SEARCH

SEARCH BY CITATION

Key points

  1. Top of page
  2. Key points
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. References
  8. Appendix
  • • 
    Changes in pH occur within neurons during nerve activity and in response to hypoxic insult.
  • • 
    Many aspects of neurophysiology are potentially influenced by intracellular pH changes.
  • • 
    At the fruit fly larval neuromuscular junction, fluorescent genetically encoded pH-indicators (GEpHIs) revealed significant cytosolic acidification of presynaptic termini during nerve activity.
  • • 
    GEpHIs revealed that presynaptic pH changes occur in live intact larvae, indicating for the first time that such pH changes are not an artifact of experimental conditions.
  • • 
    The pH changes in presynaptic termini are substantial and are likely to influence synaptic function.

Abstract  All biochemical processes, including those underlying synaptic function and plasticity, are pH sensitive. Cytosolic pH (pHcyto) shifts are known to accompany nerve activity in situ, but technological limitations have prevented characterization of such shifts in vivo. Genetically encoded pH-indicators (GEpHIs) allow for tissue-specific in vivo measurement of pH. We expressed three different GEpHIs in the cytosol of Drosophila larval motor neurons and observed substantial presynaptic acidification in nerve termini during nerve stimulation in situ. SuperEcliptic pHluorin was the most useful GEpHI for studying pHcyto shifts in this model system. We determined the resting pH of the nerve terminal cytosol to be 7.30 ± 0.02, and observed a decrease of 0.16 ± 0.01 pH units when the axon was stimulated at 40 Hz for 4 s. Realkalinization occurred upon cessation of stimulation with a time course of 20.54 ± 1.05 s (τ). The chemical pH-indicator 2′,7′-bis-(2-carboxyethyl)-5-(and-6)-carboxyfluorescein corroborated these changes in pHcyto. Bicarbonate-derived buffering did not contribute to buffering of acid loads from short (≤4 s) trains of action potentials but did buffer slow (∼60 s) acid loads. The magnitude of cytosolic acid transients correlated with cytosolic Ca2+ increase upon stimulation, and partial inhibition of the plasma membrane Ca2+-ATPase, a Ca2+/H+ exchanger, attenuated pHcyto shifts. Repeated stimulus trains mimicking motor patterns generated greater cytosolic acidification (∼0.30 pH units). Imaging through the cuticle of intact larvae revealed spontaneous pHcyto shifts in presynaptic termini in vivo, similar to those seen in situ during fictive locomotion, indicating that presynaptic pHcyto shifts cannot be dismissed as artifacts of ex vivo preparations.

Abbreviations 
βb

bicarbonate-derived buffering power

βi

intrinsic (non-bicarbonate) buffering power

βT

total buffering power

4-AP

4-aminopyridine

AM

acetyloxymethyl

BCECF

2′,7′-bis-(2-carboxyethyl)-5-(and-6)-carboxyfluorescein

BES

N,N-bis(2-hydroxyethyl)-2-aminoethanesulfonic acid

CA

carbonic anhydrase

[Ca2+]cyto

cytosolic calcium concentration

[Ca2+]e

extracellular calcium concentration

CEDA-SE

(5-(and-6))-carboxyeosin diacetate succinimidyl ester

DM

dichroic mirror

DsRed

Discosoma red fluorescent protein

em

emission wavelength

ETZ

6-ethoxy-2-benzothiazolesulfonamide

ex

excitation wavelength

GEpHI

genetically encoded pH-indicator

HL6

haemolymph-like solution number 6

LGA

l-glutamic acid

MN

motor neuron

MN13-Ib

type-I ‘big’ bouton motor neuron terminal innervating muscle 13

pHcyto

cytosolic pH

pHe

extracellular pH

PM

plasma membrane

PMCA

plasma membrane Ca2+-ATPase

PtGFP

Ptilosarcus gurneyi green fluorescent protein

RA-pHluorin

Ratiometric pHluorin

SE-pHluorin

SuperEcliptic pHluorin

SE-pHluorin 2EM

SuperEcliptic pHluorin imaged in a dual-emission wavelength mode

SNR

signal-to-noise ratio

Introduction

  1. Top of page
  2. Key points
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. References
  8. Appendix

Like other biochemical processes, the presynaptic mechanisms responsible for neurotransmitter release and its plasticity are predicted to be pH sensitive. Endocytosis is likely to be pH sensitive as clathrin light chain polymerization is highly pH sensitive near neutral pH (Ybe et al. 1998). The activation of Ca2+/calmodulin-dependent kinase II is pH sensitive (Smith et al. 1992) as are many interactions in cyclic-AMP signalling pathways (Green & Gillette, 1988), suggesting that both CaMKII and cyclic-AMP-dependent forms of short-term synaptic plasticity are pH sensitive. The pH sensitivities of voltage-gated Ca2+ channel gating (Tombaugh & Somjen, 1997), synaptic vesicle filling (Goh et al. 2011) and Ca2+ buffering (Krizaj et al. 2011) have also been documented. However, data demonstrating the pH sensitivity of the mechanisms underlying neurotransmitter release and presynaptic plasticity in situ are scarce, as cytosolic pH (pHcyto) is difficult to measure and control in situ.

The pH sensitivity of presynaptic processes would be of little interest if homeostatic mechanisms maintained pHcyto within a narrow range, but data from diverse neuronal preparations show that significant transient pHcyto shifts occur during imposed activity (Willoughby & Schwiening, 2002; Chesler, 2003; Zhang et al. 2010; Svichar et al. 2011). The ‘set-point’ for pHcyto is ultimately dictated by the relative contribution of acid-loading and acid-extruding mechanisms (Boron, 2004). The proton buffering capacity of the cytosol is conferred by bicarbonate, phosphate and imidazole residues of cytosolic proteins (Casey et al. 2010). Significant pHcyto shifts arise when proton sinks or sources overwhelm cytosolic buffering capacity and plasma membrane (PM) acid transporters. Acidification of the neuronal cytosol may arise from increased activity of the PM Ca2+-ATPase (PMCA; Trapp et al. 1996; Schwiening & Willoughby, 2002), a PM Ca2+/H+ exchanger (Niggli et al. 1982; Thomas, 2009), while alkalinization is partially mediated by PM Na+-dependent and Na+-independent Cl/HCO3 exchangers (Baxter & Church, 1996; Svichar et al. 2009, 2011), and PM acid exchangers (Luo & Sun, 2007).

Technological limitations have impeded investigation of pHcyto shifts and their consequences for presynaptic function. Currently available chemical fluorescent pH-indicators are well suited for reporting pH changes from small volumes, but they are difficult to load with specificity or use in vivo. Genetically encoded pH-indicators (GEpHIs) offer specificity of loading and the opportunity for in vivo fluorescence imaging, thereby avoiding reliance on ex vivo preparations where changes in pHcyto may arise from disruptions in extracellular pH (pHe) homeostasis (Brechenmacher & Rodeau, 2000).

We expressed three different GEpHIs in Drosophila larval motor neurons (MNs), where they revealed acidification of the presynaptic cytosol during nerve stimulation and fictive locomotion. The dextran-conjugated chemical pH-indicator 2′,7′-bis-(2-carboxyethyl)-5-(and-6)-carboxyfluorescein (BCECF-dextran) corroborated these data. In situ calibration of the GEpHIs revealed that acidification could be substantial (∼0.30 pH units), and in vivo imaging revealed pHcyto shifts similar to those seen in situ. The large and rapid acid transients we observe are likely due to proton influx mediated by the PMCA as it clears large Ca2+ loads from the termini in exchange for protons. The bicarbonate-based buffering system is weakly catalysed by carbonic anhydrase (CA) in presynaptic terminals. This system can buffer acid loading during long periods of continuous nerve activity but not rapid acid transients seen during short high-frequency bursts.

Methods

  1. Top of page
  2. Key points
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. References
  8. Appendix

Fly stocks

Flies were raised on standard medium with dry yeast at 25°C. The  w1118 strain (stock obtained from Nancy Bonini) was used as a wild-type control. Each UAS-transgene was driven in MNs of Drosophila larvae using the enhancer-trap strain w1118; P[w+, OK6::Gal4] (Aberle et al. 2002).

Solutions and chemicals

Unless indicated otherwise, chemicals were purchased from Sigma-Aldrich. Haemolymph-like solution number 6 (HL6) [pH 7.3, 2 mm Ca2+, 7 mm l-glutamic acid (LGA; Cat. no. G1626; Sigma-Aldrich), 5 mmN,N-bis(2-hydroxyethyl)-2-aminoethanesulfonic acid (BES), 10 mm NaHCO3] was used in all experiments (Macleod et al. 2002) unless noted. HL6 was reconstituted on the day of each experiment from a frozen stock (‘stock HL6’) and not bubbled with CO2 or O2 unless otherwise noted. Stock HL6 nominally contains 10 mm HCO3, but in the absence of equilibration with CO2 most of the HCO3 would be lost (see eqns (4a) and (4b) below), and thus this solution should be treated as a functionally bicarbonate-free solution buffered only by BES. Stock HL6 and modified bicarbonate-free HL6 (19 mm BES, 0 mm NaHCO3) were titrated to pH 7.3 or 6.35 as indicated in individual experiments with HCl and NaOH. Titration was not necessary for modified bicarbonate-buffered HL6 (0 mm BES, 19 mm NaHCO3, pH 7.3).

Stock solution of rhod-2 acetyloxymethyl (AM) was made in Pluronic F-127 (20% Pluronic in DMSO; Cat. no. P3000MP; Invitrogen). Rhod-2 AM, rhod-dextran and BCECF-dextran were obtained from Invitrogen (Cat. nos R1244, R34677 and D1821, respectively). (5-(and-6))-Carboxyeosin diacetate succinimidyl ester (CEDA-SE; Cat. no. C22803; Invitrogen), 6-ethoxy-2-benzothiazolesulfonamide (ETZ; Cat. no. 333328; Sigma-Aldrich) and bumetanide (Cat. no. B3023; Sigma-Aldrich) were prepared in 20% Pluronic in DMSO as 10 mm, 200 mm and 100 mm stocks, and diluted to 10 μm, 200 μm and 100 μm, respectively, in HL6 for use. A 4 m stock of 4-aminopyridine (4-AP; Cat. no. A78403; Sigma-Aldrich) was prepared in water and diluted to 4 mm in HL6 for use.

Generation of flies expressing GEpHIs

cDNAs for SuperEcliptic pHluorin (SE-pHluorin) and Ratiometric pHluorin (RA-pHluorin) were provided by Gero Miesenböck. cDNA for Ptilosarcus gurneyi green fluorescent protein (PtGFP, Schulte et al. 2006) was purchased from Nanolight Technologies (Cat. no. 102; Pinetop, AZ, USA). cDNAs were cloned into a P-element vector (pUAST) then injected into w1118 Drosophila embryos by Rainbow Transgenic Flies (Newbury Park, CA, USA). Transgenes were mapped to specific chromosomes and balanced. Flies carrying a UAS-DsRed (Discosoma red fluorescent protein) transgene were obtained from the Bloomington Stock Center (Bloomington, IN, USA; BL# 6281). A single copy of each UAS-transgene was driven by a single copy of the MN driver OK6-Gal4 in all experiments.

Wide-field imaging

Wide-field microscopy was performed on an Olympus BX51WI microscope fitted with 100× (1.0 NA) or 60× (0.9 NA) water-immersion objectives. A Prior Scientific (Rockland, MA, USA) HF110 filter wheel controlled by a Prior Scientific Proscan III control unit was used to select excitation wavelengths from an Hg lamp. Images were captured with an Andor Technology (Belfast, Northern Ireland) EMCCD camera (DV887). A Cairn Instruments (Kent, UK) Optosplit II emission beam-splitter was placed before the camera for dual-emission wavelength imaging. Filters and dichroic mirrors (DMs) were obtained from Chroma Technology (Bellows Falls, VT, USA) or Semrock (Lake Forest, IL, USA). The imaging system was controlled through a Dell PC running Andor iQ software (ver. 2.0).

All imaging was performed at room temperature (∼24°C) on MN termini of axons innervating muscle 13 of abdominal segment 4 with type-Ib ‘big’ boutons (MN13-Ib), with regions of interest drawn around at least three boutons. The boutons are exclusively synaptic. Due to tissue-specific expression (in MNs) or forward-filling of fluorescence pH-indicators, all fluorescence signals come from the presynaptic compartment, and bouton fluorescence can be quantified without contamination from background structures.

PtGFP and RA-pHluorin were each imaged using an excitation (ex) switching system with alternating 470/40 nm and 406/20 nm exciter filters, a 500 nm DM, and a 530/40 nm emission (em) filter. BCECF-dextran was imaged by switching between 470/40 nm and 436/22 nm exciter filters, a 500 nm DM, and a 530/40 nm emission filter. SE-pHluorin, in dual-emission wavelength mode (SE-pHluorin 2EM), was imaged using a 406/20 nm exciter filter, a primary 440 nm DM, a secondary 500 nm DM mounted in the emission beam-splitter, and 485/35 nm and 536/30 nm emission filters. SE-pHluorin, in single-wavelength emission mode, was imaged simultaneously with rhod-2 AM, rhod-dextran, or co-expressed with DsRed using a dual-band (470/556 nm) exciter, a primary 500/530 nm bandpass DM, a secondary 600 nm DM mounted in the emission beam-splitter, and 512/25 nm and 630/92 nm emission filters. The ex/em maxima of SE-pHluorin (475/508 nm; Miesenböck et al. 1998) are distinct from those of DsRed (558/584 nm; Baird et al. 2000) and rhod (553/576 nm; Minta et al. 1989). Immature DsRed can emit green in addition to red fluorescence (Baird et al. 2000) and, though we observed weak fluorescence at 470/512 nm ex/em in larvae expressing only DsRed, this signal did not reduce the apparent responsiveness to pH in the SE-pHluorin/DsRed ratio.

During experiments that required continuous imaging during saline exchange and drug application, we mounted preparations in a custom-made bath connected to a gravity-driven flow-through system with a flow rate of ∼6 ml min−1 to achieve one volume exchange every 4 s. A gas mixture of 5% CO2/95% O2 or 100% O2 was bubbled into the loading reservoir as described in individual experiments.

Preparation of specimens

Female wandering 3rd instar Drosophila larvae were collected and fillet dissected in chilled HL6 to expose the longitudinal muscles of the body wall and the segmental nerves that project from the ventral ganglion. When a segmental nerve was cut (all data in Figs 1, 2, 3, 6 and 7A), it was cut close to the ventral ganglion and drawn into the lumen of a glass micropipette to stimulate MN axons as described previously (Chouhan et al. 2010). In preparations stimulated with the segmental nerves intact (Figs 4, 5, and 7B and C), the posterior ventral ganglion was cut from the cerebral ganglia to stop endogenous nerve activity (Cattaert & Birman, 2001), and a loop of the nerve was drawn into the lumen of a glass micropipette with the ventral ganglion still attached. For in vivo imaging, intact larvae were restrained between a glass slide and coverslip and observed using a 60× water immersion objective.

image

Figure 1. Genetically encoded pH-indicators (GEpHIs) report changes in pHcyto in presynaptic motor neuron (MN) termini in situ A, fluorescence images of termini of two MNs (MN13-Ib and MNSNb/d-Is) expressing different GEpHIs innervating larval body-wall muscle 13. The excitation and emission wavelength combination used for ratiometric imaging is indicated for each image. Fluorescence images were collected sequentially (PtGFP, RA-pHluorin) or simultaneously (SE-pHluorin 2EM, SE-pHluorin/DsRed). B, representative plots of changes in fluorescence intensity (ΔF/F) for each indicator in response to a 20 mm NH4Cl pulse (grey column). C, representative plots of the change in fluorescence ratio (R) in response to a 20 mm NH4Cl pulse. Fluorescence image pairs were collected at 1.66 Hz. Downward deflections represent cytosolic acidification.

Download figure to PowerPoint

image

Figure 2. Fluorescent pH-indicators reveal activity-dependent cytosolic acidification of presynaptic motor neuron (MN) termini  A, representative plots of changes in fluorescence intensity (ΔF/F) for each pH-indicator in response to nerve stimulation (grey bar). Fluorescence images were collected sequentially (PtGFP, RA-pHluorin, BCECF-dextran) or simultaneously (SE-pHluorin 2EM, SE-pHluorin/DsRed). B, plots of the average change in fluorescence ratio (R) during ratiometric imaging in response to stimulation (n= 6 preparations), except for the bottom panel where an average plot of SE-pHluorin/DsRed ΔR/R is presented (n= 6 preparations). Fluorescence image pairs were collected at 1.66 Hz. Downward deflections represent cytosolic acidification.

Download figure to PowerPoint

image

Figure 3. In situ calibration of fluorescent pH-indicators  A–C, calibration curves were constructed by measuring cytosolic fluorescence levels in nigericin-permeabilized MN13-Ib termini. For SE-pHluorin/DsRed and BCECF-dextran, each preparation was first exposed to saline titrated to pH ∼7, followed by a pseudo-random sequence of solutions titrated to various pH values before being returned to pH ∼7 (representative data from single preparations are shown in the left panels of A and B). For SE-pHluorin 2EM (C), a different preparation was used for each measurement at different pH values. Data traces from SE-pHluorin/DsRed were normalized to the maximum fluorescence (pH 9) for each preparation. D, averaged plots of pHcyto shift reported by each pH-indicator in response to 40 Hz nerve stimulation for 4 s (grey bar) (n= 6 per GEpHI, same data as Fig. 2B). The SE-pHluorin/DsRed ΔR/R, SE-pHluorin 2EM emission ratio and BCECF-dextran emission ratio were converted to pHcyto using the curve fits in AC. E, pHcyto changes as reported by each pH-indicator. Each pair of empty circles represents the resting pHcyto and pHcyto immediately after stimulation in a single preparation. Filled circles are mean values. Error bars (SEM). The resting pHcyto of SE-pHluorin/DsRed was assumed to be 7.30. All pH-indicators revealed significant acidification upon stimulation (P < 0.05). F, decrease in pHcyto calculated as the difference between resting value and peak response revealed by each pH-indicator. Error bars (SEM).

Download figure to PowerPoint

image

Figure 4. Bicarbonate-derived buffering does not buffer the rapid acidification phase of brief activity-induced acid transients but can buffer slow activity-induced acid loading  A, representative trace of pHcyto changes in MN13-Ib termini of a larva expressing SE-pHluorin/DsRed induced by stepwise NH4Cl pulses. pHe was lowered to 6.35 to reduce pHcyto to ∼7.15 so as to maximize the linear range of SE-pHluorin/DsRed (pKa= 7.16). The extracellular solution was modified bicarbonate-free HL6 buffered with 19 mm BES and bubbled with 100% O2 to calculate intrinsic buffering power (βi). B, plot of βi of motor neuron (MN) termini as a function of pHcyto. Data are from 6 larvae, each exposed to the NH4Cl pulse protocol in A. Data are grouped by each NH4Cl withdrawal step. The mean pHcyto during the acidification phase of each NH4Cl withdrawal step was used as the corresponding pHcyto value for the βi calculated during the same step. Error bars (SEM). C, average traces of pHcyto changes induced by long, low-frequency stimulation (black bar; 5 Hz, 60 s) in the presence of 200 μm ETZ and vehicle control. The bar graph shows the recovery decay constant and the change in pHcyto in both conditions. Error bars (SEM). *P≤ 0.05 (paired two-way Student's t test, n= 6). D, average traces of pHcyto changes induced by short, high-frequency stimulation trains (black bar; 40 Hz, 4 s) in the presence of 200 μm ETZ and vehicle (DMSO) control. Error bars (SEM). *Time point at which the values of the ETZ and control traces become significantly different (P≤ 0.05, two-way ANOVA, n= 6).

Download figure to PowerPoint

image

Figure 5. The presence of HCO3/CO2-derived buffering decreases the time course of recovery from activity-induced acid loads, but does not change the molar flux of acid equivalents during the rapid acidification phase  A, activity-induced pHcyto shifts in MN13-Ib termini of SE-pHluorin/DsRed larvae as measured in HL6 with different buffering conditions. Stock HL6 contained 5 mm BES and 10 mm HCO3, and was not bubbled with a gas mixture. Traces are averages of 6 preparations. Stimulation (black bar) was 40 Hz, 4 s in all cases. B, the decay constant (τ, top panel) of the recovery phase (grey portion of trace in A following stimulation) and the molar flux during the stimulation rapid acidification phase (indicated in black in A, not statistically different between groups, P≥ 0.20 by one-way ANOVA, n= 6) are compared between the different buffering conditions. Error bars (SEM). *P≤ 0.05 (one-way ANOVA). All data in A and B were collected at 1.66 Hz.

Download figure to PowerPoint

Loading synthetic Ca2+- and pH-indicators in situ

BCECF-dextran, a 10,000 MW dextran-conjugated pH-indicator (pKa= 6.9) and rhod-dextran, a low-affinity 10,000 MW dextran-conjugated Ca2+-indicator (Kd= 3.0 μm) were forward-filled separately (dye applied at 3 mm in each case; Macleod et al. 2002; Rossano & Macleod, 2007). Rhod-2 AM, a high-affinity Ca2+-indicator (Kd= 570 nm), was loaded by bath application, for which a 5 mm stock was made in Pluronic F-127 and further diluted to 5 μm in stock HL6. Preparations were incubated in Rhod-2 AM in the dark for 5 min at room temperature, rinsed in chilled stock HL6, and then imaged within 5 min.

In situ calibration of GEpHIs

BCECF-dextran and SE-pHluorin/DsRed were calibrated in situ using the nigericin/K+ method (Thomas et al. 1979). Larvae were fillet dissected and the segmental nerves were cut from the ventral ganglion. Calibration solution was stock HL6 adjusted to contain 20 μm nigericin, 130 mm K+, 0 mm Ca2+ and 7 mm LGA to equilibrate intracellular and extracellular [H+] while minimizing muscle contraction. These solutions were titrated to pH ∼4, 5, 6, 7, 7.5, 8 and 9 through the addition of HCl and NaOH in the case of BCECF-dextran, and ∼5, 6, 6.5, 7, 7.5, 8 and 9 in the case of SE-pHluorin/DsRed, as DsRed fluorescence became unstable at pH 4. Images were collected after a 10 min incubation in the first solution; all subsequent incubations were 5 min each. The calibration protocol of SE-pHluorin 2EM was adjusted to account for the instability of the fluorescence signals at 406/485 nm ex/em and 406/536 nm ex/em in the presence of nigericin. Larval muscle tissue exhibits considerable autofluorescence at 406/485 nm ex/em and 406/536 nm, and this autofluorescence bleaches over the course of long imaging protocols such as those needed to step a preparation through multiple calibration solutions. Changes in muscle autofluorescence confound background correction of the imaging data, and thus we were only able to use images collected during short (<2 min) exposures to 406 nm excitation light, which necessitated using a separate preparation for each calibration point. SE-pHluorin 2EM was calibrated by exposing 38 preparations to the saline described above titrated to a pH of ∼4, 5, 6, 6.5, 7, 7.5, 8 or 9 for 10 min. The calibration curves obtained for BCECF-dextran and SE-pHluorin 2EM were constructed by fitting a Boltzmann curve to scatter plots of imposed pH versus fluorescence values. Fluorescence ratio data from SE-pHluorin/DsRed preparations were normalized to the maximum fluorescence ratio observed in each preparation before being plotted as a function of imposed pH and a curve fit was applied using eqn (1). Relative fluorescence changes in the SE-pHluorin/DsRed ratio were converted to pHcyto measurements using eqn (2) and assuming a pHrest of 7.30, as this was the average pHrest indicated by SE-pHluorin 2EM. This same calculation was used to estimate pHcyto from SE-pHluorin measurements (ΔF/F) in the absence of DsRed.

Calibration Equations (Zhang et al. 2010)

  • image(1)
  • image(2)

where a=Fmax/Fmin from eqn (1), and K is generated from eqn (1).

NH4Cl pulse experiments and estimation of intrinsic buffering power

We used a multi-step NH4Cl pulse technique to determine the intrinsic (non-bicarbonate) buffering power (βi) of MN termini (Vaughan-Jones & Wu, 1990). Larvae expressing SE-pHluorin/DsRed were dissected in a flow-through chamber and bathed in bicarbonate-free modified HL6 (19 mm BES, 0 mm Ca2+, 0 mM NaHCO3, nominally CO2-free and bubbled with 100% O2) with pH adjusted to 6.35 so as to lower the pHcyto to 7.15 ± 0.03 (n= 6, near the apparent pKa of SE-pHluorin/DsRed (7.16); Fig. 3A). This adjustment was made to ensure that SE-pHluorin/DsRed fluorescence linearly corresponded to pHcyto through the full range of pHcyto values. Preparations were then exposed to 40, 30, 20, 10 and 0 mm steps of NH4Cl (∼20 s each), and βi was calculated using eqn (3) for each drop in NH4Cl concentration (Vaughan-Jones & Wu, 1990). Each value of βi was plotted against the average pHcyto observed during the corresponding NH4Cl concentration drop to construct a linear calibration curve of pHcytovs. βi (R2= 0.92, βi=−26.53*(pHcyto) + 205.65). 4-AP (4 mm) and bumetanide (100 μm) were added to all solutions in these experiments to minimize NH4+ leakage into the termini through K+ channels and transporters (Kirsch & Drewe, 1993). We saw no fluctuations in pHcyto during the plateaus of the NH4Cl pulses. We only corrected for active pH regulatory mechanisms during the final NH4Cl withdrawal from 10 to 0 mm by extrapolating the acidification upon NH4Cl withdrawal and the realkalinization to baseline with linear and single exponential curve fits, respectively. The intersection of these extrapolations was taken as the pHcyto minimum value (Aickin & Thomas, 1977).

  • image(3)

where C= change in concentration of weak base at the withdrawal (10 mm); pHi= intracellular pH at the point of NH4Cl withdrawal; pK′= dissociation constant of NH4Cl (9.24); ΔpHi= change in pHcyto upon NH4Cl withdrawal; and pHe= extracellular pH (6.35).

Calculation of net molar acid flux

Net molar acid flux upon stimulation was calculated according to eqns (5), (6) (Chesler, 1990) and (7) (Svichar et al. 2011). However, eqn (6) only accurately describes bicarbonate-derived buffering at equilibrium. Though bicarbonate-derived buffering (βb) contributes to total buffering (βT) at equilibrium, our data (Fig. 4C and D) indicate that the reactions of the CO2/HCO3 buffering system do not approach equilibrium quickly enough to buffer rapid (4 s) acid transients in Drosophila MN termini, and thus we approximated βT as equal to βi when analysing rapid acid transients. We used the value of βi corresponding to the resting pHcyto observed 2 s before the stimulus train in the analysis of all activity-induced acid transients.

  • image(4a)
  • image(4b)
  • image(5)
  • image(6)
  • image(7)

Image analysis and data processing

Movement correction was performed using the MultiRegStack plugin in ImageJ (1.37C) software. Background subtraction was performed in all images using Andor iQ 2.0 or ImageJ software. Ratiometric analysis of fluorescent signals was performed using ImageJ, Microsoft Excel and Andor iQ 2.0 software.

Statistical analysis and data presentation

SigmaPlot (version 10.0; Systat Software) was used for statistical analysis. For pairwise comparisons, two-way Student's t tests were used. One-way ANOVA (with a Bonferroni post hoc test) was used for comparisons of net molar acid fluxes and decay constants between three buffering conditions. Associations were tested by calculating Pearson's Product Moment Correlation coefficient. Differences were considered to be statistically significant at α-values of P≤ 0.05. Values are reported as mean ± SEM. Figures were generated in SigmaPlot and imported to Canvas (version X; ACD Systems) or Adobe Illustrator CS5.1 for panel assembly and labelling.

Results

  1. Top of page
  2. Key points
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. References
  8. Appendix

GEpHIs report pHcyto in nerve termini in situ

SE-pHluorin, RA-pHluorin and PtGFP were expressed in the cytosol of Drosophila larval MNs (Fig. 1A). There were no signs of reporter aggregation or partitioning in the cytosol of MN termini. SE-pHluorin was also co-expressed and imaged simultaneously with cytosolic DsRed, which shows little pH sensitivity from pH 5 to pH 11 (Baird et al. 2000). The ratio of SE-pHluorin and DsRed fluorescence imaged in this configuration allowed for movement correction, but could only be interpreted as a relative change from an established resting pHcyto. Variable aggregation of cytosolic DsRed in individual larvae prevented comparisons of the SE-pHluorin/DsRed fluorescence ratio between larvae.

Preparations were transiently (20 s) exposed to NH4Cl to induce controlled pHcyto shifts and confirm the pH-sensitivity of the GEpHIs (Fig. 1B). All GEpHIs displayed the predicted fluorescence changes in response to a prepulse with NH4Cl and report pHcytoin situ. Each pH-sensitive wavelength of each GEpHI reported an increase in pHcyto during exposure to 20 mm NH4Cl, followed by a decrease in pHcyto to below prepulse levels upon washout of NH4Cl. The ratio of the fluorescence signals from each GEpHI also reported this pattern of pHcyto shifts (Fig. 1C). Qualitatively, SE-pHluorin in 2EM mode and as SE-pHluorin/DsRed provided the best relative change in fluorescence and signal-to-noise-ratio (SNR).

GEpHIs reveal nerve stimulation-induced acidification in the cytosol of Drosophila motor nerve termini in situ

To determine whether pHcyto changes occur within presynaptic termini during nerve activity, we stimulated segmental nerves while imaging presynaptic termini in larvae expressing GEpHIs and larvae in which the termini were loaded with BCECF-dextran, a ratiometric chemical pH-indicator. Stimulation at the peak average endogenous fictive firing frequency (40 Hz; Chouhan et al. 2010) of MN13-Ib produced fluorescence changes consistent with presynaptic cytosolic acidification in all GEpHIs (Fig. 2A). However, the SNR and relative change in the fluorescence was only sufficient in SE-pHluorin (when imaged as either SE-pHluorin 2EM or SE-pHluorin/DsRed) to provide ratiometric measurements of pHcyto (Fig. 2B). Changes in the SE-pHluorin 2EM and SE-pHluorin/DsRed ratio were corroborated as changes in pHcyto by comparing them with those observed in nerve termini forward-filled with BCECF-dextran.

Quantification of presynaptic pH changes

Ratiometric imaging of these fluorescent pH-indicators allowed for conversion of the fluorescence ratios to pH values using an in situ nigericin/K+-based calibration technique, where pHcyto is equilibrated with pHe (Thomas et al. 1979). Preparations expressing SE-pHluorin/DsRed within MN termini as well as wild-type larvae forward-filled with BCECF-dextran were incubated in pH ∼7.3 saline, then exposed to salines buffered at various pH in a randomized sequence, and finally returned to pH ∼7.3 saline to confirm stability of the indicators over time (left panels in Fig. 3A and B). The BCECF-dextran excitation fluorescence ratio and the normalized changes in the SE-pHluorin/DsRed fluorescence ratio were plotted against the imposed pHcyto. Calibration curves were constructed using eqns (1) and (2) for the SE-pHluorin/DsRed data and a Boltzmann fit for the BCECF-dextran data (Fig. 3A and B, right panels). For SE-pHluorin 2EM we collected a single fluorescence ratio measurement from 38 preparations (see Methods), where each preparation was exposed to saline buffered to pH 4–9 for 10 min, then constructed a calibration curve with a Boltzmann fit (Fig. 3C).

Calibrated fluorescence measurements from BCECF-dextran and SE-pHluorin 2EM revealed a resting pHcyto of 7.27 ± 0.02 and 7.30 ± 0.02, respectively (n= 6) in MN13-Ib termini (Fig. 3D). A resting pHcyto of 7.30 was assumed when calibrating measurements obtained using SE-pHluorin/DsRed. During stimulation (40 Hz, 4 s), the cytosol significantly acidified to an average pH of 7.14 ± 0.01, 7.11 ± 0.01 and 7.14 ± 0.03, corresponding to changes in pHcyto of −0.16 ± 0.01, −0.16 ± 0.01 and −0.17 ± 0.01 pH units (n= 6) as measured by SE-pHluorin/DsRed, BCECF-dextran and SE-pHluorin 2EM, respectively (Fig. 3E and F).

Bicarbonate-derived buffering is weakly catalysed by CA in MN termini but does not influence acid loading during rapid acid transients

We did not expect to observe such rapid and large activity-induced pHcyto shifts in MN termini, as the neuronal cytosol is thought be strongly buffered (Roos & Boron, 1981). We were initially concerned that such large acid transients were an experimental artifact caused by an unintentional decrease in βb (and thus βT). Stock HL6 used in most experiments contained 5 mm BES and 10 mm HCO3, but was not equilibrated with CO2 during mixing. Subsequent loss of HCO3 would lead to a decrease in βb and could produce artificially large acid transients. To address this issue we quantified βi of the terminal cytosol and investigated the role of βb in buffering rapid and slow activity-induced acid loads.

We quantified βi in the terminal cytosol by exposing SE-pHluorin/DsRed preparations to multiple NH4Cl pulses of decreasing concentration (Fig. 4A). The resulting pHcyto changes were used to calculate βi at the point of each decrease in NH4Cl concentration, and we plotted βi as a function of pHcyto (Fig. 4B; eqn (3)). We used linear regression on the individual values to determine the line of best fit, and this method yielded a value of βi= 11.9 mm at a pHcyto value of 7.3.

Under physiological conditions bicarbonate-derived buffering is predicted to be the primary source of buffering eqns (5) and (6). However, this will only be the case if the CO2/HCO3 buffering reactions eqn (4) reach equilibrium following an acutely imposed change in pHcyto. The forward and reverse rate constants of the hydration of CO2 to H2CO2 are of the order of 20 s−1 and 1 s−1, respectively (Maren, 1967), and thus the reaction equilibrium can take minutes to achieve. Even when the buffering reaction is catalysed by CA, bicarbonate-derived buffering can take minutes to effectively buffer an imposed intracellular pHcyto change (Leem & Vaughan-Jones, 1998) We investigated the role of bicarbonate-derived buffering in shaping activity-induced acid transients by characterizing pHcyto shifts during long, low-frequency (5 Hz, 60 s) and short, high-frequency (40 Hz, 4 s) stimulation trains in bicarbonate-buffer modified HL6 in the presence and absence of ETZ, an intracellular CA inhibitor (Fig. 4C and D).

Long trains of low-frequency nerve stimulation produced a slow acid transient in the terminal cytosol (Fig. 4C). The change in pHcyto during stimulation increased from −0.16 ± 0.01 pH units to −0.20 ± 0.02 pH units after incubation in 200 μm ETZ for 30 min (n= 6, P≤ 0.05). The decay constant of the recovery phase of the transient increased upon addition of ETZ from 29.85 ± 4.06 s to 54.35 ± 7.35 s (τ determined by single exponential fit; n= 6, P≤ 0.05). These results support the conclusion that the CO2/HCO3 buffering system is weakly catalysed by CA in the cytosol of Drosophila MN termini, and is thus unlikely to effectively buffer brief and rapid acid loads. We tested this hypothesis directly by measuring acid transients induced by short trains of high-frequency nerve stimulation (40 Hz, 4 s) in bicarbonate-buffered HL6 in the presence and absence of ETZ (Fig. 4D). The magnitude of the acid transients was unchanged upon addition of ETZ (n= 6, P≥ 0.20), but the recovery phase following the acid loading was qualitatively altered. In the presence of ETZ the recovery was well described by neither a single nor double exponential, and we determined that the recovery phase diverged from control ∼35 s after stimulation ceased. These results suggest that bicarbonate-derived buffering does not approach equilibrium during acute acid loads induced by stimulation trains at 40 Hz and, as a result, βb was not included in calculations of βT in the following section when calculating the molar flux of acid equivalents during rapid acid transients.

A better understanding of the buffering conditions that shape rapid acid transients in Drosophila MN termini allowed us to investigate the extent to which intracellular HCO3 influenced the magnitude and recovery of these transients. We evoked rapid acid transients by nerve stimulation (40 Hz, 4 s) in stock HL6, modified HL6 buffered with 19 mm BES bubbled with 100% O2, and modified HL6 buffered with 19 mm HCO3 bubbled with 5% CO2/95% O2, and compared the molar fluxes during the rapid acid transient and decay constants of the recovery period (Fig. 5A and B). Switching to bicarbonate-buffered HL6 produced an acidification from the assumed resting pHcyto of 7.30 in stock HL6 to 7.15 ± 0.02 (n= 6). Based on the correlation between pHcyto and βi (Fig. 4B), this drop in resting pH corresponded to an increase in βi from 11.95 mm to 15.94 mm. We assumed βTi as the transients were evoked using short (4 s) stimulus trains. The molar flux of acid equivalents during stimulation did not change in the three buffering conditions (0.34 ± 0.11, 0.37 ± 0.12, 0.43 ± 0.12 mm s−1 in stock, BES-buffered and bicarbonate-buffered HL6, n= 6, P≤ 0.05), and the recovery decay constant was significantly decreased in bicarbonate-buffered HL6 as compared with the BES-buffered HL6 and the stock HL6 (17.36 ± 0.82, 18.86 ± 1.16, 14.64 ± 0.69 s in stock, BES-buffered and bicarbonate-buffered HL6, n= 6, P≤ 0.05). These findings suggest that stock HL6 is functionally a bicarbonate-free saline, and that bicarbonate-derived buffering does not approach equilibrium quickly enough to influence the rapid acid loading seen during short trains of high-frequency nerve stimulation but does serve to enhance realkalinization in the recovery phase.

The PMCA is partly responsible for activity-induced presynaptic acidification

Activity-induced presynaptic acidification may be due to increased activity of the PMCA through exchange of Ca2+ for H+ (Trapp et al. 1996). PMCAs are expressed on the PM of Drosophila MNs and extrude the bulk of cytosolic Ca2+ loads accumulated during neuronal activity (Lnenicka et al. 2006). If the PMCA supplied the proton influx responsible for stimulation-induced acid transients, the magnitude of acid transients would correlate with cytosolic Ca2+ load. To investigate this possibility, we simultaneously monitored cytosolic Ca2+ levels ([Ca2+]cyto) and pHcyto by forward-filling the MN termini of SE-pHluorin expressing larvae with the synthetic Ca2+-indicator rhod-dextran. Preparations were stimulated multiple times (40 Hz, 4 s, 10 min between pulse trains) in stock HL6 containing 0.5, 1.0 and 2.0 mm extracellular calcium ([Ca2+]e) (three preparations, three stimulations per preparation in pseudo-randomized order; Fig. 6A). [Ca2+]cyto peaks were normalized to the maximum rhod-dextran fluorescence value in each preparation (during stimulation in 2 mm[Ca2+]e), and net molar acid flux (in mm s−1) was calculated from the maximum decrease in pHcyto during each stimulation and the previously calculated value of βT (see Methods and Fig. 4A). Net molar acid flux was plotted as a function of normalized rhod-dextran fluorescence, which revealed a positive linear relationship between peak [Ca2+]cyto levels and net molar acid flux (R2= 0.99; Fig. 6C).

image

Figure 6. Activity-induced acid transients correlate with cytosolic Ca2+ increase and arise due to proton influx through the plasma membrane Ca2+-ATPase (PMCA)  A, representative traces of stimulation-induced pHcyto and normalized [Ca2+]cyto changes in MN13-Ib termini in a single preparation. Larval motor neurons (MNs) expressing SE-pHluorin were forward-filled with rhod-dextran and stimulated (black bars) in stock HL6 (pHe 7.3) containing 0.5, 1.0 and 2.0 mm[Ca2+]e. The rapid acidification phase of each trace is displayed in black. B, representative traces of stimulation-induced pHcyto and normalized [Ca2+]cyto changes in a single preparation in stock HL6 containing 0.6 mm[Ca2+]e buffered to pHe 7.3 (first stimulation) and then 8.8 (second stimulation). Rhod-dextran traces are normalized to the maximum fluorescence observed in a control stimulation in stock HL6 (2 mm Ca2+, pHe buffered to 7.3, not shown). Resting pHcyto was assumed to be 7.3 and 7.65 in stock HL6 buffered to 7.3 and 8.8, respectively. C, net molar acid flux (mm s−1) plotted as a function of normalized rhod-dextran fluorescence during stimulation. Molar flux was calculated from the rapid acidification phase of the traces in A and B. Data are from 3 preparations, with each preparation stimulated once in each value of [Ca2+]e. Half-filled circles represent the same 3 preparations presented in B. The line is a linear fit (R2= 0.99) to data from preparations in HL6 with [Ca2+]e= 0.5, 1.0 and 2.0 mm. D, top panel, net molar acid fluxes from SE-pHluorin larvae forward-filled with rhod-dextran in response to stimulation (40 Hz, 4 s, [Ca2+]e= 0.6 mm) in stock HL6 buffered to pHe 7.3 (‘control’) and then 8.8 (same data from C, 10 min between stimulations). *P < 0.05. D, bottom panel, net molar acid fluxes from SE-pHluorin/DsRed larvae in response to stimulation (40 Hz, 4 s, [Ca2+]e= 0.6 mm) in stock HL6 before (‘control’) and after 30 min incubation in 100 μm CEDA-SE (n= 5 preparations). *P < 0.05. Error bars in D are SEM. A–D collected at 4.3 Hz.

Download figure to PowerPoint

If the PMCA contributes to activity-induced acid transients, then inhibition of the PMCA should both attenuate the magnitude of acid transients and increase [Ca2+]cyto peaks during stimulation, thus breaking the previously established positive relationship between peak [Ca2+]cyto levels and net molar acid flux. We tested this hypothesis by simultaneously measuring activity-induced changes in pHcyto and [Ca2+]cyto before and after preparations were incubated for 10 min in stock HL6 buffered to pHe 8.8 to inhibit the PMCA (Xu et al. 2000; Lnenicka et al. 2006; Fig. 6B). Preparations were stimulated (40 Hz, 4 s) in 2 mm[Ca2+]e stock HL6 (data not shown), 0.6 mm[Ca2+]e stock HL6 buffered to pHe 7.3, and 0.6 mm[Ca2+]e stock HL6 buffered to pHe 8.8. The stimulation in 2 mm[Ca2+]e was used to normalize the [Ca2+]cyto peaks measured in 0.6 mm[Ca2+]e. To prevent Ca2+ overload, 0.6 mm[Ca2+]e was used. Raising pHe to 8.8 caused an increase in resting SE-pHluorin fluorescence consistent with an elevation in resting pHcyto. This increase was quantified in separate SE-pHluorin/DsRed preparations incubated in stock HL6 buffered to 8.8, and the new resting pHcyto was determined to be 7.63 ± 0.05 (n= 6, data not shown). Then 7.63 was used as the assumed resting pHcyto in all preparations in HL6 buffered to 8.8. Under these conditions βi was assumed to be 3.2 mm as determined by our previous quantification of βi (Fig. 4B). Inhibiting the PMCA increased peak [Ca2+]cyto levels by 139.1 ± 0.1% relative to the observed value at pHe 7.3 and, as predicted, the net molar acid flux decreased from 0.24 ± 0.07 mm s−1 to 0.05 ± 0.04 mm s−1 (n= 3, P < 0.05; shaded circles Fig. 6C, pooled data in Fig. 6D).

Inhibiting the PMCA through elevated pHe is a non-specific manipulation, so to further test the involvement of the PMCA we applied CEDA-SE. Though eosin derivatives impair organellar as well as PM proton transporters (Kosterin et al. 1996; Watson et al. 2003; Makani & Chelser, 2010), CEDA-SE has the benefit of not changing resting pHcyto upon application (data not shown), and thus alterations in net molar flux of acid equivalents before and after CEDA-SE application can be compared without the need to account for changes in βT. We compared the stimulation-induced acid transients of SE-pHluorin/DsRed in stock HL6 (0.6 mm Ca2+) before and after a 30 min incubation in 100 μm CEDA-SE. CEDA-SE significantly reduced the net molar acid flux from 0.26 ± 0.07 mm s−1 to 0.12 ± 0.04 mm s−1 (n= 5, P < 0.05; Fig. 6D). These data corroborate our previous findings and further suggest Ca2+/H+ exchange mediated by the PMCA produces the rapid activity-induced acid transients in Drosophila MN termini.

Stimulus trains produce a large cytosolic acidification in motor nerve termini in situ

Larval MNs fire in repeated bursts to maintain peristaltic locomotion (Budnik et al. 1990). To investigate the probable minimal extent of cytosolic acidification during repetitive firing, MN13-Ib axons were stimulated in stock HL6 with a series of impulse trains (40 Hz, 2 s, one train every 10 s). Rhod-dextran was forward-filled into SE-pHluorin expressing MN termini to allow simultaneous imaging of [Ca2+]cyto and pHcyto. SE-pHluorin fluorescence revealed a progressive acidification of the cytosol with each impulse train (Fig. 7A). After 8 trains, pHcyto had dropped from the assumed resting value of 7.30 to 6.98 ± 0.03 (n= 6).

image

Figure 7. Endogenous motor activity causes substantial presynaptic acidification in situ and in vivo A, traces representing simultaneous changes in [Ca2+]cyto and pHctyo of a motor neuron (MN) terminal as reported by rhod-dextran and SE-pHluorin, in response to nerve stimulus trains (black bars, 40 Hz, 2 s each). B, traces representing spontaneous simultaneous changes in [Ca2+]cyto and pHctyo as reported by rhod-2 AM and SE-pHluorin in a MN terminal still attached to the CNS. No external stimuli were applied. In A and B, rhod-dextran or rhod-2AM fluorescence change was normalized to a value of 1 (maximum response). The rhod-2 AM trace was corrected for cytosolic dye loss and bleaching using a single exponential curve fit. No data averaging or smoothing in A or B. C, averaged plots of pHcyto shift in MN13-Ib termini reported by SE-pHluorin/DsRed in response to nerve stimulation (black bar) in preparations in which the axons had been cut (black trace, n= 6, data acquisition rate of 1.66 Hz) and left intact (grey trace, n= 6). The value of τ was calculated by fitting a single exponential to the recovery phase of each trace in each condition (R2cut= 0.93, R2intact= 0.97). D, a trace demonstrating spontaneous changes in pHctyo in a MN terminal in a restrained intact larva (in vivo). Fluorescence intensity data were acquired from presynaptic SE-pHluorin and DsRed through the cuticle. The relative change in the ratio of these fluorophores (ΔR/R) is plotted over a period of 120 s. In AD, data were acquired at a frame rate of 4.3 Hz unless noted otherwise. Resting pHcyto of SE-pHluorin/DsRed was assumed to be 7.30 in C and 7.15 in D, as this was the value observed in the presence of a CO2/HCO3-based buffering system in situ.

Download figure to PowerPoint

GEpHIs reveal activity-dependent presynaptic pH shifts in situ and in vivo during spontaneous activity

To determine whether the pHcyto shifts observed in situ during simulated motor patterns also occur during fictive locomotion in situ with the central pattern generator driving MN activity, we examined changes in GEpHI fluorescence in MN13-Ib termini without cutting or stimulating the segmental nerves. Larval preparations expressing SE-pHluorin were incubated with bath-applied rhod-2 AM to allow simultaneous monitoring of [Ca2+]cyto and pHcyto (Fig. 7B). Spontaneous bursts of nerve activity (as indicated by increases in rhod-2 AM fluorescence) were accompanied by an acidification of the cytosol from an assumed resting pHcyto of 7.30 to 7.03 ± 0.10 (n= 6 bursts from three preparations).

pHcyto shifts that occurred during spontaneous bursts of nerve activity seemed to have a faster decay time course than those observed in stimulated preparations. A systematic comparison of stimulation-induced pHcyto shifts between preparations with cut segmental nerves and those with intact segmental nerves revealed that cutting the nerves altered both the amplitude and decay time course (τ) of the pHcyto shifts (Fig. 7C). MN termini with intact axons showed a decreased pHcyto acidification (ΔpHcyto=−0.12 ± 0.01 pH units (n= 6)) and a reduced τ (13.85 ± 1.05 s (n= 6)) after stimulation (40 Hz, 4 s) when compared with larvae with cut axons (ΔpHcyto=−0.16 ± 0.01 pH units, τ= 20.54 ± 2.26 s, P < 0.05 in comparisons of ΔpHcyto and τ).

Finally, to determine whether the pHcyto shifts observed in situ were an artifact of this MN terminal preparation being ex vivo, we monitored pHcyto in intact 3rd instar larvae expressing SE-pHluorin/DsRed. Simultaneous imaging of SE-pHluorin and DsRed fluorescence in MN13-Ib termini was conducted through the cuticle (Fig. 7D). In these experiments resting pHcyto was assumed to be 7.16 as this was the pHcyto observed at rest in bicarbonate-buffered HL6 (7.15 ± 0.02, n= 6; see Fig 5A), and this is likely to represent a more physiological buffering system that mimics the endogenous haemolymph. The relative change in the SE-pHluorin/DsRed fluorescence ratio (ΔR/R) during body wall contraction was −20.6 ± 0.8% (n= 6 events from three preparations), which corresponds to a decrease in pHcyto from an assumed resting value of 7.15 to 6.97 ± 0.07. These changes confirm that the pHcyto shifts represent a real physiological phenomenon.

Discussion

  1. Top of page
  2. Key points
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. References
  8. Appendix

GEpHIs used in this study revealed cytosolic acidification of Drosophila larval MN termini during evoked and spontaneous neuronal activity. The amplitude of the pHcyto shifts measured using GEpHIs was substantial and these estimates were corroborated using the chemical pH-indicator BCECF-dextran. Depolarization-dependent acid transients of this amplitude have been documented previously in the soma of cultured frog MNs (Endres et al. 1986) and in tertiary dendrites of voltage-clamped rat cerebellar Purkinje cells (Willoughby & Schwiening, 2002), but it was not anticipated that such significant changes occur in presynaptic termini during spontaneous activity in vivo.

Localized activity-induced cytosolic acid transients have been described in many neuronal preparations, but they are often at least partially masked in the presence of extracellular HCO3. Both CA-catalysed bicarbonate-derived buffering and HCO3 influx through anion exchangers can attenuate acid transients (Zhang et al. 2010; Svichar et al. 2011). Our data demonstrate that acid loading in Drosophila larval MN termini during short trains of high-frequency nerve activity such as those that coordinate muscle contraction at the neuromuscular junction are unaltered by the absence of extracellular HCO3 and the inhibition of CA (Figs 4D and 5). These results suggest that although the Drosophila genome encodes multiple PM anion exchangers (Romero et al. 2000; Dubreuil et al. 2010), these proteins are minimally active in larval MN termini. However, CA activity does enhance the recovery from acid loads as pHcyto returns to the pre-stimulation resting level (Fig. 4C and D). This finding suggests that as the CO2/HCO3 buffering reaction approaches equilibrium it begins to serve as an effective buffer to the acid load and that this process is catalysed by CA in Drosophila MN termini.

Many Drosophila recording salines commonly used in electrophysiological and optical assays of the larval neuromuscular junction are buffered with 5 mm Hepes (standard solution of Jan & Jan, 1976), or 5 mm BES and 10 mm HCO3 (HL3 of Stewart et al. 1994 and HL6 of Macleod et al. 2002). Solutions are not commonly equilibrated with CO2 during mixing and thus may lose all HCO3 by the time of use, leaving them buffered only by zwitterionic buffers. Such a condition could result in exaggerated acid transients due to: (i) a decrease in βb owning to a lack of available HCO3 to drive HCO3/CO2 buffering reactions; and (ii) a lack of HCO3 to enter the cell through anion exchangers and drive alkalinization. Our data suggest that considerations (i) and (ii) do not affect rapid acid transients in Drosophila larval MN termini, but may influence the recovery time course following an acute acid load. CA and anion exchanger activity is also likely to vary on a subcellular scale, and thus we recommend the use of appropriately equilibrated HCO3/CO2-buffered recording saline.

The magnitude of rapid acid transients in larval MN termini correlates with the magnitude of [Ca2+]cyto transients during stimulation (Fig. 6A and C). Indeed, intracellular acidification has been previously shown to correlate with increased [Ca2+]cyto (Schwiening & Willoughby, 2002), and activity-induced acidification is thought to arise from activity of the PMCA in many neuronal preparations (Trapp et al. 1996). Inhibition of the PMCA by both increased pHe and CEDA-SE decreased the net molar acid flux in larval MN termini (Fig. 6B–D). We must note that in calculating net molar flux we assumed βT could be approximated as equivalent to βi, as we have demonstrated that βb contributed minimally to the buffering of rapid acid transients (Figs 4D and 5). Furthermore, in calculating βi we employed a stepwise NH4Cl pulse technique and used an extrapolation method to correct for active pH regulation at the final NH4Cl withdrawal step rather than attempting to block pHcyto regulation (Fig. 4A). This possibly led to an overestimation of βi at the final withdrawal step but, as these data closely followed the trend established by the previous decreases in NH4Cl, we believe this had little effect on our estimation of βi.

Increased pHe and CEDA-SE are non-specific inhibitors of the PMCA, and we cannot rule out the possibility that our manipulations inhibited other organellar acid transporters. However, our data indicate that raising pHe to 8.8 breaks the correlation between net molar acid flux and [Ca2+]cyto levels by increasing [Ca2+]cyto transients yet decreasing molar flux (Fig. 6C). These findings taken with the reduction of net molar acid flux upon addition of CEDA-SE (Fig. 6D) strongly suggest that rapid acid transients in larval MN termini arise from Ca2+/H+ exchange meditated by the PMCA.

We demonstrate for the first time that rapid acid transients occur within presynaptic termini during endogenous neuronal activity in vivo. SE-pHluorin revealed that the amplitude of presynaptic pHcyto shifts in vivo is similar to that observed in situ, though pHcyto shifts during endogenous nerve activity recover more quickly than those observed under exogenous stimulation. Exogenous stimulation produced different responses in cut and uncut nerves. Leaving the segmental nerves intact during exogenous stimulation reduced the decay time course and amplitude of the pHcyto shifts (Fig. 7C). These observations suggest that cutting the segmental nerves can disrupt the mechanisms of pHcyto homeostasis in nerve terminals and highlight the need to measure pHcyto shifts in unperturbed (intact) biological systems. GEpHIs can serve as ideal tools for such measurements as their use avoids the need for potentially damaging techniques frequently required to load chemical pH-indicators such as BCECF.

While all GEpHIs reported pHcyto shifts, SE-pHluorin proved to be the best GEpHI for measuring activity- induced pHcyto shifts in larval MN termini. SE-pHluorin 2EM provides for ratiometric estimation of pHcyto, but 406 nm excitation produces considerable autofluorescence from the underlying muscle when used in situ, and from the larval cuticle when used in vivo. SE-pHluorin, in single-emission wavelength mode, provided the best SNR and dynamic range of the GEpHIs, but co-expression with DsRed was necessary for pseudo-ratiometric analysis and movement correction, especially when used in vivo. The SE-pHluorin/DsRed fluorescence ratio obtained in individual preparations could not be directly compared with other preparations due to the inconsistent cytosolic distribution of DsRed between preparations. Therefore, estimates of pHcyto relied on calibration of the relative fluorescence change in SE-pHluorin in single-emission wavelength mode. An average resting presynaptic pHcyto of 7.30 was estimated using SE-pHluorin 2EM, and this value was assumed for calibration when calculating pH values from the SE-pHluorin/DsRed fluorescence ratio. In vivo imaging required the use of SE-pHluorin in single-emission wavelength mode, and hence the data only reveal a relative change in pHcyto from an assumed resting level of 7.16, which was determined in situ, in separate experiments. The genetic malleability of Drosophila and relative ease of in vivo imaging make the larval neuromuscular junction an ideal system for developing and testing novel GEpHIs in the future.

The extent to which pHcyto acidifies during larval locomotion likely exceeds 0.30 pH units. The series of stimulus trains applied in situ (Fig. 7A) was intended to simulate motor patterns, but the actual interval between trains (bursts) in vivo is much shorter (Klose et al. 2005). Our in situ data indicate that closely spaced bursts will lead to substantial presynaptic acidification. A caveat to this prediction is that the CO2/HCO3 buffering system present in vivo will likely approach equilibrium after sustained activity and begin to buffer further acidification. Regardless, the magnitude of rapid pH shifts suggests caution must be taken when interpreting data acquired through the use of GFP-based fluorescent reporters of other entities in such termini. Many GFP-based reporters, including commonly used genetically encoded Ca2+-indicators, are sensitive to pH fluctuations near neutral pH, in which case, activity-induced pHcyto shifts may introduce pH artifacts in any data collected using GFP-based reporters (Mank & Griesbeck, 2008).

The finding that pHcyto in MN termini acidifies substantially in response to endogenous motor activity raises many questions regarding the extent to which pHcyto shifts influence presynaptic mechanisms, and similarly the extent to which allowances for pHcyto shifts may be incorporated into presynaptic mechanisms. The correlation between the acid transients and [Ca2+]cyto levels during neuronal activity (Fig. 6C) raises the possibility that shifts in pHcyto play a role in some forms of short-term synaptic plasticity. Unfortunately direct tests of such hypotheses in situ are quite intractable because all manipulations of pHcyto, such as local application of weak membrane-permeant acids, influence both presynaptic and postsynaptic cells. The role of pHcyto shifts in presynaptic neurophysiology will remain elusive until a method is devised to control pHcyto independently of [Ca2+]cyto at a cell-specific level.

References

  1. Top of page
  2. Key points
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. References
  8. Appendix

Appendix

  1. Top of page
  2. Key points
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. References
  8. Appendix

Author contributions

All the authors contributed to the design of the experiments and interpretation of the data. All the authors collected the data. A.J.R. analysed the data. A.J.R. drafted the manuscript, and A.K.C. and G.T.M. critically revised the draft for intellectual content. All the authors approved the final version of the manuscript.

Acknowledgements

This study was supported by grants from the National Institute of Neurological Disorders and Stroke (National Institutes of Health (NIH) R01 NS061914) and National Science Foundation ((NSF) EAGER IOS-1147467) to G.T.M. We thank Bloomington Stock Center for providing fly strains, Gero Miesenbock for providing plasmids, Rosario Martinez for assistance preparing cDNA and maintaining fly stocks, and Christopher Weyand for preliminary investigation of the GEpHIs. The authors extend their appreciation to Dr. Mitchell Chesler for his critical feedback on this project.