The subthreshold-active KV7 current regulates neurotransmission by limiting spike-induced Ca2+ influx in hippocampal mossy fiber synaptic terminals

ARTICLE https://doi.org/10.1038/s42003-019-0408-4 OPEN The subthreshold-active K 7 current regulates neurotransmission by limiting spike-induced 2+ Ca influx in hippocampal mossy fiber synaptic terminals 1 2 2 3 1 Katiuscia Martinello , Elisabetta Giacalone , Michele Migliore , David A. Brown & Mala M. Shah Little is known about the properties and function of ion channels that affect synaptic terminal- resting properties. One particular subthreshold-active ion channel, the Kv7 potassium channel, is highly localized to axons, but its role in regulating synaptic terminal intrinsic excitability and release is largely unexplored. Using electrophysiological recordings together with computational modeling, we found that the K 7 current was active at rest in adult hippocampal mossy fiber synaptic terminals and enhanced their membrane conductance. 2+ 2+ The current also restrained action potential-induced Ca influx via N- and P/Q-type Ca channels in boutons. This was associated with a substantial reduction in the spike half-width and afterdepolarization following presynaptic spikes. Further, by constraining spike-induced 2+ Ca influx, the presynaptic K 7 current decreased neurotransmission onto CA3 pyramidal neurons and short-term synaptic plasticity at the mossy fiber–CA3 synapse. This is a dis- tinctive mechanism by which K 7 channels influence hippocampal neuronal excitability and synaptic plasticity. 1 2 UCL School of Pharmacy University College London, London WC1N 1AX, UK. Institute of Biophysics, National Research Council, 90146 Palermo, Italy. Department of Neuroscience, Physiology and Pharmacology, University College London, London WC1E 6BT, UK. Correspondence and requests for materials should be addressed to M.M.S. (email: [email protected]) COMMUNICATIONS BIOLOGY | (2019) 2:145 | https://doi.org/10.1038/s42003-019-0408-4 | www.nature.com/commsbio 1 1234567890():,; ARTICLE COMMUNICATIONS BIOLOGY | https://doi.org/10.1038/s42003-019-0408-4 eurotransmitter release from synaptic terminals is the (Fig. 1b), suggesting that the K 7 current was present in mossy predominant mechanism for information transfer fiber boutons. Nbetween neurons and has a fundamental role in processes To determine whether the K 7 currents recorded under whole- 1–3 such as synaptic plasticity . Diverse synaptic terminals express cell voltage-clamp conditions were generated, at least partly, in the high voltage-activated K 1 and K 3 channels that are acti- the bouton per se, we made outside-out patch recordings from V V vated during the repolarization phase of the presynaptic action the boutons (Fig. 1c). Slow de-activating currents with compar- potential. These K channels regulate the action potential width able properties to those obtained under whole-cell voltage-clamp 2+ and, consequently, presynaptic Ca influx and neurotransmis- experiments were present in outside-out patches too. These 4–16 + sion . There are, though, K channels that activate at rest (i.e., were fully inhibited by XE991 (Fig. 1c). The half voltage- at subthreshold potentials) . Much less is known about the activation (V ) and slopes of the activation curves from 1/2 properties and function of these channels in synaptic terminals. whole-cell and outside-out patch configurations were similar The K 7 channels activate at voltages positive to − 70 mV in (whole-cell and outside-out patch V = −69.6 ± 1.6 mV (n = 6) V 1/2 17–19 many neurons and form an inhibitory current at rest . and −73.9 ± 1.8 mV (n = 5), respectively; whole-cell and outside- There are 5 K 7 subunits, of which K 7.2–K 7.5 subunits are out patch activation curve slopes = 5.5 ± 1.5 (n = 6) and 7.6 ± 1.7 V V V neuronal. The K 7.2 and K 7.3 subunits contain an ankyrin (n = 5), respectively; Fig. 1d; see Methods). The current density V V G-binding motif, so they are expressed at the axon initial segment and de-activation time constants measured at − 50 mV were and nodes of Ranvier where they regulate the action potential comparable in both configurations (Fig. 1e, f). Given the 17,18,20–23 threshold and propagation . Interestingly, K 7.5 chan- remarkable similarities in K 7 current densities and biophysical V V nels are present in the giant synaptic terminals of Calyx of Held, properties between outside-out patch and whole-cell voltage- where they influence the resting membrane potential (RMP) . clamp configurations, our results strongly suggest that K 7 Immunohistochemical evidence also suggests that K 7.5 subunits channels are located on mature mossy fiber boutons synapsing are situated in GABAergic, but not glutamatergic, synaptic onto CA3 pyramidal neurons. terminals in the hippocampus . In contrast, immunohis- tochemistry suggests that K 7.2 subunits are expressed 26–28 throughout hippocampal mossy fibers . However, it is not K 7 currents limit the membrane resistance and excitability. known if hippocampal mossy fiber boutons have a K 7 current or To ascertain whether K 7 channels affect mossy fiber bouton V V whether this affects their local intrinsic excitability and neuro- intrinsic excitability, we made whole-cell current-clamp record- transmitter release. It is important to investigate this as the mossy ings from boutons in the presence of glutamate and GABA fiber giant bouton-CA3 pyramid synapse has a critical role in receptor blockers (see Methods) in the absence and presence processes such as learning and memory and in pathophysiological of XE991 (3 μM). The boutons had an average RMP of − 81.9 ± 29–33 disorders such as epilepsy . 1.2 mV (n = 8), which was unaffected by 20 min bath application Here, we show that adult mossy fiber boutons possess a K 7 of XE991 (average RMP with XE991 = −81.7 ± 1.3 mV (n = 8, current that is active at rest. This had little effect on the bouton p = 0.83, two-tailed paired t test); Fig. 2a, b). Similar findings RMP. The current, though, altered the intrinsic excitability of were also obtained with 20 min application of a second K 7 boutons by decreasing the membrane resistance at voltages channel inhibitor, linopirdine (10 μM) (Fig. 2b). Given that our positive to rest. The current also limited action potential-induced data (Fig. 1d) suggests that ~ 20% of the K 7 current is active at 2+ 2+ Ca influx via voltage-gated N- and P/Q-type Ca channels. −80 mV, this implies that other ion channels, such as the inward 42,43 This was coupled with a reduction in the presynaptic spike half- rectifier potassium channels and twin-pore potassium chan- width and an afterdepolarization that follows presynaptic spikes nels, have a larger influence on the RMP in mossy fiber boutons. in these boutons. Further, by limiting action potential-induced Depolarizing current pulses of varying magnitudes resulted in 2+ Ca influx, K 7 channels restricted neurotransmitter release only single action potentials in boutons (Fig. 2a). The inability and short-term synaptic plasticity onto CA3 pyramidal neurons. of more than one spike to be initiated with different magnitudes This is a unique mechanism by which presynaptic K 7 channels and sustained depolarization is probably owing to the specialized + + affect local excitability within adult hippocampal synaptic term- properties of the mossy fiber bouton Na and other K 14,44 inals and regulate neurotransmission. This might be an important (e.g., K 1) channels . In the presence of either XE991 or means by which K 7 channels contribute towards influencing linopirdine, though, significantly less depolarizing current was neural network rhythms and maintaining network excitability required for action potential initiation (Fig. 2a–c). Thus, K 7 34–36 in the hippocampus . channels enhance the rheobase for spike generation in mossy fiber boutons. If K 7 channels do not affect the RMP in mossy fiber boutons, Results how might they affect the ability for depolarizing pulses to result K 7 currents in mature mossy fiber boutons. As mossy fiber in spike generation here? In many other neurons, the inhibitory boutons that synapse onto CA3 proximal apical dendrites have current formed by K 7 channels at subthreshold potentials 10,31,32,37 large diameters (2–5 μm, Fig. 1a) , we made electro- limits the amount of depolarization with given current injections physiological recordings from these present in hippocampal slices (i.e., decreases the input resistance (R )), resulting in a larger obtained from mature rats. The tracer, neurobiotin, was included rheobase required to elicit action potentials. Thus, we examined and post-hoc morphological analysis was performed to positively whether these channels modified R in mossy fiber boutons too. identify boutons (Fig. 1a). To record the K 7 current, the classical To test this, we injected small hyperpolarizing and depolarizing 38–40 de-activation protocol was applied under whole-cell voltage- subthreshold current pulses at a fixed potential of − 80 mV (i.e., clamp condition in the absence and presence of the specific, near the bouton RMP) in the absence and presence of XE991 or irreversible, pharmacological K 7 channel inhibitor, XE991 (3 μM, linopirdine (Fig. 2d). R measured at potentials above −80 mV, V N a concentration that inhibits > 95% of the current ) (Fig. 1b). but not at hyperpolarizing potentials, was significantly greater This revealed slow de-activating currents that reversed at ~ − 90 in the presence of XE991 or linopirdine (Fig. 2d, e). This effect mV (i.e., near the K reversal potential, Fig. 1b). The currents was time-dependent, peaking ~20 min after application of XE991 14,45 were stable for at least 20 min, with minimal rundown and were (Fig. 2f). As mossy fiber boutons are electronically compact , maximally inhibited by 20 min bath application of XE991 the effect on R is likely to be owing to local K 7 channels in the N V 2 COMMUNICATIONS BIOLOGY | (2019) 2:145 | https://doi.org/10.1038/s42003-019-0408-4 | www.nature.com/commsbio COMMUNICATIONS BIOLOGY | https://doi.org/10.1038/s42003-019-0408-4 ARTICLE a CA3 Pyramidal Neuron Proximal Patch apical pipette dendrite Mossy fiber bouton b Whole-cell voltage-clamp recordings K 7(M-) current Control XE991 (XE991 - Control) 200 pA 50 pA 0 pA 1 s –20 mV 100 pA –110 mV 100 ms c Outside-out patch recordings K 7(M-) current Control XE991 V (XE991 - Control) 75 pA 50 pA 1 s 0 pA 100 pA 100 ms Whole-cell (n = 6) d ef Outside-out (n = 5) 1.0 0.8 100 0.6 0.4 0.2 –120 –80 –40 Voltage (mV) Fig. 1 The K 7 current is localized in mossy fiber boutons. a Schematic showing that mossy fiber boutons from which electrophysiological recordings were obtained were situated near or on the proximal dendrites of CA3 pyramidal neurons. The insert shows a confocal image of a mossy fiber bouton that had been recorded from, filled with neurobiotin and stained with streptavidin Alexa Fluor 488 conjugated antibodies. The scale bar corresponds to 2 μm. b, c Example whole-cell and outside-out voltage-clamp recordings of the K 7 current, respectively. The bouton or patch was held at −20 mV and a series of hyperpolarizing, 2 s long steps were applied as per schematic in the absence (control) and presence of 3 μM XE991. The current in the presence of XE991 was subtracted from that recorded in the absence to obtain the K 7 current. The de-activation rates of this current are shown on an expanded time scale in the inset. The scale associated with the first trace applies to all traces within the panel. The outward holding current at −20 mV under control conditions is also shown. d The activation curves of the K 7 current measured under either whole-cell or outside-out patch voltage-clamp conditions. e, f The average (bars) and individual (open squares) current (I) density and de-activation time constants (τ)in five outside-out patches and six whole-cell recordings from mossy fiber boutons, respectively. The capacitance for calculating the current density was obtained using our multiclamp amplifier. The τ values were measured for currents elicited by a hyperpolarizing step to −50 mV from a holding potential of −20 mV COMMUNICATIONS BIOLOGY | (2019) 2:145 | https://doi.org/10.1038/s42003-019-0408-4 | www.nature.com/commsbio 3 Outside-out patch recordings Whole-cell recordings Outside-out patch recordings Whole-cell recordings I/I MAX Current desnsity (pA/pF) De-activation time constant (ms) ARTICLE COMMUNICATIONS BIOLOGY | https://doi.org/10.1038/s42003-019-0408-4 Fig. 2 K 7 currents regulate the intrinsic excitability of mossy fiber boutons. Control XE991 (3 μM) a Representative whole-cell current-clamp recordings generated when a series of hyperpolarizing and depolarizing current pulses were applied to a mossy fiber bouton before and after application of 3 μM XE991 at the –82 –84 normal resting membrane potential (RMP, indicated adjacent to the traces). mV mV 20 mV The scale shown applies to both traces. b, c Graphs depicting the individual 160 pA 500 ms (open square) and mean (filled square) RMP as well as the average numbers of action potentials recorded from boutons in response to current –20 pA (I) pulses in the absence and presence of XE991 or linopirdine (10 μM) respectively. The numbers of observations are indicated in parenthesis. b (8) c (8) (4) d Example recordings obtained when 20 pA, 1 s hyperpolarizing and (4) 1.0 depolarizing square current pulses were applied at a fixed potential of − –82 0.8 * 80 mV in the absence and presence of XE991. e The individual (open – XE (n =8) 0.6 square) and mean (filled square) input resistance (R ) measured using 20 –78 + XE (n =8) pA hyperpolarizing and depolarizing current pulses at − 80 mV with and 0.4 – Lino (n =4) –74 without XE991 or linopirdine. The numbers of observations are indicated in + Lino (n =4) 0.2 * parenthesis. f The time course of the average effects of XE991 on the input –70 resistance measured using a 20 mV depolarizing step in five independent 0 50 100 150 200 mossy fiber terminals. In all panels, asterisks signify p < 0.05 when I (pA) compared with appropriate controls d e –20 mV +20 mV Control (8) +XE991 2.5 (4) by XE991 or linopirdine enhanced the ADP amplitude and decay (4) (8) (n = 4) (n = 8) time constant following a single action potential to a similar +20 pA 2.0 extent irrespective of the frequency of the spike train or the –80 1.5 position of the action potential within the train (Fig. 3b; Sup- mV 1.0 20 mV plementary Table 1). The spike width in the presence of either –20 pA XE991 or linopirdine was also broader than under control con- 0.5 500 ms ditions (Fig. 3c). Further, the spike amplitudes were smaller in the presence of K 7 channel inhibitors compared with controls (Fig. 3c; Supplementary Table 2). At frequencies of ≤ 20 Hz, under control conditions, each subsequent spike in a train occurred during the ADP generated +20 mV by the preceding action potential, resulting in a summating ADP (Fig. 3d, e). The area under the ADP was much greater after 1.5 application of XE991 or linopirdine than under control condi- tions (Fig. 3d, e). As K 7 channel inhibition at the granule cell XE991 (n = 5) 1.3 axon initial segment did not result in enhancement of an 40,46 ADP , it suggests that these channels differentially regulate 1.1 intrinsic excitability in granule cell axon subcompartments. 2+ 0.9 The K 7 conductance limits spike-induced Ca concentra- tion. To further understand the cellular mechanisms by which 0.7 K 7 channels might restrict the spike width and the ADP –5 0 5 10 15 20 25 amplitude following spikes, we generated a single compartment Time (min) model consisting predominantly of the K 7 conductance with our biophysical characteristics (Fig. 1), an inward rectifier type K conductance, the ‘A’-type (K 1) conductance, delayed rectified mossy fiber bouton. Indeed, in the granule cell somata where K 7 + + 2+ type K conductance, Na conductance and a Ca conductance channels are not present, K 7 channel inhibitors have little effect (see Supp. Table 3 and Methods). The RMP under these condi- on R . Hence, in mossy fiber boutons, K 7 channels generate N V tions was −79.1 and −78.6 mV upon removal of the K 7 con- an inhibitory current that restricts R at positive potentials to ductance. This small (0.5 mV) depolarization caused by loss of rest and thereby, limits the number of presynaptic spikes elicited the K 7 conductance is within experimental error and would by depolarization. not have been detected in experiments. Ablation of the K 7 conductance, though, enhanced the R from 1.00 GΩ to 2.33 GΩ K 7 currents reduce spike width and afterdepolarization.In when measured using depolarizing subthreshold current pulses most neurons, single or trains of spikes invade synaptic terminals, (Fig. 4a). These findings are consistent with experimental obser- 1–3 leading to neurotransmitter release . Thus, we injected very vations (Fig. 2). short current pulses (0.1 ms) to evoke single or trains of action Next, we simulated an action potential with and without the potentials at various frequencies (1, 5, 20, 50, and 100 Hz; Fig. 3). K 7 conductance (Fig. 4b). In agreement with our experimental Each spike was followed by a small intrinsically generated after- observations (Fig. 3, Supplementary Table 2), the spike width depolarization (ADP) (Fig. 3a). This ADP duration was ≥ 150 ms, broadened from 0.55 ms to 0.75 ms upon removal of the K 7 such that with spike trains generated at ≥ 20 Hz, the spikes did conductance. In the absence of the K 7 conductance (Fig. 4b), the not initiate during the ADP (Fig. 3a). Inhibition of K 7 currents spike amplitude was higher than controls. The model suggested 4 COMMUNICATIONS BIOLOGY | (2019) 2:145 | https://doi.org/10.1038/s42003-019-0408-4 | www.nature.com/commsbio Control XE991 Control Linopirdine Control XE991 Control Linopirdine Control XE991 Control Linopirdine RMP (mV) R (GOhms) R (GOhms) AP No. COMMUNICATIONS BIOLOGY | https://doi.org/10.1038/s42003-019-0408-4 ARTICLE Fig. 3 K 7 currents constrain the spike width and ADP following action Control 5 Hz train potentials in mossy fiber boutons. a Example 5 Hz train of action potentials +XE991 obtained under whole-cell current-clamp conditions before and after XE991 20 mV application at RMP (indicated adjacent to the trace). Each action potential 2 ms was elicited by applying a 0.1 ms depolarizing current injection. The last action potential of the train with the associated afterdepolarization (ADP) is shown on a larger scale on the right. The inset shows the action potential ADP shape with and without XE991. b, c Graphs depicting the ADP amplitude –81 20 mV and decay time constant (τ) as well as the action potential (AP) half-width mV 20 mV 20 ms and amplitude respectively in the absence (con) and presence of either 1 s XE991 or linopirdine. Open and filled squares represent the individual and n = 7 mean values respectively. The data shown are of the 1st action potential b n = 4 and ADP kinetics obtained for a train at 5 Hz. The numbers of observations n = 4 are indicated above each set of bars. d Representative traces showing trains n = 8 of 20 action potentials elicited at 50 Hz in the absence (control) and presence of XE991. The first action potential is shown on an expanded scale in the inset. e The individual (open square) and mean (filled square) K 7 channel inhibitor-induced ADP area associated with 20, 50, and 100 Hz trains of 20 action potentials. The numbers of observations are indicated in parenthesis. In all graphs, significance at p < 0.05 when compared with the appropriate control is indicated by asterisks (*) c n = 8 n = 4 * Removing the K 7 conductance also resulted in the generation 2.0 n = 4 n = 8 of an ADP with a decay time constant of 22.9 ms (Fig. 4b). 1.5 In agreement with our experimental observations, a train of 20 action potentials at a frequency of 50 Hz resulted in successive 1.0 action potentials being initiated near the peak of the ADP generated by the previous spike (Fig. 4d). The enhanced ADP 0.5 20 following spikes in the absence of the K 7 conductance was 2+ 0.0 0 abolished when the Ca conductance was removed (Fig. 4b, d). These findings imply that the K 7 current in boutons serves 2+ to suppress a rise in intracellular Ca concentration during an action potential and regulates the spike width and generation of an ADP succeeding the presynaptic spike. d 50 Hz train e 100 Hz (8) 20 mV BAPTA prevents K 7 inhibitor effects on spike with and ADP. 2 ms 40 To test whether the effects of K 7 channel inhibition on the spike 2+ half-width and ADP is owing to a rise in intracellular Ca , 2+ we included the Ca chelator, BAPTA (10 mM or 20 mM) in 20 Hz the intracellular patch pipette (see Methods). The findings with 50 Hz (8) Control 10 mM and 20 mM BAPTA were no different and have been (8) (4) (4) (4) +XE991 10 grouped together. Full dialysis of BAPTA into boutons occurred 20 mV within 5 min. Stable recordings could be obtained for at least 200 ms 30 min with BAPTA in the intracellular solution. Under these –82 conditions, 20 min application of XE991 had little effect on the mV single action potential half-width or amplitude (Fig. 5a, Supple- mentary Table 1). In addition, XE991 had little effect on the ADP amplitude that a reduction in spike amplitude, consistent with experiments, or decay time constant elicited by single action potentials at could be produced by reducing the Na conductance by 20% either 1 or 5 Hz when BAPTA was included in the patch pipette (Supplementary Fig. 1). (Fig. 5, Supplementary Table 1). This was independent of the 2+ We also modeled the changes in Ca concentration induced spike frequency or the position of the spike within the train 47–49 by a single spike (Fig. 4c). Consistent with previous studies , (Supplementary Table 1). Further, there was no increase in the the simulations predicted that with a K 7 conductance (i.e., ADP generated during a train of action potentials elicited at under control conditions), an action potential caused the frequencies of 20, 50, or 100 Hz in the presence of XE991 when 2+ intracellular Ca concentration to increase by 100 nM (Fig. 4c). BAPTA was present in the patch pipette solution (Fig. 5). Thus, Removal of the K 7 conductance induced a further rise in these findings further support the suggestion that K 7 channel 2+ 2+ V intracellular Ca by 60 nM (Fig. 4c). Given that the Ca inhibition in mossy fiber boutons induces a rise in intracellular 50,51 conductance is activated during a spike , the model suggests 2+ Ca that leads to spike broadening and an enhanced ADP 2+ that the rise in intracellular Ca during the spike would have following presynaptic spikes. contributed to the spike broadening in the absence of the K 7 conductance. Indeed, the spike width in our simulations without 2+ 2+ the K 7 and Ca conductance was very similar to control K 7 currents limit Ca influx to affect spike width and ADP. V V conditions (i.e., when both conductances were active; Fig. 4b). Next, we asked what might be the source of the rise in COMMUNICATIONS BIOLOGY | (2019) 2:145 | https://doi.org/10.1038/s42003-019-0408-4 | www.nature.com/commsbio 5 Con +XE991 Con +Linopirdine Con +XE991 Con Con +XE991 +Linopirdine Con +Linopirdine Con +XE991 Con +Linopirdine XE991 Linopirdine XE991 Linopirdine XE991 Linopirdine AP Half-width (ms) ADP amplitude (mV) ADP Decay τ (ms) K 7 channel inhibitor -induced V AP amplitude (mV) ADP area (mV*ms ×10 ) ARTICLE COMMUNICATIONS BIOLOGY | https://doi.org/10.1038/s42003-019-0408-4 2+ a Control (+K 7 and +Ca conductance) 2+ –K 7 conductance with Ca conductance 0.5 mV 500 ms –79.1 0.3 pA mV 2+ Control (+K 7 and +Ca conductance) 2+ –K 7 conductance with Ca conductance 2+ –K 7 conductance and –Ca conductance Action potential Action potential-induced 2+ Ca transient following removal of K 7 conductance 20 mV 1 ms 20 nM 500 ms ADP 20 mV –80 –80 2+ mV mV Control spike-induced Ca transient 10 ms Truncated action potentials 2 mV 100 ms ADP –80 mV 20 mV 100 ms Fig. 4 Computational model illustrating that the spike half-width and ADP proceeding spikes in the absence of a K 7 conductance is owing to enhanced 2+ intracellular Ca . a Simulation showing the voltage change in response to a subthreshold 1 s square depolarization in the absence and presence of a K 7 conductance. The traces under the two conditions have been superimposed. b Simulated single action potentials elicited by a short depolarizing step with 2+ and without a K 7 conductance. In addition, a single action potential in the absence of both the K 7 and Ca conductances was also generated (blue). The V V 2+ inset shows the individual action potentials on a larger scale. c Intracellular Ca changes in response to the same stimulus that elicited an action potential with (black) and without (red) the K 7 conductance. d A train of action potential waveforms at 50 Hz when the K 7 conductance was intact (black), V V 2+ 2+ following removal of the K 7 conductance in the presence of the Ca conductance (red) or after removal of both the K 7 and Ca conductances (blue). V V The inset shows the ADP following action potentials under the three conditions on an expanded scale 2+ intracellular Ca in the absence of K 7 channels? Mossy fiber (Fig. 6b, c, Supplementary Table 2), there was little difference 2+ boutons express predominantly P/Q-type voltage-gated Ca in spike width with and without the inhibitors (Fig. 6b, c, 2+ 51 channels as well as N- and R-type Ca channels . The N- and Supplementary Table 2). Further, we found that the ADP P/Q-type channels are activated during presynaptic spikes in amplitude and decay time constant following spikes occurring 51 2+ mossy fiber boutons . We, therefore, investigated whether Ca at 1 Hz and 5 Hz was significantly reduced in the presence of 2+ influx via these channels might contribute towards the spike XE991 and N- and P/Q-type Ca channel inhibitors compared broadening and increase in ADP amplitude and decay time with that controls (Fig. 6b, c, Supplementary Table 1). constant caused by K 7 channel inhibition. For this, we tested As previously found (Fig. 3d), action potential trains at the effects of co-application of 3 μM XE991 and the selective N- ≤ 20 Hz resulted in spikes being generated on the ADP under 2+ and P/Q-type Ca channel inhibitors, 100 nM ω-conotoxin control conditions (Fig. 6d). Co-application of XE991 and N- 2+ GVIA and 100 nM ω-agatoxin IVA. With these inhibitors, the and P/Q-type Ca channel inhibitors resulted in a reduction R measured by applying depolarizing, but not hyperpolarizing, of the ADP area compared with control conditions such that subthreshold steps was still enhanced compared with that prior spikes now occurred at the normal RMP (i.e., at the baseline; to application of the compounds (Fig. 6a). Fig. 6d, e). Altogether, these findings robustly support 2+ We then elicited trains of action potentials at different the notion that the K 7 current reduces Ca influx via N- type 2+ frequencies in the absence and presence of XE991, ω-conotoxin and P/Q-type Ca channels and, thereby, regulates the GVIA and ω-agatoxin IVA. Whereas the spike amplitude spike width and ADP generated following spikes in mossy fiber was still significantly reduced by application of the inhibitors boutons. 6 COMMUNICATIONS BIOLOGY | (2019) 2:145 | https://doi.org/10.1038/s42003-019-0408-4 | www.nature.com/commsbio COMMUNICATIONS BIOLOGY | https://doi.org/10.1038/s42003-019-0408-4 ARTICLE –10 pA +10 pA a 5 Hz train a Control +XE991 2+ (6) (6) + N/P/Q Ca inhibitors (6) 1.6 20 mV (6) 2 ms 20 mV 1.2 500 ms –85 0.8 + 10 pA 20 mV mV –80 20 ms 0.4 mV 20 mV BAPTA (20 mM) –10 pA BAPTA (20 mM) +XE991 1 s 0.0 n = 11 n = 11 b 5 Hz train Control 20 2+ 20 mV +XE991 + N/P/Q Ca inhibitors 5 ms 20 mV 20 ms c d 50 Hz train –85 mV 20 mV 20 mV (11) (11) 1 s 2 ms (11) c (n = 6) 0 (n = 6) (n = 6) 14 160 120 4 * (n = 6) –2 12 120 3 20 mV –4 80 60 2 200 ms –6 –85 40 1 mV BAPTA (20 mM) 0 0 0 0 BAPTA (20 mM) +XE991 2+ Fig. 5 The Ca chelator, BAPTA, prevents K 7 current inhibition-induced spike broadening and enhanced ADP following action potentials. a Example 5 Hz action potential trains when 20 mM BAPTA was included in the patch pipette in the absence and presence of 3 μM XE991. The last action potential d 50 Hz train e with the associated ADP is shown on an expanded time scale. The spike itself is shown in the inset. b Graphs depicting the individual (open squares) (6) (6) (6) and mean (filled square) values for the ADP amplitude and decay time 0.0 20 mV constant (τ) before and after application of XE991 when BAPTA was 2 ms –2.0 included in the patch pipette. c Representative 50 Hz action potential trains in the absence and presence of XE991 with 20 mM BAPTA in the patch –4.0 pipette. The first action potential is shown on an expanded time scale in the inset. d The individual (open square) and mean (filled square) area –6.0 underlying the K 7 current inhibition-induced ADP generated with trains 20 mV of action potentials at 20, 50, and 100 Hz when BAPTA was incorporated 200 ms in the patch pipette solution –85 mV Control +XE991 2+ K 7 channels reduce neurotransmission onto CA3 neurons. + N/P/Q Ca inhibitors 2+ Given that our findings suggest that K 7 channels limit Ca 2+ influx via N- and P/Q-type Ca channels, which regulate 10,51–53 synaptic release from mossy fiber boutons , we hypothe- sized that K 7 channel inhibition should enhance neuro- pyramidal neurons express K 7 channels postsynaptically ,we V V + + transmitter release from mossy fiber boutons onto CA3 replaced intracellular K in the CA3 pyramidal neuron with Cs pyramidal neurons. To investigate this, we obtained cell-attached to inhibit these channels (see Methods). We also included QX314 recordings from a mossy fiber bouton and whole-cell voltage- bromide in the CA3 pyramidal recording solution to inhibit Na clamp recordings from the CA3 pyramidal neuron whose prox- channels. As dentate–gyrus granule cells are most likely to fire imal apical dendrite the bouton was next to in the absence of action potentials phase-locked to theta or gamma rhythms 45 55 glutamate and GABA receptor blockers (see Fig. 7a ). As CA3 in vivo , we initially induced action currents at 5 Hz (i.e., theta COMMUNICATIONS BIOLOGY | (2019) 2:145 | https://doi.org/10.1038/s42003-019-0408-4 | www.nature.com/commsbio 7 BAPTA BAPTA + XE BAPTA BAPTA + XE 20 Hz 50 Hz 100 Hz ADP amplitude (mV) K 7 channel inhibitor-induced ADP area (mV*ms ×10 ) ADP decay τ (ms) 20 Hz 50 Hz 100 Hz ADP amplitude (mV) Control XE991 + 2+ Ca inhibitors ADP decay τ (ms) Control XE991 + 2+ R (GΩ) Ca inhibitors N K 7 channel inhibitor -induced ADP area 3 AP amplitude (mV) (mV*ms ×10 ) Control XE991 + 2+ Ca inhibitors Control XE991 + Control 2+ Ca inhibitors XE991 + 2+ Ca inhibitors AP half-width (ms) Control XE991 + 2+ Ca inhibitors ARTICLE COMMUNICATIONS BIOLOGY | https://doi.org/10.1038/s42003-019-0408-4 2+ 2+ Fig. 6 K 7 currents limit Ca influx partly via N- and P/Q-type Ca a 5 Hz stimulation channels to reduce the action potential half-width and ADP following spikes Action Control XE991 CA3 currents in mossy fiber boutons. a Representative recordings obtained with Stimulus pyramid subthreshold hyperpolarizing and depolarizing current pulses before and Artifact 2+ Cell- after co-application of XE991 and N- and P/Q-type Ca channel inhibitors. attached 100 pA The individual (open squares) and average (filled squares) input resistance 1 ms (R ) values obtained using this protocol are shown in the graph. The pA 100 ms numbers of observations per group are depicted above each bar. Action currents b, d Typical 5 and 50 Hz trains of action potentials under control conditions 2+ and following application of XE991 and N- and P/Q-type Ca channel 50 pA 100 ms inhibitors. The first and last action potential in the 50 and 5 Hz trains, respectively, are shown in the insets. The ADP associated with the last action potential in the 5 Hz train is also illustrated on a larger scale in the * * inset in b. c The individual (open squares) and mean (filled squares) 25 peak ADP amplitude and decay time constant (τ)) and the action potential (AP) amplitude and half-width associated with the first action potential of a 5 Hz train of action potentials in the absence and presence of XE991 2+ and N- and P/Q-type Ca channel inhibitors. e The individual neuron (open squares) and average (filled squares) area of the K 7 channel inhibitor-induced ADP with 20, 50, and 100 Hz trains of action potentials 2+ in the presence of N- and P/Q-type Ca channel inhibitors. The numbers of observations for each group are shown in parenthesis. Asterisks (*) denote significance at p < 0.05 when compared with appropriate controls c 50 Hz stimulation Truncated Stimulus frequency) in the bouton as in ref. (Fig. 7a). This resulted in Control XE991 artifacts mono-synaptic excitatory post-synaptic currents (EPSCs) in the post-synaptic CA3 neuron with a very low failure rate (% failure 100 pA = 8.33 ± 10.2%, n = 3, Fig. 7b). Subsequent application of XE991 200 ms (3 μM) enhanced the EPSC amplitude and decay time constant (Fig. 7a, b), strongly suggesting that presynaptic K 7 channels constrain neurotransmitter release from mossy fiber boutons. 50 pA As spike trains in granule cells at gamma frequency have a higher probability of initiating action potentials in CA3 pyramidal 200 ms 45,57 neurons during certain types of behavior , we also investigated how presynaptic K 7 channels would affect neurotransmitter release when the spike frequency was 50 Hz. For this, we elicited 1.6 10 action currents at 50 Hz in boutons (Fig. 7c). As has been previously described , this resulted in paired pulse facilitation of 1.2 EPSCs (Fig. 7c, d). Bath application of XE991 significantly reduced 0.8 the paired pulse ratio of evoked EPSCs (Fig. 7c, d), indicating that presynaptic K 7 channels restrict neurotransmitter release from 0.4 mossy fiber boutons independently of spike frequency. Further, these findings robustly suggest that presynaptic K 7 channels 0.0 affect short-term synaptic plasticity at this hippocampal synapse. Discussion Here, we show that mossy fiber boutons possess a K 7 current that was active at rest in these structures (Fig. 1). Despite this, electrophysiological recordings and computational modeling agreement with our premise that K 7 currents restrict spike- 2+ indicated that the current had little effect on bouton RMP (Fig. 2, induced Ca influx in mossy fiber boutons, this current sig- Fig. 4). Electrophysiological recordings and computational mod- nificantly reduced action potential-induced neurotransmitter eling, though, showed that the K 7 current acts as a shunt: release from mossy fiber boutons and short-term synaptic plas- restricting R at depolarizing potentials to rest (Figs. 2 and 4). ticity at the mossy fiber bouton–CA3 synapse (Fig. 7). These Consequently, K 7 currents influenced the probability of spikes findings indicate that K 7 channels play a substantial role in V V elicited with a given depolarization (Fig. 2). Further, K 7 currents modulating intrinsic excitability and synaptic plasticity at mature were found to restrict the spike width and ADP amplitude and synapses in the hippocampus. Our findings also imply that K 7 decay following a spike (Fig. 3). Computational modeling sug- channels are likely to be involved in processes underpinning 2+ gested that this was due to K 7 currents opposing Ca influx via information storage in the hippocampus. 2+ N- and P/Q-type Ca channels during an action potential The mossy fiber bouton K 7 current half–maximal activation 2+ (Fig. 4). Consistent with this, the presence of either the Ca voltage was more negative (−70 mV, Fig. 1d) than that in hip- 2+ chelator, BAPTA, or the N- and P/Q-type Ca channel inhibi- pocampal and neocortical pyramidal neuron somata and axon 39,58 tors prevented the spike width broadening and the increase in initial segments but consistent with that at granule cell 40 59 24 ADP after K 7 current inhibition (Figs. 5 and 6). Moreover, in somata , in peripheral axons and in Calyx of Held terminals . 8 COMMUNICATIONS BIOLOGY | (2019) 2:145 | https://doi.org/10.1038/s42003-019-0408-4 | www.nature.com/commsbio Control XE991 Control XE991 Control XE991 EPSC amplitude (pA) PPR EPSC decay τ (ms) COMMUNICATIONS BIOLOGY | https://doi.org/10.1038/s42003-019-0408-4 ARTICLE 2+ Fig. 7 Presynaptic K 7 currents reduce neurotransmitter release from Our computational model included a Ca conductance whose 2+ mossy fiber boutons onto CA3 pyramidal neurons. a Schematic illustrating decay time constants were slowed down to mimic the Ca the paired recording configuration obtained from CA3 pyramidal somata transient generated in response to a single action potential in a 2+ 47–49 and a mossy fiber bouton synapsing onto its proximal apical dendrite. Also mossy fiber terminal obtained using Ca imaging . The 2+ shown are typical single EPSCs elicited in CA3 pyramidal neurons at a fixed decay time constant of the Ca conductance was considerably 2+ potential of − 70 mV when action currents at a frequency of 5 Hz were slower than that reported for either N-, P/Q-or R-type Ca evoked in mossy fiber boutons that contacted their proximal apical currents in mossy fiber terminals and may reflect processes 2+ 2+ 48 dendrites. The EPSCs were obtained in the absence and presence of 3 μM such as Ca -induced Ca release or the effects of endogenous 2+ 47 2+ XE991. The action currents are shown on an expanded scale in the inset buffers on intracellular Ca . The presence of this Ca (red box). The scales associated with the first pair of traces apply to both conductance resulted in an ADP following K 7 conductance sets of traces. b The average (filled squares) EPSC amplitudes and decay removal in our simulations (Fig. 4), albeit smaller and faster than time constants (τ) when evoked at 5 Hz in three different mossy fiber that observed following single action potentials under our bouton–CA3 pyramid pairs before and after XE991 application. The open experimental conditions (Fig. 3). This suggests that, as in CA1 68 2+ squares represent the mean of 10 EPSC amplitudes and decay time pyramidal neurons ,aCa tail current as well as processes such 2+ 2+ constants obtained from individual mossy fiber bouton–CA3 pyramid pairs. as Ca -induced Ca release may contribute to the ADP gen- c Example recordings of 10 EPSCs evoked in a CA3 pyramidal cell by a train eration following K 7 current suppression. However, as the of 50 Hz action currents in a mossy fiber bouton synapsing onto the neuron simulated ADP amplitude and decay are clearly different from under control conditions and following XE991 application. The scales shown that observed experimentally (Fig. 3), other conductances, such 2+ on the first pair of traces apply to the second. The amplitudes of the first as Ca -activated conductances, may also contribute to the K 7 and second EPSCs in the train were measured to obtain the paired pulse current inhibition-induced ADP in terminals. ratio (PPR). d The average (filled square) and individual PPRs (open The main physiological role of presynaptic K 7 channels in squares) from three bouton–CA3 neuron pairs without and with XE991. mossy fiber boutons is to regulate synaptic transmission onto Asterisks (*) denote significance at p < 0.05 when compared with CA3 pyramidal neurons (Fig. 7). As spike broadening has been appropriate controls associated with enhanced neurotransmission from synaptic 5,7,14,73 terminals and K 7 current inhibition results in wider action potentials in mossy fiber boutons (Fig. 3), this may have contributed to the greater action potential-induced neuro- Hence, ~ 20% of the current was active at rest in mossy transmitter release in the absence of K 7 currents (Fig. 7). Fur- fiber bouton terminals. The lack of effect of the current on ther, the enhanced ADP resulting from K 7 current inhibition is 2+ the bouton RMP is probably owing to the inward rectifier probably due to augmented Ca influx and may result in ele- K current as this has been shown to exert a strong influence vated asynchronous release from these terminals. This, though, 42,43,60 on RMP in granule cells . Unlike in heterologous needs to be further tested. Notwithstanding, the mechanisms systems , though, K 7 channel inhibitors exert their effects by which a reduction in K 7 currents in mossy fiber boutons V V at negative potentials in native cells as they depolarized the might lead to greater neurotransmitter release differs from that RMP and enhanced R in hippocampal and neocortical pyr- in Calyx of Held terminals, whereby RMP depolarization was 18,24,39,58,62,63 amidal neurons . Further, axonal K 7 currents largely attributed to the increase in neurotransmission following in peripheral nerve fibers have a XE991-sensitive component K 7 current block . The effect of K 7 currents on intrinsic V V at ~ −70 mV . In mossy fiber boutons, consistent with our excitability and synaptic transmission may also vary in different computational model (Fig. 4a), XE991 increased R measured conditions. Indeed, during elevated extracellular K conditions using small subthreshold depolarizing pulses in a time-dependent when axons are already depolarized, these currents in CA3 axons manner too (Fig. 2f). have been suggested to influence Na current inactivation and 2+ K 7 currents affect mossy fiber bouton terminal excitability increase presynaptic action potential amplitude, Ca influx 24 66 distinctly from that of Calyx of Held terminals . In particular, and enhance synaptic release . Thus, K 7 currents may have unlike in Calyx of Held terminals, K 7 currents regulate the spike synapse-specific effects on intrinsic excitability which may width and ADP following spikes in mossy fiber boutons. These depend on particular conditions. effects are unlikely to be due to non-selective effects of K 7 Given that K 7 currents in mossy fiber boutons significantly V V channel inhibitors as, at the concentrations utilized, XE991 and augmented the membrane conductance (Figs. 2 and 4), it is linopirdine were at least 50-fold and three fold more potent, probable that, like in Calyx of Held terminals , the current will + 64 respectively, for K 7 than other K channels . affect excitatory synaptic potential amplitudes and shapes (i.e., K 7 currents also restrict the generation of an intrinsic analog signaling) in these terminals. As analog signaling in ADP succeeding spikes in CA1, CA3 and cortical pyramidal mossy fiber boutons influences neurotransmission , this might 58,62,65–67 neurons . In CA1 and cortical pyramidal neurons, K 7 be another mechanism by which K 7 currents might regulate V V currents counteract a persistent Na current to reduce the ADP synaptic release. Hence, K 7 currents might affect neuro- 58,62,65 2+ following spikes . In these neurons, R-type Ca tail cur- transmission by multiple mechanisms, including action potential- 68 2+ rents and Ca -activated cation channels such as TRPC channels dependent release (Fig. 7). 40,69–71 activated via G-protein coupled receptors can contribute What effect might modulation of neurotransmitter release to the ADP following spikes too. However, whilst hippocampal by K 7 channels have on CA3 neural network excitability? granule cell somata and dendrites highly express TRPC subunits, Our findings suggest that K 7 channels restricts neurotransmitter immunohistochemistry suggested that TRPC subunits are release from mossy fiber boutons elicited by trains of action present intracellularly within mossy fiber axons and boutons . potentials occurring at theta and gamma frequencies and will 2+ We cannot, though, exclude the possibility of Ca -activated con- thereby, contribute to maintaining CA3 pyramidal neuronal ductances underlying the ADP generated by K 7 current inhibition. excitability (Fig. 6). Although the giant mossy fiber bouton–CA3 Since we do not know if these are present in mossy fiber terminals pyramidal neuron connectivity is sparse (with ~ 50 granule cells or their biophysical properties, we were unable to include them contacting one CA3 neuron), each bouton contains an average 33,74,75 in our computational model (see Methods and Supplementary of 20 release sites . Thus, changes in K 7 channel activity Table 3). could have an impact on the overall CA3 pyramidal neuron COMMUNICATIONS BIOLOGY | (2019) 2:145 | https://doi.org/10.1038/s42003-019-0408-4 | www.nature.com/commsbio 9 ARTICLE COMMUNICATIONS BIOLOGY | https://doi.org/10.1038/s42003-019-0408-4 excitability. Indeed, a loss of K 7.2 subunits, which are present on decay time constant, the decay phase of the ADP was fitted with a double expo- 26–28 nential function: mossy fibers and, most likely, their boutons , led to impaired hippocampal gamma rhythms and spatial learning as well as ðÞ t=τ1 ðÞ t=τ2 ADP decay ¼ A e þ A e 1 2 spontaneous seizures in rodents, implicating hyperexcitability of cortical neural networks . Thus, the presence of these channels where τ1 and τ2 represent time constants of the initial and falling phase of the at the mossy fiber bouton, which is a conditional detonator ADP. τ2 values have been reported in Results and Figures. In addition, the area synapse, is likely to be vital for preventing CA3 neuronal and under the ADP generated during and following a train of action potentials at 20, 50, and 100 Hz was measured before and after application of XE991 (3 μM). There neural circuit hyperactivity. Further, as the K 7 current sig- was some variability (albeit non-significant) in these parameters between control nificantly affects short-term synaptic plasticity at the mossy recordings from individual mossy fiber boutons (see Fig. 3, Fig. 5, Supplementary fiber–CA3 synapse (Fig. 6), these channels at this synapse may Table 1). also affect long-term synaptic plasticity and thus, memory For paired bouton–CA3 recordings, the amplitude of the EPSCs generated in response to the action current in the bouton was measured. Both first and encoding. second EPSC amplitudes were measured from their directly preceding baselines, respectively. The 10–90% rise time of the EPSC was obtained in Clampfit 10.4 Methods using the function: Acute slice preparation. The UK Home Office approved all procedures. Hippo- campal slices were prepared as described in . In brief, 22–28 day-old male Sprague Slope ¼ A  0:8=ðÞ t2  t1 Dawley rat pups were decapitated, the brain removed and submerged in ice-cold solution (mM): 87 NaCl, 25 NaHCO , 10 glucose, 75 sucrose, 2.5 KCl, 1.25 Where A is the peak amplitude of the EPSC and t1 and t2 are the times at 10% and NaH PO , 0.5 CaCl , 7 MgCl , pH 7.3, 325 mOsm/L. The brain was hemi-sected 2 4 2 2 90% of A, respectively. The decay phase of the EPSC was fitted with the above and a cut parallel to the dorsal part of the brain made. The dorsal side brain halves double exponential equation with τ1 and τ2 represent time constants of the initial were glued onto a slice holder and 300–400 μm slices made (Leica VT1200S, Leica, and falling phase of the EPSC. Again only τ2 values are reported. Paired pulse UK). Slices were incubated in the cutting solution for 30–40 min at 35 °C and then ratios were calculated as the peak of the second EPSC divided by the peak of stored in the cutting solution at room temperature. the first EPSC. For K 7/M current voltage-clamp data, the traces obtained in the presence of the XE991 (3 μM) were subtracted from those in the absence. The subtracted Electrophysiological recordings. Slices were transferred to a submerged chamber traces were fitted with the above double exponential function with τ1 and τ2, containing external solution (mM): 125 NaCl, 25 NaHCO , 25 glucose, 2.5 KCl, representing the de-activation time constants of the initial and late phase of the 1.25 NaH PO , 2 CaCl , 1 MgCl , 0.05 CNQX, 0.05 DL-AP5, 0.01 bicuculline, 2 4 2 2 K 7 current. 0.001 CGP 55845, pH 7.3, 32–36 °C. For whole-cell current-clamp recordings, V The K 7 conductance values were generated from the normalized amplitudes of the internal pipette solution contained (mM): 120 KMeSO4, 15 KCl, 10 HEPES, 2 V the subtracted currents . For whole-cell voltage-clamp experiments, the absolute MgCl2, 0.2 EGTA, 2 Na ATP, 0.3 Tris-GTP and 14 Tris-phosphocreatinine, pH 7.3 voltage recorded was subtracted from the estimated reversal potential of K (E ). with KOH, 295–300 mOsm/L. In some experiments, 10 or 20 mM K BAPTA was K This together with the current amplitude recorded was used to calculate the added to the pipette solution. In this case, the KMeSO was reduced accordingly conductance and were plotted against the absolute voltage. The curves were fitted to 50 or 60 mM, respectively and osmolarity adjusted by adding N-methyl- using the Boltzmann equation: d-glucamine. Pipettes had resistances of 5–8MΩ. In all experiments, Neurobiotin (0.2% w/v) was included in the intracellular pipette solution. Slices were fixed in 4% paraformaldehyde and stained with streptavidin Alexa Fluor 488 conjugate y ¼ A2 þðÞ A1  A2 =ðÞ 1 þ expðÞ ðÞ x  x0 =dx (0.04 mg/ml) 24 h later . Electrophysiological recordings were made using a Multiclamp 700B amplifier where A1 and A2 are the initial and maximum values, x0 is the half-activation (Molecular Devices, UK). Current-clamp recordings were filtered at 10 kHz, and voltage and dx is the slope of the curve. sampled at 50 kHz. Protocols (including that of R (Fig. 2), action potentials initiated with depolarizing steps (Fig. 2) and trains of action potentials (Fig. 3) were Statistical analysis. Group data are expressed as mean ± SEM. In all experiments, applied every minute after application of K 7 current inhibitors to facilitate a minimum of three brain slice preparations made from three independent animals their effects during these recordings . Data were acquired using pClamp 10.0 were used. For experiments involving pharmacological drug application (i.e., (Molecular Devices, UK). Series resistance was in the order of 10–30 MΩ. 2+ XE991, linopirdine or N- and P/Q-type voltage-gated Ca channel inhibitors), Recordings were discarded if the series resistance increased by > 20%. paired t tests were used with statistical significance determined to be p < 0.05. K 7 current recordings: The external solution was supplemented with 0.001 mM Significant differences at p < 0.05 is indicated as asterisks (*) in all figures. tetrodotoxin and 0.1 mM 4-aminopyridine. The internal solution described above was present in the patch pipette. For outside-out recordings, the whole-cell configuration was first obtained and the patch pipette slowly withdrawn. Computational modeling. All simulations were carried out using the NEURON Series resistance was between 10 and 20 MΩ and was ~ 70% compensated. A simulation environment (v7.5) . All model and simulation files will be uploaded to de-activation protocol (as described in ref. ; Fig. 1b) was applied in the absence the ModelDB database (https://senselab.med.yale.edu/modeldb/ accession no. and presence of the K 7/M–channel blocker, XE991 (3 μM). Recordings were 245417). The mossy fiber synaptic bouton was modeled as a single compartment 2 2 filtered at 1 kHz and sampled at 10 kHz. (length = 3.5 μm, diameter = 2 μm, Cm = 1 μF/cm , Rm = 30 kΩ/cm , Ra = 150 Ω Paired bouton–CA3 recordings: Cell-attached recordings from mossy fiber cm). Temperature was set at 34 °C. Active properties included a transient Na + + boutons were obtained. The internal solution was as described above. Action conductance, four types of K currents (delayed rectifier type K conductance, + + currents were elicited in the cell-attached mode by applying 800 mV, 0.1 ms A-type K conductance, K 7 conductance, and inward rectifier type K 2+ 2+ pulses. The internal solution for CA3 neuron whole-cell recordings contained conductance), a Ca conductance (which is owing to all Ca conductances in 2+ (mM): 135 CsCl, 5 QX314 bromide, 10 HEPES, 2 MgCl2, 0.2 EGTA, 2 Na ATP, the bouton including N-, P/Q- and R-type Ca conductances), and a simple 2+ 0.3 Tris-GTP and 14 Tris-phosphocreatinine, pH 7.3 with CsOH, 295–300 mOsm/ Ca -extrusion mechanism with a 500 ms time constant, which is consistent with 47–49 + L. Glutamate and GABA receptor blockers were omitted from the external solution. that reported by other studies . Kinetics for the delayed rectifier type K + 2+ Voltage-clamp recordings were obtained from CA3 pyramidal cells using a conductance, A-type K conductance and Ca conductance were taken from a 39 + Multiclamp 700B amplifier (Molecular Devices, UK). Recordings were filtered previously published model (ModelDB accession no. 112546); the Na kinetics 44 + at 1 kHz, and sampled at 10 kHz. Post-synaptic series resistance was in the was implemented as in ref. ; the inward rectifier K conductance was imple- order of 10–20 MΩ. Recordings were discarded if the series resistance increased mented as in ref. (downloaded from ModelDB, accession no. 185355). The peak by > 20%. conductances used in all simulations and kinetic parameters modified with respect All reagents were purchased from Sigma-Aldrich UK apart from tetrodotoxin, to their original values are reported in Supplementary Table 3. Increasing the peak bicuculline, CGP 55845, DL-AP5 and XE991, which were obtained from Abcam conductance of the delayed rectifier conductance had little effect on the simula- 2+ Ltd (UK). Neurobiotin was acquired from Vector Laboratories Ltd and streptavidin tions. The Ca conductance peak value and decay time constant were adjusted to 2+ Alexa Fluor 488 was procured from Life Technologies. reflect Ca transient measured in response to an action potential in a mossy fiber bouton. The effects of XE991 application were modeled with a complete block of the K 7 conductance. A single action potential was elicited with a current pulse Data analysis. Clampfit (v10.4 or v10.7) was used. To calculate R , the difference of 30 pA for 0.35 ms. To simulate a 50 Hz stimulation, the model was stimulated in steady-state voltage in the last 25 ms elicited by 1 s hyperpolarizing step at every 20 ms with 30 pA 0.3 ms long current pulses. Input resistance was measured − 80 mV was divided by the applied current. Action potentials elicited by 1 s from the voltage deflection caused by a 1 s long 0.3 pA current injection. depolarizing steps were counted. Action potential height was measured from threshold to the peak, whereas action potential width was the breadth at half the height. The amplitude and decay time constant of the ADP following Reporting summary. Further information on research design is available in single action potentials evoked at 1 or 5 Hz were also measured. To obtain the the Nature Research Reporting Summary linked to this article. 10 COMMUNICATIONS BIOLOGY | (2019) 2:145 | https://doi.org/10.1038/s42003-019-0408-4 | www.nature.com/commsbio COMMUNICATIONS BIOLOGY | https://doi.org/10.1038/s42003-019-0408-4 ARTICLE Data availability 24. Huang, H. & Trussell, L. O. KCNQ5 channels control resting properties and All experimental data generated or analyzed during this study are included in this article, release probability of a synapse. Nat. Neurosci. 14, 840–847 (2011). supplementary information files, and is available from the authors upon reasonable 25. Fidzinski, P. et al. KCNQ5 K channels control hippocampal synaptic request. inhibition and fast network oscillations. Nat. Commun. 6, 6254 (2015). 26. Cooper, E. C., Harrington, E., Jan, Y. N. & Jan, L. Y. M channel KCNQ2 subunits are localized to key sites for control of neuronal network Code availability oscillations and synchronization in mouse brain. J. Neurosci. 21, 9529–9540 The complete set of files for the computational model are available on ModelDB (https:// (2001). senselab.med.yale.edu/modeldb/ accession no. 245417). 27. Geiger, J., Weber, Y. G., Landwehrmeyer, B., Sommer, C. & Lerche, H. Immunohistochemical analysis of KCNQ3 potassium channels in mouse brain. Neurosci. Lett. 400, 101–104 (2006). Received: 31 October 2018 Accepted: 29 March 2019 28. Weber, Y. G. et al. Immunohistochemical analysis of KCNQ2 potassium channels in adult and developing mouse brain. Brain Res. 1077,1–6 (2006). 29. Acsady, L. & Kali, S. Models, structure, function: the transformation of cortical signals in the dentate gyrus. Prog. Brain Res. 163, 577–599 (2007). 30. Henze, D. A., Urban, N. N. & Barrionuevo, G. The multifarious hippocampal mossy fiber pathway: a review. Neuroscience 98, 407–427 (2000). References 31. Jaffe, D. B. & Gutierrez, R. Mossy fiber synaptic transmission: 1. Schneggenburger, R. & Rosenmund, C. Molecular mechanisms governing communication from the dentate gyrus to area CA3. Prog. Brain Res. 163, 2+ Ca regulation of evoked and spontaneous release. Nat. Neurosci. 18, 109–132 (2007). 935–941 (2015). 32. Pelkey, K. A. & McBain, C. J. Target-cell-dependent plasticity within the 2. Kaeser, P. S. & Regehr, W. G. Molecular mechanisms for synchronous, mossy fibre-CA3 circuit reveals compartmentalized regulation of presynaptic asynchronous, and spontaneous neurotransmitter release. Annu. Rev. Physiol. function at divergent release sites. J. Physiol. 586, 1495–1502 (2008). 76, 333–363 (2014). 33. Rebola, N., Carta, M. & Mulle, C. Operation and plasticity of hippocampal 3. Kavalali, E. T. The mechanisms and functions of spontaneous CA3 circuits: implications for memory encoding. Nat. Rev. Neurosci. 18, neurotransmitter release. Nat. Rev. Neurosci. 16,5–16 (2015). 208–220 (2017). 4. Trimmer, J. S. Subcellular localization of K channels in mammalian brain 34. Leao, R. N., Tan, H. M. & Fisahn, A. Kv7/KCNQ channels control action neurons: remarkable precision in the midst of extraordinary complexity. potential phasing of pyramidal neurons during hippocampal gamma Neuron 85, 238–256 (2015). oscillations in vitro. J. Neurosci. 29, 13353–13364 (2009). 5. Hoppa, M. B., Gouzer, G., Armbruster, M. & Ryan, T. A. Control and 35. Peters, H. C., Hu, H., Pongs, O., Storm, J. F. & Isbrandt, D. Conditional plasticity of the presynaptic action potential waveform at small CNS nerve transgenic suppression of M channels in mouse brain reveals functions in terminals. Neuron 84, 778–789 (2014). neuronal excitability, resonance and behavior. Nat. Neurosci. 8,51–60 (2005). 6. Foust, A. J., Yu, Y., Popovic, M., Zecevic, D. & McCormick, D. A. Somatic 36. Singh, N. A. et al. Mouse models of human KCNQ2 and KCNQ3 mutations membrane potential and Kv1 channels control spike repolarization in cortical for benign familial neonatal convulsions show seizures and neuronal plasticity axon collaterals and presynaptic boutons. J. Neurosci. 31, 15490–15498 (2011). without synaptic reorganization. J. Physiol. 586, 3405–3423 (2008). 7. Begum, R., Bakiri, Y., Volynski, K. E. & Kullmann, D. M. Action potential 37. Delvendahl, I., Weyhersmuller, A., Ritzau-Jost, A. & Hallermann, S. broadening in a presynaptic channelopathy. Nat. Commun. 7, 12102 (2016). Hippocampal and cerebellar mossy fibre boutons - same name, different 8. Nakamura, Y. & Takahashi, T. Developmental changes in potassium currents function. J. Physiol. 591, 3179–3188 (2013). at the rat calyx of Held presynaptic terminal. J. Physiol. 581, 1101–1112 38. Brown, D. A. & Adams, P. R. Muscarinic suppression of a novel voltage- (2007). sensitive K+ current in a vertebrate neurone. Nature 283, 673–676 (1980). 9. Ishikawa, T. et al. Distinct roles of Kv1 and Kv3 potassium channels at the 39. Shah, M. M., Migliore, M., Valencia, I., Cooper, E. C. & Brown, D. A. calyx of Held presynaptic terminal. J Neurosci. 23, 10445–10453 (2003). Functional significance of axonal Kv7 channels in hippocampal pyramidal 10. Bischofberger, J., Engel, D., Frotscher, M. & Jonas, P. Timing and efficacy of neurons. Proc. Natl. Acad. Sci. USA 105, 7869–7874 (2008). transmitter release at mossy fiber synapses in the hippocampal network. 40. Martinello, K. et al. Cholinergic afferent stimulation induces axonal function Pflugers Arch. 453, 361–372 (2006). plasticity in adult hippocampal granule cells. Neuron 85, 346–363 (2015). 11. Dodson, P. D. et al. Presynaptic rat Kv1.2 channels suppress synaptic terminal 41. Wang, H. S. et al. KCNQ2 and KCNQ3 potassium channel subunits: hyperexcitability following action potential invasion. J. Physiol. 550,27–33 molecular correlates of the M-channel. Science 282, 1890–1893 (1998). (2003). 42. Young, C. C. et al. Upregulation of inward rectifier K (Kir2) channels in 12. Southan, A. P. & Robertson, B. Electrophysiological characterization of dentate gyrus granule cells in temporal lobe epilepsy. J. Physiol. 587, voltage-gated K(+) currents in cerebellar basket and purkinje cells: Kv1 and 4213–4233 (2009). Kv3 channel subfamilies are present in basket cell nerve terminals. J. Neurosci. 43. Gonzalez, J. C., Epps, S. A., Markwardt, S. J., Wadiche, J. I. & Overstreet- 20, 114–122 (2000). Wadiche, L. Constitutive and synaptic activation of GIRK channels 13. Alle, H., Kubota, H. & Geiger, J. R. Sparse but highly efficient Kv3 outpace differentiates mature and newborn dentate granule cells. J. Neurosci. 38, BKCa channels in action potential repolarization at hippocampal mossy fiber 6513–6526 (2018). boutons. J. Neurosci. 31, 8001–8012 (2011). 44. Engel, D. & Jonas, P. Presynaptic action potential amplification by voltage- 2+ 14. Geiger, J. R. & Jonas, P. Dynamic control of presynaptic Ca inflow by fast- gated Na+ channels in hippocampal mossy fiber boutons. Neuron 45, inactivating K channels in hippocampal mossy fiber boutons. Neuron 28, 405–417 (2005). 927–939 (2000). 45. Vyleta, N. P. & Jonas, P. Loose coupling between Ca2+ channels and release 15. Kawaguchi, S. Y. & Sakaba, T. Control of inhibitory synaptic outputs by low sensors at a plastic hippocampal synapse. Science 343, 665–670 (2014). excitability of axon terminals revealed by direct recording. Neuron 85, 46. Mateos-Aparicio, P., Murphy, R. & Storm, J. F. Complementary functions 1273–1288 (2015). of SK and Kv7/M potassium channels in excitability control and synaptic 16. Yang, Y. M. et al. Enhancing the fidelity of neurotransmission by activity- integration in rat hippocampal dentate granule cells. J. Physiol. 592, 669–693 dependent facilitation of presynaptic potassium currents. Nat. Commun. 5, (2014). 4564 (2014). 47. Jackson, M. B. & Redman, S. J. Calcium dynamics, buffering, and buffer 17. Lai, H. C. & Jan, L. Y. The distribution and targeting of neuronal voltage-gated saturation in the boutons of dentate granule-cell axons in the hilus. J. ion channels. Nat. Rev. 7, 548–562 (2006). Neurosci. 23, 1612–1621 (2003). 2+ 18. Brown, D. A. & Passmore, G. M. Neural KCNQ (Kv7) channels. Br. J. 48. Scott, R. & Rusakov, D. A. Main determinants of presynaptic Ca dynamics Pharmacol. 156, 1185–1195 (2009). at individual mossy fiber-CA3 pyramidal cell synapses. J. Neurosci. 26, 19. Jentsch, T. J. Neuronal KCNQ potassium channels: physiology and role in 7071–7081 (2006). disease. Nat. Rev. Neuerosci 1,21–30 (2000). 49. Chamberland, S., Timofeeva, Y., Evstratova, A., Volynski, K. & Toth, K. 20. Debanne, D., Campanac, E., Bialowas, A., Carlier, E. & Alcaraz, G. Axon Action potential counting at giant mossy fiber terminals gates information physiology. Physiol. Rev. 91, 555–602 (2011). transfer in the hippocampus. Proc. Natl. Acad. Sci. USA 115, 7434–7439 21. Kole, M. H. & Stuart, G. J. Signal processing in the axon initial segment. (2018). 2+ Neuron 73, 235–247 (2012). 50. Bischofberger, J., Geiger, J. R. & Jonas, P. Timing and efficacy of Ca channel 22. Devaux, J. J., Kleopa, K. A., Cooper, E. C. & Scherer, S. S. KCNQ2 is a nodal K activation in hippocampal mossy fiber boutons. J. Neurosci. 22, 10593–10602 + channel. J. Neurosci. 24, 1236–1244 (2004). (2002). 23. Pan, Z. et al. A common ankyrin-G-based mechanism retains KCNQ and 51. Li, L., Bischofberger, J. & Jonas, P. Differential gating and recruitment of P/Q-, 2+ NaV channels at electrically active domains of the axon. J. Neurosci. 26, N-, and R-type Ca channels in hippocampal mossy fiber boutons. 2599–2613 (2006). J. Neurosci. 27, 13420–13429 (2007). COMMUNICATIONS BIOLOGY | (2019) 2:145 | https://doi.org/10.1038/s42003-019-0408-4 | www.nature.com/commsbio 11 ARTICLE COMMUNICATIONS BIOLOGY | https://doi.org/10.1038/s42003-019-0408-4 2+ 52. Holderith, N. et al. Release probability of hippocampal glutamatergic 72. Nagy, G. A. et al. DAG-sensitive and Ca permeable TRPC6 channels are terminals scales with the size of the active zone. Nat. Neurosci. 15, 988–997 expressed in dentate granule cells and interneurons in the hippocampal (2012). formation. Hippocampus 23, 221–232 (2013). 2+ 53. Castillo, P. E., Weisskopf, M. G. & Nicoll, R. A. The role of Ca channels in 73. Alle, H. & Geiger, J. R. Combined analog and action potential coding in hippocampal mossy fiber synaptic transmission and long-term potentiation. hippocampal mossy fibers. Science 311, 1290–1293 (2006). Neuron 12, 261–269 (1994). 74. Chicurel, M. E. & Harris, K. M. Three-dimensional analysis of the structure 54. Jones, R. T., Faas, G. C. & Mody, I. Intracellular bicarbonate regulates action and composition of CA3 branched dendritic spines and their synaptic potential generation via KCNQ channel modulation. J. Neurosci. 34, relationships with mossy fiber boutons in the rat hippocampus. J. Comp. 4409–4417 (2014). Neurol. 325, 169–182 (1992). 55. Pernia-Andrade, A. J. & Jonas, P. Theta-gamma-modulated synaptic currents 75. Rollenhagen, A. et al. Structural determinants of transmission at large in hippocampal granule cells in vivo define a mechanism for network hippocampal mossy fiber synapses. J. Neurosci. 27, 10434–10444 (2007). oscillations. Neuron 81, 140–152 (2014). 76. Bischofberger, J., Engel, D., Li, L., Geiger, J. R. & Jonas, P. Patch-clamp 56. Vyleta, N. P., Borges-Merjane, C. & Jonas, P. Plasticity-dependent, full recording from mossy fiber terminals in hippocampal slices. Nat. Protoc. 1, detonation at hippocampal mossy fiber-CA3 pyramidal neuron synapses. Elife 2075–2081 (2006). 5, pii: e17977 (2016). 77. Huang, Z. et al. TRIP8b-independent trafficking and plasticity of adult cortical 57. Henze, D. A., Wittner, L. & Buzsaki, G. Single granule cells reliably discharge presynaptic HCN1 channels. J. Neurosci. 32, 14835–14848 (2012). targets in the hippocampal CA3 network in vivo. Nat. Neurosci. 5, 790–795 78. Hines, M. L. & Carnevale, N. T. The NEURON simulation environment. (2002). Neural Comput. 9, 1179–1209 (1997). 58. Battefeld, A., Tran, B. T., Gavrilis, J., Cooper, E. C. & Kole, M. H. Heteromeric Kv7.2/7.3 channels differentially regulate action potential initiation and Acknowledgements conduction in neocortical myelinated axons. J. Neurosci. 34, 3719–3732 This work was supported by the BBSRC (BB/L000679/1; M.M.S), Wellcome Trust (2014). (WT087363MA, M.M.S., D.A.B.), and the EU H2020 program (grant agreement 59. Schwarz, J. R. et al. KCNQ channels mediate IKs, a slow K current regulating No. 785907, HBP SGA2, M.M, EG). We thank Dr. Liyi Li (University of Freiburg, excitability in the rat node of Ranvier. J. Physiol. 573,17–34 (2006). Germany; current address: University of Basel, Switzerland) for invaluable help and 60. Stegen, M. et al. Adaptive intrinsic plasticity in human dentate gyrus advice on mossy fiber bouton patching. granule cells during temporal lobe epilepsy. Cereb. Cortex 22, 2087–2101 (2012). 61. Greene, D. L., Kang, S. & Hoshi, N. XE991 and linopirdine are state- Author contributions dependent inhibitors for Kv7/KCNQ channels that favor activated single K.M. and M.M.S. performed all experiments and analyzed experimental data. E.G. subunits. J. Pharmacol Exp Ther 362, 177–185 (2017). and M.M. performed the computational modeling. M.M.S. conceived the study and 62. Yue, C. & Yaari, Y. Axo-somatic and apical dendritic Kv7/M channels wrote the manuscript, with contributions from all authors. differentially regulate the intrinsic excitability of adult rat CA1 pyramidal cells. J. Neurophysiol. 95, 3480–3495 (2006). 63. Hu, W. & Bean, B. P. Differential control of axonal and somatic resting Additional information potential by voltage-dependent conductances in cortical layer 5 pyramidal Supplementary information accompanies this paper at https://doi.org/10.1038/s42003- 019-0408-4. neurons. Neuron 97, 1315–1326 e1313 (2018). 64. Elmedyb, P. et al. Modulation of ERG channels by XE991. Basic Clin. Pharmacol. Toxicol. 100, 316–322 (2007). Competing interests: The authors declare no competing interests. a+ 65. Yue, C., Remy, S., Su, H., Beck, H. & Yaari, Y. Proximal persistent N channels drive spike afterdepolarizations and associated bursting in adult CA1 Reprints and permission information is available online at http://npg.nature.com/ pyramidal cells. J. Neurosci. 25, 9704–9720 (2005). reprintsandpermissions/ 66. Vervaeke, K., Gu, N., Agdestein, C., Hu, H. & Storm, J. F. Kv7/KCNQ/M- Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in channels in rat glutamatergic hippocampal axons and their role in regulation of excitability and transmitter release. J. Physiol. 576, 235–256 (2006). published maps and institutional affiliations. 67. Brown, J. T. & Randall, A. D. Activity-dependent depression of the spike after- depolarization generates long-lasting intrinsic plasticity in hippocampal CA3 pyramidal neurons. J. Physiol. 587, 1265–1281 (2009). Open Access This article is licensed under a Creative Commons 68. Metz, A. E., Jarsky, T., Martina, M. & Spruston, N. R-type calcium channels Attribution 4.0 International License, which permits use, sharing, contribute to afterdepolarization and bursting in hippocampal CA1 pyramidal adaptation, distribution and reproduction in any medium or format, as long as you give neurons. J. Neurosci. 25, 5763–5773 (2005). appropriate credit to the original author(s) and the source, provide a link to the Creative 69. Yan, H. D., Villalobos, C. & Andrade, R. TRPC channels mediate a muscarinic Commons license, and indicate if changes were made. The images or other third party receptor-induced afterdepolarization in cerebral cortex. J. Neurosci. 29, material in this article are included in the article’s Creative Commons license, unless 10038–10046 (2009). indicated otherwise in a credit line to the material. If material is not included in the 70. Tai, C., Hines, D. J., Choi, H. B. & MacVicar, B. A. Plasma membrane article’s Creative Commons license and your intended use is not permitted by statutory insertion of TRPC5 channels contributes to the cholinergic plateau regulation or exceeds the permitted use, you will need to obtain permission directly from potential in hippocampal CA1 pyramidal neurons. Hippocampus 21, 958–967 the copyright holder. To view a copy of this license, visit http://creativecommons.org/ (2011). licenses/by/4.0/. 71. Park, J. Y. & Spruston, N. Synergistic actions of metabotropic acetylcholine and glutamate receptors on the excitability of hippocampal CA1 pyramidal neurons. J. Neurosci. 32, 6081–6091 (2012). © The Author(s) 2019 12 COMMUNICATIONS BIOLOGY | (2019) 2:145 | https://doi.org/10.1038/s42003-019-0408-4 | www.nature.com/commsbio