Moving from connections between individual neurons to connections between brain regions, the most widespread and valuable method delivering detailed information about directed long distance connections is neuroanatomical tract tracing (for reviews see Sawchenko and Swanson 1981; Kobbert et al., 2000; Wouterlood et al. 2002). The general approach comprises now a vast range of substances with the common feature that they are taken up by neurons and spread along their projections, where the label then can be visualized. Some substances are directly inserted intracellularly and therefore suitable for tracing of individual neurons. Most of the tracer substances, however, are applied extracellularly to the living tissue by pressure injection, iontophoresis or mechanical insertion. Most of them are actively incorporated through the neuronal membrane and transported in the cytoplasm to reveal the distant location of cell bodies (fast retrograde transport) or axonal terminals (fast and slow anterograde transport). A wide variety of substances are used for tract tracing, which differ in the direction and speed of transport, completeness and intensity of staining, sensitivity and persistence, and the mode of their detection. Among the best tracer substances are plant lectins, such as Phaseolus vulgaris leucoagglutinin (PHA-L), which bind to glycoconjugates on the neuronal membrane and are rapidly internalized and transported both antero- and retrogradely (Gerfen and Sawchenko 1984). Lectins are being used either alone or coupled to retrograde fluorescent tracers, horseradish peroxidase (HRP) or cholera toxin subunit B. Radioactive tracers, mainly tritiated amino acids (TAA), are hydrophilic tracers that are readily incorporated and transported anterogradely even across the synaptic cleft. They have now been largely replaced by dextran amines, which are easier to handle and to detect. For transneuronal tracing, viruses, such as the rabies and herpes virus, are being tried with the limitation that transgression to subsequent neurons may involve extrasynaptic sites (Boldogkoi et al. 2004; Hoshi et al. 2005). Problematic are degeneration studies after physical or toxic destruction of neuronal somata or axon dissection, with subsequent visualization of anterograde (Wallerian), retrograde, or transneuronal degeneration signs (Nauta 1957). The signs of degeneration may be difficult to detect, and large lesions tend to affect fibres of passage. Lipophilic carbocyanine dyes, such as Dil and DiO, spread by lateral diffusion within the cell membrane. Besides in living tissue these can also be employed in fixed tissue, notably for post-mortem in the human brain (Honig and Hume 1989). Although the speed of diffusion increases with temperature, the range of connections identified in fixed tissue is limited to distances of less than 1 cm and the quality of the images is comparatively poor. Thus post-mortem tracing is currently not suitable for tracing of long-distance connections in the human brain.
In in vivo tracing studies, the tracer is actively transported and maximally concentrated in the axon terminals (anterograde transport) or the cell body (retrograde transport) after appropriate survival periods of days to weeks.
The number of retrogradely marked cell bodies can be counted and the proportions in different afferent structures be compared. Axon terminals are hard to quantify leading to density measures in the best case and, more commonly, to rough rankings from sparse to dense labelling. In layered structures, such as the cerebral cortex, the laminar distribution of label at the site of transport can be observed, whereas the laminar position at the application site is often doubtful due to the size of the application and its diffusion halo, where additional uptake may have occurred. Furthermore, tracing studies provide information primarily about the sites of application and labelling after transport, whereas the course of the axonal projections is usually not well observed and hardly ever spatially reconstructed. This limitation results in the curious situation that we know much about which regions are connected by direct and directed axonal projections, but not by what route.
Several other anatomical methods have been tried to gain information about structural connectivity in the brain. Staining for myelin, particularly in fresh tissue using the Gallyas or Heidenhain-Woelcke stains, reveals the presence of myelinated axons predominantly in the white, but to some degree also in the grey matter. Myelin is ubiquitous in the white matter, which makes it impractical to identify individual fibres or to specify the connected sources and targets. Modified de-staining techniques can leave small sets of fibres visible so that their course can be followed. The difficulty of staining specific tracts limits the use of myelin stains mainly to the parcellation of brain structures on the basis of differential myelin densities.
Other methods, such as polarization microscopy, specify the mean orientation of measure fibres in small regions and allow the spatial reconstruction of local fibre bundles. Obtaining information about the three-dimensional arrangement of fibre bundles is useful to classify major tracts in the white matter, but it remains the uncertainty whether all fibres follow the overall course of the bundle. In some cases, such as the medial forebrain bundle, the bundle can be more appropriately regarded as a conduit, which groups fibres together for some part of their course with individual fibres coming in and leaving at different levels, and possibly not a single one actually continuing throughout the whole length of the bundle. By contrast, it is reasonable to assume that some fibre tracts contain a homogeneous set of fibres that originate from and terminate in a single structure (e.g. the geniculo-cortical tract in the primate visual system).
It may not be obvious to address diffusion weighted brain imaging in a section on anatomical technique. However, if one accepts imaging of calcium- or voltage-sensitive dyes as an electrophysiological method then it seems equally valid to consider the contribution of magnetic resonance imaging (MRI) to the identification of brain structure and anatomical connectivity. This inclusion is particularly relevant since diffusion weighted brain imaging is rapidly gaining the status of a versatile substitute for missing anatomical studies and is frequently being interpreted without proper validation against direct anatomical data.
Diffusion weighted magnetic resonance imaging (dMRI; more specifically diffusion tensor imaging, DTI) relies on special pulse sequences, which elicit signals that reflect the orientation of diffusion processes in soft tissue, specifically of water molecules in the brain (Le Bihan 2003). The Brownian motion of water molecules is strongly constrained by myelinated fibre tracts, which hinder any motion across them. The signals therefore indicate two main features: the deviation from randomness of diffusion within each measured voxel, which is expressed by the fractional anisotropy (FA), and the three-dimensional orientation of the diffusion tensor or the corresponding probability functions. Comparing the tensors in adjacent voxels one can concatenate the most consistent orientations ('tractography') and visualize them as lines, which correspond to major white matter fibre tracts in the human and macaque brain (Tuch et al. 2005). This abbreviated description may suffice to understand that dMRI does not directly show fibre tracts (see chapter by Alexander and Lobaugh in this volume), and that the visualization depends to a great extent on the parameters of the reconstruction algorithm. In addition, the voxel sizes are comparatively large at about 6 mm for in vivo imaging with measurement durations not exceeding 1 hour. Tractography cannot disambiguate the many possible geometric constellations within a voxel that lead to the same signal as, for example, crossing and 'kissing' of fibre tracts. Thus, dMRI results are based on major assumptions, which need to be thoroughly validated before they could be equated with directly demonstrated anatomy. Diffusion-weighted MRI is the method of choice for longitudinal in vivo studies of fibre tracts in the whole brain, whereas invasive tract tracing remains the gold standard for identifying selected neuronal connections at a single point in time. The two methods are complementary in so many respects that it is not at all obvious how they could be employed for direct cross-validation.
The characteristics of tract tracing experiments and diffusion weighted imaging are compared in Table 1.
At this time it appears that cross-validation requires a number of additional approaches:
- Combination of traditional tract tracing with corresponding paramagnetic tracers that can be visualized by in vivo whole brain imaging to demonstrate that the two techniques show corresponding results in the same animal (e.g. Saleem et al., 2002).
- In vivo imaging of paramagnetic tracers in comparison with dMRI in the same species to evaluate the reliable demonstration of non-trivial fibre tracts.
- Comparison of dMRI with other indirect measures of connectivity (such as the functional connectivity measured as ongoing "resting state" activity) to evaluate which one is the best predictor of anatomical connectivity.
Before these relationships are established and the applicability of the necessary animal studies to other species, particularly the human brain, is shown, the interpretation of indirect in vivo imaging results as demonstrating anatomical connectivity is not justified.
Table 1. Comparison of characteristic features of tract tracing and diffusion-weighted imaging
Diffusion-weighted imaging microscopic resolution any species (limited by ethics) fibres quantifiable long history and known validity invasive animal experiment post-mortem assessment few injections per brain few population studies done 3D reconstructions rarely done fibre origin + destination applicable to any brain structure can be combined with histology low spatial resolution (several mm) applicable to large brains (humans) surrogate myelin measures unclear validity non-invasive, applicable to humans longitudinal studies possible entire brain imaged population studies usually done registered in 3D coordinates shows fibre course, no direction limited to white matter can be combined with struct./fMRI
Was this article helpful?