Abstract
Myelin-derived Nogo-A protein limits axonal growth after CNS injury. One domain binds to the Nogo-66 receptor to inhibit axonal outgrowth, whereas a second domain, Amino-Nogo, inhibits axonal outgrowth and cell adhesion through unknown mechanisms. Here, we show that Amino-Nogo inhibition depends strictly on the composition of the extracellular matrix, suggesting that Amino-Nogo inhibits the function of certain integrins. Amino-Nogo inhibition can be partially overcome by antibodies that activate integrin β1 or by the addition of Mn2+, an integrin activator. Furthermore, Amino-Nogo reduces focal adhesion kinase activation by fibronectin. Analysis of various cell lines reveals that αvβ3, α5, and α4 integrins are sensitive to Amino-Nogo, but α6 integrin is not. Both αv and α5 integrins have widespread expression in adult brain and are found in axonal growth cones. Thus, inhibition of integrin signaling by Amino-Nogo contributes to the failure of CNS axon regeneration.
Introduction
The adult mammalian CNS exhibits a severely limited degree of anatomical rearrangement in response to a range of perturbations. This fact leads to persistent clinical deficits after neurological injury by preventing significant nerve fiber regeneration. Among the factors limiting axonal growth in the adult CNS is myelin, and Nogo is one of the proteins produced by oligodendrocytes that inhibits axonal growth (Chen et al., 2000; GrandPre et al., 2000; Huber and Schwab, 2000; Liu et al., 2006). Alternative promoter usage and differential splicing of the nogo gene give rise to three major transcripts (Nogo-A, -B, and -C). The loop between the two hydrophobic domains of Nogo (Nogo66) is common to all three isoforms, and it binds to a glycosylphosphatidylinositol-linked receptor (NgR) to inhibit axonal outgrowth (Fournier et al., 2001). Nogo-A contains a unique N-terminal domain (Amino-Nogo) that has been shown to inhibit both axon growth and fibroblast spreading in vitro (Fournier et al., 2001; Oertle et al., 2003). However, the mechanisms of Amino-Nogo inhibition remain unknown. Studies have suggested the existence of a receptor or receptor complex on the surface of responsive fibroblasts and in the CNS, but the identity of the receptor has not been determined (Oertle et al., 2003). Despite this uncertainty about Amino-Nogo action, antibodies directed against this domain have been effective in promoting recovery from neurological injury in rodent and primate preclinical studies (Bregman et al., 1995; Thallmair et al., 1998; Liebscher et al., 2005; Freund et al., 2006), and clinical trials of anti-Nogo-A antibodies are underway.
Here, we sought to define the molecular pathways by which Amino-Nogo inhibits cell spreading and neurite outgrowth. Identification of a high-affinity cell surface-binding protein is complicated by the observation that the ligand must be presented in a high-molecular-weight clustered form to exert biological activity, and by the fact that multiple cell types show equally robust Amino-Nogo responses. Because Amino-Nogo inhibits cell adhesion and integrins play a principal role as cell adhesion receptors, we examined an interaction between Amino-Nogo signaling and integrin signaling.
Integrins are cell surface glycoproteins composed of noncovalently linked α and β heterodimers that mediate cell–cell and cell–extracellular matrix (ECM) interactions (Hynes, 1992). Integrin engagement of ECM ligands results in the formation of adhesion complexes that provide a coupling to the actin cytoskeleton that is necessary for cell spreading and for the force generation during growth cone advance (Suter and Forscher, 1998; Beningo et al., 2001). To date, 18 different α and 8 different β mammalian integrin subunits have been identified to form 24 recognized αβ heterodimers. The interaction between integrins and ligands are diverse because some integrins bind several different ligands and some ECM ligands also bind to multiple integrins. Focal adhesion kinase (FAK) becomes activated after integrin activation and has been shown to play a key role in growth cone dynamics and axon pathfinding (Robles and Gomez, 2006).
Here we show that Amino-Nogo binds to and selectively blocks signal transduction initiated by certain integrins. The widespread expression of Nogo-sensitive integrin subunits accounts for the presence of Amino-Nogo responses in diverse cell types. Identification of integrins as mediators of Amino-Nogo inhibition of axon growth provides a basis to develop alternate means to block Amino-Nogo effects.
Materials and Methods
Cell culture and reagents.
COS-7, Jurkat, and CHO-K1 cells were obtained from American Type Culture Collection (Manassas, VA). HUVEC cells were obtained from cell culture facility at Yale University. CS1-β3 cells were kindly provided by Dr. Caroline Damsky (University of California at San Francisco, San Francisco, CA) and Dr. David Cheresh (University of California at San Diego, San Diego, CA). Integrin β1-activating antibodies and α5- and αvβ3-blocking antibodies were purchased from Millipore (Billerica, MA). Fibronectin, vitronectin, and poly-d-lysine were purchased from Sigma-Aldrich (St. Louis, MO). Mouse laminin-1 was from Invitrogen. Collagen (type I, rat tail) was obtained from BD Biosciences (San Jose, CA). VCAM-1-Ig4 fusion protein, PS2 antibodies, and Bio5192 were kindly provided by Dr. Lingling Chen and Dr. Jane K. Relton from BiogenIdec (Cambridge, MA).
Recombinant Fc fusion protein production.
For mammalian expression, human Amino-Nogo X fragment (nucleotides 540–2592) and Δ20 fragment (nucleotides 1704–2247) were cloned into pIg-tail expression system (Novagen, Madison, WI). Plasmids were transfected into HEK293T cells, and recombinant proteins were purified by protein A agarose (Sigma-Aldrich). For expression in insect cells, human Amino-Nogo X fragment (nucleotides 540–2592) and human Fc sequence were cloned in frame into pAcHLT-C baculovirus transfer vector (BD Biosciences). Baculoviruses expressing Amino-Nogo-Fc (Ng-Fc) were generated, amplified, and used to infect Sf9 insect cells. Ng-Fc was purified from the infected Sf9 cells by binding to Ni column (Invitrogen). Ng-Fc was dialyzed in PBS and clustered using goat anti-human IgG (Sigma-Aldrich) before the assays.
Cell adhesion assays.
ECM molecules of indicated concentrations were coated on 96-well tissue culture dish (Falcon, Franklin Lakes, NJ) for 1–2 h at room temperature (RT). Wells were then blocked with PBS + 3% heat-inactivated BSA for 2 h at RT. Cells were trypsinized, washed with DMEM (Invitrogen), and plated in DMEM + 25 mm HEPES. Clustered Fc or Ng-Fc was added to the media at the time of plating. After 120 min of incubation, plates were washed three times with DMEM + HEPES before fixation and staining with Hoescht and rhodamine–phalloidin (Invitrogen). All data shown are mean ± SEM (n = 2–8).
FAK phosphorylation assay.
After the adhesion assay, floating cells were collected and lysed together with attached cells in 50 mm Tris, pH 8.0, 150 mm NaCl, 1% Triton X-100 with standard proteinase inhibitors and phosphatase inhibitors. To immunoprecipitate FAK, rabbit anti-FAK antibodies (Santa Cruz Biotechnology, Santa Cruz, CA; catalog #c903) were incubated with the lysate followed by incubation with protein A/G beads. After three washes, beads were boiled and subjected to SDS-PAGE. Mouse anti-FAK (Millipore) and rabbit anti-phospho-FAK (Tyr 397) antibodies (Sigma-Aldrich) were used to blot total FAK or phospho-FAK, respectively. Image J software was used to quantify intensities of the bands on immunoblot. Percentage of FAK activation were calculated by dividing phospho-FAK signal with total FAK signal and standardized by using BSA as 0% and PBS control as 100%. Data shown are mean ± SEM (n = 3). p value was calculated using a two-tailed Student's t test.
Neurite outgrowth assay.
Dissociated embryonic day 10 (E10)–E11 chick DRG neurons in Neurobasal + B27 +NgF media or postnatal day 6 (P6)–P10 rat DRG neurons in Neurobasal-A + B27 media were plated onto 96-well tissue culture dishes coated with indicated ECM substrates and grown for 16 h. Clustered Fc or Ng-Fc was added to the media at the time of plating. Neurons were fixed and stained with rabbit anti-βIII tubulin antibodies (Covance, Princeton, NJ). Neurite length was analyzed by the Imagexpress imaging system and software (MDS Analytical Technologies, Sunnyvale, CA). All data shown are mean ± SEM [n = 2–12 for chick dorsal root ganglia (cDRGs) and n = 6–18 for rat dorsal root ganglia (rDRGs)].
Integrin staining.
Rabbit polyclonal antibodies for integrin α5 and αv were purchased from Millipore and used to stain dissociated E10–E11 chick DRG neurons or adult mouse brain sections.
Immunoprecipitation.
Adult wild-type or nogoAB−/− (Kim et al., 2003) mouse brains were homogenized in 50 mm Tris, pH 8.0, 150 mm NaCl, and 1% Triton X-100 with proteinase inhibitors and spun at 20,000 × g for 20 min. The soluble proteins were immunoprecipitated at 4°C using indicated antibodies, incubated with protein A/G beads (Santa Cruz Biotechnology), and washed three times with 50 mm Tris, pH 8.0, 150 mm NaCl, and 1% Triton X-100. The presence of Nogo in the immunoprecipitates was detected using rabbit anti-Nogo-A antibodies (Wang et al., 2002).
Results
Amino-Nogo inhibits COS-7 adhesion on fibronectin
To determine whether Amino-Nogo alters integrin signaling, we examined whether Amino-Nogo inhibition of cell adhesion can be modulated by different ECM substrates. We generated recombinant Amino-Nogo fragments fused to Fc because clustering has been shown to be important to Amino-Nogo activity (Fournier et al., 2001) and the Fc domain dimerizes by disulfide bonds. Surprisingly, we found that COS-7 adhesion on fibronectin, but not laminin or collagen, is inhibited by clustered Amino-Nogo (Fig. 1A). The X-fragment of Amino-Nogo (amino acids 181–864; X-Fc) is more inhibitory than the Δ20 fragment (amino acids 568–749; 20-Fc) (Fig. 1), suggesting that there exists a more extended inhibitory domain in the X fragment beyond the Δ20 segment. Clustered Amino-Nogo X fragment fused to Fc (referred to as Ng-Fc from here on) inhibits COS-7 adhesion on fibronectin with an IC50 of ∼40 nm (Fig. 2A). Amino-Nogo significantly inhibits COS-7 adhesion over a wide range of fibronectin concentrations, although the inhibition decreased at higher concentration of fibronectin (Fig. 2B). The dose–response curves for Amino-Nogo inhibition were determined to examine the possibility that lack of Amino-Nogo inhibition on laminin and collagen is caused by high concentration of laminin or collagen used in the initial assay. However, at all concentrations tested for collagen and laminin, Amino-Nogo is not inhibitory, with the possible exception of very low concentration of collagen (Fig. 2C,D).
Amino-Nogo inhibits COS-7 adhesion on fibronectin, but not on laminin or collagen. A, Trypsinized COS-7 cells were plated on fibronectin- (FN; 1 μg/ml), laminin- (LAM; 40 μg/ml), collagen- (CL; 50 μg/ml), and poly-d-lysine (PDL; 100 μg/ml)-coated 96-well tissue culture dishes in the presence of 40 nm Fc control, Fc fused Amino-Nogo X fragment (X-Fc), or Δ20 fragment (Δ20-Fc) in DMEM. B, Quantification of number of attached cells for experiment in A. C, Quantification of average cell area for experiment in A.
Dose–response curve of Amino-Nogo inhibition on COS-7 cells. A, Trypsinized COS-7 cells were plated on fibronectin (1 μg/ml)-coated dishes in the presence of increasing concentration of clustered Fc or Ng-Fc purified from insect cells. The number of attached cells was plotted as a function of Fc protein concentration. B, The numbers of attached cells expressed as percentage of adhesion was plotted as a function of fibronectin concentration in the presence of 40 nm clustered Fc or Ng-Fc. C, The number of attached cells expressed as percentage of adhesion was plotted as a function of collagen concentrations in the presence of 40 nm clustered Fc or Ng-Fc. D, The number of attached cells expressed as percentage of adhesion was plotted as a function of laminin concentration in the presence of 40 nm clustered Fc or Ng-Fc. E, Trypsinized COS-7 cells were plated on fibronectin- (1 μg/ml), collagen- (1 μg/ml), or laminin (20 μg/ml)-coated dishes in the presence of 20 μg/ml control, α1 integrin-blocking antibody Ha31/8 (α1), α2 integrin-blocking antibody Ha1/29 (α2), α3 integrin-blocking antibody P1B5 (α3), a5 integrin-blocking antibody SAM-1 (α5), α6 integrin-blocking antibody GoH3 (α6), or β1 integrin-blocking antibody Ha2/5 (β1). The number of attached cells was plotted as percentage of control in each condition.
Selective inhibition of adhesion by Amino-Nogo on one ECM substrate but not others might be mediated by selective Amino-Nogo interaction with ECM substrates or integrin receptor pathways. If Amino-Nogo interacts with ECM substrate directly, then Amino-Nogo inhibition would be expected to correlate best with ECM rather than with cell type. On the other hand, if Amino-Nogo alters the function of ECM receptors such as integrins, then Amino-Nogo inhibition would be expected to vary between cell types, because multiple integrins can mediate adhesion to any one ECM substrate, and different cell types use different integrins. We found that Amino-Nogo inhibits PC12 cell adhesion to laminin and collagen even though COS-7 adhesion to laminin and collagen is not altered by Amino-Nogo (data not shown). Thus, the action of Amino-Nogo is not specific for certain ECM substrates but may be specific for certain integrin receptor signaling pathways.
To determine which integrins mediate fibronectin-, laminin-, or collagen-dependent adhesion in COS-7 cells, we plated COS-7 cells onto fibronectin-, laminin-, or collagen-coated dishes in the presence of specific integrin-blocking antibodies. Integrin β1-blocking antibodies abolished fibronectin- and laminin-dependent adhesion and reduced collagen-dependent adhesion by 80%, suggesting that integrin β1 is required for adhesion to all three ECM ligands (Fig. 2E). The presence of integrin α5-blocking antibody reduced the adhesion of COS-7 cells to fibronectin by >80%, but had no or minor effect on laminin- or collagen-dependent adhesion, suggesting that the principal integrin contributing to fibronectin adhesion in COS-7 cells is α5β1 (Fig. 2E). α6β1 integrin is the primary integrin receptor for laminin in COS-7 cells because integrin α6-blocking antibody almost abolished laminin-dependent adhesion in COS-7 cells (Fig. 2E). The integrin receptor for collagen is still not clear from this analysis. Because Amino-Nogo inhibits fibronectin adhesion but not laminin adhesion in COS-7 cells, the data show that α5β1 integrin-associated adhesion pathways but not α6β1 integrin-associated adhesion mechanisms are sensitive to Amino-Nogo inhibition.
Amino-Nogo inhibits integrin signaling
To further demonstrate that Amino-Nogo affects integrin signaling, we assessed whether downstream integrin signaling pathway is altered by Amino-Nogo. FAK is phosphorylated at tyrosine 397 when integrin is activated by ECM substrate. Amino-Nogo treatment significantly suppresses fibronectin-induced FAK activation in COS-7 cells compared with control (Fig. 3A) (p < 0.01; 37% FAK activation in the presence of Amino-Nogo compared with control). These data further support the hypothesis that Amino-Nogo inhibits integrin signaling.
Amino-Nogo inhibits integrin signaling. A, Amino-Nogo affects FAK activation. Trypsinized COS-7 cells were plated on BSA only (−) or fibronectin (5 μg/ml; FN)-coated dishes in the presence of PBS (−), 120 nm Fc, or 120 nm Ng-Fc. Total FAK or phospho-FAK was analyzed using specific antibodies. B, Integrin activation can partially overcome Amino-Nogo inhibition in COS-7 cells. Trypsinized COS-7 cells were plated on fibronectin (1 μg/ml)-coated dishes in the presence of 120 nm Fc or Ng-Fc. MnCl2 (1 mm) and 20 μg/ml indicated integrin β1-activating antibodies were added when indicated. C, Integrin activation can partially overcome Amino-Nogo inhibition in Jurkat cells. Jurkat cells were plated on fibronectin (10 μg/ml)-coated dishes in the presence of 40 nm Fc or Ng-Fc. MnCl2 (1 mm) and 20 μg/ml indicated integrin β1-activating antibodies were added when indicated. Student's t test revealed a significant difference in MnCl2 and β1-activating antibodies treated conditions versus control condition (*p < 0.05; **p < 0.01).
We next examined whether integrin activation can overcome Amino-Nogo inhibition. We tested the effect of integrin-activating antibodies and Mn2+ on Amino-Nogo inhibition. Mn2+ has been shown to enhance integrin-ligand interaction (Bazzoni et al., 1995, 1998). Both Mn2+ and integrin β1-activating antibodies P4G11, HUTS4, and B44 can partially but significantly overcome Amino-Nogo inhibition of COS-7 adhesion on fibronectin (Fig. 3B), indicating that perturbation of integrin signaling plays an important role in Amino-Nogo inhibition.
Integrins inhibited by Amino-Nogo
Multiple integrins have been shown to mediate cell adhesion to fibronectin, including α5β1, α4β1, αvβ1, αvβ3, and αIIbβ3. To further determine integrin specificity of Amino-Nogo inhibition, we investigated the Amino-Nogo effect on adhesion by different cell lines. The adhesion of Jurkat T-cell on fibronectin has been shown to depend on α4β1 integrin and can be blocked by PS2, an α4 integrin-blocking antibody, or Bio5192, a small molecule inhibitor of α4β1 (Leone et al., 2003) (Fig. 3C). Amino-Nogo almost completely abolished Jurkat cell adhesion to fibronectin (Fig. 3C), suggesting that α4β1 integrin signaling can be inhibited strongly by Amino-Nogo. Furthermore, antibodies that activate β1 integrin, including P4G11, B44, and TS2/16 can substantially overcome Amino-Nogo effect (Fig. 3C), consistent with the data obtained from COS-7 cells.
To further confirm that α5β1 integrin can be inhibited by Amino-Nogo, we tested whether Amino-Nogo inhibits adhesion of CHO-K1 cells on fibronectin, because α5β1 integrin has been shown to be the major fibronectin receptor in CHO-K1 cells (Schreiner et al., 1989). Indeed, antibodies that block α5 integrin (SAM-1) abolished adhesion of CHO-K1 cells on fibronectin, but blocking antibodies for αvβ3 integrin (LM609) did not do so (Fig. 4B). Amino-Nogo inhibits the attachment and spreading of CHO-K1 cells on fibronectin, which can be overcome by the presence of Mn2+ and partially overcome by integrin β1-activating antibody B44 (Fig. 4A,B). These data demonstrate that α5β1 integrin signaling is inhibited by Amino-Nogo.
Amino-Nogo inhibits α5 integrin. A, Amino-Nogo inhibits CHO-K1 adhesion on fibronectin. Trypsinized CHO-K1 cells were plated on fibronectin (1 μg/ml)-coated dishes in the presence of 40 nm Fc or Ng-Fc. MnCl2 (1 mm) and 20 μg/ml indicated antibodies were added when indicated. B, The number of attached cells was expressed as percentage of Fc control on fibronectin without MnCl2 or antibodies added.
To determine whether αvβ3 integrin is affected by Amino-Nogo, we examined the response of HUVEC cells to Amino-Nogo. Adhesion of HUVEC cells to vitronectin is known to depend on αvβ3 integrin (Cheresh, 1987). Amino-Nogo reduced both the attachment and spreading of HUVEC cells on vitronectin, which can be overcome by the addition of Mn2+ (Fig. 5A). αvβ3 integrin-blocking antibodies LM609 interfere with the adhesion of HUVEC cells on vitronectin, verifying the specificity of the response. A combination of LM609 and Amino-Nogo seems to have additive affect, suggesting that they inhibit αvβ3 integrin through different mechanisms (Fig. 5A). To further confirm that αvβ3 integrin is inhibited by Amino-Nogo, we determined the response of CS1-β3 cells to Amino-Nogo. CS1 is a hamster melanoma cell line that grows in suspension on vitronectin substrate (Thomas et al., 1993) because it lacks receptors for vitronectin. Transfection of β3 integrin into CS1 allows these cells (CS1-β3) to express functional vitronectin receptor (αvβ3 integrin) and attach to vitronectin (Filardo et al., 1995). Adhesion on vitronectin is dependent on αvβ3 integrin in CS1-β3 cells, and can be blocked by LM609 antibodies (Fig. 5B). Amino-Nogo inhibits the attachment of CS1-β3 cells on vitronectin by 60%, which can be partially overcome by the presence of Mn2+ but not by integrin β1-activating antibodies P4G11, HUST4, and B44 (Fig. 5B). Thus, αvβ3 integrin action is inhibited by Amino-Nogo in CS1-β3 cells.
Amino-Nogo inhibits αv integrin. A, Amino-Nogo inhibits HUVEC adhesion on vitronectin. Trypsinized HUVEC cells were plated on vitronectin (1 μg/ml)-coated dishes in the presence of 40 nm Fc or Ng-Fc. MnCl2 (1 mm) and 20 μg/ml indicated antibodies were added when indicated. B, Amino-Nogo inhibits CS1-β3 adhesion on vitronectin. Trypsinized CS1-β3 cells were plated on vitronectin (1 μg/ml)-coated dishes in the presence of 40 nm Fc or Ng-Fc. MnCl2 (1 mm) and 20 μg/ml antibodies were added when indicated. Student's t test showed a significant difference in MnCl2-treated conditions versus control condition (*p < 0.05) but no significant difference (n.s.) in β1-activating antibodies-treated condition.
Amino-Nogo inhibition on axon outgrowth is modulated by ECM substrates
Amino-Nogo inhibits both cell adhesion and axon outgrowth. To determine whether Amino-Nogo inhibition of outgrowth also involves integrin inhibition, we examined whether Amino-Nogo inhibition of axon outgrowth can be modulated by ECM substrates. Neurons from embryonic cDRGs were plated onto fibronectin-, vitronectin-, laminin-, collagen-, or VCAM-1-coated dishes in the presence of Fc control or Ng-Fc. VCAM-1 is a specific ligand for α4 integrin (Yusuf-Makagiansar et al., 2002), and there is evidence documenting expression of α4 integrin in DRG neurons and regenerating nerves (Vogelezang et al., 2001). We found that Amino-Nogo exhibits greater inhibitory activity on fibronectin, vitronectin, and VCAM-1 than on collagen or laminin (Fig. 6A,B). Amino-Nogo is strongly inhibitory on fibronectin and VCAM-1 at all the concentrations of fibronectin and VCAM-1 tested (Fig. 6B). In contrast, Amino-Nogo inhibits neurons plated on low concentrations of laminin, but not on higher concentrations of laminin (Fig. 6B). The inhibition of neurons plated on collagen by Amino-Nogo shows similar concentration dependence (Fig. 6B). These data suggest that Amino-Nogo inhibition of outgrowth is also selective for certain integrin signaling pathways. A dose–response curve for Amino-Nogo inhibition on DRG neurons plated on fibronectin or VCAM-1 shows an IC50 of ∼40 nm, similar to that in COS-7 cells (Fig. 6C).
Amino-Nogo inhibition of axon outgrowth is modulated by ECM substrates. A, Dissociated E9-E11 cDRGs neurons were plated on fibronectin (10 μg/ml), vitronectin (5 μg/ml), laminin (10 μg/ml), collagen (50 μg/ml), or VCAM-1 (5 μg/ml)-coated 96-well tissue culture dishes in the presence of 40 nm Fc or Ng-Fc. B, Quantification of average neurite length per neuron expressed as percentage of Fc control on fibronectin 10 μg/ml (FN10) for experiment in A. Numbers indicate the concentration of ECM molecules used in each condition. C, Amino-Nogo inhibits cDRG outgrowth on fibronectin and VCAM-1 with IC50 of ∼40 nm. Dissociated E9-E11 cDRGs neurons were plated on fibronectin- (10 μg/ml) or VCAM-1 (5 μg/ml)-coated 96-well tissue culture dishes in the presence of increasing concentrations of Fc or Ng-Fc. Average outgrowth per well in each condition was expressed as percentage of control and plotted as a function of Fc protein concentration. D, Amino-Nogo inhibits rDRG outgrowth on various ECM substrates. Dissociated p6-p10 rDRGs neurons were plated on ECM-coated 96-well tissue culture dishes as indicated. MnCl2 (1 mm) was added when indicated (+). Average outgrowth per well in each condition was expressed as percentage of Fc control. p values are calculated using two-tailed Student's t test.
We repeated the outgrowth assay with postnatal rDRG neurons. Amino-Nogo shows greater inhibition of rDRG outgrowth on collagen or VCAM-1 than on fibronectin, laminin, or vitronectin (Fig. 6D). The differences between Amino-Nogo effect on COS-7, embryonic cDRG, and postnatal rDRG neurons is likely to be related to different integrin subunit expression levels in these types of cells. The presence of Mn2+ partially overcomes Amino-Nogo inhibition on fibronectin, collagen, and VCAM-1 (Fig. 6D), suggesting that integrin activation can overcome Amino-Nogo effect in axon outgrowth.
If the inhibition of DRG axon growth by Amino-Nogo is mediated by the reduced activation of certain integrins, then these ECM receptors should be found in DRG axons and growth cones. Indeed, immunohistologic analysis reveals the presence of at least α5 and αv integrin subunits in the growth cones of chick DRG axons (Fig. 7A). Previous studies have demonstrated that β1 integrin subunits are expressed in these axons (Tomaselli et al., 1993). Thus, integrins are available at the right context to be modulated by Amino-Nogo signaling.
Amino-Nogo expression in growth cones and CNS. A, Immunofluorescence staining of αv and α5 integrin in E10 cDRG growth cones with integrin staining in green and βIII tubulin staining in red. B, Immunofluorescence staining of αv and α5 integrin in adult mouse brain cortex and hippocampus. C, Immunofluorescence staining of αv and α5 integrin in adult mouse brain cortex with integrin staining in green and βIII tubulin staining in red. D, Nogo interacts with integrins in the brain. Adult wild-type (WT) or nogoAB −/− (KO) mouse brain lysates were immunoprecipitated in Triton X-100-containing buffer using rabbit anti-Nogo-A, rat IgG, rat anti-αv integrin, mouse anti-αvβ3 integrin, or mouse anti-α5 integrin antibodies. The presence of Nogo was detected using anti-Nogo-A antibodies (NgA).
α5 and αv integrin expression in CNS
The notion that Amino-Nogo contributes to limiting axonal growth in the adult CNS is supported by studies from Schwab and colleagues with antibodies directed against the Amino-Nogo (Liebscher et al., 2005; Freund et al., 2006). This led us to consider which integrin subunits that mediate Amino-Nogo-sensitive cell adhesion responses are available in the adult brain and spinal cord. Although α4 integrin expression has been reported in DRG neurons and regenerating sciatic nerves (Vogelezang et al., 2001), we failed to detect any discernible neuronal expression by immunofluorescence staining in the adult brain and spinal cord (data not shown). Thus, although α4 integrin can be inhibited by Amino-Nogo in vitro, this is unlikely to play a significant role in limiting adult CNS axonal growth in vivo. In contrast, both α5 integrin and αv integrin have high and widespread expression in CNS (King et al., 2001) (Fig. 7B,C). Therefore, they provide likely sites for Amino-Nogo action in vivo.
Nogo interacts physically with integrins
Because Amino-Nogo inhibits integrin signaling, we asked whether there is physical interaction between Nogo and integrins. Brain lysates from wild-type or Nogo-A/B knock-out mice were subjected to immunoprecipitation (IP) using αv and α5 integrin antibodies. Nogo-A protein is present in αv, αvβ3, and α5 integrin immunoprecipitates from wild-type brain lysates but not in control IgG immunoprecipitates (Fig. 7D), indicating that Nogo-A exists in a complex with these integrins in the brain. Nogo-A associates minimally if at all with integrin subunits under more stringent conditions in SDS-containing RIPA buffer (data not shown), suggesting an indirect and/or weak physical complex between Nogo-A and integrins. We failed to detect integrins in the anti-Nogo-A immunoprecipitates. It is quite likely that the anti-Nogo-A antibodies used for IP disrupted the Nogo-integrin interaction because the antigenic peptide is within the active region of Amino-Nogo (Wang et al., 2002). A monoclonal antibody (11B7) recognizing the same peptide is known to be functional in vivo and to improve regeneration and locomotion of spinal cord-injured rats (Liebscher et al., 2005).
Although Amino-Nogo and integrins form a complex in brain, the interaction may or may not require other intermediate proteins. Purified Amino-Nogo and integrins were tested for a direct interaction in an ELISA format. No direct binding was detected under these conditions (data not shown). It is possible that the extracted and purified integrins have lost a conformation capable of Amino-Nogo binding or that the association between integrin and Nogo is mediated through a protein or protein complex present in many cell types.
Discussion
In this study, we demonstrate that perturbation of integrin signaling mediates Amino-Nogo inhibition of cell adhesion and axon outgrowth. First, inhibition of COS-7 adhesion by Amino-Nogo is selective for fibronectin. Second, Amino-Nogo affects FAK activation on fibronectin, and Amino-Nogo inhibition of cell adhesion can be partially overcome by integrin activation. Third, we demonstrate that Amino-Nogo can inhibit α4β1-, α5-, and αvβ3 integrin-dependent adhesion, but not α6β1 integrin-dependent adhesion. Fourth, Amino-Nogo inhibition of axon outgrowth can also be modulated by ECM substrates and partially overcome by integrin activator Mn2+, suggesting that Amino-Nogo inhibition of axon outgrowth also involves inhibition of integrin signaling. Brain α5 and αv integrin subunits may interact indirectly with Nogo-A as part of a larger protein complex.
Integrin specificity of Amino-Nogo inhibition
Amino-Nogo inhibition is specific for certain integrin signaling pathways. In COS-7 cells, fibronectin-dependent adhesion mediated by integrin α5β1 is eliminated by Amino-Nogo, but laminin- or collagen-dependent adhesion is essentially unaffected. Because α6β1 integrin mediates laminin-dependent adhesion, this analysis suggests that α6β1 integrin is resistant to Amino-Nogo inhibition, whereas the fibronectin-interacting α5β1 integrin-dependent adhesion pathway is sensitive. Analysis of Amino-Nogo action in several other cell lines demonstrates that the adhesion pathways requiring integrins α4β1, α5β1, and αvβ3 are sensitive to Amino-Nogo inhibition. α4β1 integrin-dependent adhesion probably is most sensitive to Amino-Nogo effect because Amino-Nogo inhibits ∼90% of Jurkat cell adhesion to fibronectin and ∼80% of VCAM-1-dependent outgrowth.
Although laminin and collagen receptors on COS-7 cells are resistant to Amino-Nogo inhibition, there are some laminin and collagen receptors sensitive to Amino-Nogo. DRG neurons are inhibited by Amino-Nogo when cultured on a collagen substrate or on a laminin substrate. PC12 cells are sensitive to Amino-Nogo on fibronectin, laminin, or collagen substrates (data not shown). Thus, sensitivity to Amino-Nogo is determined by the expression pattern of ECM receptors on the cell surface. Complete analysis of the multiple ECM receptors expressed on various cell lines and DRG neurons will define the spectrum of Amino-Nogo inhibition.
Mechanisms of Amino-Nogo inhibition
We demonstrate that Amino-Nogo physically interacts with α5 and αv integrins of adult brain, raising the possibility that Amino-Nogo might inhibit integrin function through direct binding. However, purified Amino-Nogo and integrin did not exhibit significant affinity for one another (data not shown). The conformation of purified, detergent-extracted integrins might not be optimal for binding to Nogo, or alternatively, other proteins might be required to mediate the functional interactions between Nogo and integrins. Association of Nogo to integrin, whether direct or indirect, might reduce the binding affinity of integrin for its ligands. Consistent with this model, increasing concentrations of ECM ligands can overcome Amino-Nogo inhibition in several cases (Figs. 2B, 6B).
If Nogo-A alters the function of certain integrin subunits indirectly, this might occur via either novel cell surface-binding sites or a nonintegrin cell adhesion receptor. Cell adhesion is a cooperative process known to involve both integrin and nonintegrin receptors, so it is likely that nonintegrin adhesion receptors play a role even in cases such as COS-7 adhesion to fibronectin, in which anti-α5 integrin receptor antibodies abolish adhesion. However, if nonintegrin receptors do play a role in mediating Nogo-A action, they must obey strict integrin subunit-defined specificities of adhesion, for example, with no effect on α6 integrin-dependent adhesion to laminin.
One study indicated that Amino-Nogo activates Rho-A (Niederost et al., 2002), whereas another study did not observe RhoA activation (Fournier et al., 2003). The Rho signaling pathway has been shown to influence the dynamics of integrin-dependent adhesion sites known as point contacts at the growth cones (Woo and Gomez, 2006). The relationship between Rho activation and inhibition of integrin signaling by Amino-Nogo remains to be determined.
In vivo significance of Amino-Nogo inhibition of integrin signaling
At least two integrin subunits (α5 and αv), which are sensitive to Amino-Nogo inhibition, are widely expressed in the adult CNS. It is well known that integrins have extremely robust effects on in vitro growth of embryonic neurons, and the vast majority of outgrowth assays in the literature have been performed under conditions that depend heavily on integrin function (Condic and Letourneau, 1997). In the adult CNS, there are few data addressing a potential role for integrins in mediating axonal growth over long distances (Condic, 2001). The connection of certain integrins with Amino-Nogo inhibition enhances the need for specific studies of integrin function in axonal regeneration and recovery after CNS injury. It is recognized that adult brain synaptic plasticity can be regulated by integrin interaction with a range of ligands, including reelin, HB-GAM (heparin-binding growth-associated molecule), Narp (neuronal activity-regulated pentraxin), tenascins, and chondroitin sulfate proteoglycans (for review, see Dityatev and Schachner, 2003). At sites of adult CNS injury, several ECM integrin ligands are strongly expressed. Astrocyte-associated fibronectin is critical for axon regeneration in adult white matter (Tom et al., 2004). Injury-site specific fibronectin may be generated by brain microglia and/or invading meningeal fibroblasts (supplemental Fig. 1, available at www.jneurosci.org as supplemental material) (Egan and Vijayan, 1991). Laminin is also present at CNS trauma sites (supplemental Fig. 1, available at www.jneurosci.org as supplemental material).
Nogo-A limitation of axonal growth in the adult CNS appears to function by two independent signaling pathways. The Nogo-66 domain enhanced by an adjacent binding site stimulates NgR in concert with the myelin proteins MAG and OMgp (Hu et al., 2005). The Amino-Nogo domain inhibition of integrin function demonstrated here is independent of this pathway. Peptide, antibody, and protein therapeutics have targeted various aspects of the Nogo/NgR pathway successfully to promote improved recovery from spinal cord injury and stroke in preclinical studies (Lee et al., 2003; Liu et al., 2006). The Amino-Nogo-sensitive integrin subunits and their ECM ligands contribute additional sites to develop therapeutic approaches to stimulate axonal growth and behavioral recovery from adult neurological deficits. Such agents may be expected to function synergistically with Nogo-66/MAG/OMgp/NgR directed reagents.
Footnotes
-
This work was supported by grants to S.M.S. from the National Institutes of Health and a fellowship to F.H. from the Paralyzed Veterans of America Foundation. We thank Drs. Caroline Damsky (University of California at San Francisco, San Francisco, CA) and David Cheresh (University of California at San Diego, San Diego, CA) for providing CS1-β3 cells, and Drs. Lingling Chen and Jane K. Relton from BiogenIdec (Cambridge, MA) for providing VCAM-1, PS2, and Bio5192.
- Correspondence should be addressed to Stephen M. Strittmatter, Department of Neurology, Yale University School of Medicine, New Haven, CT 06520. stephen.strittmatter{at}yale.edu
