N-Cadherin Is an In Vivo Substrate for Protein Tyrosine Phosphatase Sigma (PTPσ) and Participates in PTPσ-Mediated Inhibition of Axon Growth

Abstract

Protein tyrosine phosphatase sigma (PTPσ) belongs to the LAR family of receptor tyrosine phosphatases and was previously shown to negatively regulate axon growth. The substrate for PTPσ and the effector(s) mediating this inhibitory effect were unknown. Here we report the identification of N-cadherin as an in vivo substrate for PTPσ. Using brain lysates from PTPσ knockout mice, in combination with substrate trapping, we identified a hyper-tyrosine-phosphorylated protein of ∼120 kDa in the knockout animals (relative to sibling controls), which was identified by mass spectrometry and immunoblotting as N-cadherin. β-Catenin also precipitated in the complex and was also a substrate for PTPσ. Dorsal root ganglion (DRG) neurons, which highly express endogenous N-cadherin and PTPσ, exhibited a faster growth rate in the knockout mice than in the sibling controls when grown on laminin or N-cadherin substrata. However, when N-cadherin function was disrupted by an inhibitory peptide or lowering calcium concentrations, the differential growth rate between the knockout and sibling control mice was greatly diminished. These results suggest that the elevated tyrosine phosphorylation of N-cadherin in the PTPσ^−/− mice likely disrupted N-cadherin function, resulting in accelerated DRG nerve growth. We conclude that N-cadherin is a physiological substrate for PTPσ and that N-cadherin (and likely β-catenin) participates in PTPσ-mediated inhibition of axon growth.

Protein tyrosine phosphatase sigma (PTPσ) (also known as LAR-PTP2, PTP-P1, CRYPα, PTP-NU3, PTP-NE3, and CPTP1 [42, 51, 58, 68, 71, 81, 84]) belongs to the LAR family of receptor PTPs (class IIa) (3, 4, 13). In mammals, this family includes LAR, PTPσ, and PTPδ, whereas DLAR is the Drosophila melanogaster orthologue of all three family members (13, 61, 62). In this class (IIa) of PTPs, the ectodomain is comprised of three immunoglobulin (Ig) domains and usually four or eight FNIII repeats, resembling cell adhesion molecules (CAMs) such as N-CAM and L1, a single pass transmembrane domain, and two tandem intracellular catalytic domains, the first of which (D1) is active and the second of which (D2) is inactive (13). Like other CAMs, LAR family PTPs have been demonstrated to play an important role in nervous system development (reviewed in references 13, 22, and 59).

In flies, DLAR and its relative PTP69D play a key role in axon guidance of motoneurons and axon pathfinding in the visual system (14, 17-19, 30, 34). Likewise, in the leech, HmLAR2 was shown to concentrate in the growth cone and direct growth of its processes (6, 26). Moreover, retinal ganglion outgrowth and growth cone integrity in developing Xenopus laevis and avian embryos require a functional PTPσ (CRYPα) (27, 31, 38, 47, 60). PTPδ is also expressed in Xenopus retinal ganglion cells during axonogenesis (27), and it acts as an attractive cue for growth cone steering and for driving axonal growth of forebrain neurons (63, 73).

In mammals, PTPσ is expressed primarily in the nervous system and in select epithelia, and its expression is developmentally regulated, generally declining with gestational age (29, 36, 52, 71, 72). In view of this developmentally dependent expression, we (69) and others (21) previously generated PTPσ knockout mice. The PTPσ knockout animals display neuroendocrine and neuronal defects, with severe reduction of growth hormone and prolactin production by the pituitary (8, 21), as well as central and peripheral nervous system (PNS) abnormalities (35, 36, 65). We previously showed that, in the central nervous system, knockout of PTPσ causes hippocampal dysgenesis, reduction in thickness of the corpus callosum and cerebral cortex, and spinal cord abnormalities (36). In the PNS, we demonstrated delayed myelination and nerve conduction velocity in the developing sciatic nerve of newborn PTPσ knockout mice (69). Moreover, our analysis of nerve regeneration of adult sciatic nerve following crush injury, nerve transection and immediate repair, or reciprocal allografting between the knockout and sibling sciatic nerve revealed accelerated nerve regeneration and axon guidance errors in the PTPσ knockout animals (35). Enhanced axon outgrowth was also shown for the regenerating facial nerves and retinal ganglion cells of PTPσ knockout mice after injury (53, 65). These findings demonstrate that PTPσ is important for axon guidance in vivo in mammals and that it is inhibitory to the rate of nerve growth. How PTPσ transmits its inhibitory signal is unknown, as its substrate(s) and downstream signaling pathway(s) have not been identified.

By use of our PTPσ knockout mice in combination with substrate trapping and mass spectrometry, we report here the identification of N-cadherin as an in vivo substrate for PTPσ and show that N-cadherin (likely in collaboration with β-catenin) participates in the downregulation of axon growth by PTPσ.

MATERIALS AND METHODS

Animals.

PTPσ knockout mice were generated (on a C57BL/6/129 background) and bred as described previously (69) The PTPσ^+/− mice are phenotypically indistinguishable from the wild-type mice. The PTPσ^−/− mice were divided into three cohorts: 60% died within 48 h after birth (cohort 1), ∼38% survived until 2 to 3 weeks of age and most then died due to a wasting syndrome (cohort 2), and the remaining 2.5% survived to adulthood but were roughly 25 to 50% smaller than the wild-type siblings by weight (cohort 3). Mice from the second cohort were used for all experiments.

Generation of PTPσ constructs.

Rat PTPσ cDNA (GenBank accession number L11587) was obtained from B. Goldstein. Flag-tagged full-length PTPσ, the first catalytic domain (D1), or the tandem first and second catalytic domains (D1D2), were cloned into the pIRES mammalian expression vector.

Generation of GST-PTPσ fusion proteins.

The sequences corresponding to the D1 domain (GenBank accession number NP_035348, amino acids 1282 to 1607) and the D1D2 domain (amino acids 1282 to 1904) of PTPσ were subcloned into pGEX-2T (Amersham). The substrate trapping mutant with an Asp-to-Ala (D1472A) mutation was made by site-directed mutagenesis (Stratagene) using PCR and verified by sequencing. Glutathione S-transferase (GST) fusion proteins of PTPσ-D1 and PTPσ-D1D2 [GST-σD1(WT), GST-σD1(DA), GST-σD1D2(WT), and GST-σD1(DA)D2] (where WT is wild type and DA is an Asp-to-Ala substitution) were generated according to standard procedure (e.g., see reference 70).

Generation of soluble N-cadherin-Fc chimeric protein.

The protocol for generation of soluble N-cadherin-Fc chimeric protein was adapted from reference 67. Briefly, COS-7 cells were plated on 150-mm tissue cultures dishes and allowed to grow until ∼70% confluent. They were then transfected with a construct expressing the extracellular domain of chicken N-cadherin fused to the Fc region of human IgG (Ncad-Fc) in the pIG1 vector (kind gift of P. Doherty), as described previously (67). For transfections, 150 μg of Ncad-Fc cDNA was mixed with 2.5 ml of 5 mM chloroquine diphosphate and then 2.5 ml of 20 μg/ml DEAE dextran. This mix was diluted into 95 ml Dulbecco's modified Eagle's medium (DMEM), and 10 ml was transferred into each of the dishes with COS-7 cells (for 4 h). The solution was aspirated, and the cells were osmotic shocked with 10 ml of 10% dimethyl sulfoxide in phosphate-buffered saline (PBS) for 2 to 3 min. Cells were then washed with DMEM to remove the dimethyl sulfoxide and left overnight in DMEM plus 10% fetal bovine serum (FBS). One day after transfection, the medium was replaced with fresh DMEM (plus 1% FBS) and the Ncad-Fc chimera was allowed to accumulate in the medium for 7 days. To purify the protein, the medium was collected, centrifuged to remove cell debris, and mixed for 2 days with 0.25 g of protein A-Sepharose (Sigma). The beads with bound Ncad-Fc were loaded onto a column and washed with 30 ml PBS. Glycine-HCl, pH 2.7 (100 mM), was used to cleave off the Ncad-Fc. Fractions (0.5 ml) were collected into tubes containing 50 μl 1 M Tris-HCl, pH 9.0, to neutralize the solution, and absorbance at 280 nm was used to identify the Ncad-Fc-containing fractions.

Preparation of mouse brain lysate.

Whole brains were dissected from 2- to 3-week-old mice, quick-frozen in liquid nitrogen, and stored at −80°C. Brain tissues were homogenized in ice-cold lysis buffer (150 mM NaCl, 50 mM HEPES, 1% Triton X-100, 10% glycerol, 1.5 mM MgCl₂, 1.0 mM EGTA, protease inhibitors [1 mM phenylmethylsulfonyl fluoride, 10 μg/ml each of aprotinin, leupeptin, and pepstatin A]) supplemented with 1 mM sodium orthovanadate and 5 mM iodoacetic acid (buffer A) to prevent tyrosine dephosphorylation of proteins. Any unreacted iodoacetic acid was inactivated with 10 mM dithiothreitol. Lysates were centrifuged at 14,000 rpm for 10 min at 4°C to remove insoluble debris. Protein concentrations in the cleared lysates were determined using the Bradford assay (Bio-Rad).

Cell culture, transfection, and preparation of cell lysates.

HEK293T cells were cultured in DMEM plus 10% FBS. Transient transfection of N-cadherin and PTPσ into HEK293T cells was performed using the calcium phosphate method. Prior to lysis, cells were treated with 1 mM pervanadate (fresh solution of H₂O₂ and sodium orthovanadate) for 10 min. After being washed with ice-cold PBS, cells were lysed in ice-cold buffer A and precleared by centrifugation at 14,000 rpm for 10 min at 4°C.

Substrate trapping assays.

Two milligrams of brain lysates from PTPσ^−/− or sibling control (PTPσ^+/+ οr PTPσ^+/−) mice was incubated with ∼20 μg each of GST, GST-PTPσD1(WT), and GST-PTPσD1(DA) fusion proteins for 1 to 2 h at 4°C. For the vanadate competition assay, the GST and GST-PTPσD1 fusion proteins were first incubated in 10 mM sodium orthovanadate for 1 h prior to incubation with lysates. Precipitates (pull-downs) were washed twice each with lysis buffer and HNTG (20 mM HEPES, pH 7.5, 150 mM NaCl, 10% glycerol, 0.1% Triton X-100), and proteins were boiled in sample buffer, separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), and either immunoblotted with antiphosphotyrosine (anti-pTyr) antibody (PY99; Santa Cruz) or excised and trypsin digested for mass spectrometry. Mass spectrometry analysis was carried out by MDS Proteomics, as described previously (23).

IPs and Western blot analysis.

For immunoprecipitations (IPs) with anti-pTyr antibody, 20 μl of the antibody-conjugated agarose was mixed with 1 to 2 mg lysates in lysis buffer for 1 h at 4°C. Beads were then washed twice each in lysis buffer and HNTG and eluted in sample buffer. For IPs with monoclonal N-cadherin and β-catenin antibodies (obtained from BD), 2 μg antibodies was incubated with 20 μl protein G-agarose and lysates as described above. Proteins were then separated by SDS-PAGE, transferred to a nitrocellulose membrane, and immunoblotted with anti-pTyr (1:2,000, blocked in 5% bovine serum albumin [BSA] in Tris-buffered saline), anti-N-cadherin (1:2,000), or anti-β-catenin (1:1,500) antibodies, with the last two blocked in 5% nonfat dry milk in Tris-buffered saline. Blots were incubated with anti-mouse horseradish peroxidase-conjugated antibody (1:20,000; Calbiochem) and visualized by ECL-Plus (PerkinElmer). For IPs with boiled lysates, prior to the IP, 2.5 mg of brain lysates was boiled in 6% SDS (for 10 min) to dissociate interacting proteins. Boiled lysates diluted in lysis buffer (<0.2% SDS, final) were then added to 20 μl of agarose-conjugated anti-pTyr antibodies and mixed overnight (for ∼18 h) at 4°C.

In vitro dephosphorylation assay.

N-cadherin and β-catenin were immunoprecipitated from lysates of HEK293T cells overexpressing N-cadherin, which were pretreated with 1 mM pervanadate to prevent Tyr dephosphorylation of N-cadherin and β-catenin in the cells. Following extensive washes of the immunoprecipitates, half of the immunoprecipitates were incubated with ∼5 μg of active GST-PTPσD1 in dephosphorylation buffer [100 mM 2-(N-morpholino)ethanesulfonic acid (MES), pH 5.5, 10 mM dithiothreitol, 150 mM NaCl, 2 mM EDTA] for 1 h at room temperature. The reaction was stopped by adding sample buffer. Proteins were then separated by SDS-PAGE, and the tyrosine phosphorylation states of the proteins were analyzed by immunoblotting with anti-pTyr antibodies.

Primary cultures of DRG.

Cervical, thoracic, and lumbar dorsal root ganglia (DRG) were isolated from 3- to 4-day-old PTPσ^−/− and sibling control PTPσ^+/+ or PTPσ^+/− mice. Isolated DRG were incubated in 0.25% trypsin (Gibco) in calcium- and magnesium-free Puck's saline G, pH 7.4 (140 mM NaCl, 5.4 mM KCl, 1 mM Na₂HPO₄ · 7H₂O, 1 mM KH₂PO₄, and 3.2 mM glucose), for 30 min at 37°C. Trypsin was then carefully aspirated, and the DRG were incubated with 0.2% collagenase (Worthington) in saline G (with 0.1 mM Ca²⁺ and 0.6 mM Mg²⁺) for 10 min at 37°C. After collagenase removal, the DRG were resuspended in 1 ml neural medium (F-12:MEM [where MEM is minimal essential medium] [Gibco]) supplemented with 10 mM HEPES, pH 7.3, 0.6% glucose, 25 ng/ml nerve growth factor (Roche), 2× N2 solution (Gibco), 110 μg/ml pyruvic acid, and 1 mg/ml ovalbumin (Sigma). For the low-calcium experiments, the MEM was replaced with S-MEM (Gibco), which contains no calcium (final calcium concentration was 0.14 mM, compared to 1.05 mM in the regular neural medium). Next, the DRG explants were gently dissociated by trituration with a glass Pasteur pipette (roughly 50 to 100 passages). Cells were then seeded at a concentration of ∼5,000 to 7,000 cells per well on six-well plates. For the N-cadherin blocking experiments, 0.1 mg/ml of inhibitory peptide (see below) was added to the neural medium during cell plating. Cells were plated on either N-cadherin-coated or poly-d-lysine plus laminin (PDLL)-coated dishes. Preparation of the N-cadherin-coated plates was performed as described elsewhere (67). Briefly, 10 μg/ml anti-human rabbit IgG, Fc specific (Jackson), in PBS was first used to coat the plates, followed by addition of 1% BSA to block nonspecific binding, PBS washes, and addition of 20 μg/ml Ncad-Fc (4°C, overnight). Ncad-Fc was rinsed off with F-12:MEM prior to adding the cells. For PDLL-coated plates, 10 μg/ml poly-d-lysine (Sigma) in water was added to the plates overnight (room temperature). The next day, the plates were rinsed with water and allowed to dry before addition of 10 μg/ml laminin (Invitrogen) in PBS for 4 h. Prior to plating cells, laminin was removed and plates rinsed with F-12:MEM.

N-cadherin blocking peptides.

The linear peptide N-acetyl-INPISGQ-NH₂, corresponding to amino acids 212 to 218 in the N-cadherin sequence (GenBank accession no. BAA23549), has previously been shown to effectively inhibit N-cadherin function (76). This peptide was generated at the Advanced Protein Technology Centre of the Hospital for Sick Children (Toronto, Ontario, Canada).

Quantification of neurite outgrowth.

Growth of DRG cells was monitored with an inverted microscope (Leica). Neurons were randomly chosen, and images were captured using a charge-coupled-device camera (QImaging) with OpenLab software (Improvision). To assess general neurite growth, the numbers of cells with and without neurites extending from their cell body were counted. A cell was counted positive if it had neurites greater than twice its body length.

Quantification of SC-axon alignment.

Schwann cells (SCs) were scored to align with axons when at least two of their processes overlapped with neurites. The results are expressed as percentages of SCs aligned with axons.

LacZ staining.

DRG cultured cells were washed twice with PBS, fixed with 0.5% glutaraldehyde (Sigma) for 15 min (room temperature), washed with PBS (3×), and stained with X-Gal solution (5 mM ferricyanide, 5 mM ferrocyanide, 2 mM MgCl₂, 1.0 mg/ml X-Gal [5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside] in PBS) for 8 h (37°C) in the dark. The reaction was stopped by rinsing the cells with PBS.

RESULTS

Hyper-tyrosine-phosphorylated proteins in brain lysates of PTPσ knockout mice.

To identify potential in vivo substrates for PTPσ, we searched for proteins that are hyperphosphorylated on tyrosine residues in the PTPσ knockout mice; these proteins are likely substrates of PTPσ because they cannot be dephosphorylated in its absence. Figure 1 demonstrates the presence of several such hyper-Tyr-phosphorylated proteins/bands in brain lysates (which highly express PTPσ) obtained from the PTPσ knockout mice relative to sibling controls (PTPσ^+/+ or PTPσ^+/−). PTPσ^+/− mice do not show any adverse phenotype and are identical to the PTPσ^+/+ animals in all aspects studied (8, 35, 69). Substrate trapping (24) with a catalytically inactive first phosphatase domain of PTPσ [PTPσ-D1(DA)] was subsequently carried out to capture these hyper-Tyr-phosphorylated proteins from brain lysates of the PTPσ knockout animals. The band at ∼120 kDa was cut out, trypsin digested, and analyzed by tandem mass spectrometry by MDS Proteomics, as detailed elsewhere (23). The sequence identified was that of N-cadherin.

FIG. 1.

Brain lysates from PTPσ^−/− mice contain several hyper-tyrosine-phosphorylated proteins. Equal amounts of protein from brain lysates of PTPσ knockout (−/−) mice and their sibling (Sib) controls (PTPσ^+/+ and PTPσ^+/−) were either separated by SDS-PAGE and immunoblotted with anti-pTyr antibodies or immunoprecipitated with anti-pTyr antibodies prior to immunoblotting with the same antibodies. The arrow indicates the protein band (∼120 kDa) that was cut out for further analysis by mass spectrometry and identified as N-cadherin.

N-cadherin is a substrate of PTPσ.

To validate the mass spectrometry identification, we tested substrate trapping of N-cadherin with PTPσ-D1(DA). PTPσ-D1(DA) or PTPσ-D1 (WT control) was incubated with brain lysates from the PTPσ^−/− mice or sibling controls, and the precipitated proteins were immunoblotted with anti-N-cadherin antibodies. As seen in Fig. 2A and B, PTPσ-D1(DA) was able to trap N-cadherin from brain lysates of the PTPσ knockout animals (Fig. 2B) but not from lysates of the sibling controls (Fig. 2A). Moreover, the control PTPσ-D1(WT) fusion protein was able to trap only a small amount of N-cadherin relative to the amount trapped by the catalytically inactive PTPσ-D1(DA) fusion protein (Fig. 2B, right panel). As expected, substrate (N-cadherin) trapping was impaired in the presence of vanadate (Fig. 2D). To verify that the N-cadherin trapped is Tyr phosphorylated, Tyr-phosphorylated proteins from brain lysates of the PTPσ knockout mice or their sibling controls were immunoprecipitated with anti-pTyr antibodies and the precipitate immunoblotted with N-cadherin antibodies. Figure 2C demonstrates that Tyr-phosphorylated N-cadherin was indeed precipitated from the PTPσ knockout animals but not from the sibling controls. In support of the substrate trapping experiments with brain lysates, we were also able to trap N-cadherin expressed in HEK293T cells with PTPσ-D1(DA), as well as with PTPσ-D1(DA)D2 (i.e., when both intracellular catalytic domains are used for trapping) (Fig. 2E). Together, these results suggest that tyrosine-phosphorylated N-cadherin binds to and is an in vivo substrate for PTPσ.

FIG. 2.

N-cadherin is an in vivo and an in vitro substrate of PTPσ. (A and B) Immobilized PTPσ substrate trapping mutant GST-σD1(DA) and the corresponding wild type, GST-σD1(WT), as well as GST alone, were incubated with brain lysates from (A) sibling (Sib) controls or (B) PTPσ knockout (−/−) mice. Precipitated proteins were then separated by SDS-PAGE and immunoblotted with anti-pTyr or anti-N-cadherin (anti-N-cad) antibodies, as indicated. (C) Tyrosine-phosphorylated N-cad is detected in brain lysates of PTPσ knockout mice but not sibling controls. Brain lysates from PTPσ knockout mice or sibling controls were immunoprecipitated with anti-pTyr antibodies and immunoblotted with anti-N-cad or anti-pTyr antibodies. (D) Vanadate (VO₄) reduces trapping of N-cad. Brain lysates from PTPσ^−/− mice were incubated with the indicated trapping constructs in the presence or absence of the PTP inhibitor vanadate (10 mM), and precipitated proteins were separated by SDS-PAGE and immunoblotted with anti-N-cad antibodies. (E) N-cad is trapped with catalytically inactive first or both PTPase domains of PTPσ. HEK293T cells transfected with N-cad were lysed, lysates were incubated with the GST fusion protein, active (WT) or inactive (DA mutant), of the first (D1) or both (D1D2) catalytic domains of PTPσ, and precipitated proteins were immunoblotted for N-cad. Since the second domain (D2) is naturally inactive, only the first domain (D1) was inactivated using the DA mutation [GST-σD1(DA) and GSTσD1(DA)D2 represent the inactive first or both catalytic domains, respectively]. For panels A, B, D, and E, the lower parts of the panels depict Ponceau S staining of the blots, shown to verify that equal amounts of substrate trapping proteins were used for precipitating N-cad. (F) Ectopic expression of PTPσ in cells leads to dephosphorylation of endogenous N-cad. Flag-tagged PTPσ (full length, D1, or D1D2; all WT) were transfected into HEK293T cells, the cells were lysed, lysates were incubated with anti-pTyr antibodies, and precipitated proteins were immunoblotted with anti-N-cad antibodies to detect the extent of N-cad dephosphorylation. The lower panel shows the levels of expression of the indicated Flag-tagged proteins. tfxn, transfection. (G) PTPσ can directly dephosphorylate N-cad. N-cad overexpressed in HEK293T cells was immunopurified with N-cad antibodies and incubated (or not) with purified, GST-tagged PTPσD1(WT). Proteins were then immunoblotted with anti-pTyr antibodies to analyze the extent of N-cad dephosphorylation or with anti-N-cad antibodies to demonstrate equal amounts of N-cad precipitated in the experiment.

To further verify that PTPσ can directly dephosphorylate N-cadherin, we transfected PTPσ into HEK293T cells (full length, the first catalytic domain [D1, which is active], or both domains [D1D2]) and tested for the ability to dephosphorylate endogenous N-cadherin. As seen in Fig. 2F, ectopic expression of PTPσ (or its intracellular catalytic domains) in these cells led to a decrease in Tyr phosphorylation of N-cadherin. Furthermore, the addition of purified PTPσ-D1 domain to immunopurified and Tyr-phosphorylated N-cadherin led to dephosphorylation of N-cadherin in vitro (Fig. 2G), suggesting that N-cadherin is a direct substrate of PTPσ. Collectively, these data suggest that PTPσ can dephosphorylate N-cadherin in vitro and in vivo.

β-Catenin is also a substrate of PTPσ.

The intracellular region of N-cadherin (and other cadherins) is known to bind to β-catenin, itself associated with α-catenin, which binds actin (1, 15). This allows a connection between N-cadherin and the actin cytoskeleton, promoting regulation of cell adhesion and motility by the cadherin-catenin complex. We first verified that N-cadherin and β-catenin can interact with each other by use of coimmunoprecipitation assays, as shown in Fig. 3A. Since the antibodies used in this experiment cannot distinguish between phosphorylated and nonphosphorylated species of N-cadherin and β-catenin, it is likely that these proteins bind each other in the nonphosphorylated state, as previously suggested (7, 10, 45, 50).

FIG. 3.

β-Catenin binds N-cadherin (N-cad) in vivo and is also a substrate for PTPσ. (A) Coimmunoprecipitation of N-cad with β-catenin (β-cat). Brain lysates from sibling (Sib) control or PTPσ^−/− (−/−) mice were immunoprecipitated with anti-N-cad antibodies. The immunoprecipitates were separated by SDS-PAGE and immunoblotted with either anti-N-cad or anti-β-cat antibodies. (B) Substrate trapping of β-cat. PTPσ^−/− brain lysates were incubated with immobilized GST or the PTPσ substrate trapping mutant GST-σD1(DA). The associated proteins were then analyzed by probing with antibodies against N-cad or β-cat. Ponceau S staining to visualize the amount of GST or GST-σD1(DA) substrate trapping proteins used is shown in the lower panel. (C and D) Hyper-tyrosine phosphorylation of β-cat in knockout mice. Brain lysates from Sib or PTPσ^−/− mice were immunoprecipitated with anti-pTyr antibodies and immunoblotted for β-cat. For panel D, lysates were boiled in 6% SDS prior to the IP. (E) Vanadate (VO₄) impairs substrate trapping of β-catenin. Brain lysates from PTPσ^−/− mice were incubated with the indicated trapping constructs in the presence or absence of the PTP inhibitor vanadate (10 mM), and precipitated proteins were separated by SDS-PAGE and immunoblotted with anti-β-cat antibodies. Ponceau S staining to verify equal amounts of trapping proteins is shown in Fig. 2D. (F) Ectopic expression of PTPσ in cells leads to dephosphorylation of endogenous β-cat. Flag-tagged PTPσ constructs (full length, D1, or D1D2; all WT) were transfected into HEK293T cells, the cells were lysed, lysates were incubated with anti-pTyr antibodies, and precipitated proteins were immunoblotted with anti-β-cat antibodies to detect β-cat dephosphorylation. Levels of expression of the indicated Flag-tagged proteins are depicted in Fig. 2F. tfxn, transfection. (G) PTPσ can directly dephosphorylate β-cat. β-Catenin from HEK293T cells overexpressing N-cad was immunopurified with β-cat antibodies and incubated (or not) with purified, GST-tagged PTPσD1(WT). Proteins were then immunoblotted with anti-pTyr antibodies to analyze β-cat dephosphorylation or with anti-β-cat antibodies to demonstrate similar amounts of precipitated β-cat in the experiment.

To test whether β-catenin itself is a substrate for PTPσ, the interaction trap experiment was repeated, where PTPσ-D1(DA) was incubated with brain lysate from the PTPσ knockout mice, leading to trapping of N-cadherin (Fig. 3B, top panel). The blot was then stripped and reprobed for β-catenin. As seen in Fig. 3B, middle panel, β-catenin was also trapped with the PTPσ-D1(DA) protein, suggesting that β-catenin itself may be a substrate for PTPσ. To test if β-catenin is indeed a subject for PTPσ-mediated dephosphorylation, we analyzed its level of Tyr phosphorylation in the lysates of our knockout animals. Figure 3C shows that although only a small fraction of β-catenin is tyrosine phosphorylated in the total cellular pool, there was more Tyr-phosphorylated β-catenin in the knockout mice than in the sibling control. This was not simply an effect of more β-catenin coprecipitating with the increased amount of Tyr-phosphorylated N-cadherin in these animals (Fig. 3B), because boiling the lysate in 6% SDS prior to the immunoprecipitation with anti-pTyr antibodies (to dissociate the N-cadherin:β-catenin complex) still revealed elevated Tyr phosphorylation of β-catenin (Fig. 3D). Similarly to results with N-cadherin, trapping of β-catenin with PTPσ-D1 was largely inhibited in the presence of vanadate (Fig. 3E). Together, these results suggest that β-catenin itself is also a substrate for PTPσ. In support, ectopic expression of PTPσ in HEK293T cells led to dephosphorylation of endogenous β-catenin (Fig. 3F); moreover, immunopurified β-catenin was directly dephosphorylated by PTPσ-D1 in vitro (Fig. 3G). Thus, in addition to N-cadherin, β-catenin appears to be a target for PTPσ-mediated dephosphorylation.

Axon growth rate of DRG neurons is accelerated in PTPσ knockout mice.

Our previous work has demonstrated an accelerated rate of axon growth in the sciatic nerve of the PTPσ knockout mice following nerve injury (35). Since DRG have been demonstrated to express high levels of N-cadherin in both neurons and Schwann cells (57, 74), we focused our studies on DRG. These cells also express PTPσ (Fig. 4A) (35), as assessed by LacZ staining reporting an endogenous pattern of expression of this phosphatase, since the knockout cassette includes the β-galactosidase gene downstream of the endogenous promoter (69). To quantify axon growth rate, we isolated DRG from the knockout animals and their sibling controls, dissociated their cells, and grew them as primary cultures on PDLL-coated plates. As seen in Fig. 4B, axon growth was faster in the DRG cells from the PTPσ knockout mice than in those from the sibling controls, and this difference was especially noticeable after 24 h of growth in culture, suggesting a lag period in the manifestation of the difference in growth rate. The different growth rates between the knockout animals and their sibling controls were apparent in terms of both the number of neurons extending neurites (Fig. 4B) and the total length of neurites per cell (not shown).

FIG. 4.

DRG neurons from PTPσ^−/− mice exhibit a faster rate of neurite outgrowth. (A) LacZ staining of PTPσ^−/− DRG culture, showing endogenous expression (blue) of PTPσ in the cell bodies of neurons (N) and, to a lesser extent, of Schwann cells (S). There was no LacZ staining seen in WT (+/+) DRG (not shown). Bar, 40 μm. (B and C) DRG cultures were prepared from sibling (Sib) controls and PTPσ^−/− mice, and cells were plated on either (B) PDLL or (C) N-cadherin (N-cad). Neurite growth was measured at the indicated times. Data are expressed as the percentages of cells counted that have neurite extensions. A cell is counted positive if it has neurites greater than twice its body length. Data are summaries of seven independent experiments, and a total of at least 750 cells were counted for each condition. Error bars represent 95% confidence intervals. **, P < 0.0001 (chi-square test).

N-cadherin differentially affects growth of DRG neurons from PTPσ^−/− mice versus sibling controls.

Since our biochemical studies described above suggest that N-cadherin is a substrate for PTPσ and since N-cadherin can support neurite growth in vitro (9, 12, 67), we investigated the effect of N-cadherin on growth of DRG neurons harvested from the knockout mice and their sibling controls by culturing the neurons on purified N-cadherin. Culturing these neurons on control substrata such as BSA or Fc antibody did not support their growth (data not shown). The accelerated growth exhibited by the PTPσ knockout neurons compared to that of the sibling control neurons when grown on PDLL (Fig. 4B) was partially attenuated (delayed) on the N-cadherin substratum (Fig. 4C). These results are consistent with the idea that N-cadherin can support neurite growth of peripheral sensory neurons. Moreover, they suggest that PTPσ may collaborate with N-cadherin to exert its inhibitory effect on neurite growth, since this inhibition was slower/less effective upon loss of PTPσ expression in the knockout animals.

To further investigate this hypothesis, we utilized the N-cadherin inhibitory peptide INPISGQ (76). This peptide is derived from the ectodomain of N-cadherin, binds to a sequence near the HAV motif necessary for N-cadherin homophilic interactions, and is unique to N-cadherin among the cadherin family (56, 76). Interference with the HAV motif by this and other peptides provides a potent antagonistic effect to N-cadherin function in cells, including DRG neurons and Schwann cells (74, 75). When the N-cadherin function was disrupted by the inhibitory peptide, sibling control DRG neurons cultured on PDLL exhibited a marked increase in neurite growth (Fig. 5A and B). In contrast, this effect was much smaller (and not statistically significant) in the PTPσ knockout neurons (Fig. 5C and D). These results suggest that the already accelerated growth of the PTPσ knockout neurons cannot be further enhanced by inhibiting N-cadherin function, suggesting that PTPσ and N-cadherin may be part of the same signaling pathway that normally inhibits (or slows down) neurite growth.

FIG. 5.

N-cadherin blocking peptide causes increased neurite outgrowth on PDLL of siblings but not PTPσ^−/− neurons. DRG neurons from (A and B) sibling (Sib) controls or (C and D) PTPσ^−/− mice were cultured on PDLL and treated with (gray bars) or without (black or white bars) 0.1 mg/ml of the N-cadherin (N-cad) inhibitory peptide INPISGQ. In panels A and C, data represent the percentages of cells counted that have neurite extensions at the indicated times, and these results are normalized to no-treatment controls in panels B and D, respectively. Data are summaries of four or five independent experiments, with a total of at least 500 cells counted for each condition. Error bars represent 95% confidence intervals. *, P < 0.001; **, P < 0.0001 (chi-square test). (E and F) As for panels A and C, only DRG were cultured on N-cad substratum. Data are summaries of four independent experiments, and a total of at least 500 cells were counted for each condition. Error bars represent 95% confidence intervals.

When the neurons were cultured on N-cadherin, the inhibitory peptide had no significant effect on the growth of either group of neurons (Fig. 5E and F), most likely because the amount of peptide used could not compete with the N-cadherin used as a substratum (larger amounts of N-cadherin inhibitory peptide were toxic to cells and hence could not be used).

Difference in growth rate observed between DRG neurons from the knockout and sibling mice is abolished in low calcium.

The role of N-cadherin as a cell-cell adhesion molecule is calcium dependent. Calcium is required to stabilize the extracellular domains of N-cadherin to allow homodimerization (64). Lowering the extracellular calcium concentration is one way to perturb N-cadherin function, and this was shown to be an effective means to disrupt Schwann cell contact with axons, a process that is dependent on N-cadherin (74). We studied the role of N-cadherin in DRG neurite outgrowth by reducing the calcium concentration from 1.05 mM in the normal neuronal medium to 0.14 mM, a concentration that would render the calcium-dependent N-cadherin nonfunctional. As expected, compared to cells cultured on N-cadherin in normal neuronal medium, the DRG neurons cultured in the low-calcium medium adhered loosely onto the surface, they did not extend neurites, and the Schwann cell processes were much shorter (see Fig S1, bottom panels, in the supplemental material). When the cells were cultured on PDLL, the effect was reversed. The number of cells extending neurites in the low-calcium medium was increased more than 1.5-fold at each time point for the sibling control cells and increased only slightly for the PTPσ knockout cells (Fig. 6A to D). Moreover, the difference that we observed in the growth rates between the sibling controls and PTPσ knockout neurons in normal calcium medium was greatly reduced under low-calcium conditions (compare Fig. 4B to Fig. 6E). This indicates that the PTPσ knockout neurons are not sensitive to lowering of extracellular calcium level, likely because the N-cadherin function/pathway in the PTPσ knockout neurons is already impaired. These results are consistent with those presented in Fig. 5A to D and suggest that disruption of N-cadherin function by lowering calcium levels or adding an N-cadherin inhibiting peptide lead to the same outcome, an acceleration of neurite growth, which is dependent on PTPσ, since it is lost in the PTPσ knockout DRG.

FIG. 6.

Low-calcium medium leads to increased neurite outgrowth on PDLL of siblings but not PTPσ^−/− neurons. Sibling (Sib) control (A and B) and PTPσ^−/− (C and D) DRG neurons grown on PDLL were cultured in normal neuronal medium (1.05 mM Ca²⁺) (black or white bars) or low-calcium medium (0.14 mM Ca²⁺) (gray bars). In panels B and D, data represent percentages of cells with neurite extensions normalized to those in panels A and C, respectively. (E). Neurite growth on PDLL in low-calcium medium (0.14 mM Ca²⁺) abolishes the difference in growth rate observed between the sibling control and PTPσ^−/− neurons. Data are summaries of six independent experiments, and a total of at least 800 cells were counted for each condition. Error bars represent 95% confidence intervals. *, P < 0.001; **, P < 0.0001 (chi-square test).

Knockout of PTPσ does not disrupt axon-Schwann cell alignment.

In the PNS, growth of neurons is supported by SCs. SC-SC and SC-axon interactions are essential during early nerve development and in the organization of the peripheral nervous tissues. These interactions have previously been shown to involve N-cadherin (32, 74). Alignment of SCs along neurites in cultured chicken DRG was disrupted upon treatment with N-cadherin blocking antibodies and by lowering the calcium concentration in the medium (32). We thus investigated whether tyrosine phosphorylation of N-cadherin and PTPσ has a role in this alignment process. DRG neurons were isolated from both sibling control and PTPσ knockout mice and cultured on PDLL or purified N-cadherin. After 2 days, alignment of SCs with axons was assessed. As expected, there was a significant reduction in SC-axon alignment when cells were cultured in low-calcium medium. In a low-calcium environment, of the SCs that made contact with axons, only about 20% aligned their processes along the axons, whereas 55 to 60% were aligned in the cultures with normal medium (Fig. 7A). No difference in alignment of neurons with SCs was observed between the sibling control and the PTPσ knockout neurons when grown on PDLL or N-cadherin substrates in normal medium (Fig. 7A and B). Neurite growth on N-cadherin of both knockout and sibling controls was inhibited in low calcium (see Fig S1, bottom panels, in the supplemental material), as expected.

FIG. 7.

Alignment of Schwann cells to axons is not affected in the PTPσ^−/− DRG. DRG cells from sibling (Sib) control and PTPσ^−/− mice were plated on either (A) PDLL or (B) N-cadherin (N-cad). Alignment of SCs to axons was assessed after the cells were allowed to grow for (A) 40 h or (B) 46 h. An SC is counted positive for alignment when at least two of its processes are aligned with axons. Data represent means ± standard deviations of percentages of SCs aligned with axons from six independent experiments. A total of at least 700 cells were counted for each condition.

Collectively, these results suggest that PTPσ and N-cadherin collaborate to regulate the rate of growth of DRG neurons but not their alignment with Schwann cells.

DISCUSSION

Our work here identified N-cadherin as an in vivo (and in vitro) substrate for PTPσ and suggests that N-cadherin (likely via β-catenin) participates in the inhibitory function of PTPσ in axon growth. We propose that dephosphorylation of N-cadherin (and possibly β-catenin) by PTPσ leads to increased adhesion and attenuated axon growth, an effect reversed in the PTPσ knockout axons (Fig. 8). This conclusion is supported by our observed loss of the difference in axon growth between the PTPσ^−/− mice and their sibling controls upon inhibition of N-cadherin function with an inhibitory peptide or by lowering Ca²⁺ levels. This correlation between our biochemical and axon growth data suggests that PTPσ and N-cadherin are part of the same signaling pathway that regulates rates of axon growth. Interestingly, we found that only the rate of axon growth, not the alignment of axons with Schwann cells (previously shown to be regulated by N-cadherin [74]), was affected in the PTPσ knockout mice, suggesting that PTPσ regulates some but not all functions of N-cadherin.

FIG. 8.

Model for PTPσ-regulated axon growth via the N-cadherin pathway. When PTPσ is present on the plasma membrane, it can dephosphorylate N-cadherin and recruit β-catenin (β-cat) to the complex. β-Catenin itself also becomes dephosphorylated, and this allows the association of the N-cadherin complex with the actin cytoskeleton, resulting in increased adhesion and reduced rate of axon growth. Conversely, when PTPσ is absent (e.g., in the knockout mice), N-cadherin and β-catenin remain phosphorylated on tyrosine residues. This prevents interaction between the two proteins, resulting in a loss of N-cadherin adhesive function and faster rate of axon growth. α-cat, α-catenin.

It is well established that the adhesive function of cadherin depends on its stable connection to the actin cytoskeleton via catenin interactions (1, 2) and that tyrosine phosphorylation of β-catenin leads to its dissociation from cadherin and loss of cadherin adhesiveness (7, 49, 50). Tyrosine phosphorylation of the cadherins themselves has not been well characterized, especially in regard to neurite growth. In this report, we have demonstrated that tyrosine (de)phosphorylation of N-cadherin itself could regulate neurite growth. The cadherin-catenin complex has been reported to play a role in the regulation of neurite growth by other receptor PTPs. LAR (39), PTPβ (37), PTPμ (11, 12), and PTPκ (25) can dephosphorylate β-catenin. Although until now there have not been any reports on cadherins being direct substrates for PTPs, PTPμ and PTP1B have been suggested to associate with N-cadherin and this association is important for the regulation of N-cadherin-dependent axon growth (12, 43). Moreover, LAR was recently shown to indirectly associate with the cadherin-catenin complex via liprin-α and GRIP (20). However, unlike PTPσ, its close relatives PTPδ and LAR were proposed to provide stimulatory cues for neurite growth in forebrain neurons and in hippocampal and DRG neurons, respectively (73, 82, 83). Since LAR can also employ the cadherin-catenin pathway and can trap N-cadherin in a substrate trapping assay (our unpublished data), it is possible that the final outcome (stimulation or inhibition of neurite growth) may depend on levels of expression or activity of the competing PTPs on the regulation of the cadherin-catenin complex.

N-cadherin is strongly expressed in the growth cone, and the growth cone has been shown to have intense phosphotyrosine staining (79). By disrupting N-cadherin connection to the actin cytoskeleton, thereby reducing adhesive strength, this might allow the growth cone to advance. Based on this model, phosphorylation of N-cadherin might result in its disconnection from the actin cytoskeleton by several possible means, which are not mutually exclusive. (i) It may lead to reduction in binding to β-catenin. (ii) It may impair cadherin homodimerization, a step that is believed to be essential in driving cadherin adhesion function. (iii) It may alter interactions with other signaling molecules.

Our work shows that PTPσ has a role in the signaling events activated by laminin. Interestingly, the PTPσ close relative LAR was shown to bind laminin via its fifth FNIII repeat, an interaction requiring the splicing of a small exon within this domain (41). Since it appears that the alternative splicing of this exon is conserved among the LAR family of PTPs, it suggests that laminin might also serve as a ligand for PTPσ. Moreover, β1 integrin has been shown to associate with tyrosine-phosphorylated proteins at the growth cone (80), and α1β1 and α3β1 integrin heterodimers are the laminin receptors in DRG neurons (66), further suggesting that tyrosine phosphorylation accompanies laminin signaling and that PTPσ might be involved in regulating this signal. It was interesting that the growth of the DRG neurons on laminin in our studies was affected by an N-cadherin peptide inhibitor. This suggests overlapping responses in the signaling events triggered by these two molecules. Indeed, there has been some evidence for cross talk between the N-cadherin and laminin signaling pathways (5, 33, 44, 48). Our observed accelerated neurite growth on laminin (of sibling controls) in the presence of the N-cadherin inhibitory peptide or upon reduction of calcium in the medium may be explained by an inhibition of N-cadherin lateral dimerization/clustering. Indeed, the monomeric inhibitory peptide we used here was previously suggested to interfere with N-cadherin clustering (77). Such inhibition leads to reduced adhesiveness (64) by affecting the actin cytoskeleton (which is shared with the laminin-integrin pathway), allowing faster neurite growth. This faster neurite outgrowth is also mimicked by loss of PTPσ, which we propose normally dephosphorylates N-cadherin, leading to increased adhesiveness.

While we show here that N-cadherin (and β-catenin) is a substrate for PTPσ, we cannot currently preclude the possibility that PTPσ has other substrates. Studies with flies have revealed genetic interaction between the tyrosine kinase Abl and DLAR (78), although it is not known if Abl is a direct substrate for this phosphatase, and our unpublished studies failed to demonstrate physical interactions between mammalian PTPσ and Abl. Several LAR interacting proteins have been identified to date, including liprins and the Rac exchange factor Trio (16, 40, 46, 54, 55), both collaborating genetically with DLAR in flies, although they are not its substrates (28). The existence of additional PTPσ substrates in mammals is suggested by the presence of several hyper-Tyr-phosphorylated proteins/bands in the brain lysates of knockout mice in addition to N-cadherin (Fig. 1). Their identification and characterization await future investigations.