INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Abl family nonreceptor tyrosine kinases regulate cellular morphogenesis in developing metazoan nervous systems. In Drosophila, the D-Abl kinase is required for normal axon fasciculation and pathfinding (Gertler et al. 1989; Giniger 1998; Wills et al. 1999a,b). In mice, the related Abl and Abl-related gene (Arg) kinases are required for proper morphogenesis of neuroepithelial cells during neurulation (Koleske et al. 1998). Abl family kinases regulate cell shape through functional interactions with proteins that control dynamic rearrangements of the actin cytoskeleton (Gertler et al. 1995; Liebl et al. 2000; Wills et al. 1999b; reviewed in Lanier and Gertler 2000). Abl and Arg also regulate the actin cytoskeleton directly via C-terminal binding domains for globular (G) and filamentous (F) actin. These actin-binding domains allow Abl and Arg to assemble F-actin into bundles in vitro (Van Etten et al. 1994; Wang et al. 2001) and to organize F-actin in vivo (Wang et al. 2001).

In addition to their roles in neuronal development, several lines of evidence suggest that Abl and Arg also contribute to synaptic function in mature neurons. Arg is most abundant in the brain, where it is concentrated in synapse-rich regions (Koleske et al. 1998). Although the brains of Arg-deficient mice appear grossly normal, these mice exhibit several behavioral abnormalities, suggesting that arg/ brains suffer from defects in synaptic function. Abl is also expressed in the adult mouse brain where it could overlap functionally with Arg, just as Abl and Arg exhibit overlapping functions in mouse development (Koleske et al. 1998).

Much recent evidence suggests that activity-dependent remodeling of actin in both the pre- and postsynaptic compartments plays an important role in synaptic plasticity (Dunaevsky et al. 2001; Fischer et al. 1998, 2000; Korkotian and Segal 2001; Star et al. 2002). Actin is the major cytoskeletal component of dendritic spines (Fifkova and Delay 1982; Matus et al. 1982) and is involved in the induction of stable long-term potentiation (LTP) in hippocampal slices (Kim and Lisman 1999; Krucker et al. 2000). Bath application of actin polymerization inhibitors increases paired-pulse facilitation (PPF) at SC-CA1 synapses of rat hippocampal slices (Kim and Lisman 1999), suggesting that dynamic rearrangements of the actin cytoskeleton are required for normal presynaptic function. Although the regulatory role played by F-actin in the presynaptic terminal remains unclear, it may involve modulation of Ca²⁺ entry through presynaptic calcium channels or the regulation of the presynaptic neurotransmitter release machinery (Furukawa et al. 1995; Morales et al. 2000). It is likely that actin can influence neurotransmitter release by acting as a substrate on which regulatory components of exo/endocytic pathways (e.g., dynamin, synaptojanin, and amphiphysin) carry out their functions (Cremona and De Camili 2001; Mundigl et al. 1998; Ochoa et al. 2000; Sakisaka et al. 1997). We hypothesized that Abl and Arg, by regulating the amount or structure of actin filaments in the nerve terminal, could contribute to the modulation of activity-dependent synaptic efficacy.

We describe here the localization and electrophysiological studies performed as a first step toward understanding a role for Abl and Arg in synaptic function. We show that both Abl and Arg localize to the presynaptic terminals and dendritic spines of neurons in the CA1 region of the hippocampus. We find that LTP and long-term depression (LTD) at SC-CA1 synapses are normal in abl/ and arg/ slices and in wild-type brain slices treated with the specific Abl family kinase inhibitor STI571. Interestingly, Abl and Arg are each required for normal PPF and appear to play nonoverlapping roles in the regulation of neurotransmitter release at excitatory SC-CA1 synapses of the hippocampus.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Immunoelectron microscopy

Immunolocalization was performed as previously described (Scheetz et al. 1997). Briefly, mice were killed with avertin and perfused transcardially with 0.1 M phosphate-buffered (pH 7.4) 4% paraformaldehyde/0.2% glutaraldehyde. After chilling to 4°C, the brains were removed from the cranium and postfixed with 0.1 M borate-buffered (pH 10.4) 4% paraformaldehyde overnight. One- to five-millimeter coronal sections of the hippocampus were cut and treated with 1% sodium borohydride for 45 min. Sections were rinsed repeatedly and cryoprotected with 20% glycerol and 15% sucrose. Sections were frozen on powdered dry ice. Samples were thawed and 60- to 100-µm-thick sagittal sections were cut using a vibratome. These sections were then treated with 10% normal serum in PBS for 1 h, followed by incubation in primary antibody. Sections were stained with antibodies to Arg (Koleske et al. 1998) and Abl (Calbiochem). Antibody/antigen interaction was detected by avidin/biotin complex coupled to HRP according to the manufacturer's instructions (Vector Labs). Sections were then processed for conventional electron microscopy. Contrast was enhanced by the application of 1% lead acetate to the Abl-stained wild-type and abl/ sections. Lead acetate was not used for any Arg-stained sections. Excitatory synapses were identified in electron micrographs based on their presence on mushroom-shaped spines with a clear postsynaptic density in apposition to a vesicle-filled presynaptic terminus.

Fluorescent microscopy

Wild-type, abl/, or arg/ mice were killed with halothane (Sigma) and perfused transcardially with 4% paraformaldehyde in PBS. The brains were removed and fixed overnight in 4% paraformaldehyde in PBS at 4°C. Brains were sagittally bisected and sagittal sections (50 µm) were cut with a vibratome. Antigen unmasking was performed by treating sections with 0.05% trypsin for 30 min at room temperature. After rinsing in PBS, sections were blocked with 2% bovine serum albumin (BSA) in PBS with 0.3% Triton X-100 (PBS-T+BSA) for 30 min. Staining for Arg was performed using purified rabbit antibodies against the SH3 and SH2 domains of Arg diluted in PBS-T+BSA for 2 h at room temperature. Staining for PSD-95 (Upstate Biotech) or synaptophysin (a gift from Pietro DeCamilli) was performed simultaneously. After rinsing in PBS-T, sections were incubated in secondary goat -rabbit biotin in PBS-T+BSA for 1 h. Arg was visualized using streptavidin-conjugated Alexa-488 (Molecular Probes), and PSD-95 or synaptophysin were visualized with Alexa-594-conjugated goat -mouse antibodies. Wild-type and arg/ sections were stained in parallel. Immunoperoxidase staining was performed similarly, but without antigen unmasking. Sections were processed using the Vectastain Elite Kit (Vector Labs) as directed by the manufacturer.

Hippocampal slice preparation

Transverse hippocampal slices (350 µm) were prepared from 4- to 6-week-old wild-type, abl/, or arg/ mice of a mixed 129Sv/J × C57Bl/6 background. All of our experiments use the abl^m2 mutant (Tybulewicz et al. 1991). Because it is a true Abl protein null, we refer to this allele as abl- throughout the present study. Each abl/ and arg/ mouse was matched with a wild-type littermate control. Briefly, mice were anesthetized with halothane (Sigma) and killed by decapitation. The brain was quickly removed and submerged in ice-cold sucrose-replaced artificial cerebrospinal fluid (ACSF) cutting solution (2 mM KCl, 2 mM MgCl₂, 1.25 mM NaH₂PO₄, 1 mM CaCl₂, 26 mM NaHCO₃, 10 mM dextrose, 248 mM sucrose). The hippocampus was dissected and 350-µm transverse slices were cut with a vibratome (Leica VT1000S). Slices from the middle third of the hippocampus were allowed to recover for a minimum of 1 h at room temperature in a submerged chamber containing ACSF (125 mM NaCl, 2.5 mM KCl, 1 mM MgCl₂, 1.25 mM NaH₂PO₄, 2 mM CaCl₂, 25 mM glucose, 26 mM NaHCO₃) bubbled with 95% O₂-5% CO₂.

Electrophysiology

Slices were transferred to a recording chamber, held submerged between two nylon nets, and constantly perfused at a rate of 2 ml/min with oxygenated ACSF containing 50 µM picrotoxin (Sigma). The recording chamber was mounted on an upright fixed-stage microscope (Olympus). A cut was made between CA1 and CA3 to prevent the propagation of epileptiform activity. Glass microelectrodes (0.5-1 M) filled with 2 M NaCl were positioned in stratum radiatum of CA1 (50-100 µm from the stratum pyramidale) to record evoked field potentials. Field potentials were amplified using a DP-301 differential amplifier (Warner Instrument), digitized at 10 kHz with pClamp 8.0 software (Axon Instruments), and analyzed using macros written using Igor Pro software. Schaffer collaterals were stimulated using bipolar tungsten electrodes (FHC) with enough current (50-µs pulses) to reliably elicit synaptic responses. Test stimuli were applied at low frequency (0.05 Hz) at a stimulus intensity that elicits a field excitatory postsynaptic potential (fEPSP) amplitude that was 33% of maximum. fEPSP magnitude was measured using the initial fEPSP slope. LTP was induced with two consecutive trains (1 s) of 100-Hz stimulation separated by 20 s. LTD was induced with 15 min of 1-Hz stimulation. PTP was induced with 1 s of 100-Hz stimulation. In all experiments, the presynaptic fiber volley was carefully monitored to ensure no change after tetanus; experiments in which the fiber volley changed were discarded. STI571 was resuspended in DMSO and bath applied at the indicated concentrations. Results are reported as mean ± SE. All electrophysiological experiments were performed and analyzed without knowledge of the experimenter of the genotype of the animals under investigation.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Localization of Abl and Arg in synapses

Immunoelectron microscopy established the presence of Abl and Arg in synapses of the mouse hippocampal CA1 area. Within synapses, staining for Abl and Arg was detected both in presynaptic terminals and in dendritic spines (Fig. 1, A-H). In control experiments, no Abl or Arg signal was detected in abl/ or arg/ hippocampal sections processed in parallel (data not shown). Staining for Abl and Arg in the presynaptic terminal was mostly limited to the contact area with the dendritic spine (Fig. 1, A, D, E, and H), although Abl was also detected throughout the axon terminal in a few synapses (Fig. 1C). In spines, Abl and Arg staining was prominent at the postsynaptic density (Fig. 1, A-C and E-G). We did not detect significant staining for Abl or Arg in axons, dendritic shafts, or spine necks. We also failed to detect Abl or Arg in glial cells.

View larger version (114K): [in this window] [in a new window]   Fig. 1. Subcellular localization of Abl and Arg in neurons. A-H: immunoelectron microscopic localization of Abl and Arg. Synapses in panels A-H are oriented in the same direction, with the presynaptic compartment toward the top of the figure and dendritic spines at the bottom. Synaptic contacts in panels A-H are delineated by arrowheads pointing toward the postsynaptic compartment. Excitatory synapses were identified based on their presence on mushroom-shaped spines with a clear PSD in apposition to a vesicle-filled presynaptic terminal. Immunopositive signal for Abl (A-D) and Arg (E-H) appears dark. Abl and Arg are localized to the cell membrane of presynaptic terminals (in panels A, C, D, E, and H). Abl and Arg are also found at the postsynaptic density of dendritic spines (between arrowheads, A-C and E-G). D and H show enlarged views of the synapses in A and E, respectively, with arrows pointing toward presynaptic staining. In A-D (all Abl-stained sections) the contrast was enhanced by staining with 1% lead acetate. Scale bar: 0.5 µm (A-C and E-G); 0.25 µm (D and H). I-P: confocal fluorescent microscopic images of Arg and PSD-95 (I-L) or Arg and synaptophysin (M-P) in stratum radiatum of the hippocampus from a wild-type mouse. Cell bodies appear dark at the top of the image. Confocal merged images of Arg and PSD-95 (K, enlarged view in L) or Arg and synaptophysin (O, enlarged in P) show colocalization of a proportion of Arg puncta with PSD-95 or synaptophysin. I and M: Arg in green. J: PSD-95 in red. N: synaptophysin in red. Scale bar: in I (applies to I-K and M-O), 48 µm; in L (applies to L and P), 15 µm. Q-S: conventional fluorescent microscopic images of Arg (Q) and GFAP (R) in stratum radiatum of the hippocampus from a wild-type mouse. Merged image (S) shows no colocalization of Arg and GFAP. Blue DAPI stain in S reveals nuclei. Scale bar in I (applies to Q-S), 121 µm. T: immunoperoxidase staining of Arg in sagittal brain sections from a wild-type (right) or arg/ (left) mouse. No Arg staining is present in the arg/ section. Scale bar, 1.7 mm.

By electron microscopy (EM), we did not detect Abl or Arg in every synapse (Table 1). However, light microscopic immunoperoxidase staining confirmed that Arg is present throughout the hippocampus (Fig. 1T) as we have reported previously (Koleske et al. 1998). This observation suggests that Arg is present in a larger proportion of CA1 synapses than is detectable by EM. Furthermore, visualization by confocal fluorescence microscopy demonstrated that most hippocampal neurons contain Arg protein, detected as tiny puncta densely occupying the neuropil but absent from dendritic shafts (Fig. 1, I and M). A proportion of these puncta colocalized with either PSD-95 (Fig. 1, J, K, and L) or synaptophysin (Fig. 1, N, O, and P), consistent with EM localization of Arg in both post- and presynaptic compartments, respectively. Arg never colocalized with GFAP, a marker for glia, ruling out the possibility that Arg is in nonneuronal cells (Fig. 1, Q, R, and S). We failed to obtain specific staining for Abl at the light microscopy level, despite repeated attempts. Our ability to detect Abl by EM, but not by light microscopy, may indicate that the Abl antigen is sensitive to the different fixation or processing conditions used for light microscopy on thin tissue sections. Together, our EM and light microscopy data suggest that Abl and Arg are present at a significant fraction of CA1 synapses.

Basal synaptic transmission is normal in abl/ånd arg/ mice

Having localized Abl and Arg to excitatory synapses in the hippocampus, we next examined whether Abl and Arg were required for normal basal synaptic transmission. We compared extracellular recordings of the SC-CA1 synapses in wild-type hippocampal slices to slices prepared from abl/ and arg/ mice. Stimulus-response curves obtained from abl/ or arg/ slices were not significantly different from wild-type (n = 16, 11, 16 slices, for abl/, arg/, and wild-type, respectively; P = 0.945, analysis of variance (ANOVA); Fig. 2A). Furthermore, the fEPSP slope corresponding to a given presynaptic fiber volley did not differ between abl/ or arg/ and wild-type slices (P = 0.159, Wilcoxon rank test; Fig. 2B).

View larger version (34K): [in this window] [in a new window]   Fig. 2. Basal synaptic transmission is normal in the absence of Abl and/or Arg function. A: stimulus-response curves of field excitatory postsynaptic potential (fEPSP) slope (mV/ms) vs. stimulus (mA) at the SC-CA1 synapses in hippocampal slices from wild-type (n = 16 slices), abl/ (n = 16 slices), and arg/ (n = 11 slices) mice. Data are presented as mean ± standard error (SE). B: plots of fEPSP slope (mV/ms) vs. presynaptic fiber volley amplitude (mV) from a random sample of slices from wild-type (n = 20 slices), abl/ (n = 20 slices), or arg/ (n = 17 slices) mice. Symbols for A and B: wild-type (), abl/ () and arg/ (). C: incubation in 0.2 µM STI571 for 30 min before recording had no effect on stimulus-response curves of fEPSP slope (mV/ms) vs. stimulus (mA) at the SC-CA1 synapses in hippocampal slices from wild-type and arg/ mice; wild-type (n = 16 slices); wild-type + STI571 (n = 16 slices); arg/ + STI571 (n = 14 slices). D: no differences were apparent in plots of fEPSP slope (mV/ms) vs. presynaptic fiber volley amplitude (mV) between wild-type (n = 20 slices), wild-type + STI571 (n = 20 slices), or arg/ + STI571 (n = 17 slices) slices. Symbols for C and D: wild-type (), wild-type + STI571 (), arg/ + STI571 (), and abl/ + STI571 (). E: wild-type slices (n = 16 slices) were stimulated at a stimulus intensity that produced 50% of the maximal fEPSP for 15 min. STI571 was then introduced into the perfusion buffer [artificial cerebrospinal fluid (ACSF)] at a concentration of 0.2 µM and responses were recorded for 1 h. No change in the fEPSP was observed. Error bars correspond to SE.

Mice lacking Abl or Arg throughout development may regulate the activity or expression of other kinases or downstream mediators to compensate for the loss of Abl or Arg. We used the Abl/Arg inhibitor STI571 to determine the effects of acute inhibition of Abl and Arg activity on basal synaptic transmission. Bath application of 0.2 or 1 µM STI571 to wild-type slices did not affect the fEPSP slope, even after 1 h in the perfusion buffer (n = 16; Fig. 2E). Previous studies have shown that these concentrations are adequate to inhibit Abl and Arg kinase activity in cells (Buchdunger et al. 1996; Okuda et al. 2001). Stimulus-response curves and plots of presynaptic fiber volley amplitudes versus fEPSP slopes were also unaffected by STI571 application (P = 0.882, ANOVA, Fig. 2C; P = 0.507, Wilcoxon rank test, Fig. 2D).

Reduced paired-pulse facilitation in abl/ and arg/ hippocampal slices

The localization of Abl to presynaptic terminals led us to evaluate whether normal presynaptic function requires Abl or Arg function. We first examined PPF by measuring fEPSP responses to two stimuli delivered at short interstimulus intervals to the Schaffer collateral inputs. PPF is a transient form of presynaptic plasticity, where the response to the second stimulus is enhanced due to residual Ca²⁺ in the presynaptic terminal following the first stimulus. Shown in Fig. 3A, at five different interstimulus intervals ranging from 50 to 250 ms, a reduction of the mean paired-pulse facilitation ratio (second fEPSP slope/first fEPSP slope) was detected in slices from both abl/ and arg/ mice as compared with wild-type (9 slices, 3 wild-type control mice; 16 slices, 5 arg/ mice; 15 slices, 5 abl/ mice). The differences between genotype groups (wild-type vs. abl/ and wild-type vs. arg/) were statistically significant (P < 0.02, ANOVA). We also detected a similar reduction in the PPF ratio in wild-type slices treated with 0.2 µM STI571 (11 slices, control; 20 slices treated with 0.2 µM STI571; P < 0.04, ANOVA; Fig. 3B). Post-hoc Student's t-test reveal that the differences were significant at time intervals from 100 to 250 ms (P < 0.01). These results indicate that both Abl and Arg functionally modulate PPF, since deletion of either Abl or Arg, or inhibition of both kinases with STI571, results in a similar reduction in PPF.

View larger version (29K): [in this window] [in a new window]   Fig. 3. Analysis of short-term synaptic plasticity in abl/ and arg/ slices. A: comparison of paired-pulse facilitation (PPF) in wild-type, abl/, and arg/ slices. Paired-pulse ratio measured at different interstimulus intervals was significantly decreased in slices from both abl/ (n = 15 slices) and arg/ (n = 16 slices) compared with wild-type (n = 9 slices) mice. Symbols: wild-type (), abl/ (), and arg/ (). Data are presented as mean ± SE. B: comparison of PPF in wild-type, arg/, or abl/ slices treated with STI571 (0.2 µM). Paired-pulse ratio was significantly decreased in wild-type slices treated with STI571; wild-type (n = 11slices); wild-type + STI571 (n = 20 slices). STI571 does not further decrease PPF when applied to arg/ or abl/ slices; arg/ + STI571 (n = 19 slices); abl/ + STI571 (n = 9 slices). Symbols: wild-type (), wild-type + STI571 (), arg/ + STI571 (), and abl/ + STI571 (). C: posttetanic potentiation (PTP) measured after a 1-s, 100-Hz tetanus in the presence of 50 µM D,L-APV is not altered in abl/ (n = 8 slices) or arg/ (n = 9 slices) slices compared with wild-type (n = 4 slices). Symbols: wild-type (), abl/ (), and arg/ (). Data are presented as mean ± SE. D, E: frequency-dependent facilitation and PTP in wild-type, abl/, and arg/ slices. Ratio of responses compared with the slope of the first fEPSP of a short stimulus train (7 stimuli of 50-µs duration at 40 Hz), followed by a single test stimulus delivered 300 ms after the burst. Responses of abl/ (n = 8 slices), arg/ (n = 7 slices), and wild-type (n = 7 slices) slices were examined in normal (2 mM, D) and low (1.3 mM, E) extracellular Ca2+. Symbols: wild-type (), abl/ (), and arg/ ().

Although STI571 has been shown to be a potent and specific inhibitor for Abl and Arg, it is possible that the decreased level of PPF might be due to inhibition of other kinases. In particular, the platelet-derived growth factor (PDGF) receptor is inhibited by STI571 with an IC₅₀ of 0.3 µM (Buchdunger et al. 1995, 1996; Druker et al. 1996). To rule out the possibility that inhibition of other kinases by STI571 was causing the decreased PPF, we applied STI571 to abl/ slices or arg/ slices and measured PPF. STI571 led to no further significant decrease in PPF when applied to abl/ or arg/ slices (19 arg/ slices; 9 abl/ slices; P = 0.217, ANOVA; Fig. 3B), demonstrating that the inhibitory effect of STI571 requires Abl or Arg function. These data strongly suggest that STI571 inhibits PPF by inhibiting Abl and/or Arg kinase activity.

As a second measure of presynaptic plasticity, we examined posttetanic potentiation (PTP) in abl/ or arg/ slices. PTP was elicited by applying a high-frequency tetanus (1 s at 100 Hz), resulting in an elevation of presynaptic Ca²⁺ and short-term enhancement of transmission due to mobilization of the reserve pool of synaptic vesicles (Zucker 1989). In the presence of D,L-2-amino-5-phosphonovaleric acid (D,L-APV, 50 µM) [an N-methyl-D-aspartate (NMDA) receptor blocker], a 100-Hz (1 s) tetanus resulted in an enhancement of the fEPSP slope in wild-type hippocampal slices, which decayed to baseline within 5 min. We observed no significant differences in the peak PTP (wild-type control 148 ± 12%, 4 slices, 2 mice; abl/ 150 ± 10%, 8 slices, 3 mice; arg/ 150 ± 8%, 9 slices, 3 mice; Student's t-test, P > 0.1) or in the time course of decay to baseline levels in abl/ or arg/ slices compared with wild-type (Fig. 3C). In addition, the field potential responses of abl/, arg/, and wild-type slices during the tetanus were qualitatively similar (data not shown). Thus the ability of the SC-CA1 synapses to respond to high-frequency stimulation is not affected by loss of Abl or Arg function.

Probability of release is altered in abl/ mice

To examine whether the probability of release is altered in abl/ or arg/ mice, we compared responses to a burst stimulation paradigm under conditions of normal (2 mM) and low (1.3 mM) extracellular Ca²⁺. The Schaffer collaterals were stimulated by a short 40-Hz train (7 stimuli), followed by a test stimulus delivered 300 ms after the end of the burst. Consistent with the findings above, in normal extracellular Ca²⁺, facilitation at the later time points during the burst (stimuli 4, 5, 6, 7) was suppressed in the abl/ or arg/ slices (n = 8, 7 slices for abl/, arg/; ANOVA, P = 0.006 and 0.04, respectively; Fig. 3D) compared with wild-type slices (n = 7 slices), while the PTP 300 ms after the burst was not significantly different (ANOVA, P = 0.82 and 0.56 abl/ or arg/ compared with wild-type, respectively). These results suggest that Abl and Arg may regulate the availability of vesicles from the readily releasable pool during repetitive stimulation.

These results support the hypothesis that Abl and Arg contribute to maintenance of the vesicle pool during repetitive stimulation, but do not rule out the possible function of the kinases in other synaptic events. For example, abl/ and arg/ mice may have more rapid inactivation of presynaptic Ca²⁺ channels or premature postsynaptic receptor desensitization, each of which could result in decreased facilitation at late time points in the burst.

In the presence of low extracellular Ca²⁺, facilitation within the burst was markedly enhanced in the wild-type and abl/ slices compared with responses in normal extracellular Ca²⁺ (n = 8, 7 slices for ab//, wild type, ANOVA, P = 0.016 and <0.0001, respectively; Fig. 3E). Lowering extracellular Ca²⁺ reduces the number of vesicles released in response to the first stimulus, leaving more vesicles available for subsequent stimuli. Lowering extracellular Ca²⁺ should therefore increase facilitation. The facilitation of responses within the burst observed in the abl/ slices did not differ from the wild-type (ANOVA, P = 0.137), strongly suggesting that Abl normally functions to conserve presynaptic vesicle stores to prevent a decrease in postsynaptic responses. In contrast, arg/ slices unexpectedly showed a slight decrease in facilitation in low Ca²⁺ when compared to slices recorded in normal extracellular Ca²⁺ (n = 7, ANOVA, P = 0.281 statistics; Fig. 3E). These data suggest that Arg normally maintains the vesicle supply during repetitive activation. The absence of Arg to maintain vesicle supply, together with decreased probability of release in low extracellular Ca²⁺, could result in decreased facilitation in arg/ slices in low compared with normal Ca²⁺ conditions. Interestingly, these results suggest that Abl and Arg have different functions in regulating neurotransmitter vesicle availability during repetitive synaptic activity.

Long-term potentiation and long-term depression are normal in abl/ and arg/ mice

In addition to Abl and Arg, several nonreceptor tyrosine kinases, including Src (Lu et al. 1998), Fyn (Grant et al. 1992), and Lyn (Hayashi et al. 1999), are highly expressed in synapses where they regulate neurotransmission and plasticity (Rostas et al. 1996; Yu et al. 1997; Yu and Salter 1999; Xiong et al. 1999). Tyrosine kinases have been implicated in the postsynaptic mechanisms underlying LTP and LTD, respectively, because broad-specificity tyrosine kinase inhibitors block the induction of LTP and LTD (Boxall et al. 1996; O'Dell et al. 1991). To examine the possible role of Abl family kinases in long-lasting forms of synaptic plasticity, we examined LTP at the SC-CA1 synapses in abl/ and arg/ hippocampal slices. LTP was induced by two 100-Hz trains of stimuli separated by 20 s delivered to the Schaffer collateral inputs. A previous study reported no change in LTP in mice homozygous for the abl^m1 allele (Grant et al. 1992). Unlike the abl/mice we used in our study, which do not express any portion of the Abl protein, the abl^m1 mutant expresses the amino-terminal half of Abl, which retains kinase activity (Schwartzberg et al. 1991). Nonetheless, consistent with the previous report, we observed no difference in LTP between abl/, arg/, or wild-type slices. The difference in the percentage potentiation observed in wild-type and mutant slices 30 min after tetanic stimulation was not significantly different (repeated measure ANOVA, P = 0.18; wild-type control 153 ± 4%, 13 slices; arg/ 152 ± 5%, 9 slices; abl/ 152 ± 3%, 13 slices; Fig. 4, A and B). Furthermore, acute inhibition of Abl and Arg by bath application of STI571 to wild-type slices had no effect on the induction or maintenance of LTP after 30 min (repeated measure ANOVA, P = 0.12; wild-type control 153 ± 4%, 13 slices; wild-type + STI571 147 ± 3%, 11 slices; Fig. 4C). These results indicate that Abl and Arg are not required for this form of hippocampal LTP. We did not examine whether other forms of LTP, such as those induced by theta stimulation or non-NMDA receptor-dependent LTP in the CA3 region, are equally unaffected by the absence of Abl or Arg.

View larger version (16K): [in this window] [in a new window]   Fig. 4. Abl and Arg are not required for long-term potentiation (LTP). A, B: LTP was induced by two 100-Hz trains for 1 s separated by 20 s. No significant difference was observed between LTP in wild-type (n = 13 slices), arg/ (n = 9 slices), or abl/ (n = 13 slices) slices. C: STI571 (0.2 µM) does not affect LTP in wild-type slices; wild-type (n = 13 slices), wild-type slices + STI571 (n = 11 slices). Symbols: wild-type (), arg/ (), abl/ (), and wild-type + STI571 (). Values are presented as a percentage of baseline (mean ± SE). A1, B1, C1: sample traces before and after LTP. Scale bars, 0.25 mV, 5 ms.

We also examined LTD in abl/ and arg/ slices. LTD was induced by applying 900 stimuli at 1 Hz to SC inputs. Both abl/ and arg/ slices showed similar levels of LTD to wild-type 30 min after the end of 1-Hz stimulation (repeated measure ANOVA, P = 0.11; wild-type control 77 ± 2%, 11 slices; arg/ 81 ± 2%, 6 slices; abl/ 77 ± 2%, 5 slices; Fig. 5, A and B). STI571-treated wild-type and arg/ slices also had normal levels of LTD (wild-type + STI571, 81 ± 2%, 11 slices; Fig. 5C; arg/ + STI571, 78 ± 3%, 4 slices; repeated measure ANOVA, P = 0.12; data not shown). Abl and Arg are therefore not required for this form of hippocampal LTD. Although this form of LTD is the most commonly studied, we did not examine other forms of LTD, such as metabotropic glutamate receptor-dependent LTD or heterosynaptic LTD, in these mutants.

View larger version (17K): [in this window] [in a new window]   Fig. 5. Abl and Arg are not required for long-term depression (LTD). A, B: LTD was induced by 900 stimuli delivered at 1 Hz. No significant difference was observed between LTD in wild-type (n = 11 slices), arg/ (n = 6 slices), or abl/ (n = 5 slices) slices. C: STI571 (0.2 µM) does not affect LTD in wild-type slices; wild-type (n = 11 slices), wild-type + STI571 (n = 11 slices). Symbols: wild-type (), arg/ (), abl/ (), and wild-type + STI571 (). Values are presented as a percentage of baseline (mean ± SE). A1, B1, C1: sample traces before and after LTD. Scale bars, 0.25 mV, 5 ms.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

We report here the presence of Abl and Arg in a large subset of excitatory synapses in the adult mouse hippocampus, where they modulate paired-pulse facilitation. These observations demonstrate a role for Abl and Arg in regulating synaptic function.

We find that Abl and Arg localize near the active zone in the presynaptic compartment. Similar localization at the active zone has been demonstrated for other proteins, including Bassoon (tom Dieck et al. 1998) and Rim (Wang et al. 1997; Koushika et al. 2001) that help align synaptic vesicles at their appropriate positions in the presynaptic compartment. The similar localization of Abl and Arg is consistent with a role for these kinases in modulating presynaptic function.

Paired-pulse stimulation is routinely used as an indirect method to study the presynaptic control of neurotransmitter release. The facilitation of the response to the second stimulus reflects a transient change in the probability of release due to the increased residual Ca²⁺ remaining in the presynaptic terminal from the first stimulus (Manabe et al. 1993; Schulz et al. 1994; Wu and Saggau 1994). Our observation of Abl and Arg near the active zone, together with the reduction of PPF in abl/ and arg/ mice, supports the hypothesis that Abl and Arg regulate neurotransmitter release. Our data are consistent with a model in which Abl limits neurotransmitter release from the presynaptic terminal. Indeed, we find that in low extracellular Ca²⁺ the frequency-dependent facilitation in abl/ slices is similar to wild-type. That is, for wild-type and abl/ slices, lowering extracellular Ca²⁺ reduces the number of vesicles released in response to the first stimulus, thereby augmenting responses to the second or subsequent stimuli within the burst. The depressed short-term plasticity observed in the abl/ mice could have resulted from a rapid depletion of neurotransmitter vesicles as a consequence of the absence of Abl's inhibitory action on neurotransmitter release in the presynaptic terminal. Abl may act presynaptically to maintain vesicle availability for release during repetitive stimulation. We cannot rule out the possible function of the kinases in other synaptic events which could give rise to similar results. For example, abl/ and arg/ mice may have more rapid inactivation of presynaptic Ca²⁺ channels or premature postsynaptic receptor desensitization, each of which could result in decreased facilitation at late time points in the burst.

Another interesting finding, that the short-term plasticity of arg/ slices was not increased but rather somewhat further decreased in low extracellular Ca²⁺, points to an important difference in the role of Abl and Arg in the regulation of the probability of neurotransmitter release from the presynaptic terminal. Arg may help to maintain the vesicle supply during repetitive activation. Therefore in normal Ca²⁺ the absence of Arg would result in decreased facilitation. In low Ca²⁺, the decreased probability of release, combined with a defect in vesicle availability, could result in a further reduction in facilitation from levels observed in normal Ca²⁺.

Our pharmacological experiments with STI571 strongly suggest that the kinase activity of Abl and/or Arg is directly required in PPF. Based on kinetic studies of Abl and Arg (Brasher and VanEtten 2000; K. Tanis and A. J. Koleske, unpublished data), it is unlikely that a phosphorylation event by Abl or Arg could be achieved in the short interstimulus intervals (50-250 ms) of our PPF experiments. A more likely possibility is that Abl and Arg kinase activity is required to maintain the presynaptic release machinery or synaptic vesicle pool in an optimal condition to allow for normal levels of PPF. Preincubating slices in STI571 for 30 min might erode this situation, leading to diminished levels of PPF.

Although our data support a presynaptic localization and function for Abl and Arg, we also find that Abl and Arg localize to the postsynaptic density in dendritic spines where they may relay signals from synaptic adhesion receptors to mediate cytoskeletal changes in the spine. Adhesion molecules such as NCAM and L1 are known to directly couple the actin cytoskeletons of adjacent neurons. This pre- to postsynaptic adhesion is essential for the maintenance, stabilization, and physiology of the synapse. Abl kinase activity is induced by integrin receptor engagement (Lewis et al. 1996) and once activated, Abl can direct dynamic rearrangements of the actin cytoskeleton (Plattner et al. 1999; Salgia et al. 1997). Similarly, the engagement of postsynaptic adhesion receptors may stimulate Abl and Arg to direct rearrangements of the postsynaptic actin cytoskeleton. A rearrangement of the spine may direct a complementary adjustment in presynaptic terminal structure, because the cytoskeletons of these two compartments are directly coupled through adhesion receptors. The identification and characterization of synaptic substrates of Abl and Arg should help to reveal how these kinases contribute to postsynaptic function.

We and others have previously shown that singly mutant abl/ or arg/ mice can live to adulthood, although abl/arg/ double-mutant mice die as embryos (Koleske et al. 1998; Schwartzberg et al. 1991; Tybulewicz et al. 1991). Thus either Abl or Arg is required for mice to mature to adulthood. These observations indicated that the closely related Abl and Arg kinases compensate functionally for each other during mouse development and suggested that the kinases might have overlapping cellular functions. In contrast, our present data indicate that both Abl and Arg are required for normal PPF. abl/ or arg/ slices, or wild-type slices treated with STI571, each exhibit similar reductions in PPF. Moreover, STI571 treatment leads to no further reduction in PPF when applied to either abl/ or arg/ slices. We also show that lowering extracellular Ca²⁺ affects frequency-dependent facilitation differently in abl/ and arg/ slices. Together, these observations demonstrate that Abl and Arg cannot functionally compensate for each other in PPF and suggest that Abl and Arg act via distinct mechanisms to promote the availability of synaptic vesicles at the active zone during conditions of repetitive stimulation.

Long-lasting forms of synaptic plasticity, such as LTP and LTD, are believed to be critical for learning and memory in the mature CNS (for review, see Martin et al. 2000; Tsien 2000). Less is known about the function of short-term forms of synaptic plasticity, such as PPF and PTP, in higher order processes such as learning and memory. Recent behavioral experiments point to a possible role for short-term plasticity in learning. In one example, analogous to the results from abl/ and arg/ mice reported here, brain slices from mice heterozygous for a mutation in -calcium/calmodulin kinase II (-CaMKII) exhibit reduced PPF, although LTP and LTD are normal. These CaMKII+/ mice exhibit a profound learning impairment in a range of behavioral tests of hippocampal function (Chapman et al. 1995; Silva et al. 1992; Wang and Kelly 1996). At least three other mouse mutants (synapsin II knockouts, synapsin I/II double knockouts, and mGluR4 knockouts) with reduced PPF or PTP exhibit deficits in learning tasks despite having apparently normal long-lasting synaptic plasticity (Pekhletski et al. 1996; Rosahl et al. 1995). These observations suggest that short-term plasticity may play a role in behavioral learning and memory tasks. It is possible that the deficits in PPF we observed in the present study contribute to the behavioral abnormalities (decreased motor skills, decreased mating and aggression, sensorineural deafness) previously observed in arg/ mice (Koleske et al. 1998).