Dielectrophoretic separation of Bacillus subtilis spores from environmental diesel particles

Abstract

Isolation of pathogenic bacteria from non-biological material of similar size is a vital sample preparation step in the identification of such organisms, particularly in the context of detecting bio-terrorist attacks. However, many detection methods are impeded by particulate contamination from the environment such as those from engine exhausts. In this paper we use dielectrophoresis—the induced motion of particles in non-uniform fields—to successfully remove over 99% of diesel particulates acquired from environmental samples, whilst letting bacterial spores of B. subtilis pass through the chamber largely unimpeded. We believe that such a device has tremendous potential as a precursor to a range of detection methods, improving the signal-to-noise ratio and ultimately improving detection rates.

---

1. Introduction

With increased threats from biological weapons against military and civilian targets, there is an urgent need for systems which can rapidly detect such harmful agents. The Centre for Disease and Control and Prevention has categorised potential pathogens based on their risk to national security,¹ and one biological agent that poses a significant risk is that of Bacillus anthracis, which is the cause of anthrax and was used against civilian targets in 2001. A surrogate bacteria for B. anthracis used in biological warfare/defence research for many years is Bacillus globigii (now more commonly known as Bacillus subtilis var niger). These particles simulate the behaviour of B. anthracis, because they are able to resist desiccation, heat and chemical treatment by forming tough spores. There have been no reports indicating significant diseases caused directly by B. globigii and, although B. subtilis has been found in hospital patients susceptible to disease, there is no clear link between this bacterium and disease.

Harmful biological agents may be intentionally released in to the environment as aerosols and can remain airborne for very long distances. There are difficulties associated with the direct detection of particles in air, hence samples collected are typically re-suspended in liquid prior to detection. Since many environmental particles are also trapped in such samples, prior to detection there still remains the issue of selectively removing unwanted particles collected in the sample.

Dielectrophoretic manipulation of bacteria was first described by Pohl in the 1970s² and has since been used for detection and separation of bacteria³ and the determination of bacterial properties after various chemical and heat treatment procedures.^4–6 It has also been used in the separation of heterogeneous particle mixtures and concentration of homogeneous particles in micro-fluidic systems.^7–11 The technique offers significant advantages for the concentration and separation of harmful bio-agents from other particles in solution.^12–14 In this paper we describe the use of dielectrophoresis for isolating and concentrating bacterial spores in a solution containing diesel particulate matter (DPM) which is taken from an airborne environmental sample. Optimisation of process parameters for the selective removal of bacterial spores is investigated for use in a dielectrophoretic micro-fluidic system.

2. Experimental

2.1 Determination of spores and DPM sample cross-over frequencies

Fabrication of pin-type electrode. Electrode geometries of a pin-type design (opposing triangles with 100 μm between opposing tips) were used for characterisation.^4,15 Electrodes were fabricated on standard microscope slides coated with 10 nm Ti/100 nm Au using standard photolithographic procedures.

The pin-type electrode slide was glued using fast curing epoxy on to a normal microscope slide, so it fitted to a standard microscope stage. Using silver epoxy, wires were secured on the electrode terminals of the pin electrodes and connected to a signal generator via crocodile clips on the end of a coaxial cable. A 4 mm diameter circle was cut out of a parafilm sheet. The sheet was then placed over the electrode slide, with the circle revealing the tips of the pin electrodes. The parafilm sheet, still on top of the electrode slide, was cured in the oven at 70 °C for 30 seconds making a bond with the surface of the slide. Final thickness of the parafilm (now the spacer) was found to be 150 μm. Samples were pippetted on to the electrodes and covered with a coverslip.

Cross-over determination. Variants of B. globigii (B. subtilis ATCC 9372; B. globigii; B. globigii lysophilized and pasteurised) obtained from Smiths Detection Ltd (Watford, UK) were each micro-centrifuged for 8 minutes at 1400 rpm, supernatant discarded and spores re-suspended in ultra-pure water (×3). A 1 ml DPM sample (obtained from a car engine exhaust via a wetted wall cyclone capture) was centrifuged for 20 minutes at 1400 rpm, solution decanted and solids re-suspended in ultra-pure water. This was performed three times to ensure all solids were collected and that the medium was at the required conductivity. Concentrations of re-suspended samples were subsequently measured using a haemocytometer. Aliquots of spores and DPM sample were then re-suspended in different medium conductivities (0.3 mS m⁻¹, 1.1 mS m⁻¹, 3.3 mS m⁻¹, 7.8 mS m⁻¹, 15.3 mS m⁻¹, 38 mS m⁻¹ and 56 mS m⁻¹). Using the pin type electrodes, cross-over frequencies of the spores and DPM were determined by observing particle motion at each decade over 5 decades (1 kHz to 10 MHz). On determining which two adjacent decades show positive and negative dielectrophoresis, manual tuning of the signal generator was performed to determine the frequency where no movement was exhibited in order to determine the cross-over frequency.¹⁶

2.3 Micro-fluidic separation of Bg spores from DPM

Micro-fluidic device. Interdigitated microelectrodes were used to trap particles from flow. The electrode array had an electrode width/spacing ratio of 500 μm : 75 μm along the axis of symmetry, and were angled at 45° to the flow. The width dimension was chosen to ensure consistent voltage across the electrodes, while the inter-electrode spacing was in line with the empirical guide of approximately 30 times particle size. Electrodes were fabricated on standard microscope slides coated with 10 nm Ti/100 nm Au using standard photolithographic procedures.

The micro-fluidic device consisted of two parts, a top part through which electrical connections are made and a bottom part where the slide (22 × 36 × 1 mm) patterned with the electrode array was placed. Both parts were mechanically joined together by six screws. The channel was sealed using a Nitrile O-ring cord (1.4 mm diameter). The micro-fluidic chamber has one inlet and one outlet (1 mm diameter bore) positioned centrally at either end of the chamber. The length between the inlet centre and outlet centre was 22.5 mm with a chamber length of 27 mm. The chamber height and width were 0.25 mm and 6 mm, respectively, giving a chamber volume of approximately 54 mm³. Horizontal channels leading to the inlet and outlet bores were 3.25 mm above the chamber floor. The channels had an inner diameter of 1 mm. Voltages were supplied to micro-electrodes via gold-plated spring-loaded pins (2.3 mm radius flat head, medium spring obtained from Coda Systems Ltd, Essex, UK) positioned around the device.

Four outer ports were created on the micro-fluidic device, two on the inlet side and two on the outlet side. Both inlet and outlet ports were designed so that they converged at a point just before entering/exiting the fluid chamber. The design resembled a ‘V’-shaped arrangement owing to the angles of the channels where they met. This design was chosen to allow medium feed to be pumped into the flow cell to remove air bubbles prior to introduction of the particle solutions via the secondary inlet port through a syringe pump, resulting in a continuous laminar flow profile throughout the fluid chamber. The outlet ports allowed collection of particulates into separate receptacles based on temporal separation/concentration strategies. Outlet or inlet ports not in use were plugged, giving a simple in and out flow-through system.

Particle separation. Using a 5 ml tapered-end syringe placed in a syringe pump (Razel), 2 ml volumes of the mixture were pumped through the flow-cell at volumetric flow-rates of 0.51, 1.02, 2.08, 3.05 and 4.08 ml h⁻¹ through inlet port 1. Tubing from the pump was filled with feed solution up to inlet port 1. To eliminate air bubble formation in the fluid chamber, buffer solution of the same conductivity was manually introduced into the flow-cell via inlet port 2 before introduction of the feed mixture into the fluid chamber. After removing the air bubbles from the system, the signal generator and syringe pump were started simultaneously.

B. globigii spores (pasteurised + lysophilized) were suspended at a concentration of 7.9 × 10⁸ spores ml⁻¹ in 30 ml ultra-pure water. The diesel exhaust sample was centrifuged for 30 min at 3000 rpm and re-suspended in 7.5 mS m⁻¹ KCl solution (×3). Final concentration of the diesel exhaust sample was found to be 1.7 × 10⁸ particles ml⁻¹. A diesel/spore mixture was made up with a v/v ratio of 7 : 1 giving a spore concentration of 9.9 × 10⁷ spores ml⁻¹ and diesel exhaust particle concentration of 2.1 × 10⁷ particles ml⁻¹. Final solution conductivity was made up to 7.5 mS m⁻¹. The micro-device efficiency was calculated by input and output concentration values obtained through manual counting in the haemocytometer. The total concentration passed through the DEP chamber was compared with the input concentration, based on the operating parameters of the DEP chamber. For collection of desired particles in the chamber, the fraction of particles exiting the system with regards to the input would be very low in the case of a high value of separation efficiency. For the flow-through of desired particles in the DEP chamber, the system is evaluated based on the number of undesirable particles not trapped in the system. The system was manually flushed through with 5 ml ethanol (70%) then distilled water three times to remove any debris which may have still been present before re-use. A schematic of the separation device is shown in Fig. 1.

Fig. 1 Representation of DEP separation strategy; white circles represent diesel particulate matter and collect at the electrode edge and black circles represent spores which experience negative DEP and are carried out of the system with fluid flow.

3. Results and discussion

3.1 Cross-over frequencies of spores and DPM

The transition from positive to negative DEP or vice versa is known as the cross-over frequency. The cross-over frequency was investigated as a function of medium conductivity for the spores and diesel particles to see whether any differences between untreated and pasteurised and lysophilized B. globigii spores were present, and to see whether distinct differences could be found for spores and DPM as a possible criterion for particle separation. Fig. 2 shows the cross-over frequency of each spore variant as a function of medium conductivity. At frequencies below the cross-over frequencies, the spores experience positive dielectrophoresis, while above the cross-over frequency curve the spores are observed to experience negative dielectrophoresis.

Fig. 2 Cross-over frequency of variant spores against medium conductivity. Line markers indicated are (+) B. globigii (cultured), (◊) B. subtilis (ATCC 9372) and (○) B. globigii (pasteurised and lysophilized).

At a medium conductivity of 0.3 mS m⁻¹ (pure water) there was a marked variation in the cross-over frequency for all 3 spore treatments. The untreated B. globigii spores were found to have a cross-over frequency at 5 MHz, whilst the pasteurised and lysophilized spores had a cross-over frequency at 83 kHz suggesting the treatment process may have altered the outer surface of the spore. The cross-over frequency for B. subtilis (ATCC 9372) in the same medium conductivity was found to occur at 1.4 MHz, signifying a difference between the properties of the untreated strain of B. globigii and the ATCC 9372 strain. As the medium conductivity increased, the cross-over frequency for untreated B. globigii showed an initial decrease at 1.1 mS m⁻¹ to 383 kHz and a final value of 330 kHz for 56 mS m⁻¹. In contrast, pasteurised and lysophilized B. globigii showed an initial rise in cross-over frequency at 1.1 mS m⁻¹ to 152 kHz, with a continual rise to 418 kHz at 56 mS m⁻¹. B. subtilis showed an initial decrease in cross-over frequency at 1.1 mS m⁻¹ to 470 kHz then began to rise to 1.4 MHz at a medium conductivity of 56 mS m⁻¹. Using the same medium conductivities, DPM were characterised in the same way. Positive dielectrophoresis of the particles were seen at all medium conductivities with a maximum frequency of 10 MHz applied. This result indicated that above the cross-over frequency for the bacterial spores, separation of spores and DPM is possible.

3.2 Continuous flow DEP separation

Since the diesel particles were observed to experience positive dielectrophoresis across the conductivity media investigated, it was decided that the most effective continuous-flow process would be to operate the system such that all spores experience negative dielectrophoresis and be repelled away from the electrode surface. At a medium conductivity of <10 mS m⁻¹, frequencies in the megahertz band could be used without causing electrode damage and with minimal attenuation of voltage amplitude across the fluidic chamber. From the cross-over frequency data of Fig. 2, at a medium conductivity of 7.6 mS m⁻¹ all spore variants were seen to experience negative dielectrophoresis at 600 kHz or greater, hence a frequency of 1 MHz 10 V_p-p was chosen to separate the spores from the diesel particulate matter in flow conditions by enriching the spores downstream of the dielectrophoretic fluidic chamber.

Table 1 shows the separation efficiencies of the process for the flow-rates used. It was found that retention of DPM in the dielectrophoretic chamber was as high as 99% for a volumetric flow-rate of 1.02 ml h⁻¹, which decreased as the volumetric flow-rate increased. As the flow-rate increased to 4.08 ml h⁻¹ the maximum fraction of spores exiting the system was found to be 66% of the input concentration, with approximately 1.7% of DPM also present in the effluent stream. In order to account for losses in the fluidic system due to particle adhesion in the tubes and potential dead spaces, spores were flowed through the DEP chamber with no field applied. It was found that the ratio of spores downstream of the DEP chamber was 48% for flow rates of 0.51 ml h⁻¹ but consistently higher at approximately 71% for all flow rates above this level. Empirical studies (data not shown) indicate the losses are largely in the tubing.

Table 1 Separation efficiencies of micro-system downstream of DEP chamber

Volumetric flow-rate/ml h^−1	DPM in effluent stream (%)	Spores in effluent stream (%)	Separation efficiency (%)
0.51	0.75	9.7	99.25
1.02	0.83	55.7	99.17
2.08	1.40	49.6	98.60
3.05	1.14	65.8	98.86
4.08	1.30	58.7	98.70

---

Particle separation at smaller volumetric flow-rates saw a decrease in spore fractions exiting the system, particularly at 0.51 ml h⁻¹ where only 10% of the spore input concentration was found downstream of the separation chamber. Particle sedimentation effects are likely to be the cause of this low fraction of spore enrichment downstream at slower flow-rates. Also, as particles flow over the electrode array, spores may become trapped in field minima electric fields between adjacent electrodes which could also inhibit fluid streams carrying the spores out of the DEP chamber. It can be seen from Fig. 3 that as the chamber flow-rate increases, DPM retained by positive DEP accumulate at the electrode edges against the direction of fluid flow. At a mean fluid velocity in the chamber of 94.1 μm s⁻¹, corresponding to a volumetric flow rate of 0.51 ml h⁻¹, DPM are also seen collecting on the surface of the electrode arrays (Fig. 3a). This effect becomes less apparent as the mean chamber velocity increases. Based on the dimensions of the device the calculated residence time of particles flowing in the inlet tubing (115 mm in length) prior to entering the separation chamber for a volumetric flow rate of 0.51 ml h⁻¹ was 51 min. This is approximately 10 times greater than the residence time within the separation chamber. In comparison, the residence time for the particles in the separation chamber in a volumetric flow-rate of 4.08 ml h⁻¹ was found to be 36 s. This suggests that as the particle’s residence time in the system is reduced, for both affluent and effluent streams and DEP chamber, the input/output ratio of spores passing through the system should increase. A critical flow-rate for this system may exist where, above this flow-rate, the fraction of DPM in the effluent stream becomes too high, indicating that fluid forces are stronger than the electric field forces and dielectrophoretic retention of DPM is no longer effective.

Fig. 3 Diesel particles retained by DEP on sections of electrode array at the end of fluid flow experiments for chamber flow-rates of (a) 0.51 ml h⁻¹; (b) 1.02 ml h⁻¹ and (c) 3.05 ml h⁻¹.

4. Conclusion

The differences in bacterial spores and DPM, based on their dielectrophoretic cross-over frequencies at particular medium conductivities, were used as the basis for a dielectrophoretic separation strategy. It was observed that over a range of frequencies and medium conductivities, DPM experienced positive dielectrophoresis only, while at the higher end of the frequency spectrum the spores were seen to experience negative dielectrophoresis. By applying appropriate processing parameters, a separation strategy has been demonstrated that removes diesel particles from spores in a dielectrophoretic separation chamber, as a precursor for bio-sensing applications. The process of spore enrichment downstream has been shown to be influenced by flow-rate, as the effluent stream still possesses contaminants of the enhanced stream. The efficiency of the process can be further increased by: (i) optimising the design of the micro-fluidic device for efficient flow-through across the array at all flow-rates; (ii) cascading devices to achieve a cumulative separation and flow-through efficiency greater than a single device at faster flow-rates; and (iii) by increasing the magnitude of the dielectrophoretic force, making the chamber penetration more effective for DPM retention at fast flow-rates.

References

1. P. F. Pilch and R. A. Zilinskas, Encyclopedia of Bioterrorism Defense, Wiley, New York, 2005.
2. H. A. Pohl, Dielectrophoresis, Cambridge University Press, Cambridge, UK, 1978.
3. W. B. Betts, The potential of dielectrophoresis for the real-time detection of microorganisms in foods, Trends Food Sci. Tech., 1995, 6, 51–58.
4. J. Johari, Y. Hübner, J. C. Hull, J. W. Dale and M. P. Hughes, Dielectrophoretic assay of bacterial resistance to antibiotics, Phys. Med. Biol., 2003, 48, N193–N198.
5. M. P. Hughes and H. Morgan, Measurement of bacterial flagellar thrust by negative dielectrophoresis, Biotechnol. Prog., 1999, 15(2), 245–249.
6. G. H. Markx, Y. Huang, X. F. Zhou and R. Pethig, Dielectrophoretic characterisation and separation of microorganism, Microbiology, 1994, 140, 585–591.
7. G. H. Markx, P. A. Dyda and R. Pethig, Dielectrophoretic separation of bacteria using a conductivity gradient, J. Biotechnol., 1996, 51, 175–180.
8. Y. Li and K. V. I. S. Kaler, Dielectrophoretic fluidic cell fractionation system, Anal. Chim. Acta, 2004, 507(1), 151–161.
9. M. Durr, J. Kentsch, T. Schnelle and M. Stelzle, Microdevices for manipulation and accumulation of micro- and nanoparticles by dielectrophoresis, Electrophoresis, 2003, 24(4), 722–731.
10. H. B. Li, Y. N. Zheng, D. Akin and R. Bashir, Characterization and modeling of a microfluidic dielectrophoresis filter for biological species, J. Microelectromech. Syst., 2005, 14, 103–112.
11. J. Auerswald and H. F. Knapp, Quantitative assessment of dielectrophoresis as a microfluidic retention and separation technique for beads and human blood erythrocytes, Microelectron. Eng., 2003, 67–68, 879–886.
12. M. P. Hughes, Strategies for dielectrophoretic separation in laboratory-on-a-chip systems, Electrophoresis, 2002, 23(16), 2569–2582.
13. P. R. C. Gascoyne and J. Vykoukal, Particle separation by dielectrophoresis, Electrophoresis, 2002, 23(13), 1973–1983.
14. F. F. Becker, X. B. Wang, Y. Huang, R. Pethig, J. Vykoukal and P. R. C. Gascoyne, Separation of human breast cancer cells from blood by differential dielectric affinity, Proc. Natl. Acad. Sci. U. S. A., 1995, 92, 860–864.
15. F. H. Labeed, H. M. Coley, H. Thomas and M. P. Hughes, Assessment of multidrug resistance reversal using dielectrophoresis and flow cytometry, Biophys. J., 2003, 85(3), 2028–2034.
16. P. R. C. Gascoyne, J. Noshari, F. F. Becker and R. Pethig, Use of dielectrophoretic collection spectra for characterizing differences between normal and cancerous cells, IEEE Trans. Ind. Appl., 1994, 30(4), 829–834.

---