The Ephs and Ephrins Hardwire the Nervous System

Signaling between Ephs and ephrins is a major component in determining how neurons wire together during development. This signaling is particularly important in the subfield of axon guidance, where the action of Ephs and ephrins has been heavily studied. The exaggerated cytoskeletal structures of neural cells, which feature extensive dendritic or axonal protrusions that extend many times the length of the cell body, terminate in easily defined structures called "growth cones" while migrating. Due to their large size and accessibility, these dynamic growth cone structures are particularly easy for investigators to study both in vivo and in cell culture systems. Accordingly, examples of the Ephs and ephrins in different aspects of neural development work well to illustrate how they function as guidance molecules during development as well as the range of responses that can come with Eph-ephrin signaling.

How does Eph-ephrin signaling influence cell organization during development? A great example comes from the neural wiring of the corticospinal tract (CST), which is a pathway that sends messages from the brain to moving limbs. This pathway depends on the crossing of neurons from one side of the central nervous system (CNS) to the other. Evidence of this anatomical crossing of the CST comes from stroke victims, whose hemorrhage on one side of the brain typically causes weakness or paralysis on the opposite side of the body. The cause is that axons extending from neurons in the motor cortex of one hemisphere of the brain typically connect to targets on the opposite side of the spinal column. Therefore, the left side of the brain controls limb movement on our right side, and the right side of the brain controls limb movement on our left side. During development, axons must grow a great distance — from the motor cortex in the brain to the spinal cord — to complete this connection. When axons of corticospinal neurons from the motor cortex cross the middle of the hindbrain, they plunge down the opposite side of the spinal cord and connect to motor neurons that control movement. In normal development, a metaphorical gate closes along the midline of the brain such that after the axons cross, they cannot go back to the side they came from. This gate closing is a fundamental point in time in development, and it helps keep the wiring of the CNS organized and specific. What does this crossover of the CST neurons mean for motor function? The ultimate result of proper CST wiring is the ability to move asymmetrically, meaning that the right and left feet and arms can move in opposite directions at the same time.

What controls how the CST crosses the midline correctly and ultimately grows to the correct target in the spinal cord? Genetic studies have shown that the signaling between Ephs and ephrins controls the proper migration of developing CST neurons. Two groups of investigators, one at the Queensland Institute for Medical Research and the other at the University of Texas Southwestern Medical Center, created transgenic mice with specific germline mutations in the Eph-ephrin signaling system. The Queensland Institute group created the mice with mutations in EphA4, and the University of Texas group created mice with mutant ephrin-B3. Both of these mutant strains showed striking phenotypes: a hopping locomotion, with their back two feet always moving in the same direction at the same time (Dottori et al. 1998; Yokoyama et al. 2001) (Figure 1). Knowing what they did about the crossover wiring in the CST, the investigators then decided to follow the path of migrating CST axons during development to see how they may have wired incorrectly. By tracing CST axons with special dyes, the investigators found that these axons were actually contacting motor neurons on both sides of the spinal column; in other words, they were not staying strictly on the opposite side. This erroneous wiring is the basis for the mutant mice's loss of ability to separate movement in their hind feet. Furthermore, when the investigators analyzed the expression patterns of Eph and ephrin proteins on specific neurons in the CST, they found that EphA4 is expressed on the surface of the migrating CST axons, whereas ephrin-B3 is expressed at the surface of cells along the spinal cord midline. What did this finding mean? The evidence suggested that the signaling between EphA4 and ephrin-B3 at the midline somehow restricts the migrating axons to stay on only the opposite side of the nervous system after they have crossed. Through follow-up studies, the investigators found that ephrin-B3 acts as a barrier, by which the ephrin ligand activates EphA4 forward signaling in migrating CST axons, with instructions to stay away from the midline after they had already crossed it. In the absence of either EphA4 or ephrin-B3, as with the mutant mice mentioned above (Figure 1), these migrating CST axons lose this critical spatial cue and erroneously migrate across the midline to innervate motoneurons on both sides of the spinal cord.

Although CST development gives a more typical example of how Eph-ephrin signaling is put to use during development, the signaling during the development of the visual system in vertebrates exemplifies the diversity that these molecules can have. Consider the challenges during eye development: All axons from the retina must properly connect to a precise spatial pattern in the brain, carrying the spatial information of the visual field captured by the eye. The process of making the correct connections starts when all retinal ganglion cells (RGCs) that carry this information from the retina to the brain must precisely funnel into the narrow optic disc in the middle of the retina to form the optic nerve. These RGC axons then navigate toward the midline of the brain at the optic chiasm, where some axons cross the midline and some do not. The proper migration of RGC axons thus contains three critical decision points: (1) Within the retina, projecting RGC axons must find and navigate through the optic disc to exit the eye and travel into the optic nerve; (2) at the optic chiasm, where the optic nerves from the two eyes connect, RGC axons must either continue migrating across the midline to their intended target in the brain or stay on the same side and travel to same-side targets in the brain; and (3) RGC axons must project directly from the retina toward other areas of the brain, such as the superior colliculus. The process of RGC pathfinding toward the superior colliculus is extremely well orchestrated, with RGC axons originating from specific fields within the retina terminating in specific zones within the superior colliculus in a highly reproducible fashion. As it turns out, researchers have shown that Eph-ephrin signaling is a major player in properly navigating these axons through all three of these RGC pathways.

Within the retina, scientists first analyzed the expression patterns of EphB receptors and ephrin-B molecules with a combination of mRNA in situ studies, lacZ reporter genes, and fluorescently labeled antibodies. All these techniques labeled the Eph and ephrin proteins or their mRNA within the retina. Interestingly, they found that EphB receptors are expressed in a gradient pattern (low dorsal to high ventral) and that ephrin-B molecules are also expressed in a gradient pattern, but with the opposite pattern (low ventral to high dorsal). These complementary expression patterns led investigators to examine the retinas of mice lacking EphB (EphB null) to see if there were defects in RGC axon pathfinding to the optic disc. They did so by tracing RGC axons from different parts of the retina with a lipophilic fluorescent dye called DiI. Using this technique, investigators from David Sretavan's laboratory at the University of California, San Francisco, found that, in the retinas of mice with null mutations for both EphB2 and EphB3, RGC axons originating from the dorsal retina (which expresses ephrin-B) did not properly funnel into the optic disc, but instead erroneously wandered to other areas of the ventral retina (Birgbauer et al. 2000). Somehow, these axons lost their way and overshot the optic disc (Figure 2). Based on the expression data of gradient patterns and the results in the null mice, the investigators proposed a scenario for the proper signaling to RGC axons: Navigating dorsal RGC axons (expressing ephrin-B) rely on the gradient of EphB receptors, and the interaction of these ephrin-B-containing axons with EphB receptors must activate reverse signaling, which normally provides a repulsive cue that prevents these axons from overshooting the optic disc.

How could these investigators test this hypothesis? One way was to examine a particular kind of mutant mouse wherein forward signaling through the EphB receptor is knocked out. Mark Henkemeyer had created such a mouse strain, called EphB2^lacZ. In this mouse, the cytoplasmic domain of EphB2 (containing all the intracellular signaling structures) was deleted and replaced with conjugated b-galactosidase, a reporter gene. If the hypothesis about EphB receptors mediating repulsive cues through reverse signaling was correct about reverse signaling, then mice with the EphB2^lacZ mutation should not show any defects in RGC pathfinding to the optic disc because they are only lacking forward signaling function. When the researchers repeated these DiI axon tracing experiments in EphB2^lacZ mutant animals, they indeed found that RGC pathfinding was normal. From these observations, they were able to conclude that proper guidance of dorsal RGCs into the optic disc is controlled by EphB activation of reverse signaling through ephrin-B receptors on RGC axons, with a cell-axon repulsive outcome.

What about RGC axons continuing into the optic chiasm? Like humans, mice have binocular vision, meaning that both eyes perceive the central portion of the visual field. To permit the accurate processing of visual signals, RGCs covering this binocular field within the retinas from both eyes should therefore come together at the same target area in the brain. To ensure that this joining occurs, a subset of RGC axons extending from the ventrotemporal region of the optic disc segregates from the rest of the RGCs at the optic chiasm and instead projects with RGCs from the retina and joins those from the opposite eye. The first suggestion that Eph-ephrin signaling was involved in this process of wiring correctly for binocular vision came from expression studies that compared different animals. These studies showed that ephrin-B2 was expressed in the optic chiasm of mammals (which have binocular vision), but not in the chiasm of fish or chickens (which do not have binocular vision) (Nakagawa et al. 2000), which suggested that ephrin-B2 was involved in pathfinding cues for binocular vision. But how was it involved? In 2003, the role of ephrin-B2 in this pathfinding process was clarified through collaborative research between the laboratories of Carol Mason at the Columbia University Medical Center and Henkemeyer at the University of Texas Southwestern Medical Center. These investigators found that the EphB1 receptor is expressed exclusively in the ventrotemporal region of the retina and that EphB1 null mice show dramatically reduced binocular projections (Williams et al. 2003). What did this finding mean? It indicated to the researchers that ephrin-B2, the ligand for these receptors, both serves as a midline barrier to activate forward signaling in EphB1-positive cells, which repels ventrotemporal RGC axons from the optic chiasm, and correctly directs them to the proper side of the brain to complete the pathway for binocular vision. In the absence of EphB1, these ventrotemporal axons are unable to detect this midline barrier and do not join this critical binocular pathway (Figure 3). This signaling mechanism is different from that in the optic disc, although both do involve repulsion events. Whereas RGC guidance into the optic disc was mediated by repulsive ephrin-B reverse signaling, at the optic chiasm it is EphB forward signaling that provides the critical repulsive instruction.

A third major axon pathfinding event in visual pathway development is the connection of RGCs to the midbrain structure called the superior colliculus. Here RGC axons must migrate to certain subregions of the superior colliculus to properly and effectively relay information about the visual field coming from the retina. Once again, DiI labeling of RGC axons helped investigators trace the projections of RGCs from the retina all the way into the superior colliculus. With this technique, investigators from Dennis O'Leary's laboratory at the Salk Institute found that mice with null mutations for EphB receptors show errors in the pattern of RGC connections to the superior colliculus (Hindges et al. 2002) (Figure 4). What kind of EphB signaling is involved in this area of the brain? Similar to the studies described above, these investigators turned to the mutant mice whose Eph receptors had only one kind of directional signaling. After examining mice with mutations in EphB2 designed to specifically remove only forward signaling through the receptor, they also found these same defects in RGC connections to the superior colliculus. They then concluded that forward signaling through the EphB receptors must be particularly important for the proper wiring of this pathway. Interestingly, the ventral RGCs, which express the highest level of EphB receptors, connect to areas of the superior colliculus that also express the highest levels of ephrin-B. To researchers, this observation suggests that proper RGC axon migration in the superior colliculus is directed through ephrin-B activation of EphB forward signaling to produce a cell attraction response at these sites of cell-axon contact.

Clearly, there is incredible variety among Eph and ephrin molecule signaling events in the axon pathfinding supporting visual development. In the retina, repulsive ephrin-B reverse signaling drives RGCs into the optic disc. At the optic chiasm, repulsive EphB forward signaling segregates ventrotemporal RGCs to the correct side of the brain for setting up binocular vision. Finally, in the superior colliculus, attractive EphB forward signaling guides RGCs to their proper termination sites.

Why are scientists so interested in how neural networks are wired? Obviously, axon guidance is a major developmental phenomenon, and knowing how it works means that we may be able to fix defects in axonal connections resulting from injury or disease. For example, spinal cord injuries often result in paralysis because axons are severed. Regrowing axons to the proper targets to heal these spinal cord wounds is theoretically possible, if the proper signals for guiding precise connectivity within the spinal cord were known. In fact, much of the research on Eph-ephrin signaling has potential relevance for developing therapeutics. When function of sensory or locomotion systems does not work properly — due to neurodegeneration in diseases such as diabetes or even due to just normal aging — this dysfunction can severely affect the quality of life. The more scientists are able to understand the molecules involved in guiding these neural structures to their intact forms, the closer we come to better orchestrating some control over repairing these same neural structures when they become damaged.