Paramecium trichocysts isolated with their membranes are stable in the presence of millimolar Ca 2+

OSCAR LIMA*, TADEUSZ GULIK-KRZYWICKI and LINDA SPERLINGf Centre de Cenetique Mole'ctiktire, Associated with the Universite Pierre et Marie Curie, CNRS, Gif-sur-Yvette, 91190 France •Present address: Laboratoire de Ge'ne'tique Physiologique, University Paris XI, Batiment 400, 91405 Orsay Cedex, France t Author for correspondence

Summary We have developed a simple and rapid procedure for the isolation of a pure fraction of Paramecium trichocysts (mature secretory vesicles) with their membranes. Since in wild-type Paramecium cells essentially all trichocysts are docked at pre-formed cortical sites, trichocysts were isolated from cells in which functional trichocysts remain free in the cytoplasm owing to a mutation, tam6, that affects the docking site. Examination of the preparations by freeze-fracture electron microscopy confirms the presence of the membranes. The distribution of particles in the membranes of the isolated trichocysts and in the membranes of wild-type trichocysts in situ are nearly identical and this argues against any rearrangement of the membranes during the isolation procedure. Although the trichocyst matrix

Introduction The description of the secretory pathway first established by Palade (1975) is perfectly relevant to lower eucaryotic organisms that provide model systems for secretion. In yeast, some 25 genes necessary for secretion have been identified by genetic analysis (Novick et al. 1980; Schekman, 1985) and by now a number of these genes have been cloned and the corresponding proteins at least partly characterized (e.g. see Segev et al. 1988; Nakano et al. 1988). Genetic analysis of secretion in Paramecium, whose mature secretory vesicles are called 'trichocysts' because of their distinctive morphology, has led to the identification of more than 30 genes necessary for secretion (cf. Adoutte, 1988, for a review). Although molecular genetics has been slower to come of age in Paramecium than in yeast, several aspects of secretion in Paramecium merit attention, in particular, it is possible to study the final step, exocytosis. Exocytosis is a universal but poorly understood process by which intracellular vesicles fuse with the plasma membrane, usually to deliver proteins to Journal of Cell Science 93, 557-564 (1989) Printed in Great Britain © The Company of Biologists Limited 1989

undergoes a dramatic structural transition in the , presence of Ca2+ and water (matrix expansion), the isolated vesicles with intact membranes are perfectly stable in the presence of millimolar free Ca z+ . This result supports a chronology in which the first step in exocytosis is membrane fusion, the swelling of vesicle contents occurring only afterwards, once the contents come into contact with the water and Ca2+ of the external medium. The role of swelling would then be to help disperse, propel or otherwise empty the contents of the vesicle outside the cell.

Key words: Paramecium, trichocysts, secretory vesicle membranes, exocytosis.

the plasma membrane or vesicle contents to the extracellular space. Secretion in yeast is constitutive so that it is not possible to separate vesicle transport or interaction with the plasma membrane from exocytosis (Burgess & Kelly, 1987). In Paramecium, however, secretion is regulated: an external stimulus triggers exocytosis of secretory vesicles already docked at pre-formed cortical sites (Planner et al. 1973; Pollack, 1974). As is the case for cortical granule exocytosis, which occurs upon activation of animal oocytes (Anderson, 1968; Kline, 1988), it is possible to trigger massive, synchronous exocytosis in Paramecium (Plattner, 1987) and a recent study presents evidence that the biological function of trichocyst exocytosis is to provide Paramecium with a defence against predators, e.g. in this study Dileptus margaritifer, a carnivorous ciliate (Harumoto & Miyake, personal communication). One striking feature of secretion in Paramecium is the elaborate architecture of the vesicle and its contents (Bannister, 1972; Hausmann, 1978). A general property of all secretory vesicles is that their contents swell. Indeed, a current hypothesis for the mechanism of 557

exocytosis is that granule swelling provides the driving force for membrane fusion (Finkelstein et al. 1986). The contents of trichocysts are crystalline and swelling is a dramatic event as it consists of a rapid, cooperative structural transition from a compact crystalline form to a second expanded form that is also crystalline (Sperling et al. 1987). The transition requires Ca 2+ and, of course, involves uptake of water by the structure. Although trichocyst crystalline contents have been purified in the past by several strategies (Steers et al. 1969; Anderer & Hausmann, 1977; Matt et al. 1978; Garofalo & Satir, 1984; Sperling et al. 1987), we present for the first time a procedure for the isolation of a pure fraction of Paramecium secretory vesicles consisting of both crystalline contents and limiting membrane. Our primary motivation in isolating intact vesicles was to develop a biochemical approach to the characterization of the products of the genes involved in the different steps of the secretory pathway. A number of these gene products have been attributed to the trichocyst compartment by microinjection experiments (Aufderheide, 1978; LefortTran et al. 1981) and are likely to be proteins of the vesicle membrane. Our initial characterization of the isolated vesicles shows that the membranes are present and intact. The distribution of intramembrane particles revealed by freeze-fracture electron microscopy is the same in the membranes of the isolated vesicles and in the membranes of trichocysts in situ, in rapidly frozen, unfixed Paramecium cells. Most significantly, the isolated vesicles are perfectly stable in the presence of millimolar free Ca + (micromolar Ca 2+ is sufficient for exocytosis), which argues against the 'osmotic hypothesis' that granule swelling provides the driving force for membrane fusion: the Ca 2+ necessary for expansion of the trichocyst matrix must come from the external medium, once the vesicle membrane has fused with the plasma membrane. Materials and methods Cells and culture conditions Wild-type Paramecium tetraurelia cells were of stock d4-2 (Sonneborn, 1974). tam6 mutant cells were originally isolated from d4-2 stock after nitrosoguanidine mutagenesis (Beisson & Rossignol, 1975). Cultures were grown at 27°C in an infusion of Wheat Grass Powder (Pines International, Lawrence, Kansas), infected with Klebsiella pneumoniae and supplemented with /3sitosterol (4f(gml~'), according to the standard procedures (Sonneborn, 1970). Isolation of trichocysts A 51 sample of tam6 culture, consisting of 30004000 cells ml" , was collected by continuous flow centrifugation and then by centrifugation in pear-shaped bottles in an oiltesting centrifuge at 300 g\ The cell pellet was washed twice at room temperature with 200 ml of 10 mM-Tris-HCl, pH 7-0, and then once with 45 ml of ice-cold 'Buffer A', which was adapted from a buffer developed for physiological studies of Paramecium mitochondria (Doussiere et al. 1979) and contains 50mM-Hepes, pH7-0, 0-25 M-sucrose, 0-5% bovine serum albumin (BSA) (Miles Laboratories), 1 mM-EGTA, 1 mMEDTA, 50,UM-PMSF (phenylmethylsulphonyl fluoride) and 558

O. Lima et al.

5f«M-leupeptin. The final pellet was resuspended to 3 ml total volume with ice-cold buffer A, transferred to a tight-fitting Dounce (teflon-glass) homogenizer and homogenized on ice with 20 strokes of the piston. The homogenate was then immediately layered over 60% Percoll (Pharmacia, Inc.)Buffer A in 10-ml centrifuge tubes, 0-5 ml of cell homogenate per tube, and centrifuged in a Beckman Ti50 rotor at 27500revsmin- 1 (50000g) for 15min at 4°C. The trichocyst bands, centered about 20 mm from the bottom of the centrifuge tube, were recovered with a Pasteur pipette that had been rinsed in Buffer A, and the Percoll partially removed by dilution with Buffer A and low-speed (5000 revs min" 1 ) centrifugation in the SS34 rotor of a Sorvall centrifuge to collect the trichocysts. Such centrifuge pellets were used directly for freeze-fracture electron microscopy sample preparation or were first transferred to an Eppendorf tube and recentrifuged at higher speed in a microfuge. We note that after the density gradient step, the trichocysts were much more stable at room temperature than at lower temperatures. In fact, if the trichocysts are stored on ice, 100 % transition to the extended form, with concomitant loss of membranes, is observed within minutes. Therefore, after the gradient step, further manipulations were carried out at 15-20°C. Determination of the density of trichocysts and mitochondria. Density marker beads (Pharmacia, Inc.) were used as suggested by the manufacturer to calibrate the self-forming Percoll gradients. Densities of trichocysts and of mitochondria were determined after centrifugation on five gradients containing Percoll concentrations ranging from 50% to 70%. Although optimal separation was found at 60 %, the positions of the bands could be measured at all of these Percoll concentrations and the apparent buoyant densities of the organelles determined with great precision (see Results). Phase-contrast light microscopy Light microscopy was carried out under phase-contrast optics and images were recorded on Kodak TMAX 400 film developed according to the manufacturer's instructions. Freeze-fracture electron microscopy The pellets of Paramecium cells or of isolated trichocysts were frozen without any pretreatment using our 'sandwich' technique (Aggergbeck & Gulik-Krzywicki, 1986). A very thin layer of the sample is spread on a thin flat copper plate and immediately covered with an identical plate. This sandwich is then rapidly plunged into liquid propane. The opening of these two plates at — 125CC, under a vacuum of about 10~ 7 Torr (lTorr = 133-3 Pa) in a freeze-fracture unit (Balzers BAF301) produces fractures in the frozen sample. Replication of the fractured surfaces was performed using platinum-carbon. The replicas were cleaned in chromic acid, washed with distilled water, and observed in a Philips 301 electron microscope.

Results Isolation of trichocysts with their membranes The isolation of intact secretory vesicles from Paramecium is a difficult task because of the dynamic properties of the vesicle contents, which are metastable protein crystals. Under a variety of conditions and particularly in the presence of Ca + , the crystals undergo a dramatic, irreversible structural transition to an extended, extra-

cellular form some eight times longer than the intracellular form of the crystal and incompatible with the presence of a limiting membrane designed to enclose a much smaller object. Conditions that permitted isolation of the trichocyst crystalline contents 'blocked' in the compact intravesicular form (detergent disruption of the cell and high-sucrose, high-EGTA, high-Mg 2+ buffers) are not suitable for membrane stability. A further obstacle in the isolation of trichocysts with intact membranes is that essentially all of the trichocysts in wild-type Paramecium cells are docked at fixed cortical sites in a 'pre-fusion' state. Removal of the trichocysts from their attachment sites without triggering exocytosis and without disrupting the vesicle membranes, is difficult if not impossible (Anderer & Hausmann, 1977). Our strategy therefore involved the use of cells from the Paramecium mutant, tam6 (Beisson & Rossignol, 1975). The trichocysts of tam6 cells are not attached at the cortex as in wild-type cells but are free in the

cytoplasm. Yet tani6 trichocysts are functional when microinjected into cells with normal cortex, while trichocysts from wild-type cells remain unattached when injected into tam6 cells (Lefort-Tran et al. 1981). The conclusion of the microinjection analysis of tani6 mutant cells is that the site of the mutation is not the trichocyst compartment but the cortex compartment. Paramecium cells gently homogenized in a buffer designed to stabilize membranes (see Materials and methods) are layered directly onto Percoll (60 %) equilibrated with the same buffer, and the isosmotic gradients formed by 15 min centrifugation at 50000 £ provide excellent separation of trichocysts from all other cellular components and in particular from the mitochondria, whose density is close to that of trichocysts. We measured buoyant densities of l-105gcm~ ± f>004 and l-124gcm~ 3 ± 0-002 for mitochondria and trichocysts, respectively. Fig. 1 is a low-magnification freeze-fracture image of a

Fig. 1. Low-magnification (X9000) freeze-fracture electron micrograph of a centrifuge pellet of isolated trichocysts. Note the homogeneity of the preparation and the absence of material other than trichocysts, with the exception of the electron-dense Percoll, which sticks to the replicas. In the inset, a higher-magnification view (X27 000) of a region of the micrograph showing characteristic EF and PF fracture faces of trichocyst tip membranes.

Trichocysts with their membranes

559

centrifuge pellet of a preparation of isolated trichocysts. Essentially all the trichocysts have intact membranes and very little material other than trichocysts is found. (The electron-dense debris is residual Percoll: 30 nm particles of silica coated with polyvinylchloride.) A trichocyst is composed of two differentiated regions, the body and the tip. The body consists of the carrot-shaped crystalline contents; the tip has a crystalline core of the same structure as the body, which is covered with two layers of fibrous material, the inner and outer sheath (for the anatomy of a trichocyst, see Bannister, 1972). The inset in Fig. 1 shows a higher-magnification view of the tip region of two trichocysts. The inner (EF) face of the membrane (trichocyst at the right of the inset) is in contact with the outer sheath of the trichocyst tip and the striations probably reflect a special arrangement of the membrane phospholipids induced by interactions with the outer sheath; this suggests that the outer sheath may be composed of filaments in a long-pitch helical arrangement. The outer (PF) face of the membrane covering the trichocyst tip (trichocyst at the left of the inset) displays many intramembrane particles whose arrangment is partly ordered. The geometrical arrangment of these particles may play a role in the assembly of the collar and perhaps other fibrous elements that link the trichocyst tip to the cortex (Pouphile et al. 1986). Asymmetric particle distribution in trichocyst membranes Allen & Hausmann (1976) first reported an asymmetric particle distribution in trichocyst membranes, which they observed by freeze-fracture of glutaraldehyde-fixed Paramedum caudatum cells. As illustrated in Fig. 2, we observe a highly asymmetric particle distribution in the . membranes of the unfixed isolated tam6 trichocysts (Fig. 2A) as well as in the membranes of wild-type trichocysts (Fig. 2B) and tam6 trichocysts (not shown) frozen in situ. In order to compare the membranes of the isolated tam6 trichocysts with the membranes of tam6 trichocysts frozen in situ, we counted the number of intramembrane particles per unit area and determined the ratio of the density of particles in the PF face with respect to the density of particles in the EF face. Particles were counted irrespective of their size; the values obtained are given in Table 1. Values for wild-type trichocysts, although based on a smaller statistical sample, are Table 1. Density of intramembrane particles in trichocyst membranes Pface

E face

PF/EF ratio

1959 ± 334 (6) 1941 ± 358 (6) 1822 ± 163 (3)

104 + 29(6) 109 ± 2 8 (8) 105 ± 1 3 (3)

18-8 17-8 17-3

Trichocysts ta>n6, isolated tani6, in situ Wild type, in situ

The values represent the average number of intramembrane particles per fim ± S.D. Particles were counted in equivalent, flat areas (0-30;«n2) of the membranes of different trichocysts to give the number of particles per unit for each given trichocyst. Average values and standard deviations were then calculated and the number of independent observations (i.e. different trichocysts examined) is given in parentheses.

560

O. Lima et al.

also given. The particle densities are very similar in the different preparations and, in each case, a particle ratio of about 18 in favour of the cytoplasmic (PF) face is found. This argues against rearrangement of the membranes during the isolation procedure. The vesicles are stable in millimolar Ca2+ Although the trichocyst membranes in our preparations appear intact and 'native' by ultrastructural criteria, we also sought a biochemical test of their integrity. As the trichocyst crystalline matrix undergoes a structural transition in the presence of Ca 2 + , which is easily observable by light microscopy (Matt et al. 1978; Garofalo & Satir, 1984), we devised a 'Ca 2+ test' of membrane integrity. As shown in Fig. 3, the isolated trichocysts remain condensed in the presence of either 1 mM free Ca 2+ or 0-1 % Triton X-100, a non-ionic detergent that dissolves the vesicle membrane. Only in the presence of both Ca 2+ and detergent is 'discharge' of all trichocysts observed. In other experiments, we raised the free Ca 2+ above 10 mM and found the trichocysts to be still intact in the absence of detergent. The preparations of isolated trichocysts are stable for several hours at room temperature in the presence of even 10mM-Ca2+. Most important, the contents retain their capacity to undergo the structural transition for, at any time, the addition of detergent leads to immediate expansion of the trichocyst matrices, provided there is Ca 2+ in the buffer. These experiments show that: (1) the membranes of the isolated vesicles are perfectly intact and they are impermeable to Ca 2+ (but permeable to water: the membranes burst when the osmolality of the buffer is reduced (data not shown)); (2) the membranes are not necessary for the crystalline matrix to remain in the compact state; (3) Ca.2+ IS necessary for the transition to the extended state. Discussion We have shown that it is possible to isolate Paramecium trichocysts with their membranes by a simple and rapid procedure. We used tam6 mutant cells whose functional trichocysts are free in the cytoplasm, and self-forming Percoll buoyant density gradients, which permit rapid, high-resolution separation of organelles under isosmotic conditions. The preparations obtained are very pure, the only contaminant being a few mitochondria, which can be largely eliminated either by reducing the amount of material charged on the gradients or by adding a second gradient purification step. The preparations are thus suitable for biochemistry, and extension of biochemical studies to the non-discharge mutants (Cohen & Beisson, 1980) should be possible after construction of the appropriate tam6 X nd double mutant strains. The trichocyst preparations are stable for hours at room temperature, as judged by the Ca 2+ test and their appearance in the light microscope, opening the possibility of physiological studies of the membrane properties of the isolated vesicles. Exocytosis The fact that the isolated trichocysts are perfectly stable

Fig. 2. Freeze-fracture electron micrographs of EF and PF faces of isolated tam6 trichocysts (A) and of wild-type trichocysts in situ (B) (magnification X60000). Note the highly asymmetric distribution of particles between the two faces in both A and B. (See Table 1 for quantification of the particle densities statistically based on examination of a number of different trichocysts.) The choice of these particular images was dictated by the proximity, in the same field, of both fracture faces of trichocysts in a parallel (or antiparallel) arrangement. Bar, 0-5 [im.

Trichocysts with their membranes

561

Ca 2 +

\

o o

\

I

X c

o

""
en. Genet. 165, 199-205. BANNISTER, L. H. (1972). Structure of trichocysts in Paramecium

caudatum.J. Cell Sci. 11, 899-929. BEISSON, J. & ROSSIGNOL, M. (1975). Movements and positioning of organelles in Paramecium aurelia. In Molecular Biology of Nucleocytoplasmic Relationships (ed. S. Puiseaux-Dao), pp. 291-294. North-Holland, Amsterdam: Elsevier. BILINSKI, M., PLATTNER, H. & MATT, H. (1981). Secretory protein

decondensation as a distinct, Ca 2+ -mediated event during the final steps of exocytosis in Paramecium cells. J. Cell Biol. 88, 179-188. BRECKENRIDGE, L. J. & ALMERS, W. (1987). Final steps in exocytosis observed in a cell with giant secretory granules. Pivc. natn. Acad. Sci. U.S.A. 84, 1945-1949. BURGESS, T . L. & KELLY, R. B. (1987). Constitutive and regulated secretion of proteins. A. Rev. Cell Biol. 3, 243—293. COHEN, F. S., AKABAS, M. H. & FINKELSTEIN, A. (1982). Osmotic

swelling of phospholipid vesicles causes them to fuse with a planar phosopholipid bilayer membrane. Science 217, 458-460. COHEN, J. & BEISSON, J. (1980). Genetic analysis of the relationships between the cell surface and the nuclei in Paramecium tetraurelia. Genetics 95, 797-818. COHEN, F., ZIMMERBERG, J. & FINKELSTEIN, A. (1980). Fusion of

phospholipid vesicles with planar phospholipid bilayer membranes. II. Incorporation of a vesicular membrane marker into a planar membrane. J. gen. Physiol. 75, 251-270. CRABB, J. H. & JACKSON, R. C. (1985). In vitro reconstitution of exocytosis from plasma meembrane and isolated secretory vesicles. J.CellBiol. 101,2263-2273. DOUSSIERE, J., SAINSARD-CHANET, A. & VIGNAIS, P. (1979). The

respiratory chain of Paramecium tetraurelia in wild type and the mutant c\\. Biochim. biophys. Ada 11, 224-235. FINKELSTEIN, A., ZIMMERBERG, J. & COHEN, F. S. (1986). Osmotic

swelling of vesicles: its role in the fusion of vesicles with planar phospholipid bilayer membranes and its possible role in exocytosis. A. Rev. Physiol. 48, 163-174. GAROFALO, R. S. & SATIR, B. H. (1984). Paramecium secretory granule content; quantitative studies on in vitro expansion and its regulation by calcium and pH. jf. Cell Biol. 99, 2193-2199. GREEN, D. P. L. (1987). Granule swelling and membrane fusion in exocytosis. J. Cell Sci. 88, 547-549. HAUSMANN, K. (1978). Extrusive organelles in protists. Int. Rev. Cytol. 52, 198-276. KERSKEN, H., MOMAYEZI, M., BRAUN, C. & PLATTNER, H. (1986).

Filamentous actin in Paramecium cells: functional and structural changes correlated with phalloidin affinity labeling in vivo. j ' . Histochem. Cytochem. 34, 455-465. KLINE, D. (1988). Calcium-dependant events at fertilization of the frog egg: injection of a calcium buffer blocks ion channel opening, exocytosis, and formation of pronuclei. Devi Biol. 126, 346-361. LEFORT-TRAN, M., AUFDERHEIDE, K., POUPHILE, M., ROSSIGNOL,

M. & BEISSON, J. (1981). Control of exocytotic processes: cytological and physiological studies of trichocyst mutants in Paramecium tetraurelia. J. Cell Biol. 88, 301-311. MATT, H., BILINSKI, M. & PLATTNER, H. (1978). Adenosine

Trichocysts with their membranes

563

triphosphate, calcium and temperature requirements for the final steps of exocytosis in Paramecium cells. J. Cell Sci. 32, 67-86. NAKANO, A., BRADA, D. & SCHEKMAN, R. (1988). A membrane

glycoprotein, secl2p, required for protein transport from the endoplasmic reticulum to the golgi apparatus in yeast. J. Cell Biol. 107, 851-863. NOVICK, P., FIELD, C. & SCHEKMAN, R. (1980). Identification of 23

complementation groups required for post-translational events in the yeast secretory pathway. Cell 21, 205-215. PALADE, H. (1975). Intracellular aspects of the process of protein synthesis. Science 189, 347-358. PLATTNER, H. (1987). Synchronous exocytosis in Paramecium cells. In Cell Fusion (ed. A. E. Sowers), pp. 69-98. New York, London: Plenum. PLATTNER, H., MILLER, F. & BACHMANN, L. (1973). Membrane

specializations in the form of regular membrane-to-membrane attachment sites in Paramecium. A correlated freeze-etching and ultrathin-sectioning analysis. J. Cell Sci. 13, 687-719. POLLACK, S. (1974). Mutations affecting the trichocysts in Paramecium aurelia. 1. Morphology and description of the 11111131118.7. Protozool. 21, 352-362. POUPHILE, M., LEFORT-TRAN, M., PLATTNER, H., ROSSIGNOL, M. &

BEISSON, J. (1986). Genetic dissection of the morphogenesis of exocytosis sites in Paramecium. Biol. Cell 56, 151-162. SATIR, B. H., BUSCH, G., VUOSO, A. & MURTAUGH, T . J. (1988).

Aspects of signal transduction in stimulus exocytosis-coupling in Paramecium. J. cell. Biochem. 36, 429-443. SCHEKMAN, R. (1985). Protein localization and membrane traffic in yeast. A. Rev. Cell Biol. 1, 115-143. SEGEV, N., MULHOLLAND, J. & BOTSTEIN, D. (1988). The yeast

564

O. Lima et al.

GTP-binding YPT1 protein and a mammalian counterpart are associated with the secretion machinery. Cell 52, 915-924. SONNEBORN, T. M. (1970). Methods in Paramecium research. Meth. CellPhysiol. 4, 241-339. SONNEBORN, T. M. (1974). Paramecium tetraurelia. In Handbook of Genetics, vol. 2 (ed. R. C. King), pp. 469-594. New York: Plenum Press. SPERLING, L., TARDIEU, A. & GULIK-KRZYWICKI, T. (1987). The

crystal lattice of Paramecium trichocysts before and after exocytosis by X-ray diffraction and freeze-fracture electron microscopy. J. Cell Biol. 105, 1649-1662. STEERS, E., BEISSON, J. & MARCHESI, J. T. (1969). A structural

protein extracted from the trichocyst of Paramecium aurelia. Expl Cell Res. 57, 392-396. WHITAKER, M. & ZIMMERBERG, J. (1987). Inhibition of secretory granule discharge during exocytosis in sea urchin eggs by polymer solutions. J. Physiol. 389, 527-539. WHITE, J., KIELIAN, M. & HELENIUS, A. (1983). Membrane fusion

proteins of enveloped animal viruses. Q. Rev. Biophvs. 16, 151-195. ZIMMERBERG, J., CURRAN, M., COHEN, F. S. & BRODWICK, M.

(1987). Simultaneous electrical and optical measurements show that membrane fusion precedes secretory granule swelling during exocytosis of beige mouse mast cells. Proc. natn. Acad. Sci. U.S.A. 84, 1585-1589. ZIMMERBERG, J. & Liu, J. (1988). Ionic and permeability requirements for exocytosis in vitro in sea urchin eggs. J. Memhr. Biol. 101, 199-207. {Received 16 January 1989 -Accepted 14 April 1989)