Multifunctional Chimeric Proteins for the Sequential Regulation of

Jan 11, 2008 - sequential regulation of growth factor signals acting on neural ... epidermal growth factor (EGF), ciliary neurotrophic factor (CNTF), ...
0 downloads 0 Views 1MB Size
Download PDF
516

Bioconjugate Chem. 2008, 19, 516–524

Multifunctional Chimeric Proteins for the Sequential Regulation of Neural Stem Cell Differentiation Tadashi Nakaji-Hirabayashi, Koichi Kato, Yusuke Arima, and Hiroo Iwata* Institute for Frontier Medical Sciences, Kyoto University, 53 Kawahara-cho, Shogoin, Sakyo-ku, Kyoto 606-8507, Japan. Received September 13, 2007; Revised Manuscript Received October 30, 2007

Controlling the dynamics of growth factor signaling is a challenge in regenerative medicine for various tissues including the central nervous system. Here, we report on the development of the biomolecular system that facilitates sequential regulation of growth factor signals acting on neural stem/progenitor cells. Recombinant technology was employed to synthesize the multifunctional chimeric protein that contained multiple domains, including epidermal growth factor (EGF), ciliary neurotrophic factor (CNTF), globular capping domain, thrombin-cleavable sequence, and substrate-binding domain with affinity for Ni(II) ions. The chimeric protein is expected to expose CNTF upon elimination of the capping domain by digestion with endogenous thrombin in vivo. When the multifunctional chimeric protein was immobilized onto a substrate through the coordination of the substratebinding domain with surface-immobilized Ni(II) ions, the substrate served to proliferate neural stem cells, maintaining the population of undifferentiated cells at 85%. This effect is primarily due to the activity of EGF, while CNTF activity is temporally veiled with the capping domain. Upon digesting the thrombin-cleavable sequence to remove the capping domain, the activity of CNTF emerged to induce differentiation of astrocytes in situ from the proliferated neural stem cells. The fraction of differentiated astrocytes reached 68% of total cells. These results demonstrate the feasibility of the system for controlling the dynamics of growth factor signals.

INTRODUCTION Transplantation of neural progenitor cells is considered a potential modality for the structural and functional restoration of the central nervous system (CNS) (1, 2). An alternative strategy is to recruit and stimulate endogeneous neural stem cells (NSCs) by the local administration of appropriate agents (3). In these approaches, the tight regulation of transplanted or recruited cell behavior is a challenging task because cell proliferation, migration, differentiation, and network formation must be regulated temporally and spatially for orchestrating the chronological processes toward tissue reconstruction (4, 5). In the mammalian CNS, a variety of neurotrophins and mitogens operate in a time-dependent manner to modulate the dynamics of cellular differentiation and tissue organization (6). Toward the regeneration of neural tissues, the transplantation of cells with biomaterials that provide microenvironments mimicking natural extracellular stimuli may improve the outcome after transplantation. To date, biodegradable scaffolds that incorporate adhesive peptides (7, 8), growth factors (9), and Notch ligands (10) were reported for this purpose. However, the biomaterials that facilitate the temporal regulation of multiple signal presentation remain to be developed for more precisely modulating the hierarchically complex process of tissue reorganization. In this study, we aimed at establishing the chimeric protein system that permits the switching of growth factor presentation for the tightly regulated proliferation and differentiation of NSCs. Our approach is to utilize the multifunctional chimeric protein (MCP) that contains two different growth factors fused to each other. To sequentially regulate the signaling of two growth factors, the terminal of the second growth factor is capped by a globular polypeptide with a protease-cleavable linker peptide in between so that the capping polypeptide * To whom correspondence should be addressed. Tel: +81-75-7514119. Fax: +81-75-751-4119. E-mail: [email protected].

temporally hinders the active site of the second growth factor until the linker is cleaved by protease. The MCP further contains a peptide sequence that binds specifically to substrates for the oriented presentation of the protein. Such a MCP was rationally designed and synthesized using recombinant technology. Our future goal is to extend this system for use in tissue engineering of the central nervous system. The MCP will be incorporated into biodegradable polymer networks carrying its binding site and used as a scaffold for neural stem/progenitor cells. Because the sequential administration of multiple growth factors requires repeated intervention to the brain, a single component system such as MCP will have an advantage in such an application. As a key component, a thrombin-cleavable peptide is incorporated to the MCP. To optimize the dynamics of growth factor actions, the peptide sequence can be readily replaced with peptides cleavable with other proteases involved in tissue reconstruction processes such as matrix metalloproteinases. To examine the feasibility of our approach, we synthesized the chimeric protein that contains epidermal growth factor (EGF) and ciliary neurotrophic factor (CNTF). It is reported that EGF exhibits mitogenic activity for NSCs (11), while CNTF promotes the glial specification of NSCs (11). To elicit the effects of growth factor switching, several types of control proteins were synthesized, and all of these proteins were subjected to biological assays using the cell-based microarray method (12). The MCP was further immobilized onto glass-based substrates for quantitative assays. The results of these assays provided a proof of principle for our approach toward the sequential regulation of NSC differentiation.

EXPERIMENTAL PROCEDURES Chimeric Proteins. Figure 1 shows the domain structure of MCP and control proteins synthesized in this study. The MCP contains full length sequences of human CNTF (200 aa) and human EGF (53 aa). These polypeptides were linked with a

10.1021/bc700355t CCC: $40.75  2008 American Chemical Society Published on Web 01/11/2008

Multifunctional Chimeric Proteins

Bioconjugate Chem., Vol. 19, No. 2, 2008 517

Figure 1. Domain structure of multifunctional chimeric proteins (CP-1) and control proteins (CP-2 to CP-5) synthesized in this study. TRX, thioredoxin; EGFP, enhanced green fluorescent protein; TCS, thrombin cleavage site; mTCS, mutated-TCS; CNTF, ciliary neurotrophic factor; EGF, epidermal growth factor.

decahistidine linker sequence (His10) that facilitates specific binding of the chimeric protein to the substrates carrying Ni(II) ions through coordination (13). As a capping polypeptide, enhanced green fluorescent protein (EGFP; 243 aa) was fused to the N-terminal of CNTF with a thrombin-cleavable site (TCS; LVPRGS) (14) incorporated in between. At the N-terminal of EGFP, thioredoxin (TRX; 109 aa) was further added because this polypeptide is frequently used to enhance the solubility of fusion partners (15). In addition, amino acid sequences, (GS)3 and (EK)5 (16), were inserted at both ends of TCS and His10, respectively, as flexible linkers. The total number of amino acid residues reached 668 aa. The MCP protein is referred to as CP1. The other control proteins include a MCP with a mutated TCS (mTCS; LVPAGS) instead of the authentic TCS described above (Figure 1b; CP-2), a MCP lacking EGF (Figure 1c; CP3), and either EGF or CNTF fused with a hexahistidine sequence (Figure 1d and e; CP-4 and CP-5, respectively). Outline of the preparation of chimeric genes is depicted in Figure S1 (Supporting Information). Sequences of primers and oligonucleotides used for the construction of the CP-1 gene are shown in Table S1 (Supporting Information). The conditions of thermal cycling employed in polymerase chain reaction (PCR) amplification are shown in Table S2 (Supporting Information). First, synthetic oligonucleotides were subjected to overlap extension and PCR to prepare DNA encoding an (EK)6-His10(EK)6 sequence containing an EcoR I site at the 5′-end and 22mer oligonucleotides at the 3′-end that overlapped with the 5′region of the EGF gene. The EGF gene was amplified by PCR from pET-22b-EGF (12) using specific primer sets (Table S1). The 5′-end of the forward primer was flanked with 20-mer oligonucleotides that overlapped with the 3′-region of (EK)6His10-(EK)6. In addition, the reverse primer contained the Xho I site at the 5′-end. The products of overlap extension PCR were annealed, extended again by overlap extension, and amplified by PCR to obtain the (EK)6-His10-(EK)6-EGF gene. The chimeric gene was digested with EcoR I and Xho I and unidirectionally ligated to the pET-32b vector (Novagen) previously digested with the same restriction enzymes. The plasmid (pET-32b-His10-EGF) was cloned in Escherichia coli (E. coli) strain DH5R and purified with Qiagen MiniPrep Purification Kit. The correctness of the chimeric gene was checked by sequencing. At the second step, cDNAs for EGFP and CNTF were amplified by PCR from pEGFP-C1 (Clontech) and pET-16bNidLBD-CNTF (18), respectively. The forward primer used for the EGFP gene contained the Msc I site, while the reverse primer contained a sequence for (GS)3-TCS-(GS)3 and a 21-mer sequence that overlapped with the 5′-region of the CNTF gene. Similarly, the forward primer used for the CNTF gene contained a 22-mer sequence that overlapped with 3′-region of the EGFP

gene and a sequence for (GS)3-TCS-(GS)3, while the reverse primer contained the EcoR I site. The PCR products were annealed, extended by overlap extension, and amplified by PCR to obtain the EGFP-(GS)3-TCS-(GS)3-CNTF gene. The chimeric gene was digested with Msc I and EcoR I and unidirectionally ligated downstream of the TRX sequence in pET-32b-His10EGF linearized by digesting with the same restriction enzymes. The plasmid was cloned in E. coli strain DH5R and purified with Qiagen MiniPrep Purification Kit. The correctness of the chimeric gene was checked by sequencing. For constructing the plasmid for CP-2, the EGFP-(GS)3mTCS-(GS)3-CNTF gene was prepared using primers with mutation in the TCS site. First, EGFP-(GS)3-mTCS and mTCS(GS)3-CNTF genes were amplified using pET-32b-EGFP-TCSCNTF-His10-EGF as a template. Amplified DNAs were extended by overlap extension PCR to obtain EGFP-(GS)3-mTCS(GS)3-CNTF and inserted into the Msc I-EcoR I site of pET32b-His10-EGF. For CP-3, EGFP-(GS)3-TCS-(GS)3-CNTF was inserted downstream of TRX in original pET-32b. The gene for CP-4 was amplified by PCR from pET-16b-NidLBD-CNTF (18) and ligated to Nde I-Xho I sites in pET-22b (Novagen). For CP-5, previously constructed pET-22b-EGF (12) was used. The MCP (CP-1) and control proteins (CP-2 to CP-5) were expressed in E. coli strain BL21-CodonPlus (Stratagene) using Overnight Express Autoinduction System (Novagen). The MCP and control proteins obtained as inclusion bodies were extracted with 8 M urea and purified by Ni-chelated affinity chromatography (His Trap HP column; Amersham) using a Prime AKTA chromatography system (Amersham). These proteins were refolded by dialyzing against a solution of reduced and oxidized forms of glutathione. The details of dialysis conditions are provided in Table S3 (Supporting Information). The purity and the molecular size of MCP and control proteins were analyzed by sodium dodecylsulfate-polyacrylamide gel electrophoresis (SDS-PAGE). The biological activity of EGF contained in MCP and control proteins was assessed from the mitogenic activity in neurosphere culture (19). The biological activity of CNTF contained in CP-1, CP-3, and CP-4 was assessed from the efficiency of glial induction from NSCs cultured on fibronectin-coated dishes (19). The assays for CNTF activity in CP-1 and CP-3 were conducted after the capping domain was removed by thrombin digestion. Thrombin Digestion. MCP and control proteins (MCP with mTCS and EGF-deficient MCP) were separately dissolved to the concentration of 50 µg/mL in 20 mM tris-HCl buffer (pH 8.4) containing 0.6 unit/mL thrombin from bovine serum (POLA Pharma, Tokyo, Japan), 150 mM NaCl, and 2.5 mM CaCl2. The mixed solutions were kept at 37 °C for typically 45 min to allow for the digestion of TCS. The reaction was stopped by adding an equal volume of SDS-PAGE sample buffer contain-

518 Bioconjugate Chem., Vol. 19, No. 2, 2008

ing 10% β-mercaptoethanol. The digestion products were analyzed by SDS-PAGE with coomassie brilliant blue (CBB) staining. Microarray Preparation. Microarrays were prepared as reported previously (10). In brief, a thin gold film was deposited on two types of glass plates: glass slides (26 × 22 × 0.5 mm; Matsunami Glass Ind., Ltd., Osaka, Japan) for cell culture and S-LAL10 glass plates (25 × 25 × 1 mm, reflective index ) 1.720) for surface plasmon resonance (SPR) imaging. The selfassembled monolayer (SAM) of 1-hexadecanethiol was prepared on gold and photolytically micropatterned to create an array of 4 × 4 circular spots (2 mm in diameter and 4 mm in centerto-center distance) presenting bare gold. Then, the SAM of 11mercapto-1-undecanoic acid was formed within the spots. The carboxylic acid within spots was derivatized with N-(5-amino1-carboxypentyl) iminodiacetic acid and then chelated with Ni(II) ions. Solutions of MCP and control proteins (7 µM) in phosphate buffered saline (PBS) were manually pipetted (1.2 µL per spot) to separate spots on a single plate under sterile conditions. The plate was kept at room temperature for 2 h to allow the binding of decahistidine or hexahistidine sequence to the surface-immobilized Ni(II) ions. Finally, the plate was washed with PBS to remove unreacted proteins. Prior to cell culture experiments, the protein-immobilized plates were blocked with 2% Pluronic F127 (Sigma) solution. Surface Plasmon Resonance (SPR). SPR imaging was performed to detect immobilized proteins on the microarray. The details of a homemade SPR imaging apparatus are reported elsewhere (20). A protein-immobilized array prepared as described above was mounted on a triangular glass prism. A p-polarized, collimated white light was radiated to the back of the array through the prism (Kretschmann configuration) at a constant incident angle ca. 0.5° smaller than the angle for the occurrence of surface plasmon resonance at the region between spots on the array. To obtain a two-dimensional (2D) reflection image, reflected light was passed through a narrow band interference filter (center wavelength: 905 nm) and then collected with a charge-coupled device (CCD; C2400–79, Hamamatsu Photonics). The reflection image represents local changes in the reflective index in the vicinity of the sample surface and hence reflects the distribution of protein over the microarray. Imaging was carried out under PBS at 30 °C. For monitoring protein immobilization and thrombin digestion, a homemade SPR sensor (21) was used. A Ni(II)-chelated surface was prepared using mixed SAM consisting of 16mercapto-1-hexadecanoic acid and (1-mercaptoundec-11-yl) triethyleneglycol (molar ratio, 1:9) formed on the entire surface of a BK7 glass plate (25 × 25 × 1 mm, reflective index ) 1.515) by chemistry similar to that described above. The plate was mounted to the SPR sensor with a flow cell, and the resonance angular shift was continuously monitored during the following procedures. After equilibrating with PBS, the surface was exposed to 10 mg/mL bovine serum albumin solution in PBS in a flow cell for 10 min, and then pure PBS was circulated for 5 min to wash the surface. PBS was switched with 20 mM sodium phosphate buffer (pH 7.4) containing 20 mM imidazole (binding buffer) and circulated for 7 min, then 50 µg/mL MCP solution in binding buffer was circulated for 45 min followed by washing with pure binding buffer for 10 min. The surface was then exposed to 0.5 unit/mL thrombin solution in binding buffer for 50 min and washed with pure binding buffer. Finally, 20 mM sodium phosphate buffer (pH 7.4) containing 500 mM imidazole was circulated for 25 min to desorb protein fragments remaining on the surface and then switched with fresh PBS. The resonance angular shift was converted to the surface density of protein, assuming the unit angular shift of 0.5 µg/cm2/DA (22).

Nakaji-Hirabayashi et al.

Cell Isolation and Culture. The striatum was isolated from the fetuses (E16) of Fischer344 rats according to the guidelines of the Animal Welfare Committee of the institute and dissociated into single cells by treating with 0.05% trypsin solution containing 0.53 mM ethylenediamine-N,N,N′,N′-tetraacetic acid (EDTA). The single cells obtained were suspended in DMEM/ F12 (1:1) (Gibco) containing 5 µg/mL heparin, 100 unit/mL penicillin, and 100 µg/mL streptomycin (base medium), supplemented with 2% B27 supplement (Gibco), 20 ng/mL basic fibroblast growth factor (bFGF), and 20 ng/mL EGF, and cultured for 4–5 days to form neurospheres. Neurospheres at passage 2 were dissociated into single cells by treating with 0.05% trypsin-EDTA solution. The cells were suspended in the base medium containing 2% B27 supplement and then seeded onto the microarrays or the protein-immobilized glass plates at 3.0 × 104 cells/cm2. The cells were cultured in an incubator at 37 °C under 5% CO2. After 3 days, 1 unit/µL thrombin solution was added to the medium to a final concentration of 0.6 unit/mL and incubated for 45 min at 37 °C under 5% CO2, followed by washing with base medium to remove thrombin. Then, fresh base medium containing 2% B27 supplement was added to the culture, and cells were incubated for an additional 2–3 days at 37 °C under 5% CO2. Finally, cells were washed gently with DMEM/F12 (1:1) to remove weakly adhering cells and observed with a phase-contrast optical microscope (DP51, Olympus Optical CO., Ltd., Tokyo, Japan). Immunocytochemistry. Cells on the substrate were fixed with PBS containing 4% paraformaldehyde and 0.1% glutaraldehyde, and permeabilized by treating with 0.5% TritonX-100 solution at room temperature for 20 min. Then, the cells were treated with 2% skim milk solution for 2 h to block nonspecific adsorption of antibodies, followed by binding of primary antibodies against nestin (1:200, mouse monoclonal Rat 401, BD Pharmingen), glial fibrillary acidic protein (GFAP, mouse monoclonal, MAB3402, Chemicon), or class III β-tubulin (βIII; 1:600, rabbit polyclonal, Covance, Princeton, NJ) for 1 h at room temperature. After washing with PBS containing 0.05% Tween20, cells were treated by Alexa Fluor 594 antimouse IgG and Alexa Fluor 488 antirabbit IgG (both from Molecular Probes) at a dilution of 1:500 for 1 h at room temperature and washed with PBS containing 0.05% Tween-20. Then, cell nuclei were counterstained with 1 mg/mL Hoechst33258. The localization of secondary antibodies was analyzed with an epifluorescent microscope (DP70, Olympus). Cells reactive for antibodies against nestin, GFAP, and βIII were counted on the microphotographs for the area of 340 × 460 µm2, which contained at most approximately 500 total cells. Cells were counted on five different microphotographs obtained from the same sample, and these data were averaged. The data are shown as the mean ( standard deviation for five independent samples. Gene Expression Analysis. Total RNA was isolated from cells cultured on the protein-immobilized surfaces using the SV Total RNA Isolation system (Promega), and 100 ng RNA was used per reverse transcription (RT) reaction using Ready-toGo You-Prime First-Strand Beads (Amersham) primed by oligo(dT)15 (Promega). First-strand cDNA was then amplified by PCR using the specific primers listed in Table S4 (Supporting Information). The number of thermal cycles was determined so as to maintain the targeted sample population in the exponential amplification phase for each gene. All samples were run in triplicate, and the products were analyzed by electrophoresis in 2% agarose gel with ethidium bromide staining. Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was also

Multifunctional Chimeric Proteins

Figure 2. SDS-PAGE analysis of MCP and control proteins expressed in E. coli and purified by metal chelate chromatography. Proteins were electrophoresed in 12.5% polyacrylamide gel at 200 V for 40 min and visualized by CBB staining. Lanes 1 and 7, molecular weight standard; lane 2, CP-1; lane 3, CP-2; lane 4, CP-3; lane 5, CP-4; and lane 6, CP-5.

amplified as the reference gene. The electrophoresed DNA was imaged with Gel Doc 2000 and Integration Control Unit (BioRad).

RESULTS Characterization of Chimeric Proteins. Figure 2 shows the result of SDS-PAGE analysis for the MCP and control proteins. The molecular weight estimated from the mobility of protein bands are as follows: CP-1 (79 kDa), CP-2 (79 kDa), CP-3 (63 kDa), CP-4 (26 kDa), and CP-5 (7.5 kDa). Because low molecular weight CP-5 could not be separated appropriately with 12.5% polyacrylamide gel (Figure 2, lane 6), SDS-PAGE was additionally performed for CP-5 with 16% polyacrylamide gel to more precisely determine the molecular size (Figure S2, Supporting Information). The molecular weights estimated for CP-1 to CP-5 are in accordance with those expected from the theoretical amino acid numbers. These chimeric proteins exhibited the bioactivity of CNTF and/or EGF when qualitatively assessed in proliferation and differentiation assays with NSCs (data not shown). In CP-5, hexahistidine sequence was fused to the C-terminal of EGF, whereas decahistidine sequence was fused to the N-terminal of EGF in CP-1 and CP-2. Our separate experiments showed that the differences in the length and the position of a histidine tag had minor effects on the bioactivity of EGF fusions (data not shown). CP-1, CP-2, and CP-3 were digested with 0.6 unit/mL thrombin at 37 °C for 45 min. Digested products were analyzed by SDS-PAGE (Figure 3). As shown in Figure 3A, digestion of CP-1 resulted in the formation of protein fragments of approximately 40 and 35 kDa, corresponding to the N- and C-terminal regions spanning from TCS, respectively. Similarly, CP-3 produced approximately 40 kDa (from N-terminal to TCS) and 25 kDa fragments (from TCS to C-terminal) (Figure 3C). In contrast, CP-2 containing mTCS (LVPAGS) remained intact even after treatment with thrombin (Figure 3B). These results are reasonable because thrombin recognizes the TCS (LVPRGS) to hydrolyze the peptide bond between arginine and glycine (14). There is no consensus sequence in the other regions of CP-1. Thrombin digestion of CP-1, CP-2, and CP-3 in a culture medium gave results similar to those in buffer solution (Figure S3, Supporting Information). The higher concentration of thrombin (5 unit/mL) or prolonged reaction (90 min at 0.6 unit/ mL) gave rise to the formation of much smaller fragments (data not shown), probably because of nonspecific hydrolysis at other

Bioconjugate Chem., Vol. 19, No. 2, 2008 519

arginine-glycine linkages contained in the proteins. In addition, NSCs cultured on the CP-1-immobilized surface (see below for immobilization) were totally detached from the substrate upon contacting with 1.5 unit/mL thrombin. From these findings, we decided to digest the TCS with 0.6 unit/mL thrombin for 45 min in the later experiments. Immobilization of Chimeric Proteins. Chimeric proteins were immobilized on the micropatterned surface through the coordination of His10 with Ni(II) ions fixed on the spots. A microarray thus obtained was analyzed with the SPR imaging technique (Figure 4A). As can be seen, brightness is higher in all of the protein-immobilized spots than in the control spot, confirming the immobilization of proteins. The intensity of reflected light is varied depending on the amount of chimeric proteins immobilized (Figure 4B), with a linear correlation to the molecular size of immobilized proteins. To examine whether the capping domain can be removed from the immobilized chimeric protein by thrombin digestion, the immobilization of CP-1 and its digestion were performed in a flow cell of the SPR sensor while resonance angular shift was monitored. As can be seen from the sensorgram (Figure 5), CP-1 was immobilized to the Ni(II)-chelated surface pretreated with BSA. The resonance angular shift due to CP-1 immobilization reaches 1.4 DA, which corresponds to the surface density of 0.7 µg/cm2 (22). The resonance angle was gradually reduced by exposing the surface to thrombin solution (g f h in Figure 5). The ratio of the resonance angular shifts due to decapping and total recovery of decapped CP-1 by imidazole exposure was determined to be 0.57, which is in accordance with the ratio (0.54) of the numbers of amino acid residues corresponding to N- and C-terminal regions spanning from TCS. These results indicate that TCS was cleaved by thrombin, leaving decapped CP-1 on the surface. Microarray-Based Assay. NSCs were cultured on the microarray that displayed MCP and control proteins. The microarray also contained spots onto which CP-3 and CP-5 or CP-4 and CP-5 were coimmobilized from their equimolar mixture. After a 3-day culture, an entire array was treated with 0.6 unit/mL thrombin for 45 min, and then cells were cultured for an additional 2 days. Markers for undifferentiated cells (nestin) and differentiated neurons (βIII) and astrocytes (GFAP) were immunologically stained, while nuclei were stained with Hoechst dye (Figure 6). The microarray-based method facilitated parallel and rapid assays under the same conditions for every experimental variation. Furthermore, the planar surface of the microarrays enabled us to distinguish reasonably small differences in marker expression, in marked contrast to the conventional culture systems with plastic dishes or microplates. As shown in Figure 6A, substantial numbers of cells are seen on the spots with immobilized CP-1, CP-2, and CP-5. In contrast, a few cells are seen on the spots with CP-3, CP-4, CP-3/CP-5, and CP-4/CP-5, being similar to the spot with bare SAM. Taking into account the structure of chimeric proteins, these results indicate that the presence of the EGF domain is crucial for the effective adhesion and proliferation of the cells. This is reasonable because NSCs are considered to be trapped initially on the surface by EGF-EGF receptor (EGFR) interactions, receiving the EGF signal for proliferation (17). However, the coimmobilization of CP-5 could not improve cell adhesion and proliferation on the spots with CP-3/CP-5 and CP-4/CP-5. Because coimmobilization requires separate binding sites for each component, the surface density of coimmobilized CP-5 on the spots with CP-3/CP-5 and CP-4/CP-5 may be smaller than that on the spot with CP-5 alone. This is confirmed by the observation (Figure 4B) that the total amount of proteins on spots with CP-3/CP-5 and CP-4/CP-5 was approximately half

520 Bioconjugate Chem., Vol. 19, No. 2, 2008

Nakaji-Hirabayashi et al.

Figure 3. SDS-PAGE analysis of chimeric proteins digested by thrombin. (A) CP-1, (B) CP-2, and (C) CP-3 were digested with 0.6 unit/mL thrombin at 37 °C for 45 min. Then the products were analyzed with SDS-PAGE with 12.5% polyacrylamide gels with CBB staining. Lanes 1, 5, and 8, molecular weight standard; lane 2, thrombin; lane 3, 6, and 9, intact protein; lane 4, 7, and 10, proteins digested with thrombin. Arrows represent the protein band for (a) intact protein, (b and c) digested fragments, and (d) thrombin.

Figure 5. Surface plasmon resonance (SPR) analysis for the immobilization and thrombin treatment of multifunctional chimeric protein. Arrows (a-k) represent the time when the following solutions were injected to the Ni(II)-chelated surface: (a) PBS, (b) 1% BSA in PBS, (c) PBS, (d) 20 mM sodium phosphate containing 20 mM imidazole (pH 7.4; binding buffer), (e) 50 µg/mL CP-1 in binding buffer, (f) binding buffer, (g) 0.6 unit/mL thrombin solution in binding buffer, (h) binding buffer, (i) 20 mM sodium phosphate containing 500 mM imidazole (pH 7.4), (j) binding buffer, and (k) PBS.

Figure 4. Surface plasmon resonance (SPR) imaging of the microarray. (A) SPR image of the microarray onto which various chimeric proteins were immobilized. Protein identification was indicated on the image. CP-3/CP-5 and CP-4/CP-5 represent the fact that equimolar mixtures of these proteins were used for immobilization. COOH-SAM represents a control spot without immobilized protein. (B) Intensity of reflected light determined on the spots. The difference from the intensity on the control spot (COOH-SAM) was plotted. The data are expressed as the mean ( standard deviation for three independent experiments.

of the sum on the spots with CP-3 and CP-5 and on the spots with CP-4 and CP-5, respectively. The results of immunocytochemical staining are shown in Figure 6B and C. The comparison between Figure 6A and B shows that the majority of cells expressed nestin regardless of the spots. A striking result is that GFAP expression is most notable on the spot with CP-1 and moderate to negligible on the other spots (Figure 6C). Given the fact that CNTF strongly induces astrocytic differentiation from NSCs (11), the result just

described gives evidence that the CNTF domain in the immobilized CP-1 was unveiled upon decapping to act effectively on the cells. Quantitative Assay of Differentiation. Cells were cultured on the glass substrate onto which CP-1 was immobilized (Figure 7). In this experiment, the entire surface of a glass plate was modified with the protein for precise quantification. The results of comparable experiments for the CP-5-immobilized substrate are given in Figure S4 (Supporting Information). From these microphotographs, the total number of cells and the fraction of cells positive for nestin, βIII, and GFAP were determined (Figure 8). On the CP-5-immobilized surface, cells proliferated rapidly, and more than 95% of total cells expressed nestin with negligible expression of βIII and GFAP (