Microchip-based enzyme-linked immunosorbent assay (microELISA) system with thermal lens detection

A microchip-based enzyme-linked immunosorbent assay (microELISA) system was developed and interferon-γ was successfully determined.

The system was composed of a microchip with a Y-shaped microchannel and a dam structure, polystyrene microbeads, and a thermal lens microscope (TLM).

All reactions required for the immunoassay were done in the microchannel by successive introduction of a sample and regents.

The enzyme reaction product, in a liquid phase, was detected downstream in the channel using the TLM as substrate solution was injected.

The antigen-antibody reaction time was shortened by the microchip integration.

The limit of the determination was improved by adopting the enzyme label.

Moreover, detection procedures were greatly simplified and required time for the detection was significantly cut.

The system has good potential to be developed as a small and automated high throughput analyzer.


Immunoassay, represented by enzyme-linked immunosorbent assay (ELISA), is a well-known and important analytical method in the life sciences.

It is an indispensable technique in biochemistry and clinical diagnosis fields because of its extremely high specificity and sensitivity.

The conventional heterogeneous immunoassay, however, has a relatively long assay time, and involves troublesome liquid-handling procedures and many expensive antibody reagents.

Moreover, automated assay systems used for clinical diagnoses require rather large apparatuses.

A microchip-based system is potentially an effective way to realize a highly sensitive analysis system which overcomes these drawbacks.

Integration of analytical systems into a microchip should bring about enhanced reaction efficiency, simplified procedures, shortened assay time, and lowered consumption of samples, reagents, and energy.

Some papers have reported on miniaturization of ELISA or immunosorbent assay systems, in which antigen and antibodies are fixed on a solid surface in a microchip.1–11

Most were preliminary reports that only mentioned the possibility of immunosorbent assay or an antigen-antibody reaction in a microfabricated device, however, we described a microchip-based immunosorbent assay system for determinations of some proteins with high sensitivity.2

In that paper, we showed the reaction time necessary for an antigen-antibody reaction was reduced to 1/90 in the integrated system, because of size effects of the liquid microspace.

Moreover, we applied this system to a clinical diagnosis system, which is based on a determination of a major tumor marker for colon cancer known as human carcinoembryonic antigen (CEA).5

The system provided rapid and sensitive determination and showed a high correlation with the conventional assay in practical measurements of patients' sera.

In addition, we succeeded in doing a simultaneous assay of multiple samples with a microchip which had branching parallel multichannels.7

The high sensitivity of our microchip-based immunoassay system was attributable to the combination of a good labeling material, i.e. colloidal gold nanoparticles, and an ultra high sensitive detector, a thermal lens microscope (TLM).

In the system, colloidal gold fixed on the surface of microbeads via an antigen-antibody complex was detected by focused laser beams.

Although the method had high sensitivity, the precision was poor because of the difficulty in focusing the laser onto the top of spherical surface and the heterogeneity of the protein adsorption and antigen-antibody reaction on the small areas of the beads.

Therefore, for determination of one sample, 5–10 beads had to be measured with the TLM while the laser was successively focused on the top of each bead and then the average was taken.

Hence, the procedures for detection were very troublesome and time-consuming and good instrumental skills were necessary.

Whereas, the reaction time required for the antigen-antibody reaction was sharply cut by microchip integration, reduction of the total assay time was limited because of the troublesome detection procedures.

Improvement of the detection scheme is effective to realize higher throughput and more precise assay.

In this paper, an ELISA system was integrated into a microchip, in which peroxidase was used as a labeling material instead of colloidal gold and an enzyme reaction product was detected in a microchannel with the TLM.

Detecting the reaction product in a solution was much easier to do than making surface measurements and determination limit was improved by adoption of the enzyme label due to amplification of the signals derived by the enzyme reaction.

Materials and methods

Assay scheme

Schematic illustrations of our bead-bed microELISA system are shown in Fig. 1.

Polystyrene beads pre-coated with a capture antibody were put into the reaction channel of the ELISA microchip (Fig. 1-1).

Next, a sample containing an antigen (Fig. 1-2), a biotinylated second antibody (Fig. 1-3), and streptavidin-peroxidase conjugate (Fig. 1-4) were introduced into the channel, successively.

Finally enzyme reaction substrate solutions were continuously pumped into the channel (Fig. 1-5), while monitoring the reaction product with the TLM positioned downstream (Fig. 1-6).

Microchip fabrication

A dam structure is necessary to pack a definite amount of the microbeads in the microchip.

In our previous paper, we reported a two-step fabrication method, i.e. the combination of laser abrasion for microchannel formation and dry etching by a fast atom beam for dam structure formation.

This method, however, was troublesome and time-consuming and required expensive large equipment.

For easy and rapid fabrication of the chip, we developed a one-step or two-step wet etching method in which the chip was fabricated by simply combining standard photolithography and wet chemical etching techniques.

The basic etching procedures were described previously.12

In the one-step method, the microchannel and dam were fabricated simultaneously.

For the dam structure, a photomask with a gapped channel pattern was used (Fig. 2A).

After photolithography, the gapped narrow window was opened on a Pyrex glass surface, and then the glass substrate was etched isotropically with HF solution.

Two separate narrow channels appeared at first; these became wider and deeper and then came to connect to one channel with a neck as etching proceeded.

By this method, microchannels and dam were simultaneously created only with a simple one-step etching.

The structure image of the dam region is shown in Fig. 2B.

Two channels with curved surface were connected slightly at the center of the image.

Beads can be retained one side of the dam region.

For the two-step method, the microchannel and dam were fabricated individually by using a different photomask for each.

The former mask had a wider gap than that of the one-step method for dam region.

The first wet etching was terminated before the two channels were connected.

Next, photoresists that remained on the dam area were removed by the second photolithography and then the appeared glass substrate was etched to become the dam structure.

Using this method, the microchannel and the dam were individually fabricated.

The etched substrate was laminated with a cover plate, which had reagent inlet and outlet holes using a thermal fusion bonding method as described previously12.


Phosphate buffer (PB; 1/15 M, pH 7.4) was prepared from phosphate buffer powder (Wako Pure Chemical, Osaka, Japan) and ultra pure water, and the buffer was filtered through a 0.2 µm membrane filter prior to use.

Polystyrene bead (Polybead polystyrene microspheres, 15 µm, 20 µm, and 25 µm in diameter with 1% CV) suspensions were obtained from Polysciences (Warrington, PA).

The beads were washed with PB prior to use.

Recombinant human interferon-γ (IFN) was obtained from R&D Systems Inc. (Minneapolis, MN).

For capture antibody, chromatographically purified mouse anti-recombinant human interferon-γ monoclonal antibody (R&D Systems Inc.) was used.

For the second antibodies, biotinylated anti-recombinant interferon-γ polyclonal antibody from goats (R&D Systems Inc.) was used.

All antibodies and the antigen were diluted with PB and used without further purification.

Anti-IFN coated beads were prepared in a 500 µL microtube.

The monoclonal anti-IFN solution (100 µL, 50 µg mL−1) was added to the polystyrene beads (10 µL solid), and the suspension was stirred gently at room temperature for 1 h, followed by overnight incubation at 4 °C.

The coated beads were washed with PB and blocked with 0.2% casein solution for 1 h at room temperature and then washed with PB and stored at 4 °C.

Streptavidin-horseradish peroxidase (HRP) conjugate was purchased from Sigma-Aldrich (St. Louis, MO).

4-aminoantipyrine (AA) and H2O2 solution were obtained from Wako Pure Chemical and N-ethyl-N-(2-hydroxy-3-sulfopropyl)-3-methylaniline, sodium salt (TOOS) was from Dojindo Laboratories (Kumamoto, Japan).

AA, H2O2 and TOOS are reacted by catalysis of HRP to produce red-colored product.

Because the product has strong absorption at ∼530 nm, it is suitable for TLM detection using aYAG laser.

To prevent nonspecific binding of proteins to the microchannel and capillary tubes, 0.2% casein (Merck, biochemistry grade) in PB, 1% bovine serum albumin (Sigma-Aldrich) in PB, or BlockAce (Dainippon Pharmaceutical, Osaka, Japan) was used as a blocking reagent.


To determine very small amounts of the enzyme reaction product in a microchannel, a highly sensitive detection method with high space resolution is indispensable.

We have developed a laser-induced thermal lens microscope (TLM), which is especially useful for ultra sensitive determination in a microspace.13,14

The TLM was comprised of a microscope with two laser-oscillation apparatuses and other optical devices.

The excitation beam was the 532 nm emission line of a YAG laser (CrystaLaser, Reno, NV, model GCL-100-S) with output power of 76 mW, and its intensity was modulated by a mechanical chopper with a modulation frequency at 1.03 kHz.

A He–Ne laser (Melles Griot, Carlsbad, CA, model 05LHP171, 15 mW) with an emission line of 632.8 nm was used for the probe beam.

The two beams were made coaxial by a dichroic mirror, and tightly focused by an objective lens (Nikon, CF IC EPI Plan 20×, N.A.


The transmitted beams were collected by a condenser lens and filtered.

Only the probe beam intensity was monitored with a photodiode.

The pre-amplified signal from the photodiode was synchronously amplified with a lock-in amplifier LI-575 (NF Corp., Yokohama, Japan).

The liquid flow was controlled with a microsyringe pump (KD Scientific, Boston, MA) and Hamilton gastight syringes with untreated fused silica capillary tubing and capillary column connectors (GL Science, Tokyo, Japan).

Fused silica capillaries were connected to inlet and outlet holes of the microchip.

The capillaries for reagent introduction were connected to syringes, and the outlet capillary was connected to a syringe for suction or to a waste reservoir.

Before the assay, the inner walls of the capillaries and the microchannel were blocked with the blocking reagent for 1 h, and then this was replaced with PB.

A change of supplied reagents could be achieved simply by changing syringes.

During the enzyme reaction, the microchip temperature was controlled with a PE120 Peltier Heating and Freezing Stage (Linkam Scientific Instruments, Surrey, UK) which was mounted on the TLM.

Analytical procedures

To introduce the beads pre-coated with anti-IFN into the microchip, an aliquot of the bead suspension was placed in one inlet hole of the microchip, and then the beads were moved to the dam region by suction from the outlet hole while the other inlet was kept closed (Fig. 3).

This procedure was repeated until the packed bead channel (1 cm in length) was completely filled.

Excess beads were washed out by flushing PB from one inlet to the other inlet.

Then, sample solution containing IFN was introduced from one inlet into the packed bead channel at 1 µL min−1 for 10 min for the first antigen-antibody reaction.

Next, PB, biotinylated anti-IFN, PB, and streptavidin-peroxidase conjugate were injected into the microchannel successively from the inlet (Fig. 1).

Solution flow rates were adjusted to 100 µL min−1 for PB washing and for solution exchange, and they were 1 µL min−1 for the other reagents during the reaction.

After the reaction, the solution in the microchannel was replaced with PB, and then substrate solutions were introduced from inlet holes.

A mixture of TOOS and H2O2 was introduced from one inlet and AA was introduced from the other, all at the same flow rate.

The resulting product was continuously detected using the TLM downstream.

Signal intensity was recorded after reaching a plateau.

After the measurement, the beads were removed from the microchip by a reverse flow of PB and the microchannel was washed with 0.1 M NaOH solution.

Thus, the microchip could be used repeatedly.

Microchip-based immunosorbent assay using colloidal gold as a labeling material was performed as described previously7.

Results and discussion

ELISA microchip

A glass microchip suitable for ELISA was designed and fabricated.

The chip had two inlet holes, a Y-shaped channel, a dam, a detection area, and an outlet hole (Fig. 3).

Injection of a reagent or a sample solution was from an inlet using the syringe pump.

Because premixing of the three substrates might bring about some undesirable progression of the reaction outside of the chip without the enzyme, substrates were introduced from two inlets separately; one solution was a mixture of TOOS and H2O2, and the other was AA.

The two solutions were mixed in the packed bead area (1 cm long) which was from the mixing point of the Y-shaped channel to the dam.

The Y-shaped channel was suitable not only for mixing of the substrate solutions, but also for packing a definite amount of beads in the microchannel easily.

The dam structure was fabricated by the one-step or two-step wet etching method.

Any dimensions of the microchannel and dam could be altered by changing the gap width of the photomask and etching time.

The one-step method required only one photomask, one photolithography and one etching process to fabricate both a microchannel and dam.

However, accurate control of the channel depth of the dam was technically difficult for a narrow neck especially at depth of less than 10 µm because of unstable etching rate.

Therefore this method was suitable for beads with a diameter greater than 20 µm.

For smaller beads, the two-step method was suitable, because the second etching process for the dam was fully controllable.

The microchannel was 200 µm and 90 µm in width and depth, respectively, whereas the depth of the dam region was dependent on each experiment.

Diameter of the beads

In our previous paper, we used microbeads with 45 µm diameter for the microchip immunoassay using colloidal gold as a labeling material, because it is very difficult to focus a laser beam on the top of small beads.

In the microELISA system, however, specific interface area (S/V), which is defined as surface area of the beads to liquid phase volume ratio, is very important for enzyme reaction rate, because S/V is directly proportional to the amount of the enzyme fixed in the chip.

Therefore a large S/V is favorable for high amplification rate of the enzyme reaction, which means high signal intensity.

Although bead diameter is apparently inversely proportional to S/V, smaller beads often led to problems in the assay.

The most critical problem was a considerable increase of backpressure at the solution introduction procedure.

Moreover, removal of air bubbles, which accidentally contaminated the packed bead region, was also very difficult in the case of smaller beads.

After detailed studies, we concluded that the optimum bead diameter was 25 µm in our system.

By using 25 µm beads, a strong signal could be obtained without any problems.

Therefore microchips made by both one-step and two-step methods could be used for the assays.

Antigen-antibody and biotin-avidin reactions

The reaction time required for the antigen-antibody reaction was remarkably cut in the bead-bed microchip-based immunoassay system as described previously.

In the present microELISA system, the time to achieve equilibrium for the antigen-antibody reaction or the biotin-avidin reaction was also 10 min (data not shown).

Therefore, in this study, every reaction time was fixed at 10 min.

For the best reproducible results, reagent concentration was set at 1 µg mL−1 or 10 µg mL−1 for the biotinylated second antibody or the streptavidin-peroxidase conjugate, respectively.

Flow rate in the enzyme reaction

In the last step of the microELISA, the reaction of the HRP with TOOS, AA, and H2O2 was carried out in the packed bead area of the microchip.

The enzyme reaction could be performed in both continuous flow and stopped flow modes.

In the stopped flow mode, after the microchannel including the packed bead area was completely filled with the substrate solutions, the enzyme reaction proceeded for an appropriate time with the liquid flow stopped.

The resulting product, dissolved in the solution, was pushed downstream by the syringe pump and detected with the TLM.

We expected that the optimum amplification could be easily obtained by changing the reaction time with minimum consumption of reagents.

Reproducible results, however, could not be obtained due to significant molecular diffusion of the product and difficulty of completely stopping the liquid flow.

Therefore, we concluded the stopped flow method was not suitable for the current assay system.

In the continuous flow mode, two substrate solutions, a mixture of TOOS and H2O2, and an AA solution, were introduced from two inlets with the same flow rate.

In our microELISA system, the enzyme reaction time was inversely proportional to the flow rate because the volume of the reaction area was fixed.

Clearly, a slower flow rate brought a longer reaction time and higher signal intensity (Fig. 4).

It is, however, very difficult to realize a constant stable flow at low flow rate, typically slower than 0.01 µL min−1 for each solution, because of poor performance of the microsyringe pump.

Therefore we concluded that the best flow rate of each solution was 0.02 µL min−1.

Enzyme reaction temperature

Turnover of the enzyme reaction is strongly dependent on the reaction temperature.

Precise control to the optimum temperature is recommended for reliable and highly sensitive determination.

We monitored the signal intensity of the enzyme reaction product at various temperatures to get the temperature dependency of the reaction rate (Fig. 5).

Not only the reaction rate, but the thermal lens signal was also affected by the solution temperature; higher temperatures led to the higher signals.

Maximum signal intensity was obtained at 55 °C.

Since the plotted temperatures were surface temperatures of the temperature control stage of the microscope, actual solution temperature might be lower.

From this result, we concluded that the optimum temperature of the stage during the enzyme reaction was 55 °C.

Determination by MicroELISA

IFN was determined with the developed microELSA system.

The antigen-antibody reactions and the biotin-avidin reaction were done with a 10 min injection of sample solution, the biotinylated second antibody solution, and the streptavidin-peroxidase conjugate solution, respectively, with a flow rate of 1 µL min−1, followed by 30 s washing with PB at 100 µL min−1.

The enzymatic reaction was performed at 55 °C with a flow rate of 0.02 µL min−1.

The calibration curves of IFN are shown in Fig. 6.

The sample containing 0.1 ng mL−1 (6 pM) IFN had a clear TLM signal, whereas the determination limit by the microchip immunoassay using colloidal gold as a labeling material was 1 ng mL−1 (60 pM).

In the colloidal gold method, measurements of 5–10 beads and averaging were necessary for one assay, because signal intensity obtained from each bead varied greatly.

Therefore 10–60 min were required for detection of one sample, and the required time was dependent on the packing conditions of the beads and instrument operator's skill.

Furthermore, because obtained data were rather dispersed, improvement of the lower limit of determination was very difficult.

On the contrary, in the microELISA system proposed in this paper, the measured signal was derived from all the beads, and then there was very little variation during the measurement.

Hence, only 10 s was required for measurement of one sample, though a 2–3 min wait was required for the signal to reach a plateau.

In total, 35 min were required for the microELISA including three reactions and detection, while several hours or one day would be required in the conventional bulk-scale ELISA or 40–90 min in the colloidal gold-based microchip immunoassay.

Moreover, detection of the enzyme reaction product in the liquid phase with the TLM was much easier than detection of the colloidal gold fixed on the surface of the beads, and no special instrumental operating skills were necessary.

Therefore, by adoption of the enzyme, the work burden was greatly reduced.


With the microELISA system, a rapid and sensitive assay of a protein could be realized.

By using an enzyme as a labeling material, laborious detection tasks were eliminated, and assay time and data error were reduced.

Additionally no special skills were needed for the assay.

Though the microELISA was demonstrated here to be useful in a single channel assay, its merits will be fully exploited in a parallel multichannel system.

In the previous colloidal gold method, the multichannel system could realize high throughput assays due to simultaneous reactions.7

However, improvement of the total throughput of the assay was limited because of difficulties with the detection.

By adopting the ELISA in the multichannel system, measurements can be made in a lot of channels laid in parallel, simultaneously or successively in a very short time.

Therefore, throughput of the detection will be much improved.

If the detection operations become easier and no special skills are required for the determination, an automated detector becomes possible.

We expect an automated multichannel analyzer can be realized on the basis of the microELISA system in the near future.

This automated microELISA system will be a small, high throughput, and high sensitivity system and it will be suitable for ubiquitous assays including point-of care testing.