ARTICLE pubs.acs.org/Langmuir
Chemical Grafting of a DNA Intercalator Probe onto Functional Iron Oxide Nanoparticles: A Physicochemical Study Laurent Bouffier, Humphrey H. P. Yiu, and Matthew J. Rosseinsky* Department of Chemistry, University of Liverpool, Liverpool, L69 7ZD, United Kingdom
bS Supporting Information ABSTRACT: Spherical magnetite nanoparticles (MNPs, ∼ 24 nm in diameter) were sequentially functionalized with trimethoxysilylpropyldiethylenetriamine (TMSPDT) and a synthetic DNA intercalator, namely, 9-chloro-4H-pyrido[4,3,2-kl]acridin-4-one (PyAcr), in order to promote DNA interaction. The designed synthetic pathway allowed control of the chemical grafting efficiency to access MNPs either partially or fully functionalized with the intercalator moiety. The newly prepared nanomaterials were characterized by a range of physicochemical techniques: FTIR, TEM, PXRD, and TGA. The data were consistent with a full surface coverage by immobilized silylpropyldiethylenetriamine (SPDT) molecules, which corresponds to ∼22 300 SPDT molecules per MNP and a subsequent (4740 2940) PyAcr after the chemical grafting step (i.e., ∼ 2.4 PyAcr/nm2). A greater amount of PyAcr (30 600) was immobilized by the alternative strategy of binding a fully prefunctionalized shell to the MNPs with up to 16.1 PyAcr/nm2. We found that the extent of PyAcr functionalization strongly affects the resulting properties and, particularly, the colloidal stability as well as the surface charge estimated by ζ-potential measurement. The intercalator grafting generates a negative charge contribution which counterbalances the positive charge of the single SPDT shell. The DNA binding capability was measured by titration assay and increases from 15 to 21.5 μg of DNA per mg of MNPs after PyAcr grafting (1420% yield) but then drops to only ∼2 μg for the fully functionalized MNPs. This highlights that even if the size of the MNPs is obviously a determining factor to promote surface DNA interaction, it is not the only limiting parameter, as the mode of binding and the interfacial charge density are essential to improve loading capability.
’ INTRODUCTION Magnetic nanoparticles (MNPs) have attracted considerable attention in bioscience research thanks to their wide range of potential biological applications, including cell separation, DNA purification, immunoassays, tumor targeting, and magnetic resonance imaging.16 From a chemical point of view, the major challenge in nanomaterials is to control the nature of the interface between the surface functional groups and the surrounding environment.7,8 An increase in control of the level of functionality of these materials could achieve a better tuning of the resulting properties and therefore more efficient applications. Synthetic preparation of MNPs involves high-temperature methods delivering highly crystalline, highly magnetic materials, i.e., solvothermal decomposition of inorganic precursors in high boiling solvents (like dibenzyl ether and relatives), in the presence of structural directing/surface capping agents. The current state of the art of nanoparticle properties suggests that control of capping agent identity, varying boiling point with respect to the solvent, binding chemistry, molar ratio to transition metal source, and concentration allow a high level of control over size, size distribution, shape, and magnetization. In that context, iron(II,III) oxide (Fe3O4) MNPs, which exhibit superparamagnetic properties when particles have a single domain size smaller than 30 nm (i.e., superparamagneticferromagnetic transition),9 are functionalizable scaffolds to promote interaction r 2011 American Chemical Society
with biomolecules.10 More recently, the shape and morphology of MNPs, which could be responsible for surface anisotropy, were also found to significantly influence the magnetic properties.11,12 Surface functionalization of nanomaterials is a key step to design the interface with both solvents and biomolecules and thus to improve biological and physicochemical properties (biocompatibility, nature and number of immobilized molecules/chemical groups, charge surface, shell degradation, and chemical stability).13 Recent advances in the surface chemistry of MNPs have enhanced the colloidal properties14,15 and generated delivery transporters or scaffolds with a tunable outer shell designed for interaction with biomolecules such as DNA, RNA, and peptides as well as proteins.16 The most obvious and common strategy to promote DNA interaction is labeling MNPs with positively charged organic molecules, which increase the electrostatic affinity for the phosphate backbone of DNA (which is negatively charged). Among the variety of possible surface anchoring groups, silanes are very promising candidates.17,18 They offer versatile functional diversity and wide availability, and more importantly, their good reactivity for grafting onto MNP surfaces as well as strong stability have been reported.15,1921 Received: November 29, 2010 Revised: March 14, 2011 Published: April 13, 2011 6185
dx.doi.org/10.1021/la104745x | Langmuir 2011, 27, 6185–6192
Langmuir In 2009, Mirkin et al. discussed the relationship between the size (radius of curvature) and the potential DNA loading in short thiolated oligonucleotide immobilized on spherical gold nanoparticles.22 These authors found that the oligonucleotide surface coverage on NPs can be expressed as a mathematical function of curvature, shape, and available surface area. This finding highlights “how size matters” and limits the DNA loading with respect to the geometry of the nanoparticles. In the present study, we investigate the size limitation in the case of spherical Fe3O4 nanoparticles by showing how a change in mode of binding affects the DNA loading capability of the MNPs. This was successfully achieved by the design of multifunctional NPs whose surface has been modified by polyamine functionalities to promote nonspecific electrostatic interaction as well as a chemically grafted synthetic DNA intercalator which specifically targets double-stranded DNA. It is generally admitted that the size of the nanoparticle or nanomaterial is the limiting factor of the potential DNA loading, but our results show that the mode of binding is at least as important.
’ EXPERIMENTAL SECTION Materials. 6,9-Dichloro-2-methoxyacridine (g98%) and iron(II, III) oxide (Nanopowder, 20 nm, 98%) were purchased from Alfa Aesar. Aminoacetaldehyde dimethyl acetal (g99%), ammonium cerium(IV) nitrate (g98.5%), and methanesulfonic acid (g99.5%) were purchased from Sigma-Aldrich. Trimethoxysilylpropyldiethylenetriamine (TMSPDT) was purchased from Fluorochem. All chemicals were used as received without any further purification. Calf thymus DNA was purchased from Sigma (activated, type XV). All aqueous solutions were prepared with ultrapure water from a Milli-Q system (Millipore, resistivity 18.2 MΩ cm at 25 °C). 9-Chloro-4H-pyrido[4,3,2-kl] acridin-4-one (PyAcr) was prepared according to literature procedures as illustrated in Scheme S1.1 in the Supporting Information.23,24 Proton NMR spectroscopy (250 MHz, CDCl3) δ = 9.45 (1H, d, J = 4.50 Hz, H-2), 9.08 (1H, d, J = 8,75 Hz, H-11), 8.27 (1H, d, J = 2.00 Hz, H-8), 8.23 (1H, d, J = 4.50 Hz, H-3), 7.91 (1H, d, J = 10.25 Hz, H-5), 7.75 (1H, dd, J = 8.75 and 2.00 Hz, H-10), 7.00 (1H, d, J = 10.25 Hz, H-6) ppm. Silylation of Fe3O4 Nanoparticles. In a typical experiment, 0.4 g of commercially available Fe3O4 nanoparticles was dispersed in dry toluene (40 mL) using an ultrasonic bath (Fisher Scientific, model FB15051, frequency = 3040 kHz) for 15 min under normal atmosphere. TMSPDT (1.0 mL, 4 mmol) was added to the Fe3O4 suspension and the reaction mixture was refluxed (115 °C) for 24 h under mechanical stirring. The solid was recovered using a rare earth NdFeB magnet (Magnet Sales; size, 10 mm D 5 mm H; strength ∼1.18 T) and washed three times with ethanol. The nanoparticles [Fe3O4SPDT (silylpropyldiethylenetriamine)] were then dried overnight at 80 °C. Intercalator Chemical Grafting. PyAcr (25 mg, 0.094 mmol) and Fe3O4SPDT (100 mg) were suspended in 40 mL of ethanol. The resulting mixture was heated at 80 °C under mechanical stirring for 5 days. The solid was recovered using an external magnet (same specifications) and washed three times with ethanol. The nanoparticles (Fe3O4SPDTPyAcr) were dried overnight at 80 °C. Chemical Functionalization of TMSPDT and Subsequent Silylation. A stoichiometric solution of PyAcr (100 mg, 0.376 mmol) and TMSPDT (0.101 mg, 0.376 mmol) in ethanol (50 mL) was stirred at 60 °C for 3 days. Then, Fe3O4 nanoparticles (100 mg) were added to the solution, and the mixture was mechanically stirred at 80 °C for 3 more days. The solid was recovered using an external magnet (same specifications) and washed three times with ethanol. The nanoparticles (Fe3O4PyAcr) were dried overnight at 80 °C.
ARTICLE
Scheme 1. Iron Oxide MNP Functionalizationa
a
(i) TMSPDT, toluene, reflux, 24 h and (ii) PyAcr, ethanol, reflux, 5 days affords particles referred to as Fe3O4SPDTPyAcr; (iii) TMSPDTPyAcr, ethanol, reflux, 3 days affords particles referred to as Fe3O4PyAcr.
Characterization. Transmission Electron Microscopy. TEM images were collected on a Tecnai electron microscope operating at an accelerating voltage of 100 kV. X-ray Powder Diffraction. X-ray powder diffraction data were collected with Co KR1 radiation (λ = 1.781 Å) on a Panalytical X’pert Pro multipurpose diffractometer. 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 cm1 were accumulated with a resolution of 4 cm1. Thermal Gravimetric Analysis. TGA was performed on a TA Instruments Q600 under air atmosphere. A 510 mg portion of nanoparticles was heated to 90 °C at 10 °C/min and kept at 90 °C for 30 min to remove all adsorbed solvent. The sample was then heated to 900 °C at 20 °C/min and kept at 900 °C for 30 min to determine the amount of of organic coating on the nanoparticle surface. CHN Elemental Anaylsis. The carbon content of the samples was analyzed with a Thermo EA1112 Flash CHNS-O Analyzer. The carbon content (C %) cited was an average of four measurements. Electrophoretic Mobility (ζ-Potential). ζ-Potential measurements were recorded on a Malvern Nanosizer with MNPs samples in solution (deionized unbuffered water). Colloidal Stability Assay (UV Spectroscopy). Absorbance of MNPs dispersed in water (0.5 mg/mL) was measured every 10 s for 10 h at a fixed wavelength of 850 nm. DNA Binding Capacity Measurement. DNA (from 50 to 300 μg) was mixed with 50 μg of nanoparticles in ultrapure water in a 1.5 mL Eppendorf microcentrifuge tube. The total volume of the suspension was kept at 1.0 mL. The suspensions were incubated overnight at room temperature under rotation (20 rpm). After high-speed centrifugations (13 000 rpm for 5 min, three times), the amount of unbound DNA in the supernatant was measured by UVvis spectroscopy at λ = 260 nm. The amount of DNA bound onto the nanoparticles 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 as well as the standard deviation were plotted as the DNA binding curve.
’ RESULTS AND DISCUSSION Design of the MNPs. The synthetic strategy is based on a twostep functionalization of commercially available spherical Fe3O4 6186
dx.doi.org/10.1021/la104745x |Langmuir 2011, 27, 6185–6192
Langmuir
ARTICLE
Figure 1. ATR FTIR spectra recorded between 2000 and 750 cm1 of magnetic nanoparticles: Fe3O4 (a), PyAcr (b), Fe3O4SPDT (c/i, black line), Fe3O4SPDTPyAcr (c/ii, gray line), and Fe3O4PyAcr (d). The corresponding full spectra (4000650 cm1) are given in the Supporting Information (S2).
MNPs (Scheme 1). Silylation with TMSPDT was achieved to form a layer of organic coating covalently linked to the iron oxide core (Scheme 1i). The prefunctionalized iron oxide MNPs (Fe3O4SPDT) which exhibit a polyaminosilane shell bearing a terminal primary amine are then postfunctionalized by the chemical grafting of pyrido[4,3,2-kl]acridin-4-one (Scheme 1ii). Such regioselective nucleophilic addition was first developed for homogeneous functionalization in solution.24,25 Nevertheless, covalent immobilization of a PyAcr moiety has already been used successfully in heterogeneous conditions (liquid/solid interface) for the preparation of modified Pt disk electrode.26 It is noteworthy that the choice of the DNA intercalator probe in this work is based on the two following criteria: simple synthetic access and high binding affinity for double-stranded DNA (K = 2.8 105 mol1).24,27 As illustrated in Scheme 1, the efficiency of that first strategy is limited by the yield of the chemical grafting, and it is very unlikely that all the available amino groups will react with the intercalator moiety. This nonquantitative functionalization could be explained on the basis of steric hindrance, heterogeneous phase conditions (solid/liquid), and partial protonation of the terminal amines, which limit the reactivity. Therefore, the inverse sequential strategy, which consists of prefunctionalizing TMSPDT with PyAcr (Supporting Information Scheme S1.2) and subsequent anchoring onto the MNPs, was also performed (Scheme 1iii) in order to achieve a full functionalization of the surface of the NPs by the DNA intercalator probe.
Physicochemical Characterization. The functionalization of the MNPs was qualitatively monitored by infrared spectroscopy (FTIR). The spectrum recorded for the starting material (Fe3O4, Supporting Information, Figure S2.2) exhibits a very large band at 3390 cm1, which has been assigned to the OH bond stretching,28 and two smaller bands centered at 1626 (H2O) and 1078 cm1, respectively. These signals confirm that the surface consists of absorbed water molecules as well as surface OH groups, which could be used for grafting organic molecules.29 Compared to the unfunctionalized MNPs, several new bands are observed in the spectrum of silane-modified Fe3O4SPDT MNPs (Figure 1c/i and Supporting Information Figure S2.3). The OH vibration intensity decreases strongly and the large band at ∼31503400 cm1 is due to the contribution of NH2 vibrations, whereas new bands at 2914 and 1457 cm1 can be assigned to CH2 stretching and bending, respectively. The observed value for ν CH2 implicates that the silane layer deposited on the ferrite MNPs does not exhibit high crystallinity.15 Although there was a large band at 1078 cm1 in the initial MNPs, new bands at 1034 and 1206 cm1 are characteristic of the silane layer and originated from SiOSi and SiC vibrations, respectively.30 All these data together confirm unambiguously the successful silane ligand anchoring. Finally, the chemical grafting of the DNA intercalator was evidenced in the spectra of Fe3O4SPDTPyAcr (Figure 1c/ii) 6187
dx.doi.org/10.1021/la104745x |Langmuir 2011, 27, 6185–6192
Langmuir
ARTICLE
Figure 3. From top to bottom, thermograms of Fe3O4, Fe3O4SPDT, Fe3O4SPDTPyAcr, and Fe3O4PyAcr. The normalized weight loss for Fe3O4SPDT, Fe3O4SPDTPyAcr, and Fe3O4PyAcr are 11.41%, 15.53%, and 32.46%, respectively.
Figure 2. TEM images of the magnetic nanoparticles: Fe3O4 (top left), Fe3O4SPDT (top right), Fe3O4SPDTPyAcr (bottom left), and Fe3O4PyAcr (bottom right). Scale bars represent 200 nm. Insets show the corresponding size distributions (in nm) calculated from about 400 nanoparticles.
Table 1. Mean Size of MNPs Calculated from PXRD Data and TEM Images mean size/nm MNPs
PXRD
TEM 24.3 ( 5.7
Fe3O4
24.0 ( 1.1
Fe3O4SPDT
23.7 ( 2.2
24.8 ( 6.4
Fe3O4SPDTPyAcr Fe3O4PyAcr
24.4 ( 0.6 24.8 ( 0.8
24.7 ( 5.4 23.9 ( 4.8
and to a larger extent for Fe3O4PyAcr (Figure 1d and Supporting Information Figures S2.3 and S2.4). Typical new bands appear at 1657, 1584, and 826 cm1, which correspond to the fingerprint region of the spectra of PyAcr (1655, 1584, and 828 cm1, respectively, as shown in Figure S2.1 in the Supporting Information). Moreover, the CH2 stretching band at 2914 cm1 indicates that the presence of the planar heteroaromatic molecule does not change the crystallinity of the corresponding silane layer. At this point, it was essential to check that the average size of the nanoparticles and the monodispersity are not affected by the chemical modification of the nanomaterials. The average size of the MNPs was evaluated after each step of functionalization by two different techniques. First of all, TEM images were recorded (Figure 2) and did not show any significant change in the morphology of the core Fe3O4 MNPs. Several supplementary TEM images (Supporting Information S3) showed that homogeneity and monodispersity do not seem to be affected by surface functionalization. It is noteworthy that all particles exhibit spherical shape, which is an important criterion for optimal magnetic properties, as shape anisotropy does strongly influence saturation magnetization value in ferrite materials.12
Moreover, TEM data allows determination of the size distribution of each batch of MNPs. In each case the mean size was calculated from about 400 MNPs, and as a result, no significant difference was observed between all four MNPs samples (Table 1). The average TEM size value is 24.4 nm with a standard deviation of about 20%. The size distribution was also evaluated by powder X-ray diffraction (Supporting Information S4). The data were collected with Co KR1 radiation and the width of the diffraction peaks were used to calculate the minimum crystalline size according to the Scherrer equation for particles of spherical shape.31 The crystallite average sizes are very consistent with the dimensions calculated from TEM data (Table 1), with a mean value varying between 23.7 and 24.8 nm. The standard deviation for PXRD size given in Table 1 was calculated with all the peaks appearing in the diffractograms except reflection [111]. One can note that the relative intensity of reflection [111] in the PXRD diffractograms is only 5.2% in comparison to the main [311] reflection (Supporting Information S4). No accurate evaluation of the width of the peak at half-maximum intensity was possible because of that low intensity. Generally, authors only use the most intense reflection to determine the particle size by Scherrer analysis. Here the six most intense reflections below 2θ = 80° are used, and a standard deviation of the resulting sizes is quoted. The functionalized MNPs have also been characterized by quantitative techniques in order to estimate the number of ligands covalently bound on the surface of the Fe3O4 materials and therefore the efficiency of the functionalization. TGA is a classical thermal analysis technique for characterization of the coating of organicinorganic nanocomposites. The measurement of the change of sample mass with the change of temperature allows the deconvolution of inorganic and organic content. However, it has only been applied recently for characterizing functional ferrite MNPs.15 The TGA thermogram of Fe3O4SPDT (Figure 3) shows a regular mass loss between 260 and 480 °C. This process takes place at slightly higher temperature than the desorption pattern reported for oleic acid ligands. For the latter, a primary mass loss at 240/261 °C followed by a second mass loss peak at 378/391 °C has been reported, depending on the size of the Fe3O4 MNPs.32,33 Moreover, these values are comparable to the 6188
dx.doi.org/10.1021/la104745x |Langmuir 2011, 27, 6185–6192
Langmuir
ARTICLE
Table 2. Surface Coverage of Functionalized Fe3O4 Nanoparticles from TGA and CHN Analysisa mole ratio of groups/Fe3O4 C % from CHN elemental analysis Fe3O4SPDT Fe3O4SPDTPyAcr Fe3O4PyAcr
6.0 7.7 19.8
TGA 0.21 0.041 0.27
no. of groups per NP
CHN
TGA
CHN
0.19
23.5 10
21.1 103
0.026
4.74 10
2.94 103
0.27
31.1 10
30.0 103
3 3 3
a
The carbon content (C %) from CHN analysis is an average of four measurements for each sample. The mole ratio of groups per Fe3O4 and the number of groups per NP were calculated from TGA and CHN analysis. These data show that both analytical results are consistent.
temperatures reported for silane ligand layers on cobalt ferrite MNPs (T = 360490 °C, depending on the type of silane).15 These higher values in comparison to oleic acid indicate the stronger covalent nature of the FeOSi linkage.32 The TGA thermogram of Fe3O4SPDTPyAcr (Figure 3) shows an almost equivalent pattern with a well-defined slope change at ∼450 °C and a larger weight loss value (17.49% against 13.37%). These data could be directly assigned to the presence of the pyridoacridine moiety, which increases the organic weight contribution (i.e., carbonization due to the aromatics). TGA can also be used to quantify the coverage of ligand molecules surrounding the MNPs.15 The mass loss of Fe3O4SPDT recorded by TGA is 13.37% between 90 and 900 °C (Figure 3), which leads to 23.5 103 SPDT molecules per NP (or 0.79 mmol of SPDT groups per gram of NPs). From that value, one can calculate the number of molecules per square centimeter (12.0 SPDT/nm2), which is in good agreement with other values reported from 2.9 to 28.9 ligands/nm2 in the case of silane ligands.15,33 This rather high surface coverage observed after silane functionalization indicates the formation of a dense silane layer, and De Palma et al. postulated the formation of polymerized silane bi- or multilayer which could correlate with the secondary mass loss occurring in the TGA.15 From the CHN elemental analysis, 6.0% carbon was recorded for the Fe3O4SPDT sample (see Table 2). This leads to 21.1 103 SPDT groups per NP (or 0.71 mmol of SPDT groups per gram of NPs). The results from these two analyses are very consistent. Using the theory from Mirkin et al.,22 the footprint of one single SPDT group is also calculated to be 8.3 Å2 from the TGA data and 9.3 Å2 from the CHN results. When we consider the tetrahedral (FeO)3SiR group, the footprint was calculated to be 7.3 Å2. This calculation suggested that the surface of the Fe3O4 nanoparticles was saturated with the SPDT groups. Indeed, the silylation occurs when the TMSPDT reacts with the FeOH groups on the surface of Fe3O4 nanoparticles and forms FeOSi bonds. However, the footprint calculation is based on all three Fe atoms at perfect positions to form a tetrahedral. This may not always happen, and some SPDT groups may only have one or two FeOSi bonds due to incomplete reaction, and residual MeOSi bonds remain instead. Also, there is the possibility of the formation of SiOSi bonds due to two adjacent silane molecules reacting with each other. The deconvolution of the TGA data of Fe3O4SPDTPyAcr (17.49% 13.37% = 4.12% between 90 and 900 °C) allows estimation of the number of chemically grafted pyridoacridine moieties as 4.7 103 per NP (2.4 PyAcr/nm2), which is consistent with CHN analysis (2.9 103 groups per NP). The low conversion values of 14 or 20% of postfunctionalization could be explained by (1) a low accessibility of the terminal amine groups of the SPDT in a multilayer environment and (2) a
high proportion of protonated (and therefore non-nucleophilic) amine groups (NH3þ) in the grafting solution, which are likely to limit the yield of the nucleophilic addition onto the electrondeficient PyAcr. Moreover, increasing steric hindrance will prevent a full functionalization of the amine groups. It is noteworthy that the MNPs are not made larger after the intercalators chemical grafting step. Indeed, the dimension of the synthetic DNA binder calculated by molecular modeling is negligible in comparison with the surface area of a spherical NP of 24 nm in diameter (see Supporting Information S5 for details). Finally, the mass loss of Fe3O4PyAcr recorded by TGA (34.42% between 90 and 900 °C) leads to about 31.1 103 SPDTPyAcr molecules per NP. Therefore, the number of molecules per square centimeter (16.1 SPDTPyAcr/nm2) was found to be slightly higher than the case of Fe3O4SPDT. Similar surface functionalization was calculated from the CHN analysis with 30.0 103 SPDTPyAcr groups per NP. The footprint size calculated from these values is 6.3 and 6.5 Å2 from TGA and CHN analyses, respectively. This is marginally smaller than the footprint size (7.3 Å2) calculated from a tetrahedron. This may be attributed to the interaction between the aromatic PyAcr moieties at the surface of the nanoparticles. However, the structure of a shell fully functionalized with DNA intercalator molecules could be thicker, because the corresponding two-step synthetic methodology allows silane polymerization versus shell formation to compete. From the TEM images in Figure 2, the structure of the SPDTPyAcr shell seems to be quite different, with a larger proportion of polymerization in the Fe3O4PyAcr case, and this is an indication of a multilayer coating on the surface of the nanoparticles. The TGA experiments were carried out in flowing air, although oxidation of Fe3O4 cores occurs and R-Fe2O3 forms. When the atmosphere of TGA experiments changed to flowing nitrogen, higher weight loss was observed from all functionalized samples. This is due to the reduction of Fe3O4 cores to form FeO. It is likely to be the result of the formation of coke from the organic groups, and this reduces the Fe3O4 core at a high temperature. However, the extent of reduction is difficult to quantify, since this does not occur for the unfunctionalized Fe3O4 nanoparticles. In the case of oxidation, since the same reaction occurs for both functionalized and unfunctionalized Fe3O4 nanoparticles, a normalized weight loss can be calculated, thereby yielding the composition of the effective coating. Each batch of MNPs was evaluated via the measurement of the electrokinetic potential (ζ-potential, given in mV), which is the electric difference of potential between the interfacial double layer around the dispersed NP and a point in the bulk solution away from the interface. ζ can be related to (1) the stability of colloidal dispersions by indicating the degree of repulsion between the suspended MNPs and (2) the magnitude of the electric surface potential at the double layer interface. 6189
dx.doi.org/10.1021/la104745x |Langmuir 2011, 27, 6185–6192
Langmuir
ARTICLE
colloidal stability (t1/2 = 1 h), whereas Fe3O4SPDTPyAcr and Fe3O4PyAcr nanoparticles have very similar half-life values (35 and 40 min, respectively). For comparison, unfunctionalized Fe3O4 MNPs has the lowest colloidal stability (t1/2 < 0.5 h). These values are very consistent with the ζ-potential measures, which confirm that surface charge does determine the colloidal stability. DNA binding assays were performed in order to address the influence of the functional organic coating on the Fe3O4 MNPs. For that, we used a calf thymus DNA of 1015 million Da, which is primarily double stranded. The amount of double-stranded DNA interacting with the MNPs was quantified by using electronic absorption spectroscopy measurements at 260 nm by difference with a reference without any MNPs. The corresponding titration curves are presented in Figure 4, together with DNA binding capability extracted from these curves and gathered in Table 3. These values can be compared to some recent data published by Matsunaga et al.37 The authors reported 1.36 μg of captured DNA per mg of aminosilane-modified magnetic MNPs (single crystals of magnetite with cuboctahedral morphology and an average 80 nm in diameter). In that study, the number of functional groups immobilized onto the surface of the MNPs was estimated at 1.1 nm2 (footprint = 91 Å2). The larger DNA binding capability measured in our study (15 μg per mg of Fe3O4SPDT MNPs) could be explained by (1) the smaller MNPs used (average 24.3 nm in diameter), which exhibit greater surface area per gram; (2) the longer chain DNA used, which favored the clustering of several MNPs around the DNA (multimeric/synergetic adsorption); and (3) the larger number of ligands in the present case, as Matsunaga et al. reported ∼1.1 105 amino group/particle for the 80 nm Fe3O4.38 The postfunctionalization with pyridoacridine further improves the DNA binding capability by ∼43% (21.5 μg per mg of Fe3O4SPDTPyAcr MNPs) and this increase could be explained on the basis of the nonreversible adsorption model between the DNA and the PyAcr intercalator. Indeed, Fe3O4SPDT MNPs only bind DNA by electrostatic interactions between the positively charge polyamino shell and the negatively charge DNA phosphate backbone. This interaction is reversible and nonspecific. On the other hand, Fe3O4 SPDTPyAcr MNPs exhibit a more complex mode of binding due to the presence of the PyAcr moiety, which strongly targets double-stranded DNA (specific interaction) and minimizes the reversibility of the DNA binding. Therefore, this 40% increase in DNA binding capability shows that the size of the nanoparticles does not solely limit the DNA binding capacity. Finally, the full functionalization with the DNA intercalator (Fe3O4PyAcr MNPs) dramatically reduces the DNA loading capacity to a value as low as 2.2 μg, which corresponds roughly to a 90% decrease. As mentioned previously, the synthetic pathway to prepare Fe3O4PyAcr MNPs leads to thicker silica shell
The initial value of unfunctionalized Fe3O4 MNPs dispersed in H2O (6.2 mV) is fairly small and reflects a low-charged surface, which correlates with the presence of OH groups only at the surface as well as a visual tendency to aggregation of the MNPs. After functionalization with SPDT, the ζ value increases to 23.8 mV. This value indicates a better colloidal stability and confirms the successful functionalization of the MNPs as the polyamine ligand SPDT is protonated in water at pH e9.34,35 Subsequent chemical grafting with PyAcr (Fe3O4SPDTPyAcr) results in a marked decrease in ζ-potential (4.9 mV), which is in good agreement with the electronic nature of the pyridoacridine. In fact, the large conjugated polyaromatic system is electron rich, and therefore, the PyAcr contribution compensates the positive charge of the protonated polyamino shell in the overall surface charge. Indeed, this decrease can be explained by two arguments: First, the pyridoacridine moiety is only partially protonated in H2O, as the pKa of nitrogen N-7 was estimated to be 7.0 according to the method of Schwarzenbach (acidbase titration using UVvis absorption).36 Therefore, the chemical grafting of the DNA intercalator does clearly affect the level of protonation of the polyamino shell. Furthermore, the MNPs repulsion is greater because the hydrophobicity of PyAcr molecules compensates the electrostatic contribution, which originates from the SPDT shell. The colloidal stability of MNPs has been evaluated by an electronic absorption assay as reported previously by others.15 When monitoring the time-evolution of the absorbance of suspensions of MNPs at a fixed wavelength of 850 nm, the main contribution is due to UV light scattering and therefore correlates with the amount of nanoparticles in suspension. The results are strongly concentration dependent but from the decay curves at a given concentration (0.5 mg/mL), one can calculate the corresponding half-life (t1/2), which gave reasonable quantitative information (see Table 2). Fe3O4SPDT MNPs have the best
Figure 4. DNA titration curves obtained with Fe3O4SPDT (black line) and Fe3O4SPDTPyAcr MNPs (gray line).
Table 3. ζ-Potential Measurement, Colloidal Stability, and DNA Binding Dataa ζ /mV
t1/2/min
DNA bindingb
MNPs per DNA
þ6.2 ( 0.3
28
0
þ23.8 ( 1.1
60
15
∼30.6
Fe3O4SPDTPyAcr
þ4.9 ( 0.3
35
21.5
∼21.4
Fe3O4PyAcr
6.8 ( 0.4
40
e2.2
Fe3O4 Fe3O4SPDT
ζ in mV and half-life (t1/2) in minutes estimated by an electronic absorption assay (i.e., colloidal stability; see Supporting Information Figure S6). b DNA binding capability expressed in μg of DNA per mg of MNPs and estimation of the number of MNPs per molecule of DNA. a
6190
dx.doi.org/10.1021/la104745x |Langmuir 2011, 27, 6185–6192
Langmuir
ARTICLE
Figure 5. Schematic comparison of the size of DNA versus MNPs at three different scales.
formation. But the large decrease in DNA loading cannot be solely explained by partial availability of the intercalators. Indeed, these MNPs were now fully functionalized by PyAcr moieties, and the drop of DNA loading capacity was then attributed to the overall negative surface charge responsible for strong electrostatic repulsion between the nanoparticles and the DNA phosphate backbone. The double-stranded DNA used in this study is extracted from calf thymus and has an average molecular weight of 12.5 million Da. This large DNA probe should not be necessarily considered as a rigid double strand, and the reality is probably highly dynamic with the long DNA molecules prone to explore a large conformational space. Furthermore, it has been found through single DNA observations in solution using fluorescence microscopy that long duplex DNA molecules with a size larger than several tens of kilobase-pairs could exhibit a discrete conformational transition from a coil state to a folded compact state under specific circumstances.39 It is always possible to calculate the theoretical DNA dimension based on a simple linear double-strand model. This leads to a 6.4 μm long double helix model consisting of about 19 000 base pairs (BP). On the other hand, the MNPs used here are on average 24.3 nm in diameter, which corresponds to the length of a 71 BP-long DNA duplex, as illustrated in Figure 5. This simple geometric model based on the molecular weight allows determination of the ratio between the numbers of DNA molecules and MNPs (Table 3). That leads to 30.6 Fe3O4SPDT MNPs and 21.4 Fe3O4SPDTPyAcr MNPs, respectively, for one molecule of DNA, which could be interpreted as a cooperative/ multivalent mode of binding of several MNPs surrounding each DNA molecule. Moreover, these values represent an average of 1 MNP every 619 (Fe3O4SPDT) or 885 BP (Fe3O4SPDT PyAcr), i.e., a distance between two nanoparticles of 210 or 300 nm based on the linear model. As mentioned previously, these numbers are just given to illustrate the huge scale gap between the probe and the target. Indeed, the real situation does not correspond to this snapshot linear model, and the long CT-DNA is likely to wrap/fold around the nanoparticles, leading to MNPs clustering around the DNA.
’ CONCLUSIONS In this paper, we have reported that silane self-assembly can be applied to 24 nm MNPs in order to generate highly functionalized scaffold to promote DNA interaction. The magnetic nanoscale material is dispersible in aqueous solutions, due to increased interparticle repulsion and colloidal stabilization. A wide range of techniques was used to fully characterized MNPs (FTIR, TEM, PXRD, TGA, and ζ measurements). The physical characteristics of the interface (i.e., the functionalized shell around the nanoparticles) were found to strongly condition the resulting
properties and particularly the charge surface and the DNA binding capacity. Fe3O4SPDT MNPs, with about (21.123.5) 103 SPDT molecules per nanoparticle, have a positively charged shell (ζ = þ23.8 mV) which binds 15 μg of DNA per mg of MNPs. Subsequent functionalization with about (2.944.74) 103 PyAcr molecules (1420% efficiency) results in an increase up to 21.5 μg of DNA per mg of MNPs, whereas a complete PyAcr grafting [(30.031.1) 103 intercalators per nanoparticle] generates a negative surface charge (ζ = 6.8 mV) and almost totally suppresses the DNA binding capability of the nanocomposite. To summarize, a reasonable amount of DNA intercalator could help and promote the interaction with DNA, but further functionalization dramatically affects the surface chemistry and inhibits the DNA loading. Interestingly, the DNA binding potential of multifunctional 24 nm spherical Fe3O4 nanocrystals was found to be about 10 times larger than those from other reports which were performed with larger MNPs (∼80 nm in diameter). The larger surface area/mass of small MNPs as well as the tuning of the binding mode (nonspecific electrostatic interaction versus specific intercalation) has been stressed to explain the improvement of the physicochemical properties. Finally, these new materials could have interesting applications in the field of biomedical and biological sciences, especially DNA sensing, DNA purification, as well as nonviral gene delivery,13,6,40 but more importantly, this complete physicochemical characterization will help to design even more efficient systems by addressing in detail the link between surface functionality and DNA binding properties.
’ ASSOCIATED CONTENT
bS
Supporting Information. Synthesis of PyAcr, chemical functionalization of TMSPDT, FTIR, supplementary TEM data, PXRD, and colloidal stability data. This material is available free of charge via the Internet at http://pubs.acs.org.
’ AUTHOR INFORMATION Corresponding Author
*Telephone: þ44 (0)151 794 3499. Fax: þ44 (0)151 794 3528. E-mail:
[email protected].
’ ACKNOWLEDGMENT This work was supported by EPSRC Complex Materials Discovery Portfolio Partnership (grant number EP/C511794). The authors thank Dr. J. Long for ζ-potential measurements. ’ REFERENCES (1) Parton, E.; De Palma, R.; Borghs, G. Solid State Technol. 2007, 50, 47–48,50,63. 6191
dx.doi.org/10.1021/la104745x |Langmuir 2011, 27, 6185–6192
Langmuir (2) Mulder, W. J. M.; Griffioen, A. W.; Strijkers, G. J.; Cormode, D. P.; Nicolay, K.; Fayad, Z. A. Nanomedicine-UK 2007, 2, 307–324. (3) De, M.; Ghosh, P. S.; Rotello, V. M. Adv. Mater. 2008, 20, 4225–4241. (4) Fang, C.; Zhang, M. J. Mater. Chem. 2009, 19, 6258–6266. (5) Frey, N. A.; Peng, S.; Cheng, K.; Sun, S. Chem. Soc. Rev. 2009, 38, 2532–2542. (6) Selvan, S. T.; Tan, T. T. Y.; Yi, D. K.; Jana, N. R. Langmuir 2010, 26, 11631–11641. (7) Wang, X.; Liu, L.-H.; Ramstrom, O.; Yan, M. Exp. Biol. Med. 2009, 234, 1128–1139. (8) Jun, Y.-w.; Lee, J.-H.; Cheon, J. Angew. Chem., Int. Ed. 2008, 47, 5122–5135. (9) Ge, J.; Hu, Y.; Biasini, M.; Beyermann, W. P.; Yin, Y. Angew. Chem., Int. Ed. 2007, 46, 4342–4345. (10) Wu, W.; He, Q.; Jiang, C. Nanoscale Res. Lett. 2008, 3, 397–415. (11) Salazar-Alvarez, G.; Qin, J.; Sepelak, V.; Bergmann, I.; Vasilakaki, M.; Trohidou, K. N.; Ardisson, J. D.; Macedo, W. A. A.; Mikhaylova, M.; Muhammed, M.; Baro, M. D.; Nogues, J. J. Am. Chem. Soc. 2008, 130, 13234–13239. (12) Joshi, H. M.; Po Lin, Y.; Aslam, M.; Prasad, P. V.; Schultz-Sikma, E. A.; Edelman, R.; Meade, T.; Dravid, V. P. J. Phys. Chem. C 2009, 113, 17761–17767. (13) Varadan, V. K., Chen, L., Xie, J. Nanomedicine: Design and Applications of Magnetic Nanomaterials, Nanosensors and Nanosystems; Wiley: New York, 2008; Chapter 4. (14) Robinson, D. B.; Persson, H. H. P.; Zeng, H.; Li, G.; Pourmand, N.; Sun, S.; Wang, S. X. Langmuir 2005, 21, 3096–3103. (15) De Palma, R.; Peeters, S.; Van Bael, M. J.; Van den Rul, H.; Bonroy, K.; Laureyn, W.; Mullens, J.; Borghs, G.; Maes, G. Chem. Mater. 2007, 19, 1821–1831. (16) McBain, S. C.; Yiu, H. H. P.; Dobson, J. Int. J. Nanomedicine-UK 2008, 3, 169–180. (17) Latham, A. H.; Williams, M. E. Acc. Chem. Res. 2008, 41, 411–420. (18) Basiruddin, S. K.; Saha, A.; Praghan, N.; Jana, N. R. J. Phys. Chem. C 2010, 114, 11009–11017. (19) Chiu, S. K.; Hsu, M.; Ku, W. C.; Tu, C. Y.; Tseng, Y. T.; Lau, W. K.; Yan, R. Y.; Ma, J. T.; Tzeng, C. M. Biochem. J. 2003, 374, 625–632. (20) Balasubramanian, S.; Xuefei, H. Chirality 2008, 20, 265–277. (21) Huang, X.; Schmucker, A.; Dyke, J.; Hall, S. M.; Retrum, J.; Stein, B.; Remmes, N.; Baxter, D. V.; Dragnea, B.; Bronstein, L. M. J. Mater. Chem. 2009, 19, 4231–4239. (22) Hill, H. D.; Millstone, J. E.; Banholzer, M. J.; Mirkin, C. A. ACS Nano 2009, 3, 418–424. (23) Fixler, N.; Demeunynck, M.; Brochier, M.-C.; Garcia, J.; Lhomme, J. Magn. Reson. Chem. 1997, 35, 697–700. (24) Bouffier, L.; Demeunynck, M.; Milet, A.; Dumy, P. J. Org. Chem. 2004, 69, 8144–8147. (25) Bouffier, L.; Baldeyrou, B.; Hildebrand, M.-P.; Lansiaux, A.; David-Cordonnier, M.-H.; Carrez, D.; Croisy, A.; Renaudet, O.; Dumy, P.; Demeunynck, M. Bioorg. Med. Chem. 2006, 14, 7520–7530. (26) Cosnier, S.; Ionescu, R. E.; Herrmann, S.; Bouffier, L.; Demeunynck, M.; Marks, R. S. Anal. Chem. 2006, 78, 7054–7057. (27) Wang, B.; Bouffier, L.; Demeunynck, M.; Mailley, P.; Roget, A.; Livache, T.; Dumy, P. Bioelectrochemistry 2004, 63, 233–237. (28) Cha, H. G.; Kim, C. W.; Kang, S. W.; Kim, B. K.; Kang, Y. S. J. Phys. Chem. C 2010, 114, 9802–9807. (29) Hubert Mutin, P.; Guerrero, G.; Vioux, A. J. Mater. Chem. 2005, 15, 3761–3768. (30) Launer, P. J. Infrared analysis of organosilicon compounds: Spectrastructure correlations. In Silicone Compounds Register and Review, 4th ed.; Anderson, R., Arkles, B., Larson, G. L., Eds.; Petrarch Systems: Bristol, PA, 1987; pp 100103. (31) Patterson, A. L. Phys. Rev. 1939, 56, 978–982. (32) Sahoo, Y.; Pizem, H.; Fried, T.; Golodnitsky, D.; Burstein, L.; Sukenik, C. N.; Markovich, G. Langmuir 2001, 17, 7907–7911.
ARTICLE
(33) Zhang, L.; He, R.; Gu, H. C. Appl. Surf. Sci. 2006, 253, 2611–2617. (34) Pan, B. F.; Gao, F.; Gu, H. C. J. Colloid Interface Sci. 2005, 284, 1–6. (35) Metwalli, E.; Haines, D.; Becker, O.; Conzone, S.; Pantano, C. G. J. Colloid Interface Sci. 2006, 298, 825–831. (36) Schwarzenbach, G.; Schwarzenbach, K. Helv. Chim. Acta 1963, 46, 1390–1400. (37) Tanaka, T.; Sakai, R.; Kobayashi, R.; Hatakeyama, K.; Matsunaga, T. Langmuir 2009, 25, 2956–2961. (38) Nakagawa, T.; Hashimoto, R.; Murayama, K.; Tanaka, T.; Takeyama, H.; Matsunaga, T. Biotechnol. Bioeng. 2006, 94, 862–868. (39) Melnikov, S. M.; Serveyev, V. G.; Yoshikawa, K. J. Am. Chem. Soc. 1995, 117, 2401–2408. (40) Kamei, K.; Mukai, Y.; Kojima, H.; Yoshikawa, T.; Yoshikawa, M.; Kiyohara, G.; Yamamoto, T. A.; Yoshioka, Y.; Okada, N.; Seino, S.; Nakagawa, S. Biomaterials 2009, 30, 1809–1814.
6192
dx.doi.org/10.1021/la104745x |Langmuir 2011, 27, 6185–6192
