Hybrids Molecularly Imprinted Polymers: The Future of Nano-

Molecularly imprinted polymers (MIPs) have been widely used in nanomedicine during 10 the last few years. However, their potential is limited by their intrinsic properties resulting, for in11 stance, in lack of control in drug release processes or complex detection for in vivo imaging. Recent 12 attempts in creating hybrid nanomaterials combining MIPs with inorganic nanomaterials succeeded 13 in providing a wide range of new interesting properties suitable for nanomedicine. Through this 14 review, we aim to illustrate how hybrids molecularly imprinted polymers may improve patient care 15 with enhanced imaging, treatments and combination of both. 16

. MIPs synthesis principle. The target used as a template is added to a polymer mixture, forming pre-polymerization complexes. After the initiation of the polymerization, the template is embedded inside the polymer matrix and can be removed, forming highly specific cavities. Created with BioRender.com. elements and even achieved picomolar limits of detection [29]. 116 Lately, MIPs mimicking biological functions [30] with, for instance, enzymes like activity [31], have been 117 designed. In 2013, an assay similar to the enzyme-linked immunosorbent assay (ELISA) was developed, replacing the 118 antibodies coating microplate wells with molecularly imprinted polymer. The sensitivity of the assay was three orders 119 of magnitude better than a previously described ELISA assay based on antibodies (2.5pM), with more stability and 120 easier implementation as no cold-chain logistic was required.
In terms of affinity and specificity, MIPs possess 121 binding characteristics similar to those of antibodies or biological receptors, which makes them often described as 122 "synthetic antibodies" [32]. Replacing antibodies with MIPs nanoparticles in ELISA-like and other similar assays has 123 been widely studied, leading to tests with a picomolar LOD similar to antibodies-based assays (LOD for biotin: 1.2pM 124 for NanoMIPs versus 2.5pM for the antibody assay, LOD for fumonisin B2: 6.1pM versus 25pM) [33,34,35]. Dissociation 125 constants (Kd), depending on the monomers used and the nature of the template are often in the nanomolar range : 126 0.48nM for NanoMIPs against vancomycin [36]. Hence, they might be used for a great number of medical applications 127 that previously required tailor-made antibodies or for which no antibodies could be developed. For example, MIPs 128 sensors can directly be applied to diagnose diseases: in 2016 Lieberzeit et al. [37] developed a MIP based QCM sensor 129 able to specifically trap lipoproteins, a cardiovascular biomarker as its concentration in human serum is inversely 130 correlated to high risk of coronary disease. This device, applied to human samples, led to similar results than a standard 131 homogenous enzymatic assay. 132 As MIPs show efficiency in biological sensors, scientists are now interested in their use in the medical field. 133 However, their synthetic process has to be modified and adapted to be compatible with the desired application, with 134 an emphasis on biocompatibility when aiming for in vivo applications. 135

136
The origin of molecularly imprinted polymers can be traced as far as the 1930s when a soviet researcher named 137 Polyakov discovered an improvement of the adsorption of benzene derivatives on silica particles when prepared in 138 their presence [38]. During the following years, the study of the formation process of those imprinted polymers led to 139 enhanced recognition properties and in the 1970s, Günter Wulff synthesized an organic polymer able to separate 140 racemates [39]. When designing an imprinted polymer, different imprinting strategies are employed, depending on the 141 application [40], as the morphology of the imprinted polymer can affect recognition properties or potential application 142 [41].
imprinted polymer able to separate amino acid derivatives [45]. With this technique, improved selectivity may be 155 obtained with mixtures of monomers [42]. This method however tends to produce polymers with a less homogenous 156 repartition of binding sites than through a covalent approach. 157 At their origin, MIPs were synthesized through bulk polymerization, forming macroporous polymer networks 158 with the ability to entrap molecules. Those polymers had to be grinded to obtain small particles and create a large 159 surface area for recapture. However, this step is time consuming and results in irregular shapes and sizes as well as 160 heterogeneities in the binding sites repartition. 161 Obtaining monodispersed polymer particles is a key condition to improve reproducibility. Conducting the 162 polymerization in a biphasic system is a great option to obtain regular shapes and sizes. such as silica during the process, enable a surfactant free-polymerization ( Figure 3 ). The process has been employed in 169 the synthesis of imprinted hydrogel microbeads able to passively release the nucleotide adenosine 5′-monophosphate 170 for an application in cosmetology [47]. In 1999, Mosbach et al. [48] first described the precipitation polymerization technique which produces uniform 173 sub-micrometer imprinted particles. The resulting molecularly imprinted microspheres were highly specific of their 174 target, theophylline or 17β-estradiol, and had higher adsorption capacities compared to the particles obtained by 175 grinding an imprinted monolith. This technique has also been successfully employed for many medical purposes such 176 as the design of a transdermal formulation of a MIP with the ability to release encapsulated nicotine [49]. The 177 polymerization process occurs in a solvent excess and stops when the particles reach a critical size which makes them 178 precipitate. However, the dilution tends to decrease interactions between the template and the functional monomers 179 leading to less selectivity.  and cross-links are homogeneously distributed [51]. However, the material presents less selectivity as the steric 190 hindrance, caused by the binding to the silica surface, prevents complementary interactions to develop [52]. This 191 technique has been adapted by Sellergren et al in 2011 [53], for the imprinting of proteins (human serum albumin and 192 immunoglobulin G) and can supposedly be used for applications in nanomedicine but to the best of our knowledge, 193 none is currently under development. 194 The molecular imprinting process being cheaper than the use of antibodies to target a given molecule, it remains 195 nonetheless expensive for biomedical applications as the cost of interesting proteins or drugs is often non-negligible. It 196 is also well known that proteins can degrade in organic solvent and that their structure, a key factor in protein 197 recognition and, hence, for an efficient protein imprinting, is altered by temperature or pH changes. The imprinting of 198 proteins, beyond the difficulties regarding their stability, tends to produce higher cross-reactivity and insufficient 199 extraction after the polymerization due to the size of the protein. The use of epitopes, small region of a protein that are 200 used as a recognition site may be employed to address those particular issues [54,55,56]. 201 Moreover, for medical applications, it is necessary to work with small and regular objects : specific methods 202 have to be designed such as the solid-phase synthesis developed by Ambrosini et al. [57] Instead of having the template 203 in solution, the proteins are immobilized onto glass beads (100µm) (step 1 Figure 4). The beads are then packed into a 204 reactor with the reaction media containing monomers, initiator and a crosslinker (step 2 Figure 4). After synthesis, the 205 residual monomers and low affinity nanoMIPs are washed away (step 3 Figure 4) before extracting high-affinity 206 nanoMIPs from the templates by hot washing or thermo-responsive swelling (step 4 Figure 4). Their attempt with 207 trypsin resulted in high specificity and selectivity toward trypsin, without any surfactant used that could alter a protein 208 conformation during the imprinting, and the release of the MIP from the functionalized glass beads is simply induced 209 by a temperature change.  Created with BioRender.com. 215 Immobilization of the template through an affinity ligand [58] or metal chelate functionalization of the glass-216 beads [59], instead of direct chemical attachment to the support, enables an oriented binding to the beads surface.

217
The solid-phase synthesis has been employed by other groups [60,61], with attempts to automatize the 218 synthesis process and to improve the binding properties of the MIPs. This process increases the binding properties as 219 non-imprinted nanoparticles are washed before releasing MIPs, less dialysis is required to remove the template and the 220 reusable aspect of the glass beads [62] tends to decrease the cost of this technique. However, the glass beads suffer 221 several drawbacks. First, their size being between 70 and 100µm, the surface/volume ratio is not optimal, leading to a 222 smaller number of proteins immobilized. Interactions between the template and the monomer were also limited by the 223 fact that stirring has to be avoided as the beads are prone to abrasion, causing leaching of the template and nanoglass 224 undesired particles. 225 As an alternative to address those issues, Sun et al. [63] replaced glass beads (70-100µm) with magnetic 226 microspheres (600-700nm). Greater immobilization of trypsin is obtained, less solid material is employed and the 227 nanoMIPs show high affinity and selectivity for the template protein while the yield increased compared to 228 conventional solid phase synthesis ( Figure 5). 230 Solid-phase synthesis hence presents two main advantages compared to historical techniques that are really 231 useful for medical applications : the fixation onto a surface, enabling more control over the orientation of the template 232 during polymerization, and the glass beads ability to be reused, reducing the cost of the imprinting process. The 233 possibility to automatize this process is also very interesting in the perspective of an industrialization of the production 234 of this type of materials. 235 However, possible applications for those materials are limited by their inherent physical and chemical 236 properties. Combining molecularly imprinting polymer to inorganic nanoparticles could lead to a novel class of hybrid 237 materials with very interesting properties for nanomedicine, especially if they respond to external stimuli. As a support, 238 those nanomaterials provide new synthesis pathways to those used for MIPs synthesis as described in the next part. In the case of a drug encapsulation inside a polymer matrix for passive delivery [65,66], the molecule 251 is protected from enzyme degradation during its transit through the body thanks to the crosslinked polymer shell. 252 It is also possible to combine passive drug release to targeting using MIPs as described by  controlled manner and quantum dots exhibiting luminescent properties. 281 Gold nanoparticles have been widely explored as therapeutic agents for displaying several coveted properties. 282 In cancer therapy, it is possible to use gold nanoparticles as antineoplastic agents as they are capable of altering cell 283 cycles [70] through their capacity to enhance radiation sensitivity, and vascularization processes [71] thanks to their 284 intrinsic property: they can interact selectively with heparin-binding glycoproteins and inhibit their activity. Gold 285 nanoparticles may also be used for photothermal therapy : the excitation of surface plasmon oscillation through light 286 exposition generates local heat release around nanoparticles [72]. This property can be used for cancer therapy [73,74]   nanoparticles and laser radiation delivered through bronchoscopy. The coupling of gold nanoparticles with molecularly 289 imprinted polymers may lead to new triggered release techniques using thermosensitive polymers and the 290 photothermal properties of the particles. 291 Another class of nanoparticles that would provide interesting properties for MIPs hybrids is superparamagnetic 292 iron oxide nanoparticles (SPIONs). Due to their size and superparamagnetic properties, strong magnetization at low 293 magnetic field without remanence when the field is off, SPIONs may be employed as MRI contrast agents. For liver 294 imaging two solutions, approved by the FDA after 1996, were commercially available : Feridex® and Resovist®, now 295 no longer employed [75]. Superparamagnetic particles are currently more studied for their ability to magnetically guide 296 carriers in vivo or generate heat upon alternating magnetic field (AMF) stimulation, so called magnetic hyperthermia 297 [76]. NanoTherm® (MagForce Ag) is a SPION based technology using magnetic hyperthermia properties [75], approved 298 in Europe for the treatment of glioblastoma and currently evaluated in the USA for the treatment of prostate cancer. 299 It is hence possible to combine such nanoparticles with thermosensitive polymers for a triggered drug release 300 or cell-targeted hyperthermia therapy. The most widely employed magnetic material to be coupled to MIP is magnetite, 301 Fe3O4, due to its excellent characteristics, such as low toxicity, easy synthesis and low production cost. MIP may also be 302 coupled to maghemite, γ-Fe2O3, less employed due to the additional oxidation step required but presenting the same 303 main characteristics as Fe3O4 on top of being chemically more stable. The main challenges with those particles are long-304 term chemical stability and functionalization. To address those issues, it is possible to add an organic (polymer, 305 dendrimer) or inorganic (silica, gold) shell before the MIP functionalization which will provide a protection against 306 oxidation as well as new functionalization pathways. 307 Another interesting property that can be added to MIP is fluorescence emitted by quantum dots. A quantum 308 dot (QD) is a nanosized crystal of inorganic semiconductor that has size-tunable, narrow, Gaussian emission spectra 309 excited at a single wavelength, enormous absorption extinction coefficients and high fluorescent quantum yields. To synthesize those hybrid materials, it is possible to do a bulk-like polymerization, where the template is added 320 with the monomer and the core, or to combine covalent and non-covalent approach by grafting the template on an 321 inorganic surface such as silica nanoparticles [83,84]. The latter, as described by Liu et al. [64], require additional steps 322 to ensure the template immobilization prior to polymerization. This method is less time-efficient but is preferred if a 323 specific orientation of the template is necessary or if the one-pot process is not successful. In the case of sialic acid 324 imprinted polymer described by Liu et al. [84] no MIP layer is formed using the bulk process, probably due to a lower 325 pH induced by the sialic acid, slowing down the polymerization process, but pre-immobilization of the template solved 326 this particular issue. Additionally, the intrinsic properties of the nanoparticles may be exploited during the synthesis : 327 coating the fluorescent CDs ( Figure 8) [85]. 329 A wide range of new applications for molecularly imprinted polymers can be developed when they are 330 combined to inorganic materials but the grafting of MIPs onto the surface of inorganic nanoparticles also results in core-331 shell structures that allow more control over the size and distribution of the material. 332

333
The combination of a polymer matrix with nanostructures usually provide additional properties [86]. However, 334 the complexity of a material may increase the number of possible side effects. The incorporation of inorganic 335 nanoparticles that are themselves already suspected to induce some toxicity [87,88] has to be considered when 336 evaluating the benefice/risk balance of such innovation. However, it has to be noted that cells experiment, as a model, are limited and that more accurate results, obtained 351 through small animal experiments, will be required to assess the potential risk of those materials. In 2017, hyaluronan and salicylic acid nanoMIPs were employed as potential indicators of pathological 381 condition to localize hyaluronan and sialylation sites on fixed and living human keratinocytes [95]. The quantum dots 382 coated with MIPs were non-cytotoxic and did not affect cell viability or morphology, offering a promising tool for 383 bioimaging on living tissue (Error! Reference source not found.).

384
Hybrid MIPs may also be employed as surface-enhanced Raman scattering (SERS) tags for bioimaging [86]. 385 First developed by Yin et al [96] to target cancer cells, their method consists in the imprint of sialic acid onto Raman-386 active silver nanoparticles. After laser excitation, the resulting SERS signals from the MIP nanoparticle were able to 387 differentiate cancer cells and tissues from normal ones. Similarly, gold nanorods coupled to EGFR imprinted polymers 388 have been employed for live cell Raman imaging [86] and could be employed for bioimaging (Figure 10

390
As MIPs can selectively and specifically adsorb molecules, they can be used for passive or active drug delivery. 391 The most impressive advances toward controlled release have been obtained through the combination of MIPs 392 with inorganic materials. Sometimes, the inorganic part only serves as a support for a clever release mechanism such as 393 developed in 2016 by Zhang et al. [97] : a silica coated MIP encapsulating DOX and containing sulfur-sulfur bonding. 394 In acidic pH and high concentration of glutathione (GSH), the S-S bonds break, facilitating drug release. Similar 395 mechanisms were then exploited in 2018 by Yang et al. [84] in a material where the silica core is removed by degradation 396 of the S-nitrosothiol polymer by GSH at low pH generated nitrous oxide which acted as an anticancer drug (Figure 397 11). 398 But beyond facilitating the synthesis, the addition of inorganic particles often provides new tools for triggered 399 drug release. 400 In 2014, UV was used to trigger drug release from asymmetric MIP-silver nanoparticles (Janus particles) 401 synthesized via a wax−water pickering emulsion. The emulsion allowed a one-sided silver coating of MIPs nanoparticle. 402 Release of the drug from the Janus MIP particles is controlled by switching on−off the UV illumination [98]. However, 403 this system lacks biocompatibility due to the application of Ag NPs and the limitation of UV illumination which 404 damages cells. Therefore, adjustments are required for practical applications. For example, we could imagine replacing 405 Ag NPs by gold nanorods which are nontoxic and more biocompatible. Magnetic MIP have also emerged as a powerful material for controlled drug delivery systems because they can 407 be localized to the delivery sites using a magnet [99] and release the drug to particular sites through passive diffusion. 408 For instance, magnetic molecularly imprinted polymers were synthesized for special recognition and slow release of 409 aspirin (ASP) [100]. The synthesized MIPs have high magnetic responding capacity, which enable themselves to be 410 separated from suspension by an external magnetic field. The magnetic MIPs exhibit good special binding and 411 selectivity capacities to the template molecule. The ASP-loaded magnetic imprinted or non-imprinted polymers show 412 a controlled release of aspirin ( Figure 12). In particular, the aspirin release is slower for the imprinted polymer, 413 highlighting the interest of the imprint for passive drug release to limit the burst release effect. However, the main interest of the molecularly imprinted polymers combination with maghemite/magnetite 417 reside in their core processing photothermia [101] and hyperthermia properties [76]. While most studies focus on passive 418 delivery, it is possible to use the properties of magnetic nanoparticles to trigger the delivery of a drug or to kill specific 419 cells through heat exposure. 420 In this context, in 2015, an innovative magnetic MIP nanomaterial for triggered cancer therapy showing active 421 control over drug release by using an alternating magnetic field was developed [102,103]. Upon AMF exposure, the 422 hydrogen bonds between the MIP and the encapsulated drug, doxorubicin, are broken and the molecule is released 423 without any significant heating of the medium. After AMF application, cancer cell viability is reduced to 60% after 1h30 424 treatment in athermal conditions while the control cells do not suffer any mortality (Error! Reference source not found.). 425 The same behavior was obtained with magnetic silica imprinted polymer. [104]

448
New systems, combining both aspects of MIPs -the ability to efficiently target and to encapsulate therapeutic 449 molecules to be delivered -are starting to spread in the theranostic field. The main idea is to combine a biological marker 450 characteristic of a pathological condition as a target and a therapeutic molecule to enhance the potency of a treatment 451 while imaging and reducing potential side effects. This approach has been widely used in cancer therapy as drugs 452 employed in chemotherapy are also toxic for healthy cells, resulting in acute side effects. 453 The recent development of technologies combining molecularly imprinted polymers with inorganic core 454 materials is so versatile that it allows a wide range treatment option. In particular, those materials could specifically 455 target any given tissue or cell types that possess a specific marker and that such marker possesses the required stability 456 to be imprinted. Like non-hybrid materials, they can be combined with a drug loading inside the polymer matrix but 457 additional features are available when using an appropriate core material. The possibilities seem endless and are 458 starting to be broadly explored in the literature. 459 In 2019, Zhang et al. [106], developed an innovative silica based material that could be used for targeting, drug 460 delivery and imaging. A biomarker, human fibroblast growth-factor-inducible based, coupled with an anticancer, 461 bleomycin, were imprinted on a silica core for its good optical properties. In vivo experiments showed an excellent 462 inhibition of the growth of BxPC-3 xenograft tumor. Similar models have been developed against other cancer cells, like 463 for instance one targeting the HER2 receptor and delivering DOX [107]. It leds to similar results in vivo, illustrating the 464 versatility of this approach. 465 It is also possible to use the magnetic properties of a material to selectively target an organ or a tumor instead 466 of imprinting a surface marker. The ability of SPIONs to accumulate in a specific site through magnetic guidance has 467 been successfully demonstrated in a study without the imprinting of molecules [108]. In a recent study, a 5-fluorouracil, 468 anticancer drug with fast degradation rates, loading novel multi core-shell structure nanocarriers based on a cross linker 469 of tannic acid has been synthesized [109]. 470 Fluorescent imaging with a small animal imaging instrument confirmed the successful conduction of the 471 carrier into the liver by applying an external magnetic field. However, as liver and kidney tend to naturally accumulate 472 nanoparticles, the choice of liver located tumors might not be the most appropriate to illustrate an efficient targeting 473 ( Figure 14). In the past, molecularly imprinted polymers have been mostly used in sensors or for analytical purposes but 481 during the last decade, great developments were made in order to use them in nanomedicine. The synthesis techniques 482 have evolved in order to ensure the biocompatibility of the material, to control the size or even the orientation of the 483 imprint. As MIPs nanoparticles are cheaper than antibodies, with an increased stability and more adapted to mass 484 production with alternatives to reuse the template proteins [110] cancer marker, to inorganic cores that heat in response to an external stimulus will produce an effective material for 497 selective cancer destruction. All-in-one materials, combining targeting, imaging, and drug delivery seems to be within 498