LETTERS
TO
NATURE
Essential
function
of
LlF
receptor
in
motor
neurons
Meng
LI*,
Mlchael
Sendtner
t "
Austin
Smlth
*t
• ,Centre for Genome Research, University
of
Ed
inburgh,
King's Buildings. West Mains Road, Edinburgh EH9
3JQ,
UK
t Clinical Research Unit for Neuroregeneration. Department
of
Neurology, University
of
Wurzburg, Josef-Schneider-Strasse
11,
97080
Wurzburg. Germany
D EVELOPME'iT and ma
intenan
ce of
th
e ma
mmali
an nCn'QU5 system
is depe
nd
e
nt
upon neurotrophic cytokines.
One
class of neuro-
tro
phi
c factor acts
thr
ough rcccptor complexes involving
th
e low-
affinity leukaemia inhibitory faclor rece
ptor
subunit (LlF_R)'
-3
.
Me
mb
ers of
thi
s
fa
mil
y of
cy
to
kin
es,
such
as
ciliary neurotro
phic
factor (eNTF') and leukae
mi
a inhibitory fac
lor
(L
IF
), ha
ve
pro-
fo
und
e
ff
ects on the sUn"ival and maintenance
of
motor neuron
s4-IO
,
Recently it was reported that
mi
ce lacking
LlF-R
di
e shortly a
ft
er
birth
l l
unlike mice lacking
CN
TF or
LlF
whi
ch are \'iable. Here
we
describe
hi
stopathological analyses of li
fr
mutants that rev
ea
l
a loss > 35% of facial motor neurons,
40
0/0
of spinal motor neurons
a
nd
50% of neurons
in
the nucleus ambig
uu
s.
Th
ese findings point
to
th
e existence
of
a
li
ga
nd
fo
r LlF-R that is required for the
no
rm
al
de
\'elo
pm
e
nt
of motor neurons
in
both bra
in
stem nuclei
a
nd
spinal cord.
A deletion
of
-20kb
of
the lifr gene was created by h
omo
lo-
gous recombination in embryonic stem (ES) cells (Fi
g.
1)
.
The
deleted sequence contains seven exons
that
en
co
de
mo
st of the
presumptive ligand-binding region. This was replaced by an
internal ribos
ome
entry sit
c-IacZ
/ neo cassette
(lRES~
geopA )1
2.
13, thereby int roducing a histochemical
reporter
in
to
: To whom corresponden
ce
should be addressed.
724
the locus.
Th
e
mutant
allele
was
established on hybrid 1
29
/ C
BA
a
nd
129
/
MFI
backgrounds
by
appr
opriate
mating
s of transmit-
ting c
himaera
s.
Breeding
of
the
mutat
ed allele
to
homozygosity resulted in
the
producti
on
of
liveborn
hom
ozygous
pup
s on both
ge
netic
backgrounds.
The
mutant
animals were
sma
ll
er
and
had a 20%
reduction
in
body
weight (1.33 ± 0.22 g,
fl
= 7)
com
p
are
d with
heterozygous
(1.67 ± 0.19 g, 1/
~
18)
and
wild-type (1.67 ± 0.22 g,
11
= 8) l
ittcnnates
at
birth.
The
y lacked vigour, did not feed and
rapidly became
moribund.
Although
they exhibited limb move-
ments, they were incapable of righting.
I n a
ll
cases
so
far, homo-
zygotes ha
ve
died within 24 h
of
birth a
nd
usua
ll
y within the
first 12 h. Simil
ar
findings were recently described by Ware
et
al.
11
•
a
,
I
,
N
'
~~
C
LlF-Rprolein
/ \
,
,
12
.5
kb
6.2
kb
,
,
/
, ,
, ,
';)
t
"'
,
I
I
,
R
8.5
kb
2.4
kb
10.5
kb
, , ,
20
kb
d~leted
~
i2.5
kb _
__
6.2k
b_
~
E 1
, ,
9Jkb
... 10.5
kb
...
9.
3kb
... a.5
kb
• ... ... 2.4 kb
I
lift
gene
W
i
ld
·ty~
Recomblnant
NATURE·
VOL
378
14
DECEMBER
1995
c
t
d
\~
(,
.
,
,,,,
'"
t
~
\'
'
~
I
"
,
,.
,
,
..
.
~;
..
'.
"
' ,
"
..
d.
•
..
\
"
•
.
~
.~
FIG.
1 Targeted distribution of
the
liff
gene. a, Deletion strategy. Top,
schematic representation of LlF-R protein showing the two cytokine
recep
tor
domains
boxed. Vertical lines indicate the conserved cysteine
residues, and sol
id
bars represent
the
WSXWS
motifs. Bottom, partial
genomic structure
of
the
liff
gene showing homology
arms
(thick lines)
included in
the
targeting vector, positions
of
EcaRI restriction sites and
sizes
of
diagnostic restriction fragments. Filled circles indicate
the
posi·
tions
of
flanking probes used
to
confirm replacement targeting events.
The deletion region contains seven exons encoding
amino
acid residues
85-444
in
the cytokine receptor
domains
of
LIF-A.
Th
is region was
replaced by
an
en-2 splice acceptor-IRES·pgeopA cassette
12
, flanked
by
EeoR
I restriction sites.
b,
Hybridization analysis
of
tail
DNA
from wild·
type, heterozygous
and
homozygous
tiff
mutant
mice.
DNA
samples
were digested with EeoRI and sequentially hybridized with a 5' genomic
probe external
to
the
targeting construct (left)
and
a cDNA fragment
(right). The genomic probe was
an
800-bp
EeoRI/BamHI fragment
shown in a. The cDNA probe was an internal
980-bp
fragment
that
contains a sequence from the deleted exons
3-9
plus a sequence from
exon
10.
The 9.3-kb band present
in
heterozygotes and homozygotes
NATU
RE
VO
L
378
14
DECEMB
ER
1995
LETTERS
TO
NATURE
FIG
. 2 Expression of integrated
~·galactosidase
reporter in the 9.5-day
mouse embryo and neonatal brain stem and spinal column.
a,
b,
Whole-
mount
in
situ hybridizations of 9.5-d mouse embryos:
a,
wild-type
embryo hybridized with antisense
lif'
probe:
b,
heterozygous embryo
hybridized with antisense
/ael
probe. c-
g,
Localization of sites of
Xgal
staining derived from expression of
~geo
integrated into the
lifr
gene.
C,
d, neonatal brainstem regions. Dotted regions mark (C) the facial and
(d) the ambiguus nuclei. Scale bar.
400
IJm.
e.
Neonatal
sp
inal cord
and dorsal root ganglia. Scale bar,
100
~lm
.
f.
Neurons of the nucleus
ambiguus of a
liff
I newborn mouse showing prominent
Xgal
staining.
Scale bar,
50
~m
.
g. Neonatal spinal motor neurons. Scale bar.
25
~Im
.
METHODS.
a,
b, Whole·mount
in
situ hybridizations were made
by
using
digoxigenin-Iabelled probes
25
•
The
liff
probe was generated byantisense
transcription of the
980
-bp cDNA fragment described
in
Fig.
1.
The
1125-nt
p·galactosidase probe was complementary to sequence down-
stream of the unique Sacl site in the lacZ gene. c- g, Tissue processing
and Xgal staining for
p.~alactosidase
on
12
-
~m
cryostat sections
was
performed as described
6.
e
'4t
¥
•
,
~'.~~
~~
'
ri
;i
..
.
';f1"
,
g
"-
corresponds to
the
recombinantjunction fragment that includes exon 10.
ME
T
HODS.
a,
liff
genomic clones were isolated from a
129Sv
genomic
library and the positions of exons determined (M,L.,
A.S
.•
A.
Cozens and
I.
Chambers, unpublished results). A promoterless targeting vector was
constructed
by
subcloning the
5,5·kb
BamHi/Sall 5' fragment and the
5.0-kb
Smai/Sacl
3' fragment into pGEMEM (Promega) and inserting the
GTIRES-pgeopA cassette. The construct was excised from the plasmid
backbone by digestion with Notl and Sfil and electroporated into
CGRB
ES
cells maintained in the absence
of
feeders
2J
• After selection
in
G418
the correct replacement event was identified in
19
/
58
colonies
by
filter
hybridization with both 5' and 3' flanking probes. Single integration
events were confirmed by hybridization with an en-2 probe.
b,
Two
clones injected into
C57BLj6
blastocysts gave rise to germline trans-
mission
of
the
mutated allele. Animals were typed by hybridization ana-
lysis of tail
DNA
and/ or by p-galactosidase·staining
of
ear biopsies
24
,
Heterozygous offspring were backcrossed to the appropriate parental
strain before intercrossing
at
the F2 and F3 generations. Homozygous
offspring were identified and the deletion was confirmed by
DNA
filter
hybridization.
725
LETTERS
TO
NATURE
The
differentiation
and
survival
of
111010r
neurons
ill
vilro
4
.
1
'/
and
ill
vivo
S,to.
1n
have been shown
to
be strongly enhanced
by
CNTF
and
UF.
indicative
ora
requiremen
t for
L1
F-R
function.
The inert condition
of
L1F-R-dcficicnt nconates was suggestive
of
possible neuronal dysfunction. Therefore expression
of
LI
F-R
within the
nervous
system
and
the
effects
of
gene
ablation
on
111010r
neuron
populati
ons
were investigated
in
morc
detail.
The fidelity
of
the reporter int
egn
ttcd into the !ifr gene was
established
by
the close similarity between
hybridization
patterns
obtained ill
sil/l
with antiscnsc
LI
F-R
and
~-galactosidase
probes
and the cor
re
spondence observed between pre
se
nce
of
IR ES-
pgeo fusion
tran
scripts
and
histochemically
detectable
p.gala
cto·
sidase activity
(Fig.
2.
and
results not
show
n). Expression in the
developing
nerv
ous syst
em
was
apparent
in
the
hind
brain.
crania
l
neural
crest
and
neural
tube
or
9.5
·d
ay
embryos
(Fig.
2a.
b).
The
distribution
or
p·ga
l
actosidase
activity
in
embryonic
( 14.4·day). n
eona
tal
and
aduh
brain
was
then
examined.
Specific
staining was evident in
cells
or
the s
ubepend
ymal
zone.
epen·
dyme
and
glia limi tans. Expression
was
pro
m
inent
in
ncurons
in the brainstem mo
tor
nuclei.
including
the
hypog
lossal
and
ra
cial nuclei
(Fig.
2e.
and
results
not
s
hown).
and
al
so
in
the
nucleus ambigulIs (Fig.
2d,f).
In
sections
through
embryonic
and
lleonata
l spinal
co
lumn.
p.galactosidase
s
taining
was
mo
st
pr
ono
un
ce
d in sen
sory
n
euro
ns
or
the
dorsa
l
root
gang
lia
but
cou
ld
also
be detected in m
os
t
of
th
e s
pinal
motor
neuron
s
(Fig.
2e.
g. ,md results
not
shown).
In
addition
so
me
non·neuronal
ce
ll
s were
stained.
no
t
ably
astrocytes
of
the glia
limitans
in the
central
nervous
systcm
and
probabl
y
also
Schwann
ce
ll
s in
peripheral nerves.
The
signifi
ca
nce
or
the
exp
ression
of
LlF·R
in s
pinal
motor
Ileurons
and
brainstem
motor
nuclei
was
investigated
by
cou
nting
motor
neurons
in the Illmbar (LI - L6)
spinal
cord
and
facia'l nuclei
or
nconata
l
homozygous.
heterozygous
and
wild·
type
littermate
s.
The
high
expressio
n
or
the
p.galactosidase
reporter
round
in the nuclells
ambigulIs
al
so
prompted
an
ana
l
ys
is
or
this
str
ucture. MOlOr
neurons
within
the
nucleus
ambiguus
innervate
the
oesophagus.
phar
ynx
and
lar
ynx.
and
coordinate
swa
ll
owing.
In ,tdditi
on
there
are
neurons
in the
nucleus
ambiguus
or
rats
and
mice
that
gene
rate
respiratory
rhythm
14
If,.
Data
presented
in Fig. 3
and
Table
1
show
that
the
numbers
of
motor
neurons
in
L1F·R-deficient
anima
ls were
reduced by >
35
% in the racial nucleus
and
by
>40
% in the
lumbar
spinal
cord.
The
nucleus
ambiguus
neurons
were
reduced
by
over
50%. Incr
eased
numbers
of
apparentl
y
dying
cells
wi
th
p
yknotic
nuclei were evident
in
both
spi
nal·cord
and
brains
t
em
FIG.
3 Neuronal morphology
in
the facial nucleus of
new-
born wild-type and lifr mutant
mice.
Facial
nucleus of wild-
type
(a, c)
and
homozygous
lifr mutant (b. d) newborn
mice from the same litter.
Scale
bars:
a,
b.
100
~Im:
c,
d,
25
~lm.
METHODS.
Paraffin sections
(7
~lm)
were
prepared from
brainstems of control and
lifr ./ mice
and
Nissl
stained
5
.
6
•
726
TABLE
1 Numbers of spinal motor neurons and neurons
in
facial a
nd
ambiguus nuclei of newborn wild-type, heterozygous and UF-R-deficient
mice
+ +
+ -
- / -
Lumbar region
(Ll-L6)
3,213 ±
230
N.O.
1.856
±25
1
In
~
3)
In
~
5)
Facial nucleus 3,108±
310
3,377 ± 394
1.951 ±
135
In
~
6)
In
~
5)
In
~
l1)
Nucleus ambiguus
883±82
705±34
406
± 32
In
~
5)
n
~
3)
In
~
5)
The
lumbar spinal cord and brainstems of newborn mice or mice
delivered
by
caesarian section at
day
20 after conception
were
fixed
by
transcardial perfusion or immersion fixation with 4% paraformaldehyde.
Paraffin serial sections
(7
pm)
were
prepared and processed
as
described
5
.
6
. After Nissl staining, the number of neurons
was
counted
in
every tenth section of the lumbar
(Ll-
L6
) region of"the spinal
cord
and every fifth section of the brainstem nuclei.
Only
neurons with distin·
guishable nucleus and nucleolus and with clearly identifiable Nissl
structure
were
counted.
Average
diameters of the nucleoli
were
as
fol·
lows. Facial motor neurons:
+/+
mice,
2.64
±
O.43~m;
+/-
mice,
2.32±0.22
pm; - / - mice, 2.14 ± 0.16
~m:
neurons of the nucleus
ambiguus:
+/
+ mice,
2.36
±
0.36~m;
- / - mice,
2.34
±
0.41~m
(means±s.d. for at least five determinations).
The
co
unts of facial and
ambiguus neuronal
nucleoli
were
corrected for double counting of split
nuc
leoli. Spinal motor·neuron counts are uncorrected.
N.D.
not deter-
mined.
Values shown are means ± s.e.m. of each group. Numbers of
neurons are different between
+/
+ and - / - mice for each group
(P<O.005, Student's Hest).
nuclei in paraffin
sec
tion
s
rrom
homozygotes
(Fig.
3d,
and
results
not
s
hown).
Thi
s suggests
that
el
evate
d
ce
ll
death
or
motor
neurons
is
taking
place
around
the
time
or
birt
h.
Lo
ss
or
racial
motor
ne
ur
o ns
sho
uld not affect the viability of
ncwborn
mice. Im
pairment
or
the
nucleus
ambiguus
is
or great
er
significance. These n
euro
ns
coo
rdi
nate
mov
e
ment
s
of
the
oesophagus
a
nd
pharynx
and
are
thererore
re
quired
for s
uckling
.
I
ntegrity
or
the
nucleus
ambiguus
is
also
essential
ror
gener
ation
of
respi
rator
y
rhythm
16. It
is
probable
that
l
oss
of
th
ese
neurons
is a
con
t
ribu
t
ory
ractor
to
the
early
mortalit
y
observed
in
LI
F·
R-deficient mice. It
wou
ld
be
or
interest
to
kn
ow
whether
this
structure
is
affected in
other
mouse
mutant
s
that
die
sh
ort
ly
arter
birth
and
exhibit
reduced
number
s
or
neuron
s in
other
brainstem
nuclei.
NA
T
URE·
VOL
378
14
DECEMBER
1995
The motor-neuron losses and deficits in the nucleus ambig-
uus
of
L1F-R-deficient mice establish
that
these neurons are
dependent on signalling through this receptor during
embryonic development. Mice lacking
CNTF
do
not show
loss
of
motor neurons before adulthoodO
and
no severe neu-
ronal defects have been observed
in
LIF-deficient mice (rers
17
,
18
and M.S., unpublished observations). Mice deficient
for both
CNTF
and
LIF
are viable and
do
not
show functional
motor defects
at
birth (M.S., unpublished observations). The
lesions present
in
newborn LIF-R-deficient animals must there-
fore reflect the activity
of
one
or
more distinct cytokines
Received 21
July;
accept
ed
27 October
1995.
1.
Gearing,
D.
P.
el
al.
EMBO
J.
10,
2839-2848
(
1991
).
2.
Gearing.
D.
P.
el
al. Science
255,
1434
-
143
7 (1992).
3.
Davis
,
S.
et al. Science
280,
1805-1808
(1993).
4.
Arakawa,
Y.,
Sendtner,
M.
& Thoenen,
H.
J.
Neurosci.
10,
3507-3515
(1990
).
5.
Sendtner,
M.,
Kreutzberg,
G.
W.
& Thoenen ,
H,
Nature
345,
44D-441
(1990).
6.
Masu,
Y.
et
al.
Nature
365,
27-32
(1993).
7.
Richards,
L.
Joo
Kilpatrick, T.
J,
Bartlett,
P.
F.
& Murphy,
M.
J.
Neurosci.
Re
s.
33,476-484
(1992).
8.
Martinou, J·C., Martinou,
I.
& Kato,
A.
C.
Neuron 8,
737-744
(1992).
9.
Zu
rn
,
A.
D.
& Werren,
F.
Devl Bioi.
163,
309
- 3
15
(
1994
).
10
. Hughes,
R.
A
..
Sendtner, M. & Thoenen, H. J. Neurosci. Res.
36,
663-671
(
1993
).
11.
Ware,
C.
B. et al. Development
121,
1283
-
12
99
(
1995
).
12. Mount/ord,
P.
et al.
Proc.
natn. Acad.
Sci.
U.SA
91,
4303-4307
(1994).
13. Mount/ord,
P.
& Smith,
A,
G.
Trends Genet.
11,
179-184
(1995).
14.
Ellenberger,
H. H.
& Feldman,
J.
L.
J.
comp. Neurol.
294,
202-211
(1990).
15. Smith,
J.
C.,
Greer,
J
J,
Liu
,
G.
& Feldman, J L. J. Neurophysiol.
64,
1149-1169
(1990).
NATURE·
VOL
378
.
14
DECEMBER
1995
LETTERS
TO
NATURE
active during embryogenesis. Candidates include the recently
described cardiotrophin-I (ref. 19)
and
a putative mouse
orthologue
of
oncostatin M
20
Interaction with the ligand may
also involve the
CNTF-receptor
a
(CNTF-Ra)
subunit, which
associates with
LIF-R
after
CNTF
binding
3
CNTF-Ra
is
expressed in the developing nervous system,
in
particular
in
motor
neurons
21
,
and
a preliminary report indicates
that
its
deletion results in developmental defects more severe than
produced by deletion
of
CNTF22. The unidentified cytokine
may
therefore be a novel ligand for
CNTF-Ra
in
addition to
LIF-R. 0
16. Paton, J
F.
R.,
Ramirez,
J·M.
& Richter, D.
W.
Pflugers Arch.
428,
250-260
(1994).
17.
Stewart.
C.
L. et al. Nature
359,
76-79
(1992).
18. Escary, J.·L .• Perreau, J
.•
Dumenil,
D.
,
Ez
ine,
S.
&
BrO
let,
P.
Nature
363,
361-364
(1993).
1
9.
Pennica,
D.
et al. J. bioi. Chem.
270,
10915-10922
(
1995
).
20.
Rose
,
T.
M. & Bruce. G.
A.
Proc.
natn. Acad. Sci.
USA.
88,
8641-8645
(
1991
).
21. lp,
N.
Y.
et al. Neuron
10,
89-102
(1993
).
22. Stahl,
N.
& Yancopoulos,
G.
D.
J.
Neurobiol.
25,
1454-1466
(1994
).
23. Smith,
A.
G.
J.
Tiss. Cult. Meth.
13,
89-94
(1991).
24. Beddington,
R.
S.
poo
Morgenstern,
Joo
Land,
H.
& Hogan,
A.
Development
106,
37-46
(1989).
25.
Wilkinson,
D.
G.
In
Situ Hybridization: a Practical Approach (
IRL
, Oxford, 1992).
26.
Skarnes,
W.
Coo
Auerbach,
B.
A.
&
Jo
yner,
A.
L.
Genes Dev. 6,
903-918
(1992).
ACKNOWLEDGEMENTS
.
M.L
and
M.S.
contributed equally to this work.
We
thank
I.
Chambers
for
LlF
-R
cDNA
, J. Nichols and
J.
Ure for chimaera production,
A.
Je
ske for expert animal
husbandry, and
H.
Doppler for e
xc
elle
nt
technical assistance;
T.
T
aga
for discussion, and
1.
Chambers and
H.
Niwa for comments
on
the manuscript. This research was supported
by
the
Biotechnology and Biological Sciences Resear
ch
Council,
UK,
and the Deutsche Forschungs·
gemeinschaft, Germany.
727