Jul 22, 1985 - George C. Saunders, James H. Jett,and John C. Martin .... z. CLINICALCHEMISTRY,Vol. 31, No. 12, 1985 2021 resuspended (0.5 mL/L) in BSA ...
CLIN. CHEM. 31/12, 2020-2023 (1985)
Amplified Flow-CytometricSeparation-Free Fluorescence Immunoassays George C. Saunders, James H. Jett, and John C. Martin An equilibrium-type competitive-binding fluorescence immunoassay with high sensitMty and excellent precision is described that obviates separation of free from bound label. In the assay relatively large (10 m diameter) antibodycoated non-fluorescent particles and very small (0.10 m diameter) antigen-coated fluorescent latex particles are used. Soluble nonlabeled antigen competes with antigen on the microspheres for antibody binding sites on the larger spheres. After equilibrium is attained, the fluorescence distribution of 5000 of the large spheres is measured in a flow cytometer. The mean values for the fluorescence distribution obtained from samples containing known concentrations of soluble antigen are used to construct a standard displacement curve. In a prototype assay for the antigen horseradish peroxidase, a sensitivity of 10-12 moIfL has been achieved. Undiluted serum can be assayed without loss of sensitivity. Preliminary experiments also indicate that double-antibody “sandwich”-type assays of very high sensitivity (10_14 mol/L) are also possible when this dual-sphere concept is exploited. AdditIonal Keyphrases: nique fluorometry
variationof the immunofluorometric techprototypetechnique
Materials and Methods Reagents Horseradish peroxidase (HRP;EC 1.11.1.7),Type VI, and serum albumin (BSA) were purchased from Sigma Chemical Co., St. Leuis, MO. Rabbit antibody (IgG fraction) toHRP was from Cappel Labs., Cochranville, PA. Uniform (10 pm in diameter, SD 0.30 j.tm) polystyrene latex particles were from Dow Chemical Company, Indianapolis, IN. Labeled green fluorescent (Xem 510 nm) polystyrene latex particles, 0.1 (SD 0.008) pm in diameter, were from Polysciences, Inc., Warrungton, PA. Rabbit serum was from Kappa Scientific Co., Escondido, CA. The standard buffer (BSA buffer) used in the assay contained 100 mmol of KH2PO4 and 50 g of bovine serum albumin per liter. The pH was adjusted to 8.2 with 100 mmol of K2HPO4 per liter. NaN3 (100 mgfL) was added as preservative. The complete buffer was ifitered through a 0.22-pm (av pore size) membrane filter. The coating buffer used to adsorb both HRP antibody and HRP to the latex particles contained 10 mmol NaHCO3 per liter; the pH was adjusted to 9.6 with 10 mmol/L Na2CO3. bovine
Apparatus Until recently, fluorescence-based immunoassays have lacked the sensitivity needed for quantifying numerous clinically important analytes. Currently there are at least two approaches that may solve this problem. In one, based on time-resolved fluorometry (1,2), the sample, containing a probe with a long lifetime such as Th3, is excited with a short light-pulse and measurement of fluorescence is started after enough time has elapsed for the background (as in serum) to decay to zero. A sensitivity of 10b0 mol/L for human immunoglobulin G (IgG) has been obtained (3).1 The other approach is a “sandwich”-type fluorescence immunoassay in which antibody-coated microspheres, sample, and soluble fluorescent antibody are reacted together as in a conventional “sandwich” immunoassay. Laser flow-cytometry obviates separation and washing steps. The gating of fluorescence on scattered light pulses allows the selective measurement of particle-associated fluorescence. A prototype assay for human IgG, performed in a 10-fold diluted serum matrix, achieved a sensitivity of 6 x 10_li mol/L (4). We have used a novel concept which takes advantage of both flow cytometi-y (5) and amplified immunomicrosphere technologies (6). Our approach is to use relatively large nonfluorescent microspheres (10 pm in diameter) as an antibody carrier and very small fluorescent microspheres (0.10 pm in diameter) as either an antigen carrier (for a competitive binding assay) or an antibody carrier (for a “sandwich”type assay). The use of this amplified fluorescent carrier has resulted in exquisite sensitivity in prototype competitive binding (10-12 mol/L) and “sandwich” (10’ mol/L) assays. Experimental Pathology Group, Life Sciences Division, Los Alamos National Laboratory, Los Alamos, NM 87545. ‘Nonstandard abbreviations: IgG, immunoglobulin G; Ab, antibody; Ag, antigen; BSA, bovine serum albumin; HRP, horseradish
peroxidase. Received July 22, 1985; accepted August 30, 1985. 2020 CLINICALCHEMISTRY, Vol. 31, No. 12, 1985
Microsphere fluorescence was measured with a modified Becton-Dickinson FACS H cell sorter/flow cytometer. The fluidics and light-collection components were used as supplied (Becton-Dickinson, Immunocytometry Systems, Mountain View, CA). The flow-cell exit orifice was 60 min diameter. The major modification was in the electronics and
data-acquisition system. The fluorescence emitted by the microspheres passing through the 457-nm laser beam was filtered by two long-passifiters (510 nm dielectric and 520 nm colored glass). Fluorescence signals detected by the photomultiplier tube were amplified and integrated by an active integrator circuit (7). Forward-scattered light was detected with the standard FACS II configuration. The light scatter and integrated fluorescence signals were digitized andstored by using a computer-based data-acquisition and analysis system (8, 9).
Procedures Preparation of immunobead reagents: The 10-pm particles were washed four times by centrifugation (1000 x g, 5 mm) in coating buffer. To 5 x io washed 10-pm particles we added 1 mL of a 100-fold dilution (in coatingbuffer) of HRPAb (containing 24 pg of specific HRP-Ab in a total of 158 pg of IgG). After 24 h at 4#{176}C, the particles were washed three times in coating buffer, then resuspended to a density of 2 x i06/mL in BSA buffer. Tubes containing 175 L of a 25 milL suspension of the fluorescent 0.10-pm particles were serially centrifuged (40000 x g, 10 mm, in an Airfuge; Beckman Instrument Co., Palo Alto, CA). After the fourth wash, the particles were resuspended (10 mL/L) in coating buffer containing 1 mg of HRP or 100 pg of HRP-Ab per milliliter. The resuspended particles were sonicated for 1 mm in an 80-W Model B-12 ultrasonic cleaner (Branson, Shelton, CT). After 24 h at 4 #{176}C the particles were again washed in coating buffer three times in the Airfuge, then
resuspended (0.5 mL/L) in BSA buffer and sonicated until a 10-FL aliquot of the suspensionappearedas a homogeneous green glow when viewed under a fluorescence microscope. This sonication process was performed before each use of the immunobead reagent. Ab and Ag immunobead reagents, so prepared, were stable for at least one year.
Generation of a Standard Competitive-Binding Displacement
Curve
We added 1 mL of BSA buffer to eachof a series of 12 x 75 mm polypropylene tubes (Falcon, no. 2063) at least 30 miii before preparing standard HRP dilutions. Serial log dilutions of HRP were made from a i0 mol/L (0.4 mg/mL) HRP stock solution. One tube in each of the replicate series received no HRP. Then, io (in 50 pL) of the 10-pmdiameter Ab-coated microspheres were addedto each tube. The reaction between I{RP and Ab bound to the large spheres was allowed to take place at room temperature. After 12 h, 30 L of a 50 pg/L suspension of the 0.1-pmdiameter HRP-coated fluorescentparticles (approximately 3 x 10) was added to each tube and incubated for another 12 h at room temperature. (In some experiments we used a single-step assay of 14 h.) Without separating free from bound l{RP-coated particles, we analyzed the large spheres in the suspension for both light scatter and fluorescence in the modified Becton-Dickinson FACS U cell sorter. The light-scatter signal from the larger spheres was used as a trigger for data acquisition. Fluorescence was measured for at least 5000 of the larger particles from each tube. The fluorescence measurement quantifies the amount of smaller Ag-coated particles bound to the large particles. To set the instrument we used a tube containing only Ab- and Agcoated particles (no soluble HRP) and the fluorescence gain was adjusted such that the linear distribution for this maximum fluorescence sample was centered between channels 160 and 190 in a 256-channel distribution. The mean of this distribution was used as the maximum (100%) fluorescence. With no change in instrument settings or sample flow rate, data for all tubes in the series were then recorded. All data were analyzed by using a multiparameter list-modeanalysis computer program in which only the fluorescence due to single (as compared with doublets and larger aggregates or debris) 10-pm particles was displayed (Figure 1). Values for the mean fluorescence intensity of each tube were converted to percent maximum fluorescence, and this value was plotted as a function of HRP concentration to obtain a standard displacement curve.
I
Generation of a Standard Curve for the SandwichType Assay We added iO HRP-Ab-coated 10-pm microspheres, in 50 L, to tubes containing known concentrations of HEP diluted in BSA buffer. After 7 h at room temperature, 50 L of a 50 mgfL suspension of fluorescent 0.10-pm HRP-Ab labeled microspheres (about 5 x 10) was added to each reaction tube. After a 12-h incubation at room temperature, 5000 of the large spheres (again without separating free from bound label) were analyzed as outlined above. We plotted the mean channel number of the linear fluorescence distribution for each tube as a function of HRP concentration. The fluorescence obtained was directly (linearly) proportional to HRP concentration.
Results and Discussion Preliminary experiments showed that nonspecific bead interactions couldbe prevented by using buffers containing high concentrations of protein (at least 20 g/L) or whole serum, or both. To generate the data for the prototype
V) I-
z w >
Li lj 0
Li
z
CHANNEL
NUMBER
Fig. 1. A. Light scatter [depicting relative size (dotted fine)]and linear fluorescence (solid fine) raw data obtainedfrom an assay sample to which no solubleHAP was added Bindingoffluorescent 0.1O-m HAP-coatedpailicles is maidmumin thissanle. Thepeak(brec*ets) In the light-scatter distribution at aboutchannel95 is due to Individual1O-m pailicles.Signalsto the left ofthis distributionrepresentsmall particulatedebrispresentin the sample; signalsto the right (centeredat about channel180)represent10-nmspheredoublets B. Reprocesseddata obtainedfrom the rawdatainAg. lAin whichthe fluorescencedue onlyto the 10-an particles(brackets)is shown The mean channel value of this fluorescencepeak is taken as the maidmum (100%)
fluorescence
C. Reprocessedfluorescencedata from a sample thatcontaIned 10b0
mol of soluble HAP per liter As cai, be seen, compared to
FIg. 18, the fluorescence peak has shifted
markedlytothe left, owing to bindingof unlabeledsoluble HPR
system presented herein, we chosea 50 gIL BSA buffer. In
one experiment we used whole rabbit serum. The ratio of fluorescentAg-coated particles to Ab-coated particles in a 1mL reaction volume that yielded a 60% mpximum saturation (in the absenceoffree Ag) at equilibrium was found to be 30000:1. The time required for equilibrium to occurat this ratio was 12 to 14 h. The binding of Ag-coated small particles to Ab-coatedlarge particles is illustrated in Figure 2. Figure 3a illustrates the shape and dynamic range of twostep displacement curves. The sensitivity of the assay has consistently been 10- 12 mol/L, which for the 1-mL reaction volume we used translates to 10-15 mol of analyte. Singlestep and two-step displacement curves are compared in Figure 3b. Although the two-step processes reportedly (10) increase the sensitivity of competitive-binding radioimmunoassays of protein by factors two- to fourfold, we saw virtually no difference in sensitivity between these two types of assay procedures. Figure 3c illustrates both reagent stability and assay performance when whole rabbit serum (instead of BSA buffer) is used as the vehicle for the HRP dilutions. This two-step curve, generated with immunobead reagents prepared one year previously,is practically identical in shape to those in Figures 3a and b. The dynamic range CLINICALCHEMISTRY,Vol. 31, No. 12, 1985 2021
Table 1. Assay Precision Intraday(n HRP, mol/L
Intorday (n = 4)
= 4)
SD
CV, %
ya
1.52 1.97 3.96 8.40
93.86 75.28 42.38 21.51
10_12
90.57
1.38
10_1I
73.58
1.45
10 10 1O
42.50 22.30
1.68 1.87
SD
CV, %
2.89 3.19 0.34 0.69
3.08 4.24 0.79 3.19
Percent of maximum fluorescence.
Fig.2. Scanning electron micrograph illustratingthe bindingof HAPcoated 0.25-pm particles to HAP-Ab-coated 10-pm particles in the presenceof 10- #{176}mol of solubleHAP per liter HRP-coated 0.25-smparticleswere usedtoobtainthis pictureonly because they givegreater photographicdefinitionthan do 0.1-urn particles 100
A
#{149}
B
U
feature of this assay; there are fewer than 20 free 0.1-/.Ini diameter spheres present in the probe volume at any time. The small probe volume should also diminish or eliminate problems caused by sample contaminants such as bilirubin, which because of their spectral properties reportedly interfere with fluoroimmunoassays (12). Also, ac coupling of thE fluorescence-amplifying electronics, together with the abii. ty to trigger on the light scatter of the large sphere, allowt the detection electronics to ignore the small dc fluorescence originating from the steady-state concentration of free beade and other fluorescent molecules, and to record only that associated with the relatively large (10-pm) particles at they pass through the probe volume. The absence of a separation step both facilitates the mechanics of the assay and improves precision because of the maintenance ol equilibrium conditions throughout the measurement proc. ess.
The relatively long(12 h) incubationtimes in the two-step
H
100
C
_______ s
0
75
0 a
50
U
a-
25
0 102
io’#{176} io’ HRP
Concontration
102
10’#{176}10’
(mol/L)
Fig. 3. A. Two-stepHAPdisplacement curve coveting the concentrationrangeiO to 10 13 moVL Thevolumeof thesampleanalyzedin the flow cytometer is