Pseudomonas aeruginosa Homoserine Lactone Activates Store-operated cAMP and Cystic Fibrosis Transmembrane Regulator-dependent Cl− Secretion by Human Airway Epithelia*

The ubiquitous bacterium Pseudomonas aeruginosa frequently causes hospital-acquired infections. P. aeruginosa also infects the lungs of cystic fibrosis (CF) patients and secretes N-(3-oxo-dodecanoyl)-S-homoserine lactone (3O-C12) to regulate bacterial gene expression critical for P. aeruginosa persistence. In addition to its effects as a quorum-sensing gene regulator in P. aeruginosa, 3O-C12 elicits cross-kingdom effects on host cell signaling leading to both pro- or anti-inflammatory effects. We find that in addition to these slow effects mediated through changes in gene expression, 3O-C12 also rapidly increases Cl− and fluid secretion in the cystic fibrosis transmembrane regulator (CFTR)-expressing airway epithelia. 3O-C12 does not stimulate Cl− secretion in CF cells, suggesting that lactone activates the CFTR. 3O-C12 also appears to directly activate the inositol trisphosphate receptor and release Ca2+ from the endoplasmic reticulum (ER), lowering [Ca2+] in the ER and thereby activating the Ca2+-sensitive ER signaling protein STIM1. 3O-C12 increases cytosolic [Ca2+] and, strikingly, also cytosolic [cAMP], the known activator of CFTR. Activation of Cl− current by 3O-C12 was inhibited by a cAMP antagonist and increased by a phosphodiesterase inhibitor. Finally, a Ca2+ buffer that lowers [Ca2+] in the ER similar to the effect of 3O-C12 also increased cAMP and ICl. The results suggest that 3O-C12 stimulates CFTR-dependent Cl− and fluid secretion in airway epithelial cells by activating the inositol trisphosphate receptor, thus lowering [Ca2+] in the ER and activating STIM1 and store-operated cAMP production. In CF airways, where CFTR is absent, the adaptive ability to rapidly flush the bacteria away is compromised because the lactone cannot affect Cl− and fluid secretion.

in this secretion, and whether Ca 2ϩ and cAMP signaling were involved. 3O-C12 increases cytosolic [Ca 2ϩ ] (Ca cyto ) in fibroblasts (7) and mast cells (13), and at least at high [3O-C12] (250 -1000 M), this resulted from Ca 2ϩ release from an internal store, possibly the endoplasmic reticulum (ER). If 3O-C12 elicited similar effects in airway epithelia, 3O-C12 might also raise cAMP by an ER store-operated cAMP mechanism recently described for colonic epithelial cells; Lefkimmiatis (15) discovered that thapsigargin (inhibitor of the Ca 2ϩ -ATPase of the ER) activated cAMP production by releasing Ca 2ϩ from the ER, lowering [Ca 2ϩ ] in the ER (Ca ER ), and activating the ERresident protein STIM1 (stromal interacting molecule 1; Ref 16) and adenylate cyclase.
The present experiments used electrophysiological and imaging methods to test whether the store-operated cyclase model (15) could explain the stimulatory effects of 3O-C12 on Cl Ϫ secretion by airway epithelia. Transepithelial electrophysiology was used in combination with CFTR-expressing and genetically matched airway epithelial cell lines to test whether 3O-C12 increased CFTR-dependent Cl Ϫ secretion. Fluid secretion by submucosal glands in intact pig tracheas was measured to determine whether 3O-C12-stimulated Cl Ϫ secretion also contributed to fluid secretion in intact tissues. Ca cyto (fura-2 imaging) and Ca ER (FRET imaging of ER-targeted cameleon) were measured during treatments with 3O-C12 and thapsigargin (selective blocker of Ca 2ϩ -ATPase in the ER) to test whether increases in Ca cyto resulted from release of Ca 2ϩ from the ER or from some other organelle. Patch clamp electrophysiology of inositol trisphosphate receptor 1 (IP 3 R1) expressed in the nuclei isolated from chicken B cells (DT40) tested whether decreases in Ca ER resulted from direct 3O-C12 activation of the IP 3 R or some other release or uptake mechanism. Total internal reflection fluorescence (TIRF) imaging was used to measure activation of STIM1, the key ER protein that has been proposed to mediate reductions in Ca ER to activation of cAMP production (15). The role of cAMP in the Cl Ϫ secretory response was tested by measuring cAMP with Epac H30 FRET imaging and then by testing inhibitors that increase [cAMP] (phosphodiesterase blocker) and inhibit protein kinase A (R p )-cAMP. Finally, cAMP and Cl Ϫ secretion were measured in cells treated with the membrane-permeant ER Ca 2ϩ buffer TPEN to determine whether specific reductions in Ca ER (i.e. without altering Ca cyto ) would increase cAMP and activate Cl Ϫ secretion.

MATERIALS AND METHODS
Reagents-Unless otherwise specified, all reagents and chemicals were obtained from Sigma. 3O-C12 (Cayman Chemical, Ann Arbor MI) was dissolved in ethanol and frozen in separate vials and then thawed for single experiments. Preliminary experiments showed that 3O-C12 lost potency with repeated thaw-freeze-thaw cycles. The cAMP-elevating agonist forskolin (Calbiochem) was prepared as a 20 mM stock solution in dimethyl sulfoxide (DMSO), and an aliquot was added at final concentrations of 2-50 M. CFTR blocker glibenclamide (17) was prepared as a 300 mM stock solution in DMSO and added to solutions at 1 mM. GLYH101 (18) and CFTRinh172 (19) were provided by Dr. Alan Verkman (University of California, San Francisco), prepared as a 20 mM stock solution in DMSO, and added to solutions at concentrations noted in the text. The Ca 2ϩ -ATPase blocker thapsigargin (20) was prepared as a 1 mM stock in DMSO and used at 5 M. TPEN was added to Ca 2ϩ -free Ringer's containing 100 M EGTA or Cl Ϫ -free ϩ Ca 2ϩ -free Ringer's at 1 mM and dissolved by continuous stirring for 1 h. TPEN was then used at either 1 or 0.5 mM as mentioned specifically in the text.
The role of CFTR in Cl Ϫ secretory responses to 3O-C12 was determined by comparing CF and CFTR-corrected CF bronchial cell lines. The parent human bronchial CF (CFBE41o Ϫ , ⌬F508/⌬F508) and CFTR-corrected (CFTR-CFBE41o Ϫ ) cell lines were generated and cultured on permeable supports as described previously (23,24). Both the parental CFBE41o Ϫ and the CFTR-corrected CFBE41o Ϫ cell lines consistently maintained an epithelial phenotype and expressed Ca 2ϩ -activated Cl Ϫ currents but no amiloride-sensitive Na ϩ absorption. The CFTR-CFBE41o Ϫ clone maintained a high level of transgene expression over all observed passages (23) and exhibited forskolin-stimulated Cl Ϫ secretion.
Transepithelial Electrophysiology-For measurements of transepithelial Cl Ϫ current, CaLu-3 cell monolayers were grown on permeable supports, washed in PBS, mounted into water-jacketed (37°C) Ussing chambers (Physiologic Instruments, San Diego), and used for electrophysiological studies (22)(23)(24). Transepithelial voltage (V t ), resistance (R t ), and short circuit current (I SC ) were measured using typical four-electrode voltage clamp with Ag/AgCl electrodes (World Precision Instruments, Sarasota, FL) connected to the solutions through agar bridges containing 1 M KCl. Positive currents were defined as cation movement from mucosa to serosa or anion movements in the opposite direction. Chamber compartments were separately perfused with 5 ml of Krebs-Henseleit solutions. A serosal-to-mucosal Cl Ϫ gradient was used to increase the electrochemical driving force for Cl Ϫ secretion across the apical membrane. The basolateral solution contained (in mM) the fol-lowing: 120 NaCl, 25 NaHCO 3 , 5 KCl, 1.2 NaH 2 PO 4 , 5.6 glucose, 2.5 CaCl 2 , and 1.2 MgCl 2 . The mucosal Cl Ϫ -free solution contained (in mM) the following: 120 sodium gluconate, 20 NaHCO 3 , 5 KHCO 3 , 1.2 NaH 2 PO 4 , 5.6 glucose, 2.5 Ca(gluconate) 2 , and 1.2 MgSO 4 . Solutions were gassed with 95% O 2 and 5% CO 2 resulting in pH 7.4. Cl Ϫ -free ϩ Ca 2ϩ -free Ringer's had the same composition as the Cl Ϫ -free Ringer's but omitted Ca 2ϩ . Transepithelial I SC measured under these conditions was termed I Cl .
Optical Measurement of Mucous Fluid Secretion from Pig Tracheal Submucosal Glands-Methods for isolating and preparing tracheas for measurements of fluid secretion were similar to those published previously (25)(26)(27)(28). Briefly, trachea and lungs were obtained from carcasses of 6 -9-week-old Yorkshire pigs used for acute experiments. No pigs were sacrificed for the present experiments. Procedures for care and euthanization of pigs were approved by the Animal Research Ethics Board at the University of Saskatchewan. Pieces of trachea or bronchus (ϳ0.5 cm 2 ) were pinned mucosal side up, and the mucosa with underlying glands was dissected from the cartilage and mounted in a Sylgard-lined Petri dish, with the serosa in the bath (ϳ1 ml volume), the mucosa in air, and maintained at 35-37°C and high humidity, using a TC-324 single channel heater and DH-35i dish incubator (Harvard Apparatus, Saint-Laurent, Quebec, Canada). Cotton swabs and air-drying were used to clean and dry the tissue surface, and 20 -30 l of watersaturated mineral oil were placed on the surface. The tissue was superfused with warmed, humidified 95% O 2 , 5% CO 2 . 3O-C12 was diluted to final concentration with warmed, gassed bath solution and added to the serosal side by complete bath replacement; the maximal DMSO concentration in the preparation was Ͻ0.1%.
Droplets of fluid ϩ mucus within the oil layer were visualized by transillumination, and images were captured using a digital camera (MiniVid USB, LWScientific, Lawrenceville, GA) mated to a 1ϫ ocular of a stereomicroscope. Each image contained an internal reference grid to compensate for minor adjustments in magnification during the experiment. Secretion volumes were calculated as described previously using the formula: V ϭ 4/3( r 3 ), where r is radius (26).
To be included in the analysis, each droplet had to meet the following criteria: (a) circular outline so that a spherical shape could be assumed; (b) clear edges to allow accurate measurement of the radius; and (c) no fusion with neighboring droplets. Viability was tested at the end of each experiment by measuring the response to carbachol (1 M); glands that did not respond to carbachol were excluded from the analysis. Although the majority of the submucosal glands was quiescent before stimulation, some produced fluid spontaneously as they were warmed from room temperature and then became quiescent again by the time they reached 37°C. Consequently, the initial fluid volume observed at 37°C was subtracted from those measured during the last 10 min, and the net (subtracted) volumes were plotted and used to calculate secretion rates. The secretion rate was calculated by fitting the volume versus time plots with straight lines using linear regression, and slopes were taken as the secretion rates expressed in nl min Ϫ1 . Linear regressions were performed using at least four points, and the r 2 value wasϾ0. 8.
Measuring Ca cyto Using Fura-2 Imaging Microscopy-Cells grown on cover glasses were incubated with growth media containing 2 M fura-2/AM for 40 -60 min at room temperature and then washed three times with Ringer's solution to remove the extra dye. The Ringer's solution had the following composition (in mM): 145 NaCl, 5 KCl, 1.2 NaH 2 PO 4 , 5.6 glucose, 1.0 CaCl 2 , 1.2 MgCl 2 and 10 mM HEPES, pH 7.4. Ca 2ϩ -free Ringer's omitted CaCl 2 and included 100 M EGTA.
Fura-2-loaded cells were mounted onto a chamber on the stage of the imaging microscope and maintained at room temperature. Treatments with agonists were made by diluting stock solutions into Ringer's solution at the concentrations stated in the text. Fluorescence ratio imaging measurements of cytosolic Ca 2ϩ (Ca cyto ) were performed using equipment and methods that have been reported previously (33)(34)(35). Briefly, a Nikon Diaphot inverted microscope was used with a 40ϫ NeoFluar objective (1.4 NA). A CCD camera collected emission (Ͼ510 nm) images during alternate excitation at 350 Ϯ 5 and 380 Ϯ 5 nm using a filter wheel (Lambda-10, Sutter Instruments, Novato, CA). Axon Imaging Workbench 4.0 (Axon Instruments, Foster City, CA) controlled both filters and collection of data. Images were corrected for background (region without cells). Data have been reported as 380:350 ratio, a measure of Ca cyto (36).
Measuring cAMP cyto and Ca ER Using Genetically Targeted FRET Imaging Microscopy-A genetically encoded cAMP sensor was used for single cell imaging of cAMP (15,37,38). This sensor (CFP-Epac(∂ DEP-CD)-YFP; called "Epac H30") is a monomeric construct that relies on FRET between YFP-and CFP-labeled portions of the cAMP-binding protein, Epac1 (38). Upon cAMP binding, the Epac probe undergoes a conformational change that alters the intermolecular distance between 3O-C12 HSL on Cl ؊ Secretion, Ca 2؉ , and cAMP the fluorophores, reflected by a change in FRET. Epac H30 has been modified to be catalytically inactive against its normal target, repressor activator protein (Rap1), and is also missing domains required for membrane association, rendering it an inert, soluble probe for cytosolic cAMP. The methods used for measuring cAMP have been described previously (15,37). Subconfluent cultures of CaLu-3 epithelial cells were grown on glass coverslips and transfected with Epac H30 using Effectene (Qiagen, Valencia, CA). After 24 -48 h, coverslips were mounted in a perfusion chamber on the stage of a Nikon TE2000 inverted microscope. Fluorescence images of cells expressing the sensor were acquired using Metafluor (Universal Imaging). Excitation pulses (440 nm) were generated using a microprocessor controlled filter wheel (Sutter Instruments, Novato, CA) and mercury light source. Pairs of fluorescence images collected alternately at 480-and 535-nm wavelengths were captured (Hamamatsu ORCA ER CCD camera) every 4 s and converted to a ratio image using the Metafluor software. FRET was expressed as ratio of CFP to YFP signals (480:535 nm). Cells were treated at the end of each experiment with 50 M forskolin ϩ 1 mM isobutylmethylxanthine (IBMX) to yield maximal [cAMP], and 480:535 ratios were compared with these maxima to yield relative increases in [cAMP] during treatments with 3O-C12 and forskolin.
ER luminal Ca 2ϩ (Ca ER ) was measured using the FRET-based D1ER cameleon (39). This probe has K d (Ca 2ϩ ) ϭ 60 M, suitable for measurements of Ca ER (39,40). Subconfluent cultures of CaLu-3 epithelial cells were grown on glass coverslips and transfected with a plasmid coding for D1ER as described above. After 24 -72 h, coverslips were mounted onto a chamber on the stage of a Nikon Diaphot microscope as described for fura-2 measurements above. The CCD camera collected alternate emission 530:470 nm images during excitation at 430 nm (S430/25ϫ, S470/30-m and S535/30-m filters from Chroma Technology, Bellows Falls, VT), controlled by Axon Imaging Workbench 4.0. All images were corrected for background (region without cells). Changes in Ca ER were expressed as the YFP-to-CFP emission ratio (530:470 nm). STIM1 Activation Measured with TIRF Microscopy-JME cells were plated on cover glasses and incubated in growth media for 24 h before co-transfection with STIM1-GFP and CD8-tagRFP. Cells were allowed to express for 48 h before experiments were performed. Total internal reflection fluorescence measurements were made to exclude fluorescence from the ER, and CD8-tagRFP was utilized to focus exclusively on the cell plasma membrane and ensure that the focal plane did not change during the course of the experiment. EGFP and tagRFP were alternately excited at 0.1 Hz by a 488-nm argon laser and 532-nm DPSS laser, respectively. Fluorescence emissions of GFP (525/50-nm bandpass filter, Chroma) and tagRFP (592/ 50-nm bandpass filter, Chroma) were acquired by an EMCCD camera (Andor iXon DV-897 BV) with 500-ms exposure per frame. Base-line images were acquired in standard Ringer's solution for 5 min before 10 M 3O-C12 (final concentration) was pipetted into the imaging chamber. 20 min after adding 3O-C12, 2 M thapsigargin (final concentration) was added, and images were taken for an additional 10 min. Fluorescence intensities were analyzed offline with ImageJ.
Statistics-Unpaired or paired t tests, Mann-Whitney test, and nonparametric repeated measurements, ANOVA and Dunn's multiple comparison test, were used to compare groups and effects, depending on the experiments; p Ͻ 0.05 was considered significant. Data have been presented as values from individual experiments or as averages Ϯ S.D. unless otherwise stated; n refers to the number of experiments.

3O-C12 Stimulates CFTR-dependent Cl Ϫ Secretion in Airway
Epithelial Cell Monolayers-3O-C12 was tested for effects on Cl Ϫ secretion by adding it to the apical side of CaLu-3 cells grown on filter inserts in Ussing chambers. Concentrations of 3O-C12 between 1 and 100 M were tested. To normalize the responses, cells were also treated with the adenylate cyclase activator forskolin to increase cytosolic cAMP and activate maximal rates of Cl Ϫ secretion. 10 M 3O-C12 caused, after a variable delay, slow increases in I Cl over the course of 45 min (Fig. 1A). These increases in I Cl were accompanied by decreases in R t , as shown by the increases in size of current pulses required to clamp transepithelial voltage from 0 mV (short cir- 3O-C12 HSL on Cl ؊ Secretion, Ca 2؉ , and cAMP NOVEMBER 5, 2010 • VOLUME 285 • NUMBER 45 cuit) to 1 mV, consistent with activation of Cl Ϫ channels (and Cl Ϫ secretion) in the cells. As also shown in Fig. 1A, subsequent treatment with a maximal dose of the adenylate cyclase activator forskolin (10 M) caused a further increase in I Cl . I Cl stimulated by 3O-C12 was largely blocked by the CFTR-specific inhibitor CFTRinh172 (Fig. 1A), indicating that the 3O-C12stimulated I Cl was likely dependent on the activity of CFTR. Further addition of the less selective CFTR blocker glibenclamide (or GLYH101, not shown) caused I Cl to decrease nearly to the initial value (Fig. 1A), indicating that there may have been another channel besides CFTR contributing to the I Cl response. These results also showed that these responses resulted from Cl Ϫ movements through the cells and not through tight junctions.
Concentrations of 3O-C12 between 10 and 50 M caused significant increases in I Cl that amounted to 50 -75% of I Cl stimulated by forskolin (Fig. 1B). However, responses to these con-centrations had large standard deviations, and although there was a trend to larger I Cl at higher [3O-C12], there were no significant differences in I Cl activated by 10, 15, 25, or 50 M 3O-C12. There were no effects of 1 M 3O-C12 on I Cl or R t . Experiments testing different [3O-C12] in single experiments were unsuccessful because responses to 3O-C12 were so slow that it was impossible to obtain traditional dose-response characteristics.
The specific role of CFTR in I Cl responses was tested using the CF cell line CFBE41o Ϫ and the stable CFTR-complemented CFBE41o Ϫ cell lines. 3O-C12 (Fig. 2, A-C) stimulated I Cl that was blocked by the CFTR inh 172 in CFTR-corrected CFBE41o Ϫ cells, but there was only a small or nonexistent stimulation of I Cl in the CF cells (CFBE41o Ϫ ). Results of experiments using the CFTR inhibitors and the CF versus CFTR-corrected airway epithelial cells showed that CFTR was mediating almost all of the Cl Ϫ secretion triggered by 3O-C12.

3O-C12 Increases Fluid Secretion by Pig Tracheal Glands-
The effect of 3O-C12 on native tissue was tested on pig airway submucosal glands. Preliminary experiments showed that submucosal glands did not respond consistently to 50 M 3O-C12, whereas 100 M 3O-C12 caused consistent stimulation. Experiments on intact airway submucosal glands therefore used 100 M 3O-C12. Fluid secretion was measured from images of surface droplets (Fig. 3A, inset). Typical time courses of fluid accumulation at the surfaces of nine glands from one piece of trachea are shown in Fig. 3A. There was no fluid secretion from any of these glands during 10 min of control measurements. 3O-C12 increased volume of bubbles with variable size (2-13 nl) during the first 10 -15 min, after which there were no further increases. Subsequent addition of carbachol (1 M) to the bath caused a further rapid increase in fluid accumulation in the bubbles, showing that the response to 3O-C12 was less than maximal as induced by this potent secretagogue. On average, the secretion rate during 10 min of control incubation was 0.04 Ϯ 0.01 nl min Ϫ1 ; 3O-C12 increased this rate to 0.37 Ϯ 0.09 nl min Ϫ1 (mean Ϯ S.E., 35 glands, 10 trachea preparations, p Ͻ 0.05 for comparison of 3O-C12 with control, Mann-Whitney Test), a rate that was 10 and 25% of the maximal secretion rates reported for the secretagogues carbachol and vasoactive intestinal peptide, respectively (28). Average results from 10 different preparations from four pigs are summarized in Fig. 3B. Because there was considerable scatter in the volumes of individual droplets, the average increase in volume did not reach statistical significance until 12 min following 3O-C12 addition.
3O-C12 and Thapsigargin Increase I Cl and Ca cyto in CaLu-3 Cells-The role of Ca cyto in mediating increases in Cl Ϫ secretion was tested by measuring I Cl during treatments with 3O-C12 and also with thapsigargin (inhibits ER Ca 2ϩ pump) to induce maximal release of Ca 2ϩ from the ER. By comparing responses to 3O-C12 and thapsigargin, we also tested the role of ER Ca 2ϩ in contributing to the 3O-C12-induced Ca cyto response. We tested 10 and 50 M 3O-C12; 10 M was the lowest [3O-C12] that elicited increases in I Cl , and 50 M was expected to elicit a larger effect. Results from experiments measuring Ca cyto in CaLu-3 cells grown on cover glasses and I Cl responses of CaLu-3 cells grown on filters to 10 or 50 M 3O-C12 and 5 M thapsigargin are shown in Fig. 4. 10 M 3O-C12 caused a small, slow increase in fura-2 ratio, and thapsigargin caused a faster, larger increase (Fig. 4A). 50 M 3O-C12 caused a larger but still slow increase in fura-2 ratio; following 50 M 3O-C12, there was only a small response to thapsigargin (Fig. 4B). When cells were treated first with thapsigargin, there were rapid, large, and transient increases in Ca cyto , and 50 M 3O-C12 had no further effect on Ca cyto (Fig. 3C). An implication of these results was that 10 M 3O-C12 released a portion of the ER Ca 2ϩ store, 50 M 3O-C12 released most of the store, and thapsigargin released all of the Ca 2ϩ store.
Effects of 3O-C12 and thapsigargin on I Cl were similar to the effects on Ca cyto . 10 and 50 M 3O-C12 both caused small, slow increases in I Cl (Fig. 4, D and E). Subsequent addition of thapsigargin caused a further increase in I Cl in cells treated with 10 M 3O-C12 (Fig. 4D) but had almost no effect on cells that had been treated with 50 M 3O-C12 (Fig. 4E). When cells were treated first with thapsigargin, there were rapid, large, and transient increases in I Cl (Fig. 4F), but 25 M 3O-C12 caused a further activation of I Cl (by average of 16.7 Ϯ 3.2 A/cm 2 , n ϭ 3) in thapsigargin-treated monolayers, indicating a Ca 2ϩ -independent component. Overall, these results indicated that 3O-C12 increased Ca cyto in CaLu-3 cells by releasing Ca 2ϩ from the thapsigargin-releasable pool of Ca 2ϩ in the ER, consistent with previous measurements in fibroblasts (7) and macrophages (13).
3O-C12 Lowers Ca ER in CaLu-3 Cells-Previous experiments on fibroblasts showed that 250 -1000 M 3O-C12 caused equivalent increases in Ca cyto in both Ca 2ϩ -containing and Ca 2ϩ -free solutions, indicating that 3O-C12 was releasing Ca 2ϩ from internal stores (7). Similar results were obtained here for CaLu-3 cells grown on cover glasses as follows: 50 M 3O-C12 caused small, slow increases in Ca cyto in both Ca 2ϩ -containing (Fig. 4) and Ca 2ϩ -free solutions (data not shown). This 3O-C12-induced increase of Ca cyto could have resulted from the Ca 2ϩ release from the ER or from the mitochondria or other organelles. The role of the ER in the Ca cyto response was tested by measuring Ca ER in CaLu-3 cells that were grown on cover glasses, transfected with ER-targeted cameleon D1 (39 -41), and imaged during treatment with 3O-C12 and thapsigargin. As shown in Fig. 5A, cameleon D1 appeared to be localized to the ER throughout the cell. As shown in Fig. 5B, cameleon FRET 530:470 ratio (proportional to Ca ER ) was reduced in a dose-dependent manner by 3O-C12, with a small effect at 10 M and a larger effect at 50 M. Subsequent addition of a maximal concentration of thapsigargin (10 M) caused a further small reduction of Ca ER . As shown in Fig. 5C, when cells were treated first with thapsigargin, there was a large drop in Ca ER , and subsequent addition of 50 M 3O-C12 had no further effect. A summary of the effects of 3O-C12 and thapsigargin on Ca ER is shown in Fig. 5D. Similar results were obtained in the CF nasal cell line JME/CF15, and in these experiments 50 and 100 M had similar effects on Ca ER , indicating that 50 M gave a maximal response (data not shown). These experiments showed that 3O-C12 released Ca 2ϩ from the thapsigargin-releasable pool of Ca 2ϩ in the ER in a dose-dependent fashion. Maximal release of Ca 2ϩ by 50 M 3O-C12 was slightly less than that released by 10 M thapsigargin.
3O-C12 Activates the IP 3 Receptor in the ER Nuclear Membrane of DT40 Cells-The 3O-C12-induced release of Ca 2ϩ from the ER could have resulted from an effect to inhibit the Ca 2ϩ pump/ATPase and/or from an effect to activate Ca 2ϩ release through the IP 3 receptor or other leak pathways in the ER. The functional consequences of 3O-C12 were explored by recording single IP 3 R channels in native ER membranes by patch-clamp electrophysiology of outer membranes of nuclei isolated from DT40-KO-r-IP 3 R-3 cells, as described previously (32,42). The pipette solution, to which the cytoplasmic face of the channels faced, contained 1 M IP 3 , 2 M Ca 2ϩ , and 0.5 mM ATP, suboptimal conditions that result in channel open probability (P o ) Ϸ0.05 (typical traces in Fig. 6A, summary in Fig. 6B). Addition of 10 M C12 to the pipette solution enhanced channel activity by ϳ5-fold (P o ϭ 0.27 Ϯ 0.05). Channel activity was further enhanced by 100 M C12 (P o ϭ 0.36 Ϯ 0.05), comparable with that activated by 1 M Bcl-x L (P o ϭ 0.47 Ϯ 0.04), a potent activator of the IP 3 R (32). In contrast, 100 M C12 was without effect in the absence of IP 3 (Fig. 6B). These data suggest that C12 activates IP 3 R channel gating by enhancing sensitivity of the channel to low levels of IP 3 .
3O-C12 Activates STIM1 in JME/CF15 Cells-Based on experiments in intestinal cells, Lefkimmiatis (15) proposed that reductions in Ca ER activate the ER resident protein STIM1, which associates with the plasma membrane and activates adenylate cyclase and production of cAMP. We tested whether the 3O-C12-induced release of Ca 2ϩ from the ER resulted in activation of STIM1 and activation of cAMP production in airway epithelial cells. We used JME/CF15 cells for most of these studies because transfections with plasmids yielded more consistent expression in these cells. Similar results were obtained with CaLu-3 cells (data not shown). JME/CF15 cells grown on cover glasses and transfected with STIM1-GFP showed ER-like expression throughout the cell (Fig. 7). STIM1-GFP migration to the plasma membrane was then measured using TIRF microscopy to assess STIM1 activation. Control cells showed only a few bright STIM1 puncta at the plasma membrane (Fig. 8A). 10 M 3O-C12 caused an increase in bright membrane-localized STIM1 puncta over the course of 10 -15 min (Fig. 8B). A further addition of 2 M thapsigargin caused a further increase in bright STIM1 puncta at the

3O-C12 HSL on Cl ؊ Secretion, Ca 2؉ , and cAMP
cell membrane (Fig. 8C). Quantitation of the time courses of these responses was performed by measuring total TIRF fluorescence during the treatments (Fig. 8D); 3O-C12 caused an increase in steady state TIRF-measured STIM1 fluorescence, and this was further increased by thapsigargin. The transient increases followed by reductions in TIRF fluorescence after each change of condition likely resulted at least in part from bleaching of the GFP associated with exposure to the laser excitation. Control experiments showed that these transients did not result from changes in fluid volume resulting from additions to the chamber.
3O-C12 Increases cAMP cyto in CaLu-3 Cells-3O-C12-induced increases in CFTR-dependent Cl Ϫ secretion were consistent with 3O-C12 stimulating cAMP/PKA-activating CFTR (43,44). ELISA-based assays were used in initial studies to measure cAMP cyto . Although low [forskolin] (100 -200 nM) rapidly stimulated small increases in I Cl (similar in magnitude to those activated by 3O-C12), there were no detectable changes in cAMP content using this method. We reasoned that the ELISA method was insufficiently sensitive to measure small changes of cAMP cyto that may occur with 3O-C12. We therefore performed single cell measurements of cAMP cyto using the FRET-based cytosolic sensor for cAMP, Epac H30. Expressed Epac H30 was distributed throughout the cytosol of CaLu-3 cells grown on cover glasses (Fig. 9A). 3O-C12 (33 M; n ϭ 19) produced a slow but significant (p Ͻ 0.001) increase in the Epac H30 480:535 nm emission ratio (a measure of cytosolic [cAMP]). As shown in Fig. 9B, this effect of 3O-C12 was smaller than that elicited by a submaximal concentration of forskolin (2 M; ratio change following 3O-C12 was 32.2 Ϯ 4.8% of forskolin, n ϭ 12; p Ͻ 0.0001) and much less (22.9 Ϯ 3.6%) than that produced by a supramaximal stimulation with forskolin (50 M) ϩ IBMX (1 mM; n ϭ 16; p Ͻ 0.0001). The average changes in ratio (absolute value) for the treatments (ϮS.E.) are summarized in Fig. 9C. Overall, these experiments showed that 3O-C12 increased cytosolic [cAMP], but these increases were slightly smaller than those elicited by 2 M forskolin and much smaller than those elicited by forskolin ϩ IBMX.
A further test of the role of cAMP in mediating 3O-C12stimulated I Cl was to incubate CaLu-3 cell monolayers grown on filters with reagents to either decrease or increase 3O-C12triggered cAMP cyto . I Cl responses to 3O-C12 were first measured in the presence of the selective cAMP antagonist (R p )-Br-cAMPS. CaLu-3 cells were left untreated (control) or treated with (R p )-Br-cAMPS (50 M) for 12 h prior to treatment with 3O-C12 (10 M). 3O-C12 (Fig. 10A) alone triggered typically small, slow increases in I Cl . Subsequent treatment with forskolin caused rapid increases followed by a reduction to a secondary, elevated plateau. Treatment with (R p )-Br-cAMPS reduced responses to 3O-C12 and 3O-C12 ϩ forskolin ( Fig. 10B; summary in Fig. 10C).
We also treated CaLu-3 cells with 100 M IBMX to block phosphodiesterase and potentially augment any 3O-C12-induced stimulations of cAMP production. Cells were treated with 3O-C12 (10 M) before or after IBMX. As shown in Fig.  11A, 3O-C12 caused typically slow increases in I Cl , and addition of IBMX caused a further rapid increase in I Cl . When IBMX was added first (Fig. 11B), there was a rapid increase in I Cl followed by a decrease to plateau. In the presence of IBMX, 3O-C12 caused a faster and larger response than occurred in the absence of IBMX. In the presence of both 3O-C12 and IBMX, forskolin (10 M) caused only a small increase in I Cl , consistent with the idea that 3O-C12 ϩ IBMX had elicited near maximal  . STIM1 localizes to the ER of JME/CF15 cells. JME/CF15 cells grown on cover glasses were transfected with a STIM1-GFP chimera plasmid and then examined 2 days later in the fluorescence microscope (excitation 490 nm, emission 520 -560 nm). Focus in the middle of the nucleus showed STIM1 expressed throughout the ER. Image typical of Ͼ10 similar transfections.
3O-C12 HSL on Cl ؊ Secretion, Ca 2؉ , and cAMP NOVEMBER 5, 2010 • VOLUME 285 • NUMBER 45 effects. A summary of the stimulatory effects of IBMX on 3O-C12-induced rate of increase of I Cl in control versus IBMXtreated cells is summarized in Fig. 11C. The more rapid stimulatory effect of 3O-C12 on I Cl in the presence of IBMX was consistent with the idea that IBMX was facilitating the stimulatory effect of 3O-C12. As summarized in Fig. 11D, there was no significant synergism in magnitudes of steady state responses between IBMX and 3O-C12, i.e. sum of the steady state I Cl responses to 3O-C12 (50%) ϩ IBMX (42%) were approximately equal to I Cl stimulated by 3O-C12 ϩ IBMX (92%) or by IBMX ϩ 3O-C12 (93%), indicating that 3O-C12 may have been eliciting its effects on I Cl mostly through effects on cAMP.

TPEN Increases cAMP cyto and I Cl in CaLu-3 Cells-Previous experiments on intestinal cells showed that the high K d Ca 2ϩ
chelator TPEN (K d (Ca 2ϩ ) ϭ 130 M) lowered Ca ER and stimulated cAMP production without altering Ca cyto (15). Similar tests were used in airway epithelial cells. Epac H30-transfected CaLu-3 cells grown on cover glasses were incubated in Ca 2ϩfree solution and treated with 1 mM TPEN. TPEN caused a slow increase in cAMP cyto that was ϳ40% of the maximum triggered by forskolin ϩ IBMX (Fig. 12A). A summary of the effects of 1 mM TPEN on cAMP cyto is shown in Fig. 12B. Further experiments showed that treatment of CaLu-3 cells with ionomycin in Ca 2ϩ -free solution, which also lowers Ca ER (Fig. 5), caused slow increases in cAMP cyto similar to those activated by TPEN (1 mM). 3 Because lowering Ca ER with 3O-C12, TPEN, or ionomycin increased cAMP cyto , it was predicted that TPEN would also stimulate I Cl . CaLu-3 monolayers grown on filters were mounted in chambers with normal Ringer's in the basolateral solution and Ca 2ϩ -free and Cl Ϫ -free apical solution so that TPEN could be added apically. As shown in Fig. 13A, 0.5 mM TPEN in the apical solution caused a slow increase in I Cl similar to that triggered by 3O-C12. Subsequent addition of 50 M 3O-C12 and then forskolin caused further small increases. The activated I Cl was blocked by CFTR inh 172, indicating a CFTR requirement. When 3O-C12 was added first (Fig. 13B), there was a typically slow increase in I Cl , and further addition of TPEN and forskolin caused further small increases. The rapid increase in I Cl when TPEN was added after 3O-C12 was consistent but unexplained. A summary of average steady state 3 I. Maiellaro and A. M. Hofer, unpublished data.

3O-C12 HSL on Cl ؊ Secretion, Ca 2؉ , and cAMP
I Cl in the different conditions is shown in Fig. 13C. TPEN and 3O-C12 elicited similar increases in I Cl , and effects of TPEN ϩ 3O-C12 were less than additive of the effects of adding TPEN or 3O-C12 individually.

DISCUSSION
3O-C12-activated Cl Ϫ and Fluid Secretion Requires CFTR Expression-A major conclusion of these experiments is that 3O-C12 stimulated CFTR-dependent Cl Ϫ secretion in airway epithelia. Consistent with this finding in the serous-like CaLu3 cell line, 3O-C12 also stimulated fluid secretion by pig tracheal submucosal glands. The results indicated that 3O-C12-stimu-lated Cl Ϫ secretion was accompanied by osmotically obliged fluid secretion from the glands, sites of a major fraction of the airway surface liquid in the upper airways (25,27,28). The secretion assay used for the present experiments does not test the role of tracheal surface cells in the response to 3O-C12, and further experiments will be needed to test this hypothesis.
Rates of fluid secretion measured optically in intact tissues in response to 100 M 3O-C12 were ϳ10 -25% of maximal rates, consistent with the electrophysiological measurements showing that 3O-C12-stimulated currents were also less than (50 -75%) of maximal rates (Fig. 1A). The need to use 100 M 3O-C12 to generate a fluid secretion response in the tracheal preparation indicated that intact tissue was less sensitive to 3O-C12 compared with tissue culture cells, which responded to [3O-C12] as low as 10 M. Although the precise [3O-C12] found in vivo remains unknown, the high lipid solubility of this molecule may ensure its equilibration into and action on cells throughout the epithelial surface in intact lungs that have been infected with P. aeruginosa.
3O-C12-stimulated Cl Ϫ secretion was inhibited by CFTR blockers in CaLu-3 cells and occurred in CFTR-corrected CFBE41o Ϫ but not CFBE41o Ϫ cells. 3O-C12-stimulated I Cl values were smaller than those elicited by high [forskolin] (2-10 M), often used to maximally stimulate CFTR in airway epithelia. These data indicated that 3O-C12 was activating CFTR-dependent Cl Ϫ secretion in airway epithelia. It has been argued (22) that airway epithelial Cl Ϫ secretion stimulated by bacterial products is an important aspect of the innate immune response in the lung airways. Fluid accumulation on the airway surface in response to 3O-C12 and other bacterial products should facilitate flushing of bacteria on the mucociliary escalator. This bacterial flushing would be largely absent in CF, but effects on other secretions, e.g. mucus (45) and IL8 (10), would be retained (34). This condition could lead to both decreased bacterial clearance and also increased white cell influx, thereby contributing to the apparent hyperinflammatory phenotype observed in CF (46).
3O-C12 Activates IP 3 R, Lowers Ca ER , and Increases Ca cyto -A second major conclusion from our experiments is that 3O-C12 activated the IP 3 R, decreased Ca ER , and increased Ca cyto , showing that 3O-C12-stimulated increases in Ca cyto resulted largely from IP 3 R-mediated release of Ca 2ϩ from the ER. Patch clamp measurements showed that 3O-C12 increased open times of IP 3 R3 in nuclear membranes of DT40 cells. DT40 cell nuclei were an appropriate model for these experiments because responses could then be attributed to IP 3 R3. Because channel activity was measured in isolated nuclei where local activation of IP 3 was unlikely, the results indicated a direct stimulatory effect of 3O-C12 on the IP 3 R, perhaps through an allosteric activation. If 3O-C12 increased the sensitivity of the IP 3 R to IP 3 , this would enable the IP 3 R to become more active under conditions of low [IP 3 ] (0.2 M) that may exist constitutively in unstimulated cells. It was notable that the stimulatory effect of 3O-C12 was large as follows: 10 M 3O-C12 increased P o to ϳ0.3 (5-6-fold over resting level) and 100 M increased P o to ϳ0.4, although maximal [IP 3 ] increased P o of IP 3 R3 to ϳ0.75 (31). 3O-C12 activation of IP 3 R in DT40 cell nuclei was consistent with cameleon D1ER FRET measurements in CaLu-3 cells showing that 10 M 3O-C12 released ϳ20% and 50 M 3O-C12 ϳ85% of ER Ca 2ϩ that was released by thapsigargin. These results indicated that in CaLu-3 cells 3O-C12 opened IP 3 R, increasing Ca 2ϩ leak from the ER and reducing Ca ER , even though the Ca 2ϩ -ATPase of the ER was still operating. When cells were treated with thapsigargin prior to 3O-C12, Ca ER was reduced to a low level (and Ca cyto was increased) so that 3O-C12 had little or no further effect on either Ca ER (or Ca cyto ). 3O-C12-induced reductions in Ca ER explain why 3O-C12 elicited Ca cyto responses in Ca 2ϩ -free solutions.
Previous experiments on fibroblasts showed that an inhibitor of phospholipase C blocked effects of 3O-C12 to elevate Ca cyto , indicating that 3O-C12 was releasing Ca 2ϩ from the ER by activating a G-protein-coupled receptor-phospholipase C complex that increased [IP 3 ] (7). One way to reconcile present and previous (7) results is to propose that the phospholipase C inhibitor reduced [IP 3 ] to low levels, so that the allosteric effect of 3O-C12 on IP 3 R was prevented. Further experiments will be required to determine the molecular details of 3O-C12 activation of IP 3 R and also whether other Ca 2ϩ regulators were being affected.
3O-C12 Activates STIM1 and Increases cAMP cyto -Two other important findings of these studies were that 3O-C12 both activated STIM1 (as seen from increases in TIRF fluores-cence in the plasma membrane) and triggered increases in cAMP cyto that were critical in mediating CFTR-dependent Cl Ϫ and fluid secretion by airway epithelia. The 3O-C12-triggered increases in cAMP cyto (Epac H30 FRET ratio) and I Cl were modest, comparable with the effects of low concentrations of forskolin. The effects of (R p )-Br-cAMPS to reduce and IBMX to increase I Cl responses to 3O-C12 were those expected if 3O-C12 were operating through cAMP/PKA.
Previous experiments in intestinal cells showed that complete emptying of the ER Ca 2ϩ store by thapsigargin activated STIM1 and cAMP (15). The present data extend these observations by showing that partial (ϳ20%) reduction of Ca ER by 10 M 3O-C12 also activated 3O-C12-triggered increases in STIM1 activity, adenylate cyclase, cAMP, PKA, and activation of CFTR. The conclusion that STIM1 activated adenylate cyclase rather than inhibiting phosphodiesterase is based on the observation that the phosphodiesterase inhibitor IBMX increased rather than decreased 3O-C12-stimulated I Cl . 3O-C12-induced increases in Ca cyto may also activate K ϩ channels, leading to hyperpolarization of the cells and increased electrical driving force for Cl Ϫ exit through CFTR across the apical membrane. Activation of STIM1 is also expected to stimulate opening of the plasma membrane Ca 2ϩ channel orai (47), which would cause sustained increases in Ca cyto . Although 3O-C12 caused similar increases in Ca cyto in Ca 2ϩ -containing and Ca 2ϩ -free solutions indicating that there was minimal activation of orai by 3O-C12 (7), further experiments are needed to test directly the role of orai activation in responses of airway epithelial cells to 3O-C12.

Store-operated cAMP Model Explains 3O-C12 Stimulation of CFTR-dependent Cl Ϫ and Fluid Secretion by Airway Epithelia-
A modification of the previously proposed store-operated adenylate cyclase-cAMP model (15) can explain the stimulatory effects of 3O-C12 on cAMP cyto and I Cl by airway epithelial cells (Fig. 14). In control conditions, I Cl is small because cAMP cyto and PKA activity are low, and CFTR is closed. As shown by the stimulatory effect of IBMX on I Cl in the absence of other stimulation, cAMP cyto is maintained low by phosphodiesterase(s) that cleaves cAMP produced constitutively. Ca ER is maintained at a normally high level (likely Ͼ300 M; see Refs. 40,48) by the Ca 2ϩ -ATPase that counters the continual loss of Ca 2ϩ through the IP 3 R or other Ca 2ϩ leak (shown by effects of thapsigargin to reduce Ca ER in otherwise untreated cells). 3O-C12 directly activates the IP 3 R, leading to loss of Ca 2ϩ from FIGURE 11. Effects of IBMX on 3O-C12-stimulated I Cl of CaLu-3 cell monolayers. Cells were treated with 3O-C12 (10 M) before or after adding IBMX (100 M). These treatments were followed by forskolin (Fsk) (10 M). A, 3O-C12 was added before IBMX (typical of six experiments). B, IBMX was added before 3O-C12 (typical of six experiments). C, summary of effects of 3O-C12 on rate of increase of I Cl in control (as in A) versus IBMXtreated (as in B) condition. Slope of I Cl versus time (A/cm 2 /min) was measured during the first 5 min of 3O-C12 treatment in control (3O-C12 alone) versus IBMX pretreatment (3O-C12 ϩ IBMX) following initial 3O-C12induced transients that occurred in some cases (see B) in the two conditions. Data are averages (ϮS.D., n ϭ 6). *, p Ͻ 0.05 for comparison with control; #, p Ͻ 0.05 for comparison with IBMX ϩ 3O-C12. D, summary of effects of 3O-C12 versus 3O-C12 ϩ IBMX and of IBMX versus IBMX ϩ 3O-C12. Averages (ϮS.D., n ϭ 6) of ⌬I Cl expressed as percent of change from control measured during 3O-C12 ϩ IBMX ϩ forskolin. ⌬I Cl (% forskolin) ϭ (I Cl (condition) Ϫ I Cl (control))/(I Cl (3O-C12 ϩ IBMX ϩ forskolin) Ϫ I Cl (control)) ϫ 100. *, p Ͻ 0.05 for comparison to 3O-C12; #, p Ͻ 0.05 for comparison to IBMX.

3O-C12 HSL on Cl ؊ Secretion, Ca 2؉ , and cAMP
the ER and reduction of Ca ER . Based on I Cl , IP 3 R patch clamp, and Ca ER measurements, it appears that the threshold concentration for 3O-C12-induced effects on cultured cells occurred at 10 M with maximal effects at 50 M. 3O-C12-induced reduction of Ca ER activates STIM1, which then migrates to the plasma membrane where it likely stimulates a still-to-be identified adenylate cyclase leading to accumulation of cAMP and activation of PKA and CFTR.
The stimulatory effects of the high K d Ca 2ϩ chelator TPEN were also consistent with the model. As shown previously in intestinal cells, TPEN lowers Ca ER and increases cAMP cyto without affecting Ca cyto (15). In CaLu-3 cells TPEN increased both cAMP cyto and I Cl . Furthermore, TPEN and 3O-C12 appeared to modulate the same pool of Ca 2ϩ . Thus, TPEN, and 3O-C12 both increased I Cl on their own, and effects of TPEN ϩ 3O-C12 were less than additive compared with the effects of adding TPEN or 3O-C12 individually.
One apparent inconsistency between our results and the model was that 3O-C12 increased I Cl in thapsigargin-treated cells (Fig. 4F), which should have produced maximal activation of STIM1 and adenylate cyclase. One possible explanation is that thapsigargin may not have released all the Ca 2ϩ from the ER, and 3O-C12 caused a further small Ca 2ϩ leak from the ER that then caused small reductions in Ca ER (Fig. 5C), increases in Ca cyto (Fig. 4C), and activation of adenylate cyclase, cAMP, and Cl Ϫ secretion (Fig. 4F). 3O-C12 may also have IP 3 R-Ca 2ϩ -in-dependent effects to stimulate I Cl . Further experiments will be required to resolve this issue.
In addition to effects on Cl Ϫ secretion, 3O-C12-triggered decreases of Ca ER and increases in Ca cyto and cAMP cyto could also contribute to multiple "downstream" effects of 3O-C12 in P. aeruginosa-infected patients, e.g. ER stress resulting from reduction in Ca ER (34,49), IL8 secretion resulting from increases in Ca cyto (33), mucous secretion resulting from increased Ca cyto and cAMP cyto (45), and apoptosis (7) resulting from Ca 2ϩ overload of mitochondria (50,51). In CF airways, the effects of 3O-C12 on IP 3 R, Ca ER , Ca cyto , STIM1, and cAMP cyto  3O-C12 HSL on Cl ؊ Secretion, Ca 2؉ , and cAMP NOVEMBER 5, 2010 • VOLUME 285 • NUMBER 45 will lead not to increased Cl Ϫ and fluid secretion to clear the bacterial product but instead to ER stress, secretions of IL8 and mucus, and apoptosis. Thus, the cross-kingdom effects of 3O-C12 on airway epithelial cell signaling could contribute to airway clearance in non-CF individuals and to pathogenesis in P. aeruginosa-infected CF patients.