ANALYST, OCTOBER 1993, VOL. 118 123s Characterization of Planar Concentration Gradients in a Sequential- injection System for Cell-perfusion Studies Cy H. Pollema and Jaromir RBfiCka" Department of Chemistry, University of Washington, Seattle, WA 98195, USA This paper describes the characterization of a perfusion chamber that is coupled with a sequential-injection system and is being designed for live-cell perfusion. The apparatus consists of a multi-port valve, a peristaltic pump, a perfusion chamber and an epifluorescence microscope. The entire system is computer controlled and temperature regulated. The parameters discussed are the concentration-time profiles with regard t o the volume of reagent used and the position of the cell i n the perfusion chamber. Other parameters discussed include the stopped-flow compliance, reproducibility and symmetry of the concentration gradients formed.The system is shown to be suitable for two modes of perfusion; the first in which all cells are exposed to the same concentration of reagent, and the second in which cells are exposed to a gradient of concentrations. All characterization is performed with use of bulk fluorescein as a tracer, and a correlation is made between the bulk flow and the response within the cellular environment by using 5-[N-(octadecanoyl)amino]fluorescein. Keywords: Flow injection; sequential injection; perfusion; live cell; planar concentration gradient Currently available cellular probes permit the measurement of physiologically relevant properties such as specific cellular component staining, measurement of intracellular calcium level, pH, membrane potential and a host of other cellular functions and structures1 by digital-imaging microscopy.The study of cellular changes requires the precise control of several experimental variables, including a probe to elicit the response, a detection system with adequate speed and sensitivity to monitor the change, and a method of changing the cellular environment to which the response under study occurs. The technology of microscopy is advancing rapidly, with higher resolution, low-light level cameras, faster confocal systems, and more computing power for the collection and storage of data. However, the control of the cellular environment often involves manual reagent introduction such as pipetting into the stagnant solution contained in a Petri dish.Automated cell-perfusion techniques have been intro- duced and are outlined in a previous study.2 Briefly, in order to determine the response to a change in the perfusing medium, the time for wash-out must be negligible with respect to the dynamics of the response being measured.3 This places a high priority on fast wash-out with little or no unswept volumes. We are introducing a simple system capable of a wide variety of protocols, with an emphasis on the flow characteristics, and based on flow-injection principles. The sequential-injection analysis (SIA) system [Fig. l(a)] consists of a bidirectional pump, a holding coil (HC), a multiposition valve QvlPV), a transfer line (TL) and a fountain cell (FC).Adherent cells are attached to a circular cover-slip, which is placed in the fountain cell for perfusion and monitoring. The function of this system is to provide the cells with a perfusing buffer at controlled temperature and flow rate (typically, 1 ml min-l). This will provide the ability to expose cells to impulses of a reagent of a defined concentration for a well-defined period. Computer control of the system allows the automated introduction of a sequence of reagents and wash solutions to cells in the chamber in a pre-programmed fashion. In this paper the perfusion system is characterized by exploring the bulk flow behaviour and its relationship to the stimulation of cells adhered within the perfusion chamber. Experimental The SIA system consisted of a Valco ten-position dead-stop selector valve (Valco Instruments, Houston, TX, USA) and * To whom correspondence should be addressed.an Alitea Type C-4V peristaltic pump (Alitea USA, Medina, WA, USA). The holding coil connecting the pump and valve was 0.8 mm i.d. poly(tetrafluoroethy1ene) (PTFE) coiled with a total calculated volume of 800 PI. Larger-diameter tubing was used on the holding coil to help prevent out-gassing during the aspiration of reagents. All other tubing was 0.5 mm i.d. PTFE. The peristaltic tubing, Masterflex (Cole-Parmer, Niles, IL, USA) No. 13, was coated on the outside with silicone oil and removed from the pump, when not in use, to minimize wear. The perfusion chamber used was of fountain cell design,4 and was constructed as previously described, and connected to the SIA systems via a 30 p1 (0.5 mm i.d.) transfer line, which was tightly knotted just before the inlet to randomize the flow.The fountain cell contained a Plexiglas insert with a collection u Pump V ' I a-D R2 Flow direction D w Fig. 1 ( a ) Sequential-injection system consisting of a bidirectional peristaltic pump, a holding coil (HC), multiposition valve (MPV), with rea ents (R) clustered at different port positions, and a transfer line (TLf connecting the valve to the fountain cell (FC). ( b ) and (c) Illustrations of the perfusion chamber showing side and overhead views, respectively; the overhead view includes illustrations of the fields observed (A-D) covering a radius from 2 to 5 mm, respectively, d being 0.5 mm and h being varied from 0.38 mm to 1.14 mm.Dotted area in inset to (b) and central shaded area in ( c ) show region of turbulence (see text for details)1236 ANALYST, OCTOBER 1993, VOL. 118 ring of radius 6.4 mm. The ring was 3.2 mm wide and 3.2 mm deep, giving it a total volume of 514 pl. The outlet from the collection ring was 0.8 mm i.d. PTFE tubing with a short (2 cm) length of 0.3 mm i.d. PTFE tubing placed downstream as a flow restrictor. This served to maintain the pressure in the fountain cell and prevent the formation of bubbles during the pressure drop associated with the radial expansion. The flow path was defined by a 381 pm PTFE spacer (Small Parts, Miami Lakes, FL, USA), yielding a detection region volume of 49 pl. The upper surface of the detection region consisted of a 25 mm round No.1 cover glass, which, in cell studies, would be coated with the cell monolayer. The cover glass was held in place with a threaded ring attached to the outer body of the fountain cell, and was easily exchanged. The detector was a Zeiss Universal microscope (Carl Zeiss, Oberkochen, Germany) equippcd with a Type 111-RS epiflu- oresccnce attachment and 50 W mercury-arc source. The source was shuttered by using a Uniblitz Model T132 controller and shutter (Vincent Associates, Rochester, NY, USA). Fluorescence signals were detected by either a Nikon PI photomultiplier tube (Nikon, Tokyo, Japan) or a COHU 6500 camera (Cohu Inc., San Diego, CA, USA) linked to an Olympus 90 mm macro lens. Photomultiplier tube detection of SIA zones involved use of a 25X planapochromatic objective (normal aperture 0.45), with a detection region of 0.345 x 0.436 mm.The video imaging of zones was accomplished by removing the objective lens and using the zoom lens for focusing. The excitation light was optimized experimentally. While uniform illumination was not obtained, this is normal with an arc source and can be compensated for by flat-field correction when quantitative results are desired. The excita- tion and emission wavelengths were defined by a dichroic mirror, allowing 450-490 nm excitation and 520 nm long-pass emission. Thermostatically controlling the system for cell perfusion work involved two separate areas, one for the perfusion buffer and reagents, and one for the stage and perfusion chamber. The perfusion buffer was kept in a water-bath at 37 "C, with the holding coil and reagents warmed by using the heated recirculating water.The reagents were kept in a smaller water- bath that was heated by using the recirculating water from the bath containing the perfusion buffer. It was found that the smaller bath maintained a temperature of approximately 35 "C, which was adequate, as these small volumes would be brought up to 37 "C when merged with the carrier and sent into the perfusion chamber. The holding coil was wrapped with similar copper tubing to help maintain the temperature of the liquids within the SIA system. The stage and perfusion chamber wcrc thermostatically controlled with use of an air- curtain incubator controlled via an Omega CN76000 pro- portional controller (Omega, Stamford, CT, USA).The temperature probe was an Omega Type 860 RTD placed directly on the cover glass with the controller indicating a temperature of 37 k 0.2 "C during experiments. Control of the SIA system and data collection from the photomultiplier tube were carried out with use of an RTD ADA1100 interface board (Real Time Devices, State College, PA, USA) in a Comtrade (City of Industry, CA, USA) 80486 computer. The data collection and control software was Atlantis (Lakeshore Technology, Chicago, IL, USA). Video imaging involved use of a Quickcapture frame grabber card (Data Translation, Marlboro, MA, USA) in a Macintosh IIci, with the image collection and manipulation controlled by National Institute of Health, USA, image system. Characterization of the SIA system was performed using 1 pg 1-1 fluorescein (Sigma, St.Louis, MO, USA) diluted in 0.01 mol I-' sodium borate buffer. The carrier was also 0.01 mol I-' sodium borate buffer and was typically used at 1 ml min- unless otherwise stated. To ensure the linearity of the response, four different dilutions (1.04.1 pg I-') were tested, using 85 p1 of fluorescein. Over this range, the response was linear. In the experiments carried out the aspirated volume was varied over the range 17-340 p1. For the cell studies, a Krebs-Ringer buffer (KRR), with MEM essential and non- essential amino acids (Gibco BRL), 1% bovine serum albumin, 20 mol I- I N'-(2-hydroxyethyl)piperazine-N-ethane- sulfonic acid and 5 mmol I-' glucose, was perfused at a flow rate of 1 ml min-I.The pH-sensitive probe 5-[N-(octadc- canoyl)amino]fluorescein(C18Fl) (Molecular Probes, Eugene, OR, USA) was used to probe the relationship between bulk flow and cell stimulation. The probe was dissolved in ethanol to form a stock solution of 2 mg ml-I. The stock solution was then diluted in the KRB to a final concentration of 10 pg ml-', and the pH of a portion of the buffer for bulk experiments was adjusted to 4. For cell studies, the cells were incubated for 10 min with the pH 7 KRB containing the probe, then perfused with medium free of the probe. The volume of pH 4 buffer was 85 pl. The cell line used for this work was a rat insulinoma cell (RIN-5AH) . Results and Discussion System Considerations The design of a sequential-injectiori system is based on a set of parameters, which are related to its physical configuration and to the concentration and volume of the material injected into the system.Starting with the configuration, the over-all dispersion in the system is the sum of variances s2, which express the individual contribution of system components S2TOT = S2Mp" + S2Hc + s2TL + S2FC (1) where TOT refers to the total system while the other subscripts are shown in Fig. l(a). Recalling that one of the aims of this design is to allow staining of all cells by the same reagent concentration (no concentration gradient), then .+,(. << s * ~ ( , ~ . Furthermore, in order to minimize the volume of solution to be used, the dispersion of the components of the SIA system also needs to be minimized. Some important considerations include the volume (i.e., length, diamctcr and ultimately s 2 ) of the transfer line and of the holding coil, and the internal volume of the valve. The configuration of the multiposition valve used in this study is such that a common ring is internal to the valve, which has a volume of approximately 10 pl; this term could dominate dispersion of some systems, making the ~ 2 ~ ~ " an important term.However, with a given valve, there are other important parameters. As each injected element moves through a section of the holding coil twice, the tube diameter can be selected in such a way as to dominate the over-all variance ( s * , ~ ~ ) ; in addition, the transfer line to the perfusion chamber also plays a crucial role. As the aim of this study is a system with low dispersion, the i.d.of the holding coil was 0.8 mm, while the i.d. of the transfer line was 0.5 mm and its length was 15 cm. These dimensions are a result of practical considerations. Use of smaller- diameter tubing tends to lead to clogging and pressure problems. The holding coil i.d. is slightly larger to prevent a high vacuum, which would lead to the formation of bubbles. To verify that the system fulfills the requirements of having low dispersion, in which thc contribution of the perfusion chamber over a given region to dispersion can be practically neglected, the depth of the chamber was varied. By using three different channel depths, the over-all dispersion of the system was calculated. This is reasonably well approximated by treating the system as a single stirred-tank models in which the dispersion coefficient D becomes a function of the injected volume as (2) D-1 = c/q, = e-'.f The dispersion coefficient is related to the maximum concentration that passes the point of observation ( c ) and the initial concentration (c,)).The dispersion of a system can also be related to the injected reagent volume, which results in a signal equal to half the signal produced by undiluted reagent ( c = 0 . 5 ~ ~ ~ ) ; this volume will be referred to as the R1,2. ThisANALYST, OCTOBER 1993. VOI,. I18 1237 volume provides a useful relationship with the dispersion of a system as low R1/2 values indicate low dispersion. The value of RIl2 can be established by injecting increasing sample volumeso and plotting the aspirated volume versus [-In (1 - c/cO)].The resulting line yields the R,,? volume when [-In (1 - c/cO)] equals 0.693. Channel depths [h in Fig. l(b)] being varied to 0.38,0.76 and 1.14 mm yielded RlI2 volumes of 33, 34 and 44 PI, respectively. Hence, the contribution of the fountain cell to the over-all dispersion of this system is negligible as large changes in the volume of the fountain cell result in small changes in the Rll2 and, therefore, the over-all system variance. Radial Flow The fountain cell [Fig. l ( h ) and (c)] exploits the symmetry of the radial flow pattern formed by a laminar stream of fluid impinging perpendicularly on a planar surface. The fluid quickly translates into radial flow, with negligible unswept volume in the transition region.It was the large viewing area and fluid flow properties that made a radial perfusion chamber design appealing. However, introduction of a radial-flow perfusion chamber requires a brief discussion of relevant previous work involving use of similar chambers. The import- ant parameters of radial flow between confined boundaries have been identified in studies on air at incompressible speeds.7 These conclusions should also describe the character- istics of non-compressible liquids. The pressure at differeirt radial locations and the flow velocity were described under conditions of either laminar or turbulent flow. The equation for the mean velocity in the radial direction, assuming laminar flow conditions, is <u> = (2 (2nrh)-' ( 3 ) where Q is the channel inlet volume flow, Y is the radial distance from the centre of the inlet, and h is the height or depth of the channel.It was also determined from this study that transition from turbulent to laminar flow occurred at a Rcynold's number of 2000 in radial flow, which is approxi- mately equal to that for flow in two-dimensional channels and circular pipes. The perfusion system being characterized operates in a range of Reynold's numbers from 5.4 at 1 mm from the inlet to 1.1 at 5 mm from the inlet. Radial flow has been used for the measurement of cell adhesion.* A study by Fowler and McKay8 exploited the change in cross-sectional area with increasing radius, which causes the linear flow velocity and hence the surface shear force to decrease linearly along the radius of the chamber.The strength of cell attachment was evaluated by measuring the critical shear radius, which is the distance from the inner edge of attached cells to the central inlet. From this the minimum distraction force, F, is shown to be F = 3 Q pk (n rhZ)-' (4) where Q is the flow rate of liquid emerging from the inlet (I min-I), p is the viscosity (cP), Y is the critical shear radius (mm), h is the distance between the cover-slip and the outlet disc (mm), and k is a constant (16.7), which allows F to be determined in N m-? (Pa).g More recently, radial flow has been used in an extensive study of the cell adhesion and detachment by using a radial-flow detachment appar- atus.h,lO.~' In this work, coated latex beads were used as model cells to evaluate the kinetics of adhesion by measuring the critical shear radius.The effects of ligand and receptor densities and the influence of pH and ionic strength of the medium were addressed. The relevance of these publications to this work is three- fold. First, they show that the flow rate within the fountain cell decreases linearly as the distance travelled from the inlet ( Y ) increases and that the pressure at any radius is inversely proportional to the fourth power of channel depth. Next, it has been demonstrated that cells remained adherent during a 3 h period at a flow rate of 500 ml min-' in a shearing apparatus having a channel depth of 1 mm. This indicates that no cells will be inadvertently detached during a sequential- injection experiment, as it is proposed here to use flow rates 500 times lower.Lastly, although radial flow chambers have been used for shear stress studies for some time, their use for cytochemical studies in conjunction with sequential injection has not been considered. In order to do so effectively, the behaviour of the concentration gradients created by sequen- tial injection in radial flow between confined boundaries needs to be studied. Also, for the effective use of this system in the study of cellular processes, a clear understanding of the flow properties is necessary. This will aid in defining the concentra- tions and contact times for which adhered cells are exposed to a selected reagent. System Characterization The characterization focused on the concentration-time profiles as observed in the fountain cell.However, it should be kept in mind that it is the geometry of the over-all system which determines the flow pattern of the concentration gradient that reaches the inlet of the fountain cell. Within the cell itself, the first important parameter to consider is the 'separation bubble' formed as the flow transitions from thc inlet into the radial chamber. [See Fig. l(6) dotted area of detail and Fig. l(c) shaded area in centre.] This transition region is formed as the flow leaves the inlet, separates from the inlet corner and then re-attaches to the back wall at some radius from the inlet. If the channel is sufficiently deep, the re- attachment will no longer occur, and a radial wall jet will form. This flow pattern causes unswept volumes that will increase the wash-out time of the chamber.I n this study, the presence of an annular separation bubble was qualitatively determined with use of 4.5 pm Fluoresbrite beads (Polysciences, Warr- ington, PA, USA), while the perfusion chamber thickness was varied by inserting additional YTFE spacers. At a depth o f 0.76 mm, which is twice the value proposed for this study, the flow reconnected approximately 0.8 mm from the centre. This region was easily identified as beads held within the separation bubble appeared to How towards the centre, while those outside flowed radially outward. Doubling this depth to 1.52 mm increased the distance for reconnection to approximately 1.6 mm. Increasing further to 2.28 mm, the reconnection was not visible. Therefore, further work was conducted with the 0.38 mm spacer, which confined the bubble to the area inside the end of the PTFE tubing and afforded well swept areas, avoiding the wall-jet effect. Mapping of the surface was also limited to the area outside of a 2 mm radius to avoid this uncharacterizcd region.The perfusion chamber can be operated in two different modes based on the aspirated volume used: (1) all cells are treated the same (uniform reagent contact), or (2) cells are treated differentially based on their position relative to the inlet (gradient reagent contact). Conditions required for each of these catagories were asccrtained by microscopic mapping of the fountain cell area, and measuring dispersion (D) values at four points A, B, C and I1 [Fig. l(c)], which were radially displaced 2 , 3 , 4 and 5 mm from the central inlet.By selecting 25X magnification, the observation area at each point was approximately 0.150 mm2. Uniform Reagent Contact By injecting 340 pl of fluorescein and monitoring the zone passage through the fountain cell at continuous flow, four traces were obtained (Fig. 2)' which show remarkable similarity in peak heights, variances and shapes. As one would expect, the peak maxima shift to longer times as the position of the observation spot moves away from the inlet. Surpris- ingly, however, the peak width does not increase as the zone moves outwards, and as the linear flow velocity is decreasing1238 ANALYST, OCTOBER 1993, VOL. 118 as a function of the radius [eqn. ( 3 ) ] , the zone appears to be focusing in the forward direction.This can be explained by the fact that a fixed volume of reagent forms a ring, which, as it expands outward, must also thin in bandwidth to occupy still the same volume; spreading is small owing to the minimal wall effects in the chamber. The over-all beneficial result is that all cells regardless of their radial distance from the inlet will be perfused by the same reagent concentration (peak height) for the same period of time (peak width), although not at the same time (as the positions of peak maxima are slightly increasing towards longer times). A summary of the mapping experiments is shown in Fig. 3, from which it follows that, by injecting a minimum of 170 pl (approximately 4R1I2), D = 1 is obtained throughout the entire area of the fountain cell.As in classical flow injection, injecting reagent volumes beyond 4R1, is a waste of reagent as it cannot increase the concentration of reagent beyond co. It should be realized that there is no need to reach the maximum reagent concentration (co) to provide well-defined reagent concentrations and contact times. The D value, once established, is a constant of the system for a given flow rate and component configuration. Hence, injection of 170 p1 of reagent (Fig. 3) into the present system is the best choice for homogeneous perfusion of the entire area. However, 85 p1 (Fig. 4) could prove suitable if the reagent is expensive and if the concentration of reagent contacting the cells [62-76% of the originally injected concentration (co)] is deemed sufficient 0 10 20 30 40 50 60 Time/s Fig.2 locations from the centre: A, 2; B, 3; C, 4; and D, 5 mm Concentration-time profiles for 340 pl zones at different radial 1 2 3 4 5 Rad ius/m m Fig. 3 Plot of radius versus dispersion coefficient for various aspiratcd volumes: A. 17; B, 85; C, 170; D, 255; and E, 340 pl This injected volume while it does yield a slight gradient, uses considerably less reagent and provides a contact time of less than 10 s at continous flow. For longer contact times the flow can be stopped at the isoconcentration point (Fig. 4, I), when all cells, regardless of their distance from the inlet, are exposed to the same reagent concentration. Experiments have been run to determine if extended stopped flow is reliable when using this system. An 85 p1 zone of fluorescein was injected into the chamber and the flow was stopped.A location 2 mm from the centre was illuminated in 750 ms impulses every 15 s to minimize photobleaching. The data collected showed no change in signal with stop periods extending up to 30 mins, indicating that extended stopped- flow contact times will allow cell contact with a constant concentration of perfusing solution. A certain minimum volume is always required to obtain these conditions. How- ever, by minimizing all the dispersing components of the system, this volume can remain fairly small. Note that this SIA system then allows for the automated selection of many different reagents, which can be reproducibly introduced and removed from the chamber over a wide range of times. During these times, cells are exposed minimally to a shear force due to the flow within the chamber, which ranges from over 5 to less than 1 x N ~ m - ~ for the given conditions.This shear range may cause shear response in some cell lines and should, therefore, be evaluated for each cell line used. Very short contact times can be obtained by flow reversal, which also allows formation of nearly square-shaped concen- tration impulses. Hence, by injecting the reagent, letting only the leading edge of the dispersed zone enter the fountain cell, and reversing the flow to withdraw the zone quickly, contact times of 100 ms can be obtained. Once withdrawn, the reagent can be aspirated into the holding coil from where it can flow to waste at an auxiliary port, or it can be returned to contact the cells again.The difficulty with the latter approach is that each flow reversal adds dispersion to the zone, and changes its character. The advantages would be in the savings of reagents through its ‘recycling’ and the speed with which the reagent could be applied and withdrawn. Gradient Reagent Contact If concentration gradients are desired, there are two ways to obtain these conditions. One is to work with small volumes that will disperse noticeably across the chamber, and the other is to stop the flow as the leading edge of the reagent zone arrives into the chamber. By decreasing the volume of the injected zone to 85 p1 of fluorescein while under the conditions of the previous experiment, a pattern is observed (Fig. 4) whereby the peak height decreases with the increase of the 0 10 20 30 40 50 60 Ti me/s Fig.4 Concentration-time profiles for 85 p1 zones at different radial locations from the centre: A, 2; B, 3; C, 4; and D, 5 mm. Also indicated is the isoconcentration point (I)1238 ANALYST, OCTOBER 1993, VOL. 118 as a function of the radius [eqn. ( 3 ) ] , the zone appears to be focusing in the forward direction. This can be explained by the fact that a fixed volume of reagent forms a ring, which, as it expands outward, must also thin in bandwidth to occupy still the same volume; spreading is small owing to the minimal wall effects in the chamber. The over-all beneficial result is that all cells regardless of their radial distance from the inlet will be perfused by the same reagent concentration (peak height) for the same period of time (peak width), although not at the same time (as the positions of peak maxima are slightly increasing towards longer times).A summary of the mapping experiments is shown in Fig. 3, from which it follows that, by injecting a minimum of 170 pl (approximately 4R1I2), D = 1 is obtained throughout the entire area of the fountain cell. As in classical flow injection, injecting reagent volumes beyond 4R1, is a waste of reagent as it cannot increase the concentration of reagent beyond co. It should be realized that there is no need to reach the maximum reagent concentration (co) to provide well-defined reagent concentrations and contact times. The D value, once established, is a constant of the system for a given flow rate and component configuration.Hence, injection of 170 p1 of reagent (Fig. 3) into the present system is the best choice for homogeneous perfusion of the entire area. However, 85 p1 (Fig. 4) could prove suitable if the reagent is expensive and if the concentration of reagent contacting the cells [62-76% of the originally injected concentration (co)] is deemed sufficient 0 10 20 30 40 50 60 Time/s Fig. 2 locations from the centre: A, 2; B, 3; C, 4; and D, 5 mm Concentration-time profiles for 340 pl zones at different radial 1 2 3 4 5 Rad ius/m m Fig. 3 Plot of radius versus dispersion coefficient for various aspiratcd volumes: A. 17; B, 85; C, 170; D, 255; and E, 340 pl This injected volume while it does yield a slight gradient, uses considerably less reagent and provides a contact time of less than 10 s at continous flow.For longer contact times the flow can be stopped at the isoconcentration point (Fig. 4, I), when all cells, regardless of their distance from the inlet, are exposed to the same reagent concentration. Experiments have been run to determine if extended stopped flow is reliable when using this system. An 85 p1 zone of fluorescein was injected into the chamber and the flow was stopped. A location 2 mm from the centre was illuminated in 750 ms impulses every 15 s to minimize photobleaching. The data collected showed no change in signal with stop periods extending up to 30 mins, indicating that extended stopped- flow contact times will allow cell contact with a constant concentration of perfusing solution. A certain minimum volume is always required to obtain these conditions.How- ever, by minimizing all the dispersing components of the system, this volume can remain fairly small. Note that this SIA system then allows for the automated selection of many different reagents, which can be reproducibly introduced and removed from the chamber over a wide range of times. During these times, cells are exposed minimally to a shear force due to the flow within the chamber, which ranges from over 5 to less than 1 x N ~ m - ~ for the given conditions. This shear range may cause shear response in some cell lines and should, therefore, be evaluated for each cell line used. Very short contact times can be obtained by flow reversal, which also allows formation of nearly square-shaped concen- tration impulses. Hence, by injecting the reagent, letting only the leading edge of the dispersed zone enter the fountain cell, and reversing the flow to withdraw the zone quickly, contact times of 100 ms can be obtained.Once withdrawn, the reagent can be aspirated into the holding coil from where it can flow to waste at an auxiliary port, or it can be returned to contact the cells again. The difficulty with the latter approach is that each flow reversal adds dispersion to the zone, and changes its character. The advantages would be in the savings of reagents through its ‘recycling’ and the speed with which the reagent could be applied and withdrawn. Gradient Reagent Contact If concentration gradients are desired, there are two ways to obtain these conditions.One is to work with small volumes that will disperse noticeably across the chamber, and the other is to stop the flow as the leading edge of the reagent zone arrives into the chamber. By decreasing the volume of the injected zone to 85 p1 of fluorescein while under the conditions of the previous experiment, a pattern is observed (Fig. 4) whereby the peak height decreases with the increase of the 0 10 20 30 40 50 60 Ti me/s Fig. 4 Concentration-time profiles for 85 p1 zones at different radial locations from the centre: A, 2; B, 3; C, 4; and D, 5 mm. Also indicated is the isoconcentration point (I)I240 bulk flow and cell response at the leading edge of the injected zone over the range covered, Liz., 2 [Fig. 6(a)], 1 [Fig. 6(h)] and 0.5 ml min-1 [Fig.6(c)J. The oscillations seen in the figures are due to the pulsatile flow generated by the peristaltic pumping. While these figures do indicate that the behaviour of the bulk flow at the leading edge is almost identical with that contacting the cells, this could not have been assumed apriori. The trailing edge of the peaks differs more significantly; however, we believe this is due mainly to photobleaching. The difference in photobleaching is seen because the bulk signal results from new fluorophores constantly streaming by, while on the cells there is no renewal possible; hence, photobleach- ing would be seen more in the cell experiment. This idea is supported in Fig. 6 ( 4 , which shows the results from a similar experiment that has been carried out on C18F1-loaded cells with the light source on continuously (solid line) or pulsed ( x).Both traces are from the same group of cells exposed to repeated impulses; with the pulsed source the signal nearly returns to the baseline. Further work is underway in our laboratory to understand better the interaction of the fluid flow environment of the cell when exposed to impulses of a stimulant. We believe that a clear idea o f how a reagent contacts and leaves a cell could provide much more infor- mation about cellular responses than is currently available. It allows the study of both the cell stimulation and recovery in a well-controlled environment. ‘The authors thank Ake Lernmark for his many helpful discussions and for providing the materials needed for the cell study. We also thank Gary Christian and Kurt Scudder for all ANALYST, OCTOBER 1093. VOL. 1 I8 their assistance. The financial support of NIH (SSS-3(5) ROl GM45260-2) is also greatly appreciated. 1 2 3 4 5 6 7 8 9 I 0 11 12 13 References Haugland, R. P., in Molcwdur Probes Hunclhook of Fluoromwt Probes und Rtwurch Clzemiculs, 5th edn., Molecular Probes Inc., Eugene, OR, 1992. Scudder, K. M., Christian. G. D., and RGiiEka, J . , Exp. Cell Kes., 1993, 205, 107. Berg, H. C., and Block, S. M . , 1. Gcw. Microbiol., 1984, 130, 291s. Scudder, K. M., Pollema, C. H., and RfiiiEka, J . , Anal. Chern., 1992, 64, 2657. KGiiEka, J . and Hanscn, E. H., in Flow Injeclion Arzulysis, Wiley-Interscience, New York. 2nd edn., 1988. Cozens-Roberts, C., Quinn, J . A., and 12auffcnburger, D. A., Biopliys. J., 1990, 58, 857. Moller, P. S . , Arvvnuuf. Q., 1963, 14, 163. Fowler. H. W., and McKay, A. J., in Microhiul Adhesion t o Surfuces, Ellis Horwood, Chichcster, 1980, pp. 143-161. Groves. B. J., and Riley, P. A., Cytohios, 1987, 52, 49. Cozens-Roberts, C., Quinn, J . A., and Lauffcnburger, D. A . , Riophys. J., I9Y0, 58, 107. Cozens-Roberts, C., Lauffenburger, D. A., and Quinn, J . A , , Biophys. J., 1990, 58, 841. Foley, M.. MacGregor, A. N., Kusel, J . R., Garland, P. B . , and Downie, T., J . Cell Biol., 1986, 103, 807. McKay, D . A.. Kuscl, J . K., and Wilkinson, P. C., J . CeIISci., 1991,100,473. Paper 3100800B Received February 9, I993 Accepted April 20, 1993