1

Φ Abstract

Neuronal specificity in the developing CNS is obtained as a result of interaction between the axon growth cones and diffusible molecules. Whereas Connectins and Fasciclins generally provide chemoattractive cues, Semaphorins and Slit-Robo provide repulsive guidance to growth cones, while in vertebrates; Netrins provide both attractive and repulsive cues. Semaphorins also assist in pruning of mismatched connections. Together, these diffusible protein molecules result in a fine-tuning of the neuronal circuitry.

______

Φ Introduction

The fully formed adult human CNS contains about 1015 neuronal cells. This number, although awe-inspiring, is elevated to sheer wonder when one considers the fact that each neuron, once it has reached its niche and has become established in the environment where it will spend its life, will make extremely precise connections with many other neurons and muscle fibers. These connections are extremely specific and through experiments conducted in the past two decades have been shown to perturb normal functionality in the CNS, even if one interlocking runs astray. This magnificent neuronal specificity has been a hot topic of research for many years in Developmental Biology. However, it is only in the past decade or so that modern molecular techniques have brought this area of research out of the realms of abstruse philosophy, to the much more tangible benches of science.

Neuronal cells find their origin in the ectoderm, which also mothers epidermal cells. Once, a cell has taken the decision to become a neuron, instead of an epidermal cell, the cell’s neuronal fate must be decided. The next decision-making process deals with whether the cell will become a motor neuron, an interneuron, or a sensory neuron. If it decides to become an interneuron, will it be a commissural neuron (one that crosses the ventral mid-line in the CNS)? All these decisions are controlled by proteins and in general, a differential gene expression. The concerns of this term paper stem next. Now that the neuron has taken its decision to become, say for instance, a commissural neuron, how will it reach its axons to its target? It is not uncommon for the neuron cell body to be far away from the target neuron or muscle fiber, forcing the axon to travel huge distances, through a variety of sub-strata offering different chemical environments. How then does the neuron obtain specificity in its connections? In this term-paper I attempt to answer this last question. In general, neurons exploit the variance in extra-cellular matrix chemicals to establish gradients, which they then follow. Sort of like sniffing bull-hounds!

______

1.0.Axon- Growth Cones

What are axon growth cones? In the developing Nervous system, axons are still in the process of reaching out to their destined targets. However, due to the likelihood of their encounter with a host of sub-strata where they are not “supposed to be”, they must scan all surfaces they come in contact with before they establish a connection. This scanning is done by a specialized structure at the axon tip, called the growth cone. The axon growth cone is a transient structure, made of microfibers, that recognizes diffusible chemicals as a direct consequence of carrying the receptors for these proteins.

2.0. Pioneer Neurons and Differential Gene Expression as mechanisms for Growth Cone Guidance.

Harrison (1) suggested that the specificity of axonal growth is due to pioneer nerve fibers, which go ahead of other axons and serve as guides for them. Harrison also noted, however, that axons must grow on a solid substrate, and he speculated that differences among embryonic surfaces might allow axons to travel in certain specified directions. The final connections would occur by complementary interactions on the target cell surface. In 1993, Goodman and Shatz (2) hypothesized that the specificity of axonal connections unfolds in three steps:

  1. Pathway selection, wherein the axons travel along a route that leads them to a particular region of the embryo.
  2. Target selection, wherein the axons, once they reach the correct area, recognize and bind to a set of cells with which they may form stable connections.
  3. Address selection, wherein the initial patterns are refined such that each axon binds to a small subset (sometimes only one) of its possible targets.

The initial pathway a growth cone follows is determined by the environment the growth cone experiences. The extracellular environment can provide substrates upon which to migrate, and these substrates can provide navigational information to the growth cone. Some of the substrates the growth cone encounters will permit it to adhere to them, and thus promote axon migration. Other substrates will cause the growth cone to retract, and will not allow its axon to grow in that direction. Some of the substrates can give extremely specific cues, recognized by only a small set of neuronal growth cones, while other cues are recognized by large sets of neurons. The migrational cues of the substrate can come from:

  1. Anatomical structures,
  2. The extracellular matrix, or from
  3. Adjacent cell surfaces.

Extracellular matrix cues often involve gradients of adhesivity. Growth cones prefer to migrate on surfaces that are more adhesive than their surroundings, and a gradient of increasingly adhesive molecules can direct them to their respective targets. This phenomenon of migrating to a target on adhesive gradients is called haptotaxis, (3, 4). Letourneau and co-workers (5) have shown that the axons of certain spinal neurons travel through the neuroepithelium over a transient laminin-coated surface that precisely marks their path. Similarly, there is a very good correlation between the elongation of retinal axons and the presence of laminin on the neuroepithelial cells and astrocytes in the embryonic mouse brain (6, 7). However since laminin may be distributes just about anywhere in the sub-stratum, specific interlocking of neurons is probably guided by other, more elaborate mechanisms. Is it possible that different neurons (and hence their axons) express different proteins, by a temporal and spatial regulation of the expression of different genes?

2.1. Differential Gene Expression as a Mechanism for Growth Cone Guidance.

The targets of motor neurons are specified before their axons extend into the periphery. This was shown by Lance-Jones and Landmesser (8), who reversed segments of the chick spinal cord so that the motor neurons were in new locations. The axons went to their original targets, not to the ones expected from their new positions. The molecular basis for this specificity resides in the members of the LIM family of proteins that are induced during neuronal migration. For instance, all motor neurons express Islet-1 and (slightly later) Islet-2. If no other LIM protein is expressed, the neurons project to the ventral body wall muscles. Thus, each group of neurons is characterized by a particular constellation of LIM transcription factors.

However true that may be it only delineates a path that the emerging axon can follow. What are the actual guidance clues that allow target selection? The most plausible answer seems to be a gradient of diffusible molecules. The existence of such a gradient will certainly not be unique to axonal-guidance. Biology offers many instances, where chemical gradients guide movements and biological processes, fertilzation, particularly in aquatic organisms; movement of Paramecia in negative environments, etc. Can such a phenomenon account for neuronal growth-cone guidance?

3. 0. Chemical Cues for Axonal Guidance

3. 1. Netrins

When viewed across phyla from insects to vertebrates, the CNS mid-line exhibits both attractive and repulsive properties for neuronal growth cones. In vertebrates, the ventral midline contains a specialized group of cells (the floor plate) that generate both attractive and repulsive axonal-guidance cues; while in Drosophila midline glia act to attract commissural growth cones while simultaneously presenting a repulsive boundary to those axons that are not supposed to cross.

3. 1. 1. Discovery

Santiago Ramón y Cajal was the first person to express the idea that specific chemical species provide axon-guidance clues in the developing nervous system. Working with commissural neurons, which are spinal-interneurons and are located dorsally, Cajal hypothesized that specific, diffusible molecules might be responsible for guiding the axonal projections of these neurons across the ventral mid-line. The discovery and identification of these “diffusible molecules” had to wait till the 1990s. It was not until 1994, when Serafini and colleagues developed an assay that would allow them to screen for such a diffusible molecule that Netrin-1 and subsequently, Netrin-2, were identified. In fact, it was later discovered that both Netrins 1 and 2 had a good amount of sequence homology to another protein first discovered in the nematode C. elegans, unc-6. Mutations in the UNC-6 gene were shown to lead to defects in cell migration and axon guidance. unc-6 expression is ventrally restricted and provides a hierarchy of guidance cues throughout the ectoderm. Such ventral restriction of netrin expression likely extends to all phyla, (9).

In the rat Netrin-1 is expressed in the ventral midline of the neural tube. Floor plate cells attract ventrally directed spinal commissural axons cells but have a long-range repulsive effect on dorsally directed trochlear motor axons. Both repulsion and attraction are mediated by Netrin-1, (10).

3. 1. 2. How do they work?

3. 1. 2. 1. Experimental Evidence for Netrin dependent Chemotaxis.

Kennedy et al (11) showed that Netrins resulted in chemotaxis, wherein axonal outgrowths followed gradients of diffusible molecules. They transformed chick fibroblast cells (which usually do not make these proteins) into netrin-producing cells using a vector containing an active netrin gene. Aggregates of these netrin-secreting fibroblast cells showed commissural axon outgrowth from dorsal rat spinal cord explants, while control cells that were given the vector without the active netrin gene did not elicit such activity. It is possible that the commissural neurons first encounter a gradient of netrin-2, which brings them into the domain of the steeper netrin-1 gradient.

Fig. 01. Transformed COS cells secreting netrins elicit commissural axon outgrowth from 11-day embryonic rat dorsal spinal cord explants. Adapted from (11).

In Drosophila Netrin genes are expressed at the ventral midline of the central nervous system. The first axons to pioneer the anterior and posterior commissures first project directly toward the midline glia and subsequently make intimate contact with them. In addition to this, Netrins also influence the peripheral projections of motor axons to their target muscles. Netrins are expressed by discrete subsets of muscles. In Netrin double mutants, axons wander over more territory than usual, and can sometimes inappropriately cross the segmental border into neighboring segments. Neurons often appear to branch inappropriately over muscle targets, to stall, to project past targets or to project into adjacent segments, (12).

3. 1. 2. 2. Spinal Sub-localization of Netrin and development of Gradients

In fact on further dissection of the effects of individual Netrins, it was observed that there exists sub-localization within the developing spinal cord, in regards to areas of Netrin 1 and 2 secretions. Netrin-1 is made by and secreted from the floor plate cells, whereas netrin-2 is synthesized in the lower region of the spinal cord, but not in the floor plate. The following diagram shows the spinal cord with areas where Netrin-1 and 2 are secreted. It is possible that the commissural neurons first encounter a gradient of netrin-2, which brings them into the domain of the steeper netrin-1 gradient.

Fig. 02. Sub-localization of Netrins in the Spinal Cord. Adapted from (30).

3. 1. 2. Attraction and repulsion: duality in Netrin clues

It must be borne in mind that in vertebrates, netrin-1 may serve as both an attractive and a repulsive signal. Colamarino and Tessier-Lavigne (10) showed that in the trochlear (fourth cranial) nerve, those axons of the trochlear nerve that originate near the floor plate of the brain stem and migrate dorsally away from the floor plate region can be prevented from being projected by placing floor plate cells or transgenic, netrin-1-secreting chick fibroblast cells within 450 μm of the dorsal portion of the explant. This dorsal outgrowth is not prevented by dorsal explants of the neural tube or by chick fibroblast cells that do not contain the active netrin-1 gene. They therefore concluded that netrins appear to be chemotactic to some neurons and chemorepulsive to others. To site another example the growth cones of Xenopus retinal neurons are attracted to netrin-1 and are guided to the optic nerve head by this diffusible factor. Once there, however, the combination of netrin-1 and laminin prevents the axons from departing from the optic nerve. Hopker et al (13) suggested that the laminin of the extracellular matrix surrounding the optic nerve converts the netrin from being an attractive molecule to being a repulsive one.

3. 1. 3. The Netrin Receptor

In the nematode C. elegans tow receptors for Netrin are known. The UNC-5 gene encodes a membrane protein of the immunoglobulin family. Misexpression experiments have shown than UNC-5 protein is necessary and sufficient for dorsally oriented cell movements that utilize the UNC-6 protein gradient. UNC-40, also called Frazzled, is another immunoglobulin superfamily member that is implicated as a receptor component for UNC-5, Harris (14).

3. 1. 4. Molecular Mechanisms of Netrin Growth-Cone Guidance.

The following signal-transduction pathway for Netrin dependent growth cone guidance seems the most likely. The pathway has not been worked out in its entirety and in fact I have collated it from a reading of various references. Apparently, the unc-5 receptor once activated by Netrins leads to activation of the L-type Calcium ion channels. The activity of the L-type channels is directly modulated by cyclic nucleotide levels. cGMP signaling, activated by an arachidonate-lipoxygenase pathway suppresses the activity of L-type channels triggered by Netrin-1. Nishiyama et al recently showed that the ratio of cyclic AMP to cyclic GMP controls polarity of Netrin dependent growth cone guidance. High AMP/GMP ratios favor attraction, whereas low ratios lead to repulsion. Hence, in the above case scenario, high GMP levels will lead to growth cone repulsion (15).

3. 1. 5. Axonal Guidance to Muscles: another feather in Netrin’s cap

In addition to their CNS midline expression and function in axon guidance the Netrins are also expressed by distinct subsets of muscles where they function as short-range target recognition molecules. Genetic analysis suggests that both types of Netrin-mediated attractive responses (i.e., pathfinding and targeting) require Frazzled, the UNC-40-like Netrin receptor. Even though they are expressed by distinct subsets of muscles and function as target recognition molecules, in Drosophila the two netrins, NetA and NetB do not act alone in specifying any one of these muscle targets. NetB is expressed by muscles 7 and 6, but NetB is not the sole attractant used to innervate these muscles. In the absence of NetB, in 35% of segments RP3 in Drosophila make the correct pathfinding decisions in the periphery but fail to innervate muscles 7 and 6 properly. However, in the other 65% of segments it does innervate muscles 7 and 6.

3. 2. Semaphorins

3. 2. 1. Axonal Growth Cone Repulsion: Semaphorins

The semaphorin proteins guide growth cones by selective repulsion. These transmembrane proteins are especially important in making “turns” where an axon does not grow in a straight line, but must change direction. Semaphorin I (also known as fasciclin IV) is one such protein that is expressed in a band of epithelial cells in the developing insect limb. This protein appears to inhibit the growth cones of the Ti1 sensory neurons from moving forward, causing them to turn. In Drosophila, semaphorin II is secreted by a single large thoracic muscle, preventing the muscle from being innervated by inappropriate axons (16). Semaphorin III, found in mammals and birds, is also known as collapsin (17) because it can collapse the growth cones of axons originating in the dorsal root ganglia.

Fig. 03. The action of semaphorin I in the developing grasshopper limb. Adapted from (30).

3. 2. 2. Uniqueness in Axonal-guidance to Muscles: another role for Semaphorins

How unique is innervation to one muscle? Muscle 33, found in the third thoracic segment (Thoracic 3, or T3) of Drosophila is a quintessential example of such specificity. A brief description of the anatomy of Drosophila musculature follows: each side of larval abdominal segments A2 to A7 contains 30 muscles. The ventral muscles in the first abdominal segment (A1) are almost identical to those in A2 to A7; an exception is the addition of Muscle 31. The A1 segment muscles are also missing two muscles otherwise found in A2 to A7. Moving forward one segment to Thoracic 3, one finds larval Muscle 33 stretching across the other thoracic segments to attach to the mouthparts. It is Muscle 33 that expresses a protein, Semaphorin II, not expressed in any other muscle cell in the developing fly. All these internal ventral muscles are innervated by branches of segmental nerve b, including muscle 33 in segment T3.

While mutations in Sema-2a are lethal, it is noteworthy that no defects are found in muscle 33 development, nor in the innervation of muscle 33. What then could be the function of Sema-2a, expressed solely in muscle 33? To answer this question Sema-2a was expressed ectopically in a different subset of ventral embryonic muscles during motoneuron pathfinding. No gross defects are apparent in muscles of Toll-Sema-2a embryos, suggesting that ectopic Sema II does not alter the differentiation of ventral muscles. However, there were abnormalities in the development of certain branches of the SNb motor nerve that normally innervate muscles 7 and 6 in all abdominal segments as well as muscle 31 in A1. The effects of ectopic Sema II expression are specific to motoneuron SNb, while other motor nerves show normal morphology and branching. SNb carries motorneurons in the RP cluster (RP1, RP3, RP4 and RP5). The innervation of muscles 7 and 6 by RP3 motoneuron is dramatically altered in Toll-Sema-2a embryos. Instead of extending towards muscles 7 and 6, the neuron stalls just external to muscles 7 and 6. Abnormalities are also found in another motoneuron, DC1, which innervates muscle 13 and normally follows the same trajectory as the RP1 motoneuron innervating muscle 13. In most Toll-Sema-2a mutant embryos, DC1 fails to innervate muscle 31. In contrast, motoneuron RP1 is apparently unaffected by Toll-Sema-2a overexpression.