Biosensors and Bioelectronics 70 (2015) 418–425

Contents lists available at ScienceDirect

Biosensors and Bioelectronics

http://d0956-56

n CorrE-m

journal homepage: www.elsevier.com/locate/bios

Ultrasensitive detection of drug resistant cancer cells in biologicalmatrixes using an amperometric nanobiosensor

Pranjal Chandra a,b, Hui-Bog Noh a, Ramjee Pallela a, Yoon-Bo Shim a,n

a Department of Chemistry and Institute of Biophysio Sensor Technology (IBST), Pusan National University, Busan 609-735, South Koreab Amity Institute of Biotechnology, Amity University Uttar Pradesh, Noida 201303, U.P., India

a r t i c l e i n f o

Article history:Received 5 February 2015Received in revised form10 March 2015Accepted 26 March 2015Available online 27 March 2015

Keywords:NanobiosensorDrug resistant cancer cellsElectrochemistryIn vitro testNanomaterials

x.doi.org/10.1016/j.bios.2015.03.06963/& 2015 Elsevier B.V. All rights reserved.

esponding author. Fax: þ82 51 514 2122.ail address: ybshim@pusan.ac.kr (Y.-B. Shim).

a b s t r a c t

Multidrug resistance (MDR) is a key issue in the failure of cancer chemotherapy and its detection will behelpful to develop suitable therapeutic strategies for cancer patients and overcome the death rates. Inthis direction, we designed a new amperometric sensor (a medical device prototype) to detect drugresistant cancer cells by sensing “Permeability glycoprotein (P-gp)”. The sensor probe is fabricated byimmobilizing monoclonal P-gp antibody on the gold nanoparticles (AuNPs) conducting polymer com-posite. The detection relies on a sandwich-type approach using a bioconjugate, where the aminophenylboronic acid (APBA) served as a recognition molecule which binds with the cell surface glycans andhydrazine (Hyd) served as an electrocatalyst for the reduction of H2O2 which are attached on multi-wallcarbon nanotube (MWCNT) (APBA-MWCNT-Hyd). A linear range for the cancer cell detection is obtainedbetween 50 and 100,000 cells/mL with the detection limit of 2372 cells/mL. The proposed im-munosensor is successfully applied to detect MDR cancer cells (MDRCC) in serum and mixed cell samples.Interferences by drug sensitive (SKBr-3 and HeLa), noncancerous cells (HEK-293 and OSE), and otherchemical molecules present in the real sample matrix are examined. The sensitivity of the proposedimmunosensor is excellent compared with the conventional reporter antibody based assay.

& 2015 Elsevier B.V. All rights reserved.

1. Introduction

Multidrug resistance (MDR) is a major factor in the failure ofcancer chemotherapy. Several mechanisms have been convincedto play vital roles in the development of the MDR in cancer cells(Krishna and Mayer, 2000). Among them, a major form is medi-ated by a cell membrane transporter “Permeability glycoprotein,(P-gp)” which is encoded by the MDR1 gene in human cells. Itincludes two nucleotide-binding and two membrane-spanningdomains, acts as energy dependent pump that decreases in-tracellular drug concentrations lower than the effective ther-apeutic levels (Bellamy, 1996; Gottesman et al., 2002). This phe-nomenon leads to the development of drug resistance in cancercells. The early detection of MDR in cancer cells can lead to theadequate chemotherapy of cancer patients and overcome themortality. Thus, it is extremely important and clinically significantto detect MDRCC at the early stage. This will help the clinicians todevelop alternate therapeutic strategies for cancer patients. Untilnow, several biological methods have been developed to detectP-gp expression for the diagnosis of MDR in cancer such as;

polymerase chain reaction (Murphy et al., 1990), im-munohistochemistry (Chan et al., 1990), flow cytometry (Lu-descher et al., 1992), and microarray (Gillet et al., 2004). Althoughthese methods can be used to detect MDRCC, they are less sensi-tive, require highly trained professionals, and have no ability to beminiaturized for the point-of-care applications. Thus, few bio-sensor-based diagnostic methods have also been reported to de-tect MDRCC. These methods, however, are indirect, nonselective,and less sensitive (Du et al., 2005; Zhang et al., 2011, 2014). Thus,the development of a sensitive, robust, and alternate method forthe detection of MDRCC directly in the biological fluids is desirableand has great clinical importance. To achieve a highly sensitivedetection method for MDRCC, a sandwich type-electrochemicalimmunosensor composed of P-gp monoclonal antibody (AntiP-gp)as a detection probe coupled with a ligand tagged-non-enzymaticcatalyst as a reporter probe is worthy to attempt due to their se-lective and highly sensitive characteristics, respectively.

The stable immobilization of biomaterials including antibodieson the sensor probe is extremely critical in the biosensor fabri-cation (Rahman et al., 2008). Electrochemical biosensors com-posed of conducting polymers-AuNPs composite are considered tobe highly stable and ultrasensitive because biomolecules can becovalently immobilized on the polymer backbone possessing–COOH or –NH2 groups (Chandra et al., 2011a, 2013; Koh et al.,

P. Chandra et al. / Biosensors and Bioelectronics 70 (2015) 418–425 419

2011; Kim and Shim, 2013; Lee et al., 2010). Compare to the re-porter antibody based conventional sandwiched type electro-chemical immunosensor (Malhotra et al. 2012; Wu et al., 2015;Zhu et al. 2014), other ligands are worthy to attempt for the in vitrodetection of cancer cells. This is important because it has beenreported that the concentration of antigen (target molecule) onthe cancer cell surface is low (Beck et al., 1996). This may result inlow antigen–antibody binding and consequently poor detection ofcancer cells. Thus, it is worthy to target molecules that are presenton the cancer cell surface in abundance. Recent studies haveclearly indicated existence of high concentration of glycans ontothe cancer cell membrane (Dube and Bertozzi, 2005; Zhang et al.,2010) due to which numerous diol-groups are present on the cellsurface. Arising from the unique capacity of boronic acids to formboronic esters with these diols (Das et al., 2011), it would be in-teresting to attempt boronic acid as an alternative recognitionmolecule (reporter probe) to detect cancer cell surface glycans.

To show the signal in an electrochemical immunosensor, anelectrochemical indicator is needed, such as a bioconjugate com-posed of enzyme or nonenzymatic catalyst. Compared to ex-pensive and easily denaturable enzyme, Hyd attached on a bio-conjugate can be used due to its small size, stability, and highcatalytic activity towards hydrogen peroxide (H2O2) reduction(Zhu et al., 2013). Thus, in this work we developed a sandwichimmunoassay approach having a chemo-nano-conjugate com-posed of APBA, MWCNT, and Hyd instead of enzyme linked re-porter antibody. In this conjugate, APBA and Hyd are used a re-porter molecule and electrocatalyst for signal generation,respectively.

In the present study, a novel amperometric immunosensor forthe detection of MDRCC has been tried to develop for the first timethrough the detection of P-gp. The immunosensor probe is fabri-cated by forming covalent bonds between AntiP-gp and carboxylicacid group-functionalized conducting polymer layer on AuNPsdeposited electrode surface. The nanocomposite was characterizedby atomic force microscopy (AFM), X-ray photoelectron spectro-scopy (XPS), electrochemical impedance spectroscopy (EIS), scan-ning electron microscopy (SEM), and the AntiP-gp immobilizationwas confirmed by quartz crystal microbalance (QCM) and XPS. Thesandwich immunosensing approach was adopted where APBA-MWCNT-Hyd was reacted with the MDRCC captured by the GCE/AuNPs/pTTBA/AntiP-gp probe. The immunoreaction was mon-itored in term of catalytic activity of Hyd towards H2O2 reduction.The experimental parameters were optimized and the detectionlimit of the MDRCC was determined. Direct detection of MDRCC inserum and in the mixed cell samples were performed to evaluatethe real clinical value of sensor. The selectivity of the biosensorwas also examined toward various non target cells, and chemicalspresent in the real sample matrix. The developed method was alsocompared with the conventional reporter antibody based method.

2. Material and methods

2.1. Materials

2,2′:5′, 2″-terthiophene-3′(p-benzoic acid) (TTBA) was synthe-sized through the Paal–Knorr pyrrole condensation reaction (Kohet al., 2011). Tetrabutylammonium perchlorate (TBAP, electro-chemical grade) was purchased from Fluka (USA) and purifiedaccording to a general method, followed by drying under vacuumat 1.33�10�3 Pa (Noh et al., 2012). 1-ethyl-3-(3-(dimethylamino)-propyl) carbodiimide (EDC), N-Hydroxysuccinimide (NHS), di-chloromethane (99.8%, anhydrous), trisodium citrate, sodium tet-rahydridoborate, HAuCl4 �3H2O, bovine serum albumin (BSA), andHydrazine sulfate were purchased from sigma Aldrich (USA).

Monoclonal p-GP antibody, amino-phenylboronic acid, indium tinoxide (ITO) glass, and H2O2 (33%) were purchased from sigmaAldrich (USA). MWCNTs (4–6 nm diameter, 95%) were obtainedfrom Iljin Nanotech (South Korea). phosphate buffer saline solu-tions (PBS) were prepared with 0.1 M of disodium Hydrogenphosphate, 0.1 M of sodium dihydrogenphosphate, and 0.9% so-dium chloride. cell culture medium, fetal bovine serum (FBS),trypsin-EDTA, penicillin/streptomycin, hank’s balance salt (HBS)solution, were purchased from Sigma-Aldrich (USA). All otherchemicals were of extra pure analytical grade and used withoutfurther purification. all aqueous solutions were prepared in ultra-pure water, which was obtained from a Milli-Q water purifyingsystem (18 MΩ cm).

2.2. Apparatus

All electrochemical experiments were performed in a threeelectrode cell. The modified glassy carbon electrode (GCE) (dia.3.0 mm), Ag/AgCl (in saturated KCl), and a platinum (Pt) wire wereused as working, reference, and counter electrodes, respectively.Voltammograms and chronoamperograms were recorded using apotentiostat/galvanostat, Kosentech, model KST P-2 (South Korea).A Multimode AFM device (Veeco Metrology) equipped with aNanoscope IV controller (Veeco) was used at ambient conditionsto get the images. The QCM experiment was performed using aSEIKO EG&G model QCA 917 and a PAR model 263A potentiostat/galvanostat (USA). An Au-coated working electrode (area:0.196 cm2; 9 MHz; AT-cut quartz crystal) was used for the QCMexperiment. The impedance spectra were obtained using a EG&GPrinceton Applied Research PARSTAT2263 at an open circuit vol-tage from 100.0 kHz down to 100.0 mHz and a sampling rate offive points per decade (AC amplitude: 10.0 mV). XPS was per-formed using a VG Scientific XPSLAB 250 XPS spectrometer and amonochromated Al Kα source with charge compensation at KBSI(Busan). The SEM images were obtained using a Cambridge Ste-reoscan 240. A JEOL JEM-2010 electron microscope (Jeol High-TechCo., Japan) with an acceleration voltage of 200 kV was used toobtain TEM images.

2.3. Preparation of APBA-MWCNT-Hyd conjugate

Firstly, MWCNTs were functionalized according to previouslyreported methods (Goldman and Lellouche, 2010; Piran et al.,2009). Briefly, 100 mg of MWCNT was treated with mixture ofconcentrated 12.0 M HNO3 and 36.0 M H2SO4 (90 °C, 2.0 h) fol-lowed by multiple rinsing with deionized water until no acid wasdetected. After drying at 80 °C under vacuum overnight, a blackpowder was obtained. Next, the treated MWCNT (3.0 mg) wasdispersed into 1.0 mL PBS (pH 7.0) containing 10.0 mM EDC/NHSsolutions and incubated at room temperature for 6 h to activatethe –COOH on the MWCNTs. The resulting mixture was separatedby centrifugation and the precipitate was washed three times.Meanwhile, an optimized 1.0 mg/mL hydrazine sulfate solutionand 10.0 mM APBA was prepared in PBS. Subsequently, the acti-vated MWCNTs were mixed with hydrazine sulfate and APBA so-lution. The mixture was stored in a refrigerator overnight at 4 °Cfollowed by centrifugation. The resulting deposit was washed fivetimes with PBS to remove any free APBA and Hyd. Finally, theprepared composite was dispersed in 1.0 mL PBS and stored in arefrigerator at 4 °C for further use.

2.4. Fabrication of immunosensor probe

The construction of the sensor is shown in Scheme 1. A layer ofthe polymer of the TTBA monomer was formed on the GCE/AuNPssurface through electropolymerization of 1.0 mM TTBA monomer

Scheme 1. Schematic representation of the immunosensor fabrication and detection principle.

P. Chandra et al. / Biosensors and Bioelectronics 70 (2015) 418–425420

solution, followed by immobilization of AntiP-gp on the surface.(details in Supporting material).

2.5. Cancer cells sample preparation

The P-gp overexpressed uterine sarcoma MDRCC lines (ATCCs

Number: CRL-1977™ and ATCCs Number: CRL-2274™) were ob-tained from the American Type Culture Collection (ATCC), Mana-ssas, USA. Control experiments and interference studies wereperformed using SKBr-3, HeLa, OSE, HEK-293 cell lines which wereobtained from Korean cell line bank. The cells were grown at 37 °Cin 5% CO2 atmosphere in the appropriate medium supplementedwith a 10% heat-inactivated FBS, 100 units/mL of penicillin andstreptomycin. The MDRCC lines were cultured in McCoy’s 5Amedium and maintained with 4.0 μg/mL adriamycin. Before eachexperiment, cells were suspended in autoclaved PBS and countedusing a disposable C-Chip (South Korea) under an optical micro-scope (Olympus).

2.6. Detection of MDR cancer cells with the sensor probe

The GCE/AuNPs/pTTBA/AntiP-gp sensor probe was incubatedwith enumerated MDRCC (CRL-1977™) for appropriate time(30 min) and then it was washed with the same buffer to removethe unbound cells. Next, the GCE/AuNPs/pTTBA/AntiP-gp/MDRCC

electrode was incubated with APBA-MWCNT-Hyd for 30 min toform GCE/AuNPs/pTTBA/AntiP-gp/MDRCC/APBA-MWCNT-Hydprobe followed by washing with the same buffer solution. The fi-nal GCE/AuNPs/pTTBA/AntiP-gp/MDRCC/APBA-MWCNT-Hyd probewas tested by cycling the potential between þ0.6 and �0.7 V at ascan rate of 50.0 mV/s. The chronoamperometric experiment wascarried out by applying a potential of �0.45 V vs. Ag/AgCl at theGCE/AuNPs/pTTBA/AntiP-gp/MDRCC/APBA-MWCNT-Hyd probeinto a deoxygenated 0.1 M PBS containing 4.0 mM of H2O2 (opti-mized) to obtained the catalytic response.

3. Results and discussion

3.1. Characterization and morphology of electrode surface

Foremost, the AuNPs were electrodeposited onto the GCE usingthe potential step method and confirmed AuNPs deposition

employing linear sweep voltammetry (LSV). During the electro-deposition of AuNPs onto bare GCE, the peak currents increased asthe number of sweep increased indicating the deposition of AuNPsand increase in the conductivity of GCE/AuNPs surface (figure notshown). The results obtained for the electrochemical deposition ofAuNPs are in agreement with our previously reported results(Chandra et al., 2011b). After that, pTTBA film was formed on theGCE/AuNPs surface by electropolymerization from a 1.0 mM TTBAmonomer solution. The monomer oxidation peak appeared atþ1.2 V during the anodic scan from 0.0 to þ1.4 V in a 1.0 mMmonomer solution. There was a distinct reduction peak at þ0.8 Vin the reverse cathodic scan from þ1.4 V, corresponding to thereduction of the oxidized polymer film formed on the GCE/AuNPssurface. These results related to nanocomposite preparation werein agreement with our previously reported results (Chandra et al.,2011a; Koh et al., 2011).

To confirm the fabrication of the sensor probe, the modificationsteps were characterized using SEM, AFM, XPS, QCM, and EIS.Supplementary material, Fig. S1(A) shows SEM images of the(a) GCE/AuNPs, (b) GCE/pTTBA, and (c) GCE/AuNPs/pTTBA elec-trodes. The SEM image of the GCE/AuNPs layer shows the ex-istence of the AuNPs of particle sizes of about 20.071.5 nm (Fig.S1(A(a))). The morphology of the GCE/AuNPs/pTTBA surface in theSEM image shows film of pTTBA over the deposited AuNPs (Fig. S1(A(c))). The TEM image also confirmed the AuNPs size to be20.071.5 nm (Fig. S1(B) in Supplementary material). Themorphologies of the polymer layers after electropolymerizationwere also observed by AFM in tapping mode. The (a) HOPG/AuNPs,(b) HOPG/pTTBA, and (c) HOPG/AuNPs/pTTBA layers were formedthrough the electropolymerization of the monomer coated on thehighly oriented pyrolytic graphite (HOPG) electrode surface (Fig. 1(A)). The polymer AuNPs-coated HOPG electrode surface exhibits ahomogeneous composition of small particles for the AuNPs/poly-mer films indicating the successful preparation of nanocomposite.The particle size of AuNPs/pTTBA film was determined to be70.5715.5 nm. The differences in the root-mean-square rough-ness of the surfaces among the AuNPs, pTTBA, and AuNPs/pTTBAfilms (3.56, 1.61, and 2.67 nm, respectively) were small. Next, wecharacterized the electrode surface using XPS. All XPS spectra weretaken after 50 s of Ar ion gas etching and calibrated with a C1speak at 284.6 eV as an internal standard. Fig. 1(B) shows the XPSspectra for the surfaces of (i) GCE/AuNPs/pTTBA, and (ii) GCE/AuNPs/pTTBA/AntiP-gp. The C1s spectrum for pTTBA exhibits two

Fig. 1. (A) Tapping mode AFM images of the (a) HOPG/AuNPs, (b) HOPG/pTTBA, and (c) HOPG/AuNPs/pTTBA films; images size is 500.0 nm�500.0 nm. (B) XPS spectra of(a) C1s, (b) S2p, and (c) N1s for (i) GCE/AuNPs/pTTBA and (ii) GCE/AuNPs/pTTBA/AntiP-gp. (C) Frequency changes during immobilization of AntiP-gp onto the Au-coated/pTTBA chip. (D) Nyquist plots obtained for bare GCE (black), GCE/AuNPs (red), GCE/pTTBA (magenta), and GCE/AuNPs/pTTBA (blue line) electrodes in a 0.1 M PBS (pH 7.4),containing 4.0 mM Fe[(CN)6]3�/4�/0.3 M NaClO4. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

P. Chandra et al. / Biosensors and Bioelectronics 70 (2015) 418–425 421

peaks at 284.6 and 286.5 eV that correspond to the C¼C, C–C, C–H,or C–S bonds, and a C¼O, C–O bonds, respectively (Fig. 1(B(a))).After immobilization of the antibody (GCE/AuNPs/pTTBA/AntiP-gp), a new C–N bond appeared at 285.1 eV due to the bond for-mation between –COOH groups of pTTBA and –NH2 groups ofAntiP-gp, the peak due to C¼O bond was shifted toward the po-sitive energy at 286.7 eV. The polymer coated film shows an S2ppeak at 163.7 eV corresponding to the C–S bond, which is due tothe sulfur component of pTTBA (Fig. 1(B (b))), while the peak at163.7 eV, however, was not observed for GCE/AuNPs surface (not

shown). After the immobilization of AntiP-gp, the peaks at 399.7and 399.1 eV appeared in the N1s spectrum corresponding to theC–N bond formation due to the covalent bond formation between–NH2 groups of AntiP-gp and –COOH groups of pTTBA indicatingsuccessful AntiP-gp immobilization (Fig. 1(B(c))). No peak at 399.7and 399.1 eV, however, was observed for GCE/AuNPs/pTTBA sur-face which clearly indicates that AntiP-gp has been successfullyimmobilized on the nanocomposite. Further studies were per-formed using quartz crystal microbalance (QCM) to confirm theimmobilization and estimation of the amount of AntiP-gp

Fig. 2. (A) CV response at GCE/AuNPs/pTTBA/AntiP-gp/MDRCC/APBA (black line)GCE/AuNPs/pTTBA/AntiP-gp/MDRCC/MWCNT-Hyd (blue line), GCE/AuNPs/pTTBA/AntiP-gp/ MDRCC/APBA-MWCNT-Hyd (red line) at the scan rate of 50.0 mV/s, 5000MDRCC/mL were used in all these experiments. (B) CVs recorded for the catalyticreduction of 4.0 mM H2O2 at GCE/AuNPs/pTTBA/AntiP-gp/MDRCC/APBA-MWCNT-Hyd sensor with increasing number of MDRCC/mL [a represents blank, b (1000), c(1500), d (2000), e (2500), f (3000), and g (3500) MDRCC/mL]. Inset shows thecalibration plot for increasing number of MDRCC/mL. (For interpretation of the re-ferences to color in this figure legend, the reader is referred to the web version ofthis article.)

P. Chandra et al. / Biosensors and Bioelectronics 70 (2015) 418–425422

immobilized on the polymer film based on the frequency change(Fig. 1(C)). In the case of AntiP-gp immobilization, a decrease in thefrequency reaches a complete steady state after about 95.0 minwith an overall frequency change (Δf) of 287 Hz, corresponding toa mass change (Δm) of 315.8712.2 ng based on a previously de-fined equation (Lee et al., 2010). These results reconfirm the suc-cessfully immobilization of AntiP-gp onto the pTTBA layer. Im-pedance spectrometry was employed to investigate the char-acteristics of probe layers at each modification step. Fig. 1(D) shows Nyquist plots obtained for bare GCE, GCE/AuNPs, GCE/pTTBA, and GCE/AuNPs/pTTBA electrodes in a 0.1 M PBS (pH 7.4),containing 4.0 mM Fe[(CN)6]3�/4�/0.3 M NaClO4. Values for Rs,Rp1, Rp2, CPE1, and CPE2 were obtained by fitting the experimentaldata to the equivalent circuit using the Zview2 impedance soft-ware. In the equivalent circuit, Rs represents the solution re-sistance, Rp1 and Rp2 represent the polarization resistances, Wrepresents the Warburg element, and CPE1 and CPE2 are theconstant-phase elements. The Rp values were determined from thecross points between the Zre axes and the extrapolation of thecurve, which were obtained from a Nyquist plot of the impedancespectroscopy. Rp values for bare GCE (636.2), GCE/AuNPs (230.9),GCE/pTTBA (10006.7), and GCE/AuNPs/pTTBA (1113.1 Ω) electrodeswere determined and the values decreased about nine times andthe conductivity of the AuNP deposited-polymer electrode in-creased because of the electrodeposition of the AuNPs.

3.2. Detection of MDR cancer cells

The GCE/AuNPs/pTTBA/AntiP-gp sensor was incubated with5000 MDRCC/mL (CRL-1977™) for 30 min. Next, the GCE/AuNPs/pTTBA/AntiP-gp/MDRCC probe was incubated with APBA-MWCNT-Hyd conjugate for 30 min followed by washing three times in thesame buffer solution. Then, cyclic voltammetry (CV) was recordedby cycling the potential between þ0.6 and �0.7 V at a scan rate of50.0 mV/s in deoxygenated 0.1 M PBS. In this case, a redox peak at�50/�90 mV was observed due to the electrochemical behavior ofHyd itself indicating that APBA-MWCNT-Hyd interacted with theGCE/AuNPs/pTTBA/AntiP-gp/MDRCC electrode successfully (Fig. 2(A), red curve). The redox peak at �50/�90 mV was directly pro-portional to the scan rate between 10.0 and 60.0 mV/s indicatingthat the electrode reaction involved a surface-confined process ofHyd. In two separate controls we performed the similar experimentas mentioned above except Hyd and APBA. In both the cases noredox current was observed (blue and black curve). This is due toabsence of Hyd, an electrochemically active group and APBA, therecognition molecule on the sensor surface. The results from thesecontrol experiments clearly suggest that APBA and Hyd both arerequired to obtain an electrochemical signal to detect the MDRCC.The ability of GCE/AuNPs/pTTBA/AntiP-gp/MDRCC/APBA-MWCNT-Hyd probe to catalyze the reduction of H2O2 was further in-vestigated. The sensor probe was dipped in a deoxygenated 0.1 MPBS containing 4.0 mM H2O2 and the CVs was recorded. A verydistinct reduction peak at �400 mV was observed due to the re-duction of H2O2 by Hyd attached on the sensor probe (Fig. 2(B)). Noreduction peak at �400 mV, however, was observed when controlexperiments were performed without APBA and Hyd. Next, westudied the H2O2 catalytic current with increasing MDRCC to es-tablish that the current arose due to the immunoreaction. Fig. 2(B) shows the CVs recorded with the increasing number of MDRCC,where the current response increases with increase in the MDRCC

from 1000 to 3500 cells/mL. In this figure, a represents blank and b(1000), c (1500), d (2000), e (2500), f (3000), and g (3500) re-presents MDRCC/mL. A linear plot is obtained (inset) with a linearregression equation expressed as follows: ΔI (μA)¼2.43þ0.002[MDRCC], and has a correlation coefficient of 0.997. These resultsclearly indicate that the developed biosensor can accurately detect

MDRCC. Next, we optimized the experimental parameters of thesensor probe to obtain the detection limit of MDRCC using chron-oamperometry (see details in Fig. S2 in Supplementary material).

3.3. Analytical performance of the sensor

Analytical performance of the GCE/AuNPs/pTTBA/AntiP-gp sensorprobe was tested for the increasing numbers of MDRCC (CRL-1977™cells) followed by binding with APBA-MWCNT-Hyd conjugate andreaction with H2O2. Under the optimized conditions chron-oamperometry was performed for GCE/AuNPs/pTTBA/AntiP-gp/MDRCC/APBA-MWCNT-Hyd probe at �0.45 V vs. Ag/AgCl (optimized)in deoxygenated PBS containing 4.0 mM H2O2. The increase in thecatalytic signal due to H2O2 reduction was obtained in chron-oamperometry which was directly proportional to the amount ofMDRCC captured by the sensor. Fig. 3(A) shows the chronoampero-gram with gradual increased in the current response with increasing

Fig. 3. (A) Amperometic responses for MDRCC detection using the biosensor withincreasing number of MDRCC/mL [a represents blank, b (50), c (100), d (500), e(1000), f (10,000), g (25,000), h (50,000), i (75,000), and j (100,000) MDRCC/mL].(B) Calibration plot based on the signal obtained from the amperogram (insetshows the magnified values between 50 and 10,000 MDRCC/mL). All experimentswere performed in deoxygenated 0.1 M PBS containing 4.0 mM H2O2 (pH 7.4),applied potential: �0.45 V vs. Ag/AgCl.

Fig. 4. (A) Histogram showing the selective detection of MDRCC using biosensor at30,000 MDRCC/mL concentration. (B) Microphotograph of MDRCC captured by ITO/pTTBA/AntiP-gp probe ((a–c)) and (d) ITO/pTTBA chip treated with MDRCC (no cellswere observed after washing the electrode).

P. Chandra et al. / Biosensors and Bioelectronics 70 (2015) 418–425 423

number of MDRCC. In this figure, a represents blank and b (50), c(100), d (500), e (1000), f (10,000), g (25,000), h (50,000), i (75,000),and j (100,000) represents MDRCC/mL. Based on the obtainedchronoamperograms a calibration curve was plotted as shown inFig. 3(B). The calibration curve shows a linear response with theMDRCC ranging from 50 to 100,000 cells/mL. The linear regressionequation is expressed as follows: ΔI¼0.9351 (70.1181)þ0.0025(70.0000067) [MDRCC] with the correlation coefficient of 0.998. Thedetection limit of MDRCC was determined to be 2372 cells/mL(RSDo4.4%) based on the standard deviation of five repeated mea-surements of the blank (95% confidence level, n¼5). The obtaineddetection limit is �4 times lower compared to the recently reportedMDRCC impedimetric biosensor based on gold nanoparticles/poly-aniline nanofibers (Zhang et al., 2014) and also to other recently re-ported chemiluminescence (Chen et al., 2014) and QCM (Shaolianet al., 2014) biosensor for cancer cell detection. We also performedsame experiment with another P-gp over expressed cell line ob-tained from ATCC, USA (CRL-2274™). The current response increasedwith increase in the CRL-2274™ cells, however, the signals in thiscase was 17% lower compared to the CRL-1977™ cell type (data not

shown). This is possibly due to the less concentration of the P-gpexpressed on the cell membrane of CRL-2274™cells.

3.4. Selectivity, reproducibility, stability, and real sample analysis

Control experiments were performed to validate the selectivityof the developed nanobiosensor. For this purpose, SKBr-3, HeLa,OSE, and HEK-293 cells were examined under the similar experi-mental conditions. The selectivity test was performed at 1000,30,000, 75,000, and 100,000 cells/mL concentrations. Fig. 4(A) shows the current response at 30,000 cells/mL concentration,where negligible signal is observed for tested cells. This was be-cause no immunoreaction occurred between AntiP-gp and cellsurface P-gp which was not present (or negligible) on these cells.We also performed control experiments to ensure the selectivity ofthe nanobiosensor by testing the compounds that are majorlypresent in the real sample matrix such as; albumin, fibrinogens,glucose. Interestingly, no current response was observed for thesecompounds indicating that the develop nanobiosensor is highlyselective towards MDRCC detection and no interference or false

Fig. 5. Histogram showing the comparative response of the APBA-MWCNT-Hydand reporter AntiP-gp-MWCNT-Hyd conjugate. The values on the y-axis shown inthe histograms are based on the amperometric signals.

P. Chandra et al. / Biosensors and Bioelectronics 70 (2015) 418–425424

positive signal is obtained. The reproducibility of the analysis wasperformed which showed the RSDo3.5% (n¼5) and electrode-to-electrode RSD was o3.2% even when the same preparation con-ditions were applied. This minor variation was possibly due to thedifference in the modified surface and/or slight variations in theoptimized conditions. We also studied the stability of the sensorprobe with respect to time. The immunosensor retained almost96% of its sensitivity for 6 weeks. The response decreased about13% after 6 weeks and then gradually decreased with time. Thus,the sensor was stable for 6 weeks. The good stability was probablycaused by the strong covalent attachment of the antibody on thestable conducting polymer–nanoparticle composite.

The biomedical application of the nanobiosensor was in-vestigated by detecting MDRCC in a human serum samples in thesimilar dynamic range (50–100,000 MDRCC/mL) as done in thebuffer solution. Serum has been widely used as a model sample todemonstrate the biomedical application of biosensors (Chandraet al., 2013; Colin et al., 2011; Joshua et al., 2007). Thus, we usedserum to examine the real clinical value of the developed sensor.MDRCC were dispersed into five times diluted serum samples andincubated with the GCE/AuNPs/pTTBA/AntiP-gp sensor probe.After washing the sensor probe was treated with the APBA-MWCNT-Hyd conjugate, and reacted with H2O2 as described ear-lier. The current response increased with increase in the MDRCC

number indicating that the developed biosensor can effectivelydetect MDRCC in serum samples, hence valuable in the biomedicalanalysis. The linear regression equation for the calibration plot inthe serum sample is expressed as follows: ΔI¼0.881þ0.0023[MDRCC] with the correlation coefficient of 0.998 indicating thatthis sensor is capable to detect MDRCC from the complex biologicalsample matrix effectively. The detection limit of MDRCC in serumsample was determined to be 2872 cells/mL (RSDo4.8%) basedon the standard deviation of five repeated measurements of theblank (95% confidence level, n¼5). The detection limit in serumsample was 5 cells/mL higher (less sensitive) compared to thedetection limit in a buffer solution (23 cells/mL). This minor var-iation in the detection limit was possibly due to negligible matrixeffect due to the serum components. We also examined the ap-plicability of the biosensor to detect MDRCC in the mixed cellsample. For this purpose, enumerated MDRCC were mixed withSKBr-3, HeLa, OSE, and HEK-293 cell and detected by the biosensorprobe. Interestingly, the sensitivity of the detection in this casewas 97% (n¼5) compared when MDRCC was tested alone. Thisresult clearly indicates that the developed biosensor can selec-tively detect MDRCC in the presence of other cancerous and non-cancerous cells effectively.

3.5. Bioimagining and comparison with conventional reporter anti-body based assay

The capturing of the MDRCC by the AntiP-gp probe was in-vestigated by bioimagining MDRCC incubated ITO/pTTBA/AntiP-gpsurface using optical microscope. The ITO/pTTBA/AntiP-gp sensorwas incubated with the MDRCC and after washing it was observedunder microscope. Fig. 4(B) shows the microscopic images of ITO/pTTBA/AntiP-gp sensor after treating it with different concentra-tions of MDRCC. With increase in the MDRCC concentrations be-tween 102 to 103 cells/mL, the number of cells captured by thesensor was increased as shown in Fig. 4(B(a–c)). These resultsclearly indicate that ITO/pTTBA/AntiP-gp sensor is able to detectMDRCC. No MDRCC, however, were observed when Anti-gp was notimmobilized on the ITO/pTTBA electrode indicating that AntiP-gpis essential to detect MDRCC (Fig. 4(B(d))). We also compared thismethod with the conventional reporter antibody based im-munoassay. In this experiment, a reporter AntiP-gp-MWCNT-Hydconjugate was prepared through the carbodiimide coupling

reaction following our earlier work (Zhu et al., 2010). This con-jugate was tested by the following same experimental steps asdescribed in the previous section. Fig. 5 shows the comparativeresponse of the APBA-MWCNT-Hyd and the reporter AntiP-gp-MWCNT-Hyd conjugate at 25,000, 50,000, and 100,000MDRCC/mL. A much higher signal was observed for APBA-MWCNT-Hyd conjugate. The detection limit of MDRCC using reporter AntiP-gp-MWCNT-Hyd conjugate was 15878 cells/mL based on thestandard deviation of five repeated measurements of the blank(95% confidence level, n¼5). This detection limit is �7 timeshigher i.e. less sensitive compared to the detection limit obtainedusing APBA-MWCNT-Hyd conjugate (2372 cells/mL). This is dueto the less immunoreaction between the reporter AntiP-gp-MWCNT-Hyd conjugate and the MDRCC captured by the sensorprobe. This less immunoreaction is due to the low concentrationsof P-gp on the MDRCC cell membrane. It has already been reportedthat variation in the concentration of cell surface antigen (e.g.:P-gp in our case) significantly affects the antibody binding(Langmuir et al., 1991; Velders et al., 1998) which may directlyinfluence the sensitivity of the biosensor. In case of APBA-MWCNT-Hyd conjugate low P-gp concentration does not affect the sensi-tivity of the biosensor because the target molecule for this con-jugate is glycan which is expressed greatly onto the cancer cellsurface (Zhang et al., 2010).

4. Conclusions

A highly sensitive and selective amperometric nanobiosensorhas been successfully designed for the detection of multidrug re-sistant cancer cells (MDRCC) in the biological matrix. The se-lectivity and sensitivity of MDRCC detection were achieved due tothe monoclonal Pg-antibody covalently immobilized onto theconduction polymer-gold nanoparticle composite and also due theAPBA-MWCNT-Hyd conjugate. The developed nanobiosensor isable to detect the MDRCC in mixed cell sample and in the presenceof other chemical molecules. The detection limit of nanobiosensoris 2372 MDRCC/mL, which is� four times more sensitive than themost recent MDR cancer cell biosensor. The developed nanobio-sensor is �7 times more sensitive compared to the conventionalreporter antibody based immunosensor which gives a new direc-tion towards sensitive detection of cancer cells. This method

P. Chandra et al. / Biosensors and Bioelectronics 70 (2015) 418–425 425

proposed herein is a generic method and can be easily adopted forother type of cancer cells simply by changing the detector anti-body. The developed sensor could be a sensitive diagnosticmethod for the detection of drug resistant cancer cell in the bio-logical fluids of cancer patients which will help clinicians to de-velop their alternate therapeutic strategies. The study clearly de-monstrates highly sensitive detection of cancer cells even thoughthe cell surface biomarkers is expressed/present in low con-centrations using a nanobiosensor. In future, the developed na-nobiosensor prototype can be translated into a miniaturized kitbased assay for the point-of-care medical applications.

Acknowledgment

This work was supported by the National Research Foundationof Korea (NRF) Grant funded by the Korea government (MSIP) (No.20100029128).

Appendix A. Supplementary material

Supplementary data associated with this article can be found inthe online version at http://dx.doi.org/10.1016/j.bios.2015.03.069.

References

Beck, W.T., Grogan, T.M., Willman, C.L., Cordon-Cardo, C., Parham, D.M., Kuttesch, J.F., Andreeff, M., Bates, S.E., Berard, C.W., Boyett, J.M., Brophy, N.A., Broxterman,H.J., Chan, H.S.L., Dalton, W.S., Dietel, M., Fojo, A.T., Gascoyne, R.D., Head, D.,Houghton, P.J., Srivastava, D.K., Lehnert, M., Leith, C.P., Paietta, E., Pavelic, Z.P.,Rimsza, L., Roninson, I.B., Sikic, B.I., Twentyman, P.R., Warnke, R., Weinstein, R.,1996. Cancer Res. 56, 3010–3020.

Bellamy, W.T., 1996. Annu. Rev. Pharmacol. Toxicol. 36, 161–183.Chan, H.S., Thorner, P.S., Haddad, G., Ling, V., 1990. J. Clin. Oncol. 8, 689–704.Chandra, P., Noh, H.-B., Shim, Y.-B., 2013. Chem. Commun. 7, 1900–1902.Chandra, P., Noh, H.-B., Won, M.-S., Shim, Y.-B., 2011a. Biosens. Bioelectron. 26,

4442–4449.Chandra, P., Zaidi, S.A., Noh, H.-B., Shim, Y.-B., 2011b. Biosens. Bioelectron. 28,

326–332.Chen, X., He, Y., Zhang, Y., Liu, M., Liu, Y., Li, J., 2014. Nanoscale 6, 11196–11203.Colin, D.M., Suwussa, B., Weihong, T., Joshua, E.S., 2011. Anal. Chem. 83, 727–734.Das, D., Kim, D.-M., Park, D.-S., Shim, Y.-B., 2011. Electroanalysis 23, 2036–2041.Du, D., Ju, H., Zhang, X., Chen, J., Cai, J., Chen, H., 2005. Biochemistry 30,

11539–11545.Dube, D.H., Bertozzi, C.R., 2005. Nat. Rev. Drug Discov. 6, 477–488.Gottesman, M.M., Fojo, T., Bates, S.E., 2002. Nat. Rev. Cancer 2, 48–58.Gillet, J.P., Efferth, T., Steinbach, D., Hamels, J., de Longueville, F., Bertholet, V., Re-

macle, J., 2004. Cancer Res. 64, 8987–8993.Goldman, D., Lellouche, J.-P., 2010. Carbon 48, 4170–4177.Joshua, E.S., Colin, D.M., Zhiwen, T., Dihua, S., Charles, L., Weihong, T., 2007. Anal.

Chem. 79, 3075–3082.Kim, D.-M., Shim, Y.-B., 2013. Anal. Chem. 85, 6536–6543.Koh, W.C.A., Chandra, P., Kim, D.-M., Shim, Y.-B., 2011. Anal. Chem. 83, 6177–6183.Krishna, R., Mayer, L.D., 2000. Eur. J. Pharm. Sci. 11, 265–283.Langmuir, V.K., McGann, J.K., Buchegger, F., Sutherland, R.M., 1991. Int. J. Radiat.

Appl. Instrum. B 18, 753–764.Lee, K.-S., Won, M.-S., Noh, H.-B., Shim, Y.-B., 2010. Biomaterials 30, 7827–7835.Wu, D., Guo, Z., Liu., Y., Guo., A., Lou., W., Fan., D., Wei, Q., 2015. Talanta 134,

305–309.Ludescher, C., Thaler, J., Drach, D., Drach, J., Spitaler, M., Gattringer, C., Huber, H.,

Hofmann, J., 1992. Br. J. Haematol. 82, 161–168.Malhotra, R., Patel, V., Chikkaveeraiah, B.V., Munge, B.S., Cheong, S.C., Zain, R.B.,

Abraham, M.T., Dey, D.K., Gutkind, J.S., Rusling, J.F., 2012. Anal. Chem. 17,6249–6255.

Murphy, L.D., Herzog, C.E., Rudick, J.B., Fojo, A.T., 1990. Biochemistry 13,10351–10356.

Noh, H.-B., Chandra, P., Moon, J.-O., Shim, Y.-B., 2012. Biomaterials 33, 2600–2607.Piran, M., Kotlyar, V., Medina, D.D., Pirlot, C., Goldmana, C., Lellouche, J.P., 2009. J.

Mater. Chem. 19, 631–638.Rahman, Md.A., Kumar, P., Park, D.-S., Shim, Y.-B., 2008. Sensors 8, 118–141.Shaolian, Z., Haihua, B., Jinmei, L., Peihui, Y., Jiye, C., 2014. Analyst 139, 6259–6265.Velders, M.P., van Rhijn, C.M., Oskam, E., Fleuren, G.J., Warnaar, S.O., 1998. Br. J.

Cancer 78, 478–483.Zhang, H., Jiang, H., Sun, F., Wang, H., Zhao, J., Chen, B., Wang, X., 2011. Biosens.

Bioelectron. 26, 3361–3366.Zhang, S., Zhang, L., Zhang, X., Yang, P., Cai, J., 2014. Analyst 16, 3629–3635.Zhang, X., Teng, Y., Fu, Y., Xu, L., Zhang, S., He, B., Wang, C., Zhang, W., 2010. Anal.

Chem. 82, 9455–9460.Zhu, Y., Chandra, P., Shim, Y.-B., 2013. Anal. Chem. 85, 1058–1064.Zhu, Y., Son, J.I., Shim, Y.-B., 2010. Biosens. Bioelectron. 26, 1002–1008.Zhu, Y.D., Peng, J., Jiang, L.P., Zhu, J.J., 2014. Analyst 7, 649–655.