THE JOURNAL OF COMPARATIVE NEUROLOGY 372:465-485 (1996)

Horseradish Peroxidase Study of the Spatial and Electrotonic Distribution of

Group Ia Synapses on Type-Identified Ankle Extensor Motoneurons in the Cat

R.E. BURKE AND L.L. GLENN Laboratory of Neural Control, National Institute of Neurological Disorders and Stroke,

National Institutes of Health, Bethesda, Maryland 20814-4455

ABSTRACT Eight functionally identified group la muscle afferents from triceps surae or plantaris

muscles were labeled intraaxonally with horseradish peroxidase (HRP) in seven adult cats. Subsequently, HRP was injected into two to six homonymous or heteronymous a-motoneurons per animal (total = 221, each identified by motor unit type and located near the site of afferent injection. The complete trajectories of labeled afferents were reconstructed, and putative synaptic contacts on HRP-labeled motoneurons were identified at high magnification. Dendritic paths from each contact were also mapped and measured. A total of 24 contact systems (the combination of a group la afferent and a postsynaptic motoneuron) were reconstructed, of which 17 were homonymous, and seven were heteronymous.

Overall, homonymous contact systems had an average of 9.6 boutons, whereas heterony- mous contact systems had an average of 5.9 boutons. The average number of boutons found on type S motoneurons in homonymous contact systems was smaller (6.4, range 3-17) than in systems involving types FF or FR motoneurons (FF: 10.4, range 4-18; FR: 11.3, range 4-32). Neither of these differences were statistically significant. In contrast to earlier reports, a majority (15/24) of contact systems included more than one collateral from the same Ia afferent. The complexity (number ofbranch points) in the arborization pathway leading to each contact (overall mean 8.4 2 3.3) was virtually identical in all contact systems, irrespective of the type of the postsynaptic motoneuron.

The three-dimensional distribution of group Ia contacts was not coextensive with the radially organized dendrites of motoneurons: Dendrites oriented in the ventromedial to dorsolateral axis had the fewest (8%) contacts, whereas rostrocaudal dendrites had the most (63%) contacts. Nevertheless, contacts were widely distributed on the motoneuron surface, with few on and near the soma ( 5 200 km radial distance from the soma) or on the most distal parts of the tree ( 2 1,000 km). The boutons in individual contact systems also showed wide spatial and estimated electrotonic distributions; only 3/24 systems had all contact located within a restricted spatial/electrotonic region. The relations between these anatomical results and existing electrophysiological data on group Ia synaptic potentials are discussed. o 1996 Wiley-Liss, Inc.

Indexing terms: spinal cord, light microscopy, anatomical reconstruction

The neurophysiology of the monosynaptic excitatory transmission between group Ia, or primary, muscle spindle afferents and a-motoneurons has been studied intensively for more than 4 decades (for reviews, see Eccles et al., 1957; Burke and Rudomin, 1977; Redman, 1979; Lu scher and Clamann, 1992). This synaptic junction is one of the archetypes for conceptual models of the operation of chemi- cal synaptic junctions within the central nervous system of mammals, including such issues as quantization of synaptic

transmission (for reviews, see Redman, 1979; Walmsley, 1991) and the factors that control the amplitude of group Ia excitatory postsynaptic potentials (EPSPs) in different

Accepted March 16,1996. L.L. Glenn is currently at the Neuromechanics Research Group, East

Tennessee State University, P.O. Box 70658, Johnson City, TN 37614-0658. Address reprint requests to Dr. Robert E. Burke, Laboratory of Neural

Control, NINDS, Building 49, Room 3A50, National Institutes of Health, Bethesda, MD 20892-4455. E-mail: Reburke@helix.nih.gov

o 1996 WILEY-LISS, INC.

466

motoneurons (for reviews, see Burke, 1981; Burke et al., 1988; Luscher and Clamann, 1992). Group Ia excitation among motoneurons within a defined motor nucleus re- mains the chief model system for quantitative study of synaptic efficacy and its relation to the phenomenon of recruitment within a functionally defined population of neurons. Consideration of these physiological issues has required assumptions about the morphology of Ia afferent arborizations and about the numbers and spatial locations of group Ia synapses onto various species of a-motoneu- rons.

The use of intraaxonal injection of horseradish peroxi- dase (HRP) has permitted detailed and elegant studies of the anatomy of the intraspinal collaterals of functionally identified group Ia fibers (Brown and Fyffe, 1978; Ishizuka et al., 1979; see also Iles, 1976). Some studies have also demonstrated the ultrastructural features of HRP-labeled Ia boutons in the spinal ventral horn (Conradi et al., 1983; Fyffe and Light, 1984; Pierce and Mendell, 1993) and in Clarke’s column (see Walmsley, 1991). However, elucida- tion of the anatomy of group Ia synapses on identified motoneurons requires labeling both pre- and postsynaptic structures, with subsequent reconstruction of presumptive bouton contacts (Burke et al., 1979a,b; Brown and Fyffe, 1981; Redman and Walmsley, 1983a,b). The present work was designed to allow identification of the postsynaptic motoneurons according to their motor unit type (Burke et al., 1973). Preliminary reports of some aspects of this work have appeared elsewhere in abstract form (Burke et al., 1980; Glenn et al., 1982; Fleshman et al., 19841, in a preliminary research report (Burke et al., 1979b), and in two review articles (Burke et al., 1979a; Burke et al., 1988).

R.E. BURKE AND L.L. GLENN

MATERIALS AND METHODS Experiments were performed on a total of 25 adult cats of

either sex (body weight 1.5-3.0 kg). Details of surgical preparation and identification of motor unit types using intracellular recording and stimulation techniques have been given elsewhere (Burke et al., 1973; Cullheim et al., 1987a,b). Briefly, all surgical procedures were done under Halothane anesthesia. After tracheotomy, cannulas were inserted into one carotid artery to monitor blood pressure and into a cephalic vein for fluid administration. The medial (MG) and lateral (LG) gastrocnemius, soleus (SOL), and plantaris (PL) muscles were freed of surrounding tissue, preserving their blood supplies, and the individual tendons were tied with no. 2 braided silk suture to small steel hooks for independent attachment to a strain gauge. The muscle nerves of MG, LG plus SOL, and PL were dissected in continuity. The lumbosacral spinal cord was exposed by laminectomy. All dorsal and ventral roots were left in continuity.

The animal was mounted in a steel frame that immobi- lized the spinal column and the left hindlimb. Muscle nerves were mounted on flexible bipolar platinum wire electrodes for stimulation. A monopolar platinum ball electrode was placed at the dorsal root entry on the left to monitor incoming afferent volleys. Before recording, halo- thane inhalation was discontinued and anesthesia was maintained with carefully graded intravenous doses of pentobarbital to maintain a surgical level of anesthesia.

Intraaxonal HRP injection into individual group Ia afferents

The left dorsal column was explored just medial to the dorsal root entry zone with glass micropipettes (tip diam- eter 1.0-1.5 pm) filled with 4% HRP in 0.2 M KCl and 0.05 M Tris, pH 8.6 (Snow et al., 1976). During tracking, the MG, LG, SOL, and PL muscles were subjected to slow (about 0.5 Hz) sinusoidal stretching by attaching their tendons to an electromechanical puller. An attempt was made to locate and characterize stretch-sensitive afferents extracellularly, but some were found only upon intraaxonal penetration. The muscle of origin of stretch-responsive afferent fibers was identified by antidromic invasion after stimulation of the individual muscle nerves, which also gave the orthodromic conduction time. The target muscle was then attached to an isometric strain gauge, and the afferent response during isometric twitches was recorded after stimulation of the muscle nerve. The afferent was then recorded during low-amplitude, high-frequency ( 5 100 pm, 100 Hz) sinusoidal muscle stretch. Merents that showed 1: 1 following, cessation of firing during the twitch, and had conduction times < 1.7 msec (implying conduction velocities 2 90 meters/second with the usual conduction distances of 2 150 mm) were judged to be group Ia afferents (Matthews, 1972).

After successful impalement of a group Ia afferent (rest- ing membrane potential shifts of from -30 to -75 mV), it was injected with HRP by iontophoresis using trains of repetitive positive current pulses (15-25 nA, 50% duty cycle, 20 msec duration), interrupted briefly once per second to examine the orthodromic action potential ampli- tude and shape. In the successful cases, current delivery continued for 20-40 minutes, and total current injections were 130-630 nA-minute (mean 376 nA-minute). Current injection was stopped when the orthodromic spike ampli- tudes declined to less than 5 mV. Because of rectification in the electrodes, it was impossible to follow the resting membrane potential once the current injection process had begun.

After completion of HRP injection, the electrode was left in place for about 5 minutes and then removed. The animal was then left undisturbed for 4-6 hours to allow initial diffusion of HRP into the intraspinal afferent arborization.

HRP injection into type-identified a-motoneurons

After the diffusion period, we attempted to impale a-mo- toneurons that innervated the muscle of origin for the injected Ia afferent (i.e., homonymous motoneurons) or of synergistic muscles (heteronymous motoneurons). Pipettes without HRP were used to locate the target motor nucleus by its antidromic field in order to minimize possible extracel- lular leakage of the tracer. The motor nuclei were ap- proached from the lateral surface of spinal cord after exposure by lifting the intact L7 and S1 dorsal roots, concentrating on the region near the site of afferent injec- tion. Micropipettes used for motoneuron recording and staining had larger tip diameters (1.5-2.5 pm) than those used for afferent penetrations, but they were filled with the same HRP solution.

After penetration of a suitable motoneuron (belonging either to the homonymous or to a synergist motor nucleus), records were made of the antidromic action potential and

MONOSYNAPTIC GROUP Ia CONTACT SYSTEMS 467

conduction time, of Ia EPSPs produced by muscle nerve stimulation, and, on occasion, of the cell input resistance (Burke, 1968). The target muscle was then attached to an isometric strain gauge and stretched to produce a constant initial force of 100 g with an electromechanical device. The impaled motoneuron was then stimulated with short (0.5 msec) depolarizing current pulses to generate single twitches and fused and unfused tetani in the muscle unit (Burke et al., 1973). Finally, a series of short tetani (13 pulses at 40 Hz) was delivered once per second for 2 minutes to deter- mine the unit’s resistance to fatigue. The presence or absence of “sag” in standardized unfused tetani and the 2-minute fatigue index calculated from the series of short tetani were used to determine the motor unit type (type FF, FR, or S), according to previously published criteria (Burke et al., 1973).

After motor unit type identification, as above, the moto- neuron was injected with HRP by iontophoresis using the same regimen as that described for afferent injections, but with shorter injection times (5-16 minutes; total currents 50-180 &/minute). Only one motoneuron was injected in each electrode track, which was photographed through the dissection microscope before electrode withdrawal for later identification in relation to spinal cord landmarks. After electrode withdrawal, the site of penetration was marked on the cord surface by a small spot of India ink. Four to six motoneurons were labeled in most experiments, separated by at least 1 mm, in a region from just caudal to the site of afferent injection to 5-6 mm rostral to it (see Figs. 1,2).

Motoneuron injections were usually accomplished in a period of less than 4 hours. Within 45 minutes of comple- tion of the last injection, additional pentobarbital was given to produce deep anesthesia, and the animal was perfused transcardially with about 800 ml of warm saline and 3 liter of dilute aldehyde fixative (1.25% glutaraldehyde and 1% paraformaldehyde in 0.1 M phosphate buffer). The lumbosa- cral spinal cord was removed immediately and placed in fresh fixative overnight at 4°C. Finally, the peripheral conduction distances from the sites of nerve stimulation to the spinal cord were measured.

Tissue preparation The spinal cord block centered on the region containing

labeled motoneurons (segments S1 to L6) was prepared after fixation. Spinal segment boundaries were identified by dorsal root entry zones and were marked with shallow razor cuts in the dorsal columns. In some cases, the base of tissue block was trimmed to lie at approximately 45” to the dorsoventral axis (see below). The final tissue block was photographed a t x 3 magnification to show the ink marks at electrode sites in relation to the block boundaries and to ensure accurate recovery of specific labeled cells. In some cases, the fixed block was then placed in 30% sucrose in phosphate buffer at 4°C for 24-36 hours and then frozen for sectioning on a sliding microtome. In other cases, serial sections were made on avibratome. All sections were cut at a thickness of 75 pm.

A few spinal cords were cut in the coronal plane (see, e.g., Fig. 5), but most were sectioned parasagittally, either in the conventional dorsoventral plane or in a plane at an angle of about 45“ to the vertical, in order to match the trajectory of group Ia collaterals as they coursed through the spinal gray matter (Figs. 1, 3, insets; see Ishizuka et al., 1979). This plane of sectioning greatly facilitated the eventual recon-

struction of the labeled Ia afferents from serial sections, because most of the collateral arbors were contained within a relatively few sections.

Serial sections were placed in separate wells in a special plastic tray that allowed histochemical staining of the entire set of sections at once. Frozen sections were pro- cessed by the cobalt-enhancement method (Adams, 1977) by using diaminobenzidine as chromogen. Vibratome sec- tions were processed by a variation of the glucose oxidase method (Itoh et al., 1979) also using cobalt enhancement. Sections were mounted serially in register on glass slides, air dried overnight, and coverslipped without counterstain.

Tissue reconstruction In all of the material to reported here, every labeled

motoneuron was positively identified according to its loca- tion in the original photographs taken during the experi- ment. All labeled afferent arborizations were identified as belonging to the stained afferent by tracing them back to the parent afferent in the dorsal column. Spinal cords in which there was any ambiguity in identification were excluded from the analysis.

The first stage in reconstruction was photographic map- ping of the regions containing labeled afferents and moto- neuron somata or dendrites. Photomicrographs of overlap- ping regions were made at moderate magnification ( x 4 objective, ~8 ocular) on Kodak Technical Pan film, devel- oped in Perfection Micrograin developer (Perfection Prod- ucts, Inc.) to maximize contrast. Negatives were printed on 8 x 10 inch sheets of Kodalith Ortho type 3 film and developed in Kodak D-76 or Edwal Litho-F developer, producing high-contrast positive images of HRP-labeled structures on an almost transparent background. These positive prints were assembled into photomontage maps from each section.

Labeled structures in the photomontage maps were drawn on tracing paper that combined the location of labeled structures in aligned serial sections. Ambiguous or overlapping structures were identified by comparing the photomontage maps with the actual sections under the microscope. In this way, it was possible to map the entire region containing labeled structures at high effective magni- fication, even though the regions in question extended over many square millimeters.

The labeled afferent in each spinal cord was traced in the dorsal column from the injection point to the most distant rostral and caudal points that were visible (up to 15 mm), noting the positions of all collaterals. The trajectory of each collateral was then followed into the ventral horn. Regions of overlap between Ia collaterals and labeled motoneuron dendrites were then examined with high magnification ( ~ 4 0 0 ; Nikon Biophot microscope), and the positions of all apparent contacts between identified Ia collaterals and HRP-labeled dendrites were noted on the maps. Each apparent “contact” was examined later under oil-immer- sion magnification and was verified by two independent observers. Although all appositions between Ia collateral axons and labeled postsynaptic structures were noted, only those appositions involving presynaptic swellings were counted as contacts (Brown and Fyffe, 1981). The final stage of mapping consisted of tracing and measuring each postsynaptic dendrite from an apparent contact back to its parent cell body (see below).

Fig.

1.

Rec

onst

ruct

ion

of a

cat

spi

nal c

ord,

in

whi

ch t

wo

sole

us (

SOL)

gro

up I

a af

fere

nts

(bla

ck a

nd r

ed t

raci

ngs)

wer

e in

ject

ed in

the

dor

sal c

olum

n (s

ites

mar

ked

with

thic

k ar

row

s),

alon

g w

ith s

ix tr

icep

s su

rae

a-m

oton

euro

ns (b

lue)

, ide

ntif

ied

by m

uscl

e an

d m

otor

uni

t typ

e.

Den

drite

s of

tw

o ce

lls (

seco

nd a

nd f

ourt

h fr

om l

eft)

wer

e su

ppre

ssed

for

cla

rity

. T

he t

otal

nu

mbe

r of

cont

acts

from

eac

h af

fere

nt to

eac

h m

oton

euro

n ar

e in

dica

ted.

Not

e th

at c

olla

tera

ls

from

the

two

affe

rent

s wer

e in

regi

ster

in so

me p

lace

s and

not

in o

ther

s. M

ain

bran

ches

from

the

two

affe

rent

s w

ere

dire

ctly

app

osed

with

in a

n af

fere

nt m

icro

bund

le a

t the

hor

izon

tal a

rrow

nea

r

the

inje

ctio

n si

te fo

r aff

eren

t 3. I

njec

tion

of a

ffer

ent 3

was

term

inat

ed p

rem

atur

ely,

and

onl

y th

e tw

o ne

ares

t co

llate

rals

wer

e fil

led

adeq

uate

ly w

ith h

orse

radi

sh p

erox

idas

e (H

RP)

. Col

late

ral

bran

chin

g at

the

leve

l of t

he in

term

edia

te n

ucle

us is

indi

cate

d by

the

labe

lled

doub

le a

rrow

. Tw

o co

llate

rals

fro

m a

ffer

ent 4

and

the

SOL

mot

oneu

ron

mar

ked

with

ast

eris

ks a

re sh

own

in m

ore

deta

il in

Fig

ure

2. In

set s

how

s the

ori

enta

tion

of th

e ob

lique

long

itudi

nal p

lane

of s

ectio

ning

(see

M

ater

ials

and

Met

hods

). L

G, l

ater

al g

astr

ocne

miu

s; M

G, m

edia

l gas

troc

nem

ius;

PL,

pla

nari

s.

MONOSYNAPTIC GROUP Ia CONTACT SYSTEMS 469

Fig. 2. Stereo drawing of two SOL collaterals (afferent 4) making contact with the SOL type S motoneuron from the spinal cord shown in Figure 1 (marked there with asterisks). The large ovals denote two terminal arbors of one collateral, and the small black circles show the locations of bouton contact regions (contact system W in Fig. 18).

Data processing Using the photomontage maps, the position of each

contact was plotted in a three-dimensional space centered on the soma of the postsynaptic motoneuron. The positions within the plane of sectioning were measured from the maps, and contact depth within the section was estimated from the original slides by using the focus calibration of the microscope at x 400 magnification. Positional accuracy in all three planes was estimated to be * l o km. These positional data were corrected for angle of sectioning in those cords that were cut at an oblique angle (see above). The final data set contained the positions of apparent contacts in three-dimensional space centered on the target motoneuron soma (see Fig. 11).

The length and diameters of all cylindrical dendritic segments on the path from each contact to the parent soma were measured (ek and dk = the length and diameter of the k-th path segment, respectively), by using the method detailed elsewhere (Cullheim et al., 1987a; Fleshman et al., 1988). These were used to estimate the electrotonic dis- tance between the contact and the soma. The characteristic length, A, of a membrane cylinder with diameter, d, is given by

A=/%! Ri 4 '

where R, is the specific resistivity of the dendritic mem- brane (a cm2), and Ri is the specific resistivity of the motoneuron cytoplasm (a cm; see Rall, 1977). The electro- tonic length, X, of a membrane cylinder with physical

length, e, is given by

By using Equation 2, the electrotonic distance, Xj, between the j-th putative Ia synapse and the motoneuron soma is given by

(3)

where n is the number of segments in the path between the contact and the soma. Thus, the electrotonic distance to any contact (Figs. 16, 17) is proportional to the summated term and can be easily calculated for arbitrary values of R, and R, by using a tabulation of X(ek dk-0.5).

No corrections for tissue shrinkage during processing were applied to the dendritic measurements. Most of the present material came from spinal cords that were sec- tioned in the parasagittal plane. Typical longitudinal shrink- age in the sagittal plane, in our hands, was about 6% (see Cullheim et al., 1987a). Nonparametric statistical tests were used for most of the comparisons in this paper, because the small data samples showed skewed distribu- tions and large variances.

RESULTS The results to be reported were based on a total of eight

group Ia afferents (two SOL afferents were successfully

470 R.E. BURKE AND L.L. GLENN

rostral + 500 p n

MNP24

MGFF MGFF S O L S ” MGFR MG S A: 10 A: 10 A 3 c: 3 none detected B: 7 C: 3 D: 1 D: 3 c: 1 D 2

A-E: Oblique longitudinal reconstruction of another spinal cord to illustrate an example of overlap between collateral territories (a rostral branch from collateral A crossed the territories of E D ) and a less than ideal

Fig. 3. relation between collaterals and the location of the labeled motoneurons (dendrites suppressed for clarity). The numbers of contacts from each collateral are indicated below each motoneuron soma.

injected in one animal; see Fig. 1) and 22 type-identified a-motoneurons (17 triceps surae: nine MG, three LG, five SOL, and five PL). This sampling resulted in a total of 24 contact systems (the combination of a given Ia afferent and a given motoneuron), with an equal number of motoneu- rons of each motor unit type (eight each for types FF, FR, and S). Two of the contact systems (systems A and B; see Fig. 17) were described in a preliminary report (Burke et al., 1979b); the third system in that report was not included in the present sample. Criteria for inclusion in the present sample were: 1) All HRP-labeled collaterals in the ventral horn were well stained in the vicinity of HRP-labeled motoneurons and were traced to the parent axon in the dorsal column; 2) the afferent injection site in the dorsal column was clearly identified and was consistent with photographs taken during the experiment; 3 ) ventral horn collaterals spatially overlapped the dendritic fields of HRP- injected motoneurons; and 4) fixation and cutting artifacts did not prevent reconstruction of all of the relevant spinal regions. When these criteria were met, putative contacts were found between one or more labeled Ia collaterals and

each homonymous or heteronymous a-motoneuron that was labeled.

General features of group Ia collateral arborizations

The general characteristics of group Ia afferent morphol- ogy have been described in detail elsewhere (Brown and Fyffe, 1978; Ishizuka et al., 19791, and the afferents in the present study were entirely consistent with these earlier reports. After bifurcation near the point of dorsal root entry, the descending afferent branch was about half the diameter of the ascending branch. All but one of the dorsal column injection sites were in the ascending branch, and labeling of collaterals from the descending branch was often less complete than that in the ascending branch. Collateral branches entering the gray matter were more closely spaced near the site of dorsal root entry than at greater rostral or caudal distances (Figs. 1, 3 ) . The territories of neighboring collaterals sometimes overlapped when they were closely spaced, and, in a few cases, some collateral branches were

MONOSYNAPTIC GROUP Ia CONTACT SYSTEMS 471

interlaced with those of neighboring collaterals over consid- erable distances (see, e.g., Fig. 3).

The reconstruction in Figure 1 shows the trajectories of two independent SOL group Ia afferents that were success- fully labeled in the same animal along with six homony- mous or heteronymous a-motoneurons. For clarity, fine preterminal branches in each collateral and the dendritic trees of two motoneurons were not drawn. The territories of collaterals belonging to the two different afferents over- lapped quite closely at several locations, whereas they were staggered in other places. In fact, major branches of the two were closely apposed in the same microbundle of afferents emerging from the dorsal columns at the point indicated by the horizontal arrow (Fig. 1). The HRP staining in afferent 3 was less complete than that in afferent 4, although both showed what appeared to be complete labeling of collaterals originating from the ascending branches in the central 4-5 mm near the injection sites. The numbers of contacts from each collateral onto the various motoneurons appeared to be dependent on the spatial coincidence between motoneu- ron dendrites and collateral territories and on the complete- ness of collateral labeling. For example, the most caudal motoneuron (MG type S) showed no identifiable contacts, because HRP label from both parent axons was too faint at this distance from the injection site.

The stereo drawing in Figure 2 (for method, see Glenn and Burke, 1981), which is from the same spinal cord shown in Figure 1, provides a three-dimensional view of the spatial overlap between two arborization of a Ia collateral from afferent 4 (Fig. 1, asterisks) and the dendritic territory of the second SOL motoneuron from the left, which re- ceived 17 contacts from this afferent. When Figure 2 is viewed stereoscopically, it is apparent that the mediolateral extent of the afferent arbors was far less than that of the motoneuron’s dendritic field.

The reconstruction in Figure 3 illustrates an example in which there was considerable overlap between the territo- ries of adjacent collaterals, particularly collaterals A and B. Collateral A also had one branch that ran rostrally for over 1 mm to overlap with the territories of collaterals C and D. In this case, all of the labeled somata were somewhat offset from the region of richest arborization, but, despite this, the most caudal motoneuron (type FF) exhibited a rela- tively large number of contacts from collaterals A, B, and C. This example again illustrates that the number and spatial location of boutons from group Ia afferents on any postsyn- aptic motoneuron depend on the spatial relation and local richness of the labeled collaterals and the dendritic terri- tory of the labeled cell. None of these features could be predicted accurately during the experiments.

Morphology of individual contact systems The term contact system will be used to refer to the

collection of synaptic boutons from a given Ia afferent that make apparent direct contact with the soma and/or den- drites of a given HRP-labeled motoneuron. Figure 4 illus- trates oil-immersion photomicrographs of putative mono- synaptic contacts (Fig. 4, arrows) between an SOL group Ia afferent and ventrolaterally directed dendrites of the SOL a-motoneuron shown in Figure 2. In each case, HRP- labeled swellings traced to a labeled arborization (see above), at high magnification, appeared to be directly apposed to HRP-labeled dendrites traceable to the motoneu- ron soma. The diagrams on the bottom (Fig. 4) show that there were two terminal boutons (A right and B) and three boutons en passant (A left and C). Although no ultrastruc-

tural confirmation was possible given the nature of these experiments, a number of studies have found that struc- tures with this appearance at the light microscopic level, in fact, do have the ultrastructural characteristics of true synaptic contacts (Fyffe and Light, 1984; Pilowski et al., 1990; FyfFe, 1991; Walmsley, 1991; Pierce and Mendell, 1993).

Each collateral that gave rise to such contacts on any labeled motoneuron was mapped from the contact to the parent afferent in the dorsal column (see above), allowing reconstruction of the pattern of branch points upstream from each contact in addition to its location on the postsyn- aptic motoneurons (see Materials and Methods). The cross- sectional reconstruction in Figure 5 shows an example of an MG group Ia collateral that made six contacts on and near the soma of a type FF MG motoneuron. The two most proximal bifurcations within the collateral produced major branches (Al, Az, and B) that ultimately supplied contacts onto the motoneuron. More distal branch points within the ventral horn generated an arborization near the target motoneuron that contributed en passant and terminal boutons to the target cell (Fig. 5, inset, branch B). This is the only example in the present work in which most or all of the contacts were juxtasomatic. However, this motoneuron also received one contact from an adjacent collateral on a quite distal dendrite, far from the soma (Fig. 17, contact system E).

The contact system in Figure 6A illustrates an example in which two adjacent collaterals produced synaptic contacts on a single motoneuron in widely dispersed regions. Five contact regions (12 boutons) were distributed to middle and relatively distal parts of two rostral dendrites from collat- eral A, and a sixth region received two boutons from the adjacent collateral B. Within each contact region, bouton contacts were closely spaced (see Fig. 41, as reported for some examples in earlier studies (Brown and Fyffe, 1981; Redman and Walmsley, 1983b). A “wiring diagram” of this contact system is shown in Figure 6B, including the relative locations and numbers (6-9) of branch points between the parent axon in the dorsal column and each putative contact.

Figure 7 is an attempt to convey an impression of the intricate network of presynaptic axons, postsynaptic den- drites, and contact regions found for three adjacent Ia collaterals from a PL group Ia afferent that established contact regions (circles) on four type FR PL motoneurons. Motoneuron dendrites and collateral arbors other than those involved in contacts were suppressed. The location and density of contact regions clearly depended on the local densities of adjacent collaterals and dendrites. There was some rostrocaudal overlap between the territory of collat- eral B with those of collaterals C and D, producing a spatial mix of adjacent contact regions from different collaterals on different motoneurons.

Figure 8 shows the wiring diagram of motoneuron 6 in the center of Figure 7, which received the largest number of contacts (n = 32) in the present sample (three adjacent collaterals making contacts onto eight of this cell’s 12 dendrites). Despite wide spatial dispersion, the electrotonic map of this contact system was relatively compact (system P in Fig. 17). The Figure 8 diagram also shows that the numbers of branch points leading to any given bouton on this cell were quite variable, ranging from 3 to 16. Like the earlier examples, the relevant branch points included some proximal to the parent axon (collateral B) as well as in preterminal arborizations.

The mean number of branch points in the paths from parent axon leading to each of the 204 contacts studied

472 R.E. BURKE AND L.L. GLENN

Fig. 4. Top: Oil-immersion photomicrographs of three contact regions (A-C) showing HRP-labeled terminal and en passant boutons (arrows; black areas in bottom diagrams) from an SOL group Ia afferent (the right collateral in Fig. 2) making contact with relatively thin,

beaded dendrites of an SOL motoneuron (stippled in bottom diagrams). Bottom: Diagrams of the photomicrographs A-C to clarify presynaptic axonal branching and the anatomy of the bouton contacts on the motoneuron dendrites (shading).

were very similar in contact systems ending on the different motor unit types (mean t S.D.; number in category): type FF: 8.6 c 3.4 (n = 70); type FR: 8.1 2 3.0 (n = 83); type S:

8.7 * 3.5 (n = 51). There was no significant difference between the numbers found in these groups (two-way analysis ofvariance; P > 0.1).

MONOSYNAF'TIC GROUP Ia CONTACT SYSTEMS 473

. ....

. . . .* .-*.".'..* *....; : ?' '%*

...'* . . * *

MNP22c3 *. ..'.., .. : : * : ..

Fig. 5. Cross-sectional reconstruction of an entire MG group Ia collateral that had six contacts (circled) on and near the soma of a type FF MG motoneuron (contact system E in Fig. 18). Note that both of the major branches (A,B) ofthis collateral, formed in the dorsal gray matter

above the intermediate nucleus, supplied contacts on this cell. The enlarged view (inset) shows that local branching also contributed preterminal axons to boutons on this cell. See text for details.

474 R.E. BURKE AND L.L. GLENN

caudal + 500 pm

Collateral B

I

I

MNPlEcS

B

I erminal bouton

0 En Passant bouton

Fig. 6. A: Semidiagrammatic reconstruction of a contact system (system C in Fig. 17) between an LG group Ia afferent and an LG type FR motoneuron (MN). Collateral A supplied 12 contacts distributed within five contact regions (circles; contact numbers given for each) on two rostra1 dendrites. Collateral B contributed two contacts on a relatively distal site on a caudally directed dendrite. B: Wiring diagram

of the contact system shown in A, with terminal boutons denoted by solid circles and en passant boutons denoted by open circles. The preterminal branching pattern from the parent axon in the dorsal column to each contact is indicated. Distances are approximations and are on different scales for the motoneuron and collaterals.

General features Of group Ia Contact systems and Fyffe, 1981). The 17 homonymous systems showed a The total numbers of contacts in each of the 24 contact

systems were unrelated to the conduction velocities (CVs) of the parent group Ia afferents (Fig. 9-4; see also Brown

much larger range (3-32) in total bouton number Per contact system than the seven heteronymous systems (4-9). There were six contact systems with more than ten

MONOSYNAPTIC GROUP Ia CONTACT SYSTEMS 475

boutons, all of which were homonymous and included motoneurons of all three motor unit types (two type FF, three type FR, and one type S). The majority of contacts were en passant in most contact systems (18/24; Fig. 9B); the reverse was true in only three cases (one type FF and two type S).

The majority (15/24) of contact systems in the present sample involved more than one collateral from the same parent Ia afferent. The scatter diagram in Figure 10 shows that there was a positive correlation between the number of collaterals involved and the number of contacts. This relation was statistically significant for the homonymous contact systems (rho = 0.71; P < 0.005, Spearman-Rank correlation test).

Table 1 gives the mean overall numbers of contacts in homonymous vs. heteronymous systems in relation to postsynaptic motor unit type. All heteronymous contact systems belonged to the triceps surae group; only one PL afferent was injected, and all five motoneurons recovered in that spinal cord were PL cells. Considered without refer- ence to motor unit type, heteronymous systems on average had about 40% fewer boutons than homonymous systems. This difference was not significant (P > 0.1, Mann-Whitney U test). The average number of boutons found in type S motoneurons was about 40% smaller than those in type FF and FR cells taken together for the 17 homonymous contact systems. This difference was of borderline significance (P = 0.08, Mann-Whitney U test). However, when the numbers for homonymous systems were split into three groups by motor unit type, the differences in contact numbers were not significant (P > 0.1, Kruskal-Wallis test). The numbers of heteronymous contacts were similar in the different motor unit types.

TABLE 1. Mean Number (kS.D.1 of Bouton Contacts in 24 Contact Systems’

System type FF FR S All

Homonymous 10.4 2 5.9 (5) 11.3 t- 9.9 (7) 6.4 2 6.0 (51 9.6 f 7.7 (17) Heteronymous 6.0 2 0.0 (3) 4 (1) 6.3 ? 2.5 (31 5.9 2 1.7 (7) Both 8.8 f 5.9 (8) 10.4 2 6.2 (8) 6.4 f 4.7 (81 8.5 f 6.7 (241

‘Numbers of contact systems are indicated in parentheses.

Three-dimensional distribution of Ia contacts and postsynaptic membrane

Although group Ia bouton contacts are widely dispersed on the surface of a-motoneurons, they are not found on those parts of the dendrit,es that extend beyond the ventral horn region reached by group Ia collaterals, which project into the ventral horn in a relatively narrow band at an angle of about 45” (Brown and Fyffe, 1978; Ishizuka et al., 1979; see also Fig. 6). Figure 11 illustrates the three-dimensional location of group Ia contacts when the data from all 24 contact systems are combined onto a common three- dimensional coordinate space centered on the soma of each motoneuron. The resulting “cloud” of Ia contacts (Fig. 11, asterisks) surrounds the “soma” of a hypothetical motoneu- ron. The cross-sectional view shows that the bouton cloud was flattened in the dorsolateral to ventromedial direction, consistent with the trajectories of triceps surae group Ia collaterals in the cat lumbosacral cord (Ishizuka et al., 1979). A rectangular region of indefinite length in the rostrocaudal direction and angled at 45” with respect to dorsoventral, with the dimensions given in Figure 11

(cross-sectional view), contained about 95% of the 204 observed boutons.

To examine the observed three-dimensional distribution of Ia boutons in relation to the location of motoneuron dendrites, we superimposed the rectangular Ia contact region shown in Figure 11 on ten completely reconstructed triceps surae a-motoneurons, including five MG or LG (two type FF, two type FR, one type S) and four SOL (all type S). Six of these cells were described by Cullheim and coworkers (1987a,b), and, subsequently, the others were recon- structed. The motoneurons that actually received Ia con- tacts in most cases were not suitable for complete recon- struction because of the density of labeled structures in those spinal cords. Figure 12 illustrates an example of a type FF LG motoneuron with soma centered within the Ia box. Most of the rostrocaudal dendrites were within the Ia box, but dendrites extending in other directions, especially in the dorsolateral (DL) to ventromedial (VM) direction, extended well beyond it (as shown in Fig. 12, cross-sectional view). The same exercise was repeated with the other nine cells in the available sample, with similar results.

The cell membrane area, including the soma, within the Ia box was calculated for each motoneuron and compared with the total membrane area of each cell. The graph in Figure 13A illustrates a comparison of the average total membrane area of the ten motoneurons (Fig. 13A, open circles, +S.D.) and the membrane area within the rectangu- lar Iacontact region (Fig. 13A, solid circles, -S.D.). The two curves diverged beyond 300 pm from the soma, as expected from the region dimensions (Fig. 11). The average mem- brane area within the contact region was approximately 60% of the total area for this sample of cells. The graph in Figure 13B compares the radial distribution the membrane area within the Ia contact region (Fig. 13B, thick curve; S.D. is indicated by the shaded area) and that of group Ia contacts (Fig. 13B, diamonds; obtained from the data illustrated in Fig. ll), both expressed as percent of the respective totals. The number of Ia contacts peaked be- tween 300 and 600 pm away from the soma, and most of the data was within one S.D. of the mean area.

Successive group Ia collaterals, as discussed and illus- trated above (see also Brown and Fyffe, 1981; Ishizuka et al., 1979), had territories that were relatively restricted rostrocaudally. The collaterals form sagittally organized, slab-like regions that encounter the radially organized dendrites of motoneurons. Accordingly, we analyzed the spatial distribution of motoneuron membrane and bouton contacts in 100 pm sagittal slabs along the longitudinal axis of the spinal cord, as indicated on the sagittal views in Figures 11 and 12. Figure 13C compares the sagittal distribution of membrane area (see also Fig. 15A) with the distribution of bouton contacts in the present sample, as in Figure 13B, but with the sagittal distance on the abscissa (area rostral and caudal to the soma combined). The distributions show a similar pattern of departure from coherence, with fewer than expected contacts close to the soma and more than expected from 200 to 500 pm along the rostrocaudal axis.

The spatial localization of Ia bouton contacts and moto- neuron membrane area were also analyzed according to radially organized pyramidal “hexants” with apices at the soma (Cullheim et al., 1987b) that permit quantitation of elements along orthogonal axes. In this case, the hexant array was rotated by 45” around the rostrocaudal axis, (Fig. 14, diagrams) to align with the trajectory of Ia collaterals.

MONOSYNAPTIC GROUP Ia CONTACT SYSTEMS 477

Fig. 8. Wiring diagram of preterminal collateral branches and contacts from three adjacent PL afferent collaterals onto a type FR PL motoneuron (cell 6 in Fig. 9; contact system P in Fig. 17). This was the largest and most complex contact system (n = 32) in the present sample.

Figure 14 shows the number of putative contacts that occurred within each of six hexants (Fig. 14, diamonds) along with the membrane area within the Ia contact region in each hexant (both as percent of sample). There was reasonable agreement between the distributions in the rostral and caudal hexants, but there was more divergence in the other directions. Over 60% of the contacts were found in the rostral and caudal hexants, and about 22% were found in the dorsomedial and ventrolateral hexants (along the axis of Ia collateral trajectories). Only 7% of contacts were found within the ventromedial and dorsolateral hex- ants, orthogonal to the collateral trajectories.

The results shown in Figures 13 and 14 indicate that the spatial distribution of group Ia contacts is approximately proportional to the amount of postsynaptic membrane area

Fig. 7. Spatial distribution of contact regions generated by three adjacent PL group Ia collaterals on four PL motoneurons (all type FR). There was some overlap between collateral territories, but it was less than the case in Figure 2. Note the density of contact regions (circles) on dendrites oriented in the rostrocaudal direction.

available within the region of overlap (the Ia contact region) between Ia collaterals and the motoneuron dendrites. The data also suggest that fewer than expected Ia contacts are formed relatively close to the cell somata, whereas there is an excess between 300 and 600 ym away (radial direction).

Collateral placement and total bouton numbers

The number of contacts on a given motoneuron, as noted above (Figs. 1,3,7) , depends in some measure on the spatial relation between the presynaptic collaterals, disposed along the rostrocaudal axis of the spinal cord, and the postsynap- tic cell. Figure 15A shows that the average membrane area of the ten comparison motoneurons fell o f f in an approxi- mately logarithmic manner with distance rostral and cau- dal to the motoneuron somata. We used this simple relation to quantitate the spatial relation between a motoneuron and the group Ia collaterals contributing contacts on it as a “distance score” (note equation) that decreased logarithmi- cally with the distance between the estimated center of each collateral and the soma of a motoneurons receivingcontacts from that collateral: the lower the score, the more distant

478 R.E. BURKE AND L.L. GLENN

A 35

30 7 ‘A

t 25 c G 6 20

SOL” , i 0 L-

90 95 100 105 110 115 120 125

Afferent Conduction Velocity ( d s )

/ /

0 / = /

/

A I/ -- I I

0 5 10 15 Number of Terminal Contacts

Fig. 9. A Plot of the number of contacts and the conduction velocity of the parent Ia afferent for 24 contact systems showing lack of systematic relation. Vertical lines connect contact systems associated with the same afferent, and the muscle of origin is given above each line. The two SOL afferents marked with asterisks were from the same spinal cord (shown in Fig. 1). Symbols denote homonymous (solid symbols) and heteronymous (open symbols) connections and motor unit types of postsynaptic motoneuron (inset legend). B: Scatter diagram comparing the number of terminal (abscissa) vs. en passant houtons (ordinate). Most (18124) contact systems had more en passant than terminal boutons (dashed line shows equality), irrespective of afferent identity or postsynaptic motoneuron type. Symbol key as in A.

the spatial interaction. When more than one collateral made contact with a given cell, the distance scores were simply added. The scatter diagram in Figure 15B shows that there was a positive correlation (r = 0.66; P < 0.01) between the number of contacts and the distance scores for homonymous contacts. The average distance score for homonymous type F (FF and FR motoneurons together) was larger (0.88) than that for type S motoneurons (0.631, but the difference was not statistically significant (P > 0.1, Mann-Whitney U test).

Electrotonic dispersion of group Ia contacts The lengths and diameters of all dendritic segments

along the path between the soma and each of the 204 group Ia contacts identified in this material (see Materials and Methods) were measured, permitting calculation of the

30

v1 Y

0 cs 0

rcl 0

e

5 20 u

3 E 2 10

0

A +

rn A

+

1 2 3

Number of Collaterals Fig. 10. Plot of number of contacts vs. the number of collaterals

involved, broken down by afferent identity and postsynaptic motoneu- ron type.

electrotonic distance of each contact from the soma, given assumed values for specific membrane (R,) and cytoplas- mic (R,) resistivities. Figure 16 shows a histogram of electrotonic distances to each contact, calculated by using R, = 70 R cm and three values for R, (2,000, 8,000, and 32,000 R cm2) that span estimated dendritic R, values for a-motoneurons (11,000-20,000 W cm2; Fleshman et al., 1988). The superimposed curve of cumulative probability (Fig. 16, circles) shows that about 60% of the contacts were made in the proximal one-fourth of the electrotonic range, and over 80% were found within the proximal half. Note that the electrotonic range did not include the most distal dendritic tips (cf., Fig. 13A).

The electrotonic dispersion for individual contact sys- tems is shown in Figure 17, separated according to motor unit type, with each group arranged in order of increasing mean electrotonic distance (Fig. 17, open arrowheads). Many contact systems exhibited a relatively wide electro- tonic distribution of contacts. Only three showed a single concentration (systems N, B, and D), although some others [e.g., systems E (see also Fig. 5 ) and Ul were concentrated, except for a single more distal bouton. There was no apparent difference in the average electrotonic distances or the pattern of dispersion based on the motor unit type of the postsynaptic motoneurons.

DISCUSSION The present results differ in several respects from earlier

reports on the anatomy of individual HRP-labeled group Ia contact systems on labeled a-motoneurons (Burke et al., 1979a,b; Brown and Fyffe, 1981; Redman and Walmsley, 198313). Two of the contact systems described in Burke et al. (1979a,b) are included in the present sample (Fig. 17, systems A and B); the numbers of contacts in system A were

MONOSYNAPTIC GROUP Ia CONTACT SYSTEMS 479

Sagi

* * *

Dorsal t,

I 00 Urn

Lateral

Fig. 11. Projections of the three-dimensional locations of houton contacts from all 24 contact systems superimposed on a coordinate space centered on the postsynaptic motoneuron soma in each case (shaded diamonds). A rectangular contact region, oriented along the spinal cord at a 45" angle with respect to the dorsoventral axis to match the general trajectory of group Ia collaterals and with the dimensions shown in the cross-sectional view (unconstrained in the rostrocaudal axis), contained about 95% of the Ia contacts.

revised downward from six to five, and one system (three SOL contacts on an LG motoneuron) was omitted from the current sample, based on the criteria given in Materials and Methods.

Numbers and spatial dispersion of group Ia contacts

The average number of contacts per motoneuron in the present sample (8.5 ? 6.7, range 3-32; Table 1) was more than twice that found by Brown and Fyffe (1981; mean = 3.4, range 2-5) for a sample of ten group Ia contact systems on triceps surae and plantaris motoneurons. Redman and Walmsley (1983a,b) reported three, four, and eight contacts in three systems in which all contacts could be traced to a

Cross-section

Fig. 12. Two views of a three-dimensional projection of the group Ia contact region (from Fig. 11) superimposed on a reconstruction of a type FF LG a-motoneuron. The dendrites within the Ia contact region are denoted by thick lines. Many dendritic branches extended beyond the limits of the contact box, especially in the ventromedial to dorsolat- eral axis.

single group Ia collateral. The apparent discrepancy in contact numbers disappears, however, when considering the number of contacts in systems in which all contacts could be traced to only one collateral (3-9; mean = 5.1 +- 1.8; see Fig. 11). In the earlier reports, including that from this laboratory, all Ia contacts found on a given motoneuron originated from a single group Ia collateral. Indeed, Brown and Fyffe (1981) reported that this was the case even when motoneuron dendrites intersected the territory of neighbor- ing collaterals.

In contrast, the majority of contact systems in the present material (15/24) involved multiple collaterals from the same parent Ia axon, and there was a significant trend toward more contacts as collateral number increased (Fig. 10). In addition, many of the individual contact systems in the present material also exhibited wider ranges of spatial dispersion than those found in the previous studies, in part, due to innervation of individual motoneurons by multiple Ia collaterals. I t seems likely that these differences may be due in part to methodological details (see also below), particularly to our use of the oblique sagittal plane of sectioning and of photomontage mapping. Both of these

RE. BURKE AND L.L. GLENN

12

480

A I l l

I

15 d

2 r j v) * 10 u

5

Id Contact Region i s

0

c 30

0

500 1000 1500

Radial Distance (pm)

0 500 1000

Sagittal Distance (pm)

A Comparison of the radial distributions of average total membrane area (open circles, tS.D.1 and average area within the group Ia contact box (solid circles, -S.D.) obtained for ten fully reconstructed a-motoneurons (see text). The cell membrane area within the Ia contact region (including the somata) averaged about 60% of the total. B: Comparison of the radial distribution of average motoneuron mem- brane area within the Ia contact region (thick curve; 2S.D. indicated by stippled region) and that of observed group la contacts in the 24 contact systems (diamonds). Both sets of data are expressed here as percent of the respective sample (ordinate). C: Comparison of average membrane area (*S.D., as in B) in 100-km-thick slabs along the rostrocaudal (sagittal) axis (note the tick marks in sagittal views in Figs. 11 and 12) and the distribution of Ia contacts within the same slabs (diamonds). Again, data are expressed as percent of the respective samples.

Fig. 13.

methods greatly facilitated the creation of complete reconstruc- tions from extensive regions that included many overlapping motoneuron dendrites and Ia collaterals (see, e.g., Fig. 1).

L Rostra1 I Caudal Hexants

u - Dorso-Medial Ventro-Lateral 4

v) 'i

5 2 4 5 E 2

1 0

Radial Distance (pm)

Fig. 14. Plots of the numbers of contacts and of average membrane areas within the Ia contact region (both as percent of respective samples; ordinate) contained within mutually orthogonal hexant subdi- visions of perisomatic space (diagrams on the right; see Cullheim et al., 1987h3. The cube on which the hexants are based was tilted 45" in the transverse plane to align with the trajectory of group Ia af€erents. The distributions of contact numbers were generally similar to those of available membrane area in the rostra1 and caudal hexant subdivisions but showed more deviation in the other directions.

On the other hand, the general morphology found in the present work for the distribution of boutons from indi- vidual collaterals was quite similar to that described by Brown and Fyffe (1981; see also Burke et al., 1979a,b; Iles, 1976). Some contacts occurred on motoneuron dendrites within a restricted contact region from relatively local branches in the terminal arborization, whereas more dis- tant branching sometimes produced quite dispersed contact regions, even from the same collateral (see, e.g., Figs. 6-8).

Collateral branching in Ia contact systems It has long been recognized that axonal branch points

could be sites for failure of orthodromic action potential propagation and consequent modulation of synaptic trans- mission (for review, see Burke and Rudomin, 1977). Al- though there is little evidence that branch point failure contributes to group Ia EPSP modulation (Lev-Tov et al., 1983; Redman and Walmsley, 1983a; Davis et al., 19851, it has been suggested that this mechanism could account for the large range in group Ia EPSP amplitudes observed in cat motoneurons (Liischer et al., 1983; Lu scher and Clamann, 1992). Liischer and Clamann (1992), in fact, have proposed a model of group Ia contact systems based on the assumption that there are fewer branch points in Ia arbors that project to small a-motoneurons (mostly type S moto- neurons) than in those projecting to larger motoneurons

MONOSYNAPTIC GROUP la CONTACT SYSTEMS 481

A I

g 0.8

c rn . 3

m 2 2 0.6

g a0 4 0.4

0 50 100 200 500 1000

Sagittal Distance (pm)

" C 30

t

/ H 2 A

3 0

9,'

P

3 El

2000 +?- 1 2 3 4 8000 -+ 0.5 1 1.5 2

32000 + 0.25 0.5 0.75 1

Electrotonic Distance

Fig. 16. Histogram of electrotonic distances between each group Ia contact and the motoneuron soma calculated by using R, = 70 CL-cm and the indicated values of R, (see Materials and Methods). The cumulative probability curve (circles; referred to right ordinate) showed that over 60% of the contacts were in the proximal one-fourth of the electrotonic range.

0 1 I I I I 0 0.5 1 1.5 2

Distance Score

Fig. 15. A Average membrane area (solid circles, ten motoneurons) in successive 100-km-thick slabs orthogonal to the rostrocaudal axis of the spinal cord (ordinate; linear scale normalized by value in the first bin) plotted against the sagittal distance (D) rostra1 and caudal to the cell soma (abscissa; note the logarithmic scaling). Shaded line indicates the equation fitted (r = 0.99) to this data: y = 2.15 - 0.7 log (D). B: Scatter plot of the number of contacts in homonymous contact systems (ordinate) and the distance score from the relation shown in A for each system (abscissa; see text for details). Dashed line denotes linear least-squares fit to the data (r = 0.66; P < 0.01).

with more complex trees (mostly types FF and FR cells: Burke et al., 1982; Ulfhake and Kellerth, 1982; Cullheim et al., 1987a). The present results provide little support for this assumption; the average numbers of branch points proximal to contacts on all of the motor unit types in the present sample were virtually identical.

Spatial distribution of Ia contacts The hypothesis that group Ia synaptic contacts are

distributed widely over the dendritic membrane of a-moto- neurons was developed almost 3 decades ago (Rall et al., 1967) on the basis of intrasomatic recordings of the shapes of single-fiber group Ia EPSPs (Burke, 1967; Jack et al., 1971; Mendell and Henneman, 1971) and on the assump- tion that group Ia synapses generate rapid conductance changes (later confirmed by Finkel and Redman, 1983). This hypothesis has been amply confirmed by the present

I

0 8

All Group Ia Contacts 30

(N = 204)

s B. 3 5

study and its predecessors (Burke et al., 1979a,b; Brown and Fyffe, 1981; Redman and Walsmley, 1983a,b). The present results refine this conclusion with the demonstra- tion that the overall distribution of group Ia boutons is not coextensive with the entire motoneuron dendritic tree but, rather, is confined to a roughly rectangular three-dimen- sional region that represents the intersection of Ia collat- eral trajectories and the radially organized motoneuron dendrites (Figs. 11,121.

The interpretation of this finding, of course, depends on the local densities of motoneuron dendrites within the various hexants. The dendritic trees of triceps surae moto- neurons, on average, are radially organized, without particu- lar bias in any direction (Cullheim et al., 1987b). The correlation between the spatial distributions of bouton numbers and available postsynaptic membrane area (Figs. 13, 14) is consistent with the hypothesis that Ia contacts occur more or less randomly on motoneurons within the Ia contact region, without strong bias for distal or proximal locations. This conclusion is quite different from that recently reached by Rose and coworkers (Rose et al., 1995) with respect to presumptive lateral vestibulospinal boutons on cervical motoneurons, which appeared to be highly selective for rostrocaudal dendrites alone. The assumption in that study was that the rest of the dendritic tree also intersects with the afferent arborizations, which assump- tion was based on the morphology of vestibulospinal termi- nations in previous studies.

Relatively few Ia contacts were found on and near the motoneuron somata, similar to the earlier study by Brown and Fyffe (1981). The system illustrated in Figure 5 was the only one in the present sample that had recognizable contacts directly on the soma. Detection of contacts can be difficult on the somata of heavily labeled motoneurons, but, in fact, there were few systems in the present sample in which the existence of somatic contacts was even suspected. Brown and Fyffe (1981) examined this issue by counting the numbers of HRP-labeled Ia boutons that occurred near the somata of counterstained motoneurons without HRP

482 R.E. BURKE AND L.L. GLENN

B I I I I I

I I I t T F w U D X K A

I I I I I I I I 1

2000 + 1 2 3 4 soon + 0.5 1 1.5 2

m o o -+ 0.25 0.5 0.75 1

Electrotonic Distance

. Fig. 17. Electrotonic distributions (as in Fig. 16) of the individual contacts (filled circles) in each contact system (rows and identifying letters) grouped according to motor unit type and arranged within each group in order of increasing average electrotonic distance (open arrow- heads). In only three cases (systems N, B, and D) were all contacts belonging to a given system grouped within a restricted electrotonic locus.

label in relation to the rest of ventral horn. They found that only 9% of ventral horn Ia boutons made apparent contact on the somata and proximal dendrites of putative motoneu- rons. The present results (Fig. 13B,C) are consistent with a relative dearth of juxtasomatic Ia contacts, which may be

due in part to the fact that the motoneuron soma and proximal dendrites present a relatively small target for spatially dispersed Ia arborizations (e.g., Fig. 2).

Electrotonic distribution of Ia contacts Early interpretations of single-fiber Ia EPSP shapes were

compatible with their generation within restricted electro- tonic loci, whether proximal or distal in the dendritic tree (Jack et al., 1971; Finkel and Redman, 1983). In this light, the wide electrotonic dispersion of boutons within many of our Ia contact systems (Fig. 17) might seem surprising. However, Walmsley and Stuklis (1989) have argued on theoretical bases that EPSP shapes expected for a re- stricted input locus can be closely mimicked by active inputs that are dispersed either in electrotonic location or activation timing or in both. A similar result was obtained in this laboratory based on the compartmental computer models that utilized the present contact distributions for individual afferents (Fleshman et al., 1984).

In an elegant experiment, Redman and Walmsley (1983b) anatomically identified group Ia bouton clusters that prob- ably produced the single-fiber EPSPs they recorded electro- physiogically in the somata of HRP-labeled motoneurons. They were able to reconcile experimental and simulated EPSPs based on the observed anatomy in two of the four cases examined, both with a few contacts in relatively restricted anatomical and electrotonic loci. Their tech- niques, however, made it difficult to exclude additional contact regions from the same afferent in the two other cases.

In a recent study, Walmsley and Bolton (Walmsley and Bolton, 1994) used spatially selective Cjuxtasomatic) block- ade of group Ia transmission with the specific AMPA- receptor blocker NBQX applied extracellularly near the somata of an intracellularly recorded motoneuron while generating a single-fiber group Ia EPSP in that cell. In 618 cases studied, juxtasomatic blockade reduced but did not eliminate the EPSPs and revealed the presence of slow EPSP components, presumably originating more distally in the dendritic tree. In the other two cases, the EPSPs were completely blocked, and their shape indices were compat- ible with exclusively juxtasomatic EPSP origin. These results suggest that many group Ia afferents generate EPSPs in motoneurons through boutons that are scattered in both space and electrotonic distance from the somata, as indicated by the present anatomical results.

Some caveats One objective of the present investigation was to deter-

mine whether there are systematic relations between the number and spatial location of group Ia boutons and the motor unit type of the postsynaptic motoneuron. This issue is important in relation to the control of group Ia EPSP amplitude (Burke, 1981; Burke et al., 1988). Although the 24 group Ia contact systems studied here represent the largest sample currently available, the material is not adequate to give definitive answers to this question. When the sample of 24 contact systems was broken down into homonymous and heteronymous sets, and these were fur- ther subdivided according to motor unit type, the resulting subsamples were too small and the data were too scattered to reach statistical significance (Table 1). An additional caveat is that the motoneuron sample represents four muscles, which, although they are functionally related, are

MONOSYNAPTIC GROUP Ia CONTACT SYSTEMS 483

not necessarily organized in exactly the same pattern with respect to group Ia input.

Spatial sampling bias also proved to be a difficult prob- lem. The spacing of group Ia collaterals along the axis of the lumbosacral cord (Iles, 1976; Brown and Fyffe, 1978; Ishizuka et al., 1979) produces the greatest concentrations of Ia contacts in the neighborhood of the dorsal root entry of the parent afferent (see also Luscher et al., 1983). With the present experimental design, the dorsal root entry zone of the parent axon was evident only after reconstruction. The positions of labeled motoneurons were chosen only with respect to the position of afferent injection site (see Materials and Methods). Thus, the spatial factors of the overlap between motoneuron position and the concentra- tions and intensity of label within Ia afferent collaterals (Figs. 1 ,3 ,7) introduced sources of uncontrolled variability. These factors are important to the following discussion.

Bouton numbers in various contact systems: Functional aspects

Taken at face value, the data in Table 1 suggest that 1) heteronymous contact systems have fewer boutons per afferent than homonymous, except in type S motoneurons; and 2) type S motoneurons have fewer boutons per homony- mous afferent than either type FF or Type FR. None of these differences reached statistical significance. The dis- tance scores (Fig. 15) that were used to estimate group differences in spatial bias suggest that this factor could account for some of the difference between F and S contact systems, although, again, statistical tests were inconclu- sive. Thus, the available data are compatible with the conclusion that all triceps surae Ia contact systems have the same average number of boutons, irrespective of these categories. Nevertheless, it seems important to consider this conclusion in relation to existing information about functional connectivity (i.e., whether or not a given afferent produces a detectable EPSP in a given motoneuron) and the range of peak amplitudes found among group Ia EPSPs.

In cat triceps surae motoneurons, there is a higher likelihood of functional connectivity from individual hom- onymous group Ia afferents than from heteronymous af€er- ents (Scott and Mendell, 1976; Webb and Cope, 1992; for review, see Henneman and Mendell, 1981). This remains true even after acute spinal cord section, when homony- mous connectivity rises to virtually 100% (Nelson and Mendell, 1978). Consequently, every triceps surae group Ia afferent must have at least one anatomical contact on every homonymous motoneuron. This is consistent with the present results, in which putative contacts were found on a-motoneurons in every case in which collateral labeling was apparently complete within the ventral horn, including heteronymous as well as homonymous contact systems. It is always possible that a larger sample of heteronymous systems might have produced examples without such con- tacts. It is of interest that two y-motoneurons were labeled in regions that interdigitated with well-labeled homony- mous Ia afferent collaterals. In neither case were contacts found (as defined in Materials and Methods), which is in keeping with the fact that monosynaptic group Ia EPSPs are not detectable in y-motoneurons (Eccles et al., 1960; Kemm and Westbury, 1978).

The situation is much more complex when trying to translate between synaptic potential amplitude and bouton number. Although one expects that larger numbers of boutons would produce larger EPSPs, transmission fail-

ures occur during group Ia EPSPs, presumably at indi- vidual synaptic boutons (for review, see Redman, 1979). The probabilities and causal mechanisms of such failures are unknown. Furthermore, transmitter release and recep- tor properties may not be the same at Ia terminals on different types of postsynaptic motoneurons (see Lev-Tov et al., 1983; Collins et al., 1984). Systematic differences in the spatial location of boutons and in postsynaptic motoneuron properties can also contribute to the control of EPSP amplitude (Segev et al., 1990). Thus, a large variety of assumptions are involved in attempts to predict EPSP amplitudes from anatomical bouton numbers, and vice versa.

In the MG motor nucleus of the cat hindlimb, the peak amplitudes of composite (i.e., produced by stimulation of all group Ia fibers in a given muscle nerve) heteronymous group Ia EPSPs are, on average, about 25-40% those of the homonymous EPSPs in the same motoneurons (Eccles et al., 1957; Burke, 1968; Burke et al., 1976). Scott and Mendell (1976) found that the amplitudes of heteronymous single-fiber Ia EPSPs, when present, were systematically smaller (by about 60% on average) than homonymous EPSPs. Such a difference would fit with a smaller average number of boutons in heteronymous vs. homonymous contact systems, as suggested (but not statistically proven) by the data in Table 1. On the other hand, Webb and Cope (1992) reported little or no difference between average amplitudes of homonymous and heteronymous single-fiber Ia EPSPs when other confounding factors were removed. The available anatomical data, unfortunately, do not per- mit resolution of this issue.

The peak amplitudes of composite group Ia EPSPs are systematically larger in type S motoneurons than in type F motoneurons (Burke, 19681, and, in cat, MG motoneurons are related to motor unit types in the following sequence: FF < FR < S (Burke et al., 1976). The same ordering with motor unit type has been found in single-fiber Ia EPSPs (Fleshman et al., 1981). This ordering of synaptic efficacy is produced by a complex mixture of pre- and postsynaptic factors, among which (other factors being equal) is the density of active synapses on the motoneuron membrane (Segev et al., 1990; see also Burke, 1981; Burke et al., 1988). The average membrane areas estimated for triceps surae a-motoneurons follows an inverse sequence to the EPSP amplitude, i.e., FF > FR > S (Burke et al., 1982). The present anatomical data are consistent with the hypothesis that the anatomical density of boutons among type S motoneurons might be scaled in the EPSP sequence, but they cannot give the functional density of active group Ia synapses. The anatomical results, however, do permit us to say that there is no evident difference in the spatial dispersion of Ia bouton contacts in the different motor unit types (Fig. 17).

A recent computer simulation study of composite group Ia EPSPs was done with the assumption that the total number of active synapses is the same on types FF, FR, and S motoneurons with realistic morphology and electrotonic architectures (Segev et al., 1990). Simulated EPSPs had shapes and peak amplitudes that were completely consis- tent with those observed electrophysiologically. The avail- able evidence is compatible with the idea that the average total number of Ia synapses is the same (probably in the range of 500-1,000 per motoneuron; see Segev et al., 1990) for all three motoneuron types. A much larger sample of

484 R.E. BURKE AND L.L. GLENN

anatomically defined group Ia contact systems will be necessary to support or refute this hypothesis.

ACKNOWLEDGMENTS The authors gratefully acknowledge the participation, at

various times over a long period of time, of the following colleagues in the experiments that produced the primary material presented in this paper: Drs. Richard P. Durn, James W. Fleshman, Jr., John Hodgson, Aharony Lev-Tov, Michael O’Donovan, Martin Pinter, and Bruce Walmsley. Without their assistance, collection of the present material would not have been possible. The reconstructions and data analysis were done by the authors, who assume the entire responsibility for these aspects of the work.

LITERATURE CITED Adams, J.C. (1977) Technical considerations on the use of horseradish

peroxidase as a neuronal marker. Neuroscience 2141-145. Brown, A.G., and R.E.W. FyfTe (1978) The morphology of group Ia afferent

fibre collaterals in the spinal cord of the cat. J. Physiol. (London) 274:lll-127.

Brown, A.G., and R.E.W. FyfTe (1981) Direct observations on the contacts made between la afferents and a-motoneurones in the cat’s lumbosacral spinal cord. J. Physiol. (London) 313:121-140.

Burke, R.E. (1967) The composite nature of the monosynaptic excitatory postsynaptic potential. J. Neurophysiol. 30:1114-1137.

Burke, R.E. (1968) Group Ia synaptic input to fast and slow twitch motor units of cat triceps surae. J. Physiol. (London) 196:605-630.

Burke, R.E. (1981) Motor units: Anatomy, physiology and functional organization. In V.B. Brooks led): Handbook of Physiology, Sect. 1: The Nervous System, Vol. 11: Motor Control, Part 1. Bethesda, MD: Ameri- can Physiological Society, pp. 345-422.

Burke, R.E., and P. Rudomin (1977) Spinal neurons and synapses. In E.R. Kandel (ed): Handbook of Physiology, Sect. 1: The Nervous System, Vol. I: The Cellular Biology of Neurons, Part 1. Bethesda, MD: American Physiological Society, pp. 877-944.

Burke, R.E., D.N. Levine, P. Tsairis, and F.E. Zajac (1973) Physiological types and histochemical profiles in motor units of the cat gastrocnemius. J. Physiol. (London) 234:723-748.

Burke, R.E., W.Z. Rymer, and J.V. Walsh (1976) Relative strength of synaptic input from short latency pathways to motor units of defined type in cat medial gastrocnemius. J. Neurophysiol. 39:447-458.

Burke, R.E., B. Walmsley, and J.A. Hodgson (1979a) HRP anatomy ofgroup Ia afferent contacts on alpha motoneurones. Brain Res. 160:347-352.

Burke, R.E., B. Walmsley, and J.A. Hodgson (1979b) Structural-functional relations in monosynaptic action on spinal motoneurons. In V.J. Wilson and H. Asanuma (eds): Integration in the Nervous System. Tokyo: Igaku-Shoin, pp. 27-45.

Burke, R.E., M.J. Pinter, A. Lev-Tov, and M.J. O’Donovan (1980) Anatomy of monosynaptic contacts from group Ia afferents to defined types of extensor a-motoneurons in the cat. Soc. Neurosci. Abstr. 6:713.

Burke, R.E., R.P. Dum, J.W. Fleshman, L.L. Glenn, A. Lev-Tov, M.J. O’Donovan, and M.J. Pinter (1982) An HRP study of the relation between cell size and motor unit type in cat ankle extensor motoneurons. J. Comp. Neuro1209:17-28.

Burke, R.E., J.W. Fleshman, and I. Segev (1988) The control of synaptic efficacy: Lessons from the Ia synapse. J. Physiol. (Paris) 83: 133-140.

Collins, W.F., M.G. Honig, and L.M. Mendell (1984) Heterogeneity of group la synapses on homonymous alpha-motoneurons as revealed by high frequency stimulation of Ia afferent fibers. J. Neurophysiol. 66:1127- 1138.

Conradi, S., S. Cullheim, L. Gollnik, and J . -0 . Kellerth (1983) Electron microscopic observations on the synaptic contacts of group Ia and muscle spindle afferents in the cat lumbosacral spinal cord. Brain Res. 265:3140.

Cullheim, S., J.W. Fleshman, L.L. Glenn, and R.E. Burke (1987a) Membrane area and dendritic structure in type-identified triceps surae alpha- motoneurons. J. Comp. Neurol. 255:68-81.

Cullheim, S., J.W. Fleshman, L.L. Glenn, and R.E. Burke (198713) Three- dimensional architecture of dendritic trees in type-identified alpha- motoneurons. J. Comp. Neurol255t82-96.

Davis, B.M., W.F. Collins, and L.M. Mendell (1985) Potentiation of transmis- sion at Ia-motoneuron connections induced by repeated short bursts of afferent activity. J. Neurophysiol. 54: 1541-1552.

Eccles, J.C., R.M. Eccles, and A. Lundberg (1957) The convergence of monosynaptic excitatory afferents on to many different species of alpha-motoneurones, J. Physiol. (London) I37:22-50.

Eccles, J.C., R.M. Eccles, A. Iggo, and A. Lundberg (1960) Electrophysiologi- cal studies on gamma motoneurones. Acta Physiol. Scand. 50:32-40.

Finkel, A.S., and S.J. Redman (1983) The synaptic current evoked in cat spinal motoneurones by impulses in single group Ia axons. J. Physiol. (London) 342:615-632.

Fleshman, J.W., J.B. Munson, and G.W. Sypert (1981) Homonymous projection of individual group la-fibers to physiologically characterized medial gastrocnemius motoneurons in the cat. J. Neurophysiol. 46:1339- 1348.

Fleshman, J.W., I . Segev, and R.E. Burke (1984) Computer simulation of single fiber EPSPs based on the morphology of a-motoneurons and group la afferents. SOC. Neurosci. Abstr. 10:743.

Fleshman, J.W., I. Segev, and R.E. Burke (1988) Electrotonic architecture of type-identified alpha-motoneurons in the cat spinal cord. J. Neuro- physiol. 60:60-85.

FyfTe, R.E.W. (1991) Spatial-distribution of recurrent inhibitory synapses on spinal motoneurons in the cat. J. Neurophysiol. 65:1134-1149.

FyfFe, R.E.W., and A.R. Light (1984) The ultrastructure of group Ia afferent fiber synapses in the lumbosacral spinal cord of the cat. Brain Res. 300:201-209.

Glenn, L.L., and R.E. Burke (1981) A simple and inexpensive method for three-dimensional visualization of neurons reconstructed from serial sections. J. Neurosci. Methods 4:127-134.

Glenn, L., R.E. Burke, J.W. Fleshman, and A. Lev-Tov (1982) Estimates of electrotonic distance of group Ia contacts on cat a-motoneurons: An HRP-morphological study. Soc. Neurosci. Ahstr. 8:995.

Henneman, E., and L.M. Mendell (1981) Functional organization of motoneu- ron pool and its inputs. In V.B. Brooks (ed): Handbook of Physiology, Sect. 1: The Nervous System, Vol. 11: Motor Control, Part 1. Bethesda, MD: American Physiological Society, Bethesda, MD, pp. 423-507.

Iles, J.F. (1976) Central terminations of muscle afferents on motoneurones in the cat spinal cord. J. Physiol. (London) 26291-117.

Ishizuka, N., H. Mannen, T. Hongo, and S. Sasaki (1979) Trajectory of group Ia afferent fibers stained with horseradish peroxidase in the lumbosacral spinal cord of the cat: Three dimensional reconstructions from serial sections. J. Comp. Neurol. 186:189-211.

Itoh, K., A. Konishi, S. Nomura, N. Mizuno, Y. Nakamura, and T. Sugimoto (1979) Application of coupled oxidation reaction to electron microscopic demonstration of horseradish peroxidase: Cobalt-glucose oxidase method. Brain Res. 175:341-346.

Jack, J.J.B., S. Miller, R. Porter, and S.J. Redman (1971) The time course of minimal excitatory post-synaptic potentials evoked in spinal motoneu- rons by group Ia afferent fibres. J. Physiol. (London) 215:353-380.

Kemm, R.E., and D.R. Westbury (1978) Some properties of spinal y-motoneu- rones in the cat, determined by microelectrode recording. J. Physiol. (London) 28259-71.

Lev-Tov, A,, M.J. Pinter, and R.E. Burke (1983) Post-tetanic potentiation of group Ia EPSPs: Possible mechanisms for differential distribution in the MG motor nucleus. J. Neurophysiol. 50:379-398.

Liischer, H.-R., and H.P. Clamann (1992) Relation between structure and function in information transfer in spinal monosynaptic reflex. Physiol. Rev. 71:71-99.

Luscher, H.-R., P. Ruenzel, and E. Henneman 11983.1 Composite EPSPs in motoneurons of different sizes before and during PTP: Implications for transmission failure and its relief in Ia projections. J. Neurophysiol. 49269-289.

Matthews, P.B.C. (1972) Mammalian Muscle Receptors and Their Central Actions. London: Arnold.

Mendell, L.M., and E. Henneman (1971) Terminals of single la fibers: Location, density, and distribution within a pool of 300 homonymous motoneurons. J. Neurophysiol. 34: 171-187.

Nelson, S.G., and L.M. Mendell (1978) Projection of single knee flexor Ia fibers to homonymous and heteronymous motoneurons. J. Neuro- physiol. 41:778-787.

Pierce, J.P., and L.M. Mendell (1993) Quantitative ultrastructure of Ia boutons in the ventral horn-Scaling and positional relationships. J. Neurosci. 13:4748-4763.

MONOSYNAPTIC GROUP Ia CONTACT SYSTEMS 485

Pilowski, P.M., D. de Castro, I. Llewellyn-Smith, J. Lipski, and M.D. Voss (1990) Serotonin immunoreactive boutons make synapses with feline phrenic motoneurons. J. Neurosci. lO:1091-1098.

Rall, W. (1977) Core conductor theory and cable properties of neurons. In E.R. Kandel (ed): The Nervous System, Vol. I: Cellular Biology of Neurons, Part 1. Bethesda, MD: American Physiological Society, pp. 39-97.

Rall, W., R.E. Burke, P.G. Nelson, T.G. Smith, and K. Frank (1967) The dendritic location of synapses and possible mechanisms for the monosyn- aptic EPSP in motoneurons. J. Neurophysiol. 30:1169-1193.

Redman, S.J. (1979) Junctional mechanisms at group Ja synapses. Progr. Neurobiol. 1233-83.

Redman, S.J., and B. Walmsley (1983a) Amplitude fluctuations in synaptic potentials evoked in cat spinal motoneurons at identified group Ia synapses. J. Physiol. (London) 343:135-145.

Redman, S.J., and B. Walmsley (198313) The time course of synaptic potentials evoked in cat spinal motoneurones at identified group Ia synapses. J. Physiol. (London) 343r117-133.

Rose, P.K., T. Jones, R. Niruls, and T. Corneil (1995) Innervation of motoneurons based on dendritic orientation. J. Neurophysiol. 73: 1319- 1322.

Scott, J.G., and L. M Mendell (1976) Individual EPSPs produced by single

triceps surae Ia afferent fibers in homonymous and heteronymous motoneurons. J. Neurophysiol. 39:679-692.

Segev, I., J.W. Fleshman, and R.E. Burke (1990) Computer simulation of group Ia EPSPs using morphologically-realistic models of cat a-motoneu- rons. J. Neurophysiol. 64:64&660.

Snow, P.J., P.K. Rose, and A.G. Brown (1976) Tracing axons and axon collaterals of spinal neurons using intracellular injection of horseradish peroxidase. Science 191:3 12-3 13.

Ulfhake, B., and J.-0. Kellerth (1982) Does a-motoneuron size correlate with motor unit type in cat triceps surae? Brain Res. 251201-208.

Walmsley, B. (1991) Central synaptic transmission: Studies at the connec- tion between primary afferent fibres and dorsal spinocerebellar tract (dsct) neurones in Clarke’s column of the spinal cord. Progi. Neurobiol. 3f5391-423.

Walsmley, B., and P.S. Bolton (1994) An in vivo pharmacological study of single group Ia fibre contacts with motoneurones in the cat spinal cord. J. Physiol. (London) 481:731-741.

Walmsley, B., and R. Stuklis (1989) Effects of spatial and temporal disper- sion of synaptic input on the time course of synaptic potentials. J. Neurophysiol. 61:681-687.

Webb, C.B., and T.C. Cope (1992) Influence of synaptic identity on single-ia- afferent connectivity and epsp amplitude in the adult cat-Homonymous versus heteronymous connections. J. Neurophysiol. 68: 1046-1052.