Non-target Signal Issue - (May/25/2010 )

Hi all,

I've been developing an sandwich ELISA for an Arabidopsis leaf protein. Below are the details of my procedure so y'all can adequately judge my method. I've just tested my method with the following kinds of samples:

1. An induced control, which shows crisp, clear bands on Western blot consistently
2. Null-mutant extract with 1 ng/uL of target protein added as a control
3. Null-mutant extract as a negative control (and to check for non-target signal)

After the ELISA I get great standard curves. But the null-mutant negative controls show equal or greater signal than the positive controls. Furthermore, I calculate the target concentration that should be 1 ng/uL as around 1.9 ng/uL. I'm really baffled by this and getting fearful my efforts have been for nothing.

Any advice at all would be hugely helpful. Thanks so much in advance!


Here are a few extra details:
----------------------------------------------------------------------
A. I just realized that I personally have never actually run the null-mutant on Western Blot but obviously someone has. However, I've blotted the whole length of the gel plenty of times with both my Primary and Secondary antibody and only very, very occasionally seen extra bands, which have always been only slightly smaller than the target and were interpreted as degraded target.

B. There was a miscalculation in my TBST recipe so that it was 0.01% Tween not the intended 0.2%... could that be a problem?


Below are the details of the procedure:
-----------------------------------------------------------------------
Protein Preparation:
1. Extract leaf proteins with SDS Buffer
2. Precipitate proteins with acetone and re-suspend in PBS (or Triton X-100 Buffer). Re-suspended samples show same level of signal on Western Blot using the same antibodies as in ELISA... single, crisp bands with no other signals at all.

Sandwich ELISA:
1. Coat overnight with 100 ul/well 1:4000 Rabbit polyclonal (recognizes N-terminus of target) primary antibody
2. Wash 3X with 200 ul/well TBST
3. Block overnight with 200 ul/well 5% Non-fat Dry Milk
4. Rotate for 3 hours with 90 ul (normally 100 uL) of sample (note that the protein standard was also added as 90 uL/well)
5. Wash 3X with 200 ul/well TBST
6. Rotate for 2 hours with 100 ul/well 1:1000 Hen polyclonal (recognizes C-terminus of target) secondary antibody labelled with Biotin
7. Wash 3X with 200 ul/well TBST
8. Rotate for 1 hour with 100 ul/well 1:1000 Streptavidin-AP
9. Wash 3X with 200 ul/well TBST
10. Develop with 100 uL ready-to-use pNPP reagent
11. Read at 405 nm!

You should run the "null mutant" (negative) and "induced" (positive) side by side in western to insure that the primary ab does not react.

Reading unknown from the dose response curve you expect 1 ug/ml but observe 1.9 ug/ml. Is this just one sample reading??
In other words are you reading 2X higher than expected for all samples? Is there a 'reference' material you can use to confirm your observations and standard value assignment? Is the matrix of the sample the same as the matrix of the standard? Although you read 'higher' do all your sample results exhibit the same trend (ie run a methods comparison plot...does the slope = 2?).

I don't believe the tween has much to do with the problem as you did not indicate you had any background problems.

sgt4boston on May 25 2010, 12:26 PM said:

SGT:

Going to run the Western like you suggested.

I expected 1 ng/uL because I took purified target protein and mixed it in with the null-mutant extract. But 12 replicates of that mix showed the absorbance expected for ~1.9 ng/uL.

There were also samples of null-mutant extract without any target protein added and they rang in at about the same absorbances, if not higher.

As for the matrix my standard was diluted in, my proteins are suspended in either PBS or a 0.8% Triton X-100 buffer. I made protein standards (and blanks) with both of these buffers and made my calculations for each sample using the buffer appropriate blank and standard. The only difference in "matrix" between my samples and the standard is the presence of protein both target and non-target!

If I were to dilute my antigen standard using the null-mutant extract it would "cost" me ~ 2 mL of extract per plate. That amount of volume is very difficult to come by so I really want/need to avoid that.

Hello,
Another quick test would be to coat some wells side by side with (1) positive control target protein (2) null extract protein "questionable negative" (3) blocking solution only and (4) if possible, a completely unrelated protein to serve as negative. All 4 should be in coated in buffer without surfactant. You can then check binding of your primary ab etc. If you see binding here and in the blotting test your ab could be cross reactive. You will have to titer your reagents so the blanks and negative read close to 0. You can also coat at different levels of protein to see the level of primary ab reaction.

You also indicated you have a nice dose response curve. To confirm the cross reactivity run the two proteins in the exact same matrix and the exact same concentrations extending the range of your standard curve. If their is 100% cross reactivity the curves will lie on top of one another. If the ab reacts more strongly to the null compared to your target, the curve will lie to the right (higher signals); if ab reacts less strongly then curve will lie to left (closer to y axis).

Similar observations are seen with the following tests: competative assays for drugs (such as amphetamines) and steroids. Also for hormones such as LH, FSH, hCG. Some of the "cross reactants" have to be several orders of magnitude higher compared to the analyte to show positive result OR much lower if the ab is poor.

I am running the Western the SGT4Boston has suggested.

In the meantime, I have one other concern I'd love feedback on:

If I did not wash thoroughly enough, could this cause the background signal?

In particular, I have read the Araibodpsis tissue contains biotin. Could it be that if tissue wasn't fully washed of the wells and there is endogenous biotin in my sample that Streptavidin-AP is binding to biotin that IS NOT attached to my secondary antibody?

How could I determine this? I have not been able to find information online about it.

Thanks,

M

Hello,
If the sample is captured by your primary ab and has active biotin it will be bound by the SA-alk phos.

The way to easily check would be coat some wells with this material. Positive controls would be ab-biotin and/or bsa-biotin. Negatives would be ab and/or bsa.

wash the wells, block, wash, react with SA-Alk phos and you will have your answer.


In your actual assay you could increase the number of washes used after the sample is added...providing that the sticking is non-specific. If the ab is capturing the protein-biotin then it will remain bound.