Dielectrophoresis-Assisted Pathogen Detection on Vertically Aligned Carbon Nanofibers Arrays in a Microfluidic Device Dielectrophoresis-Assisted Pathogen Detection on Vertically Aligned Carbon Nanofibers Arrays in a Microfluidic Device

In this chapter, we focus on utilizing nanoelectrode arrays fabricated with vertically carbon nanofibers (VACNFs) for pathogen detection based on a “point-and-lid” dielectrophoretic device in a microfluidic channel. This technique is utilized to concentrate particles from the bulk flow and detect pathogens based on fluorescence, surface-enhanced Raman spectroscopy (SERS) and impedance measurements. The advantage of VACNFs is their ultrasmall diameter (~100 nm) and the high aspect ratio (50:1). When coupled with a macroscopic indium tin oxide (ITO) electrode, it produces a large electric field gradient ( ∇ E 2 = ~10 19 − 10 20 V 2 m −3 ) which is harnessed for pathogen detection based on dielectrophoresis. Several noninfectious pathogens including bacteria Escherichia coli DHα5, inactivated vaccinia virus (species: Copenhagen strain , VC - 2 ), and Bacteriophage T4r were utilized as model species to study the size effect and kinetics of dielectrophoretic capture in this study. The comparable size of the nanoelectrode produced strong interaction with virus particles, generating striking lightning capture patterns and high detec- tion sensitivity. The dielectrophoretic capture at the nanoelectrode arrays is successfully integrated with a portable Raman probe as a microfluidic chip for ultrasensitive detec- tion of bacteria E. coli DHα5 using SERS-tagged gold nanoparticles co-functionalized with specific antibodies. a lytic act as an on-chip portable sample preparation module for poten-tially capturing pathogenic particles at concentrations approaching 1–10 particles/mL and for future downstream processing and testing of microbial samples.


Introduction
The need for rapid and reliable pathogen monitoring and detection is imperative in the food industry, biodefense, drug discovery, animal healthcare, clinical diagnosis, water, and environmental quality control. Among these, the food industry is the area where most attention has been focused on due to public health implications. In 2015, the World Health Organization (WHO) estimated that 77 million people every year fall victim to contaminated food and about 9000 deaths annually. The WHO has identified 31 agents of foodborne diseases including bacteria, virus, parasites, toxins, and chemicals, among which 95% are caused by Norovirus, Campylobacter, Escherichia coli, and non-typhoidal Salmonella [1]. In recent years, there have been considerable efforts to develop devices and methods for capturing pathogens in fluids such as blood, food matrices, soil, bodily fluids, and water for rapid detection.
The conventional pathogen identification methods are standard microbiological techniques and involve necessary steps such as preenrichment, selective enrichment, biochemical screening, and serological confirmation [2]. The traditional methods take up to 72 h to obtain confirmed results which are based on the morphological evaluation, culture growth in various media under various conditions, and enumerating colonies of the bacteria [3,4]. However, the development of polymerase chain reaction (PCR)-based molecular analysis techniques [5][6][7], the conventional biochemical methods such as enzyme-linked immunosorbent assays (ELISAs), and blot assays have led scientists to target genes, proteins, and carbohydrate moieties instead of the whole microorganisms [8] to obtain molecular fingerprints of the pathogens. These techniques despite being highly sensitive and selective require experienced personnel, expensive equipment, reagents, and long readout time, thus making the process costly and difficult for onsite applications and causing a delay in the pathogen detection, preventing immediate medical action toward infected patients. There is a keen interest in developing new rapid point-of-care biosensing systems for early detection of pathogens with high sensitivity and specificity.
In this chapter, we summarize an innovative pathogen capture and detection system based on dielectrophoresis (DEP). The device is a unique assembly of nanoelectrode arrays (NEAs) fabricated with vertically aligned carbon nanofibers (VACNFs) and a transparent macroscopic indium tin oxide (ITO) glass electrode in a "point-and-lid" geometry in which pathogens are introduced using microfluidic channels. The study of capture kinetics was accomplished using fluorescence, SERS, and impedance measurement techniques. The test pathogens utilized in this study were bacteria such as E. coli DHα5 (nonpathogenic) and viruses such as bacteriophage and inactivated vaccinia virus. There have been several reports on using microscale DEP devices for manipulation of mammalian cells (tens of microns) to bacterial cells (~1.0 micron) [40]. The DEP force is proportional to the volume of the target particles [41] and decreases rapidly when the particle size is reduced to only ~100 nm. Therefore it becomes essential to fabricate nanostructured DEP electrodes to capture virus particles due to their small sizes (ranging from 10s nm to 300 nm). We illustrate in this chapter the method to use VACNFs for fabricating stable nanoscale DEP devices. The capture of virus Bacteriophage T4r and T1 using fluorescence and impedance sensing of vaccinia virus accompanied by electroporation has been accomplished due to the high electric field focused on the tips on VACNFs. Last, we have demonstrated the specific detection of bacterial cells using SERS reporter QSY21 that is co-functionalized with polyclonal antibodies on a special type of plasmonic nanoparticles, i.e., anisotropic oval-shaped iron-oxide-gold (IO-Au) core-shell nanoparticles. This dielectrophoretic device is integrated with a portable Raman system for rapid pathogen detection in field applications. Such integrated microfluidic systems provide simultaneous concentration and identification of specific microbes in dilute samples.

Principles, design, and fabrication
The phenomenon of dielectrophoresis (DEP) is renowned as a particle manipulation technique based on the uneven electrical force on the opposite sides of polarized particles in an electric field with a high gradient produced by the electrodes. The larger the electric field gradient, the stronger the DEP force acts on the particle. This phenomenon was first described by Pohl in 1951 [42] and has been widely used in biological science to separate live and dead bacteria [43,44], viruses [45][46][47], cells [48][49][50][51][52], yeast cells [53,54], and DNA [55][56][57]. When we consider radius of the particle r, the permittivity of the suspending medium ε m , the gradient of the square of the applied electric field strength ∇E 2 and the real component of the complex Clausius-Mossotti (CM) factor Re[K(ω)], the time average DEP force (F DEP ) acting on the spherical particles by the nonuniform electric field, and the Re[K(ω)] are provided by the following equations: where: The use of physical fields for the separation of cells takes advantage of the heterogeneity of physical parameters for Eq. (2), such as ε * , representing the complex permittivity and the indices p and m referring to the particle and medium, respectively; parameter σ is the conductivity; ω is the angular frequency (ω = 2πf) of the applied electric field; and j = √−1. The direction of the force, either toward the field gradient as in positive DEP or away from it as in negative DEP, is given by the difference in complex permittivity conductivity between the particle of interest and the suspending media. In this study, the proper medium (280 mM mannitol solution) is chosen to give Re[K(ω)] > 0 so the particles experience a positive DEP (pDEP) force, directing toward higher electric field strength, which is desirable for capture bacteria and viruses at the exposed VACNF tips by selecting a proper frequency.
In the microfluidic device, a particle experiences two forces orthogonal to each other, i.e., DEP force (F DEP ) forcing the particles to capture on the tips and hydrodynamic force to carry the particles with the flow (i.e., Stokes drag force F Drag ) (as shown in Figure 1d). F DEP is proportional to the volume or cube of the radius (r 3 ) of the particle. F Drag is directly proportional to the radius of the particle by where η is the dynamic viscosity, k is a small factor accounting for the wall effects, and υ is the linear flow rate (flow velocity). Sedimentation force and Brownian force are negligible for bacteria but not for submicron particles. The advantage of nanostructured DEP devices is that the magnitude of ∇E 2 can be enhanced by orders of magnitude so even small viral particles can be captured. Figure 1a is the image of the device produced in the lab at Kansas State University. The detailed procedure of device fabrication is given in reference [58]. Figure 1a shows that the size of the devices is comparable to a US penny and illustrates the "points-and-lid" design. Figure 1b shows that the NEA comprises randomly distributed VACNFs (diameter ~100-120 nm, the density of ~2 × 10 7 exposed CNFs/cm 2 ) embedded in silicon dioxide (SiO 2 ) matrix (tip exposed) with an average spacing of ~1-2 μ. The active area exposed on NEA is 200 × 200 μm 2 , and the rest is covered with a 2-μm-thick photoresist film to shield the effect of the rest of exposed tips. The ITO glass slide containing a photolithographically fabricated 500-μm-wide microfluidic channel in an 18-μm-thick photoresist film is permanently vacuum bonded.

DEP device fabrication and setup for pathogenic particles
In the experimental setup, DEP device was placed under an upright fluorescence optical microscope (Axioskop II, Carl Zeiss) using 50 X objective lens. The microorganisms such as Bacteriophage T4r (labeled with SYBR green I dye) and E. coli DHα5 (Alexa 555) fluorescence detection filter sets were configured to an excitation wavelength of 540-552 nm and an emission wavelength of 567-647 nm (filter set 20HE, Carl Zeiss). For vaccinia virus detection, filter sets were configured to 465-505 nm excitation wavelength and an emission wavelength of 515-565 nm (filter set 17, Carl Zeiss) for 3,3′-Dioctadecyloxacarbocyanine (DiO) dye and an excitation wavelength of 620-640 nm and an emission wavelength of 640-740 nm (filter set 60, Carl Zeiss) for propidium iodide (PI) dye. The videos were recorded using Axio Cam MRm digital camera to record fluorescence videos at varying exposure times depending on the pathogen species using multidimensional acquisition mode in the Axio-vision 4.7.1 release software (Carl Zeiss MicroImaging, Inc.). To prevent biofouling, the microfluidic channel was injected with 1.0 mL bovine serum albumin (BSA) solution (2.0 gm in 100 mL of water) at a flow rate of 0.2 μl/min before performing DEP experiments. The channel was then rinsed with 2.0 mL DI water at a flow rate of 5.0 μl/min. Figure 1c shows the microfluidic design in which the particles entered from the narrow straight channel (500 μm in width) are distributed into the larger circular microchamber (2.0 mm in diameter), and only a fraction of the particles are passed over 200 × 200 μm active NEA area. Using this setup, bacteria E. coli DHα5 counting was accomplished using Axio software. In contrast, when detecting viral particles, it became difficult to distinguish the single viral particles. Hence for virus capture experiments, the integrated fluorescence intensity over the 200 × 200 μm active NEA area was recorded. For this, the initial fluorescence background (F 0 ) immediately before the V pp was applied was subtracted from the final fluorescence signal (F F ) at the end of the capture period, giving the fluorescence intensity increase (ΔF) to represent the quantity of captured virus. The counts of isolated bright spots of single viral particles were observed at much lower virus concentration and were used in some later experiments to quantify the capture efficiency during the kinetic DEP process [59].    Figure 2a shows the maximum capture frequency to be 10 kHz. Figure 2b, a plot of the integrated fluorescence intensity of captured viruses vs. the flow velocity, showed a maximum at 0.73 mm/s. At ν ≤ 0.73 mm/s, isolated bright spots were seen (Figure 2c). At ν ≥ 0.73 mm/s, viruses depicted fractal-like lightening patterns (Figure 2d). These patterns are called Lichtenberg figures, which occur when the high electric field is produced at a sharp electrode surrounded by a relatively high concentration of polarizable particles. Previously such pattern was observed using E. coli cells between interdigitated microelectrodes called as "pearl-chain-like" Suehiro et al. [59,62] This is first time that such pattern was observed with virus particles with the electric field produced at the nanoelectrode tips [59].

Detection of viruses: Bacteriophages and vaccinia virus using
The DEP kinetics dramatically changed with concentration (Figure 2e) when two diluted concentrations, i.e., 5.5 × 10 8 and 2.5 × 10 7 pfu/ml, were used. The viruses could be individually counted (40 out of 67 particles) at an extremely low concentration of Bacteriophage T1 (8.7 × 10 4 pfu/mL) when passed through the nano-DEP device as the capture was limited by mass transport giving a capture efficiency ~60% [59].

DEP capture and electroporation of vaccinia virus coupled with real-time impedance detection
Electrochemical sensors based on impedimetric measurements have emerged as an attractive low-cost portable technique for the rapid detection of pathogenic microbes and other microorganisms. In this capture study, vaccinia virus was a probe to study the impedance kinetics and electroporation of the viruses due to high electrical field gradient generated at VACNFs tips.
The details of the growth and enumeration by conventional techniques are given in a previous report [60]. Briefly, in-house stocks of vaccinia virus (Copenhagen strain, VC-2) were amplified by standard virus growth techniques of infecting HeLa cells knocked-down for an antiviral protein kinase, PKR (HeLa PKR-KD), followed by sucrose gradient centrifuge to achieve an optimal yield of 2.0 × 10 8 pfu/mL quantified via plaque assay. To move the viruses out of biosafety level 2 (BSL-2) containment, a UV-inactivation process was carried out by placing them 3-8 cm directly below a UV lamp (234 nm), and the plate was manually rocked for 10 min.  was used as the capture frequency (except few experiments). A concentration ~3 × 10 3 to 3 × 10 6 pfu/mL was employed to demonstrate concentration-dependent study with the real-time potential and measures the limit of detection of the impedance method. For this experiment, the frequency is fixed at 1.0 kHz, and the 0.5 V PP AC voltage is applied for 16 s to obtain the background impedance value. The AC voltage is then increased to 8.0 V PP for 54 s to capture the virus (indicated by the increase in the impedance signal). The AC voltage is finally reduced back to 0.5 V PP , indicating the release of the virus particles from the VACNF tips. The control experiment is done under the same conditions using blank 280 mM mannitol solution. The calibration curve (Figure 3b) shows that the logarithm of virus concentration (C) in the range from ~300 to 30,000 particles/mL is linear with percentage impedance change collected over the 200 × 200 μm active area. Using the calibration curve equation (details in Refs. [61,66]), the detection limit of vaccinia virus was calculated to be ~ 300 particles/mL [61].
Finally, to investigate the electroporation of lipophilic membrane due to the high electric field on tips of VACNF NEAs, PI dye was added to the mannitol solution containing 3.0 × 10 6 particles/mL of DiO dye-labeled vaccinia virus and observed in Neubauer chamber. The absence of the red fluorescence indicated there was no structural damage of virus due to UV inactivation [62]. For electroporation experiment in a microfluidic device, the frequency of 50.0 Hz was used. The voltage of 8.0 V pp was turned on for 65 s for maximal DEP capture with the flow velocity set at 0.05 mm/s (for maximum capture and interaction of dye and DNA). Figure 3f shows the schematic figure of electroporation of lipophilic membrane of vaccinia virus in the presence of high electric field at the VACNF tips. It is observed that the electroporation made the membrane more permeable and the DNA is likely extracted out of the membrane to interact with PI dye in the mannitol solution which increases the PI dye fluorescence intensity. There is evidence that, after the AC voltage is turned off, some PI-intercalated ds-DNAs are physically adsorbed on the VACNF tip or the NEA chip surface [61,63].

Detection of bacteria: DEP capture and identification of E. coli strain DHα5 by surface-enhanced Raman spectroscopy
DEP capture of bacterial cells was demonstrated with nontoxic E. coli strain DHα5 (18265-017, Fisher Scientific). The DEP microfluidic device with Raman setup is schematically represented in Figure 4a. The procedure for fluorescent labeling and attachment of the bacteria with a unique SERS nanotag consisting of nanooval (NOV)-shaped gold coating on spherical iron-oxide (IO) nanoparticles is reported in Ref. [62]. Gold is coated on the spherical IO nanoparticle cores (~23 nm diameter), forms an outer dimension of ~50 nm with NOVs which demonstrates high SERS enhancement factor due to Raman tag of QSY-21. The NOVs are made biocompatible, and surface active by adding carboxylic acid groups at the surface was accomplished by coating them with carboxyl-polyethylene glycol-thiol (HOOC-PEG-SH, MW 5000) and methoxy-polyethylene glycol-thiol (mPEG-SH, MW 5000). The details of the process are given in Ref. [63,66]. The carboxylic acid group aids in the formation of amide covalent bonds with secondary IgG antibody conjugated with Alexa 555 making the IO-NOVs fluorescently labeled. These secondary antibodies are complementary to IgG antibody conjugated with FITC primary antibody which was attached to bacteria E. coli DHα5. The detailed procedure is given in reference [63,64]. Figure 4b and c show TEM images of IO-Au SERS NOVs and those bound to E. coli. The typical Raman spectrum of QSY21 has prominent bands at 1333, 1584, and 1641 cm −1 which are from the xanthene ring stretching vibrations of the molecule [65]. The intensity of QSY21 marker at the Raman shift of 1496 cm −1 is visually separated from the carbon nanofiber signals at 1350 cm −1 (D-band) and 1600 cm −1 (G-band), respectively [66]. The most reliable characteristic band is seen at 1496 cm −1 as seen in Figure 4d was used to quantify the SERS signal.
To demonstrate the potential of this method, both confocal (DXR, Thermo Fisher Scientific) and portable systems (ProRaman L, Enwave Optronics. Inc) were used. The similar studies were carried out with the two spectrophotometers at varied flow velocity and frequency. Figure 4d shows the full Raman spectrum of QSY21 at different AC frequencies during the capture of bacteria. The highest peak in the full spectra, 1496 cm −1 , was used in the further calculation, and the higher capture was seen at the AC frequency of 100.0 kHz. The results between these two Raman systems were very consistent from their fluorescence and Raman intensity plots, with the maximum flow velocity at 0.4 mm/s (0.55 μl/s) (Figure 4e). To analyze the capture in complex samples, one of the representative data is shown in Figure 4f, i.e., the capture of E. coli in the chicken broth samples. Other samples such as Mott's apple juice and soil samples were also tested, and details are given in Ref. [62]. Complex matrix solution was centrifuged at 14,000 rpm for 10 min, and the supernatant was collected. Complex matrices present different challenges due to inorganic and organic substance interactions, making it difficult to isolate the target to be tested. A concentration of 5 × 10 5 cells/mL E. coli DHα5 was added into the solution of processed chicken broth [66]. The conductivity of bacteria in distilled water (pH 6.

Figure 4g
summarizes the SERS intensity of the captured NOV-labeled E. coli using the portable Raman setup, while the E. coli concentration was varied from ~10 to 1 × 10 9 cells/mL. The probe diameter at the focal point in the portable Raman system is about 100 μm (inset in Figure 4g), much larger than the 3.1 μm size in the confocal Raman microscope allowing signals to be collected from many more bacteria, and yields better statistics of the detection limit measurement [66]. The calibration curve for the detection limit measure is shown in Figure 4g.
The Raman intensity was a linear function of the logarithm of bacteria concentration when the concentration C is above ~100 cells/mL [66]: where RI was the Raman intensity increase after 50 s of DEP capture. For bacteria concentrations below the critical value, C 0 = ~100 cells/mL. There was no measurable signal detected above the background, i.e., (RI) blank = ~36 a.u. Due to the slow mass transport of bacteria to the active area, no captured bacterial cells were detected during the applied DEP period. However, at the time when the high concentration of the bacteria was passed, the Raman intensity increased. The detection limit logC dl was determined using calibration curve: where s blank (~11.7) is the standard deviation of the Raman signal for bacteria concentration below C 0 and m = 108.8 is the slope of the calibration curve. The concentration detection limit was determined to be ~210 cells/mL.

Discussion and conclusion
The physical phenomenon of DEP was observed on the tips of VACNF NEAs in microfluidic channel design due to high electric field gradient generated by the "point-and-lid" geometry acted as an effective and reversible electronic manipulation technique to rapidly (less than 60 s) concentrate bacteria and viruses into a micro-area from the solution flow. The nanoscale size of the VACNF tips has two critical features: the extremely high electrical field strength at the tip (E = ∼10 7 V m −1 ) and the large electric field gradient at the tips of nanoelectrode (giving ∇E 2 = 10 19 − 10 20 V 2 m −3 ) against ITO electrode. The polarizable pathogenic particles in the microfluidic device encounter hydrodynamic drag force along the flow direction and orthogonal (vertical) DEP forces due to the high electric field gradient. Once the pathogens are close to the VACNF tip, the lateral DEP force becomes larger than the hydrodynamic drag force, and the pathogens are captured at the nanoelectrode tip.
According to Eq. (1), the force of DEP highly depends on the volume of the particles (r 3 ) and the optimal frequency for capture bacteria, E. coli, Bacteriophage T4r, and vaccinia virus at the NEA tips was found to be 100 kHz, 10.0 kHz, and 50 Hz, respectively. This variation is due to the differences in size, structure, and molecular composition. Bacteriophage T4r virus has ~ 80-100 nm icosahedral-shaped protein capsid encapsulating ds-DNAs [67]. Vaccinia virus particles are larger spheres with dimensions of 360 × 270 × 250 nm, which consist of the lipophilic membranes encapsulating ds-DNAs [68]. Bacteria E. coli DHα5 is ~1 micron with an elongated shape and more complex internal structures [69,70].
The second drastic contrast in the capture of viruses is the formation of Lichtenburg figures which was absent during capture of the bacteria. This is due to the spatial distribution of the electrical field strength at the nanoelectrode tip. Bacteriophage T4r and vaccinia virus are similar in size and are comparable to the diameter of VACNFs, causing the viruses to be polarized to a large extent. The captured virus acts as an extended tip attracting more viruses toward it. For bacteria, the more significant size (~1 μ) and higher internal conductivity may have screened the high electric field at the nanoelectrode tip and reduced the electrical interaction with additional cells [58,59,64,70].
The device successfully captured single virus particles observed at isolated spots in the 200 × 200 μm 2 active NEA surface at an extremely dilute concentration (8.9 × 10 4 pfu/ml) in which facilitated studying the impedance kinetics of real-time DEP capture of vaccinia viral particles, yielding a detection limit of 300 particles/ml. VACNF tips have been found to cause electroporation of the lipophilic membrane of the vaccinia virus due to the large electric field produced on the tips. This electroporation phenomenon has allowed extracting the internal nucleic acid contents to the solution.
Finally, highly sensitive detection of E. coli bacteria using the SERS nanotag based on QSY21 on IO-Au NOVs proved to be highly sensitive. This was accomplished using two complementary antibodies, in which the secondary antibody was bonded to nanoovals and other to the bacteria. The attachment of the nanoovals significantly enhanced Raman signals and aided in specific recognition to E. coli DHα5 cell. The detection and kinetics of capture were studied using both a confocal Raman microscope and a portable Raman system, and the limit of detection of 210 CFU/mL was calculated by calibration curve using the portable Raman system.
All these studies revealed the exciting interplay between the highly focused electric fields at the nanoelectrode with bioparticles of comparable sizes. The device was successfully integrated with fluorescence, surface-enhanced Raman spectroscopy and electrochemical impedance sensing. All these results are very encouraging and can be further improved by optimizing the DEP design. The combined functions of DEP in concentration, detection, and electroporation make such nano-DEP devices useful to extract intracellular materials, such as DNA or proteins without a lytic agent. It can act as an on-chip portable sample preparation module for potentially capturing pathogenic particles at concentrations approaching 1-10 particles/mL and for future downstream processing and testing of microbial samples.