Although manganese can exist in multiple oxidation states, physiological conditions emphasize Mn²⁺ and Mn³⁺. Under in vitro conditions, Mn²⁺ can be oxidized to Mn³⁺ by superoxide radicals (76). Mn³⁺ is more reactive and toxic than Mn²⁺ (47, 62). The failure to detect significant amounts of Mn³⁺ within the brain mitochondria by X-ray absorption near edge structure (XANES) spectroscopy suggests that Mn³⁺ only occurs as a transient state (77). Repetitive intraperitoneal exposure of rats to equimolar doses of Mn²⁺ and Mn³⁺ using Mn(II)Cl₂ and Mn(III)pyrophosphate, respectively, revealed significant higher blood and brain concentrations after Mn³⁺ than after Mn²⁺ exposure (62). However, little is known about the distribution and possible interconversion of Mn²⁺ and Mn³⁺ in the brain. Because of its lower toxicity and biochemical similarity to Ca²⁺, only Mn²⁺ has hitherto been utilized for MEMRI.

In most cases, Mn²⁺ is administered as an aqueous solution of MnCl₂. The salt or the solutions are provided by many manufacturers. Table 7 summarizes products from Sigma Aldrich, Germany. It should be noted that the amount of MnCl₂ administered to an animal is often given in mass units. To calculate the corresponding load of Mn²⁺ in molar units, the different molecular weights of MnCl₂ (125.84 g/mol) and MnCl₂·4 H₂O (197.92 g/mol) have to be taken into account.

In situ perfusion techniques revealed a greater influx rate of ⁵⁴Mn citrate into the brain than of ⁵⁴Mn chloride. The use of pH 8.6 buffer-treated MnCl₂ which has a higher affinity to transferrin leads to a significantly lower brain manganese concentration in comparison to the use of MnCl₂ in Tris-HCl at pH 7.4. Only little is known of how different manganese salts and solvents may affect the MR contrast. So far, MEMRI studies almost exclusively employ MnCl₂ dissolved in deionized water, 5% glucose solution, or 0.9% NaCl solution.

For parenteral applications of Mn²⁺, the injected solution should be biocompatible with the organism or tissue of interest. This refers to parameters such as concentration, chemical characteristics, applied volume, and sterility. For example, the osmolarity of most extracellular body fluids is about 300 mOsM. Because MnCl₂ consists of 3 ions, a 100 mM MnCl₂ solution would result in an isotonic solution. Manganese solutions in distilled water with a lower (higher) concentration are hypotonic (hypertonic). Although iso-osmolarity should be achieved whenever possible to minimize the occurrence of cell necrosis, minor deviations are less critical for intravenous and intraperitoneal injections, because of the large ratio of the target to the injected volume and the expected rapid mixing. For subcutaneous and intracerebral injections, however, strictly isotonic solutions have to be prepared to avoid pain and tissue damage. For MnCl₂ concentrations below 100 mM this condition can be easily achieved by adding sodium chloride. For example, a 10 mM MnCl₂ solution that corresponds to 30 mOsM would have to be dissolved in a 270 mOsM solution of NaCl to reach 300 mOsM.

Another aspect is the pH of the solution. Unbuffered aqueous solutions of MnCl₂ are slightly acidic. Although this may be compensated by plasma or extracellular physiological buffer systems, the pH should be better controlled by adjustment to values near 7.4 using sodium hydroxide or by preparing manganese solutions with a buffer like bicine or Tris-HCl.

Besides pH and osmolarity the toxicity of manganese depends on the local concentration (Table 3). In cell culture, a direct cytotoxic effect of concentrations as low as 1 mM MnCl₂ has been reported, but not for MnCl₂ concentrations of 10 μM (78). The lowest possible concentration for MEMRI, however, is limited by the amount of manganese required to yield sufficient contrast in the target region and the tolerable solution volume. Table 8 summarizes the recommended maximum injection volumes for different routes of application. For subcutaneous injections the administration of manganese may be distributed to several locations in order to reduce local concentrations (see Note 1).

Delivery of Mn²⁺ to the central nervous system may be achieved using three different routes: (1) systemic applications with manganese entering the brain via the blood, (2) direct intracerebral or intraventricular injections, and (3) nostril administration. For systemic applications, the common routes are intravenous, intraperitoneal, and subcutaneous injections. In principle, oral applications are also possible, but less well controlled as absorption from the gastrointestinal tract can be influenced by many factors. These include the chemical form (79), the presence of other minerals and dietary constituents (33, 80, 81), and the age of the animal (82–84). Furthermore, adaptive changes to high dietary manganese intake lead to reduced absorption and enhanced hepatic (biliary) or pancreatic excretion of manganese (19, 85–88).

A major advantage of the intravenous application is the fast delivery to the systemic circulation. This is utilized for activity-induced MEMRI where the accumulation of MnCl₂ in conjunction with a brief opening of the BBB may be related to brain activity during stimulation. On the other hand, intravenous bolus injections of manganese can lead to a high transient blood concentration and bear the risk of altered heart function. In fact, injections of 0.01 mmol/kg Mn²⁺ over 30 s in dogs and rabbits caused prolongation of PR and QTc intervals in the electrocardiogram, while 0.2 mmol/kg Mn²⁺ led to ventricular fibrillation (89). Intravenous infusion of 3.6 μmol/min Mn²⁺ in rats resulted in a drop in blood pressure (90) and Mn²⁺ concentration of 30 μM caused immediate depression of the contractile function in isolated, perfused rat hearts (91). Considering a volume of manganese distribution of about 1.16 L/kg (23) this local concentration corresponds to a bolus injection of 0.035 mmol/kg Mn²⁺. In general, the cardiac side effects can be markedly reduced by slow infusion rates, which lower the transiently high extracellular manganese level in the myocardium. Examples for intravenous injections of MnCl₂ are given in Tables 4, 5, and 6. The most often used veins in rodents are the tail vein and the femoral vein with a recommended needle size of 27 G for rats and 30 G for mice.

Only few groups used an arterial route for the administration of manganese (92) without differences from intravenous injections (90). No significant distinctions in temporal and regional T ₁ changes were reported for subcutaneous, intraperitoneal, and intravenous injections (93, 94). The best way of administration therefore depends on the desired temporal resolution (e.g., subcutaneous injections significantly delay the peak plasma concentration), the manganese concentration (e.g., intravenous injections allow for higher concentrations, but hold a high risk of heart failure), and other practical considerations (e.g., in mice, intraperitoneal injections are much easier to perform than intravenous injections).

To reduce the acute toxicity of manganese, while retaining its potential as an MRI contrast agent, a fractionated application may be helpful (95). In rats, repeated doses of 0.15 mmol/kg Mn²⁺ every 48 h yielded a better signal enhancement than a single administration of 0.15 mmol/kg, indicating accumulation of manganese in the brain. A single administration of the cumulative dose revealed a stronger T ₁ shortening at the expense of higher acute toxicity.

Only few studies attempted to quantitatively determine the manganese concentration in the brain after systemic administration (45, 96, 97). In rats, 1 day after a single 1.4 mmol/kg Mn²⁺ injection, the respective brain concentrations were about 0.11 mM in the olfactory bulb and 0.06 mM in the cortex (45), which correspond to a range in which no toxic effects were seen in cell cultures (98). On the other hand, a single subcutaneous injection of 1.8 mmol/kg Mn²⁺ significantly increased the occurrence of biomarkers for cerebral oxidative damage (98).

A direct and localized injection of manganese into the central nervous system allows for a marked reduction of the total amount of applied manganese (Tables 4, 5, and 6). To minimize possible artifacts due to tissue damage, the manipulation should be as minimally disruptive as possible. This includes a proper choice of the needle size, infusion duration, injection volume, pH value, osmolarity, sterility, and concentration of the manganese solution. For rodents, Hamilton syringes (0.5–2.5 μL) or pressurized controlled picospritzers with a needle size of 32 G and smaller are recommended (99). Examples of recent protocols for intracerebral and intraventricular injections are given in Tables 4, 5, and 6. Although the total manganese burden for the organism is lower than that of a systemic administration, the induction of local cytotoxic effects should be kept in mind (Table 3). Primary cultures of cortical neurons that were exposed to 0.5 mM MnCl₂ for 2 h showed significant increases in F2-isoprostanes, which are markers for oxidative stress, while a fivefold lower concentration of 0.1 mM MnCl₂ did not cause significant alterations (98). For a range of volumes and concentrations, histological investigations of rats 1 week after MnCl₂ injection into the brain revealed thresholds for MnCl₂ toxicity at the injection site of 16 nmol for neuronal death and 8 nmol for astrogliosis in a tissue volume of 8–11 μL (99). Notably, most protocols for intracerebral injection exceed these values, so that a reduction of toxic side effects remains an ongoing research goal. One approach may be the use of micro-osmotic pumps that constantly infuse manganese into a targeted brain area for a period of about 48 h (99).

As with many other MRI contrast agents, Mn²⁺ ions shorten both the T ₁ and T ₂ relaxation times of the surrounding water protons. The extent of T _1,2 reduction depends on the local Mn²⁺ concentration and can be described by
where R _i = 1/T _i (i = 1, 2), [Mn²⁺] is the Mn²⁺ concentration in mM, R _i ([Mn²⁺]) is the R _i at this concentration in s⁻¹, and R _i (0) is the native R _i without manganese ([Mn²⁺] = 0).

The specific Mn²⁺ relaxivities r _i (i = 1, 2) in s⁻¹ mM⁻¹ have extensively been studied (45, 97, 100–105). In water, the very short rotational correlation time of Mn²⁺ ensures an almost constant value for r ₁ of about 7–8 s⁻¹ mM⁻¹ (100, 103), while r ₂ increases with field strength from 40 s⁻¹ mM⁻¹ at 0.47 T to 120 s⁻¹ mM⁻¹ at 9.4 T (100, 103, 104). Under in vivo conditions the intracellular binding of Mn²⁺ to macromolecules causes a much higher r ₁ of about 60 s⁻¹ mM⁻¹ for magnetic fields below 1 T (100, 102, 103). This effect is lost at higher fields of 4.7 T and above, where the in vivo r ₁ approaches the value for aqueous Mn²⁺ (45, 97, 101).

For the range of concentrations used in MEMRI studies and for field strengths greater than 4.7 T where r ₁ is similar for intra- and extracellular manganese, the linear proportionality of the R ₁ relaxation rate to the manganese concentration may be exploited to estimate the absolute amount of manganese present in tissue (45).

Due to the paramagnetism of Mn²⁺ and the fact that the native T ₁ relaxation time of tissue water protons is much longer than the T ₂ relaxation time, low concentrations of Mn²⁺ are best visualized by MRI sequences yielding T ₁ contrast (45). Thus, the majority of MEMRI studies relies on T ₁-weighted images and quantitative maps of T ₁ relaxation times.

For a conventional spin echo (or gradient echo) MRI sequence, the signal equation is given by
with ρ ₀ the spin density, TR the repetition time, and TE the echo time (for gradient echo MRI R ₂ has to be replaced by R ₂*). Unfortunately, R ₁ and R ₂ have opposite effects: while a shortening of T ₁ leads to a signal enhancement, a shortening of T₂ causes a signal decrease. In order to optimize the contrast, the shortest possible echo time will be the best choice. This depends on experimental parameters such as the receiver bandwidth, desired resolution or matrix size, and maximum gradient strength of the MRI system. If the use of an echo time with TE << T ₂ minimizes the effect of the T ₂ signal attenuation, then equation [2] simplifies to

Using equations [1] and [2], the signal difference of a region before and after manganese application can be described by

Thus, the TR for maximum signal enhancement becomes a function of the manganese concentration and the native T ₁ of the tissue water protons: The higher the local manganese concentration for a given R ₁(0), the shorter the TR that gives the best signal difference between pre- and post-contrast conditions and the higher the pre-contrast R ₁(0), the shorter the optimal TR and the lower the signal gain by a given manganese concentration (Fig. 1).

However, there is no general answer to this optimization problem as the choice of parameters always needs to reflect the actual scientific question and the experimental setting. The latter not only involves the parameters of a particular MRI sequence, for example, TR and flip angle for FLASH, but also depends on the local T ₁ change, the applied amount of Mn²⁺, and the time point of the measurement relative to the Mn²⁺ administration (see Note 2).

MEMRI findings after systemic administration need to consider modulations by the manganese pharmacokinetics and the transfer into the brain. Manganese exhibits a high blood clearance with an elimination half-life time of 1.83 h after intravenous administration in rats (23). Already 5–10 min after the beginning of a manganese infusion in rats, the MRI signals of structures outside the BBB start to increase (109). This process leads to high manganese concentrations in the choroid plexus, pituitary gland, and pineal gland after about 1 h (97, 107, 120). Additional strong signal enhancements can be found in the veins and subarachnoid space (97, 121). After about 2 h, the signal from the parenchyma surrounding the ventricles presents with an increasingly strong signal enhancement, while the signals from the vasculature and subarachnoid space start to decline (97, 109) (Fig. 2). Interestingly, in rats, a Mn²⁺ infusion rate of about 32 μmol/h resulted in a signal enhancement of the entire CSF between 10 min and 2 h after administration (109), whereas a slightly lower rate of 14 μmol/h Mn²⁺ failed to yield a signal increase of the CSF in rats and marmosets (97).

The initial enhancement of the choroid plexus and the subsequent enhancement of the periventricular tissue indicate that the BCB is the main delivery route of manganese into the brain, at least for a systemic administration at higher than physiological doses. Between 2 and 24 h after administration, manganese distributes into other brain areas accompanied by a gradual MRI signal increase (108, 121). The maximum tissue enhancement is observed about 24 h after application (122) (Fig. 2) and persists for about 24–48 h followed by a slow recovery to pre-contrast baseline over 2–3 weeks (108, 121, 123).

The clearance of manganese from the brain is rather slow. After intravenous infusion of 1.4 mmol/kg MnCl₂ in rats, the value of ΔR ₁ and the absolute manganese concentration in the brain peaked at day 1 and subsequently declined to near control levels after 28–35 days with an estimated brain half-life time of 11–12 days (from ΔR ₁) and 15–24 days (based on manganese concentration as determined by inductively coupled plasma–mass spectrometry) (45). A radioactive ⁵⁴MnCl₂ analysis, however, revealed a threefold longer brain half-life time of 51–74 days (43) indicating a long terminal elimination phase. Moreover, intravenous injection of ⁵⁴Mn even led to a gradual increase of manganese in some brain areas until day 6 after administration (107). ⁵⁴Mn studies of rhesus monkeys yielded a half-life time of more than 100 days (44, 124) for very low doses of the radioactive tracer, different from MEMRI, which relies on higher doses. The difference between these two types of studies may indeed be explained by saturation effects of manganese-binding sites (45), which result in a faster clearance rate for high doses. This view is further supported by a study where ⁵⁴Mn excretion after intramuscular injection was accelerated by increased dietary manganese (125).

For longitudinal MEMRI studies, it is important to consider potentially confounding effects from preceding injections of MnCl₂. The presence and time course of a residual signal enhancement strongly depends on the region of interest, the accumulated local amount of manganese, and the sensitivity of the applied MRI sequence. In addition, preceding manganese injections may have caused neurotoxicity with accompanying cellular and/or behavioral abnormalities that may hamper subsequent measurements.

Figure 2 shows axial, coronal, and sagittal MEMRI sections of a mouse brain in vivo 24 h after intraperitoneal injection of 0.32 mmol/kg MnCl₂. These T ₁-weighted images confirm the heterogeneous distribution of manganese in the brain. A closer look at the olfactory system in Fig. 3 reveals a signal enhancement in the glomerular layer and mitral cell layer, which are well separated from the less-enhanced granular layer and even darker external plexiform layer. This pattern can already be observed 2 h after a subcutaneous administration of 0.07 mmol/kg MnCl₂ (121). As shown in Fig. 4, the hippocampus presents with a bright signal in the dentate gyrus and CA3 region, whereas CA1 can hardly be separated from the less-enhanced subiculum. In the cerebellum, the granular layer exhibits a pronounced signal enhancement, which becomes even stronger in the layer formed by Purkinje cells (Fig. 5). In the developing cerebellum, a correlation between manganese uptake and granule cell density has been reported (122).

The combination of high-resolution MRI, for example, corresponding to 100 × 100 × 100 μm³ in mice and 50 × 50 × 750 μm³ in rats, with Mn²⁺ doses equal to or higher than 0.7 mmol/kg (109, 121) allows for a detection of cellular layers in the cerebral cortex. In particular, an increased MRI signal is observed in regions corresponding to layer II, the transition region of layers IV and V, and layer VIb, whereas layers I, III, and VIa are much less enhanced (109). This heterogeneous distribution across layers of the cerebral and piriform cortex has been exploited for an in vivo characterization of different mouse mutants (110, 126, 127). Moreover, the more pronounced shortening of T ₁ in the primary visual cortex (V1) relative to the secondary visual cortex (V2) in common marmosets enabled the in vivo delineation of the V1/V2 border (114).

The underlying causes for the differential distribution of manganese in the brain are still a matter of debate. Nevertheless, some mechanisms are generally accepted to contribute to the signal pattern. A major aspect is the location of a structure relative to the ventricles. This is because Mn²⁺ predominantly passes the BCB with uptake across the choroid plexus for the high plasma concentrations as used in MEMRI (128). Studies using radioactive ⁵⁴Mn have shown that the manganese influx in the cerebral cortex is nearly saturated for plasma concentrations of 2–7.5 μM, but the influx into the hippocampus and other regions close to the ventricular system increases with increasing plasma concentration (26). In agreement with these considerations, marmosets showed a strong enhancement in the visual cortex, which in this species is rather close to the lateral ventricles, whereas the frontal cortex due to its distance from the ventricles showed no significant enhancement (see Note 3).

A second relevant aspect is the axonal transport of manganese. For example, the distinct manganese accumulation in the globus pallidus 24 h after administration is most likely due to the anterograde transport of Mn²⁺ from the striatum (gp and str in Fig. 2) (111, 112). Similar evidence for anterograde transport exists for the habenular complex (lh and fr in Fig. 2) (129).

Further parameters influencing the cerebral pattern of manganese are the cell density and neuronal activity (130). For example, strong signal enhancements are observed for the mitral cell layer in the olfactory bulb and the Purkinje cell layer in the cerebellum, which possess high neuronal densities. Moreover, the degree of neuronal activity has been shown to alter the manganese accumulation (see Section 3.4). In addition, the density of voltage-gated Ca²⁺ channels (VGCC) may play a role. Overexpression of N-type VGCC in rats with optic neuritis was accompanied by a strong manganese-induced signal enhancement (131), which could significantly be reduced by treatment with ω-conotoxin GVIA, an N-type specific blocker (132). Furthermore, excitotoxicity in brain ischemia is characterized by enhanced Ca²⁺ influx. Consequently, ischemic lesions exhibited a marked signal enhancement after administration of manganese (133).

The effects supporting MEMRI are also related to the function of cGMP-gated channels (134). Comparison of dark-adapted and light-adapted rats revealed a higher uptake of manganese in the outer retina of the dark-adapted group, which is expected to develop higher cGMP levels resulting in a higher Ca²⁺ influx. Another factor is the presence of manganese-binding enzymes such as GS and Mn-SOD. For instance, the observation of T ₁ hyperintensities after mild ischemia in rats and humans was accompanied by endogenous manganese accumulation and Mn-SOD and GS induction in reactive astrocytes (135). Moreover, in the late phase of mild hypoxic-ischemic injury and following the administration of manganese in rats, MRI revealed enhanced gray matter lesions that correlated with increased immunoactivities of GS and Mn-SOD (136). Manganese-induced signal enhancements were also found to be co-localized with high concentrations of activated microglia (137, 138). Although suggestive, it should be noted that these correlations do not represent a final proof that the MRI findings directly or exclusively reflect the accumulation of manganese-binding proteins.

Attention should also be paid to the fact that the brain uptake of manganese (and respective signal enhancements) may be modulated by other metals. For example, low iron or calcium may result in a higher manganese uptake (80, 139). Conversely, the intravenous administration of ferric hydroxide–dextran complex and a high iron intake may reduce the brain manganese level (80). Furthermore, the amount of manganese in the animal chow may influence its clearance rate, whereas a disturbance of manganese excretion by the liver may lead to manganese deposits in the brain (140). Manganese homeostasis may also be altered in animal models of human brain disorders such as schizophrenia, Alzheimer’s disease, or Parkinson’s disease (141) (see Notes 4 and 5). Finally, in humans but not in animals, a gender difference has been reported for manganese absorption (142).

Another application of MEMRI is the mapping of functional brain activity in response to an external stimulus. This ability is based on the fact that Mn²⁺ may behave similar to Ca²⁺, which accumulates in neurons or astrocytes in an activity-dependent manner. In contrast to other functional MRI approaches that exploit the blood oxygenation level-dependent (BOLD) effect or detect changes of cerebral blood flow (CBF) or volume (CBV), MEMRI does not rely on hemodynamic responses but activity-related changes in calcium flux.

There are three gates for Ca²⁺ uptake into a cell: (1) the VGCC, (2) the ligand-gated nonspecific calcium channels (e.g., the ionotropic glutamate receptors NMDA and AMPA), and (3) the receptor-activated calcium channels (RACC) for which two main types have been described, the store-operated calcium channels and the intracellular messenger-activated nonselective channels (143, 144).

Because Mn²⁺ and Ca²⁺ have the same charge and rather similar ionic radii of 89 and 114 pm, respectively, they behave similarly in many biological systems. In experiments with synaptosomes of rat brain, it was shown that Mn²⁺ can enter the pre-synaptic nerve endings via VGCC and even induce the release of dopamine from the depolarized nerve terminals (30). Moreover, in the absence of extracellular Ca²⁺, Mn²⁺ increased the frequency of acetylcholine release from parasympathetic nerve endings during stimulation (145). In frog, the influx of Mn²⁺ (and Mg²⁺) into neuromuscular junctions was reduced by verapamil, an L-type VGCC blocker (31). Diltiazem, another L-type VGCC blocker, attenuated the signal enhancement in several MEMRI experiments (146–148). Besides VGCC, NMDA receptors also seem to play a role in Mn²⁺-dependent MRI signal changes. Application of NMDA antagonists such as MK-801 resulted in signal suppression, whereas NMDA agonists such as glutamate and NMDA increased the MEMRI signal in rats. On the other hand, NBQX, an AMPA antagonist, did not influence the signal enhancement in MEMRI experiments (94).

Mn²⁺ may also act as a competitive substrate of calcium transporters. MnCl₂ reduced the Ca²⁺ uptake and calcium-dependent sodium efflux of intact squid axons (149) and Mn²⁺ accumulated during hyperkalemic hypoxia in perfused rat myocardium via Na⁺/Ca²⁺-exchanger-mediated transport (150). The ability of Mn²⁺ to mimic Ca²⁺ has also been confirmed by the quenching of fluorescent calcium indicators following Mn²⁺ entry into various cell types (151).

Further evidence for the pertinency of MEMRI to measure neuronal activity comes from studies of c-fos expression. C-fos is an intermediate early gene and its mRNA and protein expression is a widely used functional marker of activated neurons in the CNS. MEMRI signals due to hypothalamic activation, which was induced by intracarotid arterial hypertonic NaCl infusion, correlated well with c-fos expression (152). However, the apparent manganese accumulation during osmotic stress may not only be due to calcium channels. For instance, the DMT1 receptor and transferrin iron transport system seem to be further involved. Furthermore, cGMP-gated channels (134) as well as other metal transporters have to be taken into consideration.

Despite their similarities, Ca²⁺ and Mn²⁺ control different biological functions and, in contrast to Ca²⁺, manganese can also be converted into other oxidation states. It has several similarities to iron including a similar ionic radius, oxidation states 2+ and 3+ under physiological conditions, and similar binding affinities to transferrin. It is therefore of utmost importance to further understand all factors contributing to the cellular uptake of Mn²⁺ before MEMRI may be used as a quantitative marker for Ca²⁺ fluxes (see Notes 4 and 5).

In order to map functional brain activity by MEMRI within a temporal range of seconds to minutes, MnCl₂ is intravenously or intra-arterially infused, while the stimulus of interest is applied. The mechanism relies on local increases of brain activity that lead to corresponding accumulations of Mn²⁺, which in turn cause a signal increase in affected regions. To achieve a sufficiently high manganese level under these conditions, the BBB needs to be opened. Typically, infusion of a concentrated solution of D-mannitol (∼25%) into the bloodstream yields a temporary osmotic disruption of the BBB (92, 153). The potential drawbacks of this technique are the long persistence of manganese inside the cells and the risk of the strong hyperosmolarity of the applied mannitol solution. Thus, initial increases in activation can be visualized, but further variations or deactivations cannot be resolved.

Short-term activity-induced MEMRI is usually performed on anesthetized animals, so that the depth of anesthesia will influence the functional contrast. Deep anesthesia reduces neuronal activity, whereas light anesthesia may lead to competitive brain activity without relation to the stimulus. To control for such confounding effects, it was suggested to perform four sets of measurements (92): The first after starting MnCl₂ infusion but before mannitol injection, the second after bolus injection of mannitol to estimate the Mn²⁺ accumulation caused by nonspecific effects, the third during functional stimulation, and the fourth after the end of MnCl₂ infusion and stimulation. However, additional complications may arise from different narcotic drugs and their putative modulations of manganese uptake and transport. For example, ketamine, an NMDA receptor antagonist, can reduce the signal enhancement by manganese (94), while isoflurane depresses its transsynaptic transport (41).

So far, short-term activity-induced MEMRI has been applied for mapping activation of distinct brain areas by pharmacological substances such as KCl (154), NaCl (155), glutamate (90), amphetamines, and cocaine (148) as well as in response to electrical forepaw stimulation (148, 156) and barrel activation (157). A comparison of MEMRI, CBF, and BOLD activations due to forepaw stimulation revealed a well co-localization of the respective maps in the somatosensory cortex. Interestingly, the strongest T ₁ shortening was observed in cortical layer IV exhibiting a high cell density, while BOLD MRI resulted in stronger activations near the cortical surface close to draining veins (156).

A few brain regions allow for a sufficient manganese accumulation without opening the BBB in a relatively short period of less than 3 h. In mice, these regions are adjacent to the third ventricle and include nuclei of the hypothalamus. Respective MEMRI studies have been performed on fasted and non-fasted mice as well as in response to anorexigenic agents (158, 159). Another well-accessible region is the olfactory bulb. The delivery of MnCl₂ directly into the nasal cavity leads to a signal enhancement in the olfactory bulb via uptake by the nasal epithelium and axonal transport. This approach has been used to visualize odorant representation in the glomerular and mitral cell layer (146, 160).

A second approach to activity-induced MEMRI relies on the fast influx and slow efflux of manganese into the brain and utilizes the cumulative signal change over several hours or even days. The assessment of functional brain activation involves a comparison of two groups of animals, one with and the other without stimulation or treatment. Typically, a stimulus is applied over several hours and MEMRI is performed 24 h after systemic Mn²⁺ administration. Differences in the signal enhancement between the groups are interpreted as stimulus response. Because anesthesia is only necessary during MRI, the approach yields maps of accumulated activity from awake and behaving animals. However, the longer the time between Mn²⁺ injection and MEMRI, the more axonal transport may influence the results: areas with high Mn²⁺ uptake may lead to signal enhancements in axonally connected areas independent of their own activity (161). To reduce such transport effects, shorter stimulation periods of only 9 h have been suggested (162). Due to the low manganese accumulation, this strategy is at the expense of sensitivity, especially in areas remote from the ventricles. Therefore longer periods of stimulus exposure have also been employed (163).