Flow Injection Renewable Surface Immunoassay - American Chemical

---

Anal. Chem. 1994,66, 1825-1831

Flow Injection Renewable Surface Immunoassay: A New Approach to Immunoanalysis with Fluorescence Detection Cy H. Pollema and Jaromir Ruzlcka' Department of Chemistry B G 10, University of Washington, Seattle, Washington 98 195

This paper introduces a new methodology of carrying out heterogeneous immunoassays automatically, using a flow injection technique on a renewable surface. Flow injection renewable surface immunoassay (FIRSI) relies on the use of a minute amount of beads to form a reactive surface, which is interrogated by fluorescence spectrometry. Following the assay, on-line regeneration normally used in flow based immunoassays is avoided by fluidically removing the spent reactive surface and replacing it with a new layer of beads. This allows the monitoring of antibody-antigen binding at its early stages, dramatically increases the sampling frequency of a serial assay, and eliminates the problems associated with a decrease in surface reactivity caused by repetitive use. A model system utilizing anti-mouseIgG1-coatedbeads and mouse IgCl protein is used to characterize the method with respect to reproducibility,flow rate, contact time, and amount of beads. Heterogeneous assays comprise a majority of currently performed immunoassays,' since the choice of homogeneous assays is limited by the restrictive requirement that some change in signal must occur upon binding. Use of a solid phase removes this requirement by physically separating the bound and the unbound portion of the analyte, typically by means of an appropriate antibody immobilized on the surface of a solid phase. In this way, the analyte is selectively bound to the immobilized antibody while the remaining portion of the sample can be removed by the washing steps. There are several types of solid phases used for heterogeneous immunoassays.2 Large diameter beads are used since they can be easily washed and retained in a test tube during an analysis. Immunomagnetic beads can be held in a magnetic field to facilitate washing and separation steps. Well plates and test tubes are also used in batch methods with the container walls serving as the reactive surface. This area is dominated by the microtiter well, which accounted for over 70%of the new methods in 1990.' These well formats typically use a 96-well plate or smaller well strips and are the basis for batch methods in which the reactive surface is discarded after a single use. Since the reactive surfaces are only used once and discarded, high-affinity antibodies can be immobilized to increase sensitivity and speed the binding p r o c e ~ s . ~ Yet automation of these methods requires complex instrumentat i ~ nor, ~if performed manually, are quite labor intensives5-' (1) Gosling, J. P. Clin. Chem. 1990, 36, 1408-1427. (2) Chan, D. W. Immunoassay Automation: A Practical Guide; Academic Press Inc.: San Diego, 1992. (3) Stengerg, M.; Werthen, M.; Theander, S . ; Nygren, H. J . Immunol. Methods 1988, 112, 23-29. (4) Kramer, G. W. Clin.Chem. 1990, 36, 1556-1560. ( 5 ) DeMoranville, V. E.; Ellis, J. E. Clin. Chem. 1990, 36, 1588-1590. 0003-2700~94~0366-1825$04.50/0 0 1994 American Chemical Society

In either case, a large number of assays require a considerable quantity of plastic trays which are costly and which, being contaminated, must be disposed in an environmentally acceptable and therefore expensive manner. Continuous flow methods offer several advantages for automatic immunoassay^.^^^ First, the many washing steps required are inherent in a flowing system. Second, the flow channel is enclosed and thus less prone to contamination. Finally, the flow-based assays can be carried out rapidly. However, a common drawback of many flow-based methods is the necessity of using columns packed with antibody-coated beads, which serve as the retention surface. This retention surface cannot be discarded after each use, but must be regenerated prior to the next assay. Regeneration requires breaking the antibody bond to the analyte without altering in any way the immobilized antibody, since changes in the surface reactivity will result in different binding characteristics and ultimately erroneous results. Avoiding regeneration would save time and remove the restrictions caused by having to break the antibody bond quickly. We have been investigating methods that do not require additional on-line regeneration steps by creating a renewable reaction surface that can be fluidically manipulated. Initial work involved the use of immunomagnetic beads in a sequential injection analysis (SIA) system.1° The present investigation describes a technique termed flow injection renewable surface immunoassay. FIRSI is based on two key components, the jet ring cell" and a FIA system. The jet ring cell allows the retention of a well-defined amount of beads forming a renewable reactive surface, which is monitored by an epi-illumination microscope equipped with an end on the photomultiplier tube. The FIA system provides the required fluid handling ability to handle the beads, samples, and necessary reagents required to carry out a reproducible assay. The principle of FIRSI is shown in Figure 1. The first step is to inject, by means of a valve, a well-defined volume of bead suspension, which is propelled through a transfer line into the jet ring cell. The beads are retained within the ring, while the fluid passes through the jet ring gap. The following steps depend on the type of assay. For the purpose of describing the technique, a simple competitive assay is now considered. (6) Herold, C. D.; Andree, 39, 143-147.

K.;Herold, D. A.; Felder, R. A. Clin. Chem. 1993.

(7) Amphlett, M.; Smith, D. J.; Warren, R. E. J . Med. Microbiol. 1991, 35, 249-254. (8) Gubitz, G.; Shellum, C. Anal. Chim. Acta 1993, 283, 421-428. (9) Puchades, R.; Maquieira, A.; Atienza, J.; Montoya, A. Crit. Reu. A m / .Chem.

1992, 23, 301-321. (10) Pollema, C. H.; Ruzicka, J.; Christian, G. D.; Lernmark, A. Anal. Chem. 1992,64, 1356-1361. (11) Ruzicka, J.; Pollema, C. H.; Scudder, K. M. Anal. Chem. 1993.64, 35663570.

Analytical Chemistty, Vol. 66, No. 1 1, June 1, 1994 1825

Flow Direction

Readout

1

Injected Zone

J'

Time

+

J'

Time

+

J'

Time

+

J'

Time

+

f'

Time

+

Transfer Jet Ring Line Cell

Figure 1. Principle of flow injection renewable surface immunoassay (FIRSI). Step one is the injectionof the beadsuspensionand itstransport through the transfer line into the jet ring cell, where the beads are retainedon an optical flat surfacethat is monitoredby a suitabledetector (D). Step two is the injection of a mixture of labeled (L) and unlabeled sample (S). Step three occurs as the injected zone passes through the detection region perfusing the bead layer. Since all the injected label is detected at that time peak maximum A appears. Step four is the washing of any unboundportionof the label while a plateau Bis reached, reflecting the amount of bound labeled sample. In the final step, the flow is reversed, which removes the bead suspension from the jet ring cell, thus preparing the system for the next assay.

The second step then involves injection of a mixture of label (L) and sample (S). During the third step, while the label and sample zone are passing through the reactive surface and the viewing field of the detector, a peak is observed (A), corresponding to the total amount of fluorescent label injected. Next, the unbound analyte and label are washed away, while the signal decreases to a level (B) correspondingto the amount of bound label. Finally, the flow is reversed and the layer of beads is instantaneously flushed to waste. The system is now ready for the next assay. FIRSI combines the advantages of flow-based and batch methods. The ability to provide a pristine reaction surfaced for each individual assay is maintained without the need for a slow on-line regeneration step between analyses. FIRSI methodology is also more flexible than current flow-based heterogeneous immunoassays for the following reasons. The binding capacity of the solid support can be varied by injecting different volumes of bead suspensions. It has been shown that even a partial coverage of the jet ring cell area yields reproducible results.' This allows the optimization of the binding capacity for a particular range of analyte concentrations to achieve a useful dynamic range. Next, the use of a renewable support eliminates the limitation of using only reversible binding. l 2 The reactive surface is being monitored 1826

Analytical Chemistry, Vol. 66, No. 11, June 1, 1994

and discharged after each assay and therefore does not need to be eluted and regenerated. This, as with microwell methods, allows the use of high-affinity antibodies, which would otherwise require forbiddingly long regeneration times. In addition, FIRSI offers an unprecedented versatility by offering a choice to easily assay different analytes without any system reconfiguration. Bead suspensions with different reactive surfaces can be injected and removed from the jet ring cell in any desired sequence. Finally, FIRSI has a potential for greater sensitivity since detection is performed by directly monitoring the reactive surface rather than eluting and transporting the diluted label into a flow-through detector. The other reason for added sensitivity is based on theoretical considerations,13 which indicate an inverse relationship between the minimum concentration of sample to be detected (Cmin) and the affinity constant of the antibody ( K A f f ) (a is the error in the measure):

Since FIRSI does not require regeneration, the sensitivity may be increased with the use of higher affinity antibodies. The important parameters that determine binding in a flowbased system have been described by Hage and his colleagues in their pioneering work on the design and theory of a sequential addition immunoassay (SAIA). l 4 Since there are certain similarities between their technique and FIRSI, the theory of SAIA is briefly discussed here. The current method for carrying out a SAIA involves four steps. The first step is to apply the sample to a suitable column and to allow a predetermined binding time. During this time, the analyte binds to the stationary phase, while the unbound portion of the sample is eluted. The second step is to sequentially follow the sample with a fixed amount of a labeled analyte, typically at a concentration equal to that of the binding capacity of the column. The labeled analyte binds to the stationary phase as well, but this binding is restricted by the sites already occupied by the unlabeled analyte. This allows the quantitation of the analyte in the sample by one of two methods, which comprise the third step; detection can be carried out indirectly or directly by either monitoring the unretained label or by eluting and monitoring the labeled analyte. The final step is the regeneration of the stationary phase. The behavior of this system was described using a competitive binding model. Direct binding was determined by assuming the limiting step was the adsorption of the analyte to the immobilized antibody on the column. Depending on the binding sites available in relation to the amount of sample injected, two sets of conditions can describe the fraction bound. The first in which the available sites are in excess is a linear condition expressed as

The terms in eq 2 are the fraction bound ( B ) , the adsorption (12) Stenberg, M.; Nygren, H. J. Immunol. Methods 1988, 113, 3-15. (1 3) Ekins, R. P. Alternative Immunoassays; John Wiley & Sons Ltd.: New York, 1985.

(14) Hage, D. S.; Thomas, D. H.; Beck, M. S. Anal. Chem. 1993,65,1622-1630.

HC

Figure 2. Schematics of the sequential injection system used in this work. The system comprises a peristaltic pump (PP), holding coil (HC), multiposition valve (MPV) transfer line (TL), and the jet ring cell (JR), which is coupledto an epi-illuminationmicroscope.The beadsuspension (BS) is maintained in a rotating flask (RF). Bead suspension as well as sample solutions (S) are sequentially aspirated into the HC via MPV by a controlled flow reversal of the peristaltic pump. The forward movement of the pump transports injected zones through TL into the cell. The spent beads are aspirated by a flow reversal through MPV back into the HC and then following the valve switch by means of a forward flow into the waste.

rate constant (k3), the moles of binding sites ( m ~ )and , the volumetric flow rate (F). The other condition is described as nonlinear in which the amount of sample approaches the available sites, an expression and detailed description can be found in ref 5 . Nonlinear conditions are utilized in SAIA, and several applications of the technique have been published;lq-16 however, linear conditions are pertinent to this work. The important parameters that affect binding are the relative amount of analyte added, the relative amount of labeled analyte added, and the flow rate/adsorption kinetics of the system. Increasing the amount of analyte added produced a sigmoidally shaped calibration curve, and the amount of labeled analyte added does not effect this curve as long as the labeled analyte is present at a concentration twice that of the available sites present on the column. The flow rate can change the dynamic range of the calibration curve by several orders of magnitude, the main factor being the number of accessible binding sites on the column. This study introduces the principle of FIRS1 and explores the experimental variables that have been identified as important based on SAIA as a model. These are the flow rate, the contact time, and the amount of bead suspension introduced into the detection region. Different assay protocols are also discussed as they apply to this technique. EXPERIMENTAL SECTION The jet ring cell as described previously was connected by means of the transfer line to the multiposition valve of the SIA system (Figure 2). The SIA system consisted of a Valco 10-position dead stop selector valve (Valco Instruments Co. Inc., Houston, TX) and an Alitea type C-4V peristaltic pump (Alitea USA, Medina, WA). The holding coil connecting (15) Cassidy, S. A.; Janis, L. J.; Regnier, F. E. Anal. Chem. 1992,64, 1973-1977. (16)Nilsson, M.;Hakanson, H.; Mattiasson, B. J . Chromatogr. 1992,597, 383389.

the pump and valve was a 0.8-mm-i.d. Teflon PTFE coiled with a total calculated volume of 800 pL. Larger diameter tubing was used on the holding coil to help prevent outgassing during the aspiration of reagents. All other tubing was 0.5mm-i.d. Teflon PTFE with the transfer line being 25 cm long. The peristaltic tubing, Masterflex no. 13, was coated on the outside with silicon oil and released from the pump rollers when not in use to minimize wear. The detector was a Zeiss Universal microscope (Carl Zeiss, Oberkochen, Germany) equipped with a type 111-RS epifluorescenceattachment and 75-W halogen source. The source was shuttered using a Uniblitz' Model T132 controller and shutter (Vincent Associates, Rochester, NY). Fluorescence signals were detected by a Nikon P1 photomultiplier tube (Nikon, Tokyo, Japan). PMT detection of SIA zones used a 2.5X Planapochromatic objective with detection apertured down to a region of 1.6 mm X 1.6 mm. The excitation and emission wavelengths were determined by a dichroic mirror allowing 450-490 nm excitation and a 520 nm long pass emission filter. Control of the SI system and data collection from the PMT were carried out using a RTD ADA1 100-2 interface board (Real Time Devices, State College, PA) in a Comtrade 80486 computer. The data collection and control software was Atlantis (Lakeshore Technology, Chicago, IL). The agarose beads, size approximately 35 pm, (Sigma Immunochemical,St. Louis, MO) were purchased coated with a goat anti-mouse IgGl (heavy chain specific) for a final minimum binding capacity of 0.4 mg of mouse IgG 1/mL of resin. The stock solution was diluted 1:20 in 0.01 M phosphate buffer containing 0.5 M NaCl. The beads were maintained in suspension by rotation in a square flask. The concentration of beads in the final suspension was approximately 2.0 X lo5 beads/mL. The analyte used for this model study was mouse IgG 1 monoclonal antibody (Dako, Carpinteria, CA) supplied at a protein concentration of 1.8 g L-I with a mouse IgG concentration of 100 pg mL-l- This solution was diluted over a range from 1:10 to 1:40. The labeled monoclonal antibody was also purchased from Dako as a R-Phycoerythrin conjugated mouse IgGl antibody at a concentration of 100 pg mL-' and used over a similar dilution range. Both antibodies were diluted in 0.01 M phosphate buffer containing 0.5 M NaCl and stored at 4 "C when not in use. All experiments used 0.01 M phosphate buffer containing 0.5 M NaCl as the carrier using a range of flow rates from 0.25 to 1.00 mL min-I. RESULTS AND DISCUSSION Design of the flow injection system was aimed at minimizing the volume of sample and reagent needed for an assay. The volumes of the holding coil, transfer line, and valve tubing were minimized to prevent dilution of the sample and reagent zones as they are transported to the detector. The factor which affects dilution is dispersion. The method of determining dispersion in the FI system is based on the response of the detector to a given volume of the detected species (label). By injecting progressively increasing volumes of a 10 pg mL-' solution of fluorescein (Figure 3), these responses can be experimentally determined. One way to express dispersion is based on the S 1 / 2 , which is the volume that yields a signal half Analytical Chemistry, Vol. 66, No. 11, June 1, 1994

1827

40 00-

I 30.00x

2

-p

2o.M)-

:

-5 -L

1000-

0.00 0

2

w

R

s-

Time (sec)

Figure 4. Plot showing 10 repetitive runs of the binding of mouse IgGl

R-PE labeled antigen to the renewable reaction surface, consisting of goat anti-mouse IgGl-coated agarose beads. The bead suspension volume was 42 pL with a flow rate of 1 mL min-l. Injected volume (uL)

Figure 3. Key flow parameter of the SIA system; the S2 Il value is identified at y = 0.693. The value of 45 pL reflects the volume and

flow geometry of the flow path, which was designed to minimize dispersion to effectively utilize small sample and bead volumes.

that of the undiluted label. A smaller S1p value indicates a system with less dispersion. The SI/Z is calculated by plotting the logarithm of one minus the ratio of the signal (for a given volume) ( c ) versus the signal produced by the undiluted label (cg) for a series of different volume^.'^ The volume at which the y-axis equals 0.693 ( c = 0 . 5 ~ 0is) the Sip, for this system, that volume was 45 pL. It is also the volume that was chosen as a compromise between reagent economy and minimum dilution. Further decrease in sample and label consumption (by decreasing the S1/2) is possible through changes in the design of the FI system, as discussed at the end of this paper. Introduction of a new reactive surface with a high degree of reproducibility is a key condition for a successful implementation of FIRSI. It was therefore essential to investigate not only the formation of the reaction surface and its removal but also the ability of the surface to retain the same amount of labeled analyte in each assay. This was carried out by maintaining a thoroughly suspended mixture of beads using a rotating flask while minimizing the line volume from this rotating flask to the valve. Once the suspension enters the SIA system, it needs to be completely retained within the jet ring cell and perfused by the label and by the analyte. Visual observations have confirmed that once the reactive surface was formed in the view of the detector, it remained fixed even during stop flow periods of up to 20-min duration or until a flow reversal was applied. It was also confirmed that the buildup of the bead layer always followed the same pattern, with beads first filling the circumference of the ring and then gradually filling the entire ring area with the degree of coverage depending on the volume of injected suspension. The injected bead volume was selected to be 42 pL, a choice based on visual observations as the minimum volume of bead suspension that covered the detection region. Figure 4 demonstrates 10 (17) Ruzicka, J.; Hansen, E. H. Flow Injection Analysis, 2nd ed.; Wiley-

Interscience: New York, 1988.

1828 Analytical Chemistry, Vol. 66,No. 11, June 1, 1994

sequential injections of labeled sample onto the renewed reactive surface. The curve comprises the following three sections: (A) a change in signal from baseline to peak maximum due to the labeled sample passing through the cell, (B) the tailing portion of the peak corresponding the the amount of label bound, and (C) a return to original baseline level following removal of the beads from the cell. The relative standard deviation of the plateau of the tailing portion (B) for 10 sequential experiments is 2.2%. The next area to consider is the flexibility afforded by the SIA system. There are several experimental parameters that can be easily varied to optimize the assay conditions, including the flow rate, the contact time of sample with the reactive surface which can be varied using stop flow, and the amount of reactive surface present in the detection region. All of these parameters can be changed by flow programming. Flow rate is a significant parameter affecting the amount of bound analyte and label. This has been shown to be an effective variable in SAIA and is equally important to FIRSI. The results of three different flow rates are shown in Figure 5 . The binding significantly increased by slowing the flow rate to 0.25 mL min-I, and the increase was still linear throughout five sequential injections of labeled sample indicating a large excess binding capacity for bead volumes of 42 p L ,corresponding to 0.3 mg of beads. Obviously, slowing the flow increases the contact time of the sample zone, thus increasing the amount of antibody bound. A logical extension to even more effective binding is to stop the flow for a selected period of time. Stopped flow techniques allow a choice of contact time as well as a concentration of the reagents and analytes to be brought in contact with the retained bead layer. By selecting a time which elapses between zone injection and commencement of the stop flow period, either the centroid of the dispersed injected zone or a more diluted portion of its trailing section can be held in contact with the bead layere8 In the following experiments, the stop delay time was selected (6 s) so that the centroid of the dispersed zone was held in contact with the reactive surface for a period of 25-500 s. The x-axis (Figure 6) indicates the stop flow time in seconds, while the y-axis is

40

50 uL

.25 mWmin

LUL -

30

.-y.

5 8.50 mLtmin

f: 4, bl

20

e,

P e .OOmL/min

2

G

IO

Background Level Background level I

-

I

“

*

V

I

0

Repeated Injections (#)

Figure 5. Effect of changing the flow rate over a range from 1.0 to 0.25 mL min-1 with the reactive surface formed by injecting 42 pL of the bead suspension. Labeled antigen at a concentration of 2.5 pg mL-1was repeatedlyinjectedon the same bead layer while the increase in fluorescence was monitored. 4.00

I

f

-

r

4

m

I

*

V

I

3

Repeated Injections (#)

Figure 7. Effect of changing the volume of bead suspension used to form the reactive surface. Repeatedinjectionsof R-PE-labeled antibody (2.5 pg mL-l) were sequentlaity perfused through the same bead layer to determine its binding characteristics. Four different volumes of bead suspension were used ranging from 17 to 50 I.L with a constant flow rate of 1 mL mln-l. The non-zero intercept indicates the background fluorescence level within the jet ring cell.

0

3.50 0

3.00

2.50

2.00 0

1S O

1.oo

1

High Affinity Binding

0.50

Contact Time (sec)

Figure 8. Stop flow experiments covering a contact time range from 25 to 500 seconds. The x-axis represents the stop flow Contact time of the sample with the bead layer. The y-intercept indicatesthe amount which occurs during this stop time as a ratio of the binding of binding(4 which occurs with no stop flow (60).

based on the signal obtained from the plateau level of the peak. This signal is expressed as a unitless ratio which reflects the amount of binding ( B ) relative to the amount of binding which occurs with no stop in the flow (Bo). These results indicate that high-affinity binding occurs in the first 25 s of stopped flow contact, during which a significant fraction of the antibody is preconcentrated on the beads. Increasing contact times results in a linear increase in binding, which will begin to level off after 500 s. This is indicated by considering

the relative amounts of signal produced by the plateau level ( B ) and the peak maximum ( A ) portions of the profile shown in Figure 4. Increasing the stop flow times to 500 s results in a ratio of peak height to plateau level of approximately 1 (data not shown). This indicates that extending the stop flow times would not result in significantly more binding. However, the data shown reflect the binding rate and the large capacity of the reactive surface. Since a significant portion of the binding occurs within the first 25 s, the present conditions are suitable for a rapid assay. The volume of bead suspension is the final parameter to be considered. In SAIA the column binding capacity is fixed, since it is not practical to change readily the amount of column material. In contrast, the amount of immunosorbent can easily be varied in FIRSI, by selecting the volume of the bead suspension injected into the system. The results obtained by injecting 17-50 pL of bead suspension are shown in Figure 7. The flow rate was maintained at 1 mL m i d , and five 45-pL injections of labeled antibody at a concentration of 2.5 pg mL-l were sequentially passed through each bead layer. The result was an increase of the amount of labeled antibody retained on increasing bead quantities. Doubling the bead volume (from 17 to 34 pL) doubles the signal intensity, but after full coverageof the jet ring area has been achieved (above 42 pL of bead suspension injected), not all fluorescence will reach the detector due to the thickness of the bead layer, and therefore, the change in signal will decrease. This test provides useful information about the optimum number of available binding sites, which can be monitored for a given jet ring cell dimensions, and allows for the dynamic range of the assay to be adjusted depending on the binding capacity of selected bead material. Analjlticai Chemistty, Voi. 66,No. 11, June 1, 1994

1029

1.2c

1.3

1.25

.oo

1

1.2

P 0.80

1.15

I

6

m

1.1

0.60 1.05

0.40 1

I 0.20

I

L" N

I v)

I

2

I

2

Competing Antibody (ug/mL) Figure 8. Competitive FIRSI for a sample antibody using labeled antigen (insert, labeled squares). The analyte antibody, premixed with the label, (step 1) is injected onto the bead layer containing the same immobilized antibody. The fraction of labeled antigen not bound by the anaiyte antibody is retained by the immobilizedantibody (step 2). The degree of binding Is plotted as a ratlo of the amount bound (8) versus the amount which binds when no competing antibody is present (5). The resulting calibration curve is typlcal for a competltive assay.

A competitive immunoassay was performed for an antigen analyte using a limited number of antibody binding sites on the beads and a fixed amount of competing labeled antigen. A series of sequential addition immunoassays were unsuccessfully attempted by injecting increasing concentrations of analyte (mouse IgG1) followed by a known volume of the same, but labeled, analyte. The amount of labeled analyte bound to the beads was constant, with no influence due to the previous introduction of the analyte. There are two possibilities for the failure of this approach. The first would be that the high number of available binding sites on the beads resulted in no inhibition of the binding of the label by the sample. The second would be that theoff-rate of the sample was sufficiently fast in this flow-through system to remove the sample before the label was introduced. Therefore, the next protocol was selected, which is based on an excess of available binding sites on the beads and a one-step competition. In this protocol, the analyte is an antibody identical to the antibody immobilized on the beads. This analyte is mixed with a fixed amount of labeled antigen, and the sample analyte competes with the immobilized antibody for the labeled antigen (Figure 8, insert). Again (goat anti-mouse IgG1) antibody-coated beads were used, and the analyte (goat anti-mouse IgG) was introduced premixed with a fixed amount of labeled antigen (mouse IgG1). In this way the analyte antibody binds to the labeled antigen, thus preventing the labeled antigen from binding to the reactive surface. Therefore, an increase in sample antibodies would result in a decrease in fluorescence signal. Different con1830

Analyrical Chemistry, Vol. 66, No. 11, June 1, 1994

0.95

I

I

N

I

m

I

d

I

m

Sample Antigen (ug/mL) Flgurr 9. NoncompetitiveFIRSI for an unlabeled antlgen(insert, open squares). The labeled antibody, premlxed wlth the anaiyte, (step 1) is injectedonto the bead layer with the immobiilzedantibody. This complex then blnds to the immobilized antibody (step 2). The degree of binding is plotted as a ratio of the amount bound (4 versus the amount which The resulting calibration line is binds when no antigen is present (5). typical for a noncompetitive assay.

centrations of sample antibody over a range from 10 to 1 pg mL-1 premixed with a fixed amount of 5 pg mL-' of labeled antigen and a contact time of 25 s yielded a well-resolved calibration curve (Figure 8). The range from 5 to 1 pg mL-I gives a linear response, which drops off as the analyte is in high excess of the label. Each assay was carried out in triplicate with error bars indicating the standard deviation of the method. Noncompetitive immunoassay is another format suitable for a high-capacity immobilized reaction surface. For this method, the same beads were used to perform an assay to measure antigen concentration. This antigen was determined by premixing it with a labeled antibody directed against another region of the antigen (Figure 9, insert). The restriction for this type of assay is that the analyte molecule must be large enough to have two discrete regions which the antibodies can 'sandwich' the sample antigen between. The binding was expected to be slower due to thecomplex of antigen and labeled antibody binding to the immobilized antibody, and therefore the contact time was increased to 200 s (shorter contact times did not result in a notable change in signal). While the sensitivity of this assay is lower than that of the competitive format, it does show a good linear range and proves that FIRSI is suitable for a sandwich type of assay.

CONCLUSION The present work introduces the principle and methodology of FIRSI and validates its concept on a model competitive and sandwich assay using fluorescencedetection. The binding characteristics of the chosen immunosorbent has been evalu-

ated by a series of automated experiments, whereby contact times and amounts of reactants werevaried through the change of flow rates and injected volumes. It was shown that FIRSIbased assays are very fast and economical, requiring as little as 0.3 mg of beads and 45 p L of labeled antibody (or antigen) per analysis, which lasted under 2 min for a competitive assay and less than 5 min for a noncompetitive assay. The most prominent feature of FIRSI is its versatility. It is suitable for most immunoassay formats, although a sequential competitive assay still has to be demonstrated using a sorbent of a lower binding capacity. Since the jet ring cell is capable of handling a wide variety of beads,3 the present use of commercially available agarose beads confirms, in principle, the applicability of FIRSI to a wide range of available sorbents and fluorescent labels. The ability of FIRSI in an SIA format to vary experimental conditions directly from computer keyboard, rather than by physically reconfiguring the flow system, is yet another aspect of its flexibility. The present drawback of FIRSI is the need to keep the beads well suspended in order to inject reproducibly the same amount each time. While a rotating flask performed reliably, an alternative approach might be desired. Should sample and reagent consumption be further decreased, the flow system

will have to be redesigned. Use of classical FI valve-based injection rather than volume-based sequential injection is an alternative, since the absence of flow reversal in FI will decrease the S l p of the system. Exciting topics for further research include the use of enzymatic labels, for which FIRSI seems to be ideally suited, due to its ability to monitor reaction rates continuously on the surface of the bead layer. Since cells can be grown on a suitable bead support, such as Cytodex, and since it has been shown that cytochemical reactions can be quantified by fluorescence microscopy in the jet ring cell,3 FIRSI may perhaps become useful also for the study of immunocytochemistry of living cells.

ACKNOWLEDGMENT The authors would like to thank Gary Christian for his assistance with this project. The work was supported by NIH (Grant SSS-3 (5) R 0 1 G M 45260-2). Received for review December 2, 1993.

Accepted March 9,

1994." Abstract published in Adunnce ACS Abstracts, April 15, 1994.

Analytical Chemistry, Vol. 66,No. 11, June 1. 1994

1831