|Home | About | Journals | Submit | Contact Us | Français|
Axonal branches of the trigeminal ganglion (TG) display characteristic growth and arborization patterns during development. Subsets of TG neurons express different receptors for growth factors, but these are unlikely to explain the unique patterns of axonal arborizations. Intrinsic modulators may restrict or enhance cellular responses to specific ligands and thereby contribute to the development of axon growth patterns. Protein tyrosine phosphatase receptor type O (PTPRO) which is required for Eph receptor-dependent retinotectal development in chick and for development of subsets of trunk sensory neurons in mouse, may be such an intrinsic modulator of TG neuron development. PTPRO is expressed mainly in TrkB+ and Ret+ mechanoreceptors within the TG during embryogenesis. In PTPRO mutant mice, subsets of TG neurons grow longer and more elaborate axonal branches. Cultured PTPRO−/− TG neurons display enhanced axonal outgrowth and branching in response to BDNF and GDNF compared to control neurons, indicating that PTPRO negatively controls the activity of BDNF/TrkB and GDNF/Ret signaling. Mouse PTPRO fails to regulate Eph signaling in retinocollicular development and in hindlimb motor axon guidance, suggesting that chick and mouse PTPRO have different substrate specificities. PTPRO has evolved to fine tune growth factor signaling in a cell type specific fashion and to thereby increase the diversity of signaling output of a limited number of receptor tyrosine kinases to control the branch morphology of developing sensory neurons. The regulation of Eph receptor-mediated developmental processes by protein tyrosine phosphatases has diverged between chick and mouse.
The trigeminal ganglion (TG) topographically innervates distinct facial regions through its three main branches (maxillary, mandibular and ophthalmic), and conveys mainly sensations of pain, touch and temperature (Erzurumlu et al., 2010). Work over the past decades has identified periphery-derived signals such as members of the nerve growth factor (NGF), glial cell line-derived neurotrophic factor (GDNF), semaphorin and slit families, all having some degree of growth promoting and branching activities, or guidance function (Airaksinen and Saarma, 2002, Davies, 1997, Ma and Tessier-Lavigne, 2007, Rochlin et al., 2000). However, our understanding of the development of branched neuronal morphologies is still poor. Subsets of TG neurons express different receptors for neurotrophic factors, but these factors alone cannot explain the many unique axonal arborization patterns. Receptor-associated proteins such as members of leucine-rich repeat and immunoglobulin superfamilies enhance or suppress receptor activities to control sensory axon growth, guidance or branching (Ledda et al., 2008, Mandai et al., 2009). It is possible that other intrinsic modulators of neurotrophic factor receptors contribute to the development of stereotyped axon growth patterns.
The receptor-type protein tyrosine phosphatase (RPTP) PTPRO, a member of the R3 subfamily (Matozaki et al., 2010), has previously been implicated in motor and retinal ganglion cell (RGC) axon guidance in the chick (Shintani et al., 2006, Stepanek et al., 2005, Stepanek et al., 2001). In mouse, PTPRO was shown to be required for survival and correct axonal projection of trunk sensory neurons (Gonzalez-Brito and Bixby, 2009). PTPRO has not previously been implicated in trigeminal sensory neuron development and the tyrosine-phosphorylated substrates of PTPRO in the mouse remained uncharacterized.
PTPRO can dephosphorylate TrkC, the receptor for neurotrophin-3 (NT-3) (Hower et al., 2009) and all members of the neurotrophin family - NGF, BDNF, NT-3, and NT-4 - regulate various aspects of trigeminal neuron development, including cell survival, axon growth, branching and guidance (Reichardt, 2006). Hence, Trk receptors were prime substrate candidates of PTPRO. PTPRO may also regulate Ret receptor signaling by GDNF family ligands, which are required for development of subsets of DRG (Luo et al., 2009, Luo et al., 2007) and trigeminal neurons (Airaksinen and Saarma, 2002). Finally, Eph receptor tyrosine kinases have previously been identified as PTPRO substrates in the chick retinotectal system (Shintaniet al., 2006). One member of this large receptor family, EphA4, is required for sensory innervation of vibrissae in mice (North et al., 2010), suggesting that mouse PTPRO (mPTPRO) may modulate Eph signaling in TG neurons.
Here, we describe novel functions for PTPRO as a regulator of neurotrophin receptor signaling in trigeminal axon arborization and growth during mouse embryogenesis. Genetic ablation of PTPRO expression in mice enhanced the outgrowth of trigeminal neurons, resulting in an increased arbor size and complexity of the ophthalmic nerve in vivo, and in an enhanced response to BDNF and GDNF ex vivo. Moreover, ablation of PTPRO did not change the sensitivity of trigeminal, RGC, or motor axons to ephrins, suggesting that the regulation of Eph receptors during development by RPTPs has diverged between chick and mouse.
PTPRO−/− and Hb9-GFP transgenic mice have been previously described (Wharram et al., 2000, Wichterle et al., 2002). All the mutants were maintained in a comparable mixed 129/P3J × C57Bl/6 background. Embryos used were of either sex.
E10.5, E11.5, E12.5, E15.5 embryos and newborn pups were fixed in 4 % PFA for 2 hours or overnight, and then incubated overnight in 30% sucrose. Cryostat sections of 30 µm were blocked in 4 % Goat Serum, 4 % Donkey Serum, 2 % BSA, 0.3 % triton in PBS. Primary antibodies were applied overnight in 4 % Goat Serum, 4 % Donkey Serum, 2 % BSA, 0.1 % triton at 4° C. After 45 minute washes in PBS, sections were incubated with secondary antibodies (1:200) at room temperature for 1 hour. After 45 minutes in PBS, cryosections were mounted using Dako fluorescent medium. The following primary antibodies were used: rat anti-PTPRO (kindly provided by Prof. Takashi Matozaki, 1:200), rabbit anti-Lim1 (kindly provided by Dr. Andrea Huber, 1:1000), mouse anti-Islet1 (1:50 from DSHB), rabbit anti-TrkA (1:500 from Millipore), goat anti-TrkB (1:500 from R&D), goat anti-TrkC (1:500 from R&D), goat anti-Ret (1:100 from R&D), mouse anti-NeuN (1:500 Millipore) and mouse anti-Tuj1 (1:100 from Covance). Images were acquired at the Axioplan epifluorescent microscope (Zeiss). For analysis of co-localization and to count neurons, images were acquired using the confocal microscope (Spinning Zeiss Axio Observer Z1 with a Yokagawa Spinning Disk Confocal Unit and a Cool SNAP HQ2 CCD Camera).
E11.5 and E12.5 embryos were fixed overnight in Dent`s solution (1 part DMSO; 4 parts methanol). Then, they were bleached in one part 30% H2O2 – two parts Dent’s Solution for several hours at room temperature. Three washing steps (one hour each at RT) in PBS containing 0.2% Gelatin and 1% Triton X-100 (SIGMA) were followed by incubation with the anti-neurofilament antibody (NF-160 from Sigma 1:300 in 4 parts newborn calf serum, 1 part DMSO) overnight at room temperature. Five washing steps in TBS containing 1 % Triton X-100 and 0.2 % gelatin for one hour each were followed by incubation with anti-mouse HRP-conjugated antibody (1:300 in 4 parts newborn calf serum, 1 part DMSO) overnight at room temperature. Finally, embryos were washed and developed in diaminobenzidine working solution followed by dehydration in methanol and clearing in BABB (1 part benzyl alcohol, 2 parts benzyl benzoate). Images were acquired using the DC150 camera from Leica and analyzed using ImageJ or NeuronJ. The ophthalmic nerve phenotype at E11.5 was quantified as the ratio between the area of the ophthalmic nerve arbor and the area of the maxillary nerve arbor. The ophthalmic nerve arbor complexity at E12.5 was analyzed using the Sholl analysis plug-in of NeuronJ. The hindlimb phenotype at E12.5 was quantified as ratio between the diameter of the tibial nerve and the diameter of the peroneal nerve.
Dissociated cultures of trigeminal neurons from E12.5 embryos were grown onto polyornithine/laminin coated 4-well plates. Neurons were grown for 18 hours in F12 medium supplemented with 10ng/ml NGF (R&D), and where indicated, 5ng/ml BDNF (R&D) or 5ng/ml GDNF (R&D) were added to the culture medium. Neurons were fluorescently labelled with calcein-AM (Invitrogen) and imaged with an Axiovert 200M microscope (Zeiss) using a 10X objective. Neurite length and number of branches were estimated as described in (Gutierrez and Davies, 2007). For the culture in presence of caspase inhibitors, neurons where grown onto polyornithine/laminin coated coverslips for 18 hours in F12 supplemented with 10µM Q-VD-Oph (Calbiochem) and NGF, BDNF or GDNF as indicated. Neurons were stained with cell tracker green (Invitrogen), fixed 5 minutes with 4 % PFA, and coverslips were mounted using Dako fluorescent medium. Images were acquired with Zeiss epifluorescent microscope. Explant cultures of trigeminal neurons from E12.5 embryos were grown on poly-D-lysin/laminin coated coverslips for 15 hours in F12 medium supplemented with 10ng/ml NGF.
Explant cultures of motor neurons from the lower half of E12.5 lumbar LMC were grown on poly-D-lysin/laminin coated coverslip. Neurons were grown for 15 hours in Neurobasal medium supplemented with B27, glutamine, glutamate, penicillin/streptomycin, 1ng/ml BDNF, 1ng/ml GDNF and 10ng/ml CNTF.
Trigeminal explants were stimulated for 30 minutes with 0.5µg/ml pre-clustered ephrinA5 or with 0.5µg/ml pre-clustered human IgG Fc-fragments as a control. Motor neuron explants were stimulated for 30 minutes with 0.1µg/ml and 0.5µg/ml pre-clustered ephrinA2/A5 (mixed 1:1), or with 0.1µg/ml and 0.5µg/ml pre-clustered human IgG Fc-fragments as a control. Explants were fixed twice for 30 minutes in 2 % PFA-15 % sucrose, blocked in 0.5 % Triton X-100, 1 % BSA in PBS and then stained using Phalloidin568 (1:100 Invitrogen). Coverslips were mounted using Dako fluorescent medium and images acquired with an Axioplan epifluorescence microscope (Zeiss).
HEK293 and HeLa cells were cultured in DMEM supplemented with 10 % fetal bovine serum and 1 % penicillin/streptomycin. Cells were transfected using Lipofectamine2000 (Invitrogen), according to the manufacturer’s instructions, kept at 37° C for 24 hours, and then stimulated with 50ng/ml BDNF or 50ng/ml GDNF and soluble GFRα1 (R&D), and harvested. Plasmids for mammalian expression used were mPTPRO in pFlag-CMV-5 (kindly provided by Dr. Eek-hoon Jho), cPTPRO-Flag in pcDNA3, TrkB in pMEX-neo, EphA4 in pFlag-CMV-3, Ret51 in pcDNA3. Lysis buffer (50mM Tris-HCl, pH 7.5, 150mM NaCl, 1 % Triton X-100) was supplemented with protease inhibitor cocktail and phosphatase inhibitor cocktail tablets (Roche). Proteins were separated by 7.5 % SDS-PAGE, transferred onto nitrocellulose membranes and blotted with the following primary antibodies: rabbit anti-Flag (1:1000 Sigma), goat anti-TrkB (1:1000 R&D), mouse anti-phosphotyrosine (1:1000 hybridoma clone 4G10), mouse anti-phospho p44/42 MAPK (1:2000 Cell Signaling), rabbit anti-p44/42 MAPK (1:1000 Cell Signaling), mouse anti-phosphoRet pY1062 (1:1000 R&D), goat anti-Ret (1:1000 Fitzgerald), rabbit anti-phosphoEph (1:1000 Abcam), mouse anti-EphA4 (1:5000 Zymed). The blots were then incubated with HRP-conjugated secondary antibodies (1:5000 Amersham Biosciences). Luminol (Amersham) was used for chemiluminescence detection. For immunoprecipitations, lysates were incubated overnight at 4°C with TrkB antibody (R&D). The proteinA-Sepharose beads (GE Healthcare) were then added to the mixture and incubated at 4°C for 2 hours.
Transfected HeLa cells were stimulated with 50ng/ml BDNF or GDNF/GFRα1, fixed in 4 % PFA for 20 minutes on ice. Then they were permeabilized in 0.1 % Triton X-100 in PBS, blocked in 3 % BSA in PBS and stained with rabbit anti-Flag (1:1000 Sigma) to detect PTPRO expression, rat anti-PTPRO (1:200) to detect PTPRO surface expression, goat anti-TrkB (1:500 R&D), goat anti-Ret (1:200 R&D) and mouse anti-phosphotyrosine (1:500 4G10 Millipore) and with Cell Mask Blue (1:1000 Invitrogen) to detect the cell outline. To determine the surface expression of TrkB, Ret and PTPRO cells were not permeabilized. A single plane of a confocal stack was analyzed using ImageJ plugin to determine the colocalization of TrkB and Ret with PTPRO before and after stimulation. Using Metamorph the outline of the cells was drawn based on the Cell Mask Blue staining and the intensity was determined for TrkB, Ret and the phosphotyrosine stainings.
E12.5 embryos were eviscerated and kept in DMEM/F-12 medium (Invitrogen) aerated with 5 % CO2/95 % O2. A 6 % lysine-fixable tetramethylrhodamine-dextran (molecular weight 3000, Invitrogen) solution in PBS with 0.4 % Triton X-100 was injected into the ventral and allowed to diffuse for 5–6 hours at RT.
Anterograde tracing experiments were essentially performed as described in Rashid et al., 2005 (Rashid et al., 2005). In brief, the offspring from crosses of PTPRO+/− mice was analyzed at P8, a time at which the retino-collicular projection is considered mature. Anterograde tracing of RGC axons was performed by focal injection of DiI (Molecular Probes, Eugene, OR) as a 10 % solution in dimethylformamide into the peripheral region of the nasal retina. About 24 hours later, mice were deeply anesthetized and perfused transcardially with PBS. The SC and IC as well as the injected retinae were whole mounted onto glass slides and examined under UV light. The injection sites of all retinae were verified by fluorescence imaging of flat mounts.
To understand the role of PTPRO in developing TG neurons, we examined its temporal expression pattern by immunostaining between embryonic day 10.5 (E10.5) and post-natal day 0 (P0). The specificity of the PTPRO antibody was tested on different tissues from PTPRO wild-type and PTPRO−/−embryos (data not shown; see also (Kotani et al., 2010)). At E10.5, PTPRO is barely detectable in trigeminal neurons, but by E11.5, PTPRO expression is seen on trigeminal cell bodies and axons. PTPRO expression is maintained through all later stages of embryonic development and in newborns (Figure 1A–F). TG neurons are subdivided into several subpopulations, including TrkA-expressing nociceptive, TrkB- or Ret-expressing mechanoceptive, and TrkC-expressing neurons. Later in development, TrkA+ neurons further differentiate into peptidergic and non-peptidergic neurons, and start expressing Ret (Marmigere and Ernfors, 2007). To investigate which subpopulations expressed PTPRO we performed co-immunostainings with TrkA, TrkB, TrkC and Ret antibodies at three different developmental stages: E12.5 (axon elongation and branching), E15.5 (axon arborization), and P0 (Erzurumlu et al., 2006). At E12.5 PTPRO was expressed in roughly half of TrkB+ and Ret+, in a small population of TrkC+, but rarely in TrkA+ neurons (Figure 1C–F). At E15.5, and similarly at P0, PTPRO expression strongly decreased in TrkB+ but remained high in Ret+ neurons, and did not increase in the other two populations (Figure 1B–F and data not shown). The expression pattern suggested that PTPRO was localized in mechanoceptive trigeminal neurons in the early phases of development. Since PTPRO is reportedly expressed in E16.5 DRGs (Beltran et al., 2003), we also performed immunostainings in developing lumbar DRGs. At E12.5 and E15.5, PTPRO was expressed primarily in TrkB+ and Ret+ neurons, as observed for the TG. By birth, expression decreased in TrkB+ and Ret+ and increased in TrkC+ neurons. PTPRO expression in TrkA+ neurons was low across stages (data not shown).
To test if PTPRO modulates the outgrowth and arborization of trigeminal axons in vivo, we followed the development of the three major trigeminal axon branches using whole-mount neurofilament immunostaining in PTPRO−/− embryos. No major changes were observed in the mandibular branch of E11.5 and E12.5 whole-mount stained embryos (data not shown), but we found a marked defect in one of the arbors of the ophthalmic branch. This projection normally starts to grow at E10.5, forms a highly branched arbor above the eye at E12.5 (Figure 2A,D) and becomes fully developed, covering the entire upper half of the face by E13.5 (data not shown). In E11.5 wild-type embryos the arbor had formed two main branches, whereas in stage-matched PTPRO−/− embryos there were secondary branches growing out of the main branches and the area covered by the arbor was bigger (Figure 2A,B). Similarly, the maxillary arbor covered a bigger area in PTPRO−/− embryos (Figure 2B). The difference in the ophthalmic branch was, however, greater than in the maxillary branch. Hence, the ratio measurement (ophthalmic/maxillary branch), designed to normalize for small developmental differences, also showed a significant increase (Figure 2C). No significant differences were observed between wild-type and heterozygous PTPRO+/− embryos (Figure 2B,C). At E12.5 we analyzed the ophthalmic projections by Sholl analysis (Gutierrez and Davies, 2007) and found an increased complexity in PTPRO−/− embryos compared to wild-type controls, indicating a role for PTPRO as outgrowth and/or branching inhibitor (Figure 2D,E).
Next we prepared cryosections of E12.5 embryonic heads and further analyzed the complexity of the maxillary branch to complement the quantification in the whole-mount configuration. Immunostainings for the axon marker Tuj1 revealed more numerous areas of defasciculation of the maxillary axon bundle in PTPRO−/− embryos than in wild-type littermates (Figure 2F,G), possibly caused by enhanced branching or defasciculation of these neurons. The areas of defasciculation were seen mainly in the proximal region of the nerve; more distal terminal arborizations were not affected (data not shown).
To examine if the exuberant complexity of trigeminal arbors in PTPRO−/− embryos is the result of enhanced responsiveness to growth promoting signals, we prepared primary trigeminal neuron cultures and stimulated them with different neurotrophic factors. E12.5 neurons were maintained for 18 hours in the presence of 10ng/ml NGF to promote their survival, alone or in combination with 5ng/ml BDNF or 5ng/ml GDNF (Figure 3A). Since PTPRO is mainly expressed in TrkB+ and Ret+ neurons, we expected an effect on growth and branching only in the presence of BDNF and GDNF. Indeed, stimulation with NGF alone did not show differences in outgrowth or branching between wild-type and PTPRO−/− neurons (Figure 3B,C). In contrast, in the presence of BDNF and GDNF, PTPRO−/− neurons had longer axons than their respective controls (Figure 3B). Although E12.5 neurons were mainly bipolar, BDNF stimulation triggered a significant increase in the mean number of primary branch points (Figure 3C). BDNF stimulation increased the number of branch points to the same extent in wild-type and PTPRO−/− neurons, whereas GDNF stimulation enhanced branching only in PTPRO−/− neurons (Figure 3C).
To better uncouple effects on axon growth from those on cell survival, we next cultured TG neurons in the presence of caspase inhibitors instead of NGF to promote their survival, and scored responses to neurotrophins and GDNF at different doses. Interestingly, even in the absence of extrinsic factors, PTPRO−/− axons were longer than wild-type axons (Figure 3D). As before, while treatment with NGF promoted similar outgrowth in wild-type and PTPRO−/− axons, BDNF and GDNF induced greater axon growth in PTPRO−/− neurons than in controls (Figure 3D). For all groups, growth was stimulated most at physiological levels of neurotrophic factor (1ng/ml). Axon branching in response to BDNF and GDNF was also enhanced in PTPRO−/− neurons. Strongest differences were seen at intermediate concentrations of neurotrophic factors (10ng/ml) and the responses generally plateaued by 100ng/ml (Figure 3E). Stimulation with NGF at physiological concentrations produced similar axon branching responses in wild-type and PTPRO−/− neurons (data not shown). These results indicate that embryonic PTPRO−/− neurons are more responsive to BDNF and GDNF, consistent with the expression of PTPRO in TrkB+ and Ret+ neurons.
Since cranial sensory neurons display intrinsic differences in growth rates (Davies, 1989), the enhanced growth and arborization of a sensory nerve branch may also result from a relative increase in the numbers of fast versus slow growing neurons. Such alterations may arise from changes in cell fate or cell loss of a selective subpopulation. To test this hypothesis, we counted the numbers of TrkA+, TrkB+, TrkC+, and Ret+ neurons at E12.5 and P0. At E12.5, the predominant subpopulation of trigeminal neurons was NGF-dependent and expressed TrkA, while the other three subpopulations together accounted for less than half (Huang et al., 1999b) of the overall contingent (Figure 4A,B). At P0, TrkA+ neurons were still the largest subpopulation, although reduced in number compared to E12.5; TrkB+ neurons were unchanged in numbers compared to E12.5; TrkC+ neurons were slightly reduced, and the Ret+ population had increased (Huang et al., 1999a) (Figure 4C,D). In PTPRO−/− embryos, the distribution of neuronal subpopulations was not significantly different from controls (Figure 4B), suggesting that the absence of PTPRO did not affect the cell fate of these neurons. At P0, there was a small, but significant reduction in the numbers of TrkA+ and TrkC+ neurons, and no significant change in TrkB+ and Ret+ neurons (Figure 4D). Staining for the general neuronal marker “NeuN” at E12.5 and P0 did not reveal changes in the total numbers of neurons (Figure 4A–D). Together these results suggest that changes in cell fate and survival do not contribute significantly to the exuberant growth of E12.5 embryonic trigeminal axons.
Next we asked if PTPRO interacts directly with TrkB and Ret, and exerts its growth suppressive function by suppressing TrkB and Ret kinase signaling. We tried to examine co-localization of PTPRO with RTKs (Receptor Tyrosine Kinases) in cultured neurons, but were unable to detect PTPRO with sufficient subcellular resolution with the available antibodies. As an alternative, we investigated co-localization and activation of these proteins in cell culture, over-expressing the mouse isoform of PTPRO (mPTPRO). When TrkB and mPTPRO were co-expressed, the two receptors strongly co-localized on the cell surface, but much less so in the cell interior (Figure 5A and data not shown). Over-expression of TrkB in HeLa cells led to ligand-independent activation (Shintani and Noda, 2008), as shown by anti-phosphotyrosine immunostaining (Figure 5B). In the presence of PTPRO the intensity of phosphotyrosine immunostaining was markedly reduced (Figure 5C,D). Moreover, BDNF-induced autophosphorylation of TrkB and ERK1/2 phosphorylation were strongly suppressed by co-expressed mPTPRO (Figure 5E–G).
Ret has two main isoforms: Ret9 and Ret51. Since they elicit similar responses to GDNF stimulation in sympathetic neurons (Encinas et al., 2008), we used only Ret51 for the following in vitro experiments. In the case of Ret, co-localization with mPTPRO was as strong as for TrkB and the degree of co-localization in transfected HeLa cells was enhanced by stimulation with GDNF and soluble GFRα1 (Figure 6A,D–E). However, the increase in co-localization was detectable only when cells were permeabilized, suggesting that mPTPRO and Ret may co-localize in intracellular compartments, e.g. endosomes, upon stimulation. Stimulation with GDNF and soluble GFRα1 also increased the intensity of phosphotyrosine staining and this increase was suppressed by the presence of mPTPRO (Figure 6B–D,F). In transfected HEK293 cells, basal Ret autophosphorylation, which was visualized by immunoblotting with anti-phosphotyrosine antibodies, was high and was not increased by GDNF stimulation, but was strongly suppressed by co-expression of mPTPRO (Figure 6G–I). mPTPRO effects on overall Ret tyrosine phosphorylation were significant, but stronger on Ret phosphotyrosine 1062, suggesting that PTPRO may dephosphorylate only a subset of tyrosine residues present on the receptor (Figure 6G,I). This strong inhibition of Ret kinase activity correlated well with a pronounced inhibition of GDNF-induced ERK phosphorylation (Figure 6G,J).
Finally, we checked whether PTPRO co-expression influenced surface levels of Ret and TrkB in the presence and absence of their ligands. Upon stimulation with BDNF for 5 minutes, TrkB expression on the surface slightly increased and no differences were observed if PTPRO was co-expressed. Upon stimulation with GDNF, Ret was internalized. Also in this case, we did not observe significant differences in Ret distribution when PTPRO was co-expressed (data not shown).
Together these results suggest that PTPRO regulates TrkB and Ret kinase activity and signaling, supporting the role of PTPRO as a negative regulator of BDNF- and GDNF-induced axon growth and branching.
To examine if the exuberant complexity of axonal arbors in PTPRO−/− embryos might also result from altered responsiveness to specific repellent signals, we investigated whether Eph tyrosine kinase signaling was affected. cPTPRO is a known regulator of Eph signaling in the retinotectal system (Shintaniet al., 2006), and several Eph family members are expressed in the TG (Luukko et al., 2005). TG explant cultures from E12.5 mouse embryos were stimulated with soluble ephrinA5 fused to the Fc portion of human IgG (ephrinA5-Fc) or with control Fc, and stained with phalloidin to visualize the growth cones (Figure 7A). Stimulation with ephrinA5-Fc led to a marked increase in the numbers of collapsed growth cones of TG neurons, but there was no difference in the rate of collapse between wild-type and PTPRO−/− explants (Figure 7B), arguing against an impairment of Eph signaling as the cause of the increased complexity of TG axonal arbors. To explore in vivo whether Eph signaling is regulated by PTPRO (Shintaniet al., 2006), we analyzed the retinocollicular projection in PTPRO−/− mice at post-natal day 8 (P8). Although PTPRO is expressed in the retina at P0 (data not shown), there were no obvious topographic targeting defects or ectopic branching of RGC axons in PTPRO−/− mice compared to littermate controls (Figure 7C).
These results suggested that mouse and chick PTPRO may have different substrate specificities. To test this hypothesis further, we examined the responsiveness to ephrins in another mouse neuronal population, namely limb innervating (lateral motor column, LMC) motor neurons, and the rate of EphA4 dephosphorylation by either mPTPRO or cPTPRO. We stimulated explants of lumbar motor neurons from Hb9-GFP+ transgenic embryos (Wichterleet al., 2002) with different doses of either control Fc or a 1:1 mix of ephrinA2 and ephrinA5-Fc, and ephrinB2-Fc. Both subtypes of ephrins markedly increased the fraction of collapsed growth cones in the explant, irrespective of the presence or absence of PTPRO (Figure 7D,E and data not shown). These results demonstrate that PTPRO−/− LMC motor neurons are not more sensitive to ephrin stimulation, despite the fact that PTPRO is prominently expressed in both subpopulations of LMC neurons (data not shown). Also, in vivo, guidance of the lateral cohort of LMC (LMCL) motor axons to the dorsal limb, though critically dependent on ephrinA/EphA4 signaling (Eberhart et al., 2002, Helmbacher et al., 2000, Kramer et al., 2006) is unaffected in PTPRO−/− embryos. The evidence includes quantification of axon bundle diameters in neurofilament whole-mount stained embryos and axon tracings with Rhodamine Dextran (RD) injected into the ventral or dorsal hindlimb as previously described (Dudanova et al., 2012, Krameret al., 2006) (Figure 7F–I and data not shown). To examine the substrate specificity of PTPRO, HEK293 cells co-expressing either mPTPRO or cPTPRO with mouse EphA4 were stimulated with control Fc or ephrinA4-Fc and EphA4 autophosphorylation was assessed by Western blotting. Interestingly, only cPTPRO, was able to significantly dephosphorylate EphA4 (Figure 7J,K). These results indicate that mPTPRO does not regulate Eph signaling and suggest that mPTPRO and cPTPRO have different substrate specificities.
Here we have shown that PTPRO, a receptor tyrosine phosphatase, is expressed in a large fraction of TrkB+ and Ret+ mechanoreceptors within the TG during embryogenesis. In PTPRO−/− mice, TG axons grow longer and exuberant branches, suggesting that PTPRO suppresses the response to endogenous growth factors. Furthermore, cultured PTPRO−/− TG neurons show more BDNF- and GDNF-induced axonal outgrowth and branching than control neurons, indicating that PTPRO regulates the activity of TrkB and Ret receptors. This role of PTPRO is also seen in the chick system suggesting that this function is evolutionarily conserved. Though PTPRO has been shown to regulate Eph signaling in the chick retinotectal system, we find that mouse PTPRO fails to do so in several neuronal systems in vivo and in vitro, suggesting that the regulation of Eph receptor-mediated developmental processes by RPTPs has diverged between chick and mouse.
Our work revealed a role for PTPRO as a negative modulator of TG axon growth and branching in response to specific growth factor stimulation. PTPRO is excluded from the largest population of TG neurons, TrkA+ nociceptive neurons, and the response of cultured PTPRO−/− TG neurons to exogenously added NGF is not different from wild-type. In contrast, the strong response of cultured PTPRO−/− TG neurons to BDNF and GDNF correlates well with the prominent expression of PTPRO in TrkB+ and Ret+ neurons. These results strongly suggest that PTPRO has a cell autonomous role in constraining TG neuron growth and branching. Interestingly, the loss of PTPRO affects neuron growth and branching at lower concentration of BDNF and GDNF, hinting at the requirement of the phosphatase in setting a functional threshold in response to neurotrophin stimulation. Genetic ablation of PTPRO does not change the maximal effect of the response, but rather shifts the dose-response curve to the left (higher sensitivity to neurotrophin stimulation).
Mechanistically, we show that PTPRO suppresses ligand-induced autophosphorylation of TrkB and Ret, and that Ret activation enhances co-localization with PTPRO, suggesting that these RTKs are direct substrates of PTPRO in living cells. Alternatively, PTPRO may target other tyrosine phosphorylated downstream effectors of these RTKs (see below). Moreover, we show that ligand-induced activation of ERK signaling is suppressed by co-expression of PTPRO. This result is consistent with the known function of ERK signaling in sensory axon growth (Markus et al., 2002, Zhong et al., 2007).
The function of PTPRO as a negative modulator of RTK signaling is reminiscent of RPTPσ, a class IIa RPTP, which limits neurite outgrowth of dorsal root ganglia (DRG) sensory neurons by directly dephosphorylating either TrkA and TrkC receptors (Faux et al., 2007) or the cell adhesion molecule N-cadherin (Siu et al., 2007). Similar to TG neurons in PTPRO−/−, DRG neurons in RPTPσ−/− mice exhibit faster growth rates ex vivo (Siuet al., 2007), but in vivo sensory axon development has not been analyzed. Negative modulation of neurotrophic factor receptors is not restricted to RPTPs. Lrig1, a transmembrane proteins with leucine-rich repeats (LRRs) in its ectodomain abolishes GDNF/Ret-induced axon outgrowth by inhibition of GDNF binding and recruitment of Ret to lipid rafts (Leddaet al., 2008).
RPTPs and LRR proteins can also be positive regulators of RTK signaling. Leukocyte common antigen-related (LAR) RPTP enhances TrkB signaling by dephosphorylation and activation of Src downstream of TrkB. LAR−/− hippocampal neurons display reduced TrkB signaling and diminished BDNF-induced survival (Yang et al., 2006). More recently, LAR was shown to bind heparan sulfate proteoglycans to mediate attractive guidance of sensory axons to the skin; however, the signaling mechanism underlying this function is unknown (Wang et al., 2012). The LRR protein Linx positively modulates NGF/TrkA and GDNF/Ret signaling, possibly by promoting ERK signaling. Axonal defects in Linx mutants resemble those in mice lacking NGF, TrkA and Ret (Mandai et al., 2009). The neurotrophin receptor p75 which can directly interact with Trks (Dechant, 2001), is also important for growth of the ophthalmic branch and sensory axons to the limbs (Ben-Zvi et al., 2007, Bentley and Lee, 2000); it is currently unknown whether the interaction of p75 with Trk receptors is essential for this in vivo function. PTPRO and other RTK interactors seem to have evolved to fine tune growth factor signaling in opposing manners and in a cell type specific fashion. These proteins therefore increase the diversity of signaling output of a limited number of RTKs and growth factors to control the branch morphology and connectivity of developing neurons.
We also explored the possibility that PTPRO modulates responses to repulsive guidance cues, thereby limiting axon branching. Repulsive guidance cues such as members of the Slit family have previously been shown to positively modulate sensory axon branching (Ma and Tessier-Lavigne, 2007). Moreover, PTPRO was initially characterized as negative regulator of repulsive Eph signaling in the chick retinotectal system (Shintaniet al., 2006). Gain- and loss-of-function (shRNA knockdown) studies established a role for PTPRO in retinotectal projection guidance in vivo, presumably involving Eph regulation (Shintaniet al., 2006). Several lines of evidence suggest that mouse PTPRO does not control Eph signaling: (1) Ex vivo cultures of PTPRO−/− and control TG neurons are equally sensitive to ephrin-induced growth cone collapse; (2) the retinotopic maps in PTPRO−/− mice were indistinguishable from control mice; (3) limb motor axons which are sensitive to Eph/ephrin signaling (Bonanomi and Pfaff, 2010) did not show projections defects in PTPRO− − embryos in vivo and were not more sensitive to ephrins ex vivo; (4) chick, but not mouse, PTPRO is able to dephosphorylate EphA4 in living cells. These results suggest that mPTPRO either does not control Eph activity in the retinocollicular system, or is functionally redundant with other phosphatases that regulate Eph activity (Nievergall et al., 2010). Mouse and chick PTPRO are divergent enough (80% similarity) to have different substrate specificities, as was previously shown for mouse and human RPTPσ (Hou et al., 2011). Future structure-function work to elucidate the molecular basis of substrate specificity of mouse versus chick PTPRO would be interesting to help shed light into this rather unexplored question.
Previously, PTPRO−/− mice were shown to have decreased numbers of a subset of nociceptive (CGRP+) DRG neurons at birth and as adults (Gonzalez-Brito and Bixby, 2009). In addition, the central projections of the surviving nociceptive and proprioceptive (parvalbumin+) DRG neurons were abnormal and PTPRO−/− mice performed abnormally on tests of thermal pain and sensorimotor coordination (Gonzalez-Brito and Bixby, 2009). Here we show that during early embryonic development, PTPRO is rarely co-expressed with TrkA or TrkC, antigens that are early markers of nociceptive and proprioceptive neurons. Moreover, PTPRO−/− trigeminal cultures do not show aberrant sensitivity towards exogenous NGF. However, at birth, expression of PTPRO increases in TrkC+ neurons and PTPRO−/− mice show partial loss of TrkA+ and TrkC+ sensory neurons. Thus, the observed changes in nociceptive and proprioceptive subpopulations may at least in part be due to non-cell autonomous functions of PTPRO. Whether PTPRO has a pro-survival function or affects cell differentiation remains to be clarified. PTPRO is also expressed at the spinal cord midline and in the dorsal root entry zone (DREZ) of DRG axons in the spinal cord (data not shown and (Beltranet al., 2003)) and its removal could non-cell autonomously affect the positioning of nociceptive and proprioceptive fibers. A non-cell autonomous role of PTPRO was previously suggested in RGC axon guidance, because the PTPRO ectodomain is chemorepulsive for RGC axons ex vivo (Stepaneket al., 2001). Conditional ablation of PTPRO in sensory axons versus their target fields should resolve these mechanistic questions.
In summary, our results have shown that PTPRO fine tunes growth factor signaling in a cell type specific fashion, and thereby controls the axonal branch morphology of developing TG neurons. Future work will show if this function of PTPRO also controls the connectivity of TG neurons in the adult. It will also be interesting to explore if PTPRO activity is regulated by interaction with extracellular ligands. Our work has also shown that the regulation of Eph receptor-mediated developmental processes by RPTPs has evolved differently between chick and mouse. It will be important to identify the molecular features that determine substrate specificity of cPTPRO and mPTPRO, and possibly of other protein tyrosine phosphatases.
We thank L. Gaitanos for technical assistance; D. Marinescu and S. Krinner for mouse genotyping; A. Huber-Brösamle, T. Matozaki, and Eek-hoon Jho for kindly providing reagents; T. Gaitanos, P. Klein, and S. Paixão for fruitful discussions. This study was supported by the Max-Planck Society, the European Union (MOLPARK), the Deutsche Forschungsgemeinschaft (SFB870), the NIH, and the Wellcome Trust.
The authors declare no competing financial interests.