548 NATURE BIOTECHNOLOGY VOL 18 MAY 2000 http://biotech.nature.com

RESEARCH ARTICLES

Simplifying assay protocols and reducing reagent use are importantconsiderations in developing new formats and technologies forbioaffinity measurements. Whereas simplifying protocols reducesthe manpower and the extent of automation required, reducingreagent consumption reduces costs and bioactive and packing mate-rial waste, while allowing better control of temperature levels inapplications such as PCR1. In traditional nonisotopic assay methodsthe sensitivity is almost linearly dependent on the measurement(and reaction) volume. Thus, the applicability of reduced reactionvolumes is limited by reduced signal, particularly when using elec-trochemiluminescence2 or time-resolved fluorescence3 with volumesof less than 10 µl.

To obtain a good signal-to-background ratio from microvol-umes, new microscopic techniques such as confocal microscope4,5

and fluorescence correlation spectroscopy (FCS), in both its con-focal single-photon excitation6 and two-photon excitation7 ver-sions, are promising technologies. Two-photon systems can sup-press background effectively by optical filtering, whereas confocalsystems use a pinhole to reduce unwanted background signals. Intwo-photon excitation, the excitation of fluorescent moleculestakes place only within the three-dimensional focal volume8,9. Incontrast to single-photon fluorescence systems, background sig-nals from the optical system and from the sample outside the focalvolume are not visible in the fluorescence channel of a two-pho-ton system9,10.

One of the main consequences of reduced observation volumes isthe reduced number of molecules that can be observed at any giventime. With microscopic techniques this often leads to compromisesbetween sensitivity and measuring times. Thus, special techniquessuch as spatial scanning of the sample11 are needed for detectingmolecules.

Here we have combined two-photon excitation with biochemical-ly activated 3 µm polystyrene microparticles as a solid phase to bindthe biomolecules to be studied. Each microparticle acts as a local con-centrator of the biomolecules. When a fluorescent biospecific tracer

molecule attaches to the microparticle, the degree of binding is mea-sured by observing the two-photon fluorescence signal from individ-ual microparticles. The actual assay format may be competitive orimmunometric.

ResultsTo perform the assay, the analyte and the reagent solution, composedof microparticles and a fluorescent tracer, are dispensed into a singlereaction volume. After incubation, the signal from the microparticlesis measured directly within the reaction volume. As a result of theeffective concentrating of biomolecules on the microparticles, the sig-nal from the tracer bound to each microparticle at the maximumcapacity is orders of magnitude stronger than the signal backgroundfrom the unbound tracer. In fact, the unbound tracer signal at the ana-lyte zero dose level determines the lowest limit of the working range inan assay.

The working range and sensitivity depend on the assay parame-ters: the affinity, the microparticle binding capacity, the number ofmicroparticles in an assay, the tracer concentration, and nonspecificbinding. Theoretically the dynamic range—that is, the ratio betweenthe lowest and highest measurable concentration—of four orders ofmagnitude can be reached. Consequently, no separation steps areneeded. This simple protocol is well suited for microvolume systems.In fact the whole assay could be performed even on a singlemicroparticle.

The prototype instrumental set-up is shown in Figure 1. Thereaction suspension is placed on top of the objective lens in a specialpolystyrene cuvette having a thin 0.2 mm transparent bottom. The0.75 numerical aperture (NA) objective focuses the 1.064 µm lightfrom the microchip laser to a diffraction-limited two-photon excita-tion volume of about 1 fl. A lateral beam scanning mechanism isused to locate the microparticles. A slow axial (depth) scanning (0.3Hz and 150 µm amplitude) is used to increase the total suspensionobservation volume to about 10 nl. When a microparticle appears inthe focal volume, the confocally arranged scattering detector detects

A new microvolume technique for bioaffinity assays using

two-photon excitationPekka Hänninen*, Aleksi Soini, Niko Meltola, Juhani Soini, Jori Soukka, and Erkki Soini

Laboratory of Biophysics, Institute of Biomedicine, University of Turku, P.O. Box 123, 20521 Turku, Finland. *Corresponding author (pekka.hanninen@utu.fi).

Received 12 October 1999; accepted 10 February 2000

Bioaffinity binding assays such as the immunoassay are widely used in life science research. In animmunoassay, specific antibodies are used to bind target molecules in the sample, and quantification of thebinding reaction reveals the amount of the target molecules. Here we present a method to measurebioaffinity assays using the two-photon excitation of fluorescence. In this method, microparticles are usedas solid phase in binding the target molecules. The degree of binding is then quantified from individualmicroparticles by use of two photon excitation of fluorescence. We demonstrated the effectiveness of themethod using the human α-fetoprotein (AFP) immunoassay, which is used to detect fetal disorders. Thesensitivity and dynamic range we obtained with this assay indicate that this method can provide a cost-effective and simple way to measure various biomolecules in solution for research and clinical applications.

Keywords: two-photon excitation, immunoassay, microvolume detection, optical trapping, microchip laser

© 2000 Nature America Inc. • http://biotech.nature.com©

200

0 N

atu

re A

mer

ica

Inc.

• h

ttp

://b

iote

ch.n

atu

re.c

om

NATURE BIOTECHNOLOGY VOL 18 MAY 2000 http://biotech.nature.com 549

RESEARCH ARTICLES

the particle. Once the scattering signal exceeds a preset detectionthreshold, the lateral scanning mechanism is stopped. The fluores-cence measurement is then activated without interrupting the slowaxial scanning.

During the measurement, optical forces caused by the intense two-photon excitation laser beam and its refraction by the microparti-cle12,13 draw the particle to the lateral center of the focus and slowly, inabout 50 ms, push the microparticle along the optical axis out of thefocal volume. The optical forces caused by the laser beam also over-come the forces caused by viscous “drag” of the slow axial motion ofthe focus. As the microparticle leaves the focal volume, the fluores-cence measurement ceases. The fast lateral scanning is reactivatedafter the signal processing has determined that the microparticle hastruly vanished from the focal volume. A typical microparticle mea-surement, with the associated detection threshold level and the conse-quent fluorescence integration window, is shown in Figure 2. Becauseof the slight optical differences in the scattering and fluorescencechannels, the vanishing microparticle is apparent in the fluorescencechannel for a longer interval than in the confocal scattering channel.

Variation in signal from particle to particle is reduced by the opti-cal trapping of the microparticles to the lateral center of the focus.Since the diameter of the diffraction-limited focusing beam is small(<1 µm), as compared to the microparticle size of 3 µm, the signal isproportional to the bound tracer surface density rather than to awhole-particle signal, making it less variable due to differences inmicropartical size and position.

To demonstrate the function of the recording system, we mea-sured individual particle intensity distributions of flow cytometryintensity calibration standards (2.5 µm through-colored LinearFlowGreen Cytometry Intensity Calibration Kit; Molecular Probes,Eugene, OR) and 3 µm α-fetoprotein (AFP) assay particles, asdescribed in Experimental Protocol. The assay particles were repre-sentative for surface-colored particles. The 2% and 12% relative-intensity standard particles were selected for our comparison becauseof their suitable intensity. The acquired individual particle intensitydistributions are plotted in Figure 3. With our instrument the indi-vidual-particle coefficient of variation (CV) for the through-colored

standard particles was determined to be 40% (Fig. 3, curve A) and30% (curve C), respectively. The individual-particle signal CV forAFP assay particles (curve B) was determined to be 20%.

The acquired intensity distributions were compared with intensi-ty distributions recorded with a flow cytometer (FACScan; BecktonDickinson Immunocytometry Systems, San Jose, CA). The through-colored standard particles measured in a conventional flow cytome-ter showed CVs of 20% and 15%, respectively, in the green fluores-cence channel of the instrument (data not shown). The AFP assayparticles, in contrast, exhibited a CV of 35%. Since the flow cytome-ter samples uniformly the full fluorescence of the particles, this resultis a clear indication that our instrument indeed measures micropar-ticle surface densities rather than the whole microparticle.

As a model assay to test our concept, we chose the human AFPsandwich-type immunoassay. Nominally 3.1 µm amino-modifiedpolystyrene microparticles (Bangs Laboratories, Fishers, IN) werecovalently coated with fragments (Fab’) of monoclonal mouse anti-body (MAb) anti-AFP (in-house material) and passively post-coatedwith bovine serum albumin (BSA). As a tracer we used a whole mon-oclonal MAb anti-AFP (in-house material) against another epitopeand labeled with dipyrrometheneboron difluoride dye (in-house

Figure 1. Optical set-up of the instrument. The 1.064 µm, 75 mWpassively Q-switched laser beam of 1 ns pulses at 27 kHz rate(Nanolase TP7M70, Meylan, France) is reflected (>90%) with a customdichroic mirror (DM; UAB Standa, Vilnius, Lithuania) through an opticalscanner unit and focused with an objective lens (Leica C-Plan 63x/0.75,Leica Microsystems, Bensheim, Germany) to the cuvette. The scattered(reflected) light from the microparticles is directed to a confocallyarranged pinhole (PH) in front of the photodiode-detector by thedichroica mirror and a 2% beam-splitting window (BS). Oncemicroparticle scattering has been detected at 1.064 µm, thefluorescence signal measurement is activated. The two-photon excitedfluorescence signal through the dichroic mirror (DM, >70%transmission at 530–700 nm) and the filter (F, >85% transmission at 530-650 nm, UAB Standa, Vilnius, Lithuania) is detected with thephotomultiplier tube (PMT, R 6358P; Hamamatsu Photonics KK, Japan).

*

Microparticledetector

Laser

0.75 NALens

ScannerPM

T

F

Cuvette

DM PH

BS

Focus-scanner

Figure 3. Intensity distribution histograms of three different particlecategories. The measurements of the LinearFlow 2% (curve A) and12% (curve C) green intensity standards are plotted as dashed lines.The solid-line curve (B) presents the intensity distribution histogramof the AFP assay particles at 1,000 ng ml-1 concentration level. The CVof the measurements was determined to be 40% and 30% for theLinearFlow particles and 20% for the AFP assay particles.

0

10

20

30

40

50

60

70

80

10 100 1000 10000

Intensity

Num

ber

of b

eads

Figure 2. A single-particle signal trace at 400 ng ml-1 of AFP from theconfocal scattering channel (black) and fluorescence channel (gray).A particle is detected as the signal from the scattering channel risesabove the particle threshold level (dashed line); at this point thescanners are stopped. Fluorescence is integrated in the countingwindow as long as the particle-counting threshold level is exceeded.Once the scattering signal drops below the preset particle thresholdlevel, the scanning is reactivated.

-0.4

-0.2

0

0.2

0.4

0.6

0.8

1

1.2

0 20 40 60 80 100 120 140 160

Time (ms)

Sign

al a

mpl

itude

(V

)

<----- COUNTING WINDOW ----->

Particle threshold

level

AB

C

© 2000 Nature America Inc. • http://biotech.nature.com©

200

0 N

atu

re A

mer

ica

Inc.

• h

ttp

://b

iote

ch.n

atu

re.c

om

550 NATURE BIOTECHNOLOGY VOL 18 MAY 2000 http://biotech.nature.com

RESEARCH ARTICLES

material) with single-photon excitation maximum at 530 nm. TheAFP standard was obtained from Dako A/S (Glostrup, Denmark). Arelatively large reaction and measuring volume of 25 µl was chosen tominimize the pipetting errors. The assay was performed as shown inthe Experimental Protocol section and the resulting standard curvewith the background subtracted is plotted in Figure 4. Each standardcurve point represents a normalized integral signal from about 200microparticle occurrences during 60 s measuring time. The CV of thesignal at the blank level was as good as 2%. The sensitivity of 0.1 ng ml-

1 was obtained by relating the assay response to the usual 3σ (s.d.) levelof the blank sample signal. The maximum signal response wasreached at a 400 ng ml-1 concentration level, yielding a dynamic rangeof 4,000 for this assay. From the earlier measurement of individual-particle CV of 20%, and assuming Poisson distribution of particledata, we can conclude that each particle was measured at least twotimes during the measurement cycle.

To confirm that the results of the 25 µl volume can be directlyscaled to smaller volumes, at least with regards to instrumentation,we compared signals obtained from the 25 µl and 2 µl volumes. The 2µl volume was taken into a special capillary insert to enable the mea-surement. The signal levels for 2 µl volume were identical within theexperimental error, as shown in Figure 4.

DiscussionOur experiments show that bioaffinity assays can be performed withgood sensitivity and dynamic range using two-photon fluorescenceexcitation. Although our demonstration is performed using animmunoassay, we believe that a wide range of applications for ourmethod can be found in various fields such as DNA hybridizationand high-throughput screening in drug discovery studies. Examplesof using microparticles in different applications can also be found inthe literature14.

The use of microparticles as the solid phase makes microassayformats practical and cost-effective as compared to coated-tubetechnology. Since this technique measures microparticles individu-ally, multiparameter tests can be made by use of a mix of microparti-cles with different biospecific coatings. The identification of themicroparticle categories can be made by an internal fluorescenceindicator individual to each microparticle category15.

The infrared 1.064 µm two-photon excitation laser beam is notabsorbed by biological material9, raising the expectation that mea-surements directly from complex biological matrices such as whole

blood could also be possible; we have also confirmed this in our pre-liminary tests.

Other bioactive carriers such as cells or fragments of cells couldalso be used in place of the activated microparticles, broadening theapplication field even further. Since the measurement takes placedirectly within the reaction solution, the described method is alsosuitable for real-time monitoring of reaction kinetics. Recent devel-opments in the fields of micromechanics, microfluidics, andmicrochip lasers suggest that two-photon excitation could be inte-grated into miniaturized systems for measuring bioaffinity reactionsin vitro.

Experimental protocolParticles. 3.1 µm amino-modified polystyrene microspheres (AminoModified Microspheres PA05N; Bangs Laboratories, Inc., Fishers, IN) werecovalently coated with mouse monoclonal anti-AFP IgG Fab’ fragments (in-house material; FujiRebio Inc., Tokyo, Japan) using heterobifunctional ε-maleimidocaproyloxysuccinimide linker agent.

Tracer. Mouse monoclonal anti-AFP IgG (in-house material; FujiRebio Inc.)was labeled with dipyrrometheneboron difluoride dye (in-house material).

AFP Standard. AFP standard (Dako A/S, Glostrup, Denmark) was dilutedwith assay buffer to eight different concentrations: 0.1, 0.5, 2, 20, 100, 400,1,000, and 2,000 ng ml-1. Pure assay buffer was used for zero control.

The particles and tracer were mixed into a single reagent suspension. Theparticle content of the reagent suspension was 0.08% (wt) and tracer concen-tration 6.6 nM.

AFP Assay protocol. Into each cuvette were placed 15 µl reagent suspen-sion, after which respective AFP standards were added to each cuvette. Allcuvettes were vortexed and incubated for 1h in 37°C. After vortexing again,measurements were made directly from the reaction cuvette.

AcknowledgmentsThis work was supported by the Academy of Finland and National TechnologyDevelopment Agency, TEKES. Special thanks to Mr. Honda of Fujirebio Inc.(Tokyo, Japan) and Mr. Nasu and Mr. Sakurai of SRL Inc. (Tokyo, Japan) forsupplying, preparing, and characterizing the material for AFP assay.

1. Wittwer, C.T., Fillmore, G.C. & Garling, D.J. Minimizing the time required forDNA amplification by efficient heat transfer to small samples. Anal. Biochem.186, 328–331 (1990).

2. Blackburn, G.F. et al. Electroluminescence detection for development ofimmunoassay and DNA probe assays for clinical diagnosis. Clin. Chem. 37,1534–1539 (1991).

3. Siitari H., Hemmilä, I., Soini, E., Lövgren T. & Koistinen, V. Detection of HBsAgusing time-resolved fluoroimmunoassay. Nature 301, 258–260 (1983).

4. Ekins, R.P. & Chu, F.W. Multianalyte microspot immunoassay—microanalytical“compact disk” of the future. Clin. Chem. 37, 1955–1967 (1991).

5. Swartzman, E.E., Miraglia, S.J., Mellentin-Michelotti, J., Ebangelista, L. &Yan, P. A homogeneous and multiplexed immunoassay for high throughputscreening using fluorometric microvolume assay technology. Anal. Biochem.271, 143–151 (1999).

6. Eigen, M. & Riegler, R. Sorting single molecules: application to diagnosticsand evolutionary biotechnology. Proc. Natl. Acad. Sci. USA 91, 5740–5747(1994).

7. Berland K.M., So, P.T.C. & Gratton, E. Two-photon fluorescence correlationspectroscopy: method and application to the intracellular environment.Biophys. J. 68, 694–701 (1995).

8. Denk, W., Strickler, J.H. & Webb, W.W. Two-photon laser scanning fluores-cence microscopy. Science 248, 73–76 (1990).

9. Xu, C. & Webb, W.W. Multiphoton Excitation of Molecular Fluorophores andNonlinear Laser Microscopy. In Topics in fluorescence spectroscopy, Vol. 5.Nonlinear and two-photon induced fluorescence. (ed. Lakowicz, J.R) 501–515(Plenum Press, New York, NY; 1997).

10. Lytle, F.E., Dinkel, D.M. & Gisher, W.G. Trace-level quantitation via time-resolved two-photon-excited fluorescence. Appl. Spec. 47, 2002–2006 (1993).

11. Koppel, D.E., Morgan, M., Cowan, A.E. & Carson, J.H.: Scanning concentra-tion correlation spectroscopy using the confocal laser microscope. Biophys.J. 66, 502–507 (1994).

12. Ashkin, A, Dziedzic, J.M., Bjorkholm, J.E. & Chu, S. Observation of a single-beam gradient force optical trap for dielectric particles. Opt. Lett. 11,288–290 (1986).

13. Hänninen, P.E., Schrader, M., Soini, E. & Hell, S.W. Two-photon excitation flu-orescence microscopy using a semiconductor laser. Bioimaging 3, 70–75(1995).

14. Cook, L. & Irving, D. Microsphere-based flow cytometric assays. J. Clin.Immunoassay 12, 36–39 (1989).

15. Streefkerk, J.G., Kors, N. & Boden, D. Principle of a reaction for simultaneousdetection of various antibodies using coloured antigen-coupled agarosebeads. Protides of the Biological Fluids 24, 811–814 (1976).

Figure 4. AFP background created response curves for a separation-free single-step assay measured in two volumes: 25 µl (G) and 2 µl (p).Measurements were made directly in the reaction curvette. Eachmeasuring point represents a normalized integral signal from about200 microparticle occurrences in the measurement time of 60 s.

© 2000 Nature America Inc. • http://biotech.nature.com©

200

0 N

atu

re A

mer

ica

Inc.

• h

ttp

://b

iote

ch.n

atu

re.c

om