Article pubs.acs.org/Langmuir
Comprehensive Study of DNA Binding on Iron(II,III) Oxide Nanoparticles with a Positively Charged Polyamine ThreeDimensional Coating Humphrey H. P. Yiu,† Laurent Bouffier,† Paul Boldrin,† James Long,‡ John B. Claridge,† and Matthew J. Rosseinsky*,† †
Department of Chemistry, University of Liverpool, Liverpool L69 7ZD, United Kingdom IOTA NanoSolutions Ltd, MerseyBio, Crown Street, Liverpool L69 7ZB, United Kingdom
‡
S Supporting Information *
ABSTRACT: Iron (II,III) oxide Fe3O4 nanoparticles (25 and 50 nm NPs) are grafted with amine groups through silanization in order to generate a positively charged coating for binding negatively charged species including DNA molecules. The spatial nature of the coating changes from a 2-D-functionalized surface (monoamines) through a layer of amine oligomers (diethylenetriamine or DETA, about 1 nm in length) to a 3-D layer of polyamine (polyethyleneimine or PEI, thickness ≥3.5 nm). These Fe3O4−PEI NPs were prepared by binding short-chain PEI polymers to the iodopropyl groups grafted on the NP surface. In this work, the surface charge density, or zeta potential, of the nanoparticles is found not to be the only factor influencing the DNA binding capacity, which also seems not to be affected by their buffering capacity profile in the range of pH 4−10. This study also allows the investigation of this 3-D effect on the surface of a nanoparticle as opposed to conventional 2-D amine functionalization. The flexibility of the PEI coating, which consists of only 1, 2, and 3° amines, on the nanoparticle surface has a significant influence on the overall DNA binding capacity and the binding efficiency (or N/P ratio). These polyamine-functionalized nanoparticles can be used in the purification of biomolecules and the delivery of drugs and large biomolecules.
■
INTRODUCTION Surface-functionalized iron oxide nanoparticles (NPs) have found many uses in biomedicine in recent years, including magnetic resonance imaging (MRI),1 cell labeling and tracking,2 magnetic targeting and drug delivery,3 and magnetic hyperthermia.4 Magnetite (Fe3O4) and maghemite (γ-Fe2O3) nanoparticles are the most popular choices because of their high magnetization at saturation (up to ca. 92 emu g−1 or 7.11 emu mol−1 Fe,5 which is around 70 times higher than that of hematite α-Fe2O3 (ca. 1.2 emu g−1)6 at an external field of 10 kOe) and low cytotoxicity.7 Magnetite nanparticles with a positively charged surface have already been used in DNA binding and delivery in vitro.8−10 Further development of these nanomaterials can represent a new generation of gene therapy and nanomedicine.9−11 Such a positively charged surface is usually achieved by binding organic functional groups on the surfaces of particles.12 For this purpose, positively charged organic groups, usually amines, have been used to coat various nanoparticle types, and these functionalized nanoparticles have shown effective binding not only for DNA13−15 but also for other negatively charged biomolecules such as siRNA16 and proteins (e.g., bovine serum albumin, BSA).17,18 However, this positively charged surface property is controlled by the solution pH.19 Therefore, it is essential to ensure that the nanoparticle © XXXX American Chemical Society
surface is positively charged in the binding media during the DNA binding experiment. The surface charge profile of a nanoparticle sample in various pH ranges governs not only the binding but also the release of DNA molecules.20 For DNA delivery or transfection, the release of DNA after the uptake of these DNA/nanoparticle complexes into the lysosome can be affected by the buffering capacity of the nanoparticle surface. Because the lysosomal compartment is mildly acidic, the characteristics of the nanoparticle during “neutralization” can prevent the delivered DNA molecules from being degraded.21 The sequential release of DNA is then triggered by the swelling and rupture of vesicles. However, this “proton sponge” hypothesis is still not fully understood, and in general a higher transfection rate was observed from the polycationic species (transfecting agents) with a higher buffering capacity in the lower pH range of 4.5−7.4, which is similar to a lysosomal environment.22 As a result, the pH profile of a nanoparticle surface can influence not only the DNA binding characteristics but also the release mechanism of DNA, which is directly linked to the overall transfection efficiency of the nanoparticle samples Received: March 7, 2013 Revised: August 12, 2013
A
dx.doi.org/10.1021/la400848r | Langmuir XXXX, XXX, XXX−XXX
Langmuir
Article
Scheme 1. Functionalization of Fe3O4 MNPsa
a
(a) One-step grafting of amine-silanes. (b) Grafting of methylated DETA-silane. (c) Two-step grafting of polyethyleneimine (PEI).
binding and delivery, nanoparticles with a high DNA binding capacity will be ideal because, in that case, the amount of nanomaterials to be used can be minimized.35 The latter is an indication of how many amine groups are engaging in direct interactions with the negatively charged phosphate groups on the DNA molecules. The lower the N/P number, the more efficient the binding becomes. Unfortunately, these two parameters of a functionalized nanoparticle sample are not straightforward to predict. In theory, the higher number of amine groups on the surface, which induces more positive charges at a given pH (normally at pH 6−8 for a binding experiment), the more DNA that will be bound. However, these amine groups also compete with each other for protons, and thus the overall number of positive charges on the surface is not always proportional to the number of amine groups.36,37 Also, the distribution of amine groups may be another factor affecting the DNA molecules’ binding to a particle surface because of the unique double-helix structure of the molecules. In other words, for effective DNA binding, a flexible amine surface may be necessary to accommodate the specific orientation of the phosphate groups on bulky DNA molecules. This work aims to provide a detailed study of the relationship between the surface amine groups on Fe3O4 nanoparticles to the DNA binding capacity and efficiency (or N/P ratio). A number of grafting pathways (Scheme 1) allow us to functionalize Fe3O4 nanoparticles with a 2-D amino surface to a 3-D amine network (or PEI) of a few nanometers in thickness (estimated using linear PEI with Mw = 423, see Figure S1). Two particle sizes (25 and 50 nm) were also chosen for the study of the size effect. The effect of the extent of functionality and the zeta potential of the nanoparticles on their DNA binding capacity will be studied. These nanoparticles can be potentially used in many biomedical applications in vitro and in vivo.
or the transfection agent. Moreover, the DNA binding characteristics can also be influenced by the size and shape of the nanoparticle.23−25 In our current study, we focus on the effect of the grafted amine groups on the nanoparticle surface with respect to their DNA binding properties. Amine-group-functionalized nanoparticles can be prepared via a number of methods, including the physical adsorption of a positively charged polymer (such as polyethyleneimine, PEI),26−28 cross-linking with an amino-polymer,29 and grafting using an amino-alkoxysilane.30 Without strong chemical bonds, the physical adsorption of PEI onto nanoparticles usually produces a temporary coating because of the weak interaction between the PEI molecules and the surface of the nanoparticles.28 Cross-linking of amino-polymers (dextran amine or chitosan) can trap the nanoparticles inside a polymer network, but it is likely to release the iron oxide core in the presence of digestive enzymes in vivo.31 In recent years, the grafting method using alkoxysilanes, or silanization, has become more popular because it creates more stable anchors via an −O−Si− covalent monolayer. However, this method has been underexplored. Despite the availability of organosilanes with various organic functional groups, 3-aminopropyltriethoxysilane, or APTES, is still the most popular choice because of the selfcatalyzed reaction with the basic amine groups.32 Grafting using other organosilanes may need harsher reaction conditions but is still possible. Moreover, further reactions on these grafted groups allow a wider range of functionalities on the surfaces of magnetic nanoparticles.30,33 Another important feature of silanization is that the thickness of the coating can be kept at a minimum whereas the cross-linking method usually generates a thick coating, which increases the overall particle size.11 The contrast in structure of the functionalized MNPs prepared from these two methods is depicted in Scheme S1. There are two essential physicochemical properties to be determined in assessing these systems: (i) the DNA binding capacity, which is measured as the mass of DNA bound to a unit mass of nanoparticles and (ii) the DNA binding efficiency, or N/P ratio, which is the ratio of the number of nitrogen atoms (or amine groups) on the nanoparticle surface to the number of phosphorus atoms (or phosphate groups) on the bound DNA molecules.34 For applications such as DNA
■
EXPERIMENTAL SECTION
Materials. 3-Iodopropyltrimethoxysilane (95%) was purchased from Fluka, and 3-aminopropyl-triethoxysilane (APTES, 99%), Nmethyl-3-aminopropyltrimethoxysilane (95%), (N,N-dimethyl-3aminopropyl)trimethoxysilane (95%), N-trimethoxysilylpropylN,N,N-trimethylammonium chloride (50% in methanol), and B
dx.doi.org/10.1021/la400848r | Langmuir XXXX, XXX, XXX−XXX
Langmuir
Article
(Thermo EA1112 Flash CHNS-O analyzer) and thermal gravimetric analysis (TGA, TA Instruments Q5000IR with an autosampler) were used to analyze the organic content of the functionalized samples. Xray powder diffraction data were collected with Co Kα1 radiation (λ = 1.781 Å) on a Panalytical X’pert Pro multipurpose diffractometer. The BET surface area of unfunctionalized Fe3O4 nanoparticles was measured using a Micromeritics Gemini VII 2390P unit with nitrogen as the adsorbent. The samples were activated at 120 °C under flowing nitrogen for 2 h prior to measurement. The zeta potential of the particles in unbuffered deionized water was recorded using a Malvern Nanosizer. Fourier transform infrared spectrometry (FTIR) spectra were obtained from powder samples on a Perkin-Elmer Universal ATR spectrometer. For each sample, 20 scans in the region from 650 to 4000 cm−1 were accumulated with a resolution of 4 cm−1. Buffering Capacity Measurement. The buffering capacity of the Fe3O4 nanoparticles was measured following the method published in the literature22 with modification. In a typical experiment, an Fe3O4 nanoparticle sample was diluted with 150 mM NaCl (10 mL) at a nitrogen concentration of 5 mM, which was calculated from the CHN analysis data of the sample. The pH of the solution was then adjusted to 11 with 0.1 M NaOH in order to deprotonate the amine groups fully. Finally, the solution was titrated with 0.1 M HCl, and the pH value was monitored using a pH meter (Hanna Instruments) equipped with a crystal-body pH electrode (Mettler Toledo) coupled to a data logging and collection system (Pico Technology Ltd.); calibration was performed with standard buffer solutions (Fisher Scientific). All measurements were made in triplicate. As a reference, a solution of branched PEI 1800 dissolved in 150 mM NaCl at a concentration of 5 mM nitrogen was also titrated using the same method. The buffering capacity of unfunctionalized Fe3O4 (25 and 50 nm) samples was also plotted for comparison. DNA Binding Capacity Measurement. All DNA binding experiments were carried out using unbuffered deionized water (pH 6−6.5) to minimize the competition between DNA and buffering anions. Nanoparticles (10 to 200 μg) were suspended in Millipore water in a 1.5 mL Eppendorf tube with 5 μL of DNA (at a concentration of 1 mg mL−1). The total volume of the suspension was kept at 1.0 mL, making the DNA concentration 5 μg mL−1. For samples of higher binding capacity (Fe3O4−PEI(1200) and Fe3O4− PEI(1800)), 25 μL of DNA solution (25 μg of DNA mL−1) was added instead of 5 μL. The suspensions were incubated at room temperature under rotation. After two centrifugations, the unbound DNA in the supernatant was recorded using a UV−vis spectrophotometer at 260 nm. The amount of DNA bound on the particles was calculated by subtracting the DNA concentration in the supernatant from a blank experiment. Each binding experiment was repeated three times, and average values of the three experiments were plotted as the DNA binding curve. The specific DNA binding capacity is presented as the unit weight of DNA (in micrograms) bound to the unit weight of magnetic nanoparticles (also in micrograms).
trimethoxysilylpropyldiethylenetriamine (DETA-silane, 95%) were purchased from Fluorochem. Iron(II,III) oxide (Fe3O4, magnetite) nanoparticles were purchased from Alfa Aesar (nanopowder, 20−30 nm, 99%). Iron(II) sulfate heptahydrate (FeSO4·7H2O, ACS reagent, 99%), potassium nitrate (99%), potassium hydroxide, and methyl iodide (MeI, 99%) were purchased from Sigma-Aldrich, and acetone, diethyl ether, and methanol (all GPR grade) were supplied by BDH. Four short-chain polyethyleneimine (PEI) polymers (PEI(423), linear, Mw = 423; PEI(600), branched, Mw = 600, Mn = 800; PEI(1200), branched, Mw = 1200, Mn = 1300, 50% solution in water; PEI(1800), branched, Mw = 1800, Mn = 2000, 50% solution in water) were purchased from Sigma-Aldrich. DNA from calf thymus (activated, Sigma) was used to measure the binding capacity of samples. Water was purified with a Millipore system. Toluene (GPR grade, Fisher) and N,N-dimethylformamide (DMF, GPR grade, Fisher) were dried with activated molecular sieves (4 Å, Aldrich) overnight prior to use. Ether (Et2O, GPR grade) was purchased from BDH. All other chemicals were used without further purification. Preparation of Fe3O4 50 nm Nanoparticles. Fe3O4 nanoparticles (50 nm) were prepared using the mild oxidation of iron(II) sulfate in an alkaline medium, following the method suggested by Sugimoto and Matijevic.38 FeSO4·7H2O (17.71 g, 0.06 mol) was dissolved in 200 mL of water. A potassium nitrate solution (1 M, 100 mL) and potassium hydroxide (0.5 M, 50 mL) were added to the reaction mixture with stirring. The reaction mixture was held at 90 °C under flowing nitrogen for 2 h. The black precipitate of Fe3O4 nanoparticles was separated using a rare earth NdFeB magnet and washed with deoxygenated doubly deionized water (dd H2O) until pH 7 was reached. Silanization of Fe3O4 Nanoparticles. In a typical experiment, 0.5 g of Fe3O4 nanoparticles was dispersed in toluene (dried over activated 4 Å molecular sieves, 50 mL) using an ultrasonic bath. Organosilane (0.5 mL, 2.136 mmol, ca. 10-fold excess) was added to the Fe3O4 suspension, and the reaction mixture was heated to reflux with mechanical stirring. After 24 h, the solid was recovered using a NdFeB magnet and washed with acetone five times to remove residual toluene and unreacted silane. The nanoparticles were dried under vacuum. These functionalized nanoparticles are denoted as Fe3O4−NH2 and Fe3O4−DETA for the functionalities of amine and diethyltriamine, respectively (Scheme 1a), and these nanoparticles are named as aminefunctionalized and diethyltriamine-functionalized iron(II,III) oxide nanoparticles. Methylation of DETA-Silane. For the synthesis of methylated DETA-silane, a mixture of trimethoxysilylpropyldiethylenetriamine (DETA-silane, 0.664 g, 2.5 mmol) and CH3I (1.280 g, 9.0 mmol) in DMF (2 mL) was stirred overnight at room temperature. DMF was eliminated by addition and subsequent removal of Et2O (50 mL). Solubilization in MeOH (3 mL), addition, and removal of Et2O allowed another precipitation. The product was washed with Et2O and dried overnight under vacuum to isolate the methylated DETA-silane (1.265 g, 1.83 mmol, Mw = 691.6, based on the structure shown in Scheme 1b) in 73% yield. This material was used in the silanization of Fe3O4−Me−DETA nanoparticles following the same experimental procedure described in the previous section (Scheme 1b). PEI Binding to Propyliodide-Functionalized Fe3O4 Nanoparticles (Fe3O4−I). Fe3O4−I nanoparticles (200 mg), prepared with the same silanization procedure as used for Fe3O4−NH2 but using 3iodopropyltrimethoxysilane instead of APTES, were suspended in DMF (50 mL) with sonication. PEI (0.5 mL for pure PEI, Mw = 423 and 600 and 1.0 mL for 50% aqueous solutions, Mw = 1200 and 1800) was added to the reaction mixture and heated to reflux with mechanical stirring for 24 h. The nanoparticles were recovered using a NdFeB magnet and washed with acetone five times in order to remove residual DMF and unbound PEI. Finally, the nanoparticles were dried under vacuum. These nanoparticles are denoted as Fe3O4− PEI(n) where n is the Mw of the PEI used during functionalization (Scheme 1c). Characterization of Nanoparticles. The morphology of functionalized Fe3O4 nanoparticles was studied using a transmission electron microscope (Tecnai) at 100 keV. CHN elemental analysis
■
RESULTS AND DISCUSSION Characterization of Monoamine-Functionalized Fe3O4 Nanoparticles. The degree of functionalization of the Fe3O4 nanoparticles was analyzed using CHN elemental analysis and thermogravimetric analysis (TGA) in air. Typical TGA traces for functionalized Fe3O4 samples are shown in Figure S2. Table 1 shows the number of groups per gram of Fe3O4 nanoparticle, calculated from the carbon content (CHN analysis) of the nanoparticle samples and the weight loss (120−600 °C) from the sample (TGA). There are 120−430 μmoles of amine groups per gram, and generally more groups have been grafted onto 25 nm nanoparticles than onto 50 nm nanoparticles because of the higher BET surface area exhibited from 25 nm nanoparticles (48 vs 22 m2 g−1 for 50 nm nanoparticles). In both 25 and 50 nm particles, the grafting efficiency decreases as the order of the amine increases, or 1° > 2° > 3°. This could be due to an increasing ligand size from 1 to 3° amine and C
dx.doi.org/10.1021/la400848r | Langmuir XXXX, XXX, XXX−XXX
Langmuir
Article
and from 0.1 to 0.3 nm2 (10 to 30 Å2) for 50 nm nanoparticles. This also means that the distances between groups are 5 to 7 Å and 4 to 6 Å for 25 and 50 nm nanoparticles, respectively, assuming that the footprints are perfect circles. A theoretical footprint is calculated to be 0.074 nm2 (or 7.4 Å2, see Figure S3). This indicates that the surfaces of both 25 and 50 nm Fe3O4 nanoparticles are close to saturated with functional groups under the reaction conditions employed. We can also assume a monolayer coverage of amine groups on the nanoparticle surface because there was no evidence of multilayer silane build up. FTIR spectroscopy (Figure 1) confirms the nature of the functional groups on the Fe3O4 nanoparticles. As expected, the anchoring of amino-alkyl silane could be evidence of the corresponding FTIR data (NH2 stretching at 3300 cm−1 and
Table 1. Number of Functional Groups Grafted onto the Fe3O4 Nanoparticles (25 and 50 nm) Analyzed Using CHN Microanalysis and TGAa
Fe3O4 nanoparticle sample Fe3O4−NH2 Fe3O4− NH(CH3) Fe3O4− N(CH3)2 Fe3O4− N(CH3)3+Cl− Fe3O4−DETA Fe3O4−Me− DETA Fe3O4− PEI(423) Fe3O4− PEI(600) Fe3O4− PEI(1200) Fe3O4− PEI(1800)
25 nm NPs (BET SA = 48 m2 g−1)
50 nm NPs (BET SA = 22 m2 g−1)
no. of no. of groups groups from per nm2 CHN (and calculated TGA) from CHN (μmol g−1)b (and TGA)
no. of groups from CHN (and TGA) (μmol g−1)b
no. of groups per nm2 calculated from CHN (and TGA)
430 (500) 380 (320)
5.4 (6.3) 4.8 (4.0)
350 (370) 240 (250)
9.6 (10.1) 6.6 (6.8)
190 (420)
2.4 (5.3)
120 (180)
3.3 (4.9)
330 (400)
4.1 (5.0)
340 (320)
9.3 (8.8)
370 (150) 140 (220)
4.6 (1.9) 1.8 (2.8)
350 (200) 170 (50)
9.6 (5.5) 4.7 (1.4)
90 (120)
1.1 (1.5)
20 (30)
0.5 (0.8)
40 (440)
0.5 (5.5)
10 (170)
0.3 (4.7)
170 (380)
2.1 (4.8)
50 (140)
1.4 (3.8)
110 (220)
1.4 (2.8)
40 (230)
1.1 (6.3)
a
Table S1 shows the graphical representations of the structure of the functional groups on the surface of the samples. bCalculated from %C of the sample.
consequent steric hindrance. As a result, ATPES is the most effective silanization agent for grafting monoamine groups. To examine the grafting efficiency in detail, calculating the area occupied by a grafted group, or the footprint size, can provide some useful information for understanding the distribution of groups on the surface and also the binding efficiency (or N/P ratio). For a soft material such as free PEI polymer, the number of amine groups grafted should relate directly to the DNA binding capacity of the materials. However, the efficiency of the DNA binding of a hard surface is also likely to be affected by the distance between amine groups (i.e., the footprint size). This is because DNA is a large molecule and can cause significant steric hindrance during binding. Therefore, materials with a large number of amine groups but with a small footprint size are expected to have a low binding efficiency because many amine groups do not engage with the phosphate groups on the DNA molecules. From the data shown in Table 1, there are 2.4 to 9.6 groups per nm2. When compared to a literature value reported by Lausen et al.,39 where an average of 10.5 PEG-silane groups per nm2 was estimated to be attached to 8 nm iron oxide nanoparticles using a similar grafting method, fewer groups were found to be attached to particles. Indeed, the footprint of the grafted groups can become smaller when the particle size decreases because of the curvature of the nanoparticle.40 In the current work, comparing two sets of nanoparticle samples, a larger number of groups per area was recorded for the larger 50 nm nanoparticle for all monoamine and oligo amine groups. This is likely because 50 nm particles, prepared by precipitation, have a more reactive surface than the commercial 25 nm particles, as discussed previously and shown by TGA (Figure S2a). The average footprint size of these attached groups ranges from ca. 0.2 to 0.4 nm2 (20 to 40 Å2) for 25 nm nanoparticles
Figure 1. FTIR spectra of functionalized (solid lines) and unfunctionalized Fe3O4 25 nm nanoparticles (dotted lines): (a) Fe3O4 (top) and Fe3O4−NH2 (bottom), (b) Fe3O4−DETA (top) and Fe3O4−Me−DETA (bottom), (c) Fe3O4 (top) and Fe3O4−I (bottom), and (d) from top to bottom Fe3O4, Fe3O4−PEI (423), Fe3O4−PEI (600), Fe3O4−PEI (1200), and Fe3O4−PEI (1800). The characteristic peaks at 1560 and 1460 cm−1 were due to the NH2 and CH2 bending modes, respectively. D
dx.doi.org/10.1021/la400848r | Langmuir XXXX, XXX, XXX−XXX
Langmuir
Article
Figure 2. TEM images of (a−e) 25 nm and (f−j) 50 nm Fe3O4 nanoparticles. (a) Fe3O4 25 nm unfunctionalized, (b) Fe3O4−NH2 25 nm, (c) Fe3O4−DETA 25 nm, (d) Fe3O4−Me−DETA 25 nm, (e) Fe3O4−PEI(423) 25 nm, (f) Fe3O4 50 nm unfunctionalized, (g) Fe3O4−NH2 50 nm, (h) Fe3O4−DETA 50 nm; (i) Fe3O4−Me−DETA 50 nm, and (j) Fe3O4−PEI(423) 50 nm.
Figure 3. Powder XRD patterns of functionalized and unfunctionalized Fe3O4 nanoparticles with indexes. (a−e) Diffraction patterns of Fe3O4 nanoparticles of 20 nm. (f−j) Diffraction patterns of Fe3O4 nanoparticles of 50 nm. The core size of the particle estimation with Scherrer analysis from these diffraction patterns is shown in Table S2.
bending at 1560 cm−1; CH2 stretching at 2920 cm−1 and bending at 1460 cm−1). This was observed in all samples grafted with monoamines (1, 2, 3, and 4°). Similar results were observed from Fe3O4 nanoparticles of 50 nm. Bands at at 3390 cm−1 (O−H stretching) and 1625 cm−1 (O−H bending) are due to moisture on the nanoparticle surface. The morphology and core size of the nanoparticles were analyzed using TEM and XRD. The results are shown in Figures 2 and 3, respectively. In Figure 2a, the unfunctionalized nanoparticles show a pseudospherical morphology with a narrow size distribution, and the average size is 24 to 25 nm with a standard deviation of 6 to 7 nm. The silanization process using ATPES did not affect the core size distribution or the morphology (Figure 2b). This is critical because the morphology of a nanoparticle could alter its binding characteristics.23,24 Although some aggregation was shown in these samples, this could be caused by the sample preparation procedure for TEM analysis. Figure 3 shows the XRD pattern of the unfunctionalized and functionalized particles. There is no shifting of the peaks, and the core size estimated with the Scherrer equation (from 22.0 to 25.5 nm) shows no significant change (Table S2 in Supporting Information).
In contrast, particles of increased core size were found in amine-functionalized Fe3O4 nanoparticles of 50 nm. In Figure 2g, nanoparticles with a core size of up to 100 nm were found, together with some smaller particles. This indicates that Fe3O4 nanoparticles of 50 nm have a lower thermal stability, and the grafting process at 110 °C causes both the fragmentation and sintering of particles. In general, the XRD patterns of 50 nm Fe3O4 nanoparticles (Figure 3f− j) showed narrower diffraction peaks compared to those of 25 nm particles as a result of the particles of larger core size. From the Scherrer analysis, Fe3O4− NH2 (50 nm) nanoparticles have an average of 41.4 nm in core size. Scherrer analysis provides an estimation on the volume rather than the average size of particles, so some discrepancy is expected although some change may arise from the fragmentation and formation of smaller particles, which are depicted in the TEM micrographs in Figure 2. Characterization of Oligoamine- and PolyamineFunctionalized Fe3O4 Nanoparticles. Grafting monoamine groups onto the Fe3O4 nanoparticles can create only a 2-D charged surface. To build a 3-D charged environment, diethylenetriamine (DETA) groups were grafted onto the nanoparticles surface (Scheme 1). DETA is a linear oligomer of three ethyleneamine (−CH2−CH2−NH−) units (estimated E
dx.doi.org/10.1021/la400848r | Langmuir XXXX, XXX, XXX−XXX
Langmuir
Article
Figure 4. (a) Titration curves and (b) buffering capacity plot for amine-functionalized Fe3O4 nanoparticles. (a) The titration cures were compared to those of unbound PEI1800. (b) The buffering capacity plots for Fe3O4−NH2 (gray ■), Fe3O4−DETA (▲), Fe3O4−PEI(423) (○), and Fe3O4−PEI (1800) (⧫) were compared to those for unfunctionalized Fe3O4 particles (*).
To increase the thickness of the amine coating on the Fe3O4 nanoparticles further, extending the chain length of the oligoamine groups is a logical strategy, that is, to use shortchain polyethyleneimine (PEI). In the present work, shortchain PEI molecules were chemically bound to the 25 and 50 nm Fe3O4 nanoparticles via reaction with an iodopropyl linker (Scheme 1c). To limit the overall size of the particles and the potential toxicity, short-chain PEIs of Mw = 423 (10 units), 600 (14 units), 1200 (28 units), 1800 (42 units) were chosen for this study. In general, the toxicity of PEI increases with chain length; therefore, it is desirable to keep the Mw of PEI at a minimum for functionalization if the final materials are targeting in vivo applications. In contrast to other binding methods such as amide or imine linkages, the reaction between the iodopropyl units on the particles and the amine groups on the PEI forms higher amines (2 or 3°) via C−N single bonds, which are less vulnerable to hydrolysis than amide or imine linkages. As a result, these PEI-functionalized Fe3O4 nanoparticles should be more stable in physiological environments than conventional amide- or imine-linked PEI magnetic nanoparticles as reported in the literature.35 The use of iodoalkyl groups to react with amine groups, which act as nucleophiles in this alkylation reaction, is essentially the same chemistry as the methylation of DETA employed in the preparation of Me−DETA silanes and the postsilanizated Me− DETA−Fe3O4 nanoparticles in the previous section. PEI polymers are known to be effective agents for DNA binding,41,42 with the binding efficiency increasing when the molecular weight of the PEI increases. Again, the TEM image of Fe3O4−PEI(423) for both 25 and 50 nm nanoparticles (Figure 2e) showed no significant change in the morphology of the nanoparticles. Scherrer analysis of the XRD pattern has shown only a slight increase in average
length = 1 nm) and can also be considered to be a very short chain PEI or intermediate between monoamines and shortchain PEI polymers (from 10 to 42 units), which are used in the next step (Scheme 1c). DETA was thus chosen as a candidate positive charge-bearing molecule that carries some characteristics of both PEI and monoamines. To increase the charge density further, the methylation of DETA was carried out (Scheme 1b). Quantification using CHN elemental analysis showed that the quantity of grafted DETA groups (370 μmol g−1 for 25 nm nanoparticles and 350 μmol g−1 for 50 nm nanoparticles) is very close to that of the monoamine-grafted samples (Table 1). Grafting methylated DETA−silane (Me−DETA) was less successful for both 25 and 50 nm nanoparticles; only 140 and 170 μmol g−1, respectively, were grafted. The smaller quantity of groups grafted is possibly due to the increasing size of groups from methylation. Also, an increased positive charge density may impede further silanization because the Me− DETA silanes are also positively charged. As a result, such electrostatic repulsion may produce a lower density of the positively charged Me−DETA groups grafted onto the surface of the nanoparticles. Nonetheless, the total number of groups on the shell per area (1.8 groups per nm2 for 25 nm per nanoparticle and 4.7 groups per nm2 for 50 nm per nanoparticle) is still comparable to the number of the nanoparticles grafted with monoamine groups. Also, these Fe3O4−DETA and Fe3O4−Me−DETA (both 25 and 50 nm) nanoparticles that are prepared have similar particle morphologies to their NH-grafted counterparts (Figure 2c,d,h,i). Postsynthesis methylation of Fe3O4−DETA was also carried out for comparison. Details of the experimental results can be found in Note S1 in the Supporting Information. F
dx.doi.org/10.1021/la400848r | Langmuir XXXX, XXX, XXX−XXX
Langmuir
Article
from pH 6 to 9, which may suggest that 50 nm Fe3O4−NH2 and Fe3O4−DETA can be more efficient transfecting agents than their PEI-functionalized counterparts. Nonetheless, all amine-functionalized particles have an enhanced buffering capacity when compared to unfunctionalized Fe3O4 nanoparticles, both 25 and 50 nm. Zeta Potential Measurement. To confirm that these amine-grafted nanoparticles carry positive charges, measurement of the zeta potential (or charge density) of samples provides important information for predicting their DNA binding capacity. This is a common practice in developing DNA binding reagents (details about the zeta potential can be found in Note S2).44 Although the zeta potential of a solid varies according to the solution pH, the measurements carried out in this work were in unbuffered deionized water (pH 6 to 6.5). This is because during the DNA binding experiment in the next section DNA has to be dissolved in pure deionized water to avoid competition with anions from buffering reagents such as phosphates and carbonates. Table 2 summarizes the zeta potential measurements of amine-functionalized Fe3O4 nanoparticles. For 25 nm Fe3O4
particle size for both samples (at 26 and 53 nm) possibly as a result of the reaction at higher temperature (130 °C) required for reacting PEI with the propyliodide groups. The quantity of PEI molecules grafted onto the Fe3O4 particles ranged from 10 to 170 μmol g−1 at a density of 0.3−1.4 polymer molecules per nm2 (Table 1). In general, more PEI can be attached to 25 nm Fe3O4 nanoparticles per unit mass than onto the 50 nm particles as a result of the greater surface area. Despite such a small number of molecules grafted in comparison to the monoamine-grafted samples, these PEI-functionalized particles have a large number of amine groups considering that each of the PEI molecules has up to 42 nitrogen atoms for PEI(1800), which means ca. 60 nitrogen atoms per nm2 in the case of Fe3O4−PEI(1800) nanoparticles. Theoretically, the DNA binding capacity of an amine-grafted Fe3O4 nanoparticle sample relies on the availability of the positively charged amine groups for interaction with negatively charged DNA molecules. Therefore, an increase in the quantity of amine groups should enhance the DNA binding capacity of the MNP samples. In general, silanization represents a versatile route for grafting monoamines and oligoamines onto iron oxide nanoparticles to form a monolayer coating and to avoid thick cross-linked polymer layers. By reacting PEI with iodopropyl groups, also grafted via silanization, PEI can also be chemically bound to the surface of iron oxide nanoparticles. However, these groups need to carry positive charges in a neutral environment in order to interact with negatively charged species, including DNA molecules. Buffering Capacity Measurement. To examine the surface charge profile of amine-functionalized nanoparticles under different pH conditions, the buffering capacities of the samples were measured following a standard method from the literature.22 From a potentiometric titration curve using HCl (0.1 M) as a titrant, the buffering capacity (β) of a polycationic species can be defined as β=
Table 2. Table Summarizing the Zeta Potential, Specific DNA Binding Capacity, and N/P Ratio for Amine-Grafted Fe3O4 Nanoparticles (Both 25 and 50 nm)a 25 nm NPs
d[HCl] d pH
where d[HCl] is the change in HCl concentration in the nanoparticle suspension and dpH is the change in pH.43 It has been reported that the buffering capacity of polycationic species at low pH (4.5−7.4), with the characteristic pH for the lysosomal compartment, could play an essential role in nonviral gene delivery on the basis of the proton sponge hypothesis.21 According to this hypothesis, the polycationic species (aminefunctionalized nanoparticles in this case) should be able to buffer the acidic pH of the lysosomal compartment after the uptake of the DNA/nanoparticle complexes. Therefore, the DNA molecules are protected from degradation and also released into the cytoplasm when the vesicles rupture because of osmosis.21 Figure 4a.i,a.ii shows the titration curves for Fe 3 O 4 nanoparticles functionalized with monoamine (NH2), oligoamine (DETA), and polyamine (PEI(423) and PEI(1800)), and Figure 4a.ii,b.ii shows the buffering capacities of these samples. Isoelectric points of samples determined from these titration curves are shown in Table S3. For 25 nm samples (Figure 4a.i), there is little difference in their buffering capacity observed in the region of pH 5−10. However, both the NH2 and DETA-functionalized 50 nm Fe3O4 nanoparticles show a higher buffering capacity than both the Fe3O4−PEI samples
50 nm NPs
N/P ratio
zeta potential (mV)
specific DNA binding capacity
N/P ratio
0.0066 0.0055
22 34
8.8 2.2
0.0046 0.0018
26 45
9.4
0.0062
10
4.2
0.0028
15
15.7
0.0071
16
30.0
0.010
12
21.4 34.3
0.0087 0.024
43 6.0
19.0 73.9
0.013 0.072
28 2.4
32.4
0.055
5.6
11.3
0.035
1.9
35.2
0.025
7.8
18.2
0.018
3.6
21.4
0.209
7.8
22.8
0.270
1.8
18.6
0.746
2.1
21.7
0.221
2.9
Fe3O4 nanoparticle samples
zeta potential (mV)
specific DNA binding capacity
Fe3O4−NH2 Fe3O4− NHCH3 Fe3O4− N(CH3)2 Fe3O4− N(CH3)3+Cl− Fe3O4-DETA Fe3O4−Me− DETA Fe3O4−PEI (423) Fe3O4−PEI (800) Fe3O4−PEI (1200) Fe3O4−PEI (2000)
14.2 7.6
a
The N/P ratio was calculated by assuming that the average molecular weight of a nucleic acid unit is 340 Da.
nanoparticles grafted with monoamine groups, the measured zeta potential follows the number of groups grafted, with the quaternary ammonium (4° or −N(CH3)3+) sample showing the highest zeta potential at +15.7 mV as a result of its permanent positive charges. Both 1 and 2° amine-grafted samples show a lower zeta potential despite a similar number of grafted groups. This indicates that the 1 and 2° amine groups on these samples may not be fully protonated. In general, a zeta potential of +7.6 to +15.7 mV was recorded for the 25 nm Fe3O4 nanoparticles grafted with monoamine groups. For 50 nm nanoparticles, except nanoparticles grafted with quaternary ammonium groups (4° amines), the zeta potentials recorded are lower (from only +2.2 to +8.8 mV) because of the decrease in the number of grafted groups per unit mass and the larger G
dx.doi.org/10.1021/la400848r | Langmuir XXXX, XXX, XXX−XXX
Langmuir
Article
Therefore, unbuffered deionized water was chosen to be used as a medium for DNA binding. Typical binding curves were shown in Figures 5 and S4, and the results, together with the zeta potential and the N/P ratio,
particle size, which in turn reduces the charge density. Despite the lower values, the same trend as for 25 nm samples was observed. For nanoparticles grafted with oligoamines (DETA), a higher zeta potential (+21.4 mV) is recorded when compared to monoamines as a result of an increase in the overall number of amine groups on the nanoparticles. Upon methylation, higher zeta potentials (increased from +21.4 to +34.3 mV for 25 nm nanoparticles and from +19.0 to +73.9 mV for 50 nm nanoparticles) were recorded for both Fe3O4−Me−DETA samples despite a reduced number of groups grafted on the nanoparticle surface (0.14 and 0.17 vs 0.37 and 0.35 mmol g−1). This is due to the enhanced positive charge of the methylated DETA directly grafted onto the surface of the Fe 3 O 4 nanoparticles. This also indicates that the DETA groups are not fully charged. Postsynthesis methylation of Fe3O4−DETA can also increase the zeta potential of the nanoparticles, but this procedure is not as effective as that using methylated DETA− silane for functionalization (Note S1). The zeta potential measurements of PEI-grafted Fe3O4 nanoparticle samples show that all PEI-grafted samples are positively charged. However, the zeta potential did not increase as the chain length of PEI increased. This is because in an environment with a high concentration of amine groups full protonation becomes difficult as a result of the competition between neighboring amine groups. Such an observation is not uncommon and has been reported in free PEI polymers.37 This effect will be more significant for PEI molecules immobilized on a surface because the competition for protons between neighboring polymer chains also occurs. As a result, the zeta potential for Fe3O4−PEI 25 nm particles peaked at PEI(600), which also shows the smallest number of molecules per NP, and started to decrease when the chain length increased. For 50 nm particles, the zeta potential reached a maximum with PEI(1200) because of the larger surface area per particle and the greater space between polymer branches. From the data in Table 1, the number of PEI molecules grafted onto 50 nm particles per unit mass is around 3 times fewer than that on the 25 nm particles on the basis of the CHN elemental analysis. Therefore, the competition for protons between amine groups is less severe as a result of the increased distance between amine groups from neighboring polymer branches. This observation suggested that, when grafting PEI polymers, the zeta potential does not always increase when the chain length increases. Nonetheless, grafting PEI onto Fe3O4 nanoparticles by reacting with iodopropyl-functionalized nanoparticles successfully creates a 3-D coating with positive charges. DNA Binding Capacity Measurement and Its Relation to the Zeta Potential. The DNA binding capacity of nanoparticles was measured following the method reported in the literature.45 There are numerous methods to measure the DNA binding capacity available.46,47 The data presented here were from the measurement carried out in unbuffered, deionized water because the large number of anions (e.g., Cl−, CO32−, and PO43−) present in the buffer solution will compete with the negatively charged DNA molecules for positively charged sites on the nanoparticle surface, which will induce a considerable error in the measurement. Indeed, a solution of high NaCl concentration can be used to remove bound DNA from the surface of solids.48 Because our aim is to study the intrinsic DNA binding property of the functionalized nanoparticles, competition with anions has to be minimized.
Figure 5. DNA binding curves for the determination of the DNA binding capacity (from the fitted slope) for Fe3O4−PEI(423) 50 nm (◊), Fe3O4−PEI(600) 50 nm (■), Fe3O4−PEI(1200) 50 nm (▲), and Fe3O4−PEI(2000) 50 nm (□) nanoparticles.
are summarized in Table 2. Amine-functionalized nanoparticles are known to be capable of binding DNA through electrostatic interaction,13 and in theory, the higher the zeta potential, the more DNA that will be bound to the nanoparticle surface. To visualize this relationship, a plot of the DNA binding capacity of amine-grafted Fe3O4 nanoparticles against their zeta potential is shown in Figure 6a. In general, there is a correlation between the zeta potential and the DNA binding capacity, in particular, for nanoparticles grafted with monoamines and DETA moieties (i.e., for particles with an essentially 2-D array of charges at the NP surface). On monoamine grafted samples, all amine groups are located at a level of 0.4 nm (three −CH2− groups and one −Si−O−) from the Fe3O4 surface and form a monolayer of positive charges, as illustrated in Figure S1a. However, DETAgrafted nanoparticles should have positive charges on multiple layers. Nonetheless, this did not improve the DNA binding significantly because of the small footprint size (ca. 0.2 nm2) that fails to accommodate a large molecule such as DNA, and the charges on the lower levels are unlikely to have direct interaction with DNA molecules. Indeed, the width of a doublestranded DNA molecule is about 20 Å (or 2 nm), whereas the distances between amine groups in monoamine grafted samples are from 7 to 12 Å. Therefore, the DETA-grafted samples behave as a 2-D positively charged array, similar to the monoamine grafted samples (Figure S1b). A low DNA binding capacity (from 1.8 to 13 μg of DNA per mg of nanoparticles) was recorded for all of these samples. However, Fe3O4−Me− DETA samples (both 25 and 50 nm, with a permanently positive charged surface) showed a significant increase in the DNA binding capacity, following the trend shown in Figure 6a. This is likely to be the combined effect of the higher zeta potential and the larger footprint size of the grafted groups. However, the DNA binding capacity has been enhanced when the chain length of the polyamine increases, in cases of PEI-coated MNPs. The 3-D coating (estimated to be >3.5 nm in thickness, see Figure S1) dramatically increases the specific DNA binding capacity to a value of as high as 0.746. In Figure 6a, the DNA binding capacity of both 25 and 50 nm Fe3O4− H
dx.doi.org/10.1021/la400848r | Langmuir XXXX, XXX, XXX−XXX
Langmuir
Article
Figure 6. (a) Plot of DNA binding capacity vs zeta potential, with triangles for 25 nm nanoparticles and diamonds for 50 nm nanoparticles. Filled markers (black) represent PEI samples whereas open markers (white) are used for the Fe3O4−PEI samples. (b) Data points for Fe3O4−PEI samples, 25 and 50 nm, are compared with Fe3O4−DETA samples. (c) Plot of N/P ratio vs zeta potential and (d) plot of DNA binding capacity vs N/P ratio for all samples (filled markers for Fe3O4−PEI samples and unfilled markers for the rest). Generally, a low N/P ratio indicates a higher binding efficiency in term of, on average, how many amine groups are interacting with one phosphate group on the DNA. The specific DNA binding capacity shows only the amount of DNA bound (in mg) to 1 mg of nanoparticles.
PEI(423) and Fe3O4−PEI(600) shows a significant increase and deviates upward from the relationship seen from the monoamine groups, that is, more DNA is bound than expected from the zeta potential values in the case of the 3-D polymer coatings. When the chain length increases further (Figure 6b), the values of both Fe3O4−PEI(1200) and Fe3O4−PEI(1800), 25 and 50 nm, respectively, show a further 4- to 13-fold increase from the corresponding Fe3O4−PEI(423) MNPs (Table 2). For these samples, the zeta potentials play a less influential role in DNA binding, suggesting that specific stereochemical features, identified here as the 3-D nature of the coating, are more relevant to the DNA binding capacity. The efficiency of the DNA binding capacity can also be studied by examining the N/P ratio of particle−DNA conjugates. This is a measurement of how many nitrogen atoms (positively charged sites) from grafted PEI bind to the phosphate groups (negatively charged sites) on the DNA. In an ideal situation, one amine group (+1 charge) on PEI binds to one phosphate group (−1 charge) on DNA, and N/P = 1 will be observed. In the literature regarding PEI for gene delivery, the N/P ratio varies from 4.5 to 135 with maximum levels of delivery observed at volumes of 9 to 13.5.42 However, among the nanoparticles grafted with monoamines, the DNA binding capacity value remains low and the binding efficiency, indicated by the N/P ratio, ranges from only 10 to 45, compared to a value of 5 to 6 for free PEI polymers. This suggests that the majority of the amine groups are not interacting with the phosphate groups on the DNA backbone. Indeed, free PEI polymers have the flexibility to wrap around DNA molecules whereas these monoamine groups are immobilized on a rigid 2D surface. The efficiency of binding, suggested by a high N/P ratio, has not been improved using DETA-grafted Fe3O4
nanoparticles, which is consistent with the previous study.30 In Figure 6c, a plot of N/P ratio against zeta potential shows that all PEI-grafted samples are clustered in the region of low N/P ratio (
