Manipulation of herpes simplex virus type 1 by dielectrophoresis

Manipulation of herpes simplex virus type 1 by dielectrophoresis

Biochimica et Biophysica Acta 1425 (1998) 119^126 Manipulation of herpes simplex virus type 1 by dielectrophoresis Michael P. Hughes a , Hywel Morgan...

610KB Sizes 4 Downloads 14 Views

Biochimica et Biophysica Acta 1425 (1998) 119^126

Manipulation of herpes simplex virus type 1 by dielectrophoresis Michael P. Hughes a , Hywel Morgan a; *, Frazer J. Rixon b , Julian P.H. Burt c , Ronald Pethig c a

Bioelectronics Research Centre, Department of Electronics and Electrical Engineering, University of Glasgow, Glasgow G12 8QQ, UK b MRC Virology Unit, University of Glasgow, Glasgow G12 8QQ, UK c Institute of Molecular and Bimolecular Electronics, University of Wales, Bangor, Dean Street, Bangor, Gwynedd LL57 1UT, UK Received 2 March 1998; revised 13 May 1998; accepted 19 May 1998

Abstract The frequency-dependent dielectrophoretic behaviour of an enveloped mammalian virus, herpes simplex virus type 1 is described. It is demonstrated that over the range 10 kHz^20 MHz, these viral particles, when suspended in an aqueous medium of conductivity 5 mS m31 , can be manipulated by both positive and negative dielectrophoresis using microfabricated electrode arrays. The observed transition from positive to negative dielectrophoresis at frequencies around 4.5 MHz is in qualitative agreement with a simple model of the virus as a conducting particle surrounded by an insulating membrane. ß 1998 Elsevier Science B.V. All rights reserved. Keywords: AC electrokinetics ; Dielectric property; Nanotechnology; Particle manipulation

1. Introduction

Clausius-Mossotti factor, de¢ned as:

Dielectrophoresis (DEP) is the motion of particles caused by dielectric polarisation e¡ects in non-uniform electric ¢elds [1^3]. Unlike electrophoresis, where the force acting on a particle is determined by its net charge, the DEP force depends on the volume and dielectric properties of the particle. For a spherical particle of radius r the DEP force is given by:

f CM ˆ

F DEP ˆ 2Zr3 Am Re‰f CM Š9E 2


where Am is the absolute permittivity of the suspending medium, E is the local (rms) electric ¢eld, 9 is the del vector operator and Re[fCM ] is the real part of the

* Corresponding author. Fax: +44 (141) 330 4907; E-mail: [email protected]

Ap 3Am Ap ‡ 2Am



In Eq. 2 Ap and Am are the complex permittivities of the particle and medium respectively, where A ˆ A3…jc=g† and A is the permittivity, c the conductivity, g the frequency of the applied pangular  ¢eld, and j ˆ 31. Depending on the relative magnitudes of Ap and Am , the DEP force acting on a particle can cause it to move either towards or away from high ¢eld regions at electrode edges. These two e¡ects are termed positive and negative DEP respectively. DEP has been used for the separation, characterisation and investigation of bioparticles on the micron scale, such as cells [4,5] and bacteria [6]. However, until recently the constraints of photoli-

0304-4165 / 98 / $ ^ see front matter ß 1998 Elsevier Science B.V. All rights reserved. PII: S 0 3 0 4 - 4 1 6 5 ( 9 8 ) 0 0 0 5 8 - 0

BBAGEN 24653 2-9-98


M.P. Hughes et al. / Biochimica et Biophysica Acta 1425 (1998) 119^126

thography imposed lower limits on the size of the electrodes and thus on the magnitudes of the ¢eld E and gradient 9E that could be generated using applied voltages of modest amplitude. This in turn imposed a lower limit on the size of particles that could be manipulated. Recent advances in microfabrication technology have now made it possible to construct electrodes with features small enough to generate electric ¢elds greater than 107 V m31 across gaps of 2 Wm or less. This has allowed researchers to dielectrophoretically manipulate sub-micron size latex spheres, viruses and proteins [7^12]. In this paper we present evidence for the frequency-dependent dielectrophoretic behaviour of herpes simplex virus type 1 (HSV-1) in microfabricated electrode arrays. Studies by other researchers have demonstrated positive dielectrophoretic collection at electrode edges of tobacco mosaic virus [12,13] and the negative dielectrophoretic behaviour of Sendai and in£uenza virus [7,8]. In this work we demonstrate that when HSV-1 viruses are suspended in a mannitol solution of conductivity 5 mS m31 they exhibit positive DEP below 4.5 MHz and negative DEP at higher frequencies. Furthermore, by determining the frequency dependence of the rate of collection of £uorescently labelled virus particles in a DEP trap, new information has been determined regarding the dielectric characteristics of the HSV-1 virus. HSV-1 is a human pathogen best known for caus-

ing cold sores although infection can have more serious consequences [14]. A transmission electron micrograph of a HSV-1 particle is shown in Fig. 1a. The complete particle, or virion, is approximately 200^250 nm in diameter. A diagrammatic representation of the virus particle is shown in Fig. 1b. The viral envelope is a lipid bilayer which contains a number of di¡erent glycoproteins. These extend outwards and form a layer approximately 10^20 nm thick around the envelope [15]. The envelope encloses a thick amorphous protein gel called the tegument, which in turn surrounds the capsid [16]. The capsid is a robust icosahedral protein structure which contains the viral DNA. HSV-1 is one of the largest mammalian viruses, and in electrical terms can be thought of as resembling a scaled down version of a mammalian cell, with an insulating membrane and a conducting core. 2. Materials and methods 2.1. Virus preparation HSV-1 virions were puri¢ed using protocols established by Szilagyi and Cunningham [17]. Owing to the infectious nature of HSV-1 all experimental procedures were performed in a class II containment laboratory. The virus particles were pelleted from tissue culture medium (TCM) by centrifugation at

Fig. 1. a: A transmission electron micrograph of a herpes simplex virus type 1 (HSV-1) particle. b: Schematic diagram of the HSV-1 virion showing the DNA, capsid, tegument and membrane.

BBAGEN 24653 2-9-98

M.P. Hughes et al. / Biochimica et Biophysica Acta 1425 (1998) 119^126

23 000Ug for 2 h at 4³C. They were resuspended in TCM and centrifuged through a 35 ml gradient of 5^ 15% w/w Ficoll 400 in TCM at 12 000 rpm in a Sorvall AH629 rotor. The virion band was collected, diluted in TCM, pelleted at 20 000 rpm in AH629 tubes, resuspended in an appropriate volume of TCM and frozen at 370³C for future use. The viruses were £uorescently labelled with NBDdihexadecylamine (Molecular Probes Inc). The dye was dissolved in DMF at a concentration of 2 mg/ ml and then added to the virus solution (in TCM) at a 1:100 dilution. Labelling with this dye had no effect on the infectivity of the virions. Particles were incubated for 20 min at room temperature, pelleted at 20 000 rpm for 10 min in a Sorvall TLA 100.2 rotor and resuspended in iso-osmotic (280 mM) mannitol solution. For observations of the dielectrophoretic spectra, viruses were resuspended in a 280 mM mannitol solution with the conductivity adjusted to 5 mS m31 by the addition of a small amount of EDTA. 2.2. Microelectrodes Electrode arrays based on the polynomial design [18] were used and these are shown in Fig. 2. Electrodes were fabricated with a gap of 2 Wm between

Fig. 2. A photograph of the electrode microstructures used in these experiments. The area of the electrode corresponding to the electric ¢eld simulation of Fig. 8 is shown in the box. Scale bar: 20 Wm. The electrode is 20 Wm from centre to edge and has a gap in the centre of 6 Wm.


nearest neighbour electrodes and 6 Wm gap across the centre. The arrays were made using photolithography. A microlitre volume chamber was constructed around the electrode using perspex spacers and adhesives. The chamber had a depth of approximately 50 Wm and a sample volume of the order of 1 Wl. Virus solution was pipetted into the chamber and the assembly sealed with a cover slip. 2.3. Experimental The electrodes were powered using a HewlettPackard signal generator providing 10 V peak to peak (pk-pk) sinusoidal signals over the frequency range 1 kHz^20 MHz. Potentials were applied to give a 180³ phase di¡erence between adjacent electrodes. Experiments were observed using a Nikon Microphot microscope, a Photonic Science Isis II image-intensifying camera and Sony S-VHS video recorder. Conductivity measurements were performed using a Hewlett-Packard 4192A-impedance analyser and a Sentek conductivity cell in the range 100 kHz to 1 MHz. 3. Results 3.1. Observation of DEP collection Fig. 3 shows three £uorescence micrographs of HSV-1 particles attracted to the high ¢eld regions at the electrode edges under positive DEP. These images were taken at 30-s intervals after energising the electrodes with 5 V pk-pk (equivalent to a ¢eld strength of the order of 106 V m31 ), 500 kHz, signals. Individual virions can be seen in the image, and these were observed to continuously collect at the electrode edges for as long as the ¢eld was applied. Positive DEP collection was observed at frequencies below 4 MHz (down to the lowest frequency used of 10 kHz). At frequencies above approximately 5 MHz the virions collected in the centre of the electrode array under negative DEP as shown in Fig. 4. This ¢gure depicts three photographs taken at 30-s intervals after application of 5 V, 10 MHz, signal. At this voltage, frequency and medium conductivity of 5 mS m31 , the virions were observed to be levitated at least 7 Wm above the electrodes in a column approxi-

BBAGEN 24653 2-9-98


M.P. Hughes et al. / Biochimica et Biophysica Acta 1425 (1998) 119^126

Fig. 3. A sequence of three £uorescence photographs showing the time course of herpes simplex virus type 1 particles collecting in the arms of a polynomial electrode under positive dielectrophoresis. The photographs were taken 30 (a), 60 (b) and 90 (c) s after application of a 500 kHz, 5 V peak to peak sinusoidal signal. Images were taken with an image-intensifying camera (Photonic Science ISIS II) and recorded on S-VHS video.

mately 2 Wm diameter and 50 Wm high. The column of virions grew in size over a few minutes until it stabilised at these dimensions, then virions were observed to drift out from the top of the column as new ones arrived at the bottom. The diameter of the column remained stable once ¢lled, as can be seen from the time-sequences of Fig. 4. Throughout the experiments, Brownian motion of particles was observed, with a typical particle displacement of two to three particle diameters. However, in our experimental observations, the e¡ects of Brownian motion did not appear to in£uence the dielectrophoretic manipulation of the particles.

Interesting hydrodynamic e¡ects were also observed. For example, at the voltage used in these experiments and for frequencies below about 150 kHz convection forces were found to interfere with the DEP forces in a similar manner to that reported by Mu«ller et al. [9]. Fluid £ow was observed to occur, driving particles in from the electrode edges into and along the electrode arms. In particular for 5 V peak to peak and at frequencies of approximately 30 kHz and below, virions were observed to collect at the centre of the electrode array in a manner similar to that resulting from negative DEP. However, instead of being restricted in size by the 2 Wm

Fig. 4. A sequence of £uorescence photographs showing the time course of HSV-1 particles experiencing negative dielectrophoresis and collecting in the centre of the electrode. The photographs were taken at 30-s intervals (a, b, c) after application of a 10 MHz, 5 V peak to peak sinusoidal signal. Aggregation of particles is achieved rapidly and is limited in size by the pro¢le of the potential energy well.

BBAGEN 24653 2-9-98

M.P. Hughes et al. / Biochimica et Biophysica Acta 1425 (1998) 119^126


Fig. 5. A sequence of £uorescence photographs showing the time course of HSV-1 particles collecting in the centre of the electrode 30, 60 and 90 s (a, b, c) after application of a 100 kHz, 10 V peak to peak sinusoidal signal. A large aggregate is visible at the centre of the trap, which can attain a radius in excess of 15 Wm.

diameter ¢eld funnel formed by the electrodes, the spherical aggregation of virions grew to a radius of 20 Wm or more, as illustrated in Fig. 5. This aggregation e¡ect was found to increase signi¢cantly as the applied voltage was increased to 10 V. At this voltage, virion aggregations were observed to form for frequencies up to 120 kHz. Collection of virions at the electrode edges by positive DEP was also maintained during this e¡ect. In addition to the large central accumulation of virions, small aggregations were also occasionally observed at distinct points along the inter-electrode arms, approximately 2/3 of the distance between the centre and edge of the arrays. At these points the aggregates on the electrodes were also levitated, and driven away from electrode edges by streaming currents. These aggregations are illustrated in Fig. 6. This e¡ect was also observed to occur at the much higher frequencies of 4^5 MHz, where the DEP force reversed polarity, from positive to negative.

measuring the optical absorbance of yeast solutions in AC ¢elds [19]. Measurements were performed by video-capturing images of the electrode array at ¢ve 30-s intervals after the application of 5 V pk-pk signal at a given frequency. The £uorescence excitation light was blanked o¡ between measurements to avoid photobleaching. The captured images were converted into bitmaps and an analysis was performed on the rate of particle collection, based on the light intensity and total frame area illuminated (i.e. containing a £uorescing virus). The overall rate of particle collection was determined by comparing successive frames. The mean normalised particle collection rate in the

3.2. Determination of collection rate In order to determine the relative magnitude of the DEP force on the virions as a function of frequency, the rate of their collection at the region of high ¢eld gradient was determined. By determining the rate at which particles collect, an approximation may be made to the relative force experienced by the particles. A similar method has been used previously to determine the DEP behaviour of yeast cells by

Fig. 6. A £uorescence photograph showing the aggregation of particles both at the centre of the trap, and at stable points above the inter-electrode regions, 2/3 of the distance between the centre and edge of the trap. The particles were collected using a 5 V peak to peak, 30 kHz applied sinusoidal signal.

BBAGEN 24653 2-9-98


Fig. 7. A graph of electrode trap as a was determined by the electrode array

M.P. Hughes et al. / Biochimica et Biophysica Acta 1425 (1998) 119^126

the rate of collection of virus particles in the function of applied ¢eld frequency. The rate measuring the increase in £uorescence across at 30-s intervals after applying the ¢eld.

range 100 kHz^10 MHz is shown in Fig. 7. Measurements at frequencies below 100 kHz were discarded due to interference by £uid convection e¡ects.

then: G spmem ˆ G tmem ‡ 2K s =a2 where a is the particle radius. The conductivity of an intact biological membrane is usually very low ( 6 1038 S m31 ) whilst the surface conductance is usually of the order of 1 nS [21]. Therefore, for particles of the diameter of HSV (250 nm) the surface conductance dominates, i.e. Gtmem W2K s =a2 . As shown in Fig. 8, a good ¢t to the experimental data can be obtained if the surface conductance is assigned a value of 3.5 nS ( þ 0.5 nS); and the viral interior (tegument) a conductivity of 8 mS m31 ( þ 2 mS m31 ) and a relative permittivity of 70 ( þ 5). This ¢t was obtained using a medium permittivity, Am = 78. The ¢t was found to be completely insensitive to membrane permittivity. These parameters predict a DEP spectrum that matches the experimental data, with a crossover from positive DEP to negative DEP occurring around 4^5 MHz. The interior of the virus consists of a `protein gel' so that the estimated value for the internal permittivity could re£ect the fact that this is primarily composed of water.

4. Discussion Fuhr and co-workers [7,8] have shown that viruses, when suspended in a medium of conductivity 74 mS m31 , can collect as aggregates in ¢eld cages using negative DEP. This e¡ect was observed for frequencies up to 4 MHz, and was found to be frequency-dependent. In this work we have shown that an enveloped virus (HSV-1) when suspended in a medium of conductivity 5 mS m31 can, depending on the frequency, exhibit both positive and negative DEP. The frequency-dependent DEP behaviour of particles can be characterised in terms of the so-called multishell model [5]. This model provides a theoretical prediction of how the Clausius-Mossotti factor (see Eqs. 1 and 2) varies with frequency. We have modelled the HSV-1 virus as a conducting particle surrounded by an insulating membrane. It has been demonstrated [20] that the conductivity of the membrane surrounding a cell can be represented as the sum of two components; a trans-membrane conductance, Gtmem and surface conductance component Ks . Introducing an area-speci¢c conductance [20,21]

Fig. 8. A Clausius-Mossotti plot, Eq. 2, showing the variation in the sign and magnitude of the force on a virion as a function of frequency. Values used in the simulation were: membrane conductivity 6 1038 Sm31 , surface conductance, Ks = 3.5 nS, membrane permittivity, Am = 10, tegument conductivity, ci = 5 mS m31 , tegument permittivity, Ai = 70. The medium conductivity used for this plot is cmed = 5 mS m31 and the medium permittivity, Am = 78 (the permittivity of 280 mM mannitol solution is 0.7 units less than that of pure water, but such a degree of accuracy is unwarranted in this model). This ¢gure indicates that the model is consistent with the experimental data above a frequency of 250 kHz, but does not predict the increase in the magnitude of the positive DEP force observed at low frequencies in the collection rate data.

BBAGEN 24653 2-9-98

M.P. Hughes et al. / Biochimica et Biophysica Acta 1425 (1998) 119^126

Analysis of the collection rate of virus particles in the high ¢eld regions of the electrodes indicates that the particles exhibit two dispersions ^ a low frequency dispersion at 200 kHz and a higher frequency dispersion at 2 MHz leading to the DEP crossover at 4.5 MHz. For frequencies above approximately 700 kHz, the rate measurement is in agreement with the simple interfacial polarisation model with the crossover frequency at 4^5 MHz. Below these frequencies it appears that the particle experiences an increase in the collection rate, i.e. a positive DEP force which increases with decreasing frequency, rather than a constant DEP force as predicted by the ClausiusMossotti plot of Fig. 8. It is probable that the low frequency dispersion seen in the collection rate (at 200 kHz) is associated with the relaxation of the di¡use part of the double-layer surrounding the particle. This polarisation would lead to an increase in the low frequency polarisability of the particle and thus an increase in the positive DEP force. Similar DEP e¡ects, also considered to be associated with the dynamics of the electrical double-layer, have been reported for microorganisms [22], cells [23] and latex beads [24]. In addition to DEP forces, the movement of £uid causes additional forces on the particles; forces which act with or against the DEP force. Owing to the scale of the electrode designs used here, electric ¢eld strengths greater than 106 V m31 can be generated. In certain frequency and conductivity windows the high electric ¢eld strengths give rise to local temperature gradients, which in turn cause discontinuities in the conductivity and permittivity of the liquid. These changes give rise to coulomb and/or dielectric forces on the liquid [25,26] which can induce complex patterns of £uid £ow. At low frequencies the motion of the £uid is governed by free charge in the medium and depends only on 9c and E2 . Therefore liquid moves from regions of high temperature (high ¢elds at the electrode edge) to regions of low temperature over the electrodes. In these circumstances it is the overall balance of forces (including DEP and gravity) that determines the particle behaviour. A 3-dimensional electric ¢eld plot for the polynomial electrode array is shown in Fig. 9. This plot was obtained using the Moments method, [27] with software running on a SPARC 20 WorkStation under


Fortran 77, and data analysis was performed using Matlab (The Mathworks, Inc). The ¢gure is a crosssection of the 3-D ¢eld solution in a plane 10 Wm above the electrodes, a point corresponding to the approximate levitation height of the small aggregates seen in the arms of the electrodes. The ¢gure shows how the electric ¢eld varies in magnitude across the whole of the electrode region in the x-y plane. It can be seen that there are four maxima in ¢eld strength (at this height) at discrete locations in each of the four inter-electrode regions at a distance of approx-

Fig. 9. A plot of the spatial variation of the normalised RMS electric ¢eld in a plane 10 Wm above the electrode surfaces, in the region de¢ned by the box in Fig. 2 is shown in A. The electric ¢eld is at a maximum in between the electrode arms, approximately 13 Wm from the centre of the electrode array. Also shown (B) is the variation in the DEP force across the same plane. The vectors point to the maximum in electric ¢eld as seen in A.

BBAGEN 24653 2-9-98


M.P. Hughes et al. / Biochimica et Biophysica Acta 1425 (1998) 119^126

imately 13 mm from the centre (7 mm in from the edge). Also shown in this ¢gure (Fig. 9B) is the variation in the magnitude of 9E 2 in this plane. Although the variation of 9E 2 in the z-direction is far greater than that which occurs in the in the x-y plane, these vectors show that under conditions giving positive DEP, the gradient of the DEP force is such as to move the particles inwards towards discrete locations as observed experimentally. Particles move down the gradient of DEP force to this point where both 9E 2 and the electric ¢eld is at a maximum and where a net balance of convective and dielectrophoretic forces exists. 5. Conclusion The dielectrophoretic behaviour of the mammalian virus HSV-1 in microelectrode structures has been investigated over the frequency range 10 kHz to 20 MHz. With particles suspended in an iso-osmotic EDTA/mannitol medium of conductivity 5 mS m31 both positive and negative dielectrophoretic collection was observed. The transition between these two types of behaviour at a frequency in the range of 4^5 MHz is compatible with a simple physical model of the virus. The increasing positive dielectrophoretic collection, which is observed at low frequencies, is considered to arise from the polarisation of the di¡use double-layer surrounding the virus. In addition to the DEP force an electrohydrodynamic force causes small aggregates of particles to form and levitate at points of electric ¢eld maxima. Acknowledgements The authors would like to thank Mrs. Joyce Mitchell and Ms. Mary Robertson for virus preparation, Mr. William Monaghan for electrode fabrication and Mr. Nicolas Green for valuable discussions. This work is supported by the Biotechnology and Biological Sciences Research Council (UK) grant no. 17/T05315.

References [1] H.A. Pohl, Dielectrophoresis, Cambridge University Press, Cambridge, 1978. [2] T.B. Jones, Electromechanics of Particles, Cambridge University Press, Cambridge, MA, 1995. [3] R. Pethig, Crit. Rev. Biotechnol. 16 (1996) 331^348. [4] P.R.C. Gascoyne, R. Pethig, J.P.H. Burt, F.F. Becker, Biochim. Biophys. Acta 1149 (1992) 119^126. [5] Y. Huang, R. Holzel, R. Pethig, X.-B. Wang, Phys. Med. Biol. 37 (1992) 1499^1517. [6] G.H. Markx, Y. Huang, X.-F. Zhou, R. Pethig, Microbiology 140 (1994) 585^591. [7] T. Schnelle, T. Mu«ller, S. Fiedler, S.G. Shirley, K. Ludwig, A. Hermann, G. Fuhr, B. Wagner, U. Zimmerman, Naturwissenschaften 83 (1996) 172^176. [8] T. Mu«ller, S. Fiedler, T. Schnelle, K. Ludwig, H. Jung, G. Fuhr, Biotechnol. Tech. 10 (1996) 221^226. [9] T. Mu«ller, A. Gerardino, T. Schnelle, S.G. Shirley, F. Bordoni, G. De Gasperis, R. Leoni, G. Fuhr, J. Phys. D: Appl. Phys. 29 (1996) 340^349. [10] M. Washizu, S. Suzuki, O. Kurosawa, T. Nishizaka, T. Shinohara, IEEE Trans. Ind. Appl. 30 (1994) 835^843. [11] N.G. Green, H. Morgan, J. Phys. D: Appl. Phys. 30 (1997) 2626^2633. [12] H. Morgan, N.G. Green, J. Electrostatics 42 (1997) 279^293. [13] N.G. Green, H. Morgan, J.J. Milner, J. Biochem. Biophys. Methods 35 (1997) 89^102. [14] R.J. Whitley, in: B.N. Fields et al. (Eds.), Fields' Virology, Lippincott-Raven, Philadelphia, 1996, pp. 2297^2342. [15] L.M. Stannard, A.O. Fuller, P.G. Spear, J. Gen. Virol. 68 (1987) 715^725. [16] F.J. Rixon, Sem. Virol. 4 (1993) 135^144. [17] J.F. Szilagyi, C. Cunningham, J. Gen. Virol. 72 (1991) 661^ 668. [18] Y. Huang, R. Pethig, J. Meas. Sci. Technol. 2 (1991) 1142^ 1146. [19] M.S. Talary, R. Pethig, IEE Proc. Sci. Meas. Technol. 141 (1994) 395^399. [20] W.M. Arnold, U. Zimmerman, J. Electrostatics 21 (1988) 151^191. [21] P.R.C. Gascoyne, R. Pethig, J. Satayavivad, F.F. Becker, M. Ruchirawat, Biochim. Biophys. Acta 1323 (1997) 240^252. [22] T. Inoue, R. Pethig, T.A.K. Al-Ameen, J.P.H. Burt, J.A.R. Price, J. Electrostatics 21 (1988) 215^223. [23] J.P.H. Burt, R. Pethig, P.R.C. Gascoyne, F.F. Becker, Biochim. Biophys. Acta 1034 (1990) 93^101. [24] X.-F. Zhou, G.H. Markx, R. Pethig, I.M. Eastwood, Biochim. Biophys. Acta 1245 (1995) 85^93. [25] A. Ramos, H. Morgan, N.G. Green, A. Castellanos, J. Phys. D: Appl. Phys. (1998) in press. [26] J.R. Melcher, G.I. Taylor, Annual Review of Fluid Mechanics, Palo Alto, CA, 1969. [27] M.P. Hughes, X.-B. Wang, F.F. Becker, P.R.C. Gascoyne, R. Pethig, J. Phys. D: Appl. Phys. 27 (1994) 1564^1570.

BBAGEN 24653 2-9-98