A linear-polymer-based lactoferrin- selective recognition element for an ELISA mimic: A Proof of concept

Title: A linear-polymer-based lactoferrinselective recognition element for an ELISA mimic: A Proof of concept Authors: M.A. Goicolea a*, A. Gómez-Caballero a, M. Saumell-Esnaola b, d, G. García del Caño c, d, N. Unceta a, J. Sallés b, d. e and R.J. Barrio a. a Department of Analytical Chemistry. Faculty of Pharmacy, University of the Basque Country (UPV/EHU), 01006 Vitoria-Gasteiz, Spain. b Department of Pharmacology. Faculty of Pharmacy, University of the Basque Country (UPV/EHU), 01006 Vitoria-Gasteiz, Spain. c Department of Neurosciences. Faculty of Pharmacy, University of the Basque Country (UPV/EHU), 01006 Vitoria-Gasteiz, Spain. d Instituto de Investigación Sanitaria Bioaraba, Neurofarmacología Celular y Molecular, 01008 Vitoria-Gasteiz, Spain. e Centro de Investigación Biomédica en Red de Salud Mental (CIBERSAM), 28029. Madrid, Spain. *Corresponding author. E-mail addresses: [email protected] Highlights A linear polymer for the selective recognition of lactoferrin is presented. The linear polymer works as a plastic antibody in enzyme-linked immunosorbent assays. Assay performance for lactoferrin was testedin the presence of other proteins. This is the accepted manuscript of the article that appeared in final form in Analytica Chimica Acta 1191 : (2022) // Article ID 339309, which has been published in final form at https://doi.org/10.1016/j.aca.2021.339309. © 2021 Elsevier under CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/) Abstract The synthesis of polymers with tailored properties for the recognition of macromolecules such as proteins is challenging. In this work, the synthesis of a new polymer format, a linear polymer (LP), as the selective recognition element for the globular protein lactoferrin (LF) is proposed as a proof-of-concept study. For the synthesis, a solid-phase strategy using the reversible deactivation radical polymerisation (RDRP) mechanism is proposed. This approach, which is usually used in molecular imprinting, involves the immobilisation of LF on the surface of a solid support, but, unlike classical imprinting, a cross-linker in the polymerisation mixture is not required. Consequently, the copolymer is soluble and flexible, thus overcoming the drawbacks associated with traditional synthetic polymers for macromolecule imprinting. This new polymer format has great potential for replacing natural antibodies in bioassays such as enzyme-linked immunosorbent assays (ELISA), dot blot, western blot, or pull-down. In our case, the linear polymer was used as a recognition element to replace natural antibodies in a LF-selective ELISA. The responses of the linear polymer between LF concentrations of 0.1 nM and 0.25 µM were studied, and a significant difference was observed between the non-specific signals and the signals measured in the presence of the polymeric material. Further, the response versus log concentration curves were fitted to a logistic equation, allowing estimation of the EC50 value: 11.8 ± 1.4 nM. We also confirmed the selective detection of LF using the competitive inhibition of the selective LF-biotin conjugate (LF-Bi) binding to the plastic receptor (LP) for closely related proteins (e.g. those having similar molecular weights or isoelectric points) such as human lysozyme, trypsin, and albumin, which are present in human body fluids. The system presents a cross-reactivity value or selectivity of 1.95% for lysozyme, 0.028% for trypsin, and 0.016% for albumin. The applicability of this method for the determination of urine LF levels in inflammatory and infectious diseases of the human urinary tract is also demonstrated. Keywords Linear polymer, reversible deactivation radical polymerisation, plastic antibody, lactoferrin, ELISA mimic, urine sample 1. Introduction Molecular recognition is a fundamental event in biochemical processes in which a receptor recognises and identifies one chemical species through structurally well-defined molecular interactions. The most studied molecular recognition system is the immune system of higher organisms in the form of antigen–antibody recognition. This recognition is the basis of numerous bioassays, including the enzyme-linked immunosorbent assay (ELISA), as well as dot-blot, western blotting, and immunoprecipitation assays. These recognition phenomena are crucial in biological systems, and much modern chemical research is motivated by the prospect that molecular recognition by design could lead to new technologies for the replacement of natural receptors by more durable and robust synthetic structures[1,2]. To address this challenge, many efforts have been made to design, synthesise, and test artificial materials with biomimetic properties for applications in binding assays. Supramolecular host–guest interactions represent one of the earliest non-natural molecular recognition systems. However, this now-traditional field of chemistry is relatively limited in terms of the number and function of potential receptors. With the increasing demand for multifunctional advanced materials, the molecular imprinting technique has received much attention as a strategy to incorporate specific molecular recognition sites into polymeric structures, leading to artificial polymeric receptors that mimic the mechanism of natural receptors [3]. Further, in principle, rational molecular design and chemical synthesis allow the generation of affordable materials that are easy to fabricate and use and have high selectivity and robustness[4]. Accordingly, molecularly imprinted polymers have been proposed in recent years as substitutes for natural antibodies in immunoassays[5-8]. In traditional molecular imprinting with small molecules, to preserve the cavities created during polymerisation, the imprinted polymers tend to be very dense, and the presence of a cross-linking monomer makes it possible to obtain well-defined binding cavities while maintaining recognition. To date, macromolecule imprinting technology has not yet reached its true potential because of the inherent need to perform imprinting in aqueous media. Moreover, macromolecular structures, such as peptides and proteins, can exist in a multitude of conformations, leading to the development of heterogeneous binding sites as opposed to welldefined cavities during the imprinting of small organic molecules. In addition, the large size of macromolecules can reduce the effectiveness of extraction from the polymer matrix and subsequent binding. Although some solutions to these problems in protein imprinting techniques have been published [9-11] several drawbacks remain. The main problem facing the imprinting of macromolecules is the high degree of crosslinking required to achieve recognition. As cross-linking increases, large templates, such as proteins, can become trapped in the network after polymerisation. If the template molecule cannot be extracted, the network is useless for recognition applications. In addition, a high degree of cross-linking can decrease the diffusion of these molecules in the network, resulting in slow recognition kinetics. Consequently, most research on protein imprinting has focused on the production of materials using surface molecular imprinting techniques [12]. In this strategy, the cavities have special shapes or steric effects that are complementary to parts or fragments of the target protein. As a result, mass transfer is easier and binding kinetics are less restricted, although possibly lower, and selectivity may also decrease because only part of the protein is bound or recognised. Crucially, the stability of the conformation of macromolecules during the imprinting process must be guaranteed, so functional monomers and cross-linkers that maintain the activity of the proteins must be selected [13]. Some authors have proposed the use of protein–polymer hybrid structures to achieve control of the nanostructure and orientation of the proteins while maintaining their stability and activity. For example, the conjugation of synthetic copolymers to proteins can confer beneficial properties such as tailored amphiphilicity, novel self-assembly, and phase separation behaviour to the resulting hybrids [14,15]. Copolymers are capable of complexing proteins with efficiencies depending on the copolymer composition and molecular architecture. These complexes are typically assembled using linear copolymers [16]. In the synthesis of polymers for the uptake of biomacromolecules, the degree of flexibility of the polymer must be considered. Scharader et al. [17] revealed that linear copolymers having special functions and components could exhibit high binding properties and selectivity to biomolecules. They demonstrated that, for selective protein recognition, a pre-established binding site is not a requirement as long as there is a sufficient number of interactions between the functional groups of the polymer and the amino acid residues on the protein surface. In this case, the recognition results from induced fit. Specifically, the copolymer adapts its shape according to the topology of the protein surface. In particular, linear copolymers are much simpler and suffer from less steric hindrance than three-dimensional polymeric structures; therefore, the amount of immobilised protein can be maximised. Typically, the polymerisation mixture is composed of a set of excess monomers with a large variety of functional groups. Thus, the expense depends on the different amino acid residues that predominate in the protein structure. As a result, flexible soluble polymers lacking a bulky three-dimensional structure are obtained, and these have proven advantageous for the development of new biochemical tools and bioassays [18,19]. Crucially, the linear copolymers used for molecular recognition are not imprinted, and recognition is entirely the result of induced fit, which enhances the favourable interactions between the polymer functionality and amino acid residues on the protein surface. This approach is conceptually identical to the theories postulated to account for the unlimited binding diversity of antibodies. That is, the concept of induced fit relaxes the requirement for an exact pre-existing fit between the antibody and antigen [20]. Similarly, it has been shown that random heteropolymers composed of monomers having chemical properties designed to interact with the species present on the protein surface adopt local conformations that maximise the favourable interactions with the protein, taking advantage of the conformational flexibility and designed complementarity in the polymer backbone [21,22]. Molecular recognition occurs when two molecules are both geometrically and chemically complementary; that is, when they can both “fit together” spatially, as well as bind to each other using non-covalent forces, including hydrogen bonds, electrostatic interactions, or hydrophobic interactions [14]. The recognition ability increased with the number of different types of monomer units and complementary adsorption sites of the protein. Thus, it could be said that the heteropolymer is tuned to amino acid residues on the surface of the protein [23]. In this study, a flexible linear polymer for selective binding to a whole protein was developed. As in the work of Scharader [17], the statistical copolymerisation of monomers complementary to the most characteristic surface residues of the protein is proposed, and this results in a flexible linear polymer capable of adapting to the topology of the target protein: lactoferrin (LF). The aim of this work was to increase the size and flexibility of the polymer to facilitate protein binding, taking into account the multiple interconverting conformations of the protein. In this context, we describe the development, characterisation, and biological testing of soluble linear copolymers (LPs) designed for LF. To validate these materials as substitutes for antibodies in diagnostic tests, their behaviour was studied using an ELISA mimic. Thus, as a proof of concept, the synthesis of a new LP is proposed. In this study, the monomer polymerisation was carried out in the presence of a template molecule, but, unlike traditional polymerisation for the generation of a molecularly imprinted polymers, no crosslinker was required to maintain the specific recognition. This results in soluble copolymers consisting of carbon backbones having functional side chains that are randomly distributed as a result of the statistical polymerisation of the selected functional monomers. That is, copolymers are formed in which the sequential distribution of the monomeric unit obeys statistical laws, and the processes leading to the formation of a statistical sequence of monomeric units do not occur with equal probability. Crucially, the linear and flexible polymer can adapt its shape to the flat and rough surface topologies of the target protein. As discussed above, a key to this method is the optimisation of the selection and proportion of the different functional monomers, which must be selected for their favourable interactions with the target molecule [3]. In this study, a mixture of different acrylamide monomer derivatives with various functionalities was used to prepare the LP and avoid the formation of homopolymers, which would be incapable of selective binding. We hypothesised that, to obtain good affinity and selectivity, the sequence of monomers along the polymer backbone should be directed by pre‐polymerisation driven by self-assembly around the template. To achieve this, polymerisation was conducted following a solid-phase synthesis strategy for the synthesis of imprinted materials in which LF was immobilised on the surface of glass beads, which acted as the solid phase [24,25]. These were then placed in contact with the aqueous monomer mixture, and reversible deactivation radical polymerisation (RDRP) was initiated in the presence of an iniferter. This technique offers great control over the polymerisation process, allowing the synthesis of polymers with a controlled molecular weight distribution and a more homogeneous distribution of binding sites compared to conventional radical polymerisation. In addition, RDRP enables the production of advanced materials containing latent reactivatable species that can be used for further processing in the postsynthetic phase [19,21,26]. 2. Experimental 2.1. Materials and reagents Human LF (90%), human lysozyme (≥ 100,000 units/mg), porcine pancreatic trypsin, fatty acid-free bovine serum albumin (BSA, 99%), acrylamide (AAm, 99%), N-tert-butyl acrylamide (TBAm, 97%), acrylic acid (AAc, 99%), N-(3-aminopropyl) methacrylamide hydrochloride (APMA, 98%), 3-aminopropyl triethoxysilane (APTES, 99%), 1,2bis(triethoxysilyl) ethane (BTESE, 95%), glutaraldehyde (GA, 50% w/v, ethanolamine (99.5%), dansyl chloride (99%), biotin N-hydroxysuccinimide ester (biotin-NHS) (98%), glass beads (GB, 150 µm < d < 212 µm), and silver colloidal dispersion (60 nm, 0,02 mg mL-1) were purchased from Sigma–Aldrich (Spain). Sodium hydrogen phosphate, sodium dihydrogen phosphate, sodium hydroxide, polyvinylpyrrolidone, Amicon Ultra-15 centrifugal filter units (3 and 30 KDa), and BCA protein assay kit were purchased from Merck (Spain). Sodium chloride, sodium nitrate, dry toluene, dimethyl sulphoxide (DMSO, 99.5%), and hydrochloric acid (37%) were acquired from Panreac (Spain). Pierce® Biotin Quantitation Kit, peroxidaseavidin conjugate (HRP-avidin), 3,3',5,5'-tetramethylbenzidine (TMB), Tween 20 surfactant, 96well Costar 3590 polystyrene microplates, and nitrocellulose membranes (200-nm pore size) for use as substrate for surface-enhanced Raman spectroscopy (SERS) were purchased from Thermo Fisher Scientific (Spain). Liquid chromatography–mass spectrometry grade and highperformance liquid chromatography (HPLC) grade ethanol were obtained from Schalab S.L. (Spain). Deuterated water for NMR was obtained from Eurisotop (France). The calibration of the gel permeation chromatography/refractive index (GPC/RI) system was carried out using EasiVial polyethylene glycol (PEG)/polyethylene oxide (PEO) standards compliant with the ISO9001:2008 certification and supplied by Agilent Technologies (Spain). All chemicals and solvents were of analytical or HPLC grade and were used without further purification. Each buffer solution was prepared with ultrapure water obtained from Elix20 reverse osmosis and Milli-Q water purification systems. 2.2. Equipment The silanisation of the solid support was verified by measuring the fluorescence at 365 nm using a Spectronics CM-26A cabinet equipped with a model ENF-260C lamp (USA). The immobilisation of the protein on the solid support was confirmed by spectroscopic measurements using a 1-m long, 10-mm optical path length fibre optic probe coupled to an Agilent Technologies Cary 60 UV-Vis spectrophotometer. Cary WinUV software (version 5.0) was used for data collection, storage, analysis, and display. For UV-light-induced polymerisation, a Summer Glow HB 175 lamp from Hapro (The Netherlands) was used. The preconcentration of the polymeric material was performed by evaporation and centrifugation using a refrigerated centrifuge (Allegra X-15R, Beckman Coulter, USA) and a Hei-VAP Advantage rotary evaporator (Germany). Empty 60-mL propylene cartridges purchased from Supelco (Spain) were used for solidphase extraction (SPE) tests, and process control was carried out in a vacuum manifold connected to a vacuum pump for pressure control (Varian, CA, USA). For the polymer characterisation, a 1260 Infinity gel filtration chromatography system with a 20 G1362A RI detector for concentration determination and multi-angle (15° and 90°) light scattering detector for molecular weight determination from Agilent Technologies was used. For the separation of the different fractions of the polymeric materials, a PolySep-GFCP-4000 column from Phenomenex (Torrance, California, USA) having a size range of 3,000– 400,000 Da and 0.1 M NaNO3 as the mobile phase was used. 1H-NMR spectra were acquired in a Bruker Avance 400 spectrometer operating at 400 MHz, and the spectra were processed using MNova program (Mestrelab Research, Spain). The SERS measurements were performed using a Horiba XploRa Raman microscope (Jobin-Yvon, Horiba Group, New Jersey, USA) equipped with lasers having wavelengths of 532, 638, and 785 nm. Spectral data were treated using the OMNIC 7.2 Spectra Software (Thermo Scientific, Waltham, MA, USA). The ELISA tests were conducted with temperature control and shaking using a Selecta model oven (Selecta, Spain) and an IKA VIBRAX VXR orbital shaker (Scharlab, Spain). Signal measurements were carried out on an absorbance microplate reader (Sunrise, Tecan Life Sciences, Spain). Data acquisition was performed using RdrOle4 version 4.5.1. Fittings of the binding kinetic, saturation, and competitive assay results were performed using PRISM® (GraphPad version 5.01 Software Inc., San Diego, CA). 2.3. Synthesis of the linear copolymer (LP) The synthetic strategy for the generation of the LP was based on the solid-phase approach using a photoactivated reversible deactivation radical polymerisation mechanism in the presence of an iniferter (photo RDRP). 2.3.1 Preparation of LF-modified glass beads Before copolymer synthesis, immobilisation of LF on the solid support (glass beads) was carried out. The first stage involved the activation and functionalisation of the glass beads. For this, 120 g of glass beads was suspended in 250 mL of 4 M NaOH solution. After boiling the mixture for 30 min, the glass beads were filtered through a 50-µm sieve and washed with ultrapure water until the wash water reached pH of 7. The beads were then washed with methanol and allowed to dry overnight in an oven at 100 °C. Once dry, the glass beads were silanised by immersing them in 100 mL of 95:5 (v/v) ethanol:water solution acidified with 1 mL of acetic acid and heated to 70 °C. Immediately after, 3 mL of APTES and 0.5 mL of the dipodal silaniser BTESE (10:1 molar ratio of APTES: BTESE) were added, and the mixture was left to react overnight at room temperature. Thereafter, the silanised glass beads were filtered, washed with methanol and acetone, and dried in a desiccator under vacuum. Finally, to ensure complete water removal, the samples were placed in an oven at 150 C for 1h. To determine the degree of APTES grafting, the ninhydrin test was performed using 1 mg of silanised glass beads and a 2% ninhydrin solution from Merck (Spain), revealing a grafting degree of 438.4 ± 6.7 µg-APTES/g-glass beads. For LF immobilisation on the surface of the activated and functionalised solid support, 30 g of glass beads were placed in a round-bottomed flask with 100 mL of 0.1 M phosphate buffer (pH = 7.4) containing 0.1 M NaCl (phosphate buffer saline, PBS) and GA (5%, v/v). To ensure an oxygen-free atmosphere, the mixture was subjected to repeated N2 purge–vacuum cycles using a Schlenk line. Subsequently, the mixture was allowed to react for 2 h in the dark. Subsequently, the glass beads were filtered and washed with ultrapure water. They were then transferred to a two-necked flask with 25 mL of PBS, and LF (12.5 mg) was added. The mixture was again subjected to N2 purge–vacuum cycling and kept in the dark for 2 h. Once the LF had been immobilised, 12 µL of ethanolamine was added to the mixture for 15 min to cap any unreacted aldehyde-terminal groups that had not reacted with LF. Finally, the glass beads were filtered and rinsed with Milli-Q water. 2.3.2. Synthesis and characterisation of the water-soluble iniferter 2-((diethylcarbamoyl)thio) acetic acid. Iniferters behave sequentially as initiators, transfer agents, and terminators in the radical polymerisation process. However, examples of the iniferter-induced polymerisation of biomolecules are scarce because of their poor water solubility [27–29]. In our case, the water-soluble iniferter was prepared following a procedure adapted from that of Xu et al. [30]. Briefly, in a two-necked round-bottomed flask, 5 mmol of sodium diethyldithiocarbamate trihydrate (1.265 mg) dissolved in 50 mL of water was added and deoxygenated with high-purity nitrogen. Next, 5 mmol of sodium chloroacetate (98%, 582.5 mg) dissolved in 50 mL of water was added dropwise, and the mixture was kept under a nitrogen atmosphere for 48 h. The resulting product was filtered to remove the insoluble salts. Then, 1 mL of hydrochloric acid (37%) was added to the solution to precipitate the iniferter. The resulting product was filtered, redissolved in acetone, and the solvent was evaporated with nitrogen for crystallisation. After drying at 45 °C, the crystals were stored at 4 °C in the dark. The synthesised iniferter was characterised by quadrupole time-of-flight mass spectrometry (QTOF-MS). The measurements were acquired using an Agilent 6530 hybrid in the characterisation of the polymeric material by GPC with RI and multi-angle light scattering detection. Initially, the monomer concentration was set as described in Section 2.3.3 (denoted ×1), and the influence of the UV activation time was studied in the range of 5–90 min. The synthesis time was set to 45 min (15 min × 3 times, the mixture was maintained in an ice bath) because the polymer average molecular weight did not increase with longer reaction times (Fig. 1a). In addition, an irradiation time of 45 min was set, and the influence of the monomer content in the polymerisation mixture was studied. For all the compositions tested (×1, ×2, ×5, and ×10), an increase in the average molecular weight (Mw 7,749–37,741) was observed as the monomer content in the polymerisation mixture increased (Fig.1b). The ×5 mixture containing AAc (11 mg), APMA (27.5 mg), TBAm (165 mg), and AAm (122.5 mg) was selected because the higher concentration (×10) resulted in no significant increase in polymer molecular weight and an increase in polymer heterogeneity was observed. The average molecular weight of this monomer mixture was 34,277 ± 3,219. Regarding polymer homogeneity, the polydispersity value of 1.12 ± 0.08 is close to one, as expected for a reversible deactivation radical polymerisation because the molecular weight and conversion increase with increasing monomer concentration at the expense of polydispersity [35]. The concentration of polymer present in the synthesis stock solution was determined to be 0.091 ± 0.004 mg·mL–1 based on a refractive index value (dn/dc) 0.165 mL g-1 [36]. In addition, from the results obtained from the multi-angle light scattering detector, the average radius of gyration (Rg = 79 ± 2 nm) and branching of the polymer were determined by calculating the radius of gyration contraction factor gMw = Rg (branched)/Rg (linear) = 0.92 ± 0.03. The polymer radius of gyration describes the dimensions of the polymer chain. When a polymer's radius of gyration is greater than 10 nm, it is considered to be a long polymer. On the other hand, from the values of gMw close to one (value for linear polymers) and considering the branched-ternary weight average model, which indicates a single branch point of the backbone, the average number of branches (BnMw = 0.55 ± 0.05) was calculated, assuming a random distribution of branches in the polymer. The lower the gMw value, the higher is the degree of branching. From these results, we can conclude that it is a homogeneous linear polymer with low branching. The LP was also characterised by SERS measurements, and different SERS spectra corresponding to different LP bands were obtained using a confocal Raman spectrometer. A 532-nm laser with a power of 0.05 W (5 mW) was used for SERS excitation at the sample position. A microscope attachment with a 100× objective was used to focus the laser beam onto a spot of 1 µm in diameter. The integration times varied between 2 and 10 s with 10 accumulations in the spectral range of 100–3,200 cm-1. The spectral resolution was 2 cm-1. In addition after testing with different materials, we found that a nitrocellulose (NC) membrane was a good substrate for SERS because it offers a low background, minimal preparation, and optimal position of Raman bands. The NC membrane with a 200-nm pore size allows the capture of subnanogram quantities of analyte and concentrates them in a small area from applied volumes of 0.5 µL. The subsequent staining process with colloidal silver nanoparticles (60 nm, 0.08 mg mL-1) allowed the generation of local “hot spots” showing SERS enhancements. To achieve the aggregation of the silver colloid, the reagent suspension was pre-concentrated by centrifugation at 30,000g for 30 min. This staining process was performed by incubating the sample-loaded nitrocellulose membrane with the preconcentrated silver colloid for 2 h. Fig. 2 shows a comparison of the SERS spectra of the monomers used in the polymerisation (Fig. 2a) and the LP (Fig. 2b). Fig. 2b shows two spectra for the LP obtained from different areas of the sample. The difference observed between the spectra may be due to the polymer being adsorbed on the surface in different orientations. In the LP spectrum (Fig. 2b-A), an intense signal at 1,597 cm-1 can be observed, and a band is also present in the spectrum of the acrylamide monomer (Figure 2a-B), which forms the backbone of the LP. When acrylamide polymerisation occurs, the C=C group is transformed into a single bond, but the primary amide group is maintained (band at 1,597 cm-1 Fig. 2b-A). In the spectra of N-tert-butyl acrylamide (Fig. 2a-C) and N-(3-aminopropyl) methacrylamide (Fig. 2a-D), a double signal was observed at 1,542 and 1,575 cm-1 and 1,525 and 1,560 cm-1 respectively, relatives to the secondary amide present in the structure of both monomers. After polymerisation, this band remained in the LP spectrum (Fig. 2b-A) but overlapped with the signal of the primary amide arising from the acrylamide. In the spectrum of LP (Fig. 2b-A and 2b-B), a band at 1,431 cm-1 was observed, attributable to the δC-H and δC-C deformations of the tert-butyl groups. The signals between and 1,100 and 1,300 cm-1 present in the LP spectrum (Figures 2b-A and 2b-B) are due to C-C stretching of n-alkanes, skeletal vibration of the polymer (1,050–1,150), and 1,175–1,310 ωCH2 and δCH2 n-alkanes. The C-H stretches also appeared in the spectral region between and 2,800– 3,000. At 2,900 cm-1, there is a CH2 asymmetric stretch, which could be broadened because of the overtones of the δCH2 deformation. The bands appearing at 2,800–3000 are due to C-H stretches and may be affected by the reorientation of the alkyl chains. Finally, the LP spectrum (Fig. 2b-B) shows a well-defined band at 1,613 cm-1 arising from the carboxylic acid groups of the acrylic acid monomer (Fig. 2a-A). Thus, the characteristic SERS bands of the monomers involved in LP synthesis are also present in the SERS spectrum of the copolymer. Fig. 3 shows the 1H-NMR spectrum of the LP. The strong signal at 1.3 ppm can be assigned to the protons of the tert-butyl group originating from the N-tert-butyl acrylamide monomer. The low intensity signals close to this correspond to the protons from the alkane chain formed during polymerisation. The signal at 2.1 ppm is assignable to protons from alkanes close to the carbonyl groups, whereas the signal at 1.6 ppm arises from protons in alkanes further away from the carbonyl groups originating from the acrylamide and acrylic acid monomers. 3.2. Evaluation of LP as a plastic “antibody” in the ELISA mimic The apparent dissociation constant for the interaction between the labelled ligand (LF-Bi) and the LP receptor was determined using an ELISA mimic. For this purpose, an empirical adjustment of the ELISA conditions was performed to achieve the maximum selective binding response between LF-Bi and the LP receptor. To minimise experimental error, all assays were performed in triplicate. Therefore, the signals represented are the mean values. First, a simple procedure for the immobilisation of the linear polymer as an antibody substitute in a quantitative ELISA mimic was developed. Briefly, a commonly used physical adsorption procedure for the immobilisation of antibodies onto polystyrene microplates by hydrophobic binding was used. A stable coating was achieved by allowing the polymer solution loaded in each well to evaporate to dryness overnight at 37 °C in the dark. After testing the influence of the concentration of LP on the selective binding response signal, each well was coated with 50 µL of a 0.0091 mg mL-1 solution of the LP (1:10 dilution of the stock solution, equivalent to 455 ng-LP/well). The stability of the coating was confirmed after blocking with BSA (0.1% in 10 mM PB, pH = 7.4) and several washing steps (3 × 5 min) with Tween 20 (0.05% in 10 mM PB, pH = 7.4). Blocking refers to the pre-treatment of the wells before the addition of the labelled protein (LF-Bi) to reduce non-specific binding and improve the signal-to-noise ratio. Polyvinylpyrrolidone has also been tested as a non-protein alternative to conventional blocking buffer. However, the blocking was inadequate probably because polyvinylpyrrolidone is usually used for small proteins that could be masked by BSA; in our case, the target protein (LF) is a larger protein than BSA. The influence of ionic strength on the signal was also tested, and a 65% increase in nonspecific binding was observed when 0.1 M NaCl was added to the buffered medium. This behaviour may be related to the fact that ionic strength induces a reversal of the charge of LF and results in aggregate formation [37]. Therefore, the assays were performed in PB 10 mM (pH 7.4). Washing was applied after the incubation period to remove non-bound conjugates, so that the colorimetric measurement accurately reflects the amount of LF-Bi bound to the linear polymer layer deposited in the well. A wash was also performed between the addition of the different reagents to remove residual material that could cause cross-reactivity. The proposed concentration of Tween 20 (0.05%, 4.7 × 10-4 M) is higher than its critical micellar concentration (4.88 × 10-5 M) which would prevent the adhesion of the proteins to the polystyrene [38]. Thus, for concentrations higher than 0.05%, a decrease in the selective signal was observed. After the blocking step, the coated wells were incubated with LF-Bi for 90 min at 37 °C followed by washing and specific recognition via the high-affinity binding of biotin to HRPavidin (100 µL of a 1:2,000 dilution of the reagent for 10 min). Subsequently, the sample was washed, and the colour was developed by reaction with TMB for 10 min (100 µL of reagent). The enzymatic reaction was stopped by the addition of 100 µL of H2SO4 0.5 M solution. The uncoated wells were treated in the same way as the controls. The measurement of the characteristics of the interactions between LF-Bi and LP was achieved using three types of selective response binding assays. The kinetic experiments involved the measurement of the selective binding response of one or more concentrations of LF-Bi to the linear polymer at increasing time points. The analysis of a family of association kinetics curves allows the estimation of association (kon) and dissociation (koff) rate constants, as described in Section 2.4.2.1. We used four concentrations of LF-Bi (6, 10, 60, and 100 nM) and measured the selective binding responses for incubation times up to 300 min. The fitting of the individual selective response binding curves using the appropriate equation allowed the estimation of the observed rate constant (kobs) for each concentration of LF-Bi assayed (Fig. 4a). Then, from the linear regression fit of the kobs values versus the LF-Bi concentration, the equation kobs = 0.0002 × [LF-Bi] + 0.0040 (R2 = 0.9825) was obtained (Fig. 4b). This allowed the estimation of kon (slope) = 0.0002 min-1 nM-1, and koff (y-intercept) = 0.0040 min-1, and the estimation of the apparent affinity constant Ka of 20 nM as the ratio of koff/kon. The selective LF-Bi binding response was measured for 13 increasing concentrations of LF-Bi (0.1–250 nM), and the results are plotted as a function of the logarithm of the LF-Bi concentration, and a characteristic sigmoidal curve is shown in Fig. 5a. Fitting this concentration–response curve with a logistic equation allowed estimation of the location parameter of the curve as EC50 values and the corresponding Hill coefficients (nH, slope factors). The derived EC50 values were 11.8 ± 1.4 nM, and the nH values were 1.2 ± 0.3 (n = 10). The data were also analysed using an equation adapted to the particular situation of the present work from the model of the Langmuir isotherm for one-site binding equation: B = Bmax [L] / Ka + [L] (eq. 2) where B and L represent bound and free concentrations of a ligand, Bmax is the maximum binding, and Ka is the apparent equilibrium dissociation constant. The derived Ka values were 8.4 ± 0.8 nM (n = 10). These results are compatible with the notion that LF-Bi recognises a homogeneous population of independent sites on the synthetic receptor (i.e. LP). Furthermore, these results are consistent with the Ka value (20 nM) obtained for LF-Bi in the kinetic experiments. Finally, in homologous competitive binding experiments, the selective binding of a fixed concentration of LF-Bi can be measured at equilibrium in the presence of increasing concentrations of non-labelled LF. From these competitive inhibition curves, it is possible to determine the potency of LF indirectly from the obtained IC50 values, that is, the unlabelled ligand concentration that inhibits 50% of the response to the labelled molecule, as well as the simple or complex nature of the inhibition mechanism through the obtained nH values (slope factor or Hill slope). Our results show that unlabelled LF fully inhibited the selective LF-Bi response binding (at 10 nM, a concentration close to its EC50 value), and we obtained a value of 12.1 ± 1.5 nM for the IC50 value and nH value of 1.00 (Fig. 5b). Again, the results are indicative of the absence of a heterogeneous population of sites in the synthetic LP receptor labelled by LF-Bi and are consistent with the behaviour of a noncooperative synthetic receptor. Furthermore, the IC50 value of LF indicates that LF biotinylation does not adversely affect the binding of LF to LP because the IC50 is very close to the EC50 values obtained in the concentration dependence assays. Because of its small molecular weight (244 Da), biotin labelling does not significantly increase the size of the labelled protein and does not cause steric hindrance. Specifically, by using the biotinylation procedure described in Section 2.4.1, approximately three biotin molecules were incorporated per LF molecule. Thus, the three types of LF-Bi binding response assays performed with our ELISA-mimic protocol allowed the calculation of consistent sensitivity values for the synthetic receptor (LP) for the binding of the biotinylated target protein (LF-Bi). The results obtained for the Ka, EC50, and IC50 values of LF-Bi, irrespective of the type of binding assay, are comparable, indicating that the method has a high degree of repeatability. Using our ELISA-mimic protocol, competitive inhibition binding assays were performed for several proteins (Fig. 5b). In the LP specificity assays, the selective inhibition of the LF-Bi (10 nM) binding response by other competing proteins was examined. Interestingly, lysozyme and trypsin fully inhibited the selective LF-Bi binding response, yielding slope factors close to unity. For both protein ligands, the IC50 values showed relatively low binding affinities, whereas the nH values indicate a competitive mechanism of ligand inhibition of LF-Bi selective binding to LP. However, albumin was able to inhibit only 70% of the selective LF-Bi binding at a concentration of 75 µM, the maximal albumin concentration assayed (Fig. 5b). The IC50 values derived from the full competition curves were 0.62 ± 0.03 and 43.3 ± 1.7 µM for lysozyme and trypsin. In summary, the order of potency was LF, lysozyme, trypsin, and albumin. On the basis of these results, it was possible to calculate the cross-reactivity values of these proteins in our LF-Bi/LP assay: CRLZ = 1.95%, CRTr = 0.028%, and CRBSA = 0.016%. Furthermore, the data is useful to demonstrate the anti-interference ability of LF-Bi for the determination of LF in the presence of other proteins having similar molecular weight or isoelectric point. As observed in Fig. 5b, if 10 nM of LF-Bi is used, concentrations higher than 100 nM of Lysozyme, 1,000 nM of Trypsin, or 10,000 nM of BSA are required to decrease the absorbance intensity provided by the probe, which evidences that it presents excellent anti-interference ability to detect LF in the presence of other coexisting proteins as BSA and trypsin (approximately 3–4 logarithmic units). In addition, the interfering lysozyme concentration (1 log unit) could be really important in cases where lysozyme concentrations are elevated, for instance, in the case of acute myeloblastic leukemia [39] where urinary lysozyme levels exceed at least three times the expected normal values (60–136 nM). 3.3. Validation of the LF ELISA mimic for urine analysis Urine is an abundant and easily accessible body fluid, providing an ideal route for the non-invasive diagnosis of a number of human inflammatory and infectious diseases, particularly those of the urinary tract. The usefulness of the measurement of urinary LF for the diagnosis of urinary tract infections [40-42] and as a novel urine biomarker for the diagnosis and prognosis of urothelial carcinoma of the bladder has been reported previously [43]. For example, the mean concentration of LF has been found to be 30.4 ng mL−1 in healthy urine and 3,300 ng mL−1 in the urine of individuals suffering from infections [40,44,45]. From the results obtained in the concentration-dependent assays, it was possible to establish the limit of detection and quantification of the proposed ELISA mimic in PB as 18 and 60 ng mL−1, respectively. First, the reliability of the proposed assay for the determination of LF was tested using direct or pre-concentrated (10or 20-times) urine test samples. For the direct and 10-fold concentrated test urine samples, LF levels lower than the detection limit of the method were obtained. However, in the 20-fold concentrated samples, it was possible to determine LF levels at 7.3 ± 0.3 ng mL−1 in the selected urine test samples. These results indicate that the test urine can be used as a control matrix to evaluate the matrix effects of this method. For this, control urine samples were spiked with different concentrations of LF (0.1– 250 nM), and the developed procedure was applied. The quantification limit in urine was estimated at 1.5 nM (120 ng/mL). Method accuracy was also determined by the addition of a known amount of analyte to blank urine, and besides, for comparison, the same addition was made to PB instead of urine. Recovery was performed in triplicate and for three concentration levels (a concentration close to the limit of quantification, a concentration close to the EC50 y and another one ten times higher than this). The replicates were prepared independently from the beginning of the process, avoiding serial dilutions. The accuracy is expressed as a percentage calculated from the measured analyte concentration versus the added amount of analyte. Obtained recoveries were found to be 92.1 ± 2.8% for a concentration of 2 nM, 96.8 ± 1.8% for 10 nM and 94.2 ± 1.2% for 100 nM. The intra-day precision of the method was estimated from successive measurements under the same operating conditions for the three concentration levels mentioned above. The results expressed as RSD were 5.7% for 2 nM, 2.9% for 10 nM and 3.6% for 100 nM. The precision found for different assays performed thorough several weeks (inter-day precision) was 11.4%, 8.6% and 4.2% respectively for 2 nM, 10 nM and 100 nM concentration levels. Next, competitive binding experiments in the spiked test urine were performed, and, from the obtained competitive inhibition curve, the IC50, the concentration that inhibits 50% of the response to the labelled molecule in the urine matrix, was 54.0 ± 1.3 nM. This value is approximately four times higher than that observed when the assay were performed in parallel in PB medium (Fig. 6). Based on these results, which reveal matrix effects, the LF levels in the selected test urine sample (20-time concentrated) were determined to be 22.4 ± 1.8 ng mL−1. Nevertheless, despite the observed matrix effects, the proposed method is valid for the determination of urine LF levels in human patients suffering from inflammatory and infectious diseases of the urinary tract. In addition, we determined that the detection limit of the method can be improved by pre-concentrating the urine sample with Amicon Ultra-15 centrifugal filters with a 50-kDa cut-off. 4. Conclusions As a proof of concept, the synthesis of a new linear polymer by reversible deactivation radical polymerisation is proposed. The linear and flexible polymer can adapt its shape to the flat and rough surface topology of lactoferrin, our target protein. Thus, we proposed a method to produce and validate a robust, inexpensive, non-biological polymeric antibody for LF. This polymer was, thus, used as plastic antibody in an ELISA mimic for the determination of the LF concentration in real urine samples. The selectivity of the ELISA mimic for LF in the presence of other proteins of similar molecular weight or similar isoelectric point values was also tested and found the be acceptable. The advantages of using the proposed plastic antibody are that its synthesis is fast (hours) compared to natural antibodies (months), it is stable under storage at room temperature for several years and its linear range is intermediate compared to the commercial proposals (8 ng/mL-20 g/mL) with an analysis time of 3 h 20min and a precision of 3–11% (RSD). The synthesis could be easily performed in any laboratory and the polymer may be functionalized with fluorescent, catalytic or magnetic labels to be used in different biochemical assays. Declaration of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Acknowledgements Funding for this research was provided by the Spanish Ministry of Science, Innovation and Universities (project CTQ2017-85686-R) and by the Basque Government (Research Groups of the Basque University System, Project No IT 1186-19). The authors would like to thank the Alava Central Service Analysis belonging to SGIker (UPV/EHU) for their technical assistance. References [1] V.J.B. Ruigrok, M. Levisson, M.H.M. Eppink, H. Smidt, J. Van Der Oost, Alternative affinity tools: More attractive than antibodies?, Biochem. J. 436 (2011) 1–13. https://doi.org/10.1042/BJ20101860. [2] A. Yarman, K.J. Jetzschmann, B. Neumann, X. Zhang, U. Wollenberger, A. Cordin, K. Haupt, F.W. Scheller, Enzymes as tools in MIP-sensors, Chemosensors. 5 (2017). https://doi.org/10.3390/chemosensors5020011. [3] W. Chen, X. Tian, W. He, J. Li, Y. Feng, G. Pan, Emerging functional materials based on chemically designed molecular recognition, BMC Mater. 2 (2020) 1–22. https://doi.org/10.1186/s42833-019-0007-1. [4] L. Chen, X. Wang, W. Lu, X. Wu, J. Li, Molecular imprinting: Perspectives and applications, Chem. Soc. Rev. 45 (2016) 2137–2211. https://doi.org/10.1039/c6cs00061d. [5] I. Chianella, A. Guerreiro, E. Moczko, J.S. Caygill, E. V. Piletska, I.M.P. De Vargas Sansalvador, M.J. Whitcombe, S.A. Piletsky, Direct replacement of antibodies with molecularly imprinted polymer nanoparticles in ELISA - Development of a novel assay for vancomycin, Anal. Chem. 85 (2013) 8462–8468. https://doi.org/10.1021/ac402102j. [6] C. Cáceres, F. Canfarotta, I. Chianella, E. Pereira, E. Moczko, C. Esen, A. Guerreiro, E. Piletska, M.J. Whitcombe, S.A. Piletsky, Does size matter? Study of performance of pseudo-ELISAs based on molecularly imprinted polymer nanoparticles prepared for analytes of different sizes, Analyst. 141 (2016) 1405–1412. https://doi.org/10.1039/c5an02018b. [7] J. Xu, F. Merlier, B. Avalle, V. Vieillard, P. Debré, K. Haupt, B. Tse Sum Bui, Molecularly Imprinted Polymer Nanoparticles as Potential Synthetic Antibodies for Immunoprotection against HIV, ACS Appl. Mater. Interfaces. 11 (2019) 9824–9831. https://doi.org/10.1021/acsami.8b22732. [8] K. Smolinska-Kempisty, A. Guerreiro, F. Canfarotta, C. Caceres, M.J. Whitcombe, S. Piletsky, A comparison of the performance of molecularly imprinted polymer nanoparticles for small molecule targets and antibodies in the ELISA format., Sci. Rep. 6 (2016) 37638. https://doi.org/10.1038/srep37638. [9] E. Verheyen, J.P. Schillemans, M. Van Wijk, M.A. Demeniex, W.E. Hennink, C.F. Van Nostrum, Challenges for the effective molecular imprinting of proteins, Biomaterials. 32 (2011) 3008–3020. https://doi.org/10.1016/j.biomaterials.2011.01.007. [10] S. Ambrosini, S. Beyazit, K. Haupt, B. Tse Sum Bui, Solid-phase synthesis of molecularly imprinted nanoparticles for protein recognition, Chem. Commun. 49 (2013) 6746. https://doi.org/10.1039/c3cc41701h. [11] P. Çakir, A. Cutivet, M. Resmini, B.T.S. Bui, K. Haupt, Protein-size molecularly imprinted polymer nanogels as synthetic antibodies, by localized polymerization with multi-initiators, Adv. Mater. 25 (2013) 1048–1051. https://doi.org/10.1002/adma.201203400. [12] Y. Ge, A.P.F. Turner, Too large to fit? Recent developments in macromolecular imprinting, Trends Biotechnol. 26 (2008) 218–224. https://doi.org/10.1016/j.tibtech.2008.01.001. [13] M. Pan, L. Hong, X. Xie, K. Liu, J. Yang, S. Wang, Nanomaterials-Based Surface Protein Imprinted Polymers: Synthesis and Medical Applications, Macromol. Chem. Phys. 222 (2021) 1–18. https://doi.org/10.1002/macp.202000222. [14] A. Huang, J.M. Paloni, A. Wang, A.C. Obermeyer, H. V. Sureka, H. Yao, B.D. Olsen, Predicting Protein-Polymer Block Copolymer Self-Assembly from Protein Properties, Biomacromolecules. 20 (2019) 3713–3723. https://doi.org/10.1021/acs.biomac.9b00768. [15] A. Rodriguez-Abetxuko, D. Sánchez-deAlcázar, P. Muñumer, A. Beloqui, Tunable Polymeric Scaffolds for Enzyme Immobilization, Front. Bioeng. Biotechnol. 8 (2020) 1– 27. https://doi.org/10.3389/fbioe.2020.00830. [16] A. Nieto-Orellana, M. Di Antonio, C. Conte, F.H. Falcone, C. Bosquillon, N. Childerhouse, G. Mantovani, S. Stolnik, Effect of polymer topology on non-covalent polymer-protein complexation: miktoarm versus linear mPEG-poly(glutamic acid) copolymers, Polym. Chem. 8 (2017) 2210–2220. https://doi.org/10.1039/c7py00169j. [17] S.J. Koch, C. Renner, X. Xie, T. Schrader, Tuning Linear Copolymers into ProteinSpecific Hosts, Angew. Chemie. 118 (2006) 6500–6503. https://doi.org/10.1002/ange.200601161. [18] A. Motib, A. Guerreiro, F. Al-Bayati, E. Piletska, I. Manzoor, S. Shafeeq, A. Kadam, O. Kuipers, L. Hiller, T. Cowen, S. Piletsky, P.W. Andrew, H. Yesilkaya, Modulation of Quorum Sensing in a Gram-Positive Pathogen by Linear Molecularly Imprinted Polymers with Anti-infective Properties, Angew. Chemie. 129 (2017) 16782–16785. https://doi.org/10.1002/ange.201709313. [19] Y. Wada, H. Lee, Y. Hoshino, S. Kotani, K.J. Shea, Y. Miura, Design of multi-functional linear polymers that capture and neutralize a toxic peptide: A comparison with crosslinked nanoparticles, J. Mater. Chem. B. 3 (2015) 1706–1711. https://doi.org/10.1039/c4tb01967a. [20] J.M. Rini, U. Schulze-Gahmen, I.A. Wilson, Structural evidence for induced fit as a mechanism for antibody-antigen recognition, Science (80-. ). 255 (1992) 959–965. https://doi.org/10.1126/science.1546293. [21] B. Panganiban, B. Qiao, T. Jiang, C. Delre, M.M. Obadia, T.D. Nguyen, A.A.A. Smith, A. Hall, I. Sit, M.G. Crosby, P.B. Dennis, E. Drockenmuller, M. Olvera, D. Cruz, T. Xu, Random heteropolymers preserve protein function in Foreign Environments., Science (80-. ). 1243 (2018) 1239–1243. [22] A. Alexander-Katz, R.C. Van Lehn, Random copolymers that protect proteins, Science (80-. ). 359 (2018) 1216–1217. https://doi.org/10.1126/science.aat0155. [23] Y.A. Kriksin, P.G. Khalatur, A.R. Khokhlov, Recognition of complex patterned substrates by heteropolymer chains consisting of multiple monomer types, J. Chem. Phys. 124 (2006). https://doi.org/10.1063/1.2191849. [24] A. Poma, A. Guerreiro, S. Caygill, E. Moczko, S. Piletsky, Automatic reactor for solidphase synthesis of molecularly imprinted polymeric nanoparticles (MIP NPs) in water, RSC Adv. 4 (2014) 4203–4206. https://doi.org/10.1039/c3ra46838k. [25] A. Poma, A. Guerreiro, M.J. Whitcombe, E. V. Piletska, A.P.F. Turner, S.A. Piletsky, Solid-Phase Synthesis of Molecularly Imprinted Polymer Nanoparticles with a Reusable Template-"Plastic Antibodies", Adv. Funct. Mater. 23 (2013) 2821–2827. https://doi.org/10.1002/adfm.201202397. [26] A. Bagheri, C.M. Fellows, C. Boyer, Reversible Deactivation Radical Polymerization: From Polymer Network Synthesis to 3D Printing, Adv. Sci. 8 (2021) 1–16. https://doi.org/10.1002/advs.202003701. [27] H. Zhang, Recent Advances in Macromolecularly Imprinted Polymers by Controlled Radical Polymerization Techniques, Mol. Imprinting. 3 (2016) 35–46. https://doi.org/10.1515/molim-2015-0005. [28] P. Bonomi, M.D. Attieh, C. Gonzato, K. Haupt, A New Versatile Water-Soluble Iniferter Platform for the Preparation of Molecularly Imprinted Nanoparticles by Photopolymerisation in Aqueous Media, Chem. - A Eur. J. 22 (2016) 10150–10154. https://doi.org/10.1002/chem.201600750. [29] M.J. Garcia-Soto, K. Haupt, C. Gonzato, Synthesis of molecularly imprinted polymers Figure 4 Figure 5 Figure 6