Covalent immobilization of molecularly imprinted polymer nanoparticles using an epoxy silane.

Molecularly imprinted polymers (MIPs) can be used as antibody mimics to develop robust chemical sensors. One challenging problem in using MIPs for sensor development is the lack of reliable conjugation chemistry that allows MIPs to be fixed on transducer surface. In this work, we study the use of epoxy silane to immobilize MIP nanoparticles on model transducer surfaces without impairing the function of the immobilized nanoparticles. The MIP nanoparticles with a core-shell structure have selective molecular binding sites in the core and multiple amino groups in the shell. The model transducer surface is functionalized with a self-assembled monolayer of epoxy silane, which reacts with the core-shell MIP particles to enable straightforward immobilization. The whole process is characterized by studying the treated surfaces after each preparation step using atomic force microscopy, scanning electron microscopy, fluorescence microscopy, contact angle measurements and X-ray photoelectron spectroscopy. The microscopy results show that the MIP particles are immobilized uniformly on surface. The photoelectron spectroscopy results further confirm the action of each functionalization step. The molecular selectivity of the MIP-functionalized surface is verified by radioligand binding analysis. The particle immobilization approach described here has a general applicability for constructing selective chemical sensors in different formats.


Introduction
Many biological macromolecules (e.g. antibodies, proteins, enzymes, and aptamers) exhibit a highly fine-tuned and effective molecular recognition ability, which makes them suitable for 3 chemical and biological sensors. [1][2][3] However, typically these natural receptors are costly and have problems with stability (chemical and physical) and sensitivity in non-optimal environments 4 and therefore it is often necessary to switch to synthetic receptors with high stability and cost effectiveness. A new recent development is the use of molecularly imprinted polymers (MIPs) with predesigned molecular selectivity produced by template driven polymerization. 5,6 The MIPs can be fabricated directly in situ on a transducer surface to act as a molecular recognition layer for detection of the template or its structural analogs. For example, imprinted polymer films for detection of heparin have been fabricated on an electrode in a potentiometric sensor. 7 A micro patterned thin MIP film imprinted with testosterone has also been reported in a holographic sensor. 8 A major problem associated with the in situ fabricated MIP sensors is, however, the inaccessibility of a large fraction of template sites due to the layers' small surface area. With ex situ-prepared MIP-based sensors the problem can be circumvented since these MIPs can be synthesized as particles with a large surface-to-volume ratio. In the approach the MIPs are prepared ex situ using emulsion 9 or precipitation polymerization 10 and then attached to the transducer surface either directly or through a polymer layer, as reported by Reimhult et al. for QCM-D sensors 11 and Kröger et al. for an electrochemical sensor. 12 A drawback of using polymer layer-based conjugation chemistry is the thickness of the interface layer, which can affect the MIP particles' sensitivity. Here we are looking for an approach that is designed not to affect the sensitivity of the sensing material, which is achieved by minimizing the contact between the transducer and MIP surfaces. At the same time the contact is stable and allows the formation of a smooth and uniform MIP particle layer. The contact layer consisting of highly oriented epoxy silane linker molecules is formed via a self-assembled monolayer (SAM). The structurally oriented SAM can be prepared easily with high reproducibility. 13 4 The SAMs of organosilanes have proved to be useful for functionalization of different silica surfaces like quartz, silicon wafers, glass, etc. The MIP particles can be immobilized electrostatically or covalently using SAM modified surfaces. As an example, recently an optical MIP sensor was fabricated by Felix et al. using a SAM of 3-aminopropyltriethoxysilane (APTES), where the positively charged amine allowed attachment of the negatively charged MIP particles. 14 However, electrostatic attachment does not offer sufficient particle fixation, particularly in the presence of electrolytes or in highly basic/acidic solutions. Therefore, it is necessary to develop suitable methods for the covalent immobilization of MIP particles.
Recently, we used perfluorophenylazide-mediated photoconjugation chemistry [15][16][17] to immobilize MIP particles. 18 The azide groups are activated by UV light and the generated nitrenes undergo CH insertion reactions leading to a covalent immobilization of the MIP particles. This approach results in a dense coverage of MIP particles, but the morphology is rough and difficult to control and thus not desirable for sensing applications. 19 Here we instead report another approach for covalent immobilization, with which smooth and uniform layers of MIP particles immobilized on a surface can be achieved (cf. Scheme 1). We used propranolol-imprinted MIP nanoparticles and immobilize them directly on GPTMS functionalized glass. The uniform and stable silane functionalized glass surface has epoxide terminal groups, which are available to covalently attach amine functionalized MIP core-shell nanoparticles via an epoxide ring opening reaction. 20 The covalently immobilized MIP particles are in close contact with the surface due to short length of the epoxy silane linker (< 2 nm) and retain their binding sites for the template

Methods
To study the roughness and homogeneity of the prepared surfaces tapping-mode AFM was carried out in ambient environment using an instrument by Veeco (New York, USA). The n-  XPS was performed at the spectroscopy end station of beamline I311 at the MAX IV Laboratory in Lund, Sweden. 21 The instrument with separate preparation and analysis chambers (base pressures better than 10 -10 mbar) and a sample load lock is equipped with a Scienta SES 200 hemispherical electron energy analyzer. The samples were prepared ex situ and mounted on the sample holder using double-sided adhesive carbon tape. To avoid sample charging expected for non-conducting samples the glass slides were replaced by a n-doped Si (100) wafer. Good electrical contact of the surface with ground was ensured by establishing a direct contact between the sample surface and sample holder using conductive carbon tape. The measurements were started after introduction of the sample and subsequent degasing once the pressure had recovered to 10 -8 mbar. The X-ray photoelectron (XP) spectra were recorded in normal (emission angle 0° from the surface normal) and grazing (75°) emission geometries. All spectra were calibrated by reference to the Si 2p peak for silicon dioxide peak at 103.2 eV binding energy. 22 Shirley or polynomial backgrounds were removed from all spectra.
Radioligand binding analysis was performed to study the molecular selectivity of the MIPcoated surfaces. For these experiments a two inch silicon wafer (specifications same as above) was cut into pieces of 10 × 10 mm 2 size. The pieces were then functionalized with GPTMS and incubated in MIP particle and non-imprinted polymer (NIP) suspensions for 16-18 h at room 8 temperature. The resulting surfaces were washed in 20% acetic acid solution in methanol twice and then in acetonitrile to remove any pyridine left after particle immobilization. The surfaces were then incubated for 12 h at room temperature with 3 H-labelled propranolol solution in acetonitrile (100 nM), after which they were rinsed in acetonitrile and left for a few minutes to dry. Finally, the surfaces were left for 18 h at room temperature in a cassette that ensured intimate contact between the sample surfaces and the tritium-sensitive screen. The tritiumsensitive screen was then analyzed using a FUJIFILM Fluorescent Image Analyzer FLA 3000 with a laser wavelength of 635 nm. The image analyzer generated signal patterns from the photographic film, which converts particles from the radioactive decay into photostimulable particles from the radioactive decay into photostimulable luminescence. 23 In this way the luminescence intensity is in proportion to the quantity of the radioligand bound on the sample surface. Further data analysis was carried out using the Origin Pro 9.1 software.

Synthesis of the MIP nanoparticles
Molecularly imprinted core-shell nanoparticles were synthesized in two steps according to the procedure reported by Hajizadeh et al. 24 In the first step, ( purged with nitrogen for 5 min before the mixture was allowed to polymerize for 48 h under a gentle rotation in an oven at 60 °C. After the second polymerization the reaction mixture was centrifuged at 11300 g for 15 min to collect the polymer particles. The template was removed by washing the core-shell particles with methanol containing 10% acetic acid (v/v), until no template could be detected from the washing solvent by UV spectrometry using a wavelength of 290 nm. The polymer particles were finally washed with acetone and dried in a vacuum chamber. The core-shell NIP particles were synthesized using the same protocol as described above, but without the template in the pre-polymerization mixture.

Preparation of silanized glass
The glass slides were cut into pieces of 10 × 10 mm 2 size, which were cleaned in 2 M NaOH solution for a few minutes. The slides were then rinsed using distilled water and kept in 0.

Results and discussion
Scheme 1 illustrates the approach used for immobilization of the MIP nanoparticles. In this approach, the surface was first hydroxylated and then functionalized with GPTMS. In the final step the surface was exposed to a solution of core-shell MIP particles (or core-shell NIP particles as a reference) to finish the nucleophilic substitution reaction. The density of the epoxy groups was tailored to achieve a high availability of epoxy groups for MIP particle attachment, while ensuring the uniformity of the silane layer. The ideal concentration of GPTMS was found to be 2%, at which both homogeneity and high density of epoxides on the glass surface were achieved (see Electronic Supplementary Material, Figure S1). Table 1 provides a short description of the surfaces prepared according to the scheme, which then were characterized by different methods. The table also specifies the terminology for the remainder of the text. The "glass", "G-Epoxy" and "G-Epoxy-MIP" preparations represent the single steps in Scheme 1. For benchmarking we also studied surfaces for which certain steps of the preparation had been eliminated. The "G-Epoxy-MIP*" preparation represents the entire 11 process of Scheme 1, but with the final sonication omitted, thus allowing assessment of the aggregation of physisorbed MIP particles. For the "G-MIP" surface the GPTMS functionalization was omitted, while for the "G-Epoxy-NIP" surface the entire preparation process was performed, but with NIP rather than with MIP nanoparticles. For the XPS measurements and radio ligand binding studies we replaced glass with a silicon wafer. The Si wafer surfaces were prepared in the same way as the microscopy glass slide in Scheme 1 and are represented as Si, Si-Epoxy, Si-Epoxy-MIP and Si-Epoxy-NIP (cf. Table 1). The XP spectra provide additional insight about the elemental information of the surfaces.

Tab
The overview spectrum obtained on the Si-Epoxy surface in Figure 2A shows a strong increase in C 1s and N 1s intensity upon GPTMS treatment in comparison to the spectra measured on the Si surface, for which only a small carbon contamination peak is seen. Also the Si 2s, Si 2p and O 1s lines change significantly. This confirms the functionalization of the Si slides -treated in the same way as the glass slides -with GPTMS (cf. Table 1).
The C 1s XP spectra measured in grazing and normal emission geometries are normalized to the maximum height of the signal and are shown together in Figure 2B, where their components can be compared. Three components are found, the first of which at 285.20 eV, we assign to photoemission from the carbon atoms bonded to carbon, hydrogen and silicon (denoted C-C, C-H). The second component at 286.89 eV is due to photoemission from oxygen-bonded carbon atoms and thus comprises the signal from the epoxy group. 27 Finally, at highest binding energy we find a peak that we suggest is due to carbonate impurities due to the possible reaction of the epoxide groups on the surface with atmospheric carbon dioxide at high temperature in the oven. 28,29 When measured in grazing emission, the intensity ratio between the two main components changes in comparison to the spectrum measured in normal emission geometry: the peak intensity ratio of the two main components is almost 1, whereas it decreases in the grazing emission spectrum. This is a clear indication of that the epoxide groups are available on the surface with high density. affecting the specificity and molecular recognition properties of the MIP nanoparticles towards the template. Thus the method is applicable to various amine functionalized materials for lab-onchip and other sensing applications. 37,38 The approach does not require high temperature, photoor chemical activation, which renders the method physically and chemically more robust and handy in comparison to other immobilization procedures.

Supporting Information
Water contact angle measurements, SEM images, AFM cross section profile, intensity profiles of autoradiography images.