A droplet-based novel approach for viable and low volume consumption surface plasmon resonance bio-sensing inside a polydimethylsiloxane microchip

Abstract

Over the course of last two decades, surface plasmon resonance (SPR) has emerged as a viable candidate for label-free detection and characterization for a large pool of biological interactions, ranging from hybridization of oligonucleotides to high throughput drug-screening. Conventional SPR bio-sensing involves a step-response method where the SPR sensorgram in response to a switched sequential flow of analyte and buffer is plotted in real-time and fitted to an exponential curve to extract the associative and dissociative reaction rates. Such measurement schemes involve continuous flow conditions where a substantial reagent volume is consumed and is subject to dispersive mixing at flow switching zones. In this paper, we demonstrate a new plug-train SPR technique in a microfluidic chip that separates and singulates solvent plugs in analyte and buffer by an immiscible air phase. Bio-samples are first discretized within plug droplets with volumes in order of few hundred nanoliters or less followed by pressure-driven transport onto SPR sensing sites of this hydrophobically modified SPR microdevise. The kinetic constants k_a and k_d for a model protein-small molecule interaction pair are extracted from a plug-train signal and are shown to be in reasonable agreement with our previous reports.

INTRODUCTION

Kinetic characterization of macromolecular interactions is important for quantifying biological events, such as cell adhesion, viral infection, rational drug design, and drug discovery in real-time. Optical detection method based on surface plasmon resonance (SPR) monitors the biochemical reaction in real-time and extracts information on the rate and equilibrium binding constants for such macromolecular interactions. Over the years, SPR based biomolecular interaction analysis (BIA) has gained considerable momentum in pharmaceutical industry where it is used as the secondary drug screening tool. In the instance of step-response method (SRM)^{1, 2, 3, 4} and other unconventional schemes^{5, 6, 7} of SPR detection and characterization, bio-sample volumes ranging from micro-liters (μl) to few milliliters (ml) are consumed in one experimental run, increasing the detection time and cost. If the reagent switching technique can be manipulated to transfer optimal volume to the sensing zone, the volume requirements of the experiment can be reduced manifold.

Recently, it was demonstrated^{8, 9, 10} that by manipulation of the sample introduction method, singulated plugs of a drug solution separated by an immiscible oil phase can be synthesized and transferred to surface immobilized cells for bio-reaction. This suggests that a similar technique can be used to separate plugs while delivering different compounds to a reacting SPR surface. However, while cells might be compatible with oil, the surface chemistry in case of SPR application is of delicate nature and might not be oil compatible. One way of circumventing this problem is the use of air as the inert phase for plug separation. In this paper, we report a droplet based scheme where short discrete plugs of reacting samples confined by air are used for SPR characterization. The viability and accuracy of this plug-train SPR scheme will be evaluated in discussion section.

PRINCIPLES AND SCHEMATICS

Fig. 1 shows the basic principle of plug-train technique. Unlike other SPR schemes, solution samples are first loaded as discrete plugs separated by an immiscible gas phase. When these plugs are transported by pressure to the Au sensing functionalized spots, brief association, and dissociation reactions occur for analyte and buffer droplets, respectively. Droplet volumes can be manipulated to have a wide range (50 – 500 nl). A plug-train similar to the alternating plug system of Fig. 1 is synthesized by opening the microfluidic valve that controls the flow of solvents from the reservoir source for a certain pre-chosen time period while keeping all other pressure-driven valves closed. This is followed by brief loading of air plug from its source in a similar fashion. A series of such alternating solvent plugs are synthesized and transported through the fluid channels. As we shall see, such synthesis methods can be realized only with two-level microfluidic chip architecture.

Figure 1.

Basic principle of plug-train SPR microchip scheme. The chip consists of two channel reservoirs where analyte and buffer are separated into plugs by an immiscible inert phase (air). Plugs are first loaded in the input reservoir followed by transport in the forward phase, where association and dissociation cycles are measured with all plugs ending at the output reservoir.

Similar to the oil-based singulation schemes^{8, 9, 10} and unlike other SPR schemes, the use of air as the inert separation phase requires additional considerations. Alternate plugs of the reagents that are synthesized and transported require being disconnected from each other to avoid reagent mixing by dispersion¹¹ especially along the fluidic channel walls. Findings reveal that a plug system with alternate air bubbles or short air plugs retains a thin layer of solvent on channel wall surfaces when hydrophillic.¹² Thus, our microdevise requires surface modification with a hydrophobic coat, preferred if permanent. Besides, discretization is also crucial from the standpoint of being able to control the volume of plug that is eventually delivered to the SPR sensor surface. Since its introduction in 1938, polytetrafluroethylene (PTFE), commonly known as Teflon, has been used for a wide variety of industrial coatings, chemical,¹³ and biological applications.^{14, 15} Owing to its hydrophobicity and chemical inertness, we will use an amorphous fluoropolymer (Teflon AF) resin as our devise coating material for permanent modification.

Fig. 2 shows the schematic of a plug-train SPR Microchip implementing a dual (sense + reference) chamber arrangement. Each of its chambers has four pressure driven load sources connected to buffer, analyte, regeneration, and a heterogeneous separation phase (air), respectively, and two outputs, all connected to flow channels. Flow of each of these solution sources is controlled by a set of pressure controlled microvalves. The chip microchannels in the sensing zone are lined with functionalized sensing and reference spots that record SPR sensorgrams. All driving pressure sources for the chip are regulated by external pressure regulators connected to these sources.

Figure 2.

Schematics of plug-based SPR chip. Valves O, B1 and B2, and A regulate air, buffer, and analyte plugs, respectively, while valves W1 and W2 control the transport of the plug train system to the SPR sensing channel via input storage channel.

PLUG-TRAIN SPR MICROCHIP

Fig. 3 shows a photograph of a plug-train SPR microchip. This microchip is fabricated using a modified two-level polydimethylsiloxane (PDMS) technique.^{16, 17} The two-level PDMS devise is sealed with an SF10 glass slide bearing patterned SPR gold spots using fabrication procedures similar to our previous reports.^{5, 6, 7} Briefly, molds of the two microfluidic layers are made photolithographically on silicon wafers (4 in. diameter) using photoresists AZ9260 and SU-8, respectively. PDMS is then cured on these molds, peeled, aligned to each other after an oxygen plasma exposure (100 mTorr pressure and 40 W of RF power for 20 s) using a home grown aligner, and then bonded immediately. SF10 glass substrates are cleaned with piranha solution (3:1, H₂SO₄:H₂O₂) and transferred to a TM Vacuum Sputtering instrument for deposition of a 3 nm adhesion layer of Ti/W followed by a 40 nm layer of gold. The metal is then patterned with ∼2.2 μm thick photoresist of S1813 (spin speed of 3000 rpm for 40 s) coat. The patterned Au spots have dimensions of 200 × 200 μm² and thickness ∼ 43 ± 2 nm. The patterned glass substrate is then bonded to the two-level PDMS devise using similar oxygen plasma step. Microchannels from AZ9260 molds form the upper fluid loading layer (250 μm wide and 20 μm high) and those from SU-8 resist form the lower valve layer (250 μm wide and 22 μm high) and also the flow layer. Small openings are made in areas overlapping the load and flow channels to connect them.

Figure 3.

Photograph of fabricated SPR microchip. The chip measures 1.8 by 2.2 cm.

MATERIALS AND METHODS

Reagents

Polyethylene Glycol (PEG) compounds were purchased from Laysan Bio. These include 5 kDa carboxymethyl-PEG-thiol (cm-PEG) and 2 kDa methoxy-PEG-thiol (m-PEG). N-ethyl-N′-(3-dimethylaminopropyl) carbodiimide (EDC), Sulfo-N-hydroxysuccinimide (S-NHS), and Phosphate buffered saline (PBS) tablets were purchased from Thermo Scientific. Carbonic anhydrase II (CA-II, MW ∼29 kDa), 10% Sodium dodecyl sulfate (SDS) and 4-(2-aminoethyl) benzenesulfonamide (ABS) were all purchased from Sigma-Aldrich. Sylgard 184 PDMS kit was purchased from Dow Chemicals. De-ionized 18 MΩ water (DI water), Acetone, Sulfuric acid (H₂SO₄), and Hydrogen peroxide solution (H₂O₂) and all photoresists and developers were provided by University of Utah Microfab facility. SF10 Schott glass substrates (2 × 2 in.²) were custom ordered from Schott. Teflon® AF amorphous fluoroplastic resin in solution (6% of T_g 160 °C solid resin in 3M Company FC-40 solvent) and Fluorinert (FC-3283) solvent were purchased from DuPont^TM and 3M Fluorinert^TM, respectively.

Hydrophobic modification with Teflon AF

In the plug-train SPR chip, bare gold spots are preserved from contamination by Teflon resin and its solvents prior to the formation of a hydrophilic PEG brush on the sensory gold surface. This modification step with PEG is necessary for surface functionalization and also for avoiding spurious non-specific signals from the gold surface during bio-sensing.^{4, 18} Since this sensory functionalization is carried out only after Teflon modification steps inside the microchip, the coating process with Teflon AF needs to incorporate modification steps that protect the gold surface during the coating and then de-protect it when the coating is complete. The photoresist coat of S1813 is left deliberately on patterned gold for this purpose. It must be noted that the plasma etch rate of S1813 is of the order of few hundred nanometers per minute and hence is never completely removed from gold without acetone rinsing step. Following microchip assembly, the microchannel walls are first coated with Teflon that renders the channel walls permanently hydrophobic (wetting angle ∼104°). A 0.2% Teflon-AF solution by volume dissolved in FC-3283 solvent is first prepared and flown inside the microchannel for 1 min followed by removal of excess fluorinating solvent using pressure of 20 psi. The chip is then baked at 120 °C for 12 h to complete the hydrophobic coating. The flow process of Fig. 4 shows the Teflon coating steps of PDMS microchip. The Au protective resist is next removed by acetone rinsing to expose bare gold, followed by DI water and this is repeated three times. The gold spots lined along microchannels are now ready for the SPR bio-sensing functionalization step.

Figure 4.

Fabrication process for coating channel walls of PDMS SPR microchip with Teflon.

SPR gold functionalization

The bare gold surface is modified with PEGs similar to our previous reports. We will employ a well established interaction pair of CA-II-ABS as our study model for characterizing protein-small molecule binding.¹⁹ Briefly, gold spots inside the microchannels are first rinsed with 10 mM HCL solution followed by PEGylation in PBS buffer (50 mM phosphate, 1M NaCl, pH 7.4) using cm-PEG. A short underbrushed layer of m-PEG further minimizes the non-specific adsorption.¹⁸ Excess PEG is removed by a short rinse of 50 mM NaOH. This is followed by immobilization of ABS ligand using standard amine-coupling procedure to form a ligand presenting sensing surface. A solution of Sulfo-NHS (100 mM) and EDC (400 mM) is used to facilitate crosslinking of activated cm-PEG to ABS. The sensing surface captures ABS (50 mM) while control surface is blocked with ethanolamine (50 mM) in PBS buffer (25 mM Phosphate, pH 8.4 with 0.01% SDS) for 20 min. This is followed by a 15 min PBS (5 mM phosphate, pH 7.4) rinse prior to experiments.

Experiments

The SPR chip of Fig. 3 is mounted onto a sample holding cell of SPR instrument and coupled to the prism using refractive index matching fluid. The sensorgram data are acquired using a manufacturer modified GWC Technologies SPRimager2 that accommodates our chip. Additional modifications are reported elsewhere.⁵ Post microfluidic connections to flow channels and valves, the valve control lines are filled with water to displace the air in the valves due to gas permeable properties of PDMS. Phosphate-buffered saline (PBS, 5 mM sodium phosphate, pH 7.4) is used as running buffer. The valves that regulate all the air, analyte (CA II), and buffer (PBS) loading channels are pressurized by 30 psi. The CA II samples were prepared using the previously used procedures. First, dry CA II powder was dissolved in the running buffer solution. Next, we measure the actual concentration in solution using a Spectrophotometer at a wavelength of 280 nm. The protein concentration (C) is then given by the relation C (mole/L) = A₂₈₀/(a₂₈₀·b), where A₂₈₀ is the value of absorbance at 280 nm and b being optical path length (1 cm for the given photometer). Wasabi camera control software integrated with the Hamamatsu EMCCD camera was used for analyzing the collected data. The mean intensity I(t) of the sensing and reference gold spots (900 pixels) for all the frames were extracted using this software followed by its analysis in MATLAB. The channels were flushed with regeneration solution (0.1% SDS in running buffer) between the runs.

RESULTS

Viability of plug-train method

In this chip, we first tested the feasibility of the hypothesized plug-train measurement method. Fluorescent solvent plugs with alternating air plugs were first synthesized in the chip (without SPR gold spots) which was then transported to the actual SPR sensing zone and their fluorescent intensity signal is recorded in real-time. Fig. 5 shows an example plug train where discrete solvent plugs of fluorescent solution (Fluorescein in PBS buffer) with alternating air plugs are transported through microchannels by a driving pressure source. This plug system is then plotted in real-time as illustrated by the fluorescent intensity plot of Fig. 6. Droplets have an approximate volume of ∼110 nl and plug duration of 2 s at a source pressure of 10 psi (for both solvent and air plugs). While solvent plugs of duration 1 s have an observed volume of 52 ± 3 nl, shorter plugs have reduced reproducibility and higher standard deviation. The upper limit of plug volume is, however, determined by the maximum volume permissible in the input storage channels of this fabricated devise. The total volume of the fluidic channel system in this storage chamber is about 730 nl. Generally speaking, optimal parameters for the devise of given dimensions and measurement technique is an input plug ranging from 400 to 600 nl at flow rate of 10 psi or higher. The air plugs can range from 2 to 4 s depending on the number of analytes and total input chamber volume.

Figure 5.

Snapshots of alternating solvent plugs in green (P₁₋₄ in sequence from point of injection) and air plugs (dark regions between solvent plugs) transported inside the chip (flow direction from left to right in input channel via sensing zone to the output channel downwards). The images are obtained from a fluorescent microscope.

Figure 6.

Recorded fluorescence intensity plot from a spot of 900 pixels in the sensing zone of the chip. Plugs of two different concentrations (C₀ and 2C₀) of fluorescein alternated by air plugs are flown at constant pressure of 10 psi and average flow velocity V.

Kinetic characterization by SPR

Next, we tested the characterization of the biochemical reaction using functionalized sensor and passivated reference gold spots in a SPR chip. Analyte concentrations of 1.2, 3.1, and 4.6 μM were used for the analyte plugs in separate plug-train experiments. PBS buffer was chosen for the buffer plugs. Fig. 7 (right) shows a time-domain SPR sensorgram obtained from the sense and reference gold spots for a flow of alternating analyte and buffer plugs (solvent plugs have a volume of ∼480 nl) separated by an air plug at a flow pressure of 15 psi. The kinetic constants k_a and k_d for this analyte-ligand pair were obtained by using the conventional exponential curve fitting procedure of SRM. In all the data sets, there exists fixed imager intensity I_offset for all the imaging gold spots which is subtracted from the measured signal to bring the baseline of the signal to zero. The signal from the reference spots (I_ref) is then subtracted from that of the sensor spots (I_sensor) to obtain referenced sensor signals for further processing. The reference-sensor spot pairs are approximately at the same distance downstream from both the modulators and subtraction of their signals ensure elimination of the signal component from bulk response of analyte as shown in the left of Fig. 7. The corrected time-domain responses from the experiments then follow the simple relation I(t) = I₀·s(t) where I₀ is the proportionality factor, s(t) being the test signal in analyte. The measured values of k_a and k_d are summarized in Table TABLE I. which shows a comparison of the kinetic constants obtained by the reported SPR characterization techniques. While these estimation values for the same protein-small molecule model vary for different sensing surfaces, we compare the kinetic constants from different techniques for same protocol of sensor functionalization.

Figure 7.

(Left) Image of the sensing zone in activated surface plasmon mode. The two separate channels are dedicated to sensing and control spots which are placed in proximity of each other.⁵, ⁶ (Right) Time-domain plug-based SPR sensorgram extracted from the analysis of sequence of images like one to the left. Two cycles of alternating analyte and buffer plugs (8s) with an air plug in between (3 s) are synthesized and transported. Corresponding association and dissociation for the CAII-ABS biochemical interaction system are recorded from a gold spot for analyte concentration of 3.1 μM. The final referenced data set has been baselined to 300 for the figure.

TABLE I.

Comparison of extracted rate constants.

SPR characterization method	Analyte concentration [A] (μM)	Kinetic association rate ka (M−1 s−1)	Kinetic dissociation rate kd (s−1)
Conventional SRM a	1.63	4.75 × 103	4.81 × 10−2
	3.40	8.68 × 103	4.76 × 10−2
Dual-slope a	[A; ARE] = [1.63; 54.40]	3.66 × 103	4.83 × 10−2
	[A; ARE] = [3.40; 56.67]	3.60 × 103	4.74 × 10−2
Multisine b	5.2
	10.4	9.5 × 103	6.2 × 10−2
	15.6	(Average value)	(Average value)
Plug-train	1.2	8.33 × 103	4.58 × 10−2
	3.1	7.83 × 103	4.13 × 10−2
	4.6	9.08 × 103	5.36 × 10−2

^aReference ⁶.

^bReference ⁷.

DISCUSSION

While one can argue that PDMS is hydrophobic and might not require any further modification, the use of PEG reagents in SPR functionalization steps have been observed to reduce the wetting angle. This is probably due to the physical adsorption of hydrophilic thiol-terminated PEGs on PDMS surface. In the case of chips without Teflon modification, the air plugs sandwiched between two solvent plugs can diminish in volume as the combined plug system is driven by pressure through the channels to the point where the two solvent plugs can coalesce and move together as a combined plug. This is primarily due to gas permeable properties of PDMS. As Figs. 67 indicate, however, notable or inconsistent volume reduction of air plugs do not occur throughout the transport process in the modified devices. Since air is hydrophobic, it has a stronger affinity for more hydrophobic surfaces and hence this Teflon coat on PDMS might therefore aid in withholding discretization requirements while transporting plugs.

Fig. 7 shows another interesting observation. Since air (R.I = 1) has much less R.I than buffer (R.I = 1.33), the SPR signal intensity for air plugs must be less than that of buffer and correspond to the base line value instead of the peak value in the real-time sensorgram.⁴ As observed, a quantifiable SPR value within the dynamic range of our EMCCD capture devise leads us to believe that due to unconventionally high R.I change from buffer to air and vice versa, the SPR resonance conditions undergo a reduction of angle of resonance (θ_r) in a manner that the SPR signal around buffer conditions (on the left of initial resonance dip curve at a fixed θ_r of our SPR imaging system) changes for air plugs and now corresponds to the intensity from the right side of the shifted resonance curve.

Finally, in order to validate the feasibility of such a singulated plug approach we need to account for the effect it has on the observed kinetics of the bio-reaction. Although continuous flow conditions with a flow velocity (V of Fig. 7) higher than a threshold value^{5, 19} ensures that the reaction involving the macromolecule is not diffusion limited, a switched flow condition incorporated with air plugs might affect this mass transport and alter the reaction kinetics. As mentioned earlier, Jennissen and Zumbrink¹² demonstrated that for a flow of air bubbles alternating with solvent plugs, a stationary or slow-moving nanofilm of liquid (∼200 nm thickness from the wall boundary) is retained on a hydrophilic sensing surface. This liquid film is metastable and the analyte replenishment occurs through “vortex flow” mechanism upholding the exponential kinetics. Such flow mechanism rather improves the mass transport rate by drastically reducing the Nernst diffusion layer thickness and eliminating the mass transport limitations of the bio-reaction.¹² Since our sensing gold surface bears a highly hydrophilic dense brush of PEG, it must retain such a solvent nanofilm that upholds the bio-reaction kinetics during plug-train transportation. The SPR intensity for air plugs of Fig. 7 is then a combined effect of this nanofilm and the overlying layer of air for another ∼100 nm.⁴

CONCLUDING REMARKS

We have demonstrated a novel droplet-based SPR sensing inside a microchip. In this chip, solution plugs of few hundred nanoliters or less can be synthesized and singulated by an inert, immiscible gas phase to form a short solvent stream in place of a long continuous one. The device is well suited for multi-analyte and high throughput drug screening systems where the sample consumption and reagent cost are important. Using plug-train SPR method, one can characterize an SPR bio-reaction with acceptable accuracy and manifold reduction of bio-sample volumes as compared to the conventional. It also eliminates the effects of dispersive mixing that might be a source of error in parameter estimation. On a concluding note, the numerical technique for the conventional SPR bio-sensing used in this characterization analysis is based on method of least squares³ which has been reported to be less reproducible.^{3, 6} While in this work we introduce the concept of droplet-based SPR characterization, more accurate and reproducible measurements would require SPR microchips that are tailored for more complex dual-slope type SPR measurement method.