Dissecting the Roles of Kalirin-7/PSD-95/GluN2B Interactions in Different Forms of Synaptic Plasticity

Background: Kalirin-7 (Kal7) is a multidomain scaffold and guanine nucleotide exchange factor localized to the postsynaptic density, where Kal7 is crucial for synaptic plasticity. Kal7 knockout mice exhibit marked suppression of long-term potentiation and long-term depression in hippocampus, cerebral cortex and spinal cord, with depressed surface expression of GluN2B receptor subunits and dramatically blunted perception of pain. Kal7 knockout animals show exaggerated locomotor responses to psychostimulants and self-administer cocaine more enthusiastically than wildtype mice. Results: To explore the underlying cellular and molecular mechanisms which are deranged by loss of Kal7, we infused candidate intracellular interfering peptides to acutely challenge the synaptic function(s) of Kal7 with potential protein binding partners, to determine if plasticity decits in Kal7 -/- mice are the product of developmental processes since conception, or could be produced on a much shorter time scale. We demonstrated that these small intracellular peptides disrupted normal long-term potentiation and long-term depression, strongly suggesting that maintenance of established interactions of Kal7 with PSD-95 and/or GluN2B is crucial to synaptic plasticity. Conclusions: Blockade of the Kal7-GluN2B interaction was most effective at blocking long-term potentiation, but had no effect on long-term depression. Biochemical approaches indicated that Kal7 interacted with PSD-95 at multiple sites within Kal7. t-test for post-hoc comparison. p < 0.05 was taken as a statistically signicant effect and marked with an asterisk (*). Long term potentiation and depression are always compared to untreated cells (no peptide) subjected to the same electrical stimulation pattern. All data supporting the results in this paper are presented in this paper; additional details are available from eslevine@uchc.edu, eipper@uchc.edu, or mains@uchc.edu.

Sec14 domain, spectrin repeat region and C-terminal PDZ binding motif of Kal7 play essential roles. In Kal7 knockout mice, hippocampal CA1 pyramidal neurons exhibit striking de cits in NMDA receptormediated currents, as indicated by a decrease in the NMDA/AMPA ratio (Ma, Kiraly, et al., 2008).
Attenuation of NMDA receptor-dependent currents in Kal7 knockout mice is speci c to GluN2B-containing receptors .
Since the speci c sites through which Kal7 interacts with PSD-95 and with the GluN2B subunit of the NMDA receptor have been mapped , we carried out a preliminary exploratory investigation to assess the ability of synthetic Kal7 and GluN2B peptides to disrupt synaptic plasticity evoked by different stimulation paradigms. Our earlier eld potential recordings of hippocampal CA1 pyramidal neurons demonstrated an essential role for Kal7 in LTP produced by TBS (but not by HFS) and in LTD produced by LFS . Although the stimulation paradigm needed to induce LTP in spinal projection neurons is distinctly different, inclusion of a peptide identical to the C-terminus of Kal7 in the recording patch pipette in hippocampal neurons blocked LTP (Fig.1B,1C), mimicking the electrophysiological response observed in Kal7 knockout spinal neurons (Lu et al., 2015). It is intriguing that a very early cleavage of Kal7 by calpain produces a peptide virtually identical to one of the peptides tested (Kal7-L; Table 1) (Miller, Yan, Machida, et al., 2017). Intracellular calpain cleavages are established as critical for long-term potentiation (Briz & Baudry, 2016).

Results
In our previous electrophysiological analyses of CA1 pyramidal neurons, we used eld potential recordings in acute slices to demonstrate an important role for Kal7 in the LTP that occurs in response to theta burst stimulation (TBS) and in the LTD that occurs in response to prolonged 1 Hz stimulation (low frequency stimulation, LFS). Strikingly, the absence of Kal7 did not diminish the NMDA receptorindependent LTP observed in response to 200 Hz/2sec tetanic stimulation in the presence of CPP Lemtiri-Chlieh et al., 2011;Ma, Kiraly, et al., 2008). Pharmacological and biochemical studies lent support to the conclusion that Kal7, which interacts with both PSD-95 and GluN2B, plays a unique role in GluN2B-dependent signaling Lemtiri-Chlieh et al., 2011;Ma, Kiraly, et al., 2008;Penzes et al., 2001).
Synaptic plasticity de cits observed in CA1 pyramidal neurons lacking Kal7 using patch clamp electrophysiology With the goal of acutely disrupting the interactions of Kal7 with PSD-95 or with GluN2B by injecting synthetic peptides into wild type neurons, we rst veri ed that whole cell patch clamp electrophysiology revealed similar de cits in LTP and LTD in CA1 pyramidal neurons in slices prepared from Kal7 knockout mice (Fig. 2). Two different induction protocols were used, both under voltage clamp conditions: (1) 7 train TBS and (2) HFS (two 100 Hz bursts for 2 seconds, separated by 20 seconds). For the 7 train TBS, each TBS train contained 10 bursts (200 ms interburst interval), and each burst consisted of 5 stimuli at 100 Hz. Each train was delivered with a 5 s intertrain interval. As observed using eld potential recordings , whole cell patch clamp recordings revealed the inability of Kal7 knockout CA1 pyramidal neurons to exhibit LTP in response to administration of 7 train TBS ( Figs Student's t-test, p = 0.02) and HFS (n = 5 cells; Baseline 341 pA ± 40; 35 -40 min post-HFS 544 pA ± 93; Student's t-test, p = 0.04) stimulation paradigms. Bath perfusion of CPP (3 µM), throughout the entire experiment, blocked LTP in cells subjected to 7 train TBS (Fig.2B). Whole cell patch recordings from Kal7 knockout neurons exposed to LFS (1 Hz) revealed their failure to exhibit LTD (Fig. 2E), as observed using eld recordings Lemtiri-Chlieh et al., 2011). These data justify the use of whole cell patch clamp to deliver synthetic peptides designed to disrupt acutely the interactions of Kal7 with PSD-95 or GluN2B, and then assess the effects of these peptides on the e cacy of speci c plasticity induction paradigms.
Our goal was to determine whether acute intracellular administration of peptides designed to mimic the C-terminal PDZ-binding motifs of Kal7 (Kal7-C) and GluN2B (GluN2B-C) and the PH-domain of Kal7, which interacts with the juxtamembrane domain of the GluN2B subunit (Kal7-GluN2B), had any effect on three different types of synaptic plasticity. Peptides diluted into intracellular recording solution were gradually infused into CA1 hippocampal pyramidal neurons using the whole cell patch clamp electrode.
Using a FITC-tagged Kal7-PDZ peptide, we found that signi cant lling of the patched CA1 hippocampal neuron occurred within 10 min of breaking into the cell (Figs. 1B,1C).
Control cells where no peptide was added to the intracellular recording solution showed robust LTP following induction with 7 train TBS (Fig.3A). We rst asked whether acute disruption of the Kal7-PSD-95 interaction blocked LTP induced by 7 train TBS stimulation (Fig.3A). When used at a concentration of 5 or 10 µM, Kal7-C eliminated TBS-induced LTP equally well. Infusion of 1 µM Kal7-C was without effect on TBS-induced LTP.
Control peptides do not interfere with LTP As noted above, a lot of proteins in cells have PDZ binding motifs. To test for speci city and toxicity, the C-terminal Valine residue of Kal7-C was replaced by Aspartic Acid (STYV → STYD) (Kal7-C mut ), destroying the PDZ-binding motif (Songyang et al., 1997). Following the infusion of Kal7-C mut , 7 train TBS elicited LTP that resembled the LTP observed in control cells at 35 -40 minutes post-induction at a concentration of either 1 µM (p = 0.01) or 10 µM (Fig.3B). The magnitude of LTP elicited in the presence of Kal7-C mut at 10 µM was not signi cantly different than LTP observed in control cells, despite cells lled with Kal7-C mut (10 µM) showing lower LTP. However, cells lled with Kal7-C mut (10 µM) exhibited LTP that was signi cantly greater than cells lled with the Kal7-C (10 µM) peptide. The effect of this single amino acid change supports the speci city of the Kal7-C peptide effect.
Although mice lacking Kal7 throughout gestation and maturation fail to exhibit LTP in response to TBS, HFS stimulation in Kal7 knockout mice produces LTP that is equivalent to that observed in wildtype mice . We next asked whether the acute presence of Kal7-C peptide affected LTP in response to HFS stimulation (Fig.3C). Injection of 10 µM Kal7-C peptide had no effect on the ability of CA1 hippocampal neurons to exhibit LTP in response to HFS (Fig.3C). This result is consistent with observations made using eld recordings and Kal7 knockout mice . The selectivity of the Kal7-C peptide for blocking TBS-induced LTP without affecting HFS-induced LTP argues for the speci city and lack of toxicity of the Kal7-C peptide.
In an effort to elucidate potential effects of the Kal7-C short peptide on baseline glutamatergic transmission (spontaneous and electrically-evoked), we collected data from a separate cohort of cells that included single evoked excitatory postsynaptic currents (eEPSCs) and spontaneous excitatory postsynaptic currents (sEPSCs) over the course of the rst 10 minutes after entering the whole cell recording con guration. Over a 20 minute period, eEPSC (

Acute disruption of GluN2A/B↔PSD-95 interaction blocks TBS, but not HFS, induced LTP
Biochemical and pharmacological analyses of Kal7 knockout mice have provided strong support for the importance of GluN2B in the actions of Kal7. Both GluN2B and Kal7 terminate with PDZ binding motifs known to interact with PSD-95; whether the two motifs bind simultaneously to PSD-95 or interfere with the other's binding to PSD95 is unknown. In addition to their shared interaction with PSD95, the PH1 region of Kal7 interacts directly with GluN2B ( Fig.1A) .
As expected, administration of 7 train TBS stimulation after bath application of CPP (3 µM), a selective NMDA receptor antagonist, blocked LTP in all cells examined (Fig.2B). We rst explored the role of the GluN2A/GluN2B subunit in TBS induced LTP by injecting a 6-residue peptide identical to the C-terminus of GluN2A/GluN2B (Fig.4). Using GluN2A/GluN2B-C at 10 µM, TBS was unable to elicit LTP ( We next tested the effect of the GluN2A/GluN2B peptide on LTP induced by HFS. Following the injection of 10 µM GluN2A/GluN2B-C, HFS resulted in LTP that was comparable to that observed in control cells ( Fig.4B). As seen with the Kal7-C peptide, HFS-stimulated LTP was not affected by GluN2A/GluN2B-C peptide at levels that blocked TBS-stimulated LTP, again serving as a direct control for nonspeci c effects of the peptide.

Acute disruption of GluN2B-Kal7 interaction blocks TBS, but not HFS, induced LTP
Our data suggest that both the Kal7-C and GluN2A/GluN2B peptides interact with PSD-95 or another PDZdomain protein in the dendritic spine within minutes of breaking into the cell, in order to block LTP following the 7 train TBS induction protocol. For this reason, we decided to probe the interaction between Kal7 and the juxta-membrane domain of GluN2B, at the GluN2B-speci c interaction site described by Kiraly and colleagues   (Fig.1A). At a concentration of 10 µM, the presence of Kal7-GluN2B peptide in the intracellular recording solution blocked 7 train TBS-dependent LTP in all cells examined (p = 0.43) (Fig.5A), with no effect on HFS-dependent LTP (Fig.5B). Even when tested at 1 µM, the Kal7-GluN2B peptide still blocked 7 train TBS-dependent LTP, suggesting that the Kal7/GluN2B interaction is more readily disrupted by exogenous peptide than the PDZ-domain interactions mediated by Kal7-C and GluN2A/GluN2B-C (p = 0.99).
The potent and rapid blocking effect of the Kal7-GluN2B peptide to abolish TBS-dependent LTP, even with only 1 µM peptide in the pipette and for only 10 min, while control peptides at higher concentration had no effect (Fig.3B), strongly indicates that one crucial interaction being blocked by the low amounts of infused peptide is directly between Kal7 and GluN2B. Further, the effect of Kal7-C (10 µM) was statistically signi cant when compared to cells lled with Kal7-C mut (10 µM) (Student's t-test, p = 0.03).
Acute disruption of Kal7↔GluN2A/B↔PSD-95 interactions also alters LTD Mice lacking Kal7 throughout gestation and maturation do not exhibit LTD in response to LFS . We next tested whether the acute addition of Kal7-C peptide blocked LTD in response to LFS stimulation (Fig.6). LFS produced the expected long-lasting depression of eEPSC amplitude (to approximately 50%) in control neurons (Fig.6A). The injection of Kal7-C peptide in the recording pipette abolished LTD for at least 30 min; there was noticeable recovery of LTD toward the end of the hour-long recording session. However, even at 35 -40 minutes post-induction, eEPSC amplitude was not statistically different from baseline. Thus, injection of Kal7-C peptide disrupted LFS-induced LTD seen in control neurons, which corroborates data from Kal7 KO mice.
When LFS induced LTD was tested in the presence of GluN2A/B peptide in the recording pipette (Fig.6B), LTD was initially unaffected; with increasing time, the presence of GluN2A/B gradually blocked the maintenance of LTD (beyond 20 min).
Lastly, we examined the effect of the Kal7-GluN2B peptide (which was remarkably potent at blocking LTP [ Fig.5A]) on LFS-induced LTD (Fig.6C). Unlike the Kal7-C and GluN2B peptides, cells lled with Kal7-GluN2B (10 µM) showed no differences in LTD compared to control cells at any time post-induction and expressed robust LTD following LFS. The differences in the test peptide blocking results for LTP and LTD suggest important differences in the fundamental underlying causes of these long term changes in plasticity (Zhou et al., 2018). Speci cally, GluA1 (which ends …ATGL; a Class 1 PDZ motif, as are the peptides being tested here) is crucial for LTP, while GluA2 (which ends …SVKI, a Class 2 PDZ motif) is crucial for LTD (Zhou et al., 2018).
Biochemical examination of interactions of the test peptides and proteins with PSD-95 PDZ domains are 80-90 residues in length; there are 300-400 such domains in over 150 proteins (Caillet-Saguy et al., 2015;Houslay, 2009). PDZ binding motifs typically consist of only 3-4 primary residues, primarily at the C-terminus of proteins (Caillet-Saguy et al., 2015). The initial interaction of a PDZ binding motif with a PDZ domain is generally followed by an induced t (Chi et al., 2009). PDZ-binding motifs are common; over a third of the 70+ Rho-GEFs contain PDZ-binding motifs, with similar numbers in Drosophila and C.elegans, but no known PDZ domains in yeast (Garcia-Mata & Burridge, 2006). A number of viruses capitalize upon their PDZ-binding motifs to spread within infected cells (Caillet-Saguy et al., 2015;Pim et al., 2012).
In order to address the biochemical interaction of Kal7 and GluN2B, more speci cally if GluN2B and Kal7 compete for binding sites on PSD-95, the initial biochemical attacks utilized the same interfering peptides used for the electrophysiological experiments. Because Kal7 and GluN2B both have Class 1 PDZ binding motifs that interact with PSD-95, the rst approach was to disturb PSD-95 binding to Kal7 with interfering peptides. However, no competition could be demonstrated, despite various experimental approaches that included multiple resins and a wide range of concentrations of competing peptides. Covalent binding of PDZ123 to A gel and to UltraLink Biosupport resins showed a dose-dependent saturable binding of Kal7 to the PDZ123 resin and little background binding to the BSA resin, using 5 pmol puri ed Kal7 input (Miller, Yan, Machida, et al., 2017) (Fig.7A, asterisks). However, competing GluN2B and Kal7 peptides had no effect on the binding of puri ed Kal7 to the PDZ123 resin, even though binding of Kal7 to the BSA resin was minimal (Fig.7B).
These results suggested that the original hypothesis was incorrect and perhaps PSD-95 was interacting with Kal7 and GluN2B through additional sites other than the COOH-terminal PDZ binding domains. It is possible that multiple PDZ domains were in play, that interfering with a single domain was inadequate to disrupt the overall interaction, or that immobilizing either of the binding partners was too restrictive for speci c interactions to occur (outlined in Fig.7C). We took advantage of the fact that Kal8 lacks the Cterminus PDZ binding domain of Kal7 but contains the rest of the Kal7 structure (Fig.7D), in order to address whether Kal7 and GluN2B could interact through additional sites besides the terminal PDZ binding domains. We utilized expression vectors for Kal7, Kal8 and PSD-95, separately transfected these into HEK293 cells, and made protein extracts. We then used co-immunoprecipitation to test for binding of solubilized PSD-95 with either gently solubilized Kal7 or Kal8 proteins (Fig.7D). The gels demonstrated that both Kal7 and Kal8 proteins interacted speci cally with PSD-95, regardless of whether or not they have a PDZ binding motif.

Discussion
Cell-penetrating peptide mimetics of PDZ binding motifs have been shown to disrupt protein-protein interactions as diverse as connexins with zona occludens, multiple binding proteins with AMPA receptors, and various functions of serotonin receptors (Flores, Li, Bennett, Nagy, & Pereda, 2008;Flores et al., 2012;Lee, Liu, Wang, & Sheng, 2002;Luscher et al., 1999;Luthi et al., 1999;Nishimune et al., 1998;Pichon et al., 2010). Additional experiments on spinal cord demonstrated that Kal7-derived peptides could obliterate the usual LTP and LTD seen in spinal cord preparations (Lu et al., 2015), and that cell-permeant Kal7derived peptides disrupt the appearance of dendritic spines when applied to cultured hippocampal neurons . Importantly, calpain cleavages of cytoskeletal and PSD proteins are established as crucial for LTP (Briz & Baudry, 2016), and one of the rst cleavages of the PSD protein Kal7 releases a peptide nearly identical to Kal7-L (Table 1) (Miller, Yan, Machida, et al., 2017), suggesting that some of the peptide may be important for LTP but additional peptide is deleterious to LTP.
Taken together, our current data on hippocampal slices extend previous studies that identi ed Kal7 and the GluN2B subunit-containing NMDA receptors as essential postsynaptic components of synaptic plasticity in CA1 neurons of mouse hippocampus. Further, we have integrated synthetic peptides, whole cell patch clamp electrophysiology and binding site investigations to probe the functional interactions between Kal7, GluN2B and PSD-95 and their roles in mediating LTP and LTD in pyramidal neurons in the hippocampus. The striking nding is that disrupting speci c postsynaptic interactions with synthetic peptides can affect synaptic plasticity, both LTP and LTD, within minutes after patching the neuron. This phenomenon suggests that a lifelong de cit in Kal7 (i.e. the Kal7 knockout mouse) is not necessary for at least some of the de cits in plasticity seen with the appropriate patterns of stimulation of the Schaffer collaterals. The lack of effect of control peptides, with one amino acid residue altered to ablate the PDZ binding motif, and the lack of effect of any of the peptides on HFS-induced LTP, argues that the effects blocking TBS-driven LTP are speci c, not simply a cytotoxic effect of infusing a small peptide into the recorded neuron. The increase in eEPSC amplitude following 7 train TBS observed using the Kal7-C mut peptide (Fig.3B) appears to be less than the increase observed in control neurons (no peptide). This could raise the question of whether the response to the control peptide (Kal7-C mut ) is different from the Kal7-C peptide, which completely disrupted theta burst-driven LTP (Fig.3A). However, within-cell comparisons demonstrated that neurons lled with Kal7-C mut indeed expressed LTP that was statistically signi cant when compared to baseline, while LTP was obliterated in Kal7C-infused neurons. Although the sample sizes in this exploratory study are somewhat small in select experiments, individual data points are displayed for all experiments to emphasize the low variance. Peptides infused from recording pipettes have been used to successfully investigate the electrophysiology of several other model systems (Flores et al., 2008;Flores et al., 2012;Lee et al., 2002;Lu et al., 2015;Luscher et al., 1999;Luthi et al., 1999;Nishimune et al., 1998;Pichon et al., 2010).
Disrupting Kal7/PSD-95, GluN2B/PSD-95 and GluN2B/Kal7 interactions with peptide mimics blocks 7 train theta burst stimulation (TBS)-induced long term potentiation in CA1 of mouse hippocampus, while leaving LTP produced by high frequency stimulation (HFS) untouched. Further, disruptions of Kal7/PSD-95 and NR2B/PSD-95 binding, but not disruption of NR2B/Kal7, also block low frequency stimulation (LFS)-induced long-term depression. A recent paper demonstrated, by creating transgenic knock-in mice bearing intramolecular swaps of the COOH-terminal regions of GluA1 and GluA2, that the COOH-terminal domain of GluA1 (Class 1 PDZ) was su cient to enable the expression of LTP in response to HFS, even when the GluA1 COOH-terminal was attached to the GluA2 ion channel-transmitter receptor domains (Zhou et al., 2018). Similarly, the COOH-terminal domain of GluA2 (Class 2 PDZ) was su cient to facilitate the expression of LTD in response to high frequency stimulation, even when attached to the GluA1 ion channel-transmitter receptor domains (Zhou et al., 2018). Our electrophysiological results demonstrate consistent differences between LTP in response to HFS vs. LTP in response to TBS; Theta Burst Stimulation is generally thought to mimic more closely the signaling patterns in the animal than HFS (Larson & Munkacsy, 2015;Larson et al., 1986;Vanderwolf, 1969).
We have used intracellular delivery of interfering peptides, along with binding a nity studies, to probe the role of Kalirin-7 in synaptic plasticity. The results demonstrated that the direct interactions of both Kal7 and the GluN2B subunit with a PDZ domain, presumably the very abundant PSD-95, are essential for two major forms of synaptic plasticity, TBS-induced LTP and LFS-induced LTD. Strikingly, HFS-induced LTP, as used in many studies (e.g. (Zhou et al., 2018)), was not blocked by these interfering peptides. The direct interaction of the PH domain of Kal7 with the juxtamembrane domain of GluN2B was also shown to be the most delicately poised interaction for maintaining TBS-induced LTP, but strikingly played no role in LTD. The GluN2B peptide used has only 57% match to the closest mouse protein other than GluN2B (BLAST search).
Our data illustrate that appropriate protein domain interactions are crucial for proper signal complex formation and localization to confer correct function. While speci c interactions between binding sites may be suggested at the electrophysiological level of a single cell, larger scale biochemical experiments elucidate that these molecules have more complex binding capabilities. The notion that small molecules are able to disrupt speci c interactions of PDZ domains in only a few minutes may still present potential therapeutic applications without triggering greater changes at the PSD (Houslay, 2009).

Conclusion
Kalirin-7 (Kal7) is a Rac1/RhoG GEF and multidomain scaffold localized to the postsynaptic density which plays an important role in synaptic plasticity. To address the underlying cellular and molecular mechanisms of excitatory synaptic transmission which are deranged by loss of Kal7, we infused candidate intracellular blocking peptides to acutely challenge Kal7 function at the synapse, to determine if plasticity de cits in Kal7 -/mice are the product of developmental processes since conception, or could be detected on a much shorter time scale. We found that speci c disruption of the interactions of Kal7 with PSD-95 or GluN2B resulted in signi cant suppression of long-term potentiation and long-term depression. Biochemical approaches indicated that Kal7 interacted with PSD-95 at multiple sites within Kal7.

Materials And Methods
Animal handling and slice preparation All animal procedures were conducted using animal protocols approved by the University of Connecticut Health Center Institutional Animal Care and Use Committee and adhered to the ARRIVE Guidelines.

Electrophysiology
Whole cell recordings were obtained from pyramidal neurons in the CA1 region of the hippocampus.
Pyramidal neurons were identi ed by their morphology and position under infrared differential interference contrast video microscopy. All electrical events were ltered at 2.9 kHz and digitized at > 6 kHz using a HEKA EPC9 ampli er and ITC-16 digitizer (HEKA Elektronik). Series resistance (R s ) was compensated up to 20% at 100 µs lag. Input resistance (R i ) was monitored with 10 mV (50 ms) hyperpolarizing steps at the end of each sweep. Cells were rejected from analysis if R i fell below 50 MΩ or the holding current changed by > 15% during the course of an experiment. Baseline recordings began after stable eEPSCs were obtained according to criteria outlined above. The entire duration of baseline recordings lasted 10 minutes. However, to ensure a stable patch and recording had been obtained, eEPSCs obtained within minutes 5 -10 were used as "baseline" for comparison. Said differently, the results presented in this manuscript compare the average of eEPSCs within minutes 35 -40 following various stimulation paradigms to the average of eEPSCs obtained within minutes 5 -10 of the start of each experiment, prior to stimulation.

Peptides
All synthetic peptides were purchased from Biomatik (Wilmington DE); except where indicated, the aminoand carboxyl-terminal ends of the peptides were not modi ed. Peptide names and sequences are listed in Table 1. Peptides were reconstituted in the pipette internal solution and stored as concentrated stocks.
Prior to recording experiments, each peptide was diluted in the pipette internal solution described above.

Experimental design and statistical analysis
Group data are reported as mean ± SE. Statistical comparisons were made using one-way ANOVA and Dunnett's multiple comparison test or Student's paired t-test for post-hoc comparison. p < 0.05 was taken as a statistically signi cant effect and marked with an asterisk (*). Long term potentiation and depression are always compared to untreated cells (no peptide) subjected to the same electrical stimulation pattern. All data supporting the results in this paper are presented in this paper; additional details are available from eslevine@uchc.edu, eipper@uchc.edu, or mains@uchc.edu.