Three-Dimensional Barcodes with Ultrahigh Encoding Capacities: A

ABSTRACT: Novel 3D barcodes with an extraordinarily high encoding capacity are developed through a flexible, accurate and reproducible method. Here, t...
0 downloads 4 Views 9MB Size
Article pubs.acs.org/cm

Three-Dimensional Barcodes with Ultrahigh Encoding Capacities: A Flexible, Accurate, and Reproducible Encoding Strategy for Suspension Arrays Si Lu, Ding Shengzi Zhang, Dan Wei, Ye Lin, Shunjia Zhang, Hao He, Xunbin Wei, Hongchen Gu, and Hong Xu* Shanghai Jiao Tong University Affiliated Sixth Hospital, School of Biomedical Engineering, Shanghai Jiao Tong University, Shanghai 200030, People’s Republic of China S Supporting Information *

ABSTRACT: Novel 3D barcodes with an extraordinarily high encoding capacity are developed through a flexible, accurate, and reproducible method. Here, the three dimensions refer to the size, fluorescence emission wavelength, and intensity of the barcodes. As a proof of concept, a 3D barcode library with an encoding capacity of 100 flow cytometry-distinguishable barcodes is achieved successfully by combining the multiscaled host particles with a set of guest nanoparticles owning to their individual different fluorescent intensity via an ingenious host−guest structure, where fluorescein isothiocyanate and a kind of quantum dots are separately loaded inside. Five-plexed tumor marker detection is further implemented, and the results demonstrate the strong feasibility of 3D barcodes in multiplex assays.



commercialized xMAP technology,15−17 where orange and red dyes are loaded in microscaled beads at 10 different concentrations by the swelling method. Hence, an encoding capacity of 100 barcodes has been obtained via flow cytometry. QDs, another type of optimal fluorophore for optical barcodes, have the potential to improve encoding capacity because of their unique optical properties,28−30 like single-wavelength excitation, tunable emission wavelength, and narrow emission spectra. Since the pioneering work of the Nie group,18−20 QD barcodes have been highly prevalent and have shown remarkable capabilities in optical encoding.22,24,31−37 Recently, microfluidic fabrication of QD barcodes earns an arresting reputation for its simple encoding process and high capacity.32−37 Chan et al. utilized five color QDs to achieve over 100 fluorescence spectra recognizable barcodes via successfully importing of the microfluidic technique, and the prepared barcode beads are demonstrated to be feasible in nineplex detection based on the flow cytometry platform.23,36 In summary, current optical encoding strategies mainly focus on doping different fluorophores with various concentrations into a single particle. However, this strategy has exposed several problems: (i) the encoding signals are typically limited to two dimensions: fluorescent emission wavelength and intensity of

INTRODUCTION As the field of systematic biology has progressed, the need for multiplex assays involving the simultaneous detection of multiple targets in a single sample is increasingly developing.1,2 Suspension arrays, benefiting from fast binding kinetics, flexible array preparation, and multiplexed analytical capability,3 are the most promising methods to meet this need. Barcodes, as the core elements of suspension arrays used to identify and capture multiple targets, should be enormously encoded and easily decoded.4,5 Compared with other encoding strategies,2 such as chemical encoding6 and graphical encoding,7,8 fluorescencebased optical encoding is the most universal strategy owing to its simple preparation, low cost, and well-developed decoding systems.9,10 Among the typical decoding systems used for optical barcodes,2 flow cytometry has been most widely accepted because of its high sensitivity, accuracy, and fast detecting speed.5,11 From the perspective of time efficiency and sensitivity, the tandem use of fluorescence-based optical encoding and cytometry-based decoding is the mainstream barcode platform for high-throughput multiplex assays. Nonetheless, in multiplex assays, the number of target molecules that can be assessed in parallel is restricted to the number of distinguishable barcodes in flow cytometry. To date, numerous successful cases have been developed to enlarge the encoding capacity of barcodes based on two commonly used encoding fluorophores, organic dyes12−17 or quantum dots (QDs).18−27 The most classical dye-encoding system is the © 2017 American Chemical Society

Received: September 8, 2017 Revised: November 9, 2017 Published: November 10, 2017 10398

DOI: 10.1021/acs.chemmater.7b03811 Chem. Mater. 2017, 29, 10398−10408

Article

Chemistry of Materials

Scheme 1. Schematic Showing the Three-Dimensional Encoding Strategy Used To Fabricate DMBs with Ultrahigh Encoding Capacities

the total number of 3D barcodes was given by ΣSi=1 Hi × Gi, as the assorted sizes can lead to different H and G levels, resulting in a 3D array with an enormous encoding capacity. Comprised of several building blocks like the Lego bricks, these assynthesized barcodes combined three discrete encoding signals which could be independently manipulated without interfering with each other via loading different encoding elements in separate building blocks, significantly enlarging the encoding capacity and ensuring the controllability and reproducibility of the encoding process in a flexible and simple manner. Additionally, interactions between different fluorophores were avoided, thus enhancing the encoding accuracy and simplifying the decoding procedures. Besides, we evaluated the fluorescence stability and multiplex biodetection performance of the 3D barcodes to verify their practicability and effectiveness for multiplex assays.

the barcodes,19,32,38,39 which restrains the encoding capacity. (ii) The loading space of the barcodes is too limited to accommodate large amounts of multicolor fluorophores, and the loading process among different encoding elements may restrict each other, resulting in inaccurate and uncontrollable encoding10,23,34,35 together with a low-reproducibility manufacturing process. Besides, by virtue of the short distance of different fluorophores, possible mutual interactions between different encoding elements such as fluorescence resonance energy transfer (FRET) may affect the fluorescence intensity and distribution of the resultant barcodes, resulting in entangling the encoding and complicating the decoding of the barcodes. Therefore, the preparation of flow cytometrydistinguishable optical barcodes with high encoding capacity, accuracy, and reproducibility encoding manner remains a challenge for suspension arrays. In this study, we develop a novel 3D encoding strategy, as illustrated in Scheme 1. The 3D barcode consists of two types of isolated building blocks: a QD-encoded host particle (QHP) in variable microscales and a set of dye-doped guest nanoparticles with different fluorescent intensities. After assembling two building blocks together, a strawberry-like structure is formed, denoted as a host−guest structure. More importantly, there is no restriction to the type of fluorophore loaded in each building block as long as their fluorescence emission spectra have clear distinction. Here, three dimensions refer to size, emission wavelength, and fluorescence intensity of barcodes. Thus, the 3D-encoding strategy was accomplished by three simple steps. First, magnetic spheres (MSs) with diverse sizes acted as the core substrates to attain assorted size levels (denoted as S level). Second, QHPs were prepared through a modified layer-by-layer assembly method we proposed here. Different QD fluorescence intensity levels (denoted as H level) were precisely achieved via adjusting the QD concentration and the number of QD assembly layers. Finally, dual-encoded magnetic barcodes (DMBs) were prepared, and different fluorescein isothiocyanate (FITC) fluorescence intensity levels (denoted as G level) were obtained through a host−guest structure constructed by our group previously.13,22 Therefore,



METHODS

Materials. Carboxyl-modified magnetic spheres with diameters of 2.9 and 6.2 μm were purchased from Bangs Laboratories. Poly(ethylenimine) (PEI, Mw = 750 K), branched poly(ethylenimine) (PEI, Mw = 25K), poly(ethylenimine) (PEI, Mw = 10K), poly(diallydimethylammonium chloride) solution (PDDAC, 100 000 ≤ Mw ≤ 200 000), poly(sodium 4-styrenesulfonate) (PSS, Mw = 70 000), and 2-(N-morpholino) ethanesulfonic acid hydrate (MES) were obtained from Sigma-Aldrich. Analytical-grade (AR) chloroform, nbutanol, ethanol, tetraethyl orthosilicate (TEOS), and ammonium hydroxide were purchased from Sinopharm Chemical Reagent. CdSSe/ZnS quantum dots were gained from Najing Technology Corp. (Hangzhou, China). N-(3-(Dimethylamino)propyl)-N′-ethylcarbodiimide hydrochloride (EDC) was purchased from J&K Chemical. N-Hydroxysuccinimide (NHS) was bought from Thermo Fisher Scientific. Antialpha-fetoprotein antibody (AFP) was provided from Fitzgerald (Acton, USA). Anticarcinoembryonic antigen antibody (CEA) and Antineuronspecific enolase antibody (NSE) were purchased from Shanghai Linc-Bio Science (Shanghai, China). Anticarbohydrate antigen 199 antibody (CA199) and AntiCYFRA21-1 antibody (CYFRA211) were gained from Fapon Biotech Inc. (Shenzhen, China). All reagents were used without further purification, and Millipore-purified water (18.2 MΩ·cm) was used during the whole experiment. 10399

DOI: 10.1021/acs.chemmater.7b03811 Chem. Mater. 2017, 29, 10398−10408

Article

Chemistry of Materials Characterization. The scanning electron microscopy (SEM) images were obtained using a Zeiss Ultra Plus field emission scanning electron microscope (Carl Zeiss AG, Germany). The zeta potential of the beads was detected by a Zetasizer Nano ZSP (Malvern, UK). Fluorescence spectra were recorded by a fluorescence spectrometer (F-2700, HITACHI, Japan). Confocal images were tested via a Leica TCS SP5 (Leica, Germany) instrument, where an argon-ion laser (λ = 488 nm) was adopted as excitation source, and two emission channels (channel 1, 500−530 nm; channel 2, 595−625 nm) were chosen as detector. The flow cytometry analysis was carried out on an Accuri C6 instrument (BD, USA). FL1 channel (515/20 nm) and FL2 channel (610/20 nm) were excited by a 488 nm laser, and FL4 channel (675/ 12.5 nm) was excited by a 640 nm laser. Preparation of QD-Encoded Host Particles (QHPs). Through the coordination interaction with the primary amino group of PEI, QDs were immobilized onto the surface of MSs layer by layer (LBL). First, carboxylic magnetic spheres (MSs) were functionalized with PEI. A 30 mg amount of MS, 312.4 mg of PEI (Mw = 750 K), and 10.5 mg of EDC were mixed together in 8 mL of MES (10 mM, pH = 5.0) and reacted for 3 h. The obtained beads were washed 6 times with water. Second, the MSs were encoded with the first layer of QDs by the following steps: (1) PEI-functionalized MSs (22 million L-MSs or 100 million S-MSs) were washed with ethanol 2 times and then rotated in 0.9 mL of chloroform and n-butanol mixed solvent (v:v = 1:20) containing QDs with different concentrations for 0.5 h; (2) the beads were washed by the above mixed solvent, ethanol and water in turn, and then rotated in 1 mL of PEI solution (Mw = 25K, 9 mg/mL PEI, 0.5 M NaCl) for 0.5 h; (3) the beads were washed with water 3 times. Third, additional layers of QDs were added by repeating the above steps 1, 2, and 3. Therefore, QDs-encoded magnetic spheres (MS@ QDs) were obtained. QD barcodes prepared by LBL assembly usually suffer from QD leakage since QDs are loaded on the surface of microspheres, leading to fluorescence instability, which has been one of the major bottlenecks restraining commercialization and practical use of QD barcodes. Therefore, QD-encoded beads in this study were further sealed by a composite encapsulation of silica and polyelectrolytes. MS@QDs were primarily encapsulated by a shell of silica (the obtained beads were denoted as MS@QDs@SiO2) via a condensation reaction of tetraethoxysilane (TEOS) as below: MS@QDs were washed by ethanol 2 times and rotated in a solvent mixture containing 3 mL of ethanol, 300 μL of water, and 20 μL of TEOS for 0.5 h. Then 22 μL of ammonium hydroxide was injected into the above mixture and reacted for 22 h. The product was washed with ethanol and water to obtain MS@QDs@SiO2. To enhance the protection to QDs and offer a soft substrate as well as rich binding sites for the landing of guest nanoparticles, a shell of polyelectrolytes (PEs) was further formed through electrostatic interaction. PDDA and PSS were alternately absorbed onto the surface of MS@QDs@SiO2 by the following procedures: (1) MS@ QDs@SiO2 were added and incubated in PDDA solution (5 mg/mL, in 0.5 M NaCl) for 15 min and then washed by water; (2) the beads were incubated in PSS solution (5 mg/mL, in 0.5 M NaCl) for 15 min and then washed by water again; (3) the beads were encapsulated by three layers of PDDA/PSS through repetition of steps 1 and 2 twice; (4) the beads were dispersed in PEI (Mw = 10K) solution (50 mg/mL, in 10 mM MEST), reacted for 0.5 h, and then washed by water for 3 times. As a result, MS@QDs@SiO2 was successfully encapsulated by PEs to obtain QHPs. Preparation of Dual-Color Magnetic Barcodes through Host−Guest Structure. Dual-color magnetic barcodes (DMBs) were prepared through ingenious host−guest structure. Two building blocks, amino-functionalized QHPs and carboxylated FITC-doped guest nanoparticles, were conjugated through a carbodiimide-based coupling reaction. QHPs (22 million L-QHPs or 100 million S-QHPs) were washed in 100 mM MES three times. Then QHPs were dispersed in 300 μL of MES and added dropwise into 300 μL of MES containing 5 mg of guest nanoparticles and then reacted for 0.5 h. Afterward 200 μL of MES containing 10 mg of EDC and 10 mg of NHS was injected into the reaction mixture. After 3 h reaction, the beads were

magnetically separated and then washed with sodium hydroxide (0.01M) and water, respectively; thus, DMBs were prepared successfully. On the basis of host−guest structure, different host codes and different guest codes were combined flexibly, resulting in a large number of optically distinguishable DMBs. Decoding of DMBs. The decoding of DMBs was carried out by flow cytometry detection and a further data processing. Four channels of flow cytometry were used. FSC channel is used to detect forward scattering intensity, which is related to the size of the beads. FL1 (515/ 20 nm) and FL2 channel (610/20 nm) were used to determine FITC fluorescence and QD fluorescence, respectively. To eliminate the fluorescence spillover, the FITC fluorescence intensity collected in the FL1 channel was compensated for by subtracting 0.25% of the FL2 signals, and the QD fluorescence intensity collected in the FL2 channel was compensated by subtracting 1.40% of the FL1 signals. Processing of the collected flow cytometry data was performed using homemade software coded via MATLAB (MathWorks, USA). The first step was size-based decoding. Typically, the data of forward scattering (FSC) and side scattering (SSC) was chosen to form a 2D density plot. By enclosing appropriate gates around the high-density regions in the plot, microbeads with different diameters were preliminarily extracted. Then another innovative gate on the 2D density plot of FSC−Area vs FSC−Height was used to eliminate debris and doublets of DMBs to select single DMBs. The second step was fluorescence-based decoding. A density plot of fluorescence intensity in channel FL1 vs fluorescence intensity in channel FL2 was drawn; thus, a distribution matrix with separated clusters was formed. Each cluster corresponded to one kind of DMBs with specific coding signal. Thus, the barcodes were successfully decoded. Multiplex Assays. To further certify the potential application of DMBs for multiplex assays, five typical DMBs were chosen to detect five tumor markers. Accordingly, DMBs with coding address of q3f5, q5f1, Q2F0, Q4F3, and Q6F5 were conjugated with five specific capture antibodies corresponding to anti-AFP antibody, anti-NSE antibody, anti-CEA antibody, anti-CA199 antibody, and antiCYFRA211 antibody in turn via a carbodiimide-assisted coupling reaction. First, each kind of DMB (1 mg) was activated in 400 μL of 10 mM MES buffer (pH 6.0), in which 25 mg/mL NHS and EDC were mixed for 15 min. Next, the activited DMBs were washed with phosphate buffer (10 mM, pH 7.4) twice, and then 100 μg of antibodies was added and incubated for 2 h. Finally, 400 μL of phosphate buffer containing 0.5 wt % BSA and 0.3 wt % glycine was added into the Ab-DMBs complexes and incubated overnight at 4 °C; the complexes were washed and stored in 100 μL of phosphate buffer containing 0.1% BSA at last. The Ab binding capacity was determined using a BCA protein quantification kit through the depletion method. The amounts of antibodies binded onto q3f5, q5f1, Q2F0, Q4F3, and Q6F5 were 1.161, 0.602, 4.087, 2.480, and 1.230 pg/bead, respectively. The obtained five DMBs-loaded specific capture antibodies and the five corresponding secondary antibodies labeled with allophycocyanin (APC, EM. 660 nm) were added into a 50 μL assay system containing five tumor markers with various concentrations. After 1 h incubation at 37 °C, the samples were washed with PBST two times. The barcode samples were measured by flow cytometry through FSC (size detector), FL1 (FITC detector), FL2 (QD detector), and FL4 (APC detector) channels. When the above beads passed the sensing point of flow cytometry, the first laser excited the fluorophores doped in the barcodes and the type of captured proteins was recognized by the three encoding signals of DMBs detected in FSC, FL1, and FL2 channels, while the second laser excited the APC fluorophore labeled on the analytes, and the concentration of the analytes was determined in FL4 channel. Hence, the target molecules were identified and quantified.



RESULTS AND DISCUSSIONS Preparation of QD-Encoded Host Particles by Modified Layer-by-Layer Assembly. High-performance QHPs, which show excellent brightness, uniformity, and 10400

DOI: 10.1021/acs.chemmater.7b03811 Chem. Mater. 2017, 29, 10398−10408

Article

Chemistry of Materials

Figure 1. (a) Schematic illustration showing the preparation of QD-encoded host particles. Dark gray spheres represent magnetic core spheres. Red shell with different thickness embodies the QD-doped shell comprised of different numbers of QD layers. Uttermost pink shell stands for the composite encapsulation of silica and polyelectrolytes (PEs). SEM images of L-MSs (b, f) and the resultant MS@QDs (c, g), MS@QDs@SiO2 (d, h), L-QHPs (e, i) during the preparation of L-QHPs at low magnification (b−e), middle magnification (insets), and high magnification (f−i).

fluorescence stability, are essential to improve the encoding capacity. QHPs with different fluorescence intensity levels were prepared through modified layer-by-layer assembly and further encapsulated by a composite shell of silica and polyelectrolytes, as illustrated in Figure 1a. First, instead of importing multicolor fluorophores to improve the encoding capacity as previously reported,13,22 we took advantage of the beads’ inherent property and applied size as the encoding signal, which provided an exponential increase in the encoding capacity. Here, small-sized carboxyl-functionalized magnetic spheres (D = 2.9 μm, denoted as S-MSs) and large-sized carboxylfunctionalized magnetic spheres (D = 6.2 μm, denoted as LMSs) were chosen because they can be easily distinguished by size in FSC channel via flow cytometry. To provide a foundation layer for subsequent QDs assembly, both of the MSs were functionalized with poly(ethylene imine) (PEI) in advance. Second, QDs were then immobilized onto the surface of MSs (MS@QDs) through the coordination interaction with the primary amino group of PEI. After the assembly of the first QD layer, another layer of PEI was absorbed to provide bonding sites for the following QDs assembly. Thus, QD assembly and PEI absorption were repeated several times as depicted above to obtain MS@QDs with higher fluorescence intensity. By adjustment of the QD concentration and the number of assembled QD layers during the modified layer-bylayer assembly process, different QD fluorescence intensity levels can be tuned precisely. Finally, to avoid QD leakage, MS@QDs were primarily encapsulated by a shell of silica (MS@QDs@SiO2) via a condensation reaction of tetraethoxysilane (TEOS). Then another shell of polyelectrolytes (PEs) was formed on the surface of MS@QDs@SiO2 through electrostatic interaction, further protecting the QDs and

offering a soft matrix as well as rich binding sites for the subsequent landing of guest nanoparticles. Therefore, smallsized QD-encoded host particles (S-QHPs) and large-sized QD-encoded host particles (L-QHPs) with different intensity levels were prepared successfully. Figure 1b−i shows the SEM images of the large-sized beads in corresponding stages during the preparation of L-QHPs (kindred SEM images of the small-sized beads were demonstrated in Figure S1). As shown in Figure 1b−e, the original beads and the resultant L-QHPs all possessed a uniform morphology and narrow size distribution with an approximate mean diameter of 6.2 μm, showing no aggregation. The uneven surface with nanoagglomerates of MS@QDs in Figure 1g indicates the successful assembly of QDs on the surface of MSs. In contrast to the surface morphology of MS@ QDs, the surface structure of MS@QDs@SiO2 in Figure 1h presents an additional thick shell, revealing that QDs were successfully encapsulated by silica to be isolated from the outer chemical environment. Moreover, another smooth shell with a silk analogous texture is demonstrated in Figure 1i, proving that the above beads were further sealed by polyelectrolytes, which reinforced the fluorescence stability of QHPs and provided soft landing for guest particles. As shown in Figure 2a, the zeta potential of the magnetic beads turned upside down after PEI functionalization and then declined from 45.6 to 26.3 mV after QDs assembly. The subsequent zigzag trend of the zeta potential proves QDs and PEI were alternatively assembled. The results mentioned above further demonstrate that the foundation layer of PEI and QDs was successfully coated onto the surface of MSs, which is in accordance with the SEM characterization in Figure 1b−i. Moreover, fluorescence excitation and emission spectra of the 10401

DOI: 10.1021/acs.chemmater.7b03811 Chem. Mater. 2017, 29, 10398−10408

Article

Chemistry of Materials

Figure 2. (a) Zeta potential of magnetic spheres at the corresponding assembly stages. (b) Excitation (ex 488 nm, dashed) and emission (em 595 nm, solid) spectra of QDs in chloroform and QHPs in water. (c) Linear relationship between the fluorescence intensity of MS@QDs with different sizes and the QD concentration in CHCl3/C4H9OH mixed solvent. (d) Linear relationship between fluorescence intensity of MS@QDs with different sizes and the number of assembled QD layers. For both S-MSs and L-MSs, the QD concentration used for each layer was the same. (e) Laser-scanning confocal microscopy images of q0−q7 S-QHPs and Q0−Q9 L-QHPs recoded at the hemisphere section of the beads. Images were all taken in channel 2 using the same condition (emission wavelengths ranging from 595 to 625 nm, excited at 488 nm).

the red fluorescence signals of QHPs were dramatically enhanced. Moreover, owing to the addition of QD layers, there was a slight size increase by the comparison among the sizes of q5, q6, and q7 and Q6, Q7, Q8, and Q9, which was also demonstrated by the SEM images of L-QHPs in Figure S2. In addition, the as-synthesized barcodes had excellent dispersibility in aqueous solution, and QDs were evenly distributed on the surface of MSs (see Supporting Information, Figure S3), implying that the assembly of QDs was a stochastic process, which is beneficial for reproducible encoding of QHPs. More importantly, it is shown in Figure 3a−c that the as-synthesized S-QHPs and L-QHPs were distinguished by QD fluorescence intensity since no significant overlap was observed between different peaks, further demonstrating the high identification accuracy when used as barcodes on flow cytometric platform. Compared with previously reported single-color QD barcodes,19,22,23,25,27 this coding performance is considered to be at the leading position, which is ascribed to the high QD fluorescence brightness and the narrow distribution of fluorescence intensity between the obtained barcodes. As verified in Table S2, the coefficient of variations (CVs) of the LQHP fluorescence intensity ranged from 9% to 14%, which was very small considering that the CV of the original MS surface area was 7.42%. Therefore, it is predicted that there were rare imperfections during the QDs assembly process, which is conducive to achieving an accurate and reproducible encoding procedure for the QHPs.

original QDs and the resultant MS@QDs in Figure 2b were nearly the same, implying that the fluorescence properties of QDs were quite stable during the assembly process, which may be attributed to the adoption of the mixed chloroform and nbutanol (v:v = 1:20) reaction solvent. The mixed organic solvent not only avoids damage to the QD properties inflicted by the pure n-butanol38,39 but also helps in driving the QDs onto the surface of MSs since butanol is a relatively poor solvent for QDs.18,19,38,39 As depicted in Figure 2c, the fluorescence intensity of small-sized and large-sized MS@ QDs both increase linearly in parallel with the increasing concentration of QDs. The two linear fitting curves corresponding to S-MS@QDs and L-MS@QDs both have a correlation coefficient of 0.999, indicating high tunability of QD fluorescence intensity. Notably, a certain gap exists between the linear slopes of the two fitting curves, which may be caused by their different surface areas and curvatures of S-MS and L-MS. Similarly, Figure 2d demonstrates the approximate linear increase of MS@QDs fluorescence intensity with the number of QD layers. Therefore, by means of adjusting QD concentration and the number of assembled QD layers during the assembly process, 8 barcodes of S-QHPs (denoted as qx, x = 0−7) and 10 barcodes of L-QHPs (named as Qx, x = 0−9) were prepared successfully (detailed encoding formulas and the resulting fluorescence properties of S-QHPs and L-QHPs are shown in Tables S1 and S2). As shown in Figure 2e, with the applied QD amount increased from Q0 to Q9 (or q0 to q7), 10402

DOI: 10.1021/acs.chemmater.7b03811 Chem. Mater. 2017, 29, 10398−10408

Article

Chemistry of Materials

Figure 3. (a) SEM image of the QHPs (red signals are pseudocolor). Histogram plot for the q0−q7 S-QHPs (b) and Q0−Q9 L-QHPs (c) host codes distinguished by fluorescence intensity via flow cytometry. (d) SEM image of the DMBs showing the host−guest structure (green and red signals are pseudocolor). Histogram plots for f1−f5 (e) and F0−F5 (f) guest codes distinguished by fluorescence intensity via flow cytometry. Confocal images of the q6f5 S-DMBs (g) and Q6F4 L-DMBs (h) recorded at different focalized planes from the top to the center of the cross section through different channels. Images in the first and the second row were observed at channel 2 (from 595 to 625 nm) and channel 1 (from 500 to 530 nm), respectively; third row is the merged image of the first and second row.

characterized in our previous work,13,22 were densely packed on the QHPs through a carbodiimide-assisted coupling reaction, and the obtained dual-color magnetic barcodes (DMBs) possessed a uniform morphology and narrow size distribution (see SEM images in Figure S5a1−b3). The surface coverage13 of guest particles on S-QHPs was 0.711, higher than that of LQHPs (0.607), which was attributed to the larger surface curvature of S-QHP. Here, three kinds of guest nanoparticles, none-fluorescence nanoparticles (NF), weak fluorescence nanoparticles (WF), and strong fluorescence nanoparticles (SF), were first mixed at different ratios and then conjugated onto the surface of the QHPs to achieve different FITC fluorescence intensity levels. Consequently, five guest codes of S-DMBs (f1−f5) and six guest codes of L-DMBs (F0−F5) were controllably obtained (Figure S5c−5d, Tables S3 and S4). As demonstrated in Figure 3e and 3f, the guest codes were distinguished by FITC fluorescence intensity as separate peaks in flow cytometry, indicating their high encoding resolution. It should be noted that the CV of FITC fluorescence intensity of each guest code was mostly less than 20%, demonstrating high controllability of the host−guest assembly. This may be ascribed to the well-defined encoding process and the uniform morphology of building blocks, as testified by SEM images in Figure 3d and Figure S5a1−b3. Furthermore, the resulted guest codes of S-DMBs and L-DMBs were characterized by confocal

In addition, to realize their application in real bioassays, barcodes must remain distinguishable during the storage and detection process. However, QD barcodes usually suffer from QD environmental intolerance due to the polar solvent damage to the cap ligands on the surface of QDs, leading to fluorescence instability, which has been one of the major bottlenecks restraining commercialization and practical use of QD barcodes. To address this issue, QD-encoded beads in this study were further sealed by a composite encapsulation of silica and polyelectrolytes. The fluorescence stability of the barcode beads was significantly improved after composite encapsulation, presenting excellent environmental tolerance to solutions with various kinds of salt and a wide range of pH values (Figure S4a−b). The fluorescence intensities of the QHPs with different intensity levels were almost unchanged during our observation time, i.e., over 110 days, and the mixture of 10 LQHP barcodes (Q0−Q9) remained distinguishable over this time in flow cytometry (Figure S4c−d). Therefore, we conclude that the as-synthesized QHPs have excellent fluorescence stability, assuring their application for bioassays. Preparation of Dual-Color Magnetic Barcodes through Host−Guest Structure. To further enlarge the encoding capacity, the QHPs were combined with FITC-doped guest nanoparticles based on the unique host−guest structure, as displayed in Figure 3d. Guest particles, fabricated and 10403

DOI: 10.1021/acs.chemmater.7b03811 Chem. Mater. 2017, 29, 10398−10408

Article

Chemistry of Materials

Figure 4. Laser-scanning confocal images (ex 488 nm) and the corresponding fluorescence emission spectra (ex 450 nm) of L-DMBs with the same guest particle code (F5) and different QD fluorescence intensity levels (Q0-Q4). Confocal images were taken at both channel 1 (500−530 nm) and channel 2 (595−625 nm); fluorescence intensity at 520 nm in the emission spectra was all normalized.

Figure 5. (a) Scatter plot of the S-DMBs and L-DMBs distinguished by size in the FSC channel via flow cytometry. (b and c) Scatter plot showing the two-dimensional 5 × 8 barcode library of the S-DMBs and 6 × 10 barcode library of the L-DMBs obtained from gate R1 and R2, respectively. (d) Scatter plot showing the three-dimensional barcodes library of the 100 DMBs via flow cytometry. (e) Laser-scanning confocal images of the 100 barcodes recorded at the hemisphere surface.

particles onto the surface of QHPs. As the proportion of WF or SF guest particles increased, the brightness of fluorescent ring reinforced, which well matched with the results measured by

microscopy as shown in Figure S5d. According to these images, a uniform fluorescent ring structure appeared on the beads, showing the effective and well-proportioned assembly of guest 10404

DOI: 10.1021/acs.chemmater.7b03811 Chem. Mater. 2017, 29, 10398−10408

Article

Chemistry of Materials

Figure 6. (a) Schematic of the application of the DMBs in multiplex assays. (b) Scatter plot of the DMB-based five-plexed assays. Concentrations of the tumor markers increased gradually from sample 1 to sample 10. (c) Detection performance plots of AFP, NSE, CEA, CA199, and CYFRA211 (from left to right) in the five-plexed protein assay.

flow cytometry. Not only that, the confocal images of q6f5 DMB (S-DMB comprised of q6 S-QHP and f5 guest code) and Q6F4 DMB (L-DMB comprised of Q6 S-QHP and F4 guest code) in Figure 3g and 3h manifest that the fluorescence signal of QDs and FITC were evenly distributed on the surface of the DMBs, merging into a yellow ring with a red fluorescence signal inside and a green signal outside, which clearly presents the effective union between the different-sized QHPs and the FITC-doped guest particles. To further assess the feasibility of this dual-color encoding strategy, L-QHPs with different fluorescence intensity levels were conjugated with guest particles of a fixed guest code (100% SF, F5). Confocal images and the corresponding emission spectra of the resulting Q0F5-Q9F5 L-DMBs are illustrated in Figure 4. With the amount of QD increased in LQHPs, the red fluorescence signal of confocol image gradually strengthened while the green signal remained unchanged. This phenomenon was also verified by the gradually increasing fluorescence intensity at 593 nm in the emission spectra, manifesting high consistency with the theoretical encoding design. Notably, the fluorescence intensity of Q9F5 at a peak of 593 nm (47.21) was 247 times higher than that of Q1F5 (0.19), indicating that QD fluorescence intensity could be regulated in quite wide range. It suggests that the extraordinary encoding

capacity of QHPs was well kept in the host−guest structure and the process of host−guest assembly was highly tractable. In addition, different L-DMBs made from L-QHPs with different QD fluorescence intensity levels had similar uniform morphology (see Supporting Information, Figure S2), implying the preparation of this well-defined host−guest structure was reproducible. Construction of Three-Dimensional Barcode Library. As a proof of concept, 100 3D barcodes were separately prepared, mixed together in one vial, and distinguished via flow cytometry after further data processing with an OPTICS algorithm.40 It is worth mentioning that only a single 488 nm laser was demanded to excite all 100 barcodes, and the 3D barcodes were decoded by three channels of flow cytometry: the forward-scattering (FSC) channel, the FL1 channel (515/ 20 nm), and the FL2 channel (610/20 nm) correlated to size, FITC fluorescence, and QD fluorescence, respectively. As shown in Figure 5a, the two nonoverlapped scattered clusters were consistent with the SEM images demonstrated before, and the FSC-A intensity of L-DMBs was nearly twice that of SDMBs, indicating that the S-DMBs and L-DMBs can be easily distinguished by size in the FSC channel. This imported size signal, which is dominated by the size of the host beads, will not cause any entanglement of the doping process of fluorophores. 10405

DOI: 10.1021/acs.chemmater.7b03811 Chem. Mater. 2017, 29, 10398−10408

Article

Chemistry of Materials Moreover, no extra laser or detector in the flow cytometry is required for decoding the barcode size, thus minimizing the complexity and cost of the decoding device. Rather than simply analyzing the height of the forward-scattering signal (FSC-H), the area of the forward-scattering signal (FSC-A) was analyzed simultaneously in the decoding process as described in the Methods section. Therefore, detailed size information on the barcodes, such as the width and length, was acquired.41,42 Dimmers, trimmers, or other unidentified debris, which may have similar lengths or widths to the DMBs, will not confuse the 3D scatter plot of the DMBs, thereby ensuring the precision of decoding result. The cluster with lower FSC intensity (gated in R1) was further identified as a two-dimensional plot array with 5 × 8 clusters through the FL1 and FL2 channels (Figure 5b), demonstrating the successful preparation of 40 S-DMBs based on the host−guest structure. Similarly, as shown in Figure 5c, the cluster with higher FSC intensity (gated in R2) was identified as a two-dimensional plot array with 6 × 10 clusters, presenting 60 L-DMBs with distinct coding addresses. Almost all of the clusters exhibited a circular shape rather than an oblique oval shape reported previously,19,24,33−35 which proves that the loading process of the two fluorophores is independent and uncorrelated due to the ingenious design of host−guest structure. It is worth noting that the clusters in the two plot arrays were distributed in horizontal and vertical directions, suggesting that our encoding strategy has a high encoding accuracy and predictability. Nonetheless, the clusters in Figure 5b of S-DMB made from q6 and q7 displayed a slight right skew, indicating more guest particles were immobilized on their surface which lies in the increased surface area of S-QHPs caused by multilayer QDs assembly, whereas the clusters of LDMBs did not show similar phenomena because the multilayerinduced surface area increase was insignificant. By the combinational use of the FSC, FL1, and FL2 channels, a 3D scatter plot with 100 distinctively separated clusters is successfully illustrated in Figure 5d and the video in the Supporting Information. Confocal images of the 100 3D barcodes arranged sequentially according to size, FITC fluorescence, and QD fluorescence are presented in Figure 5e. Both S-DMBs and L-DMBs had narrow size distributions and uniform fluorescent ring-like structures, indicating the outstanding controllability of the encoding process. Furthermore, each image exhibited a unique and recognizable signature. The red signal was enhanced in accordance with an increase of QDs amount in host codes, while the green signal increased in the same variation trend with red signal, which agrees with the results measured via flow cytometry. In addition, resembling QHP, the as-synthesized DMBs exhibited excellent fluorescence stability over 100 days of storage (Figure S6), which facilitates its practical application in multiplex assays. Application of DMBs in Multiplex Assay. The essence of multiplex assay lies in the simultaneous detection of multiple analytes in the same sample. Herein, five typical tumor markers were chosen as target molecules to assess the multiplex biodetection performance of the DMBs serving as typical encoding probes. Correspondingly, DMBs with coding addresses of q3f5, q5f1, Q2F0, Q4F3, and Q6F5 were assigned for AFP, NSE, CEA, CA199, and CYFRA211, respectively. As shown in Figure 6a, barcodes conjugated with captured antibodies, secondary antibodies labeled with the allophycocyanin (APC), and the five target tumor markers were mixed together, forming classical sandwich immunocomplex after 1 h of incubation. The samples were then measured by flow

cytometry and decoded through the FSC (size detector), FL1 (FITC detector), FL2 (QD detector), and FL4 (APC detector) channels. Benefiting from the superparamagnetism of the DMBs, the entire detection process was fast and efficient, and there is every reason to believe that automatic multiplex detection based on our barcodes is easy to achieve in the near future, which would notably speed up the detection process. Here, 10 samples of 5 tumor markers with different concentrations were detected (detailed concentrations in each sample are listed in Table S5). In each case, all five barcodes and five tumor markers were present in a single vial, thus confirming the multiplexing capability of DMBs. The detection results in Figure 6b and 6c show that the five tumor markers were clearly identified by the specific barcodes. Although for five different barcodes, the amounts of antibodies attached onto the beads are quite different, which should be attributed to the different barcode’s size as well as the different physical or chemical properties of antibodies that result in the different coupling efficiency. As the concentration of the five tumor markers gradually increased, the five clusters shifted from left to right along the FL4 channel, suggesting a close linear correlation between the concentration of the tumor marker and the APC fluorescence intensity within a wide detection range. These results conclusively demonstrate the huge potential of the DMBs in multiplexing assay to specifically and simultaneously recognize and quantify different target molecules in the same vial. In addition, the fluorescence properties of the DMBs were almost unchanged during the bioconjugation and biodetection procedures (Table S6), which admirably guarantees the precision of the multiplex detection.



CONCLUSION In summary, we developed a 3D encoding strategy to fabricate high-performance barcodes with an extraordinarily high encoding capacity in a splendid accurate and reproducible manner. The three dimensions refer to size, emission wavelength, and fluorescence intensity of the barcodes, which could help enlarge the encoding capacity with ease. As a result, 100 distinguishable barcodes in flow cytometry were achieved controllably, firmly demonstrating the success of this encoding strategy. Meanwhile, the DMB-based five-plexed protein detection was also studied, and the satisfactory results verify the feasibility and effectiveness of DMBs in multiplex detection. Furthermore, the encoding capacity could be exponentially increased by utilizing MSs with other sizes, employing more types of fluorophores or importing other new encoding elements into the inner space of the host particles. Therefore, this 3D encoding strategy paves a new way to fabricate enormous barcodes, which exhibit immense potential in highthroughput multiplex assays.



ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.chemmater.7b03811. SEM images, laser-scanning confocal microscopy images, encoding formula of DMBs, fluorescence tolerance test, concentrations of the five tumor markers during fiveplexed detection (PDF) Video of 3D barcode library (AVI) 10406

DOI: 10.1021/acs.chemmater.7b03811 Chem. Mater. 2017, 29, 10398−10408

Article

Chemistry of Materials



(14) Theilacker, N.; Roller, E. E.; Barbee, K. D.; Franzreb, M.; Huang, X. Multiplexed protein analysis using encoded antibodyconjugated microbeads. J. R. Soc., Interface 2011, 8, 1104−1113. (15) Lachmann, N.; Todorova, K.; Schulze, H.; Schönemann, C. Luminex® and Its Applications for Solid Organ Transplantation, Hematopoietic Stem Cell Transplantation, and Transfusion. Transfus. Med. Hemotherapy. 2013, 40, 182−189. (16) Houser, B. Bio-Rad’s Bio-Plex® suspension array system, xMAP technology overview. Arch. Physiol. Biochem. 2012, 118, 192−196. (17) Dunbar, S. a. Applications of Luminex xMAP technology for rapid, high-throughput multiplexed nucleic acid detection. Clin. Chim. Acta 2006, 363, 71−82. (18) Han, M.; Gao, X.; Su, J.; Nie, S. Quantum-dot-tagged microbeads for multiplexed optical coding of biomolecules. Nat. Biotechnol. 2001, 19, 631−635. (19) Gao, X.; Nie, S. Quantum dot-encoded mesoporous beads with high brightness and uniformity: rapid readout using flow cytometry. Anal. Chem. 2004, 76, 2406−2410. (20) Gao, X.; Nie, S. Doping mesoporous materials with multicolor quantum dots. J. Phys. Chem. B 2003, 107, 11575−11578. (21) Cheng, Y.; Zhao, Y.; Shangguan, F.; Ye, B.; Li, T.; Gu, Z. Convenient generation of quantum dot-incorporated photonic crystal beads for multiplex bioassays. J. Biomed. Nanotechnol. 2014, 10, 760− 766. (22) Zhang, D.; Jiang, Y.; Yang, H.; Zhu, Y.; Zhang, S.; Zhu, Y.; Wei, D.; Lin, Y.; Wang, P.; Fu, Q.; Xu, H.; Gu, H. Dual-Encoded Microbeads through a Host−Guest Structure: Enormous, Flexible, and Accurate Barcodes for Multiplexed Assays. Adv. Funct. Mater. 2016, 26, 6146−6157. (23) Fournier-Bidoz, S.; Jennings, T. L.; Klostranec, J. M.; Fung, W.; Rhee, A.; Li, D.; Chan, W. C. W. Facile and rapid one-step mass preparation of quantum-dot barcodes. Angew. Chem., Int. Ed. 2008, 47, 5577−5581. (24) Leng, Y.; Wu, W.; Li, L.; Lin, K.; Sun, K.; Chen, X.; Li, W. Magnetic/Fluorescent Barcodes Based on Cadmium-Free NearInfrared-Emitting Quantum Dots for Multiplexed Detection. Adv. Funct. Mater. 2016, 26, 7581−7589. (25) Hu, J.; Wen, C.-Y.; Zhang, Z.-L.; Xie, M.; Hu, J.; Wu, M.; Pang, D.-W. Optically encoded multifunctional nanospheres for one-pot separation and detection of multiplex DNA sequences. Anal. Chem. 2013, 85, 11929−11935. (26) Wang, Z.; Zong, S.; Li, W.; Wang, C.; Xu, S.; Chen, H.; Cui, Y. SERS-fluorescence joint spectral encoding using organic−metal−QD hybrid nanoparticles with a huge encoding capacity for highthroughput biodetection: putting theory into practice. J. Am. Chem. Soc. 2012, 134, 2993−3000. (27) Wang, X.; Wang, G.; Li, W.; Zhao, B.; Xing, B.; Leng, Y.; Dou, H.; Sun, K.; Shen, L.; Yuan, X.; Li, J.; Sun, K.; Han, J.; Xiao, H.; Li, Y.; Huang, P.; Chen, X. NIR emitting quantum dot-encoded microbeads through membrane emulsification for multiplexed immunoassays. Small 2013, 9, 3327−3335. (28) Bera, D.; Qian, L.; Tseng, T.-K.; Holloway, P. H. Quantum dots and their multimodal applications: a review. Materials 2010, 3, 2260− 2345. (29) Gao, X.; Chan, W. C. W.; Nie, S. Quantum-dot nanocrystals for ultrasensitive biological labeling and multicolor optical encoding. J. Biomed. Opt. 2002, 7, 532−537. (30) Alivisatos, P. The use of nanocrystals in biological detection. Nat. Biotechnol. 2004, 22, 47−52. (31) Petryayeva, E.; Algar, W. R. Multiplexed homogeneous assays of proteolytic activity using a smartphone and quantum dots. Anal. Chem. 2014, 86, 3195−3202. (32) Liu, H.; Li, G.; Sun, X.; He, Y.; Sun, S.; Ma, H. Microfluidic generation of uniform quantum dot-encoded microbeads by gelation of alginate. RSC Adv. 2015, 5, 62706−62712. (33) Kim, J.; Biondi, M. J.; Feld, J. J.; Chan, W. C. W. Clinical validation of quantum dot barcode diagnostic technology. ACS Nano 2016, 10, 4742−4753.

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. ORCID

Xunbin Wei: 0000-0002-4269-9943 Hong Xu: 0000-0002-2787-5806 Author Contributions

The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript. Funding

This work was supported by The Key Research and Development Program of Zhejiang province (2017C03005), SJTU-JOINSTAR Joint Lab Innovation Project, and SJTU funding (YG2015ZD11). Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS The authors acknowledge Junxi Gu for drawing the schemes and Yi Chen for SEM characterization.



REFERENCES

(1) Sukhanova, A.; Nabiev, I. Fluorescent Nanocrystal-Encoded Microbeads for Multiplexed Cancer Imaging and Diagnosis. Crit. Rev. Oncol. Hematol. 2008, 68, 39−59. (2) Wilson, R.; Cossins, A. R.; Spiller, D. G. Encoded microcarriers for high-throughput multiplexed detection. Angew. Chem., Int. Ed. 2006, 45, 6104−6117. (3) Jun, B. H.; Kang, H.; Lee, Y. S.; Jeong, D. H. Fluorescence-Based Multiplex Protein Detection Using Optically Encoded Microbeads. Molecules 2012, 17, 2474−2490. (4) Birtwell, S. W.; Morgan, H. Microparticle encoding technologies for high-throughput multiplexed suspension assays. IFMBE Proc. 2010, 27, 316−319. (5) Vignali, D. a a. Multiplexed particle-based flow cytometric assays. J. Immunol. Methods 2000, 243, 243−255. (6) Nam, J. M.; Thaxton, C. S.; Mirkin, C. A. Nanoparticle-Based Bio-Bar Codes for the Ultrasensitive Detection of Proteins. Science 2003, 301, 1884−1886. (7) Nicewarner-peña, A. S. R.; Freeman, R. G.; Reiss, B. D.; He, L.; Peña, D. J.; Natan, M. J. Submicrometer metallic barcodes. Science 2001, 294, 137−141. (8) Dejneka, M. J.; Streltsov, A.; Pal, S.; Frutos, A. G.; Yost, K.; Yuen, P. K.; muller, U.; Lahiri, J. Rare earth-doped glass microbarcodes. Proc. Natl. Acad. Sci. U. S. A. 2003, 100, 389−393. (9) Brazhnik, K.; Grinevich, R.; Efimov, A. E.; Nabiev, I.; Nabiev, I. Sukhanova. Development and potential applications of microarrays based on fluorescent nanocrystal-encoded beads for multiplexed cancer diagnostics. Proc. SPIE 2014, 9129, 91292C−1. (10) Wang, G.; Leng, Y.; Dou, H.; Wang, L.; Li, W.; Wang, X.; Sun, K.; Han, J.; Xiao, J.; Li, Y. Highly efficient preparation of multiscaled quantum dot barcodes for multiplexed hepatitis B detection. ACS Nano 2013, 7, 471−481. (11) Wang, H.-Q.; Liu, T.-C.; Cao, Y.-C.; Huang, Z.-L.; Wang, J.-H.; Li, X.-Q.; Zhao, Y.-D. A flow cytometric assay technology based on quantum dots-encoded beads. Anal. Chim. Acta 2006, 580, 18−23. (12) Lawrie, G. a.; Battersby, B. J.; Trau, M. Synthesis of Optically Complex Core−Shell Colloidal Suspensions: Pathways to Multiplexed Biological Screening. Adv. Funct. Mater. 2003, 13, 887−896. (13) Zhu, Y.; Xu, H.; Chen, K.; Fu, J.; Gu, H. Encoding through the host-guest structure: construction of multiplexed fluorescent beads. Chem. Commun. 2014, 50, 14041−14044. 10407

DOI: 10.1021/acs.chemmater.7b03811 Chem. Mater. 2017, 29, 10398−10408

Article

Chemistry of Materials (34) Liu, H.; Qian, X.; Wu, Z.; Yang, R.; Sun, S.; Ma, H. Microfluidic synthesis of QD-encoded PEGDA microspheres for suspension assay. J. Mater. Chem. B 2016, 4, 482−488. (35) Giri, S.; Sykes, E. a; Jennings, T. L.; Chan, W. C. W. Rapid screening of genetic biomarkers of infectious agents using quantum dot barcodes. ACS Nano 2011, 5, 1580−1587. (36) Ming, K.; Kim, J.; Biondi, M. J.; Syed, A.; Chen, K.; Lam, A.; Ostrowski, M.; Rebbapragada, A.; Feld, J. J.; Chan, W. C. W. Integrated quantum dot barcode smartphone optical device for wireless multiplexed diagnosis of infected patients. ACS Nano 2015, 9, 3060−3074. (37) Zhao, Y.; Shum, H.; Chen, H.; Adams, L.; Gu, Z.; Weitz, D. Microfluidic Generation of Multifunctional Quantum Dot Barcode Particles. J. Am. Chem. Soc. 2011, 133, 8790−8793. (38) Wang, G.; Zhang, P.; Dou, H.; Li, W.; Sun, K.; He, X.; Han, J.; Xiao, H.; Li, Y. Efficient incorporation of quantum dots into porous microspheres through a solvent-evaporation approach. Langmuir 2012, 28, 6141−6150. (39) Wilson, R.; Spiller, D. G.; Prior, I. a; Veltkamp, K. J.; Hutchinson, A. A simple method for preparing spectrally encoded magnetic beads for multiplexed detection. ACS Nano 2007, 1, 487− 493. (40) Ankerst, M.; Breunig, M. M.; Kriegel, H. P.; Sander, J. Optics: Ordering points to identify the clustering structure. Proceedings of the 1999 ACM SIGMOD international conference on management of data, New York; 1999; pp 49−60. (41) Bouvier, T.; Troussellier, M.; Anzil, A.; Courties, C.; Servais, P. Using light scatter signal to estimate bacterial biovolume by flow cytometry. Cytometry 2001, 44, 188−194. (42) Latimer, P. Light scattering vs. microscopy for measuring average cell size and shape. Biophys. J. 1979, 27, 117−126.

10408

DOI: 10.1021/acs.chemmater.7b03811 Chem. Mater. 2017, 29, 10398−10408