Interaction Between HCN and Slack Channels Regulates mPFC Pyramidal Cell Excitability and Working Memory

The ability of monkeys and rats to carry out spatial working memory tasks has been shown to depend on the persistent firing of pyramidal cells in the prefrontal cortex (PFC), arising from recurrent excitatory connections on dendritic spines. These spines express hyperpolarization-activated cyclic nucleotide-gated (HCN) channels whose open state is increased by cAMP signaling, and which markedly alter PFC network connectivity and neuronal firing. In traditional neural circuits, activation of these non-selective cation channels leads to neuronal depolarization and increased firing rate. Paradoxically, cAMP activation of HCN channels in PFC pyramidal cells reduces working memory-related neuronal firing. This suggests that activation of HCN channels may hyperpolarize rather than depolarize these neurons. The current study tested the hypothesis that Na+ influx through HCN channels activates Slack Na+-activated K+ (KNa) channels to hyperpolarize the membrane. We have found that HCN and Slack KNa channels coimmunoprecipitate in cortical extracts and that, by immunoelectron microscopy, they colocalize at postsynaptic spines of PFC pyramidal neurons. A specific blocker of HCN channels, ZD7288, reduces KNa current in pyramidal cells that express both HCN and Slack channels, but has no effect on KNa currents in an HEK cell line expressing Slack without HCN channels, indicating that blockade of HCN channels in neurons reduces K+ +current indirectly by lowering Na+ influx. Activation of HCN channels by cAMP in a cell line expressing a Ca2+ reporter results in elevation of cytoplasmic Ca2+, but the effect of cAMP is reversed if the HCN channels are co-expressed with Slack channels. Finally, we used a novel pharmacological blocker of Slack channels to show that inhibition of Slack in rat PFC improves working memory performance, an effect previously demonstrated for blockers of HCN channels. Our results suggest that the regulation of working memory by HCN channels in PFC pyramidal neurons is mediated by an HCN-Slack channel complex that links activation HCN channels to suppression of neuronal excitability.


Introduction
Neuronal networks of the prefrontal cortex (PFC) subserve working memory, the ability to retain shortterm neuronal information in the absence of sensory stimulation. This aspect of mental representation has been extensively studied in monkeys and rats using spatial working memory tasks, where the persistent ring of layer III neurons across the delay period is considered the cellular basis for spatial working memory [1]. Persistent neuronal ring arises from extensive recurrent excitatory connections, including on dendritic spines in layer III of macaque dorsolateral PFC (dlPFC) [2]. Persistent ring has also been found in rat medial PFC (mPFC) [3,4], although with a shorter duration than in primates.
Disruption of PFC pyramidal cell ring during stress or mental illness leads to working memory de cits and to inappropriate behavior [5,6]. One mechanism by which stress can disrupt working memory performance is through excessive release of catecholamines, which activate cAMP signaling to in uence the activity of ion channels that alter PFC pyramidal cell ring [7]. cAMP signaling can increase the open state of hyperpolarization-activated cyclic nucleotide-gated (HCN) channels, non-selective cation channels that ux Na + into the neuron [8]. Although HCN channels ux an inward, excitatory current, their ultimate in uences are often complex, e.g. altering membrane resistance and neuronal excitability [9,10].
In macaque dlPFC, local increases in cAMP signaling dlPFC have been shown to reduce task-related neuronal ring and impair working memory performance by opening HCN channels [11], while low dose blockade of HCN channels enhances ring [11]. Similarly, blockade of HCN channels in rat mPFC can prevent stress-induced working memory de cits [12]. Thus, understanding HCN channel mechanisms is important for revealing the etiology of cognitive de cits.
An important clue regarding HCN channel functional contributions comes from their differing subcellular locations within neurons in diverse circuits. The most common location for HCN channels on pyramidal cells is on the shaft of their distal apical dendrites, e.g. in macaque primary visual cortex [3], rodent hippocampus [13], and deep layers of rodent cortex [3], including distal dendrites of layer V mPFC [14][15][16]. In layer III dlPFC, however, HCN channels are concentrated on dendritic spines near glutamatergic-like synapses [17], where they are positioned to gate recurrent excitatory inputs needed for working memory [11]. In dlPFC, HCN channels on dlPFC spines are co-localized with multiple cAMP signaling proteins, while those on dlPFC distal dendritic shafts are not [17], suggesting differing regulation and functions.
Indeed, in contrast to some neurons where HCN channel opening is excitatory [18], cAMP opening of HCN channels in the dlPFC reduces task-related neuronal ring [11]. This nding raises the question of how activation of HCN channels, which normally produces depolarization and increased ring, reduces the ring of PFC pyramidal cells.
It is well established that the in ux of cations such as Ca 2+ or Na + can lead to hyperpolarization or reduced excitability by activation of K + conductances [19]. Experiments with olfactory mitral cells have found that Na + entry through HCN channels stimulates K + channels activated by intracellular Na + , termed K Na channels [20]. The two major known K Na channels have been termed Slack and Slick (also K Na 1.1 and K Na 1.2, encoded by the KCNT1 and KCNT2 genes respectively [21][22][23]19]. Although both are very widely expressed in the nervous system, the only cortical region in which immunoreactivity for Slack-B, a major splice isoform, is detected is the frontal cortex [24]. Human mutations that alter the function of Slack channels result in several distinct childhood epilepsies, all of which are associated with very severe intellectual impairment [25]. Moreover, in one of these conditions (ADNLFE, autosomal dominant nocturnal frontal lobe epilepsy) seizures are restricted to the frontal lobes [26-28].
We have tested the hypothesis that Na + in ux through HCN channels activates Slack-B channels. We have found that Slack channels co-immunoprecipitate with HCN1, and that HCN1 and Slack channels colocalize on the same dendritic spine of mPFC pyramidal cells. ZD7288, a speci c blocker of HCN channels, reduces K Na current in pyramidal cells but has no effect on Slack channels when they expressed without HCN1. To determine if Slack channels interact with HCN channels, we rst carried out co-immunoprecipitation (Co-IP) experiments using extracts of mouse cerebral cortex. Immunoprecipitation was carried out using a previously described antibody against the cytoplasmic N-terminal domain that is speci c to the Slack-B isoform that was generated [24,29]. An HCN1 channel band was readily detected in the Slack-B immunoprecipitates, but not in those carried out using control non-immune IgY antibodies (Fig. 1A).
To further characterize the interaction between Slack and HCN channels, we carried out immunolocalization of these channels in mPFC (Fig. 1B) of 2-month-old mice and in primary cortical cultures on DIV 14 obtained from the frontal cortex of E16-17 mouse embryos (Fig. 1C). As described previously, we found that HCN1 immunoreactivity colocalized with Slack in both mPFC neurons of mouse brain ( Fig. 1B) and cultured cortical neurons (Fig. 1C). Notably, the colocalization of HCN1 and Slack channels was observed at the dendritic spine membranes of cultured neurons (Fig. 1C), which suggests that Slack channels may complex with HCN1 channels at dendritic spines of mouse mPFC neurons.
The immunolabeling experiments were also performed in Slack −/− /Slick −/− mice [30]. Robust expression of Slack-B channels was detected in pyramidal neurons in layers II-III of wild-type mouse mPFC (Fig. 1S).
Slack and HCN1 Channels Are Co-localized within Dendritic Spines in Rat mPFC Layer II/III We further conducted post-embedding immunoEM to determine the subcellular localization of HCN1 and Slack (Slo2.2, KCNT1) channels in rat prelimbic mPFC layer II/III, the cortical sublayers most associated with working memory microcircuits in primates. We used high-resolution, non-diffusible gold immunoprobes to precisely localize HCN1 and Slack channels on dendritic spine membranes, and determine their position in relationship to synaptic specializations. Immunoreactivity for HCN1 in rat mPFC layer II/III was observed in perisynaptic compartments near axospinous glutamatergic-like asymmetric synapses ( Fig. 2A). An identical con guration has been previously observed in primate dorsolateral PFC layer III circuits, where HCN1 channels are visualized on the perisynaptic annulus of asymmetric synapses within dendritic spines, embedded at the edge of the postsynaptic density (PSD) [31]. Similarly, immunoreactivity for Slack channels was observed along the plasmalemma within dendritic spines, immediately next to asymmetric, presumed excitatory synapses in rat mPFC layer II/III ( Fig. 2B). In order to determine the spatial interactions of these channels and whether they are localized within the same dendritic spine subcompartment, we conducted dual-labeling immunogold immunoEM.
Both HCN1 and Slack channels were co-localized at dendritic spine head membranes, perisynaptically next to glutamatergic synapses (Fig. 2C). These ndings suggest that HCN1 and Slack potassium channels are strategically positioned on dendritic spines to gate excitatory synapses in mPFC layer II/III circuits.

Co-expression of Slack with HCN Channels in HEK Cells Reverses HCN-induced Depolarization
To test whether the coupling of Slack-B to HCN channels reverses the normal physiological effects of HCN channel activation, we next carried out experiments using a previously described, synthetic excitable cell type [32]. These HEK cells, termed Kuhl-H cells, express HCN channels, which are activated by increases in cAMP levels, as well as a light-sensitive adenylyl cyclase enzyme (bPAC). Thus, exposure of the cells to blue light triggers opening of the HCN channels, resulting in a depolarization of the cells that produces an increase in cytoplasmic Ca 2+ levels. The cells additionally express a Ca 2+ sensor (R-GECO1), allowing monitoring of the effects of HCN channel activation in live cells in real time. The functional components within these cells are illustrated in Fig. 3A.
As described previously, in the absence of Slack channels, application of brief pulses of blue light to these cells elevates cAMP and activates the HCN channels to depolarize the cells, resulting in a slow increase in Ca 2+ reporter uorescence ( Fig. 3B) [32]. Elavations of cAMP have been found to have no effect on Slack-B channels expressed in heterologous systems [33]. When, however, Slack-B was coexpressed with the HCN channels in the Kuhl-H cells, the response to the pulses of blue light was substantially reduced or abolished (Fig. 3B). These ndings are consistent with the hypothesis that the coupling of HCN channels to Slack-B reverses the depolarization normally produced by HCN activation.

Inhibition of HCN Channels Reduces K Na Currents in mPFC Pyramidal Neurons
To test further the hypothesis that HCN channel activation leads to the activation of K Na potassium channels, we carried out whole cell voltage clamp recordings on primary cortical neurons and on mPFC pyramidal neurons in cortical slices, and tested the effect of the HCN channel blocker ZD7288 on K Na currents induced by voltage steps. We found that treatment with ZD7288 10 µM produced a marked reduction in outward K + current in primary cortical neurons ( Fig. 4A-4C) as well as in mPFC pyramidal neurons of intact mice ( Fig. 4D-4F). However, the ZD7288 10 µM did not produce any reduction in K + current in the primary cortical neurons when blocking Slack channels with a novel Slack blocker, SLK-01 ( Fig. 4G and 4H, see next section).
The suppression of outward current by ZD7288 in neurons could in theory results from an off-target effect of this compound on K Na channels. To test this hypothesis, we carried out whole cell voltage recordings on HEK cells stably expressing the Slack channels. We found that addition of the HCN inhibitor, ZD7288 (10 µM) to the external medium had no effect on Slack currents in this cell line ( Fig. 4I-4K). Collectively, these results suggest that the reduction in K + current in mPFC pyramidal neurons by ZD7288 is mediated by its inhibition of HCN channels.

SLK-01 Is a Use-dependent Blocker of Slack Channels
Known pharmacological agents that have been used previously to inhibit Slack channels, such as quinidine [34], have signi cant non-speci c effects on additional ion channels, including other Kchannels. In an attempt to further probe the interaction of HCN channels with Slack in intact animals, we have now characterized a novel use-dependent Slack blocker. The structure of this compound, SLK-01 (originally given the identi er BMS-271723, see Discussion) is shown in Fig. 5A. In the present study, currents were recorded from HEK cells stably expressing the Slack channel using the whole cell patch clamp technique. SLK-01 was applied to these cells at concentrations of 50 nM to 40 µM (n = 7, Fig. 5B and 5C). At concentrations above 1 µM there was a concentration-dependent reduction in current amplitude accompanied by a change in kinetic behavior. On depolarization of untreated cells, Slack currents activate monotonically to a nal current level that persists throughout the duration of the depolarizing command. In contrast, in the SLK-01 treated cells, the currents in response to command potentials greater than + 20 mV rose to a peak within 50 msec followed by a rapid decline to a lower level (Fig. 5B). The amplitude of the peak current in the presence of SLK-01 to the maximal current in untreated cells was reduced with an IC 50 of ~ 20 µM (Fig. 5C). The voltage dependence of Slack currents was, however, not altered by SLK-01 (Fig. 5D). The carrier solution had no effect on Slack currents.
The decline in Slack currents following their activation in the presence of SLK-01 suggest that block of the channels by this agent is use-dependent, such that block only occurs after channels have been opened in response to depolarization. If this is the case, repeated depolarization at a high rate would be expected to promote the rate at which channels become blocked. To test this, we applied trains of 20 or 100 depolarizing commands from holding potential of -80 mV to 40 mV for 220 ms at different rates and measured the degree of block at the end of each pulse. The percent inhibition for each pulse was normalized to that for the rst pulse in each series. In untreated cells, repeated stimulation at 1or 2 Hz produced no change in current amplitude throughout the train, and stimulation at 5 Hz produced an only a small reduction in current amplitude after 20 pulses (~ 20%, Fig. 5E). After treatment with 40 µM SLK-01, the amplitude of Slack current measured at the end of each pulse progressively decreased throughout the train even at 1 Hz (Fig. 5E), and at 5 Hz, currents were suppressed by ~ 70%, con rming the usedependence of block by this agent.
As a partial test for speci city, we measured the effects of SLK-01 on BK potassium currents. BK largeconductance calcium-activated channels (KCNMA1, K Ca 1.1, Slo1) are the channels most closely related to Slack, other than its paralog Slick (KCNT2, K Na 1.2) [21,22]. Application of either 20 µM or 40 µM SLK-01 to cells stably expressing BK channels had no statistically signi cant effect on the amplitude of BK currents (Fig. 5F, G).

SLK-01 Infusion into PFC Improves Delayed Alternation Performance, Similar to HCN Channel Blockade
Previous work has demonstrated that infusion of a very low dose of the HCN channel blocker, ZD7288, into rat PFC improved delayed alternation performance of rats [11], the classic test for assessing working memory in rodents that is dependent on the integrity of the mPFC. These behavioral data were consistent with physiological data from monkeys, where low, but not high, doses of ZD7288 enhanced delay-related neuronal ring [11]. In the delayed alternation task in a T maze, the rat must remember which arm it had chosen on the previous trial, and alternate its response to the other side to receive reward (Fig. 6A). This requires both spatial working memory and response inhibition to overcome the tendency to go to a previously rewarded location, both functions of the PFC. If a major mode of action of HCN channels on mPFC pyramidal neurons is to activate Slack channels, the inhibition of Slack channels would be expected to have a similar effect to that of HCN blockade. To test this possibility, rats (n = 5) were implanted with indwelling cannula above the prelimbic PFC, and were infused with vehicle or SLK-01 (20 µM or 50 µM in 0.5µl/side). While there were no effects seen with the 20µM dose, 3 of the 5 rats were improved by the 50µM dose, and 2 were impaired. As blockade of ion channels often has an inverted U dose-response, we tested an intermediate dose, 30µM, in the two rats that were impaired by the 50µM dose, and found that they were improved compared to vehicle control (Fig. 6B). Thus, a dose of either 30 µM or 50µM SLK-01 was able to signi cantly improve performance compared to vehicle control (best dose vs. vehicle: p < 0.009). As Slack channel blockade mimics the effects of HCN channel blockade, these behavioral data are consistent with the hypothesis that the activation of HCN channels is coupled to Slack channel opening.

Discussion
We have found that Slack-B channels form a complex with HCN channels in pyramidal cells of the rodents mPFC. A clear implication of this nding is that Na + entry through HCN channels serves to activate the Na + -dependent K + current of the Slack channel. Because the reversal potential of the nonselective HCN cation channels is around − 20 mV, activation of HCN channels alone would be expected to depolarize neurons under physiological conditions, leading to increased excitability. Coupling of HCN to Slack K Na currents, however, would be expected to suppress the depolarization and result in a hyperpolarization and/or a decrease in input resistance leading to decreased membrane excitability.
Consistent with this hypothesis, we found that blockade of HCN channels with ZD7288 reduces K + current in pyramidal cells, but has no effect on K + currents in a cell line stably expressing Slack alone. Finally, infusion of the Slack channel blocker, SLK-01, into rat mPFC improved working memory performance, mimicking the bene cial effects of the HCN channel blocker, suggesting the two agents may act through a common pathway.
Our ndings may resolve the paradox that activation of HCN channels in PFC leads to a loss of taskrelated ring [11] while low dose blockade of HCN channels enhances task-related ring [11,35]. The ability to carry out spatial working memory tasks requires persistent ring of pyramidal cells arising from the recurrent excitation of layer III microcircuits that make synapses on spines [2]. These layer III spines express HCN channels [11,17], and as shown in the current study, they can be co-expressed with Slack channels on dendritic spines in rat mPFC. Thus, increased cAMP-induced opening of HCN channels in spines would lead to Slack channel opening, spine hyperpolarization, and the loss of recurrent excitation needed to sustain working-memory related neuronal ring. This is in contrast to the more classic HCN channel actions on distal dendrites where they are distant from synapses, and depolarize the membrane when the neuron becomes hyperpolarized, thus having an important excitatory in uence [9]. Layer V cortical pyramidal cells often express high levels of HCN channels on their distal dendrites, including in rodent mPFC [14][15][16], indicating circuit selective actions of these channels dependent on their location and molecular interactions.
A required role of Slack channels in the control of cortical functions such as working memory is expected based on the known effects of Slack mutations. Within the cerebral cortex, Slack-B, a major splice isoform is expressed selectively in the frontal cortex [24]. Human mutations that alter the function of Slack channels result in several distinct childhood epilepsies, all of which are associated with very severe intellectual impairment [25]. When expressed in heterologous systems, the majority of disease-causing mutations give rise to Slack currents that are increased in amplitude from 3-  [47][48][49][50][51][52]. Slack mutations have also been documented for autism [53].
As part of this study, we characterized a novel compound, SLK-01, that is an effective use-dependent inhibitor of Slack channels. This compound was originally synthesized as part of a program targeting KCNQ2/3 and BK (Slo1) channels at the Bristol-Myers Squibb (BMS) Pharmaceutical Research Institute in the Department of CNS Drug Discovery in the late 1990s/early 2000s and given the identi er BMS-271723. It was initially screened against both the primary target and select secondary K channels. During the course of this screening it was discovered that some of these lead compounds did not signi cantly activate KCNQ or BK channels, but possessed inhibitory activity at Slack channels. The modulatory action of ~ 10 of these compounds was determined using heterologous expression of Slack in Xenopus oocytes and was presented in a discussion of off-target activity discovered in compounds originally synthesized as KCNQ2/3 and BK channel modulators [54].
Our ndings indicate that SLK-01 is an open-channel blocker, suggesting that it could be more potent on human disease-relevant mutant gain-of-function Slack channels than on wild-type channels [37,26,25,55,28]. Until very recently, the only other effective compounds known to block Slack channels was quinidine, an agent that also acts on a wide variety of other channels [34]. In the past year, there have been reports of other more selective agents that block Slack channels [56][57][58][59]. Further studies will be required to determine the speci city, e cacy and species selectivity of these compounds, as well as that of SLK-01.
In summary, our studies strongly suggest that a complex containing both Slack and HNC channel plays a key regulatory role in the intrinsic excitability of medial prefrontal pyramidal cells of the cerebral cortex and in the response of their dendrites to incoming stimuli that are required. Further genetic studies will be required to establish the speci c function of this two-channel complex in the regulation of working memory performance.

Declarations Data Availability
This article contains all of the data that were generated/analyzed during this investigation.

Post-embedding Immunogold Immunohistochemistry
In order to visualize the subcellular localization and interaction between HCN1 and Slack (Slo2.2, KCNT1) channels in rat mPFC layer II/III circuits, we utilized post-embedding immunogold dual-labeling techniques. Brie y, frozen brain tissue was freeze substituted in a Leica EM AFS2 unit starting from -90°C, and gradually increasing temperature to -45°C. After 3 changes of acetone, tissue was in ltrated with Lowicryl HM20 resin. The polymerization was carried out at -20°C with ultraviolet. Hardened blocks were cut using a Leica UltraCut UC7. Sixty nanometer sections were collected on formvar/carbon-coated nickel grids and stained using 2% uranyl acetate and lead citrate. For immunolabeling of resin sections, grids were placed section side down on drops of 1% hydrogen peroxide for 5 min, rinsed, and blocked for nonspeci c binding with 3% bovine serum albumin in Tris-buffered saline (TBS) containing 1% Triton X-100 for 30 min. For single immunogold labeling, grids were incubated with anti-HCN1 or anti-Slack For Co-IP experiments, frontal cortex lysates were incubated with 5 μg anti-Slack IgY (AvesLabs) antibody [24] or IgY control (AvesLabs) overnight at 4°C. 100 μL Anti-IgY PrecipHen beads (AvesLabs) was added to sample and allowed to incubate for 2 hours, followed by wash and collection of beads. Beads were transferred to 2x Laemelli-Buffer with 5% beta-mercaptoethanol and incubated at room temperature for 30 minutes prior to Western blotting [43].
For all immunoblotting experiments, protein samples were electrophoretically separated on an SDS-PAGE gel (4%-15% gradient gel, Bio-Rad) and transferred onto PVDF membranes (0.2 μm pores, Bio-Rad, USA). Blots were blocked in 5% nonfat milk in Tris-buffered saline and Tween 20 (TBST) for 1 h at room temperature and probed with the primary antibodies to Slack (1:3000, AvesLabs) and HCN1 (1:1000, Abcam) overnight at 4°C. After overnight incubation, the blots were washed three times in TBST for 30min, followed by incubation with corresponding horseradish peroxidase-conjugated secondary antibodies (1:1000; Abcam) at room temperature for 1 h. Protein bands were visualized via enhanced chemiluminescence and quanti ed with analyzed with ImageJ (NIH) software.
For experiments, the cells were plated in 35mm plastic plates at 2ml per well (~3x10 4 cell/mL). To introduce components into the cells using baculovirus, HCN2 and the photoactivated cyclase, bPAC were packaged in BacMam (Montana Molecular, Bozeman, MT). The virus titers were bPAC (2 x 10 10 VG/mL) and HCN2 (1.57 x 10 10 VG/mL). The additional Ca 2+ biosensor in baculovirus was obtained from Montana Molecular: R-GECO (2 x 10 10 VG/mL). The cells were transduced following the manufacturer's recommended protocol, plated in 35 mm plastic plates and transduced with the appropriate mix of viruses and HDAC inhibitor sodium butyrate (2mM nal concentration). Two days later, the media was exchanged with PBS before imaging. Cells were imaged using post brief illumination from a blue LED (488nm) for rapid stimulation of bPAC. The yellow illumination was provided with a LED (560nm).
For data analysis, image data was stored in a Z-stack tiff le and loaded into the FIJI distribution of the IMAGEJ software. The cells were selected using a freehand ROI surrounding the cell of interest. The average total intensity pixel value within the ROI for each frame was collected using the time series analyzer plugin and saved as a .csv le. The raw uorescence data was then put into PRISM and ORIGIN for further analysis. The uorescence traces were rescaled from 0 -100% and normalized to zero using PRISM software.

Patch-clamp Recordings
Whole-cell patch-clamp recordings were performed with patch-clamp ampli ers (MultiClamp 700B; Molecular Devices) under the control of pClamp 11 software (Molecular Devices). Data were recorded with a sampling rate at 20 kHz and ltered at 6 kHz. Rs compensation of 70% was used. Primary cortical neurons at DIV 13-14 or mPFC pyramidal neurons in brain slices from 2-month-old mice or HEK cells were recorded at physiological temperature (37°C). Recording electrodes were pulled from lamented borosilicate glass pipettes (Sutter Instrument, CA), and had tip resistances between 4 and 6 MΩ when

Delayed Alternation Task
Rats were trained on the delayed alternation test of spatial working memory in a T-shaped maze. They were rst adapted to handling and to eating treats (highly palatable miniature chocolate chips) on the maze prior to cognitive training. In this task, the rat was placed in the start box at the bottom of the 'T'. When the gate was lifted, the rat proceeded down the stem of the maze to the choice point. On the rst trial, rats were rewarded for entering either arm, but for each subsequent trial, were rewarded only if they chose the arm that they had not visited in the previous trial. Between trials, they were picked up and returned to the start box for a prescribed delay period. The choice point was cleaned with alcohol between each trial to remove olfactory clues (scent trails) often used by rodents to mark their previous locations. Successful performance of this task requires many operations carried out by the PFC: the rats must update and maintain the spatial information over the delay period for each trial, resist the distraction of being picked up and carried to the start box, and use response inhibition to overcome the tendency to repeat a rewarded action. The delay period was adjusted for each rat, such that they were performing at a stable baseline of 60-80% before drug treatment, thus leaving room for either impairment or improvement in performance. Rats were tested by experimenters who were highly familiar with the baseline behaviors of each individual animal, but blind to drug treatment conditions. Rats were observed for any potential differences in normative behavior (e.g. grooming, distracted sni ng), physical appearance, or physiological functioning (e.g. defecation/urination).

Surgery and Drug Infusions into mPFC
Following training on the delayed alternation task, rats underwent aseptic stereotaxic surgery, under ketamine+xylazine anesthesia with metacam analgesic pretreatment, to implant cannula aimed just above the prelimbic mPFC (AP: +3.2 mm; ML: ±0.75 mm; DV: -4.2 mm). Details of the surgical procedure and post-surgical treatment can be found in [12]. After rats had fully recovered from the surgery, they were adapted to the infusion procedure to minimize stress. Once stable baseline performance was reestablished, rats received intra-PFC infusions of vehicle vs. drug (0.5μl per side) over 5 minutes; performance was assessed 10 min after the infusion by a researcher unaware of the drug treatment conditions. There was at least a one-week washout between drug infusions.

Statistical Analysis
GraphPad Prism Version 8 was used for statistical analysis. Two-tailed Student's t test (parametric) or unpaired two-tailed Kolmogorov-Smirnov-test (non-parametric) was used for single comparisons between two groups. Other data were analyzed using two-way ANOVA with Bonferroni correction (parametric). All data were expressed as mean ± SEM, with statistical signi cance determined at p values < 0.05. In details, *Indicates p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001 in all gures.  Co-localization of HCN1 and Slack channels within dendritic spines in rat mPFC layer II/III. A. Immunogold labeling (15 nm particles) reveals precise labeling of HCN1 within dendritic spines of pyramidal cells in rat prelimbic mPFC layer II/III. HCN1 channels are concentrated in the perisynaptic compartment adjacent to asymmetric, presumed glutamatergic-like synapses. B. Similar to HCN1 channel labeling, immunogold labeling (15 nm particle) for Slack channels is enriched along the plasma membranes bordering asymmetric axospinous synapses, receiving presumed glutamatergic input, within dendritic spines in rat mPFC layer II/III. C.Co-localization of HCN1 (15 nm particle) and Slack (5 nm particle) channels within dendritic spines in rat mPFC layer II/III. In dendritic spine heads, the immunoparticles for HCN1 and Slack overlap at perisynaptic locations. Synapses are between arrows.
Color-coded arrowheads point to HCN1(red) and Slack (blue) immunoreactivity. Pro les are pseudocolored for clarity. Ax, axon; Sp, dendritic spine; Mit, mitochondria. Scale bars, 200 nm. that is cAMP coupled), and R-GECO1 (a Ca 2+ sensor  Blocking HCN channels with ZD7288 reduces outward K + currents in mPFC pyramidal neurons. A. Representative current traces from whole cell voltage clamp recordings in cultured frontal cortical neurons on DIV14 induced by voltage steps from -90 to +50 mV in 10 mV increments before and after application of HCN channel blocker ZD7288 (10 μM, 10 min). B. Summary data show the outward K + currents for each voltage step in neurons before and after application of ZD7288. Data are shown as mean ± SEM (n = 12-14, two-way ANOVA). C. Maximal K + current at +50 mV for each condition. Data are shown as mean ± SEM (n = 12-14, Student's t test). D and E. Representative current traces and summary data from whole cell voltage clamp recordings in mPFC pyramidal neurons of 2-month-old mice induced by voltage steps from -120 to +200 mV in 20 mV increments before and after application of ZD7288 (10 μM, 10 min). Data are shown as mean ± SEM (n = 6, two-way ANOVA). F. Maximal K + current at +200 mV for each condition. Data are shown as mean ± SEM (n = 6, Student's t test). G and H. Representative current traces and summary data from whole cell voltage clamp recordings in cultured frontal cortical neurons on DIV14 induced by voltage steps from -90 to +50 mV in 10 mV increments before and after application of Slack channel blocker SLK-01 (10 μM, 10 min) or SLK-01 with ZD7288 (10 μM, 10 min). Data are shown as mean ± SEM (n =7-12, two-way ANOVA). I and J. Representative current traces and summary data from whole cell voltage clamp recordings in HEK cells stably expressing Slack induced by voltage steps from -120 to +80 mV in 20 mV increments before and after application of ZD7288 (10 μM, 10 min). Data are shown as mean ± SEM (n = 4, two-way ANOVA). K. Maximal K + current at +80 mV for each condition. Data are shown as mean ± SEM (n = 4, Student's t test). *p < 0.05, **p < 0.01, ***p < 0.001 and ****p < 0.0001. potential of -80 mV to +40 mV at rates of 1, 2 or 5 Hz. Plots show the progressive change in current amplitude recorded at the end of each pulse in the absence (n = 4) or presence of 40 mM SLK-01 (n = 3).
F. SLK-01 has no effect on BK currents. Representative traces of currents in CHO cells stably expressing rat BK channels before and after application of 40 mM SLK-01. Currents were evoked by voltage steps between -100 and +60 mV from a holding potential of -80 mV. G. Group data quanti ying the mean amplitude of currents recorded at +60 mV before and after SLK-01 (40 mM n = 6).

Figure 6
Blockade of Slack channels with infusion of SLK-01 into rat mPFC improved delayed alternation performance in a T maze. A. Diagram of the delayed alternation task in a T maze for testing spatial working memory performance in the rat. On the rst trial, the rat can choose either arm and be rewarded, on subsequent trials the rat must alternate its response. The rat spends the delay period in the start box, and thus must remember its previous spatial response over this delay. B. Infusion of SLK-01 (20-50 μM) produced a dose-related improvement in working memory performance, where a best dose of either 30 or 50 μM signi cantly improved performance compared to vehicle control (**p < 0.01, Student t-test).

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.