Sage Journals: Discover world-class research

Abstract

Enzyme-linked immunosorbent assay (ELISA) is a standard immunoassay to estimate protein concentrations in a variety of samples. An automated ELISA procedure was established that allows the measurement of the concentration of numerous proteins in a small sample volume. The automated ELISA procedure was tested by measuring the concentration of brain-derived neurotrophic factor in serum samples from humans.

It was shown that the reproducibility of protein concentrations in three assay plates within one automated run was greater than 92%, within independent automated experiments greater than 96%. Automated processing with a maximum capacity of six plates and 240 serum samples per run showed an average interassay precision of 91 %. Serial dilutions of randomly chosen sera displayed significant linearity of recovered concentration.

In conclusion it can be stated that the automated assay provides high flexibility with highly reproducible results and has several advantages compared to the manual method, including less operator dependence and faster sample throughput. The investigated ELISA test system seems to exhibit analytical characteristics indicating further clinical application. (JALA 2007;12:219–29)

Keywords

bioautomation ELISA BDNF

Introduction

Enzyme-linked immunosorbent assay (ELISA) represents one of the most commonly applied methods providing a simple, rapid, safe, and cost-effective way of protein determination.^{1, 2} Nevertheless, the process suffers from being highly time consuming and requires a series of liquid supplementations, washes, and incubations before detection leading to both, manpower requirements and an increased risk of failures due to the human factor. This is of special concern when laboratories have to process regularly a high number of samples. Automation of the ELISA technique and application of such methods for clinical and pharmaceutical laboratory efforts constitute, therefore, a significant part of research.^3–6 Thus, a fully automated sandwich ELISA in a 96-well format was established using a Beckman Coulter core system and peripheral devices (Fig. 1) supervised by SAMI3.5 software (Sagian Automated Method Interface, Version 3.5, Beckman Coulter) for optimal scheduling of sample processing. As a test setup, the concentration of brain-derived neurotrophic factor (BDNF) in serum samples was determined. BDNF, a member of the neurotrophin family, is a widely distributed growth and differentiation factor for neural cells that can be found in high concentrations in the human brain but also in platelets of the peripheral blood.^7–9 Altered BDNF production and secretion have been proven in a number of diseases. For instance, neurodegenerative disorders are known to be associated with decreased BDNF levels in the brain, for example, Alzheimer's and Parkinson's diseases.^10–12 There is also evidence that enhanced local BDNF production in the lung contributes to neuronal hyperreactivity and pathologic bronchoconstriction in asthma.¹³ Even in more emerging diseases such as diabetes or obesity, there seems to be a correlation with distinct BDNF serum concentrations.¹⁴ An automated BDNF–ELISA is therefore useful for clinical diagnostics and research. To establish the automated protocol, a commonly available BDNF-ELISA test kit (R&D Systems) was used.

Figure 1.

Robotic instrument setup.

System Asset

The automation core system that was used here (Fig. 1) consists of a robotic arm (optimized robot for chemical analysis, Beckman Coulter) that is able to deliver required labware to defined positions. As a central device, there is a liquid handling station, providing several pipetting tools (Biomek2000, Beckman Coulter). Moreover, there are several peripheral devices such as an automated washer (ELx405 select, Biotek), a barcode reading and labeling device (Microplate Print &Apply, Beckman Coulter), six shaking units (Variomag Teleshake, H + P, Oberschleißheim, Germany), a hotel with four stacks configured for tip boxes and deepwell plates, one stack for microtiter plates and a dark hotel (also usable as Kryohotel –80 °C) including two stacks for plate incubation (AIGmbH), and two reading device (Fluostar Galaxy, NovoStar, BMG Labtech) integrated. The device integration of reader, shaker, and dark hotel was performed in-house (AIGmbH, Rostock, Germany). The core system is driven by the SAMI3.5 superordinated software (Sagian Automated Method Interface, Version 3.5, Beckman Coulter) for optimal scheduling of sample processing in a medium throughput way.

Materials and Methods

Sample Preparation

Blood samples from cannulation of a cubital vein of 18 voluntary healthy probands were collected into additive-free containers to obtain serum samples (samples were taken several times for each volunteer during a period of 1 year). Blood was placed at rest on ice for 60 min to allow clotting and degranulation of the thrombocytes leading to the release of BDNF from platelets to serum. Subsequently, sample containers were centrifuged at 2000 × g and 10 °C for 15 min, supernatant was aliquoted and stored at –80 °C until measurement.

Elisa

For the quantitative determination of BDNF concentrations, a commercially available ELISA kit was used explicitly following the manufacturer's instructions (R&D Systems, DuoSet ELISA Development Kit). Antibodies, streptavidin-labeled Peroxidase (POX), and recombinant BDNF were supplied with the kit. Chromogen solution (3,3′,5,5′-tetrame-thylbenzidine [TMB] substrate buffer [H₂O₂], R&D Systems) and stop solution (2 N H₂SO₄) did not need further dilutions. Phosphate-buffered saline (PBS)/1% bovine serum albumin was used as reagent diluents (RD) with the exception of the capture antibody which was exclusively diluted in PBS. Serum samples were diluted 1:50 for standard procedure and 1:40–1:160 for the serial dilution. The standard maximum concentration was set to 2000 pg/mL and the standard minimum concentration was 62.5 pg/mL. All standard concentrations were measured in duplicates; samples and blanks were measured in two to eight replicates, depending on the assay layout. The calculation of protein concentrations from measured extinctions was performed automatically by a Fluorescence Reader (Fluostar Galaxy, BMG Labtech) facilitating a serial dilution standard curve.

SAMI Methods

Coating. Plate coating with capture antibody and sample preparation was performed in two separate SAMI methods each consisting of one liquid-handling step on the Biomek2000 workstation. For the coating process, coating buffer was placed on the deck, whereas lidded test plates were placed in the dark hotel. Additionally, a tip box was positioned in the appropriate hotel stack. The test plates were labeled with a barcode before transfer of capture antibody.

Sample Dilution and Layout. Before transfer of serum samples, the sample plates were labeled with a barcode. For sample preparation, undiluted serum (6 μL/sample, 10 μL for serial dilution) and the required amount of tip boxes were placed in the appropriate stack. Three different sample dilution methods were configured: the first one for sample duplicates (240 samples/six plates), the second one for four to eight replicates (48–96 samples/six plates), and the third one for serial dilution (1:40, 1:80; 1:160) in four to eight replicates (18–36 samples/six plates). The 1:50 dilution step of the samples (or the first dilution step for serial dilution, respectively) was performed within a deepwell plate. Subsequently, 60 μL of the diluted sample was subjected to the sample plates.

ELISA Workflow. Lidded sample plates and tip boxes were stored in defined hotel position groups and lidded test plates were stored in the dark hotel to avoid light exposure. The Biomek2000 deck was preconfigured with the required reagents and standards (see Fig. 2). The transfer volume into test plates was 50 μL each and 100 μL for RD, respectively. The incubation steps were either performed in the dark hotel without shaking (blocking, POX, and substrate incubation) or directly on the shaking units.

Figure 2.

Hardware setup Biomek2000 workstation. ALP, Automated Labware Positioners.

The subsequent ELISA processing comprises six distinct methods on the Biomek2000 workstation combined to one SAMI protocol (see Fig. 3, and for a schematic overview Fig. 4).

Figure 3.

Workflow of an automated ELISA.

Figure 4.

Schematic overview of a sandwich ELISA. 1: coating with capture antibody; 2: capture of antigen; 3: Secondary antibody (e.g., biotinylated) binds to the complex; 4: streptavidin-coupled peroxidase binds to biotinylated-Fc of secondary antibody; 5: addition of enzyme substrate and resulting enzymatic reaction; 6: reaction stop with 2 N sulfuric acid and measurement (photometer).

Assay workflow:.

Transfer of blocking solution (RD)

Standard and sample transfer to ELISA plate (standards for all six plates were diluted manually and provided in a deepwell plate)

Addition of biotinylated anti-BDNF antibody

Addition of streptavidin–HRP conjugates

Mixing of TMB and H₂O₂

Transfer of mixed solution to test plate

Addition of stop solution.

After each transfer step, plates were placed on one shaking unit for 30 s for optimal reagent distribution. The ELx405 was used to wash all plates in-between transfer steps with provided wash buffer. Absorption readings were taken with a Fluorescence Reader (Fluostar Galaxy) at 450 nm and 590 nm. The values obtained at 590 nm served as correction factor that was used to subtract plate imperfections.

Each experiment was accompanied by an online visualization software (in house development, AIGmbH), which generated layout-dependent “heat maps” and calculations such as average absorption (AV), standard deviation (StDev), and Z-factor,¹⁵ of each plate. For the calculation of the Z-factor, the mean and StDev of the maximum standard concentrations and blanks of the respective assay plates were consulted as positive and negative controls (for formula, please refer to Ref. 15). The assay plates were identified according to barcode, date, and time.

These three independent methods were attached to run in one time line.

Results

Running Time and Scheduling

The running time of one assay plate was determined to 6 and a half hours. Due to optimal duling of families by SAMI3.5, the whole process for six assay plates is completed within 8 hours and 45 minutes (see Fig. 5). The running time of the coating method varied between 5 min and 45 min, depending on the amount of assay plates. The effort for automated sample preparation was dependent on the assay layout (two or four replicates, serial dilution of samples) and number of sample plates, but did not exceed 1 and a half hours.

Figure 5.

Optimal SAMI3.5 scheduling process of six ELISA assay plates. Explanations of the scheme are given in white blocks, representative for the first assay plate.

Consistency of Results

A major concern while setting up ELISA assays is about the result consistency. To elucidate the test system with regard to this problem, the assay was first run with one sample with a constant concentration (1:50), including four blank controls across the entire plate (intra-assay variation). The average intra-assay variation was determined to be between 13.5% and 16.7% (four plate replicates, see Table 1). The signal to noise ratio¹⁶

(S / N = [M_{signal} 2 M_{background}] / \sqrt{{[{SD}_{signal}]}^{2} + {[{SD}_{background}]}^{2}})

Table 1

The average absorption measurement across four entire assay plates containing one serum sample in a 1:50 dilution

Plate	Average absorption 450 nm	StDev	%CV
1	0.861	0.123	14.303
2	1.050	0.142	13.537
3	1.063	0.157	14.855
4	1.415	0.235	16.666
Average	1.097	0.164	14.84

was approximately 3:1 (2.8:1) and thus within the expected range.

The consistency of the exact BDNF amount was analyzed by running a standard dilution across an entire plate (Fig. 6, Table 2). The standard was provided in one row of a deepwell plate, and then transferred onto the assay plate. Here, the absorption measurements and calculated concentrations of BDNF were in excellent correlation (R > 0.99) to the real BDNF amount in the wells. However, the deviation increased with lower protein concentrations.

Table 2

The average of calculated recombinant BDNF concentration across an entire plate

Plate row	Concentration recombinant BDNF (ng/mL)	Average calculated amount BDNF columns 1–12 (ng/mL)	StDev	%CV
1	2.000	1.954	0.240	12.289
2	1.000	1.083	0.212	19.559
3	0.500	0.541	0.066	12.135
4	0.250	0.228	0.051	22.213
5	0.125	0.095	0.030	31.831
6	0.063	0.055	0.016	29.515
7	0.000	–0.015	0.004
8	0.000	–0.006	0.015

Figure 6.

Comparison between defined BDNF concentrations (x-axis) versus the BDNF concentration from automated measurement (y-axis). Measurement was performed across an entire plate.

Comparison Between Manual and Automatically Generated Results

To estimate whether the data from an automated procedure are comparable with a manually performed ELISA, the assay was done manually step by step. Here five serum samples with a constant concentration (1:50) in four replicates were compared. Subsequently, the results were aligned with data achieved by the automated workflow.

The manually performed ELISA displayed a mean variation factor of 17.8% whereas the automated procedure showed a variation of only 14.6%. Both assays displayed a correlation coefficient greater than 0.93, and the calculated serum concentrations were comparable to each other (see Figs. 7 and 8).

Figure 7.

Comparison of estimated BDNF serum concentration resulting from manual and automated ELISA procedure. P1–5: randomly chosen serum samples.

Figure 8.

Comparison of standard curve resulting from manual and automated experiments.

Reproducibility of Data

To investigate the reproducibility of the data that have been received from an automated run, one run with three plates and three different 1:50 in RD-diluted serum samples/plate was performed and the same experiment was repeated on another day (Table 3A). Here, high reproducibility within one automated run was obtained (intra-assay variance). Although the precision varied between serum samples, it did not exceed 11% and the resulting average variability was lower than 8% (see Fig. 9 and Table 3B). The estimated BDNF serum concentration of the same samples between two independent automated runs was even better (ranging between 92% and 98.9%).

Table 3A.

Reproducibility of concentration measurements over three plates within one automated run, and comparison of two independent automated runs 1, 2, 3: randomly chosen serum samples; 1_1: assay plate_experiment

		1		2		3
	AV (ng/mL)	StDev	AV (ng/mL)	StDev	AV (ng/mL)	StDev
Plate 1_1	53.29	3.99	41.53	6.38	48.34	0.70
Plate 1_2	53.31	9.27	43.83	0.22	38.17	3.32
Plate 2_1	54.95	7.93	45.60	1.93	54.87	3.38
Plate 2_2	56.43	4.86	47.80	2.28	56.05	1.42
Plate 3_1	52.14	2.97	48.61	1.93	49.97	3.84
Plate 3_2	57.07	0.35	42.10	8.47	49.73	10.73

AV, average BDNF concentration of eight replicates of serum sample; StDev, standard deviation.

Table 3B.

Inter- and intra-assay variance of three serum samples/three plates and two experimental runs

	Intra-assay variance
Sample_exp	AV p1–3	StDev p1–3	%CV p1–3
1_1	53.55	1.99	3.71
2_1	45.25	3.55	7.85
3_1	51.06	3.40	6.66
1_2	55.60	2.01	3.62
2_2	44.58	2.92	6.56
3_2	47.98	9.07	18.90

	Interassay variance
e1 versus e2	AV p1–3	StDev p1–3	%CV p1–3
1_1–1_2	54.57	1.46	2.67
2_1–2_2	44.91	0.47	1.05
3_1–3_2	49.52	3.95	7.95

AV, average BDNF concentration of eight replicates of a serum sample; StDev, standard deviation; e, experiment; p, assay plate.

Figure 9.

Reproducibility of concentration measurements over three plates within one automated run, and comparison of two independent automated runs.

Stability of Results at Maximum Capacity

To estimate the stability of the assay results at maximum possible capacity, the ELISA method was performed with six assay plates. Here two different protocols were used:

two sample replicates and

four sample replicates.

The automated run of six assay plates with 1:50 in RD-diluted serum samples in duplicates displayed an intra-assay variability ranging from 1.8% to 21%. The average variation was 9%. The same experiment performed with four replicates have shown an intra-assay variability ranging from 7.8% to 14.8%. The mean was also nearly 9% (see Table 4). These results are in consistence with previous received data (see point 4). Even the automatically calculated Z-factor was fairly good (greater than 0.7). However, it turned out that the stability of the protein standard is a crucial factor for the consistency of the assay. The protein absorption values decreased along the time of assay processing. To ensure reproducibility, the deepwell reservoir had to be refilled with chilled standard (4 °C) during half of the transfer steps. In contrast to the BDNF standard, the serum samples did not show degradation. BDNF dissolved in neutral blood serum appears to be more stable over a longer period of time.

Table 4

Intra-assay variability

Assay	%CV of duplicates	%CV of four replicates
Plate 1	5.7	7.8
Plate 2	1.8	8.2
Plate 3	2.42	7.94
Plate 4	21	7.8
Plate 5	7.19	14.8
Plate 6	16	11
Average	9.02	9.59

Linearity of the Serial Dilution

To estimate the linearity of BDNF concentrations in different sera, a protocol for automated serial dilution was set up. Here, six assay plates were run, each consisting of three serum samples in eight replicates (all together 18 serum samples). The serial dilution resulted in a final dilution of serum samples in RD of 1:40, 1:80, and 1:160, respectively. The results are shown in Table 5 and Fig. 10. The analysis of the row data showed that the automated run is in acceptable linearity within one serum sample over three dilution steps. The StDev varied between 7% and 35% (see below). The average deviation across all sample dilutions was 14.6%. The average deviations of the distinct serum dilutions ranged from 19.8% for 1:40 sample dilution to 22.6% for 1:160 sample dilutions. However, there are some exceptions with relatively high deviations (e.g., sample 5 or 13), caused by some outliers (for a summary, see Table 5).

Table 5

Serial dilution of randomly chosen serum samples

Plate 1 serum	S1 1:40	S1 1:80	S1 1:160	S2 1:40	S2 1:80	S2 1:160	S3 1:40	S3 1:80	S3 1:160
Average	0.58	0.31	0.18	0.57	0.31	0.19	0.41	0.22	0.12
StDev	0.12	0.06	0.03	0.06	0.08	0.05	0.09	0.05	0.03
%CV	21.29	18.31	13.88	10.69	24.65	26.01	21.93	23.88	22.62
[ng/mL]	1.44	0.75	0.43	1.42	0.74	0.44	1.00	0.52	0.27
Dilution^a	57.53	59.66	68.06	56.91	59.15	69.97	40.07	41.33	42.97
Plate 2 serum	S4 1:40	S4 1:80	S4 1:160	S5 1:40	S5 1:80	S5 1:160	S6 1:40	S6 1:80	S6 1:160
Average	1.09	0.67	0.42	0.81	0.48	0.26	0.64	0.38	0.26
StDev	0.20	0.19	0.12	0.25	0.09	0.04	0.14	0.03	0.04
%CV	18.25	27.84	28.46	30.85	18.48	16.78	22.75	8.69	14.38
[ng/mL]	2.40	1.44	0.88	1.77	1.01	0.52	1.37	0.79	0.51
Dilution^a	95.87	1 15.00	141.33	70.84	80.62	83.73	54.84	63.43	81.82
Plate 3 serum	S7 1:40	S7 1:80	S7 1:160	S8 1:40	S8 1:80	S8 1:160	S9 1:40	S9 1:80	S9 1:160
Average	0.78	0.46	0.27	0.69	0.37	0.21	0.58	0.36	0.19
StDev	0.11	0.10	0.06	0.15	0.10	0.05	0.17	0.13	0.05
%CV	14.25	22.68	24.30	21.71	27.46	22.14	29.90	37.84	25.66
[ng/mL]	1.72	0.94	0.49	1.49	0.74	0.35	1.22	0.70	0.31
Dilution^a	68.80	74.96	77.81	59.51	59.27	55.71	48.91	56.00	48.86
Plate 4 serum	S10 1:40	S10 1:80	S10 1:160	S1 1 1:40	S1 1 1:80	S1 1 1:160	S12 1:40	S12 1:80	S12 1:160
Average	0.65	0.40	0.25	0.57	0.34	0.20	0.59	0.36	0.20
StDev	0.10	0.08	0.05	0.15	0.09	0.02	0.12	0.09	0.06
%CV	15.76	19.09	19.33	25.59	25.54	10.96	20.56	23.86	30.84
[ng/mL]	1.67	1.03	0.63	1.47	0.87	0.50	1.51	0.92	0.51
Dilution^a	66.60	82.19	100.71	58.78	69.63	79.65	60.31	73.53	81.78
Plate 5 serum	S13 1:40	S13 1:80	S13 1:160	S14 1:40	S14 1:80	S14 1:160	S15 1:40	S15 1:80	S15 1:160
Average	0.56	0.30	0.19	0.70	0.38	0.20	0.64	0.42	0.23
StDev	0.14	0.10	0.06	0.18	0.06	0.03	0.16	0.12	0.04
%CV	24.19	34.73	33.82	25.60	14.78	15.68	24.46	28.70	18.82
[ng/mL]	1.51	0.78	0.46	1.91	1.00	0.50	1.73	1.10	0.56
Dilution^a	60.40	62.15	73.71	76.30	80.27	80.62	69.19	88.35	90.16
Plate 6 serum	S16 1:40	S16 1:80	S16 1:160	S17 1:40	S17 1:80	S17 1:160	S18 1:40	S18 1:80	S18 1:160
Average	0.50	0.23	0.14	0.63	0.21	0.13	0.39	0.20	0.13
StDev	0.04	0.03	0.03	0.04	0.02	0.01	0.11	0.05	0.02
%CV	8.83	13.19	19.74	7.03	1 1.01	10.67	27.99	25.35	18.79
[ng/mL]	1.42	0.59	0.30	1.84	0.52	0.28	1.07	0.48	0.27
Dilution^a	56.61	47.23	47.57	73.70	41.91	44.52	42.73	38.01	42.79

Average, absorption measurement of four replicates of a serum sample; StDev, standard deviation.

Dilution: multiplicity with dilution factor.

Figure 10.

Linearity of automated sample dilution during one automated run of six assay plates. Bars: serial dilution of randomly chosen serum samples. Corresponding data are provided in Table 5. Graph: linear regression of standard curve. Error bars: standard deviation of two replicates.

Discussion

Due to the multiple applications of ELISAs in the fields of clinical diagnosis and biological research, a fully automated test procedure on a Beckman Coulter robot facility was developed. Because of its increasing relevance for the clinical diagnosis and research, the BDNF was chosen as a test target for the system.^{16, 17} The protein was detected via a commercially available ELISA, which was actually specified for manual operation.

Using the robot system, one is able to analyze 1–240 samples in replicates, provided in one to six assay plates per run, with a running time of 6 and a half hours to 8 hours and 45 minutes, respectively (see Fig. 5). After the assay is started, there is no need for manual interference, thus it is a truly walk-away setup. Nevertheless, due to the variety of assay layouts, this system also provides the flexibility which is demanded in the area of scientific and clinical applications.

The automated process was validated and the received data were reproducible. StDev of processed samples ranged from less than 1% to 21%. Additionally, there was no reproducible difference between assay plates in a series. Discrepancies arising from processing on varying daytimes where not detectable, indicating a reliable and stable procedure (see Table 4 and Fig. 9). The average intra-assay variance was 9% (see Table 4). According to recently published data, the intra- and interassay precision of automated immunochemical assays varied dependending on the discrete assay and target, but was ranging from 1.5% to 7%⁵ or to 23% coefficient of variation (CV).¹⁸ Putting these results into consideration, it can be deduced that the values represent dynamic biological ranges.

The native and the recombinant BDNF protein was detectable in all automatically handled serum samples. The calculated protein concentrations in test serum samples were comparable to those from manually operated assays (see Fig. 7).

The measured serum concentrations ranged from 3.27 to 95.9 ng/mL, the mean was 47.8 ng/mL. These data were in agreement with recently published data obtained from the largest study conducted on neurotrophin concentrations.¹⁹ According to this study, the average BDNF concentration in serum samples of healthy donors is 22.6 ng/mL, ranging from low levels (1.9 ng/mL) to 51.5 ng/mL. Nevertheless, some of the test sera showed increased levels of BDNF. The maximum determined concentration was 95.87 ng/mL, which means up to about twofold in comparison to literature values. Additional investigation on sample aliquots of the same donor resulted in comparable calculated BDNF concentration. Whether these unexpected values were caused by a failure during manual preparation or if this donor represents a specific phenotype cannot be answered here, because donor samples were given anonymously.

Using a serial dilution method for sample processing, StDev increased faintly with the dilution factor. But deviations of discrete samples were sometimes far exceeding the accepted range of agreement (over 30%). A close look at the row data revealed that StDev resulted from single mistaken data points (outliers), for example, one or two outliers of eight replicates. High deviations are always a problem in automation processes, especially when promoting miniaturization or more throughput. To decrease StDev and to increase throughput, high-throughput screening technologies are working preferably with commercially available one step assays (add reagent and read out).²⁰ Even on the overwhelming market of commercially available ELISA kits and products, companies try to develop and offer ELISA technologies with a reduced number of process steps, for example, precoated plates, no washing steps, and so on. However, an ELISA remains a multistep assay and therefore maximum throughput is kept lower to achieve reliability and good assay quality.

Another problem in automation of processes is the instability of solutions over longer periods of time.²¹ While setting up the assay it was shown that the BDNF standard degraded with increasing detention times leading to decreased extinctions during ongoing experiments. The aberration of standard values of plate one and six achieved 35% (data not shown). Eventually, other standards are more stable and this should be further evaluated using comparable assays. The problem can be avoided by decreasing the room temperature (below 20 °C) and refreshing the standard after half of the assay plates. Generally, to avoid such problems, it will be necessary to integrate a cooling automated labware positioner on the Biomek2000 deck.

The data presented here demonstrate that the automated robot-based workstation can be used for executing an ELISA under stringent conditions as a method to detect and determine immunological relevant parameters such as BDNF serum concentrations. The establishment of an automated ELISA has led to a stable and accurate application providing results that can be considered equivalent to manually operated assays. The results for recombinant BDNF standards and for BDNF concentrations that have been derived from healthy donors were reproducible for the automated run. As intended, the automation of the assay has shown to be a time-effective tool for future medium throughput analysis of protein concentrations.

Since this complex assay protocol was established, it is considerable to perform a broad range of ELISA assays without hardware changes and the technique can be extended to virtually all soluble protein targets and immunological questions, respectively.

Footnotes

Acknowledgments

The project was supported by the Federal Ministry of Education and Research.

References

Envall

; Permann

Enzyme-linked immunosorbent assay (ELISA). Quantitative assay of immunoglobin G. Immunochemistry 1971, 8, 871–874.

Van Weeman

B. K.

; Schuurs

A. H. W. M.

Immunoassay using antigen-enzyme conjugates. FEBS Lett. 1971, 15, 232–236.

Risch

; Hoefle

; Saely

; Berchthold

; Weber

; Gouya

; Rein

; Langer

; Marte

; Aczel

; Drexel

Evaluation of two fully automated novel enzyme-linked immunosorbent assays for the determination of human adiponectin in serum. Clin. Chim. Acta 2006, 373(1-2), 121.

http://www.beckman.com/resourcecenter/labresources/automatedsolutions/pdf/P-9763A.pdf; http://www.biotek.com/resources/tech_res_detail.php?id=164. Accessed January 4, 2007.

Rohner

; Zeder

; Zimmermann

M. B.

; Hurrell

R. F.

Comparison of manual and automated ELISA methods for serum ferritin analysis. J. Clin. Lab. Anal. 2005, 19(5), 196–198.

Villalta

; Tonutti

; Tampoia

; Bizzaro

; Papisch

; Tozzoli

; Stella

Analytical and diagnostic accuracy of the EliA automated enzyme fluoroimmunoassay for antineutrophil cytoplasmic autoantibody detection. Clin. Chem. Lab. Med. 2004, 42(10), 1161–1167.

Fujimura

; Altar

C. A.

; Chen

; Nakamura

; Nakahashi

; Kambayashi

Brain-derived neurotrophic factor is stored in human platelets and released by agonist stimulation. Thromb. Haemostasis 2002, 87(4), 728–734.

Tolkovsky

Neurotrophic factors in action—new dogs and new tricks. Trends Neurosci. 1997, 20, 1–3.

Croll

S. D.

; Ip

N. Y.

; Lindsay

R. M.

; Wiegand

S. J.

Expression of BDNF and trkB as a function of age and cognitive performance. Brain Res. 1998, 812(1/2), 200–208.

10.

Howells

D. W.

; Porritt

M. J.

; Wong

J. Y.

; Batchelor

P. E.

; Kalnins

; Hughes

A. J.

Reduced BDNF mRNA expression in the Parkinson's disease substantia nigra. Exp. Neurol. 2000, 166(1), 127–135.

11.

Michalski

; Fahnestock

Pro-brain-derived neurotrophic factor is decreased in parietal cortex in Alzheimer's disease. Brain Res. Mol. Brain Res. 2003, 111(1/2), 148–154.

12.

Yasutake

; Kuroda

; Yanagawa

; Okamura

; Yoneda

Serum BDNF, TNF-alpha and IL-1beta levels in dementia patients: comparison between Alzheimer's disease and vascular dementia. Eur. Arch. Psychiatry Clin. Neurosci. 2006, 256(7), 402–406.

13.

Braun

; Lommatzsch

; Neuhaus-Steinmetz

; Quarcoo

; Glaab

; McGregor

G. P.

Brain-derived neurotrophic factor (BDNF) contributes to neuronal dysfunction in a model of allergic airway inflammation. Br. J. Pharmacol. 2004, 141(3), 431–440.

14.

Suwa

; Kishimoto

; Nofuji

; Nakano

; Sasaki

; Radak

; Kumagai

Serum brain-derived neurotrophic factor level is increased and associated with obesity in newly diagnosed female patients with type 2 diabetes mellitus. Metabolism 2006, 55(7), 852–857.

15.

Zhang

; Chung

T. D.

Y.; Oldenburg, K. R. A simple statistical parameter for use in evaluation and validation of high throughput screening assays. J. Biomol. Screen. 1999, 4(2), 67–73.

16.

Nockher

W. A.

; Renz

Neurotrophins in allergic diseases: from neuronal growth factors to intercellular signaling molecules. J. Allergy Clin. Immunol. 2006, 117(3), 583–589 (Review).

17.

Obata

; Noguchi

BDNF in sensory neurons and chronic pain. Neurosci. Res. 2006, 55(1), 1–10 (Review).

18.

Nichkova

; Marco

M. P.

Biomonitoring human exposure to organohalogenated substances by measuring urinary chlorophenols using a high-throughput screening (HTS) immunochemical method. Environ. Sci. Technol. 2006, 40(7), 2469–2477.

19.

Lommatzsch

; Zingler

; Schuhbaeck

; Schloetcke