Chris Henderson and Brigitte Pettmann are at the Developmental Biology Institute of Marseille (IBDM; CIIRS, INSERM) Campus de Luminy - Case 907 13288 Marseille Cedex 09, France
chris@ibdm.univ-mrs.fr
Signalling pathways involved in axonal growth are the subject of intense study. A new report further highlights the role of the extracellular signal-regulated kinase (ERK)/p35 pathway, but identifies a surprising potential trigger: the death receptor Fas. This study illustrates the importance of cellular context in the functional outcome of a given signalling mechanism.
Cellular context is critical in determining the functional outcome of signalling through a given receptor. An excellent example is provided by the multiple phenotypic outcomes after activation of so-called 'death receptors', such as tumour necrosis factor (TNF) receptor or Fas1. When bound by their ligand, these one-pass membrane proteins induce the formation of signalling complexes at the membrane. They are best known for their ability to trigger death of the cell expressing them, but they can also affect processes as different from death as cardiomyocyte hypertrophy2. The decision of a given cell type to die or proliferate after receptor activation depends on its cycling state, its degree of differentiation or on external factors. Resting T cells treated with TNF- or Fas agonists respond by proliferating, whereas cycling cells undergo apoptosis3. Naive and memory T-cells respond in opposite ways to Fas activation, the former being induced to die and the latter to proliferate4. Hepatocytes in a normal liver die rapidly after injection of agonistic Fas antibodies, but proliferate if the injection is preceded by partial hepatectomy5.
On page 118 of this issue, Desbarats and colleagues6 report an unexpected new effect of Fas activation, namely, the stimulation of axonal growth. They begin by showing that whereas activation of Fas in a T-lymphocyte cell line results, as expected, in cell death, human SH-SY5Y neuroblastoma cells are completely resistant to Fas activation, although they too express Fas on their surface. Accordingly, neuroblastoma cells treated with Fas agonists fail to activate the characteristic downstream protease caspase-8. However, they do show significant activation of Erk phosphorylation and strong induction of p35 expression. p35 is a membrane-associated protein which functions as a cofactor of Cdk5 in regulating different aspects of cell migration and growth within the developing nervous system.
The activation of this pathway in neuroblastoma cells led the authors to test the functional consequences of Fas activation in primary cultures of sensory neurons. Indeed, they show that treating sensory ganglia with Fas agonists stimulates neurite outgrowth to the same extent as does treatment with the neurotrophic factor, nerve growth factor (NGF). Importantly, the authors demonstrate that there is a direct effect on dissociated neurons in the absence of glial cells. Unlike Fas-induced cell death, the effects of Fas on neurite growth are not mediated through caspase-8. Instead, similarly to those of NGF, they involve activation of the Erk/p35 pathway.
To determine the in vivo relevance of the phenomenon, Desbarats and her colleagues6 turned to 'nerve crush' experiments in adult mice. This involves experimental interruption of both sensory and motor axons. As these axons regenerate along the original nerve sheath (containing Schwann cells, together with inflammatory and matrix components) to recontact target muscles, the mice regain coordinated motor behaviour. Strikingly, lpr mice, in which Fas levels are strongly reduced, regenerate more slowly than wild-type mice. Even more surprisingly, a single local application of agonistic anti-Fas antibodies results in significant enhancement of both morphological and functional recovery over the next three weeks.
These results were certainly unexpected for neurons, in which death has until now been the only studied function for Fas. Activation of Fas kills motor neurons and certain cortical neurons, but not other neuronal types, such as sensory neurons, and activation of Fas in vivo has been shown to exacerbate the consequences of ischaemia7. How might these different outcomes be triggered by the same ligand−receptor system? Several possibilities come to mind.
First, the same receptor may trigger different signalling pathways (Fig. 1a). Intracellular domains of transmembrane receptors can often trigger multiple distinct signalling pathways. Signalling by members of the Fas/TNF family is most often attributed to the death domain that is loosely conserved between family members. However, Desbarats et al. show that the defective Fas allele lprcg, in which a point mutation selectively affects the death domain, can still activate Erk, and hence stimulate neurite outgrowth. Similarly, the DAXX-ASK pathway, implicated in Fas-triggered death of some cells, uses part of the cytoplasmic domain outside the death domain. Hypertrophy of cardiomyocytes after Fas activation involves phosphorylation of Akt2. Thus, it is possible that Fas itself signals differently in different cellular contexts. Given that the standard Fas knockout mice still express normal levels of Fas receptor, from which only the death domain is missing, we still perhaps have more to learn about the involvement of Fas in different processes in vivo.
Figure 1. Different functional outcomes from signalling through the Fas receptor.
a, Triggering of different signalling pathways by different parts of the Fas intracellular domain. b, Sequestration of Fas by Met prevents activation. c, Modulation of Fas signalling by the cytoplasmic protein FLIP. d, Inhibition of the Fas death signal by other survival pathway.
A second possibility is that Fas may associate with other receptors (Fig. 1b). Examples of cis interactions between cell-surface receptors are becoming more and more frequent. Through such associations, the ligand of one receptor can trigger or modulate the signalling pathway of the other. This can result in changes in intensity, nature or sense of the signal transduced8,
9. An example involving Fas has already been reported: the hepatocyte growth factor (HGF) receptor Met can sequester Fas and prevent it from signalling for apoptosis. Whether other receptors can by this means result in selective activation by Ras ligand of, say, the Erk pathway remains to be established, but this does not seem unlikely.
A third model is that Fas signalling may be modulated by cytoplasmic proteins (Fig. 1c). Pivotal molecules that interact with the cytoplasmic signalling pathway downstream of a receptor can significantly affect functional outcome. After treatment with TNF-, death ensues when the TNF-receptor interactor protein (RIP) is highly expressed, whereas proliferation occurs when RIP levels are low3. Fas triggers death through the caspase-8 pathway when levels of the caspase-8 inhibitory protein FLIP are low; in contrast, Fas triggers proliferation through an Erk pathway when levels of FLIP are high10. Is a high expression level of FLIP sufficient to render cells (for example, sensory neurons) resistant to Fas-induced death? This seems to be the case for cell types expressing moderate levels of Fas. However, in cells expressing high levels of Fas, such as T cells, resistance to Fas-induced death seems to depend more on the inability of the cell to form enough death-inducing signalling complex (DISC) than on the levels of FLIP11.
Finally, the state of activation of other signalling pathways must be considered (Fig. 1d). Survival or proliferation pathways triggered by growth factors can prevent accomplishment of the cell death programme triggered by Fas. Blocking the Erk pathway with specific Erk inhibitors is sufficient to render T cells sensitive to Fas-induced death12 and Erk activation has a dominant protecting effect over apoptotic signals. However, it is probable that the overall signalling state of the cell can influence outcome in more subtle ways. For instance, transient Erk activation can cooperate with JNK activation to induce Fas death in Fas-sensitive neuroblastoma cells.
The involvement of Raf and ERK in different aspects of axonal growth is now well established13 and overexpression of a constitutively active form of Mek induces regeneration after spinal cord injury. However, activation of this pathway may not be the only explanation for some of the results of Desbarats et al. In the nerve regeneration model in vivo, it also seems possible that Fas functions as a death receptor. It may act to trigger apoptosis in injured cells and thereby clear the lesion site of injured cells. In lpr mutant mice, this clearing would take longer and therefore regeneration would be retarded, but not prevented. This hypothesis also provides a potential explanation for the efficacy of locally applied Fas agonists at the beginning of the regeneration process.
It is tempting to speculate on possible links between the results of Desbarats et al. and those in the literature concerning the crippling neurological disease ALS (amyotrophic lateral sclerosis). ALS results in rapid progressive degeneration and death of many motor neurons, but spares most sensory neurons. As the main causes of the disease are not well understood, this specificity has remained unexplained. However, in recent years, data linking the Fas receptor to ALS have appeared. Motor, but not sensory, neurons cultured from mouse models of familial ALS were shown to be remarkably more sensitive to Fas-triggered cell death than wild-type neurons14. Moreover, sera from ALS patients were reported to contain high levels of circulating antibodies to Fas15. If Fas does indeed turn out to be an aggravating factor in human patients, the different signalling responses to Fas activation in motor versus sensory neurons might help to explain the specificity of the pathological deficit.
Overall, these results, and many others over the past few years, demonstrate the need to analyse signalling through even well-studied receptors in the context of the precise cell type in which their function is being studied. For this approach, the cellular complexity of the nervous system remains a challenge, but clearly has many more surprises in store.
or Fas agonists respond by proliferating, whereas cycling cells undergo apoptosis
