Abstract
The dunce (dnc) andrutabaga (rut) mutations ofDrosophila affect a cAMP-dependent phosphodiesterase and a Ca2+/CaM-regulated adenylyl cyclase, respectively. These mutations cause deficiencies in several learning paradigms and alter synaptic transmission, growth cone motility, and action potential generation. The cellular phenotypes either are Ca2+dependent (neurotransmission and motility) or mediate a Ca2+ rise (action potential generation). However, interrelations among these defects have not been addressed. We have established conditions for fura-2 imaging of Ca2+dynamics in the “giant” neuron culture system ofDrosophila. Using high K+depolarization of isolated neurons, we observed a larger, faster, and more dynamic response from the growth cone than the cell body. This Ca2+ increase depended on an influx through Ca2+ channels and was suppressed by the Na+ channel blocker TTX. Altered cAMP metabolism by the dnc and rut mutations reduced response amplitude in the growth cone while prolonging the response within the soma. The enhanced spatial resolution of these larger cells allowed us to analyze Ca2+ regulation within distinct domains of mutant growth cones. Modulation by a previous conditioning stimulus was altered in terms of response amplitude and waveform complexity. Furthermore, rut disrupted the distinction in Ca2+ responses observed between the periphery and central domain of growth cones with motile filopodia.
Second messenger pathways involving cAMP-dependent protein kinase (PKA) and Ca2+/CaM-dependent protein kinase (CaMKII) have been implicated in the cellular basis of learning and memory (Cline, 1991; Carew, 1996; Chen and Tonegawa, 1997; Bailey et al., 2000; Dickson et al., 2001). Ca2+/CaM stimulation of cAMP production (Cooper et al., 1995) and activation of CaMKII (Soderling, 2000) may link neuronal activity to its regulation of use-dependent morphological and physiological plasticity.
The earliest identified mutations to disrupt learning inDrosophila included dunce (dnc) andrutabaga (rut), which were later found to affect a cAMP-specific phosphodiesterase and a Ca2+/CaM-activated adenylyl cyclase (Dudai et al., 1976; Byers et al., 1981; Aceves-Pina et al., 1983; Livingstone et al., 1984; Tully and Quinn, 1985). Consequentially, their defects in developmental and activity-dependent neuronal plasticity have been studied extensively. The dnc and rut mutations increase and reduce, respectively, the rate of habituation of a jump-and-flight escape response (Engel and Wu, 1996). Abnormal mechanosensory neuron development and adaptation have also been reported in adult rut flies (Corfas and Dudai, 1990, 1991). Both mutations inhibit the activity-dependent facilitation and potentiation of neurotransmitter release at the larval neuromuscular junction (Zhong and Wu, 1991) while causing opposite changes in the extent of presynaptic arborization (Zhong et al., 1992). During development in culture, dnc and rut suppress growth cone motility (Kim and Wu, 1996) and alter the kinetics and activity-dependence of neurotransmitter release from the growth cone (Yao et al., 2000) and presynaptic terminal (Lee and O'Dowd, 2000). Furthermore, K+ channel regulation and action potential patterning are also disrupted in larval muscle and cultured neurons from these mutants (Zhong and Wu, 1993a; Zhao et al., 1995; Zhao and Wu, 1997), which may contribute to defects in neurotransmitter release and growth cone motility.
In several species, neural activity and Ca2+ regulate growth cone behaviors during development. The frequency of Ca2+transients determines the rate of neurite outgrowth inXenopus embryos (Gomez and Spitzer, 2000), whereas the navigation by guidepost cells involves Ca2+ signaling in the grasshopper embryo (Bentley et al., 1991). In culture, different levels of evoked Ca2+ can lead to the enhancement or suppression of motility and neurite outgrowth (for review, see Kater et al., 1988; Kater and Mills, 1991; Fields et al., 1993; Rehder et al., 1996). For example, a spatially restricted Ca2+ increase in filopodia and the leading edge of the lamellipodia interact with cAMP to direct growth cone turning (Ming et al., 1997; Zheng, 2000).
In Drosophila, the “giant” neuron culture system (Wu et al., 1989) has facilitated investigations of neuronal development that may be hampered by the small size of central neurons and their processes. We have established conditions for imaging Ca2+ dynamics in cultured giant neurons and exploited the increased spatial resolution to differentiate the region-specific and use-dependent Ca2+increase evoked by high K+ depolarization. Our initial exploration with the dnc and rutmutations identifies Ca2+ regulatory defects caused by abnormal cAMP metabolism. The results also demonstrate that Ca2+ imaging in the giant neuron culture system may facilitate genetic dissection of Ca2+-dependent neuronal development and plasticity.
MATERIALS AND METHODS
Animal stocks. All fly stocks were raised on standardDrosophila medium at room temperature. Two dncalleles, dnc1 (Dudai et al., 1976) anddnc2 (Bellen and Kiger, 1988), and tworut alleles, rut1 (Aceves-Pina et al., 1983; Livingstone et al., 1984) andrut2 (Bellen et al., 1987), were used in this study. The wild-type strain Canton-Special was used as the control. Results from the different mutant alleles within each locus (dnc or rut) have been pooled because they did not differ significantly.
Embryonic cell culture. Individual cultures were prepared from either single embryos (Seecof et al., 1971; Salvaterra et al., 1987; Wu et al., 1989) or homogenized embryos as described previously (Zhao and Wu, 1997; Yao and Wu, 1999; Yao et al., 2000). Cultures were plated on glass coverslips in modified Schneider's medium (Invitrogen, Gaithersburg, MD) containing 200 ng/ml insulin (Sigma, St. Louis, MO), 20% fetal bovine serum (Summit Biotechnologies, Ft. Collins, CO), 50 μg/ml streptomycin, and 50 U/ml penicillin (all from Invitrogen). The production of giant neurons (Saito and Wu, 1991; Zhao and Wu, 1997) involved the inhibition of neuroblast cytokinesis. This was achieved by adding 1–2 μg/ml of cytochalasin B (Sigma), which blocks actin polymerization and cell division, to the culture media. All cells were grown in a humidified chamber at room temperature (21–24°C), and the cytochalasin B-containing media was replaced with fresh culture media within 24 hr after plating.
Imaging. Media containing 4–6 μmfura-2 AM and 10–15 μl/ml pluronic acid was used to load 2- to 6-d-old cells in the dark at room temperature for 1.5–2 hr. Fura-2-loaded cells were perfused continuously in a chamber allowing for fast laminar flow of solution. The dead volume was <0.5 ml, and the time from valve opening to perfusion across the coverslip was on average 6 sec (up to 16 sec in early experiments). All of the salines used in this study were buffered at pH 7.1–7.2 with 5 mm HEPES (adjusted with NaOH). Normal bath saline contained (in mm): 128 NaCl, 2 KCl, 4 MgCl2, 1.8 CaCl2, and 35.5 sucrose. For high K+ depolarizing salines, the following modifications to normal saline were made: (1) 60 K saline, 60 mm KCl, and 70 mm NaCl; (2) 42 K saline, 42 mm KCl, and 88 mm NaCl; (3) 30 K saline, 30 mm KCl, and 90 mm NaCl; (4) 0 Ca saline, normal saline with 0 mm CaCl2, 5.8 mm MgCl2, and 5 mm EGTA; (5) 30 K 0 Ca and 60 K 0 Ca saline, 30 or 60 K saline with the above changes. The duration of stimulation varied between 5 and 30 sec, with the latter being more common because of the lack of responsiveness from mutant neurons. In Figures7B, 8, and 9, stimuli were used to achieve a signal-to-noise ratio of at least 5 and most often >10.
Giant neurons were imaged using an inverted Diaphot microscope equipped for epifluorescence (Nikon, Tokyo, Japan). Video images were obtained with an intensified CCD camera (IonOptix, Milton, MA) and a shutter/filter wheel (IonOptix) controlled by the IonWizard software (IonOptix). Images were collected with 40× [1.3 numerical aperture (NA)] or 100× (1.3 NA) oil-immersion objectives at 1 Hz or higher sampling rate, and single or averaged frames were saved on-line in a pseudocolor format. The camera gain was adjusted such that most cellular responses for the numerator (340 nm) and denominator (380 nm) were within a proper working range of the camera output (they approached neither the upper nor lower limits). This ensured a more linear measure and less disturbance by noise. Data collected with the 40× and 100× objectives were combined where applicable, because changes in magnification did not consistently alter the Ca2+ response. No autofluorescence caused by the media or microscope optics was observed, and compensation was made for spatial variations in the camera dark current, the camera gain, and the intensity of the excitation source. A full-frame (cell-free) background image collected before and a background zone used during the experiment helped compensate for the above systematic errors. Dye bleaching was not observed at the acquisition rates used in this study. The soma and processes differed in the intensity of basal fluorescence, so the camera gain and black level were adjusted to maintain the camera output within the optimal dynamic range, and these regions were imaged during different recording epochs.
The fura-2 binding constant and the coefficients for use in the Grynkiewicz formula (Grynkiewicz et al., 1985),KD(F0/FS)(R−Rmin)/(Rmax− R) = [Ca2+], were determined using a cell-free calibration (Williams et al., 1985).R is the ratio of fluorescence at 340 nm to the fluorescence at 380 nm, Rmin andRmax are the 340 nm/380 nm ratios in 0 and 10 mm Ca2+,KD is the effective dissociation constant (147 nm), andF0 andFS are the 380 nm excitation efficiencies in 0 mmCa2+ and at saturating Ca2+ (10 mm). The Ca2+ level was quantified by applying the Grynkiewicz formula to fura-2 fluorescence recorded from square zones that were 16 μm2 in area. Running averages of [Ca2+] taken from selected neuronal compartments were performed before analysis using a window of five points.
Electrophysiology. Whole-cell patch-clamp recording fromDrosophila giant neurons has been described previously (Saito and Wu, 1991; Zhao and Wu, 1997). Patch electrodes were pulled from glass capillaries (75 ml micropipettes; VWR, Chicago, IL) and had a tip resistance of 2–5 MΩ when measured in normal saline. Whole-cell recordings were obtained with an EPC-7 patch-clamp amplifier (Medical Systems Corp., Greenvale, NY), and electrodes were filled with (in mm): 144 KC1, 1 MgCl2, 0.5 CaCl2, 5 EGTA, 10 HEPES, and adjusted with KOH to a pH of 7.1.
RESULTS
Regional heterogeneity
Giant neuron cultures grown from cytokinesis-arrested embryonic neuroblasts were loaded with fura-2 AM and subsequently imaged during perfusion with depolarizing high K+saline. Comparable with the proportion previously found in cultures of non-cytokinesis-arrested larval neurons (Kim and Wu, 1991, 1996), ∼30% of these giant embryonic neurons contained neurites ending with large growth cones. After depolarization, a robust Ca2+ increase was imaged throughout single, isolated neurons. Thirty second depolarizations of the wild type by an increase from 2 to 60 mm[K+] (60 K saline) caused a Ca2+ increase to appear first in the growth cone and neurites (Fig.1). Later, an increase was observed throughout the soma, beginning in regions nearest the neurite (not shown in the image sequence). We quantified the intracellular Ca2+ transient within the growth cone to study how its regulation is distinct from the soma in terms of amplitude, kinetics, and Ca2+ source. Perfusion with high K+ saline caused a larger Ca2+ increase in the growth cone that peaked earlier and decayed more rapidly than in the cell body (Fig. 1B, inset).
The high K+depolarization-induced Ca2+ increase is distinct between the soma and growth cone. A, Phase-contrast and pseudocolor images of a wild-type cell are shown during and after perfusion of 60 mm K+ (60 K) from 0 to 30 sec. Depolarization caused a Ca2+ increase within discrete locations along the proximal neurite and at the growth cone, later appearing in the cell body (see color scale bar inB for Ca2+ levels). In the phase-contrast and 25 sec pseudocolor images, boxed regions indicate where fura-2 fluorescence was quantified.B, Quantification of intracellular Ca2+ over time accentuates the differences in response amplitude between the soma (S) and growth cone (GC). Accompanying the larger amplitude, the growth cone responds with faster kinetics (inset, amplitude normalization). Amplitude normalization was performed by subtracting the minimum value from every point and then dividing by the peak value. This caused the normalized value of each point of the trace to vary between 0 and 1. Scale bar (shown in A): 10 μm.Black bars in B indicate the perfusion of 60 K saline.
Involvement of Ca2+ and Na+channels in the Ca2+ increase
Drosophila cultured neurons express various ion channels, leading to several categories of electrical excitability when investigated with patch-clamp techniques applied to the soma (Zhao and Wu, 1997; Yao and Wu, 1999; Yao et al., 2000). However, the electrophysiological study of ion channels in the neurite and growth cone has been hampered by the small size of these structures [but seeSaito and Wu (1991)]. A pharmacological approach was taken to selectively block or isolate the involvement of specific ion channel types because their differential expression may underlie the temporal and spatial characteristics of the Ca2+signal. Depolarization in 0 Ca2+ saline (Fig. 2A) or in the presence of a Ca2+ channel blocker, cobalt (Fig. 2B), reversibly eliminated the response. Furthermore, a Na+ channel blockade by TTX decreased the amplitude of the Ca2+increase (Fig. 2C). From these results, we conclude that an influx through Ca2+ channels is required and Na+ channels are involved in the high K+ depolarization-induced response. It is possible that fura-2 loading alters aspects of ionic currents and spike activity that may affect Ca2+ influx and thus cytosolic Ca2+ concentration. To examine these potential effects, we compared Na+-dependent action potential generation (Wu et al., 1989; Saito and Wu, 1993) and whole-cell outward currents between neurons with and without fura-2 loading. We verified that fura-2 does not impair action potential generation (Fig.3A). In giant neuron cultures, the cell population can be classified (types 1–4) according to the proportion between the peak and sustained components and the decay time constants of the total outward current (Fig. 3B) (I.-F. Peng and C.-F. Wu, unpublished results). A similar classification has been used to describe outward currents in non-cytochalasin B-treated larval neurons in culture (Delgado et al., 1998). We did not observe a substantial change in the population distribution among the different classes (fura vs non-fura, type 1–1/24 vs 2/30, type 2–7/24 vs 6/30, type 3–8/24 vs 7/30, type 4–8/24 vs 18/33). Our data therefore imply that fura-2 does not strongly affect ionic currents underlying neuronal spike activity.
TTX-sensitive influx through voltage-gated Ca2+ channels. A, Depolarization in 0 Ca2+ saline (represented as a black bar within a white bar) eliminated the increase by 30 mm K+ saline (30 K). Images of peak Ca2+ are shown before, during, and after exchange of the bath saline with one lacking Ca2+.B, Application of 6 mm CoCl2during 60 K stimulation had a similar affect. C, The addition of 1 μm TTX only partially eliminated the Ca2+ increase during 30 sec of 60 K stimulation.
Action potential generation and whole-cell outward currents from fura-2-loaded giant neurons. A, Action potential generation was not impaired in fura-2-loaded neurons during whole-cell patch-clamp experiments. Representative action potential activities in response to current injection of two tonically firing neurons are shown, one loaded with fura-2 (+) and the other not (−).B, Whole-cell outward currents, composed mostly of various types of K+ currents, were measured from loaded and nonloaded cultures. Four categories (types 1–4) are shown based on the proportion between the peak and sustained components and the decay time-constants of the total outward current. The proportion of cells with a response that belonged to each category of current kinetics was unaffected by fura-2 (see Results).
Variation in the amplitude and kinetic properties of the Ca2+ response
Individual neurons displayed considerable variation in the high K+-induced Ca2+ increase. We analyzed a large number of cells for the regional differences in the evoked response (Fig.4). Compared with the soma, a larger number of growth cones responded to 30 K saline, some with a Ca2+ increase above 0.6 μm(growth cone, GC, 0.31 ± 0.14 vs soma, S, 0.05 ± 0.02; mean ± SEM in micrometers; sample size given in figure legends). Response amplitude distributions displayed obvious differences between 30 and 60 K stimulations (Fig. 4A) (S: 30 K, 0.05 ± 0.02, 60 K, 0.31 ± 0.07; GC: 30 K, 0.31 ± 0.14, 60 K, 0.78 ± 0.23). Using the stronger stimulus, we measured response latency from the beginning of the stimulus to 10% of the response amplitude, rise-time from 10 to 90% of the response, and duration from 50% of the rise to 50% of the decay. There was a clear trend of faster response kinetics in the growth cone compared with the soma (Figs. 1B, 4B) (see below).
Regional differences in dose dependence and response kinetics. A, Frequency distributions of response amplitudes during 30 sec depolarization with the 30 and 60 K salines. Scattered responses above 0.6 μm (600 nm) are represented by a single class. Neurons in this culture system varied in the extent of the Ca2+increase, with a subpopulation containing highly sensitive growth cones and cell bodies. In general, however, the soma required 60 K stimulation to elicit a distribution comparable to that produced by 30 K in the growth cone [growth cone (GC) at 30 K 0.31 ± 0.14 vs soma (S) at 60 K 0.31 ± 0.07; mean ± SEM in micrometers for all figures except Figures5 and 8]. B, Although the latency, rise-time, and duration vary among cells, the mean latency and rise-time are smaller in the growth cone [Latency (seconds): S23.06 ± 1.14 vs GC 13.24 ± 0.99; Student'st test; p < 0.005;Rise-Time (seconds): S 16.31 ± 0.76 vs GC 11.00 ± 0.93; p < 0.005; Duration (seconds): S 31.03 ± 2.37 vs GC 27.24 ± 2.52; p< 0.5]. Latency was measured from the beginning of the stimulus to 10% of the response amplitude, rise-time from 10 to 90% of the response, and duration from 50% of the rise to 50% of the decay. The number of zones, number of neurons, and number of cultures used for data presented in this and all figures are represented as zones, neurons, and cultures. For A and B,S 30 K: 30, 21, 13, 60 K: 101, 57, 32; GC30 K: 37, 21, 16, 60 K: 37, 26, 18.
The variation among neurons may stem from intrinsic differences between cell types, suggesting that a restricted cell population would reduce response variability. We investigated this possibility using the GAL4-UAS system (Brand and Perrimon, 1993) of Drosophila to express green fluorescent protein (GFP; in a UAS-GFP construct) in a subset of giant neurons derived from mushroom body neuroblasts with the lineage-specific driver 201Y-GAL4 (Yang et al., 1995). We found that normal Ca2+ measurements could be acquired despite GFP expression (Supplemental Fig. 1). However, such a restricted cell population did not reduce variability in response amplitude or change the characteristic differences between the soma and growth cone responses (Supplemental Fig. 2). Even within the GFP-labeled population, responses from the growth cone and soma of single neurons tend to covary from the population mean, suggesting that measurements from one area (growth cone or neuritic terminal) can serve as an indicator for responsiveness in other cellular regions.
dunce and rutabaga mutations disrupt the high K+ response
After the depolarization of dnc and rutneurons, several hallmarks were immediately noticeable. We observed a suppression of response amplitude (Fig.5A,B) and a change in response kinetics (Figs. 5C,6). However, the magnitude of these changes differed between the soma and growth cone. Representative soma and growth cone Ca2+ traces from single neurons are shown (Fig. 5A) (collected using 30 sec depolarizations with 60 K saline), indicating the effects of altered cAMP metabolism. As a population, dnc and rutgrowth cones showed a weaker response to 30 K stimulation (Fig.5B, 30 K) (WT, 0.31 ± 0.14; dnc, 0.02 ± 0.01; rut, 0.06 ± 0.02). With 60 K saline, mutant growth cone responses approached wild-type amplitudes (Fig. 5B, 60 K) (WT, 0.78 ± 0.21; dnc, 0.87 ± 0.40rut, 0.49 ± 0.12). In contrast, only rutsuppressed the high K+ response within the soma (WT, 0.31 ± 0.06; dnc, 0.38 ± 0.08; rut, 0.11 ± 0.02).
The dnc andrut mutations lower the sensitivity to high K+ depolarization more strongly in the growth cone than soma. A, Example traces from the growth cone and soma of dnc and rut neurons during 30 sec perfusion of 60 K saline (indicated by black bars).B, Response amplitudes were suppressed amongdnc and rut growth cones with 30 K andrut growth cones with 60 K (30 K: WT, 0.31 ± 0.14, dnc, 0.02 ± 0.01,rut, 0.06 ± 0.01; 60 K:WT, 0.78 ± 0.21, dnc, 0.87 ± 0.40, rut, 0.49 ± 0.12). For the soma, suppression was indicated only in rut (WT, 0.31 ± 0.07; dnc, 0.40 ± 0.08; rut, 0.13 ± 0.05). C, Cultured neurons could be categorized by regional differences in response latency (time from beginning of stimulation to 10% of the peak amplitude), with the growth cone response latency being either earlier or later than that of the soma. Samples of normalized growth cone and soma responses from the same cell of the two categories are shown. The dnc andrut mutations reversed the relationship found in the wild type, causing a larger percentage of soma responses to rise before the growth cone (quantified for 15, 15, and 10 WT,dnc, and rut neurons). ForB, 60 K (S:WT, 101, 57, 32, dnc, 89, 56, 28,rut, 91, 46, 20; GC: WT, 37, 26, 18, dnc, 51, 26, 21, rut, 68, 26, 16); 30 K (GC: WT, 37, 21, 16, dnc, 24, 11, 8, rut, 56, 16, 7).
Kinetic alterations of the Ca2+increase by dnc and rut with no change in basal Ca2+ levels. Similar to wild type,dnc and rut showed regional differences between the soma and growth cone in the rise-time and duration (Fig. 2, see legend for definition). However, the rise of Ca2+ within the growth cone was shortened byrut [Rise-Time(s): WT, 11.00 ± 0.93, dnc, 12.58 ± 1.23,rut, 7.81 ± 0.62], whereas both mutations more strongly increased the rise-time in the soma [Rise-Time(s) and Students t test:WT, 16.31 ± 0.76, dnc, 19.53 ± 0.73; p < 0.005; rut, 22.81 ± 1.84; p < 0.005]. dnc andrut increased response duration in the soma [Duration(s): WT, 31.03 ± 2.37,dnc, 48.40 ± 3.65; p < 0.0005; rut, 47.88 ± 3.22; p< 0.0005], with little effect on the growth cone (WT, 27.24 ± 2.52; dnc, 22.85 ± 1.80;rut, 26.84 ± 2.15). In contrast to these dynamic measures, the basal Ca2+ level was not affected in the mutants (values in nm; S:WT, 22.00 ± 0.91, dnc, 23.3 ± 0.81, rut, 21.51 ± 0.82; GC:WT, 24.61 ± 0.92, dnc, 22.73 ± 1.42, rut, 25.57 ± 3.07). S:WT, 101, 57, 32, dnc, 89, 56, 28,rut, 91, 46, 20; GC: WT, 37, 26, 18, dnc, 51, 26, 21, rut, 68, 26, 16.
Similar to wild-type neurons, the amplitude and kinetics of responses from the soma and growth cone of mutant cells also covaried from the population mean. The preferential suppression during low depolarizations (30 K saline) implies that the threshold for induction of the Ca2+ increase may be altered bydnc and rut, and the soma and growth cone may be differentially affected as well (Fig. 5B). As shown previously in wild-type neurons, response latency, defined as the time from beginning of the stimulus to 10% of the peak, was shorter in the growth cone than soma (Figs. 1, 4B, 5A). However, mutant growth cones often had a longer response latency than the soma (Figs. 5C). Together with stronger amplitude suppression in the growth cone (Fig. 5B), these results suggest that alterations in cAMP metabolism affect the reaction speed as well as the sensitivity to depolarization.
Region-specific effects were also revealed by comparisons of response kinetics (Fig. 6). The rise-time of the growth cone response was decreased more strongly by rut but that of the soma was increased equally by both mutations (Fig. 6, Rise-Time(s)) (S: WT, 16.31 ± 0.76, dnc, 19.53 ± 0.73, rut, 22.81 ± 1.86; GC: WT, 11.00 ± 0.93, dnc, 12.58 ± 1.23, rut, 7.81 ± 0.62]. Interestingly, neither mutation altered response duration in the growth cone, whereas the duration was increased within the soma (Fig. 6, Duration(s)) (S: WT, 31.03 ± 2.37,dnc, 48.40 ± 3.65, rut, 47.88 ± 3.22; GC: WT, 27.24 ± 2.52, dnc, 22.85 ± 1.80, rut, 26.84 ± 2.15), indicating that chronic changes in cAMP metabolism caused by these mutations disrupt the decay of the high K+ response as well as the rise. In contrast to these dynamic measures of Ca2+, dnc and rutdid not alter the basal Ca2+ levels in either region (Fig. 6).
Release of Ca2+ from intracellular stores (Friel and Tsien, 1992; Mironov et al., 1993) or the complex electrical activity caused by ion channel interactions could lead to multiple peaks during the high K+ response or to multiple time-constants during its decay. We found that the prevalence of a shoulder or second peak could further differentiate the soma and growth cone of Drosophila neurons. The above two response categories were distinguished by their local amplitudes with respect to the peak Ca2+ during 30 sec stimulation with a high K+ saline (Fig.7, see legend). In all genotypes, the complex waveforms were more common in the growth cone than soma (Fig.7A). However, the shoulder or second peak appeared to be suppressed more strongly in dnc than rut (Fig.7A). The shape of the Ca2+response was not related to the amplitude suppression described above, because the different waveform categories did not correlate with the peak Ca2+ level in any genotype (data not shown). Our results indicate that the Ca2+response is not a simple linear process and that alterations in cAMP metabolism may affect the individual processes differently in the growth cone and soma.
The Ca2+ waveform and its activity-dependent modulation in dnc andrut. A, High K+-induced Ca2+ responses were categorized into three groups: simple responses of a single peak (black) and more complex responses with a shoulder during decay (gray) or with two or more distinct peaks (white, deviation from the simple decay with >25% of the peak amplitude). Data from 30 sec, high K+stimulation indicates that the growth cone exhibits more complex waveforms than the soma of wild-type neurons. Dnc andrut appeared to suppress responses with a complex waveform, especially in the soma. B, A twin-pulse paradigm revealed the influence of previous activity on the Ca2+ increase. A pair of 30 sec stimulations (S1 and S2) were given at two interstimulus intervals (ISI) (30 and 160–185 sec), and Ca2+ dynamics were recorded in the growth cone. The Ca2+ waveform acquired a shoulder or second peak (white), did not change (gray), or lost its shoulder or second peak (black). Notably, at the 160–185 sec ISI, therut mutation blocked the modulation of response kinetics. The 30, 42, and 60 K salines were used according to the responsiveness of individual cells, and mutant growth cones were in general less responsive, so the results presented fordnc and rut contain a larger proportion of 60 K data. The kinetic categories did not correlate with amplitude (data not shown). Response categories are as follows: S:WT, 131, 70, 43, dnc, 92, 51, 29,rut, 88, 40, 22; GC: WT, 87, 38, 20,dnc, 46, 26, 15, rut, 74, 26, 16. Activity-dependent changes are as follows: 30 sec ISI:WT, 38, 20, 15, dnc, 39, 13, 12,rut, 60, 20, 15; 160–185 sec ISI: WT, 20, 12, 9, dnc, 28, 11, 9, rut, 30, 15, 14.
Use-dependent defects in dncand rut
The dnc and rut mutations alter activity-dependent regulation of action potential firing (Zhao and Wu, 1997) and neurotransmitter release from the synapse (Zhong and Wu, 1991; Lee and O'Dowd, 2000). In addition, dnc disrupts activity-dependent modulation of neurotransmitter release from the developing growth cone (Yao et al., 2000). We therefore investigated use-dependent Ca2+ regulation in the growth cone with a twin-pulse protocol (Fig. 7B). A pair of depolarizations (S1 and S2), given at two different interstimulus intervals (ISIs) (30 and 160–185 sec), revealed that previous depolarization altered the complexity of response kinetics in several ways. An increase (or decrease) of response complexity, as indicated by the emergence (or disappearance) of a shoulder or second peak, was observed in all genotypes. Noticeably, the response kinetics did not change in rut at the longer ISI in contrast with wild type and dnc (Fig.7B).
A previous depolarization also altered the amplitude of the Ca2+ increase (Fig.8A). A facilitation/suppression indicator, R, represents the log of the ratio of the second to the first response. (For example,R = 0 indicates no change, R = 1 indicates a 10-fold increase, and R = −1 indicates a 10-fold decrease.) Frequency distributions of R for the three genotypes are shown for both the short (30 sec) and long (160–185 sec) ISI (Fig. 8B). Among the population of cells, wild-type growth cones showed facilitation at both ISIs. We found that the dnc mutation caused less facilitation at the shorter ISI (WT, 0.16 ± 0.03; dnc, 0.09 ± 0.04), whereas rut suppressed the response at the longer one (WT, 0.20 ± 0.09; rut,−0.07 ± 0.06). Our results indicate that altered cAMP metabolism by dnc and rut affects at least two aspects of Ca2+ regulation after a previous depolarization. The likelihood of generating more complex waveforms is suppressed (Fig. 7B), and the facilitation of response amplitude (R > 1) seems to be decreased or in some cases converted to depression (R < 1) (Fig.8B).
Activity-dependent modulation of response amplitude in the growth cone. A, Example traces from the growth cones of two cells show facilitation and suppression of the high K+response amplitude during the twin-pulse protocol. A facilitation/suppression indicator, R, was calculated from the ratio of the two amplitudes [R = log (Amp1/Amp2),R = −1 represents a 10-fold suppression in amplitude, whereas R = 1 represents a 10-fold facilitation]. B, Frequency histograms of the facilitation/suppression indicator R showing the modulation of response amplitude in the growth cone. Wild-type cells showed slight facilitation of response amplitude regardless of the long (160–185 sec) or short (30 sec) ISI. In the case ofdnc, a tendency for less facilitation was observed, whereas for rut, suppression was observed at the long ISI. Data from different stimulus durations (2, 5, 10, and 30 sec) and magnitudes (30, 42, and 60 K) were combined because the mutants required stronger stimulation and because no correlation between the stimulus and facilitation/suppression was observed. 30 sec ISI:WT, 38, 20, 15, dnc, 39, 13, 12,rut, 60, 20, 15; 160–185 sec ISI: WT, 20, 12, 9, dnc, 28, 11, 9, rut, 30, 15, 14.
Spatial alterations of subcellular Ca2+ regulation
In many species, the lamellipodia, filopodial protrusions, and central domain of the growth cone vary their structure and movement depending on elevations in Ca2+ (for review, see Kater and Mills, 1991). Local Ca2+ increases may induce the growth of filopodia (Rehder and Kater, 1992; Lau et al., 1999) and determine the direction of growth cone turning (Zheng, 2000). InDrosophila, the dnc and rut mutations suppress the expansion and contraction of growth cone lamellipodia in dissociated larval CNS cultures (Kim and Wu, 1996) and influence the terminal arborization of motor axons at the larval neuromuscular junction (Zhong et al., 1992).
We therefore collected higher magnification images (with 100× objectives) to distinguish how dnc and rut alter Ca2+ regulation within structurally and functionally distinct regions of the growth cone (Fig.9). Growth cones of giant neurons had structures similar to those of larval neurons (Kim and Wu, 1991, 1996), but when viewed with phase-contrast optics, they were larger and appeared less motile in terms of the percentage change in lamellipodial area over time. The motility that we most often observed consisted of filopodial movements or lamellipodial expansion or retraction during 1 min recording epochs (Fig. 9A). This is similar to the motility of non-fura-loaded giant neurons (data not shown) and contrasts with more active lamellipodial ruffling and filopodial motility observed in larval cultures (Kim and Wu, 1991, 1996). However, in both embryonic giant neuron and larval CNS cultures, nonmotile growth cones often appeared phase-dark and had a bulging central domain, whereas motile growth cones appeared phase-light, contained more filopodia, and had a larger, flattened lamellipodia. We found that regional Ca2+ regulation could be correlated with motility (Fig.9A,B). Motile growth cones exhibited Ca2+ levels in the peripheral lamellipodia that were greater than in the central domain (Figs.9B). This spatial variation was less apparent in nonmotile than motile growth cones (Fig. 9C). A response amplitude ratio, edge/center, was used to facilitate the comparison with normalized values. In dnc and rut cells, some growth cones showed an anomalous relationship between the edge/center response ratio and motility. In dnc, peripheral Ca2+ levels were well beyond those found in motile growth cones of the wild type, whereas extreme ratios were seen in nonmotile rut growth cones (Fig. 9C). This is in stark contrast to the ability to predict motility based on regional Ca2+ signaling in the wild-type growth cone. These data suggest that altered cAMP metabolism may enhance peripheral Ca2+ signaling as compared with the central domain and disrupt the relationship between Ca2+ levels and motility.
The spatial distribution of Ca2+ within distinct regions of the growth cone.A, Phase images of a giant neuron and its growth cone (captured with 40 and 100× objectives, respectively) showing filopodial and lamellipodial movement (arrows). For this experiment, motility was indicated by changes in filopodial or lamellipodial shape over the course of 1 min. B, Pseudocolor images (B1) showing Ca2+ within this growth cone after a 5 sec stimulation with 60 K. Ca2+ levels in the peripheral lamellipodium (zones 1 and 2) reached a higher peak concentration than within the central domain (zone3), as quantified inB2. C, The ratio of peak Ca2+ at the leading edge to levels in the central domain of the wild type were larger in motile growth cones (●) than nonmotile growth cones (○) (e.g., the ratio of an average of 2–6 zones in the periphery to an average of 1–2 zones in the central domain). In dnc, ratios of peripheral-to-central peak Ca2+ levels in motile growth cones were well beyond those of wild-type neurons, whereas extreme ratios were seen in nonmotile rut growth cones. Upward-directed arrows indicate ratios that were outside the range of they axis. Scale bars in A are 10 μm (left) and 7 μm(right). For C, WT (cells, cultures) 16, 13; dnc, 14, 11; rut, 19, 13.
DISCUSSION
Regional differences within cultured neurons
Here, we describe the first use of fura-2 imaging for intracellular Ca2+ in a dissociated culture system of Drosophila. Unlike in vivopreparations such as the neuromuscular junction (Karunanithi et al., 1997) or the adult CNS (Rosay et al., 2001; Wang et al., 2001), optical imaging in culture can relate subcellular Ca2+ dynamics throughout different regions of single neurons (Fig. 1). The spatial resolution offered by optical imaging complements electrophysiological studies ofDrosophila giant neurons in culture (Saito and Wu, 1991,1993; Zhao and Wu, 1997; Yao and Wu, 1999) and may indicate the functional significance of membrane excitability differences in subneuronal regions. The two approaches in combination will greatly enhance the neurogenetic study of Ca2+-dependent processes involved in neuronal development and physiology.
Our initial characterization of regional differences in high K+-induced Ca2+ regulation in the soma and growth cone (Figs. 1, 4) indicates that both Ca2+and Na+ channels are involved (Fig. 2).Drosophila central neurons in culture contain two types of Ca2+ currents (L and T type) distinguished by their activation voltage, decay kinetics, voltage dependence, and underlying single-channel activities (Byerly and Leung, 1988; Leung and Byerly, 1991; Saito and Wu, 1991, 1993). In other species, electrical recordings from the growth cone (for review, see Gottman and Lux, 1995) indicate that similar L- and T-type Ca2+channels are expressed throughout the neuron, but that channel density may be higher and the channels more clustered in the growth cone. Such variation may relate to our findings that the growth cone and soma differed in sensitivity and response kinetics during depolarization (Figs. 1, 4-7). In future investigations, it will be interesting to use identified mutations in Na+ and Ca2+ channel genes to dissect their involvement in Ca2+ regulation throughout the neuron.
Local Ca2+ levels regulate filopodial formation (Rehder and Kater, 1992; Lau et al., 1999) and play an important role in directing growth cone turning (Zheng, 2000) in cultured neurons from other species. Our data show that the magnitude of the high K+ response was larger in the periphery of motile, as opposed to nonmotile, growth cones. Distinct regions within the growth cone exhibit differences in the localization of cytoskeletal elements and cytoplasmic organelles (Forscher et al., 1987; Davenport et al., 1993; Azhderian et al., 1994). It may be questioned whether the cytokinesis inhibition technique used to generate giant neurons affects the distribution of Ca2+ channels in the soma and growth cone because of actin cytoskeletal disruption. However, previous electrophysiological studies did not detect differences in action potential and ion current properties with and without the removal of cytochalasin B (Saito and Wu, 1991). This result is consistent with our preliminary study indicating that differences in response characteristics between the soma and growth cone were still evident in embryonic cultures made in the absence of CCB (data not shown). In future studies, untreated larval neurons in culture can be used to examine Ca2+ signaling in the samednc and rut growth cones that display retarded motility (Kim and Wu, 1996).
The dnc and rutCa2+ regulatory phenotypes
Our analysis of these well studied mutants by fura-2 imaging revealed previously unknown mutant phenotypes and suggests the usefulness of Ca2+ imaging for a wide range of other mutations. We observed that dnc andrut decreased sensitivity to high K+ stimulation for a cytosolic Ca2+ increase (Figs. 5, 6) while affecting both the activity dependence (Figs. 7, 8) and spatial distribution of [Ca2+] within motile and nonmotile growth cones (Fig. 9). Chronic changes in cAMP metabolism imposed by the dnc and rut mutations decreased sensitivity most strongly in the growth cone while prolonging the Ca2+ increase only in the soma (Figs. 5,6).
It is known that dnc and rut alter the modulation of K+ currents gated by the Shand eag channel subunits (Zhong and Wu, 1993b; Zhao and Wu, 1997). Moreover, enhanced spike activity has been detected with patch-clamp recordings from the soma of dnc andrut giant neurons (Zhao and Wu, 1997). Such altered excitability may explain the prolonged soma response that we observed (Fig. 6) and should be further examined during patch-clamp experiments with high K+ depolarization. Electrophysiological studies of other currents in dnc andrut neurons have been lacking, but a Ba2+ current flowing through wild-type Ca2+ channels increased during the application of cAMP analogs (Alshuaib and Byerly, 1996). Further support for Ca2+ channel defects stems from findings that dnc and rut increase and decrease an L-type (dihydropyridine-sensitive) Ca2+ current in larval muscle, phenotypes that were mimicked by short-term pharmacological manipulations on wild-type muscles (Bhattacharya et al., 1999). However, somednc and rut physiological phenotypes at the larval neuromuscular junction cannot be mimicked by acute pharmacological treatments and are attributed to long-term, developmental effects of the mutations (Renger et al., 2000). Different aspects of the Ca2+ signaling phenotypes may be caused by either acute or chronic effects. Therefore, the combination of genetic and pharmacological analyses will offer a more comprehensive picture of how the cAMP pathway regulates neuronal Ca2+.
The effects of dnc and rut on Ca2+ dynamics may be mediated by misregulation of multiple mechanisms, including channel distribution, cytoskeletal organization, and intracellular Ca2+ stores. Region-specific differences in the cAMP pathway may lead to a higher incidence of responses containing a shoulder or second peak in the growth cone than soma (Fig.7A) and affect waveform modulation by previous activity (Fig. 7B). In other culture systems, the prolonged decay of intracellular Ca2+ triggered by high K+ depolarization has been used to implicate Ca2+ uptake into and release from intracellular stores (Friel and Tsien, 1992; Mironov et al., 1993; Babcock et al., 1997; Pivovarova et al., 1999). In this respect, Ca2+ imaging of the large collection of Drosophila learning and memory mutants (Davis, 1996; Dubnau and Tully, 1998) may complement studies in other organisms designed to assay the role of intracellular Ca2+ stores in neuronal plasticity (Cormier et al., 2000; Nishiyama et al., 2000; Emptage et al., 2001).
Expansion and retraction of the growth cone have been implicated in synapse formation at the larval neuromuscular junction ofDrosophila (Halpern et al., 1991; Keshishian et al., 1993; Chiba and Keshishian 1996; Yoshihara et al., 1997). Defects in lamellipodial expansion and retraction caused by dnc andrut in cultured neurons (Kim and Wu, 1996) suggest that this altered growth cone behavior may disrupt arborization of the motor axon terminal as was described for these mutants (Zhong et al., 1992). Here we characterized defects in the spatial distribution of Ca2+ signaling in dnc andrut growth cones (Fig. 9), suggesting that the interplay between Ca2+ and the cAMP pathway may be important for growth cone behaviors. In Xenopus cultured neurons, an interaction between the cAMP pathway and Ca2+ influences the growth cone response to an external guidance cue (Ming et al., 1997). As suggested by data shown in Figure 9, dnc and rut altered the regulation of Ca2+ with reference to growth cone motility. Drosophila culture systems and available mutations can be used to link cAMP to several Ca2+-dependent aspects of growth cone behavior described in other preparations (Kater et al., 1988; Kater and Mills, 1991; Rehder and Kater, 1992; Lau et al., 1999; Zheng, 2000).
Drosophila culture systems to study the genetics of Ca2+ regulation
Highest expression of the dnc and rutproducts is found in the mushroom bodies (Crittenden et al., 1998), whereas lower levels were observed elsewhere. However, using the GAL4-UAS system (Brand and Perrimon, 1993) to label a subset of mushroom body neurons with GFP, we showed that this restricted cell population displayed similar variability and response levels (Fig.5B, compare with Supplemental Fig. 2). Previous electrophysiological data on a similar population of mushroom body neurons in larval CNS cultures (Wright and Zhong, 1995) also exhibit considerable variation in the amplitude and kinetics of voltage-gated K+ currents. Significant mutant phenotypes may not be restricted to cells with high expression levels. In fact, physiological defects have been observed in neurons with lower expression levels such as the larval neuromuscular junction (Zhong and Wu, 1991). dnc and rut defects in neuronal firing patterns (Zhao and Wu, 1997) and growth cone motility (Kim and Wu, 1996) have been documented in the general population of neurons in culture (with and without CCB treatment). Therefore, the reporteddnc and rut phenotypes in our study may reflect defects common to different types of neurons.
Given the multiple downstream targets of the cAMP pathway that may vary in spatial or temporal distribution, the phenotypes of thednc and rut mutations are necessarily complex. Mutant phenotypes within subcellular compartments may indicate localized actions of these proteins. For example, the abundance of PKA subunits varies throughout the Drosophila CNS (Kalderon and Rubin 1988; Park et al., 2000), and the cellular expression of PKA function is controlled through binding to membrane-associated attachment proteins (for review, see Gray et al., 1998). Available mutations in Drosophila offer the unique opportunity to study how different isoforms of regulatory and catalytic subunits [encoded by PKA-RI andPKA-RII (Park et al., 2000) and DC0 (Kalderon and Rubin 1988)] PKA subunits as well as anchoring proteins (AKAP200) (Li et al., 1999; Rossi et al., 1999) contribute to subcellular Ca2+ regulation. Together with pharmacological manipulations of cAMP levels and PKA activity, these studies should yield a more complete picture of how PKA and other cAMP-related mechanisms produce the phenotypes that we have described.
Footnotes
We thank Dr. Wei-Dong Yao for his generous help during the initial exploration of this work. We also thank Shetarra Walker and Daniel Choi for their assistance, Peter Taft for technical suggestions, and Drs. Atsushi Ueda, Mei-Ling Joiner, and Sudipta Saraswati for comments on this manuscript.
Correspondence should be addressed to Dr. Chun-Fang Wu, Department of Biological Sciences, University of Iowa, Iowa City, IA 52242. E-mail:cfwu{at}blue.weeg.uiowa.edu.
