Is your blocking protein also included in the buffers used to incubate sera and secondary antibodies?

Which is your blocking protein? I would suggest powder milk if you have background problems.

We use BSA for blocking. I have also considered to use skimmed milk but this is not compatible with the biotin / avidin detection system.

ELISA is more sensitive than direct assay in buffer. Always need to optimize the antibodies dilution factors in an ELISA system by the checkerboard titration with known concentration of antigen and blanks. Yes, need to use BSA blocking for biotin/avidin detection system.

One thing is to set up an assay in a buffer system, but in the presence of plasma/serum, you may need to re-evaluate Ab-concentrations. Are you sure the plasma/serum you use for spiking is devoided of this biomarker? If you know this, then I would also go for another Ab-titration experiment including plasma/serum. I assume you have diluted plasma/serum as much as you think is possible in order to measure this biomarker in samples in the future. If you use two monoclonal from same specie you may have problems with e.g. HAMAs?

The biomarker we are looking at is indeed present in the human circulation and I agree that setting up an assay in a buffer system is a different story when compared to plasma / serum. Initially, we optimized antibody concentrations in the first system but we will address this issue again in serum / plasma.

In that case you don´t know if it is background or true signal (0.6 OD). I guess depleted plasma/serum would be useful to determine real background.

It is quite conceivable that the background represents a true signal. Therefore, analysis of a depleted sample would be great but diffucult to obtain. Nevertheless, the deviation of the OD values that we observer in serum or plasma when compared to the standard curve generated in buffer does point towards a matrix effect.

Serum or plasma should be diluted in assay buffer at least in 1:10 for ELISA. Less dilution may give a matrix effect. Sample dilution effect would be another thing to look at it.

We tried different blocking agents and reevaluted antibody concentrations but we got smiliar results. Apparently, BSA is good for blocking in combination with buffer and plasma / serum. Moreover, the antibody concentrations, as determined by checkerboard analysis, are also fine for buffer and serum / plasma. We will now try to optimize the dilution for serum / plasma.

are you using monoclonals? 

They often recognizes epitopes which are not oly present at your selected antigen. They are present on many (like DNA, tetanus toxid, collagen, ...)

I checked for my PhD more than 400 different (commercial) murine monoclonals and found more than 40% of them with those multirecative properties. (see attached paper). So I was able to build an ELISA for the detection of AFP using a monoclonal against SOD at the solid phase and an HRP-labelled anti-HBsAG. The detecxtion limit was not that fare from an commercial assay. 

The attached paper refers to the human monoclonals only. There was no difference between the human and murine antibodies. 

So try if you can to use a polyclonal at least at the solid phase. Check the current used antibodies in other assay systems (western blotting using complex antigens like serum, immunhistology on different tissues), You can runcompetition assay using free DNA to compete the binding to hte solid phase antibody. If you find one antibody of your pair as speciofic, you should use this at the solid phase. 

All the other optimization experiments are not helpful. To decreas low affinity (unspecific) binding, you should inmcrease the salt concentration (0.3 mol/L NaCl, phosphate buffer is fine, the innert protein should be used in the sample and conjugate incubation buffer only. Forget all this blocking voodo. You are using a BSA which is probanly not free from your analyte. The whole blocking is done by using the washing buffer the first time at the coated plates, the detergent will do the job within a very few seconds. 

Thanks for the information and yes we use monoclonals to build our ELISA.

Check them for multireactivity. 

Try to use different concentrations of Tween 20. (i.e: Final concentration of  tween20 in the serum sample should be in the range of 0.007% - 0.5 %).

You can use a previously prepared 20% tween, in order to use a very small quantity for conditioning the sample without diluting it.

for example; 100μL of serum can be conditioned with 2.5μL of 20% Tween20, in a case of 0.5% as the final concentration of tween20.

97.5 μL serum sample + 2.5 μL 20% Tween20

You should also take into consideration the state of the serum samples, a lot of them contain more lipids or interfering components. Using very clear serum for spiking would be better. Also using a non-human serum as a negative control and for spiking is much better.

It's easier to dilute your sample in a "sample buffer". Normally a phosphate buffer is used (PBS like), it's 10 mmol/L Phosphate pH 7.2, add 300 mmol/L NaCl (This dobled salt concentration will help you to avoid unspecific reaction wich are taking place with low affinities only), 0.1% Tween 20, and an inert protein (like 3% BSA, 3% hydrolysed gealtine which is available as a drug from the pharmacy as plasma expander). Add some phenol red to see better where you are pipetting.

The washing buffer is only without the inert protein and without phenol red.

You should use the sample dilution buffer also for the dilution of the conjugated antibody.

Your ELISA should be established in a way that you can dilute your sample at least 1:10. This is to bring the detergent and the NaCl into your reaction. The detergent is also used to avoid unspecific (hydrophobic) interactions. The inert protein is used to keep the protein condition constant. The Phospahe to keep the pH constant.

To define the optimal dilution, just run in one direction (colums) of the microtiter plate the dilution (start e.g. 1:5, 1:10 ...) and in the second direction (rows) the buffer condition by variations of NaCl, inert protein concentrations, Tween concentration. For the evaluation you should use a positive sample and a negative sample so you can calculate the "positive/negative ratio" for each dilution, for each concentration of your buffer components. Just use the one with the highest value of the P/N ratio for the following experiments. By going this way, you can safe a lot of time ...

Stephan

Did you try different coating buffers? I would try several coating buffers with different pH as an isoelectric point has a strong influence to solubility. And what the biomarker is? Is it a protein? I would also try different plates for ELISA as they basically have different sorption capacity.