Subdivision of primary afferents from passive cat muscle spindles based on a single slow-adaptation parameter

Subdivision of primary afferents from passive cat muscle spindles based on a single slow-adaptation parameter

Brain ¢~12(1993) 110- I 1,4 '~;, 1993 Elsevier Science Publishers B.V. All rights reserved I)006-8993/93/$1)6.ilii ] l0 BRES 18832 Subdi...

421KB Sizes 0 Downloads 13 Views

Brain ¢~12(1993) 110- I 1,4 '~;, 1993 Elsevier Science Publishers B.V. All rights reserved I)006-8993/93/$1)6.ilii

] l0

BRES 18832

Subdivision of primary afferents from passive cat muscle spindles based on a single slow-adaptation parameter F. Awiszus and S.S. Sch~ifer Medizinische Hochschule Hannot~er, Abteilung Neurophysiologie (OE 4230), Hannot'er (FRG) (Accepted 15 December 1992)

Key words: Cat muscle spindle; Primary afferent; Slow adaptation; Intrafusal muscle fiber; lnterspike interval function

38 primary afferents originating from de-efferented cat tibialis anterior muscle spindles were investigated. Ramp-and-hold stretches of the host muscle were performed with a varying amount of muscle pre-stretch while recording the primary afferent discharges. From the discharge responses an interspike interval function was estimated. This revealed a slow adaptation during the hold phase of stretch which could be approximated quite well by a power function. The slow-adaptation power function exponent (SAE) was found to be rather independent of the amount of pre-stretch applied to the host muscle and grouped around a value characteristic for each afferent. These 'characteristic SAEs' showed a clearly bimodal distribution within the population of primaries studied. Moreover, the distribution around both modes was narrow enough to justify the subdivision of the primary afferents according to their characteristic SAE as either high-SAE (10 afferents; 26%) or Iow-SAE (28 afferents; 74%) afferents. The most likely explanation for this bimodality of slow-adaptation behavior in primary afferents is given by the assumption that the afferent discharge of the passive spindle is mainly provided from a branch innervating either the bag z (for high-SAE units) or the bag 2 and chain (for low-SAE units) intrafusal fibers.

INTRODUCTION Responses of primary muscle spindle afferents to a ramp-and-hold stretch are largely determined by properties of the intrafusal muscle fibers s'12'13. One of the most prominent features of intrafusal bag~ fibers is their ability to show a large amount of slow mechanical adaptation during the hold phase of a ramp-and-hold stretch even under passive or de-efferented conditions, whereas the other intrafusal fibers are known to show only a small amount of mechanical adapatation 7. If it is assumed that the action-potential-encoding site of a passive primary afferent is influenced mainly by current originating from afferent terminals of a single intrafusal fiber, one would expect that it should be possible to subdivide passive primary muscle spindle afferents into at least two groups based on their slowadaptation properties. One group, for which the discharge originates from the bag~ fiber, should show a large amount of slow adaptation, whereas the slow adaptation of those afferents mainly influenced by the other intrafusal fibers should be rather small.

Such a subdivision into a high-adaptation and lowadaptation group of passive primary afferents has been demonstrated recently 17. The actual amount of slow rate-of-discharge adaptation, however, could not be used as distinguishing parameter as it depends to a large extent on the discharge rate achieved at slowadaptation onset. Therefore, a distinguishing scheme based on two parameters, the amount of slow adaptation and the discharge rate at adaptation onset was necessary to achieve the subdivision w. In this paper it was tested, if a single characteristic parameter can be found for each afferent subdividing primary afferents of passive muscle spindles into two groups based on their slow-adaptation features during the hold phase of a ramp-and-hold stretch. MATERIALS AND METHODS Detailed accounts of the experimental procedures have been given in previous publications11'3'16A7, thus only a brief account is given here. Experiments were performed on cats (weight range: 2-4 kg) anesthetized with pentobarbitone sodium. The left hindlimb was denervated with the exception of the tibialis anterior muscle. After

Correspondence: F. Awiszus, Medizinische Hochschule Hannover, Abteilung Neurophysiologie (OE 4230), W-3000 Hannover 61, FRG.

111 laminectomy, the dorsal and ventral roots L 6 and L 7 were cut at their entry into the spinal cord. Single muscle spindle afferents originating from the tibialis anterior muscle were isolated in the dorsal root filaments. Only primary afferents with a conduction velocity over 80 m / s were selected. After mobilization of the tibialis anterior and its tendon without damaging the blood supply, the tendon was attached to a servo-controlled electromagnetic stretch device. Thereafter, the muscle u n d e r investigation was exposed to several series of ramp-and-hold stretches with an amplitude of 7 mm and a stretch velocity of 10 m m / s while recording the activity of the isolated primary afferents on tape for further off-line analysis. Each series consisted of 5 identical stretches with a hold-phase duration of 3 s and 7 s waiting time between individual stretches. The first series of stretches was started from the minimal physiological length of the tibialis anterior muscle (measured under maximal plantarflexion). This muscle length is henceforth referred to as 0 m m pre-stretch. Thereafter the muscle pre-stretch was increased by 3 m m and the series of 5 ramp-and-hold stretches was repeated. The whole procedure was repeated - increasing the pre-stretch in 3 m m steps - up to a pre-stretch value of 12 mm. Analog signals as played back by the tape recorder were digitized (rate: 10 kHz) by an IBM P C / A T compatible laboratory computer. Action potentials were detected using a threshold criterion and times of occurrence relative to the onset of the hold phase were calculated. From each series of stretches the response to the first stretch was discarded and the action potentials obtained during the second to fifth stretch were used to construct an interspike interval superposition plot (IISP) I'2. An lISP represents a graph consisting of dots representing all action potentials obtained during the stimulus repetitions of interest. For each dot the abscissa denotes the time relative to the start of the hold phase of stretch and the ordinate the preceding interspike interval. If the dots of an IISP represent the graph of a function, an estimate of this function is called interspike interval function or IIF for short 1'2. For all IISPs obtained in this study it was found that the IISP dots closely resembled the graph of a function as long as the median of the interspike intervals in the time interval from 250 ms to 350 ms after hold-phase onset was smaller than 25 ms (corresponding to a discharge rate of 40 i m p / s ) . IISPs showing longer interspike intervals during this period could be rather irregular and did not justify estimation of an underlying function. Consequently, such responses were excluded from further analysis. For the remaining IISPs the underlying IIF was estimated by m e a n s of a 'windowed median '15. For its calculation the ordinate value of each lISP dot is replaced by the median of that point's ordinate and the ordinates of the 2m dots nearest in abscissa• The parameter m represents the window size and determines the degree of 'smoothing' achieved by this procedure• For this study a value of 5 was chosen for the window size m.

RESULTS In total data from 54 primary afferents were obtained. However, 16 of these afferents were unable to produce a sufficiently short interspike interval during the hold phase even under the highest pre-stretch values of the host muscle. Therefore, only 38 primary afferents were included in this study. A typical example of a primary-afferent IISP with estimated IIF is shown in Fig. 1. The pre-stretch value of the host muscle was 12 mm in this case. The dots of the IISP closely resemble the graph of the function estimated by the continuous line although some scatter is present especially at longer interspike intervals. The time course of the estimated IIF looks like an upside-









• ' C .: ,.,.'." '; .'.::' ""..,. ". ....',.~,...

a, t, 10-



£ o





a 0b0 z0'00 time [ms]


Fig. 1. Primary afferent lISP with estimated IIF (a) to the rampand-hold stretch of the host muscle shown in (b). The pre-stretch value of the host muscle was 12 m m in this case.

down version of the well known instantenous frequency response of primary muscle spindle afferents to a ramp-and-hold stretch. During muscle lengthening the IIF decays showing several downward peaks that correspond to the upward peaks in the rate of discharge observed by others m'18. At hold-phase onset the IIF reaches a minimum followed by a rapid increase which in turn is followed by a more or less pronounced decay back to shorter interspike intervals. After reaching a minimum which is not very prominent in Fig. 1, the IIF increases monotonously throughout the entire hold phase of stretch and does not appear to reach a constant value. This later part of the hold-phase response will be called the slow adaptation 16. The time course of the slow adaptation in the IIF resembles that of a power function. This is confirmed in Fig. 2 showing the estimated IIFs during the middle part of the hold phase (300-2000 ms after hold phase onset) on a double-logarithmic scale for two typical primary afferents. The IIFs for four different prestretch values (3 ram: dotted line; 6 ram: dashed line; 9 ram: dashdot line; 12 mm: continuous line) of the host muscle were included for each afferent. All eight IIFs of Fig. 2 appear to be sufficiently linear on the doublelogarithmic scale. For both afferents of Fig. 2 the IIF slope on the double-logarithmic scale appears to be independent of the pre-stretch value for each particular afferent, whereas the ordinate intercept increased clearly with increasing pre-stretch of the host muscle. The main

112 40 ¸


40- b








.~ z0-







t i m e [ms]

lObO t i m e [ms]


Fig. 2. Middle part of the hold-phase llFs for two typical primary afferents (one high-SAE (a) and one Iow-SAE (b) afferent). The four different pre-stretch values of the host muscle used are indicated by different linetypes. Dotted line: 3 mm; dashed line: 6 ram; dashdot line: 9 ram; continuous line: 12 ram.

difference between the two afferents is given by the fact that for the afferent in Fig. 2a the slopes are much higher than the slopes of the afferent in Fig. 2b. It was found for all responses of all primaries included in this study that the IIF during the middle part of the hold phase was approximately linear on a double-logarithmic scale. Consequently, the slow-adaptation time course (SATC(t)) may be approximated by S A T C ( t ) = SAP. t sAE, where SAP will be called the slow-adaptation prefactor (corresponding to the ordinate intercept of the line on the double-logarithmic scale) and SAE the slow-adaptation exponent (corresponding to the slope of the line on a double-logarithmic coordinate system). For all responses of all afferents SAE and SAP were estimated using double-logarithmic linear regression of the IIFs on the time interval from 300 to 2000 ms after hold-phase onset. Fig. 3 shows the SAE estimates for the responses of Fig. 2a (filled circles) and Fig. 2b (open circles). As can be seen, the SAE value for both afferents was rather independent of the amount of pre-stretch applied to the host muscle a result also found for all other primary afferents included in this

study. Consequently, it appeared reasonable to regard the median SAE (taken over all relevant pre-stretch values) as a characteristic feature of each afferent. The afferent represented by the filled circles in Fig. 3, exhibited a median SAE of 0.21, whereas the median SAE of the open-circle afferent was 0.067. Fig. 4 shows a histogram of the median SAE values for all 38 afferents included in this study. As can be seen the distribution of this characteristic afferent feature is clearly bimodal within the population of primary afferents studied. The first mode appears at a median SAE of 0.07, whereas the second mode appears at a median SAE of 0.21. Moreover, the distributions of median SAEs around these modes is so narrow that a 'gap' (at a median SAE of 0.15) exists between median SAEs near the low mode and those near the high mode. Thus, it appears reasonable to subdivide the primary afferents into two groups, the low-SAE afferents with a median SAE below 0.15 and the high-SAE afferents with a median SAE above 0.15. Using this criterion, 28 (74%) of the investigated primary afferents were classified as low-SAE afferents, whereas the group of high-SAE afferents consisted of 10 primaries (26% of the whole population).








o p r e s t r e t e h [mm] Fig. 3. SAE estimates for the responses of Fig. 2a (filled circles) and Fig. 2b (open circles).



[-7 o13

SAE m e d i a n

Fig. 4. Histogram of the SAE median for all 38 primary afferents included in this study.

113 DISCUSSION The results given in this paper confirm the findings of Sch~ifer 17 that primary afferents of passive cat muscle spindles may be subdivided into two groups based on their slow-adaptation response to a ramp-and-hold stretch. The classification criterion of Sch~ifer 17 required measurement of two parameters, the amount of slow adaptation and the discharge rate at slow-adaptation onset and the relationship between these two parameters provided the basis for the afferent classification as lowly or highly adapting. In contrast to this two-parameter procedure the classification introduced in this paper is based on a single parameter, the median SAE, allowing a much simpler classification. Nonetheless, both classification procedures yielded identical subdivisions for the sample of primaries investigated in this paper. In particular, primaries classified as highly adapting by the method of Sch~ifer 17 were found to be high-SAE afferents, whereas lowly adapting units were classified as low-SAE afferents. To arrive at a single parameter allowing subdivision requires several transformations of the original spike train data. At first the spike trains are displayed as lISP and an underlying IIF is estimated 1'2. When compared to the 'rate of discharge superposition plot '6 or 'instantaneous-frequency display', usually employed in muscle-spindle analysis, the lISP approach emphasizes those response parts with long interspike intervals whereas those with short interspike intervals get compressed. As the slow adaptation consists of monotonously elongating interspike intervals, the lISP approach appears to be suited quite well for slow-adaptation analysis. The next analysis step consists of IIF estimation from the lISP. In general, the liSPs justified estimation of an underlying IIF reasonably well as long as the interspike interval at the beginning of the slow adaptation was short enough (about 25 ms). Responses of units that could not start their slow adaptation at such a short interspike interval even at high pre-stretch values of the host muscle, did not allow estimation of a reasonable IIF and therefore these units cannot be classified using the classification criterion introduced in this paper. For the remaining units the time course of the slow IIF adaptation could be approximated quite well by a power function (cf. Fig. 2). Moreover, the power function assumption for sensory adaptation has a profound theoretical basis in information theory 14 giving further reasoning to this kind of approach. Finally, the power function exponents of each afferent did not depend on the pre-stretch of the host

muscle and grouped around a single value specific for each afferent. As characteristic slow-adaptation exponent for each afferent the median over all SAEs at different pre-stretch values was taken. The SAE median was found to be a superior parameter than the SAE mean, as the mean is quite sensitive to the presence of outlier SAE values, whereas the median is very robust in this respect. The observed bimodality in the median SAEs should have an intrafusal basis. The most likely explanation is given by the assumption that the afferent discharge is generated mainly by the action of a particular type of intrafusal fiber even in the passive spindle. According to Banks et al. 5'4, branching trees of muscle spindle primaries may have rather specific geometries. The distribution of first order branches may be 'segregated' meaning that one branch supplies only the bag 1 fiber and the other(s) supply bag 2 and chain fibers. Although the tibialis anterior muscle was not studied by Banks et al. 5, one might speculate that tibialis anterior spindles contain mainly primary afferents with segregated branching trees. In this case the subdivision found in this paper would have a natural anatomical basis. On the other hand, most hindlimb muscles receive a high proportion of primary afferents without such a specific segragation 5. One should note, however, that even in this case the actual action potential encoding site may sample its receptor current from a single intrafusal muscle fiber 9. In particular, bag1 fibers are known to 'creep' during the hold phase of a ramp-and-hold stretch 7,8 giving rise to a large amount of mechanical adaptation. Thus, afferents sampling their receptor current mainly from a bag~ fiber should show the high adaptations observed in the high-SAE afferents of this study. On the other hand, bag 2 and chain fibers show only a small amount of mechanical adaptation during a ramp-and-hold stretch 7,8. Thus, it apears reasonable to assume that afferent terminals on these intrafusal fibers show only a rather small adaptation of their receptor current which could be the basis for the small adaptations observed in our low-SAE afferents. If these speculations are correct, one would expect that a low-SAE afferent should be transformed into a high-SAE afferent by means of a dynamic y-stimulation. Similarly, a static y-stimulation should change a high-SAE into a low-SAE afferent. If such changes, however, can occur, remains to be demonstrated. REFERENCES 1 Awiszus,F., Continuousfunctionsdeterminedby spike trains of a neuron subject to stimulation, Biol. Cybern., 58 (1988) 321-327.

114 2 Awiszus, F., On the description of neuronal output properties using spike train data, Biol. Cybern., 60 (1989) 323-333. 3 Awiszus, F. and Schiller, S.S., Re-afferent effects of individual static and dynamic y-stimuli during maintained fusimotor stimulation, Brain Res., 489 (1989) 41-48. 4 Banks, R.W., Observations on the primary sensory ending of tcnuissimus muscle spindles in the cat, Cell Tissue Res., 246 (1986) 309-319. 5 Banks, R.W., Barker, D. and Stacey, M.J., Form and distribution of sensory terminals in cat hindlimb muscle spindles, Philos. Trans. R. Soc. Lond. B, 299 (1982) 329-364. 6 Bessou, P., Laporte, Y. and Pages, B., A method of analysing the responses of spindle primary endings to fusimotor stimulation, Z Physiol., 196 (1968) 37-45. 7 Boyd, I.A., The mechanical properties of dynamic nuclear bag fibres, static nuclear bag fibres and nuclear chain fibres in isolated cat muscle spindles, Prog. Brain Res., 44 (1976) 33-50. 8 Boyd, 1.A., The isolated mammalian muscle spindle, Trends Neurosci., 3 (1980) 258-265. 9 Eagles, J.P. and Purple, R.L., Afferent fibers with multiple encoding sites, Brain Res., 77 (1974) 187-193. 10 Gioux, M., Petit, J. and Proske, U., Responses of cat muscle spindles which lack a dynamic fusimotor supply, 3. Physiol., 432 (1991) 557-571.

I 1 Holm, W., Padeken, D. and Schiller, S.S.. Characteristic curves ~1 the dynamic response of primary muscle spindle endings with and without gamma stimulation, Pfliigers Arch., 391 ( 1981 ) 163 - 171~. 12 Hulliger, M., The mammalian muscle spindle and its central control, Rec. Physiol. Biochem. Pharmacol., 101 11984) 1-t11). 13 Hunt, C.C., Mammalian muscle spindle: peripheral mechanisms, Physiol. RetJ., 70 (1990) 643-663. 14 Norwich, K.H. and Valter McConville, K.M.. An informational approach to sensory adaptation, J. O)mp. Physiol. A, 168 (19t,H) t51-157. 15 Press, W.H., Flannery, B.P., Teukolsky, S.A. and Vetterling, W.T., Numerical recipes: the art o f scientific computing, Cambridge University Press, Cambridge, 1986. 16 Sch~ifer, S.S., The fast and slow component of the receptor adaptation in the discharge frequency of cat's primary muscle spindle afferents, Z. E E G - E M G , 23 (1992) 12-19. 17 Sch~ifer, S.S., The subdivision of primary afferents of de-efferented cat muscle spindles into two groups, Z. E E G - E M G , 23 (1992) 171-177. 18 Scott, J.J.A., The 'initial burst' of muscle spindle afferents with or without terminals on the bag t intrafusal muscle fibre, Brah7 Res., 585 11992) 327-329.