Metabotropic Acetylcholine and Glutamate Receptors Mediate PI(4,5)P2 Depletion and Oscillations in Hippocampal CA1 Pyramidal Neurons in situ

The sensitivity of many ion channels to phosphatidylinositol-4,5-bisphosphate (PIP2) levels in the cell membrane suggests that PIP2 fluctuations are important and general signals modulating neuronal excitability. Yet the PIP2 dynamics of central neurons in their native environment remained largely unexplored. Here, we examined the behavior of PIP2 concentrations in response to activation of Gq-coupled neurotransmitter receptors in rat CA1 hippocampal neurons in situ in acute brain slices. Confocal microscopy of the PIP2-selective molecular sensors tubbyCT-GFP and PLCδ1-PH-GFP showed that pharmacological activation of muscarinic acetylcholine (mAChR) or group I metabotropic glutamate (mGluRI) receptors induces transient depletion of PIP2 in the soma as well as in the dendritic tree. The observed PIP2 dynamics were receptor-specific, with mAChR activation inducing stronger PIP2 depletion than mGluRI, whereas agonists of other Gαq-coupled receptors expressed in CA1 neurons did not induce measureable PIP2 depletion. Furthermore, the data show for the first time neuronal receptor-induced oscillations of membrane PIP2 concentrations. Oscillatory behavior indicated that neurons can rapidly restore PIP2 levels during persistent activation of Gq and PLC. Electrophysiological responses to receptor activation resembled PIP2 dynamics in terms of time course and receptor specificity. Our findings support a physiological function of PIP2 in regulating electrical activity.


Results
Muscarinic receptors mediate PIP 2 dynamics in CA1 neurons in situ. We began our investigation of PIP 2 dynamics in acute brain slices by examining the response of hippocampal CA1 pyramidal neurons to activation of their muscarinic ACh receptors. In order to measure Gα q induced PIP 2 dynamics in situ, we expressed genetically encoded PIP 2 sensors by stereotaxic injection of lentiviral expression vectors into the hippocampi of juvenile (P21) rats (see Methods). Two different GFP-fused sensor domains were used, tubby CT -GFP 15,40,41 and PLCd1-PH-GFP 31,32 . Both sensors work as 'translocation sensors' , i.e. their degree of membrane association is a direct measure for PIP 2 concentration and its temporal dynamics 3 .
Fluorescence of neurons in acute slices from rats (P26-32) infected with the vector encoding tubby CT -GFP indicated successful expression in CA1 pyramidal neurons. As shown in Fig. 1a GFP fluorescence was primarily localized to the plasma membrane of the soma and the dendritic tree.
Activation of mAChR receptors by application of the specific agonist, oxotremorine-M (Oxo-M), induced massive and reversible translocation of the tubby CT probes from the membrane to the cytoplasm in >95% of CA1 pyramidal cell somata examined, indicating strong PIP 2 depletion (Fig. 1c). To quantify extent and time-course of probe translocation and hence PIP 2 dynamics we measured fluorescence intensity changes in cytosolic ROIs, as cytosolic signals turned out to be less sensitive to tissue movement than when measuring from the small membrane compartment. Consequently an increase of fluorescence signal corresponds to probe dissociation from the membrane, indicating a loss of PIP 2 . Figure 1d shows a representative response to application of Oxo-M for 30 s. The mean tubby CT translocation reached a peak cytoplasmic amplitude (F/F 0 ) of 1.73 ± 0.06 (mean ± SEM; n = 28; 26 slices; 16 rats; Fig. 1e). Mean response latency was 8.8 ± 2.5 s and 90% of the peak response (t 90 ) was reached within 19.9 ± 2.4 s. Upon washout of the agonist PIP 2 levels recovered within about 100 s as indicated by re-association of the probe to the membrane. Cytoplasmic fluorescence returned to 10% of peak amplitude (t 10 , i.e. 90% recovery) in 65.1 ± 7.6 s. To explore the variability of muscarinic PIP 2 dynamics, Oxo-M was applied repetitively with subsequent stimulations separated by a time interval of >10 minutes (Fig. 1f). On average, the degree of PIP 2 depletion exhibited a slight but consistent decline in the course of repetitive stimulation. This is consistent with desensitization of muscarinic signaling previously observed in primary hippocampal neurons 11 .
Previous experiments with cultured neurons have used another sensor domain, PLCδ1-PH, to examine PIP 2 dynamics 9,17,18 . In contrast to the tubby CT sensor, however, PLCδ1-PH has a significant affinity for IP 3 , which is produced whenever PIP 2 is cleaved by PLCβ 33,34,40,[42][43][44] . Therefore the reliability of PLCδ1-PH as an indicator of PIP 2 dynamics during GqPCR/PLC signaling has remained an unresolved issue 33,34 , which provided the rationale for choosing tubby CT in the present study. Indeed, previous studies showed considerable differences in the behavior of tubby CT -GFP and PLCδ1-PH-GFP sensors in terms of translocation following PLC activation 15,34,40 . Thus, we were interested in comparing both sensors in acute brain slices. When PLCδ1-PH-GFP was expressed in CA1 neurons (Fig. 1g), stimulation of mAChRs resulted in robust translocation of fluorescence in all neurons examined (Fig. 1h), similar to the results obtained with tubby CT -GFP (PLCδ1-PH-GFP: latency 4.9 ± 1.2 s; t 90 22.3 ± 1.9 s; F/F 0 = 1.72 ± 0.07; n = 16, 11 slices, 7 rats). However, the recovery was significantly slower compared to tubby CT -GFP (t 10 = 132.7 ± 20.8 s; t-test p = 0.008), suggesting that responses of the PH sensor are co-determined by IP 3 production.
Receptor-specific PIP 2 depletion in CA1 neurons. Activation of PLCβ and subsequent hydrolysis of PIP 2 to IP 3 and DAG is the main signaling pathway of Gα q -coupled receptors. Yet it is not known if PIP 2 depletion is generally associated with the activation of Gα q coupled receptors other than M1/M3 receptors in central neurons. For example, in sympathetic ganglion neurons activity of muscarinic and purinergic receptors results in a depletion of PIP 2 , whereas bradykinin receptors generate IP 3 -dependent Ca 2+ signals without substantial changes of the PIP 2 concentration 17,18,29,45,46 . CA1 pyramidal neurons express various Gα q -coupled receptors which could potentially induce PIP 2 depletion, including group I metabotropic glutamate receptor 47 , α 1A -adrenoreceptor 48  receptors, Gq-coupled dopamine D1-like receptors, H 1 histamine receptor, or P2Y 1 receptor and monitored PIP 2 concentrations with tubby CT -GFP. Of these receptors, only mGluRs induced detectable probe translocation indicative for depletion of PIP 2 (Fig. 2a). mGluRI-induced PIP 2 depletion in neuronal somata was consistent across the population of neurons examined (F/F 0 = 1.37 ± 0.05; n = 15; 14 slices from 11 rats; Fig. 2b).  As summarized in Fig. 2c, PIP 2 levels in CA1 neurons were insensitive to activation of any of the other receptors. Importantly, subsequent control application of Oxo-M triggered robust translocation of tubby CT -GFP in each neuron, indicating proper responsiveness of the cell and appropriate sensitivity of the detection approach (Fig. 2d). Further, the prolonged application of each of the agonists (except muscarinic and glutamatergic) for up to 120 seconds or application of the endogenous ligands serotonin and dopamine did not evoke detectable responses (not shown). Equivalent results were obtained with neurons expressing the alternative PIP 2 sensor domain, PLCδ1-PH-GFP. As shown in Fig. 2f and g, the stimulation of mAChR and mGluRI but none of the other receptors examined induced translocation of PLCδ1-PH-GFP. In summary, results obtained with both sensor domains indicate that α 1A -adrenoreceptor, bradykinin, dopamin D1-like , histamin-H 1 , P2Y 1 , and 5-HT 2A/2C do not induce significant PIP 2 depletion in the soma of CA1 pyramidal neurons. Thus, PIP 2 depletion is specific to mAChR and mGluRI, at least in the context of standard experimental conditions. However, response magnitude of mGluR activation was significantly lower compared to muscarinic PIP 2 depletion in a population of cells challenged by both agonists (n = 12, 11 slices, 8 rats; paired t-test p = 0.032). While Oxo-M and DHPG have similar binding affinities for their cognate receptors 63 , EC50 values for downstream effects such as Ca 2+ responses are often higher for DHPG than for Oxo-M, raising the possibility that 10 µM of DHPG might not be sufficient to evoke a saturating PIP 2 response. However, increasing the concentration to 100 µM or the duration of agonist application of the glutamatergic agonist (DHPG) did not further increase the response mediated by mGluRs (Fig. 2e) and these responses were significantly smaller than muscarinic responses in the same neurons (p < 0.05, one way ANOVA followed by Tukey post-hoc test, n = 6, 6 slices, 3 rats).
In addition to the smaller responses, mGluRI-induced depletion of PIP 2 also differed in its time course compared to muscarinic stimulation ( Fig. 2a,b,f). As measured with tubby CT -GFP, response latencies (6.81 ± 0.89 s) and rise time (t 90 = 12.68 ± 1.15 s) where comparable, but time course of recovery was faster compared to mAChR activation (t 10 = 28.38 ± 3.57 s; t-test p = 0.0001) Remarkably, in 6 out of 15 recordings, PIP 2 levels recovered in the continued presence of the agonist DHPG, as illustrated by individual recordings shown in Fig. 2h. Similarly, PIP 2 dynamics as measured with the PLCd1-PH-GFP probe showed a much faster recovery after activation of mGluRI (t 10 = 30.44 ± 2.05 s, n = 15, 14 slices, 7 rats) when compared to mAChRs (t 10 = 132.74 ± 20.77 s; t-test p = 0.0003; Fig. 2f). As noted with mAChR activation, PIP 2 dynamics induced by mGluRIs showed slight desensitization in response to repeated application of the agonist (Fig. 2i).
Dendritic PIP 2 dynamics. Next, we were interested in the spatial pattern of PIP 2 depletion, in particular with respect to dendritic compartments. Because we probed PIP 2 dynamics with translocation sensors that require microscopic resolution of membrane versus cytoplasm the measurements were confined to dendrites with a diameter of more than 1 µm that were localized close to the slice surface allowing for good optical access. Thus we achieved recordings from the main apical dendrite and its major branches up to 300 µm and basal dendrites to 20 µm distal to the soma. The distance between stratum pyramidale to fissura hippocampi defining the total length of apical dendrites was about 400 µm for our slices.
We found robust but receptor-specific PIP 2 depletion in all dendritic compartments examined. Figure 3a shows a representative example illustrating membrane localization of tubby CT -GFP in a dendrite and its transient redistribution into the dendritic cytosol during pharmacological activation of mAChRs, indicating reversible depletion of PIP 2 . With the exception of one apical dendrite (190 µm distance from soma), all dendrites examined (n = 12) responded with the translocation of tubby CT -GFP. The time course of representative mAChR-induced dendritic PIP 2 dynamics is further shown in Fig. 3d as a kymograph and quantitatively as the change of cytosolic fluorescence intensity. The average peak amplitude derived from dendrites of 12 neurons was F/F 0 = 1.57 ± 0.06 (12 slices, 10 rats). The average latency of the responses was 14.29 ± 5.26 s. We observed a rise time t 90 of 17.52 ± 2.67 s and 90% recovery time (t 10 ) was 41.27 ± 7.25 s after the end of the application. As shown in Fig. 3a,e, activation of mGluRI also resulted in translocation of tubby CT -GFP. However, the degree of PIP 2 depletion was significantly weaker compared to activation of mAChRs (F/F 0 = 1.18 ± 0.02; paired t-test, p = 0.041, n = 3 dendrites in 3 slices from 2 rats. Similar observations were made with PLCδ1-PH-GFP as the PIP 2 probe (Fig. 3c,h). Thus, activation of muscarinic receptors by Oxo-M induced strong translocation (F/F 0 = 1.86 ± 0.07; t 90 = 21.86 ± 1.90 s; t 10 = 142 ± 30.53 s; n = 7 dendrites, 7 slices, 3 rats), whereas activation of mGluRI receptors by DHPG induced small yet reproducible translocation of PLCdδ1-PH-GFP (F/F 0 = 1.19 ± 0.02; n = 7). Stimulation of various other Gq-coupled receptors did not induce detectable translocation of PLCδ1-PH-GFP (Fig. 3c).
Prolonged receptor activation revealed complexity of PIP 2 dynamics. The occasionally observed early recovery of PIP 2 level during activation of mGluRI prompted us to examine the time course of PIP 2 dynamics during sustained stimulation. Figure 4 shows the resulting PIP 2 dynamics measured with the tubby CT sensor in neuronal somata during continuous application of the receptor agonists for 5 min. Notably, PIP 2 depletion generally showed a phasic-tonic time course with an initial strong PIP 2 depletion followed by partial recovery in the sustained presence of the agonist as apparent from the average from a larger number of neurons (n = 17 and 16 for mAChR and mGluRI activation, respectively; Fig. 4a). The initial rate of PIP 2 decrease was similar for both receptors (t 90 = 39.3 ± 5.3 s and 43.3 ± 8.8 s, respectively), but as previously seen with brief receptor stimulation (Fig. 2), the muscarinic responses had a higher average amplitude compared to glutamatergic responses (F/F 0 = 1.84 ± 0.08 and 1.55 ± 0.07, respectively; paired t-test p = 0.0002, n = 16, 14 slices; 5 rats). PIP 2 recovery after muscarinic receptor activation was more pronounced than for mGluR stimulation such that PIP 2 levels tended towards similar values at the end of receptor stimulation period. For both receptors, recovery of PIP 2 levels after removal of the receptor agonist was slower than observed with brief receptor stimulation (mAChR: t 10 243.2 ± 43.9 s; mGluRI: t 10 167.8 ± 41.5 s; cf. Fig. 2b) and thus depended on the duration of receptor activation.
While demonstrating partial desensitization of PIP 2 responses as a common pattern, the prolonged stimulation also revealed considerable variability and complexity in the time course of PIP 2 dynamics (Fig. 4b). Most strikingly, PIP 2 depletion was often multiphasic or oscillatory. Examples for such complex PIP 2 dynamics are shown in Fig. 4c-e. In these cells, PIP 2 levels returned to baseline despite sustained presence of the agonist, and moreover, multiple depletion events occurred in rapid succession (Fig. 4d,e). Altogether, 8 out of 17 neurons showed oscillatory PIP 2 concentration dynamics with Oxo-M and two out of 16 during application of DHPG. These observations suggest that PIP 2 dynamics may be subject to complex temporal regulation and indicate potent PIP 2 resynthesis capability of CA1 neurons during receptor activation.
Modulation of electrical behavior by receptor stimulation. The well-described effects of muscarinic activity on the electrical properties of CA1 neurons -including inhibition of M-currents -may (at least partially) be mediated by PIP 2 concentration dynamics. We thus were interested in differential effects on excitability of the various Gq/PLC-coupled receptors examined for their coupling to PIP 2 dynamics.
To this end, we performed patch clamp experiments in current clamp mode in acute brain slices prepared from rats at P14 to P21. Current step protocols were used to assess membrane potential, input resistance, spiking behavior, and afterpolarisation (Fig. 5a,b). Experiments were performed in the presence of inhibitors of GABA A/B and ionotropic glutamate receptors (see Methods) to exclude effects resulting from network activity.
Overall, we found pronounced changes in electrical behavior following the activation of muscarinic mAChR receptors and mGluRI but little effects of other Gq-coupled receptors. Consistent with previous findings [64][65][66][67] agonists of both mAChR (n = 11, 11 slices, 10 rats) and mGluRI (DHPG 10 µM, n = 9, 8 slices, 5 rats) induced depolarization of the resting membrane potential and an increase in firing frequency during depolarization (number of action potentials, NAP; Fig. 5a-d,f). In CA1 cells, a train of action potentials is usually followed by an afterhyperpolarization (AHP) 37,68,69 . Application of either Oxo-M or DHPG resulted in the disappearance of the AHP and the appearance of an afterdepolarisation (ADP; Fig. 5a,b,e). All of these receptor-induced changes are consistent with the deactivation of potassium conductances such as M currents 37,68,70,71 . In some neurons, activation of mAChR receptors induced sustained depolarization (plateau potentials) subsequent to the 600 ms current step, as also described previously 72 .
As shown in Fig. 5c-f, stimulation of other Gq-coupled receptors, including 5-HT 2A/C receptors, α 1 adrenergic receptors, bradykinin receptors, D 1 -like dopamine receptors, H 1 histamine receptors, and purinergic P2Y 1 receptors had no significant effects on either of these electrophysiological characteristics. Noteworthy, the mAChR and mGluRI induced depolarization did not always persist during the agonist application, as occasionally initial transient hyperpolarisation and oscillations of the membrane potential was observed during mAChR activation. During muscarinic stimulation, 9 of 11 neurons showed substantial recovery from depolarization in the presence of Oxo-M; of those, 6 neurons showed complete repolarization or even hyperpolarisation before washout. In the presence of DHPG, 3 neurons showed a substantial recovery from initial depolarization. Oscillatory behavior was observed in neurons from younger animals (P14-21), but also in slices age-matched to the PIP 2 imaging experiments, as shown in an exemplary recording (Fig. 5g) obtained from a neuron in a P27 slice.
In summary, we find that pronounced changes in membrane potential and firing rates paralleled neuronal PIP 2 depletion in terms of effect size, time course and receptor specificity. Thus, depolarization, increased spike rates and PIP 2 dynamics were largely restricted to the activation of mAChR and mGluRI receptors. Neuron type-specific PIP 2 dynamics: dentate gyrus granule cells. To extend our observations on PIP 2 dynamics to additional neuronal cell types, we measured PIP 2 dynamics following activation of the same receptors (mAChR and mGluRI) in dentate gyrus granule neurons (Fig. 6a). Granule neurons express both mAChR and mGluRI receptors at the soma 47,73,74 . However, detectable depletion of PIP 2 upon stimulation with Oxo-M was observed in only three independent experiments (n = 24; Fig. 6b,c). None of the six neurons stimulated with DHPG (10 µM, n = 2) or Glutamate (100 µM, n = 4) showed any detectable sensor translocation (Fig. 6d). Thus, the induction of PIP 2 dynamics may be highly specific between different types of neurons and this specificity seems to be dependent on mechanisms other than the expression of Gq/PLC-coupled transmitter receptors.

Direct observation of PIP 2 dynamics in central neurons in situ.
While there is good evidence for PIP 2 depletion in response to activation of Gq-coupled receptors for some types of neurons studied in the cell culture dish, surprisingly little is known about the prevalence and spatiotemporal properties of PIP 2 dynamics in central neurons under physiological conditions. PIP 2 levels and their dynamic regulation may be largely different in vivo, as embedding in the native environment and full differentiation of neurons may impact on relevant factors such as expression and spatial subcellular organization of receptors and downstream components of the signaling cascade and the enzymes that resynthesize PIP 2 . Therefore, information on PIP 2 concentration behavior in intact tissue preparations such as brain slices is required. Previous studies with organotypic slices were consistent with PIP 2 depletion in-situ triggered by synaptic release of glutamate onto cerebellar Purkinje neurons 75 or by muscarinic agonist application in cortical pyramidal cells 76 . However, both studies were not fully conclusive with  respect to PIP 2 signaling because PLCδ1-PH was used as a sensor domain, which has a similar affinity for PIP 2 and IP 3 and may report the production of IP 3 rather than depletion of PIP 2 44,77 . In fact, Okubo et al. 75 interpreted probe translocation in terms of IP 3 production rather than depletion of PIP 2 .
Here, by using tubby CT , as an alternative PIP 2 sensor insensitive to IP 3 33,34,40 , our present data now show unequivocally that muscarinic and metabotropic glutamate receptors indeed trigger PIP 2 dynamics in a prototypic central neuron in situ. Of note, in cultured cell lines tubby CT previously failed to respond or only weakly translocated upon PLC-mediated PIP 2 depletion 34,40 , which initially was attributed to a higher affinity for PIP 2 compared to PLCδ1-PH. However, quantitative titration of PIP 2 in living cells showed that its affinity to PIP 2 is actually lower, which should make it a useful PIP 2 sensor 41 . In fact, our current results demonstrate that tubby CT readily translocates in a neuronal cellular environment. The difference in behavior between experimental conditions is not yet understood, but may suggest cell type-specific segregation of PIP 2 into distinct pools selectively accessible to the different PIP 2 -binding domains 78 . In any case, our findings show that in the native neuronal system tubby CT is a much better reporter of PIP 2 dynamics than might have been anticipated 20 . Thus, using tubby CT -GFP allowed us to systematically assay PIP 2 dynamics without confounding effects of IP 3 dynamics.
Spatiotemporal properties of PIP 2 dynamics. We found that in the larger compartments accessible to measuring sensor translocation, receptor-induced PIP 2 depletion appeared largely homogenous without evidence for substantial subcellular differences. This suggested that induced PLC activity is similar in somatic and dendritic compartments. This finding was not entirely expected, because while M1 receptors show high densities throughout soma and dendrites 73 , the most prominent mGluRI receptor of CA1 neurons, mGluR 5 , has a relatively low density in soma compared to dendrites 47 . Since glutamatergic PIP 2 depletion in dendrites was not stronger than in the somatic compartment, the PIP 2 depletion pattern does not appear to correlate closely with receptor distribution. It is worth noting that the moderate degree of PIP 2 sensor translocation in dendrites did not result from the distinct dendritic geometry, because the smaller volume-to-membrane area ratio in the dendrites should rather result in a stronger relative increase of sensor fluorescence when sensors dissociate from the membrane. Also, muscarinic stimulation elicited larger dendritic responses than glutamatergic stimulation, showing that PIP 2 sensor response was not saturated by mGluR stimulation. Of note, a similar observation was made by Nakamura et al. 79 for Ca 2+ dynamics in CA1 pyramidal cells: activation of mGluRI, mAChR and 5-HT 2 R elicited comparable Ca 2+ waves despite different receptor distribution. Thus, these neurons may possess mechanisms to globalize Gq signaling including PIP 2 depletion.
To our knowledge our results for the first time demonstrate oscillations of the PIP 2 concentration in a neuron. While IP 3 , DAG and Ca 2+ are known to undergo oscillatory concentration dynamics in neurons 80 , previous observations on PIP 2 dynamics in primary dissociated neurons 16,18,33 seemed to indicate that PIP 2 concentrations essentially remained depleted during prolonged receptor activity. Oscillatory translocation of the PLCδ1-PH sensor domain observed occasionally has been understood as dynamics of the IP 3 signal picked up by the PH  23,44,81 . More recently, careful observations also including the specific tubby CT sensor showed bona-fide oscillations of PIP 2 in mast cells 82,83 . Our observations suggest that such dynamics may be a more general phenomenon with implications for neuronal biology.
The mechanisms underlying PIP 2 oscillations may include both positive and negative feedback regulation of PIP 2 cleavage by PLC. Such mechanisms have previously been shown to be involved in Ca 2+ and IP 3 oscillations and include Ca 2+ -dependent activation of PLC 80,81 as a positive feedback. In mast cells, PIP 2 oscillations are probably driven by Ca 2+ oscillations 82 . Inhibition of Gq signaling by, e.g., PKC, receptor kinases, or RGS molecules 81,84 may contribute to a negative feedback loop controlling PIP 2 degradation. Moreover, our observations reveal an impressive capability of PIP 2 replenishment, as indicated by the rapid and complete recovery of PIP 2 levels in presence of agonists. PIP 2 resynthesis may be increased during GqPCR activation 85 , providing negative feedback to PIP 2 depletion and possibly contributing to observed oscillations. Specifically, PIP 2 replenishment may involve Ca 2+ and phosphatidic acid-dependent phospholipid exchange at plasma membrane-endoplasmic reticulum (PM-ER) junctions 86 .
Whatever the mechanism underlying the oscillations is, our findings indicate that PIP 2 dynamics may provide neurons with another dimension of effector modulation beyond a simple on/off switch for downstream effectors. Although the consequences of PIP 2 oscillations for electrical neuronal activity remain to be explored, we note that indeed, neurons showed fluctuations of membrane potential and firing frequency during agonist application. It is worth mentioning that mAChR and mGluRI agonists can induce and shift gamma and theta oscillations 87 . In the light of the present data, it is tempting to speculate that PIP 2 oscillations might participate in such frequency modulation.
Neuronal ion channels as effectors of PIP 2 dynamics. Given the known high sensitivity of some ion channels to even a moderate drop in the PIP 2 concentration 5,78 , a main potential target of PIP 2 depletion are ion channels and thus electric excitability. Based on studies on isolated neurons, inhibition of Kv7 channels in sympathetic neurons as the direct consequence of PIP 2 depletion is well established 13,17,29 . Our data permit the correlation of PIP 2 dynamics and electrophysiology in situ. Activation of mAChR and mGluRI, but not other PLC-coupled receptors known to be present and functional in CA1 neurons induced robust PIP 2 depletion. The same pattern of receptor specificity was observed for modulation membrane potential, firing frequency and afterhyperpolarization, providing at least circumstantial evidence for the causation of channel regulation by PIP 2 . Simultaneous recordings of electrical activity and PIP 2 dynamics from the same neuron should be performed in the future to provide more direct evidence.
In conclusion, our data support and generalize the as yet largely hypothetical mechanism of PIP 2 dynamics as a major cellular signal in the control of neuronal activity through regulation of PIP 2 -sensitive ion channels such as Kv7. Future studies need to address this issue rigorously by manipulating PIP 2 levels in-situ 20 . Along those lines a recent study aimed at PIP 2 depletion in hippocampal slice cultures by chemically induced recruitment of a PIP 2 phosphatase 88 . While this approach did not reveal any effects on electrical properties of the neurons, the results appear inconclusive since changes in PIP 2 concentration were not verified.
One of the most intriguing unknowns are the spatiotemporal properties of PIP 2 dynamics during entirely physiological neuronal activity, i.e, during synaptic activity of the modulatory (e.g. cholinergic) and principal (i.e. glutamatergic) inputs into the hippocampal neurons and of the PIP 2 dynamics associated with intrinsic neural (network) activity. Another question is the PIP 2 signaling in the distal smaller dendritic compartments not amenable to analysis by the translocation probes used in this study. In particular, in the immediate postsynaptic compartment, i.e. spines, PIP 2 may have a role in controlling synaptic plasticity [89][90][91] .

Materials and Methods
Virus production and constructs. Lentiviral plasmids pCMVΔR8.9, pVSVG and FUGW were kindly provided by Pavel Osten (MPI for medical research Heidelberg, Germany). The PLCδ1-PH and tubby CT constructs were provided by Tamás Balla (NIH, Bethesda, USA) and Lawrence Shapiro (Columbia University, USA), respectively. Lentiviral particles were derived by triple transfection of HEK293FT cells with Lipofectamin 2000 (Invitrogen, Darmstadt, Germany). Virus purification from supernatant was achieved by 15 minute centrifugation at 3000 rpm, filtration through a Millex ® HV 0.45 µm filter (Millipore, Darmstadt, Germany) and two successive ultracentrifugation steps (25000 rpm, 1 h 30 min, 4 °C). Pellets were resuspended in TBS-5 buffer (50 mM Tris-HCl, 130 mM NaCl, 10 mM KCl, 5 mM KCl 2 ) and subjected to a final 30 s centrifugation at 5000 rpm. Aliquots were stored at −80 °C and thawed up to two times. Animals, stereotactic injection and slice preparation. Wistar rats were obtained from the animal facility of the Philipps University of Marburg (Marburg, Germany) or Charles River (Cologne, Germany) and kept and handled according to German law and institutional guidelines at the Philipps University. All procedures were approved by the Regierungspräsidium Giessen, Germany. Animals were housed with access to ad libitum water and food on a 12-h light/dark cycle. At weaning (postnatal day 21) male and female rats were anesthetized by intraperitoneal injection of a mixture of ketamine (Bela-Pharm, Vechta, Germany) and xylazine (Rompun ® , Bayer AG, Leverkusen, Germany) at a dose of 100 and 10 mg per kilogram body weight. Additionally, the mixture included 0.05 mg/kg Atropine (B. Braun, Melsungen, Germany) and 0.1 ml/10 g body weight of a 0.9% NaCl solution for injection (Diaco, Triest, Italy). Under stereotactic control, lentivirus was injected bilaterally using Paxinos and Watson 92 as a reference. Coordinates were optimized for targeting in juvenile rats by setting the adult references to x = +/−6.125, y = −6.15 and z = −6.2 mm and multiplying by the ratio of the juvenile to atlas (8.7 mm) distance of bregma to lambda. Up to 2.5 μl virus per hemisphere were injected in 500 nl portions going from ventral to dorsal in 0.3-0. 35  Wiesbaden, Germany) and sacrificed by decapitation at the ages indicated in results. The head was placed in ice cold sucrose-ACSF (sucrose-artificial cerebrospinal fluid, in mM: 87 NaCl, 25 NaHCO 3 , 25 D-glucose, 75 sucrose, 2.5 KCl, 0.5 CaCl 2 and 7 MgCl 2 , oxygenated with 95% O 2 /5% CO 2 ) and the hippocampi rapidly removed. 300 μm transversal slices were cut with a vibratome (VT1200, Leica Biosystems, Wetzlar, Germany) and placed into a chamber with 4 °C sucrose-ACSF. After a 35 min recovery period at 35 °C slices were kept at room temperature. For recordings slices were transferred to a submerged chamber and perfused with ACSF (in mM: 125 NaCl, 25 NaHCO 3 , 25 D-glucose, 2.5 KCl, 2 CaCl 2 and 1 MgCl 2 , oxygenated with 95% O 2 /5% CO 2 ) for at least 20 minutes.
Imaging, electrophysiological recording and data analysis. Confocal imaging was performed with a Zeiss LSM710 (Zeiss, Oberkochen, Germany). The sampling rate for time series experiments was 1.75 s with a pixel size of 0.13 µm. In some cases (especially dendrite measurements) the sampling rate was increased to 1 s. In all cases where the sampling rate slightly deviated the data were resampled to allow averaging across experiments. Overlay with the original was performed to ensure preservation of time scale. Average cytoplasmic fluorescence intensities were determined from regions of interest (ROI) excluding both the plasma membrane (defined as the local intensity max at the cell's border in the resting cell) and the nucleus. Distance of ROIs to the plasma membrane was >0.5 µm even when slight shifts of the cell's position occurred during the experiment. ROIs were defined post-hoc using the microscope software ZEN (Versions 2008 and 2009) and obtained average intensities were exported to Igor Pro (Version 6.03 A, Wave Metrics, Portland, OR USA). Traces were background subtracted and normalized to the last time point before beginning of a response (F/F 0 normalized to t 0 ). Measurements without evident response were corrected for photobleaching according to a biexponential fit to the decaying fluorescence signal and normalized to signal at the onset of agonist application. We found that probe translocation generally prevented the reliable determination of the time course of photobleaching. Therefore most data were not corrected for bleaching which results in apparently lower signals following transient depletion of PIP 2 , with bleaching generally being more pronounced for tubby CT -GFP than for PLCδ1-PH-GFP. Confocal images were further analyzed with ImageJ (National Institutes of Health, USA) to isolate individual images of a time series, create kymographs and set scale bars. Electrophysiological data were recorded with a HEKA EPC10USB amplifier and Patch Master software (Version 2.43 HEKA, Lambrecht, Germany) in current clamp mode. Data were low pass filtered with a 2.9 kHz Bessel filter and digitized at 20 kHz. Borosilicate recording pipettes had a resistance of 3-4 MΩ and were filled with intracellular solution containing (in mM): K-gluconate 135, KCl 20, MgCl 2 2, Na 2 -ATP 2, Na 2 -GTP 0.3, HEPES 10 and EGTA 0.1 (adjusted to pH 7.2 with KOH). Series resistance was monitored in voltage clamp mode before and after each current clamp recording, but not corrected for. Measurements with a change in series resistance >40% during the course of the experiment were discarded. Input resistance was assessed by injection of small positive and negative currents steps, followed by a depolarizing current step above action potential threshold to quantify spiking behavior and afterpolarisation (see Fig. 5a,b). Sweep length was 7 seconds. In applications of the P2Y 1 agonist ADPβS a shorter protocol without the positive 20 pA step was used. Medium afterhyperpolarisation (AHP m ) was obtained as the difference of resting V m and mean V m at 70 to 120 ms after the depolarizing current step. Changes in AHP value resulting from application of receptor agonists are given as Δafterpolarization such that positive values indicate reduction of AHP or eventually the emergence of an afterdepolarisation. Amplitudes were calculated from averaging at least 10 baseline data points and a minimum of 3 peak points, with avoidance of plateau potentials. Statistical analysis. Statistical significance was tested in Igor Pro. Randomness, equal variances and normal distribution of the data was tested with Igor's Runs, Kolmogorow-Smirnow and Jarque-Bera test. In cases where validity of a parametric test was compromised, a Wilcoxon-Mann-Whitney test was performed. Where applicable, groups of two were compared with paired and unpaired Student's t. Two tailed one-way ANOVA was followed by a Dunett test for comparing multiple groups to a single control or a Tukey test to compare all groups to each other. Unless noted otherwise all values are given ± standard error of the mean. Chemicals and perfusion system. Oxotremorine-M, DHPG, Bradykinin, SKF83959, DOI, Serotonin and Dopamine were purchased from Tocris and Methoxamine and 2-Pyridylehylamin from Sigma. All other chemicals were from Sigma/Fluka or Merck (Germany). For application of test substances a capillary of 200 to 250 µM inner diameter (TSP200350, BGB Analytik AG, Boeckten, Germany or MicroFil MF28G-5, World Precision Instruments, Berlin, Germany) was placed directly next to the hippocampal recording region. Solution exchange at the tip occurred within 1-2 s. Unless noted otherwise recordings represent first applications of each test substance. To block fast glutamatergic and GABAA/B signaling in electrophysiological recordings, receptor antagonists (4 μM NBQX, 50 μM D-AP5, 50 μM Picrotoxin and 1 μM CGP 55845, all from Tocris) were added both to the bath and local perfusion. Data availability. Most data generated or analysed during this study are included in this published article.
Additional datasets generated and analysed during the current study are available from the corresponding authors on reasonable request.