ELISA theory and saturation - (Nov/14/2011 )

I have a question about the theory of ELISA assay quantification.

It is my understanding that, using the TMB substrate (or any other HRP substrate), the analyte is actually a product of the HRP which is directly bonded to your secondary antibody. So, in a sense, you don't really read "quantity of secondary antibody" you read "quantity of analyte that is directly proportional to quantity of secondary (in the linear range of the assay).

So, what exactly causes the saturiation (end of the linear range) in an ELISA assay? Is it the capabilities of your spectrophotometer? Is that all your substrate ends up converted into the analyte by the HRP?

I don't understand what linear range means, let's talk instead about dynamic range. That is the concentration range of analyte that results in a change in response (color intensity in this example). The useful dynamic range is the analyte concentration range that produces a useful response range. The useful range can be tested by assessing control sample replicates at multiple concentrations and backcalculating their response values against a standard curve achieving good precision and accuracy. The standard curve is usually fitted by some non linear regression approach.

What affects the response range and response saturation?

In a colorimetric assay, the rate of conversion of substrate to coloured product should be proportional to the amount of enzyme within the well which in turn is proportional to the amount of bound labelled antibody, but there are limits to this. The plate reader often will not read above 3 or 4 OD units, so high concentration samples may all read the same OD, but the actual amount of colour in the well may increase with increasing analyte concentration over the range that the plate reader is maxed out. This is where fluorescence or ECL may give a wider dynamic range. Also, the enzyme ends up getting saturated after a while and development times (for TMB) more than about 20 minutes are not useful. Also, if insufficient substrate is in the well, it will all get converted before the maximum plate reader OD is reached and result in the top end of the curve flattening off.

If the labelled detection antibody is in excess (that is at a sufficiently high concentration and well above the Kd to saturate all the potential analyte binding sites) then the reaction rate will be limited by the amount of analyte captured within the well. If the labelled antibody is limiting (not enough molecules to saturate the bound antigen and or close to or lower than the Kd), the reaction rate will be capped by the available labelled antibody and maximum (now reduced) response may be shifted to lower analyte concentrations.

The reaction rate will also be affected by the concentration of the capture reagent in the well. If this is in excess relative to the analyte concentration, then we can expect all of the analyte molecules to bind. If the concentration of the capture reagent is reduced, then saturation of the capture reagent may occur at lower analyte concentrations resulting in reduced and flat signal at high analyte concentrations.

We can play around with coating antibody and detection antibody concentrations to make them limiting/in excess and this will shift the dynamic range of the assay one way or another so that the assay can be tailored to needs. The maximal potential sensitivity of an assay will be governed by the binding affinity between the capture/detection reagents and the analyte.

With all this said, the best way to figure this out is by experimentation, as many less obvious factors may influence the response at any given analyte concentration.

Ben Lomond on Tue Nov 15 03:55:29 2011 said:

Let me explain a little as to why I am asking such trivial questions:

In my previous work, I used to keep all my Absorbance values (pertaining to ELISAs) under OD = 1.0. The reason being is that I knew that in this range I was well below the plate reader's saturation level. My exposure of TMB on my wells was about 2min followed by STOP solution and I was getting really nice results.

Currently where I am now, I see ELISA results reported with Absorbances much higher; in the rage of 2.0 - 5.0 and the TMB exposure of the wells is 30min. My first reaction to this is that samples should be diluted or the TMB time reduced. But, I was wondering if the machine being used to read the plates allowed for the reading of such high OD values to be accurate. The machine currently used is the Perkin Elmer Envision.

Thus, I was wondering if the value at which the plate reader "maxes out" is determined by experimentation or if there is a standard for this value.

I actually started a topic regarding this very question over at the Envision question and answer forums:

http://www.labwrench.com/?community.posts/threadNo/11354/subject/Highest-reliable-OD-reading/

Most of my work is with Molecular Devices plate readers and these will not give values above 4 OD units, the Envision probably has a similar cut off. With assays using acid stopped TMB, with the high standards flat lined at 3.5 to 4 units, it may be possible to differentiate between them by reading the plate at 425 nm instead of 450 nm, so on the shoulder of the yellow peak. In this case, the measurement of the coloured product is limited by the reader. Also, there may well be differentiation at 650 nm prior to adding the acid to get the 450 reading.

Not an expert on enzyme kinetics, but the oxidation of TMB by HRP has an initial rate that then slows with time regardless of the HRP concentration in the well. There is a linear range for the reaction rate, and this is where kinetic readings are most useful and some TMB formulations are designed to have linear reaction rates for longer for this purpose.

For the end point TMB assays I develop, I aim for a reaction time of 10 to 20 minutes, and monitor the plate at 650 nm and add acid when the high standard reaches about 1.2 OD units. . This equates to about 2.5 units at 450-650 nm. In my experience, with the kind of plate reader that I use, ODs above 3 result in a lack of precision, which adds randomness in the identification of samples that have very high analyte concentration requiring dilution, as opposed to samples that are actually within the top of the curve range.

The problem with very rapid development (2 minutes or so) is it is difficult to monitor colour development, and drift can result from small variations in development time across the plate. Long developments of 25 minutes or more often result in a decreased signal to noise ratio.

I would suspect that ODs in the range of 5 are not robust, and would recommend monitoring the plate and stopping when the top of the standard curve reaches 1.2 units or so at 650 nm.