A dielectrophoresis-impedance method for protein detection and analysis

Dielectrophoresis (DEP) has increasingly been used for the assessment of the electrical properties of molecular scale objects including proteins, DNA, nanotubes and nanowires. However, whilst techniques have been developed for the electrical characterisation of frequency-dependent DEP response, biomolecular study is usually limited to observation using fluorescent markers, limiting its applicability as a characterisation tool. In this paper we present a label-free, impedance-based method of characterisation applied to the determination of the electrical properties of colloidal protein molecules, specifically Bovine Serum Albumin (BSA). By monitoring the impedance between electrodes as proteins collect, it is shown to be possible to observe multi-dispersion behaviour. A DEP dispersion exhibited at 400 kHz is attributable to the orientational dispersion of the molecule, whilst a second, higher-frequency dispersion is attributed to a Maxwell-Wagner type dispersion; changes in behaviour with medium conductivi...


I. INTRODUCTION
For more than fifty years, dielectrophoresis (DEP) has been used for the manipulation of a variety of particles, both organic and inorganic, and for myriad applications including separation, fabrication and analysis. 1,2First described by Herbert Pohl in 1951, DEP is the motion of a particle due to its polarization in the presence of non-uniform electric field. 3The magnitude and direction of the DEP force is regulated by the relative polarizability of the particle, which depends on the dielectric properties of both the particle and the suspending medium, as well as the frequency of the energising field.
Recently, DEP has gained significant interest for the manipulation of molecules and nanostructures.For many years it was believed that DEP force, which scales according to the volume of the manipulated particle, was too small at molecular levels to overcome Brownian motion; nevertheless, the first demonstration of DEP manipulation of biomolecules was exhibited by Masao Washizu and co-workers in 1994. 4Since this time, a number of demonstrations of biomolecular DEP have been performed, many of which are reviewed elsewhere. 5More recent developments include the use of DEP to study conduction mechanisms in DNA; 6 [9][10][11] Serum Albumin has been one of the most extensively studied proteins in blood plasma.Some of the albumins most commonly studied are human serum albumin (HSA) rat serum albumin (RSA) and bovine serum albumin (BSA), 12 which are widely available at high purity and low cost.Serum albumin is a soluble multi-domain protein, without prosthetic group subjoin of carbohydrates that is absolutely stable; the primary structure is prolate elliptical in shape, with dimensions 140x40nm 13 with a low intrinsic viscosity.Their molecular weight is about 66200 D, in cysteine group (Cys-34) and low tryptophan content.Albumin helps in transport of drugs and ligands by binding to it and so reduces the free serum concentration of these compounds.Competitive binding of drug may occur at the same site or different sites (conformational changes) for example warfarin as anticoagulant and diazepam as derivative drug.Therefore, serum albumin acts as carriers for numerous exogenous and endogenous compounds. 14The primary structure of BSA is composed of 528 amino acid residues.The sequences have 17 disulphate bonds resulting in nine formed by the bridges.BSA contains of Asparagine, Glutamine, Alanine, Leucine and Lysine as well as the four amino acid residues in the sequence of Glycine-Phenylalanine-Glycine-Asparagine. 15ypically, experiments where DEP has been used to analyse the medium conductivity-dependent behaviour of biological macromolecules have been performed using fluorescent microscopy, 5 whereas characterisation of nanoscale particles such as carbon nanotubes 16 and zinc oxide nanowires 17,18 has employed impedance measuring techniques; impedance methods have been used for the study of larger particles such as cells, 19 but have never previously been applied to biological nanoparticles.In this paper, we present for the first time and impedance-based method for the determination of the electrical properties of dissolved biomolecules, using serum albumin as a model.

A. Protein solution
Solutions of bovine serum albumin (30 mg) were prepared by dissolving into 3 ml sodium carbonate (Na 2 CO 3 ) buffer.8 mg ammonium chloride (NH 4 Cl) was added and the solution incubated for 2 hours at 4 • C. The pH of this solution was adjusted to >8.0 by adding NaOH (the iso-electric point of BSA is <5, ensuring no conformational or similar surface charge effects were observed).Media conductivity of every samples recorded varies range from 1.1 mS/m minimum to 281.0 mS/m maximum.Finally, 120 µl glycerol were mixed into protein solution to enhance stability.All of the items were purchased from Sigma-Aldrich UK.A digital conductivity meter unit was used to measure the solution.The protein solution then filtered by Millipore membrane 0.22 µm with 5 ml syringe.Twelve samples were prepared at various conductivities to investigate their behaviours.Experiments were performed in laboratory with room temperature.Each experiment was replicated 3 times.

B. Measurement apparatus
DEP impedance spectroscopy of protein collection conducted using micro-electrodes, configured with an impedance analyser device.Electrodes were of a standard quadrupole geometry and fabricated of 100 nm Au on a 10nm Ti seed layer on glass (as shown in Figure 1); full details of the geometry and fabrication methods are available elsewhere. 6A resistance was connected in series with the electrodes, and the voltage across the resistor and electrode was measured using a NumetriQ analyser phase multimeter and impedance analyser (N4L, Leicester, UK).The impedance analyser was used a 4-terminal configuration to connect to the polynomial micro-tip electrode.This allowed the NumetriQ phase analyzer multimeter and impedance analyzer to apply current to the electrode with one pair of terminals (High Drive and High Sense) and (Low Drive and Low Sense), as shown in Figure 1.Dielectrophoresis was performed by applying AC electric fields at frequencies between 100 kHz and 16 MHz logarithmic spaced with 5 points per decade, with an applied voltage of 5 Vp-p.To record a single data point the electrodes were energised at a set frequency, then a 5 µl drop of the BSA solution was deposited to the middle of paired micro-electrode.Immediately upon the solution reaching the electrodes, the impedance was observed to exhibit a negative exponential change in impedance.The time constant of the change of resistance was determined after 300 seconds of recording.A typical deposition time of 5 minutes was then followed by rinsing the sample with methanol and gently blowing the drop off the surface.The experiment was repeated on micro-electrode at five frequencies per decade.The electrodes were connected using a Kelvin clip cable with one pair of terminals, HDrive/HSense (High Drive/High Sense), LDrive/LSense (Low Drive/Low Sense) as second pair of terminal connected to the second clip.

C. Data analysis
When particles experience positive DEP, those nearest the electrodes move rapidly to collect at the electrodes, depleting the region in the electrodes' immediate vicinity; over time, particles are collected from further and further away.This leads to a reduction in particle collection over time, which has been previously phenomenologically modelled as an exponential with time constant τ; as the particles accumulate in the inter-electrode gap, they alter (lower) the inter-electrode impedance as a negative exponential, allowing the determination of τ and hence the relative force at that frequency.By mapping the variation of τ as a function of frequency (but with constant particle concentration) it is possible to determine the DEP collection rate of the particles.This technique was originally developed for carbon nanotubes 16 and nanowires, 19 where it was found that conduction around the electrical double layer 20 dominated the DEP response of particles in this size regime.
The impedance change between the micro-electrodes at 1 second intervals was modelled using the following expression: when e y = x is equal to y = ln x gives Equation 4 into Equation 5: and The time constant τ of the impedance change data was determined using a Matalb curve fitting toolbox, as described elsewhere. 16,19The inverse of the time constant was then used to approximate the DEP force, as collection rate is inversely proportional to the force causing the particles to collect.

III. RESULTS
Figure 2 shows the inverse time constant for two medium conductivities, 1.1 and 100 mS/m.As can be seen, both plots exhibit two-dispersion characteristics, similar to those seen in collection of carbon nanotubes measured using the same method. 16In that case, the populations were attributed to conducting and semiconducting fractions.Here, two dispersions could be observed in the BSA spectrum, and in different proportions at different medium conductivities. .Figure 3 shows the variation of DEP collection as a function of both medium conductivity and frequency.Both dispersions appeared to exhibit conductivity-dependent behaviour, with the "lower frequency" dispersion appearing to exhibit a resonance-type effect around 150mS/m, but reducing in intensity above this, whilst the "higher frequency" dispersion rises to a peak at around the same mark before reducing slowly.

IV. DISCUSSION
Determination of the dielectric properties of dissolved molecules using a DEP model based on Maxwell-Wagner dispersion principles requires a number of assumptions to be made about the behaviour both of the molecule and the motion of charges around it; nevertheless, we can attempt to interpret the spectrum in order to gain some meaningful insight.Our ellipsoidal molecule can be modelled as ellipsoidal solid particle with major axis r 1 and minor axis r 2, wherein the force is given by Where ε m is the suspending medium permittivity, E is the electric field, ∇ is the differential vector operator, ε * p and ε * m are the complex permittivities of the particle and medium respectively (given by ε * = ε − jσ ω where σ represents conductivity, ε permittivity, ω angular frequency and j the square root of -1, and the subscripts m and p refer to the medium and protein respectively.There are a number of possible explanations for the two observed dispersions.The two most likely are the presence of two populations, or the presence of two different dispersion mechanisms active in this frequency region.Analysis of the dielectric properties of serum albumin by Bone et al., 21 has suggested two populations representing the molecule in one of two distinct orientations towards the applied electric field.Alternatively, the dispersion could be due to other, non-Maxwell-Wagner processes.According to Grant and Pethig [22][23][24] the characterization of relaxation time of dielectric dispersion can be calculated as: Hence ε 0 is dielectric constant, β is relaxation of protein molecules and f is frequency respectively.Grant identified a dielectric dispersion in Bovine Serum Albumin and suggested that orientation dispersion occurs at 400 kHz due to the relaxation of protein molecules. 23xamination of figures 2 and 3 suggests that the "lower frequency" dispersion is centred almost exactly on 400 kHz, suggesting that this "population" more likely the orientation dispersion of the protein molecules.The orientation dispersion should not be affected by the conductivity of the medium, but this dispersion does appear to change around 100-150 mS/m; however, since the two dispersions interact, the appearance of a "resonance" effect may in fact be due to the second, "higher frequency" dispersion.This appears to exhibit a rising value of conductivity with increasing medium conductivity, a feature widely noted to be related to increased charge density in the electrical double layer with increasing frequency due to the reduction of the Debye length. 6,16,20he overall effect of each of these polarisation process (orientational and interfacial) is to produce a number of dispersions and in the dielectric permittivity of the DEP system.DEP dispersions can be affected by conductivity when a low frequency dispersion seen in the collection of nanoparticles; 20 in particular, there is a "resonant" conductivity where the conductivity of the particles increases at a certain medium conductivity.This kind of resonant frequency was observed at 100 -150mS/m in the BSA data, indicating that the higher-frequency dispersion can be considered as DEP, and the lowerfrequency dispersion is orientation.The DEP dispersion is therefore, caused by the polarisation of the double layer around the molecules, and the "resonance" occurs because the double layer becomes thinner and contains a higher charge density at higher medium conductivities."Population 1" is instead linked with the orientation of the molecule.
According to de Levine et al., 25 aqueous electrolyte solutions are ionic conductors and are hostile to electrons.Consequently, at the interface between an electrode and an aqueous electrolyte solution, there is mismatch in the type of charge carrier used.In the absence of chemical mechanism to convert one type of charge carrier into the other, the interface behaves as a capacitance. 20Moreover, AC polarization reduces the threshold electric field potential across the gap which in turn reduces the dielectrophoretic trapping.
Modelling the DEP results from dispersion 2 alone, produce the data shown in Table I.As can be see, the conductivity exhibits a rise of almost x2, peaking around 100mS/m before reducing at higher conductivities in line with previously observed behaviour.A rise in particle permittivity was also observed, which is less anticipated, but which may be due to limitations in the model in approximating the shape of the molecule, or in our understanding of the physics of the double-layer behaviour surrounding dissolved molecules in electric fields.Previous studies of molecules using DEP, such as the study of avidin molecules as a function of medium pH, 26 would benefit from this rapid and low-cost method of assessing the dielectric properties of biomolecules.

V. CONCLUSIONS
We have demonstrated that DEP manipulation can be used to determine the dielectric properties of dissolved molecules.The electrical impedance based on the time function could be a rapid label-free approach to determine protein characteristics.As presented in this study, the impedance of the electrode-electrolyte interface and the electrochemical effect (polarization of the electrode) were feasible to distinguish dispersions, which can be attributed to the orientational dispersion of the molecule, plus a Maxwell-Wagner type dispersion governed principally by the countercharge surrounding the molecule.The principle should also be applicable to the measurement of a wide variety of other macromolecules including proteins and DNA.

TABLE I .
Estimated dielectric parameters for BSA derived from modelling.