Skip to main content
NCBI home page
ACS AuthorChoice logoLink to ACS AuthorChoice
. 2020 Jun 26;12(29):32951–32960. doi: 10.1021/acsami.0c09226

Continuous Flow Reactors from Microfluidic Compartmentalization of Enzymes within Inorganic Microparticles

Tuuli A Hakala †, Friedrich Bialas ‡, Zenon Toprakcioglu †, Birgit Bräuer ‡, Kevin N Baumann †, Aviad Levin †, Gonçalo J L Bernardes †,§,*, Christian F W Becker ‡,*, Tuomas P J Knowles †,∥,*
PMCID: PMC7383928  PMID: 32589387

Abstract

graphic file with name am0c09226_0006.jpg

Compartmentalization and selective transport of molecular species are key aspects of chemical transformations inside the cell. In an artificial setting, the immobilization of a wide range of enzymes onto surfaces is commonly used for controlling their functionality but such approaches can restrict their efficacy and expose them to degrading environmental conditions, thus reducing their activity. Here, we employ an approach based on droplet microfluidics to generate enzyme-containing microparticles that feature an inorganic silica shell that forms a semipermeable barrier. We show that this porous shell permits selective diffusion of the substrate and product while protecting the enzymes from degradation by proteinases and maintaining their functionality over multiple reaction cycles. We illustrate the power of this approach by synthesizing microparticles that can be employed to detect glucose levels through simultaneous encapsulation of two distinct enzymes that form a controlled reaction cascade. These results demonstrate a robust, accessible, and modular approach for the formation of microparticles containing active but protected enzymes for molecular sensing applications and potential novel diagnostic platforms.

Keywords: silica microparticle, enzymes, encapsulation, microfluidic flow reactor, glucose sensing

Introduction

In nature, barriers and gradients modulate the behavior of a wide range of molecular species and allow them to function under optimal environmental conditions. Barriers such as membranes or shells allow the formation of spatially nonhomogeneous solution conditions, where the basic machinery of life can be shielded from external conditions. Thus, the movement of molecules through such systems requires the crossing of dynamic barriers, for example, lipid bilayers or chemical gradients of specific ions, pH values, metabolites, and physical gradients in temperature or pressure. In particular, the spatial organization of enzymes within micro- or nanoenvironments can considerably increase their effectiveness.1,2 This observation has given inspiration for the design of in vitro systems, in which enzymes are immobilized to improve their stability and recyclability and protect them from the ambient environment.3−5 Enzymes can be immobilized, for example, using surface adsorption,6 covalent bonding,7 or encapsulation.8 These techniques generate favorable microenvironments for the enzyme9 while preventing enzyme aggregation,10 dissociation,11 and rigidification of the enzyme structure via multipoint covalent attachment.12 However, all of the techniques have their own challenges. These include leaking of the enzyme with adsorption immobilization, reduced activity with covalent bonding, and creation of diffusion barriers using encapsulation.13

In commonly used enzyme-based microfluidic assays, microchannels are coated with immobilized enzymes to serve as miniaturized heterogeneous biocatalysis reactors, where the high selectivity of enzymes under continuous flow can be combined with low sample volumes. This contributes to the rapid catalytic reaction processing and allowing the enzymes to be recycled while maintaining their stability and functionality.14,15

The formation of such microreactors by immobilizing enzymes on the surfaces of microchannels16−18 can often be limited by the surface area of the reactor and by steric interference due to limited availability of the enzyme active sites. The efficiency of such in vitro microreactors can be augmented by increasing the available surface area by adding porosity to the channel19 or by packing the reactor within functionalized microbeads.20−22 However, the correct packing of microchannels with solid nonporous beads requires the use of high pressures, limiting the availability and modularity of such platforms. To resolve this issue, soft-lithography-based microfluidic approaches have been developed, wherein pillars that maintain the structural stability of the channels have been used. Loading of such beads within the confined volume of the device is achieved under low-pressure conditions,23−25 yet achieving a high level of reproducibility for such fabrication techniques remains a challenge. Mesoporous silica particles produced by biomineralization with tunable pore sizes have promised to address many of these challenges.26−28 Furthermore, in many cases, the enzymes need to be covalently linked to the surface of microbeads or channels and are exposed to the ambient environment, resulting in limited protection against degradation. Therefore, such platforms commonly cannot be used to gain stable operation in long-term experiments.

Here, we address the issues limiting the activity of immobilized-enzyme-based assays by capitalizing on a droplet microfluidic approach. Microfluidic devices allow the production of monodisperse droplets with diameters in the micrometer range.29 Using the modular nature of such platforms, microdroplets can readily be loaded with different active compounds without the need for an additional step or surface functionalization. Polymers such as alginate, agarose, or silk protein can be added to create microparticles with different melting temperatures and mechanical properties.30−33 In this study, we present a strategy relying on semipermeable silica microparticles serving as a packing material for a continuous flow microfluidic reactor. This robust and facile approach allows for the production of silica-based microparticles loaded with enzymes that maintain their activity during several cycles of activation and are simultaneously protected from degradation by exogenous factors such as proteinases. This approach mimics a remarkable naturally occurring process producing the silica shells of diatoms. These unicellular algae are encased in an intricately structured frustule. The biomineralization process in diatoms is precisely controlled through compartmentalization in silica deposition vesicles and through phase separation processes.34 A mixture of organic molecules, including long-chain polyamines and silaffin and pleuralin peptides, is found in diatom silica that play a major role in controlling the deposition process.35 Furthermore, unmodified synthetic R5 peptides comprising the repetitive unit of the silaffin polypeptide from Cylindrotheca fusiformis (H-CSSKKSGSYSGSKGSKRRIL-OH) have been shown to trigger precipitation of silica in a biomimetic process in vitro, which we envisioned to use in our microfluidic setup to generate biomimetic inorganic particles loaded with enzymes.35,36 Previous studies with silica particles and enzyme encapsulation have demonstrated minimal enzyme activity loss confirming the suitability of the approach for enzyme studies37,38 and flow reactors.37 We further demonstrate that by modulating the enzyme composition encapsulated within the microparticles, multistep reactions can be performed and monitored by the formation of the reaction products. The formed microparticles can be organized either in single droplets using microfluidic trap arrays, or in larger compartments, generating a microfluidic continuous flow reactor, which has the potential to be further utilized for diagnostic applications.

Results and Discussion

Preparation and Characterization of Silica Microparticles

Silica microparticles were generated using a double T-junction device,39 where silicic acid and phosphate buffer with or without R5 peptides were coflown at the first junction and subsequently compartmentalized into microdroplets through an additional flow of an immiscible continuous oil phase in the second junction (Figure 1a). This strategy resulted in the generation of homogeneous microdroplets with a diameter of 100 μm (Figure 1e–f). Stable silica microparticles were formed both in the presence and absence of the R5 peptide. This observation allowed us to use a simple device geometry to generate silica microparticles. The double T-junction design rapidly mixes the silicic acid and phosphate buffer, so silica is only in contact with the channel surface for a few milliseconds, which is not enough time for it to precipitate and block the device before immersion in the oil phase to form the microparticles. The precipitation of silica was very rapid, and the formation of the gel-like matrix was observed directly after droplet formation. However, all samples were incubated at least for 1 h to ensure there was complete precipitation. To identify optimal conditions for the formation of such microparticles, the relative flow rates of silicic acid and phosphate buffer were systematically changed. In addition, the effect of adding R5 peptide to the phosphate buffer, which promotes condensation of silicic acid via an amine-mediated mechanism and leads to spherical silica particles when used in simple batch experiments (Figure 1b), was studied; however, structurally different but still stable silica microparticles were formed without the R5 peptide. The formation of silica microparticles within the droplets occurred almost instantaneously.

Figure 1.

Figure 1

Open in a new tab

Microfluidic droplet formation and morphology of the resulting silica microparticles. Schematic illustration of the formation of silica microparticles using a droplet microfluidic device (a) and potential mechanism for the condensation of silicic acid mediated by the R5 peptide40 (b). Scanning electron microscopy (SEM) micrographs displaying the morphology of the formed microparticles: with (c) and without R5 (d) using different flow rate ratios (buffer inlet/silicic acid inlet) and initial silicic acid concentrations. From (c) to (e), images i, ii, and iii correspond to 270 mM silicic acid with a 9:1 ratio, 540 mM silicic acid with a 3:1 ratio, and 1080 mM silicic acid with a 1:1 flow ratio, respectively. Cross-sectional views of microparticles generated with 1080 mM silicic acid at 1:1 flow ratio with (c-iv) and without (d-iv) R5 peptide. Confocal microscopy images of silica microparticles samples generated with Cy5-labeled R5 peptide (e) and image of droplets in a bright field (f). Scale bars 100 μm, except (c-iv) and (d-iv).

To further investigate the morphology of the generated microparticles with (Figure 1c) and without R5 (Figure 1d), scanning electron microscopy (SEM) was used. The diversity in morphology between microparticles generated under different conditions is clearly evident from the micrographs. All samples either shrank considerably (Figure 1c-iii,d-ii,d-iii) or collapsed (Figure 1c-i,d-i) due to drying under the analysis conditions. Based on the SEM images obtained for R5-containing microparticles (Figure 1c-i,c-ii), in the presence of R5 in combination with low and medium silicic acid concentrations, shell-like structures that collapse under the imaging condition can be observed. However, for the medium silicic acid concentration, the situation changes and additional silica material is formed within the spheres (evidenced by the only partial collapse of the spheres found in Figure 1c-ii). This additional silica within the spheres also contains R5 that is visualized in the confocal fluorescence image in Figure 1e-ii. At the highest silicic acid concentration in the presence (Figure 1c-iii) or absence of R5 (Figure1c-ii,c-iii), microparticles only shrunk instead of collapsing, indicating a more stable shell structure or, more likely based on the micrographs seen in Figure 1c-iv,d-iv, a particle structure consisting of more evenly distributed silica (Figure S2). The latter is supported by the compact silica matrix that can be seen in particles’ cross section. Due to the affinity of the R5 peptide for silica, it colocalizes with the precipitated silica matrix throughout the microparticle, which means that fluorescently tagged R5 (Cy5-R5) can be used in further investigation of silica localization in microparticles using confocal microscopy. Images of microparticles formed in the presence of Cy5-labeled R5 show a uniform peptide distribution throughout the microparticle for medium (Figure 1e-ii) and high (Figure 1e-iii) silicic acid conditions. This resembles what has been previously demonstrated in precipitated silica spheres.40,41 However, the formation of shell-like structures with the lowest silicic acid concentration is unique to the microfluidic generation of silica particles. Furthermore, the homogeneous precipitation of the silica matrix in the case of the highest silica concentration was confirmed using FRAP (Figure S3), showing minimal recovery of the photobleached area, thus proving that the silica is precipitated also in the center of the microparticle. However, in the case of low silicic acid concentrations (Figure 1e-i), R5 fluorescence is localized on the surface of the microparticle. This supports the SEM data and indicates that by changing the silicic acid concentration morphologies from capsulelike to uniform microparticle can be achieved.

To characterize the diffusion of small and very large molecules through the formed microparticles, the release kinetics of fluorescein (332 g/mol) (Figure 2a–d) and a fluorescein isothiocyanate–bovine serum albumin (FITC–BSA) (67 kDA) (Figure 2e–h) conjugate were monitored. In the former case, even though there were some differences in the encapsulation efficiency (EE, Figure S4), all fluorescent cargo is released within 2 h and fluorescein showed only a small difference in the release kinetics from microparticles formed with or without R5. Similarly, the increase of the silicic acid concentration during microparticle formation only minimally contributes to the small molecules’ release. A more pronounced effect on the release profile is observed for the much larger FITC–BSA conjugate (Figure 2f–h), for which the EE was extremely high, between 95.7 and 99.7% for medium and high silicic acid concentrations (Figure S4). For microparticles generated at low to medium silicic acid concentrations, the presence of the R5 peptide appears to restrict protein release. However, at high silicic acid concentrations, this difference becomes negligible, while the release kinetics decreases considerably in both cases, and only ∼10% of the protein is released over a 50 h period (Figure 2h). This most likely corresponds to the population of FITC–BSA located on the surface of the microparticle. Based on this observation, the pore size of the microparticle using a 1080 mM silicic acid concentration must lie between the sizes of the fluorescein and BSA, thus around 1–6.8 nm.42 Overall, this shows that the semipermeable nature of the silica microparticles can be tuned by changing the silicic acid concentration. However, due to the moderated effect of the R5 on the release kinetics at low silicic acid concentrations, and the negligible effects at high silicic acid concentrations, R5 was not used in further experiments, which further simplified the generation of silica microparticles.

Figure 2.

Figure 2

Open in a new tab

Release kinetics from silica microparticles. Release kinetics of fluorescein (a–d) and BSA-FITC conjugate (e–h) from microparticles under different conditions: silicic acid concentrations of 270 mM (b, f), 540 mM (c, g), and 1080 mM (d, h), with (solid circles) and without R5 peptide (open circles) in the phosphate buffer. Error bars indicate the standard deviation between three separate experiments.

Encapsulation of β-Galactosidase and Protection against Proteinase K Degradation

The properties of the silica microparticles generated at high silicic acid concentrations, where small molecules can easily diffuse through the matrix, but larger molecules remain trapped, are ideal for stable compartmentalization of enzymatic reactions with small substrates. While enzymes can be trapped within the silica matrix, small-molecule substrates and products can easily diffuse in and out. As a proof of concept, β-galactosidase (β-gal) was encapsulated within the silica microparticles and its activity was measured by fluorescence of the hydrolysis product of resorufin β-d-galactopyranoside that produces fluorescent resorufin.43,44 This was achieved by simply including 1 μM β-gal in the phosphate buffer solution prior to microfluidic droplet formation. The highest silicic acid condition (1080 mM) was chosen, as it exhibited a more controlled diffusion profile. Fluorescence time-lapse microscopy shows the increasing fluorescence signal of the enzyme-loaded microparticles and diffusion of the fluorescent product to the environment over the duration of the experiment (Figure 3a and Video provided in the SI).

Figure 3.

Figure 3

Open in a new tab

Enzyme encapsulation and protection. Fluorescence microscopy time lapse of β-gal encapsulated in silica microparticles and resorufin substrate added to the surrounding solution (a). Michaelis–Menten kinetics of both free and encapsulated β-gal (b). Protection of β-gal (125 nM) from proteinase K (33 μg/mL) degradation by encapsulation within silica microparticles: schematic representation of the experimental reaction determined in the case of free β-gal (c), and results of enzyme degradation by proteinase K results in complete loss of enzymatic activity (d). Encapsulated β-gal within silica microparticles (e), retaining ∼55% of its activity (f). Error bars indicate the standard deviation between three separate experiments.

These findings demonstrate the confinement of the β-gal enzyme within silica microparticles and retention of its activity following encapsulation. The encapsulated enzyme is stabilized within a microenvironment and leads to a slower reaction than that found in the bulk assay containing a free enzyme (Figure 3b). This is due to the restriction of the substrate and product diffusion to and from the microparticle interior, where the enzyme is stabilized. However, encapsulation within the microparticle matrix not only limits the diffusion of the substrate but also protects β-gal from degrading proteases and other large molecular species that can affect its activity. To test the degree by which the silica shell stabilizes encapsulated enzymes, the β-gal assay was repeated, while the free (Figure 3c,d) and encapsulated (Figure 3e,f) enzymes were exposed to proteinase K (33 μg/ml) for 16 h. This unusually long exposure time was used to examine the long-term stability of our encapsulated enzymes. While proteinase K can completely inactivate free β-gal (Figure 3d), a large portion of the encapsulated enzyme remains active (55%, see Figure 3f). A part of the decreased activity of the encapsulated enzyme can be attributed to degradation of β-gal on the particle surface where it is accessible to proteinase K, similarly to what was observed for the release profile of BSA. Furthermore, as proteinase K’s molar mass (28.9 kDa) is less than half of BSA, a fraction can diffuse into the microparticles and destroy more β-gal (464 kDa). However, this process is either very slow or the proteinase can only penetrate to a certain extent into the microparticles considering that after the overnight incubation 55% of β-gal remains.

Sustainability and Activity of Encapsulated Enzymes at the Single Microparticle Level

Encapsulation not only protects enzymes from their environment but also allows the formation of enzyme arrays consisting of discrete microreactors and the economic usage of their cargo. To illustrate the ability of our structures to accomplish this objective, a microfluidic trapping device (Figure 4a,b)45 was employed to immobilize the β-gal-filled silica microparticles to allow sequential generation of a fluorescent product and flushing steps by flowing either the substrate solution or buffer within the device. This is particularly relevant in cases when an enzymatic cascade reaction is used where the products of the first reaction are substrates of the second step. Carrying out such a cascade reaction under microconfined conditions suppresses the loss of intermediates through diffusive disposal. Five cycles were conducted, and time-lapse fluorescent images were recorded to monitor the enzymatic activity over time (Figure 4c,d). Plotting the intensity of the fluorescent product of the enzymatic reaction within the microparticles only exhibits marginal variation in individual signals, but, more importantly, the results show that the enzyme fully retains its activity between iterative cycles of washing steps and substrate introduction. This result demonstrates the potential to recycle the encapsulated enzyme and makes it a highly sustainable process.

Figure 4.

Figure 4

Open in a new tab

Stability of enzyme-filled silica microparticles. Microparticles (MC) were trapped within a microfluidic array device where the activity of single microparticles could be investigated (a). Image of microparticles trapped within the PDMS array device (b). Five substrate cycles with intermediate washing steps were conducted to monitor the enzyme response shown in fluorescent images (c) and the corresponding fluorescent intensity plot (d). The protection against enzyme degradation by proteinase K under flow conditions after 0 and 60 min (e) and intensity kinetics (f). Activity of β-gal (500 nM/microparticle) for each individual microparticle after exposure to proteinase K (33 μg/mL) (g). Error bars indicate the standard deviation of intensity within a single microparticle. These experiments were performed three times and typical responses were shown here.

The trapping device was further used to investigate protection against enzyme degradation by proteinase K at an individual microparticle level. Here, proteinase K was added to the buffer flown through the device, while the fluorescence intensity of the product formed within the microparticles was monitored for 60 min (Figure 4e,f). Fluorescence intensity trace decreases quickly within the first 15 min, most probably due to the degradation of exposed β-gal on the microparticle surface. However, the rate of degradation reduces and remains constant after ∼30 min. Similarly, to the microparticle bulk assay (Figure 3), approximately 50% of the enzyme remains active following exposure to proteinase K (Figure 4f,g), demonstrating the protective nature of the dense silica matrix even when the continuous supply of proteinase K is introduced to the system. Individual microparticles perform very similarly in this assay, indicating an excellent homogeneity in producing the enzyme-loaded silica microparticles and high reproducibility of the assays. It should be noted that our system retained a significant enzymatic activity (>50%) even in the presence of proteinase K flow, conditions that are not present in biology. Thus, based on these findings, a flow reactor for glucose sensing was established.

Microfluidic Continuous Flow Reactor for Glucose Sensing with Glucose Oxidase

The stability and reproducibility of the reaction cascade play a key role in the development of new sensing approaches based on enzymatic reactions. The ability to pack multiple microreactors within a single device allows increased product yield by the enzymatic reaction while stabilizing enzymes through encapsulation, thus allowing for highly accurate and dynamic sensing applications. Furthermore, due to the microfluidic generation method of the silica microparticles described here and the ability to pack them into tight arrays, increased enzyme concentrations can be employed compared to surface-based systems. Here, silica microparticles containing enzymes were packed into a microfluidic chamber where a continuous supply of a product can be generated while flowing the substrate through the chamber (Figure 5a). The reactor was filled with microparticles through the designated inlet. The pillar structures were after substrate inlet and product outlet ensuring the microparticles remained within the reaction chamber (Figure 5a).

Figure 5.

Figure 5

Open in a new tab

Microfluidic continuous flow “reactor”. Schematic representation of the microfluidic continuous flow reactor (a). Conversion yield of the substrate to the product in a single-enzyme system using β-gal-loaded droplets and continuous flow of resorufin β-D-galactopyranoside with three different concentrations (b). Michaelis–Menten kinetics plot and calculated constant KM and Vmax from reactor using 500 μL/h flow rate and different substrate concentrations (c). Glucose detection with an enzymatic cascade system by coencapsulation of glucose oxidase (GOx) and horseradish peroxidase (HRP) into the silica microparticles (d–f). Schematic representation of conversion of glucose to fluorescent signal with dual enzyme silica microparticles (d). Conversion yield of glucose to the fluorescent product through the enzymatic cascade two-enzyme system with changing flow rates for 3.125, 6.25 μM, and 12.5 μM glucose (e) and Michaelis–Menten kinetics for the two-enzyme system with concentrations ranging from 1.6 to 50 μM (f). Error bars indicate the standard deviation between at least three separate experiments.

First, a single-enzyme system based on β-gal and the resorufin substrate system was used. The resorufin substrate was flown into the main chamber of the device, filled with β-gal-loaded microparticles, in different concentrations and flow rates, while the fluorescence signal of the product was determined at the end of the device (Figure 5a, detection region), showing a linear relation with the substrate concentration (Figure S7). From the intensity-based conversion yields of the substrate into the product, measured at different flow rates, substrate concentrations were calculated based on calibration curves made with different concentrations of resorufin (Figure 5b). Since the flow rate is directly proportional to the reaction time in which the substrate can be converted into the fluorescent product, higher yields are achieved at lower flow rates. The two slowest flow rates, corresponding to reaction times of 10 and 5 s, give 100% substrate conversion at the investigated concentrations. Plotting the different reaction rates at 10, 30, 90, and 180 μM substrate concentrations, the Michaelis constant for this system was found to be KM = 76.2 ± 11.1 μM (Figure 5c).

Next, the reactor was applied for sensing glucose, which has potential applications as a diagnostic tool for diabetes. Recently, a lot of interest has been shown to develop noninvasive methods for glucose monitoring.46 We sought to investigate such a suitable platform by encapsulating glucose oxidase (GOx) together with horseradish peroxidase (HRP) (Figure 5d). These serve as a two-enzyme cascade, where HRP converts the oxidation product from glucose to a measurable signal using Amplex red (for bulk kinetics, see Figure S6). A similar trend of conversion rates (Figure 5e) can be seen for this two-enzyme cascade reaction as has been found in the single-enzyme system with β-gal. However, as expected with a two-enzyme system with lower reaction rates, yields are considerably decreased. Still, the resulting signal intensity is detectable at all flow rates and has been found to be directly proportional to initial substrate concentration. These concentrations are well below the concentrations of glucose found in human blood (2–40 mM) or urine (2.7 μM–5.55 mM) but similar to glucose concentrations of sweat (0.06 μM–0.11 mM) and saliva (0.23 μM–0.38 mM).47 Similarly to the single-enzyme system, a Michaelis–Menten-like correlation of the substrate concentration to the (here glucose) fluorescent signal of amplex red was found, allowing determination of KM (54.7 ± 16.0 μM, Figure 5f).

Conclusions

Performing enzymatic assays effectively and reliably even in challenging environments is crucial for converting or detecting molecules from biological systems. Conventional enzyme-based assays lack a protective capsule and have limited area for enzyme activity. Here, we have demonstrated the facile and robust generation of silica microparticles that can be used for recyclable enzymatic assays even in degrading environments. The formation of porous microparticles with encapsulated enzymes enables the use of these enzymes at relatively high concentrations in a protected environment without the need for chemical cross-linking. Furthermore, we show that these can be tightly packed in microfluidic continuous flow reactors in which full conversion yields of encapsulated enzymes such as β-gals (with the resorufin β-D-galactopyranoside substrate) are achieved. Extension of this approach to two-enzyme systems, in which the relative amount of enzymes is important, was further investigated. We demonstrated that the two-enzyme GOx/HRP system encapsulated in silica microparticles and organized to a tight array in a microfluidic reactor can be used for glucose sensing. Such a setup has direct applications for point-of-care measurements in diabetes. The presented technique can be further expanded for the encapsulation of a wide range of molecular species while maintaining their activity under deleterious environmental conditions, thus opening new possibilities for applications in areas ranging from the food industry and enzyme-based detergents to sensing and health.

Experimental Section

Peptide Synthesis

The 20-amino-acid R5 peptide was synthesized by Fmoc-based solid-phase peptide synthesis (SPPS) on Wang resin, either manually or using a Liberty Blue peptide synthesizer (CEM, Matthews, NC). For manual synthesis, hexafluorophosphate benzotriazole tetramethyl uronium (HBTU) was used as an activator. Deprotection was performed with 20% piperidine in DMF, and peptides were cleaved using 92.5% trifluoroacetic acid (TFA) with 5% tri-isopropylsilane (TIS) and 2.5% water. The peptides were purified by reverse-phase HPLC on a C4 column. Analytical data are shown in Figure S1.

Fabrication of Microfluidic devices

All microfluidic devices used were designed with AutoCAD software and fabricated by combining standard photolithography and soft lithography steps. Specifically, a 25 μm layer of a negative photoresist (SU-8 3025, MicroChem, Westborough, MA) was applied by spin coating and then soft baked for 15 min at 95 °C. A photomask was placed onto the wafer and then exposed to a UV lamp source for 60 s. After postbaking for 5 min, the unexposed photoresist was removed using propylene glycol methyl ether acetate (Sigma-Aldrich).

Specifically, a master mold was made by spin coating a 25 μm layer of a negative photoresist (SU-8 302, MicroChem, Westborough, MA) and soft baking for 15 min at 95 °C. First, a mixture of 10:1 prepolymer PDMS to a curing agent (Sylgard 184, DowCorning, Midland, MI) was poured onto the master. Bubbles were removed under vacuum, and PDMS was cured at 65 °C for at least 1 h. The devices were cut out, and inlet and outlet holes were punched. After treatment in a plasma oven for 30 s at 40 W (Diener Electronic), the device was bonded to a glass slide, which forms the bottom of the channels. Finally, the devices were coated with a polystyrene solution (Aquapel) to create hydrophobic surfaces.

Silica Droplet Formation

Silicic acid solutions were freshly prepared from 1 mM HCl and tetramethoxysilane (TMOS, ACROS Organics). For a 270 mM solution, 40 μL of TMOS was added to 960 μL of HCl and the mixture was vortexed until clear. A solution of 1 mg/mL R5 peptide in sodium phosphate buffer was prepared one day in advance to allow the peptide to dimerize. The solutions were flown through the devices using neMESYS syringe pumps (Cetoni, Korbussen, Germany).

Silica droplets were generated with a microfluidic droplet device with three different inlets. From the outer inlet, the continuous phase, fluorinated oil (Fluorinert FC-40, Sigma-Aldrich) with 2% w/w of fluorosurfactant (RAN biotechnologies), was flown with 700 μL/h rate. This was kept constant for all different conditions. However, the flows from the middle (silicic acid inlet) and inner (the peptide/buffer inlet) inlets were varied. Inner inlet: 50 mM sodium phosphate buffer pH 7, 250 μL/h (1:1), 333 μL/h (2:1), 375 μL/h (3:1), 400 μL/h (4:1), 450 μL/h (9:1); and middle inlet: 1080 mM silicic acid (160 μL TMOS in 840 μL 1 mM HCl), 250 μL/h (1:1), 167 μl/h (2:1), 125 μL/h (3:1), 100 μl/h (4:1), 50 μl/h (9:1), where the flow ratios of the inner inlet/middle inlet are given in the parentheses. The droplets were collected in an Eppendorf tube and left to incubate in RT for at least 30 min before further use.

Scanning Electron Microscopy (SEM)

Samples for SEM were spotted onto silicon wafer shards and left to dry in ambient conditions. After drying, the samples were sputtered with 10 nm Pt and the images were obtained using a Tescan MIRA3 instrument at 5 kV acceleration voltage. ImageJ was used for image analysis.

Release Studies

Silica droplets were created as aforementioned with the addition of 1 mM fluorescein sodium salt (SIGMA) or FITC-labeled BSA. To approximately 50 μL of droplets, 100 μL of FC-40 and 100 μL of 10% perfluorooctanoic acid (PFO) in FC-40 were added. The mixture was carefully inverted, spun down, and oil residue removed. This was repeated three times, and finally, 500 μL of 50 mM sodium phosphate buffer (pH 7) was added to the droplets. The supernatant was removed at each time point and replaced with fresh buffer, and the fluorescence reading of the supernatant was measured with a plate reader (CLARIOstar, MGlabtech) with the Fluorescein-FITC preset. The results were normalized with the maximum intensity expected for 100% release; thus, the time point 0 gives the encapsulation efficiency (EE%) for each particle and the cargo molecule (see Figure S4).

Enzyme Encapsulation

Enzymes, β-galactosidase by alone or horseradish peroxidase with glucose oxidase, were encapsulated into the silica droplets by adding the desired concentration of the enzyme to the phosphate buffer prior to droplet formation. Concentrations used were 1 μM for β-gal and 15 μM and 3 for HRP and GOx, respectively. The microparticles were formed as mentioned above.

Bulk Enzyme Studies

After microparticle formation and wash, 10 μL of each droplet solution was removed and added to a well of a 96-well plate (Corning 3881) as triplicates and 90 μL of phosphate buffer was added to each well. The enzyme kinetics were followed with a plate reader (Clariostar, BMGlabtech). Michelis–Menten kinetics were performed by varying the substrate concentration from 50 μM with 0.5× serial dilutions. In the case of digestion studies, 10 μM substrate was used with the addition of 33 μg/mL of proteinase K within the well. Free enzyme (125 nM) with and without proteinase K was added as a reference. For GOx/HRP studies, 500 μM Amplex Red reagent was added to the reaction buffer and glucose solution (50 μM with 0.5 × serial dilutions) was added to initialize the reaction.

Microparticle Trapping

Microparticles were confined using a microfluidic trapping array.33 First, the whole device was filled with buffer and all possible bubbles were erased. Next, microparticles in buffer were inserted into the device through the denoted inlet. This was followed by flushing with a buffer to remove any excess microparticles. Finally, experiments were conducted by flowing either the substrate (through substrate inlet) or buffer (through microparticle inlet) while continuously imaging using fluorescent microscopy.

Microfluidic Continuous Flow Reactor Studies for Enzyme Systems

Enzyme-filled microparticles were loaded into the device while still in oil, and the washing was conducted on chip by first flowing 10% PFO followed by a buffer wash (more detailed description in Figure S5). This was done to achieve the best packing of the microparticles. The studies were conducted by flowing the substrate at the desired concentration and flow rate through the microparticle-loaded device, and the fluorescence of the product was read at the end of the device. For GOx/HRP studies, 500 μM Amplex Red was added to the running buffer with changing the glucose concentration.

The intensity was used to determine the concentration of the product by comparing it to a calibration curve. Knowing the size of the reaction chamber and the average time it takes for a molecule to travel through it based on flow rate, the reaction rate (μM/min) could be calculated. Calculating the rate with different substrate concentrations, the reaction constant KM (μM) could be calculated using the Michaelis–Menten equation.

Acknowledgments

The research leading to these results has received funding from the EU Horizon 2020 programme, Marie Skłodolwska-Curie ITN (agreement No. 675007 to T.A.H., F.B., C.F.W.B., G.J.L.B.), and European Research Council under the European Union’s Seventh Framework Programme (FP7/2007-2013) through the ERC grant PhysProt (agreement no. 337969). The authors are grateful for financial support from the FEBS Long-Term Fellowship and the Oppenheimer Early Career Fellowship (A.L.), the Royal Society (URF80019 to G.J.L.B.), FCT Portugal (iFCT IF/00624/2015 to G.J.L.B.), the BBSRC (T.P.J.K.), the Austria Wirtschaftsservice (Grant P1716433-WZP01 to C.F.W.B.), the Newman Foundation (T.P.J.K.), the Welcome Trust (T.P.J.K.), and the Cambridge Centre for Misfolding Diseases.

Glossary

Abbreviations

β-gal

β-galactosidase

BSA

bovine serum albumin

GOx

glucose oxidase

FITC

fluorescein isothiocyanate

HRP

horseradish peroxidase

MP

microparticle

SEM

scanning electron microscopy

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsami.0c09226.

  • Analytic data from R5 peptide synthesis, SEM of silica microparticle’ cross section, fluorescence recovery after photobleaching, encapsulation efficiency, operation of the microfluidic reactor, Michaelis–Menten kinetics, substrate response of the microfluidic reactor, and calculated reaction times (PDF)

  • Videos constructed from confocal microscopy images taken from different depths through the silica particles for each silicic acid concentration (270 mM with a 9:1 flow ratio, 540 mM with a 3:1 flow ratio, and 1080 mM with a 1:1 flow ratio) (AVI)

  • Microparticles_270mMSilicicAcid_Cy5R5_confocal_video (AVI)

  • Microparticles_540mMSilicicAcid_Cy5R5_confocal_video (AVI)

  • Microparticles_1080mMSilicicAcid_Cy5R5_confocal_video (AVI)

Author Contributions

⊥ T.A.H. and F.B. contributed equally.

The authors declare no competing financial interest.

Supplementary Material

am0c09226_si_001.pdf (1.5MB, pdf)
am0c09226_si_002.avi (46.2MB, avi)
am0c09226_si_003.avi (596.1KB, avi)
am0c09226_si_004.avi (651.1KB, avi)
am0c09226_si_005.avi (926.4KB, avi)

References

  1. Agapakis C. M.; Boyle P. M.; Silver P. A. Natural Strategies for the Spatial Optimization of Metabolism in Synthetic Biology. Nat. Chem. Biol. 2012, 8, 527–535. 10.1038/nchembio.975. [DOI] [PubMed] [Google Scholar]
  2. Chen W.-H.; Vázquez-González M.; Zoabi A.; Abu-Reziq R.; Willner I. Biocatalytic Cascades Driven by Enzymes Encapsulated in Metal–Organic Framework Nanoparticles. Nat. Catal. 2018, 1, 689–695. 10.1038/s41929-018-0117-2. [DOI] [Google Scholar]
  3. Beldengrün Y.; Aragon J.; Prazeres S. F.; Montalvo G.; Miras J.; Esquena J. Gelatin/Maltodextrin Water-in-Water (W/W) Emulsions for the Preparation of Cross-Linked Enzyme-Loaded Microgels. Langmuir 2018, 34, 9731–9743. 10.1021/acs.langmuir.8b01599. [DOI] [PubMed] [Google Scholar]
  4. Schachschal S.; Adler H.-J.; Pich A.; Wetzel S.; Matura A.; van Pee K.-H. Encapsulation of Enzymes in Microgels by Polymerization/Cross-Linking in Aqueous Droplets. Colloid Polym. Sci. 2011, 289, 693–698. 10.1007/s00396-011-2392-1. [DOI] [Google Scholar]
  5. Cui J.; Ren S.; Sun B.; Jia S. Optimization Protocols and Improved Strategies for Metal-Organic Frameworks for Immobilizing Enzymes: Current Development and Future Challenges. Coord. Chem. Rev. 2018, 370, 22–41. 10.1016/j.ccr.2018.05.004. [DOI] [Google Scholar]
  6. Márquez A.; Kocsis K.; Zickler G.; Bourret G. R.; Feinle A.; Hüsing N.; Himly M.; Duschl A.; Berger T.; Diwald O. Enzyme Adsorption-Induced Activity Changes: A Quantitative Study on TiO2 Model Agglomerates. J. Nanobiotechnology 2017, 15, 55. 10.1186/s12951-017-0283-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Nguyen L. T.; Yang K. L. Combined Cross-Linked Enzyme Aggregates of Horseradish Peroxidase and Glucose Oxidase for Catalyzing Cascade Chemical Reactions. Enzyme Microb. Technol. 2017, 100, 52–59. 10.1016/j.enzmictec.2017.02.007. [DOI] [PubMed] [Google Scholar]
  8. Du Y.; Gao J.; Zhou L.; Ma L.; He Y.; Huang Z.; Jiang Y. Enzyme Nanocapsules Armored by Metal-Organic Frameworks: A Novel Approach for Preparing Nanobiocatalyst. Chem. Eng. J. 2017, 327, 1192–1197. 10.1016/j.cej.2017.07.021. [DOI] [Google Scholar]
  9. Iyer P. V.; Ananthanarayan L. Enzyme Stability and Stabilization-Aqueous and Non-Aqueous Environment. Process Biochem. 2008, 43, 1019–1032. 10.1016/j.procbio.2008.06.004. [DOI] [Google Scholar]
  10. Mateo C.; Palomo J. M.; Fernandez-Lorente G.; Guisan J. M.; Fernandez-Lafuente R. Improvement of Enzyme Activity, Stability and Selectivity via Immobilization Techniques. Enzyme Microb. Technol. 2007, 40, 1451–1463. 10.1016/j.enzmictec.2007.01.018. [DOI] [Google Scholar]
  11. Fernandez-Lafuente R. Stabilization of Multimeric Enzymes: Strategies to Prevent Subunit Dissociation. Enzyme Microb. Technol. 2009, 45, 405–418. 10.1016/j.enzmictec.2009.08.009. [DOI] [Google Scholar]
  12. Hernandez K.; Fernandez-Lafuente R. Control of Protein Immobilization: Coupling Immobilization and Site-Directed Mutagenesis to Improve Biocatalyst or Biosensor Performance. Enzyme Microb. Technol. 2011, 48, 107–122. 10.1016/j.enzmictec.2010.10.003. [DOI] [PubMed] [Google Scholar]
  13. Ren S.; Li C.; Jiao X.; Jia S.; Jiang Y.; Bilal M.; Cui J. Recent Progress in Multienzymes Co-Immobilization and Multienzyme System Applications. Chem. Eng. J. 2019, 373, 1254–1278. 10.1016/j.cej.2019.05.141. [DOI] [Google Scholar]
  14. Matosevic S.; Szita N.; Baganz F. Fundamentals and Applications of Immobilized Microfluidic Enzymatic Reactors. J. Chem. Technol. Biotechnol. 2011, 86, 325–334. 10.1002/jctb.2564. [DOI] [Google Scholar]
  15. Bolivar J. M.; Wiesbauer J.; Nidetzky B. Biotransformations in Microstructured Reactors: More than Flowing with the Stream?. Trends Biotechnol. 2011, 29, 333–342. 10.1016/j.tibtech.2011.03.005. [DOI] [PubMed] [Google Scholar]
  16. Hustoft H. K.; Brandtzaeg O. K.; Rogeberg M.; Misaghian D.; Torsetnes S. B.; Greibrokk T.; Reubsaet L.; Wilson S. R.; Lundanes E. Integrated Enzyme Reactor and High Resolving Chromatography in “Sub-Chip” Dimensions for Sensitive Protein Mass Spectrometry. Sci. Rep. 2013, 3, 3511 10.1038/srep03511. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Bolivar J. M.; Tribulato M. A.; Petrasek Z.; Nidetzky B. Let the Substrate Flow, Not the Enzyme: Practical Immobilization of d -Amino Acid Oxidase in a Glass Microreactor for Effective Biocatalytic Conversions. Biotechnol. Bioeng. 2016, 113, 2342–2349. 10.1002/bit.26011. [DOI] [PubMed] [Google Scholar]
  18. Valikhani D.; Bolivar J. M.; Pfeiffer M.; Nidetzky B. Multivalency Effects on the Immobilization of Sucrose Phosphorylase in Flow Microchannels and Their Use in the Development of a High-Performance Biocatalytic Microreactor. ChemCatChem 2017, 9, 161–166. 10.1002/cctc.201601019. [DOI] [Google Scholar]
  19. Liu X.; Zhu X.; Camara M. A.; Qu Q.; Shan Y.; Yang L. Surface Modification with Highly-Homogeneous Porous Silica Layer for Enzyme Immobilization in Capillary Enzyme Microreactors. Talanta 2019, 197, 539–547. 10.1016/j.talanta.2019.01.080. [DOI] [PubMed] [Google Scholar]
  20. McDonald J. C.; Whitesides G. M. Poly(Dimethylsiloxane) as a Material for Fabricating Microfluidic Devices. Acc. Chem. Res. 2002, 35, 491–499. 10.1021/ar010110q. [DOI] [PubMed] [Google Scholar]
  21. Sotowa K.-I.; Miyoshi R.; Lee C.-G.; Kang Y.; Kusakabe K. Mixing and Enzyme Reactions in a Microchannel Packed with Glass Beads. Korean J. Chem. Eng. 2005, 22, 552–555. 10.1007/BF02706641. [DOI] [Google Scholar]
  22. Li Y.; Yan B.; Deng C.; Yu W.; Xu X.; Yang P.; Zhang X. Efficient On-Chipproteolysis System Based on Functionalized Magnetic Silica Microspheres. Proteomics 2007, 7, 2330–2339. 10.1002/pmic.200700112. [DOI] [PubMed] [Google Scholar]
  23. Ma J.; Liang Z.; Qiao X.; Deng Q.; Tao D.; Zhang L.; Zhang Y. Organic-Inorganic Hybrid Silica Monolith Based Immobilized Trypsin Reactor with High Enzymatic Activity. Anal. Chem. 2008, 80, 2949–2956. 10.1021/ac702343a. [DOI] [PubMed] [Google Scholar]
  24. Jiang S.; Zhang Z.; Li L. A One-Step Preparation Method of Monolithic Enzyme Reactor for Highly Efficient Sample Preparation Coupled to Mass Spectrometry-Based Proteomics Studies. J. Chromatogr. A 2015, 1412, 75–81. 10.1016/j.chroma.2015.07.121. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Naldi M.; Černigoj U.; Štrancar A.; Bartolini M. Towards Automation in Protein Digestion: Development of a Monolithic Trypsin Immobilized Reactor for Highly Efficient on-Line Digestion and Analysis. Talanta 2017, 167, 143–157. 10.1016/j.talanta.2017.02.016. [DOI] [PubMed] [Google Scholar]
  26. Cui J.; Jia S.; Liang L.; Zhao Y.; Feng Y. Mesoporous CLEAs-Silica Composite Microparticles with High Activity and Enhanced Stability. Sci. Rep. 2015, 5, 14203 10.1038/srep14203. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Ding L.; Zhu X.; Wang Y.; Shi B.; Ling X.; Chen H.; Nan W.; Barrett A.; Guo Z.; Tao W.; Wu J.; Shi X. Intracellular Fate of Nanoparticles with Polydopamine Surface Engineering and a Novel Strategy for Exocytosis-Inhibiting, Lysosome Impairment-Based Cancer Therapy. Nano Lett. 2017, 17, 6790–6801. 10.1021/acs.nanolett.7b03021. [DOI] [PubMed] [Google Scholar]
  28. Zhang K.; Xu L. L.; Jiang J. G.; Calin N.; Lam K. F.; Zhang S. J.; Wu H. H.; Wu G. D.; Albela B.; Bonneviot L.; Wu P. Facile Large-Scale Synthesis of Monodisperse Mesoporous Silica Nanospheres with Tunable Pore Structure. J. Am. Chem. Soc. 2013, 135, 2427–2430. 10.1021/ja3116873. [DOI] [PubMed] [Google Scholar]
  29. Xu S.; Nie Z.; Seo M.; Lewis P.; Kumacheva E.; Stone H. A.; Garstecki P.; Weibel D. B.; Gitlin I.; Whitesides G. M. Generation of Monodisperse Particles by Using Microfluidics: Control over Size, Shape, and Composition. Angew. Chem. 2005, 117, 734–738. 10.1002/ange.200462226. [DOI] [PubMed] [Google Scholar]
  30. Choi C.-H.; Jung J.-H.; Rhee Y. W.; Kim D.-P.; Shim S.-E.; Lee C.-S. Generation of Monodisperse Alginate Microbeads and in Situ Encapsulation of Cell in Microfluidic Device. Biomed. Microdevices 2007, 9, 855–862. 10.1007/s10544-007-9098-7. [DOI] [PubMed] [Google Scholar]
  31. Leng X.; Zhang W.; Wang C.; Cui L.; Yang C. J. Agarose Droplet Microfluidics for Highly Parallel and Efficient Single Molecule Emulsion PCR. Lab Chip 2010, 10, 2841. 10.1039/c0lc00145g. [DOI] [PubMed] [Google Scholar]
  32. Shimanovich U.; Ruggeri F. S.; De Genst E.; Adamcik J.; Barros T. P.; Porter D.; Müller T.; Mezzenga R.; Dobson C. M.; Vollrath F.; Holland C.; Knowles T. P. J. Silk Micrococoons for Protein Stabilisation and Molecular Encapsulation. Nat. Commun. 2017, 8, 15902 10.1038/ncomms15902. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Liu X.; Toprakcioglu Z.; Dear A. J.; Levin A.; Ruggeri F. S.; Taylor C. G.; Hu M.; Kumita J. R.; Andreasen M.; Dobson C. M.; Shimanovich U.; Knowles T. P. J. Fabrication and Characterization of Reconstituted Silk Microgels for the Storage and Release of Small Molecules. Macromol. Rapid Commun. 2019, 40, e1800898 10.1002/marc.201800898. [DOI] [PubMed] [Google Scholar]
  34. Sumper M. A Phase Separation Model for the Nanopatterning of Diatom Biosilica. Science 2002, 295, 2430–2433. 10.1126/science.1070026. [DOI] [PubMed] [Google Scholar]
  35. Lechner C.; Becker C. Silaffins in Silica Biomineralization and Biomimetic Silica Precipitation. Mar. Drugs 2015, 13, 5297–5333. 10.3390/md13085297. [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Kröger N.; Deutzmann R.; Sumper M. Polycationic Peptides from Diatom Biosilica That Direct Silica Nanosphere Formation. Science 1999, 286, 1129–1132. 10.1126/science.286.5442.1129. [DOI] [PubMed] [Google Scholar]
  37. Luckarift H. R.; Spain J. C.; Naik R. R.; Stone M. O. Enzyme Immobilization in a Biomimetic Silica Support. Nat. Biotechnol. 2004, 22, 211–213. 10.1038/nbt931. [DOI] [PubMed] [Google Scholar]
  38. Naik R. R.; Tomczak M. M.; Luckarift H. R.; Spain J. C.; Stone M. O. Entrapment of Enzymes and Nanoparticles Using Biomimetically Synthesized Silica. Chem. Commun. 2004, 4, 1684–1685. 10.1039/b404586f. [DOI] [PubMed] [Google Scholar]
  39. Toprakcioglu Z.; Levin A.; Knowles T. P. J. Hierarchical Biomolecular Emulsions Using 3-D Microfluidics with Uniform Surface Chemistry. Biomacromolecules 2017, 18, 3642–3651. 10.1021/acs.biomac.7b01159. [DOI] [PubMed] [Google Scholar]
  40. Lechner C. C.; Becker C. F. W. Exploring the Effect of Native and Artificial Peptide Modifications on Silaffin Induced Silica Precipitation. Chem. Sci. 2012, 3, 3500–3504. 10.1039/c2sc20687k. [DOI] [Google Scholar]
  41. Lechner C. C.; Becker C. F. W. Immobilising Proteins on Silica with Site-Specifically Attached Modified Silaffin Peptides. Biomater. Sci. 2015, 3, 288–297. 10.1039/C4BM00310A. [DOI] [PubMed] [Google Scholar]
  42. Yates E. V.; Müller T.; Rajah L.; De Genst E. J.; Arosio P.; Linse S.; Vendruscolo M.; Dobson C. M.; Knowles T. P. J.. Latent Analysis of Unmodified Biomolecules and Their Complexes in Solution with Attomole Detection Sensitivity 2015, 7, 802–809. 10.1038/nchem.2344. [DOI] [PubMed]
  43. Wittrup K. D.; Bailey J. E. A Single-Cell Assay of β-Galactosidase Activity InSaccharomyces Cerevisiae. Cytometry 1988, 9, 394–404. 10.1002/cyto.990090418. [DOI] [PubMed] [Google Scholar]
  44. Hadd A. G.; Raymond D. E.; Halliwell J. W.; Jacobson S. C.; Ramsey J. M. Microchip Device for Performing Enzyme Assays. Anal. Chem. 1997, 69, 3407–3412. 10.1021/ac970192p. [DOI] [PubMed] [Google Scholar]
  45. Toprakcioglu Z.; Challa P. K.; Levin A.; Knowles T. P. J. Observation of Molecular Self-Assembly Events in Massively Parallel Microdroplet Arrays. Lab Chip 2018, 18, 3303–3309. 10.1039/C8LC00862K. [DOI] [PubMed] [Google Scholar]
  46. Kim J.; Campbell A. S.; Wang J. Wearable Non-Invasive Epidermal Glucose Sensors: A Review. Talanta 2018, 177, 163–170. 10.1016/j.talanta.2017.08.077. [DOI] [PubMed] [Google Scholar]
  47. Bruen D.; Delaney C.; Florea L.; Diamond D. Glucose Sensing for Diabetes Monitoring: Recent Developments. Sensors 2017, 17, E1866. 10.3390/s17081866. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

am0c09226_si_001.pdf (1.5MB, pdf)
am0c09226_si_002.avi (46.2MB, avi)
am0c09226_si_003.avi (596.1KB, avi)
am0c09226_si_004.avi (651.1KB, avi)
am0c09226_si_005.avi (926.4KB, avi)

Articles from ACS Applied Materials & Interfaces are provided here courtesy of American Chemical Society

RESOURCES