A novel phospho-modulatory mechanism contributes to the calcium-dependent regulation of T-type Ca2+ channels

Cav3 / T-type Ca2+ channels are dynamically regulated by intracellular Ca2+ ions, which inhibit Cav3 availability. Here, we demonstrate that this inhibition becomes irreversible in the presence of non-hydrolysable ATP analogs, resulting in a strong hyperpolarizing shift in the steady-state inactivation of the residual Cav3 current. Importantly, the effect of these ATP analogs was prevented in the presence of intracellular BAPTA. Additional findings obtained using intracellular dialysis of inorganic phosphate and alkaline phosphatase or NaN3 treatment further support the involvement of a phosphorylation mechanism. Contrasting with Cav1 and Cav2 Ca2+ channels, the Ca2+-dependent modulation of Cav3 channels appears to be independent of calmodulin, calcineurin and endocytic pathways. Similar findings were obtained for the native T-type Ca2+ current recorded in rat thalamic neurons of the central medial nucleus. Overall, our data reveal a new Ca2+ sensitive phosphorylation-dependent mechanism regulating Cav3 channels, with potentially important physiological implications for the multiple cell functions controlled by T-type Ca2+ channels.


Results
At high frequency stimulation, the Ca v 3.3 current displays a Ca 2+ -dependent inhibition that is dynamically regulated 15 . During stimulations applied at 1 Hz frequency, the Ca v 3.3 current decreased ~45% from its initial value after 40 seconds (p < 0.001, Fig. 1A,B) and this decrease was associated with faster inactivation kinetics (p < 0.001, Fig. 1C). Stopping the stimulation for 3 minutes allowed a full recovery of Ca v 3.3 current amplitude, after which Ca v 3.3 current decrease at fast stimulation could be reproduced several times (Fig. 1A, 2 nd and 3 rd ). Applying this protocol 3 times (Fig. 1A) allowed us to quantify both the current inhibition and its recovery after the 1 st and the 3 rd stimulation. Quantification of current inhibition and current recovery was calculated as the ratio of the current amplitude obtained at the 3 rd stimulation before and after the 1 Hz stimulation to the initial current amplitude obtained at the 1 st stimulation (I 1 s 1 st ; Fig. 1B). Taking advantage of the reversibility and reproducibility of the Ca 2+ -dependent inhibition of the Ca v 3.3 current, we investigated the mechanisms underlying current decrease as well as current recovery.  It has been demonstrated that a prolonged Ca 2+ entry induces an internalization of the L-type channels via endocytosis, possibly leading to degradation in the lysosome 21,25,26 . We tested whether this pathway was implicated in the inhibition of Ca v 3.3 current during fast stimulation ( Supplementary Fig. 1). Endocytosis was disrupted by expressing a dominant negative mutant of dynamin 1 (Dyn 1 K44A), a GTPase required for the formation of endocytic vesicles from the plasma membrane and implicated in the endocytosis of the L-type channel 25,28 . Expression of the Dyn 1 K44A mutant ( Supplementary Fig. 1B) did not alter the inhibition of the Ca v 3.3 current during fast stimulation (~45%, p < 0.001, red bars in Supplementary Fig. 1E) as compared to WT dynamin or the control condition (p > 0.05, Supplementary Fig. 1A,E). In addition, the recovery of the Ca v 3.3 current was not different in cells expressing the Dyn 1 K44A mutant (blue bars in Supplementary Fig. 1E) as compared to WT dynamin or the control condition (p > 0.05, Supplementary Fig. 1A,E). Similar findings were obtained after expression of the adaptor protein 2 (AP-2), a central player in clathrin-mediated endocytosis 29 , which did not modify current inhibition (~45%, p < 0.01) or its recovery ( Supplementary Fig. 1C,E), as compared to the control condition (p > 0.05, Supplementary Fig. 1E). In addition, we have treated the cells with the lysosomal inhibitor bafilomycin (100 nM, 3-5 hours), using the same protocol as described for L-type channels 26 . We found that bafilomycin did not affect Ca v 3.3 current inhibition at a high frequency of stimulation, nor its recovery ( Supplementary Fig. 1D,E). The analysis of the Ca v 3.3 current also indicated no significant effect of dynamin K44A, AP2 or bafilomycin treatment on current inactivation kinetics at the beginning as well as at the end of the fast stimulation protocol, as compared to the respective control condition (p > 0.05, n = 5-6).
Calmodulin (CaM) is important for the modulation of the L-type current [19][20][21][22] and has emerged recently as a modulator of T-type channels, including Ca v 3.3 [16][17][18] . To investigate this pathway we first used a CaMKII peptide (100 µM CaMKII 290-309) that binds CaM and inhibits its function and additional calmodulin-requiring enzymes. Dialysis of this peptide did not modify the Ca v 3.3 current inhibition at 1 Hz frequency of stimulation, nor its recovery (p > 0.05, as compared to the control condition, Fig. 2A,E). We next used a CaM mutant, CaM 1234 , which is unable to bind Ca 2+ and was previously used to demonstrate the CaM regulation of the L-type channel 19,20 . Expression of this mutant was also without significant effect on Ca v 3.3 modulation (Fig. 2B,E), indicating that CaM was not involved in the decrease of Ca v 3.3 current induced by stimulation at a high frequency or in its recovery. Similar results were obtained using the CaMKII inhibitory peptide (autocamtide-2-related inhibitory peptide, AIP, 100 µM intrapipette) and STO-609, a CaMKK inhibitor (1 µM incubation for 2 hours; Fig. 2C-E). In addition, the current inactivation kinetics at the beginning and at the end of the fast stimulation protocol were not significantly modified by CaMKII and AIP peptides or CaM 1234 and STO-609, as compared to CaMKII pept.
CaMKII pept.  . Asterisks indicate significant difference in the current amplitude, as compared to the initial current amplitude. In addition, the current inhibition (red bars) and recovery (blue bars), as function of the treatment, was statistically compared to the respective control condition, as indicated. The number of cells tested is indicated into brackets. (n.s.: non-significant). the respective control condition (p > 0.05, n = 5-8). However, we found that CaM 1234 expression induced a negative shift ~5.5 mV in the steady-state inactivation of the Ca v 3.3 current (n = 8, p < 0.005) compared to the control condition (n = 5), as recently reported for the Ca v 3.2 current 17 .
It was also demonstrated that during 1 Hz frequency stimulation, the L-type current decreases similarly to the Ca v 3.3 current and that this effect is mediated in part by calcineurin 23,24 . We have therefore tested whether this pathway was implicated in the Ca v 3.3 current decrease. Cells were treated with a combination of cyclosporin A and FK506 (10 µM each) before patch-clamp experiments, a treatment that prevents the decrease and the calcium-dependent inactivation (CDI) of the L-type current 23,24 . We found that 1-2 hours incubation with cyclosporin A and FK506 did not affect the decrease of the Ca v 3.3 current, nor its recovery, as compared to vehicle (DMSO)-treated cells (p > 0.05, Fig. 3A,B,E). Similarly, inclusion of the calcineurin inhibitory peptide (CIP, 100 µM in the patch pipette), which also prevents L-type current decrease and CDI 23,24 , did not affect the Ca v 3.3 current modulation (Fig. 3C,E). We also found that both inhibition and recovery of the Ca v 3.3 current were unaffected by okadaic acid (100 nM overnight incubation and 1 µM in the patch pipette) or by deltamethrin (10 µM in the patch pipette) (Fig. 3D,E). In addition, none of the compounds tested modified the current inactivation kinetics at the beginning and at the end of the fast stimulation protocol (p > 0.05, as compared to the control condition, n = 5-9). Similarly, no significant change on the steady-state inactivation of the Ca v 3.3 current was observed (p > 0.05, as compared to the control condition, n = 7-14). Overall, these findings strongly suggest that neither endocytosis nor CaM-dependent pathways nor calcineurin play a significant role in the Ca v 3.3 current modulation at a high frequency of stimulation, which therefore requires different molecular mechanisms than those previously described for the L-type channel.
In order to investigate further the role of a phosphorylation mechanism in Ca v 3.3 current modulation, we used inorganic phosphate to compete with phosphatase activities 30,31 . Inorganic phosphate (Pi, 10 mM KH 2 PO 4 ) was applied via the patch pipette 10 minutes before stimulation of the Ca v 3.3 current. In the presence of intracellular Pi, fast stimulation induced a slight current decrease ~20% (p < 0.05, Fig. 4B,C) whereas the current decrease was ~45% in the control condition (15 mM KCl in the patch pipette, p < 0.001, Fig. 4A,B). In average, the current decrease in the presence of Pi was ~35% less than the control value at the 3 rd stimulation (p < 0.01; Fig. 4C), whereas there was a slight but not significant increase in current recovery (Fig. 4C). In addition, during the stimulation at 1 Hz frequency, the acceleration in the inactivation kinetics was less pronounced in the presence of intracellular Pi as compared to the control condition (p < 0.05, Fig. 4D). It should be noted that in addition to competing with phosphatase activities, KH 2 PO 4 may bind Ca 2+ and therefore may act as a Ca 2+ buffer. However, n.s.
Deltamethrin n.s.  www.nature.com/scientificreports www.nature.com/scientificreports/ we observed a similar decrease in Ca v 3.3 current in the cells dialyzed with an intracellular solution containing 20 mM EGTA, as compared to our standard recording condition using 10 mM EGTA (p > 0.5, n = 17), suggesting that the Pi effect does not rely on its Ca 2+ buffer property. Moreover, these results suggest that a Ca 2+ -driven phosphatase might be responsible for the current decrease (see schematic representation in Supplementary Fig. 7). In agreement with the involvement of a phosphorylation mechanism, inclusion of recombinant alkaline phosphatase (AP, 100 U/ml) in the patch pipette 32 "mimicked" the inhibition of the Ca v 3.3 current at 1 Hz frequency stimulation (Supplementary Fig. 2A,B) and induced an acceleration in the current inactivation kinetics (Supplementary Fig. 2A-C). Importantly, the remaining current after AP treatment was stable during the 1 Hz frequency stimulation protocol (Supplementary Fig. 2A,B). In addition, the steady-state inactivation of the Ca v 3.3 current was largely shifted towards negative potentials after AP dialysis (~25 mV, p < 0.001; Supplementary Fig. 2D).
Considering that current inhibition could result from a phosphatase activity, we investigated whether a putative kinase activity might be implicated in current recovery (see schematic representation in Supplementary  Fig. 7). To this purpose, the cells were incubated in the presence of 10 mM sodium azide (NaN 3 ), which inhibits mitochondrial function 33 and thereby results in a depletion of intracellular ATP (Fig. 5A,B). In the presence of NaN 3 (and in the absence of ATP in the patch pipette), the 1 Hz stimulation induced a strong inhibition of the Ca v 3.3 current (~70%, p < 0.001, Fig. 5C), which is significantly larger than that observed in the control condition (p < 0.001, Fig. 5C). Interestingly the recovery of the current at the 3 rd stimulation was completely abolished after NaN 3 treatment and the residual current displayed fast inactivation kinetics (p < 0.001, Fig. 5C,D), suggesting that the recovery of the Ca v 3.3 current relies on an ATP-dependent mechanism. We therefore investigated whether the hydrolysis of the phosphate group from ATP is needed for current recovery. To this end we tested the effect of non-hydrolysable ATP analogs (Fig. 6). When intracellular ATP in the patch pipette was replaced by AMP-PCP, the inhibition of the Ca v 3.3 current (~55%, p < 0.001) at 1 Hz frequency stimulation was not significantly modified as compared to the control condition (p > 0.05, Fig. 6A,B,E), but the recovery of the Ca v 3.3 current was completely abolished (p < 0.001; Fig. 6A,B,E). Similar findings were observed using another non-hydrolysable ATP analog, AMP-PNP, which completely abolished the recovery of the Ca v 3.3 current (p < 0.001; Fig. 6C,E) but did not modify current inhibition as compared to the control condition (p > 0.05, Fig. 6E). In the presence of AMP-PCP or AMP-PNP, the remaining current after stimulation at high frequency displayed fast inactivation kinetics (p < 0.001, Fig. 6F). Because the depression of the current recovery may represent a spontaneous loss of channel activity (rundown) in the absence of intracellular ATP, we investigated the effect of AMP-PCP on the maintenance of Ca v 3.3 current. To this purpose, Ca v 3.3 currents were recorded during 30 minutes at a low frequency of stimulation (one stimulation each 180 s / 0.0055 Hz) in the presence of intracellular ATP or intracellular AMP-PCP ( Supplementary Fig. 3). At this low frequency of stimulation, there was no significant difference in Ca v 3.3 current rundown between control and AMP-PCP dialyzed cells ( Supplementary Fig. 3A). On average  www.nature.com/scientificreports www.nature.com/scientificreports/ channel activity was 94 ± 6% (n = 6) and 89 ± 6% (n = 6) of the initial value after 30 minutes of dialysis with 3 mM intracellular ATP and with 3 mM intracellular AMP-PCP, respectively. Thus, the decrease in Ca v 3.3 current recovery was not due to faster rundown of the current but to ATP dependency of current recovery from 1 Hz frequency stimulation. Accordingly, after 30 minutes of dialysis, the application of the 1 Hz stimulation protocol induced a decrease in the Ca v 3.3 current, which was irreversible in AMP-PCP dialyzed cells (p < 0.001, Supplementary  Fig. 3A). Importantly, in the presence of intracellular BAPTA (instead EGTA), which prevents a localized increase in submembrane Ca 2+ and the Ca 2+ -dependent inhibition of the Ca v 3.3 current 15 , we found that AMP-PCP did not produce current inhibition (p > 0.05, Fig. 6D-F) strongly indicating that the effect of non-hydrolysable ATP analogs relies on a Ca 2+ -dependent phosphorylation mechanism. It should be noted however, that NaN 3 and non-hydrolysable ATP analogs may also interfere with mitochondrial function and could lead to an irreversible oxidative stress that may underlie the Ca v 3.3 current inhibition observed at fast frequency of stimulation. In order to investigate the potential role of the redox status in the current inhibition, the cells were dialyzed with an intracellular medium containing 1 mM dithiothreitol (DTT). After dialysis of DTT, the 1 Hz stimulation protocol induced inhibition of Ca v 3.3 current amplitude (~40%, p < 0.001, Supplementary Fig. 4B,C) that was associated with an acceleration of current inactivation kinetics (p < 0.001, Supplementary Fig. 4D). These effects were not different to those observed in the control condition (p > 0.05, Supplementary Fig. 4C,D). Similarly, current recovery was unaffected ( Supplementary Fig. 4C), suggesting that current modulation does not depend on the redox status of the Ca v 3.3 channel. We noted however that in the presence of DTT, the inactivation kinetics were slower at the beginning of the 1 st stimulation as compared to the control condition (p < 0.05, n = 7) but this effect was not significant at the 3 rd stimulation ( Supplementary Fig. 4D).
Several pathways regulating Ca v 3 current were identified, including PKA 34-36 , PKC 35,37,38 , Rho-associated kinase 39 , phospholipase C (PLC 40 ), and Gβγ 36,41 . We have tested whether these pathways could be implicated in the Ca v 3.3 modulation at high frequency of stimulation using dibutyryl-cAMP (db-cAMP, a PKA activator), phorbol myristate acetate (PMA, a PKC activator), chelerythrine and bisindolylmaleimide IX (BIM IX, both PKC inhibitors), fasudil (a Rho-associated kinase inhibitor), U73122 and edelfosine (both PLC inhibitors) and pertussis toxin (PTX) or expression of the βARK-Ct (both inhibiting Gβγ). In the presence of these compounds, the inhibition (~50%, p < 0.01, Supplementary Fig. 5A) of the Ca v 3.3 current was not different to that observed in the control condition (p > 0.05, Supplementary Fig. 5A). Importantly, these inhibitors did not modify the recovery of the Ca v 3.3 current as compared to the control condition (p > 0.05, Supplementary Fig. 5B). Similarly, the inhibition of PI3K and PI4K with wortmannin or the inhibition of SERCA 42 with cyclopiazonic acid (CPA) was without effect on both inhibition and recovery of the Ca v 3.3 current (p > 0.05 as compared to the control condition, Supplementary Fig. 5A,B). Collectively, these results indicate that the Ca v 3.3 current modulation at high  www.nature.com/scientificreports www.nature.com/scientificreports/ frequency of stimulation involves a yet undescribed transduction pathway. We next tested several other kinases that could possibly modulate Ca v 3.3 current. We found that the presumed inhibition of PDK1 with OSU-03012, of AKT with AKT1/2 inhibitor, of PLK1/3 with GW843682X and BI 2536, of RSK with BI-D1870, of TAK1 with 5Z-oxozeanol, of MNK1 with CGP57380, of PYK2/FAK with PF-431396, of SGK with GSK 650394, of GSK3 with CT 99021, of SRC with SU6656, of eEF2 kinase with A484954 and of MLCK with ML7, was without significant effects on both inhibition and recovery of the Ca v 3.3 current after the 1 Hz stimulation (p > 0.05 as compared to the values obtained in the control condition, Supplementary Fig. 6A,B).
We also investigated the ATP-dependency of the recombinant Ca v 3.1 current during fast stimulations (Fig. 7). In the presence of intracellular ATP, the Ca v 3.1 current decreased ~15% from its initial value during the 1 Hz frequency stimulation (p < 0.05; Fig. 7A,C) with an acceleration of its inactivation kinetics (p < 0.05; Fig. 7A,D). Importantly, the Ca v 3.1 current recovery was almost total when the stimulation ceased for 3 minutes, as illustrated at the beginning of the 3 rd stimulation (Fig. 7A,C). In contrast, in AMP-PCP dialyzed cells, the Ca v 3.1 current inhibition became irreversible (p < 0.001 as compared to the control condition, Fig. 7B,C) and the remaining current displayed faster inactivation kinetics (p < 0.01, Fig. 7B,D). In addition, Ca v 3.1 current inhibition in the presence of AMP-PCP (~35% inhibition, p < 0.001) was more potent than that observed in ATP dialyzed cells for the 1 st and the 3 rd stimulation (p < 0.05, Fig. 7B,C), indicating that current inhibition is a very dynamic process. Importantly, in cells dialyzed with AMP-PCP + BAPTA, the stimulation at 1 Hz did not produce current inhibition and the inactivation kinetics remained unchanged (p > 0.05, Fig. 7C,D).
We have previously demonstrated that current inhibition relies mainly on a hyperpolarizing shift in the steady-state inactivation of the Ca v 3.3 current 15 . Therefore, we have measured the steady-state inactivation of the Ca v 3.3 current in the presence of non-hydrolysable ATP analogs. We found that, after the 3 rd 1 Hz stimulation in the presence of AMP-PCP, there was a strong hyperpolarizing shift ~20 mV in the steady-state inactivation of the Ca v 3.3 current (Fig. 8A,B). In average, the V 0.5 values of inactivation were −69.2 ± 0.6 mV in the control condition and −88.9 ± 1.1 mV in the presence of AMP-PCP (p < 0.001; Fig. 8C). Similar findings were obtained for AMP-PNP dialyzed cells and for NaN 3 treated cells (Fig. 8C) since the V 0.5 values of the steady-state inactivation curves were −88.7 ± 2.2 mV (p < 0.001) and −88.2 ± 2.7 mV (p < 0.001), respectively. In contrast, we observed only a slight depolarizing shift in the presence of Pi (V 0.5 was −64.5 ± 2.3 mV; Fig. 8C), which was not www.nature.com/scientificreports www.nature.com/scientificreports/ significant. Importantly, we found that in the presence of intracellular BAPTA, AMP-PCP produced no significant shift in the steady-state inactivation curve of the Ca v 3.3 current (p > 0.05, Fig. 8C). Similar findings were obtained for the recombinant Ca v 3.1 current since the steady-state inactivation was significantly shifted towards negative potentials by ~8 mV in the presence of AMP-PCP and this shift was abolished in the presence of BAPTA (Fig. 8D). In average we found that the V 0.5 values of the steady-state inactivation curves for the Ca v 3.1 current were −80.6 ± 1.3 mV in the control condition, −88.3 ± 1.7 mV for the AMP-PCP dialyzed cells (p < 0.01) and −75.1 ± 0.4 mV for the cells dialyzed with AMP-PCP combined with BAPTA (p < 0.05 as compared to the control condition, and p < 0.0001 as compared to the AMP-PCP dialyzed cells).
We next explored whether a similar regulation of the native T-type current may occur in thalamic neurons. The effects of intracellular AMP-PCP and BAPTA were investigated in acute brain slices focusing on the T-type current expressed in the central medial nucleus (CeM) neurons of the thalamus, that are enriched in Ca v 3.1 currents 27 . In the presence of AMP-PCP, the T-type current amplitude in CeM neurons decreased during stimulation at the frequency of 1 Hz (Fig. 9A,B) and reached ~15% inhibition after 40 s of stimulation (p < 0.01, Fig. 9B,C). This decrease in the T-type current amplitude was prevented by the substitution of intracellular EGTA with BAPTA suggesting the role of local increase in intracellular Ca 2+ concentration in current inhibition (Fig. 9B,C). As observed for recombinant T-type channels, steady-state inactivation of the T-type current in CeM neurons was shifted towards the negative potentials in the presence of AMP-PCP (Fig. 9D,E). This shift in the steady-state inactivation was similar to the one observed when ATP was omitted in the intracellular solution (Fig. 9E), suggesting that the ATP effect relies mainly on a putative phosphorylation mechanism. In average, the V 0.5 values of the steady-state inactivation curves were −75.4 ± 1.6 mV in the control condition with intracellular ATP, −82.3 ± 1.9 mV for the cells dialyzed with an ATP-free solution (p < 0.05) and −83.7 ± 1.65 mV for the cells dialyzed with AMP-PCP (p < 0.01, Fig. 9E). Importantly, the shift in the steady-state inactivation of the T-type current did not occur when the intracellular solution contained AMP-PCP in the presence of BAPTA (p > 0.05 as compared to the control conditions, Fig. 9E). Overall, these data strongly suggest that the availability of both www.nature.com/scientificreports www.nature.com/scientificreports/ recombinant and native T-type current depends on a phosphorylation process, which is regulated by the local concentration of intracellular calcium (see schematic representation in Supplementary Fig. 7).

Discussion
Recently, we demonstrated that the availability of the Ca v 3 channels can be dynamically tuned by changes in the intracellular Ca 2+ concentration at the vicinity of these channels 15 . Here we describe that this Ca 2+ -induced modulation of the Ca v 3 current likely relies on a phosphorylation mechanism. This modulation occurs for both Ca v 3.3 and Ca v 3.1 recombinant channels, as well as for native neuronal T-type channels. Importantly, this modulation of T-type channels by intracellular Ca 2+ is novel among the Ca 2+ channel family as it involves a phosphorylation process not related to calcineurin, endocytosis and calmodulin-dependent pathways.
The regulation of the high-voltage-activated Ca v 1 and Ca v 2 Ca 2+ channels by a rise in intracellular Ca 2+ has been extensively studied and serves as a reference 13,14,[19][20][21][23][24][25][26] . For the L-type / Ca v 1.2 channel a rise in intracellular Ca 2+ induces complex effects depending on the amount of the calcium ions. Ca 2+ entry via L-type channels first induces a Ca 2+ -dependent inactivation (CDI) characterized by an acceleration of the inactivation kinetics of the L-type current that is followed, if the stimulation persists, by a decrease/rundown of the current. It was established that both CDI 14,[19][20][21]24 and rundown 23,24 of the L-type current are dependent on calmodulin activity. Clearly, the modulation of the Ca v 3.3 current described here is independent of the calmodulin-dependent pathway, contrasting with the well-established Ca 2+ -dependent mechanisms regulating the L-type channels.
Importantly recent studies have revealed that calmodulin interacts with Ca v 3 channels [16][17][18] . We have previously reported that the steady-state inactivation of the Ca v 3.2 current was negatively shifted ~5 mV towards negative potentials in the presence of the CaM 1234 mutant that does not bind Ca 2+17 , and we find here that this is also the case with the Ca v 3.3 current. However, in the presence of this mutant, the inhibition of Ca v 3.3 current at a high frequency of stimulation is preserved as well as the current recovery after the stopping of the stimulation. Similar findings were obtained using a CaMKII peptide that binds CaM and inhibits its function. In contrast with our results, another study found that in the presence of 1 µM intracellular Ca 2+ , the steady-state inactivation of www.nature.com/scientificreports www.nature.com/scientificreports/ calcineurin was also shown to interact with Ca v 3.2 channels and Ca 2+ entry via these channels induces the calcineurin-dependent activation of the transcription factor NFAT [43][44][45] . In addition, both L-type current decrease during fast stimulation and CDI were prevented after calcineurin inhibition 23,24 , suggesting that calcineurin may mediate inhibition of the Ca v 3.3 current. However, prolonged treatment with several calcineurin inhibitors was without effect on Ca v 3.3 current modulation, confirming a study on T-type currents in sensory neurons 46 . It was also demonstrated that Ca 2+ entry via Ca v 3 channels can promote a Ca 2+ -dependent CaM dissociation from the channels leading to CaMKII activation, which is blocked using the CaMKII inhibitory peptide AIP 16 . In addition, in the presence of micromolar concentrations of intracellular Ca 2+ , CaMKII can regulate Ca v 3.2 current by phosphorylation of serine residues 1198 and 1153 in the channel's 2-3 intracellular loop [47][48][49] . This regulation is abolished in the presence of the AIP peptide or by substitution of intracellular ATP by AMP-PNP 47 . However, this latter regulation of the Ca v 3.2 current is clearly different from the one described here for the Ca v 3.3 current, since Ca 2+ / CaMKII activation promotes an increase (but not inhibition) of the Ca v 3.2 current by promoting a negative shift in the activation curve without any effect on the steady-state inactivation. In addition, this Ca v 3.2 regulation involves CaMKIIγc, which is not expressed in HEK-293 cells 47 . Accordingly, we found that the dialysis of the AIP peptide had no effect on both inhibition and recovery of the Ca v 3.3 current. Therefore, although CaM/ Calcineurin/CaMKII pathways interact with Ca v 3 channels, our data strongly support the idea that the modulation of the Ca v 3.3 current at a high frequency of stimulation requires additional Ca 2+ -dependent mechanisms.
Endocytosis could be another mechanism leading to the decrease of the Ca v 3.3 current. Indeed, for the L-type channels, a prolonged current stimulation or activation of the ionotropic NMDA receptors induces the internalization of the channels and potentially their degradation in lysosomes or their recycling to the plasma membrane depending of the amount and the duration of the Ca 2+ entry 21,25,26 . These studies have highlighted the crucial role of dynamin 1, a GTPase required for the formation of endocytic vesicles from the plasma membrane 28 , in the early steps of this process 21,25,26 . In the presence of a dominant negative mutant of dynamin 1 (Dyn 1 K44A), which abolishes the endocytosis of the L-type channel 25 , the inhibition of the Ca v 3.3 current during 1 Hz stimulation is not affected as well as its recovery when the stimulation ceases. In addition, a role of lysosomal degradation is unlikely since Ca v 3.3 current inhibition and its recovery are preserved after treatment of the cells with bafilomycin, which prevents degradation of the L-type channel 26 . Therefore, although the regulation by intracellular Ca 2+ of L-and T-type channels share similar features, the mechanism underlying the Ca 2+ dependent inhibition of T-type channels during fast stimulation is clearly different from the one described for L-type channels.
We found that the Ca v 3.3 current modulation observed at a high frequency of stimulation likely relies on a phosphorylation mechanism (see schematic representation in Supplementary Fig. 7). Accordingly, the inhibition of the Ca v 3.3 current during fast stimulation is decreased in a presence of intracellular Pi, which is a competitive inhibitor of phosphatase activities 30,31 . Conversely, sodium azide, a pharmacological agent that promotes the depletion of intracellular ATP by the inhibition of cytochrome oxidase 33 , the final enzyme in the mitochondrial electron transport chain, abolishes the recovery of the Ca v 3.3 current after its inhibition induced by the fast stimulation. Therefore, we hypothesized that ATP is necessary in the recovery process. We used non-hydrolysable ATP analogs, AMP-PNP and AMP-PCP, which are competitive inhibitors of reactions requiring hydrolysable ATP 50 . The substitution of intracellular ATP with AMP-PNP or AMP-PCP, abolishes the recovery of the Ca v 3.3 current, indicating the crucial role of a phosphorylation-dependent mechanism in the Ca 2+ -induced inhibition of the Ca v 3.3 current. This depression of recovery in AMP-PCP dialyzed cells might be a consequence of a rundown of the Ca v 3.3 current during the experiment leading to an apparent inhibition of recovery after fast stimulation. However, we found a similar rundown of the Ca v 3.3 current when stimulated at a low frequency in ATP and in AMP-PCP dialyzed cells, indicating that the effect of AMP-PCP depends on the previous inhibition of the Ca v 3.3 current by a fast stimulation. Accordingly, in the presence of intracellular BAPTA, which prevents a local submembrane rise in Ca 2+ 51-53 , AMP-PCP did not produce current inhibition indicating that the effect of non-hydrolysable ATP analogs relies on a Ca 2+ -dependent mechanism.
We used different pharmacological antagonists to examine the potential involvement of various intracellular pathways underlying the Ca v 3.3 current modulation. Interestingly, the recovery of the Ca v 3.3 current after fast stimulation does not depend of the previously identified kinases that modulated the Ca v 3 current, including PKA, PKC and Rho-associated kinase [34][35][36][37]39 . The recovery of the Ca v 3.3 current is also resistant to the inhibition of PI3K / PI4K, PDK1, AKT, PLK1/3, RSK, TAK1, PYK2/FAK, SGK, GSK3, SRC, eEF2 or MLCK. Therefore, further extensive biochemical and electrophysiological studies are needed to identify the precise signaling pathways that modulate the Ca v 3.3 current.
We have previously demonstrated that a rise in submembrane Ca 2+ ions induces a hyperpolarizing shift in the steady-state inactivation of the Ca v 3.3 current, a regulation that is also found for the Ca v 3.1 current 15 . Here we show that sodium azide, AMP-PNP and AMP-PCP treatment, induced a large ~20 mV hyperpolarizing shift in the steady-state inactivation of the Ca v 3.3 current, confirming a common mechanism. Accordingly, AMP-PCP has no effect on the steady-state inactivation of the Ca v 3.3 current in cells dialyzed with BAPTA. Regarding the Ca v 3.1 current, its inhibition during stimulation at a high frequency in the presence of intracellular ATP is modest (~15%) compared to Ca v 3.3, most likely due to its rapid inactivation kinetics leading to a moderate increase in intracellular Ca 2+ 54-56 . As observed for the Ca v 3.3 current, the inhibition of the Ca v 3.1 current is irreversible in the presence of intracellular AMP-PCP whereas the inhibition is prevented in the presence of BAPTA. In addition, the steady-state inactivation curve of the Ca v 3.1 current is shifted by ~8 mV towards negative potentials in the presence of AMP-PCP. Therefore, our data reveal that the T-type channel modulation by intracellular Ca 2+ has unusual features among the Ca 2+ channel family both in its transduction and its underlying biophysical mechanisms.
Importantly, we found that the native Ca v 3.1 T-type current recorded in neurons of the central medial nucleus (CeM) of the thalamus 27 , is subject to a similar Ca 2+ -dependent regulation. When stimulated at a frequency of 1 Hz, the T-type current in CeM neurons decreases ~15% in the presence of AMP-PCP and this effect is abolished in www.nature.com/scientificreports www.nature.com/scientificreports/ the presence of intracellular BAPTA. As observed for recombinant channels, this inhibition relies on a shift of the steady-state inactivation since the native T-type current displays a ~8 mV negative shift in the steady-state inactivation curve in the presence of AMP-PCP, which is abolished in the presence of BAPTA. This shift is similar to the one observed in cells dialyzed with an ATP-free solution, indicating that ATP effect relies on a phosphorylation mechanism. Interestingly, it was previously reported in thalamocortical neurons of the ventrobasal nucleus that the steady-state inactivation of the T-type current progressively shifts towards negative potentials during the dialysis of an ATP-free solution, indicating that the ATP-dependency of the T-type channel availability is not restricted to CeM neurons 57 . Overall, our study demonstrates that the availability of both recombinant and native T-type channels depends on a phosphorylation process regulated by the local intracellular Ca 2+ concentration ( Supplementary  Fig. 7). This Ca 2+ -dependent inhibition of T-type channel would provide an important feedback mechanism during sustained neuronal activities to limit intracellular Ca 2+ overload. It should be noted however that although our study highlights the role of a putative phosphorylation mechanism, whether the Ca v 3 channel is directly phosphorylated or modulated via an intermediate protein remains an open question. Future molecular and biochemical studies will be necessary to clarify this important aspect and identify the key residues implicated in the Ca v 3 current regulation.
By demonstrating that the availability of the T-type channels critically depends on an interplay between Ca 2+ entry and the cellular phosphorylation status, we document a novel and important mechanism that tightly and dynamically controls T-type channel activity. Considering that T-type channels are widely expressed in the central nervous system where they contribute to the rebound burst firing activities 6,7 , and that an alteration in their activity is implicated in several neuronal disorders, including epilepsy, schizophrenia, autism, chronic pain and cerebellar ataxia 2,8,10-12 , this phosphorylation-dependent Ca 2+ -sensitive mechanism might be crucial to accurately balance the electrical and Ca 2+ signaling of T-type channels both in their normal physiological responses and in disease states involving these channels.
In vitro brain slice preparation from juvenile rats. Experimental procedures with animals were performed as described previously 27 according to the protocol #B-111616 (02) 1E approved by the Animal Care and Use Committee at the University of Colorado Anschutz Medical Campus. Treatments of rats adhered to guidelines set forth in the NIH Guide for the Care and Use of Laboratory Animals. All efforts were made to minimize animal suffering and to use only the number of animals necessary to produce reliable scientific data. Experiments were performed on male and female Sprague-Dawley rats (P23-P26) obtained from Envigo (Indianapolis, IN, USA). Brain slice preparation was obtained as described previously 27 . Brain slices were immediately incubated for 30 min in the following solution (in mM): 124 NaCl, 10 D-glucose, 26 NaHCO 3 , 1.25 NaH 2 PO 4 , 4 KCl, 2 CaCl 2 , 2 MgCl 2 at 37 °C before use in electrophysiology experiments, which were done at room temperature. During incubation, slices were constantly perfused with a gas mixture of 95 vol% O 2 and 5 vol% CO 2 .
Electrophysiological recordings in CeM neurons. Whole-cell recordings were performed in CeM neurons, as described previously 27  Data analysis. Current traces were analyzed using pCLAMP9 (Molecular Devices) and GraphPad Prism (GraphPad) softwares. Steady-state inactivation curves were fitted using the Boltzmann equation where I/I max = 1/(1 + exp((Vm−V 0.5 )/slope factor)). Results are presented as the mean ± SEM, and n is the number of cells. Statistical analysis were performed with two-way ANOVA combined with a Tukey post-test for multiple comparisons, excepted in Fig. 1, for which a one-way ANOVA combined with a Tukey post-test for multiple comparisons was used (*p < 0.05, **p < 0.01, ***p < 0.001).