Microcapsules as assay compartments formed through layer-by-layer deposition

We investigate the potential of using microcapsules, generated through layer-by-layer assembly, as reaction compartments for bioassays. A streptavidin-biotin and an esterase-fluorescein diacetate (FDA) assay were employed for proof-of-concept. Streptavidin was incorporated into the microcapsules via two methods: (i) physical adsorption onto MnCO 3 microcrystal templates or (ii) co-precipitation with CaCO 3 to form mixed microcrystals. The streptavidin-loaded template microparticles were then coated with five bi-layers of polyelectrolytes (polystyrene sulfonate/polyalylamine hydrochloride) (PAH/PSS) 5 and the templated cores were dissolved using EDTA to form polyelectrolyte microcapsules loaded with streptavidin. Streptavidin loading of around 77 mg streptavidin per mol of CaCO 3 was achieved using the co-precipitation method, compared to 5.1 mg streptavidin per mol of MnCO 3 using physical adsorption onto the manganese carbonate crystal surface. The microencapsulated streptavidin was allowed to bind to an analyte which was able to freely diffuse through the polyelectrolyte shell from the aqueous media. Biotin-4-fluorescein was added to the streptavidin-loaded microcapsules, with streptavidin-free capsules serving as the control sample. It was found that both types of microcapsules exhibited a similar fluorescence signal intensity, likely due to non-specific binding of biotin-4-fluorescein to charged groups on the polyelectrolyte shell. Therefore, an alternative esterase-FDA assay was investigated as fluorescence is only produced when FDA penetrates the capsule shell and reacts with the esterase enzyme inside which leads to its hydrolysis. Esterase was loaded into the LbL capsules templated on CaCO 3 via co-precipitation. FDA was added to both the esterase-loaded capsules and esterase-free capsules. It was found that the capsules containing esterase fluoresced, while the esterase-free capsules did not. This indicates that the hydrolysis reaction of FDA with esterase occurred inside the LbL microcapsules. These experiments demonstrate the potential of compartmentalised assays to be carried out inside microcapsules which could be applied in complex matrices or in multiplexing experiments, wherein different barcoded microcapsules contain different reagents to provide independent readouts without interference from larger biomolecules present in the surrounding media.


Introduction
Nano-and microcapsules have recently been introduced as drug carrier vesicles. Such capsules are fabricated by a layer-by-layer (LbL) technique involving the deposition of usually up to ten layers of oppositely charged polyelectrolytes (PE) around microparticle templates, such as inorganic microcrystals, colloidal particles (e.g. latex) or cells, which are eventually dissolved by an appropriate solvent or reagent leaving a 'hollow' polyelectrolyte capsule. 1 The size of these capsules may range from a few nanometres to hundreds of micrometres depending on the size and the nature of the template employed. [1][2][3] LbL capsules are extremely versatile; they may be fabricated using organic templates such as polystyrene 1,4,5 and melamine formaldehyde, 6,7 inorganic templates, e.g. SiO2 4,8 and carbonate particles (MnCO3, CdCO3, CaCO3), 9,10 or biological cells, e.g. red blood cells 11 or yeast cells 12,13 and even droplets. 14,15 These templates can be dissolved by hydrochloric acid (HCl), tetrahydrofuran (THF), hydrogen fluoride (HF), ethylenediaminetetraacetic acid (EDTA) and sodium hypochlorite (NaClO), respectively. The LbL capsule shell can be fabricated from diverse charged polymers such as poly(ethyleneimine) (PEI), poly(allylamine) (PAH), poly-(diallyldimethylammonium chloride) (PDADMAC) and poly-(styrenesulfonate) (PSS). 16,17 The properties of the LbL capsules can be precisely tailored by appropriate selection of the core template particles and the shell material. LbL capsules have been employed to microencapsulate a wide range of materials ranging from small molecules 18,19 to macromolecules [20][21][22] for applications in drug delivery, [23][24][25] biosensors 26,27 and bioreactors. 28,29 . Release of molecules from the capsules into the surrounding medium for drug delivery applications has been a particular research focus. More recently, researchers have started to explore the penetration of analytes through the multi-layered wall into the capsules, to use them as miniature reaction vesicles for analytes of interest to react with biological materials that are encapsulated inside the capsules 21,30,31 . For example, Caruso et al. 21 reported the encapsulation of catalase enzyme by coating catalase microcrystals with polyelectrolyte layers. The catalase enzymatic activity was preserved and measured by exposing the capsules to hydrogen peroxide in the presence of protease. The polyelectrolyte shell excluded the large protease molecules from the surrounding media, while the hydrogen peroxide diffused through the composite PE shell to interact with the catalase enzyme and produce water and oxygen. In addition, Mak et al. 30 developed a method to perform PCR inside polyelectrolyte microcapsules. In this method, the PCR reagents (polymerase, primers and template) were mixed with agarose gel and emulsifier in oil at 40 ℃ to form microdroplets. The microdroplet emulsion was cooled to form agarose gel microbeads. The microbeads containing the PCR reagents were then coated with multiple PE layers to form a polymeric capsule. During PCR, small molecular weight building blocks, nucleotides (dNTPs), were supplied externally and allowed to diffuse through the permeable capsule wall into the interior, while the resulting high molecular weight PCR products were accumulated within the microcapsule. Moreover, Price et al. reported a facile technique for RNA synthesis in LbL microcapsules. 31 Double-stranded DNA (dsDNA) PCR products were adsorbed onto silica colloids followed by LbL assembly of thiol-modified PMA (PMASH) and poly-N-vinylpyrrolidone (PVPON) via hydrogen bonding. After layering, the PMASH was crosslinked and the silica cores were dissolved. Raising the pH disrupted the hydrogen bonding between the PMASH and PVPON, releasing PVPON from the capsules. Diffusion of T7Pol into the capsules initiated transcription from the T7 promoter sequence on the dsDNA template. The synthesized RNA remained entrapped within the capsules. These examples show the potential of performing reactions inside LbL microcapsules which allow exclusion of larger reactive species from the media. The use of LbL capsules as vessels for bioassays has not yet been fully explored. Conventional immunoassays are carried out in individual reaction tubes or multi-well plates and involve sample prepurification steps. 32 Miniaturizing immunoassays in LbL microcapsules reduces the sample and reagent volumes. Using microcapsules as assay vessels also opens the door to running many individual or multiplex immunoassays in parallel. The capsule wall can act as a protective barrier to prevent larger molecules entering and interfering with the assay reaction. In the current study, we therefore aimed to investigate the potential of LbL microcapsules to act as reaction vessels for bioassays (figure 1). Streptavidin-biotin and esterase-diacetate fluorescein (FDA) assays were chosen as proof-of-principle bioassays to test this concept. Figure 1. Principle of conducting assays in PE microcapsules prepared by the LbL method using sacrificial templates. The required reagents were encapsulated during the capsule production. Small analyte molecules can penetrate through the capsule shell whilst large molecular weight inhibitors are excluded, allowing extraction of the analyte from the surrounding media for reaction with the encapsulated reagent. This approach is expected to increase the assay selectivity and sensitivity.

Preparation of microcapsules from MnCO3 microparticles
MnCO3 microparticles with a size range of 3-5 µm were obtained from PlasmaChem (Germany). The particles were suitable for capsule formation through LbL deposition of PSS and PAH followed by core dissolution as detailed in ESI1. To incorporate streptavidin into the microcapsules, streptavidin was deposited as a first layer in the LbL assembly. 50 mg of the cationic MnCO3 microcrystals were added to 1 mL of fluorescently labelled streptavidin (50 µg mL -1 ) prepared in phosphate buffered saline (PBS) solution (pH8), to ensure a negative charge on the streptavidin molecules (pI 5.8). The mixture was incubated for 1 h with continuous rotating at 40 rpm on a tube rotator (Keison Products, UK). The streptavidincoated particles were centrifuged at 5,000 rpm for 3 min, the supernatant was removed and the particles were washed three times with deionised water using centrifugation. Separately, Please do not adjust margins Please do not adjust margins 500 μg mL -1 of unlabelled streptavidin was also incubated with MnCO3 using the same protocol for use in the streptavidinbiotin assay. This was then followed by PE multilayer deposition. Briefly, particles were suspended in 0.5 mL of aqueous NaCl (0.2 M) before adding 1 mL of the cationic PAH polyelectrolyte (2 mg mL -1 ). The particles were washed three times with deionised water via centrifugation and resuspension, before adding 1 mL of anionic polyelectrolyte (PSS) solution (2 mg mL -1 ). Five bilayers of (PAH/PSS) were deposited using this procedure. The supernatant and washing solutions were kept for further analysis to determine the encapsulation efficiency from fluorescence measurements. The MnCO3 cores were dissolved with 10 mL EDTA solution (0.3 M) to yield hollow PAH/PSS microcapsules.

Preparation of CaCO3 microspheres
The synthesis of CaCO3 microparticles, the layer-by-layer deposition onto CaCO3 microparticles and core dissolution for capsule formation were performed and optimised as detailed in ESI2. To embed streptavidin in the microcapsules, the protein was co-precipitated with the calcium carbonate. First, 100 µL of fluorescently labelled streptavidin (1 mg mL -1 ) was added to 900 µL of CaCl2 solution (0.37 M). Then 1 mL of Na2CO3 solution (0.33 M) was poured rapidly into this mixture and stirred at 600 rpm for 30 s. With this approach, the streptavidin molecules became intercalated within the precipitated CaCO3 particles. Separately, to form particles for the streptavidin-biotin assay, 500 µL of unlabelled streptavidin (500 μg mL -1 ) was co-precipitated with 500 µL CaCl2 (0.66 M) and 1 mL of Na2CO3 (0.33 M), using the same protocol. Similarly, esterase was co-precipitated into CaCO3 particles by dissolving 10 mg of esterase in 1 mL of CaCl2 (0.33 M) before being rapidly mixed with 1 mL of Na2CO3 (0.33 M) and stirred for 30 s. Subsequently, particles were incubated in 1 mL of PAH (1 mg mL -1 ) under continuous rotation for 15 min. The excess PE was removed by three centrifugation and washing steps with purified water (6,000 rpm, 3 min). Next, the microparticles were dispersed into 1 mL of PSS (1 mg mL -1 ), followed by three rounds of washing. This LbL deposition was repeated until the desired number of PE layers was deposited on the microparticles. Hollow PE microcapsules were obtained by dissolving the CaCO3 core with EDTA (0.3M).

Assay and enzyme reaction protocols
For the streptavidin-biotin assay, two types of microcapsules with six layers of PAH/PSS were used: (i) microcapsules loaded with streptavidin (non-fluorescently tagged) as well as (ii) microcapsules free of streptavidin as a control. These were incubated with 500 µL (in case capsules templated on MnCO3) or 50 µL (in case capsules templated on CaCO3) of 0.25 mg mL -1 of biotin-4-fluorescein for 5 h. The microcapsules were washed with water until the fluorescent biotin was removed. For the esterase-FDA assay, 10 µL of FDA (10 mg mL -1 in acetone) was added to esterase-loaded microcapsules and esterase-free microcapsules templated on CaCO3 and then incubated for 10 min.

Streptavidin encapsulation efficiency
To determine the streptavidin encapsulation efficiency in the MnCO3 and CaCO3 particles, fluorescent streptavidinrhodamine red™-X conjugate was employed. The fluorescently labelled streptavidin was encapsulated as described above. The supernatant containing non-encapsulated streptavidin and all PE and washing solutions were collected. The amount of fluorescent streptavidin in the supernatants were quantified via a fluorescence spectrometer (Perkin Elmer, LS55). The encapsulation efficiency (EE %) was calculated as per equation 1.

EE % =
(total protein − free protein in supernatant) total protein X 100

Capsule characterization
Confocal Laser Scanning Microscopy (CLSM) images were taken using a Zeiss LSM710 inverted confocal microscope (Zeiss, UK) equipped with 20x, 40x and 63x oil immersion objectives. Samples were prepared by depositing a drop of capsule suspension onto a clean glass slide. The excitation wavelength was 570 nm for the rhodamine-red™-X conjugate, and 490 nm for fluorescein. Microcapsule suspensions were also observed under a Nikon TE-2000 inverted fluorescence microscope (Nikon, UK), equipped with five objectives (2x, 4x, 10x, 20x, 40x). A high sensitivity black and white digital CCD camera (Retiga-EXL, QImaging, Media Cybernetics, UK) and Image-Pro Plus software (Media Cybernetics, UK) were used to capture images. Scanning electron microscopy (SEM) was carried out on a Carl Zeiss, EVO 60. The particles and microcapsules were washed with deionised water to remove salts before pipetting onto a glass slide to dry. A thin carbon film was deposited on the particle suspension using a vacuum coating machine (Edwards High Vacuum, E12 series) prior to imaging. The zeta potential of the coated particle templates was monitored after each PE absorption step during the LbL assembly using a Zetasizer Nano ZS series (Malvern Instruments, UK). Particles were dispersed in Milli-Q water before the measurement. The reported zeta potential is an average of three consecutive measurements.

Results and discussion Encapsulation Efficiency of Proteins
The encapsulation efficiency of rhodamine-streptavidin was determined using two approaches. Firstly, the amount of encapsulated streptavidin was measured directly via fluorescence spectrometry of the particle suspension. Secondly, fluorescence within the PE, washing and EDTA solutions was measured as an indirect method, since some streptavidin could be released during the deposition of PEs and the washing steps, as well as during core dissolution with EDTA.
The encapsulation efficiency of streptavidin on MnCO3 microparticles was investigated by incubating them with fluorescently tagged streptavidin for 1 h. Subsequently, five bilayers of (PAH/PSS) PE were assembled on the particles. Figures 2a and 2b show bright field and fluorescence CLSM images, respectively, for the PE-coated microparticles with encapsulated fluorescent streptavidin. These illustrate the absorption of streptavidin predominately on the external surface of the MnCO3 crystals. The streptavidin did not penetrate inside the core as seen in the CLSM image (fig 2b) and the fluorescence intensity profile through the particle (fig 2c). We also produced CaCO3 particles loaded with fluorescent streptavidin through co-precipitation from Na2CO3 and CaCl2 as described in the experimental section. The streptavidin-loaded CaCO3 particles were washed with Milli-Q water to remove unbound streptavidin before visualisation. The particles were of spherical shape with a fairly narrow size distribution (4 ± 1 μm) (see SEM images, figure S-2.1 in the ESI). Figures 2d and 2e show bright-field and fluorescence microscopy images, respectively, indicating that the streptavidin was embedded successfully. The distribution of the fluorescent streptavidin was observed with CLSM ( fig 2e) showing streptavidin throughout the entire volume of the CaCO3 particles. This is further corroborated by the intensity plot (figure 2f) exhibiting a constant fluorescence intensity profile across the particle diameter. The streptavidin-loaded microparticles (MnCO3 and CaCO3) were coated with five bilayers of (PAH/PSS), followed by dissolution of the core with EDTA ( figure 3). There was no noticeable difference in the shape or dimension of microcapsules following the MnCO3 core dissolution. Interestingly, for the MnCO3 templated capsules, the streptavidin was localised in the PE shell rather than being distributed evenly throughout the capsule cavity (fig 3b). The fluorescence intensity profile across the microcapsules showed a high fluorescence intensity at the edge of the microcapsule shell, which decreased as it approached the capsule's centre (figure 3c). A similar effect was seen for the CaCO3-based capsules (fig 3 d-f).

LbL microcapsules for streptavidin-biotin assay
The streptavidin-biotin assay experiments were conducted with capsules templated on MnCO3 and CaCO3 particles. Two types of microcapsules were fabricated: (i) microcapsules not containing any streptavidin as control, and, (ii) microcapsules with (non-fluorescent) streptavidin. Both kinds of microcapsules were incubated with biotin-4-fluorescein according to the procedure outlined in the experimental section. It was found that both microcapsules containing streptavidin as well as microcapsules entirely free of streptavidin exhibited a fluorescence signal, indicating binding of fluorescently-tagged biotin to the PE capsules regardless of whether streptavidin was present or absent. The fluorescence was localised on the capsule shells, compared to the interior of the capsules, as illustrated by the fluorescence intensity profile plotted across the centre of the PE capsules (see ESI4). This would indicate that the biotin-4-fluorescein adsorbed onto the capsule PE shell rather than freely penetrating into the capsule interior to bind to streptavidin specifically. The negatively charged fluoresceintagged biotin, upon binding to the cationic polyelectrolyte component of the PE shell, would prevent the biotin molecules from entering and interacting with the streptavidin inside the capsule. To overcome this issue, we designed a different experiment with a non-ionisable molecule (fluorescein diacetate) which is hydrolysed by an esterase enzyme inside the microcapsules upon hydrolysis.

LbL microcapsules for esterase-fluorescein diacetate (FDA) reaction
The esterase enzyme was encapsulated in CaCO3 particles via co-precipitation as outlined in the experimental section. The esterase-loaded CaCO3 particles were then coated with alternating anionic and cationic PE layers and, after the desired number of PEs layers were deposited, the cores were dissolved with EDTA to form hollow PE microcapsules with entrapped esterase. As a control, PE microcapsules free of esterase were also fabricated using the same protocol. Both types of PE microcapsules, esterase-free and esterase-loaded, were incubated with a FDA solution. As can be seen in figure 4a-b, the esterase-free microcapsule exhibited no fluorescence after adding the FDA solution, whereas the esterase-containing capsules showed a fluorescence signal ( figure 4c-d). The nonfluorescent FDA molecules, after diffusion into the microcapsule, were hydrolysed by the esterase enzyme to form a fluorescent product, fluorescein. This product was not formed in the capsules free of esterase enzyme. The enzyme-filled microcapsules were imaged with a confocal fluorescence microscope (figure 4e-f). The intensity plot across a capsule (figure 4g) indicates fluorescence on/inside the shell as well as to some degree inside the capsule. The hydrolysis of FDA to fluorescein was successfully demonstrated in the microcapsules. However, it was difficult to determine enzyme localisation in the capsules. The enzyme may have been located in the capsule's centre where the FDA converted to a negatively charged fluorescein. In such a case, the fluorescein would have bound to the cationic PEs on the capsule's internal shell. Another, more realistic possibility is that the esterase enzyme is attached to the inner PE shell, just as was found with the streptavidin, and the incoming FDA is hydrolysed there, where the produced fluorescein also gets deposited. The observed result was in agreement with studies by other authors using PE microcapsules to immobilise enzymes inside a microcapsule cavity or within PE shells. 21,28,33,34 We have here shown the ability of the system to perform an assay and an enzyme reaction which can be adapted for drug concentration determination by immobilising antibody (protein) in the PE capsules, whilst allowing small analyte molecules (drug) in the sample to penetrate the PE shell to interact with the immobilized protein. Such microcapsule assays could be used for conducting a large number of individual assays in parallel on a small volume of sample for multiplexing.

Conclusions
The encapsulation of streptavidin and esterase into PE microcapsules fabricated from multiple bilayers of PAH/PSS onto inorganic crystal templates was demonstrated. A biotinstreptavidin assay and an FDA-esterase specific reaction were investigated with these capsules. The biotin-streptavidin assay revealed that biotin was found to bind non-specifically to the PE shell surface and did not penetrate freely into the capsule. For the FDA-esterase reaction, the non-fluorescent starting molecule FDA was used as an analyte model and added to both esterase-filled and esterase-free microcapsules. When the FDA diffused through the microcapsule shell, the esterase enzyme hydrolysed the incoming FDA to fluorescein. This fluorescent product was only observed in the esterase-loaded capsules. This proof-of-concept experiment shows the potential for performing assays for example for drug testing by immobilising specific antibodies inside the PE microcapsule while small sample analytes are allowed to penetrate the capsule shell. Our findings tie in with previous literature reports on enzymatic reactions carried out inside such capsules namely catalase for reaction with penetrating hydrogen peroxide 21 and PCR with penetrating nucleotide building blocks. 30 Microcapsules could react as containers for challenging matrices that may contain inhibiting substances such as environmental soil and water samples and biological samples including stool and urine samples. The method only requires a small volume of sample, the binding reaction is protecting from inhibiting larger size molecules that may be present in the sample and cannot penetrate the capsule shell. Using different coloured or fluorescently-tagged labels, the capsules have the potential to allow parallel analysis of several analytes for multiplex assays.

Conflicts of interest
There are no conflicts to declare.

Characterisation of polyelectrolyte assembly on manganese carbonate particles (MnCO 3 )
Layer-by-layer assembly was performed on MnCO3, crystals ranging in diameter from 3 to 5 μm (PlasmaChem, Germany). 50 mg of the dry particles were dispersed in 1 mL of purified water for washing. The particles were vortexed and then collected by centrifugation at 5000 rpm for 3 min. The supernatant was pipetted off before being replaced with 1 mL of fresh water. 500 μL of this particle suspension was added to 500 μL of the anionic polyelectrolyte PSS (2 mg mL -1 ) in a 1.5 mL Eppendorf tube and incubated for 20 min under continuous rotation at 40 rpm on a tube rotator. The coated microparticles were centrifuged at 600 rpm for 2 min; the supernatant was removed and the particles were washed three times with deionised water to ensure the removal of free PEs. The negatively charged coated particles were then re-suspended in 500 μL of NaCl (0.5 M) before adding the positively charged second PE, PAH-FITC (2 mg mL -1 ) (figure S1.1). The same protocol was repeated to deposit ten layers of alternatingly charged PEs (PSS/PAH-FITC) on the magnetic particles. EDTA solution (0.3 M) was used to dissolve the particle cores.
As shown in figure S1.2(a), the zeta potential alternated as oppositely charged PE layers were deposited on the particle cores. The fluorescence intensity was found to increase with the number of fluorescently-tagged PE layers (FITC-PAH) deposited ( figure S1.2(b)). SEM images of the MnCO3 cores before the LbL coating and of the collapsed hollow-shell PE capsules after the MnCO3 core removal are shown in figures S1.2(c-d). Microparticles coated with five bi-layers of FITC-PAH/PSS were also captured with fluorescence microscopy (figure S1.2(e)) and confocal microscopy (figure S1.2(f)) clearly showing the fluorescent polyelectrolyte embedded in the layers.

PSS/PAH assembly on calcium carbonate particles (CaCO 3 )
The process of LbL assembly of PE layers on the calcium carbonate particles is laid out in figure S2.2. 300 µL of CaCO3 particle stock suspension were incubated in 1 mL of PAH polyelectrolyte (1 mg mL -1 ) under continuous rotation for 15 min. The excess of PE was removed by three centrifugation and washing cycles with purified water (6,000 rpm, 3 min). Next, the microparticles were dispersed into 1 mL of PSS polyelectrolyte (1 mg mL -1 ), followed by three rounds of washing. This LbL deposition was repeated until the desired number of PE layers were deposited on the microparticles. Zeta potential measurements and microscopy images confirm the layer deposition as shown in figure S2.3.
Hollow-shell PE microcapsules were obtained by dissolving the CaCO3 core with EDTA. Different      Schematic of the method to encapsulate streptavidin inside CaCO3 particles via the co-precipitation, followed by LbL assembly of PAH/PSS polyelectrolytes. During core formation, streptavidin is co-precipitated inside the CaCO3 microparticles. PAH, a polycation PE, is deposited as the first layer. Excess PE is removed via washing before the second PE layer, PSS (a polyanion PE), is deposited. The core is then dissolved by EDTA to leave hollow capsules loaded with streptavidin.