pubs.acs.org/Langmuir © 2009 American Chemical Society
Efficient Self-Assembly of Archaeoglobus fulgidus Ferritin around Metallic Cores Joe Swift, Christopher A. Butts, Jasmina Cheung-Lau, Vijay Yerubandi, and Ivan J. Dmochowski* Department of Chemistry, University of Pennsylvania, 231 South 34th Street, Philadelphia, Pennsylvania 19104-6323 Received December 10, 2008. Revised Manuscript Received February 9, 2009 Interfacing biological systems with inorganic nanoparticles is of great interest, as it offers means of particle stabilization and spatial control in electronic or biomedical applications. We report on the particle-directed assembly of hyperthermophile Archaeoglobus fulgidus ferritin subunits around negatively charged colloidal gold. An annealing process allows rapid assembly of the protein in near-native stoichiometry. Transmission electron microscopy suggests that greater than 95% of nanoparticles are encapsulated while the self-assembly process ensures that almost 100% of the assembled ferritin cavities are occupied.
1.
Introduction
One of the challenges facing nanoscale research concerns the ability to situate structures within larger size domains. Developments in fabrication methods have allowed particles to be made with high yield and homogeneity.1 However, in order to make effective devices, particles need to be accurately positioned with respect to each other and their surroundings.2 Research into the interface of nanoscale materials with biological constructs aims to take advantage of nature’s ability to self-assemble from the atomic scale into large functional machines. Biomedical applications require a similarly precise spatial control with the additional need for biocompatibility. An increased understanding of the nano/bio interface,3 initially driven by the development and study of model systems, will be necessary to assess biological hazard4 versus the potential benefits5 of nanomedicine. Much recent work has utilized the ferritin family of proteins as an intermediate between biological and inorganic systems. *Corresponding author. E-mail:
[email protected]. (1) Masala, O.; Seshadri, R. Annu. Rev. Mater. Res. 2004, 34, 41–81. (2) Murphy, C. J.; San, T. K.; Gole, A. M.; Orendorff, C. J.; Gao, J. X.; Gou, L.; Hunyadi, S. E.; Li, T. J. Phys. Chem. B 2005, 109, 13857–13870. (3) Cedervall, T.; Lynch, I.; Lindman, S.; Berggard, T.; Thulin, E.; Nilsson, H.; Dawson, K. A.; Linse, S. Proc. Natl. Acad. Sci. U.S.A. 2007, 104, 2050–2055. (4) Lewinski, N.; Colvin, V.; Drezek, R. Small 2008, 4, 26–49. (5) Service, R. F. Science 2005, 310, 1132–1134. :: (6) Johnson, E.; Cascio, D.; Sawaya, M. R.; Gingery, M.; Schroder, I. Structure 2005, 13, 637–648. (7) Liu, X. F.; Theil, E. C. Acc. Chem. Res. 2005, 38, 167–175. (8) Hainfeld, J. F. Proc. Natl. Acad. Sci. U.S.A. 1992, 89, 11064–11068. Meldrum, F. C.; Douglas, T.; Levi, S.; Arosio, P.; Mann, S. J. Inorg. Biochem. 1995, 58, 59–68. Meldrum, F. C.; Heywood, B. R.; Mann, S. Science 1992, 257, 522–523. Meldrum, F. C.; Wade, V. J.; Nimmo, D. L.; Heywood, B. R.; Mann, S. Nature 1991, 349, 684–687. Okuda, M.; Iwahori, K.; Yamashita, I.; Yoshimura, H. Biotechnol. Bioeng. 2003, 84, 187–194. Zhang, B.; Harb, J. N.; Davis, R. C.; Kim, J. W.; Chu, S. H.; Choi, S.; Miller, T.; Watt, G. D. Inorg. Chem. 2005, 44, 3738–3745. Douglas, T.; Dickson, D. P. E.; Betteridge, S.; Charnock, J.; Garner, C. D.; Mann, S. Science 1995, 269, 54–57. Douglas, T.; Stark, V. T. Inorg. Chem. 2000, 39, 1828–1830. (9) Iwahori, K.; Yoshizawa, K.; Muraoka, M.; Yamashita, I. Inorg. Chem. 2005, 44, 6393–6400. Wong, K. K. W.; Mann, S. Adv. Mater. 1996, 8, 928– 932. Yamashita, I.; Hayashi, J.; Hara, M. Chem. Lett. 2004, 33, 1158–1159. Yoshizawa, K.; Iwahori, K.; Sugimoto, K.; Yamashita, I. Chem. Lett. 2006, 35, 1192–1193.
Langmuir 2009, 25(9), 5219–5225
The multimeric proteins have a hollow cavity of diameter between 5 and 9 nm6,7 that has been loaded with various mineral,8 semiconducting,9 or metallic10,11 cargoes by processes of ion uptake through the pores that perforate the protein shell, followed by in situ reduction. As the native purpose of ferritin is in binding iron, the pores are best suited for the uptake of divalent cations. The formation of metallic cores, for example, gold, is complicated, as reduction occurs exterior to the protein faster than the metal ions can traverse the pores. Typically, this process results in protein-enclosed metal particles far smaller than the theoretical capacity of the cavity. Bulk metal reduced exterior to the protein must be removed, resulting in diminished yields.12 Reasons to consider ferritin as a potential biological building block are manyfold: Ferritin has shown itself to be amenable to the formation of high-quality ordered arrays, through two-dimensional crystallization13 and through attachment to a surface via external mutation.14 Additionally, the ferritin-related DNA-binding protein from starved cells (Dps) has, as its name suggests, been shown to bind to DNA, giving potential for further crystalline ordering.15 As ferritin is a multimeric system, there is scope for combining distinct (10) Butts, C. A.; Swift, J.; Kang, S.; Di Costanzo, L.; Christianson, D. W.; Saven, J. G.; Dmochowski, I. J. Biochemistry 2008, 47, 12729–12739. (11) Dominguez-Vera, J. M.; Galvez, N.; Sanchez, P.; Mota, A. J.; Trasobares, S.; Hernandez, J. C.; Calvino, J. J. Eur. J. Inorg. Chem. 2007, 4823–4826. Galvez, N.; Sanchez, P.; Dominguez-Vera, J. M.; SorianoPortillo, A.; Clemente-Leon, M.; Coronado, E. J. Mater. Chem. 2006, 16, 2757–2761. Galvez, N.; Sanchez, P.; Dominguez-Vera, J. M. Dalton Trans. 2005, 2492–2494. Ueno, T.; Suzuki, M.; Goto, T.; Matsumoto, T.; Nagayama, K.; Watanabe, Y. Angew. Chem., Int. Ed. 2004, 43, 2527–2530. (12) Zhang, L.; Swift, J.; Butts, C. A.; Yerubandi, V.; Dmochowski, I. J. J. Inorg. Biochem. 2007, 101, 1719–1729. (13) Okuda, M.; Kobayashi, Y.; Suzuki, K.; Sonoda, K.; Kondoh, T.; Wagawa, A.; Kondo, A.; Yoshimura, H. Nano Lett. 2005, 5, 991–993. Yuan, Z.; Petsev, D. N.; Prevo, B. G.; Velev, O. D.; Atanassov, P. Langmuir 2007, 23, 5498–5504. Furuno, T.; Sasabe, H.; Ulmer, K. M. Thin Solid Films 1989, 180, 23–30. Yoshimura, H.; Scheybani, T.; Baumeister, W.; Nagayama, K. Langmuir 1994, 10, 3290–3295. (14) Matsui, T.; Matsukawa, N.; Iwahori, K.; Sano, K.-I.; Shiba, K.; Yamashita, I. Langmuir 2007, 23, 1615–1618. (15) Almiron, M.; Link, A.; Furlong, D.; Kolter, R. Genes Dev. 1992, 6, 2646–2654. Surguladze, N.; Thompson, K. M.; Beard, J. L.; Connor, J. R.; Fried, M. G. J. Biol. Chem. 2004, 279, 14694–14702.
Published on Web 3/4/2009
DOI: 10.1021/la8040743
5219
Article
Swift et al.
subunits to give hybrid properties.16 We have also shown that proteins from the ferritin family can endure wide-scale mutagenesis while maintaining tertiary structure and stability.10,17 Here, we report a simple procedure for the encapsulation of gold nanoparticles within the cavity of ferritin from the hyperthermophilic archaeon Archaeoglobus fulgidus.6 In contrast to methods of in situ reduction, we gain access to the hollow interior by utilizing the hyperthermophile ferritin’s propensity to reversibly disassemble at low salt concentrations, thus bypassing the narrow and selective pores as a means to load the cavity. Parallels can be drawn to earlier work in which the cavities of horse spleen ferritin were loaded with the dye neutral red,18 a gadolinium complex,19 the chemotherapeutic doxorubicin,20 or PbS quantum dots21 by disassembling the protein into subunits at low pH, followed by reassembly in the presence of the cargo. However, these methods rely on such high concentrations that the cavity might fortuitously capture a payload molecule during reconstitution. Furthermore, the pH driven disassembly-assembly process gives poor yields due to near-denaturing conditions and repeat passage through the pI of the protein. In our method, the presence of preformed gold particles drives the protein to self-assemble. A similar mechanism of self-assembly has been reported by the Dragnea group, whereby poly (ethylene oxide)-coated gold particles,22 magnetic iron,23 and semiconducting quantum dots24 were encapsulated in virus capsids. The temperature stability of hyperthermophilic ferritin can give an additional advantage to fabrication, for example, in making maghemite cores.25 In this work, we show that the thermal stability allows us to employ an annealing process that gives high yields of encapsulation with native stoichiometry of the protein assembly.
2.
Experimental Section
Starting Materials. Acros Organics: mercaptosuccinic acid (MSA) and 4-dimethylaminopyridine (DMAP); British Biocell International: 10 nm diameter citrate-capped gold nanoparticles; Difco: bacto tryptone and granulated agar; Fisher: 1 N hydrochloric acid solution, 1 N sodium hydroxide solution, sodium chloride, trishydroxymethylaminomethane (Tris), disodium ethylenediamine tetraacetate (EDTA), tris(2-carboxyethyl)phosphine hydrochloride (TCEP), sodium phosphate monobasic, and bis(p-sulfonatophenyl)phenylphosphane (BSPP); Fisher Bioreagents: biotech grade ampicillin sodium salt, yeast extract, and enzyme grade 4-(2-hydroxyethyl)-1(16) Gillitzer, E.; Suci, P.; Young, M.; Douglas, T. Small 2006, 2, 962–966. Kang, S.; Oltrogge, L. M.; Broomell, C. C.; Liepold, L. O.; Prevelige, P. E.; Young, M.; Douglas, T. J. Am. Chem. Soc. 2008, 130, 16527–16529. (17) Swift, J.; Wehbi, W. A.; Kelly, B. D.; Stowell, X. F.; Saven, J. G.; Dmochowski, I. J. J. Am. Chem. Soc. 2006, 128, 6611–6619. (18) Webb, B.; Frame, J.; Zhao, Z.; Lee, M. L.; Watt, G. D. Arch. Biochem. Biophys. 1994, 309, 178–183. (19) Aime, S.; Frullano, L.; Geninatti Crich, S. Angew. Chem., Int. Ed. 2002, 41, 1017–1019. (20) Simsek, E.; Kilic, M. A. J. Magn. Magn. Mater. 2005, 293, 509–513. (21) Hennequin, B.; Turyanska, L.; Ben, T.; Beltran, A. M.; Molina, S. I.; Li, M.; Mann, S.; Patane, A.; Thomas, N. R. Adv. Mater. 2008, 20, 3592– 3596. (22) Chen, C.; Daniel, M. C.; Quinkert, Z. T.; Stein, M.; Bowman, B.; Chipman, V. D.; Rotello, P. R.; Kao, V. M.; Dragnea, C. C. Nano Lett. 2006, 6, 611–615. (23) Huang, X.; Bronstein, L. M.; Retrum, J.; DuFort, C.; Tsvetkova, I.; Aniagyei, S.; Stein, B.; Stucky, G.; McKenna, B.; Remmes, N.; Baxter, D.; Kao, C. C.; Dragnea, B. Nano Lett. 2007, 7, 2407–2416. (24) Dixit, S. K.; Goicochea, N. L.; Daniel, M.-C.; Murali, A.; Bronstein, L. M.; Stein, M.; Rotello, B.; Kao, V. M.; Dragnea, C. C. Nano Lett. 2006, 6, 1993–1999. (25) Parker, M. J.; Allen, M. A.; Ramsay, B.; Klem, M. T.; Young, M.; Douglas, T. Chem. Mater. 2008, 20, 1541–1547.
5220
DOI: 10.1021/la8040743
piperazine-ethanesulfonic acid (HEPES); Fisher Biotech: isopropyl-β-D-thiogalactoside (IPTG); Promega: molecular biology grade guanidine hydrochloride (GuHCl); GE Healthcare: Cy3 maleimide monoreactive dye; Roche Diagnostics GmbH: protease inhibitor cocktail tablets; Teknova: NZYM Plus broth; United States Biochemical Corporation: glutathione (GSH). All solutions were prepared using deionized water purified by an Ultrapure system.
Protein Expression, Purification, And Characterization. The pAF0834 vector containing the Archaeoglobus fulgidus hyperthermophilic ferritin gene was provided by the laboratory of Dr. Eric Johnson at the California Institute of Technology. The S80C mutant for fluorescent labeling was engineered using the Stratagene QuikChange Site-Directed Mutagenesis kit. Primers were obtained from Integrated DNA Technologies: sense ð50 f30 Þ GTTGAGGAGCCACCAT GTGAGTGGGATTCGCCTTTG antisense ð50 f30 Þ CAAAGGCGAATCCCA CTCACATGGTGGCTCCTCAAC The mutated cDNA was transformed into XL1-Blue Supercompetent E. coli cells (Stratagene) according to the manufacturer’s protocol. DNA was isolated using a QIAprep Spin Miniprep kit (Qiagen). All sequencing was performed by the University of Pennsylvania DNA Sequencing Facility (Department of Medicine, University of Pennsylvania).
The protein was expressed according to a procedure modified from that of Johnson et al.6,26 The pAF0834 plasmid containing the ferritin gene was transformed into BL21-CodonPlus (DE3)-RP competent cells of E. coli (Stratagene). Cell cultures were grown in 1 L of Luria-Bertani (LB) medium supplemented with 200 μg/mL ampicillin sodium salt (Fisher BioReagents) and shaken (37 C, 180 rpm) until OD560 ≈ 1.0. The cells were induced with 1 mM IPTG (FisherBiotech) and shaken (37 C, 4 h, 180 rpm). Cells were harvested by centrifugation and resuspended in buffer (50 mL, 10 mM Tris, 10 mM NaCl, pH 8.4) with a protease inhibitor cocktail tablet (Roche Diagnostics GmbH). The cells were sonicated on ice (3 2 min) and centrifuged (5000 rpm, 10 min), and then the lysis supernatant was heat shocked (80 C, 10 min) and ultracentrifuged (25 000 rpm, 1 h). The supernatant was purified on a 5 mL HiTrap FastFlow Q anion exchange column (GE Healthcare) (20 mM Tris, pH 8.4 with a 20-600 mM NaCl gradient, 5 mL/min flow rate). The protein was equilibrated in >1 M NaCl for 24 h, concentrated (100 000 MWCO Centricon, Millipore), and purified by size exclusion chromatography using a Superdex 200 10/300 GL column (GE Healthcare) with 20 mM sodium phosphate, 800 mM NaCl, pH 7.6 buffer at 0.5 mL/min. Circular dichroism (CD) spectra were recorded on an Applied Photophysics Chirascan instrument. Dynamic light scattering (DLS) measurements were made with a Dynapro instrument from Wyatt Technology. Ferritin Functionalization. Prior to conjugation, purified S80C protein (typically 1 mg) was treated with a 20 equiv excess of TCEP and incubated at 4 C for 2 h. The reducing agent was subsequently removed with a DG10 gel filtration column (BioRad) equilibrated with 20 mM NaCl, 50 mM phosphate buffer at pH 7. Cy3 dye was attached via reaction (26) Yerubandi, V. Manipulating Ferritins for Biomaterials Applications. Master of Science Thesis, University of Pennsylvania, 2007, pp 24-38.
Langmuir 2009, 25(9), 5219–5225
Swift et al.
Article
of maleimide with the surface-exposed cysteine, the reagent added in high excess as described in the manufacturer’s protocol and stirred under argon for 24 h. The reaction mixture was subsequently passed through a DG10 column equilibrated with 800 mM NaCl, 20 mM phosphate buffer at pH 7.6 and incubated at 4 C for 24 h. The Cy3 dye labeling efficiency was determined by UV-vis spectroscopy and typically found to be around 70%. In fluorescence assays, labeled S80C subunits were mixed with wild-type subunits such that 15% of the subunits were functionalized. This fraction of labeling was found to give adequate fluorescent signal while not disrupting the tertiary assembly of the protein. This was determined by passing the labeled protein through a Superdex 200 10/300 size exclusion column and quantifying the labeled protein in the 24-meric (eluting at 11 mL) and 2-meric (eluting at 16 mL) states. Evaluation of Gold Particle Number Density. The concentration of gold in solution was measured by ICP-OES (Galbraith Laboratories Inc., TN). The distribution of gold particle sizes was measured by transmission electron microscopy (TEM; analysis performed with ImageJ, version 1.41 h, National Institutes of Health). The average number of gold atoms in a particle, Naverage, was evaluated from the Gaussian fit to the distribution of particle diameters: Naverage ¼
ðTotal volume AuÞ ¼ Vatom ðnumber of particlesÞ 2 R0 ffiffi dx π ¥ x3 exp - xp-μ 2σ ð1Þ 2 R0 ffiffi dx 6Vatom ¥ exp - xp-μ 2σ
3
where Vatom is the volume of a gold atom, 0.0169 nm . The Gaussian distribution was characterized by mean, μ (10.39 nm), and standard deviation, σ (0.78 nm). The number density of particles in the protein/metal conjugation preparations was thus calculated to be (3.8 ( 0.1) 1012 particles/mL. Protein/Metal Conjugation. Protein subunit dimers (60 μL, 20 mM NaCl, pH 7.6 phosphate buffer) were mixed with 10 nm diameter capped gold particles (1140 μL in water), typically in a ratio of 24 four-helix bundle subunits per gold particle. The mixture was annealed with a thermal cycling program comprising 50 cycles of 2.5 min periods at 55 C followed by 12 min periods at 4 C, with a ramp rate of 0.3 C/s (12.5 h total cycle time, Eppendorf Mastercycler PCR machine). Subunit Titration Assays. Protein four-helix bundle dimers were titrated into solutions containing known concentrations of gold and incubated for varying lengths of time or through a series of temperature cycles (1 mM NaCl, pH 6.4). Changes to the position of the surface plasmon absorbance maximum were recorded with a Varian Cary 100 Bio spectrophotometer. The titration was repeated with protein in which 15% of the subunits had been labeled with Cy3 dye via a cysteine mutation to the exterior S80 position. Fluorescence was monitored with a Varian Eclipse fluorimeter (excitation at 550 nm, fluorescence recorded at 570 nm). Characterization of Conjugates. Time-resolved salt aggregation experiments were performed with an Agilent 8453 spectrophotometer fitted with a magnetic stirring system. TEM images were recorded on a JEOL 1010 electron microscope operating at 80 kV; samples were negatively stained Langmuir 2009, 25(9), 5219–5225
with uranyl acetate. Image analysis was performed using ImageJ (version 1.41 h, National Institutes of Health). Exchange of Surface Ligands on Gold Particles. Ten nanometer diameter citrate-capped gold particles (10 mL, approximately 5.7 1012 particles/mL) were stirred for 24 h under argon with an excess of the new capping agent (typically a 1000:1 ratio of the new capping agent to the original citrate). New capping agents were BSPP, MSA, DMAP, and GSH. The gold particles were precipitated by centrifugation (GSH) or by adding sodium chloride solution (4 M) until the color changed from red to purple/black, followed by centrifugation (BSPP, MSA, and DMAP). The gold pellets were then resuspended in water. Calculation of Surface Potential. The electrostatic potential of the protein was mapped with APBS27 and visualized using PyMol (DeLano Scientific).28 The protein structure6 was prepared for analysis using PDB code 2PQR29 with protonation states calculated for pH 6.4 using PROPKA.30
3.
Results and Discussion
The assembly of A. fulgidus ferritin is known to be salt mediated: at low salt concentrations, the protein exists as dimers of four-helix bundles. As the salt concentration is increased to approximately 600 mM, the protein is almost entirely assembled into 24-mers.6 The ratio of dimers to 24-mers was measured at various intermediate salt concentrations using a size exclusion column (Figure 1; pages S2 and S3 in the Supporting Information). No intermediate assembly states were observed. The data were fit to the Hill equation:
f24mer ¼
c1=2 1þ c
n ! -1
ð2Þ
where the fraction of protein in the 24-meric state is a function of the concentration of NaCl, c. The halfway point of assembly occurred at c1/2 = 200 mM. A Hill coefficient value of n = 3.1 indicated that the assembly process was highly cooperative. CD spectroscopy showed that the secondary structure of the protein was independent of the assembly state (page S4 in the Supporting Information). In addition, DLS measurements were consistent with 24-mer and dimer formation (pages S5-S7 in the Supporting Information). Mixing the protein subunit dimers (60 μL, 20 mM NaCl, pH 7.6 phosphate buffer) with citrate-capped gold nanoparticles (1140 μL in water, pH 6.0, 10 nm diameter particles purchased from British BioCell International, Figure 2A) resulted in the assembly of the ferritin around the gold particles at low salt concentrations (1 mM NaCl, pH 6.4, Figure 2B). The yield of incorporation, defined in the literature as the proportion of structures that contain a nanoparticle,22 is almost 100% as the protein self-assembles (at low salt concentrations) only in the presence of the particles (see pages (27) Baker, N. A.; Sept, D.; Joseph, S.; Holst, M. J.; McCammon, J. A. Proc. Natl. Acad. Sci. U.S.A. 2001, 98, 10037–10041. Bank, R. E.; Holst, M. SIAM Rev. 2003, 45, 291–323. Holst, M. Adv. Comput. Math. 2001, 15, 139– 191. (28) DeLano, W. L. PyMOL; DeLano Scientific LLC: San Carlos, CA, 2008.Lerner, M. G.; Carison, H. A. APBS plug-in for PyMOL; University of Michigan: Ann Arbor, MI, 1996. (29) Dolinsky, T. J.; Nielsen, J. E.; McCammon, J. A.; Baker, N. A. Nucleic Acids Res. 2004, 32, 665–667. (30) Bas, D. C.; Rogers, D. M.; Jensen, J. H. Proteins 2008, 73, 765–783. Czodrowski, P.; Dramburg, I.; Sotriffer, C. A.; Klebe, G. Proteins 2006, 65, 424–437. Li, H.; Robertson, A. D.; Jensen, J. H. Proteins 2005, 61, 704–721.
DOI: 10.1021/la8040743
5221
Article
Swift et al.
Figure 1. Salt-mediated assembly of the ferritin from the hyperthermophile Archaeoglobus fulgidus. The protein exists as subunit dimers at low salt concentrations, assembling into a 24-mer as the salt concentration is increased.6 Relative concentrations of dimer and holoprotein were evaluated by integrating the absorbance as the equilibrated protein was run through a sizing column (20 mM phosphate buffer, pH 7.6); no other assembly intermediates were observed (page S2 in the Supporting Information). Fitting the data to the Hill equation (red trace) gives a Hill coefficient of 3.1, suggesting a cooperative assembly process.
Figure 2. Transmission electron micrographs of (A) ∼10 nm citrate-capped gold particles and (B) particles coated with hyperthermophile ferritin at a 1:1 24-mer to particle ratio showing complete and uniform protein coverage (the protein can be seen as a white halo surrounding the gold particles against a stain of uranyl acetate). (C) The maximum in the surface plasmon absorption of the citratecapped gold nanoparticles (black line) was shifted to longer wavelengths following encapsulation in protein (red line). No increase of absorption due to scattering was apparent in the protein-conjugated samples, suggesting that no additional aggregation occurred. S8-S12 in the Supporting Information for additional TEM images and numerical analysis). Our data suggest that the most efficient particle surface coverage is achieved when the samples are subjected to a series of temperature cycles. When the concentration of protein 24-mers is matched to the number density of nanoparticles in solution, in excess of 95% of the particles appear to be capped with protein. In 5222
DOI: 10.1021/la8040743
contrast, work by Chen et al. showed that the assembly of citrate-capped gold particles within protein cages from the brome mosaic virus gave low yields of incorporation. Part of the reason for this difference was that the assembly conditions caused aggregation of the gold particles and formation of empty cages; the problem was addressed by using gold particles functionalized with a carboxylate-terminated thioalkylated tetraethylene glycol.22 In the ferritin system, the protein assembles around the negatively charged,31 citrate-capped gold particles without the need for surface modification or destabilizing assembly conditions. The maximum in the surface plasmon absorption was found to shift to longer wavelengths following conjugation to the hyperthermophile protein (from approximately 518 to 521 nm, Figure 2C). However, there was no increase in the baseline of the UV-vis spectrum, indicating that the presence of protein did not cause the formation of large aggregates capable of scattering light. The shift in the absorption maximum can be explained by considering a change in the environment surrounding the gold particle. The theory describing interaction between light and small particles can provide some insight: modeling the effect of a 2 nm thick coating of refractive index in the range 1.5-1.6 (the experimentally determined range of refractive indices for proteins deposited on a flat surface)32 on the calculated absorbance spectrum of a gold particle resulted in a shift of the surface plasmon absorption maximum to longer wavelengths (see pages S13-S18 in the Supporting Information).33 It is difficult to achieve exact agreement between theory and experiment in calculations of plasmon absorption for a number of reasons: our particles are only roughly spherical, have rough edges, and likely contain multiple crystal domains; the complex refractive indices of metals tabulated in the literature were determined under a range of different conditions.34 Additionally, the presence of a protein coat may alter the extent of charge “spill-out” from the metal surface, typically parametrized in a correction to the bulk collisional frequency and resulting in peak broadening of the surface plasmon resonance band.35 The number density of particles in solution was calculated from the concentration of gold in solution, measured by inductively coupled plasma-optical emission spectroscopy (ICP-OES, Galbraith Laboratories Inc., TN), and the distribution of gold particle sizes, measured by TEM (Figure 3A). In order to quantify the number of ferritin subunits binding to each gold particle, protein was titrated into solutions containing known concentrations of gold and incubated for varying lengths of time or through a series of temperature cycles (1 mM NaCl, pH 6.4). The optimum thermal cycling program comprised of 50 cycles of 2.5 min periods at 55 C followed by 12 min periods at 4 C, with a ramp rate of 0.3 C/s (12.5 h total cycle time, Eppendorf Mastercycler PCR machine). Changes to the position of the surface plasmon absorbance (31) Brewer, S. H.; Glomm, W. R.; Johnson, M. C.; Knag, M. K.; Franzen, S. Langmuir 2005, 21, 9303–9307. (32) Voros, J. Biophys. J. 2004, 87, 553–561. (33) Bohren, C. F.; Huffman, D. R. Absorption and Scattering of Light by Small Particles; Wiley-InterScience: New York, 1983; pp 82-103, 181-183. Jain, P. K.; Lee, K. S.; El-Sayed, I. H.; El-Sayed, M. A. J. Phys. Chem. B 2006, 110, 7238–7248. (34) Johnson, P. B.; Christy, R. W. Phys. Rev. B 1972, 6, 4370–4379.Weaver, J. H., Krafka, C., Lynch, D., Koch, E., Eds. Optical Properties of Metals, Physik Daten 18-2; Fachinformationzentrum: Kahlsruhe, 1981; Vol. 2, pp 59-64. (35) Dalacu, D.; Martinu, L. J. Opt. Soc. Am. B 2001, 18, 85–92.
Langmuir 2009, 25(9), 5219–5225
Swift et al.
Article
Figure 3. (A) Distribution of gold core sizes, measured from a population of 317 particles imaged by TEM. The red line shows a Gaussian fit to the distribution, characterized by a mean of 10.39 nm and a standard deviation of 0.78 nm. (B) Position of the surface plasmon absorbance maximum of the gold particles at increasing concentrations of protein subunits. The samples were annealed through 50 cycles between 4 and 55 C before measurement. There are two distinct regions: the position of the maximum shifts to longer wavelengths as the protein concentration is increased to 1.0 ( 0.3 24-mer/gold particle (blue line); higher protein concentration has little effect on the plasmon resonance absorption (red line). We attribute the initial shift in absorption maximum to the change in environment around the particles as subunits adsorb to the gold surface. When maximum surface coverage is attained, subsequent protein addition does not affect the absorption. Cartoon shows protein in blue and gold in yellow. (C) Quenching of fluorescence by the gold particles at increasing concentrations of Cy3-labeled subunits. The subunits of a S80C ferritin mutant were Cys-labeled with Cy3 fluorescent dye (inset). Fluorescence before annealing is shown 60 and 1800 s after mixing. Following 50 cycles between 4 and 55 C, two regions can be clearly differentiated: Below 0.9 ( 0.1 24-mer/gold particle, the protein is adsorbed to the gold surface and the fluorescence is quenched (blue line). Beyond the point of maximal coverage of the gold surface, additional protein is free in solution and so is not quenched (red line). maximum were recorded (Figure 3B). The titration was repeated with protein in which 15% of the subunits had been labeled with Cy3 dye via a cysteine mutation to the exterior S80 position. In this experiment, quenching of the fluorescence at 570 nm was monitored (Figure 3C). Following temperature cycling, both plots showed two distinct linear regions, with the crossover between regions occurring at 1.0 ( 0.3 and 0.9 ( 0.1 24-mers/gold particle, respectively. The first region corresponded to an increasing coverage of protein on the surface of the gold particles: the surface plasmon absorbance of the gold shifted as the environment surrounding the particle changed; the fluorescence of the dye was quenched, as it was in close proximity to the gold surface. The second region occurred after the native stoichiometry of protein subunits had formed a monolayer on the surface of each gold particle. Protein added in excess of the quantity required for optimum surface coverage remained free in solution. In the fluorescence assay, the free Cy3-labeled subunits were able to fluoresce. In the surface plasmon assay, the low concentrations of gold and protein meant that the position of the absorbance maximum was unaffected by the addition of further protein. As the incubation period between gold and protein was increased, the fluorescence quenching tended toward the two-regime behavior observed for the heat-cycle sample (Figure 3C). The process of subunit dimers binding to the gold surface appears to be largely complete within a time scale on the order of minutes. However, as optimal coverage is approached, we hypothesize that there is an entropic barrier to forming a monolayer (in addition to the retarding effect of a reduced concentration of unbound protein). Annealing the sample allows rearrangement of protein on the gold surface, increasing the rate at which the energy-minima protein packing is achieved. The sharp transition between the two linear regions could not be achieved without annealing, even with incubation at 4 C for 1 week. The gold particles are slightly Langmuir 2009, 25(9), 5219–5225
larger than the cavity of the protein, as determined from the crystal structure (8.52 nm diameter, PDB code 1s3q),6 so in order to maintain the native stoichiometry of one 24-mer per particle some accommodation is required. The stability of the gold particles with respect to saltinduced aggregation is a convenient measure of surface passivation that has relevance to any eventual application under physiological conditions. Solutions of protein-coated gold nanoparticles, prepared in 1 mM NaCl solution as described previously, were rapidly mixed with a saline buffer such that the final salt concentration was 330 mM. No change in the surface plasmon resonance absorption maximum was observed over the course of 10 min. In contrast, the same assay performed in the absence of protein resulted in a rapid redshift in the position of the surface plasmon maximum, indicative of particle aggregation (Figure 4). Incubation with insufficient protein for monolayer formation (one 24-mer for every two gold particles) slowed but did not prevent aggregation. Failure to provide sufficient time for protein-gold assembly also resulted in aggregation. Over the course of 10 min, no difference was observed between protein-coated gold samples that had been subjected to heat cycling compared with those that had been incubated at 4 C for the same length of time. Samples of Cy3-labeled protein and gold in 330 mM NaCl were incubated for 1 day then centrifuged at 16 000g for 1 h. UV-vis spectroscopy showed that this process caused the precipitation of the colloidal gold. Analysis of the supernatant showed that fluorescence was not recovered by removing the gold: the protein precipitated with the gold; salt did not displace the subunits. One of the features of the hyperthermophile ferritin is the presence of four large (4.5 nm), triangular pores in the 3-fold axes.6 It is possible that some salt-induced aggregation occurs by gold-gold interactions through these pores without disassembling the ferritin 24-mer. DOI: 10.1021/la8040743
5223
Article
Swift et al.
Figure 4. Rate of aggregation of particles at high salt concentrations. The concentration of NaCl was rapidly increased from 1 to 330 mM by addition of a high-salt buffer. The position of the SPR absorption maxima was monitored by taking the ratio of absorbances at 650 and 520 nm. Particles not associated with protein quickly aggregated at elevated salt concentrations, resulting in a shift of the resonant surface plasmon absorbance to longer wavelengths (red line). The protein-coated particles (annealed with 50 cycles between 4 and 55 C) remained stable, suggesting effective passivation of the gold surface (solid black line). Incomplete protein coating (dotted black line) and insufficient incubation time (blue line) resulted in intermediate rates of aggregation. Error bars show the standard deviation from four trials; curves were fit to a double-exponential function. It is often practical to replace the citrate groups that cap the gold surface with another ligand with favorable properties. Mixing commercially purchased citrate-capped gold particles with an excess of bis(p-sulfonatophenyl)phenylphosphine (BSPP) results in rapid ligand exchange. Unlike citratecapped particles, BSPP-capped gold can be reversibly precipitated from solution at high salt concentrations, thus simplifying processes of purification and concentration.36 The protein-coated, BSPP-capped gold can be rapidly separated from unwanted impurities (excess BSPP and citrate, unbound protein) using size exclusion chromatography (Figure 5). BSPP-capped gold particles were preferable to citrate-capped particles due to their increased robustness; nonetheless, most of the protein-free, BSPP-capped gold aggregated irreversibly in the column during purification. Protein conjugated to gold particles eluted from the column at lower volumes than the apoprotein, consistent with an overall increase in external dimensions. Previous reports of the gel filtration chromatography of CdTe nanoparticles conjugated to bovine serum albumin (BSA) showed a similar effect.37 In order to understand better the nature of the binding and assembly processes, the electrostatic potential of the protein was mapped with APBS27 and visualized using PyMol (Figure 6).28 Calculations using PROPKA30 indicated that both the interior surface of the protein and the edges of the large triangular pores were predominantly negatively charged at pH 6.4. The effect of the capping agent on the efficiency of protein coating was investigated by displacing the citrate ligands with (36) Claridge, S. A.; Liang, H. W.; Basu, S. R.; Frechet, J. M. J.; Alivisatos, A. P. Nano Lett. 2008, 8, 1202–1206. Loweth, C. J.; Caldwell, W. B.; Peng, X. G.; Alivisatos, A. P.; Schultz, P. G. Angew. Chem., Int. Ed. 1999, 38, 1808–1812. (37) Wang, S. P.; Mamedova, N.; Kotov, N. A.; Chen, W.; Studer, J. Nano Lett. 2002, 2, 817–822.
5224
DOI: 10.1021/la8040743
Figure 5. Fast protein liquid chromatography (FPLC) purification of ferritin-coated gold particles. BSPP-capped gold particles (2.5 1013 particles/mL) were thermocycled (50 cycles between 4 and 55 C) with and without the presence of A. fulgidus ferritin. Following incubation, the samples were run through a Superdex S200 10/300 GL size exclusion column equilibrated with:: 1 mM NaCl buffer (pH 6.4, 0.5 mL/min flow rate; column and Akta FPLC system, GE Healthcare). Most of the uncoated gold particles aggregated in the column, although a small amount of gold eluted at 8.34 mL, the dead volume of the column (red trace). The protein-coated gold particles eluted at 8.87 mL (black trace, UV-vis spectrum of the peak shown as inset). For comparison, the assembled apoprotein eluted at 11.15 mL (blue trace). Excess BSPP and unbound protein subunits eluted at larger volumes.
Figure 6. Surface potential diagram of a hexameric section of the hyperthermophile protein, showing the interior surface of the cavity viewed along a 3-fold axis. Evaluated at pH 6.4, consistent with the solution during the gold binding procedure, the interior surface is predominantly negatively charged. The same hexameric section is also projected as part of the assembled 24-mer. Regions of subunitsubunit interface have a strong positive surface potential, suggesting the importance of salt-bridge formation in the process of 24-mer assembly. The edges of the triangular pores (the black dotted line outlines one of the pores) are weakly negatively charged (protein dielectric, 4; solvent dielectric, 80; temperature, 298 K). BSPP, DMAP, MSA, or GSH. Citrate and BSPP are both known to impart a negative charge on the metal surface,31,38 whereas DMAP leads to a positive surface charge.39 MSA and GSH both bind to the metal surface via a strong sulfur-gold interaction, with the steric bulk of GSH leading to a lower ligand density. Negatively stained TEM images of the capped gold particles coated with hyperthermophile ferritin (using the same annealing procedure as described previously) showed (38) Schmid, G.; Lehnert, A. Angew. Chem., Int. Ed. 1989, 28, 780–781. (39) Gandubert, V. J.; Lennox, R. B. Langmuir 2005, 21, 6532–6539. Vivek, J. P.; Burgess, I. J. J. Phys. Chem. C 2008, 112, 2872–2880.
Langmuir 2009, 25(9), 5219–5225
Swift et al.
Article Table 1. Summary of Data from TEM Images of Protein-Encapsulated Gold Particles with Different Capping Agentsa
capping agent citrate BSPP DMAP MSA GSH a
no. of images
total no. of particles
no. of naked particles
% naked particles
no. of empty shells
% empty shells
2 3 4 4 7
1627 561 464 378 303
70 32 23 30 10
4.3 5.7 5.0 7.9 3.3
7 15 2 1 2
0.5 2.7 0.4 0.3 0.7
See Figure 7 for sample TEM data and pages S8-S12 of the Supporting Information for a description of the analysis procedure.
ligand solvation versus that of ferritin binding.41 In support of this, the interaction between BSA (which also has a negative surface potential) is known to favor nonspecific binding to gold surfaces protected by monolayers in the following order: 31 hydrophobic > COO- > NH+ 3 > OH > ethylene glycol. In other words, the surface with the highest entropic cost to solvate is the most favored for protein binding. The fact that protein binding still occurs despite the increased strength of the MSA-gold and GSH-gold interaction would suggest that direct ligand displacement is less likely. There are possible parallels between the nature of the gold/protein assembly process and ferritin’s natural function: the presence of mineralized iron in the cavity of the protein, like the gold particle, stabilizes the 24-mer with respect to dimerization at low salt concentrations.6
Figure 7. Representative TEM images showing ∼10 nm gold particles, capped with ligands other than citrate, encapsulated in ferritin. Gold particles appear as black circles enclosed in protein, which appears as a white halo against an uranyl acetate negative stain. The nature of the ligand, including the charge it imparts on the particle, seems to have relatively little effect on the efficiency of assembly. efficient encapsulation in all cases (Figure 7). Numerical analysis of the images indicated that the fraction of enclosed gold particles was independent of the capping agent (Table 1). A slightly greater number of empty protein shells was observed in the BSPP-capped case, possibly due to a residual increase in ionic strength driving some 24-mer formation. We consider two mechanisms for assembly: the protein acts as a ligand directly to the gold surface, displacing the groups that cap the gold surface, or the interaction is electrostatic. The apparent lack of discretion for the surface charge of the gold particle, despite the negative potential of the cavity, indicates either ligand displacement or mediation by accommodating ionic interactions. Reports suggest that the surfaces of nanoparticles are accompanied by ionic double layers that could be capable of some degree of electrostatic shielding.40 The presence of the particle would increase the local concentration of salt ions, driving assembly by a similar mechanism to the process observed at elevated concentrations of [NaCl].6 There may be other factors to consider beyond the enthalpic strength of the interaction, such as the entropy change of (40) Laaksonen, T.; Ahonen, P.; Johans, C.; Kontturi, K. ChemPhysChem 2006, 7, 2143–2149.
Langmuir 2009, 25(9), 5219–5225
4. Conclusion In summary, we have shown that the ferritin from the archaeon Archaeoglobus fulgidus can be assembled around a gold core, in its native stoichiometry, by a mechanism distinct from conventional methods of in situ reduction. In contrast to much of the current literature, we have effectively utilized the entire volume of the ferritin cavity in a clean and high-yielding process. The prospect of replacing the gold cores with other metallic or semiconducting particles suggests the possibility of a modular component for self-assembling systems. This work is a step toward exploiting ferritin as an interface between inorganic materials and engineered biological systems. Acknowledgment. The authors thank Eric Johnson for providing the Archaeoglobus fulgidus ferritin gene in the pAF0834 plasmid, Raymond Meade at the University of Pennsylvania Biomedical Imaging Facility for TEM, and the laboratories of Jeffery Saven, Roderic Eckenhoff, and Jonas Johansson for use of instruments. This work was supported by the NSF (MRSEC DMR-0520020 and CAREER CHE0548188). Supporting Information Available: Chromatography traces and column calibration, circular dichroism spectra and dynamic light scattering data, example analyses of TEM images, and surface plasmon resonance calculations. This material is available free of charge via the Internet at http:// pubs.acs.org. (41) Gao, D. J.; Tian, Y.; Bi, S. Y.; Chen, Y. H.; Yu, A. M.; Zhang, H. Q. Spectrochim. Acta, Part A 2005, 62, 1203–1208.
DOI: 10.1021/la8040743
5225