The ability to modify the physical, chemical, and biologic properties of nanoparticles has led to their use as multifunctional platforms for drug delivery and diagnostic imaging applications. Typically, these applications involve functionalizing the nanoparticles with targeting agents. Antibodies remain an attractive choice as targeting agents because of their large epitope space and high affinity; however, implementation of antibody-nanoparticle conjugates is plagued by low coupling efficiencies and the high cost of reagents. Click chemistry may provide a solution to this problem, with reported coupling efficiencies nearing 100%. Although click chemistries have been used to functionalize nanoparticles with small molecules, they have not previously been used to functionalize nanoparticles with antibodies. Concerns associated with extending this procedure to antibodies are that reaction catalysts or the ligands required for cross-linking may result in loss of functionality. We evaluated the efficiency of conjugations between antibodies and superparamagnetic iron oxide nanoparticles using click chemistry as well as the functionality of the product. The results were compared with conjugates formed through carbodiimide cross-linking. The click reaction allowed for a higher extent and efficiency of labeling compared with carbodiimide, thus requiring less antibody. Further, conjugates prepared via the click reaction exhibited improved binding to target receptors.
NANOPARTICLES (NPS), commonly defined as organic or inorganic materials with at least one length dimension below 100 nm, are being widely investigated as drug delivery vehicles and/or imaging agents. Interest in NPs largely stems from their ability to carry a large therapeutic payload (or ample amounts of contrast agent), the ability to finely tune physicochemical properties, which can influence their pharmacokinetic and pharmacodynamic profiles, and the ability to functionalize their surface with molecularly specific targeting agents. For both diagnostic1 and therapeutic2 applications, it is becoming increasingly recognized that specific targeting of NPs is critical toward their effective use. When targeted, reduced amounts of therapeutic and imaging agents are required compared with systemically delivered vectors.3 Furthermore, targeting greatly reduces nonspecific background signal from accumulation of contrast agents at undesirable sites. This also leads to lower toxicity and improved efficacy of therapeutics.
Superparamagnetic iron oxide (SPIO) NPs are a widely used NP system affording T2(*)-weighted contrast for magnetic resonance imaging applications.4 Recently, SPIO NPs have been conjugated to a variety of targeting ligands to provide cellular and molecular specificity for in vitro diagnostic5 and in vivo imaging6 applications. Strategies for targeting SPIO NPs include the use of peptides, endogenous ligands, and monoclonal antibodies.7–9 Despite their relatively large size (roughly 150 kDa), antibodies remain an attractive choice because of their combinatorially large epitope space (of approximately 1015) and their high affinity for targets, with Kd’s often on the order of nanomolar.10
Currently, one of the biggest obstacles faced with the use of antibodies is their low conjugation efficiency to NPs, which can result in the need for large quantities of antibody and prohibitive costs.11 One example of a common approach used for conjugation of antibodies to NPs involves carbodiimide cross-linking.12 The zero-length cross-linker 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide hydrochloride (EDC) reacts with carboxylated NPs in the presence of sulfo-N-hydroxylfosuccinimide (sulfo-NHS) to form amine-reactive sulfo-NHS esters. Subsequent addition of antibodies results in coupling between the NPs and primary amines on the antibody via a stable amide bond. Unfortunately, the reaction is nonideal as it is highly inefficient and thus requires a high excess of antibody. Typically, only about 1 to 20% of the antibody used during the conjugation procedure will be coupled to the NPs.9,13,14
Recently, the emergence and adoption of click chemistry have had a large impact on drug discovery and materials synthesis owing to the ability to achieve conjugation efficiencies nearing 100%. The rapid rise of click chemistry is evident from its novel use in a variety of disciplines, including medicinal chemistry,15 materials and polymer science,16,17 and molecular imaging.18,19 Click chemistry refers to modular chemical conjugations with an emphasis on simple reactions that can take place under a range of conditions with stereospecificity and high efficiency.20 The most widely adopted reaction of this type, the CuI-catalyzed terminal alkyne-azide cycloaddition (CuAAC), was developed by the Sharpless and Meldal groups independently in 2002.21,22
Previously, it has been shown that the coupling of low-molecular-weight species such as peptides23 and fluorophores5 to SPIO NPs can be accomplished with CuAAC. However, evidence that functional antibodies can be coupled to SPIO NPs through CuAAC is lacking. Likewise, reference of large protein linkage to any NP system using a click approach is scarce. Sen Gupta and colleagues briefly described holotransferrin-bound viral capsids, although this was reported by way of transmission electron microscopy micrographs.24 Lipases have also been conjugated to gold NPs25; however, these enzymes are significantly smaller (≈5 times) than antibodies. Applications in which two large molecules/NPs must be coupled are where efficient conjugation is needed most, especially when one of the biologic species is available in limited quantities and/or is expensive, as is typically the case with antibodies.
A major concern with using CuAAC for NP-antibody coupling reactions is that the antibodies will be irreversibly degraded and/or modified during the reaction. For example, it has been shown that the CuAAC reaction can result in the irreversible degradation of viral capsids26 and nucleic acids.27 Further, antibodies could become nonfunctional owing to modification of the innate protein with click-reactive ligands.
In this work, we evaluated the use of CuAAC for conjugation of SPIO NPs to anti-CD20. Anti-CD20 is a clinically approved chimeric monoclonal antibody that has been approved for use in treating B cell lymphoma. This antibody has also been used previously to target SPIO NPs to CD20-expressing tumors.28 We have sought to comparatively assess the utility of the CuAAC reaction for antibody targeting of SPIO NPs against an accepted method, that of carbodiimide conjugation. Particle labeling was assessed for both strategies by means of the efficiency and extent of protein-to-particle labeling and by functional cell-targeting assays.
Azido-dPEG12 NHS ester and propargyl-dPEG NHS ester were purchased from Quanta BioDesign Ltd. (Powell, OH). The NP coating material, dextran T10, was purchased from Amersham Biosciences (now GE Healthcare, Piscataway, NJ). Reagent grade IgG from rat serum was acquired from Sigma Aldrich (St. Louis, MO), and anti-CD20 antibody, rituximab (Genentech, South San Francisco, CA), was a gift of the Eisenberg group. Human lymphoma B cells (Burkitt GA-10) were obtained from the American Type Culture Collection (Manassas, VA). Bathocuproinedisulfonic acid (BCS) was acquired from Acros Organics (Geel, Belgium). All other reagents were purchased from Thermo Fisher Scientific (Waltham, MA) unless otherwise noted.
SPIO NPs were prepared by chemical coprecipitation, as previously described.29 Briefly, 0.7313 g FeCl2 and 1.97 g FeCl3 were each dissolved in 12.5 mL diH2O and added to 25 g dextran T10 in 50 mL diH2O at 4°C. Ammonium hydroxide (15 mL) was slowly added to this mixture, turning the light yellow–colored solution black. This NP slurry was then heated to 90°C for 1 hour and cooled overnight.
Purification of SPIO NPs was accomplished by ultracentrifugation of the mixture at 20,000 relative centrifugal force (RCF) for 30 minutes. Pellets were discarded, and the supernatant was subjected to diafiltration against greater than 20 volumes of 0.02 M citrate, 0.15 M sodium chloride buffer using a 100 kDa cutoff membrane filter (GE Healthcare). The purified particles were then cross-linked by reacting the particles (10 mg Fe/mL) with 25% (v/v) 10 M NaOH and 33% epichlorohydrin. After mixing for 24 hours, the particles were briefly dialyzed and then functionalized with amines by adding 25% ammonium hydroxide. This reaction was allowed to continue for another 24 hours followed by diafiltration as above.
The hydrodynamic diameter of the NPs was measured using a Zetasizer Nano-z (Malvern Instruments, Malvern, UK) through dynamic light scattering. SPIO NPs were diluted in phosphate-buffered saline (PBS) to a concentration of approximately 0.5 mg Fe/mL and read in triplicate. The values reported for all samples are the intensity peak values.
The longitudinal (R1) and transverse (R2) relaxivity of each particle was calculated as the slope of the curves 1/T1 and 1/T2 against iron concentration, respectively. T1 and T2 relaxation times were determined using a Bruker mq60 MR relaxometer operating at 1.41 T (60 MHz). T1 measurements were performed by collecting 12 data points from 5.0 to 1000 milliseconds with a total measurement duration of 1.49 minutes.T2 measurements were made using τ = 1.5 milliseconds and two dummy echoes and fitted assuming a monoexponential decay.
The number of amines per particle was determined following the general procedure described by Zhao and colleagues.30 Briefly, iron oxide particles at a concentration of 2 mg Fe/mL were reacted with excess N-succinimidyl 3-(2-pyridyldithio) propionate (SPDP; Calbiochem, San Diego, CA) for 4 hours. SPIO NPs were washed of excess SPDP through repeated precipitation in isopropanol and resuspension in PBS. The particles were then run through a 50 kDa molecular weight cut-off centrifugal filter (YM-50, Millipore, Billerica, MA) either with or without the addition of disulfide cleavage agent tris(2-carboxyethyl)phosphine. The difference in the absorbance of these two samples at 343 nm was used to determine the concentration of SPDP in the filter flow. Adjusting for dilution, the number of amines per particle was determined.
Fluorescein isothiocyanate (FITC) was used to fluorescently label amine-functionalized cross-linked SPIO NPs by reacting at a molar ratio of 19:1 FITC to iron. The unbound FITC was washed from the FITC-SPIO NPs using a PD10 gel filtration column (GE Healthcare) equilibrated with PBS. All particles were FITC labeled prior to subsequent modification, ensuring that all particles had equal fluorescent labeling.
A schematic outlining the carbodiimide crosslinking procedure is illustrated in Figure 1A (strategy i). FITC-labeled, amine-functionalized NPs (as shown in Figure 1A) were derivatized to acid residues following reaction with an excess of succinic anhydride in basic solution. Specifically, 200 µL of NH2-SPIO NPs (5 mg Fe/mL) in 0.02 M citrate buffer, pH 8, was added to 10 µL of 1 M NaOH followed by 10 µL of 4 M succinic anhydride in dimethylformamide. The reaction was allowed to mix for 4 hours. Carboxylated SPIO NPs (COOH-SPIO NPs) were subsequently precipitated three times in four volumes of isopropanol to remove free reactants.
Carbodiimide coupling reactions were accomplished with FITC-labeled COOH-SPIO NPs in 2-(N-morpholino)ethanesulfonic acid (MES) buffer, pH 6.0. To 50 µL of SPIO NPs (2 mg Fe/mL), 10 µL each of 200 mM EDC and 500 mM sulfo-NHS was added. Solutions mixed for 1 hour at 25°C were then precipitated in 500 µL of isopropanol. A resuspension of the NPs in 100 µL of 10 mg/mL antibody solution in PBS, pH 7.4, was allowed to react at 15°C overnight. Purification was carried out using an MS magnetic column (Miltenyi Biotec, Bergisch Gladbach, Germany) in PBS.
The bicinchoninic acid protein assay was used for protein concentration of conjugate solutions, after correcting for background with unlabeled SPIO NPs. To determine the number of proteins per particle, the molarity of SPIO NPs was assessed spectrophotometrically by dissolving the NPs and oxidizing the iron with 6 M HCl and 3% H2O2. Absorbance at 410 nm was compared to a standard curve, and particle concentration was calculated assuming 8,924 Fe atoms per particle, as previously determined.29
A schematic outlining the CuAAC cross-linking procedure is found in Figure 1A (strategy ii). FITC-SPIO NPs were reacted with the amine-reactive azido-dPEG12 NHS, diluted 10 times from stock in dimethyl sulfoxide (DMSO), in 0.1 M sodium phosphate buffer, pH 9. The linker was added at 100 times molar excess to the SPIO NPs. After mixing for 4 hours, the SPIO NPs were purified twice on PD10 columns equilibrated with PBS. Azide-SPIO NPs were then concentrated on an Ultracel 30,000 centrifugal filter (Millipore, Billerica, MA) in PBS.
Alkyne functionalization of antibodies (Figure 1A), for both IgG and rituximab, was accomplished by the addition of 10% v/v propargyl-dPEG-NHS in DMSO to 10 mg/mL antibody in 0.1 M sodium phosphate buffer, pH 9. The concentration of propargyl-dPEG-NHS was varied to provide different degrees of labeling per antibody. Antibody was purified on a PD10 column equilibrated with PBS and then reconcentrated using Ultracel 30,000 filters. Antibody concentration postpurification was assessed spectrophotometrically at 280 nm using a molar extinction coefficient of 210,000 M–1cm–1.
SPIO NPs were conjugated to alkyne-functionalized IgG and anti-CD20 using the same procedures. FITC-labeled N3-SPIO NPs (3 mg Fe/mL) were mixed with varying volumes of 10 mg/mL alkynated antibody, 5 mM BCS, 1 mM CuSO4, and 5 mM sodium ascorbate. Final volumes of the reactions were brought to a constant level with the addition of PBS. The samples were mixed for 4 hours at 15°C. Samples were cleaned of the reaction additives using a YM-50 spin column. The antibody-linked NPs were then purified from unbound protein on an MS magnetic column. Protein per particle labeling was determined as above.
SPIO NPs were incubated with 100 µL of 1 × 106 cells/mL Burkitt GA-10 lymphoma B cells for 30 minutes at 37°C, 5% CO2, in a 96-well plate. SPIO NPs were added at final concentrations of 50 µg Fe/mL and 10 µg Fe/mL for carbodiimide- and click-conjugated SPIO NPs, respectively. In the antibody inhibition experiments, 50 µg/mL (final concentrations) of free anti-CD20 was added prior to the addition of the antibody-conjugated NPs. The free, unbound particles were purified from the cells through three PBS washes at 1 RCF for 5 minutes each. Cells were finally resuspended in 300 µL of PBS and placed in a 96-well plate to be read using a Guava Easycyte Plus system (Guava Technologies, Hayward, CA). Flow cytometry data were analyzed using FlowJo (TreeStar Inc., San Francisco, CA).
Click-conjugated particles at a concentration of 50 µg Fe/mL were incubated with cells as above, using 30 × 103 cells. Images were acquired with an Olympus IX 81 inverted fluorescence microscope using a LUC PLAN 40× objective (numerical aperture 0.6; Olympus) following fluorescein excitation by an X-cite 120 excitation source (EXFO, Quebec, QC). Micrographs were taken using a back-illuminated electron multiplying charge-coupled device camera (Andor Technology PLC, Belfast, Northern Ireland).
Antibodies were coupled to SPIO NPs using either conventional carbodiimide chemistry or CuAAC, as shown in Figure 1A. The SPIO NPs used throughout the present study possessed an average hydrodynamic diameter of 33.4 nm and R1 and R2 values of 13.56 1/mM*s and 71.00 1/mM*s, respectively. The SPIO NPs were also aminated (≈185 NH2 per particle), making this probe well suited to modification with dye and proteins. Initially, all CuAAC conjugations were carried out using rat IgG so that the reaction parameters could be optimized in a cost-effective manner. The first parameter to be varied was the number of alkyne groups introduced onto IgG. This was accomplished by varying the labeling ratio of alkynating reagent, CH-PEG-NHS, to IgG from 15:1 to 100:1. In a separate reaction, azide-labeled SPIO NPs were prepared by reacting aminated SPIO NPs with a 100-fold molar excess of N3-PEG-NHS, as seen in Figure 2. Following appropriate purification procedures, alkyne-labeled IgG were coupled to the azide-SPIO NPs through CuAAC in a catalytic copper solution. For this reaction, the molar ratio of IgG per SPIO NP was held constant at 17.5:1. After a 4-hour incubation, unbound IgG were removed by magnetically purifying the SPIO NPs and the average number of IgG coupled to each SPIO NP was quantified. It was found that the number of IgG per SPIO NP increased with the degree of alkyne incorporation, to a maximum labeling of 10 to 13 IgG per SPIO NP for this series of reactions. The maximum number of IgG per SPIO NP was achieved when IgG was reacted with > 35-fold molar excess of CH-PEG-NHS. Therefore, all subsequent alkynating reactions were carried out using these lowest saturating conditions.
An important advantage of CuAAC reactions has been the high efficiency of conjugations. This is a particularly desirable property when procedures involve large and expensive proteins such as antibodies. Next, we determined the conjugation efficiency between IgG and SPIO NPs for labeling ratios of alkyne-IgG to azide-SPIO NPs ranging from 2.5:1 to 35:1. As shown in Figure 3, there was nearly a 100% coupling efficiency when the labeling ratio was less than ≈20 IgG per SPIO NP. At higher labeling ratios, the number of IgG per SPIO NP did not improve, likely because there was insufficient space on the SPIO NP surface for additional conjugations. Using this information, subsequent conjugation of the clinically approved anti-CD20 was materially conserved as a labeling ratio of only 15 antibodies per SPIO NP was used.
To assess whether antibodies that have been coupled to SPIO NPs remain functionally active following the CuAAC reaction, SPIO NPs were coupled with anti-CD20 antibodies and their ability to bind to CD20-positive B cells was assessed. For comparison, the same antibodies were also coupled to SPIO NPs using carbodiimide chemistry. Each step of the carbodiimide reaction was conducted under saturating conditions to maximize the conjugation efficiency. Specifically, carboxylated SPIO NPs were reacted with an excess of EDC and sulfo-NHS in MES buffer, pH 6.0, for 1 hour in an attempt to achieve complete activation of all available carboxyl groups. To minimize subsequent hydrolysis of the NHS-activated SPIO NP, unreacted cross-linking agents were removed as rapidly as possible (≈2–5 minutes) by precipitating the SPIO NPs with isopropanol. The SPIO NPs were then resuspended directly in an excess of anti-CD20 in PBS, pH 7.4, and mixed overnight.
As shown in Figure 4B, the copper-catalyzed ligation of anti-CD20 to SPIO NPs after 4 hours yielded approximately 6.85 anti-CD20 per SPIO NP. Carbodiimide cross-linking resulted in only 2.77 anti-CD20 per SPIO NP. The labeling efficiency for each conjugation strategy, defined as a percentage of conjugated protein per SPIO NP over total protein per SPIO NP, is presented in Figure 4B. The click chemistry reaction achieved 45.6% efficiency, whereas the carbodiimide reaction resulted in a labeling efficiency of only 8.3%. It was surprising that despite a relatively high degree of labeling by the protein through CuAAC, this was significantly less than achieved with rat IgG. It is hypothesized that the reduced efficiency may stem from a lower incorporation of alkynes onto the anti-CD20 compared with IgG, perhaps owing to subtle differences between the proteins. Nonetheless, CuAAC still resulted in a 5.5-fold improvement in the efficiency of labeling compared with carbodiimide chemistry.
Binding of anti-CD20-conjugated SPIO NPs to CD20-positive B cells was assessed by flow cytometry. All SPIO NPs were fluorescently labeled prior to the respective conjugation protocols to allow for fluorescence detection. The same fluorescently labeled SPIO NPs were used for both the CuAAC and the carbodiimide reactions to ensure that total cellular fluorescence could be directly correlated to the extent of SPIO NP binding. Thus, it was possible to quantitatively compare between the equally fluorescent EDC- and CuAAC-particle labeling strategies. As shown in Figure 5 (top row), it was found that at a concentration of 10 µg Fe/mL, anti-CD20-targeted SPIO NPs prepared using click chemistry could be used to successfully target B cells. Specificity was confirmed by competitively inhibiting SPIO NP binding using an excess of free anti-CD20 antibodies. Surprisingly, anti-CD20-targeted SPIO NPs prepared using carbodiimide chemistry did not result in any appreciable B cell labeling using equivalent concentrations (10 µg Fe/mL; see Figure 5, Figure 5–Figure 5). In fact, similar levels of B cell labeling could be achieved only when the iron concentration was increased to 50 µg Fe/mL (Figure 5, bottom row). Again, for both concentrations of the carbodiimide-conjugated targeted NPs, specificity was confirmed by competitively inhibiting SPIO NP binding using an excess of free anti-CD20 antibodies. These data suggest that the anti-CD20-labeled SPIO NPs prepared using click chemistry have a higher avidity for the CD20 receptors on B cells compared with analogous NPs prepared using carbodiimide chemistry. It is likely that this increase in avidity is due to the greater number of anti-CD20 per particle. Overall, the data presented here provide strong evidence that CuAAC reactions can be used for highly efficient and effective labeling of NPs with antibodies, without any noticeable loss in antibody functionality. Fluorescent micrographs of click-conjugated SPIO NPs incubated with B cells are also shown (Figure 6). Labeling of the cells is found to be specific as IgG-SPIO NP conjugates do not bind to the cells, nor do anti-CD20-SPIO NPs inhibited with excess free antibody.
Advanced applications in the areas of diagnostic imaging and drug delivery stand to benefit from antibody- or protein-based targeting of NP carriers. Preparation of such targeted NPs can, however, be inefficient and costly. Here we have shown that the use of the CuAAC for the production of antibody-NP conjugates permits materially efficient and functionally competent targetable probes. In comparison with classic carbodiimide conjugates, CuAAC adds a valuable alternative to the toolbox of the bioconjugate chemist. This mechanistically simple and highly efficient reaction, which is generally insensitive to reaction conditions, is expected to allow for an expansion in applications for NP-targeted systems in the clinic.
We would like to acknowledge the help of Dr. Eisenberg and his laboratory at the Department of Rheumatology, School of Medicine, University of Pennsylvania.
Financial disclosure of authors: This work was supported in part by Wyeth-Ayerst Pharmaceuticals, the Transdisciplinary Program in Translational Medicine and Therapeutics, the Lupus Research Institute, the Department of Defense Breast Cancer Research Program of the Office of the Congressionally Directed Medical Research Programs (W81XWH-07-1-0457), and the National Institutes of Health–National Cancer Institute (R21-CA132658).
Financial disclosure of reviewers: None reported.
Schematic of antibody-SPIO NPs conjugation strategies.
Effect of IgG alkylation on conjugation to SPIO NP. To determine the minimally required alkyne residues on IgG necessary to achieve maximum conjugation with SPIO NPs, IgG was reacted with increasing amounts of CH-PEG-NHS. The alkyne-IgG samples were then “clicked” to N3-SPIO NPs, and the degree of labeling was assessed. The degree of antibody labeling began to plateau at an approximate starting labeling ratio of 35 CH-PEG-NHS:IgG.
Efficiency of CuAAC-based IgG-SPIO NP conjugations as a function of IgG concentration. The click reaction between protein and SPIO NP was performed with increasing molar excess of IgG to SPIO NP. The labeling Efficiency was ????100% for all conjugations up to a labeling ratio of ~20 IgG per SPIO NP. Higher labeling ratios did not result in any increase in the number of conjugated antibodies.
Comparison of SPIO NP labeling with anti-CD20 using CuAAC and carbodiimide conjugation procedures.
B-cell labeling with anti-CD20 SPIO NP. The functional targeting of SPIO NPs to B cells was accomplished for click- and carbo-diimide-cross-linked anti-CD20-SPIO NP conjugates. The top row shows the flow cytometric analysis of B cells that were incubated with 10 μg Fe/mL of CuAAC-conjugated (
Fluorescence micrographs of B cells labeled with anti-CD20 SPIO NPs. Cells were incubated for 30 minutes with 50 μg Fe/mL of (