p75 neurotrophin receptor regulates tissue fibrosis through inhibition of plasminogen activation via a PDE4/cAMP/PKA pathway

Tissue scarring, characterized by cell activation, excessive deposition of ECM, and extravascular fibrin deposition, is considered a limiting factor for tissue repair. Fibrin, the major substrate of the serine protease plasmin, is a provisional matrix deposited after vascular injury (Bugge et al., 1996). The two plasminogen activators (PAs), namely tissue plasminogen activator (tPA) and urokinase plasminogen activator (uPA) and their inhibitors, such as plasminogen activator inhibitor-1 (PAI-1), are key modulators of scar resolution by spatially and temporally regulating the conversion of plasminogen to plasmin resulting in fibrin degradation and ECM remodeling (Lijnen, 2001). In the peripheral nervous system, previous work by us and others showed that inhibition of fibrinolysis in mice deficient in plasminogen or tPA exacerbated axonal damage (Akassoglou et al., 2000) and impaired functional recovery after nerve injury (Siconolfi and Seeds, 2001). In accordance, mice deficient for fibrinogen showed increased regenerative capacity (Akassoglou et al., 2002). Studies of fibrin deposition in human diseases, in combination with experiments from mice deficient in plasminogen and PAs, have provided information about a wide range of physiological and pathological conditions that are exacerbated by defective fibrin degradation, such as wound healing, metastasis, atherosclerosis, lung ischemia, rheumatoid arthritis, muscle regeneration, and multiple sclerosis (MS) (Degen et al., 2001; Adams et al., 2004). However, the molecular mechanisms that regulate proteolytic activity remain unclear.

In our current work, we focus on the mechanisms that regulate fibrinolysis after injury. Our previous studies demonstrated a correlation between fibrin deposition and expression of p75 neurotrophin receptor (p75^NTR) after nerve injury (Akassoglou et al., 2002). Up-regulation of p75^NTR is observed in MS (Dowling et al., 1999), stroke (Park et al., 2000), and spinal cord (Beattie et al., 2002) and sciatic nerve injury (Taniuchi et al., 1986), all of which are associated with fibrin deposition. p75^NTR is also expressed in non-neuronal tissues (Lomen-Hoerth and Shooter, 1995) and is up-regulated in non-nervous system diseases associated with defects in fibrin degradation, such as atherosclerosis (Wang et al., 2000), melanoma formation (Herrmann et al., 1993), lung inflammation (Renz et al., 2004), and liver disease (Passino et al., 2007). p75^NTR has been primarily characterized as a modulator of cell death (Wang et al., 2000) and differentiation (Passino et al., 2007) in non-neuronal tissues. The expression of p75^NTR by cell types such as smooth muscle cells and hepatic stellate cells, which actively participate in tissue repair by migration, and secretion of ECM and extracellular proteases, raises the possibility for a functional role of p75^NTR in disease pathogenesis that extends beyond apoptosis and differentiation.

We find that p75^NTR is involved in the regulation of proteolytic activity and fibrin degradation. Mice deficient for p75^NTR (Lee et al., 1992) show increased proteolytic activity and decreased fibrin deposition in two disease models: sciatic nerve injury and lung fibrosis. p75^NTR regulates proteolytic activity by simultaneously down-regulating tPA and up-regulating PAI-1 via a novel cAMP/PKA pathway. p75^NTR decreases cAMP via interaction with the cAMP-specific phosphodiesterase (PDE) isoform PDE4A4/5. This is of particular note, as selective PDE4 inhibitors have an anti-inflammatory action and have potential therapeutic utility in inflammatory lung disease, as well as in a wide range of neurologic diseases such as depression, spinal cord injury, MS, and stroke (Gretarsdottir et al., 2003; Nikulina et al., 2004; Houslay et al., 2005). Overall, the regulation of plasminogen activation by p75^NTR identifies a novel pathogenic mechanism whereby p75^NTR interacts with PDE4A4/5 to degrade cAMP and thus perpetuates scar formation that could possibly render the environment hostile for tissue repair.

To examine whether p75^NTR regulates fibrin deposition in the sciatic nerve we compared fibrin levels in wild-type (wt) and p75^NTR−/− mice after injury. In wt mice, there is a dramatic increase of fibrin deposition (Fig. 1 c) and p75^NTR expression (Fig. 1 d) after injury, when compared with uninjured nerves (Fig. 1, a and b). In contrast, p75^NTR−/− mice show reduced fibrin deposition after injury (Fig. 1 e). Quantification of immunoblots reveals that p75^NTR−/− mice have decreased fibrin by threefold 3 d and fourfold 8 d after injury (Fig. 1 g). Quantification of fibrin immunostaining also reveals that p75^NTR−/− mice have significantly decreased fibrin (Fig. 1 h, P < 0.003). These results suggest that loss of p75^NTR decreases the levels of fibrin in the sciatic nerve after injury.

Analysis of total fibrinogen levels were similar in the plasma of wt and p75^NTR−/− mice (unpublished data), suggesting the decrease in fibrin deposition is not the result of hypofibrinogenemia. Because fibrin removal depends on proteolytic activity (Bugge et al., 1996), we hypothesized that the decreased fibrin in the p75^NTR−/− mice reflects an up-regulation of the proteolytic activity. p75^NTR−/− mice have increased proteolytic activity (Fig. 2 b) when compared with wt mice (Fig. 2 a) that is statistically significant (Fig. 3 i, P < 0.05). Uninjured nerves exhibit minimal proteolytic activity (Fig. 2 i), as expected (Akassoglou et al., 2000). Injured p75^NTR−/− sciatic nerves do not show lysis of fibrin in the absence of plasminogen (Fig. 2 c), suggesting that the proteolytic activity is plasminogen dependent.

The tPA/plasmin system regulates fibrin clearance after nerve injury (Akassoglou et al., 2000). A specific tPA inhibitor, tPASTOP, blocks proteolytic activity in p75^NTR−/− mice (Fig. 2 d). p75^NTR is strongly activated by withdrawal of axons (Lemke and Chao, 1988) and its expression correlates with proliferating, non-myelin producing Schwann cells (SCs) (Zorick and Lemke, 1996). After sciatic nerve injury both p75^NTR (Fig. 2 e, red) and tPA (Fig. 2 e, green) increase when compared with uninjured controls (Fig. 2 j), but show little colocalization (Fig. 2, e and h), suggesting that p75^NTR-reexpressing SCs do not express tPA. Expression of tPA (Fig. 2 k, red) and p75^NTR (Fig. 2 l, red) in SCs is confirmed using double immunofluorescence with the SC marker S100 (Fig. 2, k and l; green).

To examine genetically whether the increased proteolytic activity in the p75^NTR−/− mice was due to tPA, we crossed p75^NTR−/− mice with tPA^−/− mice and generated p75^NTR−/−tPA^−/− double-knockout mice. p75^NTR−/− mice show a decrease in fibrin deposition (Fig. 3 b) and an increase in proteolytic activity (Fig. 3 f), compared with wt control mice (Fig. 3, a and e, respectively). In contrast, p75^NTR−/−tPA^−/− mice show increased fibrin deposition (Fig. 3 c) when compared with p75^NTR−/− mice (Fig. 3 b) and no evidence of proteolytic activity (Fig. 3 g). As a control, tPA^−/− mice also show no evidence of proteolytic activity after sciatic nerve crush injury (Fig. 3 h), as described previously (Akassoglou et al., 2000). Quantification of proteolytic activity is shown in Fig. 3 i. The evidence derived from the genetic depletion of tPA in the p75^NTR−/− mice (p75^NTR−/−tPA^−/− mice, Fig. 3 g) are in accordance with the pharmacologic inhibition of tPA activity in the p75^NTR−/− sciatic nerve using tPASTOP (Fig. 2 d). Overall, these results suggest that up-regulation of proteolytic activity in the sciatic nerve of p75^NTR−/− mice is due to up-regulation of tPA.

Because SCs are a major source for tPA after injury, we isolated primary SCs from wt and p75^NTR−/− mice and cultured them on a three-dimensional (3D) fibrin gel. Wt SCs, which express high levels of p75^NTR, form a monolayer on the fibrin gel (Fig. 4 a). In contrast, p75^NTR−/− SCs degrade the fibrin gel (Fig. 4 b) and show a 2.7-fold increase of fibrin degradation (Fig. 4 c). p75^NTR−/− SCs show a sixfold increase in tPA levels, when compared with wt SCs (Fig. 4 d; P < 0.01). These results suggest that p75^NTR down-regulates tPA activity and blocks fibrin degradation in SCs in vitro.

After finding a biological function for p75^NTR in the regulation of tPA using SCs and sciatic nerves from p75^NTR−/− mice, we used stable and transient transfections of p75^NTR as well as siRNA against p75^NTR to test the properties of p75^NTR in heterologous systems. To examine whether p75^NTR could inhibit fibrin degradation, we first used NIH3T3 fibroblasts stably transfected with p75^NTR that exhibit high levels of p75^NTR (10⁵ receptors/cell) (Hsu and Chao, 1993). NIH3T3 cells on a 3D fibrin gel degrade fibrin (Fig. 5 a), whereas NIH3T3p75^NTR cells do not (Fig. 5 b). Expression of p75^NTR inhibits fibrin degradation by 12-fold (Fig. 5 c; P < 0.001). NIH3T3 cells form lytic areas (Fig. 5 d), whereas NIH3T3p75^NTR cells grow uniformly on fibrin (Fig. 5 e). NIH3T3 cells fully degrade the plasmin substrate casein (Fig. 5 f) but NIH3T3p75^NTR cells do not degrade casein (Fig. 5 g), suggesting impaired proteolysis in NIH3T3p75^NTR cells. Aprotinin, a general inhibitor of serine proteases, completely inhibits fibrin degradation by NIH3T3 cells (not depicted). In fibroblasts both tPA and uPA are involved in activation of plasminogen and fibrin degradation. tPA activity is significantly decreased in the NIH3T3p75^NTR cells (Fig. 5 h). In contrast, expression of p75^NTR has no effect on uPA activity (Fig. 5 i).

tPA is a transcriptionally regulated immediate-early gene (Qian et al., 1993). Indeed, expression of p75^NTR down-regulates tPA transcripts (Fig. 5 j). In addition, mRNA of PAI-1 is also up-regulated in NIH3T3p75^NTR cells (Fig. 5 j). Real-time quantitative PCR shows a 10.1-fold decrease in tPA mRNA, a fourfold increase in PAI-1 mRNA, and a twofold decrease in uPA mRNA in NIH3T3p75^NTR cells. Upon expression of p75^NTR, the decrease of uPA RNA does not affect uPA activity (Fig. 5 i). In contrast, the decrease of tPA RNA in NIH3T3p75^NTR cells results in a corresponding decrease in tPA activity (Fig. 5 h; P < 0.01). After injury, sciatic nerves of p75^NTR−/− mice show a fourfold increase of tPA RNA when compared with wt (Fig. 5 k). Moreover, p75^NTR−/− mice show an increase in tPA RNA in primary cerebellar granule neurons (CGNs) (Fig. S1 c), and increased proteolytic activity in the cerebellum (Fig. S1, a and b). Overall, these data suggest that expression of p75^NTR inhibits the tPA/plasmin system both in vivo in the cerebellum and after sciatic nerve injury, as well as in vitro in primary neurons, SCs, as well as fibroblasts.

Transcriptional regulation of tPA depends on the cAMP/PKA pathway (Medcalf et al., 1990). Indeed, elevation of cAMP, using dibutyryl-cAMP (db-cAMP), overcomes the inhibitory effect of p75^NTR (Fig. 6 a). Moreover, cAMP elevation, elicited using the general PDE inhibitor IBMX, elevates tPA activity in NIH3T3p75^NTR to the levels seen in NIH3T3 cells (Fig. 6 b). IBMX does not affect basal levels of tPA in NIH3T3 cells (Fig. 6 b). These data suggest that PDE activity is required for the p75^NTR-induced tPA decrease.

PKA activity is decreased in NIH3T3p75^NTR cells (Fig. 6 c, lanes 3 and 4) compared with NIH3T3 cells (Fig. 6 c, lanes 1 and 2), suggesting that p75^NTR expression reduces PKA activity. KT5720, a specific PKA inhibitor, decreases tPA activity in NIH3T3 cells (Fig. 6 b). Because the cAMP/PKA pathway enhances tPA transcription and suppresses PAI-1 secretion (Santell and Levin, 1988), we tested whether the cAMP/PKA pathway influences the p75^NTR regulation of tPA and PAI-1. Forskolin-induced cAMP elevation increases, whereas KT5720-induced PKA inhibition decreases tPA RNA in NIH3T3 cells (Fig. 6 d). Forskolin treatment of NIH3T3p75^NTR cells also increases both tPA RNA (Fig. 6 d) and activity (not depicted), whereas forskolin decreases PAI-1 RNA in both NIH3T3 and NIH3T3p75^NTR cells (Fig. 6 e).

Similar to NIH3T3 cells, elevation of cAMP increases the activity of tPA in both wt and p75^NTR−/− SCs (Fig. 6 f). Brain-derived neurotrophic factor (BDNF)/TrkB signaling has been shown to regulate tPA in primary cortical neurons (Fiumelli et al., 1999). In contrast to cortical neurons, SCs are known to express minute levels of TrkB but high levels of p75^NTR (Cosgaya et al., 2002). We show here that treatment of SCs with either BDNF or nerve growth factor (NGF) has no effect on tPA (Fig. 6 f). Similar results are obtained after treatment of SCs with pro-NGF, the high-affinity ligand of p75^NTR (Lee et al., 2001) (unpublished data). In addition, in NIH3T3 and NIH3T3p75^NTR cells, which do not express Trk receptors, the p75^NTR-mediated suppression of tPA activity occurs independent of neurotrophins or serum (unpublished data). In accordance, in NIH3T3 cells transient expression of the intracellular domain (ICD) of p75^NTR decreases tPA similar to the full-length (FL) p75^NTR (Fig. 6 g). These data suggest that neurotrophin/p75^NTR signaling is not involved in the regulation of tPA in SCs and fibroblasts and that regulation of tPA by p75^NTR is independent of neurotrophins.

Because the effects of p75^NTR were overcome by elevating cAMP, we examined whether p75^NTR reduced cAMP levels. Indeed, cAMP is decreased 7.8-fold in NIH3T3p75^NTR cells (Fig. 7 a; P < 0.0001). Transient expression of p75^NTR in NIH3T3 cells decreases levels of cAMP (Fig. 7 b; P < 0.0005).Furthermore, siRNA knockdown against p75^NTR leads to increased cAMP levels in both NIH3T3p75^NTR cells (Fig. 7, c and e; P < 0.02) and primary SCs (Fig. 7, d and f; P < 0.03). NIH3T3 cells transiently transfected with p75^NTR express fivefold less p75^NTR than the stably transfected NIH3T3p75^NTR cells (unpublished data). Differences in expression between stably and transiently transfected cells may account for the differences in the fold-decrease of cAMP and tPA between these two systems. Moreover, immunostaining with an antibody against cAMP shows increased cAMP in injured sciatic nerves from p75^NTR−/− mice (Fig. 7, g and h). In neurons BDNF elevates cAMP exclusively via TrkB (Gao et al., 2003). In NIH3T3p75^NTR cells, which do not express TrkB, stimulation with NGF or BDNF does not affect the p75^NTR-mediated suppression of cAMP (Fig. S2). Similarly, inhibition of neurotrophins by Fc-p75^NTR or BDNF by Fc-TrkB does not alter cAMP levels in NIH3T3p75^NTR cells (Fig. S2). In accordance, transient expression of the ICD of p75^NTR decreases cAMP similar to the FL p75^NTR in NIH3T3 cells (Fig. 7 b). Overall, these data suggest a neurotrophin-independent PDE4/cAMP pathway downstream of p75^NTR, which consequently leads to decreases in extracellular proteolysis.

Down-regulation of cAMP can be mediated either by inhibition of cAMP synthesis via the action of G_i, a G protein that inhibits adenylyl cyclase or via the action of PDEs. Treatment of cells with pertussis toxin (PTX) that blocks interactions between the G_i and G protein coupled receptors, does not rescue the p75^NTR-mediated down-regulation of cAMP (Fig. 7 a; P > 0.5). In contrast, the PDE inhibitor IBMX resulted in significant increase of cAMP in the NIH3T3p75^NTR cells when compared with control NIH3T3p75^NTR cells (Fig. 7 a; P < 0.000001). Use of specific chemical inhibitors for PDE isoforms shows that only rolipram, a specific inhibitor of PDE4, significantly increases cAMP levels in NIH3T3p75^NTR cells (Fig. 7 a; P < 0.000001) to the levels of NIH3T3 cells (Fig. 7 a; P = 0.051), suggesting that the p75^NTR-induced cAMP decrease is mediated via PDE4.

Recruitment of PDE4 to subcellular structures such as the plasma membrane concentrates the activity of PDEs and reduces PKA activity by enhancing degradation of cAMP (Brunton, 2003; Houslay and Adams, 2003). We therefore examined whether p75^NTR regulates cAMP via recruitment of PDE4. In NIH3T3p75^NTR cells, p75^NTR coimmunoprecipitates (co-IPs) with endogenous PDE4A (Fig. 8 a). No association is observed with the other three PDE4 sub-families, namely PDE4B, PDE4C, or PDE4D (unpublished data), suggesting that the effect was PDE4A specific. Based on the molecular weight of PDE4A at 109 kD, we determined that p75^NTR co-IPs with the PDE4A5 isoform. Endogenous co-IP in CGNs (Fig. S3 a) and in injured sciatic nerve (Fig S3 b) shows that p75^NTR and PDE4A5 interact at endogenous expression levels. Analysis of lysates shows that the levels of PDE4A are similar in NIH3T3 and NIH3T3p75^NTR cells (Fig. S3 c). These results show that p75^NTR forms a complex with PDE4A5.

A functional consequence of the p75^NTR–PDE4A5 interaction would be recruitment of PDE4A5 to the membrane resulting in decreased membrane-associated cAMP/PKA signaling. To investigate whether p75^NTR reduces membrane-associated PKA activity, we modified the genetically encoded A-kinase activity reporter, AKAR2 (Zhang et al., 2005) and generated pm-AKAR2.2, a membrane-targeted fluorescent reporter of PKA activity that generates a change in fluorescence resonance energy transfer (FRET) when it is phosphorylated by PKA in living cells (Fig. S4 a). As expected, NIH3T3 cells show a dramatic emission ratio change for the pm-AKAR2.2 in response to forskolin (Fig. 8 b). In contrast, NIH3T3p75^NTR cells show an attenuated response, revealing reduced PKA activity at the plasma membrane (Fig. 8 b). Transient transfection of p75^NTR confirmed the results observed in the stable NIH3T3p75^NTR cells using the latest generation of plasma membrane–specific PKA biosensor AKAR3 (Allen and Zhang, 2006) (Fig. S4 b). As expected, increased cAMP degradation at the plasma membrane results in decreased intracellular cAMP (Fig. S4 c; Fig. 7, a and b). Overall, our results showing reduced membrane-associated PKA activity upon expression of p75^NTR suggest that p75^NTR targets cAMP degradation to the membrane via its interaction with PDE4A5.

To verify the specificity of p75^NTR–PDE4A5 association, a series of mapping studies were conducted using deletion mutants. PDE4A5 interacts with FL p75^NTR, as well as deletions Δ3, Δ62, Δ83, but not a deletion missing the distal 151 amino acids, Δ151 (Fig. 8 c), suggesting that the interaction between p75^NTR and PDE4A5 occurs in the juxtamembrane region of p75^NTR, requiring sequences between residues 275 and 343. To explain the specificity of the interaction of p75^NTR with a single PDE4 isoform, we reasoned that p75^NTR would interact with a unique region of PDE4A5 that is not present in other PDE4s. Although the PDE4 isoforms are highly homologous, PDE4A5 contains a unique C-terminal region with a yet unknown biological function (Houslay and Adams, 2003). Co-IP experiments in HEK293 cells using the PDE4A4_δCT mutant that is missing the C-terminal region (aa 721–886) abolishes the interaction with p75^NTR (Fig. 8 d).

To examine whether p75^NTR could interact with PDE4A5 in a direct manner, we performed in vitro pull-down assays using recombinant proteins. A GST fusion protein of p75^NTR encoding the entire ICD interacts with both recombinant PDE4A5 and its human homologue PDE4A4 (Fig. 8 e). In contrast, p75^NTR ICD does not interact with recombinant PDE4D3 (Fig. 8 e). These results are in accordance with both the endogenous co-IPs in cells (Fig. 8, a and c; Fig. S3) and the PDE4A4 mutagenesis data (Fig. 8 d) because similar to PDE4A4_δCT, PDE4D3 does not contain the unique C-terminal domain of PDE4A4/5. We have used peptide array technology to define sites of direct interaction in other PDE4s (Bolger et al., 2006). Screening a peptide array library of overlapping 25-mer peptides that scanned the sequence of PDE4A4 with GST-ICD p75^NTR identified interactions with the LR1 domain, whose sequence is unique to the PDE4A subfamily (peptides 40 and 41, aa 191–220), and also to a sequence within the catalytic domain (peptides 135 and 136, aa 671–700). However, the strongest interaction was observed with sequences within the C-terminal region of PDE4A4 (peptides 172 and 173, aa 856–885). Alanine scanning mutagenesis shows that substitution of C862 abolishes the interaction of p75^NTR with the 173 peptide that is unique to PDE4A (Fig. 8 g). The p75^NTR-interacting sequences within the LR1 and C-terminal domains are highly conserved between the human PDE4A4 and the rodent PDE4A5. Indeed, peptide array screening for PDE4A5 reveals direct interaction with p75^NTR similar to that seen for PDE4A4 (unpublished data). Overall, these results suggest that the interaction of p75^NTR with PDE4A4/5 is direct and that sequences within the juxtamembrane region of p75^NTR and the unique C-terminal region of PDE4A4/5 are primarily required for the interaction (Fig. 8, a, c–g; Fig. S3).

Because expression of p75^NTR inhibits fibrinolysis in fibroblasts, we hypothesized that the role of p75^NTR as a modulator of fibrinolysis extends to tissues outside of the nervous system that express p75^NTR after injury or disease. Because p75^NTR is expressed in the lung (Ricci et al., 2004), we compared the levels of fibrin in the lung of wt and p75^NTR−/− mice in a model of lipopolysaccharide (LPS)-induced lung fibrosis (Chen et al., 2004). LPS-treated wt mice showed widespread extravascular fibrin deposition (Fig. 9 b) and decreased proteolytic activity after LPS treatment (Fig. 9 e), when compared with saline-treated wt mice (Fig. 9, a and d). In contrast, p75^NTR−/− mice show a 2.58-fold decrease of fibrin immunoreactivity (Fig. 9, c and j) and increased proteolytic activity (Fig. 9 f).

Decreased proteolytic activity in the lung after injury depends on the up-regulation of PAI-1 (Idell, 2003). Loss of PAI-1 protects from pulmonary fibrosis in LPS-induced airway disease, hyperoxia, and bleomycin-induced fibrosis (Savov et al., 2003). Because p75^NTR increases PAI-1 (Fig. 5 j and Fig. 6 e), we examined whether p75^NTR regulates expression of PAI-1 in vivo. PAI-1 is up-regulated in LPS-treated wt mice (Fig. 9 h) when compared with saline-treated wt mice (Fig. 9 g). In contrast, LPS-treated p75^NTR−/− mice show similar immunoreactivity for PAI-1 (Fig. 9 i) as saline-treated wt mice (Fig. 9 g), suggesting that p75^NTR induces up-regulation of PAI-1 after injury in the lung. Western blots show a decrease in PAI-1 in the lungs of p75^NTR−/− mice (Fig. 9 k). Similar to the p75^NTR−/− mice, rolipram reduces fibrin deposition in the lung (Fig. S5, a and b) and sciatic nerve (Fig. S5, d–f), and decreases PAI-1 in the lung (Fig. S5 c), suggesting the involvement of PDE4 in p75^NTR-mediated inhibition of fibrinolysis in vivo. Collectively, our data show that p75^NTR increases fibrin deposition via a PDE4-mediated inhibition of plasminogen activation in both LPS-induced lung fibrosis and sciatic nerve crush injury. These data suggest a role for p75^NTR/PDE4 signaling as a general regulator of plasminogen activation and fibrinolysis at sites of injury.

Our study shows a novel direct interaction of p75^NTR with PDE4A4/5, a specific PDE4 isoform, which results in the regulation of cAMP, a major intracellular signaling pathway, and mediates a major biological function of extracellular proteolysis and fibrinolysis (Fig. 10). p75^NTR is expressed in a wide range of tissue injury models, where repair depends upon both cell differentiation and ECM remodeling. For example, we recently showed that in the absence of plasminogen the effects of p75^NTR in tissue repair are protective due to its beneficial effects in cell differentiation (Passino et al., 2007). Similarly, in the flow-restricted carotid artery model of vascular injury that depends on uPA and not on tPA-mediated fibrinolysis (Kawasaki et al., 2001), p75^NTR is protective due to the induction of smooth muscle cell apoptosis (Kraemer, 2002). In the sciatic nerve p75^NTR appears to have a dual role by sustaining fibrin deposition (our study), and also promoting myelination (Cosgaya et al., 2002; Song et al., 2006). Examination of functional recovery in p75^NTR−/− mice after peripheral nerve injury would reveal the contribution of p75^NTR-mediated ECM remodeling and remyelination to the regeneration process. Overall, the biological role of p75^NTR after tissue injury would probably depend on the relative contributions of its role as a regulator of cell death and differentiation and its role as an inhibitor of fibrinolysis.

We identify regulation of cAMP as a novel signaling mechanism downstream of p75^NTR. Previous studies showed that β₂-adrenergic receptors target degradation of cAMP to the membrane via recruitment of multiple PDE4 isoforms, such as PDE4B1, PDE4B2, and PDE4Ds (Perry et al., 2002). Our finding of interaction between p75^NTR and PDE4A4/5 represents the first example of recruitment of a single PDE4 isoform to a transmembrane receptor. While interaction of β₂-adrenergic receptors with PDE4s is mediated via β-arrestin, our study suggests that the interaction of p75^NTR with PDE4A4/5 could be potentially mediated by direct binding to PDE4A domains, such as the C-terminal domain that is unique to this sub-family. In addition, in co-IP experiments we do not detect an interaction between p75^NTR and β-arrestin (unpublished data). It is possible that the unique C-terminal domain of PDE4A could regulate isoform-specific PDE4 recruitment to subcellular locations. Biological roles have been described for PDE4D in ischemic stroke (Gretarsdottir et al., 2003) and heart failure (Lehnart et al., 2005) and for PDE4B in schizophrenia (Millar et al., 2005). Our study identifies a biological function for PDE4A4/5 as a molecular mediator of p75^NTR/cAMP/PKA signaling involved in the regulation of tPA and fibrinolysis.

In spinal cord injury in rodents, elevation of cAMP via the PDE4 inhibitor, rolipram, promotes axonal regeneration and functional recovery (Nikulina et al., 2004). In the sciatic nerve, reduction of cAMP after injury is attributed primarily to up-regulation of PDE4 by SCs (Walikonis and Poduslo, 1998). Based on our findings, it is possible that reexpression of p75^NTR after injury could contribute to the activation of PDE4 and down-regulation of cAMP. BDNF, but not NGF, increases cAMP in neurons via TrkB (Gao et al., 2003). Moreover, BDNF/TrkB signaling overcomes the inhibition of nerve regeneration by myelin proteins via inhibition of PDE4 (Gao et al., 2003). We provide the first evidence for p75^NTR in the regulation of cAMP by using genetic depletion, siRNA knockdown or up-regulation of the p75^NTR. Our results suggest that p75^NTR might exert the opposite function as Trk receptors by recruiting PDE4A4/5 and decreasing cAMP. Interestingly, PDE4A has been detected as the predominant PDE4 isoform at the corticospinal tract (Cherry and Davis, 1999). Because p75^NTR can act as a coreceptor for NogoR, a mediator of the inhibition of nerve regeneration, PDE4A interaction with p75^NTR could play an inhibitory role in nerve regeneration by competing with neurotrophin signaling via Trk receptors.

It is possible that the increased expression of p75^NTR by neurons, glia, and brain endothelial cells could regulate the temporal and spatial pattern of tPA expression during brain injury or inflammation. p75^NTR might also be upstream of other non-fibrinolytic functions associated with tPA, such as neurodegeneration, synaptic plasticity, and long-term potentiation (Samson and Medcalf, 2006). Given the dependence of p75^NTR functions on the availability of ligands and coreceptors (Teng and Hempstead, 2004; Reichardt, 2006), further analysis will determine the role of p75^NTR in extracellular proteolysis and ECM remodeling in different cellular systems. We show that expression of p75^NTR can inhibit tPA in the absence of neurotrophin ligands. Constitutive expression of p75^NTR may signal in a neurotrophin-independent manner to induce neuronal apoptosis (Rabizadeh et al., 1993), activation of Akt (Roux et al., 2001), and RhoGTPase (Yamashita et al., 1999). The regulation of cAMP identified here is an effect of expression of p75^NTR that does not appear to depend on neurotrophin signaling. The cellular distribution of PDE4A4/5 would determine the involvement of p75^NTR in the regulation of cAMP. It is possible that non-neurotrophin ligands that bind directly to p75^NTR, such as β-amyloid (Teng and Hempstead, 2004), as well as the myelin/NogoR p75^NTR–dependent inhibitors of nerve regeneration (Filbin, 2003), are able to regulate both cAMP and plasminogen activation by p75^NTR. Because cAMP analogues decrease expression of p75^NTR (Baron et al., 1997), it is possible that p75^NTR by decreasing cAMP contributes to the positive regulation of its expression. Because PKA phosphorylates p75^NTR and regulates its translocation to lipid rafts (Higuchi et al., 2003), p75^NTR via regulation of PKA might regulate its own subcellular localization. Given the multiple genes regulated by cAMP and PKA, other cellular functions may be regulated by p75^NTR/cAMP signaling.

NGF/p75^NTR signaling has been suggested to enhance local neurogenic inflammation to exacerbate pulmonary disease (Renz et al., 2004). Our study suggests an additional pathway for p75^NTR as a regulator of expression of PAI-1 and a mediator of fibrosis. p75^NTR in the lung is detected mainly in basal epithelial cells of bronchioles (unpublished data). Similar to p75^NTR, PAI-1 is expressed by bronchial epithelial cells (Savov et al., 2003) and its expression results in an antifibrinolytic environment within the airway wall. Fibrin regulates both inflammation and airway remodeling (Idell, 2003; Savov et al., 2003). It is therefore possible that p75^NTR-mediated regulation of PAI-1 via PDE4 could influence inflammatory and tissue repair processes in pulmonary disease. Although in chronic obstructive pulmonary disease the PDE4A4 isoform is specifically up-regulated (Barber et al., 2004) and considered a pharmacologic target (Houslay et al., 2005), all available inhibitors target the common catalytic domain of all PDE4 isoforms resulting in unwanted side-effects. Our study suggests that the p75^NTR/PDE4A4 interaction could be a potential therapeutic target that will achieve the specific inhibition of a single PDE4 isoform and may have therapeutic potential.

Collectively, we have identified a novel cAMP-dependent signaling pathway initiated by p75^NTR that specifically regulates plasminogen activity and scar formation after sciatic nerve and lung injury. Though p75^NTR is responsible for a variety of cell survival and death decisions (Chao, 2003), our data has revealed an unrecognized property of this receptor to regulate the degradation of cAMP. This property provides a potential mechanism to account for how p75^NTR acts at sites of injury to promote ECM remodeling. The impact of high levels of p75^NTR expression upon inhibition of extracellular proteolysis indicates that the detrimental effects of p75^NTR extend beyond cell growth and axon inhibition. Finally, the dramatic inhibitory effect of p75^NTR signaling on plasminogen activation suggests that the p75^NTR/PDE4A4 interaction represents a novel target for therapeutic intervention in both neuronal and nonneuronal tissues.

p75^NTR−/− mice (Lee et al., 1992) and tPA^−/− mice (Carmeliet et al., 1994) were in C57Bl/6 background and purchased from The Jackson Laboratory. Double p75^NTR−/−tPA^−/− mice were generated by crossing p75^NTR−/− mice with tPA^−/− mice. C57Bl/6J mice were used as controls. Sciatic nerve crush was performed as described previously (Akassoglou et al., 2000). Lung fibrosis was induced as described previously (Chen et al., 2004). For the rolipram treatments, mice were administered 5 mg/kg rolipram (Calbiochem) before the LPS injection as described previously (Miotla et al., 1998). Mice were killed 4.5 h after LPS or saline administration. For rolipram treatment after sciatic nerve injury, mice were injected with rolipram (1 mg/kg) once daily for 8 d until tissue was harvested and processed for immunostaining.

Immunohistochemistry was performed as described in Akassoglou et al. (2002). Primary antibodies were sheep anti–human fibrin(ogen) (1:200; US Biologicals), rabbit anti–human tPA (1/300; Molecular Innovations), rabbit anti-p75^NTR clone 9651, (1:1,000), goat anti-p75^NTR (1/200; Santa Cruz Biotechnology, Inc.), rabbit anti–mouse PAI-1 (1:500; a gift from David Loskutoff, Scripps Research Institute, La Jolla, CA), and mouse anti-S100 (1:200; Neomarkers). For immunofluorescence, secondary antibodies were anti–rabbit FITC and anti–goat Cy3 (1:200; Jackson Immunochemicals). Images were acquired with an Axioplan II epifluorescence microscope (Carl Zeiss MicroImaging, Inc.) using dry Plan-Neofluar lenses using 10× 0.3 NA, 20× 0.5 NA, or 40× 0.75 NA objectives equipped with Axiocam HRc digital camera and the Axiovision image analysis system.

Immunoblot was performed as described previously (Akassoglou et al., 2002). Antibodies used were rabbit anti-p75^NTR clones 9992 and 9651 (1:5,000), mouse anti-fibrin (1:500; Accurate Chemical & Scientific Corp.), rabbit anti-myosin (1:1,000; Sigma-Aldrich), rabbit anti-GAPDH (1:5,000; Abcam) and rabbit anti-PAI-1 (1:5,000; a gift of David Loskutoff). Quantification was performed on the Scion NIH Imaging Software. Fibrin precipitation and quantification from lung tissues was performed exactly as described previously (Ling et al., 2004).

Co-IP was performed as described previously (Khursigara et al., 1999). Cell lysates were prepared in 1% NP-40, 200 mM NaCl, 1 mM EDTA, and 20 mM Tris-HCl, pH 8.0. IP was performed with an anti-p75^NTR antibody (9992) and immunoblot with anti-PDE4A, PDE4B, PDE4C, and PDE4D (Fabgennix). The co-IP buffer using NP-40 has been previously used to examine interactions of p75^NTR with other intracellular proteins, such as TRAF-6 (Khursigara et al., 1999) and PKA (Higuchi et al., 2003). For mapping experiments, PDE4A5 cDNA was cotransfected with HA-tagged p75^NTR deletion constructs into HEK293 cells. IP was performed with an anti-HA antibody (Cell Signaling). Cell lysates were probed with an anti-PDE4A or an anti-p75^NTR antibody (9651). For co-IP experiments using recombinant proteins, equimolar amounts (2 μM) of purified recombinant MBP-PDE4A5 (O'Connell et al., 1996), MPB-PDE4A4 (McPhee et al., 1999), MBP-PDE4D3 (Yarwood et al., 1999), and GST-p75^NTR-ICD (Khursigara et al., 2001) were mixed in binding buffer (50 mM Tris-HCl, pH 7.5, 100 mM NaCl, 2 mM MgCl₂, 1 mM DTT, 0.5% Triton X-100, and 0.5% BSA) and incubated for 1 h at 4°C. Washed glutathione-Sepharose beads were added according to the manufacturer's instructions for an additional hour. Beads were sedimented by centrifugation (10,000 g for 1 min) and washed three times. Proteins associated with the beads were eluted by boiling in loading buffer and separated by SDS-PAGE.

RT-PCR was performed as described previously (Akassoglou et al., 2002). Primers for tPA, uPA, and PAI-1 genes were used as described previously (Yamamoto and Loskutoff, 1996). Real-time PCR was performed using the Opticon DNA Engine 2 (MJ Research) and the Quantitect SYBR Green PCR kit (QIAGEN). Results were analyzed with Opticon 2 software using the comparative Ct method as described previously (Livak and Schmittgen, 2001). Data were expressed as ΔΔCt for the tPA gene normalized against GAPDH.

Quantification of tPA and uPA activity in SC and fibroblast in lysates and supernatants was performed according to the directions of the activity assay kits from American Diagnostica and Chemicon, respectively. To elevate cAMP cells were treated either with 2 mM db-cAMP (Sigma-Aldrich) or with 10 μM forskolin (Sigma-Aldrich) for 16 h. To block PKA activity, cells were treated with 200 nM KT5720 (Calbiochem). Induction with neurotrophins was performed using 100 ng/ml NGF and 50 ng/ml BDNF for 16 h before tPA assay.

Coating with fibrin was prepared as described previously (Lansink et al., 1998). To quantitate fibrin degradation, the supernatant was aspirated and the remaining gel was weighed using an analytical balance. Decrease of gel weight corresponded to increased fibrin gel degradation.

Murine SCs were isolated as described previously (Syroid et al., 2000). NIH3T3 or HEK293 cells were cotransfected either with p75^NTR FL, ICD or deletion constructs, and PDE4A5 cDNAs using Lipofectamine 2000 (Invitrogen) as described in the Results section. CGNs were isolated from P10 animals (Yamashita and Tohyama, 2003). CGNs were lysed immediately for co-IP, without plating. siRNA directed against p75^NTR (Dharmacon) was transfected into SCs and NIH3T3p75^NTR cells using Dharmafect (Dharmacon).

10⁶ fibroblasts or 500,000 SCs were lysed in 0.1 N HCl solution and cAMP was measured using a competitive binding cAMP ELISA (Assay Designs). Cells were treated with 100 ng/ml PTX for 16 h. For inhibition of PDE activity, cells were treated for 16 h with 500 μM isobutyl methylxanthine (IBMX; Calbiochem), 18.7 μM 8-methoxymethyl-3-isobutyl-1-methylxanthine (PDE1 inhibitor; Calbiochem), 80 μM erythro-9-(2-Hydroxy-3-nonyl)adenine (PDE2 inhibitor; Calbiochem), 100 nM trequinsin (PDE3 inhibitor; Calbiochem), and 10 μM rolipram (PDE4 inhibitor; Calbiochem). Cells were treated with forskolin in the presence of the inhibitors for 1 h. Because these inhibitors specifically inhibit a PDE isoform and have no effect on the other PDE isoenzymes (Beavo and Reifsnyder, 1990), they are extensively used for the identification of the specific PDE isoform that is involved in different cellular functions. Induction with neurotrophins was performed using 100 ng/ml NGF or 50 ng/ml BDNF, 750 ng/ml of FcTrkB, or 1.35 ug/ml of Fcp75^NTR for 1 h before cAMP assay. For the qualitative and quantitative PKA assay (Promega), cells were treated with 10 μM forskolin for 30 min, lysed in 1% NP-40 buffer with 150 mM NaCl, 50 mM Tris, and 1 mM EGTA, and protein concentration was determined using the Bradford Assay (Bio- Rad Laboratories). 1 μg was loaded into the PKA assay reaction mix according to the manufacturer's protocol (Promega).

In situ zymographies were performed as described previously (Akassoglou et al., 2000). Quantification of in situ zymographies was performed by measuring the area of the lytic zone surrounding each nerve, and dividing that value by the area of the nerve. Images were collected after 8 h of incubation for the sciatic nerve and 4 h of incubation for the lung. For cell zymographies, cultures were washed four times with PBS/BSA and overlaid with 200 μl of Dulbecco's minimum essential medium containing 1% LMP agarose, 2.5% boiled nonfat milk, and 25 μg/ml human plasminogen. The overlay was allowed to harden, and plates were incubated in a cell culture incubator at 37°C. Pictures of lytic zones were taken using an inverted microscope under dark field (Carl Zeiss MicroImaging, Inc.).

For the construction of pm-AKAR2.2 we used the previously described cytoplasmic PKA sensor, AKAR2 (Zhang et al., 2005). pm-AKAR2.2 consists of a cDNA containing a FRET pair, monomeric enhanced cyan fluorescent protein (ECFP), and monomeric citrine (an optimized version of YFP), fused to forkhead associated domain 1 (FHA1) (Rad53p 22–162), and the PKA substrate sequence LRRATLVD via linkers. A206K mutations were incorporated to ECFP and Citrine by the QuikChange method (Stratagene). The C-terminal sequence from K-Ras KKKKKKSKTKCVIM was added to target the construct to the plasma membrane. For expression in mammalian cells, the chimaeric proteins were subcloned into a modified pcDNA3 vector (Invitrogen) behind a Kozak sequence as described previously (Zhang et al., 2005).

For the generation of the PDE4A4_δCT, PDE4A4 was subcloned into p3XFLAG-CMV-14 using plasmid pde46 (GenBank/EMBL/DDBJ accession no. L20965) as template from Met-1 to Iso-721 (McPhee et al., 1999). A forward (5′) primer containing a HindIII restriction site immediately 5′ to the initiating Met-1 (ATG) of PDE4A4 and a reverse primer designed to the DNA sequence ending at Iso-721 (ATA) with BamHI restriction site immediately 3′ to Iso-721 was used to amplify Met-1 to Iso-721. The C terminus was removed simply by amplifying from Iso-721 instead of the final codon at the end of the full-length PDE4A4B. The C-terminally truncated PDE4A4B was cloned in-frame with three FLAG (Asp-Tyr-Lys-Xaa-Xaa-Asp) epitopes (Asp-726, Asp-733 & Asp-740) after the BamHI restriction site, therefore at the C terminus of the now-truncated PDE4A4. The stop codon (TAG) after the FLAG epitopes is located immediately after Lys-747. This strategy generates a C-terminal truncate of PDE4A4 from 1–721.

NIH3T3 cells and NIH3T3p75^NTR cells were transiently transfected with pm-AKAR2.2, AKAR3, or pm-AKAR3 (Allen and Zhang, 2006) and imaged within 24 h of transfection. Cells were rinsed once with HBSS (Cellgro) before imaging in HBSS in the dark at room temperature. An Axiovert microscope (Carl Zeiss Microimaging, Inc.) with a MicroMax digital camera (Roper-Princeton Instruments) and MetaFluor software (Universal Imaging Corp.) was used to acquire all images. Optical filters were obtained from Chroma Technologies. CFP and FRET images were taken at 15-s intervals. Dual emission ratio imaging used a 420/20-nm excitation filter, a 450-nm dichroic mirror and a 475/40-nm or 535/25-nm emission filter for CFP and FRET, respectively. Excitation and emission filters were switched in filter wheels (Lambda 10–2; Sutter Instrument Co.).

Peptide libraries were synthesized by automatic SPOT synthesis (Frank, 2002). Synthetic overlapping peptides (25 amino acids in length) were spotted on Whatman 50 cellulose membranes according to standard protocols by using Fmoc-chemistry with the AutoSpot Robot ASS 222 (Intavis Bioanalytical Instruments AG). Membranes were overlaid with 10 μg/ml recombinant GST-p75^NTR ICD. Bound recombinant GST-p75^NTR ICD (Khursigara et al., 2001) was detected using rabbit anti-GST (1:2,000; GE Healthcare) followed by secondary anti–rabbit horseradish peroxidase antibody (1:2,500; Dianova). Alanine scanning was performed as described previously (Bolger et al., 2006).

Statistical significance was calculated using JMP2 Software by unpaired t test for isolated pairs or by analysis of variance (one-way ANOVA) for multiple comparisons. Data are shown as the mean ± SEM.

Fig. S1 shows that genetic loss of p75^NTR increases tPA mRNA levels and proteolytic activity in the cerebellum. Fig. S2 demonstrates that treatment with neurotrophins has no effect on cAMP levels in NIH3T3 cells. Fig. S3 shows endogenous coimmunoprecipitations of PDE4A5 and p75^NTR from injured sciatic nerve and from primary CGNs. Fig. S4 shows results from transient transfections of NIH3T3 cells with p75^NTR and the PKA activity reporters, AKAR3 and pm-AKAR3. Fig. S5 shows that inhibition of PDE4s with rolipram decreases fibrin deposition both in LPS-induced lung fibrosis and sciatic nerve crush injury.

We thank David Loskutoff for the anti-PAI-1 antibody; Joan Heller Brown, Lawrence Brunton, Roger Y. Tsien, Barbara Hempstead, Juan-Carlos Arevalo, Hiroko Yano, and David Arthur for discussions; Paul Insel, Palmer Taylor, and Dan Littman for equipment access; Susan Taylor for assistance with the cAMP/PKA assays; Lisa Gallegos and Alexandra Newton for assistance with FRET experiments; and Binhai Zheng for advice on CGN culture. We thank Zsuzsana Pearson and Niccoló Zampieri for help with experiments and Xiaolin Tan, Andrew Maleson, and Priscila Kim for expert technical assistance.

Supported in part by the Pharmacology National Institutes of Health (NIH) training grant 5T32-GM07752 to B.D. Sachs and M.A. Passino, the DFG postdoctoral fellowship to C. Schachtrup, and the ASPET fellowship to J.R. McCall. The Alliance for Cell Signaling and NIH grant NS27177 to Roger Y. Tsien supported J. Zhang. This work was supported by NIH grants NS21072 and HD23315 to M.V. Chao; and NIH grants NS51470 and NS52189 to K. Akassoglou.

This work is dedicated to the memory of our parents Beatrice Du Chao, Evangelia Akassoglou, and Douglas Houslay, who passed away during the preparation of this manuscript.