Article | Published:

# Magnetic field remotely controlled selective biocatalysis

## Abstract

Many applications for medical therapy, biotechnology and biosensors rely on efficient delivery and release of active substances. Here, we demonstrate a platform that explores magnetic-field-responsive compartmentalization of biocatalytic reactions for well-controlled release of chemicals or biological materials on demand. This platform combines two different kinds of core–shell magnetic nanoparticle: one loaded with enzymes and another with substrate-bound therapeutic (bio)chemicals. Both cargos are shielded with a polymer brush structure of the nanoparticle shell, which prevents any enzyme–substrate interactions. The shield’s barrier is overcome when a relatively weak (a fraction of 1 T) external magnetic field is applied and the enzyme and the substrate are merged and forced to interact in the generated nanocompartment. The merged biocatalytic nanoparticles liberate the substrate-bound therapeutic drugs when the enzymes degrade the substrate. The developed platform provides a proof of concept for the remotely controlled release of drugs or (bio)chemicals using the energy of a non-invasive, weak magnetic field.

## Main

Materials that efficiently release biological molecules or therapeutic chemicals on demand using exposure to remotely controlled and safe external sources of energy, such as magnetic fields, could find applications for drug delivery1, biotechnology2,3 and biosensors4. Because live tissue and synthetic polymers are not responsive to weak magnetic fields, the development of magnetic-field-responsive soft materials has been reported by combining magnetic nanoparticles and stimuli-responsive soft materials5. Magnetic nanoparticles interact with magnetic fields and transduce magnetic field energy into physical or chemical changes in the soft material. Materials that control enzymatic processes are one example of such soft materials. Enzymes are extensively used to change or degrade colloidal particles, capsules, and their assemblies to trigger release of the cargo via biocatalytic reactions6,7.

In all eukaryotes, metabolic pathways are precisely organized and regulated. This precise control is based in part on the high selectivity of biocatalytic reactions and controlled transport of chemicals and biomacromolecules across membranes that compartmentalize cells, organelles and organs. Highly selective biocatalysis alone cannot orchestrate complex systems of biochemical reactions without the supporting role of signal-triggered synthesis, release, secretion, conversion and degrading processes that take place in different compartments in cells and organs. Despite being highly selective, enzymes cannot provide 100% selectivity. In particular, enzymes could interact with a number of substrates of a similar chemical structure (for example, proteases are highly promiscuous catalysts), be degraded by other enzymes or even by self-digestion upon secretion into a complex biological environment, or undergo undesired aggregation, crystallization or nonspecific adsorption, which would strongly damage the efficiency of the biocatalytic process. However, the overall high specificity of biocatalytic processes is strengthened by localizing the enzymatic reactions within a specific environment and spatial compartments.

Inspired by this hierarchical design in live systems, diverse stimuli-responsive functional materials have been reported, involving various architectures that respond to changes in magnetic fields8,9,10. However, it remains challenging to create a reactive system that preserves enzyme molecules from destructive environments and undesired interactions while being able to initiate the designated reaction when needed. Different approaches have been developed to preserve enzymes for storage and delivery before activating them on demand in a magnetic field at the targeted location. A number of studies aimed at controlling the kinetics of biocatalytic reactions in model systems11,12,13,14,15 have explored magnetic-field-triggered changes of the local concentration and mobility of enzymes. However, it is difficult to apply many of such approaches to live tissue because of limitations associated with degradation of many biological molecules in complex biological milieu, toxicity of the materials and a narrow variation range of physiological conditions. The most common approach is to embed magnetic nanoparticles in a thermoresponsive material and expose them to an alternating magnetic field. The local temperature rise due to transformation of electromagnetic energy into heat is used to trigger changes in the thermoresponsive material without a need to elevate the temperature of the entire system16. This scenario is not always appropriate for enzymes because of their generally poor thermal stability.

We report here a proof-of-concept study of magnetic-field-controlled biocatalysis, which does not rely on commonly used alterations of local temperature, pH, salt concentration or light absorbance. Our biocatalytic platform uses a biomimetic concept of compartmentalization, magnetic-field-controlled transport and interactions of substrates and biocatalysts across semi-permeable walls of the compartments. In our system, the biocatalytic process is achieved via magnetic-field-triggered interactions when two distinctive nanoparticles—one loaded with an enzyme (for example, protease) and another with a substrate (for example, polypeptide)—are brought into the merging vicinity of each other. The enzyme and substrate molecules bound to the nanoparticles are framed by semi-permeable barriers with gating properties—polymer brushes—which prevent interactions of both the enzyme and the substrate with other competitive molecules and with each other. This design resembles compartments that accommodate and preserve the substrate and enzyme from interactions ahead of time. The biocatalytic reaction is turned on only in the presence of a magnetic field that triggers merging of the compartments. In addition, the compartments are tailored to provide the most favourable environment for the enzymatic process, for example, an acidic environment favourable for hydrolytic reactions catalysed by many proteases could be achieved using a weak polyacid brush architecture even if the environment outside the polyacid brush was buffered at pH 717. The method proposed here could be realized by using either uniform or non-uniform magnetic fields generated by a permanent magnet or an electromagnet, with a low strength magnetic field achievable in biological systems with remote magnet positioning.

## Results

### Design of biocatalytic nanoparticles

Here, we demonstrate that papain—a highly promiscuous protease—can be utilized in a selective biocatalytic process using our concept of magnetically controlled biocatalysis. According to this concept, we perform covalent conjugation of enzymes and polymer brushes linked to nanoparticles. The conjugation has minimal effect on the chemical structure, conformation and hence the specificity of the enzyme. We change only topological aspects of the biocatalytic reaction when we secure interactions of the enzyme with the designated substrate by exploring specific architecture of the compartmentalized system. The concept is realized by using the architecture of a spherical core–shell nanoparticle (NP). The system contains two kinds of NP: E and S, where E-nanoparticles are loaded with the enzyme and S-nanoparticles are loaded with the substrate. The two NPs possess a very similar architecture: a superparamagnetic core is enveloped by a silica shell with a grafted block-copolymer brush. The grafted block is polyacrylic acid (PAA) and the external block is a polymer of poly(ethylene glycol methyl ether acrylate) macromer (PPEGMA) with an average number of ethylene glycol monomeric units per macromer of 9.3. The molecular mass of PAA-b-PPEGMA block copolymer is 17 kg mol–1 where the PAA and PPEGA blocks are 8.5 kg mol–1 each. Carbodiimide conjugation chemistry was used to covalently bind papain and the substrate—fluorescent dye (FD)-labelled bovine serum albumin (FD-BSA)—to the E- and S-nanoparticles, respectively. A cryo-transmission electron microscopy (cryo-TEM) image of a two-particle aggregate comprising E- and S-nanoparticles and a schematic of the concept are presented in Fig. 1. The nanoparticle characteristics are listed in Table 1 and Supplementary Tables 1 and 2. NPs carry 20% wt of PAA-b-PPEGMA and about 30% wt of proteins. The details of characterization can be found in Methods and Supplementary Methods.

The architecture described above enables the magnetic-field-triggered proteolysis of FD-BSA, which results in the liberation and release of FD. The latter can be detected using fluorescence spectrometry (Fig. 2). Indeed, FD release took place immediately after the magnetic field was turned on in contrast to the reference experiments when a blend of E- and S-nanoparticles experienced no magnetic field. Both E- and S-nanoparticles are mixed and coexist in the same container as a stable aqueous dispersion with no interactions if the magnetic field is off (Fig. 2b).

The fluorometric data suggest that high selectivity of the magnetically triggered biocatalytic reaction is achieved owing to the NPs’ unique architecture. The zero-field control experiments demonstrate that E- and S-nanoparticles do not interact in the aqueous suspension. Papain proteolytic behaviour is not specific to BSA. At the same time, FD-BSA is a substrate that is not specific to papain. FD-BSA could be degraded by other proteases while papain could degrade various proteins. Thus, in complex biological systems that contain various proteins, the reaction between papain and FD-BSA is not selective and not controlled. This reaction is one of many parallel and competitive reactions catalysed by different proteases, and it begins immediately upon mixing the ingredients. However, the specially crafted structure of E- and S-nanoparticles was used here to convert non-selective proteolysis into a selective reaction that rejects all other proteins and turns off the self-digestion of the proteolytic enzyme. Thus, this architecture excludes ‘unauthorized’ reactions of papain and FD-BSA (Fig. 2b, inset). Importantly, the biocatalytic activity of papain bound to PAA is not compromised, as was shown in reference experiments with the conjugated enzyme (Supplementary Figs. 10  and 11 and Supplementary Table 3). These conclusions were confirmed using gel-electrophoresis to monitor release of degraded fragments of the BSA conjugates. We observed release of BSA fragments only if a mixture of S- and E-nanoparticles was exposed to the magnetic field. The experiments with zero-field and for mixtures with native enzymes in solution did not reveal essential release of BSA fragments (Supplementary Figs. 1315).

### Magnetic-field-controlled biocatalysis

The brush-like structure of the nanoparticle shell is a key element to achieving control over the biocatalytic reaction and cargo release and to turning the papain–FD-BSA reaction into a selective biocatalytic reaction. Owing to the brush architecture, the enzyme can reach the biocatalytic nanocompartment with the designated substrate only if the magnetic field is turned on, as discussed above.

In a magnetic field, dipole–dipole interactions between superparamagnetic NPs result in the formation of aggregates that grow and form chain-like structures. The size of the aggregates depends on NP concentration and exposure time to the magnetic field as visualized by cryo-TEM, atomic-force microscopy (AFM) and dark-field optical microscopy (Fig. 3a–c). TEM images of the aggregates are presented in Supplementary Fig. 12. For a 1:1 blend of E- and S-nanoparticles, the probability of finding two different NPs that form an E–S sequence in two-particle aggregates or in the chain is 50%. While the magnetic field is on, the chain structure is growing in length and diameter, eventually consuming all the magnetic NPs in the solution. The probability of finding at least one E–S sequence in each aggregate approaches 1 for large aggregates so that virtually each particle aggregate is involved in the biocatalytic reaction. Chains of dipolar particles are subject to strong Landau–Peierls thermal fluctuations18,19,20. Theoretical analysis and experiments revealed that the field-induced aggregation process of dipolar particles is a complex phenomenon and the structure of aggregates depends on the balance of magnetic forces, thermal fluctuations and other interaction mechanisms between NPs. NPs in the aggregates may undergo rearrangements21,22,23,24. The dynamics of the aggregates facilitates an increased efficiency of biocatalytic reactions in the nanocompartments generated in the aggregates.

The mechanism of the mixed chain (made of E- and S-nano-particles) formation is illustrated in Fig. 3d–f. In this experiment, S- and E-nanoparticles were labelled with fluorescein and rhodamine-B fluorescent dyes, respectively. The dyes were encapsulated in the silica shell of the magnetic core. Both NPs form chain-like structures in the magnetic field visualized with fluorescent microscopy as green and red chains, respectively (Fig. 3d,e). In Fig. 3f, it is clearly seen that the NP blend forms yellow chains originating from the additive colour mixing of green- and red-labelled NPs in the same chain.

Within the magnetic chains, the force of dipole–dipole interactions is proportional to the strength of the magnetic field. The force generated by the magnetic field results in an attraction between the NPs and a compression of the block-copolymer brush. The compressed brush exerts a repulsive interaction between the NPs. An equilibrium compression of the brush will depend on the brush’s molecular characteristics, magnetic properties of the NPs and the strength of the magnetic field. The repulsive interaction is proportional to the molecular mass of the polymer brush and grafting density, and is reciprocal to the compressed brush thickness. The dipole–dipole interaction between the NPs is proportional to the strength of the magnetic field, size of the magnetic core and its magnetic susceptibility. Adjustments of these characteristics of the brush, magnetic NPs and magnetic field, loading and saturation of the brush with proteins in combination provide ample opportunities to optimize the biocatalytic system to achieve magnetic-field-controlled biocatalysis in real-life applications. A quantitative theoretical analysis of this adjustment has been published elsewhere25.

### Magnetic-field-controlled drug release

One important application of the developed architecture is magnetic-field-controlled drug release. It is commonly recognized that side effects of uncontrolled drug release cause severe intoxication and even death of patients. In general, targeted drug delivery should allow optimal dosages only in disease-affected areas, thus reducing toxic side effects. However, in a specific example of cancer treatment, recent analysis shows that only a small fraction of the administered drug dose is delivered to a solid tumour26 because of immune system response and organ competition mechanisms that shorten nanoparticle circulation in blood. At the same time, a prolonged circulation could cause premature drug release. The solution to this problem relies on a well-controlled initiation and duration of the dug release in the targeted location. A number of studies have demonstrated great potential of magnetic guidance of drug carriers with improved drug accumulation in solid tumours when the drug accumulation can be monitored using magnetic resonance imaging10. In those studies, magnetic guidance was combined with controlled release using heat generation in an oscillating magnetic field. The local temperature at the vicinity of the drug carrier is, however, difficult to control.

The biocatalytic system reported here is a versatile platform enabling the development of well-controlled targeted drug delivery systems. We propose two possible scenarios for drug administration. According to the first scenario, S- and E-nanoparticles are blended and injected. The particles circulate in blood and accumulate in the target tissue via passive or active (if appropriately functionalized27,28) targeting mechanisms. Application of magnetic field upon confirmed accumulation of NPs in the target will trigger the drug release. The second possible scenario is a two-step drug administration in which S-nanoparticles are magnetically guided to the target29 first. This stage could be conducted as long as needed to accumulate the therapeutic dosage of the drug carried by the S-nanoparticles. A premature release is prevented by the architecture of the S-nanoparticles, as shown in Fig. 2b inset. In the second step, E-nanoparticles are injected and guided to the same target location. The two-step approach avoids E- and S-nanoparticles interacting in blood if a magnetic field is applied prior to NP arrival in the target zone. Upon arrival, the E-nanoparticles merge with already accumulated S-nanoparticles in a magnetic field to release the drug-cargo. The particle accumulation could be confirmed using magnetic resonance imaging technology30,31. This concept is demonstrated here using a chemotherapy agent doxorubicin (DOX) loaded in S-nanoparticles via conjugation to BSA (DOX–BSA).

Fluorescence spectra, isothermal microcalorimetry and gel-electrophoresis experiments were used to monitor release of DOX from the S-nanoparticles in a buffer solution at pH 7.4. The fluorescent spectroscopy experiment was conducted in a glass cuvette with a magnet attached to the bottom (Fig. 4a). The released DOX diffused into the bulk solution in the cuvette, and the spectra were acquired at different time intervals (Fig. 4c). Initially, the S-nanoparticles with DOX–BSA were loaded in the cuvette. No DOX was detected in the solution after a waiting period, meaning no leakage of the drug in the absence of the magnetic field (Fig. 4d). Note, that the background fluorescent signal originated from the DOX molecules bound to the NPs. Then, the E-nanoparticles were added to the cuvette. No DOX spectra were detected in the mixed dispersion either. However, a burst release of DOX was observed as soon as the magnet was placed in the vicinity of the cuvette (Fig. 4d). Note, the non-zero base line is due to a weak fluorescence of quenched DOX in the conjugate. Upon release, the DOX fluorescent intensity increases and recovers32 (see Supporting Fig. 7). Quantitative analysis of the spectra revealed about 95 ± 5% release of the drug. This result was obtained amid the formation of only 50% of S–E contacts in the aggregates, thus indicating that NPs form dynamic aggregates in the magnetic field and virtually all DOX–BSA is degraded and DOX is liberated in the biocatalytic reaction.

Gel-electrophoresis experiments (Supplementary Figs. 14 and 15) with DOX–BSA loaded NPs qualitatively confirmed the conclusions derived from the spectroscopic studies: BSA and DOX–BSA are degraded in the presence of papain into small fragments (15 kDa and lower) within 1 h; S-nanoparticles release fragments of DOX–BSA in the presence of E-nanoparticles if the magnetic field is on whereas much fewer low-molecular-mass fragments are observed if the magnetic field is off; and DOX–BSA is well preserved in S-nanoparticles as concluded from the control experiment with S-nanoparticles mixed with papain in solution.

Alternatively, the magnetically triggered release of DOX was monitored using microcalorimetry. A microcalorimeter was equipped with two glass ampules, one loaded with a mixture of the E- and S-nanoparticles and another with buffer. The thermal activity of the biocatalytic reaction was recorded by subtracting a signal of the buffer ampule from the reference NP-loaded ampule. Both ampules were equipped with a special shaft for insertion of the magnet. In the control experiment, the reaction was monitored without application of the magnet and no thermal activity was recorded from the first vial. The biocatalytic reaction was initiated by moving both preloaded magnets from the dry upper compartment of the ampule to the bottom compartment. The thermal activity associated with the reaction was monitored for more than 12 h (Fig. 4e).

The feasibility of the developed drug delivery system in a biological environment was studied in vitro using 4T1 cells (murine breast cancer cell line). S- and E-nanoparticles were mixed in a 1:1 ratio and were added to the incubation medium (Fig. 5a). A calcein AM cell assay was performed after 24 hours of incubation (Fig. 5b) and the per cent cell viability was evaluated based on cell counting. Relative to the PBS control experiment, at least 70% of cells remained alive when the nanoparticle concentration was below 1.25 µg ml–1 (Fig. 5c,f). When a magnet was attached to the bottom of the cell culture dish, there was a significant drop of cancer cell viability at all particle concentrations (Fig. 5d,f). This was attributed to magnet-triggered DOX release and the following cell apoptosis. In a separate experiment, we obtained proof of the nanoparticle internalization. The 4T1 cells were incubated in the presence of fluorescein-silica-shell-labelled S-nanoparticles. Plane and 3D-confocal images clearly demonstrate the internalization of the NPs by the cells (Fig. 5e). The internalization was proved also using fluorescence properties of S-nanoparticles loaded with DOX–BSA conjugate (Supplementary Fig. 16).

The experiments with the cell culture demonstrated NP internalization and their cytotoxicity upon application of the magnetic field. The reference experiments with NPs with no polymeric shell and polymer brush decorated NPs loaded with FD-BSA and papain showed no obvious cytotoxicity at concentrations below 50 µg ml–1 (Supplementary Figs. 18 and 19). The low cytotoxicity of NPs was confirmed by multiple repetitions of the reference experiments, which were in good agreement with the literature33,34.

To exclude possible artefacts associated with cell viability we designed additional experiments that proved magnetic-field-triggered biocatalysis in the cell culture. The cells were incubated with a 1:1 mixture of the S- and rhodamine-B-silica-shell-labelled E-nanoparticles (0.05 mg ml–1 total concentration) loaded with FD-BSA and papain, respectively. FD is quenched in the FD-BSA conjugate, as shown in the reference experiment using fluorescence microscopy (Fig. 6b, bottom). However, application of the magnet led to merging NPs and initiation of the biocatalytic reaction of FD-BSA proteolysis in the nanocompartments. The latter was detected by the intense fluorescent signal of the liberated unquenched FD (Fig. 6b, top). A less intense red fluorescence of rhodamin-B embedded in the silica shell of NPs is observed for both samples with a magnet and for the control experiment (Fig. 6b). The overlaid bright field (Fig. 6a) and fluorescent microscopy (Fig. 6b) images of the same area of the cell culture confirmed that the FD was released in the areas occupied by the cells upon the application of the magnetic field (Fig. 6c). The areas occupied by NPs and released FD appear yellow owing to the combination of red NPs and green FD. A large surface area is indeed coloured green because of a much higher concentration of the released FD.

## Conclusions

We have demonstrated that our stimuli-responsive biocatalytic system of superparamagnetic NPs with a copolymer shell hosting the substrate and a complementary enzyme is a powerful platform for the remote control of biocatalytic processes. This biocatalytic platform possesses very important properties: (1) stimuli-triggered biocatalysis; (2) remote control of biocatalysis; and (3) high selectivity of biocatalysis when only authorized biocatalytic reactions are triggered. The developed system provides an example of well-controlled magnetic-field-triggered cancer drug release. Furthermore, the proposed platform is versatile and capable of delivering various biological materials or therapeutic chemical molecules.

## Methods

### Materials

Ferric chloride, ferrous chloride, copper(ii) bromide, concentrated nitric acid, hydrochloric acid, trisodium citrate, silicon tetraethoxide (TEOS), (3-aminopropyl)triethoxysilane (APS), triethylamine, α-bromoisobutyryl bromide (BIB), ethyl α-bromoisobutyrate (EBIB), N,N,N′,Nʺ,Nʺ-pentamethyldiethylenetriamine (PMDTA), ascorbic acid, tin(ii) 2-ethylhexanoate (THE), methanesulfonic acid, glutaraldehyde, 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide hydrochloride (EDC), N-hydroxysuccinimide (NHS), papain, fluorescein-labelled bovine albumin (FD-BSA), fluorescein isothiocyanate (FITC), rhodamine B isothiocyanate (RITC), suberic acid bis(N-hydroxysuccinimide ester) (SABNHS), and organic solvents ethanol, anisole, chloroform and dichloromethane were purchased from Sigma-Aldrich and used as received.

Monomers for grafting of a block copolymer of polyacrylic acid (PAA) and poly(ethylene glycol) methyl ether acrylate (PEGMA) or PAA-b-PPEGMA: tert-butyl acrylate (TBA) and PEGMA with a number average molecular mass of 480 g mol–1 were purchased from Sigma-Aldrich and purified using a flash-chromatography column containing inhibitor removers (Sigma #311340 and Sigma #311332). Doxorubicin hydrochloride (DOX) was purchased from Oakwood Chemical and used as received.

Phosphate-buffer saline (PBS), pH 7.4 at 25 °C was prepared using one Sigma pouch by dissolving it in 1 l of deionized (DI) water. 2-(N-morpholino) ethanesulfonic acid buffer (MES; 0.1 M; pH 4.7–5.0 at 25 °C) was prepared by dissolving 0.1 mol MES and 0.07 mol NaCl in 1 l DI water. A Bradford protein assay, BCA Protein Assay Kit, Zeba Spin Desalting Columns and LIVE/DEAD Viability/Cytotoxicity Kit for mammalian cells were purchased from Thermo Fisher Scientific.

Roswell Park Memorial Institute medium (RPMI 1640), 0.25% trypsin-EDTA, fetal bovine serum, and antibiotics: penicillin and streptomycin were purchased from Sigma-Aldrich. The mouse 4T1 breast tumour cells used for the present cultures were provided by J. Xie, University of Georgia, USA. A permanent magnet NdFeB, Grade N42 (K&J Magnetics, Pipersville, PA) was used in the experiments.

### Synthesis of magnetite–silica core–shell nanoparticles

The superparamagnetic NPs were synthesized using a co-precipitation method as described elsewhere35. Iron chloride salts, 4.43 g $FeCl 3 ∙6 H 2$O and 1.625 g $FeCl 2 ∙4 H 2$O, were dissolved in 190 ml of DI water with a stoichiometric ratio $Fe 3 +$:$Fe 2 +$ = 2:1 with magnetic stirring at room temperature. Then, 10 ml of 25% ammonium hydroxide was added immediately. Formation of a black precipitate was observed. The solution was stirred for an additional 10 min, then the precipitate was separated with a magnet and rinsed three times with DI water using magnetic separation. The colloidal dispersion of NPs was stabilized with citrate ions by a rapid rinsing of the precipitate twice with 2 M nitric acid aqueous solution, followed by the addition of 5 ml of 0.5 M trisodium citrate in water while maintaining pH 2.5 using a sodium hydroxide aqueous solution. After stirring for 1.5 h, NPs were magnetically separated, rinsed with DI water and diluted to obtain 100 ml (pH = 6) of the dispersion. The concentration of NP in the final stock dispersion was 2% wt.

A modified Stöber method36 was applied to coat NPs with a silica layer: 2 ml of the NP stock solution was diluted in a mixture of 160 ml ethanol and 40 ml DI water. Then, 5 ml of ammonium hydroxide was added to the NP dispersion. After 10 min of treatment in an ultrasonic bath, 1 ml of TEOS was added dropwise into the reactor. The synthesis was carried out at 0 °C and under sonication for 3 h. The reaction was terminated by the addition of several droplets of 10% HCl followed by precipitation of silica-coated NPs. The precipitate was collected by a magnet and rinsed three times with DI water. Then the precipitate was suspended in 50 ml DI water with sonication. The resulting product is a stable 2 mg ml–1 dispersion of NPs. The particle structure (TEM image) and dimensions are shown in Supplementary Figs. 1a and 3 (line 1). The NPs were characterized using a magnetometer AGM 2900 (Alternating Gradient Magnetometer by Princeton Inc.). As shown in Supplementary Fig. 1c, the NPs demonstrate superparamagnetic behaviour with no spontaneous magnetization. In the following experiments with the NPs, the external magnetic field was in a range from 0.1 to 0.2 T when magnetization of NPs is close to its saturation.

### Labelling of the nanoparticles with fluorescent dyes

Magnetic nanoparticles were labelled via inclusion of two different florescent dyes in the silica shell of NPs: 0.094 g of APS was added to 0.12 g of FITC or RITC (for green and red particles, respectively) dissolved in 10 ml of anhydrous ethyl alcohol; the reaction was carried out for 17 h by stirring the reaction mixture in dark conditions under a dry nitrogen gas. The synthesized conjugates, APS–FITC and APS–RITC, were used immediately after preparation: 4.8 ml of the stock solution of the magnetic nanoparticles was added to a mixture of 760 ml ethanol and 200 ml DI water; ammonium hydroxide (24 ml, as received) was added and the mixture was stirred and sonicated for 10 min to disperse the nanoparticles; afterwards, a mixture of 240 µl of TEOS and 300 µl of APS–RITC (or APS–FITC) were added dropwise to the particle dispersion and the mixture was stirred for 4 h in dark conditions. The silica-coated NPs were precipitated by adding hydrochloric acid, centrifuged and dispersed in ethanol. Rinsing in a fresh portion of ethanol was repeated three times to purify the nanoparticle dispersion. The dye-loaded silica shell was sealed with an additional silica coating to preserve fluorescent molecules. The nanoparticles were dispersed in a mixture of 30 ml ethanol and 7 ml DI water; 700 µl of ammonium hydroxide was added and then 3 µl of TEOS was added to the dispersion. The mixture was stirred overnight and particles were rinsed with three centrifugation–precipitation cycles in ethanol.

### Grafting of PAA-b-PPEGMA block copolymer from the nanoparticle surface

Grafting of the PAA-b-PPEGMA block copolymer from the surface of nanoparticles was conducted using the activator generated by electron transfer for atom transfer radical polymerization (AGET-ATRP) mechanism37. The polymerization was conducted in two steps. First, poly tert-butyl acrylate (PTBA) was grafted by polymerization of TBA. The AGET-ATRP of PTBA was followed by grafting of PPEGMA blocks using the same AGET-ATRP mechanism. Finally, the post polymerization treatment was applied to hydrolyse the PTBA blocks and convert them into PAA blocks. The synthetic steps are shown in Supplementary Fig. 2. The polymerization steps are described below in detail.

#### Immobilization of initiator

Silica-coated NPs were transferred to an ethanol dispersion; the stock NP solution was mixed with ethanol and the nanoparticles were extracted using magnetic separation. This was repeated several times to decrease the concentration of water in the NP ethanol dispersion. Finally, the NPs were added to a 1% APS ethanol solution and stirred for 12 h. APS immobilization was followed by three rinses with ethanol. NPs were incubated for 1 h in 100 ml dry dichloromethane with 2 ml of triethylamine and 1 ml of BIB. The initiator-functionalized particles were rinsed three times with chloroform and ethanol.

#### Polymerization

A TBA monomer was purified using a flash-chromatography column loaded with inhibitor removers. 210 µl of a 0.1 M $CuBr 2$ ethanol solution, 320 µl of a 0.5 M PMDTA ethanol solution and 37.5 µl of a 0.68 M EBIB ethanol solution were added to 45 ml of a 30% monomer solution in anisole. About 1 g (NPs by dry weight) of a concentrated slurry of the initiator-functionalized NPs was added to the solution. In this solution, EBIB (or Br-initiator in solution) was added to synthesize the block-copolymer in solution for the molecular mass analysis. The reaction mixture was deoxygenated by purging nitrogen for 20 min and then the solution was heated at 70 °C. 500 µl of 1 M ascorbic acid (or in some experiments 500 µl THE) was added to the solution and the reactor was sealed. The polymerization reaction was terminated in 76 h by opening the vial to air and rapidly cooling the reactor. The polymer from the solution was separated from the nanoparticles by centrifugation, re-precipitated three times with 30% aqueous ethanol and analysed with gel permeation chromatography (GPC). A choice of ascorbic acid versus THE was dictated by the adjustment of the polymerization rate. The polymerization reaction in the presence of THE was slower.

Grafting of the second PPEGMA block was carried out by a similar procedure: a 25% PEGMA solution in ethanol was polymerized for 60 min at room temperature. PTBA-b-PPEGMA was converted to PAA-b-PPEGMA using 1% methane sulfonic acid in ethanol. After hydrolysis, the nanoparticles were rinsed three times with chloroform, ethanol, and water and dried at 50 °C in an oven. The NP powder is easily redispersible in water and forms a stable colloidal dispersion with an average nanoparticle size of 45 nm (Supplementary Fig. 3, line 2) and zeta potential ξ = −30 mV (pH 7.4).

### Functionalization of nanoparticles with proteins and DOX

EDC-NHS conjugation was used for the conjugation of papain and FD-BSA to the polymer-coated magnetic nanoparticles. Carboxylic acid groups of the PAA polymeric layer reacted with EDC-NHS reagents to form an intermediate active ester that then reacts with primary amines of FD-BSA or papain to form covalent amide bonds. The nanoparticles were dispersed in a 10 ml MES buffer solution (pH 4.5) at a concentration of 1.6 mg ml–1. Then, 3 mg of EDC (2 mM) and 6 mg of NHS (5 mM) were added to the nanoparticle dispersion; the reaction was conducted for 20 min at 36 °C. The nanoparticles were rinsed twice with PBS buffer and then 5 ml of the dispersion was divided between two vials with a concentration of 3 mg ml–1 in each. Afterwards, 3 mg of the FD-BSA powder was added to the first vial and 125 µl of a 28 mg ml–1 papain solution was added to the second. Both enzymes were added in great excess to approach the saturation of the brush loading with the proteins. The conjugation reaction was carried out for 4 h at room temperature with delicate shaking. The nanoparticles were washed five times using centrifugation for their separation. The supernatant was periodically analysed to monitor the presence of proteins using a fluorometer. Rinsing was repeated until no trace of the labelled proteins were observed in the supernatant.

#### Conjugation of DOX

First, DOX was conjugated to BSA. 2 mg of BSA and 1 mg of DOX were added to 1 ml of 0.1% glutaraldehyde solution in PBS. DOX was dissolved first in 20 μl dimethylsulfoxide (DMSO) and then added to the aqueous solution of BSA. The reaction proceeded for 30 min at room temperature. The synthesized BSA-DOX conjugate was purified using a Zeba Spin chromatographic column. The purified BSA-DOX was analysed spectroscopically to confirm and quantitatively evaluate the conjugation, as described below. The results show that about 50% of the initially reacted amount of DOX was bound to BSA (see Supplementary Methods and Supplementary Fig. 9). Afterwards, BSA-DOX was conjugated with S-nanoparticles as described above (the same protocol as for conjugation of FD-BSA). The S-nanoparticles appeared as red-coloured material (Supplementary Fig. 6). Loading of the nanoparticles with proteins was estimated with fluorescence spectroscopy (Supplementary Figs. 7 and 8) and gravimetrically by spectra and weight change before and after loading (Supplementary Table 2). The discrepancy between the two methods was less than 5%.

### Aggregation of particles in magnetic field

Dispersions of NPs of different concentrations (0.001–1%) were prepared in PBS solutions and exposed to the magnet using several different experimental set-ups. The formed aggregates were visualized in dry conditions as shown in Supplementary Fig. 12, where exposure time increases from (a) to (e). Aggregate formation is rapid, so the exact time for each aggregate size was not determined exactly. Similar experiments were conducted using Si wafers when the aggregates were deposited and visualized using AFM (Fig. 3b). The formation of aggregates in solution was visualized using cryo-TEM (Fig. 3a) or for a greater concentration of NPs (1%) using dark-field optical microscopy (Fig. 3c). The results proved that increased time and NP concentration result in an increase of the aggregate size from 2–3-mers to larger 3D-strings of NPs.

### Smart compartmentalization

EDC-NHS conjugation is conducted at pH 4.5–5.0 when PAA acid is partially negatively charged (the pK a of PAA is a function of the degree of dissociation and ionic strength, at pH 5 and a relatively high local concentration of PAA in the brush, the pK a is in the range 4.5–5.0; ref. 38) while papain is positively charged (the isoelectric point of papain is between 8.5 and 9.5, Sigma Papain P4762 datasheet) and FD-BSA has no overall charge (the isoelectric point of FD-BSA is 4.8; ref. 39). The interaction of proteins with polyelectrolyte brushes is a complex phenomenon40. Here we partially simplify the discussion and focus on the major conclusions. Although with high ionic strength buffer solutions (about 160 mM) electrostatic interactions are largely screened, electrostatic interactions are beneficial for the introduction of enzymes in the PAA polymer brush in the first stage prior to covalent binding. The conjugation is conducted in conditions of a very high excess of proteins to saturate the PAA brush. PPEGMA block stabilizes particles sterically, however they create a relatively low barrier for the transport of small globular proteins such as papain and FD-BSA.

The performance of the brush-decorated NPs was studied in PBS buffer solutions or a cell culture medium at physiological pH 7.4. At these conditions PAA should be much more negatively charged. However, the presence of the conjugated proteins and high ionic strength of the environment results in screening of electrostatic interactions, so that the major repulsive properties of the brush are caused mainly by the osmotic pressure in the PAA blocks highly loaded with proteins. This steric repulsion mechanism is enhanced by the additional steric repulsion of PPEGMA blocks. A combination of these two effects results in the ‘insulation’ of the conjugated and hidden proteins inside the brush. We term this mechanism smart compartmentalization. The semi-permeable block-copolymer brush becomes virtually impermeable for proteins and other large molecules as soon as it becomes saturated in the conjugation step. This smart compartmentalization explains negative reference experiments demonstrated in Fig. 2b: protein in solution cannot interact with proteins loaded into the PAA brush. If magnetic field is on, the magnetic force overcomes the osmotic pressure and nanoparticles merge together to form a biocatalytic nanocompartment.

### Experiments with cell culture

#### Preparation of cell culture for the experiments

The 4T1 cell line (American Type Culture Collection) was cultured using RMPI 1640 medium supplemented with 10% fetal bovine serum and 1% penicillin–streptomycin solution. The cells were incubated  under humid conditions at 37 °C and 5% CO2. The cell line has not been authenticated or tested for mycoplasma contamination.

#### NP uptake by cells

4T1 cells were incubated in fresh medium for 72 h to allow cell attachment and proliferation. The cell culture at 100% coverage of the cell culture dish was treated with a trypsin solution. The detached cells were separated by centrifugation (1,000 g for 10 min) and seeded in fresh medium. The cell culture was used after 24 h incubation at about 40% coverage of the cell culture dish. Afterwards, the cells were rinsed with fresh medium (the unattached and dead cells were removed).

#### Internalization of NPs

FD-labelled-silica-shell S-nanoparticles (labelled with APS-FITC as described above) were added to the incubation medium (concentration of NPs in the medium was 0.1 mg ml–1). After 12 h incubation, the cell culture was rinsed in fresh medium to remove uninternalized particles. The cells were examined using a laser confocal microscope.

#### Magnetic field-triggered biocatalysis in cell culture

S- and rhodamine-B-silica-shell-labelled E-nanoparticles loaded with FD-BSA and papain, respectively, were mixed in a 1:1 ratio and were added to the incubation medium (the NP concentration in the medium was 0.05 mg ml–1). Two similar samples were incubated for 4.5 h: one sample was supplied with an attached magnet, while another sample was used for a control experiment with no magnet. The cells in both samples were rinsed in fresh medium to remove uninternalized particles. The cell samples were analysed with optical microscopy using bright field and fluorescent microscopy modes (Supplementary Figs. 16 and 17).

#### Release of therapeutic drug DOX

S- and E-nanoparticles loaded with BSA-conjugated-DOX and papain, respectively, were mixed in a 1:1 ratio and were added to the incubation medium at different concentrations, as shown in Fig. 5f. Uptake of the S-particles was evidenced by fluorescent microscopy imaging owing to the fluorescent properties of DOX (Supplementary Fig. 16). A calcein AM cell assay was performed after 24 h of incubation (Fig. 5b). The per cent cell viability was evaluated by cell counting. The assay was performed for a series of samples including samples (i) with no NPs added (control experiment), (ii) with added NPs and no magnet attached, and (iii) with added NPs and an attached magnet. The experiment was repeated three times. Representative images of two experiments are shown in Supplementary Fig. 17.

### Data availability

The data that support the plots within this paper and other findings of this study are available from the corresponding author upon reasonable request.

Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

## References

1. 1.

Torchilin, V. P. Multifunctional, stimuli-sensitive nanoparticulate systems for drug delivery. Nat. Rev. Drug Discov. 13, 813–827 (2014).

2. 2.

Kudina, O. et al. Highly efficient phase boundary biocatalysis with enzymogel nanoparticles. Angew. Chem. Int. Ed. 53, 483–487 (2014).

3. 3.

Popat, A. et al. Mesoporous silica nanoparticles for bioadsorption, enzyme immobilisation, and delivery carriers. Nanoscale 3, 2801–2818 (2011).

4. 4.

Katz, E. & Willner, I. Integrated nanoparticle-biomolecule hybrid systems: synthesis, properties, and applications. Angew. Chem. Int. Ed. 43, 6042–6108 (2004).

5. 5.

Tokarev, A., Yatvin, J., Trotsenko, O., Locklin, J. & Minko, S. Nanostructured soft matter with magnetic nanoparticles. Adv. Funct. Mater. 26, 3761–3782 (2016).

6. 6.

Hu, J. M., Zhang, G. Q. & Liu, S. Y. Enzyme-responsive polymeric assemblies, nanoparticles and hydrogels. Chem. Soc. Rev. 41, 5933–5949 (2012).

7. 7.

de la Rica, R., Aili, D. & Stevens, M. M. Enzyme-responsive nanoparticles for drug release and diagnostics. Adv. Drug Deliv. Rev. 64, 967–978 (2012).

8. 8.

Stuart, M. A. C. et al. Emerging applications of stimuli-responsive polymer materials. Nat. Mater. 9, 101–113 (2010).

9. 9.

Reddy, L. H., Arias, J. L., Nicolas, J. & Couvreur, P. Magnetic nanoparticles: design and characterization, toxicity and biocompatibility, pharmaceutical and biomedical applications. Chem. Rev. 112, 5818–5878 (2012).

10. 10.

Mura, S., Nicolas, J. & Couvreur, P. Stimuli-responsive nanocarriers for drug delivery. Nat. Mater. 12, 991–1003 (2013).

11. 11.

Katz, E., Lioubashevski, O. & Willner, I. Magnetic field effects on cytochrome c-mediated bioelectrocatalytic transformations. J. Am. Chem. Soc. 126, 11088–11092 (2004).

12. 12.

Katz, E., Weizmann, Y. & Willner, I. Magnetoswitchable reactions of DNA monolayers on electrodes: gating the processes by hydrophobic magnetic nanoparticles. J. Am. Chem. Soc. 127, 9191–9200 (2005).

13. 13.

Katz, E. & Willner, I. Switching of directions of bioelectrocatalytic currents and photocurrents at electrode surfaces by using hydrophobic magnetic nanoparticles. Angew. Chem. Int. Ed. 44, 4791–4794 (2005).

14. 14.

Tam, T. K., Ornatska, M., Pita, M., Minko, S. & Katz, E. Polymer brush-modified electrode with switchable and tunable redox activity for bioelectronic applications. J. Phys. Chem. C 112, 8438–8445 (2008).

15. 15.

Weizmann, Y., Elnathan, R., Lioubashevski, O. & Willner, I. Endonuclease-based logic gates and sensors using magnetic force-amplified readout of DNA scission on cantilevers. J. Am. Chem. Soc. 127, 12666–12672 (2005).

16. 16.

Hayashi, K. et al. Magnetically responsive smart nanoparticles for cancer treatment with a combination of magnetic hyperthermia and remote-control drug release. Theranostics 4, 834–844 (2014).

17. 17.

Uline, M. J., Rabin, Y. & Szleifer, I. Effects of the salt concentration on charge regulation in tethered polyacid monolayers. Langmuir 27, 4679–4689 (2011).

18. 18.

Halsey, T. C. & Toor, W. Fluctuation-induced couplings between defect lines or particle chains. J. Stat. Phys. 61, 1257–1281 (1990).

19. 19.

Furst, E. M. & Gast, A. P. Particle dynamics in magnetorheological suspensions using diffusing wave spectroscopy. Phys. Rev. E 58, 3372–3376 (1998).

20. 20.

Gulley, G. L. & Tao, R. Structures of an electrorheological fluid. Phys. Rev. E 56, 4328–4336 (1997).

21. 21.

Tan, Z. J., Zou, X. W., Zhang, W. B. & Jin, Z. Z. Structure transition in cluster–cluster aggregation under external fields. Phys. Rev. E 62, 734–737 (2000).

22. 22.

Silva, A. S., Bond, R., Plouraboue, F. & Wirtz, D. Fluctuation dynamics of a single magnetic chain. Phys. Rev. E 54, 5502–5510 (1996).

23. 23.

Cheng, R., Zhu, L., Huang, W. J., Mao, L. D. & Zhao, Y. P. Dynamic scaling of ferromagnetic micro-rod clusters under a weak magnetic field. Soft Matter 12, 8440–8447 (2016).

24. 24.

Townsend, J., Burtovyy, R., Galabura, Y. & Luzinov, I. Flexible chains of ferromagnetic nanoparticles. ACS Nano 8, 6970–6978 (2014).

25. 25.

Motornov, M. et al. Field-directed self-assembly with locking nanoparticles. Nano Lett. 12, 3814–3820 (2012).

26. 26.

Wilhelm, S. et al. Analysis of nanoparticle delivery to tumours. Nat. Rev. Mater. 1, 16014 (2016).

27. 27.

Sawant, R. M. et al. Nanosized cancer cell-targeted polymeric immunomicelles loaded with superparamagnetic iron oxide nanoparticles. J. Nanopart. Res. 11, 1777–1785 (2009).

28. 28.

Liao, C. D., Sun, Q. Q., Liang, B. L., Shen, J. & Shuai, X. T. Targeting EGFR-overexpressing tumor cells using cetuximab-immunomicelles loaded with doxorubicin and superparamagnetic iron oxide. Euro. J. Radiol. 80, 699–705 (2011).

29. 29.

Zhu, L. et al. Targeted delivery of methotrexate to skeletal muscular tissue by thermosensitive magnetoliposomes. Int. J. Pharm. 370, 136–143 (2009).

30. 30.

Yang, L. L. et al. Receptor-targeted nanoparticles for in vivo imaging of breast cancer. Clin. Cancer Res. 15, 4722–4732 (2009).

31. 31.

Liu, D. F. et al. Conjugation of paclitaxel to iron oxide nanoparticles for tumor imaging and therapy. Nanoscale 4, 2306–2310 (2012).

32. 32.

Bagalkot, V. et al. Quantum dot: aptamer conjugates for synchronous cancer imaging, therapy, and sensing of drug delivery based on bi-fluorescence resonance energy transfer. Nano Lett. 7, 3065–3070 (2007).

33. 33.

Karlsson, H. L., Gustafsson, J., Cronholm, P. & Moller, L. Size-dependent toxicity of metal oxide particles: a comparison between nano- and micrometer size. Toxicol. Lett. 188, 112–118 (2009).

34. 34.

Villanueva, A. et al. The influence of surface functionalization on the enhanced internalization of magnetic nanoparticles in cancer cells. Nanotechnology 20, doi:10.1088/0957-4484/20/11/115103 (2009).

35. 35.

Bumb, A. et al. Synthesis and characterization of ultra-small superparamagnetic iron oxide nanoparticles thinly coated with silica. Nanotechnology 19, 335601 (2008).

36. 36.

Deng, Y. H., Wang, C. C., Hu, J. H., Yang, W. L. & Fu, S. K. Investigation of formation of silica-coated magnetite nanoparticles via sol-gel approach. Colloids Surf. A 262, 87–93 (2005).

37. 37.

Jakubowski, W. & Matyjaszewski, K. Activator generated by electron transfer for atom transfer radical polymerization. Macromolecules 38, 4139–4146 (2005).

38. 38.

Miyajima, T., Mori, M., Ishiguro, S., Chung, K. H. & Moon, C. H. On the complexation of Cd(II) ions with polyacrylic acid. J. Colloid Interface Sci. 184, 279–288 (1996).

39. 39.

Schiller, A. A., Schayer, R. W. & Hess, E. L. Fluorescein-conjugated bovine albumin. Physical and biological properties. J. Gen. Physiol. 36, 489–505 (1953).

40. 40.

Leermakers, F. A. M., Ballauff, M. & Borisov, O. V. On the mechanism of uptake of globular proteins by polyelectrolyte brushes: a two-gradient self-consistent field analysis. Langmuir 23, 3937–3946 (2007).

## Acknowledgements

The authors would like to thank the National Science Foundation (grant number DMR 1426193) for funding. We would like to thank D. Asheghali, J. Xie and L. Xie, University of Georgia, USA for providing the mouse 4T1 breast tumour cells and assistance with cell culture experiments. We would also like to thank T. Enright, University of Georgia for assistance with NPs synthesis and functionalization and valuable discussions.

## Author information

### Affiliations

1. #### Nanostructured Materials Lab, University of Georgia, Athens, GA, USA

• Andrey Zakharchenko
•  & Sergiy Minko
2. #### Department of Chemistry and Biomolecular Science, Clarkson University, Potsdam, NY, USA

• Nataliia Guz
•  & Evgeny Katz

### Contributions

S.M. and E.K. conceived the central ideas and directed the project. A.Z. synthesized and characterized the NPs, studied their biocatalytic behaviour in a magnetic field including in the presence of cell culture; A.M.L. contributed to the characterization of the NPs and conjugation of proteins; N.G. performed experiments with the confocal microscope. All authors contributed to the analysis of the results and commented on the manuscript.

### Competing interests

The authors declare no competing financial interests.

### Corresponding author

Correspondence to Sergiy Minko.

## Electronic supplementary material

1. ### Supplementary Information

Supplementary Methods, Supplementary Figs. 1–19, Supplementary Tables 1–3, Supplementary References