Inactivation of Glycogen Synthase Kinase-3β and Up-regulation of LINGO-1 are Involved in LINGO-1 Antagonist Regulated Survival of Cerebellar Granular Neurons

Abstract

LINGO-1 has been critically implicated in the central regulation of CNS axon regeneration and oligodendrocyte maturation. We have recently demonstrated that pretreatment with LINGO-1 antagonist (LINGO-1-Fc) inhibited low potassium-induced cerebellar granular neurons (CGNs) apoptosis. In the present study, we examined the neuroprotective mechanism of LINGO-1-Fc by Western blot and in situ GST pull-down assay. CGN cultures were preincubated in medium with LINGO-1-Fc or control protein at the concentration of 10 μg/ml for 2 h and then switched to low potassium medium in the presence of corresponding proteins. Cultures were harvested at indicated time intervals for successive analysis. Several apoptosis-associated signaling factors, GSK-3β, ERK1/2, and Rho GTPases, were observed to be activated in response to potassium deprivation and the activation/dephosphorylation of GSK-3β was suppressed by LINGO-1-Fc pretreatment compared with control group. Besides, the endogenous LINGO-1 expression level of CGN cultures was augmented by low potassium stimuli and restrained by LINGO-1 antagonist treatment. Although the protein level of p75^NTR and Nogo-A were down-regulated in different patterns during apoptosis, neither of them was affected by LINGO-1-Fc application. Taken together, these results suggest a new mechanism of LINGO-1 antagonist regulated neuronal survival involving protein synthesis of LINGO-1 and inactivation of GSK-3 pathway.

Introduction

LINGO-1 is a transmembrane protein that contains 12 leucine-rich repeats (LRR) and an immunoglobulin domain. It is selectively expressed in the brain and spinal cord: neuronal LINGO-1 functions as a component of the NgR/p75 or NgR/Troy signaling complexes that regulate CNS axon growth (Mi et al. 2004), and LINGO-1 in oligodendrocyte negatively regulates the differentiation and myelination of oligodendrocyte through Fyn and RhoA GTPase (Mi et al. 2005). Soluble LINGO-1-Fc fusion protein, which acts as an antagonist for these pathways by blocking LINGO-1 binding to NgR, was reported to be effective in promoting axonal sprouting and functional recovery after spinal cord injury (Ji et al. 2006) and in promoting oligodendrocyte differentiation and myelination in vitro (Mi et al. 2005).

We have previously established that apoptosis of CGNs induced by low potassium (LK) can be inhibited and survival maintained by LINGO-1-Fc (Zhao et al. 2006). To expand upon previous observations, we analyzed the activation status of intracellular signaling molecules reported to be involved in CGN apoptosis, GSK-3β, MAPK, and Rho GTPase, and the expression levels of several related molecules, Nogo-A, p75, and LINGO-1 itself, were also compared between LINGO-1-Fc and control protein treatments.

Materials and Methods

Protein and Antibodies

LINGO-1-Fc and control Fc were prepared as described (Mi et al. 2004). Briefly, LINGO-1-Fc (residues 1-532 of human LINGO-1 fused to the hinge and Fc region of human IgG1) was expressed in CHO cells and purified on protein A Sepharose (Amersham Pharmacia, Rockville, MD). The purified protein (>95% pure) ran on SDS-PAGE with M _r = 90 kDa under reducing conditions and M _r = 180 kDa under nonreducing conditions.

To avoid reacting with the endocytic LINGO-1-Fc protein, we chose another polyclonal antibody against LINGO-1 in current study, different from the product of B&D (Zhao et al. 2007). This antibody was raised against the intracellular domain of human LINGO-1, amino acid residues 577–614, fused to GST tag, and the specificity was confirmed by immunofluorescent staining and Western blots under the control of preimmune serum. In COS-7 cells transiently transfected with full-length hlingo-1 expression plasmid (Mi et al. 2004), LINGO-1 was detected only in transfected cells with the antisera and the immunoreactant of LINGO-1 was found mainly localized on the cell membrane, which was in line with its role as a transmembrane receptor. Consistent with reports from Mi (2005), LINGO-1 antisera specifically reacted with neurons and oligodendrocytes, but not astrocytes in rat spinal cord tissue section. In Western blotting analysis, the antibody revealed a strong band at ∼80 kDa, corresponding to the results of antibody from UPSTATE that was designed against the 611–620 aa of human LINGO-1. (Data not shown)

For Western blots, the following dilutions in TBS (pH 7.4) with 0.5% (v/v) blocking solution were used: polyclonal antibody against LINGO-1, 1:12500; Nogo-A antibody (Jin et al. 2003), 1:15000; p75^NTR antibody (Promega, Madison, WI), 1:1000; MAPK antibody (Sigma, St. Louis, MO), 1:20000; anti-p-ERK1/2 (Cell Signaling Technology, Beverly, MA), 1:2000; anti-p-GSK3α/β (Ser21/9) (Cell Signaling Technology), 1:1000.

Primary Cell Culture and Protein Treatments

CGNs used in this study were prepared from 7-day-old Sprague–Dawley rat pups as described previously (Zhao et al. 2006). Briefly, rats were deeply anesthetized under standard cold operating procedure and freshly dissected cerebella was dissociated in the presence of trypsin and EDTA, then planted on poly-l-lysine coated dishes. Cells were seeded at a density of 3–4 × 10⁵ cells/cm² in Neurobasal medium supplemented with B27 and 25 mM KCl (designated as “complete medium”). After 7 days in vitro, CGN cultures were incubated in complete medium containing LINGO-1-Fc (10 μg/ml) for 2 h; afterward, complete medium was replaced by medium containing LINGO-1-Fc with [K⁺] reduced to 5 mM. Survival was assessed 8 h after K⁺-deprivation by employing a Hoechst33342 (10 μg/ml) nuclei staining method under the control of Fc protein and cells were harvested at various time points for analysis experiments. The viability of cell groups cultured in complete medium was assumed to be control and represent 100%, and the viabilities for each treatment were expressed as a percentage of control cultures.

Western Blotting

Cell cultures from different treatments were harvested in lysis buffer (PBS pH 7.4, 1% Nonidet P-40, 0.5% deoxycholate, 0.1% sodium dodecyl sulfate, 50 mM NaF, 1 mM Na₂VO₄) with a complete protease inhibitor cocktail (Boehringer Mannheim) and left on ice for 30 min. Cell lysates were clarified by centrifugation, boiled in sample buffer and separated by SDS-PAGE. The proteins were then electrotransferred to polyvinylidene difluoride membranes (PVDF, Boehringer Mannheim). Membranes were treated with 1% blocking solution in TBS (0.1 M Tris–HCl, pH 7.4, 0.1 M NaCl, 50 mM NaF, 1 mM Na₂VO₄) and incubated with primary antibody for 1 h at room temperature or overnight at 4°C (for detection of phosphorylated protein). After incubation with POD-labeled secondary antibodies, the signals were revealed by BM Chemiluminescence Western Blotting kit (Boehringer, Mannheim). Densitometry quantitation was acquired in Gel Doc 1000 system (BioRad, Hercules, CA) and analyzed using Quantity One software (BioRad).

In situ GST Pull-down Assay

GST fusion proteins of the Rhotekin GTPase binding domain (RBD) and the PAK GTPase binding domain (PBD), gifts from Dr. Xiong WC, were expressed in bacteria and purified by glutathione–agarose chromatography (Sigma). Rho GTPases activity was measured as described (Zhao et al. 2007): cell cultures were briefly washed in freshly prepared 0.05 M TBS/50 mM NaF and immediately fixed with cold methanol. Samples were then incubated with 20 μg/ml GST-RBD or GST-PBD for 1 h, washed three times in TBS/NaF and blocked in 10% donkey serum. Anti-GST antibody diluted in 1% BSA/TBS with 0.1 M l-lysine, and Alexa Fluor^® 488 (green) donkey anti-mouse IgG (Molecular Probes, Eugene, OR) were successively used to visualize active Rho. Microphotographs were obtained in Olympus BX-61 microscope system and analyzed in Image Pro-Plus software.

Results

Protective Effect of LINGO-1-Fc on Neuronal Apoptosis

In first part of the study, we examined and confirmed the neuroprotective effects of LINGO-1 antagonist by comparing viabilities between different concentrations after potassium deprivation for 8 h. Statistical analysis showed a significantly higher rate of surviving CGNs in LINGO-1-Fc group (10 μg/ml, mean ± SEM: 89 ± 4.0%) compared with Fc group (54 ± 3.8%), and LINGO-1-Fc protein prevented LK-induced apoptosis in a dose-dependent manner (Fig. 1).

Fig. 1.

Pretreatment with LINGO-1-Fc protein prevented low potassium-induced apoptosis of cerebellar granule neurons in a dose-dependent manner. CGNs were incubated with LINGO-1-Fc or Fc protein at indicated concentrations for 2 h in HK medium (25 mM KCl), and then switched to LK medium (5 mM KCl) with corresponding proteins for another 8 h. The percentage of apoptotic cells with condensed nuclei, monitored by Hoechst 33342 staining, was evaluated in each condition by counting cells from three randomly chosen fields with a 10× objective. Data were expressed as relative to the viability of HK normal cultures, which was defined as 100%. Results were analyzed by Mann–Whitney U test and presented as Mean ± SEM from three independent experiments. Asterisks indicate a significant difference from Fc control group (P < 0.01)

Inactivation of GSK3β was Associated with LINGO-1 Antagonist Regulated Neuronal Survival

We referred the results either from dose-dependent test or from previous studies that had been proved effective on promoting axonal regrowth (Ji et al. 2006) to investigate the intracellular signaling factors associated with LINGO-1-Fc treatment at the concentration of 10 μg/ml. As shown in Fig. 2, potassium deprivation treatment led to dephosphorylation/activation of GSK3β at Ser9 as early as 2 h, which persisted through the time of commitment to cell death (8 h); contrarily, treatment of granule neuron cultures with LINGO-1-Fc promoted the phosphorylation or inactivation of GSK3β compared with Fc controls at all the examined time points. Simultaneously, the lysates from granule cells cultured for 0, 2, 4, and 8 h in LK medium after LINGO-1-Fc or Fc treatment were immunoblotted with anti-phospho-ERK1/2 antibody, and the level of ERK1/2 phosphorylation was up-regulated and reached maximum at 4 h in either of the protein treatments (Fig. 2). The identical results from LINGO-1 antagonist and control treatment implicated that ERK1/2 pathway was not involved in LINGO-1 regulated neuronal survival.

Fig. 2.

Western blot analysis of GSK 3α/β and ERK1/2 activity after LINGO-1-Fc protein pretreatment and potassium withdrawal for different time courses. (a) CGNs cultured for 7 days were pretreated with LINGO-1-Fc or Fc protein, and then switched to LK medium with corresponding proteins for another 0, 2, 4, or 8 h separately. Cultures from HK (25 mM KCl) and LK medium (5 mM KCl) treatment for 8 h were included for comparison. Different cell lysates were subjected to Western blot using antibody against p-GSK3 α/β and p-ERK1/2. Blots were stripped and reprobed with an anti-MAPK antibody as internal standard. (b) Quantitation by densitometry scan revealed regulated inactivation of GSK-3β following time-dependent LINGO-1-Fc treatment. (*, Unpaired Student’s t test: P < 0.05)

Involvement of Rac/Cdc42 GTPase in Potassium Deprivation Induced CGN Apoptosis

In situ GST-pull down assay was utilized to investigate whether Rho GTPases were regulated in LINGO-1 antagonist attenuated apoptosis. CGNs possessed low basal Rac/Cdc42 GTPase activity that increased during LK-induced apoptosis (Fig. 3a) and the activation of Rac/Cdc42 GTPase was observed unaffected by either LINGO-1-Fc or Fc protein application using densitometry quantitation (Fig. 3b, c). At the same time, the staining for active Rho was very faint or absent both in depolarizing medium and in potassium withdrawing condition or after any protein treatments (Data not shown). Taken together, these results implied that small GTPases may be irrelevant to the protective effect of LINGO-1.

Fig. 3.

Rac/Cdc42 GTPases were activated in CGN cultures in response to LK stimuli but unaffected by either LINGO-1-Fc or control protein pretreatment. (a) Detected by in situ GST pull-down assay, the amount of active Rac/Cdc42 GTPases (probed with GST-PBD) in mature CGN cultures was significantly increased when switching the medium from HK (a) to LK (b), and the merged image of anti-GST with Hoechst nucleic staining showed GTPases activation in apoptotic cells as indicated by arrowheads (c). (b) CGN cultures were pretreated with LINGO-1 antagonist or control protein for 2 h and then subjected to potassium deprivation for indicated time courses. In situ GST pull-down assay detected the activated Rac/Cdc42 GTPases. Scale bar, 30 μm. (c) Densitometry quantification of Rac/Cdc42 GTPases activity. Data shown in panels are Mean ± SEM (from three independent experiments and about 200 cells for each test). Unpaired Student’s t test: P > 0.05 compared with Fc protein treatment

LINGO-1-Fc Treatment Altered Endogenous LINGO-1 Level in CGNs

In the same treatment described in Fig. 2, we showed that the protein level of LINGO-1 was promptly up-regulated upon apoptosis stimuli and antagonist application inhibited the augment during the whole observations (Fig. 4). As a proapoptotic protein and binding partner to LINGO-1, the level of p75^NTR was clearly down-regulated at eighth hour in the apoptotic courses, but LINGO-1 antagonist did not suppress the changes; similarly, the expression of Nogo-A was gradually decreased in response to LK stimuli, but unaffected by the pretreatment of any proteins (Fig. 4).

Fig. 4.

The expression levels of LINGO-1, Nogo-A, and p75 in cerebellar granule neurons undergoing apoptosis were regulated by LINGO-1-Fc in different patterns. (a) Cell lysates from cultures pretreated with LINGO-1-Fc or Fc protein and then exposed to LK treatment for indicated time period were subjected to Western blotting using antibodies against LINGO-1, Nogo-A, and p75, respectively. MAPK was selected as inner protein standard. (b) Quantitation by densitometric scans revealed the relative expression of these proteins. *P < 0.05 compared with Fc control and P value determined by unpaired t test

Discussion

Identified originally as a regulator of glycogen metabolism, glycogen synthase kinase-3 (GSK-3) is a well-established component of the Wnt signaling pathway that is essential for setting up the entire body pattern during embryonic development (Frame and Cohen 2001). There are two isoforms of GSK-3 in mammals encoded by distinct genes, GSK-3α and GSK-3β, and inhibition of GSK-3β activity is reported to be a common event in neuroprotection by different survival factors (Chin et al. 2005). Our observation that LINGO-1 antagonist promoted the phosphorylation or inactivation of GSK3β confirmed previous understanding about the role of GSK3β in the promotion of CGN apoptosis (Cross et al. 2001) as well as expanded recent results regarding the relationship between LINGO-1 and upstream regulator of GSK-3β: inhibiting LINGO-1 activities increased EGFR and p-Akt levels in the absence of myelin-associated inhibitors (Inoue et al. 2007) and the activated EGFR/Akt would inactivate downstream GSK. However, it is unclear how LINGO-1 inhibits EGFR/Akt signaling pathway function and Inoue proposes that endogenous LINGO-1 may accelerate the internalization and degradation of EGFR, thus decreasing the availability of EGFR; or that endogenous LINGO-1 directly inhibits EGFR phosphorylation and thereby reduces PI3-K/Akt signaling activity (Inoue et al. 2007).

Another candidate, known as p42/p44 MAPK, ERK1/2 is widely involved in eukaryotic mitosis and cell survival. Yamagishi (2005) reported that ERK1/2 was activated downstream of ASK1-p38 MAPK pathway during LK-induced apoptosis of cultured cerebellar granule neurons. We observed the identical alteration of ERK1/2 activity, but pretreatment with LINGO-1 antagonist did not reverse the changes, which implied that ERK pathway was irrelevant to LINGO-1 regulated neuronal survival.

As downstream effector of LINGO-1, Rho GTPases were also reported in previous studies to be involved in regulating neuronal survival in somewhat conflicting patterns. For example, Linseman (2001) observed that inhibition of Rho GTPases, specifically Rac/Cdc42, promoted apoptosis of CGNs maintained in depolarizing culture medium; blockage of the activation of RhoA protected both neurons and glial cells from apoptosis in a p75^NTR-dependent manner early on after traumatic spinal cord injury (Dubreuil et al. 2003). In the present study, we observed that the activity of Rac/Cdc42 GTPase was promptly augmented in CGN cultures stimulated with LK treatment, but LINGO-1 antagonist showed no effect on inhibiting the increase. Collectively, these data indicated that as in non-neuronal cells, Rho GTPases would play either a prosurvival or proapoptotic role in neuronal cells depending on the type of neurons studied.

Consistent with our results, recent data demonstrated that LINGO-1 was up-regulated in vivo after 6-hydroxydopamine induced neuronal damage or cell death (Inoue et al. 2007); and the up-regulation of LINGO-1 was also one characteristic of activity induced neural plasticity responses: its mRNA was strongly up-regulated in the dentate gyrus in both BDNF and KA experiments (Trifunovski et al. 2004). In particular, similar neuroprotective effects of inhibitory agents against LINGO-1 activity had been validated in other studies: LINGO-1-Fc treatment significantly increased oligodendrocyte and neuronal survival after either rubrospinal or corticospinal tract transaction (Ji et al. 2006); in LINGO-1 knockout mice, midbrain dopaminergic (DA) neuron survival was increased in animal models of Parkinson’s disease after neurotoxic lesions and similar effects were obtained in WT mice by blocking LINGO-1 activity using LINGO-1-Fc protein; besides, in vitro study implied that LINGO-1 antagonists improved DA neuron survival in response to MPP(+) involving the activation of EGFR/Akt signaling pathway (Inoue et al. 2007) and the association of GSK signaling factor with our study had been discussed above.

As binding partner of LINGO-1, many studies examining p75^NTR-mediated apoptosis and several potential concerns over the role of p75 in cell death have arisen (Hempstead et al. 2002; Harrington et al. 2002; Nykjaer et al. 2005): study on neonatal DRGs showed that p75^NTR was down-regulated in vivo after sciatic nerve transaction (Zhou et al. 2005); in contrast, an up-regulation of p75 was observed in both neurons and glia after spinal cord injury (Dubreuil et al. 2003); and treatment with antisense oligonucleotides to the p75 neurotrophin receptor decreased basal survival of cultured cerebellar neurons maintained in healthy conditions (Florez-McClure et al. 2004). In our study, the expression of p75 was detected to be decreased in advanced apoptotic stage that might act as a defensive mechanism for neurons to survive and LINGO-1 antagonist showed no impact on such changes. In addition, Reticulons have also been proposed to be involved in apoptosis (Scala et al. 2005; Li et al. 2001), then we tested whether Nogo-A expression was regulated during cerebellar granule neurons survival. Distinct from previous data that the transcript level of Nogo-A was highly sustained during CGN apoptosis (Scala et al. 2005), we observed that Nogo-A protein expression was reduced upon potassium withdrawing, which implied that degradation or other mechanism might account for the difference.

Altogether, these data suggested that signaling through LINGO-1 was a preprogrammed event tightly associated with CGN apoptosis and its neuroprotective action could be explained by a mechanism in which the activation of transcriptional factor was GSK-3 dependent. Although p75 and Nogo-A were essential for the depolarization-mediated neuronal survival, they were not the downstream targets in LINGO-1 regulated neuronal survival process. The detailed relationship elucidating LINGO-1 expression alteration and GSK-3β inactivation induced by LINGO-1 antagonist is now under our investigation.