direct ELISA help - (Jul/01/2009 )

I was assigned to develop direct ELISA to detect antigen in a sample and I encountered following problem. From my checkerboard I chose antibody dilutions which gave me high signal to noise ratio (1 in 1000) and concluded that calibration plot should be somewhere in between 156 ng/ml and 39 ng/ml as OD was 3.927 and 0.296, respectively (background OD=0.062). So in the next step i explored above mentioned range of concentrations and I got weird result. OD of all samples was too high to be measured, background was fine with OD 0.048. I repeated my checkerboard just to see if I got the same result and make sure that all reagents are fine. Checkerboard was the same as the first time so I assumed reagents are fine. I checked my calculations and don't think I made mistake. If somebody could help me to solve this problem I would be really grateful

kas on Jul 1 2009, 05:34 PM said:

I think it would help everyone if you clarify what type of assay you are using. I do not know what a "direct" ELISA is? Do you have an antigen capture assay, antibody capture method or an immunometric assay? If you don´t know how to categorize your method just give us an outline of the assay steps...

Regards

ELISA is completely new to me and I also don't come from biology/biotech background so I am not sure if the way I am performing assay is fine.I will be grateful for any suggestions.
The steps of the assay are:
1. coating the plate with insulin in carbonate/bicarbonate buffer for 2 h at 37C. Plate washed with TPBS 4 times
2.Blocking with SoluBlockT20 for 30 min at 37. Plate washed 4 times with TPBS
3. Incubation with HRP conjugated anti-insulin antibody for 1 h at 37. Plate washed 6 times with TPBS.
4. Detection by addition of TMB and incubation for 30 min at 37 following stopping the reaction with 1 M sulphuric acid. OD read at 450 nm.

kas on Jul 1 2009, 10:24 PM said:

If you are attempting to make a quantitative assay I think you could have major problems with this format. Quantitative elisa methods are most commonly immunometric (sandwich) assays.

Anyway, I think that you may be having problems with the coating step. The kinetics of protein adsorption in microwells are very slow. Thus a short 2 hour incubation may result in large between-day variations in signal. It takes >12 hours to reach some form of near equilibrium during coating, even if your plates are shaken.
You could try coating for longer an see if this improves day-to-day reproducability.

Hope this helps

Yes I realize that I might have problems when quantifying my insulin in serum, however I just would like to use it to quantify insulin in buffer during my in vitro release experiments. As I am a complete novice I started from the simplest assay just to check if it would work at all.
I have performed also checkerboard where I coated the plate overnight (16h) at 4C and I obtained similar result to the one when I coated the plate for 2 h at 37C that's why I assumed coating for 2 h at 37C is sufficient. All suggestions will be much appreciated. Thank you for help.

Insulin in commercial test kits is measured by sandwich method: 2 antibodies. The 2 antibodies usually recognize different portions of the antigen.

Here is a link: http://www.clinchem.org/cgi/content/full/52/7/1423 which may be of interest to you.

You should be able to purchase the antibodies from commercial sources.

Tests for small molecules are all competative using one ab and a conjugated antigen which competes with the ag in the sample for binding sites of the ab.

Let us know if this helps. Always refer to what commercial sources are doing...it will point you in the right direction.

Yes I agree that probably the best for me would be to develop sandwich ELISA and this was my initial thought. I even found pair of antibodies tested for sandwich Elisa to detect insulin, however the price of antibodies and especially small volume of capture antibody proved to be not cost-effective for method development and optimization. It would be actually cheaper to buy commercial kit instead, however again with number of samples to be analysed from my release studies that option was discounted. That is why I have been trying direct Elisa which works out much cheaper.
Regarding my assay I can say that I encounter strange phenomenon. Like I mentioned earlier when serial dilutions were performed result showed that linear calibration range was somewhere between 156ng/ml and 78ng/ml, but when I explored that range in separate plate OD values of all samples within that range were too high to be measured. So in another plate I i explored range from 78 ng/ml to 0 ng/ml and I found thta linear range lies between 30 and 0ng/ml. I suspect that high concentartions of insulin in other wells affected somehow Od values for low concentartions of insulin. Although I dont think the reason was cross-contamination.
I will be grateful for any thoughts.

sgt4boston on Jul 7 2009, 08:19 PM said:

I don't understand? You have abs to develop the assay in the correct format but they are too expensive so you want to develop a test in the wrong format which would be cheaper but does not work?! There are many sources of abs I am certain you should be able to find less expensive sources. (Meridian, Biospacific, Fitzgerald, Dako, etc...)


A final option for you is to contact a commercial lab or hospital lab and negotiate a one time only price for running all your samples. Thus, you will get the correct results and not have to spend much money or waste time developing a test which is already available from many sources.

dear Sir/Madam, I think there are some problems in you methods, just personal views, if you have much samples, maybe you should buy pair abs to develop sandwich ELISA kit. I know pair abs is very expensive, but if you want to obtain right results, you should do like that. if you just have a little of samples, I advise you to buy commercialized sandwich ELISA kit. Our company have insulin sandwich ELISA kit, if you need ,please leave more information for me to provide you the regarding information .Or you can write to me. my e-mail:cusabio@hotmail.com