Neuregulin and Dopamine D4 Receptors Contribute Independently to Depotentiation of Schaffer Collateral LTP by Temperoammonic Path Stimulation

Abstract Prior studies have found that dopamine (DA), acting at D4 receptors, and neuregulin (NRG), likely acting at ErbB4 receptors, are involved in a form of depotentiation of long-term potentiation (LTP) at Schaffer collateral (SC) synapses in the hippocampus. Furthermore, DA and NRG actions are intertwined in that NRG induces DA release. We previously found that low-frequency stimulation (LFS) of temperoammonic (TA) inputs to area CA1 also depotentiates previously established SC LTP through a complex signaling pathway involving endocannabinoids, GABA, adenosine, and mitogen-activated protein kinases (MAPKs), but not glutamate. In the present studies, we found that TA-induced SC depotentiation in hippocampal slices from Sprague-Dawley albino rats also involves activation of both D4 receptors and NRG-activated ErbB receptors, but that the roles of these two modulator systems are independent with D4 receptor antagonism failing to alter chemical depotentiation by NRG1β. Furthermore, a selective D4 receptor agonist was unable to depotentiate SC LTP when administered alone, suggesting that D4 receptor activation is necessary but not sufficient for TA-induced SC depotentiation. Chemical depotentiation by NRG1β was inhibited by a Pan-ErbB antagonist and by picrotoxin (PTX), an antagonist of GABA-A receptors (GABAARs), indicating that NRG likely promotes SC depotentiation via effects on GABA and interneurons. These findings have implications for understanding the role of DA and NRG in cognitive dysfunction associated with neuropsychiatric illnesses.


Introduction
Hippocampal synapses operate over a dynamic range of efficacy and are subject to both short-and long-term forms of plasticity, including long-term potentiation (LTP) and long-term depression (LTD), leading candidates as synaptic memory mechanisms (Malenka and Bear, 2004;Kandel et al., 2014, Nicoll, 2017. Because there are limits on the degree to which hippocampal synapses can po-tentiate or depress, there is interest in determining how synapses reset to baseline following the induction of stable synaptic plasticity. This interest is compounded by the fact that the hippocampus is a short-term, limited capacity storage system. Potential mechanisms for synaptic resetting include homeostatic plasticity, in which synapses adjust to changes in activity over time (Turrigiano, 2011), and homosynaptic resetting, in which the same synapses that are altered instruct their own resetting (Fujii et al., 1991;Bashir and Collingridge, 1994).
We have been interested in determining whether extrahippocampal inputs can instruct Schaffer collateral (SC) synapses to reset heterosynaptically following successful induction of stable LTP. We have found that low-frequency stimulation (LFS) of temperoammonic (TA) inputs to stratum lacunosum moleculare (SLM) in area CA1 can induce depotentiation (LTP-D) of SC synapses without persistently altering baseline SC transmission or the ability of SC synapses to undergo subsequent LTP after resetting . TA-induced LTP-D involves complex signaling including activation of GABA-A receptors (GABA A Rs), cannabinoid-1 receptors (CB1Rs) and adenosine A1 receptors (A1Rs), and activation of mitogen-activated protein kinase (MAPK) signaling, including extracellular signal-related kinase 1/2 (ERK1/2) and p38 MAPK .
Surprisingly, TA-induced LTP-D does not involve activation of AMPA-type glutamate receptors, NMDA receptors, metabotropic glutamate receptors, or L-type voltage-activated calcium channels . These latter observations led us to consider the role of other inputs in SLM. Besides direct glutamatergic inputs and long-range GABAergic inputs from entorhinal cortex (Basu et al., 2016), SLM receives input from neuromodulatory systems, including monoamines (Swanson and Hartman, 1975), and dopamine (DA) has previously been shown to dampen direct TA glutamatergic inputs from entorhinal cortex but not SC pathway responses (Otmakhova and Lisman, 1998). Other studies, however, have shown that activation of DA D4 receptors (D4Rs) can drive a form of SC depotentiation when activated within 30 min of LTP induction (Kwon et al., 2008). This form of LTP-D is induced by neuregulin-1 (NRG1) and involves activation of ErbB receptors, which in turn enhance DA release (Kwon et al., 2005(Kwon et al., , 2008. Similarly, homosynaptic activation of SC inputs by theta pulse stim-ulation can reverse SC LTP via D4Rs when administered shortly after LTP induction (Kwon et al., 2008). Other studies indicate that low frequency activation of DA fibers in the hippocampus can stimulate D4Rs to dampen SC responses via activation of parvalbumin positive interneurons (Rosen et al., 2015). Based on these observations, we examined the roles of DA and NRG in TA-induced LTP-D.

Hippocampal slices
Protocols for animal use were approved by the Washington University Animal Studies Committee in accordance with national and international guidelines. Hippocampal slices were prepared from the septal (dorsal) hippocampal region of postnatal day (P)28-P32 Sprague Dawley albino rats using previously described methods Tokuda et al., 2010). Pregnant female rats were purchased from Charles River (Crl:CD(SD), RRID:RGD_734476); male offspring were raised to age 28-32 d in an approved animal care facility. On the day of experiments, rats were anesthetized with isoflurane, decapitated, and hippocampi were dissected. Isolated hippocampi were placed in ice-cold artificial CSF (ACSF) containing 124 mM NaCl, 5 mM KCl, 2 mM MgSO 4 , 2 mM CaCl 2 , 1.25 mM NaH 2 PO 4 , 22 mM NaHCO 3 , 10 mM glucose, bubbled with 95% O 2 -5% CO 2 at 4-6°C, and cut into 450-m slices using a rotary tissue slicer. The slices were cut to include a significant portion of entorhinal cortex to keep TA inputs to SLM in the CA1 region intact to the extent possible . In the present experiments we did not monitor field potentials in SLM directly because these potentials reflect a combination of several inputs from entorhinal cortex and other regions (Basu et al., 2016;Swanson and Hartman, 1975), although prior studies indicate that repeated LFS of the TA pathway, akin to what we use in this study, produces LTD of these inputs (Dvorak-Carbone and Schuman, 1999). After preparation, slices were allowed to recover from dissection in an incubation chamber containing gassed ACSF for 1 hr at 30°C before experiments.

Hippocampal slice physiology
At the time of study, slices were transferred individually to a submersion-recording chamber. Experiments were done at 30°C with continuous ACSF perfusion at 2 ml/ min. Extracellular recordings were obtained from the apical dendritic layer (stratum radiatum) of the CA1 region for analysis of EPSPs using glass electrodes filled with 2 M NaCl (5-10 M⍀ resistance).
EPSPs were evoked using 0.1-ms constant current pulses through a bipolar stimulating electrode in the SC pathway. A second stimulating electrode was placed in the TA pathway to activate inputs to CA1 in SLM. A control input-output curve was obtained to determine stimulus intensities for subsequent studies. Responses were monitored by applying single stimuli to the SC pathway every 60 s at half maximal intensity. After establishing a stable baseline for at least 10 min, SC LTP was induced Human Services National Institutes of Health National Institute of Mental Health Gant MH101874, the Taylor Family Institute, and the Bantly Foundation.
Conflict of Interest: CFZ is a member of the Scientific Advisory Board of Sage Therapeutics. Sage Therapeutics did not fund this research and was not involved in the conduct of this research. Dr. Izumi has no conflicts of interest to disclose. There are no other competing financial interests. by a single 100 Hz ϫ 1 s high-frequency stimulus (HFS) using the same intensity stimulus. Input-output curves were repeated 60 min following tetanic stimulation. TA stimulation to induce SC depotentiation was administered as a 1 Hz ϫ 15 min LFS at half maximal intensity based on prior results .

Materials and Methods
Chemicals and pharmacological agents were obtained from Tocris or Sigma (St. Louis MO). NRG1␤ was obtained from R&D Systems. Concentrations of all agents used in this study (agonists and antagonists) and durations of exposure were based on published literature and, more specifically, on the lack of effect on baseline SC transmission under the conditions of our experiments. The exception to this was picrotoxin (PTX), a GABA A R antagonist that induced changes in basal EPSPs even at the low concentration (1 M) used for the experiments in Figure 5. Because of the changes in basal transmission, PTX was washed on at the initiation of the recordings and included in ACSF for the duration of these experiments.

Statistical analysis
Data were collected and analyzed using PClamp software (Molecular Devices). Data in the text are expressed as mean Ϯ SEM. A two-tailed Student's t test was used for comparisons between groups. Statistical comparisons were based on analysis of input-output curves at the 50% maximal point at baseline and sixty minutes following tetanic or 1-Hz stimulation, with p Ͻ 0.05 considered significant . Figure 1A shows an example of this type of analysis. The time course graphs in all figures display results from continuous monitoring of EPSPs using the 50% maximal stimulus from the baseline IO curve as the 100% response. Data presented in the text are derived from analysis of IO curves as noted above. Analyses were done using commercial software (SigmaStat, Systat Software).

Results
In prior studies, we found that 1 Hz ϫ 15 min LFS of TA (perforant path, P) inputs to area CA1 produces only a transient depression of baseline transmission of SC synapses . This same TA LFS, however, persistently depotentiates previously established LTP in the SC pathway when TA stimulation is administered an hour or so following induction of stable LTP ( Fig. 1A; 143.0 Ϯ 7.0% change in EPSP slope 60 min following HFS vs 80.7 Ϯ 8.5% change 60 min following TA LFS, p ϭ 0.0001, N ϭ 7). Because prior studies have shown that activation of D4Rs are involved in homosynaptic LTP-D and NRG1-mediated chemical depotentiation of SC LTP at short intervals following LTP induction (Kwon et al., 2008), we were interested in determining whether D4Rs also contribute to TA-induced SC LTP-D an hour or more following LTP onset. To test this, we first examined the effects of clozapine, an antipsychotic drug that inhibits D2-like DA receptors with higher affinity for . Note that data in A include control slices done in our prior manuscript  with additional slices added. The right graph in A depicts an analysis from a single slice based on changes in the IO curve. Black circles are baseline IO results, while red circles depict changes 60 min after SC HFS, and red squares show reversal of LTP following PLFS. B, The ability of PLFS to depotentiate SC LTP is completely blocked by 1 M clozapine (white bar). Upper traces to the right show representative EPSPs at baseline (black lines) and 60 min following SC HFS (redlines) while lower traces show baseline (black lines) compared to 60 min following PLFS (red lines). Calibration: 1 mV, 5 ms.

New Research
D4Rs (Sanyal and Van Tol, 1997). We found that 1 M clozapine did not alter the transient synaptic depression observed during 1-Hz TA stimulation but completely blocked the ability of this stimulation to persistently depotentiate SC synapses (143.7 Ϯ 10.5% change in EPSP slope 60 min following HFS vs 139.8 Ϯ 11.9% change 60 min following TA LFS, p ϭ 0.507, N ϭ 5; Fig. 1B).

Discussion
Hebbian plasticity in the hippocampus is thought to play a key role in learning and memory (Kandel et al., 2014;Nicoll, 2017), but has the limitation that this type of use-dependent synaptic change is saturable unless there  are mechanisms by which synapses can be reset for future plasticity and learning (Turrigiano, 2011). Hence, there has been interest in understanding mechanisms by which synapses depotentiate following LTP induction, including homeostatic changes, and forms of homosynaptic and heterosynaptic depotentiation. We previously found that LFS of direct TA (perforant path) inputs to area CA1 can depotentiate previously established SC pathway LTP, without having persisting effects either on baseline transmission or the ability of subsequent stimulation to induce LTP or LTD at SC synapses . TA-induced SC depotentiation involves complex signaling mechanisms including GABA A Rs, CB1Rs and adenosine A1Rs along with activation of two MAP kinases, ERK1/2 and p38 MAPK .Surprisingly, this form of synaptic plasticity does not appear to involve activation of glutamate receptors . In the present work, we provide evidence that TA-induced SC depotentiationinvolves activation of D4 type DA receptors and ErbB signaling.
Prior studies indicate that DA plays a key role in modulating hippocampal function, including long-term forms of synaptic plasticity (Furth et al., 2013). Intriguingly, DA innervation of area CA1, particularly afferents arising from DA cell bodies in the ventral tegmental area to SLM, is relatively sparse, with an absence of DA transporters but significant expression of DA receptors in SLM (Kwon et al., 2008;Smith and Greene, 2012;Kempadoo et al., 2016). Other work indicates that adrenergic terminals in the hippocampus can release DA and that afferents from the locus coeruleus may be critical for providing DA signals to the CA1 region (Smith and Greene, 2012; Kempa-doo et al., 2016). Additionally, norepinephrine can directly activate D4Rs providing another way that the adrenergic system can stimulate DAreceptors to modulate CA1 function (Root et al., 2015;Sánchez-Soto et al., 2016). If norepinephrine is the key monoamine transmitter driving TA-induced SC depotentiation, our present results indicate that its effects on SC LTP involve D4Rs. We previously found that exogenous norepinephrine, acting at adrenergic receptors, prevented rather than promoted homosynaptic SC depotentiation (Katsuki et al., 1997), indicating that norepinephrine alone does not mimic the effects of TA stimulation.
Buonanno and colleagues previously demonstrated a role for D4Rs in a form of chemical depotentiation of SC LTP induced by NRG1␤ and involving activation of ErbB receptors (Kwon et al., 2005(Kwon et al., , 2008. They also found that NRG1␤ promoted release of DA in the CA1 region, indicating that the effects of NRG and DA are intertwined.  D4Rs also contributed to homosynaptic SC depotentiation following theta pulse stimulation administered shortlyafter LTP was induced (Kwon et al., 2008).These forms of depotentiation, particularly NRG1␤-induced LTP-D, were observed within 30 min of LTP induction, but not 50 or more min after stable LTP had been established. Prior studies have found that LTP is more readily reversed early (Ͻ30 min) after induction rather than later after induction (an hour or more; Izumi and Zorumski, 1993;Barr et al., 1995;Stäubli and Chun, 1996;Kwon et al., 2005Kwon et al., , 2008. Several factors may contribute to the stability of LTP over time and the ease with which LTP can be reversed, including the stimuli used to induce LTP (e.g., single vs multiple HFS), the conditions under which LTP was induced (e.g., ionic conditions, age of animals; Huang et al., 1996;Huang and Kandel, 2005) and the type and duration of stimulation used to induce depotentiation (briefer HFS trains vs LFS of varying durations; Fujii et al., 1991;Izumi and Zorumski, 1993;Bashir and Collingridge,1994;Wagner and Alger, 1995;Kwon et al., 2005). In our studies we used a single 100 Hz ϫ 1 s HFS to induce LTP that remained stable for over an hour in P30 rat hippocampal slices and reversed this LTP using 15 min 1-Hz LFS; the LFS used for depotentiation was selected based on a standard LFS that has been used to induce homosynaptic LTD (Dudek and Bear, 1992) or LTP-D in the SC pathway (Izumi and Zorumski, 1993). These differences in stimulation paradigms may contribute to the fact we were able to reverse SC LTP an hour or more after induction.
In the hippocampus, ErbB4 receptors are a predominant NRG receptor  and are expressed on GABAergic interneurons Neddens and Buonanno, 2010;Bean et al., 2014). NRG1␤ disinhibits interneurons via ErbB4 and promotes release of GABA (Woo et al.2007). Consistent with this, we found that NRG1␤-induced SC depotentiation was blocked by a GABA A R antagonist, suggesting that NRG1-induced GABA release may be critical for this form of synaptic resetting. The effects of NRG1␤ on stable LTP were not reversed by a D4R antagonist, suggesting that mechanisms contributing to early and later LTP reversal by NRG1␤ and D4R activation likely differ. Other evidence indicates that D4Rs, like ErbB4, are expressed on some interneurons (Mrzljak et al., 1996), and recent work has shown that low DA release evokes feedforward inhibition in CA1 that is mediated by D4Rs on parvalbumin (PV)positive interneurons (Rosen et al., 2015). Despite the fact that TA-induced LTP-D is blocked by a D4R antagonist, we found that a D4R agonist alone failed to depotentiate SC LTP when administered 1 h after LTP induction, although NRG1␤ was effective. At present we do not know which interneurons contribute to TA-induced SC depotentiation, but note that several types of interneurons have dendrites in or extending to SLM, including PVϩ interneurons (axo-axonic and some basket cells), cholecystokininpositive interneurons and neurogliaform cells (Klausberger et al., 2003;Klausberger and Somogyi, 2008;Klausberger, 2009;Overstreet-Wadiche and McBain, 2015) along with longer-range GABAergic inputs from entorhinal cortex to SLM (Basu et al., 2016).
Based on the effects of selective agonists and antagonists, we have proposed a scheme for TA-induced SC depotentiation in which activation of ERK1/2, diacylglycerol lipase, and endocannabinoid synthesis are involved relatively early in the cascade, with activation of GABA A Rs, p38 MAPK, and A1Rs participating as more downstream effectors . Our present results indicate that D4Rs act more proximally in the depotentiation scheme than endocannabinoids, NRG1 or adenosine. D4Rs can also stimulate ERK1/2 (Bitner et al., 2006;Oak et al., 2001), which we have found is involved relatively early in the depotentiation scheme (Fig. 8). Consistent with Kwon et al. (2005Kwon et al. ( , 2008, we also found that a D4R agonist alone was not capable of inducing depotentiation when administered an hour or more after LTP had been established. These findings suggest that D4R activation is necessary but not sufficient to induce TA-mediated synaptic resetting, and that D4Rs and ErbB receptors act independently in the cascade. Based on experiments to date, adenosine is the most distal signal in the cascade yet identified. We have tested some messengers linked to A1R-induced LTD (Brust et al., 2006;Chen et al., 2014) in our prior work and found a role for p38 MAPK . However, in our experiments p38 MAPK appeared to act upstream of A1Rs and did not block the effects of chemical depotentiation by CPA.
The involvement of D4Rsand NRG1␤ in TA-induced SC depotentiation provides a possible link to the role of the DA and NRG systems in cognitive defects associated with psychiatric illnesses such as schizophrenia, major depression, substance use disorders, and attention deficit hyperactivity disorder Shamir et al., 2012;Penzes et al., 2013). D4Rs in the hippocampus modulate gamma oscillations that are important in atten- Figure 8. The diagram depicts a current scheme for TA-induced SC LTP-D based on prior studies  and the present results. Agents that promote chemical depotentiation are shown in green while agents that inhibit TAinduced SC depotentiation are shown in red.
New Research tion and information processing (Andersson et al., 2012). Similarly, by altering the ability of synapses to reset following LTP, changes in DA and NRG function could be important in driving changes in hippocampal input-output relationships observed in animal models of psychiatric illnesses (Airan et al., 2007), and perhaps in the ability to learn and remember new information. The net effects of changes in DA and NRG modulation, even within the CA1 region, likely depend on input-specific actions and the subtypes of receptors that are stimulated, as well as on the state of glutamate synapses at the time of modulator release. In Drosophila, DA plays complex roles in memory and is required for both learning and forgetting (Berry et al., 2012); furthermore, sleep has been found to promote memory in Drosophila by impairing DA-driven forgetting (Berry et al., 2015). From the perspective of D4Rs, hyperdopaminergic tone would be expected to promote reversal of hippocampal and cortical LTP, perhaps leading to defects in longer-term memory storage (Xu et al., 2009), while lower DA tone could result in more persisting LTP in the CA1 region, perhaps dampening the ability of synapses to reset for future potentiation and learning. Thus efforts to modulate D4Rs and NRG signaling could have unique effects in a range of neuropsychiatric illnesses depending on the state of glutamate synapses, perhaps leading to novel ways to dampen the cognitive dysfunction that underlies illness-related disability.