- Email: [email protected]
Is p21-ras a real G protein?
research
news
R
as, like other oncogenes, was first described as the gene in a particular RNA tumour viral genome which was essential for the transforming activity of the virus. Slightly different versions of ras were discovered in two strains [Harvey (Ha) and Kirsten (Ki)] of rat sarcoma virus, and the ability of these viruses to induce tumours and transform cells in culture was found to be due to expression of the viral ras gene products inside the cell1'2. These gene products are closely related, 21 kDa (p21) proteins. Subsequent to the description of these and other viral v-ras oncogenes and their gene products, they were observed in many types of tumour (for review see Ref. 3). Like many other oncogenes, they have been identified as mutated alleles of cellular genes transduced into the viral genome, and normal cellular ras (c-ras) genes have been detected in almost all eukaryotic cells 3'4. Known as ras proto-oncogenes, they code for cellular ras p21 proteins, which are ubiquitously expressed and generally associated with the inner surface of the plasma membrane 5. In this article I shall examine some of the possible functions of normal p21-ras.
Identification of p21-ras as a GTP-binding protein The only biochemical property that is common to all forms of p21ras protein is the ability to bind guanine nucleotides 6, and hydrolyse GTP 7. This raises the possibility that p21-ras may belong to the ever-growing family of GTP-binding or G proteins. It was known for several years that receptor stimulation of adenylate cyclase requires GTP, before the underlying mechanism was found to involve a guanine nucleotide-binding protein that coupled the receptor to stimulation of the catalytic subunit of adenylate cyclase. The structure of this protein (Gs) was found to consist of three subunits o~, i5 and 7, with the ~-subunit possessing GTPase activity (for reviews see Refs 8, 9). Other G proteins were subsequently discovered, including Gi, which couples receptors to inhibition of adenylate cyclase, Go, identified in brain membranes, and transducin (Gt), which couples TINS, Vol. 11, No. 7, 1988
Is p21-ras a real G protein? rhodopsin to activation of cyclic GMP phosphodiesterase in retinal rods 8. There is a great deal of homology between the o~-subunits of these G proteins, and their [3subunits are largely identical. A sequence data bank search revealed several regions of sequence homology between the o~-subunit of transducin and p21-ras, particularly in the guanine nucleotide-binding domains TM. Although p21-ras has a much lower molecular weight than the o~-subunit of known transducing G proteins, other mammalian GTPbinding proteins have been identified, of similar size to p21-ras but not cross-reacting with anti-ras antibodies. They are also of unknown function, and include the placental G protein (Gp), also found in brain and platelets, and ADPribosylation factor (ARF)8, among others. Examination of the differences between the cellular ras gene products and the transforming viral proteins has allowed amino acid substitutions responsible for the transforming ability to be pinpointed 3. Using the technique of site-directed mutagenesis, specific mutants can be designed with their biochemical properties altered in a predictable manner to test for the presence or absence of mitogenicity in the gene products hA2, as will be discussed below. This approach is also able to provide evidence concerning the normal function of p21-ras.
Possibility of involvement of p21-ras in the function of adenylate cyclase Because of the homology between ras proteins and the c~subunits of G proteins, and because adenylate cyclase was the first system in which the role of G proteins in signal transduction was investigated, it was thought possible that p21-ras might be involved in modulating adenylate cyclase. The yeast S. cerevisiae expresses several ras genes, one of which codes for a regulatory component of its adenylate cyclase. In ras-2yeast mutants, cyclic AMP production is impaired la, and this activity can be restored by both viral and
human ras genes 14. However, yeast adenylate cyclase is not known to be stimulated by receptor activation, and attempts to identify a direct role for ras in the mammalian adenylate cyclase system have not been successful15. Nevertheless there is clearly some indirect interaction between ras and cyclic AMP generation or metabolism, because epithelial and fibroblast cell lines transformed with Ha- or Ki-murine sarcoma virus exhibit reduced adenylate cyclase activity16.
Annette C Dolphin Departmentof Pharmacology,St George'sHospital MedicalSchool, London5W17ORE, UK.
Evidence that p21-ras functions as a signal transducing protein There is strong evidence that p21-ras functions in some capacity as a signal transducing protein, particularly with respect to growth factors. Initial evidence showed that guanine nucleotide binding to p21-ras can be influenced by growth factors, including epidermal growth factor (EGF), transferrin and insulin 17A8. EGF and insulin both increase [3H]guanine nucleotide binding to membranes in a rat kidney cell line transformed with Ha-ras 17. However, the nature of the signal transducing system(s) responsible for the effects of these growth factors remains a matter of debate. In subsequent work, an extremely useful tool in examining the possible function of p21-ras has been the Xenopus laevis oocyte. This system has several advantages in that it responds to the growth factors progesterone and insulin with readily observable morphological changes associated with maturation, and it is amenable to micro-injection with proteins or DNA 19-2~. Thus it was observed that p21-ras was able to induce oocyte maturation ~9. Insulin and progesterone stimulate maturation by different mechanisms, activation by progesterone being associated with inhibition of adenylate cyclase, and activation by insulin with tyrosine kinase phosphorylation and possibly stimulation of phospholipase C2°. Because insulin can influence the ability of p21-ras to bind GTP ~7, and the function
© 1988, ElsevierPublications, Cambridge 0378- 5912/88/$02.00
287
C terminal
Fig. 1. N terminal Primarystructureof 20 p21-ras. (Modified Amino Acids 1 I fromRef.3.)Activating mutations I **
40
60
I
I
80 I
100
120
I
I
140
160
I
I
I
GDP / GTP binding
/
I •
Membrane attachment
i
t T
Phosphoryl group
of insulin in some systems is dependent upon GTP 22, it was an obvious possibility that p21ras might be the hypothetical G protein (Gm~), responsible for transducing the signal of insulin binding to the production of its multifarious actions inside the cell. To investigate this possibility, an antibody against ras was injected into oocytes, and the ability of insulin to induce their maturation was found to be blocked, with no effect on progesterone-induced maturation 2°. The antibody also blocked the ability of p21-v-Ha-ras protein to cause oocyte maturation2°. It recognized only a single endogenous protein in oocytes, which had a molecular weight of 21 kDa. Other workers have shown that p21-ras antibodies also prevent another action of insulin - stimulation of glycogen synthesis in hepatocytes 23. It will be of interest to know whether these antibodies also inhibit insulin-stimulated breakdown of phosphatidylinositol glycan, which is the central molecule in a recently discovered second messenger system 24. However, this type of study only implicates p21-ras at some point in the cascade of intracellular events following stimulation of the insulin receptor. To establish whether p21-ras could directly subserve any of the signal transducing functions of the elusive Gm~, results from the examination of cell-free or reconstituted systems will have to be assessed. Many G proteins have been found to be substrates in vitro for phosphorylation by kinases, including protein kinase C and tyrosine kinase 25. The result of C kinase phosphorylation of Gi is its inactivation26; this may represent a mechanism for down-regulation of receptor-mediated processes. CKi-p21-ras can also be phosphorylated by C-kinase under certain conditions 27. However, O'Brien et al. 28 have observed that tyrosine
288
/
180
'1
/
-r
•
Purine ring
kinase does not phosphorylate several species of p21-ras unless they are partially denatured. In contrast, they have observed that p21-ras bound to GDP inhibits the insulin-dependent tyrosine kinase autophosphorylation of its [3subtmit. This phosphorylation has previously been shown to be essential for the subsequent physiological effects of i n s u l i n 29. A very plausible explanation put forward by the authors 28 is that p21-rasGDP normally exerts a tonic inhibition of insulin receptor tyrosine kinase activity, and when insulin binds to its receptor it enhances GDP-GTP exchange on p21-ras 17, and inhibition of the tyrosine kinase is thus relieved. This result, in addition to that described above using anti-ras antibodies, again points to the conclusion that p21-ras is essential for physiological responses to insulin. An exciting new development in the search for the signal transducing function of p21-ras is represented by the report that a transient Ca2÷ current is abolished in fibroblasts transformed with several species of ras 3°. This is particularly important in view of the ability of growth factors to regulate intracellular Ca2+.
which expression of the cellular ras proto-oncogene was under the control of an inducible promoter, and could be switched on by dexamethasone. In this way, when normal p21-ras was over-produced in this cell line, it was observed that the stimulation of polyphosphoinositide breakdown by activation of certain receptors, including those for bombesin and bradykinin, was enhanced. However, several investigators have shown that platelet-derived growth factor (PDGF)-stimulated production of inositol phosphates was attenuated in cells transformed either with v-Ha-ras, using a similar inducible promoter 32, or with EJ-ras33. In the experiments of Parries et al. 32, the bradykinin response was enhanced, but the authors attributed this to an increased number of receptors. The opposing results obtained by these investigations, albeit using different species of ras, suggest that the story is not as simple as it first appeared, and that although p21-ras influences the phospholipase C second messenger system as it does several others 16, the interaction may be an indirect one. This conclusion is supported by the work of Pouyss6gur and his colleagues 34, who failed to observe persistent phospholipase C activation in fibroblasts transformed with Is p21-ras the G protein v-Ki-ras or Ha-ras. coupling receptors to One problem with experiments phospholipase C? using ras-transformed cell lines is Receptor stimulation of inositol that the ras gene must be transphospholipid metabolism results in cribed and translated, a process the formation of inositol phosphates which may be associated with and diacylglycerol (DAG), and is alterations in the activity of other thought to involve a G protein genes. This problem can be surcoupling the receptor to a specific mounted by the technique already phospholipase C enzyme. The described of injecting the gene identity of this G protein is unclear, product p21-ras directly into the but there are several lines of cell. Although this has been perexperimentation that point in- formed particularly with oocytes, it directly to the possibility that p21- is also possible with smaller cells. In ras might serve such a function. fibroblasts injected with p21-ras Wakelam and colleagues31 have several events were triggered, investigated the normal function of including membrane ruffling and p21-ras by producing a cell line in pinocytosis, mimicking the effect of TINS, VoL 11, No. 7, 1988
mitogens. The only observed effect on phospholipid metabolism in these cells was a stimulation of lysophospholipid formation, indicating phospholipase A2 activation35. However, the difficulty with this technique is that individual cells must be injected, and the biochemical assays used may not have been sufficiently sensitive to detect a small change in phospholipase C activity in the limited amount of material available.
Properties of p2 1-ras associated with viral transformation Transformation of cells with rascontaining viruses is associated with many events whose underlying biochemical mechanism may provide a clue as to the function of p21-ras. In fibroblasts transformed by viral infection with several different species of ras, an increase was observed in the steady-state ratio of the products (DAG and inositol phosphates) to the precursors (polyphosphoinositides) of inositol lipid metabolism36. However, no evidence was presented that this effect of ras was due to direct stimulation of phospholipase C. Several more recent studies found increased steady-state DAG levels with no change in inositol phosphate production in ras-transformed fibroblastsa7'38. Furthermore, this was associated with increased steady-state levels of phosphorylethanolamine and phosphorylcholine, indicating that the source of DAG was likely to be phospholipase C-induced breakdown of phosphatidylcholine or phosphatidylethanolamine 37. Basal phosphorylation of an 80 kDa protein kinase C substrate was also increased, and some evidence was obtained in terms of a reduced phorbol ester response that protein kinase C had become downregulated 38. This phenomenon has been shown previously to occur in response to prolonged phorbol ester exposure 3a, but in this case it was presumably due to chronic elevation of endogenous diacylglycerol. Thus, ras expression might selectively activate a phospholipase C that breaks down noninositol containing lipids; and there is evidence from Exton and his colleagues that such an enzyme is under G protein regulation 4°. Thus protein kinase C-mediated events, TINS, VoL 11, No. 7, 1988
but probably not IP3-dependent processes, are selectively implicated in transformation by ras. This conclusion fits with the finding that ras-induced transformation is associated with a rise in intracellular pH, since activation of the Na+/H + antiporter is protein kinase Cmediated41. Indeed, down-regulation of protein kinase C has been found to inhibit p21-ras-induced mitogenicity37.
Use of site-directed mutagenesis in the study of ras-induced transformation It is now possible to produce mutant ras forms with single amino acid changes at any desired site, for example the GTP-binding site (Fig. 1). By this means the effect of changes in affinity of p21-ras for GTP and changes in GTPase activity have been studied on the transforming potential of the mutant ras species 3'n'12. It has previously been noted that many oncogenic mutants of p21-ras have reduced in-vitro GTPase activity although there is no apparent quantitative relationship between the two properties 7. However, the production of mutants with reduced ability to bind guanine nucleofides did not always impair the transforming capability of p21-ras n'12. One substitution (isoleucine for asparagine at position 116) had no detectable GTP-binding capability but nevertheless caused a nontransforming ras species to become transforming n. The explanation put forward was that mutations that reduce the affinity for guanine nucleotide binding are likely to do so by increasing their dissociation rate, and since the endogenous GTP:GDP ratio is high, the empty binding site so formed is more likely
to be occupied subsequently by GTP than GDP; thus p21-ras will spend more time in the activated state (Fig. 2). However, it is also conceivable that an amino acid substitution could permanently stabilize p21-ras in its activated conformation, in the absence of any guanine nucleotide in its binding site. Whatever the mechanism, transforming ability appears to be associated with activated p21-ras. Trahey and McCormick42 have recently tested this hypothesis by injecting GTP-bound normal and mutant p21-ras species into oocytes. The mutants had aspartate or valine instead of glycine at position 12 and showed reduced GTPase activity compared with normal p21-ras. In the oocyte, the difference between the species was greatly enhanced, the mutant forms remaining GTP-bound while the GTPase activity of normal p21ras was activated much more rapidly than in vitro. The results indicated that this was due to activation of the GTPase by a soluble oocyte protein that was unable to interact with the mutant p21-ras forms. Thus, not only did the oncogenic p21-ras mutants have reduced GTPase activity, but also, and possibly more importantly, this activity could not be stimulated by an endogenous factor, so that in vivo, the mutant forms were permanently GTP-activated. This may be an explanation for the much more long-lasting effects of microinjected mutant compared with normal p21-ras 35. The nature of this endogenous GTPase activating factor is unknown, but there is a precedent for such an action, in that the GTPase activity of Goo~ is stimulated (in the presence of
Fig. 2. Speciesof p2 i-ras. Oncogene formation by the mutation of proto-oncogenes may affect p2 lras in several ways: (1) increased GDP off-rate; (2) non-nucleotide bound form may assume active
conformation; and (3) reduced GTPase activity. 289
additional roles are likely4'5°'$1. Indeed, p21-ras appears in some cases to be involved in differentiation rather than division, and mimics NGF in PC12 cells, a chromaffin cell line52. Nevertheless, there is a clear association between tumour formation and mutated cellular ras proto-oncogenes 5'53-55, and it has been found that 10-15% of all human cancers are associated with ras oncogenes, a higher frequency than other oncogenes 5'53. Mitogenicity may also be associated with amplification of normal ras genes or their study 44,45. increased expression because of altered properties of the promoter. Structure and subcellular Thus, it is likely that an understanddistribution of p21-ras ing of the function of ras gene The three-dimensional structure products in the processes of signal of p21-ras has recently been transduction by growth factors will investigated by deVos et al.46 using also be a major step forward in X-ray crystallography. Its sub- understanding how the normal cellular distribution has been stud- control mechanisms in this system ied in several cell types, and normal may be disrupted and result in p21-ras appears not to be an neoplasia. More specifically, drugs integral membrane protein, but is that impair activation of the p21-ras associated with the inner surface of G protein such as inhibitors of its the plasma membrane 5, being fatty acylation or inhibitors of GTP anchored by post-translational fatty binding (e.g. GDP analogues) may acylation at its carboxy terminal 47 be efficacious in the inhibition (Fig. 1). Myristic or palmitic acid of ras-induced transformation. appear to be the fatty acids Evidence for the latter possibility involved. Mutant p21-ras proteins may be provided when compounds without an acylation site be- that can cross the plasma memcome non-transforming 5'36, but the brane and subsequently interact transforming ability can be restored with the guanine nucleotide binding by tagging onto the mutant a region site have been developed. Howfrom the src gene that codes for the ever, the staggering multiplicity of myristylation site of the src gene other oncogenes and their many product 36. This supports the idea ingenious modes of action56 lends a that p21-ras, both the normal and cautionary note to any hope for mitogenic species, are involved at novel therapy. the cell membrane in some form of signal transducfion. It is not known Selected references if the association of p21-ras with the 1 Harvey, J.J. (1964) Nature204, 1104membrane is also due to binding 1105 2 Kirsten, W. H. and Mayer, L. A. (1967) with membrane proteins, but by J. Natl Cancer Inst. 39, 311-335 analogy with other G proteins, it is 3 Barbacid, M. (1987) Annu. Rev. likely to interact through both Biochem. 56, 779-827 detector and effector domains, the 4 Furth, M. E., Aldrich, T. H. and CordonCardo, C. (1987) Oncogene 1,47-58 latter being influenced by the 5 Willumsen, B. M., Christensen, A., binding of GTP. It has been Hubbert, N. L., Papageorge, A. G. and suggested that p21-ras may interLowy, D. R. (1984) Nature 310, 583act with the cytoskeleton48, and 586 two p21-ras-binding proteins have 6 Scolnick, E. M., Papageorge, A. G. and Shih, T. Y. (1979) Proc. NatlAcad. Sci. been identified in erythrocytes 49.
Mge+ in normal intraceilular concentrations) by reassociation with 15~,-subunits to form the holoG protein 8. However, there is no evidence that p21-ras can interact with G protein [3y-subunits, and these are in any case membranebound rather than soluble. It has also been observed that the GTPase activity of the ribosomal G protein EF-Tu, though very low in vitro, can be stimulated by other ribosomal components 43. The interaction of the GTP-ase activating factor (GAP) with p21-ras has now been the subject of further
Role of p21-ras in tumour formation Although much available evidence indicates a role for p21-ras in the function of cellular growth factors and mitogens, the high levels of p21-ras in non-dividing tissues such as heart and brain suggest that 290
USA 76, 5355-5359 7 Temeles, G. L., Gibbs, J. B., D'Alonzo, J. S., Sigal, I. S. and Scolnick, E. M. (1985) Nature 313, 700-703 8 Gilman, A. G. (1987) Annu. Rev. Biochem. 56, 615-649 9 Dolphin, A. C. (1987) TrendsNeurosci. 10, 53-57 10 Hurley, J. B., Simon, M. I., Teplow, D. B., Robishaw, J. D. and Gilman, A. G. (1984) Science 226, 860-862
11 Walter, M., Clark, S. G. and Levinson, A. D. (1986) Science 233, 649-652 12 Sigal, I. S. etal. (1986) Proc. NatlAcad. Sci. USA 83, 952-956 13 Broek, D., Samiy, N., Fasano, O., Fujiyama, A. and Tamanoi, F. (1985) Cell 41,763-769 14 Todo, Y. et al. (1985) Cell 40, 27-36 15 Beckner, S. K., Hattori, S. and Shih, T. Y. (1985) Nature 317, 71-72 16 Beckner, S. K. (1984) FEBS Lett. 166, 170-174 17 Kamata, T. and Feramisco, J. R. (1984) Nature 310, 147-150 18 Finkel, T. and Cooper, G. M. (1984) Ce1136, 1115-1121 19 Birchmeier, C., Broek, D. and Wigler, M. (1985) Ce1143, 615-621 20 Korn, L. J., Siebel, C. W., McCormick, F. and Roth, R. A. (1987) Science236, 840-842 21 Lacal, J. C. et a/. (1987) Science 238, 533-536 22 Heyworth, C. M., Whetton, A. D., Wong, S., Martin, B. R. and Houslay, M. D. (1985) Biochem. J. 228, 593603 23 Deshpande, A. K. and Kung, H-F. (1987) Mol. Cell Biol. 7, 1285-1288 24 Fox, J. A., Soliz, N. M. and Saltiel, A. R. (1987) Proc. Natl Acad. Sci. USA 84, 2663-2667 25 O'Brien, R. M., Houslay, M. D., Milligan, G. and Sicldle, K. (1987) FEBS Left. 212, 281-288 26 Katada, T., Gilman, A. G., Watanabe, Y., Bauer, S. and Jakobs, K. H. (1985) Eur. J. Biochem. 151,431-437 27 Ballester, R., Furth, M. E. and Rosen, O. M. (1987) J. Biol. Chem. 262, 2688-2695 28 O'Brien, R. M., Siddle, K., Houslay, M. D. and Hall, A. (1987) FEBS Lett. 217, 253-259 29 Morgan, D. O. and Roth, R. A. (1987) Proc. Natl Acad. Sci. USA 84, 41-45 30 Chen, C., Corbley, M. J., Roberts, T. M. and Hess, P. (1988) Biophys. J. 53, 232a 31 Wakelarn, M. J. O. etal. (1986) Nature 323, 173-176 32 Parries, G., Hoebel, R. and Racker, E. (1987) Proc. Natl Acad. Sci. USA 84, 2648-2652 33 Benjamin, C. W., Tarpley, W. G. and Gorman, R. R. (1987) Biochem. Biophys. Res. Commun. 145, 1254-1259 34 Seuwen, K., Lagarde, A. and PouyssEgur, J. (1988) EMBO J. 7, 161168 35 Bar-Sagi, D. and Feramisco, J. R. (1986) Science 233, 1061-1068 36 Fleischman, L. F., Chahwala, S. B. and Cantley, L. (1986) Science 231,407410 37 Lacal, J. C., Moscat, J. and Aaronson, S. A. (1987) Nature 330, 269-272 38 Wolfman, A. and Macara, I. G. (1987) Nature 325, 359-361 39 Ballester, R. and Rosen, O. M. (1985) J. Biol. Chem. 260, 15194-15199 40 Irving, H. R. and Exton, J. H. (1987) J. Biol. Chem. 260, 14201-14207 41 Hagag, N., Lacal, J. C., Graber, M., Aaronson, S. A. and Viola, M. V. (1987) MoL Cell. Biol. 1, 1984-1988 42 Trahey, M. and McCormick, F. (1987) Science 238, 542-545 43 Gordon, J. (1969) J. BioL Chem. 244, 5680-5686 44 Cales, C., Hancock, J. F., Marshall, C. TINS, Vol. 11, No. 7, 1988
45 46 47 48
and Hall, A. (1988) Nature 332,548551 Sigal, I. S. (1988) Nature 332,485-486 deVos, K. et al. (1988) Science 239, 888-893 Fujiyama, A. and Tamanoi, F. (1986) Proc. Natl Acad. Sci. USA 83, 12661270 Haniey, M. R. and Jackson, T. (1987)
Nature 328, 668--669 49 Tanimoto, T. et al. (1988) FEBS Left. 226, 291-296 50 Tanaka, T. et al. (1986) MoL Cell. Biochem. 70, 97-104 51 Spandidos, D. A. and Dimitrov, T. (1985) Biosci. Rep. 5, 1035-1039 52 Bar-Sagi, D. and Feramisco, J. R. (1985) Cell 42, 841-848
53 Bos, J. L., Toksoz, D., Marshall, C. J., Verlaan-de-Vries, M. and Veeneman, G. H. (1985) Nature 315, 726-730 54 Bos,J. L. etal. (1987) Nature 327,293297 55 Forrester, K. etal. (1987)Nature 327, 298-303 56 Bishop,J. M. (1987) Science 235,305311
ive years ago Sternberger and F Sternberger I pubfished a semi- Neurofilamentprotein phosphorylation- where, nal paper showing that many monoclonal antibodies (mAbs) whenand why against the high molecular weight (200 kDa) component of neurofilaments (NF-H) react only with the phosphorylated form of this protein. These investigators also described antibodies that reacted exclusively with non-phosphorylated neurofilament proteins, and one that reacted with neurofilament proteins whether they were phosphorylated or not. Most importantly the Sternbergers showed that these antibodies could be used to determine the spatial distribution of these post-translationally modified forms of neurofilament proteins in nervous tissue. They found that NF-H in the perikaryon and dendrites of neurons is not phosphorylated whereas NF-H in axons, especially long projection axons, is phosphorylated. This fundamental observation has since been confirmed in a number of laboratories and several related phenomena have been described. For example, Marc et al.2 have shown that in the cerebellum of hypothyroid rats NF-H phosphorylation is delayed, in accordance with the general retardation of development. More interestingly, these workers found that basket cell axons, which in normal animals contain both phosphorylated and non-phosphorylated neurofilaments, contain only phosphorylated neurofilaments in hypothyroid animals. The antineurofilament staining also revealed that in hypothyroid animals these axons develop abnormally, forming sparse and poorly organized baskets, suggesting a possible relationship between neurofilament phosphorylation and axonal morphogenesis. Several recent papers in the Journal of Neuroscience have dealt with neurofilament phosphorylation. Foster et al. 3 examined the appearance of phosphorylated and TINS, Vol. 11, No. 7, 1988
non-phosphorylated forms of NF-H in mouse spinal cord cultures; Cohen et al. 4 have investigated the homologous 220 kDa protein in the squid giant axon; and Lee, Carden and colleagues5 have used a large battery of mAbs against NF-H and the 160 kDa neurofilament proteins (NF-M) in different states of phosphorylation to examine the rat spinal cord. The main conclusion of all three studies is the same; the phosphorylation of these high mop ecular weight neurofilament proteins occurs simultaneously with their transport into the axon. The study by Lee et al. ~ is particularly interesting from the technical point of view. By raising over 300 mAbs against neurofilament proteins they have introduced unprecedented refinement into the analysis of the protein phosphorylation states of identified molecules. On immunoblots their mAbs can distinguish three different degrees of NF-M phosphorylation, which they call P+, P + + and P+ + +. They demonstrate that non-phosphorylated neurofilamerits are present only in cell bodies and highly phosphorylated (P+ + +) neurofilaments occur only in axons. Only the lowest phosphorylation state (P+) is present in the cell body (as well as in the axon close to the perikaryon). A companion paper by Carden et al.6 uses the same mAbs to document the appearance of neurofilament protein phosphorylation in the developing spinal cord. This shows that partially phosphorylated (P+) NF-M is present 24 h before the fully phosphorylated (P + + +) form. A slightly disappointing feature of the results is that there is rather little spatial or developmental distinction between the different de-
grees of NF-H and NF-M phosphorylation. There is no distinguishable gradient of P+ + + staining within the axon so it appears that NF-M and NF-H become fully phosphorylated soon after entering the axon. This leads Carden et al. ~ to suggest that some difference in conditions between cell body and axon, such as pH or ionic composition, is responsible for the apparent activation of neurofilamentdirected kinases within the axon. Little is known about the kinases that phosphorylate neurofilaments, but by means of cDNA sequencing, potential phosphate acceptor sites are beginning to be identified. At least some of these exist as tandemly repeated peptide sequences 7,s. Lee et al. 9 have raised antibodies to synthetic peptides based on one of these sequences, in either non-phosphorylated or chemically phosphorylated form. They also used the synthetic peptides to demonstrate that 96 out of their 515 anti-NF mAbs are specific for the phosphorylated or nonphosphorylated forms of this sequence, which thus constitutes a major immunogenic neurofilament epitope. These antibodies can now be used to study the occurrence and phosphorylation state of this particular phosphate acceptor sequence in neurofilaments. This approach has great potential, since it should enable a precise analysis of when and under what circumstances specific neurofilament phosphorylation events occur in the cell. What is the functional role of the NF-H and NF-M phosphorylation in axons? Whatever it is, it appears to be of fundamental importance since Cohen et al. 4 find that essentially the same phenomenon of phosphorylation linked to transport in the axon occurs in squid as
© 1988, ElsevierPublications, Cambridge 0378- 5912/88/$0200
Andrew Matus Friedrich MiescherInstitut, PO Box 2543, 4002 Basel, Switzerland.
291
news
R
as, like other oncogenes, was first described as the gene in a particular RNA tumour viral genome which was essential for the transforming activity of the virus. Slightly different versions of ras were discovered in two strains [Harvey (Ha) and Kirsten (Ki)] of rat sarcoma virus, and the ability of these viruses to induce tumours and transform cells in culture was found to be due to expression of the viral ras gene products inside the cell1'2. These gene products are closely related, 21 kDa (p21) proteins. Subsequent to the description of these and other viral v-ras oncogenes and their gene products, they were observed in many types of tumour (for review see Ref. 3). Like many other oncogenes, they have been identified as mutated alleles of cellular genes transduced into the viral genome, and normal cellular ras (c-ras) genes have been detected in almost all eukaryotic cells 3'4. Known as ras proto-oncogenes, they code for cellular ras p21 proteins, which are ubiquitously expressed and generally associated with the inner surface of the plasma membrane 5. In this article I shall examine some of the possible functions of normal p21-ras.
Identification of p21-ras as a GTP-binding protein The only biochemical property that is common to all forms of p21ras protein is the ability to bind guanine nucleotides 6, and hydrolyse GTP 7. This raises the possibility that p21-ras may belong to the ever-growing family of GTP-binding or G proteins. It was known for several years that receptor stimulation of adenylate cyclase requires GTP, before the underlying mechanism was found to involve a guanine nucleotide-binding protein that coupled the receptor to stimulation of the catalytic subunit of adenylate cyclase. The structure of this protein (Gs) was found to consist of three subunits o~, i5 and 7, with the ~-subunit possessing GTPase activity (for reviews see Refs 8, 9). Other G proteins were subsequently discovered, including Gi, which couples receptors to inhibition of adenylate cyclase, Go, identified in brain membranes, and transducin (Gt), which couples TINS, Vol. 11, No. 7, 1988
Is p21-ras a real G protein? rhodopsin to activation of cyclic GMP phosphodiesterase in retinal rods 8. There is a great deal of homology between the o~-subunits of these G proteins, and their [3subunits are largely identical. A sequence data bank search revealed several regions of sequence homology between the o~-subunit of transducin and p21-ras, particularly in the guanine nucleotide-binding domains TM. Although p21-ras has a much lower molecular weight than the o~-subunit of known transducing G proteins, other mammalian GTPbinding proteins have been identified, of similar size to p21-ras but not cross-reacting with anti-ras antibodies. They are also of unknown function, and include the placental G protein (Gp), also found in brain and platelets, and ADPribosylation factor (ARF)8, among others. Examination of the differences between the cellular ras gene products and the transforming viral proteins has allowed amino acid substitutions responsible for the transforming ability to be pinpointed 3. Using the technique of site-directed mutagenesis, specific mutants can be designed with their biochemical properties altered in a predictable manner to test for the presence or absence of mitogenicity in the gene products hA2, as will be discussed below. This approach is also able to provide evidence concerning the normal function of p21-ras.
Possibility of involvement of p21-ras in the function of adenylate cyclase Because of the homology between ras proteins and the c~subunits of G proteins, and because adenylate cyclase was the first system in which the role of G proteins in signal transduction was investigated, it was thought possible that p21-ras might be involved in modulating adenylate cyclase. The yeast S. cerevisiae expresses several ras genes, one of which codes for a regulatory component of its adenylate cyclase. In ras-2yeast mutants, cyclic AMP production is impaired la, and this activity can be restored by both viral and
human ras genes 14. However, yeast adenylate cyclase is not known to be stimulated by receptor activation, and attempts to identify a direct role for ras in the mammalian adenylate cyclase system have not been successful15. Nevertheless there is clearly some indirect interaction between ras and cyclic AMP generation or metabolism, because epithelial and fibroblast cell lines transformed with Ha- or Ki-murine sarcoma virus exhibit reduced adenylate cyclase activity16.
Annette C Dolphin Departmentof Pharmacology,St George'sHospital MedicalSchool, London5W17ORE, UK.
Evidence that p21-ras functions as a signal transducing protein There is strong evidence that p21-ras functions in some capacity as a signal transducing protein, particularly with respect to growth factors. Initial evidence showed that guanine nucleotide binding to p21-ras can be influenced by growth factors, including epidermal growth factor (EGF), transferrin and insulin 17A8. EGF and insulin both increase [3H]guanine nucleotide binding to membranes in a rat kidney cell line transformed with Ha-ras 17. However, the nature of the signal transducing system(s) responsible for the effects of these growth factors remains a matter of debate. In subsequent work, an extremely useful tool in examining the possible function of p21-ras has been the Xenopus laevis oocyte. This system has several advantages in that it responds to the growth factors progesterone and insulin with readily observable morphological changes associated with maturation, and it is amenable to micro-injection with proteins or DNA 19-2~. Thus it was observed that p21-ras was able to induce oocyte maturation ~9. Insulin and progesterone stimulate maturation by different mechanisms, activation by progesterone being associated with inhibition of adenylate cyclase, and activation by insulin with tyrosine kinase phosphorylation and possibly stimulation of phospholipase C2°. Because insulin can influence the ability of p21-ras to bind GTP ~7, and the function
© 1988, ElsevierPublications, Cambridge 0378- 5912/88/$02.00
287
C terminal
Fig. 1. N terminal Primarystructureof 20 p21-ras. (Modified Amino Acids 1 I fromRef.3.)Activating mutations I **
40
60
I
I
80 I
100
120
I
I
140
160
I
I
I
GDP / GTP binding
/
I •
Membrane attachment
i
t T
Phosphoryl group
of insulin in some systems is dependent upon GTP 22, it was an obvious possibility that p21ras might be the hypothetical G protein (Gm~), responsible for transducing the signal of insulin binding to the production of its multifarious actions inside the cell. To investigate this possibility, an antibody against ras was injected into oocytes, and the ability of insulin to induce their maturation was found to be blocked, with no effect on progesterone-induced maturation 2°. The antibody also blocked the ability of p21-v-Ha-ras protein to cause oocyte maturation2°. It recognized only a single endogenous protein in oocytes, which had a molecular weight of 21 kDa. Other workers have shown that p21-ras antibodies also prevent another action of insulin - stimulation of glycogen synthesis in hepatocytes 23. It will be of interest to know whether these antibodies also inhibit insulin-stimulated breakdown of phosphatidylinositol glycan, which is the central molecule in a recently discovered second messenger system 24. However, this type of study only implicates p21-ras at some point in the cascade of intracellular events following stimulation of the insulin receptor. To establish whether p21-ras could directly subserve any of the signal transducing functions of the elusive Gm~, results from the examination of cell-free or reconstituted systems will have to be assessed. Many G proteins have been found to be substrates in vitro for phosphorylation by kinases, including protein kinase C and tyrosine kinase 25. The result of C kinase phosphorylation of Gi is its inactivation26; this may represent a mechanism for down-regulation of receptor-mediated processes. CKi-p21-ras can also be phosphorylated by C-kinase under certain conditions 27. However, O'Brien et al. 28 have observed that tyrosine
288
/
180
'1
/
-r
•
Purine ring
kinase does not phosphorylate several species of p21-ras unless they are partially denatured. In contrast, they have observed that p21-ras bound to GDP inhibits the insulin-dependent tyrosine kinase autophosphorylation of its [3subtmit. This phosphorylation has previously been shown to be essential for the subsequent physiological effects of i n s u l i n 29. A very plausible explanation put forward by the authors 28 is that p21-rasGDP normally exerts a tonic inhibition of insulin receptor tyrosine kinase activity, and when insulin binds to its receptor it enhances GDP-GTP exchange on p21-ras 17, and inhibition of the tyrosine kinase is thus relieved. This result, in addition to that described above using anti-ras antibodies, again points to the conclusion that p21-ras is essential for physiological responses to insulin. An exciting new development in the search for the signal transducing function of p21-ras is represented by the report that a transient Ca2÷ current is abolished in fibroblasts transformed with several species of ras 3°. This is particularly important in view of the ability of growth factors to regulate intracellular Ca2+.
which expression of the cellular ras proto-oncogene was under the control of an inducible promoter, and could be switched on by dexamethasone. In this way, when normal p21-ras was over-produced in this cell line, it was observed that the stimulation of polyphosphoinositide breakdown by activation of certain receptors, including those for bombesin and bradykinin, was enhanced. However, several investigators have shown that platelet-derived growth factor (PDGF)-stimulated production of inositol phosphates was attenuated in cells transformed either with v-Ha-ras, using a similar inducible promoter 32, or with EJ-ras33. In the experiments of Parries et al. 32, the bradykinin response was enhanced, but the authors attributed this to an increased number of receptors. The opposing results obtained by these investigations, albeit using different species of ras, suggest that the story is not as simple as it first appeared, and that although p21-ras influences the phospholipase C second messenger system as it does several others 16, the interaction may be an indirect one. This conclusion is supported by the work of Pouyss6gur and his colleagues 34, who failed to observe persistent phospholipase C activation in fibroblasts transformed with Is p21-ras the G protein v-Ki-ras or Ha-ras. coupling receptors to One problem with experiments phospholipase C? using ras-transformed cell lines is Receptor stimulation of inositol that the ras gene must be transphospholipid metabolism results in cribed and translated, a process the formation of inositol phosphates which may be associated with and diacylglycerol (DAG), and is alterations in the activity of other thought to involve a G protein genes. This problem can be surcoupling the receptor to a specific mounted by the technique already phospholipase C enzyme. The described of injecting the gene identity of this G protein is unclear, product p21-ras directly into the but there are several lines of cell. Although this has been perexperimentation that point in- formed particularly with oocytes, it directly to the possibility that p21- is also possible with smaller cells. In ras might serve such a function. fibroblasts injected with p21-ras Wakelam and colleagues31 have several events were triggered, investigated the normal function of including membrane ruffling and p21-ras by producing a cell line in pinocytosis, mimicking the effect of TINS, VoL 11, No. 7, 1988
mitogens. The only observed effect on phospholipid metabolism in these cells was a stimulation of lysophospholipid formation, indicating phospholipase A2 activation35. However, the difficulty with this technique is that individual cells must be injected, and the biochemical assays used may not have been sufficiently sensitive to detect a small change in phospholipase C activity in the limited amount of material available.
Properties of p2 1-ras associated with viral transformation Transformation of cells with rascontaining viruses is associated with many events whose underlying biochemical mechanism may provide a clue as to the function of p21-ras. In fibroblasts transformed by viral infection with several different species of ras, an increase was observed in the steady-state ratio of the products (DAG and inositol phosphates) to the precursors (polyphosphoinositides) of inositol lipid metabolism36. However, no evidence was presented that this effect of ras was due to direct stimulation of phospholipase C. Several more recent studies found increased steady-state DAG levels with no change in inositol phosphate production in ras-transformed fibroblastsa7'38. Furthermore, this was associated with increased steady-state levels of phosphorylethanolamine and phosphorylcholine, indicating that the source of DAG was likely to be phospholipase C-induced breakdown of phosphatidylcholine or phosphatidylethanolamine 37. Basal phosphorylation of an 80 kDa protein kinase C substrate was also increased, and some evidence was obtained in terms of a reduced phorbol ester response that protein kinase C had become downregulated 38. This phenomenon has been shown previously to occur in response to prolonged phorbol ester exposure 3a, but in this case it was presumably due to chronic elevation of endogenous diacylglycerol. Thus, ras expression might selectively activate a phospholipase C that breaks down noninositol containing lipids; and there is evidence from Exton and his colleagues that such an enzyme is under G protein regulation 4°. Thus protein kinase C-mediated events, TINS, VoL 11, No. 7, 1988
but probably not IP3-dependent processes, are selectively implicated in transformation by ras. This conclusion fits with the finding that ras-induced transformation is associated with a rise in intracellular pH, since activation of the Na+/H + antiporter is protein kinase Cmediated41. Indeed, down-regulation of protein kinase C has been found to inhibit p21-ras-induced mitogenicity37.
Use of site-directed mutagenesis in the study of ras-induced transformation It is now possible to produce mutant ras forms with single amino acid changes at any desired site, for example the GTP-binding site (Fig. 1). By this means the effect of changes in affinity of p21-ras for GTP and changes in GTPase activity have been studied on the transforming potential of the mutant ras species 3'n'12. It has previously been noted that many oncogenic mutants of p21-ras have reduced in-vitro GTPase activity although there is no apparent quantitative relationship between the two properties 7. However, the production of mutants with reduced ability to bind guanine nucleofides did not always impair the transforming capability of p21-ras n'12. One substitution (isoleucine for asparagine at position 116) had no detectable GTP-binding capability but nevertheless caused a nontransforming ras species to become transforming n. The explanation put forward was that mutations that reduce the affinity for guanine nucleotide binding are likely to do so by increasing their dissociation rate, and since the endogenous GTP:GDP ratio is high, the empty binding site so formed is more likely
to be occupied subsequently by GTP than GDP; thus p21-ras will spend more time in the activated state (Fig. 2). However, it is also conceivable that an amino acid substitution could permanently stabilize p21-ras in its activated conformation, in the absence of any guanine nucleotide in its binding site. Whatever the mechanism, transforming ability appears to be associated with activated p21-ras. Trahey and McCormick42 have recently tested this hypothesis by injecting GTP-bound normal and mutant p21-ras species into oocytes. The mutants had aspartate or valine instead of glycine at position 12 and showed reduced GTPase activity compared with normal p21-ras. In the oocyte, the difference between the species was greatly enhanced, the mutant forms remaining GTP-bound while the GTPase activity of normal p21ras was activated much more rapidly than in vitro. The results indicated that this was due to activation of the GTPase by a soluble oocyte protein that was unable to interact with the mutant p21-ras forms. Thus, not only did the oncogenic p21-ras mutants have reduced GTPase activity, but also, and possibly more importantly, this activity could not be stimulated by an endogenous factor, so that in vivo, the mutant forms were permanently GTP-activated. This may be an explanation for the much more long-lasting effects of microinjected mutant compared with normal p21-ras 35. The nature of this endogenous GTPase activating factor is unknown, but there is a precedent for such an action, in that the GTPase activity of Goo~ is stimulated (in the presence of
Fig. 2. Speciesof p2 i-ras. Oncogene formation by the mutation of proto-oncogenes may affect p2 lras in several ways: (1) increased GDP off-rate; (2) non-nucleotide bound form may assume active
conformation; and (3) reduced GTPase activity. 289
additional roles are likely4'5°'$1. Indeed, p21-ras appears in some cases to be involved in differentiation rather than division, and mimics NGF in PC12 cells, a chromaffin cell line52. Nevertheless, there is a clear association between tumour formation and mutated cellular ras proto-oncogenes 5'53-55, and it has been found that 10-15% of all human cancers are associated with ras oncogenes, a higher frequency than other oncogenes 5'53. Mitogenicity may also be associated with amplification of normal ras genes or their study 44,45. increased expression because of altered properties of the promoter. Structure and subcellular Thus, it is likely that an understanddistribution of p21-ras ing of the function of ras gene The three-dimensional structure products in the processes of signal of p21-ras has recently been transduction by growth factors will investigated by deVos et al.46 using also be a major step forward in X-ray crystallography. Its sub- understanding how the normal cellular distribution has been stud- control mechanisms in this system ied in several cell types, and normal may be disrupted and result in p21-ras appears not to be an neoplasia. More specifically, drugs integral membrane protein, but is that impair activation of the p21-ras associated with the inner surface of G protein such as inhibitors of its the plasma membrane 5, being fatty acylation or inhibitors of GTP anchored by post-translational fatty binding (e.g. GDP analogues) may acylation at its carboxy terminal 47 be efficacious in the inhibition (Fig. 1). Myristic or palmitic acid of ras-induced transformation. appear to be the fatty acids Evidence for the latter possibility involved. Mutant p21-ras proteins may be provided when compounds without an acylation site be- that can cross the plasma memcome non-transforming 5'36, but the brane and subsequently interact transforming ability can be restored with the guanine nucleotide binding by tagging onto the mutant a region site have been developed. Howfrom the src gene that codes for the ever, the staggering multiplicity of myristylation site of the src gene other oncogenes and their many product 36. This supports the idea ingenious modes of action56 lends a that p21-ras, both the normal and cautionary note to any hope for mitogenic species, are involved at novel therapy. the cell membrane in some form of signal transducfion. It is not known Selected references if the association of p21-ras with the 1 Harvey, J.J. (1964) Nature204, 1104membrane is also due to binding 1105 2 Kirsten, W. H. and Mayer, L. A. (1967) with membrane proteins, but by J. Natl Cancer Inst. 39, 311-335 analogy with other G proteins, it is 3 Barbacid, M. (1987) Annu. Rev. likely to interact through both Biochem. 56, 779-827 detector and effector domains, the 4 Furth, M. E., Aldrich, T. H. and CordonCardo, C. (1987) Oncogene 1,47-58 latter being influenced by the 5 Willumsen, B. M., Christensen, A., binding of GTP. It has been Hubbert, N. L., Papageorge, A. G. and suggested that p21-ras may interLowy, D. R. (1984) Nature 310, 583act with the cytoskeleton48, and 586 two p21-ras-binding proteins have 6 Scolnick, E. M., Papageorge, A. G. and Shih, T. Y. (1979) Proc. NatlAcad. Sci. been identified in erythrocytes 49.
Mge+ in normal intraceilular concentrations) by reassociation with 15~,-subunits to form the holoG protein 8. However, there is no evidence that p21-ras can interact with G protein [3y-subunits, and these are in any case membranebound rather than soluble. It has also been observed that the GTPase activity of the ribosomal G protein EF-Tu, though very low in vitro, can be stimulated by other ribosomal components 43. The interaction of the GTP-ase activating factor (GAP) with p21-ras has now been the subject of further
Role of p21-ras in tumour formation Although much available evidence indicates a role for p21-ras in the function of cellular growth factors and mitogens, the high levels of p21-ras in non-dividing tissues such as heart and brain suggest that 290
USA 76, 5355-5359 7 Temeles, G. L., Gibbs, J. B., D'Alonzo, J. S., Sigal, I. S. and Scolnick, E. M. (1985) Nature 313, 700-703 8 Gilman, A. G. (1987) Annu. Rev. Biochem. 56, 615-649 9 Dolphin, A. C. (1987) TrendsNeurosci. 10, 53-57 10 Hurley, J. B., Simon, M. I., Teplow, D. B., Robishaw, J. D. and Gilman, A. G. (1984) Science 226, 860-862
11 Walter, M., Clark, S. G. and Levinson, A. D. (1986) Science 233, 649-652 12 Sigal, I. S. etal. (1986) Proc. NatlAcad. Sci. USA 83, 952-956 13 Broek, D., Samiy, N., Fasano, O., Fujiyama, A. and Tamanoi, F. (1985) Cell 41,763-769 14 Todo, Y. et al. (1985) Cell 40, 27-36 15 Beckner, S. K., Hattori, S. and Shih, T. Y. (1985) Nature 317, 71-72 16 Beckner, S. K. (1984) FEBS Lett. 166, 170-174 17 Kamata, T. and Feramisco, J. R. (1984) Nature 310, 147-150 18 Finkel, T. and Cooper, G. M. (1984) Ce1136, 1115-1121 19 Birchmeier, C., Broek, D. and Wigler, M. (1985) Ce1143, 615-621 20 Korn, L. J., Siebel, C. W., McCormick, F. and Roth, R. A. (1987) Science236, 840-842 21 Lacal, J. C. et a/. (1987) Science 238, 533-536 22 Heyworth, C. M., Whetton, A. D., Wong, S., Martin, B. R. and Houslay, M. D. (1985) Biochem. J. 228, 593603 23 Deshpande, A. K. and Kung, H-F. (1987) Mol. Cell Biol. 7, 1285-1288 24 Fox, J. A., Soliz, N. M. and Saltiel, A. R. (1987) Proc. Natl Acad. Sci. USA 84, 2663-2667 25 O'Brien, R. M., Houslay, M. D., Milligan, G. and Sicldle, K. (1987) FEBS Left. 212, 281-288 26 Katada, T., Gilman, A. G., Watanabe, Y., Bauer, S. and Jakobs, K. H. (1985) Eur. J. Biochem. 151,431-437 27 Ballester, R., Furth, M. E. and Rosen, O. M. (1987) J. Biol. Chem. 262, 2688-2695 28 O'Brien, R. M., Siddle, K., Houslay, M. D. and Hall, A. (1987) FEBS Lett. 217, 253-259 29 Morgan, D. O. and Roth, R. A. (1987) Proc. Natl Acad. Sci. USA 84, 41-45 30 Chen, C., Corbley, M. J., Roberts, T. M. and Hess, P. (1988) Biophys. J. 53, 232a 31 Wakelarn, M. J. O. etal. (1986) Nature 323, 173-176 32 Parries, G., Hoebel, R. and Racker, E. (1987) Proc. Natl Acad. Sci. USA 84, 2648-2652 33 Benjamin, C. W., Tarpley, W. G. and Gorman, R. R. (1987) Biochem. Biophys. Res. Commun. 145, 1254-1259 34 Seuwen, K., Lagarde, A. and PouyssEgur, J. (1988) EMBO J. 7, 161168 35 Bar-Sagi, D. and Feramisco, J. R. (1986) Science 233, 1061-1068 36 Fleischman, L. F., Chahwala, S. B. and Cantley, L. (1986) Science 231,407410 37 Lacal, J. C., Moscat, J. and Aaronson, S. A. (1987) Nature 330, 269-272 38 Wolfman, A. and Macara, I. G. (1987) Nature 325, 359-361 39 Ballester, R. and Rosen, O. M. (1985) J. Biol. Chem. 260, 15194-15199 40 Irving, H. R. and Exton, J. H. (1987) J. Biol. Chem. 260, 14201-14207 41 Hagag, N., Lacal, J. C., Graber, M., Aaronson, S. A. and Viola, M. V. (1987) MoL Cell. Biol. 1, 1984-1988 42 Trahey, M. and McCormick, F. (1987) Science 238, 542-545 43 Gordon, J. (1969) J. BioL Chem. 244, 5680-5686 44 Cales, C., Hancock, J. F., Marshall, C. TINS, Vol. 11, No. 7, 1988
45 46 47 48
and Hall, A. (1988) Nature 332,548551 Sigal, I. S. (1988) Nature 332,485-486 deVos, K. et al. (1988) Science 239, 888-893 Fujiyama, A. and Tamanoi, F. (1986) Proc. Natl Acad. Sci. USA 83, 12661270 Haniey, M. R. and Jackson, T. (1987)
Nature 328, 668--669 49 Tanimoto, T. et al. (1988) FEBS Left. 226, 291-296 50 Tanaka, T. et al. (1986) MoL Cell. Biochem. 70, 97-104 51 Spandidos, D. A. and Dimitrov, T. (1985) Biosci. Rep. 5, 1035-1039 52 Bar-Sagi, D. and Feramisco, J. R. (1985) Cell 42, 841-848
53 Bos, J. L., Toksoz, D., Marshall, C. J., Verlaan-de-Vries, M. and Veeneman, G. H. (1985) Nature 315, 726-730 54 Bos,J. L. etal. (1987) Nature 327,293297 55 Forrester, K. etal. (1987)Nature 327, 298-303 56 Bishop,J. M. (1987) Science 235,305311
ive years ago Sternberger and F Sternberger I pubfished a semi- Neurofilamentprotein phosphorylation- where, nal paper showing that many monoclonal antibodies (mAbs) whenand why against the high molecular weight (200 kDa) component of neurofilaments (NF-H) react only with the phosphorylated form of this protein. These investigators also described antibodies that reacted exclusively with non-phosphorylated neurofilament proteins, and one that reacted with neurofilament proteins whether they were phosphorylated or not. Most importantly the Sternbergers showed that these antibodies could be used to determine the spatial distribution of these post-translationally modified forms of neurofilament proteins in nervous tissue. They found that NF-H in the perikaryon and dendrites of neurons is not phosphorylated whereas NF-H in axons, especially long projection axons, is phosphorylated. This fundamental observation has since been confirmed in a number of laboratories and several related phenomena have been described. For example, Marc et al.2 have shown that in the cerebellum of hypothyroid rats NF-H phosphorylation is delayed, in accordance with the general retardation of development. More interestingly, these workers found that basket cell axons, which in normal animals contain both phosphorylated and non-phosphorylated neurofilaments, contain only phosphorylated neurofilaments in hypothyroid animals. The antineurofilament staining also revealed that in hypothyroid animals these axons develop abnormally, forming sparse and poorly organized baskets, suggesting a possible relationship between neurofilament phosphorylation and axonal morphogenesis. Several recent papers in the Journal of Neuroscience have dealt with neurofilament phosphorylation. Foster et al. 3 examined the appearance of phosphorylated and TINS, Vol. 11, No. 7, 1988
non-phosphorylated forms of NF-H in mouse spinal cord cultures; Cohen et al. 4 have investigated the homologous 220 kDa protein in the squid giant axon; and Lee, Carden and colleagues5 have used a large battery of mAbs against NF-H and the 160 kDa neurofilament proteins (NF-M) in different states of phosphorylation to examine the rat spinal cord. The main conclusion of all three studies is the same; the phosphorylation of these high mop ecular weight neurofilament proteins occurs simultaneously with their transport into the axon. The study by Lee et al. ~ is particularly interesting from the technical point of view. By raising over 300 mAbs against neurofilament proteins they have introduced unprecedented refinement into the analysis of the protein phosphorylation states of identified molecules. On immunoblots their mAbs can distinguish three different degrees of NF-M phosphorylation, which they call P+, P + + and P+ + +. They demonstrate that non-phosphorylated neurofilamerits are present only in cell bodies and highly phosphorylated (P+ + +) neurofilaments occur only in axons. Only the lowest phosphorylation state (P+) is present in the cell body (as well as in the axon close to the perikaryon). A companion paper by Carden et al.6 uses the same mAbs to document the appearance of neurofilament protein phosphorylation in the developing spinal cord. This shows that partially phosphorylated (P+) NF-M is present 24 h before the fully phosphorylated (P + + +) form. A slightly disappointing feature of the results is that there is rather little spatial or developmental distinction between the different de-
grees of NF-H and NF-M phosphorylation. There is no distinguishable gradient of P+ + + staining within the axon so it appears that NF-M and NF-H become fully phosphorylated soon after entering the axon. This leads Carden et al. ~ to suggest that some difference in conditions between cell body and axon, such as pH or ionic composition, is responsible for the apparent activation of neurofilamentdirected kinases within the axon. Little is known about the kinases that phosphorylate neurofilaments, but by means of cDNA sequencing, potential phosphate acceptor sites are beginning to be identified. At least some of these exist as tandemly repeated peptide sequences 7,s. Lee et al. 9 have raised antibodies to synthetic peptides based on one of these sequences, in either non-phosphorylated or chemically phosphorylated form. They also used the synthetic peptides to demonstrate that 96 out of their 515 anti-NF mAbs are specific for the phosphorylated or nonphosphorylated forms of this sequence, which thus constitutes a major immunogenic neurofilament epitope. These antibodies can now be used to study the occurrence and phosphorylation state of this particular phosphate acceptor sequence in neurofilaments. This approach has great potential, since it should enable a precise analysis of when and under what circumstances specific neurofilament phosphorylation events occur in the cell. What is the functional role of the NF-H and NF-M phosphorylation in axons? Whatever it is, it appears to be of fundamental importance since Cohen et al. 4 find that essentially the same phenomenon of phosphorylation linked to transport in the axon occurs in squid as
© 1988, ElsevierPublications, Cambridge 0378- 5912/88/$0200
Andrew Matus Friedrich MiescherInstitut, PO Box 2543, 4002 Basel, Switzerland.
291