Flow-injection sandwich ELISA for bioprocess monitoring

A fully automated flow-injection immunoassay based on sandwich enzyme-linked immunosorbent assay (ELISA) is described for the model system: protein G-sepharose, rabbit IgG and horseradish peroxidase (HRP)-labelled protein A. After injecting rabbit IgG and HRP-labelled protein A into a cartridge containing protein G-sepharose sequentially, a mixture of hydrogen peroxide and the redox indicator, 2.2′-azino-bis(3-ethylbenzthiazoline-6-sulphonic acid) (ABTS) is passed through the cartridge. The HRP-labelled protein A bound in the cartridge is directly proportional to the concentration of rabbit IgG. The colour variation of ABTS caused during the reaction between HRP and H202 in the cartridge is detected photometrically. The whole assay procedure is controlled and evaluated by a computer. Rabbit IgG and HRP-labelled protein A are also detected by a fluorometer, which is introduced into the flow system. In the flow-injection sandwich ELISA, the slope of the calibration curve is 0.4491 in the range of 0 and 300 μg ml-1 rabbit IgG, while it is 0.1274 in the heterogeneous immunoassay. So the flow-injection sandwich ELISA system is found to be more sensitive than a heterogeneous immunoassay for the monitoring of the model protein.

G-sepharose sequentially, a mixture of hydrogen peroxide and the redox indicator, 2.2'-azino-bis(3-ethylbenzthiazoline-6-sulphonic acid) (ABTS) is passed through the cartridge. The HRPlabelled protein A bound in the cartridge is directly proportional to the concentration of rabbit IgG. The colour variation of ABTS caused during the reaction between HRP and H202 in the cartridge is detected photometrically. The whole assay procedure is controlled and evaluated by a computer. Rabbit IgG and HRPlabelled protein A are also detected by a fluorometer, which is introduced into the flow system. In the flow-injection sandwich ELISA, the slope othe calibration curve is 0.4491 in the range of 0 and 300#g mlrabbit IgG, while it is 0.1274 in the heterogeneous immunoassay. So the flow-injection sandwich ELL SA system is found to be more sensitive than a heterogeneous immunoassay for the monitoring of the model protein.
which is illustrated in figure 1. Antibodies are immobilized covalently on a support and integrated in a cartridge, which is placed in the flow-injection system (A). After injection of the sample containing antigen (B), the cartridge is washed and enzyme-labelled antibody is passed through the cartridge, which binds to the antigen (C). In a rinsing step, unbound enzyme-labelled antibody is washed out from the cartridge by the carrier buffer solution. After substrate injection, a reaction between substrate and bound enzyme-labelled antibody occurs (D). The extent of the reaction is proportional to the concentration of antigen in the sample, as measured via a photometric detector. Afterwards, both bound antigen and enzyme-labelled antibody are eluted for the next assay (E).
In our work, protein G-sepharose is utilized as support material, rabbit IgG as model analyte and HRP-labelled protein A as conjugate. For the enzymatic reaction with bound HRP-labelled protein A, hydrogen peroxide (H202) is chosen as substrate and ABTS as the redox indicator, so the colour variation of the reaction is detected photometrically. Effects of elution buffer, carrier flow stop time and binding capacity are investigated for the model system. The sensitivity of the flow-injection sandwich ELISA system is also compared with that of a heterogeneous immuno-FIA system [9].

Introduction
On-line monitoring of biotechnological processes is important for a better understanding of cell growth and for process improvements. In mammalian cell cultivation processes, the microtitre-plate enzyme-linked immunosorbent assay (ELISA) is normally used to measure concentrations of protein products. However, this ELISA technique is time consuming and labour intensive. Some efforts have been made to combine the conventional ELISA with flow-injection analysis (FIA) for the on-line monitoring of protein products [1][2][3]. De Alwes and Wilson reported a concept for both sandwich-and competitive-type assays with glucose oxidase [4,5]. Based on the competitive binding between the enzyme-labelled antibody and antigen, Lee and Meyerhoff presented a non-equilibrium flow-injection ELISA system using an immobilized secondary-antibody reactor and an ammonium ion-selective potentiometric detector [6]. Recently, Nilsson and co-workers described a competitive flow ELISA system using a spectrophotometer for the monitoring of IgG [7] and a-amylase [8]. However, there are no systematic studies on a flow-injection sandwich ELISA using fluorescence detection for the monitoring of protein products. This paper describes a flow-injection sandwich ELISA system using fluorescence detection, the principle of Experimental Apparatus Figure 2 shows a schematic set-up of a flow-injection sandwich ELISA system. Two six-way selection valves are employed. One is used to select sample and conjugate for the sample loop (S1), the other is used to switch substrate and buffer flows ($2). A cartridge containing ml protein G-sepharose is placed behind an injection

Reagents
The following reagents were used: protein G-sepharose 4B   it can be said that a 1:400 dilution (0.2 U HRP ml-) of HRP-labelled protein A can be employed for the measurement of 100 lag ml -l rabbit IgG. There was also a fluorometric output of unbound HRP-labelled protein A, when a 1:500 dilution (0.16UHRPml-l) of HRPlabelled protein A is used for the measurement of 50 lag ml -I rabbit IgG. Assay cycle Figure 3 shows a typical timing sequence, and resulting photometric and fluorometric output for 50 lag m1-1 and 100 lag ml rabbit IgG. For this assay, a 1"400 dilution (0.2 U HRP ml-l) of HRP-labelled protein A is used as conjugate. First, two small peaks were detected by the ftuorometer, resulting from unbound rabbit IgG (Fpl) and HRP-labelled protein A (Fp2). After substrate solution is introduced into the cartridge, a reaction between bound HRP-labelled protein A and substrate takes place.
The colour variation of the redox indicator (ABTS) caused by the reaction is detected by the photometer. The peak height measured is proportional to the concentration of rabbit IgG. During the elution step, both bound rabbit IgG and HRP-labelled protein A are eluted from the cartridge and produce a third peak (Fp3) which is detected fluorometrically. Enzyme-labelled antibody, i.e. conjugate, should be normally introduced into a sandwich ELISA in excess to analyte. In this work, HRP-labelled protein A binds to IgG, so that protein A should be introduced into the flow-ELISA in excess to IgG. However, the concentration of protein A in HRP-labelled protein A used is not given. Therefore, unbound HRP-labelled protein A should be detected fluorometrically in order to investigate whether HRP-labelled protein A diluted was introduced in excess to IgG. As shown in figure 3, there were fluorometric outputs for unbound HRP-labelled protein A, when a 1:400 dilution (0.2 U HRP ml-) of HRPlabelled protein A was used as conjugate. From this result

Effects of elution buffer
The regeneration extent of the cartridge plays an important role for repeated use of the cartridge containing protein G-sepharose. The influence of elution buffer on the dissociation of binding between rabbit I.G and protein G is investigated for 0 and 100 lag mlrabbit IgG. Under the conditions outlined in the flow-injection sandwich ELISA, 100 lag ml -rabbit IgG was first measured using a 1:100 dilution (0.8 U HRP ml-) of HRPlabelled protein A as conjugate. After the measurement the cartridge is eluted with an elution buffer. Next, a sample containing no rabbit IgG is injected, then conjugate and substrate serially. Figure 4 shows the peak heights of 0 and 100 lag mlrabbit IgG with four different elution buffers normally used. The largest difference of peak heights is obtained with elution buffer A (0.1 M glycine-HCl at pH 2.0). Even if the difference with elution buffer C (3 M NaSCN) is larger than that with elution buffer B (0.1 M KaPO4 at pH 12.3), it has been shown that there was noise phenomenon in the signal with elution buffer C. The dissociation of binding between protein G and rabbit IgG did not occur with elution buffer D (4 M urea). The peak height of 0 lag m1-1 rabbit IgG was a little higher than that of 100 lag ml -l rabbit IgG, because a little HRP-labelled protein A injected during the measurement of 0 lag m1-1 rabbit IgG was accumulated in the cartridge.  and their differences with different carrier flow stop times (R(50)-3 means that 50 lzg m1-1 rabbit IgG is employed as sample, and carrierjlow is stopped for 3 min). The peak height of 50 Izg m1-1 rabbit IgG is set at 100% when carrier flow is not stopped.
the flow-ELISA system can be stopped in order to increase the sensitivity of the system. However, a compromise between the assay speed and the sensitivity should be made.
Binding capacity A few IgGs, e.g. human IgG and mouse IgG, also bind to protein G. The binding capacity of these IgGs to protein G is compared for 50 lag m1-1 and 100 lag m1-1 using a 1:200 dilution (0.4 U HRP ml--) of HRP-labelled protein A as conjugate. For the measurement of each IgG, a fresh cartridge packed with protein G-sepharose is also employed. Figure 6 shows the relative peak heights of rabbit IgG, human IgG and anti-A MAB (an IgG-type produced by Boehringer Ingelheim Pharma KG). The peak height of 0 lag ml -IgG is set at 100% and used as reference. Rabbit IgG and human IgG bind to protein G very well, and the peak height of 100 lag m1-1 is about 2.5 times higher than that of 0 lag m1-1. However, there is no large difference in peak heights with anti-A MAB, so it has less binding capacity to protein G.
between rabbit IgG and protein G, and was used in our further experiments as elution buffer.

Effects of carrier flow stop time
The extent of the reaction between substrate and bound HRP-labelled protein A is related to the residence time of substrate in the cartridge. The influence of carrier flow stop time on the peak height is studied for 0 and 50 lag m1-1 rabbit IgG using a 1"500 dilution (0.16 U HRP ml-) of HRP-labelled protein A as conjugate. Figure 5 shows the relative peak heights for 0 and 50 lag m1-1 rabbit IgG. The peak height of 50 lag m1-1 rabbit IgG is set at 100% when the carrier flow has not been stopped. The

Reproducibility
The reproducibility of the system is investigated using 0 and 50 lag m1-1 rabbit IgG and a 1:250 dilution (0.32 U HRP ml-) of HRP-labelled protein A. Figure   7 shows the peak heights for 0 and 50 lag mlrabbit IgG repeatedly measured. There is a decrease in peak heights for 50 lag mlrabbit IgG and a slight increase in peak heights for 0 lag ml --I rabbit IgG. This decay of the binding capacity of protein G-sepharose in the cartridge is caused by the denaturation due to the low pH elution buffer and the substrate solution. However, the capacity decay of the cartridge can be compensated by correcting sample peak heights with reference sample, e.g. 0 lag ml-' rabbit IgG [8]. After the compensation with 0 lag ml -I rabbit IgG, the standard deviation (SD) equals 1  reproducibility of the flow-ELISA system, and that rabbit IgG can be monitored on-line very well, provided that a correction for the decay of a cartridge is adequately performed.  figure 8, the slope of the calibration curve by the photometric detection is 0.4491 mV (lag ml-])in the range of 0 and 300 lag mlrabbit IgG, while it is 0.1274 mV (lag ml-]) -1 with the fluorometric detection.
From this result, it can be seen that the flow-injection sandwich ELISA has higher sensitivity and a lower detection limit than the heterogeneous assay.

Conclusion
A flow-injection sandwich ELISA is described using the model system: protein G, rabbit IgG and HRP-labelled protein A. HRP has been shown in our studies that the flow-injection sandwich ELISA would be an attractive alternative to the on-line monitoring of protein products in biotechnological processes. Further, a flow-injection sandwich ELISA for the on-line monitoring of the rt-PA (recombinant tissue-type plasminogen activator) is envisaged using anti rt-PA as immobilizing material and HRPlabelled anti rt-PA as conjugate.