Cocaine-induced brain activation detected by dynamic manganese-enhanced magnetic resonance imaging (MEMRI)

Abstract

Dynamic manganese-enhanced magnetic resonance imaging (MEMRI) detects neuronal activity based on the passage of Mn²⁺ into active neurons. Because this mechanism is independent of any hemodynamic response, it is potentially ideal for pharmacological studies and was applied to investigate the acute CNS effects of cocaine in the rat. Dose-dependent, region-specific MEMRI signals were seen mostly in cortical and subcortical mesocorticolimbic structures. To verify the spatial accuracy and physiological mechanisms of MEMRI, neuronal activation following electrical forepaw stimulation revealed somatotopic signal enhancement in the primary and secondary somatosensory cortices, which was blocked by diltiazem, a Ca²⁺ channel antagonist. These data suggest that MEMRI may serve as a tool for investigating the effects of pharmacological agents and opens an application of MRI to study CNS drug effects at a systems level.

Substance abuse and addiction remain a serious social problem around the world (1). The adaptations that occur in brain structure and function following chronic drug exposure appear to be long-lasting and implicate multiple brain circuits, including those involved in reward, motivation, learning, inhibitory control, and executive function (2). Such neuroadaptations may be responsible, at least in part, for the generally poor outcomes after behavioral and/or pharmacological treatment for most drug addictions (3). However, the development of new pharmacological interventions requires understanding the effects of potential medications at a neural systems level, underscoring the clear and critical need for sensitive methods to globally map the complex CNS effects of pharmacological actions.

Functional MRI (fMRI) has been used to regionally map various sensory, motor and higher cognitive functions (4). It has also been applied more recently to understand the CNS sites and mechanisms of action of various pharmacological agents (for review, see ref. 5). fMRI is based on the hypothesis, originally proposed by Roy and Sherrington (6), that neuronal activity is accompanied by tightly coupled local changes in cerebral blood flow and metabolism. Recent studies (7–9) further suggest that excitatory synaptic activity, manifest as changes in local field potentials, tightly correlates with the hemodynamic signals; this coupling is predominantly nonlinear and local context-sensitive. Applying fMRI to pharmacological research introduces yet another challenge: drugs may act not only upon neuronal receptors but also potentially directly upon the cerebral vascular system or indirectly via changes in autonomic physiology (10). Such nonspecific effects could induce hemodynamic changes superimposed on those produced by specific drug-induced neuronal activity, confounding the interpretation of the fMRI signal.

Manganese-enhanced MRI (MEMRI) is an alternative approach to map neuronal activity (11) and is based on the ability of manganese (Mn²⁺) to serve as a calcium (Ca²⁺) analog. In vitro studies in isolated cells (12, 13) and in vivo mouse heart (14) demonstrate that Mn²⁺ enters cells through ligand- or voltage-gated Ca²⁺ channels, an effect blocked by Ca²⁺ channel antagonists. Once translocated, Mn²⁺ remains intracellular for a prolonged period (15). Critically, Mn²⁺ is a potent MRI relaxation agent (16). The accumulation of Mn²⁺ into activated neurons leads to signal enhancement in spin lattice relaxation time (T₁)-weighted images. Thus, unlike fMRI techniques that detect hemodynamic changes coupled to metabolic needs secondary to neuronal activity, MEMRI directly maps only those cells involved in electrical transmission. Furthermore, because MEMRI does not rely upon hemodynamic transduction, it may be especially useful to monitor drug-induced neuronal activation independent of potentially confounding vasculature effects.

Cocaine is one of the most commonly used and potent addictive drugs. Acute cocaine administration produces marked psychostimulant and rewarding effects in both humans and experimental animals (17). With repeated use, cocaine can lead to addiction, characterized by compulsive drug-seeking and intake, despite the development of significant adverse consequences (18). Extensive studies during the past several decades demonstrate that the mesocorticolimbic (MCL) dopamine (DA) system, which originates from the midbrain ventral tegmental area and projects predominantly to the nucleus accumbens (NAc) and prefrontal cortex, is the central neuronal circuitry underlying cocaine's reinforcing actions (19–22). Previous fMRI studies have demonstrated regionally restricted activations in humans and experimental animals after acute cocaine administration (23–27). Nevertheless, a major outstanding issue in these studies is the inability to separate cocaine's specific neuronal actions from its potential direct effects on the cerebrovasular system (10).

In the present study, we used MEMRI to directly map cocaine-induced neuronal activation in rat brain. We found that acute cocaine administration produced a dose-dependent neuronal activation in both cortical and subcortical brain regions consistent with its known behavioral and pharmacological effects, and presumably independent of any cocaine-induced hemodynamic changes. To examine the spatial accuracy of this technique, we also mapped neuronal activity during forepaw electrical stimulation. Localized signal enhancement in the primary and secondary somatosensory cortices was evident, the Ca²⁺ dependency of which was demonstrated by the use of diltiazem, a Ca²⁺ channel antagonist. This work provides a comprehensive demonstration that dynamic MEMRI can be applied to investigate the sites and mechanisms of action of pharmacological agents at a systems level.

Results

Signal Changes After Continuous Mn²⁺ Infusion and Bolus Mannitol Injection.

After MnCl₂ infusion, a gradual signal enhancement was seen in the muscles of the head due to the potent T₁ effect of Mn²⁺ in blood. About 10 min later, the ventricular signal started to increase. In contrast, signal changes within brain parenchyma were not observed until the blood–brain barrier (BBB) was disrupted by mannitol administration, consistent with the fact that the BBB has limited permeability to Mn²⁺ (28). Fig. 1 shows characteristic time courses from the NAc (Fig. 1A), skull muscle (Fig. 1B), and the lateral ventricle (Fig. 1C). The signal enhancements in Fig. 1 B and C before mannitol administration (blue arrow) suggest substantial increases in local Mn²⁺ concentrations.

Fig. 1.

Time course plots from three ROIs showing signal enhancement during i.v. MnCl₂ infusion. Times courses from one voxel each in the NAc (A), skull muscle (B), and lateral ventricle (C) regions are shown. During MnCl₂ infusion (indicated by the gray box), MEMRI signal remained at the baseline level in A until the BBB was disrupted by bolus injection of 25% mannitol (indicated by the blue arrow), which resulted in an abrupt signal increase, and reached a plateau 25 min thereafter. i.v. injection of 0.5 mg/kg cocaine (indicated by the red arrow) led to a further signal increase. Signals in B and C increased continuously during MnCl₂ infusion, and decreased after the cessation of Mn²⁺ infusion.

Bolus injection of 25% mannitol resulted in abrupt signal enhancement in all brain regions where the BBB was disrupted and which reached a plateau level ≈25 min after mannitol injection (signal time course between the blue and red arrows in Fig. 1A). The signal in the ventricle and skull muscles continued to increase, consistent with the continuous Mn²⁺ infusion. Cocaine administration (0.5 mg/kg) further increased the NAc signal (red arrow in Fig. 1), whereas the ventricle and muscle signal decreased after the cessation of Mn²⁺ infusion, consistent with its known rapid washout from tissue.

Acute Cocaine-Induced Neuronal Activation.

Many brain structures exhibited signal enhancement following acute cocaine administration. To quantify the MEMRI response, the plateau level after mannitol administration served as a new baseline, with each voxel's time course normalized to this new baseline and averaged across animals. Fig. 2 shows MEMRI time courses in the NAc after saline and 0.5 and 2 mg/kg of cocaine. As indicated, signal changes after saline injection were within ± 1% relative to the baseline. However, cocaine administration resulted in signal increases between 3 and 6% that lasted for the duration of the scan session.

Fig. 2.

Averaged MEMRI response time course in the NAc from animals receiving saline (n = 6) and 0.5 mg/kg (n = 5) and 2.0 mg/kg (n = 6) cocaine. All time courses were normalized to the baseline signal after bolus injection of mannitol, but before the injection of cocaine or saline. Signals before BBB disruption were not included for graphic clarity.

Fig. 3 illustrates the results of a one-way ANOVA superimposed onto high-resolution structural images (Fig. 3 Left); a stereotaxic atlas template is shown to facilitate anatomical localization (Fig. 3 Right). Regions significantly activated by cocaine were mostly cortical and subcortical limbic, and included olfactory tubercule, frontal, medial prefrontal (mPFC), prelimbic, cingulate and insular cortex, NAc, caudate-putamen, and globus pallidus. In most cases, only the BBB in the hemisphere ipsilateral to the carotid catheter was disrupted; uniform disruption of the entire brain was only observed occasionally (see Methods).

Fig. 3.

Acute cocaine-induced brain activation. Activation maps were superimposed onto T₂-weighted structural images with corresponding rat brain atlas section (53) shown on the right. Activated voxels (based on a one-way ANOVA, thresholded at P < 0.05) are clustered in the hemisphere with the BBB disrupted by hyperosmolar mannitol. The contralateral hemisphere had an intact BBB and did not show activation. Activated structures include: olfactory cortex, medial, ventral and lateral orbital cortex, prelimbic cortex, cingulate cortex, NAc, caudate putamen, ventral pallidus, external globus pallidus, agranular insular cortex, thalamus, hypothalamus, retrosplenial dysgranular cortex, hippocampus, and primary and secondary somatosensory and motor cortex.

Quantitative MEMRI responses in several anatomically defined regions of interest (ROIs) were calculated and are shown in Table 1. In each region, only activated voxels were included in the calculations. Two patterns of activation are evident: in some regions, such as the olfactory tubercule, ventral and lateral orbital cortex, prelimbic cortex, and retrosplenial dysgranular cortex, similar magnitude MEMRI responses were seen to both cocaine doses. In contrast, a second group of regions including the NAc, cingulate cortex (Cg1 and Cg2), dysgranular and granular insular cortex (DI, GI), thalamus, hypothalamus, S1FL, S1BF, and primary and secondary motor cortices (MI and MII) demonstrated consistent dose–response properties, with the MEMRI responses after 2.0 mg/kg cocaine significantly greater than that after 0.5 mg/kg, which in turn was greater than after saline administration. No regions demonstrated a decrease in signal after cocaine administration [see supporting information (SI) Fig. 7 for single dose cocaine maps].

Table 1.

Dose-dependent response of the MEMRI signal after cocaine infusion

Region	Coordinate			Vox no.	Vol, μl	Signal changes, %			Statistics
	x	y	z			Sal. (I)	COC0.5 (II)	COC2.0 (III)	I vs. II	I vs. III	II vs. III
NAc	1.8	7.5	1.64	133	9.94	0.59 ± 0.67	4.56 ± 0.89	6.29 ± 0.42	<0.005	<0.001	<0.05
CPu	2.5	5.5	0.64	161	12.04	0.1 ± 0.53	3.99 ± 0.59	4.72 ± 0.65	<0.001	<0.001	NS
Cg I/II	0.5	1.5	0.64	46	3.44	−0.15 ± 1.16	4.8 ± 1.18	8.9 ± 1.08	<0.01	<0.001	<0.05
DI	5.8	7.0	−0.36	50	3.74	1.28 ± 0.50	3.35 ± 0.74	7.45 ± 0.45	<0.02	<0.00001	<0.001
GI	4.6	6.0	1.64	54	4.04	1.28 ± 0.41	3.03 ± 0.88	7.35 ± 0.41	0.06	<0.0001	<0.001
Hipp	4.0	7.8	−4.36	57	4.26	0.37 ± 0.39	6.13 ± 0.84	6.27 ± 0.87	<0.0001	<0.0001	NS
Orb	0.8	5.5	3.64	47	3.51	−0.57 ± 0.89	4.25 ± 0.73	4.14 ± 0.78	<0.005	<0.005	NS
Olf	1.0	6.5	4.64	111	8.30	−2.2 ± 1.39	4.32 ± 1.2	5.41 ± 0.53	<0.005	<0.001	NS
Prl	0.5	4.6	2.64	35	2.62	−0.7 ± 1.06	4.74 ± 0.92	4.55 ± 0.94	0.002	<0.005	NS
IL	0.8	5.0	3.64	17	1.27	−0.43 ± 1.54	6.19 ± 0.84	4.71 ± 0.91	<0.005	<0.01	NS
GP	3.5	7.0	−1.36	53	3.96	0.12 ± 0.59	7.26 ± 0.85	5.1 ± 0.46	<0.001	<0.001	<0.05
RSD	1.2	1.5	−2.36	19	1.42	0.56 ± 1.06	9.01 ± 1.75	8.85 ± 1.32	0.001	<0.001	NS
VP	2.2	7.8	0.64	19	1.42	0.40 ± 0.69	7.72 ± 1.16	6.07 ± 0.85	<0.0005	<0.001	NS
Hypothal	0.6	9.0	−2.36	14	1.05	0.69 ± 0.39	2.22 ± 0.58	5.38 ± 0.99	0.03	<0.001	<0.01
Thal	3.6	3.8	−3.36	22	1.64	−0.2 ± 0.32	4.36 ± 0.91	6.26 ± 0.71	<0.001	<0.001	<0.05
M I/II	2.0	1.8	−0.36	174	13.01	0.29 ± 0.62	5.41 ± 1.13	8.51 ± 0.97	<0.005	<0.001	<0.05
SIBF	5.5	3.0	−2.36	84	6.28	0.62 ± 0.56	4.45 ± 0.97	9.17 ± 1.28	<0.005	<0.001	<0.01
SIFL	4.2	1.8	0.64	96	7.18	0.12 ± 0.49	5.69 ± 1.35	8.51 ± 0.97	<0.005	<0.001	<0.05
Aud	6.2	4.8	−4.36	76	5.68	0.31 ± 0.63	5.44 ± 1.15	7.86 ± 1.63	<0.005	<0.005	NS

Vox no. is the number of activated voxels in each region identified by ANOVA statistical maps; Vol is the activated volume (in microliters) calculated based on the number of activated voxels; Sal., saline administration; COC0.5, cocaine at 0.5 mg/kg; COC2.0; cocaine at 2.0 mg/kg; NS, not significant. Olf, olfactory cortex (medial); Orb, ventral and lateral orbital cortex; Prl, prelimbic cortex; Cg I/II, primary and secondary cingulate cortex; CPu, caudate putamen; DI, dysgranular insular cortex; GI, granular insular cortex; NAc, nucleus accumbens; VP, ventral pallidus; Thal, thalamus; Hypothal, hypothalamus; RSD, retrosplenial dysgranular cortex; S1BF, S1 barrel field; S1FL, S1 forelimb region; Aud, auditory cortex; M I/II, primary and secondary motor cortices. GP, globus pallidus; IL, infralimbic cortex; Hipp, hippocampus.

Forepaw Stimulation-Induced Neuronal Activation.

As a positive control procedure, MEMRI experiments were conducted after electrical forepaw stimulation. A two-tailed, voxel-wise t test between the control group (n = 6) and the stimulation group (n = 5) revealed focal signal enhancement only in the forelimb region of the primary (S1FL) and secondary somatosensory cortex (S2), and a small foci in the caudate nucleus (Fig. 4). Notably, we did not detect any activation in regions significantly activated after cocaine challenge (e.g., NAc, mPFC) (see Fig. 3), suggesting that the MEMRI technique accurately depicts stimulus-induced regional neuronal activation.

Fig. 4.

Statistical activation maps after electrical forepaw stimulation. The activation maps were superimposed onto :ml T₂-weighted structural images with the corresponding brain atlas section (53) shown on the right. Signal enhancement in the forelimb region of the somatosensory cortex (S1FL) and the secondary somatosensory cortex (S2) are evident.

Diltiazem Inhibits Forepaw-Stimulation Induced MEMRI Signal.

Diltiazem, an “L” type Ca²⁺ channel antagonist, was administered before the initiation of electrical forepaw stimulation in a third group of animals (n = 4). All other experimental conditions remained unchanged. A significant reduction in the S1FL MEMRI signal was seen after channel blockage (Fig. 5), suggesting that blocking Ca²⁺ channels substantially inhibits Mn²⁺ entry into activated cells.

Fig. 5.

Comparison of MEMRI response with and without the administration of Ca²⁺ channel antagonist. MEMRI signal resulting from electrical forepaw stimulation was significantly attenuated after the administration of diltiazem (2 mg/kg, P < 0.01), a Ca²⁺ channel antagonist. The gray box indicates stimulation period.

Discussion

Acute cocaine administration produced dose-dependent, region-specific increases in MEMRI signals predominantly in MCL cortical and subcortical systems. We also found that electrical forepaw stimulation produced a regional MEMRI signal increase somatotopically in the contralateral S1FL and S2 somatosensory cortex, which was blocked by diltiazem, a Ca²⁺ channel antagonist, suggesting that MEMRI accurately detects localized neuronal activity. Together with previous studies on isolated cells (12) and mouse heart (14), these data suggest that signal enhancement detected by MEMRI is due to neuronal activity-dependent Mn²⁺ cellular translocation via voltage-gated Ca²⁺ channels. Furthermore, Lin and Koretsky (11) previously demonstrated that the MEMRI signal was not altered by hemodynamic CO₂ challenges, suggesting that MEMRI detects direct activity-dependent neuronal activation, thus differentiating it from other fMRI techniques that require various hemodynamic or metabolic surrogates [e.g., cerebral blood flow (CBF), cerebral blood volume (CBV), CMRO₂).

MEMRI has previously been used for in vivo axonal tract tracing in mice (29), birds (30), and primates (31). It has been applied to measuring hypothalamic activation (32–34) and to localize brain activity after kindling (35) and amphetamine (36). More recently, it has been used to map tonotopic auditory activity in the mouse inferior colliculus after i.p. Mn²⁺ administration (37). Interestingly, the latter study did not report activation in the auditory cortex. In view of the current data, temporarily disrupting the BBB and allowing a sufficient amount of Mn²⁺ to enter the brain parenchyma appears to be the most viable method for acute MEMRI mapping experiments. In the present study, we refined the original experimental protocol (11, 38) and developed a method to register data in stereotaxic space, allowing accurate ROI and group-analysis of the MEMRI data, thereby effectively minimizing nonspecific activation and increasing statistical power accordingly. As a result, we were able to detect focal neuronal activity in both cortical and small subcortical regions that would otherwise likely have been very difficult, if not impossible to detect.

Cocaine-Induced Neuronal Activation.

Acute cocaine administration produced a robust MEMRI signal increase in such MCL brain regions as the anterior olfactory cortex, NAc, dorsal striatum, mPFC, and hippocampus (Fig. 3). Activation of these regions, all of which receive high-density ventral tegmental area DA projections, once again supports a critical role for the MCL system in cocaine's acute rewarding and drug-seeking properties (39). Acute cocaine also produced a significant signal increase in the primary and secondary motor cortex and the somatosensory cortex, which may be related to known cocaine-induced behavioral effects, such as enhanced alertness and stereotypic psychomotor activity (40). This activation pattern is in general agreement with previous cocaine-induced hemodynamic blood oxygen level-dependent and cerebral blood volume-weighted fMRI data (26). It is also in general agreement with early 2-deoxyglucose (41, 42) and iodoantipyrene cerebral blood flow autoradiographic studies (43, 44). For example, Porrino et al. (42) reported significant increases in the regional cerebral metabolic rate of glucose (rCMRglu) in mPFC and NAc after IV administration of 0.5 mg/kg cocaine. At doses of 1 mg/kg and above, significant increases in rCMRglu were also observed in the sensorimotor cortex, caudate putamen, and substantia nigra. Notably, we observed substantially more activated regions at 0.5 mg/kg of cocaine, which may be due to sensitivity and response time differences between these two very different techniques. It should also be mentioned that species difference in rCMRglu have been reported after acute cocaine administration, with both human (45) and nonhuman primates (46) showing decreases in metabolism. Consistent with previous rat studies, we did not see any evidence for cocaine-induced decrease in activity using MEMRI.

Substantial data have accumulated pointing to the pivotal role of the MCL system in acute cocaine's psychostimulant and rewarding effects (21). Several studies investigated dopaminergic regulation of cerebral microcirculation (47, 48). The study by Choi et al. (48) is particularly relevant. They found that increases in regional CBV (rCBV) in the frontal cortex, striatum, and thalamus that were induced by DA transporter blockers (such as cocaine) and DA releasers (such as amphetamine) were mediated through activation of D1/D5 receptors. Conversely, D2 and D3 receptor agonists produced rCBV decreases. These results suggest a complex role of DA receptors in mediating neurovascular coupling, which can lead to rCBV increases or decreases. Furthermore, chronic cocaine exposure leads to neuroadaptations in multiple brain circuits, including the DA system. Such neuroadaptations would add another confounding factor to the interpretation of the hemodynamic signal.

A number of recent studies propose fMRI as a promising tool for investigating system level actions of various pharmacological agents (23–27). These studies generally support the hypothesis of tight coupling between neuronal activity, energy metabolism and CBF. However, several studies suggest that this hypothesis may not hold universally under all pharmacological manipulations. For example, methylxanthines such as caffeine and theophylline increase glucose utilization while at the same time decreasing CBF in both humans and animals (49). It was postulated that this was due to methylxanthine competition with adenosine, an important CBF modulator in the CNS. Amphetamine also seems to differentially modify the regulating mechanism between blood flow and metabolism. Choi et al. (48) reported that amphetamine administration induces about the same percent increase in glucose utilization in the globus pallidus and the striatum; however, the globus pallidus shows minimal rCBV changes in both rats and monkeys (50), whereas rCBV changes in the striatum are prominent in both species. MEMRI may be of particular value for such studies as its signal source does not rely upon a hemodynamic response (11). In fact, we found significant MEMRI dose-response in the globus pallidus after acute cocaine challenge.

Technical Considerations.

Although MEMRI is able to image calcium activity of the rodent brain, the present MEMRI technique has several weaknesses: first, due to limited Mn²⁺ permeability, the disruption of the BBB appears to be required for dynamic MEMRI. In this study, we injected hyperosmolar mannitol into the internal carotid artery (ICA), which was then redistributed to the anterior, middle, and posterior cerebral arteries via the Circle of Willis. Thus, it appears technically and anatomically difficult for a single bolus injection of mannitol to be distributed homogenously to both hemispheres. Because we performed carotid catheterization on only one side, the BBB was completely disrupted in most cases in only one hemisphere, resulting in a false-negative outcome in the contralateral hemisphere, which maintained a more or less intact BBB. Although simultaneously injecting mannitol into both ECAs should be able to disrupt the BBB in both hemispheres uniformly, this likely would result in more surgical trauma, and was not explored in the present study. Second, the present MEMRI technique has relatively poor temporal resolution compared with other fMRI techniques due to the relatively slow accumulation of Mn²⁺ within activated neurons. It is possible to increase MEMRI speed by the use of different data acquisition schemes, such as echo-planar imaging (EPI) (51). However, EPI-based techniques generally suffer from limited spatial resolution, image distortion, and severe signal drop-out at air–tissue interfaces. In the present study, we used a traditional T₁-weighted spin echo sequence, which greatly mitigated these problems and thus achieving high spatial resolution.

We did not see any evidence for cocaine-induced decreases in activity using MEMRI. An interesting question remains whether MEMRI is able to detect such a response. Because the MEMRI signal reflects Mn²⁺ accumulation into active cells, it is conceivable that subthreshold neuronal activity and baseline firing leads to an increase in MEMRI signal, as evidenced by the saline signal in Fig. 2. Likewise, small inhibitory interneurons also must be depolarized to release their inhibitory transmitter and would also add to (albeit minimally) to the overall positive MEMRI signal. In contrast, the postsynaptic consequence of such inhibitory actions would likely result in a smaller Mn²⁺ accumulation and thus a smaller MEMRI signal increase. In principle, it may be possible to differentiate MEMRI responses under these two conditions. However, the SNR and the sensitivity of the current technique make it difficult in the present study.

In summary, we demonstrated that MEMRI can be applied to map acute cocaine-induced neuronal activation in both cortical and subcortical brain structures with good spatial accuracy and sensitivity. The MEMRI signal reflects Mn²⁺ accumulation in activated neurons, via voltage-gated Ca²⁺ channels, which was blocked by a Ca²⁺ channel antagonist. Because this technique is independent of any hemodynamic response coupling, it may prove to be a useful tool for investigating system level localization and mechanisms of action of various pharmacological agents.

Methods

Animal Preparation.

Thirty-one male Sprague–Dawley rats (250–350 g) were used in this study; data from five animals were discarded due to inhomogeneous disruption of the BBB (see below). Animal preparation procedures were generally similar to our previous report (52) and were approved by the Animal Care and Use Committee of the National Institute on Drug Abuse. Briefly, animals were anesthetized with 2% isoflurane in 1:1 mixture of oxygen and air. The femoral artery and vein were catheterized for monitoring rat physiology and drug delivery. The right external carotid artery (ECA) was catheterized for delivering hyperosmolar mannitol to disrupt the BBB. Specifically, the right common and internal carotids were temporarily clamped and a MRE-40 catheter was carefully inserted into the ECA in the direction of the common carotid artery (CCA), allowing blood flow to the ICA to remain undisturbed. A tracheotomy was performed for artificial ventilation (SAR-830 ventilator; CWE, Ardmore, PA). End-tidal CO₂ and O₂ were continuously monitored. Core body temperature was maintained at 37.5 ± 0.5°C with a water-circulating pad. After surgery, the animal's head was secured with bite bar and ear bars and positioned within the center of the magnet using a customized animal holder. After surgical preparations, anesthesia was switched to α-chloralose with an initial dose of 50 mg/kg i.v. followed by bolus 50 mg/kg each hour. A neuromuscular blocking agent (gallamine triethiodide, 80 mg/kg i.v.) was administrated to further minimize motion artifacts.

MRI Scan.

MRI experiments were carried out using a Bruker Biospin 9.4T scanner (Bruker, Karlsruhe, Germany). High-resolution T₂-weighted anatomical images were acquired by using a rapid acquisition with relaxation enhancement (RARE) sequence [reaction time (RT) = 2,000 ms, effective echo time (TE) = 50 ms, RARE factor = 8]. The anterior commissure was used as a reference point (53). T₁-weighted coronal images were acquired by using a conventional spin echo sequence. Scan parameters were: TR = 450 ms, TE = 8 ms, slice thickness = 1 mm, spectral width = 50 kHz, field of view = 3.5 cm, matrix size = 128 × 128. The in-plane resolution was 0.27 mm. A customized program was developed to reconstruct images and display voxel-wise time courses in a real-time fashion using AFNI (54). The results of BBB disruption after mannitol injection were readily evident using this display.

Experimental Paradigm.

After anatomical localization scans, three baseline T₁-weighted images were acquired. Continuous i.v. injection of 10% MnCl₂-4H₂O (Sigma, St. Louis, MO) was then initiated by using an infusion pump (Harvard Apparatus, Holliston, MA). After infusing 0.15 mmol/kg Mn²⁺ in 35 min, a bolus injection (≈25 s) of 25% mannitol (5–7 ml/kg; Sigma) was performed through the ECA catheter to disrupt the BBB. Mn²⁺ infusion continued for another 25 min at a reduced infusion rate (preliminary studies showed that the MRI time course stabilized after ≈25 min following BBB disruption). The total amount of Mn²⁺ was 0.2 mmol/kg. Because the intravascular half-life of Mn²⁺ is very short (16), continuous infusion of Mn²⁺ after BBB disruption was very important to allow the MRI signal to reach a stable plateau. Starting 10 min before and lasting until 10 min after mannitol-induced BBB disruption, 2% isoflurane was delivered to produce a deep anesthetic state. Preliminary experiments revealed that isoflurane reduced MEMRI signal resulting from nonspecific stimuli (e.g., sound), and prevented brain swelling from the hyperosmolar mannitol injection. Five minutes after MnCl₂ injection termination, either cocaine hydrochloride (0.5 mg/kg, n = 5; 2.0 mg/kg, n = 6) or saline (n = 6) was injected i.v., and T₁-weighted images were continuously acquired for 45 min. Fig. 6 summarizes the experimental procedure.

Fig. 6.

Diagram of the experimental design. Animals first underwent a 10-min baseline scan, followed by continuous MnCl₂ infusion for 60 min. Bolus injection of hyperosmolar mannitol was delivered through the ECA catheter 35 min after the initiation of MnCl₂ infusion. Five minutes after the cessation of MnCl₂ infusion, animals received either cocaine or saline, and the scan continued for another 45 min.

To examine MEMRI anatomic specificity, a forepaw stimulation experiment (n = 5) was performed by using a similar protocol as above, except that the drug challenge was replaced with electrical stimulation. The stimuli, consisting of 30 cycles of alternating 50 sec on and 10 sec off, were delivered at 3 Hz, pulse width = 3 ms, current = 3 mA. The stimulus duration was 35 min. To test the hypothesis that Mn²⁺ enters cells via voltage gated Ca²⁺ channels, diltiazem a Ca²⁺ channel blockage (Sigma) was infused 10 min before the electrical stimulation (2.5 mg/kg IV, n = 4), and continued during the entire stimulus period.

Data Analysis.

To facilitate group comparisons, images were manually registered from each animal onto a common 3D space within the AFNI framework (54). T₂-weighted images from one animal that had the best positioning were selected as template images. Five prominent anatomical reference points were identified from the template images (anterior commissure, anterior horns of the anterior commissure, posterior commissure, and the cross point of the mid-sagittal line and corpus callosum; see SI Fig. 8). The same reference points were identified in each set of structural images to be registered. A transformation matrix including linear transformation and rotational elements was generated based on these reference points and applied to the data set acquired during the MEMRI scan.

After image registration, 3D MEMRI data sets were spatially filtered using a Gaussian filter with a full width at half maximum (FWHM) of 0.4 mm. Time course data were detrended as needed. Signal change due to Mn²⁺ infusion was calculated as: 100 × (S₁ − S₀)/S₀, where S₁ was signal intensity before mannitol infusion, and S₀ was averaged signal intensity during the baseline scan. A fractional signal enhancement map was generated to identify regions that exhibited significant signal increases during Mn²⁺ infusion.

After bolus injection of hyperosmolar mannitol, the MR signal abruptly increased in regions where the BBB had been disrupted, and then stabilized within 25 min. The last three data points during the stabilization period were averaged to represent the new baseline signal (S₂). In the Ca²⁺ channel experiment, S₂ was the mean baseline signal intensity before the initiation of forepaw electrical stimulation. The last three data points of the MEMRI scan were averaged to represent signal intensities after cocaine or saline injection (S₃). Fractional signal changes resulting from cocaine or saline injection were calculated as: 100 × (S₃ − S₂)/S₂, and were subject to one-way ANOVA.

In addition to the above group analysis, ROI analyses were conducted on several a priori hypothesized cortical and subcortical limbic related brain regions. Voxel-wise fractional signal changes after cocaine and saline administration were calculated as above; only data from significant voxels within each ROI were included in the average. Data between the control group and the experiment groups were subject to one-way ANOVA and Student t tests. Data are presented as mean ± SEM unless otherwise specified. All calculations were performed in AFNI (54) and Matlab (The MathWorks, Natick, MA).

Abbreviations

fMRI

functional MRI

MEMRI

manganese-enhanced MRI

MCL

mesocorticolimbic

DA

dopamine

NAc

nucleus accumbens

BBB

blood–brain barrier

mPFC

medial prefrontal cortex

ROI

region of interest

CBV

cerebral blood volume

CBF

cerebral blood flow.