Voltage-Dependent N-Type Ca2+ Channel Activity Regulates the Interaction Between FGF-1 and S100A13 for Stress-Induced Non-Vesicular Release

Abstract

1. The Ca²⁺-mediated regulation of interaction between FGF-1 and S100A13 in NG108-15 cells was studied. When the stress by depriving B27 supplement from the culture was given, cellular levels of both proteins were decreased, while their releases were significantly increased within 3 h. These stress-induced changes were all abolished by amlexanox, an anti-allergic drug.

2. These releases were significantly inhibited by the addition of EGTA or BAPTA-AM, cellular or extracellular Ca²⁺-chelating agent, respectively. The addition of ω-conotoxin GVIA, a N-type Ca²⁺-channel blocker caused a complete inhibition of the release, while increased the cytosolic levels of both proteins. However, ω-conotoxin MVIIC, the non-N-type Ca²⁺-channel blocker was ineffective.

3. In NG108-15 cells, which had been transfected with Venus-FGF-1 and CFP-S100A13, the supplement-deprivation stress caused several spike-type fluorescence resonance energy transfer (FRET) signals, suggesting that both proteins showing interaction would be immediately released. These spikes were completely abolished by the addition of ω-conotoxin GVIA. However, the addition of amlexanox caused bell-shaped FRET signals without spikes.

4. Thus, it is suggested that the interaction between FGF-1 and S100A13 responsible for stress-induced non-vesicular release is dependent of Ca²⁺-influx through N-type Ca²⁺-channels.

INTRODUCTION

FGF-1 in the central nervous system plays important roles in neuroprotection and angiogenesis during development and repair process from ischemic damages (Reuss and von Bohlen und Halbach, 2003). This polypeptide is probably released upon cell damage and the underlying mechanism is supposed to be performed by endoplasmic reticulum (ER)-Golgi-independent non-vesicular or non-classical release (Cleves, 1997; Prudovsky et al., 2003; Nickel, 2005), since it lacks conventional signal sequence in its amino terminal. Molecular machineries for the non-vesicular release of FGF-1 upon heat shock stress are in part characterized to be associated with Ca²⁺-binding protein S100A13 in peripheral cells, in experiments where it had been over-expressed with exogenous protein components (Mouta Carreira et al., 1998; Landriscina et al., 2001a; Prudovsky et al., 2002). In such studies, FGF-1 is thought to be released as a multiprotein aggregate containing FGF-1, S100A13 and p40 synaptotagmin-1, and this multiprotein complex is formed in the presence of high concentration of Cu²⁺ (1 –100 mM) in a cell-free system, which is induced by the formation of the homodimer of FGF-1 due to the oxidation of Cys³⁰ by Cu²⁺. In addition, Cu²⁺-chelating agent prevents the heat shock-induced release of FGF-1 in NIH-3T3 cells. Landriscina et al. (2001a) suggested that conformational change of FGF-1 would be necessary for the interaction between FGF-1 and p40 synaptotagmin, but not for the interaction between FGF-1 and S100A13. S100A13 protein has been identified as a target of anti-allergic drug, amlexanox, which inhibits degranulation of mast cells (Shishibori et al., 1999) and co-release of FGF-1 and S100A13 upon heat-shock stress in NIH-3T3 cells (Mouta Carreira et al., 1998; Prudovsky et al., 2002), suggesting that S1000A13 is a cargo molecule for the FGF-1 release. However, the trigger for the interaction between FGF-1 and S100A13 is unclear. Here we report the role of Ca²⁺ on the non-vesicular release and interaction between FGF-1 and S100A13.

MATERIALS AND METHODS

Materials

BAPTA-AM was purchased from Nacalai Tesque (Kyoto, Japan). Ammonium tetrathiomolybdate (TTM), ω-conotoxin GVIA and MVIIC were purchased from SIGMA (St. Louis, MO). Amlexanox was kindly provided by Takeda Pharmaceutical Company Ltd. (Osaka, Japan).

Antibodies

Rabbit anti-FGF-1 was purchased from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA). Anti-FGF-1 was biotinylated by micro biotinylation kit (SIGMA). Rabbit anti-S100A13 was kindly provided by Dr. T. Maciag (Center for Molecular Medicine, Maine Medical Center Institute, Scarborough).

Expression Constructs

Genes for CFP and Venus, a variant of YFP (Nagai et al., 2002) were kindly provided by Dr. A. Miyawaki (Brain Science Institute, RIKEN, Saitama, Japan). Each gene was amplified by PCR. The primers used were: Fluorescent proteins-F (Venus-F and CFP-F, both fluorescent proteins have the same sequence at 5′-terminal): 5′-AAGGATCCACCATGGTGAGCAAGGGCG-3′ (containing a BamHI site); Venus-R: 5′-AAGAATTCCTTGTACAGCTCGTCCATGC-3′ (containing a EcoRI site); CFP-R: 5′-AAGAATTCGGCGGCGGTCACGAACTCCAG-3′ (containing a EcoRI site). Amplified genes were cloned in-flame into the BamHI- Eco RI sites of pcDNA3.1 (+) (Invitrogen, Tokyo, Japan). The rat FGF-1 and S100A13 genes were amplified from the cDNA derived from rat embryonic brain. PCR primers used were FGF-1-F: 5′-AGAATTCATGGCCGAAGGGGAGATCAC-3′ (containing a Eco RI site); FGF-1-R: 5′-ACTCGAGTTAGTCAGAAGATACCGGG-3′ (containing a XhoI site); S100A13-F: 5′-AAGAATTCATGGCAGCAGAGCCCCCGAC-3′ (containing a EcoRI site); S100A13-R: 5′-AAGAATTCTTACTTCTTGCGAATCGCCAGG-3′ (containing a EcoRI site). Amplified FGF-1 and S100A13 genes were cloned in in-flame into the EcoRI-XhoI site of Venus gene cloned pcDNA3.1 (+) and the EcoRI site of CFP gene cloned pcDNA 3.1 (+), respectively. After cloning, each construct was verified by sequencing. DH10B E. coli strain was transformed with all these constructs.

Cell Culture

NG108-15 cells were routinely grown in regular tissue culture flasks with Dulbecco's Modified Eagle Medium (DMEM) supplemented with 10% fetal bovine serum in a CO₂ incubator at 37°C. For experiments, the medium was replaced with serum-free DMEM supplemented with B27 (Invitrogen) and cultured for 3 days prior to use. Cells were then replated at a density of 6–8×10⁴ cells cm⁻² onto Lab-Tek II glass bottom chamber (Nalge Nunc International Corp., Naperville, IL), which had been thin-coated with growth factor reduced Matrigel (BD Falcon, San Jose, CA).

Immunocytochemistry

Cells were fixed in 4% paraformaldehyde in PBS for 30 min and permeabilized with methanol for 10 min at RT. Fixed cells were washed with PBS and incubated in blocking buffer (1% BSA in PBS) for 2 h at 4°C, and then incubated overnight at 4°C in the presence of 1:300 first antibodies in blocking buffer. Secondary antibody conjugated to FITC (Santa Cruz Biotechnology, Inc.) or rhodamine (Chemicon International Inc., Temecula, CA), and streptavidin-Alexa 488 (Molecular Probes, Eugene, OR) or streptavidin-rhodamine (Chemicon International Inc.) were used at 1:500 in blocking buffer. Finally, the nuclei were stained with Hoechst 33342 (Molecular Probes). Images were collected using a BZ-8000 microscope (KEYENCE, Osaka, Japan) and a 20× Plan APO lens (Nikon, Tokyo, Japan).

Immunoprecipitation

For the determination of protein release, the conditioned medium (CM) from cultured NG108-15 cells (2×10⁶ cells) was used after the removal of insoluble debris by centrifugation at 10,000×g. Anti-FGF-1, anti-S100A13 and Protein G-Sepharose™ beads (Amersham Biosciences, Tokyo, Japan) were added to CM and incubated on a rotor for 2 h at 4°C. The beads were washed three times with serum-free medium and quenched with 50 μl of SDS sample buffer.

Fluorescence Resonance Energy Transfer (FRET)

NG108-15 cells were grown on 24-well plate for 1 day, and Venus-FGF-1 and CFP-S100A13 were transfected by a calcium phosphate/DNA coprecipitation procedure. Twenty-four h after the transfection, NG108-15 cells were grown on Lab-Tek glass bottom chamber (Nalge Nunc International Corp.) in serum-free medium supplemented with B27 for 3 days prior to use. Cells were imaged in Hanks buffer. Images were collected with a TE 300 inverted microscope with a 40 × S Fluor lens (Nikon). Cells were excited at 433 nm and 535/480 nm ratiometric images were acquired every 30 sec for 2 h. Dual-emission ratiometric imaging was analyzed by AQUACOSMOS software (Hamamatsu Photonics, Shizuoka, Japan). For the acceptor photo-bleach, FRET analysis was performed after quenching the Venus-derived fluorescence by excitation at 515 nm.

RESULTS

As shown in Fig. 1A, FGF-1 is distributed in the nuclei and cytosol, while S100A13 is throughout the NG108-15 cell. When the starvation stress by depriving B27 supplement from the culture was given, the FGF-1 level in the cytosol was almost disappeared, and the level in the nuclei was largely decreased at 3 h. On the other hand, S100A13 levels were slightly decreased by this stress. These decreases were abolished by the addition of 100 μM amlexanox, which had been added 30 min before the stress. It should be noted that FGF-1 was observed in larger size than the nuclei.

Fig. 1.

Stress-induced release of FGF-1 and S100A13. (A) Supplement-deprivation stress-induced FGF-1 release. Results represent the immunocytochemistry of FGF-1 and S100A13 in NG1008-15 cells. Cultured NG108-15 cells were subjected to supplement-deprivation stress (3 h) in the presence or absence of amlexanox (100 μM, pretreatment for 30 min). Scale bar represents 20 μm. (B) Immunoblot analysis of FGF-1 and S100A13 levels in CM. The CM from supplement-deprived NG108-15 cells in each time point was immunoprecipitated with anti-FGF-1 IgG or S100A13 IgG. Abbreviations: IB; Immunoblot, i.p.p.t; Immunoprecipitation. Data are represented as the mean ± S.E.M.^* p< 0.01, vs. the corresponding 0  h.#p< 0.01, vs. the corresponding 3 h. (Student's t-test).

Quite consistent results were observed when the release of FGF-1 and S100A13 to the conditioned medium (CM) was measured (Fig. 1B). Although both proteins were found in CM without starvation stress, these levels were time-dependently increased up to 3 h after the stress. At 3 h, FGF-1 release came up to the level of 43% of total cell contents, while S100A13 release to the level of 22%. These releases were also abolished by amlexanox.

The starvation stress-induced decrease in cellular FGF-1 levels was markedly reversed by the addition of 100 nM ω-conotoxin GVIA, a voltage-dependent N-type Ca²⁺-channel inhibitor (Hillyard et al., 1992), but not of 1 μM ω-conotoxin MVIIC, a voltage-dependent non-N-type Ca²⁺-channel inhibitor (Ellinor et al., 1993), as shown in Fig. 2A. It should be noted that the addition of ω-conotoxin GVIA rather increased the cytosolic levels of FGF-1, compared with the case without treatment (Fig. 1A). The addition of intracellular Ca²⁺-chelating reagent BAPTA-AM at 10 μM, a maximum concentration (Strayer et al., 1999), also reversed it, but the level was less than the case with ω-conotoxin GVIA. On the other hand, TTM at 250 μM, a maximum concentration to chelate Cu²⁺ (Lang and Tatsumi, 1998; Landriscina et al., 2001b) showed only slight recovery. Similar, but less marked changes by these reagents were also observed with the S100A13 release.

Fig. 2.

N-type Ca²⁺ channel involvement in the stress-induced FGF-1 and S100A13 release. (A) Immunocytochemistry of FGF-1 and S100A13 in NG1008-15 cells. (B) Immunoblot analysis of FGF-1 and S100A13 levels in CM. Test drugs are ω-conotoxin GVIA (100 nM), MVIIC (1 μM), BAPTA-AM (10 μM, pretreatment for 30 min), TTM (250 μM, pretreatment for 30 min) and EGTA (10 mM). Abbreviations: IB; Immunoblot, i.p.p.t; Immunoprecipitation. Data are presented as the mean ± S.E.M. ^* p< 0.05,^** p< 0.01, vs. the corresponding Vehicle treatment (Student's t-test).

When the stress-induced release of FGF-1 and S100A13 was examined by the immunoblot after the immunoprecipitation, significant decrease was observed with EGTA (10 mM), BAPTA-AM, TTM and ω-conotoxin GVIA, but not with ω-conotoxin MVIIC. The treatment with ω-conotoxin GVIA completely abolished both releases, being consistent with the cytochemical observation in Fig. 2A.

To evaluate the FGF-1-S100A13 interaction in living cells, we used NG108-15 cells transiently expressing Venus-FGF-1 and CFP-S100A13. In such cells, CFP (donor) emission at ∼475 nm is quenched, while Venus (acceptor) emission at ∼530 nm increases when Venus-FGF-1 and CFP-S100A13 come within sufficient proximity. Thus, an increase in the emission ratio of 530/475 nm (Venus/CFP) reflects a close association of the two proteins (less than 100 angstroms). In this FRET analysis, the cells showed some spike-like FRET signals (Fig. 3Aa). When B27 supplement was deprived from the culture, more and higher spikes were observed (Fig. 3Ab). When the number of spikes which are higher than 5 emission ratio was calculated, FRET signals tend to increase by supplement deprivation stress (Fig. 3B). When the number of FRET signals (Ratio > 5) was plotted against the FRET ratio, the ratio of most cells showed below 10 in untreated cells. However, the supplement-deprived cells showed higher ranges of ratios. It is evident that these FRET signals were true, since no FRET signal was observed by the photo-bleaching of Venus (Fig. 3B).

Fig. 3.

Ca²⁺-involvement in the stress-induced interaction between FGF-1 and S100A13 in living cells. (A) Representative FRET-imaging of the Venus-FGF-1 - CFP-S100A13 interaction. Venus-FGF-1 and CFP-S100A13 were transiently expressed in NG108-15 cells. The FRET ratio represents the emission ratio of 535/480-nm. (Aa) FRET signals in the cell without supplement deprivation stress. (Ab) Supplement-deprivation stress-induced increase in FRET signals. (Ac and Ad) Decrease in stress-induced FRET signals by ω-conotoxin GVIA and TTM, respectively. (B) Comparison of frequency of FRET spikes (Ratio > 5, n=7). Data are presented as the mean ± S.E.M. ^* p< 0.05,^** p< 0.01, vs. Vehicle (One-factor ANOVA, Tukey's test). (C) Ratio-distribution. Results represent the number of FRET signals (Ratio>5, n=7) against the FRET ratio. (Ca and Cb) FRET signals in the presence and absence of supplement, respectively. (Cc and Cd) FRET signals after supplement-deprivation stress in the presence of ω-conotoxin GVIA and TTM, respectively. (D) FRET signals in the presence of amlexanox (100 μM, pretreatment for 30 min).

In accord with the results with FGF-1 release, stress-induced FRET signals were completely inhibited by ω-conotoxin GVIA (Fig. 3Ac, B, Cc), and less evident inhibition was observed with TTM (Fig. 3Ad, B, Cd).

Quite different changes were observed when amlexanox was added to the culture without supplement. The FRET signals were bell-shaped, and spikes became less evident (Fig. 3D).

DISCUSSION

The first important pieces of evidence obtained in the present study are that the endogenous FGF-1 mostly found in the nuclei of NG108-15 cells is extracellularly released upon supplement-deprivation-induced (or starvation) stress (Fig. 1). As FGF-1 has no apparent signal peptide sequence at its N-terminal, the manner of release is evident to differ from the classical vesicular release. Although the heat-shock stress-induced non-vesicular release of FGF-1 from the peripheral cells overexpressing recombinant FGF-1 has been previously reported (Mouta Carreira et al., 1998; Landriscina et al., 2001a,b), the present finding would be the first demonstration in terms of starvation stress-induced endogenous FGF-1 release from neuronal cells. In this experiment, we used NG108-15 cells cultured in the presence B27 supplement, since such treated cells are differentiated to have more neuronal properties (Kowtha et al., 1993).

The starvation-induced release of FGF-1 and S100A13 was very similar to each other in the time-course and the sensitivity to various inhibitors including amlexanox, EGTA, BAPTA-AM, TTM and ω-conotoxin GVIA, but not ω-conotoxin MVIIC. These findings strongly support the previous report that S100A13 is a cargo protein to mediate the non-vesicular release of FGF-1 (Landriscina et al., 2001a). However, the ion-selectivity was a little different from that report, in which the authors demonstrated that the same concentration (250 μM) of TTM completely inhibits the heat-shock stress-induced co-release of both proteins from NIH3T3 cells over-expressing recombinant proteins. In the present study, on the other hand, TTM-induced inhibition of the release was only partial and less than any inhibitors, which decrease cellular Ca²⁺ levels. This difference may be attributed to the cell species and stress selectivity and to the fact that over-expressed FGF-1 was mostly found in the cytosol. In the presence of ω-conotoxin GVIA, the cytosol FGF-1 level was markedly increased. This finding strongly suggests that FGF-1 is re-distributed from the nuclei to the cytosol upon the starvation-stress under the condition preventing Ca²⁺-influx-mediated non-vesicular release. This is consistent with the finding that FGF-1 was found in larger size than the nuclei in the presence of amlexanox, which inhibits non-vesicular release. However, the detailed mechanisms underlying stress-induced re-distribution remain to be determined.

It has been proposed that the non-vesicular release of FGF-1 is mediated through an interaction with S100A13 upon the stress. Prudovsky et al. (2003) speculated that Cu²⁺ is required for the interaction between FGF-1 and S100A13, but it remains unclear whether Ca²⁺ is involved in the interaction. The present study firstly demonstrated that the stress-induced interaction between both proteins is mediated through Ca²⁺-influx through voltage-dependent N-type Ca²⁺-channels in the FRET analysis. Although untreated cells show some spike-like FRET signals, as shown in Fig. 3, the starvation stress markedly increased the frequency and ratio. The addition of ω-conotoxin GVIA completely inhibited the FRET signals, being consistent with the fact that FGF-1 and S100A13 are released to some extent without stress (Fig. 1B). On the other hand, TTM inhibited them to the basal level without stress in terms of the frequency and ratio distribution. It is of interest that such FRET signals were spike-like. This suggests that the Venus-FGF-1-CFP-S100A13 complex showing FRET is immediately released. Indeed, the spike-like FRET signals were mostly disappeared, but instead a bell-shaped signal pattern was observed in the presence of amlexanox, which inhibits the release. The declined phase of FRET may be related to the Ca²⁺ homeostasis, in which Ca²⁺ gated through channels will be soon reverted to the resting level.

In conclusion, we demonstrated that the starvation stress induced non-vesicular release of FGF-1 and S100A13 through an activation of voltage-dependent N-type Ca²⁺ channels in neuronal cells.

Abbreviations

FGF

fibroblast growth factor

ER

endoplasmic reticulum

CM

conditioned medium

EGTA

ethylene glycol-bis (β-aminoethylether)-N,N,N′,N′-tetraacetic acid

TTM

ammonium tetrathiomolybdate

PCR

polymerase chain reaction

DMEM

dulbecco's modified eagle medium

BSA

bovine serum albumin

PBS

phosphate buffered saline

YFP

yellow fluorescent protein

CFP

cyan fluorescent protein

FRET

fluorescence resonance energy transfer