Differences of Ca2+ handling properties in identified central olfactory neurons of the antennal lobe

Cell Calcium 46 (2009) 87–98

Contents lists available at ScienceDirect

Cell Calcium journal homepage: www.elsevier.com/locate/ceca

Differences of Ca2+ handling properties in identified central olfactory neurons of the antennal lobe Andreas Pippow a , Andreas Husch a , Christophe Pouzat b , Peter Kloppenburg a,∗ a Institute of Zoology and Physiology, Center for Molecular Medicine Cologne (CMMC) and Cologne Excellence Cluster in Aging Associated Diseases (CECAD), University of Cologne, Weyertal 119, 50931 Cologne, Germany b Laboratoire de Physiologie Cerebrale, CNRS UMR 8118, UFR biomedicale de l’Universite Paris V, 45 rue des Saints Peres, 75006 Paris, France

a r t i c l e

i n f o

Article history: Received 17 March 2009 Received in revised form 15 May 2009 Accepted 19 May 2009 Available online 9 July 2009 Keywords: Antennal lobe Calcium binding ratio Calcium buffering Chemosensory Fura-2 Local interneurons Olfaction Projection neurons

a b s t r a c t Information processing in neurons depends on highly localized Ca2+ signals. The spatial and temporal dynamics of these signals are determined by a variety of cellular parameters including the calcium influx, calcium buffering and calcium extrusion. Our long-term goal is to better understand how intracellular Ca2+ dynamics are controlled and contribute to information processing in defined interneurons of the insect olfactory system. The latter has served as an excellent model to study general mechanisms of olfaction. Using patch-clamp recordings and fast optical imaging in combination with the ‘added buffer approach’, we analyzed the Ca2+ handling properties of different identified neuron types in Periplaneta americana’s olfactory system. Our focus was on two types of local interneurons (LNs) with significant differences in intrinsic electrophysiological properties: (1) spiking LNs that generate ‘normal’ Na+ driven action potentials and (2) non-spiking LNs that do not express voltage-activated Na+ channels. We found that the distinct electrophysiological properties from different types of central olfactory interneurons are strongly correlated with their cell specific calcium handling properties: non-spiking LNs, in which Ca2+ is the only cation that enters the cell to contribute to membrane depolarization, had the highest endogenous Ca2+ binding ratio and Ca2+ extrusion rate. © 2009 Elsevier Ltd. All rights reserved.

1. Introduction The insect olfactory system has served as an extremely successful model for olfactory information processing. Due to the many striking similarities in structural organization and physiological function between vertebrate and invertebrate olfactory systems, data from the insect olfactory system have contributed significantly to the construction of reasonable general models for olfactory processing on the network level [1–6]. This study further analyzes the cellular parameters that determine intracellular calcium dynamics in central olfactory interneurons, to better understand the mechanisms that mediate olfactory information processing on the cellular and sub-cellular level. In neurons, calcium regulates a multitude of cellular functions, and many aspects of information processing in single neurons are dependent on highly localized calcium domains. Besides being a charge carrier, which contributes directly to the membrane potential, calcium serves as a second messenger that controls a variety of cellular processes including synaptic release, membrane excitability, enzyme activation and activity-dependent gene acti-

∗ Corresponding author. Tel.: +49 221 470 5950; fax: +49 221 470 4889. E-mail address: [email protected] (P. Kloppenburg). 0143-4160/$ – see front matter © 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.ceca.2009.05.004

vation. Selective activation of these functions is achieved through the spatial and temporal distribution of the calcium signals [7]. The spatio-temporal Ca2+ dynamics are determined by several cellular parameters, including: the calcium influx, the geometry of the cell, the location of the calcium source, the calcium buffering, the calcium extrusion and the locally changing diffusion coefficients [8–13]. The voltage-activated Ca2+ currents, an important Ca2+ entry pathway, have been studied in many insect neuron types [14–20] including central olfactory neurons [21,22]. Other factors that influence calcium dynamics, however, are not as well studied and understood. Our long-term goal is to better understand how intracellular calcium dynamics contribute to information processing in olfactory interneurons of the insect antennal lobe (AL). Specifically, we are interested in how different electrophysiological characteristics of neurons are correlated with differences in their Ca2+ handling properties. This is of special interest in the AL, since we have recently identified two types of local interneurons (LNs) with fundamental differences in intrinsic firing properties: type I LNs generate ‘normal’ Na+ driven action potentials, whereas type II LNs do not express voltage-activated sodium channels. In type II LNs, Ca2+ is the only cation that enters the cell to contribute to membrane depolarization [22]. We consider the characterization of cellular parameters that determine cytosolic Ca2+ dynamics, as an important step towards a

88

A. Pippow et al. / Cell Calcium 46 (2009) 87–98

detailed understanding of the cellular basis of olfactory information processing at the single cell level. Such data are moreover important to help interpret time constants of Ca2+ signals from experiments in which Ca2+ measurements are performed to monitor global neuronal activity [23–32]. Here we used patch-clamp recordings and fast optical imaging in combination with the ‘added buffer approach’ [8] to characterize the Ca2+ handling properties of identified interneurons in the Periplaneta americana olfactory system. The Ca2+ handling properties were significantly different between different identified neuron types, with a clear correlation between the Ca2+ handling properties and the intrinsic electrophysiological characteristics. 2. Methods 2.1. Animals and materials P. americana were reared in crowded colonies at 27 ◦ C under a 13:11 h light/dark photoperiod regimen and reared on a diet of dry dog food, oatmeal and water. Neurons for cell culture were obtained from adult animals of both sexes, while the in situ recordings were performed with adult males. Before dissection, the animals were anesthetized by CO2 or cooling (4 ◦ C) for several minutes. To obtain neurons for cell culture, the animals were then adhered in a plastic tube with adhesive tape and the heads were immobilized using dental modeling wax (S-U Modellierwachs, Schuler-Dental, Ulm, Germany) with a low solidification point (57 ◦ C). For experiments with the intact brain, the animals were placed in a custom built holder, and the head was immobilized with dental wax. The antennae were placed in small tubes on a plastic ring that was later used to transfer the preparation to the recording chamber. All chemicals, unless stated otherwise, were obtained from Aplichem (Darmstadt, Germany) or Sigma–Aldrich (Taufkirchen, Germany) in ‘pro analysis’ purity grade. 2.2. Cell culture To examine the electrophysiological properties of isolated unidentified antennal lobe neurons, cells were dissociated and cultured using protocols reported previously [21]. The head capsule of anesthetized animals was opened and the antennal lobes were dissected with fine forceps. Typically, ALs from eight animals were pooled in sterile ‘culture’ saline (kept on ice) containing (in mM): 185 NaCl, 4 KCl, 6 CaCl2 , 2 MgCl2 , 35 d-glucose, 10 HEPES and 5% fetal bovine serum (S-10, c.c.pro, Neustadt, Germany), adjusted to pH 7.2 (with NaOH), which resulted in an osmolarity of ∼420 mOsm. For dissociation the ALs were transferred for 2 min at 37 ◦ C into 500 ␮l Hanks Ca2+ and Mg2+ free buffered salt solution (14170, GIBCO, Invitrogen, Karlsruhe, Germany) containing (in mM): 10 HEPES, 130 sucrose, 8 units ml−1 collagenase (LS004194, Worthington, Lakewood, NJ, USA) and 0.7 units ml−1 dispase (LS02100, Worthington), adjusted to pH 7.2 (with NaOH) and to 450 mOsm (with sucrose). Dissociation of neurons was aided by careful trituration with a fire-polished Pasteur pipette for 3–5 min. Enzyme treatment was terminated by cooling and centrifuging the cells twice through 6 ml of culture medium (4 ◦ C, 480 g, 5 min). The culture medium consisted of five parts Schneider’s Drosophila medium (21720, GIBCO) and four parts Minimum Essential Medium (21575, GIBCO) to which was added (in mM): 10 HEPES, 15 glucose, 10 fructose, 60 sucrose, 5% fetal bovine serum adjusted to pH 7.5 (with NaOH) and 430 mOsm (with sucrose). After centrifugation, the cells were resuspended in a small volume of culture medium (100 ␮l per dish; eight ALs were plated in five to six dishes), and allowed to settle for 2 h to adhere to the surface of the culture dishes coated with concanavalin A (C-2010, Sigma, 0.7 mg ml−1 dissolved in H2 O). The cultures were placed in an incubator at 26 ◦ C, and

then used for electrophysiological experiments on the same day. For recordings, the cells were visualized with an inverted microscope (IX71, Olympus, Hamburg, Germany) using a 40× water objective (UAPO 40×, 1.15 NA, 0.25 mm WD, Olympus) and phase contrast optics. 2.3. Intact brain preparation The intact brain preparation was based on an approach described previously [18,22,33], in which the entire olfactory network was left intact. The animals were anesthetized by CO2 , placed in a custom build holder and the head with antennae was immobilized with tape (Tesa ExtraPower Gewebeband, Tesa, Hamburg, Germany). The head capsule was opened by cutting a window between the two compound eyes at the bases of the antennae. The brain with antennal nerves and antennae attached was dissected from the head capsule in ‘normal saline’ (see below) and pinned in a Sylgard-coated (Dow Corning Corp., Midland, MI, USA) recording chamber. To gain access to the recording site and facilitate the penetration of pharmacological agents into the tissue, we desheathed parts of the AL using fine forceps. Some preparations were also enzyme-treated with a mixture of papain (0.3 mg ml−1 , P4762, Sigma) and l-cysteine (1 mg ml−1 , 30090, Fluka/Sigma) dissolved in ‘normal’ saline (∼3 min, RT). The AL neurons were visualized with a fixed stage upright microscope (BX51WI, Olympus, Hamburg, Germany) using a 20× water-immersion objective (XLUMPLFL 20×, 0.95 NA, 2 mm WD, Olympus) with a 4× magnification changer (U-TVAC, Olympus) and IR-DIC optics [34]. 2.4. Whole-cell recordings Whole-cell recordings were performed at 24 ◦ C following the methods described by Hamill et al. [35]. Electrodes with tip resistance between 3 and 5 M were fashioned from borosilicate glass (0.86 mm OD, 1.5 mm ID, GB150-8P, Science Products, Hofheim, Germany) with a temperature controlled pipette puller (PIP5, HEKA-Elektronik, Lambrecht, Germany), and filled with a solution containing (in mM): 210 CsCl, 10 NaCl, 2 MgCl2 , 10 HEPES and 0.2 fura-2 (F1200, Molecular Probes, OR, USA) adjusted to pH 7.2 (with CsOH), resulting in an osmolarity of ∼420 mOsm. During the experiments, if not stated otherwise, the cells were superfused constantly with ‘normal’ saline solution containing (in mM): 185 NaCl, 4 KCl, 6 CaCl2 , 2 MgCl2 , 10 HEPES, 5 Glucose. The solution was adjusted to pH 7.2 (with NaOH) and to 430 mOsm (with glucose). To isolate the Ca2+ currents under voltage-clamp, we used a combination of pharmacological blockers and ion substitution that has been shown to be effective in central olfactory neurons of the cockroach [21,22] and in other insect preparations [17,18]. Transient voltage-gated sodium currents were blocked by tetrodotoxin (10−7 to 10−4 M, TTX, T-550, Alomone, Jerusalem, Israel). In AL neurons 10−7 M TTX blocked voltage-gated sodium currents effectively. When only small areas of the brains were desheathed TTX diffusion was hindered and it could take more than 30 min to block sodium currents. In such cases we sometimes used higher TTX concentrations to accelerate the block. 4-Aminopyridine (4 × 10−3 M, 4-AP, A78403, Sigma) was used to block transient K+ currents (IA ) and tetraethylammonium (20 × 10−3 M, TEA, T2265, Sigma) blocked sustained K+ currents (IK(V) ) as well as Ca2+ activated K+ currents (IK(Ca) ). To further reduce outward currents the pipette solution contained cesium instead of potassium. Whole-cell voltage-clamp recordings were made with an EPC9 patch-clamp amplifier (HEKA-Elektronik) that was controlled by the program Pulse (version 8.63, HEKA-Elektronik) running under Windows. The electrophysiological signals were sampled at intervals of 100 ␮s (10 kHz). The recordings were low pass filtered at 2 kHz with a 4-pole Bessel-Filter. Compensation of the offset

A. Pippow et al. / Cell Calcium 46 (2009) 87–98

potential and capacitance were performed using the ‘automatic mode’ of the EPC9 amplifier. Whole-cell capacitance was determined by using the capacitance compensation (C-slow) of the EPC9. The calculated liquid junction potential between intracellular and extracellular solution [36] of 4.8 mV (calculated with Patcher’s-Power-Tools plug-in from http://www.mpibpc.gwdg.de/ abteilungen/140/software/index.html for Igor Pro 6 [Wavemetrics, Portland, OR, USA]) was also compensated. To remove uncompensated leakage and capacitive currents, a p/6 protocol was used [37]. Voltage errors due to series resistance (RS ) were minimized using the RS -compensation of the EPC9. RS was compensated between 30% and 70% with a time constant () of 2 ␮s. Stimulus protocols used for each set of experiments are provided in the results. 2.5. Fluorimetric Ca2+ measurements Intracellular Ca2+ concentrations were measured with the Ca2+ indicator fura-2, a ratiometric dye suitable for absolute Ca2+ concentration determination once calibrated [38,39]. The imaging setup for fluorimetric measurements consisted of an Imago/SensiCam CCD camera with a 640 × 480 chip (Till Photonics, Planegg, Germany) and a Polychromator IV (Till Photonics) that was coupled via an optical fibre into the IX71 inverted microscope or the fixed stage BX51 WI upright microscope (both from Olympus, for details see above). The camera and the polychromator were controlled by the software Vision (version 4.0, Till Photonics) run on a Windows PC. The neurons were loaded with fura-2 via the patch pipette (200 ␮M in the pipette) and illuminated during data collection with 340, 360 or 380 nm light from the polychromator that was reflected onto cells with a 410 nm dichroic mirror (DCLP410, Chroma, Gräfeling, Germany). Emitted fluorescence was detected through a 440 nm long-pass filter (LP440). Data were acquired as 40 × 30 frames using 8 × 8 on-chip binning for fast kinetic measurements and 80 × 60 frames using 4 × 4 binning for slow loading measurements. Images were recorded in analog-to-digital units (ADUs) and stored and analyzed as 12 bit grayscale images. For all calculations of kinetics, the mean value of ADU within a region of interest (ROI) from the center of a cell body was used. The ROI was adjusted to each cell (Fig. 1). For ‘background subtraction’ in cell culture an image was obtained in on-cell mode (before break-in) for each excitation wavelength. Background fluorescence was then subtracted from each image of the time series. In the intact brain preparation for the whole time series an adjacent, second ROI was chosen and its mean fluorescence at each excitation wavelength was subtracted from the fura-2 fluorescence (see [40]).

89

2.6. Calibration The free intracellular Ca2+ concentrations were determined as in Grynkiewicz et al. [38]: [Ca2+ ]i = Kd,Fura

F380,min R − Rmin F380,max Rmax − R

(1)

[Ca2+ ]i is the free intracellular Ca2+ concentration for the background subtracted fluorescence ratio R from 340 and 380 nm excitation. Rmin and Rmax are the ratios at a Ca2+ concentration at virtually 0 M (i.e. ideally no fura-2 molecules are bound to Ca2+ ) and at saturating Ca2+ concentrations (i.e. ideally all fura-2 molecules are saturated with Ca2+ ), respectively. Kd,Fura is the dissociation constant of fura-2. F380,min /F380,max is the ratio between the emitted fluorescence of Ca2+ free dye and the emitted fluorescence of Ca2+ saturated dye at 380 nm excitation, reflecting the dynamic range of the indicator. We used two methods for calibration: (a) in vivo (in living cells) and (b) in vitro (in solution). The term Kd,Fura × (F380,min /F380,max ) in Eq. (1) is dependent on the dye concentration and is substituted with the effective dissociation constant Kd,Fura,eff , which is independent of the dye concentration and specific for each experimental setup [41]: Kd,Fura,eff = [Ca2+ ]i

Rmax − R R − Rmin

(2)

For in vivo calibration Kd,Fura,eff was determined by measuring fura2 fluorescence ratios for Rmax , Rmin and R = Rdef . Rdef is the ratio for a defined [Ca2+ ]i , which was set to 140 nM (see below for the preparation of the solutions). Kd,Fura,eff was then calculated from Eq. (2). For the in vitro calibration we used a correction factor (P) for Rmax and Rmin to account for environmental differences between the cytoplasmic milieu and in vitro conditions, as suggested by Poeni [39]: Kd,Fura,eff = [Ca2+ ]i

Rmax P − R R − Rmin P

(3)

P was determined as described by Poeni [39]. First the fluorescence (peak) deflection at 340 nm was divided by that at 380 nm excitation from voltage-induced intracellular calcium transients (Rd,cell ). Second, the ratio (Rd,vitro ) from pairs of calibration solutions was determined by dividing (F340,max − F340,min )/(F380,min − F380,max ). The correction factor P is the fraction of Rd,cell /Rd,vitro .

Fig. 1. Fura-2 loading and Ca2+ decay kinetics: (A and B) fluorescence images from a cultured antennal lobe neuron were acquired at 360 nm excitation (isosbestic point of fura-2) in analog-to-digital units (ADU) every 30 s during loading with fura-2. (A) Series of three images that were acquired at times indicated in (B) to demonstrate increasing indicator concentration. (B) Fura-2 loading curve of the neuron shown in (A). The loading curve starts with establishing the whole cell configuration and is measured in the ROI that is marked in (A). (C) Transients of free Ca2+ concentrations at increasing intracellular fura-2 concentrations. Ca2+ transients were elicited by 50 ms voltage pulses to −5 mV from a holding potential of −60 mV. The fura-2 concentrations are given in (B) and were determined as described in Section 2 and in Fig. 3. The free intracellular Ca2+ concentration was determined as described in Section 2.

90

A. Pippow et al. / Cell Calcium 46 (2009) 87–98

The fluorescence ratio R of an intracellular transient can then be converted to [Ca2+ ]i using: [Ca2+ ]i = Kd,Fura,eff

R − Rmin (for in vivo calibration) Rmax − R

(4)

R − Rmin P (for in vitro calibration) Rmax P − R

(5)

and [Ca2+ ]i = Kd,Fura,eff

To determine the endogenous Ca2+ binding ratio, the dissociation constant (Kd,Fura ) has to be known. To formulate a relationship between Kd,Fura and Kd,Fura,eff that is independent of the dye concentration, an isocoefficient ˛ was introduced by Zhou and Neher [42]: Kd,Fura = Kd,Fura,eff

Rmin + ˛ Rmax + ˛

(6)

˛ is the isocoefficient, for which the sum: Fi (t) = F340 (t) + ˛F380 (t)

(7)

Ca2+

is independent of concentration. F340 and F380 are the background subtracted fluorescence signals measured during a brief Ca2+ transient. Both traces show an excursion, which is dependent on Ca2+ concentration. The correct isocoefficient is the value of ˛ which minimizes this excursion making it independent of Ca2+ concentration (Fig. 2). Once the isocoefficient has been determined, Kd,Fura can be calculated from Eq. (6). 2.6.1. Calibration solutions The free Ca2+ concentrations of the calibration solutions were adjusted by using appropriate proportions of Ca2+ and the Ca2+ chelator EGTA. The ability of EGTA to bind calcium is highly dependent on parameters such as ionic strength, temperature, pH and the concentrations of other metals that compete for binding [43,44]. To account for these parameters we calculated the appropriate concentrations by using the program MaxChelator (WinMaxc32 version 2.50, http://www.stanford.edu/∼cpatton/maxc.html), which is based on earlier work [45–48]. The necessary amount of Ca2+ , Mg2+ and EGTA was computed to set the free Ca2+ and Mg2+ concentration for the experimental conditions (22 ◦ C; pH 7.2; ionic strength [without divalent ions and Ca2+ chelator]: 219 mM for Rmin and Rmax , 214 mM for Rdef ). Calibration solutions were prepared as follows (in mM): Rmax : 205 CsCl, 10 NaCl, 2 MgCl2 , 10 HEPES, 10 CaCl2 and 0.2 fura-2; Rmin : 205 CsCl, 10 NaCl, 2.7 MgCl2 , 10 HEPES, 10 EGTA and 0.2 fura-2; Rdef : 200 CsCl, 10 NaCl, 2.4 MgCl2 , 4.1 CaCl2 , 10 HEPES, 10 EGTA and 0.2 fura-2, yielding a free Ca2+ concentration of 140 nm. All solutions were adjusted to pH 7.2 with CsOH resulting in an osmolarity of ∼430 mOsm (Rmax ) and ∼440 mOsm (Rmin , Rdef ), respectively. 2.7. Single compartment model of calcium buffering: the ‘added buffer approach’ For measurements of intracellular Ca2+ concentrations with Ca2+ chelator-based indicators, the amplitude and time course of the signals are dependent on the concentration of the Ca2+ indicator (in this case fura-2) that acts as an exogenous (added) Ca2+ buffer and competes with the endogenous buffer [8,9,49,50]. With the ‘added buffer approach’ [8], the capacity of the immobile endogenous Ca2+ buffer in a cell is determined by measuring the decay of the Ca2+ signal at different concentrations of ‘added buffer’ and by extrapolating to conditions in which only the endogenous buffer is present. This approach is based on a single compartment model assuming that a big patch-pipette with a known (clamped) concentration of a Ca2+ indicator (fura-2) is connected (whole-cell configuration) to a spherical cell. The model assumes that the cell contains an

Fig. 2. Isocoefficient ˛: the isocoefficient ˛ must be known to calculate the dissociation constant of the Ca2+ indicator fura-2 (Kd,Fura ) (Eq. (6)). The isocoefficient is the factor for which Fi (t) = F340 (t) + ˛F380 (t) is independent of Ca2+ . F340 and F380 are the fluorescence signals for a Ca2+ transient measured at 340 and 380 nm excitation, respectively. ˛ is determined by iteration, optimizing Fi (t) to a horizontal line for the decay of the signals, as shown in (C). t = 0 s is the start of the fluorescence signal.

immobile endogenous Ca2+ buffer (S), of which the total concentration, dissociation constant Kd,S , and its on (k+ ) and off (k− ) rates are unknown. When the whole-cell configuration is established (break in) fura-2 (in its bound and unbound forms) starts to diffuse into the cell, which is considered small enough to be always in a ‘well mixed’ state (i.e., there are no gradients of chemical species within the cell). In this model three cellular parameters determine the cytosolic Ca2+ dynamics: (1) Ca2+ sources, (2) Ca2+ binding ratio and (3) Ca2+ extrusion. 2.7.1. Ca2+ sources In the model, two forms of Ca2+ sources are defined: (1) the Ca2+ influx from the pipette (jin,pipette ) and (2) the Ca2+ entering the cell via voltage-activated channels from a brief Ca2+ pulse, which can be described by a delta function (jin,stim ) [8,50]: jin,pipette = ([BCap ] − [BCa]) jin,stim = nCa ı(t − tstim ) =

DB  Rp

QCa ı(t − tstim ) 2F

(8)

(9)

A. Pippow et al. / Cell Calcium 46 (2009) 87–98

[BCa] is the concentration of the exogenous buffer (fura-2) in its Ca2+ bound form (the subscript p is for quantities within the pipette), DB is the diffusion coefficient of the exogenous buffer,  is the specific resistance of the pipette filling solution, Rp is the pipette resistance, nCa is the Ca2+ influx induced by the stimulus, ı(t − tstim ) is the delta function with tstim as the time of stimulus, QCa is the Ca2+ charge influx and F is the Faraday’s constant. Ca2+ binding ratio: In the cell Ca2+ is bound to the Ca2+ buffers fura-2 and to the immobile endogenous buffer, which are both assumed to be always in equilibrium with free Ca2+ and non saturated. The ability of the experimentally introduced exogenous buffer to bind Ca2+ is described by its Ca2+ binding ratio that is defined as the ratio of the change in buffer-bound Ca2+ over the change in free Ca2+ : B =

[BT ]Kd,B d[BCa] = 2 d[Ca2+ ]i ([Ca2+ ]i + Kd,B )

(10)

[BT ] is the total concentration of the exogenous buffer B and Kd,B is its dissociation constant for Ca2+ . An analogous expression exists for the Ca2+ binding ratio of the endogenous buffer S (S ). 2.7.2. Ca2+ extrusion The model assumes a linear extrusion mechanism for Ca2+ [8]: jout =  [Ca2+ ]i = ([Ca2+ ]i − [Ca2+ ]i,∞ )

(11)

[Ca2+ ]i,∞ is the steady state intracellular Ca2+ concentration [Ca2+ ]i , to which a transient decays. According to Neher and Augustine [8]  ‘reflects the combined action of pumps, exchange carriers and membrane conductances’ and has the dimension l s−1 . In the following we refer to it as . ˜ In more recent publications  is defined as the extrusion rate for Ca2+ measured in s−1 [10,13,40,49,51–54]. Accordingly we use (see also [55]): =

˜ V

(12)

where V is the volume of the cell. Dynamics of Ca2+ transients: the mechanisms described in Eqs. (8)–(11) are combined to the model that describes the dynamics of the Ca2+ decay after a brief Ca2+ influx: jin,pipette + jin,stim − jout d[Ca2+ ]i d[BCa] d[SCa] + + = V dt dt dt

(13)

Substituting the time dependent terms for the buffers B and S with their Ca2+ binding ratios B and S , respectively, [BCa] with its equilibrium value [BCa] = [Ca2+ ]i [BT ]/(Kd + [Ca2+ ]i ) and  loading = (VRP )/(DB ) we get [8]: −1 · ([BCap ] − ([Ca2+ ]i [BT ])/(Kd + [Ca2+ ]i )) loading

d[Ca2+ ] dt

i

=

+(/V ˜ )([Ca2+ ]i,∞ − [Ca2+ ]i ) + ((nCa ı(t − t0 ))/V ) 1 + B + S

(14)

Neher and Augustine [8] linearized Eq. (14) to simplify the solution for this equation for the case in which the baseline for [Ca2+ ]i is constant, the concentration of the exogenous buffer [BT ] is constant and does not influence the baseline Ca2+ , and when the amplitudes of the Ca2+ transients are small. Tank et al. [9] defined ‘small’ as smaller than 0.5Kd,Fura . When the Ca2+ pulse is short compared to the decay time constant the delta function equals zero. Under these conditions the decay of a voltage-induced Ca2+ transient can be described as transient =

1 + B + S for :  − ((1 + S )/loading )

baseline = const [BT ] = const small [Ca2+ ]i

(15)

 transient is proportional to B and a linear fit to the data has its negative x-axis intercept at 1 + S , yielding the endogenous Ca2+ binding ratio of the cell. The slope of this fit is the inverse of the

91

Ca2+ extrusion rate  (Eq. (12)). The y-axis intercept yields the time constant  endo for the decay of the Ca2+ transient as it would appear in the cell without exogenous buffer.

2.8. Analysis of calcium buffering After establishing the whole-cell configuration, neurons were voltage-clamped at their resting potentials (∼−60 mV) and intracellular dye loading was monitored at 360 nm excitation, the isosbestic point of fura-2. Frames were taken at 30 s intervals (7 ms exposure time). We estimated the intracellular fura-2 concentration for different times during the loading curve, assuming that cells were fully loaded when the fluorescence reached a plateau and the fura2 concentration in the cell and in the pipette is equal (200 ␮M) [40]. To control the physiological status of the cell, loading curves were also measured at 380 nm excitation. Divergence between the 360 and 380 nm loading curve indicated rising intracellular Ca2+ concentration, related to beginning cell death. During fura-2 loading, voltage-activated Ca2+ influx was induced by stepping the voltage-clamped membrane potential for 50 ms to −5 mV, which is in the range where ICa has its maximum [21,22]. To monitor the induced Ca2+ transients ratiometrically, pairs of images at 340 nm (15 ms exposure time) and 380 nm (6 ms exposure time) excitation were acquired at 13.3 Hz for ∼12 s. Typically 3 Ca2+ transients were induced during loading at different intracellular fura-2 concentrations approximately 50, 200 and 500 s after establishing the whole-cell configuration. Fitting the loading curve: the increase in fluorescence was measured at the ‘isosbestic wavelength’ of fura-2 to monitor loading of fura-2 into somata of the neurons (Fig. 1A and C). The loading curves were fit with the exponential function: Y = Ymax (1 − e(−(t−tbi )/loading ) )

(16)

Y is the fluorescence intensity in ADU or, after rescaling, the concentration of fura-2. Ymax is the plateau of fluorescence or the maximum concentration of fura-2 (200 ␮M),  loading is the loading time constant, tbi is the time of break-in. Initially, the data were fit with Eq. (16) using the least squares method implemented by the ‘R function’ nls (Nonlinear Least Squares, R Development Core Team [2007], R: A language environment for statistical computing [R Foundation for Statistical Computing, Vienna, Austria], http://www.R-project.org). Reasonable ‘initial guesses’ were obtained by the ‘R function’ SSasympOff (Asymptotic Regression Model with an Offset, R Development Core Team [2007]). To provide the absolute fura concentrations during the time course of the experiment the fluorescence intensity was rescaled to fura-2 concentrations (Ymax = 200 ␮M) and the rescaled data were fit again with Eq. (16). This fitting procedure describes the overall loading process very well. On a shorter time scale, however, the data sometimes fluctuated around the fit (Fig. 3A), probably due to changes in access resistance [10]. Improved estimates of the fura-2 concentration at any time throughout the loading could be obtained from subsequent smoothing splines fits [56] (Fig. 3B), which were computed with the ‘R function’ smooth.spline (R Development Core Team [2007]). For the analysis of calcium buffering the fura-2 concentrations obtained from the smoothing spline fit were used. Fitting the transients: fluorescence ratios (R) from 340 and 380 nm excitation (13.3 Hz) after a brief voltage-activated Ca2+ influx were used to monitor the time course of these transients. Values for each wavelength were acquired in ADU from a rectangular region of interest. These ADU counts were used to compute ratios that were then transformed to absolute Ca2+ concentrations as described above. The decay of the transient was fit with the

92

A. Pippow et al. / Cell Calcium 46 (2009) 87–98

a linearized one compartment model were implemented in R (R Development Core Team [2007], http://www.R-project.org). 2.10. Simulations Ca2+ decay kinetics were numerically computed from the differential equation (Eq. (14)) that describes the Ca2+ dynamics in a single compartment model. The computations were performed with the ‘R function’ lsoda implemented in the odesolve library in R (odesolve: Solvers for Ordinary Differential Equations, R package version 0.5-20). Values were taken from the previous analysis of calcium buffering so that the resulting kinetics represented typical neurons from this study. The values used are indicated in the figure caption. 3. Results Fig. 3. Fitting the loading curves: (A and B) loading curves were acquired as in Fig. 1. (A) The data (in ADU) were first fit with a monoexponential function (Eq. (16)). To provide the fura-2 concentration the data were rescaled to fura-2 concentrations with Ymax = 200 ␮M, which is the fura-2 concentration in the recording pipette. The rescaled data were refit (Eq. (16)) (for details see Section 2). This fit describes the overall loading process very well. On a shorter timescale, however, the data sometimes fluctuated around the fit, probably caused by changes in the access resistance. Improved, time resolved estimates of the fura-2 concentrations could be obtained from a subsequent smoothing spline fit as shown in (B). The residuals demonstrate the improvement of the fits.

monoexponential function: [Ca2+ ]i (t) = Sdrift t + [Ca2+ ]i,0 e−t/transient

(17)

Sdrift t is a linear drift term taking bleaching into account (with Sdrift for the slope and t for time), [Ca2+ ]i,0 the amplitude of the signal at its peak and  is the decay constant. Parameters of Eq. (17) were estimated using the ‘R function’ nls (R Development Core Team [2007]). Starting values for Sdrift were obtained from the slope of the baseline, for [Ca2+ ]i,0 by subtracting the baseline of the signal from the peak amplitude, and for  from the time point where the amplitude had decreased e-fold. Fitting the linearized model: time constants of transients were plotted as a function of B values and fit with a linear function Y = ˇ0 + ˇ1 x using the ‘R function’ lm (R Development Core Team [2007]). To estimate the variance of the slope (extrusion rate ), the y-axis intercept (endogenous decay constant ) and the negative x-axis intercept (endogenous Ca2+ binding ratio S ) we used the bootstrap method [57] implemented in the boot library in R (fixed-x resampling, 1000 bootstrap samples, boot: Bootstrap R Functions, R package version 1.2-27). This resulted in bootstrap distributions (n = 1000) for each of the parameters (i.e. extrusion rate , endogenous decay constant  and endogenous Ca2+ binding ratio S ). The distributions were log-transformed to make them closer to a Gaussian. To determine differences in means between the different cell types, ANOVA was performed; post hoc pairwise comparisons were performed using t-tests with the Holm method for p-value adjustment. A significance level of 0.05 was accepted for all tests. 2.9. Tools for data analysis Electrophysiological data were analyzed with the software Pulse (version 8.63, HEKA-Elektronik) and Igor Pro 6 (Wavemetrics, including the Patcher’s Power Tools plug-in, http://www.mpibpc. gwdg.de/abteilungen/140/software/index.html). The MaxChelator program was used to compute the free metal concentrations (Mg2+ and Ca2+ ) in EGTA buffered solutions, which were used for calibration of the Ca2+ indicator fura-2 (WinMaxc32 version 2.50, http://www.stanford.edu/∼cpatton/maxc.html). All functions (Eqs. (15)–(17)) that were used to fit the Ca2+ buffering related data with

In this study, we analyzed the cellular parameters of insect olfactory interneurons of the antennal lobe that determine the intracellular Ca2+ dynamics. The experiments were performed on somata of (a) isolated olfactory interneurons from the AL of adult P. americana in short term cell culture and (b) identified uniglomerular projection neurons (uPN) and two types of previously described local interneurons (type I LN and type II LN, [22]) in an intact brain preparation from adult male P. americana. In the brain, neurons were first identified using IR-DIC by the position and size of the somata. The uPNs are located in the ventral portion of the ventrolateral somata group (VSG [22,58]). The two types of LNs are situated in the dorsal portion of the VSG [22]. Their identity was subsequently confirmed by electrophysiological and morphological characteristics. The latter were revealed by the single cell label provided by the Ca2+ indicator fura-2. For the electrophysiological classification, we used the distinct calcium currents of each cell type as described by Husch et al. [22] (Fig. 4). Using the ‘added buffer approach’ in combination with electrophysiological recordings and optical imaging, we aimed to quantitatively analyze the Ca2+ influx (jin,stim ), the Ca2+ binding ratio S and the extrusion rate , which are all cellular parameters that determine cytosolic Ca2+ dynamics, in identified cell types. 3.1. Ca2+ resting level and Ca2+ influx The neurons were voltage-clamped at their resting potential (∼−60 mV). Ca2+ transients were elicited by stepping the voltageclamped membrane potential for 50 ms to −5 mV, which is in the range where ICa is maximal [21,22]. Using ion substitution and pharmacological blockers it has also been shown that this current is indeed carried by Ca2+ [21]. Thus, assuming a single compartment model, the integrated current could be used to estimate the amount of Ca2+ that entered the cell via voltage-activated Ca2+ channels. Monitored by optical Ca2+ imaging the Ca2+ influx induced a clear and reproducible Ca2+ signal. Resting Ca2+ concentration: The concentrations of free Ca2+ were determined from the ratios of the imaging signals applying the approach from Grynkiewicz et al. [38] (Eq. (2)). The calibration was performed in vivo (in living cells) and in vitro (in solution). We used in vivo calibration in the first series of experiments that was performed with cultured neurons. For all experiments on cells in the intact brain, which were used for statistical comparisons between cell types, we performed in vitro calibrations. The advantages and disadvantages of both methods are discussed below. The in vivo calibration solutions (Rmin : no Ca2+ ; Rdef : 0.14 ␮M Ca2+ ; Rmax : 10 mM Ca2+ ) were introduced into the cell via the patch pipette. When the fluorescence, imaged at the isosbestic point of fura-2 (360 nm), had reached equilibrium, ratio images (from 340 and 380 nm excitation) were acquired

A. Pippow et al. / Cell Calcium 46 (2009) 87–98

93

Fig. 4. Calcium currents from identified antennal lobe neurons: (A) ICa from a uPN, a type I and a type II LN. ICa was induced by 50 ms voltage steps from a holding potential of −80 to 60 mV in 10 mV increments. (B) I–V relations of calcium currents recorded from uPNs, types I and II LNs. Note the differences in the voltage dependence of ICa between the neuron types and that the current amplitude is larger in type II LNs than type I LNs or uPNs. The results agree with a previous study [21] and the distinct properties of ICa for each cell type were used to help identifying the neuron types.

for each solution. The acquired values were: Rmax = 2.701 ± 0.479 (n = 10); Rmin = 0.136 ± 0.009 (n = 6); Rdef = 0.231 ± 0.066 (n = 13); Kd,Fura,eff = 3.637 ± 2.573 ␮M and isocoefficient ˛ = 0.205 ± 0.025, resulting in a Kd for fura of Kd,Fura = 0.427 ± 0.311 ␮M. For in vitro calibration, a drop of each calibration solution (75 ␮l, Rmin : no Ca2+ ; Rdef : 0.14 ␮M Ca2+ ; Rmax : 10 mM Ca2+ ) was placed on a sylgard coated recording chamber. Ratio images (from 340 to 380 nm excitation) were acquired for each solution. The acquired values were: correction factor [39] P = 0.77 ± 0.19, resulting in: Rmax = 1.130 ± 0.157 (n = 9); Rmin = 0.132 ± 0.005 (n = 9); Rdef = 0.290 ± 0.011 (n = 9); Kd,Fura,eff = 0.744 ± 0.147 ␮M and isocoefficient ˛ = 0.212 ± 0.038. According to Eq. (6) the Kd for fura-2 was Kd,Fura = 0.191 ± 0.047 ␮M. The mean resting level of free Ca2+ just before the stimulation was in cultured cells 0.21 ± 0.22 ␮M (n = 20), in uPNs 0.05 ± 0.04 ␮M (n = 11), in type I LNs 0.17 ± 0.09 ␮M (n = 10) and in type II LNs 0.12 ± 0.05 ␮M (n = 8). The difference in mean resting Ca2+ concentration was significant between the uPNs and the two LN types (p < 0.001) and between types I and II LNs (p = 0.002, Fig. 5A). 3.1.1. Cell volume Assuming the cell bodies are homogeneous spheres, the volume (V) was estimated from the whole-cell capacitance: C = εε0 A/d. ε is the dielectric constant of the cell membrane, ε0 is the polarizability

Fig. 5. Ca2+ resting level, cell volume and Ca2+ influx of uPNs: n = 11; type I LNs: n = 10; type II LNs: n = 8. (A) Mean resting Ca2+ concentration. The Ca2+ resting levels were determined from the baseline of the signals before stimulation. uPNs: [Ca2+ ]i,∞ = 0.05 ± 0.04 ␮M; type I LNs: [Ca2+ ]i,∞ = 0.17 ± 0.09 ␮M; type II LNs: [Ca2+ ]i,∞ = 0.12 ± 0.05 ␮M. The differences in Ca2+ resting levels were significant between all neuron types. (B) Mean cell volume determined from the whole cell capacity measurements. uPNs: V = 13 ± 2 pL; type I LNs: V = 23 ± 9 pL; type II LNs: V = 57 ± 18 pL. The differences in volume were significant between all neuron types. (C) Mean ICa induced by 50 ms voltage steps from −60 mV to −5 mV from uPNs, type I LNs and type II LNs. (D) Mean charge influx from ICa shown in (C). The Ca2+ influx was determined from ICa using nCa = QCa /2F. uPNs: nCa = 0.30 ± 0.15 fmol; type I LNs: nCa = 0.12 ± 0.08 fmol; type II LNs: nCa = 0.50 ± 0.17 fmol. The differences in Ca2+ influx were significant between all neuron types. (E) Increase in total intracellular Ca2+ concentration. uPNs: [Ca2+ ]tot = 25 ± 14 ␮M; type I LNs: [Ca2+ ]tot = 6 ± 4 ␮M; type II LNs: [Ca2+ ]tot = 9 ± 3 ␮M. The differences in total intracellular Ca2+ concentrations between uPNs and both LN types were significant. The differences between type I LNs and type II LNs were not significant. Post hoc pairwise t-tests were used to assess statistical significance. ***p < 0.001; **p < 0.005; n.s., not significant.

94

A. Pippow et al. / Cell Calcium 46 (2009) 87–98

of free space, d its thickness and A is the area of the cell membrane. Using values given in Hille [59] (ε = 2.1; ε0 = 8.85 × 10−12 C V−1 m−1 ; d = 2.3 nm), the specific capacitance of the cell membrane √ is 0.01 pF ␮m−2 . Accordingly the volume is V = A3/2 /6 . The whole-cell capacitance was measured using the capacitance compensation (C-slow) of the EPC9 (HEKA-Elektronik). All calculations were performed with the Patcher’s-Power-Tools extension for Igor Pro. The mean whole-cell membrane capacitance CM of uPNs was 26.0 ± 3.3 pF, of type I LNs 39.1 ± 10.1 pF and of type II LNs 70.1 ± 15 pF, which is well in the range previously determined by Husch et al. [22]. This corresponds to the somata volume for uPNs as 13 ± 2 pL (n = 11), for type I LNs 23 ± 9 pL (n = 10) and for type II LNs it was 57 ± 18 pL (n = 8). Differences in cell volume were significant for all cell types (p < 0.001, Fig. 5B). The mean volume for unidentified dissociated cell bodies was 17 ± 6 pL (n = 20). 3.1.2. Ca2+ influx The charge influx during a voltage pulse is given by ʃdICa /dt = Qca . Accordingly the Ca2+ influx is: nCa = QCa /2F. The Ca2+ influx during a 50 ms voltage step from −60 mV to −5 mV (Fig. 5C) was 0.22 ± 0.05 fmol (n = 20) in the unidentified cultured AL neurons, 0.30 ± 0.15 fmol (n = 11) in uPNs, 0.12 ± 0.08 fmol (n = 10) in type I LNs and 0.50 ± 0.17 fmol (n = 8) in type II LNs. The differences in Ca2+ influx between all cell types were significant (p < 0.001, Fig. 5D). 3.1.3. Ca2+ concentration The increase of total intracellular Ca2+ concentration ([Ca2+ ]tot = QCa /2FV) during the voltage pulse was determined from the Ca2+ influx and the cell volume. The change in total calcium concentration ([Ca2+ ]tot ) reflects the sum of the change in free calcium [Ca2+ ]i and the change of bound calcium [CaS]. The total increase in cytosolic Ca2+ concentration was in dissociated cell bodies 14 ± 5 ␮M (n = 20); in uPNs, the total cytosolic Ca2+ concentration increased by 25 ± 14 ␮M (n = 11), in type I LNs by 6 ± 4 ␮M (n = 10) and in type II LNs by 9 ± 3 ␮M (n = 8). The differences in increase of total Ca2+ concentration ([Ca2+ ]tot ) were significant between the uPNs and the two LN types (p < 0.001), but not significant between type I and type II LNs (p = 0.21, Fig. 5E). 3.2. Dye concentration from loading curves The intracellular concentration of the calcium indicator at any time during the experiment was determined from the loading curves. The cells were loaded via the patch pipette whose solution contained 200 ␮M fura-2. During dye loading, the fluorescence of the cell body was monitored at 360 nm excitation to determine the dye concentration and at 380 nm excitation to control the condition of the cell. After establishing the whole-cell configuration, fluorescence in the cell bodies appeared within seconds and fluorescence intensity increased until it reached a stable value. During loading, fluorescence images were obtained in 30 s intervals. The time course of increasing fluorescence (loading curve) were fit with an exponential function (Eq. (16), for details see Section 2). Such a loading curve of a cell body is shown in Fig. 3. The average time constants for dye loading determined from the exponential fits (Eq. (16)) were 667 ± 314 s for the cultured neurons (n = 19) and 1018 ± 525 s (n = 25) for in situ recordings. On a shorter time scale the data sometimes fluctuated around the exponential fit, probably due to variations in access resistance [10]. To account for these variations and to determine the instantaneous dye concentration more precisely, we fit the data with a smoothing spline. The improvement that is achieved by using this approach is demonstrated in Fig. 3, in which the residuals of the two fitting procedures are compared. (Table 1).

Table 1 Summary of symbol definitions. Symbol

Meaning

[BCa] [BT ] [Ca2+ ]i [Ca2+ ]i,∞ [SCa] [ST ] ADU Fx Kd,Fura Kd,S nCa P QCa R V ˛  ˜ [Ca2+ ]i [Ca2+ ]tot B S  endo  loading  transient

Concentration of calcium bound buffer B Total concentration of exogenous buffer B Free intracellular calcium ion concentration Steady state (resting) level of free calcium Concentration of calcium bound buffer S Total concentration of endogenous buffer S Analog-to-digital units Fluorescence at excitation wavelength x Dissociation constant of fura-2 Dissociation constant of endogenous buffer S Calcium influx Correction factor for in vitro calibration Calcium charge influx Ratio of Fx /Fy Accessible cell volume Isocoefficient Calcium extrusion rate  ·V Amplitude of a calcium transient ([Ca2+ ]i − [Ca2+ ]i,∞ ) Increase of total calcium concentration Calcium binding ratio of exogenous buffer B Calcium binding ratio of endogenous buffer S Endogenous decay time constant Fura-2 loading time constant Decay constant of a calcium transient

3.3. Ca2+ handling properties The kinetics of cytosolic Ca2+ signals are strongly dependent on the endogenous and exogenous (added) Ca2+ buffers of the cell. The amplitude and decay rate of the free intracellular Ca2+ changes with increasing exogenous buffer concentration: the amplitude of free Ca2+ is decreasing and the time constant  transient of the decay is increasing, as shown in Fig. 1. If the buffer capacity of the added buffer is known, the time constant of decay ( transient ) can be used to estimate by extrapolation the Ca2+ signal to conditions, with only endogenous buffers present (−B = 1 + S ). The model used for this study (Eq. (15) and [8]) assumes that the decay time constants  transient are a linear function of the Ca2+ binding ratios (B and S ). S was determined from the negative x-axis intercept of this plot (Fig. 6). The point of intersection of the linear fit with the y-axis denotes the endogenous decay time constant  endo (no exogenous Ca2+ buffer in the cell). The slope of the fit is the inverse of the linear extrusion rate (). To estimate the variability of these parameters, which were determined by linear fits, we used a bootstrap approach (n = 1000) as described in Section 2. The endogenous Ca2+ binding ratio was in cultured neurons 317 ± 144, in uPNs 418 ± 152, in type I LNs 154 ± 49 and in type II LNs 672 ± 280. The endogenous decay time constant was in cultured neurons 3.7 ± 0.5 s, in uPNs 1.8 ± 0.3 s, in type I LNs 1.1 ± 0.2 s and in type II LNs 2.2 ± 0.2 s. The extrusion rate was determined with 84 ± 29 s−1 in cultured neurons, in uPNs it was 221 ± 44 s−1 , in type I LNs 141 ± 19 s−1 and in type II LNs 305 ± 101 s−1 . The differences in endogenous Ca2+ binding ratio, endogenous decay time constant and extrusion rate were significant between the three cell types (p < 0.001, Fig. 7). A summary of the presented data is contained in Table 2. 4. Discussion In this study, we used the ‘added buffer’ approach in combination with patch-clamp recordings and fast optical imaging to analyze the cellular parameters that determine the dynamics of intracellular calcium signaling in central olfactory neurons of P. americana. The goal was to investigate possible correlations between Ca2+ handling properties and electrophysiological charac-

A. Pippow et al. / Cell Calcium 46 (2009) 87–98

95

Fig. 6. Ca2+ binding ratios and Ca2+ extrusion rates: (A) transients of free Ca2+ concentrations at increasing fura-2 concentrations in different types of AL neurons. Ca2+ transients were elicited by 50 ms voltage pulses to -5 mV from a holding potential of −60 mV. The fura-2 concentrations are given in subscripts of the decay time constants. The decay of the free Ca2+ concentration could be well fit with a monoexponential function (Eq. (17); see residuals). The time constants ( transient ) for each fit are given in the graphs. (B) The decay time constants of Ca2+ transients were plotted as a function of the Ca2+ binding ratio (B ), which was calculated from the intracellular fura-2 concentration, the Kd of fura-2 and the resting concentration of free intracellular Ca2+ (see Section 2). The data were fit with a linear function (Eq. (15)) to determine the endogenous time constant ( endo ) for the Ca2+ decay (y-axis intercept), the endogenous Ca2+ binding ratio (S ) (negative x-axis intercept) and the extrusion rate () (slope). Dotted lines indicate 95% confidence bands of the fits.

be advantageous to handle the high Ca2+ influx necessary for information processing in this non-spiking cell type and might prevent a global uncontrolled (possibly toxic) increase of intracellular Ca2+ . Accordingly a high Ca2+ influx leads only to a relative moderate and damped global increase in free intracellular Ca2+ . In type I LNs the Ca2+ binding ratio is significantly lower. Accordingly, Ca2+ would spread more easily across the cell. Currently it is not clear, if this is of functional significance for information processing, or if a high binding ratio is not necessary because the Ca2+ influx during normal neuronal function is lower in these cells. To better understand and illustrate how the intracellular Ca2+ dynamics are quantitatively affected by these properties and differ between cell types, we simulated the dynamics of free Ca2+ in the soma with different sets of values for the increase in total Ca2+ concentration, Ca2+ binding ratio and extrusion rate. The scheme in Fig. 8A demonstrates which parameters have been determined in this study and Fig. 8B shows simulated Ca2+ decays of a uPN, a type I LN and a type II LN using parameter values as they were determined for each neuron type in this study. Type II LNs had the highest and type I LNs had the lowest values for the Ca2+ influx (nCa ), the endogenous Ca2+ binding ratio (S ), the extrusion rate () and

teristics. The focus of this study was on the quantitative comparison between identified cell types in the intact brain. However, we first performed a series of experiments on unidentified AL neurons in culture. Generally the data from these neurons showed a larger scatter in the raw data (Fig. 6), which was caused at least in part by the heterogeneity of the neuron sample. We show these data along the data from the identified neurons to demonstrate how a reliable identification and selection of neurons increases the homogeneity of the sample (Fig. 6). We compared uniglomerular projection neurons and two physiologically different types of local interneurons. uPNs generate Na+ driven action potentials upon odor stimulation or depolarizing current injection. Type I LNs also generate Na+ driven action potentials, whereas type II LNs do not possess voltage-activated Na+ channels and Ca2+ is the only cation that enters the cell to contribute to membrane depolarizations. Our results show significant differences in Ca2+ handling properties between these electrophysiologically distinct neuron types. Type II LNs, for example, which strongly depend on Ca2+ for membrane depolarization, have the highest voltage-dependent Ca2+ influx, Ca2+ binding ratio and Ca2+ extrusion rate. This might Table 2 Summary of buffering related parameters.

nCa (fmol) V (pL) [Ca2+ ]tot (␮M) [Ca2+ ]i,∞ (␮M)  (s−1 )  endo (s) S n Values are means ± SD.

cell culture

uPN

Type I LN

Type II LN

0.22 ± 0.05 17 ± 6 14 ± 5 0.21 ± 0.22 84 ± 29 3.7 ± 0.5 317 ± 144 20

0.30 ± 0.15 13 ± 2 25 ± 14 0.05 ± 0.04 221 ± 44 1.8 ± 0.3 418 ± 152 11

0.12 ± 0.08 23 ± 9 6±4 0.17 ± 0.09 141 ± 19 1.1 ± 0.2 154 ± 49 10

0.50 ± 0.17 57 ± 18 9±3 0.12 ± 0.05 305 ± 101 2.2 ± 0.2 672 ± 280 8

96

A. Pippow et al. / Cell Calcium 46 (2009) 87–98

4.1. Comparison of data between different studies The absolute values that we determined for the parameters that control Ca2+ handling in antennal lobe neurons are well in the range that have been described in other vertebrate and invertebrate neurons. In this context, however, it is important to note that a meaningful comparison is impossible since in most previous studies no measure of variability was associated to the estimated parameters (for further discussion see below). In our study, the resting levels were 0.05 ␮M for uPNs, 0.12 ␮M for type II LNs and 0.17 ␮M for type I LNs, which is in the range observed in other preparations (0.145–0.26 ␮M in rat neurohypophysial nerve endings [60], 0.157 ␮M in motoneurons of the nucleus hypoglossus from mouse [51], 0.02–0.12 in snail neurons of H. pomatia [61], 0.046 ␮M in the calyx of Held [49]). The Ca2+ binding ratios, S , determined with the linearized Neher–Augustine method (see Eq. (15)), were 154 for type I LNs, 418 for uPNs and 672 for type II LNs. This is well in the range of estimates from 40 in the calyx of Held [49] or adrenal chromaffin cells [42] to 900 in cerebellar Purkinje cells of 6-day-old rats (for review see [62,63]). The time constants for the decay of the calcium signals were 1.1 s for type I LNs, 1.8 s for uPNs and 2.2 s for type II LNs. Previous estimates for decay time constants ranged from 1 to 5 s in dissociated neurons from the rat nucleus basalis [64], 0.7 s in the soma of hypoglossal motoneurons from mouse [51], 90 ms in the calyx of Held [49] to 70 ms in dendritic regions of cortical layer V neurons [10]. Decay time constants in somata seem to be much larger than in dendritic or synaptic regions. This is physiologically sound, because in dendritic or synaptic regions, signals with higher spatio-temporal resolution are necessary for information processing. The relatively slow decay constants in somata are, at least in part, caused by the small ratio of surface/volume. Corresponding to the long decay time constants, AL neurons have a relatively low Ca2+ extrusion rate between 141 s−1 (type I LNs) and 305 s−1 (type II LNs). Previously determined extrusion rates range from 60 s−1 in mouse hypoglossal motoneurons [51] and 900 s−1 in the calyx of Held [49] to 2000 s−1 in dendritic regions of cortical layer V neurons [10]. Fig. 7. Significance of the differences in the endogenous calcium binding ratio (S ), the endogenous Ca2+ decay time constant ( endo ) and the Ca2+ extrusion rate () between the neuron types: data were obtained from the graphs in Fig. 6B as described in the text. (A–C) To estimate the variance of the slope (extrusion rate ), the y-axis intercept (endogenous decay time constant  endo ) and the negative x-axis intercept (endogenous Ca2+ binding ratio S ) we used the bootstrap method (with 1000 bootstrap samples), which provided (mock) distributions (n = 1000) of the parameters for uPNs, type I LNs and type II LNs. The vertical lines indicate the means. (A) Histograms of  endo . (B) Histograms of . 19 counts between 600 and 1200 s−1 for type II LNs are not shown. (C) Histograms of S . 19 counts between 1500 and 3100 for type II LNs are not shown. Subsequently the distributions were log-transformed to make them closer to a Gaussian and ANOVA was performed to determine differences in the means between different neuron types. Post hoc pairwise t-tests were used to assess statistical significance. ***p < 0.001.

the endogenous decay constant ( endo ). However, the increase in total Ca2+ concentration [Ca2+ ]tot was not significantly different due to differences in cell volume. Compared to both LN types uPNs had an intermediate Ca2+ influx, Ca2+ binding ratio, extrusion rate and endogenous decay constant. However, because of their small volume the increase of total Ca2+ was significantly higher. Fig. 8C.1 shows the Ca2+ decays simulated after the same increase in total Ca2+ concentration for all three cell types. Fig. 8C.2–C.4 shows the Ca2+ signals simulated in each case with two (out of three) parameters set to a constant value. The third parameter was set to the specific value for each cell type (see Fig. 8B). These simple simulations demonstrate convincingly the strong dependence of the intracellular free Ca2+ concentration on these parameters.

4.2. Methodical implications In this and in many previous studies,  endo , S , and  have been determined by fitting data points of several experiments, resulting in discrete values usually without any measure for variability or error, which hinders comparison between cell types. We applied a bootstrap approach to estimate the variability of  endo , S , and  for a further statistical comparison. For this study, in which each experiment was performed in exactly the same way, we consider this as a valid approach to determine if significant differences between cell types exist. However, it must be noted that the bootstrap approach does not account for systematic errors in these complex measurements, and for a comparison between different studies a thorough error analysis would be desirable. We determined the intracellular Ca2+ dynamics and absolute 2+ Ca concentrations using the ratiometric Ca2+ indicator fura-2, which was calibrated in vivo or in vitro using the approach from Grynkiewicz et al. [38] and Neher [41]. We used in vivo calibration for the series of experiments with the cultured neurons and in vitro calibration for the comparison of different cell types in the intact brain. Often in vivo calibration is considered preferable. It is argued that the parameters from in vitro calibration may give unrealistic results, because the properties of fluorescent indicators change in the intracellular environment [39,65,66]. While this is certainly true, it has also to be considered that the binding properties of the Ca2+ chelators used to control the free Ca2+ concentration in the calibration solutions may also change in the intracellular environment, which would change the concentration of free Ca2+ in the calibra-

A. Pippow et al. / Cell Calcium 46 (2009) 87–98

97

Fig. 8. Simulated decays of free intracellular Ca2+ , [Ca2+ ]i : (A) scheme of the analyzed intracellular parameters (abbreviations see Table 1), which determine the intracellular Ca2+ dynamics. (B and C) Simulated decays of [Ca2+ ]i in uPNs, type I and type II LNs with cell-specific experimentally derived sets of values for increase of total Ca2+ concentration ([Ca2+ ]tot ), endogenous Ca2+ binding ratio (S ) and Ca2+ extrusion rate (). The specific values are given in each graph. All other parameter values were set to ‘typical values’ and kept constant in all simulations: [BT ] = 0 ␮M; [BCap ] = 200 ␮M;  loading = 1000 s; Kd,Fura = 0.191 ␮M; [Ca2+ ]i,∞ = 0 ␮M. Decays were computed with the differential Eq. 14. (B) Simulated Ca2+ decays with parameter values for [Ca2+ ]tot , S and  as they were determined for each neuron type in this study. (C) To demonstrate the dependence of the decay kinetics from [Ca2+ ]tot , S and , decays were simulated with one or two of the three parameters set to the same value for all neuron types, while the other parameters were kept neuron specific. For the parameters that were set to the same value we used the parameter values that were determined for the uPNs in this study. Parameters that were set to the same value in the simulated decays: (C.1): [Ca2+ ]tot ; (C.2): [Ca2+ ]tot and ; (C.3): [Ca2+ ]tot and S ; (C.4): S and .

tion solutions. We also observed that the properties of the dissolved Ca2+ indicator change significantly within a few days, which potentially can induce large calibration errors because a series of the in vivo calibrations cannot be finished within a short time. This might be reflected in the larger variability of the Kd,Fura when it is determined with in vivo calibration compared to in vitro calibration. Considering these arguments we conclude that in vivo calibration is not necessarily preferable compared to in vitro calibration. Taken together we have shown that the distinct intrinsic electrophysiological properties from different types of central olfactory neurons in the insect AL are strongly correlated with their cell specific Ca2+ handling properties as measured in the somata. We consider this work as a significant step to better understand the mechanisms for processing olfactory information on the cellular level. The next important step in this context will be the analysis of the calcium handling properties in more distal cellular compartments, directly where synaptic processing occurs. Our data are also important to help interpret and understand time constants of Ca2+ signals from experiments in which Ca2+ measurements are performed to monitor neuronal activity. Conflict of interest There are no conflicts of interest.

Acknowledgements We thank Helmut Wratil for excellent technical assistance and Ron Harris-Warrick for valuable comments to earlier versions of the manuscript. This work was supported by a fellowship of the Konrad-Adenauer-Stiftung (AP) and grants of the Deutsche Forschungsgemeinschaft (PK). References [1] R.L. Davis, Olfactory learning, Neuron 44 (2004) 31–48. [2] A. Fiala, Olfaction and olfactory learning in Drosophila: recent progress, Curr. Opin. Neurobiol. 17 (2007) 720–726. [3] J.G. Hildebrand, G.M. Shepherd, Mechanisms of olfactory discrimination: converging evidence for common principles across phyla, Annu. Rev. Neurosci. 20 (1997) 595–631. [4] G. Laurent, A systems perspective on early olfactory coding, Science 286 (1999) 723–728. [5] N.J. Strausfeld, J.G. Hildebrand, Olfactory systems: common design, uncommon origins? Curr. Opin. Neurobiol. 9 (1999) 634–639. [6] R.I. Wilson, Z.F. Mainen, Early events in olfactory processing, Annu. Rev. Neurosci. 29 (2006) 163–201. [7] G.J. Augustine, F. Santamaria, K. Tanaka, Local calcium signaling in neurons, Neuron 40 (2003) 331–346. [8] E. Neher, G.J. Augustine, Calcium gradients and buffers in bovine chromaffin cells, J. Physiol. 450 (1992) 273–301. [9] D.W. Tank, W.G. Regehr, K.R. Delaney, A quantitative analysis of presynaptic calcium dynamics that contribute to short-term enhancement, J. Neurosci. 15 (1995) 7940–7952.

98

A. Pippow et al. / Cell Calcium 46 (2009) 87–98

[10] F. Helmchen, K. Imoto, B. Sakmann, Ca2+ buffering and action potentialevoked Ca2+ signaling in dendrites of pyramidal neurons, Biophys. J. 70 (1996) 1069–1081. [11] P. Kloppenburg, W.R. Zipfel, W.W. Webb, R.M. Harris-Warrick, Highly localized Ca2+ accumulation revealed by multiphoton microscopy in an identified motoneuron and its modulation by dopamine, J. Neurosci. 20 (2000) 2523–2533. [12] M. Maravall, Z.F. Mainen, B.L. Sabatini, K. Svoboda, Estimating intracellular calcium concentrations and buffering without wavelength ratioing, Biophys. J. 78 (2000) 2655–2667. [13] B.L. Sabatini, T.G. Oertner, K. Svoboda, The life cycle of Ca2+ ions in dendritic spines, Neuron 33 (2002) 439–452. [14] P. Benquet, J. Le Guen, G. Dayanithi, Y. Pichon, F. Tiaho, omega-AgaIVA-sensitive (P/Q-type) and -resistant (R-type) high-voltage-activated Ba2+ currents in embryonic cockroach brain neurons, J. Neurophysiol. 82 (1999) 2284–2293. [15] F. Grolleau, B. Lapied, Two distinct low-voltage-activated Ca2+ currents contribute to the pacemaker mechanism in cockroach dorsal unpaired median neurons, J. Neurophysiol. 76 (1996) 963–976. [16] E. Heidel, H.J. Pfluger, Ion currents and spiking properties of identified subtypes of locust octopaminergic dorsal unpaired median neurons, Eur. J. Neurosci. 23 (2006) 1189–1206. [17] P. Kloppenburg, M. Hörner, Voltage-activated currents in identified giant interneurons isolated from adult crickets gryllus bimaculatus, J. Exp. Biol. 201 (1998) 2529–2541. [18] P. Kloppenburg, B.S. Kirchhof, A.R. Mercer, Voltage-activated currents from adult honeybee (Apis mellifera) antennal motor neurons recorded in vitro and in situ, J. Neurophysiol. 81 (1999) 39–48. [19] J.D. Mills, R.M. Pitman, Electrical properties of a cockroach motor neuron soma depend on different characteristics of individual Ca components, J. Neurophysiol. 78 (1997) 2455–2466. [20] D. Wicher, H. Penzlin, Ca2+ currents in central insect neurons: electrophysiological and pharmacological properties, J. Neurophysiol. 77 (1997) 186–199. [21] A. Husch, S. Hess, P. Kloppenburg, Functional parameters of voltage-activated Ca2+ currents from olfactory interneurons in the antennal lobe of Periplaneta americana, J. Neurophysiol. 99 (2008) 320–332. [22] A. Husch, M. Paehler, D. Fusca, L. Paeger, P. Kloppenburg, Calcium current diversity in physiologically different local interneuron types of the antennal lobe, J. Neurosci. 29 (2009) 716–726. [23] S. Sachse, C.G. Galizia, Role of inhibition for temporal and spatial odor representation in olfactory output neurons: a calcium imaging study, J. Neurophysiol. 87 (2002) 1106–1117. [24] J.W. Wang, A.M. Wong, J. Flores, L.B. Vosshall, R. Axel, Two-photon calcium imaging reveals an odor-evoked map of activity in the fly brain, Cell 112 (2003) 271–282. [25] A.M. Wong, J.W. Wang, R. Axel, Spatial representation of the glomerular map in the Drosophila protocerebrum, Cell 109 (2002) 229–241. [26] M.A. Carlsson, P. Knusel, P.F. Verschure, B.S. Hansson, Spatio-temporal Ca2+ dynamics of moth olfactory projection neurons, Eur. J. Neurosci. 22 (2005) 647–657. [27] V. Egger, K. Svoboda, Z.F. Mainen, Dendrodendritic synaptic signals in olfactory bulb granule cells: local spine boost and global low-threshold spike, J. Neurosci. 25 (2005) 3521–3530. [28] C.G. Galizia, B. Kimmerle, Physiological and morphological characterization of honeybee olfactory neurons combining electrophysiology, calcium imaging and confocal microscopy, J. Comp. Physiol. A: Neuroethol. Sens. Neural Behav. Physiol. 190 (2004) 21–38. [29] C.M. Root, J.L. Semmelhack, A.M. Wong, J. Flores, J.W. Wang, Propagation of olfactory information in Drosophila, Proc. Natl. Acad. Sci. U.S.A. 104 (2007) 11826–11831. [30] S. Sachse, P. Peele, A.F. Silbering, M. Guhmann, C.G. Galizia, Role of histamine as a putative inhibitory transmitter in the honeybee antennal lobe, Front. Zool. 3 (2006) 22. [31] A.F. Silbering, C.G. Galizia, Processing of odor mixtures in the Drosophila antennal lobe reveals both global inhibition and glomerulus-specific interactions, J. Neurosci. 27 (2007) 11966–11977. [32] P. Szyszka, M. Ditzen, A. Galkin, C.G. Galizia, R. Menzel, Sparsening and temporal sharpening of olfactory representations in the honeybee mushroom bodies, J. Neurophysiol. 94 (2005) 3303–3313. [33] P. Kloppenburg, D. Ferns, A.R. Mercer, Serotonin enhances central olfactory neuron responses to female sex pheromone in the male sphinx moth manduca sexta, J. Neurosci. 19 (1999) 8172–8181. [34] H.U. Dodt, W. Zieglgansberger, Infrared videomicroscopy: a new look at neuronal structure and function, Trends Neurosci. 17 (1994) 453–458. [35] O.P. Hamill, A. Marty, E. Neher, B. Sakmann, F.J. Sigworth, Improved patchclamp techniques for high-resolution current recording from cells and cell-free membrane patches, Pflugers Arch. 391 (1981) 85–100. [36] E. Neher, Correction for liquid junction potentials in patch clamp experiments, Methods Enzymol. 207 (1992) 123–131.

[37] C.M. Armstrong, F. Bezanilla, Charge movement associated with the opening and closing of the activation gates of the Na channels, J. Gen. Physiol. 63 (1974) 533–552. [38] G. Grynkiewicz, M. Poenie, R.Y. Tsien, A new generation of Ca2+ indicators with greatly improved fluorescence properties, J. Biol. Chem. 260 (1985) 3440–3450. [39] M. Poenie, Alteration of intracellular Fura-2 fluorescence by viscosity: a simple correction, Cell Calcium 11 (1990) 85–91. [40] S.H. Lee, C. Rosenmund, B. Schwaller, E. Neher, Differences in Ca2+ buffering properties between excitatory and inhibitory hippocampal neurons from the rat, J. Physiol. 525 (2000) 405–418. [41] E. Neher, Combined fura-2 and patch clamp measurements in rat peritoneal mast cells, in: L.C. Sellin, R. Libelius, S. Thesleff (Eds.), Neuromuscular Junction, Elsevier Science Publishers, Amsterdam, 1989, pp. 65–76. [42] Z. Zhou, E. Neher, Mobile and immobile calcium buffers in bovine adrenal chromaffin cells, J. Physiol. 469 (1993) 245–273. [43] S.M. Harrison, D.M. Bers, The effect of temperature and ionic strength on the apparent Ca-affinity of EGTA and the analogous Ca-chelators BAPTA and dibromo-BAPTA, Biochim. Biophys. Acta 925 (1987) 133–143. [44] S.M. Harrison, D.M. Bers, Correction of proton and Ca association constants of EGTA for temperature and ionic strength, Am. J. Physiol. 256 (1989) C1250–C1256. [45] C. Patton, S. Thompson, D. Epel, Some precautions in using chelators to buffer metals in biological solutions, Cell Calcium 35 (2004) 427–431. [46] D.M. Bers, C.W. Patton, R. Nuccitelli, A practical guide to the preparation of Ca2+ buffers, Methods Cell Biol. 40 (1994) 3–29. [47] T.J. Schoenmakers, G.J. Visser, G. Flik, A.P. Theuvenet, Chelator: an improved method for computing metal ion concentrations in physiological solutions, Biotechniques 12 (1992) 870–874, 876–879. [48] S.P. Brooks, K.B. Storey, Bound and determined: a computer program for making buffers of defined ion concentrations, Anal. Biochem. 201 (1992) 119–126. [49] F. Helmchen, J.G. Borst, B. Sakmann, Calcium dynamics associated with a single action potential in a CNS presynaptic terminal, Biophys. J. 72 (1997) 1458–1471. [50] F. Helmchen, D.W. Tank, A single-compartment model of calcium dynamics in nerve terminals and dendrites, in: R. Yuste, A. Konnerth (Eds.), Imaging in Neuroscience and Development, Cold Spring Harbor Laboratory Press, Cold Spring Harbor, New York, 2005, pp. 265–275. [51] M.B. Lips, B.U. Keller, Endogenous calcium buffering in motoneurones of the nucleus hypoglossus from mouse, J. Physiol. 511 (1998) 105–117. [52] S.H. Lee, B. Schwaller, E. Neher, Kinetics of Ca2+ binding to parvalbumin in bovine chromaffin cells: implications for [Ca2+ ] transients of neuronal dendrites, J. Physiol. 525 (Pt 2) (2000) 419–432. [53] J. Palecek, M.B. Lips, B.U. Keller, Calcium dynamics and buffering in motoneurones of the mouse spinal cord, J. Physiol. 520 (Pt 2) (1999) 485–502. [54] B.K. Vanselow, B.U. Keller, Calcium dynamics and buffering in oculomotor neurones from mouse that are particularly resistant during amyotrophic lateral sclerosis (ALS)-related motoneurone disease, J. Physiol. 525 (Pt 2) (2000) 433–445. [55] M.B. Jackson, S.J. Redman, Calcium dynamics, buffering, and buffer saturation in the boutons of dentate granule-cell axons in the hilus, J. Neurosci. 23 (2003) 1612–1621. [56] P. Green, B.W. Silverman, Nonparametric Regression and Generalized Linear Models: A Roughness Penalty Approach, 1st ed., Chapman & Hall/CRC, 1994. [57] B. Efron, 1977 Rietz lecture—bootstrap methods—another look at the jackknife, Ann. Stat. 7 (1979) 1–26. [58] P.G. Distler, C. Gruber, J. Boeckh, Synaptic connections between GABAimmunoreactive neurons and uniglomerular projection neurons within the antennal lobe of the cockroach, Periplaneta americana, Synapse 29 (1998) 1–13. [59] B. Hille, Ion Channels of Excitable Membranes, Sinauer Associates, Inc., Sunderland, MA, USA, 2001. [60] E.L. Stuenkel, Regulation of intracellular calcium and calcium buffering properties of rat isolated neurohypophysial nerve endings, J. Physiol. 481 (1994) 251–271. [61] C.J. Schwiening, R.C. Thomas, Relationship between intracellular calcium and its muffling measured by calcium iontophoresis in snail neurones, J. Physiol. 491 (1996) 621–633. [62] E. Neher, The use of fura-2 for estimating Ca buffers and Ca fluxes, Neuropharmacology 34 (1995) 1423–1442. [63] L. Fierro, I. Llano, High endogenous calcium buffering in Purkinje cells from rat cerebellar slices, J. Physiol. 496 (1996) 617–625. [64] H. Tatsumi, Y. Katayama, Regulation of the intracellular free calcium concentration in acutely dissociated neurones from rat nucleus basalis, J. Physiol. 464 (1993) 165–181. [65] M. Konishi, A. Olson, S. Hollingworth, S.M. Baylor, Myoplasmic binding of fura2 investigated by steady-state fluorescence and absorbance measurements, Biophys. J. 54 (1988) 1089–1104. [66] E.D. Moore, P.L. Becker, K.E. Fogarty, D.A. Williams, F.S. Fay, Ca2+ imaging in single living cells: theoretical and practical issues, Cell Calcium 11 (1990) 157–179.