A genetically-encoded chloride and pH sensor for dissociating ion dynamics in the nervous system

Within the nervous system, intracellular Cl− and pH regulate fundamental processes including cell proliferation, metabolism, synaptic transmission, and network excitability. Cl− and pH are often co-regulated, and network activity results in the movement of both Cl− and H+. Tools to accurately measure these ions are crucial for understanding their role under physiological and pathological conditions. Although genetically-encoded Cl− and pH sensors have been described previously, these either lack ion specificity or are unsuitable for neuronal use. Here we present ClopHensorN—a new genetically-encoded ratiometric Cl− and pH sensor that is optimized for the nervous system. We demonstrate the ability of ClopHensorN to dissociate and simultaneously quantify Cl− and H+ concentrations under a variety of conditions. In addition, we establish the sensor's utility by characterizing activity-dependent ion dynamics in hippocampal neurons.


INTRODUCTION
Chloride (Cl − ) and hydrogen (H + ) ions are fundamental to a wide range of processes within the nervous system including cell division, volume regulation, migration, metabolism, synaptic vesicle loading, network excitability, and fast synaptic inhibition (Tabb et al., 1992;Denker and Barber, 2002;Putney and Barber, 2003;Farrant and Kaila, 2007). Cl − and H + are often co-regulated (Russell and Boron, 1976;Kaila et al., 1989) and activity-dependent neuronal processes typically involve either the related, or independent, flux of both Cl − and H + ions (Chesler, 2003;Farrant and Kaila, 2007). Whilst intracellular Cl − concentration ([Cl − ] i ) and the negative logarithm of intracellular H + ion concentration (pH i ) are known to affect network excitability, network activity itself can generate shifts in the intracellular concentrations of these two ions (Isomura et al., 2003;Raimondo et al., 2012b). This reciprocal relationship means that tools to accurately and independently measure [Cl − ] i and pH i are important for understanding the separate and combined roles that these ions play during physiological and pathological network states.
Cl − or pH sensitive microelectrodes generated some of the earliest measurements of [Cl − ] i and pH i . However, due to their size and potential effect upon cell integrity, the use of these microelectrodes has been most successful in large invertebrate neurons. Fluorescent dyes have also been widely used to report both Cl − and pH. However, the most sensitive Cl − dyes, such as N-(ethoxycarbonylmethyl)-6-methoxyquinolinium bromide (MQAE), are not ratiometric and so do not measure absolute Cl − concentrations. In addition, these dyes tend to suffer from problems associated with toxicity, rapid bleaching, and cell leakage (Bregestovski et al., 2009). In contrast, the most popular pH sensitive dyes, such as 2 -7 -bis(carboxyethyl)-5(6)carboxyfluoroscein (BCECF) and the seminaphthorhodafluors (SNARFs), are well tolerated and offer ratiometric estimation of absolute pH. Nonetheless, these dyes can interfere with endogenous H + ion transport mechanisms and cannot be genetically targeted to cell types or subcellular compartments (Gatto and Milanick, 1993).
The discovery that green fluorescent protein (GFP) demonstrates inherent Cl − and pH fluorescence sensitivity marked the beginning of an effort to develop genetically-encoded reporters of Cl − or pH (Jayaraman et al., 2000;Kuner and Augustine, 2000). One approach has been to create fusions of two GFP mutants: yellow fluorescent protein (YFP) and cyan fluorescent protein (CFP). Whereas YFP emission is reduced by Cl − or H + binding, CFP fluorescence is relatively unaffected and serves as a fluorescence resonance energy transfer (FRET) donor for YFP. These fusion proteins are therefore useful as ratiometric reporters of either [Cl − ] i or pH i , and proteins within this group include Clomeleon, Cl-sensor, YFpH, and the pHlameleons (Kuner and Augustine, 2000;Awaji and Hirasawa, 2001;Esposito et al., 2008;Markova et al., 2008). Unfortunately, YFP's dual sensitivity to Cl − and pH complicates the interpretation of measurements using these reporters and means that if the concentration of both ions change, it is not possible to dissociate the underlying ion fluxes. This is compounded by the fact that cellular processes often involve concomitant changes in both Cl − and H + ion concentration (Russell and Boron, 1976;Kaila et al., 1989). Whilst other genetically-encoded pH indicators have been developed, such as pHlourin (Miesenböck et al., 1998) (RaGFP) and the deGFPs (Hanson et al., 2002), their susceptibility to artifacts based on Cl − sensitivity has not been well-characterized. This issue was addressed by the introduction of ClopHensor (Arosio et al., 2010)-a fusion protein that is able to simultaneously measure concentrations of both Cl − and H + ions and has been shown to function in heterologous cell lines (Arosio et al., 2010;Mukhtarov et al., 2013). ClopHensor is composed of the Cl − and pH sensitive GFP mutant, E 2 GFP, fused to monomeric DsRed. Here we identify problems associated with the use of ClopHensor in the nervous system. We therefore reengineer and improve this reporter, and present "ClopHensorN," a new, genetically-encoded ratiometric Cl − and pH sensor that is optimized to dissociate ion dynamics in neuronal cell types. We demonstrate the ability of ClopHensorN to simultaneously quantify Cl − and pH fluxes under a variety of conditions. Expression of ClopHensorN in hippocampal neurons enables us to dissect changes in intracellular Cl − and pH in response to network activity, which would not be possible with previous reporters.

DNA CONSTRUCTS AND SUBCLONING
The original ClopHensor construct (Arosio et al., 2010), and ClopHensor fused to two palmitoylation sites ("PalmPalm-ClopHensor"), were kindly provided by Daniele Arosio (University of Trento; Addgene plasmids 25938 and 25940). ClopHensor was moved into the expression vector pBJ1 under the control of a chicken beta actin (CAG) promoter and the DsRed monomer was replaced with either mCherry or tandem dimer tomato (tdTomato). The new constructs maintained the 20 amino acid linker sequence RGSASGGGGGLVPRGSASGA between E 2 GFP and the red fusion protein, with the addition of two amino acids (TG) at the C-terminal end of the linker. Western blotting experiments confirmed that both the original ClopHensor and the new ClopHensorN construct generated >90% full-length fusion proteins (data not shown). Upon publication the constructs presented here will be made freely available from the non-profit service Addgene (http:// www.addgene.org/).

SLICE PREPARATION AND DNA TRANSFECTION
Rat organotypic hippocampal slice cultures were prepared using a method similar to that described by Stoppini et al. (1991). Briefly, 7 day old male Wistar rats were killed in accordance with the UK Animals Scientific Procedures Act 1986. The brains were extracted and placed in cold (4 • C) Geys Balanced Salt Solution (GBSS), supplemented with D-glucose (34.7 mM). All reagents were purchased from Sigma-Aldrich, unless stated. The hemispheres were separated and individual hippocampi were removed and immediately sectioned into 350 μm thick slices on a McIlwain tissue chopper. Slices were rinsed in cold dissection media, placed onto Millicell-CM membranes and maintained in culture media containing 25% EBSS, 50% MEM, 25% heat-inactivated horse serum, glucose, and B27 (Invitrogen). Slices were incubated at 36 • C in a 5% CO 2 humidified incubator before transfection. Neurons were biolistically transfected after 4-7 days in vitro using a Helios Gene Gun (120 psi; Bio-Rad). Fifty microgram of target DNA was precipitated onto 25 mg of 1.6 μm diameter gold microcarriers and bullets generated in accordance with the manufacturer's instructions (Bio-Rad). Biolistic delivery of target DNA resulted in sparse transfection rates (typically less than 10 cells per slice) and recordings were performed 2-7 days post-transfection. For a subset of experiments ClophensorN was expressed in mouse cortex by in utero electroporation. In utero electroporation was carried out on E14.5 C57BL6/J (Jackson Laboratory) mouse embryos. Dams were anesthetized using isoflorane (3% for induction, 2-2.5% for surgery) and the uterine horns exposed by laparotomy. Each embryo was injected through the uterine wall with 0.5-1 ul ClopHensorN plasmid (2 ug/ul) in PBS with 0.03% fast green (Sigma) using a thin-walled glass pipette (WPI) pulled to a ∼50 μm tip. Paddle electrodes (Nepagene, CUY650) were used to deliver five 50 ms, 42 V pulses at 1 Hz from a square pulse generator (BTX, ECM 830). Embryos were kept moist during the surgery by applying warm sterile PBS. Following electroporation the uterine horns were replaced and the dam allowed to recover and litter as normal. At postnatal day 21 the mice were killed and the brain rapidly removed and placed in ice-cold (0 to +4 • C) artificial cerebro-spinal fluid (ACSF), bubbled with 95% O 2 /5% CO 2 . Coronal cortical slices (350-400 μm thickness) were cut using a vibrating microtome (Microm HM650V, Carl Zeiss Ltd) and slices were maintained in an interface chamber between humidified carbogen gas (95% O 2 , 5% CO 2 ) and ACSF (at 20-25 • C). After recovering for at least 1 h, the slices were mounted on coverslips (coated with 0.1% poly-L-lysine in ultrapure H 2 O) before being transferred to the recording chamber for imaging.
Pyramidal neurons within the CA1 and CA3 regions were visualized under a 40× water-immersion objective (Leica) and targeted for recording. Patch-clamp recordings were made using an Axopatch 1D or Axoclamp 2B amplifier (Axon Instruments). Data was acquired with WinWCP Strathclyde Whole Cell Analysis software (V.3.9.7; University of Strathclyde) before being exported to the MATLAB environment (MathWorks) for further analysis using customized scripts. Some statistical analysis was performed using GraphPad Prism version 5.00 (GraphPad Software). Data are reported as mean ± SEM. GABA A receptors were activated either by exogenous application of GABA or by electrical stimulation of GABAergic afferents. Short "puffs" of GABA (200 μM) were applied via patch pipette positioned close to the soma and connected to a picospritzer (20 psi for 20 ms; General Valve). Synaptic GABA A receptor activation was achieved by stimulating afferents using a bipolar stimulating electrode (Frederick Haer Company) placed in stratum radiatum, 300-400 μm from the recorded cell (Pouille and Scanziani, 2001). Simultaneous activation of glutamatergic receptors was prevented using bath application of kynurenic acid (2 mM).

IMAGING AND CALIBRATING INTRACELLULAR Cl − AND pH
Imaging was performed using an upright Leica SP2 AOBS laser scanning confocal microscope equipped with a 40x water immersion objective (NA 0.8). To determine whether expression of ClopHensor and ClopHensor-derived constructs (including PalmPalm-ClopHensor) resulted in corresponding expression of E 2 GFP and the fused red fluorophore (either DsRed, mCherry or TdTomato), cells were excited at 488 nm (for E 2 GFP, "green channel") and at 594 nm (for the red fluorophores, "red channel"). Emission was collected by separate photomultiplier tubes (PMTs): between 500 and 550 nm for the green channel and between 650 and 700 nm for the red channel. In a blinded manner, cells expressing each construct were counted. The number of cells demonstrating aggregates (examples Figures 1B,C) as a fraction of total expressing cells was recorded. Cells were counted as aggregated if there was an area of increased red fluorescence that was not matched by an increase in green fluorescence.
For intracellular Cl − and pH imaging ClopHensorN was used as a ratiometric indicator by excitation and was excited sequentially at 458, 488, and 594 nm. Emission was collected between 500 and 550 nm with a single PMT when excited at 458 and 488 nm, but between 650 and 700 nm with a second PMT when excited at 594 nm. Images were exported to the MATLAB programming environment where background was subtracted and fluorescence averaged within regions of interest selected from the soma of individual cells. To correct for fluctuations in laser intensity, a photodiode (sample rate 10 kHz) recorded laser power output during imaging (Zucker and Price, 2001;Arosio et al., 2010) and the resulting data was used to correct fluorescence ratios offline by a factor α (for R pH ) and α 2 (for R Cl ), see Figure 2B.
and α 2 = 0.5 ÷ (I 458 /I 594 ) where I 458 , I 488 , and I 594 are the laser powers measured by the photodiode for excitation with the 458, 488, and 594 nm lasers, respectively.
In order to determine pH i using ClopHensorN, laser power corrected excitation fluorescence ratios (αR pH ) were measured as follows: αR pH = F 488 /F 458 F 488 and F 458 are the fluorescence recorded using excitation with the 488 and 458 nm lasers. The formation of a 1:1 analyte-sensor complex results in an equilibrium described by the Grynkiewicz equation (Grynkiewicz et al., 1985;Arosio et al., 2010), which can be written for ClopHensorN as follows: αR A and αR B are the values of αR pH for ClopHensorN in its most acidic and basic forms, respectively. Likewise, F 458,A and F 458,B reflect the emission upon excitation at 458 nm, when the ratiometric indicator is in its acidic and basic form. pK a is the acid dissociation constant of the indicator. Due to the fact that ClopHensorN has a pH isobestic point at 458 nm, fluorescence is pH insensitive when excited at this wavelength. That is, F 458,A = F 458,B . Calibration data was therefore fitted using the following rearranged version of Equation 1: this allowed the pKa of ClopHensorN to be determined and pH i to be calculated from measured fluorescence ratios (αR pH ) during subsequent experiments.
Having determined pH i , [Cl − ] i could be calculated using the laser power corrected excitation fluorescence ratio (α 2 R Cl ) α 2 R Cl = F 458 /F 594 respectively. The Grynkiewicz equation for using ClopHensorN as a Cl − indicator can be written as: F 594,free and F 594,bound reflect the fluorescence after excitation with the 594 nm when the ClopHensorN is in its Cl − free and Cl − bound forms. However, as the fluorescence of TdTomato is insensitive to Cl − , F 594,free F 594,bound = 1 and Equation 3 can be simplified to: (Arosio et al., 2010). α 2 R free is the maximum value of α 2 R Cl , reflecting α 2 R Cl where no Cl − is bound to ClopHensorN. In contrast, α 2 R bound [pH i ] is the minimum value of α 2 R Cl , reflecting α 2 R Cl when ClopHensorN is saturated with Cl − . As demonstrated in Figure 2F, α 2 R bound [pH i ] was also found to depend Lasers exhibit fluctuations over seconds and minutes, which affects real-time ratiometric imaging by excitation and requires offline correction, as described previously (Zucker and Price, 2001;Arosio et al., 2010). A hippocampal neuron expressing ClopHensorN was imaged in the presence of ionophores and an ACSF that "clamps" intracellular pH (pH i ) and intracellular Cl − concentration ([Cl − ] i ) at 7.8 pH units and 8 mM, respectively, (see Materials and Methods). Despite pH being held at this constant value, significant apparent changes in R pH (F 488 /F 458 ) were recorded over a period of 330 s (top trace). A photodiode placed below the tissue was used to continuously record the emission power of the lasers, which were then used to calculate the intensity ratio for the 488 laser and the 458 laser (I 488 /I 458 , middle trace). Note the strong correlation between the fluctuations in this ratio and R pH . Bottom trace (red), R pH was corrected offline by a factor α, where α = 0.5÷ (I 488 /I 458 ). This resulted in a stable estimate of pH under these "clamped" conditions and confirmed the efficacy of this correction process. In all experiments αR pH was used to determine intracellular pH, and in a similar manner, α 2 was used to correct R Cl in order to determine [Cl − ] i (see Materials and Methods). (C) ClopHensorN calibration curve relating pH i to the pH sensitive excitation fluorescence ratio (αR pH ), following correction for laser power fluctuations. pH i was manipulated by adjusting extracellular pH in the presence of a H + permeable ionophore, data was fit using established equations (Grynkiewicz et al., 1985;Arosio et al., 2010) (Bizzarri et al., 2006;Arosio et al., 2010), pH changes influence the affinity of ClopHensorN for Cl − . K Cl d shifted from 8.3 mM with a fitted 95% CI between 3.6 and 13.0 mM (pH range 5.8-6.1) to 31.3 mM and 41.7 mM at pH ranges 6.8-7.2 and 7.3-7.6, respectively. The fitted 95% CI intervals for these K d s were 4.9 to 57.8 mM and −2.3 to 85.7 mM. The inset demonstrates the calculated dependence of K Cl d on pH (using Equation 6, See Materials and Methods). The 3 K Cl d values for the 3 pH ranges (horizontal bars) are shown. Utilizing this calibration data, ClopHensorN could be used to accurately and simultaneously report [Cl − ] i and pH i . on pH. Calibration data was then fitted using the following rearranged version of Equation 4: This allowed K Cl d [pH i ], α 2 R free and α 2 R bound [pH i ] to be determined by performing Cl − calibrations at different pH i values (see Figure 2F). K Cl d [pH i ] can be described according to the following equation (Arosio et al., 2007(Arosio et al., , 2010: 1 K Cl d reflects the Cl − dissociation constant when ClopHensorN is in its most acidic form (fully protonated) and was determined by fitting Equation 5 with K Cl d [pH i ] data derived from Equation 4. This allowed K Cl d [pH i ] to be determined for any pH i . α 2 R free was constant irrespective of pH i . The relationship between α 2 R bound [pH i ] and pH i was assumed to be linear and also fit using data from Equation 4.
No weighting was used during the fitting procedures. Calibration data was acquired from ClopHensorN-expressing cells in hippocampal slices. Intracellular pH and Cl − were controlled by equilibrating extra-and intracellular ion concentrations using the K + /H + exchanger nigericin (10 μM) and the Cl − /OH − exchanger tributyltinchloride (10 μM) in a high K + containing ACSF, according to the method described previously by Boyarsky et al. (1988). (2). Using these two solutions high K + containing ACSF

Recently Arosio et al. (2010) engineered a novel fusion protein capable of independently and simultaneously measuring
Cl − and pH. Named ClopHensor, this reporter is based on the fusion of a well-described pH and Cl − sensitive GFP mutant E 2 GFP (Bizzarri et al., 2006), with the pH and Cl − insensitive monomer DsRed ( Figure 1A). The corresponding intracellular expression of E 2 GFP and DsRed is critical to ClopHensor's function as a ratiometric reporter of intracellular ion concentration. To test whether ClopHensor could be used to measure Cl − and H + in neurons, rat hippocampal brain slices were biolistically transfected with ClopHensorexpressing plasmids (Figure 1). Expression of ClopHensor or PalmPalm-ClopHensor in hippocampal pyramidal neurons often resulted in uniform E 2 GFP expression, but highly heterogeneous DsRed expression that occurred as dense intracellular aggregations (Figures 1B,E,F). This was associated with compromised cellular morphology and is consistent with previous reports that DsRed is susceptible to aggregation, particularly within fusion proteins (Shaner et al., 2005). In an attempt to address this issue we designed new fusion proteins where the DsRed monomer in the original ClopHensor was replaced with either mCherry or tandem dimer tomato (TdTomato; Figures 1A,C,D). Interestingly, like the original ClopHensor, the E 2 GFP-mCherry fusion exhibited intracellular aggregations and a lack of colocalization between the two fluorophores ( Figures 1C,E,F). In contrast, the E 2 GFP-tdTomato fusion protein exhibited uniform and corresponding expression between the two fluorophores ( Figures 1D-F). Whilst the other fusion proteins showed aggregations in more than 60% of the cells examined, this was not the case in any of the E 2 GFP-tdTomato-expressing cells (Figures 1E,F). To further confirm that E 2 GFP-tdTomato is welltolerated, we delivered the construct in vivo by in utero electroporation and confirmed that this resulted in uniform and widespread expression in mature cortical neurons ( Figure 1G). Together these data confirm that the E 2 GFP-tdTomato construct is well-suited for studies in the nervous system and has the potential to afford ratiometric simultaneous reporting of neuronal Cl − and H + . We have named this new fusion protein "ClopHensorN."

ClopHensorN ALLOWS INDEPENDENT MEASUREMENT OF INTRACELLULAR Cl − AND pH
We then assessed the ability of ClopHensorN to independently report steady-state Cl − and pH in hippocampal pyramidal neurons. ClopHensorN was used as a ratiometric indicator of Cl − and pH by excitation (Arosio et al., 2010) and was excited sequentially via excitation at 458, 488 and 594 nm using a confocal microscope (Figure 2A). To compensate for fluctuations in laser intensity, a photodiode was used to record laser power output during imaging (Zucker and Price, 2001;Arosio et al., 2010) and the resulting data was used to correct fluorescence ratios offline by a factor α (for R pH ) and α 2 (for R Cl ) (see Figure 2B and Materials and Methods). Calibration of the reporter was performed by systematically varying extracellular Cl − and pH in the presence of H + and Cl − permeable ionophores, which are known to equilibrate intra-and extracellular concentrations of these two ions. αR pH was shown to depend upon intracellular pH with a pKa of 7.45 ( Figure 2C). Importantly, systematically varying the intracellular Cl − across the physiological range (0-65.5 mM), did not affect the αR pH measurements. There was no correlation between intracellular Cl − and αR pH for the three pH values tested (for pH 6.85-6.94, P = 0.15, for pH 7.14-7.22, P = 0.73 and for pH 7.14-7.22, P = 0.11, Pearson Correlation, Figure 2E). Thus, consistent with previous work, E 2 GFP and hence ClopHensorN provides a Cl − insensitive readout of pH (Bizzarri et al., 2006;Arosio et al., 2010;Raimondo et al., 2012b). In order to report Cl − , ClopHensor derived proteins make use of the fact that E 2 GFP possesses a pH isobestic point at 458 nm (Arosio et al., 2010). Fluorescence emission at this excitation wavelength is Cl − sensitive, but not pH sensitive. Like DsRed, TdTomato's insensitivity to both Cl − and pH (Shaner et al., 2004) make it a suitable partner for ratiometric imaging. ClopHensorN was therefore used to report Cl − by calculating the ratio of fluorescence emission when E 2 GFP was excited at 458 nm, over the emission when TdTomato was excited at 594 nm (R Cl = F 458 /F 594 ) and corrected for laser power fluctuations to generate α 2 R Cl . Consistent with our predictions, α 2 R Cl was strongly sensitive to [Cl − ] i ( Figure 2D). In agreement with previous studies using E 2 GFP, pH changes influenced the affinity of ClopHensorN for Cl − (Bizzarri et al., 2006;Arosio et al., 2010), which could be accounted for by using the family of calibration curves (Figure 2F; see Materials and Methods). At low pH, (range 5.8-6.1), whereby the sensor approaches full protonation, the affinity for Cl − is high with a K Cl d of 8.3 mM. Due to the fact that E 2 GFP requires a proton to be bound in order to bind Cl − , at higher intracellular pH the affinity of E 2 GFP for Cl − drops (Arosio et al., 2010). Indeed, at more physiological pH levels K Cl d was 31.3 mM (pH range 6.8-7.2) and 41.7 mM (pH range 7.3-7.6). In summary, ClopHensorN allowed absolute neuronal Cl − concentration and pH to be determined simultaneously from fluorescence ratios (αR Cl and α 2 R pH ), in a manner that is independent of protein expression levels. This was achieved by: (1) calculating pH in a region of interest from αR pH , (2) using this value to calculate the pH-dependent Cl − binding parameters of ClopHensorN, and (3)

FUNCTIONAL DISSOCIATION OF INTRACELLULAR Cl − AND pH MEASUREMENTS WITH ClopHensorN
Genetically-encoded reporters of Cl − , which are currently available for use within the nervous system, are unable to discriminate between changes in [Cl − ] i and pH i (Kuner and Augustine, 2000;Markova et al., 2008;Raimondo et al., 2012b). To verify that ClopHensorN is able to dissociate [Cl − ] i and pH i we performed a series of experiments in ClopHensorN-expressing CA1 and CA3 hippocampal pyramidal neurons, in which we induced different activity-dependent ion fluxes. First, as GABA A receptors are primarily permeable to Cl − and to a lesser extent HCO − 3 , strong GABA A receptor activation is predicted to generate a transient increase in [Cl − ] i and an acidic pH transient due to HCO − 3 efflux (Kaila et al., 1989). Consistent with this prediction, both agonist activation of GABA A receptors by application of GABA (200 μM, 20 ms) and synaptic activation of GABA A receptors via electrical stimulation of GABAergic afferents, resulted in a transient increase in [Cl − ] i and a modest acidification (Figures 3A,B). Second, we performed ion substitution experiments in which synchronous network activity was elicited by bathing the hippocampal slices in either a Mg 2+ -free ACSF or a Cl − -free ACSF; (see Materials and Methods). As the substitution of extracellular Cl − would remove the driving force for this ion to enter cells, we predicted that the two ACSF solutions would significantly alter the ion dynamics measured via ClopHensorN. Whilst imaging the ClopHensorN-expressing pyramidal neurons, synchronous network activity was monitored via a whole cell patch recording from a nearby pyramidal neuron (<200 μm between somata). Consistent with previous reports, network activity in the Mg 2+free ACSF generated intracellular acidic transients in the imaged neuron (Raimondo et al., 2012b) and robust accumulation of intracellular Cl − (Figure 3C), presumably due to Cl − infux via activated GABA A receptors (Ilie et al., 2012). In contrast, when the same neuron was imaged in the Cl − -free ACSF, synchronous network activity continued to generate acidic transients but the activity-dependent Cl − influxes were now replaced by smaller Cl − effluxes ( Figure 3D). This is consistent with the predicted change in transmembrane driving force for Cl − . Taken together, these sets of experiments confirmed the ability of ClopHensorN to dissociate changes in Cl − and H + in the nervous system.

ClopHensorN DISSOCIATES ACTIVITY-DEPENDENT Cl − AND pH DYNAMICS IN HIPPOCAMPAL NEURONS
Having established the potential of ClopHensorN to dissociate simultaneous ion dynamics in the nervous system, we conducted a quantitative study of the magnitude and kinetics of activity-dependent changes in [Cl − ] i and pH i . Synchronous network activity was induced in hippocampal slices by removing Mg 2+ from the slice perfusate. ClopHensorN-expressing CA1 and CA3 pyramidal neurons were imaged concurrently with whole cell patch-clamp recordings from neighboring cells to provide a simultaneous readout of network activity. Network events resulted in a highly significant positive shift in [Cl − ] i (P < 0.0001, t-test, n = 75 network events from 16 neurons) and an acidic shift in pH i (P < 0.0001, t-test, Figures 4A,B). Even short periods of network activity caused detectable shifts in intracellular Cl − and pH ( Figure 4A). For instance, network events lasting 1-3 s caused a peak increase in [Cl − ] i of 2.2 ± ] i and pH i measurements using ClopHensorN. Synchronous network events were induced in hippocampal brain slices via bath application of a Mg 2+ -free ACSF. A ClopHensorN-expressing hippocampal neuron was imaged whilst network events were monitored via a whole-cell patch recording from a nearby neuron (<200 μm between somata; black trace, top). The onset of network events (vertical dashed lines) correlated with the onset of intracellular Cl − accumulation (blue trace, middle) and intracellular acidification (red trace, bottom). (D) [Cl − ] i and pH i measurements from the same neuron in "c," following the replacement of the Mg 2+ -free ACSF with a Cl − -free ACSF. Network events (black trace, top) were again correlated with intracellular acidification. However, rather than Cl − influxes, smaller Cl − effluxes were now associated with the network events. 0.7 mM and a 0.017 ± 0.003 pH unit decrease in pH i (P = 0.008 and P < 0.0001, t-test). Furthermore, the magnitude of the peak increase in [Cl − ] i and decrease in pH i was tightly correlated with the length of the recorded network event ([Cl − ] i : r = 0.5792, P < 0.0001, pH i : r = −0.7487, P < 0.0001, Pearson Correlation, Figure 4B). The slope of the linear fit revealed a 0.43 mM increase in peak [Cl − ] i per second of network activity. Conversely, the slope of the linear fit for peak change in pH i revealed a decrease of 0.008 pH units per second of network activity ( Figure 4B).
Utilizing ClopHensorN we were able to detect statistically significant increases in neuronal [Cl − ] i within 1.5 s of the onset of network activity (the maximum sample rate; see Materials and Methods; P = 0.0007, n = 37 events, t-test, Figure 5A). Likewise intracellular acidification could also be detected within 1.5 s of the onset of elevated network activity (P = 0.0007 for neurons, t-test, Figure 5A). This demonstrates the rapidity with which intracellular Cl − and pH changes could be monitored using ClopHensorN under our experimental conditions. Lastly, simultaneous, dynamic readout of [Cl − ] i and pH i allowed us to independently compare the kinetics of activity-dependent changes to neuronal Cl − and pH levels. Whilst the time to peak shift of intracellular ion concentration was not different between the two ions (96.5 ± 4.7 and 104.6 ± 4.7% of network event duration, P = 0.089, t-test, Figure 5B), [Cl − ] i recovered to baseline levels significantly faster than pH i (219.6 ± 15.4 vs. 319.8 ± 14.7% of network event duration, P < 0.0001, t-test, Figure 5B). These experiments demonstrate the utility of using ClopHensorN for quantifying the size and temporal properties of Cl − and H + ion dynamics in neurons.

DISCUSSION
Here we present ClopHensorN-a genetically-encoded ratiometric Cl − and pH sensor that is optimized for use in the nervous system. ClopHensorN is shown to provide dynamic, simultaneous quantification of intracellular Cl − and H + concentrations under a variety of conditions. Furthermore, our measurements with ClopHensorN have identified previously unrecognized differences in the temporal dynamics of Cl − and pH concentrations, which become evident during periods of network activity. These features distinguish ClopHensorN from currently available reporters and highlight that ClopHensorN represents an important resource for monitoring ion dynamics in the nervous system. This reporter offers new opportunities to understand the regulation and roles played by Cl − and H + in processes such as synaptic transmission, cell morphology changes and metabolism.
ClopHensorN displays the cellular expression profile desired for a ratiometric reporter of intracellular ion concentration and exhibits signal to noise characteristics that are comparable to the best currently available reporters of either Cl − or pH (Raimondo et al., 2012b). ClopHensorN allowed us to detect significant intracellular Cl − and pH changes within our minimum sampling interval (1.5 s). In addition, the reporter displays good photostability with measurements exhibiting negligible drift due to photobleaching, which has been a limitation for other reporters (Bregestovski et al., 2009). These advantageous properties make ClopHensorN suitable for the measurement of both baseline and dynamic changes in Cl − and pH. As with other fluorescent proteins, there is the opportunity to use genetic techniques to target ClopHensorN to specific cell types or subcellular compartments, and to combine the reporter with other molecular tools.
Cl − and H + ions are often co-regulated (Russell and Boron, 1976;Kaila et al., 1989) and neuronal processes can generate fluxes of both ions (Chesler, 2003;Farrant and Kaila, 2007). Therefore, a fundamental advantage of ClopHensorN is that it generates measurements of both ion species and makes it possible to dissociate concentration changes that are occurring simultaneously. This feature of ClopHensorN was confirmed through ion substitution experiments and by detecting the intracellular Cl − and H + accumulation that accompanies intense GABA A receptor activation (Kaila et al., 1989). We then leveraged this capability to make the first simultaneous recordings of activitydependent intracellular Cl − and pH changes in the hippocampus. ClopHensorN revealed that network activity results in both intracellular Cl − and H + accumulation in neurons, but also revealed differences in the kinetics of these ionic shifts, with pH requiring longer to recover than Cl − changes. This is consistent with the idea that distinct cellular mechanisms contribute to the ongoing regulation of these ions and further illustrates the potential for the reporter to dissociate ion concentrations.
A further advantage of ClopHensorN over other genetic reporters is that it affords measurements of Cl − that are not confounded by pH. Currently, the two most popular geneticallyencoded Cl − sensors are Clomeleon and the Clomeleon variant Cl-sensor (Kuner and Augustine, 2000;Markova et al., 2008). Cl − estimates using both of these sensors are affected by pH, which means that concomitant H + fluxes will affect measurements. Using ClopHensorN we provide the first corrected measurements of [Cl − ] i within hippocampal neurons. Synchronous network activity was observed to cause rapid intracellular Cl − accumulation in neurons. The intraneuronal Cl − accumulation that accompanies synchronous network activity will depolarize the reversal potential for GABA A receptors (Raimondo et al., 2012a). And it has been suggested that this activitydependent shift in inhibition ("ionic plasticity") (Rivera et al., 2005;Raimondo et al., 2012c) may regulate NMDA-dependent mechanisms of synaptic plasticity (Staley et al., 1995), improve the coherence of gamma oscillations (Vida et al., 2006) or result in the conversion of physiological activity into pathological epileptiform states (Wright et al., 2011;Lillis et al., 2012).
A disadvantage of ClopHensorN is that it is a reporter by excitation, which means that laser excitation intensity should be measured and correction algorithms used to account for independent fluctuations in the power of confocal laser lines. Lastly, the fact that pH affects the affinity of ClopHensorN for Cl − means that relatively complex procedures (see Materials and Methods) are required to correct Cl − ratios for concurrent pH fluctuations, with the potential for measurement artifacts to occur if performed incorrectly.
Nonetheless, ClopHensorN is the only available geneticallyencoded sensor that is able to measure both Cl − and pH within the nervous system-the location for some of the most dynamic changes in intracellular ion concentrations. We foresee future work in which ClopHensorN is used to dissect Cl − and H + fluxes in specific cellular and subcellular compartments, and in the context of different processes in the nervous system.

ACKNOWLEDGMENTS
This manuscript is dedicated to the memory of the late Theresa Schlagheck, a special colleague and friend. We thank Daniele Arosio (University of Trento) for providing DNA constructs. We also thank the members of the Akerman lab for providing insightful comments and critically reading the manuscript. This work was supported by a grant from the Medical Research Council (G0601503), the Oxford Stem Cell Institute and the research leading to these results has received funding from the European Research Council under the European Community's Seventh Framework Programme (FP7/2007(FP7/ -2013, ERC grant agreement number 243273. Joseph V. Raimondo was supported by a Rhodes Scholarship, a NRF Innovation Postdoctoral Fellowship and a UCT URC Postdoctoral Fellowship. Bradley Joyce was supported by the Anatomical Society of Great Britain and Sarah E. Newey was supported by a Royal Society Dorothy Hodgkin Fellowship.