ting axons usually returned to the original end-plates. Polyneuronal

J.

Phytiol.

(1976),

261,

pp.

387-422

387

With

2

plates

and

13

text-figure.

Printed

in

Great

Britain

POLYNEURONAL

INNERVATION

OF

SKELETAL

MUSCLE

IN

NEW-BORN

RATS

AND

ITS

ELIMINATION

DURING

MATURATION

BY

M.

C.

BROWN,*

J.

K.

S.

JANSEN

AND

D.

VAN

ESSENt

From

the

Inrtitute

of

Physiology,

Univer8ity

of

0810,

Karl

Johan8

Gate

47,

0810

1,

Norway

(Received

27

February

1976)

SUMMARY

1.

The

events

taking

place

during

the

elimination

of

polyneuronal

inner-

vation

in

the

soleus

muscle

of

new-born

rats

have

been

studied

using

a

combination

of

electrophysiological

and

anatomical

techniques.

2.

Each

immature

muscle

fibre

is

supplied

by

two

or

more

motor

axons

which

converge

on

to

a

single

end-plate.

There

was

no

sign

of

electrical

coupling

between

muscle

fibres

receiving

multiple

synaptic

inputs.

By

the

end

of

the

second

week

after

birth

virtually

all

muscle

fibres

are

inner-

vated

by

only

a

single

motor

axon.

3.

The

average

tension

produced

by

individual

motor

units,

measured

in

terms

of

the

percentage

of

the

total

muscle

twitch

tension,

declined

dramatically

during

the

first

2

weeks

after

birth.

During

this

period

there

was

no

significant

change

in

the

number

of

motor

neurones

innervating

the

soleus

muscle.

Thus,

the

disappearance

of

polyneuronal

innervation

reflects

a

decrease

in

the

number

of

peripheral

synapses

made

by

each

motor

neurone.

4.

The

decline

in

motor

unit

size

was

delayed,

but

not

ultimately

pre-

vented,

by

the

early

surgical

removal

of

all

but

a

few

motor

axons

to

the

soleus

muscle.

This

procedure

also

caused

a

delay

in

the

removal

of

poly-

neuronal

innervation

involving

the

remaining

motor

units.

5.

Following

a

crush

of

the

soleus

nerve

in

neonatal

animals,

regenera-

ting

axons

usually

returned

to

the

original

end-plates.

Polyneuronal

inner-

vation

was

extensive

at

early

stages

of

re-innervation

and

it

disappeared

during

the

second

week

after

birth

just

as

in

normal

muscles.

6.

Cross-innervation

of

neonatal

muscles

by

an

implanted

foreign

nerve

caused

a

rapid

disappearance

of

cholinesterase

at

denervated

original

*

Present

address:

University

Laboratory

of

Physiology,

Parks

Road,

Oxford.

t

Present

address:

Department

of

Anatomy,

University

College

London,

Gower

Street,

London

WC

1.

388

M.

C.

BROWN,

J.

K.

S.

JANSEN

AND

D.

VAN

ESSEN

end-plates

and

in

most

fibres

prevented

re-innervation

by

the

original

nerve.

In

the

small

proportion

of

fibres

that

did

become

innervated

through

both

the

foreign

and

original

nerves

the

end-plates

were

more

than

1

mm

apart,

and

both

foreign

and

original

nerve

end-plates

could

persist

indefinitely.

7.

Many

cross-innervated

fibres

received

multiple

inputs

through

the

foreign

nerve.

Some

foreign

end-plates

were

separated

by

distances

ranging

up

to

1

mm.

Polyneuronal

innervation

through

the

foreign

nerve

was

completely

eliminated

during

maturation

but

over

a

slightly

longer

period

than

in

normal

muscles.

Apparently

the

elimination

process

can

act

over

a

distance

up

to

but

not

much

more

than

1

mm.

8.

These

observations

suggest

that

there

are

several

factors

influencing

the

elimination

of

redundant

inputs

in

immature

muscles.

Individual

motor

neurones

appear

to

have

an

inherent

tendency

to

withdraw

the

majority

of

their

original

complement

of

peripheral

terminals.

The

deter-

mination

of

which

particular

synapses

are

to

survive,

however,

seems

to

be

made

in

the

periphery

by

a

selection

among

all

the

synapses

that

inner-

vate

a

limited

region

of

each

muscle

fibre.

There

may

be

a

competitive

interaction

among

synapses

in

which

those

belonging

to

smaller

motor

units

are

less

likely

to

be

eliminated,

thereby

leading

to

a

relatively

uni-

form

size

of

the

motor

units

in

the

soleus.

INTRODUCTION

Mature

mammalian

skeletal

muscle

is

innervated

in

a

remarkably

simple

way

by

its

motor

axons,

in

the

sense

that

each

muscle

fibre

is

supplied

by

only

a

single

axon

terminal

situated

approximately

midway

along

the

length

of

the

fibre.

It

thus

came

as

a

surprise

when

Redfern

(1970)

demon-

strated

that

the

end-plate

potentials

(e.p.p.s)

of

muscle

fibres

from

new-

born

rats

consisted

of

several

discrete

components,

indicating

that

each

muscle

fibre

initially

receives

an

input

from

several

motor

axons.

During

the

first

few

weeks

after

birth

multiple

innervation

disappears

and

the

adult

pattern

of

single

innervation

of

each

muscle

fibre

is

achieved.

These

observations

have

been

confirmed

and

extended

by

Bennett

&

Pettigrew

(1974a);

Bagust,

Lewis

&

Westerman

(1973)

have

shown

that

a

similar

phenomenon

occurs

in

kitten

muscles.

The

accessibility

of

muscles

even

in

neonatal

animals

to

experimental

procedures

such

as

micro-electrode

recording,

partial

denervation,

re-

innervation,

and

cross-innervation

has

made

it

possible

to

ask

a

variety

of

questions

concerning

the

process

of

synapse

elimination.

For

example,

where

are

the

synaptic

terminals

located?

What

factors

determine

the

time

at

which

synapses

disappear:

is

it

related

specifically

to

the

age

of

the

muscle,

the

motor

neurone,

or

the

synapse?

Does

synapse

elimination

ELIMINATION

OF

SYNAPSES

involve

only

terminals

situated

very

close

to

one

another,

or

can

the

inter-

action

take

place

between

spatially

separated

terminals?

Do

immature

fibres

show

a

preference

for

retaining

innervation

from

their

original

nerve

supply

rather

than

from

a

foreign

nerve?

In

addition

to

providing

answers

to

some

of

these

and

related

questions,

our

results

give

clues

to

the

control

mechanisms

involved

in

producing

the

end

result

of

one

and

only

one

synapse

per

muscle

fibre.

Some

of

the

observations

have

been

presented

in

preliminary

communications

(Brown,

Jansen

&

Van

Essen

1975,

1976).

METHODS

Phy&iological

recordings.

Experiments

were

made

on

the

soleus

and

diaphragm

muscles

of

rats

of

different

ages,

starting

as

early

as

the

first

day

after

birth.

We

refer

to

muscles

from

animals

under

2

weeks

of

age

as

immature,

because

up

until

that

time

they

retain

a

pattern

of

innervation

clearly

different

from

that

seen

in

the

adult.

For

physiological

recordings

the

muscle

and

its

motor

nerve

supply

were

isolated,

pinned

out

in

a

small

chamber

coated

with

transparent

Sylgard

resin,

and

perfused

at

room

temperature

with

a

well-oxygenated

Ringer

solution

of

the

fol-

lowing

composition

(in

mM):

NaCl,

137;

KCl,

5;

CaCl1,

10;

MgCl2,

1;

glucose,

11;

Tris

Cl,

10;

buffered

to

pH

7-4.

The

high

Ca

in

the

solution

improved

the

stability

and

quality

of

micro-electrode

penetrations.

Intracellular

recordings

were

made

using

glass

micro-electrodes

filled

with

4

M

K-acetate

and

having

resistances

of

20-60

Mil.

Using

dark-field

illumination

it

was

possible

to

resolve

individual

muscle

fibres

even

in

the

small

immature

muscles.

Nerves

were

stimulated

through

glass-

tipped

suction

electrodes.

In

many

experiments

the

muscle

was

curarized

by

adding

just

enough

D-tubocurarine

(0-5-2

jug/ml.)

to

block

nerve-evoked

contractions.

In

experiments

involving

iontophoresis

of

acetylcholine

(ACh)

the

ACh

was

delivered

from

a

micropipette

filled

with

1

M

ACh

having

a

resistance

of

200-400

MKI.

A

backing

current

of

1-3

nA

was

used

to

minimize

the

desensitization

of

ACh

receptors.

Measurement

of

motor

unit

tension.

In

one

series

of

experiments

we

counted

the

total

number

of

motor

axons

supplying

the

soleus

muscle

and

the

tension

generated

by

individual

motor

units.

The

nerve

supply

to

the

soleus

muscle

was

dissected

free

all

the

way

back

to

spinal

roots

L5

and

L4.

The

preparation

was

then

transferred

to

a

chamber

having

two

compartments

separated

by

a

thin

plastic

partition.

The

soleus

nerve

was

placed

through

a

slot

in

the

partition,

and

the

proximal

tendon

of

the

muscle

was

pinned

to

the

bottom

of

the

chamber

close

to

the

partition.

The

distal

tendon

was

connected

by

fine

surgical

thread

to

a

sensitive

strain

gauge.

Activity

in

the

soleus

nerve

was

monitored

by

recording

differentially

from

the

two

compart-

ments,

which

were

perfused

independently.

Each

ventral

root

was

split

into

several

filaments

that

were

stimulated

individually.

The

splitting

was

continued

until

each

filament

contained

at

most

four,

and

usually

only

one

or

two

soleus

motor

axons.

We

occasionally

saw

action

potentials

in

the

soleus

nerve

not

followed

by

measurable

tension

changes

in

the

muscle,

but

these

were

discounted

because

they

were

pre-

sumably

either

from

y

motor

axons

or

from

gastrocnemius

motor

axons

extending

unusually

far

into

the

soleus

nerve.

For

direct

stimulation

current

pulses

were

passed

between

coarse

electrodes

placed

on

opposite

sides

of

the

muscle.

In

the

course

of

most

experiments

there

was

an

unexplained

slow

and

parallel

decline

of

10-30%

in

the

tensions

produced

by

direct

muscle

stimulation

and

by

maximal

stimulation

of

the

nerve.

When

this

occurred

the

individual

motor

unit

tensions

were

expressed

as

a

percentage

of

the

original

total.

The

decline

in

muscle

tension,

389

390

M.

C.

BROWN,

J.

K.

S.

JANSEN

AND

D.

VAN

ESSEN

whatever

its

cause,

would

have

only

a

slight

effect

on

our

estimates

of

average

motor

unit

size.

It

might,

however,

seriously

affect

our

measurements

of

the

total

range

in

motor

unit

size

if

the

over-all

tension

decline

involved

large

losses

by

only

a

few

motor

units.

This

seems

unlikely,

though,

as

small

motor

units

were

encountered

even

in

experiments

where

little

or

no

tension

loss

was

seen

and

also

at

early

stages

of

the

other

experiments,

before

the

tension

decline

had

set

in.

Hieological

techniques.

End-plate

cholinesterase

was

stained

using

the

procedure

of

Karnovsky

(1964).

Muscles

were

prefixed

for

10-15

min

in

a

solution

containing

2-

5

%

glutaraldehyde,

2

%

paraformaldehyde,

and

0-1

IM

cacodylate

buffer

at

pH

7-5.

They

were

then

incubated

in

the

staining

solution

at

room

temperature

for

30

min,

fixed

in

the

same

fixative

solution

for

1-24

hr,

and

stored

in

H20.

The

distri-

bution

of

end-plates

throughout

the

muscle

could

be

seen

easily

in

whole

mount

preparations

viewed

under

a

dissecting

microscope.

In

order

to

count

the

number

of

end-plates

on

individual

fibres

we

found

it

necessary

to

tease

out

single

fibres

over

the

entire

region

where

cholinesterase

was

visible

in

the

whole

mount.

Fibres

were

dissected

free

with

the

aid

of

electrolytically

sharpened

tungsten

needles

and

fine

dissecting

forceps.

Each

fibre

was

placed

in

Aquamount

under

a

cover-slip

and

viewed

at

high

power

in

order

to

resolve

individual

end-plates

and

to

ensure

that

only

a

single

fibre

had

been

isolated.

Nerve

terminals

and

their

preterminal

axons

were

stained

using

the

zinc

iodide-

osmium

technique

(Akert

&

Sandri

1968).

Muscles

were

incubated

16-18

hr

in

the

staining

solution

and

washed

in

H20

for

at

least

2

hr.

Frozen

sections

of

30

jam

thick-

ness

were

cut

parallel

to

the

length

of

the

muscle

and

mounted

in

Aquamount.

The

staining

of

nerve

terminals

was

more

consistent

in

the

diaphragm

than

in

the

soleus

muscle,

presumably

because

the

diaphragm

is

a

much

thinner

sheet

of

muscle.

Surgical

procedures.

Operations

were

carried

out

under

ether

anaesthesia,

usually

on

the

first

or

second

day

after

birth.

The

soleus

muscle

was

denervated

by

crushing

the

nerve

next

to

its

entry

into

the

muscle

and

cross-innervated

by

placing

the

superficial

branch

of

the

fibular

nerve

on

to

the

proximal

surface

of

the

muscles.

The

effectiveness

of

the

nerve

crush

in

completely

interrupting

the

original

nerve

supply

was

demonstrated

by

showing

that

at

early

stages

(up

to

a

week

after

the

operation)

most

muscle

fibres

were

still

denervated,

and

that

those

which

were

in-

nervated

had

end-plate

potentials

with

abnormally

long

latencies

owing

to

the

slower

conduction

velocity

of

regenerated

axons.

Furthermore,

cross-innervation,

which

did

not

occur

when

the

original

nerve

was

left

intact,

took

place

just

as

effectively

after

a

nerve

crush

as

after

a

cut.

Nerve

crushes

were

preferred

for

most

experiments

because

re-innervation

took

place

more

quickly

and

reliably

than

after

a

nerve

cut.

In

one

series

of

experiments

the

soleus

muscle

was

partially

denervated

between

day

3

and

day

7

after

birth

by

cutting

ventral

root

L5,

which

contains

all

but

a

few

of

the

motor

axons

supplying

the

muscle.

The

root

was

cut

either

within

or

just

outside

the

spinal

column.

Most

animals

survived

the

operation

but

since

it

was

not

possible

to

recognize

the

nerve

roots

individually

during

the

operation

L5

was

successfully

cut

and

L4

left

undamaged

in

only

nine

animals

out

of

more

than

fifty

upon

which

we

operated.

RESULTS

Normal

neonatal

muscle

The

compound

end-plate

potential

In

muscles

taken

from

animals

less

than

10

days

old

virtually

all

muscle

fibres

receive

synaptic

inputs

from

several

motor

axons.

Text-

fig.

1

shows,

for

example,

an

intracellular

recording

from

a

fibre

in

a

ELIMINATION

OF

S

YNAPSES

5-day-old

soleus

muscle

paralysed

with

curare.

By

grading

the

strength

of

the

stimulus

to

the

nerve

it

was

possible

to

demonstrate

three

distinct

components

of

the

e.p.p.

The

nerve

was

stimulated

twice

in

each

of

the

six

superimposed

traces

in

the

Text-figure.

The

first

stimulus

was

varied

in

strength,

progressively

recruiting

one,

two

and

then

three

discrete

inputs;

the

second

stimulus

was

kept

supramaximal

to

indicate

the

degree

of

variability

in

maximal

e.p.p.

amplitude

from

one

trial

to

the

next.

The

Aft~~~~~~~~~~

Graded

Maximal

4mV

stimuli

stimuli

20

msec

Text-fig.

1.

The

compound

e.p.p.

Intracellular

recording

(a.c.)

from

a

soleus

muscle

fibre

of

a

5-day-old

rat.

The

soleus

nerve

was

stimulated

twice

during

each

of

six

superimposed sweeps.

The

first

shock

was

graded

in

strength,

while

the

second

was

kept

supramaximal.

Each

of

the

three

different

e.p.p.

components

was

recruited

at

sharply

defined

threshold

levels.

The

muscle

was

paralyzed

with

D-tubucurarine,

1

jcglml.

different

components

of

the

e.p.p.

invariably

had

very

similar

rise

times

which

were

as

short

as

2

msec

when

the

recording

electrode

was

situated

focally.

Often

all

of

the

components

had

a

similar

amplitude,

as

in

Text-

fig.

1,

but

occasionally

their

sizes

differed

by

as

much

as

a

factor

of

ten.

These

observations

are

in

basic

agreement

with

those

of

Redfern

(1970)

and

of

Bennett

&

Pettigrew

(1974a),

except

that

Bennett

&

Pettigrew

occa-

sionally

saw

e.p.p.

components

having

different

time

courses

in

muscles

examined

before

the

end

of

gestation.

Absence

of

electrical

coupling

between

muscle

fibres

One

simple

explanation

for

the

presence

of

several

components

in

the

e.p.p.

is

that

immature

muscle

fibres

might

be

electrically

coupled

to

one

another,

a

phenomenon

which

has

been

demonstrated

in

regenerating

salamander

muscle

(Dennis,

1975)

and

in

developing

amphibian

muscle

(Blackshaw

&

Warner,

1976).

In

the

rat

the

coupling

would

have

to

be

strong

in

order

to

account

for

the

equality

in

rise

times

and

similarity

in

amplitudes

of

the

different

components

of

the

e.p.p.

We

were

not,

how-

ever,

able

to

detect

any

signs

of

coupling

between

immature

muscle

fibres

in

extensive

surveys

of

several

muscles

examined

in

the

first

2

weeks

391

392

M.

C.

BROWN,

J.

K.

S.

JANSEN

AND

D.

VAN

ESSEN

after

birth.

When

two

adjacent

muscle

fibres

were

impaled

with

separate

micro-electrodes,

current

injected

into

one

fibre

never

caused

a

measur-

able

potential

change

in

the

neighbouring

fibre,

even

when

the

stimulus

was

sufficient

to

set

up

an

action

potential

in

the

first

fibre.

This

result

rules

out

any

widespread

electrical

coupling

in

the

immature

soleus

muscle,

but

a

more

restricted

alternative,

that

of

coupling

within

small

groups

of

muscle

fibres,

remained

a

possibility

if

one

supposed

that

most

of

the

fibres

within

a

group

were

too

small

to

be

clearly

resolved

in

the

dissecting

microscope.

We

therefore

carried

out

other

experiments

to

elucidate

the

pattern

of

innervation

of

neonatal

muscle

fibres.

A

10

mV/

20

msec

Text-fig.

2.

The

effect

of

prostigmine

on

the

compound

e.p.p.

A,

rapidly

rising

and

decaying

e.p.p.s

(single

component

on

left,

double

on

right)

in

the

absence

of

prostigmine

in

a

10-day

muscle

paralysed

with

curare,

1

jug/ml.

B,

slowly

rising

and

decaying

e.p.p.

components

recorded

from

the

same

fibre

several

minutes

after

applying

prostigmine,

2

x

10-6

gfml.

Multiple

innervation

of

single

end-plate

sites

Staining

of

a

neonatal

soleus

muscle

for

cholinesterase

reveals

that

the

end-plates

are

distributed

along

a

narrow

band

that

lies

roughly

midway

along

the

length

of

the

muscle,

just

as

is

found

in

adult

muscles.

In

whole

mounts

and

in

sectioned

material,

it

is

difficult

to

tell

whether

adjacent

spots

of

cholinesterase

are

on

the

same

or

on

neighbouring

fibres.

Lubinska

&

Zelena

(1966)

found,

however,

only

a

single

spot

of

cholinesterase

on

single

muscle

fibres

teased

from

the

new-born

rat

diaphragm.

We

have

confirmed

this

point

for

the

immature

soleus

muscle

(PI.

1

B).

ELIMINATION

OF

SYNAPSES

We

tested

for

the

possibility

that

some

synapses

might

lack

end-plate

cholinesterase

altogether

by

examining

the

effects

of

prostigmine

on

the

time

course

of

e.p.p.s

in

partially

curarized

muscles.

In

the

experiment

illustrated

in

Text-fig.

2

the

application

of

prostigmine

(2

,Ug/m1.)

to

the

bathing

solution

more

than

doubled

the

rise

times

and

amplitudes

of

both

the

low-threshold

e.p.p.

component

elicited

by

the

first

stimulus

and

the

maximal

e.p.p.

evoked

by

the

second

stimulus.

Prostigmine

had

a

similar

effect

on

all

components

of the

e.p.p.

in

every

muscle

fibre

tested,

indi-

cating

that

there

is

cholinesterase

associated

with

all

of

the

synapses

present

in

immature

muscles.

Since

only

one

spot

on

each

fibre

stains

for

cholinesterase,

it

is

likely

that

all

its

terminals

are

situated

at

this

one

end-plate.

Further

evidence

for

convergent

innervation

from

several

motor

axons

came

from

examining

muscles

whose

nerve

terminals

had

been

stained

using

the

zinc

iodide-osmium

technique.

In

well-stained

sections

from

immature

muscles

(both

the

soleus

and

the

diaphragm)

it

was

usually

possible

to

see

two

or

more

axons

leading

into

each

end-plate

(PI.

1A,

arrows).

In

older

muscles

examined

20

or

more

days

after

birth,

when

physiological

signs

of

multiple

innervation

have

virtually

disappeared,

there

was

only

one

axon

supplying

each

end-plate.

These

observations,

which

are

in

agreement

with

those

made

by

Bennett

&

Pettigrew

(1974a)

on

silver-stained

muscles,

provide

strong

support

for

the

idea

that

each

end-plate

initially

receives

an

input

from

several

motor

axons.

Strictly

speaking,

though,

the

anatomical

evidence

is

not

by

itself

conclusive

because

we

were

not

able

to

trace

the

preterminal

axons

to

each

fibre

far

enough

back

towards

the

main

nerve

to

be

certain

that

they

were

no

branches

of

the

same

parent

axon.

Our

final

piece

of

evidence

concerning

the

distribution

of

synapses

in

immature

muscles

comes

from

experiments

in

which

localized

application

of

ACh

from

a

micropipette

was

used

to

set

an

upper

limit

to

the

separation

between

synapses

on

each

fibre.

The

principle

of

the

experiment

was

to

see

whether

desensitization

of

the

ACh

receptors

over

a

small

region

of

a

muscle

fibre

had

an

equal

effect

on

all

components

of

the

e.p.p.

The

re-

cording

micro-electrode

was

first

inserted

into

a

muscle

fibre

near

the

synaptic

region

in

a

curarized

muscle

and

multiple

components

of

the

e.p.p.

were

demonstrated

by

the

paired

stimulation

technique

described

above

(Text-fig.

3A).

The

ACh

micropipette

was

then

moved

in

small

steps

along

the

fibre

until

a

spot

was

found

having

a

moderately

high

sensitivity

to

ACh

(>

10

mV/nC

in

the

presence

of

curare,

1-2

x

10-6

g/ml.).

At

this

point

a

steady

positive

current

of

1-5

nA

through

the

ACh

pipette

invariably

caused

a

large

reduction

or

even

a

complete

abolition

of

all

components

of

the

e.p.p.

(Text-fig.

3

B).

Such

a

parallel

reduction

393

394

M.

C.

BROWN,

J.

K.

S.

JANSEN

AND

D.

VAN

ESSEN

of

all

components

of

the

e.p.p.

was

seen

in

every

experiment.

The

ACh

application

also

caused

a

steady

conductance

change

and

membrane

depolarization

(Text-fig.

3,

lower

traces),

both

of

which

would

contribute

indirectly

to

the

reduction

in

e.p.p.

amplitude.

The

membrane

potential

always

remained

well

below

the

reversal

potential

for

the

e.p.p.,

however,

even

during

ACh

applications

that

completely

abolished

the

e.p.p.;

moreover,

considerable

reductions

in

e.p.p.

amplitude

could

be

obtained

by

ACh

applications

that

depolarized

the

membrane

by

only

a

few

milli-

volts.

It

is

clear,

therefore,

that

the

parallel

effects

of

ACh

application

on

all

components

of

the

e.p.p.

must

have

resulted

primarily

from

a

direct

desensitization

of

ACh

receptors

at

the

end-plate.

A

B

C

0

.

F

~~~~~~~~~~~

~~~5

mV

Eli

-40

20

msec

Text-fig.

3.

Desensitization

of

end-plate

receptors

by

ACh

iontophoresis.

A,

compound

e.p.p.

from

a

10-day-old

curarized

soleus

muscle

fibre.

High

gain,

a.c.

records

above;

low

gain,

d.c.

records

below.

B,

all

components

of

the

e.p.p.

were

abolished

during

steady

iontophoretic

application

of

ACh

to

the

end-plate

region.

C,

both

e.p.p.

components

recovered

completely

several

seconds

after

the

cessation

of

ACh

ionophoresis.

Preparation

curarized,

1

ug/ml.

Careful

positioning

of

the

ACh

pipette

was

necessary

in

order to

obtain

this

desensitizing

effect.

Movements

of

the

pipette

by

30-40

/Um

in

either

direction

along

the

length

of

the

fibre

or

over

to

adjacent

fibres

greatly

reduced

or

abolished

the

effect.

Although

the

spatial

resolution

provided

by

the

technique

as

we

used

it

was

somewhat

coarser

than

the

dimensions

of

an

end-plate

or

the

diameter

of

an

immature

muscle

fibre,

it

is

neverthe-

less

sufficient

to

demonstrate

that

all

of

the

synapses

on

an

immature

muscle

fibre

are

situated

within

about

50

gsm

of

one

another.

The

elimination

of

polyneuronal

innervation

The

percentage

of

muscle

fibres

receiving

more

than

one

synaptic

input

declined

rapidly

during

the

second

week

after

birth.

Text-fig.

4

shows

the

results

from

twenty-six

muscles

examined

between

the

second

and

nine-

teenth

day

after

birth.

At

least

twenty

fibres

from

each

muscle

were

tested

for

multiple

inputs

in

the

manner

illustrated

in

Text-fig..

1.

During

the

5

day

period

from

day

10

to

15

the

percentage

of

fibres

having

multiple

ELIMINATION

OF

SYNAPSES

inputs

dropped

from

91

%

(forty

of

forty-four)

fibres

to

25

%

(one

of

forty

fibres).

The

incidence

of

fibres

having

three

or

more

components

was

clearly

higher

during

the

first

week

after

birth

than

in

the

second

week,

but

we

did

not

examine

this

point

systematically

because

the

fluctuations

in

maximal

e.p.p.

amplitude

(which

were

usually

greater

than

that

shown

in

Text-fig.

1)

often

made

it

difficult

to

determine

exactly

how

many

components

were

present.

The

elimination

of

multiple

synaptic

inputs

100

_

-@

--

*@@

°,

50_

.

0~~~~~~~

25_\

>

75

0

5

10

15

20

Age

(days)

Texrt-fig.

4.

Percentage

of

soleus

muscle

fibres

innervated

by

more

than

one

axron

at

different

ages,

determined

from

intracellular

recordings

in

curar-

ized

muscles.

At

least

twenty

fibres

were

examined

in

each

muscle.

Treor

more

e.p.p.

components

were

seen

in

many

fibres,

especially

at

early

ages,

suggesting

that

the

loss

of

extra

synapses

may

be

well

under

way

before

day

10.

Continuous

line

drawn

by

eye

to

fit

the

observations.

thus

appears

to

be

an

ongoing

process

that

starts

sometime

during

the

first

week

after

birth

and

continues

until

the

end

of

the

second

week

(see

also

Bennet

&;

Pettigrew,

1974a

and

Texrt-fig.

6

below).

After

this

time

we

found

a

low

level

of

multiple

innervation

remaining

for

a

few

days,

but

even

this

must

eventually

disappear

(see

Jansen

&;

Van

Essen,

1975).

During

the

period

when

multiple

synaptic

inputs

were

being

eliminated

there

was

no

sign

that

significant

numbers

of

muscle

fibres

became

tran-

siently

denervated.

Once

the

recording

micro-electrode

was

placed

in

a

region

where

focal

e.p.p.s.

could

be

recorded

it

was

usually

possible

to

record

long

sequences

of

twenty

or

more

fibres

all

having

normal

e.p.p.s

395

396

M.

C.

BROWN,

J.

K.

S.

JANSEN

AND

D.

VAN

ESSEN

with

one

or

more

components.

In

the

few

cases

where

one

or

more

fibres

showing

no

response

to

nerve

stimulation

were

encountered

it

was

possible

to

attribute

the

change

either

to

a

sudden

shift

in

the

location

of

the

end-

plates

or

to

a

widespread

block

in

neuromuscular

transmission.

This

result

suggests

that

if

there

is

a

period

of

transient

denervation

during

the

matur-

ation

of

the

muscle,

it

either

involves

very

few

fibres

or

else

lasts

for

a

very

short

time

before

the

fibre

disappears

or

is

re-innervated.

Number

and

size

of

motor

units

The

removal

of

hyperinnervation

during

maturation

could

be

associated

with

the

complete

loss

of

whole

motor

neurones,

by

a

reduction

in

the

number

of

peripheral

terminations

made

by

each

motor

neurone,

or

by

some

combination

ofthe

two.

Degeneration

of

presumptive

motor

neurones

is

a

well

known

phenomenon

during

early

stages

of

development

(Hughes,

1968;

Landmesser

&

Pilar,

1974),

and

Bennett

&

Pettigrew

(1974a)

suggested

that

this

might

explain

the

elimination

of

synapses

during

the

early

post-natal

period.

The

question

was

settled

by

determining

the

size

and

the

total

number

of

motor

units

supplying

the

soleus

muscle

at

differ-

ent

ages.

On

account

of

the

rapid

growth

of

the

muscle

during

matura-

tion,

motor

unit

size

is

defined

operationally

as

the

percentage

of

the

total

muscle

twitch

tension

produced

by

that

unit.

Since

fibre

diameters

are

relatively

uniform

in

the

soleus

the

tension

measurements

should

provide

a

reasonable

indication

of

the

percentage

of

muscle

fibres

innervated

by

each

motor

unit.

In

order

to

count

the

total

number

of

soleus

motor

units,

the

muscle

was

dissected

free

along

with

its

nerve

supply

all

the

way

back

to

the

spinal

cord

and

mounted

in

a

chamber

that

allowed

impulse

activity

in

the

soleus

nerve

and

tension

in

the

muscle

to

be

recorded

while

small

filaments

of

the

ventral

roots

were

stimulated

electrically

(see

Methods).

We

found

that

the

total

number

of

motor

units

remained

virtually

constant

throughout

the

period

when

polyneuronal

innervation

was

being

eliminated.

This

number

was

between

twenty-one

and

twenty-five

for

the

seven

muscles

between

3

and

42

days

of

age

for

which

we

obtained

complete

counts.

There

was

no

significant

difference

in

the

number

of

motor

units

for

the

three

muscles

examined

before

day

10

(mean

22-7

units)

and

the

four

muscles

examined

after

that

time

(mean

23x5

units).

A

few

motor

units

may

have

been

missed

in

some

of

the

preparations;

in

fact,

our

value

for

mature

muscle

is

slightly

lower

than

the

estimates

of

others

based

on

in

vivo

tension

measurements

(28-30

units;

Close,

1967)

or

fibre

counts

in

deafferented

motor

nerves

(thirty-two

a

motor

axons;

Gutman

&

Hanz-

llkova',

1966).

Nevertheless,

it

is

unlikely

that

we

missed

a

much

greater

number

of

units

in

younger

than

in

older

rats

and

we

therefore

conclude

ELIMINATION

OF

S

YNAPSES

that

the

number

of

motor

units

to

the

soleus

stays

relatively

constant

after

birth.

In

contrast

to

the

constancy

of

the

number

of

motor

units,

there

were

dramatic

changes

in

motor

unit

size

shortly

after

birth.

Text-fig.

5,

for

example,

shows

a

single

motor

unit

from

an

immature

muscle

that

gener-

ated

a

tension

greater

than

one

fifth

of

the

total

muscle

tension.

Text-fig.

5B

is

a

recording

from

the

soleus

nerve

at

a

fast

sweep

speed

to

demon-

strate

that

the

stimulus

activated

only

a

single

soleus

motor

axon.

Text-

fig.

5A

shows,

at

a

much

slower

sweep

speed,

the

tension

generated

by

the

Maximal

tension

g

wt..

I

_LZ

Single

motor

unit

tension

2001

PV

|

Nerve

500

msec

50

msec

Text-fig.

5.

Measurement

of

motor

unit

size

in

an

immature

muscle.

A,

shows

the

tension

generated

in

the

soleus

muscle

of

a

3-day-old

rat

by

stimulation

of

a

single

ventral

root

filament

and,

for

comparison,

the

ten-

sion

produced

by

maximal

stimulation

of

the

whole

nerve.

B,

shows,

at

a

fast

sweep

speed,

the

unitary

action

potential

recorded

en

pa88age

from

the

soleus

nerve

(see

Methods)

after

stimulation

of

the

ventral

root

filament.

Note

the

very

slow

time

course

of

the

contraction,

which

is

characteristic

of

immature

muscle.

single

motor

unit

and,

for

comparison,

the

total

muscle

tension

produced

by

maximal

stimulation

of

the

nerve.

(Direct

electrical

stimulation

of

the

muscle

also

produced

the

same

maximal

tension.)

Our

observations

on

motor

unit

tensions

are

summarized

in

Text-fig.

6,

which

gives

the

relative

tension

of

the

motor

units

in

ten

different

muscles

as

a

function

of

the

age

of

the

animal.

The

tensions

are

displayed

as

per-

centages

of

the

total

twitch

tension

evoked

by

direct

stimulation

of

the

muscle.

Each

vertical

line

represents

the

observations

in

one

animal,

and

the

mean

values

(filled

circles)

as

well

as

the

values

for

individual

units

(horizontal

lines)

are

presented

for

each

muscle.

Tension

measurements

397

398

M.

C.

BROWN,

J.

K.

S.

JANSEN

AND

D.

VAN

ESSEN

were

included

only

for

the

lowest

threshold

unit

in

each

ventral

root

filament

because

there

was,

as

expected,

non-linear

summation

of

tension

values

when

more

than

one

unit

was

stimulated.

The

open

circles

show

the

average

tension

that

would

be

expected

if

there

were

no

multiply

inner-

35

.

30

0

C

~25

0

20

20

0

CI

10

0

10

Is

20

4

Age

(days)

Text-fig.

6.

Size

and

number

of

soleus

motor

units

at

different

ages.

Each

vertical

line

represents

observations

on

one

animal.

The

ordinate

gives

the

size

of

motor

units

expressed

as

their

percentage

of

the

maxima

twitch

to

direct

stimulation.

Filled

circles

(X)

give

the

mean

size

of

motor

units

in

each

muscle;

horizontal

lines

indicate

the

individual

measurements

for

the

lowest-threshold

motor

unit

in

each

ventral

root

filament.

Open

circles

(0)

show

the

average

motor

unit

size

one

would

expect

in

the

absence

of

poly-

neuronal

innervation

[100

x

(total

number

of

units-l)].

In

three

muscles

(2,

5

and

17

days)

this

value

is

not

given

because

a

partial

nerve

block

in

the

region

of

the

ventral

roots

prevented

completion

of

the

motor

unit

count.

vated

fibres.

This

value,

obtained

simply

by

taking

the

inverse

of

the

total

number

of

motor

units,

was

between

4

and

5

%

for

the

seven

muscles

for

which

complete

motor

unit

counts

were

available.

Text-fig.

6

shows

that

motor

units

were

about

five

times

larger

at

early

ELIMINATION

OF

SYNAPSES

times

(mean

23

%

at

days

2

and

3)

than

at

later

times

(mean

5-1

%

at

days

15-18

and

4.4

%

at

day

42).

This

reduction

in

motor

unit

size

is

large

enough

to

account

entirely

for

the

disappearance

of

polyneuronal

innerva-

tion

in

immature

muscles.

In

fact,

the

average

degree

of

polyneuronal

innervation

estimated

from

the

motor

unit

measurements

(obtained

by

taking

the

ratio

of

the

filled

circle

to

the

open

circle

values

for

each

muscle),

is

about

five

at

days

2

and

3.

An

average

value

of

five

synapses

per

fibre

is

higher

than

either

we

or

Bennett

&

Pettigrew

(1974a)

obtained

by

count-

ing

the

number

of

e.p.p.

components

in

individual

fibres

but

the

dis-

crepancy

is

not

surprising

in

view

of

the

inherent

inaccuracies

in

each

type

of

estimate.

A

safe

conclusion

would

be

that

there

are

many

separate

synaptic

inputs

to

each

neonatal

muscle

fibre,

each

one

of

which

is

capable

of

activating

the

muscle

fibre

by

itself.

It

is

interesting

that

there

was

considerably

more

scatter

in

the

size

of

motor

units

throughout

the

period

of

maturation

than

in

adult

muscles.

The

total

range

was

about

a

factor

of

three

in

the

one

adult

muscle

we

examined

and

in

the

larger

sample

of

motor

units

obtained

by

Close

(1967),

whereas

the

range

was

about

tenfold

in

several

of

the

immature

muscles.

At

all

ages

there

were

some

motor

units

that

were

within

the

range

seen

in

the

adult.

In

the

15-

and

17-day-old

muscles

a

few

of the

motor

units

were

actually

smaller

than

any

seen

in

the

adult.

It

is

possible

that

the

tension

values

were

spuriously

low

owing

to

an

axonal

conduction

block

or

to

a

failure

in

synaptic

transmission.

This

explanation

seems

unlikely,

however,

because

the

maximal

indirect

and

direct

twitches

were

equal

and

in

the

same

muscles

there

were

still

other

motor

units

larger

than

those

seen

in

the

adult.

If

the

tension

measurements

provide

a

reasonably

accurate

measure

of

motor

unit

size

in

terms

of

numbers

of

muscle

fibres

innervated,

then

the

results

suggest

that

at

a

time

when

almost

all

polyneuronal

innervation

has

been

eliminated,

some

readjustments

in

motor

unit

sizes

remain

to

be

completed.

Modifying

the

time

course

of

synarpse

removal

The

effects

of

partial

denervation

If

one

could

remove

all

but

a

few

of

the

motor

axons

to

the

soleus

muscle

shortly

after

birth,

a

reduction

in

size

of

the

remaining

ones

would

have

to

take

place

at

the

expense

of

leaving

some

fibres

completely

de-

nervated.

A

priori

it

seemed

plausible

that

this

might

influence

or

even

reverse

the

normal

reduction

in

motor

unit

size.

The

experiment

was

performed

by

sectioning

the

lower

of

the

two

ventral

roots

supplying

the

soleus

muscle.

The

remaining

root

(L4)

usually

contains

at

least

one

and

sometimes

as

many

as

nine

soleus

motor

axons.

The

operation

was

carried

out

successfully

on

nine

animals.

They

were

examined

between

days

15

and

43

after

birth.

The

results

were

quite

striking:

partial

denervation,

on

399

400

M.

C.

BROWN,

J.

K.

S.

JANSEN

AND

D.

VAN

ESSEN

the

one

hand,

delayed

the

normal

reduction

in

motor

unit

size

and

even

the

elimination

of

polyneuronal

innervation;

on

the

other

hand,

it

did

not

seem

to

prevent

the

eventual

shrinkage

of

motor

units

to

approximately

their

normal

adult

size.

Five

muscles

with

five

to

nine

remaining

motor

units

were

examined

between

days

15

and

17,

just

after

the

time

when

most

polyneuronal

innervation

normally

has

been

eliminated.

The

histogram

in

Text-fig.

7

l

l

'5

0

0

10

20

30

Motor

unit

sizes

Text-fig.

7.

Motor

unit

sizes

at

intermediate

times

after

partial

denerva-

tion.

Continuous

lines:

motor

unit

sizes,

determined

from

tension

measure-

ments

(as

in

Text-fig.

6),

of

twenty

motor

units

in

five

muscles

examined

15-17

days

after

birth

following

partial

denervation

on

day

3-5

(three

on

day

3,

one

each

on

days

4

and

5).

In

each

animal

there

were

between

four

and

nine

motor

units

in

the

remaining

ventral

root

(L4)

but

tension

meas-

urements

were

not

obtained

for

all

of

these.

Dashed

lines:

motor

unit

sizes

in

two

normal

soleus

muscles

aged

15

and

17

days

(from

Text-fig.

6).

shows

the

tensions

generated

by

twenty

of

the

motor

units

remaining

in

these

five

muscles.

The

sizes

of

seventeen

units

from

two

normal

muscles

of

comparable

age

are

shown

for

comparison.

The

mean

size

of

the

motor

units

from

the

partially

denervated

muscles

(20

%)

was

much

larger

than

that

of

the

normal

muscles

(5-6

%)

and

was,

in

fact,

very

close

to

the

mean

size

of

motor

units

at

the

time

of

the

initial

operation

(21

%;

results

from

days

3

and

5

in

Text-fig.

6).

The

abnormally

large

motor

units

did

not

appear

to

result

simply

from

hypertrophy

of

innervated

muscle

fibres

or

to

atrophy

of

the

denervated

ones.

The

muscles

were

not

noticeably

atrophic

ELIMINATION

OF

SYNAPSES

and

they

generated

a

normal

tension

to

nerve

stimulation.

In

addition,

cross-sections

of

two

of

the

muscles

showed

that

they

contained

close

to

the

normal

number

of

muscle

fibres

(2500-2900)

having

relatively

uniform

diameters.

Thus,

it

appears

that

the

motor

units

remaining

after

partial

denervation

innervated

a

much

larger

number

of

muscle

fibres

than

they

normally

would

have

at

this

age.

There

were

two

other

obvious

abnormalities

in

the

partially

denervated

muscles

examined

between

days

15

and

17.

The

first

was

that the

twitch

contractions

to

stimulation

of

single

motor

units

fatigued

very

rapidly,

even

during

stimulation

at

frequencies

as

low

as

0

5-1

Hz.

After

a

few

stimuli

the

twitch

tension

was

often

less

than

half

of

its

initial

value,

after

which

it

remained

stable.

In

normal

muscles

there

was

little

tension

decre-

ment

even

during

stimulation

at

rates

of

2-3

Hz.

The

rapid

fatigue

in

the

partially

denervated

muscles

was

not

related

to

an

impairment

in

the

con-

tractile

properties

of

the

muscle

as

there

was

no

fatigue

during

direct

muscle

stimulation.

A

likely

explanation

for

the

fatigue

is

that

transmission

at

many

synapses

was

only

marginally

above

threshold

and

was

readily

brought

below

threshold

by

repetitive

stimulation.

The

other

abnormality

in

this

group

of

partially

denervated

muscles

was

that

there

was

extensive

polyneuronal

innervation

remaining

2

weeks

after

the

operation.

We

recorded

e.p.p.s.

having

more

than

one

component

in

about

half

of

the

innervated

fibres

in

each

muscle

(range

22-65

%

in

five

muscles).

This

was

not

a

general

'systemic'

consequence

of the

opera-

tion,

because

in

every

experiment

we

found

that

polyneuronal

innerva-

tion

in

the

control

soleus

muscle

from

the

opposite

leg

had

been

reduced

to

its

normal

level

of

5

%

or

less.

The

situation

was

quite

different

in

muscles

examined

at

later

times

after

partial

denervation.

The

clearest

illustration

of

this

came

from

a

muscle

partially

denervated

four

days

after

birth

and

examined

on

day

33.

There

were

only

two

motor

axons

supplying

the

soleus

muscle

through

the

intact

ventral

root

and

the

muscle

itself

had

atrophied

considerably.

A

cross-section

of

the

muscle,

shown

in

P1.

2

B,

revealed

that

a

minority

of

the

muscle

fibres,

321

in

all,

had

escaped

atrophy

and

were

actually

larger

than

normal.

The

other

fibres

were

very

small

and

difficult

to

resolve

individually.

For

comparison,

P1.

2A

shows

the

relative

uniformity

of

fibre

diameters

in

a

normal

muscle

of

similar

age.

We

found

no

multiply

innervated

fibres

during

a

micro-electrode

survey

of

the

partially

dener-

vated

muscle

after

it

was

curarized.

Hence,

a

reasonable

estimate

of

motor

unit

sizes

can

be

obtained

by

assuming

that

the

very

small

fibres

in

the

muscle

were

all

denervated

and

then

simply

dividing

up

the

number

of

larger

fibres

in

accordance

with

the

relative

sizes

of

the

motor

unit

ten-

sions.

The

two

motor

units

in

this

muscle

contributed

40

and

60

%

to

the

401

402

M.

C.

BROWN,

J.

K.

S.

JANSEN

AND

D.

VAN

ESSEN

total

muscle

tension,

suggesting

that

they

innervated

approximately

130

and

190

muscle

fibres,

respectively.

These

values

are

about

4

and

6

%

of

the

total

of

approximately

3000

fibres

found

in

normal

muscles

(Chiakulas

&

Pauly,

1965;

Frank,

Jansen,

Lomo

&

Westgaard,

1975).

A

similar

estimate

was

obtained

indirectly

by

comparing

the

absolute

values

of

the

motor

unit

tensions

obtained

from

normal

muscles

of

comparable

age.

The

total

twitch

tension

from

the

partially

denervated

muscle

was

about

15

%

of

that

from

normal

muscles

of

a

similar

age

(12

g

wt.

vs.

6'9-10

g

wt.

for

five

normal

4-6-week-old

muscles),

and

the

individual

motor

unit

tensions

were

roughly

6

and

9

%

of

the

normal

total.

Thus,

even

without

a

correction

for

the

effects

of

the

obvious

hypertrophy

of

innervated

muscle

fibres,

the

tension

measurements

support

the

suggestion

that

the

number

of

muscle

fibres

innervated

by

each

motor

unit

was

closer

to

the

mean

value

in

normal

adult

muscle,

than

to

the

mean

value

at

the

time

of

the

initial

operation

or

after

2

weeks

of

partial

denervation.

Owing

to

the

large

scatter

in

individual

motor

unit

sizes

in

immature

muscles,

however,

it

is

not

possible

to

conclude

that

these

particular

motor

units

had

actually

decreased

in

size,

as

opposed

to

starting

out

small

and

not

expanding.

Similar

results

were

obtained

for

two

of

the

three

other

muscles

exam-

ined

4-6

weeks

after

partial

denervation.

One

of

these,

partially

dener-

vated

on

day

7

and

examined

on

day

43,

had

two

remaining

motor

units

and

408

large

fibres

visible

in

a

cross-section

of

the

muscle.

The

individual

motor

unit

tensions

were

approximately

equal

in

amplitude

(Text-fig.

8A:

upper

record,

one

motor

unit;

lower

record,

both

motor

units)

but

they

were

several

times

larger

in

absolute

value

than

those

in

a

normal

muscle

of

the

same

age,

owing

in

part

to

the

hypertrophy

of

the

inner-

vated

fibres.

The

total

tension

produced

by

stimulating

the

nerve

close

to

the

muscle

(Text-fig.

8B)

was

no

larger

than

that

from

combined

stimula-

tion

of

the

two

motor

units,

indicating

that

there

were

no

other

large

motor

units

in

the

nerve.

Moreover,

the

tension

from

direct

muscle

stimulation

was

the

same

as

the

nerve-evoked

tension,

suggesting

that

non-innervated

fibres

(presumably

all

the

atrophied

ones)

contributed

little

to

the

total

muscle

tension.

A

low

level

of

multiple

innervation

was

found

in

this

muscle

(four

of

twenty

fibres

tested).

Accordingly,

the

estimated

size

of

each

motor

unit,

based

on

the

muscle

fibre

counts

and

allowing

for

20

%

of

the

large

fibres

being

doubly

innervated,

was

about

245

fibres,

or

8

%

of

the

normal

total.

In

another

moscle,

partially

denervated

on

day

7

and

examined

on

day

41,

only

one

surviving

motor

unit

was

counted

(plus

a

few

very

small

motor

units

that

had

regenerated

through

the

cut

nerve);

we

estimated

that

it

innervated

280

muscle

fibres,

or

9

%

of

the

normal

total.

Finally,

one

muscle

having

six

surviving

motor

axons

was

examined

at

ELIMINATION

OF

SYNAPSES

a

slightly

earlier

time

(31

days

v8.

33,

41

and

43

days);

it

had

not

under-

gone

changes

as

extensive

as

in

the

other

muscles.

A

cross-section

of

the

muscle

appeared

normal,

with

no

hypertrophied

fibres

and

no

obvious

population

of

small,

presumably

denervated

fibres.

The

motor

unit

ten-

sions

were

all

between

12

%

and

16

%

of

the

total

for

the

five

units

for

A

B

An

I

I~~~~~~~~~~~~~~~~~~~3

g

wt.

4

msec

200

msec

Text-fig.

8.

Two

remaining

motor

units

in

a

43-day-old

soleus

muscle

par-

tially

denervated

by

extraspinal

section

of

L5

on

day

7.

A,

records

of

nerve

impulses

(upper

traces,

fast

sweep

speed)

and

twitch

tensions

(lower

traces,

slow

sweep

speed)

after

stimulation

of

the

intact

ventral

root.

Low

intensity

stimuli

activated

a

single

soleus

motor

axon

(upper

records);

maximal

stimuli

activated

only

two

motor

axons

(lower

records).

The

slow

wave

in

the

right

half

of

the

nerve

records

is

an

artifactual

pick-up

of

the

gross

muscle

action

potential.

B,

the

maximal

twitch

to

stimulation

of

the

whole

soleus

nerve

near

the

muscle.

Note

that

the

tension

is

no

larger

than

that

from

activating

just

the

two

motor

units

in

A.

The

maximal

twitch

to

direct

stimulation

was

the

same

as

the

maximal

nerve-evoked

twitch.

There

were

408

large

fibres

counted

in

a

cross-section

of

this

muscle.

which

individual

measurements

were

obtained.

The

larger

size

of

the

motor

units

in

this

muscle,

compared

to

those

seen

in

the

three

preceding

ex-

amples,

might

have

been

a

chance

occurrence,

or

it

might

be

related

to

the

younger

age

of

the

muscle

or

to

the

fact

that

it

happened

to

have

a

larger

number

of

surviving

motor

units.

These

results

are

summarized

in

Text-fig.

9,

which

compares

the

motor

unit

sizes

for

the

four

muscles

partially

denervated

for

4-6

weeks

403

404

M.

C.

BROWN,

J.

K.

S.

JANSEN

AND

D.

VAN

ESSEN

(continuous

lines)

with

those

seen

around

the

time

of

the

initial

operation

(days

3,

5

and

7,

dashed

lines).

The

mean

size

of

motor

units

was

consid-

erably

lower

in

the

partially

denervated

muscles

(10.6

vs.

15-9

%),

es-

pecially

if

one

considers

only

the

three

muscles

with

an

obvious

mixture

of

hypertrophied

and

atrophied

fibres

(hatched

area,

mean

7

%),

excluding

the

remaining

muscle

(dotted

area)

on

the

grounds

that

its

histological

uniformity

suggests

that

it

might

still

have

been

in

a

transitional

state.

10

0

.......I

I

_I

I

I

I

4J

I

I

0

1

0

3

M

_

~~I

I

e

_

11

.o5

1

z

I

I

0

10

20

30

Motor

unit

sizes

Text-fig.

9.

Histogram

of

unit

sizes

at

late

times

after

partial

denervation.

Hatched

area:

motor

unit

sizes

for

five

units

in

three

soleus

muscles

examined

at

33,

41

and

43

days

after

birth

following

partial

denervation

at

4,

7

and

7

days,

respectively.

The

estimates

were

based

on

counts

of

large,

non-atrophied

muscle

fibres

seen

in

histological

cross-sections,

expressed

as

a

percentage

of

the

normal

value

of

3000

fibres.

Dotted

area:

motor

unit

sizes,

based

on

tension

measurements

for

five

units

in

a

muscle

partially

denervated

on

day

4

and

examined

on

day

31.

Dashes

lines:

histogram

of

motor

unit

sizes

for

twenty-six

motor

units

in

three

normal

muscles

aged

3,

5,

and

7

days

(from

Text-fig.

6).

In

any

event,

whether

or

not

the

one

muscle

is

excluded

from

the

sample,

the

results

are

consistent

with

a

genuine

shrinkage

of

motor

units

during

maturation

after

partial

denervation.

They

are

not

conclusive,

however,

owing

to

the

relatively

large

scatter

in

motor

unit

sizes

and

to

the

uncer-

tainties

that

result

from

comparisons

involving

two

different

methods

for

measuring

motor

unit

sizes

(tension

measurements

and

muscle

fibre

ELIMINATION

OF

SYNAPSES

counts).

We

conclude

that,

at

the

very

least,

there

is

no

widespread

sprouting

of

axon

collaterals

to

take

over

large

numbers

of

denervated

fibres

after

partial

denervation

in

immature

rats.

To

the

contrary,

it

is

prob-

able

that

some

motor

units

are

not

able

to

maintain

their

original

comple-

ment

of

peripheral

synapses

even

though

some

muscle

fibres

consequently

are

stripped

of

their

remaining

inputs.

Re-innervation

and

cro88-innervation

of

immature

mwdcle

By

examining

the

events

that

occur

during

re-innervation

of

the

im-

mature

soleus

muscle

by

its

original

nerve

and

cross-innervation

by

a

foreign

nerve

it

has

been

possible

to

study

several

questions

that

would

be

difficult

or

impossible

to

investigate

in

a

normal

muscle.

For

example,

we

were

interested

in

whether

newly

formed

terminals

can

take

part

in

the

process

of

synapse

elimination,

whether

interactions

between

synapses

situated

at

a

distance

from

one

another

can

lead

to

the

elimination

of

one

of

them

and

whether

there

is

any

selectivity

in

the

choice

of

which

synapses

are

to

survive

and

which

are

to

be

removed.

Re-innervation

by

the

original

nerve

Re-innervation

of

the

immature

soleus

muscle

occurred

relatively

quickly

and

completely

following

a

crush

of

the

nerve

just

at

its

entry

into

the

muscle.

Substantial

contractions

to

nerve

stimulation

were

seen

within

a

week

following

denervation

on

the

first

or

second

day

after

birth.

By

days

11-12

more

than

half

of

the

muscle

had

been

re-innervated,

as

assessed

by

micro-electrode

surveys

of

superficial

muscle

fibres

and

by

comparing

nerve-evoked

and

direct

muscle

tensions

(three

experiments).

Three

weeks

after

birth

re-innervation

was

nearly

complete

(80-100

%

in

two

experiments);

muscle

cross-sections

taken

at

this

time

showed

a

normal

number

of

muscle

fibres

having

relatively

uniform

diameters,

indicating

that

there

had

not

been

a

loss

or

severe

atrophy

of

large

num-

bers

of

fibres.

During

the

first

two

weeks

of

re-innervation

many

muscle

fibres

were

innervated

by

more

than

one

motor

axon.

In

some

muscles

as

many

as

80-90

%

of

the

re-innervated

fibres

had

e.p.p.s

with

more

than

one

com-

ponent.

Text-fig.

10

shows

the

time

course

of

elimination

of

polyneuronal

innervation

in

the

re-innervated

muscles.

Most

of

the

synapses

that

were

eliminated

during

days

10-15

were

presumably

formed

during

the

prece-

ding

week

when

re-innervation

was

taking

place

most

rapidly

and

were

therefore

much

younger

than

synapses

in

normal

muscles,

which

are

formed

in

the

final

week

of

gestation

(Bennett

&

Pettigrew

1974a).

It

is

remarkable

how

closely

the

decline

in

multiple

innervation

followed

that

seen

in

normal

muscles

(dashed

line,

taken

from

Text-fig.

4).

The

405

406

M.

C.

BROWN,

J.

K.

S.

JANSEN

AND

D.

VAN

ESSEN

elimination

process

seemed

to

run

its

normal

course

even

though

the

synapses

it

acted

upon

were

about

half

their

normal

age.

In

the

great

majority

of

fibres

examined

in

the

re-innervated

muscles

all

components

of

the

e.p.p.

had

similar

rapid

time

courses,

just

as

in

normal

immature

muscles.

This

indicated

that

they

were

all

associated

with

end-

plate

cholinesterase,

and

that

they

were

relatively

close

to

one

another.

Occasional

e.p.p.s.

had

components

of

clearly

different

rise

times,

suggest-

ing

that

they

might

be

spatially

separated.

In

order

to

examine

the

spatial

100

-

0

%

C

75

-

*741

o.S

50

Co

I,

25

_-

o

{~~~

1

._

NO

~~~~~~0

41

U

.00_

O

'

'

J

//L

'^-

' '

5

10

15

20

25

30

50

75

100

Age

(days)

Text-fig.

10.

Incidence

of

multiple

innervation

during

re-innervation

by

the

soleus

nerve.

Each

point

shows

the

percentage

of

re-innervated

fibres

in

which

more

than

one

e.p.p.

component

was

detected.

At

least

twenty

fibres

were sampled

in

each

muscle.

The

soleus

nerve

was

crushed

next

to

its

entry

into

the

muscle

on

the

first

or

second

day

after

birth.

Some

muscles

had

also

been

cross-innervated

through

the

fibular

nerve

but

these

mea-

surements

were

made

in

parts

of

the

muscle

well

away

from

the

zone

of

foreign

innervation.

Note

that

very

occasional

multiple

inputs

were

seen

even

at

long

times

after

birth.

The

interrupted

line

shows

the

time

course

of

elimination

of

multiple

inputs

in

normal

muscles

(from

Text-fig.

4).

distribution

of

synapses

more,

closely

we

looked

at

the

disposition

of

cholinesterase

in

single

muscle

fibres

isolated

from

the

re-innervated

muscles.

All

of

the

end-plates

were

situated

in

the

vicinity

of

the

original

end-plate

band,

presumably

because

the

regenerating

axons

were

guided

there

along

the

old

nerve

branches.

Most

fibres

had

only

a

single

end-plate,

even

in

muscles

where

the

majority

of

fibres

were

multiply

innervated.

For

example,

in

one

10-day-old

re-innervated

muscle,

eighteen

of

twenty

fibres

tested

physiologically

were

multiply

innervated,

whereas

only

eight

ELIMINATION

OF

SYNAPSES

out

of

twenty-one

fibres

isolated

from

the

same

region

of

the

muscle

had

more

than

one

end-plate.

Thus,

it

is

likely

that

some

of

the

regenerating

axons,

converged

on

to

the

same

end-plate

site,

just

as

in

normal

immature

muscle.

This

site

is

probably

the

original

end-plate

present

before

dener-

vation

but

this

is

not

certain

in

view

of

the

lability

of

end-plate

cholin-

esterase

in

immature

fibres

(see

below).

The

presence

of

some

multiple

cholinesterase

spots

(twenty-five

double

sites

andtwo

triple

sites

in

ninety-

two

fibres

from

six

muscles

aged

10

to

21

days)

suggests

that

not

all

axons

returned

exactly

to

the

old

end-plates

and

that

some

fibres

might

have

been

innervated

at

more

than

one

location.

It

was

difficult,

however,

to

estimate

the

frequency

of

occurrence

of

distributed

end-plates

because

of

the

problem

in

knowing

whether

the

old

end-plate

had

been

re-innervated,

whether

it

was

present

but

still

denervated,

or

whether

it

had

been

elimi-

nated.

A

better

understanding

of

the

spatial

distribution

of

newly-formed

synapses

came

from

examining

muscles

that

had

been

cross-innervated

by

a

foreign

nerve

in

a

region

well

outside

the

original

end-plate

zone.

Cross-innervation

by

a

foreign

nerve

Cross-innervation

of

the

immature

soleus

muscle

took

place

only

after

disruption

of

the

original

nerve

supply,

just

as

in

adult

muscle

(Elsberg

1917;

Jansen,

Lomo,

Nicolaysen

&

Westgaard,

1973).

In

three

new-born

rats

the

superficial

branch

of

the

fibular

nerve

was

transplanted

on

to

the

soleus

muscle

without

interrupting

the

original

nerve.

At

the

time

of

the

final

experiment,

nineteen

days

after

birth,

there

were

no

contractions

to

stimulation

of

the

foreign

nerve

in

any

of

the

muscles,

even

though

the

nerve

had

grown

extensively

over

the

surface

of

the

muscle;

nor

were

any

sub-threshold

e.p.p.s

detected

in

the

one

muscle

where

a

micro-electrode

survey

was

made.

In

contrast,

innervation

by

the

foreign

nerve

took

place

readily

when

the

soleus

nerve

was

cut

or

crushed.

As

in

the

adult

(Frank

et

al.

1975)

the

foreign

nerve

never

took

over

more

than

about

a

third

of

the

muscle,

even

when

re-innervation

was

delayed

by

cutting

or

re-secting

the

original

nerve.

Two

types

of

polyneuronal

innervation

involving

the

foreign

nerve

were

seen.

Some

fibres

received

separate

inputs

from

the

foreign

and

original

nerves,

while

other

fibres

received

more

than

one

input

through

the

foreign

nerve

alone.

These

two

forms

of

polyneuronal

innervation

will

be

con-

sidered

separately,

since

our

methods

for

testing

for

them

differed

slightly

and,

more

importantly,

because

there

was

a

striking

difference

in

the

persistence

of the

two

types

of

multiple

innervation

during

maturation.

407

408

M.

C.

BROWN,

J.

K.

S.

JANSEN

AND

D.

VAN

ESSEN

The

interaction

between

foreign

and

original

nerves

The

extent

of

innervation

by

each

nerve

was

assayed

by

recording

action

potentials

and

e.p.p.s.

intracellularly

from

a

series

of

superficial

muscle

fibres,

noting

whether

each

fibre

was

innervated

by

one,

both

or

neither

nerve.

Twenty-one

muscles

were

examined

at

times

ranging

from

8

to

99

days

after

soleus

nerve

crush

and

fibular

nerve

implant.

The

usual

result

from

this

survey

was

that

the

foreign

nerve

tended

to

innervate

only

the

60

L

*

~~B

C

C

so

.

~~0

L,,

3.0

2

mV

~

~ ~ ~ ~

~~

[

20

20

msec

a

b

0

25

50

75

100

Age

(days)

Text-fig.

11.

Dual

innervation

of

individual

soleus

muscle

fibres

by

both

original

and

transplanted

foreign

(fibular)

nerve.

A,

intracellular

record

of

end-plate

potentials

to

both

soleus

nerve

stimulation

(a)

and

fibular

nerve

stimulation

(b)

from a

9-day-old

animal

in

which

the

superficial

fibular

nerve

was

transplanted

and

the

soleus

nerve

crushed

8

days

before.

B,

per-

centage

of

cross-innervated

fibres

that

were

also

re-innervated

by

the

original

nerve

in

muscles

of

different

ages.

The

initial

operation

was

per-

formed

at

days

1

or

2

in

all

animals.

Each

point

represents

the

observations

in

one

soleus

muscle

in

which

at

least

twenty

cross-innervated

muscle

fibres

were

examined.

lateral

one

third

to

one

half

of

the

muscle,

and

in

this

region

very

few

fibres

had

in

addition

been

re-innervated

by

the

original

nerve.

Fibres

in

the

medial

half

of

the

muscle

were

innervated

only

by

the

soleus

nerve.

In

the

border

region

between

the

two

territories

many

fibres

received

an

input

from

both

nerves.

Text-fig.

11

A

shows

an

example

of

a

fibre

with

a

dual

input

from

the

two

nerves,

taken

from

a

9-day-old

muscle.

Stimula-

tion

of

the

soleus

nerve

evoked

an

e.p.p.

having

two

components

with

slightly

different

latencies

and

stimulation

of

the

fibular

nerve

evoked

an

e.p.p.

of

similar

amplitude.

As

an

index

of

the

frequency

of

dual

innervation

by

the

two

nerves

we

determined

what

percentage

of

muscle

fibres

innervated

by

the

foreign

ELIMINATION

OF

SYNAPSES

nerve

had

in

addition

an

input

from

the

original

nerve.

Text-fig.

11

B

shows

the

results

for

twenty-one

muscles

in

which

ten

to

fifty

fibres

with

foreign

innervation

were

examined

in

each

muscle.

The

percentage

of

dually

innervated

fibres

varied

over

a

wide

range

but

it

did

not

appear

to

change

significantly

with

time.

The

average

percentage

for

muscles

under

25

days

of

age

(19.5

%)

is

not

significantly

different

from

that

for

older

muscles

(18

%),

nor

was

there

a

significant

difference

if

any

other

age

was

chosen

as

the

dividing

point.

The

particular

values

obtained

for

the

per-

centage

of

dually

innervated

fibres

in

each

muscle

may

not

be

very

accurate

owing

to

the

sampling

problems

that

arose

because

these

fibres

tended

to

lie

in

a

narrow

strip

between

the

foreign

and

original

nerve

territories.

Nevertheless,

the

results

demonstrate

that

substantial

numbers

of

fibres

with

both

foreign

and

original

nerve

synapses

can

be

found

at

all

ages,

and

that

there

is

no

dramatic

reduction

in

their

incidence

as

a

function

of

time.

It

is

surprising

that

the

majority

of

fibres

innervated

by

the

foreign

nerve

never

became

re-innervated

through

the

original

nerve.

The

failure

to

re-innervate

cross-innervated

muscle

fibres

in

the

immature

animals

was

not

due

simply

to

a

lowered

over-all

capacity

for

regeneration

by

the

soleus

nerve,

because

in

the

absence

of

the

foreign

nerve

the

soleus

nerve

was

capable

of

completely

re-innervating

its

muscle,

as

mentioned

above.

It

thus

seems

that

the

presence

of

foreign

innervation

can

inhibit

re-

innervation

of

old

end-plates

in

immature

rats.

The

separation

between

foreign

and

original

synapses

on

dually-

innervated

fibres

was

determined

by

staining

the

muscle

for

end-plate

cholinesterase.

In

all

but

one

experiment

the

region

of

foreign

end-plates

was

in

the

proximal

part

of

the

muscle,

well

away

from

the

original

end-

plate

band,

just

as

is

seen

in

adult

soleus

muscles

after

cross-innervation

(Frank

et

al.

1976).

The

minimal

separation

between

the

original

end-plate

band

and

the

closest

foreign

end-plates

was

at

least

1

mm

and

usually

more.

In

the

one

exceptional

case

the

foreign

nerve

had

grown

directly

into

the

original

end-plate

region

and

made

synapses

there.

It

is

perhaps

significant

that

this

was

the

one

muscle

in

which

we

did

not

find

any

fibres

receiving

an

input

from

both

foreign

and

original

nerves.

There

was

usually

a

sharp

transition

in

the

appearance

of

the

original

end-plate

band

just

at

the

border

between

foreign

and

original

nerve

territories.

The

band

was

continuous

and

of

normal

appearance

in

the

re-

innervated

zone

while

it

was

much

less

dense

and

consisted

of

scattered

isolated

end-plates

in

that

part

of

the

muscle

which

was

cross-innervated

by

the

foreign

nerve.

Examination

of

single,

isolated

fibres

confirmed

that

the

original

end-plate

cholinesterase

had

often

completely

disappeared

from

cross-innervated

fibres.

On

those

fibres

which

had

both

foreign

and

409

410

M.

C.

BROWN,

J.

K.

S.

JANSEN

AND

D.

VAN

ESSEN

original

end-plates

the

distance

between

them

was,

as

expected,

always

1

mm

or

more.

The

disappearance

of

the

cholinesterase

at

the

original

end-plates

was

not

simply

a

consequence

of

the

end-plate

itself

not

being

re-innervated.

Following

simple

denervation

of

neonatal

muscles

cholinesterase

remains

at

the

original

end-plate

for

at

least

a

month

(Lubinska

&

Zelena

1966).

In

contrast,

we

found

that

the

cholinesterase

at

denervated

end-plates

on

fibres

innervated

elsewhere

by

a

foreign

nerve

disappeared

entirely

within

about

a

week.

A

clear

illustration

of

this

came

from

a

13-day-old

muscle

in

which

the

fibular

nerve

had

been

implanted

on

the

first

day

after

birth

and

the

soleus

nerve

cut

far

enough

from

the

muscle

to

prevent

re-

innervation

by

the

time

of

the

final

experiment.

In

the

region

of

the

muscle

innervated

by

the

foreign

nerve

only

scattered

end-plates

were

visible

in

the

original

end-plate

band,

whereas

outside

the

region

of

cross-innerva-

tion

the

old

denervated

end-plates

appeared

in

the

whole

mount

as

a

continuous

dense

band

of

cholinesterase.

Isolation

of

individual

fibres

teased

from

the

cross-innervated

region

of

the

muscle

demonstrated

that

seven

out

of

eight

fibres

having

foreign

nerve

end-plates

had

completely

lost

their

original

end-plates;

the

one

fibre

having

both

foreign

and

original

end-plates

presumably

had

been

cros-innervated

for

a

shorter

time

than

the

others.

We

did

not

investigate

in

detail

the

time

required

for

foreign

innervation

to

suppress

cholinesterase

at

the

original

end-plate

but

it

is

clear

that

the

process

cannot

take

much

more

than

a

week.

The

lability

of

end-plate

cholinesterase

indicated

by

this

experiment

might

be

related

to

the

inability

of

regenerating

soleus

nerve

axons

to

re-innervate

most

cross-

innervated

muscle

fibres.

Polyneuronal

innervation

through

the

foreign

nerve

At

early

times

after

cross-innervation,

graded

stimulation

of

the

fibular

nerve

showed

that

many

muscle

fibres

were

innervated

by

more

than

one

foreign

motor

axon.

Two

lines

of

evidence

indicate

that

many

of

the

foreign

synapses

on

multiply

innervated

fibres

were

situated

at

an

appreciable

distance

from

one

another.

First,

the

different

components

of

the

e.p.p.

often

had

clearly

different

time

courses.

For

example,

in

the

fibre

illus-

trated

in

Text-fig.

12A

a

low

intensity

stimulus

(left)

to

the

fibular

nerve

elicited

an

e.p.p.

with

a

rise

time

of

about

12

msec,

while

a

stronger

stim-

ulus

(right)

evoked

an

additional

component

whose

rise

time

was

5

msec.

A

likely

explanation

for

the

difference

in

time

course

is

that

the

two

synapses

were

located

at

different

distances

from

the

recording

micro-electrode.

More

direct

evidence

for

spatially

separated

end-plates

came

from

exam-

ining

the

distribution

of

cholinesterase

on

individual

fibres

isolated

from

ELIMINATION

OF

SYNAPSES

411

the

region

of

foreign

innervation.

Many

fibres

had

two

distinct

end-plates

separated

by

distances

of

up

to

1

mm;

an

example

of

a

fibre

having

two

end-plates

about

150

,um

apart

is

shown

in

P1.

20

(arrows).

A

few

fibres

had

three

separate

end-plates

in

the

region

of

the

fibular

nerve,

and

on

one

fibre

five

closely

spaced

but

distinct

end-plates

were

found.

Multiple

fibular

end-plates

were

seen

in

more

than

a

quarter

of

the

fibres

isolated

from

four

cross-innervated

muscles

between

10

and

13

days

old

(thirty

out

of

107

fibres).

In

the

same

four

muscles

the

incidence

of

polyneuronal

innervation

estimated

physiologically

was

about

one

fibre

in

three

(twenty-

eight

of

seventy-nine

fibres).

This

suggests

that

most

but

perhaps

not

all

polyneuronal

innervation

mediated

by

the

foreign

nerve

occurs

at

spa-

tially

separate

synapses.

A

100

_

B

0>

75

"C

4)0

LLJ

50

msec

0

5

10

15

20

25

3050

75100

Age

(days)

Text-fig.

12.

Polyneuronal

innervation

by

a

foreign

nerve.

A,

multiple

e.p.p.

components

evoked

by

fibular

nerve

stimulation

in

a

19-day-old

soleus

muscle

cross-innervated

on

day

2.

Low

intensity

stimuli

(left)

produced

a

small

e.p.p.

with

rise

time

about

twice

as

long

as

that

of

the

higher

threshold

component

(right).

B,

percentage

of

foreign

innervated

fibres

in

which

more

than

one

e.p.p.

component

was

detected

during

graded

stimulation

of

the

foreign

nerve.

Each

point

gives

the

results

for

one

muscle

in

which

at

least

ten

and

usually

twenty

or

more

cross-innervated

fibres

were

ex-

amined.

The

interrupted

line

shows

the

time course

of

synapse

elimination

in

normal

muscles

(from

Text-fig.

4).

All

physiological

signs

of

polyneuronal

innervation

through

the

foreign

nerve

eventually

disappeared

in

the

cross-innervated

muscles.

Text-fig.

12B

shows

the

time

course

of

elimination

of

multiple

innervation

as

deter-

mined

physiologically

with

intracellular

recordings.

Each

point

represents

the

observations

in

one

muscle

and

gives

the

percentage

of

fibres

with

more

than

one

input

from

the

fibular

nerve.

The

incidence

of

polyneuronal

in-

nervation

was

quite

high

up

to

about

the

tenth

day

after

birth,

with

80

%

or

more

multiple

innervation

in

some

muscles.

The

disappearance

of

multiple

innervation

initially

was

similar

to

that

seen

in

normal

muscles

412

M.

C.

BROWN,

J.

K.

S.

JANSEN

AND

D.

VAN

ESSEN

(dashed

line)

but

a

substantial

percentage

of

the

multiple

inputs

persisted

for

about

a

week

longer

than

normal.

The

delay

in

removal

of

extra

inputs

was

not

due

simply

to

the

initial

surgical

interference

with

the

muscle

because

there

was

no

significant

prolongation

of

the

synaptic

elimination

process

during

re-innervation

by

the

original

nerve

(Text-fig.

10).

No

physiological

signs

of

multiple

innervation

through

the

foreign

nerve

remained

in

muscles

examined

25

days

or

more

after

birth

(135

fibres

in

eight

muscles).

Thus

the

process

that

leads

to

the

elimination

of

convergent

inputs

in

normal

muscles

evidently

is

able

to

act

in

muscle

fibres

where

the

synapses

are

spatially

distributed.

so

s

20

2

A

60

B

60

40

40

-

is

.0

0

0

10.

-

~~~~~10

E

z

55

0

~~~~~~~0

SingleO

0-1

0-2

0-3

04

05-10

Single

0

0-1

02

03

0-4

0-5

0-6

Separation

of

Separation

of

end-plates

(mm)

end-plates

(mm)

Text-fig.

13.

Histograms

of

the

separation

between

multiple

foreign

end-

plates

in

cross-innervated

soleus

muscles.

Open

column

shows

the

number

of

fibres

having

only

a

single

end-plate

in

the

region

of

foreign

innervation.

Hatched

columns

indicate

the

separation

between

multiple

foreign

end-

plates.

A,

there

were

thirty

fibres

(out

of

108)

in

the

four

young

muscles

(10-13

days)

having

more

than

one

foreign

end-plate;

two

of these

had

three

and

five

end-plates,

respectively,

which

were

included

in

the

histo-

gram

according

to

the

distance

to

their

nearest

neighbour.

B,

none

of

the

fibres

from

the

three

older

muscles

(77-99

days)

had

more

than

two

foreign

end-plates.

In

order

to

see

whether

there

was

an

elimination

of

multiple

end-plates

corresponding

to

the

disappearance

of

polyneuronal

synaptic

inputs

we

isolated

single

muscle

fibres

from

the

region

of

foreign

innervation

in

several

muscles

that

had

been

cross-innervated

for

long

periods

(11-14

ELIMINATION

OF

SYNAPSES

weeks)

and

compared

the

results

with

those

obtained

from

younger

muscles

(10-13

days).

It

was

surprising

to

find

that

despite

a

clear

reduc-

tion

in

the

incidence

of

spatially

separated

end-plates

compared

with

that

seen

in

younger

muscles,

there

was

still

a

substantial

number

of

fibres

that

retained

more

than

one

foreign

end-plate.

Text-fig.

13

shows

histograms

of

the

distance

between

multiple

foreign

end-plates

in

young

muscles

(A)

and

in

older

muscles

(B).

Almost

all

of

the

double

end-plates

that

persisted

in

the

older

muscles

were

within

100

,um

of

one

another

(e.g.

P1.

2D),

whereas

many

of

the

end-plates

in

the

younger

muscles

were

separated

by

distances

up

to

1

mm.

The

reduction

in

the

incidence

of

end-

plates

separated

by

more

than

100

#tm

provides

evidence

that

the

process

of

synapse

elimination

can

work

on

synapses

that

were

sufficiently

mature

to

have

induced

the

incorporation

of

cholinesterase

at

a

new

end-plate.

DISCUSSION

The

location

of

synapses

on

immature

muscle

fibres

The

argument

that

each

immature

muscle

fibre

receives

a

direct

and

convergent

input

from

more

than

one

motor

neurone

depends

upon

several

lines

of

evidence,

none

of

which

is

by

itself

conclusive.

Our

electro-

physiological

experiments

show

that

all

of

the

synaptic

inputs

to

any

particular

fibre

are

at

most

about

50

/tm

apart

(from

the

desensitizing

effects

of

iontophoretically

applied

ACh),

and

that

all

of

the

synapses

must

be

associated

with

end-plate

cholinesterase

(from

the

effects

of

prostig-

mine

on

e.p.p.

time

course).

Since

there

is

only

one

cholinesterase

spot

per

immature

muscle

fibre,

it

follows

that

either

the

multiple

inputs

all

con-

verge

on

to

a

single

end-plate,

or

that

there

is

strong

electrical

coupling

between

neighbouring

fibres

whose

end-plates

are

immediately

adjacent

to

one

another.

We

found

no

sign

of

widespread

electrical

coupling

between

immature

muscle

fibres,

although

the

possibility

was

not

excluded

that

there

might

be

coupling

within

small

groups

of

fibres

linked

closely

enough

to

appear

as

a

single

fibre

in

the

dissecting

microscope.

On

the

other

hand,

each

end-plate

in

an

immature

muscle

is

supplied

by

several

preterminal

axons

and

all

but

one

of

these

disappear

during

maturation.

Although

we

could

not

show

directly

that

each

preterminal

branch

belonged

to

a

differ-

ent

motor

axon,

it

seems

very

likely

that

the

multiple

preterminal

axons

are

the

anatomical

basis

for

the

presence

of

several

e.p.p.

components

in

each

immature

muscle

fibre.

Finally,

it

is

clear

that

the

multiple

synaptic

inputs

demonstrated

by

graded

stimulation

of

the

nerve

do

not

come

about

because

of

a

division

of

each

motor

axon

into

several

separate

branches

that

subsequently

converge

on

to

the

same

set

of

muscle

fibres.

This

is

because

the

tensions

produced

by

stimulating

individual

motor

units

in

413

PH

Y

26i

I4

414

M.

C.

BROWN,

J.

K.

S.

JANSEN

AND

D.

VAN

ESSEN

ventral

root

filaments

are

so

large

that

there

must

inevitably

be

a

large

overlap

in

the

innervation

of

different

motor

units.

We

therefore

conclude,

in

agreement

with

Redfern

(1970)

and

Bennett

&

Pettigrew

(1974a)

that

there

is

genuine

polyneuronal

innervation

in

immature

skeletal

muscles

of

the

rat.

In

muscles

that

are

only

a

few

days

old

all

muscle

fibres

receive

at

least

two

inputs

and

estimates

of

the

average

degree

of

multiple

innerva-

tion

range

from

about

three

(Bennett

&

Pettigrew,

1974a)

to

about

five

(Text-fig.

6,

above).

It

is

remarkable

that

all

of

the

inputs

are

crowded

into

an

area

that

is

smaller

than

an

end-plate

on

an

adult

muscle

fibre,

yet

each

synapse

is

able

to

excite

the

muscle

fibre

on

its

own.

Presumably

the

efficiency

of

transmission

is

related

in

part

to

the

higher

input

resistance

of

smaller

muscle

fibres.

Polyneuronal

innervation

has

also

been

seen

in

the

soleus

muscle

of

kittens

(Bagust

et

al.

1973),

suggesting

that

it

may

be

a

widespread

or

even

universal

event

in

the

maturation

of

mammalian

skeletal

muscle.

The

extent

of

hyperinnervation

which

exists

in

new-born

animals

will,

how-

ever,

presumably

vary

according

to

the

degree

of

maturity

at

birth.

It

will

be

interesting

to

know

whether

in

a

particular

species

the

loss

of

poly-

neuronal

innervation

occurs

synchronously

in

all

muscles, as

is

the

case

for

the

soleus

and

diaphragm

muscles

of

the

rat.

The

elimination

of

synapses

during

normal

maturation

During

the

period

when

polyneuronal

innervation

disappears

the

num-

ber

of

motor

axons

to

the

soleus

muscle

remains

constant,

while

the

aver-

age

size

of

motor

units

decreases

severalfold.

The

elimination

of

multiple

innervation

and

the

reduction

in

motor

unit

size

conceivably

might

come

about

by

muscle

fibres

dividing

in

such

a

way

as

to

leave

only

one

synapse

on

each

fibre.

Chiakulas

&

Pauly

(1965)

have

shown

that

there

may

be

an

increase

of

about

50

%

in

the

number

of

soleus

muscle

fibres

after

the

first

post-natal

week.

They

suggest

that

the

new

fibres

originate

from

myoblasts

rather

than

from

division

of

pre-existing

fibres

but

even

if

some

division

took

place

it

would

be

quantitatively

insufficient

to

account

fully

for

the

loss

of

multiple

innervation.

Hence,

it

is

evident

that

many

terminals

must

eventually

disappear

by

undergoing

either

degeneration

or

retrac-

tion.

H.

Korneliussen

(personal

communication)

found

no

obvious

signs

of

terminal

degeneration

in

electron

micrographs

of

synapses

from

immature

muscles,

suggesting

that

simple

retraction

of

extra

synapses

is

the

more

likely

alternative.

It

would

be

interesting

to

know

whether

e.p.p.

components

that

are

particularly

small

relative

to

other

components

on

the

same

fibre,

or

which

have

a

particularly

long

latency

following

nerve

stimulation

(e.g.

Fig.

2

in

Redfern,

1970)

might

be

in

the

process

of

being

eliminated

but

it is

difficult

to

see

how

to

test

these

possibilities

ELIMINATION

OF

SYNAPSES

directly.

There

are

numerous

other

changes

in

muscle

and

motor

nerve

properties

that

take

place

during

the

time

that

neuromuscular

connexions

are

maturing.

In

addition

to

growing

in

length

and

cross-sectional

area

the

muscle

develops

fast,

or

slow

contractile

properties,

or

both

(Close,

1964),

and

the

nerve

becomes

myelinated

and

gains

in

conduction

velocity.

It

is

not

yet

known

whether

the

removal

of

multiple

innervation

is

specifically

dependent

on

any

of

these

changes.

The

total

spread

of

motor

unit

sizes

within

a

given

muscle

was

con-

siderably

greater

in

young

animals

than

in

the

adult,

with

a

few

motor

units

being

within

the

adult

range

at

all

ages

we

examined.

This

suggests

that

the

shrinkage

of

motor

units

does

not

come

about

simply

by

the

synchronous

loss

by

each

motor

neurone

of

a

fixed

percentage

of

its

peripheral

terminals.

Either

there

must

be

considerable

variability

in

the

times

at

which

different

motor

units

reach

their

peak

size,

or

else

some

motor

units

never

become

much

larger

than

their

eventual

adult

size.

In

either

case,

it

appears

that

at

any

given

time

the

elimination

process

acts

predominantly

on

the

terminals

of

the

largest

motor

units.

This

could

come

about

if

the

terminals

of

small

motor

units

somehow

had

a

selective

advantage

over

the

terminals

of

larger

ones.

One

can

imagine,

for

example,

that

as

a

motor

neurone

gradually

lost

some

of

its

synapses

it

might

be

able

to

supply

its

remaining

terminals

with

some

material

or

quality

that

improved

their

chances

for

survival.

Synapse

elimination

following

partial

denervation

The

removal

of

all

but

a

few

motor

axons

to

the

soleus

muscle

leaves

the

remaining

neurones

with

the

opportunity

to

innervate

large

numbers

of

muscle

fibres

while

facing

only

minimal

competition

at

the

level

of

peri-

pheral

synaptic

interactions.

One

might

expect

to

find

that

whatever

residual

polyneuronal

innervation

was

present

would

be

removed

nor-

mally,

while

the

remaining

motor

units

would

either

remain

constant

in

size

or

even

expand

by

sprouting

on

to

neighbouring

denervated

fibres.

It

thus

came

as

a

surprise

to

find

that

partial

denervation

resulted,

on

the

one

hand,

in

a

delay

in

the

removal

of

polyneuronal

innervation

among

sur-

viving

motor

units

and

that,

on

the

other

hand,

a

reduction

in

motor

unit

size

nevertheless

eventually

took

place.

It

is

puzzling

that

multiple

innervation

should

persist

on

some

muscle

fibres

just

because

most

of

the

motor

input

to

the

muscle

has

been

re-

moved.

The

delay

might

be

an

indirect

effect

of

the

surgical

interference

within

the

spinal

cord

but

this

seems

unlikely

because

multiple

innervation

disappeared

with

its

normal

time

course

in

the

control

soleus

muscle

on

the

opposite

side.

A

more

interesting

possibility

is

that

the

delay

might

be

related

to

the

large

size

of

the

remaining

motor

units:

for

example,

if

the

14-2

415

416

M.

C.

BROWN,

J.

K.

S.

JANSEN

AND

D.

VAN

ESSEN

maturation

of

synaptic

terminals

proceeded

more

slowly

in

larger

motor

units

and

if

the

process

of

synapse

elimination

on

each

fibre

did

not

begin

until

the

favoured

terminal

reached

a

certain

absolute,

rather

than

relative

maturity,

then

one

would

expect

partial

denervation

to

delay

the

removal

of

multiple

innervation.

The

eventual

reduction

in

motor

unit

size

that

took

place

after

partial

denervation

was

usually

associated

with

gross

hypertrophy

of

the

remain-

ing

innervated

fibres

and

atrophy,

or

at

least

failure

to

grow,

of

the

dener-

vated

fibres.

Many

motor

units

were

closer

to

their

normal

adult

level

than

to

their

mean

initial

size;

this

happened

too

frequently

to

be

accounted

for

by

a

chance

selection

of

small

motor

units

at

the

time

of

the

initial

operation.

Moreover,

it

is

difficult

to

account

for

the

shrinkage

simply

in

terms

of

the

elimination

of

residual

multiple

innervation,

since

small

motor

units

were

seen

in

muscles

having

only

one

or

two

motor

units

left.

Hence

it

seems

likely

that

the

withdrawal

of

synapses

can

continue

even

though

many

previously

innervated

fibres

actually

become

denervated.

It

is

not

surprising

that

there

should

be

a

natural

limit

to

how

many

muscle

fibres

a

motor

neurone

can

innervate.

What

is

interesting,

though,

is

that

there

should

be

an

actual

reduction

in

the

upper

limit

of

motor

unit

size

during

maturation.

This

might

come

about

if

the

total

synaptic

area

that

a

motor

neurone

could

supply

grew

more

slowly

than

the

mean

area

of

individual

end-plates.

Presumably

the

natural

tendency

of

each

motor

unit

to

reduce

its

initial

size

plays

a

major

role

in

the

elimination

of

hyper-

innervation.

It

cannot

be

the

sole

mechanism

involved,

however,

because

there

must

be

some

way

of

insuring

that

each

muscle

fibre

ends

up

with

one

and

only

one

synapse.

There

is

no

inconsistency

between

the

occurrence

of

axonal

sprouting

after

partial

denervation

of

adult

muscles

(Edds,

1953;

Guth,

1962)

and

our

failure

to

detect

sprouting

of

immature

motor

units.

The

lack

of

sprouting

in

immature

muscles

is

presumably

related

to

the

large

initial

size

of

motor

units.

The

existence

of

an

upper

limit

to

motor

neurone

size

is

entirely

compatible

with

a

moderate

degree

of

sprouting

in

the

adult.

Our

results

suggest

that

the

limiting

size

of

mature

motor

units

might

be

substantially

larger

(i.e.

a

factor

of

two

or

so)

than

their

normal

adult

size.

The

relative

uniformity

in

adult

motor

unit

sizes

seen

in

the

soleus

muscle

is

not

found

in

other

muscles

such

as

the

fast-twitch

extensor

digitorum

longus,

where

there

is

about

a

tenfold

range

in

size

(Close,

1967).

It

would

be

interesting

to

know

whether

this

greater

variability

is

related

to

larger

inherent

differences

among

fast

motor

neurones.

ELIMINATION

OF

SYNAPSES

The

elimination

of

synapses

after

re-innervation

and

cross-innervation

Multiple

innervation

through

regenerated

axons

of

the

original

nerve

disappeared

with

a

time

course

similar

to

that

seen

in

normal

muscles.

This

shows

that

the

process

of

synapse

elimination

is

not

heavily

influenced

by

either

the

age

of

the

terminals

involved

or

the

period

of

muscle

inactivity

before

re-innervation

took

place.

The

lack

of

an

effect

is

remarkable

in

view

of

the

delay

in

the

removal

of

hyperinnervation

following

several

other

procedures

that

might

seem

to

be

less

disruptive

to

the

synapses

involved.

For

example,

partial

denervation

leads

to

a

prolonged

innerva-

tion

overlap

among

motor

units

that

were

not

directly

affected

by

the

operation.

In

addition

Benoit

&

Changeux

(1975)

have

shown

that

teno-

tomy

causes

a

delay

in

the

removal

of

polyneuronal

innervation,

and

we

have

seen

a

similar

effect

following

the

injection

of

botulinum

toxin

into

immature

muscles

(M.

C.

Brown,

J.

K.

S.

Jansen

and

D.

Van

Essen,

un-

published

observations).

The

events

that

took

place

during

cross-innervation

by

a

foreign

nerve

differed

in

several

important

respects

from

those

that

occurred

during

re-

innervation

and

during

the

maturation

of

normal

muscles.

Firstly,

many

synapses

on

fibres

multiply

innervated

through

the

foreign

nerve

were

situated

at

separate

sites

up

to

k/mm

apart.

Distributed

synapses

were

never

seen

in

normal

muscles,

anid

were

probably

not

the

most

common

form

of

multiple

innervation

during

re-innervation.

Presumably,

the

difference

between

re-innervation

and

cross-innervation

is

that

regenera-

ting

axons

from

the

original

nerve

tend

to

be

guided

back

to

the

sur-

viving

old

end-plates.

Secondly,

a

moderate

level

of

polyneuronal

inner-

vation

through

the

foreign

nerve

persisted

through

the

third

week

after

birth.

That

these

multiple

inputs

were

event

ally

eliminated

means

that

the

interactions

leading

to

synapse

elimination

can

act

over

considerable

distances.

The

delay

in

synapse

removal

suggests

that

the

process

might

take

longer

when

the

synapses

are

some

distance

apart.

Indirect

support

for

this

comes

from

the

observation

that

there

was

a

higher

incidence

of

fibres

having

e.p.p.

components

with

different

rise

times

in

the

2-3-week-

old

cross-innervated

muscles

than

in

the

younger

ones.

s

final

major

difference

seen

after

foreign

innervation

was

that

some

multiple

end-plates

persisted

indefinitely

in

cross-innervated

muscles.

This

came

about

in

two

distinct

ways:

some

fibres

received

separate

functional

inputs

from

the

foreign

and

original

nerves;

while

other

fibres

had

two

closely

spaced

end-plates

in

the

region

of

foreign

innervation,

even

though

they

had

only

one

input

that

could

be

demonstrated

physiologically.

The

distance

between

synapses

from

the

foreign

and

original

nerves

was

al-

ways

1

mm

or

more,

suggesting

that

the

degree

of

separation

between

417

418

M.

C.

BROWN,

J.

K.

S.

JANSEN

AND

D.

VAN

ESSEN

synapses

may

determine

whether

one

of

them

will

be

eliminated.

If

dis-

tance

is

indeed

an

important

factor,

then

the

basis

for

synapse

elimination

cannot

be

a

process

that

affects

the

entire

muscle

fibre

equally,

such

as

the

action

potential

or

contractile

activity.

It

is

interesting

that

the

distance

of

about

1

mm

suggested

by

the

present

experiments

is

similar

to

the

spacing

between

synapses

in

slow

muscles

of

the

adult

chicken.

The

development

of

the

adult

pattern

of

innervation

in

the

chicken

appears

to

come

about

by

a

selective

elimination

of

syn-

apses

which

initially

are

more

closely

spaced

than

in

the

adult

(Bennett

&

Pettigrew,

1974b).

Slow

muscles

in

the

chicken

do

not

conduct

action

potentials

and

it

has

been

suggested

that

the

decremental

spread

of

syn-

aptic

potentials

might

be

involved

in

determining

the

spacing

between

synapses

(Gordon,

Perry,

Tuffery

&

Vrbova,

1974).

This

cannot,

however,

be

the

sole

explanation

for

mammalian

skeletal

muscles,

which

conduct

action

potentials

throughout

the

period

when

synapse

elimination

takes

place.

The

persistence

of

double

esterase-stained

end-plates

in

the

region

of

foreign

innervation

appears

to

be

quite

a

different

phenomenon

than

the

dual

innervation

through

foreign

and

original

nerves

that

was

seen

in

the

same

muscles.

Foreign

and

original

end-plates

on

the

same

muscle

fibre

remained

functional

indefinitely,whereas

we

never

saw

physiological

signs

of

inputs

from

more

than

one

foreign

motor

axon

at

long

times

after

cross-

innervation.

If

there

were

functional

synapses

at

both

foreign

end-plates,

they

must

both

have

belonged

to

the

same

parent

axon.

This

is

an

intriguing

possibility,

because

it

suggests

that

a

muscle

fibre

is

not

inherently

resistant

to

maintaining

more

than

one

closely-spaced

end-

plate,

provided

that

they

are

all

supplied

by

one

axon.

The

key

to

protection

of

multiple

synapses

from

elimination

might

be

the

synchrony

in

the

patterns

of

synaptic

activity,

or

it

might

be

some

other

specific

factor

common

to

all

terminals

of

a

particular

motor

neurone.

Differences

between

immature

and

adult

muscle

There

are

several

important

differences

in

the

events

that

take

place

during

re-innervation

and

cross-innervation

of

immature

v8.

adult

muscles.

Perhaps

the

most

striking

are

the

changes

at

the

original

end-plate

region.

Our

results

show

that

immature,

cross-inervated

muscle

fibres,

but

not

fibres

which

are

simply

denervated,

lose

the

cholinesterase

at

the

original

end-plate

quickly

and

completely.

Associated

with

this

is

a

loss

in

most

cross-innervated

fibres

of

the

ability

to

accept

re-innervation

by

the

original

nerve.

In

adult

cross-innervated

soleus

muscle

fibres

the

original

end-plate

properties

are

much

more

stable.

The

intensity

of

cholinesterase

staining

declines

but

never

disappears

and

original

end-plates

will

accept

ELIMINATION

OF

SYNAPSES

re-innervation

for

many

weeks

in

spite

of

foreign

innervation

of

the

same

fibres

(Frank

et

al.

1975).

The

degree

of

polyneuronal

innervation

seen

at

various

stages

of

re-

and

cross-innervation

is

also

different

for

immature

and

adult

rats.

Multiple

innervation

was

seen

in

only

a

small

percentage

(10-15

%)

of

fibres

examined

in

the

first

week

of

re-innervation

of

the

adult

rat

dia-

phragm

(Jansen

&

Van

Essen,

1975).

It

is

not

known

whether

the

multiple

inputs

were

spatially

distributed

and

whether

they

would

eventually

be

removed.

In

adult

cross-innervated

muscles,

polyneuronal

innervation

through

the

foreign

nerve

occurs

commonly,

even

many

months

after

the

operation

(Frank

et

al.

1975).

However,

recent

experiments

by

T.

Lomo

and

C.

R.

Slater

and

by

H.

Sommerschild

(personal

communications)

suggest

that

there

is

a

removal

of

closely

spaced

foreign

synapses

even

in

the

adult

rat.

Possible

mechanism

of

synapse

elimination

The

elimination

of

peripheral

synapses

could

in

principle

be

controlled

by

a

variety

of

different

mechanisms

involving

any

combination

of

the

three

cell

types

present

at

the

neuromuscular

junction:

neurones,

Schwann

cells

and

muscle

fibres.

Our

results

serve

to

restrict

the

number

of

accep-

table

hypotheses

by

specifying

more

explicitly

the

conditions

under

which

synapse

elimination

can

or

cannot

take

place.

The

most

important

res-

traints

are

the

following:

(1)

the

end

result

of

maturation

is

invariably

that

each

muscle

fibre

is

innervated

by

a

single

motor

axon;

(2)

during

normal

maturation

there

are

never

a

substantial

number

of

muscle

fibres

that

are

completely

denervated

but

after

partial

denervation

withdrawal

of

terminals

continues

even

though

some

muscle

fibres

consequently

are

left

without

innervation;

(3)

terminals

from

small

motor

units

are

more

likely

to

survive

than

those

of

larger

units;

(4)

the

elimination

process

has

a

restricted

range:

it

can

act

over

distances

of

hundreds

of

microns

but

not

over

millimetres.

The

elimination

of

synapses

does

not

appear

to

be

a

random

process,

either

in

the

sense

that

a

particular

motor

neurone

loses

a

randomly

selected

proportion

of

its

terminals

(since

this

would

lead

to

transient

denervation

of

many

fibres),

or

in

the

sense

that

on

each

muscle

fibre

there

is

a

random

choice

of

which

terminal

is

to

survive

(since

terminals

of

smaller

units

survive

preferentially).

Instead,

there

seems

to

be

some

sort

of

competition

among

the

synapses

on

each

fibre.

It

may

be

of

interest

to

speculate

on

the

cellular

mechanisms

that

might

operate

in

a

way

con-

sistent

with

these

restraints.

Selection

by

Schwann

cells.

Since

all

of

the

terminals

on

a

normal

im-

mature

muscle

fibre

are

crowded

into

a

single

end-plate,

it

is

possible

to

419

420

M.

C.

BROWN,

J.

K.

S.

JANSEN

AND

D.

VAN

ESSEN

imagine

that

a

single

Schwann

cell,

or

row

of

Schwann

cells,

could

select

one

of

the

terminals

to

be

the

survivor.

This

could

be

accomplished

by

the

myelination

of

only

one

preterminal

axon,

providing

that

the

remaining

unmyelinated

axons

were

eventually

withdrawn.

The

selection

of

which

axon

to

myelinate

could

be

made

on

the

basis

of

some

quality

such

as

axon

diameter,

in

which

case

the

smaller

motor

units

might

attain

a

selective

advantage

simply

because

their

terminals

become

larger.

The

myelination

hypothesis

cannot,

however,

account

for

the

events

related

to

the

elimination

of

spatially

separated

end-plates

in

any

simple

way.

Interactions

between

nerve

terminate.

The

elimination

process

might

in-

volve

interactions

only

between

the

nerve

terminals

co-existing

on

a

fibre,

without

any

direct

participation

of

the

muscle

fibre

or

of

Schwann

cells.

The

interactions

might

take

the

form

of

one

terminal

physically

crowding

out

the

others,

or

of

a

biochemical

signal

passing

from

one

terminal

to

the

others

to

induce

the

withdrawal

of

synapses.

In

either

case

this

type

of

hypothesis

suffers

from

the

same

disadvantage

as

the

myelination

hypothe-

sis,

in

that

it

is

difficult

to

explain

direct

and

specific

interactions

between

terminals

separated

by

hundreds

of

microns.

Interactions

directly

involving

the

muscle

fibre.

In

order

to

account

for

the

elimination

of

distributed

inputs

in

seems

necessary

to

invoke

some

type

of

signal

passing

from

the

muscle

fibre

to

the

synapses

upon

it.

It

is

not

clear

what

the

nature

of

the

communication

between

muscle

and

nerve

could

be.

There

might,

for

example,

be

a

signal

from

the

muscle

that

actively

induced

the

removal

of

unwanted

terminals.

A

different

type

of

explanation

is

that

synapses

might

compete

for

some

substance,

provided

in

limited

amounts

by

the

muscle

fibre

and

needed

for

the

survival

of

the

synapse.

The

persistence

of

distant

multiple

synapses

might

then

reflect

the

limits

over

which

synapses

could

compete

for

this

hypothetical

substance.

It

is

evident

that

we

are

now

at

the

stage

of

understanding

only

a

few

of

the

rules

of

the

process

whereby

neuromuscular

connexions

reach

their

mature

state.

These

rules

serve

to

restrict

the

types

of

mechanisms

that

could

control

the

elimination

of

synapses

but

as

yet

they

provide

only

indirect

hints

concerning

the

specific

cells

and

intercellular

signals

in-

volved.

A

major

impetus

behind

the

elimination

process

seems

to

be

a

natural

inability

of

the

motor

neurone

to

maintain

more

than

a

fraction

of

its

original

complement

of

synapses.

The

choice

of

which

synapses

are

to

be

removed

results

from

a

competitive

interaction

between

all

of

the

synapses

within

a

limited

distance

on

each

muscle

fibre.

The

nature

of

the

competition

is

still

not

known

but

it

is

probably

geared

to

favour

the

survival

of

terminals

from

smaller

motor

units

and

it

appears

to

involve

a

signal

from

muscle

to

nerve.

ELIMINATION

OF

SYNAPSES

We

thank

Mr

HAvard

T0nnesen

for

invaluable

technical

assistance.

Drs

Eric

Frank,

Terje

Lomo

and

Wesley

Thompson

have

all

given

good

advice

at

various

stages.

D.V.E.

was

supported

by

a

Helen

Hay

Whitney

fellowship.

M.C.B.

was

European

exchange

fellow

of

the

Royal

Society.

REFERENCES

AKERT,

K.

&

SANDRi,

C.

(1968).

An

electron-microscopic

study

of

zine

iodide-

osmium

impregnation

of

neurons.

I.

Staining

of

synaptic

vesicles

at

cholinergic

junctions.

Brain

Re8.

7,

286-295.

BAGUST,

J.,

LEWIs,

D.

M.

&

WEsTERwAN,

R.

A.

(1973).

Polyneuronal

innervation

of

kitten

skeletal

muscle.

J.

Physiol

229,

241-255.

BENNETr,

M.

R.

&

PErioGRiw,

A.

G.

(1974a).

The

formation

of

synapses

in

striated

muscle

during

development.

J.

Phygiol.

241,

515-545.

BENNETT,

M.

R.

&

PE=IGREW,

A.

G.

(1974b).

The

formation

of

synapses

in

re-

innervated

and

cross-innervated

striated

muscle

during

development.

J.

Phy8iol.

241,

547-573.

BENOrT,

P.

&

CHANGEux,

J.-P.

(1975).

Consequences

of

tenotomy

on

the

evolution

of

multiinnervation

in

developing

rat

soleus

muscle.

Brain

Re8.

99,

345-358.

BLAcKsHAw,

S.

W.

&

WARNER,

A.

E.

(1976).

Low

resistance

junctions

between

mesoderm

cells

during

development

of

trunk

muscles.

J.

Phy8iol.

255,

209-230.

BROWN,

M.

C.,

JANSEN,

J.

K.

S.

&

VAN

EssEN,

D.

(1975).

A

large-scale

reduction

in

motor

neurone

peripheral

fields

during

post-natal

development

in

the

rat.

Acta

phy8iol.

8cand.

95,

3-4A.

BROWN,

M.

C.,

JANSEN,

J.

K.

S.

&

VAN

ESSEN,

D.

(1976).

Polyneuronal

cross-

innervation

of

the

immature

rat

soleus

muscle.

Acta

phygiol.

8cand.

96,

20A.

CHAKuLAs,

J.

J.

&

PAULY,

J.

E.

(1965).

A

study

of

postnatal

growth

of

skeletal

muscle

in

the

rat.

Anat.

Rec.

152,

55-61.

CLOSE,

R.

(1964).

Dynamic

properties

of

fast

and

slow

skeletal

muscles

of

the

rat

during

development.

J.

Phy&iol.

173,

74-95.

CLOSE,

R.

(1967).

Properties

of

motor

units

in

fast

and

slow

skeletal

muscles

of

the

rat.

J.

Phy8iol.

193,

45-55.

DENNIs,

M.

(1975).

Physiological

properties

of

junctions

between

nerve

and

muscle

developing during

salamander

limb

regeneration.

J.

Physiol.

244,

683-702.

EDDS,

M.

V.,

JR

(1953).

Collateral

nerve

regeneration.

Q.

Rev.

Biol.

28,

260-276.

ELSBERG,

C.

A.

(1917).

Experiments

on motor

nerve

regeneration

and

the

direct

neurotization

of

paralyzed

muscles

by

their

own

and

by

foreign

nerves.

Science,

N.Y.

45,

318-320.

FRANK,

E.,

GAUTVIK,

K.

&

SOMMERScHILD,

H.

(1976).

Persistence

of

junctional

acetylcholine

receptors

at

mammalian

neuromuscular

synapses

following

dener-

vation.

Cold

Spring

Harb.

Symp.

quant.

Biol.

40,

275-281.

FiLNK,

E.,

JANSEN,

J.

K.

S.,

L0Mo,

T.

&

WESTGAARD,

R.

H.

(1975).

The

interaction

between

foreign

and

original

motor

nerves

innervating

the

soleus

muscle

of

rats.

J.

Phy8iol.

247,

725-743.

GORDON,

T.,

PERRY,

R.,

TuFEERY,

A.

R.

&

VRBOVA,

G.

(1974).

Possible

mechanisms

determining

synapse

formation

in

developing

skeletal

muscle

of

the

chick.

Cell

&

Tissue

Re8.

155,

13-25.

GumT,

L.

(1962).

Neuromuscular

function

after

regeneration

of

interrupted

nerve

fibres

into

partially

denervated

muscle.

Expl

Neurol.

6,

129-141.

GumAN,

E.

&

HANITJRKovI,

V.

(1966).

Motor

unit

in

old

age.

Nature,

Lond.

209,

921-922.

HUGHES,

A.

F.

W.

(1968).

A8pects

of

Neural

Ontogeny,

pp.

249.

London:

Lagos

Press.

421

422

M.

C.

BROWN,

J.

K.

S.

JANSEN

AND

D.

VAN

ESSEN

JANsEN,

J.

K.

S.,

L0Mo,

T.,

NIcOLAYsEN,

K.

&

WEsTGAARn,

R.

H.

(1973).

Hyper-

innervation

of

skeletal

muscle

fibres:

dependence

on

muscle

activity.

Science,

N.Y.

181,

559-561.

JANwsi,

J.

K.

S.

&

VAN

EssEN,

D.

(1975).

Re-innervation

of

rat

skeletal

muscle

in

the

presence

of

a-bungarotoxin.

J.

Phy8iol.

250,

651-667.

KARNovsKY,

M.

(1964).

The

localization

of

cholinesterase

in

rat

cardiac

muscle

by

electron

microscopy.

J.

CeU

Bid.

23,

217-232.

LIimmsssER,

L.

&

PUAR,

G.

(1974).

Synaptic

transmission

and

cell

death

during

normal

ganglionic

development.

J.

Phyoiol.

241,

737-749.

LImBINsKA,

L.

&

ZEimENA,

J.

(1966).

Formation

of

new

sites

of

acetylcholine

esterase

activity

in

denervated

muscles

of

young

rats.

Nature,

Lond.

210,

39-41.

REDFERN,

P.

A.

(1970).

Neuromuscular

transmission

in

new-born

rats.

J.

Phy8sol.

209,

701-709.

EXPIANATION

OF

PLATES

PLATE

1

A,

multiple

preterminal

axons

(arrows)

to

single

end-plates

in

an

8-day-old

rat

diaphragm

stained

with

zinc

iodide-osmium.

The

muscle

fibres,

which

run

horizon-

tally

in

the

photograph,

stain

lightly

and

are

not

easily

resolvable.

B,

single

end-plate

stained

for

cholinesterase

on

a

fibre

isolated

from

a

6-day-

old

soleus

muscle.

PLATE

2

A,

cross-section

from

the

middle

of

a

32-day-old

normal

soleus

muscle,

stained

with

Toluidine

blue.

B,

cross-section

of

a

33-day-old

soleus

muscle

partially

denervated

since

day

4.

There

were

321

large

fibres

in

this

muscle,

which

was

innervated

by

only

two motor

axons.

The

scale

applies

to

A

and

B.

C,

multiple'foreign

end-plates

on

a

cross-innervated

muscle

fibre.

The

fibre

was

isolated

from

a

13-day-old

cross-innervated

muscle

after

staining

for

cholinesterase.

Two

end-plates

are

visible

at

a

separation

of

about

150

1m.

D,

double

foreign

end-plates

on an

isolated

fibre

from

a

72-day-old

cross-inner-

vated

muscle

stained

for

cholinesterase.

No

fibres

with

multiple

foreign

e.p.p.

com-

ponents

were

seen

in

this

muscle.

The

Journal

of

Physiology,

Vol.

261,

No.

2

Plate

1

A

~~~~~~~~I

_

50

Um

M.

C.

BROWN,

J.

K.

S.

JANSEN

AND

D.

VAN

EVEN

(Facing

p.

422)

Uhe

Journal

of

Physiology,

Vol.

261,

No.

2

Plate

2

A

.~~~~~~~nw

__

200

pm

C

50

,m

100pam

M.

C.

BROWN,

J.

K.

S.

JANSEN

AND

D.

VAN

ESSEN