Relapse to Cocaine-Seeking After Hippocampal Theta Burst Stimulation

See allHide authors and affiliations

Science  11 May 2001:
Vol. 292, Issue 5519, pp. 1175-1178
DOI: 10.1126/science.1058043


Treatment efforts for cocaine addiction are hampered by high relapse rates. To map brain areas underlying relapse, we used electrical brain stimulation and intracranial injection of pharmacological compounds after extinction of cocaine self-administration behavior in rats. Electrical stimulation of the hippocampus containing glutamatergic fibers, but not the medial forebrain bundle containing dopaminergic fibers, elicited cocaine-seeking behavior dependent on glutamate in the ventral tegmental area. This suggests a role for glutamatergic neurotransmission in relapse to cocaine abuse. The medial forebrain bundle electrodes supported intense electrical self-stimulation. These findings suggest a dissociation of neural systems subserving positive reinforcement (self-stimulation) and incentive motivation (relapse).

Cocaine addiction is a chronic brain disorder with psychosocial and neurobiological determinants (1). Treatment efforts are hampered by relapse (2). Imaging techniques have been applied to study the neural substrates of cocaine craving (3–6). These studies, although informative, address subjective craving, not actual relapse. They are correlational, not causal, and they take place in laboratory settings, not the actual context of the cocaine experience. Complementary approaches to mapping brain areas underlying relapse are therefore desirable.

Reinstatement of cocaine-seeking behavior in the rat has high validity for relapse in the human addict (7–9). Reinstatement and relapse are operant responses. The rat and human share common triggers of relapse: cocaine (10,11), stress (12, 13), and stimuli conditioned to cocaine (6, 14, 15). Thus, reinstatement in the rat is an objective measure of relapse with predictive value for the human.

The environmental context of the cocaine experience is a powerful determinant of cocaine-seeking behavior in the rat and human (15, 16). The hippocampus underlies the learning of associations between the environmental context and unconditioned stimuli (17–19) (e.g., cocaine). Stimulation of the ventral subiculum (VSUB) of the hippocampal formation induces long-lasting dopamine (DA) release in the nucleus accumbens (NAC) (20–23) and enhances the firing of mesolimbic DA neurons that originate in the ventral tegmental area (VTA) (23, 24). NAC DA has been implicated in reinstatement (25–27). Hence, VSUB stimulation may elicit reinstatement.

We catheterized the jugular vein of Long-Evans rats and implanted an electrode into the VSUB, cerebellum (CBL), or medial forebrain bundle (MFB) (28). The rats self-administered cocaine intravenously (i.v.) during daily 3-hour sessions in operant chambers equipped with one “active” and one “inactive” lever. Active lever presses resulted in cocaine delivery (1.0 mg/kg per infusion) and a light signal; inactive lever presses were counted but had no further consequence. After 1 week of stable cocaine self-administration, saline was substituted for cocaine. The lever pressing progressively diminished, a behavioral phenomenon called “extinction.” The extinction period varied between 7 and 20 days. When the extinction criterion was met (three consecutive sessions with fewer than 10 lever presses), priming electrical stimulations were administered (28).

Neither sham nor 2-Hz VSUB stimulation elicited reinstatement (Fig. 1B). However, brief (8 s) theta burst stimulation mimicking rhythms recorded electroencephalographically in the hippocampus (29) elicited reinstatement (Fig. 1, A and B). All reinstated lever presses occurred after the end of electrical stimulation; no lever presses occurred during the stimulation (Fig. 1A).

Figure 1

(A) Effect of VSUB theta burst stimulation (arrow) on reinstatement in an individual rat. Upward bars: active lever presses; downward bars: inactive lever presses. For clarity, only the first hour of the 3-hour session is shown. (B) Effect of different patterns of VSUB electrical stimulation in a group of rats (n = 9). The black bars show “active” lever presses (mean ± SEM), the white bars “inactive” lever presses (mean ± SEM). During “sham” stimulation, no actual stimulation was delivered. 2 Hz: 2-Hz repetitive stimulation; THETA: stimulation in “theta burst” rhythm. Asterisk indicates significant difference compared with sham and 2-Hz groups (*P < 0.00001). There were no significant differences in inactive lever presses among sham, 2-Hz, and theta burst treatment groups.

VSUB stimulation enhances VTA DA neuron firing (23,24). NAC DA increase after VSUB stimulation depends on VTA glutamate (GLU), because the NAC DA increase is blocked by the nonselective ionotropic GLU receptor antagonist kynurenic acid (KYN) applied into the VTA (23). Therefore, reinstatement after VSUB stimulation may depend on VTA GLU neurotransmission. Microinjection of KYN (50 nmol/0.5 μl), but not 0.5 μl vehicle (VEH), into the VTA blocked reinstatement after VSUB burst stimulation (Fig. 2) (30). This suggests the involvement of the AMPA and/orN-methyl-d-aspartate (NMDA) ionotropic GLU receptor in reinstatement. The ionotropic GLU receptor agonist NMDA in the VTA enhances NAC DA (31, 32). Microinjection of NMDA (83 pmol/0.5 μl), but not 0.5 μl VEH, into the VTA of the same subjects elicited reinstatement (Fig. 2).

Figure 2

Effect of microinjection of VEH (0.5 μl) or KYN (50 nmol/0.5 μl) into the VTA on reinstatement elicited by VSUB theta burst stimulation (n = 6); effect of microinjection of VEH (0.5 μl) or NMDA (83 pmol/0.5 μl) into the VTA on reinstatement (n = 6). VEH: vehicle; KYN: kynurenic acid; E: theta burst stimulation. Legend as in Fig. 1B. Asterisks indicate significant difference compared with the VEH only group (*P < 0.00001) (active lever presses).

An anatomical control group received CBL stimulation (28) (Fig. 3). The CBL supports electrical self-stimulation (33–35) and is activated after exposure to cues eliciting cocaine craving (4, 5). Neither 2-Hz nor theta burst stimulation elicited reinstatement (Fig. 3). Reinstatement did occur after a noncontingent i.v. cocaine prime (2.0 mg/kg) (Fig. 3), indicating capability of reinstatement.

Figure 3

Effect of different electrical stimulation patterns in a group of rats (n = 6) with electrodes in CBL. COC: cocaine prime (2.0 mg/kg i.v.). Legend as in Fig. 1B. The only significant effect occurred after i.v. cocaine prime (*P < 0.00001) (active lever presses).

Another group received MFB stimulation (28) (Fig. 4, A and B). The MFB supports self-stimulation, which depends on NAC DA (36). MFB stimulation induces brief NAC DA release (37–40). Neither 2-Hz nor theta burst stimulation nor a stimulation pattern mimicking the intrinsic burst rhythm of DA neurons (41–43) elicited reinstatement. The cocaine prime control (2.0 mg/kg i.v.) induced reinstatement (Fig. 4B). The MFB electrode placements were physiologically relevant, because they supported self-stimulation (41) (Fig. 4C). MFB stimulation shown to increase running speed for MFB stimulation reward (44) also failed to elicit reinstatement (Fig. 4B).

Figure 4

(A) Effect of MFB theta burst stimulation (arrow) on reinstatement in an individual rat. Only the first hour is shown. Legend as in Fig. 1A. (B) Effect of different stimulation patterns on reinstatement in a group (n = 6) with MFB electrodes. INTR: intrinsic burst rhythm occurring in mesolimbic DA neurons. ICSS: stimulation pattern based on these rats' performance during MFB self-stimulation [see (C)]. Otherwise legend as in Figs. 1 and 2. The only significant difference occurred after a cocaine prime (2.0 mg/kg i.v.) (*P < 0.00001) (active lever presses). (C) Rate-frequency curve of rats with MFB electrodes generated during MFB self-stimulation after finishing the experiment shown in (B) (n = 6, mean ± SEM). All rats showed stable self-stimulation patterns. The optimal self-administered current intensities varied between −251 and −158 μA.

The hippocampus subserves contextual learning (17–19). Reinstatement after VSUB stimulation may reflect the read-out of an encoded association between the context of the cocaine experience (i.e., the operant chamber) and (the previously available) cocaine (45–47). Reinstatement after VTA activation (Fig. 2) (27) concurs with the proposed function of VTA DA neurons as reward predictors (48, 49). Thus, VSUB stimulation may harness a neural substrate involving the VTA that is predictive of cocaine reward.

VSUB burst stimulation elicited reinstatement and preceded lever pressing (Fig. 1, A and B). Thus, VSUB stimulation has predictive or incentive (50–52) properties that facilitate the initiation of lever-press responding. MFB stimulation had no such predictive or incentive properties, because it failed to initiate reinstatement (Fig. 4, A and B). In contrast, during MFB self-stimulation, electrical stimulation followed and positively reinforced (53, 54) lever pressing (Fig. 4C). The hippocampus is much less effective at supporting self-stimulation than the MFB (55–57); i.e., VSUB (self-) stimulation is much less reinforcing than MFB stimulation. Therefore, separate neural systems may subserve positive reinforcing (self-stimulation: MFB) and incentive (reinstatement: VSUB) electrical brain stimulation. Separation of neural systems subserving positive reinforcing and incentive stimuli has been proposed previously (58–60).

The two neural systems presumably share the mesolimbic DA projection, because pharmacological stimulation of mesolimbic DA neurotransmission triggers reinstatement (Fig. 2) (7,25–27) and because VSUB stimulation enhances mesolimbic DA neurotransmission (20–24). Although both MFB and VSUB stimulation increase NAC DA, we propose that the long-lasting (about 30 min) DA release after VSUB stimulation (22)—as opposed to the brief (less than 5 s) release after MFB stimulation (37–40)—is critical for reinstatement. Our finding that reinstatement depends on VTA GLU agrees with neurochemical data (23).

Because the VSUB contains GLU (61–63) and because reinstatement could be blocked by KYN and elicited by NMDA in the VTA (Fig. 2), GLU seems to be involved in cocaine-seeking behavior (27, 64). Therefore, GLU agents appear worthy of pursuit as potential pharmacotherapeutic candidates for cocaine addiction.

  • * To whom correspondence should be addressed. E-mail: robvorel{at}


View Abstract

Navigate This Article