Reflexes Mediated by Non-Impulsive Afferent Neurones of Thoracic-Coxal Muscle Receptor Organs in the Crab, Carcinus Maenas: II. Reflex Discharge Evoked by Current Injection

In the jointed crustacean limb, proprioceptors are associated with all joints and provide the central nervous sytem with information concerning the position of limb segments, the rate and direction of joint movements, and, in the case of tension receptors (Macmillan, 1976, for review), information on the muscular forces produced during movements is also provided. It is generally accepted that the central nervous system, in the absence of sensory feedback, can generate the cyclical patterns of neuronal activity that form the basis of locomotion (Kennedy & Davis, 1977). Nevertheless, proprioceptive feedback is known to modulate motoneurone (MN) output in several arthropod locomotory systems (e.g. Wilson & Gettrup, 1963; Davis, 1969; Pearson, 1972; Paul, 1976). Resistance reflexes are an important consequence of proprioceptor activation, notwithstanding the fact that, opposing as they do an ongoing locomotory cycle, they are evidently suppressed during active movements (Barnes, Spirito & Evoy, 1972; Field, 1974; Vedel & Clarac, 1975). It is arguable, therefore, that resistance reflexes in the Crustacea are functionally restricted to controlling posture (Fields, 1966; Sokolove, 1973; Macmillan, 1975); to dampening joint movement at the extremes of joint angle during walking (Evoy & Cohen, 1971 ; Fourtner & Evoy, 1973), and to facilitating the initiation of an antagonistic component of a movement cycle as proposed for the lobster swimmeret system (Davis, 1969); to load compensation during unexpected deviations in locomotory movements (Barnes et al. 1972); and finally, in the case of ‘distributed’ (intersegmental) reflex effects, such reflexes can be seen as contributing to the coordination of the activity of muscles having an essentially synergistic function in limb walking movements (Ayers & Davis, 1977, 1978). A more recent finding is the unambiguous (cf. Moody, 1970) reflex modulation by one proprioceptor (C-B chordotonal organ) of the efferent input to a muscle receptor organ (the accessory flexor muscle of the M-C myochor-dotonal organ) in a different leg segment (Bush, Vedel & Clarac, 1978).

In order to fully comprehend the contribution of proprioception, and hence proprioceptive reflexes, to the neural regulation of movement or posture, a necessary prerequisite is the establishment of the qualititative and quantitative relationships between sensory feedback and motoneurone output. Moreover, such a study should extend to the innervation patterns, and the electrical and mechanical properties of muscle fibres in the limb muscle reflexly activated (Cannone & Bush, unpublished observations; cf. Atwood, 1967, 1976, for reviews). In the case of the thoracic-coxal muscle receptor organ (T–C MRO), sensorimotor input-output relationships are of added interest by virtue of the non-impulsive (graded), decrementally conducting nature of the sensory feedback to the thoracic ganglion (Ripley, Bush & Roberts, 1968; Bush & Roberts, 1971; Bush, 1976, 1977; Mirolli, 1979). In this (Bush & Cannone, 1973; Bush, 1976, 1977), and in only a few other preparations (Mendelson, 1971; Pearson & Fourtner, 1975; Maynard, & Walton, 1975; Baylor & Fettiplace, 1976; Burrows & Siegler, 1978) has it been possible to examine information transfer between neurones where the absence of impulses has not been artificially induced. In a recent detailed investigation Blight & Llinás (1980), recording pre-and post-synaptically within the thoracic ganglion in the T sensory fibre and in promotor MNs in the crab Callinectes while passing current across a sucrose gap into the T fibre, analysed the transmission characteristics of this ‘tonic’ sensorimotor synapse.

In our previous paper (Cannone & Bush, 1980a) the close similarity between the trapezoid stretch-induced receptor potential waveform of the in-series T sensory fibre and the ‘frequency envelope’of reflexly activated promotor MNs was established. Several promotor MNs were found to discharge in response to receptor stretch with a predictable order of recruitment dependent upon effective stimulus parameters such as receptor muscle length, stretch velocity and stretch amplitude, all of these parameters being reflected jointly or on their own in the degree of depolarization of the T fibre. Moreover, in the most ‘sensitive’ preparations and fairly early on in the course of an experiment, the three commonly activated (lowest threshold) MNs, Pmi-3 are tonically active at maintained receptor lengths. The series of experiments described here, involving the injection of constant current pulses of both depolarizing and hyperpolarizing sign into the S and T sensory fibres, were designed to answer several fundamental questions such as (a) a possible role for the displacement sensitive S fibre in the resistance reflex; (b) whether the tonic activity of promotor MNs is determined purely by receptor length prescribed sensory fibre ‘resting’ potentials; (c) the shape and slope or ‘sensitivity’ of the input-output relationships for individual MNs; (d) the numbers of promotor MNs reflexly activated; (e) the characteristics of adaptation of MNs during constant value sensory fibre depolarizations. In addition, discharge patterning in the form of discrete impulse couplets in one of the promotor MNs (Pm2) is described.

Crabs were prepared as reported previously (Cannone & Bush, 1980a). Direct intracellular stimulation of the S and T fibres was effected using a constant current pulse generator with a stimulus isolation facility. Five current ranges between 0·1 nA and 1·0 mA could be selected, a multi-turn digital readout potentiometer giving a resolution of 10⁻³ within each of these ranges.

Two well-spaced intracellular recording micropipettes in the same sensory fibre allowed determination of the attentuation factor from the decrement in the afferent responses to a stretch stimulus applied to the receptor muscle. From this the fibre length constant, λ, was estimated, using simple infinite cable assumptions (cf. Roberts & Bush, 1971). This in turn can be expressed as the dimensionless electrotonic length, L(= fibre length/λ; Rail, 1969), of the fibre, allowing a more meaningful assessment of current induced reflex activity.

Subsequently, one of the two microelectrodes was used to inject current into the fibre, the membrane potential being monitored by the other. Usually the current passing electrode was positioned proximal (i.e. nearer the thoracic ganglion) to the recording electrode. This ensured that the current flowing from the stimulating micropipette to the synaptic area in the thoracic ganglion did not meet an area of damaged membrane with a consequently lowered resistance giving exaggerated attenuation, as might have resulted from a more proximal recording pipette. The distance between stimulating and recording electrodes was chosen to approximate that between the former and the edge of the ganglion (Fig. 1). Assuming uniform electrical properties along the extra-ganglionic length of the fibre, the signal attenuation between current electrode and ganglion would then be similar to that ‘seen’ by the distal recording electrode. Thus only the unknown intra-ganglionic attenuation need be taken into account in assessing the true input-output relations, including the gain or ‘sensitivity’ of the sensory-motoneurone reflex. This central component of the overall attenuation may well be greater than the peripheral contribution, owing to the evident narrowing and branching of the sensory fibres within the ganglion (Bush, 1976) and any associated changes in electrical properties (but cf. Blight & Llinás, 1980).

As expected from previous work (Bush & Roberts, 1968) and the first paper of this series (Cannone & Bush, 1980a), depolarization of the T fibre evokes reflex discharge of one or more promotor MNs (Fig. 2). In view of the close similarity between the stretch-evoked reflex pattern and the T fibre response waveform, it was surprising to find that S fibre depolarization also resulted in reflex firing of promotor MNs. The T fibre, however, is considerably more potent in its effects than the S fibre, and the individual motoneurones show characteristic differences in their responsiveness to either sensory input. Furthermore, hyperpolarization of the T fibre but not the S fibre inhibits any ongoing or reflexly evoked promotor activity. The input-output properties of the sensory-promotor system have been analysed quantitatively by means of graded current pulses, usually within the range 0–200 nA and of either polarity, injected into the T and S fibres in turn, so as to determine the relationship between sensory fibre membrane potential and resultant reflex discharge frequency for each afferent and efferent unit.

Typical reflex frequency curves (Fig. 3) reveal a sigmoid voltage-frequency relationship for promotor motoneurones Pm1, Pm2 and Pm3, although a linear relationship for both Pm2 and Pm3 is sometimes obtained. Usually, the increases in discharge frequency of these three MNs with incrementally increasing depolarizing currents are comparable, the slopes for Pm1, Pm2 and Pm3 being similar (Fig. 3A). Sometimes however, the slopes of the curves for Pm2 and Pm3 may be considerably steeper than that for Pm1, and may even overlap that of Pm1 (Fig. 3B). The ‘gain’ (or ‘sensitivity’) of the reflex pathways varies from one preparation to another. Over the steepest part of the voltage-frequency curves the output-input relationships obtained in 19 preparations tested in this way ranged from 3·6, 4·2 and 3·2 Hz/mV (Fig. 4B) to as high as 31, 47·5, 33 Hz/mV (Figs 2A, 3B) for Pm1, Pm2 and Pm3, respectively. Factors that could be contributing to such marked variation in sensitivity are the relative state of reflex excitability, or ‘ liveliness’ (cf. Cannone & Bush, 1980 a), of a particular preparation at the time of graded current injections, and the varied electrotonic lengths found for sensory fibres in different preparations (Table 1).

Preparations in which fibre Pmi was tonically active provided the opportunity of investigating the effects of T fibre hyperpolarization on such tonic activity. Even with low tonic frequencies, large hyperpolarizing currents are required to abolish this activity (Fig. 4), of much larger intensities than the depolarizing currents required to add equivalent frequency increments to the initial tonic discharge. The transition from hyperpolarized to depolarized states in the voltage-frequency relationship is usually smooth (Fig. 4A), but occasionally displayed a ‘step’ or plateau in the relationship (Fig. 4B). Such a step might result from nonlinear summation of subthreshold inputs from sources other than the T fibre, including for instance undetermined central influences and/or a small S fibre contribution. Since the voltagefrequency curves of Fig. 4B were obtained with the receptor muscle at its maximum in situ length (12 mm in this preparation), the S fibre would have been considerably depolarized, probably beyond its activation threshold for Pm1 (see below). In any event, the fact that hyperpolarizing the T fibre from its ‘resting’ membrane potential suppresses the receptor length-dependent tonic discharge of Pm1 (cf. inset to Fig. 4B) clearly implicates the T fibre as providing the dominant, if not the sole, sensory drive to Pm1-as also to Pm2 and Pm3.

Depolarizing the S fibre (Fig. 5 A) elicits firing of the same three motoneurones as are reflexly activated by stretch alone or by T fibre depolarization. However, the membrane potential threshold for activation of motoneurones by current induced S fibre depolarizations are usually considerably higher than those for the T fibre. Not only are the activation thresholds higher, but the maximum slopes or ‘gains’ of the voltage-frequency relationships of the individual reflex pathways are much less for the S fibre than they are for the T fibre (Fig. 5 B). This is particularly striking for Pmi, and when direct comparisons are made between the S and T fibre induced reflexes in the same preparation within a short period (Fig. 6).

In further contrast to the situation in the T fibre, hyperpolarizing current, however strong, injected into the S fibre has no effect on any ongoing tonic promotor activity. This is perhaps not surprising, however, since the S fibre depolarization thresholds for promotor activation were invariably well beyond its ‘resting potential’ at RM lengths within the in situ physiological length range (Fig. 6). Indeed, in most experi-ments these thresholds were higher even than the peak potential achieved by the stretch-induced S fibre receptor potential at maximal receptor lengths (except at high stretch velocities). These observations are consistent with the view expressed in the previous paper (Cannone & Bush, 1980a), that the S fibre probably does not contribute directly to the stretch reflex. However, the presence of (direct or indirect) synaptic connexions between the S fibre and promotor motoneurones Pm1–3, implied by the excitatory action of (strong) S-fibre depolarization, leaves open the possibility of summation of S and T inputs to these MNs in vivo, particularly at the more extended receptor lengths. This possibility is reinforced by the observation that the only experimental S fibre whose depolarization threshold for Pmi activity was within the range of stretch-evoked receptor potentials (Fig. 5B), was also the only one with a high length constant (preparation 5, Table 1). This allows greater confidence in the accuracy of the measurements on this preparation-notwithstanding the still high activation thresholds for fibres Pm2 and Pm3, of 25–30 mV more depolarized than the – 50 mV resting potential recorded at the moderate receptor length of 9 mm. The question of possible summation of S and T fibre reflex effects will be discussed later.

The reflex response to a normally effective stretch of the receptor muscle can be partially or completely suppressed by applying graded hyperpolarizations to the T fibre during the period of the stretch stimulus (Fig. 7A). In contrast, hyperpolarizing the S fibre has no effect upon the stretch-evoked reflex, however large the hyperpolarising current (Fig. 7B). This constitutes decisive evidence that it is the in-series T fibre rather than the in-parallel S fibre that provides the principal if not sole sensory drive for the stretch reflex (see Cannone & Bush, 1980a). As expected, the amplitude of the afferent fibre response to stretch increases progressively with increasing hyperpolarization, since the displacement of the membrane potential is away from the equilibrium potential for ions contributing to the receptor potential, and the ‘driving force’ (Hodgkin & Huxley, 1952) therefore increases.

An intriguing effect of T fibre hyperpolarization in preparations exhibiting spontaneous, centrally originating promotor MN activity is the evident suppression, partial if not complete, of a centrally originating burst of impulses (Fig. 7C, 1st record). This again raises the question of how hyperpolariziation of the T fibre exerts an inhibitory effect, in this case upon a centrally determined firing pattern rather than upon a peripherally determined tonic one (see Discussion).

For both the S and T fibres, injecting depolarizing current pulses of constant intensity at different resting lengths of the receptor muscle emphasizes the role of level of depolarization rather than voltage change per se in determining the reflex response. Thus an identical current pulse (or stretch: Cannone & Bush, 1980a) applied at an extended receptor length results in a stronger reflex response compared to that evoked at a more relaxed receptor length, manifested as an increase in discharge frequency and number of activated promotor MNs. This is notwithstanding the fact that current pulses of constant intensity injected into the afferent fibre at increasing receptor lengths result in membrane potential changes of decreasing magnitude (Fig. 8 A, B). This implies a decrease in input resistance of the fibre with increase in receptor length (Fig. 8D), due presumably to an increase in conductance of the transducer terminals with stretch (cf. Lowe, Bush & Ripley, 1978), as in other stretch receptors (e.g. Brown, Ottoson & Rydqvist, 1978). The functional significance of this effect can be appreciated when considering the reflex response to receptor displacement of a particular magnitude superimposed upon a variable receptor length, for thoracic-coxal joint angle (Cannone & Bush, 1980a).

Adaptation of the sensory response at presynaptic stages (mechanical and transduction levels) has been dealt with previously (Bush & Roberts, 1971 ; Bush & Godden, 1974; Bush, Godden & Macdonald, 1975; Cannone & Bush, 1980a). There remains the possibility of postsynaptic adaptation at the levels of the central synapses in the reflex pathways and the site of impulse generation in the motoneurones. In many preparations, fibre Pmi is tonically active at frequencies related to the length of the receptor muscle, and fibres Pm2 and Pmj may also fire tonically, evidently at least partly as a result of afferent input within the physiological length range of the receptor muscle (fig. 3, Cannone & Bush, 1980a). It is therefore important to determine the ability of the central transmission process to maintain a constant frequency impulse output in the motoneurones in response to an invariant membrane potential in the sensory fibres, particularly in the case of the T fibre. A postural role for the peripherally driven ‘tonic’ motoneurone output cannot be convincingly argued if in fact there is a significant decline in impulse frequencies in the face of a constant sensory input.

It is clear that the motoneurone response is particularly sensitive to the early phase of a current induced depolarization (Figs. 8, 9). Even when an initial fast transient is absent in the intracellular recordings, as in the T and S fibre responses at the longer receptor lengths (Fig. 8), the reflex response nevertheless shows a high-frequency burst during the early phase of current pulse. This is commonly followed by some depression in frequency before a second increase to a relatively constant level maintained over a longer time course. Further, during the first few seconds of a prolonged (e.g. 20 s duration) constant current pulse, the T fibre membrane potential may decline by several millivolts (Fig. 9A) so that an accurate indication of the rate of central adaptation of motoneurone impulses during this early phase cannot be obtained from this data. With the T fibre membrane potential fairly stable during the remainder of a 20 s pulse however, the impulse frequency of all three promotor MNs under consideration (Pmi, Pm2 and Pm3) still declines steadily. The greater the level of depolarization of the T fibre, and consequently the higher the resulting reflex frequencies, the greater the rate of adaptation in all three MNs (Table 2). With smaller, more physiological depolarizations of the T fibre, however, such as would presumably occur following slight postural adjustments of the thoracic-coxal joint in the living animal, the rate of adaptation may be very small (Fig. 9B).

The three promotor MNs, Pm1, Pm2 and Pm3, differ in their rates of adaptation, these differences being more evident with larger current induced depolarizations (especially at unphysiologically high levels of depolarization). Thus, both within and beyond probably normal physiological limits, Pmi shows the least and Pm2 the greatest rate of adaptation to constant current pulses. Accordingly, of the three motoneurones Pmi has properties best fitted for a tonic role, a role appropriate for the maintenance of the postural stance of the thoracic-coxal joint. The continued discharge of fibres Pmz and Pm3 over equally long periods of time with only slightly greater rates of adaptation, suggests that they too may contribute towards this function. This view is reinforced by the observation in one particularly ‘excitable’ preparation (see Fig. 3, Cannone & Bush, 1980a) of maintained tonic discharge of all three motoneurones at the maximum in situ length of the receptor muscle.

In a few exceptionally ‘lively’ preparations, receptor stretch alone has revealed that motoneurones of higher thresholds than Pm3 can be activated (Fig. 10A). These fibres, although active only during the positive dynamic phase of the trapezoid stimulus, cannot on this basis alone be characterised as rapidly adapting or phasically responding. Injection of current pulses into the S and T fibres has proved more valuable in establishing (a) the total number of motoneurones that may be reflexly activated, (b) the differential recruitment of certain of these motoneurones by depolarization of the S and T fibres, and (c) the phasic nature of the response characteristics of at least some of these motoneurones.

A fourth promotor MN can occasionally be discriminated in the reflex responses to moderate depolarizing currents applied to either S or T fibres, in addition to the three commonly active MNs Pm1–3 (Fig. 10 B, D). The recorded impulses of this motoneurone (Pm4) are smaller than those of Pm1, so that relatively high gain, good signal-to-noise recordings are needed to see it. It has a higher threshold than Pm1–3 for either S or T fibre depolarization (hence Pm4), and in the absence of current stimuli has never been observed to be tonically active within the physiological range of receptor lengths.

Large depolarizing currents injected into the T fibre will, in some fresh and active preparations, evoke discharges of large-amplitude spikes that decay in frequency fairly rapidly. By simultaneously recording T fibre membrane potentials, promotor nerve activity, and intracellular junctional potentials in different promotor muscle fibres, it can be clearly established that these larger spikes represent further MNs recruited in addition to Pm 1–4, and are not merely the result of compounding effects at the high reflex frequencies in force (Fig. 10 C). At least three separate junctional events, each attributable to one of the additional large-amplitude impulses (termed ‘Pm5’, ‘Pm7’ and ‘Pm8’; pairs of solid arrows in each of three different sections of the record), can be identified in the same polyneuronally innervated muscle fibre. Moreover, in this recording a further motoneurone (‘Pm6’) can be identified on the basis of impulse amplitude (broken arrows) although it does not appear to innervate this particular muscle fibre. Other muscle fibres recorded from in the same preparation all responded with large amplitude junction potentials to one or more of these additional large amplitude impulses. Significantly, whereas fibres Pm1, Pm2 and Pm3 discharge at high maintained frequencies throughout the long duration current pulse, the rarge amplitude impulses adapt rapidly in frequency and are absent altogether at the end of the stimulus pulse.

Thus the same T fibre depolarization recruited at least eight different motoneurones (Pm 1–8), each identifiable on the basis of clearly distinguishable impulse amplitudes and/or muscle junction potentials. It is also clear that for many muscle fibres of the promotor muscle it is the high threshold, rapidly adapting, large-amplitude impulses (presumably large-diameter motoneurones) that produce the most marked electrical and mechanical responses. The large-amplitude muscle junction potentials may rapidly summate at the beginning of the current pulse, and may even cause a depolarization of the muscle fibre membrane supra-threshold for the generation of a muscle action potential (Fig. 10E), in turn resulting in a visually apparent twitchlike contraction of the muscle fibre.

Of the eight motoneurones (Pm1–8) recruited by T fibre depolarization, four (Pm1–4) are common to S fibre depolarizations. In addition, the S fibre uniquely activates another high threshold motoneurone (‘Pm9’), impulses of which are smaller than those of P1T14 (Fig. 10D). The four large-amplitude, rapidly adapting impulses (Pm5–8) detected by strong T fibre stimulation are not evoked by S-fibre depolarization.

The generation of a patterned discharge has been well documented for a variety of crustacean motor systems (e.g. Wilson & Davis, 1965; Burrows & Horridge, 1968; Gillary & Kennedy, 1969 a). It has also been established that grouped motor impulses can produce enhanced postsynaptic effects in crustacean muscle fibres, compared to evenly spaced unpatterned outputs at the same average frequency (Wiersma & Adams, 1950; Ripley & Wiersma, 1953; Gillary & Kennedy 1969b). In the present study, evidence has been obtained of pattern generation in fibre Pm2, though the effects of such patterning on the development of tension in the innervated muscle fibres has not yet been evaluated. Pattern generation was observed in relatively few preparations, suggesting that the central mechanism or motoneurone properties responsible for pattern generation are particularly labile under the experimental conditions in force.

Patterned output in response to either receptor stretch (Fig. 11 A), or to depolarizing currents injected into the T sensory fibre (Fig. 11B), is usually in the form of impulse couplets with an interval of 7–8 ms between impulses of a couplet. Occasionally, however, patterning may take the form of well-separated bursts of impulses, each burst comprising 2–4 impulses (Fig. 11C). Recording the electrical response of a promotor muscle fibre to a current induced patterned output of Pm2 has shown a step-like depolarization of the membrane which may, provided high couplet frequencies are evoked, lead to the generation of a muscle action potential triggering a twitch contraction of the muscle fibre (Fig. 11D).

Graded depolarizations of the T fibre allow a quantitative analysis of impulse patterning (Fig. 12). This series of histograms portrays the intervals between Pm2 impulses for a series of current injections of 55–100 nA. In this particular experiment, with mean impulse frequencies of up to 32 Hz (55–75 nA stimulus pulses), an overall mean intra-couplet impulse interval of 7·3 ms with little variation was retained. Larger currents, and therefore more elevated mean impulse frequencies, result not only in the mean inter-couplet impulse intervals becoming progressively shorter, but also in a gradual increase in the intra-couplet interval. Eventually the two interspike interval ‘populations’ (intra-couplet and inter-couplet intervals) merge, at a current intensity of 95 nA, obscuring the couplets altogether. Thus, the higher the mean impulse frequency the less rigid the interval between impulses of a couplet.

While these experiments do not reveal the mechanism responsible for pattern generation in motoneurone Pm2, they do indicate that patterning can be experimentally altered in a manner that possibly also occurs in vivo. Moreover, patterning in this sytem is almost certainly a property of the motoneurone itself, since it is restricted to only one of eight or more reflexly activated motoneurones.

More so than in the case of the T fibre, S fibre depolarization often leads to variable reflex response frequencies even when spontaneous, centrally originating impulse activity is not evident. This is more readily apparent when constant intensity current pulses are injected repetitively at the same receptor muscle length. For example, an overt pre-stimulus central drive onto Pm1 is reflected not only in an increased discharge of Pm1 during a current stimulus, but also in the effect of elevating Pm2 discharge frequencies where no pre-stimulus discharge of this MN occurred (Table 3). A likely explanation for this behaviour is that the central nervous system imposes a variable, often sub-threshold excitatory influence upon promotor MNs. This effect can therefore be seen to be related to the previously described loose coupling between central drive and the phenomenon of post-stimulus excitation (Cannone & Bush, 1980 a). Such variability in the reflex response, occurring as it does during sensory fibre activation, can have important implications if it can be shown to occur in the in vivo situation.

The experiments described here and in our previous paper (Cannone & Bush, 1980a) make it abundantly clear that there is a quantitative relationship between sensory response and reflex strength. By implication, as the T–C joint angle increases on remotion, there is both a recruitment of reflexly active promotor MNs and an increase in their discharge frequencies, similar to the situation in the more distal leg segments (e.g. Evoy & Cohen, 1969). The quantitative basis for these effects is the level of depolarization achieved by the in-series T sensory fibre. However, such a simplistic view fails to take into account the fact that muscle receptor organs not only respond to joint movement, but also to efferent modulation of receptor muscle tension. Although in some conditions the S fibre may respond to strong isometric contraction of the receptor muscle (cf. Cannone & Bush, 1980 c), it is the T fibre that is specially responsive to active tension (Bush & Godden, 1974) generated by impulse discharge in the two excitatory motoneurones innervating the receptor muscle (Bush & Cannone, 1974).

In general, signal attenuation in the S fibre was found to be much greater than that in the T fibre, with an electrotonic length (L) of 0·4 (A, 11 mm) being an exceptional value compared to ‘normal’ values > 1·0 (λ, 3·5 mm). Earlier determinations of length constants for the S fibre (Roberts & Bush, 1971 ; Bush, 1976) were similarly significantly lower than our upper values for the T fibre (λ, 16 mm; L, 0·3). The apparent non-contribution of the S fibre to the stretch-evoked reflex (Cannone & Bush, 1980a; Fig. 7, this paper) needs to be reconciled with the established central connectivity between the S fibre and at least five promotor MNs. Several factors need to be taken into account. Firstly, the progressive decline in reflex excitability of these highly dissected preparations (cf. Cannone & Bush, 1980a) would suggest that the thresholds for motoneurone activation by the S fibre in vivo could be lower than those determined experimentally. Secondly, although the dynamic components of the stretch-evoked receptor potentials in the S fibre are relatively small in amplitude due to a low sensitivity to velocity of stretch, there is a substantial change in the steadystate membrane potential of the S fibre over the physiological length range of the receptor muscle ; this can be as much as 30–40 mV depolarization over an effective length change of 4–6 mm (cf. Fig. 5, Cannone & Bush, 1980a). Thirdly, in the one experimental S fibre for which a relatively large length constant was found (λ = 11 mm, cf. Table 1), stretch-evoked receptor potentials at the more extended receptor lengths achieved a level of depolarization greater than the activation threshold for Pm1 as determined by injecting currents into the same S fibre. Finally, S fibre connectivity has been established for the low threshold, slowly adapting promotor MNs, Pm 1 – 4, but not for the higher threshold, rapidly adapting promotor MNs, Pm5 – 8. Taking all these factors into consideration, we conclude that the S fibre almost certainly contributes to the resistance reflex in vivo. Any such contribution is probably a small one, however, and might well be limited to the more extreme, ‘remoted’ joint positions when the receptor muscle approaches its fully extended length in situ. Moreover, the fact that S fibre connectivity seems to be confined to the lower threshold, slowly adapting promotor MNs suggests that it may be primarily concerned with a postural role of the T-C MRO. In any event, it must be concluded from this and the preceding paper that the promotor resistance reflex in vivo is predominantly, but probably not exclusively, mediated by the T fibre. Accordingly the gain and effective range of the reflex will be much more amenable to central control via the receptor motor supply than if the S fibre provided the principal sensory input (cf. Cannone & Bush, 1980b, c).

That as many as nine promotor MNs are reflexly recruited by T and S fibre sensory input is beyond dispute. In contrast, it is not clear whether the sensorimotor connexions are monosynaptic and whether they are chemically transmitting (Bush, 1977; Cannone & Bush, 1980 a). However, in the absence of any evidence to the contrary, chemical synaptic transmission in this reflex system is indicated (Emson, Bush & Joseph, 1976), and monosynaptic connexions occur in Callinectes(Blight & Llinás, 1980).

There is considerably less doubt as to the source of the tonic firing of Pmi-3, since T fibre receptor-length-induced depolarization adds to the frequency thereof, and T fibre hyperpolarization decreases this tonic output, at least for Pmi. This latter effect is explicable in terms of our proposal that there is a continuous ‘tonic’ release of transmitter by the T fibre even at fully relaxed receptor lengths (Cannone & Bush, 1980a), in much the same way that locust interneurones release transmitter at their ‘resting’ potentials (Burrows & Siegler, 1978). However, the reason why T fibre hyperpolarization can suppress the frequency of an apparently centrally originating spontaneous burst of impulses is not immediately obvious if purely chemical synaptic transmission is assumed. In this context too the failure of T fibre hyperpolarization-as during a pronounced negative dynamic response to a trapezoid length change-to affect the ‘post-stimulus excitation’, a phenomenon coupled to spontaneous central drive (Cannone & Bush, 1980a), needs to be accounted for. Although spontaneous bursts in promotor MNs probably do not reflect in vivo patterns of centrally determined activity (cf. Cannone & Bush, 1980a), reconciling these observations should provide a useful insight into central connectivity patterns and transmission characteristics.

Assuming that synaptic transmission is chemical, the simplest explanation of the suppression of centrally originating bursts of impulses in promotor MNs is that there is a continuous subthreshold excitatory drive onto these promotor MNs by the T fibre. This is not indicated for the S fibre since there is no effect of S fibre hyperpolarization on centrally originating spontaneous activity. The T fibre drive could conceivably sum with a variable, even subthreshold, central drive, explaining why such central bursts of impulses involve only those promotor MNs which are active in that particular preparation. Hyperpolarization of the T fibre would then remove the peripheral component of synaptic drive onto promotor MNs, thereby effectively raising the threshold for the central component of synaptic drive onto MNs so affected. The concept that a major contribution of sensory input associated with movements is to provide the central nervous system with a source of (generalized) excitatory input from which movement can be ‘moulded’ (Cohen, 1965) is consistent with the subthreshold excitatory effects proposed here for the T fibre input.

Adaptation to maintained stretch of the receptor muscle, as evidenced by a decline in the T fibre level of depolarization, can be at least partly attributed to viscoelasticity of the receptor muscle (Bush & Roberts, 1971; Cannone & Bush, 1980a). Similarly, receptor muscle properties are responsible for the time-dependent variability seen in repetitive responses to identical stretch stimuli. Thus, all peripheral components contributing to adaptation are duly reflected in the reflex output, the T fibre response waveform with its prominent positive dynamic component dictating a ‘phasic’ and a ‘tonic’ character thereto (Cannone & Bush, 1980a).

Clearly, information transfer within a non-impulsive sensory neurone is advantageous in the sense that there is no conversion from analogue to digital signalling, and thus no loss of information content at a spike initiation site. Certain constraints are imposed on such a neurone, however, by virtue of the passive electrical properties dictating signal attenuation and distortion with distance. Nevertheless, in a comparative treatment of non-impulsive neurones, Pearson (1976) concludes that in preparations such as the T – C MRO where high-output frequencies are achieved by small changes in the input, the discharge rate of motoneurones can be precisely regulated by an input element that is non-impulsive, rather than by a spiking driver neurone producing discrete excitatory postsynaptic potentials. Thus, in a non-spiking sensory system such as the T – C MRO a ‘requirement’ for high amplification between sensory feedback and reflex response (cf. Fig. 3) could conceivably have contributed to the evolution of this purely analogue sensory signalling system.

The nature of the T – C MRO reflex response may be additionally determined by factors other than the above, and the possible summatory role of the S fibre input. These could include centrally imposed subthreshold excitatory effects imposing a degree of variability on the reflex output to a constant sensory input; the degree of frequency adaptation within a particular MN ; the extent to which impulse patterning favours muscular contraction compared to a non-patterned impulse regime; and finally, neuromuscular properties and innervation patterns of the individual muscle fibres making up the promotor muscle (Cannone & Bush, unpublished observations). All of these factors could determine to a large extent the efficacy of the resistance reflex.

The MNs supplying the tonic flexor and extensor muscles of the crayfish abdomen, and the ‘slow’ (tonic) MNs of crustacean leg muscles, are normally concerned with postural regulation and slow movements (Atwood & Walcott, 1965; Kennedy & Takeda, 1965a; Wilson & Davies, 1965). These MNs often show a high level of spontaneous discharge. By contrast, the MNs supplying the ‘fast’ (phasic) flexor and extensor muscles, and those supplying certain ‘fast’ crab leg muscles, show little or no spontaneous tonic activity (Kennedy & Takeda, 1965 b). On the basis of evidence presented here (section 3), we conclude that promotor MNs reflexly activated can themselves be categorized as tonic (Pm1 – 3, and probably Pm4) and phasic (Pm5 – 8), partly on the basis of the observed rates of frequency adaptation, and partly on the basis of the relative activation thresholds of these two groups of MNs. The low rates of frequency adaptation of Pm1 – 3, particularly at moderate (physiological) levels of T fibre depolarization, supports our contention that the reflex activation of these MNs could serve a postural role (Cannone & Bush, 1980a). The high threshold, larger-amplitude impulses of Pm5 – 8 adapt rapidly in frequency in response to a high, constant amplitude, T fibre depolarization. Nevertheless, the large-amplitude, facilitating and summating muscle junction potentials so evoked can lead to the initiation of muscle action potentials, and hence twitch contractions of muscle fibres innervated by these MNs. Thus, during the dynamic phase of an imposed remotion when T fibre depolarization is at its maximum, the recruitment of the phasic category of promotor MNs over and above the firing of the tonic category can be expected to lead to the development of a powerful muscular resistance at the very onset of the movement.

A final consideration in the context of the efficacy of the resistance reflex and the factors that contribute thereto is the stretch-induced positive feedback to the receptor muscle (Bush & Cannone, 1974), and the suggested tonic excitatory input to the receptor muscle (Cannone & Bush, 19806). Efferent input to the receptor muscle during an imposed movement can therefore be expected to lead to an enhancement of the gain of the reflex loop, or at least to extend the range of receptor length over which the reflex is activated, and to abolish ambiguities in the reflex response imposed by viscoelastic properties of the receptor muscle (Cannone & Bush, 1980a, c).

This work was supported by a project grant from the Medical Research Council to B.M.H.B.