|Home | About | Journals | Submit | Contact Us | Français|
Brucellosis is a globally significant zoonosis, the control of which is difficult and resource intensive. Serological tests form a vital part of a multifactorial approach to control and are often performed in large numbers. The aim of the present study was to develop a new assay to improve the efficiency, ease, and effectiveness of serological testing. An existing competitive enzyme-linked immunosorbent assay (cELISA) was adapted to a completely homogeneous time-resolved fluorescent resonance energy transfer (TR-FRET) assay. This was achieved by labeling an anti-Brucella monoclonal antibody with a long-lifetime donor fluorophore and Brucella smooth lipopolysaccharide with a compatible acceptor and optimizing the reading conditions. The assay was performed in a 96-well plate with a single 30-min incubation period and no separation (wash) steps and was concluded by a single plate-reading step. The performance of the assay was evaluated with a panel of serum samples from infected (n = 73) and uninfected (n = 480) sources and compared to the performance of the parent cELISA, an indirect ELISA (iELISA), and fluorescence polarization assay (FPA). The performance of the TR-FRET assay matched the performance of the iELISA, which had 100% diagnostic sensitivity and specificity, and surpassed the performance of the cELISA and the FPA. The results also demonstrated that the TR-FRET technique is effective with poor-quality serum samples from the field. To the knowledge of the authors, this is the first homogeneous TR-FRET assay to detect antibodies raised against an infectious disease. The technique appears to be sufficiently adaptable to meet the needs of many other similar testing requirements to identify infectious diseases.
Brucellosis is a zoonosis of widespread distribution and significance caused by species of the genus Brucella. The disease is known to be especially prevalent in the Middle East and Mediterranean Basin (33), and there are disturbing signs of its reemergence across large areas of the globe, especially in central Asia. Few countries have successfully managed to eradicate the disease. In many of these countries, although the livestock sector is disease free, a significant wildlife reservoir remains and presents a risk of reintroduction (11).
The principle etiologic agents of brucellosis are the classical smooth species Brucella abortus, B. melitensis, and B. suis. These species all have smooth lipopolysaccharide (sLPS), which is a major virulence factor (17, 35). Each of these species has a preferred host; however, many animals, including humans, are susceptible to each of the classical species, although the underlying mechanisms for host preference are not clearly understood (5). Brucellosis in ruminants is mainly manifested by reproductive failure due to abortion. There are few other clinical signs, and this causes difficulties with diagnosis. In humans, the disease is mainly presented as an undulating febrile condition, although there may be other, more serious complications. The symptoms of human brucellosis are particularly nonspecific, and this again presents serious diagnostic complications (2).
There are many aspects to the effective control of the disease in both the human and the animal populations, including educational programs, effective animal tracing, vaccination of animals, and intersectorial cooperation. Within and, indeed, prior to the implementation of a holistic control program, establishing the prevalence of disease and identifying infected animals are crucial. Owing to a lack of specific symptoms, the most effective means of doing this is through serological testing, followed by, if possible, the isolation of Brucella from serologically positive animals. In areas where the disease has been eradicated, a surveillance system is vital in order to maintain freedom. Once again, serology also plays a vital role in this.
The Organization International des Epizooties (OIE) prescribed and alternative serological tests for the diagnosis of brucellosis due to infection with smooth strains largely rely upon the detection of antibodies to the O antigen of sLPS (10, 32). The classical tests include the Rose Bengal test, the complement fixation test (CFT), and the serum agglutination test (SAT), all of which employ a whole-cell antigen as the key diagnostic reagent. More recently developed techniques, such as the indirect enzyme-linked immunosorbent assay (iELISA), competitive ELISA (cELISA), and the fluorescence polarization assay (FPA), use purified sLPS or O antigen. The immunodominance of the sLPS O antigen is the basis for the generally good sensitivity of these assays. The use of these antigens can lead to false-positive serological test results when animals are infected with bacteria possessing O antigens with a structure similar to that of the O antigen of Brucella species (7), such as Yersinia enterocolitica O:9. Owing to the widespread use of the S19 and Rev 1 vaccines, such tests also fail to reliably differentiate between vaccinated and infected animals.
In all effective brucellosis control scenarios, the number of samples tested is high, and therefore, optimizing the efficiency of the testing regimen is critical to limit costs. ELISAs are readily amenable to high-throughput testing due to the standardized nature of the technology and reagents. This allows for many efficiency savings, including the introduction of automation (20). Although ELISAs have advantages over classical tests in this regard, they still require several steps to be completed, including separation (wash) steps. Although these steps can be automated, they are a vital part of the assay yet present a frequent source of imprecision, error, mechanical breakdown, and additional cost. Assays which have the advantages of the ELISA, such as assays that use a 96-well format, and that have an objective means of assessment of the results and good sensitivities and specificities but that reduce the burden of work and opportunity for error are clearly desirable.
The aim of the project described here was to improve the efficiency of serological testing by developing a homogeneous homologue of the Brucella cELISA (from the Veterinary Laboratories Agency, Weybridge, United Kingdom) by using the principles of time-resolved fluorescent resonance energy transfer (TR-FRET). FRET occurs when two fluorophores (a donor and an acceptor) with the appropriate spectral properties transfer energy between them if they are within sufficient proximity to each other (9). The degree to which complementary antigens and antibodies have bound (and are therefore within close proximity) can be detected by labeling each with an appropriate fluorophore and measuring the amount of energy transfer produced after the initial excitation of the donor.
Use of a donor fluorophore with a long fluorescent lifetime enables the specific transfer of energy to persist long after the nonspecific background fluorescence, due to the initial excitation, has ceased. Initiating fluorescence intensity (FI) measurements after the background fluorescence has reduced improves the sensitivity of the FRET technique (26). The fluorescence of both the donor and the acceptor fluorophores can be measured, and this endows the method with additional resistance to variable fluorescent effects due to sample matrices, such as serum.
The introduction of competing agents, such as specific antigens or serum antibodies, that alter the degree of binding between the labeled reagents will be manifested as a change in the donor and the acceptor intensities. This process can be conducted with no solid-phase reagents or wash steps and with only a single incubation and a single read step.
The antigen used in the cELISA was the sLPS antigen derived from B. melitensis strain 16 M by phenol extraction (32, 43). For adaptation of the sLPS to the TR-FRET assay, it was prepared as described above and then labeled with fluorescein isothiocyanate (FITC; catalog no. F-7250; Sigma). This was performed by using an adaptation of a previously described method (18). Approximately 3 mg of sLPS was dissolved in 1,200 μl of 0.1 N NaOH, and the mixture was incubated at 37°C for 1 h. Then, 600 μl of a freshly prepared solution of FITC in dimethyl sulfoxide (25 mg/ml) was added, and the components were mixed well and incubated for 1 h at 37°C. The labeled sLPS was separated from the free FITC into TBS (50 mM Tris HCl, pH 7.45, 150 mM NaCl) by using PD-10 columns (GE Healthcare), in accordance with the manufacturer's instructions.
The anti-Brucella sLPS monoclonal antibody (MAb) used in the cELISA, BM40 (13), was labeled with terbium (Tb; LanthaScreen amine reactive Tb chelate; catalog no. PV3582 Invitrogen), in accordance with the manufacturer's instructions. Briefly, 1 mg of BM40 was buffer exchanged into 0.1 M sodium carbonate buffer by using a 2 ml Zebra desalting column (Pierce) and concentrated to a final volume of 0.5 ml. This was added to 100 μg of the Tb chelate and incubated at room temperature in the dark on a rotary shaker for 6 h. The labeled BM40 was separated from the unlabeled Tb and buffer exchanged into TBS by using a 2-ml Zebra column, as described above.
The TR-FRET assay was performed in half-area of black polystyrene non-binding-surface 96-well plates (catalog no. 3686; Corning). The assay result was read with a GENios Pro instrument (Tecan) fitted with a 340-nm filter (bandwidth, 60 nm) for excitation, a 488-nm filter (bandwidth, 10 nm) for measurement of the Tb emission, and a 520-nm filter (bandwidth, 10 nm) for measurement of the fluorescein emission.
The optimal concentrations of labeled antigen, labeled BM40 MAb, and serum were determined by checkerboard titration with a small panel of positive and negative serum samples. The optimal lag time (the period between excitation and the initiation of emission measurement) and integration time (the duration of emission measurement) were systematically determined by evaluating the individual and cumulative signals generated from sequential increments of the measurement parameters. The optimal incubation time was determined by reading the fluorescence outputs at different time points. The samples were read on multiple occasions, such as at different incubation, lag, and integration times, with no detrimental effects.
The final optimized protocol was as follows: to each well, 20 μl of test/control sera, 40 μl of labeled BM40 MAb (5 nM in TBS), and 40 μl of labeled antigen (at 1/700 in TBS) were added in the given order. The plate was incubated on the bench and then read at between 30 to 60 min by using the filters described above and lag and integration settings of 80 and 50 μs, respectively. Each sample was tested singularly.
Each plate was tested with a high-titer and a low-titer positive control, a negative control, and a conjugate control (for which 20 μl of serum was replaced with 20 μl TBS). The low-titer positive control was calibrated so that the titer was equal to the titer of the Brucella OIE ELISA weak positive standard sera (OIEELISAWPSS). The controls were tested in duplicate on each plate. The emission intensity of the fluorescein (520 nm) was divided by the emission intensity of the Tb (488 nm) for all samples. The results were then obtained by expressing the calculated data for the test samples as a percentage of the data calculated for the low-titer positive control.
Both the cELISA (23) and the iELISA with samples from cattle (21) were performed as described previously. The iELISA for use with sheep and goat serum samples was performed as described for the iELISA for use with the cattle serum samples but with B. melitensis 16 M sLPS instead of B. abortus S99 sLPS. The sheep and goat assay uses an anti-goat antibody conjugate and a serum dilution of 1/200. All assays conform to OIE ELISA requirements (10, 32). The determination of a positive or a negative result for the serum samples tested followed the standard means of interpretation for each ELISA method used.
To assess the diagnostic sensitivity (DSn) of the assay, single serum samples (from our serum archive) from 32 cattle and 41 sheep and goats (small ruminants) were tested. Of the samples from cattle, 8 were from naturally infected culture-positive animals, 2 were from culture-positive animals experimentally infected with B. abortus strain 544, 10 were from culture-positive animals, and a further 12 were from serologically positive (by CFT and SAT) animals from a herd with a culturally confirmed outbreak of brucellosis. Of the 41 samples from small ruminants, 2 were from naturally infected culture-positive animals, 5 were from culture-positive animals experimentally infected with B. melitensis, 9 were from animals that were serologically positive (by CFT and SAT) and from a herd with a culturally confirmed outbreak of brucellosis, and the remaining 25 were from animals from a herd involved in a suspected outbreak of brucellosis from an area of endemicity.
To assess the diagnostic specificity (DSp) of the TR-FRET assay, single serum samples from 240 randomly selected cattle from Great Britain (which has officially been brucellosis free since 1985) were collected. In addition, single serum samples from 240 randomly selected sheep and goats from Great Britain were also collected.
The analytical sensitivity of the TR-FRET assay was measured by adding unlabeled MAb BM40 and testing a dilution series (prepared in negative control sera) of the Brucella OIE ELISA strong positive standard sera (OIEELISASPSS).
Some of the cattle serum samples used were hemolyzed. This was clearly visible by eye, but to quantify this objectively, all the samples were diluted 1/10 in phosphate-buffered saline (pH 7.4) and 100 μl was added to individual wells of a clear 96-well polystyrene plate. The plates were read at 450 nm on a colorimetric plate reader to provide optical density (OD) data for each sample.
The analysis of all data was performed with Microsoft Excel software. The optimal TR-FRET assay positive-negative cutoff threshold was determined by using two-way receiver operator curves (12). The confidence intervals for the DSn and DSp values derived from the evaluation panel were calculated by using binomial distribution models (4). Correlation coefficients were calculated by using the Pearson product-moment equation and were tested for significance by Student's t test.
The summary results for the serum evaluation panel for all the assays are shown in Table Table1.1. The TR-FRET assay results have been presented for readings taken after incubation periods of 15, 30, and 60 min. The TR-FRET assay had 100% DSn and 100% DSp for samples from cattle for all three incubation periods. For the samples from sheep and goats, the TR-FRET assay had 100% DSn and 100% DSp for the 30- and 60-min incubation times; the values obtained for the 15-min incubation time did not match those results, as there was one sample with a false-positive result. In order to obtain the optimal DSn and DSp values after 15 min of incubation, the use of a cutoff different from that used to obtain the optimal values after 30 and 60 min of incubation was required. Table Table11 also shows the range of TR-FRET test values that can be selected as the positive-negative cutoff while maintaining the DSn and DSp values shown. This range grows as the incubation time increases. At 30 and 60 min, it is possible to set the positive-negative cutoff at a test value of 120%, which results in 100% DSn and 100% DSp for serum samples from cattle and from sheep and goats. This is the positive-negative cutoff that has been used for the rest of the analysis. Although the test continues to improve with time, the majority of the samples react quickly and can swiftly be determined to be positive (data not shown). After 60 min, there is no further change in the data (results not shown). The DSn and DSp values for the TR-FRET assay are equivalent to those of the iELISA but superior to those of the cELISA and the FPA. The only significant difference, as determined from the confidence intervals, was that the FPA showed a lower DSp for samples from cattle than the other assays used.
The TR-FRET assay results for all the sera in the evaluation panel are presented in Fig. Fig.1,1, which plots the results after 30 min of incubation against the results for the same sample after 15 and 60 min of incubation. This shows that there was a consistent response against time, as all the results fit close to a straight line (correlation coefficients for 30 min against 15 min of incubation and 30 min against 60 min of incubation, 0.991 and 0.993, respectively). Figure Figure11 also shows that there was good separation between the results for the infected and the noninfected animals, as they mainly fit into the bottom left and top right quadrants, respectively, with the quadrants having been formed by plotting the 120% cutoff value. The separation was perfect for the data for the 30- and 60-min incubations but not for the data for the 15-min incubation, in which there were two false-positive results and one false-negative result by use of a 120% cutoff. This can be seen more closely in the inset in Fig. Fig.1.1. Fig. Fig.11 shows that, in general, samples from infected animals are more positive (have higher titers) and those from noninfected animals are more negative (have lower titers) after 60 min of incubation than after 15 min of incubation.
The TR-FRET assay (30 min of incubation) and cELISA results for all the sera in the evaluation panel are potted against each other in Fig. Fig.2.2. The respective positive and negative cutoffs for each assay are also plotted. Figure Figure22 shows that for both assays, the majority of the positive and the negative samples are well clustered in two regions distinct from each other. For the TR-FRET assay, although there was perfect DSn and DSp, the result for one sample from an infected animal was particularly close to the 120% cutoff. The cELISA results showed one false-positive result and one false-negative result. There was a significant (P < 0.001) positive correlation between the positive results for both assays.
The raw data from the 30-min TR-FRET assay are plotted in Fig. Fig.3.3. The emission FI of the donor (Tb) is shown on the x axis, and the FI of the acceptor (fluorescein) is shown on the y axis. The values for the test controls for the different plates tested have also been plotted. There was good reproducibility between these intertest replicates. Most, but not all, of the samples from noninfected animals had a higher acceptor FI than the samples from infected animals. The samples with the low acceptor FIs also had low donor FIs; therefore, by expressing the test results as a function of the acceptor and donor FIs, it is possible to obtain perfect DSn and DSp, as demonstrated by the diagonal dashed line in Fig. Fig.33.
The ODs of the samples from noninfected cattle are plotted against the TR-FRET (30 min of incubation) results in Fig. Fig.4.4. The majority of the samples had ODs less than 0.5, but a significant number of samples had greater ODs. Figure Figure44 shows that there was a general downward trend in the TR-FRET assay result as the OD increased, although there were exceptions. Figure Figure44 also shows that samples with low acceptor FIs (520 nm) tended to be those that had higher ODs. This can be related back to Fig. Fig.3,3, in which samples with acceptor FIs of less than 2,000 had donor FIs proportionally higher than the average for the noninfected sample population. However, these samples remained negative, despite their high ODs. The sample with the highest OD did have a low, if not extreme, acceptor FI value of 3,363 compared to the average (for cattle), which was 4,952.
The analytical sensitivity, as measured by the detection of unlabeled MAb BM40 added to the reaction mixture, is shown in Fig. Fig.5.5. The data from the TR-FRET assay are plotted alongside the data from the cELISA and the AlphaLISA (PerkinElmer), which have been reported previously (22). The positive-negative cutoff for each assay is also shown in the respective dashed horizontal lines. The TR-FRET assay exhibited a dose-response curve similar in shape to that of the AlphaLISA, whereas the curve for the cELISA was less steep. The approximate MAb concentrations at the positive-negative cutoffs for each assay were 4 nM for the TR-FRET assay, 2 nM for the cELISA, and 0.25 to 0.125 nM for the AlphaLISA. Against these results it should be noted that the final serum dilution factors for each of these assays were 1/5 for the TR-FRET assay, 1/6 for the cELISA, and 1/150 for the AlphaLISA.
The TR-FRET assay was also used to test the OIEELISASPSS, and the results are presented in Fig. Fig.6.6. The results for the 15-, 30-, and 60-min incubations are shown. Each dilution was tested in triplicate, and the average was plotted. The error bars, shown only for the data from the 30-min incubation, show the maximum and the minimum values. The result for the 1/8 dilution was consistently classified as positive, and the result for the 1/16 dilution was consistently classified as negative. As could be seen in Fig. Fig.1,1, the results demonstrate that the assay becomes more sensitive and specific over time, although the data for the 30- and 60-min incubations are closer than those for the 15- and 30-min incubations. The coefficient of variation (CV) for all the data for the 30-min incubation (the average of the CV for the three replicates at each dilution) was 3.61%.
The underlying principle of FRET is that the level of energy transfer is dependent upon the proximity and the orientation of the donor and the acceptor fluorophores. This technique, sometimes referred to as Förster resonance energy transfer, has been known for more than 60 years (9) and has been applied under many circumstances, including real-time PCR for the detection of Brucella DNA (37). The proximity dependence of FRET enables binding reactions to be measured without a separation or wash step.
In order to improve the specificity of FRET detection, time-resolved techniques were developed to remove, by time gating, the majority of nonspecific (non-FRET) fluorescence (26). This approach was improved considerably by the development of lanthanide cryptates (including Tb) with exceptionally long fluorescent lifetimes and high quantum yields (19, 38). This enabled the use of donor fluorophores capable of sustaining energy transfer for a substantial period of time, therefore enabling fluorescence detection when the non-FRET fluorescence had substantially declined, resulting in increased sensitivity (15).
Owing to the efficiency and the sensitivity of homogeneous lanthanide-based TR-FRET, the technique has been widely applied across a number of disciplines for the detection of a range of analytes within many matrices. It is particularly well suited to high-throughput screening in the drug discovery industry (25, 46) and in clinical settings (36). It has successfully been applied to the detection of analytes within human sera (3, 19) and has been shown to work in either the competitive format (6) or the sandwich format (1). It has also been applied for the detection of human hormones (42) and cytokines (8). Time-resolved fluorescent techniques have been used for the detection of bacterial food pathogens (14) and toxins (34), but these methods use only one fluorophore type and are not homogeneous.
The homogeneous TR-FRET approach has a strong track record of success, yet until now it has not been developed for the specific detection of infectious diseases in either the human or the veterinary sector. The aim of the project described here was to develop a new, efficient, and effective assay for the serodetection of anti-Brucella antibodies by converting an effective existing cELISA to a homogeneous TR-FRET format.
The underlying biological principle of this new assay is the same as that for the cELISA, in that serum antibodies specific for the O-antigen component of the Brucella sLPS outcompete a labeled MAb that is also specific for this antigen. The MAb used in this example, BM40, is specific to the M epitope on the Brucella O antigen (13). Although this epitope is found largely in the O antigen of M epitope-dominant Brucella strains (such as strain 16 M used to make the sLPS for this assay), antibodies to epitopes common to all smooth Brucella stains will inhibit the binding of BM40 to the O antigen of 16 M, possibly due to the steric hindrance caused by the overlapping of these epitopes on the O antigen (45) or conformational changes in the O antigen due to antibody binding (16). This inhibitory property is further demonstrated by the inability to eliminate false-positive reactions due to infections with Yersinia enterocolitica O:9 through the use of Brucella-specific MAbs in a cELISA (22, 44). This is therefore also likely to be the case with this TR-FRET assay, although further testing is planned. The effect of this competition is measured by an enzyme-driven color change in the final stage of the cELISA. In the TR-FRET assay, this competition will reduce the binding between the MAb and the antigen, and therefore, their conjugates will be less able to exchange energy. It is therefore possible to indirectly measure the degree of binding by measuring the specific fluorescence intensities of the conjugates.
Preliminary studies with control sera enabled the development and optimization of the assay. It was found that a lag time of 80 μs was sufficient to exclude the majority of the non-FRET signal that is due to any intrinsic fluorescence of the matrix (in this case, serum), the plates, and the direct stimulation of fluorescein. A subsequent reading period of 50 μs maximized the difference in the proportional donor-to-acceptor emission intensities between low-titer positive and negative samples.
A larger evaluation panel was used to establish the optimal incubation time for the assay developed. An advantage of the TR-FRET technique is that it is possible to perform multiple reads on the same plate to measure the kinetics of the reaction, and this simplified optimization of the assay. The data from Fig. Fig.11 show that the reaction reached optimal DSn and DSp values from 30 min onwards, and there was little difference from 30 to 60 min. No difference in performance was found within the next 60 min. From a practical perspective, this offers the user a large window of opportunity to read the assay and adds flexibility to a large testing regimen. After a 15-min incubation period, the assay had already correctly classified 99.46% of the samples (when a 120% cutoff was applied). It is possible that by shaking the plates, the reaction speed may increase, yet by showing that the reaction is effective without shaking demonstrates the simplicity of the working protocol and its amenability to high-throughput testing.
The results from the TR-FRET assay are plotted against the results from the parent cELISA in Fig. Fig.2.2. As expected, there was a good concordance between the two sets of data. This is borne out by the relative DSn and DSp values for the two assays and demonstrates that none of the performance of the cELISA has been lost in the conversion to the TR-FRET platform. As with the cELISA, it is possible to set a common positive-negative cutoff for all species of ruminants tested here, and this cutoff can be applied universally from 30 min onwards. As with the cELISA, the TR-FRET assay is not test subject species specific.
There was a general trend for the TR-FRET assay values for the samples from noninfected cattle to be lower than those for the samples from noninfected small ruminants; this was not apparent in the cELISA. This may be due to the intrinsic effects of the serum on fluorescence, and this is more apparent in Fig. Fig.3.3. From Fig. Fig.33 it is clear that samples from uninfected cattle have generally lower FI values. The fact that large differences in the FI values between sera have not had a detrimental effect on the results is testimony to the robustness of the technique. Expressing the results as a function of the donor and acceptor FIs and then as a percentage of the values for low-titer positive control compensates for these interserum differences and helps normalize the assay for interplate variations. This property enables the protocol to be completed with just one read (both donor and acceptor emission intensities can be measured in a single read).
The difference in the distribution of the results for the noninfected cattle and the noninfected small ruminants was most likely due to the relatively poor condition of the samples from the cattle than species-level differences. Figure Figure44 demonstrates that a significant proportion of the samples from noninfected cattle had high ODs which corresponded to the observations by eye that the samples had hemolyzed (the samples from the small ruminants were in a visibly better condition). Such samples are unsuitable for testing by CFT and SAT (39). Previous test development projects have also determined that samples with OD values greater than 0.5 also produce unreliable results by AlphaLISA (22). Although there was a weak negative relationship between the OD and the TR-FRET result, this did not affect the qualitative results.
It was possible to measure the analytical sensitivity of the TR-FRET assay by using unlabeled MAb BM40 as the target analyte. The results were compared with those previously reported for the Brucella AlphaLISA and cELISA (the same method of comparison is not possible for the FPA, as it is not a competitive method and the O antigen used is not M-epitope dominant). Upon initial inspection of the results, the TR-FRET assay appears to be the least sensitive of the three methods. These results are of interest, as they clearly demonstrate the underlying competitive mechanism of the reaction and the resulting test output. The data from all three assays suggest that once serum dilution factors have been taken into account, a serum titer equivalent to approximately 10 to 20 nM of BM40 MAb is an appropriate point at which to optimize the dynamic range for the assay. Yet the data are limited in their relevance to serodetection. The TR-FRET assay was optimized to maximize the difference between low-titer positive and negative samples rather than analytical sensitivity and has the ability to deal with confounding, non-antibody-mediated factors in the serum.
The performance of the TR-FRET assay was measured against the OIE ELISA standards used to standardize the performance of ELISA methods for the serological diagnosis of brucellosis (32). These results showed that the TR-FRET assay passed the OIE criteria set for ELISAs, as the result for the low-titer positive control was always positive (by definition, it is 100%), and this is calibrated to equal the titer of the OIEELISAWPSS. The 1/8 predilution (in negative sera) of the OIEELISASPSS was also positive (the OIEELISAWPSS was prepared as a 1/8 dilution [in negative sera] of the OIEELISASPSS). This provides further evidence to support the strong performance of the TR-FRET assay and demonstrates that use of these standards could be appropriate for standardization of this test. The data showed the good reproducibility of the method, with the average CV of the replicates being just 3.61%.
Owing to the relatively small size of the evaluation panel, it is appropriate to compare the diagnostic performance of the TR-FRET assay relative to that of the ELISAs rather than to consider the presented data in isolation. The diagnostic characteristics of the ELISAs themselves are strong, with the performance indices both being in excess of 194% (23). The results from Table Table11 demonstrate that the TR-FRET assay had nearly perfect performance with the evaluation panel. The performance of the TR-FRET assay with the 30- and 60-min incubation periods was equal to that of the iELISA, and both assays were able to correctly differentiate between samples from infected and noninfected animals in 100% of cases. The performance of the TR-FRET assay was superior to that of its parent assay, the cELISA. Although the two assays share the same underlying biological principle, the reason for the difference in performance may be due to the superior precision of the TR-FRET assay. The elimination of coating and separation steps not only saves considerable time and money but also reduces many potential sources for variation.
The TR-FRET assay also demonstrated a performance significantly superior to that of the FPA. The FPA has previously been reported to have excellent DSn and DSp values (24, 28, 31), so the differences presented here would require additional evidence before they are considered conclusive. Both assays are homogeneous and offer some considerable advantages to the user because of this. Both are based on the measurement of fluorescent signals from a labeled antigen, and therefore, both must be able to deal with the intrinsic variability of the target sample. The TR-FRET assay does this in a single read step by taking two measurements which are affected by the test matrix proportionally. The FPA handles this problem by taking two measurements separately, one before the addition of the labeled antigen and one after. This adds a step to the protocol compared to the number of steps in the TR-FRET assay. This is counterbalanced by the use of just three components in the FPA (sample, buffer, and antigen), whereas the TR-FRET assay has four components, as the labeled MAb must also be added.
In theory, it is quicker to perform the FPA than the TR-FRET assay, yet in practice, the additional read step and time restrictions add a complexity and a lack of flexibility when large numbers of samples are tested, which is not the case with the TR-FRET assay. A successful FPA is dependent on the detection of changes in the rate of spin of a labeled molecule that are due to changes in the effective size of that molecule. The rate of spin is also dependent on temperature and viscosity, and both of these add a source of non-antibody-mediated variation to the test result (24, 27). The TR-FRET assay is also much less affected by such changes.
Owing to the short lifetimes of high-quantum-yield polarized fluorophores, there is a relatively small range over which changes in molecular size can be reliably detected by FPA. The TR-FRET assay is more adaptable in this regard, as it is only the distances between the labels themselves that are of importance. The AlphaLISA is able to work over even greater distances than typical FRET systems (40, 41); however, the bead-based nature of the assay means that it is slower than both the truly homogeneous FPA and TR-FRET assay. It has also been shown to be vulnerable to sample matrix effects (22).
This study showed that by conjugating the biological reagents of the Veterinary Laboratories Agency Brucella cELISA with specific donor and acceptor fluorescent reagents and applying the principles of TR-FRET, an effective new assay was developed. This assay showed all of the benefits of the cELISA, including high sensitivity and specificity. The TR-FRET assay is very simple to perform, is robust, and has an excellent serodetection capability. It can be performed with equipment already in place in laboratories which have 96-well-plate FPA capabilities and offers an attractive alternative option. The technique is highly suitable to high-throughput testing but also presents a valid option for low-throughput tests. The fluorescent mechanism of the technique suggests that it may also be possible to develop tests that use whole-blood samples (29) and milk samples (30), as has been the case for the FPA. Likewise, a portable single-tube assay could be developed. Furthermore, the adaptability of the technique offers the possibility of developing a range of similar assays for the detection of other infectious diseases.
This work was supported by the InterAct partnership (though funding provided by the Department for Innovation Universities and Skills, Great Britain) and the Department for Environment, Food, and Rural Affairs, Great Britain.
Published ahead of print on 5 August 2009.