- Email: [email protected]
The DnaJ-like cysteine string protein and exocytotic neurotransmitter release
REVIEW
M. Frotscher et al. – Sprouting in the hippocampus
59 Stanfield, B. and Cowan, W.M. (1979) Anat. Embryol. 156, 37–52 60 Köhler, C. (1985) Neurosci. Lett. 56, 13–19 61 Deller, T. et al. (1996) J. Neurosci. 16, 3322–3333 62 Sutula, T. et al. (1988) Science 239, 1147–1150 63 Sutula, T. et al. (1989) Ann. Neurol. 26, 321–330 64 Ramón y Cajal (1911) Histologie du Système Nerveux de l’Homme et des Vertébrés (Vol. II), Maloine 65 Steward, O. and Levy, W.B. (1982) J. Neurosci. 2, 284–291
66 Steward, O. (1995) Curr. Opin. Neurobiol. 5, 55–61 67 Del Rio, J.A. et al. (1996) J. Neurosci. 16, 6896–6907 68 Liu, Y., Fujise, N. and Kosaka, T. (1996) Exp. Brain Res. 108, 389–403 69 Ogawa, M. et al. (1995) Neuron 14, 899–912 70 D’Arcangelo, G. et al. (1995) Nature 374, 719–723 71 Del Rio, J.A. et al. (1997) Nature 385, 70–74 72 Zhou, C-F., Li, Y. and Raisman, G. (1989) Neuroscience 32, 349–362
The DnaJ-like cysteine string protein and exocytotic neurotransmitter release Erich Buchner and Cameron B. Gundersen The fast,tightly regulated release of neurotransmitters from presynaptic nerve terminals is effected by a complex molecular apparatus. The precise roles of the various proteins involved remain largely conjectural. Cysteine string proteins (CSPs) are novel synaptic vesicle components that have been conserved in evolution.They are characterized by an N-terminus ‘J’-domain and a central, multiply palmitoylated string of cysteine residues. Vertebrate CSPs have been implicated in a functional interaction of synaptic vesicles with presynaptic Ca2+ channels. Genetic ‘knockout’ of CSPs in Drosophila results in a temperature-sensitive breakdown of elicited transmitter release. Here we try to integrate these observations into speculative functional models on the role of this new protein family in synaptic vesicle exocytosis. Trends Neurosci. (1997) 20, 223–227
I
N THE NERVOUS SYSTEM information is being transmitted, processed, stored and retrieved in the form of electrical and chemical signals. The presynaptic nerve terminal represents a prominent interface between these two kinds of signals as membrane depolarization is transformed into the finely tuned release of short- and long-acting chemical messengers. Non-linear spatial interactions and temporal modifications at synaptic sites are central to virtually all aspects of nervous system function including, in particular, feature abstraction and learning and memory. It is for this reason that the detailed analysis of exocytotic neurotransmitter release is of prominent interest to neuroscientists. The gradually evolving pattern of subcellular events and protein–protein interactions associated with the various steps of the transmitter release process has been extensively reviewed1–4, but regular updates are advised as more interactions are observed and new components are described. This article reviews the information concerning a new, highly conserved protein family found in apparently all conventional chemical synapses as well as in a variety of nonneuronal tissues, and offers hypotheses concerning its possible function. Cysteine string proteins (CSPs) were discovered as synapse-associated antigens in the nervous system of Drosophila5 and, independently, in Torpedo as candidate functional components of presynaptic, voltagesensitive N-type Ca2+ channels which were expressed Copyright © 1997, Elsevier Science Ltd. All rights reserved. 0166 - 2236/97/$17.00
ectopically in frog oocytes6,7. In an effort to identify synaptic proteins in Drosophila, monoclonal antibodies that specifically recognized most or all synaptic terminals in immunohistochemical preparations8 (A. Hofbauer, Habilitationsschrift, University of Würzburg, 1991) were used for screening expression libraries of Drosophila head cDNAs. One antibody led to the characterization of three alternatively spliced transcripts coding for inferred proteins of 26.9 kDa, 24.3 kDa and 23.8 kDa calculated molecular weight (cMW), respectively5,9 [Drosophila-CSP (D-CSP1–3) in Fig. 1]. The precise mechanisms by which these three transcripts produce the four isoforms detected by anti-D-CSP antibodies in western blots of fly head homogenates9 are not known (but see below). Two features of the predicted proteins were noted: (1) the significant homology of an N-terminus region of the derived amino acid sequence with the ‘J’-domain of bacterial DnaJ protein10,11; and (2) the highly unusual string of 11 cysteines flanked by two additional pairs of cysteine residues which gave these proteins their name5 (K.E. Zinsmaier, PhD Thesis, University of Würzburg, 1990) (Fig. 1). The approach by which the first vertebrate cDNA was cloned prompted speculation for a possible molecular function of CSPs: when size-fractionated mRNA isolated from the electric organ of Torpedo californica was injected into frog oocytes, it induced the expression of an ω-conotoxin-sensitive Ca2+-channel not found in non-injected oocytes. By injecting cloned PII: S0166-2236(96)10082-5
TINS Vol. 20, No. 5, 1997
Erich Buchner is at the Lehrstuhl für Genetik, Biozentrum der Universität, 97074 Würzburg, Germany. Cameron B. Gundersen is at the Dept of Molecular and Medical Pharmacology, UCLA School of Medicine, Los Angeles, CA 90095, USA.
223
REVIEW
E. Buchner and C.B. Gundersen – Cysteine string proteins are required for synaptic function
‘J’
CS
CE-CSP D-CSP1 D-CSP2 D-CSP3 T-CSP R-CSP B-CSP1 B-CSP2 H-CSP1 H-CSP2 Fig. 1. Domain structure of presently known members of the cysteine string protein family. CE, Caenorhabditis elegans; D, Drosophila melanogaster; T, Torpedo califormica; R, rat; B, Bos; H, human. The ‘J’-domain characteristic for the DnaJ-like protein family is hatched, the cysteine string (CS) is cross-hatched. A domain of 21 amino acids present in D-CSP1 (striped) is deleted in D-CSP2 and D-CSP3. The different isoforms characterized so far in Drosophila, Bos and human are generated by alternative splicing.
antisense cRNAs derived from this same mRNA fraction, one clone was isolated whose antisense cRNA selectively inhibited Ca2+-channel expression but left co-expressed K+-channels undisturbed6,7. The inferred protein, deduced from the corresponding sense transcript, had a cMW of 21.8 kDa and displayed, in 132 out of 195 amino acids, 75% identity with the Drosophila cysteine string protein. This homology includes both the J-domain and the cysteine string, emphasizing the functional relevance of these regions throughout much of metazoan evolution. The sequence linking the J-domain and the cysteine string is also highly conserved, whereas little similarity can be observed in the region downstream of the cysteine string. More recently, cDNAs for a rat cysteine string protein have been cloned12,13. The inferred sequence of 198 amino acids (cMW 22.1 kDa) shows 84% identity to the Torpedo protein, most of the differences being due to conservative substitutions. In addition, two vertebrate CSP splice variants have been identified by homology cloning using bovine adrenal medullary chromaffin cell mRNA (Ref. 14) and human brain mRNA (Ref. 15). One isolated cDNA predicts a homolog to the known Torpedo and rat CSP, while a second predicted isoform is terminated prematurely due to an inserted exon and lacks the C-terminus 30 amino acids. Finally, as part of the Caenorhabditis elegans genome project16, a cosmid has been sequenced containing an open reading frame of a nematode CSP (Fig. 1).
Biochemical properties of CSPs Subcellular fractionation studies of Torpedo CSPs (T-CSPs) revealed that the native protein is membraneassociated while in vitro translated T-CSP behaved as soluble protein in Triton X-114 partitioning17. This suggested that CSPs are attached to membranes by post-translational modifications of the protein. Indeed, it was demonstrated that native CSP is extensively fatty acylated18, a modification that is also carried out when in vitro translated T-CSP (27 kDa) is 224
TINS Vol. 20, No. 5, 1997
injected into frog oocytes: the unmodified isoform remains in the soluble fraction while the modified 34 kDa form is found exclusively in membranes18. It was also found that after treatment with the deacylating agents hydroxylamine or methanolic KOH, the 34 kDa form of T-CSP could revert to the 27 kDa form, and it could be demonstrated that [3H]palmitic acid was specifically incorporated into the 27 kDa form of T-CSP to yield the 34 kDa form when oocytes were incubated with this labeled fatty acid. Additional experiments specified that at least 11 of the 13 cysteine residues of T-CSP are palmitoylated18. These data, coupled with immunoprecipitation results19, led to the hypothesis that CSPs are tethered to the external leaflet of synaptic vesicles via the fatty acylated cysteine string domain19. Deacylation of Drosophila membrane fractions revealed basically similar properties of D-CSPs (Refs 20,21). Unexpectedly, however, deacylation in hydroxylamine does not detach D-CSPs from membranes while treatment of membranes with 0.1 M sodium carbonate at pH 11.5 displaces substantial amounts of D-CSPs to the soluble fraction without detectably changing their electrophoretic mobilities20. These latter results have been interpreted as suggesting that D-CSPs are anchored to membranes perhaps by a second kind of post-translational modification, independent of cysteine acylation20. Whether the apparent discrepancies between data from Drosophila and Torpedo are due to intrinsic structural and functional differences between insect and vertebrate CSPs, or mainly reflect variations in experimental protocols, remains to be clarified. The question whether CSPs undergo cycles of acylation and deacylation has been investigated for vertebrates and Drosophila by pharmacological and genetic techniques, respectively. At present, there is no evidence for a change in the extent of cysteine acylation that might occur in conjunction with the life cycle of synaptic vesicles during elicited transmitter secretion20,21.
Distribution of CSPs in cells and tissues Several lines of evidence show that CSPs are components of the synaptic vesicle membrane. Thus, highly purified synaptic vesicles from Torpedo electric organ were obtained by ultracentrifugation followed by chromatography19. Cysteine string proteins and SV2 (an established synaptic vesicle component used as a control) co-distributed with occluded ACh, the neurotransmitter of the electric organ. Semiquantitative western blots show that CSPs constitute approximately 1% of the total synaptic vesicle protein, indicating that each vesicle accommodates about eight (error range, 2–20) molecules of CSP. Antibodies to both the N- and C-termini of T-CSP immunoprecipitate T-CSP, SV2 immunoreactivity and occluded ACh. These data suggest that the protein is attached to the synaptic vesicle membrane in such a way that both the N- and C-termini are accessible to antibodies from the cytoplasmic side19. Immunoprecipitation9 and sedimentation velocity fractionation of Drosophila membranes in a glycerol gradient22 also support the conclusion that CSPs are synaptic vesicle proteins. In the temperature-sensitive paralytic Drosophila mutant shibire ts1, synaptic vesicle endocytosis is blocked at elevated temperatures resulting in total depletion of synaptic vesicles. The concomitant
REVIEW
E. Buchner and C.B. Gundersen – Cysteine string proteins are required for synaptic function
translocation of CSPs (as well as other vesicle proteins) from vesicular membrane fractions of the glycerol gradient to fractions containing plasma membrane proteins again indicates that CSPs are synaptic vesicle proteins22. Finally, subcellular fractionation of rat brain reveals that CSPs partition with synaptosome and synaptic vesicle fractions, further supporting their association with synaptic vesicles12,13. In non-neural cells CSPs have so far been localized subcellularly to membranes of bovine chromaffin granules, the vesicles for release of catecholamines from adrenal medulla23, and to membranes of rat zymogen granules, exocrine secretory vesicles of pancreatic acinar cells13. Tissue distribution of CSP protein isoforms and mRNAs has been investigated both in Drosophila and in vertebrates. In flies all synaptic terminals apparently contain CSPs as shown by the binding of two anti-D-CSP monoclonal antibodies to all known chemical synapses5,9. Correspondingly, csp mRNAs are detected in all neurons of the adult fly5. Two of the four known D-CSP isoforms are specifically recognized by a third antibody9. These isoforms are expressed at high levels in photoreceptor cells and in selected regions of the CNS, but are found only at low abundance in the remaining CNS and in motoneurons9 (K.K. Eberle, unpublished observations). Outside the adult nervous system high concentrations of CSPs are detected in a cell layer of the cardia, in the follicle cells of the ovaria, and in the testes (E. Buchner, unpublished observations). During development, weak D-CSP immunoreactivity is already found in neuroblasts of six-hour old embryos, and high levels are observed in all neuropil regions as soon as these are discernible (H. Dürr, unpublished observations). Specificity of all immunochemical reactions described has been rigorously tested by the use of csp knockout flies (see below). In vertebrates, purified antibodies against the C-terminus undecapeptide of T-CSP crossreact with a rat CSP isoform of 35 kDa (Ref. 17), but they do not recognize the truncated 26 kDa isoform14 (cf. Fig. 1). These antibodies strongly bind to neuromuscular junctions and to most or all neuropil regions in the brain. Brain areas lacking synapses, such as white matter and vasculature, are essentially devoid of immunoreactivity. Particularly strong immunostaining by these antibodies is observed in many welldefined brain areas24. Whether this selective abundance distribution is specific for the 35 kDa isoform or reflects regional variations in the general expression level of the rat csp gene is unclear. In addition, these antibodies detect prominent CSP immunoreactivity in rat chromaffin cells, while the steroid-secreting adrenal cortical cells do not stain specifically24. Recently, anti-CSP antibodies have also been used to demonstrate CSP immunoreactivity in human blood samples15. Moreover, mRNA analysis of various rat tissues and western blots developed with a different antiserum against full-length recombinant rat CSP led to the proposal of a widespread expression of the csp gene in vertebrates14. This view is further supported by the detection of csp mRNAs in all human tissues examined15.
Molecular interactions and function of CSPs To date, direct evidence for the identity of proteins interacting with CSPs has not been published. The
cloning strategy that had led to the isolation of a vertebrate cDNA suggested a functional interaction of CSPs with presynaptic voltage-sensitive N-type Ca2+ channels6,7, but no further experimental support for this inference exists. Efforts biochemically to demonstrate an interaction of CSPs with P/Q-type Ca2+ channels were negative25, and at present comparable data for v-conotoxin-sensitive N-type channels are unavailable. However, a challenge for these experiments is that any CSP–Ca2+-channel interaction is likely to be sub-stoichiometric and transient, making its detection very difficult19. Important insight into the cellular role of CSPs has been obtained by a genetic approach. By targeted mutagenesis, the csp gene of Drosophila has been deleted. Null mutants display a phenotype that is characterized by slow development, short adult life span with progressing neurological symptoms such as sluggishness, spasmic jumping, shaking and uncoordinated locomotion, and a temperature-sensitive blockage of synaptic transmission, resulting in reversible paralysis at elevated temperatures9 (K.K. Eberle, unpublished observations). Detailed electrophysiological analysis of neuromuscular transmission in csp null mutants has shown that the defect impairs the coupling of presynaptic membrane depolarization to neurotransmitter release26. Spontaneous quantal release, as measured by miniature excitatory junction potentials (EJPs) (Fig. 2A), axonal propagation of action potentials and responses of the muscles to externally applied glutamate, are not affected by csp mutation. At room temperature, quantal content of elicited EJPs in the mutants is reduced to about 50% while the apparent Ca2+ co-operativity for exocytosis is not affected: when the datapoints of EJP quantal content measured at various external Ca2+ concentrations are fitted by a linear function in a log–log diagram, the line is shifted to the right for the mutant compared to the wild type, but its slope is unaffected (Fig. 2B). At 30°C, however, EJPs in the mutants gradually decline over a period of several minutes until they fail completely. Return to room temperature restores EJPs (Fig. 2C). While such results are compatible with a defect in Ca2+-channel opening in csp mutants, they do not exclude other explanations. All mutant defects can be ‘rescued’ by transforming the mutants with the wild-type csp gene9,26.
Models of CSP function in neurotransmitter release We discuss here two hypothetical models concerning the involvement of CSPs in synaptic function. One scheme emphasizes the J-domain by adapting current understanding of DnaJ-like proteins10,11, the other scheme concentrates on the second prominent feature of CSPs, that is, their remarkable, cysteine-rich ‘cysteine string’ domain. These two classes of models should be considered complementary rather than mutually exclusive, and their goal is to stimulate the design of new experiments. The first model assigns to CSPs a role as a molecular chaperone in the presently envisaged scheme of a protein–protein interaction cascade for Ca2+-triggered exocytosis4. Recent co-expression experiments27 demonstrate an inactivation of presynaptic Ca2+ channels by the presynaptic membrane protein syntaxin 1A. We19 and others27 speculate that inactivation TINS Vol. 20, No. 5, 1997
225
REVIEW
E. Buchner and C.B. Gundersen – Cysteine string proteins are required for synaptic function
Fig. 2. Excitatory junction potentials in controls and cysteine string protein null mutants. (A) Intracellular recordings of excitatory junction potentials (EJPs; main traces) in larval body wall muscles elicited by stimulation of the corresponding motoneurons, and spontaneous miniature EJPs (MEJPs; inset traces). Recordings are from Drosophila wild type, from a functional csp null mutant (cspX1), and from this mutant after germ line transformation with the wild-type gene (rescue). At high temperatures (30°C) EJPs are abolished in the mutant (small spike represents stimulus artefact), while MEJPs remain essentially unchanged. After return to room temperature, mutant EJPs recover. Temperature shifts have little effects in the wild type and rescue preparations. (B) Quantal content (area of EJP divided by area of meansized MEJP) as a function of extracellular Ca2+ concentrations [Ca2+]o for wild type (crosses), two csp mutants (cspX1, open circles; cspR1, open triangles), and rescue transformants (closed circles). (C) Timecourse of EJP breakdown and recovery in cspX1 mutant. Numbered arrows in temperature trace indicate time points at which the response of the muscle to a single elicited action potential in the motoneuron was recorded (lower traces). The data demonstrate the temperature-dependent breakdown of depolarization secretion coupling at the presynaptic nerve terminals of mutants lacking cysteine string proteins. Modified, with permission, from Ref. 26. Acknowledgements This work was supported by travel grant Bu566/8-1 to EB from the DFG. The kind and stimulating hospitality of the members of the Dept of Molecular and Medical Pharmacology of UCLA during the sabbatical visit of EB is gratefully acknowledged.
226
of Ca2+ channels might be relieved when a synaptic vesicle docks near a Ca2+ channel. Such a mechanism would allow Ca2+ entry only at sites where it can efficiently trigger exocytosis. From the properties of CSPs and the csp mutant phenotype it is tempting to assume that the vesicle-mediated disinhibition is effected by CSPs, either directly or via an as yet unidentified interacting partner. The J-domains of previously analysed members of the DnaJ-like protein family interact with an isoform of the heat shock protein family, HSP70, and regulate their intrinsic ATPase activity. In a variety of organisms different DnaJlike–HSP70-like protein pairs participate as molecular chaperones in diverse intracellular events10,11. Cysteine string proteins together with their postulated HSP70 chaperone partners might thus help to dissociTINS Vol. 20, No. 5, 1997
ate syntaxin (or some other protein) from presynaptic Ca2+ channels and relieve their inactivation. However, clearly it is necessary to have more direct data before elaborating further on this hypothesis. The second model dwells upon the amphipathic nature of CSPs resulting from the hydrophobic fatty acylated cysteine-string domain which is flanked by highly polar stretches of amino acids. Here a role for CSPs in the biophysical process of membrane fusion has been proposed28. These proposals blend in a speculative manner knowledge of CSP structure with constraints imposed by the known features of exocytotic fusion. In these models28, CSPs are hypothesized to operate at the synaptic vesicle interface with the plasma membrane and catalyse membrane fusion28. Again, the participation of CSPs in this process remains highly conjectural, and future investigations are important to clarify the role(s) of CSPs. While present subcellular data identify CSPs on vesicular membranes as participating in regulated exocytosis, their detection in many non-secretory tissues, both in Drosophila and in vertebrates, suggests that this new family of proteins might have a more general function. They could co-operate with a molecular chaperone in a rather ubiquitous biochemical cascade whose function might vary in different cell types, being specified by components in the cascade downstream of CSPs. In the synapse this downstream target of CSP might be a syntaxin–Ca2+-channel complex. Alternatively, CSPs could participate in a general mechanism to overcome energy barriers in membrane fusion, which must proceed in all cells but is of particular relevance in regulated exocytosis. The principal value of schemes such as those outlined above lies in the clarification of the main open questions about the function of CSPs: What kind of post-translational modification is responsible for membrane association in Drosophila CSPs if it is not the cysteine residue acylation? What are the molecular partners that presumably interact with CSPs via the J-domain and, perhaps, other highly conserved domains? With what kind of subcellular structures are CSPs associated in non-secretory tissues? In the synapse, is there a direct interaction of CSPs (or their J-partner) with Ca2+ channels? Can conformational changes in CSPs be detected that might occur during high Ca2+ transients? Can additional components of the functional cascade that involves CSPs be identified? These and related questions are presently being pursued in several laboratories using biochemical, biophysical and genetic techniques. Selected references 1 Bark, I.C. and Wilson, M.C. (1994) Proc. Natl. Acad. Sci. U. S. A. 91, 4621–4624 2 Scheller, R.H. (1995) Neuron 14, 893–897 3 Schweizer, F.E., Betz, H. and Augustine, G.J. (1995) Neuron 14, 689–696 4 Südhof, T.C. (1995) Nature 375, 645–653 5 Zinsmaier, K.E. et al. (1990) J. Neurogenet. 7, 15–29 6 Umbach, J.A. and Gundersen, C.B. (1991) Ann. New York Acad. Sci. 635, 443–446 7 Gundersen, C.B. and Umbach, J.A. (1992) Neuron 9, 527–537 8 Buchner, E. et al. (1988) Cell Tissue Res. 253, 357–370 9 Zinsmaier, K.E. et al. (1994) Science 263, 977–980 10 Silver, P.A. and Way, J.C. (1993) Cell 74, 5–7 11 Cyr, D.M., Langer, T. and Douglas, M.G. (1994) Trends Biochem. Sci. 19, 176–181 12 Mastrogiacomo, A. and Gundersen, C.B. (1995) Mol. Brain Res. 28, 12–18 13 Braun, J.E.A. and Scheller, R.H. (1995) J. Mol. Pharmacol. 34,
REVIEW
E. Buchner and C.B. Gundersen – Cysteine string proteins are required for synaptic function
1361–1369 14 Chamberlain, L.H. and Burgoyne, R.D. (1996) J. Biol. Chem. 271, 7320–7323 15 Coppola, T. and Gundersen, C.B. (1996) FEBS Lett. 391, 269–272 16 Wilson, R. et al. (1994) Nature 368, 32–38 17 Mastrogiacomo, A., Evans, C.J. and Gundersen, C.B. (1994) J. Neurochem. 62, 873–880 18 Gundersen, C.B. et al. (1994) J. Biol. Chem. 269, 19197–19199 19 Mastrogiacomo, A. et al. (1994) Science 263, 981–982 20 van de Goor, J. and Kelly, R.B. (1996) FEBS Lett. 380, 251–256 21 Gundersen, C.B., Umbach, J.A. and Mastrogiacomo, A. (1996)
Life Sci. 58, 2037–2040 22 van de Goor, J., Ramaswami, M. and Kelly, R. (1995) Proc. Natl. Acad. Sci. U. S. A. 92, 5739–5743 23 Chamberlain, L.H., Henry, J. and Burgoyne, R.D. (1996) J. Biol. Chem. 271, 19514–19517 24 Kohan, S.A. et al. (1995) J. Neurosci. 15, 6230–6238 25 Martin-Moutot, M. et al. (1996) J. Biol. Chem. 271, 6567–6570 26 Umbach, J.A. et al. (1994) Neuron 13, 899–907 27 Bezprozvanny, I., Scheller, R.H. and Tsien, R.W. (1995) Nature 378, 623–626 28 Gundersen, C.B., Mastrogiacomo, A. and Umbach, J.A. (1995) J. Theor. Biol. 172, 269–277
The c-Jun transcription factor – bipotential mediator of neuronal death, survival and regeneration Thomas Herdegen, Pate Skene and Mathias Bähr Axon interruption elicits a complex neuronal response that leaves neurons poised precariously between death and regeneration. The signals underlying this dichotomy are not fully understood. The transcription factor c-Jun is one of the earliest and most consistent markers for neurons that respond to nerve-fiber transection, and its expression can be related to both degeneration and survival including target re-innervation. In vitro experiments have demonstrated that expression of c-Jun can kill neonatal neurons but, in the adult nervous system, c-Jun might also be involved in neuroprotection and regeneration. The functional characteristics of c-Jun offer a model for the ability of a single molecule to serve as pivotal regulator for death or survival, not only in the response of the cell body to axonal lesions but also following neurodegenerative disorders. In this model, the fate of neurons is determined by a novel transcriptional network comprising c-Jun, ATF-2 (activating transcription factor-2) and JNKs (c-Jun N-terminal kinases). Trends Neurosci. (1997) 20, 227–231
I
NTERRUPTION OF MATURE AXONS activates a cascade of events in neuron cell bodies that can lead to apoptotic cell death or, paradoxically, vigorous regeneration of the injured axon1. Correlative evidence suggests that the signaling pathways leading to these contrasting responses might be closely linked. For example, in mammalian CNS neurons axotomy close to the neuron cell body leads to an earlier onset and greater extent of cell death than distal axotomy2–4. Neurons cut proximally, however, are far more effective in regenerating their axons through a favorable growth environment, such as a grafted segment of a peripheral nerve, than those cut distally. On the other hand, CNS neurons lesioned far from their cell bodies rarely regenerate even with access to supportive environments, but simultaneously they are somehow protected from cell death2,5,6. Other conditions that provoke extensive cell death in the adult CNS, such as transient ischemia and seizure activity, also elicit axonal sprouting. These observations suggest that intraneuronal messengers activated by axotomy do not lead directly to the degeneration of CNS neurons, but instead overlap in their initial stages and prepare these cells to activate either programmed cell death or Copyright © 1997, Elsevier Science Ltd. All rights reserved. 0166 - 2236/97/$17.00
regeneration in response to subsequent signals, such as activation of microglia7 or transynaptic input8. How can a common initial signaling pathway lead to the apparently opposed responses of cell death and axon regeneration? And what determines which of these responses predominate? Recent studies on the transcriptional regulator c-Jun, which responds to axon injury and neurodegenerative disorders in adult neurons, provide a model for such conditional signaling by a single molecule (Fig. 1).
c-Jun as a common, early and lasting response to axotomy Cellular responses to environmental change are often mediated by inducible transcription factors, which serve to co-ordinate the activation and repression of downstream genes to carry out an adaptive alteration in cell behavior. Among a variety of inducible transcription factors examined in axotomized neurons, only the proto-oncogene-encoded c-Jun protein and its family member JunD are expressed consistently following axon interruption. Moreover, c-Jun is the predominant transcription factor known so far that is also expressed following other S0166-2236(96)01000-4
TINS Vol. 20, No. 5, 1997
Thomas Herdegen is at the Dept of Pharmacology, University of Kiel, Hospitalstrasse 4, 24105 Kiel, Germany. Pate Skene is at the Dept of Neurobiology, Duke University, Durham, NC 27710, USA. Mathias Bähr is at the Dept of Neurology, University of Tübingen, HoppeSeyler-Strasse 3, 72076 Tübingen, Germany.
227
M. Frotscher et al. – Sprouting in the hippocampus
59 Stanfield, B. and Cowan, W.M. (1979) Anat. Embryol. 156, 37–52 60 Köhler, C. (1985) Neurosci. Lett. 56, 13–19 61 Deller, T. et al. (1996) J. Neurosci. 16, 3322–3333 62 Sutula, T. et al. (1988) Science 239, 1147–1150 63 Sutula, T. et al. (1989) Ann. Neurol. 26, 321–330 64 Ramón y Cajal (1911) Histologie du Système Nerveux de l’Homme et des Vertébrés (Vol. II), Maloine 65 Steward, O. and Levy, W.B. (1982) J. Neurosci. 2, 284–291
66 Steward, O. (1995) Curr. Opin. Neurobiol. 5, 55–61 67 Del Rio, J.A. et al. (1996) J. Neurosci. 16, 6896–6907 68 Liu, Y., Fujise, N. and Kosaka, T. (1996) Exp. Brain Res. 108, 389–403 69 Ogawa, M. et al. (1995) Neuron 14, 899–912 70 D’Arcangelo, G. et al. (1995) Nature 374, 719–723 71 Del Rio, J.A. et al. (1997) Nature 385, 70–74 72 Zhou, C-F., Li, Y. and Raisman, G. (1989) Neuroscience 32, 349–362
The DnaJ-like cysteine string protein and exocytotic neurotransmitter release Erich Buchner and Cameron B. Gundersen The fast,tightly regulated release of neurotransmitters from presynaptic nerve terminals is effected by a complex molecular apparatus. The precise roles of the various proteins involved remain largely conjectural. Cysteine string proteins (CSPs) are novel synaptic vesicle components that have been conserved in evolution.They are characterized by an N-terminus ‘J’-domain and a central, multiply palmitoylated string of cysteine residues. Vertebrate CSPs have been implicated in a functional interaction of synaptic vesicles with presynaptic Ca2+ channels. Genetic ‘knockout’ of CSPs in Drosophila results in a temperature-sensitive breakdown of elicited transmitter release. Here we try to integrate these observations into speculative functional models on the role of this new protein family in synaptic vesicle exocytosis. Trends Neurosci. (1997) 20, 223–227
I
N THE NERVOUS SYSTEM information is being transmitted, processed, stored and retrieved in the form of electrical and chemical signals. The presynaptic nerve terminal represents a prominent interface between these two kinds of signals as membrane depolarization is transformed into the finely tuned release of short- and long-acting chemical messengers. Non-linear spatial interactions and temporal modifications at synaptic sites are central to virtually all aspects of nervous system function including, in particular, feature abstraction and learning and memory. It is for this reason that the detailed analysis of exocytotic neurotransmitter release is of prominent interest to neuroscientists. The gradually evolving pattern of subcellular events and protein–protein interactions associated with the various steps of the transmitter release process has been extensively reviewed1–4, but regular updates are advised as more interactions are observed and new components are described. This article reviews the information concerning a new, highly conserved protein family found in apparently all conventional chemical synapses as well as in a variety of nonneuronal tissues, and offers hypotheses concerning its possible function. Cysteine string proteins (CSPs) were discovered as synapse-associated antigens in the nervous system of Drosophila5 and, independently, in Torpedo as candidate functional components of presynaptic, voltagesensitive N-type Ca2+ channels which were expressed Copyright © 1997, Elsevier Science Ltd. All rights reserved. 0166 - 2236/97/$17.00
ectopically in frog oocytes6,7. In an effort to identify synaptic proteins in Drosophila, monoclonal antibodies that specifically recognized most or all synaptic terminals in immunohistochemical preparations8 (A. Hofbauer, Habilitationsschrift, University of Würzburg, 1991) were used for screening expression libraries of Drosophila head cDNAs. One antibody led to the characterization of three alternatively spliced transcripts coding for inferred proteins of 26.9 kDa, 24.3 kDa and 23.8 kDa calculated molecular weight (cMW), respectively5,9 [Drosophila-CSP (D-CSP1–3) in Fig. 1]. The precise mechanisms by which these three transcripts produce the four isoforms detected by anti-D-CSP antibodies in western blots of fly head homogenates9 are not known (but see below). Two features of the predicted proteins were noted: (1) the significant homology of an N-terminus region of the derived amino acid sequence with the ‘J’-domain of bacterial DnaJ protein10,11; and (2) the highly unusual string of 11 cysteines flanked by two additional pairs of cysteine residues which gave these proteins their name5 (K.E. Zinsmaier, PhD Thesis, University of Würzburg, 1990) (Fig. 1). The approach by which the first vertebrate cDNA was cloned prompted speculation for a possible molecular function of CSPs: when size-fractionated mRNA isolated from the electric organ of Torpedo californica was injected into frog oocytes, it induced the expression of an ω-conotoxin-sensitive Ca2+-channel not found in non-injected oocytes. By injecting cloned PII: S0166-2236(96)10082-5
TINS Vol. 20, No. 5, 1997
Erich Buchner is at the Lehrstuhl für Genetik, Biozentrum der Universität, 97074 Würzburg, Germany. Cameron B. Gundersen is at the Dept of Molecular and Medical Pharmacology, UCLA School of Medicine, Los Angeles, CA 90095, USA.
223
REVIEW
E. Buchner and C.B. Gundersen – Cysteine string proteins are required for synaptic function
‘J’
CS
CE-CSP D-CSP1 D-CSP2 D-CSP3 T-CSP R-CSP B-CSP1 B-CSP2 H-CSP1 H-CSP2 Fig. 1. Domain structure of presently known members of the cysteine string protein family. CE, Caenorhabditis elegans; D, Drosophila melanogaster; T, Torpedo califormica; R, rat; B, Bos; H, human. The ‘J’-domain characteristic for the DnaJ-like protein family is hatched, the cysteine string (CS) is cross-hatched. A domain of 21 amino acids present in D-CSP1 (striped) is deleted in D-CSP2 and D-CSP3. The different isoforms characterized so far in Drosophila, Bos and human are generated by alternative splicing.
antisense cRNAs derived from this same mRNA fraction, one clone was isolated whose antisense cRNA selectively inhibited Ca2+-channel expression but left co-expressed K+-channels undisturbed6,7. The inferred protein, deduced from the corresponding sense transcript, had a cMW of 21.8 kDa and displayed, in 132 out of 195 amino acids, 75% identity with the Drosophila cysteine string protein. This homology includes both the J-domain and the cysteine string, emphasizing the functional relevance of these regions throughout much of metazoan evolution. The sequence linking the J-domain and the cysteine string is also highly conserved, whereas little similarity can be observed in the region downstream of the cysteine string. More recently, cDNAs for a rat cysteine string protein have been cloned12,13. The inferred sequence of 198 amino acids (cMW 22.1 kDa) shows 84% identity to the Torpedo protein, most of the differences being due to conservative substitutions. In addition, two vertebrate CSP splice variants have been identified by homology cloning using bovine adrenal medullary chromaffin cell mRNA (Ref. 14) and human brain mRNA (Ref. 15). One isolated cDNA predicts a homolog to the known Torpedo and rat CSP, while a second predicted isoform is terminated prematurely due to an inserted exon and lacks the C-terminus 30 amino acids. Finally, as part of the Caenorhabditis elegans genome project16, a cosmid has been sequenced containing an open reading frame of a nematode CSP (Fig. 1).
Biochemical properties of CSPs Subcellular fractionation studies of Torpedo CSPs (T-CSPs) revealed that the native protein is membraneassociated while in vitro translated T-CSP behaved as soluble protein in Triton X-114 partitioning17. This suggested that CSPs are attached to membranes by post-translational modifications of the protein. Indeed, it was demonstrated that native CSP is extensively fatty acylated18, a modification that is also carried out when in vitro translated T-CSP (27 kDa) is 224
TINS Vol. 20, No. 5, 1997
injected into frog oocytes: the unmodified isoform remains in the soluble fraction while the modified 34 kDa form is found exclusively in membranes18. It was also found that after treatment with the deacylating agents hydroxylamine or methanolic KOH, the 34 kDa form of T-CSP could revert to the 27 kDa form, and it could be demonstrated that [3H]palmitic acid was specifically incorporated into the 27 kDa form of T-CSP to yield the 34 kDa form when oocytes were incubated with this labeled fatty acid. Additional experiments specified that at least 11 of the 13 cysteine residues of T-CSP are palmitoylated18. These data, coupled with immunoprecipitation results19, led to the hypothesis that CSPs are tethered to the external leaflet of synaptic vesicles via the fatty acylated cysteine string domain19. Deacylation of Drosophila membrane fractions revealed basically similar properties of D-CSPs (Refs 20,21). Unexpectedly, however, deacylation in hydroxylamine does not detach D-CSPs from membranes while treatment of membranes with 0.1 M sodium carbonate at pH 11.5 displaces substantial amounts of D-CSPs to the soluble fraction without detectably changing their electrophoretic mobilities20. These latter results have been interpreted as suggesting that D-CSPs are anchored to membranes perhaps by a second kind of post-translational modification, independent of cysteine acylation20. Whether the apparent discrepancies between data from Drosophila and Torpedo are due to intrinsic structural and functional differences between insect and vertebrate CSPs, or mainly reflect variations in experimental protocols, remains to be clarified. The question whether CSPs undergo cycles of acylation and deacylation has been investigated for vertebrates and Drosophila by pharmacological and genetic techniques, respectively. At present, there is no evidence for a change in the extent of cysteine acylation that might occur in conjunction with the life cycle of synaptic vesicles during elicited transmitter secretion20,21.
Distribution of CSPs in cells and tissues Several lines of evidence show that CSPs are components of the synaptic vesicle membrane. Thus, highly purified synaptic vesicles from Torpedo electric organ were obtained by ultracentrifugation followed by chromatography19. Cysteine string proteins and SV2 (an established synaptic vesicle component used as a control) co-distributed with occluded ACh, the neurotransmitter of the electric organ. Semiquantitative western blots show that CSPs constitute approximately 1% of the total synaptic vesicle protein, indicating that each vesicle accommodates about eight (error range, 2–20) molecules of CSP. Antibodies to both the N- and C-termini of T-CSP immunoprecipitate T-CSP, SV2 immunoreactivity and occluded ACh. These data suggest that the protein is attached to the synaptic vesicle membrane in such a way that both the N- and C-termini are accessible to antibodies from the cytoplasmic side19. Immunoprecipitation9 and sedimentation velocity fractionation of Drosophila membranes in a glycerol gradient22 also support the conclusion that CSPs are synaptic vesicle proteins. In the temperature-sensitive paralytic Drosophila mutant shibire ts1, synaptic vesicle endocytosis is blocked at elevated temperatures resulting in total depletion of synaptic vesicles. The concomitant
REVIEW
E. Buchner and C.B. Gundersen – Cysteine string proteins are required for synaptic function
translocation of CSPs (as well as other vesicle proteins) from vesicular membrane fractions of the glycerol gradient to fractions containing plasma membrane proteins again indicates that CSPs are synaptic vesicle proteins22. Finally, subcellular fractionation of rat brain reveals that CSPs partition with synaptosome and synaptic vesicle fractions, further supporting their association with synaptic vesicles12,13. In non-neural cells CSPs have so far been localized subcellularly to membranes of bovine chromaffin granules, the vesicles for release of catecholamines from adrenal medulla23, and to membranes of rat zymogen granules, exocrine secretory vesicles of pancreatic acinar cells13. Tissue distribution of CSP protein isoforms and mRNAs has been investigated both in Drosophila and in vertebrates. In flies all synaptic terminals apparently contain CSPs as shown by the binding of two anti-D-CSP monoclonal antibodies to all known chemical synapses5,9. Correspondingly, csp mRNAs are detected in all neurons of the adult fly5. Two of the four known D-CSP isoforms are specifically recognized by a third antibody9. These isoforms are expressed at high levels in photoreceptor cells and in selected regions of the CNS, but are found only at low abundance in the remaining CNS and in motoneurons9 (K.K. Eberle, unpublished observations). Outside the adult nervous system high concentrations of CSPs are detected in a cell layer of the cardia, in the follicle cells of the ovaria, and in the testes (E. Buchner, unpublished observations). During development, weak D-CSP immunoreactivity is already found in neuroblasts of six-hour old embryos, and high levels are observed in all neuropil regions as soon as these are discernible (H. Dürr, unpublished observations). Specificity of all immunochemical reactions described has been rigorously tested by the use of csp knockout flies (see below). In vertebrates, purified antibodies against the C-terminus undecapeptide of T-CSP crossreact with a rat CSP isoform of 35 kDa (Ref. 17), but they do not recognize the truncated 26 kDa isoform14 (cf. Fig. 1). These antibodies strongly bind to neuromuscular junctions and to most or all neuropil regions in the brain. Brain areas lacking synapses, such as white matter and vasculature, are essentially devoid of immunoreactivity. Particularly strong immunostaining by these antibodies is observed in many welldefined brain areas24. Whether this selective abundance distribution is specific for the 35 kDa isoform or reflects regional variations in the general expression level of the rat csp gene is unclear. In addition, these antibodies detect prominent CSP immunoreactivity in rat chromaffin cells, while the steroid-secreting adrenal cortical cells do not stain specifically24. Recently, anti-CSP antibodies have also been used to demonstrate CSP immunoreactivity in human blood samples15. Moreover, mRNA analysis of various rat tissues and western blots developed with a different antiserum against full-length recombinant rat CSP led to the proposal of a widespread expression of the csp gene in vertebrates14. This view is further supported by the detection of csp mRNAs in all human tissues examined15.
Molecular interactions and function of CSPs To date, direct evidence for the identity of proteins interacting with CSPs has not been published. The
cloning strategy that had led to the isolation of a vertebrate cDNA suggested a functional interaction of CSPs with presynaptic voltage-sensitive N-type Ca2+ channels6,7, but no further experimental support for this inference exists. Efforts biochemically to demonstrate an interaction of CSPs with P/Q-type Ca2+ channels were negative25, and at present comparable data for v-conotoxin-sensitive N-type channels are unavailable. However, a challenge for these experiments is that any CSP–Ca2+-channel interaction is likely to be sub-stoichiometric and transient, making its detection very difficult19. Important insight into the cellular role of CSPs has been obtained by a genetic approach. By targeted mutagenesis, the csp gene of Drosophila has been deleted. Null mutants display a phenotype that is characterized by slow development, short adult life span with progressing neurological symptoms such as sluggishness, spasmic jumping, shaking and uncoordinated locomotion, and a temperature-sensitive blockage of synaptic transmission, resulting in reversible paralysis at elevated temperatures9 (K.K. Eberle, unpublished observations). Detailed electrophysiological analysis of neuromuscular transmission in csp null mutants has shown that the defect impairs the coupling of presynaptic membrane depolarization to neurotransmitter release26. Spontaneous quantal release, as measured by miniature excitatory junction potentials (EJPs) (Fig. 2A), axonal propagation of action potentials and responses of the muscles to externally applied glutamate, are not affected by csp mutation. At room temperature, quantal content of elicited EJPs in the mutants is reduced to about 50% while the apparent Ca2+ co-operativity for exocytosis is not affected: when the datapoints of EJP quantal content measured at various external Ca2+ concentrations are fitted by a linear function in a log–log diagram, the line is shifted to the right for the mutant compared to the wild type, but its slope is unaffected (Fig. 2B). At 30°C, however, EJPs in the mutants gradually decline over a period of several minutes until they fail completely. Return to room temperature restores EJPs (Fig. 2C). While such results are compatible with a defect in Ca2+-channel opening in csp mutants, they do not exclude other explanations. All mutant defects can be ‘rescued’ by transforming the mutants with the wild-type csp gene9,26.
Models of CSP function in neurotransmitter release We discuss here two hypothetical models concerning the involvement of CSPs in synaptic function. One scheme emphasizes the J-domain by adapting current understanding of DnaJ-like proteins10,11, the other scheme concentrates on the second prominent feature of CSPs, that is, their remarkable, cysteine-rich ‘cysteine string’ domain. These two classes of models should be considered complementary rather than mutually exclusive, and their goal is to stimulate the design of new experiments. The first model assigns to CSPs a role as a molecular chaperone in the presently envisaged scheme of a protein–protein interaction cascade for Ca2+-triggered exocytosis4. Recent co-expression experiments27 demonstrate an inactivation of presynaptic Ca2+ channels by the presynaptic membrane protein syntaxin 1A. We19 and others27 speculate that inactivation TINS Vol. 20, No. 5, 1997
225
REVIEW
E. Buchner and C.B. Gundersen – Cysteine string proteins are required for synaptic function
Fig. 2. Excitatory junction potentials in controls and cysteine string protein null mutants. (A) Intracellular recordings of excitatory junction potentials (EJPs; main traces) in larval body wall muscles elicited by stimulation of the corresponding motoneurons, and spontaneous miniature EJPs (MEJPs; inset traces). Recordings are from Drosophila wild type, from a functional csp null mutant (cspX1), and from this mutant after germ line transformation with the wild-type gene (rescue). At high temperatures (30°C) EJPs are abolished in the mutant (small spike represents stimulus artefact), while MEJPs remain essentially unchanged. After return to room temperature, mutant EJPs recover. Temperature shifts have little effects in the wild type and rescue preparations. (B) Quantal content (area of EJP divided by area of meansized MEJP) as a function of extracellular Ca2+ concentrations [Ca2+]o for wild type (crosses), two csp mutants (cspX1, open circles; cspR1, open triangles), and rescue transformants (closed circles). (C) Timecourse of EJP breakdown and recovery in cspX1 mutant. Numbered arrows in temperature trace indicate time points at which the response of the muscle to a single elicited action potential in the motoneuron was recorded (lower traces). The data demonstrate the temperature-dependent breakdown of depolarization secretion coupling at the presynaptic nerve terminals of mutants lacking cysteine string proteins. Modified, with permission, from Ref. 26. Acknowledgements This work was supported by travel grant Bu566/8-1 to EB from the DFG. The kind and stimulating hospitality of the members of the Dept of Molecular and Medical Pharmacology of UCLA during the sabbatical visit of EB is gratefully acknowledged.
226
of Ca2+ channels might be relieved when a synaptic vesicle docks near a Ca2+ channel. Such a mechanism would allow Ca2+ entry only at sites where it can efficiently trigger exocytosis. From the properties of CSPs and the csp mutant phenotype it is tempting to assume that the vesicle-mediated disinhibition is effected by CSPs, either directly or via an as yet unidentified interacting partner. The J-domains of previously analysed members of the DnaJ-like protein family interact with an isoform of the heat shock protein family, HSP70, and regulate their intrinsic ATPase activity. In a variety of organisms different DnaJlike–HSP70-like protein pairs participate as molecular chaperones in diverse intracellular events10,11. Cysteine string proteins together with their postulated HSP70 chaperone partners might thus help to dissociTINS Vol. 20, No. 5, 1997
ate syntaxin (or some other protein) from presynaptic Ca2+ channels and relieve their inactivation. However, clearly it is necessary to have more direct data before elaborating further on this hypothesis. The second model dwells upon the amphipathic nature of CSPs resulting from the hydrophobic fatty acylated cysteine-string domain which is flanked by highly polar stretches of amino acids. Here a role for CSPs in the biophysical process of membrane fusion has been proposed28. These proposals blend in a speculative manner knowledge of CSP structure with constraints imposed by the known features of exocytotic fusion. In these models28, CSPs are hypothesized to operate at the synaptic vesicle interface with the plasma membrane and catalyse membrane fusion28. Again, the participation of CSPs in this process remains highly conjectural, and future investigations are important to clarify the role(s) of CSPs. While present subcellular data identify CSPs on vesicular membranes as participating in regulated exocytosis, their detection in many non-secretory tissues, both in Drosophila and in vertebrates, suggests that this new family of proteins might have a more general function. They could co-operate with a molecular chaperone in a rather ubiquitous biochemical cascade whose function might vary in different cell types, being specified by components in the cascade downstream of CSPs. In the synapse this downstream target of CSP might be a syntaxin–Ca2+-channel complex. Alternatively, CSPs could participate in a general mechanism to overcome energy barriers in membrane fusion, which must proceed in all cells but is of particular relevance in regulated exocytosis. The principal value of schemes such as those outlined above lies in the clarification of the main open questions about the function of CSPs: What kind of post-translational modification is responsible for membrane association in Drosophila CSPs if it is not the cysteine residue acylation? What are the molecular partners that presumably interact with CSPs via the J-domain and, perhaps, other highly conserved domains? With what kind of subcellular structures are CSPs associated in non-secretory tissues? In the synapse, is there a direct interaction of CSPs (or their J-partner) with Ca2+ channels? Can conformational changes in CSPs be detected that might occur during high Ca2+ transients? Can additional components of the functional cascade that involves CSPs be identified? These and related questions are presently being pursued in several laboratories using biochemical, biophysical and genetic techniques. Selected references 1 Bark, I.C. and Wilson, M.C. (1994) Proc. Natl. Acad. Sci. U. S. A. 91, 4621–4624 2 Scheller, R.H. (1995) Neuron 14, 893–897 3 Schweizer, F.E., Betz, H. and Augustine, G.J. (1995) Neuron 14, 689–696 4 Südhof, T.C. (1995) Nature 375, 645–653 5 Zinsmaier, K.E. et al. (1990) J. Neurogenet. 7, 15–29 6 Umbach, J.A. and Gundersen, C.B. (1991) Ann. New York Acad. Sci. 635, 443–446 7 Gundersen, C.B. and Umbach, J.A. (1992) Neuron 9, 527–537 8 Buchner, E. et al. (1988) Cell Tissue Res. 253, 357–370 9 Zinsmaier, K.E. et al. (1994) Science 263, 977–980 10 Silver, P.A. and Way, J.C. (1993) Cell 74, 5–7 11 Cyr, D.M., Langer, T. and Douglas, M.G. (1994) Trends Biochem. Sci. 19, 176–181 12 Mastrogiacomo, A. and Gundersen, C.B. (1995) Mol. Brain Res. 28, 12–18 13 Braun, J.E.A. and Scheller, R.H. (1995) J. Mol. Pharmacol. 34,
REVIEW
E. Buchner and C.B. Gundersen – Cysteine string proteins are required for synaptic function
1361–1369 14 Chamberlain, L.H. and Burgoyne, R.D. (1996) J. Biol. Chem. 271, 7320–7323 15 Coppola, T. and Gundersen, C.B. (1996) FEBS Lett. 391, 269–272 16 Wilson, R. et al. (1994) Nature 368, 32–38 17 Mastrogiacomo, A., Evans, C.J. and Gundersen, C.B. (1994) J. Neurochem. 62, 873–880 18 Gundersen, C.B. et al. (1994) J. Biol. Chem. 269, 19197–19199 19 Mastrogiacomo, A. et al. (1994) Science 263, 981–982 20 van de Goor, J. and Kelly, R.B. (1996) FEBS Lett. 380, 251–256 21 Gundersen, C.B., Umbach, J.A. and Mastrogiacomo, A. (1996)
Life Sci. 58, 2037–2040 22 van de Goor, J., Ramaswami, M. and Kelly, R. (1995) Proc. Natl. Acad. Sci. U. S. A. 92, 5739–5743 23 Chamberlain, L.H., Henry, J. and Burgoyne, R.D. (1996) J. Biol. Chem. 271, 19514–19517 24 Kohan, S.A. et al. (1995) J. Neurosci. 15, 6230–6238 25 Martin-Moutot, M. et al. (1996) J. Biol. Chem. 271, 6567–6570 26 Umbach, J.A. et al. (1994) Neuron 13, 899–907 27 Bezprozvanny, I., Scheller, R.H. and Tsien, R.W. (1995) Nature 378, 623–626 28 Gundersen, C.B., Mastrogiacomo, A. and Umbach, J.A. (1995) J. Theor. Biol. 172, 269–277
The c-Jun transcription factor – bipotential mediator of neuronal death, survival and regeneration Thomas Herdegen, Pate Skene and Mathias Bähr Axon interruption elicits a complex neuronal response that leaves neurons poised precariously between death and regeneration. The signals underlying this dichotomy are not fully understood. The transcription factor c-Jun is one of the earliest and most consistent markers for neurons that respond to nerve-fiber transection, and its expression can be related to both degeneration and survival including target re-innervation. In vitro experiments have demonstrated that expression of c-Jun can kill neonatal neurons but, in the adult nervous system, c-Jun might also be involved in neuroprotection and regeneration. The functional characteristics of c-Jun offer a model for the ability of a single molecule to serve as pivotal regulator for death or survival, not only in the response of the cell body to axonal lesions but also following neurodegenerative disorders. In this model, the fate of neurons is determined by a novel transcriptional network comprising c-Jun, ATF-2 (activating transcription factor-2) and JNKs (c-Jun N-terminal kinases). Trends Neurosci. (1997) 20, 227–231
I
NTERRUPTION OF MATURE AXONS activates a cascade of events in neuron cell bodies that can lead to apoptotic cell death or, paradoxically, vigorous regeneration of the injured axon1. Correlative evidence suggests that the signaling pathways leading to these contrasting responses might be closely linked. For example, in mammalian CNS neurons axotomy close to the neuron cell body leads to an earlier onset and greater extent of cell death than distal axotomy2–4. Neurons cut proximally, however, are far more effective in regenerating their axons through a favorable growth environment, such as a grafted segment of a peripheral nerve, than those cut distally. On the other hand, CNS neurons lesioned far from their cell bodies rarely regenerate even with access to supportive environments, but simultaneously they are somehow protected from cell death2,5,6. Other conditions that provoke extensive cell death in the adult CNS, such as transient ischemia and seizure activity, also elicit axonal sprouting. These observations suggest that intraneuronal messengers activated by axotomy do not lead directly to the degeneration of CNS neurons, but instead overlap in their initial stages and prepare these cells to activate either programmed cell death or Copyright © 1997, Elsevier Science Ltd. All rights reserved. 0166 - 2236/97/$17.00
regeneration in response to subsequent signals, such as activation of microglia7 or transynaptic input8. How can a common initial signaling pathway lead to the apparently opposed responses of cell death and axon regeneration? And what determines which of these responses predominate? Recent studies on the transcriptional regulator c-Jun, which responds to axon injury and neurodegenerative disorders in adult neurons, provide a model for such conditional signaling by a single molecule (Fig. 1).
c-Jun as a common, early and lasting response to axotomy Cellular responses to environmental change are often mediated by inducible transcription factors, which serve to co-ordinate the activation and repression of downstream genes to carry out an adaptive alteration in cell behavior. Among a variety of inducible transcription factors examined in axotomized neurons, only the proto-oncogene-encoded c-Jun protein and its family member JunD are expressed consistently following axon interruption. Moreover, c-Jun is the predominant transcription factor known so far that is also expressed following other S0166-2236(96)01000-4
TINS Vol. 20, No. 5, 1997
Thomas Herdegen is at the Dept of Pharmacology, University of Kiel, Hospitalstrasse 4, 24105 Kiel, Germany. Pate Skene is at the Dept of Neurobiology, Duke University, Durham, NC 27710, USA. Mathias Bähr is at the Dept of Neurology, University of Tübingen, HoppeSeyler-Strasse 3, 72076 Tübingen, Germany.
227