Protocol Online logo
Top : New Forum Archives (2009-): : Immunology

Problems with ELISA...very low levels in human serum - (Mar/08/2010 )


I have a few problems with an assay which I'm trying to optimize. The assay originally is for cell culture. The production company (R&D) suggests to increase blocking buffer concentration to 10% BSA for serum samples which is what I did. The primary antibody is mouse anti-human protein. The secondary is goat anti-human protein. The biggest problem I had was enormously high background (1.2). After some reading and re-experimentation, I've made some changes. I switched to 5% horse serum block (in PBS), 10% horse serum in standards. Decreased primary antibody coating. The secondary is prepared in 2% goat serum. It uses TMP substrate. I repeated the experiment and got readings above 500 pg/mL, but below, absorbance values were inconsistent. Turns out the protein is less than 500 pg/mL in human serum, whereas it's 2-9 ng in BSA (which might explain the high background in the first runs). In my last run, I spiked the samples directly before plating w/ 500 pg of the protein (100 uL sample with 3.5 uL protein; background was still high, but I saw the trend I wanted (the later samples (it was a bleeding time course, and we expected to see levels decrease with time) had values below 500 pg, which makes no sense!) I'm not sure exactly what's going on. I don't know if it's the blocking, the secondary diluent, spiking issues? I've read some stuff about heterophile antibodies, and it seems like it might be that. Another thing, which I've never questioned...why do I dilute the secondary (goat anti-human protein) in goat serum? My understanding is that the antibodies in the serum won't bind the goat antibody. Is this right? Should I be blocking with goat serum? mouse serum? human serum? For spiking, should I dilute the protein in BSA or something so that I can ensure it won't stick to the sides of the tubes before I plate? I would really appreciate any help!


You have lots of information here.

Firstly since your analyte is in BSA you can not use that for blocking or in buffer matrices since it will reaact with your abs.

2nd you kit was developed for cell culture...nice clean samples with primarily the analyte being in the matrix along with tissue culture fluid. You are now using that kit for human sera analysis.

1. the matrix for the standards should be as similar as possible to the sample matrix. You can probably find a stripped human sera to use. If you don't use this you will have a matrix issue and if you do a methods comparison your slope and intercept will be different.

2. For blocking you can try casein or even a purified BSA that does not have your analyte or synthetic blocker or fish protein (East Coast Biologics). Try to avoid adding your ag back into the system.

3. Your abs are both mouse and goat (mab and poly). As you wish to have high sensitivity you may encounter heterophile abs (can be very very high titer). To block these you should use mixture of mouse and goat sera or even better non specific mouse and goat IgG abs. The object being to block human abs from reacting with abs in your system.

4. Pre-incubate your sample with the blocking agents before analysis and if possible dilute your sample 1:10 before analysis to dilute out the heterophiles.

5. Test you curve with the human sera matrix in parallel with a buffer matrix.

let us know how you are doing.
good luck


your problem is that you are using a secondary that is not against your primary, it is against your sample. this gives high background.

your secondary should be goat anti-mouse igg.


Thank you for the excellent suggestions, sgt4boston and mdfenko! mdfenko, I'm not sure how that could work, since the Fc region of the mouse anti human protein is stuck to the plate. I'm using the sandwich elisa method, but, if you happen to know of a modification, I'm all ears, hehe! I repeated the assay today with 2 different conditions.

Condition 1:
Block with 5% goat serum in PBS
Standards prepared in 5% goat serum
Samples undiluted
Secondary prepared in 2% goat serum

Condition 2:
Block with 5% human serum in PBS
Standards prepared in 5% human serum
Samples diluted 1:20 (effectively 5% human serum, as in standards)
Secondary prepared in 2% goat serum

For condition 1, my rationale was that if something in the human serum is interfering, perhaps heterophile Ab, then blocking with goat serum would rid human Ab against the goat "stuff." Won't have to worry about heterophile ab in secondary.

For condition 2, my rationale was that if something in the human serum is interfering, then blocking with human serum would not have a heterophile effect (though this might bring trouble w/ secondary since it's in goat serum.

Turns out, condition 1 worked. Not just worked, but gave me the best results ever! Low background (.3), perfect standard curve, and absolutely the trend I wanted to see in the sample data (and the values were well within the standard curve).

Condition 2 was another complete failure (background of .8 and negative values for sample data.

My question now, is my rationale for condition 1 correct? Is there something else that I'm missing? I really want to understand why it worked so well. I will repeat the assay by the end of the weekend (I have to get ready for Finals, unfortunately >.<). Any input would be appreciated!

Attached File

Attached File


I think there is confusion in the terms being used. You are using the MAB as a capture on the plate and your detection is polyclonal which is conjugated to enzyme? (did I make the correct assumption that your mab was on the plate). FYI adsorbed abs stick in random fashion, fc, heavy, light, etc not all orientation is with binding portion up.

Your human samples may also have anti-goat abs and anti-mouse abs. Heterophiles can cause both elevated and depressed signals.

You indicated good curve and results. the only way you will know this is if you do a methods comparison study...are the values correct vs. reference method? Also a spike and recovery; spike high sample into a lower and mid level you recover ~~90%?


Yes, the MAB is capture; the polyclonal is biotinylated goat anti-analyte. For methods comparison, do you mean compare to the original protocol results? I will try the spike and recover. Thanks, sgt4boston!


By method comparison I mean plot your test results (y axis) versus another method or reference method results for the same analytes. Ideal result would be slope of 1 intercept of 0. This is a test for accuracy.

Spike and recovery: if insufficient volume of sample you can pool high, med, low analyze the pools and spike high into each. Recovery should be > 90%.

Also, you can serially dilute the high or medium sample/pool and the high standard in parallel with one another and test...they should dilute linearly and the lines should be parallel with one another...ideal they would be on top of one another. This is a test for matrix effects.

You should also run replicates of several samples and your standard and check for %CV...should be < 10% <5% ideal; a test for precision.

If you have no reference method you can plot result versus clinical condition, time, etc where values should increase or decrease accordingly.

These tests will confirm the validity of your test results.