The Effect of Metabolic Inhibitors On the Cockroach Nerve-Muscle Synapse

Recent studies involving vertebrate neuromuscular systems have indicated that the maintenance of the intrinsic ionic composition of nerve terminal axoplasm plays a critical role in determining the spatial distribution of the synaptic vesicles (Blioch, Glagoleva, Liberman & Nenashev, 1968; Heuser, Katz & Miledi, 1971) and that this ionic homeostasis may be metabolically maintained by the axon terminal mitochondria (Blioch et al. 1968; Lieberman, Palmer & Collins, 1967).

Rees & Usherwood (1972) observed that, following neurotomy in the locust retractor unguis muscle, agglutination of the synaptic vesicles invariably occurred in axon terminals, the mitochondria exhibiting a derangement of their cristal cyto-architecture. Rees (1971) and Rees & Usherwood (1972) have postulated that the agglutination of synaptic vesicles in degenerating axon terminals of locusts might be correlated with an increase in the axoplasmic levels of adenosine diphosphate (ADP) and/or calcium ions as a consequence of metabolic dysfunction in the degenerative mitochondria.

In the present study, metabolic inhibitors of known biochemical specificity have been applied to cockroach nerve-muscle synapses in an effort to elucidate, in terms of metabolic dysfunction, the phenomenon of synaptic vesicle agglutination and anomalous miniature potential activity (Usherwood, 1963) previously observed in degenerating axon terminals of the locust Schistocerca gregaria.

The nerve-muscle preparation used in this study was the metathoracic retractor unguis muscle of the cockroaches Periplaneta americana and Blaberus giganteus (Rees, 1974). The ipsi- and contralateral retractor unguis nerve-muscle preparations were dissected out and secured in adjacent troughs of a perfusion chamber. One of these preparations served as a ‘control’ in each experiment. The nerve-muscle preparations were perfused (0·5 ml/min) with their respective salines for a period of 2 h to allow the muscle fibres to attain ionic equilibration before obtaining electrical recordings. The ionic composition of the cockroach salines used in these experiments was as follows:

• P. americana: 15 IDM-K+; 190 mm-Na+; 4 mm-Ca²⁺; 213 mm-Cl⁻.

• B. giganteus: 10 mm-K+; 190 mm-Na⁴⁻; 4mm-Ca²+; 208 mm-Cl⁻.

For those investigations conducted on the retractor unguis of B. giganteus in experimental solutions at pH 7·1, 1 mm of N, N-bis(2-hydroxyethyl)-2-aminoethane-sulphonic acid (BES) (see Good et al. 1966) was added to saline (b) and adjusted to pH 7·1 with 0·2 N-NaOH.

Stock solutions of antimycin A₁ (Calbiochem. Ltd., Calif., U.S.A.) and A₃ (ICN Nutritional Biochemicals, Ohio, U.S.A.) were made up in ethanol at concentrations of 10⁻³ M and stored at −8°C. Stock solutions of antimycin A prepared in this manner and stored at 4°C are alleged to retain their biochemical activity for periods of up to 18 months (Lockwood, Leben & Keitt, 1954). Stock solutions of rotenone, oligomycin and 5-Cl-3-t-butyl-2′-Cl-4′-nitrosalicylanilide were similarly prepared in ethanol and stored at 6°C. The final concentrations of the metabolic inhibitors were prepared by dilution of the stock solution with the respective salines and freshly made up for each experiment. All experiments were conducted at 21°C.

The recording techniques employed in this study on the effects of metabolic inhibitors on cockroach nerve-muscle synapses are described elsewhere (Rees, 1974).

A relative measure of the miniature potential discharge frequencies recorded from ‘control’ (f_c) and ‘experimental’ (f_e) muscle fibres has been adopted to indicate the degree of change with time of the miniature potntial activity observed in muscle fibres following application of the metabolic inhibitor, antimycin A₃, and is referred to in the text as f_e/ f_c (see Fig. 2 a, b).

Muscle fibres which exhibited either high frequency (20–100 sec⁻¹) miniature potential discharges (see Rees, 1974) or low resting membrane potentials after the 2 h ionic equilibration period, were not used in this study.

The average number of events counted in each sample of miniature potentials was about 400 and the computed values expressed as the mean + S.D. of mean. The Student t test was used to test the significance between the means of the miniature potential discharge and accepted at the 1% significance level.

The fixatives and ‘wash’ solutions used in the preparation of control’ and ‘experimental’ retractor unguis muscles for electron microscopic examination were adjusted to an osmolarity and pH equivalent to that of the experimental saline with sucrose and 0·2 N-NaOH respectively. Fixation of ‘control’ or ‘experimental’ preparations in isosmotic 2% glutaraldehyde produced no noticeable increase in the miniature potential discharge rate. During the various phases of miniature potential activity (see Results), nerve-muscle preparations were prefixed in 2% glutaraldehyde at 8 °C for an initial fixation period of 10 min and then transected into proximal, medial and distal segments and fixed for a further 2 h in 2% glutaraldehyde at 6°C. The nerve and muscle segments were washed overnight and post-fixed in 1% osmic acid for 30 min at 18 °C. The tissues were then washed for 1 h in a sucrose ‘wash’ solution and dehydrated in 70%, 90% and 100% ethanol. Infiltration of the tissue with 10% MNA-TAAB resin (Taab Lab., Reading, England) was carried out overnight and the resin blocks polymerized at 60°C for 24 h (Rees & Usher wood, 1972). Ultrathin sections were cut on a Porter-Blum MT-2 ultratome and mounted on uncoated or coated grids. The sections were stained with alcoholic uranyl acetate (2–5 min) and lead citrate (1–4 min) and examined with a Phillips EM 300.

The resting membrane potential of ‘white’ and ‘red’ muscle fibres in the retractor unguis of P. americana was about 50·0 mV. Intracellular recordings from both types of muscle fibre indicated that the majority of miniature excitatory post-synaptic potentials (m-EPSPs) occurred as discrete synaptic events (see, for example, Fig. 1a). The mean rate of discharge of miniature potentials recorded from ‘red’ muscle fibres (t = 0·313 ±0·069 sec) was significantly lower (P < 0·001) than that recorded from ‘white’ fibres (t = 0·083 ±0·017 sec) in this insect skeletal muscle (Rees, 1974).

Electron microscopic examination of the multi-terminally innervated retractor unguis muscle in P. americana indicated that the nerve-muscle synapses possessed comparable ultrastructural features to those of other insect skeletal muscles (e.g. Atwood, Smyth & Johnston, 1969; Edwards, Ruska & De Harven, 1958; Rees & Usherwood, 1972). Consequently the cyto-architectural organization of nerve-muscle synapses on ‘white’ and ‘red’ muscle fibres in this insect muscle were similar except that axon terminals associated with the former were generally larger in diameter (i.e. 1·0–3·0μm) than those associated with the latter (ca. 1·0–1·6/mi). Furthermore, it appeared that the number of mitochondria/cross-sectional area, in axon terminals innervating ‘white’ muscle fibres, was greater than in those axon terminals associated with ‘red’ muscle fibre (cf. Fig. 3a, b, Plate 1). The density of synaptic vesicles in nerve-muscles synapses on these skeletal muscles was greatest near the synaptic sites (see Fig. 3 a, b, Plate 1, arrows). The mitochondria were invariably located centrally within the axon terminals and appeared in an ‘orthodox’ configuration: the cristae appearing maximally contracted, a characteristic of mitochondria fixed ‘in situ’ and in a ‘non-energized’ metabolic state (Hackenbrock, 1966, 1972; Williams et al. 1970).

Four phases of miniature potential activity were recorded from ‘white’ and ‘red’ muscle fibres in the retractor unguis of both P. americana and B. giganteus following the application of low concentrations of the respiratory inhibitor, antimycin A₃. A summary of the electrophysiological characteristics of these phases is presented in Table 1.

The initial phase of miniature potential activity occurred during the first 6090 min after the application of 10⁻¹¹ M antimycin A₃: the frequency of the m-EPSPs either increasing slightly (ca. 10%) or, more usually, decreasing to about 40% of the initial ‘control’ discharge rate (see Fig. 2 a). During phase I the miniature potentials tended to occur in small, summated bursts with their mean amplitude increasing from the ‘control’ value of 0·34±10·12 to 0·61 ±0·36 mV (Fig. 1b).

During phase II the frequency of the miniature potential discharge increased (t = 0·036 ± 0·009 sec) and was accompanied by an increase in the duration of the bursts of miniature potentials (Fig. 1 c, arrows). Phases I and II both occurred within the first 2 h following the application of 10⁻¹¹ M antimycin A₃ (see Fig. 2a). The resting membrane potentials of the ‘white’ muscle fibres remained relatively constant during this period of time.

Electron microscopic examination of nerve-muscle synapses on ‘white’ muscle fibres during the phases (I and II) of sporadic bursting showed that the majority of synaptic vesicles tended to exist in groups away from the synaptic sites (Fig. 4a, Plate 2). The terminal mitochondria during these first two phases were comparable in ultrastructural appearance to those observed in ‘control’ axon terminals. However, some mitochondria exhibited an ‘aggregated’ configuration which is characteristic of ‘de-energized’ mitochondria in the presence of antimycin (Hackenbrock, 1972; Williams et al. 1970).

A dramatic increase (t = 0·006 ± 0·001 sec) in the discharge rate of m-EPSPs occurred during phase III. The miniature potentials occurred in sustained discharges, with each discharge lasting from several hundred to several thousand milliseconds (Fig. 1 d). The miniature potentials during this phase tended to summate and form ‘composite’ potentials of between 2 and 4 mV in amplitude (see Fig. 1d, arrows). The mean amplitude of the m-EPSPs was 0·93 + 0·38 mV. The time of onset and duration of phase III varied from fibre to fibre (see Fig. 2a), but tended to subside more quickly in those muscle fibres which had initially displayed higher than average miniature potential discharge rates.

During the phase (III) of sustained miniature potential activity a dramatic change occurred in the spatial distribution of synaptic vesicles and ultrastructural morphology of the terminal mitochondria, in the majority of nerve-muscle synapses examined. Aggregate masses of synaptic vesicles appeared within the axon terminals and were predominantly located at the synaptic sites (Fig. 4b, Plate 2). The cyto-architectural organization of the terminal mitochondria appeared to be disrupted with subsequent derangement of the cristal membranes. The muscle mitochondria appeared unchanged and were comparable in ultrastructural appearance to those observed in ‘control’ muscle fibres (cf. Fig. 3 a, Plate 1; Fig. 4 b, Plate 2).

The ultimate phase (IV) of the effect of antimycin A₃ on the miniature potential activity became apparent upon termination of the sustained miniature potential discharge phase (III), the rate of miniature potential discharge decreasing gradually (i.e. 10–30 min) and then ceased completely, resulting in a failure in neuromuscular transmission. Prolonged perfusion (ca. 1 h) of antimycin A₃-inhibited muscle fibres, during phase IV, with inhibitor-free saline indicated that the failure in synaptic transmission was irreversible. At the time of failure of neuromuscular transmission the resting membrane potentials of the muscle fibres fell to about 50% of their initial ‘control’ values. The fibres at this time were inexcitable to neural stimulation, but were still responsive to direct (d.c.) stimulation applied with external electrodes. The insertion of one microelectrode into the proximal region of a muscle fibre and another into either the medial or distal region of the same muscle fibre showed that each phase of miniature potential activity invariably occurred simultaneously in each of these regions, indicating that the drug was simultaneously affecting all synaptic sites on this muscle fibre.

Electron microscopic examination of the retractor unguis muscle during phase IV indicated that there was a paucity of synaptic vesicles in the majority of the axon terminals examined, with the few remaining vesicles invariably found juxtaposed to the synaptic sites (see Fig. 4c, Plate 2). No apparent change was observed in the organization of the aposynaptic apparatus up to and including the time of neuromuscular transmission failure. The muscle mitochondria during phase IV appeared distended and in a ‘de-energized’ (aggregate) state.

Comparable phases of miniature potential activity and correlated ultrastructural changes to those reported above for ‘white’ muscle fibres were similarly observed in ‘red’ muscle fibres following application of antimycin A₃ (see Table 1). However, to obtain these phases within the same time periods (i.e. f_e/ f_c) as observed in ‘white’ muscle fibres required 10–100 times the concentration of antimycin A₃ (see Fig. 2b). The observed variability, with respect to the time of onset, of the four phases of miniature potential activity in ‘white’ and ‘red’ muscle fibres in P. americana appeared, upon electron microscopical examination, to be causally related to the size of the axon terminals, although other factors may also be involved. Ultrastructural changes in axon terminals associated with ‘white’ muscle fibres always occurred before those of the smaller axon terminals innervating ‘red’ muscle fibres. Furthermore, in proximal, medial or distal regions of the same muscle fibre in either ‘white’ or ‘red’ fibres, the larger diameter axon terminals usually exhibited morphological changes in preference to the smaller diameter axon terminals.

Electron microscopic examination of the peripheral axon branches (ca. 3–9 μ m diam.) innervating the retractor unguis muscle of P. americana (at the time of neuromuscular transmission failure) indicated that while the glial cells appeared disrupted and vacuolated, the morphology of the axon branches was relatively unchanged. Some of the mitochondria, however, appeared in a ‘condensed’ state, a characteristic of antimycin-treated mitochondria (Hackenbrock, 1972).

In three experiments the addition of 30 mm Mg²+ to 10⁻¹⁰ M antimycin A₃-saline perfusing of the nerve-muscle preparation did not prevent, or delay, the phase (III) of sustained miniature potential discharge (t = 0·009 + 0·003 sec) in ‘red’ muscle fibres (see Fig. 2b).

A recovery of neuromuscular transmission was achieved on three separate occasions following perfusion of the metabolically inhibited retractor unguis muscles with 10⁻¹⁰ M antimycin A₃ at pH 9·2. Approximately 20 min after the commencement of perfusion there was a resumption of the spontaneous miniature potential activity and EPSPs could be evoked by electrical stimulation of the retractor unguis motor axons. The resting membrane potentials of the muscle fibres then returned to their initial ‘control’ values. It has been reported that the degree of insensitivity of succinoxidase respiration in heart-muscle mitochondria to antimycin A progressively increased as the alkalinity of the substrate medium was increased from pH 7·2–9·1 (Estabrock, 1962).

Equimolar concentrations of antimycin A₁took longer to produce neuromuscular transmission failure than antimycin A₃ in the retractor unguis muscle of P. americana (see Miller & Rees, 1973). The respiratory inhibitor, rotenone, and oxidative phosphorylation uncouplers, oligomycin and 5-Cl-3-t-butyl-2′-Cl-4′-nitrosalicylanilide, produced comparable electrophysiological changes in ‘white’ and ‘red’ muscle fibres of the retractor unguis of P. americana, but required higher concentrations (i.e. 10⁻⁶ M) than antimycin A₁ or A₃.

The resting membrane potentials of ‘white’ and ‘red’ muscle fibres in the retractor unguis of Blaberus were approximately 60·0 mV in either unbuffered or buffered (pH 7·1) 10 mm-K+-saline. No significant differences were observed in the discharge rates of miniature potentials (m-EPSPs) from muscle fibres in unbuffered or buffered salines.

The mean interval between spontaneously occurring miniature potentials recorded intracellularly from ‘white’ muscle fibres (t = 0·35 ± 0·01sec) was significantly shorter (P <0·01) than that observed in ‘red’ muscle fibres (t = 0·54 + 0·03 sec) (Rees, 1974).

The effect of the aforementioned metabolic inhibitors on spontaneous m-EPSPs in the retractor unguis of B. giganteus was similar to that previously described for P. americana (see Table 1). However, the time of occurrence of phases I-IV was notably longer at equivalent concentrations of the metabolic inhibitors. The delay in the onset of the phases of miniature potential activity is depicted in Fig. 2 (b), where the effect of io⁻¹⁰ M anticycin A₃ on the relative frequency (f_e/ f_c.) of m-EPSPs from ‘red’ muscle fibres in P. americana and B. giganteus may be compared.

All four phases of miniature potential activity observed in muscle fibres of P. americana (Table 1) were similarly observed in the retractor unguis of B. giganteus, following the application of either unbuffered or buffered (pH 7·1) solutions of antimy-cin A₁ or A₃. Inhibition by antimycin A on the rate of oxygen uptake in heart-muscle mitochondria is reported to occur ten times faster at pH 7·2 than at pH 9·1 (Estabrock, 1962).

At the time of neuromuscular transmission failure in the retractor unguis of B. giganteus the muscle fibres became inexcitable to neural stimulation. These muscle fibres, like those observed in P. americana, were still mechanically responsive to direct (d.c.) stimulation with external electrodes.

The significantly higher rates of miniature potential discharges (Rees, 1974) and greater responsiveness of axon terminals on ‘white’ muscle fibres to metabolic inhibitors is suggestive of differences in basal metabolism between axon terminals on ‘white’ and ‘red’ muscle fibres in P. americana and B. giganteus. Usherwood & Machili (1968) reported that ‘white’ fibres fatigued quicker than ‘red’ fibres, in the retractor unguis muscle of P. americana and B. giganteus, upon repetitive neural stimulation.

Electron microscopic examination of nerve-muscle synapses in the retractor unguis of P. americana has indicated that axon terminals associated with ‘white’ muscle fibres were larger in diameter and contained a greater density of mitochondria than those associated with ‘red’ muscle fibres. In the extensor tibiae muscles of P. americana the ‘fast ‘axon terminals were found to be significantly larger in diameter and to possess more synaptic ‘sites’ per terminal than the ‘slow’ axon terminals (Atwood et al. 1969). McArdle & Albuquerque (1973) have reported that the mean frequency of miniature potentials in the ‘fast’ extensor muscle was significantly greater than that observed in the ‘slow’ soleus muscle of the rat. These authors suggested that the lower miniature potential discharge of the soleus was causally related to the smaller size of the neuromuscular junctions observed on this ‘slow’ muscle.

At crustacean (Sherman & Atwood, 1972) and vertebrate neuromuscular junctions (Kuno, Turkanis & Weakly, 1971) positive correlations have been obtained between the size of the neuromuscular junction, quantal content and the frequency of miniature potentials.

Treatment of cockroach axon terminals with metabolic inhibitors induced comparable electrophysiological and ultrastructural changes to those observed in axon terminals on insect (Rees & Usherwood, 1972) and vertebrate muscles (Birks, Katz & Miledi, 1960; Mildei & Slater, 1968, 1970) following neurotomy. Of interest in the present investigation was the observation that agglutination of synaptic vesicles occurred only in those axon terminals in which the mitochondria appeared in a metabolically ‘uncoupled’ state. This observation is analogous to that observed in degenerating axon terminals of the locust (Rees & Usherwood, 1972). The changes in cyto-architectural organization observed in cockroach axon terminals, following treatment with metabolic inhibitors, have similarly been observed in vertebrate synapses following the application of a variety of chemical agents. Agglutination of synaptic vesicles and disruption of the mitochondrial cyto-architecture in vertebrate axon terminals has been observed following application of β-bungarotoxin (Chen & Lee, 1972) and black widow spider venom (Clark, Hurlbut & Mauro, 1972). Administration of black-widow spider venom to insect nerve-muscle synapses is reported to cause comparable electrophysiological and ultrastructural effects to those at vertebrate neuromuscular junctions (Cull-Candy, Neal & Usherwood, 1973). It is of interest to note that in the studies cited above a disruption of the cyto-architecture of the mitochondria was a predominant feature of the axon terminals.

The passage of electrons through the electron transfer system in mitochondria results in the generation of intramitochondrial ATP and its subsequent hydrolysis facilitates the phosphorylation of extramitochondrial ADP (Green & Baum, 1970).

Respiratory inhibitors interrupt this flow of electrons and cause a decrease in the ATP/ADP ratio. Antimycin A₁ and A₃ are extremely potent respiratory inhibitors (Rieske, 1967) and their site of action in both insect (Gilmour, 1961) and vertebrate mitochondria (Harold, 1972) is thought to be localized between cytochromes b and q in complex III of the electron transfer chain. The net result of oxidative phosphorylation uncoupling by both oligomycin and the nitrosalicylanilide analogue (Williamson & Metcalf, 1967) is to inhibit phosphorylation of the intermediate adenine nucleotides (e.g. AMP, ADP), so as to induce an increase in the extramitochondrial levels of inorganic phosphate and ADP (Hackenbrock, Rehn, Weinbach & Lemaster, 1971).

The energy derived from transduction processes in the mitochondrion is used to translocate Ca²⁺ into the matrix, a process which takes precedence over oxidative phosphorylation for all available respiratory energy (Lehninger, 1970). The energy-linked processes of calcium accumulation (‘matrix loading’) and binding of calcium to the inner mitochondrial membranes (‘membrane loading’) are extremely sensitive to respiratory and oxidative phosphorylation inhibitors. Mitochondria loaded with calcium ions exhibit an immediate discharge of membrane-bound Ca²⁺, in the presence of metabolic inhibitors and absence of a permanent anion (Lehninger, 1972). It is postulated that under conditions of metabolic inhibition there will be an increase in the intracellular levels of Ca²+ and ADP, with a concomitant decrease in ATP. It has been postulated that the maintenance of the intrinsic ionic composition of nerve terminal axoplasm plays a critical role in determining the spatial distribution of the synaptic vesicles at vertebrate axosomatic synapses (Blioch et al. 1968). This ionic homeostasis may be metabolically maintained by the axon terminal mitochondria (Blioch et al. 1968; Lieberman et al. 1967).

In this study it was observed that the agglutination of synaptic vesicles in cockroach axon terminals coincided with a metabolic dysfunction of the terminal mitochondria, as evidenced by their anomalous ultrastructural configurations. It is suggested that the agglutination phenomenon may be induced by increased intra-terminal levels of ADP and/or Ca²⁺. Indirect evidence is available to support the hypothesis that increased levels of intra-terminal Ca²⁺ could cause a decrease in the number of negative charges around synaptic vesicle membranes and so facilitate the adherence of the vesicles in the metabolically inhibited axon terminals of the cockroach. The progressive increase in spontaneous miniature potential activity (i.e. phases I—III), in metabolically inhibited cockroach muscle fibres, may be caused by a gradual increase in intra-terminal Ca²⁺ levels. This would increase the degree of cohesion between vesicles and presynaptic membranes, thereby facilitating increased transmitter release. Blioch et al. (1968) have similarly reported an increase in the miniature potential discharges from frog neuromuscular junctions, following administration of uncouplers of oxidative phosphorylation, and have similarly inferred that this effect was caused by increased levels of intra-terminal Ca²⁺. The dramatic increase in the frequency of miniature potentials observed in cockroach nerve-muscle synapses, following the application of metabolic inhibitors, has in addition been reported to occur at vertebrate neuromuscular junctions following application of metabolic inhibitors (Blioch et al. 1968; Katz & Edwards, 1973), isotonic Ca²⁺-salines (Heuser et al. 1971; Katz & Miledi, 1969) and under hypoxic conditions (Hubbard & Løyning, 1966).

The current ‘co-operative’ hypothesis of transmitter release postulates that release site’ with four Ca²⁺ attached is the most efficient Ca²⁺ complex in facilitating transmitter release (Cooke, Okamoto & Quastel, 1973; Dodge & Rahamin-off, 1967; Rahaminoff, 1968). It is postulated that in the presence of increased levels of intra-terminal Ca²⁺, the probability of all ‘release sites’ having a full co-operative number of Ca²⁺ will be increased. Under such conditions it is envisaged that the probability of spontaneous transmitter release will similarly be increased. Llinás, Blinks & Nicholson (1972) have, indeed, demonstrated (using a bioluminescent protein sensitive to calcium) that a transient increase of intra-terminal Ca²⁺ does occur during synaptic transmission in the squid ‘giant’ synapse.

The increased levels of intra-terminal ADP, like that of Ca²⁺, might also be involved in the agglutination of the synaptic vesicles in metabolically inhibited cockroach axon terminals. It has been reported that the addition of exogenous ADP to isolated cell systems (e.g. chick embryo fibroblasts (Jones, 1966) and mammalian platelets (Gaader et al. 1961; Mason et al. 1972)) caused these cells to form aggregate masses.

From the present study and the literature cited it would appear that the terminal mitochondria play a fundamental role in maintaining the intrinsic ionic composition of both vertebrate and invertebrate nerve-muscle synapses. This ionic homeostasis appears to be a basic necessity for maintaining optimal charge distributions on the phospholipid membranes of both the synaptic vesicles and the axon terminal in the immediate vicinity of the synaptic sites.

The effectiveness with which the various metabolic inhibitors induced changes in both the electrophysiological and ultrastructural characteristics of cockroach nervemuscle synapses supports the hypothesis that the agglutination of synaptic vesicles in degenerating axon terminals on the locust retractor unguis muscle may be causally related to a metabolic dysfunction of the terminal mitochondria.

I wish to thank Professors T. A. Miller and T. R. Fukuto for enabling me to conduct this study. My thanks also to Professors T. A. Miller, R. Olsen, J. A. Simpson and W. W. Thompson for their discussions on the content and presentation of this study. The technical assistance of K. Platt and M. Adams was most appreciated. My grateful thanks to C. Sprehn and K. Atherton for typing this manuscript.