Abstract
The leukocyte common antigen-related (LAR) subfamily of receptor protein tyrosine phosphatases (RPTPs), LAR, RPTP-σ, and RPTP-δ, regulate neuroendocrine development, axonal regeneration, and hippocampal long-term potentiation in mammals. In Drosophila, RPTPs are required for appropriate axon targeting during embryonic development. In contrast, deletion of any one of the three LAR-RPTP family members in mammals does not result in gross axon targeting defects. Both RPTP-σ and RPTP-δ are highly expressed in the developing mammalian nervous system, suggesting they might be functionally redundant. To test this hypothesis, we generated RPTP-σ and RPTP-δ (RPTP-σ/δ) double-mutant mice. Although embryonic day 18.5 RPTP-σ and RPTP-δ single-mutant embryos were viable, RPTP-σ/δ double mutants were paralyzed, were never observed to draw a breath, and died shortly after cesarean section. RPTP-σ/δ double mutants exhibit severe muscle dysgenesis and severe loss of motoneurons in the spinal cord. Detailed analysis of the projections of phrenic nerves in RPTP-σ/δ double mutants indicated that these motoneuron axons emerge normally from the cervical spinal cord, but stall on reaching the diaphragm. Our results demonstrate that RPTP-σ and RPTP-δ complement each other functionally during mammalian development, and reveal an essential contribution of RPTP-σ and RPTP-δ to appropriate motoneuron axon targeting during mammalian axonogenesis.
- protein tyrosine phosphatase
- tyrosine phosphorylation
- axon targeting
- axon guidance
- motoneuron (motor neuron)
- phrenic
- σ
- δ
Introduction
Receptor protein tyrosine phosphatases (RPTPs) are key regulators of axon guidance in Drosophila (Johnson and Van Vactor, 2003). For example, loss-of-function mutants of leukocyte common antigen-related (LAR) subfamily RPTPs [Drosophila LAR (DLAR)] exhibit defects in motor axon guidance, midline axon crossing, and photoreceptor axon guidance (Desai et al., 1996; Krueger et al., 1996; Sun et al., 2000; Clandinin et al., 2001; Maurel-Zaffran et al., 2001; Desai and Purdy, 2003). There are three vertebrate homologs of DLAR: PTPRS (RPTP-σ), PTPRD (RPTP-δ), and PTPRF (LAR) (Johnson and Holt, 2000; Johnson and Van Vactor, 2003).
Type IIa RPTPs are characterized by variable numbers of extracellular Ig-like and type-III fibronectin (FNIII) repeats and two intracellular tandem phosphatase domains. The extracellular domains of this family resemble cell adhesion molecules (CAMs), suggesting that they play roles in cell–cell and cell–matrix interactions (Brady-Kalnay and Tonks, 1995; Beltran and Bixby, 2003).
In vertebrates, type IIa RPTPs are present in axons and growth cones (Tian et al., 1991; Desai et al., 1994; Stoker et al., 1995; Sommer et al., 1997; Schaapveld et al., 1998; Thompson et al., 2003). Although RNA interference-mediated knock-down has provided evidence that they contribute to motor axon guidance in chick (Stepanek et al., 2005), a role for LAR subfamily members in axon guidance during mammalian development has not been reported. Limited levels of LAR expression in the CNS are detected during development, and its absence results in minor defects in cholinergic innervation of the hippocampal dentate gyrus (Yeo et al., 1997; Van Lieshout et al., 2001). In contrast, RPTP-σ and RPTP-δ are highly expressed in the developing mammalian nervous system (Yan et al., 1993; Mizuno et al., 1994; Sommer et al., 1997; Schaapveld et al., 1998). RPTP-σ-deficient mice exhibit abnormal development of the pituitary and neurological defects that include spastic movements and abnormal limb flexion (Elchebly et al., 1999; Wallace et al., 1999). Additionally, the absence of RPTP-σ results in enhanced axon regeneration in both the peripheral and central nervous systems (McLean et al., 2002; Thompson et al., 2003; Saphieha et al., 2005). Mutant mice lacking an essential catalytic domain of RPTP-δ exhibit early growth retardation and neonatal mortality. In this case, the high mortality is likely attributable to insufficient food intake caused by motor dysfunction. RPTP-δ-deficient mice that reach adulthood show enhanced hippocampal long-term potentiation and impaired spatial learning ability (Uetani et al., 2000).
The overlapping expression patterns of RPTP-σ and RPTP-δ suggest that the phenotypes observed in single gene knock-outs may underestimate the contribution of type IIa RPTPs to mammalian neural development. Consistent with this, RNA interference knock-down suggested overlapping functions for LAR-RPTPs in cultured rat hippocampal neurons (Dunah et al., 2005). To address the possibility of functional redundancy in vivo, mice deficient for both RPTP-σ and RPTP-δ were generated and characterized. Our findings demonstrate that RPTP-σ and RPTP-δ are essential for the appropriate innervation of muscles by motoneurons during development of the murine peripheral nervous system.
Materials and Methods
Generation of RPTP-σ/δ double-mutant mice.
Mutant mice lacking either RPTP-σ or RPTP-δ have been described previously (Elchebly et al., 1999; Uetani et al., 2000). These animals were backcrossed for at least six generations to C57BL/6J. Double homozygous mutant mice were obtained by intercrossing double heterozygous parents. Noon on the day of the detection of the vaginal plug was considered embryonic day 0.5 (E0.5). To genotype RPTP-σ and RPTP-δ alleles, tissue lysates from mice were used for PCR analysis with allele-specific primers: Sigma-WT-R2, 5′-TAG TCA GAG CCC TCC ACA CCG-3′; Sigma-KO-F1, 5′-CAG CGA CCT CCA ACA CAC AAG-3′; Sigma-WT-F2, 5′-CAT TTC CAG TTC ACG GCA TGG C-3′; Delta-WT-R1, 5′-CTG GAA TTG TCT CAC TTT CCT C-3′; Delta-WT-F1, 5′-CCA GCA GAG GCA CAG AAA CTC-3′; Delta-PGK#5, 5′-GAC TGC CTT GGG AAA AGC GCC TCC-3′.
RPTP-σ wild-type and knock-out alleles are predicted to produce 781 and 1000 bp PCR products, respectively. RPTP-δ wild-type and knock-out alleles are predicted to produce 620 and 380 bp bands, respectively. Amplified bands were the expected size. Mice were kept under specific pathogen-free conditions in an environmentally controlled clean room at the Animal Resource Center of McGill University. All experiments were conducted according to the Canadian Council of Animal Care ethical guidelines for animal experiments.
Animals identified as “control” in this study include littermates with the following genotypes (RPTP-σ+/+/δ+/+, RPTP-σ+/−/δ+/+, RPTP-σ+/+/δ+/−, RPTP-σ+/−/δ+/−) with all presenting identical normal (wild-type) phenotypes. Hence they were tabulated together for statistical analysis.
Histological analysis.
Tissues were dissected and fixed in phosphate buffered 4% paraformaldehyde (PFA), left overnight in the same fixative, and then embedded in paraffin wax according to standard procedure. Serial 6 μm paraffin sections were made and stained with hematoxylin and eosin (H&E) or cresyl violet (Nissl stain). For every fourth section at the forelimb level (E18.5), the thickness of the dorsal epidermis and dermis was measured at three locations using AxioVision, version 4.0, software (Carl Zeiss, Oberkochen, Germany). The thickness of the diaphragm was measured on E16.5 sagittal sections at eye level in the same way. A total of three sections from each animal was measured and the mean value of these measurements was used as a single data value for statistical analysis by Student's t test. Skeletal muscle was visualized in paraffin sections using an antibody against myosin heavy chain (MY32; Sigma, St. Louis, MO; 1:100 dilution; a biotinylated anti-mouse secondary antibody and peroxidase-labeled streptavidin; LSAB kit; DakoCytomation, Carpinteria, CA). The color reaction was developed using 3-3′-diaminobenzidine tetrahydrochloride (DAB) as a chromogen (SigmaFast DAB tablets; Sigma). Histological analyses were performed on E16.5 RPTP-σ/δ double mutants (n = 2) with control littermates (n = 2), and on E18.5 RPTP-σ/δ double mutants (n = 3) with their control littermates (n = 3), RPTP-σ−/−/δ+/+ (n = 2), RPTP-σ+/+/δ−/− (n = 2), RPTP-σ+/−/δ−/− (n = 2), and RPTP-σ−/−/δ+/− (n = 2).
Quantification of motor neurons.
E13 and E18.5 embryos were fixed in 4% PFA overnight. Whole E13 embryos were embedded in OCT compound (TissueTek; Sakura Finetek, Torrance, CA). Cervical spinal cords (C3–C5) were dissected from E18.5 embryos and embedded in paraffin. Ten micrometer serial cryostat and paraffin sections were prepared. The location of C3–C5 at E13 was determined morphologically based on surrounding tissue. Mouse anti-Islet-1/2 antibodies (39.4D5; Developmental Studies Hybridoma Bank, Iowa City, IA; supernatant, 1:50 dilution) and goat anti-choline acetyltransferase (ChAT) antibodies (Chemicon, Temecula, CA; 1:100 dilution) were used to identify motoneurons in the E13 and E18.5 spinal cords, respectively. Sections were visualized using secondary antibodies coupled to Alexa 488 (Invitrogen, San Diego, CA), and photographed using a Zeiss Axioskop2 Plus microscope.
Islet-1/2- and ChAT-immunopositive motoneurons were counted independently on right and left hemiventral columns in every fourth section. A total of 10 hemiventral columns from each animal were counted, and the mean value of these counts was used as an individual data point for statistical analysis by Student's t test. At E13, RPTP-σ/δ double mutants (n = 3) were analyzed with their control (n = 5) littermates. At E18.5, RPTP-σ/δ double mutants (n = 3) were analyzed with their littermates RPTP-σ+/+/δ+/+ (n = 4), RPTP-σ+/+/δ−/− (n = 3), RPTP-σ−/−/δ+/+ (n = 2), RPTP-σ+/−/δ−/− (n = 2), and RPTP-σ−/−/δ+/− (n = 4).
Whole-mount diaphragm staining.
To visualize peripheral nerves, whole E12 embryos and thoracic cages dissected from E13.5 and E15.5 embryos were fixed in methanol/dimethylsulfoxide (DMSO) (4:1) at 4°C overnight. Tissues were then immersed in methanol/DMSO/H2O2 (4:1:1) at room temperature for 5–10 h to block endogenous peroxidase activity, and rehydrated in 50% methanol and PBS. E12 embryos were embedded in OCT compound, and sagittal sections were cut with a cryostat until heart and liver were visible. The remaining E12 embryonic tissue was immersed in PBS to remove excess OCT and then blocked in PBS containing 5% skimmed milk, 0.5% Triton X-100 (PBSMT). The embryos were incubated in 2 ml of PBSMT with the 2H3 monoclonal antibody against neurofilament (ascites; 1:2000 dilution; 4°C overnight; Developmental Studies Hybridoma Bank) (Dodd et al., 1988). After washing with 4 ml of PBSMT for 1 h at room temperature three times, the embryos were incubated with horseradish peroxidase-conjugated anti-mouse IgG (1:10,000 dilution) (Jackson ImmunoResearch, West Grove, PA) in PBSMT overnight at 4°C. They were then washed with 4 ml of PBSMT for 1 h at room temperature six times, and thoracic cages were soaked in DAB substrate solution to develop the color reaction (SigmaFast DAB tablets; Sigma). Stained diaphragms were dissected from E13.5 and E15.5 thoracic cages in PBS, and dehydrated in an ethanol series. After xylene substitution, diaphragms were mounted on slides. Stained E12 embryos were then dehydrated in methanol and substituted in benzyl alcohol (Sigma) and benzyl benzonate (Sigma) (1:2). Stained diaphragms were photographed and branch points on the sternocostal phrenic nerve were counted using AxioVision, version 4.0, software. E12 RPTP-σ/δ double mutants (n = 3) were analyzed with control (n = 4) littermates. E13.5 RPTP-σ/δ double mutants (n = 2) were analyzed with control (n = 2), RPTP-σ−/−/δ+/+ (n = 2), RPTP-σ+/+/δ−/− (n = 3), RPTP-σ−/−/δ+/− (n = 1), and RPTP-σ+/−/δ−/− (n = 2) littermates. E15.5 RPTP-σ/δ double mutants (n = 2) were analyzed with control (n = 3), RPTP-σ−/−/δ+/+ (n = 3), RPTP-σ+/+/δ−/− (n = 2), RPTP-σ−/−/δ+/− (n = 2), and RPTP-σ+/−/δ−/− (n = 3) littermates.
Results
RPTPσ/δ double-mutant mice die at birth
RPTP-σ and RPTP-δ (RPTP-σ/δ) double-mutant mice were generated by breeding adults heterozygous for both RPTP-σ and RPTP-δ (RPTP-σ+/−/δ+/− × RPTP−σ+/−/δ+/−). Pups were genotyped using PCR with gene-specific primers (Fig. 1A). Although each of the single knock-outs is viable at birth, live RPTP-σ/δ double mutants were not found in litters of newborn pups at postnatal day 0 (P0) (n = 217) (Table 1). Furthermore, RPTP-σ−/−/δ+/− and RPTPσ+/−/δ−/− mice had slightly higher mortality rates compared with either single mutant. Within 4 d of birth, 55% of RPTP-σ−/−/δ+/− (n = 18) and 71% of RPTP-σ+/−/δ−/− (n = 14) pups died, whereas 44.4% of RPTP-σ−/−/δ+/+ (n = 18) and 14.3% of RPTP-σ+/+/δ−/− (n = 21) died in the same period. Mice heterozygous for either gene had mortality rates not significantly different from wild-type littermates (RPTP-σ+/+/δ+/+, 8.7%; RPTP-σ+/−/δ+/+, 13.6%; RPTP-σ+/+/δ+/−, 5.9%). These findings suggest that RPTP-σ and RPTP-δ exhibit some functional complementarity during development.
To investigate whether RPTP-σ/δ double-mutant mice died before or after birth, cesarean sections were performed at E18.5. Approximately one-sixteenth (5.48%; n = 73) were RPTP-σ/δ double mutant, slightly less than the expected Mendelian ratio, indicating that the RPTP-σ/δ double mutants are viable to late embryogenesis. While, non-double-mutant E18.5 littermates begin to breathe immediately after cesarean section, RPTP-σ/δ double-mutant embryos were never observed to take a breath and died shortly after cesarean section. Furthermore, RPTP-σ/δ double mutants were never observed to move and display kyphosis (hunchback) and carpoptosis (wrist drop) with 100% penetrance (n = 12) as shown in Figure 1C. Histological analysis revealed that the alveoli of E18.5 RPTP-σ/δ double-mutant mice (n = 3) were tightly compressed (Fig. 2C), consistent with their inability to breathe. We conclude that RPTPσ/δ double mutants die immediately after birth and that death is likely attributable to respiratory failure.
Histological analysis of skin, adipose tissue and skeletal muscle
To examine the effects of the mutations on tissue structure, an extensive histological survey was performed. The skin of RPTP-σ/δ double-mutant embryos was abnormal with a semitransparent appearance that was obvious from E15. Although the thickness of the epidermis was not altered in RPTP-σ/δ double mutants (controls, 42 ± 8 μm, n = 18; RPTP-σ/δ double mutants, 43 ± 9 μm, n = 18), the dermis of RPTP-σ/δ double mutants was thinner (control, 109 ± 21 μm, n = 18; RPTP-σ/δ double mutants, 83 ± 14 μm, n = 18; p < 0.001) and hair follicles were not fully developed at E18.5 in comparison with control non-double-mutant littermates (Fig. 3A–D). Subcutaneous tissues were enlarged, and the compartmentalization of brown adipose tissue was less obvious in RPTP-σ/δ double mutants (Fig. 3E,G, indicated by ∗∗). Furthermore, immunohistochemical analyses using anti-myosin heavy chain antibodies revealed a reduction in the mass of dorsal axial skeletal muscles and limb muscles (Fig. 3E–J). In E18.5 RPTP-σ/δ double-mutant embryos, we observed that skeletal muscle fibers were smaller in diameter, with their nuclei centrally localized along the fiber, suggesting the muscles remained immature with their development halted at an early stage of myotube formation (Fig. 3K,L). These phenotypes, revealed using histological and immunohistochemical staining, were identical in each of the three E18.5 RPTP-σ/δ double-mutant samples analyzed. Although only two E16.5 embryos were analyzed for each genotype, the mean thickness of the diaphragm of these RPTP-σ/δ double-mutant mice was thinner compared with those of wild-type littermates (RPTP-σ+/+/δ+/+, 197 ± 15 μm, n = 27; RPTP-σ/δ double mutants, 110 ± 17 μm, n = 27; p < 0.001) (Fig. 3M–P). Together, these results suggest that the respiratory insufficiency in the double mutants was attributable to skeletal muscle defects, likely caused by the absence of motoneuron innervation, as described below.
Histological differences were not detected in H&E-stained E18.5 heart, liver, kidney, colon, thymus, or spleen of RPTP-σ/δ double mutants (data not shown). In addition, sagittal sections of E18.5 brain were also examined by H&E stain. Although RPTP-σ and RPTP-δ are highly expressed in the developing brain, the gross anatomy of the brain including olfactory bulb, forebrain, hippocampus, cortex, and cerebellum appeared normal in RPTP-σ/δ double mutants (data not shown).
Loss of spinal motoneurons in RPTP-σ/δ double mutants
RPTP-σ/δ double-mutant embryos did not move after cesarean section and appeared paralyzed (Fig. 1C). This phenotype is often associated with motoneuron defects, including aberrant neuromuscular junction formation (Misgeld et al., 2002; Brandon et al., 2003). RPTP-σ and RPTP-δ are both highly expressed by spinal motoneurons during embryogenesis (Sommer et al., 1997; Schaapveld et al., 1998). To explore the possibility that motoneuron development may be disrupted by loss of RPTP-σ and RPTP-δ function, transverse sections of spinal cords were examined by Nissl stain (Fig. 4). At E18.5, motoneurons were readily identified by their large cell bodies, large nucleolus, and enriched cellular Nissl bodies in the spinal cords of wild-type embryos (Fig. 4C).
In contrast, only a few morphologically abnormal motoneurons, exhibiting smaller cytoplasm with Nissl bodies and an indistinct nucleolus, were found in E18.5 RPTP-σ/δ double mutants (Fig. 4D, arrows). No obvious abnormalities were observed in the dorsal root ganglia (DRGs) (Fig. 4E,F) or sympathetic ganglia (SGs) (Fig. 4G,H). Expression of the motoneuron-specific markers, Islet-1/2 and ChAT (Tsuchida et al., 1994; Pfaff et al., 1996; Helmbacher et al., 2000), was then examined on transverse sections of E13 and E18.5 spinal cords at the level of the third (C3) and fifth (C5) cervical vertebrae, respectively (Fig. 5A–D). Although the number of Islet-1/2-immunopositive motoneurons per hemiventral motor column appeared normal in RPTP-σ/δ double mutants at E13 (Fig. 5E), the number of ChAT-immunopositive motoneurons detected in RPTP-σ/δ double mutants at E18.5 was significantly reduced (Fig. 5F, ∗∗p < 0.0001). These findings identify a major motoneuron deficit in RPTP-σ/δ double mutants that occurs late in embryogenesis between E13 and E18.5, during motoneuron axon extension. Additionally, a smaller but significant decrease in the number of ChAT-immunopositive motoneurons was also detected in RPTP-σ−/−/δ+/− embryos compared with wild-type or embryos mutant for either RPTP-σ or RPTP-δ alone (Fig. 5F, ∗p < 0.001). That a single gene dose of RPTP-σ is sufficient for levels of motoneuron survival in RPTP-σ+/−/δ−/− that are not different from control, suggests that RPTP-σ may make a larger essential contribution to motoneuron survival than does RPTP-δ.
Altered innervation of the diaphragm by the phrenic nerve in RPTP-σ/δ double mutants
Motoneurons depend on target-derived trophic support to survive. During late embryogenesis, cell death matches the number of motoneurons with the size of the muscle targets being innervated (Banks and Noakes, 2002). We hypothesized that defects in the ability of motoneurons to reach and innervate muscle targets might contribute to the motoneuron loss detected in RPTP-σ/δ double mutants (Figs. 4, 5). To test this, we examined the projection of the phrenic nerve at various developmental stages using whole-mount immunochemistry and a neurofilament (NF) antibody (2H3). The phrenic nerve contains motor axons that originate in the spinal cord between C3 and C5 and extend to the diaphragm. It then forms three characteristic branches on the surface of the diaphragm: two sternocostal branches and one crural branch (a schematic is included in the supplemental material, available at www.jneurosci.org). In E10.5 RPTP-σ/δ double mutants, the phrenic nerve descends normally from the spinal cord (data not shown) and reaches the diaphragm at E12 (n = 3) (Fig. 6A). However, in E13.5 RPTP-σ/δ double-mutant embryos, the phrenic motor axons extend aberrantly on the surface of the diaphragm (Fig. 6E,F) (n = 2). Although the phrenic nerve appears to reach the appropriate initial location on the diaphragm, the two main sternocostal branches that normally defasciculate from the main nerve appear to have retracted (Fig. 6F, arrowhead), and only a tiny residual projection extends toward the crus (Fig. 6E,F, white arrowheads). These phrenic nerve abnormalities were identical in the two individual E13.5 RPTP-σ/δ double mutants examined. Furthermore, analyses of E15.5 (n = 2) and E18.5 (n = 3) double mutants revealed the complete absence of sternocostal branches on the diaphragm (E15.5) (Fig. 7E,F) (E18.5) (data not shown). Each of the total of seven RPTP-σ/δ double mutants analyzed (E13.5, E15.5, and E18.5) exhibited aberrant phrenic nerve phenotypes, consistent with the conclusion that these phrenic nerve defects are 100% penetrant in the absence of RPTP-σ/δ function. At E13.5 and E15.5, the capillaceous nerves normally defasciculate from the major sternocostal branches (Figs. 6C, 7A,B); however, capillaceous nerves were seldom observed in either the RPTP-σ−/−/δ+/− (data not shown) or RPTP-σ+/−/δ−/− genotypes (Figs. 6D, 7C,D). In contrast, the innervation of the diaphragm in RPTP-σ and RPTP-δ single-mutant littermates appeared normal (Fig. 7G,H). To further substantiate these observations, we performed quantitative analysis on anti-NF-stained diaphragms at E15.5 by counting each nerve branch point. As shown in Figure 7I, the total number of branch points was decreased in RPTP-σ+/−/δ−/− (n = 3) and RPTP-σ−/−/δ+/− (n = 2) embryos compared with non-double mutants, including the RPTP-σ (n = 3) and RPTP-δ (n = 2) single mutants.
Comparison of these findings with Figure 5F indicates that the defect in branching is more sensitive to RPTP loss of function than is motoneuron survival: E18.5 RPTP-σ+/−/δ−/− exhibit no significant difference in motoneuron number compared with control, whereas E18.5 RPTP-σ−/−/δ+/− exhibit a modest, albeit significant, reduction of the number of motoneurons, ∼70% of control (Fig. 5F). In contrast, E18.5 double knock-out RPTP-σ−/−/δ−/− exhibit severely reduced numbers of motoneurons, ∼20% of control, and no branching from the sternocostal phrenic nerve. Although these findings suggest a relationship between innervation and survival, we have not ruled out the possibility that these two phenotypes may be dissociable. The intermediate nerve branch phenotype found in RPTP-σ−/−/δ+/− and RPTP-σ+/−/δ−/− embryos provides evidence that RPTP-σ and RPTP-δ have at least partially redundant functions and normally make an additive contribution to the branching of phrenic projections. We conclude that RPTP-σ and RPTP-δ make an essential functional contribution to normal muscle innervation by motoneurons. Table 2provides a summary of the quantitative analyses of the various phenotypes described.
Discussion
Here, we demonstrate that RPTP-σ and RPTP-δ make a complementary contribution to motoneuron axon targeting during mammalian development. In the absence of either RPTP-σ or RPTP-δ, mice exhibit severe growth retardation, increased postnatal lethality, and motor dysfunction (Elchebly et al., 1999; Wallace et al., 1999; Uetani et al., 2000). Unlike in invertebrates, in which it has been shown that LAR-RPTP loss of function causes defects in axon extension by motoneurons (Desai et al., 1996, 1997; Krueger et al., 1996), photoreceptors (Garrity et al., 1999; Newsome et al., 2000), and commissural neurons (Sun et al., 2000), mutation of RPTP-σ or RPTP-δ has not been found to generate gross defects in axon guidance in mammals. Among the three members of the mammalian LAR-RPTP subfamily, LAR is most broadly expressed and, in particular, is expressed in many non-neural tissues. In contrast, RPTP-σ and RPTP-δ are both expressed in the CNS during neural development, often by the same cell types (Yan et al., 1993; Mizuno et al., 1994; Sommer et al., 1997; Schaapveld et al., 1998). We postulated that RPTP-σ and RPTP-δ likely play important roles during neural development but that they may compensate for each other functionally in single RPTP mutant mice. To test this hypothesis, we generated mice lacking both RPTP-σ and RPTP-δ. Here, we report that RPTP-σ/δ double-mutant mice exhibit a dramatic neuromuscular phenotype, with complete absence of movement, absence of breathing, and death shortly after birth. Histological analysis confirmed that the lungs fail to inflate in the double mutants (Fig. 2), likely because of inadequate innervation of the diaphragm (Figs. 6, 7). Notably, both RPTP-σ−/−/δ+/− and RPTPσ+/−/δ−/− mice exhibit reduced average life spans compared with either single mutant, suggesting some complementary function. RPTP-σ/δ double mutants also exhibited histological defects in the dermis, brown adipose tissue, and skeletal muscle (Fig. 3). Aberrant innervation by motoneurons is a likely cause of the skeletal muscle defects found in RPTP-σ/δ double mutants. However, we have not excluded the possibility of other roles for RPTP-σ and RPTP-δ, because both are expressed by mesodermal tissues during embryogenesis (Schaapveld et al., 1998).
Consistent with the complete absence of movement in RPTP-σ/δ double mutants, histological examination of the embryonic spinal cord revealed that the majority of ChAT-immunopositive motoneurons were lost by E18.5 (Fig. 5D,F). However, at E13, the number of Islet-1/2-immunopositive motoneurons observed in RPTP-σ/δ double mutants was indistinguishable from wild type (Fig. 5E), suggesting that motoneuron loss is linked to defects in the late stages of axon extension, rather than altered motoneuron differentiation or development. This hypothesis is supported by our findings that phrenic axons in the double mutant appear to exit the spinal cord normally at E12 and appropriately extend toward the diaphragm (Fig. 6B), yet at E13.5, the terminal axonal branches that normally contact the diaphragm are severely stunted in the double mutants (Fig. 6E,F). By E15.5, the branches are absent (Fig. 7E,F) and presumably have been retracted. This interpretation is supported by our observation of a similar distribution of residual branches in the two E13.5 RPTP-σ/δ double mutants examined (Fig. 6F, black arrowheads). Consistent with retraction of these axons at E15.5, the loss of motoneuron cell bodies was clearly evident by E18.5 (Fig. 5D,F). These results suggest that spinal motoneurons are lost in the double mutants as a consequence of their inability to establish appropriate contacts with muscle targets.
Our analyses reveal the absence of specific phrenic nerve branches in RPTP-σ/δ double mutants and reduced branching in RPTP-σ+/−/δ−/− and RPTP-σ−/−/δ+/− (Fig. 7I). These results suggest that the absence of RPTP-σ and RPTP-δ may suppress the ability of axons to defasciculate from the phrenic nerve. Fasciculation is regulated by the relative affinity of the interactions between neighboring axons, and the affinity between an axon and its immediate environment. Interestingly, axon guidance defects with similarities to what we report here have been observed as a result of RPTP loss-of-function mutations in Drosophila. Loss of DLAR function, the Drosophila ortholog of both RPTP-σ and RPTP-δ, causes SNb motoneurons to fail to recognize appropriate muscle targets and generates errors in synapse formation (Krueger et al., 1996). Furthermore, mutation of the Drosophila RPTPs DPTP99A and DPTP69D results in misdirected motoneuron axon growth, defects in motoneuron axon defasciculation, and causes growth cones to stall as they approach muscle targets (Desai et al., 1996). These RPTP loss-of-function mutations in Drosophila produced defects in motoneuron axon guidance similar to those produced by CAM gain of function mutations (Lin and Goodman, 1994; Nose et al., 1994). Moreover, using RNA interference in the chick embryo, Stepanek et al. (2005) provided evidence that dorsal motor nerve fasciculation is disrupted by knock-down of either RPTP-σ or RPTP-δ. Thus, it is tempting to speculate that RPTP-σ and RPTP-δ play a conserved role during mouse neural development, tempering adhesive interactions, thereby allowing motoneurons to appropriately defasciculate, extend, and make synaptic connections.
Quantitative analyses indicated that RPTP-σ+/−/δ−/− and RPTP-σ−/−/δ+/− mutants exhibited intermediate phenotypes in motoneuron survival and phrenic nerve branching, whereas no significant defect in either of the RPTP-σ or RPTP-δ single mutants was detected (Figs. 5, 7I). These results provide strong evidence that RPTP-σ and RPTP-δ complement each other functionally during axon targeting by mammalian motoneurons.
In contrast to the severe motoneuron defects reported, we have not observed abnormal phenotypes in sensory neurons, such as DRG neurons, nor in sympathetic neurons. RPTP-σ is expressed by both motoneurons and sensory neurons. Motoneurons express RPTP-δ but not LAR, whereas the converse is the case for sensory neurons (Sommer et al., 1997; Schaapveld et al., 1998), suggesting that LAR may be able to compensate for the loss of RPTP-σ in sensory neurons in the double mutant. Interestingly, recent findings indicate that RPTP-σ and LAR influence axon regeneration in the adult CNS. The absence of LAR delayed axonal regeneration in vivo after injury to the sciatic nerve, a mixed sensory and motor nerve (Xie et al., 2001; Van der Zee et al., 2003). In contrast, in the RPTP-σ mutant mice, axon regeneration in the sciatic, facial, and optic nerves was enhanced after axotomy (McLean et al., 2002; Thompson et al., 2003; Saphieha et al., 2005). These findings imply that, although LAR-RPTPs may sometimes exhibit redundant functions, the expression of different LAR-RPTPs may also have different functional consequences in some neuronal cell types.
Although the ligands and signal transduction mechanisms associated with LAR-RPTPs are not well understood, recent studies have provided evidence that the extracellular domain of cRPTP-σ binds to matrix-associated heparan sulfate proteoglycans (HSPGs) (Haj et al., 1999; Aricescu et al., 2002), cues that influence axon guidance and play multiple essential roles during neural development (Inatani et al., 2003). The cRPTP-σ extracellular domain also binds developing skeletal myotubes, although the putative ligand involved has not been identified (Sajnani-Perez et al., 2003). The extracellular matrix laminin-nidogen complex has also been identified as a ligand for the LAR-FNIII domain 5 (FN5) (O'Grady et al., 1998). This domain of LAR is highly conserved in all LAR-RPTPs, implicating laminin-nidogen as a potential ligand for RPTP-σ and RPTP-δ as well. Notably, laminins make numerous contributions to neural development, including influencing nerve–muscle synapse formation (Colognato and Yurchenco, 2000).
The intracellular domains of LAR-RPTPs are reported to interact with α-Liprins (LAR interacting proteins) (Pulido et al., 1995) and β-catenin (Kypta et al., 1996; Muller et al., 1999), and these interactions are implicated in the maintenance of hippocampal excitatory synapses (Dunah et al., 2005). The other known mechanisms of downstream signal transduction regulated by LAR-RPTPs point to a role in the control of actin cytoskeletal organization. For example, the DH-GEF (Dbl-homology guanine nucleotide exchange factor) Trio, a key regulator of the small Rho GTPase, regulates actin dynamics in axon growth cones (Lin and Greenberg, 2000) and is functionally downstream of Drosophila DLAR (Bateman et al., 2000). Furthermore, Trio binds Abelson tyrosine kinase (Abl) and Enabled (Ena), both of which are DLAR substrates (Wills et al., 1999). Reducing Trio activity potentiates the motor axon guidance defects produced by DLAR loss of function, which is consistent with Trio collaborating with DLAR during axon guidance (Bateman et al., 2000). Interestingly, mice lacking Trio exhibit phenotypic defects similar to those reported here for RPTP-σ/δ double mutants. These include deficiencies in muscle development (O'Brien et al., 2000), which are consistent with the RPTP-σ/δ double-mutant phenotypes, suggesting that the functional relationship between these genes may be conserved in mammalian neural development. We provide evidence that RPTP-σ and RPTP-δ play functionally complementary roles during murine neural development and together make an essential contribution to appropriate motoneuron targeting. These findings identify a conserved role for the LAR subfamily of RPTPs, regulating motoneuron axon extension in both invertebrates and vertebrates.
Footnotes
-
This work was supported by funding from the Canadian Institutes of Health Research (CIHR) (MOP-42482 to M.L.T.). M.L.T. is a Chercheur National du Fonds de la Recherche en Santé du Québec (FRSQ) and a holder of the Jeanne and Jean-Louis Levesque Chair in Cancer Research. T.E.K. holds a FRSQ Chercheur-Boursier award. N.U. is a recipient of a CIHR-funded Chemical Biology Postgraduate Training award. M.J.C. is a recipient of a FRSQ Doctoral Training Award. We thank Eva Migon and Lisa Michelle Read for their help with excellent animal care, and Cortney Trott for genotyping mice. We also thank all members of the Tremblay laboratory for their valuable discussions. The monoclonal antibodies against Islet-1/2 (39.4D5) developed by Dr. Thomas M. Jessell and neurofilament (2H3) developed by Drs. Thomas M. Jessell and Jane Dodd were obtained from the Developmental Studies Hybridoma Bank developed under the auspices of the National Institute of Child Health and Human Development and maintained by the Department of Biological Sciences, University of Iowa (Iowa City, IA).
- Correspondence should be addressed to Michel L. Tremblay, McGill Cancer Centre, McIntyre Medical Sciences Building, McGill University, 3655 Promenade Sir-William-Osler, Room 701, Montreal, Quebec, Canada H3G 1Y6. Email: michel.tremblay{at}mcgill.ca