Surface Molecular Imprinting with Bacteria: Visualizing Re-Binding and Selectivity

Abstract

Introduction Surface molecular imprinting [1,2] is one of the preferred techniques when it comes to generating biomimetic recognition for sensing bioanalytes. Limiting oneself to the surface of the respective molecularly imprinted polymer (MIP) usually is no problem, because biospecies are usually large enough that few binding events lead to measurable sensor responses, e.g. for mass-sensitive bacteria detection [3]. However, the detection limits of quartz crystal microbalances (QCM) do usually not meet the needs for detecting contamination of real-life samples when it comes to pathogen detection. The (geometrical) limitation of surface MIP in that regard makes it necessary to think about suitable systems for pre-concentration prior to the sensor measurement. Herein, we report Raman Microscopy studies on bacteria surface MIP aiming at visualizing different areas of the polymer and distinguishing between different species bound to the surface. Furthermore, we assess the binding properties of MIP resulting from Pickering emulsion polymerization [4] as a possible approach to selectively pre-concentrate bacteria. Experimental For molecular imprinting, we chose a non-pathogenic strain of Escherichia coli (E. coli) as a model species. We prepared surface MIP of two different morphologies based on polystyrenes cross-linked with divinyl benzene utilizing azo-bisisobutyronitrile (AIBN) as a radical initiator. For MIP thin films, we prepared respective mixtures of monomer and crosslinker, pre-polymerized at 70°C, and then spin-coated onto the respective devices. In parallel, we prepared stamps comprising E. coli on their surfaces. Then, we pressed the stamp into the oligomer thin films on the respective surface and cured the system at 80°C over night. We took all spectra on a confocal Raman Microscopy system (WiTec alpha300RAS) at an excitation wavelength of l=532nm. Chemometric data evaluation relied on the software package Solo&Mia by Eigenvector. For Pickering emulsion, we prepared mixtures of styrene and divinylbenzene and used bacteria to emulsify them in either distilled water, or a suitable buffer system (e.g. 10mM PBS) and polymerized the bacteria stabilized droplets at 37°C. Then, we removed E. coli cells from the polymer surface via solvent extractions in acetic acid + SDS, water and methanol. Results and Discussion In a first step it was essential to clarify if one can use the Raman spectra of different bacteria species and polymer, to differentiate between those, respectively. Polymer and bacteria spectra obviously differ from each other enough to distinguish them by the naked eye. However, this is not the case for the Raman emission spectra of different bacteria species, because they all contain nucleic acids, proteins, and lipids. Therefore it is necessary to rely on chemometric strategies – modified partial least squares discriminant analysis (PLS-DA) in the concrete case – to achieve such goal. This is indeed possible: Fig. 1A shows the outcome for four different bacteria species, namely Escherichia coli, Lactococcus lactis, Bacillus cereus and Staphylococcus epidermidis that clearly demonstrates separation of the species according to their spectral properties. Knowing the specific spectral ranges also allows for generating false-color images showing which areas of a bacteria MIP surfaces reveal pure polymer and which parts contain bacteria of different species. It is even possible to go one step further: Comparing AFM and false color so-called RamanTV images (see Fig. 1B) for an example of the latter) demonstrates that the Raman microscope is even useful for distinguishing polymer surface and imprinted cavities. For such micrometer-sized analytes, the combination of Raman microscopy and AFM hence allows for generating maps showing all possible types of surfaces, i.e. polymer matrix as well as both unoccupied cavities and different bacteria species present at the polymer surface. This opens the perspective of directly visualizing occupancy and selectivity of the respective MIP. AFM turned out useful to visualize such imprints not only on flat surfaces, but also on curved ones: Figure 1C) and 1D) reveal an imprinted site on the surface of a miroparticle resulting from emulsion polymerization, Fig. 1E) an SEM image showing occupied and empty cavities. One can generate such cavities in different polymer matrices, for instance when replacing divinyl benzene as a cross-linker by trimethyolylpropane trimethacrylate (TRIM), or ethylene glycol dimethacrylate (EGDMA). In both cases, the respective imprinted surface reveal cavities that fit the size and shape of the template bacteria. Scanning electron microscopy also revealed that bacteria re-occupy those cavities if one exposes the particles to to bacteria suspensions. This leads to two conclusions: first, one can tune polymer properties without destroying the emulsion structure necessary for Pickering emulsion. This is of interest for bringing the artificial system closer to physiological conditions. Second, those particles are inherently suitable for selectively pre-concentrating those species from larger samples. Figure 1