Coronal slices of the frontal cortex were prepared from Sprague-Dawley (SD) rats (Orient Bio Co., Ltd., Seongnam, Korea) on days 21–27 after birth, which were raised under standard conditions (23°C ± 1°C, 12/12 h/ light/dark cycle). Animal care and surgical procedures were conducted by the institutional animal care and use committee (IACUC) of the Catholic University of Korea (IACUC approval No. 2019-0024-01) and were consistent with the National Institutes of Health Guidelines for the Care and Use of Laboratory Animals. Animals were sedated with chloral hydrate (400 mg/kg, i.p.). Brains were rapidly isolated after decapitation and immersed in ice-cold artificial cerebrospinal fluid (ACSF) containing (in mM) 125 NaCl, 2.5 KCl, 25 NaHCO₃, 1.25 NaH₂PO₄, 1 CaCl₂, 2 MgSO₄, and 10 D-glucose bubbled with 95% O₂ and 5% CO₂. Coronal slices (300-µm thick) were prepared on a vibratome (Microm HM650V, Walldorf, Germany) and incubated in a submerged slice chamber for 30 min at 36.5°C, and then left at room temperature (22°C–24°C) until the recording was performed. Slices were transferred to the recording chamber and continuously perfused with the same ACSF (bubbling with 95% O₂ and 5% CO₂) except for the addition of 2 mM CaCl₂ and 1 mM MgSO₄ (1–1.5 ml/min).

The whole-cell patch-clamp recording was performed as described previously [7]. Recording electrodes (4–6 MΩ) were pulled from borosilicate glass pipettes (World Precision Instruments, Inc., Sarasota, FL, USA) using a micropipette puller (MODEL P-97; Sutter Instrument Co., Novato, CA, USA). Excitatory postsynaptic potential (EPSP) was recorded with an electrode filled with a solution containing (in mM) 130 K-gluconate, 10 KCl, 4 Mg-ATP, 10 Na₂-phosphocreatine, 0.3 Na₃-GTP, and 10 HEPES (pH 7.25 with KOH). The shape of L2/3 PyNs (~300–350 µm from pia) in the AI area of the PFC (Paxinos and Watson [29]) was visually identified with an infrared-differential interference contrast (IR-DIC) video microscope coupled to an upright microscope (BX50-WI fitted with 40x/0.80 NA immersion objective; Olympus, Tokyo, Japan) and their regular spiking patterns were confirmed. Whole-cell recording signals were acquired using an amplifier (IX2-700; Dagan, Minneapolis, MN, USA) and digitized at 20 kHz (Digidata 1322A; Axon Instruments, Foster City, CA, USA). Using pCLAMP 10.3 software (Axon Instruments), the signals were saved to a computer hard drive (Pentium PC). The data were low-pass filtered at 5 kHz. Membrane potentials were not corrected for ~14 mV calculated junction potential. The typical access resistance was 12–15 MΩ, and the bridge balance set was optimized in the experiment. The input resistance was measured by injecting negative current (–50 pA, 150 ms) into the soma (every 60 sec). Bipolar tungsten electrode (100 µm in diameter, SNE-100; Kopf Instruments, Tujunga, CA, USA) and glass electrode (20-µm tip diameter, filled with ACSF) were located at layer 3 (L3, ~50 µm downward and ~100 µm laterally from the soma) and corresponding L1 (~50 µm from pia), respectively, for the local electrical stimulations. EPSPs were evoked by paired-pulse stimulation (40-ms interval) with a brief square current pulse (0.2-ms duration) using a constant-current isolation unit (A360; World Precision Instruments). The EPSP amplitude changes were monitored at 0.05 Hz. The stimulus intensity was adjusted to obtain stable baseline EPSPs (5–8 mV). The average amplitude of EPSP was measured at 30–40 min after wash-out of serotonin or carbachol (CCh) and compared with that of baseline EPSPs. In some experimental sets, 6,7-dinitroquinoxaline-2,3-dione disodium salt (DNQX, 20 µM) and d-(-)-2-amino-5-phosphonopentanoic acid (D-AP5, 50 µM) were added to ACSF to confirm glutamatergic synaptic transmission. The paired-pulse ratio (PPR) was determined as the amplitude of the second EPSP compared with that of the first EPSP evoked paired-pulse stimulation (40-ms interval). The amplitude of the second EPSP was determined from the single exponential fit of the first EPSP.

The unloading protocol of fluorescence dye, FM1-43 was similar to that described previously [8]. The changes in fluorescence intensity of FM1-43 indicate synaptic vesicle recycling at presynaptic terminals. Synaptic stimulation releases the FM1-43 from presynaptic vesicles resulting in decreased fluorescence intensity of FM1-43 [30]. To load FM1-43 (Molecular Probes, Eugene, OR, USA) into synaptic vesicles, slices were exposed to high-potassium (40 mM) ACSF for 2 min with 8 µM FM1-43. To reduce nonspecific binding of the FM1-43 to tissue, slices were washed with 1 mM ADVASEP-7 for 2 min in ACSF and further perfused with 0.1 mM ADVASEP-7 in ACSF for 50 min. To unload FM1-43, electric stimulation (2 Hz) was delivered to either L1 or L3 for 2 min with glass pipettes or tungsten bipolar electrodes with 2- to 3-fold stimulus intensity for baseline EPSP amplitude. All loading and unloading procedures of FM1-43 were conducted using ACSF containing ionotropic glutamatergic blockers D-AP5 (50 µM) and DNQX (20 µM) to prevent excitotoxic damage. High-potassium ACSF was applied for 5 min at the end of experiments for the background subtraction and normalization of fluorescence signal. Fluorescence signals were acquired on the microscope for whole-cell patch-clamp recording equipped with a monochromator (Polychrome V; TILL Photonics GmbH, Gräfelfing, Germany), a long-pass filter (510 nm), and a CCD camera (Retiga-2000RV; QImaging, Surrey, BC, Canada). Wavelengths for excitation and emission of FM1-43 dye were 480 nm and 532 nm, respectively. Fluorescence images were collected at 5-sec intervals with 10–70 ms light exposure and captured using TILLvisION software (TILL Photonics).

Drugs including 5-Hydroxytryptamine (5-HT), DNQX, 8-hydroxy-2-dipropylaminotetralin hydrobromide (8-OH-DPAT), N-(piperidin-1-yl)-5-(4-iodophenyl)-1-(2,4-dichlorophenyl)-4-methyl-1H-pyrazole-3-carboxamide (AM251), D-AP5, 1-(m-chlorophenyl) biguanide hydrochloride (mCPBG), NAN-190, ketanserin, and 1,2-bis(2-aminophenoxy)ethane-N,N,N',N'-tetraacetic acid (BAPTA) were purchased from Tocris Bioscience (Bristol, UK). ondansetron hydrochloride, α-methyl-5-hydroxytryptamine (α-me-5-HT), CCh, and atropine were purchased from Sigma-Aldrich (St Louis, MO, USA). BAPTA was included in the pipette solution. NAN-190 and AM251 were dissolved in dimethyl sulfoxide (DMSO, 0.1%).

Data are expressed as mean ± standard error. Statistical comparisons were performed using paired or unpaired two-tailed Student’s t-test and two-way ANOVA with Tukey’s multiple comparisons. The level of significance was set at p < 0.05.

Whole-cell patch-clamp recording was performed in L2/3 PyNs (300–400 µm from the pia) of the AI cortex, a part of OFC [14,16], in 3-week-old rats with local electrical stimulation of corresponding L1 and L3. In the primary visual cortex, as reported previously, these local stimulations activated non-overlapping synapses located in distal apical dendrites and basal dendrites, respectively, using unloading experiment of FM1-43 dye [8]. In the present study, we analyzed the spatial profile of unloaded FM1-43 dye by stimulation of L1 or L3 in the OFC (Fig. 1A–C). Stimulation of L1 resulted in an apparent decrease in fluorescence almost limited to L1 (F_{8, 36} = 103.5, p < 0.001), and the stimulation was reduced far less in deeper L2/3 near the soma of PyNs. However, the repeated stimulation of L3 reduced fluorescence near the soma (300–600 µm from the pia). These results demonstrated that local electrical stimulation of L1 and L3 mainly activated synapses in distal apical dendrites and perisomatic dendrites, respectively. This distance profile of stimulation was similar to that of the primary visual cortex [8]. Based on these results, the two stimuli were used to evoke layer-specific activation of synaptic transmission in L2/3 PyNs of the OFC.

Next, we investigated whether local electrical stimulation of L1 and L3 activated non-overlapping array of synapses (Fig. 1D). The amplitude of EPSPs evoked by simultaneous stimulation of L1 and L3 (9.5 ± 0.3 mV, n = 12) was similar to that of the arithmetic sum of each L1-EPSP and L3-EPSP (9.9 ± 0.4 mV, p = 0.107). The peak time (12.1 ± 1.2 ms) and decay time constant (58.8 ± 10.9 ms) of L1-EPSPs were longer than those of L3-EPSPs (6.4 ± 0.4 ms, p < 0.001 and 41.3 ± 5.2 ms, p < 0.05). Consistent with layer-specific activation of axon terminals demonstrated in the imaging experiment, these results indicate that the stimulation of L1 and L3 specifically activated synapses in distal apical dendrites and perisomatic dendrites including basal dendrites, respectively, in L2/3 PyNs of the OFC.

To investigate whether EPSP was mediated by AMPA- and NMDA-type glutamate receptors, we recorded EPSPs in the presence of the AMPA receptor blocker DNQX and the NMDA receptor blocker D-AP5 (Fig. 2). We evoked EPSPs with 5–8 mV amplitude in the present study. Application of D-AP5 (50 µM) did not change the amplitude of EPSPs (L1 stimulation: 100.6% ± 3.9%, n = 6, p = 0.877; L3 stimulation: 98.8% ± 1.3%, n = 5, p = 0.437). Co-application of DNQX (20 µM) with D-AP5 abolished synaptic response (L1 stimulation: 2.6% ± 0.6%, p < 0.001; L3 stimulation: 3.8% ± 0.4%, p < 0.001). These results indicate that EPSPs evoked by local electrical stimulation resulted mostly from the activation of AMPA receptors in L2/3 PyNs of the OFC.

AMPA receptor-mediated EPSPs were recorded by applying paired-pulse electrical stimulation (40-ms interval) of L1 and L3 alternating in a single PyN. During bath application of 5-HT (10 µM, 10 min) the amplitude of L1-EPSPs decreased by ~30% and then recovered to the baseline level 30–40 min after washout (98.6% ± 1.8%, n = 9, p = 0.632) (Fig. 3A, B). Conversely, the amplitude of L3-EPSP decreased from the baseline to 83.0% ± 2.9% (n = 9, p < 0.001). Therefore, serotonergic stimulation induced input-specific LTD activated by L3 stimulation, but not by stimulation. PPR was not changed in L1-EPSP (1.37 ± 0.06 to 1.33 ± 0.06, n = 9, p = 0.163) and L3-EPSP (1.29 ± 0.11 to 1.32 ± 0.10, n = 9, p = 0.561) (Fig. 3B). In a previous study, combining 5-HT application with tetanic stimulation induced ~25% LTD in the PFC [25]. In the present study, we also applied tetanic stimulation at the end of 5-HT treatment, which induced a rather smaller magnitude of LTD (91.8% ± 1.6% from the baseline, n = 6, p < 0.01). To further investigate the activity dependence of LTD, the test stimulation was turned off during the application of 5-HT and for 10 min after washout (Fig. 3A, B) [7,31]. Without synaptic activity during 5-HT application, no LTD was induced in L3-EPSP (99.1% ± 2.4%, n = 6, p = 0.736) or L1-EPSP (99.5% ± 4.86%, n = 6, p = 0.931). These results indicate that 5-HT induced LTD in a layer-specific and an activity-dependent manner. The concentration-dependent response of 5-HT to L3-EPSP yielded EC₅₀ values of 3.57 µM (n = 5–9) (Fig. 3C). Thus, 10 µM concentration of 5-HT was used to investigate the cellular mechanism of L3 LTD in subsequent experiments.

The 5-HT_1A and 5-HT_2A receptors are the most prevalent 5-HT receptor subtypes in PyNs of the PFC [10]. In the following experiment, we investigated the 5-HT receptor subtypes involved in layer-specific serotonergic induction of LTD using specific antagonists and agonists for 5-HT_1A, 5-HT₂ and 5-HT₃ receptors (Fig. 4). The 5-HT_1A receptor agonist 8-OH-DPAT (10 µM) and the 5-HT₃ receptor agonist mCPBG (10 µM) failed to induce LTD in L3-EPSP (100.1% ± 2.5%, n = 6, p = 0.969; 99.0% ± 2.3%, n = 6, p = 0.730; respectively). In contrast, the 5-HT₂ receptor agonist α-me-5-HT (10 µM) mimicked the LTD induction by 5-HT (82.1% ± 3.4%, n = 6, p < 0.01). Both NAN-190 (10 µM), a 5-HT_1A receptor antagonist (87.9% ± 3.6%, n = 5, p < 0.05) and ondansetron (10 µM), a 5-HT₃ receptor antagonist (84.9% ± 4.6%, n = 6, p < 0.05), failed to inhibit LTD induced by 5-HT application, respectively. However, 5-HT induction of L3-LTD is blocked by ketanserin (10 µM), a 5-HT₂ receptor antagonist (98.1% ± 2.5%, n = 7, p = 0.474). Moreover, ketanserin abolished the effect of α-me-5-HT (98.1% ± 2.3%, n = 6, p = 0.444). Neither 5-HT receptor agonists nor antagonists changed the amplitude of L1-EPSP (Fig. 4A). These results suggest that 5-HT₂ receptors mediate the serotonergic induction of L3-LTD in L2/3 PyNs of the OFC.

We next investigated the involvement of intracellular Ca²⁺ and NMDA receptors in LTD induction. Intracellular application of BAPTA (10 mM), a Ca²⁺ chelating agent, inhibited L3-LTD induced by 5-HT (99.2% ± 1.8%, n = 6, p = 0.681) while L1-EPSP was not changed (100.6% ± 3.0%, n = 6, p = 0.830) (Fig. 5A). NMDA receptor antagonist D-AP5 (50 µM) inhibited L3-LTD (98.6% ± 2.1%, n = 6, p = 0.547), but not L1-EPSP (98.2% ± 1.0%, n = 5, p = 0.157) (Fig. 5B). These results suggest that the increase in intracellular Ca²⁺ via NMDA receptors might mediate 5-HT induction of L3-LTD. We also determined whether endocannabinoids (eCBs) were involved in the serotonergic induction of LTD in the OFC, since eCBs generated by the increase in intracellular Ca²⁺ mediated muscarinic LTD in perisomatic synapses of the visual cortex [8]. Application of the cannabinoid (CB) receptor antagonist AM251 (5 µM) failed to inhibit L3-LTD (83.7% ± 4.5%, n = 6, p < 0.05), while the amplitude of L1-EPSP was not changed (97.5% ± 3.4%, n = 5, p = 0.517) (Fig. 5C). These data suggest that eCB signaling played no role in serotonergic LTD in L2/3 PyNs of the OFC.

CCh (10 µM, 10 min) was bath-applied to induce cholinergic LTD and compared with our previous results in the primary visual cortex [7,8]. CCh decreased the amplitude of the baseline amplitude of L1-EPSP to 81.1% ± 3.4% (n = 5, p < 0.01) and L3-EPSP to 86.8% ± 2.5% (n = 6, p < 0.01), respectively (Fig. 6). PPR in L1-LTD (1.11 ± 0.01 to 1.23 ± 0.06, n = 5, p = 0.091) and in L3-LTD (1.04 ± 0.09 to 1.13 ± 0.10, n = 6, p = 0.210) remained unchanged, indicating postsynaptic form (Fig. 6B). The muscarinic ACh receptor (mAChR) blocker atropine (10 µM) in the bath solution blocked the induction of L1-LTD (97.4% ± 2.8% of the baseline, n = 7, p = 0.389) and L3-LTD (98.7% ± 1.4% of the baseline, n = 5, p = 0.454). Therefore, the activation of ACh receptors induced muscarinic LTD in both synaptic inputs on distal apical and perisomatic dendrites in L2/3 PyN of the OFC, similar to the visual cortex. However, contrary to layer-specific serotonergic induction of LTD, cholinergic stimulation induced LTD in both inputs in the OFC.

In L2/3 PyNs of the PFC, distal apical dendrites in L1 receive nonspecific thalamocortical inputs whereas the perisomatic dendritic area including basal dendrites in L2/3 receive various synaptic inputs from other cortex, striatum, amygdala, and hippocampus [17-21]. Thus, L2/3 PyNs are well suited for the integration of these diverse synaptic inputs. In the present study, we used local electrical stimulation at L1 and L3 to activate synapses located in distal apical and perisomatic dendrites, respectively. Although we demonstrated that local electrical stimulation activated nearby dendritic synapse in the primary visual cortex in our previous study [8], it was not clear whether this experimental configuration could be applied in the PFC. In the present study, we demonstrated that local electrical stimulation of L1 and L3 activated synapses located in distal apical and perisomatic dendritic areas, respectively, using FM1-43 dye unloading from axon terminals (Fig. 1A–C). Consistently, rapid dynamics of L3-EPSP than L1-EPSP coincided with the distance of synaptic location of respective dendrites from the soma. Furthermore, we showed that L1-EPSP and L3-EPSP were independently evoked by the corresponding stimulation (Fig. 1D). Thus, we established that local electrical stimulation of L1 and L3 independently activated synapses located in distal apical and perisomatic dendrites, respectively, in L2/3 PyNs of the PFC. Inputs to L2/3 PyNs from other brain areas are segregated and terminated onto different dendritic compartments [1,32]. Various active membrane properties are specific to dendritic locations and compartments [33]. Thus, the local electrical stimulation could be an efficient experimental technique to investigate the integration properties of the PFC as well as the primary visual cortex.

In the PFC, 5-HT is involved in various forms of long-term synaptic plasticity. LTP in hippocampo-mPFC was augmented in 5-HT-depleted rats [28]. Combining 5-HT treatment with tetanic stimulation induced LTD in layers 5/6 PyNs of the PFC [25]. A 20-min exposure of 5-HT has been reported to induce LTD via activation of 5-HT_2A receptors in L5 PyNs of the mPFC [34]. In our previous study, 5-HT inhibited LTP induced by theta-burst stimulation and LTD induced by low-frequency stimulation during the critical period in L2/3 PyNs of the primary visual cortex [35,36]. However, 5-HT reinstated the induction of NMDA receptor-dependent LTP in 8- to 10-week-old rats [37]. In the cat visual cortex, 5-HT promotes the probability of LTP in 5-HT_2C receptor-rich zones and facilitates LTD induction in 5-HT_2C receptor-poor zones [38]. Thus, 5-HT induces LTP and LTD in a complex manner, depending on the different 5-HT receptor subtypes, brain regions and developmental stages [24,39,40]. In the present study, we demonstrated that a 10-min application of 5-HT (10–30 µM) induced LTD with a 17% decrease in amplitude in the OFC. In a previous study, 5-HT (40 µM) alone reduced excitatory postsynaptic current (EPSC) amplitude in L5 PyNs of the PFC less than 10%, whereas 5-HT treatment combed with tetanic stimulation decreased EPSC amplitude by 25% [25]. In the present study, we also applied tetanic stimulation combined with 5-HT (10 µM), although a smaller amplitude LTD (91.8%) was obtained. Since 5-HT expression varies in receptor subtypes and density depending on cortical areas and layers [40-42], the difference might be attributed to the different areas (mPFC vs. OFC) and layers (L2/3 vs. L5 PyNs).

Among the 14 different 5-HT receptor subtypes, 5-HT_1A, 5-HT_2A and 5-HT₃ are the most abundantly expressed receptors in the PFC. 5-HT₂ receptors are coupled to phospholipase C (PLC) and increase intracellular Ca²⁺. The 5-HT_2A/C receptors are highly expressed at the proximal dendrites including the soma according to mRNA expression and immunohistochemical localization [40,42]. In previous studies, 5-HT₂ receptor-induced LTD in L5 PyNs of the mPFC was dependent on metabotropic glutamate receptors, in synapses activated by stimulation of either L5 [25] or L1 [34]. Spike timing-dependent LTD (t-LTD) in L1 synapses of L5 PyNs of the mPFC was induced by the activation of presynaptic NMDA receptors upon the activation of presynaptic 5-HT_2A receptors [43]. In the present study, 5-HT-induced LTD was mediated by 5-HT₂ receptor activation in L2/3 PyNs of the OFC. The 5-HT-induced LTD was NMDA receptor-dependent. We demonstrated that NMDA receptors did not contribute to the peak EPSP amplitude of 5–8 mV upon stimulation of L1 and L3. However, L3-EPSPs showed faster decay kinetics in the presence of D-AP5 in some PyNs (Fig. 2). Since the activation kinetics of NMDA receptors is slower than that of AMPA receptors [44], the small number of NMDA receptors might be activated in perisomatic dendrites by L3 stimulation. Furthermore, the local EPSP amplitude of distal synapses might be large enough to activate NMDA receptors especially in spine heads [45,46]. Moreover, silencing synaptic activation during 5-HT application failed to induce LTD. Thus, the co-activation of 5-HT₂ receptors and NMDA receptors is critical for the induction of LTD in OFC L2/3 PyNs. The discrepancy between previous studies in the mPFC and the present study might be attributed to the different cortical areas and layers [47].

In the current study, chelating intracellular Ca²⁺ with BAPTA blocked L3-LTD. Influx of extracellular Ca²⁺ from NMDA receptors under 5-HT₂ receptor activation could be the source of increased intracellular Ca²⁺. Otherwise, PLC activation by 5-HT₂ receptor-coupled Gq protein could increase intracellular Ca²⁺ via release from IP₃-sensitive stores. In the primary visual cortex, 5-HT₂ receptor activation triggered Ca²⁺ release in the soma and the IP₃-mediated facilitation of inhibitory currents was restricted to the perisomatic area, but not L1 [9]. Thus, this previous findings suggest that Ca²⁺ release from IP₃-sensitive stores via activation of 5-HT₂ receptors might lead to the induction of L3-LTD in the OFC.

Cholinergic regulation of synaptic plasticity in the brain is associated with essential brain functions such as information processing, learning, memory, and attention [31,48,49]. LTD at glutamatergic synapses of L5 PyNs was mediated by mAChRs in the adult mPFC [26,27]. In the present study, in contrast to layer-specific 5-HT induction of LTD, muscarinic stimulation induced LTD in both inputs activated by L1 and L3 stimulation in L2/3 PyNs of the OFC. The magnitude of LTD (~19%) was similar to that of L3-LTD induced by 5-HT application. In our previous study, the magnitude of muscarinic-induced LTD in perisomatic inputs of the primary visual cortex was ~25% [8]. Thus, these findings suggest that the low magnitude of LTD might be an intrinsic property of L2/3 PyNs in the OFC. Although muscarinic and serotonergic stimulation induced LTD, they showed different layer specificity. Thus, these results suggest that different neuromodulators play a role in the induction of long-term synaptic plasticity depending on their own layer-specific patterns. Differential expression of receptors, ion channels and intracellular signaling molecules between dendritic compartments could contribute to the variable effect of neuromodulators.

Cortical PyNs extend their dendrites to the other layers and integrate the inputs from other brain structures in different layers. The L2/3 and L5 PyNs carry multiple perisomatic dendrites including basal dendrites around the soma and a single apical dendrite extending up to L1 as dendritic tufts [2,33]. These dendrites integrate synaptic inputs as compartments with characteristic membrane distribution of ion channels and intracellular Ca²⁺ release [7,33]. In L2/3 PyNs of the primary sensory cortex, top-down associative inputs from other cortical areas and bottom-up sensory inputs from the thalamus terminate mainly in distal apical dendrites and basal dendrites, respectively [32]. Likewise, in the PFC, feedback connections from other cortical areas as well as the nonspecific thalamus target dendrites in L1 while inputs from amygdala and hippocampus and feedforward cortical inputs terminate in perisomatic dendrites [18,20,21]. In the mPFC, L1 even receives thalamocortical projection and corticocortical inputs in a sublaminar fashion [19]. Thus, the layer-specific nature of synaptic transmission and plasticity of cortical PyNs is important in understanding the cortical integration of information. It is suggested that neuromodulators control synaptic transmission in a pathway-specific manner depending on brain states such as attention, sleep and mood [50]. We previously demonstrated layer-specific muscarinic and serotonergic modulation of synaptic transmission in the primary visual cortex [7-9]. In the present study, 5-HT induced LTD in perisomatic inputs but not in distal apical inputs located in L1 of the OFC while muscarinic stimulation induced LTD in both inputs. This layer-specific neuromodulation may represent an important mechanism of neuromodulatory control of information processing depending on the brain state.