October 04, 2004
Author: Paul Held PhD, Senior Scientist, BioTek Instruments, Inc
Today’s biomedical testing has resulted in sample sizes becoming smaller and smaller, driving the need to measure samples with ever-lower detection limits. The quantitation of nucleic acids is no exception. While other methods may be more sensitive, the ease, simplicity, and noninvasiveness of direct UV absorbance measurements has made it the most popular method of nucleic acid quantitation. The use of UV absorbance measurement to quantitate nucleic acids is the de facto standard by which all other methods are measured. Here we describe the limit of detection for nucleic acid quantitation using BioTek’s UV-Vis microplate spectrophotometers.
Limit of Detection
There are several different ways to define the limit of detection for an assay-instrument combination. Regardless of the method, the intent is to provide a means to reliably insure a differentiation of a sample containing an analyte and the buffer only control. One method used is interpolating a concentration curve with the value of the blank mean plus 3 times the standard deviation of the blank. The resultant interpolated concentration is said to be the limit of detection. Samples with signals greater than the interpolated value can be detected reliably. Another related method is known as a signal to noise ratio - the signal of a known standard minus the blank divided by the total standard deviation of the standard and the blank. Concentrations that result in a signal to noise ratio of 2.0 or greater are considered to be significantly different from the blank. This is the method that BioTek uses to identify detection limits for fluorescent compounds. A less stringent test often used is the Student’s T-test, which compares the probability that two populations are different. This test also uses the means and standard deviation of blanks and samples to determine if the populations of data are different. Probabilities of 0.05 (p< 0.05) or less are considered being different.
Theoretical Lowest Concentration
The smallest absorbance unit capable of being reported by the BioTek Microplate Spectrophotometers (e.g. PowerWave™ HT, PowerWave™ XS, Synergy™ HT, and µQuant™) is 0.001 ODs. If we use BioTek’s 1 cm quartz Bio-Cell™ to make this measurement then 0.001 would also be the smallest measurable absorbance reading for a 1 cm pathlength. If we imagine a sample in the Bio-Cell™ that absorbs 0.001 OD above a blanked value and multiply by the extinction value of 50 (1 OD equals 50 µg/ml), the concentration becomes 0.050 µg/ml or 50 ng/ml. However, in a well, the pathlength is usually less than 1-cm, but 0.001 OD is still the minimum value to be reported. For example, if a 200µl sample is used, the pathlength is about 0.5 cm so the conversion to a 1-cm pathlength would result in a corrected absorbance two times the raw value. The minimum reported value is still 0.001 regardless so the pathlength corrected value would be 0.002 and the concentration would be 0.002 x 50 or 0.100 µg/ml or 100 ng/ml which equates to 20 ng/well (100 ng/ml x 0.2 ml). However, these calculations assume that each measurement is perfect, that there are no errors inherent with the measurements, and each measurement is completely repeatable.
Hypothetical DNA Measurements from Different Well Contents Using Various Fluid Volumes
300 µl volume
200 µl volume
|100 µl volume|
|(ng/well)||Conc. (ng/ml)||Raw OD||Corr.* OD||Conc. (ng/ml)||Raw OD||Corr.* OD||Conc. (ng/ml)||Raw OD||Corr.* OD|
*Based on an estimated path length of 0.75 cm, 0.5 cm and 0.25 cm for 300 µl, 200 µl, and 100 µl well volumes respectively.
Concentration vs. Content
A common misconception occurs when comparing detection limits. This misconception revolves around the differences between concentration and well content. Concentration is an amount of absorbing material per unit of volume (e.g. µg/ml or ng/µl). Content is an amount (weight) in the well (ng/well) and is independent of volume. The two are often referred to interchangeably when they actually are not interchangeable. The subtle difference between the two is that content will change with an increasing volume of a constant concentration. For example a well that has 200 µl of a 1000-ng/ml solution has a content of 200 ng/well, while the same well that has 100 µl of a 1000-ng/ml solution will have a content of 100 ng (Table 1). Note that the raw ODs for each well will be different, while the pathlength corrected absorbance and concentration calculation will theoretically return the same values, as pathlength correction normalizes everything to a common 1 cm pathlength. Likewise, two wells that have different concentrations, but have the same content, would have the same raw absorbance. A well that has a content of 500 ng/well in 300 µl (1667 ng/ml) might have an absorbance of 0.025 ODs and another well also has a content of 500 ng/well, but in 200 µl would also return an absorbance measurement of 0.025 ODs, but the calculated concentration would be 2500 ng/ml (Table 1).
Figure 1. Absorbance of Various DNA concentrations with Different Sample Volumes. The absorbance at 260 nm (after blanking) of various DNA concentrations were plotted and linear regression analysis performed for two different sample volumes. Error bars represent a +/- 0.002 ODs from the mean at each determination.
Larger Volume per well equals greater pathlength
When trying to measure low concentrations, the longer the pathlength the better the measurement will be. Longer pathlengths result in a larger signal if the concentration is the same. In microplates the pathlength of the sample is dependent on the volume of fluid in the well. This means that a well with 300 µl of a DNA solution will have three times the absorbance as a well with 100 µl of the same solution after background subtraction. Besides increasing the volume of the well, one can reduce the size of the well to increase pathlength. Half area 96-well plates or 384-well plates will provide a greater fluid pathlength than a standard plate with the same fluid volume.
Blanking: Pre-read vs. identified blanks
Due to the background absorbance of the microplate, it is absolutely necessary to perform some sort of blanking on the raw data at 260 nm in order to achieve accurate results. In regards to blanking the plate there are two options: pre-read an empty microplate (preferably the plate about to be used in the assay) prior to measuring samples experimentally or define one or more wells as blanks using the plate layout. While it is more convenient to define one or more wells as a blank, this method assumes that the background for each well is the same. In reality there may be as much as a 0.005 difference between the highest and lowest well. In general, the commonly used diluents used in conjunction with nucleic acids do not have any appreciable absorbance at 260 nm, with all the background absorbance the result of the microplate itself. As a matter of practice using a blank well or two will provide acceptable results in regards to nucleic acid quantitation. If possible it is recommended that both a pre-read and a blank well(s) be defined. This combination will accommodate variations in plate background as well as any influence that the sample buffer may provide.
Errors can add up
While the theoretical limit to the measurement is 0.001 OD units, there are a number of errors inherent to the measurement. These errors have the potential to add up. Reader specification for repeatability is +/-1% +/- 0.010 ODs, which at very low ODs means that the 1% component essentially drops out of the equation, which leaves an error of +/- 0.010 ODs. Typically BioTek’s readers perform much better than specification so we can estimate accuracy at +/-0.002 for this paper. In addition to the reader variability, well-to-well variability in the plate is approximately +/- 0.002 ODs, which represents a total variation of 0.004 OD across the plate. The meniscus at the air-liquid interface has the potential for a variability of +/- 0.001 ODs as well. Note that the error caused by the well-to-well variability and the meniscus can be corrected for using the pre-read functionality and blank well subtraction respectively. This effectively limits the error to the error of the reader, which we have estimated at +/- 0.002 ODs.
Figure 2. Absorbance of Wells with Different Total DNA content. Data from wells with differing volumes and DNA concentrations were pooled to provide a plot of DNA content versus absorbance. Error bars represent a +/- of 0.002 ODs from the mean of each observation.
Pathlength, sample volume and detection limits
Relationship between sample volume and pathlength can confuse the issue in regards to detection limits as measured by concentration. Because of this, the preferential method is to use well content (ng/well) as a means by which detection limits can be defined. With vertical photometry the absorbance of a well content is the same regardless of the volume. In other words, the absorbance of 500 ng of DNA in a well is the same if it is in 100, 200 or 300 µl. Note that the DNA “concentration” would be 5 µg/ml, 2.5 µg/ml, and 1.67 µg/ml respectively. As demonstrated in Figure 1, when various concentrations of DNA are quantitated, 300-µl samples can be distinguished from the buffer only blank at lower concentrations than the 100-µl samples. Please note the use of “concentration” in this example. Again this relates to the total DNA content on a per well basis, as large volumes of a fixed concentration will result in a larger content of absorbing material.
The Bottom Line
As seen in Figure 2, when the absorbance was plotted against the total DNA content on a per well basis, a linear relationship between content and absorbance is observed. Using the estimated error of measurement of +/- 0.002 OD one observes that the upward error bars from the blank do not overlap the downward error bars of the 100 ng/well samples. Therefore, in regards to detecting DNA in solution using absorbance, the minimum amount that can be detected is about 100 ng per well under the best of circumstances. While samples with lower per well content have a mean absorbance that is higher than that observed with the blanks, the difference is not outside the error of measurement. Nucleic acids, such as RNA or ssDNA can be measured using absorbance in the same manner as dsDNA. Because these compounds have a higher extinction coefficient than dsDNA, the detection limits would be expected to be lower. RNA has an extinction coefficient 20% higher than dsDNA. This suggests that the detection limit for RNA would be approximately 20% lower than dsDNA or 80 ng/well. Likewise ssDNA such as oligonucleotides have an even higher extinction coefficient. The detection for these molecules can be calculated to be approximately 67 ng/well.
There are a number of factors that have the ability to prevent one from achieving these detection limits. High quality DNA is absolutely essential for good detection limits. DNA solutions and buffers containing protein contamination or particulates will often result in a greater variability of replicates, which is observed as higher standard deviation or %CV values. Using sufficient volumes for the well type being used is also critical. Volumes that do not completely fill and cover the bottom of the well will result in aberrant results. Standard 96-well plate wells require a minimum of 100 µl. Half area 96-well plates can be used with as little as 50 µl and samples in 384-well plates can be measured using 25 µl. Dirty or scratched microplates can affect the variability of samples as well, particularly if the plate pre-read feature of the PC software (KC4 or KCjunior) is not utilized. Proper storage of the unused plates can minimize this potential problem. Inaccurate pipetting of samples can lead to unintended inaccuracies that cannot always be corrected using pathlength correction. Good laboratory practices, such as maintaining buffers and solutions contaminate free, are also essential for obtaining accurate and precise nucleic acid quantitation. This is particularly important at very low concentration levels.