Mono-ADP-ribosylation by ARTD10 reduces PKCδ activity
To confirm PKCδ as a substrate of ARTD10, we performed in vitro ADP-ribosylation assays with the purified catalytic domain of ARTD10 fused to GST (GST-ARTD10 CAT) and of PKCδ fused to a 6-His-tag (His-PKCδ; Fig. 1a). As expected, ARTD10 MARylated itself. In addition, PKCδ was also MARylated, while BSA as a negative control was not. Although MARylation of PKCδ was less effective than auto-MARylation of ARTD10 (Fig. 1a), this result confirms PKCδ as a substrate of ARTD10 in vitro.
Next, we analyzed PKCδ MARylation in cells. GFP-PKCδ and HA-ARTD10 or a catalytically inactive mutant (HA-ARTD10-G888W) [25] were transiently expressed in HEK293 cells, GFP-PKCδ was enriched using a GFP-TRAP, and MARylation was analyzed by immunoblotting. PKCδ was MARylated in the presence of ARTD10 but not when PKCδ was expressed alone or in the presence of ARTD10-G888W (Fig. 1b). Moreover, MARylation of PKCδ when co-expressed with ARTD10 was reduced in the presence of the selective ARTD10 inhibitor OUL35 [26]. These results demonstrate MARylation of PKCδ by ARTD10 in living cells. Further, ARTD10 as well as ARTD10-G888W were co-immunoprecipitated with PKCδ using the GFP-TRAP from lysates (Fig. 1b). Thus, this interaction was independent of catalytic activity of ARTD10. In the absence of PKCδ no ARTD10 was trapped, indicating that the interaction was specific.
HeLa cells express the classical PKC isoform PKCα, the atypical isoform PKCζ, as well as the novel isoform PKCδ [27]. Phosphorylation at tyrosine 311 (Y311) between the regulatory and catalytic domains of PKCδ is a critical step for its activation [28] and can be used as a read-out of PKCδ activity. To further corroborate MARylation of PKCδ in living cells, we therefore used HeLa cells stably expressing either wildtype (WT) ARTD10 or mutant ARTD10-G888W [25] and compared the phosphorylation status of PKCδ in these cells with that of control HeLa cells. While the total abundance of PKCδ was similar in all three cell lines, the abundance of PKCδ phosphorylated at Y311 was reduced in ARTD10-WT expressing cells compared with control HeLa cells or cells expressing ARTD10-G888W (Fig. 1c). This finding indicates that MARylation of PKCδ causes dephosphorylation of PKCδ. It is reminiscent of the ARTD10 effect on GSK3β, which leads to decreased phosphorylation and decreased kinase activity [7]. Given that phosphorylation at Y311 activates PKCδ [28], our results therefore suggest that MARylation of PKCδ by ARTD10 reduces its catalytic activity.
PKCδ reduces the proportion of inactivating Kv1.1 currents via dephosphorylation at S446 of the α subunit
To investigate the modulation of the voltage-gated K+ channel Kv1.1 by PKC in HeLa cells, we transiently expressed Kv1.1 in HeLa cells and evoked K+ currents by a depolarization from − 80 mV to + 40 mV. Transient expression of Kv1.1α alone induced robust non-inactivating K+ currents; additional co-expression of the β1.1 subunit conferred fast and partial inactivation to these currents. We quantified the proportion of the non-inactivating current component by building the ratio of the current amplitude at the end of a 200-ms pulse to +40 mV (Isteady-state) to the peak current at the beginning of this pulse (Ipeak; Fig. 2a). Inactivation was best fit with a bi-exponential function, yielding a fast τ1 = 7.2 ± 0.3 ms and a slow τ2 = 82.9 ± 8.1 ms (n = 6). Peak currents were similar with and without the β subunit (Fig. 2a). In cells co-expressing Kv1.1α and Kvβ1.1, 5 min application of IBMX and forskolin, to increase cAMP levels and activate PKA, decreased the proportion of the non-inactivating K+ current (Isteady-state/Ipeak, from 0.55 ± 0.08 to 0.42 ± 0.09, n = 8, p = 0.005, paired Student’s t test), while 5-min application of the phorbol ester phorbol 12-myristate 13-acetate (PMA) to activate PKC increased it (Isteady-state/Ipeak, from 0.40 ± 0.04 to 0.49 ± 0.05, n = 8, p = 0.022, paired Student’s t test; Fig. 2b). This is in agreement with previous reports [21]. To confirm the importance of phosphorylation at S446 for the inactivation, we co-expressed the phosphorylation-deficient mutant Kv1.1α-S446A together with Kvβ1.1. Compared to wild-type Kv1.1α, the proportion of the non-inactivating component was strongly increased for Kv1.1α-S446A (Isteady-state/Ipeak, 0.85 ± 0.02 vs. 0.51 ± 0.09, n = 5, p = 0.006; Fig. 2c), although the β subunit binds equally well to Kv1.1α-S446A and to wild-type Kv1.1α [23]. These results confirm that phosphorylation of S446 promotes inactivation [23]. Peak current amplitudes were similar (Fig. 2c). Moreover, a phosphatase inhibitor cocktail strongly decreased the proportion of non-inactivating K+ currents in HeLa cells expressing Kv1.1 (Isteady-state/Ipeak, 0.31 ± 0.05 vs. 0.62 ± 0.05, n = 5, p = 0.002; Fig. 2d), suggesting that phosphorylation/dephosphorylation of S446 is dynamic in these cells.
To directly show enhanced Kv1.1 phosphorylation after stimulation of PKA, we analyzed Kv1.1α by Western blotting. It has been reported that the phosphorylated form of Kv1.1 migrates at 57 kD and the un-phosphorylated form at 54 kD [21, 23]. After expression in HeLa cells, two bands with an apparent molecular weight of approximately 60 kDa were observed (Fig. 2e). The ratio between the higher band, corresponding to phosphorylated Kv1.1α, and the lower band, corresponding to un-phosphorylated Kv1.1α, was strongly increased in HeLa cells stimulated with IBMX and forskolin compared to control cells (Fig. 2e). PMA application only slightly reduced phosphorylation. In contrast, in cells expressing the phosphorylation-site mutant Kv1.1-S446A, only the smaller band, corresponding to un-phosphorylated Kv1.1α, was observed (Fig. 2e). These results confirm the phosphorylation of Kv1.1α at S446 by PKA.
In HeLa cells, phorbol esters activate PKCδ and PKCα, but not the atypical isoform PKCζ [27]. To confirm that PKCδ is the subtype of PKC, which was responsible for the reduced extent of inactivation after PMA treatment, we co-expressed Kv1.1 together either with a constitutively active catalytic domain of PKCδ (PKCδ-CAT), or with a dominant negative PKCδ (PKCδ-DN) [29]. Cells co-expressing PKCδ-CAT had a large non-inactivating current component while cells co-expressing PKCδ-DN had a small non-inactivating component (Isteady-state/Ipeak, PKCδ-CAT = 0.77 ± 0.04, vs. PKCδ-DN = 0.39 ± 0.07; n = 10 for PKCδ-CAT, n = 11 for PKCδ-DN, p < 0.001; Fig. 3a), in agreement with a modulation of the extent of inactivation by PKCδ. The peak current amplitude was not significantly different between both groups (PKCδ-CAT = 1.94 ± 0.31 nA vs. PKCδ-DN = 1.86 ± 0.35 nA p = 0.99; Fig. 3a). When the phosphorylation-deficient mutant Kv1.1-S446A was co-expressed with either PKCδ-CAT or PKCδ-DN, both groups of cells had a similarly large proportion of non-inactivating currents (Fig. 3b), confirming that PKCδ reduces inactivation of Kv1.1 via inducing dephosphorylation of Kv1.1α at S446 in HeLa cells.
ARTD10 increases the inactivating current component of Kv1.1 via phosphorylation at S446
To investigate the effect of PKCδ MARylation on Kv1.1, we transiently expressed Kv1.1 in control HeLa cells, in HeLa cells stably expressing ARTD10-WT and in HeLa cells stably expressing ARTD10-G888W. The non-inactivating current component of Kv1.1 was significantly smaller in cells expressing ARTD10-WT than in either control cells or cells expressing ARTD10-G888W (Isteady-state/Ipeak, ARTD10 = 0.32 ± 0.03 vs. control = 0.56 ± 0.08, n = 6, and ARTD10-G888W = 0.56 ± 0.06, n = 7, p = 0.04; Fig. 4a). This is consistent with a stronger phosphorylation of the Kv1.1α at S446 and thus with a decreased PKCδ activity in ARTD10-expressing cells. Moreover, there was a tendency for a larger peak current amplitude in ARTD10-expressing cells than in control cells (5.12 ± 0.61 nA vs. 3.82 ± 0.49 nA, n = 6, p = 0.13); amplitudes of Kv currents in ARTD10-G888W-expressing cells were significantly smaller than in cells expressing ARTD10 wild-type (2.88 ± 0.66 nA, n = 7, p = 0.03) (Fig. 4a). These results suggest increased Kv1.1 activity after phosphorylation at S446, as has previously been observed [23]. When the phosphorylation-deficient mutant Kv1.1-S446A was expressed, no difference between groups was observed (Fig. 4b), demonstrating that phosphorylation at S446 was indeed important for the modulation of Kv1.1 inactivation by ARTD10. This observation also supports the conclusion that the ARTD10 effect is mediated by PKCδ.
To directly show enhanced Kv1.1 phosphorylation in ARTD10-expressing cells, we analyzed Kv1.1α by Western blot. In ARTD10-expressing HeLa cells, the ratio between the higher band, corresponding to phosphorylated Kv1.1, and the lower band, corresponding to un-phosphorylated Kv1.1, was significantly larger compared to control or to ARTD10-G888W-expressing cells (Fig. 4c), directly demonstrating that the presence of ARTD10 increases phosphorylation of Kv1.1.
An ARTD10 inhibitor decreases inactivation of K+ currents and increases excitability of hippocampal neurons
Kv1.1 together with other Kv1α and Kvβ subunits are widely expressed in the brain [10]. Kv1 channels have been extensively studied in hippocampal CA1 neurons, where they mediate the D-type K+ current (“delay current”) [11, 30]. Although different Kv1α subunits may contribute to the D current, it has been reported that in hippocampal neurons cultivated for 6–11 days in vitro (DIV) Kv1.1 is the only Kv1 protein expressed [31]. ARTD10 is expressed at low levels in many brain regions, including hippocampus. PKCδ has a relatively low expression in mouse hippocampus at postnatal days 2–6, but its expression increases strongly afterwards [32]. We, therefore, focused on hippocampal neurons to investigate the modulation of Kv1 channel function and neuronal excitability by ARTD10. We treated isolated hippocampal neurons in culture with the selective inhibitor of ARTD10, OUL35 (3 μM) [26], at DIV 7 and performed patch clamp recordings after 3 to 5 days (DIV 10-12). K+ currents were evoked by depolarizing neurons from − 80 mV to + 40 mV. K+ currents elicited in hippocampal neurons resembled Kv1.1 currents recorded in HeLa cells, but inactivated slightly more slowly (τ1 = 17.8 ± 1.4 ms, τ2 = 175.5 ± 35.5 ms, n = 7). Strikingly, voltage-gated K+ currents in neurons incubated with OUL35 had significantly decreased peak amplitudes (1.22 ± 0.13 nA, n = 6, vs. 1.99 ± 0.33 nA, n = 7, p = 0.04) and an increased proportion of non-inactivating currents (Isteady-state/Ipeak, OUL35 = 0.67 ± 0.04 vs. DMSO control = 0.55 ± 0.03, p = 0.02; Fig. 5a), which is expected if inhibition of ARTD10 would increase PKCδ activity in these neurons and concomitantly reduce phosphorylation at S446 of Kv1.1. Thus, voltage-clamp experiments suggested that the ARTD10-inhibitor OUL35 affected the D-current in hippocampal neurons.
We then assessed the intrinsic excitability of cultured hippocampal neurons recorded under current clamp in a perforated patch configuration. First, we noticed that neurons treated with OUL35 had a significantly depolarized membrane potential as compared to untreated neurons (− 61.8 ± 2.4 mV vs. − 69.5 ± 2.7 mV, p = 0.04; Fig. 5b left). Although Kv1.1 is not expected to strongly control the resting membrane potential, it has previously been reported that PKA phosphorylation of Kv1.1 negatively shifts the resting membrane potential [22]. The positive shift of the membrane potential that we observed in OUL35 treated neurons would thus be compatible with a reduced phosphorylation of Kv1.1. The slightly depolarized membrane potential is expected to bring neurons closer to the threshold potential. Indeed, almost half of the OUL35-treated neurons showed spontaneous action potentials (APs), while only 17% of neurons without treatment showed spontaneous APs (p = 0.028, Fisher’s exact test; Fig. 5c). Within the pool of spontaneously active neurons, neurons incubated with OUL35 also had a tendency for an increased AP frequency (3.1 ± 1.3 spikes/s vs. 1.15 ± 0.33 spikes/s, p = 0.27; Fig. 5b right). When we excluded strongly depolarized cells with a resting membrane potential more positive than − 60 mV, only 33% of the OUL35-treated neurons showed spontaneous APs (n = 15) compared with 15% of the control neurons (n = 27; p = 0.24, Fisher’s exact test) and there was still a tendency for an increased AP frequency (1.87 ± 0.78 spikes/s vs. 2.91 ± 1.10 spikes/s, p = 0.25).
To further test the excitability of hippocampal neurons, we used current clamp and a step protocol to apply current pulses of gradually increasing amplitude and determined the current that elicited the first AP (rheobase). Neurons incubated with OUL35 for 72 h displayed a robust increase in excitability with a significant decrease in rheobase (11 ± 2.9 pA, n = 9, vs. 23.8 ± 3.9 pA, n = 13, p = 0.02; Fig. 6a). Moreover, counting the numbers of APs generated by current injections of higher amplitude revealed that the neurons treated with OUL35 had an increased AP frequency compared with the control (Fig. 6b). In both groups, AP frequency started to plateau at current pulses of 40 pA and decreased at higher amplitude pulses, probably due to a depolarization block. The amplitude of APs was increased in OUL35-treated neurons (Fig. 6b), which is consistent with a reduced repolarizing current.
To ascertain the importance of Kv1 channels for the increase in excitability of hippocampal neurons by OUL35, we inhibited D-type K+ currents with the snake toxin α-dendrotoxin (DTX; 100 nM), a specific blocker of channels containing Kv1.1, Kv1.2 or Kv1.6 subunits [33]. Like previously reported [34], in control neurons, DTX application decreased the rheobase (from 16.38 ± 2.52 pA to 11.75 ± 2.31 pA, p = 0.008, paired Student’s t test; Fig. 6c left). In contrast, in neurons treated with OUL35, DTX had no effect on the rheobase (5.83 ± 1.27 pA vs. 6.57 ± 1.51 pA, p = 0.54; Fig. 6c). OUL35, however, robustly decreased the rheobase (6.57 ± 1.51 pA vs. 16.38 ± 2.52 pA, p = 0.006; Fig. 6c), as we observed before (Fig. 6a). These data show that OUL35 occluded DTX-sensitivity of hippocampal neurons, suggesting that OUL35 reduced the activity of native DTX-sensitive Kv1 channels. The stronger reduction in rheobase by OUL35 than by DTX (Fig. 6c) might indicate that OUL35 affected rheobase by Kv1.1 and another unknown mechanism. In addition to the reduced rheobase, there was a tendency of increased AP frequency by DTX-treatment in both control and OUL35 treated group (Fig. 6c).
We confirmed the specific inhibition of ARTD10 by OUL35 in hippocampal neurons with another ARTD10 inhibitor, compound 20 (Fig. 7). This compound, which will be described in a future manuscript, had similar effects on excitability of hippocampal neurons as OUL35: it increased the proportion of the non-inactivating component of K+ currents (Isteady-state/Ipeak, 0.72 ± 0.03 vs. 0.61 ± 0.03, p = 0.007; Fig. 7a), it depolarized the resting membrane potential (− 64.7 ± 3.5 mV vs. − 82.7 ± 4.1 mV, p = 0.005; Fig. 7c), it strongly reduced the rheobase (14.6 ± 6 pA vs. 51.5 ± 13.3 pA, p = 0.04; Fig. 7b), and it increased the frequency of evoked APs (p = 0.007, unpaired Student’s t test; Fig. 7d). These results confirm that inhibiting ARTD10 in hippocampal neurons change the inactivation of K+ channels and increase neuronal excitability.