Gene Detection in Complex Biological Media Using Semiconductor

Sep 18, 2015 - Gene Detection in Complex Biological Media Using Semiconductor ... Microfluidic systems allow miniaturization and integration of multip...
0 downloads 0 Views 4MB Size
Article pubs.acs.org/ac

Gene Detection in Complex Biological Media Using Semiconductor Nanorods within an Integrated Microfluidic Device Xinyan Bi,† Giulia Adriani,† Yang Xu,‡ Sabyasachi Chakrabortty,† Giorgia Pastorin,§ Han Kiat Ho,§ Wee Han Ang,*,† and Yinthai Chan*,†,‡ †

Department of Chemistry, National University of Singapore, 3 Science Drive 3, Singapore 117543, Singapore Institute of Materials Research & Engineering A*STAR, 3 Research Link, Singapore 117602, Singapore § Department of Pharmacy, National University of Singapore, 18 Science Drive 4, Singapore 117543, Singapore ‡

S Supporting Information *

ABSTRACT: The salient optical properties of highly luminescent semiconductor nanocrystals render them ideal fluorophores for clinical diagnostics, therapeutics, and highly sensitive biochip applications. Microfluidic systems allow miniaturization and integration of multiple biochemical processes in a single device and do not require sophisticated diagnostic tools. Herein, we describe a microfluidic system that integrates RNA extraction, reverse transcription to cDNA, amplification and detection within one integrated device to detect histidine decarboxylase (HDC) gene directly from human white blood cells samples. When anisotropic semiconductor nanorods (NRs) were used as the fluorescent probes, the detection limit was found to be 0.4 ng of total RNA, which was much lower than that obtained using spherical quantum dots (QDs) or organic dyes. This was attributed to the large action cross-section of NRs and their high probability of target capture in a pull-down detection scheme. The combination of large scale integrated microfluidics with highly fluorescent semiconductor NRs may find widespread utility in point-of-care devices and multitarget diagnostics.

D

In order to improve basophil detection, we sought to develop new PCR-based strategies capable of semiquantitatively determining expression levels of biomarkers such as HDC. At present, real-time PCR represents the industry gold standard in clinical diagnosis. It combines PCR amplification with a fluorescent readout, typically using organic dyes that specifically bind double-stranded DNA (ds-DNA) as fluorogenic probes. However, it suffers from inherent drawbacks such as the need for skilled operators, laborious sample preparation, and relatively expensive reagents and equipment, making its implementation in POC testing impractical. Multilayer microfluidics can overcome the above-mentioned challenges by allowing for parallelized, automated processing of nanoliter sample volumes which cannot be achieved via benchtop pipetting.4−7 Indeed, it has been shown that miniaturizing the PCR process to such small volumes not only minimizes reagent costs but also facilitates shorter reaction times.8,9 Furthermore, sample loss is minimized and crosscontamination averted since minimal user intervention is required. There are several reported examples leveraging the advantages of microfluidic architecture specifically for RNA

iseases arising from the body’s immunological response toward allergens constitute some of the most common maladies afflicting people living in the developed world. The etiology of allergic diseases can be both environmental and genetic, but they can manifest in symptoms that are mild, e.g., rhinitis, or severe, e.g., asthma, sometimes with deadly consequences like anaphylaxis. One of the main chemical mediators of allergic diseases is histamine, a biogenic amine that is synthesized by histidine decarboxylase (HDC) via the decarboxylation of endogenous histidine. Within blood, histamine is synthesized and stored in basophils, a rare leukocyte that is present in less than 1% of the peripheral blood leukocytes. Quantification of circulating blood basophils may shed light on the pathogenesis of these diseases.1−3 For example, basopenia, the depletion of blood basophils, is linked to the chronic urticaria as these basophils are being recruited into urticarial wheals during disease activities. Similarly, a concomitant decrease in basophil count in the blood accompanied by an increase in the bronchoalveolar lavage follows an asthma attack due to chemotaxis. On the other hand, rapid increase in basophil levels is indicative of acute infection. Current methods for basophil detection, based on cell staining or flow cytometry, are insensitive, time-consuming, inaccurate, and ultimately incompatible with point-of-care (POC) diagnostics. © XXXX American Chemical Society

Received: May 25, 2015 Accepted: September 18, 2015

A

DOI: 10.1021/acs.analchem.5b01942 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

streptavidin functionalized NRs (SA-NRs) for fluorescent readout. By carrying out the detection on a solid-phase support, unbounded species are easily removed by washing steps. The main advantage of this system is that amplification by SP-PCR allows low abundance gene targets to be detected. Additionally, the efficiency of the detection scheme is not dependent on the length of the targets, allowing for both short and long gene sequences to be detected. This presents a distinct advantage over substrate-free detection methods based on Forster resonance energy transfer (FRET),19 which is highly sensitive to the distance between the acceptor and donor chromophore that is bridged by the gene target. Lastly, this scheme benefits from the fact that binding of SA-NRs can occur directly onto the microbeads following SP-PCR. This reduces incubation and washing steps required which can dramatically shorten overall detection time and significantly reduce the engineering complexity required to construct the microfluidic device.

detection and sensing.10−12 Carrying out real-time PCR within the micron-sized chambers of a microfluidic device, however, imposes drastic limitations on the level of fluorescence that can be generated because the detection volume is constrained. In situations where target abundance is low, binding events between amplified material and the chromophore are rare and thus the number of bound chromophore within the small detection volume is markedly minute. Classical organic dyes, such as SYBR and TaqMan, have relatively low action crosssection and are prone to photobleaching,13 requiring sensitive photodetectors too costly for POC diagnostics. On the other hand, colloidal quantum dots (QDs) which are known for their flexible surface chemistry, size-tunable emission and broad excitation profiles, possess high resistance to photobleaching and in recent times near unity quantum yields.14−16 Typical molar absorption coefficient values for QDs are 105−106 M−1 cm−1 whereas for organic dyes they are ∼2.5 × 104−2.5 × 105 M−1 cm−1.13 This means that, for an equivalent quantum yield (QY), a QD can in principle be ∼4 to 40 times brighter than a dye molecule. Recently, core seeded semiconductor nanorods (NRs) with molar absorption coefficients nearly 1 order of magnitude larger than QDs have been successfully synthesized,17,18 thus showing even more promise as fluorescent labels. We demonstrate in this work that the convergence of multilayer microfluidics and highly fluorescent NRs can potentially overcome many of the key limitations surrounding current POC diagnostic platforms. We validated the efficiency of our NR-based microfluidic diagnostic platform in the detection of the human decarboxylase gene present in low abundance basophils from human white blood cells samples. The device was capable of sample-to-answer operation within 3.5 h with a detection limit of 0.4 ng RNA, which is on the order of fmol. We devised a gene detection scheme designed to harness both the unique physicochemical properties of highly emissive semiconductor NRs as well as the excellent specificity of detection conferred by PCR, as illustrated in Figure 1. In this strategy, immobilized primers are extended using solid-phase PCR (SP-PCR) in the presence of the gene target to form DNA amplicons that are biotin-terminated. These solidsupported biotinylated amplicons can then be used to anchor



EXPERIMENTAL SECTION Reagents and Chemicals. Cadmium acetylacetonate (Cd(acac)2, 99.9%), cadmium oxide (CdO, 99.5%), selenium (Se, 99.99%), 1,2-hexadecanediol (HDDO, 90%), 1-hexadecylamine (HDA, 90%), sulfur (S, reagent grade), trioctylphosphine oxide (TOPO, 90%), 1-octadecene (1-ODE, 90%), octylamine, and poly(acrylic acid) (PAA) were purchased from Sigma-Aldrich. Diisooctylphosphinic acid (DIPA, 90%) was purchased from Fluka. Trioctylphosphine (TOP, 97%), noctylphosphonic acid (ODPA, 97%), trioctylphosphine oxide (TOPO, 99%), and n-hexylphosphonic acid (HPA, 97%) were purchased from Strem. SP-PCR reagents, including Taq DNA polymerase with buffer and dNTP mix were obtained from New England Biolabs, Inc. (Ipswich, MA, USA). Synthetic oligonucleotides, labeled with biotin or amine, were purchased from Integrated DNA Technologies (Coralville, IA, USA). Streptavidin (SA) was purchased from Invitrogen, Thermo Fisher Scientific. Polystyrene (PS) microbeads with PEG COOH surface functional groups (50 mg/mL, size 3 μm) were purchased from micromod Partikeltechnologie GmbH (Germany). SA-FITC, bovine serum albumin (BSA) and 1ethyl-3-(3-dimethylamino propyl) carbodiimide hydrochloride (EDC) were purchased from Sigma Chemical Co. (St. Louis, MO, USA) and used as received. Water was purified through a Milli-Q system (Millipore). PDMS (Sylgard 184A/B) was purchased from Dow Corning Corporation. Negative photoresist SU8-2025 was purchased from MicroChem Corp. Posivite photoresist AZ50XT was purchased from AZ Electronic Materials plc. Synthesis of Spherical Core−Shell CdSe/CdS QDs. The synthesis of highly fluorescent spherical CdSe/CdS QDs was carried out via modification of an established procedure for the synthesis of CdSe/CdZnS QDs.20 Briefly, 207 mg of Cd(acac)2 and 345 mg of HDDO was mixed in 4 mL of 1-ODE and degassed at 130 °C for 1 h. Separately, 21 mg of sulfur powder (S) was dissolved in 4 mL of 1-ODE and degassed at 160 °C for 1 h. In a 100 mL four neck RBF, 9 g of TOPO and 6 g of HDA were degassed in at 80 °C for 0.5 h before addition of CdSe seeds in minimal amount of hexane. The hexane was removed under vacuum at 80 °C for 0.5 h. After degassing for 1 h at 120 °C, the flask was filled with N2. A solution of Cd(acac)2 (∼0.75 mL) was injected when the temperature reached 230 °C followed by addition of an equivalent volume of sulfur solution dropwise at the same temperature. The

Figure 1. Gene detection scheme based on SP-PCR and capture of fluorescent NRs as readout. Extracted total RNA is reverse transcribed to yield a complex cDNA library, which is then amplified by SP-PCR to yield biotin-terminated amplicons. The immobilized biotinterminated amplicons on polystyrene microbeads serve as a pulldown for streptavidin coated semiconductor NRs. B

DOI: 10.1021/acs.analchem.5b01942 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

Modification of PS Microbeads with Amine-Labeled Reverse Primer. First, 20 μL of PEG-COOH functionalized PS beads, 10 μL of amine-labeled reverse primer 5′-NH2(C6)TTTTTT TTTTCC TTGACC CAGAAC CCAGTA (100 μM), and 0.6 mg of EDC were added to 170 μL of 2-(Nmorpholino)ethanesulfonic acid (MES, 0.1 M). The mixture was vortexed and incubated for 20 min at room temperature. Next, 0.6 mg of EDC was added and the mixture was vortexed for 20 min. Then, 0.6 mg of EDC was added and the overall mixture was vortexed for 80 min. The resulting mixture was subsequently centrifuged, and the supernatant was removed carefully. The pellet was resuspended in PBS-T (1 × PBS containing 0.1% Tween-20) to remove any unreacted reverse primer. After centrifuging, the pellet was resuspended in tris(hydroxymethyl)aminomethane hydrochloride-ethylenediaminetetraacetic acid (TE) buffer to quench any unreacted EDC. Finally, the PS beads were washed three times with PBS before subsequent use. On-Chip SP-PCR and Optical Detection. Reverse primer functionalized PS beads with standard Taq reaction buffer, Taq DNA polymerase, dNTPs, biotin-labeled forward primer 5′Biotin-TEG-CTCCAC ATCGAT GCTGCT TA, BSA, and Tween-20 were allowed to flow into the chamber of the microfluidic chip, where they were mixed with cDNA. The chip was put in a thermo cycler (PCR Genemate Service) to perform SP-PCR using the following conditions: initial denaturation at 95 °C for 30 s, followed by 25× three-step cycles (95 °C for 30 s, 60 °C for 30 s and 68 °C for 45 s) and a final extension at 68 °C for 5 min. All fluorescence images of the microfluidic chip were captured using a TECAN LR reloaded scanner at 488 nm excitation and a 575 nm bandpass filter set. Flow Simulation. Fluid dynamic simulations were carried out using COMSOL Multiphysics. To reduce computation time, a small cross-sectional volume of the actual microfluidic chamber (7.8 × 14.4 × 20 μm) was modeled with a porosity of about 0.43. The creeping flow, which adequately describes fluid motion through a packed column of beads within a microfluidic chamber, was simulated by solving the incompressible Navier− Stokes and continuity equations. The boundary conditions were as follows: parabolic inlet velocity profile, outlet at atmospheric pressure, no-slip wall for the top and bottom walls (perpendicular to the y direction) and periodic boundary conditions on the lateral walls (perpendicular to the x direction). Additionally, the hydrodynamic reaction forces acting on a NR and a spherical QD in a linear shear flow were computed for eight different orientations of the particles and wall shear rates (S) ranging between 600 and 7000 s−1.

remaining cadmium solution and sulfur solution were injected dropwise in succession when the temperature was raised to 240 °C. The resulting mixture was left to stir for 20 min at 240 °C, yielding core−shell CdSe/CdS QDs. The as-synthesized QDs were processed by 3 cycles of precipitation in a cosolvent of butanol and methanol and redispersion in hexane. The resulting QDs were then dispersed in CHCl3 for subsequent use. Synthesis of CdSe Seeded CdS NRs. The synthesis of highly fluorescent colloidal CdSe seeded CdS NRs was carried out via the core seeded approach previously reported by Carbone et al.18 Briefly, in a 50 mL three neck RBF, a mixture containing 3 g of TOPO, 65 mg of CdO, 290 mg of ODPA, and 80 mg of HPA are degassed at 150 °C for about 1.5 h. Separately, the S stock solution was prepared by dissolving 80 mg of S in 1.8 mL of TOP at 50 °C before adding 200 μL of the prepared CdSe stock solution. This S stock solution was degassed at 50 °C for 0.5 h. The temperature of the reaction mixture in the three neck RBF was raised to 360 °C under N2 atmosphere. Upon reaching the desired temperature, 1.8 mL of TOP was added, and the temperature was allowed to recover to 360 °C before the mixture of S, TOP and CdSe was swiftly injected. The growth of the anisotropic CdS shell was complete after 6−8 min reaction at 360 °C. The solution was allowed to cool to 80 °C. The NRs were purified by 3 cycles of precipitation in methanol and redispersion in toluene. The resulting NRs were then dispersed in CHCl3 for subsequent use. Rendering QDs and NRs Dispersible in Water. The dispersion of as-synthesized hydrophobic QDs and NRs in water was achieved via slight modifications of a reported method.21 Briefly, ∼40% of the carboxyl groups of PAA was modified with octylamine via a standard EDC coupling reaction in DMF. The QDs and NRs were encapsulated with this amphiphilic polymer, imparting water dispersibility. Conjugation of SA to both QD-COOH and NR-COOH were carried out using a standard EDC coupling reaction as described below. Coupling of Streptavidin to QDs and NRs. Typically, 20 μL of 10 μM polymer coated QDs (or NRs) was diluted to 250 μL with 10 mM borate buffer, pH 7.4 and mixed well with 30 μL of SA stock solution (10 mg/mL in 10 mM borate buffer, pH 7.4). The coupling reaction was initiated by addition of 8 μL of fresh prepared EDC solution (10 mg/mL in cold water). The mixture was left under gentle shaking for 2 h at room temperature to complete the conjugation. Excess unbound proteins and EDC were removed by at least 5 times wash with 50 mM borate buffer (pH 8.3) using a clean centrifugal ultrafiltration unit (100 kDa cutoff). The purified streptavidin coated QDs (SA-QDs) and NRs (SA-NRs) was dispersed in PBS (pH 7.4) and stored at 4 °C for further usage. Microfluidic Device Fabrication. All devices were fabricated using the process of multilayer soft lithography (MSL). Devices were composed of three layers of PDMS bonded to a glass slide with push-up valve geometry. Negative master molds were patterned with 20 000 dpi transparency masks (Infinite Graphics Pte Ltd.) and fabricated out of photoresist by standard photolithography. SU8-2025 (∼24 μm high) was utilized for the control channel molds. The flow channel molds were made from AZ 50XT and SU8-2025. The on−off microfluidic valves within each device were actuated by an array of independently addressable solenoid valves attached to a pressurized air source. The solenoid valves were controlled by a Labview graphical interface.



RESULTS AND DISCUSSION As mentioned earlier, the synthesis of highly fluorescent colloidal CdSe seeded CdS NRs was carried out via the core seeded approach previously reported by Carbone et al.18 This synthesis procedure results in the asymmetric growth of CdS over core CdSe seeds, yielding uniform rod-like nanostructures of ∼5 nm in diameter and ∼32 nm in length. Longer rod lengths may be obtained by tuning the amount of CdSe seeds added and the ratio of Cd to S precursor. For comparison, we also synthesized ∼5 nm diameter spherical core−shell CdSe/ CdS QDs emitting at a similar wavelength. The hydrophobic as-synthesized QDs and NRs were rendered dispersible in water by encapsulating with an amphiphilic polymer which comprised an octylamine modified poly(acrylic acid).21 The C

DOI: 10.1021/acs.analchem.5b01942 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

electron microscope images of SA-NRs and SA-QDs, respectively, where it may be seen that the particles were well-spaced apart, indicating that they did not suffer from any noticeable agglomeration. Using dynamic light scattering (DLS), their corresponding hydrodynamic diameters were determined to be 42.1 ± 4.2 and 32.8 ± 3.3 nm for NRs and QDs, respectively. If we approximate the NRs as prolate ellipsoids, their dimensions may be estimated based on the Perrin’s relations22 as a ≈ 27.8 nm and b ≈ 16.4 nm, where a and b are the major and minor semiaxes of the ellipsoid. A cartoon schematic depicting the average dimensions of each type of particle is given in Figure 2c. In addition to action-cross section, the pull-down efficiency of the chromophore is critical to determining the overall detection efficiency of the device. To maximize the surface area available for target DNA capture, we used a packed chamber of microbeads that were surface functionalized with primers containing a 30 base recognition sequence (as illustrated in Figure 1). Successful capture of the target DNA allows for SPPCR to be carried out, resulting in biotinylated amplicons immobilized on the microbead surface. Finally, pull-down of SA-QDs or SA-NRs completes the detection process. It is not well understood how differences in shape between NRs and spherical QDs affect their pull-down while being transported through the small voids of a packed column of microbeads. To gain some level of understanding of the effects of nanocrystal shape from the standpoint of fluid dynamics, the reaction forces of a NR and QD in a linear shear flow were computed on the basis that the reaction forces are equal and opposite to the hydrodynamic forces exerted on the particles by the flow. The range of S in consideration was chosen from the shear rate values calculated by modeling a portion of a microfluidic chamber containing beads with a porosity of 0.43 (Figure 3a). The fluid motion through the beads was simulated by solving the incompressible Navier−Stokes and continuity equations.

octyl chains of the polymer interact strongly with the native hydrophobic ligands of the nanoparticle while the carboxylic acid residues on the exterior backbone of the polymer facilitate dispersion in water. Subsequently, SA was conjugated to the carboxylic acid groups of the polymer coated NRs and QDs via EDC coupling to yield aqueous dispersions of SA-NRs and SAQDs in PBS. The absolute QYs of the synthesized SA-NRs and SA-QDs were generally in the range of 10−15% and 20−30% respectively, which is expected after ligand-exchange with the amphiphilic polymer. Despite their seemingly low QY, the action cross section of these NRs at an excitation wavelength of 350 nm is ∼10 times larger than fluorescein at its excitation maximum. Figure 2a,b shows representative transmission

Figure 2. TEM images of SA-conjugated, polymer coated CdSe seeded CdS NRs (a) and CdSe/CdS QDs (b). (c) Schematic of a SA-NR and SA-QD with average dimensions obtained via TEM and hydrodynamic sizes obtained via DLS measurements.

Figure 3. (a) Inlet velocity profile, velocity contours in the zy plane [m/s], wall shear rates on beads [s−1]. (b) Velocity contours [m/s] and (c) xvorticity contours [1/s], considering a nanosphere and a nanorod in eight orientations, at 60 nm from the wall. (d) Reaction torques of a nanorod and a nanosphere in a linear shear flow considering eight different orientations and wall shear rates S ranging between 600 and 7000 s−1. D

DOI: 10.1021/acs.analchem.5b01942 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry The inertia terms were discarded due to a low Reynolds number (Re ≪ 1) as determined in the Supporting Information. Although Brownian diffusion is expected to play a role in mass transport within the chamber, a calculated Péclet number >1 (Supporting Information) suggested that convection dominated the transport dynamics. Mass transport by convection was therefore the primary consideration in this analysis. The shear rate and velocity contours, as illustrated in Figure 3a, showed that the velocity of the fluid was zero at the sphere surface (no-slip condition) and increased away from the sphere with S up to ∼5000 s−1. S was found to be even higher in the gap between the beads and walls of the channel (Figure 3a). Therefore, values of S ranging between 600 and 7000 s−1 were considered for the computation of the reaction forces. Representative velocity and vorticity contours are shown in Figure 3b,c for S = 600 s−1. Analysis of these results revealed that the torque on the NR increased with S. Importantly, it was found that the NR undergoes a higher torque compared to the sphere for most of the eight angular orientations and for all the wall shear rates considered (Figure 3d). We predicted based on these calculations that due to its elongated shape, rotating a NR doubles its area, thereby making it effectively three times larger than a QD. Therefore, our choice of employing a packed column of microbeads as opposed to a flat surface for pulldown detection not only increases the effective surface area for binding to the target but also increases S, which in turn promotes tumbling in NRs and leads to higher sampling rates at the microbead surface. Based on these considerations, and the fact that the action cross-section of NRs can be much larger than that of QDs, it may be expected that the use of NRs in our detection scheme would allow for lower detection limits to be achieved than QDs. In broader terms, it may be hypothesized that core seeded NRs can serve as more efficient fluorescent labels than core−shell QDs for in vitro diagnostics. To automate the detection scheme described in Figure 1, we constructed a microfluidic device capable of delivering sampleto-answer operation within nanoliter sample volumes. Sample processing and reagent dispensing capabilities were incorporated into the detection device for minimal user intervention. Our POC device was based on a microfluidic large-scale integration (mLSI) design that was microfabricated with polydimethylsiloxane (PDMS) using MSL.23 Each device comprised of three chambers (ca. 60 nL each) in sequential order for (A) total RNA extraction, (B) reverse transcription (RT), and (C) SP-PCR and fluorescence detection (as exemplified in Figure 4a). Fluid manipulation through the inlet/outlet channels and chambers (yellow, green, red) were controlled via pressurized push-up valves (blue) actuated by a computer-controlled solenoid valve array. This allowed for clinical samples to be processed and transferred between chambers in an automated and highly controlled fashion. Reagents required for each process and buffer solutions used for washing steps were also introduced directly to each reaction chamber. A schematic of the device operation is given in Figure 4b. Sieve valves (dark crimson, Figure 4a) were utilized in chambers (A) and (C) to trap microbeads used for RNA extraction and SP-PCR. Positive detection of the gene target was indicated by the observation of NR or QD fluorescence from chamber (C) upon irradiation with light of shorter wavelengths (blue or UV in this case).

Figure 4. (a) Photograph of the device: Food dyes are placed in the channels for better visualization (scale bar, 1 cm). Chambers for RNA extraction (yellow, A), RT (green, B), SP-PCR/detection (red, C). The control layer is filled with blue color food dye. The sieve valve is filled with cochineal color food dye. (b) Schematic illustration of device.

The microfluidic device was primed for gene detection after blocking the nonspecific reactive sites of the PDMS walls using 1 mg/mL of BSA solution prior to each experiment. Materials from commercial RNA extraction kit (Qiagen RNeasy Mini Kit) was adapted for use in the device to carry out total RNA extraction. First, the silica matrix obtained from the kit was loaded into chamber (A) before human white blood cell lysate was introduced. Biomolecules, including RNA, were thus bound onto the silica matrix. The silica matrix was washed successively with washing buffer to remove nonspecifically bound biomolecules and contaminating salts, before RNA was eluted into the chamber (B) using RNase-free water. We validated the efficient RNA extraction by sampling for RNA content in the eluted total RNA solution using UV−vis spectroscopy (Nanodrop). Next, SuperScript III reverse transcriptase, oligo-(dT)20 primers, dNTPs and buffer was introduced to chamber (B) to initiate the synthesis of the first strand cDNA library. Specifically, the primer and dNTPs was mixed with the RNA solution, containing mRNA templates, in chamber (B) at 65 °C for 5 min and the chip was placed atop an ice bath for 1 min. Thereafter, the reverse transcriptase enzyme was loaded into the chamber with buffer and the transcription mixture incubated at room temperature for 10 min, at 50 °C for 50 min and at 85 °C for 5 min to carry out cDNA synthesis. RNase H was then introduced to remove the mRNA template at 37 °C for 20 min. This complex cDNA library was transferred to chamber (C), containing reverse primer-functionalized microbeads, for SP-PCR detection against HDC. To ensure efficient hybridization for SP-PCR, 10 additional dT nucleotides were added in the reverse primer design at the 5′-end served as a spacer between the priming sites and the bead surface. Taq DNA polymerase, dNTPs, biotin-labeled forward primer were introduced to initiate PCR. SP-PCR was E

DOI: 10.1021/acs.analchem.5b01942 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry then carried out over 25 cycles and completed in 50 min. Excess SA-QDs or SA-NRs were introduced to chamber (C) and incubated for 60 min to conjugate with the microbeads via SA-biotin interaction. The beads were washed with PBS-T buffer against the sieve valves and imaged via the TECAN scanner. Using treated human white blood lysate, we observed a strong fluorescence signal (Figure S2a and S2b) from the semiconductor nanoparticles upon completion of the detection cycle. In contrast, when DI water was used in place of the white blood lysate as a negative control, no fluorescence response was observed (Figure S2c and S2d). Taken together, these results indicate that (i) the integrated mLSI device allows for the extraction, RT, amplification and specific detection of the HDC gene from complex human white blood samples on a single chip; (ii) the SA-QDs or SA-NRs did not adsorb nonspecifically onto the surface of the microbeads. To validate our earlier hypothesis that core-seeded NRs are more effective fluorescent labels than conventional core−shell QDs, we compared the signal intensities obtained from identical microfluidic devices in which either SA-NRs or SAQDs were used. As a control, we used FITC-labeled forward primers to model a situation when a commercial organic dye is used as the fluorophore for direct amplicon detection. Emission of the FITC-labeled amplicon was quantitated with a 535 nm bandpass filter corresponding to its peak emission wavelength. The same RNA mixture at different quantities was added to chamber (B) while chamber (C) of the different devices was exposed to different chromophores. The devices were operated in accordance to the devised workflow without involving total RNA extraction and each experiment was performed in triplicates. Figure 5a shows the mean fluorescence intensity versus total RNA. It was found that the SA-NRs (Figure 5b) and SA-QDs generated strong fluorescence intensity that could be adequately imaged with 12.8 ng of total RNA. The fluorescence intensity from FITC, on the other hand, was comparatively weaker as illustrated in Figure 5c. When the quantity of RNA was reduced 4-fold to 3.2 ng, only the SA-NRs yielded a discernible fluorescence response. To evaluate the performance of our device at its limit of detection (LOD), we used total RNA quantities of 3.2, 1.5, and 0.4 ng and measured the fluorescence response from the SANRs in triplicates. Based on our measurements, the LOD was determined to be 0.4 ng (S/N ratio = 4:1), which is on the order of femtomoles of total RNA. This exceptional sensitivity and linear quantification exceeds the levels required for making quantitative assessment on pathogenic levels. When SA-QDs were used, the LOD was nearly 1 order of magnitude larger. The much lower LOD facilitated by the NRs cannot be entirely accounted for by differences in action cross-section, which was approximately 4:1 (NRs:QDs) for the particles used in this measurement. The number of SA groups per unit surface area of both NRs and QDs were comparable, and therefore cannot be used to explain differences in target capture efficiency. It is likely that the elongated shape of the NRs aids in more efficient pull-down, as elaborated earlier, ultimately leading to better overall detection. In line with our expectations, these results indicated that SA-NRs were more effective fluorophore labels compared to SA-QDs and classical FITC molecules. To vindicate our use of microfluidics and the device architecture we adopted, we performed the entire procedure, with the exception of total RNA extraction, using standard benchtop techniques in 0.5 mL eppendorf centrifuge tubes. The beads exposed to SA-NRs were then injected into a

Figure 5. (a) Sample-to-answer operation within the integrated microfludic device for different amounts of total RNA. Three fluorescent labels, including NRs, QDs, and FITC, are employed. (b,c) The TECAN scanner images at 12.8 ng of total RNA using NRs and FITC as labels, respectively. (d) Graph of fluorescence intensity versus total RNA in sample obtained from a microfluidic chip (orange) and Eppendorf tubes (red) via standard benchtop techniques. A linear relationship between mean intensity of microbeads and the amount of total RNA with correlation coefficient R = 0.98 was found for the chip. For both experiments, SA-NRs were used to conjugate with the biotinlabeled amplicons on the microbeads surface.

microfluidic device and imaged for comparison. Figure 5d plots the integrated fluorescence intensity as a function of total RNA, where it is readily seen that the device performed significantly better than the manual protocol which yielded very low fluorescence intensities. The chip yielded fluorescence intensities 35× and 12× that of the manual protocol when total RNA quantities of 3.2 and 1.5 ng were used, respectively. We reasoned that the better performance was attributable to (1) the integration of multiple sample-processing steps on a single microfluidic chip which reduced sample loss and contamination and (2) nanoliter reaction volumes which enhanced SP-PCR efficiency by allowing a drastic reduction of sample volume and diffusion time of the reagents.



CONCLUSIONS In summary, we described here the use of fluorescent CdSe seeded CdS NRs as labels within a microfluidic chip capable of seamlessly processing biological samples and carrying gene detection. The chip was designed to automate multiple tasks such as total RNA extraction, reverse transcription to cDNA, amplification, and detection. It exhibited remarkable LOD attributable to the nanoliter reaction volumes within the chip as well as the photophysical properties of the NRs as fluorophores. We demonstrated using a combination of experimental and computational results that the anisotropic nature of these core seeded NRs greatly boosted the device performance on gene detection. This was ascribed not only to the higher action crossF

DOI: 10.1021/acs.analchem.5b01942 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

(16) Chen, O.; Zhao, J.; Chauhan, V. P.; Cui, J.; Wong, C.; Harris, D. K.; Wei, H.; Han, H.-S.; Fukumura, D.; Jain, R. K.; Bawendi, M. G. Nat. Mater. 2013, 12, 445−451. (17) Talapin, D. V.; Nelson, J. H.; Shevchenko, E. V.; Aloni, S.; Sadtler, B.; Alivisatos, A. P. Nano Lett. 2007, 7, 2951−2959. (18) Carbone, L.; Nobile, C.; De Giorgi, M.; Sala, F. D.; Morello, G.; Pompa, P.; Hytch, M.; Snoeck, E.; Fiore, A.; Franchini, I. R.; Nadasan, M.; Silvestre, A. F.; Chiodo, L.; Kudera, S.; Cingolani, R.; Krahne, R.; Manna, L. Nano Lett. 2007, 7, 2942−2950. (19) Jares-Erijman, E. A.; Jovin, T. M. Nat. Biotechnol. 2003, 21, 1387−1395. (20) Snee, P. T.; Chan, Y.; Nocera, D. G.; Bawendi, M. G. Adv. Mater. 2005, 17, 1131−1136. (21) Chen, Y. C.; Thakar, R.; Snee, P. T. J. Am. Chem. Soc. 2008, 130, 3744−3745. (22) Pecora, R. Dynamic Light Scattering: Applications of Photon Correlation Spectroscopy; Plenum Press: New York, 1985. (23) Unger, M. A. Science 2000, 288, 113−116.

section relative to QDs that made them more emissive, but also their rod-like structure that resulted in a higher hydrodynamic profile and more efficient capture onto the solid support. Although the present work illustrates how the convergence of semiconductor NRs and multilayer microfluidics can address the challenge of detecting basophils from a complex biological sample, it may be envisaged that such an integrated detection platform may be harnessed to tackle other biological systems where in vitro detection has proven difficult.



ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.analchem.5b01942. UV−vis and fluorescence spectra, fluorescence images, the proof of concept of the integrated microfluidic device, and determination of Reynolds and Péclet numbers (PDF)



AUTHOR INFORMATION

Corresponding Authors

*E-mail: [email protected]. Tel: 65165131. Fax: 67791691. *E-mail: [email protected]. Tel: 65166788. Fax: 67791691. Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS This work was supported by a Visiting Investigator Program grant from the Joint Council Office of the Agency for Science, Technology and Research in Singapore (Project No. 1235e00049).



REFERENCES

(1) Waterer, G. Curr. Opin. Infect. Dis. 2011, 24, 130−136. (2) Butler, D. Nature 2012, 490, 20−20. (3) Graham, R. L.; Baric, R. S. J. Virol. 2010, 84, 3134−3146. (4) Squires, T.; Quake, S. R. Rev. Mod. Phys. 2005, 77, 977−1026. (5) Hansen, C.; Quake, S. R. Curr. Opin. Struct. Biol. 2003, 13, 538− 544. (6) Mairhofer, J.; Roppert, K.; Ertl, P. Sensors 2009, 9, 4804−4823. (7) Park, S.; Zhang, Y.; Lin, S.; Wang, T. H.; Yang, S. Biotechnol. Adv. 2011, 29, 830−839. (8) Sanchez-Freire, V.; Ebert, A. D.; Kalisky, T.; Quake, S. R.; Wu, J. C. Nat. Protoc. 2012, 7, 829−838. (9) Lee, D.-S.; Park, S. H.; Yang, H.; Chung, K.-H.; Yoon, T. H.; Kim, S.-J.; Kim, K.; Kim, Y. T. Lab Chip 2004, 4, 401−407. (10) Shaw, K. J.; Hughes, E. M.; Dyer, C. E.; Greenman, J.; Haswell, S. J. Lab. Invest. 2013, 93, 961−966. (11) Cao, Q.; Mahalanabis, M.; Chang, J.; Carey, B.; Hsieh, C.; Stanley, A.; Odell, C. A.; Mitchell, P.; Feldman, J.; Pollock, N. R.; Klapperich, C. M. PLoS One 2012, 7, e33176. (12) Foudeh, A. M.; Brassard, D.; Tabrizian, M.; Veres, T. Lab Chip 2015, 15, 1609−1618. (13) Resch-Genger, U.; Grabolle, M.; Cavaliere-Jaricot, S.; Nitschke, R.; Nann, T. Nat. Methods 2008, 5, 763−775. (14) Medintz, I. L.; Uyeda, H. T.; Goldman, E. R.; Mattoussi, H. Nat. Mater. 2005, 4, 435−446. (15) Michalet, X.; Pinaud, F.; Bentolila, L. A.; Tsay, J.; Doose, S.; Li, J.; Sundaresan, G.; Wu, A. M.; Gambhir, S. S.; Weiss, S. Science 2005, 307, 538−544. G

DOI: 10.1021/acs.analchem.5b01942 Anal. Chem. XXXX, XXX, XXX−XXX