Crystal Growth & Design - ACS Publications - American Chemical


Crystal Growth & Design - ACS Publications - American Chemical...

0 downloads 81 Views 5MB Size

Article pubs.acs.org/crystal

Influence of Alternating Electric Fields on Protein Crystallization in Microfluidic Devices with Patterned Electrodes in a Parallel-Plate Configuration Fei Li and Richard Lakerveld* Department of Chemical and Biomolecular Engineering, The Hong Kong University of Science and Technology, Hong Kong, China ABSTRACT: Improved control over protein nucleation is important to advance the design and operation of protein separation and purification processes. The influence of nonuniform electric fields induced by patterned indium tin oxide (ITO) electrodes of various shapes and surface areas on protein nucleation in microfluidic devices with parallel electrodes is investigated experimentally. In particular, the impact of various electrode designs and electric-field properties on the solid-state form, induction time, nucleation rate, and the location of lysozyme crystals is described. The results demonstrate that both enhanced and inhibited protein crystallization can be observed depending on the specifics of electrode shape and surface area and electric-field properties. The nucleation location of crystals is found to be influenced by both the ITO layer as a template and the nonuniform electric field induced by specific designs of electrodes. The optimal electric-field conditions and electrode design for enhancement of lysozyme crystallization could be extended to insulin crystallization experiments, which showed the formation of an insulin crystal near the electrode, whereas control experiments did not show any crystals.



INTRODUCTION The global market for biopharmaceuticals is growing rapidly with current revenues exceeding $100 billion.1 The competitive business environment calls for shorter development times, improved manufacturing flexibility, and reduction of batch-tobatch variability. A typical challenge in the manufacture of biopharmaceuticals is separating a protein from a solution with impurities. Chromatography is the current industrial workhorse for separating proteins. Unfortunately, chromatography suffers from inherent challenges in terms of scaling and throughput.1,2 Crystallization of proteins is widely regarded as an alternative technology to replace or supplement chromatography for the separation of proteins.1−4 Crystallization is a highly selective process, yielding high purities in a single processing step that can operate over a range of scales and throughput. Furthermore, a crystalline pharmaceutical product is generally purer and more stable than other forms. Unfortunately, the design and control of crystallization processes involving large biomolecules remains notoriously difficult. Fundamental challenges include the random nature of nucleation, the lack of control over various physical phenomena in the process, and limited means of manipulating protein crystallization at the relevant scales. Therefore, improved methods to control protein crystallization are needed. Crystallization of proteins is frequently applied for structural analysis of proteins with X-ray crystallography.5 The general approach is to promote supersaturation in the system by adjusting global variables such as concentration, temperature,6 pH,7 or solvent composition.8 Furthermore, several studies © 2017 American Chemical Society

have been conducted to analyze and optimize protein crystallization processes with both mathematical models and small-scale experiments.9−12 A key challenge in such processes is sufficient spatial and dynamic control over protein nucleation. Methods of initiating protein nucleation by using templates,13−15 solvent freeze-out,16 a focused laser beam,17 ultrasound,18 magnetic fields,19 direct-current electric fields,20,21 alternating-current electric fields,22−25 and combinations of magnetic and electric fields have been reported.26 In general, clear evidence has been presented that under a broad range of conditions such methods can deliver faster nucleation times and larger, higher quality crystals. Electric fields are of particular interest for process design and control because of their tunable direction and strength, which creates the potential of improving control over protein nucleation in a flexible way. The utilization of electric fields has been demonstrated to enhance protein crystallization by controlling the nucleation rate and location.27 The first proof of principle on the use of an electric field to enhance protein crystallization was delivered for lysozyme in a vapor diffusion setup.28,29 It was found that an external direct current (DC) electric field creates fewer crystals with larger sizes near the droplet surface on the cathode side of the electrode. Subsequently, different configurations have been investigated, and theoretical explanations have been proposed.30,31 The method has been applied to batch protein Received: December 16, 2016 Revised: April 13, 2017 Published: May 10, 2017 3062

DOI: 10.1021/acs.cgd.6b01846 Cryst. Growth Des. 2017, 17, 3062−3070

Crystal Growth & Design



crystallization by bringing a supersaturated solution into direct20,22,32 or indirect33,34 contact with electrodes and by using DC or alternating-current (AC) potential differences for a number of devices.35−37 In principle, high flexibility for design and operation of electric-field-assisted protein crystallization can be achieved as the potential energy landscape can be shaped by adapting different types of electrodes, e.g., wire electrode,38 parallel-plate electrodes,37 and coplanar electrodes,23 by varying the spatial arrangement of the electrodes,39 and by manipulating the electric-field properties such as the frequency and strength.27 Most studies conducted so far use uniform electrodes. However, patterned electrodes provide the opportunity to design a potential-energy landscape flexibly. Hou and Chang23 constructed interdigitated and quadrupole Ti/Au electrodes for protein crystallization and found that the number of nucleation sites was reduced. Furthermore, crystal quality was improved under the influence of the nonuniform electric field by the patterned electrodes. In addition, it was observed that for quadrupole electrodes crystals tended to form at the field minima as a result of a dielectrophoretic force. Typically, indium tin oxide (ITO) coated glass slides are used for patterning, having the advantage of optical transparency, high electrical conductivity, and easy fabrication following standard lithographic methods.40,41 In principle, patterned ITO electrodes can impact protein nucleation in at least two ways. First, ITO can act as a template for nucleation. Indeed, it has been reported that an ITO surface can lead to a shorter crystal growth time and a larger final size compared to other metal surfaces or poly(methyl methacrylate).42 Second, electrokinetic phenomena induced by nonuniform electric fields may manipulate local conditions in a cell to enhance or suppress nucleation. Parallel patterned electrodes have been used to control directed self-assembly processes of colloidal particles by exploiting various electrokinetic phenomena43−45 but have not been investigated for use in protein crystallization to the best of our knowledge. The question is whether a parallel type of electrode configuration can also manipulate protein crystallization using nonuniform electric fields. The objective of this work is to investigate how AC electric fields induced by electrodes with different patterns and surface areas can influence protein crystallization in a parallel-plate configuration. Lysozyme and insulin are used as model proteins, which were crystallized in a transparent electricfield-assisted microfluidic device. The crystallization behavior is characterized in terms of induction time, crystal number, and location using an optical microscope. A particular focus will be placed on the role of ITO as a template for nucleation in addition to its role as an electrode. Patterned electrodes are particularly suitable to identify any templating effect of ITO when operated in the absence of electric fields since any bias in nucleation location can be identified. First, optimal experimental conditions for enhancement of lysozyme crystallization are determined in terms of electrode design and electric-field conditions. Subsequently, those optimal conditions are applied to insulin crystallization to investigate to what extent the results for lysozyme crystallization translate to a different protein. In principle, many different proteins could be investigated in the proposed experimental setup. However, insulin was selected because of its practical relevance and since it has a much lower isoelectric point (5.4) compared to lysozyme (11), which is expected to affect the crystallization behavior in an electric field.

Article

EXPERIMENTAL SECTION

Materials. Lysozyme from chicken egg white (≥98%, L4919), insulin solution (human, 9.5−11.5 mg/mL, I9278), and acetate buffer solution (pH 4.65) were purchased from Sigma-Aldrich. Sodium chloride, citric acid (anhydrous), trisodium citrate, and acetone were purchased from VWR. Zinc sulfate heptahydrate was purchased from BBI Life Sciences. All the chemicals were used without further purification. Solution Preparation. Sodium chloride was used as the precipitant for crystallizing lysozyme. 6% (w/v) sodium chloride in acetate buffer was mixed with 60 mg/mL of lysozyme dissolved in the same buffer at 1:1 ratio after filtration through 0.22 μm syringe filters.37 Citrate buffer (0.48 M, pH = 6) containing 10% (v/v) acetone and 1% (w/v) zinc sulfate was used as the precipitant solution for insulin crystallization by mixing with an equal volume of ∼3 mg/mL insulin solution.46 For insulin solutions with a concentration higher than 5 mg/mL, nucleation was observed almost instantaneously at room temperature, and different solid states were observed. Therefore, a lower insulin concentration (∼3 mg/mL) was used to achieve a relatively long induction time so that any effect of the electric field could be better investigated. Fabrication of Patterned Electrodes. Standard microfabrication techniques were used to make the patterned electrodes. Figure 1(a)

Figure 1. (a) Patterns used for fabricating the patterned ITO electrodes on glass slides and (b) a microfluidic device connected to a function generator via copper tape. shows eight electrode patterns with six duplicates each, which were all designed in LayoutEditor (juspertor GmbH, Germany) to fabricate the mask for photolithography. Table 1 shows the ITO area ratio φITO, the ratio of ITO surface area to the total area of the spherical sample well (diameter of 8 mm), for each pattern. A 4″ × 4″ ITO-coated glass (KINTEC) was cleaned by sonicating in Contrad70, acetone, isopropyl alcohol, and DI water for 15 min each. Then, the ITOcoated glass was primed with HMDS, spin-coated with positive photoresist (HPR504, thickness of ∼1.5 μm), and exposed to UV light under the mask. The exposed photoresist was dissolved in FHD-5 3063

DOI: 10.1021/acs.cgd.6b01846 Cryst. Growth Des. 2017, 17, 3062−3070

Crystal Growth & Design

Article

Table 1. Description of Experiments with an Electric Field no.

protein

pattern no.

ITO area ratio ϕITO

amplitude (Vpp)

frequency (Hz)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18

lysozyme lysozyme lysozyme lysozyme lysozyme lysozyme lysozyme lysozyme lysozyme lysozyme lysozyme lysozyme lysozyme lysozyme lysozyme lysozyme lysozyme insulin

1 2 3 4 5 6 7 8 uniform 1 2 8 uniform 1 2 8 uniform 8

0.51 0.41 0.28 0.21 0.18 0.15 0.18 0.27 1.00 0.51 0.41 0.27 1.00 0.51 0.41 0.27 1.00 0.27

2 2 2 2 2 2 2 2 2 10 10 10 10 2 2 2 2 10

1M 1M 1M 1M 1M 1M 1M 1M 1M 1M 1M 1M 1M 100 100 100 100 1M

Figure 2. Different crystal modifications observed: (a) tetragonal; (b) hexagonal; (c,d) needle-shaped for experiments in MFDs without an electric field.

developer uncovering the ITO layer underneath that was, subsequently, etched away in the aqua regia. Finally, the remaining photoresist was stripped in MS-2001, and the patterned ITO-coated glass was cleaned and cut into 2″ × 1″ pieces. Experimental Setup. The microfluidic device (MFD) consisted of an ITO-patterned bottom glass slide and a uniformly ITO-coated top glass slide that acted as electrodes. The electrodes were separated by a double-sided adhesive silicone isolator (thickness of ∼0.8 mm, Grace Bio-Laboratories). Six sample wells were cut in the rubber spacer using a biopsy punch (8 mm in diameter, Integra Miltex), which matched the repeated patterns on the bottom electrode such that six tests could be conducted in parallel for each device. After adding the mixture of protein solution and precipitating agent into the sample wells (∼40 μL solution per well), the device was closed carefully to avoid the formation of bubbles. Subsequently, each glass slide was connected to a function generator (33500B, Keysight Technologies) via copper tape with conductive adhesive (Agar Scientific) for 7 h. An example of a microfluidic device can be seen in Figure 1(b). All the experiments were carried out at room temperature and kept under observation for at least 3 days using an optical microscope (Nikon, Ni−U) with camera (Nikon, Digital Sight Qi2 cooled camera system). The devices were inspected for crystals approximately every 2 h during the daytime for the first 2 days of the experiments and once more at 72 h after the start of the experiments to estimate the induction time for each well. Table 1 shows all the experiments conducted with an electric field for lysozyme and insulin. The optimal conditions for enhancement of lysozyme crystallization were duplicated for insulin crystallization (experiment no. 18). For all experiments, corresponding control experiments were conducted in the absence of an electric field (for lysozyme and insulin) and in devices made of glass only (for lysozyme). Considering the stochastic nature of nucleation, the experiments were each repeated producing a total of 12 tests (wells) for each condition. Crystals that did not nucleate in the bulk solution, but on fibers, surfaces of bubbles, or on the inside of the rubber spacers, were neglected. Finally, the location of each crystal was determined by observing whether the center of a crystal was located on the ITO or on the glass surface.

temperature.6 It is known that lysozyme can also crystallize as a needle, which was reported by Alderton and Fevold47 for lysozyme crystallization at a pH between 7.0 and 11. Needles can also be formed at a higher salt concentration (5% NaCl) and a lower pH of 4.548 or when using polymer-induced heterogeneous nucleation. 49 Yu et al.50 measured the solubilities of tetragonal crystals and needle crystals at different pH with a salt concentration of 4%. It was found that at a pH of 4.5 the solubility of tetragonal lysozyme crystals is lower compared to that of needle crystals, indicating the tetragonal form is more stable. No solubility information is available to the best of our knowledge for the exact conditions used in this work (3% NaCl at a pH of 4.65), but it can be estimated that lysozyme crystals with a needle-shaped morphology are metastable at the conditions used in this work, which is consistent with our results. However, in a minority of cases, besides tetragonal and hexagonal crystals, needle-shaped crystals were also formed in several tests where patterned electrodes were used. In some cases, after 2 days of the experiments, gas bubbles could be observed inside the wells (e.g., see Figure 2(c)). The formation of gas bubbles increased the protein concentration, which may have led to the formation of needle-like crystals in addition to the tetragonal or hexagonal crystals when the concentration surpassed the solubility of both solid-state forms. Furthermore, in a small number of cases, needle-like crystals were observed when no bubbles formed in the wells (e.g., see Figure 2(d)). In principle, heterogeneous nucleation on the patterned ITO surface might have induced the formation of the metastable needle morphology, but a detailed investigation on this topic is considered outside the scope of the present work. Figure 3(a,b) illustrates the different time periods at which crystals appeared in the wells and the average number of crystals in each well at 72 h as a function of the ITO area ratio for crystallization in MFDs with different electrode designs in the absence of an electric field. Here, the first group with ϕITO =



RESULTS AND DISCUSSION Nucleation of Lysozyme on Patterned Electrodes in the Absence of Electric Fields. Lysozyme crystals with a tetragonal or hexagonal morphology were typically observed at different locations within the wells (see Figure 2(a,b)), which is expected for experiments conducted with a pH of 4.65 at room 3064

DOI: 10.1021/acs.cgd.6b01846 Cryst. Growth Des. 2017, 17, 3062−3070

Crystal Growth & Design

Article

Figure 3. (a) Number of wells that had crystals during different induction times, (b) average number of crystals in each well at 72 h, and (c) average fraction of crystals nucleated on the ITO surface as a function of the ITO area ratio for experiments in MFDs made with different electrode designs in the absence of an electric field.

Figure 4. (a) Number of wells that had crystals during different induction times, (b) average number of crystals in each well at 72 h, and (c) average fraction of crystals nucleated on the ITO surface as a function of the ITO area ratio for experiments in MFDs made with different electrode designs with an electric field of 2 Vpp and 1 MHz.

0.18 refers to experiment using pattern 5, while the second group refers to experiment using pattern 7 (same in Figure 4(a,b)). Note that measured induction times are grouped within certain time periods to account for the discrete nature of the observations for nucleated crystals, and the solid vertical lines correspond to the 90% confidence intervals of all 12 observations. In general, the variation in nucleation rate was quite high with most error bars in Figure 3(b) overlapping, which is typical for protein nucleation in the absence of any specific controls. In addition, as shown in Figure 3(a), for tests under the same conditions, the induction times ranged from 7 h up to 72 h or even longer (beyond the observation time). Nevertheless, some general trends can be observed. In particular, in the MFDs made out of glass with no ITO

electrode (ϕITO = 0), just one crystal was observed in 1 out of the 12 wells at around 47 h after the start of the experiment. However, the nucleation rate was increased when ITO was present in any form. For the cases with a uniform or patterned ITO surface, the average induction time significantly decreased, and the average number of crystals significantly increased compared to the control case without ITO. By introducing patterned electrodes, crystals were observed as early as 7 h after the start of the experiments, and crystals were observed at least in some of the parallel wells within 24 h for almost all of the patterns. Although the average number of crystals for experiments using certain patterns varied between different patterns, statistically, significant differences cannot be observed as the error bars in most cases overlapped. A possible exception 3065

DOI: 10.1021/acs.cgd.6b01846 Cryst. Growth Des. 2017, 17, 3062−3070

Crystal Growth & Design

Article

Figure 5. (a−h) Microscope images taken for lysozyme crystals nucleated on ITO surface (except (f) for pattern 6 where the crystal nucleated on glass) in MFDs with patterned electrodes (patterns 1−8 from left to right) subject to an electric field of 2 Vpp and 1 MHz. The dashed lines illustrate the edges of electrodes.

eight patterns with different shapes and ITO area ratios were tested for lysozyme crystallization with the same electric-field properties (2 Vpp and 1 MHz, experiments 1−8). Like the control experiments conducted without an electric field (see previous section), needle-shaped crystals were also observed after the formation of bubbles in one test when an electric field of 2 Vpp and 1 MHz was applied. Figure 4(a) illustrates the number of wells that had crystals formed during different time periods after the start of the experiments as a function of the ITO area ratio for crystallization in MFDs with different electrode designs with an electric field of 2 Vpp and 1 MHz. For most conditions, crystals were observed at least in some of the wells within 24 h of the experiments regardless of pattern and electric field. For MFDs with uniform electrodes (ϕITO = 1.00), the mean induction time was decreased when an electric field of 2 Vpp and 1 MHz was applied, while for MFDs with patterned electrodes the results in general varied between the different patterns. For example, crystallization was promoted by an electric field of 2 Vpp and 1 MHz for patterns 4 (ϕITO = 0.21) and 8 (ϕITO = 0.27) illustrated by the high number of wells that had crystals and the reduced mean induction time. With the same ϕITO (0.18), pattern 5 led to a faster nucleation rate compared to pattern 7. Figure 4(b) shows the average number of crystals observed in each well at 72 h as a function of the ITO area ratio for crystallization in MFDs with different electrode designs with an electric field of 2 Vpp and 1 MHz. The number of crystals observed was mostly limited up to 10, although in some wells the number could be more than 20. When applying an electric field of 2 Vpp and 1 MHz, in general, the average number of crystals formed was not obviously changed for experiments using patterned electrodes in comparison with the control using uniform electrodes, considering the overlapping error bars. More crystals were formed in MFDs using patterns 4 (ϕITO = 0.21) and 8 (ϕITO = 0.27) compared to that using pattern 2 (ϕITO = 0.41), which was contrary to the case without the electric field. An electric field of 2 Vpp and 1 MHz was found to enhance protein crystallization clearly for patterns 4 (ϕITO = 0.21) and 8 (ϕITO = 0.27), not only by resulting in more wells

is the patterned electrode with the highest surface fraction of ITO (i.e., pattern 1 with ϕITO = 0.51), which showed a statistically relevant increase in nucleation rate compared to some of the patterns with a smaller surface fraction of ITO (i.e., patterns with ϕITO = 0.15, ϕITO = 0.21, and ϕITO = 0.27). The results discussed above suggested that ITO acts as a template for nucleation. To further verify this hypothesis, the average fraction of crystals nucleated on the ITO surface was compared to the surface fraction of ITO for all the eight patterns (see Figure 3(c)). Here, the result for experiment using pattern 7 was shifted a bit to the right to avoid overlap with the result for experiment using pattern 5, and the solid vertical lines indicate the 90% Clopper−Pearson binomial confidence intervals (MATLAB R2015a, binofit, same in Figure 4(c)). In case no templating effect of ITO is present, one would expect such a confidence interval to cross the solid diagonal line in Figure 3(c). While the confidence interval did cross the diagonal line for some experiments (patterns 5 (ϕITO = 0.18), 4 (ϕITO = 0.21), and 2 (ϕITO = 0.41)), for several patterns the mean fraction of crystals nucleated on the ITO surface was clearly higher than the electrode ITO area ratio with some statistical significance as indicated by the confidence intervals that were either completely above or showed only a slight overlap with the diagonal line. Furthermore, none of the cases has an error bar completely below the diagonal line. The tendency of crystals to preferentially nucleate on the ITOcoated part of the bottom slide in the absence of an electric field suggests that ITO can act as a better template for nucleation during our experiments compared to glass. Lysozyme has an isoelectric point around 11 and is positively charged at a pH of 4.65. The surface charge of ITO was reported to be controlled by both the pH and the composition, e.g., the ratio of Sn/In.51 Therefore, the ITO-coated glass used in our experiments might be slightly negatively charged at the specific pH and composition so that lysozyme molecules tend to adsorb on the surface, leading to enhanced nucleation on ITO compared to nucleation on glass. Nucleation of Lysozyme on Patterned Electrodes in the Presence of Electric Fields. As described in Table 1, 3066

DOI: 10.1021/acs.cgd.6b01846 Cryst. Growth Des. 2017, 17, 3062−3070

Crystal Growth & Design

Article

with crystals in shorter induction time, as mentioned previously, but also by producing a larger number of crystals on average in each well. In addition, inhibited crystallization with longer induction time and a smaller mean number of crystals was observed for patterns 6 (ϕITO = 0.15), 2 (ϕITO = 0.41), and 1 (ϕITO = 0.51) when an electric field of 2 Vpp and 1 MHz was applied. Even though pattern 5 (ϕITO = 0.18) led to a much higher mean number of crystals compared to pattern 7 (ϕITO = 0.18) with the electric field, the shape of electrodes used did not influence the results significantly as the error bars overlapped. The discussion above suggested that the nonuniform electric field induced by the patterned electrodes can manipulate protein crystallization with enhanced or inhibited effect, depending on the specific design of an electrode. To further investigate the effect of an electric field on protein nucleation, we analyzed the nucleation location of crystals by comparing the average fraction of crystals nucleated on the ITO surface with an electric field of 2 Vpp and 1 MHz to the surface fraction of ITO for all of the eight patterns described in Table 1 (see Figure 4(c)). When applying an electric field, it is expected that ITO may still act as a template for nucleation in addition to any effect on nucleation of the electric field itself. Although for some patterns there appeared to be a tendency for crystals to nucleate on ITO, the general trend was weakened in the presence of an electric field. Figure 5 illustrates the microscope images taken for lysozyme crystals nucleated on the ITO surface (except (f) for pattern 6 where the crystal nucleated on glass) in MFDs with patterned electrodes subject to an electric field of 2 Vpp and 1 MHz. When applying the electric field, cases where no preference of the nucleation location was observed in the absence of an electric field (e.g., for patterns 5 (ϕITO = 0.18) and 2 (ϕITO = 0.41)) indicated a tendency for crystals to preferentially nucleate on the ITO surface in the presence of the electric field, whereas other cases where crystals preferentially nucleated on the ITO surface in the absence of the electric field (e.g., patterns 6 (ϕITO = 0.15), 3 (ϕITO = 0.28), and 1 (ϕITO = 0.51)) showed no preference of the nucleation location when applying the electric field. Furthermore, the tendency for crystals to nucleate on the ITO surface was enhanced by an electric field of 2 Vpp and 1 MHz for patterns 7 (ϕITO = 0.18) and 8 (ϕITO = 0.27) with some statistical significance as the confidence intervals were completely above the electrode ITO area ratio. In summary, although variation clearly exists, ITO seems to have a general templating effect leading to preferential nucleation of lysozyme on ITO, which can be enhanced or reduced by the electric fields for certain patterns. To further illustrate the effect of specific electrode designs, Table 2 summarizes the effect of an electric field of 2 Vpp and 1 MHz for various electrode designs on the induction time, crystal number, and nucleation location compared to the control cases without an electric field for lysozyme crystallization. Several trends can be observed: (i) for experiments using pattern 1 (ϕITO = 0.51) or 6 (ϕITO = 0.15), an electric field of 2 Vpp and 1 MHz clearly inhibited crystallization with longer induction time and fewer crystals as well as preference for nucleation on glass; (ii) for experiment using pattern 2 (ϕITO = 0.41), the same electric field also inhibited crystallization with longer induction time and fewer crystals but with preference for nucleation on ITO; (iii) for experiment using pattern 8 (ϕITO = 0.28), the electric field enhanced crystallization with shorter induction time, more crystals, and

Table 2. Influence of an Electric Field of 2 Vpp and 1 MHz on Lysozyme Crystallization for Various Electrode Designsa no.

pattern no.

ITO area ratio

average induction time

average number of crystals

preference to nucleate on ITO

1 2 3 4 5 6 7 8 9

1 2 3 4 5 6 7 8 uniform

0.51 0.41 0.28 0.21 0.18 0.15 0.18 0.27 1.00

↑ ↑ ↑ ↓ ↑ ↑ ↑ ↓ ↓

↓ ↓ ↔ ↑ ↔ ↓ ↔ ↑ ↔

↓ ↑ ↓ ↔ ↑ ↓ ↔ ↑ N/A

Here, ↑ refers to a statistically relevant increase; ↓ refers to a statistically relevant decrease; and ↔ indicates no statistically relevant effect.

a

preference for nucleation on ITO. Studies have showed that an external electric field can affect protein nucleation rate by varying the difference between the electrical permittivity of the solid and liquid phases.22 Thus, the inhibited or enhanced effect of an electric field on nucleation as summarized in Table 2 may be a result of the different change in the electrical permittivity for different patterned electrodes. Lysozyme is polarized due to its molecular composition and solvent−protein interactions so that it will migrate in response to an external electric field, which is a phenomenon known as dielectrophoresis (DEP).52 The DEP response can be positive with molecules attracted to a high electric-field region or negative where molecules are attracted to a low electric-field region. The electric potential map and electric-field map at the cross section of MFDs using pattern 6 are simulated with COMSOL Multiphysics (see Figure 6) as an example for the parallel-plate microfluidic devices used in this work and show that the lowest electric-field region is on the glass and the highest electric-field region at the edges of ITO, which is consistent with simulations reported in the literature for similar devices.43 As indicated by Table 2, MFDs with certain patterns such as 1, 3, and 6 showed a preference to nucleate on the glass surface compared to the control without an electric field, which implied a negative DEP response of lysozyme. MFDs with some other patterns such as 2, 5, and 8 showed a preference to nucleate on the ITO surface, and this does not indicate a positive DEP as crystals that nucleated on both the ITO surfaces and ITO edges were counted. In general, local features of the electric field, e.g., local minima or maxima, may lead to different results for different patterns. To investigate whether the electric-field properties can change this observed inhibition or enhancement, different amplitude voltage and frequency were applied to MFDs with patterned electrodes (patterns 1, 2, and 8) and uniform electrode (see Table 1, experiments 10−17). Similar to the control experiments without an electric field, needle-shaped crystals were also observed in MFDs with patterned electrodes with the applied electric fields in a minority of cases. As shown in Figure 7(a,b), for MFDs with uniform electrodes (ϕITO = 1.00) or pattern 2 (ϕITO = 0.41), the average induction time and average number of crystals varied with the electric-field properties. For MFDs using pattern 8 (ϕITO = 0.27), an electric field of 10 Vpp and 1 MHz led to the largest crystallization enhancement, while for MFDs using pattern 1 (ϕITO = 0.51) the absence of an electric field gave the least inhibition. Similar 3067

DOI: 10.1021/acs.cgd.6b01846 Cryst. Growth Des. 2017, 17, 3062−3070

Crystal Growth & Design

Article

Figure 6. Simulation of the electric field at the cross section of MFDs using pattern 6 with COMSOL Multiphysics. (a) Geometry and boundary conditions; (b) electric potential map and electric-field map; (c) zoom-in at the square region in (b).

response of lysozyme like many other proteins such as avidin and streptavidin.54,55 In conclusion, by shaping the potential energy landscape with the electrode pattern and electric-field property, controllable nucleation can in principle be achieved. Among all the tested cases presented above, the combination of MFD with pattern 8 and an electric field of 10 Vpp and 1 MHz led to the maximum enhancement of lysozyme crystallization. Insulin Crystallization. With an initial concentration of 5 or 10 mg/mL, insulin spherulites as well as crystals were observed at room temperature in MFDs made of glass. The spherulites were metastable and dissolved eventually during the growth of insulin crystals (see Figure 8(a,b)). The optimal experimental condition for enhancement of lysozyme crystallization was implemented for insulin crystallization (experiment 18, Table 1). For the control experiments without an electric field, no crystals were observed in any of the 12 wells during 3 days of observation. In contrast, when an electric field of 10 Vpp and 1 MHz was applied, an insulin crystal was observed in 1 out of the 12 wells at around 3 h after the start of the experiment, which was during the application of the electric field (see Figure 8(c,d)). The crystal was found to nucleate on the glass surface near the interface of the ITO and glass layers. Although the low nucleation rate prevents any rigorous analysis, the formation of an insulin crystal during the application of the electric field suggests some enhancement of a nonuniform electric field on insulin crystallization as well. Enhanced nucleation has been reported for insulin crystallization with a

results have been observed for batch crystallization in the presence of AC electric fields induced by uniform electrodes, where the rate of nucleation can be either increased or decreased by adjusting the electric-field properties.22 Moreover, our results indicated that applying an electric field led to consistent enhancement of nucleation rate (shorter induction time and more crystals) for MFDs using pattern 8 (ϕITO = 0.27) and consistent inhibition of nucleation rate for MFDs using pattern 1 (ϕITO = 0.51) regardless of the amplitude voltage and frequency. This implies that the nonuniform AC electric field can promote or inhibit protein crystallization based on the specifics of the electrode designs. Figure 7(c) shows the nucleation location of lysozyme in MFDs made with different electrode designs with different electric-field properties. Here, the results for experiments with electric fields (AC1, AC2, and AC3) were shifted slightly to the right to avoid overlapping. For patterns 8 (ϕITO = 0.27) and 2 (ϕITO = 0.41), the preference to nucleate on ITO or glass surface was found to depend on the frequency of the applied electric field, with lysozyme tending to nucleate on the ITO surface at a high frequency (1 MHz) and on the glass surface at a low frequency. For pattern 1 (φITO = 0.51), the results indicated a preference to nucleate on the glass surface (negative DEP) at a fixed frequency, and this agreed with theoretical studies that the DEP response of protein is frequency-dependent.52,53 Positive DEP of lysozyme was predicted to occur in the MHz to GHz range,53 thus a cross over frequency may exist for the DEP 3068

DOI: 10.1021/acs.cgd.6b01846 Cryst. Growth Des. 2017, 17, 3062−3070

Crystal Growth & Design

Article

Figure 8. Microscope images for insulin crystals and spherulites observed in experiments using MFDs made of glass with an initial concentration of 10 mg/mL at (a) 27 h and (b) 35 h (spherulites disappeared almost completely) and for insulin crystals observed in experiment 18 described in Table 1 at (c) 3 h (crystal inside the square) and (d) 72 h.

conditions showed a statistically significant effect, nucleation can be either enhanced or inhibited in terms of induction time or number of crystals formed depending on the specific electricfield properties and electrode designs compared to control experiments in the absence of an electric field. Close inspection of the location of the formed crystals on the patterned electrodes in the absence of an electric field suggests that ITO has a tendency to act as a better template for lysozyme nucleation compared to glass. In general, either an increased or decreased preference for nucleation on ITO could be observed when applying an electric field. For some cases, a preference to nucleate on glass (lowest electric-field region) was found, indicating a negative DEP response of lysozyme. Overall, among all the tested cases, the nonuniform electric field of 10 Vpp and 1 MHz induced by a specific pattern (8) led to the maximum enhancement of lysozyme crystallization, which could at least to some extent also be applied for the enhancement of the nucleation of insulin. Future work may include translation of the obtained results to other proteins in flow devices to use electric fields for improved control over protein crystallization to develop better continuous protein separation and purification processes.

Figure 7. (a) Number of wells that had crystals during different induction times, (b) average number of crystals in each well at 72 h, and (c) average fraction of crystals nucleated on the ITO surface as a function of the ITO area ratio for experiments in MFDs made with different electrode designs with an electric field of 2 Vpp and 1 MHz (AC1), 10 Vpp and 1 MHz (AC2), and 2 Vpp and 100 Hz (AC3) and in the absence of an electric field (NO).

uniform AC electric field,56 which shows consistency with our result from a patterned electrode. In general, the application of a nonuniform electric field induced by patterned electrodes in a parallel-plate configuration to promote crystallization for both lysozyme and insulin shows prospects for extending our approach to other proteins.



AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected].



ORCID

CONCLUSIONS Patterned electrodes in a parallel-plate configuration can be used to manipulate lysozyme nucleation in microfluidic devices. In general, although variation inevitably exists and not all tested

Richard Lakerveld: 0000-0001-7444-2678 Notes

The authors declare no competing financial interest. 3069

DOI: 10.1021/acs.cgd.6b01846 Cryst. Growth Des. 2017, 17, 3062−3070

Crystal Growth & Design



Article

(28) Taleb, M.; Didierjean, C.; Jelsch, C.; Mangeot, J. P.; Capelle, B.; Aubry, A. J. Cryst. Growth 1999, 200, 575−582. (29) Taleb, M.; Didierjean, C.; Jelsch, C.; Mangeot, J. P.; Aubry, A. J. Cryst. Growth 2001, 232, 250−255. (30) Giegé, R. FEBS J. 2013, 280, 6456−6497. (31) Flores-Hernández, E.; Stojanoff, V.; Arreguín-Espinosa, R.; Moreno, A.; Sánchez-Puig, N. J. Appl. Crystallogr. 2013, 46, 832−834. (32) Hammadi, Z.; Astier, J.-P.; Morin, R.; Veesler, S. Cryst. Growth Des. 2009, 9, 3346−3347. (33) Nanev, C. N.; Penkova, A. Colloids Surf., A 2002, 209, 139−145. (34) Boltsis, I.; Lagoumintzis, G.; Chatzileontiadou, D. S. M.; Giastas, P.; Tzartos, S. J.; Leonidas, D. D.; Poulas, K. Cryst. Growth Des. 2014, 14, 4347−4354. (35) Koizumi, H.; Fujiwara, K.; Uda, S. Cryst. Growth Des. 2010, 10, 2591−2595. (36) Koizumi, H.; Uda, S.; Fujiwara, K.; Tachibana, M.; Kojima, K.; Nozawa, J. J. Appl. Crystallogr. 2013, 46, 25−29. (37) Wakamatsu, T. Rev. Sci. Instrum. 2015, 86, 015112. (38) Moreno, A.; Sazaki, G. J. Cryst. Growth 2004, 264, 438−444. (39) Nieto-Mendoza, E.; Frontana-Uribe, B. A.; Sazaki, G.; Moreno, A. J. Cryst. Growth 2005, 275, e1437−e1446. (40) Konry, T.; Marks, R. S. Thin Solid Films 2005, 492, 313−321. (41) Tanamoto, R.; Shindo, Y.; Miki, N.; Matsumoto, Y.; Hotta, K.; Oka, K. J. Neurosci. Methods 2015, 253, 272−278. (42) Mauge-Lewis, K.; Gordon, B.; Syed, F.; Syed, S.; Bonyi, E.; Mohammed, M.; Toth, E. A.; Seifu, D.; Aslan, K. Nano Biomed. Eng. 2016, 8, 60−71. (43) Bhatt, K. H.; Grego, S.; Velev, O. D. Langmuir 2005, 21, 6603− 6612. (44) Xie, R.; Liu, X.-Y. Adv. Funct. Mater. 2008, 18, 802−809. (45) Melvin, E. M.; Moore, B. R.; Gilchrist, K. H.; Grego, S.; Velev, O. D. Biomicrofluidics 2011, 5, 034113. (46) Parambil, J. V.; Schaepertoens, M.; Williams, D. R.; Heng, J. Y. Y. Cryst. Growth Des. 2011, 11, 4353−4359. (47) Alderton, G.; Fevold, H. L. J. Biol. Chem. 1946, 164, 1−5. (48) Bhamidi, V.; Skrzypczak-Jankun, E.; Schall, C. A. J. Cryst. Growth 2001, 232, 77−85. (49) Grzesiak, A. L.; Matzger, A. J. Cryst. Growth Des. 2008, 8, 347− 350. (50) Yu, X.; Ulrich, J.; Wang, J. Cryst. Res. Technol. 2015, 50, 179− 187. (51) Liu, B.; Liu, J. Langmuir 2015, 31, 371−377. (52) Nakano, A.; Ros, A. Electrophoresis 2013, 34, 1085−1096. (53) Matyushov, D. V. J. Chem. Phys. 2012, 136, 085102. (54) Bakewell, D. J. G.; Hughes, M. P.; Milner, J. J.; Morgan, H. Ann. Int. IEEE Embs. 1998, 20, 1079−1082. (55) Liao, K.-T.; Chou, C.-F. J. Am. Chem. Soc. 2012, 134, 8742− 8745. (56) Koizumi, H.; Tomita, Y.; Uda, S.; Fujiwara, K.; Nozawa, J. J. Cryst. Growth 2012, 352, 155−157.

ACKNOWLEDGMENTS The authors thank Mr. Henry Chun Fai Yeung and Mr. Michael Hon Chiu Kwok from the Nanosystem Fabrication Facility of the Hong Kong University of Science and Technology for providing support in device fabrication. The work described in this paper was supported by a grant from the Research Grants Council of the Hong Kong Special Administrative Region, China (Project No. 16242916).



REFERENCES

(1) Rosa, P. A. J.; Ferreira, I. F.; Azevedo, A. M.; Aires-Barros, M. R. J. Chromatogr. A 2010, 1217, 2296−2305. (2) Przybycien, T. M.; Pujar, N. S.; Steele, L. M. Curr. Opin. Biotechnol. 2004, 15, 469−478. (3) Gottschalk, U. Biotechnol. Prog. 2008, 24, 496−503. (4) Horváth, Z.; Horosanskaia, E.; Lee, J. W.; Lorenz, H.; Gilmore, K.; Seeberger, P. H.; Seidel-Morgenstern, A. Org. Process Res. Dev. 2015, 19, 624−634. (5) McPherson, A.; Gavira, J. A. Acta Crystallogr., Sect. F: Struct. Biol. Commun. 2014, 70, 2−20. (6) Liang, M.; Jin, F.; Liu, R.; Yu, Y.; Su, R.; Wang, L.; Qi, W.; He, Z. Bioprocess Biosyst. Eng. 2013, 36, 91−99. (7) Judge, R. A.; Jacobs, R. S.; Frazier, T.; Snell, E. H.; Pusey, M. L. Biophys. J. 1999, 77, 1585−1593. (8) Liu, Y.; Zhang, X.-F.; Zhang, C.-Y.; Guo, Y.-Z.; Xie, S.-X.; Zhou, R.-B.; Cheng, Q.-D.; Yan, E.-K.; Liu, Y.-L.; Lu, X.-L.; Lu, Q.-Q.; Lu, H.M.; Ye, Y.-J.; Yin, D.-C. CrystEngComm 2015, 17, 5488−5495. (9) Sang-Il Kwon, J.; Nayhouse, M.; Christofides, P. D.; Orkoulas, G. AIChE J. 2013, 59, 2317−2327. (10) Liu, J. J.; Hu, Y. D.; Wang, X. Z. Comput. Chem. Eng. 2013, 57, 133−140. (11) Sleutel, M.; Van Driessche, A. E. S. Proc. Natl. Acad. Sci. U. S. A. 2014, 111, E546−E553. (12) Adawy, A.; van der Heijden, E. G. G.; Hekelaar, J.; van Enckevort, W. J. P.; de Grip, W. J.; Vlieg, E. Cryst. Growth Des. 2015, 15, 1150−1159. (13) Paxton, T. E.; Sambanis, A.; Rousseau, R. W. J. Cryst. Growth 1999, 198-199, 656−660. (14) Paxton, T. E.; Sambanis, A.; Rousseau, R. W. Langmuir 2001, 17, 3076−3079. (15) Liu, Y.-X.; Wang, X.-J.; Lu, J.; Ching, C.-B. J. Phys. Chem. B 2007, 111, 13971−13978. (16) Ryu, B. H.; Ulrich, J. Cryst. Growth Des. 2012, 12, 6126−6133. (17) Iefuji, N.; Murai, R.; Maruyama, M.; Takahashi, Y.; Sugiyama, S.; Adachi, H.; Matsumura, H.; Murakami, S.; Inoue, T.; Mori, Y.; Koga, Y.; Takano, K.; Kanaya, S. J. Cryst. Growth 2011, 318, 741−744. (18) Kakinouchi, K.; Adachi, H.; Matsumura, H.; Inoue, T.; Murakami, S.; Mori, Y.; Koga, Y.; Takano, K.; Kanaya, S. J. Cryst. Growth 2006, 292, 437−440. (19) Sazaki, G.; Yoshida, E.; Komatsu, H.; Nakada, T.; Miyashita, S.; Watanabe, K. J. Cryst. Growth 1997, 173, 231−234. (20) Hammadi, Z.; Astier, J.-P.; Morin, R.; Veesler, S. Cryst. Growth Des. 2007, 7, 1472−1475. (21) Gil-Alvaradejo, G.; Ruiz-Arellano, R. R.; Owen, C.; RodríguezRomero, A.; Rudiño Piñera, E.; Antwi, M. K.; Stojanoff, V.; Moreno, A. Cryst. Growth Des. 2011, 11, 3917−3922. (22) Koizumi, H.; Fujiwara, K.; Uda, S. Cryst. Growth Des. 2009, 9, 2420−2424. (23) Hou, D.; Chang, H.-C. Appl. Phys. Lett. 2008, 92, 223902. (24) Pan, W.; Xu, H.; Zhang, R.; Xu, J.; Tsukamoto, K.; Han, J.; Li, A. J. Cryst. Growth 2015, 428, 35−39. (25) Wakamatsu, T.; Toyoshima, S.; Shimizu, H. Appl. Phys. Lett. 2011, 99, 153701. (26) Sazaki, G.; Moreno, A.; Nakajima, K. J. Cryst. Growth 2004, 262, 499−502. (27) Hammadi, Z.; Veesler, S. Prog. Biophys. Mol. Biol. 2009, 101, 38−44. 3070

DOI: 10.1021/acs.cgd.6b01846 Cryst. Growth Des. 2017, 17, 3062−3070