Anal. Chem. 1995, 67, 1486-1490
Configurations of a Flow Injection System for Perfusion Studies of Adherent Cells Pamela J. Baxter, Lars Hallgren,t Cy H. Pollema, Michael Tmka, and Jaromir Ruzicka" Department of Chemistry, University of WashingtonJSeaitleJWashington 98195
This paper describes three configurations of a flow injection apparatus designed to be the fluidic drive for perfusion studies of cultured adherent cells. The apparatus was coupled to a flow-throughperfusion chamber that was specificallydesigned for live cell perfusion using fluorescence microscopy as the detector. The instrument consists of two linear syringe pumps and a multiposition selector valve which, under computer control, allowed sequential injection, fluid switching, and flow injection to be performedwithminimal system recodiguration. When the apparatus is coupled to a dual inlet perfusion chamber, target cells can be exposed to very steep reagent pulses, while the traditional single inlet perfusion chamber allows more flexibility and provides a more gradual increase in reagent concentration. The most significant salient feature of the system is the ability to generate very steep pulses-a desirable feature for cell perfusion studies. When performing cytochemical studies by fluorescence microscopy, it is often desirable to quickly and transiently expose a population of cells held stationary on a coverslip or in a culture chamber to a stimulant under consideration.lI2 Removal of the stimulus would allow the cells to return to basal levels, possibly allowing the same cells to be either restimulated or stimulated by multiple reagents. This is not possible when the cells are stimulated by pipetting a bolus into a culture chamber, since the stimulant cannot be removed rapidly or reproducibly. Although a wide range of detectors and sensors have been integrated with flow injection systems: pulses with subsecond pulses and rise times have not been reported. This work is aimed at fulfilling this need by designing and characterizing a versatile system capable of generating a wide variety of concentration gradients from an injected reagent zone. Additionally, it is believed that exploration of several instrument configurations is of a general interest for other areas of instrumental analysis and chemical sensor technology. The existing fluidic systems have their operations limited to either flow injection (FI) or sequential injection (SI) mode. The present work explores different component configurations in order to design an instrument with the ability to perform both flow and sequential injection and to allow the delivery of a wide variety of pulses with emphasis on steep and narrow concentration gradi+ Present address: Novo Nordisk A/S, Novo Ale, Bagsvaerd, Copenhagen, Denmark. (1) Ruzicka, J.; Lindberg, W. Anal. Chem. 1992.64.537A-545A (2) Scudder, K.M.; Pollema, C. H.; Ruzicka, J. Anal. Chem. 1992,64,26572660. (3) Ruzicka, J.; Hansen, E. H. Flow Injection Analysis, 2nd ed.; WileyInterscience: New York, 1988.
1486 Analytical Chemisfry, Vol. 67, No. 8, April 15, 1995
U
A
U
B
CONFLUENCE
C
Carrier
I
Figure 1. Three instrument configurations characterized in this study. A six-port selection valve, two high-precision piston pumps (each equipped with a two-way valve), a holding coil, the fountain cell, and the fluorescence microscope. (A) Sequential injection mode. (B) Fluid switching mode; inset, FS mode with two-lumen fountain cell. (C) Flow injection mode.
ents. The instrument Figure 1) includes two linear stepper motor driven syringe pumps (syringe volumes from 50 pL to 25 mL available) and a six-port selector valve. Each pump has an auxiliary three-position valve for fluid aspirating and dispensing. The three modes of operation and two perfusion chambers are discussed here (Figures 1and 2). The first mode, a SI mode, uses one linear syringe pump and the multiport selection valve Figure M). The second mode is a fluid switching (FS) mode which uses two linear syringe pumps whose lines were teed at the confluence point in front of the transfer line (Figure 1B). The fluid switching mode was also configured without the transfer line by using a dual inlet fountain cell Figure lB, inset, and Figure 2). The last mode is the FI mode, which uses both linear syringe pumps and the multiport selection valve (Figure 1C). Although 0 1995 American Chemical Society 0003-2700/95/0367-1486$9.00/0
B
C
Figure 2. Two-lumen fountain cell. (a) Top view of the chamber: SS, stainless steel holder; D, drain; V, viewing area (Plexiglas or Teflon); 11, 12, two-lumen inlets; Z, injected sample zone. The small circles indicatethe data collectionposition. Position 1 was 2 mm from the inlet, and position 2 was 5 mm from the inlet. The respective chamber volumes were 5 and 30 pL. The optical flat is secured by a threaded ring that screws into the stainless steel holder. (b) Diagram of the viewing area as seen if two different colored solutions flow through each lumen simultaneously at the same flow rate. The sharp boundry is due to the laminar flow in the radial direction. (c) Diagram of the viewing area as seen if two different colored solutions flow sequentially through each lumen at the same or differentflow rates.
this instrument does not look like a conventional flow injection system, it can be used as such, since it provides all the functions necessary: a well-defined zone is injected into a flowing stream of carrier, which is merged with a second stream of a reagent at a confluence tee, where it proceeds to a dete~tor.~ EXPERIMENTAL SECTION
Perfusion Chamber. The fountain cell was developed in our laboratory to provide an optical geometry suitable for brightfield or fluorescence microscopy and the flow geometry necessary for perfusion of cultured adherent cells.2 It has been extensively chara~terized~*~-~ and applied to study of live cultured cell^.^^^ The fountain cell differs considerably from all perfusion chambers described to date8-13in that it has a central inlet from which the (4) Ruzicka, J. Anal. Chim. Acta 1992,261,3-10. (5) Pollema, C. H.; Lemmark, A.; Ruzicka, J. Cjtometry 1995,19,70-76. (6) Scudder, K.M.;Ruzicka, J.; Christian, G. D. Erp. Cell Res. 1993,205,197204. (7)Pollema, C. H.; Ruzicka, J. Analyst 1993,118,1235-1240. (8) Ince, C.; Beekman, R E.; Verschragen, G.J. Immunol. Methods 1990,128, 227-234. (9) Pentz, S.; Horler, H. J. Microsc. 1992, 167,97-103. (10) Walcerz, D.B.;Diller, K. RJ. Microsc. 1991, 161,297-311. (11) Irion, G.; Ochsenfeld, L.; Naujok, A; Zimmermann, H. W. Histochemistry 1993,99,75-83. (12) Salih, V.; Greenwald, S. E.; Chong, C. F.; Berry, C. L. Int. J. E@. Puthol. 1992, 73,625-632.
delivered fluid emits radially toward the fountain cell's circumference through a thin gap formed by placing a 0.015in. Teflon spacer between the fountain cell body and a round coverslip on which the observed cells have been grown2J4(Figures 1and 2). The total volume of the fountain cell was 43 pL. Hardware. A six-port selector valve @LSmart Valve, Cavro Scientific Instruments, Sunnyvale, CA) and two high-resolution O O 0 modular digital (24000 steps) linear syringe pumps W pump, Cavro Scientific Instruments) were equipped with 1.0-mL syringes. well-defined, precise volumes (1.0 pL to 1.0 mL) of one or several reagents can be aspirated and delivered into the detector with flow rates from 100 pL min-l to 50 mL min-l. The minimum flow rate is determined by the syringe size, since the maximum stroke time is 20 min. These components are powered by a 24V switching power supply @@-Key, Thief River Falls, MN) and controlled serially by a software package written in our laboratory using Visual Basic Professional (Microsoft,Redmond, WA). The detector was a Zeiss Axiovert 100 inverted microscope (Carl Zeiss, Inc., San Leandro, CA) equipped with a mechanical stage on which the perfusion chamber was mounted. Excitation light was provided by a dc stabilized halogen source (built inhouse containing a 12-V, 50-W tungsten halogen lamp powered by a regulated 12-Vpower supply). The microscopewas equipped with a lox objective (Zeiss Fluar 10/0.50) resulting in a 0.7-mmdiameter field of view. The excitation filter was a 450-490-nm band-pass filter, and the emission filter was a 520-nm long-pass filter. Single channel fluorescence signals were measured with a Nikon P1 microscope photometry system (current-measuring photomultiplier tube, Nikon, Tokyo, Japan). Additionally, the zones were observed using a CCD camera (Cohu 4990 series, San Diego, CA). Chemicals. The carrier solution used was a 0.01 M sodium borate 0. T. Baker, Phillipsburg, NJ) buffer. The dye solution was 1 ppm fluorescein (Sigma Chemical Co., St. Louis, MO) prepared in the carrier solution. All solutions were prepared in deionized and distilled water. RESULTS AND DISCUSSION
The key parameter defining the steepness of the concentration gradient generated by the instrument in any of the three possible configurations is the SI12 value. This value is the injected dye volume necessary in order to reach a signal that is half of the steady-state response, resulting in a dispersion factor of 2.3 Thus smaller SI12 values result in zones with a squarer leading edge. To measure the S1/2 for the SI configuration (Figure lA),all lines were first filled with the carrier solution, followed by a reverse movement of the syringe to aspirate the dye zone into the holding coil. By forward movement of the syringe, the dye zone was propelled to and through the fountain cell. Trpical response curves were obtained for dye injections ranging from 100 to 10 pL while monitoring fluorescence at a position 2 mm from the center of the fountain cell (position 1on figure 2). Previous work has shown that monitoring of any spot within the circumference of the fountain cell is representative of all spots of the same size within the cell area? The maximum signal was measured by a 250-pL dye zone. This experiment was repeated for the FS and FI modes in order to determine the S1/2 value for each of these (13) Kirchoff, F.;Ohlemeyer, C.; Kettenmann, H. PfEiigersArch. 1992,420,573577. (14) Ruzicka, J.; Pollema, C. H.; Scudder, K. M. Anal. Chem. 1993,65,35663570.
Analytical Chemistry, Vol. 67, No. 8, April 15, 1995
1487
2.50
2.00
T
-3 4: ”
v
4
1.w
1 0
I
25
5
7.5
10
12.5
15
17.5
20
225
Time (sec) 0.50
5
0
10 15 Time (secl
20
2?
Flgure 4. Rise times to reach steady state by. injecting . - a 250-,u 1 ppm fluorescein zone at different carrier flow rates. Respective .,-e times (to reach 95% maximum signal) for each flow rate: flow rate 0.1 mL min-l, 18.55 s;0.5 mL min-l, 5.29 s; 1 mL min-l, 1.61 s; 5 mL min-’, 0.87 s; and 10 mL min-l, 0.44 s.
0.K
10
20
30 volume (pL)
40
50
60
Figure 3. Determination of 5112 for the three different modes. A,,,, maximum signal obtained from an undiluted sample plug; A,,, maximum signal obtained from the sample injected. The S V for ~ the multilumen cell was determined using the FS mode. (0)SI mode (5112 = 22 pL); (0)FS mode (5112 = 25 pL); (0)FI mode (SW = 22 pL); (A) two-lumen cell connected to the fluid switching mode. Inset: typical response curves used to calculate the 5112 values. From minimum to maximum response: 5, 10, 20, 30,40,50, and lOOpL.
Table 1. Summary of Rate Usedd.
flow rate (mLmin-l) 0.1 0.5 1.0 5.0 10.0
t112
and t0.95 Values for Each Flow t l / Z (s)
9.20 1.54 0.29 0.14 0.08
t0.95
(s)
18.55 5.29 1.61 0.87 0.44
Sample size, 250 pL.
instrument configurations (Figure 3), keeping the volume of the transfer line, VI (45 pL), the same. These experiments showed that the value depends entirely on the volume of VI and was independent of the instrument configuration used. This is true provided the volume of the transfer line is much larger than the volume of the detection chamber and internal volume of the valve and that the same tube diameter is being used. Indeed, the SI/Z values in the present work are very close (22 pL, SI; 25 pL, Fs; and 22 pL, FI), confuming that for systems with identical transfer lines and comparable flow rates, S1/2 need only be determined for one of the configurations. Fluid Switching Mode. This mode (Figure 1B) offers a capability to generate very steep impulses of a reagent for live cell stimulation. The disadvantage of this system is the necessity to fill one of the syringes with the reagent. This is a significant drawback if the reagent is expensive or if several reagents need to be used, as each would require a separate syringe. While the dispersion of the leading edge of the generated step is a function of the transfer line volume, the m i n i u m time necessary for the transition from one solution to another is a combination of the S1/2 value of the system and the flow rate applied (Figure 4). Fast flow rates (e.g., 10 mL min-9 result in a sudden exposure of the cells to the reagent, while moderate flow rates (1.0 mL min-I) yield a slower exposure, and the slowest flow rate (0.1 mL min-I) gives a gradual exposure. The parameter describing the rate of exposure is therefore the tl/z value, which is directly related to the flow rate of the carrier stream. Here, the tl/z values varied from 80 ms (flow rate of 10 mL min-l) to 9.2 s (flow rate of 0.1 mL min-I). Table 1 summarizes the t1/2 and t0.95 values for each flow rate. The total time during which the cells are exposed to the reagent is then defined by the injected zone volume and the 1488 Analytical Chemistry, Vol. 67, No. 8, April 15, 1995
use of a stopped flow period if extended exposure times are needed. In must be pointed out, however, that the use of fast flow rates is wasteful of reagent and that it also may expose cells to unwanted shear stress. Therefore, minimizing the SI/Zvalue is the only viable alternative if steep concentration gradients are desirable. Dual Inlet Fountain Cell. It follows from the preceding discussion that it is necessary to eliminate VI if the steepest possible impulse is to be realized. A fountain cell with multiple central inlets (Figure 2) allows the limes to be controlled independently, thereby eliminating the volume of the common transfer line, VI. The two separate lines allow the chamber to be perfused in two different modes: the two streams can be pumped (1) simultaneously or (2) in sequence. When fluids are pumped through both inlets simultaneously and at equal flow rates, a distinctive flow pattern is observed (Figure 2B): the volume of the fountain cell splits into two equal areas separated by a sharp boundary. If the flow rates from the two inlets are different, the &&ion is equal to the fraction of the total flow represented. This separation is due to the laminar flow radiating toward the chambers circumference and confirms that flow in the chamber does in fact follow a radial path.15 When fluid switching is performed in sequence such that a well-defined volume of dye is injected through one inlet followed by the carrier stream delivered through the second inlet, the same zone pattern is observed as in the single inlet perfusion chamber (Figure 2C), provided all of the zone has been delivered into the chamber before flow resumes (15) Scudder K. Ph.D. Thesis. University of Washington, Seattle, WA, 1994.
401 ,
1
0
I
2
I
4
6
8
0
I
10
4
t
8
0
4
Figure 5. Three fluorescein zones (5 pL) injected into the multilumen perfusion chamber at a rate of 5.0 mL min-l. The signal was collected at position 2 (5 mm from the inlet) (see Figure 2).
from the other inlet. Using this mode, a very steep pulse can be generated by injection of a small zone (5 pL) of dye followed by perfusion through the fountain cell with carrier flowing through the second inlet at a flow rate of 5 mL min-' (Figure 5). The data were collected 5 mm (position 2 on Figure 2) from the center of the chamber, where the volume of the chamber was large enough to accommodate the 5 p L zone. The following describes the salient features of the response. Before reaching point a, the dye is entering the perfusion chamber until the leading edge enters the detector's field of view. At point a, the reagent zone has been delivered to the perfusion chamber and the flow stopped. The first increase in fluorescence (from 0 to the flat baseline at a) is due to fluorescence from the leading edge of the dye zone emerging from the inlet. The elapsed time between points a and b is the time taken to switch syringe valves so that the dye zone can be propelled past the detector with the carrier stream. At point b, the flow is resumed (at a rate of 5 mL min-l), and the dye zone is pushed across the monitored spot and into the waste. The time between points b and c is the time during which the adherent cell would be exposed to the zone of stimulant. Point d represents the time at which all stimulant has been washed out of the observed area. The maximum exposure time a target cell would experience at this observation spot and in this regime is based on the peak width at the base line. Ideally, this would be a time difference (b-c) of 113 ms. However, the mechanics of the valving system do not allow instantaneous valve switching, which results in a stopped flow period, increasing the total interval (a-d) to more than 2 s. In spite of this artifact, three stimulant impulses can be propelled through the perfusion chamber within 10 s, while the reproducibility of the peak heights (RSD = 6.1% for 12 injections) is satisfactory for biological studies. Since the SIIZof the dual inlet perfusion chamber can be safely assumed to be zero, in theory, this mode gives the ability to expose live cells to the steepest possible impulses. The analysis of the plot used to determine the S1/2 shows that the tank-in-series model does not apply (Figure 3), which, of course, is the direct result of the elimination of VI. However, the inability of hardware to switch the streams with no delay caused an unexpected problem (Figure 6). Here, steep impulses were obtained by injecting increasing dye volumes (3,5,10, and 20 pL) and monitoring from the 2 mm position. As already observed, the square shape of the response
8
12
16
0
Time (sec)
Time (sec)
Figure 6. Responsesfrom the FS mode in the multilumen perfusion chamber. The signal was collected at position 1 (2 mm from the inlet). Increasing volumes of the 1 ppm fluorescein solution were injected into the multilumen cell and pushed through by the carrier at 1.O mL min-I. From highest to lowest fluorescence intensity: 20, 10, 5, and 3 pL. The square shape of the curve was caused by stopped flow, which occurred when the valve on the piston pump changed positions. The explanation for the slight downward slope on these square waves is not fluid movement or dye dilution but rather photobleaching of the fluorescein. For details, see the text.
curve is caused by the inability of the hardware to switch fast enough from dye stream to carrier stream, causing a 2-s stopped flow period. (The slight downward slope on the steady-state portion of the peak was found to be due to photobleaching of the fluorescein.) To conclude, subsecond exposure times could only be obtained under the following conditions: (1) the dual inlet perfusion chamber is used and (2) only a small zone of dye (less than the volume of the perfusion chamber at the point where it is being monitored) is being injected, such that its leading edge will not yet reach the monitored spot before switching to the carrier syringe. Larger injected zone volumes will more likely penetrate the monitored spot as it is pushed to the perfusion chamber and toward the circumference. This is because, as the zone volume increases (Figures 5 and 6), part of the zone will inevitably reach the monitored spot, yielding exposure times which are increased due to inability of the present hardware to switch faster between the two streams. Flow Injection Mode. Since the present system utilizes two fluid drivers, it can also be configured with a confluence point at which a reagent is supplied along the entire length of a sample zone (Figure 1C). Two independent drivers would also allow the acceptor and donor stream to be moved independently, as required for gas diffusion and dialy~is.~J~ CONCLUSIONS
Flow system configurations explored in this work were primarily designed for cell perfusion studies with the goal of producing short exposure times, It was found that the subsecond pulses could be generated using the dual inlet perfusion chamber only when injecting small volumes of dye. In the single inlet perfusion chamber operated in SI or FS mode, such a problem does not occur since the injected zone is accommodated within the transfer line. Therefore,the volume of the transfer line should (16) Fang, Z. Flow Injection Separation and Preconcentration; VCH. New York, 1993; Chapters 5 and 6.
Analytical Chemistry, Vol. 67, No. 8, April 15, 1995
1489
be a compromise between the volume of the stimulant to be injected and the S1/2 value desired. It is well established3 that for an injected volume of one Sl/z, the peak width at the baseline (96%clearance) equals five t l l z intervals. Thus, Table 1indicates that at a flow rate of 5 mL min-', the cell exposure time will be 0.7 s, while at a flow rate of 1mL min-l, the exposure time will be 1.5 s. Since the sequential injection mode requires only one syringe while allowing the flexibility of injecting several different reagents, its combination with a single inlet perfusion chamber is recommended. Finally, it should be mentioned that there is no upper limit on the pulse duration (by using stopped flow), and less steep concentration gradients can be created with ease. Thus, in flow injection microscopy, a practical limit for shortest cell contact time (the t 0 . 9 ~ )to a reagent or stimulant is approximately 500 ms, a value shorter than has been previously rep~rted.*,~JlJ~ For cell contact times down to the 5@msrange, a rapid mixer for flow injection c y t ~ m e t r yhas ~ ~been developed. Flow injection is a general solution handling technique which has found wide application in many areas of instrumental analysis.
As such, its potential is being explored in increasingly more extreme conditions; in this case its ability to create very steep and short pulses was examined. As always, the ultimate performance of any instrument is a combination of theoretical limits and hardware limitations. It is believed that the experience gained here will be of value to other uses of flow injection in instrumental analysis. ACKNOWLEDQMENT We would like to express our gratitude to Prof. Gary D. Christian for his continuing interest in this project. Zymogenetics has also supported our work by providing us with the microscope. This work was supported by the NIH (Grant SSS3 (5) RO1 GM 452W05). Received for review September 12, 1994. Februaty 6, 1995.@
Accepted
AC940906L
(17) Scampavia, L;Blankenstein, G.; Ruzicka, J.; Christian, G. Anal. Chem.,
submitted.
1490 Analytical Chemistry, Vol. 67,
@Abstractpublished in Advance ACS Abstracts, March 15, 1995.
No. 8, April 15, 1995
