Receptor for Advanced Glycation End Products Plays a More Important Role in Cellular Survival than in Neurite Outgrowth during Retinoic Acid-induced Differentiation of Neuroblastoma Cells*

The receptor for advanced glycation end products (RAGE), a member of the immunoglobulin superfamily, is known to interact with amphoterin. This interaction has been proposed to play a role in neurite outgrowth and process elongation during neurodifferentiation. However, there is as yet no direct evidence of the relevance of this pathway to neurodifferentiation under physiological conditions. In this study we have investigated a possible role of RAGE and amphoterin in the retinoic acid-induced differentiation of neuroblastoma cells. The functional inactivation of RAGE by dominant negative and antisense strategies showed that RAGE is not required for process outgrowth or differentiation, although overexpression of RAGE accelerates the elongation of neuritic processes. Using the antisense strategy, amphoterin was shown to be essential for process outgrowth and differentiation, suggesting that amphoterin may interact with other molecules to exert its effect in this context. Interestingly, the survival of the neuroblastoma cells treated with retinoic acid was partly dependent on the expression of RAGE, and inhibition of RAGE function partially blocked the increase in anti-apoptotic protein Bcl-2 following retinoic acid treatment. Based on these results we propose that a combination therapy using RAGE blockers and retinoic acid may prove as a useful approach for chemotherapy for the treatment of neuroblastoma.

. RA is a natural morphogen that determines anterioposterior axial patterning and induces neuronal differentiation during embryogenesis (4,5). Human and mouse neuroblastoma cells extend neurites and elongate axons following RA exposure. RA accomplishes most of its biological functions through interaction with two classes of nuclear receptors, the RA receptors (RAR) and retinoid X receptors (RXR) (6 -8). Of the naturally occurring retinoids, all-trans-retinoic acid (ATRA) binds to both RAR and RXR, whereas 9-cis-retinoic acid binds to RXR. The signal is then transduced by the formation of RXR/ RAR heterodimers, which bind to RA response elements to activate the transcription of RA-responsive genes.
Retinoic acid-induced differentiation leads ultimately to apoptosis in many cell types indicating close links between the molecular pathways of cell differentiation and those of cell death. Indeed, ATRA reduces the growth of human neuroblastoma by inducing differentiation and apoptosis (3,9,10) together with growth arrest at the G 1 phase of the cell cycle (11,12). This makes retinoic acid treatment an attractive approach for differentiation therapy of cancers that resist surgery. Clinical trials are in progress to determine the efficacy of retinoids in cancer such as neuroblastoma, which accounts for 15% of cancer deaths in children (13,14). The receptor for advanced glycation end products (RAGE) is a member of the immunoglobulin superfamily, which binds to amphoterin, a member of high mobility group protein, to initiate neurite outgrowth in cortical neurons (15,16). The RAGE-amphoterin interaction elicits Rac and cdc42 activation and stimulates neurodifferentiation in RAGE-expressing cells cultured on amphoterincoated plates (17). RAGE-amphoterin is localized at the leading edge of differentiating neurons, suggesting a role in axonal and dendritic elongation.
Although retinoic acid-induced neurodifferentiation is mediated by interaction with nuclear receptors, the molecular details of how exactly the nuclear message leads to axonal and dendritic outgrowth at the cell membrane level is still not clearly understood. An improved knowledge of the pathways that link differentiation and apoptosis during retinoic acidinduced differentiation would provide a basis for the development of better therapeutic approaches for neuroblastoma treatment. The present investigation was designed to study the role of RAGE in retinoic acid-induced neurodifferentiation. The mouse neuroblastoma Neuro2a and human SH-SY5Y cell lines chosen as the model for this study extend processes and also undergo apoptosis following retinoic acid treatment. The results presented here show that RAGE-amphoterin interaction plays a critical role in growth retardation and survival of neuroblastoma cells following retinoic acid treatment. The results also suggest that RAGE has a supplementary rather than an essential role in process outgrowth during retinoic acid-induced differentiation.

EXPERIMENTAL PROCEDURES
Plasmids and Oligonucleotides-Human RAGE cDNA was a gift from Dr. David Stern, Columbia University, New York. The cytoplasmic deletion mutant of RAGE was prepared by as previously described (14).
Phosphorothioate antisense oligonucleotides were obtained from Sigma-Genosys Ltd., UK. The oligos were 5Ј-labeled with fluorescein to follow transfection efficiency. The names, sequences, and orientation of the sense and antisense oligos used in this study are presented below in Table I.
Cell Culture and Transfections-The mouse neuroblastoma cell line Neuro2a and the human neuroblastoma cell line SH-SY5Y (Deutsche Sammlung von Mikroorganismen und Zellkulturen GmbH, Germany) were maintained in DMEM containing heat-inactivated 10% FCS, Lglutamine, penicillin, and streptomycin. Endogenous RAGE expression in both the cell lines was confirmed by Western and Northern analyses.
Construction of Stable Cell Lines-Transfection of Neuro2a cells was carried out with FuGENE 6 transfection reagent (Roche Molecular Biochemicals, Germany) and plasmid at the ratio of 3:1. Stable cell lines expressing full-length RAGE (RAGE-FL), RAGE cytoplasmic deletion mutant (RAGE-⌬), and mock cells harboring the expression vector pcDNA3 were prepared by selection in medium containing 400 g/ml G418 (Invitrogen). After 6 weeks of selection, nine independent clones were picked and analyzed for RAGE and RAGE dominant negative expression. Of these, three independent clones of RAGE-FL and RAGE-⌬ cells were used for further experiments.
Antisense Experiments-Antisense oligos (see Table I) were included in the medium for passive cellular uptake. The cells were pre-treated for 24 h with the required concentration of oligos, the presence of oligos was maintained throughout the duration of the experiment, and fresh medium containing oligos was added every second day. Uptake of oligos was monitored using a fluorescence microscope. In general, it was observed that the cellular uptake started from 12 h and the nucleotides were detectable undegraded for 60 -72 h after the initial treatment.
Differentiation Analyses-Neuro2a cells were seeded at a density of 2 ϫ 10 5 cells/well in a six-well plate and cultured for 24 h. The medium was then replaced with DMEM containing 2% serum and 20 M ATRA (Sigma). After the required time period of incubation, the percentage of differentiation was analyzed by counting the number of cells that showed neurite-like processes. Cells that showed bidirectional or mul-tidirectional outgrowths were counted. Five random fields were counted for each observation. The percentage of differentiation was calculated from the number of cells that showed process outgrowth divided by the total number of cells in each field. Cells bearing neurite-like processes longer than the diameter of cell body were considered differentiated after 48 h, and cells bearing neurites with length at least double that of cell diameter were considered differentiated after 7 days.
Neurite outgrowth was also studied by staining F-actin with FITC-Phalloidin. Neuro2a cells at a density of 5 ϫ 10 4 per well were seeded in two-well chamber slides (Nunc Lab-Tek) and cultured for 24 h. The medium was then replaced with DMEM containing 2% FCS and 20 M ATRA. After the required time period of incubation the cells were fixed in 4% paraformaldehyde. The fixed cells were subsequently permeabilized with 0.2% Triton X-100 blocked with 1% FCS and stained with FITC-labeled phalloidin (Sigma). The results were analyzed in a conventional fluorescence microscope (Zeiss, Axiovert 25).
Protein Extraction and Western Blots-Immediately after the incubation period, the cells were washed twice with PBS and scraped off the plates in lysis buffer (50 mM Tris, pH 7.4, 150 mM NaCl, Triton X-100 1.0%, SDS 0.1% ϩ protease inhibitor mixture from Roche Molecular Biochemicals) with a cell scrapper. The protein concentration of the lysates was determined by using the BCA protein assay reagent (Pierce). Equal amounts of protein were subjected to SDS-PAGE and transferred to nitrocellulose membrane in a semi-dry blotting apparatus (Sigma). The membranes were blocked with synthetic blocking agent Roti-Block (Roth, Germany) for RAGE detection and 5% (w/v) nonfat dry milk for all other proteins. The membrane was then incubated with the primary antibody followed by secondary antibody conjugated with alkaline phosphatase. The bands were subsequently detected following the addition of the sensitive coloring agents nitro blue tetrazolium and 5-bromo-4-chloro-3-indolyl phosphate-p-toluidine. Primary antibodies used: Anti-RAGE antibody was a gift from Dr. Neeper (Merck Sharp & Dohme Research Laboratories, Germany). Antiamphoterin antibody and recombinant rat amphoterin were produced and purified as has been described previously (18). Anti-Bcl-2, Bcl-X L , and proliferation cell nuclear antigen (PCNA) were purchased from PharMingen. Dr. Larry Denner (Texas Biotechnology Corp.) donated the Anti-sRAGE antibody.
Immunocytochemistry-Cells grown on 2-well chamber slides were fixed with 4% paraformaldehyde for 10 min, washed with PBS, pH 7.4, permeabilized with 0.1% Triton X-100, blocked in PBS with 1% FCS, and then incubated with primary antibody for 1 h. The slides were washed with PBS and incubated with secondary antibody, Alexa fluor 488 (Molecular Probes, The Netherlands) for 1 h. Slides were then washed with PBS and mounted with mountant containing 1% pphenylenediamine (Sigma) prior to analysis using a fluorescence microscope.
Cell Viability and Apoptosis Detection-Cell viability was studied by the MTT assay, and apoptosis was detected by analyses of DNA fragmentation by agarose gel electrophoresis.
MTT Reduction Assay-At the end of each experiment, 10 l of MTT (5 mg/ml) was added to each well and incubated at 37°C in 95% air/5% CO 2 for 4 -5 h. The insoluble formazan formed was dissolved in isopropanol/0.01 M HCl, and the absorbance was measured in a spectrophotometer at 570 nm with a background reading of 660 nm. The percentage of cell survival was calculated relative to control (taken as 100%).
Agarose Gel for DNA Fragmentation-The detached cells and the adherent cells were processed together for DNA extraction. Detached cells floating in the medium were centrifuged at 500 ϫ g for 10 min and washed with PBS once. Adherent cells were trypsinized, washed with PBS, and pelleted. The pellet was digested in proteinase K and RNase A at 50°C for overnight. After addition of DNA extraction solution, DNA was obtained by isopropanol and ethanol precipitation. DNA (10 g) was examined on 1.5% agarose gel, stained with ethidium bromide. DNA size calibration was performed using a 100-bp marker.

RAGE Is Not Required for Retinoic Acid-induced Neurodifferentiation, but RAGE Overexpression Increases the Length of Process Outgrowth
Previous reports (17,19) on the effect of RAGE on neurite outgrowth in neuroblastoma cells were made using cells that do not express endogenous RAGE. In this study, we chose cells that express endogenous RAGE to study its contribution to cellular differentiation and outgrowth under the influence of a physiological agent, retinoic acid. We constructed stable Neurite Outgrowth after 48 h of Differentiation-Neuro2a cells treated with ATRA showed sprouting within 12 h, and process outgrowth was clearly observed from 24 to 48 h in agreement with previous findings (20). FITC-phalloidin staining to visualize filamentous actin ( Fig. 2A) showed no significant difference in morphology between mock (pcDNA3), RAGE-FL (RAGE fulllength), and RAGE-⌬ (RAGE cytoplasmic deletion mutant) cells. Compared with untreated control, ATRA treatment for 48 h resulted in 10.0 Ϯ 1.1% of differentiation in mock cells (Fig. 2B). RAGE-⌬ cells, which are known to function as a dominant negative mutant for RAGE (21), also extended neuritic processes and were equally well differentiated (10.5 Ϯ 1.1%) as the mock cells ( Fig. 2, A (panel c) and B). However, the percentage of differentiation was significantly higher in RAGE-FL cells (15.2 Ϯ 1.7%) (Fig. 2B), that is, the number of cells undergoing differentiation was significantly increased when RAGE is overexpressed. As a complementary approach, inhibition of RAGE expression by antisense oligos (Table I) (Fig. 1C) in Neuro2a cells also failed to show any effect either on the morphology (data not shown) or frequency of differentiation following retinoic acid treatment. After 48 h of differentiation, Neuro2a cells treated with antisense oligos (AS-mRAGE1) showed a frequency of 11.0 Ϯ 1.3% of differentiated cells. It was noted that treatment of Neuro2a cells with antisense oligos (AS-mRAGE) resulted in 50 -60% inhibition of RAGE expression (Fig. 1C).
Neurite Outgrowth after 7 Days of Differentiation-Continuing the ATRA treatment for a week resulted in extensive out-growth and elaborate network of the processes (Fig. 3A), and the percentage of differentiation remained significantly higher in RAGE-FL (20.8 Ϯ 3.1%) compared with mock (15.6 Ϯ 1.0%) or RAGE-⌬ cells (14.4 Ϯ 1.4%) (Fig. 3B). In addition, some of the RAGE-FL cells exhibited extensive elongation of bidirectional processes (Fig. 4A). Cells bearing processes more than 10 times the diameter of cell body were present in significantly higher number among RAGE-FL (8.1 Ϯ 2.1%) compared with mock cells (2.9 Ϯ 1.4%) (Fig. 4B), demonstrating that overexpressed RAGE can contribute to the extension of retinoic acidinduced outgrowths even though its absence may not inhibit the initiation of outgrowth.

Survival of Retinoic Acid-treated Cells Is RAGE dependent; RAGE Plays a Role in Bcl-2 Production during Retinoic acid-induced Neurodifferentiation
Although RAGE-⌬ cells showed no significant difference in retinoic acid-induced differentiation or neurite outgrowth, compared with the control after 7 days of retinoic acid treatment, we did observe a reduction in the number of cells remaining on the plates ( Fig. 3A (panel c)) suggesting that RAGE-⌬ cells may detach from the plate or undergo cell death much earlier than mock or RAGE-FL cells. To address the question whether RAGE-⌬ cells are susceptible to accelerated growth arrest and/or increased cell death, we analyzed the expression of the proliferation marker PCNA as an index of growth arrest, and MTT reduction assay and DNA fragmentation pattern analyses were carried out to study cell death. Western analysis of the expression of PCNA after 24 and 48 h of ATRA treatment is shown in Fig.  5A. Mock cells showed a mildly reduced expression of PCNA after 24 h and no detectable expression at 48 h. In RAGE-⌬ cells, PCNA expression was already strongly reduced within 24 h of ATRA addition, implying that the functional inactivation of RAGE accelerates growth arrest in ATRA-treated cells.
RAGE-⌬ Cells Undergo Accelerated Cell Death following ATRA Addition-As detected by MTT reduction analyses, after 7 days of ATRA treatment, compared with untreated control cells (taken as 100%), only 11.5 Ϯ 0.7% of RAGE-⌬ cells survived against 43.0 Ϯ 1.4% of mock cells (Fig. 5B). Analyses of the DNA fragmentation pattern provided additional evidence for the increased susceptibility of RAGE-⌬ cells to retinoic acid-induced apoptosis. Agarose gel electrophoretic analysis of DNA extracted from RAGE-⌬ cells revealed laddering from day 3 onward, whereas no clear laddering was noted in DNA extracted from mock and RAGE-FL cells even after 5 days (Fig.  5C). Similarly, inhibition of RAGE expression by antisense oligos AS-mRAGE1 (Table I)  RAGE Transmits an Anti-apoptotic Signal during Retinoic Acid-induced Differentiation-The level of expression of antiapoptotic molecules of the Bcl-2 family proteins is a critical factor known to influence the susceptibility of a cell to undergo apoptosis. Changes in the expression levels of Bcl-2 family proteins have been reported (22)(23)(24) during the retinoic acidinduced differentiation of neuroblastoma cells and are likely to contribute to the altered susceptibility of these cells to apoptosis (24). We therefore studied the expression of Bcl-X L /Bcl-2 proteins to examine the molecular mechanism underlying the increased susceptibility of RAGE-⌬ cells to apoptosis. Western analyses revealed the endogenous expression of Bcl-2 and Bcl-X L in Neuro2a cells (Fig. 6, A-D), and the level of expression of Bcl-2 was increased from 24 h of ATRA treatment in mock cells (Fig. 6B) whereas that of Bcl-X L remained unchanged (Fig. 6A). A similar increase in Bcl-2 expression was noted in RAGE-FL cells (Fig. 6C). However, Bcl-2 expression did not increase in RAGE-⌬ cells (Fig. 6C). A densitometric analysis confirmed that, after 36 h of ATRA treatment, mock and RAGE-FL cells had elevated the level of Bcl-2 ϳ2.5 Ϯ 0.3and 2.8 Ϯ 0.26-fold, respectively, relative to the level expressed at 0 h, whereas no elevation was observed in RAGE-⌬ cells. These findings imply that RAGE may transmit an anti-apoptotic signal through the increased expression of Bcl-2 during retinoic acid-induced differentiation, thereby influencing the susceptibility of these cells to apoptosis.

Antisense Inhibition of Amphoterin Expression Prevents Neurite Outgrowth and Reduces Survival of Retinoic Acid-treated Neuroblastoma Cells
RAGE-mediated neurite outgrowth is known to be propagated by its interaction with the ligand amphoterin (15)(16)(17). Because the down-regulation of RAGE showed no significant effect on neurite outgrowth and differentiation, we then studied the role of amphoterin in this process. An antisense strategy was employed to inhibit the expression of amphoterin. Immunostaining revealed high concentrations of amphoterin in the nucleus (Fig. 7A), indicating its role as a chromatin-associated protein. Treatment of Neuro2a cells with AS-AMP (3.0 M) (Table I), for 48 h significantly down-regulated amphoterin expression (Ͼ70%) as indicated by the reduced intensity of fluorescence (Fig. 7A). The effect of amphoterin inhibition on retinoic acid-induced neurite outgrowth and differentiation was subsequently analyzed. The inhibition of amphoterin expression prevented retinoic acid-induced differentiation and neurite outgrowth, as detected by FITC-phalloidin staining (Fig. 7B). However, AS-AMP-treated cells showed early signs of differentiation (i.e. sprouting), demonstrating that amphoterin inhibition prevents the elongation of neurites rather than the early events. Neuro2a cells treated with sense oligos, such as S-AMP, showed 11.9 Ϯ 1.0% of differentiation, whereas cells treated with antisense oligos showed only 4.9 Ϯ 0.8% (Fig. 7C). This shows that, although RAGE is not essential for retinoic acid-induced neurite outgrowth, its putative ligand amphoterin does play an essential role. It also suggests that the functions  Because RAGE-FL cells showed extensive process outgrowth and elongation of neuritic processes, we speculated that retinoic acid might increase the expression or secretion of amphoterin during differentiation. To clarify this the expression of amphoterin following ATRA addition was monitored by Western analysis. However, we were unable to detect changes in the expression following ATRA treatment (Fig. 7D).
We then addressed the role of amphoterin in cellular survival during retinoic acid-induced differentiation. As observed by the MTT reduction assay, cells treated with sense oligos (S-AMP) showed 48.0 Ϯ 3.2% (of retinoic acid-untreated control) of survival whereas amphoterin inhibition using antisense oligos (AS-AMP) significantly reduced the percentage of survival to 35.5 Ϯ 0.7% (Fig. 8A). This result was further confirmed by DNA fragmentation analyses, wherein characteristic DNA laddering was observed in AS-AMP-treated cells following 5 days of ATRA treatment (Fig. 8B). Consistent with the results obtained from RAGE-⌬ cells, antisense inhibition of amphoterin expression also led to early growth arrest of Neuro2a cells following ATRA treatment. Expression of the proliferation marker, PCNA, was almost completely inhibited within 24 h of ATRA addition (Fig. 5A). These results parallel those obtained with the RAGE-⌬ cells, suggesting that the effect of RAGE on cellular survival during retinoic acid-induced differentiation is achieved through its interaction with amphoterin.
Both dominant negative inhibition of RAGE and antisense inhibition of amphoterin resulted in significant reduction in the survival of Neuro2a cells following ATRA treatment. However, there is a discrepancy between the absolute values obtained from MTT reduction analyses of the individual experiments. Treatment with ATRA for 7 days resulted in more than 90% inhibition of cellular survival of RAGE-⌬ cells, whereas under similar conditions, antisense inhibition of amphoterin resulted in only 65% inhibition, compared with untreated control cells. Such a difference in percentage of survival is unexpected if RAGE and amphoterin function in the same pathway through their interaction. One possible explanation could be the difference in the strategy employed to inhibit the function of the respected molecules. Antisense oligo-mediated inhibition of amphoterin expression may not reach a maximum level whereas the dominant negative strategy used in RAGE-⌬ cells could exert a maximum inhibition of RAGE function, as reported earlier (21).

Involvement of RAGE in Retinoic Acid-induced Differentiation and Survival in Neuro2a Cells Is Reproduced in SH-SY5Y Cells
Having studied the role of RAGE in the murine Neuro2a cells, we wanted to establish whether or not the function of RAGE in cellular survival during retinoic acid-induced differentiation is also observed in the human neuroblastoma cell line. To this end we chose the human neuroblastoma cell line SH-SY5Y, which is well established as a model for in vitro neurodifferentiation studies (25,26). An antisense strategy was employed to block the expression of RAGE. As shown in Fig. 9A inclusion of phosphorothioate oligonucleotides, AS-hRAGE (3 and 5 M), in the medium for 48 h resulted in 50 -70% inhibition of the expression of RAGE in SH-SY5Y cells (Fig. 9A). ATRA treatment in the presence of sense oligos, S-hRAGE, for 7 days showed extensive differentiation of SH-SY5Y cells, as judged by shrinkage of cell body and extension of neuritic processes (Fig. 9B). Consistent with the results obtained with Neuro2a cells, inhibition of RAGE expression by AS-hRAGE oligos showed no effect on differentiation, as judged from the morphology and neurite extension of differentiated cells (Fig. 9B). The notion that RAGE plays a role in cellular survival during retinoic acid-induced differentiation received additional support from the cellular survival experiments conducted on SH-SY5Y cells. Treatment of SH-SY5Y cells with ATRA for 7 days reduced the number of surviving control cells to 46.0 Ϯ 2.8% of an untreated control. Under similar conditions inhibition of RAGE by antisense oligos reduced survival to 16.7 Ϯ 0.7% (Fig. 9D). The observation was further supported by the analysis of expression of Bcl-2: Inhibition of RAGE reduced the extent of Bcl-2 induction in ATRA-treated SH-SY5Y cells (Fig. 9C).

RAGE Blocking Antibodies Reduce Growth of Neuroblastoma Cells Treated with Retinoic Acid
Having established the role of RAGE in cellular survival during retinoic acid-induced differentiation of neuroblastoma cells, we then asked if the RAGE blocking antibodies, raised against the soluble RAGE (sRAGE) could in principle serve as an agent, together with ATRA, for combination therapy to treat neuroblastoma tumors. Taguchi et al. (21) have recently demonstrated, using a mouse model, that RAGE blocking antibodies prevent tumor growth and metastasis by inhibiting RAGEamphoterin signaling. Therefore, we hypothesize that a combination therapy using retinoic acid and RAGE blockers may exert the double effect of early cell death and prevention of metastasis.
Inclusion of anti-sRAGE antibodies at a concentration of 50 g/ml together with ATRA significantly reduced the cellular survival of Neuro2a as determined by MTT reduction analysis. Treatment of Neuro2a cells with ATRA alone for 4 days reduced the percentage of survival to 52.0 Ϯ 5.6% of untreated control cells (taken to be 100%). The presence of anti-sRAGE antibodies under similar conditions significantly reduced the percentage of survival to 33.0 Ϯ 4.2% (Fig. 10A). A similar effect was observed in the RAGE-overexpressing RAGE-FL cells: The percentage of survival was reduced to 37.0 Ϯ 4.2% in these cells. Similar results were obtained with the human neuroblastoma cells SH-SY5Y. However, these cells required a higher concentration of antibody to demonstrate the effect.  (Table I)  Treatment with ATRA for 8 days reduced the percentage of survival to 55.0 Ϯ 2.1% of untreated control cells whereas the inclusion of 100 g/ml anti-sRAGE antibodies reduced the percentage of survival to 43.0 Ϯ 2.1% (Fig. 10B). The difference in the required concentration of antibody may be due to the difference in the time period required for the cell lines to undergo differentiation, because Neuro2a cells undergo differentiation within 48 h of ATRA treatment, whereas the SH-SY5Y cells took more than a week. DISCUSSION Some recent reports (15,17) have suggested that the interaction between RAGE and amphoterin plays a prominent role in neuritic process extension and outgrowth in neuroblastoma cells and/or neurons . It should be noted that these observations were made by culturing RAGE-expressing cells on amphoterin-coated plates and hence establishing a direct contact between RAGE and amphoterin in vitro. The role of RAGEamphoterin interaction in neurite outgrowth or axonal elongation induced by physiological differentiation inducers, such as retinoic acid, nerve growth factor, and others, remains unknown. Such experiments on the role of RAGE-amphoterin interaction in neurite outgrowth under the influence of physiological agents are necessary to assess the contribution of RAGE in neural development. A detailed understanding of the role of RAGE-amphoterin interaction in neurite outgrowth under the influence of physiological regulators would provide profound knowledge on the molecular biology of neuronal differentiation. Therefore, the purpose of this investigation was to study the role of RAGE and amphoterin in retinoic acid-induced neurodifferentiation, a process widely believed to play an essential role during early neural development (3-5, 27, 28). To achieve this, we have used dominant negative and antisense strategies to functionally inactivate RAGE and an antisense strategy to suppress amphoterin expression in Neuro2a cells. The effect of RAGE was further confirmed by antisense strategy in human neuroblastoma SH-SY5Y cells.
It is widely accepted that the nuclear receptors for retinoic acid, RAR and RXR, mediate most of the initiating events in retinoic acid-induced neurodifferentiation (8,29). Yet, the various molecular mechanisms that translate the nuclear message to axonal and dendritic outgrowth are poorly understood. Data presented here show that amphoterin, which has been shown to localize to the leading edge of a migrating cell (30), is essential for retinoic acid-induced process outgrowth. The blockage of amphoterin expression by antisense oligos significantly prevented process outgrowth. However, functional inactivation of RAGE, the putative receptor for amphoterin, either by dominant negative or antisense strategies, failed to prevent retinoic acid-induced neurite outgrowth, raising the possibility that amphoterin may interact with one or more other molecules to exert its effect on outgrowth. Indeed, previous reports have shown that amphoterin does in fact interact with other mole-  cules, including syndecan (31) and sulfoglycolipids (32), although the relevance of these interactions in neurite outgrowth remains uncertain. The possibility of amphoterin interacting with an as yet unidentified molecule may not be ruled out. The observation made in this study, that blockage of amphoterin expression prevented only neurite elongation without any effect on early sprouting, indicates that the early signals of retinoic acid-induced differentiation are controlled by mechanisms not linked to amphoterin. Middlemas et al. (33) have shown that ATRA induces differentiation of neuroblastoma cells by establishing an autocrine loop involving brain-derived neurotrophic factor and trkB. Early sprouting of neurites may be controlled by trkB activation followed by G protein activation, and the subsequent messages may be transmitted to a pathway that requires amphoterin for neurite outgrowth. On the other hand, because the overexpression of RAGE dramatically increased the length of neurites after retinoic acid treatment, it is possible that RAGE participates in the process outgrowth but the role of RAGE might be in the elongation rather than the initiation phase of neurite growth.
The observation that the overexpression of RAGE stimulates extensive outgrowth of neurites also suggests that RAGE may provide an additive effect on axonal elongation, provided its expression is increased. RAGE is known to be able to increase its own expression through an NFB-dependent positive feedback mechanism (34). Data on in vivo expression of RAGE during early neural development supports this hypothesis. It has been shown that neurons undergoing differentiation and functional maturity during early development express significantly higher levels of RAGE and amphoterin than do adult neurons (15,35). Therefore an increased expression of these molecules during the early development could possibly contribute to retinoic acid-induced neurodifferentiation in vivo.
The most interesting observation made in this study is the role of RAGE-amphoterin interaction in the survival of retinoic acid-treated neuroblastoma cells, which may have implications for chemotherapy, because differentiation therapy by retinoic acid treatment is one of the widely tested current modes of cancer treatment (13). The data presented here show that RAGE-amphoterin interaction plays a positive regulatory role in the survival of retinoic acid-treated neuroblastoma cells. Inhibition of RAGE resulted in early growth arrest and cell death in both Neuro2a and SH-SY5Y cells. Because the expression levels of anti-apoptotic molecules have been shown to increase during differentiation of neuroblastoma cells (22)(23)(24), we hypothesized that RAGE-amphoterin interaction may transmit signals necessary for the increased expression of antiapoptotic molecules. Recently, Huttunen et al. (19) have shown that RAGE-amphoterin interaction in N18 neuroblastoma cells favors cell survival through the expression of Bcl-2. Consistent with this we observed that RAGE blockage in cells resulted in failure to substantially increase expression of Bcl-2 in retinoic acid-treated neuroblastoma cells. These observations suggest that RAGE-amphoterin interaction contributes to the survival of cells undergoing retinoic acid-induced neurodifferentiation.
If RAGE and/or amphoterin contribute to Bcl-2 expression during retinoic acid-induced differentiation, what could be the possible mechanism? Signal transduction pathways involving protein kinase C (PKC) activation have been implicated in Bcl-2 production during differentiation (24,36). Modulators of PKC activation have been shown to alter the expression levels of Bcl-2 family proteins and thus modulate susceptibility to apoptosis (24). In SH-SY5Y cells, the enhanced expression of Bcl-2 after retinoic acid treatment was blocked by PKC inhibitors, staurosporine or calphostin C (36). In addition, PKC activation is also reported to result in phosphorylation of Bcl-2 in myeloid leukemia cells thus rendering the cells resistant to apoptosis (37). The available literature suggests a prominent role for PKC in the expression and phosphorylation of Bcl-2 protein. Against this background, it is interesting to note that Melloni et al. (38) have shown that an amphoterin-like protein activates PKC-␣ in murine erythroleukemia cells. A recent report (39) showed that PKC inhibitors block RAGE-mediated transendothelial migration of monocytes upon ligation with ␤-amyloid, another ligand for RAGE, suggesting that RAGE activation may activate PKC. These reports suggest a possible involvement of RAGE or amphoterin activation in PKC activation, however, additional experimental evidence will be required to clarify whether or not the PKC pathway is involved in the regulation of Bcl-2 family proteins by RAGE-amphoterin interaction.
One interesting question that arises from the present study is whether the signal that passes through RAGE during retinoic acid-induced differentiation is involved in cellular growth arrest or cell death. The observation that inactivation of RAGE results in reduced PCNA expression and appearance of DNA laddering clearly suggests that RAGE is involved in both pathways. It is widely known that the molecular mechanisms involved in growth arrest and apoptosis are closely linked. However, the modulatory roles of retinoic acid in these pathways remain to be elucidated. Further studies in this area would provide insight into signaling pathways that are propagated through RAGE during retinoic acid-induced growth arrest and apoptosis. The increased expression of Bcl-2 family proteins during differentiation reduces the susceptibility of neuroblastoma cells to apoptosis (24,40). On the other hand, the antisense inhibition of Bcl-2 expression induces retinoic acid-induced cell death in human neural precursor cells (41). Therefore, improved knowledge of the mechanisms that inhibit increased expression of anti-apoptotic proteins following retinoic acid treatment would provide better therapeutic strategies for neuroblastoma, which accounts for 15% of cancer-related deaths in children (14). Although the presently available chemotherapy for neuroblastoma is partly successful, the majority of the patients subsequently relapse. Alternative therapeutic approaches using retinoic acid and other agents like histone deacetylase inhibitors as combination therapy are being studied (42). Based on our present results we hypothesized that a combination of retinoic acid and RAGE blockers may substantially increase cell death in neuroblastoma cells. As a preliminary approach, we studied the effect of a combination of RAGE blocking antibody and retinoic acid treatment on Neuro2a and SH-SY5Y cells and observed that RAGE blockage significantly prevents the growth and survival of these cells. If these results are further proved in other neuroblastoma or tumor cells, a combination therapy of retinoic acid and RAGE blockers may be a useful approach for the treatment of neuroblastoma.
In conclusion, the results presented in this work suggest that RAGE is not required for retinoic acid-induced neurite outgrowth but plays a significant role in the elongation of neurites and cellular survival during the process of differentiation. However, amphoterin, a known ligand of RAGE, is essential for both neurite outgrowth and cellular survival, raising the possibility that amphoterin might interact with other molecules to exert its effect on neurite outgrowth. Thus there is a disassociation between the effects of RAGE and amphoterin upon retinoic acid-induced neurite outgrowth but not in cellular sur-vival in Neuro2a cells. Based on these results it is proposed that blocking the function of RAGE together with retinoic acid treatment may be used as a therapeutic approach for early growth arrest and cell death in neuroblastoma.