the metabotropic glutamate receptor activates the lipid kinase

INVESTIGATION

The Metabotropic Glutamate Receptor Activates the LipidKinase PI3K in Drosophila Motor Neurons Through theCalcium/Calmodulin-Dependent Protein Kinase II and

the Nonreceptor Tyrosine Protein Kinase DFakCurtis Chun-Jen Lin, James B. Summerville, Eric Howlett,1 and Michael Stern2

Department of Biochemistry and Cell Biology, Rice University, Houston, Texas 77005

ABSTRACT Ligand activation of the metabotropic glutamate receptor (mGluR) activates the lipid kinase PI3K in both the mammaliancentral nervous system and Drosophila motor nerve terminal. In several subregions of the mammalian brain, mGluR-mediated PI3Kactivation is essential for a form of synaptic plasticity termed long-term depression (LTD), which is implicated in neurological diseasessuch as fragile X and autism. In Drosophila larval motor neurons, ligand activation of DmGluRA, the sole Drosophila mGluR, similarlymediates a PI3K-dependent downregulation of neuronal activity. The mechanism by which mGluR activates PI3K remains in-completely understood in either mammals or Drosophila. Here we identify CaMKII and the nonreceptor tyrosine kinase DFak ascritical intermediates in the DmGluRA-dependent activation of PI3K at Drosophila motor nerve terminals. We find that transgene-induced CaMKII inhibition or the DFakCG1 null mutation each block the ability of glutamate application to activate PI3K in larvalmotor nerve terminals, whereas transgene-induced CaMKII activation increases PI3K activity in motor nerve terminals in a DFak-dependent manner, even in the absence of glutamate application. We also find that CaMKII activation induces other PI3K-dependenteffects, such as increased motor axon diameter and increased synapse number at the larval neuromuscular junction. CaMKII, butnot PI3K, requires DFak activity for these increases. We conclude that the activation of PI3K by DmGluRA is mediated by CaMKIIand DFak.

METABOTROPIC glutamate receptors (mGluRs), whichare G protein-coupled receptors for which glutamate is

ligand, mediate aspects of synaptic plasticity in several sys-tems. In several regions of the mammalian brain, includingthe hippocampus, the cerebellum, the prefrontal cortex,and others, ligand activation of group I mGluRs inducesa long-term depression of synaptic activity, termed mGluR-mediated long-term depression (LTD) (Luscher and Huber2010). Induction of mGluR-mediated LTD both activatesand requires the activation of the lipid kinase PI3 kinase(PI3K) and the downstream kinase Tor (Hou and Klann2004). Several genetic diseases of the nervous system are

predicted to increase sensitivity to activation of mGluR-mediated LTD. For example, increased sensitivity to induc-tion of mGluR-mediated LTD has been observed in themouse model for fragile X (Bear et al. 2004). Furthermore,the genes affected in tuberous sclerosis (Tsc1 and Tsc2) andneurofibromatosis (Nf1) encode proteins that downregulateTor activity (Gao et al. 2002; Dasgupta et al. 2005). Theseobservations raise the possibility that hyperactivation ofmGluR-mediated LTD plays a causal role in the neurologicalphenotypes of fragile X, neurofibromatosis and tuberoussclerosis (Kelleher and Bear 2008). Because these diseasesare each associated with an extremely high incidence ofautism spectrum disorders (ASDs), and because several linesof evidence suggest that elevated PI3K activity is associatedwith ASDs (Serajee et al. 2003; Kwon et al. 2006; Mills et al.2007; Cusco et al. 2009), it has been hypothesized thathyperactivation of this pathway might be responsible forASDs as well. Thus it would be of interest to identify addi-tional molecular components by which mGluR activation

Copyright © 2011 by the Genetics Society of Americadoi: 10.1534/genetics.111.128561Manuscript received March 8, 2011; accepted for publication April 5, 20111Present address: Department of Radiation Oncology, Virginia CommonwealthUniversity, 401 College Street, Richmond, VA 23298.

2Corresponding author: Department of Biochemistry and Cell Biology MS-140, RiceUniversity, 6100 Main St., Houston, TX 77005. E-mail: [email protected]

Genetics, Vol. 188, 601–613 July 2011 601

mailto:[email protected]

activates PI3K, and yet despite recent advances, this mech-anism remains incompletely understood.

In Drosophila larval motor neurons, glutamate activa-tion of the single mGluR, called DmGluRA, downregulatesneuronal excitability (Bogdanik et al. 2004); glutamate bothactivates PI3K and requires PI3K activity for this downregu-lation (Howlett et al. 2008). Because glutamate is the excit-atory neurotransmitter at the Drosophila neuromuscularjunction (NMJ) (Jan and Jan 1976), it was hypothesizedthat this DmGluRA-mediated downregulation of neuronalexcitability carried out a negative feedback on activity:glutamate released from motor nerve terminals would ac-tivate DmGluRA autoreceptors, which would then depressexcitability.

Here we identify additional molecular components thatmediate the activation of PI3K by DmGluRA in Drosophilalarval motor nerve terminals. We find that activity of thecalcium/calmodulin-dependent kinase II (CaMKII) is neces-sary for glutamate application to activate PI3K, and expres-sion of the constitutively active CaMKIIT287D (Jin et al. 1998)is sufficient both to activate PI3K even in the absence ofglutamate and to confer several other neuronal phenotypesconsistent with PI3K hyperactivation. We also find thatCaMKIIT287D requires the nonreceptor tyrosine kinase DFakfor this PI3K activation: the DFakCG1 null mutation (Grabbeet al. 2004) blocks the ability of glutamate application toactivate PI3K and prevents CaMKIIT287D from hyperactivat-ing PI3K. Finally, CaMKIIT287D expression completely sup-presses the hyperexcitability conferred by the DmGluRAnull mutation DmGluRA112b. We conclude that ligand acti-vation of DmGluRA activates PI3K via CaMKII and DFak.

Materials and Methods

Drosophila stocks

For all experiments, Drosophila larvae were reared on stan-dard cornmeal/agar media at 22–23�. The D42 Gal4 driver(Brand and Perrimon 1993; Parkes et al. 1998), whichexpresses in motor neurons, was provided by Tom Schwarz(Harvard Medical School, Boston, MA). Flies carrying theUAS-PI3KDN (D954A) and UAS-PI3K-CAAX transgenes (Leeverset al. 1996), encoding dominant-negative and constitutivelyactive PI3K, respectively, were provided by Sally Leevers(London Research Institute, London, UK). Flies carryingthe UAS-ala, UAS-CaMKIIT287A, and UAS-CaMKIIT287D trans-genes, which encode a CaMKII inhibitor peptide, calcium/calmodulin-dependent CaMKII, and calcium/calmodulin-independent CaMKII, respectively (Griffith et al. 1994; Parket al. 2002), were provided by Leslie C. Griffith (BrandeisUniversity, Waltham, MA). Flies carrying the DFakCG1 dele-tion mutation and UAS-DFak1 transgene (Grabbe et al. 2004)were provided by Ruth Palmer (Umeå University, Umeå,Sweden). Flies carrying the DmGluRA112b deletion mutation(Bogdanik et al. 2004) were provided by Marie-Laure Par-mentier (Unité Propre de Recherche Centre, Montpellier,

France). All other fly stocks were provided by the DrosophilaStock Center (Bloomington, IN).

Immunocytochemistry

Larvae were grown to the wandering third instar stage inuncrowded half-pint bottles. Larvae were collected for ex-perimentation only during the first and second days afterthe initial wandering third instar larvae appeared. For basalphosphorylated Akt (p-Akt) measurements, larvae weredissected in Schneider’s Drosophila media (Gibco) and fixedin Schneider’s Drosophila media containing 4% paraformal-dehyde. For experiments involving glutamate application,larvae were dissected and then incubated for 1 min inSchneider’s Drosophila media containing 100 mM glutamateprior to the fixation procedure described above. Fixed larvaltissues were incubated with a rabbit anti-Drosophila p-Akt(Ser505) polyclonal primary antibody (1:500 dilution; CellSignaling Technologies). The antibody used was from a dif-ferent lot from the antibody used in a previous report(Howlett et al. 2008) and thus might not have identicalproperties. Larvae were then incubated with RhodamineRed conjugated goat antirabbit IgG (1:500 dilution; JacksonImmunoResearch), and Cy-2 conjugated antibodies againsthorseradish peroxidase (1:200 dilution; Jackson Immuno-Research). Immunolabeled larval tissues in standard phos-phate buffered saline (PBS) (0.128 M NaCl, 2.0 mM KCl,4.0 mM MgCl2, 0.34 M sucrose, 5.0 mM Hepes, pH 7.1, and0.15 mM CaCl2) containing 50% glycerol were mountedonto slides. Care was taken to treat all samples identicallyduring this procedure. NMJs from muscles 6 and 7 in seg-ments A3, A4, A5, or A6 were imaged on a Zeiss LSM 510confocal microscope system (Zeiss, Oberkochen, Germany)with a ·20 objective. Optical sections were 10 mm thick,which encompassed the entire NMJ. Optical parameters, in-cluding pinhole, gain, contrast, and brightness, were heldconstant for each experimental set. NMJs (marked by anti-HRP) were traced using the ImageJ (National Institutes ofHealth, Bethesda, MD) freehand selection tool and the se-lection was transferred to the anti–p-Akt image where pixelintensity value was quantified. Background was obtained byaveraging p-Akt intensity from muscles 6 and 7 from thesame abdominal hemisegment, excluding the muscle nucleiand the neuromuscular junction. This background was thensubtracted from the mean motor nerve terminal p-Akt pixelintensity. This p-Akt pixel intensity was then normalized tothe mean p-Akt pixel intensity of wild-type motor nerveterminals obtained in experiments performed in parallel.

To measure synaptic bouton number, larvae were grown,selected, and dissected as described above, fixed in standardPBS containing 4% paraformaldehyde and labeled with Cy-2conjugated antibodies against horseradish peroxidase (1:200dilution). Images were obtained as described above. The LSMImage Browser was used for quantifying bouton numbers andobtaining surface area of muscles 6 and 7 from abdominalsegments A3 and A4. The number of boutons was countedmanually.

602 C. Chun-Jen Lin et al.

Transmission electron microscopy

Transmission electron microscopy was performed as describedpreviously (Lavery et al. 2007; Howlett et al. 2008). Briefly,larvae were grown, selected, and dissected as described above,fixed with glutaraldehyde and paraformaldehyde, stainedwith 0.5% OsO4 and 2% uranyl acetate, and embedded inan eponate 12-araldite mixture. Ultrathin cross-sectionalslices (pale gold, 120 nm thick) were cut with an ultrami-crotome, poststained with uranyl acetate and Reynoldslead citrate, and analyzed with a JEOL 1230 (JEOL, Tokyo,Japan) transmission electron microscope at 80 kV. Carl ZeissAxioversion software (Zeiss, Oberkochen, Germany) wasused for analyzing axon diameter. Each experimental setwas analyzed from the five largest axons (motor axons)from more than five different nerves from at least two dif-ferent larvae.

Electrophysiology

Larvae were grown and selected as described above anddissected in Jan’s buffer (128 mM NaCl, 2.0 mM KCl,4.0 mM MgCl2, 34 mM sucrose, 4.8 mM Hepes, pH 7.1,and CaCl2 concentration as specified in the text). Peripheralnerves were cut immediately posterior to their exit from theventral ganglion and were stimulated with a suction elec-trode at a 5-V stimulus intensity. Muscle recordings weretaken from muscle 6 in abdominal segments A3, A4, orA5. Stimulus duration, �0.05 msec, was adjusted to 1.5 timesthreshold, which reproducibly stimulates both axons innervat-ing muscles 6 and 7. Intracellular recording electrodes formuscle potentials were pulled with a Flaming/Brown micropi-pette puller to a tip resistance of 10–40 MV and filled with 3MKCl. Rate of onset of long-term facilitation (LTF) was reportedas geometric means because the data show a positive skew.

For all LTF experiments, the bath solution contained0.15 mM [Ca21] and 100 mM quinidine, which is a potassiumchannel blocker that sensitizes the motor neuron and enablesLTF to occur and be measured even in hypoexcitable neurons.

Results

CaMKII regulates PI3K activity in Drosophila motornerve terminals

Metabotropic glutamate receptors (mGluRs) are G protein-coupled receptors for which glutamate is ligand. In Drosoph-ila, the DmGluRA112b null mutation in the single mGluRincreases motor neuron excitability by preventing activationof the lipid kinase PI3K (Bogdanik et al. 2004; Howlett et al.2008). Given that glutamate is the major excitatory neuro-transmitter at NMJs in insects such as Drosophila (Jan andJan 1976), it was suggested that DmGluRA in motor neu-rons participate in a negative feedback: glutamate releasedfrom motor nerve terminals as a consequence of neuronalactivity activates DmGluRA autoreceptors in motor nerveterminals. DmGluRA activation downregulates subse-quent neuronal activity by activating PI3K. In this view,

DmGluRA112b increases excitability by preventing PI3K ac-tivation and thus severing this negative feedback. This view issupported by the observation that transgene-induced PI3Kinhibition increases neuronal excitability in a manner similarto DmGluRA112b, whereas transgene-induced PI3K activationdecreases neuronal excitability (Howlett et al. 2008). How-ever, the molecular mechanism by which glutamate/DmGluRAactivates PI3K was not elucidated.

We initially hypothesized that CaMKII might be anessential intermediate in DmGluRA-mediated PI3K activa-tion. Although the possibility that CaMKII can activate PI3Khas only recently been reported (Ma et al. 2009), alteredCaMKII activity alters certain motor neuron phenotypesin Drosophila in a manner similar to altered PI3K activity(Griffith et al. 1994; Park et al. 2002). One way that CaMKIIactivity can be altered is by expression of the constitutivelyactive CaMKIIT287D: this variant is rendered active even inthe absence of calcium or calmodulin by the phosphomi-metic T287D substitution in the autophosphorylation site(Fong et al. 1989). It was previously found that motor neu-ron expression of CaMKIIT287D increased synapse numberand decreased excitability in motor neurons (Park et al.2002); both of these phenotypes are also exhibited in larvaeexpressing the constitutively active PI3K-CAAX (Martin-Penaet al. 2006; Howlett et al. 2008). Similarly, CaMKII inhibitionaccomplished by transgenic expression of the CaMKII inhibi-tory peptide “ala” (Joiner and Griffith, 1997) increases motorneuron excitability and decreases synapse number (Park et al.2002); both of these phenotypes are also exhibited in trans-genic larvae in which PI3K activity is blocked. Taken together,these results support the notion that levels of PI3K activity arespecified, at least in part, by activity of CaMKII.

To test the prediction that altered CaMKII activity wouldalter PI3K activity in Drosophila larval motor nerve termi-nals, we used the D42 motor neuron Gal4 driver to expresstransgenes encoding either the nonactivated CaMKIIT287A

(which acts equivalently to wild type in all assays performedto date), the constitutively active CaMKIIT287D, or the CaMKIIinhibitory peptide ala in motor neurons, and then evaluatedeffects of these transgenes on PI3K activity. We monitoredPI3K activity with an antibody against the phosphorylatedform of the kinase Akt (p-Akt), which increases upon PI3Kactivation and is a well-established readout for PI3K activity(Colombani et al. 2005; Dionne et al. 2006; Palomero et al.2007). We found that expressing the activated CaMKIIT287D,but not the nonphosphorylatable CaMKIIT287A, increasedp-Akt levels in motor nerve terminals �40% (Figure 1, P ¼0.002, n ¼ 29), which is similar to, but less extreme, thanthe twofold increase in p-Akt conferred by transgene-induced PI3K activation (Howlett et al. 2008). The mag-nitude of this response is similar to the magnitude of thep-Akt increase induced during mGluR-mediated LTD inthe mouse hippocampus (Hou and Klann 2004), consis-tent with the possibility that the p-Akt increases in theDrosophila and mammalian systems occur via a commonpathway.

Activation of PI3K by CaMKII and DFak 603

The previous study in which this approach was used(Howlett et al. 2008) was not able to distinguish possibledifferences in basal p-Akt levels among genotypes because ofhigh sample errors. However, by significantly increasing sam-ple size and by including data normalization to accommodatesignal variability, we were able to show that CaMKII inhibitionvia ala expression decreased basal p-Akt �25% (Figure 1, P ¼0.033, n ¼ 16). These results support the notion that CaMKIIactivity specifies PI3K activity in larval motor neurons.

Focal adhesion kinase (DFak) is required for CaMKII toactivate PI3K

Next we wanted to identify molecules that might serve asintermediates linking activated CaMKII with PI3K activation.We hypothesized that the nonreceptor tyrosine kinase focaladhesion kinase (DFak) might play such a role: DFak is thesingle Drosophila ortholog of the family which in mammalsconsists of both Fak and the closely related Pyk2. Inmammals, CaMKII phosphorylates Pyk2 on multiple serineresidues on the C terminus, which leads to Pyk2 tyrosinephosphorylation and activation by mechanisms that are in-completely understood (Della Rocca et al. 1997; Soltoff1998; Zwick et al. 1999; Heidinger et al. 2002; Fan et al.2005; Montiel et al. 2007). CaMKII also phosphorylatesmammalian Fak on C-terminal serine residues, althoughthe functional significance of these phosphorylation eventsremains unknown (Fan et al. 2005). Furthermore, both Pyk2and Fak are capable of activating PI3K (Chen and Guan1994; Schlaepfer et al. 1994; Guinebault et al. 1995; Chenet al. 1996; Dikic et al. 1996; Avraham et al. 2000; Rocicet al. 2001; Montiel et al. 2007). Two CaMKII phosphoryla-tion sites on Fak, serines 843 and 910, are conserved inDFak, raising the possibility that CaMKII phosphorylatesDFak at these serine positions, which leads to DFak and thenPI3K activation (Grabbe et al. 2004).

If DFak acts downstream of CaMKII, then the DFak nullmutation DFakCG1 is predicted to decrease motor nerve ter-minal PI3K activation, and prevent CaMKIIT287D-dependentPI3K activation. As shown in Figure 1, PI3K activity in motornerve terminals, as monitored by levels of p-Akt, was de-creased in DFakCG1 to 80% of wild type (P ¼ 0.008, n ¼70). Furthermore, this decrease was rescued by motor neu-ron expression of a DFak1 transgene (Figure 1), demonstrat-ing that this effect of DFakCG1 resulted from loss of DFakspecifically within motor neurons. Furthermore, DFakCG1

larvae expressing CaMKIIT287D in motor neurons did notexhibit the increased p-Akt levels in motor nerve terminalsobserved when CaMKIIT287D was expressed in an otherwisewild-type background (Figure 1, P , 0.001). Rather, p-Aktlevels were decreased to levels similar to those observedin DFakCG1 larvae in the absence of CaMKIIT287D expres-sion (P ¼ 0.4). Similarly, coexpression with the dominant-negative PI3KDN (Leevers et al. 1996) blocked the ability ofCaMKIIT287D to increase p-Akt levels (Figure 1). Taken to-gether, these results confirm that both DFak and PI3K activ-ities are required for CaMKII to increase p-Akt levels.

Figure 1 CaMKII activity increases levels of p-Akt in larval motor nerveterminals in a DFak-dependent manner. (A) Representative confocalimages of third instar larval neuromuscular junctions of wild type (WT),D42 . CaMKIIT287D, and DFakCG1 stained with anti-pAkt (top), anti-HRP(middle), and merged images (bottom). All images are taken frommuscles 7 and 6 from abdominal segments A3, A4, A5, or A6. Bar, 20 mm.(B) Mean normalized p-Akt intensity (y axis) 6 SEM for the follow-ing genotypes (x axis): D42.1, DFakCG1, D42 . CaMKIIT287D, D42 .CaMKIIT287A, D42 . ala, DFakCG1; D42 . DFak1, DFakCG1; D42 .CaMKIIT287D, and D42 . CaMKIIT287D, PI3KDN. Anti-HRP was appliedto outline nerve terminals. Pixel intensities were quantified using ImageJsoftware and background subtraction was performed as described inMATERIALS ANDMETHODS. The normalized p-Akt intensities are shown asa ratio change compared to wild type. From left to right, n ¼ 65, 70, 29,30, 16, 84, 35, and 18, respectively, for each genotype. One-way ANOVAand Fisher’s LSD were performed for statistical analysis and gave thefollowing differences: For D42.1; vs. DFakCG1, P ¼ 0.008; vs. D42 .CaMKIIT287D, P ¼ 0.002; vs. D42 . CaMKIIT287A, P ¼ 0.62; vs. D42 .ala, P ¼ 0.033; vs. DFakCG1; D42 . DFak1, P ¼ 0.34. For DFakCG1; vs.DFakCG1; D42 . DFak1, P ¼ 0.018; vs. DFakCG1; D42 . CaMKIIT287D,P ¼ 0.40. For D42 . CaMKIIT287D; vs. DFakCG1; D42 . CaMKIIT287D, P ,0.001; vs. D42 . CaMKIIT287D, PI3KDN, P ¼ 0.001. P values ,0.05 wereconsidered statistically significant.


CaMKII and DFak are required for PI3K activation byglutamate application

It was previously demonstrated that application of gluta-mate to larval NMJs in filleted third instar larval prepara-tions activates PI3K, as monitored by glutamate-inducedincreases in p-Akt levels in motor nerve terminals (Howlettet al. 2008). Furthermore, these glutamate-evoked p-Aktincreases were blocked by the null mutation DmGluRA112b

or by presynaptic inhibition of either DmGluRA or PI3K(Howlett et al. 2008), which demonstrated that glutamateactivated PI3K via DmGluRA autoreceptors. If CaMKII andDFak are critical intermediates in glutamate/DmGluRA-induced PI3K activation, then inhibition of either moleculeis predicted to block this glutamate-induced PI3K activation.To test this possibility, we applied 0.1 mM glutamate toDFakCG1 larvae, or to larvae expressing the CaMKII inhibi-tory peptide ala in motor neurons. We found that wild-typelarvae showed a robust glutamate-induced PI3K activation,as monitored by immunocytochemistry against p-Akt (Fig-ure 2, P ¼ 0.001), following 1 min of glutamate application.This percent increase is similar to the increase reported dur-ing mGluR-mediated LTD in the mouse hippocampus (Houand Klann 2004) in which a 5-min application of a glutamateanalog evoked an �50% increase in p-Akt levels. In contrast,blocking either CaMKII or DFak blocked this increase (Fig-ure 2, P ¼ 0.64 and P ¼ 0.77, respectively). Furthermore,the motor neuron expression of UAS-DFak1 restored normalglutamate sensitivity to DFakCG1 (Figure 2).

CaMKII promotes synapse formation viaDFak-dependent PI3K activation

CaMKII activity was previously shown to promote arboriza-tion and increase synapse number at the larval NMJ (Parket al. 2002) by phosphorylating discs large and thus decreas-ing levels of Fas2 (Koh et al. 1999). The observation thatCaMKII acts downstream of integrins for this control of syn-apse structure (Beumer et al. 2002) is consistent with a rolefor DFak in this process as well. In addition, arborization

and synapse number are regulated by PI3K activity: activat-ing PI3K by expressing the constitutively active PI3K-CAAXincreases synapse number �2.5-fold, whereas blocking PI3Kby expressing the dominant-negative PI3KDN decreasessynapse number about twofold (Martin-Pena et al. 2006;Howlett et al. 2008). From these data and the resultsshown above, we hypothesized that CaMKII promotes syn-apse formation at least in part by activating PI3K in a DFak-dependent manner. We confirmed the previous findings ofPark et al. (2002) that expressing CaMKIIT287D, but notCaMKIIT287A, in motor neurons increased bouton number�30% (Figure 3, P ¼ 0.001), an increase similar to, thoughless extreme, than the increase conferred by PI3K-CAAX.We found that this CaMKIIT287D-dependent increase re-quired PI3K activity: in larvae coexpressing PI3KDN andCaMKIIT287D, the CaMKIIT287D-dependent increase in bou-ton number was completely suppressed and in fact wasdecreased to a value indistinguishable from that conferredby PI3KDN alone (Figure 3).

To determine whether DFak is a required intermediate inthis CaMKII-dependent PI3K activation, we measured syn-apse number in DFakCG1 larvae both in the presence andabsence of CaMKIIT287D expression. We found that in theabsence of CaMKIIT287D expression, synapse number inDFakCG1 larvae was highly variable and therefore differenceswith wild type failed to reach statistical significance (P ¼0.051). However, there was a tendency for DFakCG1 synapsenumber to decrease, rather than increase, compared to wildtype (Figure 3). Thus we were unable to confirm the findingof Tsai et al. (2008), who reported that DFak inhibitionincreases synapse number at the larval NMJ. However,DFakCG1 completely suppressed the increased synapse numberconferred by CaMKIIT287D expression (Figure 3, P ¼ 0.001);synapse number in DFakCG1 larvae expressing CaMKIIT287D wasindistinguishable from synapse number in DFakCG1 alone (Fig-ure 3, P ¼ 0.89). These data support the notion that CaMKIIactivates PI3K in a DFak-dependent manner.

Figure 2 The glutamate-evoked increase in presynapticp-Akt levels requires CaMKII and DFak. (A) Representativeconfocal images of third instar larval neuromuscular junctionsof wild type (WT or1/1), DFakCG1 and DFakCG1; D42. UAS-DFak1 either prior to (2) or 1 min following (1) application of100 mM glutamate. Larvae were stained with anti-p-Akt (left)and anti-HRP (right). All images are from muscles 7 and 6 ofabdominal segments A3, A4, A5, or A6. Bar, 20mm. (B) Meannormalized p-Akt intensity (y axis)6 SEM, both prior to andfollowing 1-min glutamate application, for the following gen-otypes (x axis): D42.1, DFakCG1, D42 . ala, and DFakCG1;D42 . DFak1. Anti-HRP was applied to outline nerve termi-nals. Pixel intensities were quantified using ImageJ softwareand background subtraction was performed as described inMaterials and Methods. The normalized p-Akt intensities areshown as a ratio change compared to wild type. From left toright, n ¼ 65, 56, 70, 48, 84, 24, 32, and 36, respectively, foreach genotype. One-way ANOVA and Fisher’s LSD were per-formed for statistical analysis. P values,0.05 were consideredstatistically significant.


We also found that blocking CaMKII in motor neurons byexpressing the CaMKII-inhibitor peptide ala decreased syn-apse number about twofold (Figure 4, P , 0.001). If thisdecrease were a consequence of failure to activate PI3K,then expression of the constitutively active (and henceCaMKII independent) PI3K-CAAX would be predicted tosuppress these effects of CaMKII inhibition. We confirmedthis prediction: the ala-dependent decrease in synapse num-ber was completely suppressed by coexpression with PI3K-

CAAX: synapse number in larvae expressing both ala andPI3K-CAAX was significantly increased compared to synapsenumber in larvae expressing ala alone (Figure 4, P , 0.001)and was indistinguishable from synapse number in larvaeexpressing PI3K-CAAX alone (Figure 4, P ¼ 0.53). In addi-tion, we found that synapse number in larvae expressingPI3K-CAAX was similarly unaffected by the presence ofDFakCG1 (Figure 4). Taken together, these results suggest thatCaMKII increases synapse number by the DFak-dependentactivation of PI3K.

Certain genotypes exhibited altered bouton morphologyin addition to altered bouton number. For example, boutonsin DFakCG1 mutants often appear to be larger in size thanwild type. The DmGluRA112b mutation was similarly foundto increase bouton size (Bogdanik et al. 2004). These obser-vations are consistent with the possibility that DFak activa-tion is attenuated in the absence of DmGluRA.

CaMKII activity increases motor axon diameter by theDFak-dependent activation of PI3K

Larval motor axon diameter is specified by the level of PI3Kactivity: expressing constitutively active PI3K-CAAX increasesmotor axon diameter �75% (0.8–1.4 mm), whereas express-ing the dominant-negative PI3KDN decreases motor axon di-ameter (Howlett et al. 2008). To determine whether CaMKIIsimilarly regulates motor axon diameter, we expressedCaMKIIT287D in motor axons and found as predicted signif-icantly increased (�40%) motor axon diameters (Figure 5,P ¼ 0.001), an increase similar to, but less extreme than,the increase conferred by PI3K-CAAX. In addition, theCaMKIIT287D-dependent increase, but not the PI3K-CAAX-dependent increase, in axon diameter was blocked byDFakCG1 (Figure 5, P ¼ 0.004). Taken together, these resultssuggest that CaMKII increases motor axon diameter by theDFak-dependent activation of PI3K.

Altered CaMKII and DFak activities alter motorneuron excitability

PI3K activity regulates motor neuron excitability: transgene-induced PI3K activation decreases neuronal excitability,whereas transgene-induced PI3K inhibition increases neu-ronal excitability (Howlett et al. 2008). Additional reportshave demonstrated that CaMKII and DFak also regulate Dro-sophila neuronal excitability (Griffith et al. 1994; Park et al.2002; Ueda et al. 2008), raising the possibility that CaMKII,DFak, and PI3K might regulate excitability via a commonpathway. However, in each previous report, different sets ofexcitability assays were used, which complicates attempts toassign the effects of these three genes to specific pathways.

To address this complication, we monitored motorneuron excitability in DFakCG1 and in larvae expressing theconstitutively active CaMKIIT287D with the same assay pre-viously employed to monitor excitability in larvae with al-tered PI3K activity. With this assay, the larval neuromuscularpreparation (Jan and Jan 1976) transmitter release frommotor neurons is evoked by nerve stimulation, and the

Figure 3 The increase in motor neuron synapse number conferred byCaMKIIT287D expression requires DFak and PI3K. (A) Representative con-focal images of third instar larval neuromuscular junctions of D42.1,D42 . CaMKIIT287D, DFakCG1, D42 . CaMKIIT287D, PI3KDN, DFakCG1;D42 . DFak1, and DFakCG1; D42 . CaMKIIT287D. Nerve terminals werevisualized by staining with anti-HRP. Bar, 20 mm. (B) Means 6 SEM ofsynaptic bouton number normalized to the total surface area of muscle (yaxis) for the following genotypes (x axis): D42.1, D42 . CaMKIIT287A,D42 . CaMKIIT287D, D42 . PI3KDN, D42 . PI3KDN, CaMKIIT287D,DFakCG1, DFakCG1; D42 . DFak1, and DFakCG1; D42 . CaMKIIT287D.All measurements were performed on nerve terminals on muscles 6and 7 from abdominal segment A3. From left to right, n ¼ 25, 12, 18,8, 16, 26, 28, and 17, respectively, for each genotype. One-way ANOVAand Fisher’s LSD were performed for statistical analysis.


amount of transmitter released is inferred from the ampli-tude of the muscle depolarization (termed the excitatoryjunction potential, or EJP). Using this assay, we employedthe same readouts to evaluate neuronal excitability inDFakCG1 larvae and in larvae expressing CaMKIIT287D thatwere previously employed for altered PI3K (Howlett et al.2008).

First, we measured rate of onset of LTF (Jan and Jan1978). LTF is a form of synaptic plasticity induced whena larval motor neuron is subjected to a train of repetitivenerve stimulations at low external [Ca21]. At a certain pointduring the stimulus train, a threshold is reached, and sub-sequent stimulations elicit EJPs greatly increased in ampli-tude and duration. This long-term facilitation results fromincreased and asynchronous neurotransmitter release, whichin turn results from increased duration of nerve terminaldepolarization (Jan and Jan 1978). The number of stimula-tions required to reach this LTF threshold (LTF onset rate) isdecreased by increased neuronal excitability (Jan and Jan1978; Stern and Ganetzky 1989; Stern et al. 1990; Mallartet al. 1991; Poulain et al. 1994; Schweers et al. 2002). Boththe DmGluRA112b null mutation and transgene-induced in-hibition of PI3K greatly decrease the number of nerve stim-ulations required to reach LTF and thus greatly increase LTFonset rate (Bogdanik et al. 2004; Howlett et al. 2008). Thisincreased LTF onset rate indicates increased excitabilityin these genotypes. The possibility that DFak is requiredfor DmGluRA-dependent PI3K activation predicted thatDFakCG1 would prevent this PI3K activation and similarlyincrease the LTF onset rate. This prediction was confirmed(Figure 6A). This increased LTF onset rate was rescued bymotor neuron-dependent expression of DFak1, confirmingthat DFak is required in the motor neuron for this effect.In contrast, the possibility that CaMKIIT287D constitutivelyactivates PI3K would predict that CaMKIIT287D expression

would decrease LTF onset rate, which we also confirmed(Figure 6A).

Second, we measured EJP amplitude evoked by singlenerve stimulations at low external [Ca21]. At these low[Ca21], Ca21 is limiting for transmitter release, and muta-tions that alter excitability, and hence Ca21 influx, have sig-nificant effects on EJP amplitude (Ganetzky and Wu 1982;Huang and Stern 2002). In particular, transgene-inducedPI3K activation decreases EJP amplitude, whereas transgene-induced PI3K inhibition increases EJP amplitude (Howlettet al. 2008). The possibility that DFak is required forDmGluRA-dependent PI3K activation predicted that DFakCG1

would prevent this PI3K activation and similarly increaseEJP amplitude. We found that DFakCG1 significantly increasedEJP amplitude at 0.15 mM and 0.2 mM [Ca21] (Figure 6B),but was not significantly different from wild type at 0.1 mM[Ca21], at which transmitter release is the smallest anderrors are the greatest. This increased EJP amplitude wasrescued by motor neuron-dependent expression of DFak1,confirming that DFak is required in the motor neuron forthis effect.

To confirm that these effects on EJP amplitude reflectedaltered transmitter release rather than altered responsive-ness of the muscle to transmitter, we compared the amplitudeof miniature EJPs (mEJPs), which reflect the spontaneousrelease of individual vesicles of neurotransmitter, in wild-type larvae vs. DFakCG1 larvae and DFakCG1 larvae express-ing DFak1 in motor neurons (Figure 6C). We found thatmEJP amplitude was unaffected in these three genotypes.Thus the increased EJP amplitude observed in DFakCG1

reflects increased neurotransmitter release.The possibility that CaMKIIT287D constitutively activates

PI3K would predict that CaMKIIT287D expression would de-crease EJP amplitude. We found that as predicted, CaMKIIT287D

expression decreases EJP amplitude at each [Ca21] tested

Figure 4 Constitutive PI3K activation suppressesthe decrease in synapse number conferredby CaMKII or DFak inhibition. (A) Representa-tive confocal images of third instar larvalneuromuscular junctions of D42.1, D42 .ala, D42 . ala, PI3K-CAAX and DFakCG1;D42 . PI3K-CAAX. Nerve terminals were visu-alized by staining with anti-HRP. Bar, 20 mm. (B)Means 6 SEM of synaptic bouton number nor-malized to the total surface area of muscle(y axis) for the following genotypes (x axis):D42.1, D42 . ala, D42 . PI3K-CAAX, D42.ala, PI3K-CAAX, and DFakCG1; D42 . PI3K-CAAX.All measurements were performed on nerveterminals on muscles 6 and 7 from abdominalsegment A3. From left to right, n ¼ 25, 12, 15,8, and 10, respectively, for each genotype.One-way ANOVA and Fisher’s LSD were per-formed for statistical analysis.


(Figure 6B), while having no significant effect on mEJP am-plitude (Figure 6C). This observation supports the conclu-sion that decreased neurotransmitter release underlies thedecreased EJP amplitude conferred by CaMKIIT287D.

Activating CaMKII prevents hyperexcitability of theDmGluRA112b null mutant

The DmGluRA112b null mutant exhibits extreme neuronalhyperexcitability, as shown by a greatly increased rate ofonset of LTF (Bogdanik et al. 2004; Howlett et al. 2008).This hyperexcitability is completely suppressed by expres-sion of the constitutively active PI3K-CAAX and thus reflectsfailure to activate PI3K (Howlett et al. 2008). If this failureto activate PI3K is a consequence of failure to activateCaMKII, then this hyperexcitability is predicted to be sup-pressed by transgene-induced expression of the constitutivelyactive CaMKIIT287D. To test this prediction, we compared rateof onset of LTF in DmGluRA112b mutants and in DmGluRA112b

mutants expressing CaMKIIT287D in motor neurons. We foundthat CaMKIIT287D expression completely suppressed the hy-perexcitability of DmGluRA112b, and decreased excitability toa level indistinguishable from the hypoexcitability in larvaein which CaMKIIT287D was expressed in a DmGluRA1 back-ground (Figure 7). These results support the notion thatCaMKII is a critical intermediate in the ability of ligand-activated DmGluRA to activate PI3K.

Discussion

In both mammalian central synapses and Drosophila larvalmotor neurons, activation by glutamate of the metabotropicglutamate receptor (mGluR) activates the lipid kinase PI3K,but the mechanism by which this activation occurs has not

been elucidated. Here we identify CaMKII as a critical in-termediate in the ability of the single Drosophila mGluR(DmGluRA) to activate PI3K and show that the ability ofboth activated DmGluRA and CaMKII to activate PI3Krequires the nonreceptor tyrosine kinase, DFak (Figure 8).These results provide novel insights into the mechanism bywhich DmGluRA activation triggers the observed downregu-lation of subsequent neuronal activity in Drosophila motorneurons. These results might also be relevant to the mech-anism by which mGluR activates PI3K in mammalian centralsynapses, a process implicated in fragile X, ASDs, neurofibro-matosis, and tuberous sclerosis (Kelleher and Bear 2008).

How might CaMKII lead to the DFak-dependent activa-tion of PI3K? Although the ability of CaMKII to activate PI3Khas only recently been reported (Ma et al. 2009), it has beenwell established in mammals that CaMKII phosphorylatesboth Fak and Pyk2 on multiple serines on the C terminus.These phosphorylation events can activate Pyk2 by enablingsubsequent tyrosine phosphorylations (particularly at Tyr402)via mechanisms that are incompletely understood (DellaRocca et al. 1997; Soltoff 1998; Zwick et al. 1999; Heidingeret al. 2002; Fan et al. 2005; Montiel et al. 2007). It has alsobeen well established that Fak and Pyk2, when activated bytyrosine phosphorylation, are each able to activate PI3K:tyrosine-phosphorylated Fak binds p85, the PI3K regulatorysubunit, via both the SH3 and SH2 domains (Chen andGuan 1994; Guinebault et al. 1995; Chen et al. 1996). Inaddition, both tyrosine-phosphorylated Fak and Pyk2 arecapable of activating Ras via the conserved Grb2-SoS path-way (Schlaepfer et al. 1994; Dikic et al. 1996; Avraham et al.2000; Rocic et al. 2001), which could in principle lead to theRas-dependent, p85-independent activation of PI3K. Theseobservations raise the possibility that Drosophila CaMKII

Figure 5 DFak is required for the increased motor axondiameter conferred by CaMKII, but not PI3K. (A) Repre-sentative transmission electron micrographs of cross-sections of peripheral nerves from D42.1 and D42 .CaMKIIT287D larvae. The �80 motor and sensory axonswithin the nerve are surrounded by at least two layers ofglia. Diameter of the motor axons, the largest axons in thenerve, are considerably increased in D42 . CaMKIIT287D.An axon is marked with a red arrow. Bar, 2 mm. (B) Meanaxon diameter 6 SEM of the five largest axons (which aremotor axons) from each nerve (y axis) for the followinggenotypes (x axis): D42.1, D42 . CaMKIIT287D, DFakCG1,DFakCG1; D42 . CaMKIIT287D, DFakCG1; D42 . DFak1,and DFakCG1; D42 . PI3K-CAAX. Nerves from at leasttwo different larvae were scored for these axon measures.From left to right, n ¼ 6, 7, 5, 5, 7, and 5 nerves, respec-tively, for each genotype. One-way ANOVA and Fisher’sLSD were performed for statistical analysis.


might similarly activate PI3K by directly phosphorylat-ing and activating DFak. Alternatively, DFak might functionin a more indirect fashion, perhaps as a scaffold linkingCaMKII and PI3K in a signaling complex. This alternativepossibility would suggest that additional intermediateslinking CaMKII and PI3K activation exist but are currentlyunidentified.

The observation that DmGluRA-mediated activation ofPI3K requires CaMKII implies that DmGluRA activationincreases intracellular Ca21 levels in Drosophila motor nerveterminals as a necessary step in PI3K activation. The sourceof Ca21 for this activation is not known. However in mam-

mals, activation of group I mGluRs, which are responsiblefor mGluR-mediated LTD in the hippocampus and cerebel-lum, induce phospholipase C and IP3-mediated Ca21 tran-sients (Pin et al. 1994), which are essential intermediates incerebellar mGluR-mediated LTD (Ito 2001). Although theDrosophila DmGluRA is most similar to mammalian groupII mGluRs, which are not known to activate Ca21 transients,given that DmGluRA is the sole mGluR in Drosophila, itseems possible that DmGluRA might carry out many of thefunctions carried out by each of the three groups of mGluRsin mammals, as suggested previously (Pan et al. 2008). Al-ternatively, it is possible that DmGluRA activation might

Figure 6 CaMKII and DFak regulate motorneuron excitability. The larval neuromuscularpreparation (Jan and Jan 1976) was used for allrecordings. (A, left) Representative traces show-ing increased rate of onset of LTF in DFakCG1,and decreased rate of onset of LTF in larvaeoverexpressing CaMKIIT287D in motor neurons.Arrowheads indicate the increased and asyn-chronous EJPs, indicative of onset of LTF. Thebath solution contained 0.15 mM [Ca21] and100 mM quinidine. (Right) Geometric means 6SEM of the number of stimulations required toevoke LTF (y axis) at the indicated stimulus fre-quencies (x axis) for the following genotypes:D42.1, DFakCG1, D42 . CaMKIIT287D, andDFakCG1; D42 . DFak1. From top to bottom,n ¼ 7, 8, 18, and 6, respectively, for each ge-notype. One-way ANOVA and Fisher’s LSD gavethe following differences, at 10, 7, 5, and3 Hz, respectively: for D42.1 vs. DFakCG1,P ¼ 0.028, 0.022, 0.016, 0.127; vs. D42 .CaMKIIT287D, P ¼ 0.001, 0.006, 0.007, 0.011;vs. DFakCG1; D42 . DFak1, P ¼ 0.503, 0.876,0.859, 0.327. (B, left) Representative tracesshowing the increased excitatory junction po-tential (EJP) amplitude in DFakCG1 larvae andthe decreased EJP amplitude in larvae overex-pressing CaMKIIT287D in motor neurons, ata bath [Ca21] of 0.15 mM. (Right) Mean EJPamplitudes 6 SEM (y axis) at the indicated[Ca21] (x axis), from the following genotypes:D42.1, DFakCG1, D42 . CaMKIIT287D, andDFakCG1; D42 . DFak1. Larval nerves werestimulated at a frequency of 1 Hz, and 10responses were measured and averaged fromeach nerve. n ¼ (6, 6, 6), (14, 8, 9), (6, 6, 6),and (10, 6, 6), respectively, for each genotypeand [Ca21] (0.2, 0.15, and 0.1 mM). One-wayANOVA and Fisher’s LSD gave the followingdifferences at 0.2, 0.15, and 0.1 mM [Ca21],respectively: for D42.1 vs. DFakCG1, P ¼0.021, 0.012, 0.21; vs. D42 . CaMKIIT287D,P ¼ 0.001, 0.005, 0.484; vs. DFakCG1; D42 .DFak1, P ¼ 0.375, 0.941, 0.040. (C, left) Rep-

resentative traces showing mini EJPs (mEJP, marked by arrows) in D42.1, DFakCG1, D42 . CaMKIIT287D, and DFakCG1; D42 . DFak1. Two traces fromeach genotype are from two different larvae. The asterisks indicate EJPs evoked by single stimulations. (Right) Mean mEJP amplitudes 6 SEM (y axis) forthe indicated genotypes (x axis): D42.1, DFakCG1, D42 . CaMKIIT287D, and DFakCG1; D42 . DFak1. Recordings were collected from five differentlarvae from each genotype in a bath [Ca21] of 0.15 mM, and 30 responses were measured and averaged from each larva. One-way ANOVA and Fisher’sLSD gave the following differences: For D42.1 vs. DFakCG1, P ¼ 0.127; vs. D42 . CaMKIIT287D, P ¼ 0.265; vs. DFakCG1; D42 . DFak1, P ¼ 0.354. ForDFakCG1 vs. D42 . CaMKIIT287D, P ¼ 0.788; vs. DFakCG1; D42 . DFak1, P ¼ 0.332. For D42 . CaMKIIT287D vs. DFakCG1; D42 . DFak1, P ¼ 0.572. Pvalues ,0.05 were considered statistically significant.


increase intracellular Ca21 via the ryanodine receptor, whichwas previously shown to be an essential activator of CaMKIIin Drosophila larval motor nerve terminals (Shakiryanovaet al. 2007).

The ability of CaMKII to activate PI3K requires thenonreceptor tyrosine kinase DFak; the DFakCG1 null muta-tion completely blocks the ability of glutamate applied tomotor nerve terminals to activate PI3K, completely sup-presses the increase in basal p-Akt levels conferred byCaMKIIT287D, and blocks the ability of CaMKIIT287D to confertwo additional PI3K-dependent phenotypes: increased syn-apse number at the NMJ and increased motor axon diame-ter. These results identify DFak as an essential intermediatein PI3K activation by DmGluRA and CaMKII. However,DFakCG1 mutants fail to exhibit other phenotypes conferredby decreased PI3K activity: in an otherwise wild-type back-ground, DFakCG1 larvae exhibit only minor effects on NMJsynapse number or motor axon diameters, which are eachsignificantly decreased by decreased PI3K (Martin-Penaet al. 2006; Howlett et al. 2008). These results raise thepossibility that, whereas PI3K activation by DmGluRA andCaMKII is blocked in DFakCG1, total PI3K activity is notstrongly decreased because other significant routes to PI3Kactivation are DFak independent. Alternatively, DFak mightparticipate in signaling pathways distinct from the CaMKII-DFak-PI3K pathway we have identified, which would op-

pose the effects of PI3K on synapse number and axon di-ameter. In this view, CaMKII would preferentially promotethe ability of DFak to activate PI3K, rather than other DFak-dependent pathways.

In several subregions of the mammalian brain, ligandactivation of group I mGluRs induces LTD, a type of synapticplasticity. This induction both activates and requires theactivity of PI3K as well as the PI3K-activated kinase Tor(Hou and Klann 2004; Ronesi and Huber 2008). Severallines of evidence have led to the proposal that increasedsensitivity to mGluR-mediated LTD induction might underliespecific neurogenetic disorders. In particular, mice null forthe gene affected in fragile X, which is associated with anextremely high incidence of autism as well as other cognitivedeficits, exhibit increased sensitivity to mGluR-mediatedLTD induction in both the hippocampus (Huber et al.2002) and cerebellum (Koekkoek et al. 2005). Furthermore,the genes identified in two additional diseases associatedwith a high incidence of autism, neurofibromatosis (Nf1)and tuberous sclerosis (Tsc1 and Tsc2), each encode nega-tive regulators of the PI3K pathway: Nf1 encodes a RasGTPase activator protein (Xu et al. 1990), which inhibitsthe PI3K activator Ras (Rodriguez-Viciana et al. 1994),whereas the Tsc proteins are Tor inhibitors that are in turninhibited by PI3K activity (Gao et al. 2002; Inoki et al.2002). Thus loss of Nf1 or Tsc might also increase sensitivityto mGluR-mediated LTD. Finally, several lines of direct

Figure 7 The increased excitability conferred by DmGluRA112b is sup-pressed by expression of the activated CaMKIIT287D. Mean number ofstimulations (geometric means 6 SEM) required to induce LTF (y axis)at the indicated stimulus frequencies (x axis). The bath solution contained0.15 mM [Ca21] and 100 mM quinidine. From top to bottom (D42.1,DmGluRA112b; D42.1, DmGluRA112b, D42 . CaMKIIT287D, andDmGluRA112b; D42 . CaMKIIT287D), n ¼ 7, 8, 6, 18, and 18, respectively,for each genotype. One-way ANOVA and Fisher’s LSD gave the followingdifferences, at 10, 7, 5, and 3 Hz, respectively: For D42.1 vs.DmGluRA112b, P ¼ 0.001, ,0.001, ,0.001, ,0.001; vs. DmGluRA112b;D42.1, P ¼ 0.031, 0.007, 0.002, ,0.001; vs. D42 . CaMKIIT287D, P ¼0.001, 0.006, 0.007, 0.011; vs. DmGluRA112b; D42 . CaMKIIT287D, P ¼0.016, 0.013, 0.045, 0.011. For DmGluRA112b; D42.1 vs. DmGluRA112b,P ¼ 0.203, 0.256, 0.218, 0.650; vs. DmGluRA112b; D42 . CaMKIIT287D,P , 0.001, ,0.001, ,0.001, ,0.001. For DmGluRA112b; D42 .CaMKIIT287D vs. D42 . CaMKIIT287D, P ¼ 0.725, 0.745, 0.642, 0.897.P values ,0.05 were considered statistically significant.

Figure 8 A proposed mechanism for the DmGluRA-dependent activa-tion of PI3K via CaMKII and DFak. We suggest that glutamate releasedfrom motor nerve terminals as a consequence of neuronal activityactivates DmGluRA autoreceptors located in motor nerve terminals. ThisDmGluRA activation then activates CaMKII, which phosphorylates andactivates DFak. The formation of phosphotyrosylated DFak recruits andactivates PI3K, either via the p85 regulatory subunit or via Ras. The abilityof DmGluRA to activate CaMKII implies that DmGluRA activationincreases intracellular [Ca21], although this possibility has not beentested. As shown previously (Howlett et al. 2008), PI3K activationdecreases neuronal excitability by inhibiting the transcription factor Foxoand increases synapse number by activating S6 kinase.


http://flybase.org/reports/FBgn0015269.html

evidence suggest that PI3K hyperactivation plays a causalrole in autism (Serajee et al. 2003; Butler et al. 2005; Kwonet al. 2006; Mills et al. 2007). For example, DNA copy num-ber variants observed in individuals with autism but notunaffected individuals identify at high frequency PI3K sub-units or regulators, and each genetic change is predicted toelevate PI3K activity (Cusco et al. 2009). In addition, a trans-location that increases expression of the translation factoreIF-4E, which is known to be activated by the PI3K pathway(Miron et al. 2003; Puig et al. 2003), plays a direct, causalrole in autism (Neves-Pereira et al. 2009). The potential in-volvement of mGluR-mediated LTD in these neurogeneticdisorders increases interest in identifying the molecularintermediates that participate in this pathway, but theseintermediates are for the most part unidentified. Thus, thepossibility that CaMKII and Fak might participate in mGluR-mediated PI3K activation in mammals as well as Drosophilamight have significant medical interest.

Acknowledgments

We are grateful to James McNew and Daniel S. Wagner forcomments on the manuscript, Sally Leevers, Ruth Palmer,Leslie Griffith, Tom Schwarz, Marie-Laure Parmentier, andthe Drosophila stock center at Bloomington, IN for providingfly stocks. This work was funded by grants W81XWH-04-1-0272 and W81XWH-09-1-0106 from the Departmentof Defense Neurofibromatosis Research Program and grantIOS-0820660 from the National Science Foundation (toM.S.).

Literature Cited

Avraham, H., S. Y. Park, K. Schinkmann, and S. Avraham,2000 RAFTK/Pyk2-mediated cellular signalling. Cell Signal12: 123–133.

Bear, M. F., K. M. Huber, and S. T. Warren, 2004 The mGluRtheory of fragile X mental retardation. Trends Neurosci. 27:370–377.

Beumer, K., H. J. Matthies, A. Bradshaw, and K. Broadie, 2002 Inte-grins regulate DLG/FAS2 via a CaM kinase II-dependent pathwayto mediate synapse elaboration and stabilization during postem-bryonic development. Development 129: 3381–3391.

Bogdanik, L., R. Mohrmann, A. Ramaekers, J. Bockaert, Y. Grauet al., 2004 The Drosophila metabotropic glutamate receptorDmGluRA regulates activity-dependent synaptic facilitation andfine synaptic morphology. J. Neurosci. 24: 9105–9116.

Brand, A. H., and N. Perrimon, 1993 Targeted gene expression asa means of altering cell fates and generating dominant pheno-types. Development 118: 401–415.

Butler, M. G., M. J. Dasouki, X. P. Zhou, Z. Talebizadeh, M. Brownet al., 2005 Subset of individuals with autism spectrum disor-ders and extreme macrocephaly associated with germline PTENtumour suppressor gene mutations. J. Med. Genet. 42: 318–321.

Chen, H. C., P. A. Appeddu, H. Isoda, and J. L. Guan, 1996 Phos-phorylation of tyrosine 397 in focal adhesion kinase is requiredfor binding phosphatidylinositol 3-kinase. J. Biol. Chem. 271:26329–26334.

Chen, H. C., and J. L. Guan, 1994 Association of focal adhesionkinase with its potential substrate phosphatidylinositol 3-kinase.Proc. Natl. Acad. Sci. USA 91: 10148–10152.

Colombani, J., L. Bianchini, S. Layalle, E. Pondeville, C. Dauphin-Villemant et al., 2005 Antagonistic actions of ecdysone andinsulins determine final size in Drosophila. Science 310: 667–670.

Cusco, I., A. Medrano, B. Gener, M. Vilardell, F. Gallastegui et al.,2009 Autism-specific copy number variants further implicatethe phosphatidylinositol signaling pathway and the glutamater-gic synapse in the etiology of the disorder. Hum. Mol. Genet. 18:1795–1804.

Dasgupta, B., Y. Yi, D. Y. Chen, J. D. Weber, and D. H. Gutmann,2005 Proteomic analysis reveals hyperactivation of the mam-malian target of rapamycin pathway in neurofibromatosis 1-associated human and mouse brain tumors. Cancer Res. 65:2755–2760.

Della Rocca, G. J., T. van Biesen, Y. Daaka, D. K. Luttrell, L. M.Luttrell et al., 1997 Ras-dependent mitogen-activated proteinkinase activation by G protein-coupled receptors. Convergenceof Gi- and Gq-mediated pathways on calcium/calmodulin, Pyk2,and Src kinase. J. Biol. Chem. 272: 19125–19132.

Dikic, I., G. Tokiwa, S. Lev, S. A. Courtneidge, and J. Schlessinger,1996 A role for Pyk2 and Src in linking G-protein-coupledreceptors with MAP kinase activation. Nature 383: 547–550.

Dionne, M. S., L. N. Pham, M. Shirasu-Hiza, and D. S. Schneider,2006 Akt and FOXO dysregulation contribute to infection-induced wasting in Drosophila. Curr Biol 16: 1977–1985.

Fan, R. S., R. O. Jacamo, X. Jiang, J. Sinnett-Smith, and E. Rozengurt,2005 G protein-coupled receptor activation rapidly stimulatesfocal adhesion kinase phosphorylation at Ser-843. Mediation byCa21, calmodulin, and Ca21/calmodulin-dependent kinase II.J. Biol. Chem. 280: 24212–24220.

Fong, Y. L., W. L. Taylor, A. R. Means, and T. R. Soderling,1989 Studies of the regulatory mechanism of Ca21/calmodulin-dependent protein kinase II. Mutation of threonine 286 to alanineand aspartate. J. Biol. Chem. 264: 16759–16763.

Ganetzky, B., and C. F. Wu, 1982 Drosophila mutants with oppos-ing effects on nerve excitability: genetic and spatial interactionsin repetitive firing. J. Neurophysiol. 47: 501–514.

Gao, X., Y. Zhang, P. Arrazola, O. Hino, T. Kobayashi et al.,2002 Tsc tumour suppressor proteins antagonize amino-acid-TOR signalling. Nat. Cell. Biol. 4: 699–704.

Grabbe, C., C. G. Zervas, T. Hunter, N. H. Brown, and R. H. Palmer,2004 Focal adhesion kinase is not required for integrin func-tion or viability in Drosophila. Development 131: 5795–5805.

Griffith, L. C., J. Wang, Y. Zhong, C. F. Wu, and R. J. Greenspan,1994 Calcium/calmodulin-dependent protein kinase II andpotassium channel subunit eag similarly affect plasticity in Dro-sophila. Proc. Natl. Acad. Sci. USA 91: 10044–10048.

Guinebault, C., B. Payrastre, C. Racaud-Sultan, H. Mazarguil, M.Breton et al., 1995 Integrin-dependent translocation of phos-phoinositide 3-kinase to the cytoskeleton of thrombin-activatedplatelets involves specific interactions of p85 alpha with actinfilaments and focal adhesion kinase. J. Cell. Biol. 129: 831–842.

Heidinger, V., P. Manzerra, X. Q. Wang, U. Strasser, S. P. Yu et al.,2002 Metabotropic glutamate receptor 1-induced upregula-tion of NMDA receptor current: mediation through the Pyk2/Src-family kinase pathway in cortical neurons. J. Neurosci. 22:5452–5461.

Hou, L., and E. Klann, 2004 Activation of the phosphoinositide3-kinase-Akt-mammalian target of rapamycin signaling pathwayis required for metabotropic glutamate receptor-dependentlong-term depression. J. Neurosci. 24: 6352–6361.

Howlett, E., C. C. Lin, W. Lavery, and M. Stern, 2008 A PI3-kinase-mediated negative feedback regulates neuronal excit-ability. PLoS Genet. 4: e1000277.


Huang, Y., and M. Stern, 2002 In vivo properties of the Drosoph-ila inebriated-encoded neurotransmitter transporter. J. Neuro-sci. 22: 1698–1708.

Huber, K. M., S. M. Gallagher, S. T. Warren, and M. F. Bear,2002 Altered synaptic plasticity in a mouse model of fragileX mental retardation. Proc. Natl. Acad. Sci. USA 99: 7746–7750.

Inoki, K., Y. Li, T. Zhu, J. Wu, and K. L. Guan, 2002 TSC2 isphosphorylated and inhibited by Akt and suppresses mTOR sig-nalling. Nat. Cell. Biol. 4: 648–657.

Ito, M., 2001 Cerebellar long-term depression: characteriza-tion, signal transduction, and functional roles. Physiol. Rev.81: 1143–1195.

Jan, L. Y., and Y. N. Jan, 1976 Properties of the larval neuromus-cular junction in Drosophila melanogaster. J. Physiol. 262: 189–214.

Jan, Y. N., and L. Y. Jan, 1978 Genetic dissection of short-termand long-term facilitation at the Drosophila neuromuscularjunction. Proc. Natl. Acad. Sci. USA 75: 515–519.

Jin, P., L. C. Griffith, and R. K. Murphey, 1998 Presynaptic calcium/calmodulin-dependent protein kinase II regulates habituationof a simple reflex in adult Drosophila. J. Neurosci. 18: 8955–8964.

Joiner, M. L. A., and L. C. Griffith, 1997 CaM kinase II and visualinput modulate memory formation in the neuronal circuit con-trolling courtship conditioning. J. Neurosci. 17: 9384–9391.

Kelleher, R. J. 3rd, and M. F. Bear. 2008 The autistic neuron:troubled translation? Cell 135: 401–406.

Koekkoek, S. K., K. Yamaguchi, B. A. Milojkovic, B. R. Dortland, T.J. Ruigrok et al., 2005 Deletion of FMR1 in Purkinje cellsenhances parallel fiber LTD, enlarges spines, and attenuatescerebellar eyelid conditioning in Fragile X syndrome. Neuron47: 339–352.

Koh, Y. H., E. Popova, U. Thomas, L. C. Griffith, and V. Budnik,1999 Regulation of DLG localization at synapses by CaMKII-dependent phosphorylation. Cell 98: 353–363.

Kwon, C. H., B. W. Luikart, C. M. Powell, J. Zhou, S. A. Mathenyet al., 2006 Pten regulates neuronal arborization and socialinteraction in mice. Neuron 50: 377–388.

Lavery, W., V. Hall, J. C. Yager, A. Rottgers, M. C. Wells et al.,2007 Phosphatidylinositol 3-kinase and Akt nonautonomouslypromote perineurial glial growth in Drosophila peripheralnerves. J. Neurosci. 27: 279–288.

Leevers, S. J., D. Weinkove, L. K. MacDougall, E. Hafen, and M. D.Waterfield, 1996 The Drosophila phosphoinositide 3-kinaseDp110 promotes cell growth. EMBO J. 15: 6584–6594.

Luscher, C., and K. M. Huber, 2010 Group 1 mGluR-dependentsynaptic long-term depression: mechanisms and implications forcircuitry and disease. Neuron 65: 445–459.

Ma, S., Y. Yang, C. Wang, N. Hui, L. Gu et al., 2009 Endogenoushuman CaMKII inhibitory protein suppresses tumor growth byinducing cell cycle arrest and apoptosis through down-regulationof the phosphatidylinositide 3-kinase/Akt/HDM2 pathway. J.Biol. Chem. 284: 24773–24782.

Mallart, A., D. Angaut-Petit, C. Bourret-Poulain, and A. Ferrus,1991 Nerve terminal excitability and neuromuscular transmis-sion in T(X;Y)V7 and Shaker mutants of Drosophila mela-nogaster. J. Neurogenet. 7: 75–84.

Martin-Pena, A., A. Acebes, J. R. Rodriguez, A. Sorribes, G. G. dePolavieja et al., 2006 Age-independent synaptogenesis byphosphoinositide 3 kinase. J. Neurosci. 26: 10199–10208.

Mills, J. L., M. L. Hediger, C. A. Molloy, G. P. Chrousos, P. Manning-Courtney et al., 2007 Elevated levels of growth-related hor-mones in autism and autism spectrum disorder. Clin. Endocrinol.(Oxf) 67: 230–237.

Miron, M., P. Lasko, and N. Sonenberg, 2003 Signaling from Aktto FRAP/TOR targets both 4E-BP and S6K in Drosophila mela-nogaster. Mol. Cell. Biol. 23: 9117–9126.

Montiel, M., J. Quesada, and E. Jimenez, 2007 Activation ofcalcium-dependent kinases and epidermal growth factor recep-tor regulate muscarinic acetylcholine receptor-mediated MAPK/ERK activation in thyroid epithelial cells. Cell Signal 19: 2138–2146.

Neves-Pereira, M., B. Muller, D. Massie, J. H. Williams, P. C. O’Brienet al., 2009 Deregulation of EIF4E: a novel mechanism forautism. J. Med. Genet. 46: 759–765.

Palomero, T., M. L. Sulis, M. Cortina, P. J. Real, K. Barnes et al.,2007 Mutational loss of PTEN induces resistance to NOTCH1inhibition in T-cell leukemia. Nat. Med. 13: 1203–1210.

Pan, L., E. Woodruff 3rd, P. Liang, and K. Broadie, 2008 Mech-anistic relationships between Drosophila fragile X mental retar-dation protein and metabotropic glutamate receptor Asignaling. Mol. Cell. Neurosci. 37: 747–760.

Park, D., M. J. Coleman, J. J. Hodge, V. Budnik, and L. C. Griffith,2002 Regulation of neuronal excitability in Drosophila by con-stitutively active CaMKII. J. Neurobiol. 52: 24–42.

Parkes, T. L., A. J. Elia, D. Dickinson, A. J. Hilliker, J. P. Phillipset al., 1998 Extension of Drosophila lifespan by overexpressionof human SOD1 in motorneurons. Nat. Genet. 19: 171–174.

Pin, J. P., C. Joly, S. F. Heinemann, and J. Bockaert,1994 Domains involved in the specificity of G protein activa-tion in phospholipase C-coupled metabotropic glutamate recep-tors. EMBO J. 13: 342–348.

Poulain, C., A. Ferrus, and A. Mallart, 1994 Modulation of type AK1 current in Drosophila larval muscle by internal Ca21;effects of the overexpression of frequenin. Pflugers Arch. 427:71–79.

Puig, O., M. T. Marr, M. L. Ruhf, and R. Tjian, 2003 Control of cellnumber by Drosophila FOXO: downstream and feedback regu-lation of the insulin receptor pathway. Genes Dev. 17: 2006–2020.

Rocic, P., G. Govindarajan, A. Sabri, and P. A. Lucchesi, 2001 Arole for PYK2 in regulation of ERK1/2 MAP kinases and PI3-kinase by ANG II in vascular smooth muscle. Am. J. Physiol.Cell. Physiol. 280: C90–C99.

Rodriguez-Viciana, P., P. H. Warne, R. Dhand, B. Vanhaesebroeck,I. Gout et al., 1994 Phosphatidylinositol-3-OH kinase as a di-rect target of Ras. Nature 370: 527–532.

Ronesi, J. A., and K. M. Huber, 2008 Homer interactions are nec-essary for metabotropic glutamate receptor-induced long-termdepression and translational activation. J. Neurosci. 28: 543–547.

Schlaepfer, D. D., S. K. Hanks, T. Hunter, and P. van der Geer,1994 Integrin-mediated signal transduction linked to Raspathway by GRB2 binding to focal adhesion kinase. Nature372: 786–791.

Schweers, B. A., K. J. Walters, and M. Stern, 2002 The Drosophilamelanogaster translational repressor pumilio regulates neuronalexcitability. Genetics 161: 1177–1185.

Serajee, F. J., R. Nabi, H. Zhong, and A. H. Mahbubul Huq,2003 Association of INPP1, PIK3CG, and TSC2 gene variantswith autistic disorder: implications for phosphatidylinositol sig-nalling in autism. J. Med. Genet. 40: e119.

Shakiryanova, D., M. K. Klose, Y. Zhou, T. Gu, D. L. Deitcher et al.,2007 Presynaptic ryanodine receptor-activated calmodulin ki-nase II increases vesicle mobility and potentiates neuropeptiderelease. J. Neurosci. 27: 7799–7806.

Soltoff, S. P., 1998 Related adhesion focal tyrosine kinase andthe epidermal growth factor receptor mediate the stimulationof mitogen-activated protein kinase by the G-protein-coupledP2Y2 receptor. Phorbol ester or [Ca21]i elevation can sub-stitute for receptor activation. J. Biol. Chem. 273: 23110–23117.

Stern, M., and B. Ganetzky, 1989 Altered synaptic transmission inDrosophila hyperkinetic mutants. J. Neurogenet. 5: 215–228.


Stern, M., R. Kreber, and B. Ganetzky, 1990 Dosage effects ofa Drosophila sodium channel gene on behavior and axonal ex-citability. Genetics 124: 133–143.

Tsai, P. I., H. H. Kao, C. Grabbe, Y. T. Lee, A. Ghose et al., 2008 Fak56functions downstream of integrin alphaPS3betanu and sup-presses MAPK activation in neuromuscular junction growth.Neural Dev. 3: 26.

Ueda, A., C. Grabbe, J. Lee, J. Lee, R. H. Palmer et al., 2008 Mu-tation of Drosophila focal adhesion kinase induces bang-sensitivebehavior and disrupts glial function, axonal conduction and syn-aptic transmission. Eur. J. Neurosci. 27: 2860–2870.

Xu, G. F., B. Lin, K. Tanaka, D. Dunn, D. Wood et al., 1990 Thecatalytic domain of the neurofibromatosis type 1 gene productstimulates ras GTPase and complements ira mutants of S. cer-evisiae. Cell 63: 835–841.

Zwick, E., C. Wallasch, H. Daub, and A. Ullrich, 1999 Distinctcalcium-dependent pathways of epidermal growth factor recep-tor transactivation and PYK2 tyrosine phosphorylation in PC12cells. J. Biol. Chem. 274: 20989–20996.

Communicating editor: T. C. Kaufman


the metabotropic glutamate receptor activates the lipid kinase

Documents