IPP5 inhibits neurite growth in primary sensory neurons by maintaining TGF-β/Smad signaling

Neurite growth is a critical step for successful nerve regeneration after injury. Extracellular factors surrounding the ends of injured axons were previously considered the major reasons affecting axonal growth (Filbin, 2003; Schwab, 2004; Yiu and He, 2006). However, recent evidence indicates that the activation of intrinsic molecules also play an important role (Abe and Cavalli, 2008; Chen et al., 2007; Liu et al., 2011; Sun and He, 2010). Several protein kinases, transcription factors and growth-associated molecules have been shown to be intrinsic molecules that positively regulate axon growth (Chen et al., 2007; Gao et al., 2004; Lorber et al., 2009; MacGillavry et al., 2009; Moore et al., 2009; Seijffers et al., 2007). Downregulation or low activity of these positive molecules causes reduced axonal growth and failure of nerve regeneration. However, intrinsic molecules that negatively regulate axonal growth are also considered important when the cell is unable to eliminate or even activate these molecules following nerve injury (Abe and Cavalli, 2008; Liu et al., 2011). The transforming growth factor-β (TGF-β)/Smad signaling pathway inhibits axon growth (de la Torre-Ubieta and Bonni, 2011). Knocking down Smad2 or blocking the TGF-β/Smad signaling pathway with SB431542 or specific antibodies enables axons to override myelin inhibition and promotes functional recovery after spinal cord injury (He and Wang, 2006; Kohta et al., 2009; Stegmüller et al., 2008).

Protein phosphatase 1 (PP1) is a eukaryotic protein serine/threonine phosphatase that is important in regulating neuronal morphology. Inactivation of PP1 by phosphorylation at Thr³²⁰ or with a specific inhibitor, calyculin A, dramatically affects nerve growth factor-induced neurite growth in PC12 cells (Li et al., 2007; Reber and Bouron, 1995). In the central nervous system, F-actin-associated PP1 promotes spine development and axonal growth (Bielas et al., 2007; Oliver et al., 2002). Generally, each PP1 holoenzyme is composed of a catalytic subunit and a regulatory subunit. The regulatory subunit determines the substrate specificity, subcellular localization, and diverse cellular functions (Bollen et al., 2010). Previous studies report that the TGF-β/Smad signaling pathway can be specifically regulated by PP1. Under the control of the regulatory subunit GADD34, PP1 is targeted to the type I transforming growth factor-β receptor (TβRI) and inactivates this Ser/Thr kinase receptor through dephosphorylating the juxta-membrane region (Shi et al., 2004).

IPP5, a newly identified regulatory subunit of PP1, belongs to the protein phosphatase 1 regulatory subunit 1 (PPP1R1) family, which has a highly conserved PP1-docking motif and a PP1-inhibiting motif. IPP5 inhibits the enzymatic activity of PP1 after being phosphorylated within its PP1-inhibiting motif at Thr³⁴ (Wang et al., 2008). IPP5 promotes tumor cell cycle progression by accelerating the G1-S transition in a PP1-dependent manner (Wang et al., 2008), inhibits anchorage-dependent growth and induces apoptosis of HeLa cells (Zeng et al., 2009). However, the function of IPP5 in the nervous system has not been investigated. In the present study, IPP5 was shown to be selectively expressed in the dorsal root ganglion (DRG) neurons of rats. The DRG contains pseudounipolar sensory neurons that extend their peripheral terminals to peripheral tissues and their central terminals to the dorsal horn of the spinal cord to convey somatic sensory signals. Generally, DRG neurons cultured in vitro generate multiple neurites, which reflects the growth ability of these sensory neurons. We found that IPP5 inhibits neurite growth in cultured DRG neurons by inhibiting PP1 activity and maintaining TGF-β/Smad signaling. Blocking TGF-β/Smad signaling and PP1 function impaired the IPP5-induced inhibition of neurite growth. Thus, IPP5 is a novel intrinsic molecule that negatively regulates neurite growth in primary sensory neurons.

We searched the Unigene library of the National Center for Biotechnology Information (NCBI) to identify genes that are highly expressed in the DRG but are expressed at relatively low levels in other tissues. In this process, we identified IPP5. By applying the reverse transcription polymerase chain reaction (PCR), we further confirmed that IPP5 was highly expressed in the DRG of rats but showed no detectable expression in other regions of the central nervous system, including the cortex, cerebellum, hypothalamus, hippocampus, medulla oblongata, pituitary gland and spinal cord (Fig. 1A). Amino acid sequence alignment showed that IPP5 belonged to the PPP1R1 family, possessing a highly conserved PP1-docking motif K⁸IQF¹¹ and a PP1-inhibiting motif R³¹RPTPA³⁶ (Fig. 1B). We also analyzed the expression patterns of the other two members of the PPP1R1 family, inhibitor-1, and dopamine and cAMP-regulated phosphoprotein M_r 32 kDa (DARPP-32), which exhibited universal distributions in nervous system but low expression in the DRG of rats (Fig. 1A).

To further explore the cell-type characterization of IPP5 in the DRG, we produced a rabbit polyclonal antibody against IPP5 and confirmed its specificity with antigen preabsorption (supplementary material Fig. S1A). Immunohistochemistry showed that 54.4±2.5% of the DRG neurons were IPP5 positive (n = 3856 out of 7414 counted neurons). We then assessed IPP5 expression in two major subsets of small DRG neurons with unmyelinated C-fibers or myelinated Aδ-fibers, which are known as isolectin B4 (IB4)-positive nonpeptidergic neurons and calcitonin gene-related peptide (CGRP)-positive peptidergic neurons, respectively. Double immunofluorescence staining showed that 55.9±2.1% of IPP5-positive neurons were IB4 positive (n = 2820 out of 5073 counted neurons) and 25.7±1.2% were CGRP positive (n = 1143 out of 4397 counted neurons; Fig. 1C). Of these, 49.3±1.7% of the IPP5-positive neurons (n = 1664 out of 3358 counted neurons) were labeled with P2X₃ receptor, a purinergic receptor that is selectively expressed in nonpeptidergic small DRG neurons (supplementary material Fig. S1C,D). Additionally, 14.4±0.9% of IPP5-positive neurons (n = 606 out of 4179 counted neurons) contained neurofilament 160/200 (NF160/200), which labels large DRG neurons (diameter >35 µm) with myelinated Aβ-fibers (Fig. 1C). Consistently, double immunofluorescence staining in the dorsal horn of the spinal cord showed abundant IPP5-positive nerve fibers in laminae I–II and a few IPP5-positive nerve fibers in laminae III–IV (Fig. 1D). Most IPP5-positive nerve fibers were located in the inner laminae II, an area that exhibited IB4 binding (Fig. 1D). Thus, IPP5 is predominantly expressed in small primary sensory neurons that mainly project to laminae I–II of the spinal cord.

We performed both real-time PCR and immunoblotting to analyze the expression of IPP5 in L4 and L5 DRGs of rats after sciatic nerve axotomy. The quantitative data from real-time PCR showed that the mRNA level of IPP5 in the DRGs was reduced to approximately 60% at post-nerve injury day 2 and maintained at the same low levels until day 28 (Fig. 1E). Immunoblotting further revealed that the reduction of IPP5 mRNA in the DRGs was accompanied by decreased protein levels of IPP5. The immunoblotting band with a molecular weight of approximately 16 kD was confirmed to be IPP5 by antigen preabsorption (supplementary material Fig. S1B). The protein levels in the DRGs were reduced to approximately 20% at post-nerve injury day 7 and gradually recovered to approximately 60% at post-nerve injury day 14 (Fig. 1F). Thus, IPP5 is remarkably downregulated in DRG neurons after nerve injury and may serve as a neurite growth inhibitory molecule in primary sensory neurons.

To examine the effect of IPP5 on neurite growth, we used a plasmid-based method of IPP5 shRNA (Fig. 2A) to knock down endogenous IPP5 in rat primary sensory neurons. We electroporated scrambled or IPP5 shRNA plasmids into dissociated DRG neurons. The IPP5 protein levels were significantly decreased two days after transfection with IPP5 shRNA plasmids (Fig. 2B), indicating that the construct efficiently interfered with endogenous IPP5 expression. We also constructed a shRNA-resistant (R) IPP5 (IPP5^R, Fig. 2A) that was not silenced by IPP5 shRNA. The expression level of IPP5 was unchanged after coexpressing the IPP5 shRNA plasmid with IPP5^R in ND7-23 cells (Fig. 2C), indicating that IPP5^R is not targeted by IPP5 shRNA. In cultured DRG neurons, the effect of IPP5 shRNA on expression of endogenous IPP5 could be rescued by IPP5^R as well (Fig. 2B).

Importantly, two days after knocking down IPP5 in cultured DRG neurons, the length of the longest neurite was increased by approximately 24% (Fig. 2D,E), the total neurite length was increased by approximately 35% (Fig. 2D,F), and the number of neurite ends was increased by approximately 18% (Fig. 2D,G). The cumulative frequency curves of all measures were significantly right-shifted in DRG neurons transfected with IPP5 shRNA (Fig. 2E–G). Coexpression of untargeted IPP5^R corrected the length of the longest neurite, the total neurite length and the number of neurite ends, which were increased by IPP5 shRNA (Fig. 2D–G), indicating that the phenotype caused by silencing IPP5 is not due to off-target effects. These results suggest that IPP5 is a negative regulator of neurite growth in primary sensory neurons.

IPP5 has been identified as a novel inhibitory subunit of PP1 that contains both the PP1-docking motif and the PP1-inhibiting motif (Wang et al., 2008). The PP1-docking motif K⁸IQF¹¹ is required for IPP5 binding to PP1, and the threonine at position 34 (Thr³⁴) within the PP1-inhibiting motif is presumed to bind to the active site of PP1 when phosphorylated (Svenningsson et al., 2004; Wang et al., 2008). A co-immunoprecipitation (co-IP) experiment confirmed that IPP5 interacts with PP1 in both rat DRGs and HEK293T cells expressing IPP5 (Fig. 3B,C). Consistent with the shRNA knockdown experiment, IPP5 overexpression in cultured rat DRG neurons (supplementary material Fig. S2A,B) reduced the length of the longest neurite to ∼63% of the control (Fig. 3D,E), reduced the total neurite length to approximately 72% of the control (Fig. 3D,F), and reduced the number of neurite ends to 84% of the control (Fig. 3D,G). In addition, we also tested the effect of IPP5 on the percentage of neurons bearing neurites. Overexpressing IPP5 for 24 hr in cultured DRG neurons significantly decreased the percentage of neurons bearing neurites, including IB4-binding and CGRP-positive small DRG neurons and large DRG neurons with a diameter >35 µm (supplementary material Fig. S3). To further test whether the regulation of neurite growth by IPP5 depends on its interaction with PP1, we produced an IPP5 mutant (IPP5-M) in which the PP1-docking motif K⁸IQF¹¹ was replaced with alanine (Fig. 3A). Co-IP experiments showed that the interaction between PP1 and IPP5-M was dramatically reduced compared with that between PP1 and wild-type IPP5 in HEK293T cells (Fig. 3C). Correspondingly, IPP5-M had no significantly inhibitory effect on the length of the longest neurite, the total neurite length or the number of neurite ends (Fig. 3D–G). These data indicate that the interaction of IPP5 with PP1 is essential for its inhibition of neurite growth.

Previous in vitro protein phosphatase assays have shown that IPP5 phosphorylated at Thr³⁴ and IPP5^T34D (Fig. 3A), a phosphorylation-mimicking mutant, were able to effectively inhibit PP1 activity with IC₅₀ values of 45 nM and 110 nM, respectively, while IPP5^T34A (Fig. 3A), a phosphorylation-deficient mutant, had no effect on PP1 (Wang et al., 2008). Our in vitro autoradiograph assay showed that IPP5 was phosphorylated by PKA but IPP5^T34D was not (Fig. 4A). The phosphorylation of IPP5 was also detected by a site-specific monoclonal antibody against phosphorylated Thr³⁴, which was blocked by the PKA pharmacological inhibitor H89 (Fig. 4B). Further pharmacological assays showed that the phosphorylation level of IPP5 at Thr³⁴ was dramatically elevated by treating cultured DRG neurons with the cAMP/PKA signaling pathway activator forskolin and was further enhanced by the protein phosphatase 2B (PP2B) inhibitor cyclosporin A (Fig. 4C). Thus, Thr³⁴ within the PP1-inhibiting motif of IPP5 is phosphorylated by PKA and dephosphorylated by PP2B, similar to the other PPP1R1 family members, inhibitor-1 and DARPP-32 (Endo et al., 1996; Nairn et al., 2004; Weiser et al., 2004).

To further investigate whether the regulation of neurite growth by IPP5 depends on its inhibition against PP1, dissociated DRG neurons were electroporated with IPP5 or its two phosphorylation mutants, IPP5^T34A and IPP5^T34D. Similar to the cultured DRG neurons expressing IPP5, the length of the longest neurite, the total neurite length and the number of neurite ends were markedly decreased in neurons expressing IPP5^T34D but not IPP5^T34A (Fig. 4D–G). Taken together, these data indicate that both the interaction with and inhibition of PP1 are indispensable for the negative regulation of neurite growth by IPP5.

Given that PP1 mediates the effect of IPP5 on neurite growth in primary sensory neurons, we searched for the downstream pathway. Several signaling molecules have been reported to be both regulated by PP1 and involved in the regulation of neurite growth (Abe et al., 2010; Bito et al., 1996; Gao et al., 2004; Hur et al., 2011; Morfini et al., 2004; Shi et al., 2004; Stegmüller et al., 2008; Thayyullathil et al., 2011; Xiao et al., 2010; Zhou et al., 2004). Dephosphorylation of Akt/glycogen synthase kinase 3 beta (GSK3β), cAMP-response element binding protein (CREB) and TβRI by PP1 regulates the phosphatidylinositol 3-kinase (PI3K), cAMP/PKA and TGF-β/Smad signaling pathways, respectively. To ascertain whether these signaling molecules are regulated by IPP5, we expressed IPP5 or its phosphorylation mutants, IPP5^T34A and IPP5^T34D, in HEK293T cells. The basal phosphorylation levels of CREB, Akt and GSK3β were high and unchanged in cells expressing IPP5 or its mutants (Fig. 5A). However, the basal phosphorylation levels of Smad2 and Smad3, two molecules downstream of TβRI, were almost undetectable in HEK293T cells. We treated cells with 10 ng/ml TGF-β1 for 60 min to increase the phosphorylation level of Smad2/3 and found that Smad2/3 phosphorylation was upregulated in cells expressing IPP5 and IPP5^T34D compared to cells expressing the control vector (Fig. 5B). Importantly, the phosphorylation level of Smad2/3 was not affected in cells expressing IPP5-M or IPP5^T34A (Fig. 5B). This result suggests that IPP5 positively regulates the TGF-β/Smad signal pathway, and the interaction with and inhibition of PP1 are also essential for IPP5-mediated Smad2/3 activation.

We also tested whether IPP5 regulates the TGF-β signaling pathway in primary sensory neurons. Immunocytochemistry showed that IPP5 colocalized with PP1 and TβRI in the region near the plasma membrane in dissociated DRG neurons (Fig. 5C). After TGF-β1 stimulation, the distributions of IPP5, PP1 and TβRI were not obviously changed, but the phosphorylation level of Smad2/3 within the nucleus was significantly increased (supplementary material Fig. S4A,B). IPP5 siRNA was employed to efficiently knock down endogenous IPP5 in order to perform biochemical assays in cultured DRG neurons. The phosphorylation level of Smad2/3 after treatment with 10 ng/ml TGF-β1 for 60 min was reduced in cultured DRG neurons transfected with IPP5 siRNA compared to control siRNA (Fig. 5D). These data suggest that IPP5 regulates the TGF-β/Smad signaling pathway in primary sensory neurons.

Signaling by TGF-β family members mainly occurs via type I and type II serine/threonine kinase receptors. Binding of TGF-β1 dimers to TβRII leads to the recruitment of TβRI and the formation of a tetrameric complex. Constitutively active TβRII activates TβRI by specifically phosphorylating its serine and threonine residues in the juxta-membrane region. Activated TβRI propagates the signal downstream by directly phosphorylating Smad2 and Smad3 (ten Dijke and Hill, 2004). Previous studies have reported that PP1 interacts with TβRI and inhibits its function via dephosphorylation (Shi et al., 2004). Our co-IP experiments showed that IPP5 interacts with both TβRI and PP1 in cultured DRG neurons. This interaction was not enhanced by TGF-β treatment (Fig. 5E). Reverse-IPs in HEK293T cells transiently expressing IPP5 and HA-TβRI also showed an interaction of TβRI with IPP5 and PP1 (Fig. 5F). Thus, IPP5 forms a protein complex with TβRI and PP1, which may maintain TGF-β/Smad signaling by inhibiting the dephosphorylation of TβRI by PP1.

The TGF-β/Smad signaling pathway has been reported to be an intrinsic negative regulator of neurite growth (He and Wang, 2006; Hellal et al., 2011; Kohta et al., 2009; Ng, 2008; Stegmuller et al., 2008). Considering the effect of IPP5 on TGF-β/Smad signaling, we investigated the involvement of the TGF-β/Smad pathway in the IPP5-induced inhibition of neurite growth. Reverse transcription PCR showed that TGF-β1, TGF-β2, and TGF-β3 were expressed in both DRGs and cultured DRG neurons (supplementary material Fig. S4C). We then performed neurite growth assays in the presence of TGF-βs. The data showed that the length of the longest neurite, the total neurite length and the number of neurite ends were dramatically inhibited by TGF-βs (Fig. 6; supplementary material Fig. S4D–G). TGF-β1-induced inhibition of the length of the longest neurite and the total neurite length were further enhanced by IPP5 overexpression, but this effect was not observed with regard to the number of neurite ends (Fig. 6). Treatment with 5 nM tautomycin, a PP1 inhibitor, prevented the enhancement of the TGF-β1-induced neurite growth inhibition by IPP5 (Fig. 6). Similarly, IPP5-M, which lacks the ability to bind PP1, did not enhance the TGF-β1-induced inhibition of neurite growth (Fig. 6). These data show that IPP5 enhances the TGF-β-induced inhibition of neurite growth in a PP1-dependent manner.

A drug (SB-431542) has been reported to be a potent and specific inhibitor of the TGF-β receptor superfamily. It dramatically inhibits the kinase activity of TβRI, the activin type I receptor and the nodal type I receptor, but it has little effect on other protein kinases (Inman et al., 2002). Considering expression of TGF-βs in cultured DRG neurons (supplementary material Fig. S4C), activation of TGF-β signaling was maintained in basal condition (Fig. 5C). Inhibition of TGF-β signaling with 5 µM SB431542 increased the length of the longest neurite by approximately 27%, the total neurite length by approximately 37%, and the number of neurite ends by approximately 12.7% in cultured rat DRG neurons (Fig. 7). Importantly, the IPP5-induced inhibition of neurite growth was blocked by 5 µM SB431542 treatment (Fig. 7). To search for a more specific way to block the activation of the TGF-β/Smad signaling pathway, we selected a Smad2 shRNA to knock down endogenous Smad2 in DRG neurons (supplementary material Fig. S5A). In cultured DRG neurons expressing Smad2 shRNA, the inhibition of the TGF-β signaling increased the length of the longest neurite by approximately 23% and the total neurite length by approximately 25% (Fig. 8A–C). Furthermore, the IPP5-induced inhibition of neurite growth was blocked in cultured DRG neurons coexpressing Smad2 shRNA (Fig. 8A–C). Taken together, these data indicate that the activation of the TGF-β/Smad signaling pathway is necessary for the IPP5-induced inhibition of neurite growth in primary sensory neurons.

IPP5 is a PP1-inhibiting protein that has been shown to regulate tumor cell cycle, growth and apoptosis (Wang et al., 2008; Zeng et al., 2009). Here, we report a critical role of IPP5 in the inhibition of neurite growth in primary sensory neurons. IPP5 was selectively expressed in the DRG in the nervous system, specifically in small primary sensory neurons. If the ability of IPP5 to interact with or inhibit PP1 was disrupted, the inhibitory effects of IPP5 on neurite growth were impaired. IPP5 formed a protein complex with TβRI and PP1 and maintained TGF-β/Smad signaling by inhibiting the function of PP1 (Fig. 8D). Furthermore, IPP5 enhanced the TGF-β-induced inhibition of neurite growth. Blockage of TGF-β/Smad signaling disrupted the IPP5-induced inhibition of neurite growth. These findings extend our understanding about the function of the regulatory subunits of PP1 in the nervous system.

Phosphorylation and dephosphorylation of cellular proteins by protein kinases and phosphatases are important mechanisms for controlling many cellular events. In the nervous system, protein phosphatases are contained in highly dynamic complexes localized within specialized subcellular compartments to ensure the temporally and spatially controlled dephosphorylation of multiple neuronal phosphoproteins (Mansuy and Shenolikar, 2006). Approximately ten different regulatory subunits of PP1 have been identified in the nervous system (Cohen, 2002). Neurabin-I and spinophilin are both regulatory subunits targeting PP1 to F-actin during neuronal development, which regulates spine development or axonal growth by affecting the cytoskeletal dynamics in hippocampal and cortical neurons (Bielas et al., 2007; Oliver et al., 2002). Scapinin is another PP1-inhibiting protein that inhibits axonal growth without affecting neurite branching in primary rat cortical neurons (Farghaian et al., 2011). As a regulatory subunit of PP1, IPP5 is selectively expressed in primary sensory neurons localized in the DRG. Knocking down IPP5 significantly elevated neurite growth in cultured DRG neurons, indicating that IPP5 functions as a negative regulator of neurite growth. This conclusion is further supported by the evidence that IPP5 overexpression significantly inhibited neurite growth in cultured DRG neurons. IPP5 is the first regulatory subunit of PP1 identified in primary sensory neurons that exhibits this function.

Previous studies reported that members of the PPP1R1 family function by regulating PP1-dependent signaling pathways and inhibiting PP1 activity (Bollen et al., 2010; Nairn et al., 2004). Our study also suggests that IPP5 inhibits neurite growth in a PP1-dependent manner. Being a member of the PPP1R1 family, IPP5 possesses a highly conserved PP1-docking motif K⁸IQF¹¹ in the N-terminus and a PP1-inhibiting motif R³¹RPTPA³⁶. Loss of IPP5 binding or inhibitory activity regarding PP1 and blockage of PP1 activity disrupted the IPP5-induced inhibition of neurite growth in cultured DRG neurons. In contrast, increasing IPP5 inhibitory activity regarding PP1 by expressing IPP5^T34D inhibited neurite growth. However, the inhibitory effect of IPP5^T34D on neurite growth was slightly weaker than that of wild-type IPP5, which is consistent with previous evidence showing lower inhibitory activity of IPP5^T34D towards PP1 compared to wild-type IPP5 (Wang et al., 2008). The weaker inhibition of PP1 caused by IPP5^T34D might be due to a conformational difference from phosphorylated IPP5 at Thr³⁴, which leads to a reduced affinity for the PP1 catalytic subunit (Weiser et al., 2004).

Regulatory subunits endow PP1 with distinct substrate specificities and restricted subcellular locations (Cohen, 2002). It is important to identify the phosphoprotein substrates of the IPP5-PP1 holoenzyme complex. The phosphorylation levels of CREB, Akt and GSK3β were not altered in cells expressing IPP5 and its mutants. Importantly, in HEK293T cells expressing IPP5 or IPP5^T34D, the phosphorylation level of Smad2/3 was significantly elevated after TGF-β stimulation. In cultured DRG neurons, the TGF-β-induced phosphorylation of Smad2/3 was reduced after knocking down IPP5. Thus, IPP5 positively regulates TGF-β/Smad signaling. PP1 has been shown to dephosphorylate activated TβRI but not Smad2/3, which is an effective negative feedback mechanism for regulating TGF-β/Smad signaling (Lin et al., 2006; Shi et al., 2004). IPP5, TβRI and PP1 were found to colocalize in the region near the plasma membrane in dissociated DRG neurons. Importantly, IPP5 was shown to interact with TβRI and PP1, which formed a complex in DRG neurons. In HEK293T cells expressing the IPP5 mutants that lack PP1-docking or inhibiting activities, the phosphorylation levels of Smad2/3 were not further elevated after TGF-β stimulation. These data suggest a model in which IPP5 inhibits PP1 activity, keeps TβRI activity, and maintains TGF-β/Smad signaling (Fig. 8D).

During the development of the nervous system, TGF-β/Smad signaling is critical for establishing neuronal polarity and axonal identity (Awasaki et al., 2011; Farkas et al., 2003; Ng, 2008; Stegmüller et al., 2008; Yi et al., 2010). Loss of the TGF-β receptor homolog Baboon resulted in neurite overextension in neurons in drosophila mushroom body (Ng, 2008). Blockage of the TGF-β signaling pathway with Taxol facilitated axonal regeneration by decreasing scar formation and enhancing intrinsic axonal growth (Hellal et al., 2011). In cultured cortical neurons, TGF-β-induced inhibition of axonal growth was eliminated by SB431542 treatment or Smad2 knock-down (Stegmüller et al., 2008; Ylera et al., 2009). However, in RGC-5 cells transformed form retinal ganglion cells TGF-β promoted neurite growth through a noncanonical TGF-β/Smad signaling pathway (Walshe et al., 2011). In our study, TGF-βs exhibited an inhibitory effect on neurite growth and blocking TGF-β/Smad signaling with the TβRI inhibitor SB431542 or Smad2 shRNA also increased neurite growth, which support a role for TGF-β signaling in the negative regulation of growth in DRG neurons. Moreover, IPP5 enhanced the TGF-β-induced inhibition of neurite growth and blocking TGF-β/Smad signaling with SB431542 or Smad2 shRNA impaired the IPP5-induced inhibition of neurite growth. Thus, the TGF-β/Smad signaling pathway is essential to the function of IPP5 in primary sensory neurons.

In this study, IPP5 was found to be highly expressed in the peripheral nervous system but not in the central nervous system, including the cortex, cerebellum, hypothalamus, hippocampus, medulla oblongata, pituitary gland and spinal cord. Two other members of the PPP1R1 family, inhibitor-1 and DARPP-32, exhibited universal distributions in the nervous system but low expression in the DRG. This expression pattern suggests that IPP5 is a member of the PPP1R1 family that is expressed specifically in primary sensory neurons to help PP1 recognize specific substrates and restrict its subcellular locations.

Importantly, primary sensory neurons that expressed IPP5 projected to peripheral tissues and lamina I–II of the dorsal horn in the spinal cord. Axonal growth in the spinal cord is dramatically inhibited after injury, even before the physical obstacle of the glial scar is formed. This suggests that the spinal cord provides an inhibitory environment for injured axons because of the activation of several inhibitory signaling pathways. TGF-β is an important factor secreted by local astrocytes at injury sites to stimulate the proliferation of astrocytes and form a glial scar. TGF-β also inhibits the growth of injured axons (Kohta et al., 2009; Stegmüller et al., 2008). The enrichment of IPP5 in the afferent fibers of primary sensory neurons in the spinal cord might aggravate the inhibition of axonal growth by maintaining activated TGF-β signaling. After nerve injury, IPP5 in the DRG is remarkably downregulated, indicating the activation of a mechanism to increase the intrinsic capacity for neurite growth in primary sensory neurons. Interestingly, IPP5 is dominantly expressed in small primary sensory neurons. Other inhibitory subunits expressed in large primary sensory neurons may also regulate PP1 function.

The expression construct of IPP5 was generated by inserting coding sequence of the PCR-amplified IPP5 (NM_001109200) from DRG cDNA of Sprague-Dawley rat into pMyc vector (Clontech Laboratories, Palo Alto, CA) in which Myc tag was substituted for EGFP. The primers for amplifying coding sequence of IPP5 from DRG were as follows: 5′-tacaagcttatggagcccaac-3′ and 5′-tacggatccatggttccactt-3′. Myc-IPP5^T34, Myc-IPP5^T34D, Myc-IPP5-M and Myc-1PP5^R plasmids were generated by PCR from Myc-IPP5 using KOD-Plus-mutagenesis kit (Toyobo, Osaka, Japan) with following primers: 5′-gatcaggaaaagaagacctgccccagcatcccttgtgattc-3′ and 5′-gaatcacaagggatgctggggcaggtcttcttttcctgatc-3′; 5′-gatcaggaaaaaagacctgacccagcatcccttgtgattc-3′ and 5′-gaatcacaagggatgctgggtcaggtcttcttttcctgatc-3′; 5′-atggagcccaacagccccaaagcagctgcagctgctgtgcctttattccag-3′ and 5′-ctggaataaaggcacagcagctgcagctgctttggggctgttgggctccat-3′; 5′-cggaagaagaagaatcagcgtcggagagagaagaaaagtgg-3′ and 5′-ccacttttcttctctctccgacgctgattcttcttcttccg-3′, respectively. The coding sequence of IPP5 and its mutants were also subcloned into pIRES-EGFP (Clontech Laboratories) or pCAG-IRES-EGFP, a modified vector of pDC316 (VGTC, Beijing, China) in which the CMV promoter was replaced by CAG promoter. The scramble shRNA and IPP5 shRNA were cloned into pSuper RNAi vector (Oligoengine, Seattle, WA) with primers: 5′-gatccccgagtgagagaacacagagattcaagagatctctgtgttctctcactctttttggaaa-3′ and 5′-agcttttccaaaaagagtgagagaacacagagatctcttgaatctcttgttctctcactcggg-3′; 5′-gatccccgcgcaagtgaaagagaagattcaagagatcttctctttcactgcgctttttggaaa-3′ and 5′-agcttttccaaaaagcgcaagtgaaagagaagatctcttgaatcttctctttcacttgcgcggg-3′. The control shRNA and Smad2 shRNA were cloned into U6 GFP RNAi vector (GenePharma, Shanghai, China) with primers: 5′-caccgttctccgaacgtgaacgtgtcacgtcaagagattacgtgacacgttcggagaatttttg-3′ and 5′-gatccaaaaaattctccgaacgtgtcacgtaatctcttgacgtgacacgttcggagaac-3′; 5′-caccgcgatcgagaactgcgaatacttcaagagagtattcgcagttctcgatcgcttttttg-3′ and 5′-gatccaaaaagcgatcgagaactgcgaatactctcttgaagtattcgcagttctcgatcgc-3′. For the expression and purification of GST-fused proteins, coding sequences of IPP5 and IPP5^T34D were subcloned into pGEX-4T1 vector (Amersham Pharmacia Biotech, Piscataway, NJ) with primers 5′-gcggatccatggagcccaacagcccc-3′ and 5′-ccgaattcttaatggttccacttttcttc-3′.

Adult rats (body weight 250 g; Shanghai Center for Experimental Animals, Chinese Academy of Sciences, China) were used according to the policy of the Society for Neuroscience (USA) regarding the use of animals. The experiment was approved by the Committee for the Use of Laboratory Animals and Common Facilities, Institute of Biochemistry and Cell Biology, Chinese Academy of Sciences. For animal modeling of a peripheral nerve injury, a 5 mm portion of rat sciatic nerve was transected at mid-thigh level. The rats were killed after 2, 7, 14 or 28 days (10 rats for each time point).

Total RNA was isolated from the tissues of adult rats using TRIzol reagent (Invitrogen, Carlsbad, CA). The first-strand cDNA was generated using SuperScript®II Reverse Transcriptase (Invitrogen) for reverse transcription PCR, and the products were analyzed on a 1% agarose gel. Glyceraldehyde 3-phosphate dehydrogenase (GAPDH) was used as an internal control. The real-time PCR was performed in triplicate with SYBR Premix Ex Taq (TaKaRa, Shiga, Japan) on an ABI 7500 real-time PCR system (Applied Biosystems, Carlsbad, CA), and the endogenous mRNA values of IPP5 were normalized to that of GAPDH. The primers were as follows: 5′-cctacaccagcatcccttgt-3′ and 5′-ttgcgctttcttcttcttcc-3′ for IPP5; 5′-aagaacctgagggagccact-3′ and 5′-tgggaatccagtggtagcat-3′ for Inhibitor-1; 5′-cacccaaagtcgaagagacc-3′ and 5′-tcatcctcgtcctcatcctc-3′ for DARPP-32; and 5′-ggcaagttcaacggcacag-3′ and 5′-cgccagtagactccacgac-3′ for GAPDH.

Adult rats were anesthetized and perfused through the ascending aorta with saline followed by 4% paraformaldehyde containing 0.02% picric acid. Lumbar (L)4 and L5 DRGs and the L4–L5 segment of the spinal cord were isolated, post-fixed, and cryoprotected in 20% sucrose. For both the DRG and the spinal cord, 14 µm sections were cut with a cryostat (Leica, Heidelberg, Germany). The sections were incubated overnight with a rabbit antibody against IPP5 (1∶5000; homemade) mixed with a mouse antibody against NF160/200 (1∶10,000; Sigma, St Louis, MO) or a goat antibody against CGRP (1∶1000; Biogenesis, Poole, UK) or a guinea pig antibody against P2X₃ receptor (1∶2000, Chemicon, Temecula, CA) or guinea pig antibody against substance P (SP, 1∶1000; Neuromics, Minneapolis, MN). Then, the sections were incubated with secondary antibodies conjugated to FITC and Cy3 (1∶100; Jackson ImmunoResearch, West Grove, PA) and examined under a Leica TCS SP5 MP confocal microscope (Leica, German). To label the IB4-positive small DRG neurons, the sections were incubated with fluorescein-labeled IB4 (1∶100; Vector Laboratories, Burlingame, CA). Two sections from each DRG were quantitatively analyzed for each rat, and the data were collected from at least three animals. To determine the percentage of IPP5-positive neurons, the number of stained neurons was divided by the total number of neurons. The percentage of IPP5-positive neurons within a subset of primary sensory neurons was also determined.

DRG neurons cultured on coverslips were fixed in 4% paraformaldehyde for 15 min and incubated overnight with a rabbit antibody against IPP5 mixed with a mouse antibody against PP1 (1∶1000; Epitomics) or with a mouse antibody against IPP5 (1∶5000; Abcam, Cambridge, UK) mixed with a rabbit antibody against TβRI (1∶500; Santa Cruz Biotechnology, Santa Cruz, CA) or phosphorylated Smad2/3 (1∶1000; Epitomics, Burlingame, CA) (supplementary material Fig. S5B) followed by secondary antibodies.

HEK293T cells (American Type Culture Collection, Manassas, VA) were cultured in MEM (Invitrogen, Carlsbad, CA) containing 10% fetal bovine serum (Biochrom, Berlin, Germany) and 100 U/ml penicillin/100 pg/ml streptomycin mixture (Invitrogen). ND7-23 cells (European Collection of Cell Cultures, Porton Down, UK) were cultured in DMEM (Invitrogen) with 10% fetal bovine serum, 100 U/ml penicillin/100 pg/ml streptomycin mixture, and 2 mM L-glutamine (Invitrogen). Transient expression was performed with Lipofectamine 2000 (Invitrogen) according to the manufacturer’s protocol.

The rats (body weight 100–120 g) were anesthetized and euthanized. The DRGs were dissected, digested with 1 mg/ml collagenase type 1A, 0.4 mg/ml trypsin type I, and 0.1 mg/ml DNase I (all Sigma) in DMEM (Invitrogen) at 37°C for 30 min, and then triturated. The dissociated DRG neurons were transfected by electroporation with Nucleofector II (Amaxa Biosystems, Cologne, Germany) using a nucleofection kit and then cultured in DMEM containing 10% fetal bovine serum. After 6 hours, the culture medium was replaced with DMEM/F12 (1∶1) containing 1% N2 supplement (Life Technologies, Grand Island, NY), and the neurons were maintained for further experiments. To test the knockdown efficiency of IPP5 shRNA and Smad2 shRNA, the freshly isolated DRG neurons were harvested 48 hr after transfection. For biochemical assays to test the regulation of IPP5 on the TGF-β/Smad signaling pathway, the dissociated DRG neurons were incubated with 4 µM cytosine-1-β-D-arabinofuranoside to inhibit glial proliferation and transfected with control siRNA or IPP5 siRNA using Lipofectamine^TM RNAiMAX reagent (Invitrogen) 18 hr later. The culture medium was replaced with DMEM/F12 (1∶1) containing 1% N2 supplement after 6 hr, and these neurons were harvested for immunoblotting after 48 h. The oligonucleotides for control siRNA and IPP5 siRNA were as follows: 5′-uucuccgaacgugucacgutt-3′ and 5′-acgugacacguucggagaatt-3′; 5′-cccuugugauucucaaugatt-3′ and 5′-ucauugagaaucacaagggat-3′.

The dissociated DRG neurons were transfected by electroporation with the indicated plasmids together with the GFP plasmid (1∶1) if necessary, plated on poly-D-lysine (Sigma)-coated cover glasses, and fixed for imaging after 48–72 h. For each experiment, 30–50 neurons expressing GFP were selected randomly for each group; however, nearly the same number of neurons were measured from each group for one independent experiment. The length of the longest neurite, the total neurite length, and the number of neurite ends per neuron were measured and analyzed with Neurolucida software (MBF Bioscience, Williston, VT). The values were then averaged for each experiment, and the data were pooled from three independent experiments and normalized to the control. The cumulative frequency was calculated from the proportion of the total events. For the representative images, we further converted the RGB images to grayscale, inverted, and adjusted the contrast globally to get ones with black neurites and white background.

After electroporation, most of the large DRG neurons died; ∼30% of the surviving DRG neurons were transfected, but only ∼80% of these transfected neurons were IPP5 positive (supplementary material Fig. S2C,D). Therefore, for the experiments using IPP5 shRNA or overexpression, ∼20% of the transfected DRG neurons in absent of endogenous IPP5 expression have been taken into account which may have led to an underestimation of the effect on neurite growth.

For the measurements of neurite growth, the DRG neurons were incubated with 5 µM SB431542 (Santa Cruz Biotechnology); or 10 ng/ml human TGF-β1 (Chemicon, Temecula, CA), TGF-β2 and TGF-β3 (Peprotech, Rocky Hill, NJ); or 5 nM tautomycin (Calbiochem, Nottingham, UK) immediately after transfection and maintained for 48 h. To detect the phosphorylation levels, co-IP efficiency and subcellular localization, HEK293T cells and cultured DRG neurons were treated for 60 min with 10 ng/ml TGF-β1 48 h after transfection with control, IPP5 or IPP5 mutant plasmids. To detect the IPP5 phosphorylation, cultured DRG neurons were treated for 10 min with 10 µM forskolin (Sigma) with or without 2 µM cyclosporin A (Sigma).

Freshly isolated DRGs from adult rats or dissociated DRG neurons or transfected HEK293T cells were lysed in ice-cold immunoprecipitation buffer (10 mM Tris-HCl, pH 7.4, 150 mM NaCl, 1 mM EDTA, 1% Triton X-100 and 10% glycerol). The lysate was immunoprecipitated with 0.5 µg of a mouse antibody against IPP5 or HA tag (Sigma) or rabbit antibody against Flag tag (Sigma) and then incubated with protein G-Sepharose beads (Amersham Biosciences, Heidelberg, Germany). For immunoblotting, the lysates or beads were incubated in SDS-PAGE loading buffer. The samples were separated on an SDS-PAGE gel, transferred, and probed with homemade antibodies against IPP5 (1∶10000); PP1 (1∶1000; Cell Signaling Technology, Boston, MA); phosphorylated Akt ser473 (1∶1000; Cell Signaling Technology); phosphorylated CREB Ser133 (1∶1000; Cell Signaling Technology); GAPDH (1∶50,000; Abcam); Smad2/3 (1∶5000; Santa Cruz Biotechnology); GFP (1∶10,000; Roche, Burlington, NC); phosphorylated Thr³⁴ of IPP5 (1∶1000, Cell Signaling Technology); or rabbit antibodies against phosphorylated GSK3β Ser9 (1∶1000; Cell Signaling Technology); GSK3β (1∶1000; Cell Signaling Technology); Akt (1∶1000, Abcam); CREB (1∶1000, Cell Signaling Technology); phosphorylated Smad2/3 (1∶10,000); TβRI (1∶1000); HA tag (1∶1000; Sigma); Flag tag (1∶5000; Sigma). The immunoreactive bands were then detected with horseradish peroxidase-conjugated secondary antibodies, visualized with enhanced chemiluminescence (Amersham Biosciences) and quantified with Image-Pro Plus software (Media Cybernetics Inc., Bethesda, MD). Each experiment was repeated at least three times.

Briefly, Escherichia coli BL21 were transformed with constructs encoding GST-fused IPP5 or IPP5^T34D, and protein production was induced with 0.1 mM isopropyl-β-D-thiogalactopyranoside (Amresco, Solon, OH). The fusion proteins were loaded onto a column packed with glutathione-Sepharose beads (Amersham Biosciences). Instead of eluting the fusion protein with glutathione buffer, a solution of 2.5 U/µl thrombin (Amersham Biosciences) was loaded onto the column and allowed to incubate at room temperature for 3 h. After elution from the column, the reaction was terminated by adding 1 mM PMSF, heating in a boiling water bath for 5 min and centrifuging at 40,000 g for 20 min. The supernatant was collected and quantitatively analyzed by the Bradford assay (Sigma). For the measurement with autoradiograph, purified IPP5 or IPP5^T34D (200 ng) was suspended in 30 µl reaction buffer (50 mM Tris-HCl, pH 7.5 and 10 mM MgCl₂) and incubated with 200 µM ATP, 50 U purified PKA (New England BioLabs, Ipswich, MA) and 5 µCi [γ-³²P]ATP (PerkinElmer, Waltham, MA) at 30°C for 20 min. The phosphorylation status was analyzed by SDS-PAGE and autoradiography. For the measurement with site-specific monoclonal antibody against phosphorylated Thr³⁴ of IPP5, purified GST-IPP5 (800 ng) was suspended in 25 µl reaction buffer and incubated with 25 µM ATP and 100 U purified PKA with or without 50 µM H89 (Calbiochem) at 30°C for 4 hr. Phosphorylation status was analyzed by SDS-PAGE and immunoblotting.

The data are shown as means ± s.e.m. Statistical significance was calculated using unpaired or paired Student’s t-tests. The Kolmogorov–Smirnov test (KS-test) was performed to determine the significance between two groups for the cumulative frequency of the length of the longest neurite, the total neurite length and the number of neurite ends. Differences were considered significant at P<0.05.

We thank Dr Yeguang Chen for providing the plasmids of FLAG-tagged Smad2 and Smad3, and HA-tagged TβRI.