We investigated the afferents and intracortical synaptic organization of the anterior cingulate cortex (ACC) during noxious electrical stimulation. Extracellular field potentials were recorded simultaneously from 16 electrodes spanning all layers of the ACC in male Sprague–Dawley rats anesthetized by halothane inhalation. Laminar-specific transmembrane currents were calculated with the current source density analysis method. Two major groups of evoked sink currents were identified: an early group (latency = 54.04 ± 2.12 ms; 0.63 ± 0.07 mV/mm2) in layers V–VI and a more intense late group (latency = 80.07 ± 4.85 ms; 2.16 ± 0.22 mV/mm2) in layer II/III and layer V. Multiunit activities were evoked mainly in layer V and deep layer II/III with latencies similar to that of the early and late sink groups. The evoked EPSP latencies of pyramidal neurons in layers II/III and V related closely with the sink currents. The sink currents were inhibited by intracortical injection of CNQX (1 mM, 1 μl), a glutaminergic receptor antagonist, and enhanced by intraperitoneal (5 mg/kg) and intracortical (10 μg/μl, 1 μl) injection of morphine, a μ-opioid receptor agonist. Paired-pulse depression was observed with interpulse intervals of 50 to 1,000 ms. High-frequency stimulation (100 Hz, 11 pulses) enhanced evoked responses in the ACC and evoked medial thalamic (MT) unit activities. MT lesions blocked evoked responses in the ACC. Our results demonstrated that two distinct synaptic circuits in the ACC were activated by noxious stimuli and that the MT is the major thalamic relay that transmits nociceptive information to the ACC.
Pain is a multidimensional experience including sensory-discriminative, affective, motivational, and evaluative components. Neuroimaging studies in humans have shown that the anterior cingulate cortex (ACC) is consistently activated by painful stimuli (Alkire et al. 2004; Bantick et al. 2002; Brooks et al. 2002; Casey 1999; Casey et al. 2001; Coghill et al. 1999; Craig et al. 1996; Davis et al. 1997; Derbyshire et al. 1998; Ibinson et al. 2004; Kwan et al. 2000; Ploner et al. 2002; Talbot et al. 1991; Vogt et al. 1996). As experimental subjects became more aware of unpleasant feelings increased responses in the ACC were observed; furthermore, hypnotic suggestions could influence ACC responses to external painful stimulation (Kulkarni et al. 2005; Rainville et al. 1997, 1999; Tolle et al. 1999). Emotional and personality changes were previously reported in patients that underwent cingulotomy for chronic intractable pain (Cohen et al. 2001). Thus it appears that the ACC may be involved in the emotional or affective components of multidimensional pain frameworks (Price 2000; Rainville 2002; Treede et al. 1999; Vogt 2005). Nociceptive ACC neurons recorded from human, monkey, and rat were not organized somatotopically, had large receptive fields, and responded to noxious mechanical or heat stimulation on both sides of the body (Hutchison et al. 1999; Kuo and Yen 2005; Wang et al. 2003; Yamamura et al. 1996). These data provided a neurophysiological basis for the view that the ACC may serve an affective function in pain processing.
Somatosensory-evoked population excitatory postsynaptic potentials (EPSPs) or unit activities in the ACC after electrical or laser-pulse stimulation of the hind paw were previously investigated (Kuo and Yen 2005; Wang et al. 2004; Wei and Zhuo 2001). The evoked field potential consists of a sequence of positive and negative components. These distinct and reproducible responses in the ACC are likely important for mediating the affective components associated with nociceptive sensory stimulus. There is not yet a clear consensus for how extracellularly recorded evoked field potentials or unit activities should be interpreted regarding localization of intracortical components. Negative extracellular field potentials may result from extracellular current flow in the vicinity of synaptic activation sites. Anatomical studies have shown that specific medial thalamic (MT) afferent terminals form asymmetric synapses with dendritic spines arising from the apical dendrites of pyramidal cells whose somata reside in layers III and V of the prefrontal cortex (Kuroda et al. 1998). Our previous examination of medial dorsal thalamic projections revealed a high density of terminals in layers II and III of the ACC (Wang and Shyu 2004). Functional columnar organization in the ACC is derived from dendritic bundles of pyramidal neurons in layers II/III, V, and VI of medial prefrontal cortex (Gabbott et al. 2005). The modular organization of cingulate cortical architecture is thought to be associated with the initial stages of cortical processing of afferent signals (Gabbott et al. 2003). Peripheral nociceptive signals to the ACC are relayed principally through MT nuclei (Kuroda et al. 1998; Marini et al. 1996; Sikes and Vogt 1992; Treede et al. 1999; Vogt 2005; Vogt et al. 1993).
Although anatomical connectivity from the MT to the ACC has been firmly established (Berendse and Groenewegen 1991; Groenewegen 1988; Krettek and Price 1977; Kuroda et al. 1998; Wang and Shyu 2004), there is still a lack of precise information about the microcircuitry within each cingulate cortical column. Current source density (CSD) analyses ofspatially recorded field potentials may reveal the laminar location of current sinks—putative synapse locations—underlying somatosensory-evoked responses; such data would be helpful in elucidating the precise pattern and sequence of intracortical synaptic events underlying evoked field potential waveforms in the ACC. The use of a microelectrode array with 16 recording channels distributed among all of the cortical layers allows field potentials from all channels to be recorded simultaneously. This technique also helps to avoid field potential variation between channels arising from sampling bias. To record evoked field potentials consistently with the same latency, it is crucial to use a reliable method for activation of noxious inputs. Thus we adopted a previously established method for electrically exciting nociceptive afferents in the sciatic nerve (Chang and Shyu 2001).
The aim of the present study was to investigate the functional synaptic organization in the ACC by analyzing responses evoked by noxious electrical stimulation of the sciatic nerve in vivo. Once anesthetized, rats were subjected to the following different experimental procedures. Intracortical extracellular field potentials were first recorded in the S1 and the ACC simultaneously by multichannel electrodes and were analyzed by the CSD method. The optimal intensity for ACC activation was determined by evaluating responses to graded peripheral stimuli. The recording angle across the ACC subareas specific for obtaining the CSD profiles was also established during this first experiment. We then proceeded to investigate intra-ACC neuronal circuit activity and plasticity in response to noxious stimulation by examining the spatiotemporal distribution of evoked CSD patterns, multiunit activities, and intracellularly recorded responses after application of the noxious stimulus. To examine the plastic changes in the ACC, CSD profiles and MT unit activities were recorded simultaneously after paired and high-frequency pulses applied to the sciatic nerve. Finally, we applied pharmacological agents or electrical lesion methods to elucidate the factors that influence evoked synaptic activation in the ACC and to determine the contribution of MT afferents to ACC circuit activation.
Preparation of animals
Male Sprague–Dawley rats (300–400 g) were housed in an air-conditioned room (21–23°C, humidity 50%, 12-h light/dark cycle starting at 08:00 h) with free access to food and water. All experiments were carried out in accordance with the guidelines of the Academia Sinica Institutional Animal Care and Utilization Committee. Rats were initially anesthetized with 4% halothane (in pure O2) in an acrylic box. A PE-240 tube was inserted by tracheotomy and EMLA Cream (lidocaine 2.5% and prilocaine 2.5% cream; AstraZeneca, Mölndal, Sweden) was smeared over the wound. Animals were then placed in a stereotaxic apparatus and maintained under anesthesia with 2% halothane in 30%/70% nitrous oxide/oxygen during the surgery. Rats' body temperatures were maintained at ≥36.5°C with a homeothermic blanket system (Model 50–7079, Harvard Apparatus, Holliston, MA). Craniotomies were performed over the two target areas. To expose the ACC, a craniotomy centered 2.5 mm anterior to the bregma and 1 mm lateral of the midline was performed; to expose the primary somatosensory cortex (S1), a craniotomy centered 1 mm posterior to the bregma and 3 mm lateral of the midline was performed. A small piece of the dura over each target area was carefully removed and warm paraffin was applied to keep the cortical surface moist. Rats' heart rates were monitored by electrocardiography. Animals were subsequently kept anesthetized with 0.75–1.5% halothane in a nitrous oxide/oxygen mixture (10%/90%) during the recording session. Depth of anesthesia was checked and maintained periodically by pinching the tail and confirming that no overt reflexive movements or heart accelerations were observed.
Electrical stimulation of sciatic nerve
The sciatic nerves were exposed bilaterally and then a custom-made stainless steel cuff electrode was attached to each nerve. Biphasic electrical current (0.03–10 mA, 0.5-ms duration, 0.1 Hz) was delivered by an isolated pulse stimulator (Model 2100, A-M Systems, Carlsborg, WA). The anode electrode was placed about 1 cm distal to the cathode electrode. Minimal intensity for inducing S1 responses was regarded as the threshold value. Stimulation was delivered at intensities that were multiples of the threshold current value. We previously showed that innocuous A-β fibers are excited by twice the intensity at which muscle twitches are elicited and that nociceptive A-δ and C fibers are recruited and excited by 10 × and 20 × threshold, respectively (Chang and Shyu 2001).
Electrical stimulation of the medial thalamus
The stimulation procedures used were similar to those described previously (Kung and Shyu 2002). Monopolar tungsten electrodes (A-M Systems) with an impedance of 12 MΩ were placed in the MT. Most of the thalamic units that responded to peripheral stimulation were detected by electrodes placed about 2.5 mm posterior to bregma, 1 mm lateral of the midline, and 5 mm below the dura. Square-wave constant-current pulses (0.2 ms, 100–300 μA) were delivered to the MT from an isolated pulse stimulator (Model 2100, A-M Systems). For paired-pulse stimulation, the interpulse interval was set to 80–120 ms.
A 2-mm-diameter stainless steel rod was attached to the voice coil of an 8-Ω, 15-W loudspeaker. Tactile stimuli were applied to the hind paw triggered by a transistor–transistor logic pulse from a pulse generator (Model 2100, A-M Systems). The movement of the coil transmitted to the rod produced an outward excursion of 5 mm. The pressure exerted by the protrusion of the rod onto the paw surface was measured by a transducer indicator (Model 1601C, IITC Life Science, Woodland Hills, CA). The exerted force was 1.2 g when a 10-V, 1-ms square wave was applied to the coil. This pressure did not produce any pain sensation when applied to the experimenter's fingertip.
Recording of evoked multichannel field potentials and unit activities
A Michigan probe with 16 contact points (150-μm-interval spacing) was used to record the extracellular field potentials in the left ACC (about 2.5 mm anterior and 1 mm lateral to bregma; probe inserted 40° from the vertical line). Another Michigan probe was used to record extracellular field potentials in the hind paw projection area in the left S1 (about 1 mm posterior and 3 mm lateral to bregma; probe inserted perpendicular to the cortical surface). An Ag–AgCl reference electrode was placed in the nasal cavity. The sampling rate of recorded analog signals was 6 kHz and the data were processed in a multichannel data acquisition system (TDT, Alachua, FL) for a PC.
Multiunit activities of the evoked responses were obtained by filtering the multichannel field potentials (200–3,000 Hz). The square root of the mean basal activity over a 100-ms period before stimulation was determined, multiplied by 4, and the product was set as the spike threshold value. Spikes that exceeded the threshold were marked as the digital signal “1”. Subthreshold signals were marked as the digital signal “0”. Twenty multichannel multiunit activity trials were summated and were presented in a poststimulus histogram. A contour plot of multichannel multiunit activities was constructed to present the spatial distribution of unit activities across the cortical layers.
Current source density analysis
We used CSD analysis to determine the underlying current generators of the evoked field potentials (Freeman and Nicholson 1975; Mitzdorf 1985). The ACC is dominated by parallel-aligned pyramidal cells whose apical dendrites extend along the axis across cortical layers. Sciatic nerve stimulation produced synchronous excitation of ensembles of pyramidal cells, which resulted in a major current flow perpendicular to the cortical layers. Under the assumptions of homogeneous cortical activity and constant extracellular electrical conductivity, CSD can be estimated from the second spatial derivative of the recorded field potentials in the axis parallel to the cortical layers. Thus the procedure used to obtain CSD data was to record the field potential at equidistant, linearly positioned electrode contacts using Michigan probe electrodes that vertically penetrated the cortical layers. Field potentials evoked by peripheral stimulation were recorded for 1 s in each channel per trial. Twenty to 40 trials of evoked field potentials were averaged for each channel. With regard to the time span and the sampling variations in each recording session, we adopted a five-point formula (Freeman and Nicholson 1975) to smooth the spatial sampling variability. The membrane current Im was derived from the second spatial derivations of the extracellular field potentials φ and was calculated by the finite-difference formula where h is the distance between successive measuring points (150 μm in the present investigation) and x is the coordinate perpendicular to the cortical layer. The remaining constants were as follows: n = 2, k = 4, a0 = −2, a±1 = 0, and a±2 = 1.
The μ-opioid receptor agonist morphine (5 mg/kg) and the μ-opioid receptor antagonist naloxone (0.7 mg/kg) were administered by intraperitoneal injection. For intracortical injections, a microfilament was inserted vertically into the ACC region (2.5 mm anterior to bregma and 1 mm lateral of the midline). 6-Cyano-7-nitroquinoxaline-2,3-dione (CNQX, 1 mM, 1 μl), an α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA)/kainate glutaminergic receptor antagonist, and morphine (10 μg/μl, 1 μl) were perfused intracortically at a speed of 0.3 μl/min by a perfusion pump (Model 55–2222, Harvard Apparatus).
Electrolytic lesion of medial thalamic nucleus
The location of the MT was functionally identified by recording responses to electrical stimulation of the contralateral sciatic nerve. A Michigan probe was inserted vertically into the MT ipsilateral to the ACC being recorded. Thalamic activities were monitored and examined as the electrode advanced and electrical stimulation was applied to the sciatic nerve. Once the locations of stimulus-responding thalamic unit responses were found, the Michigan probe was withdrawn. A tungsten electrode was reintroduced along the same track and lowered to the depth previously reached by the Michigan probe. To deactivate thalamic responses, a 100-μA, 100-s DC was delivered by the constant-current pulse generator (AM-Systems).
Verification of recording and lesion sites
At the end of the experiment a small lesion was made by passing an anodal current (30 μA for 5 s) to the 16th contact lead, the deepest electrode of the probe. Another lesion was made at the same lead after the Michigan probe was withdrawn 1,000 μm. Rats were fixed by perfusion with normal saline followed by 10% formalin. The brains were cut in 60-μm-thick coronal sections using a cryostat and the sections were stained with cresyl violet. The rat atlas of Paxino and Waston (1998) was used as a reference to estimate the cortical layer structures during histological examination. The positions of the remaining recording sites in the cortex were estimated by determining their distances from the two lesions.
The peak latencies and amplitudes of sink currents evoked by electrical, mechanical, and laser stimuli were determined from CSD data. The changes of sink currents were measured and compared before and after drug applications. The sink amplitudes between groups were analyzed with Student's t-test. One-way or two-way ANOVA was used to analyze the effects of drugs, nitrous oxide, or graded stimuli on the evoked sink currents. Tukey's post hoc tests were used to detect the sources of group differences revealed by the ANOVAs.
Determining laminar specific CSD profiles in the S1 and the ACC
Two Michigan probes were inserted in the S1 and ACC with angles perpendicular to the cortical surface (schematic diagram and photomicrographs in Fig. 1A). The lesion markers (arrowheads in the photomicrographs of S1 and ACC) were used to identify the loci of the recording electrodes relative to the cortical layers. The threshold level at which responses in S1 were observed was used as a reference to grade the stimulus intensity used in the subsequent experiments. CSD values were calculated from averaged field potentials recorded simultaneously in all six layers in the left S1 during 1 × (threshold intensity) and 20 × threshold electrical stimuli applied to the rat right sciatic nerve (Fig. 1B, left). Similar to our previous work, we observed two groups of sinks evoked in S1 (Sun et al. 2006b). The first group (sink 1) was in layer IV and the other (sink 2) was in layer VI. CSD values were calculated from averaged field potentials recorded simultaneously in all layers of the left ACC during 20 × threshold electrical stimulation applied to the rat right sciatic nerve (Fig. 1B, right). The most prominent group of sinks was in the lower part of layer II/III (circled). One small sink current was observed in layer V (white arrowhead) with a shorter latency than that in layer II/III.
Comparison of evoked sink currents in the ACC and S1
We examined responses in the S1 and ACC to stimuli across a range of intensities to evaluate the minimal effective stimulus intensity for ACC activation. The amplitude of sink 1 in S1 increased as electrical stimulation intensity was increased until the maximum response was reached with a stimulation tenfold stronger than the threshold stimulation. Responses were evoked in the ACC by stimulation at tenfold of the intensity of the threshold stimulation level for the S1 and maximal amplitude activation in the ACC required stimulation 20-fold more intense than the S1 threshold level. The evoked sink 1 in S1 was much weaker with ipsilateral stimulation than with contralateral stimulation (Fig. 2A). On the other hand sink currents in the ACC were similar with stimulation of either side (Fig. 2B). It should be noted that the maximal evoked amplitude in the ACC differed from that in the S1 by roughly one order of magnitude. Thus stronger electrical intensity is required to evoke responses in the ACC than in S1. Such high-intensity electrical stimulation may recruit many different sensory modalities. To rule out the possible contribution of innocuous afferent inputs to the ACC, response to a rapid pressure stimulation delivered by a mechanical device was tested. When an innocuous mechanical stimulus was applied to the hind paw, responses in S1 were similar to those induced by 5 × threshold electrical stimulation (Sun et al. 2006b), but no responses were elicited in the ACC by the same mechanical stimulus (n = 3, data not shown).
Effects of nitrous oxide
To determine to what extent the nitrous oxide mixture used for maintaining anesthesia may suppress cortical evoked responses, we compared the amplitudes of evoked responses recorded in the S1 and ACC under varied nitrous oxide concentration in 1% halothane anesthesia. We found that the S1 sink current evoked at different electrical intensities did not change significantly under different nitrous oxide concentrations [F(3,90) = 0.422, P = 0.738, Table 1]. However, stimuli delivered at more than double threshold intensity affected the evoked sink currents in the S1 [F(4,90) = 39.156, P < 0.001, Table 1]. Simultaneously recorded ACC evoked sink currents did differ with changes in nitrous oxide concentration [F(3,84) = 3.857, P = 0.012] or with electrical intensity [F(4,84) = 21.970, P < 0.001]; however, these effects of nitrous oxide concentration arose solely from differences between 0 and 30% nitrous oxide and between 10 and 30% nitrous oxide with stimulus applied at tenfold the threshold level (P < 0.05, Table 2). A significant difference between responses elicited at tenfold the threshold and those elicited at lower intensities when 0% nitrous oxide (P < 0.01) or 10% nitrous oxide (P < 0.05) was applied was observed (Table 2). Thus it appeared that nitrous oxide has no detectable effect on S1 evoked responses and concentration at 10%, as used in our experiments, does not produce a significant suppressive effect on evoked ACC responses after graded electrical stimulation.
Recording angles and cortical subareas specific for obtaining CSD profiles
The Michigan probe was inserted with different angles and in different subareas of the medial prefrontal cortex to evaluate the CSD profiles evoked by peripheral stimuli. The angle chosen for the recording in area 1 of the ACC (Cg1) was perpendicular to the layers and about 40° oblique from the midline. We obtained a typical response with strong sink currents activated in layer II/III (Fig. 3B). When the deeper prelimbic cortex (PrL) was reached by inserting at a wider insertion angle, the responses were much smaller than those in Cg1 (Fig. 3A). No sink/source components were discernible from the probes inserted vertically passing through layer II/III of secondary motor cortex (M2), Cg1, and PrlL (Fig. 3C). Evoked responses in primary motor cortex (M1) were located mainly in layer V (Fig. 3D). This pattern of evoked cortical sink/source currents across subareas was consistent among multiple rats (PrL, n = 6; ACC obliquely, n = 7; ACC vertically, n = 3; and M1, n = 4).
Delineating CSD profiles in the ACC
The CSD results reported herein were obtained from Cg1 with the insertion angle verified histologically. Seven individual CSD-evoked responses in the ACC after 20 × S1 threshold stimulation of the contralateral sciatic nerve are presented in Fig. 4, B–H and reconstructions of the relative positions of the Michigan probe tracks are shown in Fig. 4A. Several prominent features of the CSD profiles were identified in layers II/III and V and these sink components were reproducible across animals. We thus constructed a representative grand average from individual animals to identify individual sinks and sources. Direct averaging of the respective CSD sweeps from different animals was not justifiable because of differences in recording sites with respect to cortical layer and slight variation in the thickness of the cortical layers across animals.
We determined the precise histological location of the electrode positions according to the lesion markers in each animal. There were three predominant sinks that appeared along the border between layer II/III and layer V in all of the animals. Therefore this border was used as the reference for aligning the CSD profiles. The cortical thickness of layers I and II/III was consistent among the seven animals, whereas that of layers V and VI was somewhat variable. In four cases (Fig. 4, B–E) layer V spanned four electrodes, in two cases (Fig. 4, F and G) it covered five electrodes, and in one case (Fig. 4H) it spanned six electrodes. So we chose the four-electrode coverage as the standard thickness of layer V and the extra channels in layer V in the three other cases were not considered, but rather were replaced with neighboring channels in layer VI (Fig. 4, F–H). After readjustment of CSD sweeps, CSDs of the seven animals were averaged on a trace-by-trace basis. CSD sweeps demonstrated little variation (Fig. 5 A, black lines = means, gray lines = means ± SE). All the sinks and sources apparent in the grand average CSD could also be identified in the individual CSDs, demonstrating the high consistency of the CSD response across animals. An early group of sinks (latency = 54.04 ± 2.12 ms, n = 7, Fig. 5A, arrowhead) was located entirely within the deep layers (V–VI) and its earliest component was in layer VI. The second early group of sinks (latency = 80.07 ± 4.85 ms, n = 7, Fig. 5A, arrows) was located in the deeper part of layer II/III and the upper part of layer V and its amplitude was larger than that of the earlier group (2.16 ± 0.22 vs. 0.63 ± 0.07 mv/mm2, n = 7). Two late sink groups located in the lower part of layer VI and the upper part of layer II/III may represent separate intracortical polysynaptic connections to the two early sink groups. The amplitudes and latencies of major sink currents identified in the different layers are given in Table 3.
CSD and multiunit activities in the ACC
Seven examples of multiunit activities in the ACC evoked by 20 × S1 threshold stimulation applied to the sciatic nerve are shown in Fig. 4, B–H (right panels). The data were aligned and adjusted as described for the CSD sweeps. The summation of these multiunit activities is shown in Fig. 5B. The contour plot of multiunit activities superimposed with a color map of the grand average CSD is shown in Fig. 5C. The strongest multiunit activities occurred in upper layer V of the ACC. The white-arrowed dashed lines in the CSD color maps follow the probable depth–time course with respect to the two sink current flows. The largest multiunit activities spanned the entire layer V and the lower part of layer III.
Intracellular recording of cingulate neurons
A total of 21 neurons were recorded in the medial prefrontal cortex and six of them responded to 20 × S1 threshold stimulation (Fig. 6A, top). The mean membrane potential of these six neurons was −57.83 ± 3.52 mV and the mean membrane resistance was 16.41 ± 2.90 mΩ. Two pyramidal neurons recorded in layer II/III and layer V and their camera lucida drawings are illustrated in Fig. 6A, a and b, respectively. Their synaptic responses were recorded simultaneously with multichannel sink currents as shown in Fig. 6B, a and b. The top and middle panels of Fig. 6B, a and b represent, respectively, the evoked CSD pattern and the superimposed plot of the evoked EPSP of the pyramidal neuron. Figure 6B, bottom shows the perievent histogram of the evoked action potentials of this neuron after application of 20 × S1 threshold stimulation to the sciatic nerve. Response latencies of the evoked EPSP ranged from 27.8 to 121.0 ms (67.8 ± 41.3 ms, n = 6). Evoked peak responses in the EPSP sweeps of these two pyramidal neurons in layer II/III and layer V of ACC appeared to match the sink in layer II/III after the noxious electrical stimulation (vertical dashed lines in Fig. 6B).
The conduction velocities of the peripheral afferent fibers mediating the sink currents in the ACC and S1 were measured. Conduction velocity was calculated by dividing the distance between the two stimulation sites (sciatic nerve vs. digits) by the difference of the latencies of sink currents in the ACC evoked by the respective peripheral stimulation sites. The conduction velocity of the layer V sink current in the ACC was 8.69 ± 1.43 m/s (n = 5) and 5.42 ± 0.62 m/s (n = 3) for contralateral and ipsilateral stimulation, respectively. The conduction velocity of the layer II/III sink current in the ACC was 5.25 ± 0.79 m/s (n = 6) and 6.11 ± 0.33 m/s (n = 4) for contralateral and ipsilateral stimulation, respectively. The conduction velocity of the electrically evoked sink 1 current in S1 was 33.96 ± 4.06 m/s (n = 6) for contralateral stimulation. The conduction velocities of the major sinks in the ACC produced by electrical stimulation were in the A-δ fiber range, whereas the conduction velocities of the fast responses in S1 were in the A-β fiber range.
Paired-pulse and high-frequency stimulation
To examine short-term plasticity of evoked intra-ACC responses, paired-pulse and high-frequency titanic stimulation protocols were applied to the sciatic nerve. Two stimuli of identical duration and amplitude were applied with interstimulus intervals varying from 50 to 2,000 ms (Fig. 7B). Paired-pulse depression was observed in ACC layer II/III CSD responses and MT multiunit activities evoked by sciatic nerve stimuli with the second pulses delivered 50 to 1,000 ms after the first response (Fig. 7B, n = 7 for ACC and n = 16 for MT). Delivery of high-frequency stimulation (11 pulses, 100 Hz) to the sciatic nerve potentiated the evoked response of sink currents in layer II/III and layer V (Fig. 7C). There was a strong correlation between MT unit activities and the integrated layer II/III sink currents (Fig. 7C, inset; r = 0.91, n = 14, P < 0.001). Paired-pulse facilitation was observed in the ACC layer II/III CSD responses evoked by direct electrical stimulation of the mediodorsal (MD) nucleus (Fig. 7F; Fig. 7D shows schematic diagram) and significant changes relative to the first response were found with a 50- to 100-ms interpulse interval (Fig. 7, E and G; n = 5). Sink currents in the late phase were appreciably potentiated during high-frequency stimulation (100 Hz, 11 pulses).
Effects of intracortical CNQX
To confirm that the evoked sink currents in layer II/III and layer V of the ACC were elicited by glutamatergic neurotransmission, 1 μl of CNQX (1 mM) was injected into the ACC 300–400 μm from the Michigan probe (Fig. 8A, schematic diagram). The sink currents in layer II/III and layer V were significantly suppressed after injection of CNQX into the cortex and took about 2 h to recover (Fig. 8C, n = 4). Meanwhile, infusion of artificial cerebrospinal fluid alone (n = 3) did not affect the amplitudes of the layer II/III sink currents (Fig. 8B).
Effects of morphine administration
Sink currents in S1 were not affected by intraperitoneal injection of morphine (5 mg/kg) (Fig. 9A, n = 5). On the other hand, systemic morphine increased sink current amplitudes in layer II/III and slightly enhanced sink currents in layer V of the ACC (Fig. 9B, n = 5). Direct infusion of 1 μl morphine (10 μg/μl) into the ACC greatly enhanced the sink currents in layer II/III and slightly enhanced the sink currents in layer V (Fig. 9C, n = 7 for layer II/III and n = 3 for layer V).
The evoked multiunit activities in MT and the effect of MT deactivation
Multiunit activities could be evoked in several nuclei in the MT in response to noxious electrical stimulation. Figure 10A shows 15 tracks of multichannel electrodes and their verified position in the MT. Recording leads that detected evoked multiunit activities are indicated by filled circles and recording leads that did not detect evoked multiunit activities are represented by open circles. The evoked multiunit activities were predominantly recorded in MD. However, we could also detect evoked multiunit activities in the central lateral (CL), paracentral (PC), ventral lateral (VL), and parafascicular (PF) nuclei. The averaged multiunit activities of these respective thalamic nuclei are shown in Fig. 10A.
To evaluate the relative contribution of the MT in relaying nociceptive inputs to the ACC, electrolytic lesions were made. The lesions generally encompassed the MD, CL, PC, VL, and PF nuclei (Fig. 10B). Evoked CSD currents were completely blocked by MT lesions (Fig. 10C; n = 4).
In the present study, we explored the activation of intra-ACC circuits after high-intensity electrical stimulation of the sciatic nerve. Our results were the first to show that CSD analysis performed on evoked laminar field potentials in the ACC revealed a discrete spatiotemporal distribution of the underlying sinks and sources. Interpretation of CSD profiles requires consideration of the local anatomy and the characteristics of multiunit and intracellular responses evoked by the same stimuli. Thalamocingulate afferents from the MD and PF nuclei terminate on the dendrites of cingulate pyramidal neurons in layer II/III and layer V; these synapses may account for the initial evoked sink currents at those depths (Berendse and Groenewegen 1991; Kuroda et al. 1998; Marini et al. 1996). Our intracellular recordings revealed orthodromically driven EPSPs in pyramidal neurons in layers II/III and V that exhibited a poststimulus latency equivalent to that of the evoked sink current latency. Strong multiunit activities were found along the layer III–V borders where a preponderance of action potentials fired and propagated along intracortical pathways. These observations suggest that thalamic inputs excited by intense peripheral electrical stimuli may activate pyramidal neurons of layer II/III and layer V in the ACC, which may in turn polysynaptically activate both superficial and deep-layer intracortical circuitry.
Nociceptive responses in the ACC
At least tenfold of the S1 threshold intensity was required to elicit clear sink/source currents in the ACC. The early group of evoked sink currents in layer II/III of the ACC had a longer latency than that of layer IV evoked sink currents in the S1. Although these differences could be partly the result of central processes relayed by different nuclei in the ascending somatosensory pathways, comparison of the thresholds and conduction velocities of peripheral event signals strongly suggests that the differences originated, at least in part, in the periphery. We previously determined that nociceptive A-δ and C fibers are excited by stimulation 10 × and 20 × the somatosensory threshold intensity, respectively (Chang and Shyu 2001). The early group of evoked sink currents in the present study had conduction velocities in the A-δ fiber range. Intense electrical current applied to the sciatic nerve may excite cutaneous A-δ afferents and group III deep receptor afferents, which are both considered nociceptive (Adreani et al. 1997; Leem et al. 1993). Thus it is very likely that the afferents exciting intra-ACC circuits are nociceptive. Our findings are consistent with previous demonstrations that ACC neurons respond to noxious stimuli (Devinsky et al. 1995; Sikes and Vogt 1992; Vogt et al. 1979; Wei and Zhuo 2001; Zhang et al. 2004). Although the ACC receives bilateral somatic inputs by A-δ afferents, this area apparently lacks input from the A-β afferents found in S1. This suggests the ACC processes somatosensory information that is distinct from that processed in S1.
Sink/source profile and synaptic organization in the ACC
Peripheral noxious electrical stimulation evoked two distinct spatiotemporal patterns of intracortical sink source currents in the ACC. The aggregation of sink currents across different cortical layers is a product of a laminar arrangement of the excitatory synapses. The spatial extent of each sink current and its complementary source current reveals that excited neurons extending their dendritic processes are involved in the successive intracortical excitatory networks. The close approximation of intracellularly recorded pyramidal neuron EPSPs and the sink currents strongly suggest that these early evoked sink currents constitute essentially ensemble neuronal activity (Mitzdorf 1985). Our CSD analysis revealed that early synaptic activations occurred first and separately in layers II/III and V, and were relayed transynaptically to both more superficial and deeper layers. The amplitude of the former is several times greater than that of the latter, reflecting the net strengths of intracortical synaptic connections at the sites of respective sink currents.
The early sink current in layer II/III was accompanied by a more superficial source current. This arrangement suggests that afferent fibers terminate on the spines of vertically oriented ascending dendrites of layer II/III pyramidal neurons, creating local sink currents and a distal source current in their apical dendrites (Castro-Alamancos and Connors 1996; Cauller and Connors 1994; Kuroda et al. 1998). Sinks located in upper layer II/III have successively longer latencies than that of this earlier sink. These sinks likely represent the intracortical synaptic current flows sweeping upward and forming polysynaptic connections.
The earliest sink in layer V, located near the layer V–VI border, has a shorter latency than that in layer II/III and likely reflects the activation by thalamocortical afferent projections of neurons in a deeper layer (Berendse and Groenewegen 1991; Groenewegen 1988; Krettek and Price 1977; Kuroda et al. 1998; Marini et al. 1996). This sink is complemented by sources in the lower part of layer II/III. This sink/source arrangement and the onset latencies of the corresponding currents suggest that a similar configuration of excitatory synaptic actions of layer II/III sinks exists. Nociceptive-specific neurons in ACC layers V and VI were demonstrated in several studies (Sikes and DeFrance 1985; Tsai et al. 2004; Yamamura et al. 1996). Subsequent sink currents in deeper layers have a longer latency, indicating a polysynaptic connection, possibly from layer VI neurons that have axon collaterals that terminate on layer V neurons (Branchereau et al. 1996; Pirot et al. 1996). This recurrent excitation may contribute to the late deep-layer sink current, resulting in a long-duration recurrent excitation in cingulate neurons.
The strong activation of the layer II/III sink current by MT stimulation found in the present and previous studies suggests that sink currents in layer II/III may receive direct excitatory synaptic inputs from the MT (Sikes and DeFrance 1985; Sun et al. 2006a). In concordance, we recently demonstrated with anterograde labeling that MT afferents terminate predominantly in layers II and III of the ACC (Wang and Shyu 2004). Similar studies have demonstrated major thalamic projections to layers II/III and V in this brain region (Berendse and Groenewegen 1991; Groenewegen 1988; Krettek and Price 1977; Kuroda et al. 1998).
Intracortical glutamate synaptic transmission
The observed blockade of intra-ACC sink currents with CNQX strongly indicates that AMPA/kainate glutamate receptors mediate the excitatory drive in thalamic inputs that are presynaptic to the cingulate neurons. Glutamatergic thalamocortical pathways were described previously (Greengard et al. 1991; Hedberg and Stanton 1995; Kirkwood et al. 1993; Wei et al. 1999) and findings from several electrophysiological studies indicate that these thalamic inputs are the primary source of glutamatergic input to cingulate neurons (Gemmell and O'Mara 2002; Gigg et al. 1992; Greengard et al. 1991; Hedberg and Stanton 1995; Kirkwood et al. 1993; Kung and Shyu 2002; Pirot et al. 1996; Wei et al. 1999).
Morphine effects on the ACC
In contrast to our previous finding that systemic morphine preferentially affected a C-fiber–evoked long-latency sink current in S1, here we found an excitatory effect of morphine on the nociceptive-driven sink current in the ACC (Sun et al. 2006b). The excitation of intra-ACC evoked synaptic responses observed in the present study was likely the result of a collective action of morphine on multiple levels of the ascending nociceptive pathway. Nevertheless, the robust excitatory effects of local intra-ACC injection of morphine confirmed that morphine has direct effects within the ACC. The more pronounced effect of systemic morphine relative to locally applied morphine may be explained by the influence of morphine on other brain areas that modulate ACC activity.
Moderate to high levels of μ- and δ-opioid receptors are present in the ACC (Mansour et al. 1987). δ-Opioid–receptor activation hyperpolarizes some pyramidal neurons and also inhibits presynaptically the release of excitatory amino acids and γ-aminobutyric acid (GABA) at synapses on pyramidal neurons (Tanaka and North 1994). Local application of a μ-opioid receptor agonist attenuates cingulate neuronal firing evoked by local application of glutamate (Giacchino and Henriksen 1998). We recently confirmed a previous report showing that morphine treatment attenuates prefrontal neuronal responses evoked by MD thalamic afferents (Giacchino and Henriksen 1998; Sun et al. 2006a). Thus the excitatory effect of morphine on cingulate neurons may be mediated by a different pathway or cellular circuits. Evidence for morphine enhancement of excitatory synaptic transmission through inhibition of GABAergic interneurons was previously found in the hippocampus and the periaqueductal gray (Akaishi et al. 2000; Chiou and Huang 1999; McQuiston and Saggau 2003). Thalamocingulate terminals synapse on GABAergic interneurons as well as principal neurons in the ACC (Kuroda et al. 1998). This arrangement may enable GABAergic interneurons to inhibit cingulate principal neurons by feedforward inhibition; moreover, the presence of GABAergic terminals both pre- and postsynaptic of thalamocingulate synapses may enable disinhibition of the interneurons after the activation of thalamocingulate afferents (Gigg et al. 1992; Kuroda et al. 2004).
Direct evidence of the action of morphine on GABAergic interneurons in the cingulate cortex is still lacking. However, it is noteworthy that nonpyramidal neurons in the rat ACC were hyperpolarized after activation of μ-opioid receptors (Tanaka and North 1994) and some prefrontal neurons have been reported to increase responses to MD thalamic stimulation in the presence of morphine (Giacchino and Henriksen 1998). Thus it is possible that morphine may enhance evoked sink currents in the ACC through inactivation of GABAergic interneurons, which ultimately decreases the level of inhibition on principal neurons.
Activation of opioid receptors in the ACC can produce powerful antinociception effects (LaGraize et al. 2006; Lee et al. 1999). A recent human neuroimaging study showed that analgesia induced by delivery of fentayl, a μ-opioid receptor agonist, was accompanied by an increase of regional cerebral blood flow in the ACC (Casey et al. 2000). The supraspinal antinociceptive action of morphine may be mediated, at least in part, by activation of descending modulatory systems from the ACC to subcortical areas and the spinal cord (Calejesan et al. 2000; Senapati et al. 2005).
Short-term synaptic plasticity
Homosynaptic short-term facilitation was induced by paired-pulse stimulation of MT–ACC afferents. A clear increase in the amplitude of the sink currents in layer II/III was observed. Our results are consistent with previous reports that examined short-term neuronal plasticity in the ACC (Gemmell and O'Mara 2002; Kung and Shyu 2002; Sikes and DeFrance 1985; Sikes and Vogt 1992). The loci of potentiation are likely the initial terminal fields of MT afferents on the apical and basal dendrites of deep-layer pyramidal neurons (Horikawa et al. 1988; Swanson and Cowan 1977; Vogt and Miller 1983). Meanwhile high-frequency stimulation of MT inputs enhanced the sink currents. Our findings strongly indicate that the mono- and polysynaptic circuitries in the ACC are subject to plastic changes in response to the constant inflow of the thalamic inputs. In contrast to the effects exerted by direct activation of thalamic inputs, paired-pulse stimulation of peripheral inputs produced a depressive effect. Because there is a decrease in MT unit activities evoked by the second pulse that corresponds with this depressive effect, it is likely that the depression may be mediated by actions at the thalamus or at multiple levels of the ascending nociceptive pathway.
Thalamic transmission relay
We present in this report a convergence of two lines of evidence that suggest that nociceptive responses in ACC are conveyed by A-δ nociceptive afferents transmitted primarily by MT nuclei. First, we observed a correlation between thalamic unit activities and ACC evoked sink currents; second, we observed direct elicitation of ACC responses by MT stimulation. Indeed previous electrophysiological studies demonstrated that neurons in the MD and CL nuclei in the MT respond to nociceptive stimuli (Berkley et al. 1995; Dostrovsky and Guilbaud 1990; Guilbaud et al. 1986; Shyu et al. 1992). Furthermore, we demonstrated that a selective MT lesion could completely block transmission of the A-δ nociceptive pathway to the ACC, which further underscores the critical role played by the MT in transmitting affective nociceptive information in the medial pain system (Vogt et al. 1993).
The present findings are consistent with the view that the ACC may have an essential role in the affective aspects of pain. Recent human studies indicate that there is a parallel mode of pain processing most probably subserved by parallel thalamocortical projections to multiple cortical areas, including the ACC, and that pain with a strong affective component is closely associated with activity in the ACC (Ploner et al. 2002; Schnitzler and Ploner 2000). Our CSD analysis results indicate that A-δ fiber–activated intracortical synaptic currents have a pattern distinct from that of A-β fiber–activated currents in S1 that have been linked to discriminative function. For example, the ascending A-δ spino-ACC pathways exhibited bilateral activation and were less intensity dependent than the A-β fiber system. Consistent with a more general role of the ACC in affective aspect of pain, previous electrophysiological studies demonstrated that nociceptive ACC neurons have large receptive fields (Sikes and Vogt 1992; Yamamura et al. 1996).
The layer II/III sink currents initiate an intracortical polysynaptic excitation in the superficial layers thought to be associated with synaptic interactions in intercortical area synaptic networks (Conde et al. 1995; Gabbott et al. 2005; Rao et al. 1999). The downward sweeps of synaptic currents in the deep layers indicate that the cortical signals may be relayed to cingulatefugal pathways to subcortical structures such as the MT and periaqueductal gray (Royce 1983; Wyss and Sripanidkulchai 1984). Both the spinal pathway and cortical circuitry make it possible for the ACC to integrate information transmitted by nociceptive A-δ fibers from both sides of the body. Thus the bilateral excitation of the ACC from nociceptive A-δ fibers and the more diffused cortical and subcortical activation of the descending output from the ACC may subserve the role of forwarding nociceptive signals from the ACC to subcortical areas for further nociceptive information processing.
This study was supported by grants from the National Science Council and the Academia Sinica, Taiwan, Republic of China. The multichannel silicon probes were provided by the University of Michigan Center for Neural Communication Technology, sponsored by National Institute of Biomedical Imaging and Bioengineering, Division of Research Resources Grant P41-RR-09754.
The authors thank M.-J. Gan for technical assistance.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2006 by the American Physiological Society