Heme Protein Assisted Dispersion of Gold Nanoparticle Multilayers on

Apr 9, 2009 - The AuNP ERE chips are shown to have high sensitivity and specificity for quantitative detection of ERE binding with its two transcripti...
0 downloads 0 Views 2MB Size
Anal. Chem. 2009, 81, 4076–4081

Heme Protein Assisted Dispersion of Gold Nanoparticle Multilayers on Chips: From Stabilization to High-Density Double-Stranded DNAs Fabricated in Situ for Protein/DNA Binding Yu-Ting Li,† Chun-Wei Li,‡ Wang-Chou Sung,† and Shu-Hui Chen*,† Department of Chemistry, National Cheng Kung University, No. 1 College Road, Tainan, 701, Taiwan, and Department of Medical Imaging and Radiological Sciences, College of Health Science, Kaohsiung Medical University, Kaohsiung, Taiwan Heme proteins in general are shown to be an effective linking agent in stabilizing gold nanoparticles (AuNPs) and thus facilitate the fabrication of three-dimensional (3D) AuNP multilayers on a chip, resulting in a higher coating density than that on polymer linker anchored surfaces for analytical applications. With the use of electron spectroscopy for chemical analysis (ESCA) measurements, a lower oxidation state of Au(0) and dramatic changes among multiple chemical states of N1s are detected upon coating AuNPs with heme proteins but not detected upon coating AuNPs with non-heme proteins. Thus, we propose that the stabilization power arises from π-conjugation between AuNPs and the heme group. We also propose that such conjugation must be facilitated by the exposure of the heme group through a conformational change of the protein as well as interactions of other functional groups with AuNPs to bring the heme moiety to a close face-to-face distance with the AuNPs. A highdensity double-stranded DNA (dsDNA) composed of a sequence of estrogen response element (ERE) is then fabricated on heme protein anchored chips. An in situ hybridization and tracking method is developed based on hybridization-induced fluorescence restoration associated with AuNPs and assists in the subsequent detection of DNA/protein binding on the same chip. The AuNP ERE chips are shown to have high sensitivity and specificity for quantitative detection of ERE binding with its two transcription factor isoforms, estrogen receptors r and β (ERr and ERβ), in cell lysates with reduced reagents and reaction time. Immobilization of gold nanoparticles (AuNPs) on a substrate is highly demanded in many analytical applications since they can be easily modified by a thiolated molecule. A stable and high density of AuNPs on a substrate can be used to fabricate highdensity ligands with high capacity and high sensitivity for sensing, whereas by current methods using polymer linkers such as * Corresponding author. † National Cheng Kung University. ‡ Kaohsiung Medical University.

4076

Analytical Chemistry, Vol. 81, No. 10, May 15, 2009

aminopropyltriethoxysilane (APTES)1-3 and (3-mercaptopropyl)trimethoxysilane (MPTS),4,5 the coating density of AuNPs on a substrate is low and the stability is also poor, requiring surface activation to create hydroxyl groups. In a recent study,6 myoglobin (Mb) was demonstrated as an effective linking agent for the layerby-layer fabrication of AuNPs/protein nanocomposite films in which AuNPs are spatially separated from one another; the film was stable in air and capable of withstanding harsh acid/base immersions. It was generally believed that under acidic conditions, Mb molecules carry multiple positive charges (NH3+) on their outer surface that can strongly adsorb onto negatively charged subjects, such as glass substrates or nanoparticles, to form stable composites or films.7-9 Since Mb is a heme protein, however, we suspect its heme moiety plays a role in stabilizing the nanocomposites on substrates through π-conjugation because the heme moiety is an analogue of the porphyrin molecule which is one of the most important π-conjugated compounds. Recent studies indicate that porphyrin can form a self-organized monolayer on the Au(111) surface due to the coordination of the porphyrin macrocycle to Au(111).10 Macrocyclic porphyrin ligands have also been synthesized to form quite stable Au(0) porphyrin nanocomplexes, with porphyrin rings parallel to the AuNP surface.11 The existence of the Soret band, tunable by the distance between the ring and the AuNP surface, implies an electronic interaction between the AuNP and porphyrin.11 Thus, we suspect that such unique properties should be applicable to other heme (1) Wang, Y.; Qian, W. P.; Tan, Y.; Ding, S. H. Biosens. Bioelectron. 2008, 23, 1166–1170. (2) Das, J.; Huh, C. H.; Kwon, K.; Park, S.; Jon, S.; Kim, K.; Yang, H. Langmuir 2009, 25, 235–241. (3) Endo, T.; Kerman, K.; Nagatani, N.; Takamura, Y.; Tamiya, E. Anal. Chem. 2005, 77, 6976–6984. (4) Huang, H. Z.; Liu, Z. G.; Yang, X. R. Anal. Biochem. 2006, 356, 208–214. (5) Jena, B. K.; Raj, C. R. Chem.sEur. J. 2006, 12, 2702–2708. (6) Qi, Z. M.; Honma, I.; Ichihara, M.; Zhou, H. S. Adv. Funct. Mater. 2006, 16, 377–386. (7) Lvov, Y.; Ariga, K.; Ichinose, I.; Kunitake, T. J. Am. Chem. Soc. 1995, 117, 6117–6123. (8) Qi, Z. M.; Matsuda, N.; Takatsu, A.; Kato, K. Langmuir 2004, 20, 778– 784. (9) Lee, J. E.; Saavedra, S. S. Langmuir 1996, 12, 4025–4032. (10) Katsonis, N.; Vicario, J.; Kudernac, T.; Visser, J.; Pollard, M. M.; Feringa, B. L. J. Am. Chem. Soc. 2006, 128, 15537–15541. (11) Kanehara, M.; Takahashi, H.; Teranishi, T. Angew. Chem., Int. Ed. 2008, 47, 307–310. 10.1021/ac900295j CCC: $40.75  2009 American Chemical Society Published on Web 04/09/2009

proteins in general and used as a versatile means for device fabrication. However, unlike porphyrins, multiple interaction forces could happen for a protein due to its multifunctionalties, and more investigations are required to make a good use of heme proteins as linking agents. In addition to feasible surface immobilization, AuNPs could form a resonance energy transfer system12-14 with organic dye molecules, leading to conformational-induced fluorescence quenching/restoration.15 We intend to take such advantage in developing an in situ hybridization and tracking method to synthesize highdensity double-stranded DNAs (dsDNAs) on a AuNP multilayer chip for sensing protein-DNA interaction. Protein-DNA interaction is related to networks formed by transcription factors, DNA, or chromatin regulatory proteins and their target genes. Finding transcriptional binding will help to unravel disease mechanisms. Conventionally, gel-based electrophoretic mobility shift assay (EMSA) is used to investigate protein-DNA binding,16 which, however, requires the use of the hazardous 32P isotopes, and it is also slow, labor-intensive, and quantification is difficult. EMSA was also demonstrated on microchip electrophoresis using fluorescence detection.17 In both EMSA methods, hybridization must be performed in a tube for forming dsDNAs through a heating/cooling process to minimize mismatch or hair loop before electrophoresis. Hybridization on an immobilized chip could reduce the chance of mismatch, and a tracking method for the process could improve the efficiency and assist in subsequent analyses on the same chip. In this study, a dsDNA sequence composed of estrogen response element is fabricated as a model to evaluate the AuNPs ERE chips by detecting the binding with its activated transcription factors (ERR and ERβ) from the total lysate of MCF7 and A549 cells. EXPERIMENTAL SECTION Fabrication of the Protein-Anchored AuNP Multilayer on Glass Chips. AuNPs with the size of 13 nm were prepared by sodium citrate reduction at a concentration around 11.6 nM, as reported earlier.18 Each protein (Mb, hemoglobin, cytochrome c, β-casein (Cas), bovine serum albumin, R-lactalbumin, lysozyme, and avidin) solution was prepared in deionized (DI) water at a concentration around 0.1 mM. For cross-linking, 0.6 mg of hemin was added to a 2 mL solution containing 40 mM N-(3-dimethylaminopropyl)-N′-ethylcarbodiimide hydrochloride (EDC), 10 mM (N-hydroxysuccinimide (NHS), and 30 mM NaCl. A volume of 0.5 mL of the prepared hemin solution was added into 100 mL of Cas solution (0.1 mM), and the mixture was allowed to react for 30 min at room temperature. Poly(dimethylsiloxane) (PDMS) was (12) Gueroui, Z.; Libchaber, A. Phys. Rev. Lett. 2003, 93, 166108-1-4. (13) Maxwell, D. J.; Taylor, J. R.; Nie, S. M. J. Am. Chem. Soc. 2002, 124, 9606– 9612. (14) Karvinen, J.; Laitala, V.; Makinen, M. L.; Mulari, O.; Tamminen, J.; Hermonen, J.; Hurskainen, P.; Hemmila, I. Anal. Chem. 2004, 76, 1429– 1436. (15) Li, Y. T.; Liu, H. S.; Lin, H. P.; Chen, S. H. Electrophoresis 2005, 26, 4743– 4750. (16) Hyder, S. M.; Nawaz, Z.; Chiappetta, C.; Stancel, G. M. Cancer Res. 2000, 60, 3183–3190. (17) Chuang, Y. J.; Huang, J. W.; Makamba, H.; Tsai, M. L.; Li, C. W.; Chen, S. H. Electrophoresis 2006, 27, 4158–4165. (18) Izrailev, S.; Stepaniants, S.; Balsera, M.; Oono, Y.; Schulten, K. Biophys. J. 1997, 72, 1568–1581.

used to form a 4 × 3 array wall, which was subsequently bonded on a glass slide using a casting/oxidation method as described previously.19 For layer-by-layer topping, each well (0.5 cm in diameter) was first soaked in the protein solution for 20 min and then in AuNP solution for 20 min; the soaking sequence was repeated several times until the desired number of layers was reached. The multilayer surfaces were investigated visually as well as by UV-vis spectrometry and scanning electron microscopy (SEM: surface scanning mode; voltage, 10 kV; amplification factor, 100 000-fold; vacuum value, 5.1 × 10-4 Pa; conduct layer, coating 30 nm thickness of Pt) for possible aggregation. The curve fitting for electron spectroscopy for chemical analysis (ESCA) spectra was performed with XPSPEAK41 software. Fabrication of Estrogen Response Element Probes on the Au Surface. The estrogen response element (ERE) DNA probes ((5′-Cy3-ACTTT GATCA GGTCA CTGTG ACCTG ACTTT GGAC-SH-3′) were dissolved in phosphate-buffered saline (PBS) buffer containing 8 mM sodium phosphate, 2 mM potassium phosphate, 140 mM NaCl, and 10 mM potassium chloride (pH 7.4) at a final concentration of 10 ppm. The prepared solutions were pipetted into each glass well immobilized with proteinanchored AuNP multilayers and incubated overnight. The modified wells were washed three times with the PBS buffer to remove unreacted reagents. A concentration of 20 ppm of the samples including the complementary strand (5′-GTCCA AAGTC AGGTC ACAGT GACCT GATCA AAGT-3′) and noncomplementary strand (5′-GTCCA AAGTC AATCG CCAGC ACGAT GATCA AAGT-3′) dissolved in PBS buffer were added into each individual well. The PBS buffer was also added for comparison. The samples were left to hybridize for 30 min at room temperature, and the modified wells were washed three times with the PBS buffer to remove unhybridized DNAs. Then, the fluorescence image of the wells was taken before and after the hybridization using a reflection fluorescence microscope (model BX40, Olympus, Tokyo, Japan) to ensure that dsERE was formed on the chips. Transcription Factor Binding. To detect the transcription factors that bind to ERE fabricated on chips, ERR and ERβ in MCF7 and A549 cells were the target protein to be detected and quantified by AuNPs ERE chips. Four dishes of cultured cells were starved for 18 h, then treated with dimethyl sulfoxide (DMSO) (vehicle control) or 17β-estradiol (10-8 M) for 24 h. PBST buffer (0.05% Tween 20 in the PBS buffer) was used as the wash buffer to remove the unbound analytes between each step. The AuNPs ERE chips were added with the blocking buffer (0.05% Tween 20 in the PBS buffer with 5% w/w milk) to reduce nonspecific binding. Subsequently, the human recombined ERR (Invitrogen, California), treated MCF-7, and A549 cells were loaded into individual wells and then incubated for 2 h. For enzyme-linked immunosorbent assay (ELISA), the polyclonal capture antibody (mouse anti-ERR and anti-ERβ, diluted 500fold with PBST buffer) and the secondary antibodies conjugated with peroxidase (diluted 1000-fold with PBST buffer) as well as the binding substrate 3,3′,5,5′-tetramethylbenzidine were sequentially added to produce the emission signal for detection using a photomultiplier tube (PMT) placed on a microscope or an ELISA reader (TECAN, Austria) at a wavelength of 450 (19) Sung, W. C.; Chang, C. C.; Makamba, H.; Chen, S. H. Anal. Chem. 2008, 80, 1529–1535.

Analytical Chemistry, Vol. 81, No. 10, May 15, 2009

4077

Figure 1. Protein-anchored AuNP multilayers (n ) 4) and APTESanchored monolayer on glass.

nm. The intensity was digitized by ImageJ software version 1.41 (http://rsb.info.nih.gov/ij/download.html), and each condition was repeated at least three times for quantification.

Figure 2. SEM and photo images of (a) Mb-anchored AuNP monolayer, (b) Mb-anchored AuNP multilayer (n ) 4), (c) Casanchored AuNP monolayer, and (d) H-Cas-anchored AuNP monolayer on glass.

RESULTS AND DISCUSSION In order to investigate whether heme proteins in general are unique anchoring agents for forming stable AuNP multilayers on chips, we first examined several heme proteins (Mb, hemoglobin, cytochrome c) and non-heme acid proteins (Cas, bovine serum albumin, R-lactalbumin) as well as a non-heme basic protein (lysozyme). As a comparison, we also investigated APTESfunctionalizedsurfaces.MultiplefactorsinvestigatedforAuNP-linker molecule interaction and their consequences are summarized in Table 1, and detailed descriptions are in the following sections. As shown in Figure 1, for all heme protein anchored surfaces, AuNPs could be uniformly coated with pink or deep pink color, and there was no significant aggregation throughout the multilayer (n g 4) formation, indicating that all heme proteins investigated could act as an effective linking agent for forming stable AuNP multilayers. However, on non-heme protein anchored surfaces, AuNPs were either coated with very low density such as Cas and R-lactalbumin or appeared to show a dark blue color as an indication of aggregations such as lysozyme and bovine serum albumin. Lysozyme is a strong basic protein with an isoelectric point around 11 and was suspected to be a good linking agent for AuNPs6 due to electrostatic interactions with the negatively charged citrate-protected AuNP surface. However, instead of

stabilization, our observations suggest that such strong electrostatic interactions arising from charged protein functional groups could lead to aggregation. On the APTES-functionalized substrate, AuNPs were coated with a low density as observed for non-heme proteins. In order to examine whether the heme group could assist in AuNP dispersion, we cross-linked Cas with hemin, an ironchelated porphyrin, and used the cross-linked hemin-β-casein (H-Cas) as the linking agent for AuNPs. On the H-Cas surface, AuNPs were uniformly coated on both the monolayer and the multilayers as on the Mb surface (Figure 1), indicating the contribution of heme groups in AuNPs stabilization. SEM was also used to examine the monolayer and multilayer surfaces. As shown in Figure 2, parts a and b, the coating density of AuNPs on the Mb-anchored surface greatly increased from monolayer to multilayers with good dispersion. Moreover, the heme protein anchored multilayers were all stable when exposed to air for more than 1 week. As shown in Figure 2c, AuNPs on the Cas-anchored monolayer are much lower in density than those on the Mbanchored surface (Figure 2a), and some local aggregation was noted, whereas on the H-Cas surface, AuNPs were densely coated and well dispersed on the monolayer (Figure 1) as well as on the multilayers without notable aggregation. These results strongly suggest that the heme group of anchored proteins play

Table 1. Multiple Factors for the Linker-AuNP Interactions linker type linker name

protein hemin cross-linked non-heme protein layer examined monolayer and multilayer monolayer and multilayer functional groups 1. NH2 or NH3+ 1. NH2 or NH3+ 2. COO- or COOH 2. COO- or COOH 3. heme 3. heme stability stable and high coating density spectroscopic evidence (a) one additional chemical state with lower energy for AuNP-linker interaction associated with both 5/2Au and 7/2Au peaks

comments

4078

heme proteins

(b) variations in the chemical states of N1s peak, particularly for the additional state near 401 eV π-conjugation facilitated by multiple interaction forces and conformational changes of the protein to expose the heme moiety, leading to stabilization

Analytical Chemistry, Vol. 81, No. 10, May 15, 2009

non-heme proteins monolayer and multilayer 1. NH2 or NH3+ 2. COO- or COOH

polymer SAM APTES monolayer 1. NH2 or NH3+

instable or low coating density low coating density (a) no change in chemical state (a) no change in chemical associated with both 5/2Au state associated with both and 7/2Au peaks and 7/2Au peaks (b) no change in the chemical states of N1s peak electrostatic interactions a low density of silane between protein and AuNPs SAM created on the glass which is either too surface for AuNPs weak or too strong immobilization

5/2

Au

Figure 3. 5/2Au and 7/2Au ESCA spectra for (a) Mb-AuNPs, (b) H-Cas-AuNPs, (c) Cas-AuNPs, and (d) aminosilane-functionalized AuNPs.

Figure 4. N1s core level of ESCA spectra for (a) Mb, (b) Mb-AuNP, (c) Cas, (d) Cas-AuNP, (e) H-Cas, and (f) H-Cas-AuNP on glass.

an important role in stabilizing AuNP composites on a surface. However, we had also investigated the surface coated with hemin molecules alone and found a low coating density as obtained from non-heme proteins, which we think multiple forces to bring the heme group to a closer proximity to AuNPs is the key for effective π-conjugation to occur. ESCA spectra were measured to investigate the molecular interactions between AuNPs and heme proteins to better understand the cause of stabilization. As shown in Figure 3, both 5/2Au and 7/2Au bands were detected from various AuNP surfaces, anchored by molecules with or without heme groups. Notably, for heme proteins, Mb-, and H-Cas-anchored AuNPs, the fitted spectra show two chemical states for each Au band; however, for non-heme molecules, APTES-, and Cas-anchored AuNPs, the fitted spectra exhibit only the elemental Au bands (5/2Au and 7/2Au). Apparently, cross-linking Cas with hemin resulted in an additional chemical state for AuNPs. These results indicate that the additional chemical state of Au is associated with the heme groups. Since the binding energy of the additional state is lower than that of the elemental states, oxidized Au ions are excluded for such heme-induced chemical states. Instead, Au with lower oxidation states, such as π-orbital interactions, is likely to occur to reduce the tunneling resistance of the surrounding ligands. To the best of our knowledge, there has been no report on the red-shift of binding energy for Au or AuNPs. With unique surface plasma, however, we expect AuNPs to exhibit π-coordination with ligands. Another way to exploit possible π-coordination between AuNPs and ligands is to examine the ESCA spectra of atoms that could serve as Lewis bases, such as N and O on proteins. As shown in

Figure 4, parts a and b, topping Mb with AuNPs resulted in a dramatic increase in the relative intensity between the 399 and 400 eV peaks and an additional peak at 401 eV in the N1s spectra. However, as shown in Figure 4, parts c and d, topping Cas with AuNPs resulted in only little changes in the peak profile. By comparison with the N1s spectra of amino acids, we concluded that the two chemical states (400 and 399 eV) of N1s for Cas, shown in Figure 4, parts c and d, correspond to the N atoms on the ε-amino group and the N-terminus of the amide bond, respectively. On the other hand, the N1s peaks for heme groups of Mb could not be unambiguously assigned based on comparisons with the reported N1s spectra of porphyrin derivatives10,11 because of the overlapping of the bond energy with amide nitrogens. For porphyrin derivatives on Au(111) and AuNPs, pyrrolic (-NH-) nitrogen was reported to be 400.2 eV, which could overlap with N for the ε-amino group; free iminic (-CdN-) nitrogen was reported to be 397.5 eV. The important evidence for the coordination between the metal surface and iminic nitrogens was reported to be 399.3 eV, which, however, could overlap with N in the N-terminus of the amide bond. Nevertheless, the substantial change induced by AuNPs was found to be associated with Mb but not with Cas, indicating specific interactions between AuNPs and heme proteins. If Cas is cross-linked with hemin, as shown in Figure 4e for H-Cas, the N1s spectra change dramatically (compared to Figure 4c). Since hemin is derivatized to lysine residues which are most likely to expose to the protein surface, the derivatized hemin is expected to expose to the outer surface of Cas, leading to the dramatic change in Analytical Chemistry, Vol. 81, No. 10, May 15, 2009

4079

Figure 5. In situ hybridization-induced fluorescence restoration for sensing transcription factor binding on heme protein stabilized AuNP multilayers.

N1s spectra. Moreover, as shown in Figure 4f, topping H-Cas with AuNPs resulted in a significant increase in the intensity of the peak near 401 eV, which was also detected as an additional peak upon topping Mb with AuNPs, as shown in Figure 4b. Therefore, the 401 eV peak could be related to the interaction between AuNPs and the exposed heme groups. Mb molecules embedded in a nanocomposite film have been reported to exhibit conformational changes that cause shrinkage in air and swelling in water.6 Thus, we suspect that Mb undergoes a relatively slow conformational change to expose its heme groups in order to coordinate with the surface of AuNPs through π-conjugation. Such a conformational change could involve the replacement of heme-coordinated amino acid residues such as histidine with AuNPs and lead to a lower oxidation state of Au(0) as well as population variations among different chemical states of N1s. Because of the multiple chemical states of N1s shown in Figure 3b, we suspect that other functional groups of Mb could have interactions with the AuNPs surface to bring the heme group and AuNPs face to face to a distance over which coordination could occur through the π-electrons of the iminic (-CdN-) nitrogens. Hemin binding on the nanoparticle surface was also reported to occur through its carboxylic acid groups.20 Thus, we suspect that the carboxylic acids of the heme group or other residues of proteins could have interactions with AuNPs. Moreover, the positive charges (NH3+) of amino acid residues could also strongly adsorb to the citrate-protected surfaces of the AuNPs, which bear negative charges. In the previous report,11 a porphyrin ring was linked to AuNPs through thiol groups derivatized on the porphyrin ring, and its distance to the ring surface, which was reported to affect the Soret band intensity, was adjusted by the chain length of the linker. For heme (20) Tom, R. T.; Pradeep, T. Langmuir 2005, 21, 11896–11902.

4080

Analytical Chemistry, Vol. 81, No. 10, May 15, 2009

proteins, the three-dimensional (3D) structure could exhibit the existence of multiple bonds between Mb molecules and AuNPs. Once the AuNPs and heme rings were brought close enough by other functional groups, heme groups could densely protect AuNPs in a face-coordination fashion to form very stable Au(0) heme complexes through π-coordination. Whether or not the chelated iron of the heme groups was replaced by the coordinated AuNPs, however, cannot be confirmed from this study, since no iron signal could be detected by ESCA in heme proteins we studied with or without AuNPs. We then used the stable AuNPs multilayers anchored by Mb for fabricating dsDNA containing the sequence of ERE by an in situ method that can be tracked via hybridization-induced fluorescence restoration.15 ERE is the cis regulatory sequence of estrogen receptors (ERs) located in the promoter region of the target genes and the ERE-ER binding is the key for transcriptional activation. As depicted in Figure 5, DNA probes were bonded to the monolayer via thiol groups at one end, and a fluorophore dye (Cy3) was attached to the other end of the probe. The construct created is spontaneously assembled into a constrained archlike conformation on the particle surface, and the fluorescence of Cy3 is quenched by AuNPs. Hybridization of target DNAs results in a conformational change of the construct and then restores the fluorescence, which serves as a tracking method for the formation process of dsDNA. Such a synthesis and tracking method for the hybridization of dsDNAs on chips is very useful for total analyses on a chip, since it is in situ and requires few amounts of reagent and few buffer exchange steps. In comparison to hybridization in a tube, a constant temperature can be used on a chip, eliminating tedious temperature control steps to prevent mismatch, and because of the tracking capability, subsequent analyses can be performed without delay on the same chip, which

Figure 6. Hybridization-induced fluorescence (Cy3) restoration for the fabrication of dsDNAs containing ERE sequences. No fluorescence signal was detected when noncomplementary DNA and blank solution were added.

Figure 7. Transcription factor binding assays using Mb-AuNP chips immobilized with ERE dsDNAs. The emission signal for sensing the binding of ERR and ERβ was detected from the control rERR sample as well as 17β-estradiol-treated MCF-7 and A549 cells. Signals detected from three repeated measurements were quantified with their standard deviations indicated by arrow bars.

is a kind of “one-pot reaction”. Such a continuous scheme will be very useful in developing micrototal analyses on a chip. As shown in Figure 6, the fluorescence was restored when the cDNA was hybridized on chip, and no fluorescence could be detected when buffers or noncomplementary DNA was added. Because of the 3D multilayer, ERE chips fabricated on the AuNPs multilayer are expected to have a relatively high bonding density compared to the polymer linker anchored surfaces. The coating density for Mbanchored AuNPs multilayer which shows good stability was estimated to be around 600 particles/µm2. Since one AuNP can bind to one or two fluorescein-DNA probes,13 the coating density for dsDNAs was estimated to be around 600-1200 molecules/µm2. Using such functionalized surface, we expect to gain detection sensitivity that is capable of sensing lowabundance proteins in cells. Two cell lines, breast cancer cell MCF7 and lung cancer cell A549, which are known to express two different isoforms of the estrogen receptor (ERR and ERβ), were tested. MCF-7 cells are

known to express more ERR than ERβ; A549 cells are known to express ERβ but very little ERR. The ligand-mediated activation of the nuclear receptors (ERR and ERβ) causes a direct binding of ER with ERE in the promoters of target genes and recruits various coactivators to mediate transcriptional regulation. Because the molecular aspects of the estrogen-related malignancy including cancers remain poorly understood, only a few drugs are currently used. Resolving ERE binding partners and binding strength will help to understand related molecular mechanisms as well as help to screen potential lead compounds. Both cells were treated with 17β-estradiol for 24 h, and the binding of activated ERR and ERβ to ERE on chips was detected using the ELISA. As expected, on the basis of the comparison with the signal detected from the control and the standard made of recombinant ERR, Figure 7 shows that MCF-7 cells were detected with more ERR than ERβ, and A549 cells were detected with ERβ but very little ERR. These results suggest that the ERE chips fabricated on heme protein anchored AuNP multilayers are suitable for real-world quantitative analysis with acceptable sensitivity and specificity. CONCLUSIONS We have shown that heme proteins in general can stabilize AuNPs and assist in the dispersion of AuNP multilayers on a chip, resulting in a higher coating density than those on polymer linker anchored surfaces. On the basis of spectroscopic studies, the stabilization power is believed to arise from π-conjugation between AuNPs and the heme groups, which are exposed to the surface by a conformational change of the heme protein and brought to a close face-to-face distance by other functional groups. Such binding assembly is formed through a self-activated 3D structural change of the heme protein and cannot be achieved by the heme molecule alone. For the first time, we have demonstrated a lower oxidation state of Au(0) induced upon conjugation with heme proteins. We suspect such conjugations could also occur between AuNPs and proteins containing many aromatic residues such as histidine and tryptophan, and these are currently under investigation. We have further demonstrated an in situ hybridization method on heme protein anchored AuNP multilayer chips with tracking capability for the process via hybridization-induced fluorescence restoration using a DNA sequence with Cy3 dye that could form a resonance energy transfer system with AuNPs. The chips are successfully demonstrated for sensing the transcriptional binding of estrogen receptors R and β in different cells quantitatively with high sensitivity and specificity. The reported methods are expected to hold great potential for many other applications in detecting protein-DNA binding. ACKNOWLEDGMENT The work was supported by the National Science Council in Taiwan. We would like to thank the technical support of the NSC Instrument Center NCKU for ESCA measurements. Received for review February 7, 2009. Accepted March 23, 2009. AC900295J

Analytical Chemistry, Vol. 81, No. 10, May 15, 2009

4081