Computational methods and challenges for large-scale circuit mapping

The connectivity architecture of neuronal circuits is essential to understand how brains work, yet our knowledge about the neuronal wiring diagrams remains limited and partial. Technical breakthroughs in labeling and imaging methods starting more than a century ago have advanced knowledge in the field. However, the volume of data associated with imaging a whole brain or a significant fraction thereof, with electron or light microscopy, has only recently become amenable to digital storage and analysis. A mouse brain imaged at light-microscopic resolution is about a terabyte of data, and 1mm(3) of the brain at EM resolution is about half a petabyte. This has given rise to a new field of research, computational analysis of large-scale neuroanatomical data sets, with goals that include reconstructions of the morphology of individual neurons as well as entire circuits. The problems encountered include large data management, segmentation and 3D reconstruction, computational geometry and workflow management allowing for hybrid approaches combining manual and algorithmic processing. Here we review this growing field of neuronal data analysis with emphasis on reconstructing neurons from EM data cubes.     

Volume scanning electron microscopy for imaging biological ultrastructure

Electron microscopy (EM) has been a key imaging method to investigate biological ultrastructure for over six decades. In recent years, novel volume EM techniques have significantly advanced nanometre‐scale imaging of cells and tissues in three dimensions. Previously, this had depended on the slow and error‐prone manual tasks of cutting and handling large numbers of sections, and imaging them one‐by‐one with transmission EM. Now, automated volume imaging methods mostly based on scanning EM (SEM) allow faster and more reliable acquisition of serial images through tissue volumes and achieve higher z‐resolution. Various software tools have been developed to manipulate the acquired image stacks and facilitate quantitative analysis. Here, we introduce three volume SEM methods: serial block‐face electron microscopy (SBEM), focused ion beam SEM (FIB‐SEM) and automated tape‐collecting ultramicrotome SEM (ATUM‐SEM). We discuss and compare their capabilities, provide an overview of the full volume SEM workflow for obtaining 3D datasets and showcase different applications for biological research.     

Computational methods for ultrastructural analysis of synaptic complexes

Electron microscopy (EM) provided fundamental insights about the ultrastructure of neuronal synapses. The large amount of information present in the contemporary EM datasets precludes a thorough assessment by visual inspection alone, thus requiring computational methods for the analysis of the data. Here, I review image processing software methods ranging from membrane tracing in large volume datasets to high resolution structures of synaptic complexes. Particular attention is payed to molecular level analysis provided by recent cryo-electron microscopy and tomography methods.     

Correlative Stochastic Optical Reconstruction Microscopy and Electron Microscopy

Correlative fluorescence light microscopy and electron microscopy allows the imaging of spatial distributions of specific biomolecules in the context of cellular ultrastructure. Recent development of super-resolution fluorescence microscopy allows the location of molecules to be determined with nanometer-scale spatial resolution. However, correlative super-resolution fluorescence microscopy and electron microscopy (EM) still remains challenging because the optimal specimen preparation and imaging conditions for super-resolution fluorescence microscopy and EM are often not compatible. Here, we have developed several experiment protocols for correlative stochastic optical reconstruction microscopy (STORM) and EM methods, both for un-embedded samples by applying EM-specific sample preparations after STORM imaging and for embedded and sectioned samples by optimizing the fluorescence under EM fixation, staining and embedding conditions. We demonstrated these methods using a variety of cellular targets.     

Fast three-dimensional fluorescence imaging of activity in neural populations by objective-coupled planar illumination microscopy

Unraveling the functions of the diverse neural types in any local circuit ultimately requires methods to record from most or all of its cells simultaneously. One promising approach to this goal is fluorescence imaging, but existing methods using laser-scanning microscopy (LSM) are severely limited in their ability to resolve rapid phenomena, like neuronal action potentials, over wide fields. Here we present a microscope that rapidly sections a three-dimensional volume using a thin illumination sheet whose position is rigidly coupled to the objective and aligned with its focal plane. We demonstrate that this approach allows exceptionally low-noise imaging of large neuronal populations at pixel rates at least 100-fold higher than with LSM. Using this microscope, we studied the pheromone-sensing neurons of the mouse vomeronasal organ and found that responses to dilute urine are largely or exclusively restricted to cells in the apical layer, the location of V1r-family-expressing neurons.     

Wet electron microscopy with quantum dots

Wet electron microscopy (EM) is a new imaging method with the potential to allow higher spatial resolution of samples. In contrast to most EM methods, it requires little time to perform and does not require complicated equipment or difficult steps. We used this method on a common murine macrophage cell line, IC-21, in combination with various stains and preparations, to collect high resolution images of the actin cytoskeleton. Most importantly, we demonstrated the use of quantum dots in conjunction with this technique to perform light/electron correlation microscopy. We found that wet EM is a useful tool that fits into a niche between the simplicity of light microscopy and the high spatial resolution of EM.     

Anterograde transport of herpes simplex virus capsids in neurons by both separate and married mechanisms

Anterograde transport of herpes simplex virus (HSV) from neuronal cell bodies into, and down, axons is a fundamentally important process for spread to other hosts. Different techniques for imaging HSV in axons have produced two models for how virus particles are transported in axons. In the Separate model, viral nucleocapsids devoid of the viral envelope and membrane glycoproteins are transported in axons. In the Married model, enveloped HSV particles (with the viral glycoproteins) encased within membrane vesicles are transported in the anterograde direction. Earlier studies of HSV-infected human neurons involving electron microscopy (EM) and immunofluorescence staining of glycoproteins and capsids supported the Separate model. However, more-recent live-cell imaging of rat, chicken, and mouse neurons produced evidence supporting the Married model. In a recent EM study, a mixture of Married (75%) and Separate (25%) HSV particles was observed. Here, we studied an HSV recombinant expressing a fluorescent form of the viral glycoprotein gB and a fluorescent capsid protein (VP26), observing that human SK-N-SH neurons contained both Separate (the majority) and Married particles. Live-cell imaging of rat superior cervical ganglion (SCG) neuronal axons in a chamber system (which oriented the axons) also produced evidence of Separate and Married particles. Together, our results suggest that one can observe anterograde transport of both HSV capsids and enveloped virus particles depending on which neurons are cultured and how the neurons are imaged.     

Semi-automatic classification of skeletal morphology in genetically altered mice using flat-panel volume computed tomography

Rapid progress in exploring the human and mouse genome has resulted in the generation of a multitude of mouse models to study gene functions in their biological context. However, effective screening methods that allow rapid noninvasive phenotyping of transgenic and knockout mice are still lacking. To identify murine models with bone alterations in vivo, we used flat-panel volume computed tomography (fpVCT) for high-resolution 3-D imaging and developed an algorithm with a computational intelligence system. First, we tested the accuracy and reliability of this approach by imaging discoidin domain receptor 2- (DDR2-) deficient mice, which display distinct skull abnormalities as shown by comparative landmark-based analysis. High-contrast fpVCT data of the skull with 200 microm isotropic resolution and 8-s scan time allowed segmentation and computation of significant shape features as well as visualization of morphological differences. The application of a trained artificial neuronal network to these datasets permitted a semi-automatic and highly accurate phenotype classification of DDR2-deficient compared to C57BL/6 wild-type mice. Even heterozygous DDR2 mice with only subtle phenotypic alterations were correctly determined by fpVCT imaging and identified as a new class. In addition, we successfully applied the algorithm to classify knockout mice lacking the DDR1 gene with no apparent skull deformities. Thus, this new method seems to be a potential tool to identify novel mouse phenotypes with skull changes from transgenic and knockout mice on the basis of random mutagenesis as well as from genetic models. However for this purpose, new neuronal networks have to be created and trained. In summary, the combination of fpVCT images with artificial neuronal networks provides a reliable, novel method for rapid, cost-effective, and noninvasive primary screening tool to detect skeletal phenotypes in mice.     

Direct imaging electron microscopy (EM) methods in modern structural biology: Overview and comparison with X‐ray crystallography and single‐particle cryo‐EM reconstruction in the studies of large macromolecules

Determining the structure of macromolecules is important for understanding their function. The fine structure of large macromolecules is currently studied primarily by X‐ray crystallography and single‐particle cryo‐electron microscopy (EM) reconstruction. Before the development of these techniques, macromolecular structure was often examined by negative‐staining, rotary‐shadowing and freeze‐etching EM, which are categorised here as ‘direct imaging EM methods’. In this review, the results are summarised by each of the above techniques and compared with respect to four macromolecules: the ryanodine receptor, cadherin, rhodopsin and the ribosome–translocon complex (RTC). The results of structural analysis of the ryanodine receptor and cadherin are consistent between each technique. The results obtained for rhodopsin vary to some extent within each technique and between the different techniques. Finally, the results for RTC are inconsistent between direct imaging EM and other analytical techniques, especially with respect to the space within RTC, the reasons for which are discussed. Then, the role of direct imaging EM methods in modern structural biology is discussed. Direct imaging methods should support and verify the results obtained by other analytical methods capable of solving three‐dimensional molecular architecture, and they should still be used as a primary tool for studying macromolecule structure in vivo.     

Investigation of HIV‐1 Assembly and Release Using Modern Fluorescence Imaging Techniques

The replication of HIV‐1, like that of all viruses, is intimately connected with cellular structures and pathways. For many years, bulk biochemical and cell biological methods were the main approaches employed to investigate interactions between HIV‐1 and its host cell. However, during the past decade advancements in fluorescence imaging technologies opened new possibilities for the direct visualization of individual steps occurring throughout the viral replication cycle. Electron microscopy (EM) methods, which have traditionally been employed for the study of viruses, are complemented by fluorescence microscopy (FM) techniques that allow us to follow the dynamics of virus–cell interaction. Subdiffraction fluorescence microscopy, as well as correlative EM/FM approaches, are narrowing the fundamental gap between the high structural resolution provided by EM and the high temporal resolution and throughput accomplished by FM. The application of modern microscopy to the study of HIV‐1–host cell interactions has provided insights into the biology of the virus which could not easily, or not at all, have been gained by other methods. Here, we review how modern fluorescence imaging techniques enhanced our knowledge of the dynamic and structural changes involved in HIV‐1 particle formation.      

Fabrication of carbon films with ∼500 nm holes for cryo-EM with a direct detector device

Single particle electron cryomicroscopy (cryo-EM) is often performed using EM grids coated with a perforated or holey layer of amorphous carbon. Regular arrays of holes enable efficient cryo-EM data collection and several methods for the production of micropatterned holey-carbon film coated grids have been described. However, a new generation of direct detector device (DDD) electron microscope cameras can benefit from hole diameters that are smaller than currently available. Here we extend a previously proposed method involving soft lithography with a poly(dimethylsiloxane) (PDMS) stamp for the production of holey-carbon film coated EM grids. By incorporating electron-beam (e-beam) lithography and modifying the procedure, we are able to produce low-cost high-quality holey-carbon film coated EM grids with ∼500 nm holes spaced 4 μm apart centre-to-centre. We demonstrate that these grids can be used for cryo-EM. Furthermore, we show that by applying image shifts to obtain movies of the carbon regions beside the holes after imaging the holes, the contrast transfer function (CTF) parameters needed for calculation of high-resolution cryo-EM maps with a DDD can be obtained efficiently.     

Spotiton: a prototype for an integrated inkjet dispense and vitrification system for cryo-TEM

Over the last three decades, Cryo-TEM has developed into a powerful technique for high-resolution imaging of biological macromolecules in their native vitrified state. However, the method for vitrifying specimens onto EM grids is essentially unchanged - application of ～3 μL sample to a grid, followed by blotting and rapid plunge freezing into liquid ethane. Several trials are often required to obtain suitable thin (few hundred nanometers or less) vitrified layers amenable for cryo-TEM imaging, which results in waste of precious sample and resources. While commercially available instruments provide some level of automation to control the vitrification process in an effort to increase quality and reproducibility, obtaining satisfactory vitrified specimens remains a bottleneck in the Cryo-TEM pipeline. We describe here a completely novel method for EM specimen preparation based on small volume (picoliter to nanoliter) dispensing using inkjet technology. A first prototype system (Spotiton v0.5) demonstrates feasibility of this new approach for specimen vitrification. A piezo-electric inkjet dispenser is integrated with optical real-time cameras (100 Hz frame rate) to analyze picoliter to nanoliter droplet profiles in-flight and spreading dynamics on the grid, and thus provides a method to optimize timing of the process. Using TEM imaging and biochemical assays we demonstrate that the piezo-electric inkjet mechanism does not disrupt the structural or functional integrity of macromolecules. These preliminary studies provide insight into the factors and components that will need further development to enable a robust and repeatable technique for specimen vitrification using this novel approach.     

Multi-resolution correlative focused ion beam scanning electron microscopy: Applications to cell biology

Efficient correlative imaging of small targets within large fields is a central problem in cell biology. Here, we demonstrate a series of technical advances in focused ion beam scanning electron microscopy (FIB–SEM) to address this issue. We report increases in the speed, robustness and automation of the process, and achieve consistent z slice thickness of ∼3 nm. We introduce “keyframe imaging” as a new approach to simultaneously image large fields of view and obtain high-resolution 3D images of targeted sub-volumes. We demonstrate application of these advances to image post-fusion cytoplasmic intermediates of the HIV core. Using fluorescently labeled cell membranes, proteins and HIV cores, we first produce a “target map” of an HIV infected cell by fluorescence microscopy. We then generate a correlated 3D EM volume of the entire cell as well as high-resolution 3D images of individual HIV cores, achieving correlative imaging across a volume scale of 10⁹ in a single automated experimental run.     

A compact multiphoton 3D imaging system for recording fast neuronal activity

We constructed a simple and compact imaging system designed specifically for the recording of fast neuronal activity in a 3D volume. The system uses an Yb:KYW femtosecond laser we designed for use with acousto-optic deflection. An integrated two-axis acousto-optic deflector, driven by digitally synthesized signals, can target locations in three dimensions. Data acquisition and the control of scanning are performed by a LeCroy digital oscilloscope. The total cost of construction was one order of magnitude lower than that of a typical Ti:sapphire system. The entire imaging apparatus, including the laser, fits comfortably onto a small rig for electrophysiology. Despite the low cost and simplicity, the convergence of several new technologies allowed us to achieve the following capabilities: i) full-frame acquisition at video rates suitable for patch clamping; ii) random access in under ten microseconds with dwelling ability in the nominal focal plane; iii) three-dimensional random access with the ability to perform fast volume sweeps at kilohertz rates; and iv) fluorescence lifetime imaging. We demonstrate the ability to record action potentials with high temporal resolution using intracellularly loaded potentiometric dye di-2-ANEPEQ. Our design proffers easy integration with electrophysiology and promises a more widespread adoption of functional two-photon imaging as a tool for the study of neuronal activity. The software and firmware we developed is available for download at http://neurospy.org/ under an open source license.     

Mechanistic Mathematical Modeling Tests Hypotheses of the Neurovascular Coupling in fMRI

Functional magnetic resonance imaging (fMRI) measures brain activity by detecting the blood-oxygen-level dependent (BOLD) response to neural activity. The BOLD response depends on the neurovascular coupling, which connects cerebral blood flow, cerebral blood volume, and deoxyhemoglobin level to neuronal activity. The exact mechanisms behind this neurovascular coupling are not yet fully investigated. There are at least three different ways in which these mechanisms are being discussed. Firstly, mathematical models involving the so-called Balloon model describes the relation between oxygen metabolism, cerebral blood volume, and cerebral blood flow. However, the Balloon model does not describe cellular and biochemical mechanisms. Secondly, the metabolic feedback hypothesis, which is based on experimental findings on metabolism associated with brain activation, and thirdly, the neurotransmitter feed-forward hypothesis which describes intracellular pathways leading to vasoactive substance release. Both the metabolic feedback and the neurotransmitter feed-forward hypotheses have been extensively studied, but only experimentally. These two hypotheses have never been implemented as mathematical models. Here we investigate these two hypotheses by mechanistic mathematical modeling using a systems biology approach; these methods have been used in biological research for many years but never been applied to the BOLD response in fMRI. In the current work, model structures describing the metabolic feedback and the neurotransmitter feed-forward hypotheses were applied to measured BOLD responses in the visual cortex of 12 healthy volunteers. Evaluating each hypothesis separately shows that neither hypothesis alone can describe the data in a biologically plausible way. However, by adding metabolism to the neurotransmitter feed-forward model structure, we obtained a new model structure which is able to fit the estimation data and successfully predict new, independent validation data. These results open the door to a new type of fMRI analysis that more accurately reflects the true neuronal activity.     

A Context-Aware Delayed Agglomeration Framework for Electron Microscopy Segmentation

Electron Microscopy (EM) image (or volume) segmentation has become significantly important in recent years as an instrument for connectomics. This paper proposes a novel agglomerative framework for EM segmentation. In particular, given an over-segmented image or volume, we propose a novel framework for accurately clustering regions of the same neuron. Unlike existing agglomerative methods, the proposed context-aware algorithm divides superpixels (over-segmented regions) of different biological entities into different subsets and agglomerates them separately. In addition, this paper describes a “delayed” scheme for agglomerative clustering that postpones some of the merge decisions, pertaining to newly formed bodies, in order to generate a more confident boundary prediction. We report significant improvements attained by the proposed approach in segmentation accuracy over existing standard methods on 2D and 3D datasets.     

ECS Dynamism and Its Influence on Neuronal Excitability and Seizures

Seizure activity is governed by changes in normal neuronal physiology that lead to a state of neuronal hyperexcitability and synchrony. There is a growing body of research and evidence suggesting that alterations in the volume fraction (α) of the brain’s extracellular space (ECS) have the ability to prolong or even initiate seizures. These ictogenic effects likely occur due to the ECS volume being critically important in determining both the concentration of neuroactive substances contained within it, such as ions and neurotransmitters, and the effect of electric field-mediated interactions between neurons. Changes in the size of the ECS likely both precede a seizure, assisting in its initiation, and occur during a seizure, assisting in its maintenance. Different cellular ion and water transporters and channels are essential mediators in determining neuronal excitability and synchrony and can do so through alterations in ECS volume and/or through non-ECS volume related mechanisms. This review will parse out the relationships between how the ECS volume changes during normal physiology and seizures, how those changes might alter neuronal physiology to promote seizures, and what ion and water transporters and channels are important in linking ECS volume changes and seizures.

     

Isolation and reconstruction of cardiac mitochondria from SBEM images using a deep learning-based method

Mitochondrial morphological defects are a common feature of diseased cardiac myocytes. However, quantitative assessment of mitochondrial morphology is limited by the time-consuming manual segmentation of electron micrograph (EM) images. To advance understanding of the relation between morphological defects and dysfunction, an efficient morphological reconstruction method is desired to enable isolation and reconstruction of mitochondria from EM images. We propose a new method for isolating and reconstructing single mitochondria from serial block-face scanning EM (SBEM) images. CDeep3M, a cloud-based deep learning network for EM images, was used to segment mitochondrial interior volumes and boundaries. Post-processing was performed using both the predicted interior volume and exterior boundary to isolate and reconstruct individual mitochondria. Series of SBEM images from two separate cardiac myocytes were processed. The highest F1-score was 95% using 50 training datasets, greater than that for previously reported automated methods and comparable to manual segmentations. Accuracy of separation of individual mitochondria was 80% on a pixel basis. A total of 2315 mitochondria in the two series of SBEM images were evaluated with a mean volume of 0.78 µm³. The volume distribution was very broad and skewed; the most frequent mitochondria were 0.04–0.06 µm³, but mitochondria larger than 2.0 µm³ accounted for more than 10% of the total number. The average short-axis length was 0.47 µm. Primarily longitudinal mitochondria (0–30 degrees) were dominant (54%). This new automated segmentation and separation method can help quantitate mitochondrial morphology and improve understanding of myocyte structure–function relationships.     

Structural analysis of vimentin and keratin intermediate filaments by cryo-electron tomography

Intermediate filaments are a large and structurally diverse group of cellular filaments that are classified into five different groups. They are referred to as intermediate filaments (IFs) because they are intermediate in diameter between the two other cytoskeletal filament systems that is filamentous actin and microtubules. The basic building block of IFs is a predominantly alpha-helical rod with variable length globular N- and C-terminal domains. On the ultra-structural level there are two major differences between IFs and microtubules or actin filaments: IFs are non-polar, and they do not exhibit large globular domains. IF molecules associate via a coiled-coil interaction into dimers and higher oligomers. Structural investigations into the molecular building plan of IFs have been performed with a variety of biophysical and imaging methods such as negative staining and metal-shadowing electron microscopy (EM), mass determination by scanning transmission EM, X-ray crystallography on fragments of the IF stalk and low-angle X-ray scattering. The actual packing of IF dimers into a long filament varies between the different families. Typically the dimers form so called protofibrils that further assemble into a filament. Here we introduce new cryo-imaging methods for structural investigations of IFs in vitro and in vivo, i.e., cryo-electron microscopy and cryo-electron tomography, as well as associated techniques such as the preparation and handling of vitrified sections of cellular specimens.     

Advanced Correlative Light/Electron Microscopy: Current Methods and New Developments Using Tokuyasu Cryosections

Microscopy is an essential tool for analysis of cellular structures and function. With the advent of new fluorescent probes and super-resolution light microscopy techniques, the study of dynamic processes in living cells has been greatly facilitated. Fluorescence light microscopy provides analytical, quantitative, and three-dimensional (3D) data with emphasis on analysis of live cells using fluorescent markers. Sample preparation is easy and relatively inexpensive, and the use of appropriate tags provides the ability to track specific proteins of interest. Of course, only electron microscopy (EM) achieves the highest definition in terms of ultrastructure and protein labeling. To fill the gap between light microscopy and EM, correlative light and electron microscopy (CLEM) strategies have been developed. In particular, hybrid techniques based upon immuno-EM provide sensitive protein detection combined with high-resolution information on cell structures and protein localization. By adding the third dimension to EM with electron tomography (ET) combined with rapid freezing, CLEM techniques now provide additional tools for quantitative 3D analysis. Here, we overview the major methods applied and highlight the latest advances in the field of CLEM. We then focus on two selected techniques that use cryosections as substrate for combined biomolecular imaging. Finally, we provide a perspective of future developments in the field. (J Histochem Cytochem 57:1103–1112, 2009)