Search

Automated sample preparation and electron microscopy enables acquisition of very large image data sets. These technical advances are of special importance to the field of neuroanatomy, as 3D reconstructions of neuronal processes at the nm scale can provide new insight into the fine grained structure of the brain. Segmentation of large-scale electron microscopy data is the main bottleneck in the analysis of these data sets. In this paper we present a pipeline that provides state-of-the art reconstruction performance while scaling to data sets in the GB-TB range. First, we train a random forest classifier on interactive sparse user annotations. The classifier output is combined with an anisotropic smoothing prior in a Conditional Random Field framework to generate multiple segmentation hypotheses per image. These segmentations are then combined into geometrically consistent 3D objects by segmentation fusion. We provide qualitative and quantitative evaluation of the automatic segmentation and demonstrate large-scale 3D reconstructions of neuronal processes from a 27,000 μm(3) volume of brain tissue over a cube of 30 μm in each dimension corresponding to 1000 consecutive image sections. We also introduce Mojo, a proofreading tool including semi-automated correction of merge errors based on sparse user scribbles., (Copyright © 2015 Elsevier B.V. All rights reserved.)

The connectivity among neurons holds the key to understanding brain function. Mapping neural connectivity in brain circuits requires imaging techniques with high spatial resolution to facilitate neuron tracing and high molecular specificity to mark different cellular and molecular populations. Here, we tested a three-dimensional (3D), multicolor super-resolution imaging method, stochastic optical reconstruction microscopy (STORM), for tracing neural connectivity using cultured hippocampal neurons obtained from wild-type neonatal rat embryos as a model system. Using a membrane specific labeling approach that improves labeling density compared to cytoplasmic labeling, we imaged neural processes at 44 nm 2D and 116 nm 3D resolution as determined by considering both the localization precision of the fluorescent probes and the Nyquist criterion based on label density. Comparison with confocal images showed that, with the currently achieved resolution, we could distinguish and trace substantially more neuronal processes in the super-resolution images. The accuracy of tracing was further improved by using multicolor super-resolution imaging. The resolution obtained here was largely limited by the label density and not by the localization precision of the fluorescent probes. Therefore, higher image resolution, and thus higher tracing accuracy, can in principle be achieved by further improving the label density.

We present a novel semi-automatic method for segmenting neural processes in large, highly anisotropic EM (electron microscopy) image stacks. Our method takes advantage of sparse scribble annotations provided by the user to guide a 3D variational segmentation model, thereby allowing our method to globally optimally enforce 3D geometric constraints on the segmentation. Moreover, we leverage a novel algorithm for propagating segmentation constraints through the image stack via optimal volumetric pathways, thereby allowing our method to compute highly accurate 3D segmentations from very sparse user input. We evaluate our method by reconstructing 16 neural processes in a 1024 x 1024 x 50 nanometer-scale EM image stack of a mouse hippocampus. We demonstrate that, on average, our method is 68% more accurate than previous state-of-the-art semi-automatic methods.

This paper presents a new algorithm for extracting the centerlines of the axons from a 3D data stack collected by a confocal laser scanning microscope. Recovery of neuronal structures from such datasets is critical for quantitatively addressing a range of neurobiological questions such as the manner in which the branching pattern of motor neurons change during synapse elimination. Unfortunately, the data acquired using fluorescence microscopy contains many imaging artifacts, such as blurry boundaries and non-uniform intensities of fluorescent radiation. This makes the centerline extraction difficult. We propose a robust segmentation method based on probabilistic region merging to extract the centerlines of individual axons with minimal user interaction. The 3D model of the extracted axon centerlines in three datasets is presented in this paper. The results are validated with the manual tracking results while the robustness of the algorithm is compared with the published repulsive snake algorithm.

Dynamic processes of neural development, such as migrations of precursor cells, growth of axons and dendrites, and formation and modification of synapses, can be fully analyzed only with techniques that monitor changes over time. Although there has been long-standing motivation for following cellular and synaptic events in vivo (intravital microscopy), until recently few preparations have been studied, and then often only with great effort. Innovations in low-light and laser-scanning microscopies, coupled with developments of new dyes and of genetically encoded indicators, have increased both the breadth and depth of in situ imaging approaches. Here we present the motivations and challenges for dynamic imaging methods, offer some illustrative examples and point to future opportunities with emerging technologies.

Studies of populations of receptor proteins suggest that their number and location are highly regulated. Single-particle tracking of glycine receptors now reveals the direct movement of receptors between different clusters of the anchoring protein gephyrin.

We generated transgenic mice in which red, green, yellow, or cyan fluorescent proteins (together termed XFPs) were selectively expressed in neurons. All four XFPs labeled neurons in their entirety, including axons, nerve terminals, dendrites, and dendritic spines. Remarkably, each of 25 independently generated transgenic lines expressed XFP in a unique pattern, even though all incorporated identical regulatory elements (from the thyl gene). For example, all retinal ganglion cells or many cortical neurons were XFP positive in some lines, whereas only a few ganglion cells or only layer 5 cortical pyramids were labeled in others. In some lines, intense labeling of small neuronal subsets provided a Golgi-like vital stain. In double transgenic mice expressing two different XFPs, it was possible to differentially label 3 neuronal subsets in a single animal.

We describe a technique for rapid labeling of a large number of cells in the nervous system with many different colors. By delivering lipophilic dye-coated particles to neuronal preparations with a "gene gun," individual neurons and glia whose membranes are contacted by the particles are quickly labeled. Using particles that are each coated with different combinations of various lipophilic dyes, many cells within a complex neuronal network can be simultaneously labeled with a wide variety of colors. This approach is most effective in living material but also labels previously fixed material. In living material, labeled neurons continue to show normal synaptic responses and undergo dendritic remodeling. This technique is thus useful for studying structural plasticity of neuronal circuits in living preparations. In addition, the Golgi-like labeling of neurons with many different colors provides a novel way to study neuronal connectivity.

In early postnatal life, multiple motor axons converge at individual neuromuscular junctions. However, during the first few weeks after birth, a competitive mechanism eliminates all the inputs but one. This phenomenon, known as synapse elimination, is thought to result from competition based on interaxonal differences in patterns or levels of activity (for review, see Lichtman,1995). Surprisingly, experimental data support two opposite views of the role of activity: that active axons have a competitive advantage (Ribchester and Taxt, 1983; Ridge and Betz, 1984; Balice-Gordon and Lichtman, 1994) and that inactive axons have a competitive advantage (Callaway et al., 1987, 1989). To understand this paradox, we have formulated a mathematical model of activity-mediated synapse elimination. We assume that the total amount of transmitter released, rather than the frequency of release, mediates synaptic competition. We further assume that the total synaptic area that a neuron can support is metabolically constrained by its activity level and size. This model resolves the paradox by showing that a competitive advantage of higher frequency axons early in development is overcome at later stages by greater synaptic efficacy of axons firing at a lower rate. This model both provides results consistent with experiments in which activity has been manipulated and an explanation for the origin of the size principle (Henneman, 1985).

The enormous range of animal size raises a fundamental problem: How do larger animals maintain adequate control of peripheral structures that are many times more massive and extensive than the homologous structures in smaller animals? To explore this question, we have determined neuronal number, the number of axons that innervate each neuron (convergence) and the number of neurons innervated by each axon (divergence), in a peripheral sympathetic pathway of several mammals (mouse, hamster, rat, guinea pig, and rabbit). The average adult weights of these species vary over approximately a 65-fold range. However, the number of superior cervical ganglion cells increases by only a factor of 4 between the smallest of these animals (mice; about 25 gm) and the largest (rabbits; about 1700 gm); the number of spinal preganglionic neurons that innervate the ganglion increases by only a factor of 2. Thus, the number of nerve cells in the sympathetic system does not increase in proportion to animal size. On the other hand, our results indicate that there are systematic differences across these species in the number of axons that innervate each ganglion cell and in the number of ganglion cells innervated by each axon. We suggest that modulation of convergence and divergence in sympathetic ganglia allows this part of the nervous system to effectively activate homologous peripheral targets over a wide range of animal size.

Preganglionic axons arising from different levels of the mammalian spinal cord make preferential connexions with different classes of superior cervical ganglion cells (Langley, 1892, 1900; Njå & Purves, 1977a). For example, preganglionic axons from the first thoracic segment (T1) make relatively strong connexions with ganglion cells activating end-organs of the eye; conversely, axons arising from T4 selectively innervate ganglion cells projecting to the ear. In the present work we have asked whether this selectivity reflects the function of the pre- and post-synaptic cells, and aspect of their respective positions, or some other criterion. 1. End-organs with different functions at the same locus (the eye) respond to stimulation of the same ventral roots; end-organs of a single modality (hairs or blood vessels) at different positions, however, tend to be activated by different spinal segments. Thus the segmental innervation of ganglion cells is correlated with the position rather than the function of post-ganglionic targets. 2. The role of target position in ganglion cell innervation was examined directly by recording from neurones sending axons to different destinations. Superior cervical ganglion cells running dorso-medially in a spinal nerve receive, on average, innervation from more caudal segments than cells projecting ventro-laterally. 3. These selective connexions do not depend on intraganglionic cell position: neurones located at different points along the major axes of the superior cervical ganglion receive, on average, the same segmental innervation. In accord with this observation, retrogradely labelled neurones innervating a particular target such as the eye or ear are widely and randomly distributed within a large portion of the ganglion. Thus the importance of post-ganglionic target position in ganglion cell innervation is not simply a reflexion of ganglionic topography. 4. We conclude that one purpose of the selective connexions in the superior cervical ganglion is to bring together preganglionic axons arising from different levels of the spinal cord and ganglion cells whose axons innervate particular regions of the superior cervical territory.

Synaptic boutons on the surface of identified autonomic ganglion cells were visualized by methylene blue staining at intervals of 1-2 months following denervation to assess whether regenerating axon terminals reoccupy original synaptic sites. The distribution of synapses observed on the same neuronal cell bodies was almost always different in appearance after reinnervation. These results are at odds with the conclusions of earlier workers, who have argued that mammalian neurons bear a fixed number of synaptic sites, which are reoccupied during reinnervation.

1. Simultaneous intracellular recordings were made from pairs of submandibular ganglion cells to examine why each of these neurons is generally innervated by a single preganglionic axon. 2. Impalements made within isolated clusters of two to fifty cells showed that a preganglionic axon typically innervates several neurons within a group. 3. The neurones innervated by a particular axon tended to be intermingled with other neurons; some neurons were innervated by an axon that was not shared by any of their immediate neighbours. 4. These results indicate a mechanism that causes a preganglionic axon to innervate exclusively many of the neurones it contacts, presumably by successful competition with nearby axons that initially innervate the same cells and continue to provide innervation to neighbouring neurones. 5. Re-innervation of adult ganglion cells shows that this mechanism persists or can be reactivated in maturity.

The dendritic arbors of sympathetic neurons in different species of mammals vary systematically: the superior cervical ganglion cells of smaller mammals have fewer and less extensive dendrites than the homologous neurons in larger animals. This difference in dendritic complexity according to body size is reflected in the convergence of ganglionic innervation; the ganglion cells of progressively larger mammals are innervated by progressively more axons. These relations have implications both for the function of homologous neural systems in animals of different sizes and for the regulation of neuronal geometry during development.