In vivo diffusion tensor imaging (DTI) of brain subdivisions and vocal pathways in songbirds

Abstract

The neural substrate for song behavior in songbirds, the song control system (SCS), is thus far the best-documented brain circuit in which to study neuroplasticity and adult neurogenesis. Not only does the volume of the key song control nuclei change in size, but also the density of the connections between them changes as a function of seasonal and hormonal influences. This study explores the potentials of in vivo Diffusion-Tensor MRI (DT-MRI or DTI) to visualize the distinct, concentrated connections of the SCS in the brain of the starling (Sturnus vulgaris). In vivo DTI on starling was performed on a 7T MR system using sagittal and coronal slices. DTI was accomplished with diffusion gradients applied in seven non-collinear directions. Fractional Anisotropy (FA)-maps allowed us to distinguish most of the grey matter and white matter-tracts, including the laminae subdividing the avian telencephalon and the tracts connecting the major song control nuclei (e.g., HVC with RA and X). The FA-maps also allowed us to discern a number of song control, auditory and visual nuclei. Fiber tracking was implemented to illustrate the discrimination of all tracts running from and to RA. Because of the remarkable plasticity inherent to the songbird brain, the successful implementation of DTI in this model could represent a useful tool for the in vivo exploration of fiber degeneration and regeneration and the biological mechanisms involved in brain plasticity.

Introduction

The diffusion weighted image contrast in tissues is a well-accepted supplementary tool for tissue characterization. It is based on the restricted molecular movement of water within and between cells. Using Diffusion Weighted-proton (H⁺) MRI, moving proton spins will undergo a phase difference resulting in a signal intensity attenuation on MR-images (Stejskal and Tanner, 1965). In a sample where the barriers are not coherently oriented, diffusion is the same in all directions and is termed isotropic diffusion. However, if diffusion depends on direction, as in a sample with highly oriented barriers, it is termed anisotropic diffusion. In this way, structural subtypes can be identified simply based on their diffusion characteristics and the anisotropy is directly related to the geometry of the fibers. Moseley et al. (1990) confirmed that water diffusion in cat brain was anisotropic in normal white matter whereas diffusion was isotropic in grey matter. This was also confirmed for the human brain by Ahlhelm et al. (2004). The visualization of bundles of axons connecting distant brain regions (in vivo Fiber-tracking) is accomplished by measuring the diffusion tensor (using different images with diffusion weighting along non-collinear gradient axes) of the endogenous water along the axons (Mori and Barker, 1999). Diffusion Tensor Imaging (DTI) is therefore increasingly used for brain imaging in human studies. Recently, in vivo studies on small animals have been performed, although all studies thus far have been restricted to the mammalian brain, i.e., in rats (Xue et al., 1999, Lin et al., 2001), in mice (Xue et al., 1999, Lin et al., 2001, Zhang et al., 2002, Song et al., 2002, Song et al., 2003, Song et al., 2004, Sun et al., 2003) and in cats (Ronen et al., 2003, Kim et al., 2003).

There are strong indications that most functional sensory, motor and cognitive regions found in the mammalian telencephalon are also present in the avian telencephalon, although both animal groups display remarkable differences in telencephalic anatomy (Reiner et al., 2004, Jarvis et al., 2005). The telencephalon of birds consists of conglomerations of grey matter separated by thin lamina of white matter. This is in contrast to the cerebral organization in mammals, where a superficial thin layer of grey matter (the laminated cortex) is clearly separated from the underlying structure of grey matter (the basal ganglia) by a thick mass of myelinated axons, the internal capsule.

This mammalian brain anatomy has proved very accessible to conventional MRI methods (T₂-, T₁- or Proton Density-weighted MRI), providing superior contrast differences between white and grey matter. As a result, interesting applications have appeared, such as morphometrical MRI measurements evaluating discrepancies in the distribution of grey and white matter between subjects with different behavioral backgrounds (e.g., taxi drivers or professional musicians (Maguire et al., 2000, Gaser and Schlaug, 2003). Recently, DTI has contributed to these human studies by providing more detailed and 3D information on the white matter fiber tracts in the brain (Yamazaki et al., 2004). In contrast, the telencephalon of birds is quite different in its gross appearance from that of mammals and a clear morphometrical distinction between grey matter and myelinated axons cannot be made using conventional intrinsic MRI contrast settings, even at high resolution (Van der Linden et al., 1998, Verhoye et al., 1998). This lack of intrinsic contrast in avian brain tissue not only applies to the subtle laminae consisting of fibers dividing telencephalic brain regions, but also to inherent differences in cytoarchitecture and even to more obvious differences between specific nuclei, such as those delineating the song control system (SCS) of songbirds (Tramontin et al., 1998, Tramontin and Brenowitz, 2000, Thompson and Brenowitz, 2005). This system is responsible for the learning and production of learned complex vocalizations (song) and is the most well documented part of the songbird brain. It provides a unique and excellent model for the study of brain plasticity, learning and the neural substrate of an easily quantifiable behavior (song production). The telencephalic part of this system shows a remarkable sexual dimorphism in species with sexual dimorphism in song behavior (Del Negro and Edeline, 2001, Van der Linden et al., 2002, Riters and Ball, 2002, Thompson and Brenowitz, 2005) and a remarkable seasonal plasticity in species where song output changes with seasons (Ball et al., 2004, Brenowitz, 2004). Some parts of the circuit develop in a period during which the juvenile learns its songs from adult conspecific birds (Aamodt et al., 1992).

The non-invasive, in vivo nature of MRI renders it an exquisite tool with which to study brain plasticity, brain behavior interactions and associated phenomena. However, previous MRI studies on the song bird brain using T₂- and Proton Density-weighted datasets (Van der Linden et al., 1998, Verhoye et al., 1998) failed to discern the song control nuclei and the fiber bundles that connect them. More recently, the use of an in vivo tract tracing technique based on the stereotaxic injection of paramagnetic Mn²⁺ in HVC, a key nucleus of the song control system (HVC is the name, not an abbreviation), allowed successful labeling of its two targets, namely the robust nucleus of the arcopallium (RA) and area X of the basal ganglia (Van der Linden et al., 2002, Van der Linden et al., 2004, Tindemans et al., 2003), but still failed to show the fiber bundles that connect them. By adopting dynamic measurements, known as Dynamic Manganese enhanced (DME)-MRI, we evaluated the functional status of the song control circuit while birds where exposed to conspecific song (Tindemans et al., 2003) or under different endocrine conditions that changed the physiological status of the circuit (Van Meir et al., 2004). However, despite its advantages for analyzing functional changes in physiological status using repeated observations within individual subjects, the DME-MRI method has certain drawbacks. One of these is that the interpretation of changes in neuronal activity, which may come about as a result of the accumulation of Mn²⁺ in the target regions, is blurred by potential morphological changes in connectivity and cell density, especially if the measurements are performed over long time periods. Also, the ME-MRI technique is not completely non-invasive and small brain lesions can still occur. DTI could therefore be a complementary non-invasive tool that quantifies the morphological changes in the song control system occurring during seasonal and endocrine changes.

This paper represents the first detailed evaluation of the application of DTI to the songbird brain and explores the potential of further applications in this model. Data will be discussed in the light of future technical developments in order to establish a quantitative technique that can be applied in this remarkable animal model exhibiting seasonal changes in song behavior with concomitant functional and morphological plasticity of the brain regions involved.