Calcium current homeostasis and synaptic deficits in hippocampal neurons from Kelch-like 1 knockout mice

Kelch-like 1 (KLHL1) is a neuronal actin-binding protein that modulates voltage-gated CaV2.1 (P/Q-type) and CaV3.2 (α1H T-type) calcium channels; KLHL1 knockdown experiments (KD) cause down-regulation of both channel types and altered synaptic properties in cultured rat hippocampal neurons (Perissinotti et al., 2014). Here, we studied the effect of ablation of KLHL1 on calcium channel function and synaptic properties in cultured hippocampal neurons from KLHL1 knockout (KO) mice. Western blot data showed the P/Q-type channel α1A subunit was less abundant in KO hippocampus compared to wildtype (WT); and P/Q-type calcium currents were smaller in KO neurons than WT during early days in vitro, although this decrease was compensated for at late stages by increases in L-type calcium current. In contrast, T-type currents did not change in culture. However, biophysical properties and western blot analysis revealed a differential contribution of T-type channel isoforms in the KO, with CaV3.2 α1H subunit being down-regulated and CaV3.1 α1G up-regulated. Synapsin I levels were also reduced in the KO hippocampus and cultured neurons displayed a concomitant reduction in synapsin I puncta and decreased miniature excitatory postsynaptic current (mEPSC) frequency. In summary, genetic ablation of the calcium channel modulator resulted in compensatory mechanisms to maintain calcium current homeostasis in hippocampal KO neurons; however, synaptic alterations resulted in a reduction of excitatory synapse number, causing an imbalance of the excitatory-inhibitory synaptic input ratio favoring inhibition.


INTRODUCTION
The actin-binding protein KLHL1 is expressed in the cytosol, axons and dendrites of neurons and in glia in the nervous system (Prag and Adams, 2003;Jiang et al., 2007;Chen et al., 2008). He and collaborators confirmed KLHL1's physiological role in cerebellar function using targeted deletion of KLHL1 in Purkinje neurons, which resulted in dendritic deficits, mild cerebellar atrophy, abnormal gait and progressive loss of motor coordination in mice (He et al., 2006). The KLHL1 gene locus lies near genes implicated in Spinocerebellar ataxia type 8 (SCA8; Ikeda et al., 2007), a rare and slowly progressive neurological disease characterized by the presence of trinucleotide expansions in two genes: ataxin 8 (ATXN8) and ATXN8 opposite strand (ATXN8OS). Mutations in ATXN8 and ATXN8OS genes cause disruptions in multiple cellular pathways; it has been proposed that the characteristic SCA pathogenesis does not begin until calcium signaling pathways are disrupted, and calcium blockers and stabilizers have been suggested as potential treatment in the disease (Kasumu and Bezprozvanny, 2012). The 5'region of the KLHL1 gene overlaps with the 5' end of ATXN8OS (Koob et al., 1999;Nemes et al., 2000;Ikeda et al., 2007). The expression pattern of the KLHL1 antisense strand suggests a regulatory role of KLHL1, and trans-RNA interference assays support that ATXN8OS may function as an endogenous negative regulator of KLHL1 (Nemes et al., 2000;Lin et al., 2007;Chen et al., 2008). Therefore, understanding the physiological functions of KLHL1 will help frame our understanding of the regulatory role of its antisense.
KLHL1 functions as a modulator of neurite outgrowth and process elongation, and it is also a modulator of ion channel activity (Seng et al., 2006;Aromolaran et al., 2007Aromolaran et al., , 2010Jiang et al., 2007). KLHL1 positively modulates Ca V 2.1 (P/Q-type, α 1A ) and 3.2 (T-type α 1H ) voltage-gated calcium channel function, but it does not affect Ca V 3.1 (T-type α 1G ) or 3.3 (T-type α 1I ) channels, nor Ca V 1.2, 1.3, or 2.2 (Perissinotti et al., 2014). In vitro studies demonstrated that KLHL1 interacts with the polymerized actin cytoskeleton and with the pore-forming subunit of the Ca V 3.2 channel. This interaction increases its recycling via the recycling endosome, increasing the number of active channels at the plasma membrane thus up-regulating T-type calcium current density (Aromolaran et al., 2010). Adenoviral delivery of shRNA against KLHL1 down-regulates KLHL1, which in turn decreases expression levels and current density of Ca V 3.2 and Ca V 2.1 channels in cultured rat hippocampal neurons; KLHL1 also reduces synapsin I levels, miniature excitatory postsynaptic current (mEPSC) frequency and the number of excitatory and inhibitory synapses (Perissinotti et al., 2014).
To corroborate the effects of endogenous KLHL1 we assessed the status of voltage-gated calcium currents and synaptic properties in cultured hippocampal neurons from the KLHL1 KO mouse model (He et al., 2006). We found a homeostatic upregulation of α 1C (Ca V 1.2) and α 1G (Ca V 3.1) concomitant to decreased expression of α 1A and α 1H subunits in the KO mice, confirming the role of KLHL1 in the positive modulation of P/Q-and T-type calcium currents. Calcium current homeostasis resulted in maintenance of total somatic calcium current levels in KO hippocampal neurons. However, the synaptic deficit could not be compensated for, resulting in decreased ratio of excitatory/inhibitory synaptic inputs onto KLHL1 KO neurons, confirming a presynaptic function of KLHL1 most likely due to its effects on synapsin I, presynaptic P/Q-type channels and T-type channels.

MATERIALS AND METHODS
The animal protocols used in this study were reviewed and approved by an independent Institutional Animal Care and Use Committee (IACUC).

MOUSE MODEL
KLHL1-knockout mice were generated in the Laboratory Medicine and Pathology from the University of Minnesota (He et al., 2006).
Cells with series resistance (R s ) <20 MΩ were used; R s was compensated on line (>80%). Data were acquired with Clampex 10 and analyzed with Clampfit 10 software (Molecular Devices). Voltage control was enhanced by increasing cell impedance using extracellular TEA and intracellular cesium to block K + conductances. I-V curve properties such as its negative slope and reversal potential were monitored for appropriate voltage control. For the study of calcium current properties, we avoided recording from neurons older than 10 DIV because the possibility of space-clamp problems.

Current-voltage relationships (I-V curves)
Currents were elicited from a holding potential (V H ) = −90 mV or −50 mV and depolarized for 150 ms to a test potential (V T ) = −70 to + 60 mV, in 10 mV increments. The low voltage activated (LVA) current component contribution was determined by the subtraction method (Bean, 1985): average high voltage activated (HVA) currents obtained at V H = −50 mV were subtracted from those obtained at V H = −90 mV at each test potential.

Calcium influx
Calcium influx was determined by the current integral evoked by an action potential waveform (APW; influx normalization by capacitance was not necessary as it produced similar results as the integral values reported). The LVA component was isolated by the subtraction method. The APW consisted of a digitized hippocampal action potential with a resting potential of −70 mV, an upstroke of 116 mV/ms to a peak voltage of + 50 mV, followed by a repolarizing downstroke at −65 mV/ms to a hyperpolarizing potential of −90 mV. The repolarization from after-hyperpolarization slope was 0.78 mV/ms to resting conditions (Aromolaran et al., 2012). This protocol was applied from V H of −50 or −90 mV prior to clamping the voltage at −70 mV for 10 ms before the triggering of the AP, so no changes in the driving force should be expected.

Current kinetics
Current activation and inactivation were assessed by stepping the neuron from V H = −90 mV to V T = −50 to −20 mV for 150 ms. The rate of activation was estimated from the current rise time from 10% to 90% of its maximum value (rise time 10-90%); these measurements were favored over the calculation of the time constant of activation (τ on ) to minimize the contribution of the inactivation process on the activation rate value. Time constant of inactivation (τ off ) was obtained from mono-exponential fits. The time constant of deactivation (τ deactivation ) was measured by fitting with a mono-exponential function the decaying phase of tail currents elicited from V H = −90 mV to V T = −30 mV and back to V T = −60 to −140 mV.

Steady-state analysis
Steady-state activation (SSA) was analyzed with protocols stepping from V H = −90 (or −50) mV to V T up to 0 mV (∆V = 10 mV) for 12 ms followed by repolarization to −100 mV to evoke inward tail currents. Data were fitted by a single Boltzmann function of the form, where I max is maximal current, V 50 is half-voltage of activation, k is slope factor, and m is baseline. Steady-state inactivation (SSI) was determined by stepping the membrane potential to various pre-pulse voltage levels (V pre = −110 to 0 mV, ∆V = 10 mV) for 1 s before depolarization to a fixed test level (−30 mV) to evoke channel opening. The resulting data were also fitted to a Boltzmann function.

STATISTICS
Results are presented as mean ± SEM. Statistical analysis was performed with the Sigma Plot 11 Software. Statistical significance was determined by P < 0.05 using Student's t-test, unless otherwise noted. Analysis of Variance with post-hoc Duncan's Method (ANOVA) was performed when more than 2 data sets were compared. Portions of the results reported here have been

HOMEOSTATIC REGULATION OF VOLTAGE DEPENDENT CALCIUM CURRENT IN KLHL1 KO HIPPOCAMPAL NEURONS
We measured calcium current density from WT and KO neurons during early and late stages in culture (4-6 DIV and 8-10 DIV, respectively), using established electrophysiological protocols to assess the HVA and LVA current components (Methods).

HVA calcium current
The I-V curves in Figure 1A show that the HVA peak current density at 4-6 DIV was 33% smaller in KO than in WT neurons, but this difference was no longer detected at 8-10 DIV. In contrast, there were no differences in the calcium influx evoked by an APW between WT and KO at any stage in culture. However, the decay time and half-width of the APW-induced currents were smaller in KO neurons at 8-10 DIV than in WT (Figure 1B), suggesting that different HVA channels were contributing to the total HVA current during the development of KO and WT hippocampal neurons. Sequential application of calcium channel blockers was used to identify individual HVA current components. Figure 2A shows an example where peak current was plotted vs. time during drug application, along with representative current traces before (a) and after application of: (b) 1 µM nifedipine; (c) 200 nM ωagatoxin IVA; and (d) 2 µM ω-conotoxin GVIA. Figure 2B shows the fraction of current blocked by each of these antagonists at 4-6 DIV: 55 ± 5% for nifedipine (n = 11), 21 ± 5% for ωagatoxin IVA (n = 6) and 15 ± 3% for ω-conotoxin GVIA (n = 5) for WT neurons. The fraction of current blocked by nifedipine was not significantly different in KO neurons (47 ± 4%, n = 8), whereas ω-agatoxin IVA and ω-conotoxin GVIA blocked 10 ± 3% (n = 6) and 32 ± 6% (n = 5), respectively; these fractions were significantly different from WT. In contrast, at 8-10 DIV the fraction of current blocked by nifedipine was higher in KO neurons (58 ± 4%, n = 16) compared to WT (36 ± 5%, n = 12), whereas ω-agatoxin IVA blocked a smaller fraction of current in the KLHL1-KO (13 ± 3%, n = 6) than in WT (31 ± 7%, n = 7). No differences between genotypes were observed for ω-conotoxin GVIA at 8-10 DIV (32 ± 5% and 26 ± 6% for WT and KO, n = 9 for each genotype). These results show that decreased levels in P/Q-type channels in KLHL1-KO neurons observed at 4-6 and 8-10 DIV were compensated for by the functional increases of N-type and L-type channels at each stage, respectively. The expression levels of L-type α 1C (Ca V 1.2) and α 1D (Ca V 1.3), P/Q-type α 1A (Ca V 2.1), and N-type α 1B (Ca V 2.2) poreforming subunits were determined in whole adult hippocampus by western blot (Figure 2C). KO hippocampus had lower expression of α 1A subunit (n = 6), and α 1C subunit levels showed a statistical tendency to higher expression compared to WT (n = 5, P = 0.05), supporting the electrophysiology data. Neither expression levels of α 1B (n = 3) subunit or α 1D (n = 4) changed in the KO. Figure 3A shows I-V curves at early and late stages of culture. Both HVA and LVA current components can be detected in this experiment, the HVA peak component at +10 mV and the LVA peak at ∼−30 mV. LVA peak current density (I peak ) did not change during the time of culture, nor between genotypes. The LVA peak voltage (V peak ) changed from −25 mV at 4-6 DIV to ∼−32 mV at 8-10 DIV in WT neurons, but it remained unchanged in the KO. Thus, this ∼7 mV left shift in the peak voltage suggests that the reported developmental change in expression of LVA channel isoforms occurred in WT but not in KO neurons (McRory et al., 2001).

LVA calcium current
Western blot analysis of whole hippocampus confirmed that the KO exhibits down-regulation of the Ca V 3.2 α 1H isoform and a concomitant homeostatic increment in Ca V 3.1 α 1G expression ( Figure 3B); Ca V 3.3 α 1I protein levels remained constant.
LVA current kinetics and steady-state properties were also analyzed to assess the T-type isoform expression changes (Cain and Snutch, 2010; Figure 4). Current activation (rise time 10-90%) did not change between genotypes or during the time of culture ( Figure 4A). τ off was similar between genotypes and stage in culture, except for the value at −40 mV for WT at 8-10 DIV ( Figure 4B). In contrast, τ deactivation values were significantly slower in KO neurons compared to WT at 4-6 DIV (5.3 ± 0.3 vs. 4.1 ± 0.2 ms at −60 mV), but not at later stages of culture where both genotypes displayed slower τ deactivation , similar to those seen in KO neurons at early stages (6.1 ± 0.7 for WT and 5.7 ± 0.3 for KO) ( Figure 4C). Steady-state activation (SSA) and inactivation (SSI) V 50 properties did not change between genotypes at 4-6 DIV; however WT V 50 values significantly shifted to negative potentials at later stages of culture ( Figure 4D).
Calcium influx was assessed using APWs ( Figure 5). Calcium influx via LVA channels was 34% smaller at 4-6 DIV in the KO compared to WT neurons, but this difference was no longer detected at 8-10 DIV. Furthermore, whereas the time constants of APW-induced LVA currents (rise time, decay time, half-width and time to peak) were slower in KO neurons at 4-6 DIV compared to WT, no differences in these constants were observed at 8-10 DIV.

ALTERED SYNAPSE NUMBER IN KLHL1 KO HIPPOCAMPAL NEURONS
shRNA-induced KLHL1 KD in rat hippocampal neurons results in reduction of the number of synaptic puncta, synapsin I protein levels and decreased mPSC frequency (Perissinotti et al., 2014); thus we tested the impact of the genetic ablation of KLHL1 on these synaptic properties in mice neurons.
Analysis of the number of synaptic puncta per area was performed at 11 DIV. KLHL1 KO cultures exhibited significantly fewer puncta compared to WT (P < 0.05). The example shown in Figure 6A depicts neurons stained for synapsin I (red) merged with the neuronal marker MAP2 (blue). Synaptic contacts decreased from 1661.2 ± 493.1 puncta per frame in WT to 809.2 ± 94.6 in KO neurons. In a separate line of experiments, quantification of synapsin I by western blot showed that protein levels were ∼25% lower in the adult KO hippocampus (0.73 ± 0.126) ( Figure 6B). In line with these results, mPSC properties in the absence of blockers showed lower frequency in KO neurons (0.25 ± 0.03 Hz) compared to WT (0.63 ± 0.16) (Figure 6C), confirming a decrease of synapse number in the KO. mPSC amplitude was not altered (WT: −12.0 ± 1.9 pA, KO: −12.1 ± 1.8 pA).
To identify if the decrease in synapse number was typespecific we characterized excitatory synapses by co-localization of synapsin I (Syn I; Figure 7A), or the presynaptic glutamate vesicle transporter 1 VGlut-1 (Figure 7B) with the postsynaptic scaffolding protein PSD-95 (Woods and Bryant, 1993;Gomperts, 1996). Mander's correlation coefficient analysis showed KO neurons exhibited less co-localization between pre-and post-synaptic excitatory markers. WT neurons exhibited 35.9 ± 2.9% colocalization of Syn I with PSD95 compared to 28.4 ± 2.0% in the KO; and 54.3 ± 6.3% co-localization of VGlut-1 with PSD-95 in WT compared to 38.3 ± 5.8% in the KO. Inhibitory synapses were identified by the presence of synapsin I (Figure 7C) or the presynaptic enzyme glutamic acid decarboxylase (GAD67, Figure 7D) and the postsynaptic scaffolding protein gephyrin (Köhler and Chan-Palay, 1983;Danglot et al., 2003). There were no differences in colocalization of Syn I with gephyrin, however signal co-localization between GAD67 and gephyrin was significantly higher in the KO (36.9 ± 3.8%) compared to WT (26.6 ± 3.0%).
Excitatory and inhibitory activities were then studied in isolation using specific inhibitors (Figures 8A-D). Excitatory signals recorded in the presence of 20 µM Bic were in accordance with the co-localization experiments; excitatory activity was much lower in the KO (17% of WT), with a mEPSC frequency of 0.20 ± 0.05 Hz (n = 7) compared to 1.21 ± 0.43 Hz in the WT, confirming a decrease in excitatory synapse number (Figures 8A,B). Again, no changes in mEPSC amplitude were detected (−18.7 ± 2.8 pA, WT vs. −15.1 ± 1.5 pA, KO, Figure 8C). mIPSC activity was studied in presence of 10 µM APV and 10 µM NBQX. As expected from the Syn I-gephyrin co-localization experiments, no differences were observed in either mIPSC frequency between KO and WT genotypes (Figures 8A-C). KLHL1 KO neurons had a mean mIPSC frequency of 0.43 ± 0.21 Hz compared to 0.43 ± 0.15 Hz in the WT. mIPSC amplitude did not change: 16.3 ± 2.9 pA and 17.2 ± 1.3 pA. mIPSCs exhibited slower kinetics than mEPSCs, as expected (Shao et al., 2003;Szczot et al., 2010), but no significant changes in the kinetic properties of either excitatory or inhibitory currents were detected between groups ( Figure 8D). Overall, the mEPSC/mIPSC frequency ratio was 2.8 in WT neurons, compared to 1 in the KO (Figure 8A).

DISCUSSION
KLHL1 function as an endogenous modulator of P/Q-and Ttype calcium channels was first detected by acute KLHL1 KD experiments in rat neurons (Perissinotti et al., 2014) and is now confirmed in the KLHL1 knockout mouse model. Hippocampal HVA currents were comprised of L-type, P/Qtype, and N type currents, as previously reported (Isaeva et al., 1998;Chameau et al., 1999;Perissinotti et al., 2014), in line with a differential contribution of HVA calcium channel subtypes at different stages in vitro (Chameau et al., 1999;Nudler et al., 2003). KLHL1 ablation resulted in a 33% decrease in HVA current density at early culture stages, which was compensated for by 8-10 days in culture. KO neurons exhibited decreased contribution by P/Q-type current since early stages of culture, and decreased α 1A levels were detected in adult hippocampus. This data is in agreement with KD experiments of KLHL1 using shRNA in cultured rat hippocampal neurons (Perissinotti et al., 2014), and also with over-expression experiments of KLHL1 in vitro (Aromolaran et al., 2007). N-type and L-type currents increased in KO neurons during early and late stages of culture, respectively, likely as compensatory mechanisms for the decrement in P/Q-type currents. α 1B subunit protein levels were similar between adult WT and KO hippocampal samples, which are representative of late stage hippocampal culture conditions, where adult neuronal phenotypes are observed. There was a possible up-regulation of the α 1C subunit of the L-type channel, its protein level tended to be statiscally significant by western blot analysis (p = 0.05); therefore increased L-type current activity may be due to post-translational mechanisms and/or increased protein levels.
LVA calcium current density was constant during the time in culture in WT neurons, as reported for cultured mouse hippocampal neurons (Chameau et al., 1999), and in contrast with acutely dissociated and cultured rat hippocampal neurons during postnatal development (Kortekaas and Wadman, 1997;Perissinotti et al., 2014). All three T-type channel isoforms (Ca V 3.1-3.3 or α 1G-I ) are present in hippocampal neurons (McKay et al., 2006), and despite the lack of change in total current, it is known that the expression pattern of Ca V 3.3 (α 1I ) mRNA is higher in immature compared to mature stages, whereas Ca V 3.1 and Ca V 3.2 expression (α 1G and α 1H ) predominates in mature hippocampal neurons (McRory et al., 2001). Regarding their biophysical properties, Ca V 3.3 (α 1I ) channels have peak currents at more positive potential, display slower kinetics of activation and inactivation and have faster kinetics of deactivation compared to Ca V 3.1 and Ca V 3.2. α 1I currents also display more positive steady state V 50 values than α 1H /α 1G currents (Klöckner et al., 1999;McRory et al., 2001;Chemin et al., 2002;Perez-Reyes, 2003). Analysis of protein levels demonstrates the KO mouse model exhibited slight down-regulation of the . Bottom right, mPSC amplitude distribution. Inset, mean amplitude value for each population (pA) (n = 10, 9). *p < 0.05, Student's t-test; **p < 0.05, paired Student's t-test.
Ca V 3.2 α 1H isoform, in contrast with KD experiments where α 1H down-regulation was substantial (Perissinotti et al., 2014); this difference is likely due to the manifestation of compensatory mechanisms in the KO, although differences in LVA current modualtion/expression between mouse and rat models could also be involved. Ca V 3.1 α 1G levels were increased and no changes in the level of Ca V 3.3 α 1I expression were detected in the KO model, in line with the function of KLHL1 as a positive modulator of Ca V 3.2 (Aromolaran et al., , 2010Perissinotti et al., 2014). Similar to previous reports, we did not detect differences in total LVA current density between stages in culture, nor between genotypes; however analysis of biophysical properties showed V peak , SSA and SSI curves were shifted toward more negative values in WT neurons, with slower deactivation rates at 8-10 DIV, consistent with a decrease in the relative contribution of the α 1I component in WT neurons at later stages in vitro. In contrast, neither V peak , SSA nor SSI curves changed in KO neurons during their time in culture suggesting that down-regulation of α 1H allowed the electrophysiological detection of α 1I current properties, even at late stages of neuronal culture when they are normally not detectable due the preponderance of α 1G and α 1H . LVA currents in KO neurons had properties similar to those seen in WT neuronal cultures at early stages in vitro, when α 1I is more abundant in hippocampal neurons. This result is in agreement with our KLHL1 knockdown study in rat hippocampal neurons (Perissinotti et al., 2014).
Ca V 3.1 current is not modulated by KLHL1 (Aromolaran et al., 2010); and interestingly, the KO model elicited significant overexpression of α 1G subunit in the hippocampus. Accordingly, LVA currents in KO neurons exhibited slow time constant of deactivation from early days in culture; suggesting the presence of α 1G masked the contribution of the faster deactivation kinetics of α 1I , the normally predominant component at early stages of development. The rate of deactivation was unaffected by the increase in the relative contribution of α 1G at later stages likely due to the concomitant decrease in α 1H levels, as both α 1G and α 1H have similar deactivation constants.
The unique biophysical properties of the LVA currents are more noticeable in APW experiments as previously reported (McCobb and Beam, 1991), where LVA-mediated calcium influx was 34% smaller in KO neurons at early stage, despite the similar LVA current densities detected in WT and KO. Interestingly, expression of Ca V 3.3 in immature cultures cannot effectively provide comparable calcium influx as Ca V 3.1 and/or Ca V 3.2 do in mature neurons (McRory et al., 2001). The slow activation kinetics of α 1I channels results in small currents in response to a single action potential (Kozlov et al., 1999), uncovering the deficiency in α 1H subunit expression in the KO. At late stages α 1G and α 1H activate quickly and display similar current amplitudes in response to a single action potential (Kozlov et al., 1999). In agreement, we did not observe differences in the calcium influx between KO and WT genotypes at mature stages, even though these cultures displayed different α 1H /α 1G expression ratios. Calcium entry through T-type channels occurs mainly during the repolarization phase of the AP, therefore the deactivation kinetics are important in tuning the current decay (Djouhri and Lawson, 1999;Chemin et al., 2002). Current deactivation and its decay time constant in response to an APW were increased in the KO neurons at early days of culture, likely due to the contribution of the slow α 1G component which contributes to a slowly decaying current (Chemin et al., 2002). Overall, the data show a differential contribution of T-type isoforms during development of WT and KLHL1-KO neurons in vitro, and a homeostatic regulation of LVA currents in the KO by increased α 1G levels upon down-regulation of α 1H .
KLHL1 KD in cultured rat hippocampal neurons resulted in decreased synapsin I contents and decreased number of excitatory synapses, confirmed by decreased number of synaptic puncta and decreased mEPSC frequency. The number of inhibitory synapses was also reduced, however electrophysiological confirmation of mIPSC frequency could not be obtained in that system due Inhibitory synapses, identified by Syn I and gephyrin (C) and GAD67 and gephyrin (D) antibodies; left, quantification of the % co-localization of Syn I with gephyrin (n = 27, 29) or GAD67 with gephyrin (n = 9, 11) for WT and KO. *p < 0.05, Student's t-test. Scale bars = 20 µm.
to experimental difficulty (Perissinotti et al., 2014). Decreased synapsin I levels and decreased synaptic inputs onto cultured hippocampal neurons also occurred in the KLHL1 KO model. However, the reduction in the number of synapses was specific for excitatory neurons in the KO. This synaptic alteration resulted in abnormal mEPSC/mIPSC frequency ratio, favoring inhibition. In contrast to the KD model, there were no differences in inhibitory synapse number and the frequency of mIPSCs between genotypes; in fact we observed an increase in the % co-localization between GAD67 and gephyrin, suggesting increased expression of GAD-67 as a likely compensatory mechanism for reduced glutamatergic inputs. Synaptic homeostasis mechanisms have been thoroughly demonstrated to maintain neuronal excitability (Davis and Bezprozvanny, 2001;Turrigiano, 2012) and GAD67 is a target for activity-dependent regulation (Lau and Murthy, 2012); however changes in expression of gephyrin has not been ruled out. The inability to compensate the excitatory/inhibitory imbalance in KO neurons is likely due to the combined altered expression of synapsin I and α 1A and α 1H Ca V channels. Synapsin phosphoproteins play an important role in the regulation of axonogenesis, synaptogenesis and the modulation of neurotransmitter release (Hilfiker et al., 1999;Ferreira and Rapoport, 2002;Evergren et al., 2007;Bykhovskaia, 2011). In parallel, pre-and post-synaptic Ca V channel subunits are directly involved in the formation and maintenance of synapses by interacting with synapse organizers (Brodbeck et al., 2002;Nishimune et al., 2004;Eroglu et al., 2009;Kurshan et al., 2009;Nishimune, 2011). We studied the status of somatic calcium currents only, but western blot results are indicative of global Ca V protein levels in pre-and post-synaptic sites altogether. P/Q-type channels are preferentially located in presynaptic terminals throughout the brain; they have a prominent role in controlling neurotransmitter release that increases during development (Nudler et al., 2003;Inchauspe et al., 2004;Giugovaz-Tropper et al., 2011); thus decreased presynaptic α 1A could alter the number of excitatory synapses in the KLHL1 KO. Q and N-type calcium channels mediate synaptic transmission in the hippocampus (Scholz and Miller, 1995;Wheeler et al., 1995;Pravettoni et al., 2000), and P/Q-, N-and R-type channel openings account for ∼50% of all spontaneous glutamate release at rat cultured hippocampal synapses (Scholz and Miller, 1995;Ermolyuk et al., 2013); on the other hand, N-and L-type but not P/Q-type channel activity generate the spontaneous release of GABA in hippocampal granule neurons (Goswami et al., 2012). Therefore decreased P/Q channel levels and activity could affect excitatory neurons to a greater extent than inhibitory neurons. Furthermore, regulation of spontaneous synaptic release of glutamate by presynaptic Ca V 3.2 channels has been reported for dorsal horn neurons, raising the possibility that alterations in T-type current activity could also have a deleterious role in pre-synaptic function of excitatory hippocampal neurons (Bao et al., 1998;Jacus et al., 2012;Rozanski et al., 2013;Fekete et al., 2014). In contrast, Ltype channels are preferentially located at postsynaptic sites and have less effect on neurotransmitter release (Kamiya et al., 1988;Pravettoni et al., 2000), therefore up-regulation of L-type currents unlikely compensated for deficits in presynaptic calcium currents in the KO. Finally, altered astrocyte function due to the absence of KLHL1 could be possible and has not been ruled out as a factor contributing to the synaptic deficits observed in the KLHL1 KO (Ullian et al., 2001;Christopherson et al., 2005;Jiang et al., 2007;Südhof, 2012). The KLHL1 KO mouse model highlights the importance of KLHL1 in the modulation of Ca V 2.1 and Ca V 3.2-mediated calcium currents, and revealed compensatory mechanisms to maintain calcium homeostasis. Overall, ablation of KLHL1 leads to an initial decrease in calcium current in cultured neurons, which was compensated for by homeostatic up-regulation of voltage dependent calcium currents not subjected to modulation by KLHL1, such as L-, N-and α 1G T-type currents. However, despite the upregulation of somatic HVA and LVA current densities, possible alterations in calcium channel composition at presynaptic sites and down-regulation of synapsin I resulted in decreased number of excitatory synapses and altered balance between excitatory and inhibitory inputs onto hippocampal neurons, corroborating a synaptic function for KLHL1 (Aromolaran et al., 2007(Aromolaran et al., , 2010Perissinotti et al., 2014). These deficits may become significant during process of sub-optimal function, such as aging or in disease. Erika S. Piedras-Rentería prepared figures; Erika S. Piedras-Rentería and Paula P. Perissinotti wrote the manuscript; Paula P. Perissinotti and Erika S. Piedras-Rentería edited and revised the manuscript. Erika S. Piedras-Rentería approved the final version of the manuscript.