INTRODUCTION

Oligodendrocytes are remarkable cells. In vertebrate evolution, the advent of oligodendrocytes and myelination transformed the CNS by allowing fast and energy efficient communication between neurons, ultimately fostering the evolution of animals with complex, highly integrated motor, sensory and cognitive functions. In humans, myelination underlies most of the early developmental neurological milestones, and new myelination continues to be important into the third and fourth decades. Human diseases involving oligodendrocyte dysfunction are devastating, and those such as multiple sclerosis (MS) account for a significant proportion of neurological disease. Since the first studies of myelin protein biochemistry in the late nineteenth century, myelin proteins and lipids have received intense experimental investigation. Extensive reviews of the biochemistry, genetics, immunogenicity and localizations of the major myelin proteins and lipids have been published. Recent genomic and proteomic studies have begun to provide a complete list of myelin and oligodendrocyte-enriched molecules. The goal of this chapter is to consider the contributions of different myelin proteins and lipids to (1) the structure of central nervous system (CNS) myelin, (2) the cell biology of myelin formation and (3) their roles in vital interactions between oligodendrocyte and axons. The emerging picture of oligodendrocyte myelination is a process that is extremely fault tolerant and inextricably intertwined with axonal function.

OLIGODENDROCYTES HAVE A HIGHLY POLARIZED SHAPE

Few cells have as extreme a shape as myelinating oligodendrocytes. Before discussing their molecular organization, it is therefore important to have a clear picture of oligodendrocytes and their myelin membranes.

INTRODUCTION

Over the past decade, there has been tremendous progress in our understanding of oligodendrocyte biology. Many of the factors that are important for the specification of oligodendrocytes during development have been identified (Kessaris et al., 2008). In addition, the emergence of genomic technologies has greatly improved our understanding of the factors that affect oligodendrocytes during development and in disease. In this chapter, we highlight some of the significant discoveries that have advanced our understanding of the genetic factors that are important in oligodendrocytes, and we identify some of the contributions that were made to this field using genomic technologies.

OLIGODENDROCYTE SPECIFICATION

A great deal of progress has been made in understanding the earliest events of oligodendrocyte specification. The spinal cord has received the most attention, and soluble signaling molecules such as sonic hedgehog (Shh) (Cai et al., 2005) and bone morphogenic proteins (BMPs) are key early regulators of oligodendrocyte development in this part of the central nervous system (CNS). Shh is secreted from the ventral floor plate and is required for the specification of oligodendrocytes (Orentas et al., 1999). Shh signaling induces the expression of the basic helix-loop-helix (bHLH) transcription factors Olig1 and Olig2 in a region of the ventral spinal cord referred to as the motor neuron progenitor (pMN) domain (Lu et al., 2000; Zhou et al., 2000). Shh and Olig gene expression is required for oligodendrocyte specification in the spinal cord and brain (Zhou and Anderson, 2002).

Most computational models of conditioning adopt associative strength (V) as the variable that tracks learning about the association between a conditioned stimulus (CS) or action and the reinforcing unconditioned stimulus (US). Many of these models make very simple assumptions about the arguments of associations – the CSs and USs themselves. For example, the Rescorla–Wagner model treats these stimuli as singular units such that, during learning, a single connection strengthens between the CS unit and US unit (Rescorla & Wagner, 1972; Wagner & Rescorla, 1972).

Within-compound associations: models and data

Pavlov (1927) discovered that both positive and negative mediation effects can occur when a target cue (X) is presented during training in conjunction with a nontarget cue (A). Positive mediation effects refer to situations in which the presence of A during training results in more excitatory behavioral control by X than if X was trained elementally. An example of positive mediation is second-order conditioning, which occurs when A–unconditioned stimulus (A–US) pairings (Phase 1) precede X–A pairings (Phase 2, which presumably establishes an X–A within-compound association), resulting in more excitatory conditioned responding to X than in a control condition lacking one or the other phase (Pavlov, 1927).

INTRODUCTION

Glutamate is the major excitatory neurotransmitter in the central nervous system (CNS). Glutamate exerts its role by activating ionotropic (Dingledine et al., 1999) and metabotropic glutamate receptors (Conn and Patel, 1994). Ionotropic glutamate receptors are ligand-gated ion channels consisting of α-amino-3-hydroxyl-5-methyl-4-isoxazole-propionate (AMPA), N-methyl-d-aspartate (NMDA), and kainate receptors. Metabotropic glutamate receptors are not ion channels, but rather G-protein-coupled receptors that activate biochemical cascades that can indirectly influence ion channels. Glutamate transporters are responsible for regulating the concentration of glutamate in the vicinity of synaptic and extrasynaptic glutamate receptors (Tzingounis and Wadiche, 2007). Excessive accumulation of extracellular glutamate causes prolonged activation of glutamate receptors resulting in excitotoxicity (Lipton and Rosenberg, 1994). High levels of glutamate cause an excessive influx of calcium into the cytosol via glutamate receptors coupled to ion channels, resulting in the pathological activation of a number of enzymes including phospholipases, endonucleases and proteases, such as calpain. Increased glutamate levels also result in mitochondrial dysfunction (Choi, 1988). Excitotoxicity is a common pathway of injury in many neurological diseases leading to neuronal cell death. Excitotoxicity has also been implicated in oligodendrocyte death in disorders of cerebral white matter, including periventricular leukomalacia (PVL).

PERIVENTRICULAR LEUKOMALACIA

Periventricular leukomalacia (PVL) is the major brain pathology, in long-term survivors of prematurity. Approximately 5–10% of babies born with a birth weight of less than 1500 g develop cerebral palsy. Nearly 5500 new cases of cerebral palsy in premature infants are diagnosed each year in the United States.

In the late 1970s, Wagner (1976; 1979) presented some views about associative learning that were centered upon the notion that events that are already represented, or “primed” in active memory, are not as effectively processed as they otherwise would be. One of several forms of evidence for this supposition was the so-called conditioned diminution of the UR, a phenomenon originally reported by Kimble and Ost (1961), in the context of human eyeblink conditioning.

NEUROGLIA

For the past 160 or so years the cells of the nervous system have been divided into two main categories: neurons and glia (Kettenmann and Verkhratsky, 2008). Prior to this, ever since the first image of a neuron was published in 1836 by Gabriel Valentin, the nerve cell had been in a class of its own (Lopez-Munoz et al., 2006). Some 20 years later in 1856 the term neuroglia was introduced by the German physician Rudolph Virchow. Virchow, also known as the “Pope of pathology” (Kettenmann and Ransom, 2005; Magner, 2002), described a “connective substance … in which nervous system elements are embedded” and referred to it as “nervenkitt” (or nerve putty). This description led to the use of the term “neuroglia,” which derives from archaic Greek, meaning something sticky or clammy. The notion that neuroglia were there merely as neural putty was treated with the reverence usually reserved for a bona fide papal encyclical and as such neuroglia remained sidelined for decades to come. Even though Virchow was responsible for the term neuroglia coming into use, at this stage he did not recognize that it was made up of cells rather than an acellular connective tissue.

Birds do it, bees do it; maybe ordinary fleas do it. They all learn from experience. But why is learning so ubiquitous? Why not just be born already knowing how to behave? That would save a lot of time and a lot of error. Presumably, we are born ignorant either because evolution is unfinished or because what we need to know is too complex to be fully coded in the genome. Either way, it seems that evolution has cleverly found a mechanism for dealing with the birth of ignorance; a mechanism that we call learning.

Of course, it may be that learning is merely something that organisms do for fun in their spare time. Perhaps there is not much adaptive value in learning, and little cost, and therefore no selective pressure on the mechanisms of learning.

Over a century's worth of research on animal conditioning has generated a wealth of empirical evidence relating to the influence of associative history on future learning. Historically, in developing models of learning, theorists have tended to concentrate on one such influence and build a model centered on that aspect. For example (and to anticipate), the model developed by Mackintosh (1975) deals exclusively with examples of positive transfer of the processing of cues; Pearce and Hall's (1980) model was developed purely on the basis of examples of negative transfer of cue processing; and Rescorla and Wagner's (1972) model deals with competition between cues on the basis of the surprisingness of the outcome.

This book contains the presentations given during the Duke Symposium on Computational Models of Conditioning, which took place between May 15th and May 17th of 2009 at the Duke Campus in Durham, N.C. The meeting was sponsored by the Duke Department of Psychology and Neuroscience, the Duke Office of the Vice Provost for International Affairs, and the Duke Arts and Sciences Research Council. All the participants and I are indebted for their generous support.

The meeting was organized with the assistance of my friend and former Ph.D. advisor Professor John Moore (University of Massachusetts at Amherst). I am particularly thankful to John for helping me in finding a group of participants who contributed both well-established and novel theories of classical conditioning. I am also grateful to Munir Gunes Kutlu for his help in running many aspects of the meeting.

The models

John Kruschke and Rick Hullinger (Indiana University, USA) prepared the chapter on “The evolution of learned attention.” In this chapter, the authors use simulated evolution, with adaptive fitness measured as overall accuracy during a lifetime of learning, and show that evolution converges to architectures that incorporate attentional learning. They also describe the specific training environments that encourage this evolutionary trajectory, and how we assess attentional learning in the evolved learners. Interestingly, the resulting attentional mechanism is similar to that proposed by Mackintosh (1975).

INTRODUCTION

Louis-Antoine Ranvier first described myelin in 1878 (Ranvier, 1878); however it was not until 1928 that del Rio Hortega described the oligodendrocyte – the myelin-forming cell of the CNS (Hortega, 1928). Myelin pathology present in spinal cord injury was first mentioned in 1907 (Holmes, 1907) and had become more widely recognized by the 1960s (Bunge et al., 1960). Today the oligodendrocyte with its compact and non-compact myelin components has become an avidly pursued therapeutic target, yet our understanding of the physical and functional changes in the oligodendrocyte myelin with injury are relatively understudied and misunderstood.

Spinal cord injury (SCI) is a devastating condition currently affecting 250 000 people in the United States, with 11 000 new cases recorded each year (2005). SCI is a complex neurological disorder with varied pathological patterns. The majority of injuries result in permanent anatomical and functional deficits. The most frequent type of injury – contusion injuries in the cervical and thoracic regions – is caused by a bone or disk displacement into the spinal cord due to bone dislodgement or fracture of the spinal column (Rothman and Simeone, 1992). Few injuries are caused by laceration or transection of the spinal cord. Contusion injury is typified by a necrotic core or cavity, which is partly surrounded by a rim of remaining white matter containing spared descending and ascending axons (Balentine, 1978a; Blight, 1983). SCI results in a loss of function due to the disruption of sensory and motor pathways and glial cell support.

INTRODUCTION

Myelin in the peripheral nervous system (PNS) and central nervous system (CNS) is formed by Schwann cells and oligodendrocytes respectively. The myelin formed by the two glial cell types electrically insulates the axons and restricts the generation of action potentials to the nodes of Ranvier, myelin-free regions separating the myelin segments, allowing for rapid propagation of action potentials along the axons. While there are striking overall similarities in the mechanisms of myelination and the structure and molecular composition of PNS and CNS myelin, there are also important differences between the two glial cell types. For instance, Schwann cells elaborate a single myelin segment around a single axon, while oligodendrocytes may myelinate up to 60 different axons (Figure 2.1). Furthermore, Schwann cells – but not oligodendrocytes – are surrounded by a basal lamina that is continuous with the adjacent internode. In addition, nodes of Ranvier in the PNS are covered by Schwann cell processes, while axons at CNS nodes are bare. A characteristic feature of CNS myelin not present in PNS myelin is the so-called radial component, interlamellar claudin-11-positive strands spanning the myelin. There is also a slight difference in the periodicity of mature compact CNS and PNS myelin, and important differences in the molecular composition of CNS and PNS myelin. For instance, proteolipid protein (PLP) is the major myelin protein in the CNS, while P0 and peripheral myelin protein 22 (PMP22) are exclusively found in PNS myelin.

INTRODUCTION

Oligodendrocytes, as the myelin-producing cells of the central nervous system (CNS), are exactly what their Greek-derived name “oligodendroglia” suggests: they, alongside astrocytes, the non-neural microglia and ependymal cells, have been characterized as the “glue” that holds together the intricate apparatus of our brain. The fact that oligodendrocyte and astrocyte cells outnumber neurons by ten to one illustrates their importance, which is particularly highlighted by the oligodendrocyte's role in accelerating transmission of axonal action potentials. On the other hand, oligodendrocytes are involved in a number of serious diseases of viral, metabolic and immunological origin. This chapter tries to shed light on the immunobiological properties of oligodendroglial cells in the healthy and diseased CNS. We will begin with an overview of diseases featuring oligodendrocyte/immune system interactions and will then, in the second part, focus on the molecular repertoire that allows these cells to interact directly or indirectly with immune cells. Subsequently, we will discuss oligodendrocytes as antigen-presenting cells and finally we will present data on direct oligodendroglial/immune cell interactions.

IMMUNE-MEDIATED DISEASES AFFECTING OLIGODENDROCYTES

Multiple sclerosis

Multiple sclerosis (MS), which was first described by the French neurologist Jean-Martin Charcot in 1868 (Charcot, 1868), is a chronic inflammatory disease of the CNS of unknown etiology (Hemmer et al., 2006). Although an ideal system for the classification of different MS stages does not yet exist (Van der Valk and De Groot, 2000) there is broad consensus that loss of myelin due to oligodendrocyte damage or death together with axonal degeneration leading to reactive glial scar formation are the key hallmarks of this disease (Trapp and Nave, 2008).

INTRODUCTION

The oligodendrocyte lineage is among the best studied and most well understood of all the lineages in the vertebrate central nervous system (CNS). Oligodendrocytes are the cells responsible for the formation of CNS myelin. Structurally compact as distinct from non-compact myelin appears as concentric wraps of modified plasma membrane that ensheathe individual lengths, i.e. internodes, of neuronal processes or axons, that connect neurons with their targets. This compact internodal myelin is discontinuous as each of the up to 50 oligodendrocyte processes forms only one internode. The myelination by the oligodendrocyte processes results in accelerated conduction of signals along the axons and lowering of the threshold for propagating such signals. In the adult CNS, myelination is critical for the normal functioning of the CNS such that diseases and injury that result in the loss of myelin lead to functional deficits. The ability to isolate and grow cells in culture that will generate oligodendrocytes has allowed researchers to understand some of the fundamental steps that lead from a neural stem cell to a myelinating oligodendrocyte in the adult CNS. The identification of oligodendrocyte-lineage-specific cell surface molecules, growth factor receptors and transcription factors has facilitated the labeling and manipulation of the lineage both in vivo and in culture. These studies have revealed extensive plasticity in the development of oligodendrocytes and provided critical insights into the behavior of oligodendrocytes and their oligodendrocyte precursor cells during development in a variety of different regions of the CNS as well as under a number of pathological conditions.