biosensors Article

Asymmetric Mach–Zehnder Interferometer Based Biosensors for Aflatoxin M1 Detection Tatevik Chalyan 1, *,† , Romain Guider 1,† , Laura Pasquardini 2,† , Manuela Zanetti 2 , Floris Falke 3 , Erik Schreuder 3 , Rene G. Heideman 3 , Cecilia Pederzolli 2 and Lorenzo Pavesi 1 Received: 30 October 2015; Accepted: 30 December 2015; Published: 6 January 2016 Academic Editor: Yuliya Semenova 1 2 3

* †

Nanoscience Laboratory, Department of Physics, University of Trento, Via Sommarive 14, 38123 Povo (TN), Italy; [email protected] (R.G.); [email protected] (L.P.) LaBSSAH, Fondazione Bruno Kessler, Via Sommarive 18, 38123 Povo, Italy; [email protected] (L.P.); [email protected] (M.Z.); [email protected] (C.P.) LioniX B.V., Enschede 7522, The Netherlands; [email protected] (F.F.); [email protected] (E.S.); [email protected] (R.G.H.) Correspondence: [email protected]; Tel.: +39-04-6128-2493; Fax: +39-04-6128-2967 These authors contributed equally to this work.

Abstract: In this work, we present a study of Aflatoxin M1 detection by photonic biosensors based on Si3 N4 Asymmetric Mach–Zehnder Interferometer (aMZI) functionalized with antibodies fragments (Fab1 ). We measured a best volumetric sensitivity of 104 rad/RIU, leading to a Limit of Detection below 5 ˆ 10´7 RIU. On sensors functionalized with Fab1 , we performed specific and non-specific sensing measurements at various toxin concentrations. Reproducibility of the measurements and re-usability of the sensor were also investigated. Keywords: Mach–Zehnder interferometer; biosensor; Aflatoxin M1; limit of detection; Fab1

1. Introduction In recent years, integrated optical sensors have shown very promising results for label-free detection [1]. Even if Surface Plasmon Resonance (SPR) is the most commonly used technique for label-free analysis [2], devices like ring resonators [3,4] and Mach–Zehnder interferometers [5] are showing high sensitivities and miniaturization abilities, which allow realizing a complete lab-on-chip device [6]. We are here interested in a biosensor for a fast and comprehensive detection of Aflatoxin M1 (AFM1) in milk. In fact, the AFM1 mycotoxin is a milk contaminant and a potent carcinogen substance that is regulated by the European Commission (EC No. 1881/2006). At present, the level of AFM1 in milk is analyzed using screening tests like Enzyme-Linked ImmunoSorbent Assay (ELISA). In addition, High-Performance Liquid Chromatography is needed when samples show suspect concentration of AFM1. The combination of these analyses is expensive, time-consuming and bulky. The sensor, which we are here proposing, is based on an asymmetric Mach–Zehnder Interferometer (aMZI), with silicon nitride (Si3 N4 ) as core material and SiO2 as cladding material. They are fabricated by standard CMOS processing and can be assembled with fluidics in a compact package. The basic idea behind the sensor is that the interference at the output of an aMZI is affected by the phase difference between the light which travels along the two interferometer arms. By opening a window in the SiO2 and exposing one of the two arms to an analyte, the change in the refractive index in the window region is measured as an interference fringe shift in the aMZI output signal. From the fringe shift, the phase shift suffered by the light that travels in the exposed arm can be determined. From the phase shift, the change of the refractive index is measured, i.e., the sensor can sense the

Biosensors 2016, 6, 1; doi:10.3390/bios6010001

www.mdpi.com/journal/biosensors

Biosensors 2016, 6, 1

2 of 10

analyte. The Mach–Zehnder interferometer is made asymmetric by the addition of a small path length difference between both arms. This difference in path length, results in a wavelength dependency of the output phase. Scanning the aMZI over a small bandwidth results in a phase plot of which the phase shift, due to a refractive index change, can be monitored with high accuracy. One key point of a biosensor is its specificity. In order to develop a biosensor specific to the detection of AFM1, a functionalization process based on antigen-binding fragments (Fab1 ) was applied on the aMZI surface. One benefit of using antibodies is the extreme specificity associated with their antigen binding sites [7]. The immobilization of antibodies includes whole antibodies, F(ab)2 , and Fab1 fragments. Antibodies need to be immobilized in a controlled and oriented fashion, in order to maximize the number of available antigen-binding sites and prevent surface degradation. Although significant progress has been made with whole antibody immobilization, methods for the immobilization of Fab1 fragments have become relevant especially for biosensor development [8–10]. Fab1 fragment immobilization offers several benefits over whole antibodies. Firstly, bound Fab1 fragments are able to maintain a higher amount of binding sites per unit area compared to whole antibodies due to their smaller size. Moreover, Fab1 fragments are immobilized easily in an oriented conformation via the nucleophilic sulfide near their C-terminals. The characteristics of Fab1 fragments make them ideal candidates for biosensing, also because the currently well-developed knowledge of the available immobilization chemistries for many different materials that are frequently used as biosensing platforms (e.g., gold, Si, Si alloys, plastic, and inorganic substrates). In this work, we measured and analyzed the selectivity and regeneration of functionalized asymmetric Mach–Zehnder interferometric biosensors with purified solutions. First, we setup the functionalization procedure of the sensor. Then, we performed sensing measurements using two mycotoxins, Aflatoxin M1 and Ochratoxin, in buffer solution for various concentrations, and, finally, we performed regeneration measurements using glycine-methanol solutions. Our results are confirmed by experiments on functionalized Si3 N4 flat substrates. 2. Experimental Section 2.1. Materials 3-mercaptopropyltrimethoxysilane (MPTMS, 99%) purchased from Gelest Ltd. (Maidstone, Kent, UK), was used without any further purification. Toluene anhydrous (99.8%), toluene, dithiothreitol (DTT) and all powders for buffered solutions were purchased from Sigma-Aldrich s.r.l. (Milan, Italy). Methoxypolyethyleneglycolthiol (mPEG-SH) with 2000 and 5000 molecular weight were purchased from Nektar Therapeutics AL (Huntsville, AL, USA). A rabbit polyclonal anti-AFM1 antibody and an horseradish peroxidase (HRP)-conjugated Aflatoxin M1 (AFM1-HRP) contained in the I’screen Afla M1 milk Elisa kit were purchased from Tecna s.r.l. (Padua, Italy), while Aflatoxin M1 and Ochratoxin were purchased from Sigma-Aldrich s.r.l. (Milan, Italy). SuperSignal West Femto Chemiluminescent Substrate was purchased from Thermo Scientific (Rockford, IL, USA). 2.2. Functionalization Process F(ab1 )2 fragments were generated by protease digestion (Immobilized Papain Thermo Scientific) of 20 µL of 1 mg/mL anti-AFM1 polyclonal rabbit IgG according to manufacturer’s instructions. Then, a 13.3 µM F(ab1 )2 solution was mixed with 10 mM DTT to reduce the disulfide bond in the hinge region, and the mixture was incubated for two hours at room temperature. The mixture was poured into a centrifugal filter unit (Microcon YM-10, MWCO 10000, Millipore Corp., Billerica, MA, USA) to remove the excess of DTT. The Fab1 were immobilized on the Si3 N4 surface adapting the protocol described in Yoshimoto et al., [11] with some modifications. In order to introduce thiol groups able to react with the cysteine groups on Fab1 fragments, the silicon nitride surface was functionalized in wet conditions with mercaptosilane (MPTMS) [12]. The surfaces (both aMZI chip and flat samples) were cleaned with

Biosensors 2016, 6, 0001

3 of 10

The Fab’ were immobilized on the Si3N4 surface adapting the protocol described in Yoshimoto et al., [11] with some modifications. In order to introduce thiol groups able to react with the cysteine Biosensors 2016,Fab’ 6, 1 fragments, the silicon nitride surface was functionalized in wet conditions3 with of 10 groups on mercaptosilane (MPTMS) [12]. The surfaces (both aMZI chip and flat samples) were cleaned with an argon plasma (6.8(6.8 W, W, oneone min) to remove organic contaminants and and to hydroxylate the surface and an argon plasma min) to remove organic contaminants to hydroxylate the surface were then immersed in a 1% v/v solution of MPTMS in toluene anhydrous at 60 °C for 10 min. ˝ and were then immersed in a 1% v/v solution of MPTMS in toluene anhydrous at 60 C for 10 min. Silane-coated substrates substrateswere wererinsed rinsed several times toluene anddried then in dried in aofstream of Silane-coated several times withwith toluene and then a stream nitrogen. nitrogen. The immobilization of Fab’ fragments onto the surface was carried out by deposition of 80 1 The immobilization of Fab fragments onto the surface was carried out by deposition of 80 µL of a μL of a Fab 0.331 in μM in 10 mM buffer phosphate buffer with 10 mM ethylenediaminetetraacetic acid 0.33 µM 10 Fab’ mM phosphate with 10 mM ethylenediaminetetraacetic acid (EDTA). After (EDTA). After minutes, the surface was PEGylated first of 200 μM final two minutes, thetwo surface was PEGylated by addition first of 200by µMaddition final concentration of mPEG-SH concentration of mPEG-SH 5000 (for 30 min on orbital shaker at 80 rpm) and then of 200 μM on of 5000 (for 30 min on orbital shaker at 80 rpm) and then of 200 µM of mPEG-SH 2000 (for 60 min mPEG-SH 2000 (forThe 60 min on shaker at 80 rpm). The surface was finallybuffer. cleaned using a PBS-EDTA shaker at 80 rpm). surface was finally cleaned using a PBS-EDTA The concentration of buffer. The concentration of the Fab’ was determined by measuring their absorbance with a 1 the Fab was determined by measuring their absorbance with a Nanodrop instrument, (Extinction Nanodrop (0.1%) instrument, coefficient = 1.35(Extinction at 280 nm).coefficient (0.1%) = 1.35 at 280 nm). 2.3. Mach–Zehnder Fabrication Process and Design

The aMZI aMZI devices devices are are based based on on the the TriPlex TriPlex technology technology [13]. [13]. A μm thick thick silicon silicon The A 4” 4” inch inch 525 525 µm substrate is is is formed onon thethe surface. Then, a single 103 substrate is oxidized oxidized until until88micron micronthermal thermaloxide oxidelayer layer formed surface. Then, a single nm thick LPCVD (low-pressure chemical vapor deposition) Si 3 N 4 layer (refractive index 2.02) is 103 nm thick LPCVD (low-pressure chemical vapor deposition) Si3 N4 layer (refractive index 2.02) deposited onto the is deposited onto thethermal thermaloxide oxidefollowed followedby byaathin thinLPCVD LPCVDSiO SiO22layer. layer. This Thislayer layer stack stack is is patterned patterned by using photolithography, dry etching (RIE) and subsequently resist removal. The waveguides are by using photolithography, dry etching (RIE) and subsequently resist removal. The waveguides are then covered with a thick 6 μm LPCVD SiO 2 cladding. To enable an interaction between the then covered with a thick 6 µm LPCVD SiO2 cladding. To enable an interaction between the evanescent evanescent fieldpropagating of the light through propagating through the and theofliquid sample of cladding interest, field of the light the waveguide andwaveguide the liquid sample interest, the top the top cladding is locally removed by opening the sensing window. This is accomplished a is locally removed by opening the sensing window. This is accomplished by a photolithography by step photolithography and BHFSiwet and BHF wet etch step down to the layerdown [14]. to the Si3N4 layer [14]. 3 N4 etch

Sketch of asymmetric the asymmetric Mach–Zehnder interferometer chips (inset) measured; (inset) Figure 1.1.Sketch of the Mach–Zehnder interferometer chips measured; photograph photograph of the chip and a one euro-cent coin for comparison. The sensor here is an eight of the chip and a one euro-cent coin for comparison. The sensor here is an eight aMZI sensor. aMZI sensor.

The design of the sensor is reported in Figure 1. In this design, four aMZI are integrated in a single chip. A same input signal is sent to four1.aMZI bydesign, a one tofour fouraMZI channel splitter. Other The design of the sensor is reported inthe Figure In this are integrated in a designs haveAupsame to eight aMZI integrated in the chip (see inset Figure 1). The long optical single chip. input signal is sent to the same four aMZI by the a one to to four channel splitter. Other path length arms is 6.25 mm, and it ischip achieved byinset a spiral waveguide minimize designs haveofupthe to sensing eight aMZI integrated in the same (see the to Figure 1). Theto long optical required surface area. Therearms is a small initthe length reference arm to the path length of the sensing is 6.25addition mm, and is path achieved byina the spiral waveguide to obtain minimize asymmetry. The additional length determines the free spectral range (FSR) of the aMZI and is chosen required surface area. There is a small addition in the path length in the reference arm to obtain the such that it matched with the length bandwidth of the used Surface-Emitting (VCSEL). asymmetry. The additional determines the Vertical-Cavity free spectral range (FSR) of theLaser aMZI and is Openings arethat shown in the figure bybandwidth a blue hatched Three out of the four aMZI have the chosen such it matched with the of theregion. used Vertical-Cavity Surface-Emitting Laser sensing window on top of the sensing arm. The fourth aMZI is left covered by the cladding in order to (VCSEL). Openings are shown in the figure by a blue hatched region. Three out of the four aMZI isolate it from the microfluidic chamber and, therefore, to be used as the reference sensor (baseline sensor). This aMZI is used as an internal reference 3/10both for the input signal intensity as well as for setting the temperature reference. The area of and pitch between the sensors are chosen such that

Biosensors 2016, 6, 1

4 of 10

each individual sensor can be functionalized by a spotter with a different chemistry allowing internal consistency tests or multianalyte detection. Note that the input and output waveguides are all on the same edge of the chip. 2.4. Experimental Setup To characterize the sensors, we used a fiber array to waveguide alignment setup where a 12 fiber array unit with 250 micron pitch between the fibers is placed on multiaxis piezo-controlled translation stages. The light signal polarization is controlled by a two-paddle polarization controller. An optical microscope coupled to a visible/IR camera is used for alignment and imaging. For the detection, we connected the fibers to Si transimpedance amplified photodetectors interfaced to a PicoScope 4824 (an 8 channel USB oscilloscope). Finally, a ULM850-B2-PL VCSELs from Philips Technologies GmbH U-L-M Photonics connected to a single mode visible fiber is used as light source. Wavelength variation is achieved by current tuning the VCSEL with a periodic current ramp which also triggers the time scan of the oscilloscope. In this way, a wavelength scan of the four aMZI output waveguides is recorded by the oscilloscope and transferred to the control computer. The data are analyzed online by an automatic routine to extract the time dependent phase shift signal of each four aMZI. At the end, a live recording of the phase shift of the signal light propagating in the four aMZI is achieved with a VCSEL modulation frequency of 20 Hz, and a data acquisition of 50,000 points per spectrum. For sensitivity and sensing measurements, the chip is sealed on a homemade polydimethylsiloxane (PDMS) microfluidic flow cell with a volume chamber of less than 0.5 µL connected to a VICI M6 liquid handling pump. 3. Results and Discussion 3.1. Optimization of the Functionalization Process The Fab1 density on the silanized Si3 N4 flat surfaces was optimized incubating different concentrations using the protocol reported in Section 2.2. After immobilization, the surfaces were incubated with an HRP-conjugated Aflatoxin M1 (AFM1-HRP) stock solution diluted 80 times in 50 mM MES buffer pH 6.6 for one hour, washed twice in buffer and transferred to a black microplate, where the developer solution was added. HRP in presence of a suitable substrate develops a chemiluminescence signal that can be easily detected. After five minutes incubation, the signal was recorded with a ChemiDoc MP system (Biorad). Figure 2a shows the chemiluminescence signal generated by AFM1-HRP immobilized on functionalized surfaces as a function of the Fab1 concentration. The immobilization protocol described in Section 2.2 was applied, incubating the Fab1 solution for 2 min before PEG addition. No signal is recorded on the PEG passivated surface (without Fab1 solution, data not shown). Fitting the data with a Langmuir equation that describes the adsorption of molecules on a surface, it is possible to determine the surface saturation. A 0.33 µM Fab1 concentration was selected for the following experiments. In order to optimize the functionalization process, Fab1 are immobilized changing the incubation time from 2 to 60 min before PEG addition. The delayed addition of PEG after Fab1 immobilization is crucial in gold surface functionalization since it prevents a wrong Fab1 orientation [11] while on mercaptosilanized surfaces, we have demonstrated that this is not true. As reported in Figure 2b, in fact, no significant variation in the AFM1-HRP detection is observed, suggesting that 2 min of delay in the PEG addition is enough to obtain a good Fab1 orientation, although no significant effect in the antigen recognition is observed.

Biosensors 2016, 6, 1 Biosensors 2016, 6, 0001

5 of 10 5 of 10

1 prepared Figure 2. (a) of of AFM1-HRP incubated on Fab’ prepared surface. The Figure (a)Chemiluminescence Chemiluminescencedetection detection AFM1-HRP incubated on Fab surface. 1 protocol described in Section 2.2 was applied changing the Fab’ concentration (incubation time The protocol described in Section 2.2 was applied changing the Fab concentration (incubation time2 2min). min).(b) (b)For Fora a0.33 0.33μM µMFab’ Fab1concentration, concentration,the theAFM1-HRP AFM1-HRP recognition recognition was was evaluated as a function of the the delay delay before before PEG PEG addition addition (incubation (incubation time). time). The reported values represent represent the the mean mean value value on of two different differentexperiments, experiments,and anderror errorbars bars reported standard deviation. Exposition times two areare reported as as thethe standard deviation. Exposition times are equal to 0.2 and 1 s for (b), respectively. are equal tos0.2 s and 1 s(a) forand (a) and (b), respectively.

3.2.1. Sensing Measurements 3.2. Sensing Measurements In order order to to define define the the performances performances of of our our photonic photonic sensors, sensors, we we characterized characterized the the volume volume In Sensitivity (S) (S) of of the the uncovered uncovered aMZI. aMZI. To To calculate calculate this this parameter, parameter, we we monitored monitored in in real-time real-time the the Sensitivity phase shift of the aMZI, as one arm of the sensor was exposed to glucose-water solutions of various phase shift of the aMZI, as one arm of the sensor was exposed to glucose-water solutions of various concentrations. Similar group on on ring ring resonators, resonators, and and concentrations. Similar experiments experiments were were already already performed performed by by our our group more details detailson onthis this experiment available in [15]. Figure 3 represents theshifts phaseasshifts as a more experiment areare available in [15]. Figure 3 represents the phase a function function ofrefractive the bulk refractive index variations measured simultaneously on three aMZI. note of the bulk index variations measured simultaneously on three aMZI. We note thatWe adding thatglucose-water adding the glucose-water solution causes aphase significant phase shift which similar the three the solution causes a significant shift which is similar foris the threefor sensors on sensors on the chip. The initial phase is recovered when only water is flown. From the relation the chip. The initial phase is recovered when only water is flown. From the relation between the betweenconcentration the glucose concentration (i.e., refractive index of the cladding liquid) thewe phase shift, glucose (i.e., refractive index of the cladding liquid) and the phaseand shift, deduced we sensitivity. deduced theWe sensitivity. We found of 10,600 rad/RIU an error 3%. Ifthe welimit define the found sensitivity of sensitivity 10,600 rad/RIU with an errorwith of 3%. If we of define of the limit of detection (LOD) as the minimum input quantity that can be distinguished with more detection (LOD) as the minimum input quantity that can be distinguished with more than 99% fidelity, than LOD 99% fidelity, then LOD 3σ/S, where σ is the output uncertainty, given as the measured phase then = 3σ/S, where σ is =the output uncertainty, given as the measured phase standard deviation standard on deviation on repeated measurements of no blank solution (when In no our analyte is obtained repeatedobtained measurements of blank solution (when analyte is present). sensor, −7 RIU. These values for sensitivity and LOD show good ´ 7 present). In our sensor, LOD = 5 × 10 LOD = 5 ˆ 10 RIU. These values for sensitivity and LOD show good performances of our sensors in performances of our sensors in comparison with other MZI platforms [16,17]. comparison with other MZI platforms [16,17].

5/10

Biosensors 2016, 6, 1 Biosensors 2016, 6, 0001

6 of 10 6 of 10

Figure3.3.Volume Volume sensitivity measurements. Evaluation ofbulk the sensitivity; bulk sensitivity; phase shift Figure sensitivity measurements. Evaluation of the (inset)(inset) phase shift curve curve forthe one of the aMZI sensors during the injection of the water-glucose solutions (glucose for one of aMZI sensors during the injection of the water-glucose solutions (glucose concentration concentration in %w/w labelled on the plot). in %w/w labelled on the plot).

3.2.2. Aflatoxin Measurements 3.3. Aflatoxin M1 M1 Measurements Toperform performsensing sensingmeasurements measurementson onthese thesesensors, sensors,we weinitially initiallyfilled filledthe themicrofluidic microfluidicchamber chamber To withaabuffer buffersolution. solution. In In our ourcase, case,the thebuffer buffersolution solutionwas was50 50mM mMMES MESbuffer bufferpH pH6.6. 6.6. We Wethen then with injected7575µLµof L aofsolution a solution containing the targeted mycotoxin or Ochratoxin, injected containing the targeted mycotoxin (AFM1(AFM1 or Ochratoxin, which is which anotheris another mycotoxin similar molecular atconcentration a known concentration order to measure the mycotoxin of similarof molecular weight) atweight) a known in order toin measure the evolution thetoaMZI due toofthe of the from the functionalized aMZI. ofevolution the phaseofofthe thephase aMZI of due the capture thecapture toxin from thetoxin functionalized aMZI. Note that the Note that the functionalization aims at detecting AFM1 and not Ochratoxin, so this experiment functionalization aims at detecting AFM1 and not Ochratoxin, so this experiment allows testing the allows testing specificity of the sensor response. The solution is inserted into the microfluidic specificity of thethe sensor response. The solution is inserted into the microfluidic chamber using an chamberloop using injection loop order to avoid the formation of air bubbles in the microfluidic injection in an order to avoid theinformation of air bubbles in the microfluidic chamber during the chamber duringAll themeasurements buffer exchange. Alldone measurements with a flow rate of 3 µ L/min. buffer exchange. were with a flowwere rate done of 3 µL/min. Asisisthe thecase caseofof sensitivity measurements, monitored phase to binding the binding As sensitivity measurements, wewe monitored the the phase shiftshift duedue to the of theof the toxins to the surface. results the when three asensors a 100 nM toxins to the surface. Figure 4aFigure shows 4a theshows results the for the three for sensors 100 nM when concentration of concentration of AFM1 (black lines) or of Ochratoxin (red lines) is added to the buffer solution. The AFM1 (black lines) or of Ochratoxin (red lines) is added to the buffer solution. The phase shifts in time phase shifts time following the kinetics of the to binding of the toxins the antibody on the surface following the in kinetics of the binding of the toxins the antibody on theto surface of the exposed aMZI of the exposed aMZI arms. When the toxin flow is stopped, the phase decreases according tothe the arms. When the toxin flow is stopped, the phase decreases according to the dissociation kinetics of dissociation kinetics of the toxin from the antibodies. These measurements were done on the same toxin from the antibodies. These measurements were done on the same chip after one regeneration chip for after regeneration for the AFM1 two for order to allow cycle theone AFM1 and two forcycle Ochratoxin, in orderand to allow theirOchratoxin, comparisonin(see Section 3.4). their comparison (see Section 3.2.3). We can appreciate the clear dependence of these signals as a function of the injected toxin We can appreciate the clear dependence signals as ainfunction of AFM1, the injected toxin concentration. The functionalization is shown toof bethese specific. In fact, the case of after MES concentration. The functionalization is shown to be In fact, the case while of AFM1, after rinsing, the phase shift is 2 radians larger than the onespecific. before the toxinininjection, in the caseMES of rinsing, the phase shift is 2 radians larger than the one before the toxin injection, while in the case of Ochratoxin, it is only about 0.25 radians. Ochratoxin, only aboutthe 0.25lowest radians. In orderittois determine concentration of Aflatoxin M1 that could be detected, we In order to determine the lowest concentration of Aflatoxin M1. M1 Figure that could be detected, we performed measurements with different concentrations of Aflatoxin 4b shows the results performed measurements with different concentrations of Aflatoxin M1. Figure 4b shows the results for 50 nM and 10 nM Aflatoxin M1 concentrations. Considering the molecular weight of Aflatoxin forof50328.27 nM and 10 nM Aflatoxin M1 concentrations. Considering the molecular of Aflatoxin M1 g/mol, 10 nM corresponds to 3 ng/mL. This value is higher than the weight value permitted by M1 of 328.27 g/mol, 10 nM corresponds to 3 ng/mL. This value is higher than the value permitted by European regulations (50 ng/L in milk). In order to decrease the limit of detection of the sensor, European regulations (50 ng/L in milk). In order to decrease the limit of detection of the sensor, a pre-concentration module is needed. Combining the sensor sensitivity and specificity with aa pre-concentrationmodule, moduletheisrequired needed.levels Combining sensor sensitivity and specificity with a pre-concentration could bethe accomplished. pre-concentration module, the required levels could be accomplished.

6/10

Biosensors 2016, 6, 1 Biosensors 2016, 6, 0001

7 of 10 7 of 10

Figure4.4.(a)(a)Sensorgram Sensorgramrecorded recordedononaMZI aMZIsensors sensorsbybyflowing flowingAFM1 AFM1(black) (black)and andOchratoxin Ochratoxin(red) (red) Figure throughthe themicrofluidic microfluidicchamber. chamber. At At t = = 0, sensors. The toxin was through 0, MES MESbuffer bufferisisflowing flowingthrough throughthe the sensors. The toxin then injected at at t =t 2.5 min is stopped stopped and andthe theMES MESbuffer bufferwas was was then injected = 2.5 minand andatatt t==25 25min. min. The The toxin toxin flow is injectedagain; again;(b) (b)Sensorgram Sensorgramrecorded recordedononone oneaMZI aMZIsensor sensorbybyflowing flowinga a5050nM nM(black) (black)and and1010nM nM injected (red) AFM1 solutions microfluidic chamber. phase is proportional the Aflatoxin (red) AFM1 solutions in in thethe microfluidic chamber. TheThe phase shiftshift is proportional to thetoAflatoxin M1 concentration. The actual phasephase shift isshift different from the onethe in one (a) since different sensor was used. M1 concentration. The actual is different from in (a)a since a different sensor was used.

3.4. Regeneration Measurements 3.4. Regeneration Measurements In order to test the reusability of the biosensor, we also investigated the regeneration of the functionalized samples by usingof100 glycine-HCl 2.3 with 10% v/v of methanol In order to test the reusability themM biosensor, we alsopH investigated the regeneration of the (glycine-methanol solution). injected M1 solutions in the microfluidic in functionalized samples by We using 100 Aflatoxin mM glycine-HCl pH 2.3 with 10% v/v chamber, of methanol order to link the toxin to the functionalized surface ofM1 the solutions sensor. We injected the MES bufferin (glycine-methanol solution). We injected Aflatoxin inthen the microfluidic chamber, again microfluidic chamber for several minutes, order to restore a stable signal, orderintothe link the toxin to the functionalized surface ofinthe sensor. We then injected theand MESfinally buffer we injected solution, in orderminutes, to breakinthe Aflatoxin-antibody and remove again in theglycine-methanol microfluidic chamber for several order to restore a stablebond signal, and finally allwe the linkedglycine-methanol toxins from the sensor surface while keeping antibodies in place: aim toall injected solution, in order to break the the Aflatoxin-antibody bondi.e., andwe remove regenerate sensor. We the repeated procedure times the same in chip, andi.e., analyzed theto the linkedthe toxins from sensorthis surface whileseveral keeping the on antibodies place: we aim sensor response. ResultsWe arerepeated reported this in Figure 5 for aseveral 100 nM Aflatoxin concentration. regenerate the sensor. procedure times on theM1 same chip, and analyzed the We response. can clearly observe a significant decrease of the of concentration. the sensor after the sensor Results are reported in Figure 5 for a 100 nM sensitivity Aflatoxin M1 first glycine-methanol injection, and again after the second one. of The We can clearly observe a significant decrease of the sensitivity theregeneration sensor after of the the first antibody-functionalized surfacesand can beagain obtained using are mainly of based glycine-methanol injection, after thedifferent secondapproaches. one. TheThese regeneration the onantibody-functionalized solutions at low pH (HCl,surfaces Glycine-HCl), at high pH (NaOH) or at high ionic strength (MgCl2) [18]. can beorobtained using different approaches. These are mainly From five to 39 regeneration cyclesGlycine-HCl), are reported for thehigh whole while based onup solutions at low pH (HCl, or at pHantibody (NaOH) molecule or at high[18,19], ionic strength for Fab1 molecules, a regeneration four cycles has been [20].for Different solutions were tested (MgCl2) [18]. From five up to 39of regeneration cycles arefound reported the whole antibody molecule 1 -based platform. Among them, the acidic one revealed itself to be the most effective. on our Fab [18,19], while for Fab’ molecules, a regeneration of four cycles has been found [20]. Different 1 regions resulting in an unfolding of However, solutions the thermal stability of Fab solutionsacidic were tested on decrease our Fab’-based platform. Among them, the acidic one revealed itself to be the domains, and this could explain our poor regenerability. This behavior was regions also confirmed theFab most effective. However, acidic solutions decrease the thermal stability of Fab’ resulting in an unfolding of the Fab domains, and this could explain our poor regenerability. This behavior

7/10

Biosensors 2016, 6, 0001 Biosensors 2016, 6, 0001 Biosensors 2016, 6, 1

8 of 10 8 of 10 8 of 10

in an unfolding of the Fab domains, and this could explain our poor regenerability. This behavior in an unfolding of the Fab domains, and this could explain our poor regenerability. This behavior was also confirmed by experiments performed on functionalized Si3N4 flat substrate. In these, the was also confirmed by experiments performed on functionalized Si3N4 flat substrate. In these, the by experiments performed on functionalized Si3 N4 flat Inas these, the Fab immobilization Fab’ immobilization on silanized flat Si3N4 surfaces wassubstrate. performed reported in 1Section 2.2. After Fab’ immobilization on silanized flat Si3N4 surfaces was performed as reported in Section 2.2. After on silanized flat Si N surfaces was performed as reported in Section 2.2. After immobilization, immobilization, the were incubated with AFM1-HRP as described in Section 3.1. After the the 3 surfaces 4 immobilization, the surfaces were incubated with AFM1-HRP as described in Section 3.1. After the surfaces were incubated withwere AFM1-HRP Section 3.1.then After the first incubation, the first incubation, the surfaces washedas in described PBS-EDTAinbuffer and regeneration with 100 mM first incubation, the surfaces were washed in PBS-EDTA buffer and then regeneration with 100 mM surfaces weresolution washed with in PBS-EDTA buffer and then with 100 mM solution glycine-HCl 10% v/v of methanol wasregeneration applied for one hour. Afterglycine-HCl three washing steps glycine-HCl solution with 10% v/v of methanol was applied for one hour. After three washing steps with 10% v/v of methanol was applied for one hour. After three washing steps in MES buffer, a in MES buffer, a new incubation with AFM1-HRP was performed. As reported in Figure new 6, a in MES buffer, a new incubation with AFM1-HRP was performed. As reported in Figure 6, a incubation was capture performed. As reported 6, aofdecrease in the ability to again decrease inwith the AFM1-HRP ability to again Aflatoxin M1 asina Figure function regeneration procedure was decrease in the ability to again capture Aflatoxin M1 as a function of regeneration procedure was capture Aflatoxin M1 as is a function of regeneration procedure This trend is in rather recorded. This trend in rather good agreement withwas therecorded. measurements performed on recorded. This trend is in rather good agreement with the measurements performed on good agreementaMZI with the performedthe on different functionalized aMZI (see Figure 5), used considering functionalized (seemeasurements Figure 5), considering regeneration procedures for the functionalized aMZI (see Figure 5), considering the different regeneration procedures used for the the regeneration procedures used for the two (for aMZI, regeneration was twodifferent measurements (for aMZI, the regeneration wasmeasurements performed flowing the the solution for 20 min, two measurements (for aMZI, the regeneration was performed flowing the solution for 20 min, performed flowing the solution for 20 min, while for flat substrates an orbital shaking was applied while for flat substrates an orbital shaking was applied for 1 h). while for flat substrates an orbital shaking was applied for 1 h). for 1 h).

Figure 5. Sensorgram recorded on one single aMZI sensor by flowing a 100 nM AFM1 solution in the Figure 5. 5. Sensorgram Sensorgram recorded recorded on on one one single single aMZI aMZI sensor sensor by by flowing flowing aa 100 100 nM nM AFM1 AFM1 solution solution in in the the Figure microfluidic chamber. The first curve is the response of the fresh sensor, (black line), the second one microfluidic chamber. chamber. The The first first curve curve is is the the response response of of the the fresh fresh sensor, sensor, (black (black line), one microfluidic line), the the second second one after one glycine –methanol injection (red line) and the third one after two injections (blue line). after one one glycine glycine –methanol after –methanol injection injection (red (red line) line) and and the the third third one one after after two two injections injections (blue (blue line). line).

Figure 6. Chemiluminescence detection of AFM1 on Si3 N4 substrates after regeneration cycles with Figure 6. Chemiluminescence detection of AFM1 on Si3N4 substrates after regeneration cycles with 100 mM6.glycine-methanol solution. The of data are reported a mean value of three samples andwith the Figure Chemiluminescence detection AFM1 on Si3N4 as substrates after regeneration cycles 100 mM glycine-methanol solution. The data are reported as a mean value of three samples and the errormM barsglycine-methanol are reported as standard 100 solution.deviations. The data are reported as a mean value of three samples and the error bars are reported as standard deviations. error bars are reported as standard deviations.

4. Conclusions 4. Conclusions 4. Conclusions In 44 biosensors In this this article, article, we wedesigned designedand andtested testedSiSi3 N 3N biosensors based based on onasymmetric asymmetric Mach–Zehnder Mach–Zehnder ´7 RIU) In this article, we designed and tested Si3N 47 biosensors based on asymmetric interferometers. We achieved a low LOD (LOD = ˆ 10 and demonstrated a Mach–Zehnder high selectivity interferometers. We achieved a low LOD (LOD = 7 × 10−7−7 RIU) and demonstrated a high selectivity to interferometers. We1achieved a low LOD (LOD =The 7 × 10 RIU) and demonstrated a high selectivity to to AFM1 when Fab functionalization is used. measured minimum concentration of AFM1 is AFM1 when Fab’ functionalization is used. The measured minimum concentration of AFM1 is still AFM1 when Fab’ functionalization is used. The measured minimum concentration of AFM1 is still still higher minimum concentration prescribed regulations, whichindicates indicates the need need of higher thanthan thethe minimum concentration prescribed bybyregulations, which the of aa higher than the minimum concentration prescribed by regulations, which indicates the are need of to a preconcentration module. Other AFM1 sensors have been reported in the literature which able preconcentration module. Other AFM1 sensors have been reported in the literature which are able to preconcentration module. Other AFM1 sensors have been reported in the literature which are able to

8/10 8/10

Biosensors 2016, 6, 1

9 of 10

detect AFM1 concentration in the low ppt interval [21,22]. However, the sensor here proposed has the advantage of being compact, easily integrable and scalable to multi-analyte possibility. We finally performed regeneration sensing measurements on functionalized sensors using glycine-methanol solutions and observed a clear decrease of the efficiency of the sensors after the first cleaning process. These regeneration issues were also confirmed by chemical analysis. Therefore, the proposed sensors are suitable for disposable sensors. If one would like to use them as reusable sensors, the chemistry of the regeneration should be improved with respect to the procedure here presented. Let us clarify that the measurements we did were on purified samples. Further work still needs to be carried out if the sensor should be used with raw milk samples due to the complexity of the milk matrix. Filter and preconcentration stages should be considered to expose the sensor to a purified and concentrated sample of milk. Acknowledgments: This work was supported by the FP7 EU project “Symphony” (Grant agreement No: 610580). Author Contributions: Floris Falke and Erik Schreuder fabricated the samples. Laura Pasquardini and Manuela Zanetti functionalized the samples. Romain Guider and Tatevik Chalyan carried out the measurements. Romain Guider, Tatevik Chalyan and Laura Pasquardini analyzed the data. Romain Guider, Laura Pasquardini and Floris Falke wrote the paper. Lorenzo Pavesi, Cecilia Pederzolli and Rene G. Heideman led the project. All authors discussed the results and commented on the manuscript. Conflicts of Interest: The authors declare no conflict of interest.

References 1.

2. 3.

4.

5.

6. 7. 8. 9.

10.

11.

Sun, X.; Dai, D.; Thylén, L.; Wosinski, L. High-sensitivity liquid refractive-index sensor based on a Mach-Zehnder interferometer with a double-slot hybrid plasmonic waveguide. Opt. Express 2015, 23, 25688–25699. [CrossRef] [PubMed] Homola, J. Surface plasmon resonance sensors for detection of chemical and biological species. Chem. Rev. 2008, 108, 462–493. [CrossRef] [PubMed] Iqbal, M.; Gleeson, M.A.; Spaugh, B.; Tybor, F.; Gunn, W.G.; Hochberg, M.; Baehr-Jones, T.; Bailey, R.C.; Gunn, L.C. Label-free biosensor arrays based on silicon ring resonators and high-speed optical scanning instrumentation. IEEE J. Sel. Top. Quantum Electron. 2010, 16, 654–661. [CrossRef] Guider, R.; Gandolfi, D.; Chalyan, T.; Pasquardini, L.; Samusenko, A.; Pucker, G.; Pederzolli, C.; Pavesi, L. Design and Optimization of SiON Ring Resonator-Based Biosensors for Aflatoxin M1 Detection. Sensors 2015, 15, 17300–17312. [CrossRef] [PubMed] Prieto, F.; Sepulveda, B.; Calle, A.; Llobera, A.; Domínguez, C.; Abad, A.; Montoya, A.; Lechuga, L.M. An integrated optical interferometric nanodevice based on silicon technology for biosensor applications. Nanotechnology 2003, 14. [CrossRef] Estevez, M.C.; Alvarez, M.; Lechuga, L.M. Integrated optical devices for lab-on-a-chip biosensing applications. Laser Photon. Rev. 2012, 6, 463–487. [CrossRef] Crivianu-Gaita, V.; Thompson, M. Immobilization of Fab1 fragments onto substrate surfaces: A survey of methods and applications. Biosens. Bioelectron. 2015, 70, 167–180. [CrossRef] [PubMed] Ertürk, G.; Uzun, L.; Tümer, M.A.; Say, R.; Denizli, A. Fab fragments imprinted SPR biosensor for real-time human immunoglobulin G detection. Biosens. Bioelectron. 2011, 28, 97–104. [CrossRef] [PubMed] Mustafaoglu, N.; Alves, N.J.; Bilgicer, B. Oriented Immobilization of Fab Fragments by Site-Specific Biotinylation at the Conserved Nucleotide Binding Site for Enhanced Antigen Detection. Langmuir 2015, 31, 9728–9736. [CrossRef] [PubMed] Nguyen, A.H.; Lee, J.; Choi, H.I.; Kwak, H.S.; Sim, S.J. Fabrication of Plasmon length-based surface enhanced Raman scattering for multiplex detection on microfluidic device. Biosens. Bioelectron. 2015, 70, 358–365. [CrossRef] [PubMed] Yoshimoto, K.; Nishio, M.; Sugasawa, H.; Nagasaki, Y. Direct Observation of Adsorption-Induced Inactivation of Antibody Fragments Surrounded by Mixed-PEG Layer on a Gold Surface. J. Am. Chem. Soc. 2010, 132, 7982–7989. [CrossRef] [PubMed]

Biosensors 2016, 6, 1

12.

13. 14. 15.

16.

17.

18.

19.

20.

21.

22.

10 of 10

Pasquardini, L.; Pancheri, L.; Potrich, C.; Ferri, A.; Piemonte, C.; Lunelli, L.; Napione, L.; Comunanza, V.; Alvaro, M.; Vanzetti, L.; et al. SPAD aptasensor for the detection of circulating protein biomarkers. Biosens. Bioelectron. 2015, 68, 500–507. [CrossRef] [PubMed] Wörhoff, K.; Heideman, R.G.; Leinse, A.; Hoekman, M. TriPleX: A versatile dielectric photonic platform. Adv. Opt. Technol. 2015, 4, 189–207. [CrossRef] Heideman, R.; Hoekman, M.; Schreuder, E. TriPleX-based integrated optical ring resonators for lab-on-a-chip and environmental detection. IEEE J. Sel. Top. Quantum Electron. 2012, 18, 1583–1596. [CrossRef] Gandolfi, D.; Guider, R.; Chalyan, T.; Pavesi, L.; Pasquardini, L.; Pederzolli, C.; Samusenko, A.; Pucker, G. Sensitivity and Limit of detection of biosensors based on ring resonators. In Proceedings of the AISEM Annual Conference 2015 XVIII, Trento, Italy, 3–5 February 2015. Brosinger, F.; Freimuth, H.; Lacher, M.; Ehrfeld, W.; Gedig, E.; Katerkamp, A.; Spener, F.; Cammann, K. A label-free affinity sensor with compensation of unspecific protein interaction by a highly sensitive integrated optical Mach–Zehnder interferometer on silicon. Sens. Actuators B Chem. 1997, 44, 350–355. Liu, Q.; Tu, X.; Kim, K.W.; Kee, J.S.; Shin, Y.; Han, K.; Yoon, Y.J.; Lo, G.Q.; Park, M.K. Highly sensitive Mach–Zehnder interferometer biosensor based on silicon nitride slot waveguide. Sens. Actuators B Chem. 2013, 188, 681–688. [CrossRef] Jantra, J.; Kanatharana, P.; Asawatreratanakul, P.; Wongkittisuksa, B.; Limsakul, C.; Thavarungkul, P. Detection of staphylococcal enterotoxin A (SEA) at picogram level by a capacitive immunosensor. J. Environ. Sci. Health, Part A 2011, 46, 560–568. [CrossRef] [PubMed] Zhao, Q.; Duan, R.; Yuan, J.; Quan, Y.; Yang, H.; Xi, M. A reusable localized surface plasmon resonance biosensor for quantitative detection of serum squamous cell carcinoma antigen in cervical cancer patients based on silver nanoparticles array. Int. J. Nanomed. 2014, 9, 1097–1104. Edupuganti, S.R.; Edupuganti, O.P.; Hearty, S.; O’Kennedy, R. A highly stable, sensitive, regenerable and rapid immunoassay for detecting aflatoxin B1 in corn incorporating covalent AFB1 immobilization and a recombinant Fab antibody. Talanta 2013, 115, 329–335. [CrossRef] [PubMed] Liu, D.; Huang, Y.; Wang, S.; Liu, K.; Chen, M.; Xiong, Y.; Yang, W.; Lai, W. A modified lateral flow immunoassay for the detection of trace aflatoxin M1 based on immunomagnetic nanobeads with different antibody concentrations. Food Control. 2015, 51, 218–224. [CrossRef] Istamboulié, G.; Paniel, N.; Zara, L.; Reguillo, G.L.; Barthelmebs, L.; Noguer, T. Development of an impedimetric aptasensor for the determination of aflatoxin M1 in milk. Talanta 2016, 146, 464–469. [CrossRef] [PubMed] © 2016 by the authors; licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons by Attribution (CC-BY) license (http://creativecommons.org/licenses/by/4.0/).