| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1 Department of Biotechnology, University of Turku, Tykistökatu 6, 20520 Turku, Finland.
2 Central Laboratory of Turku University Hospital, Kiinamyllynkatu 4-8, 20521 Turku, Finland.
aAuthor for correspondence. Fax 358-2-3338050; e-mail qiqin{at}utu.fi.
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Methods: The competitive homogeneous assay used an albumin-specific monoclonal antibody labeled with a stable fluorescent europium chelate as donor and an albumin labeled with cyanine 5 (Cy5) as acceptor. The assay was performed at room temperature in single microtitration wells that contained all the required dry-form reagents. The close proximity between the two labels in the immune complex allowed fluorescence resonance energy to be transferred from the pulse-excited europium chelate to the acceptor Cy5. The emission of long-lived energy transfer signal from the sensitized Cy5 was measured at 665 nm with time-resolved fluorometry that eliminated short-lived background.
Results: The assay procedure required 12 min for a 10-µL urine sample. The working range was from 10 to 
320 mg/L, and the lower limit of detection was 5.5 mg/L. The within- and between-run CVs were 6.9–10% and 7.5–13%, respectively. Recovery was 103–122%. The assay correlated well (r2 = 0.98; n = 37) with a laboratory-based immunoassay, although mean (SD) results were 7 (29)% lower.
Conclusions: The speed and ease of performance of this assay recommend it for near-patient use. The assay is the first to combine a fluorescence resonance energy transfer-type rapid competitive assay with an all-in-one dry reagent.
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Although antibody and antigen are in contact in an immune complex, efficient energy transfer occurs when the distance between the labeling sites on the antibody and antigen is less than R0 (the distance at which the energy transfer efficiency is 50%). This fact has allowed FRET to be exploited in immunoassays with a homogeneous format (3)(4).
Unfortunately, FRET assays that do not use time delay have limited sensitivity because of their inability to distinguish acceptor emission that is excited by energy transfer from the emission induced by direct light absorbance by the acceptor (background signal) (5)(6). The use of a long-lifetime donor fluorophore and a short-lifetime acceptor fluorophore together with pulsed-laser excitation and time-resolved detection (temporal resolution) are effective in reducing the background signal (7).
The fluorescent lanthanide chelates of europium and terbium are attractive FRET donors. Their long fluorescence lifetimes make double differentiation of the emitted light possible through spectral and temporal resolution. Apart from that, large Förster distance R0 values (up to 9 nm) can be achieved when lanthanide donors are used in combination with acceptor fluorophores such as cyanine 5 (Cy5) and XL665, which compare favorably with conventional FRET pairs based solely on organic dyes (3)(8)(9). Several time-resolved FRET assays using highly fluorescent lanthanide chelates have been reported (10)(11)(12)(13)(14).
Diabetic nephropathy is a serious microvascular complication of diabetes that leads to end-stage renal disease. It has been demonstrated that microalbuminuria, defined as an increased urinary albumin excretion (30–300 mg/24 h) above the reference interval for healthy individuals but still clinically undetectable by the usual screening tests for proteinuria, is the earliest clinical sign of diabetic nephropathy in individuals with non-insulin- or insulin-dependent diabetes (15)(16). The appearance of microalbuminuria indicates that the disease is at the stage of incipient nephropathy. In this situation, intensive care and aggressive therapies, such as restricting the intake of dietary protein, use of angiotensin-converting enzyme inhibitors, and good control of blood glucose, can slow or even reverse the progression of incipient nephropathy to overt nephropathy (15)(16). Regular monitoring of albumin excretion can therefore detect diabetic nephropathy at an early, potentially reversible stage. It is now generally agreed that all diabetic patients should have their urine tested for albumin excretion annually (16).
Urine albumin in the microalbuminuric range is currently measured by a variety of immunochemical methods, such as RIA, ELISA, immunonephelometric, and immunoturbidimetric assays (17). Because of their formats, most of these assays can be performed only in central laboratories, and they are not applicable in near-patient surroundings, e.g., at a doctor’s office. A few immunoassay-based tests for point-of-care use have appeared, such as the Bayer Clinitek 50 (18) and the DCA 2000(19). These tests provide immediate information to the doctor, allowing clinical decisions to be made at the time of the patient’s visit.
We report here the development of a rapid homogeneous assay for quantitative determination of urine albumin based on time-resolved FRET and an all-in-one, dry-reagent immunoassay concept (20).
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
reagents
The ITC-TEKES fluorescent europium chelate of 4-[2-(4-isothiocyanatophenyl)ethynyl]-2,6,-bis{[N,N-bis(carboxymethyl)-amino]methyl}pyridine was obtained from Perkin-Elmer Life Science, Wallac Oy. The monofunctional N-hydroxysuccinimide ester of Cy5 was from Amersham Life Science. Low-fluorescence single wells and Maxisorp microtitration strips (ultraviolet-quenched) were purchased from Nunc. Bovine serum albumin (BSA) was purchased from Intergen. NAP-5TM and NAP-10TM columns were from Pharmacia Biotech. Assay buffer and wash solution were prepared as described previously (21). All other chemicals were of analytical grade.
monoclonal antibody
One monoclonal antibody against human albumin was used in this study. The antibody was a kind gift of Christina Lundqvist (Medix Biochemica, Kauniainen, Finland).
proteins, metabolites, and vitamins
Several urine proteins, metabolites, and vitamins were chosen for testing for possible assay interference, including human IgG (Calbiochem); human immunoglobulin light chains 
and 
(Calbiochem); human hemoglobin (homemade); vitamins B1, B2, B6, and B12 (Sigma); vitamin C (J.T. Baker); acetone (Labscan Ltd.); creatinine (ICN); glucose (J.T. Baker); sodium nitrite (Fluka); and urea (Sigma).
calibrators
Calibrators were made from a highly purified human albumin product purchased from Sigma (cat. no. A-8763; lot no. 88H-7602). The calibrators for the FRET assay were dissolved in a normal urine pool (filtered through a 0.22 µm Millipore filter) made from six human urine samples with albumin concentrations <2.1 mg/L and had absorbance values of 0.12 at 340 nm. The normal urine pool was stored at -20 °C in aliquots. The calibrators were freshly made from a stock solution of human albumin (6.4 g/L) with the pooled urine.
labeling of antibody with the fluorescent EU3+ chelate
As described previously (22), labeling of the antibody was conducted by incubation overnight at 4 °C in 50 mmol/L sodium carbonate buffer (pH 9.6) containing a 50-fold molar excess of ITC-TEKES Eu3+ chelate. BSA and trehalose were added to the labeled antibody solution to final concentrations of 2 and 125 g/L, respectively. Under these conditions, the labeled monoclonal antibody contained 2.7 Eu3+ molecules per IgG molecule. The antibody with this labeling degree was used in the assay unless otherwise stated. Other labeling degrees were produced by varying the amount of chelate and the temperature in the protocol described above.
labeling of albumin with cy5
We added 6 mg of human albumin dissolved in 400 µL of 0.1 mol/L sodium carbonate buffer (pH 9.3) to a vial containing 1 mg of dried Cy5 and mixed the solution thoroughly. The labeling reaction was then incubated at room temperature in the dark for 30 min with additional mixing every 5 min. The labeled albumin was separated from the free dye by gel filtration on a PD-10 column, equilibrated and eluted with a buffer of 50 mmol/L Tris-HCl (pH 7.8) containing 0.15 mol/L NaCl and 0.5 g/L NaN3 (Tris-saline-azide). Fractions containing the labeled albumin were pooled, and their absorbance at 650 nm was measured. The albumin concentration was determined by a Coomassie protein assay (Pierce). Under these conditions, the labeled albumin product contained 1.6 Cy5 molecules per albumin molecule. The albumin with this labeling degree was used in the assay if not stated otherwise. Other labeling degrees were produced by varying the amount of albumin and the reaction volume in the protocol described above.
preparation of fret assay dry-reagent wells
The procedure used for preparation of dry-reagent wells for the FRET assay was essentially as described previously for heterogeneous all-in-one, dry-reagent assays (22)(23). The dried wells kept within microwell frames were stored at 4 °C in sealed airtight bags with desiccant until use.
all-in-one, dry-reagent fret assay procedure
To each FRET dry-reagent well we added 10 µL of urine-based calibrators or urine samples and 100 µL of MilliQ water (Millipore). After a 10-min incubation at room temperature with slow shaking, the fluorescence in the wells was measured in the 1420 multilabel counter (Victor) at 615 and at 655 nm. The concentrations of unknown samples were obtained by calibrating their response-normalized fluorescence signals (described below) against a calibration curve derived from the calibrator wells by the MultiCalc immunoassay program (Perkin-Elmer Life Sciences, Wallac Oy).
sample materials
We obtained 90 urine samples as part of the typical hospital routine and stored them for 1 week at 4 °C before the measurements. These 90 urine samples were used for a comparison study between two methods, i.e., the current FRET assay and an immunonephelometric assay (Dade Behring).
conditions for double-wavelength measurement
Energy transfer emission from Cy5 was measured with an emission filter at 665 nm, a delay time of 80 µs, a window time of 100 µs, and a cycling time of 2 ms. The europium chelate was excited at 340 nm with a light pulse, and its emission was measured at 615 nm with a delay time of 400 µs, a window time of 400 µs, and a cycling time of 1ms.
correction program for sample matrix interference
A program similar to one mentioned previously (12), but modified, was used to correct for interfering effects seen at the emission wavelength of the Cy5-labeled albumin attributable to absorbance of urine samples at the excitation wavelength of 340 nm. Briefly, the minimum background signal was derived from the wells that contained the Cy5-labeled albumin and buffer, whereas the maximum background signal came from the wells that contained the europium-labeled monoclonal antibody and buffer. The real background signal of each sample varied in its counts at 615 nm and could be derived from its relation to the counts at 615 nm from the wells that contained the europium-labeled antibody and buffer. After background subtraction, the FRET signal of each sample was normalized by multiplying a relevant ratio that was obtained by comparing the counts at 615 nm from the wells containing the Cy5-labeled albumin, Eu3+-labeled antibody, and buffer (i.e., B0 wells) with the counts at 615 nm from the wells containing this urine sample. The calibrators were treated in the same way as the urine samples. It was this background-subtracted, normalized signal that served as response in the FRET assay.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
125 µs longer than that of the nonsensitized acceptor Cy5 emission but shorter than that of the donor Eu3+ emission (418 µs). Therefore, in the FRET assay, the emission of the energy transfer-excited acceptor could be specifically measured with use of temporal resolution and spectral resolution that effectively reduced interference from the matrix, directly excited acceptor, and the emission of the long-lived donor.
effects of extent of conjugation with EU3+ and cy5
Experiments evaluating the effects of Eu3+ and Cy5 conjugates were performed in liquid at room temperature without shaking with incubation times of 5, 10, 30, 60, and 120 min. The 5-min data were compared with the 60-min data for assay kinetics and signal-to-background ratio. As can be seen in Table 1
, antibodies with different europium labeling degrees had a modest effect on the signal-to-background ratio. Up to a labeling degree of 5, higher labeling degrees produced higher signal-to-background ratios. With a labeling degree >5, the signal-to-background ratio was slightly decreased. In contrast, the degree of Cy-5 labeling on the albumin substantially affected the signal-to-background ratio: the lower the degree of Cy5 labeling, the higher the signal-to-background ratio. The best signal-to-background ratio was at a labeling degree of 1.6 (see Table 1
). In addition to the effects on the signal-to-background ratio, the conjugation degrees of both the donor and the acceptor affected the assay kinetics. The degree of europium labeling had a marked effect on the assay kinetics. Lower degrees of labeling produced quicker assay kinetics. The antibody with a labeling degree of 2.7 generated the quickest assay kinetics. Compared with the degree of europium labeling, the degree of Cy5 labeling had only a minor effect on assay kinetics: higher labeling degrees only slightly improved the assay kinetics. A europium labeling degree of 2.7 and a Cy5 labeling degree of 1.6 were selected for the assay because under these conditions, the assay had the fastest kinetics and a fairly high signal-to-background ratio.
|
assay kinetics
Results from the kinetic experiments showed that both the low- and high-concentration urine-based calibrators reached 90% and 100% saturation in 5 and 10 min, respectively (see Fig. 1
). Urine samples had virtually the same kinetics as the urine-based calibrators irrespective of the albumin concentrations. A 10-min incubation time at room temperature was thus chosen for the assay.
|
performance characteristics
The present FRET assay is a reagent-limited competitive assay. To perform the assay, only the sample or calibrator and water need to be added (Fig. 2
). After a 10-min incubation at room temperature with shaking, the fluorescence signals are measured at 615 and at 665 nm with time-resolved fluorometry.
|
The calibration curve for urine-based calibrators was sigmoidal, as is typical for competitive assays. The imprecision (CV) was <10% over the range from 10 to 320 mg/L (Fig. 3
).
|
The detection limit, estimated as the concentration of albumin giving a fluorescence signal equivalent to that of the mean of 20 replicates of the 0 calibrator minus 3 SD, was 5.5 mg/L.
Within- and between-run CVs were determined with three urine samples with low, medium, and high albumin concentrations. As shown in Table 2
, the within- and between-run CVs were 6.9–10% and 7.5–13%, respectively.
|
Analytical recovery was examined by analyzing urine samples containing added human albumin. The analytical recovery was 103–122% in three urine samples with low, medium and high concentrations of albumin (Table 3
).
|
Parallelism was examined by diluting the three samples with the 10 g/L BSA Tris-saline-azide buffer, and these diluted samples were measured again as new samples. Agreement between the expected and measured values was good (Fig. 4
).
|
We investigated three possible types of interference: (a) possible interference with the immune reaction; (b) pH interference in assay performance; and (c) interference from urine matrix via absorbance of excitation energy. To investigate the first type of interference, we tested the urine metabolites, proteins, and vitamins mentioned earlier, which are usually seen in nonpathologic or pathologic conditions, in the buffer-based assay system. The concentrations added were much higher than the maximum concentrations encountered in clinical practice. None of the substances significantly interfered with the assay. To investigate possible interference from pH on assay performance, we tested urine samples as well as different buffers (0.1 mol/L) with pH values from 3 to 12. When the FRET assay was performed in 0.05 mol/L Tris buffer (pH 7.8), it was more sensitive to changes in pH on the acidic side than on the basic side (data not shown), but when the assay was performed in 0.25 mol/L Tris buffer (pH 7.8), the buffer we use for the assay, we observed no interference even for some urine samples with extremely low (5.03) or high (8.59) pH values. When we evaluated the third type of interference, i.e., interference from urine matrix attributable to absorbance of excitation energy, we observed that urine samples absorbed the excitation energy at 340 nm to various degrees (Table 4
). This would certainly reduce the donor emission, with a direct effect on acceptor emission. As a result, concentrations were overestimated accordingly, as shown in Table 4
. However, the extent of this interference could be quantified by comparing the signal obtained at 615 nm from wells containing the urine sample and the Eu3+-labeled antibody with the signal obtained from wells containing the normal urine pool and the Eu3+-labeled antibody. The results obtained after the quenching correction were very close to those achieved with the immunonephelometric assay (see Table 4
).
|
comparison of the fret assay with the immunonephelometric assay
We compared the test results obtained with the FRET assay with those obtained with the immunonephelometric assay (Dade Behring) for 90 urine samples. Samples with albumin concentrations in the range 5–320 mg/L were used for correlation and difference analysis because this range was close to the working range for the FRET assay. As shown in Fig. 5A
, the results achieved with two methods correlated very well (r = 0.989) for the whole range and reasonably well (r = 0.922) at albumin concentrations <50 mg/L. The difference plot (Fig. 5B
) for the FRET assay and immunonephelometry, however, revealed a mean (SD) difference of -7 (29)%. The differences were larger at low concentrations. The difference was statistically significant (P <0.05, two-tailed t-test).
|
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Homogeneous assays can be based on different principles, such as fluorescence polarization (24), scintillation proximity (25), fluorescence quenching (26), and resonant energy transfer. Among these assay techniques, FRET is the most suitable technique for assays of macromolecules. For conventional FRET assays based solely on organic dyes, however, the lifetimes of the donor fluorophores are short, as are the emissions of the sensitized acceptors, which in turn make the assays vulnerable to interference from sample autofluorescence and from direct excitation of the acceptors. With fluorescent lanthanide chelates as the long-lifetime donors, the emitted long-lived energy transfer signals can be clearly distinguished from the direct emission of the acceptors and autofluorescence by temporal resolution. Thus, much improved signal-to-background ratios can be achieved (7)(8).
In this study, a monoclonal antibody labeled with stable fluorescent Eu3+ chelate was used as donor, and Cy5-labeled albumin was used as acceptor. The europium chelate used here has a long decay time, rather high quantum yield, and narrowly picked emission wavelengths (27). These are important for both high-efficiency energy transfer and generation of long-lived energy transfer signals. Furthermore, the europium chelate emits at 613620 nm, which allows the use of acceptors emitting in the red or near-infrared range. Cy5 has high molar absorptivity at the wavelength of the europium emission and emits red light at 665670 nm, which is different from the europium emission (28). The system thus permits the use of spectral resolution and temporal resolution for the detection of the sensitized acceptor signal. Furthermore, because the system is operated in the red spectrum, biological interference should be much lower. This is also important for the homogeneous assay because the assay mixture is not separated from the tested urine.
A competitive assay format was used in the assay. When the Eu3+-labeled antibody and the Cy5-labeled albumin are specifically combined, the close proximity between the two labels in the immunocomplex allows energy transfer to take place between the pulse-excited europium chelate and the acceptor Cy5. As a result, the europium signal is decreased (quenching), and a prolonged energy transfer signal is emitted from the sensitized Cy5. Addition of unlabeled albumin dissociates the labeled antigen-antibody complex. This leads to reduced energy transfer and a lower signal. The extent to which the energy transfer signal is reduced is dose-dependent on the unlabeled albumin.
We found that both the degree of Eu3+ labeling of antibody and the degree of Cy5 labeling of albumin had effects on the signal-to-background ratio of the assay, but the degree of Cy5 labeling obviously played a dominant role in this regard. An increase in the degree of Cy5 labeling dramatically decreased the signal-to-background ratio. This may be attributable to the formation of nonfluorescent Cy5 dimer. It was previously demonstrated that an increasing Cy5/protein ratio would lead to the formation of nonfluorescent Cy5 dimer within one protein molecule, causing a decrease of the quantum yield (28). The highest signal-to-background ratio (32.7) was obtained with albumin labeled with 1.2–1.6 Cy5 molecules/molecule of albumin. In addition, the signal-to-background ratios increased with the degree of Eu3+ labeling of the antibody up to 5 Eu3+ moieties/antibody molecule. This is also understandable because more Eu3+ moieties are at positions within energy transfer distance from the antigen-binding site as the Eu3+/antibody ratio increases. It seems that the immunoreactivity of the antibody is basically unaffected at labeling degrees <5. However, further increases in the Eu3+/antibody ratio will not only lead to some gain in energy transfer but also some loss in antibody immunoreactivity. Therefore, the final signal-to-background ratio is the result of both actions.
In addition to the effects on the signal-to-background ratio of the assay, the degree of Eu3+ labeling of the antibody and Cy5 labeling of albumin also had effects on the assay kinetics. Increasing the degree of Cy5 labeling only slightly improved the assay kinetics, whereas increasing the degree of Eu3+ labeling substantially slowed the assay kinetics. The conditions selected allow the assay to achieve the fastest assay kinetics with a high signal-to-background ratio.
The use of spectral and temporal resolution restricted the maximum background signal of the current dry assay to <1000 counts, which is comparable to many ordinary time-resolved immunofluorometric assays. Furthermore, the assay is not affected by metabolites, vitamins, and proteins usually seen in nonpathologic or pathologic conditions. However, because the signal is measured in the original solution containing urine, the urine constituents do interfere with donor excitation energy at 340 nm and perhaps also with donor emission energy around 615 nm, but the latter is believed to interfere to only a minor extent because light scattering and tissue absorbance are minimal at longer wavelengths (29). The combined effect of the interferences is to lower the quantum yield of the donor, decreasing the energy transfer signal seen at the emission wavelength of the acceptor (665 nm). The extent of the interference can, however, be accurately quantified and corrected by the methods mentioned above.
A time-resolved FRET assay is less sensitive than a heterogeneous time-resolved immunofluorometric assay (12). Moreover, a competitive assay is usually less sensitive than a noncompetitive assay (30). It is thus expected that a homogeneous assay with a noncompetitive assay format will be rather insensitive. However, when an analyte is present in the circulation or in the body fluids at high concentrations, an insensitive assay is actually required. Homogeneous FRET assays with a competitive format have reportedly been used for the measurement of serum albumin (6)(31), IgG (7), and thyroxine (32), which are all present at high concentrations in the circulation.
The usual concentrations of albumin in urine are reasonably high (0.516.2 mg/L) (33), whereas albumin concentrations in the microalbuminuric range (30–300 mg/L) (15)(34) are even higher than those in normal urine. The dry assay has a reliable working range from 10 to 320 mg/L, which covers the crucial region (15–40 mg/L) usually used as cutoff limits to distinguish normoalbuminuria and microalbuminuria (35). Furthermore, the dry-reagent-based assay allows direct use of urine samples and provides rapid test results comparable to those obtained by the other, laboratory-based assay.
Compared with conventional laboratory assays, dry-reagent-based assays are performed in single microtitration wells that contain all of the required dry-form reagents. In the protocols of dry-reagent assays, only samples and assay solution are added to the wells to start the immune reaction (20)(22)(23). The performance data shown here were obtained by manually performing the dry-reagent assay. The assay variation (CV) was relatively high but still within the range of acceptability. Because the urine albumin excretion rate can vary highly (by up to 40%) from day to day (36), an analytical CV equal to or less than one-half of the intraindividual biological variation (<18%) would be considered as desirable performance (37). In practical settings, the dry-reagent assay will be fully automated, which will help reduce the assay variation observed here.
Because of wide variations in the urinary albumin excretion rate during the day and from day to day, caution should be exercised when measuring urine albumin. The appropriate specimens for the assay are a timed specimen and a random specimen if the human urine albumin value of the latter is corrected by the creatinine concentration of the same sample. The diagnosis of microalbuminuria requires results from at least two urine collections over months (38).
In conclusion, the assay presented here appears to be suitable for the measurement of urine albumin in near-patient settings to screen for microalbuminuria in diabetic and hypertensive patients (39). The assay is easy to perform and provides results comparable to those of an existing laboratory immunoassay. The rapid and quantitative test results obtained directly from urine samples could allow clinical decisions to be made at the time of the patient’s clinical visit, potentially providing economic savings and better clinical care.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
assay using fluorescence resonance energy transfer. Cytokine 1998;10:495-499.[CrossRef][ISI][Medline]
[Order article via Infotrieve]
The following articles in journals at HighWire Press have cited this article:
|
|
 |
|
  J. R. Lakowicz, J. Malicka, E. Matveeva, I. Gryczynski, and Z. Gryczynski Plasmonic Technology: Novel Approach to Ultrasensitive Immunoassays Clin. Chem., October 1, 2005; 51(10): 1914 - 1922. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
, low calibrator; 
, high calibrator; 
, low-albumin urine; 
, medium-albumin urine; 
, high-albumin urine.
, Eu3+-labeled antibody; 
, Cys-labeled human albumin; 
, endogenous human albumin.
