Magnetic resonance imaging of cortical connectivity in vivo

Abstract

Magnetic resonance imaging of neuronal connectivity in vivo opens up the possibility of performing longitudinal investigations on neuronal networks. This is one main reason for the attention that paramagnetic ion manganese (Mn²⁺) has attracted as a potential anterograde neuronal tracer for MRI experiments. However, the correct and possibly repeated use of this tracer – or of any tracer for that matter, including heavy metals – requires the development of an administration strategy that minimizes its toxic effects. Here we first investigated the conditions that maximize the tracing efficiency of Mn²⁺ and preserve viability and tissue architectonics in combined MRI and histology experiments in rats. We demonstrate that most common protocols for neuronal tract tracing using Mn²⁺ result in large neuronal and glial lesions. The toxicity of manganese is distinct during intracortical injections and blocks the transfer of the tracer. After optimizing the technique, we could show that extensive cortical connectivity maps can be generated, with no sign of neuronal damage. Importantly, preservation of tissue viability improves the efficiency of Mn²⁺ in tracing neuronal connections. We have successfully used this technique to trace corticofugal somatosensory and motor pathways in individual animals and describe a connectivity index (CnI) based on Mn²⁺ transport that quantitatively reveals cortical heterogeneities in interhemispheric communication. Finally, we have significantly improved the resolution of the technique by continuously infusing very low concentrations of Mn²⁺ into the target area using osmotic pumps coupled to chronically implanted brain cannulae. The specific, nontoxic and quantitative nature of the neuronal tracings described here indicates the value of this tracer for chronic studies of development and plasticity as well as for studies of brain pathology.

Introduction

One of the most striking characteristics of the brain in general and the cerebral cortex in particular is its profuse connectivity. The level of intercommunication is so that is misleading to think about its function by studying small parts in isolation. The so-called cortical wiring diagram, which describes how information is distributed in such system, comprises local columnar and laminar microcircuits as well as long-range circuits. Since neocortical microcircuits are thought to be highly stereotyped and similar across functionally distinct cortical areas (Braitenberg and Schuz, 1991, Douglas and Martin, 2004), the characteristic function of a cortical area is defined by its particular inputs and outputs and their dynamic properties. The dynamic nature of this information flow is most likely determined by the effective connectivity in a particular brain state or behavioral task and represented, for instance, by short and long-term synaptic plasticity, adaptation and neuromodulation. An understanding of this long-range communication and information processing is one of the major challenges for systems neuroscience. In order to achieve this objective, it is mandatory to obtain anatomical and functional information simultaneously from interconnected groups of neurons in behaving animals.

Countless investigations using classical neuroanatomical techniques like degeneration and tracing methods (Baizer et al., 1991, Cowan et al., 1972, Felleman and Van Essen, 1991, Jones and Powell, 1970, Lanciego and Wouterlood, 2000, Mesulam, 1982, Saint-Cyr et al., 1990, Seltzer and Pandya, 1978) have contributed valuable descriptions of connectivity in the mammalian brain. These very important studies, however, require fixed, processed tissue for data analysis and therefore cannot be applied to an animal participating in longitudinal investigations, where consecutive studies examining an entire circuit are carried out in the same subjects. The recently introduced manganese technique for studying neuronal connectivity in vivo by means of magnetic resonance imaging (MRI) (Pautler et al., 1998) offers an excellent opportunity to perform such experiments. In contrast to studies using conventional techniques, volume imaging with MRI-visible neuronal tracers provides complete descriptions of large-scale three-dimensional (3-D) networks. Furthermore, this nondestructive technique can be repeatedly applied in the same experimental animal to visualize different neuronal pathways and to generate individualized connectivity maps that could guide, for instance, multiple and targeted electrophysiological recordings. Last but not least the technique is invaluable for longitudinal studies, such as those on plasticity and reorganization or on neurodegenerative processes.

The manganese-enhanced MRI (MEMRI) (Koretsky and Silva, 2004) technique is based on the fact that Mn²⁺ ions reduce the longitudinal (i.e. spin-lattice) relaxation times, T1, of water protons. Consequently T1-weighted (T1-W) MR images show enhanced signal intensity at those locations where Mn²⁺ ions accumulate; Mn²⁺ concentration can be estimated based on the change in relaxation rate (1/T1) (Kang and Gore, 1984, Nordhoy et al., 2003, Nordhoy et al., 2004, Silva et al., 2004), adding quantitative information to the connectivity pattern. Once in the extracellular fluid, Mn²⁺ enters the neurons through voltage-dependent Ca²⁺ channels (Drapeau and Nachshen, 1984, Narita et al., 1990) and is anterogradely transported to the distant presynaptic terminals, revealing the first stage in the circuit (Leergaard et al., 2003, Pautler et al., 1998, Saleem et al., 2002, Sloot and Gramsbergen, 1994, Tjalve et al., 1995, Tjalve et al., 1996, Van der et al., 2002, Watanabe et al., 2004). Subsequently, Mn²⁺ is released in the synaptic cleft (Takeda et al., 1998), from where it is taken up into the postsynaptic element and a new cycle of transport starts. In this way Mn²⁺ is transported across synapses and can help identify polysynaptic networks (Murayama et al., 2006, Pautler et al., 1998, Saleem et al., 2002).

However, the technique has several drawbacks that can reduce its applicability, the most important being the potential toxicity of the ion in the tissues. It is well known that an excessive accumulation of Mn²⁺ decreases energy metabolism, increases the production of free radicals, and induces cell death of both neurons and glia in experimental models (Hazell, 2002, Takeda, 2003). In humans, chronic manganese exposure is associated with a syndrome called manganism that resembles Parkinson's disease (Couper, 1837). Perturbations of the neuronal circuits under study due to such toxicity would eliminate the value of Mn²⁺ as an in vivo neuronal tracer, particularly for functional studies, for quantitative 3-D analysis of connectivity, or for repetitive applications investigating the same pathway dynamically over time.

Several factors can contribute to the toxicity of manganese (Aschner and Aschner, 1991). When Mn²⁺ concentrations in the tissue rise above a certain threshold, which has not yet been well defined in vivo, it decreases the neuronal oxidative metabolism increasing the lactate flux (Zwingmann et al., 2003, Zwingmann et al., 2004), inhibits the activity of detoxifying enzymes like the superoxide dismutase and glutathione peroxidase (Hazell, 2002, Takeda, 2003), and alters protein expression (Hazell, 2002). Additional sources of toxicity are some of the physicochemical properties of the injected solution. Firstly, concentrations of MnCl₂ in the extracellular fluid higher than 100 mM are hyperosmolar. We can expect highly concentrated solutions to induce hypertonic stress, contributing to the toxicity (Somjen, 2004), even when the total amount of Mn²⁺ is not by itself toxic. Similarly, pH can indirectly contribute to toxicity because Mn²⁺ solutions in water have a pH (≈ 5.5) well below the physiological range (≈ 7.3). By affecting both pH buffering and osmotic concentration equilibration, the infusion rate of the Mn²⁺ solution into the brain can also contribute to the final toxicity. Although all these indirect effects are transient, dissipating as the ion is diluted into the extracellular fluid, their potential contribution to brain damage must not be dismissed (Somjen, 2004).

In the present work we first investigated the factors contributing to Mn²⁺ toxicity in combined MRI and immunohistological experiments. We demonstrate that most protocols for neuronal tract tracing based on Mn²⁺ and MRI induce severe brain lesions that compromise the interpretation of the experimental results. By taking into account all the cellular and physicochemical considerations mentioned above, we were able to optimize the technique, eliminating the toxic effects and at the same time improving the efficiency of Mn²⁺ in tracing neuronal connections. Second, we successfully used the technique to noninvasively trace and quantify the efferent connectivity of somatosensory and motor areas in individual animals. Finally, we have improved the resolution of the technique by continuously infusing very low concentrations of Mn²⁺ in a target area using osmotic pumps coupled to chronically implanted brain cannulae.