Water-Dispersible Superparamagnetic Microspheres Adorned with

ARTICLE pubs.acs.org/Biomac

Water-Dispersible Superparamagnetic Microspheres Adorned with Two Types of Surface Chains Zhihan Zhou, Guojun Liu,* and Liangzhi Hong Department of Chemistry, Queen’s University, 90 Bader Lane, Kingston, Ontario, Canada K7P 3E5

bS Supporting Information ABSTRACT: Water-dispersible superparamagnetic polymer/γ-Fe2O3 composite microspheres adorned with two types of surface polymer chains are prepared and characterized. To prepare these spheres, we first synthesize uniform γ-Fe2O3 nanoparticles that are covered by poly(2-cinnamoyloxyethyl methacrylate)-blockpoly(acrylic acid) (PCEMA-b-PAA). These nanoparticles are then mixed with a PCEMA homopolymer in CHCl3 to form an oil phase. The oil phase is dispersed into water under vigorous stirring with the help of two diblock copolymer surfactants, PGMA-b-PCEMA and PSGMA-b-PCEMA. Here PGMA and PSGMA denote poly(glyceryl monomethacrylate) and succinated PGMA, respectively. Solid microspheres with cores composed of PCEMA and PCEMA-b-PAA-covered γ-Fe2O3 nanoparticles are obtained after CHCl3 evaporation and PCEMA photo-cross-linking. Under certain conditions, the coronal PGMA and PSGMA chains become segregated, thus producing surface bumps, ridges, and valleys. The PSGMA chains preferentially cover the protruding regions. The PSGMA carboxyl groups are used to immobilize bovine serum albumin (BSA). The immobilized BSA retains its activity and binds with anti-BSA. These spheres should be useful in immunoassays.

I. INTRODUCTION Immunoassays allow the quantification of a given species in a biological fluid using the specific binding between the species and a ligand.1 Here the biological fluid can be blood serum or urine, and the binding can be between an antigen and antibody, a heptan and antibody, or the complementary strands of DNA or RNA. Immunoassays do not rely on expensive instruments and are therefore cost-effective. When automated, they can process samples quickly. Because of these, immunoassays currently play a leading role in clinical laboratories. The rapid results generated from automated immunoassays help guide the early intervention in clinical situations. Heterogeneous immunoassays, involving the use of superparamagnetic beads or spheres,2 are one effective type of automated immunoassay. In these assays, competitive and sandwichtype (immunometric) methods1 are normally used to quantify analytes. Of these methods, the sandwich-type assays are used mainly to analyze antibodies and antigens. To analyze an antigen in a biological fluid, two antibodies are required. The “capturing” antibody is immobilized onto the surfaces of magnetic spheres, and the “signal” antibody bears light-absorbing, fluorescent, radioactive, or chemiluminescent groups or an enzyme. These two antibodies are mixed with a biological fluid to allow sandwich complex formation between the antigen and the two antibodies. A magnet placed next to the vial containing this sample draws the immobilized antigen to the side that is closest to the magnet. The supernatant containing many interfering species is then discarded, and the solid is rinsed with an aqueous buffer and redispersed in a buffer for antigen quantification by reading the r 2011 American Chemical Society

signal produced by the signaling antibody. The signal can be an absorbance, radioactivity, fluorescence, or chemiluminescence signal3 or the demonstrated activity of the enzyme label.4 Sandwich assays impart the best results if the labeled antibody binds specifically with the antigen and negligibly with interfering species. This is achieved only under well-controlled conditions because biomolecules do adsorb significantly onto most tested surfaces, with only a few exceptions.5-7 For example, lyzozyme adsorbs to negatively charged surfaces at pH 7.4 because this protein is positively charged at this pH.5 Fibrinogen bears negative charges at pH 7.4 and deposits onto positively charged surfaces.5 If the signal antibody binds nonspecifically with the magnetic spheres, one normally corrects for this by subtracting a control signal from the signal produced by the sample. The control is prepared under otherwise identical conditions but without the antigen present. Evidently, this signal subtraction process deteriorates data accuracy. This is especially true at low antigen concentrations when the background signal dominates. Therefore, nonspecific binding narrows the detection range of heterogeneous immunoassays and is highly undesirable. Nonspecific binding can, in principle, be reduced by using magnetic spheres that bear two types of surface chains. One type can include a neutral species, such as poly(ethylene glycol) (PEG), to which proteins adsorb only minimally.5 The other type can bear carboxyl or amino groups, which are charged Received: November 29, 2010 Revised: January 1, 2011 Published: February 04, 2011 813

dx.doi.org/10.1021/bm101432e | Biomacromolecules 2011, 12, 813–823

Biomacromolecules

ARTICLE

depending on the pH of the solution but are necessary for biomolecule immobilization. Despite this, most magnetic microspheres reported in the literature bear one type of surface chain. Commercial Seradyn8-10 and Dynal11-13 particles bear poly(acrylic acid) (PAA) surface chains. We also reported a method for preparing magnetic microspheres bearing PAA surface chains.14 For a review on the preparation of water-dispersible microspheres bearing one type of surface chain, readers should consult ref 14. We report in this Article the preparation of magnetic microspheres adorned with two types of surface chains, which represent our success in the first step toward preparing microspheres that may potentially have reduced nonspecific binding. Of the chains, one type bears carboxyl groups, and the other chain is neutral and bears hydroxyl groups. We further show that the carboxyl-bearing chains can be used to immobilize bovine serum albumin (BSA) and that the immobilized BSA retains its activity and binds with its antibody.

Scheme 1. Structures for PCEMA, PAA, PGMA, and PSGMA

The polymers used were carefully characterized by 1H NMR in CDCl3 and a size-exclusion chromatograph (SEC) equipped with a Dawn Helios II light scattering detector and Optilab rEX refractive index detector. The eluent used for SEC analysis was THF. The refractive indices of the polymer solutions were measured using the Optilab rEX refractive index detector. The specific refractive index increments (dnr/ dc) of the samples were determined from the slope of the straight line obtained by plotting polymer solution refractive index nr against concentration. To ensure the solubility of all of the blocks in CDCl3 and THF, the polymers were characterized in the forms of PCEMA55-bPSMA460, PCEMA65-b-PSMA650, PCEMA320, and PCEMA30-b-PtBA4. The characterization results are summarized in Table 1. All of the polymers used had low polydispersities, as we expected. Successful derivatization of these polymers was confirmed by 1H NMR analysis of these samples in pyridine-d5. γ-Fe2O3 Nanoparticles. The preparation of γ-Fe2O3 nanoparticles involved the thermal decomposition of iron(III) oleate (Fe(Oleate)3).19 To prepare Fe(Oleate)3, we first charged 1.44 g of FeCl3 and 8.20 g of sodium oleate in a 250 mL round-bottomed flask. Water (8.0 mL), methanol (11 mL), and hexane (19 mL) were then added. The salts dissolved immediately, and the color of the hexane layer became dark red, indicating the formation of iron(III) oleate. The mixture was heated to 70 °C for 3 h before it was cooled to room temperature and washed thrice with 50 mL of water. The hexane layer was collected and dried over anhydrous magnesium sulfate overnight. Hexane was removed from the filtered solution by rotorary evaporation. The iron(III) oleate was vacuum-dried initially at 70 °C for 3 h and then at 25 °C for 20 h to give 7.49 g of iron(III) oleate as a dark brown viscous liquid with a 94% yield. To prepare γ-Fe2O3 nanoparticles, 3.0 g of iron(III) oleate was mixed with 12.0 mL of 1-octadecene in a 100 mL flask. The temperature of the mixture was increased at 3.3 °C/min to 320 °C, and the color of the mixture turned black. The mixture was annealed at this temperature for 60 min. After the reaction mixture cooled to room temperature, 3.0 mL of the reaction mixture was withdrawn. To the rest of the mixture was added, under nitrogen flow, 0.50 g of trimethylamine oxide. As the mixture was heated to 180 at 5 °C/min, its color gradually turned from black to brown. The temperature was maintained at 180 °C for 2 h before the mixture was cooled to room temperature. To purify the iron oxide nanoparticles, we mixed 4.0 mL of the nanoparticle dispersion with 16 mL of ethanol. The precipitate was captured by a 0.47 T magnet, and the supernatant was decanted. To the precipitate was added 4.0 mL of hexane to redisperse the particles. These particles were precipitated again with the addition of 10 mL of acetone. This process was repeated thrice to yield 120 mg of γ-Fe2O3 nanoparticles. Ligand Exchange. The γ-Fe2O3 nanoparticles (120 mg) were redispersed in 4.0 mL of chloroform. To the dispersion was added

II. EXPERIMENTAL SECTION Materials. Cinnamoyl chloride (98%), iron(III) chloride (FeCl3, reagent grade), calcium hydride (95%), magnesium sulfate (MgSO4, anhydrous, reagent grade), sodium chloride (reagent grade), sodium hydroxide (95%), N-ethyl-N0 -(3-dimethylaminopropyl)carbodiimide hydrochloride (EDCI, commercial grade), and N-hydroxysuccinimide (NHS, 97%) were purchased from Sigma-Aldrich and were used as received. Phosphate-buffered saline (PBS) tablets were purchased from Sigma-Aldrich. Dissolving one of these tablets in 200 mL of deionized water yielded a PBS pH 7.4 solution containing 10 g/L of NaCl, 2.5 mg/L of KCl, 1.8 g/L of Na2HPO4, and 0.3 g/L of KH2PO4. ACS grade chloroform, methanol, hexane, diethyl ether, pyridine, and tetrahydrofuran (THF) as well as hydrochloric acid (36.5-38% or 12 M) were purchased from Fisher Scientific. Pyridine was refluxed over calcium hydride and freshly distilled before use. ACS grade acetone was purchased from Caledon Laboratories. Sodium oleate and 1-octadecene (90%) were purchased from TCI and were used as received. Dialysis tubing (Spectra/Por, 12000 cut off and 3500 cut off) was purchased from VWR. BSA (fatty acid free) was supplied by Boval Company (Cleburme, Texas). A micro BCA protein assay kit (Thermo Scientific) was purchased from Fisher Scientific. This kit contained three reagents: reagent A consisted of sodium carbonate and sodium bicarbonate in 0.2 N NaOH; reagent B was bicinchoninic acid (4.0 wt %) in water; reagent C was 4.0 wt % cupric sulfate pentahydrate in water. Sheep anti-BSA labeled by fluorescein isothiocyanate (molecular weight ∼150 kDa) at a concentration of 1.0 mg/mL in PBS was purchased from Immunology Consultants Laboratory. Polymers. Three diblock copolymers and one homopolymer, poly(2-cinnamoyloxytheyl methacrylate) or PCEMA, were used to prepare the microspheres. These were PCEMA55-b-PSGMA460, PCEMA65-b-PGMA650, PCEMA320, and PCEMA30-b-PAA4, where the subscripts refer to the numbers of repeat units for the different blocks. PCEMA55-b-PSGMA460 and PCEMA65-b-PGMA650 were used as dispersants for the magnetic microspheres, PCEMA320 was used as the binder for the magnetic γ-Fe2O3 nanoparticles, and PCEMA30b-PAA4 was used to coat the γ-Fe2O3 nanoparticles. Scheme 1 shows the structures of the concerned polymer blocks. The precursors to these polymers were prepared by anionic polymerization. PCEMA was derived from the reaction between poly(2-hydroxyethyl methacrylate) (PHEMA) and cinnamoyl chloride. The precursor for PHEMA was poly(2-trimethylsiyloxyethyl methacrylate),15 that for PGMA was poly(solketal methacrylate) (PSMA),16 and that for PAA was poly(tert-butyl acrylate) (PtBA).17 PSGMA was obtained from the reaction between PGMA and excess succinic anhydride.18 814

dx.doi.org/10.1021/bm101432e |Biomacromolecules 2011, 12, 813–823

Biomacromolecules

ARTICLE

Table 1. Molecular Characteristics of the Polymers SEC results

a

polymer

dnr/dc (mL/g)

Mw

Mn

Mw/Mn

n/m from 1H NMRa

PCEMA55-b-PSMA460

0.096

108 000

106 000

1.02

1.0/8.3

PCEMA65-b-PSMA650

0.088

152 000

148 000

1.03

1.0/1.0

PCEMA320

0.162

84000

82000

1.02

PCEMA30-b-PtBA4

0.121

8300

8000

1.04

8.0/1.0

Ratio of the number of repeat units between the first and second blocks.

Scheme 2. Structure of the Cuþ-Bicinchoninic Acid Complex

240 mg of PCEMA30-b-PAA4. The mixture was then transferred to a dialysis tube with a molecular weight cutoff of 3500 g/mol and dialyzed against 50 mL of chloroform. This was done over 3 days, and the CHCl3 solution was changed six times. To remove the excess PCEMA30-b-PAA4 from the final mixture, 6.0 mL of diethyl ether was added to the dispersion to precipitate the nanoparticles. These nanoparticles were attracted to the side of the flask by a magnet, and the supernatant was removed by a glass pipet. The precipitate was redispersed in 4.0 mL of chloroform, and 4.8 mL of diethyl ether was then added to induce nanoparticle precipitation. The supernatant was subsequently removed by magnetic decantation again. This process was repeated four times before the nanoparticles were dried under vacuum to yield 97 mg of product.

Optimized Procedure for Magnetic Microsphere Preparation and Purification. To a glass vial were added 1.00 mL of a

was placed next to a 0.47 T magnet to capture the magnetic spheres, and the supernatant was decanted. The spheres were redispersed in 1.50 mL of PBS buffer. To the vial containing these spheres was then added 0.10 mL of a freshly prepared EDCI (2.0 mg) and NHS (3.0 mg) solution in PBS buffer. The spheres were stirred for 15 min to activate the carboxyl groups20 before the supernatant was removed by magnetic decantation. The spheres were then washed twice with 1.5 mL of the PBS buffer. To the magnetic spheres 1.0 mL of a BSA solution (10.0 mg/mL in PBS) was added. The mixture was left in a refrigerated room maintained at 4 °C and was stirred for 20 h to complete the protein coupling. The obtained magnetic spheres were then washed thrice with 1.5 mL of PBS buffer by magnetic decantation. The resulting sample was dispersed in 1.5 mL of PBS buffer. BSA Coupling Capacity. The BSA amount attached to the magnetic microspheres was determined using a standard protocol and a commercial assay kit.21 This kit contained bicinchoninic acid, Cu2þ, and a buffer solution. After the assay kit was mixed with BSA, some of the Cu2þ was reduced to Cuþ by reductant(s) present in the protein. The Cuþ then reacted with bicinchoninic acid to produce a purple 1:2 (metal:ligand) complex (Scheme 2) that had strong absorbance at 562 nm. The amount of Cuþ produced was determined spectrophotometrically and was proportional to the BSA amount present in the system. To determine accurately the concentration of BSA, we obtained a calibration curve by performing the assay on BSA solutions with known concentrations. This involved first the preparation of a bicinchoninic acid working solution from mixing the reagents A, B, and C provided via the commercial kit at a volume ratio of 25:24:1 (A:B:C). Several BSA solutions with concentrations at 40, 20, 10, 5, and 0 μg/mL were then prepared in PBS buffer. To 1.00 mL of each of the standard solutions, 1.00 mL of the bicinchoninic acid working solution was added. The mixtures were then heated in an oil bath at 60 °C for 1 h. This was followed by absorbance measurements at 562 nm at room temperature. Magnetic microspheres bearing BSA or lacking BSA were analyzed analogously. After the incubation of the magnetic solutions with the bicinchoninic acid working solution, the magnetic spheres were captured by placing the samples next to a 0.47 T magnet for 3 min, and the supernatant was then collected for absorbance analysis. The amount of

PCEMA55-b-PSGMA460 solution (5.0 mg/mL) in water, 0.50 mL of a PCEMA65-b-PGMA650 solution in water (10.0 mg/mL), 0.100 mL of a MgSO4 solution (20.0 mg/mL), and 0.08 mL of a NaOH solution (0.20 M). The resultant mixture was transferred to a 250 mL doublenecked round-bottomed flask that contained 11.0 mL of deionized water that was saturated with chloroform. The solution pH was adjusted to 7.0 with 0.10 M HCl and 0.20 M NaOH before the side neck of the flask was sealed with a septum. To the solution was added 0.15 mL of chloroform by a syringe to saturate the vapor phase. The PCEMA-b-PAA-coated γ-Fe2O3 nanoparticles were redispersed in CHCl3 at a concentration of 50 mg/mL, and 0.50 mL of this solution was mixed with 16.0 mg of a PCEMA320 homopolymer. The oil phase was then added dropwise by syringe over a time period of 2 min into the aqueous phase, which was mechanically stirred at 1200 rpm. The stirring was continued for another 30 min at room temperature before the septum was removed to allow the evaporation of chloroform. Five minutes later, the mixture was immersed in an oil bath. The emulsion was heated to 45 °C in 15 min and was left at this temperature for another 10 min. The temperature was further increased to 75 °C and maintained at this temperature for 30 min to remove the chloroform fully. To “lock in” the core structure of the spheres, the dispersion (about 8.0 mL) was transferred to a 25 mL photolysis cell. The sample was exposed to a UV beam for 3 h to cross-link the PCEMA. The light beam was produced by a 500 W mercury lamp in an Oriel 6140 lamp housing, powered by an Oriel 6128 power supply. Short-wavelength light was removed by passing the beam through a 270 nm cutoff filter. To remove the large particles from the microspheres, the cross-linked spheres were centrifuged at 500 rpm (25 g) for 2 min, and the precipitate was discarded. This procedure was repeated once again. To remove the smaller particles and the particles that were either less magnetized or nonmagnetized, the supernatant placed in glass vial was placed next to a 0.47 T magnet for 2 min to capture the large magnetic spheres, and the supernatant was discarded. This process was repeated thrice. The final sample was stored as a dispersion in deionized water at a concentration of 3.0 mg/mL. Bovine Serum Albumin Immobilization. A 1.50 mL sample of the above magnetic sphere dispersion, at a concentration of 3.0 mg/mL, 815

dx.doi.org/10.1021/bm101432e |Biomacromolecules 2011, 12, 813–823

Biomacromolecules

ARTICLE

coupled BSA was determined to be 22 μg of BSA per mg of spheres based on the calibration curve. Binding between Anti-BSA and Immobilized BSA. The fluoresceine-labeled anti-BSA was diluted to 10 μg/mL by PBS buffer before use. Magnetic decantation was used to remove the solvent from the magnetic microspheres (2.2 mg) which were either bearing BSA or lacking BSA. The magnetic spheres were then mixed with 2.00 mL of the anti-BSA solution, and the mixture was incubated at room temperature for 2 h. Subsequently, the supernatant was collected by magnetic decantation and its fluorescence intensity was measured at 518 nm. The excitation wavelength used was 494 nm. Analogously, the fluorescence intensity was measured for a control sample that was prepared under otherwise identical conditions, except that the control sample was not mixed with anti-BSA. The fluorescence intensity of an anti-BSA reference sample that was not equilibrated with magnetic microspheres was also measured. Techniques. Thermal gravimetric analysis (TGA) was performed using a TA-Q-500 TGA instrument. The sample was first heated under N2 atmosphere from 25 to 200 °C at a rate of 10 °C/min. The temperature was then increased from 200 to 500 °C at a rate of 5 °C/min. Finally, the sample was heated from 500 to 800 °C at a rate of 10 °C/min. Fourier transform infrared (FT-IR) analysis was performed using a Varian 1000 FT-IR instrument. The sample was first dispersed in chloroform. The concentration of the polymer sample was ∼5.0 mg/ mL, whereas the concentration of the nanoparticle sample was ∼15 mg/ mL. The sample dispersion (0.2 mL) was dropped onto a KBr crystal and was allowed to form a uniform layer on the crystal after the solvent was vaporized. The IR spectra were obtained in the transmission mode. Quantitative analysis was performed using a Varian 600-FT-IR instrument. The dried sample was mixed with 100 mg of KBr powder. The sample and the mixture were then ground together and transferred to a sample holder. The sample was analyzed in diffuse reflectance Fourier transform infrared (DRIFT-IR) mode. The absorbance of the sample was recorded using the Kubelka-Munk function. A series of PCEMA-bPAA samples with different polymer/KBr ratios was prepared and analyzed to obtain a calibration curve. The amount of PCEMA-b-PAA present in the ligand exchanged nanoparticles was calculated on the basis of this calibration curve. Transmission electron microscopy (TEM) images were obtained using a Hitachi H-7000 instrument with an accelerating voltage of 75 kV. The iron oxide samples were sprayed onto carbon-coated copper grids and were subsequently analyzed without further staining. The emulsion spheres were sprayed onto nitrocellulose-coated copper grids. To identify the polymer samples, we stained the PSGMA domains by mixing with uranyl acetate for 20 min. The sample was rinsed with water 10 times to remove the excess uranyl acetate. The X-ray diffraction (XRD) analysis of the iron oxide particles was performed using a Philips X’Pert Pro MPD diffractometer using Co KR (λ = 1.7890 Å) radiation.

Figure 1. TEM images of (a) oleic-acid-coated and (b) PCEMA-bPAA-coated γ-Fe2O3 nanoparticles.

Figure 2. X-ray diffraction pattern of oleic-acid-coated γ-Fe2O3 12.0 nm particles.

Oleic-Acid-Coated γ-Fe2O3 Nanoparticles. Oleic-acid-coated γ-Fe2O3 nanoparticles were prepared using a method based on that of Hyeon and coworkers.19 This method involved the hightemperature iron(III) oleate decomposition into iron oxide nanoparticles. According to Hyeon and coworkers,19 these particles should contain only maghemite (γ-Fe2O3) and magnetite (Fe3O4) phases. A later study by Bronstein and coworkers22 revealed that the particles also contained a w€ustite nonmagnetic phase if the particles were prepared under conditions deprived of oxidants such as oxygen. To annihilate the w€ustite phase, we reacted the crude iron oxide nanoparticles with a mild oxidant, trimethylamine oxide, to oxidize the w€ustite and magnetite phases into maghemite.23 Figure 1a shows a TEM image of iron oxide nanoparticles prepared using this method. The particles are uniformly sized. The average TEM diameter was 12.0 ( 0.9 nm. Figure 2 shows an XRD pattern for the prepared iron oxide nanoparticles. The peak positions agree with those reported by Hyeon and coworkers for γ-Fe2O3 nanocrystals.23 We believe that the nanoparticles should be γ-Fe2O3 because of the excess trimethylamine oxide used to oxidize the precursory nanoparticles. However, the presence of some magnetite phases inside the particles cannot be ruled out because the XRD profiles of maghemite and magnetite are indistinguishable because of peak broadening. We will call these particles γ-Fe2O3 nonetheless for convenience. We analyzed the γ-Fe2O3 nanoparticles by TGA (Figure 1S of the Supporting Information). There was a gradual and small weight loss between room temperature and 200 °C. This likely arose from the removal of volatile components, such as sorbed water from the system. More abrupt weight losses occurred at ∼250 and ∼380 °C. These have been previously associated with

III. RESULTS AND DISCUSSION The spheres were prepared from an oil-in-water emulsification process. The PCEMA-b-PAA-coated γ-Fe2O3 nanoparticles were prepared first. These nanoparticles were dispersed together with PCEMA into CHCl3 to form an oil phase. Under vigorous stirring, this oil phase was then added to an aqueous phase containing the surfactants PGMA-b-PCEMA and PSGMA-b-PCEMA to yield oil droplets. Heating the system evaporated CHCl3, yielding solid microspheres. Irradiating these microspheres with light with wavelength >270 nm cross-linked the PCEMA domains and produced permanent microspheres. The next sections will discuss different aspects of magnetic microsphere preparation. We will also show the use of the spheres for BSA immobilization. 816

dx.doi.org/10.1021/bm101432e |Biomacromolecules 2011, 12, 813–823

Biomacromolecules

ARTICLE

oleic acid desorption (from the particle surfaces) and oleic acid decomposition temperatures, respectively.24,25 The total weight loss between 200 and 450 °C was 12.5%. This should be equal to the decomposed oleic acid content in the nanoparticles.26 Figure 1b shows a TEM image of the γ-Fe2O3 nanoparticles after ligand exchange. The PCEMA-b-PAA-coated particles appeared the same as the oleic-acid-coated particles. This was reasonable because the ligand exchange was achieved under mild conditions, and no aggregation or morphological changes should occur for the nanoparticles under these conditions. PCEMA-b-PAA-Coated γ-Fe2O3 Nanoparticles. Oleic-acidcoated γ-Fe2O3 nanoparticles did not mix well with the PCEMA320 binder during magnetic microsphere preparation, and their use led to a precipitate rather than water-dispersible microspheres. We replaced the oleic acid ligand by PCEMA30-b-PAA4 to prepare PCEMA-b-PAA-coated γ-Fe2O3 nanoparticles. After PCEMA30-b-PAA4 was added to an oleic-acid-coated γ-Fe2O3 nanoparticle solution in CHCl3, a ligand exchange equilibrium should eventually become established. We fancied that the equilibrium should favor PCEMA30-b-PAA4 adsorption because the polymer contained several carboxyl units per chain and thus the polymer was a multidentate ligand.27 Furthermore, the PCEMA block was not too long, so it should not have created an unruly repulsion among the coronal chains, which could hinder PAA deposition. The PCEMA block was chosen to be 30 units in length so that it was long enough to stabilize the γ-Fe2O3 nanoparticles. To favor further PCEMA30-b-PAA4 deposition, we used an excess amount of PCEMA30-b-PAA4 (mass ratio between oleic-acid-coated γ-Fe2O3 and PCEMA30-b-PAA4 was 1/2). Last but not least, the ligand was exchanged in a dialysis tube, which had a molecular weight cutoff of 3500 Da. Evidently, the molecular weight of oleic acid was lower than and the average molecular weight of PCEMA30-b-PAA4 was higher than 3500 Da. Therefore, oleic acid should preferentially diffuse out of the dialysis tube into the external solvent, which was replaced with fresh CHCl3 at regular time intervals. The frequent changing of the external solvent kept the oleic acid concentration low in the solvent phase and thus favored oleic acid dissociation from the particle surfaces. This should have greatly facilitated the exchange of oleic acid with PCEMA30-b-PAA4. A successful ligand exchange was supported by the following evidence: First, the ligand exchange experiment proceeded as planned. We performed 1H NMR analyses of the nonvolatile material, which came out of the dialysis tube into the external solvent after CHCl3 evaporation. No PCEMA30-b-PAA4 but only oleic acid was detected in the external solvent phase. Dialysis was stopped only after no further oleic acid was drained from the dialysis tube. Second, the ligand exchange process resulted in a change in the behavior of the particles. Before ligand exchange, the γ-Fe2O3 nanoparticles were dispersible in diethyl ether as well as chloroform. After ligand exchange, they could be dispersed in chloroform but not diethyl ether. This change could be readily explained by the fact that both chloroform and diethyl ether were good solvents for the aliphatic tail of oleic acid, whereas PCEMA was soluble only in chloroform and not diethyl ether. Third, our TGA experiments revealed weight losses of 12.5 and 31% between 200 and 450 °C for the oleic-acid-coated and PCEMA-b-PAA-coated nanoparticles, respectively (Figure 1S of the Supporting Information). This agrees with the expectation that the organic content of the nanoparticles should increase after oleic acid displaced PCEMA-b-PAA. One should also note that this organic component weight increase should derive mainly

Figure 3. Comparison of FT-IR spectra of PCEMA-b-PAA (bottom), oleic-acid-coated γ-Fe2O3 nanoparticles (top), and γ-Fe2O3 nanoparticles after ligand exchange (middle). After ligand exchange, the broad oleic carboxylate peak at 1621 cm-1 disappeared.

from the ligated PCEMA-b-PAA because the free chains were mostly removed by the fractionational precipitation treatment described in the Experimental Section. Ligand exchange was also supported by our FT-IR results. Figure 3 compares the FTIR spectra for PCEMA30-b-PAA4 and for the γ-Fe2O3 nanoparticles before and after ligand exchange. The CdO stretching of the γ-Fe2O3-bound oleic carboxylate groups was responsible for the broad peak observed at 1621 cm-1 in the spectrum of the oleic-acid-coated nanoparticles.19,24 After ligand exchange, the PCEMA CdO and CdC stretching peaks at 1716 and 1640 cm-1 dominated, whereas the oleate carbonyl stretching peak at 1621 cm-1 was not seen. These spectral changes were possible only if most of the oleate groups were replaced by the PAA groups of PCEMA30-b-PAA4. We measured at 1716 cm-1 the absorbance of several samples consisting of different amounts of PCEMA30-b-PAA4 in 100 mg of KBr and constructed a calibration curve. This curve was then used to quantify the amount of PCEMA30-b-PAA4 present in a sample of PCEMA30-b-PAA4-coated γ-Fe2O3 nanoparticles. The weight fraction of PCEMA30-b-PAA4 in the ligand-exchanged γ-Fe2O3 nanoparticles was 28%, which agreed well with 31% determined by TGA. Whereas the evidence presented above suggests that most of the oleic acid ligand was replaced by PCEMA30-b-PAA4, we could not rule out the presence of some residual oleic acid ligand. Despite this, we will address the ligand-exchanged nanoparticles as PCEMA-b-PAA-coated γ-Fe2O3 nanoparticles for convenience. Magnetic Microsphere Preparation. The two surfactants used for microsphere preparation were PCEMA55-b-PGMA460 and PCEMA65-b-PSGMA650. These polymers with relatively long hydrophilic blocks and short hydrophobic PCEMA blocks were used because we suspected that they would be effective surfactants. Despite this, the PCEMA blocks should be sufficiently long to facilitate their chemical attachment to the microspheres during the PCEMA cross-linking step. Our prior study indicated that Fe3O4-containing microspheres with an average diameter of ∼300 nm were optimal for diagnostic applications.14 Spheres that were too small did not respond swiftly to an applied magnetic field, whereas those that were too large had less surface area for biomolecule immobilization. Therefore, we decided to prepare spheres with a diameter ∼300 nm. To prepare the spheres, we first dispersed, under vigorous stirring, an oil phase consisting of CHCl3, PCEMA320, and PCEMA-b-PAA-coated γ-Fe2O3 nanoparticles into water using 817

dx.doi.org/10.1021/bm101432e |Biomacromolecules 2011, 12, 813–823

Biomacromolecules

ARTICLE

the two surfactants. CHCl3 was then evaporated from the oil droplets to yield solid particles. We hypothesized that the diameter of the final solid particles would be determined by that of the oil droplets formed under shearing and the amount of solid materials contained in the oil droplets before CHCl3 evaporation. This assumption should be correct because both PCEMA and PCEMA-b-PAA-coated γ-Fe2O3 nanoparticles had negligible solubility in water. Therefore, exchange of these highly hydrophobic materials between different emulsion droplets by diffusion through the aqueous phase should be difficult, and these droplets should behave as mini-emulsion droplets.28,29 We verified this hypothesis previously for another system by monitoring, using dynamic light scattering, the changes of the diameters of emulsion droplets during solvent evaporation.30 Therefore, the key to tuning the sizes of the magnetic microspheres lied in controlling that of the oil droplets. According to past studies,31-33 the addition of oil into stirring water initially leads to the stretching of the oil phase into columns.31 Undulation then develops on these columns or Rayleigh instability sets in, and the columns break up into spherical droplets within 1 s. This is followed by a slower droplet equilibrium process and a further reduction in droplet diameter. At equilibrium, the mean diameter d0 of the oil droplets is31,32,34 d0 µ ðγp0:20 Þ=ðητÞ

We also examined the need for PCEMA320. A typical microsphere preparation protocol excluding PCEMA320 use yielded a precipitate. This suggests that the PCEMA blocks of PCEMA30b-PAA4 and the surfactants did not mix well. At a PCEMA30-bPAA4 weight fraction of 31% of the PCEMA-b-PAA-coated γ-Fe2O3 nanoparticles, the PCEMA30 chains should form a brush layer on the γ-Fe2O3 nanoparticle surfaces when in CHCl3.35 Polymer brushes have long been known to repel each other, and this repulsion has been used to stabilize pigment particles and other additives in paint.36 In a similar manner, the PCEMA30-bPAA4-covered γ-Fe2O3 nanoparticles repelled one another in the oil droplets immediately after their formation. They still repelled each other even after most of the CHCl3 had been evaporated because theory predicted37 and experimental results38 have shown that the different brushes would rather contract than mix. The PCEMA55 and PCEMA65 chains would not mix well with the PCEMA30 brushes because the former PCEMA chains were longer than the brush-forming PCEMA30 chains. They refused to enter the PCEMA30 brush so that they could avoid chain stretching and configurational entropy loss.37,39,40 The term “dry brush regime” has been traditionally used to describe this immiscible situation. Using PCEMA320 as a binder resolved this incompatibility issue. PCEMA320 and the PCEMA30 brush chains should not mix extensively with one another either. However, this did not prevent the entrapment of the PCEMA30-b-PAA4-coated γ-Fe2O3 nanoparticles inside the PCEMA320 matrix. The PCEMA320, PCEMA65, and PCEMA55 chains should mix well with one another because we do not see a reason for brush formation from any of these chain types. As will be discussed later, the PGMA and PSGMA chains of the two surfactants segregated on the microsphere surfaces when an optimal amount of MgSO4 was added to the aqueous phase during microsphere preparation. Therefore, the optimized microsphere preparation protocol also included MgSO4 addition to the aqueous phase. Using this optimized protocol and [MgSO4] = 0.16 mg/mL, we prepared several batches of microspheres. Analysis of more than 100 spheres from one sample yielded an average diameter of 294 ( 91 nm based on TEM measurements. Therefore, the spheres were rather polydisperse. To obtain the standard deviation of 91 nm, we excluded spheres with diameters of 6.0 for successful microsphere preparation. A precipitate was produced when the pH was decreased to