Neurotrophic Factors Ensure That Adequate Numbers of Neurons Survive

A critical factor that determines whether a given neuron survives or dies during development is its success in accumulating neurotrophic factors of specific kinds, different kinds for different neuronal types (Fig. 24-1). Neurotrophic factors are produced in limited amounts by target tissues (e.g., muscle, glands, other neurons), gobbled up by presynaptic endings, and transported back to the cell body. There they act to prevent apoptosis (programmed cell death) and to promote growth. Going back to the spinal cord example, more dorsal root ganglion cells and motor neurons survive at levels where there’s a lot of target tissue in the periphery (e.g., lower cervical) than at levels where there’s less (e.g., midthoracic). But this isn’t restricted to the spinal cord—throughout the nervous system, something like half of all the neurons produced during development die before birth.

Figure 24-1 Competition for trophic factors and its effects on neuronal survival during development.

Axonal Branches Are Pruned to Match Functional Requirements

Long after neurons finish competing with each other for survival, they continue to compete for neurotrophic factors in an effort to preserve their connections (Fig. 24-2).

Figure 24-2 Competition for trophic factors leading to survival and proliferation of axonal branches during development.

The best-known example is the innervation of skeletal muscle by lower motor neurons. Early on, individual muscle fibers receive inputs from multiple motor neurons. By about the time of birth, all but one input has been pruned away and the sole survivor develops into a single elaborate neuromuscular junction (THB6 Figure 8-11, p. 184). Similar pruning goes on throughout the nervous system; depending on the area involved, this is a process that may continue well after birth.

There Are Short-Term and Long-Term Adjustments of Synaptic Strength

Some of the short-term changes in synaptic strength follow naturally from normal synaptic function (Fig. 24-3). A little extra Ca²⁺ hanging around in a presynaptic terminal after transmitter release, for example, can result in potentiation of transmitter release in response to the next action potential. High-frequency stimulation of a presynaptic ending can cause depletion of synaptic vesicles, resulting in depression of subsequent release for a little while. Fast-acting retrograde messengers, such as nitric oxide, can also cause short-term changes.

Figure 24-3 A few mechanisms of short-term changes in synaptic strength.

Longer-term changes can involve almost any conceivable part of presynaptic or postsynaptic elements (THB6 Figure 24-10, p. 616). One prominent example is the insertion or removal of postsynaptic transmitter receptors, resulting in long-term potentiation (LTP) or long-term depression (LTD). This can be triggered by postsynaptic Ca²⁺ entry through NMDA receptors (Fig. 24-4). NMDA receptors (named for N-methyl-d-aspartate, which binds to them) are glutamate receptors with some special properties. First, they only open when they bind glutamate and the membrane is already depolarized, making them great detectors of simultaneous activity at multiple synapses—something that could be a building block for memory formation. Second, they are less selective than other ion channels and let Ca²⁺ through (in addition to Na⁺ and K⁺). Small amounts of Ca²⁺ entry cause LTD, and larger amounts cause LTP. Subsequent Ca²⁺-initiated communication with the nucleus can make these changes very long lasting.

Figure 24-4 Induction of LTP and LTD by Ca²⁺ entry through NMDA channels. A, If only the synapse on the right is active, depolarization is not adequate to dislodge Mg²⁺ from its binding site on the NMDA receptor (1) and the channel does not open. B, Simultaneous activity at the synapse on the left dislodges the Mg²⁺ (2), allowing it to open when glutamate is bound. The Ca²⁺ that enters through the NMDA channel has both local (3) and long-range (4) effects. Local changes include insertion or removal of neurotransmitter receptors (causing LTP or LTD, respectively). Long-range effects include interactions with the nucleus, resulting in changes in the synthesis of neurotransmitter receptors and other synaptic molecules.

Modifications of synaptic strength play a big role in lifelong adjustments of cortical maps. These maps were traditionally considered to be static, but in fact the concentrated use of some body part or input (e.g., intense practice with a musical instrument) causes the corresponding part of the map to expand. Conversely, restricted use (e.g., having an arm in a cast) causes the representation to shrink. The changes in maps begin to happen within hours, much too quickly to involve growth of new connections.

Multiple Memory Systems Depend on Adjustments of Synaptic Strength

Most of us associate memory with things like learning lists of items before taking a test or writing a book, but there are actually multiple types of memory, each related to particular parts of the CNS (Fig. 24-5). Memories of facts (e.g., whose picture is on a dollar bill) and events (e.g., the particulars of the last handball game) are declarative memories, meaning you can declare them to be true. Memories of facts, also called semantic memories, span a range of categories, including historical facts, mathematical relationships, and the meanings of words; episodic memories include the specifics of events, including their timing. Nondeclarative memories, such as skills, patterns of behavior, and emotional reactions, are learned over time but show up more subconsciously.

Figure 24-5 Major categories of long-term memory and parts of the CNS prominently involved in each.

A simple view of how we form memories, especially declarative memories, is that two qualitatively different processes are involved. The first is short-term memory, a process that depends on directed attention and continuous neuronal activity. Disruption of this process, for example by temporary loss of consciousness, causes loss of all items in short-term memory. Short-term memories are gradually copied, or consolidated, into long-term memory, which probably involves permanent structural and physiological changes in brain synapses. Once something has been copied into long-term memory, it can survive lack of attention and even loss of consciousness and may persist for a lifetime.

The Hippocampus and Nearby Cortical Regions Are Critical for Declarative Memory

The hippocampus is a distinctive area of cerebral cortex folded into the temporal lobe. It is made up of the dentate gyrus and the hippocampus proper, two interlocking strips of three-layered cortex (unlike the six-layered neocortex described in Chapter 22), together with the subiculum, a transition zone between the hippocampus proper and temporal lobe neocortex (Fig. 24-6). The anterior part of the parahippocampal gyrus (entorhinal cortex) is the major interface between the hippocampus and vast areas of association cortex (see Figs. 24-7 and 24-8), allowing the hippocampus to somehow serve as a key link underlying declarative memory. Bilateral damage to the hippocampus and neighboring areas of cortex, or to the diencephalic areas they are interconnected with, causes anterograde amnesia, in which new memories for facts and events cannot be formed. The retrograde amnesia following such damage is not as severe, indicating that long-term memories mostly live outside the hippocampus.

Figure 24-6 Arrangement of cells in a cross section of the hippocampus. Information flows mostly in one direction, arriving from entorhinal cortex and passing through several sequential hippocampal synapses before reaching the subiculum. The fimbria (“fringe”) of the hippocampus is the collection of fibers that will leave (or arrive) through the fornix. DG, Dentate gyrus; HP, hippocampus proper.

Figure 24-7 Inputs to the hippocampus.

Figure 24-8 Outputs from the hippocampus.

Inputs to entorhinal cortex, and from there to the hippocampus, come from widespread unimodal, multimodal, and limbic areas (Fig. 24-7). In addition, modulatory cholinergic inputs from the septal nuclei reach the hippocampus directly by traveling “backward” through the fornix, which is a major output route from the hippocampus (Fig. 24-8). Finally, there are direct projections from the amygdala to the hippocampus. As discussed later in this chapter, the amygdala is important for marking the emotional significance of situations and events; this connection affects the probability that something will be recorded as a declarative memory, depending on our emotional reaction to it.

The hippocampus then projects back, by way of entorhinal cortex, to widespread unimodal, multimodal, and limbic areas (Fig. 24-8). Hippocampal outputs also reach limbic cortex indirectly, by way of projections to the mammillary bodies through the fornix. The fornix curves around with the lateral ventricle; it separates from the hippocampus near the splenium of the corpus callosum, travels forward along the inferior edge of the septum pellucidum, turns downward in front of the interventricular foramen, and enters the hypothalamus (THB6 Figure 24-17, p. 623). The mammillary bodies project to the anterior nucleus of the thalamus through the mammillothalamic tract, and this link forms part of a hippocampal loop known as the Papez circuit (hippocampus → mammillary body → anterior nucleus → cingulate and parahippocampal gyri → hippocampus). The relative roles of these direct and indirect outputs to limbic cortex in the formation of declarative memories are still not understood.

The Amygdala Is Centrally Involved in Emotional Memories

One form of learning critical for survival is figuring out which situations are likely to be enjoyable and which are likely to be unpleasant or dangerous. The connections of the amygdala (see Fig. 23-7Fig. 23-8) suit it well for a role in such emotional learning. The amygdala receives inputs about things that are intrinsically pleasant or unpleasant (a full stomach, a painful jab) and is hardwired to have such events cause autonomic reactions (through the hypothalamus) and conscious feelings (through projections to limbic cortex). Information from the thalamus and unimodal sensory areas reaches the same neurons in the amygdala; pairing an otherwise neutral stimulus (e.g., the sight of chocolate or a scorpion) with pleasant or unpleasant outcomes causes activation of NMDA receptors, LTP, and an emotional response to subsequent experience with the previously neutral stimuli.

The Basal Ganglia Are Important for Some Forms of Nondeclarative Memory

Both the basal ganglia and cerebellum collaborate with motor areas of cerebral cortex as we learn to move more rapidly and accurately, as in learning to play a musical instrument; this collaboration is based on the long loops from cortex → basal ganglia or cerebellum → thalamus → cortex (see Fig. 19-2Fig. 20-7). The basal ganglia, however, have a broader role in learning patterns of behavior. We routinely make decisions about how to act based on subconsciously accumulated experiences that allow “educated guesses” about probable outcomes (e.g., deciding whether to bring an umbrella on the basis of what the sky looks like in the morning). Learning to make decisions this way is based largely on interconnections between the caudate nucleus and association cortex (see Fig. 19-4).

The Cerebellum Is Important for Some Forms of Nondeclarative Memory

The cerebellum has a special role in learning how to adjust movements to fit changing circumstances. This ranges from adjusting the gain of reflexes (e.g., the flocculus changes the gain of the vestibulo-ocular reflex to compensate for wearing eyeglasses) to modifying limb movements (THB6 Figure 20-25, p. 518). The widespread interconnections of the lateral cerebellar hemispheres with nonmotor areas of cerebral cortex indicate that they probably have a broader role in learning various mental skills as well (THB6 Figure 24-24, p. 630).

The powerful excitatory synapses made by climbing fibers on Purkinje cells give the inferior olivary nucleus a particularly important role in cerebellar learning functions.

Peripheral Nerve Fibers Can Regrow After Injury

Following transection of an axon in the PNS, Schwann cells proliferate and secrete trophic factors. In response, the affected neuron ramps up its protein-synthesis machinery and starts to grow a new axon. In the process, even though RNA synthesis increases, Nissl bodies spread out and the cell seems to lose much of its basophilia (chromatolysis—“loss of color”). If the regrowing axon successfully reaches its former target, anterograde (e.g., muscle atrophy) and retrograde changes are reversed, and function is restored.

CNS Glial Cells Impede Repair after Injury

The story after transection of an axon in the CNS is different, for two main reasons. First, astrocytes and oligodendrocytes do not secrete trophic factors in response, so the affected neuron is unable to switch on growth-related genes. Second, astrocytes and oligodendrocytes actually interfere with regrowth. Both increase their production of growth-inhibiting factors that normally help stabilize connections in adult nervous systems. Astrocytes hypertrophy and wall off the injured area, forming a glial scar that blocks the progress of any regrowing axons. As a result of all this, there is usually a limited chromatolytic response and the axon fails to regrow.

Limited Numbers of New Neurons Are Added to the CNS throughout Life

The conventional wisdom for a long time was that humans are born with all the neurons they will ever have. Recent evidence, however, indicates that there are latent stem cells—cells with the potential to produce new neurons—near the walls of all the ventricles, and that there are two locations where new neurons are produced throughout life. Stem cells in the dentate gyrus produce new dentate neurons that may participate in the formation of declarative memories. Stem cells in the wall of the anterior horn of the lateral ventricle produce new inhibitory interneurons that migrate to the olfactory bulb. Coaxing latent or active stem cells to differentiate in controlled ways may make it possible some day to replace lost neurons.