Abstract
Synapse formation is tightly associated with neuronal excitability. We found striking synaptic overgrowth caused by Drosophila K+-channel mutations of the seizure and slowpoke genes, encoding Erg and Ca2+-activated large-conductance (BK) channels, respectively. These mutants display two distinct patterns of “satellite” budding from larval motor terminus synaptic boutons. Double-mutant analysis indicates that BK and Erg K+ channels interact with separate sets of synaptic proteins to affect distinct growth steps. Post-synaptic L-type Ca2+ channels, Dmca1D, and PSD-95-like scaffold protein, Discs large, are required for satellite budding induced by slowpoke and seizure mutations. Pre-synaptic cacophony Ca2+ channels and the NCAM-like adhesion molecule, Fasciclin II, take part in a maturation step that is partially arrested by seizure mutations. Importantly, slowpoke and seizure satellites were both suppressed by rutabaga mutations that disrupt Ca2+/CaM-dependent adenylyl cyclase, demonstrating a convergence of K+ channels of different functional categories in regulation of excitability-dependent Ca2+ influx for triggering cAMP-mediated growth plasticity.
Introduction
Altered synaptic activity leads to changes in synaptic morphology, providing an effective means to adjust synaptic efficacy. For example, after long-term stimulation simulating the tonic activity patterns, the shape and ultrastructure of pre-synaptic terminals of phasic motoneurons at crayfish neuromuscular junctions (NMJs) are transformed into those resembling tonic terminals (Lnenicka et al., 1986). In addition, prolonged stimulation of lobster motor axons is correlated with pre-synaptic ultrastructural modifications, including increased numbers of active zones (Chiang and Govind, 1986). In parallel, long-term potentiation (LTP) induced by high-frequency nerve stimulation in the mammalian hippocampus can trigger dendritic spine formation in tens of minutes (Engert and Bonhoeffer, 1999).
In line with this activity-dependent synaptic growth, increased excitability from dysfunction of 4-aminopyridine-sensitive, voltage-activated K+ channels (Kv1 and Kv4) leads to aberrant synaptic growth. In developing Xenopus retinal ganglion cells, 4-aminopyridine blockade of Kv channels alters branch outgrowth (McFarlane and Pollock, 2000). Similarly, mutations of Kv1 in Caenorhabditis elegans induce abnormal sensory axon branching (Peckol et al., 1999). At Drosophila larval NMJs, Shaker (Sh) mutations affecting Kv1 channels lead to synaptic overgrowth when combined with ether a go-go (eag) K+-channel mutations (Budnik et al., 1990) or after modest increase in rearing temperature (Zhong and Wu, 2004). However, beyond 4-aminopyridine-sensitive channels including Sh, little is known about the roles of other K+ channels in activity-dependent synaptic growth.
Here we report striking synaptic overgrowth indicated by abundant small “satellite” boutons in slowpoke (slo) (Elkins et al., 1986; Komatsu et al., 1990; Atkinson et al., 1991) and seizure (sei) (Elkins and Ganetzky, 1990; Titus et al., 1997; Wang et al., 1997) mutants, defective in Ca2+-activated large conductance (BK) and voltage-dependent Erg channels, respectively. Slo (BK) K+ channels, known to colocalize with Ca2+ channels in excitable cells, provide negative feedback onto Ca2+-influx-regulated events, including neurotransmitter release (Salkoff et al., 2006). Sei (Erg) K+ channels in vertebrates affect cardiac action potential repolarization (Sanguinetti et al., 1995) and regulate neuronal firing frequency adaptation (Sacco et al., 2003).
Importantly, disruption of BK and Erg channels led to abundance of two types of satellites distinct in their morphology, suggesting that they mirror structures differentiated from distinct synaptic growth intermediates trapped by preferential effects of these mutations. These growth steps consisting of initial satellite budding followed by maturation into boutons have been proposed previously based on time-lapse studies of intact larvae (Zito et al., 1999). Recent live imaging of larval NMJs also demonstrated dynamic growth and molecular differentiation of smaller boutons after intense stimulation or high K+-induced depolarization (Ataman et al., 2008). Our double-mutant analysis further revealed functional associations of BK and Erg channels with distinct pre- and post-synaptic molecules and, more importantly, with Ca2+-influx-induced cAMP signaling in proposed growth and differentiation steps. Together with Sh-induced synaptic overgrowth (Zhong et al., 1992; Zhong and Wu, 2004), our data thus indicate that the cAMP signaling pathway acts as a common mediator in excitability-regulated synaptic growth and that dysfunctions of K+ channels of different properties and distributions can result in modifications of distinct aspects of NMJ growth.
Materials and Methods
Fly stocks.
The fly stocks used in this study include the following: wild-type (WT) Canton-S and K+-channel mutants, slowpoke (slo1, slo98, slo1/slo4, and slo98/slo4; slo4 provided by Dr. N. Atkinson, University of Arizona, Tucson, AZ) and seizure (seits2 and seits2/seits1). The following single mutants and their corresponding double-mutant combinations with slo and sei were examined in this study: fasciclin II (fasII; fasIIe76 and fasIIe76/fasIIeB112; Dr. V. Budnik, University of Massachusetts Medical School, Worcester, MA), fasII;;slo (fasIIe76;;slo1), fasII;sei (fasIIe76; seits2), discs large (dlg; dlgX1–2/Y or dlgX1–2/Df(1)N71, both of which lack all three PDZ domains; Dr. V. Budnik), dlg;;slo (dlgX1–2/Y;;slo1), and dlg;sei (dlgX1–2/Y;seits2); Ca2+-channel mutants cacophony (cac; cacs and cacNT27/cacs; Dr. R. Ordway, Pennsylvania State University, University Park, PA), cac;;slo (cacs;;slo1, cacs;;slo98, and cacs;;slo4), cac;sei (cacs;seits2), Dmca1D (Dmca1DAR66; Dr. D. Eberl, University of Iowa, Iowa City, IA), Dmca1D;slo (Dmca1DAR66;slo1 and Dmca1DAR66;slo98), and Dmca1D sei (Dmca1DAR66 seits2; three independent recombinant lines); Na+-channel mutants malelessnap (napts), nap;slo (napts;slo1 and napts;slo98), and nap sei (napts seits2; three independent recombinant lines); cAMP mutants and transgenic constructs rutabaga (rut; rut1, and rut1084), rut;;slo (rut1;;slo1), rut;sei (rut1;seits2), dunce (dnc; dncM14), dnc;;slo (dncM14;;slo1), dnc;sei (dncM14;seits2), UAS-dunce+ (UAS-dnc+; Dr. H. Atwood, University of Toronto, Toronto, Ontario, Canada), UAS-dnc+/Y;C164-GAL4/+;slo1, UAS-dnc+/+;mef2-GAL4/+;slo1/slo98, and UAS-dnc+/C155-GAL4;seits2. Pan-neuronal and motoneuron-specific drivers C155-GAL4 (Bloomington Drosophila Stock Center, Indiana University, Bloomington, IN) and C164-GAL4 (Dr. J. Wang, University of California, La Jolla, CA), respectively, were used for expression of a transgene construct, UAS-dnc+, in pre-synaptic neurons. For expression of UAS-dnc+ in post-synaptic muscle, mef2-GAL4 (Bloomington Drosophila Stock Center) was used in the slo mutant background. C155-GAL4 was also used to drive pre-synaptic expression of a mutated sei transgene (UAS-seits2; a generous gift from Drs. J. Simpson, Janelia Farm Research Campus, Ashburn, VA and B. Ganetzky at University of Wisconsin–Madison, Madison, WI). UAS-RNAi transgenes specifically targeting slo and sei (UAS-slo-RNAi, #6723 and #104421; UAS-sei-RNAi, #104698 and #3606; Vienna Drosophila RNAi Center) were used in combinations with multiple neuronal and muscular GAL4 drivers (C155 and C164 for neuronal drivers; Mhc, C179, and mef2 for muscular drivers) to reproduce mutant phenotypes in the WT background. Whenever possible, we used multiple alleles or independent isolates of each gene in phenotypic characterization to minimize the contributions from genetic variants of unidentified loci. Our results indicate overall similarities between multiple alleles in the numbers of mature boutons and satellites (supplemental Table S1, available at www.jneurosci.org as supplemental material).
All stocks were raised on conventional fly medium and maintained at room temperature (RT). In experiments for long-term high-temperature (HT) rearing, adult flies were kept at RT in fresh food vials for egg laying for 1 d, which were subsequently placed in a 29°C incubator for additional 4 to 5 d to cover entire larval development. For a short-term HT treatment, vials were kept at RT for 5–7 d until the majority of the larval population reached the third-instar stage and were then placed in a 29°C incubator for 5 h before dissection.
Immunohistochemistry.
Larvae were prepared as described previously (Lee et al., 2008). The primary antibodies used include monoclonal Fasciclin II [FasII; 1D4; 1:4; Developmental Studies Hybridoma Bank (DSHB)], Bruchpilot (Brp; NC82; 1:250; DSHB), and α-tubulin antibodies (clone 1A; 1:400; Sigma-Aldrich), and polyclonal Discs large (Dlg; 1:20,000; Dr. V. Budnik), glutamate receptor subunit IIC (DGIuRIIC; 1:1000; Dr. A. DiAntonio, Washington University, St. Louis, MO), Nervous wreck (Nwk; 1:2000; Dr. B. Ganetzky), and dynamin-associated protein, 160 kDa (Dap160; 1:200; Dr. H. Bellen, Baylor College of Medicine, Houston, TX). The pre-synaptic terminals were visualized by FITC- or Texas Red-conjugated horseradish peroxidase (HRP) antibody (1:50; Jackson ImmunoResearch).
Imaging and analysis.
Images of NMJs on muscle 4 from abdominal segments 3 through 7 were collected using an M1024 confocal microscope (Bio-Rad). The collapsed Z-stacked images of type Ib NMJs recognized by anti-HRP signals were compiled by ImageJ software (NIH Image) and analyzed for the numbers of primary mature boutons, satellites, and terminal branches. A satellite was defined as a bouton of smaller size than adjacent mature boutons deviating from the main trajectory of terminal branches. Two types of satellites were classified based on the following criteria: “type B” satellites resemble yeast budding without a clear constriction, whereas “type M” satellites display a neck-like constriction separating mature boutons and satellites. A branch was defined as a projection that contains two or more boutons in series separated by clear interbouton regions. A branch segment was indicated as a region between two adjacent branching points, and ascending orders were given to the next level of branch segments at each branching point. The arbitrary branch index reflecting the complexity of branching patterns was calculated based on the following formula: branch index = 1 × (the number of first-order branch segments) + 2 × (the number of second-order branch segments) + 3 × (the number of third-order branch segments) + 4 × (the number of branch segments of an order higher than the third).
For comparison of distributions of FasII and Dlg between type B and M satellites, each NMJ was manually traced for FasII and Dlg immunoreactivity with the pixel intensities measured from each satellite and four to five adjacent mature boutons within the same NMJ. The relative intensity ratios (percentage intensity from satellites/mature boutons) were then compared between the entire populations of type B and M satellites for slo and sei mutants.
Electron microscopy.
Larvae dissected in hemolymph-like (HL) 3.1 saline were superfused for 15 min with fixative (0.1 m sodium cacodylate buffer, 2.5% glutaraldehyde, and 2.5% formaldehyde, pH 7.2) and then stored in a vial for 2 h at RT. After fixation, the tissue was rinsed in the buffer solution followed by postfixation in buffered 1% osmium tetroxide for 1 h and a 15 min rinse in buffer. Subsequent dehydration was performed in 30–100% ethanol series followed by clearing in propylene oxide and embedding in plastic (Polybed 812; Polysciences). The plastic block cured in a 60°C oven for 1–2 d was sectioned, placed on copper grids, and double stained with uranyl acetate and lead citrate. NMJ electron micrographs were taken at a magnification of 10,000 to 20,000.
Statistical analysis.
For comparison between two genotypes, two-tailed t test was routinely performed against the null hypothesis of equal values between two genotypes, unless indicated otherwise. A p value <0.05 was considered significantly different. All statistical analyses were performed using Origin software (version 6.0; OriginLab).
Results
Aberrant synaptic growth patterns in slowpoke and seizure K+-channel mutants
The stereotypic innervation patterns of Drosophila larval NMJs, with identifiable motoneurons and muscle fibers, facilitate detection of morphological modifications and have enabled quantitative studies of the regulatory mechanisms underlying activity- and excitability-dependent synaptic growth (Budnik et al., 1990; Sigrist et al., 2003). In this study, we investigated striking NMJ phenotypes in slo and sei mutants (Fig. 1), which displayed abundant small boutons, termed “satellites,” budding from the larger primary boutons along the branch axis (Fig. 1B, arrows and arrowheads). Such aberrant outgrowth was found at both type Ib and Is NMJs (Johansen et al., 1989; Atwood et al., 1993) in different muscles, e.g., 6/7 (Fig. 1A) and 4 (Fig. 1B, type Ib only). To facilitate quantitative analysis, we focused on type Ib NMJs in muscle 4, which displayed a stereotypic branching pattern with fewer boutons of larger sizes. During closer examination, two distinct types of satellites were observed, one without a clear constriction between satellites and primary boutons, resembling yeast budding (Fig. 1B, arrowheads) (hereafter, “type B satellite”) and the other with a short but clear constriction or “neck” (Fig. 1B, arrows) (hereafter, “type M satellite”). Our results indicated more abundant type B satellites in slo mutants in contrast to a majority of type M satellites in sei mutants (Fig. 1C). These alterations were accompanied by increases in the synaptic bouton number and terminal branching for both mutants (Fig. 1D). Both sei and slo mutants were first identified as temperature-sensitive mutants because of more evident behavioral manifestations at HT (Elkins et al., 1986; Elkins and Ganetzky, 1990). However, our experiments performed at RT revealed abundant satellites in both mutants, demonstrating the severe effects of dysfunctions of the Erg and BK channels, separately encoded by the two genes.
The satellites abundant in slo and sei mutants are morphologically similar to primordial boutons detected during normal development, which mature into primary boutons and form new branches (Zito et al., 1999). Such similarity raised the possibility that satellites may represent arrested structural intermediates accumulated during synaptic growth that are rarely seen in mature third-instar WT larvae (Fig. 1C). We proposed a working hypothesis in an attempt to link these putative structural intermediates in a sequential growth process. Similar to the scheme proposed for normal bouton development (Zito et al., 1999), we assumed that type B satellites are derived from initial budding, whereas type M satellites reflect derivatives from intermediates of a more mature state that further differentiate and transform into full-grown boutons (Fig. 1E).
Given that type B satellites recapitulate the initial budding process, two possibilities may be invoked to explain their abundance in slo mutants. First, Slo channel activity may promote the formation of type M satellites (Fig. 1E, open arrow, step b). A retarded transformation in slo mutants should lead to accumulation of type B satellites, but with a fewer number of type M satellites and mature boutons. This scenario contradicts our observations of a significant fraction of type M satellites together with more mature boutons and terminal branches in slo (Fig. 1C,D). Conversely, Slo channels may normally serve as a restraining factor, preventing excessive budding from mature boutons (Fig. 1E, closed bar, step a). Thus, slo mutations may unleash the growth, leading to accumulation of large proportions of type B satellites. This would also enhance conversion of type B satellites into type M satellites or even mature boutons at a limited rate, which appears to be consistent with the observed slo phenotypes (Fig. 1C,D).
In contrast, sei mutants were characterized by more abundant type M than B satellites, along with significantly enhanced bouton and branch formation (Fig. 1C,D for seits2) (to a lesser extent in seits2/seits1 mutants; data not shown). Abundant type B satellites in sei mutants comparable to slo suggest a similarly unleashed growth in the earlier step (Fig. 1E, step a), indicating a restraining action of Sei as well as Slo K+ channels. However, the excessive accumulation of type M satellites implicates an additional action of Sei channels in later steps. First, a restraining action on transformation of type B into M satellites may be considered (Fig. 1E, closed bar, step b). In this case, mutations in Sei channels would result in more type M satellites and mature boutons with depletion of type B satellites to some extent. However, this prediction contradicts the observed sei phenotype with abundant type B satellites comparable to slo (Fig. 1C). Our result thus favors an alternative possibility of Sei channels in facilitating maturation of type M satellites into mature boutons and terminal branches (Fig. 1E, open arrow, step c). Thus, mutations of Sei channels may partially stunt the development and lead to excessive accumulation of type M satellites, along with enhanced, although less extreme, overgrowth of type B satellites, consistent with the observed sei phenotypes (Fig. 1). Notably, sei mutants displayed significantly enhanced branching patterns along with robust increase in mature bouton numbers (Fig. 1D), indicating that sei mutations do not fully prevent type M satellites from entering the successive growth steps. These hypotheses were further examined with additional manipulations, such as HT treatments and double-mutant combinations (see below).
We then examined sei;slo double mutants to further explore the differential effects of sei and slo in the proposed stepwise growth processes. When compared with single mutants, the combined mutational effects of the two genes in sei;slo were not simply additive, with a further enhancement of some slo, but not sei, phenotypes. Apparently, satellite abundance was less extreme in sei;slo double mutants compared to sei mutants. There was a slight, but not significant, reduction in the numbers of both types B and M satellites [p > 0.10, mean ± SEM (n), sei;slo vs sei, 5.25 ± 0.86 (20) vs 7.63 ± 0.80 (38) for type B; 7.42 ± 1.33 (20) vs 10.58 ± 1.02 (38) for type M]. In contrast, the numbers of type M (but not type B) satellites and mature boutons were significantly greater in sei;slo compared to slo [mean ± SEM (n), sei;slo vs slo, 7.42 ± 1.33 (20) vs 4.09 ± 0.32 (107), p < 0.01, for type M; 5.25 ± 0.86 (20) vs 7.18 ± 0.37 (107), p > 0.10, for type B satellites; 28.70 ± 1.84 vs 24.67 ± 0.80, p < 0.05, for mature boutons]. The enhanced overgrowth of these two structures in double mutants beyond the levels in slo is consistent with the idea that Sei K+ channels exert influences on the later steps of the synaptic bouton growth processes (Fig. 1E). The lack of further overgrowth in sei;slo beyond the levels in sei suggests the existence of a saturable capacity, or an optimal activity level for promotion of the excitability-associated synaptic growth examined here. Possible explanations include potential overlapping actions of Sei and Slo K+ channels onto activity-dependent, common downstream effectors such as Ca2+ and cAMP (see below).
Dynamic growth process revealed by high rearing temperature in slo and sei mutants
HT treatment on slo and sei mutants revealed the time scales in the dynamic growth process described above. Although 2 h exposures to HT (29°C) failed to induce significant morphological changes (data not shown), both types of satellites in these mutants were apparently suppressed by longer incubation (5 h) (see Materials and Methods) (Fig. 2A). Whereas the frequency of type B, rather than type M, satellites was drastically reduced in slo mutants, type M satellites were reduced to a greater extent in sei mutants (Fig. 2B), in line with the relative abundance of each type of satellites in these two mutants (Fig. 1C) and the proposed major actions of Slo and Sei channels at RT (Fig. 1E). Importantly, HT treatment for 5 h further increased the number of branches in slo mutants (Fig. 2C). Unlike slo and sei mutants, WT failed to respond to the same 5 h HT treatment (Fig. 2), consistent with a previous report (Sigrist et al., 2003). These results thus indicate that these mutations provided a sensitized condition to expose the temperature dependency of individual growth steps (Fig. 1E).
In parallel, chronic HT treatment (5 d, sufficient to cover the entire larval development) induced significant increases in both bouton and branch numbers in WT, consistent with previous reports (Sigrist et al., 2003; Zhong and Wu, 2004), despite no significant changes in the rare occurrence of satellites (Fig. 2). Unlike WT, chronic HT-treated slo mutants displayed significantly reduced frequency of both types of satellites (Fig. 2B) along with increased mature bouton and branch numbers (Fig. 2C). These results thus corroborate the above short-term HT effects, suggesting facilitated transformation of slo-induced satellites into primary boutons and terminal branches after raising the temperature.
In chronic HT-treated sei mutants, we observed a unusual population of terminal branches consisting of thin strings of type M (and occasionally type B) satellites that were absent in HT-treated WT or slo larvae (Fig. 2A, asterisks, B, hatched bars), contributing to enhanced branching complexity (Fig. 2C, hatched bar). This observation is in line with a specific role of Sei channels in later growth steps (compare Fig. 1E, step c), but we also detected a contrasting decrease in primary bouton number in chronic HT-treated sei larvae (Fig. 2C). How the further enhanced branching took place in the expense of bouton formation remains unclear, awaiting further investigation.
Together, these acute and chronic HT-induced modifications indicate that elevated temperature may promote synaptic restructuring or growth, providing a force to overcome barriers at specific steps caused by dysfunction of Sei and Slo K+ channels.
Dominant mutational effects of sei and slo on synaptic growth
Sei Erg and Slo BK channels, like many other K+ channels, are thought to be composed of four identical subunits (Hille, 2001). Thus, a mixture of mutant and wild-type subunits in a multimeric assembly in heterozygous mutants could lead to a dominant-negative effect, altering the function of heteromeric channels that contain one or more mutant subunit and sparing only a minority of channels that consist strictly of WT subunits. For example, such dominant effect on HT-induced seizure and paralytic behavior was evident in heterozygous seits2/+ flies carrying a point mutation in the pore domain of the channel (Titus et al., 1997; Wang et al., 1997), comparable to those observed in the homozygote (seits2/ seits2) (Jackson et al., 1985). Therefore, we have examined the morphological phenotypes of seits2/+ and slo/+ in comparison with seits2/ seits2 and slo/slo, respectively, at larval NMJs.
Importantly, we observed dominant effects on satellite and bouton formation in seits2/+as well as slo/+ heterozygotes (supplemental Table S2, available at www.jneurosci.org as supplemental material). Larvae heterozygous for two alleles of slo (slo1/+ and slo98/+) showed increases in all of the parameters we examined, including the numbers of both types of satellites, mature boutons, and branch segments, to levels comparable to the homozygous counterpart (slo/slo). However, larvae heterozygous for a null allele, slo4/+ (Atkinson et al., 1991), did not display significant enhancement in these growth parameters except for bouton formation (supplemental Table S2, available at www.jneurosci.org as supplemental material). Such contrasting phenotypes between null (slo4/+) and non-null alleles (slo1/+ and slo98/+) of slo in the heterozygote are consistent with the idea that dominant-negative effects can result from the incorporation of mutated subunits in the combinatorial assembly of a multimeric channel and indicate a high sensitivity of these growth processes to dysfunction of Slo K+ channels. Considering a tetrameric subunit assembly of Slo channels, the fraction of intact channels strictly composed of WT subunits in slo1/+ or slo98/+ larvae could be significantly lower than that expected (50%) in slo4/+, consistent with their relative phenotypic severity (supplemental Table S2, available at www.jneurosci.org as supplemental material). Similar interference by mutated Sh subunits beyond the simple gene-dosage effects of null mutations has been documented previously in voltage-clamp measurements for Sh K+ current amplitude in Drosophila larval muscle fibers (Haugland and Wu, 1990).
In the case of seits2/+ larvae, we observed a significant increase in the number of mature boutons and type B satellites, but the number of type M satellites and branch segments remained comparable to WT (supplemental Table S2, available at www.jneurosci.org as supplemental material). These results indicate differential degrees of dominant mutational effects on each growth step, i.e., the initial budding step with a higher sensitivity to Sei channel dysfunction versus the later growth steps such as branch formation more resistant to the same degrees of dysfunction in these heterozygous larvae.
Synaptic ultrastructure and protein distribution in slo and sei satellites
We then examined the potential structural distinctions between two types of satellites, indicative of regulatory mechanisms differentially affected by slo and sei mutations. Both types of satellites contained synaptic vesicles, electron-dense synaptic area and T-bars, and subsynaptic reticulum, a specialized folding of post-synaptic muscle membranes (Jan and Jan, 1976) (Fig. 3A), similar to normal growing boutons (Zito et al., 1999) and mature synapses (Atwood et al., 1993). The presence of active zones in these satellites was also detected when an antibody (NC82) against Bruchpilot, a protein important for integrity of T-bars (Wagh et al., 2006), was used for visualizing active zones at a light-microscopic level (Fig. 3C). In addition to the similarity in ultrastructure, immunostaining demonstrated post-synaptic association of glutamate receptor clusters containing DGluRIIC subunits with both types of satellites (supplemental Fig. S1, top, available at www.jneurosci.org as supplemental material).
Aside from these similarities, distinct microtubule distributions between the two types of satellites were indicated by α-tubulin immunostaining (Fig. 3B), in contrast to comparable phalloidin-labeled actin networks (data not shown). Although there was no detectable microtubule structure in type B satellites (Fig. 3B, arrowheads), some type M satellites displayed a loop-like microtubule arrangement (Fig. 3B, arrows) that is known to be associated with structurally stable, mature synaptic boutons (Roos and Kelly, 1998). When the data from multiple satellites were compared for the presence of filamentous or loop-like tubulin structures, we were unable to detect any type B satellites with these structures in both slo (0 of 60) and sei (0 of 28) mutants. In contrast, such features were evident in subsets of type M satellites in both mutants, with a higher fraction in sei than slo (15 of 59 for sei vs 4 of 54 for slo). This is consistent with our model that type M satellites, abundant in sei mutants, mirror more advanced structural intermediates than type B satellites (Fig. 1E).
In previous studies, satellites are also frequently observed in mutants defective in Nwk and Dap160, which affect actin dynamics and membrane recycling (Coyle et al., 2004; Koh et al., 2004; Marie et al., 2004), as well as in other endocytic mutants (Dickman et al., 2006). We thus investigated potential alterations in their distributions in slo and sei mutants. However, immunoreactivities against Nwk and Dap160 were clearly detected in both types of satellites in mutants, similar to those in adjacent primary boutons (supplemental Fig. S1, available at www.jneurosci.org as supplemental material), suggesting that satellite formation induced by these K+-channel mutations is not likely attributable to alterations in actin cytoskeleton and membrane recycling caused by gross differences in the levels of Nwk and Dap160. Whether slo and sei mutations alter fine-tuning of these cellular processes during development awaits further investigation.
It is interesting to note that a previous live-imaging study has documented the sighting of rapid growth of synaptic terminals at larval NMJ within minutes after either high-frequency electrical stimulation or high K+-induced depolarization (Ataman et al., 2008). However, these newly formed boutons of a smaller size, referred to as “ghost boutons,” were not equipped with pre- and post-synaptic markers, indicating a lack of synaptic differentiation. Accumulation of putative pre-synaptic active zones markers and post-synaptic glutamate receptor clusters was evident only after hours to days, indicating slower differentiation in the maturation and stabilization processes. Significantly, such a slow differentiation time course is consistent with HT-induced modulation we observed over the span of hours and days (Fig. 2).
Additional exploration revealed abundant satellites well differentiated with pre- and post-synaptic markers, including Dlg and FasII in addition to DGluRIIC, contrasting with “ghost boutons” in their youth (see below). Thus, slo and sei satellites may reflect the trapped transient growth steps undergoing subsequent molecular and ultrastructural differentiation. The abundance of these satellites may be derived from stabilization and preservation of growth intermediates described in time-lapse imaging of developing NMJs in intact larvae (Zito et al., 1999).
Differential contributions of FasII and Dlg to satellite formation in slo and sei mutants
In addition to microtubule networks described above, two types of satellites were further distinguished by immunostaining of a scaffold protein, Dlg (Lahey et al., 1994) and cell-adhesion molecule FasII (Bastiani et al., 1987), two synaptic proteins interacting during synaptic growth in Drosophila embryos (Kohsaka et al., 2007). Within the same NMJ preparation, heavy decoration of Dlg immunoreactivity was observed in type B satellites as well as adjacent mature boutons (Fig. 4A, arrowheads), whereas it was significantly reduced around type M satellites in both slo and sei mutants (Fig. 4A, arrows; supplemental Fig. S2, available at www.jneurosci.org as supplemental material) (p < 0.001 between type B and M satellites). In contrast, the level of FasII immunofluorescence was significantly higher for both types of satellites compared to that in mature boutons (supplemental Fig. S2, available at www.jneurosci.org as supplemental material), which may be attributable in part to a larger surface-to-volume ratio of satellites. FasII levels were comparable between two types of satellites despite a small, but statistically significant, difference in immunoreactivity of type M satellites at sei NMJs (supplemental Fig. S2, available at www.jneurosci.org as supplemental material).
These results demonstrate differential distribution of Dlg, but not FasII, between type B and M satellites, and thus suggest potential differences in Dlg- and FasII-dependent regulation in the sequential steps of synaptic growth. We examined double-mutant combinations and found striking differential effects of fasII mutations on slo and sei satellites. Significantly, satellite frequency in fasII;;slo remained comparable to slo, whereas satellites in fasII;sei were significantly reduced from the sei level, more pronouncedly for type M than type B satellites (Fig. 4C). This is in contrast to the observation of similar distribution profiles of FasII immunoreactivity in both slo and sei single mutants. Such differential suppression of sei phenotypes by fasII was also reflected in the frequency of mature boutons (supplemental Table S1, available at www.jneurosci.org as supplemental material) and complexity of terminal branches (Fig. 4B, branch index) (see Materials and Methods) when fasII;;slo and fasII;sei were compared. The bouton number was slightly reduced in fasII;;slo double mutants, but a similar reduction was seen in fasII (compared to WT, p < 0.05) (cf. Schuster et al., 1996). However, far more reduction was observed in fasII;sei (supplemental Table S1, available at www.jneurosci.org as supplemental material). This preferential modification of sei rather than slo phenotypes by fasII, i.e., significant suppression of type M satellite formation and its maturation, suggests a tight interaction between FasII and Sei channels at the later steps in the growth process (Fig. 1E, step c). Additionally, significantly reduction of type B satellites in fasII;sei double mutants indicates that interaction of FasII and Erg channels occur in an earlier growth step as well (Fig. 1E, step a).
In contrast to fasII, a dlg mutation reduced not only type B, but also type M satellites in both dlg;;slo and dlg;sei mutants close to the WT level (Fig. 4C). Furthermore, such reduced satellite frequency in double mutants was correlated with fewer mature boutons (supplemental Table S1, available at www.jneurosci.org as supplemental material) (p < 0.001 for slo vs dlg;;slo and sei vs dlg;sei) and with simpler branching patterns indicated by the reduced branching index when compared to slo and sei single mutants (Fig. 4B). These results thus imply that Dlg may be involved in the initial step of satellite formation induced by both sei and slo mutations (Fig. 1E, step a). Interfering with initial formation of type B satellites, the dlg mutation could prevent their transformation into type M satellites and mature boutons, consistent with our results (Fig. 4; supplemental Table S1, available at www.jneurosci.org as supplemental material).
Along with significant suppression of satellites by dlg mutations (Fig. 4C), heavier decoration of Dlg in type B than in type M satellites in both slo and sei NMJs (Fig. 4A; supplemental Fig. S2, available at www.jneurosci.org as supplemental material) also raised a possibility that a manipulation of the Dlg level may differentially modify type B and M satellites. This idea was tested in slo mutants, which display higher frequencies of type B than type M satellites (Fig. 1C), by overexpressing WT dlg transgene tagged with EGFP (UAS-dlg+-EGFP). Since the majority of immunoreactivity against endogenous Dlg is detected from the post-synaptic muscles surrounding nerve terminals at NMJs, we expressed this transgene using a muscle-specific GAL4 driver, mef2 (Gossett et al., 1989; Lilly et al., 1994). Indeed, the expression profiles of UAS-dlg+ recognized by EGFP closely resembled those observed for endogenous Dlg in WT larvae (data now shown). Notably, in contrast to the severe suppression of both type B and M satellites in dlg;;slo (Fig. 4C), overexpression of dlg+ induced a more pronounced reduction in type B satellites (supplemental Fig. S3, available at www.jneurosci.org as supplemental material) (slo vs UAS-dlg+(in muscle);slo; p < 0.001 for type B vs p > 0.05 for type M). As a result, the numbers of remaining type B and M satellites became similar, and there were fewer satellites in total (supplemental Fig. S3, available at www.jneurosci.org as supplemental material). More importantly, in contrast to a reduction observed in dlg;;slo compared to slo, the number of mature boutons was significantly enhanced in slo larvae expressing post-synaptic dlg+ (supplemental Fig. S3, available at www.jneurosci.org as supplemental material). Selective decrease in type B satellite frequency coupled with enhanced bouton formation raises a possibility that excessive Dlg expressed in the post-synaptic muscle may provide a platform to promote conversion of type B satellites to type M, which can mature into boutons. This represents an interesting parallel to the modulation induced by HT treatment in slo mutants (compare Fig. 2B,C, with supplemental Fig. S3, available at www.jneurosci.org as supplemental material), demonstrating the similar outcomes for slo-induced satellite formation caused by two different treatments that facilitate synaptic growth.
Differential contributions of pre- and post-synaptic Ca2+ channels to satellite formation in slo and sei mutants
Dlg and its vertebrate homologs, including PSD-95, play important roles in scaffolding various post-synaptic proteins, such as FasII, glutamate receptor, and Sh K+ channels (Lahey et al., 1994; Kim and Sheng, 2004), whereas pre-synaptic regulation of FasII is particularly important for growth plasticity at Drosophila larval NMJs (Schuster et al., 1996). Significantly, activity-dependent accumulation of intracellular Ca2+ has been implicated in modulation of both PSD-95 and FasII (Schuster et al., 1996; Kim and Sheng, 2004). The differential dlg- and fasII-dependent modulation (Fig. 4) thus may reflect distinct Ca2+ signaling processes, including Ca2+ channels, in regulation of slo and sei satellites.
Two genes, Dmca1D and cacophony, have been identified to encode voltage-activated Ca2+ channels in Drosophila. The Dmca1D channels, sharing homology to vertebrate L-type Ca2+ channels, are thought to be the predominant Ca2+ channels in post-synaptic muscles at NMJs (Zheng et al., 1995; Ren et al., 1998). In contrast, the Cac channels serve similar functions as vertebrate N-type channels, playing a major role in pre-synaptic neurotransmitter release (Smith et al., 1996; Kawasaki et al., 2000, 2002, 2004). Our double-mutant analysis indicated that separate pre- and post-synaptic actions of Cac and Dmca1D channels, respectively, exert differential influences on slo and sei satellites (Fig. 5). Since null mutations of these two Ca2+ channels are lethal, we used viable hypomorphic alleles to construct double mutants with slo and sei (see Materials and Methods). Nevertheless, the results showed clear suppressive effects on satellite formation in double-mutant combinations.
Significantly, the Dmca1D mutation suppressed both type B and M satellites in Dmca1D;slo and Dmca1D sei double mutants nearly to the WT level. This parallels with dlg-induced suppression of slo and sei satellites (compare Figs. 4B and 5B), suggesting a strong post-synaptic influence on the initial budding step (Fig. 1E, step a). Such suppression in satellite formation led to the morphology of Dmca1D;slo resembling that of Dmca1D single mutants (supplemental Table S1, available at www.jneurosci.org as supplemental material).
In contrast to Dmca1D, cac-induced suppression of satellite frequency was more drastic in sei than slo mutants, mirroring the fasII action (compare Figs. 4 and 5). Although cac mutations by themselves [cacs (Fig. 5); cacNT27/cacs (data not shown)] caused no significant changes in the morphological parameters examined, except for an increase in type M satellites, cac;sei (but not cac;;slo) double mutants displayed drastic decreases in both type B and M satellites (Fig. 5B), along with even more pronounced reduction in the number of mature boutons and terminal branches (supplemental Table S1, supplemental Fig. S4A, available at www.jneurosci.org as supplemental material). These results thus suggest a tight functional relation of Sei, rather than Slo, K+ channels with Cac pre-synaptic Ca2+ channels in regulation of Ca2+ influx for initial satellite formation as well as maturation (Fig. 1E, steps a–c).
Such tight association between Sei K+ channels and pre-synaptic regulators of membrane excitability was further supported by the double-mutant phenotypes when the napts (malelessnap) mutation was combined with either sei or slo. The nap mutation down-regulates the paralytic (para) gene encoding voltage-activated Na+ channels in Drosophila, weakening neuronal excitability and causing paralysis at high temperature (Wu et al., 1978; Kernan et al., 1991). Our results demonstrated suppression in satellite growth, in particular the type M, in nap sei double mutants (Fig. 5B). In contrast, the number of both types of satellites remained unchanged in nap;slo double mutants compared to slo single mutants (Fig. 5B). Since Drosophila muscle is devoid of Na+ channels (Singh and Wu, 1999), reduced excitability by nap presumably affects only pre-synaptic growth regulation associated with Sei- but not Slo-associated post-synaptic mechanisms.
Role of cAMP in pre- and post-synaptic regulation of satellite formation
The preferential effects of nap, cac, and Dmca1D mutations in suppressing sei and slo satellites prompted us to examine the role of rutabaga adenylyl cyclase (AC) that produces cAMP in a Ca2+/CaM-dependent manner (Livingstone et al., 1984; Levin et al., 1992). As reported previously, rut suppresses synaptic overgrowth induced by eag Sh hyperexcitability and HT treatments (Zhong et al., 1992; Zhong and Wu, 2004), but itself does not induce significant changes in the synaptic growth parameters, including the rare occurrence of satellites (Fig. 6; supplemental Table S1, available at www.jneurosci.org as supplemental material). Then, we examined the effects of rut mutations on slo and sei satellites using double-mutant combinations.
Despite their abundance in slo single mutants, both type B and M satellites were suppressed in rut;;slo double mutants to the WT level, along with a slight reduction in mature bouton number (Fig. 6; supplemental Table S1, available at www.jneurosci.org as supplemental material), paralleling the effects of dlg and Dmca1D mutations on satellite formation. In contrast to rut;;slo showing equal suppression, rut;sei double mutants displayed selective suppression of type M, but not type B, satellites (Fig. 6B), along with drastically reduced mature bouton (supplemental Table S1, available at www.jneurosci.org as supplemental material) and branch formation (supplemental Fig. S4B, available at www.jneurosci.org as supplemental material).
In parallel, we also examined the effects of mutations in dunce (dnc), a gene encoding a cAMP-specific phosphodiesterase (PDE) (Byers et al., 1981). Loss of PDE actions in dnc mutants results in accumulation of cAMP, in contrast to its depletion in rut. Despite such counteractive influences of rut and dnc mutations on cAMP metabolism, we found similar trends of changes in satellite frequency and overall bouton formation caused by dnc mutations. Significantly, dnc;;slo double mutants displayed drastic suppression of both types of satellites and mature boutons, in contrast to a relatively selective reduction in type M satellites and mature boutons in dnc;sei double mutants (Fig. 6; supplemental Table S1, available at www.jneurosci.org as supplemental material). These results thus suggest that fine-tuning of an optimal level, rather than a simple reduction or elevation, of cAMP is an important aspect of regulating satellite formation as well as overall synaptic growth. This is an interesting parallel to the previous reports that defects in cAMP synthesis or degradation caused by rut and dnc, respectively, can lead to similar abnormalities at multiple levels, despite their opposite effects on cAMP concentration. These mutant phenotypes include poor learning ability of adult flies (Tully and Quinn, 1985), defects in growth cone morphology and motility (Kim and Wu, 1996) and disrupted spike-firing patterns (Zhao and Wu, 1997) in cultured embryonic neuroblasts, and impaired synaptic function at larval NMJs (Renger et al., 2000). Our study thus further supports the idea of “optimal cAMP levels” required for proper neuronal function and structure.
We further investigated differential effects of rut and dnc mutations on slo and sei satellites with targeted alterations of cAMP metabolism in the pre- and post-synaptic compartments using the GAL4-UAS expression system (Brand and Perrimon, 1993). Among the possible GAL4 and UAS lines to be introduced into the slo and sei mutant backgrounds, two combinations were selected for the ease of genetic schemes that allow targeted pre-synaptic expression of a wild-type dnc transgene (UAS-dnc+). Since dnc PDE hydrolyzes cAMP, forced expression of PDE could lead to decreased cAMP levels, resembling rut-induced reduction in cAMP levels.
Interestingly, pre-synaptic expression of PDE (UAS-dnc+ driven by pan-neuronal C155-GAL4) in the sei background closely mimicked the rut effects on sei satellite formation, i.e., drastically reduced type M, but not type B, satellites (Fig. 6B) along with a milder decrease in primary bouton (supplemental Table S1, available at www.jneurosci.org as supplemental material) and branch numbers (supplemental Fig. S4B, available at www.jneurosci.org as supplemental material). These results thus support a major effect of sei mutations that is associated with pre-synaptic cAMP regulation.
In contrast to sei, we found that in the slo background, pre-synaptic expression of PDE (UAS-dnc+ driven by motoneuron-specific C164-GAL4) failed to phenocopy rut;;slo (Fig. 6B), consistent with confined cAMP-dependent regulatory mechanisms to the post-synaptic compartment in slo satellite formation. Indeed, post-synaptic dnc+ expressed in slo was effective in suppression of both type B and M satellites (supplemental Fig. S5, available at www.jneurosci.org as supplemental material), comparable to those in rut;;slo double mutants (compare Fig. 6B and supplemental Fig. S5, available at www.jneurosci.org as supplemental material). However, the number of mature boutons remained higher in slo mutants expressing post-synaptic dnc+ (supplemental Fig. S5; for slo and rut;;slo, see supplemental Table S1, available at www.jneurosci.org as supplemental material), indicating that the phenotypes of rut;;slo can be partially reproduced by overexpression of post-synaptic, but not pre-synaptic, dnc+ in slo mutants.
It should be noted that differential suppression of slo and sei satellites by pre-synaptic PDE expression resembles the effects of fasII and cac mutations (compare Figs. 4, 5, 6). However, rut- and pre-synaptic dnc+-induced suppression of sei satellites was restricted to type M satellites (Fig. 6B), whereas both fasII and cac mutations caused significant suppression of type M and, to a lesser extent, type B satellites in double-mutant combinations with sei (Figs. 4C, 5B). Therefore, the promotion of earlier growth steps in sei mutants (Fig. 2) may involve additional Ca2+-activated signaling pathways, such as CaMKII and PKG, that are known to affect synaptic growth at larval NMJs (Wang et al., 1994; Renger et al., 1999).
Discussion
Sequential growth process revealed by preferential effects of BK and Erg K+-channel mutations and the associated pre- and post-synaptic regulatory mechanisms
Distinct satellite patterns induced by slo and sei mutations support the notion that the two K+ channels act on separate growth steps in concert with localized molecular partners. Our double-mutant analysis leads to a minimal model involving functional interactions of Slo and Sei K+ channels with distinct assemblies of pre- and post-synaptic regulators in the sequential steps of synaptic growth and differentiation (Fig. 7). Expression of slo mutant phenotypes depends on scaffold protein, Dlg, and post-synaptic Dmca1D Ca2+ channels, both of which appear to be important for initial budding of satellites (Figs. 4, 5, 7). Double-mutant analysis reveals a tight association between Sei, but not Slo, K+ channels and adhesion molecule, FasII, and pre-synaptic Cac Ca2+ and Para Na+ channels in initial satellite formation as well as the ensuing process (Figs. 4, 5, 7). In the same vein, manipulations of pre-synaptic cAMP affect only sei-induced satellite formation, whereas slo satellites are more susceptible to modulations in post-synaptic cAMP signaling (Figs. 6, 7; supplemental Fig. S5, available at www.jneurosci.org as supplemental material).
Whereas these pre- and post-synaptic molecules can contribute to the initial growth of satellites in slo and sei mutants, they may also be important for further differentiation and stabilization of such intermediate structures. The stabilized satellites could accumulate over time and would facilitate their capture in fixed preparations. Our immunohistochemical and electron-microscopic analyses indicate that the majority of slo and sei satellites are well differentiated in molecular composition and ultrastructure (Figs. 3, 4; supplemental Fig. S1, available at www.jneurosci.org as supplemental material). As live imaging studies have demonstrated, differentiation of early “ghost boutons” occurs at a slow rate, taking hours to days (Ataman et al., 2008). Consistently, our preliminary live imaging indicates type B and M satellites abundant in mutants as stable structures with no active morphological changes over the observation period up to 1 h, during which new satellites were sighted budding from primary boutons after high K+ stimulation (data not shown). Thus, the synaptic differentiation process involving Slo or Sei K+ channels and their interacting partners may occur at a slower time scale (Fig. 7, dashed arrows).
Our results demonstrate a more profound influence of post-synaptic molecules on initial induction of satellite formation and major pre-synaptic contribution in subsequent steps (Fig. 7). This picture is in line with potential retrograde signaling during the sequential growth process. Recent studies at Drosophila larval NMJs have revealed significant contributions of retrograde factors, such as bone morphogenetic protein, to synaptic development and function (Keshishian and Kim, 2004). It will be important to examine whether and how these factors take part in particular steps of the proposed sequential growth process.
Separate interacting partners of BK and Erg K+ channels in regulation of synaptic growth
There has been emerging evidence for colocalization of post-synaptic BK channels with L-type Ca2+ channels (Hui et al., 1991) and with PSD-95 scaffold protein (Sailer et al., 2006) at vertebrate synapses. Our genetic analysis thus demonstrates the functional significance of the homologous post-synaptic macromolecular association (Slo BK, Dmca1D/L-type Ca2+ channels, and Dlg/PSD-95) in synaptic growth at the Drosophila NMJ. Whether interactions among these players are also important for regulation of synaptic transmission awaits further investigations.
It has been shown that seits1 mutants display increased spontaneous activities in the giant-fiber neuron (Elkins and Ganetzky, 1990) and enhanced synaptic growth at larval NMJs (Guan et al., 2005) when exposed to high temperature. However, we observed synaptic overgrowth even at room temperature in seits2 (Fig. 1) and, to a lesser extent, seits2/seits1 (data not shown). DNA sequencing predicts truncated versus full-length polypeptides in the seits1 and seits2 alleles, respectively (Titus et al., 1997; Wang et al., 1997), which could explain the observed allele-dependent differences. Notably, altered pre-synaptic Ca2+ and cAMP regulation drastically suppressed sei phenotypes, but was ineffective on slo-induced overgrowth (Figs. 5, 6), suggesting significant interactions between Erg K+ channels and these pre-synaptic components, although pre-synaptic interaction of the Sei K+ channel with the Ca2+/cAMP pathway has not been well established in Drosophila. DNA sequence analysis suggests a putative cyclic nucleotide-binding domain in Sei K+ channels, similar to Eag that belongs to the same K+-channel family (Wang et al., 1997). Whether cAMP-dependent modification of sei phenotypes (Fig. 6) is related to the action of this putative domain should be further investigated in future studies.
Multimeric assembly and transgenic manipulations of Slo and Sei K+ channels
Multimeric assembly of K+ channels, including Sei Erg and Slo BK, has been implicated in regulating the channel properties (Hille, 2001). Indeed, seits2/+ and slo/+ larvae, presumably containing a mixture of mutated and WT subunits in their Erg and BK channels, display dominant mutational effects on satellite formation and associated synaptic growth (supplemental Table S2, available at www.jneurosci.org as supplemental material). Importantly, pre-synaptic expression of a mutated sei transgene (UAS-seits2) in WT led to a similar, but less extreme, phenotype (supplemental Table S2, available at www.jneurosci.org as supplemental material), confirming the pre-synaptic action of seits2 and its dominant effects in multimeric Sei channels.
It is interesting to ask whether simply reducing the amount of Sei and Slo channel proteins may produce phenotypes similar to heterozygous seits2/+ and slo/+ animals. We employed the RNA interference (RNAi) technique to test this possibility, using multiple combinations of GAL4 drivers and UAS-slo/sei-RNAi constructs, with Dicer-2 to facilitate RNA interference in some combinations (Lee et al., 2004; Pham et al., 2004) (supplemental Table S3, available at www.jneurosci.org as supplemental material). However, none of these combinations caused characteristic behavioral and physiological abnormalities of sei and slo (Jackson et al., 1984; Elkins et al., 1986; Elkins and Ganetzky, 1990; Gho and Ganetzky, 1992; Warbington et al., 1996; Atkinson et al., 2000; Lee, 2008; Lee et al., 2008). We observed only marginal and inconsistent synaptic growth phenotypes among these combinations. For instance, the expression of slo and sei RNAi in motoneurons with the driver C164-GAL4 led to a slightly elevated satellite frequency, but the pan-neuronal driver C155-GAL4 produced even less overgrowth. Bouton formation was enhanced in these GAL4-UAS-RNAi combinations but not significantly above the elevated levels intrinsic to individual GAL4 and RNAi lines (supplemental Table S3, available at www.jneurosci.org as supplemental material).
The results suggest that dysfunctions induced by RNAi knockdown may not reproduce all aspects of mutant phenotypes. A match in protein levels or altered protein properties may be required to produce the phenotype of interest. At this time, the efficiency of these RNAi lines has not been documented. Since we were unable to measure the levels of Slo and Sei proteins because of a lack of appropriate antibodies, it is not possible to determine the levels of each RNAi knockdown. The slo and sei mutations induced by a chemical mutagen, ethyl methanesulfonate, may affect the properties and/or the amount of the gene product. For example, seits2 mutants carry a point mutation near the pore domain of the channels (Titus et al., 1997; Wang et al., 1997), and thus may act as neomorphs that confer dominant effects in the heterozygote, a property difficult to be mimicked by RNAi knockdown.
cAMP-dependent regulation of synaptic overgrowth induced by BK and Erg K+-channel dysfunction
Our results point out the critical role of cAMP signaling in the expression of both slo and sei mutant phenotypes (Fig. 7) and further highlight the profound functional consequences of altered excitability in neuronal plasticity. Activation of rut AC by activity-dependent accumulation of intracellular Ca2+ is pivotal in several forms of synaptic plasticity. For instance, in the Aplysia siphon-gill withdrawal reflex model, sensitizing stimuli increase cAMP levels and subsequently enhance transmission efficacy at sensorimotor synapses (Bernier et al., 1982), and repeated conditioning induces sensory varicosity growth (Bailey and Chen, 1983). Similarly, cAMP-dependent activation of protein kinase A in hippocampal slices is required for late-phase LTP that involves formation of new dendritic spines (Nguyen and Kandel, 1996).
At Drosophila larval NMJs, altered cAMP metabolism in rut and dnc mutants impairs synaptic transmission stability (Renger et al., 2000) and post-tetanic potentiation (Zhong and Wu, 1991). In addition, fewer docked vesicles (Renger et al., 2000) and retarded reserve pool mobilization (Kuromi and Kidokoro, 2000) have been documented in these mutants, indicating vesicle targeting and cycling defects. Thus, it will be interesting to examine the possibility that suppression of slo and sei satellites by rut is associated with alterations in membrane recycling. Such studies can be facilitated by relevant mutations, such as shibire defective in Dynamin, which is responsible for vesicle pinch-off (Koenig and Ikeda, 1983; Kim and Wu, 1987; Stimson and Ramaswami, 1999), or drp1 (Dynamin-related protein 1) defective in reserve pool mobilization (Verstreken et al., 2005).
In summary, our observations reveal distinct patterns of satellite formation induced by sei and slo mutations affecting two separate categories of K+ channels, which are apparently regulated by pre- and post-synaptic Ca2+/cAMP signaling, respectively. Together with previous studies, convergence on the Ca2+/CaM-activated cAMP synthesis by rut AC in the regulation of synaptic growth induced by a variety of K+ channel mutations (Budnik et al., 1990; Zhong et al., 1992; Zhong and Wu, 2004) further establishes a central role of rut AC in activity-dependent plasticity of synaptic function and growth.
Footnotes
This work was supported by National Institutes of Health Grant NS 26528 to C.F.W. We thank Mr. D. Abel and Dr. A. Ueda for technical assistance in TEM experiments. We also thank Drs. N. Atkinson, H. Atwood, H. Bellen, V. Budnik, A. DiAntonio, D. Eberl, B. Ganetzky, R. Ordway, J. Simpson, and J. Wang for providing fly stocks and antibodies and Dr. T. Littleton for providing confocal microscope facility for live imaging.
- Correspondence should be addressed to Dr. Jihye Lee, The Picower Institute for Learning and Memory, Massachusetts Institute of Technology, Cambridge, MA 02139. jihyelee{at}mit.edu