“Current Pathology Techniques” Symposium Review: Advances and Issues in Neuropathology

Abstract

Our understanding of the mechanisms that incite neurological diseases has progressed rapidly in recent years, mainly owing to the advent of new research instruments and our increasingly facile ability to assemble large, complex data sets acquired across several disciplines into an integrated representation of neural function at the molecular, cellular, and systemic levels. This mini-review has been designed to communicate the principal technical advances and current issues of importance in neuropathology research today in the context of our traditional neuropathology practices. Specific topics briefly addressed in this paper include correlative biology of the many facets of the nervous system; conventional and novel methods for investigating neural structure and function; theoretical and technical issues associated with investigating neuropathology end points in emerging areas of concern (developmental neurotoxicity, neurodegenerative conditions); and challenges and opportunities that will face pathologists in this field in the foreseeable future. We have organized this information in a manner that we hope will be of interest not only to professionals with a career focus in neuropathology, but also to general pathologists who occasionally face neuropathology questions.

Introduction

The nervous system is the main substrate whereby vertebrates regulate their physiologic integrity and interact with their environment. The fine control exerted by the nervous system requires that neurons and neural regions are highly differentiated to a degree where each can perform a finite repertoire of actions. Such specialization provides exquisite control when the nervous system is finely tuned and functioning within normal parameters. However, such specialization also retards or even precludes efforts to repair neural injury, particularly in the central nervous system (CNS). Damage to one or more neural structures can have substantial impact on an organism’s capacity to function for extended periods, or even to survive at all.

The meager regenerative capacity of the nervous system coupled with the recognition that neurological damage is a major occupational health hazard (Connelly and Malkin 2007; Gobba 2003) and a common congenital deficit (Bearer 2001) has resulted in a wide-ranging international effort to identify and ameliorate neurological syndromes. Extensive resources have been allocated in recent years to exploring neurodegenerative (e.g., Alzheimer’s and Parkinson’s diseases), neurogenetic (e.g., Down’s syndrome), neuroimmunologic (e.g., multiple sclerosis), and neurotoxic (e.g., ethanol and lead exposures in children, and organic solvent neuropathies in adults) conditions, to name just a few examples. In all these instances, the integrated anatomic, clinical, and functional expertise of neuropathologists plays a key factor in seeking to understand the mechanisms and develop treatments.

The current review has been formatted to highlight the roles and opportunities for pathologists working as collaborators in a neurobiological research team. This information was assembled by choosing “Advances and Issues in Neuropathology” as the focus of the 2005 “Current Pathology Techniques” symposium, a meeting sponsored by the International Life Sciences Institute (ILSI) to familiarize practitioners with current thinking in the given subdiscipline of pathology. This paper is composed of brief sections prepared by the symposium speakers along with supplementary topics that were not presented because of time constraints. The sections cover presently accepted practice in neuropathology research, give guidance regarding current challenges and opportunities in the field, and provide a list of recent citations for further study.

Correlative Neurobiology

The nervous system in vertebrates actually consists of multiple interacting hierarchies of body control elements. The simplest classification scheme can define two divisions based on the manner in which cell interactions occur: neural (with direct communication between adjacent cells via chemical or electrical signals) and neuroendocrine (in which chemical signals between remote cells must travel more diffusely over greater distance). Another simple means of segregating the nervous system is by location, in which two main divisions are evident: the central nervous system (CNS, i.e., those neural elements encased in the skull and vertebral column, the brain and spinal cord), and the peripheral nervous system (PNS, i.e., the nerves and ganglia residing outside the skull and vertebral column that convey signals between the CNS and outlying body regions). Yet a third means of partitioning the nervous system is by function: cognitive (domains involved in complex associative and creative behaviors [“thinking”]), somatic (regions that control motor and sensory activities), and autonomic (centers that regulate internal physiologic conditions). The nervous system also can be parsed to an increasingly fine degree (parasympathetic and sympathetic divisions of the autonomic nervous system, to take one example). The pathologist who engages in neurobiological research must be aware of these views of the nervous system because each is associated with its own unique anatomy, cell biology, and neurochemistry. The most effective practice of modern neuropathology requires that one build an integrated mental map in which all these data are superimposed for ready access.

General Neurobiology

The principal functional units of the nervous system are highly differentiated, excitable neurons that have adapted to conduct action potentials rapidly along specialized cellular extensions called axons. The conduction velocity is accelerated in both the CNS and the PNS by the presence of insulating myelin sheaths around neuronal axons.

Information from one neuron flows to other neurons across small gaps (synapses) that separate them. Synapses consist of three elements: a presynaptic terminal that contains neurotransmitter-laden vesicles, mitochondria, and other organelles; a postsynaptic ending that contains membrane-bound receptors for neurotransmitters; and an intervening synaptic cleft that separates the presynaptic and postsynaptic endings.

Central Nervous System (CNS)

Physiology:The CNS controls or influences virtually all physiological processes. The primary functions of the CNS include (a) detection of sensory inputs, (b) processing and generating responses to signals, and (c) regulation of endocrine output. Within the adult human brain there are approximately 130 billion neurons forming 150 trillion synapses (Saver 2006). Different neuron populations exist within the CNS and are often distinguishable by their shape (e.g., round vs. pyramidal neurons), primary neurotransmitter (e.g., cholinergic vs. dopaminergic neurons), and/or location (e.g., cortical vs. cerebellar neurons). Neurons in general have a high metabolic rate, which makes them exquisitely susceptible to certain global toxic insults that impair intracellular energy metabolism (Calabresi et al. 2000). In addition, some neuronal populations experience an enhanced vulnerability to certain toxicants owing to cell-type–specific metabolic pathways (e.g., 6-hydroxydopamine and its effects on dopaminergic neurons) (Kostrzewa and Brus 1998).

In the CNS, each neuron is sustained by approximately ten to fifty glial cells (Aschner et al. 2002). There are three types of glia, each of which has a unique set of responsibilities. The astrocyte provides structural support to neurons, participates in the metabolism of certain chemical neurotransmitters, maintains extracellular ionic conditions, stores and transfers metabolites from capillaries to neurons, and is involved in repair. Oligodendrocytes form the myelin sheaths around neuronal axons in the CNS; myelin serves as an electrical insulator that accelerates nerve conduction velocity. Each oligodendroglial cell envelops segments of several axons. Microglial cells are the tissue-resident histocytes of monocytic lineage. Glia respond in different ways to toxicant-induced injury. Astrocytes and microglia may exhibit hypertrophy and hyperplasia. In addition, astrocytes may express an increased quantity of their unique cytoskeletal element, the intermediate filament protein glial fibrillary acidic protein (GFAP) (Eng et al. 2000; Pekny and Pekna 2004), whereas microglia increase production of proteins that participate in phagocytic activities (Dheen et al. 2007). Damaged oligodendrocytes often swell. Proliferation of glial cells (gliosis) may also occur as a secondary response to primary neuronal injury.

The delicate CNS is supported and protected by a three-layered membrane of fibrous tissue, the meninges. The outer meningeal layer, or dura mater, is adherent to the skull and restricts the movement of the brain within the skull. A potential subdural space separates the dura mater from the thin avascular middle layer, called the arachnoid mater. The inner layer, or pia mater, is a highly vascular membrane that is closely adhered to neural tissue. Although the dura appears grossly to be the most imposing structural layer of the meninges, the tight junctions in the arachnoid layer actually provide the most resistance to diffusion between the brain/spinal cord and potential toxicants on the other side. The meninges are relatively resistant to toxicants.

The choroid plexus is a highly vascular, thin-walled tissue with an extensive surface area. It is found along the floor of the lateral ventricles and on the roof of the third and fourth ventricles. The choroid plexus produces cerebrospinal fluid (CSF), a low-protein filtrate of blood plasma that is replenished several times a day; the plexus also accumulates and transports many chemicals (Redzic and Segal 2004). The CSF flows through the brain ventricular system, central canal of the spinal cord, and subarachnoid space, acting to provide additional mechanical cushioning of the CNS. Its circulation is driven by cardiac-inspired pulsations of the choroid plexus and facilitated by ciliated ependymal cells lining the ventricles and central canal. The CSF is ultimately absorbed by the arachnoid villi. The fluid also plays a critical role in the removal of brain metabolites, redistribution of hypothalamic peptide hormones, and maintenance of normal CNS pH (Redzic and Segal 2004). Components in the CSF are in contact with the brain interstitial fluid (ISF) that bathes neurons and glia. The ISF flows throughout the neural parenchyma, thereby serving as a conduit for the dispersion and clearance of fluid as well as entities (cells, drugs, hormones) that are carried in it (Abbott 2004).

Anatomy:Upon gross examination, the CNS can be divided into anatomically distinct structures including the forebrain (cerebral cortex, basal ganglia, corpus callosum, and hippocampus), diencephalon (thalamus and hypothalamus), mesencephalon (substantia nigra, cranial and caudal colliculi, tectum, tegmentum), hindbrain (cerebellum, pons, medulla), and spinal cord (cervicothoracic, lumbosacral). Each of these defined neuroanatomic structures may, in fact, have multiple subdivisions, each of which may serve a unique function (Bolon 2000).

Visual inspection of the brain from various classes of mammals (e.g., rodents, carnivores, and primates) reveals marked species differences in size and gross anatomy (Bolon 2000). Brain weight generally rises with an increase in phylogenetic level and body size. For example, the adult human brain weighs approximately 1300 to 1400 g, whereas the adult rat brain weighs only 1.5 to 2 g. Other striking differences are the surface topography of the brain, the prominence of various cortical regions, and the size of certain spinal tracts. The arrangement of sulci and gyri varies with both age and species, and within a species it is often asymmetric between the right and left hemispheres. In the spinal cord, many white matter tracts, including the dorsal funiculus (the main conduit for transmission of most tactile and proprioceptive sensory impulses) and corticospinal pathways (which mediate cortical control of voluntary motor functions), are largest in primates, of intermediate size in carnivores, and smallest in rodents. In the cerebral cortex, the main functional domains have comparable locations across all mammalian species, although the size of these zones varies among species. For instance, in the carnivore the primary sensory and motor areas and olfactory cortex account for 80% of the cerebral cortex, whereas these same zones in the primate represent less than 20% of the cortical tissue. Another telling interspecies difference in cerebrocortical function is the ratio of paleocortex (which is largely controlled by the limbic system) to that of neocortex (which participates in associative and cognitive activities). The rodent cortex consists primarily of paleocortex, whereas the neocortex accounts for a very large portion of the primate brain.

Within the CNS, specific anatomic regions harbor distinct functional domains. For example, the dorsal domains of the brain and spinal cord coordinate afferent impulses, whereas ventral structures control effector pathways. The motor system consists of a somatic motor function, which conveys impulses to skeletal muscles, and a visceral motor portion, which supplies cardiac and smooth muscles. Likewise, the sensory system contains a somatic sensory component, which receives signals from the skin and body wall, and a visceral sensory domain that innervates the internal organs. Neurons with similar functions are often arranged in stereotypical patterns that are shared among different animal species. In the mammalian forebrain, the primary somatic centers typically are found in the dorsolateral cerebral cortex, with the motor area located rostral to the sensory area. The spinal cord exhibits a similar functional segregation, with input from the somatic and visceral sensory domains entering the dorsal gray matter, whereas signals from the somatic motor domain originate in the ventral gray matter. The visceral sensory and motor centers are located in nuclei of the midbrain, brain stem, and intermediate column of gray matter in the spinal cord. Many homeostatic activities are unconsciously regulated in the limbic system (especially the hypothalamus), midbrain, and brain stem. The use of functional magnetic resonance imaging (fMRI) and other neuroimaging modalities has dramatically improved our understanding of the neuroanatomical substrates associated with these different activities.

Different regions of the nervous system may be uniquely sensitive to a given xenobiotic. Regional responses may reflect variations in local blood flow, neuron and neurotransmitter distribution, and compartmentalization of other biochemical or metabolic functions. For example, substantia nigral dopaminergic neurons play a critical role in the pathogenesis of Parkinson’s disease and schizophrenia. These neurons control movement and are also involved in feelings of reward, alertness, and purposeful behavior. Cocaine, heroin, amphetamine, alcohol, and nicotine elevate synaptic levels of dopamine and increase locomotor activity in animals. The ability of the nervous system to recover from chemical-induced injury also demonstrates regional differences. In the CNS, domains have a quite limited ability to regenerate and generally must compensate instead for neurotoxic damage. Younger individuals appear to have more capacity for both anatomical and functional adaptation (plasticity) than older individuals, suggesting that the aged may be at greater risk from neurotoxic exposure to certain chemicals.

Peripheral Nervous System

The normal peripheral nerve is a small, white trunk composed of numerous neuronal processes (axons) surrounded by glia (Schwann cells) and embedded in layers of connective tissue. The connective tissue surrounding the outer surface of each nerve trunk is called the epineurium. The connective tissue partitions that separate groups of axons into distinct bundles (or fascicles) are termed the perineurium. This perineurial covering is a tough membrane with tight junctions that serves (in conjunction with the specialized endothelium of capillaries inside nerves) to provide an effective diffusion barrier; greater permeability of the perineurium in certain PNS structures (e.g., dorsal root ganglia, parasympathetic ganglia, and sensory and motor nerve endings) is thought to contribute to the enhanced susceptibility of these regions to certain toxic agents (Anthony et al. 2002). The connective tissue inside the fascicles that supports individual nerve fibers is called the endoneurium. Additional cells scattered within the endoneurium include fibroblasts, endothelial cells, macrophages, and mast cells.

Peripheral nerves contain both myelinated and nonmyelinated axons (Figure 1A). The architecture of myelinated axons is adequately evaluated in thin (1 μm) plastic sections stained with toluidine blue, though thicker (1.5 μm) sections are often cut to achieve improved contrast. Myelinated fibers are usually 2 to 20 μm in diameter. In standard thin sections stained with toluidine blue, the axoplasm (axonal cytoplasm) typically appears light blue owing to its protein components. A variety of organelles are dispersed in the axoplasm: large mitochondria (greatest dimension, 150 to 300 nm), axoplasmic reticulum (SER), neurofilaments (10 nm), and microtubules (also known as neurotubules, 20 nm). Neurofilaments, which are an intermediate filament unique to neurons, are constructed from the noncovalent association of three subunits; they are very argyrophilic and typically represent the major structures stained by certain silver-based methods. In normal nerves, axonal components such as mitochondria, neurofilaments, neurotubules, and axoplasmic reticulum are evenly distributed throughout the axoplasm. In peripheral neuropathies, in which neurofilament proteins accumulate in the axoplasm, mitochondria and neurotubules may be displaced to the periphery of the axon (Graham et al. 1995; Sills et al. 1998). Microfilaments (4 to 6 nm) composed of actin are typically lost during processing but are associated with the axolemma, a three-layer membrane (approximately 8 nm thick) separating the axoplasm from the myelin. Unmyelinated axons are much smaller than their myelinated counterparts and are thus best evaluated in thin sections (about 90 nm thick) by electron microscopy.

Myelin in the PNS is formed by a special glial element, the Schwann cell. These glia are of two kinds, myelinating and nonmyelinating. The myelinating Schwann cell is surrounded by a basement membrane (basal lamina), and the majority of its cytoplasm is located near the nucleus and at the nodes of Ranvier (Figure 1B). The Schwann cell cytoplasm contains mitochondria, Golgi bodies, axoplasmic reticulum, and polyribosomes; resting cells contain very little rough endoplasmic reticulum (RER). The mesaxon is a Schwann cell sheet-like extension that wraps around the axon, after which the cytoplasmic compartment collapses to form layered (termed compact) myelin consisting of a continuous lamina characterized by alternating major dense lines (intracellular) and intraperiod lines (extracellular) (Figure 1C). Multiple myelinating Schwann cells are associated with any single axon; however, each Schwann cell envelops only one axon, acting to produce the myelin for one internode (demarcated on each end by a node of Ranvier that does not possess myelin). The internodal length is proportional to axonal diameter and remains relatively constant for a given fiber. A unique feature of myelin in the PNS is the presence of Schmidt-Lanterman incisures (Figure 1D), which are continuous spirals of Schwann cell cytoplasm connecting outer perinuclear and inner adaxonal cytoplasmic compartments. The intraperiod line also opens at these incisures, providing communication of endoneurial space with periaxonal space. Schmidt-Lanterman incisures are more prevalent in larger fibers. The presence of normal cytoplasmic features in these structures serves to differentiate them from the abnormal myelin splitting that occurs in certain pathologic states.

Unmyelinated axons are located in recesses of nonmyelinating Schwann cells, and multiple axons are contained within the basal lamina of a single Schwann cell (Figure 1E). The unmyelinated axons together with their associated Schwann cell are termed Remak fibers. Unmyelinated axons typically range from 0.2 to 3.0 μm in diameter and exhibit a unimodal size distribution. Their axoplasm contains a similar complement of structures to those present in myelinated axons.

Relative to the CNS, an important characteristic of the PNS is the ability of damaged components to undergo fairly efficient regeneration. Whether the primary insult is to the axon or the Schwann cell, if the injury is not repaired, the other element will degenerate as a secondary event. The disintegrating debris is efficiently removed by resident endoneurial and recruited hematogenous macrophages. The basement membrane surrounding the axon–Schwann cell unit remains after degeneration of the axon and dedifferentiation of myelinating Schwann cells, providing conduits (termed bands of Bungner; Figure 1F) thought to provide physical guidance and concentrate Schwann-cell–secreted trophic factors important for regenerating axons. Over time, new axons from centrally located neurons can grow through the Schwann cells to reestablish their capacity to signal peripheral targets. The major difference between the initial nerve fibers and the regenerated fibers is that the internodal distance is smaller in repaired nerve trunks.

Pathology Analysis of the Central Nervous System in the Preclinical Setting

Structural evaluation of the nervous system has been undertaken in an organized fashion by anatomists and pathologists for over a century. Even general pathologists are familiar with the standard battery of procedures used to perform a routine postmortem anatomic assessment of neural integrity. However, newer techniques combining structural evaluation with region-specific molecular analysis have substantially improved the potential for gaining data suitable for mechanistic and regulatory purposes. This section will reiterate the common methods used in conventional neuropathology examinations of the CNS in vertebrate species used in neurotoxicity testing. The same procedures are readily adapted in the diagnostic setting when clinical signs point to the presence of neurological disease in a patient.

Study Design

The experimental approach must be balanced between a need for a thorough assessment of neural tissues and the desire to maximize the efficiency with which scarce research resources (e.g., money, personnel time) are expended. The generally accepted list of neural structures to be evaluated in a conventional toxicity study (Solleveld and Boorman 1990) provides a suitable survey of the nervous system (Table 1). This list is necessary because an abbreviated sampling protocol is not conducive to a full characterization of potential neurotoxicity given the tremendous morphologic and physiologic heterogeneity of the brain (Bolon 2000). For dedicated neurotoxicity studies, the generally accepted list of neural tissues to analyze is greatly expanded (Table 1), but even so, some neural structures are missed or scanned in a superficial manner (e.g., inner ear, peripheral autonomic nerves) unless prior evidence suggesting a potential deficit originating in these sites results in their inclusion in the final tissue list.

Each study design must be developed specifically for the test article and system. That said, consistent use of a standard study design provides investigators and regulatory scientists with confidence that any morphologic effects present in the CNS and PNS will be detected (within the limitations of light and/or electron microscopic examination). Particular attention must be given to proper dose selection, appropriate sampling times, adequate sectioning (standardized between animals and ideally across experiments), evaluation of a sufficient number of animals, methods to decrease artifactual change, examination of appropriate sites (particularly within the brain), and appropriate embedding and staining techniques (Broxup 1991).

Acquisition of CNS Tissues

Timing:Standard four-, fourteen-, twenty-eight-, or ninety-day toxicity studies may not reveal the full spectrum of neurotoxicity for a given test article. When possible, necropsies should be scheduled to take advantage of any known or suspected times when morphologic changes may occur. For example, a single dose of the N-methyl-d-aspartate (NMDA) receptor antagonist MK-801 incites vacuolation in neurons of the cingulate gyrus (retrosplenial cortex) four to eight hours after exposure, which progresses to localized neuronal necrosis one or more days after dosing (Fix et al. 1996). Following kainic acid administration, cellular degeneration is most prominent between one and four days after exposure, but synaptic terminal degeneration is more pronounced four to fourteen days postdosing (Schmued and Hopkins 2000a).

Fixation:Parameters for successfully fixing the nervous system have been dealt with previously in a comprehensive manner (Bagnell 1995; Cassella et al. 1997; Fix and Garman 2000). Current regulatory guidelines for adult (EPA 1998a; OECD 1997) and juvenile (EPA 1998b; OECD 2007) rats as well as accepted neuropathology practice (Fix et al. 1996; Garman 1990; Mattsson et al. 1990) advocate that neural tissues are best preserved by intravascular perfusion fixation followed by an additional period of immersion postfixation. Options for achieving such perfusions have been detailed formerly for both adult (Fix and Garman 2000) and juvenile (Bolon et al. 2006) rats. The same principles apply to other vertebrate species as well (Krinke et al. 1997). Two notable artifacts reduced by intravascular infusion of an aldehyde fixative are neuropil vacuolation and induction of dark neurons (Garman 1990).

The choice of fixative depends on the study design. Neutral buffered 10% formalin (NBF, a 3.7% formaldehyde solution, pH 7.6, made by adding one volume of concentrated [37%] formaldehyde to nine volumes of water), 4% formaldehyde (prepared fresh from paraformaldehyde powder), variable concentrations of glutaraldehyde (typically 1% to 4%), or combinations of glutaraldehyde and formaldehyde are all suitable fixatives for neural tissues. Each has advantages and disadvantages. Formaldehyde is the preferred choice for routine light microscopic studies because its small size allows ready penetration of dense neural tissues, whereas the reactive aldehyde group rapidly initiates fixation. Commercial formaldehyde solutions may be used if they have not been adulterated with stabilizing chemicals (e.g., methanol), which can induce artifacts. Glutaraldehyde is a larger molecule and thus penetrates tissues more slowly, but the presence of two aldehyde groups facilitates cross-linking of proteins and preservation of the lipids. Therefore, glutaraldehyde or a glutaraldehyde–formaldehyde mixture is preferred when the primary focus is evaluation of myelinated components of the central or peripheral nervous systems or whenever an ultrastructural (electron microscopic [EM]) examination is to be undertaken. Fixation with glutaraldehyde is improved when the vascular system is flushed with buffer prior to infusing the fixative; preflushing is recommended but not absolutely required for formaldehyde-based fixatives. Regardless of the delivery method (gravity drip or perfusion pump), the perfusate should be delivered at a pressure that is approximately equal to systolic blood pressure (120 mm Hg). Intermittent interruptions in intravascular pressure—even brief ones—may cause premature collapse of small blood vessels and suboptimal fixation. Osmolarity, pH, buffering capacity, and temperature are all important characteristics of the flushing and fixative solutions (Fix and Garman 2000).

Since the majority of routine toxicity studies employ NBF immersion as a means of tissue preservation, perfusion-fixed specimens of neural tissues will not always be available for evaluation. This absence does not necessarily prevent a thorough evaluation of neural structures, but the investigators and pathologist must be aware that immersion fixation increases the possibility of false positives (artifactual changes being interpreted as real lesions) and false negatives (failure to observe or correctly interpret subtle changes that may be masked by artifactual changes) (Garman 2003; Jortner 2006).

Some immunohistochemical procedures and other evaluations including polymerase chain reaction (PCR) may require the collection of unfixed tissue. Perfusion with ice-cold buffered saline may reduce artifactual changes by slowing postmortem autolysis until the tissues can be frozen. Reduced-strength fixatives (e.g., 1% or 2% formaldehyde) may also be used if a higher concentration of fixatives will prevent the retrieval of specific antigens. In such instances it is wise to conduct a preliminary study to check that the proposed fixation method will not interfere with the chosen end points.

Postfixation of aldehyde-fixed neural tissues in osmium tetroxide is recommended for ultrastructural (EM) examination and is specifically preferred for evaluation of myelin (Bagnell 1995).

Tissue Collection:Neural tissues are prone to artifacts if handled roughly or excessively during necropsy. For instance, prompt removal of the brain from the cranial vault, even following intravascular perfusion with fixative, can increase the propensity for artifactual cell shrinkage and increased basophilia of some neurons (so called “dark” neurons) (Fix and Garman 2000; Garman 1990; Jortner 2006). A better technique for brain collection in rodents is to carefully remove the skull cap following fixative perfusion and then immerse the entire head (with brain left in place in the cranial cavity) into fresh fixative for at least twenty-four hours before brain removal. The practice of immersing the perfused brain in a saline solution may elicit surface vacuolar change (Fix and Garman 2000). Instead, immediate immersion (after rapid weighing) into a suitable fixative is preferred.

Blocking CNS Tissues

A dedicated neuropathology examination, whether planned prospectively or conducted retrospectively, should use some version of this sampling scheme for all vertebrates (Table 1). For convenience in visualizing the many facets of a dedicated neuropathology necropsy, the following discussion has been arranged from rostral to caudal, and from central to peripheral.

Many tissues included in this ideal embedding scheme may be unavailable for retrospective examinations, thus necessitating that the principal scientist use his professional judgment to decide if the overall tissue availability will satisfactorily meet the scientific and regulatory needs of a post hoc study. This situation can be avoided by collecting and embedding the entire nervous system up front as a routine practice, whether or not sections are to be taken during the initial evaluation. An additional advantage of this holistic approach is that neural tissues from all treatment groups will be handled in the same time frame, thereby eliminating systematic variations that may be associated with processing (particularly the length of time in fixative) that can complicate morphometric measurements.

Brain:Examine coronal (“transverse”) sections of one or both hemispheres taken at regular intervals. At least eight sections should be used for rodents, and ten to twelve for larger animals (Garman 2003). If necessary, place complete sections on oversized slides or subdivide brain regions into multiple blocks. Emphasize areas of the brain known to be susceptible to neurotoxicants (e.g., cerebral cortex, hippocampus, cerebellum), but do not forget to sample other regions like olfactory lobes, which are a potential target for both ingested (Crews et al. 2000) and inhaled (Colin-Barenque et al. 1999) toxicants. For studies where a test agent has been delivered directly into the central nervous system (e.g., intraparenchymal injection, intracerebroventricular or intrathecal administration), examine at least eight full coronal sections for all species with additional emphasis on brain regions near the site of test article administration. The use of oversized slides to hold full coronal brain sections from animals larger than rodents offers the great diagnostic advantage of immediately comparing one side of the neuroaxis to the opposite side. This capability is especially important for experiments that employ direct central delivery.

Cranial Nerves:For all species, evaluate the trunks of cranial nerve II (optic nerve [see below]) and the trunks and ganglia for cranial nerve V (trigeminal nerve), which bracket the pituitary gland on the base of the skull. Cranial nerve V may be unavailable for retrospective analyses of studies performed in large animals, as it is usually discarded with the skull at necropsy. It is important to remember that the optic nerve is actually an extension of the brain and therefore is myelinated by oligodendrocytes to or near the optic disk. Cranial nerve VIII, especially in primates, is typically centrally myelinated for a greater distance from the brain than the other cranial nerves.

Spinal Cord:A thorough examination includes evaluation in transverse and longitudinal planes. That said, opinions differ among neuropathologists regarding the best practices for sampling the spinal cord. The two principal debates regard the levels to be evaluated and the preferred orientation of the longitudinal plane.

The optimal assessment in all species would encompass blocks from all three major levels (cervical, thoracic, lumbar); common practice in rodents is to conduct an abbreviated analysis only in the cervical and lumbar regions. The standard sites are generally the cervical (C₄−C₇) and lumbar (L₄−L ₅) intumescences (which house the large motor neurons that innervate the muscles of the forelimb and hind limb, respectively) and the middle portion of the thoracic region. Some pathologists choose the cranial cervical segment (C₁−C₂), which houses the terminal portions of the ascending sensory tracts in the dorsal funiculus; this site is known to be exquisitely sensitive to neurotoxic agents that injure the distal axon (Gottfried et al. 1985; Schaeppi and Krinke 1985; Sills et al. 1998). It may work best to include a section from the C₁−C₂area with sections from the caudal medulla oblongata. Sites in which therapeutic candidates have been directly delivered (e.g., intrathecal delivery, spinal implants) also require evaluation of the cord at the site of drug administration (level at which the catheter tip/implant rests), as well as a transverse section through the cauda equina.

The longitudinal section is included because this plane affords a better ability to detect subtle changes in nerve fiber tracts that run for considerable lengths in the white matter. Such sections can be taken as true longitudinal preparations (in a parasagittal plane 1 to 2 mm to one side of the midline) or as an oblique preparation (made using a dorsoventral cut that angles across the midline). Oblique sections have the advantage of maintaining the overall architecture of the spinal cord (Eisenbrandt et al. 1990).

Spinal Nerve Roots and Dorsal Root Ganglia:Although technically components of the PNS, the trimming procedure for these structures is considered here because they are typically extracted in conjunction with the spinal cord. The ganglia and their associated nerve roots should be collected from the same levels at which the spinal cord is harvested. A retrospective analysis may not be possible without careful planning of the necropsy, as these tissues are often discarded with the spinal column. Some guidelines require PNS tissues to be embedded in “plastic” (EPA 1998a).

Eyes and Optic Nerve:The globe is fixed without trimming and then embedded in such a fashion that it is sectioned through the horizontal axis of the eye from cornea to optic nerve (cranial nerve II). Eyes of large animals may need to be trimmed prior to embedding. In this event, fixation in Bouin’s solution helps to harden the soft tissues sufficiently to enhance trimming.

Staining Tissues of the CNS

The initial evaluation of CNS tissues will generally be performed with sections stained with hematoxylin and eosin (HE), after which a variety of special neurohistology stains are routinely employed (Fix et al. 1996) (Figure 2). One common procedure to evaluate the general integrity of neurons throughout the CNS is application of a Nissl stain, a basic dye to detect RNA of the rough endoplasmic reticulum (RER or Nissl substance) in neuronal cytoplasm. Loss of Nissl substance is evidence that neuron function has been compromised. Cresyl fast violet and thionine are two common Nissl stains. Myelin structure may be demonstrated using either a histochemical stain (such as Luxol Fast Blue [LFB]) or an immunohistochemical (IHC) method (e.g., directed against myelin basic protein [MBP]). Combined stains (e.g., cresyl violet/LFB) are often applied to the same section to allow simultaneous visualization of neuronal and glial elements. Silver impregnation techniques are used to define subtle details of neuronal architecture by highlighting both the cell body and fine axonal processes; examples include the Bielschowsky, Bodian, Holmes, and Sevier-Munger methods. The standard means of detecting markers for specific cell types (e.g., glial fibrillary acidic protein [GFAP] for astrocytes) or physiologic processes (e.g., proliferation or apoptotic markers) is generally an IHC method, many of which can be used across all mammalian species. Immunohistochemistry is limited only by the availability of well-characterized antibodies and the availability of tissue handled in a fashion that preserves antigenicity. The advent of dual-labeling IHC techniques for demonstrating multiple cell populations or co-localization of antigenic determinants to a single cell population has improved the ability of pathologists to understand neurotoxic mechanisms.

Although neuronal degeneration and necrosis can be detected (sometimes quite easily) on HE-stained slides, the use of a selective neuronal degeneration stain can assist in the speed and accuracy of such an evaluation. In this respect a notable advance of great relevance to the neuropathologist is the improvement in the strength and long-term stability of immunofluorescent dyes, which bind preferentially to damaged neurons, such as Fluoro-Jade B (Schmued and Hopkins 2000a) and Fluoro-Jade C (Schmued et al. 2005). Two major advantages of Fluoro-Jade B as a stain for degenerating neurons (Figure 2D) are the ease with which the stain is performed and the ability to stain routinely fixed, paraffin-embedded tissues (Schmued and Hopkins 2000b). Fluoro-Jade B may also aid the identification of degenerating dendrites, axons, and synaptic terminals; these structures are difficult to identify with HE, since the neuropil stains a relatively uniform pink. The exact entity labeled by Fluoro-Jade B is unknown (Ye et al. 2001). The alternative to using a fluorescent dye is to employ a silver degeneration stain (notably cupric silver), which will dramatically highlight affected neurons (Fix et al. 1996; Switzer 1991; Switzer 2000). However, silver degeneration methods have a reputation for being difficult to master (Switzer 2000), likely because proper staining requires special fixation (intravascular perfusion with cacodylate-buffered solutions) and the preparation of frozen sections. Thus, silver degeneration procedures are suitable only for prospective evaluations. Cupric silver stains may be more effective than Fluoro-Jade at demonstrating persistent segments of disintegrating neurites and synaptic terminals, although both stains do allow visualization of these processes (Switzer 1991). The use of a stain specific for neuronal degeneration can actually save time when numerous brain sections must be evaluated because these stains are sensitive and easily assessed. These stains can also help sort out the question of artifactual change from actual neuronal degeneration.

Another important stain in the neurohistology armamentarium is anti-GFAP (Figure 2E). Enhanced expression of this astrocytic filament is considered a robust marker for astrocytic response to neurotoxicity (O’Callaghan and Sriram 2005). In fact, increased GFAP expression appears to be one of the most persistent markers of cell necrosis in the CNS (Fix et al. 1996) and often represents the only durable evidence of a prior injury. IHC stains permitting the identification of a specific neuronal population, such as tyrosine hydroxylase staining to identify dopaminergic neurons (Figure 2F), are useful selectively, especially for quantifying, via stereologic methods, cell numbers in a given region.

The following scheme defines a neurohistology staining battery that long experience has proven suitable for preclinical neuropathology studies. This strategy might appear excessive at first glance, but this battery often reveals changes not readily apparent with HE staining. The decision regarding whether or not to apply these stains broadly or selectively will be based on the experience of the examining pathologist, the time frame for the submission of the neuropathology report, and the available resources (scientific and financial). The various stains are listed in order of priority (with highest priority first). It must be reemphasized that a thorough evaluation (i.e., an adequate number of sections from brain, spinal cord, and peripheral nerves) is more important than the use of special stains other than HE.

Brain:HE (for general morphology); Fluoro-Jade B or a silver degeneration stain (to detect neuronal degeneration); anti-GFAP stain (as a marker for astrocytic hyperplasia associated with neurotoxic injury); and a silver impregnation stain (e.g., Bielschowsky’s) or an anti-neurofilament (NFP) stain (to visualize the integrity of neuronal processes). A myelin stain such as LFB may be added, although paraffin embedding produces extensive myelin artifact.

Spinal Cord:HE, Fluoro-Jade B (or C), anti-GFAP, and silver impregnation (e.g., Bielschowsky’s) or anti-neurofilament stains.

Pathology Analysis of the Peripheral Nervous System in the Preclinical Setting

Even pathologists acquainted with the procurement and analysis of CNS tissues often experience reservations when faced with the need for a detailed examination of the PNS elements. This section will review the standard methods used in neuropathology examinations of the PNS in vertebrate species used in neurotoxicity testing. More detailed coverage of this topic can be found in several texts (Dyck and Thomas 2005; King 1999).

Acquisition and Blocking of PNS Tissues

The extensive ramification of peripheral nerves extending from the spinal cord to effector and sensory organs necessitates that samples be collected at multiple sites. The selective vulnerability of various PNS sites to toxic injury (Anthony et al. 2002) requires that a general survey be performed to ascertain the potential to produce PNS injury. The following sampling scheme is a useful means for ensuring that relevant regions are collected. The nerve should be handled gently and by one end to avoid tension-induced damage (particularly to myelin). The end used for this purpose should be clearly identified, since it will be crushed and thus will not be suitable for microscopic analysis.

Somatic Nervous System:The minimal analysis in paraffin blocks requires evaluation of transverse and longitudinal sections of the sciatic (at two levels [proximal and distal]), posterior tibial (including its muscular branches), and sural nerves to provide a reasonable assessment of proximal to distal locations within the nervous system. This sampling scheme can be performed with relative ease given enough practice. Longitudinal sections are particularly important for evaluating myelin if the samples are to be embedded in paraffin or soft plastic (e.g., glycol methacrylate [GMA]). Tissues trimmed to show both orientations can typically be performed in the same block. Cross (transverse) sections alone are generally adequate for specimens postfixed in osmium and then embedded in a hard resin (e.g., diglycidyl ether [Spurr’s]) suitable for preparing ultrastructural sections because of the available morphologic detail and greater preservation of myelin. Cross-sections typically provide the most straightforward evaluation of nerve components. The addition of a longitudinal nerve section in paraffin or GMA offers the potential to evaluate a much larger expanse of nerve, thus allowing greater opportunity to find a region with few processing-associated artifacts.

Autonomic Nervous System:Certain agents specifically target either the sympathetic or parasympathetic systems (Anthony et al. 2002). The cranial mesenteric and cervicothoracic ganglia are the most readily accessible and consistently recognized sympathetic ganglia to locate consistently in larger animals. The cranial mesenteric ganglion is the easiest to harvest in rodents. These ganglia will generally be unavailable for retrospective analysis. The parasympathetic ganglia are located inside organs and are especially apparent (microscopically) in the muscle layers of the digestive tract.

Skeletal Muscle:This effector organ is sampled because loss of neural input leads to atrophy. Muscle bellies should be evaluated in transverse and longitudinal orientations. The usual sites for collecting samples are the biceps femoris (through the middle) and/or the gastrocnemius (proximal third).

Comparative Merits of PNS Evaluation Techniques

Routine Processing:Standard methods (e.g,, immersion fixation in NBF followed by either paraffin embedding or cryosectioning of frozen tissues) are appropriate for immunohistochemistry, histochemistry, and in situ hybridization, but not for detailed structural or ultrastructural assessments. These embedding media are suitable for evaluating specific nerve components using a large assortment of special stains. However, two limitations must be considered when using conventional procedures. The relatively thick sections (3 to 8 μm) obtained from paraffin-embedded material yield poor resolution of fine structures (Figure 3A). In addition, lipids in myelin are not preserved by formalin, and their subsequent dissolution by xylene during the paraffin embedding process results in large artifactual vacuoles (“bubbling”) within the myelin (Figure 1D). An advantage of immersion fixation is that other tissue samples to be used for biochemical analysis can be acquired from the same animal, thus serving to both reduce the number of animals required and to provide direct comparisons of biochemical changes to structural changes.

Special Techniques:The standard light microscopic evaluation of the PNS is done using “special” fixation and embedding procedures (Dyck and Thomas 2005; King 1999). Initial fixation by transcardial perfusion is preferred to prevent postmortem changes that might impact the analysis, particularly those that can be accelerated by manipulation (such as destruction of myelin, which has little mechanical strength to withstand rough handling). Glutaraldehyde is the fixative of choice to preserve proteins, especially when ultrastructural studies are planned; it does not fully protect myelin. Employ EM-grade glutaraldehyde or distill the solution before use to ensure that the fixative contains no contaminants which might impact structural preservation of the tissue. The perfusion solutions can be prepared in a PIPES (piperazine-1,4-bis[2ethanesulphonic acid])-, sodium cacodylate-, or phosphate-based buffer but should be isotonic and at physiologic pH. Our preferred choice of buffer is PIPES or sodium cacodylate because phosphate buffer requires careful washing to prevent formation of precipitates in subsequent processing steps. The slow rate of glutaraldehyde penetration relative to NBF (particularly given the good perfusion barrier afforded by the perineurium) requires that nerve specimens be postfixed by immersion in fresh glutaraldehyde for four to twenty-four hours before further processing. The tissue is then postfixed again in osmium tetroxide (1% in 0.1 M buffer, pH 7.4 at 20°C for two hours) to preserve myelin, and then carefully dehydrated. The osmium fixation and dehydration steps are critical in obtaining sections suitable for myelin evaluation, as deficiencies at these two stages will yield myelin bubbling or even more drastic alterations in myelin structure, which preclude a detailed observation.

These specialized fixation techniques serve as a prelude to embedding the nerve tissue in hard resin, the medium required to obtain the 1-μm–thick sections that are recommended for routine analysis of peripheral nerve structure (Figures 1Aand 3B). In addition, thin (about 90-nm–thick) sections can be cut from the same resin blocks, thus allowing for ultrastructural evaluation. Low-viscosity epoxy resins (e.g., diglycidyl ether [Spurr’s]) are the preferred resins for this purpose, as they provide good infiltration and allow for thinner sections. However, they cannot be used for most special stains.

Staining PNS Tissues:Initial evaluation of the PNS, as with the CNS, can generally be performed with paraffin embedding and HE stain. The stains used to accentuate findings in PNS nerves and ganglia are comparable to those used for the CNS. Standard stains used in specimens embedded in paraffin or glycolmethacrylate include HE, Bielschowsky’s silver, and LFB. Immunoperoxidase stains may also be used on sections cut from the paraffin blocks. To perform detailed assessment of axons and myelination, some blocks of tissue are often postfixed in osmium tetroxide and embedded in hard resins. The major stains used for sections cut from hard resin are toluidine blue for the “thick” (1 μm) sections to be evaluated by bright-field microscopy and uranyl acetate for the “thin” (~90 nm) ones that will be viewed by transmission electron microscopy (TEM).

Teased Fiber Preparations:Structural evaluation of isolated nerve fibers facilitates the assessment of a single axon and its associated myelin over long distances (up to 10 mm). This method allows for inspection of the nodes of Ranvier, the internodal distance, and the condition of myelin on adjacent segments (Figure 3C). The appearance of such features can help discern between states of demyelination and remyelination, or axonal atrophy and axonal regeneration (Dyck and Thomas 2005; King 1999). Prior to teasing, nerves are fixed by immersion for twenty-four hours in an appropriate aldehyde, separated into individual nerve bundles by removing the connective tissue sheath, and then separated into smaller bundles of fibers. The bundles are rinsed in phosphate-buffered sucrose and fixed in 1% osmium tetroxide for one hour, rinsed again in phosphate-buffered sucrose, dehydrated in graded alcohols (30% to 100%), and then placed in cedar wood oil or glycerin for storage until analysis. Individual myelinated axons are teased apart under a stereomicroscope using forceps, mounted on slides, dried overnight at 60°C, and then cover-slipped with Permount. The individual fibers are examined under a stereomicroscope.

Morphometry:Determination of axonal density and diameter distribution for a given cross-section of peripheral nerve is a useful means for automated quantification of structural changes in peripheral nerves (Romero et al. 2000). Results must be interpreted with the knowledge that a single cross-section represents only a focal sampling and that artifacts may be introduced during processing (e.g., shrinkage by hyper-osmolar fixatives). In normal mature nerve, myelinated fibers range from 2 μm to 20 μm in diameter and demonstrate a bimodal distribution. With axonal atrophy, loss of damaged large-diameter fibers produces a shift to smaller axonal sizes exhibiting a unimodal distribution. In peripheral nerve the axon diameter is correlated with the myelin thickness. The g ratio, the ratio of axonal diameter to the total fiber diameter, falls between 0.4 and 0.7 for most fibers in normal nerves. Fibers with an abnormally thin myelin sheath will have a larger g ratio, and those with abnormally thick myelin will have a small ratio. Calculation of the density of fibers within a given area can also be useful when compared to controls to assess for axonal loss. Lesions (typically degenerated axons rather than myelin defects) occur at very low rates in control animals, and their incidence can increase with age. Therefore, statistical analysis is required to determine whether or not changes in a given study are treatment related.

Fundamental Neuropathology

Neuropathology assessment remains a hallmark of conventional neurotoxicity testing and a critical adjunct procedure for clinical neurology practice. Sampling strategies are designed to assess different cell populations as reflected by their shape (e.g., round vs. pyramidal neurons), neurochemical composition (e.g., cholinergic vs. dopaminergic neurons), location (e.g., cerebellar vs. cortical neurons), and function (neurons vs. astrocytes vs. myelinating cells). Neuropathology procedures can examine only a small fraction of the myriad potential target cells and must rely on comprehensive assessment of a limited number of regions in the nervous system. Where available, the use of antemortem clinical findings to target specific CNS and PNS regions for neuropathology assessment is an important means of refining the sampling scheme.

Toxicant-induced histopathologic changes in the nervous system are typically classified by cell type and affected subcellular site. For example, neuronal damage is categorized by sub-cellular location: neuronopathy (injured cell body), axonopathy (axonal abnormality), peripheral neuropathy (aberration in a peripheral nerve or at neuromuscular junctions), or myelinopathy (abnormal myelin sheaths). Such categories can be further split according to the mechanism that launched the structural lesion; for instance, axonopathies include axonal atrophy (decreased axon diameter), dystrophy (intra-axonal accumulation of cytoskeletal elements or organelles), and degeneration (axon disintegration), among others. A primary lesion at one site can yield secondary changes in other locations (e.g., Wallerian degeneration of the axons and dendrites attached to a damaged neuron) or in other cell types (glial reactivity in response to damage in adjacent neurons). Some changes are readily evident on routinely prepared sections, whereas others can only be demonstrated reliably by using special stains (e.g., silver staining or neurotransmitter-specific immunohistochemistry to reveal nerve terminal degeneration). However, not all neurotoxicants elicit appreciable neuropathology. Some, such as agents that interfere with synaptic transmission (e.g., the glycine antagonist strychnine), engender a biochemical abnormality that is typically not associated with structural lesions in conventional neuropathology preparations.

Most chemicals evaluated for neurotoxicity have been evaluated in rats or other rodents. Despite significant neuroanatomical differences, the response of humans and rodents to many neurotoxicants is remarkably concordant (Dorman 2000). This pattern is usually true for chemicals affecting CNS neurons (e.g., trimethyltin [TMT]), PNS nerves (e.g., acrylamide), and myelin (e.g., triethyltin [TET]). However, the similarity is not universal, as some agents that are profoundly toxic to humans (and nonhuman primates) produce much less toxicity in rodents (e.g., manganese in the CNS, methanol in the retina).

Neuroimaging Applications

Noninvasive Assessment of the CNS

During the past decade, many sophisticated imaging technologies have been introduced into biomedical research. The list includes “standard” (MRI) and functional (fMRI) magnetic resonance imaging, computerized tomography (CT), ultrasound (US), laser Doppler flowmetry, optical imaging, single photon emission computed tomography (SPECT), magnetic resonance spectroscopy (MRS), and positron-emission tomography (PET) (Chatziioannou 2002; Green et al. 2001; Heckl et al. 2004; Maeda et al. 1999). These innovative methods offer unprecedented in-life opportunities for in situ measurements of anatomic (MRI, CT, US) and biochemical (optical imaging, SPECT, MRS, PET) end points in small laboratory animals, including mice and rats. Like their human counterparts, the small animal instruments are noninvasive and can, in most cases, be used to study the same animal at multiple times over a protracted period. Neurobiology has been a particularly fruitful arena for investigation, particularly with respect to MRI (Benveniste et al. 2000; Hesselbarth et al. 1998; Kornguth et al. 1994; Sills et al. 2004) and PET (Suehiro et al. 1993; Toyama et al. 2005) applications. One major advantage of MRI is that the entire brain can be scanned for lesions without disrupting the tissue integrity for subsequent light microscopic evaluations (Maronpot et al. 2004). The data acquired using these applications, either alone or together, have provided astonishing new insights into the relationship between structure and function, which previously were beyond the ability of traditional pathology techniques.

The imaging modalities that provide anatomic data (MRI, CT, US) rely primarily on distinct endogenous properties of different organs to form images. For example, US depends on reflection of acoustic energy from boundaries between structures with dissimilar acoustic properties. However, although US can define the extent of diseased tissues, the images formed have no mechanistic relationship to a specific metabolic aberration that caused the anatomic change; instead, alterations in organs represent the sum of myriad biochemical changes that occurred well before the divergence of structural and acoustic properties from the norm.

In contrast, the imaging techniques that provide functional data (optical imaging, SPECT, MRS, PET) illuminate a specific biochemical pathway within a given set of structures, thereby offering the possibility of direct and quantitative examination of deviant homeostatic mechanisms during all stages of disease evolution. For instance, for PET the subject—human or animal—will be given (usually via the vascular system) a positron-emitting agent (e.g., the glucose analog ¹⁸F-fluo-rodeoxyglucose [FDG]), which will necessarily follow a particular metabolic pathway in discrete body regions; any medically relevant biochemical pathway can be sampled if a suitable compound can be identified and labeled with a positron-emitting isotope. The PET images allow the amount of compound at any spatial location to be measured quantitatively and dynamically as the material traverses the selected pathway; a common application for neurobiological experiments uses FDG to characterize metabolic activity in different brain regions, particularly in relation to various sensory stimuli (Shimoji et al. 2004; Toyama et al. 2004). Additional uses for small animal PET in both academic and industrial laboratories are to develop and evaluate the efficacy of novel radiolabeled diagnostic agents and therapies for neural diseases. This latter endeavor has been applied particularly to new genetically engineered mouse models that mimic human neurological disorders (Toyama et al. 2005). Added advantages of PET for brain imaging are its exquisite sensitivity in detecting nanomolar changes in concentrations of various receptor molecules (Fujita et al. 2007) and its ability to yield accurate estimates of drug concentration in structures of interest (Toyama et al. 2004). Further advances in later-generation PET instruments (Seidel et al. 2003; Tai et al. 2001) have substantially improved spatial resolution, sensitivity, and functionality to levels largely thought impossible only a decade ago, thereby holding the promise that small animal PET will find even wider application in neuropathology experiments in the future.

In like manner, MRS is gaining ground as a clinical tool for guiding neuropathology analysis, chiefly in the arena of serial assessment of brain neoplasms. This method provides detailed spectroscopic information about focal abnormalities that can be correlated point-to-point with pathological findings, creating the potential for serial detailed observations of CNS lesions. In addition to anatomic information, MRS follows physiologic parameters related to metabolites as well as hemodynamic and diffusion variables; the molecule N-acetyl aspartate (NAA) is a marker for normal tissue, whereas choline-related signals correlate well with highly cellular areas of a neoplasm (Law et al. 2004; Marcus et al. 2007). Furthermore, the MRS data are well correlated with neuropathology findings from biopsy materials (Gujar et al. 2005). The ability to evaluate individual specimens and provide point-to-point correlative data with in vivo imaging methods overcomes the obstacles of microheterogeneity, which is prevalent in CNS disease and which is compounded in human clinical medicine by the typically small number of subjects.

Noninvasive Assessment of the PNS

Morphological examination of the PNS is complicated by the need for intricate processing techniques and the relatively small amount of tissue that can be analyzed because of the extensive ramification of the PNS. Noninvasive imaging methods are of great interest owing to their capacity to sample more sites in more rapid fashion and at multiple time points during life. Conventional MRI has been used to map the T2 values of the major water compartments in peripheral nerve to the axoplasm, endoneurium, and myelin (Does and Snyder 1996). Injuries that alter these compartments will be reflected by a change in their respective contribution to the total T2 map, whether induced by mechanical injury to axon integrity (Webb et al. 2003) or by toxicant-induced damage to myelin (Valentine et al. 2007). Future challenges for this approach involve evaluating its utility in vivo, determining its sensitivity for detecting early changes, and correlating spectral changes with specific structural changes.

Current Practices and Challenges in Developmental Neuropathology

Scope of Developmental Exposure to Neurotoxicants

Children are exposed to numerous xenobiotic agents both before and shortly after birth. Approximately 85,000 industrial chemicals are manufactured in the United States each year, and another 2000 to 3000 new chemicals are registered annually (Goldman 1998). One estimate suggests that 3% to 5% of chemicals will be neurotoxic (Claudio 1992). However, a recent survey of high-production-volume chemicals (those made yearly in quantities exceeding one million pounds) suggests that 67% have never been tested for neurotoxicity (Environmental Defense Fund 1997). Many pharmaceutical agents taken by pregnant and nursing women have neuroactive properties or affect hormonal systems.

The nervous system is the damaged system in approximately half of the children born each year in the United States with one or more anatomic and/or functional deficit (Bearer 2001). Structural defects include gross malformations such as neural tube defects, which afflict one infant per 1000 live births in most countries but can increase markedly (up to 8 per 1000 births) in regions with substantial pollution (Anonymous 1991), and microscopic neural lesions with concordant functional deficits (Purpura 1974). Many chemicals (ethanol, lead, methyl mercury, polychlorinated biphenyls) and pharmaceutical agents (nicotine, phenytoin) have been shown to elicit such neural lesions. The potential also exists that subtle developmental neurotoxicity will predispose the adult nervous system to premature senescence (Reuhl 1991) or early-onset degeneration (Godfrey and Barker 2001), or even to dysfunction following stress (Buelke-Sam and Kimmel 1979) or a later neurotoxic challenge.

Current Practices in Developmental Neuropathology

The pattern of neural development is generally comparable among mammals (Butler and Juurlink 1987), although species-specific traits are apparent (Rice and Barone 2000). The elements of the nervous system at both the cellular and organ level develop at different times both during gestation and after birth (Rodier 1980). Thus, damage induced by developmental neurotoxicants will be limited to specific sites depending on the timing of the exposure. In general, the consequences of chemical entry are frank anomalies or death if during early gestation, microscopic lesions and functional deficits if during late gestation or the neonatal period, and functional abnormalities alone if during adolescence.

Conventional developmental neurotoxicity testing (DNT) is performed in rats to identify hazards and assess risk. The standard study design (Bolon et al. 2006; Kaufmann 2003) incorporates maternal oral exposure from gestational day 6 to postnatal day (PND) 10 for multiple dose groups, including negative (vehicle) and positive (chemical or nonchemical (Maurissen and Marable 2005) control groups. Sometimes an untreated control group is added to assess the impact of maternal stress on developmental toxicity (Chernoff et al. 1987). Each dose group consists of sixty to eighty animals divided into cohorts (n = 10 individuals per sex per group, using one or two individuals per litter), with each subset being used for a particular procedure. Neural tissues are typically acquired from two or three cohorts dedicated to neuropathology analysis—one each terminated on PND 11 (the day after exposure ends), PND 21 or 22 (weaning), and PND 60. The PND 60 cohort is always included, whereas institutions may choose to use only one of the earlier time points depending on the guidelines issued by various regulatory agencies. Tissues from PND 11 are fixed by immersion in Bouin’s solution or buffered formalin (followed by Bouin’s solution to harden tissues), whereas those of PND 21 and PND 60 are typically fixed in situ by formalin perfusion followed by additional immersion fixation. Brains are weighed and measured prior to histologic processing. Specimens from the control and high-dose groups are prepared according to conventional techniques (Garman et al. 2001) and then analyzed qualitatively as well as quantitatively. All of the tissues recommended for neuropathology evaluation in adult animals are also evaluated in young adult rats (PND 60 to 75).

Current national (EPA 1998b) and international (OECD 2007) DNT neuropathology guidelines are adapted from the comparable documents set forth for adult neurotoxicity studies (EPA 1998a; OECD 1997). Multiple coronal (transverse) brain sections of each rat are produced in a routine DNT study. These sections must be homologous so that comparable structures are available among all animals for morphometric analysis; such standardization is best achieved by trimming tissues using external anatomic landmarks rather than brain molds. Some differences of opinion exist among pathologists regarding the number of levels to examine. The main consideration is that the DNT neuropathology analysis should sample at least the major structures correlated to known functional and neurochemical domains (Bolon et al. 2006).

The regulatory guidelines relating to the DNT neuropathology evaluation of the PNS allow some latitude in tissue sampling. In general, peripheral nerves routinely taken for DNT evaluation are harvested from the hind limb. The nerves may be fixed in situ or following removal, as long as care is taken not to stretch the nerves. Nerves are embedded in both cross and longitudinal orientations.

Multidisciplinary Approaches and Future Directions In Neurotoxicity Assessment

Evaluation of the nervous system in vertebrates is complex, requiring an integrated approach using techniques from several disciplines to provide even cursory insights into the impact of potential toxicants on neural structure and function. Even general pathologists are familiar if not expert with the usual battery of procedures used to perform a postmortem anatomic assessment of neural integrity. However, newer techniques combining structural evaluation with region-specific molecular and functional analyses have substantially improved the potential for assembling a truly integrated picture of the nervous system in health and disease (Sills et al. 2005). It behooves us as biomedical researchers to have an understanding of the strengths and limitations of these newer techniques.

Clinical Assessment of the Nervous System

The clinical function of the CNS can be assessed in many ways. Behavioral end points are generally noninvasive and can be used to assess subjects repeatedly during the course of an experiment. End points commonly used in neurotoxicity bioassays include the functional observational battery (FOB), tests for motor and sensory behaviors, and assays of learning and memory (Boyes 2001; Moser 1990). Importantly, such tests often lack specificity for the nervous system, since changes may also occur indirectly from systemic toxicity. In addition, behavioral tests generally evaluate the coordinated activity of large networks of neurons, so that changes in CNS function can rarely be attributed to a single neuroanatomical location.

For example, the FOB is commonly employed to assess the functional neurological status of laboratory animals, particularly rats (Moser 1990; Moser 2000). This basic screen is simple, rapid, inexpensive, and noninvasive; its main features, including home cage observations and manipulative tests that assess various reflexes and the animal’s reactivity to various sensory stimuli, are comparable to those performed in a standard clinical examination undertaken on patients in a general neurology practice. Functional abnormalities in the basic clinical examination and FOB can provide clues to help localize neuroanatomic lesions (Morgan et al. 2004; Moser et al. 1998), so the neuropathologist must have a general understanding about what regions of the nervous system control given behaviors. Functional effects may result from injury to either a peripheral nerve trunk or to a CNS nucleus or through a biochemical mechanism.

Sequential Anatomic Evaluation of the Nervous System

Noninvasive and conventional neuropathology techniques can be used in series to provide more information from a given sample than can be garnered by a single analytical modality. An example of this strategy is found in a recent collaborative investigation of carbonyl sulfide (COS) neurotoxicity in rats (Figure 4).

Parallel studies of neuropathology using light microscopy and magnetic resonance microscopy (MRM) catalogued the distribution of COS-induced brain lesions (Sills et al. 2004). A major advantage of this tandem analytical approach was that MRM of the intact brain was able to localize necrotic lesions in the majority of rats without disrupting the brain parenchyma. The importance of this advantage cannot be overstated. The primary sensitive site in all subjects was found to be the caudal (posterior) colliculus, a neural region known for its high metabolic rate and high capillary density (Sokoloff 1981; Sokoloff et al. 1977). Energy use in this area was impaired, as indicated by a substantial, dose-dependent reduction in cytochrome oxidase activity (Morgan et al. 2004). The caudal colliculus composes a deep nucleus of the midbrain and is usually not sampled during routine assessment of the CNS. Thus, this lesion might have been missed during routine histopathology examination in the absence of the prior MRM data. In turn, the excellent resolution of conventional light and EM microscopes can find and catalog subtle cellular and subcellular changes that might have been masked by the limited resolution (20 μm to 30 μm at present) of MRM. These paired neuropathology methods were also able to detect lesions in the auditory brain stem that were not found using standard behavioral screens (e.g., FOB) but were correlated with electrophysiological anomalies demonstrated in the brain stem auditory-evoked response (BAERS) (Herr et al. 2007; Morgan et al. 2004).

The demonstrated success of this approach warrants further investigation of such multimodality, serial anatomic assessments in future neurotoxicological studies.

Novel Techniques for Correlating Anatomic and Molecular Pathology End Points

Surrogate end points to evaluate neuropathologic damage (e.g., measurement of striatal dopamine levels in brain homogenates as an indirect index of neuronal loss in the substantia nigra) do not represent true anatomic gauges of neural disease. Instead, the most reliable means to accomplish such neuropathology analyses are direct structural assessments (e.g., counts of dopamine neurons in step sections using unbiased stereology). Such direct methods are slow and unwieldy for experiments in which a large number of treatment conditions or animals are to be compared, even if the target cell population to be evaluated has been emphasized using a special stain (e.g., an anti-tyrosine hydroxylase immunostain to label dopamine-containing neurons in substantia nigra), because labeling cannot mask the difficulties in interpreting count data resulting from heterogeneity of neuronal populations in different portions of the substantia nigra.

In human cases of neurotoxicity, such grossly invasive techniques cannot be undertaken with impunity. Sampling limitations owing to the exquisitely refined localization of CNS functions pose unique problems for the field of neuropathology and the classification of CNS diseases. However, rapid advances in molecular biology and bioinformatics have created unprecedented opportunity for progress in classifying neural diseases, particularly in the CNS. Thus, new approaches in neuropathology rely on more correlative analyses compared to other areas of pathobiology. Tools being used to overcome the obstacles include point-to-point correlative pathology with sophisticated imaging modalities (as previously discussed), laser capture microdissection (LCM) to harvest specific cell subpopulations from heterogeneous samples, and gene expression and gene silencing methodologies.

Access to tissue represents an obstacle to progress in clinical neuropathology. Biopsies tend to be small in size, and repeat craniotomies are avoided, preventing sequential analysis of diseases over time. The use of LCM as a tool for addressing cell-specific events in neurobiological research has accelerated rapidly in the last decade (Böhm et al. 2005). By isolating individual cell types, and in some instances individual cellular structures, LCM has allowed the identification of pathogens and genetic events within distinct cell subsets. In this fashion, sections of neural tissue can be evaluated by standard light microscopy (including acquisition of relevant photomicrographs), after which specific cells can be removed and submitted for biochemical analysis for DNA, RNA, and/or protein; such molecular end points are equivalent to large-scale measurements acquired from tissue homogenates made from manual microdissection of the nervous system. The main advantage to LCM followed by small-scale biochemical analysis is that anatomic lesions can be directly correlated to the molecular pathogenesis in the same patient.

Molecular classification of CNS diseases is a science in its infancy, but it is growing rapidly. Genetic assessment in neuropathology has been approached with two basic paradigms: measurement of individual genes whose expression or silencing is known or suspected to be involved in a given disease process, and microarray assessment of numerous genes to identify those which are linked to specific diseases. Common techniques for evaluating (and, in most cases, quantifying) a single gene or a few genes include loss of heterozygosity (LOH), fluorescent in situ hybridization (FISH) (Fuller and Perry 2002), and real-time polymerase chain reaction (PCR) (Aldape et al. 2002). Tissue arrays are a high-throughput mechanism for evaluating pathology samples, facilitating the accumulation of gene expression data. High-density tissue arrays allow the evaluation of several hundred genes at a time. This latter technology will likely see increased use in neuropathology research as automated analytical techniques and reliable commercially available arrays become more prevalent.

Microarray platforms have led to the discovery of combinations of gene expression or gene silencing, which are related to neuropathology classification and prediction of outcomes, especially in tumor biology of the human CNS. For medulloblastomas, microarray data have allowed molecular distinction of medulloblastoma from desmoplastic medulloblastoma, malignant glioma, and atypical rhabdoid tumors (Mischel et al. 2004). In addition, combinations of gene silencing or gene expression identify subsets of patients that respond to individual therapies, thereby providing the capacity to tailor more appropriate treatment regimens to specific individuals. Similar approaches are being used to identify genes involved in neurotoxic responses. An example is the time-dependent expression of genes in the initial stages of methamphetamine exposure (Thomas et al. 2004).

The rapid refinements in neuropathologic classification of CNS diseases afforded by these interdisciplinary techniques will likely lead to reliable molecular subclassifications of common neural disease in the near future, with correlations to etiology, response to therapies, and clinical outcomes.

Summary

The goal of making real-time connections between a specific type of anatomic disruption and the toxicant-induced biochemical and molecular alterations that are engendered is fast approaching. Recent innovations will add considerable power to neuropathology evaluations conducted in support of toxicologic research programs going forward. However, for the time being the conventional tests that have served so well for so long will not be displaced, but merely supplemented. Neuropathologists are in a unique position to contribute as leaders in further advancing this field. All we need to do is remember what we already have learned, and to be receptive to new scientific and technical opportunities.