail to fire during their refractory period, neurotransmitter release is exhausted or acinar cell Ca²⁺ signaling is perturbed, during this likely non-physiological stimulation. Over the range of physiological stimulus strengths, the extent of fluid secretion was strongly correlated with the magnitude of the peak Ca²⁺ rise, rather than necessarily the frequency of Ca²⁺ transients. This conclusion is consistent with our previous modeling and experimental work that has indicated that the integrated increase in [Ca²⁺] during stimulation is the primary driving factor for efficient fluid secretion (Sneyd et al., 2014).

Local, non-propagating Ca²⁺ signals are observed in isolated pancreatic acinar cell in response to IP₃, low concentrations of secretagogues and physiological concentrations of the gut hormone cholecystokinin (Ashby et al., 2003; Thorn et al., 1993; Thorn et al., 1996; Thorn and Petersen, 1993). It should be noted that the majority of these studies have been performed in whole-cell patch clamped isolated acinar cells. Ca²⁺ buffering is established by the internal pipette solution in these experiments and thus could conceivably alter the spatial characteristics of the signal. Future in vivo experiments in the pancreas will address whether the spatial properties of Ca²⁺ signals differ from those observed in acutely isolated cells. Nevertheless, the spatial characteristics of the Ca²⁺ signal observed in vivo in the present study represents a major difference to previous work in isolated salivary acinar cells. The failure of the signal to propagate and standing [Ca²⁺] gradients observed are presumably established by SERCA, PMCA and mitochondrial buffering to efficiently reduce the local [Ca²⁺] to levels that do not support substantial CICR through IP₃R and RYR that are presumed to be localized in the ER distant from the apical trigger zone. An obvious question is, what factors or mechanisms underlie these differences? At the level of the SMG lobule, the polarity of individual cells and their organization within an acinus is vital for function. The complex architecture is maintained by junctional complexes and interactions with the extracellular matrix and stromal cells (Baker, 2010). Clearly, these interactions may be disturbed when cells are isolated from an excised gland by enzymic digestion. An apparent consequence may be a disruption of the exquisite polarity of acinar cells. Perturbed localization of the Ca²⁺ signaling machinery could disrupt the balance between Ca²⁺ release and clearance to ultimately result in an alteration of the spatiotemporal characteristics of the signal. Conceivably, alterations in the polarization of the ER, such that IP₃R or SERCA are not maintained in the extreme apical ER (Pages et al., 2019), or a change in distribution of mitochondria (Bruce et al., 2004) to disrupt ER-mitochondrial junctions and effective Ca²⁺ sequestration from release sites could contribute to the altered Ca²⁺ signals observed in vitro.

A further important consideration when interpreting the characteristics of these Ca²⁺ signals is the obvious difference in delivery of the secretagogue in these experiments. In contrast to superfusion of agonists in vitro, that results in a defined ‘square’ pulse of stimulating agent, in vivo, acinar cells likely experience a complex, stimulus intensity, and time-dependent gradient of neurotransmitter. In our studies, the concentration of ACh released from neurons would be predicted to rapidly increase and then decay as it is hydrolyzed by acetylcholinesterases. As the frequency of stimulation is increased, the residual concentration of ACh present in the vicinity of acinar cells would be predicted to increase with a ‘saw-tooth’ profile as the amount of ACh release exceeds the rate of degradation following more frequent, repetitive cycles of neurotransmitter exocytosis. In this scenario, when a threshold concentration of neurotransmitter is reached that generates sufficient IP₃, an increase [Ca²⁺]_i, will be triggered. The time to reach this threshold would define the latent period for that particular cell and the relative sensitivity to neurotransmitter will predicate whether an individual cell is responsive to a given stimulation. At optimum stimulation frequencies, the residual ACh concentration would further increase and remain above the threshold for the majority of acinar cells to respond. This would facilitate greater Ca²⁺ release manifested as an increased peak and Ca²⁺ oscillations by the entire field. Notably, neural stimulation does not simply ‘pace’ cells, as the frequency of oscillations does not correlate with the stimulus frequency. It is therefore likely, that the oscillations result from the inherent properties of IP₃R and SERCA pumping activity but where the kinetics are influenced by the dynamic changes in neurotransmitter concentrations. This may contribute to the differences in oscillation frequency observed when comparing in vivo and in vitro data.

We observed considerable spatial heterogeneity in the responsiveness and peak amplitude of cells across the field of view. Specifically, there were sections of the field which exhibit the greatest sensitivity and largest responses at all stimulus intensities. Conceptually, this may reflect acini with expression of a higher density of cell surface receptors. Alternatively, highly responsive regions may represent cells physically closer to synapsing neurons and thus experiencing higher concentrations of ACh. The innervation of SMG lobules visualized in a reporter mouse expressing fluorescent protein in parasympathetic neurons has been reported to be fairly homogeneous (Sheu et al., 2017). Therefore, a further possibility is that spatial homogeneity arises in part because increasing numbers of axons are recruited to fire as stimulus intensity increases thus directly innervating a larger proportion of cells.

Clearly, our previous models describing how Ca²⁺ signals promote salivary secretion are not entirely consistent with the new in vivo data presented here. Fundamentally, if there is no obligatory global, propagating intracellular Ca²⁺ wave, activating the KCa channels in the basal PM would be predicted to be inefficient. Previously, however we have demonstrated, both functionally and physically, that the KCa channels, K_Ca3.1 (‘IK’) and K_Ca1.1 (‘BK’) together with Na/K ATPases are present in the apical PM of salivary acinar cells (Almássy et al., 2018; Almassy et al., 2012). Notably, we reported that KCa_3.1 colocalized with apical markers including IP₃R3 and Zona occludens-1. Further, focal, highly localized uncaging of Ca²⁺, directly below the apical PM resulted in activation of a K⁺ conductance with pharmacology consistent with K_Ca3.1. Additionally, apical uncaging of Ca²⁺ evoked a smaller current in acini prepared from K_Ca1.1 knockout transgenic mice (Almassy et al., 2012). At the time these results were puzzling, as there was no obvious reason, according to the current model, why the apical PM should express these transport proteins. Nevertheless, a previous modeling study has suggested that expression of a proportion of KCa to the apical PM could optimize fluid secretion. The new data, together with the localization of apical KCa, combined with simulations by the new model, strongly suggests the physiological importance of the apical KCa and Na/K-ATPases. Specifically, as [Ca²⁺] elevations occur and are maintained at greater magnitude in close apposition to apical KCa channels, it would seem likely that their activation contributes to maintaining the electrochemical driving force for Cl^- secretion. Given the localization of IP₃R to apical ER, ~ 50–100 nm from the apical PM (Sneyd et al., 2021), a tight coupling between Ca²⁺ release and activation of both major effectors would be reasonably predicted to promote efficient fluid secretion. The importance of apical KCa, while difficult to explicitly address experimentally, is supported by the new model which indicates that optimal secretion is achieved when KCa are expressed in both the apical and basolateral membranes. On the other hand, since the primary saliva has a low [K⁺], this K⁺ must be removed from the lumen. This is the proposed function of the Na/K-ATPase in the apical membrane.

Taken together, the present studies describe the spatial and temporal characteristics of physiological Ca²⁺ signals driving salivary fluid secretion. Further, our findings highlight that caution should be exerted in extrapolating conclusions from ex vivo studies to physiological Ca²⁺ signals and function in vivo. Finally, it is envisioned that the present studies will provide a framework for investigating if Ca²⁺ signaling is disrupted in disease states such as Sjögren's syndrome or radiation-induced xerostomia which result in pronounced hyposecretion of saliva (Teos et al., 2016; Baum et al., 2012).

Tamoxifen, corn oil, ketamine, and Xylazine were purchased from Sigma Chemical, St Louis, Mo. Physostigmine was purchased from Tocris Minneapolis, Mn.

All data generated or analysed during this study are included in the manuscript and supporting files. Source data files have been provided for Figures 2,3,5,6,7,10,11.Python scripts are available at https://github.com/jrugis/cell_tools (copy archived at https://archive.softwareheritage.org/swh:1:rev:c8fced9a9d8f3b7526a8f8eabfe547c618f2374b).

1. 1. Almassy J
   2. Won JH
   3. Begenisich TB
   4. Yule DI

   (2012) Apical Ca2+-activated potassium channels in mouse parotid acinar cells

   Journal of General Physiology 139:121–133.

   https://doi.org/10.1085/jgp.201110718

   • PubMed
   • Google Scholar
2. 1. Almássy J
   2. Siguenza E
   3. Skaliczki M
   4. Matesz K
   5. Sneyd J
   6. Yule DI
   7. Nánási PP

   (2018) New saliva secretion model based on the expression of Na⁺-K⁺ pump and K⁺ channels in the apical membrane of parotid acinar cells

   Pflügers Archiv - European Journal of Physiology 470:613–621.

   https://doi.org/10.1007/s00424-018-2109-0

   • PubMed
   • Google Scholar
3. 1. Ambudkar IS

   (2011) Dissection of calcium signaling events in exocrine secretion

   Neurochemical Research 36:1212–1221.

   https://doi.org/10.1007/s11064-011-0465-7

   • PubMed
   • Google Scholar
4. 1. Ambudkar IS

   (2016) Calcium signalling in salivary gland physiology and dysfunction

   The Journal of Physiology 594:2813–2824.

   https://doi.org/10.1113/JP271143

   • PubMed
   • Google Scholar
5. 1. Ambudkar IS
   2. de Souza LB
   3. Ong HL

   (2017) TRPC1, Orai1, and STIM1 in SOCE: Friends in tight spaces

   Cell Calcium 63:33–39.

   https://doi.org/10.1016/j.ceca.2016.12.009

   • PubMed
   • Google Scholar
6. 1. Arreola J
   2. Melvin JE
   3. Begenisich T

   (1996a) Activation of calcium-dependent chloride channels in rat parotid acinar cells

   Journal of General Physiology 108:35–47.

   https://doi.org/10.1085/jgp.108.1.35

   • PubMed
   • Google Scholar
7. 1. Arreola J
   2. Park K
   3. Melvin JE
   4. Begenisich T

   (1996b) Three distinct chloride channels control anion movements in rat parotid acinar cells

   The Journal of Physiology 490:351–362.

   https://doi.org/10.1113/jphysiol.1996.sp021149

   • PubMed
   • Google Scholar
8. 1. Ashby MC
   2. Camello-Almaraz C
   3. Gerasimenko OV
   4. Petersen OH
   5. Tepikin AV

   (2003) Long distance communication between muscarinic receptors and Ca2+ release channels revealed by carbachol uncaging in cell-attached patch pipette

   Journal of Biological Chemistry 278:20860–20864.

   https://doi.org/10.1074/jbc.M302599200

   • PubMed
   • Google Scholar
9. 1. Badura A
   2. Sun XR
   3. Giovannucci A
   4. Lynch LA
   5. Wang SS

   (2014) Fast calcium sensor proteins for monitoring neural activity

   Neurophotonics 1:025008.

   https://doi.org/10.1117/1.NPh.1.2.025008

   • PubMed
   • Google Scholar
10. 1. Baker OJ

    (2010) Tight junctions in salivary epithelium

    Journal of Biomedicine & Biotechnology 2010:278948.

    https://doi.org/10.1155/2010/278948

    • PubMed
    • Google Scholar
11. 1. Baum BJ
    2. Alevizos I
    3. Zheng C
    4. Cotrim AP
    5. Liu S
    6. McCullagh L
    7. Goldsmith CM
    8. Burbelo PD
    9. Citrin DE
    10. Mitchell JB
    11. Nottingham LK
    12. Rudy SF
    13. Van Waes C
    14. Whatley MA
    15. Brahim JS
    16. Chiorini JA
    17. Danielides S
    18. Turner RJ
    19. Patronas NJ
    20. Chen CC
    21. Nikolov NP
    22. Illei GG

    (2012) Early responses to adenoviral-mediated transfer of the aquaporin-1 cDNA for radiation-induced salivary hypofunction

    PNAS 109:19403–19407.

    https://doi.org/10.1073/pnas.1210662109

    • PubMed
    • Google Scholar
12. 1. Begenisich T
    2. Melvin JE

    (1998) Regulation of chloride channels in secretory epithelia

    Journal of Membrane Biology 163:77–85.

    https://doi.org/10.1007/s002329900372

    • PubMed
    • Google Scholar
13. 1. Bruce JI
    2. Shuttleworth TJ
    3. Giovannucci DR
    4. Yule DI

    (2002) Phosphorylation of inositol 1,4,5-trisphosphate receptors in parotid acinar cells. A mechanism for the synergistic effects of cAMP on Ca2+ signaling

    The Journal of biological chemistry 277:1340–1348.

    https://doi.org/10.1074/jbc.M106609200

    • PubMed
    • Google Scholar
14. 1. Bruce JI
    2. Giovannucci DR
    3. Blinder G
    4. Shuttleworth TJ
    5. Yule DI

    (2004) Modulation of [Ca2+]i signaling dynamics and metabolism by perinuclear mitochondria in mouse parotid acinar cells

    Journal of Biological Chemistry 279:12909–12917.

    https://doi.org/10.1074/jbc.M309070200

    • PubMed
    • Google Scholar
15. 1. Cao P
    2. Donovan G
    3. Falcke M
    4. Sneyd J

    (2013) A stochastic model of calcium puffs based on single-channel data

    Biophysical Journal 105:1133–1142.

    https://doi.org/10.1016/j.bpj.2013.07.034

    • PubMed
    • Google Scholar
16. 1. Cao P
    2. Tan X
    3. Donovan G
    4. Sanderson MJ
    5. Sneyd J

    (2014) A deterministic model predicts the properties of stochastic calcium oscillations in airway smooth muscle cells

    PLOS Computational Biology 10:e1003783.

    https://doi.org/10.1371/journal.pcbi.1003783

    • PubMed
    • Google Scholar
17. 1. Chen TW
    2. Wardill TJ
    3. Sun Y
    4. Pulver SR
    5. Renninger SL
    6. Baohan A
    7. Schreiter ER
    8. Kerr RA
    9. Orger MB
    10. Jayaraman V
    11. Looger LL
    12. Svoboda K
    13. Kim DS

    (2013) Ultrasensitive fluorescent proteins for imaging neuronal activity

    Nature 499:295–300.

    https://doi.org/10.1038/nature12354

    • PubMed
    • Google Scholar
18. 1. Cheng KT
    2. Liu X
    3. Ong HL
    4. Swaim W
    5. Ambudkar IS

    (2011a) Local Ca²+ entry via Orai1 regulates plasma membrane recruitment of TRPC1 and controls cytosolic Ca²+ signals required for specific cell functions

    PLOS Biology 9:e1001025.

    https://doi.org/10.1371/journal.pbio.1001025

    • PubMed
    • Google Scholar
19. 1. Cheng KT
    2. Ong HL
    3. Liu X
    4. Ambudkar IS

    (2011b) Contribution of TRPC1 and Orai1 to Ca(2+) entry activated by store depletion

    Advances in experimental medicine and biology 704:435–449.

    https://doi.org/10.1007/978-94-007-0265-3_24

    • PubMed
    • Google Scholar
20. 1. Evans RL
    2. Park K
    3. Turner RJ
    4. Watson GE
    5. Nguyen HV
    6. Dennett MR
    7. Hand AR
    8. Flagella M
    9. Shull GE
    10. Melvin JE

    (2000) Severe impairment of salivation in Na+/K+/2Cl- cotransporter (NKCC1)-deficient mice

    Journal of Biological Chemistry 275:26720–26726.

    https://doi.org/10.1016/S0021-9258(19)61435-3

    • PubMed
    • Google Scholar
21. 1. Foskett JK

    (1990) [Ca2+]i modulation of Cl- content controls cell volume in single salivary acinar cells during fluid secretion

    American Journal of Physiology-Cell Physiology 259:C998–C1004.

    https://doi.org/10.1152/ajpcell.1990.259.6.C998

    • PubMed
    • Google Scholar
22. 1. Futatsugi A
    2. Nakamura T
    3. Yamada MK
    4. Ebisui E
    5. Nakamura K
    6. Uchida K
    7. Kitaguchi T
    8. Takahashi-Iwanaga H
    9. Noda T
    10. Aruga J
    11. Mikoshiba K

    (2005) IP3 receptor types 2 and 3 mediate exocrine secretion underlying energy metabolism

    Science 309:2232–2234.

    https://doi.org/10.1126/science.1114110

    • PubMed
    • Google Scholar
23. 1. Gallacher DV
    2. Petersen OH

    (1983)

    Stimulus-secretion coupling in mammalian salivary glands

    International review of physiology 28:1–52.

    • PubMed
    • Google Scholar
24. 1. Giovannucci DR
    2. Bruce JI
    3. Straub SV
    4. Arreola J
    5. Sneyd J
    6. Shuttleworth TJ
    7. Yule DI

    (2002) Cytosolic Ca(2+) and Ca(2+)-activated Cl(-) current dynamics: insights from two functionally distinct mouse exocrine cells

    The Journal of physiology 540:469–484.

    https://doi.org/10.1113/jphysiol.2001.013453

    • PubMed
    • Google Scholar
25. 1. Hong JH
    2. Li Q
    3. Kim MS
    4. Shin DM
    5. Feske S
    6. Birnbaumer L
    7. Cheng KT
    8. Ambudkar IS
    9. Muallem S

    (2011) Polarized but differential localization and recruitment of STIM1, Orai1 and TRPC channels in secretory cells

    Traffic 12:232–245.

    https://doi.org/10.1111/j.1600-0854.2010.01138.x

    • PubMed
    • Google Scholar
26. 1. Kasai H
    2. Augustine GJ

    (1990) Cytosolic Ca2+ gradients triggering unidirectional fluid secretion from exocrine pancreas

    Nature 348:735–738.

    https://doi.org/10.1038/348735a0

    • PubMed
    • Google Scholar
27. 1. Kidd JF
    2. Thorn P

    (2000) Intracellular Ca2+ and Cl- channel activation in secretory cells

    Annual review of physiology 62:493–513.

    https://doi.org/10.1146/annurev.physiol.62.1.493

    • PubMed
    • Google Scholar
28. 1. Larina O
    2. Thorn P

    (2005) Ca2+ dynamics in salivary acinar cells: distinct morphology of the acinar lumen underlies near-synchronous global Ca2+ responses

    Journal of Cell Science 118:4131–4139.

    https://doi.org/10.1242/jcs.02533

    • PubMed
    • Google Scholar
29. 1. Lee MG
    2. Xu X
    3. Zeng W
    4. Diaz J
    5. Wojcikiewicz RJ
    6. Kuo TH
    7. Wuytack F
    8. Racymaekers L
    9. Muallem S

    (1997) Polarized expression of Ca2+ channels in pancreatic and salivary gland cells. Correlation with initiation and propagation of [Ca2+]i waves

    The Journal of biological chemistry 272:15765–15770.

    https://doi.org/10.1074/jbc.272.25.15765

    • PubMed
    • Google Scholar
30. 1. Lewis RS
    2. Cahalan MD

    (1989) Mitogen-induced oscillations of cytosolic Ca2+ and transmembrane Ca2+ current in human leukemic T cells

    Cell Regulation 1:99–112.

    https://doi.org/10.1091/mbc.1.1.99

    • PubMed
    • Google Scholar
31. 1. Maruyama Y
    2. Gallacher DV
    3. Petersen OH

    (1983) Voltage and Ca2+-activated K+ channel in baso-lateral acinar cell membranes of mammalian salivary glands

    Nature 302:827–829.

    https://doi.org/10.1038/302827a0

    • PubMed
    • Google Scholar
32. 1. Maruyama EO
    2. Aure MH
    3. Xie X
    4. Myal Y
    5. Gan L
    6. Ovitt CE

    (2016) Cell-Specific Cre Strains For Genetic Manipulation in Salivary Glands

    PLOS ONE 11:e0146711.

    https://doi.org/10.1371/journal.pone.0146711

    • PubMed
    • Google Scholar
33. 1. Matsui M
    2. Motomura D
    3. Karasawa H
    4. Fujikawa T
    5. Jiang J
    6. Komiya Y
    7. Takahashi S
    8. Taketo MM

    (2000) Multiple functional defects in peripheral autonomic organs in mice lacking muscarinic acetylcholine receptor gene for the M3 subtype

    PNAS 97:9579–9584.

    https://doi.org/10.1073/pnas.97.17.9579

    • PubMed
    • Google Scholar
34. 1. Melvin JE

    (1991)

    Saliva and dental diseases

    Current opinion in dentistry 1:795–801.

    • PubMed
    • Google Scholar
35. 1. Melvin JE
    2. Koek L
    3. Zhang GH

    (1991) A capacitative Ca2+ influx is required for sustained fluid secretion in sublingual mucous acini

    American Journal of Physiology-Gastrointestinal and Liver Physiology 261:G1043–G1050.

    https://doi.org/10.1152/ajpgi.1991.261.6.G1043

    • PubMed
    • Google Scholar
36. 1. Melvin JE

    (1999) Chloride channels and salivary gland function

    Critical Reviews in Oral Biology & Medicine 10:199–209.

    https://doi.org/10.1177/10454411990100020601

    • PubMed
    • Google Scholar
37. 1. Melvin JE
    2. Yule D
    3. Shuttleworth T
    4. Begenisich T

    (2005) Regulation of fluid and electrolyte secretion in salivary gland acinar cells

    Annual Review of Physiology 67:445–469.

    https://doi.org/10.1146/annurev.physiol.67.041703.084745

    • PubMed
    • Google Scholar
38. 1. Nakamoto T
    2. Srivastava A
    3. Romanenko VG
    4. Ovitt CE
    5. Perez-Cornejo P
    6. Arreola J
    7. Begenisich T
    8. Melvin JE

    (2007) Functional and molecular characterization of the fluid secretion mechanism in human parotid acinar cells

    American Journal of Physiology-Regulatory, Integrative and Comparative Physiology 292:R2380–R2390.

    https://doi.org/10.1152/ajpregu.00591.2006

    • PubMed
    • Google Scholar
39. 1. Nakamoto T
    2. Romanenko VG
    3. Takahashi A
    4. Begenisich T
    5. Melvin JE

    (2008) Apical maxi-K (KCa1.1) channels mediate K+ secretion by the mouse submandibular exocrine gland

    American Journal of Physiology-Cell Physiology 294:C810–C819.

    https://doi.org/10.1152/ajpcell.00511.2007

    • PubMed
    • Google Scholar
40. 1. Pages N
    2. Vera-Sigüenza E
    3. Rugis J
    4. Kirk V
    5. Yule DI
    6. Sneyd J

    (2019) A Model of [Formula: see text] Dynamics in an Accurate Reconstruction of Parotid Acinar Cells

    Bulletin of mathematical biology 81:1394–1426.

    https://doi.org/10.1007/s11538-018-00563-z

    • PubMed
    • Google Scholar
41. 1. Palk L
    2. Sneyd J
    3. Shuttleworth TJ
    4. Yule DI
    5. Crampin EJ

    (2010) A dynamic model of saliva secretion

    Journal of Theoretical Biology 266:625–640.

    https://doi.org/10.1016/j.jtbi.2010.06.027

    • PubMed
    • Google Scholar
42. 1. Palk L
    2. Sneyd J
    3. Patterson K
    4. Shuttleworth TJ
    5. Yule DI
    6. Maclaren O
    7. Crampin EJ

    (2012) Modelling the effects of calcium waves and oscillations on saliva secretion

    Journal of Theoretical Biology 305:45–53.

    https://doi.org/10.1016/j.jtbi.2012.04.009

    • PubMed
    • Google Scholar
43. 1. Park HS
    2. Betzenhauser MJ
    3. Zhang Y
    4. Yule DI

    (2012) Regulation of Ca²⁺ release through inositol 1,4,5-trisphosphate receptors by adenine nucleotides in parotid acinar cells

    American Journal of Physiology-Gastrointestinal and Liver Physiology 302:G97–G104.

    https://doi.org/10.1152/ajpgi.00328.2011

    • PubMed
    • Google Scholar
44. 1. Pedersen AM
    2. Bardow A
    3. Jensen SB
    4. Nauntofte B

    (2002) Saliva and gastrointestinal functions of taste, mastication, swallowing and digestion

    Oral Diseases 8:117–129.

    https://doi.org/10.1034/j.1601-0825.2002.02851.x

    • PubMed
    • Google Scholar
45. 1. Putney JW

    (1982) Cellular regulation

    Science 216:1403–1404.

    https://doi.org/10.1126/science.216.4553.1403

    • PubMed
    • Google Scholar
46. 1. Romanenko VG
    2. Nakamoto T
    3. Srivastava A
    4. Begenisich T
    5. Melvin JE

    (2007) Regulation of membrane potential and fluid secretion by Ca2+-activated K+ channels in mouse submandibular glands

    The Journal of Physiology 581:801–817.

    https://doi.org/10.1113/jphysiol.2006.127498

    • PubMed
    • Google Scholar
47. 1. Romanenko VG
    2. Catalán MA
    3. Brown DA
    4. Putzier I
    5. Hartzell HC
    6. Marmorstein AD
    7. Gonzalez-Begne M
    8. Rock JR
    9. Harfe BD
    10. Melvin JE

    (2010a) Tmem16A encodes the Ca2+-activated Cl- channel in mouse submandibular salivary gland acinar cells

    Journal of Biological Chemistry 285:12990–13001.

    https://doi.org/10.1074/jbc.M109.068544

    • PubMed
    • Google Scholar
48. 1. Romanenko VG
    2. Thompson J
    3. Begenisich T

    (2010b) Ca2+-activated K channels in parotid acinar cells: The functional basis for the hyperpolarized activation of BK channels

    Channels 4:278–288.

    https://doi.org/10.4161/chan.4.4.12197

    • PubMed
    • Google Scholar
49. Software

    1. Rugis J

    (2021) plot_apical, version swh:1:rev:c8fced9a9d8f3b7526a8f8eabfe547c618f2374b

    Software Heritage.

    https://archive.softwareheritage.org/swh:1:dir:41686eb1f83bbdd30202e8e5d33847dd22bdc68a;origin=https://github.com/jrugis/cell_tools;visit=swh:1:snp:e6f03588cdf88966f5ed61ce247a58d536989a60;anchor=swh:1:rev:c8fced9a9d8f3b7526a8f8eabfe547c618f2374b
50. 1. Sheu SH
    2. Tapia JC
    3. Tsuriel S
    4. Lichtman JW

    (2017) Similar synapse elimination motifs at successive relays in the same efferent pathway during development in mice

    eLife 6:e23193.

    https://doi.org/10.7554/eLife.23193

    • PubMed
    • Google Scholar
51. 1. Siekmann I
    2. Wagner LE
    3. Yule D
    4. Crampin EJ
    5. Sneyd J

    (2012) A kinetic model for type I and II IP3R accounting for mode changes

    Biophysical Journal 103:658–668.

    https://doi.org/10.1016/j.bpj.2012.07.016

    • PubMed
    • Google Scholar
52. 1. Sneyd J
    2. Crampin E
    3. Yule D

    (2014) Multiscale modelling of saliva secretion

    Mathematical Biosciences 257:69–79.

    https://doi.org/10.1016/j.mbs.2014.06.017

    • PubMed
    • Google Scholar
53. 1. Sneyd J
    2. Han JM
    3. Wang L
    4. Chen J
    5. Yang X
    6. Tanimura A
    7. Sanderson MJ
    8. Kirk V
    9. Yule DI

    (2017a) On the dynamical structure of calcium oscillations

    PNAS 114:1456–1461.

    https://doi.org/10.1073/pnas.1614613114

    • PubMed
    • Google Scholar
54. 1. Sneyd J
    2. Means S
    3. Zhu D
    4. Rugis J
    5. Won JH
    6. Yule DI

    (2017b) Modeling calcium waves in an anatomically accurate three-dimensional parotid acinar cell

    Journal of Theoretical Biology 419:383–393.

    https://doi.org/10.1016/j.jtbi.2016.04.030

    • PubMed
    • Google Scholar
55. 1. Sneyd J
    2. Vera-Sigüenza E
    3. Rugis J
    4. Pages N
    5. Yule DI

    (2021) Calcium Dynamics and Water Transport in Salivary Acinar Cells

    Bulletin of Mathematical Biology 83:31.

    https://doi.org/10.1007/s11538-020-00841-9

    • PubMed
    • Google Scholar
56. 1. Takemura H
    2. Hughes AR
    3. Thastrup O
    4. Putney JW

    (1989)

    Activation of calcium entry by the tumor promoter thapsigargin in parotid acinar cells. Evidence that an intracellular calcium pool and not an inositol phosphate regulates calcium fluxes at the plasma membrane

    The Journal of biological chemistry 264:12266–12271.

    • PubMed
    • Google Scholar
57. 1. Teos LY
    2. Zheng CY
    3. Liu X
    4. Swaim WD
    5. Goldsmith CM
    6. Cotrim AP
    7. Baum BJ
    8. Ambudkar IS

    (2016) Adenovirus-mediated hAQP1 expression in irradiated mouse salivary glands causes recovery of saliva secretion by enhancing acinar cell volume decrease

    Gene Therapy 23:572–579.

    https://doi.org/10.1038/gt.2016.29

    • PubMed
    • Google Scholar
58. 1. Thorn P
    2. Lawrie AM
    3. Smith PM
    4. Gallacher DV
    5. Petersen OH

    (1993) Local and global cytosolic Ca2+ oscillations in exocrine cells evoked by agonists and inositol trisphosphate

    Cell 74:661–668.

    https://doi.org/10.1016/0092-8674(93)90513-P

    • PubMed
    • Google Scholar
59. 1. Thorn P
    2. Moreton R
    3. Berridge M

    (1996) Multiple, coordinated Ca2+ -release events underlie the inositol trisphosphate-induced local Ca2+ spikes in mouse pancreatic acinar cells

    The EMBO Journal 15:999–1003.

    https://doi.org/10.1002/j.1460-2075.1996.tb00436.x

    • PubMed
    • Google Scholar
60. 1. Thorn P
    2. Petersen OH

    (1993) Calcium oscillations in pancreatic acinar cells, evoked by the cholecystokinin analogue JMV-180, depend on functional inositol 1,4,5-trisphosphate receptors

    Journal of Biological Chemistry 268:23219–23221.

    https://doi.org/10.1016/S0021-9258(19)49451-9

    • PubMed
    • Google Scholar
61. 1. Thurley K
    2. Smith IF
    3. Tovey SC
    4. Taylor CW
    5. Parker I
    6. Falcke M

    (2011) Timescales of IP(3)-evoked Ca(2+) spikes emerge from Ca(2+) puffs only at the cellular level

    Biophysical Journal 101:2638–2644.

    https://doi.org/10.1016/j.bpj.2011.10.030

    • PubMed
    • Google Scholar
62. 1. Thurley K
    2. Skupin A
    3. Thul R
    4. Falcke M

    (2012) Fundamental properties of Ca2+ signals

    Biochimica et Biophysica Acta (BBA) - General Subjects 1820:1185–1194.

    https://doi.org/10.1016/j.bbagen.2011.10.007

    • PubMed
    • Google Scholar
63. 1. Thurley K
    2. Tovey SC
    3. Moenke G
    4. Prince VL
    5. Meena A
    6. Thomas AP
    7. Skupin A
    8. Taylor CW
    9. Falcke M

    (2014) Reliable encoding of stimulus intensities within random sequences of intracellular Ca2+ spikes

    Science Signaling 7:ra59.

    https://doi.org/10.1126/scisignal.2005237

    • PubMed
    • Google Scholar
64. 1. Tojyo Y
    2. Tanimura A
    3. Matsumoto Y

    (1997) Imaging of intracellular Ca2+ waves induced by muscarinic receptor stimulation in rat parotid acinar cells

    Cell Calcium 22:455–462.

    https://doi.org/10.1016/S0143-4160(97)90073-7

    • PubMed
    • Google Scholar
65. 1. Turner RJ
    2. Sugiya H

    (2002) Understanding salivary fluid and protein secretion

    Oral Diseases 8:3–11.

    https://doi.org/10.1034/j.1601-0825.2002.10815.x

    • PubMed
    • Google Scholar
66. 1. Vera-Sigüenza E
    2. Catalán MA
    3. Peña-Münzenmayer G
    4. Melvin JE
    5. Sneyd J

    (2018) A Mathematical Model Supports a Key Role for Ae4 (Slc4a9) in Salivary Gland Secretion

    Bulletin of Mathematical Biology 80:255–282.

    https://doi.org/10.1007/s11538-017-0370-6

    • PubMed
    • Google Scholar
67. 1. Vera-Sigüenza E
    2. Pages N
    3. Rugis J
    4. Yule DI
    5. Sneyd J

    (2019) A Mathematical Model of Fluid Transport in an Accurate Reconstruction of Parotid Acinar Cells

    Bulletin of Mathematical Biology 81:699–721.

    https://doi.org/10.1007/s11538-018-0534-z

    • PubMed
    • Google Scholar
68. 1. Vera-Sigüenza E
    2. Pages N
    3. Rugis J
    4. Yule DI
    5. Sneyd J

    (2020) A Multicellular Model of Primary Saliva Secretion in the Parotid Gland

    Bulletin of Mathematical Biology 82:38.

    https://doi.org/10.1007/s11538-020-00712-3

    • PubMed
    • Google Scholar
69. 1. Warner JD
    2. Peters CG
    3. Saunders R
    4. Won JH
    5. Betzenhauser MJ
    6. Gunning WT
    7. Yule DI
    8. Giovannucci DR

    (2008) Visualizing form and function in organotypic slices of the adult mouse parotid gland

    American Journal of Physiology-Gastrointestinal and Liver Physiology 295:G629–G640.

    https://doi.org/10.1152/ajpgi.90217.2008

    • PubMed
    • Google Scholar
70. 1. Weiss SJ
    2. McKinney JS
    3. Putney JW

    (1982) Regulation of phosphatidate synthesis by secretagogues in parotid acinar cells

    Biochemical Journal 204:587–592.

    https://doi.org/10.1042/bj2040587

    • PubMed
    • Google Scholar
71. 1. Won JH
    2. Cottrell WJ
    3. Foster TH
    4. Yule DI

    (2007) Ca2+ release dynamics in parotid and pancreatic exocrine acinar cells evoked by spatially limited flash photolysis

    American Journal of Physiology-Gastrointestinal and Liver Physiology 293:G1166–G1177.

    https://doi.org/10.1152/ajpgi.00352.2007

    • PubMed
    • Google Scholar
72. 1. Zhang X
    2. Xin P
    3. Yoast RE
    4. Emrich SM
    5. Johnson MT
    6. Pathak T
    7. Benson JC
    8. Azimi I
    9. Gill DL
    10. Monteith GR
    11. Trebak M

    (2020) Distinct pharmacological profiles of ORAI1, ORAI2, and ORAI3 channels

    Cell Calcium 91:102281.

    https://doi.org/10.1016/j.ceca.2020.102281

    • PubMed
    • Google Scholar

Acceptance summary:

This paper applies intravital 2-photon microscopy to describe the intracellular calcium signals that drive fluid secretion from salivary gland cells in vivo. Secretion of saliva depends on calcium-dependent chloride and potassium channels, and based on studies of isolated salivary gland cells in vitro, these were previously thought to reside in different regions of the cell and require a propagating calcium wave for their activation. This study reveals a much more localized pattern of calcium signals in the intact organ, providing the first view of how calcium is coupled to secretion in vivo and new insights into how ion channel activity is coordinated to optimize secretion.

Decision letter after peer review:

Thank you for submitting your article "The characteristics of intracellular Ca²⁺ signals in vivo necessitate a new model for salivary fluid secretion" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, one of whom is a member of our Board of Reviewing Editors, and the evaluation has been overseen Kenton Swartz as the Senior Editor. The following individual involved in review of your submission has agreed to reveal their identity: Peter Thorn (Reviewer #3).

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

Essential revisions:

The major significance of this work rests on the novel pattern of Ca²⁺ signals associated with secretion in vivo and demonstrating a consequent need for a revised model of apical K channel activation to maintain the driving force for Cl^-efflux across the apical membrane and fluid secretion.

1. Is a new model needed to describe secretion in vivo? The argument for a new model in which Ca²⁺ activates KCa channels in the apical membrane is based on the observation of low levels of secretion (<5 Hz stimulation) associated with apically localized Ca oscillations. However, these signals are not strictly localized to the apical membrane and would likely activate KCa channels in the basolateral membrane that are close to the tight junctions. In addition, a large increase in secretion is seen for stimuli (5-10 Hz) that propagate Ca to the basal pole, consistent with previous models incorporating activation of basal KCa channels. The contribution of KCa channels in the apical membrane needs to be demonstrated experimentally. This could be addressed by applying blockers to the apical surface, although this may be difficult. Overall, to justify the model, a need for apical KCa channels or evidence of their contribution to secretion, needs to be shown.

2. The apical Ca signals are ascribed solely to release through IP3R, but Orai1 channels, known to be expressed in these cells, are likely to contribute as well. Preferably, Orai1 inhibitors applied systemically (as was done for the cholinesterase inhibitor) should be tested for effects on Ca²⁺ signals and secretion. At the least the contributions of Orai1 channels should be discussed, taking into account previously published work.

3. The computational model does not adequately reproduce the main features of the Ca²⁺ response. It shows a large initial spike upon stimulation (Figure 11B) that is not seen experimentally (Figure 9). It also oscillates at a much lower frequency (<0.2 Hz, similar to in vitro) than measured in vivo (0.5-1 Hz). In addition, the model needs to be tested under more conditions to confirm that it can account for the experimental data; e.g., does it predict the intracellular [Ca²⁺]i gradients, and how increasing stimulation frequency changes the Ca²⁺ response? More data are needed to justify the choice of model parameter values.

4. The absence of local apical Ca²⁺ signals in vitro is not the case for all exocrine acinar cells. The authors should acknowledge that local apical responses have been seen in other acinar cells (e.g. pancreatic acinar cells) in vitro, upon stimulation with physiological concentrations of hormones, low concentrations of neurotransmitter, or directly by IP3.

5. The current data do not allow a fair comparison of the spatiotemporal aspects of Ca²⁺ signaling in vivo and in vitro. While the use of GCaMP6f is a good way to rule out effects of different indicators (e.g., fura-2), other factors are likely to contribute and can be easily tested:

(a) The in vivo experiments were conducted at 37 deg C while the in vitro imaging seems to have been done at room temperature (temperature was not indicated in the Methods). Cooler temperature is likely to affect oscillation frequency as well as the rate of pumping or sequestration which could affect propagation.

(b) The concentrations of ACh used in vitro may not relate to the neural stimulation frequencies in vivo. The 100 nM response in vitro is quite large, so it may not be fair to compare this with low- frequency stimulation in vivo where signals are confined to the apical pole. Perhaps more localized signals would result from lower ACh concentrations.

6. The method for standard deviation (SD) image processing should be described. This approach highlights areas that oscillate, but may de-emphasize areas of tonic elevation which would also be expected to drive secretion. At 5 Hz stimulation it shows that oscillations are confined mostly to the apical region (Figure 7C), but at 10 Hz the cells don't seem to oscillate (Figure 8), so it is not clear why SD images are presented here. It would be more informative to plot a cross-section of GCaMP6 signal from apical to basal before and during stimulation.

Reviewer #1:

This work offers a new look at the calcium signals that drive secretion in salivary gland cells. Based on studies of isolated cells in vitro, the common view was that calcium signals originating in the apical pole of the cell propagate to the basal pole, so that chloride channels allow chloride ions to exit across the apical membrane while K channels in the basal membrane are opened to maintain the electrical driving force for chloride exit. Through an elegant use of 2-photon microscopy to monitor the spatial spread of calcium signals in intact glands in vivo, the authors describe a different pattern, in which the signals do not propagate but are much more confined to the apical pole of the cell, and they show that secretion occurs even without significant spread of calcium to the basal pole. These results are significant as they challenge current mechanistic models for secretion, and they imply a functional role for K channels in the apical membrane, which until now have been commonly considered to only function in the basolateral membrane.

However, the paper has several shortcomings that need to be addressed to better support a need for revising the current secretion model. It is not clear whether KCa channels in proximal regions of the basolateral membrane are activated by the apically confined Ca²⁺ and therefore are sufficient to support secretion. A modified computational model intended to support the need for apical KCa channels does not offer strong support, as it does not faithfully reproduce several of the experimentally observed Ca²⁺ responses. Finally, the differences between the spatiotemporal aspects of Ca²⁺ signaling in vivo and in vitro are difficult to assess as critical conditions such as temperature and agonist concentration were not matched in the two experimental scenarios.

Reviewer #2:

In this study Dr. Yule and his colleagues describe very impressive and painstaking work in which they have monitored for the first time Ca²⁺ signals and salivary secretion in vivo in a mouse model which expressed the Ca²⁺ indicator protein, GCamp6F, in salivary glands. Previously reported studies measured fluid secretion in vivo by injecting the animals with pilocarpine and collecting the oral secretion or by using protocols where the gland is perfused. However, cytosolic [Ca²⁺] changes, which are correlated with, and required for, salivary fluid secretion have been predominantly measured in isolated acinar or cell cluster preparations and some in tissue slices. By using intravital multiphoton (MP) microscopy together with stimulation of the neural input to the gland, the authors were able to measure the spatiotemporal properties of Ca²⁺ signals within the acinar cells and collect saliva that is secreted into the oral cavity. Thus, this study is an important breakthrough in this field as it reveals the exact [Ca²⁺] changes associated with salivary gland fluid secretion.

This study shows that major spatiotemporal properties of the Ca²⁺ signals detected in vivo are markedly different from those previously reported using isolated acinar cells (by this group and others) which demonstrated that Ca²⁺ signals are initiated apically but propagate as a wave globally across the cell without generation of Ca²⁺ gradients. The present findings show that following moderate stimulation, apically localized oscillatory Ca²⁺ signals are generated that do not propagate globally and cause minimal secretion. However, at stimulation intensities optimum for secretion, there is a spread of Ca²⁺ across the cell although with an apical-basal standing Ca²⁺ gradient. Based on these differences the authors have proposed a revised model which suggests that the apical region contains all the machinery needed for saliva secretion and that Ca²⁺ released predominately in the apical region, via IP3R, can regulate the function of locally situated ion channels without requirement for a propagated wave of increased [Ca²⁺] across the cell. As I have noted above the data described in this manuscript provide new understanding of the physiological response of salivary gland to neuronal stimulation and how fine-tuning of the spatiotemporal properties of the Ca²⁺ signals regulate fluid secretion.

The main question I have is regarding stimulus strengths that induce optimal levels of secretion (10 and 30Hz) where there is propagation of Ca²⁺ across the cell and almost 3-fold higher secretion relative to that at 5Hz. While increase in % responding cells at higher stimulus strengths could result in higher total secretion, possible contribution of Ca²⁺ influx pathways in the spread of Ca²⁺ in individual acinar cells, e.g. via Orai1 that is localized in the apical region of the cell, cannot be ruled out. Also not clear is whether the spread enhances secretion. Thus, when stimulus is low, secretion is minimal and supported only by release and oscillatory apical [Ca²⁺]. At higher stimulus strengths there might be an additional Ca²⁺ influx component which is activated very early after stimulation, likely during IP3-mediated Ca²⁺ release. Since the basolateral membrane extends all the way to the tight junction, is it possible that several components shown basally localized in the model could actually be quite close to the apical membrane region, near the tight junctions.

Thus, it is important to distinguish between initial stimulation of secretion vs regulation of sustained secretion. For example; the last sentence in the abstract clearly, and correctly, states "salivary secretion can be efficiently stimulated by apically localized Ca²⁺ signals".

Reviewer #3:

This paper describes a nice set of experiments where the authors have measured calcium responses in salivary acinar cells from living animals and, importantly, correlated these with measures of salivary secretion. The work is very well performed and the data interesting and definitely an advance in the field. I cannot think of another study that combines, so well, cellular calcium responses with genuine physiological output.

Strengths

Intravital images are very good and the combination with fluid secretion gives a nice correlation between calcium responses and physiological output – something which is impossible in vitro.

Good initial analysis of the images and convincing data where the calcium responses are further analysed.

The conclusions that local calcium responses can drive physiological responses is interesting in the context of salivary glands.

Weaknesses

The authors claim that local responses are not seen in vitro is dependent on taking a very narrow view of exocrine cells. Local responses are readily seen in other exocrine cells, like pancreatic acinar cells. These local responses can be driven directly by IP3, low concentrations of neurotransmitter, or physiological concentrations of hormones. Furthermore, these responses drive enzyme secretion. Therefore, although these local responses have not been measured intravitally, it is highly likely that they are the native response in the pancreas.

The new model put forward is conjecture. It is built entirely around the observations of localised calcium responses but has no supporting data on changes in membrane potential or activation of specific channels.

The in vitro GCaMP data is not convincing. GCaMP and Fura have different affinities for calcium and so this is a serious potential confounder in their experiments. For the in vitro work, the isolated cells are clearly fragmented and the cell chosen is at the edge of the cell cluster. It would be expected that the cells will completely lose their polarisation and structural factors that might limit the calcium response.

While I like the analysis, I think that much of it is unnecessary. The authors, for example, show that regions that respond well at low stimulation also respond well at high levels of stimulation. That's fine, but the authors then do not explore what the basis of these differences are.

[Editors' note: further revisions were suggested prior to acceptance, as described below.]

Thank you for resubmitting your work entitled "The characteristics of intracellular Ca²⁺ signals in vivo necessitate a new model for salivary fluid secretion" for further consideration by eLife. Your revised article has been evaluated by Kenton Swartz (Senior Editor) and a Reviewing Editor.

Thank you for the revised manuscript with the additional data and clarifications. Your responses have addressed most of the major concerns of the reviewers. The only remaining issue is whether KCa channels in the apical membrane are required for secretion. I think this is central to the significance of the paper. The study shows beautifully how Ca signals are mostly apical and do not propagate to the basal membrane in vivo, but the most interesting implication is that it necessitates having KCa channels near or in the apical membrane. To broaden the appeal of your paper to more than exocrine gland physiologists it will be important to establish this point clearly.

We understand that an experimental approach (luminal application of blockers) to establish the role of apical KCa channels is not feasible. Thus, the model offers the best opportunity to test this idea. The model needs a fuller description, not including all the details and equations, but rather a description of the salient features that will allow the reader to grasp how it works, and what assumptions it makes. The comments below summarize the uncertainties in the model's operation that need to be addressed:

1. How are apical, cytoplasmic, and basal regions defined in the model, and how do they correspond to the same regions in the experimental data? Throughout the paper there are references to "apical PM" or "apical region." The apical PM is the membrane bounded by tight junctions, but what does "apical region" mean – does this include lateral membrane between the apical PM and the nucleus? The model cartoon in Figure 13 does not explain the distinction between apical and basolateral. Likewise, on p. 12: "We emphasise that, in the model, KCa channels and Na/K- ATPases are also present in the basal membrane, and thus the secretory machinery is not restricted to the apical membrane." Is "basal" here referring to lateral membrane as well? I imagine that some of this vagueness is because the anatomy is complex and the localization of channels is not precisely known, but this needs to be clarified, particularly for non-expert readers.

2. How is the KCa current simulated? This is a crucial part of the model, but it is not clear how this was done. The only description is (p. 25) "The apical/basal K⁺ current ratio was determined by the apical/basal surface area ratio." After Ca²⁺ is released from the ER, is it assumed to diffuse to the lateral membrane where it activates KCa channels? (lateral membrane is defined here as basolateral membrane between the tight junction and the nucleus, corresponding to the "cytoplasmic" region in the experimental Ca measurements.) Or is only the basal Ca²⁺ signal being used to determine KCa current? The data in Figure 10 show that there is significant Ca²⁺ elevation in the cytoplasmic region below the apical membrane even at low stimulation frequencies, which could activate channels in the lateral membrane. What spatial distribution of KCa channels is assumed in the model? These things need to be clearly and simply explained.

3. The model shows that with no apical KCa channels (but the same total number) you still get ~80% of the maximal response (Figure 14F). This is almost maximal and implies only a modest effect of true apical PM channels. Figure 14F Y axis should be plotted from 0 rather than 60, in order to give a truer impression of the effect of KCa channel localization. There is also some confusion here: Figure 14B shows only 50% of max response if you remove apical KCa channels – but I assume that this is because the total number of channels was not constant in this case, resulting in fewer KCa channels than in Figure 14F. If so, that is not a fair comparison.

4. On p. 17 there is a reference to Supp. Figure 4, but that figure was not included.

5. I would suggest some rewording of the title, as it seems too focused on what was done, rather than the most significant result. Perhaps you can revise it to focus on localized calcium signaling that engages apical K channels in vivo.

Essential revisions:

The major significance of this work rests on the novel pattern of Ca²⁺ signals associated with secretion in vivo and demonstrating a consequent need for a revised model of apical K channel activation to maintain the driving force for Cl^-efflux across the apical membrane and fluid secretion.

1. Is a new model needed to describe secretion in vivo? The argument for a new model in which Ca²⁺ activates KCa channels in the apical membrane is based on the observation of low levels of secretion (<5 Hz stimulation) associated with apically localized Ca oscillations. However, these signals are not strictly localized to the apical membrane and would likely activate KCa channels in the basolateral membrane that are close to the tight junctions. In addition, a large increase in secretion is seen for stimuli (5-10 Hz) that propagate Ca to the basal pole, consistent with previous models incorporating activation of basal KCa channels. The contribution of KCa channels in the apical membrane needs to be demonstrated experimentally. This could be addressed by applying blockers to the apical surface, although this may be difficult. Overall, to justify the model, a need for apical KCa channels or evidence of their contribution to secretion, needs to be shown.

While a direct demonstration that apical KCa channels participate in the secretory mechanism would be ideal, we believe that the experiments proposed (local application of blockers to the apical face of acini) is not practically feasible. Specifically, given the architecture of an acinus, it is not possible to effectively isolate the apical plasma membrane (~1-5 % of PM in the center of a group of cells) to apply a pharmacological agent locally with any degree of certainty without it acting basolaterally. In addition, it is difficult to imagine how this could be performed in every apical domain in the gland to assess its effect on fluid secretion into the oral cavity following neural stimulation. Further, this problem is not obviously amenable to any straightforward genetic approach. Our previous data has demonstrated the presence of apical KCa channels in salivary gland acinar cells and we provide a more complete discussion of evidence for their functional importance.

2. The apical Ca signals are ascribed solely to release through IP3R, but Orai1 channels, known to be expressed in these cells, are likely to contribute as well. Preferably, Orai1 inhibitors applied systemically (as was done for the cholinesterase inhibitor) should be tested for effects on Ca²⁺ signals and secretion. At the least the contributions of Orai1 channels should be discussed, taking into account previously published work.

We have performed new in vivo experiments using the Orai channel blocker, GSK7975A, which show that the sustained phase of the Ca²⁺ signal following extended stimulation with physiological stimulus strengths is blunted by the Orai antagonist. The initial peak is largely unaffected, consistent with the Ca²⁺ response to sustained stimulation being mediated by Ca²⁺ release and subsequent Ca²⁺ influx.

3. The computational model does not adequately reproduce the main features of the Ca²⁺ response. It shows a large initial spike upon stimulation (Figure 11B) that is not seen experimentally (Figure 9). It also oscillates at a much lower frequency (<0.2 Hz, similar to in vitro) than measured in vivo (0.5-1 Hz). In addition, the model needs to be tested under more conditions to confirm that it can account for the experimental data; e.g., does it predict the intracellular [Ca²⁺]i gradients, and how increasing stimulation frequency changes the Ca²⁺ response? More data are needed to justify the choice of model parameter values.

The model has been modified and the computational aspects of these studies have been expanded to test more conditions. The modified model now also more faithfully reflects the characteristics of the Ca²⁺ signals observed experimentally. A more detailed discussion of the model and predictions is also now provided, specifically to explicitly state that basolateral KCa channels in the original (and present) model are present throughout the PM to regions directly adjacent to tight junctions. We now test the model responses under more conditions. We show that increasing stimulation increases the frequency of oscillations, and decreases the latency (new Figure 14A), in agreement with data. We also compare model and experimental line scans and show that the [Ca²⁺] gradient through the cell in the model is qualitatively similar to that observed experimentally. We test model behaviour for lower values of the IP₃ diffusion coefficient, and show that the behaviour remains unchanged (results not shown, as they are so similar to the results obtained with a larger IP₃ diffusion coefficient. However, these results can be included as supplementary data should this be required). We simulated the model with periodic stimulation of PLC activity, to mimic the effects of periodic neural stimulation, and showed that the [Ca²⁺] traces and secretion remain essentially unchanged (although there are some minor quantitative differences). Again, because of the similarity with the results already shown here we do not include these results explicitly, we just state the result. We now give a more detailed discussion of where the model parameter values are derived. Since these parameter values have been developed over many years, and have already been presented in multiple papers, we do not give full details again in the present submission. For example, the parameters that govern our model of the IPR were determined by fitting the model to single-channel data, and this work is described in a series of papers between 2009 and 2019 (Refs 1-7 below). However, any interested reader is now told more precisely where further details about parameter values can be found.

1. Gin, E., L.E. Wagner, D.I. Yule, and J. Sneyd, Inositol trisphosphate receptor and ion channel models based on single-channel data. Chaos, 2009. 19(3): p. 037104.

2. Gin, E., M. Falcke, L.E. Wagner, 2nd, D.I. Yule, and J. Sneyd, A kinetic model of the inositol trisphosphate receptor based on single-channel data. Biophys J, 2009. 96(10): p. 4053-62.

3. Gin, E., M. Falcke, L.E. Wagner, D.I. Yule, and J. Sneyd, Markov chain Monte Carlo fitting of single-channel data from inositol trisphosphate receptors. J Theor Biol, 2009. 257(3): p. 460-74.

4. Siekmann, I., L.E. Wagner, 2nd, D. Yule, E.J. Crampin, and J. Sneyd, A kinetic model for type I and II IP3R accounting for mode changes. Biophys J, 2012. 103(4): p. 658-68.

5. Siekmann, I., L.E. Wagner, D. Yule, E.J. Crampin, and J. Sneyd, A Park/Drive Model for the Inositol-Trisphosphate Receptor (IPR). Biophysical Journal, 2012. 102(3): p. 110a-110a.

6. Cao, P., X. Tan, G. Donovan, M.J. Sanderson, and J. Sneyd, A deterministic model predicts the properties of stochastic calcium oscillations in airway smooth muscle cells. PLoS Comput Biol, 2014. 10(8): p. e1003783.

7. Cao, P., G. Donovan, M. Falcke, and J. Sneyd, A stochastic model of calcium puffs based on single-channel data. Biophys J, 2013. 105(5): p. 1133-42.

4. The absence of local apical Ca²⁺ signals in vitro is not the case for all exocrine acinar cells. The authors should acknowledge that local apical responses have been seen in other acinar cells (e.g. pancreatic acinar cells) in vitro, upon stimulation with physiological concentrations of hormones, low concentrations of neurotransmitter, or directly by IP3.

A discussion of the numerous reports of apically localized Ca²⁺ signals in acutely isolated pancreatic acinar cells is now included.

5. The current data do not allow a fair comparison of the spatiotemporal aspects of Ca²⁺ signaling in vivo and in vitro. While the use of GCaMP6f is a good way to rule out effects of different indicators (e.g., fura-2), other factors are likely to contribute and can be easily tested:

(a) The in vivo experiments were conducted at 37 deg C while the in vitro imaging seems to have been done at room temperature (temperature was not indicated in the Methods). Cooler temperature is likely to affect oscillation frequency as well as the rate of pumping or sequestration which could affect propagation.

(b) The concentrations of ACh used in vitro may not relate to the neural stimulation frequencies in vivo. The 100 nM response in vitro is quite large, so it may not be fair to compare this with low- frequency stimulation in vivo where signals are confined to the apical pole. Perhaps more localized signals would result from lower ACh concentrations.

We have now provided new in vitro data that demonstrates that acini are more sensitive to agonist stimulation and oscillations are more rapid at 37^oC, but the Ca²⁺ are still invariably global at physiological temperature.

6. The method for standard deviation (SD) image processing should be described. This approach highlights areas that oscillate, but may de-emphasize areas of tonic elevation which would also be expected to drive secretion. At 5 Hz stimulation it shows that oscillations are confined mostly to the apical region (Figure 7C), but at 10 Hz the cells don't seem to oscillate (Figure 8), so it is not clear why SD images are presented here. It would be more informative to plot a cross-section of GCaMP6 signal from apical to basal before and during stimulation.

We describe the method for SD image generation. While we contend that this method is an effective means of visualizing the sub-cellular gradients (“average” images generated from the time series are essentially indistinguishable from the SD image and highlight apical-basal gradients), we now provide plots of multiple ROI across the acinus to better visualize the gradients (Figures 8/9). We also provide a new figure where line scans have been performed across an acinus (figure 10). Both these approaches demonstrate that the magnitude of the basal Ca²⁺ signal is ~20-30% of that in the apical domain.

Reviewer #1:

[…] The paper has several shortcomings that need to be addressed to better support a need for revising the current secretion model. It is not clear whether KCa channels in proximal regions of the basolateral membrane are activated by the apically confined Ca²⁺ and therefore are sufficient to support secretion.

We appreciate the opportunity to clarify this point. In the model the basolateral membrane extends all the way to the tight junction, and thus a fraction of the basal KCa channels are close enough to the apical membrane to be activated by Ca²⁺. This fact has been explicitly stated in the new version of the manuscript. The effect of this localization can be seen in the blue curve in new Figure 14B, which shows how, in the old model modified to have no propagated apical-to-basal Ca²⁺ wave, there is still an increase in secretion upon stimulation. However, this increase is only about 50% of the saliva secretion attained if apical KCa channels are included, as in the new version of the model (new Figure 14B, red curve).

A modified computational model intended to support the need for apical KCa channels does not offer strong support, as it does not faithfully reproduce several of the experimentally observed Ca²⁺ responses.

The large initial peak in the first version of the model has now been corrected by making a slight change to the way in which we model the activity of PLC. In the first version of the model PLC activity was a function of [Ca²⁺]. In the new version of the model this dependence on [Ca²⁺] has been removed (as described on new page 24). This change has removed the artifact of the large initial spike, but has made no difference to the conclusions drawn from the model results. The model simulations now closely resemble the data, in that they are oscillations (with the correct frequency of around 0.5 to 1 Hz at high stimulation) superimposed on an increasing baseline as seen in new figure 14 A and discussed on page 12.

Finally, the differences between the spatiotemporal aspects of Ca²⁺ signaling in vivo and in vitro are difficult to assess as critical conditions such as temperature and agonist concentration were not matched in the two experimental scenarios.

We agree with the reviewer and have now included data which investigates the characteristics of the Ca²⁺ signals at 37^oC in isolated GCamp6F expressing acinar cells. We now show that the individual cells are more sensitive to secretagogue stimulation and exhibit more rapid oscillations at 37^oC. Nevertheless, at all concentrations of secretagogues, the signals originate in the apical aspects of the cell and invariably globalize and thus the indicator or experimental temperature cannot explain the in vivo data.

Reviewer #2:

[…] The main question I have is regarding stimulus strengths that induce optimal levels of secretion (10 and 30Hz) where there is propagation of Ca²⁺ across the cell and almost 3-fold higher secretion relative to that at 5Hz. While increase in % responding cells at higher stimulus strengths could result in higher total secretion, possible contribution of Ca²⁺ influx pathways in the spread of Ca²⁺ in individual acinar cells, e.g. via Orai1 that is localized in the apical region of the cell, cannot be ruled out.

This is an important point because it is highly likely that Ca²⁺ influx pathways play a major role in maintaining the Ca²⁺ signal, and thus secretion, especially during prolonged stimulation. We now include data which shows that systemic administration of an Orai channel blocker GSK7579A markedly inhibits in vivo Ca²⁺ signals to physiologically optimum nervous stimuli. Notably, but only after extended stimulation (~5 min) and has little impact on the initial phase of the Ca²⁺ rise. These data are included as new figure 4.

Also not clear is whether the spread enhances secretion. Thus, when stimulus is low, secretion is minimal and supported only by release and oscillatory apical [Ca²⁺]. At higher stimulus strengths there might be an additional Ca²⁺ influx component which is activated very early after stimulation, likely during IP3-mediated Ca²⁺ release. Since the basolateral membrane extends all the way to the tight junction, is it possible that several components shown basally localized in the model could actually be quite close to the apical membrane region, near the tight junctions.

Thus, it is important to distinguish between initial stimulation of secretion vs regulation of sustained secretion. For example; the last sentence in the abstract clearly, and correctly, states "salivary secretion can be efficiently stimulated by apically localized Ca²⁺ signals".

We obviously accept this idea and as stated in response to the Reviewer 1, our model originally incorporated KCa channels in the basolateral membrane immediately abutting the tight-junctions and incorporates the evidence that some will be activated by the reduced Ca²⁺ signal seen in the basal region. Secretion predictions from the model varying the ratio of apical:basal KCa are now included and support the idea that secretion is optimum with a proportion of apical KCa, activated by the large increase in Ca²⁺ which occur in the extreme apical domain.

Reviewer #3:

[…] The authors claim that local responses are not seen in vitro is dependent on taking a very narrow view of exocrine cells. Local responses are readily seen in other exocrine cells, like pancreatic acinar cells. These local responses can be driven directly by IP3, low concentrations of neurotransmitter, or physiological concentrations of hormones. Furthermore, these responses drive enzyme secretion. Therefore, although these local responses have not been measured intravitally, it is highly likely that they are the native response in the pancreas.

The Authors acknowledge that there is a rich literature in isolated pancreatic acinar cells demonstrating that local Ca²⁺ signals occur at low agonist concentrations and exposure to IP₃. Additional references have been added to the text to reflect this. A discussion point is that the majority of these studies were performed in whole-cell patch clamped isolated cells, where cytoplasmic buffering is imposed by the internal pipette solution and which itself could influence the spatial characteristics of the Ca²⁺ signals and thus it would still be informative to observe Ca²⁺ signals stimulated in vivo in the exocrine pancreas.

The new model put forward is conjecture. It is built entirely around the observations of localised calcium responses but has no supporting data on changes in membrane potential or activation of specific channels.

The model has been developed over a number of years and is faithfully built on experimental data. More details are given regarding the model parameters in the revised paper, which have been derived experimentally and published over many years. While it is correct that the present study did not monitor any effectors of the local Ca²⁺ signal in vivo, a previous study (Almassy et al., JGP 139. 121-133. 2012) exhaustively demonstrated that there were apical KCa3.1 and KCa1.1 in the apical PM of salivary acinar cells, which were activated by local uncaging of Ca²⁺. A more extensive discussion of this previous data has been added to the text in the revised manuscript.

The in vitro GCaMP data is not convincing. GCaMP and Fura have different affinities for calcium and so this is a serious potential confounder in their experiments. For the in vitro work, the isolated cells are clearly fragmented and the cell chosen is at the edge of the cell cluster. It would be expected that the cells will completely lose their polarisation and structural factors that might limit the calcium response.

We agree that a more thorough characterization of the attributes of the Ca²⁺ signal in vitro would strengthen our case that there is a real difference between in vivo and isolated cells. However, we stress that in 20 years working on Ca²⁺ signaling in salivary gland acinar cells, stimulated by secretagogue or IP₃ directly, we have never observed an apical confined Ca²⁺ rise. We now include additional data, from experiments performed at 37^oC and over a wider range of secretagogue concentrations.

While I like the analysis, I think that much of it is unnecessary. The authors, for example, show that regions that respond well at low stimulation also respond well at high levels of stimulation. That's fine, but the authors then do not explore what the basis of these differences are.

The Authors contend that this analysis establishing the “baseline” characteristics of in vivo Ca²⁺ signals in salivary acinar cells for the first time is important per se. The response heterogeneity observed is a feature which can only really be meaningful when studied in vivo, given that any variability in response in isolated cells could be due to the isolation procedure. Exploring further the underlying mechanisms will be the subject of future study.

[Editors' note: further revisions were suggested prior to acceptance, as described below.]

Thank you for the revised manuscript with the additional data and clarifications. Your responses have addressed most of the major concerns of the reviewers. The only remaining issue is whether KCa channels in the apical membrane are required for secretion. I think this is central to the significance of the paper. The study shows beautifully how Ca signals are mostly apical and do not propagate to the basal membrane in vivo, but the most interesting implication is that it necessitates having KCa channels near or in the apical membrane. To broaden the appeal of your paper to more than exocrine gland physiologists it will be important to establish this point clearly.

We understand that an experimental approach (luminal application of blockers) to establish the role of apical KCa channels is not feasible. Thus, the model offers the best opportunity to test this idea. The model needs a fuller description, not including all the details and equations, but rather a description of the salient features that will allow the reader to grasp how it works, and what assumptions it makes. The comments below summarize the uncertainties in the model's operation that need to be addressed:

The Authors wish to thank the Reviewers and Editor for their constructive comments. In response we have made changes to the text and modified figures. We believe this has further clarified our description of the architecture of the acini as it relates to the experimental and computational approaches with the overarching goal of making the paper more accessible to a wider readership. Specifically, this includes a more extensive description of the model, including an additional supplemental figure and additional results of the model simulations. We provide a point-by-point response to the comments provided below.

1. How are apical, cytoplasmic, and basal regions defined in the model, and how do they correspond to the same regions in the experimental data? Throughout the paper there are references to "apical PM" or "apical region." The apical PM is the membrane bounded by tight junctions, but what does "apical region" mean – does this include lateral membrane between the apical PM and the nucleus? The model cartoon in Figure 13 does not explain the distinction between apical and basolateral. Likewise, on p. 12: "We emphasise that, in the model, KCa channels and Na/K- ATPases are also present in the basal membrane, and thus the secretory machinery is not restricted to the apical membrane." Is "basal" here referring to lateral membrane as well? I imagine that some of this vagueness is because the anatomy is complex and the localization of channels is not precisely known, but this needs to be clarified, particularly for non-expert readers.

In the introduction, we now provide a paragraph describing the general architecture of exocrine acinar cells, including making it clear that basal and lateral membranes are functionally identical. The cartoon in Figure 13 has been amended to illustrate this architecture. In an effort to define the “apical region” we also highlight published data demonstrating that the apical ER and apical plasma membrane are intimately associated and can be considered a region of privileged communication. In the experimental figures, the apical domain is defined as in the line scan data, where the apical domain is considered to be 3 μm from the apical PM. The line scan and positioning of ROIs in experimental data is designed primarily to illustrate the spatial heterogeneity of the Ca²⁺ signal. It is possible this volume encompasses some lateral membrane in the region defined as “cytoplasmic”, however this does not impact the secretion model because of how this Ca²⁺ release is modelled (see below).

In the model the apical, basal and lateral membranes of the model are clearly defined by anatomical reconstructions from Z stacks in which the TMEM channels and the Na/K ATPases are labelled. This procedure is discussed in detail in previous papers, but here, for clarity, we now include a three-dimensional plot of the cell (supplemental figure 5) on which all the computations in this paper were performed. The acinar membranes are labelled in red and the basolateral membranes in blue. Thus, the reader can now obtain a much clearer intuitive understanding of the model domain, without having to read our previous papers. Although there is no such clear-cut experimentally-based definition of apical or basal “regions”, this is unimportant for the model, in which the concept of apical and basal “regions” plays essentially no role. This is because the IPR flux is modelled as a boundary flux through the apical membrane, while PLC activity is restricted to the basal membrane. Thus, the spatial positioning of these two critical Ca pathways is determined directly (and reasonably precisely) by experimental data. Of course, this requires some simplifying assumptions in the model; for example, the IPR Ca²⁺ flux is not exactly a boundary flux, but comes through IPR which we know are situated very close (within approximately 50 nm) of the TMEM channels, which are certainly on the apical membrane. However, this simplifying assumption appears to introduce no significant quantitative differences; we know this because we have also run simulations in which the IPR Ca²⁺ flux occurs in an apical “region” consisting of all the grid points within 3 (or 2, or 1, or 0.5) microns of the apical membrane, and the model simulations show no significant quantitative differences in any of these cases. We now explain this in more detail on new pages 24 and 25.

Although the model does technically make a distinction between the basal membranes and the lateral membranes (i.e., the lateral surface triangles are labelled differently, can be colored differently, and are constructed so as to be conformal with the lateral surface triangles of the immediately adjacent cell) this distinction is of no importance in

the current simulations, which (without exception) assume that all the ion channels and transporters (with the exception of TMEM, which is restricted to the apical membrane) are uniformly distributed across both the basal and lateral membranes. This means that Ca²⁺ released from the apical “region” (which, in the model, means Ca²⁺ influx across the surface triangles corresponding to the apical membrane) can diffuse to neighboring basolateral surface triangles and activate KCa channels on the basolateral surface. This is why the model exhibits substantial fluid flow even when there are no KCa channels in the apical membrane.

2. How is the KCa current simulated? This is a crucial part of the model, but it is not clear how this was done. The only description is (p. 25) "The apical/basal K⁺ current ratio was determined by the apical/basal surface area ratio." After Ca²⁺ is released from the ER, is it assumed to diffuse to the lateral membrane where it activates KCa channels? (lateral membrane is defined here as basolateral membrane between the tight junction and the nucleus, corresponding to the "cytoplasmic" region in the experimental Ca measurements.) Or is only the basal Ca²⁺ signal being used to determine KCa current? The data in Figure 10 show that there is significant Ca²⁺ elevation in the cytoplasmic region below the apical membrane even at low stimulation frequencies, which could activate channels in the lateral membrane. What spatial distribution of KCa channels is assumed in the model? These things need to be clearly and simply explained.

We now explain in much more detail (new page 25) exactly how activation of the KCa channels works in the model. Briefly (and as we describe above also), the channels on each surface triangle are affected only by the average [Ca²⁺] in the tetrahedron that contains that triangle. Thus Ca²⁺ released from the apical region can diffuse a short distance to the neighboring basolateral membrane to activate KCa channels there.

3. The model shows that with no apical KCa channels (but the same total number) you still get ~80% of the maximal response (Figure 14F). This is almost maximal and implies only a modest effect of true apical PM channels. Figure 14F Y axis should be plotted from 0 rather than 60, in order to give a truer impression of the effect of KCa channel localization. There is also some confusion here: Figure 14B shows only 50% of max response if you remove apical KCa channels – but I assume that this is because the total number of channels was not constant in this case, resulting in fewer KCa channels than in Figure 14F. If so, that is not a fair comparison.

In the previous resubmission we acknowledged that the model predicts that the absence of apical KCa channels reduces salivary secretion by ~25%. We contend that this is not insignificant. Nevertheless, there is indeed only a partial effect from the presence of apical KCa channels, as the model is capable of significant fluid secretion in their absence. We now emphasize more strongly throughout that the model predicts that apical KCa channels, although not necessary for fluid flow, appear to be required for optimal fluid flow, hence the change in the title for the manuscript. We have modified the figure as requested by the reviewer and added simulations for additional levels of stimulus strengths.

We also now explain at greater length (and, we hope, more clearly) that the curve with no apical KCa in Figure 14B corresponds to simulations from the old model, not the new one, which explains the quantitative differences with Figure 14F. The curves in Figure 14 B are designed to show how the old model is poor at generating fluid flow in the absence of apical KCa, and thus a new model is needed. Of course, we note that even the old model can generate some fluid flow even in the absence of apical KCa. In both models it is simply a matter of how much fluid flow there is, not whether fluid flow exists at all.

4. On p. 17 there is a reference to Supp. Figure 4, but that figure was not included.

Supp Figure 4- (Figure 5-Supp Figure 4) is now included.

5. I would suggest some rewording of the title, as it seems too focused on what was done, rather than the most significant result. Perhaps you can revise it to focus on localized calcium signaling that engages apical K channels in vivo.

We have amended the title to “Highly Localized intracellular Ca²⁺ signals promote optimal salivary gland fluid secretion”, which we feel better reflects the data.

Author details

1. Takahiro Takano

   Department of Pharmacology and Physiology, University of Rochester, Rochester, United States

   Contribution

   Formal analysis, Investigation

   Competing interests

   No competing interests declared

2. Amanda M Wahl

   Department of Pharmacology and Physiology, University of Rochester, Rochester, United States

   Contribution

   Investigation

   Competing interests

   No competing interests declared

3. Kai-Ting Huang

   Department of Pharmacology and Physiology, University of Rochester, Rochester, United States

   Contribution

   Conceptualization, Methodology

   Competing interests

   No competing interests declared

4. Takanori Narita

   Department of Veterinary Medicine, College of Bioresource Sciences, Nihon University, Fujisawa, Japan

   Contribution

   Methodology

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0003-0295-5014

5. John Rugis

   Department of Mathematics, University of Auckland, Auckland, New Zealand

   Contribution

   Software, Visualization, Writing - review and editing

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0002-8009-4152

6. James Sneyd

   Department of Mathematics, University of Auckland, Auckland, New Zealand

   Contribution

   Conceptualization, Investigation, Methodology, Writing - original draft

   Competing interests

   No competing interests declared

7. David I Yule

   Department of Pharmacology and Physiology, University of Rochester, Rochester, United States

   Contribution

   Conceptualization, Supervision, Funding acquisition, Writing - original draft, Project administration

   For correspondence

   david_yule@urmc.rochester.edu

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0002-6743-0668

• 1,274

  Page views

• 184

  Downloads

• 9

  Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.