Abstract
Neurotrophins are growth factors that promote cell survival, differentiation, and cell death. They are synthesized as proforms that can be cleaved intracellularly to release mature, secreted ligands. Although proneurotrophins have been considered inactive precursors, we show here that the proforms of nerve growth factor (NGF) and the proforms of brain derived neurotrophic factor (BDNF) are secreted and cleaved extracellularly by the serine protease plasmin and by selective matrix metalloproteinases (MMPs). ProNGF is a high-affinity ligand for p75NTR with high affinity and induced p75NTR-dependent apoptosis in cultured neurons with minimal activation of TrkA-mediated differentiation or survival. The biological action of neurotrophins is thus regulated by proteolytic cleavage, with proforms preferentially activating p75NTR to mediate apoptosis and mature forms activating Trk receptors to promote survival.
The neurotrophin family of growth factors, including NGF, BDNF, and neurotrophins-3 and -4 (NT-3, NT-4) regulates neuronal survival and synaptic plasticity (1). They are synthesized as precursors (proneurotrophins) that are proteolytically cleaved to mature, biologically active neurotrophins (2). Because neurotrophins are normally expressed at low levels, little is known about their processing and secretion by neurons and non-neuronal cells in vivo. However, when expressed in heterologous cells, proneurotrophins are secreted as well as cleaved intracellularly by furin or proconvertases at a highly conserved dibasic amino acid cleavage site for secretion as stable, noncovalent dimers (3, 4). Mature neurotrophins selectively bind to members of the Trk family of receptor tyrosine kinases, but they also interact with low affinity to a structurally distinct receptor, p75NTR. Although p75NTR can increase the affinity and specificity of Trk-neurotrophin interactions, p75NTR can also induce apoptosis in oligodendrocytes, neurons, and vascular smooth muscle cells when Trk activation is reduced or absent (5). High doses of neurotrophin elicit cell death through p75NTR, indicating that p75NTR signaling is relatively limited in comparison to that of other receptors of the tumor necrosis factor (TNF) receptor superfamily (5).
Defined functions for the neurotrophin prodomains have been limited to promoting the folding of the mature domain (6–8) and to sorting neurotrophins to either constitutive or regulated secretory pathways (9). However, sequence comparison of proneurotrophins revealed regions of the prodomain that are highly conserved across species, suggesting that they may mediate additional biological actions (10). To assess this possibility, the expression of proNGF and proBDNF in adult mouse tissues was determined (11). Using antibodies specific for the mature domains of NGF or BDNF (11), immunoreactive proteins with molecular masses of 18 to 30 kD were detected, in addition to the 13.5-kD mature forms of NGF and BDNF (11, 12). This suggested that tissue-specific proteolytic processing of proneurotrophins occurs in vivo.
Because proneurotrophins are secreted by cells (4,11, 12) and contain consensus sites for cleavage by plasmin and by MMP-3 and MMP-7 (13), we considered whether secreted proneurotrophins could be cleaved extracellularly by such proteases. Both plasmin and MMPs exhibit expression patterns consistent with neurotrophin action: at the synapse, where mature BDNF can activate presynaptic and postsynaptic TrkB receptors (14), and on endothelial cell surfaces, where mature BDNF promotes TrkB-mediated endothelial cell survival (15).
BDNF was harvested from the media of 293 or endothelial cells that were infected with recombinant adenovirus encoding BDNF (16). In addition to 13.5-kD mature BDNF, proBDNF forms of molecular mass 30, 28, and 24 kD, and those of mass 28 kD, were detected by a BDNF-specific antibody in the media of 293 cells and endothelial cells, respectively (Fig. 1, A and B). Addition of plasmin, MMP-3, or MMP-7 to the harvested media reduced the amount of the 30-kD proBDNF (Fig. 1A). The 28-kD proBDNF from endothelial cell media was also cleaved by plasmin and MMP-7 but not MMP-3, possibly reflecting differences in expression of tissue inhibitors of metalloproteinases by endothelial and 293 cells (Fig. 1B). MMP-7 treatment of secreted proBDNF from 293 or endothelial cells yielded a 17-kD proBDNF form, suggesting that cleavage occurred approximately 90 amino acids from the NH2-terminus (11). Incubation of plasmin or MMP-7 with specific inhibitors verified the specificity of these proteolytic events.
Cleavage of secreted proBDNF and proNGF by candidate proteinases. Harvested media of BDNF-expressing 293 cells (A), or murine endothelial cells (B) was incubated with the indicated proteinases, and cleavage products were probed in Western blot analysis with antibody to BDNF (16). To confirm enzymatic specificity, plasmin was incubated with the inhibitor aprotinin or MMPs were incubated with a peptide inhibitor. (C) Endothelial cells expressing BDNF were incubated for 48 hours in the presence of plasmin with or without (+/−) aprotinin. Media were harvested and analyzed by Western blot. (D) Purification of mature NGF (M) and cleavage-resistant proNGF (Pro) from media from 293 cells stably expressing native NGF, of a furin-resistant form of NGF with a His tag at the COOH-terminus, or of the vector alone (V) with the use of Ni-column chromatography (17). Proteins eluted with imidazole were analyzed by SDS–polyacrylamide gel electrophoresis (SDS-PAGE) and silver staining. (E) Purified furin-resistant proNGF was incubated with the indicated proteinases with or without (+/−) inhibitors and proteolytic products were detected by Western blot analysis with antibody to NGF. The inability of MMP-3 to cleave proNGF, as compared to proBDNF, may reflect sequence differences at the putative MMP site.
To confirm the extracellular cleavage of proBDNF, endothelial cells expressing BDNF were incubated with plasmin before media collection (Fig. 1C). Inhibition of plasmin activity by the cell-impermeable inhibitor aprotinin reduced cleavage of proBDNF, suggesting that these higher molecular weight forms are released from cells and then cleaved by plasmin at the cell surface. Thus, in addition to furin, cleavage of proneurotrophins can be regulated selectively by plasmin and MMPs.
To analyze the cleavage of secreted proNGF, a point mutation in the dibasic site used by furin was generated to impair intracellular proteolysis (17). Furin-resistant proNGF was expressed in 293 cells (Fig. 1D), and, when secreted, was cleaved by plasmin to a 13-kD form (Fig. 1E). In addition, incubation with MMP-7 but not MMP-2, -3, or -9 resulted in cleavage of the 30-kD proNGF to the 17-kD form (Fig. 1E).
To evaluate whether proneurotrophins selectively bind and activate Trk and p75NTR receptors, furin-resistant proNGF and mature NGF were purified from the media of transfected 293 cells (Fig. 1D), and their ability to bind TrkA or p75NTR was determined (Fig. 2) (18). Cells expressing either TrkA or p75NTR were treated with radiolabeled mature NGF and either unlabeled proNGF or unlabeled mature NGF. Unlabeled mature NGF displaced binding of radiolabeled mature NGF from TrkA expressing cells with an IC50 (concentration of inhibitor that reduced binding by 50%) of 1.2 nM, in agreement with equilibrium binding studies [dissociation constant (K d) = 1 nM] (19,20). However, furin-resistant proNGF was ineffective in displacing mature NGF bound to TrkA, with an IC50 of greater than 5 nM. Mature NGF also displaced radiolabeled mature NGF bound to p75NTR with an IC50 of 1.0 nM, consistent with equilibrium binding studies (K d = 1.3 nM) (19). However, furin-resistant proNGF bound to p75NTR with five times greater affinity than mature NGF, with an IC50 of 0.2 nM. These results demonstrate that proNGF is the preferred ligand for p75NTR but not TrkA, and they suggest that p75NTR-dependent cellular processes might be more efficiently induced by proNGF than by mature NGF.
Binding analysis of mature and cleavage-resistant proNGF to TrkA or p75NTR receptors. Dilutions of purified furin-resistant proNGF (Fig. 1D), or commercial mature NGF were assayed for their ability to displace 125I-radiolabeled commercial mature NGF (1 nM) from 293 cells expressing TrkA (A) or from A875 cells expressing p75NTR (B) (28). Squares, commercial mature NGF; open circles, cleavage resistant proNGF. Competition with purified mature NGF yielded results that were comparable to those obtained with commercial NGF (18).
To determine whether the higher affinity binding of proNGF to p75NTR resulted in enhanced p75NTR-mediated apoptosis, we used a vascular smooth muscle cell line expressing p75NTR but not Trk receptors, which exhibits dose-dependent apoptosis upon addition of NGF (21). Cells were exposed to purified mature NGF or furin-resistant proNGF (22). Treatment of cells with mature NGF from commercial sources (predominantly mature NGF) induced apoptosis in 20% of the cells at 2 nM concentration (∼50 ng/ml) (5, 21) (Fig. 3A). In contrast, treatment of cells with proNGF was at least 10 times more potent than mature NGF in inducing apoptosis in 18% of cells at 0.1 nM. This result, together with the binding data (Fig. 2B), suggests that occupancy of less than 30% of p75NTR receptors with furin-resistant proNGF can induce apoptosis.
Distinct biological activities are elicited by mature and cleavage-resistant proNGF. (A) Induction of apoptosis in p75NTR-expressing smooth muscle cells (p75Ts-Tag-SMC) (21) by mature NGF and furin-resistant proNGF. Cells were cultured at 39.5°C for 96 hours to permit differentiation and were then incubated with the indicated concentration of commercial mature NGF (open circles), purified mature NGF (solid circles), purified furin-resistant proNGF (solid squares), or column eluates from vector transfected cells (open squares). After 18 hours, cells were stained with 4′,6′-diamidino-2-phenylindole (DAPI) to visualize nuclei, were fixed, and were subjected to TUNEL analysis (Roche Molecular Biochemicals, Indianapolis, Indiana). At least 400 cells per condition were counted, and the result is representative of two independent experiments with different preparations of NGFs. (B) Mature NGF is more active than proNGF in inducing TrkA phosphorylation. TrkA-expressing 293 cells were cultured in reduced serum for 18 hours, followed by either the addition of diluent (–) or of mature NGF, proNGF, or commercial NGF (cNGF) at the indicated concentrations for 10 min. Immunoprecipitations using antibody to Trk were analyzed by Western blot with antibody to phosphotyrosine. (C and D) Effects of mature and cleavage resistant proNGF on neuritogenesis using PC12 cells (C) or SCG neurons (D). Cells were treated with mature NGF (solid circles), cleavage-resistant proNGF (solid squares), or commerical mature NGF (open circles) at the indicated concentration or with the control column eluates from vector-transfected cells (open squares) for 48 hours in serum-free media, and cells were evaluated for neurite processes greater than one cell body in length. SCG neurons were treated with 0.4 nM mature or proNGF for 36 hours and were fixed. Trk was detected using rhodamine-conjugated antibody to Trk, TUNEL analysis performed with fluorescein isothiocyanate–deoxyuridine 5′-triphosphate (FITC-dUTP), and nuclei were detected by DAPI staining. The percentage of neurite-expressing cells (white typeface) and apoptotic cells (green typeface) is indicated. Cells treated with diluent alone yielded 30 ± 10% apoptosis. Scale bar, 30 μm. At least 200 cells per condition were scored, and results are representative of three independent experiments using different preparations of NGFs.
To determine the relative activities of mature NGF and furin-resistant proNGF in activating Trk-mediated cellular responses, we assessed TrkA autophosphorylation in dose-response studies (22) (Fig. 3B). Mature NGF and commercial NGF induced TrkA autophosphorylation at a concentration of 0.2 nM. However, furin-resistant proNGF did not induce TrkA phosphorylation even at concentrations of 1 nM, consistent with the observed reduction in TrkA binding (Fig. 2A). To assess the functional consequences of proNGF-TrkA interactions, TrkA-expressing PC12 cells and dissociated superior cervical ganglia (SCG) neurons were used in neurite outgrowth assays (Fig. 3, C and D). Treatment of PC12 cells with purified mature NGF or commercial NGF induced neurite elongation at 0.2 to 0.4 nM concentration. However, reduced neurite outgrowth was observed in PC12 cells or SCG neurons that were treated with the furin-resistant proNGF at concentrations of 0.2 to 0.8 nM, consistent with prior reports that proneurotrophins are less active than the mature forms in promoting Trk-mediated neuronal survival (6) and in activating Trk receptors (3). Treatment of SCG neurons, which coexpress both p75NTR and TrkA (23), with proNGF resulted in cell death (Fig. 3D). Taken together, these results suggest that the cleavage-resistant proform of NGF is a high-affinity, functional ligand for the pro-apoptotic p75NTR receptor, whereas the proteolytically cleaved mature NGF is the preferred ligand for TrkA.
These studies shed light on the often-conflicting roles for p75NTR in mediating apoptosis (23–25) and in augmenting Trk-induced survival and differentiation (5). The selectivity of proNGF for p75NTR suggests that its local secretion may determine whether apoptotic or survival actions predominate. Thus, despite widespread p75NTR expression, an evaluation of proneurotrophin expression should clarify the spatial and temporal restriction of apoptosis in injured neuronal tissues (after seizures, inflammation, or degeneration) and in injured blood vessels where neurotrophins, p75NTR, and Trk receptors are coexpressed (1, 26). The regulated activity of proteases such as plasmin and selective MMPs may further define these proapoptotic or prosurvival effects of neurotrophins. Examination of potential biological activities for the precursor forms of other growth factors, such as glial derived neurotrophic factor and neuregulins, whose cleaved forms bind to multicomponent receptors, may uncover differential activation of individual receptor components.
↵* To whom correspondence should be addressed: E-mail: blhempst{at}med.cornell.edu
