glial cell-line derived neurotrophic factor-dependent fusimotor neuron survival during development
TRANSCRIPT
Glial cell-line derived neurotrophic factor-dependent fusimotor
neuron survival during development
Jennifer Whiteheada,d, Cynthia Keller-Pecke, Jan Kuceraf,g, Warren G. Tourtellottea,b,c,d,*
aDepartment of Pathology (Neuropathology), Northwestern University, Chicago, IL 60611, USAbDepartment of Neurology, Northwestern University, Chicago, IL 60611, USA
cThe Institute for Neuroscience, Northwestern University, Chicago, IL 60611, USAdVeterans Administration Lakeside Division in Chicago, Chicago, IL 60611, USA
eDepartment of Anatomy and Neurobiology, Washington University, 660 S Euclid Ave., St Louis, MO 63110, USAfDepartment of Neurology, Boston University School of Medicine, Boston, MA, USA
gThe Veterans Administration Medical Center in Boston, Boston MA, USA
Received 30 July 2004; received in revised form 13 September 2004; accepted 14 September 2004
Available online 6 October 2004
Abstract
Glial cell-line derived neurotrophic factor (GDNF) is a potent survival factor for motor neurons. Previous studies have shown that some
motor neurons depend upon GDNF during development but this GDNF-dependent motor neuron subpopulation has not been characterized.
We examined GDNF expression patterns in muscle and the impact of altered GDNF expression on the development of subtypes of motor
neurons. In GDNF hemizygous mice, motor neuron innervation to muscle spindle stretch receptors (fusimotor neuron innervation) was
decreased, whereas in transgenic mice that overexpress GDNF in muscle, fusimotor innervation to muscle spindles was increased. Facial
motor neurons, which do not contain fusimotor neurons, were not changed in number when GDNF was over expressed by facial muscles
during their development. Taken together, these data indicate that fusimotor neurons depend upon GDNF for survival during development.
Since the fraction of cervical and lumbar motor neurons lost in GDNF-deficient mice at birth closely approximates the size of the fusimotor
neuron pool, these data suggest that motor neuron loss in GDNF-deficient mice may be primarily of fusimotor neuron origin.
q 2004 Elsevier Ireland Ltd. All rights reserved.
Keywords: Glial cell-line derived neurotrophic factor; Motor neuron; Muscle spindle; Fusimotor
1. Introduction
Glial cell-line derived neurotrophic factor (GDNF) is a
potent motor neuron survival factor for developing and
mature motor neurons (Henderson et al., 1994; Li et al.,
1995; Yan et al., 1995; Zurn et al., 1994). It is best known for
its ability to rescue motor neurons from death after injury
(Henderson et al., 1994; Matheson et al., 1997; Oppenheim
et al., 1995; Ramer et al., 2000; Vejsada et al., 1998) and in
animal models of primary motor neuron degeneration (e.g.
amyotrophic lateral sclerosis; ALS) (Kaspar et al., 2003;
0925-4773/$ - see front matter q 2004 Elsevier Ireland Ltd. All rights reserved.
doi:10.1016/j.mod.2004.09.003
* Corresponding author. Address: Departments of Pathology (Neuro-
pathology), Neurology, The Institute for Neuroscience, Northwestern
University, 303 E. Chicago Ave., Chicago, IL 60611, USA. Tel.: C1 312
503 2415; fax: C1 312 503 2459.
E-mail address: [email protected] (W.G. Tourtellotte).
Mohajeri et al., 1999). Comparatively little is known about
its role during motor neuron development but it appears to
function as a target derived neurotrophic factor for motor
neurons in vivo because it is expressed in developing and
adult target muscles (Trupp et al., 1996; Wright and Snider,
1996), it is retrogradely transported in motor axons (Tomac
et al., 1995; Yan et al., 1995) and a fraction of spinal motor
neurons appear to co-express the preferred GDNF signaling
receptor complex, GFRa1/c-ret (Leitner et al., 1999;
Oppenheim et al., 2000). GDNF is among a large group of
motor neuron trophic factors that include other GDNF
family members (Neurturin and Persephin) (Kotzbauer
et al., 1996; Milbrandt et al., 1998), brain-derived neuro-
trophic factor (BDNF) (Oppenheim et al., 1992; Sendtner
et al., 1992a; Yan et al., 1992), neurotrophin-4 (NT4)
(Koliatsos et al., 1994), insulin-like growth factor-1 (IGF-1)
(Lewis et al., 1993; Neff et al., 1993), ciliary neurotrophic
Mechanisms of Development 122 (2005) 27–41
www.elsevier.com/locate/modo
J. Whitehead et al. / Mechanisms of Development 122 (2005) 27–4128
factor (CNTF) (Sendtner et al., 1992b), leukemia inhibitory
factor (LIF) (Cheema et al., 1994), cardiotrophin-1 (Pennica
et al., 1996) and hepatocyte growth factor/scatter factor
(HGF/SF) (Yamamoto et al., 1997). GDNF is one of
the most potent motor neuron survival factors identified
to date but like all other known motor neuron survival
factors, only a fraction of motor neurons depend upon it
during development or are rescued by it after injury. In
GDNF-deficient and GFRa1-deficient mice approximately
22–35% of spinal motor neurons are lost by birth,
demonstrating the existence of GDNF-dependent and
GDNF-independent motor neuron populations (Cacalano
et al., 1998; Moore et al., 1996; Oppenheim et al., 2000;
Sanchez et al., 1996).
The GDNF-dependent motor neurons in the spinal cord
have remained poorly characterized. Studies focused on
identifying the spinal motor neurons lost in GFRa1-
deficient or GDNF-deficient mice have shown that they
are distributed in a complex manner throughout the rostral-
caudal extent of the spinal cord, in the medial and lateral
motor columns, and that most of them express the preferred
GFRa1/c-ret GDNF signaling receptor complex (Garces
et al., 2000; Homma et al., 2003; Oppenheim et al., 2000).
There is no single trophic factor identified that all motor
neurons depend upon, but it is clear that motor neurons
derive survival and differentiation signals from muscle
during development since mutant mice lacking all skeletal
muscles due to impaired myogenesis also lack somatic
motor neurons (Grieshammer et al., 1998; Kablar and
Rudnicki, 1999). Accordingly, GDNF-dependent motor
neurons probably derive GDNF from a specific population
of target muscle fibers once innervation is initiated. Indeed,
this appears to be the case for a particular population of
brachial motor neurons that innervate the latisssimus dorsi
(LD) and cutaneous maximus muscles (CM) in mice
(Haase et al., 2002; Helmbacher et al., 2003). These
muscles selectively express GDNF during their innervation
where it regulates the expression of the Ets related
transcription factor Pea3 in the CM and LD motor neuron
pools in the spinal cord. Interestingly, GDNF plays a
critical role in establishing the motor neuron pools that
innervate CM and LD at embryonic day 12.5 (E12.5) but it
is not required for motor neuron survival at this develop-
mental stage since spinal motor neuron numbers are not
altered at E12.5 in GDNF-deficient mice (Haase et al.,
2002; Oppenheim et al., 2000). At later stages of
development however, GDNF clearly has an additional
role related to motor neuron survival that is more
widespread than its highly selective role in regulating
Pea3 expression in brachial motor neurons that innervate
CM and LD muscles.
We hypothesized that GDNF-dependent spinal motor
neurons could be further characterized by examining
whether they represent functionally similar motor neurons
that derive GDNF from a related population of target
muscle fibers. Here we demonstrate that GDNF is
expressed in a sub-population of muscle fibers known as
intrafusal muscle fibers within muscle spindle stretch
receptors and that fusimotor neurons innervating intrafusal
muscle fibers depend upon trophic signals from spindles
for survival during development. Moreover, fusimotor
innervation is selectively reduced in GDNF haploinsuffi-
cient (GDNFC/lacz) mice and it is increased in mice
expressing GDNF in all muscle fibers (MyoGDNF). These
results demonstrate a role for GDNF in the survival of
fusimotor neurons, an important class of motor neurons
involved in proprioception. Thus, we propose that spinal
motor neuron loss in GDNF-deficient mice is, at least in
part, attributed to loss of fusimotor neurons.
2. Results
2.1. GDNF expression is localized to intrafusal muscle
fibers in skeletal muscle spindle stretch receptors
To understand the role of GDNF as a motor neuron
survival factor for a subset of spinal motor neurons, we
examined the expression of GDNF in developing muscles
using heterozygous mice (GDNFC/lacz) that express a loss-
of-function n-terminal GDNF/b-galactosidase (lacz) fusion
protein (Moore et al., 1996). In E12.5 embryos, GDNF/
lacz expression was identified in proximal forelimb and
hindlimb buds, and in mesenchymal cells positioned along
the ventral rostral-caudal extent of the spinal cord as
previously described (Haase et al., 2002; Wright and
Snider, 1996). GDNF expression in muscle was examined
at later developmental timepoints since widespread
motor neuron death occurs between E14.5 and E15.5 in
GDNF-deficient mice (Oppenheim et al., 2000). GDNF/
lacz positive skeletal muscle fibers were first identified in
E15.5 embryos (data not shown) and by E18.5, the
individual GDNF/lacz expressing muscle fibers were
clearly identifiable as single intrafusal muscle fibers within
muscle spindle stretch receptors. The first intrafusal fiber to
form after the onset of spindle morphogenesis is the
nuclear bag2 fiber, which can be identified by its selective
expression of slow-developmental myosin heavy chain
(Kucera and Walro, 1995). In nascent spindles, GDNF/lacz
was expressed in intrafusal fibers that formed after the
bag2 fiber, in either a nuclear bag1 or a nuclear chain
intrafusal fiber (Fig. 1A,B). At E18.5, the GDNF/lacz
reaction product clearly delineated the entire length of
single intrafusal fibers (Fig. 1C, arrows). In adult spindles,
GDNF/lacz was identified in all intrafusal muscle fibers
within each spindle throughout the longitudinal extent of
the fibers (Fig. 1D,E). Whereas all intrafusal muscle fibers
expressed high levels of GDNF, a small number of
extrafusal muscle fibers were noted to express low levels
of GDNF in late embryonic, perinatal and adult GDNF/
lacz heterozygous mice.
Fig. 1. GDNF/lacz expression in muscle spindle intrafusal fibers during development in mice. (A) Shortly after the onset of spindle morphogenesis in mouse
skeletal muscle, GDNFlacz is expressed in single intrafusal muscle spindle fibers (E18.5, lacz histochemistry; arrow pointing to single transversely sectioned
blue fiber). (B) An adjacent tissue section shows slow-developmental myosin heavy chain expression by the first intrafusal muscle fiber to form during spindle
morphogenesis, the nuclear bag2 fiber (S46 immunohistochemistry; arrow pointing to single transversely sectioned red fiber). At this developmental stage,
GDNF/lacz and slow-developmental myosin heavy chain expression are not colocalized to the same intrafusal fibers, indicating that GDNF/lacz is initially
expressed in an intrafusal fiber that forms after the nuclear bag2 fiber (either nuclear bag1 or nuclear chain fiber). Thus, GDNF/lacz expression is initiated in
intrafusal fibers well after the onset of spindle morphogenesis. (C) GDNF/lacz expression in intrafusal muscle fibers in intercostal muscles from E18.5
GDNF/lacz heterozygous mice. Single GDNF/lacz expressing intrafusal fibers are running in anti-parallel directions (blue fibers), reflecting the anti-parallel
orientation of the two intercostal muscle layers that interconnect the ribs (r). (D and E) In spindles from adult skeletal muscle, GDNF/lacz expression is
identified in all intrafusal fibers. The same spindle is transversely sectioned through (D) the equatorial and (E) the polar region demonstrating GDNF/lacz
reaction product throughout the longitudinal extent of the fibers. (Magnification barsZ1 mm.)
J. Whitehead et al. / Mechanisms of Development 122 (2005) 27–41 29
2.2. Fusimotor neuron innervation depends upon intact
spindles during development
Somatic motor neurons require signals from skeletal
muscles for survival during development but the relation-
ship between particular trophic factors and the motor
neurons that depend upon them remains very poorly defined
(Grieshammer et al., 1998; Kablar and Rudnicki, 1999). To
test the hypothesis that the innervation of spindles by
fusimotor neurons depends upon spindle-derived trophic
signals, we examined mice deficient in the transcription
factor Egr3. Egr3 and GDNF are expressed by spindles
during their morphogenesis (Fig. 2A–D) and adult
Egr3-deficient mice lack muscle spindles (Tourtellotte
and Milbrandt, 1998). Cervical and lumbar ventral spinal
roots were examined in wild type and Egr3-deficient mice
to identify large diameter skeletomotor and small
diameter fusimotor axons (Kawamura et al., 1977;
McHanwell and Biscoe, 1981). In Egr3-deficient mice,
O95% of the small diameter fusimotor axons (g) were
Fig. 2. Muscle spindles produce trophic signals that are essential for fusimotor innervation. (A) In newborn skeletal muscle (PN0.5), expression of the zinc-finger
transcription factor Egr3 colocalizes (brown labeling) with (B) GDNF/lacz expression (blue labeling) in a similar subset of intrafusal muscle fibers. (C) In adult
spindles, Egr3 is expressed (brown labeling) in all intrafusal muscle fibers similar to (D) the expression pattern of GDNF/lacz (blue labeling). (E) Extrafusal
muscle fibers are innervated by large diameter motor neurons that give rise to large diameter myelinated (a, skeletomotor) axons, whereas intrafusal fibers are
innervated by small diameter motor neurons that give rise to small diameter (g, fusimotor) axons. In wild type mice (top), a bimodal axon diameter-frequency
distribution distinguishes fusimotor from skeletomotor axons, accounting for approximately 35% and 65%, respectively, of the motor axons innervating skeletal
muscle from this ventral spinal root. In adult Egr3-deficient mice (bottom), greater than 95% of the fusimotor axons are missing while there is no significant
change in the number of skeletomotor axons. Fusimotor axon loss as a consequence of absent muscle targets (spindles) in Egr3-deficient mice indicates that
spindles produce trophic signals that are required to support fusimotor neuron innervation during development. (Magnification barZ5 mm (A–D) and 10 mm (E).)
J. Whitehead et al. / Mechanisms of Development 122 (2005) 27–4130
absent while no significant difference in the number of
large diameter skeletomotor axons (a) was identified
(Fig. 2E). That fusimotor neuron innervation to abnormal
Egr3-deficient spindles is present at birth (Tourtellotte
et al., 2001) and that fusimotor axons degenerate after
spindles disassemble in the periphery, indicates that
fusimotor neurons require intact spindles for their survival
during development.
J. Whitehead et al. / Mechanisms of Development 122 (2005) 27–41 31
2.3. GDNF/lacz haploinsufficient mice have abnormal
fusimotor but not skeletomotor innervation
We observed that fusimotor innervation depends upon
intact muscle spindles and that GDNF expression is highly
restricted to intrafusal muscle fibers within spindles. To
determine whether fusimotor neurons specifically depend
upon GDNF produced by spindles, we examined fusimotor
and skeletomotor axons in ventral spinal nerve roots from
adult GDNF/lacz heterozygous mice. Since there are
presently no molecular markers available that distinguish
fusimotor from skeletomotor neurons in the spinal cord,
Fig. 3. Fusimotor neurons depend upon muscle-derived GDNF for target innervatio
contain skeletomotor (large diameter) and fusimotor (small diameter) axons. In GD
decreased in the ventral spinal roots (L4 shown here). (B) The axon diameter–frequ
roots defines the fusimotor axon (%3.5 mm in diameter) and skeletomotor axon
diameters. (C) L4 ventral roots from GDNF/lacz heterozygous mice contained 11
roots. The axon loss was entirely attributed to a 46% loss of fusimotor axons (P!0
skeletomotor axons between wild type and GDNF/lacz heterozygous adult L4 ve
and the differential diameter-frequency relationship
between skeletomotor and fusimotor axons is poorly
established at birth, we examined ventral roots from adult
GDNF/lacz heterozygous and wild type mice (Fig. 3A). It
was necessary to examine heterozygous mice since GDNF/
lacz homozygous mice suffer from renal agenesis and die
shortly after birth (Moore et al., 1996; Pichel et al., 1996;
Sanchez et al., 1996). To define whether fusimotor or
skeletomotor neurons were preferentially affected, we
analyzed the diameter and frequency of myelinated axons
in cervical (C4) and lumbar (L4) ventral roots from adult
wild type and GDNF/lacz heterozygous mice. The axon
n. (A) Ventral spinal roots (L4 shown here) from adult wild type mice (left)
NF/lacz heterozygous mice (right), the number of fusimotor axons appears
ency analysis from adult wild type and GDNF/lacz heterozygous L4 ventral
(O3.5 mm in diameter) populations from the bimodal distribution of axon
% fewer total myelinated axons (P!0.05, Student’s t-test) than wild type
.01, Student’s t-test), as there was no significant difference in the number of
ntral roots. (Magnification barsZ10 mm.)
Fig. 4. Axon diameter–frequency analysis demonstrates a selective loss of
fusimotor axons in muscle nerves from GDNF/lacz heterozygous mice. (A)
Adult medial gastrocnemius muscle nerves had 9.1% fewer axons (P!0.05, Student’s t-test) in GDNF/lacz heterozygous mice relative to wild
type mice. The axon loss was attributed to a 29% loss of small diameter
axons (P!0.05, Student’s t-test), the majority of which were fusimotor
axons. (B) Similarly, soleus muscle nerves from GDNF/lacz heterozygous
mice had 11% fewer axons (P!0.05, Student’s t-test) with all of the axon
loss attributed to a 31% loss of small diameter fusimotor axons (P!0.01,
Student’s t-test). (C) A small branch of the soleus nerve (the ‘propriocep-
proprioceptive’ branch) innervates Golgi tendon organs and spindles but
contains no skeletomotor axons. There was a 43% loss of axons (P!0.05,
Student’s t-test) in the proprioceptive branch of the soleus muscle nerve,
which was entirely attributed to a 63% loss of small diameter axons (P!0.0001, Student’s t-test), the vast majority of which are fusimotor axons.
J. Whitehead et al. / Mechanisms of Development 122 (2005) 27–4132
diameter measurements showed a characteristic bimodal
distribution that identified fusimotor axons (‘small’ diam-
eter myelinated axons %3.5 mm in diameter) and skeleto-
motor axons (‘large’ diameter myelinated axons O3.5 mm
in diameter) (Fig. 3B) (Kawamura et al., 1977; Ringstedt et
al., 1998; Tourtellotte et al., 2001; Tourtellotte and
Milbrandt, 1998). In L4 ventral roots, 34% of the motor
axons originated from fusimotor neurons and of those, 46%
were lost in GDNF/lacz heterozygous mice (P!0.01;
Student’s t-test; Fig. 3C). Analysis of cervical (C4) ventral
nerve roots showed similar results (not shown). Surpris-
ingly, the motor axon loss in GDNF/lacz heterozygous mice
was entirely attributed to a loss of fusimotor axons since
there was no significant change in the number of
skeletomotor axons (Fig. 3C).
Individual hindlimb muscle nerves supplied by other
lumbar and sacral spinal roots were also examined to
confirm the results obtained from the analysis of lumbar
(L4) and cervical (C4) ventral roots. In the medial
gastrocnemius muscle nerve (supplied by L4, L5 and the
first sacral (S1) nerve roots) there was a 9.1% decrease in the
number of axons entering the muscle in GDNF/lacz
heterozygous mice (P!0.05; Student’s t-test). Moreover,
diameter-frequency analysis of axons in the medial gastro-
cnemius muscle nerve showed a selective loss of small
myelinated axons (29%, P!0.05; Student’s t-test) and no
loss of large diameter myelinated axons in GDNF/lacz
heterozygous mice (Fig. 4A). Similarly, in the soleus
muscle nerve (also supplied by L4, L5 and S1 spinal
roots) there was an 11% decrease (P!0.05; Student’s t-test)
in the total number of myelinated axons in GDNF/lacz
heterozygous mice and a selective loss of small (31%,
P!0.01; Student’s t-test) but not large diameter myelinated
axons (Fig. 4B). Unlike the analysis performed on lumbar
and cervical ventral spinal roots that contain primarily
motor axons, examination of the muscle nerves was
confounded by the fact that they contained a mixture of
sensory and motor axons. Some small diameter myelinated
sensory axons were included in the morphometry analysis of
the muscle nerves, which may have contributed to
decreasing the magnitude of small diameter axon loss
relative to the loss identified in ventral spinal roots from
GDNF/lacz heterozygous mice. This interpretation was
validated by analyzing a purely proprioceptive branch of the
soleus muscle nerve, which innervates only Golgi tendon
organs and muscle spindles. This nerve branch contains a
mixture of myelinated axons that consist of large diameter
sensory axons (Group Ia, II and Ib) and small diameter
axons (primarily fusimotor efferents) that innervate 2–4
muscle spindles and three Golgi tendon organs, but no
extrafusal muscle fibers in the soleus muscle (see Section 4)
(Tourtellotte et al., 2001; Vult von Steyern et al., 1999).
Morphometric analysis of the proprioceptive branch of the
soleus muscle nerve from adult GDNF/lacz heterozygous
mice showed a 43% loss of the total number of myelinated
axons (P!0.05; Student’s t-test), a 63% loss of small
J. Whitehead et al. / Mechanisms of Development 122 (2005) 27–41 33
diameter axons (P!0.0001; Student’s t-test) and no change
in the number of large diameter axons (Fig. 4C). These data
confirm the observations that GDNF is essential for
fusimotor neuron innervation of intrafusal muscle fibers
during development in mice.
2.4. Increased fusimotor innervation in transgenic mice
that over-express GDNF in muscle
Increased GDNF expression in skeletal muscle of
transgenic mice (MyoGDNF) or chronic perinatal GDNF
administration leads to increased numbers of spinal and
cranial somatic motor neurons (Oppenheim et al., 2000; Sun
et al., 2003). The results demonstrating that fusimotor
neurons require GDNF produced by target intrafusal fibers
raises the possibility that at least some of the somatic motor
neuron increase observed in GDNF treated or MyoGDNF
transgenic mice could be attributed to an increase in the
number of fusimotor neurons that was never considered in
any of the previously published reports. MyoGDNF mice
were examined at PN28, a time point when the transient
perinatal neuromuscular junction polyinnervation had
decreased to control levels (Nguyen et al., 1998). The
innervation pattern in muscle spindles from hindlimb
muscles was examined using confocal microscopy to
characterize axons innervating spindles from whole mount
muscle preparations. Muscle spindles from MyoGDNF
mice were hyperinnervated by small diameter axons
entering the equatorial zone of spindles when compared to
wild type littermate mice. The small diameter axons were
identified as fusimotor axons (g) since they terminated on a-
bungarotoxin labeled motor endplates near the polar region
of the intrafusal muscle fibers (Fig. 5A,B). Hindlimb and
paraspinal muscles were embedded in resin and serially
sectioned to quantify the fusimotor innervation to individual
spindles (see Section 4). In MyoGDNF mice, medial
gastrocnemius muscle spindles were innervated by 63%
more fusimotor axons (P!0.0001, Student’s t-test), soleus
muscle spindles were innervated by 68% more fusimotor
axons (P!0.0001, Student’s t-test) and paraspinal muscle
spindles were innervated by 118% more fusimotor axons
(P!0.005, Student’s t-test) compared to wild type mice
(Fig. 5C). No change in the sensory innervation to spindles
was identified. Fusimotor hyperinnervation of spindles was
further confirmed by examining the proprioceptive branch
of the soleus nerve in MyoGDNF and wild type mice.
Consistent with the results obtained from the individual
spindle analyses, there was an increase in the total number
of axons (38% increase; P!0.01, Student’s t-test) and small
diameter myelinated (primarily fusimotor) axons (50%
increase; P!0.05, Student’s t-test) but no difference in
the large diameter (sensory proprioceptive) axons within the
nerve branch (Fig. 5D). The increase in fusimotor
innervation to spindles was primarily due to an increase in
fusimotor innervation from separate axons rather than
intramuscular preterminal axon branching since there was
an increase in the number of small diameter (fusimotor)
axons in the proprioceptive branch of the soleus nerve
(Fig. 5D) as well as the medial gastrocnemius and tibialis
anterior muscle nerves (not shown). These results demon-
strate that GDNF overexpression in muscle during devel-
opment leads to a sustained increase in fusimotor
innervation of spindles.
Newborn MyoGDNF mice are reported to have approxi-
mately 45% more lumbar and brachial spinal motor neurons
than wild type mice (Oppenheim et al., 2000). Since there
are no selective molecular markers for fusimotor neurons, it
was not possible to directly examine the relative proportion
of fusimotor and skeletomotor neurons present in the spinal
cord of neonatal MyoGDNF mice. To examine whether
fusimotor neurons were selectively increased by muscle
specific GDNF expression, we examined cervical and
lumbar ventral spinal roots from PN28 MyoGDNF and
wild type mice. Since the observations were similar, we
focused our analysis on the L4 ventral roots. Compared to
wild type ventral roots (Fig. 6A, left), the cross-sections
from lumbar (L4) ventral roots of MyoGDNF mice (Fig. 6A,
right) showed a large increase in the number of small-
diameter myelinated axons. A characteristic bimodal axon
diameter distribution was noted in the ventral roots
(Fig. 6B). However, there were 105% more total myelinated
axons in MyoGDNF L4 ventral roots than wild type L4
ventral roots (P!0.001, Student’s t-test). The small
diameter myelinated axons were increased by 324% relative
to wild type (P!0.0005, Student’s t test) whereas the
large diameter myelinated axons were decreased by 37%
(P!0.05, Student’s t-test). The axon counts were an
accurate reflection of the number of motor neurons in this
spinal cord segment since there was no detectable axon
branching within the ventral roots when proximal and distal
sections of the roots were compared. The marked increase in
the number of small diameter myelinated axons parallels the
increased fusimotor innervation that was directly observed
in muscle spindles from MyoGDNF mice. However, it was
not possible to exclude the possibility that a portion of the
small diameter axons could have originated from atrophic
skeletomotor neurons or that some small diameter axons
ended blindly in muscle.
2.5. Facial motor neurons which lack fusimotor neurons
are not increased in MyoGDNF mice
The facial motor nerve innervates facial muscles and
lacks fusimotor axons since facial muscles are devoid
of spindles. In GDNF-deficient mice, previous reports
indicated that there is no loss of facial motor neurons
(Mikaels et al., 2000; Moore et al., 1996; Sanchez et al.,
1996), which may reflect the fact that there are no fusimotor
neurons present. To examine whether facial motor neurons
respond to GDNF during development, we examined facial
motor neurons and axons in wild type and MyoGDNF mice.
We found no difference in the number of facial motor
Fig. 5. Muscle specific over expression of GDNF in transgenic mice (Myo-GDNF) leads to fusimotor hyperinnervation of muscle spindles. (A) Motor and
sensory innervation to individual muscle spindles was observed in intact PN28 skeletal muscles using confocal microscopy. In wild type muscles, small
diameter fusimotor axons (g; neurofilament immunohistochemistry, green) entered the spindle capsule near the equatorial region and terminated on a-
bungarotoxin labeled motor endplates (red). The intrafusal muscle fibers were identified by their specific myonuclear expression of Egr3 (Egr3; Egr3
immunohistochemistry; blue). Note the large diameter group Ia sensory axon (Ia; neurofilament immunohistochemistry, green) forming an anulospiral ending
around an intrafusal fiber. (B) Spindles from Myo-GDNF mice showed obvious hyperinnervation by small caliber myelinated axons (g; neurofilament
immunohistochemistry, green) that terminated on a-bungarotoxin motor endplates (red), the latter feature defining them as fusimotor axons. (C) Serial sections
from resin embedded muscle tissues were used to count the number of fusimotor axons that innervated individual spindles by tracing the axons terminating on
the polar region of intrafusal fibers to their entry into the spindle capsule. Spindles from Myo-GDNF medial gastrocnemius muscle were innervated by 63%
more fusimotor axons (P!0.0001, Student’s t-test), spindles from soleus muscle were innervated by 68% more fusimotor axons (P!0.0001, Student’s t-test)
and spindles from lumbar paraspinal muscles were innervated by 118% more fusimotor axons (P!0.005, Student’s t-test) relative to spindles from the same
muscles in age matched wild type mice. (D) The proprioceptive branch of the soleus muscle nerve from Myo-GDNF mice had 38% more axons (P!0.01,
Student’s t-test) than wild type mice. The increase in the number of axons was entirely attributed to a 50% increase (P!0.05, Student’s t-test) in the number of
small diameter fusimotor axons. (Magnification barsZ100 mm.)
J. Whitehead et al. / Mechanisms of Development 122 (2005) 27–4134
neurons between PN28 wild type and MyoGDNF mice
(Student’s t-test; PZ0.73, Fig. 7A). To confirm these results,
we counted the number of axons in facial motor nerves.
Consistent with the facial motor neuron counts, no
differences were observed in the number of facial motor
axons between PN28 wild type and MyoGDNF mice
(Student’s t-test; PZ0.67, Fig. 7A). Moreover, axon
diameter-frequency analysis showed no significant differ-
ences in the axon diameter distributions within the facial
nerve (Fig. 7B and morphometry data not shown). Thus, in
sharp contrast to spinal motor innervation (Fig. 6A), facial
motor innervation (Fig. 7B) is unaffected in the context of
increased GDNF expression in muscle during development.
While muscle GDNF transgene expression and protein
content have been confirmed in upper hindlimb muscles
from PN3 MyoGDNF mice (Nguyen et al., 1998), GDNF
transgene expression has not been confirmed in facial
muscles from these mice. Apart from the fact that facial
motor neurons are devoid of fusimotor neurons, an
alternative explanation for a lack of facial motor neuron
Fig. 6. Altered distribution of axon types and numbers in lumbar ventral roots from Myo-GDNF mice. (A) In lumbar ventral roots from PN28 wild type mice
(left), a normal number of large diameter myelinated skeletomotor and small diameter myelinated fusimotor axons were noted. In L4 ventral roots from PN28
Myo-GDNF mice (right), there was a large increase in small diameter myelinated axons and a decrease in large diameter myelinated axons. (B) The axon
diameter-frequency analysis defined the two different axon populations that in wild type mice correspond to fusimotor and skeletomotor axons. (C) In Myo-
GDNF mice, there was a 105% increase (P!0.001, Student’s t-test) in the total number of myelinated axons. The small diameter myelinated axons,
indistinguishable from fusimotor axons in wild type mice, were increased by 324% (P!0.0005, Student’s t-test) while large diameter axons were decreased by
37% (P!0.05, Student’s t-test). The small diameter axon increase is consistent with a selective increase in survival and/or proliferation of fusimotor neurons in
the spinal cord. The selective decrease in large diameter myelinated axons is consistent with an increase in the motor unit size within skeletal muscles in
Myo-GDNF mice that has been previously reported. (Magnification barsZ10 mm.)
J. Whitehead et al. / Mechanisms of Development 122 (2005) 27–41 35
response to increased muscle produced GDNF could be that
the MyoGDNF transgene is not expressed in facial muscles.
We examined GDNF expression using semi-quantitative
RT-PCR analysis and found that GDNF was expressed at
comparable levels in facial (posterior digastric) and hindlimb
(gastrocnemius) muscles in PN28 and adult MyoGDNF
mice, and that the expression was markedly elevated relative
to muscle from wild type mice (Fig. 7C). Thus, despite the
fact that facial motor neurons are exposed to elevated levels
of GDNF in MyoGDNF mice, they dramatically differ in
their response to GDNF relative to spinal motor neurons. The
results are consistent with the hypothesis that fusimotor
neurons which are present in the spinal cord, but not in the
facial motor nucleus, are preferentially rescued from
programmed cell death by GDNF during development.
3. Discussion
Previous attempts to characterize GDNF-dependent
motor neurons have met with little success, largely because
molecular markers that distinguish functional groups of
neurons are limited in number (Garces et al., 2000; Homma
et al., 2003; Oppenheim et al., 2000). GDNF-dependent
Fig. 7. Unlike axons in spinal ventral roots, facial motor axons were not affected by over expression of GDNF in muscle in Myo-GDNF mice. Facial muscles do
not contain muscle spindles and hence facial motor neurons are devoid of fusimotor neurons. (A) Neither the number of facial motor neurons (left) nor the
number of facial motor axons (right) was altered in adult Myo-GDNF mice relative to wild type littermate mice. (B) Unlike the axon changes that occurred in
lumbar ventral roots, there were no obvious differences between wild type and Myo-GDNF facial motor axons. (C) Semi-quantitative PCR analysis
demonstrated equivalent levels of GDNF expression in facial (posterior digastric; lanes 3 and 5) and hindlimb (gastrocnemius; lanes 4 and 6) muscles in PN28
and adult Myo-GDNF mice. GDNF was expressed at far lower levels in both posterior digastric (lane 1) and gastrocnemius (lane 2) muscles from wild type
muscles. The cycle number was reduced to maintain linear amplification of highly expressed GDNF in transgenic muscle samples (lanes 3–6) thereby reducing
the sensitivity of the assay to a level that was unable to detect the low abundance endogenous GDNF expression in muscle from PN28 wild type mice (lanes 1
and 2). PCR reactions performed with no template added (water only) showed no amplification of GDNF or GAPDH (lane C). (Magnification barsZ5 mm.)
J. Whitehead et al. / Mechanisms of Development 122 (2005) 27–4136
motor neurons appear to express GFRa1, c-ret and isl-1
which is consistent with the fact that they express the
components of the preferred GDNF signaling receptor
complex (GFRa1 and c-ret) and have a motor neuron
phenotype (isl-1) (Oppenheim et al., 2000). It was not
possible to directly characterize GDNF-dependent motor
neurons further so we explored whether they were
functionally related by examining GDNF expression in
muscle. We found GDNF expression localized to intrafusal
muscle fibers shortly after the induction of spindle
morphogenesis at E15.5–E16.5 and low levels of GDNF
expression in relatively few extrafusal muscle fibers.
Fusimotor neurons were dependent upon intact spindles
since Egr3-deficient mice with defective spindle morpho-
genesis lacked fusimotor innervation. Taken together, these
results indicate that fusimotor neurons depend upon trophic
signals produced by spindles and that GDNF is a candidate
trophic factor required for their survival during develop-
ment. Since it was not possible to study fusimotor neurons
directly in GDNF-deficient mice without fusimotor neuron
specific markers, we used axon morphometry to distinguish
small diameter fusimotor axons from large diameter
skeletomotor axons in ventral spinal roots and motor nerves.
Fusimotor innervation to muscle spindles was decreased by
46% in GDNF haploinsufficient mice which closely agreed
with the predicted 50% that would be lost if fusimotor
neurons were dependent upon GDNF produced in limiting
amount by target muscle fibers. Moreover, the morpho-
metric analysis demonstrated that fusimotor axons rep-
resented 34% of the L4 motor axons which was similar to
the 22–30% lumbar motor neuron loss previously reported
in newborn GDNF-deficient mice (Oppenheim et al., 2000;
Sanchez et al., 1996). From these data, the calculated
average net loss of motor axons would be 16% (100%!0.34!0.46) in GDNF/lacz heterozygous mice, which
compared within 5% to the measured total axon loss of
11% (P!0.05; Student’s t-test, Fig. 3C). Accordingly, these
results would predict that 32% (16%!2) of the motor
axons, attributed entirely to fusimotor axons, would be lost
if it were possible to study GDNF/lacz-deficient mice using
these methods. These results also closely corresponded to
the amount of motor neuron loss previously reported in
GDNF-deficient mice. That fusimotor neurons appear to
depend upon target derived GDNF from intrafusal muscle
fibers is similar to neurotrophin dependent sensory inner-
vation of intrafusal muscle fibers. Intrafusal muscle fibers
J. Whitehead et al. / Mechanisms of Development 122 (2005) 27–41 37
receive both sensory (group Ia and II afferent) and motor
(fusimotor efferent) innervation and they express both
Neurotrophin-3 (NT3) and GDNF. Mice lacking either NT3
or TrkC, the preferred tyrosine kinase signaling receptor for
NT3, also lack group Ia sensory neurons (Ernfors et al.,
1994; Klein et al., 1994; Kucera et al., 1995) and fusimotor
neurons. However, in NT-3 deficient mice, fusimotor
neuron loss is dependent upon intact Ia-afferents and
spindles rather than trophic activity from NT-3 (Ringstedt
et al., 1998). Thus GDNF and NT3 appear to have analogous
but distinct roles in establishing and/or maintaining motor
and sensory innervation to intrafusal muscle fibers.
It was surprising that there was no skeletomotor axon
loss in GDNF haploinsufficient mice since it has been
widely assumed that skeletomotor neurons are lost in
GDNF-deficient mice. This concept has been largely
promulgated by evidence that GDNF has potent survival
activity for injured skeletomotor neurons and in dissociated
motor neuron cultures (Henderson et al., 1994; Matheson
et al., 1997). However, GDNF application to injured motor
neurons, to motor neurons cultured in vitro, or by systemic
administration or abnormal tissue expression all reflect
abnormal cellular contexts that may not recapitulate the role
of GDNF during normal development. In this study, we
sought to identify the motor neuron population that depends
upon GDNF for survival in an anatomically relevant
developmental context. Skeletomotor neurons do not appear
to require GDNF for survival during development, but since
our conclusions are based upon results obtained from
haploinsufficient mice, we cannot rule out the possibility
that other motor neuron survival factors produced by muscle
may have masked the affects of decreased GDNF available
to skeletomotor neurons. By contrast, the results clearly
demonstrate that fusimotor neurons depend upon GDNF
since unequivocal fusimotor axon loss was associated with
GDNF haploinsufficiency. Moreover, this interpretation was
further supported by GDNF expression which is primarily
restricted to intrafusal muscle fibers, the developmental
timing of motor neuron loss in GDNF-deficient mice that is
well correlated with the timing of GDNF expression in
intrafusal muscle fibers and the fact that the magnitude of
L4 motor neuron loss in GDNF-deficient mice is similar to
the size of the L4 fusimotor neuron pool. While our detailed
studies were mainly focused on the analysis of L4 motor
neurons, similar results were obtained when we analyzed
axon loss in muscle nerves arising from other lumbar and
sacral roots and when we examined cervical (C4) ventral
roots.
To examine whether GDNF could influence the distri-
bution of skeletomotor and fusimotor neurons, we examined
transgenic mice that express GDNF in all muscle fibers
(MyoGDNF) and found a marked increase in small
myelinated axons and a decrease in large myelinated
axons. The large myelinated (skeletomotor) axon loss in
MyoGDNF mice (e.g. 37% loss in L4 ventral root
corresponding to 63% of the axons remaining to innervate
the same number of extrafusal muscle fibersZ1.6-fold
increase in the average motor unit size) closely compared to
previously published results indicating that skeletomotor
unit size is increased by approximately 1.5-fold in both
MyoGDNF mice and mice treated postnatally with GDNF
(Keller-Peck et al., 2001; Nguyen et al., 1998). However,
the large increase in the number of motor neurons that
appears to correspond to an increase in small diameter axons
is not easily explained (Oppenheim et al., 2000). It is
possible that the increased number of small myelinated
axons corresponded to atrophic skeletomotor axons but in
agreement with what has been previously reported in
MyoGDNF mice, polyinnervation of extrafusal muscle
fibers by separate axons was not observed and thus seems
unlikely to account for a large skeletomotor neuron/axon
increase (Keller-Peck et al., 2001; Nguyen et al., 1998). The
axon increase must have corresponded 1:1 with the increase
in the number of motor neurons since axons numbers in
proximal and distal segments of ventral roots and muscle
nerves showed no evidence of axon branching. Thus, it is
tempting to speculate that GDNF expression in muscle leads
to a selective increase in fusimotor neurons. Indeed,
hyperinnervation of intrafusal muscle fibers was confirmed
in individually examined spindles from several different
muscles indicating that at least some of the small myelinated
axons terminated on intrafusal fibers. It was not possible to
resolve all of the individual axons within the intramuscular
nerve bundles with our imaging technique, but some axons
may have ended blindly and not terminated on any muscle
fibers. Thus, the increased number of small diameter
myelinated axons must have originated from newly formed
neurons or neurons rescued from programmed cell death in
MyoGDNF mice. Despite the fact that the small diameter
myelinated axons were morphologically indistinguishable
from fusimotor axons, until specific molecular markers that
distinguish fusimotor from skeletomotor neurons become
available, it will not be possible to directly study these
motor neuron populations within the spinal cord.
Fusimotor neurons may be similar to skeletomotor
neurons in their response to GDNF and the fact that they
appear to differentially respond to GDNF during develop-
ment may be largely dictated by the fact that fusimotor
neurons derive GDNF from intrafusal muscle fiber targets
that express high levels of GDNF, whereas skeletomotor
neurons that innervate extrafusal muscle fibers do not.
GDNF-dependent motor neurons express GFRa1 and c-ret
whereas many GDNF-independent motor neurons express
additional receptors such as GFRa2 that may allow them to
respond to other trophic factors and not depend strictly upon
GDNF (Mikaels et al., 2000; Oppenheim et al., 2000). A
subset of spinal and cranial neurons are GDNF-dependent
with the exception of facial motor neurons that do not
appear to depend upon GDNF during development (Mikaels
et al., 2000; Moore et al., 1996; Sanchez et al., 1996). One
potential explanation for why facial motor neurons do not
depend upon GDNF during development is that they do not
J. Whitehead et al. / Mechanisms of Development 122 (2005) 27–4138
contain any fusimotor neurons since facial muscles are
devoid of muscle spindles. Accordingly, unlike ventral
spinal roots and muscle nerves, we found no change in the
number of facial motor neurons or axons in MyoGDNF
mice, despite high levels of GDNF expression in facial
muscles. We speculate that this differential responsiveness
to muscle derived GDNF, which is dramatically different for
spinal motor neurons, is due to the fact that there are no
fusimotor neurons in the facial nucleus that require GDNF
during development.
There are many motor neuron trophic factors that have
been described but a single trophic factor that all motor
neurons depend upon has never been identified. Since many
trophic factors have motor neuron survival activity, there
may be a high degree of redundancy such that the function of
one trophic factor could compensate for loss of another.
Current data would seem to support some aspects of this
model since many motor neurons express receptors for more
than a single trophic factor (Garces et al., 2000; Homma et al.,
2003; Oppenheim et al., 2000). In part, this may explain why
administration of a single trophic factor to motor neurons in
vitro or after injury may rescue comparably more neurons
than are lost in vivo in the trophic factor deficient mice. For
example, c-ret, GFRa1 (the preferred GDNF co-receptor)
and GFRa2 (the preferred Neurturin co-receptor) are
expressed in most facial motor neurons (Mikaels et al.,
2000). Facial motor neurons do not require GDNF in vivo
suggesting that Neurturin or another neurotrophin may
compensate for loss of GDNF. Thus, while facial motor
neurons are rescued from programmed cell death by GDNF
administered after injury or when GDNF is supplied to
dissociated neurons in vitro (Henderson et al., 1994;
Matheson et al., 1997; Oppenheim et al., 1995; Yan et al.,
1995), the motor neurons do not depend upon GDNF because
other survival signaling mechanisms may be utilized.
Accordingly, it is possible that fusimotor neurons express
c-ret and GFRa1, but not other receptors for potential trophic
factors expressed by intrafusal fibers, making them strictly
dependent upon GDNF for survival during development.
While trophic factor redundancy may be one mechanism for
ensuring motor neuron survival, motor neurons that require
single trophic factors, or a discreet combination of trophic
factors, may have a specific relationship to the muscle targets
that are innervated and the functions that the neurons
subserve. Thus, the myriad motor neuron trophic factors
that have been described may reflect a unique relationship
between particular motor neurons and specific muscle fiber
types or functionally related muscle groups that are not
currently appreciated. For example, there is precedence for
this organizational scheme in different classes of sensory
neurons that mediate proprioception and nociception.
Proprioceptive and nociceptive neurons in the dorsal root
ganglion require NT3 and nerve growth factor (NGF),
respectively, from their targets for survival, which closely
relates to the sensory neuron subtype and function. Likewise,
fusimotor neurons may derive GDNF from intrafusal muscle
fibers that participate in maintaining or establishing their
interaction with intrafusal muscle fibers. As more consider-
ation is given to exploring the use of GDNF and other
potential motor neuron trophic factors for treating motor
neuron diseases, it will be essential to thoroughly character-
ize the motor neuron populations upon which these factors
act. Our data indicate that GDNF, which has a transient
trophic influence on skeletomotor neurons and sustained
trophic affects on fusimotor neurons, may not be the best
choice for treatment of motor neuronopathies, such as ALS,
that primarily affect skeletomotor neurons (Kawamura et al.,
1981; Tsukagoshi et al., 1979).
4. Methods
Animals. Egr3-deficient mice were generated as pre-
viously described(Tourtellotte and Milbrandt, 1998) and
genotyped by PCR using the following primers: Egr3sense:
GTCAACCCACCCCCTATTACCCCA, Egr3antisense: ATG
TGAGTGGTGAGGTGGTCGCT and Neoantisense: ATA-
TTGGCTGCAGGGTCGCTCGG. GDNF/lacz-deficient
mice were generated as previously described (Moore et al.,
1996) and genotyped by PCR with the following primers:
GDNFsense: CAGCGCTTCCTCGAAGAGAGAGGAA-
TCGGC, GDNFantisense: CATGCCTGGCCTACCTTGTCA
and Laczantisense: GATGGGCGCA TCGTAACCGTG-
CATCT. Myo-GDNF transgenic mice were generated as
previously described (Nguyen et al., 1998) and genotyped by
PCR with the following primers: hGHsense: GCCTA-
TATCCCAAAGGAACAGAAG and hGHantisense:
CCTTCCTCTAGGTCCTTTAGGAGG. For all PCR geno-
typing, the reaction conditions were: 95 8C!3 0; (95 8C!30 00; 60 8C!30 00; 72 8C!1 0)!35 cycles; 72 8C!10 0. The
use of all of the animals was approved by the Animal Care
and Use Committee of Northwestern University.
Histology. Hindlimbs or dissected muscles from newborn
(PN0.5), adolescent (PN28) or adult (PNO60) mice were
obtained after intracardiac aldehyde perfusion (4% parafor-
maldehyde, 0.1 M phosphate buffer (PB), pHZ7.4). Periph-
eral nerves and ventral spinal roots were dissected from PN28
and adult mice after intracardiac perfusion with gluteralde-
hyde/paraformaldehyde fixative (2.5% gluteraldehyde, 2%
paraformaldehyde, 0.1 M cacodylate buffer, pHZ7.2).
Muscles and hindlimbs were cryoprotected in 30% sucro-
se/PB, embedded in OCT and sectioned at 15-mm thickness
on a cryostat. Peripheral nerves and ventral spinal roots were
fixed overnight in gluteraldehyde/paraformaldehyde fixa-
tive, washed in cacodylate buffer, osmicated, dehydrated and
embedded in Epon 812 (EMS). 1-mm thick resin embedded
tissue sections were stained with Toluidine Blue.
Immunohistochemistry. Immunohistochemistry was per-
formed on either frozen tissue sections from hindlimbs and
muscles or on whole unsectioned muscles (whole mount
preparation). Tissue sections were incubated with 3%
normal serum/0.1% Triton X-100 in PBS (blocking buffer)
J. Whitehead et al. / Mechanisms of Development 122 (2005) 27–41 39
for 1 h at room temperature. The sections were incubated in
primary antibody diluted in blocking buffer overnight at
4 8C. The sections were washed in PBS/0.1% Triton X-100
and incubated with either an appropriate fluorescent
secondary antibodies (Cy3, Cy5, Jackson Immunoresearch
or Alexa-488, Molecular Probes) or a horseradish peroxi-
dase (HRP) conjugated secondary antibody for 1 h at room
temperature. When HRP-conjugated secondary antibodies
were used, the immunocomplexes were visualized using
DAB histochemistry. The whole mount immunohistochem-
istry was essentially the same except that the sections were
washed for a longer time between antibody incubations
(3–4 h) and the detergent in the blocking and antibody
diluents was increased (0.4% Triton X-100). The following
reagents were used: rabbit anti-Egr3 antibody (SC-191,
Santa Cruz Biotechnology, 1:1000), mouse anti-neurofila-
ment antibody (SMI-312, Sternberger Monoclonals,
1:1000), mouse anti-slow developmental myosin heavy
chain (S46, Developmental Studies Hybroma Bank,
University of Iowa, 1:1000) and Cy3- conjugated a-bungar-
bungarotoxin (Molecular Probes, 1:200). Fluorescent
images were captured with a Nikon E600 microscope
equipped with a Spot digital camera (Research Diagnostics).
Spindles within whole mount muscle preparations were
visualized with a Zeiss LS510 confocal microscope using a
100! oil immersion lens. Individual microscopic fields
were assembled into large field montages using Photoshop
(Adobe) to generate high-resolution images of spindles and
their innervating axons.
b-Galactosidase (lacz) histochemistry. Lacz histochem-
istry on 10-mm aldehyde-fixed frozen muscle sections and
newborn intact mice was performed as previously described
(Hogan, 1994).
Axon morphometry. Axon morphometry of ventral spinal
roots and peripheral nerves was performed as previously
described (Tourtellotte and Milbrandt, 1998). Diameter–
frequency measurements of individual axons and total axon
numbers were determined for ventral roots and peripheral
nerves. To examine whether axons branched within the
roots or nerves the number of axons and axon diameter-
frequency from proximal and distal portions of many nerves
or ventral roots were compared.
Analysis of the proprioceptive branch of the soleus
muscle nerve. The proprioceptive branch of the soleus nerve
was analyzed as described previously (Tourtellotte et al.,
2001). In the mouse, the soleus muscle nerve divides into a
large and a small branch approximately 200–500 mm
proximal to its entry into the muscle (Vult von Steyern et
al., 1999). The small proprioceptive branch supplies sensory
(Ia, Ib, II) and fusimotor axons to 2–4 spindles and three
Golgi tendon organs (GTO) located in the proximal part of
the muscle. There are no skeletomotor axons in this nerve
branch. The composition of this purely proprioceptive
branch of the soleus muscle nerve was determined by
retrograde tracing using serial transverse sections of resin
embedded muscle to identify the constituent axons from
their spindle or GTO terminations. Large caliber axons
terminating in the equatorial region of spindles were group
Ia afferents and axons terminating in the juxtaequatorial
regions were group II afferents. Axons innervating GTOs
were identified as Ib afferents. Axons other than Ia, Ib and II
(‘large diameter’ axons) represented a mixed group
comprising fusimotor efferents and group III non-proprio-
ceptive afferents (‘small diameter’ axons). The total number
of small diameter axons (primarily fusimotor efferents) in
the nerve branch was determined and compared between
wild type, Myo-GDNF and GDNF/lacz heterozygous mice.
Analysis of fusimotor innervation of spindles. Serial
sections were examined from transversely cut resin
embedded muscles. Fusimotor efferents were identified as
small diameter myelinated axons within spindles that
terminated on intrafusal muscle fibers in the polar region
of spindles. Individual fusimotor axons were traced back in
the serial sections to their entry point at the equatorial region
of the spindle and counted in individual spindles from wild
type and Myo-GDNF mice. A total of 3–10 spindles were
serially reconstructed for each genotype and for several
different muscles.
Quantification of facial motor neurons and facial motor
axons. Mice were perfused through the heart with 4%
paraformaldehyde/0.1 M phosphate buffer (pHZ7.4). The
brainstems from PN28 wild type (NZ4, transgene negative)
and Myo-GDNF (NZ4, transgene positive) mice were
dissected, processed in paraffin and 8-mm serial sections
were cut in the transverse plane. The tissue sections were
stained with Cresyl Violet and motor neurons within the
discreet facial motor nucleus were counted. Starting with
the first section containing facial motor neurons, every third
section was counted throughout the rostral-caudal extent of
the nucleus. To avoid counting neurons more than once,
they were scored only if they contained nuclear profiles with
an intact nucleolus. For facial motor nerve axon counts and
morphometry, PN28 wild type (NZ4) and Myo-GDNF
(NZ4) mice were perfused with 2.5% gluteraldehyde, 2%
paraformaldehyde, 0.1 M cacodylate buffer (pHZ7.2) and
the facial nerve was dissected as it existed the skull. This
portion of the facial nerve represented the majority of axons
projecting from the facial motor nucleus but did not
represent axons that innervated posterior auricular muscu-
lature and the stapedius muscle. Accordingly, the number of
motor axons counted in this portion of the facial nerve was
always less than the number of motor neurons counted in the
brainstem.
Semi-quantitative RT-PCR. The posterior digastric
muscle (a representative facial muscle) and gastrocnemius
muscle (a representative upper hindlimb muscle) from adult
and PN28 wild type and Myo-GDNF mice were dissected
and snap frozen in liquid nitrogen. The tissues were lysed in
Trizol (Invitrogen) and total RNA was extracted. For each
muscle and time point examined, total RNA was pooled
from three animals. Reverse transcription using Superscript
II reverse transcriptase (Invitrogen) was performed
J. Whitehead et al. / Mechanisms of Development 122 (2005) 27–4140
according to the manufacturer’s protocol. The cDNA was
subjected to PCR using gene specific primers for GDNF
(GDNFsense: AGCGCTTCCTCGAAGAGAGAGGAA-
TCGGC, GDNFantisense: CATGCCTGGCCTACCTTGTCA
and glyceraldehydes-3-phosphate dehydrogenase (GAPDH;
GAPDHsense: GCTTTCCAGAGGGGCCATCCACAG,
GAPDHantisense: GCTTTCCAGAGGGGCCATCCACAG).
The PCR reaction was cycled in the linear range and the
products probed by Southern blotting with cDNA probes for
GDNF and GAPDH.
Statistical analysis. For all quantitative analyses, at least
three animals for each genotype and tissue were analyzed.
For many nerves analyzed by morphometry, tissue sections
from proximal and distal regions of the nerve were
compared to determine whether there was significant axon
branching within the nerve. Statistical significance was
determined using Student’s t-test with significance set at
P!0.05.
Acknowledgements
This work was supported by grants to W.G.T. from the
VA Medical Research Service (Merit Review) and NIH
(NS40748). J.K. received funding from the Veterans
Administration and the National Science Foundation. We
thank Dr. T. Jessell, Columbia University, NY for providing
the PEA3 antibody, Dr. A. Parsadanian, Washington
University, St. Louis, for providing the GDNF cDNA,
Dr. H. Phillips, Genentec, San Francisco, CA for providing
the GDNF/lacz knockout mice and Dr. W. Snider, UNC,
Chapel Hill, NC for providing the Myo-GDNF transgenic
mice.
References
Cacalano, G., Farinas, I., Wang, L.C., Hagler, K., Forgie, A., Moore, M.,
et al., 1998. GFRalpha1 is an essential receptor component for GDNF in
the developing nervous system and kidney. Neuron 21, 53–62.
Cheema, S.S., Richards, L.J., Murphy, M., Bartlett, P.F., 1994. Leukaemia
inhibitory factor rescues motoneurones from axotomy-induced cell
death. Neuroreport 5, 989–992.
Ernfors, P., Lee, K.F., Kucera, J., Jaenisch, R., 1994. Lack of neurotrophin-
3 leads to deficiencies in the peripheral nervous system and loss of limb
proprioceptive afferents. Cell 77, 503–512.
Garces, A., Haase, G., Airaksinen, M.S., Livet, J., Filippi, P.,
deLapeyriere, O., 2000. GFRalpha 1 is required for development of
distinct subpopulations of motoneuron. J. Neurosci. 20, 4992–5000.
Grieshammer, U., Lewandoski, M., Prevette, D., Oppenheim, R.W.,
Martin, G.R., 1998. Muscle-specific cell ablation conditional upon
Cre-mediated DNA recombination in transgenic mice leads to
massive spinal and cranial motoneuron loss. Dev. Biol. 197,
234–247.
Haase, G., Dessaud, E., Garces, A., de Bovis, B., Birling, M., Filippi, P.,
et al., 2002. GDNF acts through PEA3 to regulate cell body positioning
and muscle innervation of specific motor neuron pools. Neuron 35,
893–905.
Helmbacher, F., Dessaud, E., Arber, S., deLapeyriere, O., Henderson, C.E.,
Klein, R., Maina, F., 2003. Met signaling is required for recruitment of
motor neurons to PEA3-positive motor pools. Neuron 39, 767–777.
Henderson, C.E., Phillips, H.S., Pollock, R.A., Davies, A.M., Lemeulle, C.,
Armanini, M., et al., 1994. GDNF: a potent survival factor for
motoneurons present in peripheral nerve and muscle. Science 266,
1062–1064.
Hogan, B., 1994. Manipulating the Mouse Embryo: A Laboratory Manual.
Cold Spring Harbor Laboratory Press, Plainview, NY pp. xvii, 497.
Homma, S., Yaginuma, H., Vinsant, S., Seino, M., Kawata, M., Gould, T.,
et al., 2003. Differential expression of the GDNF family receptors RET
and GFRalpha1. 2, and 4 in subsets of motoneurons: a relationship
between motoneuron birthdate and receptor expression. J. Comp.
Neurol. 456, 245–259.
Kablar, B., Rudnicki, M.A., 1999. Development in the absence of skeletal
muscle results in the sequential ablation of motor neurons from the
spinal cord to the brain. Dev. Biol. 208, 93–109.
Kaspar, B.K., Llado, J., Sherkat, N., Rothstein, J.D., Gage, F.H., 2003.
Retrograde viral delivery of IGF-1 prolongs survival in a mouse ALS
model. Science 301, 839–842.
Kawamura, Y., Okazaki, H., O’Brien, P.C., Dych, P.J., 1977. Lumbar
motoneurons of man: I. Number and diameter histogram of alpha and
gamma axons of ventral root. J. Neuropathol. Exp. Neurol. 36, 853–860.
Kawamura, Y., Dyck, P.J., Shimono, M., Okazaki, H., Tateishi, J., Doi, H.,
1981. Morphometric comparison of the vulnerability of peripheral
motor and sensory neurons in amyotrophic lateral sclerosis.
J. Neuropathol. Exp. Neurol. 40, 667–675.
Keller-Peck, C.R., Feng, G., Sanes, J.R., Yan, Q., Lichtman, J.W.,
Snider, W.D., 2001. Glial cell line-derived neurotrophic factor
administration in postnatal life results in motor unit enlargement and
continuous synaptic remodeling at the neuromuscular junction.
J. Neurosci. 21, 6136–6146.
Klein, R., Silos-Santiago, I., Smeyne, R.J., Lira, S.A., Brambilla, R.,
Bryant, S., et al., 1994. Disruption of the neurotrophin-3 receptor gene
trkC eliminates la muscle afferents and results in abnormal movements.
Nature 368, 249–251.
Koliatsos, V.E., Cayouette, M.H., Berkemeier, L.R., Clatterbuck, R.E.,
Price, D.L., Rosenthal, A., 1994. Neurotrophin 4/5 is a trophic factor
for mammalian facial motor neurons. Proc. Natl Acad. Sci. USA 91,
3304–3308.
Kotzbauer, P.T., Lampe, P.A., Heuckeroth, R.O., Golden, J.P.,
Creedon, D.J., Johnson Jr., E.M., Milbrandt, J., 1996. Neurturin,
a relative of glial-cell-line-derived neurotrophic factor. Nature 384,
467–470.
Kucera, J., Walro, J.M., 1995. Origin of intrafusal fibers from a subset of
primary myotubes in the rat. Anat. Embryol. (Berlin) 192, 149–158.
Kucera, J., Fan, G., Jaenisch, R., Linnarsson, S., Ernfors, P., 1995.
Dependence of developing group Ia afferents on neurotrophin-3.
J. Comp. Neurol. 363, 307–320.
Leitner, M.L., Molliver, D.C., Osborne, P.A., Vejsada, R., Golden, J.P.,
Lampe, P.A., et al., 1999. Analysis of the retrograde transport of glial
cell line-derived neurotrophic factor (GDNF), neurturin, and persephin
suggests that in vivo signaling for the GDNF family is GFRalpha
coreceptor-specific. J. Neurosci. 19, 9322–9331.
Lewis, M.E., Neff, N.T., Contreras, P.C., Stong, D.B., Oppenheim, R.W.,
Grebow, P.E., Vaught, J.L., 1993. Insulin-like growth factor-I: potential
for treatment of motor neuronal disorders. Exp. Neurol. 124, 73–88.
Li, L., Wu, W., Lin, L.F., Lei, M., Oppenheim, R.W., Houenou, L.J., 1995.
Rescue of adult mouse motoneurons from injury-induced cell death by
glial cell line-derived neurotrophic factor. Proc. Natl Acad. Sci. USA
92, 9771–9775.
Matheson, C.R., Wang, J., Collins, F.D., Yan, Q., 1997. Long-term survival
effects of GDNF on neonatal rat facial motoneurons after axotomy.
Neuroreport 8, 1739–1742.
McHanwell, S., Biscoe, T.J., 1981. The sizes of motoneurons supplying
hindlimb muscles in the mouse. Proc. R. Soc. Lond. B Biol. Sci. 213,
201–216.
J. Whitehead et al. / Mechanisms of Development 122 (2005) 27–41 41
Mikaels, A., Livet, J., Westphal, H., De Lapeyriere, O., Ernfors, P., 2000. A
dynamic regulation of GDNF-family receptors correlates with a specific
trophic dependency of cranial motor neuron subpopulations during
development. Eur. J. Neurosci. 12, 446–456.
Milbrandt, J., de Sauvage, F.J., Fahrner, T.J., Baloh, R.H., Leitner, M.L.,
Tansey, M.G., et al., 1998. Persephin, a novel neurotrophic factor
related to GDNF and neurturin. Neuron 20, 245–253.
Mohajeri, M.H., Figlewicz, D.A., Bohn, M.C., 1999. Intramuscular grafts
of myoblasts genetically modified to secrete glial cell line-derived
neurotrophic factor prevent motoneuron loss and disease progression in
a mouse model of familial amyotrophic lateral sclerosis. Hum. Gene
Ther. 10, 1853–1866.
Moore, M.W., Klein, R.D., Farinas, I., Sauer, H., Armanini, M.,
Phillips, H., et al., 1996. Renal and neuronal abnormalities in mice
lacking GDNF. Nature 382, 76–79.
Neff, N.T., Prevette, D., Houenou, L.J., Lewis, M.E., Glicksman, M.A.,
Yin, Q.W., Oppenheim, R.W., 1993. Insulin-like growth factors:
putative muscle-derived trophic agents that promote motoneuron
survival. J. Neurobiol. 24, 1578–1588.
Nguyen, Q.T., Parsadanian, A.S., Snider, W.D., Lichtman, J.W., 1998.
Hyperinnervation of neuromuscular junctions caused by GDNF over-
expression in muscle. Science 279, 1725–1729.
Oppenheim, R.W., Yin, Q.W., Prevette, D., Yan, Q., 1992. Brain-derived
neurotrophic factor rescues developing avian motoneurons from cell
death. Nature 360, 755–757.
Oppenheim, R.W., Houenou, L.J., Johnson, J.E., Lin, L.F., Li, L., Lo, A.C.,
et al., 1995. Developing motor neurons rescued from programmed and
axotomy-induced cell death by GDNF. Nature 373, 344–346.
Oppenheim, R.W., Houenou, L.J., Parsadanian, A.S., Prevette, D.,
Snider, W.D., Shen, L., 2000. Glial cell line-derived neurotrophic
factor and developing mammalian motoneurons: regulation of pro-
grammed cell death among motoneuron subtypes. J. Neurosci. 20,
5001–5011.
Pennica, D., Arce, V., Swanson, T.A., Vejsada, R., Pollock, R.A.,
Armanini, M., Dudley, K., et al., 1996. Cardiotrophin-1, a cytokine
present in embryonic muscle, supports long-term survival of spinal
motoneurons. Neuron 17, 63–74.
Pichel, J.G., Shen, L., Sheng, H.Z., Granholm, A.C., Drago, J.,
Grinberg, A., et al., 1996. Defects in enteric innervation and kidney
development in mice lacking GDNF. Nature 382, 73–76.
Ramer, M.S., Priestley, J.V., McMahon, S.B., 2000. Functional regener-
ation of sensory axons into the adult spinal cord. Nature 403, 312–316.
Ringstedt, T., Copray, S., Walro, J., Kucera, J., 1998. Development
of fusimotor innervation correlates with group Ia afferents but
is independent of neurotrophin-3. Brain Res. Dev. Brain Res. 111,
295–300 bin/cas/tree/store/bresd/cas_sub/browse/browse.cgi?yearZ1998&volumeZ111&issueZ2&aidZ60721.
Sanchez, M.P., Silos-Santiago, I., Frisen, J., He, B., Lira, S.A.,
Barbacid, M., 1996. Renal agenesis and the absence of enteric neurons
in mice lacking GDNF. Nature 382, 70–73.
Sendtner, M., Holtmann, B., Kolbeck, R., Thoenen, H., Barde, Y.A., 1992a.
Brain-derived neurotrophic factor prevents the death of motoneurons in
newborn rats after nerve section. Nature 360, 757–759.
Sendtner, M., Schmalbruch, H., Stockli, K.A., Carroll, P.,
Kreutzberg, G.W., Thoenen, H., 1992b. Ciliary neurotrophic factor
prevents degeneration of motor neurons in mouse mutant progressive
motor neuronopathy. Nature 358, 502–504.
Sun, W., Gould, T.W., Vinsant, S., Prevette, D., Oppenheim, R.W., 2003.
Neuromuscular development after the prevention of naturally occurring
neuronal death by Bax deletion. J. Neurosci. 23, 7298–7310.
Tomac, A., Widenfalk, J., Lin, L.F., Kohno, T., Ebendal, T., Hoffer, B.J.,
Olson, L., 1995. Retrograde axonal transport of glial cell line-derived
neurotrophic factor in the adult nigrostriatal system suggests a trophic
role in the adult. Proc. Natl Acad. Sci. USA 92, 8274–8278.
Tourtellotte, W.G., Milbrandt, J., 1998. Sensory ataxia and muscle spindle
agenesis in mice lacking the transcription factor Egr3. Nat. Genet. 20,
87–91.
Tourtellotte, W.G., Keller-Peck, C., Milbrandt, J., Kucera, J., 2001. The
transcription factor Egr3 modulates sensory axon-myotube interactions
during muscle spindle morphogenesis. Dev. Biol. 232, 388–399.
Trupp, M., Arenas, E., Fainzilber, M., Nilsson, A.S., Sieber, B.A.,
Grigoriou, M., et al., 1996. Functional receptor for GDNF encoded
by the c-ret proto-oncogene. Nature 381, 785–789.
Tsukagoshi, H., Yanagisawa, N., Oguchi, K., Nagashima, K.,
Murakami, T., 1979. Morphometric quantification of the cervical
limb motor cells in controls and in amyotrophic lateral sclerosis.
J. Neurol. Sci. 41, 287–297.
Vejsada, R., Tseng, J.L., Lindsay, R.M., Acheson, A., Aebischer, P.,
Kato, A.C., 1998. Synergistic but transient rescue effects of BDNF
and GDNF on axotomized neonatal motoneurons. Neuroscience 84,
129–139.
Vult von Steyern, F., Martinov, F., Rabben, V., Nja, I., de Lapeyriere, O.,
Lomo, T., 1999. The homeodomain transcription factors Islet 1 and
HB9 are expressed in adult alpha and gamma motoneurons identified by
selective retrograde tracing. Eur. J. Neurosci. 11, 2093–2102.
Wright, D.E., Snider, W.D., 1996. Focal expression of glial cell line-
derived neurotrophic factor in developing mouse limb bud. Cell Tissue
Res. 286, 209–217.
Yamamoto, Y., Livet, J., Pollock, R.A., Garces, A., Arce, V.,
deLapeyriere, O., Henderson, C.E., 1997. Hepatocyte growth factor
(HGF/SF) is a muscle-derived survival factor for a subpopulation of
embryonic motoneurons. Development 124, 2903–2913.
Yan, Q., Elliott, J., Snider, W.D., 1992. Brain-derived neurotrophic factor
rescues spinal motor neurons from axotomy-induced cell death. Nature
360, 753–755.
Yan, Q., Matheson, C., Lopez, O.T., 1995. In vivo neurotrophic effects
of GDNF on neonatal and adult facial motor neurons. Nature 373,
341–344.
Zurn, A.D., Baetge, E.E., Hammang, J.P., Tan, S.A., Aebischer, P., 1994.
Glial cell line-derived neurotrophic factor (GDNF), a new neurotrophic
factor for motoneurones. Neuroreport 6, 113–118.