Dopaminergic tone persistently regulates voltage-gated ion current densities through the D1R-PKA axis, RNA polymerase II transcription, RNAi, mTORC1, and translation

Long-term intrinsic and synaptic plasticity must be coordinated to ensure stability and flexibility in neuronal circuits. Coordination might be achieved through shared transduction components. Dopamine (DA) is a well-established participant in many forms of long-term synaptic plasticity. Recent work indicates that DA is also involved in both activity-dependent and -independent forms of long-term intrinsic plasticity. We previously examined DA-enabled long-term intrinsic plasticity in a single identified neuron. The lateral pyloric (LP) neuron is a component of the pyloric network in the crustacean stomatogastric nervous system (STNS). LP expresses type 1 DA receptors (D1Rs). A 1 h bath application of 5 nM DA followed by washout produced a significant increase in the maximal conductance (Gmax) of the LP transient potassium current (IA) that peaked ~4 h after the start of DA application; furthermore, if a change in neuronal activity accompanied the DA application, then a persistent increase in the LP hyperpolarization activated current (Ih) was also observed. Here, we repeated these experiments with pharmacological and peptide inhibitors to determine the cellular processes and signaling proteins involved. We discovered that the persistent, DA-induced activity-independent (IA) and activity-dependent (Ih) changes in ionic conductances depended upon many of the same elements that enable long-term synaptic plasticity, including: the D1R-protein kinase A (PKA) axis, RNA polymerase II transcription, RNA interference (RNAi), and mechanistic target of rapamycin (mTOR)-dependent translation. We interpret the data to mean that increasing the tonic DA concentration enhances expression of a microRNA(s) (miRs), resulting in increased cap-dependent translation of an unidentified protein(s).


INTRODUCTION
Dopaminergic systems use volume transmission to modulate cognitive and motor functions (Zoli et al., 1998;Schultz, 2007;Oginsky et al., 2010). Tonic and burst firing neurons release Dopamine (DA) that can then diffuse and act predominantly at remote extra-synaptic receptors before reuptake by DA transporters. As a result, target neurons are tonically exposed to DA; e.g., approximately tens of nM in the striatum and prefrontal cortex (Owesson-White et al., 2012;Nirogi et al., 2013;Zuo et al., 2013), and superimposed upon this baseline are periodic fluctuations in DA that can transiently rise to ∼µM levels near the release sites of bursting DA neurons (Park et al., 2011;Rice et al., 2011;Owesson-White et al., 2012).
Phasic and tonic DA have distinct roles in the CNS. Phasic DA may encode reward prediction error (Steinberg et al., 2013), provide sustained motivational drive (Howe et al., 2013) and modulate motor behaviors (Gerfen and Surmeier, 2011). On the other hand, tonic DA is thought to have an enabling function because tonic administration of drugs, such as L-dopa or neuroleptics, can enable motor, motivational and cognitive behaviors (Schultz, 2007). The effects of tonic DA have largely been attributed to D2Rs, but all receptors can show high and low affinity states and there is increasing evidence that tonic DA acting at high affinity type 1 DA receptors (D1Rs) may also enable and shape circuit output over the long-term (Trantham-Davidson et al., 2004;Rodgers et al., 2011a,b;Wall et al., 2011;Saba et al., 2012).
We previously showed that the sole lateral pyloric (LP) neuron in the stomatogastric nervous system (STNS) of the spiny lobster, Panulirus interruptus, expressed high and low affinity D1Rs but not D2Rs (Zhang et al., 2010;Rodgers et al., 2011a,b;Krenz et al., 2013). Low affinity LP D1Rs were activated by µM DA to produce immediate and reversible alterations in the biophysical properties of LP voltage gated ionic currents (Harris-Warrick et al., 1995;Johnson et al., 2003;Kloppenburg et al., 2007;Zhang et al., 2010). High affinity LP D1Rs activated by nM DA produced effects over two time scales. They rapidly conferred activity-dependence upon LP I h to maintain a conductance ratio and its activity correlate , and they also acted through a slower process(es) to persistently influence ion current densities. A 1 h application of 5 nM DA or saline (control) to the superfusate bathing LP, followed by a 4 h washout and subsequent voltage clamp to measure LP I A showed that LP I A G max was significantly increased by 25% in the DA-treated relative to control preparations (Rodgers et al., 2011b). If the experiment was repeated, but LP activity was altered during the 1 h 5 nM DA (or saline) application, then LP I h was also significantly increased by 55% in DA-treated preparations relative to saline controls (Rodgers et al., 2011a). Here we examine the cellular processes mediating DA' s persistent effects and show that many of the same elements involved in long-term synaptic plasticity underpin DA-induced long-term intrinsic plasticity.

ANIMALS
California spiny lobsters, Panulirus interruptus, were purchased from Marinus Scientific (Long Beach, CA) and Catalina Offshore Products (San Diego, CA). Lobsters were maintained at 16 • C in aerated and filtered seawater. Animals were anesthetized on ice before dissection.

EXPERIMENTAL PREPARATION
The STNS was dissected and pinned in a Sylgard lined Petri dish using standard techniques (Selverston et al., 1976). The stomatogastric ganglion (STG) was desheathed and isolated with a Vaseline well. The STG was superfused with saline consisting of (in mM) 479 NaCl, 12.8 KCl, 13.7 CaCl 2 , 39 Na 2 SO 4 , 10 MgSO 4 , 2 glucose, 4.99 HEPES, 5 TES at pH 7.4. Intracellular somatic recordings used to identify neurons were obtained with sharp high resistance glass microelectrodes filled with 3 M KCl (20-30 MΩ) and an Axoclamp 2B amplifier (Axon Instruments, Foster City, CA). Neurons were identified by correlating action potentials from somatic intracellular recordings with extracellularly recorded action potentials on identified motor nerves, and by their characteristic shape and timing of oscillations. The process of dissection and cell identification usually took 3-5 h.

SOMATIC TWO-ELECTRODE VOLTAGE CLAMP (TEVC)
For two-electrode voltage clamp (TEVC) of LP I h , the well surrounding the STG was superfused for 1 h with blocking saline: saline containing 10 −6 M picrotoxin to block inhibitory glutamatergic synaptic inputs (Marder and Eisen, 1984;Cleland and Selverston, 1995), 10 −7 M TTX to block voltage-gated Na + channels, 2 × 10 −2 M tetraethylammonium (TEA) to block voltagegated K + channels, 2 × 10 −4 M cadmium chloride (CdCl 2 ) to block Ca 2+ -and Ca 2+ -dependent channels. The LP neuron was next impaled with two low resistance voltage clamp micropipettes (8-10 MΩ when filled with 3 M KCl) connected to Axoclamp 2B or 900A amplifiers (Molecular Devices, Foster City, CA). LP was clamped to a −50 mV holding potential using pClamp software. I h was elicited using a series of 4 s hyperpolarizing voltage steps, from −60 mV to −120 mV in 10 mV increments with 6 s between steps. Steady state peak currents were measured by fitting the current trace back to the beginning of the hyperpolarizing voltage step or by subtracting the initial fast leak current from the slowly developing peak of I h at the end of each negative voltage step. Currents were converted to conductance (G = I peak /(V m -V rev ) and fitted to a first order Boltzmann equation. V rev I h = −35 mV (Kiehn and Harris-Warrick, 1992). For TEVC measurement of LP I A the command potential was stepped from −50 mV to −90 mV for 200 ms to remove resting inactivation. The deinactivating prepulse was immediately followed by a 400 ms testpulse to activate the channels. Activation pulses ranged from −40 to +40 mV in 10 mV increments. To subtract the leak current, the hyperpolarizing prepulse was omitted and instead the prepulse was set to −40 mV to remove I A activation from the −50 mV holding potential. Currents were converted to conductance (G = I peak /(V m -V rev ) and fitted to a first order Boltzmann equation. V rev I A = −86 mV (Eisen and Marder, 1982). TEVC experiments were done at 19-22 • C as measured with a probe in the bath. Temperature did not change by more than 1 • C during any given experiment.

CLONING AND SEQUENCING LOBSTER ARGONAUTE 1 (AGO1)
Total RNA was isolated from the lobster nervous system using TRIzol (Ambion, Austin, TX) and converted to cDNA using Superscript (Life Technologies, Grand Island, NY) according to manufacturers' instructions. Degenerate primers were generated based on alignments with Drosophila melanogaster (Genbank accession: AB035447), Penaeus monodon (Genbank accession: DQ343133), and Daphnia pulex (wfleabase: NCBI_GNO_68324) and are shown in Table 1. Degenerate polymerase chain reactions (PCRs) were performed with Advantage Taq (Clontech, Mountain View, CA) as previously described (Baro et al., 1994). PCR products were cloned with a TA cloning kit (Qiagen, Valencia, CA) using the manufacturer's instructions. The 3 end was obtained with lobster specific primers, S. For 1 ( Table 1) and a SMARTer RACE kit (Clontech) using instructions provided. The 5 end was obtained with lobster specific primer, S. Rev 2 ( Table 1) and a FirstChoice RLM RACE Kit (Ambion) using instructions provided. All sequencing was performed by the GSU DNA core facility. Sequences were analyzed and manipulated with the Lasergene 10 suite of DNASTAR software (Madison, WI).

PEPTIDE INJECTION
The his-tagged hook (HHHHHHPDNGTSAWGEPNESSPG-WGEMD) and mutant hook (HHHHHHPDNGTSvavEPNESSP-vavEMD) peptides were diluted in water to a working concentration of 10 ng/ml and fast green was added to 0.04% to visualize injections. Microloaders (Eppendorf) were used to directly fill glass pipettes (8-15 MΩ when filled with 3 M KCl) with the solution (i.e., no backfilling). Because of the high resistance of the peptide solution, pipette tips were broken before injection by gently touching a Kim wipe. The peptide was pressure injected into LP neurons using a Picospritzer III (General Valve/Parker Hannifin). Only two pressure pulses (on average 32 psi and 47 ms) separated by 30 s were applied. Intracellular recording during the injection showed that the injection procedure had no effect on LP voltage envelope and firing properties. Extracellular recordings were used to continuously monitor the activity of the LP neuron before, during and for 1 h after peptide injection.

STATISTICAL ANALYSES
The data were checked for normality and analyzed using parametric statistical tests including Student t-tests and ANOVAs. In the one case where data were not normally distributed, a nonparametric Kruskal-Wallis test was used. All data were analyzed using Prism Statistical software package (Graphpad). Significance threshold was set at p < 0.05 in all cases. Statistical outliers were excluded if the values fell greater than two standard deviations from the mean and this resulted in exclusion of one experiment. Means and standard errors are presented unless otherwise noted. ANOVAs were usually followed by Tukey's post hoc tests that make all pairwise comparisons.

EXPERIMENTAL MODEL
A persistent activity-dependent increase in LP I h G max was elicited by two coincident events: an activation of high affinity LP D1Rs and a reduction in LP burst duration (Rodgers et al., 2011a). We used a simple experimental model to coincidently elicit these events and study the cellular processes involved in long-term intrinsic plasticity: the spiny lobster STNS was dissected and pinned in a dish ( Figure 1A). The STG, which contains the LP neuron, was continuously superfused with saline. Intracellular and extracellular recordings were used to identify the sole LP neuron as described in Section Materials and Methods. Both in vivo (Heinzel et al., 1993) and in situ (Figure 1B), the LP neuron undergoes spontaneous slow oscillations in membrane potential (∼20 mV at 1-2 Hz) with a burst of spikes riding on the depolarized plateau of each oscillation. The standard experimental protocol used to elicit the persistent increase in LP I h G max is diagrammed in Figure 1C. LP activity was altered during a 1 h application of DA followed by washout of DA. At the end of the  wash, the preparation was superfused with blocking saline for 1 h to prevent spontaneous activity, and LP I h was then measured with somatic TEVC ( Figure 1D). We previously demonstrated that in the absence of DA, LP I h G max does not exhibit rapid activity-dependent changes ; and, measures of LP I h before and after the block indicate that it does not change appreciably during the block (LP I h G max before block = 0.125 + 0.013 µS; LP I h G max after 1 h block = 0.120 + 0.012 µS, n = 7, Student t-test, p = 0.796). Three methods were previously used to elicit a persistent ∼55% increase in LP I h G max by simultaneously activating high affinity D1Rs while altering LP activity (Rodgers et al., 2011a). The first two methods used a 1 h application of 5 nM DA to activate high affinity D1Rs and either concurrent application of TTX to block activity or concurrent injection of a hyperpolarizing bias current into LP to reduce LP burst duration and decrease LP duty cycle (burst duration/period). These treatments were followed by a 2.5 h saline wash, a 1 h block and TEVC measurement of LP I h . The fact that both methods produced the same persistent ∼55% increase in LP I h G max suggested that the specific change in activity did not determine the magnitude of the alteration in LP I h G max measured 3.5 h after the treatment ended (other time points were not examined). In the absence of a change in activity, 5 nM DA had no effect; and, TTX had no significant effect in the absence of 5 nM DA. The third method used to elicit a persistent ∼55% increase in LP I h G max was a 1 h application of 5 µM DA alone, which activates both high affinity D1Rs to permit activity-dependent regulation of LP I h G max and low affinity D1Rs to decrease LP burst duration and reduce LP duty cycle ( Figure 1B, compare top and bottom panels). When TTX was included with 5 µM DA, the same 55% increase in LP I h G max was observed, again suggesting that the magnitude of the persistent change in LP I h G max measured 3.5 h after the treatment was not strictly correlated with the magnitude of the change in activity. However, a change in activity was required because, if a depolarizing bias current was injected into LP to prevent the 5 µM DA-induced decrease in LP burst duration and duty cycle, then there was no persistent change in LP I h G max in the presence of 5 µM DA. The most parsimonious interpretation of these data is that 5 µM DA and 5 nM DA + TTX acted through the same pathway to produce a persistent ∼55% increase in LP I h G max . Therefore, these two treatments are used interchangeably to study the processes involved.

TIME COURSE OF THE PERSISTENT INCREASE IN LATERAL PYLORIC (LP) I h G max
Previous experiments showed that a 1 h DA application accompanied by a change in activity produced a 55% increase in LP I h G max measured after a 2.5 h DA washout followed by a 1 h block (Rodgers et al., 2011a). To gain insight into the mechanism involved, we examined the time course of the increase. The experiments are diagrammed in Figure 2A. For the DA-treated group, the STG was superfused with 5 µM DA for 1 h followed by washout with saline for 0-6 h. At the end of the washout, the STG was treated with blocking saline for 1 h followed by TEVC to measure LP I h . Control experiments were performed in which the STG was superfused with saline for 0 h (acute) or 3.5 h (control) followed by a 1 h block and TEVC to measure LP I h . The measured LP I h G max for each experiment was divided by the mean LP I h G max value for control experiments, and the resulting normalized LP I h G max was plotted ( Figure 2B). The data indicated that the increase in LP I h G max developed slowly, peaked within 2-3 h of the start of DA application and then slowly declined over a similar time course. In the absence of 5 µM DA, LP I h G max did not change significantly over time (compare acute and control treatment groups).
In order to further demonstrate that the persistent activitydependent increase in LP I h was enabled by activation of high affinity D1Rs, and not washout of 5 µM DA, we performed one additional experiment (Figure 2A, orange bar). After dissection and cell identification, STGs were superfused with 5 nM DA + TTX for 3 h followed immediately by TEVC measures of LP I h . The data were normalized as described above and plotted ( Figure 2B, orange stars). The results indicated that the persistent increase in LP I h G max did not depend upon DA washout. The mean fold-changes in LP I h G max for the two 3 h treatment groups were 1.39 + 0.07 (3 h 5 nM DA + TTX) vs. 1.42 + 0.14 (1 h 5 µM DA + 1 h wash + 1 h block). These means were not significantly different from one another, but both were significantly increased relative to control. Since we previously showed that neither 5 nM DA nor a change in LP activity produced a significant longterm change in LP I h G max relative to saline controls on its own (Rodgers et al., 2011a), we interpret the data presented here to mean that tonic activation of high affinity D1Rs enables a slow cellular process(es) that permits activity-dependent regulation of LP I h G max .

THE TYPE 1 DA RECEPTOR (D1R)-PROTEIN KINASE A (PKA) AXIS IS REQUIRED FOR THE PERSISTENT INCREASE IN LATERAL PYLORIC (LP) I h G max
Experiments were next performed to determine if the persistent increase in LP I h G max was mediated by high affinity D1Rs acting through PKA (Figure 3). The experiment is diagrammed in Figure 3A: from t = −10-60 min, the STG was superfused with saline that in some cases contained TTX with or without a pharmacological reagent. In some experiments, 5 nM DA was added to the superfusate from t = 0-60 min. From t = 1 h-3.5 h, the STG was superfused with saline alone. The preparation was then blocked for 1 h and LP I h was measured with TEVC. Previous work showed that under these conditions, superfusing TTX alone from t = −10-60 min had no significant effect on LP I h G max relative to saline controls (Rodgers et al., 2011a). Flupenthixol antagonizes LP D1Rs (Zhang et al., 2010;Rodgers et al., 2011b) and in these experiments10 µM flupenthixol blocked the increase in LP I h G max elicited by 5 nM DA + TTX but had no effect on its own ( Figure 3B). Similarly, a competitive antagonist for cAMP binding to PKA, Rp-cAMPS, completely blocked the DAand activity-dependent persistent increase in LP I h G max , but had no effect in the absence of DA ( Figure 3C). These data are consistent with the idea that D1Rs act through PKA to persistently alter LP I h G max ; however, Rp-cAMPS can potentially antagonize other cAMP binding proteins including exchange protein activated by cAMP (epac) and hyperpolarization activated cyclic nucleotide-gated (HCN) channels (Shabb, 2011  For all treatment groups a single measure was obtained for each preparation using two-electrode voltage clamp (TEVC) at the end of the experiment, i.e., LP I h was not repeatedly measured over time within a given preparation; rather, terminal measurements from DA-treated preparations were compared to terminal measurements from control preparations and 68 animals were used for all of the experiments shown. Note that for the DA-treated group, the length of the saline wash varied across time points. (B) Plot of normalized LP I h G max for each experiment in every treatment group. Each symbol is a discrete experiment; e.g., the preparations in the 1 h DA-treatment group are different from the preparations in the 1.5 h DA-treatment group. Each y-value represents the LP I h G max for that experiment divided by the mean for the control experiments. The solid horizontal lines represent the means. Note that means will not be accurate at later time points where n ≤ 3, and they are only meant to show a decreasing trend over time. The numbers on the x-axis correspond to the hours that elapsed between the beginning of the DA application and the beginning of the block, i.e., x = 1 means that there was no saline wash before application of blocking saline; x = 2 indicates a 1 h saline wash, etc. Blue asterisks indicate significant differences relative to the control group as determined with a one-way ANOVA followed by Dunnett's post hoc tests that compared the control treatment group to the acute and no washout treatment groups and each time point in the DA-treated group except those time points with n ≤ 3: F (7,50) = 3.921, p = 0.0018.
involvement, the experiment was repeated with the specific membrane permeable PKA blocker, myristoylated PKI (14−22) , which specifically binds to and inactivates the catalytic subunit of PKA (Wen and Taylor, 1994;Shabb, 2011). PKI also blocked the DAand activity-dependent persistent increase in LP I h G max but had no effect in the absence of DA ( Figure 3D). Together these data suggested that a functional D1R-PKA axis was necessary for the persistent activity-dependent increase in LP I h G max .

MECHANISTIC TARGET OF RAPAMYCIN (mTOR)-DEPENDENT TRANSLATION IS REQUIRED FOR THE PERSISTENT INCREASE IN LATERAL PYLORIC (LP) I h G max
Mechanistic target of rapamycin (mTOR) is a conserved serine threonine kinase that functions as part of the protein complex, mTORC1, to regulate cap-dependent translation in all eukaryotic cells (Foster and Fingar, 2010). We used the mTORC1 specific blocker, rapamycin, and the translation blocker, anisomycin, to determine if mTORC1 and translation were also necessary for the DA-and activity-dependent increase in LP I h G max (Figure 4). In these experiments, from t = 0-60 min, the STG was superfused with saline that did or did not (control) contain 5 µM DA, followed by a 1 h wash with saline, a 1 h block and TEVC to measure LP I h . Either 100 nM rapamycin or 30 µM anisomycin was also superfused from t = −10-120 min. The data indicated that both mTOR and translation were necessary to produce the DA-and activitydependent persistent increase in LP I h G max . In the presence of either blocker, 5 µM DA could no longer elicit a significant , had no effect on its own, but prevented the DA-and activity-dependent persistent increase in LP I h G max . LP I h G max is plotted for every treatment group; each symbol represents one experiment, and the horizontal bars represent the means. Asterisks indicate significant differences as determined using a one-way ANOVA with Tukey's post hoc tests that made all pairwise comparisons: F (3,21) = 6.642, p = 0.0025. (C) The PKA inhibitor, Rp-cAMPS (1 mM), had no effect on its own, but prevented the DA-and activity-dependent persistent increase in LP I h G max . Asterisks indicate significant differences as determined using a one-way ANOVA with Tukey's multiple comparisons post hoc tests: F (3,24) = 5.9, p = 0.0036. (D) The PKA inhibitor, myristoylated PKI (14−22) (5 µM), prevented the DA-and activity-dependent persistent increase in LP I h G max , but had no effect on its own. Asterisks indicate significant differences as determined using a one-way ANOVA with Tukey's multiple comparisons post hoc tests: F (3,22) = 10.38, p = 0.0002.
increase in LP I h G max , but the blockers had no effect on their own.

THE RNAi PATHWAY IS REQUIRED FOR THE PERSISTENT INCREASE IN LATERAL PYLORIC (LP) I h G max
Activity-dependent intrinsic plasticity involving mTORC1 often requires additional regulatory elements that bind mRNA, including microRNA(s) (miRs) (Goldie and Cairns, 2012). The RNAi pathway processes miRs and mediates their actions (Finnegan and Pasquinelli, 2013). We next asked if a functional RNAi pathway was necessary for the persistent DA-and activity-dependent increase in LP I h G max . The experimental logic is diagrammed in Figure 5A. The RNA interference silencing complex (RISC) is an essential component of the RNAi pathway. RISC comprises several proteins including members of the Argonaute (Ago) and TNRC6/GW182 families. Dimerization occurs between members of the Ago and TNRC6 families, and disrupting this interaction prevents RISC formation and blocks the RNAi pathway and miR effects (Till et al., 2007). The minimal Ago binding domain from TNRC6 proteins has been identified as a continuous stretch of 22 amino acids, termed the Ago hook (Figure 5A, purple). An excess of the hook peptide can outcompete endogenous TNRC6 proteins for binding to endogenous Ago1 and 2 in human tissue culture cell lines (Till et al., 2007) (Figure 5A, panel ii). Altering amino acids in the Ago hook (termed mutant hook) prevented it from binding to Ago. Ago and TNRC6 proteins dimerized in the presence of an excess of the mutant hook ( Figure 5A, panel iii).
Ago is highly conserved across species, and the human Ago hook has been used successfully to disrupt the effects of a Drosophila miR in an in vitro translation assay and to isolate Drosophila Ago1 and yeast Ago in pull-down assays (Till et al., 2007). The Ago amino acids that are necessary to bind the Ago hook have been identified (Till et al., 2007), and are indicated in orange in Figures 5A, B. In order to determine if the amino acids involved Frontiers in Cellular Neuroscience www.frontiersin.org February 2014 | Volume 8 | Article 39 | 6

FIGURE 4 | The DA-and activity-dependent persistent increase in LP I h G max is mediated by an mTOR-dependent translational mechanism. (A) Diagram of the experimental protocol. (B)
The mTORC1 inhibitor, rapamycin (100 nM), prevented the increase in LP I h G max normally elicited by 5 µM DA but had no effect on its own. LP I h G max is plotted for each treatment group; each symbol represents one experiment; the horizontal bars represent the means. Asterisks indicate significant differences as determined using a one-way ANOVA with Tukey's post hoc tests that made all pairwise comparisons, F (3,26) = 5.015, p = 0.0071. (C) The translation inhibitor, anisomycin (30 µM), had no effect on its own but prevented the persistent increase in LP I h G max elicited by 5 µM DA. Asterisks indicate significant differences as determined using a one-way ANOVA with Tukey's multiple comparison post hoc tests, F (3,27) = 5.976, p = 0.0029.
in binding the Ago hook were conserved in lobster, we cloned lobster Ago1, which shares 83% identity with Drosophila Ago1, and compared it to each of the four human Ago proteins. These comparisons indicated that lobster Ago1 was ≥72% identical to each human Ago. An alignment of lobster and human Ago1 proteins indicated that they shared 74% identity over their entire length; and, 16 of the 17 amino acids known to be involved in binding the Ago hook were identical with the single amino acid change being conservative ( Figure 5B). Together, the existing data suggested that the previously validated human Ago hook and mutant hook peptides could be used in our experiments to test if a functional RNAi pathway was necessary for the persistent increase in LP I h G max . Experiments involving peptide injections into LP neurons are shown in Figure 5C. We pressure injected hook or mutant hook peptides into LP neurons as described in Section Materials and Methods. The STG was then superfused for 1 h to allow the injected peptide to compete with endogenous proteins for binding to LP Ago1. No change in rhythmic LP activity was observed during or after peptide injection. We next superfused the STG with 5 nM DA + TTX or TTX (control) for 1 h, followed by a 3 h wash with saline, a 1 h block and TEVC to measure LP I h . The data indicated that peptide injections had no effect on LP I h G max . In the absence of 5 nM DA, LP I h G max was not significantly different between uninjected neurons and neurons injected with Ago hook or mutant hook peptides (one-way ANOVA, F (2,23) = 0.3245, p = 0.7264). On the other hand, injection of the Ago hook, but not the mutant hook, prevented the usual persistent increase in LP I h G max in the presence of 5 nM DA + TTX ( Figure 5C). These data indicated that the RNAi pathway was necessary to elicit the DA-and activity-dependent persistent increase in LP I h G max .

TRANSCRIPTION IS NECESSARY FOR THE DOPAMINE (DA)-AND ACTIVITY-DEPENDENT PERSISTENT INCREASE IN LATERAL PYLORIC (LP) I h G max
Activity-dependent processes can involve transcriptional regulation of mRNAs and/or miRs (Krol et al., 2010;Wibrand et al., 2010;Kandel, 2012). RNA polymerase II transcribes both mRNAs and miRs (Pawlicki and Steitz, 2010). In order to determine if RNA Polymerase II-dependent transcription was necessary for the DA-and activity-dependent increase in LP I h G max , we first employed pharmacological agents that acted on RNA Polymerase II to prevent transcription (Figure 6). The STG was superfused with or without (control) 5 µM DA from t = 0-60 min. This was followed by a 2 h wash with saline, then a 1 h block and TEVC to measure LP I h . Either 100 nM flavopiridol or 100 µM was superfused from t = −10-60 min. The results indicated that the drugs blocked the persistent increase in LP I h G max (Figures 6B, C). These drugs act by inhibiting cyclin dependent kinases (CDKs) that phosphorylate RNA polymerase II and thereby promote transcript elongation (Bensaude, 2011); however, CDKs are known to regulate a number of other proteins. For this reason, we repeated the experiments with a third transcription blocker, actinomycin D, which acts by intercalating into the DNA (Bensaude, 2011). Inclusion of 50 µM actinomycin D in the superfusate had no effect on its own, but blocked the DA-and activity-dependent persistent increase in LP I h G max (Figure 6D). Together these data suggested that RNA Polymerase II transcription was necessary for the DA-and activity-dependent increase in LP I h G max . Finally, to test the previously stated assumption that the persistent effects of 5 nM DA + TTX and 5 µM DA on LP I h G max were mediated by the same pathway, we repeated the flavopiridol experiment with 5 nM DA + TTX. Consistent with our hypothesis, flavopiridol blocked the persistent 55% increase in LP I h G max elicited by 5 nM DA + TTX ( mean + SEM LP I h G max in 5 nM DA + TTX = 0.155 + 0.01 µS, n = 9; in flavopiridol + 5 nM + TTX, = 0.108 + 0.008 µS, n = 4; Student's t-test p = 0.015)

THE SAME SLOW PROCESSES ARE NECESSARY FOR THE 5 nM DA INDUCED, ACTIVITY-INDEPENDENT INCREASE IN LATERAL PYLORIC (LP) I h G max
Thus far we have examined the cellular processes underpinning the persistent increase in LP I h G max without regard for other voltage-gated ionic conductances; however, LP I A and I h can be co-regulated (MacLean et al., 2005;Temporal et al., 2012;Krenz et al., 2013). We previously demonstrated that a 1 h application of 5 nM DA produced a persistent increase in LP I A that was dependent upon the D1R-PKA axis and mTORdependent translation; but, unlike LP I h , the persistent increase in LP I A was activity-independent (Rodgers et al., 2011b(Rodgers et al., , 2013. In order to better understand the signaling network that co-regulates LP I A and I h , we asked if RNAi and transcription were also necessary for the activity-independent persistent increase in LP I A G max (Figure 7). We repeated the hook injection experiment diagrammed in Figure 5A using 5 nM DA without TTX and LP I h G max is plotted for each treatment group; each symbol is one experiment; the horizontal bars represent the means. Asterisk indicates a significant difference as determined using t-tests to compare DA and saline treatment groups in preparations with (p = 0.701) and without (p = 0.011) flavopiridol. Note that an ANOVA could not be performed due to unequal variances between +/− flavopiridol groups (F -test, p < 0.03). (C) DRB (100 µM) blocks the persistent increase in LP I h G max elicited by 5 µM DA. Asterisk indicates a significant difference as determined using a one-way ANOVA with Tukey's post hoc tests that made all pairwise comparisons, F (3,25) = 3.827, p < 0.022. (D) Actinomycin D (50 µM) blocks the persistent increase in LP I h G max elicited by 5 µM DA but has no effect alone. Asterisks indicate significant differences as determined using a one-way ANOVA with Tukey's post hoc tests that made all pairwise comparisons, F (3,26) = 7.611, p = 0.0008.
measured LP I A . The hook blocked the DA induced increase in LP I A G max ( Figure 7A); thus, the RNAi pathway was necessary for the persistent increase in LP I A . We next repeated the experiments with the transcription blockers diagrammed in Figure 6A. DRB alone significantly increased LP I A G max relative to saline controls (t-test, p = 0.026, n > 5 per treatment group), and was not considered further. On the other hand, both flavopiridol ( Figure 7B) and actinomycin D ( Figure 7C) blocked the DA-induced increase in LP I A G max . Consistent with the idea that 5 µM DA and 5 nM DA acted through the same pathway, flavopiridol also blocked the persistent ∼25% increase in LP I A G max elicited by 5 nM DA + TTX (mean + SEM LP I A G max in 5 nM DA + TTX = 3.1 + 0.2 µS, n = 8; in flavopiridol + 5 nM + TTX = 2.08 + 0.23 µS, n = 4; Student's t-test p = 0.005). We concluded that RNA polymerase II transcription was also necessary for the DA-induced persistent increase in LP I A .

DISCUSSION
The main finding of the work presented here is that tonic nM DA can act at high affinity D1Rs to permit a persistent, activitydependent increase in LP I h G max through a signaling network that relies on the canonical D1R-PKA axis, RNA Polymerase II transcription, components of the RNAi pathway, mTORC1 and translation. All of these same elements are also necessary for the activity-independent, persistent increase in LP I A G max elicited by tonic nM DA.

POTENTIAL MECHANISMS FOR HOW 5 nM DA PERSISTENTLY REGULATES LATERAL PYLORIC (LP) I A AND LP I h
Modulatory tone continuously influences ion current density: Washout of modulatory tone reduced LP I A G max and adding 5 nM DA back to the bath prevented the decrease and could even produce a persistent increase (Rodgers et al., 2013). The mechanism involved did not rely on alterations in the number of Kv4 transcripts (Rodgers et al., 2011b) that encode the poreforming subunits of the channels mediating LP I A (Baro et al., 1997(Baro et al., , 2000. If bath application of 5 nM DA was accompanied by a significant change in LP slow wave activity, then a persistent increase in LP I h was also observed (Rodgers et al., 2011a). In the simplest case, high affinity D1Rs regulate both LP I A and I h through the same mechanism, and activity-dependence is bestowed upon LP I h through an additional process. RNA polymerase II transcription is essential for the persistent increase in LP I A and I h elicited by 5 nM DA. Both mRNAs and miRs are transcribed by RNA polymerase II. Our data suggest miR expression is regulated by dopaminergic tone. The RNAi pathway, which processes miRs and mediates their effects, is necessary for the DA-induced persistent increases in LP I A and I h G max . Injecting the Ago hook to sequester endogenous Ago1, and thereby obstruct RNAi, did not appear to alter LP I A or I h over the long-term (several hours); however, Ago hook injections did block the persistent increase in LP I A and I h G max elicited by 5 nM DA. The most parsimonious interpretation of these Frontiers in Cellular Neuroscience www.frontiersin.org February 2014 | Volume 8 | Article 39 | 9

FIGURE 7 | miR transcription is required for the DA-dependent persistent increase in LP I h G max . (A)
The Ago hook blocks the persistent increase in LP I A G max elicited by 5 nM DA. The Ago hook injection experiments described in Figure 5A were repeated without TTX, and LP I A G max was measured. Each symbol represents one experiment, and the horizontal bars represent the means. The asterisk indicates a significant difference between the saline and DA-treated preparations, as determined using Student t-tests for the non-injected preparations (p < 0.006) and the hook-injected preparations (p = 0.244). (B) Flavopiridol (100 nM) blocked the persistent increase in LP I A G max elicited by 5 µM DA. Experiments diagrammed in Figure 6A were repeated with flavopiridol except that LP I A was measured and plotted for each treatment group. Each symbol is one experiment; horizontal bars are the means. Asterisk indicates a significant difference as determined using t-tests to compare DA and saline treatment groups in preparations with (p = 0.969) and without (p = 0.004) flavopiridol. Note that unequal variances between +/− flavopiridol groups prevented analysis with an ANOVA (F -test, p < 0.008). (C) Actinomycin D (50 µM) blocked the persistent increase in LP I A G max elicited by 5 µM DA. Experiments diagrammed in Figure 6A were repeated with Actinomycin D, except that LP I A was measured and plotted for each treatment group. Each symbol is one experiment; horizontal bars are the means. Asterisks indicate significant differences as determined using a Kruskal-Wallis test with Dunn's multiple comparison posthoc tests, p = 0.0014. data is that DA regulates miR expression, although there are other explanations (Pinder and Smibert, 2013). If DA suppressed miR expression, then Ago hook injections should have occluded the DA effect. Since Ago hook injections blocked rather than occluded DA's effect, it is more likely that DA enhanced miR expression. Consistent with this hypothesis, activation of high affinity D1Rs has been shown to enhance miR-181a expression in hippocampal neurons (Saba et al., 2012). The half-lives of miRs are variable, ranging from minutes to hours (Bail et al., 2010;Krol et al., 2010). MiR expression can be regulated by altering rates of transcription (Fiore et al., 2009;Impey et al., 2010;Nudelman et al., 2010), processing (Heinrich et al., 2013;Massirer and Pasquinelli, 2013) and/or degradation (Chatterjee and Grosshans, 2009;Krol et al., 2010;Wibrand et al., 2010;Grosshans and Chatterjee, 2011). DA could be acting on one or all three of these processes to enhance miR expression. The D1R-PKA axis could directly increase transcription rates through the cAMP response element binding protein (CREB), a transcription factor known to augment the expression of several miRs (Vo et al., 2005;Tan et al., 2012a,b). Monoamines can also regulate the expression of Piwi-interacting RNAs (piRs), an additional class of small noncoding RNAs that can promote long-term neuronal plasticity by regulating transcription factor expression (Rajasethupathy et al., 2012). Thus, it is possible that DA could indirectly influence miR transcription by regulating piRs. It should be noted that although both Piwi and Ago1 proteins possess PIWI domains, the amino acids necessary for binding to the Ago hook are not preserved in Piwi proteins (Parker et al., 2004), and the Ago hook does not pull down Piwi proteins (Till et al., 2007). Theoretically, DA could also regulate the processing or stabilization of nascent miRs, but to the best of our knowledge, this has not yet been demonstrated. For the remainder of this discussion, we assume the same miR(s) controls both LP I A and I h densities in order to permit their co-regulation; however, it is also possible that distinct miRs regulate LP I A and I h densities, and in this case, both DA and a change in activity may be required to increase the expresson of the miR regulating I h density (Wibrand et al., 2010;Cohen et al., 2011;Eacker et al., 2011). Both mTORC1 and translation are necessary for the DAinduced persistent increases in LP I A and I h G max . Many cellular processes are regulated by mTORC1, including cap-dependent translation (Laplante and Sabatini, 2012). The most parsimonious interpretation of the data is that DA directly or indirectly enhances mTORC1-dependent translation of a protein(s) because, the mTORC1 inhibitor, rapamycin, and the translation inhibitor, anisomycin, had no effect on their own, but each prevented the persistent increases in LP I A and I h elicited by 5 nM DA. The identity of the transcript(s) undergoing enhanced mTORC1-dependent translation is unknown. Increased translation of ion channel subunits could augment ion channel surface expression and LP maximal conductances, including the pore-forming subunits that mediate I A (Kv4) and I h (HCN) or the auxiliary subunits that regulate channel conductance and trafficking Zhang et al., 2003;Santoro et al., 2009;Lin et al., 2010;Santoro et al., 2011). Additional candidates for altered translation include a wide variety of proteins involved in ion channel translation, trafficking and surface expression. Despite the fact that there are many potential targets, for the ease of discussion, here we further consider only Kv4 and HCN transcripts. How might the increased expression of a miR lead to increased mTORC1-dependent translation of Kv4 and HCN transcripts? RNA binding proteins (RBPs) act in a combinatorial fashion to repress or enhance translation of the transcript to which they bind (Darnell and Richter, 2012;Darnell, 2013). miRs remodel the RBP complexes bound to transcripts and thereby either inhibit or facilitate their translation (Lee and Vasudevan, 2013). We hypothesize that 5 nM DA promotes expression of a miR that can reconfigure the RBP complexes on Kv4 and HCN transcripts to facilitate their translation. There are a number of ways that this could occur: the miR could act as a decoy and compete with Kv4 and HCN transcripts for binding to a repressive RBP (Eiring et al., 2010); or, the miR could compete with a more repressive RBP for binding to Kv4 and HCN transcripts (Ma et al., 2010). Then again, the miR could noncompetitively bind Kv4 and HCN transcripts and recruit RBPs that promote translation (Vasudevan et al., 2007;Tsai et al., 2009). Alternatively, the miR could de-repress Kv4 and HCN transcripts by reducing the number of available repressive RBPs; for example, the miR could bind repressive RBP transcripts and block their translation initiation (Djuranovic et al., 2012;Meijer et al., 2013) and/or elongation (Graber et al., 2013a) and/or promote their degradation (Djuranovic et al., 2011;Fukaya and Tomari, 2012). Since a given transcript is regulated by multiple elements, the aforementioned models could account for both the activity-dependent and -independent regulation of LP I h and I A , respectively, if we postulate activity-dependent remodeling of an additional RBP complex on HCN transcripts. Although these hypotheses have the advantage of being simple and straightforward, they are highly speculative. It is also possible that the miR(s) indirectly alters RBP complexes on Kv4 and HCN transcripts by regulating transcripts encoding other types of proteins. For example, Kv1 transcripts in hippocampal neurons compete with CAMKIIα and other transcripts for binding to a limited number of Hu/embryonic lethal, abnormal vision (ELAV) RBPs that promote translation; and, Kv1 transcripts bind these facilitatory RBPs and are translated only when competitor transcripts (e.g., CAMKIIα are destabilized and degraded (Sosanya et al., 2013), suggesting that the shared RBPs may promote switching between two distinct programs/states.

COMMONALITIES BETWEEN ACTIVITY-DEPENDENT REGULATION OF LATERAL PYLORIC (LP) I h AND SYNAPTIC PLASTICITY
Learning and memory depend upon coordinated intrinsic and synaptic plasticity (Sehgal et al., 2013). Coordination can be achieved through shared transduction components. In this regard, many of the cellular processes underpinning long-term activity-dependent regulation of LP I h G max and synaptic plasticity are similar. First, miRs can contribute to long-term synaptic plasticity in multiple species. Throughout the mammalian brain, miRs participate in activity-dependent synaptic remodeling and regulate cognition by controlling components of the postsynaptic density, spine volume and synaptic cytoskeletal proteins (Schratt, 2009;Eacker et al., 2013;Hansen et al., 2013). miRs are also linked to synaptic plasticity and long-term memory in Drosophila (Ashraf et al., 2006;McCann et al., 2011). In Aplysia, serotonin can down-regulate expression of a miR that normally constrains synaptic plasticity (Rajasethupathy et al., 2009). Second, mTOR-dependent translation is necessary for long-term synaptic plasticity in a number of systems (Hoeffer and Klann, 2009;Gkogkas et al., 2010;Graber et al., 2013b). In rat hippocampal neurons, the D1R-PKA axis permits local mTORC1dependent translation of the glutamate receptor subunit, GluR1, in an activity-dependent fashion (Smith et al., 2005). D1Rs mediate memory consolidation in the gerbil auditory cortex through mTOR-dependent protein synthesis (Schicknick et al., 2008). In Aplysia, long-term facilitation of a sensory-motor synapse relies on serotonin-enabled local mTORC1-dependent translation (Yanow et al., 1998;Casadio et al., 1999;Wang et al., 2009). Similarly, long-term facilitation at a crayfish neuromuscular synapse required local mTOR-dependent translation (Beaumont et al., 2001). While synaptic and intrinsic activity-dependent processes employ similar mechanisms, it is important to note that modulatory tone also utilizes the same elements to persistently regulate ion current density in an activity-independent fashion (Rodgers et al., 2011b).

DOPAMINERGIC TONE ACTS OVER TWO DISTINCT TIME SCALES TO CO-REGULATE I A AND I h
The balance of ion conductances, rather than the absolute number of ion channels, can determine certain features of neuronal activity (Marder, 2011). It appears that several mechanisms can control the balance of the same conductance pair. Different mechanisms may predominate in each cell type; for example, GABA A receptors and HCN1 channels co-vary to maintain hippocampal neuron resting membrane potential (Bonin et al., 2013), but in cortical pyramidal neurons, these two conductances vary inversely to maintain excitatory post synaptic potential summation (Chen et al., 2010). Even within the same cell type, two conductances can be co-regulated by multiple mechanisms that act over distinct time scales. In LP, I A and I h densities are coordinated by at least three distinct mechanisms in order to maintain the timing of LP activity; and, for two of the mechanisms, dopaminergic tone was shown to play a permissive role. In the first, most rapid mechanism, activation of high affinity D1Rs conferred activitydependence upon LP I h . Alterations in LP activity that advanced LP firing phase largely due to a decrease in LP I A triggered a rapid compensatory decrease in LP I h to restore the timing of the LP activity phase . Activation of high affinity LP D1Rs also enabled co-regulation of LP I A and I h through a second, slower process described here. Collectively, our work shows that an increase in dopaminergic tone produces a slow increase in LP I A G max , independent of LP I h ; however, when LP activity changes, then the same DA-enabled mechanism is engaged to increase LP I h G max . In another study, overexpression of Kv4 channels in LP neurons increased LP I A over days in organ culture and triggered a compensatory increase in LP I h through a third, activity-independent mechanism (MacLean et al., , 2005. Descending modulatory inputs were intact in the latter study, but it is unclear if modulatory tone played a role. It has been demonstrated that other modulators can Frontiers in Cellular Neuroscience www.frontiersin.org February 2014 | Volume 8 | Article 39 | 11 maintain activity and conductance ratios over the long-term, and removal of modulators appears to change the ratios that are maintained (Rezer and Moulins, 1992;Simmers, 1998, 2002;Khorkova and Golowasch, 2007). Taken together, the data suggest that modulatory tone may influence neuronal identity by determining which homeostatic mechanisms are in play.

DOPAMINERGIC TONE MAY PERSISTENTLY REGULATE VOLTAGE-GATED CONDUCTANCES IN OTHER CELL TYPES
If regulation of voltage-gated conductances by modulatory tone is widespread, then the findings presented here could have important implications for neurological and psychiatric disorders involving disruptions in dopaminergic tone. For example, in a mouse model of Parkinson's disease, dopaminergic tone was severely attenuated and I h was persistently reduced in globus pallidus neurons (Chan et al., 2011). Since DA receptors are expressed in rodent globus pallidus neurons (Mansour et al., 1990;Marshall et al., 2001;Araki et al., 2007), the reduction in I h could potentially be explained by a lack of normal DA-enabled, activity-dependent compensation.