Biosensors and Bioelectronics 81 (2016) 23–31

Contents lists available at ScienceDirect

Biosensors and Bioelectronics journal homepage: www.elsevier.com/locate/bios

Total internal reflection plasmonic scattering-based fluorescence-free nanoimmunosensor probe for ultra-sensitive detection of cancer antigen 125 Suresh Kumar Chakkarapani a, Peng Zhang a, Sujin Ahn a, Seong Ho Kang a,b,n a b

Department of Chemistry, Graduate School, Kyung Hee University, Yongin-si, Gyeonggi-do 17104, Republic of Korea Department of Applied Chemistry and Institute of Natural Sciences, Kyung Hee University, Yongin-si, Gyeonggi-do 17104, Republic of Korea

art ic l e i nf o

a b s t r a c t

Article history: Received 22 December 2015 Received in revised form 29 January 2016 Accepted 30 January 2016 Available online 17 February 2016

Highly sensitive detection of cancer antigen 125 (CA125) on nanoarray chips was carried out by means of total internal reflection (TIR) microscopy based on fluorescent labeling (i.e., TIR fluorescence microscopy; TIRFM) and fluorescent-free labeling (TIR scattering microscopy; TIRSM). TIR plasmonic scattering of nanoparticles (NPs) as a fluorescence-free immunosensor probe potentially superior to fluorescent probes was applied to quantify CA125 on a nanoarray chip. NP-labeled CA125 (NP-CA125) was immunoreacted on chips, and the TIR scattering illumination of NP-CA125 allowed quantitative TIRSM measurement of wavelength-dependent plasmonic scattering detection of CA125. In addition, Alexafluor 488–labeled CA125 was immunoreacted on the same chips for comparison of detection sensitivity. TIRSM showed less photobleaching and higher photostability and detection sensitivity than TIRFM, as well as a lower limit of detection (LOD), 0.0018 U/mL. This LOD was
144 times lower than that of previously reported detection methods. These results demonstrated that the wavelength-dependent TIR plasmon NPs can be used as an enhanced nanoimmunosensor probe, providing ultra-sensitive fluorescence-free biomolecule detection to enable earliest-stage disease diagnosis. & 2016 Elsevier B.V. All rights reserved.

Keywords: Cancer antigen 125 (CA125) Nanoimmunosensor TIR plasmonic scattering Nanoparticle

1. Introduction A challenging task in the field of biomedical science is to detect diseases at their earliest stages to allow effective clinical treatment (Etzioni et al., 2003). The stages of several types of cancers, including ovarian cancer, can be tested by measuring the serum level of cancer antigen 125 (CA125) in blood samples (Bast et al., 1981). So far, CA125 has been studied extensively for earliest-stage detection of cancer by quantitatively determining its lowest possible limit of detection (LOD) (Huckabay et al., 2013; Pal et al., 2015; Raamanathan et al., 2012). Owing to their smaller required sample volumes (Chen and Li, 2007), greater sensitivity (Lee et al., 2004), and wide dynamic range (Lee and Kang, 2014), nanoarray chips have emerged as an excellent immunoassay detection platform, supporting earliest-stage detection of disease-related biomolecules at the single-molecule level. Biomolecule detection by immunoassay is highly specific, as it involves a complimentary binding of the target antigen to its corresponding antibody (Chung et al., 2005; Lee et al., 2013). Several techniques have been n

Corresponding author. E-mail address: [email protected] (S.H. Kang).

http://dx.doi.org/10.1016/j.bios.2016.01.094 0956-5663/& 2016 Elsevier B.V. All rights reserved.

reported for biomolecule detection by immunoassay on nanoarray chips (Aroonyadet et al., 2015; Lee et al., 2013; Lee and Kang 2014; Shi et al., 2015). Each technique has its own strengths and weaknesses, especially with regard to the lowest possible LOD. Recent studies on biomolecule detection by total internal reflection (TIR) microscopy have demonstrated greater sensitivity of detection for disease-related biomolecules and have achieved lower LODs than previously reported detection methods for the same biomolecules (Lee et al., 2015; Lee and Kang, 2013). In TIR microscopy (Fig. 1A), laser light is directed to generate TIR (Fig. 1A inset) on the glass prism/liquid interface, which in turn creates an evanescent field layer (EFL) on the aqueous medium (Axelrod et al., 1984). The electromagnetic field in the EFL is strongest at the interface and declines exponentially with distance from the interface (Axelrod, 2013). Thus, a label of the target molecule closer to the interface will yield greater signal intensity than one farther away, resulting in a high signal-to-noise (S/N) ratio (Prieve et al., 1987). The EFL can cause fluorophores present at the interface to produce fluorescence emission by means of interaction with the electromagnetic field (Prieve et al., 1987). The EFL can also cause plasmon NPs to undergo plasmonic scattering by means of interaction between the electromagnetic field and

24

S.K. Chakkarapani et al. / Biosensors and Bioelectronics 81 (2016) 23–31

Fig. 1. (A) Photographs of the physical layout of TIR microscopy system; (inset) TIR of the laser in a Dove prism. (B) Schematic illustration of immunoassay detection of CA125 conjugated with primary antibody (X306) and secondary antibody (X75), on GNCs labeled with 20 nm SNPs for TIRSM detection and with Alexafluor 488 fluorescent dye for TIRFM detection. Abbreviations used: EM-CCD, electron-multiplying cooled charge-coupled device camera; SNP, silver nanoparticle; CA125, cancer antigen 125; 2nd Ab, secondary antibody, 1st Ab, primary antibody; PAG, protein A/G; DSP, dithiobis(succinimidyl propionate); GNC, gold-nanopatterned chip.

free conduction electrons on the NPs' surfaces (Hoppener and Novotny, 2012). Therefore, TIR microscopy is a versatile tool for imaging both fluorophore- and NP-labeled biomolecules as a fluorescent-free detection method. However, the sensitivity and the detection limit of TIR microscopy depends completely on the type of labeling of the target molecules. Over the years, organic fluorophores have become the most widely used type of label for biomolecule detection (Goncalves, 2009; He et al., 2015; Maxwell et al., 2002). Nevertheless, the use of fluorophores in TIR microscopy for biomolecule detection carries the considerable drawbacks of rapid photobleaching and photofluctuation (Mattheyses et al., 2010; Waters, 2009). The use of oxygen scavengers can reduce but not prevent photobleaching of the fluorophores (Rasnik et al., 2006). Semiconductor quantum dots (Qdots) (Gao et al., 2004, 2005; Rakovich et al., 2015) have been developed as a potential alternative for fluorophores, but the relatively higher toxicity of Qdots (Tsoi et al., 2013) remains a concern for biomolecule detection. Recently, plasmon NPs have served as a better alternative for biomolecule labeling due to their greater scattering intensity (Lee et al., 2015; Nallathamby et al., 2008), higher photostability (Huang et al., 2008), and larger scattering cross section (Fan et al., 2014) compared to organic fluorophores. In plasmon NPs, evanescent waves induce conduction electrons to resonate at a particular frequency, resulting in light scattering. The intensity of scattered light in the EFL is sufficient for effective quantitative detection of biomolecules at the single-particle level. In the present work, a nanoimmunosensor probe for detection

of CA125 was investigated for the first time using TIR plasmonic scattering on a nanoarray chip. By comparing the TIR fluorescence (TIRF) with fluorescent-free TIR scattering (TIRS) in the immunoassay (Fig. 1B), we demonstrated the possibility of TIRSM based on plasmon NP probes for highly sensitive detection of biomolecules on nanoarray chips. The detection method is applicable to various types of biomolecules; CA125 was used as a model disease-related protein in the present work. However, the wider importance of this work was to show the application and versatility of TIR microscopy with fluorescent and fluorescent-free labeling for ultra-sensitive biomolecule detection to enable earliest-stage detection of various types of diseases.

2. Materials and methods 2.1. Functionalization of silver NPs The method used to functionalize silver NPs (SNPs) was slightly modified from a previously reported method (Lee et al., 2013). A colloidal solution of 20 nm SNPs (7.00  1010 particles/mL) purchased from BBI Solutions (Llanishen, Cardiff, UK) was functionalized before bioconjugating with a secondary antibody. For functionalization, 10 mM 11-mercaptoundecanoic acid (95% MUA) and 30 mM 6-mercapto-1-hexanol (97% MCH) purchased from SigmaAldrich Inc. (St. Louis, MO, USA) were dissolved in ethanol and added to the SNP solution. The resulting SNP-MUA-MCH mixture

S.K. Chakkarapani et al. / Biosensors and Bioelectronics 81 (2016) 23–31

was sonicated for 10 min and allowed to react at room temperature for 150 min, during which time MUA-MCH formed monolayers on the SNP surfaces. The resulting mixture was washed with deionized water by means of centrifugation (12,000 rpm, 90 min, 4 °C) and the residue of this mixture was dissolved in 50 mM 2(morphozlino)ethanesulfonic acid (MES, Sigma-Aldrich). All aqueous solutions were prepared from deionized water purified with a water purification system (Human Power 1 þ, Human Corporation, Seoul, Korea). 2.2. SNP and fluorescent dye labeling of secondary antibody 1-Ethyl-3-(3-dimethylamino-propyl)carbodimide hydrochloride (EDC, Sigma-Aldrich), N-hydroxysulfosuccimide (NHSS, Pierce, Rockford, IL, USA) and 1  phosphate-buffered saline (PBS, Sigma-Aldrich) were used for the conjugation of SNPs with secondary antibody X75 (HyTest, Joukahaisenkatu, Turku, Finland) for TIRS. Four microliter of each EDC (2 mg/mL in 50 mM MES) and NHSS (2 mg/mL in 1  PBS) were added to the SNP-MUA-MCH mixture and stirred for 40 min at room temperature. The solution was washed by means of centrifugation (12,000 rpm) at 4 °C for 30 min and the residue was dissolved in PBS buffer. Monoclonal secondary antibody X75 (10 μg/mL in 1  PBS) was added to the mixture and was allowed to react in a rotatory shaker for 4 h at room temperature, followed by incubation for 12 h at 4 °C. The mixture was centrifuged (12 000 rpm, 30 min, 4 °C) to remove the unreacted antibody, and the residue was suspended in PBS buffer. Separately, for TIRF, X75 was labeled with Alexafluor 488 donkey anti-mouse IgG (Life Technologies, Carlsbad, CA, USA) before immunoreaction with the target antigen (Pal et al., 2015). Twenty microliters of Alexafluor 488 (2 μg/mL in 1  PBS) was added to 20 μL of X75 (2 μg/mL in 1  PBS) and allowed to react in a rotatory shaker for 1 h at room temperature; the resulting labels were stored at  20 °C. 2.3. Immunoreaction on gold-nanopatterned chips Gold nanopad arrays fabricated according to a previously reported method (Lee and Kang, 2013) were obtained from the Korea Advanced Nano Fab Center (KANC) (Suwon, South Korea). These arrays were 4  4 arrays of Au spots with 500 nm spot diameter and 10 mm pitch. Before fabrication, each 10 mm2 glass wafer substrate was cleaned using a piranha solution (3:1 v/v of H2SO4 and 30% H2O2) and coated with a 100 nm–thick layer of ZEP520A electron beam resist (Zeon Co., Chiyoda-ku, Tokyo, Japan). The wafer was then subjected to electron beam lithography by using JBX-9300FS (JEOL, Tokyo, Japan) equipment operated at a beam current of 100 pA and an accelerating voltage of 100 keV. Au/Cr pads of 20/5 nm thickness were deposited on the glass by means of a thermal evaporation method, and the resist was then removed by treatment with dimethylacetamide. Each fabricated gold-nanopatterned chip (GNC) was cleaned before immunoreaction by immersing the chip for 30 min in a piranha solution (1:1 v/ v of H2SO4 and 30% H2O2) and then for 30 s each in acetone (99.5% purity) and isopropyl alcohol (99.9% purity). Immunoreactions were performed at room temperature according to a previously reported procedure (Lee et al., 2015). GNC was made to react in a stepwise manner with 4 mg/mL of dithiobis(succinimidyl propionate) (DSP, Pierce) in dimethyl sulfoxide (DMSO, Sigma-Aldrich) for 30 min, and with 0.1 mg/mL of protein A/G (Pierce) for 1 h; the unreacted succinimide was then blocked with 10 mM Tris (pH 7.5) and 1 M glycine for 30 min. The chip was immersed in a stabilizing guard (SurModics, Eden Prairie, MN, USA) for 30 min and then reacted with 20 μL of 2 μg/mL primary antibody X306 (HyTest) for 1 h, followed by reaction with 20 μL of target antigen CA125 (HyTest, 80 U/mL to 0.0001 U/mL) for 1 h. After each step, the chip

25

was washed with deionized water. 2.4. Custom-made TIR microscopy system A TIR microscopy system (Fig. 1A) was equipped on an upright Olympus BX 51 microscope (Olympus Optical Co., Ltd., Shinjukuku, Tokyo, Japan) with an electron multiplying cooled chargecoupled device (EM-CCD) camera (512  512 pixel imaging array, QuantEM 512 SC, Photometrics, Tucson, AZ, USA). An additional laser was added to the previously reported setup of this system (Lee et al., 2015). Three lasers were used: a 671 nm laser (SDL-671040T, Shanghai Laser & Optics Century Co., Ltd., Shanghai, China), a 405 nm laser (Compact-30G-405, World Star Tech., Toronto, ON, Canada), and a 473 nm laser (SL-473 nm laser-200 mW, Korea Lasertronix, INC., Seoul, Korea). Wavelength-dependent TIRS/TIRF images were acquired using an objective lens with numerical aperture (NA) 0.6–1.3 (UPLANFLN,  100) and using filter sets 406/ 15 nm, 473/10 nm, 520/15 nm, and 680/10 nm (Semrock Inc., Rochester, NY, USA). The exposure time of 100 ms was maintained using a shutter controller and data was analyzed using MetaMorph 7.1 software (Molecular Devices, LLC, Sunnyvale, CA, USA). 2.5. Strategy for determination of LOD LOD was determined for both TIRS and TIRF, by calculating the intensity value for various CA125 concentrations ranges from 80 U/mL to 0.0001 U/mL. To calculate the intensity, a posterior (after acquisition) background correction was carried for each spot with a region of interest (ROI) tool. The ROI was constructed manually around each spot (500 nm GNC) to calculate the intensity value. The background ROI (same shape and number of pixels as the ROI of the GNC) was translated in x/y to a nearby region of the spot that was the representative of the background intensity. Thereby, we subtracted the intensity of the background from the signal of each spots. We averaged the intensity of all the 16 spots in the nanoarray chip for each concentration, and LOD was typically determined to be in the region where the S/N was three (S/N ¼3).

3. Results and discussion 3.1. Wavelength-dependent TIRS detection of NPs The higher refractive index sensitivity of Au (44 nm/refractive index unit (RIU)) (Alvarez et al., 1997) and of Ag (161 nm/ RIU) (Jain and El-Sayed, 2010) makes them a valuable tool in biomolecule detection by means of TIRS. In this work, the 20 nm SNPs (fluorescence-free immunosensor probe) and the 500 nm GNC (nanoarray chip) were the two plasmonic materials used for TIRS detection of the target biomolecule CA125. The scattering signals of both the 20 nm SNPs and the 500 nm GNC were analyzed separately before their immunoreaction with CA125. Plasmonic scattering of NPs depends on their areal density, the radius of the plasmon, the dielectric constant of the surrounding medium, absorbed radiation, and the imaginary and real portions of the NPs' relative permittivities (Negre and Sanchez, 2013). By keeping all the above parameters constant, the scattering signal of the GNC was observed under various incident wavelengths of electromagnetic radiation, using bandpass filters (Supplementary Fig. S1). The observed scattering intensity was greatest for the incident wavelength of 671 nm (680/10 nm bandpass filter; Supplementary Fig. S1A), followed by that for the incident wavelengths of 473 nm (473/10 nm bandpass filter; Supplementary Fig. S1B) and 405 nm (406/15 nm bandpass filter; Supplementary Fig. S1C). Scattering signals were observed when the central wavelength of the

26

S.K. Chakkarapani et al. / Biosensors and Bioelectronics 81 (2016) 23–31

bandpass filter matched the incident wavelength. The maximum scattering efficiency of the 500 nm GNC of around 670 nm agreed with the previously reported resonant scattering peak wavelength obtained by TIR spectroscopy (Lee et al., 2013). SNPs were labeled with the secondary antibody X75 before immunoreaction with the target antigen (CA125) on GNC. Silver and gold NPs possess strong absorption in the visible region, and surface modification of these NPs leads to unique changes in their extinction in the UV–vis region. The dependence of the extinction wavelength maximum upon surface modification of the plasmonic NPs can be expressed as follows (Hammond et al., 2014):

⎡ ⎛ 2d ⎞⎤ Δλ max = mΔn⎢1 − exp⎜− ⎟⎥, ⎝ ld ⎠⎦ ⎣ where λmax is the NP extinction wavelength maximum; m and n are the refractive indexes of the bulk NP and the absorbed surface layer, respectively; d is the effective thickness of the adsorbed layer, and ld is the characteristic electromagnetic decay length. The absorption maxima of the SNPs were observed by means of UV– visible spectroscopy both before and after binding with the secondary antibody (Supplementary Fig. S2). The absorption maximum of 407 nm for the bare SNP shifted to 421 nm after bioconjugation with the secondary antibody X75. This bathochromic shift of about 14 nm in the visible region confirms the bioconjugation of the SNPs with X75. The scattering efficiency of the 20 nm SNP was calculated by

(A) SPR of SNP

means of discrete dipole approximation (DDA) to be maximal under incident light of wavelength 396 nm (Zhang, 2011). Therefore, a laser of the similar wavelength of 405 nm, equipped with a 406/15 nm bandpass filter, was used in the present work to acquire TIRS images of SNPs. Plasmon scattering of the metal nanoparticles was altered by their surface modification (Fig. 2A and B). Bioconjugation on the surface of the SNP changes its local refractive index, thereby changing the scattering efficiency of the SNP. Therefore, the relative scattering intensities of the SNPs before (Fig. 2C) and after (Fig. 2D) binding with the secondary antibody X75 were observed on a bare cover glass under the incident wavelength of 405 nm at 30 mW laser power. The scattering intensity of the SNPs decreases 24% from its initial value after bioconjugation with the secondary antibody X75. This scattering intensity difference and an accompanying bathochromic shift in the UV–vis spectra confirmed the bioconjugation of the SNPs with X75. Moreover, the scattering intensities of the SNPs were observed with respect to the different concentration of the X75; the scattering intensity decreased linearly with increasing concentration of X75 (Supplementary Fig. S3). 3.2. Nanoimmunoassay of CA125 on GNCs The primary antibody, the target antigen, and the SNP/Alexafluor 488–labeled secondary antibody were immunoreacted on the GNC and analyzed using TIR microscopy. Immunoassay detection of biomolecules by TIR microscopy was highly wavelength-

(B) SPR of bio-conjugated SNP

E

E

Bio-conjugated 20 nm SNP

20 nm SNP

20 18 16 14 12 10 8 6 4 2 0

(D) TIRS of bio-conjugated SNP

RSI × 10 2 (a.u.)

RSI × 10 2 (a.u.)

(C) TIRS of SNP

0

1

2

Pixels

3

4

5

20 18 16 14 12 10 8 6 4 2 0

0

1

2

3

4

5

Pixels

Fig. 2. (A, B) Schematic illustrations of electric field (E) in surface plasmon resonance of SNP (A) before and (B) after bioconjugation with secondary antibody X75. (C, D) Relative scattering intensity (RSI) of 20 nm SNPs (C) before and (D) after bioconjugation with X75. SNPs were illuminated with a 30 mW 405 nm laser and imaging was carried with a 406/15 nm bandpass filter (n¼ 11). Scale bar¼ 100 nm.

S.K. Chakkarapani et al. / Biosensors and Bioelectronics 81 (2016) 23–31

dependent. To confirm immunoreaction on a GNC, the scattering intensity of the GNC was compared before and after immunoreaction under incident radiation by 405 nm, 473 nm, and 671 nm lasers, including the use of various bandpass filters between the sample and the imager (Lee et al., 2015). As discussed earlier, the GNC of 500 nm spot diameter possessed higher scattering intensity under 671 nm illumination and comparatively lower scattering intensity under 405 nm illumination (Supplementary Fig. S1). After immunoreaction, the scattering intensity of the GNC was reduced by 3.7 times under 671 nm illumination (Lee et al., 2015) and increased to 7.2 times the initial value under 405 nm incident radiation (Supplementary Fig. S4A). Thus, the present results confirmed the immunoreaction of target molecule CA125 labeled with 20 nm SNP for TIRSM biomolecule detection. In addition, the 473 nm laser was used for immunoassay detection by TIRFM. Before immunoreaction, no fluorescence from the GNC was apparent when viewed through a 520/15 nm bandpass filter. However, after immunoreaction of CA125 labeled with Alexafluor 488, fluorescence from the GNC was clearly detected through the same bandpass filter. This result confirms the immunoreaction of CA125 on GNC with Alexafluor 488 labeling (Supplementary Fig. S4B). Once the immunoreaction was confirmed for both TIRSM and TIRFM detection, these two immunosensors were compared to assess their sensitivity for biomolecule detection.

27

3.3. Comparison of photobleaching and photofluctuation Photostability is the most important property for a biomolecule label to possess to enable accurate and precise quantitative analysis by means of TIR microscopic methods. The drawback of using an organic fluorophore as a fluorescent dye for biomolecule detection is the irreversible photobleaching of the fluorescence. To compare the photobleaching resistance of plasmonic NPs and fluorophores, GNCs were immunoreacted with the target antigen (CA125) and labeled separately with 20 nm SNPs and Alexafluor 488. The scattering intensity of the GNC labeled with 20 nm SNPs was observed under 30 mW 405 nm laser illumination (Fig. 3A) and the fluorescence intensity of the GNC labeled with Alexafluor 488 was observed under 30 mW 473 nm laser illumination (Fig. 3B). The scattering intensity of the GNC labeled with 20 nm SNPs was greater than the respective fluorescence intensity for the fluorophore-labeled GNC under the same concentration of the target antigen CA125. After 60 s of irradiation, the fluorescence intensity decreased by 2.63% of the initial value; it was decreased by 41.23% after 300 s and by 82.31% after 450 s. Contrastingly, the scattering intensity of the SNP-labeled GNC was more stable, decreasing by only 3.65% from the initial intensity after 450 s (Fig. 3C). Thus, the average scattering intensity after 450 s was 9.7 times the corresponding average fluorescence intensity.

(A)TIRSM 0s

60 s

120 s

180 s

240 s

300 s

360 s

450 s

(B)TIRFM

(C) Photobleaching 30

30

25

N.I × 102 (a.u.)

25

N.I × 102 (a.u.)

TIRSM

3.65%

20 TIRSM 15

TIRFM

10

TIRFM

20 15

82.31%

10 5

5

0

0 0

60

120

180

240

Time (s)

300

360

420

0

450

0

450

Time (s)

Fig. 3. (A, B) TIR microscope images of labeled CA125 immunoreacted on GNC arrays of 500 nm Au spots: (A) TIRS images with 20 nm SNP labels and (B) TIRF images with Alexafluor 488 labels. (C) Normalized intensity of TIRS (blue) and TIRF (green), demonstrating the difference in photobleaching and photostability between the biomarkers. TIRS intensity decreased 3.65% after 450 s relative to the initially observed intensity, whereas TIRF intensity decreased 82.31% relative to the initial intensity. The TIRS image was acquired using 30 mW 405 nm laser illumination and the TIRF was acquired using 30 mW 473 nm laser illumination. Scale bar ¼1 μm. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

28

S.K. Chakkarapani et al. / Biosensors and Bioelectronics 81 (2016) 23–31

Photobleaching of fluorophores occurs due to irreversible covalent modification when they react with other molecules, such as excited-state oxygen (Bernas et al., 2005; Greenbaum et al., 2000). In the case of SNPs, photobleaching occurs due to photooxidation of the silver (Han et al., 2011). Even though SNPs are susceptible to photooxidization and photobleaching under prolonged irradiation by a laser of specific wavelength, the present results showed that their rate of photobleaching was much less

than that of the fluorophore. Thus, surface electrons of the plasmonic NPs are scattered upon illumination with the laser, and this scattering continues even after long exposures to the same laser source. The intensity of both the fluorophore and the NPs depended upon the laser power; thus, the same laser power of different wavelengths was used to compare the photobleaching resistance of the NPs and the fluorophore. The photostability was assessed in terms of the normalized intensity of the 20 nm SNPs

(A) TIRSM 0.07 U/mL

0.05 U/mL

0.03 U/mL

0.05 U/mL

0.03 U/mL

(B) TIRFM 0.07 U/mL

(C) Intensity comparison 30

0.05 U/mL

Intensity (a.u.)

0.07 U/mL

0.03 U/mL

25 20 15 10

0

10

20

Pixels

30

40

0

10

20

Pixels

30

40

0

10

20

30

40

Pixels

Fig. 4. (A) TIRS and (B) TIRF images of GNCs immunoreacted with different concentrations of the target antigen CA125 (n¼ 11). Scale bar¼ 500 nm. (C) Plot shows the scattering intensity (blue) of NP-CA125 and the fluorescence intensity (green) of Alexafluor 488-CA125 on GNCs. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

S.K. Chakkarapani et al. / Biosensors and Bioelectronics 81 (2016) 23–31

and the Alexafluor 488 after background subtraction. The SNPs remained nearly constant in intensity over time; meanwhile, the Alexafluor 488 fluctuated considerably from the mean intensity. These results demonstrated that the plasmonic NPs were both photoresistant and photostable, and do not suffer photobleaching and photofluctuation like organic fluorophores. Thus, plasmonic NP labeling as a fluorescence-free probe gave a more accurate and precise result than fluorophore labeling of the biomolecules.

a wide range of concentrations. For LOD experiments, the target antigen CA125 was diluted to various concentrations from 0.0001 to 80 U/mL, with same concentrations of primary and secondary antibody. The antigen was immunoreacted on a GNC and labeled with Alexafluor 488 for TIRF detection and 20 nm SNPs for TIRS detection. To calculate the scattering intensity for TIRSM and the fluorescence intensity for TIRFM, the intensity of all 16 Au spots in the GNC array was averaged after subtracting the background signal for each spot. Because of the higher photostability and sensitivity of SNPs over the fluorophore, the SNP labeling was expected to yield a lower LOD than the fluorophore labeling. Under 30 mW illumination by a 473 nm laser, the LOD of CA125 by Alexafluor 488 labeling was measured to be 0.003 U/mL (Fig. 5A and Supplementary Fig. S5A), which was 86 times lower than a previously reported LOD value achieved by using surface plasmon resonance spectroscopy (Pal et al., 2015). Detection with TIRSM was carried out using the same concentrations and optical setup, and using 30 mW illumination by a 405 nm laser. The LOD achieved by using the 20 nm SNP labels was 0.0018 U/mL (Fig. 5B and Supplementary Fig. S5B), nearly half the TIRFM LOD and 144 times lower than the previously reported value (Pal et al., 2015) in which a magnetic nanoparticle-antibody conjugate was used and detected by spectrophotometer. Moreover, the relative standard deviation (RSD) values for different concentrations of CA125 (Supplementary Table S1) shows the reliability of the LODs calculated with TIRS and TIRF detection. The results show the sensitive of the biomolecule detection without any surface modification or scattering intensity enhancement of the labeled NPs.

3.4. Sensitivity comparison of nanoimmunosensors The molar extinction coefficients of the 20 nm SNPs and of Alexafluor 488 were 9.21  108 M  1 cm  1 and 7.3  104 M  1 cm  1, respectively. Either the molar extinction coefficient is the amount of photons absorbed by the fluorophore or the extinction (summation of scattering and absorption) of the nanoparticle at a given wavelength (Liu et al., 2007). Hence, the NPs were more sensitive than the fluorophore. To compare the detection sensitivity of the GNCs based on the NP and fluorophore labels, GNCs were immunoreacted with different concentrations of target antigen CA125 and labeled separately with 20 nm SNPs (Fig. 4A) and Alexafluor 488 (Fig. 4B). A single spot was selected on each array and the scattering intensity and fluorescence intensity were compared for each concentration; in each case, the scattering intensity was approximately 1.14 times higher than the fluorescence intensity. This result demonstrated the higher sensitivity of nanoparticles over fluorophores at lower concentrations of target antigen. Therefore, plasmonic NPs could be suitable labeling tools for highly sensitive detection of biomolecules. As the sensitivity of NP was observed to be greater than that of the Alexafluor 488, plasmon NP labeling allowed detection of lower concentrations of target antigen relative to fluorophore labeling.

3.6. Quantification of CA125 in serum samples by TIRSM To show the clinical application of the CA125 immunoassay detection on chip, serum samples were spiked with the known concentrations of antigen CA125 and quantified using the TIRSM (Fig. 6) (See ESI for Section 2). The RSI value obtained for the normal serum sample (without spiking) was correlated with the proposed calibration curve (Fig. 5) and the concentration of CA125 was quantified to be 0.0974 70.0012 U/mL (n ¼11). Known concentrations of CA125 (0.01 U/mL, 1 U/mL, 50 U/mL) were then spiked with the serum samples and RSI values were obtained for each concentration (Fig. 6B). The RSI values were correlated with

3.5. Comparison of quantitative analysis by TIRSM and TIRFM The detection sensitivity of the TIR microscopy system was demonstrated by measuring the LODs for the target antigen. A key advantage of TIR microscopy as a detection tool is its low LODs combined with its wide linear dynamic ranges. The concentration of cancer antigen CA125 indicates the various stages of ovarian cancer; accordingly, it is important to be able to detect CA125 over

(B) TIRSM

12 11 10 S/N = 3 9 8 7 6 LOD = 0.0030 U/mL 5 4 3 2 1 0 -3 -2 -1

RSI (×10 3 )

(A) TIRFM

RFI (×10 3 )

29

y = 2273.5x +5508.5 R = 0.9952

0

1

Log [CA125 (U/mL)]

2

3

13 12 11 10 S/N = 3 9 8 7 LOD = 0.0018 U/mL 6 5 y = 2389.6x +6304.1 R = 0.9911 4 3 2 1 0 3 -3 -2 -1 0 1 2

Log [CA125 (U/mL)]

Fig. 5. Standard calibration curves obtained by labeling the target antigen CA125 with (A) a fluorophore, Alexafluor 488, and (B) a fluorescence-free probe, plasmonic NPs (20 nm SNPs), both for CA125 concentrations ranging from 0.0001 to 80 U/mL.

30

S.K. Chakkarapani et al. / Biosensors and Bioelectronics 81 (2016) 23–31

(A) TIRSM of spiked serum 0 U/mL

0.01 U/mL

RSI (× 10 3 )

(B) RSI of spiked serum

1 U/mL

(C) % Recovery Spiked conc. (U/mL)

12 11 10 9 8 7 6 5 4 3 2 1 0

Theoretic al conc. (U/mL)

0.01

1

50

Mean recovered conc. ± S.D. (U/mL)

Recovery ± S.D. (%)

0.0974 ± 0.0012

0

0

50 U/mL

0.01

0.1074

0.1014 ± 0.0008

94.41 ± 0.74

1

1.0974

1.0702 ± 0.0005

97.52 ± 0.04

50

50.0974

50.0124 ± 0.0002

99.83 ± 0.01

Spiked conc. (U/mL)

Fig. 6. (A) TIRS images (Scale bar¼ 1 μm) and (B) RSI value of serum samples spiked with different concentrations (0 U/mL, 0.01 U/mL, 1 U/mL, 50 U/mL) of CA125 (n¼ 11). (C) Percentage recovery of the CA125 in the spiked samples.

proposed calibration curve to quantify the concentrations recovered from the spiked samples. TIRSM quantification of the spiked samples showed an excellent recovery percentage of 94.41 70.74%, 97.52 70.04%, and 99.83 70.01% with respect to 0.01 U/mL, 1 U/mL, 50 U/mL spiked samples (Fig. 6C). Thus, the proposed calibration curve can be used as a standard for detecting CA125 in serum samples with high accuracy and sensitivity. The results shows that the TIRSM-based fluorescence-free nanoimmunosensor probe for CA125 detection on chip was the most appropriate tool for detecting CA125 in serum samples with high sensitive and reliability.

4. Conclusions Quantitative analysis of biomolecules requires a highly sensitive imaging tool with adequate labeling probes. Disease-related biomolecules must be detected with the lowest possible LOD and over a wide range of concentrations; for the present example of CA125, the stage of ovarian cancer can be detected based on the concentration of CA125 in blood. Various immunosensors have been previously developed with different labels for detecting disease-related biomolecules with lowest possible LODs. Nanoimmunosensing based on TIRS microscopy is a highly sensitive biomolecule detection method, but requires a suitable label for

effective detection. In this work, target antigen CA125 was immunoreacted on GNCs and labeled separately with 20 nm SNPs and Alexafluor 488 to study its detection by means of TIRSM and TIRFM, respectively. Compared to the fluorophore Alexafluor 488, the SNPs showed much greater resistance to photobleaching, better photostability, and higher S/N. Moreover, the LOD of the SNP labeling was nearly half that of the Alexafluor 488 labeling. The ability of TIR microscopy to detect a wide range of biomolecule concentrations was demonstrated with both SNP and Alexafluor 488 labeling, showing the reliability of the method for detecting the serum level of the target antigen CA125 as a means to detect ovarian cancer and monitor its stage. These results demonstrated the versatility of the TIR microscopic technique and the significant potential of plasmonic NP labeling over fluorophore labeling for highly sensitive biomolecule detection. CA125 was demonstrated as a model biomolecule; this method could be applied well to the detection of any biomolecule, paving the way for earliest-stage disease diagnosis.

Acknowledgments This research was supported by the Basic Science Research Program through the National Research Foundation of Korea

S.K. Chakkarapani et al. / Biosensors and Bioelectronics 81 (2016) 23–31

(NRF), funded by the Ministry of Education, Science, and Technology (No. 2015R1A2A2A01003839).

Appendix A. Supplementary material Supplementary data associated with this article can be found in the online version at http://dx.doi.org/10.1016/j.bios.2016.01.094.

References Alvarez, M.M., Khoury, J.T., Schaff, T.G., Shaffigullin, M.N., Vezmor, I., Whetten, R.L., 1997. J. Phys. Chem. B 101, 3706–3712. Axelrod, D., Burghardt, T.P., Thompson, N.L., 1984. Annu. Rev. Biophys. Bioeng. 13, 247–268. Axelrod, D., 2013. Biophys. J. 104, 1401–1409. Aroonyadet, N., Wang, X., Song, Y., Chen, H., Cote, R.J., Thompson, M.E., Datar, R.H., Zhou, C., 2015. Nano Lett. 15, 1943–1951. Bast, R.C., Feeney, M., Lazarus, H., Nadler, L.M., Colvin, R.B., Knapp, R.C., 1981. Clin. Investig. 68, 1331–1337. Bernas, T., Robinson, J.P., Asem, E.K., Rajwa, B., 2005. J. Biomed. Opt. 10, 064015. Chen, H., Li, J., 2007. Methods Mol. Biol. 381, 411–436. Chung, J.W., Kim, S.D., Bernhardt, R., Pyun, J.C., 2005. Sens. Actuators B: Chem. 111, 416–422. Etzioni, R., Urban, N., Ramsey, S., McIntosh, M., Schwartz, S., Reid, B., Radich, J., Anderson, G., Hartwell, L., 2003. Nat. Rev. Cancer 3, 243–252. Fan, X., Zheng, W., Singh, D.J., 2014. Light Sci. Appl. 3, e179. Gao, X., Cui, Y., Levenson, R.M., Chung, L.W.K., Nie, S., 2004. Nat. Biotechnol. 22, 969–976. Gao, X., Yang, L., Petros, J.A., Marshall, F.F., Simons, J.W., Nie, S., 2005. Curr. Opin. Biotechnol. 16, 63–72. Goncalves, M.S., 2009. Chem. Rev. 109, 190–212.

31

Greenbaum, L., Rothmann, C., Lavie, R., Malik, Z., 2000. Biol. Chem. 381, 1251–1258. Han, Y., Lupitskyy, R., Chou, T.-M., Stafford, C.M., Du, H., Sukhishvili, S., 2011. Anal. Chem. 83, 5873–5880. Hammond, J.L., Bhalla, N., Rafiee, S.D., Estrela, P., 2014. Biosensors 4, 172–188. He, X.P., Zang, Y., James, T.D., Li, J., Chen, G.R., 2015. Chem. Soc. Rev. 44, 4239–4248. Hoppener, C., Novotny, L., 2012. Q. Rev. Biophys 45, 209–255. Huang, T., Nallathamby, P.D., Xu, X.-H.N., 2008. J. Am. Chem. Soc. 130, 17095–17105. Huckabay, H.A., Wildgen, S.M., Dunn, R.C., 2013. Biosens. Bioelectron. 45, 223–229. Jain, P.K., El-Sayed, M.A., 2010. Chem. Phys. Lett. 487, 153–164. Lee, K.B., Kim, E.Y., Mirkin, C.A., Wollinsky, S.M., 2004. Nano Lett. 4, 1869–1872. Lee, S., Park, G., Chakkarapani, S.K., Kang, S.H., 2015. Biosens. Bioelectron. 63, 444–449. Lee, S., Kang, S.H., 2013. Analyst 138, 3478–3482. Lee, S., Kang, S.H., 2014. Biosens. Bioelectron. 60, 45–51. Lee, S., Yu, H., Kang, S.H., 2013. Chem. Commun. 49, 8335–8337. Liu, X., Atwater, M., Wang, J., Huo, Q., 2007. Colloids Surf. B: Biointerfaces 58, 3–7. Mattheyses, A.L., Simon, S.M., Rappoport, J.Z., 2010. J. Cell Sci. 123, 3621–3628. Maxwell, D.J., Taylor, J.R., Nie, S., 2002. J. Am. Chem. Soc. 124, 9606–9612. Nallathamby, P.D., Lee, K.J., Xu, X.-H.N., 2008. ACS Nano 2, 1371–1380. Negre, C.F.A., Sanchez, C.G., 2013. Optical properties of metal nanoclusters from an atomistic point of view. In: Mariscal, M.M., Oviedo, O.A., Leiva, E.P.M. (Eds.), Metal Clusters and Nanoalloys: From Modelling to Applications. Springer, New York, NY, USA, pp. 105–157. Prieve, D.C., Luo, F., Lanni, F., 1987. Faraday Discuss. Chem. Soc. 83, 297–307. Pal, M.K., Rashid, M., Bisht, M., 2015. Biosens. Bioelectron. 73, 146–152. Raamanathan, A., Simmons, G.W., Christodoulides, N., Floriano, P.N., Furmaga, W.B., Redding, S.W., Lu, K.H., Bast, R.C., McDevitt, J.T., 2012. Cancer Prev. Res. (Phila.) 5, 706–716. Rakovich, A., Albella, P., Maier, S.A., 2015. ACS Nano 9, 2648–2658. Rasnik, I., McKinney, S.A., Ha, T., 2006. Nat. Methods 3, 891–893. Shi, M., Zheng, J., Tan, Y., Li, J., Li, Y., Li, X., Zhou, Z., Yang, R., 2015. Anal. Chem. 87, 2734–2740. Tsoi, K.M., Dai, Q., Alman, B.A., Chan, W.C., 2013. Acc. Chem. Res. 46, 662  –671. Waters, J.C., 2009. J. Cell Biol. 185, 1135–1148. Zhang, Y.J., 2011. Plasmonics 6, 393–397.