A Sensitive Membrane-Targeted Biosensor for Monitoring Changes in Intracellular Chloride in Neuronal Processes

Background Regulation of chloride gradients is a major mechanism by which excitability is regulated in neurons. Disruption of these gradients is implicated in various diseases, including cystic fibrosis, neuropathic pain and epilepsy. Relatively few studies have addressed chloride regulation in neuronal processes because probes capable of detecting changes in small compartments over a physiological range are limited. Methodology/Principal Findings In this study, a palmitoylation sequence was added to a variant of the yellow fluorescent protein previously described as a sensitive chloride indicator (YFPQS) to target the protein to the plasma membrane (mbYFPQS) of cultured midbrain neurons. The reporter partitions to the cytoplasmic face of the cellular membranes, including the plasma membrane throughout the neurons and fluorescence is stable over 30–40 min of repeated excitation showing less than 10% decrease in mbYFPQS fluorescence compared to baseline. The mbYFPQS has similar chloride sensitivity (k50 =  41 mM) but has a shifted pKa compared to the unpalmitoylated YFPQS variant (cytYFPQS) that remains in the cytoplasm when expressed in midbrain neurons. Changes in mbYFPQS fluorescence were induced by the GABAA agonist muscimol and were similar in the soma and processes of the midbrain neurons. Amphetamine also increased mbYFPQS fluorescence in a subpopulation of cultured midbrain neurons that was reversed by the selective dopamine transporter (DAT) inhibitor, GBR12909, indicating that mbYFPQS is sensitive enough to detect endogenous DAT activity in midbrain dopamine (DA) neurons. Conclusions/Significance The mbYFPQS biosensor is a sensitive tool to study modulation of intracellular chloride levels in neuronal processes and is particularly advantageous for simultaneous whole-cell patch clamp and live-cell imaging experiments.


Introduction
GABA and glycine are the main inhibitory neurotransmitters in the nervous system. These neurotransmitters activate ionotropic receptors that flux chloride to produce inhibition of many cell types. However, in some cells, activation of these receptors produce excitatory effects through various mechanisms, including differences in chloride/bicarbonate permeability of GABA A channels [1,2,3], differences in chloride gradients [4,5,6] and altered intracellular chloride homeostasis [7,8,9,10,11,12], as well as synaptic integration mechanisms [2,3,13,14]. Depolarizing GABA A responses are crucial in early development for establishing and maintaining synaptic connections in neurons throughout the brain [15,16] and become hyperpolarizing after the induction of chloride cotransporter subtype 2 (KCC2) expression during postnatal development [11,17,18]. Modulation of chloride cotransporter function has also been implicated in nervous system disorders, including temporal lobe epilepsy [19] and neuropathic pain [20,21]. Clearly, regulation of intracellular chloride is becoming recognized as an important neuronal process in synaptic plasticity of neuronal circuits.
Many of the neurotransmitter transporters, such as the dopamine transporter (DAT), norepinephrine transporter and excitatory amino acid transporters have been demonstrated to elicit uncoupled chloride conductances in the presence of substrates [22,23,24,25]. Very little is known about the physiological role of these chloride conductances. Although the DATmediated chloride current has been shown to increase firing of DA neurons [25], it has been difficult to study this current using wholecell patch-clamp recordings from the soma when DAT proteins are localized to small neuronal processes [26]. Thus, development of a sensitive fluorescent biosensor for chloride that could be used for monitoring changes in chloride in processes of neurons would significantly enhance information obtained with simultaneous whole-cell patch-clamp experiments.
Fluorescence imaging of cellular events provides a noninvasive window into cellular function not available with other approaches. Fluorescent tools include both synthetic dyes and the GFP (AvGFP) derived fluorescent proteins used as both cellular markers and biosensors. Synthetic chloride-sensitive dyes include 6-methoxy-N-(3-sulfopropyl)quinolinium (SPQ), 6-methoxy-N-ethyquinolinium chloride (MEQ) and N-(ethoxycarbonylmethyl)-6-methoxyquinolinium bromide (MQAE). These dyes have been used to monitor intracellular chloride in neurons in vitro [2,10,27,28]. The fluorescence emission of yellow fluorescent protein (YFP), a derivative of GFP, has been shown to respond rapidly and reversibly to changes in the concentration of small anions such as Cl 2 and has been used as a genetically encoded biosensor for chloride [29,30,31]. Clomeleon, a ratiometric chloride biosensor, is a chimeric protein combining the chloride-sensitive YFP with the chloride-insensitive cyan-fluorescent protein (CFP) and uses fluorescence resonance energy transfer (FRET) to monitor intracellular chloride concentrations in neurons and transgenic animals [31,6]. Point mutations in the YFP gene have been discovered that dramatically increase the affinity of YFP to chloride [30,32] and have been used to monitor the chloride conductance of cystic fibrosis transmembrane conductance regulator (CFTR) [33] and for high-throughput fluorescence screening of CFTR active compounds [34]. These mutations have also been incorporated into a chimera analogous to Clomeleon [31,35].
Genetically encoded biosensors have advantages when compared to currently available chloride-sensitive dyes. These advantages include excitation at visible wavelengths that avoid issues associated with phototoxicity caused by excitation wavelengths near the UV range, good signal-to-noise ratios at low concentrations, and insensitivity to gluconate and other physiological anions [32,35]. Other advantages arise from the fact that these biosensors can be specifically targeted to subcellular compartments, including the plasma membrane with addition of genetically encoded subcellular localization signal peptides.
The goal of this study was to demonstrate that attaching a palmitoylation site to the YFP variant with enhanced sensitivity to chloride developed by Galietta and colleagues (YFP-H148Q/ V163S or YFPQS [32]), targets the YFPQS protein to the plasma membrane (mbYFPQS) and that this provides enhanced functionality of the protein. Anchoring the biosensor to the membrane decreases leakage or diffusion of the fluorescent protein into the recording pipette during whole-cell recordings. In situ calibration of mbYFPQS expressed in midbrain neurons indicates that this construct has enhanced sensitivity to low chloride concentrations and is less influenced by changes in cellular pH compared to YFP and the FRET-based biosensor Clomeleon. These results confirm the increased chloride sensitivity of YFPQS developed by Galietta and colleagues and demonstrate the usefulness of this particular variant for detecting changes in intracellular chloride concentrations in neurons. In addition, the results show that mbYFPQS is suitable for monitoring changes in intracellular chloride concentrations in both the soma and processes of midbrain neurons in response to the GABA A agonist muscimol and DAT substrates.

Properties of mbYFPQS as a Chloride Sensor
In this study, the chloride sensor described by Galietta and colleagues, YFP-H148Q/V163S (YFPQS [32]) has been further modified by the addition of the N-terminal signal peptide of Neuromodulin. The addition of this peptide results in posttranslational palmitoylation of the protein, which facilitates the anchoring of the protein to the cytoplasmic face of membranes, including the plasma membrane [36]. Figure 1 shows a comparison of the cytoplasmic construct (cytYFPQS) and the membranetargeted construct (mbYFPQS) expressed in midbrain neurons.
To determine the sensitivity range of mbYFPQS to changes in intracellular chloride, midbrain cultures were perfused with an extracellular solution of varying chloride concentrations in the presence of ionophores to equilibrate extracellular and intracellular chloride concentrations as described (see Methods) [37]. The fluorescence of mbYFPQS was inversely dependent upon the concentration of chloride in the extracellular solution, with significantly reduced fluorescence at the highest chloride concentration (158 mM Cl 2 ) and greatest fluorescence in the absence of extracellular chloride (0 mM Cl 2 ). This is depicted in a plot of steady state fluorescence as a function of chloride concentration (Figure 2A). The average k 50 for all cells imaged was 41 6 3 mM (n = 17) and did not vary substantially from the k 50 measured from distal processes of 52 6 6 mM. The non-palmitoylated cytYFPQS had a k 50 = 59 6 18 mM (n = 8) when expressed in cultured midbrain neurons, comparable to both mbYFPQS and purified YFPQS [32].
pH Sensitivity of mbYFPQS A potential complication with YFP-based assays is that YFP fluorescence emission is also affected by pH because the anionsensing mechanism of YFP involves a shift in chromophore pK a [29,30]. To determine the contribution that pH fluctuations might have on intracellular chloride measurements using this sensor, mbYFPQS was expressed in midbrain neurons and the intracellular pH was adjusted by equilibration with ionophorecontaining buffers of varying pH in either 8 mM or 30 mM chloride ( Figure 2C). These pH concentrations cover the estimated intracellular chloride concentrations within midbrain neurons [38]. The apparent pK a s for mbYFPQS were 8.0 6 0.2 and 8.5 6 0.2 at 8 mM and 30 mM Cl 2 , respectively. These apparent pK a s were not significantly changed when permeabilizing the cells with b-escin (80 mM [39]) instead of ionophores with pK a s = 7.8 6 0.1 and 8.4 6 0.4 at 8 mM and 30 mM Cl 2 , respectively. The cytYFPQS had apparent pK a s of 7.4 6 0.1 and 7.6 6 0.1 at 8 mM and 30 mM Cl 2 , respectively when expressed in midbrain neurons, similar to results determined for purified YFPQS (pK a = 7.23 [32] and the apparent pK a of 7.3-7.8 for the new CFP-YFP ratiometric chloride indicator over the chloride concentration range of 10-60 mM [35]). Thus, the pKa of mbYFPQS is shifted to the right compared to the cytoplasmic constructs and only small differences in the fluorescence intensity were observed over a typical intracellular pH range (pH 7.5 to 7.2; Figure 2C).

Loss of Fluorescence during Whole-cell Patch-clamp Recordings
One advantage of the membrane-targeted construct is that it will not readily diffuse during whole-cell patch-clamp recordings. In order to test this directly, whole-cell recordings from HEK-293 cells expressing either cytYFPQS or mbYFPQS were made and the fluorescence intensity (200 ms exposure at 0.2 Hz) was monitored over time. The tau of fluorescence decay in recordings from cytYFPQS-expressing cells (76 6 17 s, n = 7) was significantly faster than when recording from mbYFPPQSexpressing cells (633 6 50 s, n = 4; t(9) = 13.01, p , 0.01; Figure 3). The faster decay suggests that the intracellular pipette solution dialyzes the cytYFPQS more readily than the mbYFPQS.

Detection of Changes in Intracellular Chloride with Activation of GABA A Receptors in Midbrain Neurons
In order to test the utility of mbYFPQS as a tool to monitor changes in intracellular chloride concentrations, we activated the GABA A -mediated chloride flux with bath application of muscimol (20 mM; Figure 4) in the presence of physiological extracellular solution (158 mM Cl 2 ). Muscimol superfusion of neurons expressing mbYFPQS resulted in both fluorescence increases and decreases highlighting the ability of mbYFPQS to detect fluorescence changes in both directions. In the majority of the cultured neurons (18/20), muscimol superfusion resulted in increased mbYFPQS fluorescence ( Figure 4A & B) suggesting that the chloride flux is outward, as expected from cultures of neurons taken from developmentally immature P2-P4 rats [11]. In contrast, the fluorescence decrease observed in a small subpopulation of cells is indicative of inward chloride flux, perhaps indicating a maturing cellular phenotype or a different sub- population of neurons ( Figure 4C & D). The muscimol-mediated fluorescence changes were inhibited by superfusion of the GABA A antagonist, bicuculline (10 mM) ( Figure 4D). In addition, the absolute value of the change in mbYFPQS fluorescence induced by muscimol in cells and processes of the midbrain neurons were comparable demonstrating that there is no substantial difference in the sensitivity of the sensor when expressed in these two regions. The muscimol-induced change in fluorescence was significantly reduced and/or reversed by the chloride exchange inhibitor bumetanide (25 mM; data not shown) suggesting that active transport of chloride establishes the chloride gradients observed in the neurons.

Detection of Amphetamine-mediated Changes in Intracellular Chloride
We previously demonstrated that DAT substrates stimulate an uncoupled chloride current associated with the DAT in midbrain DA neurons [25]. Since DAT are expressed in axons and dendrites of DA neurons, we were interested in determining if chloride changes to DAT substrates could be detected with mbYFPQS. Amphetamine (1-20 mM) superfusion elicited increases in mbYFPQS fluorescence in a subpopulation of the cultured midbrain neurons ( Figure 5A) that were reversed by the DAT-specific inhibitor, GBR129009 (10 mM). The change in mbYFPQS fluorescence elicited by amphetamine was dosedependent in both soma and processes of the responding neurons ( Figure 5B).

Stability of mbYFPQS for Simultaneous Imaging and Whole-cell Patch-clamp Recordings
In order to determine if mbYFPQS will be useful for simultaneous imaging and whole-cell patch-clamp experiments, cultured midbrain neurons were transfected with mbYFPQS and control experiments tested the photobleaching by repeated exposure to light excitation wavelengths for 30-40 min ( Figure 6A). Less than 10 6 5% (n = 5) decrease in baseline fluorescence occurred when imaging at a rate of 0.2 Hz and that rate of photobleaching was not significantly different during recordings with potassium gluconate internal solution (8 6 4% decrease; n = 4; Figure 6B).
The YFPQS construct has previously been shown to be differentially sensitive to different anions [32]. The DAT-mediated chloride current is also sensitive to anion substitution in the intracellular recording solutions [25]. For example, the DATmediated current is blocked by intracellular substitution with gluconate but methylsulfate ions have a similar permeability to chloride ions [25]. However, it is not known how these intracellular anions interact with the fluorophore in mbYFPQS. In order to test the sensitivity of mbYFPQS to these anions, cells were incubated in the cocktail of ionophores used for chloride calibration experiments (see Figure 2) but the extracellular solution was replaced with our typical internal solutions (pH = 7.35). The fluorescence of mbYFPQS in the presence of these solutions was compared to the quenching observed with the normal extracellular solution (158 mM Cl 2 ) after ionophore perforation of the membrane ( Figure 6D). The CsCl internal solution (134.2 mM 0.05). R square values for fits were 0.993 and 0.985 for the soma and process, respectively. C. pH calibration of mbYFPQS and cytYFPQS expressed in midbrain neurons. Following incubation of neurons in ionophores as in A & B, neurons were exposed to high potassium extracellular solutions containing either 8 mM or 30 mM Cl 2 at various pH (n = 5-8/group). Fluorescence of mbYFPQS was normalized to the fluorescence at pH 8.3 and plotted versus pH. Data were fit with sigmoidal curves to estimate apparent pK a . doi:10.1371/journal.pone.0035373.g002 Cl 2 ) slightly increased the fluorescence by 6 6 2% (n = 22). Potassium gluconate internal solution (12.6 mM Cl 2 ) increased fluorescence by 43 6 6% (n = 17) and potassium methylsulfate substitution (12.6 mM Cl 2 ) increased fluorescence by 56 6 7% (n = 17). The increase in fluorescence with potassium methylsulfate substitution is the expected increase from calibration experiments shown in Figure 2 with a shift in chloride concentrations from 132 mM to 12 mM. These results indicate that the gluconate and methylsulfate ions have little effect on mbYFPQS fluorescence and will be useful anion substitutions for whole-cell patch-clamp experiments.

Discussion
The fluorescence emission of yellow fluorescent protein (YFP) has been shown to respond rapidly and reversibly to changes in the concentration of small anions such as Cl 2 and has been used as a genetically encoded biosensor for chloride [30,40]. The results in this paper demonstrate that the point mutations in YFP described by Galietta and colleagues [32] increase the chloride sensitivity and are useful for monitoring changes in intracellular chloride. Intracellular chloride concentrations have been estimated to be 5-17 mM in the soma of midbrain neurons [38] but concentration gradients of .20 mM between soma and dendrites also have been observed in some neurons [6] suggesting that intracellular chloride gradients from 5-30 mM may be observed. We have further adapted the YFP variant (YFPQS) by addition of a palmitoylation sequence (mbYFPQS) that facilitates anchoring of the protein to the plasma membrane and have demonstrated that this construct has increased sensitivity to chloride (k 50 ,40 mM) after expression in cultured midbrain neurons compared to the parent YFP molecule (k 50 = 168 mM) and the FRET-based CFP-YFP chimera, Clomeleon (k 50 = 160 mM [6,31]). This increased sensitivity is similar to the cytYFPQS [32] and a variant with a triple mutation in YFP (k 50 ,30 mM; [35] and also is comparable to the best chloride dye that is commercially available (MQAE; k 50 = 2-40 mM [10,28]). In addition, the mbYFPQS has a shifted pKa compared to the cytYFPQS and the FRETbased sensors [35]. Finally, we demonstrate that the membrane targeting is advantageous for stabilizing the fluorophore during whole-cell patch-clamp recordings.
The advantages of a genetically encoded chloride sensor are significant when compared to conventional chloride-sensitive dyes, such as the quinoline-based fluorescent dyes SPQ, MEQ and MQAE. First, mbYFPQS can be directed to different compartments of the cell by generating a genetically encoded fusion protein with the biosensor and specific subcellular localization sequences. In this case, the mbYFPQS construct encodes a signal peptide from neuromodulin, which results in post-translational palmitoylation of the protein that facilitates attachment to the cytoplasmic face of cellular membranes. This is a particular advantage when opting for simultaneous fluorescence imaging and whole-cell patch-clamp electrophysiological experiments because the mbYFPQS does not diffuse into the pipette or leak out of cells similarly to cytoplasmic constructs or dyes. Although the quinoline-based dyes have good sensitivity in cells (SPQ k 50 = ,80 mM [27]; MEQ = ,50 mM [11,41]; MQAE = 2-40 mM [10,28]), insensitivity to pH and relatively low biological toxicity, they are problematic during long experiments because of fluorescence reduction due to leaking across the membrane and bleaching of the fluorophore. Thus, the most successful experiments monitoring changes in chloride with these dyes have been done using 2-photon microscopy [28,42]. On the other hand, the fluorescence of mbYFPQS is very stable over time compared to the chloride-sensitive dyes, MQAE [10] and MEQ [41] and the ratiometric FRET-based Clomeleon [6]. It is bright, very sensitive and the fact that the fluorescence excitation is within the visible light range limits cellular damage during experiments. The bright fluorescence also allows short exposure times (100-300 ms), a contributing factor to the minimal bleaching of the fluorescence during experiments using conventional epifluorescence wide-field microscopy.
The main disadvantage of all of the YFP-based chloride sensors is that the fluorophore is also sensitive to changes in pH in a near physiological range. Purified YFPQS has a pKa of 7.27 at 0 mM Clthat increases to 7.97 at 75 mM Cl - [32]. Clomeleon has a pKa range of 5.2 (0 mM Cl 2 ) to 6.5 (150 mM Cl 2 ) [31] and the triple mutation biosensor has a pKa of 7.1 to 8.0 over the same range [35]. Calibration of mbYFPQS expressed in midbrain neurons determined apparent pK a s of 8.0 and 8.5 at 8 mM and 30 mM Cl 2 , respectively. These pKas are shifted to the right compared to the cytYFPQS expressed in midbrain neurons and purified YFPQS [32], as well as to Clomeleonbased chloride sensors [31,35] suggesting that small physiologically relevant pH changes in the neurons will not significantly affect mbYFPQS fluorescence. We determined that direct measurements of changes in pH in the cultured midbrain neurons due to muscimolinduced GABA A currents were negligible using a GFP-based pH sensor (data not shown). In actuality, intracellular pH is tightly regulated [43] and may not contribute greatly to the fluorescence changes observed for mbYFPQS. However, it is possible that pathological processes may cause larger pH changes so direct measurements with selective pH sensors should be considered for each application.
In addition to the important role in development, alterations in chloride homeostasis have been shown to be important in regulating neuronal network activity in adults (for review, see [44]). Adult olfactory sensory neurons [28] and photoreceptors [45,46] have higher intracellular chloride concentrations than many cell types and typically display depolarizing effects of GABA A and glycine agonists. Even more interesting is the demonstration in several areas that GABA A responses can be modulated by transcriptional or translational regulation of chloride co-transporters. Some cells have a chloride gradient between the soma and axons [31] and/or dendrites [6,47], while the neurons in the suprachiasmatic nucleus can shift GABA responses on a circadian cycle [48]. Other studies have also shown that synaptic activity is sufficient to alter intracellular chloride concentrations [49] or membrane voltage to result in depolarizing responses to GABA [50]. Finally, disruption of chloride homeostasis due to downregulation of KCC2 may play an important role in abnormal neuronal activity associated with temporal lobe epilepsy [51] and neuropathic pain [52]. These studies suggest that there are multiple mechanisms by which neuronal activity is modulated by intracellular chloride.
Few studies have looked at chloride homeostasis in midbrain neurons even though GABA neurotransmission regulates activity of both dopaminergic and GABAergic neurons in the substantia nigra (SN) and ventral tegmental area (VTA) of the midbrain [38,53]. GABAergic neurons of the SN have been shown to be more sensitive to GABA A -mediated inhibition compared to dopaminergic neurons [54,55] because the GABAergic neurons express KCC2 whereas dopaminergic neurons do not [38]. These results suggest that adult dopaminergic neurons regulate chloride homeostasis differently than other midbrain neurons. This data is consistent with observations that the chloride equilibrium potential for GABA A receptor activation was more hyperpolarized than for the chloride current mediated by the dopamine transporter in processes of midbrain dopamine neurons [25]. Further studies will use expression of mbYFPQS in midbrain dopamine neurons to facilitate monitoring of intracellular chloride levels associated with DAT activity in neuronal processes combined with simultaneous whole-cell patch-clamping to study chloride homeostasis and endogenous DAT-mediated signaling.

Ethics Statement
Experiments were conducted in accordance with the animal care and use guidelines outlined in the Guide for the Care and Use of Laboratory Animals and approved by respective Institutional Animal Care and Use Committees (Washington State University ASAF#3711 and OHSU#1192).

Generation of Chloride Biosensor
The YFP with enhanced sensitivity to low chloride concentrations (YFPQS) was generated from pEYFP-mem (Clontech), which encodes the N-terminal palmitoylation signal sequence of neuromodulin fused to EYFP, by the following amino acid substitutions in the EYFP protein, H148Q (CAC R CAG) and V163S (GTG R AGC). The mutagenesis was verified by fully sequencing the EYFP gene in plasmid p-mbYFPQS.
Human Embryonic Kidney 293 (HEK293) cells were maintained in Advanced DMEM media (Gibco) supplemented with 10% fetal bovine serum and 100 mg/mL streptomycin and 100 U/mL penicillin at 37uC and 5% CO 2 in a humidified atmosphere. 24 hours prior to transfection, cells were trypsinized and plated on 12 mm coverslips coated in poly-L-lysine (BD

Imaging
Measurements of mbYFPQS fluorescence were performed by placing a coverslip of cultured cells in a continuous perfusion chamber on an inverted microscope equipped with a PixelFly cooled CCD camera, a Sutter High Speed Filterwheel and software control via InCyte acquisition software (Intracellular Imaging, Inc.) or a setup comprising an upright AxioExaminer, AxioCam MRm and Axiovision software for data acquisition (Carl Zeiss, Inc.) equipped with a Xenon (Intracellular Imaging, Inc.) or Mercury lamp (XCite 120Q, Lumen Dynamics Group) and filter sets for YFP fluorescence (Excitation 500/dichroic VIS .500-600/Emission 545; Omega Optical). Cells were bath perfused in normal extracellular buffer containing in mM: 146 NaCl, 5 KCl, 5 HEPES, 2.5 CaCl 2 and 1.2 MgCl 2 at pH 7.35 and room temperature.

Calibration of pH Dependence
For pH calibration, the high potassium solution with added ionophores was made at final chloride concentrations of 8 mM and 30 mM and varying pH. The fluorescence was normalized to fluorescence intensity at pH 8.3 and calibration curves were fit with a sigmoidal curve to estimate apparent pKa [35] for mbYFPQS under the two different chloride concentrations.

Determination of Whole-cell Patch-clamp-induced Quenching of mbYFPQS
The YFPQS construct has previously been shown to be differentially sensitive to different anions [32]. Since the wholecell patch-clamp recording technique introduces the pipette solution to the intracellular environment of the recorded neuron, three typical internal solutions were compared for their ability to quench mbYFPQS fluorescence. A cesium chloride internal solution (total chloride = 134.2 mM) contained in mM: 130 CsCl, 10 HEPES, 1.1 EGTA, 2 MgCl 2 and 0.1 CaCl 2 , pH 7.35. The potassium gluconate solution (total chloride = 12.6 mM) contained in mM: 138 KGluconate, 10 HEPES, 1 EGTA, 10 KCl, 1 MgCl 2 and 0.3 CaCl 2 , pH 7.35 and the potassium methylsulfate solution (total chloride = 12.6 mM) substituted potassium methylsulfate for the potassium gluconate.

Data Analyses
Average pixel intensity in each region of interest (ROI) was normalized to the baseline fluorescence intensity of each ROI by subtracting the fluorescence of a background ROI in the field and dividing by the initial fluorescence of each ROI. Percent (%) change in fluorescence was determined after background subtraction by [(fluorescence in drug)-(fluorescence in control)]/ (fluorescence in control) * 100%. Calibration data were plotted and curve fits to equations (above) were used to obtain values of k 50 and pKa (GraphPad Prism). All results are expressed as mean 6 STE and ANOVA with post-hoc tests when appropriate (GraphPad Prism).