We've updated our Privacy Policy to make it clearer how we use your personal data. We use cookies to provide you with a better experience. You can read our Cookie Policy here.

Advertisement

8 Top Tips for ELISA Success

A diagram of three different types of ELISA.

Favored for its simplicity, specificity and cost-effectiveness, ELISA is a useful technique used in many laboratories globally. ELISA kits for a wide range of analytes can be purchased commercially, although these tend to be expensive and money is often tight, particularly in a research setting. This leads many laboratories to develop their own assays, particularly when working to a budget or where many samples need to be analyzed. Despite the ease of use of the technique as a whole, there are still a number of factors to be considered in order to design robust and reliable assays. Here are some that I have experienced along the way and would like to share in the hopes of saving you time, stress and money, and helping you to achieve the best results possible!


1.Choose your surface wisely

Due to the diversity of proteins used for coating in ELISA, there are a wide range of specialist microplates available which have a variety of binding properties and capacities. When developing a new assay, a range of surfaces should be assessed to determine which is optimal for the antibody/antigen to be tested. Most plates are made from the same types of plastics such as polyurethane or polyvinyl chloride, but they are made with different binding properties. It is worth trying several plastics with low, medium and high binding capabilities at the very least. Cost is also an important consideration when selecting a plastic, particularly if a large number of samples are to be tested. Furthermore, the molecule used for coating the plate should be titrated properly to determine the best concentration to use in the assay.

2. Chequerboard

Whilst it may seem sensible to use high concentrations of antibodies in an assay, particularly when the molecule being detected is not very abundant in a sample, this can be counterintuitive. Proper titration of all antibodies to be used in the assay, as well as the sample, ensures that the signal to noise ratio is optimal; the aim is to have the lowest possible background with the highest possible optical density (OD) for a well-designed and sensitive assay. Where too much antibody is used, the background noise will be high due to unspecific binding. Where concentrations are too low, the assay will lose sensitivity. An effective way to test this is to carry out a range of serial dilutions of the molecule to be used for coating, the sample, and the secondary antibodies (plus any other antibodies, such as when designing a sandwich ELISA) and test these against each other, with a negative control for each condition. This allows the optimal concentration of each to be determined in several simple experiments. Multiple types of plastics can also be analyzed at the same time.

3. Proper sample dilution is key
 

Proper dilution of samples is often overlooked when investigating assay to assay variability. Where a sample needs to be diluted before use, such as when using serum, stepwise dilution is pivotal for minimizing pipetting error. For example, if a chequerboard experiment has determined that sera should be diluted 1:100 before addition to the plate, it is wise to carry out a 1:10 dilution of the sample, followed by a subsequent 1:10 dilution. Whilst this may seem obvious, this is often not carried out and can create huge variation between experiments, which is frustrating for the researcher and can be avoided very simply, improving assay reliability. Improper dilutions can also create misleading results when comparing samples in the same experiment. For instance, if two serum samples are being compared to determine whether seroconversion has occurred, improperly diluted samples could mean that seroconversion is missed if too much of the pre-conversion sample is added due to poor dilution technique. 

4. All antibodies are not created equal
 

Whilst it may not always be possible, where more than one antibody to the protein of interest is available, it is best to test each one. There are many considerations to make when choosing antibodies. Polyclonal antibodies have a wider target range, and are usually cheaper, but may decrease the specificity of an assay. Monoclonal antibodies are often more expensive but are more specific provided that the epitope used by the antibody is visible on the antigen of interest (but may not bind at all if it isn’t). Often the only way to test this is to try a range of antibodies from a variety of manufacturers. Thankfully, many manufacturers offer small sample sizes for this purpose, or are willing to provide these if asked. When selecting pairs to use in sandwich ELISAs, this choice becomes even more important, as the antibodies must recognize different epitopes to each other.

5. Beware of blocking surprises

Many investigators trial a range of blocking agents during the development stages of an assay and make the choice to use that which gives the best signal to noise ratio. A factor which is often under considered when making this choice is batch to batch variation of particular blockers. Blockers are often biological agents, such as bovine serum albumin (BSA). BSA is a very effective blocker in a broad range of assays, but due to the nature of the reagent it can vary hugely between batches, with some batches containing large amounts of cross-reactive antibodies that will interfere with an experiment, and some lots containing little to none. Therefore, if a blocker such as BSA or foetal calf serum is chosen it is prudent to test several batches to check for consistency in your results before committing to a given one. This is an important consideration, and for unlucky individuals can often be the cause of many failed experiments once the batch tested initially has been used up! Where batch to batch variation is seen, but blocker choice cannot be modified, often manufacturers are happy to hold particular batches aside if they are consulted and have an idea of how much will be used. 

6. Batch testing of secondary antibodies

Similar to blocking agents, batches of secondary antibodies can also vary. Whilst their overall activity may not change, the amount needed can vary between lots. When buying in another batch of secondary antibody, it is worth running a simple experiment with the previous lot to make sure the concentration used is still appropriate. The assumption that the same dilution can be used can sometimes mean that background noise suddenly becomes much higher when the batch is changed, which can be a nasty surprise!

7. Don’t forget to normalize!
 

Often overlooked, normalization takes any variation between plates/experiments and minimizes its effects on overall results. A simple way to normalize between experiments where serum is used as the analyte, is to compare the result for each serum sample to that of the negative serum control. By doing this, any variation from outside factors such as room temperature and development time is minimized. By dividing the OD of the test sample by the OD of the control, results can be expressed as a fold change. Whilst this isn’t absolute quantification, often an absolute number isn’t necessary, but a robust set of results can still be obtained, and samples can be compared to each other.

8. Know your dynamic range

When designing an experiment where a standard curve is used to determine the concentration of samples, it is important to stay within the dynamic range of the instrument you are using to read your plates. These values will be specified in the manufacturers manual; the upper limit for many machines is ~1.5 OD. When preparing a standard curve, the top standard (i.e. the most concentrated) must have an OD below the highest of the dynamic range. By using standards that sit above this range, extrapolation of results against the standard curve becomes risky, as any OD above this number may be inaccurate Many newer models of plate reader are able to adjust the dynamic range, but if you are using a model where this is not possible, maximal OD will need to be taken into account in standard curve design.  All samples must also fall within this dynamic range in order to be accurately measured and the concentration ascertained, so sample dilution too must be considered. This may be frustrating initially, but once the operator has a feel for the concentration range in a batch of samples, it is likely a standard dilution of sample can be used that fits within this range and gives good results for future experiments.