Effects of μ-opioid receptor modulation on the hippocampal network activity of sharp wave and ripples

Slice preparation

Animals were deeply anaesthetized with diethyl ether and decapitated immediately after they stopped breathing. The brain was removed from the skull, placed in a chilled (2–4°C) artificial cerebrospinal fluid containing (in mM): 124 NaCl, 4 KCl, 2 MgSO₄, 2 CaCl₂, 1.25 NaH₂PO₄, 26 NaHCO₃ and 10 glucose. The solution was equilibrated with 95% O₂ and 5% CO₂ gas mixture to achieve pH 7.4. The two hippocampi were excised free inside the chilled medium, then transverse slices of 500–550 μm thick were prepared from the ventral hippocampus, extending between 1 and 4 mm from the end of the structure using a McIlwain tissue chopper (Mickle Laboratory Engineering Co. Ltd, Gomshall, Surrey, UK). The slices were immediately transferred to an interface type recording chamber where they were maintained at a constant temperature of 31 ± 0.5°C. Slices were continuously supplied with a humidified mixed gas containing 95% O₂ and 5% CO₂ and they were perfused with standard artificial cerebrospinal fluid containing (in mM): 124 NaCl, 4 KCl, 2 MgSO₄, 2 CaCl₂, 1.25 NaH₂PO₄, 26 NaHCO₃, and 10 glucose, pH 7.4, equilibrated with 95% O₂ and 5% CO₂ gas mixture.

Recordings and data processing

Recordings started approximately 3 h after the slices were placed in the recording chamber to allow them to recover and organize spontaneous activity. Recordings of spontaneous and evoked extracellular potentials were made from the CA1 stratum pyramidale using carbon fibre electrodes (diameter 7 μm; Kation Scientific, Minneapolis, MN, USA). Evoked synaptic responses consisting of field EPSPs and population spikes (PSs) were recorded by delivering electrical pulses (intensity 0.1–0.6 mA, duration 0.1 ms) at the Schaffer collaterals using a bipolar platinum/iridium wire electrode (wire diameter of 25 μm; World Precision Instruments, Sarasota, FL, USA). PS with amplitude of about 2 mV was evoked every 30 s and continuously monitored in order to determine the stability of the response. Only slices with a stable response for at least 10 min were selected for further experimentation. Signals were recorded with a Neurolog amplifier (Digitimer Limited, Welwyn Garden City, Hertforshire, UK), band-pass filtered at 0.5 Hz–2 kHz, digitized at 5 kHz, and stored in a computer disc using the CED 1401-plus interface and the Spike5 and Signal software (Cambridge Electronic Design, Cambridge, UK).

As described in Results section, episodes of spontaneous network potentials identified as SWR activity consisted of either isolated single events or sequences of multiple consecutive events termed clusters (see Figure 1A). We call the first event in a sequence the primary event, while the subsequent events are called secondary events. Events were further analysed into slow and fast components, the slow sharp wave and the high-frequency ripple oscillation (which, in rats, peaks at 110–200 Hz). The following parameters were quantified: the inter-episode interval (IEpI) determined as the time interval between episodes, that is, the interval between the last event of an episode and the first event of the following episode; the inter-event interval (IEI) determined as the time interval between successive individual SPWs regardless of whether they appeared as isolated events or clusters; the intra-cluster interval (ICI, the mean time interval between consecutive events inside a cluster); the probability of occurrence of clusters of SWRs (probability of clusters); the number of events per minute; and the number of episodes per minute. Additionally, the amplitude, the duration at half amplitude and the maximum duration of the primary event were also measured. IEI, IEpI, ICI, amplitude and duration were measured after down sampling (at 1–2 kHz) and then low-pass filtering (at 30 Hz) original data. Individual sharp waves were detected by setting a threshold at a level where all putative events were marked as verified by visual inspection. Measurements of the above parameters were made during the last 2 min before applying the drug and during the last 2 min of drug application. High-frequency ripple oscillation was revealed by applying fast Fourier transform or filtering original records at 90–300 Hz. We used power spectra to measure the power of the peak of the ripple oscillation. In filtered recordings, ripples were detected by setting a threshold at four times the SD of event-free baseline noise; we measured (i) the number of troughs of a ripple event; (ii) the duration of the ripple event; (iii) the amplitude of the ripple event; and (iv) the mean intra-ripple frequency during each event. In episodes consisting of multiple events (i.e. clusters), measurements of ripple parameters were obtained from the primary event only. For accuracy reasons, all measurements were made manually, with the exception of ripple power that was measured using plots of fast Fourier transforms. In a given slice, the values obtained from the measurement of all sharp waves contained in the 2 min periods of control/drug conditions were averaged and the resulting mean value was the representative value from this particular slice. Manual measures of ripple characteristics from each slice were made from an epoch of 30 consecutive events inside the 2 min periods. Mean values of each ripple parameter were calculated by averaging the values of all 30 ripple episodes in a particular slice.

In order to examine the effects of opioid agonists on neuronal excitability and on GABA_A receptor-mediated synaptic inhibition, we constructed input/output curves of the stimulation strength and the amplitude of PS, as well as of EPSP and PS before and during drug application. Neuronal excitability was quantified by measuring the strength of the stimulation required to produce half-maximal PS (I-50), the postsynaptic excitation required to produce half-maximal PS (EPSP₅₀), and the ratio between PS and the corresponding EPSP₅₀ (PS/EPSP). We quantified changes in excitability by measuring the changes of I-50, EPSP₅₀ and PS/EPSP. PS was quantified from its amplitude, measured as the length of the projection of the negative peak on the line connecting the two positive peaks of the PS waveform. EPSP was quantified from the maximal slope of its rising phase.

Synaptic GABA_A receptor-mediated recurrent inhibition was evoked using a protocol of paired-pulse stimulation consisting of two consecutive stimuli of identical intensity delivered at Shaffer collaterals with an inter-pulse interval of 10 ms. The strength of so-produced paired-pulse inhibition (PPI) was quantified using the ratio response evoked by the second (test) stimulus (PS2)/response evoked by the first (conditioning) stimulus (PS1). Based on the constructed input/output curves and previous observations (Petrides et al., 2007), the strength of PPI and its changes were measured at two specific points of the input/output curves. More specifically, the stimulation strengths causing 50% of the maximum PS were chosen to calculate the PS2/PS1 ratio. The effect of opioids on PPI was quantified from the change produced in the I₅₀ (shift of I-50) as well as the change of the PS2/PS1 ratio.

Data analysis

Values are expressed as mean ± SEM, and ‘n’ indicates the number of slices included in the analysis. In several instances, the number of animals is also given. For statistical comparisons (using paired and independent t-tests) of the drug effects on SWRs, measurements were made immediately before drug application and during the last 2–5 min of drug application. The EC₅₀ were calculated from log concentration–response curves after sigmoidal or non-linear (dose–response) fitting to experimental data. In case of lack of fit with the previous methods, polynomial fit was applied. For data analysis and preparation of figures we used the software SPSS 14 (SPSS Inc., Chicago, IL, USA) and Origin 7.5 (OriginLab Corporation, Northampton, MA, USA).

Materials

We used three opioid receptor agonists – morphine, the synthetic analogue of endogenous enkephalin [D-Ala²,N-Me-Phe⁴,Gly-ol⁵]enkephalin (DAMGO), and fentanyl. All three agonists exhibit selectivity for MORs (Waldhoer et al., 2004; Crooks et al., 2006). We also used the opioid receptor antagonist naloxone and the highly selective MOR antagonist, D-Phe-Cys-Tyr-D-Trp-Orn-Thr-Pen-Thr-NH₂ (CTOP). Morphine, fentanyl and naloxone were obtained from Sigma-Aldrich, Taufkirchen, Germany. DAMGO and CTOP were purchased from Tocris Cookson Ltd, Bristol, UK. All drugs were first prepared as stock solutions and then were dissolved in the artificial cerebrospinal fluid and bath applied. Drug effects on the parameters of the spontaneous and evoked activity were measured 30 min after adding them to the perfusion medium. However, when low drug concentrations were used, measurements were also obtained at 60 min of application.

Effects of μ-opioids on network excitability and recurrent inhibition

It is thought that the action of opioids in the hippocampus is produced mainly by a disinhibitory effect on the principal cells that results in an enhanced neuronal excitability of the network (Zieglgansberger et al., 1979; Nicoll et al., 1980; Masukawa and Prince, 1982; Swearengen and Chavkin, 1987). We thus examined the effects of morphine (5 nM, 10 μM) and DAMGO (10 nM, 1 μM) on the neuronal network excitability and paired-pulse recurrent inhibition in 41 slices obtained from 27 animals. In general, both drugs affected network excitability and inhibition, yet some differences in their action were revealed. Figures 7 and 8 show examples of the effects of morphine and DAMGO respectively. Summary results are shown in Figure 9. As shown by the excitability curves for morphine (Figure 7A-a, B-a) and DAMGO (Figure 8A-a, B-a) applied at relatively low concentrations (5–100 nM), neither of the two drugs significantly altered network excitability (see also Figure 9). However, at higher concentrations both morphine (1 and 10 μM, Figure 7C-a, D-b) and DAMGO (10 μM, Figure 8C-a) enhanced network excitability as shown by the leftward shift in the input/output curves, the corresponding decrease in I₅₀ and EPSP₅₀, and the increase in PS/EPSP ratio (Figure 9). The opioid-induced shift to the left of the stimulus-to-PS and EPSP-to-PS curves is consistent with previous reports (Dingledine, 1981; Watson and Lanthorn, 1993). Additionally, DAMGO was more potent than morphine in increasing network excitability, as shown by the lower mean EC₅₀ value (i.e. from all three excitability indexes) of DAMGO (209) compared with morphine (517). Neither of the two drugs produced a consistent effect on the amplitude of maximal PS and maximal EPSP.

Striking results were obtained from the drug effects on inhibition. Their effects on PPI were measured using the decrease in the rightward displacement of the input/output curve of the second PS (PS2) in relation to the first PS (PS1) evoked by the paired-pulse stimulation. This effect was reflected in the lower shift of I₅₀ and in the higher PS2/PS1 ratio. Although both drugs reduced PPI they displayed different profiles of action. As shown in Figures 7 and 9, morphine reduced PPI only at 10 μM, whereas DAMGO reduced PPI at a much lower concentration (100 nM) (Figures 8B and 9) and yielded a lower EC₅₀ value for inhibition (60.4), compared with that for morphine (2033) (see Table 1). Thus, DAMGO was more potent than morphine in reducing inhibition. Furthermore, the mean EC₅₀ value of DAMGO for inhibition was more than threefold lower than that for excitability (60.4 vs. 209), whereas morphine was less potent in reducing inhibition than in increasing excitability (mean EC₅₀ values, 2033 and 517 respectively).

In addition to the changes in the excitability indexes derived from the input/output curve, both morphine and DAMGO at relatively high concentrations induced secondary PSs following a primary PS (Figure 7C, D-d and Figure 8C). This is in line with previous observations that opioid agonists induce secondary PSs in hippocampal slices (Dingledine, 1981; Swearengen and Chavkin, 1987; Watson and Lanthorn, 1993). DAMGO was more effective than morphine in generating secondary spikes as its maximal effect was more than threefold greater than that of morphine (Figure 9); this effect was produced at a lower concentration (1 μM vs. 10 μM). In addition to the above, DAMGO began inducing secondary spikes at a lower concentration than morphine (100 nM vs. 1 μM). This difference in potency was also expressed in the considerably lower EC₅₀ value for DAMGO (372.5) when compared with morphine (1578). In keeping with previous reports (Dingledine, 1981), we found that the secondary spikes were completely blocked by naloxone (Figures 7D-e and 8C-b). We also found that the secondary spikes induced by the MOR agonists were partly blocked by an antagonist of NMDA receptor CPP (Figure 8C-c), as previously reported (Swearengen and Chavkin, 1987). The remaining component of secondary spikes that was not blocked by CPP was abolished by naloxone (Figure 8C-d). It has been previously shown that morphine reduces the post-spike afterhyperpolarization (Faulkner et al., 1999). This reduction may lead to increased excitability in pyramidal cells and the appearance of multiple spikes (Azouz et al., 1996). Consequently, the greater ability of DAMGO to induce secondary spikes may reflect a greater effect of this drug in enhancing NMDA receptor activity and/or reducing post-spike afterhyperpolarization when compared with morphine.

The difference in the effects of morphine and DAMGO on excitability and inhibition suggests that their mechanisms of action on the local neuronal circuit are not identical. Morphine and DAMGO have high affinity for MORs but they also bind to δ, and κ opioid receptors (Mansour et al., 1987; Waldhoer et al., 2004; Crooks et al., 2006). M, δ and κ opioid receptors can be found in the CA1 field of the ventral hippocampus (Mansour et al., 1987), the part of the hippocampus that has been used in the current study. Because these opioid receptors elicit similar cellular events (Pan, 1998), the differences of the drug potencies observed here could be interpreted as differences in the affinity/efficacy of the drugs for the various opioid receptors (Zernig et al., 1995; Crooks et al., 2006), as well as different expression of the receptors from physiologically different cell types (Pan, 1998). DAMGO's higher potency to suppress inhibition, compared with morphine, could be accounted for by its higher affinity and efficacy for MORs (Chaturvedi et al., 2000; Crooks et al., 2006) and possibly for the various MOR subtypes (Pasternak, 2001).

Interpretation of drug effects – implications for the mechanisms underlying SWRs

The similar effects of MOR agonists on the amplitude of sharp waves, the initiation of episodes and the incidence of sequences suggested a common mechanism of action. Because changes were observed at concentrations as low as those of endogenous μ-opioids in the rat brain (Lam et al., 2008), these changes might reflect the modulation of SWRs by endogenous opioids. In the hippocampus, MORs are almost exclusively located on GABAergic inhibitory interneurons (Drake and Milner, 2002) and opioid agonists modulate inhibition (Nicoll et al., 1980; Lupica, 1995) and increase the excitability of the pyramidal cells (Zieglgansberger et al., 1979; Masukawa and Prince, 1982; Swearengen and Chavkin, 1987). In CA1, opioids suppress synaptic inhibition by hyperpolarizing GABAergic interneurons especially those that project to the perisomatic domain of principal neurons (Madison and Nicoll, 1988; McQuiston, 2008). In keeping with these observations, we found that MOR agonists enhanced network excitability and decreased inhibition. However, as those effects on SWRs produced by low opioid concentrations (<100 nM) were observed in the absence of change on the overall neuronal network excitability and inhibition, we would propose that the modulation of SWRs requires relatively small and specific alterations in the GABAergic transmission which cannot be easily detected by stimulation that engages a rather large part of the local network. The different potencies observed between the three MOR agonists might result from the different affinities/efficacies of the three compounds for the MORs (Zernig et al., 1995; Crooks et al., 2006).

The reduction in the occurrence of SWR episodes is consistent with the suppressive effect of MOR agonists on the activity of GABAergic interneurons, and recent findings suggest that the initiation of episodes of SWRs is favoured by slight enhancement of GABA_A receptor-mediated transmission (Ellender et al., 2010; Koniaris et al., 2011).

The generation of the ripple oscillation requires an accurate balance between excitation and inhibition in the local network. Interestingly, in the present study MOR agonists suppressed ripple oscillation at the same concentration (100 nM) that they disturbed the excitation/inhibition balance. This suppressive effect is consistent with the selective localization of MORs on PV cells (Drake and Milner, 2002) and the critical role these cells play in the generation of ripples (Klausberger et al., 2004). In addition to the strong activation of PV cells, recent observations suggest that ripples also require a rather mild increase in principal cell excitation (Racz et al., 2009; Koniaris et al., 2011; Maier et al., 2011). This may be related to our finding that low concentrations of morphine enhanced ripples. We found that DAMGO suppressed inhibition more readily than it increased excitability. Morphine displayed a much lower potency in reducing inhibition than in enhancing excitability. That is, at relatively low concentrations morphine (but not DAMGO) changed network excitability inducing a shift of the excitation/inhibition balance towards excitation. Although the effect of 5 nM morphine on the evoked responses was not statistically significant, it is plausible that small changes in the excitation/inhibition balance may also occur in the nanomolar range leading to moderate increase of the excitation in the local network and consequently to augmentation of ripples.

The enhancement of clusters by MOR agonists is consistent with their effect on excitation/inhibition balance and agrees with previous observations showing that even small enhancements in GABA_A receptor-mediated transmission suppress clusters (Papatheodoropoulos and Koniaris, 2011). The generation of sequences requires activation of NMDA receptors (Papatheodoropoulos, 2010) whose activity can be uncovered by MORs (Swearengen and Chavkin, 1987; McQuiston, 2008). In the present study, DAMGO produced a marked increase in the probability of cluster occurrence at the same concentration (1 μM) that elicited NMDA-dependent secondary spikes. Thus, network disinhibition and NMDA receptor activation resulting from the increased activation of MORs is a condition that favours the generation of clusters.

We found that all three MOR agonists increased the amplitude of sharp waves. In the CA1 area, sharp waves correspond to GABA_A receptor-mediated IPSPs (Papatheodoropoulos and Kostopoulos, 2002; Wu et al., 2002; Maier et al., 2003) and their amplitude is correlated with the amplitude of IPSPs on pyramidal neurons (Papatheodoropoulos, 2010). In view of the suppressive effect of MOR agonists on IPSPs (Nicoll et al., 1980; Masukawa and Prince, 1982; Capogna et al., 1993; McQuiston, 2008), these agonists should suppress rather than increase the amplitude of sharp waves in CA1. However, although MOR agonists act mainly on PV cells (Drake and Milner, 2002), the latter are not the only contributors to the generation of SWRs. Bistratified cells are also involved in SWRs (Klausberger et al., 2004). Additionally, SWRs in CA1 are triggered by excitatory input from CA3 (Csicsvari et al., 2000) and sharp waves in CA3 pyramidal cells correspond to excitatory rather than inhibitory synaptic activity (Wu et al., 2002; Colgin et al., 2004; Behrens et al., 2005). Thus, disinhibition of CA3 pyramidal cells by MOR agonists could lead to higher excitation of their targets in CA1 including pyramidal cells (Dingledine, 1981; McQuiston, 2008) and inhibitory bistratified cells (Freund and Buzsaki, 1996). Consequently, the enhancing effect of MOR agonists on the amplitude of sharp waves in CA1 could also be explained by a higher activity of bistratified inhibitory cells in this field, following their higher activation from disinhibited CA3 principal cells.

Recent findings show that MORs readily control the activity of ivy and neurogliaform GABAergic cells in the CA1 area (Krook-Magnuson et al., 2011). However, their contribution to the modulation of SWRs is uncertain because ivy and neurogliaform cells do not change their firing activity during SWRs (Fuentealba et al., 2008a).

Implications for the drug effects on memory

The effects of opioids on learning and memory appear diverse because of the variety of opioid receptors (Pan, 1998; Jamot et al., 2003), the learning task and animal species used (Morris and Johnston, 1995), as well as the interaction of opioids with other neurotransmitter/neuromodulator systems (McGaugh, 2002; Hadjiconstantinou and Neff, 2011). A consistent pattern of effects is disclosed when the time of drug administration is taken into account; morphine impairs memory when present during encoding (Castellano, 1975) but facilitates memory when administered after the training/learning sessions (White et al., 1978; Mondadori and Waser, 1979; Staubli and Huston, 1980). Furthermore, memory is more consistently facilitated by intrahippocampal/intraventricular (Meilandt et al., 2004; Hunsaker et al., 2007) than peripheral drug injections (Izquierdo, 1979; Veselis et al., 1994). The facilitatory effect may represent drug effect on the memory consolidation, a process thought to encompass hippocampo-neocortical information transfer through the activity of SWRs (Chrobak and Buzsaki, 1994; Wierzynski et al., 2009). SWRs might arise from an assembly of hippocampal cells interconnected by synapses that have been strengthened during encoding of new information, and such an assembly may underlie a particular memory trace in the hippocampus (Buzsaki, 1989). The SWR-associated, offline reactivation of this assembly may then contribute to further strengthening of hippocampal and neocortical synapses, assisting memory consolidation (Buzsaki, 1989). Recent evidence showing that synaptic strengthening accompanies the emergence (Behrens et al., 2005) and potentiation (Papatheodoropoulos, 2010) of SWRs supports this hypothesis. Similarly, the observed increase in amplitude of sharp waves induced by MOR agonists may contribute to synaptic strengthening inside a sharp wave-associated cell assembly assisting the establishment/maintenance of memory trace. Hence, the facilitatory effect of activating MORs on hippocampus-dependent memory may involve long-term synaptic modification. Indeed, activation of MORs facilitates the induction of long-term potentiation in hippocampal synapses (Martinez et al., 2011).

The effect of MOR agonists on the sequences of SWRs might have important consequences on memory function as the former appear to reflect specific aspects of information processing by the hippocampal circuit. Recent observations suggest that sequences may represent compressed activity that corresponds to long paths the animal traverses in space, and thus they may be involved in the memorization of prolonged experiences (Davidson et al., 2009).

It is of note that, although moderate to relatively high concentrations (≥100 nM) of all the MOR agonists tested reduced ripples, low concentrations of morphine (5–10 nM) produced a considerable increase in the ripple oscillation supporting the idea that endogenous opioid levels may positively modulate memory. An increase in ripple quantity and power has been found after learning, and has been correlated with increased hippocampus-dependent memory performance (Ramadan et al., 2009). Additionally, reduced performance was observed during pharmacologically induced reduction of the ripple power (Robbe et al., 2006). Interestingly, low concentrations of the MOR agonists did not alter the total number of SWR events. Stability in the quantity of SWRs has been observed in parallel with long-term increase in the amplitude of sharp waves and frequency of clusters, following θ-burst stimulation (Papatheodoropoulos, 2010). Therefore, the enhancing effects of MOR agonists on SWRs are consistent with the behavioural data that show MOR-induced facilitation of memory consolidation, and they suggest that this effect may entail the modulation of opioids on SWRs. Moreover, the fact that MOR-mediated modulation of SWRs apparently involves only subtle changes in the excitation/inhibition balance is consistent with the hypothesized role of SWRs in the processes of transfer and consolidation of discrete memory traces given that these processes have to be selective and accurate employing small neuronal circuits, and thus keeping fidelity (Sutherland and McNaughton, 2000).

In this study, we have determined the roles played by MOR agonists in the hippocampal activity of SWRs. Morphine, DAMGO and fentanyl significantly modulated the activity enhancing the amplitude and the sequencing of sharp waves and dampening the initiation of distinct episodes. Low-level activation of MORs, not affecting network excitability and recurrent inhibition, produced an augmentation of the ripple oscillation. These results show for the first time that MOR agonists at concentrations comparable to endogenous levels of opioids exerted a dynamic regulation on SWRs, presumably by finely tuning the balance between excitation and inhibition. To some extent, such modulation may underlie the actions of MOR agonists on hippocampus-dependent memory.