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.

Tetrodotoxin (TTX), flupenthixol and myristoylated PKI_(14–22) were purchased from Tocris Bioscience (Bristol, UK), flavopiridol was from Selleckchem (Houston, TX), and all other chemicals were purchased from Sigma-Aldrich (St. Louis, MI). Peptides were synthesized by Biomatik (Wilmington, DE). DA was made fresh every 30 min to minimize oxidation. In all experiments, antagonists were administered 10 min before DA application. Rp-cAMPS (1 mM) effectively blocks protein kinase A (PKA) in several arthropod models such as Drosophila and crustaceans, including Panulirus (Erxleben et al., 1995; Kuromi and Kidokoro, 2000; Zhang et al., 2010). PKI is an effective blocker of the PKA catalytic subunit in crustaceans (Dixon and Atwood, 1989). Dosages of rapamycin (100 nM), anisomycin (30 µM) and actinomycin D (50 µM) were previously demonstrated to be effective in several invertebrate models including Panulirus (Rodgers et al., 2011a). Concentrations of flavopiridol (100 nM) and 5, 6-dichloro-1-β-D-ribobenzimidazole (DRB, 100 µM) were chosen based on previously demonstrated effective dosages (Chao and Price, 2001; Bensaude, 2011; Yuan and Burrell, 2013).

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 CaCl₂, 39 Na₂SO₄, 10 MgSO₄, 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.

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 × 10⁻²M tetraethylammonium (TEA) to block voltage-gated K⁺ channels, 2 × 10⁻⁴M cadmium chloride (CdCl₂) to block Ca²⁺- and Ca²⁺-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.

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).

The his-tagged hook (HHHHHHPDNGTSAWGEPNESSPGWGEMD) and mutant hook (HHHHHHPDNGTSvavEPNESSPvavEMD) 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.

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 non-parametric 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.

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 (Krenz et al., 2013); 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.

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 activity-dependent 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 long-term 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.

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 DA- and 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). To confirm PKA 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 DA- and 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) 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 activity-dependent persistent increase in LP I_h G_max. In the presence of either blocker, 5 µM DA could no longer elicit a significant increase in LP I_h G_max, but the blockers had no effect on their own.

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 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.

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)

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 mTOR-dependent 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 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.

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 pore-forming subunits of the channels mediating LP I_A (Baro et al., 1997, 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 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 DA-induced 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 (An et al., 2000; 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.

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 post-synaptic 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 mTORC1-dependent 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).

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 activity-dependence 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 (Krenz et al., 2013). 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., 2003, 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 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; Thoby-Brisson and 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.

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.