The Journal of Neuroscience, January 1993, 73(l): l-12

Activation of Single Neurons in the Rat during Self-Stimulation of the

M. Wolske,’ P.-P. Rompre,2 R. A. Wise,2 and M. 0. West’ ‘Department of Psychology, Rutgers, the State University, New Brunswick, New Jersey 08903 and ‘Center for Studies in Behavioral Neurobiology and Department of Psychology, Concordia University, Montreal, Quebec, Canada H3G lM8

Single neurons (n = 76) were recorded in the nucleus ac- [Key words: brain stimulation reward, self-stimulation, me- cumbens septi (NAS) of rats self-stimulating the ipsilateral dial forebrain bundle, ventral tegmental area, single-unit re- (MFB) at the level of the ventral cording, nucleus accumbens, fimbria, hippocampus] tegmental area (VTA). Responses evoked by rewarding trains of stimulus pulses fell into five categories. The first category Animals rapidly learn to perform a variety of operant tasksin (40% of the sample) was characterized by a single discharge order to obtain electrical stimulation of the medial forebrain at invariant latency in response to individual pulses of the bundle (MFB). Considerableevidence suggeststhat responsive- train, and hence was termed “tightly time locked” (TTL). Two nessto brain stimulation reward (BSR) is sensitive to changes TTL neurons were collision tested, and both showed colli- in the function of central (DA) neurons (Wise and sion, suggesting that self-stimulation of the VTA may involve Rompre, 1989). Among the several forebrain nuclei innervated antidromic, and thus direct, activation of a substantial num- by DA neurons, the nucleusaccumbens septi (NAS) appearsto ber of NAS . The second category (26%) was char- play an important role in reward processes.Neurochemical stud- acterized by discharges that varied in latency from pulse to ies have shown increasedDA levels in the NAS following food pulse and hence was termed “loosely time locked” (LTL). consumption (Hemandez and Hoebel, 1988; Radhakishun et Responses of the remainder of the sample showed no cou- al., 1988; Josephand Hodges, 1990) during administration of pling to individual pulses but were categorized based on drugs abusedby humans(Di Chiara and Imperato, 1988; Her- general firing patterns during the train: excited (7%), inhib- nandez and Hoebel, 1988) and during self-stimulation of the ited (4%), and no change (23%). Irrespective of category, MFB (Nakahara et al., 1989; Blaha and Phillips, 1990). Am- immediately after the self-stimulation session, the likelihood phetamine (Hoebel et al., 1983) and electrical stimulation (Pra- of evoked discharge at monosynaptic latency by single pulse do-Alcala and Wise, 1984) are self-administereddirectly into stimulation of the ipsilateral fimbria was reduced (relative to the NAS. Intra-accumbens amphetamineincreases the reward- pre-session level), concurrent with elevations in mean firing ing impact of MFB self-stimulation (Colle and Wise, 1988). In rate and motor activity. contrast, the rewarding efficacy of MFB self-stimulation is de- NAS neurons thus exhibit vigorous activation, apparently creased by pharmacologic blockade of DA receptors via mi- both antidromically and orthodromically, in response to VTA croinjection into the NAS, but not other DA terminal fields self-stimulation. The responses of certain LTL and TTL neu- (Stellar and Corbett, 1989). Moreover, 6-hydroxydopamine de- rons increased as a function of pulse number in the train, pletions of NAS DA attenuate MFB self-stimulation (Stellarand suggestive of integrative mechanisms important for brain Corbett, 1989) block acquisition of amphetamineself-admin- stimulation reward. Conduction velocities of directly acti- istration (Lyness et al., 1979), block maintenance of am- vated (TTL) axons (0.41-0.65 m/set) were slower than those phetamine (Lyness et al., 1979) and (Roberts et al., previously reported for first-stage, reward-relevant axons. 1980; Pettit et al., 1984) self-administration, and block in- Nonetheless, an implication of direct activation of NAS (and creasedresponding for conditioned produced by other MFB) axons is that rewarding stimulation triggers ac- intra-accumbens (Taylor and Robbins, 1984, tion potentials that could invade all axonal branches, in- 1986). cluding those between the stimulation site and the soma, While pharmacologicalstudies strongly suggestthat DA trans- and send synaptic signals to target neurons. Such signals missionin the NAS plays an important role in BSR, experiments from NAS neurons could contribute to the increased motor using psychophysical methods adapted for studying self-stim- behavior accompanying MFB self-stimulation, and/or could ulation have provided evidence that axons of DA neurons do interact with dopamine-mediated signals projected to the not constitute a significant portion of the directly stimulated, NAS from reward circuitry. first-stageaxons. The latter have beenshown to (1) be small in diameter and myelinated, (2) exhibit refractory periods in the range of 0.4-I .2 msec, (3) exhibit conduction velocities in the Received Mar. IO, 1992; revised June 9, 1992; accepted June 24, 1992. rangeof l-8 m/set, and (4) descendtoward the mesencephalon, This work was supported by NIDA Grant DA 0455 1, NSFGrant BNS-8708523, all of which are characteristics incompatible with those of as- and U.S. Public Health Service Grant RR 07058 to M.O.W., and by NIDA Grant cending DA neurons (Yeomans, 1979; Yim and Mogenson, DA 01720 to R.A.W. Correspondence should be addressed to M. 0. West, Department of Psychology, 1980a,b; Bielajew and Shizgal, 1982, 1986; Yeomans et al., Rutgers University, New Brunswick, NJ 08903. 1988; Shizgalet al., 1989).It has beensuggested that descending Copyright 0 I993 Society for Neuroscience 0270-6474193113000 I-12$05.00/0 reward-relevant axons transsynaptically activate DA neurons 2 Wolske et al. * Resrxnse of Accumbens Neurons to VTA Self-Stimulation

(Wise, 1980; Yeomans, 1982; Bielajew and Shizgal, 1986), which Recording. Each recording day, the microdrive, equipped with a tung- in turn may produce a rewarding effect, in part, by modulating sten recording microelectrode (l-10 MQ; Frederick Haer), was attached to the animal, without anesthesia, and the electrode was lowered through the activity of NAS neurons via robust ascending projections cortex and . To ensure that only NAS neurons were studied, (Ungerstedt, 1971; Yim and Mogenson, 1980a; Totterdell and each neuron was tested for responsiveness to single pulse stimulation Smith, 1989; Sesack and Pickel, 1990). Using the BSR model, of the fimbria. Since subicular inputs project (via the fimbria) heavily the objective of the present study was to provide an initial to the NAS but only sparsely to the striatum (Groenewegen et al., 1987; characterization of single-cell activity in the NAS during self- Mogenson, 1987), only those neurons activated by, or bracketed dorsally and ventrally by neurons activated by, fimbria stimulation at mono- stimulation of the MFB at the level of the ventral tegmental synaptic latencies (5-11 msec) were studied (DeFrance et al., 1985). area (VTA). Specifically, two primary questions were addressed. Extracellular action potentials were led through a harness and comu- First, are NAS neurons synaptically activated, at latencies sug- tator, amplified, filtered (l-9 kHz bandpass), and sent to the computer gestive of disynaptic or polysynaI&ic input from reward cii- (AST 386-C), which controlled all phases of the experiment (DISCOVERY software, Brainwave Systems Corp). cuitty? Second, because some NAS axons descend in the MFB After the last recording in each animal, an electrolytic lesion (0.05 to terminate in the VTA (Nauta et al., 1978) on DA neurons mA. 10 set) was made with a 0.01 inch stainless steel insulated wire (Yim and Mogenson, 1980b), thus potentially satisfying one mounted in the microdrive to verify the placement of recording elec- criterion of first-stage axons, .are NAS neurons antidrimically trodes histologically. Cells recorded-at depths not verified to be-in the activated? Stimulation of axons results in both antidromic and NAS were excluded from the sample. Electrolytic lesions were also made through fimbria and VTA stimulating electrodes to verify placement. orthodromic nerve impulses, and thus antidromic responses Procedure. After a neuron in the NAS was isolated, the animal was identify neurons whose fibers might carry directly activated, allowed access to the lever. If lever nressine did not commence within reward-relevant signals orthodromically toward the mesen- 2 min, up to five “priming” trains were delitered noncontingently, after cephalon. which all rats began pressing. Each neuron was studied for 5 min of lever pressing (150-200 trains). Afterward, the Plexiglas wall was re- Preliminary results have been previously reported (Wolske et placed to deny access to the lever. The electrode was then lowered to al., 1990). record additional NAS neurons similarly. One measure of NAS activitv was to comoare firine durine the train (0.5 set duration) with that dt&ng the 0.5 set period ?mmed?ately pre- Materials and Methods ceding each train (see Data analysis, below). High rates of lever pressing Subjects. Adult, male Long-Evans rats (Charles River Laboratory) (known to exceed three per second in self-stimulating rats) would in- weighing 35wOO gm at time of surgery were housed individually, given terfere with this measure. This was prevented in three animals by using ad lib food and water, and maintained on a reversed 12 hr light cycle a fixed-interval 1.5 set schedule, ensuring a 1 set interval without stim- (on 20:00, off 08:OO) so that daily experiments were conducted during ulation following each train. Within five sessions (- 5000 trials), animals their active period. exhibited a response scallop typical of fixed-interval schedules, after Surgery. Rats were surgically implanted using modifications of a tech- which recording commenced. A second strategy was to use a tone dis- nique described previously (Deadwyler et al., 1979; West and Wood- crimination paradigm for the fourth animal. A tone (1 kHz, 65 dB, 7 ward, 1984). Rats were anesthetized with sodium pentobarbital(50 mg/ set limited hold) signaling the availability of reward was presented at kg, i.p.), and a microelectrode drive (microdrive) base was surgically irregular intervals (range, 3-15 set). The first press after tone onset implanted on the (level) skull above the right nucleus accumbens at 2.2 resulted in cessation of the tone and delivery of one train. This animal mm AP and 1.5 mm ML relative to bregma (Paxinos and Watson, required six sessions (approximately 4500 trials) to reach criterion (less 1986). A bipolar stimulating electrode (0.005 inch Teflon-insulated than 10% incorrect presses between tone presentations) before recording stainless steel wire) was implanted in the ipsilateral fimbria (-2.2 mm commenced. AP and 2.8 mm ML from bregma, 4.2 mm DV from skull). A monopolar A second measure of NAS activity was the firing evoked by single stimulating electrode (0.01 inch Teflon-insulated stainless steel wire) pulses within each train. In order to allow this analysis, pulses were was implanted in the ipsilateral VTA (-5.8 mm AP and 0.6-0.8 mm delivered at frequencies below 33 Hz, providing a minimum interpulse ML from bregma, 7.8-8.1 mm DV from the cortical surface). The tip interval of 30 msec. of the VTA stimulating wire was sharpened to a point (15’) whose Additional tests were performed on some neurons. Two NAS neurons uninsulated conical surface area was 0.453 mm* (tip length, 1.0 mm). were tested to assess whether the VTA-evoked discharge was antidrom- A tinned copper wire (22 or 26 gauge) wrapped around three skull screws ic, by means of the collision test. Immediately after the self-stimulation posterior to bregma was used as the return path from the VTA stim- session, a window discriminator (World Precision Instruments model ulating electrode. Following surgery, rats were given a 1 week recovery 12 1) was used to discriminate the neuronal waveform (separately from period before testing. the discrimination performed by BRAINWAVE during the session; see next Recording chamber. Experiments were performed in a clear Plexiglas section). The output of the discriminator triggered the VTA stimulator recording chamber (L 35 cm x W 17 cm x H 40 cm). A speaker, to deliver one pulse identical to the pulses used during the session. A through which a computer-controlled, 1000 Hz, 60 dB tone could be further test (20 neurons) was to assess alterations in NAS neuronal delivered, was positioned above the chamber at one end. At the same responsiveness to monosynaptic input from the hippocampus. This end of the chamber, a black Plexiglas lever protruded through a hole in consisted of delivering approximately 50 fimbria pulses (0.33 Hz) im- a moveable, clear Plexiglas wall such that the length of lever available mediately before and after the self-stimulation session. A single stimulus to the animal could be varied. Once animals had been conditioned to intensity was selected that evoked neuronal discharges with approxi- lever press, the wall was positioned so that only the paw could contact mately a 50% likelihood (range, 0.5-6.0 mA; 0.1 msec duration). the lever. A second moveable wall, made of black Plexiglas, could be Data analysis. Using BRAIN~AVE software, neuronal waveforms were placed in front of the lever when it was desirable to deny the animal digitized (32 kHz sampling frequency) and time stamped as they oc- access to the lever. A low-level room light remained lit throughout the curred, and stored for off-line analysis. Different waveforms recorded experiment to allow visual observation by the experimenter. at single electrode positions were separated based on specific parameters. Stimulation. Each lever press resulted in the immediate delivery of These were peak amplitude, valley amplitude, overall amplitude, du- one train of DC pulses (0.1 msec pulse duration, 0.5 set train duration) ration (peak to peak), and polarity. The boundaries for each parameter to the VTA stimulating electrode. Maximum pulse frequency was 33 were selected so as to yield pure neuronal waveforms and to reject Hz (to permit analysis of firing between pulses; see below), resulting in stimulus artifact and other artifacts (i.e., related to grooming, chewing, a maximum of 16 pulses per train. During the first recording in each etc.). Waveforms recorded throughout the time course of each experi- animal, a single stimulus current was identified as the lowest (threshold) ment, and their parameters were displayed to ensure that no change in current required to maintain lever pressing (> 100 presseslmin). Re- recorded signal occurred as a function of time, due to a spurious shift ductions from threshold of less than 10% (0.03 mA) resulted in cessation in electrode position. of lever pressing. This threshold value was the only stimulus current Peristimulus histograms (PSHs) were generated for each neuron. Since used throughout the study. For the four animals, stimulus current ranged each count in the PSH corresponded to one occurrence of the waveform from 0.3 to 0.5 mA. Stimulus pulses were delivered by computer-con- ofthat neuron, PSHs were free ofall artifact (including stimulus artifact). trolled Grass SD9 stimulators. Each histogram represented the summed activity of the neuron across The Journal of Neuroscience, January 1993, 13(l) 3 approximately 150 train deliveries during a 5 min session. Two PSHs duration of the responsedisplayed in the PSH was severalmil- with different time bases were used to study each neuron’s response to liseconds.Such responseswere termed LTL and are consistent VTA self-stimulation. One PSH displayed neural firing during the 250 msec prior to train delivery, the 500 msec of train delivery, and the 750 with synaptic activation. Across the 14 LTL neurons,a broad msec following the end of the train (total time base, 1.5 set). Numerical range of responsetopographies were exhibited. Responsemag- data derived from these histograms were analyzed statistically and also nitudes of the 14 LTL neurons, in terms of evoked period:base were analyzed to quantify changes in evoked discharge as a function of period firing ratio, ranged from 0.52 to 0.96 (where 0.96 rep- increasing pulse number in the train. A second PSH displayed neuronal resentsan increaseof >20-fold in the evoked period; Fig lk). discharges on an expanded time base showing activity forward 50 msec and backward 50 msec from individual pulses of the train. All pulses Responsedurations rangedfrom 8 to 18 msec.Response latency from each train were summed across all trains (16 pulses per train x ranged from 17 msec (Fig. lc) to 35 msec (Fig. 11) the latter 150 trains = 2400 sweeps). Finally, PSHs were used to display mono- exceeding the interpulse interval (seenalso in Fig. 1k, which synaptic responses to fimbria stimulation (50 msec time base). “Spon- illustrates that two responsesoccurred after the final pulsede- taneous” firing rates were calculated from PSHs of longer time base displaying firing 1.5 set prior to each fimbria pulse (delivered once per livery). One exception was observed, that is, an LTL neuron 3 set). (Fig. lc) that respondedat a much shorter latency (4 msec)than Terminology. The baseline (“base period”) firing rate was the mean the others. This short-latency dischargewas present in PSHs rate (in Hz) during the 0.5 set period immediately prior to train delivery. (not shown) displaying the responseto only the first pulse of “Train period” firing rate was the mean rate during the 0.5 set of train the train; moreover, PSHs synchronized to the last pulseof the delivery. During the train period, an additional measure, termed “evoked period” firing rate, was determined for neuronsexhibiting evoked dis- train showed the short-latency responsebut did not show a chargestime lockedto individual pulses.A “post-period”firing rate second,long-latency (> 33 msec)response (as seen for the neu- wascalculated as the meanrate duringthe 0.5 set periodafter the end ron in Fig. 1k,f). of the train. Four LTL neurons exhibited distinct changesin evoked pe- Comparison of firing rates between these periods was based on a ratio that wascalculated for eachneuron as follows.For any two periods riod firing as a function of pulse number in the train. In all four beingcompared, the firingrate of oneperiod was divided by the summed cases,responsiveness increased as the train progressed.In each firing rates of both periods. For example, the “evoked period:base period case,the number of dischargesin the evoked period increased firing ratio” wasthe ratio of evokedperiod firing to the sum evoked in a nearly linear manner. To obtain a simple quantification of period firing + base period firing. Ratios thus were limited to a range these increases,the responseto the last pulse was compared between 0 and 1. These ratios evenly weighted increases and decreases, and minimized variance in firing rate among neurons, both of which with that to the first. Increasesshown by the four neuronswere were necessary for statistical analysis. Classification of non-time-locked 100%(Fig. lb), 114%, 155%, and 950% (Fig. lk). PSHs of the neuronswas made arbitrarily, basedon train period:baseperiod firing remaining 10LTL neuronsexhibited no visible changein evoked ratios. That is, neurons with a ratio between 0.33 and 0.66 were classified period firing as a function of pulsenumber (e.g., Fig. 1e.h). as no change.Those above 0.66 (i.e., firing duringthe train periodwas more than twice that of base period) were classified as excited, and those During the period between time-locked evoked responses below 0.33 (i.e., train periodfiring wasless than half that of baseperiod within the train, firing decreasedfor five LTL neuronsrelative firing)were classified as inhibited. to the baseperiod (range, 9-44% of baseperiod firing). Figure Statisticalanalysis. Version 6.06 of sAs (SAS Institute Inc.) was used 1b illustrates an LTL neuron that exhibited this decrease.For to performstatistical analyses on an IBM 308 I mainframe computer. T statisticswere generated using a t test for unequalvariances, except this neuron, evoked period firing did not exceed baseperiod as noted in the Results. No analyses were performed when there were firing, but consistedof a time-locked dischargewithin a period fewer than five observations for a particular independent variable. of generally decreasedfiring during the train (Fig. lc). During the post-period, one LTL neuron (Fig. 1k) exhibited an initial rebound excitation followed by a gradual recovery to Results the baseperiod firing rate. Neuronal activity in the sample TTL neurons.For 2 1 “tightly time-locked” neurons,a single Histological analysis confirmed that all neurons (n = 76) re- dischargeat an invariant latency characterized the responseto corded from the four animals were located in the NAS (described individual pulsesof the train, consistent with antidromic acti- below). Twenty-three of the 76 neurons exhibited firing rates vation. That is, all evoked dischargesoccurred in one 1 msec too low to allow analysis (never exceeding0.3 Hz) and were not bin of the expanded PSH (Fig. 2). (For any given neuron, vari- analyzed further. For 53 neurons analyzed, base period firing ations in responselatency of lessthan 0.4 msecresulted in the wasnot different from post-period firing [T(84.3) = -0.721. In display of all neuronal dischargesin one bin.) Latency to dis- contrast, train period firing (including evoked period firing for charge for TTL neurons ranged from 11 to 16 msec. time-locked responses)was significantly different from basepe- Following the 5 min sessionof self-stimulation, two TTL riod firing [T(47.5) = -3.71; p < O.OOl] and from post-period neurons were tested for high-frequency following and collision, firing [T(48.1) = 3.61; p < O.OOl]. Therefore, further analysis using a stimulus pulseidentical to those usedduring self-stim- of train period firing was undertaken. ulation. Both neuronsfollowed paired stimuli separatedby less than 2.0 msec;one showed a refractory period of 1.2 msec(Fuller Categorization of neuronsby PSH topography and Schlag, 1976). Both neurons also showed collision. The Visual inspection of PSHs acrossthe sample revealed diverse probability ofevoked dischargefor thesetwo neuronswas stable but recurring topographiesof neuronal responseto self-admin- at 0.4 and 0.9, respectively, when the collision interval was isteredtrains of VTA stimulation. This prompted classification greater than the conduction time. When the collision interval of each neuron into one of five categories:loosely time locked was lessthan the conduction time, both neurons showed an (LTL), tightly time locked (TTL), excited, inhibited, and no abrupt cessationof evoked discharge.The conduction velocities change.Except where noted, neuronal data did not differ across for these two neurons were 0.41 and 0.43 m/set (latency, 16 rats, and data therefore were pooled. and 15 msec,respectively). The range for all 2 1 TTL neurons, LTL neurons. Fourteen neurons showed excitatory evoked assumingthat these invariant latency responseswere also an- responsesthat were time locked to individual pulsesin the train, tidromic, was quite narrow, that is, from 0.41 to 0.65 m/set. at latenciesthat varied from one pulse to the next. Thus, the Over half the TTL neurons (13 of 21, 62%) were recorded 4 Wolske et al. * Response of Accumbens Neurons to VTA Self-Stimulation

LOOSELY TIlME-LOCKED

Figure 1. Four examples of LTL neu- ronal responses. Eachneuron is repre- sentedin horizontal arrangementof threegraphs (e.g., u-c). Graphsa, d, g, and j are overlaid records of all occur- rences ofeach neuron’s extracellular ac- tion potential waveform during 5 min d.75 0.+5 ’ 0 0 session of self-stimulation (negative is TIME IN SECONDS TIME IN SECONDS up; calibration: 0.2 msec, 0.05 mV). Each occurrence of waveform corre- sponds with one count entered into each PSH at the right. Thus, PSHscontain no stimulusartifacts (which obscured neural waveforms for I msec-i.e., first bin after time 0 in right column). PSHs b, e, h, and k show discharges during 250msec prior to train of VTA pulses, 500msec of train delivery (horizontal bar), and 750 msec following train de- livery. All occurrencesof last (16th) TIME IN SECONDS pulse in train (time 0) were used to gen- TIME IN SECONDS eratePSH (- 150sweeps in each).The ordinate represents firing rate (Hz) per bin(3 msec/bin). PSHs c,f; i, andlshow responsesto individual pulsesin train. All pulses(1-16) in all trainsof 5 min session (solid arrow at time 0) were used to generatePSHs (-2400 sweepsin each). The second solid arrow repre- sents next pulse (2-16) in train. The ordinate represents,for each1 msecbin, the numberof dischargesper stimulus -675 0.15 ’ b ’ ’ 0!05 pulse (i.e., likelihood, from 0 to 1, of TIME IN SECONDS TIME IN SECONDS responding at a particular latency). The openarrow showsmean baseline firing rate (base period) converted into dis- charges per pulse, for reference lo PSH at the left. Note that response latency in PSHs k and I was 35 msec, as shown by presence of two evoked responses after the end of train in k. Thus, re- sponse (peak) at the right in I is the true response to pulse at time zero; the peak at the left in I is response to the pre- -6.75 6 0.15 b 0h5- cedingpulse. TIME IN SECONDS TIME IN SECONDS

from one animal. These 13 neurons constituted 65% of the total Evoked period firing rateswere substantially higher than base neuronsrecorded from this animal. Placementof the VTA elec- period rates for TTL neurons. The evoked period:baseperiod trode for this rat was verified as -4.80 mm AP and 1.4 mm firing ratio ranged from 0.89 to 1.00. That is, the smallestin- ML (from bregma)and 8.7 mm from skull surface. In contrast, creasewas greater than eightfold. The largest increase(1.00 a different animal exhibited only one TTL neuron (8% of the ratio) reflected one neuron’s absenceof spontaneousfiring in total recorded from this animal). Placement of the VTA elec- the baseperiod. trode was verified at -6.04 mm AP and 0.5 mm ML (from The evoked period dischargeof six TTL neuronschanged as bregma)and 8.8 mm from skull surface. Although all neurons a function of pulsenumber in the train. In eachcase, the change from both animals were recorded from the central region of the took the form of a shift, within one to three pulses,from one AP extent of the NAS, recordingsobtained from the latter an- fairly steady level of responseto another. To obtain a simple imal (exhibiting one TTL response;see Fig. 5, track D) were quantification of thesechanges, the mean responseto all pulses lateral and ventral to those from the former (seeFig. 5, track after the shift was compared with that to all pulsesbefore the C). Overall, 18 TTL neurons were observed medially (40% of shift. (Onset of the shift rangedfrom pulse4 to pulse 10). Five all neurons recorded medially), whereas only 3 TTL neurons neuronsshowed increases of 54% (Fig. 2e), 63% (Fig. 2/z), 80%, were observed laterally (10% of all neurons recorded laterally). 83%, and 187%,and one neuron showeda decreaseof 64% (Fig. Due to technical problems, histological verification of the VTA 26). Of three neurons that were recorded simultaneously,one electrode placement for the other two animals could not be increased(Fig. 2e), one decreased(Fig. 26), and one showedno performed. change(not shown)across the train, suggestingthat suchchanges The Journal of Neuroscience, January 1993, 73(l) 5

TIGHTLY TIME-LOCKED

0.05 TIME IN SECONDS TIME IN SECONDS

g “I- h l vi

11 0.76 0 C TIME IN SECONDS TIME IN SECONDS

F&we 2. Four examples of TTL neu- ronal responses. TTL responses were characterized by invariant latency: all discharges occurred within one 1 msec bin (c, 1; i, and I). (Note that stimulus artifact is not displayed.) Different neu- -0.15 rons responded at different latencies. TIME IN SECONDS TIME IN SECONDS Details are as in Figure 1.

were not due to spurious fluctuations in pulse intensity during (see Materials and Methods) as excited (n = 4), inhibited (n = the train. The remaining 15 TTL neurons did not show any 2), and no change (n = 12). Figure 3 illustrates the responses of identifiable pattern of change in evoked firing across the train three excited neurons. Firing rates increased beginning approx- (e.g., Fig. 2k). imately 25 msec after the first pulse, continued throughout the During the period between time-locked evoked responses duration of the train, and extended 100-400 msec into the post- within the train, four TTL neurons exhibited increased firing period (Fig. 3b,e,h). Figure 4 illustrates the responses of the two rates relative to base period firing (range, 21011200% of base inhibited neurons. Firing rates decreased beginning 35-60 msec period level), seven showed decreased firing rates (range, < l- following the first pulse in the train, continued throughout the 46% of base period level), and seven did not change (no deter- duration of the train, and extended 50-350 msec into the post- mination was possible for the remaining three neurons). Also, period. Neither excited nor inhibited neurons exhibited any five TTL neurons exhibited changes during the post-period. The change in responsiveness across the train period. decreases over the course of the post-period on the part of three neurons appeared to represent recovery from elevated firing Comparisons of the categories between evoked periods of the train (e.g., Fig. 2/c). The increases The mean firing rate for neurons in each category is shown in shown by the other two neurons appeared to represent rebound Table 1. Orthogonal comparisons were made using base period from suppressed firing between evoked periods of the train. firing as the dependent variable. Base period firing for LTL Non-time-locked neurons. Eighteen neurons exhibited general neurons was significantly higher than that ofthe other categories patterns of firing that were not time locked to individual pulses combined [T( 12.4) = 2.33; p < 0.051. Two LTL neurons showed of the train. The responses of these neurons were categorized unusually high base period firing rates (11.27 Hz and 12.08 Hz) 6 Wolske et al. - Response of Accumbens Neurons to VTA Se&Stimulation

EXCITED

6 TIUE IN SECONDS TIME IN sEmNns

0 0 6 TIME IN SECONDS TIME IN SECONDS

Figure 3. Threeexamples of excited, non-time-locked NAS neuronal re- sponses.Evoked firing of excitedneu- ronsconsisted of generalincrease dur- ing train period (horizontal bar over graphs b, e, andh) that showedno ob- vious synchronyto individual pulses (solid arrows in graphsc, J andi). De- 6 tailsare as in Figure 1. TIME IN SECONDS TIME IN SECONDS

as compared with other LTL neurons (0.65-4.53 Hz) and the single pulsestimulation of the ipsilateral fimbria, from 0.58 to other categories.Base period firing of neuronsshowing no change 0.25 discharges/pulse[T(19) = -3.24; p < 0.011. was not different from that of excited, inhibited, and TTL neu- rons combined [T(18.1) < 11. No further comparisons were Discussion possible. Thesefindings constitute the first characterization at the single- cell level in self-stimulating animals of neural activity in the Histology NAS. The major conclusion of this study is that NAS neurons Of 76 neurons in the sample, 33 were located in the central exhibit vigorous responsesto self-administeredtrains of VTA region of the AP axis of the NAS, 5 were anterior, and 38 were posterior. Figure 5 showsthe locations in the NAS of all neurons Table 1. Mean firing rates (+SEM) of the five categories during in the sample,in terms of responsecategory. base, train, and post periods Comparison of pre- versuspost-session levels of spontaneous Period andJimbria-evoked firing Category Base Train Post N (53) For 20 NAS neurons, spontaneousfiring rate and responseto fimbria stimulation (verified histologically) were measuredim- LTL 3.5 f 1.1 10.5 + 2.5 4.6 f 1.6 14 mediately after the 5 min sessionof VTA self-stimulation and TTL 0.9 + 0.2 42.1 k 10.2 1.2 f 0.3 21 compared to pre-sessionlevels. Mean spontaneousfiring rate Excited 1.0 + 0.3 3.5 + 1.1 1.3 t 0.4 4 following the sessionwas significantly increased(Fig. 6), from Inhibited 1.0 + 0.1 0.3 f 0.1 1.4 k 0.8 2 2.15 to 2.69 Hz [T(19) = 3.22; p < 0.01, paired t test]. Con- No change 1.2 + 0.3 1.0 + 0.3 1.2 + 0.3 12 currently, a significantdecrease was observed in meanlikelihood For time-locked response categories, “Train” refers specifically to evoked period. of evoking dischargeat monosynaptic latency (5-l 1 msec) by N = number of neurons in each category. The Journal of Neuroscience, January 1993. 13(l) 7

INHIBITED

0 -6.76 0.95 TlMElNSECOND.3 TlME IN SECONDS

13 Figure 4. Graphs illustrating the two d inhibited, non-time-locked neuronal responses. Evoked activity of inhibited neurons consisted of general decrease in firing during train period that showed no synchrony to individual pulses. The neuron at the top showed steady recov- ery during the post-period (graph b); the 0 neuron at the bottom showed rebound/ -6.75 0.i5 recovery in post-period (graph e). De- TIMEINSECONDS TIME IN aEcoNlx4 tails are as in Figure 1.

stimulation. Although the magnitude of the responses, for ex- sidered the most reliable measure of antidromic discharge (Ful- ample, Figures 1 and 2, might be an expected result of stimu- ler and Schlag, 1976), is the collision test. Both TTL neurons lating the VTA, a major afferent and efferent of the NAS, such tested, using a stimulus pulse identical to those used during self- responsiveness was not predicted by the results of 2-14C-deoxy- stimulation, showed collision. Both neurons also faithfully fol- glucose (2-DG) autoradiographic studies by Gomita and Gal- lowed high-frequency (> 500 Hz) stimulation, thus meeting all listel(1982) and Gallistel et al. (1985). These investigators found criteria for antidromic activation. These results provide strong that 2-DG utilization was not altered in the NAS following a evidence that some neurons (up to 40% of the present sample) 45 min session of MFB stimulation. However, a recent study in the NAS are antidromically, and thus directly, activated by by Porrino et al. (1990) showed that 2-DG utilization was in- self-stimulation of the VTA, consistent with anatomical (Nauta creased in the NAS following self-administered trains of MFB et al., 1978) and electrophysiological (Maeda and Mogenson, stimulation. These authors suggested that modifications in the 1980; Yim and Mogenson, 1980b) evidence for a descending 2-DG technique by Gallistel and colleagues accounted for the projection from the NAS to the VTA. differences in the results. In agreement with the present findings, LTL responses exhibited durations, evident in PSHs, ranging several other studies have demonstrated increased NAS activity from 8 to 18 msec. Analysis of responses to single pulses (raster in response to MFB stimulation, by means of a variety of meth- displays not shown) revealed either single or multiple discharges ods, including acute electrophysiology (Rompre and Shizgal, that varied in poststimulus latency from one pulse to the next. 1986; Shizgal et al., 1989), acute (Gratton et al., 1988) and Such responses are consistent with synaptic activation. We can- chronic (Blaha and Phillips, 1990) electrochemistry, microdi- not rule out the possibility that some recordings involved more alysis (Nakahara et al., 1989), and tyrosine hydroxylase activity than one neuron with waveforms indistinguishable from one (Phillips et al., 1987). another, using the present template matching technology. This could account for the variable latencies, as well as the high firing Possible mechanisms mediating NAS responses to self- rates (in some cases), of LTL neurons. However, the long du- stimulation rations of the evoked response period (up to 18 msec) almost Based on the topography of PSHs, NAS responses to VTA self- certainly do not represent the discharges of up to 18 indistin- stimulation were separated into four categories. These topog- guishable, antidromically activated (TTL) neurons, each at a raphies were, in turn, based on neurophysiological and neuro- different latency. Therefore, the most important conclusion chemical principles. drawn from LTL discharges is sustained, that is, that they rep- TTL neurons all met the first criterion for demonstrating resent a distinct category of NAS responses mediated synapti- antidromic excitation, that is, that neuronal discharges be evoked cally, not antidromically. Orthodromic transmission to the NAS at a constant latency across trials (variance of less than 0.1-0.5 could involve direct projections from the VTA (Ungerstedt, msec; Lipski, 1981). Indeed, the present rigid invariance in la- 1971; Beckstead et al., 1979; Nauta et al., 1978; Sesack and tency militates strongly against the alternative explanation, that Pickel, 1990). Judging from the topographies of most LTL re- TTL responses were mediated synaptically. A second test, con- sponses, fairly rapid on-off excitatory neurotransmission ap- 8 Wolske et al. * Response of Accumbens Neurons to VTA Self-Stimulation

A !3 l-r U EU o- L - N&J T - NUU L

TUU U - LLEN L - ‘: T E - Figure 5. Summary of locations of N lTr TIU x NAS neurons across animals. Top, Each column (A-E) illustrates dorsoventral I” positions of neurons within electrode B l- tracks whose positions are shown in co- LNUU b”u ronal planes at the bottom (A-E). Each - “track” represents one to three individ- :: Lb” ual electrode penetrations at that lo- TU cation. Thus, each neuron can be vi- TLN sualized with respect to its location in the NAS, but not with respect to its spatial relation to other neurons. Each neuron is represented by letter desig- nating its response category (r, TTL; L, LTL, E, excited; I, inhibited; N, no change; CJ,unidentified; -, neurons not meeting criteria for study; blank spaces represent absence of single cell activi- ty). “Unidentified” neurons are those with firma rates too low to classify. The vertical scale (fop right) is divide-d into 100 pm intervals and labeled every 1.O mm; 0 represents point of entry into NAS. Bottom, Tracks change from bro- ken to solid line at point of entry into NAS. Ventralmost portions of some tracks (e.g., track B) were located in NAS shell. Numbers indicate AP distance (in mm) from bregma (Paxinos and Wat- son, 1986). Each coronal plane is partial view of one hemisphere (medial at left; lateral at right). Hatched areas, anterior commissure. Tick marks on tracks are at 1.0 mm intervals.

peared to be involved (Maeda and Mogenson, 1980; Thierry et stageneurons make synaptic connectionswith second-stageneu- al., 1980; Yim and Mogenson, 1980a). No conclusionscan be rons that integrate the signal. Integration may take placeat one drawn regarding the conduction velocities of axons mediating or more additional synaptic stagesin the relevant circuitry, but LTL responses(given an unknown number of synaptic delays), ultimately the signal reachesa conversion site where the re- or whether theseresponses involve DA. warding effect is generated. Non-time-locked changesin firing lastingthroughout the train Could the observed NAS firing patterns representactivity of period were exhibited by excited and inhibited neurons, and first-stageneurons? Psychophysical methodsadapted for study- alsoby certain LTL and TTL neuronsduring the period between ing self-stimulation have revealed that the axons of first-stage time-locked evoked firing. Such patterns could reflect asyn- (directly activated) neurons are myelinated, descendingaxons chronous polysynaptic input, or possibly the action of slow- (Bielajew and Shizgal, 1986) with conduction velocities of l-8 acting transmitters or modulators. Neurons in non-time-locked m/s, diametersof 0.3-l .5 pm (Bielajew and Shizgal, 1982) and responsecategories may representdistinct populations of NAS refractory periods between0.4 and 1.2 msec(Yeomans, 1979). neurons or may have been misalignedwith the portion of the One category in this study consistedof neurons that were an- MFB activated in a given animal; a different alignment might tidromically, and thus directly activated, that is, TTL neurons. produce time-locked discharges. In one case,a determination of refractory period wasmade, and this value, 1.2 msec, is within the appropriate range, as were Possiblefunctions of NAS responsepatterns in self-stimulation values obtained from a portion of NAS neuronsin acute studies Do these findings fit present conceptualizations of activity in (Rompre and Shizgal, 1986; Shizgal et al., 1989).TTL neurons reward circuitry during self-stimulation?According to the model are further implicated as candidatesby studiesindicating that advanced by Gallistel et al. (198l), action potentials are gen- someNAS axons descendin the MFB to terminate in the VTA erated initially in a bundle of reward-relevant axons. Thesefirst- (Nauta et al., 1978)on DA neurons(Yim and Mogenson,1980b). The Journal of Neuroscience, January 1993, 13(i) 9

larly the LTL neurons, in which responses appeared to be or- thodromic, exhibited a pattern that might be expected of neurons involved in the integration of rewarding signals. To our knowl- edge, this is the first such demonstration. Whether this pattern is unique to the NAS is an issue awaiting further study; no such pattern has been reported in electrophysiological studies of other structures (Rolls, 197 la,b; Segal and Bloom, 1976). Finally, it should be noted that changes across the train shown by TTL neurons, assuming these were antidromically activated, could possibly reflect changes in terminal excitability (cf. Ryan et al., 1986, 1989). FIM FIM PRE POST Relevance of response patterns to DA transmission in the NAS Recent evidence strongly suggests that DA transmission is an important component of MFB self-stimulation. When DA levels are increased in the NAS through local injections of amphet- amine during self-stimulation, reward, measured independently from motor performance, is enhanced (Colle and Wise, 1988). * Conversely, when DA function is temporarily attenuated via 1 local injection of neuroleptics into the NAS, reward, but not motor performance, is decremented (Stellar and Corbett, 1989). That this is a specific function of DA in the NAS and not other forebrain DA terminal fields is supported by the finding that local injections of neuroleptics into the striatum, amygdala, or +d---l prefrontal cortex do not affect reward (Stellar and Corbett, 1989). Complementary to these reports, electrochemical studies have 04 I I I I shown that NAS DA levels correlate directly with MFB self- SPONT SPONT stimulation using both microdialysis (Nakahara et al., 1989) PRE POST and in vivo voltametry (Blaha and Phillips, 1990). Taken to- Figure 6. Concurrentincrease in meantiring rate anddecrease in fim- gether, these data provide strong support for a role of NAS DA bria-evokeddischarge (n = 20 NAS neurons)immediately following in BSR (see also Wise and Rompre, 1989). session of VTA self-stimulation. Top, Mean number of discharges per The question of whether DA or other neurotransmitters af- pulse evoked at monosynaptic latency by single pulse stimulation of ipsilateral fimbria immediately before (PRE) and after (POST) session. ferent to the NAS (Maeda and Mogenson, 1980; Yim and Mo- Bottom, Mean spontaneous firing rate of same neurons during corre- genson, 1980a) are involved in actually mediating any of the sponding time periods. Asterisks indicatesignificant changes (p < 0.01) present NAS response patterns was not addressed in this study. relative to pre-period. Nonetheless, the significance of these findings is that they were obtained during self-stimulation of the MFB, and therefore must Thus, these neuronsproject caudally to the and ex- be considered integral with the activity of a structure important hibit thresholdslow enough to be excited by rewarding stimu- for this behavior. Direct, antidromic activation of NAS neurons lation parameters.However, conduction velocities of TTL neu- on an apparently large scale must almost certainly interact with rons ranged from 0.41 to 0.65 m/set-below the values putative reward-relevant activity in the NAS. For determined for first-stage, reward-relevant axons in studies example, action potentials conducted antidromically to the NAS (Yeomans, 1989) using stimulus parameters similar to those could invade short, probably GABAergic collaterals (Smith usedin the present study. Thus, TTL neurons do not meet all and Bolam, 1990) and potentially modify responses of neigh- criteria for first-stageneurons (Shizgal et al., 1989). boring neurons (Somogyi et al., 1981) to incoming synaptic Are the recorded neuronal responsesconsistent with the pos- signals from reward circuitry. Or, for example, synapses upon sibility of a role for the NAS at a later stagein reward circuitry? a TTL NAS neuron that are activated (orthodromically) by It is known that one or two brief (0.1 msec) pulses are not reward circuitry concurrently with antidromic activation of that sufficient to support self-stimulation, but that a train of several same neuron could undergo changes, such as Hebbian poten- pulsesis required. This suggeststhe need for an additive (in- tiation. Such interactions could facilitate-or interfere with- tegrative) processacross the train of pulses, which might be neural activity mediating MFB stimulation reward. Whatever evidenced as a changein the responsesof neurons at an inte- the case, the inevitable conclusion is that, during self-stimula- grator site. The evoked dischargeof 10 NAS neurons (of 33 tion of the MFB at the level of the VTA, NAS output includes examined) did show changesacross the train. Nine (four LTL the response patterns shown here, either separate from, or in and five TTL neurons) of these 10 showed increasedrespon- combination with, any DA-mediated processing. Although the siveness,possibly reflective of an additive process.These changes form and the effects of such interactions must remain specu- did not appearto be due to unintended fluctuations in stimulus lative at present, knowledge of these NAS response patterns pulse intensity acrossthe train (e.g., due to electrode polariza- may prove necessary for ultimately understanding neural circuit tion), becausesimultaneously recorded neurons showed changes activity underlying BSR. in opposite directions (Fig. 26,e). Moreover, sincethe response Another implication of direct, antidromic activation of NAS to each pulsewas a summation across > 150 trials, it was con- neurons during MFB self-stimulation is that all target cells of sidereda reliable measure.Thus, some NAS neurons, particu- antidromically activated neurons, not just those that are efferent 10 Wolske et al. f Response of Accumbens Neurons to VTA Self-Stimulation

to the stimulating electrode, may receive synaptic inputs via peared to mediate the TTL responses of this study, two ofwhich collateral projections. That is, rewarding stimulation triggers were confirmed by the collision test. One ofthe neurons showing action potentials in directly activated axons both orthodromi- collision, as well as 16 of 16 other TTL neurons tested, respond- tally and antidromically, the latter of which could potentially ed at monosynaptic latency to stimulation of the fimbria. These invade axonal branches that leave the main axon between the findings demonstrate that hippocampal neurons have a disy- stimulation site and the soma. Presumably, this applies to all naptic route of accessing VTA neurons-potentially DA neu- directly activated axons originating in the many nuclei, includ- rons, which have been shown to receive synaptic input from ing the NAS, that contribute to the MFB (Nieuwenhuys et al., NAS axons (Yim and Mogenson, 1980b). The data (Fig. 6) 1982). In addition, TTL firing was well within natural limits. suggest that the efficacy ofthe connection from the hippocampus TTL discharges showed an orderly relation to stimulus pulses to the NAS and VTA may be reduced, at the level of the NAS, (either one discharge or no discharge in response to each stim- by activation of the VTA itself, with possible implications for ulus pulse, delivered at 33 Hz or less). Thus, TTL discharges limbic-motor integration (Lopes da Silva et al., 1984; Mogen- did not represent aberrant neural firing, characteristic of epi- son, 1987). leptiform activity or high-frequency “injury discharge.” Taken together, these observations suggest that signals projected or- Behavior thodromically by branches of first-stage axons to target neurons, Increased motor behavior (e.g., sniffing, gnawing, rearing, lo- whether the latter are second-stage neurons of reward circuitry comotor activity, head bobbing) was associated with self-stim- or not, are likely to be of consequence. For example, signals ulation. As typically observed during self-stimulation of the projected from NAS TTL neurons to targets in locomotor cen- MFB (Olds and Milner, 1954; Trowill et al., 1969), these be- ters (Mogenson, 1987) could contribute to the robust stimula- haviors commenced upon delivery of the first VTA train of each tion of locomotor behavior that accompanies MFB self-stim- session and continued for 34 min after the last train delivery. ulation. (Locomotor stimulation also could result from the Mean firing rates of NAS neurons during this 34 min period apparently orthodromically mediated activity of LTL, excited, following the session were significantly higher than those im- and/or inhibited neurons in the NAS.) mediately prior to self-stimulation. (It was not possible to assess mean firing rates during the session itself.) Finding elevated Fimbria stimulation firing rates of NAS neurons during elevated motor behavior is The decreased likelihood of evoking monosynaptic NAS dis- consistent with the previous observation in our laboratory that charges by stimulating the fimbria immediately following the 95% of NAS neurons fired faster during locomotion than during session of VTA self-stimulation was not the result of a gener- rest (West and Wolske, 1987). Increased firing of NAS neurons alized decrease in excitability of NAS neurons, since mean firing following VTA self-stimulation may be related, among several rates were elevated at this time. A plausible explanation of this possibilities, to concurrent residual elevation in DA levels in decrease derives from work reported by Yang and Mogenson the NAS following VTA self-stimulation (Blaha and Phillips, (1986). They demonstrated that subicular neurons antidromi- 1990). This may implicate elevated firing of NAS neurons as a tally activated by stimulation ofthe NAS exhibited an increased mechanism by which DA action at the level of the NAS stim- likelihood of being activated by NAS stimulation following trains ulates motor behavior (cf. Kelley et al., 1989; Delfs et al., 1990). of VTA stimulation. Their results suggested that the mechanism Alternatively, if VTA stimulation results in residual DA ele- responsible for this effect was a D2 receptor-mediated increase vation within the VTA as well, autoreceptor-mediated inhibi- in subicular terminal excitability: elevated DA levels, via D2 tion of VTA DA neurons could, in turn, disinhibit NAS neurons heteroreceptors, depolarized the subicular terminals in the NAS, (White and Wang, 1986). resulting in presynaptic inhibition, that is, decreased release of neurotransmitter upon arrival of an orthodromic action poten- References tial at the subicular terminal. The present decrease in fimbria- Beckstead RM, Domesick VB, Nauta WJH (1979) Efferent connec- evoked NAS neuronal discharge, which occurred within the tions of the substantia nigra and ventral tegmental area in the rat. same time interval (several minutes) following trains of VTA Brain Res 175:191-217. stimulation as did enhanced terminal excitability (Yang and Bielajew C, Shizgal P (1982) Behaviorally derived measures of con- duction velocity in the substrate for rewarding medial forebrain bun- Mogenson, 1986), could involve a similar depolarization ofsub- dle stimulation. Brain Res 237:107-l 19. icular terminals in the NAS. Indeed, a residual elevation in DA Bielajew C, Shizgal P (1986) Evidence implicating descending fibers levels in the NAS during a similar time period after VTA self- in self-stimulation of the medial forebrain bundle. J Neurosci 6:9 19- stimulation was recently reported (Blaha and Phillips, 1990). 929. Moreover, ultrastructural evidence indicates that the arrange- Blaha CD, Phillips AG (1990) Application of in vivoelectrochemistry to the measurement of changes in dopamine release during intracra- ment of dopaminergic terminals in the NAS could potentially nial self-stimulation. J Neurosci Methods 34: 125-133. modulate hippocampal synaptic input (Totterdell and Smith, Colle LM, Wise RA (1988) Effects ofnucleus accumbens amphetamine 1989; Sesack and Pickel, 1990; cf. Freund et al., 1984). It has on lateral hypothalamic brain stimulation reward. Brain Res 459: also been shown that activation of VTA neurons by opiates 36 l-368. Deadwyler SA, Biela J, Rose G, West MO, Lynch G (1979) A mi- reduced NAS neuronal discharge evoked by fimbria stimulation, crodrive for use with glass or metal microelectrodes in recording from but this effect did not appear to be mediated by DA (Hakan and freely-moving rats. Electroencephalogr Clin Neurophysiol 47:752- Hemiksen, 1989). Other possible explanations cannot be ruled 754. out; for example, increased activity of fimbria fibers after the DeFrance JF, Marchand JF, Sikes RW, Chronister RB, Hubbard JI session could have resulted in refractoriness to stimulation. (1985) Characterization of fimbria input to nucleus accumbens. J Neurophysiol 54: 1553-I 567. Anatomical projections from the NAS to the VTA have been Delfs JM, Schreiber L, Kelley AE (1990) Microinjection of cocaine demonstrated (Nauta et al., 1978; Maeda and Mogenson, 1980; into the nucleus accumbens elicits locomotor activation in the rat. J Walaas and Fonnum, 1980). Their antidromic activation ap- Neurosci 10:303-3 10. The Journal of Neuroscience, January 1993, 13(l) 11

Di Chiara G, Imperato A (1988) Drugs abused by humans preferen- dopamine in the nucleus accumbens selectively attenuates cocaine tially increase synaptic dopamine concentrations in the mesolimbic but not heroin self-administration in rats. Psychopharmacology 84: system of freely moving rats. Proc Nat1 Acad Sci USA 85:5274-5278. 167-173. Freund TF, Powell JF, Smith AD (1984) Tyrosine hydroxylase-im- Phillips AG, Jakubovic A, Fibiger HC (1987) Increased in vivo tyrosine munoreactive boutons in synaptic contact with identified striatonigral hydroxylase activity in rat telencephalon produced by self-stimulation neurons, with particular reference to dendritic spines. Neuroscience of the ventral tegmental area. Brain Res 402: 109-I 16 13:1189-1215. Porrino LJ, Huston-Lyons D, Bain G, Sokoloff L, Kometsky C (1990) Fuller JH, Schlag JD (1976) Determination of antidromic excitation The distribution of changes in local cerebral energy metabolism as- bv the collision test: problems of interpretation. Brain Res 112:283- sociated with brain stimulation reward to the medial forebrain bundle 298. of the rat. Brain Res 5 11: l-6. Gallistel CR, Shizgal P, Yeomans JS (198 1) A portrait ofthe substrate Prado-Alcala R, Wise RA (1984) Brain stimulation reward and do- for self-stimulation. Psvchol Rev 88:228-273. pamine terminal fields. I. Caudate-putamen, nucleus accumbens and Gallistel CR, Gomita Y,- Yadin E, Campbell KA (1985) Forebrain amygdala. Brain Res 297~265-273. origins and terminations ofthe medial forebrain bundle metabolically Radhakishun FS, van Ree JM, Westerink BHC (1988) Scheduled eat- activated by rewarding stimulation or reward-blocking doses of pi- ing increases dopamine release in the nucleus accumbens of food- mozide. J Neurosci 5: 1246-I 26 1. deprived rats as assessed with on-line brain dialysis. Neurosci Lett Gomita Y, Gallistel CR (1982) Effects ofreinforcement-blocking doses 85:35 l-356. of pimozide on neural systems driven by rewarding stimulation of Roberts DCS, Koob GF, Klonoff P, Fibiger HC (1980) Extinction and the MFB: a I%-2-deoxyglucose analysis. Pharmacol Biochem Behav recovery of cocaine self-administration following 6-hydroxydopa- 17:841-845. mine lesions of the nucleus accumbens. Pharmacol Biochem . Behav-_.__. Gratton A, Hoffer BJ, Gerhardt GA (1988) Effects of electrical stim- 12:78 l-787. ulation of brain reward sites on release of dopamine in rat: an in vivo Rolls ET (197 1a) Involvement of brainstem units in medial forebrain electrochemical study. Brain Res Bull 2 I:3 19-324. bundle self-stimulation. Physiol Behav 7:297-3 10. Groenewegen HJ, Vermeulen-Van der Zee E, Te Kortschot A, Witter Rolls ET (197lb) Contrasting effects of hypothalamic and nucleus MP (1987) Organization of the projections from the subiculum to accumbens septi self-stimulation on brain stem single unit activity the ventral striatum in the rat. A study using anterograde transport and cortical arousal. Brain Res 31:275-285. of Phaseolus vulgaris leucoagglutinin. Neuroscience 23: 103-I 20. Rompre P-P, Shizgal P (1986) Electrophysiological characteristics of Hakan RL, Henriksen SJ (1989) Opiate influences on nucleus accum- neurons in forebrain regions implicated in self-stimulation of the bens neuronal electrophysiology: dopamine and non-dopamine mech- medial forebrain bundle in the rat. Brain Res 364:338-349. anisms. J Neurosci 9:3538-3546. Ryan LJ, Young SJ, Groves PM (1986) Substantia nigra stimulation Hemandez L, Hoebel BG (1988) Food reward and cocaine increase evoked antidromic responses in rat neostriatum. Exp Brain Res 63: extracellular dopamine in the nucleus accumbens as measured by 449460. microdialysis. Life Sci 42: 1705-I 7 12. Ryan LJ, Diana M, Young SJ, Groves PM (1989) Dopamine Dl Hoebel BG, Monaco AP, Hemandez L, Aulisi EF, Stanley BG, Lenard heteroreceptors on striatonigral axons are not stimulated by endoe- L (1983) Self-injection of amphetamine directly into the brain. Psy- enous dopamine either tonically or after amphetamine: evidence from chopharmacology 8 1: 158-I 63. terminal excitabilitv. Exp Brain Res 77:161-165. Joseph MH, Hodges H (1990) Lever pressing for food reward and Segal M, Bloom FE (1976) The action of in the rat changes in dopamine turnover and uric acid in rat caudate and nucleus hippocampus. III. Hippocampal cellular responses to locus coeruleus accumbens studied chronically by in vivo voltammetry. J Neurosci stimulation in the awake rat. Brain Res 107:499-5 1 1. Methods 34:143-149. Sesack SR, Pickel VM (1990) In the rat medial nucleus accumbens, Kellev AE, Gauthier AM, Lang CG (1989) Amphetamine microin- hippocampal and catecholaminergic terminals converge on spiny neu- jections into distinct striatal subregions cause dissociable effects on rons and are in apposition to each other. Brain Res 527:266-279. motor and ingestive behavior. Behav Brain Res 35:27-39. Shizgal P, Schindler D, Rompre P-P (1989) Forebrain neurons driven Lipski J (198 I)” Antidromic activation of neurones as an analytic tool by rewarding stimulation of the medial forebrain bundle in the rat: in the study of the . J Neurosci Methods 4: l- comparison of psychophysical and electrophysiological estimates of 32. refractory periods. Brain Res 499:234-248. Lopes da Silva FH, Amolds DEAT, Neijt HC (1984) A functional Smith AD,-Bolam JP (1990) The neural network of the basal ganglia link between the limbic cortex and ventral striatum: physiology of as revealed by the study of synaptic connections of identified struc- the subiculum accumbens pathway. Exp Brain Res 55:205-214. tures. Trends Neurosci 13:259-265. Lyness WH, Friedle NM, Moore KE (1979) Destruction of dopami- Somogyi P, Bolam JP, Smith AD (198 1) Monosynaptic cortical input nergic nerve terminals in nucleus accumbens: effect on d-amphet- and local axon collaterals of identified striatonigral neurons. A light amine self-administration. Pharmacol Biochem Behav 11:553-556. and electron microscopic study using the Golgi-peroxidase transport- Maeda H, Mogenson GJ (1980) An electrophysiological study of in- degeneration procedure. J Comp Neurol 195:567-584. puts to neurons of the ventral tegmental area from the nucleus ac- Stellar JR, Corbett D (1989) Regional neuroleptic microinjections cumbens and medial preoptic anterior hypothalamic areas. Brain Res indicate a role for nucleus accumbens in lateral hypothalamic self- 1971365-377. stimulation reward. Brain Res 477: 126-143. Mogenson GJ (1987) Limbic-motor integration. Prog Psychobiol Taylor JR, Robbins TW (1984) Enhanced behavioural control by Physiol Psycho1 12: 117-I 70. conditioned reinforcers following microinjections of d-amphetamine Nakahara D, Ozaki N, Miura Y, Miura H, Nagatsu T (1989) Increased into the nucleus accumbens. Psychopharmacology 84:4054 12. dopamine and metabolism in the rat nucleus accumbens Taylor JR, Robbins TW (1986) 6-Hydroxydopamine lesions of the produced by intracranial self-stimulation of medial forebrain bundle nucleus accumbens, but not of the caudate nucleus, attenuate en- as measured by in vivo microdialysis. Brain Res 495: 178-l 8 1. hanced responding with reward-related stimuli produced by intra- Nauta WJH, Smith GP, Faull RLM, Domesick VB (1978) Efferent accumbens d-amphetamine. Psychopharmacology 90:39&397. connections and nigral afferents of the nucleus accumbens septi in Thierry AM, Deniau JM, Herve D, Chevalier G (1980) Electrophys- the rat. Neuroscience 3:385-t0 1. iological evidence for non-dopaminergic mesocortical and mesolim- Nieuwenhuys R, Geeraedts LMG, Veening JG (1982) The medial bit neurons in the rat. Brain Res 201:2 10-214. forebrain bundle of the rat: I. General introduction. J Comp Neurol Totterdell S, Smith AD (1989) Convergence of hippocampal and do- 206:49-8 1. paminergic input onto identified neurons in the nucleus accumbens Olds J, Milner P (1954) Positive reinforcement produced by electrical of the rat. J Chem Neuroanat 2:285-298. stimulation of septal area and other regions of rat brain. J Comp Trowill JA, Panksepp J, Gandelman R (1969) An incentive model of Physiol Psycho1 45:555-564. rewarding brain stimulation. Psycho1 Rev 76:264-28 I. Paxinos G, Watson C (1986) The rat brain in stereotaxic coordinates, Ungerstedtt (197 1) Stereotaxic mapping ofthe monoamine pathways 2d ed. New York: Academic. in the rat brain. Acta Physiol Stand [Suppl] 367: 148. Pettit HO, Ettenberg A, Bloom FE, Koob GF (1984) Destruction of Walaas I, Fonnum F (1980) Biochemical evidence for gamma-ami- 12 Wolske et al. * Response of Accumbens Neurons to VTA Self-Stimulation

nobutyrate containing fibres from the nucleus accumbens to the sub- ability of hippocampal-accumbens neurons via D2 receptor: role of stantia nigra and ventral tegmental area in the rat. Neuroscience 5: dopamine in presynaptic inhibition. J Neurosci 6:247C-2478. 63-72. Yeomans JS (i979) Absolute refractory periods of self-stimulation West MO, Wolske M (1987) Electrophysiological characterization of neurons. Physiol Behav 22:9 1 l-9 19. single unit activity in the nucleus accumbens of the freely moving rat. Yeomans JS (1982) The cells and axons mediating medial forebrain Sot Neurosci Abstr 14:923. bundle reward. In: The neural basis of feeding and reward (Hoebel West MO, Woodward DJ (1984) A technique for microiontophoretic BG, Novin D, eds), pp 405-417. Brunswick, ME: Haer Institute. study of single neurons in the freely moving rat. J Neurosci Methods Yeomans JS (1989) Two substrates for medial forebrain bundle self- 11:179-186. stimulation: myehnated axons and dopamine axons. Neurosci Biob- White FJ, Wang RY (1986) Electrophysiological evidence for the ex- ehav Rev 13:9 l-98. istence of both D-l and D-2 dopamine receptors in the rat nucleus Yeomans JS, Maidment NT, Bunney BS (1988) Excitability properties accumbens. J Neurosci 6:274-280. of medial forebrain bundle axons of A9 and A10 dopamine cells. Wise RA (1980) Action of drugs of abuse on brain reward systems. Brain Res 450:86-93. Pharmacol Biochem Behav l3ISupol 11:213-223. Yim CY, Mogenson GJ (1980a) Electrophysiological studies of neu- Wise RA, Rompre P-P (1989) Brain hopamine and reward. Annu Rev rons in the ventral tegmental area of Tsai. Brain Res 18 1:301-3 13. Psycho1 40: 19 1-225. Yim CY, Mogenson GJ (1980b) Effect of picrotoxin and nipecotic Wolske M, Rompre P-P, Wise RA, West MO (1990) Nucleus accum- acid on inhibitory response of dopaminergic neurons in the ventral bens single unit activity in the rat during VTA self-stimulation. Sot tegmental area to stimulation of the nucleus accumbens. Brain Res Neurosci Abstr 16:592. 1991466472. Yang CR, Mogenson GJ (1986) Dopamine enhances terminal excit-