Research Article www.acsami.org
High-Precision Dispensing of Nanoliter Biofluids on Glass Pedestal Arrays for Ultrasensitive Biomolecule Detection Xiaoxiao Chen,† Yang Liu,‡,§ QianFeng Xu,*,†,‡ Jing Zhu,‡ Sébastien F. Poget,‡,§ and Alan M. Lyons*,†,‡,§ †
ARL Designs LLC, 215 West 125th Street, New York, New York 10027, United States Department of Chemistry, College of Staten Island, City University of New York, 2800 Victory Boulevard, Staten Island, New York 10314, United States § Ph.D. Program in Chemistry, The Graduate Center, City University of New York, 365 Fifth Avenue, New York, New York 10314, United States ‡
S Supporting Information *
ABSTRACT: Precise dispensing of nanoliter droplets is necessary for the development of sensitive and accurate assays, especially when the availability of the source solution is limited. Conventional approaches are limited by imprecise positioning, large shear forces, surface tension effects, and high costs. To address the need for precise and economical dispensing of nanoliter volumes, we developed a new approach where the dispensed volume is dependent on the size and shape of defined surface features, thus freeing the dispensing process from pumps and fine-gauge needles requiring accurate positioning. The surface we fabricated, called a nanoliter droplet virtual well microplate (nVWP), achieves highprecision dispensing (better than ±0.5 nL or ±1.6% at 32 nL) of 20−40 nL droplets using a small source drop (3−10 μL) on isolated hydrophilic glass pedestals (500 μm on a side) bonded to arrays of polydimethylsiloxane conical posts. The sharp 90° edge of the glass pedestal pins the solid−liquid−vapor triple contact line (TCL), averting the wetting of the glass sidewalls while the fluid is prevented from receding from the edge. This edge creates a sufficiently large energy barrier such that microliter water droplets can be poised on the glass pedestals, exhibiting contact angles greater >150°. This approach relieves the stringent mechanical alignment tolerances required for conventional dispensing techniques, shifting the control of dispensed volume to the area circumscribed by the glass edge. The effects of glass surface chemistry and dispense velocity on droplet volume were studied using optical microscopy and high-speed video. Functionalization of the glass pedestal surface enabled the selective adsorption of specific peptides and proteins from synthetic and natural biomolecule mixtures, such as venom. We further demonstrate how the nVWP dispensing platform can be used for a variety of assays, including sensitive detection of proteins and peptides by fluorescence microscopy or MALDI-TOF. KEYWORDS: nanodroplets, microplate/microarray, chemical functionalization, ELISA, fluorescence, label-free MALDI-TOF
1. INTRODUCTION
dispensed volume decreases from microliters to picoliters. Alternatively, noncontact dispensing systems create a jet of nanoliter droplets and so relieve the tolerances imposed on positioning and substrate planarity but increase the cost and complexity of the delivery system used,14−21 as well as subjecting the solution to high temperatures and/or shear forces that can damage large molecules and cells. Dispensing errors associated with jetting systems remain high, approximately ±10% with 20 nL droplets.22 Another approach to dispensing is to control the size and wettability of the surface itself. Virtual microwells for highthroughput screening applications have been described where
Precise dispensing of a large number of nanoliter droplets containing bioreagents1−5 is one of the most crucial steps for achieving reliable assay results and is highly desired for highthroughput/content screening of new drugs or biomarkers.6 The need for precise dispensing of nanoliter (nL) quantities is especially acute when the source sample volume is limited, such as naturally occurring venom from snakes and spiders or other natural products.7 Conventional contact dispensing techniques do not offer sufficient precision for dispensing droplets less than 50 nL,8,9 and microwell plates are too large to handle such small volumes. Dispensing volumes below 50 nL is challenging because the dispensing process is dominated by interfacial adhesion3,10 and factors such as surface tension, capillary forces, and local microstructures that affect the transferred volume.11−13 As a result, error increases significantly as the © XXXX American Chemical Society
Received: February 28, 2016 Accepted: April 12, 2016
A
DOI: 10.1021/acsami.6b02487 ACS Appl. Mater. Interfaces XXXX, XXX, XXX−XXX
Research Article
ACS Applied Materials & Interfaces
Figure 1. Process schematic for the fabrication of nVWPs.
fluorescently labeled proteins deposited on the surface after droplet evaporation. The effects of glass surface chemistry and dispense parameters (e.g., lift-up velocity) on dispensed volume were systematically studied. We further demonstrate how the high-precision nVWP dispensing platform can be used for a variety of assays, including sensitive detection of proteins and peptides by both fluorescence microscopy as well as MALDI-TOF. In addition, the glass pedestal surface was functionalized to enable the selective adsorption of specific peptides/proteins from biomolecule mixtures. Biotin and KcsA ion channels were bound to PEGylated or nickel-chelate resin coated glass pedestals, and these surfaces were shown to selectively adsorb NeutrAvidin [from a mixture with bovine serum albumin (BSA)] or Tx7335 peptide from snake venom.
liquid was transferred to hydrophilic arrays patterned within a hydrophobic substrate.23 The liquid sample is constrained by the boundaries between the hydrophobic and the hydrophilic surface. To dispense fluids, either a special cover with matching arrays of features or a microdispenser as described above is required. Thus, although this surface aids in the formation of small droplets, dispensing still requires specialized materials and equipment to achieve high precision. To facilitate the dispensing of aqueous fluids by using differences in wetting, some researchers have dispersed hydrophilic regions on a superhydrophobic substrate.24,25 Salvinia molesta, a plant that floats on water, uses a combination of hydrophilic and superhydrophobic features to generate a high free energy barrier.26 Although these types of hydrophilic/superhydrophobic surfaces facilitate droplet positioning, they do not enhance control of the dispensed volume because the energy barrier between the two regions is too small; dispensing accuracy still relies on accurate dispensing tools. For the surface to influence the dispensed volume, the activation barrier between hydrophilic and hydrophobic regions must be sufficiently large that the solid−liquid−vapor triple contact line (TCL) is pinned. To achieve such a high energy barrier, a sharp edge at the boundary of the hydrophilic region is required. Superhydrophobic surfaces have been prepared with such high energy barriers including nanonail27 and microhoodoo28,29 structures. Inspired by these studies, we previously developed a superhydrophobic microumbrella surface made from hydrophilic polymer materials with re-entrant features.30 The high energy barrier pinned the TCL, causing a concave meniscus to form; on a hydrophobic surface of the same geometry, the water meniscus is convex. Thus, a combination of a hydrophilic surface with an abrupt boundary (i.e., sharp edge) results in a structure with a pinned TCL and a sufficiently high energy barrier to prevent wetting beyond the edge of the surface. Although these high energy barrier surfaces have been demonstrated on micrometer length scales, they are expensive to fabricate and are easily damaged due to their fragility. Thus, there is a need for a surface that is inexpensive to fabricate, mechanically robust, and of the appropriate size that can be used to precisely dispense arrays of nanoliter droplets. In this paper, we report a novel nanoliter droplet virtual well microplate (nVWP) for precisely dispensing nanodroplets on top of isolated glass pedestals. High precision (better than ±1.6%) was achieved through the combination of local surface chemistry and the geometry of the pedestals, i.e., the sharp edge created at the glass−air interface. Such a device relieves the mechanical alignment tolerances required for dispensing compared to conventional dispensing techniques and provides a significantly improved method to accurately dispense and manipulate nanoliter droplets from source sample volumes as small as 3 μL. The wetting behavior of nanodroplets dispensed on the nVWP and the position of the solid−liquid−vapor TCL interface were studied using optical microscopy during dispensing as well as by determining the position of
2. MATERIALS AND METHODS 2.1. Fabrication of the nVWP. The fabrication of virtual well microplates is shown schematically in Figure 1. An array of polydimethylsiloxane (PDMS) posts was printed onto a glass substrate using a robotic dispensing system (Janome 2203N and EFD Performus syringe dispenser) as previously described.31,32 Briefly, PDMS silicone resin with thixotropic properties (ELASTOSIL LR 3003 50A/50B, Wacker) was loaded and degassed in 10 cm3 syringes fitted with a 22 gauge tapered tip mounted to the robot. The robot was programmed to bring the tip to the first location at a controlled height of 300 μm above the glass microscope substrate. The robot triggers the syringe dispenser to deposit a controlled amount of PDMS (85 psi and 0.3 s) and then lifts vertically from the surface. This forms a single cylindrical cone of PDMS with a base diameter of 700 μm and a height of 1.2 mm. An array is created by repeating this procedure on a 1.2 mm pitch. After printing, the conical posts were planarized during cure (165 °C for 5 min) by contacting the tips with a flat, Teflon-coated plate, which was mounted on a stage that could move up and down in the Z direction. The height of the posts was controlled by a 1 mm spacer attached to the Teflon-coated plate. Since the printed silicone posts can flow easily upon contact with the Teflon plate, no additional pressure was required. After curing, the plate was easily released, exposing the tapered posts with flat tops. The diameter of the flat top was 150−200 μm, as shown schematically in Figure 1b. Glass pedestals measuring 500 × 500 × 100 μm (diced from glass coverslips by Valley Design Corp.) were mounted onto the flat tops using a roomtemperature vulcanizing hydroxyl-terminated dimethylsiloxane (DAP, Dow Corning) adhesive, as shown schematically in Figure 1c. The glass pedestals were first placed into a 3D printed alignment fixture with square well arrays. The adhesive-coated PDMS posts were aligned and brought into contact with these glass pieces under a microscope with the aid of an x,y,z stage. In this way, an array of glass pedestals can be bonded to the PDMS post array in one step. In the future, automatic pick-and-place tools can be used for assembling large volumes of arrays using a practical large-scale manufacturing technique. Tools similar to those used for LED and other highthroughput semiconductor manufacturing33,34 are readily available. 2.2. Glass Pedestal Preparation. Glass pedestals were dismounted from the carrier substrates by soaking in acetone solution for 1 h in an ultrasonic bath (Fisher Scientific Inc.) to remove the dicing tape. The dismounted glass pedestals were rinsed twice in DI water and dried. The glass pieces were then thoroughly cleaned using the following procedure: (1) washed in an aqueous solution of 4 wt % B
DOI: 10.1021/acsami.6b02487 ACS Appl. Mater. Interfaces XXXX, XXX, XXX−XXX
Research Article
ACS Applied Materials & Interfaces
Figure 2. Schematic for the chemical surface modification of glass pedestals: (a) biotinylated and PEGylated glass pedestals, (b) purely PEGylated glass pedestals, and (c) DMDCS-treated glass pedestals. hydrogen peroxide and 4 wt % ammonium hydroxide at 80 °C for 10 min, (2) rinsed with DI water three times, (3) washed with an aqueous solution consisting of 4 wt % hydrogen peroxide and 1.5 wt % HCl at 80 °C for 10 min, (4) rinsed with DI water three times, and (5) dried in an oven for 12 h at 60 °C. The glass pieces were then ready to assemble onto the PDMS post arrays, or the surfaces were further modified as described below. 2.3. Glass Pedestal Surface Modification. 2.3.1. Grafting Biotin and PEG onto Glass Pedestals. Biotin and PEG were grafted onto the surface of glass pedestals using a solution immersion method,35 as shown schematically in Figure 2a,b. Briefly, silane−PEG−biotin (MW = 600, Nanocs) was first grafted onto clean glass pedestals by incubating in 20 mg of the reagent dissolved in 1 mL of 95:5 w/w ethanol/water for 1 h at room temperature. The modified glass pedestals were rinsed three times with DI water and dried. To block any exposed/unreacted glass, neat silane−PEG [CH3(CH2CH2O)n(CH2)3Si(OCH3)3, n = 6−9, Gelest] was then grafted onto the biotinylated glass pedestals at room temperature for 30 min, rinsed with DI water, and dried in air. Treatment of glass surfaces with neat silane−PEG was shown to effectively block nonspecific protein adsorption. 2.3.2. Hydrophobization of Glass Pedestals. A chemical vapor deposition (CVD) method36,37 was used to hydrophobize the surface,
as shown schematically in Figure 2c. An excess amount (1 mL) of dimethyldichlorosilane (DMDCS, Sigma-Aldrich) was added into a bottle and put in a jar containing cleaned glass pedestals. The jar was sealed, evacuated under vacuum for 2 min, and heated at 60 °C for 15 min to accelerate the vaporization of the DMDCS. The jar was then stored at room temperature overnight. The glass pedestals were cleaned by sonication in toluene for 5 min to remove any unreacted silane deposited on the surfaces and dried. Static water contact angle measurements, before and after treatment, were used to verify the surface modification. 2.3.3. Nickel-Chelate-Resin-Coated Glass Pedestals. Nickel-chelate resin is well-known to selectively bind biomolecules labeled with 6-his tags.38,39 Nickel-chelate-resin-coated glass, 180 μm thick, was purchased from Xenopore, scribed, and broken into small squares measuring approximately 500 μm on a side. 2.4. Dispensing Nanoliter Droplets onto the nVWP. 2.4.1. Manual Dispensing of Microliter Droplets. Large droplets with volumes between 0.5 and 2 μL were deposited onto the 500 × 500 μm glass pedestals using a hand-held Eppendorf pipet. The droplet was first formed at the tip of the adjustable pipet, and the bottom of the droplet was brought into contact with the top of the glass pedestal. The pipet was manually lifted up, transferring the droplet to the glass pedestal. C
DOI: 10.1021/acsami.6b02487 ACS Appl. Mater. Interfaces XXXX, XXX, XXX−XXX
Research Article
ACS Applied Materials & Interfaces
Figure 3. Schematic of the automated dispensing process. 2.4.2. Manual Dispensing of Nanoliter Droplets. A drop of 10 μL source solution was loaded onto the tip of a polystyrene rod, using an Eppendorf pipet. The droplet was brought into contact with the glass pedestals and lifted-up by hand to transfer nanodroplets onto the nVWP surface. 2.4.3. Automated Dispensing of Nanoliter Droplets. Automated dispensing was conducted using a robot (Janome-2203N) with a positioning accuracy of ±10 μm, as shown in Figure S1 of the Supporting Information (SI). A 2.5 mm diameter polymer rod was mounted on the robotic arm. A fixed volume (3 μL) of the source solution was placed onto the flat bottom of the rod using an Eppendorf pipet, as shown in Figure S1a (SI). The robotic dispensing process consists of three steps as illustrated in Figure 3: the rod with source solution was aligned with a glass pedestal; the rod was then moved downward at a preset speed (2 mm/s), enabling the source solution to contact the glass pedestal at a fixed height such that the bottom of the droplet was brought 200 μm below the top surface of the glass pedestal; last, the rod was lifted-off from the surface at a predetermined speed (15 mm/s unless otherwise specified) and translated to the next dispensing location. The dispensing process was monitored using a high-speed camera (Vision Research, Phantom V7.3) with a Mitutoyo 5X Plan APO objective coupled to a Tokina AT-X 100 mm f/2.8 macro lens operated at 5000 frames/s. The images/videos were analyzed with PCC 2.5 software. 2.4.4. Measurement of Dispensed Droplet Volume. An initial volume (Vi) of 3 μL of DI water was placed on the flat tip of the cylindrical dispensing rod (2.50 mm diameter) using an Eppendorf pipet. The volume of this spherical cap-shaped drop was confirmed by calculating the volume using the equation
V = πh(3a 2 + h2)/6
To assess the precision of Vd, the height of nanoliter droplets dispensed on pedestals was measured using a Centrimax long distance microscope with MX-5 lens and Pixilink PL-B681C USB camera with a resolution of 2 μm/pixel on 10 sample droplets. The standard deviation of height values was converted to a volume standard deviation using eq 1. 2.5. Detection Limit and Selectivity of Fluorescently Labeled Proteins Deposited on nVWP Surfaces from Nanoliter Droplets. 2.5.1. Detection of Fluorescently Labeled Protein Using nVWP. NeutrAvidin protein from Thermo Scientific was labeled using a Dylight antibody labeling kit following the procedure obtained from the manufacturer. The labeled NeutrAvidin−Dylight 488 had a concentration of 2 mg/mL in PBS buffer solution, which was used as a stock solution from which further dilutions were made. Droplets with a volume of 32 nL and with specific concentrations (0.1, 0.5, 1, 5, 10, 20, and 50 μg/mL) were automatically dispensed (i.e., using the robot) on the glass pedestals of the nVWP and allowed to evaporate. The fluorescence intensity was then measured using a Zeiss Cell Observer microscope using the same light source and exposure time (3 s) to enable comparisons between droplets and concentrations. Four glass pedestals/droplet samples were measured for each concentration. Confocal microscopy (Leica TCS SP8) with high resolution (400 nm/scan) under a 10× lens was used to detect protein absorbed onto the surfaces and edges of the glass pedestals. 2.5.2. Selective Adsorption of NeutrAvidin from a Mixed Protein Solution. A series of specific concentrations of NeutrAvidin−Dylight 488 (0.01, 0.1, 0.5, 1, and 5 μg/mL) in PBS buffer solution were prepared for the biotin−NeutrAvidin binding assay study. Fluoresceinconjugated BSA (BSA−Alexa Fluor 555, from Invitrogen) was dissolved to prepare a 2 mg/mL stock solution in PBS buffer and stored in the dark at 4 °C. To study the selective binding of NeutrAvidin, the stock solution of BSA−Alexa Fluor 555 was diluted to 1 μg/mL using PBS buffer and mixed with different concentrations of labeled NeutrAvidin. Two samples for each NeutrAvidin concentration were prepared and analyzed. Before imaging, the nVWP was rinsed with PBS buffer twice to remove unbound protein. The fluorescence intensity of glass pedestals was imaged using a Zeiss Cell Observer microscope (10× lens and 488 excitation laser) with NeutrAvidin−Dylight 488 (λex(max) = 494 nm, λem(max) = 518 nm). 2.6. MALDI-TOF Detection of Proteins on nVWP Surfaces Deposited from Nanoliter Droplets. An nVWP was directly fabricated onto a stainless steel MALDI plate (MSP 96 target, Bruker Daltonics). The MALDI plate was machined to create a recess 1.0 mm deep onto which the PDMS posts were printed. In this way, the top surface of the glass pedestals was at the same height as the MALDI plate surface. To deposit protein onto the glass pedestals, an ultrathinlayer sample preparation technique was used.40 The glass pedestals were first coated with 0.5 μL of a matrix solution (α-cyano-4hydroxycinnamic acid (4-HCCA) saturated in 1:2:1 0.1% trifluoroacetic acid:acetonitrile:water) diluted 1:4 into 2-propanol and dried. Droplets of a mixture of NeutrAvidin and a second matrix solution (4HCCA saturated in 3:1:2 formic acid:water:2-propanol) in a 1:10 ratio were dispensed onto the coated glass and dried. MALDI-TOF spectra were recorded on a Bruker Microflex MALDI-TOF instrument.
(1)
where h is the height of the drop and a is the radius of the water drop base. Values of a and h were measured from the optical image profile of the drop acquired using the high-speed-camera system with a resolution of 7 μm/pixel. The average volume of droplets dispensed on the nVWP surface was determined at five different lift-up velocities (0.5, 5, 15, 50, and 100 mm/s). For each velocity, 9−11 drops were dispensed onto different pedestals on an nVWP surface. After dispensing, the volume remaining on the dispensing rod (Vf) was determined by measuring the source drop width and height and calculating the volume of the spherical cap as described above. To account for evaporation, the volume of an initial 3 μL drop was calculated after the time required for dispensing (Vfe) using measured values of the height and radius, and the volume lost to evaporation (Ve) was determined by subtraction (Vi − Vfe). The average dispensed volume (Vd) was determined by Vd = (Vi − Vf − Ve)/n, where n is the total number of dispensed droplets. Plotting Vd vs h reveals a linear relationship (Figure S3, SI) that fits the equation
Vd = 173.4h − 8.4
(2)
with R2 = 0.979. This linear relationship was used to determine droplet volumes from height measurements of manually dispensed nanoliter droplets. D
DOI: 10.1021/acsami.6b02487 ACS Appl. Mater. Interfaces XXXX, XXX, XXX−XXX
Research Article
ACS Applied Materials & Interfaces 2.7. Detection of Selectively Adsorbed Peptides on NickelChelate-Treated Glass nVWP Surfaces. Nickel-chelate-resincoated glass pedestals (500 × 500 × 180 μm) were used to anchor KcsA ion channel proteins to selectively adsorb Tx7335 peptide from snake venom. The KcsA ion channel, modified to contain a 6-his tag, was prepared in buffer containing 50 mm Tris pH 7.5, 150 mM KCl, and 10 mM of the surfactant n-decyl β-D-maltopyranoside (DM) to form ion channel micelles. The 6-his-tagged KcsA micelles were anchored onto the Ni-chelate-treated glass pedestals by depositing 1 μL droplets of the KcsA solution onto the pedestals, incubating for 5 min, and removing the droplet with vacuum. After anchoring, excess ion channels were washed away by rinsing two times with DM buffer. Crude lyophilized venom from the Eastern green mamba snake (Dendroaspis angusticeps) was redissolved in the same buffer at a concentration of 2 mg/mL and predepleted of most nonspecifically binding toxins by passing it over a Ni2+ affinity column (GE Life Sciences Ni Sepharose 6). A 1 μL droplet of prepared venom solution was deposited onto the nickel-chelate-coated glass pedestal of an nVWP and incubated for 2 min. Excess venom was washed away by rinsing with DM buffer for 6 times. A 1 μL droplet of matrix solution (4-HCCA saturated in 3:1:2 formic acid:water:2-propanol) was then added onto the surface and allowed to dry for MALDI-TOF detection.
the material costs (e.g., glass slide, PDMS, glass pedestals) are inherently low. However, the technology for placing small integrated circuits is well established and can be leveraged to assemble nVWPs at low cost. The technology to pick-and-place small, 500 μm, die for electronic components has been used commercially for decades.34 Recently, a detailed cost assessment for packaging small, thin die has shown35 that the packaging cost for an active die 500 μm wide is
