Author’s Accepted Manuscript Fabrication of plasmon length-based surface enhanced Raman scattering for multiplex detection on microfluidic device Anh H. Nguyen, Jeewon Lee, Hong Il Choi, Ho Seok Kwak, Sang Jun Sim www.elsevier.com/locate/bios

PII: DOI: Reference:

S0956-5663(15)00216-X http://dx.doi.org/10.1016/j.bios.2015.03.064 BIOS7556

To appear in: Biosensors and Bioelectronic Received date: 18 January 2015 Revised date: 15 March 2015 Accepted date: 25 March 2015 Cite this article as: Anh H. Nguyen, Jeewon Lee, Hong Il Choi, Ho Seok Kwak and Sang Jun Sim, Fabrication of plasmon length-based surface enhanced Raman scattering for multiplex detection on microfluidic device, Biosensors and Bioelectronic, http://dx.doi.org/10.1016/j.bios.2015.03.064 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting galley proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Fabrication of plasmon length-based surface enhanced Raman scattering for multiplex detection on microfluidic device Anh H. Nguyena, Jeewon Leea, Hong Il Choia, Ho Seok Kwaka, Sang Jun Sima,b*

[*] Prof. Sim S.J.; Nguyen A.H.; Prof. Lee J.; Choi H.I.; Kwak H.S. a

Department of Chemical and Biological Engineering

b

Green School, Korea University, Seoul 136-713 (Korea) Tel.: +82-31-3290-4853;

Fax: +82-2-926-6102.

E-mail: [email protected] ABSTRACT The length of bioreceptors plays an important role in signal enhancement of surface-enhanced Raman scattering (SERS) due to amplification of electromagnetic fields generated by the excitation of localized surface plasmons. Herein, intact antibodies (IgG) and Fab fragments conjugated onto gold nanostar were used to fabricate two kinds of immunosensors for measurement of their SERS signals. Using CA125 as the antigen and Rhodamine-6G (R6G)-conjugated immunogolds, a SERS immunosensor was selfassembled by antigen-antibody interaction. The results showed that the SERS signal from the Fab immunosensor was 2.4 times higher than that of the IgG immunosensor. Furthermore, increased hot-spots by silver atom deposition onto the IgG and Fab immunosensor showed 2.1 and 1.4 times higher signals than before enhancement, respectively. For application, based on the Fab immunosensor, a SERScompatible microfluidic system was designed for multiplex assays to overcome the drawbacks of conventional assays. This system can measure biological specimens directly from bio fluids instead of using a complex microfluidic device containing separation and detection elements. Four approved biomarkers of breast cancer, including cancer antigen (CA125), HER2, epididymis protein (HE4), and Eotaxin-1, were detected from patient-mimicked serum with limits of 15 fM, 17 fM, 21 fM, and 6.5 fM, respectively. The results indicated that the lengths and geometry of the bioreceptors determined the

1

intensity of SERS signal from the interface and cavity of the sandwich immunosensor. Silver atom deposition at the cavity of the immunosensor increased the SERS signal. Finally, the SERS immunosensor built-in microfluidic system improved the performance of multiplex diagnostics.

Keywords: SERS, Multiplex detection, Microfluidics, Plasmonic nanoparticles, Cancer biomarkers

1. Introduction DNA-based plasmon lengths have been reported from hierarchical fabrication of gold nanoparticles which have plasmon resonance in the visible range and do not blink or bleach (Sönnichsen et al., 2005). Such fabrications have been made with different structures to evaluate the effects of size and gap in the gold nanostructure, resulting in increased application for distance decay of plasmon coupling (plasmonic ruler) (Ringe et al., 2012; Jain et al., 2007), bio-imaging (Sokolov et al., 2003) and sensing (Alivisatos, 2004; Haes et al., 2004). Antibody-based plasmon lengths, however, could also be evaluated from the variable lengths of antibody structures for determination of non-amplified analytes. As proof-of-concept, antibodies including intact IgG (13 nm in length) and Fab fragments (3 nm in length) bound to antigens with the same affinity (Raab et al., 1999), and were successfully conjugated on thiol-modified AuNPs for therapeutics and sensing (Cai et al., 2008; Bogart et al., 2014). The short-arm, conformational homogeneity, and potentiality for directed and dense of immobilization on the interfaces of Fab compared with those of intact IgG provide advantages over use of intact IgG in applications. In surface plasmon resonance and quartz crystal microbalance sensors, the sensitivity was significantly improved by conjugating the sensor surface with antigen binding sites using the reduced C-terminal SH groups of fragment antigen-binding (Fab) (Adamczykt et al., 2000). Moreover, the C-terminal SH groups of Fab are ready to react with maleimide-functionalized nanoparticles to form a directed self-assembly of monolayers. Thus, Fab fragments provide benefits over intact IgG molecules, such as being shorter in length, and the possibility of thiol-maleimide conjugation for directed and dense immobilization on the surface (Shen et al., 2012). Antibodies containing different functional groups (amine, thiol and carboxyl) will self-assembly into

2

an ordered monolayer at a modified gold nanostructure to make immunogolds (Shen et al., 2012). Individual immunogolds are self-assembled by well-known biochemical interactions, such as antibodyantigen coupling. Antibody-antigen interaction occurs far from nanostructure surface, due to the length of the antibody. The electromagnetic field principle of SERS, plasmon lengths generated by the length of the antibody, and direction of capture of the antibodies determine the sensitivity and stability of the SERS signal. This pattern can also be used to form desired assemblies of a sensing platform, for instance, sandwich SERS immunobiosensors. The fabrication of SERS-based immunosensors has attracted much interest as a platform for highthroughput, fast and reliable detection of low concentrations of disease biomarkers present in bio-fluids (Le Ru et al., 2007). Due to the almost label-free detection and distinct spectra of the analytes, SERS is advantageous for diagnostic settings, disease tracking, and research in proteomics (Sharma et al., 2012). SERS signals of absorbed or closed-contact molecules on rough surfaces can result in enhancement of the signal by a factor of as much as 109 – 1010, making the technique sensitive and selective enough to detect single molecules (Le Ru et al., 2007). The SERS sensitivity depends on the shape, size, and surface structure of the nanoparticles and the plasmon length in sandwich immunosensor, which generates hot spots enhance the electromagnetic field, enabling enhancement of the Raman signal (Goh et al., 2012). Microfluidic devices have been successfully applied in micro diagnostic systems for the manipulation of solutions, suspended nanoparticles and biological specimens, and are capable of connection with optical or electronic parts (Chrimes et al., 2013; Chin et al., 2011). The integration of SERS and microfluidics enables the development of highly sensitive and specific devices in a very short time (in seconds) without sample processing, fluid handling and signal amplification (Franke et al., 2008). The combination of microfluidics and immunosensor is the most useful technique, with improved performance including substantial reduction of background, and amplification of the signal via a series of incubation and washing steps (Chin et al., 2011). Here, we characterized the plasmonic length-dependent sensitivity of SERS-based immunosensor with Fab and intact IgG, which served as a platform for multiplex detection on a microfluidic device.

3

Comparison was made between the characteristics of two kinds of SERS platforms: IgG-immunosensor and Fab-immunosensor. Moreover, the number of hot spots on the immunosesnor was increased by filling the cavity of its surface and gaps in sandwich immunosensor with silver atoms. Finally, the immunosensor was applied for multiplex detection of an approved panel of biomarkers (CA125, HE4, CEA, Eotaxin-1) for breast cancer diagnostics with simplicity, a rapid mode, and high sensitivity. The increased SERS signal after the formation of new hot spots increased by 2.1 and 1.4 times for the Fabimmunosensor and IgG-immunosensor platforms, respectively. The detection limit of the device was determined to be 15 fM for cancer antigen (CA125), 17 fM for HER2, 21 fM for epididymis protein 4 (HE4), and 6.5 fM for Eotaxin-1.

2. Materials and Methods 2.1. Materials Antigens and antibodies against cancer antigen (CA125), HER-2, epididymis protein 4 (HE4) and Eotaxin-1 were purchased from Fitzgerald (Germany). Gold (III) chloride trihydrate (≥ 99.9%), Sodium Citrate (99%), poly-(vinylpyrrolidone) (PVP, MW = 10,000), N, N-dimethylformamide (DMF, 99.8%), Ethanol (99.8%), Sodium Sulfide (Na2S·9H20, 98%), p-MercaptoBenzoic Acid (p-MBA), furan and thiolpolyethylene glycol were purchased from Sigma Aldrich. HS-PEG-NH2 and sulfo-SMCC were purchased from Creative PEG Works (USA) and Thermo Scientific, respectively.

2.2. Synthesis of gold nanostar (AuNSs) Synthesis of the gold nanostars in size-controlled patterns was introduced in a previous report (Khoury and T. Vo-Dinh, 2008). Briefly, 100 μL of 50 mM HAuCl4 was added to a growth solution containing 50 μL of 12-nm PVP-coated gold seeds (Supplementary Fig. S1) in ethanol (1.5 mM) with an in 15 mL DMF containing 10 mM PVP (10,000 MW). The reaction was performed under vigorous stirring (1500 rpm) for 20 min till the solution became stable deep green, and 60-nm gold nanostars were achieved. The reaction was halted by addition of 10 μl of 4.2 mM Na2S in DMF. The nanostar solution was washed by centrifugation to remove the excess DMF and PVP. The samples were analyzed via UV-

4

Visible spectroscopy (UV-3600, Shimadzu) and high-resolution transmission electron microscopy (JEM2100F, 200 kV, JEOL).

2.3. Preparation of gold nanostar conjugates as detectors Polyclonal and monoclonal antibodies were used for capturing and detecting specific antigens. Four Fab fragments were prepared from the antibodies against four antigens (CA125, HER2, HE4 and Eotaxin1) using Fab preparation kits (Thermo Scientific). The Fab fragments were then conjugated into maleimido nanostar particles (5 mg/ml ~ 4.2 x 1012 particles/ml ) in 1 mL of 0.1 M PBS, pH 6.0, containing 5 mM EDTA. Briefly, the maleimido gold nanostars (60 nm) were functionalized with the Fab fragments by adding the fragments (5 μM, pH 6.0) to the solution and shaking for 20 min, after which Rhodamine 6G (R6G)-modified alkylthiol-capped oligonucleotide (HS-A5-R6G), with an OD of 0.25 at 260 nm, was added to the solution to form R6G-immunogold. After 12h, 10% BSA solution (300 μl) was added to the solution to further passivate the surface of the gold nanostar. The solution was allowed to stand for 10 min, and then the R6G-immunogold particles were purified by centrifugation (10,000 rpm, 15 min) and separated from the supernatant. The immunogold particles were then redispersed in PBS buffer solution (Sigma tablet), pH 7.4.

2.4. Fabrication of microfluidic device The microfluidic device was designed by computer graphics using Auto CAD software and then printed out as a metal masks with the resolution of 50 μm. Fabrication of the microfluidic device using polydimethylsiloxane (PDMS) followed a previously published method (Fujii, 2002). Glass slides (76 x 26 x 1 mm, Marienfeld) were used as a support for the microfluidic devices to build inlet and outlet connections for the SERS immunosensor. The device allowed sample movement from the input to the detection chamber (upper chamber), which ended at the outlet (lower chamber) (Fig. 2A and Supplementary Fig. S7). The chamber was cleaned with 200 μL isopropanol and DI water prior to surface treatment. SEM images of the meander were also obtained with a JSM-7610F (Jeol, Japan).

5

2.5. Surface modification, thiol functionalization and preparation of chip The reaction chambers on the device were first incubated with 100 μl of absolute ethanol for 5 min, and then washed extensively three times with 200 μl deionized water (DIW). The device was then incubated with 100 μl of 1% (wt/vol) KOH in DIW for 10 min and washed extensively three times with DIW. The KOH-activated surface chamber was treated with oxygen plasma in a plasma cleaner at the plasma strength of 30 W. A vacuum pump with a pump speed of 0.91 ft3 min-1 was used to generate the pressure of 0.31 mbar. The plasma-oxidized plate well surface was then incubated with 100 μl of 2% (vol/vol) MPTS in DIW in a vacuum desiccator at 80o C for 1 hour (Dixit et al., 2011). After equilibration of the chamber to room temperature, the thiol-modified chamber was extensively washed three times with DIW. The preparation of maleimide-functionalized 60 nm gold nanostars was carried out as reported in a previous study (Ba et al., 2010), with a minor modification in the thiol glass functionalization. Briefly, the gold nanostars (4.2 x 1012 particles/ml) were incubated to allow formation of a self-assembled immunosensor on the surface of the reaction chamber. The immunosensor was then incubated with heterobifunctional (HS-PEG-NH2) and monofunctional (mPEG-SH) poly(ethylene glycol) (ratio 1:9) to guarantee efficient surface grafting and particle stability in PBS, and incubated for 10 h. The chamber was then washed with three times with PBS buffer. The modified immunosensor (AuNS-mPEG-SH/HS-PEGNH2) was reacted for 1 h with 1 ml of a fresh 100 μM solution of the maleimide crosslinker, sulfo-SMCC, in PBS, pH 7.6. Thereafter, the immunosensor was washed three times with PBS buffer (pH 7.6) to remove excessive crosslinker. The maleimide-crosslinker-modified immunosensor was ready for directed conjugation of specific antibodies. For antibody conjugation, 100 ng of the prepared Fab (2 nM) (after treatment with 50 mM DTT) was dissolved in 1 mL of 0.1 M PBS, pH 6.0, containing 5 mM EDTA. The reduced Fab was injected and incubated for 1 h. The unconjugated Fab was then washed with PBS until no Fab was present in the washing fractions (tested by ELISA).

2.6. Excitation of localized surface plasmons of plasmonic length-based coupling Amplification of electromagnetic fields generated by the excitation of localized surface plasmons

6

plays an important role in increasing the SERS sensitivity (Sharma et al., 2012). At the level of individual gold nanostars, excitation of the localized surface plasmon due to plasmon length-based coupling was measured by localized surface plasmon resonance (LSPR) (Jain et al., 2007). Briefly, gold nanostars were immobilized onto the MPTS-modified chamber and functionalized via the same procedure as presented in section 2.5. After injection of the serum samples and immungold detectors, excitation of the localized surface plasmons from plasmon coupling depending on the length of the antibodies (IgG and Fab) was measured with a home-built Raman and Rayleigh integrated microscope (supporting information “optical configuration”).

2.7. Increase of hot spots on the immunosensor surface The creation of more SERS active hot spots on the immunosensor surface produces increased SERS detection sensitivity (Wei et al., 2013; Goh et al., 2012). The increase was performed by deposition of silver atoms in the cavity of the dual layers and inter-distance between the gold nanostar and the second layer. The deposition of Ag atoms generated from AgNO3 was performed by reduction of 10 fM to 100 μM AgNO3 using 10 μM hydroquinone on the surface of the immunosensor, which was stopped by addition of 1.25 μM Na2S. The immunosensor was then washed three times with Milli-Q water, and the SERS signal was measured.

2.8. SERS for multiplex detection Multiplex detection was performed with an approved panel of cancer biomarkers, cancer antigen (CA125), epidermal growth factor receptor (HER2), epididymis protein 4 (HE4) and Eotaxin-1. The antigens (10-3 fM – 104 fM) were mixed with 50 μl human serum (Sigma Aldrich). Each chamber was incubated in triplicate for individual samples for 30 min, and then 5x10-3 mg/ml of R6G-immunogold were injected and incubated for 5 min. The chambers were scanned from left to right on the X-axis by a home-built Raman microscope (Fig. 2A) after washing with 0.3 M NaCl PBS, pH 7.4.

7

3. Results and Discussion 3.1. Plasmonic length-dependent sandwich immunosensor Plasmonic length-dependent SERS-based immunosensor was constructed according to the length of intact IgG and Fab molecules placed between sandwich structures. The immunogolds were made of a conjugation of intact IgG and Fab onto gold nanostars to produce IgG-immunogold (IgG-gold) and Fabimmunogold (Fab-gold), respectively. Like the sandwich ELISA principle, polyclonal antibody (pAb) was used for antigen capture while monoclonal antibody (mAb) was used for antigen detection. Self-assembly of the capturing immunogold into an MPTS-modified surface array for two kinds of platforms was achieved by incubation of 60 nm (5 mg/ml ~ 4.2 x 1012 particles/ml) of the polyclonal IgG-gold or FabIgG (Fig. 1A and 1B). The Fab fragment was prepared with a Fab preparation kit (Thermo) (Supplementary Fig. S2). The capturing immunogold was exposed to 50 fM of specific antigens. The detecting immunogold was achieved by incubation of the monoclonal IgG-gold and Fab-gold (4.1 mg/ml ~ 3.7 x 1012 particles/ml). After incubation, the resulting immunosensor were observed in a dark-field image. The color of immunosensor surface was orderly changed by formation of capture monolayer, sandwich immunosensor, and after enhancement (Supplementary Fig. S3). The immunosensor was evident from the SEM image (Fig. 1C), which showed formation of the sandwich immunosensor of selfassembled immunogolds. The shapes and sizes of the gold nanostars used for immunosensor fabrication were determined by TEM (Fig. 1D). Sandwich immunosensor was constructed in this work owing to their simplicity, high sensitivity and cost-effectiveness for protein detection. The immunosensor was set up in an integrated Raman and microfluidics system (Fig. 2A). Microfluidics were manufactured with precise features at 3 μm for the channel and 0.2 cm for the reaction chamber with different meanders, made by polydimethylsiloxane (PDMS) (Fig. 2B). The assembly of immunosensor depends on the spontaneous binding of antigenantibody interactions. Through the use of Fab molecules, reduction of the distance in the sandwich structure and minimum of steric hindrance in the immunosensor surface for detection could be achieved,

8

thus resulting in SERS signal enhancement. Moreover, using Fab-gold increases the density compared with IgG-gold, resulting in reduced inter-distance between the immunogolds due to minimization, directed and dense immobilization on the interfaces (Backmann et al., 2005). The number of immunogolds self-assembled into the immunosensor were in range of 1.8 x 109 to 1.1 x 1010 particles, as calculated by UV/vis spectroscopy. The excitation of localized surface plasmons due to closed-contact plasmon coupling generates amplification of the electromagnetic fields, enhancing the SERS signal (Sharma et al., 2012; Wei et al., 2013). Plasmonic coupling from two immunogolds was determined by the LSPR spectra. The LSPR of the IgG-gold coupling was recorded at about 49 nm (Fig. 2C and inlet), while a 153 nm shift was recorded for Fab-gold (Fig. 2D and inlet). Moreover, the LSPR of nanoparticles can be excited to produce highly intense localized electromagnetic fields, which can strongly enhance the Raman scattering of cross sections of molecules in their proximity (Stockman et al., 2011). Decrease of the inter-particle distance increases the electric near-field, with a consequent shift of the plasmon resonance peak (Reinhard et al., 2005). Moreover, the electromagnetic field of LSPR decays exponentially away from the surface of the nanoparticles, extending to only a few nanometers (Lu et al., 2009, Nguyen et al., 2015), which positions the analytes outside of the hot-spots for SERS. The capturing immunogolds were made of polyclonal antibody to capture CA125, while the detecting immunogolds were monoclonal R6G-capped-immunogold. After incubation, immunosensor with the inter-distance (cavity) of 13 nm and 3 nm or intact IgG and the Fab fragments, respectively, was formed. SERS measurement was then directly performed with an acquisition time of 0.3 s per spectrum in the reaction chamber (Fig. 3A and 3B). The SERS signal from the Fab-gold was 2.4 times higher than that from the IgG-gold array, as shown in Fig. 3C. This results from increase of the immunogold density on the Fab-gold surface, which led to enhancement of Raman signal. As expected, a decrease in antibody length resulted in an increase in the total SERS signal. For example, comparing the intensities of the 1512.6 cm-1 vibrational band for 50 fM CA125 on the two kinds of immunosensor, the peak intensity of the Fab-gold increased by 2.4 times compared to that of the IgG-gold. Moreover, the antigen

9

concentration (1 aM to 10 pM) was directly proportional to the SERS signal at the 1512 cm-1 vibrational band (Fig. 3D). High stability of the immunosensor and plasmonic resonance between the immunogolds were advanced characteristics for detection at low concentration, and for reproducibility. Fig.1 and Fig. 2

3.2. Increased hot-spots by silver atom deposition The surface roughness and electromagnetic coupling between immunogolds can produce active hot spots that contain enhanced electromagnetic fields in intersections, which increases the sensitivity of the SERS-based immunosensor (Goh et al., 2012; Alonso-González et al., 2012). Cavities are present in the sandwich immunosensors. By using a silver deposition to fill the cavities, an attempt was made to increase the number of hot-spots on the immunosensor surface for increased SERS signal. To accomplish this, the immunosensor was subjected to silver deposition by reduction of AgNO3 (10 fM – 100 μM) with 3 μM hydroquinone. Deposition on the surface created “hot spots”, based on silver atom deposition (Fig. 4A and Fig. 4B). As the concentration of the detecting immunogolds is directly related to the concentration of the antigen through immunoreaction, the presence of the immunogolds acted as a solid phase for the reduction catalysis of silver ions to silver atoms. Examination of the SERS intensity of the immunosensor demonstrated that the silver deposition increased the SERS signal of the Fab-gold and IgG-gold by 2.1 and 1.4 times, respectively, compared to those observed before enhancement (Fig. 4C) due to increased number of SERS hot spots and resonant Raman cross sections (Meyer et al., 2010). After enhancement, EDX data showed that the silver content was deposited onto the immunosensor accounting for 19.66% (atomic %) (Supplementary Fig. S4). In a range of silver concentrations, amplified SERS signal was detected at the value of 1 nM AgNO3 (Fig. 4D). The deposition of silver atoms on the immunosensor also depends on the reaction rate of hydroquinone. When applying this mechanism for universal increase in the number of hot spots, low concentrations of hydroquinone in the range of 1 nM to 3 μM showed increased SERS sensitivity and reproducibility (Red box, supplementary Fig. S5). Use of over 8.1 μM of hydroquinone resulted in unstable and non-reproducible SERS signals (Cyan box, supplementary Fig. S5). At high concentrations of hydroquinone, the abundant reducing agent favors

10

nucleation instead of epitaxial growth, causing free-standing silver nanocrystals to be obtained (Pérez et al., 2008). The hot-spots formed during self-assembly depend on the plasmonic length and inter-distance of the immunogolds. The SERS intensity was dominated by that of the R6G occupying sites of the large Raman cross section. R6G molecules residing in the hot spots of the sandwich immunosensor are very sensitive compared with those outside of hot spots (Wei and Xu, 2013). The deposition of silver atoms on the surface caused the formation and transition of new hot spots that increased the field enhancement within the hot spots. Although the distribution of hot-spots for enhancement of SERS accounts for only 0.0063% (63 hot spot sites out of 1,000,000 sites) (enhancement factor > 109), their contribution to the total SERS signal is 24% (Fang et al., 2008). In the resultant silver deposition, the resonant intensity is a function of the concentration of the biomarkers, partially filling the cavity in the immunosensor. Fig. 3 and Fig. 4

3.3. SERS-microfluidics for multiplex detection SERS-integrated microfluidic devices were fabricated as described in Fig. 2A, and effective conjugation of the Fab antibodies was assayed by ELISA (Supplementary Fig. S6). The microfluidics were designed with four reaction chambers, each of which was developed for one approved biomarker on the device (Fig. 2B and Supplementary Fig. S7). First, we wanted to develop a device for the multiplex detection of microfluidic cassettes with cost-effective, SERS scanning-based high performance. In the channel of the device, the reagents (serum, PBS 0.1% Tween, immunogolds, silver nitrate and hydroquinone solution) and biological samples were delivered by micropump (Warner Instrument). This approach allowed the use of precise volumes and incubation times for the assay (Whitesides, 2006). Since breast cancer is highly heterogeneous (Stingl and Caldas, 2007), measurement of a single biomarker is not reliable for diagnosis and classification. Using the SERS-integrated microfluidic device, multiplex detection was performed with a panel of biomarkers for breast cancer, including cancer antigen (CA125), HER2, epididymis protein 4 (HE4) and Eotaxin-1. Under optimized conditions, the sensitivity of the immunosensor was estimated by the SERS signal in concentration ranges from 1 aM to 10 pM. The

11

SERS intensity at the 1521 cm-1 peak of the R6G spectral imprinting patterns was considerably increased with increasing concentration of the biomarkers, which corresponded to self-assembly of the sandwich immunosensor (Fig. 5A). The increase of the electromagnetic field enhances the SERS intensity, which was determined by self-assembly (Ko et al., 2008). The SERS intensity at 1512 cm-1 peak increased when the number of Fab-gold increased due to corresponding increase in the total electromagnetic field (Fig 2D and Fig. 5B). The feature SERS peak of R6G at 1512 cm-1 was used to determine the limits of detection (LOD) for the panel of biomarkers. The LOD values were determined to be 15 fM for cancer antigen (CA125), 17 fM for HER2, 21 fM in the case of epididymis protein 4 (HE4) and 6.5 fM for Eotaxin-1, using a formula which was described in the supporting information (Ma et al., 2014) (Fig. 5C). Detection of distinct SERS spectra from the four immunnosensors on the microfluidic device using mimic serum samples suggested that the immnunosensor may be convenient for clinical use to evaluate the levels of these biomarkers. The specificity of the immunosensor was determined for other substrates including PSA, BSA, APOE4, β amyloid 42, commercial human serum (Sigma Aldrich) and lysozyme at 10 nM. The SERS imprinting spectra of the samples at 1512 cm-1 and 1364 cm-1, which are the typical Raman peaks of R6G (Jensen et al., 2006), showed no clear variant with these unspecific substrate at 10 nM (Fig. 5D). The results confirmed that the unspecific targets were not specific to the immunosensor. Thus, this platform had great selectivity for the detection of the panel of biomarkers for breast cancer.

Fig. 5 4. Conclusions In summary, SERS effects and signals based on the plasmonic lengths of sandwich SERS immunosensor and Rayleigh scattering of plasmonic coupling were determined by using intact IgG and Fab molecules as bioreceptors. The Fab-immunosensor demonstrated 2.4 times higher signal compared with that of the IgG-immunosensor. Moreover, kinetic analysis of the increased hot spots on the immunosensor for SERS application was performed, resulting in 2.1 and 1.4 times higher signal for the

12

Fab- and IgG-immunosensors, respectively, compared to before enhancement. Based on the results, an integrated SERS-microfluidic device was fabricated for multiplex detection of a panel of breast cancer biomarkers from patient-mimicked serum. The platform demonstrated LOD of 15 fM, 17 fM, 21 fM and 6.5 fM for CA125, HER2, HE4 and Eotaxin-1, respectively. Thanks to its specificity, applicability and fast-scanning, the platform could be used for clinical testing or for screening of rare biomarkers.

Acknowledgments This study was supported by the Korea Institute of Energy Technology Evaluation and Planning and Ministry of Trade, Industry & Energy of in “Energy Efficiency & Resources Technology R&D” project Korea (20122010200010-11-2-100), the National Research Foundation of Korea (NRF) grants (grant No. NRF-2013R1A2A1A01015644/2010-0027955/2012R1A2A1A01008085),

University-Institute

Cooperation Program (2013), and grants (2014M1A8A1049278) from Korea CCS R&D Center of the NRF funded by the Ministry of Science, ICT, & Future Planning of Korea.

References Adamczyk, M., Moore, J.A., Yu, Z., 2000. Methods, 20, 319-328. Alivisatos, A.P., 2004. Nat. Biotechnol., 22, 47-52. Alonso-González, P., Albella, P., Schnell, M., Chen, J., Huth, F., García-Etxarri, A., Casanova, F., Golmar, F., Arzubiaga, L., Hueso, L.E., Aizpurua, J., Hillenbrand, R., 2012. Nat. Commun., 3, 684 doi:10.1038/ncomms1674. Ba, H., Rodríguez-Fernández, J., Stefani, F.D., Feldmann, J., 2010. Nano Lett., 10, 3006-3012. Backmann, N., Zahnd, C., Huber, F., Bietsch, A., Plückthun, A., Lang, H.P., Güntherodt, H.J., Hegner, M., Gerber, C., 2005, PNAS., 102, 14587-14592. Bogart, L.K., Pourroy, G., Murphy, C.J., Puntes, V., Pellegrino, T., Rosenblum, D., Peer, D., Lévy, R., 2014. ACS Nano, 8, 3017-3122.

13

Cai, W., Gao, T., Hong H., Sun, J., 2008. Nanotechnol. Sci. Appl., 1, 17-32. Chin, C.D., Laksanasopin, T., Cheung, Y.K., Steinmiller, D., Linder, V., Parsa, H., Wang, J., Moore, H., Rouse, R., Umviligihozo, G., Karita, E., Mwambarangwe, L., Braunstein, S.L., J. v. d. Wijgert, R. Sahabo, J. E. Justman, W. El-Sadr and S. K. Sia, 2011. Nat. Med.,17, 1015-1019. Chrimes, A.F., Khoshmanesh, K., Stoddart, P.R., Mitchella, A., Zadeha, K.K., 2013. Chem. Soc. Rev., 42, 5880 -5906. Dixit, C.K., Vashist S. K., MacCraith, B.D., O’Kennedy. K., 2011. Nat. Protoc., 6, 439-445. Fang, Y., ,Seong, N. H., Dlott, D. D., 2008. Science, 321, 388-392. Franke T.A and Wixforth, A., 2008. Chem.Phys.Chem., 9, 2140 – 2156. Fujii, T., 2002. Microelectron. Eng, 61, 907–914. Goh, M.S., Lee, W.H., Pedireddy, S., Phang, I.Y., Tjiu, W.W., Tan, J.M.R, Ling, X.Y., 2012. Langmuir, 28, 14441-14449. Haes, A.J., Hall, W.P., Chang, L., Klein W.L., Van Duyne R.P., 2004. Nano Lett., 4, 1029-1034. Jain, P.K and El-Sayed, M.A., 2007. Nano Lett., 7, 2854-2858. Jain, P.K., Huang, W., El-Sayed, M.A., 2007. Nano Lett., 7, 2080-2088. Jensen, L and Schatz, G.C., 2006. J. Phys. Chem. A, 110, 5973-5977. Khoury C.G and Vo-Dinh, T., 2008. J. Phys. Chem. C. Nanomater Interfaces, 112, 18849–18859. Ko, H., Singamaneni, S., Vladimir, V., Tsukruk, V. V., 2008. Small, 4, 1576–1599. Le Ru, E.C., Blackie, E., Meyer, M., Etchegoin, P.G., 2007. J. Phys. Chem. C, 111, 13794–13803. Lu, X., Rycenga, M., Skrabalak, S.E., Wiley, B., Xia, Y., 2009. Annu. Rev. Phys. Chem., 60, 167–192. Ma, W., Yin, H., Xu, L., Wu, X., Kuang, H., Wang, L., Xu, C., 2014. Chem. Commun., 50, 9737-9740. Meyer, S.A., Le Ru, E.C., Etchegoin, P. G., 2010. J. Phys. Chem. A, 114, 5515–5519. Nguyen, A.H., Ma, X., Sim S.J., 2015. Biosens. Bioelectron., 66, 497-503. Pérez, M.A., Moiraghi, R., Coronado, E.A., Macagno, V.A., 2008. Cryst. Growth Des., 8, 1377-1383. Raab, A., Han, W., Badt, D., Smith-Gill, S.J., Lindsay, S.M., Schindler, H., Hinterdorfer, P., 1999. Nat. Biotechnol., 17, 901-905.

14

Reinhard, B.M., Siu, M., Agarwal, H., Alivisatos, A.P., Liphardt, L., 2005. Nano. Lett., 5, 2246-2252. Ringe, E., Langille, M.R., Sohn, K., Zhang, J., Huang, J., Mirkin, C.A., Van Duyne, R. P., Marks, L.D., 2012. J. Phys. Chem. Lett., 3, 1479−1483. Sharma, B., Frontiera, R.R., Henry, A.I., Ringe, E., Van Duyne, R.P., 2012. Mater. Today, 15, 16-25. Shen B.Q., Xu, K., Liu, L., Raab, H., Bhakta, S., Kenrick, M., Parsons-Reponte, K.L., Tien, J., Yu, S.F., Mai, E., Li, D., Tibbitts, J., Baudys, J., Saad, O.M., Scales, S.J.,

McDonald, P.J., Hass, P.E.,

Eigenbrot, C., Nguyen, T., Solis, W.A., Fuji, R.N., Flagella, K.M., Patel, D., Spencer, S.D., Khawli, L.A., Ebens, A., Wong, W.L., Vandlen, R., Kaur, S., Sliwkowski, M.X., Scheller, R.H., Polakis, P., and Junutula, J.R., 2012. Nat. Biotechnol., 30, 184-189. Sokolov, K., Follen, M., Aaron, J., Pavlova, I., Malpica, A., Lotan R., Richards-Kortum, R., 2003. Cancer Res., 63, 1999-2004. Stockman, M.I., 2011. Phys. Today, 64, 39-44. Stingl, J and Caldas, C., 2007. Nat. Rev. Cancer., 10, 791-799. Sönnichsen, C., Reinhard, B.M., Liphardt, L., Alivisatos, A.P., 2005. Nat. Biotechnol., 23, 741 – 745. Wei, H and Xu, H., 2013. Nanoscale, 5, 10794-10805. Whitesides, G. M., 2006. Nature, 442, 368-373. Zhang, L., Jin, Y., Mao, H., Zheng, L., Zhao, J., Peng, Y., Du, S., Zhang, Z., 2014, Biosens Bioelectron., 58, 165-171.

15

Figure captions Fig. 1. Surface modification, thiol functionalization and preparation of chip (A) Using intact IgG molecules (13 nm). (B) Using Fab molecules (3 nm). (C) SEM image of sandwich immunosensor surface. (D) TEM image of capture monolayer. Scale: 100 nm for TEM and 500 nm for SEM. Fig. 2. Excitation of localized surface plasmons of plasmonic length-based coupling (A) Home-built optical system for Rayleigh and Raman scattering on microfluidic device. (B) SEM image of a part of microfluidics device (reaction chamber and meander). Excitation of localized surface plasmon of plasmonic length-based coupling: (C) from IgG-gold (coupling model in inlet) and (D) FabY

gold (coupling model in inlet).

IgG-gold,

antigen (CA125),

Fab-gold.

Fig. 3. Plasmonic length-dependent SERS signal of sandwich immunosensor SERS intensity of R6G Raman dye for CA125 detection (10-2 fM to 104 fM) on (A) IgG-gold array and (B) Fab-gold array. (C) Signal to noise ratio of IgG-gold and Fab-gold array, Fab-gold array showed 2.4 times higher SERS intensity at 1512 cm-1 peak than IgG-gold array (inlet). (D) Regression of SERS intensity at 1512 cm-1 for Fab-gold (black line) array and IgG-gold array (red line). Fig. 4. Increased hot-spot sites by silver atom deposition (A) Model of silver deposition on the immunosensor. (B) SEM image of immunosensor surface after silver deposition. (C) Enhancement of SERS intensity based on silver concentration for Fabimmunosensor and IgG-immunosensor. (D) Regression of SERS intensity at 1512 cm-1 on immunosensor before and after enhancement. Fig. 5. SERS-microfluidics for multiplex detection (A) Multiplex detection for CA-125, HER-2, HE-4 and Eotaxin-1. (B) The linear relationship between the concentration of the biomarkers and the SERS signal. (C) Proportional correlation of immunogold concentration with fixed antigen concentration (200 fM) on immunosesnor surface. (D) Unspecific assay for the immunosensor.

16

Highlights (for submission) -

Fabrication of plasmon length based SERS immunosensor based on IgG and Fab molecules

-

Fab immunogold array showed stables, sensitivity in integrated SERS-microfluidics device

-

Increased hot spots numbers on the immunosensor surface in microfluidic device

-

Multiplex detection for a panel of breast cancer biomarkers at ultralow concentration

17

A

CA125

HER2

Thiol oligo- R6G

HE4

B

Eotaxin1

Capturing IgG Immunogold

HER2

Thiol oligo- R6G

HE4

Eotaxin1

Capturing Fab Immunogold

Detecting Fab Immunogold

Detecting IgG Immunogold

ScanningC

CA125

ScanningD

Fig. 1.

18

Sandwich SERS

B

C

D

Y 800

900

Fig. 2.

19

500

600

700

6 nm

700

Y

26 nm

600

6 nm

500

26 nm

Y

Y

A

800

900

B

883 (a.u) Intensity (a.u)

1325 (a.u)

Intensity (a.u)

A

104 103 2 10 101 100

104 103 102 101 10o

10-1 10-2 10-3

D

Raman shift, cm-1

SERS intensity at 1512 cm-1 (a.u)

C

y = 150.3 log(x) + 742.2

Signal to Noise ratio (S/R)

10 8

y = 121.1 log(x) + 400.9 6 4 2 0

IgG-gold array

104

103

102

101

10-1 10-2 10-3

100

10-1

10-2

Fab-gold array

10-3

[CA125, fM]

Fig. 3.

20

Laser

A

SERS B

Ag+ Hydroquinone

C

D

102

101

100 10-1 10-2 10-3 [AgNO Log [AgNO 3, μM] 3, μM]

10-4

10-5

Fig. 4.

21

A

B

C

D

Log

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

Raman Shift (cm-1)

= 136.6x + 64.6 y =y 167.2 Log(X) + 821.6 y =y 159.5 Log (X)+ 886.3 = 90.6x + 79.1 y = 114.3 Log(X) + 672.7 = 82.8x + 55.5 y =y147.6 Log(X) + 1097.6 y= 62.4x + 93.2

PSA

Fig. 5.

22

BSA

APOE4

Aβ42

Normal Lysozyme Human Serum