AppNote Evaluating Microplate Detection Instruments Introduction Important features to consider when evaluating a microplate reader are the following: sensitivity, dynamic range, throughput, integration (into software and hardware of current lab), ease-of-use, reliability/serviceability, and analytical flexibility. This AppNote helps you to design objective tests that evaluate some of these features so you can accurately compare different readers. Evaluating instrument specifications will not necessarily predict whether your assay will perform with the desired sensitivity, speed, and dynamic range. Data from designed experiments that test your assay needs are required. Follow these general guidelines in your overall evaluation plan: Map out the overall performance test plan in advance as time constraints can limit test refinement. Control as many variables as possible e.g., microplate background, pipetting precision. Choose the instrument settings, interference filters, and dichroics very carefully. Evaluate instrument performance by objectively testing the critical features mentioned above. Refrain from using assay development or unique instrument modalities as the basis of a test plan. Note Often the motivation for evaluating a new instrument is the availability of a new feature or modality. Rather than embarking on a development project to test the instrument s performance using the new modality (a costly endeavor), discuss your interests with the vendor. They may be able to suggest test methods or refer you to other customers who have had experience with the relevant features. Evaluation Strategy Determine the Evaluation Criteria When determining the evaluation criteria consider the following requirements for current and future applications: sensitivity, dynamic range, throughput, hardware and software compatibility, and a target budget for both acquisition and on-going costs. Evaluating Microplate Detection Instruments 1
Plan the Evaluation When planning the evaluation, do the following: Write down your requirements. Define the experiments you will need to gather the required information. Estimate the evaluation time available and prioritize the features to be tested. Eliminate redundancies as much as possible. Try to evaluate multiple features using a single microplate. Design Objective Tests When evaluating multiple microplate readers, design objective tests as follows: Run the same test plates on each system, if possible. Compare instruments by designing easily reproducible, objective tests. Use objective measurements such as lower limit of detection (LLD), imprecision, linearity, and read speed. These are useful because they are readily reproduced at a later date on other instruments. Use assay plates that will be available for evaluating other systems later on. Design Controlled Tests When evaluating microplate readers, design controlled tests as follows: Determine sensitivity and dynamic range by using serial dilutions that extend beyond the expected working range. Include positive and negative controls. Include background wells that combine all assay components except fluorophore. If the plate contains wells with a variety of compositions, in addition to variations in label concentration, include multiple background wells representing these variations.for example, if there are several concentrations of (unlabeled) cellular membranes present, there should be a background well for each concentration. Collect Statistically Significant Data One of the most important parts of an evaluation is analyzing the precision of measurements. Collect statistically significant samples by running an appropriate number of replicates. This is accomplished by calculating the standard deviation or the coefficient of variation of replicate measurements (CV in % = 100 * SD/ mean). It is important to weigh the statistical significance of the estimate of precision, as the estimate of the noise can be very noisy. For example, if you find an 8% CV for triplicate measurements on instrument A and only 2% for the same wells on instrument B, how confident are you that B is better than A? A fairly rigorous statistical analysis indicates that for triplicate measurements, the estimate of the standard deviation is likely to be off by about 50% of the calculated value, or 4% in this case. The precision estimates for the two instruments differed only by 8%-5%=3%, so the evidence that B is better than A is quite weak. Although there are rigorous statistical treatments of these issues, the essential point is that precision is harder to measure than mean value. The uncertainty in the measured precision is large when the number of replicates is small; try to follow the guidelines below. Read as many replicates per plate as possible. Make multiple plates for critical precision measurements. Look for overall trends in precision. The uncertainty of the precision estimate for 8 replicates is about 25% of its measured value; for 16 it is about 18%. A few outlier wells can mask the true instrumental performance, particularly at low label concentrations Evaluating Microplate Detection Instruments 2
where microplate background is relatively important. For example, in a series of 8 replicates with a 5% CV, if you replace just one of the data points with an outlier that is 25% high, the CV more than doubles. Distinguish Background from Noise Background is not the same as noise. The background signal is the contribution to the measurement from sources other than the fluorescent label. It is most easily seen taking a measurement on a well containing all test components except the fluorescent label. Background signals may arise from the microplate plastic, solution contaminants, leakage of light through the optical filters, or other sources generated in the instrument. Noise relates to the uncertainty in measurements and backgrounds are not necessarily very noisy. If the background is highly predictable i.e., constant from well to well, it can largely be eliminated by subtracting the signal from a control well lacking fluorescent label. Subtracting this constant signal will yield useful information from wells generating signals that are close to background levels. As a practical matter however, all background signals have some noise. Noise usually determines the ultimate sensitivity of an instrument. A common measure of the sensitivity of a fluorescence-intensity measurement is the statistical LLD for a fluorescent label. The LLD is usually defined as the lowest concentration of the label producing a signal that is statistically distinguishable from background: a signal that is equal to three standard deviations of the background signal, after subtraction of the mean background signal. Use Appropriate Optical Filters Filters are a critical component of any optical microplate reader. When choosing a filter follow these recommendations: Make sure that you have interference filters with an appropriate wavelength and bandpass for the test application. Reduce crosstalk by using filters with a small bandwidth. Choose the correct filter very carefully with fluorophores that have small Stokes shifts. These fluorophores such as the Cy dyes, have excitation and emission spectra that are very close together. Use equivalent filters when you are comparing two instruments. Some filters may need to be special-ordered; give the vendor time to obtain the correct filter sets. Examine Assay Cost The two primary ways to reduce the cost-per-assay is to scale the volume down and reduce the concentration of compounds and reagents. For example, a simple volume reduction from 200 µl to 50 µl can reduce per-assay costs 75% without making any changes to the protocol. Greater savings are possible by reducing the concentration of the overall assay, reducing the amount of compounds, fluorophore-labeled reagents and other costly components. Test instrument performance at several assay volumes by preparing master 96-well plates and dispensing into 96- and 384-well plates with lower volumes. Check with the microplate vendor to determine what volumes are appropriate for each plate type. Estimate the per-assay cost savings by estimating the total number of assays per year and multiplying by the reduced volume. Remember that not all reagent costs scale linearly due to batch sizes and other efficiencies. Evaluating Microplate Detection Instruments 3
Test Model Systems and Real Assays Should you test with simple model systems or real assays? Each has its advantages and disadvantages; a comprehensive test includes both. The hallmarks of a good model system benchmark test are the following: Use of universally available reagents Preparation with a minimum of manipulation Design with inherent well-to-well precision An example of a model system would be testing the sensitivity of fluorescence-intensity measurements using a serial dilution of sodium fluorescein. The aim would be to obtain as pure and reproducible a measurement of instrumental performance as possible, uncontaminated by well-to-well or day-to-day imprecision in the test chemistry. Although the advantages of testing with a real assay are obvious, there are some real disadvantages. Some of the potential problems are the following: Typical assays may not test instrumental performance under as wide a range of conditions as a well-crafted benchmark test. Assay plates are likely to show more variability than those in a benchmark test. Some guidelines for testing with a real assay are the following: Wherever possible, work with well characterized reagents to minimize assay plate variability. If you are forced to work with poorly characterized reagents, try to arrange a side-by-side evaluation among competitive instruments. To control for plate-to-plate variation, compare instruments side by side, reading the same plate in both instruments. The important point to remember is that when you compare the performance of two instruments on separate test plates, you risk having the outcome of the evaluation depend on the relative pipetting precision on the plates as much as on the relative performance of the instruments. Evaluating Microplate Detection Instruments 4
Example 1: Evaluating fluorescence intensity and fluorescence polarization using a serial dilution of sodium fluorescein Introduction This example describes how to make up a serial dilution of sodium fluorescein from 1 pm to 100 nm in half-log intervals, suitable for testing fluorescence intensity and fluorescence polarization in several microplates with 8 or 16 replicates per plate. Equipment and Materials Preparation of dilutions The following materials and equipment are required: 1 ml of 10 mm sodium fluorescein in deionized H 2 0 or phosphate buffered saline (PBS), ph 7.4, 10 to 50 mm phosphate Note The intensity of fluorescein fluorescence is ph dependent (with a pk just below 7.0), therefore a buffer with a higher ph will improve sensitivity at the expense of moving out of the physiological ph range. About 150 ml of PBS buffer One 96-well black flat bottom plate (e.g., Costar # 3915) One 384-well black flat bottom plate (e.g., Costar # 3710) 50-mL tubes for making dilutions 20-mL tubes for making dilutions 5-mL tubes for making dilutions 1.5-mL tubes for making dilutions Pipets suitable for dispensing in the range 0.010 ml to 10 ml Vortex mixer This preparation yields 10 ml each of 0 nm and 0.001 through 100 nm in half-log intervals. To prepare dilutions : 1 Label twelve 20-mL tubes with the numbers 1 12. 2 Label one 1.5-mL tube 100 µm. 3 Add 10 ml of PBS to tubes 1 11. 4 Add 14.6 ml of PBS to tube 12. 5 Add 0.990 ml of PBS to the 100 µm tube labeled in step 2. 6 Create a 100 µm working stock solution by adding 10 µl of the 10 mm stock to the 100 µm labeled tube. Evaluating Microplate Detection Instruments 5
To prepare dilutions (Continued): 7 Follow the table below to make the remaining dilutions. Mix each tube thoroughly after adding fluorescein dilution. Tube # Volume to add of PBS (ml) Tube # or stock to use Volume to add of tube or stock Final conc. (nm) 12 14.6 100 µm 0.0146 100 11 10 12 4.625 32 10 10 11 4.625 10 9 10 10 4.625 3.2 8 10 9 4.625 1 7 10 8 4.625 0.32 6 10 7 4.625 0.1 5 10 6 4.625 0.032 4 10 5 4.625 0.01 3 10 4 4.625 0.0032 2 10 3 4.625 0.001 1 10-0 0 Preparation of Microplates To prepare a 96-well plate dilution series: 1 Choose a volume between 50 300 µl to be dispensed into each well. 2 From tube 1 dispense the chosen volume into each well in column 1 of the plate. 3 From tube 2 dispense the chosen volume into each well in column 2 of the plate. 4 Repeat for tubes 3 12 into columns 3 12 respectively. To prepare a 384-well plate dilution series: 1 Choose a volume between 20 100 µl to be dispensed into each well. 2 From tube 1 dispense the chosen volume into each well in columns 1 and 2. Evaluating Microplate Detection Instruments 6
To prepare a 384-well plate dilution series: (Continued) 3 From tube 2 dispense the chosen volume into each well in columns 3 and 4. 4 Repeat for tubes 3 12 and columns 5 24, respectively. Typical Results Figures 1 and 2 show typical results: 1.E+08 6.0 Fluorescence Intensity (relative Units) 1.E+07 1.E+06 1.E+05 1.E+04 1.E+03 Intensity Intensity - Background CV(%) 5.0 4.0 3.0 2.0 1.0 CV (%) 1.E+02 0.0 0.001 0.01 0.1 1.0 10 100 Fluorescein (nm) Figure 1. Measurement of fluorescence intensity on serial dilution of sodium fluorescein in a 96-well plate using an Analyst instrument. The instrument was set in comparator conversion mode with no neutral-density attenuation, a maximum integration time of 100,000 µsec, and 300 µl /well. Background was defined as the reading from a well without fluorophore. The LLD, defined as the concentration of fluorescein producing a signal three standard deviations above background, was <3 pm. 50 Fluorescence Polarization (mp) 40 30 20 10 0 0.01 0.1 1.0 10 100 Fluorescein (nm) Figure 2. Measurement of fluorescence polarization on serial dilution of sodium fluorescein in a 96-well plate using the Analyst instrument. The instrument was set in comparator mode with no neutral-density attenuation, a maximum integration time of 100,000 µsec, and a well volume of 300 µl. Error bars indicate one standard deviation. Polarization values were calculated by subtracting background from the intensity measurements. Errors became large at low concentration as background noise dominated the intensities. The rise in noise at high concentrations was related to the high light intensities reaching the photomultiplier tube and could be suppressed by including one of the instrument s neutral-density filters in the optical path under these conditions. Evaluating Microplate Detection Instruments 7
Example 2: Evaluating the uniformity of fluorescence intensity and fluorescence polarization across a microplate using sodium fluorescein Introduction This example describes how to make up a dilution of sodium fluorescein in quantities suitable for testing the uniformity of fluorescence intensity and fluorescence polarization across several microplates. Some types of imprecision, such as pipetting error, variations in microplate background, and electronic noise, will usually not have a consistent spatial pattern. Others, such as errors in microplate registration and movement, will have a distinct pattern. If you wish to emphasize instrumental sensitivity, use a low concentration of fluorophore; if you wish to emphasize positioning accuracy, use a high concentration of fluorophore. The example provided here uses an intermediate concentration, 300 pm fluorescein, which gives a signal at least 100X background in Costar #3915 96-well microplates using an Analyst instrument. Materials and Equipment Preparation of Dilutions The following materials and equipment are required: 1 ml of 10 mm sodium fluorescein in deionized H 2 0 or phosphate buffered saline (PBS), ph 7.4, 10 to 50 mm phosphate. Note The intensity of fluorescein fluorescence is ph dependent (with a pk just below 7.0), therefore using a buffer at a higher ph improves sensitivity at the expense of moving out of the physiological ph range. 60 ml of PBS buffer One 96-well black flat bottom plate (e.g., Costar # 3915) One 384-well black flat bottom plate (e.g., Costar # 3710) 50-mL tubes for making dilutions 1.5-mL tubes for making dilutions Pipets suitable for dispensing in the range 0.010 ml to 10 ml Vortex mixer The following preparation is enough for at least one 384-well or two 96- well plates. To prepare the dilutions : 1 Label three 1.5-mL tubes 300 µm, 3 µm, and 30 nm. Label one 50-mL tube 300 pm. 2 Add 0.970 ml PBS to the 300 µm tube. 3 Add 0.990 ml PBS to the 3 µm and 30 nm tubes. 4 Add 49.5 ml PBS to the 300 pm tube and mix thoroughly. 5 Add 30 µl of the 10 mm fluorescein to the 300 µm tube and mix thoroughly. 6 Add 10 µl from the 300 µm tube to the 3 µm and mix thoroughly. Evaluating Microplate Detection Instruments 8
To prepare the dilutions (Continued): 7 Add 10 µl from the 3 µm labeled tube to the 30 nm tube and mix thoroughly. 8 Add 0.5 ml from the 30 nm tube to the 300 pm tube and mix thoroughly. Preparation of Microplates To prepare a microplate: 1 Choose a well volume and a microplate. 2 Pipet the same volume of solution from the 300 pm tube into each well. 3 Check to make sure that the meniscus are uniform and even them out by gentle tapping if necessary. Note This is supposed to be a test of instrumental performance, not your pipetting precision, so pipet carefully! Alternatively, if you would like to see how fluorescence polarization corrects for meniscus effects compared to fluorescence intensity, purposely introduce meniscus variations. Typical Results Figure 3 shows typical results: 6 % Deviation from the Mean 4 2 0-2 -4-6 A1 P24 Wells (by rows) Figure 3. Variations in fluorescence intensity measured across a 384-well microplate Data were taken with 300 pm fluorescein, 100 µl/well, using an Analyst instrument in photon-counting mode, and 100,000 µsec integration time. The order of the wells is A1, A2, A24, B1, B2, P24. The standard deviation of the entire set of measurements was 1.6%. Evaluating Microplate Detection Instruments 9
Summary The purpose of this AppNote is to help you with the design of a test strategy for evaluating microplate readers. Included in this publication are examples of practical experiments that can be used in the evaluation process. The essential elements of a microplate reader evaluation are given below. Examine the following features when evaluating a microplate reader: sensitivity, dynamic range, throughput, integration (into software and hardware of current lab), ease-of-use, reliability/serviceability, and analytical flexibility. Use the following strategies to assess the essential microplate features: determine the essential criteria of the evaluation, plan the evaluation, design objective tests, design controlled tests, collect statistically significant data, distinguish background from noise, use appropriate optical filters, examine assay cost, and test model systems and real assays. Molecular Devices Corporation 1311 Orleans Drive Sunnyvale, CA 94089 USA Email: info@moldev.com www.moleculardevices.com Sales Offices USA 800-635-5577 UK +44-118-944-8000 Germany +49-89-9620-2340 Japan +81-797-32-2877 Check our web site for a current listing of worldwide distributors. Lit# 0120-1251 Analyst, Acquest, Criterion, HE, HEFP, ScreenStation, and SmartOptics are trademarks of Molecular Devices Corp. 2001 Molecular Devices Corporation. Printed in USA. 3./01 Evaluating Microplate Detection Instruments 10