1. Introduction

Digital holographic microscopy (DHM) is a combination of digital holography and microscopic imaging technology that provides surface information including intensity and quantitative phase information for samples, acquiring the refractive index or thickness of transparent samples, as well as the surface morphology of scattering samples [1–3]. Based on above properties, it is widely used in the field of biomedicine for the study of cell morphology and cell dynamic monitoring [4–7]. Since the digital holography technology has the ability to obtain the thickness of sample correctly [8,9], micro-quartz pieces (MQPs) with different thicknesses, instead of the conventional microbeads, can be used as the encoding carriers of suspension array (SA), achieving the accurate encoding and decoding analysis.

Over the past decades, SA has developed as one of the most promising technologies for high-throughput multiplexed bioassays [10–12]. Compared with the traditional analytical methods, like the planar biochip, SA is more suitable for simultaneous detection of multiple targets in rare samples due to its better biocompatibility and higher sensitivity [13,14]. Currently, several kinds of encoding methods have been developed, including graphical, physical, optical techniques, and even combinational encoding ways that integrate different kinds of signal [15–19]. While the microbead-based optical encoding method, especially encoded with fluorescence materials, is most widely used for its flexible encoding ability [20–22]. For the decoding optical system, a multi-color flow cytometer is utilized to acquire the quantitative information of label signal [23]. Nevertheless, some restrictions, including overlap of fluorescence spectra [18], crosstalk between encoding signal and label signal [24] and also the complicated decoding system [25], provide poor stability and complex process for fluorescence encoding method. In our previous work, a dual-wavelength digital holographic phase and fluorescence microscopy (DW-DHPFM) system was developed for the MQPs-based optical thickness encoding method [26]. With the economical and stable coding scheme, DHM provides an encoding channel which is independent with the fluorescent label, avoiding the crosstalk during immunoassays. Fluorescence microscopy enables a focal plane imaging manner to quantify the fluorescence materials-labeled analytes with the advantage of effectiveness and convenience. Therefore, qualitative and quantitative analysis of targets can be achieved simultaneously with the DW-DHPFM system. Considering the manufacturing technique of the MQPs and the measurement error of the optical system, the method is limited in the number of encoding. In order to meet the increasing requirements of detecting multiple targets from one rare sample at the same time, multiplexed detection with high accuracy is necessary to develop, which can be realized by finding a new encoding channel and compositing with the optical thickness encoding channel. For achieving the combination of two encoding channels, it should be indicated that the new one should has the above encoding advantages and would not affect the accuracy of optical thickness decoding.

Raman spectrum, as a fingerprint spectrum, is specific for each material [27–29]. With the development of laser technology and the discovery of enhanced Raman signal caused by surface plasmon resonance effect of noble metal nanoparticles, surface enhanced Raman scattering (SERS) has been highly valued and gradually popularized [19,30]. Since materials with different functional groups have various characteristic peaks, SERS spectra have the encoding potential [31–33]. Compared with the fluorescence encoding method, which has the spectral overlap problem [24,34–36], the narrow bandwidths of SERS spectra provide more accurate signal and encoding space [18,37], and the number of codes increases exponentially with the addition of encoding material types [18]. Similarly, the encoding channel is independent with the labeled fluorescence signal, and such property makes it no need to suffer from the crosstalk between encoding and label signal. In order to obtain the enhanced Raman signal, Raman reporter molecules are combined with gold nanorods (AuNRs) by Au-S bond [38,39]. For the combination of two channels, the bonded AuNRs and Raman reporter molecules (AuNRs-Ra), decorated with positive charges, can be easily assembled on the negative charges decorated MQPs. Since the AuNRs are nanometer-sized and dispersed evenly, this assembly method would not affect the thickness detecting of MQPs with micron-size physical thickness.

In this paper, DW-DHPFM integrating Raman spectroscopy is introduced for the detection and analysis to biomolecules. Considering that the digital holography-based optical thickness encoding channel and the SERS spectra encoding channel both have good stability and do not interfere with the labeled fluorescence signal, this dual-channel encoded method has the advantages of high accuracy and no crosstalk. By assembling Raman reporter molecules on the surface of MQPs, the combination of two encoding channels is achieved with a considerable encoding capacity. Furthermore, this method breaks the limitation of the traditional fluorescence encoding method which utilizes the microbeads as carriers. After being captured by SA, the fluorescence-labeled targets are analysed with fluorescence imaging method for providing quantitative information, which simplifies the operation complexity and reduces the detecting expenses.

2. Materials and methods

2.1 Optical system

The schematic of DW-DHPFM system is shown in Fig. 1(a). The phase images are obtained by a dual-wavelength digital holographic phase microscopy (DW-DHPM), and the fluorescence images are acquired from the fluorescence microscope (FM). For the phase imaging path, the DW-DHPM system uses a dual-wavelength mode to achieve a large axial measurement range. The two wavelengths, λ₁ = 830.0 nm and λ₂ = 833.4 nm, are attained from a Superluminescent Diodes (SLD, Inphenix, center wavelength 844 nm, 3 dB bandwidth 50.8 nm, 8 mW) using two laser line filters (Semrock; LLF1, LL01-830; LLF2, LL01-852) with different angle. Specifically, the wavelength of 833.4 nm is achieved with leaning LLF2 at a certain inclination to the vertical. The incident light from the SLD is filtered and divided into sample arm and reference arm by a beam splitter (BS). A linear polarizer (LP) and quarter-wave plate (QWP) are used to maximize the contrast of hologram. Two optical delay lines (ODLs) are designed to adjust the difference of optical path length between the reference arm and the sample arm. In the reference arm, after expanding the light beam, the reference light is adjusted by three mirrors to achieve the off-axis DHM. This arrangement only requires one hologram for each operating wavelength, and the complex amplitude of the sample light can be retrieved.

 

Fig. 1 (a) Schematic of the DW-DHPFM system. SLD, superluminescent diode; SMF, single-mode fiber; MMF, multi-mode fiber; CL, collimator; LLF, laser line filter; LP, linear polarizer; BS, beam splitter; QWP, quarter-wave plate; ODL, optical delay line; OL, objective lens; DM, dichroic mirror; LF, long-pass filter; SF, short-pass filter; TL, tube lens. (b) Schematic of the confocal Raman microscope. NF, notch filter; (c) The synthesis and detection scheme of dual-channel encoded SA.

Download Full Size | PDF

In the sample arm, to eliminate the spherical aberration caused by the objective lens (OL) [OL1, Olympus, MPlan N 5 × ], it is necessary to adjust the back focal plane of OL1 to coincide with the front focal plane of the tube lens (TL) [TL1, f = 125 mm]. Eventually, the reference light and the sample light interfere on the receiving surface and captured by a monochrome charge-coupled device (CCD) [CCD1, ZWO, ASI174MM]. While collecting data, switch the two filters to collect the hologram with each wavelength.

For the fluorescence imaging path, a 405 nm laser is used as an excitation source. After reflected by a dichroic mirror (DM) [DM2, Semrock, FF484-FDi01], the incident light is defocused on the sample, and the sample is uniformly excited to achieve the fluorescence imaging of the whole plane. Then the emitted fluorescent signal penetrates the DM2 and reflected by the DM1 (Daheng Optics, GCC-414008). Further passing through a long pass filter (LF, Thorlabs, FELH0500), a short pass filter (SF, Daheng Optics, GCC-211002)) and TL2 (f = 100 mm), the target signal is finally recorded by a color CCD (CCD2, ZWO, ASI071MC).

For the Raman spectroscopy decoding, the schematic of optical system is shown in Fig. 1(b). A CMOS camera (CMOS, IDS, UI-1460LE-C-BG) is utilized in this system for monitoring the target samples, ensuring a correspondence between Raman signal and thickness information for each MQP. A 785 nm laser, as the excitation light, is reflected by DM and focused on the sample through the OL. Then the scattered light is spread out by spectrometer at different spatial positions, and the resulting Raman spectrum is recorded by the CCD (CCD3, HORIBA JOBIN YVON, Synapse CCD).

2.2 Materials

Cetyltrimethylammonium bromide (CTAB) and hydrogen tetrachloroaurate (III) hydrate (HAuCl₄·3H₂O) were purchased from Sigma-Aldrich. Sodium borohydride (NaBH₄) was purchased from Damao Chemical Reagent Factory (Tianjin). Silver nitrate (AgNO₃) was purchased from Sinopharm Chemical Reagent Co., Ltd. Ascorbic acid (AA) was purchased from Aladdin Biochemical Technology Co., Ltd. (shanghai, China). (3-Aminopropyl)triethoxysilane (APTES) was purchased from Shanghai Macklin Biochemical Co., Ltd (shanghai, China). Polyethyleneimine (PEI), poly(sodium-pstyrenesulfonate) (PSS), 2-Naphthalenethiol (2-NT), 4-Chlorothiophenol (4-CBT), 4-Mercaptophenol (4-HBT) and 5,5′-Dithio bis-(2-nitrobenzoic acid) (DTNB), glutaraldehyde were purchased from Aladdin Industrial Corporation, China. Phosphate buffer saline (PBS) was purchased from Beijing Solarbio Science & Technology Co., Ltd. Sodium chloride (NaCl) was purchased from Tianjin Zhiyuan Reagent Co., Ltd. Non-protein blocking solution was purchased from Beijing Bai’aolaibo Technological Development Corporation, China. Three types of IgGs (mouse, rabbit and human) were purchased from Bioss Biotechnology, China. Three types of QDs-labeled anti-IgGs (565 nm QDs-labeled goat anti mouse IgG, 585 nm QDs-abeled goat anti rabbit IgG and 610 nm QDs-labeled goat anti human IgG) were purchased from Hangzhou NajingTech Co., Ltd, China.

2.3 Synthesis and detection principle of dual-channel encoded SA

Based on DW-DHPFM and Raman spectroscopy, we propose a new dual-channel encoded SA analytical method. The SAs are prepared with bio-probes grafted dual-channel encoded microcarriers (DCEMs), which are attained by assembling Raman reporter molecules on the surface of the MQPs. Then the confocal Raman microscope and home-built DW-DHPFM system are used to provide qualitative and quantitative information of the target analytes, achieving the multiplexed detection to biomolecules. Figure 1 shows the scheme of synthesis, detection, decoding and analysis of our dual-channel encoded SA. Specifically, the synthesis and detection principle of SA are shown in Fig. 1(c).

Firstly, Raman reporter molecules are bonded with AuNRs as the AuNRs-Ra, followed by the modification of APTES solution (AuNRs-Ra@APTES) for providing positive charges. Meanwhile, MQPs are decorated with negative charges by using PEI and PSS solution sequentially. Then, the AuNRs-Ra@APTES are assembled on the surface of above MQPs by electrostatic adsorption, and the glutaraldehyde solution is used for encapsulation and probe grafting. After that, the excess groups that not grafted with the probes are covered by non-protein blocking solution. With the above process, the two-channel encoded SAs are achieved and can be used to detect the QDs-labeled analytes. Because of the specific target analytes are only captured by their corresponding probe-grafted SAs, multiple kinds of analytes can be detected simultaneously. After the multiplexed immunoassay experiment, the sample is prepared on the glass slide for decoding and analysis.

As an analysis platform, the fluorescence imaging path [Fig. 1(a)] is used to acquire the quantitative information of the target analytes, and the phase imaging path [Fig. 1(a)] is used to reconstruct the phase map of reacted SAs, thereby obtaining the decoding optical thickness. The other decoding information, SERS spectrum of the same SA is provided by the confocal Raman microscope [Fig. 1(b)].

2.4 Preparation of the DCEMs

In our experiments, MQPs with a physical size of 100 μm × 100 μm and thickness range of approximately 30 - 70 μm were used for optical thickness encoding. The MQPs (5000 pieces for each type) were subjected to washing with acetone and ethanol repeatedly, and immersed in piranha solution for 12 h to sufficiently clean. Hydroxyl groups were also modified on the surface of MQPs with such step. After that, it was washed with deionized water for five times and dispersed in deionized water. Then the MQPs were layer-by-layer self-assembled and finally with the negative charges on its surface, which is used for the assembling of Raman reporter molecules. The hydroxyl-modified MQPs were dispersed in 1.5 mL PEI solution (4 mg/mL, 0.5 M NaCl), stirred for 1.5 h, and washed with deionized water for three times. Next, the above MQPs were dispersed into 1.5 mL of PSS solution (2 mg/mL, 0.5 M NaCl), stirred for 1.5 h, followed by washing with deionized for three times and dispersed in deionized water.

In order to improve the encoding performance with Raman spectrum, SERS spectrum with high signal-to-noise ratio (SNR) is obtained by using AuNRs. Firstly, AuNRs were prepared with the seed growth method. Then, the Raman reporter molecules with characteristic peaks were combined with AuNRs thorough the Au-S bond. After that, AuNRs-Ra were decorated with positive charges for assembling with the surface of MQPs. To the preparation of AuNRs, seeds were prepared at first. 4 mL CTAB (0.1M) was added to the round bottom flask, stirred for 2 min. Then added 40 μL of 24.28 mM HAuCl₄ solution and stirred for 2 min. After that, 24 μL of 0.1 M fresh NaBH₄ solution prepared with ice water was added quickly. The disperse solution were high-speed stirred for 90 s, followed by standing at room temperature for 2-5 h. Next is the preparation of AuNRs. 20 mL of 0.2 M CTAB solution was took into a round bottom flask and added 400 μL of 24.28 mM HAuCl₄ solution, 150 μL of 5 M HCl, 60 μL of 40 mM AgNO₃ solution, 160 μL of 0.1 M ascorbic acid solution and 28 μL of the seed solution prepared above gradually at intervals of 2 min. The reacted solution was allowed to stand at 30°C for 12 h, then the resulting solution was centrifuged at 8500 rpm for 10 minutes to remove excess CTAB on the surface of the AuNRs, and the finally obtained AuNRs were redispersed in ethanol. In the second step, 100 mg of Raman reporter molecules were added to 2 mL of ethanol and sonicated for 10 min. In the third step, the above solution was added to the AuNRs dispersed ethanol solution, stirred for 30 min, and then left to incubate for 24 h at the room temperature. After that, the excess Raman report molecules were removed by centrifugation (4000 rpm) with ethanol for twice, and the AuNRs-Ra were dispersed in 30% (v/v) APTES ethanol solution for 1.5 h. After the surface modifying, the excess APTES was washed with ethanol and deionized water respectively, and the precipitate was redispersed in deionized water.

The negative charges modified MQPs and the above-mentioned AuNRs-Ra@APTES were mixed in an aqueous solution and stirred at room temperature for 3 h. To remove the unbonded AuNRs-Ra@APTES, the reacted MQPs were washed with deionized water for 3 times. Then the precipitate was dispersed in 5% (v/v) glutaraldehyde PBS solution (0.01 M, pH 7.4) for 1.5 h, followed by washing with PBS. Until that, the DCEMs were prepared.

2.5 Probe grafting and protein detection

The above DCEMs were dispersed in 50 μL IgG solution (1 mg/mL), incubated for 2 h at 37°C. After that, it was washed with PBS for several times and the unreacted groups were blocked with the non-protein blocking solution at 4°C for 8 h, thus the dual-channel encoded SAs were obtained. The QDs-labeled anti-IgG was regarded as the target analyte, incubated with the SAs for 1 h at 37°C. To remove the unreacted QDs-labeled anti-IgG, the SAs were washed with PBS for three times, and then dispersed in PBS solution for the further decoding and analysis.

For the proof-of-concept experiment of the dual-channel encoded SA in multiplexed immunoassays, three kinds of DCEMs grafted with three different IgG bio-probes were prepared as follows: (1) SA1: DCEM1 (80.37 ± 1.45 μm, DTNB) grafted with mouse IgG; (2) SA2: DCEM2 (69.98 ± 1.54 μm, 2-NT) grafted with rabbit IgG; (3) SA3: DCEM3 (52.10 ± 0.87 μm, DTNB + 4-CBT) grafted with human IgG. The three kinds of IgG-grafted SAs were used to capture the corresponding QDs-labeled anti-IgGs in a mixed solution.

For the quantitative analysis, 565 nm QDs-labeled goat anti mouse IgG solutions with concentrations of 30, 15, 7.5, 3.75, 1.875 nM, and 937.5, 468.75 and 0 pM were tested as analytes to demonstrate the response of SA3.

2.6 Characterization

Scanning electron microscopy (SEM) was performed on a ZEISS SUPRA 55, which was operated at an accelerating voltage of 3 kV. Samples were dispersed in silicon slices and sprayed with platinum. Transmission electron microscope (TEM) image was obtained with a FEI spirit T12 TEM. Zeta potential measurements were performed on Zetasizer Nano ZS90. Using 785 nm laser excitation (7.8mW at the sample position), Raman spectra were attained by a confocal Raman microscope (HORIBA LabRAM HR800) with a 50X microscope objective. The acquisition time was 5s with 2 repetitions per spectrum. The phase and fluorescence images were acquired by a home-built DW-DHPFM system.

3. Results and discussion

3.1 Theory of dual-wavelength linear regression phase unwrapping

For the optical thickness encoding method, we apply the dual-wavelength linear regression phase unwrapping to acquire the thickness information of encoded microcarriers. The optical thickness of the sample can be described by the optical path difference map $D\left(x,y\right)$, where $D\text{=}n\cdot T$, $n$ is refractive index and $T$ is the physical thickness of sample. The optical path difference map is determined by the phase of holographic reconstruction after the light passing through the sample, expressed as:

If the variation of optical path difference after passing through the sample is greater than $\lambda$, the phase map of the sample becomes blurred and contains a discontinuous point of $2\pi$, so the unwrapping process is required. For decades, many software algorithms are used to unwrap phase, but most of them are computationally complex, and cannot handle the complicated topologies such as porous materials [40]. Even a relatively mature algorithm software for phase unwrapping might misinterpret areas with weaker light intensity as multiple phase step points and produce unnecessary errors. Since the discontinuities of $2\pi$ are unlikely to occur at the same location in phase map with different wavelengths, multi-wavelength techniques are developed to unwrap phase [41–43], which is based on the comparation of phase maps with different wavelengths.

Assuming the sample is imaged by two different wavelengths and both of them are less than the optical thickness of the sample, the optical path difference map can be expressed as:

Since the final phase map is obtained by subtracting the phase map from two different wavelengths, when the result is negative, $2\pi$ could be added, then a new sample phase map corresponding to the beat wavelength of $\Lambda$ is obtained. However, this method increases the phase noise even though it improves the measurable range of phase [41]. For overcoming the above shortcomings, another phase unwrapping method is developed [44]. Supposing that $m_1$ could be expressed by an integer $m_2$, then $m_1$ can be rewritten according to Eq. (2) as:

As the Eq. (4) shows, theoretically, the linear equation of the integer $m_1$ expressed by the integer $m_2$ can be accurately solved without considering the measurement uncertainty of phase and wavelength. Moreover, excepting that there is a common divisor between the two wavelengths (for example, the wavelengths are 600 and 700 nm), the measurement range of optical path difference in this method will be extended to infinity, because only one set of integers $m_1$, $m_2$ can meet the Eq. (4). For the practical phase measurements, the linear equation of $m_1$ with a variable of $m_2$ can be rounded to determine the most probable integers of $m_1$ and $m_2$. It can be seen that the Eq. (3) is a special case of Eq. (4) where the difference between $m_1$ and $m_2$ is equal to 0 or 1. However, under normal conditions, the optical path difference caused by the sample is different from the beat wavelength, and it is not necessary to set the limit for the value of $m_2$. The range of values for $m_2$ can be set according to the optical path difference caused by the sample, then the wrapped phase map is observed to count the number of phase jumps, estimating the optical thickness of the sample. Different from the previous method, by finding the value of $m_2$ where $m_1$ is closest to the integer, we can ascertain the correct number of wavelengths and recreate the optical thickness of the original sample, which avoids the excessive amplification of phase noise.

In this research, two wavelengths of 830.0 nm and 833.4 nm are selected as ${\lambda }_1$ and ${\lambda }_2$. According to the principle of dual-wavelength synthesis, the beat wavelength is calculated as 203.4 μm, which can be regarded as the axial measurement range of our DW-DHPM system.

3.2 Decoding performance of DW-DHPFM system to MQPs

To exhibit the property of our proposed DW-DHPFM system, three kinds of different optical thicknesses encoded MQPs are cleaned and then mixed together as the sample for optical thickness imaging, the results from the DW-DHPFM system are showed in Fig. 2. Figure 2(a) is the original hologram acquired with the wavelength of 830 nm with high spatial carrier frequency. By amplifying the red square area of Fig. 2(a), the interference fringe can be clearly seen as shown in Fig. 2(b). The corresponding hologram obtained with the wavelength of 833.4 nm is similar to Fig. 2(a). For phase reconstruction process, owing to the effects of phase differences, blank region without the MQPs is imaged firstly as the aberration data to each wavelengths. Then the aberration data is subtracted from the phase image of sample area to obtain an aberration-compensated phase image of each wavelength. According to the theory of angular spectrum reconstruction, the phase maps of the sample corresponding to the two wavelengths are reconstructed respectively as shown in Figs. 2(c) and 2(d). Finally, according to the algorithm of two-wavelength linear regression phase wrapping introduced above, the optical thickness image of the MQPs is obtained as shown in Fig. 2(e). The distribution of optical thickness at the white line is shown in Fig. 2(h). Three different MQPs with the optical thickness values of 80.37 ± 1.45 μm, 69.98 ± 1.54 μm and 52.10 ± 0.87 μm are corresponding to the three different color in Fig. 2(e). It should be noted that the final value of optical thickness is equal to the mean value of the corresponding pixels (excluding edge pixels of each MQP), hence the existence of little noise as shown in Fig. 2(e) would not affect the accuracy of results.

 

Fig. 2 Dual-wavelength reconstruction process using DW-DHPFM system. (a) A hologram captured at 830 nm. (b) The red square area of (a). (c), (d) Reconstructed phase images of 830 and 833.4 nm respectively after compensating the aberration. (e) Reconstructed optical thickness image of MQPs and (f) cross-sectional profile of the white line in (e). Size of field of view of (a), (c)–(e): 1.26 mm × 1.26 mm.

Download Full Size | PDF

Through the above experiments, we have confirmed that different optical thicknesses of MQPs can be distinguished exactly by our DW-DHPFM system. Since the number of distinguishable MQPs in the system determines the encoding capacity, we need to test the axial resolution of our system and the standard deviation of multiple MQPs from the same optical thickness. First of all, for the axial resolution test, we chose a clean glass slide as the sample, the optical thickness distribution of the single layer interface is obtained according to the above data processing flow. In the edge portion of the hologram, due to the weak coherence of the broadband light, contrast of the interference fringe is not good enough to reconstruct high quality images, resulting the optical thickness profile obtained by synthesizing phase maps with two wavelengths is relatively noisy in these regions. We calculated the standard deviation of the region 1 mm × 1 mm in the center of the image and obtained the axial resolution of the system as 0.53 μm.

Then, the dual-wavelength phase imaging process is performed to the MQPs from the same optical thickness, and the optical thickness distribution of MQPs with 100 pieces is calculated. After the statistics of the above three optical thicknesses of MQPs, the standard deviations distribute in the range of 0.87~1.54μm, that is, the nonuniformity makes the MQPs nominally from the same optical thickness distributes in a range in fact. The results confirmed that the limitation of decoding system lies in the manufacturing techniques of MQPs. Meanwhile, the axial dimension of sample is required no more than 203.4 μm with the principle of dual-wavelength synthesis. Therefore, the number of codes that can be achieved is around only 100. Compared with the traditional microbead-based fluorescence encoding method, this MQP-based optical thickness encoding method is limited in the number of codes.

3.3 Characterization and performance of MQPs assembled with SERS signal

In order to verify the feasibility of dual-channel encoding method, we need to prove firstly that the integration with SERS encoding channel does not affect the optical thickness decoding to MQPs. The cleaned initial MQPs are shown in Figs. 3(a) and 3(b) with good uniformity in size and smooth distribution in surface. As the Fig. 3(c) shows, the prepared AuNRs, which have long axis of 62 nm and aspect ratio of 3.1: 1, exhibit good dispersity and basically own the same size. After the Raman reporter molecules and AuNRs bonded together by the Au-S bond, they are encapsulated and positive charges modified with APTES solution [45,46]. Figures 3(d) and 3(e) are the zeta potential and particle size distribution of AuNRs-Ra@APTES, where R1 is AuNRs-DTNB@APTES and R2 is AuNRs-4-CBT@APTES. The column values from the histograms in Figs. 3(d) and 3(e) indicate that AuNRs-Ra@APTES have stable positive potential and ideal size within 100 nm. As shown in Fig. 3(f), the AuNRs-Ra@APTES can be evenly assembled on the surface of MQP without aggregation after the electrostatic adsorption. Figure 3(g) is the optical thickness illustration of the MQP obtained by the DW-DHPFM system before and after R1 and R2 assembled. From all the above results, the little difference in optical thickness after assembling has no influence with the optical thickness detection of MQPs, which demonstrate the feasibility of dual-channel encoded method. For the next step, we need to verify the stability of SERS encoding method with various Raman reporter molecules assembled on the surface of MQPs individual and simultaneously.

 

Fig. 3 (a), (b) SEM images of initial MQPs. (c) TEM image of AuNRs. (d) Zeta potential of R1 and R2. (e) Size of R1 and R2. (f) The surface of DCEM. (g) The optical thickness of MQPs before and after the assembling of R1 and R2.

Download Full Size | PDF

In this article, we used four kinds of materials with thiol group and unique spectral peaks as the Raman reporter molecules. Figure 4 is the SERS spectra obtained by assembling single or multiple kinds of Raman reporter molecules on the surface of the MQPs. Figure 4(a) shows the SERS spectra obtained after a single kind of Raman reporter molecule assembled. As we labeled in the figure, the specific peak of 4-HBT is 388 cm⁻¹, 4-CBT is characteristic by the peaks of 342 and 542 cm⁻¹, the peaks of 518 and 1383 cm⁻¹ are used to recognize 2-NT and the characteristic spectrum of DTNB is dominated by the peak of 1340 cm⁻¹. Figure 4(b) is the result obtained after two kinds of Raman reporter molecules assembled simultaneously on the same MQP. By mixing the two kinds of Raman reporter molecules in a specific ratio, we can obtain the SERS spectra with all characteristic peaks of this two Raman reporter molecules. Similarly, the multiple kinds of Raman reporter molecules with different ratios are mixed and assembled on the same MQP. As the Fig. 4(c) shows, all the characteristic peaks of encoding materials can be seen clearly in the decoding SERS spectra. All these results demonstrate that various combinations of SERS encoding can be achieved by this assembling method.

 

Fig. 4 Decoding Raman spectra of (a) single kind of Raman reporter molecules, (b) two kinds of Raman reporter molecules and (c) multiple kinds of Raman reporter molecules.

Download Full Size | PDF

3.4 Principle and encoding capacity of composite encoding method

Since the two encoding channels are independent with each other, the composite encoding method would not cause interference between the two channels theoretically. Following that assumption, with the combination of two channels, the total number of codes can be obtained by directly multiply the codes of each channel, which improve the encoding capacity greatly. The number of composited codes can be calculated as follows:

For the MQP optical thickness encoding channel, the number of codes $W_1$ is

According to the previous theory, for the MQP optical thickness encoding channel, the encoding capacity is near 100. It can be seen from the Eq. (6) that the encoding capacity of the SERS encoding channel increases exponentially with the addition of the type of encoding materials. Therefore, by combining the SERS encoding channel with the MQP optical thickness encoding channel, the total number of codes is increased in a trend of 100 times exponentially grown with the increased types of the Raman reporter molecules, which is sufficient for the normal multiplexed detection.

In order to exhibit the performance of dual-channel encoding method, we select two kinds of Raman reporter molecules (2-NT and 4-CBT) and two optical thicknesses of MQPs (T1: 55 μm; T2: 66 μm) for achieving the composite encoding. According to the Eq. (7), when $m\text{=}2$ and $n\text{=}2$, $W\text{=}6$, which means six composite encoding combinations can be attained. The decoding results are shown in Fig. 5. A and B are Raman reporter molecules, and C and D are two kinds of encoded MQPs with different optical thicknesses. In the binary sequences, the character “1” indicates that the encoding element exists, and “0” gives the opposite meaning. The sequence and corresponding decoded data are shown as the figure, SERS spectra and optical thickness images exhibit good consistency and stability. Through all the above experiments, we have proved that the two channels of the composite encoding method do not interfere with each other, and the number of codes after the combination of encoding channels can be increased greatly. Therefore, this composite encoding method has great potential in multiplexed biological detection.

 

Fig. 5 The binary sequence of codes and components, SERS spectra and optical thickness information of the DCEMs. 2-NT and 4-CBT are two types of Raman reporter molecules, T1 and T2 are two kinds of encoding optical thickness of MQPs. Size of field of view of reconstructed optical thickness image: 1.26 mm × 1.26 mm.

Download Full Size | PDF

3.5 Application of dual-channel encoded SA in multiplexed protein detection

For confirming the practical application of the dual-channel encoded SA in multiplexed biological detection, three kinds of SAs were designed to achieve the selective capture. Specifically, SA1 SA2, SA3 were prepared with mouse IgG grafted DCEM1, rabbit IgG grafted DCEM2 and human IgG grafted DCEM3 respectively. Then the corresponding three kinds of QDs-labeled anti-IgGs, 565 nm QDs-labeled goat anti mouse IgG, 585 nm QDs-labeled goat anti rabbit IgG and 610 nm QDs-labeled goat anti human IgG, were mixed together as the analytes. After specific binding experiment, the fluorescence and decoding results are shown in Fig. 6. Three different colors in the fluorescence image as shown in Fig. 6(a) are produced by three kinds of QDs (565 nm, 585 nm, 610 nm), which indicate that the SAs can selectively capture the analytes corresponding to their specific probes. The intensity of fluorescence signifies the quantitative information of target analytes. The optical thickness image, as the first decoding channel is shown in Fig. 6(b), exhibit the three kinds of optical thicknesses of the reacted SAs with three different colors. SERS spectra as the second decoding channel are shown in Fig. 6(d). Each kind of the SERS spectrum is obtained from the reacted SAs with the same color marked in Fig. 6(c). For example, in Fig. 6(c), the red circle marked SA is SA3, which is prepared by MQP3 (52.10 ± 0.87 μm, DTNB + 4-CBT) grafted with human IgG. The decoding SERS spectrum is shown in Fig. 6(d) with red color, the optical thickness of MQP is manifested in Fig. 6(b) with green color. Since human IgG is the probe, 610 nm QDs-labeled goat anti human IgG is captured by SA3, thus the fluorescence image reveals red color. The immunoassay results show good performance of the dual-channel encoded SA with high accuracy and no interference, which verified the promising application of our proposed dual-channel encoded method in multiplexed detection and analysis.

 

Fig. 6 (a) The fluorescence image and (b) reconstructed optical thickness image of the reacted SA1, SA2, SA3 in multiplexed analysis. (c) The field of view in decoding Raman spectra and (d) the corresponding Raman spectrum of reacted SA1, SA2, SA3. Size of field of view: 1.26 mm × 1.26 mm.

Download Full Size | PDF

3.6 Concentration response of dual-channel encoded SA

For the quantitative analysis, 565 nm QDs-labeled goat anti mouse IgG solutions with concentrations of 30, 15, 7.5, 3.75, 1.875 nM, and 937.5, 468.75 and 0 pM were tested as analytes to demonstrate the response of SA3. Figures 7(a)-7(h) show the fluorescence images of a group of analyte-captured SAs measured by DW-DHPFM system at each concentration gradient. It can be seen from the figure that the fluorescence intensity is gradually decreased with the reduced concentration of target analytes. As shown in Fig. 7(g), almost no fluorescence is visible when the concentration of target analytes reaches as low as 468.75 pM. By fitting the average fluorescence intensity with the concentration of analyte, we get a good linear relationship as shown in Fig. 7(i). The average fluorescence intensity indicates the fluorescence intensity averaged with each pixel excluding the edge region for each MQP. The average and standard deviation are calculated from 50 different MQPs. With the obtained standard deviation of 0.2913(A.U.) to blank samples, the limit of detection, according to the principle of triple standard difference method, is calculated to be 2.1346 × 10⁻⁹ M. Since the limit of detection is connected with the fluorescence intensity of label, it is worthy to note that better performance could be achieved by adopting QDs in label with higher quantum yields.

 

Fig. 7 (a-h) Fluorescence images of SA3 after being reacted with 565 nm QDs-labeled goat anti mouse IgG solution with the concentrations of 30, 15, 7.5, 3.75, 1.875 nM, and 937.5, 468.75 and 0 pM. (i) The concentration response curve is obtained by fitting average fluorescence intensity with the analytes concentration. Error bars indicate the standard deviations from 50 MQPs at different concentrations.

Download Full Size | PDF

4. Conclusions

In summary, it has been verified that DW-DHPFM combining with Raman spectroscopy can be served as a decoding and detecting platform for multiple biomolecules. To achieve the multiplexed detection, a new dual-channel encoding method is utilized by assembling Raman reporter molecules on the surface of MQPs. For the two encoding channels, both the optical thickness of MQPs and the SERS spectra are independent with the fluorescent label, which provide high accuracy and stability for the analytical method. Meanwhile, the crosstalk between encoding signal and label signal can be avoided. Using the fluorescence microscopy to acquire the quantitative information of analytes makes the detection and analysis process much economical and convenient. Furthermore, the combination of the above two channels gives the encoding number a considerable increase in magnitude. The multiplexed detection and gradient experiments have proved the ability of this method for achieving selective capture and quantitative analysis of target analytes. With all these abilities and advantages, we anticipate that the proposed optical system and encoding method have powerful application of multiple biomolecule detection in the field of medical diagnostics.